0 files changed, 0 insertions, 0 deletions
diff --git a/drivers/gpu/Makefile b/drivers/gpu/Makefile
index e9ed439a5b65..36a54d456630 100644
--- a/drivers/gpu/Makefile
+++ b/drivers/gpu/Makefile
@@ -1,6 +1,8 @@
+# SPDX-License-Identifier: GPL-2.0-only
 # drm/tegra depends on host1x, so if both drivers are built-in care must be
 # taken to initialize them in the correct order. Link order is the only way
 # to ensure this currently.
-obj-$(CONFIG_TEGRA_HOST1X)	+= host1x/
-obj-y			+= drm/ vga/
+obj-y			+= host1x/ drm/ vga/
 obj-$(CONFIG_IMX_IPUV3_CORE)	+= ipu-v3/
+obj-$(CONFIG_TRACE_GPU_MEM)		+= trace/
+obj-$(CONFIG_NOVA_CORE)		+= nova-core/
diff --git a/drivers/gpu/drm/Kconfig b/drivers/gpu/drm/Kconfig
index 83cb2a88c204..7e6bc0b3a589 100644
--- a/drivers/gpu/drm/Kconfig
+++ b/drivers/gpu/drm/Kconfig
@@ -1,3 +1,4 @@
+# SPDX-License-Identifier: GPL-2.0-only
 #
 # Drm device configuration
 #
@@ -7,12 +8,15 @@
 menuconfig DRM
 	tristate "Direct Rendering Manager (XFree86 4.1.0 and higher DRI support)"
 	depends on (AGP || AGP=n) && !EMULATED_CMPXCHG && HAS_DMA
+	select DRM_PANEL_ORIENTATION_QUIRKS
 	select HDMI
-	select FB_CMDLINE
 	select I2C
-	select I2C_ALGOBIT
 	select DMA_SHARED_BUFFER
 	select SYNC_FILE
+# gallium uses SYS_kcmp for os_same_file_description() to de-duplicate
+# device and dmabuf fd. Let's make sure that is available for our userspace.
+	select KCMP
+	select VIDEO
 	help
 	  Kernel-level support for the Direct Rendering Infrastructure (DRI)
 	  introduced in XFree86 4.0. If you say Y here, you need to select
@@ -22,95 +26,154 @@ menuconfig DRM
 	  details.  You should also select and configure AGP
 	  (/dev/agpgart) support if it is available for your platform.
 
+menu "DRM debugging options"
+depends on DRM
+source "drivers/gpu/drm/Kconfig.debug"
+endmenu
+
+if DRM
+
+config DRM_MIPI_DBI
+	tristate
+	depends on DRM
+	select DRM_KMS_HELPER
+
 config DRM_MIPI_DSI
 	bool
 	depends on DRM
 
-config DRM_DP_AUX_CHARDEV
-	bool "DRM DP AUX Interface"
+config DRM_KMS_HELPER
+	tristate
 	depends on DRM
+	select FB_CORE if DRM_FBDEV_EMULATION
 	help
-	  Choose this option to enable a /dev/drm_dp_auxN node that allows to
-	  read and write values to arbitrary DPCD registers on the DP aux
-	  channel.
-
-config DRM_DEBUG_MM
-	bool "Insert extra checks and debug info into the DRM range managers"
-	default n
-	depends on DRM=y
-	depends on STACKTRACE_SUPPORT
-	select STACKDEPOT
-	help
-	  Enable allocation tracking of memory manager and leak detection on
-	  shutdown.
-
-	  Recommended for driver developers only.
+	  CRTC helpers for KMS drivers.
 
-	  If in doubt, say "N".
+config DRM_DRAW
+	bool
+	depends on DRM
 
-config DRM_DEBUG_MM_SELFTEST
-	tristate "kselftests for DRM range manager (struct drm_mm)"
+config DRM_PANIC
+	bool "Display a user-friendly message when a kernel panic occurs"
 	depends on DRM
-	depends on DEBUG_KERNEL
-	select PRIME_NUMBERS
-	select DRM_LIB_RANDOM
-	default n
+	select FONT_SUPPORT
+	select DRM_DRAW
 	help
-	  This option provides a kernel module that can be used to test
-	  the DRM range manager (drm_mm) and its API. This option is not
-	  useful for distributions or general kernels, but only for kernel
-	  developers working on DRM and associated drivers.
-
+	  Enable a drm panic handler, which will display a user-friendly message
+	  when a kernel panic occurs. It's useful when using a user-space
+	  console instead of fbcon.
+	  It will only work if your graphic driver supports this feature.
+	  To support Hi-DPI Display, you can enable bigger fonts like
+	  FONT_TER16x32
+
+config DRM_PANIC_FOREGROUND_COLOR
+	hex "Drm panic screen foreground color, in RGB"
+	depends on DRM_PANIC
+	default 0xffffff
+
+config DRM_PANIC_BACKGROUND_COLOR
+	hex "Drm panic screen background color, in RGB"
+	depends on DRM_PANIC
+	default 0x000000
+
+config DRM_PANIC_DEBUG
+	bool "Add a debug fs entry to trigger drm_panic"
+	depends on DRM_PANIC && DEBUG_FS
+	help
+	  Add dri/[device]/drm_panic_plane_x in the kernel debugfs, to force the
+	  panic handler to write the panic message to this plane scanout buffer.
+	  This is unsafe and should not be enabled on a production build.
 	  If in doubt, say "N".
 
-config DRM_KMS_HELPER
-	tristate
-	depends on DRM
+config DRM_PANIC_SCREEN
+	string "Panic screen formatter"
+	default "user"
+	depends on DRM_PANIC
 	help
-	  CRTC helpers for KMS drivers.
-
-config DRM_KMS_FB_HELPER
-	bool
-	depends on DRM_KMS_HELPER
-	select FB
-	select FRAMEBUFFER_CONSOLE if !EXPERT
-	select FRAMEBUFFER_CONSOLE_DETECT_PRIMARY if FRAMEBUFFER_CONSOLE
-	select FB_SYS_FOPS
-	select FB_SYS_FILLRECT
-	select FB_SYS_COPYAREA
-	select FB_SYS_IMAGEBLIT
-	select FB_CFB_FILLRECT
-	select FB_CFB_COPYAREA
-	select FB_CFB_IMAGEBLIT
-	select FB_DEFERRED_IO
+	  This option enable to choose what will be displayed when a kernel
+	  panic occurs. You can choose between "user", a short message telling
+	  the user to reboot the system, or "kmsg" which will display the last
+	  lines of kmsg.
+	  This can also be overridden by drm.panic_screen=xxxx kernel parameter
+	  or by writing to /sys/module/drm/parameters/panic_screen sysfs entry
+	  Default is "user"
+
+config DRM_PANIC_SCREEN_QR_CODE
+	bool "Add a panic screen with a QR code"
+	depends on DRM_PANIC && RUST
+	select ZLIB_DEFLATE
+	help
+	  This option adds a QR code generator, and a panic screen with a QR
+	  code. The QR code will contain the last lines of kmsg and other debug
+	  information. This should be easier for the user to report a kernel
+	  panic, with all debug information available.
+	  To use this panic screen, also set DRM_PANIC_SCREEN to "qr_code"
+
+config DRM_PANIC_SCREEN_QR_CODE_URL
+	string "Base URL of the QR code in the panic screen"
+	depends on DRM_PANIC_SCREEN_QR_CODE
+	help
+	  This option sets the base URL to report the kernel panic. If it's set
+	  the QR code will contain the URL and the kmsg compressed with zlib as
+	  a URL parameter. If it's empty, the QR code will contain the kmsg as
+	  uncompressed text only.
+	  There is a demo code in javascript, to decode and uncompress the kmsg
+	  data from the URL parameter at https://github.com/kdj0c/panic_report
+
+config DRM_PANIC_SCREEN_QR_VERSION
+	int "Maximum version (size) of the QR code."
+	depends on DRM_PANIC_SCREEN_QR_CODE
+	default 40
 	help
-	  FBDEV helpers for KMS drivers.
+	  This option limits the version (or size) of the QR code. QR code
+	  version ranges from Version 1 (21x21) to Version 40 (177x177).
+	  Smaller QR code are easier to read, but will contain less debugging
+	  data. Default is 40.
 
-config DRM_FBDEV_EMULATION
-	bool "Enable legacy fbdev support for your modesetting driver"
-	depends on DRM
-	select DRM_KMS_HELPER
-	select DRM_KMS_FB_HELPER
-	default y
+config DRM_DEBUG_DP_MST_TOPOLOGY_REFS
+        bool "Enable refcount backtrace history in the DP MST helpers"
+	depends on STACKTRACE_SUPPORT
+        select STACKDEPOT
+        select DRM_KMS_HELPER
+        depends on DEBUG_KERNEL
+        depends on EXPERT
+        help
+          Enables debug tracing for topology refs in DRM's DP MST helpers. A
+          history of each topology reference/dereference will be printed to the
+          kernel log once a port or branch device's topology refcount reaches 0.
+
+          This has the potential to use a lot of memory and print some very
+          large kernel messages. If in doubt, say "N".
+
+config DRM_DEBUG_MODESET_LOCK
+	bool "Enable backtrace history for lock contention"
+	depends on STACKTRACE_SUPPORT
+	depends on DEBUG_KERNEL
+	depends on EXPERT
+	select STACKDEPOT
+	default y if DEBUG_WW_MUTEX_SLOWPATH
 	help
-	  Choose this option if you have a need for the legacy fbdev
-	  support. Note that this support also provides the linux console
-	  support on top of your modesetting driver.
+	  Enable debug tracing of failures to gracefully handle drm modeset lock
+	  contention. A history of each drm modeset lock path hitting -EDEADLK
+	  will be saved until gracefully handled, and the backtrace will be
+	  printed when attempting to lock a contended lock.
 
-	  If in doubt, say "Y".
+	  If in doubt, say "N".
 
-config DRM_FBDEV_OVERALLOC
-	int "Overallocation of the fbdev buffer"
-	depends on DRM_FBDEV_EMULATION
-	default 100
+config DRM_CLIENT
+	bool
+	depends on DRM
 	help
-	  Defines the fbdev buffer overallocation in percent. Default
-	  is 100. Typical values for double buffering will be 200,
-	  triple buffering 300.
+	  Enables support for DRM clients. DRM drivers that need
+	  struct drm_client_dev and its interfaces should select this
+	  option. Drivers that support the default clients should
+	  select DRM_CLIENT_SELECTION instead.
+
+source "drivers/gpu/drm/clients/Kconfig"
 
 config DRM_LOAD_EDID_FIRMWARE
 	bool "Allow to specify an EDID data set instead of probing for it"
-	depends on DRM_KMS_HELPER
+	depends on DRM
 	help
 	  Say Y here, if you want to use EDID data to be loaded from the
 	  /lib/firmware directory or one of the provided built-in
@@ -118,91 +181,118 @@ config DRM_LOAD_EDID_FIRMWARE
 	  monitor are unable to provide appropriate EDID data. Since this
 	  feature is provided as a workaround for broken hardware, the
 	  default case is N. Details and instructions how to build your own
-	  EDID data are given in Documentation/EDID/HOWTO.txt.
+	  EDID data are given in Documentation/admin-guide/edid.rst.
+
+source "drivers/gpu/drm/display/Kconfig"
 
 config DRM_TTM
 	tristate
 	depends on DRM && MMU
+	select SHMEM
 	help
 	  GPU memory management subsystem for devices with multiple
 	  GPU memory types. Will be enabled automatically if a device driver
 	  uses it.
 
-config DRM_GEM_CMA_HELPER
-	bool
+config DRM_EXEC
+	tristate
 	depends on DRM
 	help
-	  Choose this if you need the GEM CMA helper functions
+	  Execution context for command submissions
 
-config DRM_KMS_CMA_HELPER
-	bool
+config DRM_GPUVM
+	tristate
 	depends on DRM
-	select DRM_GEM_CMA_HELPER
-	select DRM_KMS_FB_HELPER
-	select FB_SYS_FILLRECT
-	select FB_SYS_COPYAREA
-	select FB_SYS_IMAGEBLIT
+	select DRM_EXEC
 	help
-	  Choose this if you need the KMS CMA helper functions
-
-config DRM_VM
-	bool
-	depends on DRM && MMU
+	  GPU-VM representation providing helpers to manage a GPUs virtual
+	  address space
 
-source "drivers/gpu/drm/i2c/Kconfig"
+config DRM_GPUSVM
+	tristate
+	depends on DRM && DEVICE_PRIVATE
+	select HMM_MIRROR
+	select MMU_NOTIFIER
+	help
+	  GPU-SVM representation providing helpers to manage a GPUs shared
+	  virtual memory
 
-source "drivers/gpu/drm/arm/Kconfig"
+config DRM_BUDDY
+	tristate
+	depends on DRM
+	help
+	  A page based buddy allocator
 
-config DRM_RADEON
-	tristate "ATI Radeon"
-	depends on DRM && PCI && MMU
-	select FW_LOADER
-        select DRM_KMS_HELPER
-        select DRM_TTM
-	select POWER_SUPPLY
-	select HWMON
-	select BACKLIGHT_CLASS_DEVICE
-	select BACKLIGHT_LCD_SUPPORT
-	select INTERVAL_TREE
+config DRM_VRAM_HELPER
+	tristate
+	depends on DRM
 	help
-	  Choose this option if you have an ATI Radeon graphics card.  There
-	  are both PCI and AGP versions.  You don't need to choose this to
-	  run the Radeon in plain VGA mode.
+	  Helpers for VRAM memory management
 
-	  If M is selected, the module will be called radeon.
+config DRM_TTM_HELPER
+	tristate
+	depends on DRM
+	select DRM_TTM
+	select DRM_KMS_HELPER if DRM_FBDEV_EMULATION
+	select FB_CORE if DRM_FBDEV_EMULATION
+	select FB_SYSMEM_HELPERS_DEFERRED if DRM_FBDEV_EMULATION
+	help
+	  Helpers for ttm-based gem objects
 
-source "drivers/gpu/drm/radeon/Kconfig"
+config DRM_GEM_DMA_HELPER
+	tristate
+	depends on DRM
+	select DRM_KMS_HELPER if DRM_FBDEV_EMULATION
+	select FB_CORE if DRM_FBDEV_EMULATION
+	select FB_DMAMEM_HELPERS_DEFERRED if DRM_FBDEV_EMULATION
+	help
+	  Choose this if you need the GEM DMA helper functions
 
-config DRM_AMDGPU
-	tristate "AMD GPU"
-	depends on DRM && PCI && MMU
-	select FW_LOADER
-        select DRM_KMS_HELPER
-        select DRM_TTM
-	select POWER_SUPPLY
-	select HWMON
-	select BACKLIGHT_CLASS_DEVICE
-	select BACKLIGHT_LCD_SUPPORT
-	select INTERVAL_TREE
+config DRM_GEM_SHMEM_HELPER
+	tristate
+	depends on DRM && MMU
+	select DRM_KMS_HELPER if DRM_FBDEV_EMULATION
+	select FB_CORE if DRM_FBDEV_EMULATION
+	select FB_SYSMEM_HELPERS_DEFERRED if DRM_FBDEV_EMULATION
 	help
-	  Choose this option if you have a recent AMD Radeon graphics card.
+	  Choose this if you need the GEM shmem helper functions
+
+config DRM_SUBALLOC_HELPER
+	tristate
+	depends on DRM
+
+config DRM_SCHED
+	tristate
+	depends on DRM
 
-	  If M is selected, the module will be called amdgpu.
+source "drivers/gpu/drm/sysfb/Kconfig"
+
+source "drivers/gpu/drm/arm/Kconfig"
+
+source "drivers/gpu/drm/radeon/Kconfig"
 
 source "drivers/gpu/drm/amd/amdgpu/Kconfig"
 
 source "drivers/gpu/drm/nouveau/Kconfig"
 
+source "drivers/gpu/drm/nova/Kconfig"
+
 source "drivers/gpu/drm/i915/Kconfig"
 
+source "drivers/gpu/drm/xe/Kconfig"
+
+source "drivers/gpu/drm/kmb/Kconfig"
+
 config DRM_VGEM
 	tristate "Virtual GEM provider"
-	depends on DRM
+	depends on DRM && MMU
+	select DRM_GEM_SHMEM_HELPER
 	help
 	  Choose this option to get a virtual graphics memory manager,
 	  as used by Mesa's software renderer for enhanced performance.
 	  If M is selected the module will be called vgem.
 
+source "drivers/gpu/drm/vkms/Kconfig"
 
 source "drivers/gpu/drm/exynos/Kconfig"
 
@@ -218,15 +308,11 @@ source "drivers/gpu/drm/ast/Kconfig"
 
 source "drivers/gpu/drm/mgag200/Kconfig"
 
-source "drivers/gpu/drm/cirrus/Kconfig"
-
 source "drivers/gpu/drm/armada/Kconfig"
 
 source "drivers/gpu/drm/atmel-hlcdc/Kconfig"
 
-source "drivers/gpu/drm/rcar-du/Kconfig"
-
-source "drivers/gpu/drm/shmobile/Kconfig"
+source "drivers/gpu/drm/renesas/Kconfig"
 
 source "drivers/gpu/drm/sun4i/Kconfig"
 
@@ -236,8 +322,6 @@ source "drivers/gpu/drm/tilcdc/Kconfig"
 
 source "drivers/gpu/drm/qxl/Kconfig"
 
-source "drivers/gpu/drm/bochs/Kconfig"
-
 source "drivers/gpu/drm/virtio/Kconfig"
 
 source "drivers/gpu/drm/msm/Kconfig"
@@ -254,109 +338,94 @@ source "drivers/gpu/drm/bridge/Kconfig"
 
 source "drivers/gpu/drm/sti/Kconfig"
 
-source "drivers/gpu/drm/amd/amdkfd/Kconfig"
-
 source "drivers/gpu/drm/imx/Kconfig"
 
+source "drivers/gpu/drm/ingenic/Kconfig"
+
+source "drivers/gpu/drm/v3d/Kconfig"
+
 source "drivers/gpu/drm/vc4/Kconfig"
 
-source "drivers/gpu/drm/etnaviv/Kconfig"
+source "drivers/gpu/drm/loongson/Kconfig"
 
-source "drivers/gpu/drm/arc/Kconfig"
+source "drivers/gpu/drm/etnaviv/Kconfig"
 
 source "drivers/gpu/drm/hisilicon/Kconfig"
 
-source "drivers/gpu/drm/mediatek/Kconfig"
+source "drivers/gpu/drm/logicvc/Kconfig"
 
-source "drivers/gpu/drm/zte/Kconfig"
+source "drivers/gpu/drm/mediatek/Kconfig"
 
 source "drivers/gpu/drm/mxsfb/Kconfig"
 
 source "drivers/gpu/drm/meson/Kconfig"
 
-source "drivers/gpu/drm/tinydrm/Kconfig"
+source "drivers/gpu/drm/tiny/Kconfig"
 
 source "drivers/gpu/drm/pl111/Kconfig"
 
-# Keep legacy drivers last
+source "drivers/gpu/drm/tve200/Kconfig"
 
-menuconfig DRM_LEGACY
-	bool "Enable legacy drivers (DANGEROUS)"
-	depends on DRM && MMU
-	select DRM_VM
-	help
-	  Enable legacy DRI1 drivers. Those drivers expose unsafe and dangerous
-	  APIs to user-space, which can be used to circumvent access
-	  restrictions and other security measures. For backwards compatibility
-	  those drivers are still available, but their use is highly
-	  inadvisable and might harm your system.
+source "drivers/gpu/drm/xen/Kconfig"
 
-	  You are recommended to use the safe modeset-only drivers instead, and
-	  perform 3D emulation in user-space.
+source "drivers/gpu/drm/vboxvideo/Kconfig"
 
-	  Unless you have strong reasons to go rogue, say "N".
+source "drivers/gpu/drm/lima/Kconfig"
 
-if DRM_LEGACY
+source "drivers/gpu/drm/panfrost/Kconfig"
 
-config DRM_TDFX
-	tristate "3dfx Banshee/Voodoo3+"
-	depends on DRM && PCI
-	help
-	  Choose this option if you have a 3dfx Banshee or Voodoo3 (or later),
-	  graphics card.  If M is selected, the module will be called tdfx.
+source "drivers/gpu/drm/panthor/Kconfig"
 
-config DRM_R128
-	tristate "ATI Rage 128"
-	depends on DRM && PCI
-	select FW_LOADER
-	help
-	  Choose this option if you have an ATI Rage 128 graphics card.  If M
-	  is selected, the module will be called r128.  AGP support for
-	  this card is strongly suggested (unless you have a PCI version).
-
-config DRM_I810
-	tristate "Intel I810"
-	# !PREEMPT because of missing ioctl locking
-	depends on DRM && AGP && AGP_INTEL && (!PREEMPT || BROKEN)
-	help
-	  Choose this option if you have an Intel I810 graphics card.  If M is
-	  selected, the module will be called i810.  AGP support is required
-	  for this driver to work.
-
-config DRM_MGA
-	tristate "Matrox g200/g400"
-	depends on DRM && PCI
-	select FW_LOADER
-	help
-	  Choose this option if you have a Matrox G200, G400 or G450 graphics
-	  card.  If M is selected, the module will be called mga.  AGP
-	  support is required for this driver to work.
-
-config DRM_SIS
-	tristate "SiS video cards"
-	depends on DRM && AGP
-	depends on FB_SIS || FB_SIS=n
-	help
-	  Choose this option if you have a SiS 630 or compatible video
-	  chipset. If M is selected the module will be called sis. AGP
-	  support is required for this driver to work.
+source "drivers/gpu/drm/aspeed/Kconfig"
 
-config DRM_VIA
-	tristate "Via unichrome video cards"
-	depends on DRM && PCI
-	help
-	  Choose this option if you have a Via unichrome or compatible video
-	  chipset. If M is selected the module will be called via.
+source "drivers/gpu/drm/mcde/Kconfig"
+
+source "drivers/gpu/drm/tidss/Kconfig"
+
+source "drivers/gpu/drm/adp/Kconfig"
+
+source "drivers/gpu/drm/xlnx/Kconfig"
 
-config DRM_SAVAGE
-	tristate "Savage video cards"
-	depends on DRM && PCI
+source "drivers/gpu/drm/gud/Kconfig"
+
+source "drivers/gpu/drm/sitronix/Kconfig"
+
+source "drivers/gpu/drm/solomon/Kconfig"
+
+source "drivers/gpu/drm/sprd/Kconfig"
+
+source "drivers/gpu/drm/imagination/Kconfig"
+
+source "drivers/gpu/drm/tyr/Kconfig"
+
+config DRM_HYPERV
+	tristate "DRM Support for Hyper-V synthetic video device"
+	depends on DRM && PCI && HYPERV_VMBUS
+	select DRM_CLIENT_SELECTION
+	select DRM_KMS_HELPER
+	select DRM_GEM_SHMEM_HELPER
 	help
-	  Choose this option if you have a Savage3D/4/SuperSavage/Pro/Twister
-	  chipset. If M is selected the module will be called savage.
+	 This is a KMS driver for Hyper-V synthetic video device. Choose this
+	 option if you would like to enable drm driver for Hyper-V virtual
+	 machine. Unselect Hyper-V framebuffer driver (CONFIG_FB_HYPERV) so
+	 that DRM driver is used by default.
 
-endif # DRM_LEGACY
+	 If M is selected the module will be called hyperv_drm.
+
+# Separate option as not all DRM drivers use it
+config DRM_PANEL_BACKLIGHT_QUIRKS
+	tristate
 
 config DRM_LIB_RANDOM
 	bool
 	default n
+
+config DRM_PRIVACY_SCREEN
+	bool
+	default n
+
+endif
+
+# Separate option because drm_panel_orientation_quirks.c is shared with fbdev
+config DRM_PANEL_ORIENTATION_QUIRKS
+	tristate
diff --git a/drivers/gpu/drm/Kconfig.debug b/drivers/gpu/drm/Kconfig.debug
new file mode 100644
index 000000000000..05dc43c0b8c5
--- /dev/null
+++ b/drivers/gpu/drm/Kconfig.debug
@@ -0,0 +1,117 @@
+config DRM_USE_DYNAMIC_DEBUG
+	bool "use dynamic debug to implement drm.debug"
+	default n
+	depends on BROKEN
+	depends on DRM
+	depends on DYNAMIC_DEBUG || DYNAMIC_DEBUG_CORE
+	depends on JUMP_LABEL
+	help
+	 Use dynamic-debug to avoid drm_debug_enabled() runtime overheads.
+	 Due to callsite counts in DRM drivers (~4k in amdgpu) and 56
+	 bytes per callsite, the .data costs can be substantial, and
+	 are therefore configurable.
+
+config DRM_WERROR
+	bool "Compile the drm subsystem with warnings as errors"
+	depends on DRM && EXPERT
+	depends on !WERROR
+	default n
+	help
+	  A kernel build should not cause any compiler warnings, and this
+	  enables the '-Werror' flag to enforce that rule in the drm subsystem.
+
+	  The drm subsystem enables more warnings than the kernel default, so
+	  this config option is disabled by default.
+
+	  If in doubt, say N.
+
+config DRM_HEADER_TEST
+	bool "Ensure DRM headers are self-contained and pass kernel-doc"
+	depends on DRM && EXPERT && BROKEN
+	default n
+	help
+	  Ensure the DRM subsystem headers both under drivers/gpu/drm and
+	  include/drm compile, are self-contained, have header guards, and have
+	  no kernel-doc warnings.
+
+	  If in doubt, say N.
+
+config DRM_DEBUG_MM
+	bool "Insert extra checks and debug info into the DRM range managers"
+	default n
+	depends on DRM
+	depends on STACKTRACE_SUPPORT
+	select STACKDEPOT
+	help
+	  Enable allocation tracking of memory manager and leak detection on
+	  shutdown.
+
+	  Recommended for driver developers only.
+
+	  If in doubt, say "N".
+
+config DRM_KUNIT_TEST_HELPERS
+	tristate
+	depends on DRM && KUNIT
+	select DRM_KMS_HELPER
+	help
+	  KUnit Helpers for KMS drivers.
+
+config DRM_KUNIT_TEST
+	tristate "KUnit tests for DRM" if !KUNIT_ALL_TESTS
+	depends on DRM && KUNIT && MMU
+	select DRM_BRIDGE_CONNECTOR
+	select DRM_BUDDY
+	select DRM_DISPLAY_DP_HELPER
+	select DRM_DISPLAY_HDMI_STATE_HELPER
+	select DRM_DISPLAY_HELPER
+	select DRM_EXEC
+	select DRM_EXPORT_FOR_TESTS if m
+	select DRM_GEM_SHMEM_HELPER
+	select DRM_KUNIT_TEST_HELPERS
+	select DRM_LIB_RANDOM
+	select DRM_SYSFB_HELPER
+	select PRIME_NUMBERS
+	default KUNIT_ALL_TESTS
+	help
+	  This builds unit tests for DRM. This option is not useful for
+	  distributions or general kernels, but only for kernel
+	  developers working on DRM and associated drivers.
+
+	  For more information on KUnit and unit tests in general,
+	  please refer to the KUnit documentation in
+	  Documentation/dev-tools/kunit/.
+
+	  If in doubt, say "N".
+
+config DRM_TTM_KUNIT_TEST
+	tristate "KUnit tests for TTM" if !KUNIT_ALL_TESTS
+	default n
+	depends on DRM && KUNIT && MMU && (UML || COMPILE_TEST)
+	select DRM_TTM
+	select DRM_BUDDY
+	select DRM_EXPORT_FOR_TESTS if m
+	select DRM_KUNIT_TEST_HELPERS
+	default KUNIT_ALL_TESTS
+	help
+	  Enables unit tests for TTM, a GPU memory manager subsystem used
+	  to manage memory buffers. This option is mostly useful for kernel
+	  developers. It depends on (UML || COMPILE_TEST) since no other driver
+	  which uses TTM can be loaded while running the tests.
+
+	  If in doubt, say "N".
+
+config DRM_SCHED_KUNIT_TEST
+	tristate "KUnit tests for the DRM scheduler" if !KUNIT_ALL_TESTS
+	select DRM_SCHED
+	depends on DRM && KUNIT
+	default KUNIT_ALL_TESTS
+	help
+	  Choose this option to build unit tests for the DRM scheduler.
+
+	  Recommended for driver developers only.
+
+	  If in doubt, say "N".
+
+config DRM_EXPORT_FOR_TESTS
+	bool
diff --git a/drivers/gpu/drm/Makefile b/drivers/gpu/drm/Makefile
index a8acc197dec3..4b2f7d794275 100644
--- a/drivers/gpu/drm/Makefile
+++ b/drivers/gpu/drm/Makefile
@@ -1,72 +1,187 @@
+# SPDX-License-Identifier: GPL-2.0
 
 # Makefile for the drm device driver.  This driver provides support for the
 # Direct Rendering Infrastructure (DRI) in XFree86 4.1.0 and higher.
 
-drm-y       :=	drm_auth.o drm_bufs.o drm_cache.o \
-		drm_context.o drm_dma.o \
-		drm_file.o drm_gem.o drm_ioctl.o drm_irq.o \
-		drm_lock.o drm_memory.o drm_drv.o \
-		drm_scatter.o drm_pci.o \
-		drm_sysfs.o drm_hashtab.o drm_mm.o \
-		drm_crtc.o drm_fourcc.o drm_modes.o drm_edid.o \
-		drm_info.o drm_encoder_slave.o \
-		drm_trace_points.o drm_global.o drm_prime.o \
-		drm_rect.o drm_vma_manager.o drm_flip_work.o \
-		drm_modeset_lock.o drm_atomic.o drm_bridge.o \
-		drm_framebuffer.o drm_connector.o drm_blend.o \
-		drm_encoder.o drm_mode_object.o drm_property.o \
-		drm_plane.o drm_color_mgmt.o drm_print.o \
-		drm_dumb_buffers.o drm_mode_config.o drm_vblank.o \
-		drm_syncobj.o
+CFLAGS-$(CONFIG_DRM_USE_DYNAMIC_DEBUG)	+= -DDYNAMIC_DEBUG_MODULE
 
+# Unconditionally enable W=1 warnings locally
+# --- begin copy-paste W=1 warnings from scripts/Makefile.extrawarn
+subdir-ccflags-y += -Wextra -Wunused -Wno-unused-parameter
+subdir-ccflags-y += $(call cc-option, -Wrestrict)
+subdir-ccflags-y += -Wmissing-format-attribute
+subdir-ccflags-y += -Wold-style-definition
+subdir-ccflags-y += -Wmissing-include-dirs
+subdir-ccflags-y += $(call cc-option, -Wunused-but-set-variable)
+subdir-ccflags-y += $(call cc-option, -Wunused-const-variable)
+subdir-ccflags-y += $(call cc-option, -Wpacked-not-aligned)
+subdir-ccflags-y += $(call cc-option, -Wformat-overflow)
+# FIXME: fix -Wformat-truncation warnings and uncomment
+#subdir-ccflags-y += $(call cc-option, -Wformat-truncation)
+subdir-ccflags-y += $(call cc-option, -Wstringop-truncation)
+# The following turn off the warnings enabled by -Wextra
+ifeq ($(findstring 2, $(KBUILD_EXTRA_WARN)),)
+subdir-ccflags-y += -Wno-missing-field-initializers
+subdir-ccflags-y += -Wno-type-limits
+subdir-ccflags-y += -Wno-shift-negative-value
+endif
+ifeq ($(findstring 3, $(KBUILD_EXTRA_WARN)),)
+subdir-ccflags-y += -Wno-sign-compare
+endif
+# --- end copy-paste
+
+# Enable -Werror in CI and development
+subdir-ccflags-$(CONFIG_DRM_WERROR) += -Werror
+
+drm-y := \
+	drm_atomic.o \
+	drm_atomic_uapi.o \
+	drm_auth.o \
+	drm_blend.o \
+	drm_bridge.o \
+	drm_cache.o \
+	drm_color_mgmt.o \
+	drm_connector.o \
+	drm_crtc.o \
+	drm_displayid.o \
+	drm_drv.o \
+	drm_dumb_buffers.o \
+	drm_edid.o \
+	drm_eld.o \
+	drm_encoder.o \
+	drm_file.o \
+	drm_fourcc.o \
+	drm_framebuffer.o \
+	drm_gem.o \
+	drm_ioctl.o \
+	drm_lease.o \
+	drm_managed.o \
+	drm_mm.o \
+	drm_mode_config.o \
+	drm_mode_object.o \
+	drm_modes.o \
+	drm_modeset_lock.o \
+	drm_plane.o \
+	drm_prime.o \
+	drm_print.o \
+	drm_property.o \
+	drm_rect.o \
+	drm_syncobj.o \
+	drm_sysfs.o \
+	drm_trace_points.o \
+	drm_vblank.o \
+	drm_vblank_work.o \
+	drm_vma_manager.o \
+	drm_writeback.o
+drm-$(CONFIG_DRM_CLIENT) += \
+	drm_client.o \
+	drm_client_event.o \
+	drm_client_modeset.o
 drm-$(CONFIG_DRM_LIB_RANDOM) += lib/drm_random.o
-drm-$(CONFIG_DRM_VM) += drm_vm.o
 drm-$(CONFIG_COMPAT) += drm_ioc32.o
-drm-$(CONFIG_DRM_GEM_CMA_HELPER) += drm_gem_cma_helper.o
-drm-$(CONFIG_PCI) += ati_pcigart.o
 drm-$(CONFIG_DRM_PANEL) += drm_panel.o
 drm-$(CONFIG_OF) += drm_of.o
-drm-$(CONFIG_AGP) += drm_agpsupport.o
-drm-$(CONFIG_DEBUG_FS) += drm_debugfs.o drm_debugfs_crc.o
+drm-$(CONFIG_PCI) += drm_pci.o
+drm-$(CONFIG_DEBUG_FS) += \
+	drm_debugfs.o \
+	drm_debugfs_crc.o
+drm-$(CONFIG_DRM_LOAD_EDID_FIRMWARE) += drm_edid_load.o
+drm-$(CONFIG_DRM_PRIVACY_SCREEN) += \
+	drm_privacy_screen.o \
+	drm_privacy_screen_x86.o
+drm-$(CONFIG_DRM_ACCEL) += ../../accel/drm_accel.o
+drm-$(CONFIG_DRM_PANIC) += drm_panic.o
+drm-$(CONFIG_DRM_DRAW) += drm_draw.o
+drm-$(CONFIG_DRM_PANIC_SCREEN_QR_CODE) += drm_panic_qr.o
+obj-$(CONFIG_DRM)	+= drm.o
+
+obj-$(CONFIG_DRM_PANEL_ORIENTATION_QUIRKS) += drm_panel_orientation_quirks.o
+obj-$(CONFIG_DRM_PANEL_BACKLIGHT_QUIRKS) += drm_panel_backlight_quirks.o
+
+#
+# Memory-management helpers
+#
+#
+obj-$(CONFIG_DRM_EXEC) += drm_exec.o
+obj-$(CONFIG_DRM_GPUVM) += drm_gpuvm.o
+
+drm_gpusvm_helper-y := \
+	drm_gpusvm.o\
+	drm_pagemap.o
+obj-$(CONFIG_DRM_GPUSVM) += drm_gpusvm_helper.o
+
+obj-$(CONFIG_DRM_BUDDY) += drm_buddy.o
+
+drm_dma_helper-y := drm_gem_dma_helper.o
+drm_dma_helper-$(CONFIG_DRM_FBDEV_EMULATION) += drm_fbdev_dma.o
+drm_dma_helper-$(CONFIG_DRM_KMS_HELPER) += drm_fb_dma_helper.o
+obj-$(CONFIG_DRM_GEM_DMA_HELPER) += drm_dma_helper.o
+
+drm_shmem_helper-y := drm_gem_shmem_helper.o
+drm_shmem_helper-$(CONFIG_DRM_FBDEV_EMULATION) += drm_fbdev_shmem.o
+obj-$(CONFIG_DRM_GEM_SHMEM_HELPER) += drm_shmem_helper.o
+
+drm_suballoc_helper-y := drm_suballoc.o
+obj-$(CONFIG_DRM_SUBALLOC_HELPER) += drm_suballoc_helper.o
 
-drm_kms_helper-y := drm_crtc_helper.o drm_dp_helper.o drm_probe_helper.o \
-		drm_plane_helper.o drm_dp_mst_topology.o drm_atomic_helper.o \
-		drm_kms_helper_common.o drm_dp_dual_mode_helper.o \
-		drm_simple_kms_helper.o drm_modeset_helper.o \
-		drm_scdc_helper.o drm_gem_framebuffer_helper.o
+drm_vram_helper-y := drm_gem_vram_helper.o
+obj-$(CONFIG_DRM_VRAM_HELPER) += drm_vram_helper.o
 
+drm_ttm_helper-y := drm_gem_ttm_helper.o
+drm_ttm_helper-$(CONFIG_DRM_FBDEV_EMULATION) += drm_fbdev_ttm.o
+obj-$(CONFIG_DRM_TTM_HELPER) += drm_ttm_helper.o
+
+#
+# Modesetting helpers
+#
+
+drm_kms_helper-y := \
+	drm_atomic_helper.o \
+	drm_atomic_state_helper.o \
+	drm_bridge_helper.o \
+	drm_crtc_helper.o \
+	drm_damage_helper.o \
+	drm_flip_work.o \
+	drm_format_helper.o \
+	drm_gem_atomic_helper.o \
+	drm_gem_framebuffer_helper.o \
+	drm_kms_helper_common.o \
+	drm_modeset_helper.o \
+	drm_plane_helper.o \
+	drm_probe_helper.o \
+	drm_self_refresh_helper.o \
+	drm_simple_kms_helper.o
 drm_kms_helper-$(CONFIG_DRM_PANEL_BRIDGE) += bridge/panel.o
-drm_kms_helper-$(CONFIG_DRM_LOAD_EDID_FIRMWARE) += drm_edid_load.o
 drm_kms_helper-$(CONFIG_DRM_FBDEV_EMULATION) += drm_fb_helper.o
-drm_kms_helper-$(CONFIG_DRM_KMS_CMA_HELPER) += drm_fb_cma_helper.o
-drm_kms_helper-$(CONFIG_DRM_DP_AUX_CHARDEV) += drm_dp_aux_dev.o
-
 obj-$(CONFIG_DRM_KMS_HELPER) += drm_kms_helper.o
-obj-$(CONFIG_DRM_DEBUG_MM_SELFTEST) += selftests/
 
-CFLAGS_drm_trace_points.o := -I$(src)
+#
+# Drivers and the rest
+#
 
-obj-$(CONFIG_DRM)	+= drm.o
+obj-y			+= tests/
+
+obj-$(CONFIG_DRM_MIPI_DBI) += drm_mipi_dbi.o
 obj-$(CONFIG_DRM_MIPI_DSI) += drm_mipi_dsi.o
-obj-$(CONFIG_DRM_ARM)	+= arm/
+obj-y			+= arm/
+obj-y			+= clients/
+obj-y			+= display/
 obj-$(CONFIG_DRM_TTM)	+= ttm/
-obj-$(CONFIG_DRM_TDFX)	+= tdfx/
-obj-$(CONFIG_DRM_R128)	+= r128/
-obj-$(CONFIG_HSA_AMD) += amd/amdkfd/
+obj-$(CONFIG_DRM_SCHED)	+= scheduler/
 obj-$(CONFIG_DRM_RADEON)+= radeon/
 obj-$(CONFIG_DRM_AMDGPU)+= amd/amdgpu/
-obj-$(CONFIG_DRM_MGA)	+= mga/
-obj-$(CONFIG_DRM_I810)	+= i810/
+obj-$(CONFIG_DRM_AMDGPU)+= amd/amdxcp/
 obj-$(CONFIG_DRM_I915)	+= i915/
+obj-$(CONFIG_DRM_XE)	+= xe/
+obj-$(CONFIG_DRM_KMB_DISPLAY)  += kmb/
 obj-$(CONFIG_DRM_MGAG200) += mgag200/
+obj-$(CONFIG_DRM_V3D)  += v3d/
 obj-$(CONFIG_DRM_VC4)  += vc4/
-obj-$(CONFIG_DRM_CIRRUS_QEMU) += cirrus/
-obj-$(CONFIG_DRM_SIS)   += sis/
-obj-$(CONFIG_DRM_SAVAGE)+= savage/
 obj-$(CONFIG_DRM_VMWGFX)+= vmwgfx/
-obj-$(CONFIG_DRM_VIA)	+=via/
 obj-$(CONFIG_DRM_VGEM)	+= vgem/
+obj-$(CONFIG_DRM_VKMS)	+= vkms/
 obj-$(CONFIG_DRM_NOUVEAU) +=nouveau/
+obj-$(CONFIG_DRM_NOVA) += nova/
 obj-$(CONFIG_DRM_EXYNOS) +=exynos/
 obj-$(CONFIG_DRM_ROCKCHIP) +=rockchip/
 obj-$(CONFIG_DRM_GMA500) += gma500/
@@ -74,29 +189,64 @@ obj-$(CONFIG_DRM_UDL) += udl/
 obj-$(CONFIG_DRM_AST) += ast/
 obj-$(CONFIG_DRM_ARMADA) += armada/
 obj-$(CONFIG_DRM_ATMEL_HLCDC)	+= atmel-hlcdc/
-obj-$(CONFIG_DRM_RCAR_DU) += rcar-du/
-obj-$(CONFIG_DRM_SHMOBILE) +=shmobile/
+obj-y			+= renesas/
 obj-y			+= omapdrm/
 obj-$(CONFIG_DRM_SUN4I) += sun4i/
 obj-y			+= tilcdc/
 obj-$(CONFIG_DRM_QXL) += qxl/
-obj-$(CONFIG_DRM_BOCHS) += bochs/
 obj-$(CONFIG_DRM_VIRTIO_GPU) += virtio/
 obj-$(CONFIG_DRM_MSM) += msm/
 obj-$(CONFIG_DRM_TEGRA) += tegra/
 obj-$(CONFIG_DRM_STM) += stm/
 obj-$(CONFIG_DRM_STI) += sti/
-obj-$(CONFIG_DRM_IMX) += imx/
+obj-y 			+= imx/
+obj-$(CONFIG_DRM_INGENIC) += ingenic/
+obj-$(CONFIG_DRM_LOGICVC) += logicvc/
 obj-$(CONFIG_DRM_MEDIATEK) += mediatek/
 obj-$(CONFIG_DRM_MESON)	+= meson/
-obj-y			+= i2c/
 obj-y			+= panel/
 obj-y			+= bridge/
 obj-$(CONFIG_DRM_FSL_DCU) += fsl-dcu/
 obj-$(CONFIG_DRM_ETNAVIV) += etnaviv/
-obj-$(CONFIG_DRM_ARCPGU)+= arc/
 obj-y			+= hisilicon/
-obj-$(CONFIG_DRM_ZTE)	+= zte/
-obj-$(CONFIG_DRM_MXSFB)	+= mxsfb/
-obj-$(CONFIG_DRM_TINYDRM) += tinydrm/
+obj-y			+= mxsfb/
+obj-y			+= sysfb/
+obj-y			+= tiny/
 obj-$(CONFIG_DRM_PL111) += pl111/
+obj-$(CONFIG_DRM_TVE200) += tve200/
+obj-$(CONFIG_DRM_ADP) += adp/
+obj-$(CONFIG_DRM_XEN) += xen/
+obj-$(CONFIG_DRM_VBOXVIDEO) += vboxvideo/
+obj-$(CONFIG_DRM_LIMA)  += lima/
+obj-$(CONFIG_DRM_PANFROST) += panfrost/
+obj-$(CONFIG_DRM_PANTHOR) += panthor/
+obj-$(CONFIG_DRM_TYR) += tyr/
+obj-$(CONFIG_DRM_ASPEED_GFX) += aspeed/
+obj-$(CONFIG_DRM_MCDE) += mcde/
+obj-$(CONFIG_DRM_TIDSS) += tidss/
+obj-y			+= xlnx/
+obj-y			+= gud/
+obj-$(CONFIG_DRM_HYPERV) += hyperv/
+obj-y			+= sitronix/
+obj-y			+= solomon/
+obj-$(CONFIG_DRM_SPRD) += sprd/
+obj-$(CONFIG_DRM_LOONGSON) += loongson/
+obj-$(CONFIG_DRM_POWERVR) += imagination/
+
+# Ensure drm headers are self-contained and pass kernel-doc
+hdrtest-files := \
+	$(shell cd $(src) && find . -maxdepth 1 -name 'drm_*.h') \
+	$(shell cd $(src) && find display lib -name '*.h')
+
+always-$(CONFIG_DRM_HEADER_TEST) += \
+	$(patsubst %.h,%.hdrtest, $(hdrtest-files))
+
+# Include the header twice to detect missing include guard.
+quiet_cmd_hdrtest = HDRTEST $(patsubst %.hdrtest,%.h,$@)
+      cmd_hdrtest = \
+		$(CC) $(c_flags) -fsyntax-only -x c /dev/null -include $< -include $<; \
+		 PYTHONDONTWRITEBYTECODE=1 $(KERNELDOC) -none $(if $(CONFIG_WERROR)$(CONFIG_DRM_WERROR),-Werror) $<; \
+		touch $@
+
+$(obj)/%.hdrtest: $(src)/%.h FORCE
+	$(call if_changed_dep,hdrtest)
diff --git a/drivers/gpu/drm/adp/Kconfig b/drivers/gpu/drm/adp/Kconfig
new file mode 100644
index 000000000000..9fcc27eb200d
--- /dev/null
+++ b/drivers/gpu/drm/adp/Kconfig
@@ -0,0 +1,17 @@
+# SPDX-License-Identifier: GPL-2.0-only OR MIT
+config DRM_ADP
+	tristate "DRM Support for pre-DCP Apple display controllers"
+	depends on DRM && OF && ARM64
+	depends on ARCH_APPLE || COMPILE_TEST
+	select DRM_KMS_HELPER
+	select DRM_BRIDGE_CONNECTOR
+	select DRM_DISPLAY_HELPER
+	select DRM_KMS_DMA_HELPER
+	select DRM_GEM_DMA_HELPER
+	select DRM_PANEL_BRIDGE
+	select VIDEOMODE_HELPERS
+	select DRM_MIPI_DSI
+	help
+	  Chose this option if you have an Apple Arm laptop with a touchbar.
+
+	  If M is selected, this module will be called adpdrm.
diff --git a/drivers/gpu/drm/adp/Makefile b/drivers/gpu/drm/adp/Makefile
new file mode 100644
index 000000000000..8e7b618edd35
--- /dev/null
+++ b/drivers/gpu/drm/adp/Makefile
@@ -0,0 +1,5 @@
+# SPDX-License-Identifier: GPL-2.0-only OR MIT
+
+adpdrm-y := adp_drv.o
+adpdrm-mipi-y := adp-mipi.o
+obj-$(CONFIG_DRM_ADP) += adpdrm.o adpdrm-mipi.o
diff --git a/drivers/gpu/drm/adp/adp-mipi.c b/drivers/gpu/drm/adp/adp-mipi.c
new file mode 100644
index 000000000000..cba7d32150a9
--- /dev/null
+++ b/drivers/gpu/drm/adp/adp-mipi.c
@@ -0,0 +1,277 @@
+// SPDX-License-Identifier: GPL-2.0-only
+
+#include <linux/component.h>
+#include <linux/iopoll.h>
+#include <linux/of.h>
+#include <linux/platform_device.h>
+
+#include <drm/drm_bridge.h>
+#include <drm/drm_mipi_dsi.h>
+
+#define DSI_GEN_HDR 0x6c
+#define DSI_GEN_PLD_DATA 0x70
+
+#define DSI_CMD_PKT_STATUS 0x74
+
+#define GEN_PLD_R_EMPTY BIT(4)
+#define GEN_PLD_W_FULL BIT(3)
+#define GEN_PLD_W_EMPTY BIT(2)
+#define GEN_CMD_FULL BIT(1)
+#define GEN_CMD_EMPTY BIT(0)
+#define GEN_RD_CMD_BUSY BIT(6)
+#define CMD_PKT_STATUS_TIMEOUT_US 20000
+
+struct adp_mipi_drv_private {
+	struct mipi_dsi_host dsi;
+	struct drm_bridge bridge;
+	struct drm_bridge *next_bridge;
+	void __iomem *mipi;
+};
+
+#define mipi_to_adp(x) container_of(x, struct adp_mipi_drv_private, dsi)
+
+static int adp_dsi_gen_pkt_hdr_write(struct adp_mipi_drv_private *adp, u32 hdr_val)
+{
+	int ret;
+	u32 val, mask;
+
+	ret = readl_poll_timeout(adp->mipi + DSI_CMD_PKT_STATUS,
+				 val, !(val & GEN_CMD_FULL), 1000,
+				 CMD_PKT_STATUS_TIMEOUT_US);
+	if (ret) {
+		dev_err(adp->dsi.dev, "failed to get available command FIFO\n");
+		return ret;
+	}
+
+	writel(hdr_val, adp->mipi + DSI_GEN_HDR);
+
+	mask = GEN_CMD_EMPTY | GEN_PLD_W_EMPTY;
+	ret = readl_poll_timeout(adp->mipi + DSI_CMD_PKT_STATUS,
+				 val, (val & mask) == mask,
+				 1000, CMD_PKT_STATUS_TIMEOUT_US);
+	if (ret) {
+		dev_err(adp->dsi.dev, "failed to write command FIFO\n");
+		return ret;
+	}
+
+	return 0;
+}
+
+static int adp_dsi_write(struct adp_mipi_drv_private *adp,
+			 const struct mipi_dsi_packet *packet)
+{
+	const u8 *tx_buf = packet->payload;
+	int len = packet->payload_length, pld_data_bytes = sizeof(u32), ret;
+	__le32 word;
+	u32 val;
+
+	while (len) {
+		if (len < pld_data_bytes) {
+			word = 0;
+			memcpy(&word, tx_buf, len);
+			writel(le32_to_cpu(word), adp->mipi + DSI_GEN_PLD_DATA);
+			len = 0;
+		} else {
+			memcpy(&word, tx_buf, pld_data_bytes);
+			writel(le32_to_cpu(word), adp->mipi + DSI_GEN_PLD_DATA);
+			tx_buf += pld_data_bytes;
+			len -= pld_data_bytes;
+		}
+
+		ret = readl_poll_timeout(adp->mipi + DSI_CMD_PKT_STATUS,
+					 val, !(val & GEN_PLD_W_FULL), 1000,
+					 CMD_PKT_STATUS_TIMEOUT_US);
+		if (ret) {
+			dev_err(adp->dsi.dev,
+				"failed to get available write payload FIFO\n");
+			return ret;
+		}
+	}
+
+	word = 0;
+	memcpy(&word, packet->header, sizeof(packet->header));
+	return adp_dsi_gen_pkt_hdr_write(adp, le32_to_cpu(word));
+}
+
+static int adp_dsi_read(struct adp_mipi_drv_private *adp,
+			const struct mipi_dsi_msg *msg)
+{
+	int i, j, ret, len = msg->rx_len;
+	u8 *buf = msg->rx_buf;
+	u32 val;
+
+	/* Wait end of the read operation */
+	ret = readl_poll_timeout(adp->mipi + DSI_CMD_PKT_STATUS,
+				 val, !(val & GEN_RD_CMD_BUSY),
+				 1000, CMD_PKT_STATUS_TIMEOUT_US);
+	if (ret) {
+		dev_err(adp->dsi.dev, "Timeout during read operation\n");
+		return ret;
+	}
+
+	for (i = 0; i < len; i += 4) {
+		/* Read fifo must not be empty before all bytes are read */
+		ret = readl_poll_timeout(adp->mipi + DSI_CMD_PKT_STATUS,
+					 val, !(val & GEN_PLD_R_EMPTY),
+					 1000, CMD_PKT_STATUS_TIMEOUT_US);
+		if (ret) {
+			dev_err(adp->dsi.dev, "Read payload FIFO is empty\n");
+			return ret;
+		}
+
+		val = readl(adp->mipi + DSI_GEN_PLD_DATA);
+		for (j = 0; j < 4 && j + i < len; j++)
+			buf[i + j] = val >> (8 * j);
+	}
+
+	return ret;
+}
+
+static ssize_t adp_dsi_host_transfer(struct mipi_dsi_host *host,
+				     const struct mipi_dsi_msg *msg)
+{
+	struct adp_mipi_drv_private *adp = mipi_to_adp(host);
+	struct mipi_dsi_packet packet;
+	int ret, nb_bytes;
+
+	ret = mipi_dsi_create_packet(&packet, msg);
+	if (ret) {
+		dev_err(adp->dsi.dev, "failed to create packet: %d\n", ret);
+		return ret;
+	}
+
+	ret = adp_dsi_write(adp, &packet);
+	if (ret)
+		return ret;
+
+	if (msg->rx_buf && msg->rx_len) {
+		ret = adp_dsi_read(adp, msg);
+		if (ret)
+			return ret;
+		nb_bytes = msg->rx_len;
+	} else {
+		nb_bytes = packet.size;
+	}
+
+	return nb_bytes;
+}
+
+static int adp_dsi_bind(struct device *dev, struct device *master, void *data)
+{
+	return 0;
+}
+
+static void adp_dsi_unbind(struct device *dev, struct device *master, void *data)
+{
+}
+
+static const struct component_ops adp_dsi_component_ops = {
+	.bind	= adp_dsi_bind,
+	.unbind	= adp_dsi_unbind,
+};
+
+static int adp_dsi_host_attach(struct mipi_dsi_host *host,
+			       struct mipi_dsi_device *dev)
+{
+	struct adp_mipi_drv_private *adp = mipi_to_adp(host);
+	struct drm_bridge *next;
+	int ret;
+
+	next = devm_drm_of_get_bridge(adp->dsi.dev, adp->dsi.dev->of_node, 1, 0);
+	if (IS_ERR(next))
+		return PTR_ERR(next);
+
+	adp->next_bridge = next;
+
+	drm_bridge_add(&adp->bridge);
+
+	ret = component_add(host->dev, &adp_dsi_component_ops);
+	if (ret) {
+		pr_err("failed to add dsi_host component: %d\n", ret);
+		drm_bridge_remove(&adp->bridge);
+		return ret;
+	}
+
+	return 0;
+}
+
+static int adp_dsi_host_detach(struct mipi_dsi_host *host,
+			       struct mipi_dsi_device *dev)
+{
+	struct adp_mipi_drv_private *adp = mipi_to_adp(host);
+
+	component_del(host->dev, &adp_dsi_component_ops);
+	drm_bridge_remove(&adp->bridge);
+	return 0;
+}
+
+static const struct mipi_dsi_host_ops adp_dsi_host_ops = {
+	.transfer = adp_dsi_host_transfer,
+	.attach = adp_dsi_host_attach,
+	.detach = adp_dsi_host_detach,
+};
+
+static int adp_dsi_bridge_attach(struct drm_bridge *bridge,
+				 struct drm_encoder *encoder,
+				 enum drm_bridge_attach_flags flags)
+{
+	struct adp_mipi_drv_private *adp =
+		container_of(bridge, struct adp_mipi_drv_private, bridge);
+
+	return drm_bridge_attach(encoder, adp->next_bridge, bridge, flags);
+}
+
+static const struct drm_bridge_funcs adp_dsi_bridge_funcs = {
+	.attach	= adp_dsi_bridge_attach,
+};
+
+static int adp_mipi_probe(struct platform_device *pdev)
+{
+	struct adp_mipi_drv_private *adp;
+
+	adp = devm_drm_bridge_alloc(&pdev->dev, struct adp_mipi_drv_private,
+				    bridge, &adp_dsi_bridge_funcs);
+	if (IS_ERR(adp))
+		return PTR_ERR(adp);
+
+	adp->mipi = devm_platform_ioremap_resource(pdev, 0);
+	if (IS_ERR(adp->mipi)) {
+		dev_err(&pdev->dev, "failed to map mipi mmio");
+		return PTR_ERR(adp->mipi);
+	}
+
+	adp->dsi.dev = &pdev->dev;
+	adp->dsi.ops = &adp_dsi_host_ops;
+	adp->bridge.of_node = pdev->dev.of_node;
+	adp->bridge.type = DRM_MODE_CONNECTOR_DSI;
+	dev_set_drvdata(&pdev->dev, adp);
+	return mipi_dsi_host_register(&adp->dsi);
+}
+
+static void adp_mipi_remove(struct platform_device *pdev)
+{
+	struct device *dev = &pdev->dev;
+	struct adp_mipi_drv_private *adp = dev_get_drvdata(dev);
+
+	mipi_dsi_host_unregister(&adp->dsi);
+}
+
+static const struct of_device_id adp_mipi_of_match[] = {
+	{ .compatible = "apple,h7-display-pipe-mipi", },
+	{ },
+};
+MODULE_DEVICE_TABLE(of, adp_mipi_of_match);
+
+static struct platform_driver adp_mipi_platform_driver = {
+	.driver = {
+		.name = "adp-mipi",
+		.of_match_table = adp_mipi_of_match,
+	},
+	.probe = adp_mipi_probe,
+	.remove = adp_mipi_remove,
+};
+
+module_platform_driver(adp_mipi_platform_driver);
+
+MODULE_DESCRIPTION("Apple Display Pipe MIPI driver");
+MODULE_LICENSE("GPL");
diff --git a/drivers/gpu/drm/adp/adp_drv.c b/drivers/gpu/drm/adp/adp_drv.c
new file mode 100644
index 000000000000..54cde090c3f4
--- /dev/null
+++ b/drivers/gpu/drm/adp/adp_drv.c
@@ -0,0 +1,613 @@
+// SPDX-License-Identifier: GPL-2.0-only
+
+#include <linux/component.h>
+#include <linux/iopoll.h>
+#include <linux/of.h>
+#include <linux/platform_device.h>
+
+#include <drm/drm_atomic.h>
+#include <drm/drm_atomic_helper.h>
+#include <drm/drm_bridge.h>
+#include <drm/drm_bridge_connector.h>
+#include <drm/drm_drv.h>
+#include <drm/drm_fb_dma_helper.h>
+#include <drm/drm_framebuffer.h>
+#include <drm/drm_gem_atomic_helper.h>
+#include <drm/drm_gem_dma_helper.h>
+#include <drm/drm_gem_framebuffer_helper.h>
+#include <drm/drm_of.h>
+#include <drm/drm_probe_helper.h>
+#include <drm/drm_vblank.h>
+
+#define ADP_INT_STATUS 0x34
+#define ADP_INT_STATUS_INT_MASK 0x7
+#define ADP_INT_STATUS_VBLANK 0x1
+#define ADP_CTRL 0x100
+#define ADP_CTRL_VBLANK_ON 0x12
+#define ADP_CTRL_FIFO_ON 0x601
+#define ADP_SCREEN_SIZE 0x0c
+#define ADP_SCREEN_HSIZE GENMASK(15, 0)
+#define ADP_SCREEN_VSIZE GENMASK(31, 16)
+
+#define ADBE_FIFO 0x10c0
+#define ADBE_FIFO_SYNC 0xc0000000
+
+#define ADBE_BLEND_BYPASS 0x2020
+#define ADBE_BLEND_EN1 0x2028
+#define ADBE_BLEND_EN2 0x2074
+#define ADBE_BLEND_EN3 0x202c
+#define ADBE_BLEND_EN4 0x2034
+#define ADBE_MASK_BUF 0x2200
+
+#define ADBE_SRC_START 0x4040
+#define ADBE_SRC_SIZE 0x4048
+#define ADBE_DST_START 0x4050
+#define ADBE_DST_SIZE 0x4054
+#define ADBE_STRIDE 0x4038
+#define ADBE_FB_BASE 0x4030
+
+#define ADBE_LAYER_EN1 0x4020
+#define ADBE_LAYER_EN2 0x4068
+#define ADBE_LAYER_EN3 0x40b4
+#define ADBE_LAYER_EN4 0x40f4
+#define ADBE_SCALE_CTL 0x40ac
+#define ADBE_SCALE_CTL_BYPASS 0x100000
+
+#define ADBE_LAYER_CTL 0x1038
+#define ADBE_LAYER_CTL_ENABLE 0x10000
+
+#define ADBE_PIX_FMT 0x402c
+#define ADBE_PIX_FMT_XRGB32 0x53e4001
+
+static int adp_open(struct inode *inode, struct file *filp)
+{
+	/*
+	 * The modesetting driver does not check the non-desktop connector
+	 * property and keeps the device open and locked. If the touchbar daemon
+	 * opens the device first, modesetting breaks the whole X session.
+	 * Simply refuse to open the device for X11 server processes as
+	 * workaround.
+	 */
+	if (current->comm[0] == 'X')
+		return -EBUSY;
+
+	return drm_open(inode, filp);
+}
+
+static const struct file_operations adp_fops = {
+	.owner          = THIS_MODULE,
+	.open           = adp_open,
+	.release        = drm_release,
+	.unlocked_ioctl = drm_ioctl,
+	.compat_ioctl   = drm_compat_ioctl,
+	.poll           = drm_poll,
+	.read           = drm_read,
+	.llseek         = noop_llseek,
+	.mmap           = drm_gem_mmap,
+	.fop_flags      = FOP_UNSIGNED_OFFSET,
+	DRM_GEM_DMA_UNMAPPED_AREA_FOPS
+};
+
+static int adp_drm_gem_dumb_create(struct drm_file *file_priv,
+					struct drm_device *drm,
+					struct drm_mode_create_dumb *args)
+{
+	args->height = ALIGN(args->height, 64);
+	args->size = args->pitch * args->height;
+
+	return drm_gem_dma_dumb_create_internal(file_priv, drm, args);
+}
+
+static const struct drm_driver adp_driver = {
+	.driver_features = DRIVER_GEM | DRIVER_MODESET | DRIVER_ATOMIC,
+	.fops = &adp_fops,
+	DRM_GEM_DMA_DRIVER_OPS_VMAP_WITH_DUMB_CREATE(adp_drm_gem_dumb_create),
+	.name = "adp",
+	.desc = "Apple Display Pipe DRM Driver",
+	.major = 0,
+	.minor = 1,
+};
+
+struct adp_drv_private {
+	struct drm_device drm;
+	struct drm_crtc crtc;
+	struct drm_encoder *encoder;
+	struct drm_connector *connector;
+	struct drm_bridge *next_bridge;
+	void __iomem *be;
+	void __iomem *fe;
+	u32 *mask_buf;
+	u64 mask_buf_size;
+	dma_addr_t mask_iova;
+	int be_irq;
+	int fe_irq;
+	struct drm_pending_vblank_event *event;
+};
+
+#define to_adp(x) container_of(x, struct adp_drv_private, drm)
+#define crtc_to_adp(x) container_of(x, struct adp_drv_private, crtc)
+
+static int adp_plane_atomic_check(struct drm_plane *plane,
+				    struct drm_atomic_state *state)
+{
+	struct drm_plane_state *new_plane_state;
+	struct drm_crtc_state *crtc_state;
+
+	new_plane_state = drm_atomic_get_new_plane_state(state, plane);
+
+	if (!new_plane_state->crtc)
+		return 0;
+
+	crtc_state = drm_atomic_get_crtc_state(state, new_plane_state->crtc);
+	if (IS_ERR(crtc_state))
+		return PTR_ERR(crtc_state);
+
+	return drm_atomic_helper_check_plane_state(new_plane_state,
+						   crtc_state,
+						   DRM_PLANE_NO_SCALING,
+						   DRM_PLANE_NO_SCALING,
+						   true, true);
+}
+
+static void adp_plane_atomic_update(struct drm_plane *plane,
+				    struct drm_atomic_state *state)
+{
+	struct adp_drv_private *adp;
+	struct drm_rect src_rect;
+	struct drm_gem_dma_object *obj;
+	struct drm_framebuffer *fb;
+	struct drm_plane_state *new_state = drm_atomic_get_new_plane_state(state, plane);
+	u32 src_pos, src_size, dst_pos, dst_size;
+
+	if (!plane || !new_state)
+		return;
+
+	fb = new_state->fb;
+	if (!fb)
+		return;
+	adp = to_adp(plane->dev);
+
+	drm_rect_fp_to_int(&src_rect, &new_state->src);
+	src_pos = src_rect.x1 << 16 | src_rect.y1;
+	dst_pos = new_state->dst.x1 << 16 | new_state->dst.y1;
+	src_size = drm_rect_width(&src_rect) << 16 | drm_rect_height(&src_rect);
+	dst_size = drm_rect_width(&new_state->dst) << 16 |
+		drm_rect_height(&new_state->dst);
+	writel(src_pos, adp->be + ADBE_SRC_START);
+	writel(src_size, adp->be + ADBE_SRC_SIZE);
+	writel(dst_pos, adp->be + ADBE_DST_START);
+	writel(dst_size, adp->be + ADBE_DST_SIZE);
+	writel(fb->pitches[0], adp->be + ADBE_STRIDE);
+	obj = drm_fb_dma_get_gem_obj(fb, 0);
+	if (obj)
+		writel(obj->dma_addr + fb->offsets[0], adp->be + ADBE_FB_BASE);
+
+	writel(BIT(0), adp->be + ADBE_LAYER_EN1);
+	writel(BIT(0), adp->be + ADBE_LAYER_EN2);
+	writel(BIT(0), adp->be + ADBE_LAYER_EN3);
+	writel(BIT(0), adp->be + ADBE_LAYER_EN4);
+	writel(ADBE_SCALE_CTL_BYPASS, adp->be + ADBE_SCALE_CTL);
+	writel(ADBE_LAYER_CTL_ENABLE | BIT(0), adp->be + ADBE_LAYER_CTL);
+	writel(ADBE_PIX_FMT_XRGB32, adp->be + ADBE_PIX_FMT);
+}
+
+static void adp_plane_atomic_disable(struct drm_plane *plane,
+				     struct drm_atomic_state *state)
+{
+	struct adp_drv_private *adp = to_adp(plane->dev);
+
+	writel(0x0, adp->be + ADBE_LAYER_EN1);
+	writel(0x0, adp->be + ADBE_LAYER_EN2);
+	writel(0x0, adp->be + ADBE_LAYER_EN3);
+	writel(0x0, adp->be + ADBE_LAYER_EN4);
+	writel(ADBE_LAYER_CTL_ENABLE, adp->be + ADBE_LAYER_CTL);
+}
+
+static const struct drm_plane_helper_funcs adp_plane_helper_funcs = {
+	.atomic_check = adp_plane_atomic_check,
+	.atomic_update = adp_plane_atomic_update,
+	.atomic_disable = adp_plane_atomic_disable,
+	DRM_GEM_SHADOW_PLANE_HELPER_FUNCS
+};
+
+static const struct drm_plane_funcs adp_plane_funcs = {
+	.update_plane = drm_atomic_helper_update_plane,
+	.disable_plane = drm_atomic_helper_disable_plane,
+	DRM_GEM_SHADOW_PLANE_FUNCS
+};
+
+static const u32 plane_formats[] = {
+	DRM_FORMAT_XRGB8888,
+};
+
+#define ALL_CRTCS 1
+
+static struct drm_plane *adp_plane_new(struct adp_drv_private *adp)
+{
+	struct drm_device *drm = &adp->drm;
+	struct drm_plane *plane;
+
+	plane = __drmm_universal_plane_alloc(drm, sizeof(struct drm_plane), 0,
+					     ALL_CRTCS, &adp_plane_funcs,
+					     plane_formats, ARRAY_SIZE(plane_formats),
+					     NULL, DRM_PLANE_TYPE_PRIMARY, "plane");
+	if (IS_ERR(plane)) {
+		drm_err(drm, "failed to allocate plane");
+		return plane;
+	}
+
+	drm_plane_helper_add(plane, &adp_plane_helper_funcs);
+	return plane;
+}
+
+static void adp_enable_vblank(struct adp_drv_private *adp)
+{
+	u32 cur_ctrl;
+
+	writel(ADP_INT_STATUS_INT_MASK, adp->fe + ADP_INT_STATUS);
+
+	cur_ctrl = readl(adp->fe + ADP_CTRL);
+	writel(cur_ctrl | ADP_CTRL_VBLANK_ON, adp->fe + ADP_CTRL);
+}
+
+static int adp_crtc_enable_vblank(struct drm_crtc *crtc)
+{
+	struct drm_device *dev = crtc->dev;
+	struct adp_drv_private *adp = to_adp(dev);
+
+	adp_enable_vblank(adp);
+
+	return 0;
+}
+
+static void adp_disable_vblank(struct adp_drv_private *adp)
+{
+	u32 cur_ctrl;
+
+	cur_ctrl = readl(adp->fe + ADP_CTRL);
+	writel(cur_ctrl & ~ADP_CTRL_VBLANK_ON, adp->fe + ADP_CTRL);
+	writel(ADP_INT_STATUS_INT_MASK, adp->fe + ADP_INT_STATUS);
+}
+
+static void adp_crtc_disable_vblank(struct drm_crtc *crtc)
+{
+	struct drm_device *dev = crtc->dev;
+	struct adp_drv_private *adp = to_adp(dev);
+
+	adp_disable_vblank(adp);
+}
+
+static void adp_crtc_atomic_enable(struct drm_crtc *crtc,
+				   struct drm_atomic_state *state)
+{
+	struct adp_drv_private *adp = crtc_to_adp(crtc);
+
+	writel(BIT(0), adp->be + ADBE_BLEND_EN2);
+	writel(BIT(4), adp->be + ADBE_BLEND_EN1);
+	writel(BIT(0), adp->be + ADBE_BLEND_EN3);
+	writel(BIT(0), adp->be + ADBE_BLEND_BYPASS);
+	writel(BIT(0), adp->be + ADBE_BLEND_EN4);
+	drm_crtc_vblank_on(crtc);
+}
+
+static void adp_crtc_atomic_disable(struct drm_crtc *crtc,
+				    struct drm_atomic_state *state)
+{
+	struct adp_drv_private *adp = crtc_to_adp(crtc);
+	struct drm_crtc_state *old_state = drm_atomic_get_old_crtc_state(state, crtc);
+
+	drm_atomic_helper_disable_planes_on_crtc(old_state, false);
+
+	writel(0x0, adp->be + ADBE_BLEND_EN2);
+	writel(0x0, adp->be + ADBE_BLEND_EN1);
+	writel(0x0, adp->be + ADBE_BLEND_EN3);
+	writel(0x0, adp->be + ADBE_BLEND_BYPASS);
+	writel(0x0, adp->be + ADBE_BLEND_EN4);
+	drm_crtc_vblank_off(crtc);
+}
+
+static void adp_crtc_atomic_flush(struct drm_crtc *crtc,
+				  struct drm_atomic_state *state)
+{
+	u32 frame_num = 1;
+	unsigned long flags;
+	struct adp_drv_private *adp = crtc_to_adp(crtc);
+	struct drm_crtc_state *new_state = drm_atomic_get_new_crtc_state(state, crtc);
+	u64 new_size = ALIGN(new_state->mode.hdisplay *
+			     new_state->mode.vdisplay * 4, PAGE_SIZE);
+
+	if (new_size != adp->mask_buf_size) {
+		if (adp->mask_buf)
+			dma_free_coherent(crtc->dev->dev, adp->mask_buf_size,
+					  adp->mask_buf, adp->mask_iova);
+		adp->mask_buf = NULL;
+		if (new_size != 0) {
+			adp->mask_buf = dma_alloc_coherent(crtc->dev->dev, new_size,
+							   &adp->mask_iova, GFP_KERNEL);
+			memset(adp->mask_buf, 0xFF, new_size);
+			writel(adp->mask_iova, adp->be + ADBE_MASK_BUF);
+		}
+		adp->mask_buf_size = new_size;
+	}
+	writel(ADBE_FIFO_SYNC | frame_num, adp->be + ADBE_FIFO);
+	//FIXME: use adbe flush interrupt
+	if (crtc->state->event) {
+		struct drm_pending_vblank_event *event = crtc->state->event;
+
+		crtc->state->event = NULL;
+		spin_lock_irqsave(&crtc->dev->event_lock, flags);
+
+		if (drm_crtc_vblank_get(crtc) != 0)
+			drm_crtc_send_vblank_event(crtc, event);
+		else
+			adp->event = event;
+
+		spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
+	}
+}
+
+static const struct drm_crtc_funcs adp_crtc_funcs = {
+	.destroy = drm_crtc_cleanup,
+	.set_config = drm_atomic_helper_set_config,
+	.page_flip = drm_atomic_helper_page_flip,
+	.reset = drm_atomic_helper_crtc_reset,
+	.atomic_duplicate_state = drm_atomic_helper_crtc_duplicate_state,
+	.atomic_destroy_state = drm_atomic_helper_crtc_destroy_state,
+	.enable_vblank = adp_crtc_enable_vblank,
+	.disable_vblank = adp_crtc_disable_vblank,
+};
+
+
+static const struct drm_crtc_helper_funcs adp_crtc_helper_funcs = {
+	.atomic_enable = adp_crtc_atomic_enable,
+	.atomic_disable = adp_crtc_atomic_disable,
+	.atomic_flush = adp_crtc_atomic_flush,
+};
+
+static int adp_setup_crtc(struct adp_drv_private *adp)
+{
+	struct drm_device *drm = &adp->drm;
+	struct drm_plane *primary;
+	int ret;
+
+	primary = adp_plane_new(adp);
+	if (IS_ERR(primary))
+		return PTR_ERR(primary);
+
+	ret = drm_crtc_init_with_planes(drm, &adp->crtc, primary,
+					NULL, &adp_crtc_funcs, NULL);
+	if (ret)
+		return ret;
+
+	drm_crtc_helper_add(&adp->crtc, &adp_crtc_helper_funcs);
+	return 0;
+}
+
+static const struct drm_mode_config_funcs adp_mode_config_funcs = {
+	.fb_create = drm_gem_fb_create_with_dirty,
+	.atomic_check = drm_atomic_helper_check,
+	.atomic_commit = drm_atomic_helper_commit,
+};
+
+static int adp_setup_mode_config(struct adp_drv_private *adp)
+{
+	struct drm_device *drm = &adp->drm;
+	int ret;
+	u32 size;
+
+	ret = drmm_mode_config_init(drm);
+	if (ret)
+		return ret;
+
+	/*
+	 * Query screen size restrict the frame buffer size to the screen size
+	 * aligned to the next multiple of 64. This is not necessary but can be
+	 * used as simple check for non-desktop devices.
+	 * Xorg's modesetting driver does not care about the connector
+	 * "non-desktop" property. The max frame buffer width or height can be
+	 * easily checked and a device can be reject if the max width/height is
+	 * smaller than 120 for example.
+	 * Any touchbar daemon is not limited by this small framebuffer size.
+	 */
+	size = readl(adp->fe + ADP_SCREEN_SIZE);
+
+	drm->mode_config.min_width = 32;
+	drm->mode_config.min_height = 32;
+	drm->mode_config.max_width = ALIGN(FIELD_GET(ADP_SCREEN_HSIZE, size), 64);
+	drm->mode_config.max_height = ALIGN(FIELD_GET(ADP_SCREEN_VSIZE, size), 64);
+	drm->mode_config.preferred_depth = 24;
+	drm->mode_config.prefer_shadow = 0;
+	drm->mode_config.funcs = &adp_mode_config_funcs;
+
+	ret = adp_setup_crtc(adp);
+	if (ret) {
+		drm_err(drm, "failed to create crtc");
+		return ret;
+	}
+
+	adp->encoder = drmm_plain_encoder_alloc(drm, NULL, DRM_MODE_ENCODER_DSI, NULL);
+	if (IS_ERR(adp->encoder)) {
+		drm_err(drm, "failed to init encoder");
+		return PTR_ERR(adp->encoder);
+	}
+	adp->encoder->possible_crtcs = ALL_CRTCS;
+
+	ret = drm_bridge_attach(adp->encoder, adp->next_bridge, NULL,
+				DRM_BRIDGE_ATTACH_NO_CONNECTOR);
+	if (ret) {
+		drm_err(drm, "failed to init bridge chain");
+		return ret;
+	}
+
+	adp->connector = drm_bridge_connector_init(drm, adp->encoder);
+	if (IS_ERR(adp->connector))
+		return PTR_ERR(adp->connector);
+
+	drm_connector_attach_encoder(adp->connector, adp->encoder);
+
+	ret = drm_vblank_init(drm, drm->mode_config.num_crtc);
+	if (ret < 0) {
+		drm_err(drm, "failed to initialize vblank");
+		return ret;
+	}
+
+	drm_mode_config_reset(drm);
+
+	return 0;
+}
+
+static int adp_parse_of(struct platform_device *pdev, struct adp_drv_private *adp)
+{
+	struct device *dev = &pdev->dev;
+
+	adp->be = devm_platform_ioremap_resource_byname(pdev, "be");
+	if (IS_ERR(adp->be)) {
+		dev_err(dev, "failed to map display backend mmio");
+		return PTR_ERR(adp->be);
+	}
+
+	adp->fe = devm_platform_ioremap_resource_byname(pdev, "fe");
+	if (IS_ERR(adp->fe)) {
+		dev_err(dev, "failed to map display pipe mmio");
+		return PTR_ERR(adp->fe);
+	}
+
+	adp->be_irq = platform_get_irq_byname(pdev, "be");
+	if (adp->be_irq < 0)
+		return adp->be_irq;
+
+	adp->fe_irq = platform_get_irq_byname(pdev, "fe");
+	if (adp->fe_irq < 0)
+		return adp->fe_irq;
+
+	return 0;
+}
+
+static irqreturn_t adp_fe_irq(int irq, void *arg)
+{
+	struct adp_drv_private *adp = (struct adp_drv_private *)arg;
+	u32 int_status;
+	u32 int_ctl;
+
+	int_status = readl(adp->fe + ADP_INT_STATUS);
+	if (int_status & ADP_INT_STATUS_VBLANK) {
+		drm_crtc_handle_vblank(&adp->crtc);
+		spin_lock(&adp->crtc.dev->event_lock);
+		if (adp->event) {
+			int_ctl = readl(adp->fe + ADP_CTRL);
+			if ((int_ctl & 0xF00) == 0x600) {
+				drm_crtc_send_vblank_event(&adp->crtc, adp->event);
+				adp->event = NULL;
+				drm_crtc_vblank_put(&adp->crtc);
+			}
+		}
+		spin_unlock(&adp->crtc.dev->event_lock);
+	}
+
+	writel(int_status, adp->fe + ADP_INT_STATUS);
+
+
+	return IRQ_HANDLED;
+}
+
+static int adp_drm_bind(struct device *dev)
+{
+	struct drm_device *drm = dev_get_drvdata(dev);
+	struct adp_drv_private *adp = to_adp(drm);
+	int err;
+
+	writel(ADP_CTRL_FIFO_ON, adp->fe + ADP_CTRL);
+
+	adp->next_bridge = drmm_of_get_bridge(&adp->drm, dev->of_node, 0, 0);
+	if (IS_ERR(adp->next_bridge)) {
+		dev_err(dev, "failed to find next bridge");
+		return PTR_ERR(adp->next_bridge);
+	}
+
+	err = adp_setup_mode_config(adp);
+	if (err < 0)
+		return err;
+
+	err = request_irq(adp->fe_irq, adp_fe_irq, 0, "adp-fe", adp);
+	if (err)
+		return err;
+
+	err = drm_dev_register(&adp->drm, 0);
+	if (err)
+		return err;
+
+	return 0;
+}
+
+static void adp_drm_unbind(struct device *dev)
+{
+	struct drm_device *drm = dev_get_drvdata(dev);
+	struct adp_drv_private *adp = to_adp(drm);
+
+	drm_dev_unregister(drm);
+	drm_atomic_helper_shutdown(drm);
+	free_irq(adp->fe_irq, adp);
+}
+
+static const struct component_master_ops adp_master_ops = {
+	.bind	= adp_drm_bind,
+	.unbind = adp_drm_unbind,
+};
+
+static int compare_dev(struct device *dev, void *data)
+{
+	return dev->of_node == data;
+}
+
+static int adp_probe(struct platform_device *pdev)
+{
+	struct device_node *port;
+	struct component_match *match = NULL;
+	struct adp_drv_private *adp;
+	int err;
+
+	adp = devm_drm_dev_alloc(&pdev->dev, &adp_driver, struct adp_drv_private, drm);
+	if (IS_ERR(adp))
+		return PTR_ERR(adp);
+
+	dev_set_drvdata(&pdev->dev, &adp->drm);
+
+	err = adp_parse_of(pdev, adp);
+	if (err < 0)
+		return err;
+
+	port = of_graph_get_remote_node(pdev->dev.of_node, 0, 0);
+	if (!port)
+		return -ENODEV;
+
+	drm_of_component_match_add(&pdev->dev, &match, compare_dev, port);
+	of_node_put(port);
+
+	return component_master_add_with_match(&pdev->dev, &adp_master_ops, match);
+}
+
+static void adp_remove(struct platform_device *pdev)
+{
+	component_master_del(&pdev->dev, &adp_master_ops);
+	dev_set_drvdata(&pdev->dev, NULL);
+}
+
+static const struct of_device_id adp_of_match[] = {
+	{ .compatible = "apple,h7-display-pipe", },
+	{ },
+};
+MODULE_DEVICE_TABLE(of, adp_of_match);
+
+static struct platform_driver adp_platform_driver = {
+	.driver = {
+		.name = "adp",
+		.of_match_table = adp_of_match,
+	},
+	.probe = adp_probe,
+	.remove = adp_remove,
+};
+
+module_platform_driver(adp_platform_driver);
+
+MODULE_DESCRIPTION("Apple Display Pipe DRM driver");
+MODULE_LICENSE("GPL");
diff --git a/drivers/gpu/drm/amd/acp/Kconfig b/drivers/gpu/drm/amd/acp/Kconfig
index e503e3d6d920..216d932a7831 100644
--- a/drivers/gpu/drm/amd/acp/Kconfig
+++ b/drivers/gpu/drm/amd/acp/Kconfig
@@ -1,11 +1,13 @@
+# SPDX-License-Identifier: MIT
 menu "ACP (Audio CoProcessor) Configuration"
+	depends on DRM_AMDGPU
 
 config DRM_AMD_ACP
-       bool "Enable AMD Audio CoProcessor IP support"
-       depends on DRM_AMDGPU
-       select MFD_CORE
-       select PM_GENERIC_DOMAINS if PM
-       help
+	bool "Enable AMD Audio CoProcessor IP support"
+	depends on DRM_AMDGPU
+	select MFD_CORE
+	select PM_GENERIC_DOMAINS if PM
+	help
 	Choose this option to enable ACP IP support for AMD SOCs.
 	This adds the ACP (Audio CoProcessor) IP driver and wires
 	it up into the amdgpu driver.  The ACP block provides the DMA
diff --git a/drivers/gpu/drm/amd/acp/Makefile b/drivers/gpu/drm/amd/acp/Makefile
index 8a08e81ee90d..d4176a3fb706 100644
--- a/drivers/gpu/drm/amd/acp/Makefile
+++ b/drivers/gpu/drm/amd/acp/Makefile
@@ -1,4 +1,25 @@
 #
+# Copyright 2017 Advanced Micro Devices, Inc.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+#
 # Makefile for the ACP, which is a sub-component
 # of AMDSOC/AMDGPU drm driver.
 # It provides the HW control for ACP related functionalities.
diff --git a/drivers/gpu/drm/amd/acp/include/acp_gfx_if.h b/drivers/gpu/drm/amd/acp/include/acp_gfx_if.h
index a72ddb2f69ac..b26710cae801 100644
--- a/drivers/gpu/drm/amd/acp/include/acp_gfx_if.h
+++ b/drivers/gpu/drm/amd/acp/include/acp_gfx_if.h
@@ -19,13 +19,12 @@
  * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
  * OTHER DEALINGS IN THE SOFTWARE.
  *
-*/
+ */
 
 #ifndef _ACP_GFX_IF_H
 #define _ACP_GFX_IF_H
 
 #include <linux/types.h>
-#include "cgs_linux.h"
 #include "cgs_common.h"
 
 int amd_acp_hw_init(struct cgs_device *cgs_device,
diff --git a/drivers/gpu/drm/amd/amdgpu/Kconfig b/drivers/gpu/drm/amd/amdgpu/Kconfig
index 26682454a446..1acfed2f92ef 100644
--- a/drivers/gpu/drm/amd/amdgpu/Kconfig
+++ b/drivers/gpu/drm/amd/amdgpu/Kconfig
@@ -1,9 +1,50 @@
+# SPDX-License-Identifier: MIT
+
+config DRM_AMDGPU
+	tristate "AMD GPU"
+	depends on DRM && PCI
+	depends on !UML
+	select FW_LOADER
+	select DRM_CLIENT
+	select DRM_CLIENT_SELECTION
+	select DRM_DISPLAY_DP_HELPER
+	select DRM_DISPLAY_DSC_HELPER
+	select DRM_DISPLAY_HDMI_HELPER
+	select DRM_DISPLAY_HDCP_HELPER
+	select DRM_DISPLAY_HELPER
+	select DRM_KMS_HELPER
+	select DRM_SCHED
+	select DRM_TTM
+	select DRM_TTM_HELPER
+	select POWER_SUPPLY
+	select HWMON
+	select I2C
+	select I2C_ALGOBIT
+	select CRC16
+	select BACKLIGHT_CLASS_DEVICE
+	select INTERVAL_TREE
+	select DRM_BUDDY
+	select DRM_SUBALLOC_HELPER
+	select DRM_EXEC
+	select DRM_PANEL_BACKLIGHT_QUIRKS
+	# amdgpu depends on ACPI_VIDEO when ACPI is enabled, for select to work
+	# ACPI_VIDEO's dependencies must also be selected.
+	select INPUT if ACPI
+	select ACPI_VIDEO if ACPI
+	# On x86 ACPI_VIDEO also needs ACPI_WMI
+	select X86_PLATFORM_DEVICES if ACPI && X86
+	select ACPI_WMI if ACPI && X86
+	help
+	  Choose this option if you have a recent AMD Radeon graphics card.
+
+	  If M is selected, the module will be called amdgpu.
+
 config DRM_AMDGPU_SI
 	bool "Enable amdgpu support for SI parts"
 	depends on DRM_AMDGPU
 	help
 	  Choose this option if you want to enable experimental support
-	  for SI asics.
+	  for SI (Southern Islands) asics.
 
 	  SI is already supported in radeon. Experimental support for SI
 	  in amdgpu will be disabled by default and is still provided by
@@ -15,7 +56,8 @@ config DRM_AMDGPU_CIK
 	bool "Enable amdgpu support for CIK parts"
 	depends on DRM_AMDGPU
 	help
-	  Choose this option if you want to enable support for CIK asics.
+	  Choose this option if you want to enable support for CIK (Sea
+	  Islands) asics.
 
 	  CIK is already supported in radeon. Support for CIK in amdgpu
 	  will be disabled by default and is still provided by radeon.
@@ -26,18 +68,33 @@ config DRM_AMDGPU_CIK
 config DRM_AMDGPU_USERPTR
 	bool "Always enable userptr write support"
 	depends on DRM_AMDGPU
+	select HMM_MIRROR
 	select MMU_NOTIFIER
 	help
-	  This option selects CONFIG_MMU_NOTIFIER if it isn't already
-	  selected to enabled full userptr support.
+	  This option selects CONFIG_HMM and CONFIG_HMM_MIRROR if it
+	  isn't already selected to enabled full userptr support.
+
+config DRM_AMD_ISP
+	bool "Enable AMD Image Signal Processor IP support"
+	depends on DRM_AMDGPU && ACPI
+	select MFD_CORE
+	select PM_GENERIC_DOMAINS if PM
+	help
+	Choose this option to enable ISP IP support for AMD SOCs.
+	This adds the ISP (Image Signal Processor) IP driver and wires
+	it up into the amdgpu driver.  It is required for cameras
+	on APUs which utilize mipi cameras.
 
-config DRM_AMDGPU_GART_DEBUGFS
-	bool "Allow GART access through debugfs"
+config DRM_AMDGPU_WERROR
+	bool "Force the compiler to throw an error instead of a warning when compiling"
 	depends on DRM_AMDGPU
-	depends on DEBUG_FS
+	depends on EXPERT
+	depends on !COMPILE_TEST
 	default n
 	help
-	  Selecting this option creates a debugfs file to inspect the mapped
-	  pages. Uses more memory for housekeeping, enable only for debugging.
+	  Add -Werror to the build flags for amdgpu.ko.
+	  Only enable this if you are warning code for amdgpu.ko.
 
 source "drivers/gpu/drm/amd/acp/Kconfig"
+source "drivers/gpu/drm/amd/display/Kconfig"
+source "drivers/gpu/drm/amd/amdkfd/Kconfig"
diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile
index 658bac0cdc5e..64e7acff8f18 100644
--- a/drivers/gpu/drm/amd/amdgpu/Makefile
+++ b/drivers/gpu/drm/amd/amdgpu/Makefile
@@ -1,47 +1,113 @@
 #
+# Copyright 2017-2024 Advanced Micro Devices, Inc. All rights reserved.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+#
 # Makefile for the drm device driver.  This driver provides support for the
 # Direct Rendering Infrastructure (DRI) in XFree86 4.1.0 and higher.
 
 FULL_AMD_PATH=$(src)/..
+DISPLAY_FOLDER_NAME=display
+FULL_AMD_DISPLAY_PATH = $(FULL_AMD_PATH)/$(DISPLAY_FOLDER_NAME)
 
 ccflags-y := -I$(FULL_AMD_PATH)/include/asic_reg \
 	-I$(FULL_AMD_PATH)/include \
 	-I$(FULL_AMD_PATH)/amdgpu \
-	-I$(FULL_AMD_PATH)/scheduler \
-	-I$(FULL_AMD_PATH)/powerplay/inc \
-	-I$(FULL_AMD_PATH)/acp/include
+	-I$(FULL_AMD_PATH)/pm/inc \
+	-I$(FULL_AMD_PATH)/acp/include \
+	-I$(FULL_AMD_DISPLAY_PATH) \
+	-I$(FULL_AMD_DISPLAY_PATH)/include \
+	-I$(FULL_AMD_DISPLAY_PATH)/modules/inc \
+	-I$(FULL_AMD_DISPLAY_PATH)/dc \
+	-I$(FULL_AMD_DISPLAY_PATH)/amdgpu_dm \
+	-I$(FULL_AMD_PATH)/amdkfd
+
+# Locally disable W=1 warnings enabled in drm subsystem Makefile
+subdir-ccflags-y += -Wno-override-init
+subdir-ccflags-$(CONFIG_DRM_AMDGPU_WERROR) += -Werror
 
 amdgpu-y := amdgpu_drv.o
 
 # add KMS driver
-amdgpu-y += amdgpu_device.o amdgpu_kms.o \
+amdgpu-y += amdgpu_device.o amdgpu_doorbell_mgr.o amdgpu_kms.o \
 	amdgpu_atombios.o atombios_crtc.o amdgpu_connectors.o \
 	atom.o amdgpu_fence.o amdgpu_ttm.o amdgpu_object.o amdgpu_gart.o \
 	amdgpu_encoders.o amdgpu_display.o amdgpu_i2c.o \
-	amdgpu_fb.o amdgpu_gem.o amdgpu_ring.o \
-	amdgpu_cs.o amdgpu_bios.o amdgpu_benchmark.o amdgpu_test.o \
-	amdgpu_pm.o atombios_dp.o amdgpu_afmt.o amdgpu_trace_points.o \
+	amdgpu_gem.o amdgpu_ring.o \
+	amdgpu_cs.o amdgpu_bios.o amdgpu_benchmark.o \
+	atombios_dp.o amdgpu_afmt.o amdgpu_trace_points.o \
 	atombios_encoders.o amdgpu_sa.o atombios_i2c.o \
-	amdgpu_prime.o amdgpu_vm.o amdgpu_ib.o amdgpu_pll.o \
+	amdgpu_dma_buf.o amdgpu_vm.o amdgpu_vm_pt.o amdgpu_vm_tlb_fence.o \
+	amdgpu_ib.o amdgpu_pll.o \
 	amdgpu_ucode.o amdgpu_bo_list.o amdgpu_ctx.o amdgpu_sync.o \
-	amdgpu_gtt_mgr.o amdgpu_vram_mgr.o amdgpu_virt.o amdgpu_atomfirmware.o \
-	amdgpu_queue_mgr.o amdgpu_vf_error.o
+	amdgpu_gtt_mgr.o amdgpu_preempt_mgr.o amdgpu_vram_mgr.o amdgpu_virt.o \
+	amdgpu_atomfirmware.o amdgpu_vf_error.o amdgpu_sched.o \
+	amdgpu_debugfs.o amdgpu_ids.o amdgpu_gmc.o amdgpu_mmhub.o amdgpu_hdp.o \
+	amdgpu_xgmi.o amdgpu_csa.o amdgpu_ras.o amdgpu_vm_cpu.o \
+	amdgpu_vm_sdma.o amdgpu_discovery.o amdgpu_ras_eeprom.o amdgpu_nbio.o \
+	amdgpu_umc.o smu_v11_0_i2c.o amdgpu_fru_eeprom.o amdgpu_rap.o \
+	amdgpu_fw_attestation.o amdgpu_securedisplay.o \
+	amdgpu_eeprom.o amdgpu_mca.o amdgpu_psp_ta.o amdgpu_lsdma.o \
+	amdgpu_ring_mux.o amdgpu_xcp.o amdgpu_seq64.o amdgpu_aca.o amdgpu_dev_coredump.o \
+	amdgpu_cper.o amdgpu_userq_fence.o amdgpu_eviction_fence.o amdgpu_ip.o
+
+amdgpu-$(CONFIG_PROC_FS) += amdgpu_fdinfo.o
+
+amdgpu-$(CONFIG_PERF_EVENTS) += amdgpu_pmu.o
 
 # add asic specific block
-amdgpu-$(CONFIG_DRM_AMDGPU_CIK)+= cik.o cik_ih.o kv_smc.o kv_dpm.o \
-	ci_smc.o ci_dpm.o dce_v8_0.o gfx_v7_0.o cik_sdma.o uvd_v4_2.o vce_v2_0.o \
-	amdgpu_amdkfd_gfx_v7.o
+amdgpu-$(CONFIG_DRM_AMDGPU_CIK)+= cik.o cik_ih.o \
+	dce_v8_0.o gfx_v7_0.o cik_sdma.o uvd_v4_2.o vce_v2_0.o
+
+amdgpu-$(CONFIG_DRM_AMDGPU_SI)+= si.o gmc_v6_0.o gfx_v6_0.o si_ih.o si_dma.o dce_v6_0.o \
+	uvd_v3_1.o
 
-amdgpu-$(CONFIG_DRM_AMDGPU_SI)+= si.o gmc_v6_0.o gfx_v6_0.o si_ih.o si_dma.o dce_v6_0.o si_dpm.o si_smc.o
+amdgpu-y += \
+	vi.o mxgpu_vi.o nbio_v6_1.o soc15.o emu_soc.o mxgpu_ai.o nbio_v7_0.o vega10_reg_init.o \
+	vega20_reg_init.o nbio_v7_4.o nbio_v2_3.o nv.o arct_reg_init.o mxgpu_nv.o \
+	nbio_v7_2.o hdp_v4_0.o hdp_v5_0.o aldebaran_reg_init.o aldebaran.o soc21.o soc24.o \
+	sienna_cichlid.o smu_v13_0_10.o nbio_v4_3.o hdp_v6_0.o nbio_v7_7.o hdp_v5_2.o lsdma_v6_0.o \
+	nbio_v7_9.o aqua_vanjaram.o nbio_v7_11.o lsdma_v7_0.o hdp_v7_0.o nbif_v6_3_1.o \
+	cyan_skillfish_reg_init.o
 
+# add DF block
 amdgpu-y += \
-	vi.o mxgpu_vi.o nbio_v6_1.o soc15.o mxgpu_ai.o nbio_v7_0.o
+	df_v1_7.o \
+	df_v3_6.o \
+	df_v4_3.o \
+	df_v4_6_2.o \
+	df_v4_15.o
 
 # add GMC block
 amdgpu-y += \
 	gmc_v7_0.o \
 	gmc_v8_0.o \
-	gfxhub_v1_0.o mmhub_v1_0.o gmc_v9_0.o
+	gfxhub_v1_0.o mmhub_v1_0.o gmc_v9_0.o gfxhub_v1_1.o mmhub_v9_4.o \
+	gfxhub_v2_0.o mmhub_v2_0.o gmc_v10_0.o gfxhub_v2_1.o mmhub_v2_3.o \
+	mmhub_v1_7.o gfxhub_v3_0.o mmhub_v3_0.o mmhub_v3_0_2.o gmc_v11_0.o \
+	mmhub_v3_0_1.o gfxhub_v3_0_3.o gfxhub_v1_2.o mmhub_v1_8.o mmhub_v3_3.o \
+	gfxhub_v11_5_0.o mmhub_v4_1_0.o gfxhub_v12_0.o gmc_v12_0.o
+
+# add UMC block
+amdgpu-y += \
+	umc_v6_0.o umc_v6_1.o umc_v6_7.o umc_v8_7.o umc_v8_10.o umc_v12_0.o umc_v8_14.o
 
 # add IH block
 amdgpu-y += \
@@ -50,36 +116,68 @@ amdgpu-y += \
 	iceland_ih.o \
 	tonga_ih.o \
 	cz_ih.o \
-	vega10_ih.o
+	vega10_ih.o \
+	vega20_ih.o \
+	navi10_ih.o \
+	ih_v6_0.o \
+	ih_v6_1.o \
+	ih_v7_0.o
 
 # add PSP block
 amdgpu-y += \
 	amdgpu_psp.o \
 	psp_v3_1.o \
-	psp_v10_0.o
-
-# add SMC block
-amdgpu-y += \
-	amdgpu_dpm.o \
-	amdgpu_powerplay.o
+	psp_v10_0.o \
+	psp_v11_0.o \
+	psp_v11_0_8.o \
+	psp_v12_0.o \
+	psp_v13_0.o \
+	psp_v13_0_4.o \
+	psp_v14_0.o
 
 # add DCE block
 amdgpu-y += \
 	dce_v10_0.o \
-	dce_v11_0.o \
-	dce_virtual.o
+	amdgpu_vkms.o
 
 # add GFX block
 amdgpu-y += \
 	amdgpu_gfx.o \
+	amdgpu_rlc.o \
 	gfx_v8_0.o \
-	gfx_v9_0.o
+	gfx_v9_0.o \
+	gfx_v9_4.o \
+	gfx_v9_4_2.o \
+	gfx_v9_4_3.o \
+	gfx_v10_0.o \
+	imu_v11_0.o \
+	gfx_v11_0.o \
+	gfx_v11_0_3.o \
+	imu_v11_0_3.o \
+	gfx_v12_0.o \
+	imu_v12_0.o
 
 # add async DMA block
 amdgpu-y += \
+	amdgpu_sdma.o \
 	sdma_v2_4.o \
 	sdma_v3_0.o \
-	sdma_v4_0.o
+	sdma_v4_0.o \
+	sdma_v4_4.o \
+	sdma_v4_4_2.o \
+	sdma_v5_0.o \
+	sdma_v5_2.o \
+	sdma_v6_0.o \
+	sdma_v7_0.o
+
+# add MES block
+amdgpu-y += \
+	amdgpu_mes.o \
+	mes_v11_0.o \
+	mes_v12_0.o \
+
+# add GFX userqueue support
+amdgpu-y += mes_userqueue.o
 
 # add UVD block
 amdgpu-y += \
@@ -94,24 +192,101 @@ amdgpu-y += \
 	vce_v3_0.o \
 	vce_v4_0.o
 
-# add VCN block
+# add VCN and JPEG block
 amdgpu-y += \
 	amdgpu_vcn.o \
-	vcn_v1_0.o
+	vcn_sw_ring.o \
+	vcn_v1_0.o \
+	vcn_v2_0.o \
+	vcn_v2_5.o \
+	vcn_v3_0.o \
+	vcn_v4_0.o \
+	vcn_v4_0_3.o \
+	vcn_v4_0_5.o \
+	vcn_v5_0_0.o \
+	vcn_v5_0_1.o \
+	amdgpu_jpeg.o \
+	jpeg_v1_0.o \
+	jpeg_v2_0.o \
+	jpeg_v2_5.o \
+	jpeg_v3_0.o \
+	jpeg_v4_0.o \
+	jpeg_v4_0_3.o \
+	jpeg_v4_0_5.o \
+	jpeg_v5_0_0.o \
+	jpeg_v5_0_1.o
+
+# add VPE block
+amdgpu-y += \
+	amdgpu_vpe.o \
+	vpe_v6_1.o
+
+# add UMSCH block
+amdgpu-y += \
+	amdgpu_umsch_mm.o \
+	umsch_mm_v4_0.o
+
+#
+# add ATHUB block
+amdgpu-y += \
+	athub_v1_0.o \
+	athub_v2_0.o \
+	athub_v2_1.o \
+	athub_v3_0.o \
+	athub_v4_1_0.o
+
+# add SMUIO block
+amdgpu-y += \
+	smuio_v9_0.o \
+	smuio_v11_0.o \
+	smuio_v11_0_6.o \
+	smuio_v13_0.o \
+	smuio_v13_0_3.o \
+	smuio_v13_0_6.o \
+	smuio_v14_0_2.o
+
+# add reset block
+amdgpu-y += \
+	amdgpu_reset.o
+
+# add MCA block
+amdgpu-y += \
+	mca_v3_0.o
 
 # add amdkfd interfaces
+amdgpu-y += amdgpu_amdkfd.o
+
+# add gfx usermode queue
+amdgpu-y += amdgpu_userq.o
+
+ifneq ($(CONFIG_HSA_AMD),)
+AMDKFD_PATH := ../amdkfd
+include $(FULL_AMD_PATH)/amdkfd/Makefile
+amdgpu-y += $(AMDKFD_FILES)
 amdgpu-y += \
-	 amdgpu_amdkfd.o \
-	 amdgpu_amdkfd_gfx_v8.o
+	amdgpu_amdkfd_fence.o \
+	amdgpu_amdkfd_gpuvm.o \
+	amdgpu_amdkfd_gfx_v8.o \
+	amdgpu_amdkfd_gfx_v9.o \
+	amdgpu_amdkfd_arcturus.o \
+	amdgpu_amdkfd_aldebaran.o \
+	amdgpu_amdkfd_gc_9_4_3.o \
+	amdgpu_amdkfd_gfx_v10.o \
+	amdgpu_amdkfd_gfx_v10_3.o \
+	amdgpu_amdkfd_gfx_v11.o \
+	amdgpu_amdkfd_gfx_v12.o
+
+ifneq ($(CONFIG_DRM_AMDGPU_CIK),)
+amdgpu-y += amdgpu_amdkfd_gfx_v7.o
+endif
+
+endif
 
 # add cgs
 amdgpu-y += amdgpu_cgs.o
 
 # GPU scheduler
-amdgpu-y += \
-	../scheduler/gpu_scheduler.o \
-	../scheduler/sched_fence.o \
-	amdgpu_job.o
+amdgpu-y += amdgpu_job.o
 
 # ACP componet
 ifneq ($(CONFIG_DRM_AMD_ACP),)
@@ -126,12 +301,27 @@ endif
 amdgpu-$(CONFIG_COMPAT) += amdgpu_ioc32.o
 amdgpu-$(CONFIG_VGA_SWITCHEROO) += amdgpu_atpx_handler.o
 amdgpu-$(CONFIG_ACPI) += amdgpu_acpi.o
-amdgpu-$(CONFIG_MMU_NOTIFIER) += amdgpu_mn.o
+amdgpu-$(CONFIG_HMM_MIRROR) += amdgpu_hmm.o
 
-include $(FULL_AMD_PATH)/powerplay/Makefile
+include $(FULL_AMD_PATH)/pm/Makefile
 
 amdgpu-y += $(AMD_POWERPLAY_FILES)
 
-obj-$(CONFIG_DRM_AMDGPU)+= amdgpu.o
+ifneq ($(CONFIG_DRM_AMD_DC),)
 
-CFLAGS_amdgpu_trace_points.o := -I$(src)
+RELATIVE_AMD_DISPLAY_PATH = ../$(DISPLAY_FOLDER_NAME)
+include $(FULL_AMD_DISPLAY_PATH)/Makefile
+
+amdgpu-y += $(AMD_DISPLAY_FILES)
+
+endif
+
+# add isp block
+ifneq ($(CONFIG_DRM_AMD_ISP),)
+amdgpu-y += \
+	amdgpu_isp.o \
+	isp_v4_1_0.o \
+	isp_v4_1_1.o
+endif
+
+obj-$(CONFIG_DRM_AMDGPU)+= amdgpu.o
diff --git a/drivers/gpu/drm/amd/amdgpu/ObjectID.h b/drivers/gpu/drm/amd/amdgpu/ObjectID.h
index 06192698bd96..a0f0a17e224f 100644
--- a/drivers/gpu/drm/amd/amdgpu/ObjectID.h
+++ b/drivers/gpu/drm/amd/amdgpu/ObjectID.h
@@ -119,6 +119,7 @@
 #define CONNECTOR_OBJECT_ID_eDP                   0x14
 #define CONNECTOR_OBJECT_ID_MXM                   0x15
 #define CONNECTOR_OBJECT_ID_LVDS_eDP              0x16
+#define CONNECTOR_OBJECT_ID_USBC                  0x17
 
 /* deleted */
 
@@ -136,6 +137,7 @@
 #define GENERIC_OBJECT_ID_PX2_NON_DRIVABLE        0x02
 #define GENERIC_OBJECT_ID_MXM_OPM                 0x03
 #define GENERIC_OBJECT_ID_STEREO_PIN              0x04        //This object could show up from Misc Object table, it follows ATOM_OBJECT format, and contains one ATOM_OBJECT_GPIO_CNTL_RECORD for the stereo pin
+#define GENERIC_OBJECT_ID_BRACKET_LAYOUT          0x05
 
 /****************************************************/
 /* Graphics Object ENUM ID Definition               */
@@ -714,6 +716,13 @@
                                                  GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
                                                  GENERIC_OBJECT_ID_STEREO_PIN << OBJECT_ID_SHIFT)
 
+#define GENERICOBJECT_BRACKET_LAYOUT_ENUM_ID1    (GRAPH_OBJECT_TYPE_GENERIC << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 GENERIC_OBJECT_ID_BRACKET_LAYOUT << OBJECT_ID_SHIFT)
+
+#define GENERICOBJECT_BRACKET_LAYOUT_ENUM_ID2    (GRAPH_OBJECT_TYPE_GENERIC << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 GENERIC_OBJECT_ID_BRACKET_LAYOUT << OBJECT_ID_SHIFT)
 /****************************************************/
 /* Object Cap definition - Shared with BIOS         */
 /****************************************************/
diff --git a/drivers/gpu/drm/amd/amdgpu/aldebaran.c b/drivers/gpu/drm/amd/amdgpu/aldebaran.c
new file mode 100644
index 000000000000..9569dc16dd3d
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/aldebaran.c
@@ -0,0 +1,469 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "aldebaran.h"
+#include "amdgpu_reset.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_dpm.h"
+#include "amdgpu_job.h"
+#include "amdgpu_ring.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_psp.h"
+#include "amdgpu_xgmi.h"
+
+static bool aldebaran_is_mode2_default(struct amdgpu_reset_control *reset_ctl)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle;
+
+	if ((amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 2) &&
+	     adev->gmc.xgmi.connected_to_cpu))
+		return true;
+
+	return false;
+}
+
+static struct amdgpu_reset_handler *
+aldebaran_get_reset_handler(struct amdgpu_reset_control *reset_ctl,
+			    struct amdgpu_reset_context *reset_context)
+{
+	struct amdgpu_reset_handler *handler;
+	struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle;
+	int i;
+
+	if (reset_context->method == AMD_RESET_METHOD_NONE) {
+		if (aldebaran_is_mode2_default(reset_ctl))
+			reset_context->method = AMD_RESET_METHOD_MODE2;
+		else
+			reset_context->method = amdgpu_asic_reset_method(adev);
+	}
+
+	if (reset_context->method != AMD_RESET_METHOD_NONE) {
+		dev_dbg(adev->dev, "Getting reset handler for method %d\n",
+			reset_context->method);
+		for_each_handler(i, handler, reset_ctl) {
+			if (handler->reset_method == reset_context->method)
+				return handler;
+		}
+	}
+
+	dev_dbg(adev->dev, "Reset handler not found!\n");
+
+	return NULL;
+}
+
+static inline uint32_t aldebaran_get_ip_block_mask(struct amdgpu_device *adev)
+{
+	uint32_t ip_block_mask = BIT(AMD_IP_BLOCK_TYPE_GFX) |
+				 BIT(AMD_IP_BLOCK_TYPE_SDMA);
+
+	if (adev->aid_mask)
+		ip_block_mask |= BIT(AMD_IP_BLOCK_TYPE_IH);
+
+	return ip_block_mask;
+}
+
+static int aldebaran_mode2_suspend_ip(struct amdgpu_device *adev)
+{
+	uint32_t ip_block_mask = aldebaran_get_ip_block_mask(adev);
+	uint32_t ip_block;
+	int r, i;
+
+	amdgpu_device_set_pg_state(adev, AMD_PG_STATE_UNGATE);
+	amdgpu_device_set_cg_state(adev, AMD_CG_STATE_UNGATE);
+
+	for (i = adev->num_ip_blocks - 1; i >= 0; i--) {
+		ip_block = BIT(adev->ip_blocks[i].version->type);
+		if (!(ip_block_mask & ip_block))
+			continue;
+
+		r = amdgpu_ip_block_suspend(&adev->ip_blocks[i]);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+static int
+aldebaran_mode2_prepare_hwcontext(struct amdgpu_reset_control *reset_ctl,
+				  struct amdgpu_reset_context *reset_context)
+{
+	int r = 0;
+	struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle;
+
+	dev_dbg(adev->dev, "Aldebaran prepare hw context\n");
+	/* Don't suspend on bare metal if we are not going to HW reset the ASIC */
+	if (!amdgpu_sriov_vf(adev))
+		r = aldebaran_mode2_suspend_ip(adev);
+
+	return r;
+}
+
+static void aldebaran_async_reset(struct work_struct *work)
+{
+	struct amdgpu_reset_handler *handler;
+	struct amdgpu_reset_control *reset_ctl =
+		container_of(work, struct amdgpu_reset_control, reset_work);
+	struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle;
+	int i;
+
+	for_each_handler(i, handler, reset_ctl)	{
+		if (handler->reset_method == reset_ctl->active_reset) {
+			dev_dbg(adev->dev, "Resetting device\n");
+			handler->do_reset(adev);
+			break;
+		}
+	}
+}
+
+static int aldebaran_mode2_reset(struct amdgpu_device *adev)
+{
+	/* disable BM */
+	pci_clear_master(adev->pdev);
+	adev->asic_reset_res = amdgpu_dpm_mode2_reset(adev);
+	return adev->asic_reset_res;
+}
+
+static int
+aldebaran_mode2_perform_reset(struct amdgpu_reset_control *reset_ctl,
+			      struct amdgpu_reset_context *reset_context)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle;
+	struct list_head *reset_device_list = reset_context->reset_device_list;
+	struct amdgpu_device *tmp_adev = NULL;
+	int r = 0;
+
+	dev_dbg(adev->dev, "aldebaran perform hw reset\n");
+
+	if (reset_device_list == NULL)
+		return -EINVAL;
+
+	if (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 2) &&
+	    reset_context->hive == NULL) {
+		/* Wrong context, return error */
+		return -EINVAL;
+	}
+
+	list_for_each_entry(tmp_adev, reset_device_list, reset_list) {
+		mutex_lock(&tmp_adev->reset_cntl->reset_lock);
+		tmp_adev->reset_cntl->active_reset = AMD_RESET_METHOD_MODE2;
+	}
+	/*
+	 * Mode2 reset doesn't need any sync between nodes in XGMI hive, instead launch
+	 * them together so that they can be completed asynchronously on multiple nodes
+	 */
+	list_for_each_entry(tmp_adev, reset_device_list, reset_list) {
+		/* For XGMI run all resets in parallel to speed up the process */
+		if (tmp_adev->gmc.xgmi.num_physical_nodes > 1) {
+			if (!queue_work(system_unbound_wq,
+					&tmp_adev->reset_cntl->reset_work))
+				r = -EALREADY;
+		} else
+			r = aldebaran_mode2_reset(tmp_adev);
+		if (r) {
+			dev_err(tmp_adev->dev,
+				"ASIC reset failed with error, %d for drm dev, %s",
+				r, adev_to_drm(tmp_adev)->unique);
+			break;
+		}
+	}
+
+	/* For XGMI wait for all resets to complete before proceed */
+	if (!r) {
+		list_for_each_entry(tmp_adev, reset_device_list, reset_list) {
+			if (tmp_adev->gmc.xgmi.num_physical_nodes > 1) {
+				flush_work(&tmp_adev->reset_cntl->reset_work);
+				r = tmp_adev->asic_reset_res;
+				if (r)
+					break;
+			}
+		}
+	}
+
+	list_for_each_entry(tmp_adev, reset_device_list, reset_list) {
+		mutex_unlock(&tmp_adev->reset_cntl->reset_lock);
+		tmp_adev->reset_cntl->active_reset = AMD_RESET_METHOD_NONE;
+	}
+
+	return r;
+}
+
+static int aldebaran_mode2_restore_ip(struct amdgpu_device *adev)
+{
+	struct amdgpu_firmware_info *ucode_list[AMDGPU_UCODE_ID_MAXIMUM];
+	uint32_t ip_block_mask = aldebaran_get_ip_block_mask(adev);
+	struct amdgpu_firmware_info *ucode;
+	struct amdgpu_ip_block *cmn_block;
+	struct amdgpu_ip_block *ih_block;
+	int ucode_count = 0;
+	int i, r;
+
+	dev_dbg(adev->dev, "Reloading ucodes after reset\n");
+	for (i = 0; i < adev->firmware.max_ucodes; i++) {
+		ucode = &adev->firmware.ucode[i];
+		if (!ucode->fw)
+			continue;
+		switch (ucode->ucode_id) {
+		case AMDGPU_UCODE_ID_SDMA0:
+		case AMDGPU_UCODE_ID_SDMA1:
+		case AMDGPU_UCODE_ID_SDMA2:
+		case AMDGPU_UCODE_ID_SDMA3:
+		case AMDGPU_UCODE_ID_SDMA4:
+		case AMDGPU_UCODE_ID_SDMA5:
+		case AMDGPU_UCODE_ID_SDMA6:
+		case AMDGPU_UCODE_ID_SDMA7:
+		case AMDGPU_UCODE_ID_CP_MEC1:
+		case AMDGPU_UCODE_ID_CP_MEC1_JT:
+		case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL:
+		case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM:
+		case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM:
+		case AMDGPU_UCODE_ID_RLC_G:
+			ucode_list[ucode_count++] = ucode;
+			break;
+		default:
+			break;
+		}
+	}
+
+	/* Reinit NBIF block */
+	cmn_block =
+		amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_COMMON);
+	if (unlikely(!cmn_block)) {
+		dev_err(adev->dev, "Failed to get BIF handle\n");
+		return -EINVAL;
+	}
+	r = amdgpu_ip_block_resume(cmn_block);
+	if (r)
+		return r;
+
+	if (ip_block_mask & BIT(AMD_IP_BLOCK_TYPE_IH)) {
+		ih_block = amdgpu_device_ip_get_ip_block(adev,
+							 AMD_IP_BLOCK_TYPE_IH);
+		if (unlikely(!ih_block)) {
+			dev_err(adev->dev, "Failed to get IH handle\n");
+			return -EINVAL;
+		}
+		r = amdgpu_ip_block_resume(ih_block);
+		if (r)
+			return r;
+	}
+
+	/* Reinit GFXHUB */
+	adev->gfxhub.funcs->init(adev);
+	r = adev->gfxhub.funcs->gart_enable(adev);
+	if (r) {
+		dev_err(adev->dev, "GFXHUB gart reenable failed after reset\n");
+		return r;
+	}
+
+	/* Reload GFX firmware */
+	r = psp_load_fw_list(&adev->psp, ucode_list, ucode_count);
+	if (r) {
+		dev_err(adev->dev, "GFX ucode load failed after reset\n");
+		return r;
+	}
+
+	/* Resume RLC, FW needs RLC alive to complete reset process */
+	adev->gfx.rlc.funcs->resume(adev);
+
+	/* Wait for FW reset event complete */
+	r = amdgpu_dpm_wait_for_event(adev, SMU_EVENT_RESET_COMPLETE, 0);
+	if (r) {
+		dev_err(adev->dev,
+			"Failed to get response from firmware after reset\n");
+		return r;
+	}
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!(adev->ip_blocks[i].version->type ==
+			      AMD_IP_BLOCK_TYPE_GFX ||
+		      adev->ip_blocks[i].version->type ==
+			      AMD_IP_BLOCK_TYPE_SDMA))
+			continue;
+
+		r = amdgpu_ip_block_resume(&adev->ip_blocks[i]);
+		if (r)
+			return r;
+	}
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!(adev->ip_blocks[i].version->type ==
+			      AMD_IP_BLOCK_TYPE_GFX ||
+		      adev->ip_blocks[i].version->type ==
+			      AMD_IP_BLOCK_TYPE_SDMA ||
+		      adev->ip_blocks[i].version->type ==
+			      AMD_IP_BLOCK_TYPE_COMMON))
+			continue;
+
+		if (adev->ip_blocks[i].version->funcs->late_init) {
+			r = adev->ip_blocks[i].version->funcs->late_init(
+				&adev->ip_blocks[i]);
+			if (r) {
+				dev_err(adev->dev,
+					"late_init of IP block <%s> failed %d after reset\n",
+					adev->ip_blocks[i].version->funcs->name,
+					r);
+				return r;
+			}
+		}
+		adev->ip_blocks[i].status.late_initialized = true;
+	}
+
+	amdgpu_device_set_cg_state(adev, AMD_CG_STATE_GATE);
+	amdgpu_device_set_pg_state(adev, AMD_PG_STATE_GATE);
+
+	return r;
+}
+
+static int
+aldebaran_mode2_restore_hwcontext(struct amdgpu_reset_control *reset_ctl,
+				  struct amdgpu_reset_context *reset_context)
+{
+	struct list_head *reset_device_list = reset_context->reset_device_list;
+	struct amdgpu_device *tmp_adev = NULL;
+	struct amdgpu_ras *con;
+	int r;
+
+	if (reset_device_list == NULL)
+		return -EINVAL;
+
+	if (amdgpu_ip_version(reset_context->reset_req_dev, MP1_HWIP, 0) ==
+		    IP_VERSION(13, 0, 2) &&
+	    reset_context->hive == NULL) {
+		/* Wrong context, return error */
+		return -EINVAL;
+	}
+
+	list_for_each_entry(tmp_adev, reset_device_list, reset_list) {
+		amdgpu_set_init_level(tmp_adev,
+				AMDGPU_INIT_LEVEL_RESET_RECOVERY);
+		dev_info(tmp_adev->dev,
+			 "GPU reset succeeded, trying to resume\n");
+		/*TBD: Ideally should clear only GFX, SDMA blocks*/
+		amdgpu_ras_clear_err_state(tmp_adev);
+		r = aldebaran_mode2_restore_ip(tmp_adev);
+		if (r)
+			goto end;
+
+		/*
+		 * Add this ASIC as tracked as reset was already
+		 * complete successfully.
+		 */
+		amdgpu_register_gpu_instance(tmp_adev);
+
+		/* Resume RAS, ecc_irq */
+		con = amdgpu_ras_get_context(tmp_adev);
+		if (!amdgpu_sriov_vf(tmp_adev) && con) {
+			if (tmp_adev->sdma.ras &&
+				tmp_adev->sdma.ras->ras_block.ras_late_init) {
+				r = tmp_adev->sdma.ras->ras_block.ras_late_init(tmp_adev,
+						&tmp_adev->sdma.ras->ras_block.ras_comm);
+				if (r) {
+					dev_err(tmp_adev->dev, "SDMA failed to execute ras_late_init! ret:%d\n", r);
+					goto end;
+				}
+			}
+
+			if (tmp_adev->gfx.ras &&
+				tmp_adev->gfx.ras->ras_block.ras_late_init) {
+				r = tmp_adev->gfx.ras->ras_block.ras_late_init(tmp_adev,
+						&tmp_adev->gfx.ras->ras_block.ras_comm);
+				if (r) {
+					dev_err(tmp_adev->dev, "GFX failed to execute ras_late_init! ret:%d\n", r);
+					goto end;
+				}
+			}
+		}
+
+		amdgpu_ras_resume(tmp_adev);
+
+		/* Update PSP FW topology after reset */
+		if (reset_context->hive &&
+		    tmp_adev->gmc.xgmi.num_physical_nodes > 1)
+			r = amdgpu_xgmi_update_topology(reset_context->hive,
+							tmp_adev);
+
+		if (!r) {
+			amdgpu_set_init_level(tmp_adev,
+					      AMDGPU_INIT_LEVEL_DEFAULT);
+			amdgpu_irq_gpu_reset_resume_helper(tmp_adev);
+
+			r = amdgpu_ib_ring_tests(tmp_adev);
+			if (r) {
+				dev_err(tmp_adev->dev,
+					"ib ring test failed (%d).\n", r);
+				r = -EAGAIN;
+				tmp_adev->asic_reset_res = r;
+				goto end;
+			}
+		}
+	}
+
+end:
+	return r;
+}
+
+static struct amdgpu_reset_handler aldebaran_mode2_handler = {
+	.reset_method		= AMD_RESET_METHOD_MODE2,
+	.prepare_env		= NULL,
+	.prepare_hwcontext	= aldebaran_mode2_prepare_hwcontext,
+	.perform_reset		= aldebaran_mode2_perform_reset,
+	.restore_hwcontext	= aldebaran_mode2_restore_hwcontext,
+	.restore_env		= NULL,
+	.do_reset		= aldebaran_mode2_reset,
+};
+
+static struct amdgpu_reset_handler
+	*aldebaran_rst_handlers[AMDGPU_RESET_MAX_HANDLERS] = {
+		&aldebaran_mode2_handler,
+		&xgmi_reset_on_init_handler,
+	};
+
+int aldebaran_reset_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_reset_control *reset_ctl;
+
+	reset_ctl = kzalloc(sizeof(*reset_ctl), GFP_KERNEL);
+	if (!reset_ctl)
+		return -ENOMEM;
+
+	reset_ctl->handle = adev;
+	reset_ctl->async_reset = aldebaran_async_reset;
+	reset_ctl->active_reset = AMD_RESET_METHOD_NONE;
+	reset_ctl->get_reset_handler = aldebaran_get_reset_handler;
+
+	INIT_WORK(&reset_ctl->reset_work, reset_ctl->async_reset);
+	/* Only mode2 is handled through reset control now */
+	reset_ctl->reset_handlers = &aldebaran_rst_handlers;
+
+	adev->reset_cntl = reset_ctl;
+
+	return 0;
+}
+
+int aldebaran_reset_fini(struct amdgpu_device *adev)
+{
+	kfree(adev->reset_cntl);
+	adev->reset_cntl = NULL;
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/aldebaran.h b/drivers/gpu/drm/amd/amdgpu/aldebaran.h
new file mode 100644
index 000000000000..a07db5454d49
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/aldebaran.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __ALDEBARAN_H__
+#define __ALDEBARAN_H__
+
+#include "amdgpu.h"
+
+int aldebaran_reset_init(struct amdgpu_device *adev);
+int aldebaran_reset_fini(struct amdgpu_device *adev);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/aldebaran_reg_init.c b/drivers/gpu/drm/amd/amdgpu/aldebaran_reg_init.c
new file mode 100644
index 000000000000..28e6c9ab8767
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/aldebaran_reg_init.c
@@ -0,0 +1,54 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "soc15.h"
+
+#include "soc15_common.h"
+#include "aldebaran_ip_offset.h"
+
+int aldebaran_reg_base_init(struct amdgpu_device *adev)
+{
+	/* HW has more IP blocks,  only initialized the block needed by our driver  */
+	uint32_t i;
+	for (i = 0 ; i < MAX_INSTANCE ; ++i) {
+		adev->reg_offset[GC_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i]));
+		adev->reg_offset[HDP_HWIP][i] = (uint32_t *)(&(HDP_BASE.instance[i]));
+		adev->reg_offset[MMHUB_HWIP][i] = (uint32_t *)(&(MMHUB_BASE.instance[i]));
+		adev->reg_offset[ATHUB_HWIP][i] = (uint32_t *)(&(ATHUB_BASE.instance[i]));
+		adev->reg_offset[NBIO_HWIP][i] = (uint32_t *)(&(NBIO_BASE.instance[i]));
+		adev->reg_offset[MP0_HWIP][i] = (uint32_t *)(&(MP0_BASE.instance[i]));
+		adev->reg_offset[MP1_HWIP][i] = (uint32_t *)(&(MP1_BASE.instance[i]));
+		adev->reg_offset[DF_HWIP][i] = (uint32_t *)(&(DF_BASE.instance[i]));
+		adev->reg_offset[OSSSYS_HWIP][i] = (uint32_t *)(&(OSSSYS_BASE.instance[i]));
+		adev->reg_offset[SDMA0_HWIP][i] = (uint32_t *)(&(SDMA0_BASE.instance[i]));
+		adev->reg_offset[SDMA1_HWIP][i] = (uint32_t *)(&(SDMA1_BASE.instance[i]));
+		adev->reg_offset[SDMA2_HWIP][i] = (uint32_t *)(&(SDMA2_BASE.instance[i]));
+		adev->reg_offset[SDMA3_HWIP][i] = (uint32_t *)(&(SDMA3_BASE.instance[i]));
+		adev->reg_offset[SDMA4_HWIP][i] = (uint32_t *)(&(SDMA4_BASE.instance[i]));
+		adev->reg_offset[SMUIO_HWIP][i] = (uint32_t *)(&(SMUIO_BASE.instance[i]));
+		adev->reg_offset[THM_HWIP][i] = (uint32_t *)(&(THM_BASE.instance[i]));
+		adev->reg_offset[UMC_HWIP][i] = (uint32_t *)(&(UMC_BASE.instance[i]));
+		adev->reg_offset[VCN_HWIP][i] = (uint32_t *)(&(VCN_BASE.instance[i]));
+	}
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index 103635ab784c..2a0df4cabb99 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -28,6 +28,20 @@
 #ifndef __AMDGPU_H__
 #define __AMDGPU_H__
 
+#ifdef pr_fmt
+#undef pr_fmt
+#endif
+
+#define pr_fmt(fmt) "amdgpu: " fmt
+
+#ifdef dev_fmt
+#undef dev_fmt
+#endif
+
+#define dev_fmt(fmt) "amdgpu: " fmt
+
+#include "amdgpu_ctx.h"
+
 #include <linux/atomic.h>
 #include <linux/wait.h>
 #include <linux/list.h>
@@ -35,20 +49,21 @@
 #include <linux/rbtree.h>
 #include <linux/hashtable.h>
 #include <linux/dma-fence.h>
+#include <linux/pci.h>
 
-#include <drm/ttm/ttm_bo_api.h>
-#include <drm/ttm/ttm_bo_driver.h>
+#include <drm/ttm/ttm_bo.h>
 #include <drm/ttm/ttm_placement.h>
-#include <drm/ttm/ttm_module.h>
-#include <drm/ttm/ttm_execbuf_util.h>
 
-#include <drm/drmP.h>
-#include <drm/drm_gem.h>
 #include <drm/amdgpu_drm.h>
+#include <drm/drm_gem.h>
+#include <drm/drm_ioctl.h>
 
 #include <kgd_kfd_interface.h>
+#include "dm_pp_interface.h"
+#include "kgd_pp_interface.h"
 
 #include "amd_shared.h"
+#include "amdgpu_utils.h"
 #include "amdgpu_mode.h"
 #include "amdgpu_ih.h"
 #include "amdgpu_irq.h"
@@ -59,39 +74,111 @@
 #include "amdgpu_sync.h"
 #include "amdgpu_ring.h"
 #include "amdgpu_vm.h"
-#include "amd_powerplay.h"
 #include "amdgpu_dpm.h"
 #include "amdgpu_acp.h"
 #include "amdgpu_uvd.h"
 #include "amdgpu_vce.h"
 #include "amdgpu_vcn.h"
-
-#include "gpu_scheduler.h"
+#include "amdgpu_jpeg.h"
+#include "amdgpu_vpe.h"
+#include "amdgpu_umsch_mm.h"
+#include "amdgpu_gmc.h"
+#include "amdgpu_gfx.h"
+#include "amdgpu_sdma.h"
+#include "amdgpu_lsdma.h"
+#include "amdgpu_nbio.h"
+#include "amdgpu_hdp.h"
+#include "amdgpu_dm.h"
 #include "amdgpu_virt.h"
+#include "amdgpu_csa.h"
+#include "amdgpu_mes_ctx.h"
 #include "amdgpu_gart.h"
+#include "amdgpu_debugfs.h"
+#include "amdgpu_job.h"
+#include "amdgpu_bo_list.h"
+#include "amdgpu_gem.h"
+#include "amdgpu_doorbell.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_discovery.h"
+#include "amdgpu_mes.h"
+#include "amdgpu_umc.h"
+#include "amdgpu_mmhub.h"
+#include "amdgpu_gfxhub.h"
+#include "amdgpu_df.h"
+#include "amdgpu_smuio.h"
+#include "amdgpu_fdinfo.h"
+#include "amdgpu_mca.h"
+#include "amdgpu_aca.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_cper.h"
+#include "amdgpu_xcp.h"
+#include "amdgpu_seq64.h"
+#include "amdgpu_reg_state.h"
+#include "amdgpu_userq.h"
+#include "amdgpu_eviction_fence.h"
+#if defined(CONFIG_DRM_AMD_ISP)
+#include "amdgpu_isp.h"
+#endif
+
+#define MAX_GPU_INSTANCE		64
+
+#define GFX_SLICE_PERIOD_MS		250
+
+struct amdgpu_gpu_instance {
+	struct amdgpu_device		*adev;
+	int				mgpu_fan_enabled;
+};
+
+struct amdgpu_mgpu_info {
+	struct amdgpu_gpu_instance	gpu_ins[MAX_GPU_INSTANCE];
+	struct mutex			mutex;
+	uint32_t			num_gpu;
+	uint32_t			num_dgpu;
+	uint32_t			num_apu;
+};
+
+enum amdgpu_ss {
+	AMDGPU_SS_DRV_LOAD,
+	AMDGPU_SS_DEV_D0,
+	AMDGPU_SS_DEV_D3,
+	AMDGPU_SS_DRV_UNLOAD
+};
+
+struct amdgpu_hwip_reg_entry {
+	u32		hwip;
+	u32		inst;
+	u32		seg;
+	u32		reg_offset;
+	const char	*reg_name;
+};
+
+struct amdgpu_watchdog_timer {
+	bool timeout_fatal_disable;
+	uint32_t period; /* maxCycles = (1 << period), the number of cycles before a timeout */
+};
+
+#define AMDGPU_MAX_TIMEOUT_PARAM_LENGTH	256
 
 /*
  * Modules parameters.
  */
 extern int amdgpu_modeset;
-extern int amdgpu_vram_limit;
+extern unsigned int amdgpu_vram_limit;
 extern int amdgpu_vis_vram_limit;
 extern int amdgpu_gart_size;
 extern int amdgpu_gtt_size;
 extern int amdgpu_moverate;
-extern int amdgpu_benchmarking;
-extern int amdgpu_testing;
 extern int amdgpu_audio;
 extern int amdgpu_disp_priority;
 extern int amdgpu_hw_i2c;
 extern int amdgpu_pcie_gen2;
 extern int amdgpu_msi;
-extern int amdgpu_lockup_timeout;
+extern char amdgpu_lockup_timeout[AMDGPU_MAX_TIMEOUT_PARAM_LENGTH];
 extern int amdgpu_dpm;
 extern int amdgpu_fw_load_type;
 extern int amdgpu_aspm;
 extern int amdgpu_runtime_pm;
-extern unsigned amdgpu_ip_block_mask;
+extern uint amdgpu_ip_block_mask;
 extern int amdgpu_bapm;
 extern int amdgpu_deep_color;
 extern int amdgpu_vm_size;
@@ -100,26 +187,69 @@ extern int amdgpu_vm_fragment_size;
 extern int amdgpu_vm_fault_stop;
 extern int amdgpu_vm_debug;
 extern int amdgpu_vm_update_mode;
+extern int amdgpu_exp_hw_support;
+extern int amdgpu_dc;
 extern int amdgpu_sched_jobs;
 extern int amdgpu_sched_hw_submission;
-extern int amdgpu_no_evict;
-extern int amdgpu_direct_gma_size;
-extern unsigned amdgpu_pcie_gen_cap;
-extern unsigned amdgpu_pcie_lane_cap;
-extern unsigned amdgpu_cg_mask;
-extern unsigned amdgpu_pg_mask;
-extern unsigned amdgpu_sdma_phase_quantum;
+extern uint amdgpu_pcie_gen_cap;
+extern uint amdgpu_pcie_lane_cap;
+extern u64 amdgpu_cg_mask;
+extern uint amdgpu_pg_mask;
+extern uint amdgpu_sdma_phase_quantum;
 extern char *amdgpu_disable_cu;
 extern char *amdgpu_virtual_display;
-extern unsigned amdgpu_pp_feature_mask;
-extern int amdgpu_vram_page_split;
-extern int amdgpu_ngg;
-extern int amdgpu_prim_buf_per_se;
-extern int amdgpu_pos_buf_per_se;
-extern int amdgpu_cntl_sb_buf_per_se;
-extern int amdgpu_param_buf_per_se;
-extern int amdgpu_job_hang_limit;
+extern uint amdgpu_pp_feature_mask;
+extern uint amdgpu_force_long_training;
 extern int amdgpu_lbpw;
+extern int amdgpu_compute_multipipe;
+extern int amdgpu_gpu_recovery;
+extern int amdgpu_emu_mode;
+extern uint amdgpu_smu_memory_pool_size;
+extern int amdgpu_smu_pptable_id;
+extern uint amdgpu_dc_feature_mask;
+extern uint amdgpu_freesync_vid_mode;
+extern uint amdgpu_dc_debug_mask;
+extern uint amdgpu_dc_visual_confirm;
+extern int amdgpu_dm_abm_level;
+extern int amdgpu_backlight;
+extern int amdgpu_damage_clips;
+extern struct amdgpu_mgpu_info mgpu_info;
+extern int amdgpu_ras_enable;
+extern uint amdgpu_ras_mask;
+extern int amdgpu_bad_page_threshold;
+extern bool amdgpu_ignore_bad_page_threshold;
+extern struct amdgpu_watchdog_timer amdgpu_watchdog_timer;
+extern int amdgpu_async_gfx_ring;
+extern int amdgpu_mcbp;
+extern int amdgpu_discovery;
+extern int amdgpu_mes;
+extern int amdgpu_mes_log_enable;
+extern int amdgpu_mes_kiq;
+extern int amdgpu_uni_mes;
+extern int amdgpu_noretry;
+extern int amdgpu_force_asic_type;
+extern int amdgpu_smartshift_bias;
+extern int amdgpu_use_xgmi_p2p;
+extern int amdgpu_mtype_local;
+extern int amdgpu_enforce_isolation;
+#ifdef CONFIG_HSA_AMD
+extern int sched_policy;
+extern bool debug_evictions;
+extern bool no_system_mem_limit;
+extern int halt_if_hws_hang;
+extern uint amdgpu_svm_default_granularity;
+#else
+static const int __maybe_unused sched_policy = KFD_SCHED_POLICY_HWS;
+static const bool __maybe_unused debug_evictions; /* = false */
+static const bool __maybe_unused no_system_mem_limit;
+static const int __maybe_unused halt_if_hws_hang;
+#endif
+#ifdef CONFIG_HSA_AMD_P2P
+extern bool pcie_p2p;
+#endif
+
+extern int amdgpu_tmz;
+extern int amdgpu_reset_method;
 
 #ifdef CONFIG_DRM_AMDGPU_SI
 extern int amdgpu_si_support;
@@ -127,19 +257,33 @@ extern int amdgpu_si_support;
 #ifdef CONFIG_DRM_AMDGPU_CIK
 extern int amdgpu_cik_support;
 #endif
+extern int amdgpu_num_kcq;
+
+#define AMDGPU_VCNFW_LOG_SIZE (32 * 1024)
+#define AMDGPU_UMSCHFW_LOG_SIZE (32 * 1024)
+extern int amdgpu_vcnfw_log;
+extern int amdgpu_sg_display;
+extern int amdgpu_umsch_mm;
+extern int amdgpu_seamless;
+extern int amdgpu_umsch_mm_fwlog;
+
+extern int amdgpu_user_partt_mode;
+extern int amdgpu_agp;
+extern int amdgpu_rebar;
 
-#define AMDGPU_DEFAULT_GTT_SIZE_MB		3072ULL /* 3GB by default */
+extern int amdgpu_wbrf;
+extern int amdgpu_user_queue;
+
+#define AMDGPU_VM_MAX_NUM_CTX			4096
+#define AMDGPU_SG_THRESHOLD			(256*1024*1024)
 #define AMDGPU_WAIT_IDLE_TIMEOUT_IN_MS	        3000
 #define AMDGPU_MAX_USEC_TIMEOUT			100000	/* 100 ms */
 #define AMDGPU_FENCE_JIFFIES_TIMEOUT		(HZ / 2)
-/* AMDGPU_IB_POOL_SIZE must be a power of 2 */
-#define AMDGPU_IB_POOL_SIZE			16
 #define AMDGPU_DEBUGFS_MAX_COMPONENTS		32
 #define AMDGPUFB_CONN_LIMIT			4
 #define AMDGPU_BIOS_NUM_SCRATCH			16
 
-/* max number of IP instances */
-#define AMDGPU_MAX_SDMA_INSTANCES		2
+#define AMDGPU_VBIOS_VGA_ALLOCATION		(9 * 1024 * 1024) /* reserve 8MB for vga emulator and 1 MB for FB */
 
 /* hard reset data */
 #define AMDGPU_ASIC_RESET_DATA                  0x39d5e86b
@@ -161,26 +305,36 @@ extern int amdgpu_cik_support;
 #define AMDGPU_RESET_VCE			(1 << 13)
 #define AMDGPU_RESET_VCE1			(1 << 14)
 
-/* GFX current status */
-#define AMDGPU_GFX_NORMAL_MODE			0x00000000L
-#define AMDGPU_GFX_SAFE_MODE			0x00000001L
-#define AMDGPU_GFX_PG_DISABLED_MODE		0x00000002L
-#define AMDGPU_GFX_CG_DISABLED_MODE		0x00000004L
-#define AMDGPU_GFX_LBPW_DISABLED_MODE		0x00000008L
+/* reset mask */
+#define AMDGPU_RESET_TYPE_FULL (1 << 0) /* full adapter reset, mode1/mode2/BACO/etc. */
+#define AMDGPU_RESET_TYPE_SOFT_RESET (1 << 1) /* IP level soft reset */
+#define AMDGPU_RESET_TYPE_PER_QUEUE (1 << 2) /* per queue */
+#define AMDGPU_RESET_TYPE_PER_PIPE (1 << 3) /* per pipe */
 
 /* max cursor sizes (in pixels) */
 #define CIK_CURSOR_WIDTH 128
 #define CIK_CURSOR_HEIGHT 128
 
+/* smart shift bias level limits */
+#define AMDGPU_SMARTSHIFT_MAX_BIAS (100)
+#define AMDGPU_SMARTSHIFT_MIN_BIAS (-100)
+
+/* Extra time delay(in ms) to eliminate the influence of temperature momentary fluctuation */
+#define AMDGPU_SWCTF_EXTRA_DELAY		50
+
+struct amdgpu_xcp_mgr;
 struct amdgpu_device;
-struct amdgpu_ib;
-struct amdgpu_cs_parser;
-struct amdgpu_job;
 struct amdgpu_irq_src;
 struct amdgpu_fpriv;
+struct amdgpu_bo_va_mapping;
+struct kfd_vm_fault_info;
+struct amdgpu_hive_info;
+struct amdgpu_reset_context;
+struct amdgpu_reset_control;
 
 enum amdgpu_cp_irq {
-	AMDGPU_CP_IRQ_GFX_EOP = 0,
+	AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP = 0,
+	AMDGPU_CP_IRQ_GFX_ME0_PIPE1_EOP,
 	AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP,
 	AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE1_EOP,
 	AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE2_EOP,
@@ -193,13 +347,6 @@ enum amdgpu_cp_irq {
 	AMDGPU_CP_IRQ_LAST
 };
 
-enum amdgpu_sdma_irq {
-	AMDGPU_SDMA_IRQ_TRAP0 = 0,
-	AMDGPU_SDMA_IRQ_TRAP1,
-
-	AMDGPU_SDMA_IRQ_LAST
-};
-
 enum amdgpu_thermal_irq {
 	AMDGPU_THERMAL_IRQ_LOW_TO_HIGH = 0,
 	AMDGPU_THERMAL_IRQ_HIGH_TO_LOW,
@@ -211,18 +358,25 @@ enum amdgpu_kiq_irq {
 	AMDGPU_CP_KIQ_IRQ_DRIVER0 = 0,
 	AMDGPU_CP_KIQ_IRQ_LAST
 };
-
-int amdgpu_set_clockgating_state(struct amdgpu_device *adev,
-				  enum amd_ip_block_type block_type,
-				  enum amd_clockgating_state state);
-int amdgpu_set_powergating_state(struct amdgpu_device *adev,
-				  enum amd_ip_block_type block_type,
-				  enum amd_powergating_state state);
-void amdgpu_get_clockgating_state(struct amdgpu_device *adev, u32 *flags);
-int amdgpu_wait_for_idle(struct amdgpu_device *adev,
-			 enum amd_ip_block_type block_type);
-bool amdgpu_is_idle(struct amdgpu_device *adev,
-		    enum amd_ip_block_type block_type);
+#define MAX_KIQ_REG_WAIT       5000 /* in usecs, 5ms */
+#define MAX_KIQ_REG_BAILOUT_INTERVAL   5 /* in msecs, 5ms */
+#define MAX_KIQ_REG_TRY 1000
+
+int amdgpu_device_ip_set_clockgating_state(void *dev,
+					   enum amd_ip_block_type block_type,
+					   enum amd_clockgating_state state);
+int amdgpu_device_ip_set_powergating_state(void *dev,
+					   enum amd_ip_block_type block_type,
+					   enum amd_powergating_state state);
+void amdgpu_device_ip_get_clockgating_state(struct amdgpu_device *adev,
+					    u64 *flags);
+int amdgpu_device_ip_wait_for_idle(struct amdgpu_device *adev,
+				   enum amd_ip_block_type block_type);
+bool amdgpu_device_ip_is_valid(struct amdgpu_device *adev,
+			      enum amd_ip_block_type block_type);
+int amdgpu_ip_block_suspend(struct amdgpu_ip_block *ip_block);
+
+int amdgpu_ip_block_resume(struct amdgpu_ip_block *ip_block);
 
 #define AMDGPU_MAX_IP_NUM 16
 
@@ -245,115 +399,28 @@ struct amdgpu_ip_block_version {
 struct amdgpu_ip_block {
 	struct amdgpu_ip_block_status status;
 	const struct amdgpu_ip_block_version *version;
+	struct amdgpu_device *adev;
 };
 
-int amdgpu_ip_block_version_cmp(struct amdgpu_device *adev,
-				enum amd_ip_block_type type,
-				u32 major, u32 minor);
-
-struct amdgpu_ip_block * amdgpu_get_ip_block(struct amdgpu_device *adev,
-					     enum amd_ip_block_type type);
-
-int amdgpu_ip_block_add(struct amdgpu_device *adev,
-			const struct amdgpu_ip_block_version *ip_block_version);
-
-/* provided by hw blocks that can move/clear data.  e.g., gfx or sdma */
-struct amdgpu_buffer_funcs {
-	/* maximum bytes in a single operation */
-	uint32_t	copy_max_bytes;
-
-	/* number of dw to reserve per operation */
-	unsigned	copy_num_dw;
-
-	/* used for buffer migration */
-	void (*emit_copy_buffer)(struct amdgpu_ib *ib,
-				 /* src addr in bytes */
-				 uint64_t src_offset,
-				 /* dst addr in bytes */
-				 uint64_t dst_offset,
-				 /* number of byte to transfer */
-				 uint32_t byte_count);
-
-	/* maximum bytes in a single operation */
-	uint32_t	fill_max_bytes;
-
-	/* number of dw to reserve per operation */
-	unsigned	fill_num_dw;
-
-	/* used for buffer clearing */
-	void (*emit_fill_buffer)(struct amdgpu_ib *ib,
-				 /* value to write to memory */
-				 uint32_t src_data,
-				 /* dst addr in bytes */
-				 uint64_t dst_offset,
-				 /* number of byte to fill */
-				 uint32_t byte_count);
-};
-
-/* provided by hw blocks that can write ptes, e.g., sdma */
-struct amdgpu_vm_pte_funcs {
-	/* copy pte entries from GART */
-	void (*copy_pte)(struct amdgpu_ib *ib,
-			 uint64_t pe, uint64_t src,
-			 unsigned count);
-	/* write pte one entry at a time with addr mapping */
-	void (*write_pte)(struct amdgpu_ib *ib, uint64_t pe,
-			  uint64_t value, unsigned count,
-			  uint32_t incr);
-	/* for linear pte/pde updates without addr mapping */
-	void (*set_pte_pde)(struct amdgpu_ib *ib,
-			    uint64_t pe,
-			    uint64_t addr, unsigned count,
-			    uint32_t incr, uint64_t flags);
-};
+int amdgpu_device_ip_block_version_cmp(struct amdgpu_device *adev,
+				       enum amd_ip_block_type type,
+				       u32 major, u32 minor);
 
-/* provided by the gmc block */
-struct amdgpu_gart_funcs {
-	/* flush the vm tlb via mmio */
-	void (*flush_gpu_tlb)(struct amdgpu_device *adev,
-			      uint32_t vmid);
-	/* write pte/pde updates using the cpu */
-	int (*set_pte_pde)(struct amdgpu_device *adev,
-			   void *cpu_pt_addr, /* cpu addr of page table */
-			   uint32_t gpu_page_idx, /* pte/pde to update */
-			   uint64_t addr, /* addr to write into pte/pde */
-			   uint64_t flags); /* access flags */
-	/* enable/disable PRT support */
-	void (*set_prt)(struct amdgpu_device *adev, bool enable);
-	/* set pte flags based per asic */
-	uint64_t (*get_vm_pte_flags)(struct amdgpu_device *adev,
-				     uint32_t flags);
-	/* get the pde for a given mc addr */
-	u64 (*get_vm_pde)(struct amdgpu_device *adev, u64 addr);
-	uint32_t (*get_invalidate_req)(unsigned int vm_id);
-};
+struct amdgpu_ip_block *
+amdgpu_device_ip_get_ip_block(struct amdgpu_device *adev,
+			      enum amd_ip_block_type type);
 
-/* provided by the ih block */
-struct amdgpu_ih_funcs {
-	/* ring read/write ptr handling, called from interrupt context */
-	u32 (*get_wptr)(struct amdgpu_device *adev);
-	void (*decode_iv)(struct amdgpu_device *adev,
-			  struct amdgpu_iv_entry *entry);
-	void (*set_rptr)(struct amdgpu_device *adev);
-};
+int amdgpu_device_ip_block_add(struct amdgpu_device *adev,
+			       const struct amdgpu_ip_block_version *ip_block_version);
 
 /*
  * BIOS.
  */
 bool amdgpu_get_bios(struct amdgpu_device *adev);
 bool amdgpu_read_bios(struct amdgpu_device *adev);
-
-/*
- * Dummy page
- */
-struct amdgpu_dummy_page {
-	struct page	*page;
-	dma_addr_t	addr;
-};
-int amdgpu_dummy_page_init(struct amdgpu_device *adev);
-void amdgpu_dummy_page_fini(struct amdgpu_device *adev);
-
-
+bool amdgpu_soc15_read_bios_from_rom(struct amdgpu_device *adev,
+				     u8 *bios, u32 length_bytes);
+void amdgpu_bios_release(struct amdgpu_device *adev);
 /*
  * Clocks
  */
@@ -368,39 +435,10 @@ struct amdgpu_clock {
 	uint32_t default_mclk;
 	uint32_t default_sclk;
 	uint32_t default_dispclk;
-	uint32_t current_dispclk;
 	uint32_t dp_extclk;
 	uint32_t max_pixel_clock;
 };
 
-/*
- * GEM.
- */
-
-#define AMDGPU_GEM_DOMAIN_MAX		0x3
-#define gem_to_amdgpu_bo(gobj) container_of((gobj), struct amdgpu_bo, gem_base)
-
-void amdgpu_gem_object_free(struct drm_gem_object *obj);
-int amdgpu_gem_object_open(struct drm_gem_object *obj,
-				struct drm_file *file_priv);
-void amdgpu_gem_object_close(struct drm_gem_object *obj,
-				struct drm_file *file_priv);
-unsigned long amdgpu_gem_timeout(uint64_t timeout_ns);
-struct sg_table *amdgpu_gem_prime_get_sg_table(struct drm_gem_object *obj);
-struct drm_gem_object *
-amdgpu_gem_prime_import_sg_table(struct drm_device *dev,
-				 struct dma_buf_attachment *attach,
-				 struct sg_table *sg);
-struct dma_buf *amdgpu_gem_prime_export(struct drm_device *dev,
-					struct drm_gem_object *gobj,
-					int flags);
-int amdgpu_gem_prime_pin(struct drm_gem_object *obj);
-void amdgpu_gem_prime_unpin(struct drm_gem_object *obj);
-struct reservation_object *amdgpu_gem_prime_res_obj(struct drm_gem_object *);
-void *amdgpu_gem_prime_vmap(struct drm_gem_object *obj);
-void amdgpu_gem_prime_vunmap(struct drm_gem_object *obj, void *vaddr);
-int amdgpu_gem_debugfs_init(struct amdgpu_device *adev);
-
 /* sub-allocation manager, it has to be protected by another lock.
  * By conception this is an helper for other part of the driver
  * like the indirect buffer or semaphore, which both have their
@@ -425,213 +463,14 @@ int amdgpu_gem_debugfs_init(struct amdgpu_device *adev);
  * alignment).
  */
 
-#define AMDGPU_SA_NUM_FENCE_LISTS	32
-
 struct amdgpu_sa_manager {
-	wait_queue_head_t	wq;
-	struct amdgpu_bo	*bo;
-	struct list_head	*hole;
-	struct list_head	flist[AMDGPU_SA_NUM_FENCE_LISTS];
-	struct list_head	olist;
-	unsigned		size;
-	uint64_t		gpu_addr;
-	void			*cpu_ptr;
-	uint32_t		domain;
-	uint32_t		align;
-};
-
-/* sub-allocation buffer */
-struct amdgpu_sa_bo {
-	struct list_head		olist;
-	struct list_head		flist;
-	struct amdgpu_sa_manager	*manager;
-	unsigned			soffset;
-	unsigned			eoffset;
-	struct dma_fence	        *fence;
-};
-
-/*
- * GEM objects.
- */
-void amdgpu_gem_force_release(struct amdgpu_device *adev);
-int amdgpu_gem_object_create(struct amdgpu_device *adev, unsigned long size,
-				int alignment, u32 initial_domain,
-				u64 flags, bool kernel,
-				struct drm_gem_object **obj);
-
-int amdgpu_mode_dumb_create(struct drm_file *file_priv,
-			    struct drm_device *dev,
-			    struct drm_mode_create_dumb *args);
-int amdgpu_mode_dumb_mmap(struct drm_file *filp,
-			  struct drm_device *dev,
-			  uint32_t handle, uint64_t *offset_p);
-int amdgpu_fence_slab_init(void);
-void amdgpu_fence_slab_fini(void);
-
-/*
- * VMHUB structures, functions & helpers
- */
-struct amdgpu_vmhub {
-	uint32_t	ctx0_ptb_addr_lo32;
-	uint32_t	ctx0_ptb_addr_hi32;
-	uint32_t	vm_inv_eng0_req;
-	uint32_t	vm_inv_eng0_ack;
-	uint32_t	vm_context0_cntl;
-	uint32_t	vm_l2_pro_fault_status;
-	uint32_t	vm_l2_pro_fault_cntl;
-};
-
-/*
- * GPU MC structures, functions & helpers
- */
-struct amdgpu_mc {
-	resource_size_t		aper_size;
-	resource_size_t		aper_base;
-	resource_size_t		agp_base;
-	/* for some chips with <= 32MB we need to lie
-	 * about vram size near mc fb location */
-	u64			mc_vram_size;
-	u64			visible_vram_size;
-	u64			gart_size;
-	u64			gart_start;
-	u64			gart_end;
-	u64			vram_start;
-	u64			vram_end;
-	unsigned		vram_width;
-	u64			real_vram_size;
-	int			vram_mtrr;
-	u64                     mc_mask;
-	const struct firmware   *fw;	/* MC firmware */
-	uint32_t                fw_version;
-	struct amdgpu_irq_src	vm_fault;
-	uint32_t		vram_type;
-	uint32_t                srbm_soft_reset;
-	bool			prt_warning;
-	uint64_t		stolen_size;
-	/* apertures */
-	u64					shared_aperture_start;
-	u64					shared_aperture_end;
-	u64					private_aperture_start;
-	u64					private_aperture_end;
-	/* protects concurrent invalidation */
-	spinlock_t		invalidate_lock;
-};
-
-/*
- * GPU doorbell structures, functions & helpers
- */
-typedef enum _AMDGPU_DOORBELL_ASSIGNMENT
-{
-	AMDGPU_DOORBELL_KIQ                     = 0x000,
-	AMDGPU_DOORBELL_HIQ                     = 0x001,
-	AMDGPU_DOORBELL_DIQ                     = 0x002,
-	AMDGPU_DOORBELL_MEC_RING0               = 0x010,
-	AMDGPU_DOORBELL_MEC_RING1               = 0x011,
-	AMDGPU_DOORBELL_MEC_RING2               = 0x012,
-	AMDGPU_DOORBELL_MEC_RING3               = 0x013,
-	AMDGPU_DOORBELL_MEC_RING4               = 0x014,
-	AMDGPU_DOORBELL_MEC_RING5               = 0x015,
-	AMDGPU_DOORBELL_MEC_RING6               = 0x016,
-	AMDGPU_DOORBELL_MEC_RING7               = 0x017,
-	AMDGPU_DOORBELL_GFX_RING0               = 0x020,
-	AMDGPU_DOORBELL_sDMA_ENGINE0            = 0x1E0,
-	AMDGPU_DOORBELL_sDMA_ENGINE1            = 0x1E1,
-	AMDGPU_DOORBELL_IH                      = 0x1E8,
-	AMDGPU_DOORBELL_MAX_ASSIGNMENT          = 0x3FF,
-	AMDGPU_DOORBELL_INVALID                 = 0xFFFF
-} AMDGPU_DOORBELL_ASSIGNMENT;
-
-struct amdgpu_doorbell {
-	/* doorbell mmio */
-	resource_size_t		base;
-	resource_size_t		size;
-	u32 __iomem		*ptr;
-	u32			num_doorbells;	/* Number of doorbells actually reserved for amdgpu. */
+	struct drm_suballoc_manager	base;
+	struct amdgpu_bo		*bo;
+	uint64_t			gpu_addr;
+	void				*cpu_ptr;
 };
 
 /*
- * 64bit doorbell, offset are in QWORD, occupy 2KB doorbell space
- */
-typedef enum _AMDGPU_DOORBELL64_ASSIGNMENT
-{
-	/*
-	 * All compute related doorbells: kiq, hiq, diq, traditional compute queue, user queue, should locate in
-	 * a continues range so that programming CP_MEC_DOORBELL_RANGE_LOWER/UPPER can cover this range.
-	 *  Compute related doorbells are allocated from 0x00 to 0x8a
-	 */
-
-
-	/* kernel scheduling */
-	AMDGPU_DOORBELL64_KIQ                     = 0x00,
-
-	/* HSA interface queue and debug queue */
-	AMDGPU_DOORBELL64_HIQ                     = 0x01,
-	AMDGPU_DOORBELL64_DIQ                     = 0x02,
-
-	/* Compute engines */
-	AMDGPU_DOORBELL64_MEC_RING0               = 0x03,
-	AMDGPU_DOORBELL64_MEC_RING1               = 0x04,
-	AMDGPU_DOORBELL64_MEC_RING2               = 0x05,
-	AMDGPU_DOORBELL64_MEC_RING3               = 0x06,
-	AMDGPU_DOORBELL64_MEC_RING4               = 0x07,
-	AMDGPU_DOORBELL64_MEC_RING5               = 0x08,
-	AMDGPU_DOORBELL64_MEC_RING6               = 0x09,
-	AMDGPU_DOORBELL64_MEC_RING7               = 0x0a,
-
-	/* User queue doorbell range (128 doorbells) */
-	AMDGPU_DOORBELL64_USERQUEUE_START         = 0x0b,
-	AMDGPU_DOORBELL64_USERQUEUE_END           = 0x8a,
-
-	/* Graphics engine */
-	AMDGPU_DOORBELL64_GFX_RING0               = 0x8b,
-
-	/*
-	 * Other graphics doorbells can be allocated here: from 0x8c to 0xef
-	 * Graphics voltage island aperture 1
-	 * default non-graphics QWORD index is 0xF0 - 0xFF inclusive
-	 */
-
-	/* sDMA engines */
-	AMDGPU_DOORBELL64_sDMA_ENGINE0            = 0xF0,
-	AMDGPU_DOORBELL64_sDMA_HI_PRI_ENGINE0     = 0xF1,
-	AMDGPU_DOORBELL64_sDMA_ENGINE1            = 0xF2,
-	AMDGPU_DOORBELL64_sDMA_HI_PRI_ENGINE1     = 0xF3,
-
-	/* Interrupt handler */
-	AMDGPU_DOORBELL64_IH                      = 0xF4,  /* For legacy interrupt ring buffer */
-	AMDGPU_DOORBELL64_IH_RING1                = 0xF5,  /* For page migration request log */
-	AMDGPU_DOORBELL64_IH_RING2                = 0xF6,  /* For page migration translation/invalidation log */
-
-	/* VCN engine use 32 bits doorbell  */
-	AMDGPU_DOORBELL64_VCN0_1                  = 0xF8, /* lower 32 bits for VNC0 and upper 32 bits for VNC1 */
-	AMDGPU_DOORBELL64_VCN2_3                  = 0xF9,
-	AMDGPU_DOORBELL64_VCN4_5                  = 0xFA,
-	AMDGPU_DOORBELL64_VCN6_7                  = 0xFB,
-
-	/* overlap the doorbell assignment with VCN as they are  mutually exclusive
-	 * VCE engine's doorbell is 32 bit and two VCE ring share one QWORD
-	 */
-	AMDGPU_DOORBELL64_UVD_RING0_1             = 0xF8,
-	AMDGPU_DOORBELL64_UVD_RING2_3             = 0xF9,
-	AMDGPU_DOORBELL64_UVD_RING4_5             = 0xFA,
-	AMDGPU_DOORBELL64_UVD_RING6_7             = 0xFB,
-
-	AMDGPU_DOORBELL64_VCE_RING0_1             = 0xFC,
-	AMDGPU_DOORBELL64_VCE_RING2_3             = 0xFD,
-	AMDGPU_DOORBELL64_VCE_RING4_5             = 0xFE,
-	AMDGPU_DOORBELL64_VCE_RING6_7             = 0xFF,
-
-	AMDGPU_DOORBELL64_MAX_ASSIGNMENT          = 0xFF,
-	AMDGPU_DOORBELL64_INVALID                 = 0xFFFF
-} AMDGPU_DOORBELL64_ASSIGNMENT;
-
-
-void amdgpu_doorbell_get_kfd_info(struct amdgpu_device *adev,
-				phys_addr_t *aperture_base,
-				size_t *aperture_size,
-				size_t *start_offset);
-
-/*
  * IRQS.
  */
 
@@ -644,104 +483,12 @@ struct amdgpu_flip_work {
 	uint64_t			base;
 	struct drm_pending_vblank_event *event;
 	struct amdgpu_bo		*old_abo;
-	struct dma_fence		*excl;
 	unsigned			shared_count;
 	struct dma_fence		**shared;
 	struct dma_fence_cb		cb;
 	bool				async;
 };
 
-
-/*
- * CP & rings.
- */
-
-struct amdgpu_ib {
-	struct amdgpu_sa_bo		*sa_bo;
-	uint32_t			length_dw;
-	uint64_t			gpu_addr;
-	uint32_t			*ptr;
-	uint32_t			flags;
-};
-
-extern const struct amd_sched_backend_ops amdgpu_sched_ops;
-
-int amdgpu_job_alloc(struct amdgpu_device *adev, unsigned num_ibs,
-		     struct amdgpu_job **job, struct amdgpu_vm *vm);
-int amdgpu_job_alloc_with_ib(struct amdgpu_device *adev, unsigned size,
-			     struct amdgpu_job **job);
-
-void amdgpu_job_free_resources(struct amdgpu_job *job);
-void amdgpu_job_free(struct amdgpu_job *job);
-int amdgpu_job_submit(struct amdgpu_job *job, struct amdgpu_ring *ring,
-		      struct amd_sched_entity *entity, void *owner,
-		      struct dma_fence **f);
-
-/*
- * Queue manager
- */
-struct amdgpu_queue_mapper {
-	int 		hw_ip;
-	struct mutex	lock;
-	/* protected by lock */
-	struct amdgpu_ring *queue_map[AMDGPU_MAX_RINGS];
-};
-
-struct amdgpu_queue_mgr {
-	struct amdgpu_queue_mapper mapper[AMDGPU_MAX_IP_NUM];
-};
-
-int amdgpu_queue_mgr_init(struct amdgpu_device *adev,
-			  struct amdgpu_queue_mgr *mgr);
-int amdgpu_queue_mgr_fini(struct amdgpu_device *adev,
-			  struct amdgpu_queue_mgr *mgr);
-int amdgpu_queue_mgr_map(struct amdgpu_device *adev,
-			 struct amdgpu_queue_mgr *mgr,
-			 int hw_ip, int instance, int ring,
-			 struct amdgpu_ring **out_ring);
-
-/*
- * context related structures
- */
-
-struct amdgpu_ctx_ring {
-	uint64_t		sequence;
-	struct dma_fence	**fences;
-	struct amd_sched_entity	entity;
-};
-
-struct amdgpu_ctx {
-	struct kref		refcount;
-	struct amdgpu_device    *adev;
-	struct amdgpu_queue_mgr queue_mgr;
-	unsigned		reset_counter;
-	spinlock_t		ring_lock;
-	struct dma_fence	**fences;
-	struct amdgpu_ctx_ring	rings[AMDGPU_MAX_RINGS];
-	bool preamble_presented;
-};
-
-struct amdgpu_ctx_mgr {
-	struct amdgpu_device	*adev;
-	struct mutex		lock;
-	/* protected by lock */
-	struct idr		ctx_handles;
-};
-
-struct amdgpu_ctx *amdgpu_ctx_get(struct amdgpu_fpriv *fpriv, uint32_t id);
-int amdgpu_ctx_put(struct amdgpu_ctx *ctx);
-
-uint64_t amdgpu_ctx_add_fence(struct amdgpu_ctx *ctx, struct amdgpu_ring *ring,
-			      struct dma_fence *fence);
-struct dma_fence *amdgpu_ctx_get_fence(struct amdgpu_ctx *ctx,
-				   struct amdgpu_ring *ring, uint64_t seq);
-
-int amdgpu_ctx_ioctl(struct drm_device *dev, void *data,
-		     struct drm_file *filp);
-
-void amdgpu_ctx_mgr_init(struct amdgpu_ctx_mgr *mgr);
-void amdgpu_ctx_mgr_fini(struct amdgpu_ctx_mgr *mgr);
-
 /*
  * file private structure
  */
@@ -750,508 +497,92 @@ struct amdgpu_fpriv {
 	struct amdgpu_vm	vm;
 	struct amdgpu_bo_va	*prt_va;
 	struct amdgpu_bo_va	*csa_va;
+	struct amdgpu_bo_va	*seq64_va;
 	struct mutex		bo_list_lock;
 	struct idr		bo_list_handles;
 	struct amdgpu_ctx_mgr	ctx_mgr;
-	u32			vram_lost_counter;
-};
-
-/*
- * residency list
- */
-struct amdgpu_bo_list_entry {
-	struct amdgpu_bo		*robj;
-	struct ttm_validate_buffer	tv;
-	struct amdgpu_bo_va		*bo_va;
-	uint32_t			priority;
-	struct page			**user_pages;
-	int				user_invalidated;
-};
-
-struct amdgpu_bo_list {
-	struct mutex lock;
-	struct rcu_head rhead;
-	struct kref refcount;
-	struct amdgpu_bo *gds_obj;
-	struct amdgpu_bo *gws_obj;
-	struct amdgpu_bo *oa_obj;
-	unsigned first_userptr;
-	unsigned num_entries;
-	struct amdgpu_bo_list_entry *array;
-};
-
-struct amdgpu_bo_list *
-amdgpu_bo_list_get(struct amdgpu_fpriv *fpriv, int id);
-void amdgpu_bo_list_get_list(struct amdgpu_bo_list *list,
-			     struct list_head *validated);
-void amdgpu_bo_list_put(struct amdgpu_bo_list *list);
-void amdgpu_bo_list_free(struct amdgpu_bo_list *list);
-
-/*
- * GFX stuff
- */
-#include "clearstate_defs.h"
-
-struct amdgpu_rlc_funcs {
-	void (*enter_safe_mode)(struct amdgpu_device *adev);
-	void (*exit_safe_mode)(struct amdgpu_device *adev);
-};
-
-struct amdgpu_rlc {
-	/* for power gating */
-	struct amdgpu_bo	*save_restore_obj;
-	uint64_t		save_restore_gpu_addr;
-	volatile uint32_t	*sr_ptr;
-	const u32               *reg_list;
-	u32                     reg_list_size;
-	/* for clear state */
-	struct amdgpu_bo	*clear_state_obj;
-	uint64_t		clear_state_gpu_addr;
-	volatile uint32_t	*cs_ptr;
-	const struct cs_section_def   *cs_data;
-	u32                     clear_state_size;
-	/* for cp tables */
-	struct amdgpu_bo	*cp_table_obj;
-	uint64_t		cp_table_gpu_addr;
-	volatile uint32_t	*cp_table_ptr;
-	u32                     cp_table_size;
-
-	/* safe mode for updating CG/PG state */
-	bool in_safe_mode;
-	const struct amdgpu_rlc_funcs *funcs;
-
-	/* for firmware data */
-	u32 save_and_restore_offset;
-	u32 clear_state_descriptor_offset;
-	u32 avail_scratch_ram_locations;
-	u32 reg_restore_list_size;
-	u32 reg_list_format_start;
-	u32 reg_list_format_separate_start;
-	u32 starting_offsets_start;
-	u32 reg_list_format_size_bytes;
-	u32 reg_list_size_bytes;
-
-	u32 *register_list_format;
-	u32 *register_restore;
-};
-
-#define AMDGPU_MAX_COMPUTE_QUEUES KGD_MAX_QUEUES
-
-struct amdgpu_mec {
-	struct amdgpu_bo	*hpd_eop_obj;
-	u64			hpd_eop_gpu_addr;
-	struct amdgpu_bo	*mec_fw_obj;
-	u64			mec_fw_gpu_addr;
-	u32 num_mec;
-	u32 num_pipe_per_mec;
-	u32 num_queue_per_pipe;
-	void			*mqd_backup[AMDGPU_MAX_COMPUTE_RINGS + 1];
-
-	/* These are the resources for which amdgpu takes ownership */
-	DECLARE_BITMAP(queue_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
-};
-
-struct amdgpu_kiq {
-	u64			eop_gpu_addr;
-	struct amdgpu_bo	*eop_obj;
-	struct mutex		ring_mutex;
-	struct amdgpu_ring	ring;
-	struct amdgpu_irq_src	irq;
-};
-
-/*
- * GPU scratch registers structures, functions & helpers
- */
-struct amdgpu_scratch {
-	unsigned		num_reg;
-	uint32_t                reg_base;
-	uint32_t		free_mask;
-};
-
-/*
- * GFX configurations
- */
-#define AMDGPU_GFX_MAX_SE 4
-#define AMDGPU_GFX_MAX_SH_PER_SE 2
-
-struct amdgpu_rb_config {
-	uint32_t rb_backend_disable;
-	uint32_t user_rb_backend_disable;
-	uint32_t raster_config;
-	uint32_t raster_config_1;
-};
-
-struct gb_addr_config {
-	uint16_t pipe_interleave_size;
-	uint8_t num_pipes;
-	uint8_t max_compress_frags;
-	uint8_t num_banks;
-	uint8_t num_se;
-	uint8_t num_rb_per_se;
-};
-
-struct amdgpu_gfx_config {
-	unsigned max_shader_engines;
-	unsigned max_tile_pipes;
-	unsigned max_cu_per_sh;
-	unsigned max_sh_per_se;
-	unsigned max_backends_per_se;
-	unsigned max_texture_channel_caches;
-	unsigned max_gprs;
-	unsigned max_gs_threads;
-	unsigned max_hw_contexts;
-	unsigned sc_prim_fifo_size_frontend;
-	unsigned sc_prim_fifo_size_backend;
-	unsigned sc_hiz_tile_fifo_size;
-	unsigned sc_earlyz_tile_fifo_size;
-
-	unsigned num_tile_pipes;
-	unsigned backend_enable_mask;
-	unsigned mem_max_burst_length_bytes;
-	unsigned mem_row_size_in_kb;
-	unsigned shader_engine_tile_size;
-	unsigned num_gpus;
-	unsigned multi_gpu_tile_size;
-	unsigned mc_arb_ramcfg;
-	unsigned gb_addr_config;
-	unsigned num_rbs;
-	unsigned gs_vgt_table_depth;
-	unsigned gs_prim_buffer_depth;
-
-	uint32_t tile_mode_array[32];
-	uint32_t macrotile_mode_array[16];
-
-	struct gb_addr_config gb_addr_config_fields;
-	struct amdgpu_rb_config rb_config[AMDGPU_GFX_MAX_SE][AMDGPU_GFX_MAX_SH_PER_SE];
-
-	/* gfx configure feature */
-	uint32_t double_offchip_lds_buf;
-};
-
-struct amdgpu_cu_info {
-	uint32_t max_waves_per_simd;
-	uint32_t wave_front_size;
-	uint32_t max_scratch_slots_per_cu;
-	uint32_t lds_size;
-
-	/* total active CU number */
-	uint32_t number;
-	uint32_t ao_cu_mask;
-	uint32_t ao_cu_bitmap[4][4];
-	uint32_t bitmap[4][4];
-};
-
-struct amdgpu_gfx_funcs {
-	/* get the gpu clock counter */
-	uint64_t (*get_gpu_clock_counter)(struct amdgpu_device *adev);
-	void (*select_se_sh)(struct amdgpu_device *adev, u32 se_num, u32 sh_num, u32 instance);
-	void (*read_wave_data)(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t *dst, int *no_fields);
-	void (*read_wave_vgprs)(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t thread, uint32_t start, uint32_t size, uint32_t *dst);
-	void (*read_wave_sgprs)(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t start, uint32_t size, uint32_t *dst);
-};
-
-struct amdgpu_ngg_buf {
-	struct amdgpu_bo	*bo;
-	uint64_t		gpu_addr;
-	uint32_t		size;
-	uint32_t		bo_size;
-};
-
-enum {
-	NGG_PRIM = 0,
-	NGG_POS,
-	NGG_CNTL,
-	NGG_PARAM,
-	NGG_BUF_MAX
-};
-
-struct amdgpu_ngg {
-	struct amdgpu_ngg_buf	buf[NGG_BUF_MAX];
-	uint32_t		gds_reserve_addr;
-	uint32_t		gds_reserve_size;
-	bool			init;
-};
+	struct amdgpu_userq_mgr	userq_mgr;
 
-struct amdgpu_gfx {
-	struct mutex			gpu_clock_mutex;
-	struct amdgpu_gfx_config	config;
-	struct amdgpu_rlc		rlc;
-	struct amdgpu_mec		mec;
-	struct amdgpu_kiq		kiq;
-	struct amdgpu_scratch		scratch;
-	const struct firmware		*me_fw;	/* ME firmware */
-	uint32_t			me_fw_version;
-	const struct firmware		*pfp_fw; /* PFP firmware */
-	uint32_t			pfp_fw_version;
-	const struct firmware		*ce_fw;	/* CE firmware */
-	uint32_t			ce_fw_version;
-	const struct firmware		*rlc_fw; /* RLC firmware */
-	uint32_t			rlc_fw_version;
-	const struct firmware		*mec_fw; /* MEC firmware */
-	uint32_t			mec_fw_version;
-	const struct firmware		*mec2_fw; /* MEC2 firmware */
-	uint32_t			mec2_fw_version;
-	uint32_t			me_feature_version;
-	uint32_t			ce_feature_version;
-	uint32_t			pfp_feature_version;
-	uint32_t			rlc_feature_version;
-	uint32_t			mec_feature_version;
-	uint32_t			mec2_feature_version;
-	struct amdgpu_ring		gfx_ring[AMDGPU_MAX_GFX_RINGS];
-	unsigned			num_gfx_rings;
-	struct amdgpu_ring		compute_ring[AMDGPU_MAX_COMPUTE_RINGS];
-	unsigned			num_compute_rings;
-	struct amdgpu_irq_src		eop_irq;
-	struct amdgpu_irq_src		priv_reg_irq;
-	struct amdgpu_irq_src		priv_inst_irq;
-	/* gfx status */
-	uint32_t			gfx_current_status;
-	/* ce ram size*/
-	unsigned			ce_ram_size;
-	struct amdgpu_cu_info		cu_info;
-	const struct amdgpu_gfx_funcs	*funcs;
-
-	/* reset mask */
-	uint32_t                        grbm_soft_reset;
-	uint32_t                        srbm_soft_reset;
-	bool                            in_reset;
-	/* s3/s4 mask */
-	bool                            in_suspend;
-	/* NGG */
-	struct amdgpu_ngg		ngg;
-};
-
-int amdgpu_ib_get(struct amdgpu_device *adev, struct amdgpu_vm *vm,
-		  unsigned size, struct amdgpu_ib *ib);
-void amdgpu_ib_free(struct amdgpu_device *adev, struct amdgpu_ib *ib,
-		    struct dma_fence *f);
-int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
-		       struct amdgpu_ib *ibs, struct amdgpu_job *job,
-		       struct dma_fence **f);
-int amdgpu_ib_pool_init(struct amdgpu_device *adev);
-void amdgpu_ib_pool_fini(struct amdgpu_device *adev);
-int amdgpu_ib_ring_tests(struct amdgpu_device *adev);
-
-/*
- * CS.
- */
-struct amdgpu_cs_chunk {
-	uint32_t		chunk_id;
-	uint32_t		length_dw;
-	void			*kdata;
-};
-
-struct amdgpu_cs_parser {
-	struct amdgpu_device	*adev;
-	struct drm_file		*filp;
-	struct amdgpu_ctx	*ctx;
-
-	/* chunks */
-	unsigned		nchunks;
-	struct amdgpu_cs_chunk	*chunks;
-
-	/* scheduler job object */
-	struct amdgpu_job	*job;
-
-	/* buffer objects */
-	struct ww_acquire_ctx		ticket;
-	struct amdgpu_bo_list		*bo_list;
-	struct amdgpu_bo_list_entry	vm_pd;
-	struct list_head		validated;
-	struct dma_fence		*fence;
-	uint64_t			bytes_moved_threshold;
-	uint64_t			bytes_moved_vis_threshold;
-	uint64_t			bytes_moved;
-	uint64_t			bytes_moved_vis;
-	struct amdgpu_bo_list_entry	*evictable;
-
-	/* user fence */
-	struct amdgpu_bo_list_entry	uf_entry;
-
-	unsigned num_post_dep_syncobjs;
-	struct drm_syncobj **post_dep_syncobjs;
-};
-
-#define AMDGPU_PREAMBLE_IB_PRESENT          (1 << 0) /* bit set means command submit involves a preamble IB */
-#define AMDGPU_PREAMBLE_IB_PRESENT_FIRST    (1 << 1) /* bit set means preamble IB is first presented in belonging context */
-#define AMDGPU_HAVE_CTX_SWITCH              (1 << 2) /* bit set means context switch occured */
-
-struct amdgpu_job {
-	struct amd_sched_job    base;
-	struct amdgpu_device	*adev;
-	struct amdgpu_vm	*vm;
-	struct amdgpu_ring	*ring;
-	struct amdgpu_sync	sync;
-	struct amdgpu_sync	dep_sync;
-	struct amdgpu_sync	sched_sync;
-	struct amdgpu_ib	*ibs;
-	struct dma_fence	*fence; /* the hw fence */
-	uint32_t		preamble_status;
-	uint32_t		num_ibs;
-	void			*owner;
-	uint64_t		fence_ctx; /* the fence_context this job uses */
-	bool                    vm_needs_flush;
-	unsigned		vm_id;
-	uint64_t		vm_pd_addr;
-	uint32_t		gds_base, gds_size;
-	uint32_t		gws_base, gws_size;
-	uint32_t		oa_base, oa_size;
-
-	/* user fence handling */
-	uint64_t		uf_addr;
-	uint64_t		uf_sequence;
+	/* Eviction fence infra */
+	struct amdgpu_eviction_fence_mgr evf_mgr;
 
+	/** GPU partition selection */
+	uint32_t		xcp_id;
 };
-#define to_amdgpu_job(sched_job)		\
-		container_of((sched_job), struct amdgpu_job, base)
-
-static inline u32 amdgpu_get_ib_value(struct amdgpu_cs_parser *p,
-				      uint32_t ib_idx, int idx)
-{
-	return p->job->ibs[ib_idx].ptr[idx];
-}
 
-static inline void amdgpu_set_ib_value(struct amdgpu_cs_parser *p,
-				       uint32_t ib_idx, int idx,
-				       uint32_t value)
-{
-	p->job->ibs[ib_idx].ptr[idx] = value;
-}
+int amdgpu_file_to_fpriv(struct file *filp, struct amdgpu_fpriv **fpriv);
 
 /*
  * Writeback
  */
 #define AMDGPU_MAX_WB 1024	/* Reserve at most 1024 WB slots for amdgpu-owned rings. */
 
+/**
+ * amdgpu_wb - This struct is used for small GPU memory allocation.
+ *
+ * This struct is used to allocate a small amount of GPU memory that can be
+ * used to shadow certain states into the memory. This is especially useful for
+ * providing easy CPU access to some states without requiring register access
+ * (e.g., if some block is power gated, reading register may be problematic).
+ *
+ * Note: the term writeback was initially used because many of the amdgpu
+ * components had some level of writeback memory, and this struct initially
+ * described those components.
+ */
 struct amdgpu_wb {
+
+	/**
+	 * @wb_obj:
+	 *
+	 * Buffer Object used for the writeback memory.
+	 */
 	struct amdgpu_bo	*wb_obj;
-	volatile uint32_t	*wb;
-	uint64_t		gpu_addr;
-	u32			num_wb;	/* Number of wb slots actually reserved for amdgpu. */
-	unsigned long		used[DIV_ROUND_UP(AMDGPU_MAX_WB, BITS_PER_LONG)];
-};
 
-int amdgpu_wb_get(struct amdgpu_device *adev, u32 *wb);
-void amdgpu_wb_free(struct amdgpu_device *adev, u32 wb);
+	/**
+	 * @wb:
+	 *
+	 * Pointer to the first writeback slot. In terms of CPU address
+	 * this value can be accessed directly by using the offset as an index.
+	 * For the GPU address, it is necessary to use gpu_addr and the offset.
+	 */
+	uint32_t		*wb;
 
-void amdgpu_get_pcie_info(struct amdgpu_device *adev);
+	/**
+	 * @gpu_addr:
+	 *
+	 * Writeback base address in the GPU.
+	 */
+	uint64_t		gpu_addr;
 
-/*
- * SDMA
- */
-struct amdgpu_sdma_instance {
-	/* SDMA firmware */
-	const struct firmware	*fw;
-	uint32_t		fw_version;
-	uint32_t		feature_version;
-
-	struct amdgpu_ring	ring;
-	bool			burst_nop;
-};
+	/**
+	 * @num_wb:
+	 *
+	 * Number of writeback slots reserved for amdgpu.
+	 */
+	u32			num_wb;
 
-struct amdgpu_sdma {
-	struct amdgpu_sdma_instance instance[AMDGPU_MAX_SDMA_INSTANCES];
-#ifdef CONFIG_DRM_AMDGPU_SI
-	//SI DMA has a difference trap irq number for the second engine
-	struct amdgpu_irq_src	trap_irq_1;
-#endif
-	struct amdgpu_irq_src	trap_irq;
-	struct amdgpu_irq_src	illegal_inst_irq;
-	int			num_instances;
-	uint32_t                    srbm_soft_reset;
-};
+	/**
+	 * @used:
+	 *
+	 * Track the writeback slot already used.
+	 */
+	unsigned long		used[DIV_ROUND_UP(AMDGPU_MAX_WB, BITS_PER_LONG)];
 
-/*
- * Firmware
- */
-enum amdgpu_firmware_load_type {
-	AMDGPU_FW_LOAD_DIRECT = 0,
-	AMDGPU_FW_LOAD_SMU,
-	AMDGPU_FW_LOAD_PSP,
+	/**
+	 * @lock:
+	 *
+	 * Protects read and write of the used field array.
+	 */
+	spinlock_t		lock;
 };
 
-struct amdgpu_firmware {
-	struct amdgpu_firmware_info ucode[AMDGPU_UCODE_ID_MAXIMUM];
-	enum amdgpu_firmware_load_type load_type;
-	struct amdgpu_bo *fw_buf;
-	unsigned int fw_size;
-	unsigned int max_ucodes;
-	/* firmwares are loaded by psp instead of smu from vega10 */
-	const struct amdgpu_psp_funcs *funcs;
-	struct amdgpu_bo *rbuf;
-	struct mutex mutex;
-
-	/* gpu info firmware data pointer */
-	const struct firmware *gpu_info_fw;
-};
+int amdgpu_device_wb_get(struct amdgpu_device *adev, u32 *wb);
+void amdgpu_device_wb_free(struct amdgpu_device *adev, u32 wb);
 
 /*
  * Benchmarking
  */
-void amdgpu_benchmark(struct amdgpu_device *adev, int test_number);
-
-
-/*
- * Testing
- */
-void amdgpu_test_moves(struct amdgpu_device *adev);
-
-/*
- * MMU Notifier
- */
-#if defined(CONFIG_MMU_NOTIFIER)
-int amdgpu_mn_register(struct amdgpu_bo *bo, unsigned long addr);
-void amdgpu_mn_unregister(struct amdgpu_bo *bo);
-#else
-static inline int amdgpu_mn_register(struct amdgpu_bo *bo, unsigned long addr)
-{
-	return -ENODEV;
-}
-static inline void amdgpu_mn_unregister(struct amdgpu_bo *bo) {}
-#endif
-
-/*
- * Debugfs
- */
-struct amdgpu_debugfs {
-	const struct drm_info_list	*files;
-	unsigned		num_files;
-};
-
-int amdgpu_debugfs_add_files(struct amdgpu_device *adev,
-			     const struct drm_info_list *files,
-			     unsigned nfiles);
-int amdgpu_debugfs_fence_init(struct amdgpu_device *adev);
-
-#if defined(CONFIG_DEBUG_FS)
-int amdgpu_debugfs_init(struct drm_minor *minor);
-#endif
-
-int amdgpu_debugfs_firmware_init(struct amdgpu_device *adev);
-
-/*
- * amdgpu smumgr functions
- */
-struct amdgpu_smumgr_funcs {
-	int (*check_fw_load_finish)(struct amdgpu_device *adev, uint32_t fwtype);
-	int (*request_smu_load_fw)(struct amdgpu_device *adev);
-	int (*request_smu_specific_fw)(struct amdgpu_device *adev, uint32_t fwtype);
-};
-
-/*
- * amdgpu smumgr
- */
-struct amdgpu_smumgr {
-	struct amdgpu_bo *toc_buf;
-	struct amdgpu_bo *smu_buf;
-	/* asic priv smu data */
-	void *priv;
-	spinlock_t smu_lock;
-	/* smumgr functions */
-	const struct amdgpu_smumgr_funcs *smumgr_funcs;
-	/* ucode loading complete flag */
-	uint32_t fw_flags;
-};
+int amdgpu_benchmark(struct amdgpu_device *adev, int test_number);
 
 /*
  * ASIC specific register table accessible by UMD
@@ -1261,6 +592,64 @@ struct amdgpu_allowed_register_entry {
 	bool grbm_indexed;
 };
 
+/**
+ * enum amd_reset_method - Methods for resetting AMD GPU devices
+ *
+ * @AMD_RESET_METHOD_NONE: The device will not be reset.
+ * @AMD_RESET_LEGACY: Method reserved for SI, CIK and VI ASICs.
+ * @AMD_RESET_MODE0: Reset the entire ASIC. Not currently available for the
+ *                   any device.
+ * @AMD_RESET_MODE1: Resets all IP blocks on the ASIC (SDMA, GFX, VCN, etc.)
+ *                   individually. Suitable only for some discrete GPU, not
+ *                   available for all ASICs.
+ * @AMD_RESET_MODE2: Resets a lesser level of IPs compared to MODE1. Which IPs
+ *                   are reset depends on the ASIC. Notably doesn't reset IPs
+ *                   shared with the CPU on APUs or the memory controllers (so
+ *                   VRAM is not lost). Not available on all ASICs.
+ * @AMD_RESET_LINK: Triggers SW-UP link reset on other GPUs
+ * @AMD_RESET_BACO: BACO (Bus Alive, Chip Off) method powers off and on the card
+ *                  but without powering off the PCI bus. Suitable only for
+ *                  discrete GPUs.
+ * @AMD_RESET_PCI: Does a full bus reset using core Linux subsystem PCI reset
+ *                 and does a secondary bus reset or FLR, depending on what the
+ *                 underlying hardware supports.
+ *
+ * Methods available for AMD GPU driver for resetting the device. Not all
+ * methods are suitable for every device. User can override the method using
+ * module parameter `reset_method`.
+ */
+enum amd_reset_method {
+	AMD_RESET_METHOD_NONE = -1,
+	AMD_RESET_METHOD_LEGACY = 0,
+	AMD_RESET_METHOD_MODE0,
+	AMD_RESET_METHOD_MODE1,
+	AMD_RESET_METHOD_MODE2,
+	AMD_RESET_METHOD_LINK,
+	AMD_RESET_METHOD_BACO,
+	AMD_RESET_METHOD_PCI,
+	AMD_RESET_METHOD_ON_INIT,
+};
+
+struct amdgpu_video_codec_info {
+	u32 codec_type;
+	u32 max_width;
+	u32 max_height;
+	u32 max_pixels_per_frame;
+	u32 max_level;
+};
+
+#define codec_info_build(type, width, height, level) \
+			 .codec_type = type,\
+			 .max_width = width,\
+			 .max_height = height,\
+			 .max_pixels_per_frame = height * width,\
+			 .max_level = level,
+
+struct amdgpu_video_codecs {
+	const u32 codec_count;
+	const struct amdgpu_video_codec_info *codec_array;
+};
+
 /*
  * ASIC specific functions.
  */
@@ -1272,6 +661,7 @@ struct amdgpu_asic_funcs {
 			     u32 sh_num, u32 reg_offset, u32 *value);
 	void (*set_vga_state)(struct amdgpu_device *adev, bool state);
 	int (*reset)(struct amdgpu_device *adev);
+	enum amd_reset_method (*reset_method)(struct amdgpu_device *adev);
 	/* get the reference clock */
 	u32 (*get_xclk)(struct amdgpu_device *adev);
 	/* MM block clocks */
@@ -1282,95 +672,60 @@ struct amdgpu_asic_funcs {
 	void (*set_pcie_lanes)(struct amdgpu_device *adev, int lanes);
 	/* get config memsize register */
 	u32 (*get_config_memsize)(struct amdgpu_device *adev);
+	/* flush hdp write queue */
+	void (*flush_hdp)(struct amdgpu_device *adev, struct amdgpu_ring *ring);
+	/* invalidate hdp read cache */
+	void (*invalidate_hdp)(struct amdgpu_device *adev,
+			       struct amdgpu_ring *ring);
+	/* check if the asic needs a full reset of if soft reset will work */
+	bool (*need_full_reset)(struct amdgpu_device *adev);
+	/* initialize doorbell layout for specific asic*/
+	void (*init_doorbell_index)(struct amdgpu_device *adev);
+	/* PCIe bandwidth usage */
+	void (*get_pcie_usage)(struct amdgpu_device *adev, uint64_t *count0,
+			       uint64_t *count1);
+	/* do we need to reset the asic at init time (e.g., kexec) */
+	bool (*need_reset_on_init)(struct amdgpu_device *adev);
+	/* PCIe replay counter */
+	uint64_t (*get_pcie_replay_count)(struct amdgpu_device *adev);
+	/* device supports BACO */
+	int (*supports_baco)(struct amdgpu_device *adev);
+	/* pre asic_init quirks */
+	void (*pre_asic_init)(struct amdgpu_device *adev);
+	/* enter/exit umd stable pstate */
+	int (*update_umd_stable_pstate)(struct amdgpu_device *adev, bool enter);
+	/* query video codecs */
+	int (*query_video_codecs)(struct amdgpu_device *adev, bool encode,
+				  const struct amdgpu_video_codecs **codecs);
+	/* encode "> 32bits" smn addressing */
+	u64 (*encode_ext_smn_addressing)(int ext_id);
+
+	ssize_t (*get_reg_state)(struct amdgpu_device *adev,
+				 enum amdgpu_reg_state reg_state, void *buf,
+				 size_t max_size);
 };
 
 /*
  * IOCTL.
  */
-int amdgpu_gem_create_ioctl(struct drm_device *dev, void *data,
-			    struct drm_file *filp);
 int amdgpu_bo_list_ioctl(struct drm_device *dev, void *data,
 				struct drm_file *filp);
 
-int amdgpu_gem_info_ioctl(struct drm_device *dev, void *data,
-			  struct drm_file *filp);
-int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
-			struct drm_file *filp);
-int amdgpu_gem_mmap_ioctl(struct drm_device *dev, void *data,
-			  struct drm_file *filp);
-int amdgpu_gem_wait_idle_ioctl(struct drm_device *dev, void *data,
-			      struct drm_file *filp);
-int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
-			  struct drm_file *filp);
-int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,
-			struct drm_file *filp);
 int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp);
+int amdgpu_cs_fence_to_handle_ioctl(struct drm_device *dev, void *data,
+				    struct drm_file *filp);
 int amdgpu_cs_wait_ioctl(struct drm_device *dev, void *data, struct drm_file *filp);
 int amdgpu_cs_wait_fences_ioctl(struct drm_device *dev, void *data,
 				struct drm_file *filp);
 
-int amdgpu_gem_metadata_ioctl(struct drm_device *dev, void *data,
-				struct drm_file *filp);
-
 /* VRAM scratch page for HDP bug, default vram page */
-struct amdgpu_vram_scratch {
+struct amdgpu_mem_scratch {
 	struct amdgpu_bo		*robj;
-	volatile uint32_t		*ptr;
+	uint32_t			*ptr;
 	u64				gpu_addr;
 };
 
 /*
- * ACPI
- */
-struct amdgpu_atif_notification_cfg {
-	bool enabled;
-	int command_code;
-};
-
-struct amdgpu_atif_notifications {
-	bool display_switch;
-	bool expansion_mode_change;
-	bool thermal_state;
-	bool forced_power_state;
-	bool system_power_state;
-	bool display_conf_change;
-	bool px_gfx_switch;
-	bool brightness_change;
-	bool dgpu_display_event;
-};
-
-struct amdgpu_atif_functions {
-	bool system_params;
-	bool sbios_requests;
-	bool select_active_disp;
-	bool lid_state;
-	bool get_tv_standard;
-	bool set_tv_standard;
-	bool get_panel_expansion_mode;
-	bool set_panel_expansion_mode;
-	bool temperature_change;
-	bool graphics_device_types;
-};
-
-struct amdgpu_atif {
-	struct amdgpu_atif_notifications notifications;
-	struct amdgpu_atif_functions functions;
-	struct amdgpu_atif_notification_cfg notification_cfg;
-	struct amdgpu_encoder *encoder_for_bl;
-};
-
-struct amdgpu_atcs_functions {
-	bool get_ext_state;
-	bool pcie_perf_req;
-	bool pcie_dev_rdy;
-	bool pcie_bus_width;
-};
-
-struct amdgpu_atcs {
-	struct amdgpu_atcs_functions functions;
-};
-
-/*
  * CGS
  */
 struct cgs_device *amdgpu_cgs_create_device(struct amdgpu_device *adev);
@@ -1382,51 +737,254 @@ void amdgpu_cgs_destroy_device(struct cgs_device *cgs_device);
 typedef uint32_t (*amdgpu_rreg_t)(struct amdgpu_device*, uint32_t);
 typedef void (*amdgpu_wreg_t)(struct amdgpu_device*, uint32_t, uint32_t);
 
+typedef uint32_t (*amdgpu_rreg_ext_t)(struct amdgpu_device*, uint64_t);
+typedef void (*amdgpu_wreg_ext_t)(struct amdgpu_device*, uint64_t, uint32_t);
+
+typedef uint64_t (*amdgpu_rreg64_t)(struct amdgpu_device*, uint32_t);
+typedef void (*amdgpu_wreg64_t)(struct amdgpu_device*, uint32_t, uint64_t);
+
+typedef uint64_t (*amdgpu_rreg64_ext_t)(struct amdgpu_device*, uint64_t);
+typedef void (*amdgpu_wreg64_ext_t)(struct amdgpu_device*, uint64_t, uint64_t);
+
 typedef uint32_t (*amdgpu_block_rreg_t)(struct amdgpu_device*, uint32_t, uint32_t);
 typedef void (*amdgpu_block_wreg_t)(struct amdgpu_device*, uint32_t, uint32_t, uint32_t);
 
+struct amdgpu_mmio_remap {
+	u32 reg_offset;
+	resource_size_t bus_addr;
+	struct amdgpu_bo *bo;
+};
+
+/* Define the HW IP blocks will be used in driver , add more if necessary */
+enum amd_hw_ip_block_type {
+	GC_HWIP = 1,
+	HDP_HWIP,
+	SDMA0_HWIP,
+	SDMA1_HWIP,
+	SDMA2_HWIP,
+	SDMA3_HWIP,
+	SDMA4_HWIP,
+	SDMA5_HWIP,
+	SDMA6_HWIP,
+	SDMA7_HWIP,
+	LSDMA_HWIP,
+	MMHUB_HWIP,
+	ATHUB_HWIP,
+	NBIO_HWIP,
+	MP0_HWIP,
+	MP1_HWIP,
+	UVD_HWIP,
+	VCN_HWIP = UVD_HWIP,
+	JPEG_HWIP = VCN_HWIP,
+	VCN1_HWIP,
+	VCE_HWIP,
+	VPE_HWIP,
+	DF_HWIP,
+	DCE_HWIP,
+	OSSSYS_HWIP,
+	SMUIO_HWIP,
+	PWR_HWIP,
+	NBIF_HWIP,
+	THM_HWIP,
+	CLK_HWIP,
+	UMC_HWIP,
+	RSMU_HWIP,
+	XGMI_HWIP,
+	DCI_HWIP,
+	PCIE_HWIP,
+	ISP_HWIP,
+	MAX_HWIP
+};
+
+#define HWIP_MAX_INSTANCE	44
+
+#define HW_ID_MAX		300
+#define IP_VERSION_FULL(mj, mn, rv, var, srev) \
+	(((mj) << 24) | ((mn) << 16) | ((rv) << 8) | ((var) << 4) | (srev))
+#define IP_VERSION(mj, mn, rv)		IP_VERSION_FULL(mj, mn, rv, 0, 0)
+#define IP_VERSION_MAJ(ver)		((ver) >> 24)
+#define IP_VERSION_MIN(ver)		(((ver) >> 16) & 0xFF)
+#define IP_VERSION_REV(ver)		(((ver) >> 8) & 0xFF)
+#define IP_VERSION_VARIANT(ver)		(((ver) >> 4) & 0xF)
+#define IP_VERSION_SUBREV(ver)		((ver) & 0xF)
+#define IP_VERSION_MAJ_MIN_REV(ver)	((ver) >> 8)
+
+struct amdgpu_ip_map_info {
+	/* Map of logical to actual dev instances/mask */
+	uint32_t 		dev_inst[MAX_HWIP][HWIP_MAX_INSTANCE];
+	int8_t (*logical_to_dev_inst)(struct amdgpu_device *adev,
+				      enum amd_hw_ip_block_type block,
+				      int8_t inst);
+	uint32_t (*logical_to_dev_mask)(struct amdgpu_device *adev,
+					enum amd_hw_ip_block_type block,
+					uint32_t mask);
+};
+
+enum amdgpu_uid_type {
+	AMDGPU_UID_TYPE_XCD,
+	AMDGPU_UID_TYPE_AID,
+	AMDGPU_UID_TYPE_SOC,
+	AMDGPU_UID_TYPE_MAX
+};
+
+#define AMDGPU_UID_INST_MAX 8 /* max number of instances for each UID type */
+
+struct amdgpu_uid {
+	uint64_t uid[AMDGPU_UID_TYPE_MAX][AMDGPU_UID_INST_MAX];
+	struct amdgpu_device *adev;
+};
+
+struct amd_powerplay {
+	void *pp_handle;
+	const struct amd_pm_funcs *pp_funcs;
+};
+
+struct ip_discovery_top;
+
+/* polaris10 kickers */
+#define ASICID_IS_P20(did, rid)		(((did == 0x67DF) && \
+					 ((rid == 0xE3) || \
+					  (rid == 0xE4) || \
+					  (rid == 0xE5) || \
+					  (rid == 0xE7) || \
+					  (rid == 0xEF))) || \
+					 ((did == 0x6FDF) && \
+					 ((rid == 0xE7) || \
+					  (rid == 0xEF) || \
+					  (rid == 0xFF))))
+
+#define ASICID_IS_P30(did, rid)		((did == 0x67DF) && \
+					((rid == 0xE1) || \
+					 (rid == 0xF7)))
+
+/* polaris11 kickers */
+#define ASICID_IS_P21(did, rid)		(((did == 0x67EF) && \
+					 ((rid == 0xE0) || \
+					  (rid == 0xE5))) || \
+					 ((did == 0x67FF) && \
+					 ((rid == 0xCF) || \
+					  (rid == 0xEF) || \
+					  (rid == 0xFF))))
+
+#define ASICID_IS_P31(did, rid)		((did == 0x67EF) && \
+					((rid == 0xE2)))
+
+/* polaris12 kickers */
+#define ASICID_IS_P23(did, rid)		(((did == 0x6987) && \
+					 ((rid == 0xC0) || \
+					  (rid == 0xC1) || \
+					  (rid == 0xC3) || \
+					  (rid == 0xC7))) || \
+					 ((did == 0x6981) && \
+					 ((rid == 0x00) || \
+					  (rid == 0x01) || \
+					  (rid == 0x10))))
+
+struct amdgpu_mqd_prop {
+	uint64_t mqd_gpu_addr;
+	uint64_t hqd_base_gpu_addr;
+	uint64_t rptr_gpu_addr;
+	uint64_t wptr_gpu_addr;
+	uint32_t queue_size;
+	bool use_doorbell;
+	uint32_t doorbell_index;
+	uint64_t eop_gpu_addr;
+	uint32_t hqd_pipe_priority;
+	uint32_t hqd_queue_priority;
+	bool allow_tunneling;
+	bool hqd_active;
+	uint64_t shadow_addr;
+	uint64_t gds_bkup_addr;
+	uint64_t csa_addr;
+	uint64_t fence_address;
+	bool tmz_queue;
+	bool kernel_queue;
+};
+
+struct amdgpu_mqd {
+	unsigned mqd_size;
+	int (*init_mqd)(struct amdgpu_device *adev, void *mqd,
+			struct amdgpu_mqd_prop *p);
+};
+
+struct amdgpu_pcie_reset_ctx {
+	bool in_link_reset;
+	bool occurs_dpc;
+	bool audio_suspended;
+	struct pci_dev *swus;
+	struct pci_saved_state *swus_pcistate;
+	struct pci_saved_state *swds_pcistate;
+};
+
+/*
+ * Custom Init levels could be defined for different situations where a full
+ * initialization of all hardware blocks are not expected. Sample cases are
+ * custom init sequences after resume after S0i3/S3, reset on initialization,
+ * partial reset of blocks etc. Presently, this defines only two levels. Levels
+ * are described in corresponding struct definitions - amdgpu_init_default,
+ * amdgpu_init_minimal_xgmi.
+ */
+enum amdgpu_init_lvl_id {
+	AMDGPU_INIT_LEVEL_DEFAULT,
+	AMDGPU_INIT_LEVEL_MINIMAL_XGMI,
+	AMDGPU_INIT_LEVEL_RESET_RECOVERY,
+};
+
+struct amdgpu_init_level {
+	enum amdgpu_init_lvl_id level;
+	uint32_t hwini_ip_block_mask;
+};
+
 #define AMDGPU_RESET_MAGIC_NUM 64
+#define AMDGPU_MAX_DF_PERFMONS 4
+struct amdgpu_reset_domain;
+struct amdgpu_fru_info;
+
+enum amdgpu_enforce_isolation_mode {
+	AMDGPU_ENFORCE_ISOLATION_DISABLE = 0,
+	AMDGPU_ENFORCE_ISOLATION_ENABLE = 1,
+	AMDGPU_ENFORCE_ISOLATION_ENABLE_LEGACY = 2,
+	AMDGPU_ENFORCE_ISOLATION_NO_CLEANER_SHADER = 3,
+};
+
 struct amdgpu_device {
 	struct device			*dev;
-	struct drm_device		*ddev;
 	struct pci_dev			*pdev;
+	struct drm_device		ddev;
 
 #ifdef CONFIG_DRM_AMD_ACP
 	struct amdgpu_acp		acp;
 #endif
-
+	struct amdgpu_hive_info *hive;
+	struct amdgpu_xcp_mgr *xcp_mgr;
 	/* ASIC */
 	enum amd_asic_type		asic_type;
 	uint32_t			family;
 	uint32_t			rev_id;
 	uint32_t			external_rev_id;
 	unsigned long			flags;
+	unsigned long			apu_flags;
 	int				usec_timeout;
 	const struct amdgpu_asic_funcs	*asic_funcs;
 	bool				shutdown;
-	bool				need_dma32;
+	bool				need_swiotlb;
 	bool				accel_working;
-	struct work_struct		reset_work;
 	struct notifier_block		acpi_nb;
+	struct notifier_block		pm_nb;
 	struct amdgpu_i2c_chan		*i2c_bus[AMDGPU_MAX_I2C_BUS];
-	struct amdgpu_debugfs		debugfs[AMDGPU_DEBUGFS_MAX_COMPONENTS];
-	unsigned			debugfs_count;
-#if defined(CONFIG_DEBUG_FS)
-	struct dentry			*debugfs_regs[AMDGPU_DEBUGFS_MAX_COMPONENTS];
-#endif
-	struct amdgpu_atif		atif;
-	struct amdgpu_atcs		atcs;
+	struct debugfs_blob_wrapper     debugfs_vbios_blob;
+	struct debugfs_blob_wrapper     debugfs_discovery_blob;
 	struct mutex			srbm_mutex;
 	/* GRBM index mutex. Protects concurrent access to GRBM index */
 	struct mutex                    grbm_idx_mutex;
 	struct dev_pm_domain		vga_pm_domain;
 	bool				have_disp_power_ref;
+	bool                            have_atomics_support;
 
 	/* BIOS */
 	bool				is_atom_fw;
 	uint8_t				*bios;
 	uint32_t			bios_size;
-	struct amdgpu_bo		*stolen_vga_memory;
 	uint32_t			bios_scratch_reg_offset;
 	uint32_t			bios_scratch[AMDGPU_BIOS_NUM_SCRATCH];
 
@@ -1436,6 +994,7 @@ struct amdgpu_device {
 	void __iomem			*rmmio;
 	/* protects concurrent MM_INDEX/DATA based register access */
 	spinlock_t mmio_idx_lock;
+	struct amdgpu_mmio_remap        rmmio_remap;
 	/* protects concurrent SMC based register access */
 	spinlock_t smc_idx_lock;
 	amdgpu_rreg_t			smc_rreg;
@@ -1446,6 +1005,12 @@ struct amdgpu_device {
 	amdgpu_wreg_t			pcie_wreg;
 	amdgpu_rreg_t			pciep_rreg;
 	amdgpu_wreg_t			pciep_wreg;
+	amdgpu_rreg_ext_t		pcie_rreg_ext;
+	amdgpu_wreg_ext_t		pcie_wreg_ext;
+	amdgpu_rreg64_t			pcie_rreg64;
+	amdgpu_wreg64_t			pcie_wreg64;
+	amdgpu_rreg64_ext_t			pcie_rreg64_ext;
+	amdgpu_wreg64_ext_t			pcie_wreg64_ext;
 	/* protects concurrent UVD register access */
 	spinlock_t uvd_ctx_idx_lock;
 	amdgpu_rreg_t			uvd_ctx_rreg;
@@ -1466,23 +1031,22 @@ struct amdgpu_device {
 	spinlock_t audio_endpt_idx_lock;
 	amdgpu_block_rreg_t		audio_endpt_rreg;
 	amdgpu_block_wreg_t		audio_endpt_wreg;
-	void __iomem                    *rio_mem;
-	resource_size_t			rio_mem_size;
 	struct amdgpu_doorbell		doorbell;
 
 	/* clock/pll info */
 	struct amdgpu_clock            clock;
 
 	/* MC */
-	struct amdgpu_mc		mc;
+	struct amdgpu_gmc		gmc;
 	struct amdgpu_gart		gart;
-	struct amdgpu_dummy_page	dummy_page;
+	dma_addr_t			dummy_page_addr;
 	struct amdgpu_vm_manager	vm_manager;
 	struct amdgpu_vmhub             vmhub[AMDGPU_MAX_VMHUBS];
+	DECLARE_BITMAP(vmhubs_mask, AMDGPU_MAX_VMHUBS);
 
 	/* memory management */
 	struct amdgpu_mman		mman;
-	struct amdgpu_vram_scratch	vram_scratch;
+	struct amdgpu_mem_scratch	mem_scratch;
 	struct amdgpu_wb		wb;
 	atomic64_t			num_bytes_moved;
 	atomic64_t			num_evictions;
@@ -1501,34 +1065,50 @@ struct amdgpu_device {
 
 	/* display */
 	bool				enable_virtual_display;
+	struct amdgpu_vkms_output       *amdgpu_vkms_output;
 	struct amdgpu_mode_info		mode_info;
-	struct work_struct		hotplug_work;
+	/* For pre-DCE11. DCE11 and later are in "struct amdgpu_device->dm" */
+	struct delayed_work         hotplug_work;
 	struct amdgpu_irq_src		crtc_irq;
+	struct amdgpu_irq_src		vline0_irq;
+	struct amdgpu_irq_src		vupdate_irq;
 	struct amdgpu_irq_src		pageflip_irq;
 	struct amdgpu_irq_src		hpd_irq;
+	struct amdgpu_irq_src		dmub_trace_irq;
+	struct amdgpu_irq_src		dmub_outbox_irq;
 
 	/* rings */
 	u64				fence_context;
 	unsigned			num_rings;
 	struct amdgpu_ring		*rings[AMDGPU_MAX_RINGS];
+	struct dma_fence __rcu		*gang_submit;
 	bool				ib_pool_ready;
-	struct amdgpu_sa_manager	ring_tmp_bo;
+	struct amdgpu_sa_manager	ib_pools[AMDGPU_IB_POOL_MAX];
+	struct amdgpu_sched		gpu_sched[AMDGPU_HW_IP_NUM][AMDGPU_RING_PRIO_MAX];
 
 	/* interrupts */
 	struct amdgpu_irq		irq;
 
 	/* powerplay */
 	struct amd_powerplay		powerplay;
-	bool				pp_enabled;
-	bool				pp_force_state_enabled;
-
-	/* dpm */
 	struct amdgpu_pm		pm;
-	u32				cg_flags;
+	u64				cg_flags;
 	u32				pg_flags;
 
-	/* amdgpu smumgr */
-	struct amdgpu_smumgr smu;
+	/* nbio */
+	struct amdgpu_nbio		nbio;
+
+	/* hdp */
+	struct amdgpu_hdp		hdp;
+
+	/* smuio */
+	struct amdgpu_smuio		smuio;
+
+	/* mmhub */
+	struct amdgpu_mmhub		mmhub;
+
+	/* gfxhub */
+	struct amdgpu_gfxhub		gfxhub;
 
 	/* gfx */
 	struct amdgpu_gfx		gfx;
@@ -1536,18 +1116,27 @@ struct amdgpu_device {
 	/* sdma */
 	struct amdgpu_sdma		sdma;
 
-	union {
-		struct {
-			/* uvd */
-			struct amdgpu_uvd		uvd;
+	/* lsdma */
+	struct amdgpu_lsdma		lsdma;
+
+	/* uvd */
+	struct amdgpu_uvd		uvd;
+
+	/* vce */
+	struct amdgpu_vce		vce;
+
+	/* vcn */
+	struct amdgpu_vcn		vcn;
+
+	/* jpeg */
+	struct amdgpu_jpeg		jpeg;
 
-			/* vce */
-			struct amdgpu_vce		vce;
-		};
+	/* vpe */
+	struct amdgpu_vpe		vpe;
 
-		/* vcn */
-		struct amdgpu_vcn		vcn;
-	};
+	/* umsch */
+	struct amdgpu_umsch_mm		umsch_mm;
+	bool				enable_umsch_mm;
 
 	/* firmwares */
 	struct amdgpu_firmware		firmware;
@@ -1558,87 +1147,313 @@ struct amdgpu_device {
 	/* GDS */
 	struct amdgpu_gds		gds;
 
+	/* for userq and VM fences */
+	struct amdgpu_seq64		seq64;
+
+	/* UMC */
+	struct amdgpu_umc		umc;
+
+	/* display related functionality */
+	struct amdgpu_display_manager dm;
+
+#if defined(CONFIG_DRM_AMD_ISP)
+	/* isp */
+	struct amdgpu_isp		isp;
+#endif
+
+	/* mes */
+	bool                            enable_mes;
+	bool                            enable_mes_kiq;
+	bool                            enable_uni_mes;
+	struct amdgpu_mes               mes;
+	struct amdgpu_mqd               mqds[AMDGPU_HW_IP_NUM];
+	const struct amdgpu_userq_funcs *userq_funcs[AMDGPU_HW_IP_NUM];
+
+	/* xarray used to retrieve the user queue fence driver reference
+	 * in the EOP interrupt handler to signal the particular user
+	 * queue fence.
+	 */
+	struct xarray			userq_xa;
+
+	/* df */
+	struct amdgpu_df                df;
+
+	/* MCA */
+	struct amdgpu_mca               mca;
+
+	/* ACA */
+	struct amdgpu_aca		aca;
+
+	/* CPER */
+	struct amdgpu_cper		cper;
+
 	struct amdgpu_ip_block          ip_blocks[AMDGPU_MAX_IP_NUM];
+	uint32_t		        harvest_ip_mask;
 	int				num_ip_blocks;
 	struct mutex	mn_lock;
 	DECLARE_HASHTABLE(mn_hash, 7);
 
 	/* tracking pinned memory */
-	u64 vram_pin_size;
-	u64 invisible_pin_size;
-	u64 gart_pin_size;
+	atomic64_t vram_pin_size;
+	atomic64_t visible_pin_size;
+	atomic64_t gart_pin_size;
 
-	/* amdkfd interface */
-	struct kfd_dev          *kfd;
+	/* soc15 register offset based on ip, instance and  segment */
+	uint32_t		*reg_offset[MAX_HWIP][HWIP_MAX_INSTANCE];
+	struct amdgpu_ip_map_info	ip_map;
 
 	/* delayed work_func for deferring clockgating during resume */
-	struct delayed_work     late_init_work;
+	struct delayed_work     delayed_init_work;
 
 	struct amdgpu_virt	virt;
 
-	/* link all shadow bo */
-	struct list_head                shadow_list;
-	struct mutex                    shadow_list_lock;
-	/* link all gtt */
-	spinlock_t			gtt_list_lock;
-	struct list_head                gtt_list;
-	/* keep an lru list of rings by HW IP */
-	struct list_head		ring_lru_list;
-	spinlock_t			ring_lru_list_lock;
-
 	/* record hw reset is performed */
 	bool has_hw_reset;
 	u8				reset_magic[AMDGPU_RESET_MAGIC_NUM];
 
-	/* record last mm index being written through WREG32*/
-	unsigned long last_mm_index;
+	/* s3/s4 mask */
+	bool                            in_suspend;
+	bool				in_s3;
+	bool				in_s4;
+	bool				in_s0ix;
+	suspend_state_t			last_suspend_state;
+
+	enum pp_mp1_state               mp1_state;
+	struct amdgpu_doorbell_index doorbell_index;
+
+	struct mutex			notifier_lock;
+
+	int asic_reset_res;
+	struct work_struct		xgmi_reset_work;
+	struct list_head		reset_list;
+
+	long				gfx_timeout;
+	long				sdma_timeout;
+	long				video_timeout;
+	long				compute_timeout;
+	long				psp_timeout;
+
+	uint64_t			unique_id;
+	uint64_t	df_perfmon_config_assign_mask[AMDGPU_MAX_DF_PERFMONS];
+
+	/* enable runtime pm on the device */
+	bool                            in_runpm;
+	bool                            has_pr3;
+
+	bool                            ucode_sysfs_en;
+
+	struct amdgpu_fru_info		*fru_info;
+	atomic_t			throttling_logging_enabled;
+	struct ratelimit_state		throttling_logging_rs;
+	uint32_t                        ras_hw_enabled;
+	uint32_t                        ras_enabled;
+	bool                            ras_default_ecc_enabled;
+
+	bool                            no_hw_access;
+	struct pci_saved_state          *pci_state;
+	pci_channel_state_t		pci_channel_state;
+
+	struct amdgpu_pcie_reset_ctx	pcie_reset_ctx;
+
+	/* Track auto wait count on s_barrier settings */
+	bool				barrier_has_auto_waitcnt;
+
+	struct amdgpu_reset_control     *reset_cntl;
+	uint32_t                        ip_versions[MAX_HWIP][HWIP_MAX_INSTANCE];
+
+	bool				ram_is_direct_mapped;
+
+	struct list_head                ras_list;
+
+	struct ip_discovery_top         *ip_top;
+
+	struct amdgpu_reset_domain	*reset_domain;
+
+	struct mutex			benchmark_mutex;
+
+	bool                            scpm_enabled;
+	uint32_t                        scpm_status;
+
+	struct work_struct		reset_work;
+
+	bool                            dc_enabled;
+	/* Mask of active clusters */
+	uint32_t			aid_mask;
+
+	/* Debug */
+	bool                            debug_vm;
+	bool                            debug_largebar;
+	bool                            debug_disable_soft_recovery;
+	bool                            debug_use_vram_fw_buf;
+	bool                            debug_enable_ras_aca;
+	bool                            debug_exp_resets;
+	bool                            debug_disable_gpu_ring_reset;
+	bool                            debug_vm_userptr;
+	bool                            debug_disable_ce_logs;
+
+	/* Protection for the following isolation structure */
+	struct mutex                    enforce_isolation_mutex;
+	enum amdgpu_enforce_isolation_mode	enforce_isolation[MAX_XCP];
+	struct amdgpu_isolation {
+		void			*owner;
+		struct dma_fence	*spearhead;
+		struct amdgpu_sync	active;
+		struct amdgpu_sync	prev;
+	} isolation[MAX_XCP];
+
+	struct amdgpu_init_level *init_lvl;
+
+	/* This flag is used to determine how VRAM allocations are handled for APUs
+	 * in KFD: VRAM or GTT.
+	 */
+	bool                            apu_prefer_gtt;
+
+	struct list_head		userq_mgr_list;
+	struct mutex                    userq_mutex;
+	bool                            userq_halt_for_enforce_isolation;
+	struct amdgpu_uid *uid_info;
+
+	/* KFD
+	 * Must be last --ends in a flexible-array member.
+	 */
+	struct amdgpu_kfd_dev		kfd;
 };
 
-static inline struct amdgpu_device *amdgpu_ttm_adev(struct ttm_bo_device *bdev)
+static inline uint32_t amdgpu_ip_version(const struct amdgpu_device *adev,
+					 uint8_t ip, uint8_t inst)
+{
+	/* This considers only major/minor/rev and ignores
+	 * subrevision/variant fields.
+	 */
+	return adev->ip_versions[ip][inst] & ~0xFFU;
+}
+
+static inline uint32_t amdgpu_ip_version_full(const struct amdgpu_device *adev,
+					      uint8_t ip, uint8_t inst)
+{
+	/* This returns full version - major/minor/rev/variant/subrevision */
+	return adev->ip_versions[ip][inst];
+}
+
+static inline struct amdgpu_device *drm_to_adev(struct drm_device *ddev)
+{
+	return container_of(ddev, struct amdgpu_device, ddev);
+}
+
+static inline struct drm_device *adev_to_drm(struct amdgpu_device *adev)
+{
+	return &adev->ddev;
+}
+
+static inline struct amdgpu_device *amdgpu_ttm_adev(struct ttm_device *bdev)
 {
 	return container_of(bdev, struct amdgpu_device, mman.bdev);
 }
 
+static inline bool amdgpu_is_multi_aid(struct amdgpu_device *adev)
+{
+	return !!adev->aid_mask;
+}
+
 int amdgpu_device_init(struct amdgpu_device *adev,
-		       struct drm_device *ddev,
-		       struct pci_dev *pdev,
 		       uint32_t flags);
-void amdgpu_device_fini(struct amdgpu_device *adev);
+void amdgpu_device_fini_hw(struct amdgpu_device *adev);
+void amdgpu_device_fini_sw(struct amdgpu_device *adev);
+
 int amdgpu_gpu_wait_for_idle(struct amdgpu_device *adev);
 
-uint32_t amdgpu_mm_rreg(struct amdgpu_device *adev, uint32_t reg,
+void amdgpu_device_mm_access(struct amdgpu_device *adev, loff_t pos,
+			     void *buf, size_t size, bool write);
+size_t amdgpu_device_aper_access(struct amdgpu_device *adev, loff_t pos,
+				 void *buf, size_t size, bool write);
+
+void amdgpu_device_vram_access(struct amdgpu_device *adev, loff_t pos,
+			       void *buf, size_t size, bool write);
+uint32_t amdgpu_device_wait_on_rreg(struct amdgpu_device *adev,
+			    uint32_t inst, uint32_t reg_addr, char reg_name[],
+			    uint32_t expected_value, uint32_t mask);
+uint32_t amdgpu_device_rreg(struct amdgpu_device *adev,
+			    uint32_t reg, uint32_t acc_flags);
+u32 amdgpu_device_indirect_rreg_ext(struct amdgpu_device *adev,
+				    u64 reg_addr);
+uint32_t amdgpu_device_xcc_rreg(struct amdgpu_device *adev,
+				uint32_t reg, uint32_t acc_flags,
+				uint32_t xcc_id);
+void amdgpu_device_wreg(struct amdgpu_device *adev,
+			uint32_t reg, uint32_t v,
 			uint32_t acc_flags);
-void amdgpu_mm_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v,
-		    uint32_t acc_flags);
-u32 amdgpu_io_rreg(struct amdgpu_device *adev, u32 reg);
-void amdgpu_io_wreg(struct amdgpu_device *adev, u32 reg, u32 v);
-
-u32 amdgpu_mm_rdoorbell(struct amdgpu_device *adev, u32 index);
-void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v);
-u64 amdgpu_mm_rdoorbell64(struct amdgpu_device *adev, u32 index);
-void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v);
+void amdgpu_device_indirect_wreg_ext(struct amdgpu_device *adev,
+				     u64 reg_addr, u32 reg_data);
+void amdgpu_device_xcc_wreg(struct amdgpu_device *adev,
+			    uint32_t reg, uint32_t v,
+			    uint32_t acc_flags,
+			    uint32_t xcc_id);
+void amdgpu_mm_wreg_mmio_rlc(struct amdgpu_device *adev,
+			     uint32_t reg, uint32_t v, uint32_t xcc_id);
+void amdgpu_mm_wreg8(struct amdgpu_device *adev, uint32_t offset, uint8_t value);
+uint8_t amdgpu_mm_rreg8(struct amdgpu_device *adev, uint32_t offset);
+
+u32 amdgpu_device_indirect_rreg(struct amdgpu_device *adev,
+				u32 reg_addr);
+u64 amdgpu_device_indirect_rreg64(struct amdgpu_device *adev,
+				  u32 reg_addr);
+u64 amdgpu_device_indirect_rreg64_ext(struct amdgpu_device *adev,
+				  u64 reg_addr);
+void amdgpu_device_indirect_wreg(struct amdgpu_device *adev,
+				 u32 reg_addr, u32 reg_data);
+void amdgpu_device_indirect_wreg64(struct amdgpu_device *adev,
+				   u32 reg_addr, u64 reg_data);
+void amdgpu_device_indirect_wreg64_ext(struct amdgpu_device *adev,
+				   u64 reg_addr, u64 reg_data);
+u32 amdgpu_device_get_rev_id(struct amdgpu_device *adev);
+bool amdgpu_device_asic_has_dc_support(struct pci_dev *pdev,
+				       enum amd_asic_type asic_type);
+bool amdgpu_device_has_dc_support(struct amdgpu_device *adev);
+
+void amdgpu_device_set_sriov_virtual_display(struct amdgpu_device *adev);
+
+int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev,
+				 struct amdgpu_reset_context *reset_context);
+
+int amdgpu_do_asic_reset(struct list_head *device_list_handle,
+			 struct amdgpu_reset_context *reset_context);
+
+int amdgpu_device_reinit_after_reset(struct amdgpu_reset_context *reset_context);
+
+int emu_soc_asic_init(struct amdgpu_device *adev);
 
 /*
  * Registers read & write functions.
  */
-
-#define AMDGPU_REGS_IDX       (1<<0)
 #define AMDGPU_REGS_NO_KIQ    (1<<1)
+#define AMDGPU_REGS_RLC	(1<<2)
+
+#define RREG32_NO_KIQ(reg) amdgpu_device_rreg(adev, (reg), AMDGPU_REGS_NO_KIQ)
+#define WREG32_NO_KIQ(reg, v) amdgpu_device_wreg(adev, (reg), (v), AMDGPU_REGS_NO_KIQ)
+
+#define RREG32_KIQ(reg) amdgpu_kiq_rreg(adev, (reg), 0)
+#define WREG32_KIQ(reg, v) amdgpu_kiq_wreg(adev, (reg), (v), 0)
 
-#define RREG32_NO_KIQ(reg) amdgpu_mm_rreg(adev, (reg), AMDGPU_REGS_NO_KIQ)
-#define WREG32_NO_KIQ(reg, v) amdgpu_mm_wreg(adev, (reg), (v), AMDGPU_REGS_NO_KIQ)
+#define RREG8(reg) amdgpu_mm_rreg8(adev, (reg))
+#define WREG8(reg, v) amdgpu_mm_wreg8(adev, (reg), (v))
 
-#define RREG32(reg) amdgpu_mm_rreg(adev, (reg), 0)
-#define RREG32_IDX(reg) amdgpu_mm_rreg(adev, (reg), AMDGPU_REGS_IDX)
-#define DREG32(reg) printk(KERN_INFO "REGISTER: " #reg " : 0x%08X\n", amdgpu_mm_rreg(adev, (reg), 0))
-#define WREG32(reg, v) amdgpu_mm_wreg(adev, (reg), (v), 0)
-#define WREG32_IDX(reg, v) amdgpu_mm_wreg(adev, (reg), (v), AMDGPU_REGS_IDX)
+#define RREG32(reg) amdgpu_device_rreg(adev, (reg), 0)
+#define DREG32(reg) printk(KERN_INFO "REGISTER: " #reg " : 0x%08X\n", amdgpu_device_rreg(adev, (reg), 0))
+#define WREG32(reg, v) amdgpu_device_wreg(adev, (reg), (v), 0)
 #define REG_SET(FIELD, v) (((v) << FIELD##_SHIFT) & FIELD##_MASK)
 #define REG_GET(FIELD, v) (((v) << FIELD##_SHIFT) & FIELD##_MASK)
+#define RREG32_XCC(reg, inst) amdgpu_device_xcc_rreg(adev, (reg), 0, inst)
+#define WREG32_XCC(reg, v, inst) amdgpu_device_xcc_wreg(adev, (reg), (v), 0, inst)
 #define RREG32_PCIE(reg) adev->pcie_rreg(adev, (reg))
 #define WREG32_PCIE(reg, v) adev->pcie_wreg(adev, (reg), (v))
 #define RREG32_PCIE_PORT(reg) adev->pciep_rreg(adev, (reg))
 #define WREG32_PCIE_PORT(reg, v) adev->pciep_wreg(adev, (reg), (v))
+#define RREG32_PCIE_EXT(reg) adev->pcie_rreg_ext(adev, (reg))
+#define WREG32_PCIE_EXT(reg, v) adev->pcie_wreg_ext(adev, (reg), (v))
+#define RREG64_PCIE(reg) adev->pcie_rreg64(adev, (reg))
+#define WREG64_PCIE(reg, v) adev->pcie_wreg64(adev, (reg), (v))
+#define RREG64_PCIE_EXT(reg) adev->pcie_rreg64_ext(adev, (reg))
+#define WREG64_PCIE_EXT(reg, v) adev->pcie_wreg64_ext(adev, (reg), (v))
 #define RREG32_SMC(reg) adev->smc_rreg(adev, (reg))
 #define WREG32_SMC(reg, v) adev->smc_wreg(adev, (reg), (v))
 #define RREG32_UVD_CTX(reg) adev->uvd_ctx_rreg(adev, (reg))
@@ -1667,14 +1482,16 @@ void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v);
 		tmp_ |= ((val) & ~(mask));			\
 		WREG32_PLL(reg, tmp_);				\
 	} while (0)
-#define DREG32_SYS(sqf, adev, reg) seq_printf((sqf), #reg " : 0x%08X\n", amdgpu_mm_rreg((adev), (reg), false))
-#define RREG32_IO(reg) amdgpu_io_rreg(adev, (reg))
-#define WREG32_IO(reg, v) amdgpu_io_wreg(adev, (reg), (v))
 
-#define RDOORBELL32(index) amdgpu_mm_rdoorbell(adev, (index))
-#define WDOORBELL32(index, v) amdgpu_mm_wdoorbell(adev, (index), (v))
-#define RDOORBELL64(index) amdgpu_mm_rdoorbell64(adev, (index))
-#define WDOORBELL64(index, v) amdgpu_mm_wdoorbell64(adev, (index), (v))
+#define WREG32_SMC_P(_Reg, _Val, _Mask)                         \
+	do {                                                    \
+		u32 tmp = RREG32_SMC(_Reg);                     \
+		tmp &= (_Mask);                                 \
+		tmp |= ((_Val) & ~(_Mask));                     \
+		WREG32_SMC(_Reg, tmp);                          \
+	} while (0)
+
+#define DREG32_SYS(sqf, adev, reg) seq_printf((sqf), #reg " : 0x%08X\n", amdgpu_device_rreg((adev), (reg), false))
 
 #define REG_FIELD_SHIFT(reg, field) reg##__##field##__SHIFT
 #define REG_FIELD_MASK(reg, field) reg##__##field##_MASK
@@ -1692,6 +1509,7 @@ void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v);
 #define WREG32_FIELD_OFFSET(reg, offset, field, val)	\
 	WREG32(mm##reg + offset, (RREG32(mm##reg + offset) & ~REG_FIELD_MASK(reg, field)) | (val) << REG_FIELD_SHIFT(reg, field))
 
+#define AMDGPU_GET_REG_FIELD(x, h, l) (((x) & GENMASK_ULL(h, l)) >> (l))
 /*
  * BIOS helpers.
  */
@@ -1699,27 +1517,13 @@ void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v);
 #define RBIOS16(i) (RBIOS8(i) | (RBIOS8((i)+1) << 8))
 #define RBIOS32(i) ((RBIOS16(i)) | (RBIOS16((i)+2) << 16))
 
-static inline struct amdgpu_sdma_instance *
-amdgpu_get_sdma_instance(struct amdgpu_ring *ring)
-{
-	struct amdgpu_device *adev = ring->adev;
-	int i;
-
-	for (i = 0; i < adev->sdma.num_instances; i++)
-		if (&adev->sdma.instance[i].ring == ring)
-			break;
-
-	if (i < AMDGPU_MAX_SDMA_INSTANCES)
-		return &adev->sdma.instance[i];
-	else
-		return NULL;
-}
-
 /*
  * ASICs macro.
  */
-#define amdgpu_asic_set_vga_state(adev, state) (adev)->asic_funcs->set_vga_state((adev), (state))
+#define amdgpu_asic_set_vga_state(adev, state) \
+    ((adev)->asic_funcs->set_vga_state ? (adev)->asic_funcs->set_vga_state((adev), (state)) : 0)
 #define amdgpu_asic_reset(adev) (adev)->asic_funcs->reset((adev))
+#define amdgpu_asic_reset_method(adev) (adev)->asic_funcs->reset_method((adev))
 #define amdgpu_asic_get_xclk(adev) (adev)->asic_funcs->get_xclk((adev))
 #define amdgpu_asic_set_uvd_clocks(adev, v, d) (adev)->asic_funcs->set_uvd_clocks((adev), (v), (d))
 #define amdgpu_asic_set_vce_clocks(adev, ev, ec) (adev)->asic_funcs->set_vce_clocks((adev), (ev), (ec))
@@ -1730,103 +1534,93 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring)
 #define amdgpu_asic_read_bios_from_rom(adev, b, l) (adev)->asic_funcs->read_bios_from_rom((adev), (b), (l))
 #define amdgpu_asic_read_register(adev, se, sh, offset, v)((adev)->asic_funcs->read_register((adev), (se), (sh), (offset), (v)))
 #define amdgpu_asic_get_config_memsize(adev) (adev)->asic_funcs->get_config_memsize((adev))
-#define amdgpu_gart_flush_gpu_tlb(adev, vmid) (adev)->gart.gart_funcs->flush_gpu_tlb((adev), (vmid))
-#define amdgpu_gart_set_pte_pde(adev, pt, idx, addr, flags) (adev)->gart.gart_funcs->set_pte_pde((adev), (pt), (idx), (addr), (flags))
-#define amdgpu_gart_get_vm_pde(adev, addr) (adev)->gart.gart_funcs->get_vm_pde((adev), (addr))
-#define amdgpu_vm_copy_pte(adev, ib, pe, src, count) ((adev)->vm_manager.vm_pte_funcs->copy_pte((ib), (pe), (src), (count)))
-#define amdgpu_vm_write_pte(adev, ib, pe, value, count, incr) ((adev)->vm_manager.vm_pte_funcs->write_pte((ib), (pe), (value), (count), (incr)))
-#define amdgpu_vm_set_pte_pde(adev, ib, pe, addr, count, incr, flags) ((adev)->vm_manager.vm_pte_funcs->set_pte_pde((ib), (pe), (addr), (count), (incr), (flags)))
-#define amdgpu_vm_get_pte_flags(adev, flags) (adev)->gart.gart_funcs->get_vm_pte_flags((adev),(flags))
-#define amdgpu_ring_parse_cs(r, p, ib) ((r)->funcs->parse_cs((p), (ib)))
-#define amdgpu_ring_test_ring(r) (r)->funcs->test_ring((r))
-#define amdgpu_ring_test_ib(r, t) (r)->funcs->test_ib((r), (t))
-#define amdgpu_ring_get_rptr(r) (r)->funcs->get_rptr((r))
-#define amdgpu_ring_get_wptr(r) (r)->funcs->get_wptr((r))
-#define amdgpu_ring_set_wptr(r) (r)->funcs->set_wptr((r))
-#define amdgpu_ring_emit_ib(r, ib, vm_id, c) (r)->funcs->emit_ib((r), (ib), (vm_id), (c))
-#define amdgpu_ring_emit_pipeline_sync(r) (r)->funcs->emit_pipeline_sync((r))
-#define amdgpu_ring_emit_vm_flush(r, vmid, addr) (r)->funcs->emit_vm_flush((r), (vmid), (addr))
-#define amdgpu_ring_emit_fence(r, addr, seq, flags) (r)->funcs->emit_fence((r), (addr), (seq), (flags))
-#define amdgpu_ring_emit_gds_switch(r, v, db, ds, wb, ws, ab, as) (r)->funcs->emit_gds_switch((r), (v), (db), (ds), (wb), (ws), (ab), (as))
-#define amdgpu_ring_emit_hdp_flush(r) (r)->funcs->emit_hdp_flush((r))
-#define amdgpu_ring_emit_hdp_invalidate(r) (r)->funcs->emit_hdp_invalidate((r))
-#define amdgpu_ring_emit_switch_buffer(r) (r)->funcs->emit_switch_buffer((r))
-#define amdgpu_ring_emit_cntxcntl(r, d) (r)->funcs->emit_cntxcntl((r), (d))
-#define amdgpu_ring_emit_rreg(r, d) (r)->funcs->emit_rreg((r), (d))
-#define amdgpu_ring_emit_wreg(r, d, v) (r)->funcs->emit_wreg((r), (d), (v))
-#define amdgpu_ring_emit_tmz(r, b) (r)->funcs->emit_tmz((r), (b))
-#define amdgpu_ring_pad_ib(r, ib) ((r)->funcs->pad_ib((r), (ib)))
-#define amdgpu_ring_init_cond_exec(r) (r)->funcs->init_cond_exec((r))
-#define amdgpu_ring_patch_cond_exec(r,o) (r)->funcs->patch_cond_exec((r),(o))
-#define amdgpu_ih_get_wptr(adev) (adev)->irq.ih_funcs->get_wptr((adev))
-#define amdgpu_ih_decode_iv(adev, iv) (adev)->irq.ih_funcs->decode_iv((adev), (iv))
-#define amdgpu_ih_set_rptr(adev) (adev)->irq.ih_funcs->set_rptr((adev))
-#define amdgpu_display_vblank_get_counter(adev, crtc) (adev)->mode_info.funcs->vblank_get_counter((adev), (crtc))
-#define amdgpu_display_vblank_wait(adev, crtc) (adev)->mode_info.funcs->vblank_wait((adev), (crtc))
-#define amdgpu_display_backlight_set_level(adev, e, l) (adev)->mode_info.funcs->backlight_set_level((e), (l))
-#define amdgpu_display_backlight_get_level(adev, e) (adev)->mode_info.funcs->backlight_get_level((e))
-#define amdgpu_display_hpd_sense(adev, h) (adev)->mode_info.funcs->hpd_sense((adev), (h))
-#define amdgpu_display_hpd_set_polarity(adev, h) (adev)->mode_info.funcs->hpd_set_polarity((adev), (h))
-#define amdgpu_display_hpd_get_gpio_reg(adev) (adev)->mode_info.funcs->hpd_get_gpio_reg((adev))
-#define amdgpu_display_bandwidth_update(adev) (adev)->mode_info.funcs->bandwidth_update((adev))
-#define amdgpu_display_page_flip(adev, crtc, base, async) (adev)->mode_info.funcs->page_flip((adev), (crtc), (base), (async))
-#define amdgpu_display_page_flip_get_scanoutpos(adev, crtc, vbl, pos) (adev)->mode_info.funcs->page_flip_get_scanoutpos((adev), (crtc), (vbl), (pos))
-#define amdgpu_display_add_encoder(adev, e, s, c) (adev)->mode_info.funcs->add_encoder((adev), (e), (s), (c))
-#define amdgpu_display_add_connector(adev, ci, sd, ct, ib, coi, h, r) (adev)->mode_info.funcs->add_connector((adev), (ci), (sd), (ct), (ib), (coi), (h), (r))
-#define amdgpu_emit_copy_buffer(adev, ib, s, d, b) (adev)->mman.buffer_funcs->emit_copy_buffer((ib),  (s), (d), (b))
-#define amdgpu_emit_fill_buffer(adev, ib, s, d, b) (adev)->mman.buffer_funcs->emit_fill_buffer((ib), (s), (d), (b))
-#define amdgpu_gfx_get_gpu_clock_counter(adev) (adev)->gfx.funcs->get_gpu_clock_counter((adev))
-#define amdgpu_gfx_select_se_sh(adev, se, sh, instance) (adev)->gfx.funcs->select_se_sh((adev), (se), (sh), (instance))
-#define amdgpu_gds_switch(adev, r, v, d, w, a) (adev)->gds.funcs->patch_gds_switch((r), (v), (d), (w), (a))
-#define amdgpu_psp_check_fw_loading_status(adev, i) (adev)->firmware.funcs->check_fw_loading_status((adev), (i))
+#define amdgpu_asic_flush_hdp(adev, r) \
+	((adev)->asic_funcs->flush_hdp ? (adev)->asic_funcs->flush_hdp((adev), (r)) : (adev)->hdp.funcs->flush_hdp((adev), (r)))
+#define amdgpu_asic_invalidate_hdp(adev, r) \
+	((adev)->asic_funcs->invalidate_hdp ? (adev)->asic_funcs->invalidate_hdp((adev), (r)) : \
+	 ((adev)->hdp.funcs->invalidate_hdp ? (adev)->hdp.funcs->invalidate_hdp((adev), (r)) : (void)0))
+#define amdgpu_asic_need_full_reset(adev) (adev)->asic_funcs->need_full_reset((adev))
+#define amdgpu_asic_init_doorbell_index(adev) (adev)->asic_funcs->init_doorbell_index((adev))
+#define amdgpu_asic_get_pcie_usage(adev, cnt0, cnt1) ((adev)->asic_funcs->get_pcie_usage((adev), (cnt0), (cnt1)))
+#define amdgpu_asic_need_reset_on_init(adev) (adev)->asic_funcs->need_reset_on_init((adev))
+#define amdgpu_asic_get_pcie_replay_count(adev) ((adev)->asic_funcs->get_pcie_replay_count((adev)))
+#define amdgpu_asic_supports_baco(adev) (adev)->asic_funcs->supports_baco((adev))
+#define amdgpu_asic_pre_asic_init(adev) (adev)->asic_funcs->pre_asic_init((adev))
+#define amdgpu_asic_update_umd_stable_pstate(adev, enter) \
+	((adev)->asic_funcs->update_umd_stable_pstate ? (adev)->asic_funcs->update_umd_stable_pstate((adev), (enter)) : 0)
+#define amdgpu_asic_query_video_codecs(adev, e, c) (adev)->asic_funcs->query_video_codecs((adev), (e), (c))
+
+#define amdgpu_inc_vram_lost(adev) atomic_inc(&((adev)->vram_lost_counter))
+
+#define BIT_MASK_UPPER(i) ((i) >= BITS_PER_LONG ? 0 : ~0UL << (i))
+#define for_each_inst(i, inst_mask)        \
+	for (i = ffs(inst_mask); i-- != 0; \
+	     i = ffs(inst_mask & BIT_MASK_UPPER(i + 1)))
 
 /* Common functions */
-int amdgpu_gpu_reset(struct amdgpu_device *adev);
-bool amdgpu_need_backup(struct amdgpu_device *adev);
-void amdgpu_pci_config_reset(struct amdgpu_device *adev);
-bool amdgpu_need_post(struct amdgpu_device *adev);
-void amdgpu_update_display_priority(struct amdgpu_device *adev);
+bool amdgpu_device_has_job_running(struct amdgpu_device *adev);
+bool amdgpu_device_should_recover_gpu(struct amdgpu_device *adev);
+int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
+			      struct amdgpu_job *job,
+			      struct amdgpu_reset_context *reset_context);
+void amdgpu_device_pci_config_reset(struct amdgpu_device *adev);
+int amdgpu_device_pci_reset(struct amdgpu_device *adev);
+bool amdgpu_device_need_post(struct amdgpu_device *adev);
+bool amdgpu_device_seamless_boot_supported(struct amdgpu_device *adev);
+bool amdgpu_device_should_use_aspm(struct amdgpu_device *adev);
 
 void amdgpu_cs_report_moved_bytes(struct amdgpu_device *adev, u64 num_bytes,
 				  u64 num_vis_bytes);
-void amdgpu_ttm_placement_from_domain(struct amdgpu_bo *abo, u32 domain);
-bool amdgpu_ttm_bo_is_amdgpu_bo(struct ttm_buffer_object *bo);
-int amdgpu_ttm_tt_get_user_pages(struct ttm_tt *ttm, struct page **pages);
-int amdgpu_ttm_tt_set_userptr(struct ttm_tt *ttm, uint64_t addr,
-				     uint32_t flags);
-bool amdgpu_ttm_tt_has_userptr(struct ttm_tt *ttm);
-struct mm_struct *amdgpu_ttm_tt_get_usermm(struct ttm_tt *ttm);
-bool amdgpu_ttm_tt_affect_userptr(struct ttm_tt *ttm, unsigned long start,
-				  unsigned long end);
-bool amdgpu_ttm_tt_userptr_invalidated(struct ttm_tt *ttm,
-				       int *last_invalidated);
-bool amdgpu_ttm_tt_is_readonly(struct ttm_tt *ttm);
-uint64_t amdgpu_ttm_tt_pte_flags(struct amdgpu_device *adev, struct ttm_tt *ttm,
-				 struct ttm_mem_reg *mem);
-void amdgpu_vram_location(struct amdgpu_device *adev, struct amdgpu_mc *mc, u64 base);
-void amdgpu_gart_location(struct amdgpu_device *adev, struct amdgpu_mc *mc);
-void amdgpu_ttm_set_active_vram_size(struct amdgpu_device *adev, u64 size);
-int amdgpu_ttm_init(struct amdgpu_device *adev);
-void amdgpu_ttm_fini(struct amdgpu_device *adev);
-void amdgpu_program_register_sequence(struct amdgpu_device *adev,
+int amdgpu_device_resize_fb_bar(struct amdgpu_device *adev);
+void amdgpu_device_program_register_sequence(struct amdgpu_device *adev,
 					     const u32 *registers,
 					     const u32 array_size);
 
-bool amdgpu_device_is_px(struct drm_device *dev);
+int amdgpu_device_mode1_reset(struct amdgpu_device *adev);
+int amdgpu_device_link_reset(struct amdgpu_device *adev);
+bool amdgpu_device_supports_atpx(struct amdgpu_device *adev);
+bool amdgpu_device_supports_px(struct amdgpu_device *adev);
+bool amdgpu_device_supports_boco(struct amdgpu_device *adev);
+bool amdgpu_device_supports_smart_shift(struct amdgpu_device *adev);
+int amdgpu_device_supports_baco(struct amdgpu_device *adev);
+void amdgpu_device_detect_runtime_pm_mode(struct amdgpu_device *adev);
+bool amdgpu_device_is_peer_accessible(struct amdgpu_device *adev,
+				      struct amdgpu_device *peer_adev);
+int amdgpu_device_baco_enter(struct amdgpu_device *adev);
+int amdgpu_device_baco_exit(struct amdgpu_device *adev);
+
+void amdgpu_device_flush_hdp(struct amdgpu_device *adev,
+		struct amdgpu_ring *ring);
+void amdgpu_device_invalidate_hdp(struct amdgpu_device *adev,
+		struct amdgpu_ring *ring);
+
+void amdgpu_device_halt(struct amdgpu_device *adev);
+u32 amdgpu_device_pcie_port_rreg(struct amdgpu_device *adev,
+				u32 reg);
+void amdgpu_device_pcie_port_wreg(struct amdgpu_device *adev,
+				u32 reg, u32 v);
+struct dma_fence *amdgpu_device_get_gang(struct amdgpu_device *adev);
+struct dma_fence *amdgpu_device_switch_gang(struct amdgpu_device *adev,
+					    struct dma_fence *gang);
+struct dma_fence *amdgpu_device_enforce_isolation(struct amdgpu_device *adev,
+						  struct amdgpu_ring *ring,
+						  struct amdgpu_job *job);
+bool amdgpu_device_has_display_hardware(struct amdgpu_device *adev);
+ssize_t amdgpu_get_soft_full_reset_mask(struct amdgpu_ring *ring);
+ssize_t amdgpu_show_reset_mask(char *buf, uint32_t supported_reset);
+
 /* atpx handler */
 #if defined(CONFIG_VGA_SWITCHEROO)
 void amdgpu_register_atpx_handler(void);
 void amdgpu_unregister_atpx_handler(void);
 bool amdgpu_has_atpx_dgpu_power_cntl(void);
 bool amdgpu_is_atpx_hybrid(void);
-bool amdgpu_atpx_dgpu_req_power_for_displays(void);
 bool amdgpu_has_atpx(void);
 #else
 static inline void amdgpu_register_atpx_handler(void) {}
 static inline void amdgpu_unregister_atpx_handler(void) {}
 static inline bool amdgpu_has_atpx_dgpu_power_cntl(void) { return false; }
 static inline bool amdgpu_is_atpx_hybrid(void) { return false; }
-static inline bool amdgpu_atpx_dgpu_req_power_for_displays(void) { return false; }
 static inline bool amdgpu_has_atpx(void) { return false; }
 #endif
 
@@ -1836,22 +1630,23 @@ static inline bool amdgpu_has_atpx(void) { return false; }
 extern const struct drm_ioctl_desc amdgpu_ioctls_kms[];
 extern const int amdgpu_max_kms_ioctl;
 
-bool amdgpu_kms_vram_lost(struct amdgpu_device *adev,
-			  struct amdgpu_fpriv *fpriv);
-int amdgpu_driver_load_kms(struct drm_device *dev, unsigned long flags);
+int amdgpu_driver_load_kms(struct amdgpu_device *adev, unsigned long flags);
 void amdgpu_driver_unload_kms(struct drm_device *dev);
-void amdgpu_driver_lastclose_kms(struct drm_device *dev);
 int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv);
 void amdgpu_driver_postclose_kms(struct drm_device *dev,
 				 struct drm_file *file_priv);
-int amdgpu_suspend(struct amdgpu_device *adev);
-int amdgpu_device_suspend(struct drm_device *dev, bool suspend, bool fbcon);
-int amdgpu_device_resume(struct drm_device *dev, bool resume, bool fbcon);
-u32 amdgpu_get_vblank_counter_kms(struct drm_device *dev, unsigned int pipe);
-int amdgpu_enable_vblank_kms(struct drm_device *dev, unsigned int pipe);
-void amdgpu_disable_vblank_kms(struct drm_device *dev, unsigned int pipe);
-long amdgpu_kms_compat_ioctl(struct file *filp, unsigned int cmd,
-			     unsigned long arg);
+void amdgpu_driver_release_kms(struct drm_device *dev);
+
+int amdgpu_device_ip_suspend(struct amdgpu_device *adev);
+int amdgpu_device_prepare(struct drm_device *dev);
+void amdgpu_device_complete(struct drm_device *dev);
+int amdgpu_device_suspend(struct drm_device *dev, bool fbcon);
+int amdgpu_device_resume(struct drm_device *dev, bool fbcon);
+u32 amdgpu_get_vblank_counter_kms(struct drm_crtc *crtc);
+int amdgpu_enable_vblank_kms(struct drm_crtc *crtc);
+void amdgpu_disable_vblank_kms(struct drm_crtc *crtc);
+int amdgpu_info_ioctl(struct drm_device *dev, void *data,
+		      struct drm_file *filp);
 
 /*
  * functions used by amdgpu_encoder.c
@@ -1873,22 +1668,141 @@ struct amdgpu_afmt_acr {
 struct amdgpu_afmt_acr amdgpu_afmt_acr(uint32_t clock);
 
 /* amdgpu_acpi.c */
+
+struct amdgpu_numa_info {
+	uint64_t size;
+	int pxm;
+	int nid;
+};
+
+/* ATCS Device/Driver State */
+#define AMDGPU_ATCS_PSC_DEV_STATE_D0		0
+#define AMDGPU_ATCS_PSC_DEV_STATE_D3_HOT	3
+#define AMDGPU_ATCS_PSC_DRV_STATE_OPR		0
+#define AMDGPU_ATCS_PSC_DRV_STATE_NOT_OPR	1
+
 #if defined(CONFIG_ACPI)
 int amdgpu_acpi_init(struct amdgpu_device *adev);
 void amdgpu_acpi_fini(struct amdgpu_device *adev);
 bool amdgpu_acpi_is_pcie_performance_request_supported(struct amdgpu_device *adev);
+bool amdgpu_acpi_is_power_shift_control_supported(void);
 int amdgpu_acpi_pcie_performance_request(struct amdgpu_device *adev,
 						u8 perf_req, bool advertise);
+int amdgpu_acpi_power_shift_control(struct amdgpu_device *adev,
+				    u8 dev_state, bool drv_state);
+int amdgpu_acpi_smart_shift_update(struct amdgpu_device *adev,
+				   enum amdgpu_ss ss_state);
 int amdgpu_acpi_pcie_notify_device_ready(struct amdgpu_device *adev);
+int amdgpu_acpi_get_tmr_info(struct amdgpu_device *adev, u64 *tmr_offset,
+			     u64 *tmr_size);
+int amdgpu_acpi_get_mem_info(struct amdgpu_device *adev, int xcc_id,
+			     struct amdgpu_numa_info *numa_info);
+
+void amdgpu_acpi_get_backlight_caps(struct amdgpu_dm_backlight_caps *caps);
+bool amdgpu_acpi_should_gpu_reset(struct amdgpu_device *adev);
+void amdgpu_acpi_detect(void);
+void amdgpu_acpi_release(void);
 #else
 static inline int amdgpu_acpi_init(struct amdgpu_device *adev) { return 0; }
+static inline int amdgpu_acpi_get_tmr_info(struct amdgpu_device *adev,
+					   u64 *tmr_offset, u64 *tmr_size)
+{
+	return -EINVAL;
+}
+static inline int amdgpu_acpi_get_mem_info(struct amdgpu_device *adev,
+					   int xcc_id,
+					   struct amdgpu_numa_info *numa_info)
+{
+	return -EINVAL;
+}
 static inline void amdgpu_acpi_fini(struct amdgpu_device *adev) { }
+static inline bool amdgpu_acpi_should_gpu_reset(struct amdgpu_device *adev) { return false; }
+static inline void amdgpu_acpi_detect(void) { }
+static inline void amdgpu_acpi_release(void) { }
+static inline bool amdgpu_acpi_is_power_shift_control_supported(void) { return false; }
+static inline int amdgpu_acpi_power_shift_control(struct amdgpu_device *adev,
+						  u8 dev_state, bool drv_state) { return 0; }
+static inline int amdgpu_acpi_smart_shift_update(struct amdgpu_device *adev,
+						 enum amdgpu_ss ss_state)
+{
+	return 0;
+}
+static inline void amdgpu_acpi_get_backlight_caps(struct amdgpu_dm_backlight_caps *caps) { }
 #endif
 
-struct amdgpu_bo_va_mapping *
-amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser,
-		       uint64_t addr, struct amdgpu_bo **bo);
-int amdgpu_cs_sysvm_access_required(struct amdgpu_cs_parser *parser);
+#if defined(CONFIG_ACPI) && defined(CONFIG_SUSPEND)
+bool amdgpu_acpi_is_s3_active(struct amdgpu_device *adev);
+bool amdgpu_acpi_is_s0ix_active(struct amdgpu_device *adev);
+#else
+static inline bool amdgpu_acpi_is_s0ix_active(struct amdgpu_device *adev) { return false; }
+static inline bool amdgpu_acpi_is_s3_active(struct amdgpu_device *adev) { return false; }
+#endif
+
+#if defined(CONFIG_DRM_AMD_ISP)
+int amdgpu_acpi_get_isp4_dev(struct acpi_device **dev);
+#endif
+
+void amdgpu_register_gpu_instance(struct amdgpu_device *adev);
+void amdgpu_unregister_gpu_instance(struct amdgpu_device *adev);
+
+pci_ers_result_t amdgpu_pci_error_detected(struct pci_dev *pdev,
+					   pci_channel_state_t state);
+pci_ers_result_t amdgpu_pci_mmio_enabled(struct pci_dev *pdev);
+pci_ers_result_t amdgpu_pci_slot_reset(struct pci_dev *pdev);
+void amdgpu_pci_resume(struct pci_dev *pdev);
+
+bool amdgpu_device_cache_pci_state(struct pci_dev *pdev);
+bool amdgpu_device_load_pci_state(struct pci_dev *pdev);
+
+bool amdgpu_device_skip_hw_access(struct amdgpu_device *adev);
+
+int amdgpu_device_set_cg_state(struct amdgpu_device *adev,
+			       enum amd_clockgating_state state);
+int amdgpu_device_set_pg_state(struct amdgpu_device *adev,
+			       enum amd_powergating_state state);
+
+static inline bool amdgpu_device_has_timeouts_enabled(struct amdgpu_device *adev)
+{
+	return amdgpu_gpu_recovery != 0 &&
+		adev->gfx_timeout != MAX_SCHEDULE_TIMEOUT &&
+		adev->compute_timeout != MAX_SCHEDULE_TIMEOUT &&
+		adev->sdma_timeout != MAX_SCHEDULE_TIMEOUT &&
+		adev->video_timeout != MAX_SCHEDULE_TIMEOUT;
+}
 
 #include "amdgpu_object.h"
+
+static inline bool amdgpu_is_tmz(struct amdgpu_device *adev)
+{
+       return adev->gmc.tmz_enabled;
+}
+
+int amdgpu_in_reset(struct amdgpu_device *adev);
+
+extern const struct attribute_group amdgpu_vram_mgr_attr_group;
+extern const struct attribute_group amdgpu_gtt_mgr_attr_group;
+extern const struct attribute_group amdgpu_flash_attr_group;
+
+void amdgpu_set_init_level(struct amdgpu_device *adev,
+			   enum amdgpu_init_lvl_id lvl);
+
+static inline int amdgpu_device_bus_status_check(struct amdgpu_device *adev)
+{
+       u32 status;
+       int r;
+
+       r = pci_read_config_dword(adev->pdev, PCI_COMMAND, &status);
+       if (r || PCI_POSSIBLE_ERROR(status)) {
+		dev_err(adev->dev, "device lost from bus!");
+		return -ENODEV;
+       }
+
+       return 0;
+}
+
+void amdgpu_device_set_uid(struct amdgpu_uid *uid_info,
+			   enum amdgpu_uid_type type, uint8_t inst,
+			   uint64_t uid);
+uint64_t amdgpu_device_get_uid(struct amdgpu_uid *uid_info,
+			       enum amdgpu_uid_type type, uint8_t inst);
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c
new file mode 100644
index 000000000000..9b3180449150
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c
@@ -0,0 +1,984 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/list.h>
+#include "amdgpu.h"
+#include "amdgpu_aca.h"
+#include "amdgpu_ras.h"
+
+#define ACA_BANK_HWID(type, hwid, mcatype) [ACA_HWIP_TYPE_##type] = {hwid, mcatype}
+
+typedef int bank_handler_t(struct aca_handle *handle, struct aca_bank *bank, enum aca_smu_type type, void *data);
+
+static struct aca_hwip aca_hwid_mcatypes[ACA_HWIP_TYPE_COUNT] = {
+	ACA_BANK_HWID(SMU,	0x01,	0x01),
+	ACA_BANK_HWID(PCS_XGMI, 0x50,	0x00),
+	ACA_BANK_HWID(UMC,	0x96,	0x00),
+};
+
+static void aca_banks_init(struct aca_banks *banks)
+{
+	if (!banks)
+		return;
+
+	memset(banks, 0, sizeof(*banks));
+	INIT_LIST_HEAD(&banks->list);
+}
+
+static int aca_banks_add_bank(struct aca_banks *banks, struct aca_bank *bank)
+{
+	struct aca_bank_node *node;
+
+	if (!bank)
+		return -EINVAL;
+
+	node = kvzalloc(sizeof(*node), GFP_KERNEL);
+	if (!node)
+		return -ENOMEM;
+
+	memcpy(&node->bank, bank, sizeof(*bank));
+
+	INIT_LIST_HEAD(&node->node);
+	list_add_tail(&node->node, &banks->list);
+
+	banks->nr_banks++;
+
+	return 0;
+}
+
+static void aca_banks_release(struct aca_banks *banks)
+{
+	struct aca_bank_node *node, *tmp;
+
+	if (list_empty(&banks->list))
+		return;
+
+	list_for_each_entry_safe(node, tmp, &banks->list, node) {
+		list_del(&node->node);
+		kvfree(node);
+		banks->nr_banks--;
+	}
+}
+
+static int aca_smu_get_valid_aca_count(struct amdgpu_device *adev, enum aca_smu_type type, u32 *count)
+{
+	struct amdgpu_aca *aca = &adev->aca;
+	const struct aca_smu_funcs *smu_funcs = aca->smu_funcs;
+
+	if (!count)
+		return -EINVAL;
+
+	if (!smu_funcs || !smu_funcs->get_valid_aca_count)
+		return -EOPNOTSUPP;
+
+	return smu_funcs->get_valid_aca_count(adev, type, count);
+}
+
+static struct aca_regs_dump {
+	const char *name;
+	int reg_idx;
+} aca_regs[] = {
+	{"CONTROL",		ACA_REG_IDX_CTL},
+	{"STATUS",		ACA_REG_IDX_STATUS},
+	{"ADDR",		ACA_REG_IDX_ADDR},
+	{"MISC",		ACA_REG_IDX_MISC0},
+	{"CONFIG",		ACA_REG_IDX_CONFIG},
+	{"IPID",		ACA_REG_IDX_IPID},
+	{"SYND",		ACA_REG_IDX_SYND},
+	{"DESTAT",		ACA_REG_IDX_DESTAT},
+	{"DEADDR",		ACA_REG_IDX_DEADDR},
+	{"CONTROL_MASK",	ACA_REG_IDX_CTL_MASK},
+};
+
+static void aca_smu_bank_dump(struct amdgpu_device *adev, int idx, int total, struct aca_bank *bank,
+			      struct ras_query_context *qctx)
+{
+	u64 event_id = qctx ? qctx->evid.event_id : RAS_EVENT_INVALID_ID;
+	int i;
+
+	if (adev->debug_disable_ce_logs &&
+	    bank->smu_err_type == ACA_SMU_TYPE_CE &&
+	    !ACA_BANK_ERR_IS_DEFFERED(bank))
+		return;
+
+	RAS_EVENT_LOG(adev, event_id, HW_ERR "Accelerator Check Architecture events logged\n");
+	/* plus 1 for output format, e.g: ACA[08/08]: xxxx */
+	for (i = 0; i < ARRAY_SIZE(aca_regs); i++)
+		RAS_EVENT_LOG(adev, event_id, HW_ERR "ACA[%02d/%02d].%s=0x%016llx\n",
+			      idx + 1, total, aca_regs[i].name, bank->regs[aca_regs[i].reg_idx]);
+
+	if (ACA_REG__STATUS__SCRUB(bank->regs[ACA_REG_IDX_STATUS]))
+		RAS_EVENT_LOG(adev, event_id, HW_ERR "hardware error logged by the scrubber\n");
+}
+
+static bool aca_bank_hwip_is_matched(struct aca_bank *bank, enum aca_hwip_type type)
+{
+
+	struct aca_hwip *hwip;
+	int hwid, mcatype;
+	u64 ipid;
+
+	if (!bank || type == ACA_HWIP_TYPE_UNKNOW)
+		return false;
+
+	hwip = &aca_hwid_mcatypes[type];
+	if (!hwip->hwid)
+		return false;
+
+	ipid = bank->regs[ACA_REG_IDX_IPID];
+	hwid = ACA_REG__IPID__HARDWAREID(ipid);
+	mcatype = ACA_REG__IPID__MCATYPE(ipid);
+
+	return hwip->hwid == hwid && hwip->mcatype == mcatype;
+}
+
+static int aca_smu_get_valid_aca_banks(struct amdgpu_device *adev, enum aca_smu_type type,
+				       int start, int count,
+				       struct aca_banks *banks, struct ras_query_context *qctx)
+{
+	struct amdgpu_aca *aca = &adev->aca;
+	const struct aca_smu_funcs *smu_funcs = aca->smu_funcs;
+	struct aca_bank bank;
+	int i, max_count, ret;
+
+	if (!count)
+		return 0;
+
+	if (!smu_funcs || !smu_funcs->get_valid_aca_bank)
+		return -EOPNOTSUPP;
+
+	switch (type) {
+	case ACA_SMU_TYPE_UE:
+		max_count = smu_funcs->max_ue_bank_count;
+		break;
+	case ACA_SMU_TYPE_CE:
+		max_count = smu_funcs->max_ce_bank_count;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	if (start + count > max_count)
+		return -EINVAL;
+
+	count = min_t(int, count, max_count);
+	for (i = 0; i < count; i++) {
+		memset(&bank, 0, sizeof(bank));
+		ret = smu_funcs->get_valid_aca_bank(adev, type, start + i, &bank);
+		if (ret)
+			return ret;
+
+		bank.smu_err_type = type;
+
+		/*
+		 * Poison being consumed when injecting a UE while running background workloads,
+		 * which are unexpected.
+		 */
+		if (type == ACA_SMU_TYPE_UE &&
+		    ACA_REG__STATUS__POISON(bank.regs[ACA_REG_IDX_STATUS]) &&
+		    !aca_bank_hwip_is_matched(&bank, ACA_HWIP_TYPE_UMC))
+			continue;
+
+		aca_smu_bank_dump(adev, i, count, &bank, qctx);
+
+		ret = aca_banks_add_bank(banks, &bank);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static bool aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank, enum aca_smu_type type)
+{
+	const struct aca_bank_ops *bank_ops = handle->bank_ops;
+
+	/* Parse all deferred errors with UMC aca handle */
+	if (ACA_BANK_ERR_IS_DEFFERED(bank))
+		return handle->hwip == ACA_HWIP_TYPE_UMC;
+
+	if (!aca_bank_hwip_is_matched(bank, handle->hwip))
+		return false;
+
+	if (!bank_ops->aca_bank_is_valid)
+		return true;
+
+	return bank_ops->aca_bank_is_valid(handle, bank, type, handle->data);
+}
+
+static struct aca_bank_error *new_bank_error(struct aca_error *aerr, struct aca_bank_info *info)
+{
+	struct aca_bank_error *bank_error;
+
+	bank_error = kvzalloc(sizeof(*bank_error), GFP_KERNEL);
+	if (!bank_error)
+		return NULL;
+
+	INIT_LIST_HEAD(&bank_error->node);
+	memcpy(&bank_error->info, info, sizeof(*info));
+
+	mutex_lock(&aerr->lock);
+	list_add_tail(&bank_error->node, &aerr->list);
+	aerr->nr_errors++;
+	mutex_unlock(&aerr->lock);
+
+	return bank_error;
+}
+
+static struct aca_bank_error *find_bank_error(struct aca_error *aerr, struct aca_bank_info *info)
+{
+	struct aca_bank_error *bank_error = NULL;
+	struct aca_bank_info *tmp_info;
+	bool found = false;
+
+	mutex_lock(&aerr->lock);
+	list_for_each_entry(bank_error, &aerr->list, node) {
+		tmp_info = &bank_error->info;
+		if (tmp_info->socket_id == info->socket_id &&
+		    tmp_info->die_id == info->die_id) {
+			found = true;
+			goto out_unlock;
+		}
+	}
+
+out_unlock:
+	mutex_unlock(&aerr->lock);
+
+	return found ? bank_error : NULL;
+}
+
+static void aca_bank_error_remove(struct aca_error *aerr, struct aca_bank_error *bank_error)
+{
+	if (!aerr || !bank_error)
+		return;
+
+	list_del(&bank_error->node);
+	aerr->nr_errors--;
+
+	kvfree(bank_error);
+}
+
+static struct aca_bank_error *get_bank_error(struct aca_error *aerr, struct aca_bank_info *info)
+{
+	struct aca_bank_error *bank_error;
+
+	if (!aerr || !info)
+		return NULL;
+
+	bank_error = find_bank_error(aerr, info);
+	if (bank_error)
+		return bank_error;
+
+	return new_bank_error(aerr, info);
+}
+
+int aca_error_cache_log_bank_error(struct aca_handle *handle, struct aca_bank_info *info,
+				   enum aca_error_type type, u64 count)
+{
+	struct aca_error_cache *error_cache = &handle->error_cache;
+	struct aca_bank_error *bank_error;
+	struct aca_error *aerr;
+
+	if (!handle || !info || type >= ACA_ERROR_TYPE_COUNT)
+		return -EINVAL;
+
+	if (!count)
+		return 0;
+
+	aerr = &error_cache->errors[type];
+	bank_error = get_bank_error(aerr, info);
+	if (!bank_error)
+		return -ENOMEM;
+
+	bank_error->count += count;
+
+	return 0;
+}
+
+static int aca_bank_parser(struct aca_handle *handle, struct aca_bank *bank, enum aca_smu_type type)
+{
+	const struct aca_bank_ops *bank_ops = handle->bank_ops;
+
+	if (!bank)
+		return -EINVAL;
+
+	if (!bank_ops->aca_bank_parser)
+		return -EOPNOTSUPP;
+
+	return bank_ops->aca_bank_parser(handle, bank, type,
+					 handle->data);
+}
+
+static int handler_aca_log_bank_error(struct aca_handle *handle, struct aca_bank *bank,
+				      enum aca_smu_type type, void *data)
+{
+	int ret;
+
+	ret = aca_bank_parser(handle, bank, type);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static int aca_dispatch_bank(struct aca_handle_manager *mgr, struct aca_bank *bank,
+			     enum aca_smu_type type, bank_handler_t handler, void *data)
+{
+	struct aca_handle *handle;
+	int ret;
+
+	if (list_empty(&mgr->list))
+		return 0;
+
+	list_for_each_entry(handle, &mgr->list, node) {
+		if (!aca_bank_is_valid(handle, bank, type))
+			continue;
+
+		ret = handler(handle, bank, type, data);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int aca_dispatch_banks(struct aca_handle_manager *mgr, struct aca_banks *banks,
+			      enum aca_smu_type type, bank_handler_t handler, void *data)
+{
+	struct aca_bank_node *node;
+	struct aca_bank *bank;
+	int ret;
+
+	if (!mgr || !banks)
+		return -EINVAL;
+
+	/* pre check to avoid unnecessary operations */
+	if (list_empty(&mgr->list) || list_empty(&banks->list))
+		return 0;
+
+	list_for_each_entry(node, &banks->list, node) {
+		bank = &node->bank;
+
+		ret = aca_dispatch_bank(mgr, bank, type, handler, data);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static bool aca_bank_should_update(struct amdgpu_device *adev, enum aca_smu_type type)
+{
+	struct amdgpu_aca *aca = &adev->aca;
+	bool ret = true;
+
+	/*
+	 * Because the UE Valid MCA count will only be cleared after reset,
+	 * in order to avoid repeated counting of the error count,
+	 * the aca bank is only updated once during the gpu recovery stage.
+	 */
+	if (type == ACA_SMU_TYPE_UE) {
+		if (amdgpu_ras_intr_triggered())
+			ret = atomic_cmpxchg(&aca->ue_update_flag, 0, 1) == 0;
+		else
+			atomic_set(&aca->ue_update_flag, 0);
+	}
+
+	return ret;
+}
+
+static void aca_banks_generate_cper(struct amdgpu_device *adev,
+				    enum aca_smu_type type,
+				    struct aca_banks *banks,
+				    int count)
+{
+	struct aca_bank_node *node;
+	struct aca_bank *bank;
+	int r;
+
+	if (!adev->cper.enabled)
+		return;
+
+	if (!banks || !count) {
+		dev_warn(adev->dev, "fail to generate cper records\n");
+		return;
+	}
+
+	/* UEs must be encoded into separate CPER entries */
+	if (type == ACA_SMU_TYPE_UE) {
+		struct aca_banks de_banks;
+
+		aca_banks_init(&de_banks);
+		list_for_each_entry(node, &banks->list, node) {
+			bank = &node->bank;
+			if (bank->aca_err_type == ACA_ERROR_TYPE_DEFERRED) {
+				r = aca_banks_add_bank(&de_banks, bank);
+				if (r)
+					dev_warn(adev->dev, "fail to add de banks, ret = %d\n", r);
+			} else {
+				if (amdgpu_cper_generate_ue_record(adev, bank))
+					dev_warn(adev->dev, "fail to generate ue cper records\n");
+			}
+		}
+
+		if (!list_empty(&de_banks.list)) {
+			if (amdgpu_cper_generate_ce_records(adev, &de_banks, de_banks.nr_banks))
+				dev_warn(adev->dev, "fail to generate de cper records\n");
+		}
+
+		aca_banks_release(&de_banks);
+	} else {
+		/*
+		 * SMU_TYPE_CE banks are combined into 1 CPER entries,
+		 * they could be CEs or DEs or both
+		 */
+		if (amdgpu_cper_generate_ce_records(adev, banks, count))
+			dev_warn(adev->dev, "fail to generate ce cper records\n");
+	}
+}
+
+static int aca_banks_update(struct amdgpu_device *adev, enum aca_smu_type type,
+			    bank_handler_t handler, struct ras_query_context *qctx, void *data)
+{
+	struct amdgpu_aca *aca = &adev->aca;
+	struct aca_banks banks;
+	u32 count = 0;
+	int ret;
+
+	if (list_empty(&aca->mgr.list))
+		return 0;
+
+	if (!aca_bank_should_update(adev, type))
+		return 0;
+
+	ret = aca_smu_get_valid_aca_count(adev, type, &count);
+	if (ret)
+		return ret;
+
+	if (!count)
+		return 0;
+
+	aca_banks_init(&banks);
+
+	ret = aca_smu_get_valid_aca_banks(adev, type, 0, count, &banks, qctx);
+	if (ret)
+		goto err_release_banks;
+
+	if (list_empty(&banks.list)) {
+		ret = 0;
+		goto err_release_banks;
+	}
+
+	ret = aca_dispatch_banks(&aca->mgr, &banks, type,
+				 handler, data);
+	if (ret)
+		goto err_release_banks;
+
+	aca_banks_generate_cper(adev, type, &banks, count);
+
+err_release_banks:
+	aca_banks_release(&banks);
+
+	return ret;
+}
+
+static int aca_log_aca_error_data(struct aca_bank_error *bank_error, enum aca_error_type type, struct ras_err_data *err_data)
+{
+	struct aca_bank_info *info;
+	struct amdgpu_smuio_mcm_config_info mcm_info;
+	u64 count;
+
+	if (type >= ACA_ERROR_TYPE_COUNT)
+		return -EINVAL;
+
+	count = bank_error->count;
+	if (!count)
+		return 0;
+
+	info = &bank_error->info;
+	mcm_info.die_id = info->die_id;
+	mcm_info.socket_id = info->socket_id;
+
+	switch (type) {
+	case ACA_ERROR_TYPE_UE:
+		amdgpu_ras_error_statistic_ue_count(err_data, &mcm_info, count);
+		break;
+	case ACA_ERROR_TYPE_CE:
+		amdgpu_ras_error_statistic_ce_count(err_data, &mcm_info, count);
+		break;
+	case ACA_ERROR_TYPE_DEFERRED:
+		amdgpu_ras_error_statistic_de_count(err_data, &mcm_info, count);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int aca_log_aca_error(struct aca_handle *handle, enum aca_error_type type, struct ras_err_data *err_data)
+{
+	struct aca_error_cache *error_cache = &handle->error_cache;
+	struct aca_error *aerr = &error_cache->errors[type];
+	struct aca_bank_error *bank_error, *tmp;
+
+	mutex_lock(&aerr->lock);
+
+	if (list_empty(&aerr->list))
+		goto out_unlock;
+
+	list_for_each_entry_safe(bank_error, tmp, &aerr->list, node) {
+		aca_log_aca_error_data(bank_error, type, err_data);
+		aca_bank_error_remove(aerr, bank_error);
+	}
+
+out_unlock:
+	mutex_unlock(&aerr->lock);
+
+	return 0;
+}
+
+static int __aca_get_error_data(struct amdgpu_device *adev, struct aca_handle *handle, enum aca_error_type type,
+				struct ras_err_data *err_data, struct ras_query_context *qctx)
+{
+	enum aca_smu_type smu_type;
+	int ret;
+
+	switch (type) {
+	case ACA_ERROR_TYPE_UE:
+		smu_type = ACA_SMU_TYPE_UE;
+		break;
+	case ACA_ERROR_TYPE_CE:
+	case ACA_ERROR_TYPE_DEFERRED:
+		smu_type = ACA_SMU_TYPE_CE;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	/* update aca bank to aca source error_cache first */
+	ret = aca_banks_update(adev, smu_type, handler_aca_log_bank_error, qctx, NULL);
+	if (ret)
+		return ret;
+
+	/* DEs may contain in CEs or UEs */
+	if (type != ACA_ERROR_TYPE_DEFERRED)
+		aca_log_aca_error(handle, ACA_ERROR_TYPE_DEFERRED, err_data);
+
+	return aca_log_aca_error(handle, type, err_data);
+}
+
+static bool aca_handle_is_valid(struct aca_handle *handle)
+{
+	if (!handle->mask || !list_empty(&handle->node))
+		return false;
+
+	return true;
+}
+
+int amdgpu_aca_get_error_data(struct amdgpu_device *adev, struct aca_handle *handle,
+			      enum aca_error_type type, struct ras_err_data *err_data,
+			      struct ras_query_context *qctx)
+{
+	if (!handle || !err_data)
+		return -EINVAL;
+
+	if (aca_handle_is_valid(handle))
+		return -EOPNOTSUPP;
+
+	if ((type < 0) || (!(BIT(type) & handle->mask)))
+		return  0;
+
+	return __aca_get_error_data(adev, handle, type, err_data, qctx);
+}
+
+static void aca_error_init(struct aca_error *aerr, enum aca_error_type type)
+{
+	mutex_init(&aerr->lock);
+	INIT_LIST_HEAD(&aerr->list);
+	aerr->type = type;
+	aerr->nr_errors = 0;
+}
+
+static void aca_init_error_cache(struct aca_handle *handle)
+{
+	struct aca_error_cache *error_cache = &handle->error_cache;
+	int type;
+
+	for (type = ACA_ERROR_TYPE_UE; type < ACA_ERROR_TYPE_COUNT; type++)
+		aca_error_init(&error_cache->errors[type], type);
+}
+
+static void aca_error_fini(struct aca_error *aerr)
+{
+	struct aca_bank_error *bank_error, *tmp;
+
+	mutex_lock(&aerr->lock);
+	if (list_empty(&aerr->list))
+		goto out_unlock;
+
+	list_for_each_entry_safe(bank_error, tmp, &aerr->list, node)
+		aca_bank_error_remove(aerr, bank_error);
+
+out_unlock:
+	mutex_destroy(&aerr->lock);
+}
+
+static void aca_fini_error_cache(struct aca_handle *handle)
+{
+	struct aca_error_cache *error_cache = &handle->error_cache;
+	int type;
+
+	for (type = ACA_ERROR_TYPE_UE; type < ACA_ERROR_TYPE_COUNT; type++)
+		aca_error_fini(&error_cache->errors[type]);
+}
+
+static int add_aca_handle(struct amdgpu_device *adev, struct aca_handle_manager *mgr, struct aca_handle *handle,
+			  const char *name, const struct aca_info *ras_info, void *data)
+{
+	memset(handle, 0, sizeof(*handle));
+
+	handle->adev = adev;
+	handle->mgr = mgr;
+	handle->name = name;
+	handle->hwip = ras_info->hwip;
+	handle->mask = ras_info->mask;
+	handle->bank_ops = ras_info->bank_ops;
+	handle->data = data;
+	aca_init_error_cache(handle);
+
+	INIT_LIST_HEAD(&handle->node);
+	list_add_tail(&handle->node, &mgr->list);
+	mgr->nr_handles++;
+
+	return 0;
+}
+
+static ssize_t aca_sysfs_read(struct device *dev,
+			      struct device_attribute *attr, char *buf)
+{
+	struct aca_handle *handle = container_of(attr, struct aca_handle, aca_attr);
+
+	/* NOTE: the aca cache will be auto cleared once read,
+	 * So the driver should unify the query entry point, forward request to ras query interface directly */
+	return amdgpu_ras_aca_sysfs_read(dev, attr, handle, buf, handle->data);
+}
+
+static int add_aca_sysfs(struct amdgpu_device *adev, struct aca_handle *handle)
+{
+	struct device_attribute *aca_attr = &handle->aca_attr;
+
+	snprintf(handle->attr_name, sizeof(handle->attr_name) - 1, "aca_%s", handle->name);
+	aca_attr->show = aca_sysfs_read;
+	aca_attr->attr.name = handle->attr_name;
+	aca_attr->attr.mode = S_IRUGO;
+	sysfs_attr_init(&aca_attr->attr);
+
+	return sysfs_add_file_to_group(&adev->dev->kobj,
+				       &aca_attr->attr,
+				       "ras");
+}
+
+int amdgpu_aca_add_handle(struct amdgpu_device *adev, struct aca_handle *handle,
+			  const char *name, const struct aca_info *ras_info, void *data)
+{
+	struct amdgpu_aca *aca = &adev->aca;
+	int ret;
+
+	if (!amdgpu_aca_is_enabled(adev))
+		return 0;
+
+	ret = add_aca_handle(adev, &aca->mgr, handle, name, ras_info, data);
+	if (ret)
+		return ret;
+
+	return add_aca_sysfs(adev, handle);
+}
+
+static void remove_aca_handle(struct aca_handle *handle)
+{
+	struct aca_handle_manager *mgr = handle->mgr;
+
+	aca_fini_error_cache(handle);
+	list_del(&handle->node);
+	mgr->nr_handles--;
+}
+
+static void remove_aca_sysfs(struct aca_handle *handle)
+{
+	struct amdgpu_device *adev = handle->adev;
+	struct device_attribute *aca_attr = &handle->aca_attr;
+
+	if (adev->dev->kobj.sd)
+		sysfs_remove_file_from_group(&adev->dev->kobj,
+					     &aca_attr->attr,
+					     "ras");
+}
+
+void amdgpu_aca_remove_handle(struct aca_handle *handle)
+{
+	if (!handle || list_empty(&handle->node))
+		return;
+
+	remove_aca_sysfs(handle);
+	remove_aca_handle(handle);
+}
+
+static int aca_manager_init(struct aca_handle_manager *mgr)
+{
+	INIT_LIST_HEAD(&mgr->list);
+	mgr->nr_handles = 0;
+
+	return 0;
+}
+
+static void aca_manager_fini(struct aca_handle_manager *mgr)
+{
+	struct aca_handle *handle, *tmp;
+
+	if (list_empty(&mgr->list))
+		return;
+
+	list_for_each_entry_safe(handle, tmp, &mgr->list, node)
+		amdgpu_aca_remove_handle(handle);
+}
+
+bool amdgpu_aca_is_enabled(struct amdgpu_device *adev)
+{
+	return (adev->aca.is_enabled ||
+		adev->debug_enable_ras_aca);
+}
+
+int amdgpu_aca_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_aca *aca = &adev->aca;
+	int ret;
+
+	atomic_set(&aca->ue_update_flag, 0);
+
+	ret = aca_manager_init(&aca->mgr);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+void amdgpu_aca_fini(struct amdgpu_device *adev)
+{
+	struct amdgpu_aca *aca = &adev->aca;
+
+	aca_manager_fini(&aca->mgr);
+
+	atomic_set(&aca->ue_update_flag, 0);
+}
+
+int amdgpu_aca_reset(struct amdgpu_device *adev)
+{
+	struct amdgpu_aca *aca = &adev->aca;
+
+	atomic_set(&aca->ue_update_flag, 0);
+
+	return 0;
+}
+
+void amdgpu_aca_set_smu_funcs(struct amdgpu_device *adev, const struct aca_smu_funcs *smu_funcs)
+{
+	struct amdgpu_aca *aca = &adev->aca;
+
+	WARN_ON(aca->smu_funcs);
+	aca->smu_funcs = smu_funcs;
+}
+
+int aca_bank_info_decode(struct aca_bank *bank, struct aca_bank_info *info)
+{
+	u64 ipid;
+	u32 instidhi, instidlo;
+
+	if (!bank || !info)
+		return -EINVAL;
+
+	ipid = bank->regs[ACA_REG_IDX_IPID];
+	info->hwid = ACA_REG__IPID__HARDWAREID(ipid);
+	info->mcatype = ACA_REG__IPID__MCATYPE(ipid);
+	/*
+	 * Unfied DieID Format: SAASS. A:AID, S:Socket.
+	 * Unfied DieID[4:4] = InstanceId[0:0]
+	 * Unfied DieID[0:3] = InstanceIdHi[0:3]
+	 */
+	instidhi = ACA_REG__IPID__INSTANCEIDHI(ipid);
+	instidlo = ACA_REG__IPID__INSTANCEIDLO(ipid);
+	info->die_id = ((instidhi >> 2) & 0x03);
+	info->socket_id = ((instidlo & 0x1) << 2) | (instidhi & 0x03);
+
+	return 0;
+}
+
+static int aca_bank_get_error_code(struct amdgpu_device *adev, struct aca_bank *bank)
+{
+	struct amdgpu_aca *aca = &adev->aca;
+	const struct aca_smu_funcs *smu_funcs = aca->smu_funcs;
+
+	if (!smu_funcs || !smu_funcs->parse_error_code)
+		return -EOPNOTSUPP;
+
+	return smu_funcs->parse_error_code(adev, bank);
+}
+
+int aca_bank_check_error_codes(struct amdgpu_device *adev, struct aca_bank *bank, int *err_codes, int size)
+{
+	int i, error_code;
+
+	if (!bank || !err_codes)
+		return -EINVAL;
+
+	error_code = aca_bank_get_error_code(adev, bank);
+	if (error_code < 0)
+		return error_code;
+
+	for (i = 0; i < size; i++) {
+		if (err_codes[i] == error_code)
+			return 0;
+	}
+
+	return -EINVAL;
+}
+
+int amdgpu_aca_smu_set_debug_mode(struct amdgpu_device *adev, bool en)
+{
+	struct amdgpu_aca *aca = &adev->aca;
+	const struct aca_smu_funcs *smu_funcs = aca->smu_funcs;
+
+	if (!smu_funcs || !smu_funcs->set_debug_mode)
+		return -EOPNOTSUPP;
+
+	return smu_funcs->set_debug_mode(adev, en);
+}
+
+#if defined(CONFIG_DEBUG_FS)
+static int amdgpu_aca_smu_debug_mode_set(void *data, u64 val)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)data;
+	int ret;
+
+	ret = amdgpu_ras_set_aca_debug_mode(adev, val ? true : false);
+	if (ret)
+		return ret;
+
+	dev_info(adev->dev, "amdgpu set smu aca debug mode %s success\n", val ? "on" : "off");
+
+	return 0;
+}
+
+static void aca_dump_entry(struct seq_file *m, struct aca_bank *bank, enum aca_smu_type type, int idx)
+{
+	struct aca_bank_info info;
+	int i, ret;
+
+	ret = aca_bank_info_decode(bank, &info);
+	if (ret)
+		return;
+
+	seq_printf(m, "aca entry[%d].type: %s\n", idx, type ==  ACA_SMU_TYPE_UE ? "UE" : "CE");
+	seq_printf(m, "aca entry[%d].info: socketid:%d aid:%d hwid:0x%03x mcatype:0x%04x\n",
+		   idx, info.socket_id, info.die_id, info.hwid, info.mcatype);
+
+	for (i = 0; i < ARRAY_SIZE(aca_regs); i++)
+		seq_printf(m, "aca entry[%d].regs[%d]: 0x%016llx\n", idx, aca_regs[i].reg_idx, bank->regs[aca_regs[i].reg_idx]);
+}
+
+struct aca_dump_context {
+	struct seq_file *m;
+	int idx;
+};
+
+static int handler_aca_bank_dump(struct aca_handle *handle, struct aca_bank *bank,
+				 enum aca_smu_type type, void *data)
+{
+	struct aca_dump_context *ctx = (struct aca_dump_context *)data;
+
+	aca_dump_entry(ctx->m, bank, type, ctx->idx++);
+
+	return handler_aca_log_bank_error(handle, bank, type, NULL);
+}
+
+static int aca_dump_show(struct seq_file *m, enum aca_smu_type type)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)m->private;
+	struct aca_dump_context context = {
+		.m = m,
+		.idx = 0,
+	};
+
+	return aca_banks_update(adev, type, handler_aca_bank_dump, NULL, (void *)&context);
+}
+
+static int aca_dump_ce_show(struct seq_file *m, void *unused)
+{
+	return aca_dump_show(m, ACA_SMU_TYPE_CE);
+}
+
+static int aca_dump_ce_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, aca_dump_ce_show, inode->i_private);
+}
+
+static const struct file_operations aca_ce_dump_debug_fops = {
+	.owner = THIS_MODULE,
+	.open = aca_dump_ce_open,
+	.read = seq_read,
+	.llseek = seq_lseek,
+	.release = single_release,
+};
+
+static int aca_dump_ue_show(struct seq_file *m, void *unused)
+{
+	return aca_dump_show(m, ACA_SMU_TYPE_UE);
+}
+
+static int aca_dump_ue_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, aca_dump_ue_show, inode->i_private);
+}
+
+static const struct file_operations aca_ue_dump_debug_fops = {
+	.owner = THIS_MODULE,
+	.open = aca_dump_ue_open,
+	.read = seq_read,
+	.llseek = seq_lseek,
+	.release = single_release,
+};
+
+DEFINE_DEBUGFS_ATTRIBUTE(aca_debug_mode_fops, NULL, amdgpu_aca_smu_debug_mode_set, "%llu\n");
+#endif
+
+void amdgpu_aca_smu_debugfs_init(struct amdgpu_device *adev, struct dentry *root)
+{
+#if defined(CONFIG_DEBUG_FS)
+	if (!root)
+		return;
+
+	debugfs_create_file("aca_debug_mode", 0200, root, adev, &aca_debug_mode_fops);
+	debugfs_create_file("aca_ue_dump", 0400, root, adev, &aca_ue_dump_debug_fops);
+	debugfs_create_file("aca_ce_dump", 0400, root, adev, &aca_ce_dump_debug_fops);
+#endif
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h
new file mode 100644
index 000000000000..38c88897e1ec
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h
@@ -0,0 +1,232 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_ACA_H__
+#define __AMDGPU_ACA_H__
+
+#include <linux/list.h>
+
+struct ras_err_data;
+struct ras_query_context;
+
+#define ACA_MAX_REGS_COUNT	(16)
+
+#define ACA_REG_FIELD(x, h, l)			(((x) & GENMASK_ULL(h, l)) >> l)
+#define ACA_REG__STATUS__VAL(x)			ACA_REG_FIELD(x, 63, 63)
+#define ACA_REG__STATUS__OVERFLOW(x)		ACA_REG_FIELD(x, 62, 62)
+#define ACA_REG__STATUS__UC(x)			ACA_REG_FIELD(x, 61, 61)
+#define ACA_REG__STATUS__EN(x)			ACA_REG_FIELD(x, 60, 60)
+#define ACA_REG__STATUS__MISCV(x)		ACA_REG_FIELD(x, 59, 59)
+#define ACA_REG__STATUS__ADDRV(x)		ACA_REG_FIELD(x, 58, 58)
+#define ACA_REG__STATUS__PCC(x)			ACA_REG_FIELD(x, 57, 57)
+#define ACA_REG__STATUS__ERRCOREIDVAL(x)	ACA_REG_FIELD(x, 56, 56)
+#define ACA_REG__STATUS__TCC(x)			ACA_REG_FIELD(x, 55, 55)
+#define ACA_REG__STATUS__SYNDV(x)		ACA_REG_FIELD(x, 53, 53)
+#define ACA_REG__STATUS__CECC(x)		ACA_REG_FIELD(x, 46, 46)
+#define ACA_REG__STATUS__UECC(x)		ACA_REG_FIELD(x, 45, 45)
+#define ACA_REG__STATUS__DEFERRED(x)		ACA_REG_FIELD(x, 44, 44)
+#define ACA_REG__STATUS__POISON(x)		ACA_REG_FIELD(x, 43, 43)
+#define ACA_REG__STATUS__SCRUB(x)		ACA_REG_FIELD(x, 40, 40)
+#define ACA_REG__STATUS__ERRCOREID(x)		ACA_REG_FIELD(x, 37, 32)
+#define ACA_REG__STATUS__ADDRLSB(x)		ACA_REG_FIELD(x, 29, 24)
+#define ACA_REG__STATUS__ERRORCODEEXT(x)	ACA_REG_FIELD(x, 21, 16)
+#define ACA_REG__STATUS__ERRORCODE(x)		ACA_REG_FIELD(x, 15, 0)
+
+#define ACA_REG__IPID__MCATYPE(x)		ACA_REG_FIELD(x, 63, 48)
+#define ACA_REG__IPID__INSTANCEIDHI(x)		ACA_REG_FIELD(x, 47, 44)
+#define ACA_REG__IPID__HARDWAREID(x)		ACA_REG_FIELD(x, 43, 32)
+#define ACA_REG__IPID__INSTANCEIDLO(x)		ACA_REG_FIELD(x, 31, 0)
+
+#define ACA_REG__MISC0__VALID(x)		ACA_REG_FIELD(x, 63, 63)
+#define ACA_REG__MISC0__OVRFLW(x)		ACA_REG_FIELD(x, 48, 48)
+#define ACA_REG__MISC0__ERRCNT(x)		ACA_REG_FIELD(x, 43, 32)
+
+#define ACA_REG__SYND__ERRORINFORMATION(x)	ACA_REG_FIELD(x, 17, 0)
+
+/* NOTE: The following codes refers to the smu header file */
+#define ACA_EXTERROR_CODE_CE			0x3a
+#define ACA_EXTERROR_CODE_FAULT			0x3b
+
+#define ACA_ERROR_UE_MASK		BIT_MASK(ACA_ERROR_TYPE_UE)
+#define ACA_ERROR_CE_MASK		BIT_MASK(ACA_ERROR_TYPE_CE)
+#define ACA_ERROR_DEFERRED_MASK		BIT_MASK(ACA_ERROR_TYPE_DEFERRED)
+
+#define mmSMNAID_AID0_MCA_SMU		0x03b30400	/* SMN AID AID0 */
+#define mmSMNAID_XCD0_MCA_SMU		0x36430400	/* SMN AID XCD0 */
+#define mmSMNAID_XCD1_MCA_SMU		0x38430400	/* SMN AID XCD1 */
+#define mmSMNXCD_XCD0_MCA_SMU		0x40430400	/* SMN XCD XCD0 */
+
+#define ACA_BANK_ERR_IS_DEFFERED(bank)                                \
+	(ACA_REG__STATUS__POISON((bank)->regs[ACA_REG_IDX_STATUS]) || \
+	 ACA_REG__STATUS__DEFERRED((bank)->regs[ACA_REG_IDX_STATUS]))
+
+enum aca_reg_idx {
+	ACA_REG_IDX_CTL			= 0,
+	ACA_REG_IDX_STATUS		= 1,
+	ACA_REG_IDX_ADDR		= 2,
+	ACA_REG_IDX_MISC0		= 3,
+	ACA_REG_IDX_CONFIG		= 4,
+	ACA_REG_IDX_IPID		= 5,
+	ACA_REG_IDX_SYND		= 6,
+	ACA_REG_IDX_DESTAT		= 8,
+	ACA_REG_IDX_DEADDR		= 9,
+	ACA_REG_IDX_CTL_MASK		= 10,
+	ACA_REG_IDX_COUNT		= 16,
+};
+
+enum aca_hwip_type {
+	ACA_HWIP_TYPE_UNKNOW = -1,
+	ACA_HWIP_TYPE_PSP = 0,
+	ACA_HWIP_TYPE_UMC,
+	ACA_HWIP_TYPE_SMU,
+	ACA_HWIP_TYPE_PCS_XGMI,
+	ACA_HWIP_TYPE_COUNT,
+};
+
+enum aca_error_type {
+	ACA_ERROR_TYPE_INVALID = -1,
+	ACA_ERROR_TYPE_UE = 0,
+	ACA_ERROR_TYPE_CE,
+	ACA_ERROR_TYPE_DEFERRED,
+	ACA_ERROR_TYPE_COUNT
+};
+
+enum aca_smu_type {
+	ACA_SMU_TYPE_INVALID = -1,
+	ACA_SMU_TYPE_UE = 0,
+	ACA_SMU_TYPE_CE,
+	ACA_SMU_TYPE_COUNT,
+};
+
+struct aca_hwip {
+	int hwid;
+	int mcatype;
+};
+
+struct aca_bank {
+	enum aca_error_type aca_err_type;
+	enum aca_smu_type smu_err_type;
+	u64 regs[ACA_MAX_REGS_COUNT];
+};
+
+struct aca_bank_node {
+	struct aca_bank bank;
+	struct list_head node;
+};
+
+struct aca_banks {
+	int nr_banks;
+	struct list_head list;
+};
+
+struct aca_bank_info {
+	int die_id;
+	int socket_id;
+	int hwid;
+	int mcatype;
+};
+
+struct aca_bank_error {
+	struct list_head node;
+	struct aca_bank_info info;
+	u64 count;
+};
+
+struct aca_error {
+	struct list_head list;
+	struct mutex lock;
+	enum aca_error_type type;
+	int nr_errors;
+};
+
+struct aca_handle_manager {
+	struct list_head list;
+	int nr_handles;
+};
+
+struct aca_error_cache {
+	struct aca_error errors[ACA_ERROR_TYPE_COUNT];
+};
+
+struct aca_handle {
+	struct list_head node;
+	enum aca_hwip_type hwip;
+	struct amdgpu_device *adev;
+	struct aca_handle_manager *mgr;
+	struct aca_error_cache error_cache;
+	const struct aca_bank_ops *bank_ops;
+	struct device_attribute aca_attr;
+	char attr_name[64];
+	const char *name;
+	u32 mask;
+	void *data;
+};
+
+struct aca_bank_ops {
+	int (*aca_bank_parser)(struct aca_handle *handle, struct aca_bank *bank, enum aca_smu_type type, void *data);
+	bool (*aca_bank_is_valid)(struct aca_handle *handle, struct aca_bank *bank, enum aca_smu_type type,
+				  void *data);
+};
+
+struct aca_smu_funcs {
+	int max_ue_bank_count;
+	int max_ce_bank_count;
+	int (*set_debug_mode)(struct amdgpu_device *adev, bool enable);
+	int (*get_valid_aca_count)(struct amdgpu_device *adev, enum aca_smu_type type, u32 *count);
+	int (*get_valid_aca_bank)(struct amdgpu_device *adev, enum aca_smu_type type, int idx, struct aca_bank *bank);
+	int (*parse_error_code)(struct amdgpu_device *adev, struct aca_bank *bank);
+};
+
+struct amdgpu_aca {
+	struct aca_handle_manager mgr;
+	const struct aca_smu_funcs *smu_funcs;
+	atomic_t ue_update_flag;
+	bool is_enabled;
+};
+
+struct aca_info {
+	enum aca_hwip_type hwip;
+	const struct aca_bank_ops *bank_ops;
+	u32 mask;
+};
+
+int amdgpu_aca_init(struct amdgpu_device *adev);
+void amdgpu_aca_fini(struct amdgpu_device *adev);
+int amdgpu_aca_reset(struct amdgpu_device *adev);
+void amdgpu_aca_set_smu_funcs(struct amdgpu_device *adev, const struct aca_smu_funcs *smu_funcs);
+bool amdgpu_aca_is_enabled(struct amdgpu_device *adev);
+
+int aca_bank_info_decode(struct aca_bank *bank, struct aca_bank_info *info);
+int aca_bank_check_error_codes(struct amdgpu_device *adev, struct aca_bank *bank, int *err_codes, int size);
+
+int amdgpu_aca_add_handle(struct amdgpu_device *adev, struct aca_handle *handle,
+			  const char *name, const struct aca_info *aca_info, void *data);
+void amdgpu_aca_remove_handle(struct aca_handle *handle);
+int amdgpu_aca_get_error_data(struct amdgpu_device *adev, struct aca_handle *handle,
+			      enum aca_error_type type, struct ras_err_data *err_data,
+			      struct ras_query_context *qctx);
+int amdgpu_aca_smu_set_debug_mode(struct amdgpu_device *adev, bool en);
+void amdgpu_aca_smu_debugfs_init(struct amdgpu_device *adev, struct dentry *root);
+int aca_error_cache_log_bank_error(struct aca_handle *handle, struct aca_bank_info *info,
+				   enum aca_error_type type, u64 count);
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c
index a52795d9b458..4926996f94da 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c
@@ -24,10 +24,13 @@
  */
 
 #include <linux/irqdomain.h>
+#include <linux/pci.h>
 #include <linux/pm_domain.h>
 #include <linux/platform_device.h>
 #include <sound/designware_i2s.h>
 #include <sound/pcm.h>
+#include <linux/acpi.h>
+#include <linux/dmi.h>
 
 #include "amdgpu.h"
 #include "atom.h"
@@ -35,41 +38,57 @@
 
 #include "acp_gfx_if.h"
 
-#define ACP_TILE_ON_MASK                0x03
-#define ACP_TILE_OFF_MASK               0x02
-#define ACP_TILE_ON_RETAIN_REG_MASK     0x1f
-#define ACP_TILE_OFF_RETAIN_REG_MASK    0x20
-
-#define ACP_TILE_P1_MASK                0x3e
-#define ACP_TILE_P2_MASK                0x3d
-#define ACP_TILE_DSP0_MASK              0x3b
-#define ACP_TILE_DSP1_MASK              0x37
-
-#define ACP_TILE_DSP2_MASK              0x2f
-
-#define ACP_DMA_REGS_END		0x146c0
-#define ACP_I2S_PLAY_REGS_START		0x14840
-#define ACP_I2S_PLAY_REGS_END		0x148b4
-#define ACP_I2S_CAP_REGS_START		0x148b8
-#define ACP_I2S_CAP_REGS_END		0x1496c
-
-#define ACP_I2S_COMP1_CAP_REG_OFFSET	0xac
-#define ACP_I2S_COMP2_CAP_REG_OFFSET	0xa8
-#define ACP_I2S_COMP1_PLAY_REG_OFFSET	0x6c
-#define ACP_I2S_COMP2_PLAY_REG_OFFSET	0x68
-
-#define mmACP_PGFSM_RETAIN_REG		0x51c9
-#define mmACP_PGFSM_CONFIG_REG		0x51ca
-#define mmACP_PGFSM_READ_REG_0		0x51cc
-
-#define mmACP_MEM_SHUT_DOWN_REQ_LO	0x51f8
-#define mmACP_MEM_SHUT_DOWN_REQ_HI	0x51f9
-#define mmACP_MEM_SHUT_DOWN_STS_LO	0x51fa
-#define mmACP_MEM_SHUT_DOWN_STS_HI	0x51fb
-
-#define ACP_TIMEOUT_LOOP		0x000000FF
-#define ACP_DEVS			3
-#define ACP_SRC_ID			162
+#define ST_JADEITE 1
+#define ACP_TILE_ON_MASK			0x03
+#define ACP_TILE_OFF_MASK			0x02
+#define ACP_TILE_ON_RETAIN_REG_MASK		0x1f
+#define ACP_TILE_OFF_RETAIN_REG_MASK		0x20
+
+#define ACP_TILE_P1_MASK			0x3e
+#define ACP_TILE_P2_MASK			0x3d
+#define ACP_TILE_DSP0_MASK			0x3b
+#define ACP_TILE_DSP1_MASK			0x37
+
+#define ACP_TILE_DSP2_MASK			0x2f
+
+#define ACP_DMA_REGS_END			0x146c0
+#define ACP_I2S_PLAY_REGS_START			0x14840
+#define ACP_I2S_PLAY_REGS_END			0x148b4
+#define ACP_I2S_CAP_REGS_START			0x148b8
+#define ACP_I2S_CAP_REGS_END			0x1496c
+
+#define ACP_I2S_COMP1_CAP_REG_OFFSET		0xac
+#define ACP_I2S_COMP2_CAP_REG_OFFSET		0xa8
+#define ACP_I2S_COMP1_PLAY_REG_OFFSET		0x6c
+#define ACP_I2S_COMP2_PLAY_REG_OFFSET		0x68
+#define ACP_BT_PLAY_REGS_START			0x14970
+#define ACP_BT_PLAY_REGS_END			0x14a24
+#define ACP_BT_COMP1_REG_OFFSET			0xac
+#define ACP_BT_COMP2_REG_OFFSET			0xa8
+
+#define mmACP_PGFSM_RETAIN_REG			0x51c9
+#define mmACP_PGFSM_CONFIG_REG			0x51ca
+#define mmACP_PGFSM_READ_REG_0			0x51cc
+
+#define mmACP_MEM_SHUT_DOWN_REQ_LO		0x51f8
+#define mmACP_MEM_SHUT_DOWN_REQ_HI		0x51f9
+#define mmACP_MEM_SHUT_DOWN_STS_LO		0x51fa
+#define mmACP_MEM_SHUT_DOWN_STS_HI		0x51fb
+
+#define mmACP_CONTROL				0x5131
+#define mmACP_STATUS				0x5133
+#define mmACP_SOFT_RESET			0x5134
+#define ACP_CONTROL__ClkEn_MASK			0x1
+#define ACP_SOFT_RESET__SoftResetAud_MASK	0x100
+#define ACP_SOFT_RESET__SoftResetAudDone_MASK	0x1000000
+#define ACP_CLOCK_EN_TIME_OUT_VALUE		0x000000FF
+#define ACP_SOFT_RESET_DONE_TIME_OUT_VALUE	0x000000FF
+
+#define ACP_TIMEOUT_LOOP			0x000000FF
+#define ACP_DEVS				4
+#define ACP_SRC_ID				162
+
+static unsigned long acp_machine_id;
 
 enum {
 	ACP_TILE_P1 = 0,
@@ -79,9 +98,9 @@ enum {
 	ACP_TILE_DSP2,
 };
 
-static int acp_sw_init(void *handle)
+static int acp_sw_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	adev->acp.parent = adev->dev;
 
@@ -93,9 +112,9 @@ static int acp_sw_init(void *handle)
 	return 0;
 }
 
-static int acp_sw_fini(void *handle)
+static int acp_sw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (adev->acp.cgs_device)
 		amdgpu_cgs_destroy_device(adev->acp.cgs_device);
@@ -103,386 +122,487 @@ static int acp_sw_fini(void *handle)
 	return 0;
 }
 
-/* power off a tile/block within ACP */
-static int acp_suspend_tile(void *cgs_dev, int tile)
-{
-	u32 val = 0;
-	u32 count = 0;
-
-	if ((tile  < ACP_TILE_P1) || (tile > ACP_TILE_DSP2)) {
-		pr_err("Invalid ACP tile : %d to suspend\n", tile);
-		return -1;
-	}
-
-	val = cgs_read_register(cgs_dev, mmACP_PGFSM_READ_REG_0 + tile);
-	val &= ACP_TILE_ON_MASK;
-
-	if (val == 0x0) {
-		val = cgs_read_register(cgs_dev, mmACP_PGFSM_RETAIN_REG);
-		val = val | (1 << tile);
-		cgs_write_register(cgs_dev, mmACP_PGFSM_RETAIN_REG, val);
-		cgs_write_register(cgs_dev, mmACP_PGFSM_CONFIG_REG,
-					0x500 + tile);
-
-		count = ACP_TIMEOUT_LOOP;
-		while (true) {
-			val = cgs_read_register(cgs_dev, mmACP_PGFSM_READ_REG_0
-								+ tile);
-			val = val & ACP_TILE_ON_MASK;
-			if (val == ACP_TILE_OFF_MASK)
-				break;
-			if (--count == 0) {
-				pr_err("Timeout reading ACP PGFSM status\n");
-				return -ETIMEDOUT;
-			}
-			udelay(100);
-		}
-
-		val = cgs_read_register(cgs_dev, mmACP_PGFSM_RETAIN_REG);
-
-		val |= ACP_TILE_OFF_RETAIN_REG_MASK;
-		cgs_write_register(cgs_dev, mmACP_PGFSM_RETAIN_REG, val);
-	}
-	return 0;
-}
-
-/* power on a tile/block within ACP */
-static int acp_resume_tile(void *cgs_dev, int tile)
-{
-	u32 val = 0;
-	u32 count = 0;
-
-	if ((tile  < ACP_TILE_P1) || (tile > ACP_TILE_DSP2)) {
-		pr_err("Invalid ACP tile to resume\n");
-		return -1;
-	}
-
-	val = cgs_read_register(cgs_dev, mmACP_PGFSM_READ_REG_0 + tile);
-	val = val & ACP_TILE_ON_MASK;
-
-	if (val != 0x0) {
-		cgs_write_register(cgs_dev, mmACP_PGFSM_CONFIG_REG,
-					0x600 + tile);
-		count = ACP_TIMEOUT_LOOP;
-		while (true) {
-			val = cgs_read_register(cgs_dev, mmACP_PGFSM_READ_REG_0
-							+ tile);
-			val = val & ACP_TILE_ON_MASK;
-			if (val == 0x0)
-				break;
-			if (--count == 0) {
-				pr_err("Timeout reading ACP PGFSM status\n");
-				return -ETIMEDOUT;
-			}
-			udelay(100);
-		}
-		val = cgs_read_register(cgs_dev, mmACP_PGFSM_RETAIN_REG);
-		if (tile == ACP_TILE_P1)
-			val = val & (ACP_TILE_P1_MASK);
-		else if (tile == ACP_TILE_P2)
-			val = val & (ACP_TILE_P2_MASK);
-
-		cgs_write_register(cgs_dev, mmACP_PGFSM_RETAIN_REG, val);
-	}
-	return 0;
-}
-
 struct acp_pm_domain {
-	void *cgs_dev;
+	void *adev;
 	struct generic_pm_domain gpd;
 };
 
 static int acp_poweroff(struct generic_pm_domain *genpd)
 {
-	int i, ret;
 	struct acp_pm_domain *apd;
+	struct amdgpu_device *adev;
 
 	apd = container_of(genpd, struct acp_pm_domain, gpd);
-	if (apd != NULL) {
-		/* Donot return abruptly if any of power tile fails to suspend.
-		 * Log it and continue powering off other tile
-		 */
-		for (i = 4; i >= 0 ; i--) {
-			ret = acp_suspend_tile(apd->cgs_dev, ACP_TILE_P1 + i);
-			if (ret)
-				pr_err("ACP tile %d tile suspend failed\n", i);
-		}
-	}
+	adev = apd->adev;
+	/* call smu to POWER GATE ACP block
+	 * smu will
+	 * 1. turn off the acp clock
+	 * 2. power off the acp tiles
+	 * 3. check and enter ulv state
+	 */
+	amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_ACP, true, 0);
 	return 0;
 }
 
 static int acp_poweron(struct generic_pm_domain *genpd)
 {
-	int i, ret;
 	struct acp_pm_domain *apd;
+	struct amdgpu_device *adev;
 
 	apd = container_of(genpd, struct acp_pm_domain, gpd);
-	if (apd != NULL) {
-		for (i = 0; i < 2; i++) {
-			ret = acp_resume_tile(apd->cgs_dev, ACP_TILE_P1 + i);
-			if (ret) {
-				pr_err("ACP tile %d resume failed\n", i);
-				break;
-			}
-		}
-
-		/* Disable DSPs which are not going to be used */
-		for (i = 0; i < 3; i++) {
-			ret = acp_suspend_tile(apd->cgs_dev, ACP_TILE_DSP0 + i);
-			/* Continue suspending other DSP, even if one fails */
-			if (ret)
-				pr_err("ACP DSP %d suspend failed\n", i);
-		}
-	}
+	adev = apd->adev;
+	/* call smu to UNGATE ACP block
+	 * smu will
+	 * 1. exit ulv
+	 * 2. turn on acp clock
+	 * 3. power on acp tiles
+	 */
+	amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_ACP, false, 0);
 	return 0;
 }
 
-static struct device *get_mfd_cell_dev(const char *device_name, int r)
+static int acp_genpd_add_device(struct device *dev, void *data)
+{
+	struct generic_pm_domain *gpd = data;
+	int ret;
+
+	ret = pm_genpd_add_device(gpd, dev);
+	if (ret)
+		dev_err(dev, "Failed to add dev to genpd %d\n", ret);
+
+	return ret;
+}
+
+static int acp_genpd_remove_device(struct device *dev, void *data)
 {
-	char auto_dev_name[25];
-	struct device *dev;
+	int ret;
 
-	snprintf(auto_dev_name, sizeof(auto_dev_name),
-		 "%s.%d.auto", device_name, r);
-	dev = bus_find_device_by_name(&platform_bus_type, NULL, auto_dev_name);
-	dev_info(dev, "device %s added to pm domain\n", auto_dev_name);
+	ret = pm_genpd_remove_device(dev);
+	if (ret)
+		dev_err(dev, "Failed to remove dev from genpd %d\n", ret);
 
-	return dev;
+	/* Continue to remove */
+	return 0;
+}
+
+static int acp_quirk_cb(const struct dmi_system_id *id)
+{
+	acp_machine_id = ST_JADEITE;
+	return 1;
 }
 
+static const struct dmi_system_id acp_quirk_table[] = {
+	{
+		.callback = acp_quirk_cb,
+		.matches = {
+			DMI_EXACT_MATCH(DMI_BOARD_VENDOR, "AMD"),
+			DMI_EXACT_MATCH(DMI_PRODUCT_NAME, "Jadeite"),
+		}
+	},
+	{
+		.callback = acp_quirk_cb,
+		.matches = {
+			DMI_EXACT_MATCH(DMI_BOARD_VENDOR, "IP3 Technology CO.,Ltd."),
+			DMI_EXACT_MATCH(DMI_PRODUCT_NAME, "ASN1D"),
+		},
+	},
+	{
+		.callback = acp_quirk_cb,
+		.matches = {
+			DMI_EXACT_MATCH(DMI_BOARD_VENDOR, "Standard"),
+			DMI_EXACT_MATCH(DMI_PRODUCT_NAME, "ASN10"),
+		},
+	},
+	{}
+};
+
 /**
  * acp_hw_init - start and test ACP block
  *
- * @adev: amdgpu_device pointer
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
  *
  */
-static int acp_hw_init(void *handle)
+static int acp_hw_init(struct amdgpu_ip_block *ip_block)
 {
-	int r, i;
-	uint64_t acp_base;
-	struct device *dev;
-	struct i2s_platform_data *i2s_pdata;
-
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int r;
+	u64 acp_base;
+	u32 val = 0;
+	u32 count = 0;
+	struct i2s_platform_data *i2s_pdata = NULL;
 
-	const struct amdgpu_ip_block *ip_block =
-		amdgpu_get_ip_block(adev, AMD_IP_BLOCK_TYPE_ACP);
-
-	if (!ip_block)
-		return -EINVAL;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	r = amd_acp_hw_init(adev->acp.cgs_device,
 			    ip_block->version->major, ip_block->version->minor);
 	/* -ENODEV means board uses AZ rather than ACP */
-	if (r == -ENODEV)
-		return 0;
-	else if (r)
-		return r;
-
-	r = cgs_get_pci_resource(adev->acp.cgs_device, CGS_RESOURCE_TYPE_MMIO,
-			0x5289, 0, &acp_base);
-	if (r == -ENODEV)
+	if (r == -ENODEV) {
+		amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_ACP, true, 0);
 		return 0;
-	else if (r)
+	} else if (r) {
 		return r;
-	if (adev->asic_type != CHIP_STONEY) {
-		adev->acp.acp_genpd = kzalloc(sizeof(struct acp_pm_domain), GFP_KERNEL);
-		if (adev->acp.acp_genpd == NULL)
-			return -ENOMEM;
-
-		adev->acp.acp_genpd->gpd.name = "ACP_AUDIO";
-		adev->acp.acp_genpd->gpd.power_off = acp_poweroff;
-		adev->acp.acp_genpd->gpd.power_on = acp_poweron;
-
-
-		adev->acp.acp_genpd->cgs_dev = adev->acp.cgs_device;
-
-		pm_genpd_init(&adev->acp.acp_genpd->gpd, NULL, false);
 	}
 
-	adev->acp.acp_cell = kzalloc(sizeof(struct mfd_cell) * ACP_DEVS,
-							GFP_KERNEL);
+	if (adev->rmmio_size == 0 || adev->rmmio_size < 0x5289)
+		return -EINVAL;
 
-	if (adev->acp.acp_cell == NULL)
+	acp_base = adev->rmmio_base;
+	adev->acp.acp_genpd = kzalloc(sizeof(struct acp_pm_domain), GFP_KERNEL);
+	if (!adev->acp.acp_genpd)
 		return -ENOMEM;
 
-	adev->acp.acp_res = kzalloc(sizeof(struct resource) * 4, GFP_KERNEL);
+	adev->acp.acp_genpd->gpd.name = "ACP_AUDIO";
+	adev->acp.acp_genpd->gpd.power_off = acp_poweroff;
+	adev->acp.acp_genpd->gpd.power_on = acp_poweron;
+	adev->acp.acp_genpd->adev = adev;
+
+	pm_genpd_init(&adev->acp.acp_genpd->gpd, NULL, false);
+	dmi_check_system(acp_quirk_table);
+	switch (acp_machine_id) {
+	case ST_JADEITE:
+	{
+		adev->acp.acp_cell = kcalloc(2, sizeof(struct mfd_cell),
+					     GFP_KERNEL);
+		if (!adev->acp.acp_cell) {
+			r = -ENOMEM;
+			goto failure;
+		}
 
-	if (adev->acp.acp_res == NULL) {
-		kfree(adev->acp.acp_cell);
-		return -ENOMEM;
-	}
+		adev->acp.acp_res = kcalloc(3, sizeof(struct resource), GFP_KERNEL);
+		if (!adev->acp.acp_res) {
+			r = -ENOMEM;
+			goto failure;
+		}
 
-	i2s_pdata = kzalloc(sizeof(struct i2s_platform_data) * 2, GFP_KERNEL);
-	if (i2s_pdata == NULL) {
-		kfree(adev->acp.acp_res);
-		kfree(adev->acp.acp_cell);
-		return -ENOMEM;
-	}
+		i2s_pdata = kcalloc(1, sizeof(struct i2s_platform_data), GFP_KERNEL);
+		if (!i2s_pdata) {
+			r = -ENOMEM;
+			goto failure;
+		}
 
-	switch (adev->asic_type) {
-	case CHIP_STONEY:
 		i2s_pdata[0].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET |
-			DW_I2S_QUIRK_16BIT_IDX_OVERRIDE;
+				      DW_I2S_QUIRK_16BIT_IDX_OVERRIDE;
+		i2s_pdata[0].cap = DWC_I2S_PLAY | DWC_I2S_RECORD;
+		i2s_pdata[0].snd_rates = SNDRV_PCM_RATE_8000_96000;
+		i2s_pdata[0].i2s_reg_comp1 = ACP_I2S_COMP1_CAP_REG_OFFSET;
+		i2s_pdata[0].i2s_reg_comp2 = ACP_I2S_COMP2_CAP_REG_OFFSET;
+
+		adev->acp.acp_res[0].name = "acp2x_dma";
+		adev->acp.acp_res[0].flags = IORESOURCE_MEM;
+		adev->acp.acp_res[0].start = acp_base;
+		adev->acp.acp_res[0].end = acp_base + ACP_DMA_REGS_END;
+
+		adev->acp.acp_res[1].name = "acp2x_dw_i2s_play_cap";
+		adev->acp.acp_res[1].flags = IORESOURCE_MEM;
+		adev->acp.acp_res[1].start = acp_base + ACP_I2S_CAP_REGS_START;
+		adev->acp.acp_res[1].end = acp_base + ACP_I2S_CAP_REGS_END;
+
+		adev->acp.acp_res[2].name = "acp2x_dma_irq";
+		adev->acp.acp_res[2].flags = IORESOURCE_IRQ;
+		adev->acp.acp_res[2].start = amdgpu_irq_create_mapping(adev, 162);
+		adev->acp.acp_res[2].end = adev->acp.acp_res[2].start;
+
+		adev->acp.acp_cell[0].name = "acp_audio_dma";
+		adev->acp.acp_cell[0].num_resources = 3;
+		adev->acp.acp_cell[0].resources = &adev->acp.acp_res[0];
+		adev->acp.acp_cell[0].platform_data = &adev->asic_type;
+		adev->acp.acp_cell[0].pdata_size = sizeof(adev->asic_type);
+
+		adev->acp.acp_cell[1].name = "designware-i2s";
+		adev->acp.acp_cell[1].num_resources = 1;
+		adev->acp.acp_cell[1].resources = &adev->acp.acp_res[1];
+		adev->acp.acp_cell[1].platform_data = &i2s_pdata[0];
+		adev->acp.acp_cell[1].pdata_size = sizeof(struct i2s_platform_data);
+		r = mfd_add_hotplug_devices(adev->acp.parent, adev->acp.acp_cell, 2);
+		if (r)
+			goto failure;
+		r = device_for_each_child(adev->acp.parent, &adev->acp.acp_genpd->gpd,
+					  acp_genpd_add_device);
+		if (r)
+			goto failure;
 		break;
-	default:
-		i2s_pdata[0].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET;
 	}
-	i2s_pdata[0].cap = DWC_I2S_PLAY;
-	i2s_pdata[0].snd_rates = SNDRV_PCM_RATE_8000_96000;
-	i2s_pdata[0].i2s_reg_comp1 = ACP_I2S_COMP1_PLAY_REG_OFFSET;
-	i2s_pdata[0].i2s_reg_comp2 = ACP_I2S_COMP2_PLAY_REG_OFFSET;
-	switch (adev->asic_type) {
-	case CHIP_STONEY:
-		i2s_pdata[1].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET |
-			DW_I2S_QUIRK_COMP_PARAM1 |
-			DW_I2S_QUIRK_16BIT_IDX_OVERRIDE;
-		break;
 	default:
-		i2s_pdata[1].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET |
-			DW_I2S_QUIRK_COMP_PARAM1;
-	}
+		adev->acp.acp_cell = kcalloc(ACP_DEVS, sizeof(struct mfd_cell),
+					     GFP_KERNEL);
 
-	i2s_pdata[1].cap = DWC_I2S_RECORD;
-	i2s_pdata[1].snd_rates = SNDRV_PCM_RATE_8000_96000;
-	i2s_pdata[1].i2s_reg_comp1 = ACP_I2S_COMP1_CAP_REG_OFFSET;
-	i2s_pdata[1].i2s_reg_comp2 = ACP_I2S_COMP2_CAP_REG_OFFSET;
-
-	adev->acp.acp_res[0].name = "acp2x_dma";
-	adev->acp.acp_res[0].flags = IORESOURCE_MEM;
-	adev->acp.acp_res[0].start = acp_base;
-	adev->acp.acp_res[0].end = acp_base + ACP_DMA_REGS_END;
-
-	adev->acp.acp_res[1].name = "acp2x_dw_i2s_play";
-	adev->acp.acp_res[1].flags = IORESOURCE_MEM;
-	adev->acp.acp_res[1].start = acp_base + ACP_I2S_PLAY_REGS_START;
-	adev->acp.acp_res[1].end = acp_base + ACP_I2S_PLAY_REGS_END;
-
-	adev->acp.acp_res[2].name = "acp2x_dw_i2s_cap";
-	adev->acp.acp_res[2].flags = IORESOURCE_MEM;
-	adev->acp.acp_res[2].start = acp_base + ACP_I2S_CAP_REGS_START;
-	adev->acp.acp_res[2].end = acp_base + ACP_I2S_CAP_REGS_END;
-
-	adev->acp.acp_res[3].name = "acp2x_dma_irq";
-	adev->acp.acp_res[3].flags = IORESOURCE_IRQ;
-	adev->acp.acp_res[3].start = amdgpu_irq_create_mapping(adev, 162);
-	adev->acp.acp_res[3].end = adev->acp.acp_res[3].start;
-
-	adev->acp.acp_cell[0].name = "acp_audio_dma";
-	adev->acp.acp_cell[0].num_resources = 4;
-	adev->acp.acp_cell[0].resources = &adev->acp.acp_res[0];
-
-	adev->acp.acp_cell[1].name = "designware-i2s";
-	adev->acp.acp_cell[1].num_resources = 1;
-	adev->acp.acp_cell[1].resources = &adev->acp.acp_res[1];
-	adev->acp.acp_cell[1].platform_data = &i2s_pdata[0];
-	adev->acp.acp_cell[1].pdata_size = sizeof(struct i2s_platform_data);
-
-	adev->acp.acp_cell[2].name = "designware-i2s";
-	adev->acp.acp_cell[2].num_resources = 1;
-	adev->acp.acp_cell[2].resources = &adev->acp.acp_res[2];
-	adev->acp.acp_cell[2].platform_data = &i2s_pdata[1];
-	adev->acp.acp_cell[2].pdata_size = sizeof(struct i2s_platform_data);
-
-	r = mfd_add_hotplug_devices(adev->acp.parent, adev->acp.acp_cell,
-								ACP_DEVS);
-	if (r)
-		return r;
+		if (!adev->acp.acp_cell) {
+			r = -ENOMEM;
+			goto failure;
+		}
+
+		adev->acp.acp_res = kcalloc(5, sizeof(struct resource), GFP_KERNEL);
+		if (!adev->acp.acp_res) {
+			r = -ENOMEM;
+			goto failure;
+		}
+
+		i2s_pdata = kcalloc(3, sizeof(struct i2s_platform_data), GFP_KERNEL);
+		if (!i2s_pdata) {
+			r = -ENOMEM;
+			goto failure;
+		}
 
-	if (adev->asic_type != CHIP_STONEY) {
-		for (i = 0; i < ACP_DEVS ; i++) {
-			dev = get_mfd_cell_dev(adev->acp.acp_cell[i].name, i);
-			r = pm_genpd_add_device(&adev->acp.acp_genpd->gpd, dev);
-			if (r) {
-				dev_err(dev, "Failed to add dev to genpd\n");
-				return r;
-			}
+		switch (adev->asic_type) {
+		case CHIP_STONEY:
+			i2s_pdata[0].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET |
+				DW_I2S_QUIRK_16BIT_IDX_OVERRIDE;
+			break;
+		default:
+			i2s_pdata[0].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET;
 		}
+		i2s_pdata[0].cap = DWC_I2S_PLAY;
+		i2s_pdata[0].snd_rates = SNDRV_PCM_RATE_8000_96000;
+		i2s_pdata[0].i2s_reg_comp1 = ACP_I2S_COMP1_PLAY_REG_OFFSET;
+		i2s_pdata[0].i2s_reg_comp2 = ACP_I2S_COMP2_PLAY_REG_OFFSET;
+		switch (adev->asic_type) {
+		case CHIP_STONEY:
+			i2s_pdata[1].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET |
+				DW_I2S_QUIRK_COMP_PARAM1 |
+				DW_I2S_QUIRK_16BIT_IDX_OVERRIDE;
+			break;
+		default:
+			i2s_pdata[1].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET |
+				DW_I2S_QUIRK_COMP_PARAM1;
+		}
+
+		i2s_pdata[1].cap = DWC_I2S_RECORD;
+		i2s_pdata[1].snd_rates = SNDRV_PCM_RATE_8000_96000;
+		i2s_pdata[1].i2s_reg_comp1 = ACP_I2S_COMP1_CAP_REG_OFFSET;
+		i2s_pdata[1].i2s_reg_comp2 = ACP_I2S_COMP2_CAP_REG_OFFSET;
+
+		i2s_pdata[2].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET;
+		switch (adev->asic_type) {
+		case CHIP_STONEY:
+			i2s_pdata[2].quirks |= DW_I2S_QUIRK_16BIT_IDX_OVERRIDE;
+			break;
+		default:
+			break;
+		}
+
+		i2s_pdata[2].cap = DWC_I2S_PLAY | DWC_I2S_RECORD;
+		i2s_pdata[2].snd_rates = SNDRV_PCM_RATE_8000_96000;
+		i2s_pdata[2].i2s_reg_comp1 = ACP_BT_COMP1_REG_OFFSET;
+		i2s_pdata[2].i2s_reg_comp2 = ACP_BT_COMP2_REG_OFFSET;
+
+		adev->acp.acp_res[0].name = "acp2x_dma";
+		adev->acp.acp_res[0].flags = IORESOURCE_MEM;
+		adev->acp.acp_res[0].start = acp_base;
+		adev->acp.acp_res[0].end = acp_base + ACP_DMA_REGS_END;
+
+		adev->acp.acp_res[1].name = "acp2x_dw_i2s_play";
+		adev->acp.acp_res[1].flags = IORESOURCE_MEM;
+		adev->acp.acp_res[1].start = acp_base + ACP_I2S_PLAY_REGS_START;
+		adev->acp.acp_res[1].end = acp_base + ACP_I2S_PLAY_REGS_END;
+
+		adev->acp.acp_res[2].name = "acp2x_dw_i2s_cap";
+		adev->acp.acp_res[2].flags = IORESOURCE_MEM;
+		adev->acp.acp_res[2].start = acp_base + ACP_I2S_CAP_REGS_START;
+		adev->acp.acp_res[2].end = acp_base + ACP_I2S_CAP_REGS_END;
+
+		adev->acp.acp_res[3].name = "acp2x_dw_bt_i2s_play_cap";
+		adev->acp.acp_res[3].flags = IORESOURCE_MEM;
+		adev->acp.acp_res[3].start = acp_base + ACP_BT_PLAY_REGS_START;
+		adev->acp.acp_res[3].end = acp_base + ACP_BT_PLAY_REGS_END;
+
+		adev->acp.acp_res[4].name = "acp2x_dma_irq";
+		adev->acp.acp_res[4].flags = IORESOURCE_IRQ;
+		adev->acp.acp_res[4].start = amdgpu_irq_create_mapping(adev, 162);
+		adev->acp.acp_res[4].end = adev->acp.acp_res[4].start;
+
+		adev->acp.acp_cell[0].name = "acp_audio_dma";
+		adev->acp.acp_cell[0].num_resources = 5;
+		adev->acp.acp_cell[0].resources = &adev->acp.acp_res[0];
+		adev->acp.acp_cell[0].platform_data = &adev->asic_type;
+		adev->acp.acp_cell[0].pdata_size = sizeof(adev->asic_type);
+
+		adev->acp.acp_cell[1].name = "designware-i2s";
+		adev->acp.acp_cell[1].num_resources = 1;
+		adev->acp.acp_cell[1].resources = &adev->acp.acp_res[1];
+		adev->acp.acp_cell[1].platform_data = &i2s_pdata[0];
+		adev->acp.acp_cell[1].pdata_size = sizeof(struct i2s_platform_data);
+
+		adev->acp.acp_cell[2].name = "designware-i2s";
+		adev->acp.acp_cell[2].num_resources = 1;
+		adev->acp.acp_cell[2].resources = &adev->acp.acp_res[2];
+		adev->acp.acp_cell[2].platform_data = &i2s_pdata[1];
+		adev->acp.acp_cell[2].pdata_size = sizeof(struct i2s_platform_data);
+
+		adev->acp.acp_cell[3].name = "designware-i2s";
+		adev->acp.acp_cell[3].num_resources = 1;
+		adev->acp.acp_cell[3].resources = &adev->acp.acp_res[3];
+		adev->acp.acp_cell[3].platform_data = &i2s_pdata[2];
+		adev->acp.acp_cell[3].pdata_size = sizeof(struct i2s_platform_data);
+
+		r = mfd_add_hotplug_devices(adev->acp.parent, adev->acp.acp_cell, ACP_DEVS);
+		if (r)
+			goto failure;
+
+		r = device_for_each_child(adev->acp.parent, &adev->acp.acp_genpd->gpd,
+					  acp_genpd_add_device);
+		if (r)
+			goto failure;
 	}
 
+	/* Assert Soft reset of ACP */
+	val = cgs_read_register(adev->acp.cgs_device, mmACP_SOFT_RESET);
+
+	val |= ACP_SOFT_RESET__SoftResetAud_MASK;
+	cgs_write_register(adev->acp.cgs_device, mmACP_SOFT_RESET, val);
+
+	count = ACP_SOFT_RESET_DONE_TIME_OUT_VALUE;
+	while (true) {
+		val = cgs_read_register(adev->acp.cgs_device, mmACP_SOFT_RESET);
+		if (ACP_SOFT_RESET__SoftResetAudDone_MASK ==
+		    (val & ACP_SOFT_RESET__SoftResetAudDone_MASK))
+			break;
+		if (--count == 0) {
+			dev_err(&adev->pdev->dev, "Failed to reset ACP\n");
+			r = -ETIMEDOUT;
+			goto failure;
+		}
+		udelay(100);
+	}
+	/* Enable clock to ACP and wait until the clock is enabled */
+	val = cgs_read_register(adev->acp.cgs_device, mmACP_CONTROL);
+	val = val | ACP_CONTROL__ClkEn_MASK;
+	cgs_write_register(adev->acp.cgs_device, mmACP_CONTROL, val);
+
+	count = ACP_CLOCK_EN_TIME_OUT_VALUE;
+
+	while (true) {
+		val = cgs_read_register(adev->acp.cgs_device, mmACP_STATUS);
+		if (val & (u32) 0x1)
+			break;
+		if (--count == 0) {
+			dev_err(&adev->pdev->dev, "Failed to reset ACP\n");
+			r = -ETIMEDOUT;
+			goto failure;
+		}
+		udelay(100);
+	}
+	/* Deassert the SOFT RESET flags */
+	val = cgs_read_register(adev->acp.cgs_device, mmACP_SOFT_RESET);
+	val &= ~ACP_SOFT_RESET__SoftResetAud_MASK;
+	cgs_write_register(adev->acp.cgs_device, mmACP_SOFT_RESET, val);
 	return 0;
+
+failure:
+	kfree(i2s_pdata);
+	kfree(adev->acp.acp_res);
+	kfree(adev->acp.acp_cell);
+	kfree(adev->acp.acp_genpd);
+	return r;
 }
 
 /**
  * acp_hw_fini - stop the hardware block
  *
- * @adev: amdgpu_device pointer
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
  *
  */
-static int acp_hw_fini(void *handle)
+static int acp_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	int i, ret;
-	struct device *dev;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 val = 0;
+	u32 count = 0;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	/* return early if no ACP */
-	if (!adev->acp.acp_cell)
+	if (!adev->acp.acp_genpd) {
+		amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_ACP, false, 0);
 		return 0;
+	}
 
-	if (adev->acp.acp_genpd) {
-		for (i = 0; i < ACP_DEVS ; i++) {
-			dev = get_mfd_cell_dev(adev->acp.acp_cell[i].name, i);
-			ret = pm_genpd_remove_device(&adev->acp.acp_genpd->gpd, dev);
-			/* If removal fails, dont giveup and try rest */
-			if (ret)
-				dev_err(dev, "remove dev from genpd failed\n");
+	/* Assert Soft reset of ACP */
+	val = cgs_read_register(adev->acp.cgs_device, mmACP_SOFT_RESET);
+
+	val |= ACP_SOFT_RESET__SoftResetAud_MASK;
+	cgs_write_register(adev->acp.cgs_device, mmACP_SOFT_RESET, val);
+
+	count = ACP_SOFT_RESET_DONE_TIME_OUT_VALUE;
+	while (true) {
+		val = cgs_read_register(adev->acp.cgs_device, mmACP_SOFT_RESET);
+		if (ACP_SOFT_RESET__SoftResetAudDone_MASK ==
+		    (val & ACP_SOFT_RESET__SoftResetAudDone_MASK))
+			break;
+		if (--count == 0) {
+			dev_err(&adev->pdev->dev, "Failed to reset ACP\n");
+			return -ETIMEDOUT;
+		}
+		udelay(100);
+	}
+	/* Disable ACP clock */
+	val = cgs_read_register(adev->acp.cgs_device, mmACP_CONTROL);
+	val &= ~ACP_CONTROL__ClkEn_MASK;
+	cgs_write_register(adev->acp.cgs_device, mmACP_CONTROL, val);
+
+	count = ACP_CLOCK_EN_TIME_OUT_VALUE;
+
+	while (true) {
+		val = cgs_read_register(adev->acp.cgs_device, mmACP_STATUS);
+		if (val & (u32) 0x1)
+			break;
+		if (--count == 0) {
+			dev_err(&adev->pdev->dev, "Failed to reset ACP\n");
+			return -ETIMEDOUT;
 		}
-		kfree(adev->acp.acp_genpd);
+		udelay(100);
 	}
 
+	device_for_each_child(adev->acp.parent, NULL,
+			      acp_genpd_remove_device);
+
 	mfd_remove_devices(adev->acp.parent);
 	kfree(adev->acp.acp_res);
+	kfree(adev->acp.acp_genpd);
 	kfree(adev->acp.acp_cell);
 
 	return 0;
 }
 
-static int acp_suspend(void *handle)
-{
-	return 0;
-}
-
-static int acp_resume(void *handle)
+static int acp_suspend(struct amdgpu_ip_block *ip_block)
 {
-	return 0;
-}
+	struct amdgpu_device *adev = ip_block->adev;
 
-static int acp_early_init(void *handle)
-{
+	/* power up on suspend */
+	if (!adev->acp.acp_cell)
+		amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_ACP, false, 0);
 	return 0;
 }
 
-static bool acp_is_idle(void *handle)
+static int acp_resume(struct amdgpu_ip_block *ip_block)
 {
-	return true;
-}
+	struct amdgpu_device *adev = ip_block->adev;
 
-static int acp_wait_for_idle(void *handle)
-{
+	/* power down again on resume */
+	if (!adev->acp.acp_cell)
+		amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_ACP, true, 0);
 	return 0;
 }
 
-static int acp_soft_reset(void *handle)
+static bool acp_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	return 0;
+	return true;
 }
 
-static int acp_set_clockgating_state(void *handle,
+static int acp_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 				     enum amd_clockgating_state state)
 {
 	return 0;
 }
 
-static int acp_set_powergating_state(void *handle,
+static int acp_set_powergating_state(struct amdgpu_ip_block *ip_block,
 				     enum amd_powergating_state state)
 {
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = (state == AMD_PG_STATE_GATE);
+
+	amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_ACP, enable, 0);
+
 	return 0;
 }
 
 static const struct amd_ip_funcs acp_ip_funcs = {
 	.name = "acp_ip",
-	.early_init = acp_early_init,
-	.late_init = NULL,
 	.sw_init = acp_sw_init,
 	.sw_fini = acp_sw_fini,
 	.hw_init = acp_hw_init,
@@ -490,14 +610,11 @@ static const struct amd_ip_funcs acp_ip_funcs = {
 	.suspend = acp_suspend,
 	.resume = acp_resume,
 	.is_idle = acp_is_idle,
-	.wait_for_idle = acp_wait_for_idle,
-	.soft_reset = acp_soft_reset,
 	.set_clockgating_state = acp_set_clockgating_state,
 	.set_powergating_state = acp_set_powergating_state,
 };
 
-const struct amdgpu_ip_block_version acp_ip_block =
-{
+const struct amdgpu_ip_block_version acp_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_ACP,
 	.major = 2,
 	.minor = 2,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
index 57afad79f55d..6c62e27b9800 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
@@ -1,3 +1,4 @@
+// SPDX-License-Identifier: MIT
 /*
  * Copyright 2012 Advanced Micro Devices, Inc.
  *
@@ -23,33 +24,130 @@
 
 #include <linux/pci.h>
 #include <linux/acpi.h>
+#include <linux/backlight.h>
 #include <linux/slab.h>
+#include <linux/xarray.h>
 #include <linux/power_supply.h>
 #include <linux/pm_runtime.h>
+#include <linux/suspend.h>
 #include <acpi/video.h>
-#include <drm/drmP.h>
-#include <drm/drm_crtc_helper.h>
+#include <acpi/actbl.h>
+
 #include "amdgpu.h"
 #include "amdgpu_pm.h"
+#include "amdgpu_display.h"
 #include "amd_acpi.h"
 #include "atom.h"
 
+/* Declare GUID for AMD _DSM method for XCCs */
+static const guid_t amd_xcc_dsm_guid = GUID_INIT(0x8267f5d5, 0xa556, 0x44f2,
+						 0xb8, 0xb4, 0x45, 0x56, 0x2e,
+						 0x8c, 0x5b, 0xec);
+
+#define AMD_XCC_HID_START 3000
+#define AMD_XCC_DSM_GET_NUM_FUNCS 0
+#define AMD_XCC_DSM_GET_SUPP_MODE 1
+#define AMD_XCC_DSM_GET_XCP_MODE 2
+#define AMD_XCC_DSM_GET_VF_XCC_MAPPING 4
+#define AMD_XCC_DSM_GET_TMR_INFO 5
+#define AMD_XCC_DSM_NUM_FUNCS 5
+
+#define AMD_XCC_MAX_HID 24
+
+struct xarray numa_info_xa;
+
+/* Encapsulates the XCD acpi object information */
+struct amdgpu_acpi_xcc_info {
+	struct list_head list;
+	struct amdgpu_numa_info *numa_info;
+	uint8_t xcp_node;
+	uint8_t phy_id;
+	acpi_handle handle;
+};
+
+struct amdgpu_acpi_dev_info {
+	struct list_head list;
+	struct list_head xcc_list;
+	uint32_t sbdf;
+	uint16_t supp_xcp_mode;
+	uint16_t xcp_mode;
+	uint16_t mem_mode;
+	uint64_t tmr_base;
+	uint64_t tmr_size;
+};
+
+struct list_head amdgpu_acpi_dev_list;
+
+struct amdgpu_atif_notification_cfg {
+	bool enabled;
+	int command_code;
+};
+
+struct amdgpu_atif_notifications {
+	bool thermal_state;
+	bool forced_power_state;
+	bool system_power_state;
+	bool brightness_change;
+	bool dgpu_display_event;
+	bool gpu_package_power_limit;
+};
+
+struct amdgpu_atif_functions {
+	bool system_params;
+	bool sbios_requests;
+	bool temperature_change;
+	bool query_backlight_transfer_characteristics;
+	bool ready_to_undock;
+	bool external_gpu_information;
+};
+
+struct amdgpu_atif {
+	acpi_handle handle;
+
+	struct amdgpu_atif_notifications notifications;
+	struct amdgpu_atif_functions functions;
+	struct amdgpu_atif_notification_cfg notification_cfg;
+	struct backlight_device *bd;
+	struct amdgpu_dm_backlight_caps backlight_caps;
+};
+
+struct amdgpu_atcs_functions {
+	bool get_ext_state;
+	bool pcie_perf_req;
+	bool pcie_dev_rdy;
+	bool pcie_bus_width;
+	bool power_shift_control;
+};
+
+struct amdgpu_atcs {
+	acpi_handle handle;
+
+	struct amdgpu_atcs_functions functions;
+};
+
+static struct amdgpu_acpi_priv {
+	struct amdgpu_atif atif;
+	struct amdgpu_atcs atcs;
+} amdgpu_acpi_priv;
+
 /* Call the ATIF method
  */
 /**
  * amdgpu_atif_call - call an ATIF method
  *
- * @handle: acpi handle
+ * @atif: atif structure
  * @function: the ATIF function to execute
  * @params: ATIF function params
  *
  * Executes the requested ATIF function (all asics).
  * Returns a pointer to the acpi output buffer.
  */
-static union acpi_object *amdgpu_atif_call(acpi_handle handle, int function,
-		struct acpi_buffer *params)
+static union acpi_object *amdgpu_atif_call(struct amdgpu_atif *atif,
+					   int function,
+					   struct acpi_buffer *params)
 {
 	acpi_status status;
+	union acpi_object *obj;
 	union acpi_object atif_arg_elements[2];
 	struct acpi_object_list atif_arg;
 	struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL };
@@ -70,17 +168,26 @@ static union acpi_object *amdgpu_atif_call(acpi_handle handle, int function,
 		atif_arg_elements[1].integer.value = 0;
 	}
 
-	status = acpi_evaluate_object(handle, "ATIF", &atif_arg, &buffer);
+	status = acpi_evaluate_object(atif->handle, NULL, &atif_arg,
+				      &buffer);
+	obj = (union acpi_object *)buffer.pointer;
 
-	/* Fail only if calling the method fails and ATIF is supported */
-	if (ACPI_FAILURE(status) && status != AE_NOT_FOUND) {
+	/* Fail if calling the method fails */
+	if (ACPI_FAILURE(status)) {
 		DRM_DEBUG_DRIVER("failed to evaluate ATIF got %s\n",
 				 acpi_format_exception(status));
-		kfree(buffer.pointer);
+		kfree(obj);
 		return NULL;
 	}
 
-	return buffer.pointer;
+	if (obj->type != ACPI_TYPE_BUFFER) {
+		DRM_DEBUG_DRIVER("bad object returned from ATIF: %d\n",
+				 obj->type);
+		kfree(obj);
+		return NULL;
+	}
+
+	return obj;
 }
 
 /**
@@ -95,15 +202,12 @@ static union acpi_object *amdgpu_atif_call(acpi_handle handle, int function,
  */
 static void amdgpu_atif_parse_notification(struct amdgpu_atif_notifications *n, u32 mask)
 {
-	n->display_switch = mask & ATIF_DISPLAY_SWITCH_REQUEST_SUPPORTED;
-	n->expansion_mode_change = mask & ATIF_EXPANSION_MODE_CHANGE_REQUEST_SUPPORTED;
 	n->thermal_state = mask & ATIF_THERMAL_STATE_CHANGE_REQUEST_SUPPORTED;
 	n->forced_power_state = mask & ATIF_FORCED_POWER_STATE_CHANGE_REQUEST_SUPPORTED;
 	n->system_power_state = mask & ATIF_SYSTEM_POWER_SOURCE_CHANGE_REQUEST_SUPPORTED;
-	n->display_conf_change = mask & ATIF_DISPLAY_CONF_CHANGE_REQUEST_SUPPORTED;
-	n->px_gfx_switch = mask & ATIF_PX_GFX_SWITCH_REQUEST_SUPPORTED;
 	n->brightness_change = mask & ATIF_PANEL_BRIGHTNESS_CHANGE_REQUEST_SUPPORTED;
 	n->dgpu_display_event = mask & ATIF_DGPU_DISPLAY_EVENT_SUPPORTED;
+	n->gpu_package_power_limit = mask & ATIF_GPU_PACKAGE_POWER_LIMIT_REQUEST_SUPPORTED;
 }
 
 /**
@@ -120,20 +224,16 @@ static void amdgpu_atif_parse_functions(struct amdgpu_atif_functions *f, u32 mas
 {
 	f->system_params = mask & ATIF_GET_SYSTEM_PARAMETERS_SUPPORTED;
 	f->sbios_requests = mask & ATIF_GET_SYSTEM_BIOS_REQUESTS_SUPPORTED;
-	f->select_active_disp = mask & ATIF_SELECT_ACTIVE_DISPLAYS_SUPPORTED;
-	f->lid_state = mask & ATIF_GET_LID_STATE_SUPPORTED;
-	f->get_tv_standard = mask & ATIF_GET_TV_STANDARD_FROM_CMOS_SUPPORTED;
-	f->set_tv_standard = mask & ATIF_SET_TV_STANDARD_IN_CMOS_SUPPORTED;
-	f->get_panel_expansion_mode = mask & ATIF_GET_PANEL_EXPANSION_MODE_FROM_CMOS_SUPPORTED;
-	f->set_panel_expansion_mode = mask & ATIF_SET_PANEL_EXPANSION_MODE_IN_CMOS_SUPPORTED;
 	f->temperature_change = mask & ATIF_TEMPERATURE_CHANGE_NOTIFICATION_SUPPORTED;
-	f->graphics_device_types = mask & ATIF_GET_GRAPHICS_DEVICE_TYPES_SUPPORTED;
+	f->query_backlight_transfer_characteristics =
+		mask & ATIF_QUERY_BACKLIGHT_TRANSFER_CHARACTERISTICS_SUPPORTED;
+	f->ready_to_undock = mask & ATIF_READY_TO_UNDOCK_NOTIFICATION_SUPPORTED;
+	f->external_gpu_information = mask & ATIF_GET_EXTERNAL_GPU_INFORMATION_SUPPORTED;
 }
 
 /**
  * amdgpu_atif_verify_interface - verify ATIF
  *
- * @handle: acpi handle
  * @atif: amdgpu atif struct
  *
  * Execute the ATIF_FUNCTION_VERIFY_INTERFACE ATIF function
@@ -141,15 +241,14 @@ static void amdgpu_atif_parse_functions(struct amdgpu_atif_functions *f, u32 mas
  * (all asics).
  * returns 0 on success, error on failure.
  */
-static int amdgpu_atif_verify_interface(acpi_handle handle,
-		struct amdgpu_atif *atif)
+static int amdgpu_atif_verify_interface(struct amdgpu_atif *atif)
 {
 	union acpi_object *info;
 	struct atif_verify_interface output;
 	size_t size;
 	int err = 0;
 
-	info = amdgpu_atif_call(handle, ATIF_FUNCTION_VERIFY_INTERFACE, NULL);
+	info = amdgpu_atif_call(atif, ATIF_FUNCTION_VERIFY_INTERFACE, NULL);
 	if (!info)
 		return -EIO;
 
@@ -179,8 +278,7 @@ out:
 /**
  * amdgpu_atif_get_notification_params - determine notify configuration
  *
- * @handle: acpi handle
- * @n: atif notification configuration struct
+ * @atif: acpi handle
  *
  * Execute the ATIF_FUNCTION_GET_SYSTEM_PARAMETERS ATIF function
  * to determine if a notifier is used and if so which one
@@ -188,15 +286,16 @@ out:
  * where n is specified in the result if a notifier is used.
  * Returns 0 on success, error on failure.
  */
-static int amdgpu_atif_get_notification_params(acpi_handle handle,
-		struct amdgpu_atif_notification_cfg *n)
+static int amdgpu_atif_get_notification_params(struct amdgpu_atif *atif)
 {
 	union acpi_object *info;
+	struct amdgpu_atif_notification_cfg *n = &atif->notification_cfg;
 	struct atif_system_params params;
 	size_t size;
 	int err = 0;
 
-	info = amdgpu_atif_call(handle, ATIF_FUNCTION_GET_SYSTEM_PARAMETERS, NULL);
+	info = amdgpu_atif_call(atif, ATIF_FUNCTION_GET_SYSTEM_PARAMETERS,
+				NULL);
 	if (!info) {
 		err = -EIO;
 		goto out;
@@ -240,9 +339,74 @@ out:
 }
 
 /**
+ * amdgpu_atif_query_backlight_caps - get min and max backlight input signal
+ *
+ * @atif: acpi handle
+ *
+ * Execute the QUERY_BRIGHTNESS_TRANSFER_CHARACTERISTICS ATIF function
+ * to determine the acceptable range of backlight values
+ *
+ * Backlight_caps.caps_valid will be set to true if the query is successful
+ *
+ * The input signals are in range 0-255
+ *
+ * This function assumes the display with backlight is the first LCD
+ *
+ * Returns 0 on success, error on failure.
+ */
+static int amdgpu_atif_query_backlight_caps(struct amdgpu_atif *atif)
+{
+	union acpi_object *info;
+	struct atif_qbtc_output characteristics;
+	struct atif_qbtc_arguments arguments;
+	struct acpi_buffer params;
+	size_t size;
+	int err = 0;
+
+	arguments.size = sizeof(arguments);
+	arguments.requested_display = ATIF_QBTC_REQUEST_LCD1;
+
+	params.length = sizeof(arguments);
+	params.pointer = (void *)&arguments;
+
+	info = amdgpu_atif_call(atif,
+		ATIF_FUNCTION_QUERY_BRIGHTNESS_TRANSFER_CHARACTERISTICS,
+		&params);
+	if (!info) {
+		err = -EIO;
+		goto out;
+	}
+
+	size = *(u16 *) info->buffer.pointer;
+	if (size < 10) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	memset(&characteristics, 0, sizeof(characteristics));
+	size = min(sizeof(characteristics), size);
+	memcpy(&characteristics, info->buffer.pointer, size);
+
+	atif->backlight_caps.caps_valid = true;
+	atif->backlight_caps.min_input_signal =
+			characteristics.min_input_signal;
+	atif->backlight_caps.max_input_signal =
+			characteristics.max_input_signal;
+	atif->backlight_caps.ac_level = characteristics.ac_level;
+	atif->backlight_caps.dc_level = characteristics.dc_level;
+	atif->backlight_caps.data_points = characteristics.number_of_points;
+	memcpy(atif->backlight_caps.luminance_data,
+	       characteristics.data_points,
+	       sizeof(atif->backlight_caps.luminance_data));
+out:
+	kfree(info);
+	return err;
+}
+
+/**
  * amdgpu_atif_get_sbios_requests - get requested sbios event
  *
- * @handle: acpi handle
+ * @atif: acpi handle
  * @req: atif sbios request struct
  *
  * Execute the ATIF_FUNCTION_GET_SYSTEM_BIOS_REQUESTS ATIF function
@@ -250,14 +414,15 @@ out:
  * (all asics).
  * Returns 0 on success, error on failure.
  */
-static int amdgpu_atif_get_sbios_requests(acpi_handle handle,
-		struct atif_sbios_requests *req)
+static int amdgpu_atif_get_sbios_requests(struct amdgpu_atif *atif,
+					  struct atif_sbios_requests *req)
 {
 	union acpi_object *info;
 	size_t size;
 	int count = 0;
 
-	info = amdgpu_atif_call(handle, ATIF_FUNCTION_GET_SYSTEM_BIOS_REQUESTS, NULL);
+	info = amdgpu_atif_call(atif, ATIF_FUNCTION_GET_SYSTEM_BIOS_REQUESTS,
+				NULL);
 	if (!info)
 		return -EIO;
 
@@ -287,14 +452,14 @@ out:
  *
  * Checks the acpi event and if it matches an atif event,
  * handles it.
- * Returns NOTIFY code
+ *
+ * Returns:
+ * NOTIFY_BAD or NOTIFY_DONE, depending on the event.
  */
 static int amdgpu_atif_handler(struct amdgpu_device *adev,
-			struct acpi_bus_event *event)
+			       struct acpi_bus_event *event)
 {
-	struct amdgpu_atif *atif = &adev->atif;
-	struct atif_sbios_requests req;
-	acpi_handle handle;
+	struct amdgpu_atif *atif = &amdgpu_acpi_priv.atif;
 	int count;
 
 	DRM_DEBUG_DRIVER("event, device_class = %s, type = %#x\n",
@@ -303,48 +468,51 @@ static int amdgpu_atif_handler(struct amdgpu_device *adev,
 	if (strcmp(event->device_class, ACPI_VIDEO_CLASS) != 0)
 		return NOTIFY_DONE;
 
+	/* Is this actually our event? */
 	if (!atif->notification_cfg.enabled ||
-	    event->type != atif->notification_cfg.command_code)
-		/* Not our event */
-		return NOTIFY_DONE;
-
-	/* Check pending SBIOS requests */
-	handle = ACPI_HANDLE(&adev->pdev->dev);
-	count = amdgpu_atif_get_sbios_requests(handle, &req);
-
-	if (count <= 0)
-		return NOTIFY_DONE;
-
-	DRM_DEBUG_DRIVER("ATIF: %d pending SBIOS requests\n", count);
+	    event->type != atif->notification_cfg.command_code) {
+		/* These events will generate keypresses otherwise */
+		if (event->type == ACPI_VIDEO_NOTIFY_PROBE)
+			return NOTIFY_BAD;
+		else
+			return NOTIFY_DONE;
+	}
 
-	if (req.pending & ATIF_PANEL_BRIGHTNESS_CHANGE_REQUEST) {
-		struct amdgpu_encoder *enc = atif->encoder_for_bl;
+	if (atif->functions.sbios_requests) {
+		struct atif_sbios_requests req;
 
-		if (enc) {
-			struct amdgpu_encoder_atom_dig *dig = enc->enc_priv;
+		/* Check pending SBIOS requests */
+		count = amdgpu_atif_get_sbios_requests(atif, &req);
 
-			DRM_DEBUG_DRIVER("Changing brightness to %d\n",
-					req.backlight_level);
+		if (count <= 0)
+			return NOTIFY_BAD;
 
-			amdgpu_display_backlight_set_level(adev, enc, req.backlight_level);
+		DRM_DEBUG_DRIVER("ATIF: %d pending SBIOS requests\n", count);
 
-#if defined(CONFIG_BACKLIGHT_CLASS_DEVICE) || defined(CONFIG_BACKLIGHT_CLASS_DEVICE_MODULE)
-			backlight_force_update(dig->bl_dev,
-					       BACKLIGHT_UPDATE_HOTKEY);
-#endif
+		if (req.pending & ATIF_PANEL_BRIGHTNESS_CHANGE_REQUEST) {
+			if (atif->bd) {
+				DRM_DEBUG_DRIVER("Changing brightness to %d\n",
+						 req.backlight_level);
+				/*
+				 * XXX backlight_device_set_brightness() is
+				 * hardwired to post BACKLIGHT_UPDATE_SYSFS.
+				 * It probably should accept 'reason' parameter.
+				 */
+				backlight_device_set_brightness(atif->bd, req.backlight_level);
+			}
 		}
-	}
-	if (req.pending & ATIF_DGPU_DISPLAY_EVENT) {
-		if ((adev->flags & AMD_IS_PX) &&
-		    amdgpu_atpx_dgpu_req_power_for_displays()) {
-			pm_runtime_get_sync(adev->ddev->dev);
-			/* Just fire off a uevent and let userspace tell us what to do */
-			drm_helper_hpd_irq_event(adev->ddev);
-			pm_runtime_mark_last_busy(adev->ddev->dev);
-			pm_runtime_put_autosuspend(adev->ddev->dev);
+
+		if (req.pending & ATIF_DGPU_DISPLAY_EVENT) {
+			if (adev->flags & AMD_IS_PX) {
+				pm_runtime_get_sync(adev_to_drm(adev)->dev);
+				/* Just fire off a uevent and let userspace tell us what to do */
+				drm_helper_hpd_irq_event(adev_to_drm(adev));
+				pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+				pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+			}
 		}
+		/* TODO: check other events */
 	}
-	/* TODO: check other events */
 
 	/* We've handled the event, stop the notifier chain. The ACPI interface
 	 * overloads ACPI_VIDEO_NOTIFY_PROBE, we don't want to send that to
@@ -359,14 +527,15 @@ static int amdgpu_atif_handler(struct amdgpu_device *adev,
 /**
  * amdgpu_atcs_call - call an ATCS method
  *
- * @handle: acpi handle
+ * @atcs: atcs structure
  * @function: the ATCS function to execute
  * @params: ATCS function params
  *
  * Executes the requested ATCS function (all asics).
  * Returns a pointer to the acpi output buffer.
  */
-static union acpi_object *amdgpu_atcs_call(acpi_handle handle, int function,
+static union acpi_object *amdgpu_atcs_call(struct amdgpu_atcs *atcs,
+					   int function,
 					   struct acpi_buffer *params)
 {
 	acpi_status status;
@@ -390,7 +559,7 @@ static union acpi_object *amdgpu_atcs_call(acpi_handle handle, int function,
 		atcs_arg_elements[1].integer.value = 0;
 	}
 
-	status = acpi_evaluate_object(handle, "ATCS", &atcs_arg, &buffer);
+	status = acpi_evaluate_object(atcs->handle, NULL, &atcs_arg, &buffer);
 
 	/* Fail only if calling the method fails and ATIF is supported */
 	if (ACPI_FAILURE(status) && status != AE_NOT_FOUND) {
@@ -419,12 +588,12 @@ static void amdgpu_atcs_parse_functions(struct amdgpu_atcs_functions *f, u32 mas
 	f->pcie_perf_req = mask & ATCS_PCIE_PERFORMANCE_REQUEST_SUPPORTED;
 	f->pcie_dev_rdy = mask & ATCS_PCIE_DEVICE_READY_NOTIFICATION_SUPPORTED;
 	f->pcie_bus_width = mask & ATCS_SET_PCIE_BUS_WIDTH_SUPPORTED;
+	f->power_shift_control = mask & ATCS_SET_POWER_SHIFT_CONTROL_SUPPORTED;
 }
 
 /**
  * amdgpu_atcs_verify_interface - verify ATCS
  *
- * @handle: acpi handle
  * @atcs: amdgpu atcs struct
  *
  * Execute the ATCS_FUNCTION_VERIFY_INTERFACE ATCS function
@@ -432,15 +601,14 @@ static void amdgpu_atcs_parse_functions(struct amdgpu_atcs_functions *f, u32 mas
  * (all asics).
  * returns 0 on success, error on failure.
  */
-static int amdgpu_atcs_verify_interface(acpi_handle handle,
-					struct amdgpu_atcs *atcs)
+static int amdgpu_atcs_verify_interface(struct amdgpu_atcs *atcs)
 {
 	union acpi_object *info;
 	struct atcs_verify_interface output;
 	size_t size;
 	int err = 0;
 
-	info = amdgpu_atcs_call(handle, ATCS_FUNCTION_VERIFY_INTERFACE, NULL);
+	info = amdgpu_atcs_call(atcs, ATCS_FUNCTION_VERIFY_INTERFACE, NULL);
 	if (!info)
 		return -EIO;
 
@@ -477,7 +645,7 @@ out:
  */
 bool amdgpu_acpi_is_pcie_performance_request_supported(struct amdgpu_device *adev)
 {
-	struct amdgpu_atcs *atcs = &adev->atcs;
+	struct amdgpu_atcs *atcs = &amdgpu_acpi_priv.atcs;
 
 	if (atcs->functions.pcie_perf_req && atcs->functions.pcie_dev_rdy)
 		return true;
@@ -486,6 +654,18 @@ bool amdgpu_acpi_is_pcie_performance_request_supported(struct amdgpu_device *ade
 }
 
 /**
+ * amdgpu_acpi_is_power_shift_control_supported
+ *
+ * Check if the ATCS power shift control method
+ * is supported.
+ * returns true if supported, false if not.
+ */
+bool amdgpu_acpi_is_power_shift_control_supported(void)
+{
+	return amdgpu_acpi_priv.atcs.functions.power_shift_control;
+}
+
+/**
  * amdgpu_acpi_pcie_notify_device_ready
  *
  * @adev: amdgpu_device pointer
@@ -496,19 +676,13 @@ bool amdgpu_acpi_is_pcie_performance_request_supported(struct amdgpu_device *ade
  */
 int amdgpu_acpi_pcie_notify_device_ready(struct amdgpu_device *adev)
 {
-	acpi_handle handle;
 	union acpi_object *info;
-	struct amdgpu_atcs *atcs = &adev->atcs;
-
-	/* Get the device handle */
-	handle = ACPI_HANDLE(&adev->pdev->dev);
-	if (!handle)
-		return -EINVAL;
+	struct amdgpu_atcs *atcs = &amdgpu_acpi_priv.atcs;
 
 	if (!atcs->functions.pcie_dev_rdy)
 		return -EINVAL;
 
-	info = amdgpu_atcs_call(handle, ATCS_FUNCTION_PCIE_DEVICE_READY_NOTIFICATION, NULL);
+	info = amdgpu_atcs_call(atcs, ATCS_FUNCTION_PCIE_DEVICE_READY_NOTIFICATION, NULL);
 	if (!info)
 		return -EIO;
 
@@ -531,18 +705,15 @@ int amdgpu_acpi_pcie_notify_device_ready(struct amdgpu_device *adev)
 int amdgpu_acpi_pcie_performance_request(struct amdgpu_device *adev,
 					 u8 perf_req, bool advertise)
 {
-	acpi_handle handle;
 	union acpi_object *info;
-	struct amdgpu_atcs *atcs = &adev->atcs;
+	struct amdgpu_atcs *atcs = &amdgpu_acpi_priv.atcs;
 	struct atcs_pref_req_input atcs_input;
 	struct atcs_pref_req_output atcs_output;
 	struct acpi_buffer params;
 	size_t size;
 	u32 retry = 3;
 
-	/* Get the device handle */
-	handle = ACPI_HANDLE(&adev->pdev->dev);
-	if (!handle)
+	if (amdgpu_acpi_pcie_notify_device_ready(adev))
 		return -EINVAL;
 
 	if (!atcs->functions.pcie_perf_req)
@@ -550,7 +721,7 @@ int amdgpu_acpi_pcie_performance_request(struct amdgpu_device *adev,
 
 	atcs_input.size = sizeof(struct atcs_pref_req_input);
 	/* client id (bit 2-0: func num, 7-3: dev num, 15-8: bus num) */
-	atcs_input.client_id = adev->pdev->devfn | (adev->pdev->bus->number << 8);
+	atcs_input.client_id = pci_dev_id(adev->pdev);
 	atcs_input.valid_flags_mask = ATCS_VALID_FLAGS_MASK;
 	atcs_input.flags = ATCS_WAIT_FOR_COMPLETION;
 	if (advertise)
@@ -562,7 +733,7 @@ int amdgpu_acpi_pcie_performance_request(struct amdgpu_device *adev,
 	params.pointer = &atcs_input;
 
 	while (retry--) {
-		info = amdgpu_atcs_call(handle, ATCS_FUNCTION_PCIE_PERFORMANCE_REQUEST, &params);
+		info = amdgpu_atcs_call(atcs, ATCS_FUNCTION_PCIE_PERFORMANCE_REQUEST, &params);
 		if (!info)
 			return -EIO;
 
@@ -596,6 +767,438 @@ int amdgpu_acpi_pcie_performance_request(struct amdgpu_device *adev,
 }
 
 /**
+ * amdgpu_acpi_power_shift_control
+ *
+ * @adev: amdgpu_device pointer
+ * @dev_state: device acpi state
+ * @drv_state: driver state
+ *
+ * Executes the POWER_SHIFT_CONTROL method to
+ * communicate current dGPU device state and
+ * driver state to APU/SBIOS.
+ * returns 0 on success, error on failure.
+ */
+int amdgpu_acpi_power_shift_control(struct amdgpu_device *adev,
+				    u8 dev_state, bool drv_state)
+{
+	union acpi_object *info;
+	struct amdgpu_atcs *atcs = &amdgpu_acpi_priv.atcs;
+	struct atcs_pwr_shift_input atcs_input;
+	struct acpi_buffer params;
+
+	if (!amdgpu_acpi_is_power_shift_control_supported())
+		return -EINVAL;
+
+	atcs_input.size = sizeof(struct atcs_pwr_shift_input);
+	/* dGPU id (bit 2-0: func num, 7-3: dev num, 15-8: bus num) */
+	atcs_input.dgpu_id = pci_dev_id(adev->pdev);
+	atcs_input.dev_acpi_state = dev_state;
+	atcs_input.drv_state = drv_state;
+
+	params.length = sizeof(struct atcs_pwr_shift_input);
+	params.pointer = &atcs_input;
+
+	info = amdgpu_atcs_call(atcs, ATCS_FUNCTION_POWER_SHIFT_CONTROL, &params);
+	if (!info) {
+		DRM_ERROR("ATCS PSC update failed\n");
+		return -EIO;
+	}
+
+	kfree(info);
+	return 0;
+}
+
+/**
+ * amdgpu_acpi_smart_shift_update - update dGPU device state to SBIOS
+ *
+ * @adev: amdgpu device pointer
+ * @ss_state: current smart shift event
+ *
+ * returns 0 on success,
+ * otherwise return error number.
+ */
+int amdgpu_acpi_smart_shift_update(struct amdgpu_device *adev,
+				   enum amdgpu_ss ss_state)
+{
+	int r;
+
+	if (!amdgpu_device_supports_smart_shift(adev))
+		return 0;
+
+	switch (ss_state) {
+	/* SBIOS trigger “stop”, “enable” and “start” at D0, Driver Operational.
+	 * SBIOS trigger “stop” at D3, Driver Not Operational.
+	 * SBIOS trigger “stop” and “disable” at D0, Driver NOT operational.
+	 */
+	case AMDGPU_SS_DRV_LOAD:
+		r = amdgpu_acpi_power_shift_control(adev,
+						    AMDGPU_ATCS_PSC_DEV_STATE_D0,
+						    AMDGPU_ATCS_PSC_DRV_STATE_OPR);
+		break;
+	case AMDGPU_SS_DEV_D0:
+		r = amdgpu_acpi_power_shift_control(adev,
+						    AMDGPU_ATCS_PSC_DEV_STATE_D0,
+						    AMDGPU_ATCS_PSC_DRV_STATE_OPR);
+		break;
+	case AMDGPU_SS_DEV_D3:
+		r = amdgpu_acpi_power_shift_control(adev,
+						    AMDGPU_ATCS_PSC_DEV_STATE_D3_HOT,
+						    AMDGPU_ATCS_PSC_DRV_STATE_NOT_OPR);
+		break;
+	case AMDGPU_SS_DRV_UNLOAD:
+		r = amdgpu_acpi_power_shift_control(adev,
+						    AMDGPU_ATCS_PSC_DEV_STATE_D0,
+						    AMDGPU_ATCS_PSC_DRV_STATE_NOT_OPR);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return r;
+}
+
+#ifdef CONFIG_ACPI_NUMA
+static inline uint64_t amdgpu_acpi_get_numa_size(int nid)
+{
+	/* This is directly using si_meminfo_node implementation as the
+	 * function is not exported.
+	 */
+	int zone_type;
+	uint64_t managed_pages = 0;
+
+	pg_data_t *pgdat = NODE_DATA(nid);
+
+	for (zone_type = 0; zone_type < MAX_NR_ZONES; zone_type++)
+		managed_pages +=
+			zone_managed_pages(&pgdat->node_zones[zone_type]);
+	return managed_pages * PAGE_SIZE;
+}
+
+static struct amdgpu_numa_info *amdgpu_acpi_get_numa_info(uint32_t pxm)
+{
+	struct amdgpu_numa_info *numa_info;
+	int nid;
+
+	numa_info = xa_load(&numa_info_xa, pxm);
+
+	if (!numa_info) {
+		struct sysinfo info;
+
+		numa_info = kzalloc(sizeof(*numa_info), GFP_KERNEL);
+		if (!numa_info)
+			return NULL;
+
+		nid = pxm_to_node(pxm);
+		numa_info->pxm = pxm;
+		numa_info->nid = nid;
+
+		if (numa_info->nid == NUMA_NO_NODE) {
+			si_meminfo(&info);
+			numa_info->size = info.totalram * info.mem_unit;
+		} else {
+			numa_info->size = amdgpu_acpi_get_numa_size(nid);
+		}
+		xa_store(&numa_info_xa, numa_info->pxm, numa_info, GFP_KERNEL);
+	}
+
+	return numa_info;
+}
+#endif
+
+/**
+ * amdgpu_acpi_get_node_id - obtain the NUMA node id for corresponding amdgpu
+ * acpi device handle
+ *
+ * @handle: acpi handle
+ * @numa_info: amdgpu_numa_info structure holding numa information
+ *
+ * Queries the ACPI interface to fetch the corresponding NUMA Node ID for a
+ * given amdgpu acpi device.
+ *
+ * Returns ACPI STATUS OK with Node ID on success or the corresponding failure reason
+ */
+static acpi_status amdgpu_acpi_get_node_id(acpi_handle handle,
+				    struct amdgpu_numa_info **numa_info)
+{
+#ifdef CONFIG_ACPI_NUMA
+	u64 pxm;
+	acpi_status status;
+
+	if (!numa_info)
+		return_ACPI_STATUS(AE_ERROR);
+
+	status = acpi_evaluate_integer(handle, "_PXM", NULL, &pxm);
+
+	if (ACPI_FAILURE(status))
+		return status;
+
+	*numa_info = amdgpu_acpi_get_numa_info(pxm);
+
+	if (!*numa_info)
+		return_ACPI_STATUS(AE_ERROR);
+
+	return_ACPI_STATUS(AE_OK);
+#else
+	return_ACPI_STATUS(AE_NOT_EXIST);
+#endif
+}
+
+static struct amdgpu_acpi_dev_info *amdgpu_acpi_get_dev(u32 sbdf)
+{
+	struct amdgpu_acpi_dev_info *acpi_dev;
+
+	if (list_empty(&amdgpu_acpi_dev_list))
+		return NULL;
+
+	list_for_each_entry(acpi_dev, &amdgpu_acpi_dev_list, list)
+		if (acpi_dev->sbdf == sbdf)
+			return acpi_dev;
+
+	return NULL;
+}
+
+static int amdgpu_acpi_dev_init(struct amdgpu_acpi_dev_info **dev_info,
+				struct amdgpu_acpi_xcc_info *xcc_info, u32 sbdf)
+{
+	struct amdgpu_acpi_dev_info *tmp;
+	union acpi_object *obj;
+	int ret = -ENOENT;
+
+	*dev_info = NULL;
+	tmp = kzalloc(sizeof(struct amdgpu_acpi_dev_info), GFP_KERNEL);
+	if (!tmp)
+		return -ENOMEM;
+
+	INIT_LIST_HEAD(&tmp->xcc_list);
+	INIT_LIST_HEAD(&tmp->list);
+	tmp->sbdf = sbdf;
+
+	obj = acpi_evaluate_dsm_typed(xcc_info->handle, &amd_xcc_dsm_guid, 0,
+				      AMD_XCC_DSM_GET_SUPP_MODE, NULL,
+				      ACPI_TYPE_INTEGER);
+
+	if (!obj) {
+		acpi_handle_debug(xcc_info->handle,
+				  "_DSM function %d evaluation failed",
+				  AMD_XCC_DSM_GET_SUPP_MODE);
+		ret = -ENOENT;
+		goto out;
+	}
+
+	tmp->supp_xcp_mode = obj->integer.value & 0xFFFF;
+	ACPI_FREE(obj);
+
+	obj = acpi_evaluate_dsm_typed(xcc_info->handle, &amd_xcc_dsm_guid, 0,
+				      AMD_XCC_DSM_GET_XCP_MODE, NULL,
+				      ACPI_TYPE_INTEGER);
+
+	if (!obj) {
+		acpi_handle_debug(xcc_info->handle,
+				  "_DSM function %d evaluation failed",
+				  AMD_XCC_DSM_GET_XCP_MODE);
+		ret = -ENOENT;
+		goto out;
+	}
+
+	tmp->xcp_mode = obj->integer.value & 0xFFFF;
+	tmp->mem_mode = (obj->integer.value >> 32) & 0xFFFF;
+	ACPI_FREE(obj);
+
+	/* Evaluate DSMs and fill XCC information */
+	obj = acpi_evaluate_dsm_typed(xcc_info->handle, &amd_xcc_dsm_guid, 0,
+				      AMD_XCC_DSM_GET_TMR_INFO, NULL,
+				      ACPI_TYPE_PACKAGE);
+
+	if (!obj || obj->package.count < 2) {
+		acpi_handle_debug(xcc_info->handle,
+				  "_DSM function %d evaluation failed",
+				  AMD_XCC_DSM_GET_TMR_INFO);
+		ret = -ENOENT;
+		goto out;
+	}
+
+	tmp->tmr_base = obj->package.elements[0].integer.value;
+	tmp->tmr_size = obj->package.elements[1].integer.value;
+	ACPI_FREE(obj);
+
+	DRM_DEBUG_DRIVER(
+		"New dev(%x): Supported xcp mode: %x curr xcp_mode : %x mem mode : %x, tmr base: %llx tmr size: %llx  ",
+		tmp->sbdf, tmp->supp_xcp_mode, tmp->xcp_mode, tmp->mem_mode,
+		tmp->tmr_base, tmp->tmr_size);
+	list_add_tail(&tmp->list, &amdgpu_acpi_dev_list);
+	*dev_info = tmp;
+
+	return 0;
+
+out:
+	if (obj)
+		ACPI_FREE(obj);
+	kfree(tmp);
+
+	return ret;
+}
+
+static int amdgpu_acpi_get_xcc_info(struct amdgpu_acpi_xcc_info *xcc_info,
+				    u32 *sbdf)
+{
+	union acpi_object *obj;
+	acpi_status status;
+	int ret = -ENOENT;
+
+	obj = acpi_evaluate_dsm_typed(xcc_info->handle, &amd_xcc_dsm_guid, 0,
+				      AMD_XCC_DSM_GET_NUM_FUNCS, NULL,
+				      ACPI_TYPE_INTEGER);
+
+	if (!obj || obj->integer.value != AMD_XCC_DSM_NUM_FUNCS)
+		goto out;
+	ACPI_FREE(obj);
+
+	/* Evaluate DSMs and fill XCC information */
+	obj = acpi_evaluate_dsm_typed(xcc_info->handle, &amd_xcc_dsm_guid, 0,
+				      AMD_XCC_DSM_GET_VF_XCC_MAPPING, NULL,
+				      ACPI_TYPE_INTEGER);
+
+	if (!obj) {
+		acpi_handle_debug(xcc_info->handle,
+				  "_DSM function %d evaluation failed",
+				  AMD_XCC_DSM_GET_VF_XCC_MAPPING);
+		ret = -EINVAL;
+		goto out;
+	}
+
+	/* PF xcc id [39:32] */
+	xcc_info->phy_id = (obj->integer.value >> 32) & 0xFF;
+	/* xcp node of this xcc [47:40] */
+	xcc_info->xcp_node = (obj->integer.value >> 40) & 0xFF;
+	/* PF domain of this xcc [31:16] */
+	*sbdf = (obj->integer.value) & 0xFFFF0000;
+	/* PF bus/dev/fn of this xcc [63:48] */
+	*sbdf |= (obj->integer.value >> 48) & 0xFFFF;
+	ACPI_FREE(obj);
+	obj = NULL;
+
+	status =
+		amdgpu_acpi_get_node_id(xcc_info->handle, &xcc_info->numa_info);
+
+	/* TODO: check if this check is required */
+	if (ACPI_SUCCESS(status))
+		ret = 0;
+out:
+	if (obj)
+		ACPI_FREE(obj);
+
+	return ret;
+}
+
+static int amdgpu_acpi_enumerate_xcc(void)
+{
+	struct amdgpu_acpi_dev_info *dev_info = NULL;
+	struct amdgpu_acpi_xcc_info *xcc_info;
+	struct acpi_device *acpi_dev;
+	char hid[ACPI_ID_LEN];
+	int ret, id;
+	u32 sbdf;
+
+	INIT_LIST_HEAD(&amdgpu_acpi_dev_list);
+	xa_init(&numa_info_xa);
+
+	for (id = 0; id < AMD_XCC_MAX_HID; id++) {
+		sprintf(hid, "%s%d", "AMD", AMD_XCC_HID_START + id);
+		acpi_dev = acpi_dev_get_first_match_dev(hid, NULL, -1);
+		/* These ACPI objects are expected to be in sequential order. If
+		 * one is not found, no need to check the rest.
+		 */
+		if (!acpi_dev) {
+			DRM_DEBUG_DRIVER("No matching acpi device found for %s",
+					 hid);
+			break;
+		}
+
+		xcc_info = kzalloc(sizeof(struct amdgpu_acpi_xcc_info),
+				   GFP_KERNEL);
+		if (!xcc_info) {
+			DRM_ERROR("Failed to allocate memory for xcc info\n");
+			return -ENOMEM;
+		}
+
+		INIT_LIST_HEAD(&xcc_info->list);
+		xcc_info->handle = acpi_device_handle(acpi_dev);
+		acpi_dev_put(acpi_dev);
+
+		ret = amdgpu_acpi_get_xcc_info(xcc_info, &sbdf);
+		if (ret) {
+			kfree(xcc_info);
+			continue;
+		}
+
+		dev_info = amdgpu_acpi_get_dev(sbdf);
+
+		if (!dev_info)
+			ret = amdgpu_acpi_dev_init(&dev_info, xcc_info, sbdf);
+
+		if (ret == -ENOMEM)
+			return ret;
+
+		if (!dev_info) {
+			kfree(xcc_info);
+			continue;
+		}
+
+		list_add_tail(&xcc_info->list, &dev_info->xcc_list);
+	}
+
+	return 0;
+}
+
+int amdgpu_acpi_get_tmr_info(struct amdgpu_device *adev, u64 *tmr_offset,
+			     u64 *tmr_size)
+{
+	struct amdgpu_acpi_dev_info *dev_info;
+	u32 sbdf;
+
+	if (!tmr_offset || !tmr_size)
+		return -EINVAL;
+
+	sbdf = (pci_domain_nr(adev->pdev->bus) << 16);
+	sbdf |= pci_dev_id(adev->pdev);
+	dev_info = amdgpu_acpi_get_dev(sbdf);
+	if (!dev_info)
+		return -ENOENT;
+
+	*tmr_offset = dev_info->tmr_base;
+	*tmr_size = dev_info->tmr_size;
+
+	return 0;
+}
+
+int amdgpu_acpi_get_mem_info(struct amdgpu_device *adev, int xcc_id,
+			     struct amdgpu_numa_info *numa_info)
+{
+	struct amdgpu_acpi_dev_info *dev_info;
+	struct amdgpu_acpi_xcc_info *xcc_info;
+	u32 sbdf;
+
+	if (!numa_info)
+		return -EINVAL;
+
+	sbdf = (pci_domain_nr(adev->pdev->bus) << 16);
+	sbdf |= pci_dev_id(adev->pdev);
+	dev_info = amdgpu_acpi_get_dev(sbdf);
+	if (!dev_info)
+		return -ENOENT;
+
+	list_for_each_entry(xcc_info, &dev_info->xcc_list, list) {
+		if (xcc_info->phy_id == xcc_id) {
+			memcpy(numa_info, xcc_info->numa_info,
+			       sizeof(*numa_info));
+			return 0;
+		}
+	}
+
+	return -ENOENT;
+}
+
+/**
  * amdgpu_acpi_event - handle notify events
  *
  * @nb: notifier block
@@ -638,47 +1241,181 @@ static int amdgpu_acpi_event(struct notifier_block *nb,
  */
 int amdgpu_acpi_init(struct amdgpu_device *adev)
 {
-	acpi_handle handle;
-	struct amdgpu_atif *atif = &adev->atif;
-	struct amdgpu_atcs *atcs = &adev->atcs;
+	struct amdgpu_atif *atif = &amdgpu_acpi_priv.atif;
+
+	if (atif->notifications.brightness_change) {
+		if (adev->dc_enabled) {
+#if defined(CONFIG_DRM_AMD_DC)
+			struct amdgpu_display_manager *dm = &adev->dm;
+
+			if (dm->backlight_dev[0])
+				atif->bd = dm->backlight_dev[0];
+#endif
+		} else {
+			struct drm_encoder *tmp;
+
+			/* Find the encoder controlling the brightness */
+			list_for_each_entry(tmp, &adev_to_drm(adev)->mode_config.encoder_list,
+					    head) {
+				struct amdgpu_encoder *enc = to_amdgpu_encoder(tmp);
+
+				if ((enc->devices & (ATOM_DEVICE_LCD_SUPPORT)) &&
+				    enc->enc_priv) {
+					struct amdgpu_encoder_atom_dig *dig = enc->enc_priv;
+
+					if (dig->bl_dev) {
+						atif->bd = dig->bl_dev;
+						break;
+					}
+				}
+			}
+		}
+	}
+	adev->acpi_nb.notifier_call = amdgpu_acpi_event;
+	register_acpi_notifier(&adev->acpi_nb);
+
+	return 0;
+}
+
+void amdgpu_acpi_get_backlight_caps(struct amdgpu_dm_backlight_caps *caps)
+{
+	struct amdgpu_atif *atif = &amdgpu_acpi_priv.atif;
+
+	memcpy(caps, &atif->backlight_caps, sizeof(*caps));
+}
+
+/**
+ * amdgpu_acpi_fini - tear down driver acpi support
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Unregisters with the acpi notifier chain (all asics).
+ */
+void amdgpu_acpi_fini(struct amdgpu_device *adev)
+{
+	unregister_acpi_notifier(&adev->acpi_nb);
+}
+
+/**
+ * amdgpu_atif_pci_probe_handle - look up the ATIF handle
+ *
+ * @pdev: pci device
+ *
+ * Look up the ATIF handles (all asics).
+ * Returns true if the handle is found, false if not.
+ */
+static bool amdgpu_atif_pci_probe_handle(struct pci_dev *pdev)
+{
+	char acpi_method_name[255] = { 0 };
+	struct acpi_buffer buffer = {sizeof(acpi_method_name), acpi_method_name};
+	acpi_handle dhandle, atif_handle;
+	acpi_status status;
 	int ret;
 
-	/* Get the device handle */
-	handle = ACPI_HANDLE(&adev->pdev->dev);
+	dhandle = ACPI_HANDLE(&pdev->dev);
+	if (!dhandle)
+		return false;
 
-	if (!adev->bios || !handle)
-		return 0;
+	status = acpi_get_handle(dhandle, "ATIF", &atif_handle);
+	if (ACPI_FAILURE(status))
+		return false;
 
-	/* Call the ATCS method */
-	ret = amdgpu_atcs_verify_interface(handle, atcs);
+	amdgpu_acpi_priv.atif.handle = atif_handle;
+	acpi_get_name(amdgpu_acpi_priv.atif.handle, ACPI_FULL_PATHNAME, &buffer);
+	DRM_DEBUG_DRIVER("Found ATIF handle %s\n", acpi_method_name);
+	ret = amdgpu_atif_verify_interface(&amdgpu_acpi_priv.atif);
 	if (ret) {
-		DRM_DEBUG_DRIVER("Call to ATCS verify_interface failed: %d\n", ret);
+		amdgpu_acpi_priv.atif.handle = 0;
+		return false;
 	}
+	return true;
+}
+
+/**
+ * amdgpu_atcs_pci_probe_handle - look up the ATCS handle
+ *
+ * @pdev: pci device
+ *
+ * Look up the ATCS handles (all asics).
+ * Returns true if the handle is found, false if not.
+ */
+static bool amdgpu_atcs_pci_probe_handle(struct pci_dev *pdev)
+{
+	char acpi_method_name[255] = { 0 };
+	struct acpi_buffer buffer = { sizeof(acpi_method_name), acpi_method_name };
+	acpi_handle dhandle, atcs_handle;
+	acpi_status status;
+	int ret;
+
+	dhandle = ACPI_HANDLE(&pdev->dev);
+	if (!dhandle)
+		return false;
 
-	/* Call the ATIF method */
-	ret = amdgpu_atif_verify_interface(handle, atif);
+	status = acpi_get_handle(dhandle, "ATCS", &atcs_handle);
+	if (ACPI_FAILURE(status))
+		return false;
+
+	amdgpu_acpi_priv.atcs.handle = atcs_handle;
+	acpi_get_name(amdgpu_acpi_priv.atcs.handle, ACPI_FULL_PATHNAME, &buffer);
+	DRM_DEBUG_DRIVER("Found ATCS handle %s\n", acpi_method_name);
+	ret = amdgpu_atcs_verify_interface(&amdgpu_acpi_priv.atcs);
 	if (ret) {
-		DRM_DEBUG_DRIVER("Call to ATIF verify_interface failed: %d\n", ret);
-		goto out;
+		amdgpu_acpi_priv.atcs.handle = 0;
+		return false;
 	}
+	return true;
+}
 
-	if (atif->notifications.brightness_change) {
-		struct drm_encoder *tmp;
-
-		/* Find the encoder controlling the brightness */
-		list_for_each_entry(tmp, &adev->ddev->mode_config.encoder_list,
-				head) {
-			struct amdgpu_encoder *enc = to_amdgpu_encoder(tmp);
-
-			if ((enc->devices & (ATOM_DEVICE_LCD_SUPPORT)) &&
-			    enc->enc_priv) {
-				struct amdgpu_encoder_atom_dig *dig = enc->enc_priv;
-				if (dig->bl_dev) {
-					atif->encoder_for_bl = enc;
-					break;
-				}
-			}
-		}
+
+/**
+ * amdgpu_acpi_should_gpu_reset
+ *
+ * @adev: amdgpu_device_pointer
+ *
+ * returns true if should reset GPU, false if not
+ */
+bool amdgpu_acpi_should_gpu_reset(struct amdgpu_device *adev)
+{
+	if ((adev->flags & AMD_IS_APU) &&
+	    adev->gfx.imu.funcs) /* Not need to do mode2 reset for IMU enabled APUs */
+		return false;
+
+	if ((adev->flags & AMD_IS_APU) &&
+	    amdgpu_acpi_is_s3_active(adev))
+		return false;
+
+	if (amdgpu_sriov_vf(adev))
+		return false;
+
+#if IS_ENABLED(CONFIG_SUSPEND)
+	return pm_suspend_target_state != PM_SUSPEND_TO_IDLE;
+#else
+	return true;
+#endif
+}
+
+/*
+ * amdgpu_acpi_detect - detect ACPI ATIF/ATCS methods
+ *
+ * Check if we have the ATIF/ATCS methods and populate
+ * the structures in the driver.
+ */
+void amdgpu_acpi_detect(void)
+{
+	struct amdgpu_atif *atif = &amdgpu_acpi_priv.atif;
+	struct amdgpu_atcs *atcs = &amdgpu_acpi_priv.atcs;
+	struct pci_dev *pdev = NULL;
+	int ret;
+
+	while ((pdev = pci_get_base_class(PCI_BASE_CLASS_DISPLAY, pdev))) {
+		if ((pdev->class != PCI_CLASS_DISPLAY_VGA << 8) &&
+		    (pdev->class != PCI_CLASS_DISPLAY_OTHER << 8))
+			continue;
+
+		if (!atif->handle)
+			amdgpu_atif_pci_probe_handle(pdev);
+		if (!atcs->handle)
+			amdgpu_atcs_pci_probe_handle(pdev);
 	}
 
 	if (atif->functions.sbios_requests && !atif->functions.system_params) {
@@ -690,8 +1427,7 @@ int amdgpu_acpi_init(struct amdgpu_device *adev)
 	}
 
 	if (atif->functions.system_params) {
-		ret = amdgpu_atif_get_notification_params(handle,
-				&atif->notification_cfg);
+		ret = amdgpu_atif_get_notification_params(atif);
 		if (ret) {
 			DRM_DEBUG_DRIVER("Call to GET_SYSTEM_PARAMS failed: %d\n",
 					ret);
@@ -700,21 +1436,131 @@ int amdgpu_acpi_init(struct amdgpu_device *adev)
 		}
 	}
 
-out:
-	adev->acpi_nb.notifier_call = amdgpu_acpi_event;
-	register_acpi_notifier(&adev->acpi_nb);
+	if (atif->functions.query_backlight_transfer_characteristics) {
+		ret = amdgpu_atif_query_backlight_caps(atif);
+		if (ret) {
+			DRM_DEBUG_DRIVER("Call to QUERY_BACKLIGHT_TRANSFER_CHARACTERISTICS failed: %d\n",
+					ret);
+			atif->backlight_caps.caps_valid = false;
+		}
+	} else {
+		atif->backlight_caps.caps_valid = false;
+	}
 
-	return ret;
+	amdgpu_acpi_enumerate_xcc();
+}
+
+void amdgpu_acpi_release(void)
+{
+	struct amdgpu_acpi_dev_info *dev_info, *dev_tmp;
+	struct amdgpu_acpi_xcc_info *xcc_info, *xcc_tmp;
+	struct amdgpu_numa_info *numa_info;
+	unsigned long index;
+
+	xa_for_each(&numa_info_xa, index, numa_info) {
+		kfree(numa_info);
+		xa_erase(&numa_info_xa, index);
+	}
+
+	if (list_empty(&amdgpu_acpi_dev_list))
+		return;
+
+	list_for_each_entry_safe(dev_info, dev_tmp, &amdgpu_acpi_dev_list,
+				 list) {
+		list_for_each_entry_safe(xcc_info, xcc_tmp, &dev_info->xcc_list,
+					 list) {
+			list_del(&xcc_info->list);
+			kfree(xcc_info);
+		}
+
+		list_del(&dev_info->list);
+		kfree(dev_info);
+	}
 }
 
+#if IS_ENABLED(CONFIG_SUSPEND)
 /**
- * amdgpu_acpi_fini - tear down driver acpi support
+ * amdgpu_acpi_is_s3_active
  *
- * @adev: amdgpu_device pointer
+ * @adev: amdgpu_device_pointer
  *
- * Unregisters with the acpi notifier chain (all asics).
+ * returns true if supported, false if not.
  */
-void amdgpu_acpi_fini(struct amdgpu_device *adev)
+bool amdgpu_acpi_is_s3_active(struct amdgpu_device *adev)
 {
-	unregister_acpi_notifier(&adev->acpi_nb);
+	return !(adev->flags & AMD_IS_APU) ||
+		(pm_suspend_target_state == PM_SUSPEND_MEM);
+}
+
+/**
+ * amdgpu_acpi_is_s0ix_active
+ *
+ * @adev: amdgpu_device_pointer
+ *
+ * returns true if supported, false if not.
+ */
+bool amdgpu_acpi_is_s0ix_active(struct amdgpu_device *adev)
+{
+	if (!(adev->flags & AMD_IS_APU) ||
+	    (pm_suspend_target_state != PM_SUSPEND_TO_IDLE))
+		return false;
+
+	if (adev->asic_type < CHIP_RAVEN)
+		return false;
+
+	if (!(adev->pm.pp_feature & PP_GFXOFF_MASK))
+		return false;
+
+	/*
+	 * If ACPI_FADT_LOW_POWER_S0 is not set in the FADT, it is generally
+	 * risky to do any special firmware-related preparations for entering
+	 * S0ix even though the system is suspending to idle, so return false
+	 * in that case.
+	 */
+	if (!(acpi_gbl_FADT.flags & ACPI_FADT_LOW_POWER_S0)) {
+		dev_err_once(adev->dev,
+			      "Power consumption will be higher as BIOS has not been configured for suspend-to-idle.\n"
+			      "To use suspend-to-idle change the sleep mode in BIOS setup.\n");
+		return false;
+	}
+
+#if !IS_ENABLED(CONFIG_AMD_PMC)
+	dev_err_once(adev->dev,
+		      "Power consumption will be higher as the kernel has not been compiled with CONFIG_AMD_PMC.\n");
+	return false;
+#else
+	return true;
+#endif /* CONFIG_AMD_PMC */
+}
+#endif /* CONFIG_SUSPEND */
+
+#if IS_ENABLED(CONFIG_DRM_AMD_ISP)
+static const struct acpi_device_id isp_sensor_ids[] = {
+	{ "OMNI5C10" },
+	{ }
+};
+
+static int isp_match_acpi_device_ids(struct device *dev, const void *data)
+{
+	return acpi_match_device(data, dev) ? 1 : 0;
+}
+
+int amdgpu_acpi_get_isp4_dev(struct acpi_device **dev)
+{
+	struct device *pdev __free(put_device) = NULL;
+	struct acpi_device *acpi_pdev;
+
+	pdev = bus_find_device(&platform_bus_type, NULL, isp_sensor_ids,
+			       isp_match_acpi_device_ids);
+	if (!pdev)
+		return -EINVAL;
+
+	acpi_pdev = ACPI_COMPANION(pdev);
+	if (!acpi_pdev)
+		return -ENODEV;
+
+	*dev = acpi_pdev;
+
+	return 0;
 }
+#endif /* CONFIG_DRM_AMD_ISP */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_afmt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_afmt.c
index 3889486f71fe..80771b1480ff 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_afmt.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_afmt.c
@@ -25,7 +25,7 @@
  */
 #include <linux/hdmi.h>
 #include <linux/gcd.h>
-#include <drm/drmP.h>
+
 #include <drm/amdgpu_drm.h>
 #include "amdgpu.h"
 
@@ -100,6 +100,7 @@ struct amdgpu_afmt_acr amdgpu_afmt_acr(uint32_t clock)
 	amdgpu_afmt_calc_cts(clock, &res.cts_32khz, &res.n_32khz, 32000);
 	amdgpu_afmt_calc_cts(clock, &res.cts_44_1khz, &res.n_44_1khz, 44100);
 	amdgpu_afmt_calc_cts(clock, &res.cts_48khz, &res.n_48khz, 48000);
+	res.clock = clock;
 
 	return res;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
index 5432af39a674..a2879d2b7c8e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
@@ -1,3 +1,4 @@
+// SPDX-License-Identifier: MIT
 /*
  * Copyright 2014 Advanced Micro Devices, Inc.
  *
@@ -21,170 +22,322 @@
  */
 
 #include "amdgpu_amdkfd.h"
+#include "amd_pcie.h"
 #include "amd_shared.h"
-#include <drm/drmP.h>
+
 #include "amdgpu.h"
 #include "amdgpu_gfx.h"
+#include "amdgpu_dma_buf.h"
+#include <drm/ttm/ttm_tt.h>
 #include <linux/module.h>
+#include <linux/dma-buf.h>
+#include "amdgpu_xgmi.h"
+#include <uapi/linux/kfd_ioctl.h>
+#include "amdgpu_ras.h"
+#include "amdgpu_umc.h"
+#include "amdgpu_reset.h"
+
+/* Total memory size in system memory and all GPU VRAM. Used to
+ * estimate worst case amount of memory to reserve for page tables
+ */
+uint64_t amdgpu_amdkfd_total_mem_size;
 
-const struct kgd2kfd_calls *kgd2kfd;
-bool (*kgd2kfd_init_p)(unsigned int, const struct kgd2kfd_calls**);
+static bool kfd_initialized;
 
 int amdgpu_amdkfd_init(void)
 {
+	struct sysinfo si;
 	int ret;
 
-#if defined(CONFIG_HSA_AMD_MODULE)
-	int (*kgd2kfd_init_p)(unsigned int, const struct kgd2kfd_calls**);
+	si_meminfo(&si);
+	amdgpu_amdkfd_total_mem_size = si.freeram - si.freehigh;
+	amdgpu_amdkfd_total_mem_size *= si.mem_unit;
 
-	kgd2kfd_init_p = symbol_request(kgd2kfd_init);
+	ret = kgd2kfd_init();
+	kfd_initialized = !ret;
 
-	if (kgd2kfd_init_p == NULL)
-		return -ENOENT;
+	return ret;
+}
 
-	ret = kgd2kfd_init_p(KFD_INTERFACE_VERSION, &kgd2kfd);
-	if (ret) {
-		symbol_put(kgd2kfd_init);
-		kgd2kfd = NULL;
+void amdgpu_amdkfd_fini(void)
+{
+	if (kfd_initialized) {
+		kgd2kfd_exit();
+		kfd_initialized = false;
 	}
+}
 
-#elif defined(CONFIG_HSA_AMD)
-	ret = kgd2kfd_init(KFD_INTERFACE_VERSION, &kgd2kfd);
-	if (ret)
-		kgd2kfd = NULL;
+void amdgpu_amdkfd_device_probe(struct amdgpu_device *adev)
+{
+	bool vf = amdgpu_sriov_vf(adev);
 
-#else
-	ret = -ENOENT;
-#endif
+	if (!kfd_initialized)
+		return;
 
-	return ret;
+	adev->kfd.dev = kgd2kfd_probe(adev, vf);
 }
 
-void amdgpu_amdkfd_fini(void)
+/**
+ * amdgpu_doorbell_get_kfd_info - Report doorbell configuration required to
+ *                                setup amdkfd
+ *
+ * @adev: amdgpu_device pointer
+ * @aperture_base: output returning doorbell aperture base physical address
+ * @aperture_size: output returning doorbell aperture size in bytes
+ * @start_offset: output returning # of doorbell bytes reserved for amdgpu.
+ *
+ * amdgpu and amdkfd share the doorbell aperture. amdgpu sets it up,
+ * takes doorbells required for its own rings and reports the setup to amdkfd.
+ * amdgpu reserved doorbells are at the start of the doorbell aperture.
+ */
+static void amdgpu_doorbell_get_kfd_info(struct amdgpu_device *adev,
+					 phys_addr_t *aperture_base,
+					 size_t *aperture_size,
+					 size_t *start_offset)
 {
-	if (kgd2kfd) {
-		kgd2kfd->exit();
-		symbol_put(kgd2kfd_init);
+	/*
+	 * The first num_kernel_doorbells are used by amdgpu.
+	 * amdkfd takes whatever's left in the aperture.
+	 */
+	if (adev->enable_mes) {
+		/*
+		 * With MES enabled, we only need to initialize
+		 * the base address. The size and offset are
+		 * not initialized as AMDGPU manages the whole
+		 * doorbell space.
+		 */
+		*aperture_base = adev->doorbell.base;
+		*aperture_size = 0;
+		*start_offset = 0;
+	} else if (adev->doorbell.size > adev->doorbell.num_kernel_doorbells *
+						sizeof(u32)) {
+		*aperture_base = adev->doorbell.base;
+		*aperture_size = adev->doorbell.size;
+		*start_offset = adev->doorbell.num_kernel_doorbells * sizeof(u32);
+	} else {
+		*aperture_base = 0;
+		*aperture_size = 0;
+		*start_offset = 0;
 	}
 }
 
-void amdgpu_amdkfd_device_probe(struct amdgpu_device *adev)
+
+static void amdgpu_amdkfd_reset_work(struct work_struct *work)
 {
-	const struct kfd2kgd_calls *kfd2kgd;
+	struct amdgpu_device *adev = container_of(work, struct amdgpu_device,
+						  kfd.reset_work);
 
-	if (!kgd2kfd)
-		return;
+	struct amdgpu_reset_context reset_context;
 
-	switch (adev->asic_type) {
-#ifdef CONFIG_DRM_AMDGPU_CIK
-	case CHIP_KAVERI:
-		kfd2kgd = amdgpu_amdkfd_gfx_7_get_functions();
-		break;
-#endif
-	case CHIP_CARRIZO:
-		kfd2kgd = amdgpu_amdkfd_gfx_8_0_get_functions();
-		break;
-	default:
-		dev_info(adev->dev, "kfd not supported on this ASIC\n");
-		return;
+	memset(&reset_context, 0, sizeof(reset_context));
+
+	reset_context.method = AMD_RESET_METHOD_NONE;
+	reset_context.reset_req_dev = adev;
+	reset_context.src = adev->enable_mes ?
+			    AMDGPU_RESET_SRC_MES :
+			    AMDGPU_RESET_SRC_HWS;
+	clear_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
+
+	amdgpu_device_gpu_recover(adev, NULL, &reset_context);
+}
+
+static const struct drm_client_funcs kfd_client_funcs = {
+	.unregister	= drm_client_release,
+};
+
+int amdgpu_amdkfd_drm_client_create(struct amdgpu_device *adev)
+{
+	int ret;
+
+	if (!adev->kfd.init_complete || adev->kfd.client.dev)
+		return 0;
+
+	ret = drm_client_init(&adev->ddev, &adev->kfd.client, "kfd",
+			      &kfd_client_funcs);
+	if (ret) {
+		dev_err(adev->dev, "Failed to init DRM client: %d\n",
+			ret);
+		return ret;
 	}
 
-	adev->kfd = kgd2kfd->probe((struct kgd_dev *)adev,
-				   adev->pdev, kfd2kgd);
+	drm_client_register(&adev->kfd.client);
+
+	return 0;
 }
 
 void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
 {
 	int i;
 	int last_valid_bit;
-	if (adev->kfd) {
+
+	amdgpu_amdkfd_gpuvm_init_mem_limits();
+
+	if (adev->kfd.dev) {
 		struct kgd2kfd_shared_resources gpu_resources = {
-			.compute_vmid_bitmap = 0xFF00,
+			.compute_vmid_bitmap =
+				((1 << AMDGPU_NUM_VMID) - 1) -
+				((1 << adev->vm_manager.first_kfd_vmid) - 1),
 			.num_pipe_per_mec = adev->gfx.mec.num_pipe_per_mec,
-			.num_queue_per_pipe = adev->gfx.mec.num_queue_per_pipe
+			.num_queue_per_pipe = adev->gfx.mec.num_queue_per_pipe,
+			.gpuvm_size = min(adev->vm_manager.max_pfn
+					  << AMDGPU_GPU_PAGE_SHIFT,
+					  AMDGPU_GMC_HOLE_START),
+			.drm_render_minor = adev_to_drm(adev)->render->index,
+			.sdma_doorbell_idx = adev->doorbell_index.sdma_engine,
+			.enable_mes = adev->enable_mes,
 		};
 
 		/* this is going to have a few of the MSBs set that we need to
-		 * clear */
-		bitmap_complement(gpu_resources.queue_bitmap,
-				  adev->gfx.mec.queue_bitmap,
-				  KGD_MAX_QUEUES);
-
-		/* remove the KIQ bit as well */
-		if (adev->gfx.kiq.ring.ready)
-			clear_bit(amdgpu_gfx_queue_to_bit(adev,
-							  adev->gfx.kiq.ring.me - 1,
-							  adev->gfx.kiq.ring.pipe,
-							  adev->gfx.kiq.ring.queue),
-				  gpu_resources.queue_bitmap);
+		 * clear
+		 */
+		bitmap_complement(gpu_resources.cp_queue_bitmap,
+				  adev->gfx.mec_bitmap[0].queue_bitmap,
+				  AMDGPU_MAX_QUEUES);
 
 		/* According to linux/bitmap.h we shouldn't use bitmap_clear if
-		 * nbits is not compile time constant */
+		 * nbits is not compile time constant
+		 */
 		last_valid_bit = 1 /* only first MEC can have compute queues */
 				* adev->gfx.mec.num_pipe_per_mec
 				* adev->gfx.mec.num_queue_per_pipe;
-		for (i = last_valid_bit; i < KGD_MAX_QUEUES; ++i)
-			clear_bit(i, gpu_resources.queue_bitmap);
+		for (i = last_valid_bit; i < AMDGPU_MAX_QUEUES; ++i)
+			clear_bit(i, gpu_resources.cp_queue_bitmap);
 
 		amdgpu_doorbell_get_kfd_info(adev,
 				&gpu_resources.doorbell_physical_address,
 				&gpu_resources.doorbell_aperture_size,
 				&gpu_resources.doorbell_start_offset);
 
-		kgd2kfd->device_init(adev->kfd, &gpu_resources);
+		/* Since SOC15, BIF starts to statically use the
+		 * lower 12 bits of doorbell addresses for routing
+		 * based on settings in registers like
+		 * SDMA0_DOORBELL_RANGE etc..
+		 * In order to route a doorbell to CP engine, the lower
+		 * 12 bits of its address has to be outside the range
+		 * set for SDMA, VCN, and IH blocks.
+		 */
+		if (adev->asic_type >= CHIP_VEGA10) {
+			gpu_resources.non_cp_doorbells_start =
+					adev->doorbell_index.first_non_cp;
+			gpu_resources.non_cp_doorbells_end =
+					adev->doorbell_index.last_non_cp;
+		}
+
+		adev->kfd.init_complete = kgd2kfd_device_init(adev->kfd.dev,
+							&gpu_resources);
+
+		amdgpu_amdkfd_total_mem_size += adev->gmc.real_vram_size;
+
+		INIT_WORK(&adev->kfd.reset_work, amdgpu_amdkfd_reset_work);
 	}
 }
 
-void amdgpu_amdkfd_device_fini(struct amdgpu_device *adev)
+void amdgpu_amdkfd_device_fini_sw(struct amdgpu_device *adev)
 {
-	if (adev->kfd) {
-		kgd2kfd->device_exit(adev->kfd);
-		adev->kfd = NULL;
+	if (adev->kfd.dev) {
+		kgd2kfd_device_exit(adev->kfd.dev);
+		adev->kfd.dev = NULL;
+		amdgpu_amdkfd_total_mem_size -= adev->gmc.real_vram_size;
 	}
 }
 
 void amdgpu_amdkfd_interrupt(struct amdgpu_device *adev,
 		const void *ih_ring_entry)
 {
-	if (adev->kfd)
-		kgd2kfd->interrupt(adev->kfd, ih_ring_entry);
+	if (adev->kfd.dev)
+		kgd2kfd_interrupt(adev->kfd.dev, ih_ring_entry);
+}
+
+void amdgpu_amdkfd_suspend(struct amdgpu_device *adev, bool suspend_proc)
+{
+	if (adev->kfd.dev) {
+		if (adev->in_s0ix)
+			kgd2kfd_stop_sched_all_nodes(adev->kfd.dev);
+		else
+			kgd2kfd_suspend(adev->kfd.dev, suspend_proc);
+	}
+}
+
+int amdgpu_amdkfd_resume(struct amdgpu_device *adev, bool resume_proc)
+{
+	int r = 0;
+
+	if (adev->kfd.dev) {
+		if (adev->in_s0ix)
+			r = kgd2kfd_start_sched_all_nodes(adev->kfd.dev);
+		else
+			r = kgd2kfd_resume(adev->kfd.dev, resume_proc);
+	}
+
+	return r;
+}
+
+void amdgpu_amdkfd_suspend_process(struct amdgpu_device *adev)
+{
+	if (adev->kfd.dev)
+		kgd2kfd_suspend_process(adev->kfd.dev);
+}
+
+int amdgpu_amdkfd_resume_process(struct amdgpu_device *adev)
+{
+	int r = 0;
+
+	if (adev->kfd.dev)
+		r = kgd2kfd_resume_process(adev->kfd.dev);
+
+	return r;
 }
 
-void amdgpu_amdkfd_suspend(struct amdgpu_device *adev)
+int amdgpu_amdkfd_pre_reset(struct amdgpu_device *adev,
+			    struct amdgpu_reset_context *reset_context)
 {
-	if (adev->kfd)
-		kgd2kfd->suspend(adev->kfd);
+	int r = 0;
+
+	if (adev->kfd.dev)
+		r = kgd2kfd_pre_reset(adev->kfd.dev, reset_context);
+
+	return r;
 }
 
-int amdgpu_amdkfd_resume(struct amdgpu_device *adev)
+int amdgpu_amdkfd_post_reset(struct amdgpu_device *adev)
 {
 	int r = 0;
 
-	if (adev->kfd)
-		r = kgd2kfd->resume(adev->kfd);
+	if (adev->kfd.dev)
+		r = kgd2kfd_post_reset(adev->kfd.dev);
 
 	return r;
 }
 
-int alloc_gtt_mem(struct kgd_dev *kgd, size_t size,
-			void **mem_obj, uint64_t *gpu_addr,
-			void **cpu_ptr)
+void amdgpu_amdkfd_gpu_reset(struct amdgpu_device *adev)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
-	struct kgd_mem **mem = (struct kgd_mem **) mem_obj;
+	if (amdgpu_device_should_recover_gpu(adev))
+		amdgpu_reset_domain_schedule(adev->reset_domain,
+					     &adev->kfd.reset_work);
+}
+
+int amdgpu_amdkfd_alloc_gtt_mem(struct amdgpu_device *adev, size_t size,
+				void **mem_obj, uint64_t *gpu_addr,
+				void **cpu_ptr, bool cp_mqd_gfx9)
+{
+	struct amdgpu_bo *bo = NULL;
+	struct amdgpu_bo_param bp;
 	int r;
+	void *cpu_ptr_tmp = NULL;
 
-	BUG_ON(kgd == NULL);
-	BUG_ON(gpu_addr == NULL);
-	BUG_ON(cpu_ptr == NULL);
+	memset(&bp, 0, sizeof(bp));
+	bp.size = size;
+	bp.byte_align = PAGE_SIZE;
+	bp.domain = AMDGPU_GEM_DOMAIN_GTT;
+	bp.flags = AMDGPU_GEM_CREATE_CPU_GTT_USWC;
+	bp.type = ttm_bo_type_kernel;
+	bp.resv = NULL;
+	bp.bo_ptr_size = sizeof(struct amdgpu_bo);
 
-	*mem = kmalloc(sizeof(struct kgd_mem), GFP_KERNEL);
-	if ((*mem) == NULL)
-		return -ENOMEM;
+	if (cp_mqd_gfx9)
+		bp.flags |= AMDGPU_GEM_CREATE_CP_MQD_GFX9;
 
-	r = amdgpu_bo_create(adev, size, PAGE_SIZE, true, AMDGPU_GEM_DOMAIN_GTT,
-			     AMDGPU_GEM_CREATE_CPU_GTT_USWC, NULL, NULL, 0,
-			     &(*mem)->bo);
+	r = amdgpu_bo_create(adev, &bp, &bo);
 	if (r) {
 		dev_err(adev->dev,
 			"failed to allocate BO for amdkfd (%d)\n", r);
@@ -192,79 +345,568 @@ int alloc_gtt_mem(struct kgd_dev *kgd, size_t size,
 	}
 
 	/* map the buffer */
-	r = amdgpu_bo_reserve((*mem)->bo, true);
+	r = amdgpu_bo_reserve(bo, true);
 	if (r) {
 		dev_err(adev->dev, "(%d) failed to reserve bo for amdkfd\n", r);
 		goto allocate_mem_reserve_bo_failed;
 	}
 
-	r = amdgpu_bo_pin((*mem)->bo, AMDGPU_GEM_DOMAIN_GTT,
-				&(*mem)->gpu_addr);
+	r = amdgpu_bo_pin(bo, AMDGPU_GEM_DOMAIN_GTT);
 	if (r) {
 		dev_err(adev->dev, "(%d) failed to pin bo for amdkfd\n", r);
 		goto allocate_mem_pin_bo_failed;
 	}
-	*gpu_addr = (*mem)->gpu_addr;
 
-	r = amdgpu_bo_kmap((*mem)->bo, &(*mem)->cpu_ptr);
+	r = amdgpu_ttm_alloc_gart(&bo->tbo);
+	if (r) {
+		dev_err(adev->dev, "%p bind failed\n", bo);
+		goto allocate_mem_kmap_bo_failed;
+	}
+
+	r = amdgpu_bo_kmap(bo, &cpu_ptr_tmp);
 	if (r) {
 		dev_err(adev->dev,
 			"(%d) failed to map bo to kernel for amdkfd\n", r);
 		goto allocate_mem_kmap_bo_failed;
 	}
-	*cpu_ptr = (*mem)->cpu_ptr;
 
-	amdgpu_bo_unreserve((*mem)->bo);
+	*mem_obj = bo;
+	*gpu_addr = amdgpu_bo_gpu_offset(bo);
+	*cpu_ptr = cpu_ptr_tmp;
+
+	amdgpu_bo_unreserve(bo);
 
 	return 0;
 
 allocate_mem_kmap_bo_failed:
-	amdgpu_bo_unpin((*mem)->bo);
+	amdgpu_bo_unpin(bo);
 allocate_mem_pin_bo_failed:
-	amdgpu_bo_unreserve((*mem)->bo);
+	amdgpu_bo_unreserve(bo);
 allocate_mem_reserve_bo_failed:
-	amdgpu_bo_unref(&(*mem)->bo);
+	amdgpu_bo_unref(&bo);
 
 	return r;
 }
 
-void free_gtt_mem(struct kgd_dev *kgd, void *mem_obj)
+void amdgpu_amdkfd_free_gtt_mem(struct amdgpu_device *adev, void **mem_obj)
 {
-	struct kgd_mem *mem = (struct kgd_mem *) mem_obj;
+	struct amdgpu_bo **bo = (struct amdgpu_bo **) mem_obj;
+
+	if (!bo || !*bo)
+		return;
+
+	(void)amdgpu_bo_reserve(*bo, true);
+	amdgpu_bo_kunmap(*bo);
+	amdgpu_bo_unpin(*bo);
+	amdgpu_bo_unreserve(*bo);
+	amdgpu_bo_unref(bo);
+}
+
+int amdgpu_amdkfd_alloc_gws(struct amdgpu_device *adev, size_t size,
+				void **mem_obj)
+{
+	struct amdgpu_bo *bo = NULL;
+	struct amdgpu_bo_user *ubo;
+	struct amdgpu_bo_param bp;
+	int r;
+
+	memset(&bp, 0, sizeof(bp));
+	bp.size = size;
+	bp.byte_align = 1;
+	bp.domain = AMDGPU_GEM_DOMAIN_GWS;
+	bp.flags = AMDGPU_GEM_CREATE_NO_CPU_ACCESS;
+	bp.type = ttm_bo_type_device;
+	bp.resv = NULL;
+	bp.bo_ptr_size = sizeof(struct amdgpu_bo);
+
+	r = amdgpu_bo_create_user(adev, &bp, &ubo);
+	if (r) {
+		dev_err(adev->dev,
+			"failed to allocate gws BO for amdkfd (%d)\n", r);
+		return r;
+	}
+
+	bo = &ubo->bo;
+	*mem_obj = bo;
+	return 0;
+}
 
-	BUG_ON(mem == NULL);
+void amdgpu_amdkfd_free_gws(struct amdgpu_device *adev, void *mem_obj)
+{
+	struct amdgpu_bo *bo = (struct amdgpu_bo *)mem_obj;
 
-	amdgpu_bo_reserve(mem->bo, true);
-	amdgpu_bo_kunmap(mem->bo);
-	amdgpu_bo_unpin(mem->bo);
-	amdgpu_bo_unreserve(mem->bo);
-	amdgpu_bo_unref(&(mem->bo));
-	kfree(mem);
+	amdgpu_bo_unref(&bo);
 }
 
-uint64_t get_vmem_size(struct kgd_dev *kgd)
+uint32_t amdgpu_amdkfd_get_fw_version(struct amdgpu_device *adev,
+				      enum kgd_engine_type type)
 {
-	struct amdgpu_device *adev =
-		(struct amdgpu_device *)kgd;
+	switch (type) {
+	case KGD_ENGINE_PFP:
+		return adev->gfx.pfp_fw_version;
+
+	case KGD_ENGINE_ME:
+		return adev->gfx.me_fw_version;
+
+	case KGD_ENGINE_CE:
+		return adev->gfx.ce_fw_version;
+
+	case KGD_ENGINE_MEC1:
+		return adev->gfx.mec_fw_version;
+
+	case KGD_ENGINE_MEC2:
+		return adev->gfx.mec2_fw_version;
+
+	case KGD_ENGINE_RLC:
+		return adev->gfx.rlc_fw_version;
 
-	BUG_ON(kgd == NULL);
+	case KGD_ENGINE_SDMA1:
+		return adev->sdma.instance[0].fw_version;
 
-	return adev->mc.real_vram_size;
+	case KGD_ENGINE_SDMA2:
+		return adev->sdma.instance[1].fw_version;
+
+	default:
+		return 0;
+	}
+
+	return 0;
 }
 
-uint64_t get_gpu_clock_counter(struct kgd_dev *kgd)
+void amdgpu_amdkfd_get_local_mem_info(struct amdgpu_device *adev,
+				      struct kfd_local_mem_info *mem_info,
+				      struct amdgpu_xcp *xcp)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+	memset(mem_info, 0, sizeof(*mem_info));
+
+	if (xcp) {
+		if (adev->gmc.real_vram_size == adev->gmc.visible_vram_size)
+			mem_info->local_mem_size_public =
+					KFD_XCP_MEMORY_SIZE(adev, xcp->id);
+		else
+			mem_info->local_mem_size_private =
+					KFD_XCP_MEMORY_SIZE(adev, xcp->id);
+	} else if (adev->apu_prefer_gtt) {
+		mem_info->local_mem_size_public = (ttm_tt_pages_limit() << PAGE_SHIFT);
+		mem_info->local_mem_size_private = 0;
+	} else {
+		mem_info->local_mem_size_public = adev->gmc.visible_vram_size;
+		mem_info->local_mem_size_private = adev->gmc.real_vram_size -
+						adev->gmc.visible_vram_size;
+	}
+	mem_info->vram_width = adev->gmc.vram_width;
+
+	pr_debug("Address base: %pap public 0x%llx private 0x%llx\n",
+			&adev->gmc.aper_base,
+			mem_info->local_mem_size_public,
+			mem_info->local_mem_size_private);
+
+	if (adev->pm.dpm_enabled) {
+		if (amdgpu_emu_mode == 1)
+			mem_info->mem_clk_max = 0;
+		else
+			mem_info->mem_clk_max = amdgpu_dpm_get_mclk(adev, false) / 100;
+	} else
+		mem_info->mem_clk_max = 100;
+}
 
+uint64_t amdgpu_amdkfd_get_gpu_clock_counter(struct amdgpu_device *adev)
+{
 	if (adev->gfx.funcs->get_gpu_clock_counter)
 		return adev->gfx.funcs->get_gpu_clock_counter(adev);
 	return 0;
 }
 
-uint32_t get_max_engine_clock_in_mhz(struct kgd_dev *kgd)
+uint32_t amdgpu_amdkfd_get_max_engine_clock_in_mhz(struct amdgpu_device *adev)
+{
+	/* the sclk is in quantas of 10kHz */
+	if (adev->pm.dpm_enabled)
+		return amdgpu_dpm_get_sclk(adev, false) / 100;
+	else
+		return 100;
+}
+
+int amdgpu_amdkfd_get_dmabuf_info(struct amdgpu_device *adev, int dma_buf_fd,
+				  struct amdgpu_device **dmabuf_adev,
+				  uint64_t *bo_size, void *metadata_buffer,
+				  size_t buffer_size, uint32_t *metadata_size,
+				  uint32_t *flags, int8_t *xcp_id)
+{
+	struct dma_buf *dma_buf;
+	struct drm_gem_object *obj;
+	struct amdgpu_bo *bo;
+	uint64_t metadata_flags;
+	int r = -EINVAL;
+
+	dma_buf = dma_buf_get(dma_buf_fd);
+	if (IS_ERR(dma_buf))
+		return PTR_ERR(dma_buf);
+
+	if (dma_buf->ops != &amdgpu_dmabuf_ops)
+		/* Can't handle non-graphics buffers */
+		goto out_put;
+
+	obj = dma_buf->priv;
+	if (obj->dev->driver != adev_to_drm(adev)->driver)
+		/* Can't handle buffers from different drivers */
+		goto out_put;
+
+	adev = drm_to_adev(obj->dev);
+	bo = gem_to_amdgpu_bo(obj);
+	if (!(bo->preferred_domains & (AMDGPU_GEM_DOMAIN_VRAM |
+				    AMDGPU_GEM_DOMAIN_GTT)))
+		/* Only VRAM and GTT BOs are supported */
+		goto out_put;
+
+	r = 0;
+	if (dmabuf_adev)
+		*dmabuf_adev = adev;
+	if (bo_size)
+		*bo_size = amdgpu_bo_size(bo);
+	if (metadata_buffer)
+		r = amdgpu_bo_get_metadata(bo, metadata_buffer, buffer_size,
+					   metadata_size, &metadata_flags);
+	if (flags) {
+		*flags = (bo->preferred_domains & AMDGPU_GEM_DOMAIN_VRAM) ?
+				KFD_IOC_ALLOC_MEM_FLAGS_VRAM
+				: KFD_IOC_ALLOC_MEM_FLAGS_GTT;
+
+		if (bo->flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED)
+			*flags |= KFD_IOC_ALLOC_MEM_FLAGS_PUBLIC;
+	}
+	if (xcp_id)
+		*xcp_id = bo->xcp_id;
+
+out_put:
+	dma_buf_put(dma_buf);
+	return r;
+}
+
+int amdgpu_amdkfd_get_pcie_bandwidth_mbytes(struct amdgpu_device *adev, bool is_min)
+{
+	int num_lanes_shift = (is_min ? ffs(adev->pm.pcie_mlw_mask) :
+							fls(adev->pm.pcie_mlw_mask)) - 1;
+	int gen_speed_shift = (is_min ? ffs(adev->pm.pcie_gen_mask &
+						CAIL_PCIE_LINK_SPEED_SUPPORT_MASK) :
+					fls(adev->pm.pcie_gen_mask &
+						CAIL_PCIE_LINK_SPEED_SUPPORT_MASK)) - 1;
+	uint32_t num_lanes_mask = 1 << num_lanes_shift;
+	uint32_t gen_speed_mask = 1 << gen_speed_shift;
+	int num_lanes_factor = 0, gen_speed_mbits_factor = 0;
+
+	switch (num_lanes_mask) {
+	case CAIL_PCIE_LINK_WIDTH_SUPPORT_X1:
+		num_lanes_factor = 1;
+		break;
+	case CAIL_PCIE_LINK_WIDTH_SUPPORT_X2:
+		num_lanes_factor = 2;
+		break;
+	case CAIL_PCIE_LINK_WIDTH_SUPPORT_X4:
+		num_lanes_factor = 4;
+		break;
+	case CAIL_PCIE_LINK_WIDTH_SUPPORT_X8:
+		num_lanes_factor = 8;
+		break;
+	case CAIL_PCIE_LINK_WIDTH_SUPPORT_X12:
+		num_lanes_factor = 12;
+		break;
+	case CAIL_PCIE_LINK_WIDTH_SUPPORT_X16:
+		num_lanes_factor = 16;
+		break;
+	case CAIL_PCIE_LINK_WIDTH_SUPPORT_X32:
+		num_lanes_factor = 32;
+		break;
+	}
+
+	switch (gen_speed_mask) {
+	case CAIL_PCIE_LINK_SPEED_SUPPORT_GEN1:
+		gen_speed_mbits_factor = 2500;
+		break;
+	case CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2:
+		gen_speed_mbits_factor = 5000;
+		break;
+	case CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3:
+		gen_speed_mbits_factor = 8000;
+		break;
+	case CAIL_PCIE_LINK_SPEED_SUPPORT_GEN4:
+		gen_speed_mbits_factor = 16000;
+		break;
+	case CAIL_PCIE_LINK_SPEED_SUPPORT_GEN5:
+		gen_speed_mbits_factor = 32000;
+		break;
+	}
+
+	return (num_lanes_factor * gen_speed_mbits_factor)/BITS_PER_BYTE;
+}
+
+int amdgpu_amdkfd_submit_ib(struct amdgpu_device *adev,
+				enum kgd_engine_type engine,
+				uint32_t vmid, uint64_t gpu_addr,
+				uint32_t *ib_cmd, uint32_t ib_len)
+{
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	struct amdgpu_ring *ring;
+	struct dma_fence *f = NULL;
+	int ret;
+
+	switch (engine) {
+	case KGD_ENGINE_MEC1:
+		ring = &adev->gfx.compute_ring[0];
+		break;
+	case KGD_ENGINE_SDMA1:
+		ring = &adev->sdma.instance[0].ring;
+		break;
+	case KGD_ENGINE_SDMA2:
+		ring = &adev->sdma.instance[1].ring;
+		break;
+	default:
+		pr_err("Invalid engine in IB submission: %d\n", engine);
+		ret = -EINVAL;
+		goto err;
+	}
+
+	ret = amdgpu_job_alloc(adev, NULL, NULL, NULL, 1, &job, 0);
+	if (ret)
+		goto err;
+
+	ib = &job->ibs[0];
+	memset(ib, 0, sizeof(struct amdgpu_ib));
+
+	ib->gpu_addr = gpu_addr;
+	ib->ptr = ib_cmd;
+	ib->length_dw = ib_len;
+	/* This works for NO_HWS. TODO: need to handle without knowing VMID */
+	job->vmid = vmid;
+	job->num_ibs = 1;
+
+	ret = amdgpu_ib_schedule(ring, 1, ib, job, &f);
+
+	if (ret) {
+		DRM_ERROR("amdgpu: failed to schedule IB.\n");
+		goto err_ib_sched;
+	}
+
+	/* Drop the initial kref_init count (see drm_sched_main as example) */
+	dma_fence_put(f);
+	ret = dma_fence_wait(f, false);
+
+err_ib_sched:
+	amdgpu_job_free(job);
+err:
+	return ret;
+}
+
+void amdgpu_amdkfd_set_compute_idle(struct amdgpu_device *adev, bool idle)
+{
+	enum amd_powergating_state state = idle ? AMD_PG_STATE_GATE : AMD_PG_STATE_UNGATE;
+	if ((IP_VERSION_MAJ(amdgpu_ip_version(adev, GC_HWIP, 0)) == 11 &&
+	    ((adev->mes.kiq_version & AMDGPU_MES_VERSION_MASK) <= 64)) ||
+		(IP_VERSION_MAJ(amdgpu_ip_version(adev, GC_HWIP, 0)) == 12)) {
+		pr_debug("GFXOFF is %s\n", idle ? "enabled" : "disabled");
+		amdgpu_gfx_off_ctrl(adev, idle);
+	} else if ((IP_VERSION_MAJ(amdgpu_ip_version(adev, GC_HWIP, 0)) == 9) &&
+		(adev->flags & AMD_IS_APU)) {
+		/* Disable GFXOFF and PG. Temporary workaround
+		 * to fix some compute applications issue on GFX9.
+		 */
+		struct amdgpu_ip_block *gfx_block = amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_GFX);
+		if (gfx_block != NULL)
+			gfx_block->version->funcs->set_powergating_state((void *)gfx_block, state);
+	}
+	amdgpu_dpm_switch_power_profile(adev,
+					PP_SMC_POWER_PROFILE_COMPUTE,
+					!idle);
+}
+
+bool amdgpu_amdkfd_is_kfd_vmid(struct amdgpu_device *adev, u32 vmid)
+{
+	if (adev->kfd.dev)
+		return vmid >= adev->vm_manager.first_kfd_vmid;
+
+	return false;
+}
+
+bool amdgpu_amdkfd_have_atomics_support(struct amdgpu_device *adev)
+{
+	return adev->have_atomics_support;
+}
+
+void amdgpu_amdkfd_debug_mem_fence(struct amdgpu_device *adev)
+{
+	amdgpu_device_flush_hdp(adev, NULL);
+}
+
+bool amdgpu_amdkfd_is_fed(struct amdgpu_device *adev)
+{
+	return amdgpu_ras_get_fed_status(adev);
+}
+
+void amdgpu_amdkfd_ras_pasid_poison_consumption_handler(struct amdgpu_device *adev,
+				enum amdgpu_ras_block block, uint16_t pasid,
+				pasid_notify pasid_fn, void *data, uint32_t reset)
+{
+	amdgpu_umc_pasid_poison_handler(adev, block, pasid, pasid_fn, data, reset);
+}
+
+void amdgpu_amdkfd_ras_poison_consumption_handler(struct amdgpu_device *adev,
+	enum amdgpu_ras_block block, uint32_t reset)
+{
+	amdgpu_umc_pasid_poison_handler(adev, block, 0, NULL, NULL, reset);
+}
+
+int amdgpu_amdkfd_send_close_event_drain_irq(struct amdgpu_device *adev,
+					uint32_t *payload)
+{
+	int ret;
+
+	/* Device or IH ring is not ready so bail. */
+	ret = amdgpu_ih_wait_on_checkpoint_process_ts(adev, &adev->irq.ih);
+	if (ret)
+		return ret;
+
+	/* Send payload to fence KFD interrupts */
+	amdgpu_amdkfd_interrupt(adev, payload);
+
+	return 0;
+}
+
+int amdgpu_amdkfd_check_and_lock_kfd(struct amdgpu_device *adev)
+{
+	return kgd2kfd_check_and_lock_kfd(adev->kfd.dev);
+}
+
+void amdgpu_amdkfd_unlock_kfd(struct amdgpu_device *adev)
+{
+	kgd2kfd_unlock_kfd(adev->kfd.dev);
+}
+
+
+u64 amdgpu_amdkfd_xcp_memory_size(struct amdgpu_device *adev, int xcp_id)
+{
+	s8 mem_id = KFD_XCP_MEM_ID(adev, xcp_id);
+	u64 tmp;
+
+	if (adev->gmc.num_mem_partitions && xcp_id >= 0 && mem_id >= 0) {
+		if (adev->gmc.is_app_apu && adev->gmc.num_mem_partitions == 1) {
+			/* In NPS1 mode, we should restrict the vram reporting
+			 * tied to the ttm_pages_limit which is 1/2 of the system
+			 * memory. For other partition modes, the HBM is uniformly
+			 * divided already per numa node reported. If user wants to
+			 * go beyond the default ttm limit and maximize the ROCm
+			 * allocations, they can go up to max ttm and sysmem limits.
+			 */
+
+			tmp = (ttm_tt_pages_limit() << PAGE_SHIFT) / num_online_nodes();
+		} else {
+			tmp = adev->gmc.mem_partitions[mem_id].size;
+		}
+		do_div(tmp, adev->xcp_mgr->num_xcp_per_mem_partition);
+		return ALIGN_DOWN(tmp, PAGE_SIZE);
+	} else if (adev->apu_prefer_gtt) {
+		return (ttm_tt_pages_limit() << PAGE_SHIFT);
+	} else {
+		return adev->gmc.real_vram_size;
+	}
+}
+
+int amdgpu_amdkfd_unmap_hiq(struct amdgpu_device *adev, u32 doorbell_off,
+			    u32 inst)
+{
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[inst];
+	struct amdgpu_ring *kiq_ring = &kiq->ring;
+	struct amdgpu_ring_funcs *ring_funcs;
+	struct amdgpu_ring *ring;
+	int r = 0;
+
+	if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
+		return -EINVAL;
+
+	if (!kiq_ring->sched.ready || amdgpu_in_reset(adev))
+		return 0;
+
+	ring_funcs = kzalloc(sizeof(*ring_funcs), GFP_KERNEL);
+	if (!ring_funcs)
+		return -ENOMEM;
+
+	ring = kzalloc(sizeof(*ring), GFP_KERNEL);
+	if (!ring) {
+		r = -ENOMEM;
+		goto free_ring_funcs;
+	}
+
+	ring_funcs->type = AMDGPU_RING_TYPE_COMPUTE;
+	ring->doorbell_index = doorbell_off;
+	ring->funcs = ring_funcs;
+
+	spin_lock(&kiq->ring_lock);
+
+	if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->unmap_queues_size)) {
+		spin_unlock(&kiq->ring_lock);
+		r = -ENOMEM;
+		goto free_ring;
+	}
+
+	kiq->pmf->kiq_unmap_queues(kiq_ring, ring, RESET_QUEUES, 0, 0);
+
+	/* Submit unmap queue packet */
+	amdgpu_ring_commit(kiq_ring);
+	/*
+	 * Ring test will do a basic scratch register change check. Just run
+	 * this to ensure that unmap queues that is submitted before got
+	 * processed successfully before returning.
+	 */
+	r = amdgpu_ring_test_helper(kiq_ring);
+
+	spin_unlock(&kiq->ring_lock);
+
+free_ring:
+	kfree(ring);
+
+free_ring_funcs:
+	kfree(ring_funcs);
+
+	return r;
+}
+
+/* Stop scheduling on KFD */
+int amdgpu_amdkfd_stop_sched(struct amdgpu_device *adev, uint32_t node_id)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+	if (!adev->kfd.init_complete)
+		return 0;
+
+	return kgd2kfd_stop_sched(adev->kfd.dev, node_id);
+}
 
-	/* The sclk is in quantas of 10kHz */
-	return adev->pm.dpm.dyn_state.max_clock_voltage_on_ac.sclk / 100;
+/* Start scheduling on KFD */
+int amdgpu_amdkfd_start_sched(struct amdgpu_device *adev, uint32_t node_id)
+{
+	if (!adev->kfd.init_complete)
+		return 0;
+
+	return kgd2kfd_start_sched(adev->kfd.dev, node_id);
+}
+
+/* check if there are KFD queues active */
+bool amdgpu_amdkfd_compute_active(struct amdgpu_device *adev, uint32_t node_id)
+{
+	if (!adev->kfd.init_complete)
+		return false;
+
+	return kgd2kfd_compute_active(adev->kfd.dev, node_id);
+}
+
+/* Config CGTT_SQ_CLK_CTRL */
+int amdgpu_amdkfd_config_sq_perfmon(struct amdgpu_device *adev, uint32_t xcp_id,
+	bool core_override_enable, bool reg_override_enable, bool perfmon_override_enable)
+{
+	int r;
+
+	if (!adev->kfd.init_complete)
+		return 0;
+
+	r = psp_config_sq_perfmon(&adev->psp, xcp_id, core_override_enable,
+					reg_override_enable, perfmon_override_enable);
+
+	return r;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
index 8d689ab7e429..9e120c934cc1 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
@@ -25,55 +25,526 @@
 #ifndef AMDGPU_AMDKFD_H_INCLUDED
 #define AMDGPU_AMDKFD_H_INCLUDED
 
+#include <linux/list.h>
 #include <linux/types.h>
-#include <linux/mmu_context.h>
+#include <linux/mm.h>
+#include <linux/kthread.h>
+#include <linux/workqueue.h>
+#include <linux/mmu_notifier.h>
+#include <linux/memremap.h>
 #include <kgd_kfd_interface.h>
+#include <drm/drm_client.h>
+#include "amdgpu_sync.h"
+#include "amdgpu_vm.h"
+#include "amdgpu_xcp.h"
+
+extern uint64_t amdgpu_amdkfd_total_mem_size;
+
+enum TLB_FLUSH_TYPE {
+	TLB_FLUSH_LEGACY = 0,
+	TLB_FLUSH_LIGHTWEIGHT,
+	TLB_FLUSH_HEAVYWEIGHT
+};
 
 struct amdgpu_device;
+struct kfd_process_device;
+struct amdgpu_reset_context;
+
+enum kfd_mem_attachment_type {
+	KFD_MEM_ATT_SHARED,	/* Share kgd_mem->bo or another attachment's */
+	KFD_MEM_ATT_USERPTR,	/* SG bo to DMA map pages from a userptr bo */
+	KFD_MEM_ATT_DMABUF,	/* DMAbuf to DMA map TTM BOs */
+	KFD_MEM_ATT_SG		/* Tag to DMA map SG BOs */
+};
+
+struct kfd_mem_attachment {
+	struct list_head list;
+	enum kfd_mem_attachment_type type;
+	bool is_mapped;
+	struct amdgpu_bo_va *bo_va;
+	struct amdgpu_device *adev;
+	uint64_t va;
+	uint64_t pte_flags;
+};
 
 struct kgd_mem {
+	struct mutex lock;
 	struct amdgpu_bo *bo;
-	uint64_t gpu_addr;
-	void *cpu_ptr;
+	struct dma_buf *dmabuf;
+	struct hmm_range *range;
+	struct list_head attachments;
+	/* protected by amdkfd_process_info.lock */
+	struct list_head validate_list;
+	uint32_t domain;
+	unsigned int mapped_to_gpu_memory;
+	uint64_t va;
+
+	uint32_t alloc_flags;
+
+	uint32_t invalid;
+	struct amdkfd_process_info *process_info;
+
+	struct amdgpu_sync sync;
+
+	uint32_t gem_handle;
+	bool aql_queue;
+	bool is_imported;
+};
+
+/* KFD Memory Eviction */
+struct amdgpu_amdkfd_fence {
+	struct dma_fence base;
+	struct mm_struct *mm;
+	spinlock_t lock;
+	char timeline_name[TASK_COMM_LEN];
+	struct svm_range_bo *svm_bo;
+};
+
+struct amdgpu_kfd_dev {
+	struct kfd_dev *dev;
+	int64_t vram_used[MAX_XCP];
+	uint64_t vram_used_aligned[MAX_XCP];
+	bool init_complete;
+	struct work_struct reset_work;
+
+	/* Client for KFD BO GEM handle allocations */
+	struct drm_client_dev client;
+
+	/* HMM page migration MEMORY_DEVICE_PRIVATE mapping
+	 * Must be last --ends in a flexible-array member.
+	 */
+	struct dev_pagemap pgmap;
+};
+
+enum kgd_engine_type {
+	KGD_ENGINE_PFP = 1,
+	KGD_ENGINE_ME,
+	KGD_ENGINE_CE,
+	KGD_ENGINE_MEC1,
+	KGD_ENGINE_MEC2,
+	KGD_ENGINE_RLC,
+	KGD_ENGINE_SDMA1,
+	KGD_ENGINE_SDMA2,
+	KGD_ENGINE_MAX
+};
+
+
+struct amdkfd_process_info {
+	/* List head of all VMs that belong to a KFD process */
+	struct list_head vm_list_head;
+	/* List head for all KFD BOs that belong to a KFD process. */
+	struct list_head kfd_bo_list;
+	/* List of userptr BOs that are valid or invalid */
+	struct list_head userptr_valid_list;
+	struct list_head userptr_inval_list;
+	/* Lock to protect kfd_bo_list */
+	struct mutex lock;
+
+	/* Number of VMs */
+	unsigned int n_vms;
+	/* Eviction Fence */
+	struct amdgpu_amdkfd_fence *eviction_fence;
+
+	/* MMU-notifier related fields */
+	struct mutex notifier_lock;
+	uint32_t evicted_bos;
+	struct delayed_work restore_userptr_work;
+	struct pid *pid;
+	bool block_mmu_notifications;
 };
 
 int amdgpu_amdkfd_init(void);
 void amdgpu_amdkfd_fini(void);
 
-void amdgpu_amdkfd_suspend(struct amdgpu_device *adev);
-int amdgpu_amdkfd_resume(struct amdgpu_device *adev);
+void amdgpu_amdkfd_suspend(struct amdgpu_device *adev, bool suspend_proc);
+int amdgpu_amdkfd_resume(struct amdgpu_device *adev, bool resume_proc);
+void amdgpu_amdkfd_suspend_process(struct amdgpu_device *adev);
+int amdgpu_amdkfd_resume_process(struct amdgpu_device *adev);
 void amdgpu_amdkfd_interrupt(struct amdgpu_device *adev,
 			const void *ih_ring_entry);
 void amdgpu_amdkfd_device_probe(struct amdgpu_device *adev);
 void amdgpu_amdkfd_device_init(struct amdgpu_device *adev);
-void amdgpu_amdkfd_device_fini(struct amdgpu_device *adev);
+void amdgpu_amdkfd_device_fini_sw(struct amdgpu_device *adev);
+int amdgpu_amdkfd_check_and_lock_kfd(struct amdgpu_device *adev);
+void amdgpu_amdkfd_unlock_kfd(struct amdgpu_device *adev);
+int amdgpu_amdkfd_submit_ib(struct amdgpu_device *adev,
+				enum kgd_engine_type engine,
+				uint32_t vmid, uint64_t gpu_addr,
+				uint32_t *ib_cmd, uint32_t ib_len);
+void amdgpu_amdkfd_set_compute_idle(struct amdgpu_device *adev, bool idle);
+bool amdgpu_amdkfd_have_atomics_support(struct amdgpu_device *adev);
+
+bool amdgpu_amdkfd_is_kfd_vmid(struct amdgpu_device *adev, u32 vmid);
+
+int amdgpu_amdkfd_pre_reset(struct amdgpu_device *adev,
+			    struct amdgpu_reset_context *reset_context);
+
+int amdgpu_amdkfd_post_reset(struct amdgpu_device *adev);
+
+void amdgpu_amdkfd_gpu_reset(struct amdgpu_device *adev);
 
-struct kfd2kgd_calls *amdgpu_amdkfd_gfx_7_get_functions(void);
-struct kfd2kgd_calls *amdgpu_amdkfd_gfx_8_0_get_functions(void);
+int amdgpu_queue_mask_bit_to_set_resource_bit(struct amdgpu_device *adev,
+					int queue_bit);
 
+struct amdgpu_amdkfd_fence *amdgpu_amdkfd_fence_create(u64 context,
+				struct mm_struct *mm,
+				struct svm_range_bo *svm_bo);
+
+int amdgpu_amdkfd_drm_client_create(struct amdgpu_device *adev);
+#if defined(CONFIG_DEBUG_FS)
+int kfd_debugfs_kfd_mem_limits(struct seq_file *m, void *data);
+#endif
+#if IS_ENABLED(CONFIG_HSA_AMD)
+bool amdkfd_fence_check_mm(struct dma_fence *f, struct mm_struct *mm);
+struct amdgpu_amdkfd_fence *to_amdgpu_amdkfd_fence(struct dma_fence *f);
+void amdgpu_amdkfd_remove_all_eviction_fences(struct amdgpu_bo *bo);
+int amdgpu_amdkfd_evict_userptr(struct mmu_interval_notifier *mni,
+				unsigned long cur_seq, struct kgd_mem *mem);
+int amdgpu_amdkfd_bo_validate_and_fence(struct amdgpu_bo *bo,
+					uint32_t domain,
+					struct dma_fence *fence);
+#else
+static inline
+bool amdkfd_fence_check_mm(struct dma_fence *f, struct mm_struct *mm)
+{
+	return false;
+}
+
+static inline
+struct amdgpu_amdkfd_fence *to_amdgpu_amdkfd_fence(struct dma_fence *f)
+{
+	return NULL;
+}
+
+static inline
+void amdgpu_amdkfd_remove_all_eviction_fences(struct amdgpu_bo *bo)
+{
+}
+
+static inline
+int amdgpu_amdkfd_evict_userptr(struct mmu_interval_notifier *mni,
+				unsigned long cur_seq, struct kgd_mem *mem)
+{
+	return 0;
+}
+static inline
+int amdgpu_amdkfd_bo_validate_and_fence(struct amdgpu_bo *bo,
+					uint32_t domain,
+					struct dma_fence *fence)
+{
+	return 0;
+}
+#endif
 /* Shared API */
-int alloc_gtt_mem(struct kgd_dev *kgd, size_t size,
-			void **mem_obj, uint64_t *gpu_addr,
-			void **cpu_ptr);
-void free_gtt_mem(struct kgd_dev *kgd, void *mem_obj);
-uint64_t get_vmem_size(struct kgd_dev *kgd);
-uint64_t get_gpu_clock_counter(struct kgd_dev *kgd);
+int amdgpu_amdkfd_alloc_gtt_mem(struct amdgpu_device *adev, size_t size,
+				void **mem_obj, uint64_t *gpu_addr,
+				void **cpu_ptr, bool mqd_gfx9);
+void amdgpu_amdkfd_free_gtt_mem(struct amdgpu_device *adev, void **mem_obj);
+int amdgpu_amdkfd_alloc_gws(struct amdgpu_device *adev, size_t size,
+				void **mem_obj);
+void amdgpu_amdkfd_free_gws(struct amdgpu_device *adev, void *mem_obj);
+int amdgpu_amdkfd_add_gws_to_process(void *info, void *gws, struct kgd_mem **mem);
+int amdgpu_amdkfd_remove_gws_from_process(void *info, void *mem);
+uint32_t amdgpu_amdkfd_get_fw_version(struct amdgpu_device *adev,
+				      enum kgd_engine_type type);
+void amdgpu_amdkfd_get_local_mem_info(struct amdgpu_device *adev,
+				      struct kfd_local_mem_info *mem_info,
+				      struct amdgpu_xcp *xcp);
+uint64_t amdgpu_amdkfd_get_gpu_clock_counter(struct amdgpu_device *adev);
+
+uint32_t amdgpu_amdkfd_get_max_engine_clock_in_mhz(struct amdgpu_device *adev);
+int amdgpu_amdkfd_get_dmabuf_info(struct amdgpu_device *adev, int dma_buf_fd,
+				  struct amdgpu_device **dmabuf_adev,
+				  uint64_t *bo_size, void *metadata_buffer,
+				  size_t buffer_size, uint32_t *metadata_size,
+				  uint32_t *flags, int8_t *xcp_id);
+int amdgpu_amdkfd_get_pcie_bandwidth_mbytes(struct amdgpu_device *adev, bool is_min);
+int amdgpu_amdkfd_send_close_event_drain_irq(struct amdgpu_device *adev,
+					uint32_t *payload);
+int amdgpu_amdkfd_unmap_hiq(struct amdgpu_device *adev, u32 doorbell_off,
+				u32 inst);
+int amdgpu_amdkfd_start_sched(struct amdgpu_device *adev, uint32_t node_id);
+int amdgpu_amdkfd_stop_sched(struct amdgpu_device *adev, uint32_t node_id);
+int amdgpu_amdkfd_config_sq_perfmon(struct amdgpu_device *adev, uint32_t xcp_id,
+	bool core_override_enable, bool reg_override_enable, bool perfmon_override_enable);
+bool amdgpu_amdkfd_compute_active(struct amdgpu_device *adev, uint32_t node_id);
 
-uint32_t get_max_engine_clock_in_mhz(struct kgd_dev *kgd);
 
+/* Read user wptr from a specified user address space with page fault
+ * disabled. The memory must be pinned and mapped to the hardware when
+ * this is called in hqd_load functions, so it should never fault in
+ * the first place. This resolves a circular lock dependency involving
+ * four locks, including the DQM lock and mmap_lock.
+ */
 #define read_user_wptr(mmptr, wptr, dst)				\
 	({								\
 		bool valid = false;					\
 		if ((mmptr) && (wptr)) {				\
+			pagefault_disable();				\
 			if ((mmptr) == current->mm) {			\
 				valid = !get_user((dst), (wptr));	\
-			} else if (current->mm == NULL) {		\
-				use_mm(mmptr);				\
+			} else if (current->flags & PF_KTHREAD) {	\
+				kthread_use_mm(mmptr);			\
 				valid = !get_user((dst), (wptr));	\
-				unuse_mm(mmptr);			\
+				kthread_unuse_mm(mmptr);		\
 			}						\
+			pagefault_enable();				\
 		}							\
 		valid;							\
 	})
 
+/* GPUVM API */
+#define drm_priv_to_vm(drm_priv)					\
+	(&((struct amdgpu_fpriv *)					\
+		((struct drm_file *)(drm_priv))->driver_priv)->vm)
+
+int amdgpu_amdkfd_gpuvm_acquire_process_vm(struct amdgpu_device *adev,
+					struct amdgpu_vm *avm,
+					void **process_info,
+					struct dma_fence **ef);
+uint64_t amdgpu_amdkfd_gpuvm_get_process_page_dir(void *drm_priv);
+size_t amdgpu_amdkfd_get_available_memory(struct amdgpu_device *adev,
+					uint8_t xcp_id);
+int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
+		struct amdgpu_device *adev, uint64_t va, uint64_t size,
+		void *drm_priv, struct kgd_mem **mem,
+		uint64_t *offset, uint32_t flags, bool criu_resume);
+int amdgpu_amdkfd_gpuvm_free_memory_of_gpu(
+		struct amdgpu_device *adev, struct kgd_mem *mem, void *drm_priv,
+		uint64_t *size);
+int amdgpu_amdkfd_gpuvm_map_memory_to_gpu(struct amdgpu_device *adev,
+					  struct kgd_mem *mem, void *drm_priv);
+int amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu(
+		struct amdgpu_device *adev, struct kgd_mem *mem, void *drm_priv);
+int amdgpu_amdkfd_gpuvm_dmaunmap_mem(struct kgd_mem *mem, void *drm_priv);
+int amdgpu_amdkfd_gpuvm_sync_memory(
+		struct amdgpu_device *adev, struct kgd_mem *mem, bool intr);
+int amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel(struct kgd_mem *mem,
+					     void **kptr, uint64_t *size);
+void amdgpu_amdkfd_gpuvm_unmap_gtt_bo_from_kernel(struct kgd_mem *mem);
+
+int amdgpu_amdkfd_map_gtt_bo_to_gart(struct amdgpu_bo *bo, struct amdgpu_bo **bo_gart);
+
+int amdgpu_amdkfd_gpuvm_restore_process_bos(void *process_info,
+					    struct dma_fence __rcu **ef);
+int amdgpu_amdkfd_gpuvm_get_vm_fault_info(struct amdgpu_device *adev,
+					      struct kfd_vm_fault_info *info);
+int amdgpu_amdkfd_gpuvm_import_dmabuf_fd(struct amdgpu_device *adev, int fd,
+					 uint64_t va, void *drm_priv,
+					 struct kgd_mem **mem, uint64_t *size,
+					 uint64_t *mmap_offset);
+int amdgpu_amdkfd_gpuvm_export_dmabuf(struct kgd_mem *mem,
+				      struct dma_buf **dmabuf);
+void amdgpu_amdkfd_debug_mem_fence(struct amdgpu_device *adev);
+int amdgpu_amdkfd_get_tile_config(struct amdgpu_device *adev,
+				struct tile_config *config);
+void amdgpu_amdkfd_ras_poison_consumption_handler(struct amdgpu_device *adev,
+			enum amdgpu_ras_block block, uint32_t reset);
+
+void amdgpu_amdkfd_ras_pasid_poison_consumption_handler(struct amdgpu_device *adev,
+			enum amdgpu_ras_block block, uint16_t pasid,
+			pasid_notify pasid_fn, void *data, uint32_t reset);
+
+bool amdgpu_amdkfd_is_fed(struct amdgpu_device *adev);
+bool amdgpu_amdkfd_bo_mapped_to_dev(void *drm_priv, struct kgd_mem *mem);
+void amdgpu_amdkfd_block_mmu_notifications(void *p);
+int amdgpu_amdkfd_criu_resume(void *p);
+int amdgpu_amdkfd_reserve_mem_limit(struct amdgpu_device *adev,
+		uint64_t size, u32 alloc_flag, int8_t xcp_id);
+void amdgpu_amdkfd_unreserve_mem_limit(struct amdgpu_device *adev,
+		uint64_t size, u32 alloc_flag, int8_t xcp_id);
+
+u64 amdgpu_amdkfd_xcp_memory_size(struct amdgpu_device *adev, int xcp_id);
+
+#define KFD_XCP_MEM_ID(adev, xcp_id) \
+		((adev)->xcp_mgr && (xcp_id) >= 0 ?\
+		(adev)->xcp_mgr->xcp[(xcp_id)].mem_id : -1)
+
+#define KFD_XCP_MEMORY_SIZE(adev, xcp_id) amdgpu_amdkfd_xcp_memory_size((adev), (xcp_id))
+
+
+#if IS_ENABLED(CONFIG_HSA_AMD)
+void amdgpu_amdkfd_gpuvm_init_mem_limits(void);
+void amdgpu_amdkfd_gpuvm_destroy_cb(struct amdgpu_device *adev,
+				struct amdgpu_vm *vm);
+
+/**
+ * @amdgpu_amdkfd_release_notify() - Notify KFD when GEM object is released
+ *
+ * Allows KFD to release its resources associated with the GEM object.
+ */
+void amdgpu_amdkfd_release_notify(struct amdgpu_bo *bo);
+void amdgpu_amdkfd_reserve_system_mem(uint64_t size);
+#else
+static inline
+void amdgpu_amdkfd_gpuvm_init_mem_limits(void)
+{
+}
+
+static inline
+void amdgpu_amdkfd_gpuvm_destroy_cb(struct amdgpu_device *adev,
+					struct amdgpu_vm *vm)
+{
+}
+
+static inline
+void amdgpu_amdkfd_release_notify(struct amdgpu_bo *bo)
+{
+}
+#endif
+
+#if IS_ENABLED(CONFIG_HSA_AMD_SVM)
+int kgd2kfd_init_zone_device(struct amdgpu_device *adev);
+#else
+static inline
+int kgd2kfd_init_zone_device(struct amdgpu_device *adev)
+{
+	return 0;
+}
+#endif
+
+/* KGD2KFD callbacks */
+int kgd2kfd_quiesce_mm(struct mm_struct *mm, uint32_t trigger);
+int kgd2kfd_resume_mm(struct mm_struct *mm);
+int kgd2kfd_schedule_evict_and_restore_process(struct mm_struct *mm,
+						struct dma_fence *fence);
+#if IS_ENABLED(CONFIG_HSA_AMD)
+int kgd2kfd_init(void);
+void kgd2kfd_exit(void);
+struct kfd_dev *kgd2kfd_probe(struct amdgpu_device *adev, bool vf);
+bool kgd2kfd_device_init(struct kfd_dev *kfd,
+			 const struct kgd2kfd_shared_resources *gpu_resources);
+void kgd2kfd_device_exit(struct kfd_dev *kfd);
+void kgd2kfd_suspend(struct kfd_dev *kfd, bool suspend_proc);
+int kgd2kfd_resume(struct kfd_dev *kfd, bool resume_proc);
+void kgd2kfd_suspend_process(struct kfd_dev *kfd);
+int kgd2kfd_resume_process(struct kfd_dev *kfd);
+int kgd2kfd_pre_reset(struct kfd_dev *kfd,
+		      struct amdgpu_reset_context *reset_context);
+int kgd2kfd_post_reset(struct kfd_dev *kfd);
+void kgd2kfd_interrupt(struct kfd_dev *kfd, const void *ih_ring_entry);
+void kgd2kfd_set_sram_ecc_flag(struct kfd_dev *kfd);
+void kgd2kfd_smi_event_throttle(struct kfd_dev *kfd, uint64_t throttle_bitmask);
+int kgd2kfd_check_and_lock_kfd(struct kfd_dev *kfd);
+void kgd2kfd_unlock_kfd(struct kfd_dev *kfd);
+int kgd2kfd_start_sched(struct kfd_dev *kfd, uint32_t node_id);
+int kgd2kfd_start_sched_all_nodes(struct kfd_dev *kfd);
+int kgd2kfd_stop_sched(struct kfd_dev *kfd, uint32_t node_id);
+int kgd2kfd_stop_sched_all_nodes(struct kfd_dev *kfd);
+bool kgd2kfd_compute_active(struct kfd_dev *kfd, uint32_t node_id);
+bool kgd2kfd_vmfault_fast_path(struct amdgpu_device *adev, struct amdgpu_iv_entry *entry,
+			       bool retry_fault);
+
+#else
+static inline int kgd2kfd_init(void)
+{
+	return -ENOENT;
+}
+
+static inline void kgd2kfd_exit(void)
+{
+}
+
+static inline
+struct kfd_dev *kgd2kfd_probe(struct amdgpu_device *adev, bool vf)
+{
+	return NULL;
+}
+
+static inline
+bool kgd2kfd_device_init(struct kfd_dev *kfd,
+				const struct kgd2kfd_shared_resources *gpu_resources)
+{
+	return false;
+}
+
+static inline void kgd2kfd_device_exit(struct kfd_dev *kfd)
+{
+}
+
+static inline void kgd2kfd_suspend(struct kfd_dev *kfd, bool suspend_proc)
+{
+}
+
+static inline int kgd2kfd_resume(struct kfd_dev *kfd, bool resume_proc)
+{
+	return 0;
+}
+
+static inline void kgd2kfd_suspend_process(struct kfd_dev *kfd)
+{
+}
+
+static inline int kgd2kfd_resume_process(struct kfd_dev *kfd)
+{
+	return 0;
+}
+
+static inline int kgd2kfd_pre_reset(struct kfd_dev *kfd,
+				    struct amdgpu_reset_context *reset_context)
+{
+	return 0;
+}
+
+static inline int kgd2kfd_post_reset(struct kfd_dev *kfd)
+{
+	return 0;
+}
+
+static inline
+void kgd2kfd_interrupt(struct kfd_dev *kfd, const void *ih_ring_entry)
+{
+}
+
+static inline
+void kgd2kfd_set_sram_ecc_flag(struct kfd_dev *kfd)
+{
+}
+
+static inline
+void kgd2kfd_smi_event_throttle(struct kfd_dev *kfd, uint64_t throttle_bitmask)
+{
+}
+
+static inline int kgd2kfd_check_and_lock_kfd(struct kfd_dev *kfd)
+{
+	return 0;
+}
+
+static inline void kgd2kfd_unlock_kfd(struct kfd_dev *kfd)
+{
+}
+
+static inline int kgd2kfd_start_sched(struct kfd_dev *kfd, uint32_t node_id)
+{
+	return 0;
+}
+
+static inline int kgd2kfd_start_sched_all_nodes(struct kfd_dev *kfd)
+{
+	return 0;
+}
+
+static inline int kgd2kfd_stop_sched(struct kfd_dev *kfd, uint32_t node_id)
+{
+	return 0;
+}
+
+static inline int kgd2kfd_stop_sched_all_nodes(struct kfd_dev *kfd)
+{
+	return 0;
+}
+
+static inline bool kgd2kfd_compute_active(struct kfd_dev *kfd, uint32_t node_id)
+{
+	return false;
+}
+
+static inline bool kgd2kfd_vmfault_fast_path(struct amdgpu_device *adev, struct amdgpu_iv_entry *entry,
+				      bool retry_fault)
+{
+	return false;
+}
+
+#endif
 #endif /* AMDGPU_AMDKFD_H_INCLUDED */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c
new file mode 100644
index 000000000000..7e9f7a280c1b
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c
@@ -0,0 +1,197 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+#include "amdgpu.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_amdkfd_arcturus.h"
+#include "amdgpu_amdkfd_gfx_v9.h"
+#include "amdgpu_amdkfd_aldebaran.h"
+#include "gc/gc_9_4_2_offset.h"
+#include "gc/gc_9_4_2_sh_mask.h"
+#include <uapi/linux/kfd_ioctl.h>
+
+/*
+ * Returns TRAP_EN, EXCP_EN and EXCP_REPLACE.
+ *
+ * restore_dbg_registers is ignored here but is a general interface requirement
+ * for devices that support GFXOFF and where the RLC save/restore list
+ * does not support hw registers for debugging i.e. the driver has to manually
+ * initialize the debug mode registers after it has disabled GFX off during the
+ * debug session.
+ */
+uint32_t kgd_aldebaran_enable_debug_trap(struct amdgpu_device *adev,
+					    bool restore_dbg_registers,
+					    uint32_t vmid)
+{
+	uint32_t data = 0;
+
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, TRAP_EN, 1);
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_EN, 0);
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_REPLACE, 0);
+
+	return data;
+}
+
+/* returns TRAP_EN, EXCP_EN and EXCP_REPLACE. */
+static uint32_t kgd_aldebaran_disable_debug_trap(struct amdgpu_device *adev,
+						bool keep_trap_enabled,
+						uint32_t vmid)
+{
+	uint32_t data = 0;
+
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, TRAP_EN, keep_trap_enabled);
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_EN, 0);
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_REPLACE, 0);
+
+	return data;
+}
+
+static int kgd_aldebaran_validate_trap_override_request(struct amdgpu_device *adev,
+							uint32_t trap_override,
+							uint32_t *trap_mask_supported)
+{
+	*trap_mask_supported &= KFD_DBG_TRAP_MASK_FP_INVALID |
+				KFD_DBG_TRAP_MASK_FP_INPUT_DENORMAL |
+				KFD_DBG_TRAP_MASK_FP_DIVIDE_BY_ZERO |
+				KFD_DBG_TRAP_MASK_FP_OVERFLOW |
+				KFD_DBG_TRAP_MASK_FP_UNDERFLOW |
+				KFD_DBG_TRAP_MASK_FP_INEXACT |
+				KFD_DBG_TRAP_MASK_INT_DIVIDE_BY_ZERO |
+				KFD_DBG_TRAP_MASK_DBG_ADDRESS_WATCH |
+				KFD_DBG_TRAP_MASK_DBG_MEMORY_VIOLATION;
+
+	if (trap_override != KFD_DBG_TRAP_OVERRIDE_OR &&
+			trap_override != KFD_DBG_TRAP_OVERRIDE_REPLACE)
+		return -EPERM;
+
+	return 0;
+}
+
+/* returns TRAP_EN, EXCP_EN and EXCP_RPLACE. */
+static uint32_t kgd_aldebaran_set_wave_launch_trap_override(struct amdgpu_device *adev,
+					uint32_t vmid,
+					uint32_t trap_override,
+					uint32_t trap_mask_bits,
+					uint32_t trap_mask_request,
+					uint32_t *trap_mask_prev,
+					uint32_t kfd_dbg_trap_cntl_prev)
+
+{
+	uint32_t data = 0;
+
+	*trap_mask_prev = REG_GET_FIELD(kfd_dbg_trap_cntl_prev, SPI_GDBG_PER_VMID_CNTL, EXCP_EN);
+	trap_mask_bits = (trap_mask_bits & trap_mask_request) |
+		(*trap_mask_prev & ~trap_mask_request);
+
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, TRAP_EN, 1);
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_EN, trap_mask_bits);
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_REPLACE, trap_override);
+
+	return data;
+}
+
+uint32_t kgd_aldebaran_set_wave_launch_mode(struct amdgpu_device *adev,
+					uint8_t wave_launch_mode,
+					uint32_t vmid)
+{
+	uint32_t data = 0;
+
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, LAUNCH_MODE, wave_launch_mode);
+
+	return data;
+}
+
+#define TCP_WATCH_STRIDE (regTCP_WATCH1_ADDR_H - regTCP_WATCH0_ADDR_H)
+static uint32_t kgd_gfx_aldebaran_set_address_watch(
+					struct amdgpu_device *adev,
+					uint64_t watch_address,
+					uint32_t watch_address_mask,
+					uint32_t watch_id,
+					uint32_t watch_mode,
+					uint32_t debug_vmid,
+					uint32_t inst)
+{
+	uint32_t watch_address_high;
+	uint32_t watch_address_low;
+	uint32_t watch_address_cntl;
+
+	watch_address_cntl = 0;
+	watch_address_low = lower_32_bits(watch_address);
+	watch_address_high = upper_32_bits(watch_address) & 0xffff;
+
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			MODE,
+			watch_mode);
+
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			MASK,
+			watch_address_mask >> 6);
+
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			VALID,
+			1);
+
+	WREG32_RLC((SOC15_REG_OFFSET(GC, 0, regTCP_WATCH0_ADDR_H) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_high);
+
+	WREG32_RLC((SOC15_REG_OFFSET(GC, 0, regTCP_WATCH0_ADDR_L) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_low);
+
+	return watch_address_cntl;
+}
+
+const struct kfd2kgd_calls aldebaran_kfd2kgd = {
+	.program_sh_mem_settings = kgd_gfx_v9_program_sh_mem_settings,
+	.set_pasid_vmid_mapping = kgd_gfx_v9_set_pasid_vmid_mapping,
+	.init_interrupts = kgd_gfx_v9_init_interrupts,
+	.hqd_load = kgd_gfx_v9_hqd_load,
+	.hiq_mqd_load = kgd_gfx_v9_hiq_mqd_load,
+	.hqd_sdma_load = kgd_arcturus_hqd_sdma_load,
+	.hqd_dump = kgd_gfx_v9_hqd_dump,
+	.hqd_sdma_dump = kgd_arcturus_hqd_sdma_dump,
+	.hqd_is_occupied = kgd_gfx_v9_hqd_is_occupied,
+	.hqd_sdma_is_occupied = kgd_arcturus_hqd_sdma_is_occupied,
+	.hqd_destroy = kgd_gfx_v9_hqd_destroy,
+	.hqd_sdma_destroy = kgd_arcturus_hqd_sdma_destroy,
+	.wave_control_execute = kgd_gfx_v9_wave_control_execute,
+	.get_atc_vmid_pasid_mapping_info =
+				kgd_gfx_v9_get_atc_vmid_pasid_mapping_info,
+	.set_vm_context_page_table_base = kgd_gfx_v9_set_vm_context_page_table_base,
+	.get_cu_occupancy = kgd_gfx_v9_get_cu_occupancy,
+	.enable_debug_trap = kgd_aldebaran_enable_debug_trap,
+	.disable_debug_trap = kgd_aldebaran_disable_debug_trap,
+	.validate_trap_override_request = kgd_aldebaran_validate_trap_override_request,
+	.set_wave_launch_trap_override = kgd_aldebaran_set_wave_launch_trap_override,
+	.set_wave_launch_mode = kgd_aldebaran_set_wave_launch_mode,
+	.set_address_watch = kgd_gfx_aldebaran_set_address_watch,
+	.clear_address_watch = kgd_gfx_v9_clear_address_watch,
+	.get_iq_wait_times = kgd_gfx_v9_get_iq_wait_times,
+	.build_dequeue_wait_counts_packet_info = kgd_gfx_v9_build_dequeue_wait_counts_packet_info,
+	.program_trap_handler_settings = kgd_gfx_v9_program_trap_handler_settings,
+	.hqd_get_pq_addr = kgd_gfx_v9_hqd_get_pq_addr,
+	.hqd_reset = kgd_gfx_v9_hqd_reset,
+	.hqd_sdma_get_doorbell = kgd_gfx_v9_hqd_sdma_get_doorbell
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.h
new file mode 100644
index 000000000000..a7bdaf8d82dd
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.h
@@ -0,0 +1,27 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+uint32_t kgd_aldebaran_enable_debug_trap(struct amdgpu_device *adev,
+					bool restore_dbg_registers,
+					uint32_t vmid);
+uint32_t kgd_aldebaran_set_wave_launch_mode(struct amdgpu_device *adev,
+					uint8_t wave_launch_mode,
+					uint32_t vmid);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c
new file mode 100644
index 000000000000..1105a09e55dc
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c
@@ -0,0 +1,424 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+#include <linux/module.h>
+#include <linux/uaccess.h>
+#include <linux/firmware.h>
+#include "amdgpu.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_amdkfd_arcturus.h"
+#include "amdgpu_reset.h"
+#include "sdma0/sdma0_4_2_2_offset.h"
+#include "sdma0/sdma0_4_2_2_sh_mask.h"
+#include "sdma1/sdma1_4_2_2_offset.h"
+#include "sdma1/sdma1_4_2_2_sh_mask.h"
+#include "sdma2/sdma2_4_2_2_offset.h"
+#include "sdma2/sdma2_4_2_2_sh_mask.h"
+#include "sdma3/sdma3_4_2_2_offset.h"
+#include "sdma3/sdma3_4_2_2_sh_mask.h"
+#include "sdma4/sdma4_4_2_2_offset.h"
+#include "sdma4/sdma4_4_2_2_sh_mask.h"
+#include "sdma5/sdma5_4_2_2_offset.h"
+#include "sdma5/sdma5_4_2_2_sh_mask.h"
+#include "sdma6/sdma6_4_2_2_offset.h"
+#include "sdma6/sdma6_4_2_2_sh_mask.h"
+#include "sdma7/sdma7_4_2_2_offset.h"
+#include "sdma7/sdma7_4_2_2_sh_mask.h"
+#include "v9_structs.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "amdgpu_amdkfd_gfx_v9.h"
+#include "gfxhub_v1_0.h"
+#include "mmhub_v9_4.h"
+#include "gc/gc_9_0_offset.h"
+#include "gc/gc_9_0_sh_mask.h"
+
+#define HQD_N_REGS 56
+#define DUMP_REG(addr) do {				\
+		if (WARN_ON_ONCE(i >= HQD_N_REGS))	\
+			break;				\
+		(*dump)[i][0] = (addr) << 2;		\
+		(*dump)[i++][1] = RREG32(addr);		\
+	} while (0)
+
+static inline struct v9_sdma_mqd *get_sdma_mqd(void *mqd)
+{
+	return (struct v9_sdma_mqd *)mqd;
+}
+
+static uint32_t get_sdma_rlc_reg_offset(struct amdgpu_device *adev,
+				unsigned int engine_id,
+				unsigned int queue_id)
+{
+	uint32_t sdma_engine_reg_base = 0;
+	uint32_t sdma_rlc_reg_offset;
+
+	switch (engine_id) {
+	default:
+		dev_warn(adev->dev,
+			 "Invalid sdma engine id (%d), using engine id 0\n",
+			 engine_id);
+		fallthrough;
+	case 0:
+		sdma_engine_reg_base = SOC15_REG_OFFSET(SDMA0, 0,
+				mmSDMA0_RLC0_RB_CNTL) - mmSDMA0_RLC0_RB_CNTL;
+		break;
+	case 1:
+		sdma_engine_reg_base = SOC15_REG_OFFSET(SDMA1, 0,
+				mmSDMA1_RLC0_RB_CNTL) - mmSDMA1_RLC0_RB_CNTL;
+		break;
+	case 2:
+		sdma_engine_reg_base = SOC15_REG_OFFSET(SDMA2, 0,
+				mmSDMA2_RLC0_RB_CNTL) - mmSDMA2_RLC0_RB_CNTL;
+		break;
+	case 3:
+		sdma_engine_reg_base = SOC15_REG_OFFSET(SDMA3, 0,
+				mmSDMA3_RLC0_RB_CNTL) - mmSDMA3_RLC0_RB_CNTL;
+		break;
+	case 4:
+		sdma_engine_reg_base = SOC15_REG_OFFSET(SDMA4, 0,
+				mmSDMA4_RLC0_RB_CNTL) - mmSDMA4_RLC0_RB_CNTL;
+		break;
+	case 5:
+		sdma_engine_reg_base = SOC15_REG_OFFSET(SDMA5, 0,
+				mmSDMA5_RLC0_RB_CNTL) - mmSDMA5_RLC0_RB_CNTL;
+		break;
+	case 6:
+		sdma_engine_reg_base = SOC15_REG_OFFSET(SDMA6, 0,
+				mmSDMA6_RLC0_RB_CNTL) - mmSDMA6_RLC0_RB_CNTL;
+		break;
+	case 7:
+		sdma_engine_reg_base = SOC15_REG_OFFSET(SDMA7, 0,
+				mmSDMA7_RLC0_RB_CNTL) - mmSDMA7_RLC0_RB_CNTL;
+		break;
+	}
+
+	sdma_rlc_reg_offset = sdma_engine_reg_base
+		+ queue_id * (mmSDMA0_RLC1_RB_CNTL - mmSDMA0_RLC0_RB_CNTL);
+
+	pr_debug("RLC register offset for SDMA%d RLC%d: 0x%x\n", engine_id,
+			queue_id, sdma_rlc_reg_offset);
+
+	return sdma_rlc_reg_offset;
+}
+
+int kgd_arcturus_hqd_sdma_load(struct amdgpu_device *adev, void *mqd,
+			     uint32_t __user *wptr, struct mm_struct *mm)
+{
+	struct v9_sdma_mqd *m;
+	uint32_t sdma_rlc_reg_offset;
+	unsigned long end_jiffies;
+	uint32_t data;
+	uint64_t data64;
+	uint64_t __user *wptr64 = (uint64_t __user *)wptr;
+
+	m = get_sdma_mqd(mqd);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev, m->sdma_engine_id,
+					    m->sdma_queue_id);
+
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL,
+		m->sdmax_rlcx_rb_cntl & (~SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK));
+
+	end_jiffies = msecs_to_jiffies(2000) + jiffies;
+	while (true) {
+		data = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_CONTEXT_STATUS);
+		if (data & SDMA0_RLC0_CONTEXT_STATUS__IDLE_MASK)
+			break;
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("SDMA RLC not idle in %s\n", __func__);
+			return -ETIME;
+		}
+		usleep_range(500, 1000);
+	}
+
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_DOORBELL_OFFSET,
+	       m->sdmax_rlcx_doorbell_offset);
+
+	data = REG_SET_FIELD(m->sdmax_rlcx_doorbell, SDMA0_RLC0_DOORBELL,
+			     ENABLE, 1);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_DOORBELL, data);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR,
+				m->sdmax_rlcx_rb_rptr);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR_HI,
+				m->sdmax_rlcx_rb_rptr_hi);
+
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_MINOR_PTR_UPDATE, 1);
+	if (read_user_wptr(mm, wptr64, data64)) {
+		WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_WPTR,
+		       lower_32_bits(data64));
+		WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_WPTR_HI,
+		       upper_32_bits(data64));
+	} else {
+		WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_WPTR,
+		       m->sdmax_rlcx_rb_rptr);
+		WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_WPTR_HI,
+		       m->sdmax_rlcx_rb_rptr_hi);
+	}
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_MINOR_PTR_UPDATE, 0);
+
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_BASE, m->sdmax_rlcx_rb_base);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_BASE_HI,
+			m->sdmax_rlcx_rb_base_hi);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR_ADDR_LO,
+			m->sdmax_rlcx_rb_rptr_addr_lo);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR_ADDR_HI,
+			m->sdmax_rlcx_rb_rptr_addr_hi);
+
+	data = REG_SET_FIELD(m->sdmax_rlcx_rb_cntl, SDMA0_RLC0_RB_CNTL,
+			     RB_ENABLE, 1);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL, data);
+
+	return 0;
+}
+
+int kgd_arcturus_hqd_sdma_dump(struct amdgpu_device *adev,
+			     uint32_t engine_id, uint32_t queue_id,
+			     uint32_t (**dump)[2], uint32_t *n_regs)
+{
+	uint32_t sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev,
+			engine_id, queue_id);
+	uint32_t i = 0, reg;
+#undef HQD_N_REGS
+#define HQD_N_REGS (19+6+7+10)
+
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
+	if (*dump == NULL)
+		return -ENOMEM;
+
+	for (reg = mmSDMA0_RLC0_RB_CNTL; reg <= mmSDMA0_RLC0_DOORBELL; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+	for (reg = mmSDMA0_RLC0_STATUS; reg <= mmSDMA0_RLC0_CSA_ADDR_HI; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+	for (reg = mmSDMA0_RLC0_IB_SUB_REMAIN;
+	     reg <= mmSDMA0_RLC0_MINOR_PTR_UPDATE; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+	for (reg = mmSDMA0_RLC0_MIDCMD_DATA0;
+	     reg <= mmSDMA0_RLC0_MIDCMD_CNTL; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+
+	WARN_ON_ONCE(i != HQD_N_REGS);
+	*n_regs = i;
+
+	return 0;
+}
+
+bool kgd_arcturus_hqd_sdma_is_occupied(struct amdgpu_device *adev,
+				void *mqd)
+{
+	struct v9_sdma_mqd *m;
+	uint32_t sdma_rlc_reg_offset;
+	uint32_t sdma_rlc_rb_cntl;
+
+	m = get_sdma_mqd(mqd);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev, m->sdma_engine_id,
+					    m->sdma_queue_id);
+
+	sdma_rlc_rb_cntl = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL);
+
+	if (sdma_rlc_rb_cntl & SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK)
+		return true;
+
+	return false;
+}
+
+int kgd_arcturus_hqd_sdma_destroy(struct amdgpu_device *adev, void *mqd,
+				unsigned int utimeout)
+{
+	struct v9_sdma_mqd *m;
+	uint32_t sdma_rlc_reg_offset;
+	uint32_t temp;
+	unsigned long end_jiffies = (utimeout * HZ / 1000) + jiffies;
+
+	m = get_sdma_mqd(mqd);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev, m->sdma_engine_id,
+					    m->sdma_queue_id);
+
+	temp = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL);
+	temp = temp & ~SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK;
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL, temp);
+
+	while (true) {
+		temp = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_CONTEXT_STATUS);
+		if (temp & SDMA0_RLC0_CONTEXT_STATUS__IDLE_MASK)
+			break;
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("SDMA RLC not idle in %s\n", __func__);
+			return -ETIME;
+		}
+		usleep_range(500, 1000);
+	}
+
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_DOORBELL, 0);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL,
+		RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL) |
+		SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK);
+
+	m->sdmax_rlcx_rb_rptr = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR);
+	m->sdmax_rlcx_rb_rptr_hi =
+		RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR_HI);
+
+	return 0;
+}
+
+/*
+ * Helper used to suspend/resume gfx pipe for image post process work to set
+ * barrier behaviour.
+ */
+static int suspend_resume_compute_scheduler(struct amdgpu_device *adev, bool suspend)
+{
+	int i, r = 0;
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		struct amdgpu_ring *ring = &adev->gfx.compute_ring[i];
+
+		if (!amdgpu_ring_sched_ready(ring))
+			continue;
+
+		/* stop secheduler and drain ring. */
+		if (suspend) {
+			drm_sched_stop(&ring->sched, NULL);
+			r = amdgpu_fence_wait_empty(ring);
+			if (r)
+				goto out;
+		} else {
+			drm_sched_start(&ring->sched, 0);
+		}
+	}
+
+out:
+	/* return on resume or failure to drain rings. */
+	if (!suspend || r)
+		return r;
+
+	return amdgpu_device_ip_wait_for_idle(adev, AMD_IP_BLOCK_TYPE_GFX);
+}
+
+static void set_barrier_auto_waitcnt(struct amdgpu_device *adev, bool enable_waitcnt)
+{
+	uint32_t data;
+
+	WRITE_ONCE(adev->barrier_has_auto_waitcnt, enable_waitcnt);
+
+	if (!down_read_trylock(&adev->reset_domain->sem))
+		return;
+
+	amdgpu_amdkfd_suspend(adev, true);
+
+	if (suspend_resume_compute_scheduler(adev, true))
+		goto out;
+
+	data = RREG32(SOC15_REG_OFFSET(GC, 0, mmSQ_CONFIG));
+	data = REG_SET_FIELD(data, SQ_CONFIG, DISABLE_BARRIER_WAITCNT,
+						!enable_waitcnt);
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSQ_CONFIG), data);
+
+out:
+	suspend_resume_compute_scheduler(adev, false);
+
+	amdgpu_amdkfd_resume(adev, true);
+
+	up_read(&adev->reset_domain->sem);
+}
+
+/*
+ * restore_dbg_registers is ignored here but is a general interface requirement
+ * for devices that support GFXOFF and where the RLC save/restore list
+ * does not support hw registers for debugging i.e. the driver has to manually
+ * initialize the debug mode registers after it has disabled GFX off during the
+ * debug session.
+ */
+static uint32_t kgd_arcturus_enable_debug_trap(struct amdgpu_device *adev,
+				bool restore_dbg_registers,
+				uint32_t vmid)
+{
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	kgd_gfx_v9_set_wave_launch_stall(adev, vmid, true);
+
+	set_barrier_auto_waitcnt(adev, true);
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_TRAP_MASK), 0);
+
+	kgd_gfx_v9_set_wave_launch_stall(adev, vmid, false);
+
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+/*
+ * keep_trap_enabled is ignored here but is a general interface requirement
+ * for devices that support multi-process debugging where the performance
+ * overhead from trap temporary setup needs to be bypassed when the debug
+ * session has ended.
+ */
+static uint32_t kgd_arcturus_disable_debug_trap(struct amdgpu_device *adev,
+					bool keep_trap_enabled,
+					uint32_t vmid)
+{
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	kgd_gfx_v9_set_wave_launch_stall(adev, vmid, true);
+
+	set_barrier_auto_waitcnt(adev, false);
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_TRAP_MASK), 0);
+
+	kgd_gfx_v9_set_wave_launch_stall(adev, vmid, false);
+
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+const struct kfd2kgd_calls arcturus_kfd2kgd = {
+	.program_sh_mem_settings = kgd_gfx_v9_program_sh_mem_settings,
+	.set_pasid_vmid_mapping = kgd_gfx_v9_set_pasid_vmid_mapping,
+	.init_interrupts = kgd_gfx_v9_init_interrupts,
+	.hqd_load = kgd_gfx_v9_hqd_load,
+	.hiq_mqd_load = kgd_gfx_v9_hiq_mqd_load,
+	.hqd_sdma_load = kgd_arcturus_hqd_sdma_load,
+	.hqd_dump = kgd_gfx_v9_hqd_dump,
+	.hqd_sdma_dump = kgd_arcturus_hqd_sdma_dump,
+	.hqd_is_occupied = kgd_gfx_v9_hqd_is_occupied,
+	.hqd_sdma_is_occupied = kgd_arcturus_hqd_sdma_is_occupied,
+	.hqd_destroy = kgd_gfx_v9_hqd_destroy,
+	.hqd_sdma_destroy = kgd_arcturus_hqd_sdma_destroy,
+	.wave_control_execute = kgd_gfx_v9_wave_control_execute,
+	.get_atc_vmid_pasid_mapping_info =
+				kgd_gfx_v9_get_atc_vmid_pasid_mapping_info,
+	.set_vm_context_page_table_base =
+				kgd_gfx_v9_set_vm_context_page_table_base,
+	.enable_debug_trap = kgd_arcturus_enable_debug_trap,
+	.disable_debug_trap = kgd_arcturus_disable_debug_trap,
+	.validate_trap_override_request = kgd_gfx_v9_validate_trap_override_request,
+	.set_wave_launch_trap_override = kgd_gfx_v9_set_wave_launch_trap_override,
+	.set_wave_launch_mode = kgd_gfx_v9_set_wave_launch_mode,
+	.set_address_watch = kgd_gfx_v9_set_address_watch,
+	.clear_address_watch = kgd_gfx_v9_clear_address_watch,
+	.get_iq_wait_times = kgd_gfx_v9_get_iq_wait_times,
+	.build_dequeue_wait_counts_packet_info = kgd_gfx_v9_build_dequeue_wait_counts_packet_info,
+	.get_cu_occupancy = kgd_gfx_v9_get_cu_occupancy,
+	.program_trap_handler_settings = kgd_gfx_v9_program_trap_handler_settings,
+	.hqd_get_pq_addr = kgd_gfx_v9_hqd_get_pq_addr,
+	.hqd_reset = kgd_gfx_v9_hqd_reset,
+	.hqd_sdma_get_doorbell = kgd_gfx_v9_hqd_sdma_get_doorbell
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.h
new file mode 100644
index 000000000000..756c1a5679c0
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+int kgd_arcturus_hqd_sdma_load(struct amdgpu_device *adev, void *mqd,
+			     uint32_t __user *wptr, struct mm_struct *mm);
+int kgd_arcturus_hqd_sdma_dump(struct amdgpu_device *adev,
+			     uint32_t engine_id, uint32_t queue_id,
+			     uint32_t (**dump)[2], uint32_t *n_regs);
+bool kgd_arcturus_hqd_sdma_is_occupied(struct amdgpu_device *adev,
+				void *mqd);
+int kgd_arcturus_hqd_sdma_destroy(struct amdgpu_device *adev, void *mqd,
+				unsigned int utimeout);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_fence.c
new file mode 100644
index 000000000000..1ef758ac5076
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_fence.c
@@ -0,0 +1,188 @@
+/*
+ * Copyright 2016-2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include <linux/dma-fence.h>
+#include <linux/spinlock.h>
+#include <linux/atomic.h>
+#include <linux/stacktrace.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <linux/sched/mm.h>
+#include "amdgpu_amdkfd.h"
+#include "kfd_svm.h"
+
+static const struct dma_fence_ops amdkfd_fence_ops;
+static atomic_t fence_seq = ATOMIC_INIT(0);
+
+/* Eviction Fence
+ * Fence helper functions to deal with KFD memory eviction.
+ * Big Idea - Since KFD submissions are done by user queues, a BO cannot be
+ *  evicted unless all the user queues for that process are evicted.
+ *
+ * All the BOs in a process share an eviction fence. When process X wants
+ * to map VRAM memory but TTM can't find enough space, TTM will attempt to
+ * evict BOs from its LRU list. TTM checks if the BO is valuable to evict
+ * by calling ttm_device_funcs->eviction_valuable().
+ *
+ * ttm_device_funcs->eviction_valuable() - will return false if the BO belongs
+ *  to process X. Otherwise, it will return true to indicate BO can be
+ *  evicted by TTM.
+ *
+ * If ttm_device_funcs->eviction_valuable returns true, then TTM will continue
+ * the evcition process for that BO by calling ttm_bo_evict --> amdgpu_bo_move
+ * --> amdgpu_copy_buffer(). This sets up job in GPU scheduler.
+ *
+ * GPU Scheduler (amd_sched_main) - sets up a cb (fence_add_callback) to
+ *  nofity when the BO is free to move. fence_add_callback --> enable_signaling
+ *  --> amdgpu_amdkfd_fence.enable_signaling
+ *
+ * amdgpu_amdkfd_fence.enable_signaling - Start a work item that will quiesce
+ * user queues and signal fence. The work item will also start another delayed
+ * work item to restore BOs
+ */
+
+struct amdgpu_amdkfd_fence *amdgpu_amdkfd_fence_create(u64 context,
+				struct mm_struct *mm,
+				struct svm_range_bo *svm_bo)
+{
+	struct amdgpu_amdkfd_fence *fence;
+
+	fence = kzalloc(sizeof(*fence), GFP_KERNEL);
+	if (fence == NULL)
+		return NULL;
+
+	/* This reference gets released in amdkfd_fence_release */
+	mmgrab(mm);
+	fence->mm = mm;
+	get_task_comm(fence->timeline_name, current);
+	spin_lock_init(&fence->lock);
+	fence->svm_bo = svm_bo;
+	dma_fence_init(&fence->base, &amdkfd_fence_ops, &fence->lock,
+		   context, atomic_inc_return(&fence_seq));
+
+	return fence;
+}
+
+struct amdgpu_amdkfd_fence *to_amdgpu_amdkfd_fence(struct dma_fence *f)
+{
+	struct amdgpu_amdkfd_fence *fence;
+
+	if (!f)
+		return NULL;
+
+	fence = container_of(f, struct amdgpu_amdkfd_fence, base);
+	if (f->ops == &amdkfd_fence_ops)
+		return fence;
+
+	return NULL;
+}
+
+static const char *amdkfd_fence_get_driver_name(struct dma_fence *f)
+{
+	return "amdgpu_amdkfd_fence";
+}
+
+static const char *amdkfd_fence_get_timeline_name(struct dma_fence *f)
+{
+	struct amdgpu_amdkfd_fence *fence = to_amdgpu_amdkfd_fence(f);
+
+	return fence->timeline_name;
+}
+
+/**
+ * amdkfd_fence_enable_signaling - This gets called when TTM wants to evict
+ *  a KFD BO and schedules a job to move the BO.
+ *  If fence is already signaled return true.
+ *  If fence is not signaled schedule a evict KFD process work item.
+ *
+ *  @f: dma_fence
+ */
+static bool amdkfd_fence_enable_signaling(struct dma_fence *f)
+{
+	struct amdgpu_amdkfd_fence *fence = to_amdgpu_amdkfd_fence(f);
+
+	if (!fence)
+		return false;
+
+	if (dma_fence_is_signaled(f))
+		return true;
+
+	if (!fence->svm_bo) {
+		if (!kgd2kfd_schedule_evict_and_restore_process(fence->mm, f))
+			return true;
+	} else {
+		if (!svm_range_schedule_evict_svm_bo(fence))
+			return true;
+	}
+	return false;
+}
+
+/**
+ * amdkfd_fence_release - callback that fence can be freed
+ *
+ * @f: dma_fence
+ *
+ * This function is called when the reference count becomes zero.
+ * Drops the mm_struct reference and RCU schedules freeing up the fence.
+ */
+static void amdkfd_fence_release(struct dma_fence *f)
+{
+	struct amdgpu_amdkfd_fence *fence = to_amdgpu_amdkfd_fence(f);
+
+	/* Unconditionally signal the fence. The process is getting
+	 * terminated.
+	 */
+	if (WARN_ON(!fence))
+		return; /* Not an amdgpu_amdkfd_fence */
+
+	mmdrop(fence->mm);
+	kfree_rcu(f, rcu);
+}
+
+/**
+ * amdkfd_fence_check_mm - Check whether to prevent eviction of @f by @mm
+ *
+ * @f: [IN] fence
+ * @mm: [IN] mm that needs to be verified
+ *
+ * Check if @mm is same as that of the fence @f, if same return TRUE else
+ * return FALSE.
+ * For svm bo, which support vram overcommitment, always return FALSE.
+ */
+bool amdkfd_fence_check_mm(struct dma_fence *f, struct mm_struct *mm)
+{
+	struct amdgpu_amdkfd_fence *fence = to_amdgpu_amdkfd_fence(f);
+
+	if (!fence)
+		return false;
+	else if (fence->mm == mm  && !fence->svm_bo)
+		return true;
+
+	return false;
+}
+
+static const struct dma_fence_ops amdkfd_fence_ops = {
+	.get_driver_name = amdkfd_fence_get_driver_name,
+	.get_timeline_name = amdkfd_fence_get_timeline_name,
+	.enable_signaling = amdkfd_fence_enable_signaling,
+	.release = amdkfd_fence_release,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gc_9_4_3.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gc_9_4_3.c
new file mode 100644
index 000000000000..89a45a9218f3
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gc_9_4_3.c
@@ -0,0 +1,559 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+#include "amdgpu.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_amdkfd_gfx_v9.h"
+#include "amdgpu_amdkfd_aldebaran.h"
+#include "gc/gc_9_4_3_offset.h"
+#include "gc/gc_9_4_3_sh_mask.h"
+#include "athub/athub_1_8_0_offset.h"
+#include "athub/athub_1_8_0_sh_mask.h"
+#include "oss/osssys_4_4_2_offset.h"
+#include "oss/osssys_4_4_2_sh_mask.h"
+#include "v9_structs.h"
+#include "soc15.h"
+#include "sdma/sdma_4_4_2_offset.h"
+#include "sdma/sdma_4_4_2_sh_mask.h"
+#include <uapi/linux/kfd_ioctl.h>
+
+static inline struct v9_sdma_mqd *get_sdma_mqd(void *mqd)
+{
+	return (struct v9_sdma_mqd *)mqd;
+}
+
+static uint32_t get_sdma_rlc_reg_offset(struct amdgpu_device *adev,
+					unsigned int engine_id,
+					unsigned int queue_id)
+{
+	uint32_t sdma_engine_reg_base =
+		SOC15_REG_OFFSET(SDMA0, GET_INST(SDMA0, engine_id),
+				 regSDMA_RLC0_RB_CNTL) -
+		regSDMA_RLC0_RB_CNTL;
+	uint32_t retval = sdma_engine_reg_base +
+		  queue_id * (regSDMA_RLC1_RB_CNTL - regSDMA_RLC0_RB_CNTL);
+
+	pr_debug("RLC register offset for SDMA%d RLC%d: 0x%x\n", engine_id,
+							queue_id, retval);
+	return retval;
+}
+
+static int kgd_gfx_v9_4_3_hqd_sdma_load(struct amdgpu_device *adev, void *mqd,
+				 uint32_t __user *wptr, struct mm_struct *mm)
+{
+	struct v9_sdma_mqd *m;
+	uint32_t sdma_rlc_reg_offset;
+	unsigned long end_jiffies;
+	uint32_t data;
+	uint64_t data64;
+	uint64_t __user *wptr64 = (uint64_t __user *)wptr;
+
+	m = get_sdma_mqd(mqd);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev, m->sdma_engine_id,
+							m->sdma_queue_id);
+
+	WREG32(sdma_rlc_reg_offset + regSDMA_RLC0_RB_CNTL,
+		m->sdmax_rlcx_rb_cntl & (~SDMA_RLC0_RB_CNTL__RB_ENABLE_MASK));
+
+	end_jiffies = msecs_to_jiffies(2000) + jiffies;
+	while (true) {
+		data = RREG32(sdma_rlc_reg_offset + regSDMA_RLC0_CONTEXT_STATUS);
+		if (data & SDMA_RLC0_CONTEXT_STATUS__IDLE_MASK)
+			break;
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("SDMA RLC not idle in %s\n", __func__);
+			return -ETIME;
+		}
+		usleep_range(500, 1000);
+	}
+
+	WREG32(sdma_rlc_reg_offset + regSDMA_RLC0_DOORBELL_OFFSET,
+		m->sdmax_rlcx_doorbell_offset);
+
+	data = REG_SET_FIELD(m->sdmax_rlcx_doorbell, SDMA_RLC0_DOORBELL,
+				ENABLE, 1);
+	WREG32(sdma_rlc_reg_offset + regSDMA_RLC0_DOORBELL, data);
+	WREG32(sdma_rlc_reg_offset + regSDMA_RLC0_RB_RPTR,
+					m->sdmax_rlcx_rb_rptr);
+	WREG32(sdma_rlc_reg_offset + regSDMA_RLC0_RB_RPTR_HI,
+					m->sdmax_rlcx_rb_rptr_hi);
+
+	WREG32(sdma_rlc_reg_offset + regSDMA_RLC0_MINOR_PTR_UPDATE, 1);
+	if (read_user_wptr(mm, wptr64, data64)) {
+		WREG32(sdma_rlc_reg_offset + regSDMA_RLC0_RB_WPTR,
+			lower_32_bits(data64));
+		WREG32(sdma_rlc_reg_offset + regSDMA_RLC0_RB_WPTR_HI,
+			upper_32_bits(data64));
+	} else {
+		WREG32(sdma_rlc_reg_offset + regSDMA_RLC0_RB_WPTR,
+			m->sdmax_rlcx_rb_rptr);
+		WREG32(sdma_rlc_reg_offset + regSDMA_RLC0_RB_WPTR_HI,
+			m->sdmax_rlcx_rb_rptr_hi);
+	}
+	WREG32(sdma_rlc_reg_offset + regSDMA_RLC0_MINOR_PTR_UPDATE, 0);
+
+	WREG32(sdma_rlc_reg_offset + regSDMA_RLC0_RB_BASE, m->sdmax_rlcx_rb_base);
+	WREG32(sdma_rlc_reg_offset + regSDMA_RLC0_RB_BASE_HI,
+			m->sdmax_rlcx_rb_base_hi);
+	WREG32(sdma_rlc_reg_offset + regSDMA_RLC0_RB_RPTR_ADDR_LO,
+			m->sdmax_rlcx_rb_rptr_addr_lo);
+	WREG32(sdma_rlc_reg_offset + regSDMA_RLC0_RB_RPTR_ADDR_HI,
+			m->sdmax_rlcx_rb_rptr_addr_hi);
+
+	data = REG_SET_FIELD(m->sdmax_rlcx_rb_cntl, SDMA_RLC0_RB_CNTL,
+				RB_ENABLE, 1);
+	WREG32(sdma_rlc_reg_offset + regSDMA_RLC0_RB_CNTL, data);
+
+	return 0;
+}
+
+static int kgd_gfx_v9_4_3_hqd_sdma_dump(struct amdgpu_device *adev,
+				 uint32_t engine_id, uint32_t queue_id,
+				 uint32_t (**dump)[2], uint32_t *n_regs)
+{
+	uint32_t sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev,
+							engine_id, queue_id);
+	uint32_t i = 0, reg;
+#undef HQD_N_REGS
+#define HQD_N_REGS (19+6+7+12)
+#define DUMP_REG(addr) do {				\
+		if (WARN_ON_ONCE(i >= HQD_N_REGS))      \
+			break;				\
+		(*dump)[i][0] = (addr) << 2;            \
+		(*dump)[i++][1] = RREG32(addr);         \
+	} while (0)
+
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
+	if (*dump == NULL)
+		return -ENOMEM;
+
+	for (reg = regSDMA_RLC0_RB_CNTL; reg <= regSDMA_RLC0_DOORBELL; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+	for (reg = regSDMA_RLC0_STATUS; reg <= regSDMA_RLC0_CSA_ADDR_HI; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+	for (reg = regSDMA_RLC0_IB_SUB_REMAIN;
+	     reg <= regSDMA_RLC0_MINOR_PTR_UPDATE; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+	for (reg = regSDMA_RLC0_MIDCMD_DATA0;
+	     reg <= regSDMA_RLC0_MIDCMD_CNTL; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+
+	WARN_ON_ONCE(i != HQD_N_REGS);
+	*n_regs = i;
+
+	return 0;
+}
+
+static bool kgd_gfx_v9_4_3_hqd_sdma_is_occupied(struct amdgpu_device *adev, void *mqd)
+{
+	struct v9_sdma_mqd *m;
+	uint32_t sdma_rlc_reg_offset;
+	uint32_t sdma_rlc_rb_cntl;
+
+	m = get_sdma_mqd(mqd);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev, m->sdma_engine_id,
+							m->sdma_queue_id);
+
+	sdma_rlc_rb_cntl = RREG32(sdma_rlc_reg_offset + regSDMA_RLC0_RB_CNTL);
+
+	if (sdma_rlc_rb_cntl & SDMA_RLC0_RB_CNTL__RB_ENABLE_MASK)
+		return true;
+
+	return false;
+}
+
+static int kgd_gfx_v9_4_3_hqd_sdma_destroy(struct amdgpu_device *adev, void *mqd,
+				    unsigned int utimeout)
+{
+	struct v9_sdma_mqd *m;
+	uint32_t sdma_rlc_reg_offset;
+	uint32_t temp;
+	unsigned long end_jiffies = (utimeout * HZ / 1000) + jiffies;
+
+	m = get_sdma_mqd(mqd);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev, m->sdma_engine_id,
+							m->sdma_queue_id);
+
+	temp = RREG32(sdma_rlc_reg_offset + regSDMA_RLC0_RB_CNTL);
+	temp = temp & ~SDMA_RLC0_RB_CNTL__RB_ENABLE_MASK;
+	WREG32(sdma_rlc_reg_offset + regSDMA_RLC0_RB_CNTL, temp);
+
+	while (true) {
+		temp = RREG32(sdma_rlc_reg_offset + regSDMA_RLC0_CONTEXT_STATUS);
+		if (temp & SDMA_RLC0_CONTEXT_STATUS__IDLE_MASK)
+			break;
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("SDMA RLC not idle in %s\n", __func__);
+			return -ETIME;
+		}
+		usleep_range(500, 1000);
+	}
+
+	WREG32(sdma_rlc_reg_offset + regSDMA_RLC0_DOORBELL, 0);
+	WREG32(sdma_rlc_reg_offset + regSDMA_RLC0_RB_CNTL,
+		RREG32(sdma_rlc_reg_offset + regSDMA_RLC0_RB_CNTL) |
+		SDMA_RLC0_RB_CNTL__RB_ENABLE_MASK);
+
+	m->sdmax_rlcx_rb_rptr =
+			RREG32(sdma_rlc_reg_offset + regSDMA_RLC0_RB_RPTR);
+	m->sdmax_rlcx_rb_rptr_hi =
+			RREG32(sdma_rlc_reg_offset + regSDMA_RLC0_RB_RPTR_HI);
+
+	return 0;
+}
+
+static int kgd_gfx_v9_4_3_set_pasid_vmid_mapping(struct amdgpu_device *adev,
+			u32 pasid, unsigned int vmid, uint32_t xcc_inst)
+{
+	unsigned long timeout;
+	unsigned int reg;
+	unsigned int phy_inst = GET_INST(GC, xcc_inst);
+	/* Every two XCCs share one AID */
+	unsigned int aid = phy_inst / 2;
+
+	/*
+	 * We have to assume that there is no outstanding mapping.
+	 * The ATC_VMID_PASID_MAPPING_UPDATE_STATUS bit could be 0 because
+	 * a mapping is in progress or because a mapping finished
+	 * and the SW cleared it.
+	 * So the protocol is to always wait & clear.
+	 */
+	uint32_t pasid_mapping = (pasid == 0) ? 0 : (uint32_t)pasid |
+			ATC_VMID0_PASID_MAPPING__VALID_MASK;
+
+	WREG32(SOC15_REG_OFFSET(ATHUB, 0,
+		regATC_VMID0_PASID_MAPPING) + vmid, pasid_mapping);
+
+	timeout = jiffies + msecs_to_jiffies(10);
+	while (!(RREG32(SOC15_REG_OFFSET(ATHUB, 0,
+			regATC_VMID_PASID_MAPPING_UPDATE_STATUS)) &
+			(1U << vmid))) {
+		if (time_after(jiffies, timeout)) {
+			pr_err("Fail to program VMID-PASID mapping\n");
+			return -ETIME;
+		}
+		cpu_relax();
+	}
+
+	WREG32(SOC15_REG_OFFSET(ATHUB, 0,
+		regATC_VMID_PASID_MAPPING_UPDATE_STATUS),
+		1U << vmid);
+
+	reg = RREG32(SOC15_REG_OFFSET(OSSSYS, 0, regIH_VMID_LUT_INDEX));
+	/* Every 4 numbers is a cycle. 1st is AID, 2nd and 3rd are XCDs,
+	 * and the 4th is reserved. Therefore "aid * 4 + (xcc_inst % 2) + 1"
+	 * programs _LUT for XCC and "aid * 4" for AID where the XCC connects
+	 * to.
+	 */
+	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, regIH_VMID_LUT_INDEX),
+		aid * 4 + (phy_inst % 2) + 1);
+	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, regIH_VMID_0_LUT) + vmid,
+		pasid_mapping);
+	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, regIH_VMID_LUT_INDEX),
+		aid * 4);
+	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, regIH_VMID_0_LUT_MM) + vmid,
+		pasid_mapping);
+	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, regIH_VMID_LUT_INDEX), reg);
+
+	return 0;
+}
+
+static inline struct v9_mqd *get_mqd(void *mqd)
+{
+	return (struct v9_mqd *)mqd;
+}
+
+static int kgd_gfx_v9_4_3_hqd_load(struct amdgpu_device *adev, void *mqd,
+			uint32_t pipe_id, uint32_t queue_id,
+			uint32_t __user *wptr, uint32_t wptr_shift,
+			uint32_t wptr_mask, struct mm_struct *mm, uint32_t inst)
+{
+	struct v9_mqd *m;
+	uint32_t *mqd_hqd;
+	uint32_t reg, hqd_base, hqd_end, data;
+
+	m = get_mqd(mqd);
+
+	kgd_gfx_v9_acquire_queue(adev, pipe_id, queue_id, inst);
+
+	/* HQD registers extend to CP_HQD_AQL_DISPATCH_ID_HI */
+	mqd_hqd = &m->cp_mqd_base_addr_lo;
+	hqd_base = SOC15_REG_OFFSET(GC, GET_INST(GC, inst), regCP_MQD_BASE_ADDR);
+	hqd_end = SOC15_REG_OFFSET(GC, GET_INST(GC, inst), regCP_HQD_AQL_DISPATCH_ID_HI);
+
+	for (reg = hqd_base; reg <= hqd_end; reg++)
+		WREG32_XCC(reg, mqd_hqd[reg - hqd_base], inst);
+
+
+	/* Activate doorbell logic before triggering WPTR poll. */
+	data = REG_SET_FIELD(m->cp_hqd_pq_doorbell_control,
+			     CP_HQD_PQ_DOORBELL_CONTROL, DOORBELL_EN, 1);
+	WREG32_SOC15_RLC(GC, GET_INST(GC, inst), regCP_HQD_PQ_DOORBELL_CONTROL, data);
+
+	if (wptr) {
+		/* Don't read wptr with get_user because the user
+		 * context may not be accessible (if this function
+		 * runs in a work queue). Instead trigger a one-shot
+		 * polling read from memory in the CP. This assumes
+		 * that wptr is GPU-accessible in the queue's VMID via
+		 * ATC or SVM. WPTR==RPTR before starting the poll so
+		 * the CP starts fetching new commands from the right
+		 * place.
+		 *
+		 * Guessing a 64-bit WPTR from a 32-bit RPTR is a bit
+		 * tricky. Assume that the queue didn't overflow. The
+		 * number of valid bits in the 32-bit RPTR depends on
+		 * the queue size. The remaining bits are taken from
+		 * the saved 64-bit WPTR. If the WPTR wrapped, add the
+		 * queue size.
+		 */
+		uint32_t queue_size =
+			2 << REG_GET_FIELD(m->cp_hqd_pq_control,
+					   CP_HQD_PQ_CONTROL, QUEUE_SIZE);
+		uint64_t guessed_wptr = m->cp_hqd_pq_rptr & (queue_size - 1);
+
+		if ((m->cp_hqd_pq_wptr_lo & (queue_size - 1)) < guessed_wptr)
+			guessed_wptr += queue_size;
+		guessed_wptr += m->cp_hqd_pq_wptr_lo & ~(queue_size - 1);
+		guessed_wptr += (uint64_t)m->cp_hqd_pq_wptr_hi << 32;
+
+		WREG32_SOC15_RLC(GC, GET_INST(GC, inst), regCP_HQD_PQ_WPTR_LO,
+			lower_32_bits(guessed_wptr));
+		WREG32_SOC15_RLC(GC, GET_INST(GC, inst), regCP_HQD_PQ_WPTR_HI,
+			upper_32_bits(guessed_wptr));
+		WREG32_SOC15_RLC(GC, GET_INST(GC, inst), regCP_HQD_PQ_WPTR_POLL_ADDR,
+			lower_32_bits((uintptr_t)wptr));
+		WREG32_SOC15_RLC(GC, GET_INST(GC, inst), regCP_HQD_PQ_WPTR_POLL_ADDR_HI,
+			upper_32_bits((uintptr_t)wptr));
+		WREG32_SOC15_RLC(GC, GET_INST(GC, inst), regCP_PQ_WPTR_POLL_CNTL1,
+			(uint32_t)kgd_gfx_v9_get_queue_mask(adev, pipe_id, queue_id));
+	}
+
+	/* Start the EOP fetcher */
+	WREG32_SOC15_RLC(GC, GET_INST(GC, inst), regCP_HQD_EOP_RPTR,
+	       REG_SET_FIELD(m->cp_hqd_eop_rptr, CP_HQD_EOP_RPTR, INIT_FETCHER, 1));
+
+	data = REG_SET_FIELD(m->cp_hqd_active, CP_HQD_ACTIVE, ACTIVE, 1);
+	WREG32_SOC15_RLC(GC, GET_INST(GC, inst), regCP_HQD_ACTIVE, data);
+
+	kgd_gfx_v9_release_queue(adev, inst);
+
+	return 0;
+}
+
+/* returns TRAP_EN, EXCP_EN and EXCP_REPLACE. */
+static uint32_t kgd_gfx_v9_4_3_disable_debug_trap(struct amdgpu_device *adev,
+						bool keep_trap_enabled,
+						uint32_t vmid)
+{
+	uint32_t data = 0;
+
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, TRAP_EN, 1);
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_EN, 0);
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_REPLACE, 0);
+
+	return data;
+}
+
+static int kgd_gfx_v9_4_3_validate_trap_override_request(
+				struct amdgpu_device *adev,
+				uint32_t trap_override,
+				uint32_t *trap_mask_supported)
+{
+	*trap_mask_supported &= KFD_DBG_TRAP_MASK_FP_INVALID |
+				KFD_DBG_TRAP_MASK_FP_INPUT_DENORMAL |
+				KFD_DBG_TRAP_MASK_FP_DIVIDE_BY_ZERO |
+				KFD_DBG_TRAP_MASK_FP_OVERFLOW |
+				KFD_DBG_TRAP_MASK_FP_UNDERFLOW |
+				KFD_DBG_TRAP_MASK_FP_INEXACT |
+				KFD_DBG_TRAP_MASK_INT_DIVIDE_BY_ZERO |
+				KFD_DBG_TRAP_MASK_DBG_ADDRESS_WATCH |
+				KFD_DBG_TRAP_MASK_DBG_MEMORY_VIOLATION |
+				KFD_DBG_TRAP_MASK_TRAP_ON_WAVE_START |
+				KFD_DBG_TRAP_MASK_TRAP_ON_WAVE_END;
+
+	if (trap_override != KFD_DBG_TRAP_OVERRIDE_OR &&
+			trap_override != KFD_DBG_TRAP_OVERRIDE_REPLACE)
+		return -EPERM;
+
+	return 0;
+}
+
+static uint32_t trap_mask_map_sw_to_hw(uint32_t mask)
+{
+	uint32_t trap_on_start = (mask & KFD_DBG_TRAP_MASK_TRAP_ON_WAVE_START) ? 1 : 0;
+	uint32_t trap_on_end = (mask & KFD_DBG_TRAP_MASK_TRAP_ON_WAVE_END) ? 1 : 0;
+	uint32_t excp_en = mask & (KFD_DBG_TRAP_MASK_FP_INVALID |
+				KFD_DBG_TRAP_MASK_FP_INPUT_DENORMAL |
+				KFD_DBG_TRAP_MASK_FP_DIVIDE_BY_ZERO |
+				KFD_DBG_TRAP_MASK_FP_OVERFLOW |
+				KFD_DBG_TRAP_MASK_FP_UNDERFLOW |
+				KFD_DBG_TRAP_MASK_FP_INEXACT |
+				KFD_DBG_TRAP_MASK_INT_DIVIDE_BY_ZERO |
+				KFD_DBG_TRAP_MASK_DBG_ADDRESS_WATCH |
+				KFD_DBG_TRAP_MASK_DBG_MEMORY_VIOLATION);
+	uint32_t ret;
+
+	ret = REG_SET_FIELD(0, SPI_GDBG_PER_VMID_CNTL, EXCP_EN, excp_en);
+	ret = REG_SET_FIELD(ret, SPI_GDBG_PER_VMID_CNTL, TRAP_ON_START, trap_on_start);
+	ret = REG_SET_FIELD(ret, SPI_GDBG_PER_VMID_CNTL, TRAP_ON_END, trap_on_end);
+
+	return ret;
+}
+
+static uint32_t trap_mask_map_hw_to_sw(uint32_t mask)
+{
+	uint32_t ret = REG_GET_FIELD(mask, SPI_GDBG_PER_VMID_CNTL, EXCP_EN);
+
+	if (REG_GET_FIELD(mask, SPI_GDBG_PER_VMID_CNTL, TRAP_ON_START))
+		ret |= KFD_DBG_TRAP_MASK_TRAP_ON_WAVE_START;
+
+	if (REG_GET_FIELD(mask, SPI_GDBG_PER_VMID_CNTL, TRAP_ON_END))
+		ret |= KFD_DBG_TRAP_MASK_TRAP_ON_WAVE_END;
+
+	return ret;
+}
+
+/* returns TRAP_EN, EXCP_EN and EXCP_REPLACE. */
+static uint32_t kgd_gfx_v9_4_3_set_wave_launch_trap_override(
+				struct amdgpu_device *adev,
+				uint32_t vmid,
+				uint32_t trap_override,
+				uint32_t trap_mask_bits,
+				uint32_t trap_mask_request,
+				uint32_t *trap_mask_prev,
+				uint32_t kfd_dbg_trap_cntl_prev)
+
+{
+	uint32_t data = 0;
+
+	*trap_mask_prev = trap_mask_map_hw_to_sw(kfd_dbg_trap_cntl_prev);
+
+	data = (trap_mask_bits & trap_mask_request) |
+	       (*trap_mask_prev & ~trap_mask_request);
+	data = trap_mask_map_sw_to_hw(data);
+
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, TRAP_EN, 1);
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_REPLACE, trap_override);
+
+	return data;
+}
+
+#define TCP_WATCH_STRIDE (regTCP_WATCH1_ADDR_H - regTCP_WATCH0_ADDR_H)
+static uint32_t kgd_gfx_v9_4_3_set_address_watch(
+				struct amdgpu_device *adev,
+				uint64_t watch_address,
+				uint32_t watch_address_mask,
+				uint32_t watch_id,
+				uint32_t watch_mode,
+				uint32_t debug_vmid,
+				uint32_t inst)
+{
+	uint32_t watch_address_high;
+	uint32_t watch_address_low;
+	uint32_t watch_address_cntl;
+
+	watch_address_cntl = 0;
+	watch_address_low = lower_32_bits(watch_address);
+	watch_address_high = upper_32_bits(watch_address) & 0xffff;
+
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			MODE,
+			watch_mode);
+
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			MASK,
+			watch_address_mask >> 7);
+
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			VALID,
+			1);
+
+	WREG32_XCC((SOC15_REG_OFFSET(GC, GET_INST(GC, inst),
+			regTCP_WATCH0_ADDR_H) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_high, inst);
+
+	WREG32_XCC((SOC15_REG_OFFSET(GC, GET_INST(GC, inst),
+			regTCP_WATCH0_ADDR_L) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_low, inst);
+
+	return watch_address_cntl;
+}
+
+static uint32_t kgd_gfx_v9_4_3_clear_address_watch(struct amdgpu_device *adev,
+				uint32_t watch_id)
+{
+	return 0;
+}
+
+static uint32_t kgd_gfx_v9_4_3_hqd_sdma_get_doorbell(struct amdgpu_device *adev,
+						     int engine, int queue)
+{
+	uint32_t reg_offset = get_sdma_rlc_reg_offset(adev, engine, queue);
+	uint32_t status = RREG32(regSDMA_RLC0_CONTEXT_STATUS + reg_offset);
+	uint32_t doorbell_off = RREG32(regSDMA_RLC0_DOORBELL_OFFSET + reg_offset);
+	bool is_active = !!REG_GET_FIELD(status, SDMA_RLC0_CONTEXT_STATUS, SELECTED);
+
+	return is_active ? doorbell_off >> 2 : 0;
+}
+
+const struct kfd2kgd_calls gc_9_4_3_kfd2kgd = {
+	.program_sh_mem_settings = kgd_gfx_v9_program_sh_mem_settings,
+	.set_pasid_vmid_mapping = kgd_gfx_v9_4_3_set_pasid_vmid_mapping,
+	.init_interrupts = kgd_gfx_v9_init_interrupts,
+	.hqd_load = kgd_gfx_v9_4_3_hqd_load,
+	.hiq_mqd_load = kgd_gfx_v9_hiq_mqd_load,
+	.hqd_sdma_load = kgd_gfx_v9_4_3_hqd_sdma_load,
+	.hqd_dump = kgd_gfx_v9_hqd_dump,
+	.hqd_sdma_dump = kgd_gfx_v9_4_3_hqd_sdma_dump,
+	.hqd_is_occupied = kgd_gfx_v9_hqd_is_occupied,
+	.hqd_sdma_is_occupied = kgd_gfx_v9_4_3_hqd_sdma_is_occupied,
+	.hqd_destroy = kgd_gfx_v9_hqd_destroy,
+	.hqd_sdma_destroy = kgd_gfx_v9_4_3_hqd_sdma_destroy,
+	.wave_control_execute = kgd_gfx_v9_wave_control_execute,
+	.get_atc_vmid_pasid_mapping_info =
+				kgd_gfx_v9_get_atc_vmid_pasid_mapping_info,
+	.set_vm_context_page_table_base =
+				kgd_gfx_v9_set_vm_context_page_table_base,
+	.get_cu_occupancy = kgd_gfx_v9_get_cu_occupancy,
+	.program_trap_handler_settings =
+				kgd_gfx_v9_program_trap_handler_settings,
+	.build_dequeue_wait_counts_packet_info =
+				kgd_gfx_v9_build_dequeue_wait_counts_packet_info,
+	.get_iq_wait_times = kgd_gfx_v9_get_iq_wait_times,
+	.enable_debug_trap = kgd_aldebaran_enable_debug_trap,
+	.disable_debug_trap = kgd_gfx_v9_4_3_disable_debug_trap,
+	.validate_trap_override_request =
+			kgd_gfx_v9_4_3_validate_trap_override_request,
+	.set_wave_launch_trap_override =
+			kgd_gfx_v9_4_3_set_wave_launch_trap_override,
+	.set_wave_launch_mode = kgd_aldebaran_set_wave_launch_mode,
+	.set_address_watch = kgd_gfx_v9_4_3_set_address_watch,
+	.clear_address_watch = kgd_gfx_v9_4_3_clear_address_watch,
+	.hqd_get_pq_addr = kgd_gfx_v9_hqd_get_pq_addr,
+	.hqd_reset = kgd_gfx_v9_hqd_reset,
+	.hqd_sdma_get_doorbell = kgd_gfx_v9_4_3_hqd_sdma_get_doorbell
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c
new file mode 100644
index 000000000000..0239114fb6c4
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c
@@ -0,0 +1,1123 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+#include "amdgpu.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_amdkfd_gfx_v10.h"
+#include "gc/gc_10_1_0_offset.h"
+#include "gc/gc_10_1_0_sh_mask.h"
+#include "athub/athub_2_0_0_offset.h"
+#include "athub/athub_2_0_0_sh_mask.h"
+#include "oss/osssys_5_0_0_offset.h"
+#include "oss/osssys_5_0_0_sh_mask.h"
+#include "soc15_common.h"
+#include "v10_structs.h"
+#include "nv.h"
+#include "nvd.h"
+#include <uapi/linux/kfd_ioctl.h>
+
+enum hqd_dequeue_request_type {
+	NO_ACTION = 0,
+	DRAIN_PIPE,
+	RESET_WAVES,
+	SAVE_WAVES
+};
+
+static void lock_srbm(struct amdgpu_device *adev, uint32_t mec, uint32_t pipe,
+			uint32_t queue, uint32_t vmid)
+{
+	mutex_lock(&adev->srbm_mutex);
+	nv_grbm_select(adev, mec, pipe, queue, vmid);
+}
+
+static void unlock_srbm(struct amdgpu_device *adev)
+{
+	nv_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static void acquire_queue(struct amdgpu_device *adev, uint32_t pipe_id,
+				uint32_t queue_id)
+{
+	uint32_t mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	uint32_t pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+	lock_srbm(adev, mec, pipe, queue_id, 0);
+}
+
+static uint64_t get_queue_mask(struct amdgpu_device *adev,
+			       uint32_t pipe_id, uint32_t queue_id)
+{
+	unsigned int bit = pipe_id * adev->gfx.mec.num_queue_per_pipe +
+			queue_id;
+
+	return 1ull << bit;
+}
+
+static void release_queue(struct amdgpu_device *adev)
+{
+	unlock_srbm(adev);
+}
+
+static void kgd_program_sh_mem_settings(struct amdgpu_device *adev, uint32_t vmid,
+					uint32_t sh_mem_config,
+					uint32_t sh_mem_ape1_base,
+					uint32_t sh_mem_ape1_limit,
+					uint32_t sh_mem_bases, uint32_t inst)
+{
+	lock_srbm(adev, 0, 0, 0, vmid);
+
+	WREG32_SOC15(GC, 0, mmSH_MEM_CONFIG, sh_mem_config);
+	WREG32_SOC15(GC, 0, mmSH_MEM_BASES, sh_mem_bases);
+	/* APE1 no longer exists on GFX9 */
+
+	unlock_srbm(adev);
+}
+
+static int kgd_set_pasid_vmid_mapping(struct amdgpu_device *adev, u32 pasid,
+					unsigned int vmid, uint32_t inst)
+{
+	/*
+	 * We have to assume that there is no outstanding mapping.
+	 * The ATC_VMID_PASID_MAPPING_UPDATE_STATUS bit could be 0 because
+	 * a mapping is in progress or because a mapping finished
+	 * and the SW cleared it.
+	 * So the protocol is to always wait & clear.
+	 */
+	uint32_t pasid_mapping = (pasid == 0) ? 0 : (uint32_t)pasid |
+			ATC_VMID0_PASID_MAPPING__VALID_MASK;
+
+	pr_debug("pasid 0x%x vmid %d, reg value %x\n", pasid, vmid, pasid_mapping);
+
+	pr_debug("ATHUB, reg %x\n", SOC15_REG_OFFSET(ATHUB, 0, mmATC_VMID0_PASID_MAPPING) + vmid);
+	WREG32(SOC15_REG_OFFSET(ATHUB, 0, mmATC_VMID0_PASID_MAPPING) + vmid,
+	       pasid_mapping);
+
+#if 0
+	/* TODO: uncomment this code when the hardware support is ready. */
+	while (!(RREG32(SOC15_REG_OFFSET(
+				ATHUB, 0,
+				mmATC_VMID_PASID_MAPPING_UPDATE_STATUS)) &
+		 (1U << vmid)))
+		cpu_relax();
+
+	pr_debug("ATHUB mapping update finished\n");
+	WREG32(SOC15_REG_OFFSET(ATHUB, 0,
+				mmATC_VMID_PASID_MAPPING_UPDATE_STATUS),
+	       1U << vmid);
+#endif
+
+	/* Mapping vmid to pasid also for IH block */
+	pr_debug("update mapping for IH block and mmhub");
+	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_VMID_0_LUT) + vmid,
+	       pasid_mapping);
+
+	return 0;
+}
+
+/* TODO - RING0 form of field is obsolete, seems to date back to SI
+ * but still works
+ */
+
+static int kgd_init_interrupts(struct amdgpu_device *adev, uint32_t pipe_id,
+				uint32_t inst)
+{
+	uint32_t mec;
+	uint32_t pipe;
+
+	mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+	lock_srbm(adev, mec, pipe, 0, 0);
+
+	WREG32_SOC15(GC, 0, mmCPC_INT_CNTL,
+		CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK |
+		CP_INT_CNTL_RING0__OPCODE_ERROR_INT_ENABLE_MASK);
+
+	unlock_srbm(adev);
+
+	return 0;
+}
+
+static uint32_t get_sdma_rlc_reg_offset(struct amdgpu_device *adev,
+				unsigned int engine_id,
+				unsigned int queue_id)
+{
+	uint32_t sdma_engine_reg_base[2] = {
+		SOC15_REG_OFFSET(SDMA0, 0,
+				 mmSDMA0_RLC0_RB_CNTL) - mmSDMA0_RLC0_RB_CNTL,
+		/* On gfx10, mmSDMA1_xxx registers are defined NOT based
+		 * on SDMA1 base address (dw 0x1860) but based on SDMA0
+		 * base address (dw 0x1260). Therefore use mmSDMA0_RLC0_RB_CNTL
+		 * instead of mmSDMA1_RLC0_RB_CNTL for the base address calc
+		 * below
+		 */
+		SOC15_REG_OFFSET(SDMA1, 0,
+				 mmSDMA1_RLC0_RB_CNTL) - mmSDMA0_RLC0_RB_CNTL
+	};
+
+	uint32_t retval = sdma_engine_reg_base[engine_id]
+		+ queue_id * (mmSDMA0_RLC1_RB_CNTL - mmSDMA0_RLC0_RB_CNTL);
+
+	pr_debug("RLC register offset for SDMA%d RLC%d: 0x%x\n", engine_id,
+			queue_id, retval);
+
+	return retval;
+}
+
+#if 0
+static uint32_t get_watch_base_addr(struct amdgpu_device *adev)
+{
+	uint32_t retval = SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_ADDR_H) -
+			mmTCP_WATCH0_ADDR_H;
+
+	pr_debug("kfd: reg watch base address: 0x%x\n", retval);
+
+	return retval;
+}
+#endif
+
+static inline struct v10_compute_mqd *get_mqd(void *mqd)
+{
+	return (struct v10_compute_mqd *)mqd;
+}
+
+static inline struct v10_sdma_mqd *get_sdma_mqd(void *mqd)
+{
+	return (struct v10_sdma_mqd *)mqd;
+}
+
+static int kgd_hqd_load(struct amdgpu_device *adev, void *mqd,
+			uint32_t pipe_id, uint32_t queue_id,
+			uint32_t __user *wptr, uint32_t wptr_shift,
+			uint32_t wptr_mask, struct mm_struct *mm, uint32_t inst)
+{
+	struct v10_compute_mqd *m;
+	uint32_t *mqd_hqd;
+	uint32_t reg, hqd_base, data;
+
+	m = get_mqd(mqd);
+
+	pr_debug("Load hqd of pipe %d queue %d\n", pipe_id, queue_id);
+	acquire_queue(adev, pipe_id, queue_id);
+
+	/* HQD registers extend from CP_MQD_BASE_ADDR to CP_HQD_EOP_WPTR_MEM. */
+	mqd_hqd = &m->cp_mqd_base_addr_lo;
+	hqd_base = SOC15_REG_OFFSET(GC, 0, mmCP_MQD_BASE_ADDR);
+
+	for (reg = hqd_base;
+	     reg <= SOC15_REG_OFFSET(GC, 0, mmCP_HQD_PQ_WPTR_HI); reg++)
+		WREG32_SOC15_IP(GC, reg, mqd_hqd[reg - hqd_base]);
+
+
+	/* Activate doorbell logic before triggering WPTR poll. */
+	data = REG_SET_FIELD(m->cp_hqd_pq_doorbell_control,
+			     CP_HQD_PQ_DOORBELL_CONTROL, DOORBELL_EN, 1);
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL, data);
+
+	if (wptr) {
+		/* Don't read wptr with get_user because the user
+		 * context may not be accessible (if this function
+		 * runs in a work queue). Instead trigger a one-shot
+		 * polling read from memory in the CP. This assumes
+		 * that wptr is GPU-accessible in the queue's VMID via
+		 * ATC or SVM. WPTR==RPTR before starting the poll so
+		 * the CP starts fetching new commands from the right
+		 * place.
+		 *
+		 * Guessing a 64-bit WPTR from a 32-bit RPTR is a bit
+		 * tricky. Assume that the queue didn't overflow. The
+		 * number of valid bits in the 32-bit RPTR depends on
+		 * the queue size. The remaining bits are taken from
+		 * the saved 64-bit WPTR. If the WPTR wrapped, add the
+		 * queue size.
+		 */
+		uint32_t queue_size =
+			2 << REG_GET_FIELD(m->cp_hqd_pq_control,
+					   CP_HQD_PQ_CONTROL, QUEUE_SIZE);
+		uint64_t guessed_wptr = m->cp_hqd_pq_rptr & (queue_size - 1);
+
+		if ((m->cp_hqd_pq_wptr_lo & (queue_size - 1)) < guessed_wptr)
+			guessed_wptr += queue_size;
+		guessed_wptr += m->cp_hqd_pq_wptr_lo & ~(queue_size - 1);
+		guessed_wptr += (uint64_t)m->cp_hqd_pq_wptr_hi << 32;
+
+		WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_LO,
+		       lower_32_bits(guessed_wptr));
+		WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_HI,
+		       upper_32_bits(guessed_wptr));
+		WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_POLL_ADDR,
+		       lower_32_bits((uint64_t)wptr));
+		WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_POLL_ADDR_HI,
+		       upper_32_bits((uint64_t)wptr));
+		pr_debug("%s setting CP_PQ_WPTR_POLL_CNTL1 to %x\n", __func__,
+			 (uint32_t)get_queue_mask(adev, pipe_id, queue_id));
+		WREG32_SOC15(GC, 0, mmCP_PQ_WPTR_POLL_CNTL1,
+		       (uint32_t)get_queue_mask(adev, pipe_id, queue_id));
+	}
+
+	/* Start the EOP fetcher */
+	WREG32_SOC15(GC, 0, mmCP_HQD_EOP_RPTR,
+	       REG_SET_FIELD(m->cp_hqd_eop_rptr,
+			     CP_HQD_EOP_RPTR, INIT_FETCHER, 1));
+
+	data = REG_SET_FIELD(m->cp_hqd_active, CP_HQD_ACTIVE, ACTIVE, 1);
+	WREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE, data);
+
+	release_queue(adev);
+
+	return 0;
+}
+
+static int kgd_hiq_mqd_load(struct amdgpu_device *adev, void *mqd,
+			    uint32_t pipe_id, uint32_t queue_id,
+			    uint32_t doorbell_off, uint32_t inst)
+{
+	struct amdgpu_ring *kiq_ring = &adev->gfx.kiq[0].ring;
+	struct v10_compute_mqd *m;
+	uint32_t mec, pipe;
+	int r;
+
+	m = get_mqd(mqd);
+
+	acquire_queue(adev, pipe_id, queue_id);
+
+	mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+	pr_debug("kfd: set HIQ, mec:%d, pipe:%d, queue:%d.\n",
+		 mec, pipe, queue_id);
+
+	spin_lock(&adev->gfx.kiq[0].ring_lock);
+	r = amdgpu_ring_alloc(kiq_ring, 7);
+	if (r) {
+		pr_err("Failed to alloc KIQ (%d).\n", r);
+		goto out_unlock;
+	}
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_MAP_QUEUES, 5));
+	amdgpu_ring_write(kiq_ring,
+			  PACKET3_MAP_QUEUES_QUEUE_SEL(0) | /* Queue_Sel */
+			  PACKET3_MAP_QUEUES_VMID(m->cp_hqd_vmid) | /* VMID */
+			  PACKET3_MAP_QUEUES_QUEUE(queue_id) |
+			  PACKET3_MAP_QUEUES_PIPE(pipe) |
+			  PACKET3_MAP_QUEUES_ME((mec - 1)) |
+			  PACKET3_MAP_QUEUES_QUEUE_TYPE(0) | /*queue_type: normal compute queue */
+			  PACKET3_MAP_QUEUES_ALLOC_FORMAT(0) | /* alloc format: all_on_one_pipe */
+			  PACKET3_MAP_QUEUES_ENGINE_SEL(1) | /* engine_sel: hiq */
+			  PACKET3_MAP_QUEUES_NUM_QUEUES(1)); /* num_queues: must be 1 */
+	amdgpu_ring_write(kiq_ring,
+			  PACKET3_MAP_QUEUES_DOORBELL_OFFSET(doorbell_off));
+	amdgpu_ring_write(kiq_ring, m->cp_mqd_base_addr_lo);
+	amdgpu_ring_write(kiq_ring, m->cp_mqd_base_addr_hi);
+	amdgpu_ring_write(kiq_ring, m->cp_hqd_pq_wptr_poll_addr_lo);
+	amdgpu_ring_write(kiq_ring, m->cp_hqd_pq_wptr_poll_addr_hi);
+	amdgpu_ring_commit(kiq_ring);
+
+out_unlock:
+	spin_unlock(&adev->gfx.kiq[0].ring_lock);
+	release_queue(adev);
+
+	return r;
+}
+
+static int kgd_hqd_dump(struct amdgpu_device *adev,
+			uint32_t pipe_id, uint32_t queue_id,
+			uint32_t (**dump)[2], uint32_t *n_regs, uint32_t inst)
+{
+	uint32_t i = 0, reg;
+#define HQD_N_REGS 56
+#define DUMP_REG(addr) do {				\
+		if (WARN_ON_ONCE(i >= HQD_N_REGS))	\
+			break;				\
+		(*dump)[i][0] = (addr) << 2;		\
+		(*dump)[i++][1] = RREG32_SOC15_IP(GC, addr);		\
+	} while (0)
+
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
+	if (*dump == NULL)
+		return -ENOMEM;
+
+	acquire_queue(adev, pipe_id, queue_id);
+
+	for (reg = SOC15_REG_OFFSET(GC, 0, mmCP_MQD_BASE_ADDR);
+	     reg <= SOC15_REG_OFFSET(GC, 0, mmCP_HQD_PQ_WPTR_HI); reg++)
+		DUMP_REG(reg);
+
+	release_queue(adev);
+
+	WARN_ON_ONCE(i != HQD_N_REGS);
+	*n_regs = i;
+
+	return 0;
+}
+
+static int kgd_hqd_sdma_load(struct amdgpu_device *adev, void *mqd,
+			     uint32_t __user *wptr, struct mm_struct *mm)
+{
+	struct v10_sdma_mqd *m;
+	uint32_t sdma_rlc_reg_offset;
+	unsigned long end_jiffies;
+	uint32_t data;
+	uint64_t data64;
+	uint64_t __user *wptr64 = (uint64_t __user *)wptr;
+
+	m = get_sdma_mqd(mqd);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev, m->sdma_engine_id,
+					    m->sdma_queue_id);
+
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL,
+		m->sdmax_rlcx_rb_cntl & (~SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK));
+
+	end_jiffies = msecs_to_jiffies(2000) + jiffies;
+	while (true) {
+		data = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_CONTEXT_STATUS);
+		if (data & SDMA0_RLC0_CONTEXT_STATUS__IDLE_MASK)
+			break;
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("SDMA RLC not idle in %s\n", __func__);
+			return -ETIME;
+		}
+		usleep_range(500, 1000);
+	}
+
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_DOORBELL_OFFSET,
+	       m->sdmax_rlcx_doorbell_offset);
+
+	data = REG_SET_FIELD(m->sdmax_rlcx_doorbell, SDMA0_RLC0_DOORBELL,
+			     ENABLE, 1);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_DOORBELL, data);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR,
+				m->sdmax_rlcx_rb_rptr);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR_HI,
+				m->sdmax_rlcx_rb_rptr_hi);
+
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_MINOR_PTR_UPDATE, 1);
+	if (read_user_wptr(mm, wptr64, data64)) {
+		WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_WPTR,
+		       lower_32_bits(data64));
+		WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_WPTR_HI,
+		       upper_32_bits(data64));
+	} else {
+		WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_WPTR,
+		       m->sdmax_rlcx_rb_rptr);
+		WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_WPTR_HI,
+		       m->sdmax_rlcx_rb_rptr_hi);
+	}
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_MINOR_PTR_UPDATE, 0);
+
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_BASE, m->sdmax_rlcx_rb_base);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_BASE_HI,
+			m->sdmax_rlcx_rb_base_hi);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR_ADDR_LO,
+			m->sdmax_rlcx_rb_rptr_addr_lo);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR_ADDR_HI,
+			m->sdmax_rlcx_rb_rptr_addr_hi);
+
+	data = REG_SET_FIELD(m->sdmax_rlcx_rb_cntl, SDMA0_RLC0_RB_CNTL,
+			     RB_ENABLE, 1);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL, data);
+
+	return 0;
+}
+
+static int kgd_hqd_sdma_dump(struct amdgpu_device *adev,
+			     uint32_t engine_id, uint32_t queue_id,
+			     uint32_t (**dump)[2], uint32_t *n_regs)
+{
+	uint32_t sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev,
+			engine_id, queue_id);
+	uint32_t i = 0, reg;
+#undef HQD_N_REGS
+#define HQD_N_REGS (19+6+7+10)
+
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
+	if (*dump == NULL)
+		return -ENOMEM;
+
+	for (reg = mmSDMA0_RLC0_RB_CNTL; reg <= mmSDMA0_RLC0_DOORBELL; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+	for (reg = mmSDMA0_RLC0_STATUS; reg <= mmSDMA0_RLC0_CSA_ADDR_HI; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+	for (reg = mmSDMA0_RLC0_IB_SUB_REMAIN;
+	     reg <= mmSDMA0_RLC0_MINOR_PTR_UPDATE; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+	for (reg = mmSDMA0_RLC0_MIDCMD_DATA0;
+	     reg <= mmSDMA0_RLC0_MIDCMD_CNTL; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+
+	WARN_ON_ONCE(i != HQD_N_REGS);
+	*n_regs = i;
+
+	return 0;
+}
+
+static bool kgd_hqd_is_occupied(struct amdgpu_device *adev,
+				uint64_t queue_address, uint32_t pipe_id,
+				uint32_t queue_id, uint32_t inst)
+{
+	uint32_t act;
+	bool retval = false;
+	uint32_t low, high;
+
+	acquire_queue(adev, pipe_id, queue_id);
+	act = RREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE);
+	if (act) {
+		low = lower_32_bits(queue_address >> 8);
+		high = upper_32_bits(queue_address >> 8);
+
+		if (low == RREG32_SOC15(GC, 0, mmCP_HQD_PQ_BASE) &&
+		   high == RREG32_SOC15(GC, 0, mmCP_HQD_PQ_BASE_HI))
+			retval = true;
+	}
+	release_queue(adev);
+	return retval;
+}
+
+static bool kgd_hqd_sdma_is_occupied(struct amdgpu_device *adev, void *mqd)
+{
+	struct v10_sdma_mqd *m;
+	uint32_t sdma_rlc_reg_offset;
+	uint32_t sdma_rlc_rb_cntl;
+
+	m = get_sdma_mqd(mqd);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev, m->sdma_engine_id,
+					    m->sdma_queue_id);
+
+	sdma_rlc_rb_cntl = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL);
+
+	if (sdma_rlc_rb_cntl & SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK)
+		return true;
+
+	return false;
+}
+
+static int kgd_hqd_destroy(struct amdgpu_device *adev, void *mqd,
+				enum kfd_preempt_type reset_type,
+				unsigned int utimeout, uint32_t pipe_id,
+				uint32_t queue_id, uint32_t inst)
+{
+	enum hqd_dequeue_request_type type;
+	unsigned long end_jiffies;
+	uint32_t temp;
+	struct v10_compute_mqd *m = get_mqd(mqd);
+
+	if (amdgpu_in_reset(adev))
+		return -EIO;
+
+#if 0
+	unsigned long flags;
+	int retry;
+#endif
+
+	acquire_queue(adev, pipe_id, queue_id);
+
+	if (m->cp_hqd_vmid == 0)
+		WREG32_FIELD15(GC, 0, RLC_CP_SCHEDULERS, scheduler1, 0);
+
+	switch (reset_type) {
+	case KFD_PREEMPT_TYPE_WAVEFRONT_DRAIN:
+		type = DRAIN_PIPE;
+		break;
+	case KFD_PREEMPT_TYPE_WAVEFRONT_RESET:
+		type = RESET_WAVES;
+		break;
+	case KFD_PREEMPT_TYPE_WAVEFRONT_SAVE:
+		type = SAVE_WAVES;
+		break;
+	default:
+		type = DRAIN_PIPE;
+		break;
+	}
+
+#if 0 /* Is this still needed? */
+	/* Workaround: If IQ timer is active and the wait time is close to or
+	 * equal to 0, dequeueing is not safe. Wait until either the wait time
+	 * is larger or timer is cleared. Also, ensure that IQ_REQ_PEND is
+	 * cleared before continuing. Also, ensure wait times are set to at
+	 * least 0x3.
+	 */
+	local_irq_save(flags);
+	preempt_disable();
+	retry = 5000; /* wait for 500 usecs at maximum */
+	while (true) {
+		temp = RREG32(mmCP_HQD_IQ_TIMER);
+		if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, PROCESSING_IQ)) {
+			pr_debug("HW is processing IQ\n");
+			goto loop;
+		}
+		if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, ACTIVE)) {
+			if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, RETRY_TYPE)
+					== 3) /* SEM-rearm is safe */
+				break;
+			/* Wait time 3 is safe for CP, but our MMIO read/write
+			 * time is close to 1 microsecond, so check for 10 to
+			 * leave more buffer room
+			 */
+			if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, WAIT_TIME)
+					>= 10)
+				break;
+			pr_debug("IQ timer is active\n");
+		} else
+			break;
+loop:
+		if (!retry) {
+			pr_err("CP HQD IQ timer status time out\n");
+			break;
+		}
+		ndelay(100);
+		--retry;
+	}
+	retry = 1000;
+	while (true) {
+		temp = RREG32(mmCP_HQD_DEQUEUE_REQUEST);
+		if (!(temp & CP_HQD_DEQUEUE_REQUEST__IQ_REQ_PEND_MASK))
+			break;
+		pr_debug("Dequeue request is pending\n");
+
+		if (!retry) {
+			pr_err("CP HQD dequeue request time out\n");
+			break;
+		}
+		ndelay(100);
+		--retry;
+	}
+	local_irq_restore(flags);
+	preempt_enable();
+#endif
+
+	WREG32_SOC15(GC, 0, mmCP_HQD_DEQUEUE_REQUEST, type);
+
+	end_jiffies = (utimeout * HZ / 1000) + jiffies;
+	while (true) {
+		temp = RREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE);
+		if (!(temp & CP_HQD_ACTIVE__ACTIVE_MASK))
+			break;
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("cp queue preemption time out.\n");
+			release_queue(adev);
+			return -ETIME;
+		}
+		usleep_range(500, 1000);
+	}
+
+	release_queue(adev);
+	return 0;
+}
+
+static int kgd_hqd_sdma_destroy(struct amdgpu_device *adev, void *mqd,
+				unsigned int utimeout)
+{
+	struct v10_sdma_mqd *m;
+	uint32_t sdma_rlc_reg_offset;
+	uint32_t temp;
+	unsigned long end_jiffies = (utimeout * HZ / 1000) + jiffies;
+
+	m = get_sdma_mqd(mqd);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev, m->sdma_engine_id,
+					    m->sdma_queue_id);
+
+	temp = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL);
+	temp = temp & ~SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK;
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL, temp);
+
+	while (true) {
+		temp = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_CONTEXT_STATUS);
+		if (temp & SDMA0_RLC0_CONTEXT_STATUS__IDLE_MASK)
+			break;
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("SDMA RLC not idle in %s\n", __func__);
+			return -ETIME;
+		}
+		usleep_range(500, 1000);
+	}
+
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_DOORBELL, 0);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL,
+		RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL) |
+		SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK);
+
+	m->sdmax_rlcx_rb_rptr = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR);
+	m->sdmax_rlcx_rb_rptr_hi =
+		RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR_HI);
+
+	return 0;
+}
+
+static bool get_atc_vmid_pasid_mapping_info(struct amdgpu_device *adev,
+					uint8_t vmid, uint16_t *p_pasid)
+{
+	uint32_t value;
+
+	value = RREG32(SOC15_REG_OFFSET(ATHUB, 0, mmATC_VMID0_PASID_MAPPING)
+		     + vmid);
+	*p_pasid = value & ATC_VMID0_PASID_MAPPING__PASID_MASK;
+
+	return !!(value & ATC_VMID0_PASID_MAPPING__VALID_MASK);
+}
+
+static int kgd_wave_control_execute(struct amdgpu_device *adev,
+					uint32_t gfx_index_val,
+					uint32_t sq_cmd, uint32_t inst)
+{
+	uint32_t data = 0;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	WREG32_SOC15(GC, 0, mmGRBM_GFX_INDEX, gfx_index_val);
+	WREG32_SOC15(GC, 0, mmSQ_CMD, sq_cmd);
+
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		INSTANCE_BROADCAST_WRITES, 1);
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		SA_BROADCAST_WRITES, 1);
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		SE_BROADCAST_WRITES, 1);
+
+	WREG32_SOC15(GC, 0, mmGRBM_GFX_INDEX, data);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+static void set_vm_context_page_table_base(struct amdgpu_device *adev,
+		uint32_t vmid, uint64_t page_table_base)
+{
+	if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid)) {
+		pr_err("trying to set page table base for wrong VMID %u\n",
+		       vmid);
+		return;
+	}
+
+	/* SDMA is on gfxhub as well for Navi1* series */
+	adev->gfxhub.funcs->setup_vm_pt_regs(adev, vmid, page_table_base);
+}
+
+/*
+ * GFX10 helper for wave launch stall requirements on debug trap setting.
+ *
+ * vmid:
+ *   Target VMID to stall/unstall.
+ *
+ * stall:
+ *   0-unstall wave launch (enable), 1-stall wave launch (disable).
+ *   After wavefront launch has been stalled, allocated waves must drain from
+ *   SPI in order for debug trap settings to take effect on those waves.
+ *   This is roughly a ~3500 clock cycle wait on SPI where a read on
+ *   SPI_GDBG_WAVE_CNTL translates to ~32 clock cycles.
+ *   KGD_GFX_V10_WAVE_LAUNCH_SPI_DRAIN_LATENCY indicates the number of reads required.
+ *
+ *   NOTE: We can afford to clear the entire STALL_VMID field on unstall
+ *   because current GFX10 chips cannot support multi-process debugging due to
+ *   trap configuration and masking being limited to global scope.  Always
+ *   assume single process conditions.
+ *
+ */
+
+#define KGD_GFX_V10_WAVE_LAUNCH_SPI_DRAIN_LATENCY	110
+static void kgd_gfx_v10_set_wave_launch_stall(struct amdgpu_device *adev, uint32_t vmid, bool stall)
+{
+	uint32_t data = RREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_WAVE_CNTL));
+	int i;
+
+	data = REG_SET_FIELD(data, SPI_GDBG_WAVE_CNTL, STALL_VMID,
+							stall ? 1 << vmid : 0);
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_WAVE_CNTL), data);
+
+	if (!stall)
+		return;
+
+	for (i = 0; i < KGD_GFX_V10_WAVE_LAUNCH_SPI_DRAIN_LATENCY; i++)
+		RREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_WAVE_CNTL));
+}
+
+uint32_t kgd_gfx_v10_enable_debug_trap(struct amdgpu_device *adev,
+				bool restore_dbg_registers,
+				uint32_t vmid)
+{
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	kgd_gfx_v10_set_wave_launch_stall(adev, vmid, true);
+
+	/* assume gfx off is disabled for the debug session if rlc restore not supported. */
+	if (restore_dbg_registers) {
+		uint32_t data = 0;
+
+		data = REG_SET_FIELD(data, SPI_GDBG_TRAP_CONFIG,
+				VMID_SEL, 1 << vmid);
+		data = REG_SET_FIELD(data, SPI_GDBG_TRAP_CONFIG,
+				TRAP_EN, 1);
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_TRAP_CONFIG), data);
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_TRAP_DATA0), 0);
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_TRAP_DATA1), 0);
+
+		kgd_gfx_v10_set_wave_launch_stall(adev, vmid, false);
+
+		mutex_unlock(&adev->grbm_idx_mutex);
+
+		return 0;
+	}
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_TRAP_MASK), 0);
+
+	kgd_gfx_v10_set_wave_launch_stall(adev, vmid, false);
+
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+uint32_t kgd_gfx_v10_disable_debug_trap(struct amdgpu_device *adev,
+					bool keep_trap_enabled,
+					uint32_t vmid)
+{
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	kgd_gfx_v10_set_wave_launch_stall(adev, vmid, true);
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_TRAP_MASK), 0);
+
+	kgd_gfx_v10_set_wave_launch_stall(adev, vmid, false);
+
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+int kgd_gfx_v10_validate_trap_override_request(struct amdgpu_device *adev,
+					      uint32_t trap_override,
+					      uint32_t *trap_mask_supported)
+{
+	*trap_mask_supported &= KFD_DBG_TRAP_MASK_DBG_ADDRESS_WATCH;
+
+	/* The SPI_GDBG_TRAP_MASK register is global and affects all
+	 * processes. Only allow OR-ing the address-watch bit, since
+	 * this only affects processes under the debugger. Other bits
+	 * should stay 0 to avoid the debugger interfering with other
+	 * processes.
+	 */
+	if (trap_override != KFD_DBG_TRAP_OVERRIDE_OR)
+		return -EINVAL;
+
+	return 0;
+}
+
+uint32_t kgd_gfx_v10_set_wave_launch_trap_override(struct amdgpu_device *adev,
+					      uint32_t vmid,
+					      uint32_t trap_override,
+					      uint32_t trap_mask_bits,
+					      uint32_t trap_mask_request,
+					      uint32_t *trap_mask_prev,
+					      uint32_t kfd_dbg_trap_cntl_prev)
+{
+	uint32_t data, wave_cntl_prev;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	wave_cntl_prev = RREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_WAVE_CNTL));
+
+	kgd_gfx_v10_set_wave_launch_stall(adev, vmid, true);
+
+	data = RREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_TRAP_MASK));
+	*trap_mask_prev = REG_GET_FIELD(data, SPI_GDBG_TRAP_MASK, EXCP_EN);
+
+	trap_mask_bits = (trap_mask_bits & trap_mask_request) |
+		(*trap_mask_prev & ~trap_mask_request);
+
+	data = REG_SET_FIELD(data, SPI_GDBG_TRAP_MASK, EXCP_EN, trap_mask_bits);
+	data = REG_SET_FIELD(data, SPI_GDBG_TRAP_MASK, REPLACE, trap_override);
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_TRAP_MASK), data);
+
+	/* We need to preserve wave launch mode stall settings. */
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_WAVE_CNTL), wave_cntl_prev);
+
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+uint32_t kgd_gfx_v10_set_wave_launch_mode(struct amdgpu_device *adev,
+					uint8_t wave_launch_mode,
+					uint32_t vmid)
+{
+	uint32_t data = 0;
+	bool is_mode_set = !!wave_launch_mode;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	kgd_gfx_v10_set_wave_launch_stall(adev, vmid, true);
+
+	data = REG_SET_FIELD(data, SPI_GDBG_WAVE_CNTL2,
+			VMID_MASK, is_mode_set ? 1 << vmid : 0);
+	data = REG_SET_FIELD(data, SPI_GDBG_WAVE_CNTL2,
+			MODE, is_mode_set ? wave_launch_mode : 0);
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_WAVE_CNTL2), data);
+
+	kgd_gfx_v10_set_wave_launch_stall(adev, vmid, false);
+
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+#define TCP_WATCH_STRIDE (mmTCP_WATCH1_ADDR_H - mmTCP_WATCH0_ADDR_H)
+#define SQ_WATCH_STRIDE (mmSQ_WATCH1_ADDR_H - mmSQ_WATCH0_ADDR_H)
+uint32_t kgd_gfx_v10_set_address_watch(struct amdgpu_device *adev,
+					uint64_t watch_address,
+					uint32_t watch_address_mask,
+					uint32_t watch_id,
+					uint32_t watch_mode,
+					uint32_t debug_vmid,
+					uint32_t inst)
+{
+	/* SQ_WATCH?_ADDR_* and TCP_WATCH?_ADDR_* are programmed with the
+	 * same values.
+	 */
+	uint32_t watch_address_high;
+	uint32_t watch_address_low;
+	uint32_t tcp_watch_address_cntl;
+	uint32_t sq_watch_address_cntl;
+
+	watch_address_low = lower_32_bits(watch_address);
+	watch_address_high = upper_32_bits(watch_address) & 0xffff;
+
+	tcp_watch_address_cntl = 0;
+	tcp_watch_address_cntl = REG_SET_FIELD(tcp_watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			VMID,
+			debug_vmid);
+	tcp_watch_address_cntl = REG_SET_FIELD(tcp_watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			MODE,
+			watch_mode);
+	tcp_watch_address_cntl = REG_SET_FIELD(tcp_watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			MASK,
+			watch_address_mask >> 7);
+
+	sq_watch_address_cntl = 0;
+	sq_watch_address_cntl = REG_SET_FIELD(sq_watch_address_cntl,
+			SQ_WATCH0_CNTL,
+			VMID,
+			debug_vmid);
+	sq_watch_address_cntl = REG_SET_FIELD(sq_watch_address_cntl,
+			SQ_WATCH0_CNTL,
+			MODE,
+			watch_mode);
+	sq_watch_address_cntl = REG_SET_FIELD(sq_watch_address_cntl,
+			SQ_WATCH0_CNTL,
+			MASK,
+			watch_address_mask >> 6);
+
+	/* Turning off this watch point until we set all the registers */
+	tcp_watch_address_cntl = REG_SET_FIELD(tcp_watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			VALID,
+			0);
+	WREG32((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_CNTL) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			tcp_watch_address_cntl);
+
+	sq_watch_address_cntl = REG_SET_FIELD(sq_watch_address_cntl,
+			SQ_WATCH0_CNTL,
+			VALID,
+			0);
+	WREG32((SOC15_REG_OFFSET(GC, 0, mmSQ_WATCH0_CNTL) +
+			(watch_id * SQ_WATCH_STRIDE)),
+			sq_watch_address_cntl);
+
+	/* Program {TCP,SQ}_WATCH?_ADDR* */
+	WREG32((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_ADDR_H) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_high);
+	WREG32((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_ADDR_L) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_low);
+
+	WREG32((SOC15_REG_OFFSET(GC, 0, mmSQ_WATCH0_ADDR_H) +
+			(watch_id * SQ_WATCH_STRIDE)),
+			watch_address_high);
+	WREG32((SOC15_REG_OFFSET(GC, 0, mmSQ_WATCH0_ADDR_L) +
+			(watch_id * SQ_WATCH_STRIDE)),
+			watch_address_low);
+
+	/* Enable the watch point */
+	tcp_watch_address_cntl = REG_SET_FIELD(tcp_watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			VALID,
+			1);
+	WREG32((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_CNTL) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			tcp_watch_address_cntl);
+
+	sq_watch_address_cntl = REG_SET_FIELD(sq_watch_address_cntl,
+			SQ_WATCH0_CNTL,
+			VALID,
+			1);
+	WREG32((SOC15_REG_OFFSET(GC, 0, mmSQ_WATCH0_CNTL) +
+			(watch_id * SQ_WATCH_STRIDE)),
+			sq_watch_address_cntl);
+
+	return 0;
+}
+
+uint32_t kgd_gfx_v10_clear_address_watch(struct amdgpu_device *adev,
+					uint32_t watch_id)
+{
+	uint32_t watch_address_cntl;
+
+	watch_address_cntl = 0;
+
+	WREG32((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_CNTL) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_cntl);
+
+	WREG32((SOC15_REG_OFFSET(GC, 0, mmSQ_WATCH0_CNTL) +
+			(watch_id * SQ_WATCH_STRIDE)),
+			watch_address_cntl);
+
+	return 0;
+}
+#undef TCP_WATCH_STRIDE
+#undef SQ_WATCH_STRIDE
+
+
+/* kgd_gfx_v10_get_iq_wait_times: Returns the mmCP_IQ_WAIT_TIME1/2 values
+ * The values read are:
+ *     ib_offload_wait_time     -- Wait Count for Indirect Buffer Offloads.
+ *     atomic_offload_wait_time -- Wait Count for L2 and GDS Atomics Offloads.
+ *     wrm_offload_wait_time    -- Wait Count for WAIT_REG_MEM Offloads.
+ *     gws_wait_time            -- Wait Count for Global Wave Syncs.
+ *     que_sleep_wait_time      -- Wait Count for Dequeue Retry.
+ *     sch_wave_wait_time       -- Wait Count for Scheduling Wave Message.
+ *     sem_rearm_wait_time      -- Wait Count for Semaphore re-arm.
+ *     deq_retry_wait_time      -- Wait Count for Global Wave Syncs.
+ */
+void kgd_gfx_v10_get_iq_wait_times(struct amdgpu_device *adev,
+					uint32_t *wait_times,
+					uint32_t inst)
+
+{
+	*wait_times = RREG32(SOC15_REG_OFFSET(GC, 0, mmCP_IQ_WAIT_TIME2));
+}
+
+void kgd_gfx_v10_build_dequeue_wait_counts_packet_info(struct amdgpu_device *adev,
+						uint32_t wait_times,
+						uint32_t sch_wave,
+						uint32_t que_sleep,
+						uint32_t *reg_offset,
+						uint32_t *reg_data)
+{
+	*reg_data = wait_times;
+
+	if (sch_wave)
+		*reg_data = REG_SET_FIELD(*reg_data,
+				CP_IQ_WAIT_TIME2,
+				SCH_WAVE,
+				sch_wave);
+	if (que_sleep)
+		*reg_data = REG_SET_FIELD(*reg_data,
+				CP_IQ_WAIT_TIME2,
+				QUE_SLEEP,
+				que_sleep);
+
+	*reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_IQ_WAIT_TIME2);
+}
+
+static void program_trap_handler_settings(struct amdgpu_device *adev,
+		uint32_t vmid, uint64_t tba_addr, uint64_t tma_addr,
+		uint32_t inst)
+{
+	lock_srbm(adev, 0, 0, 0, vmid);
+
+	/*
+	 * Program TBA registers
+	 */
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSQ_SHADER_TBA_LO),
+			lower_32_bits(tba_addr >> 8));
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSQ_SHADER_TBA_HI),
+			upper_32_bits(tba_addr >> 8) |
+			(1 << SQ_SHADER_TBA_HI__TRAP_EN__SHIFT));
+
+	/*
+	 * Program TMA registers
+	 */
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSQ_SHADER_TMA_LO),
+			lower_32_bits(tma_addr >> 8));
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSQ_SHADER_TMA_HI),
+			upper_32_bits(tma_addr >> 8));
+
+	unlock_srbm(adev);
+}
+
+uint64_t kgd_gfx_v10_hqd_get_pq_addr(struct amdgpu_device *adev,
+				     uint32_t pipe_id, uint32_t queue_id,
+				     uint32_t inst)
+{
+	return 0;
+}
+
+uint64_t kgd_gfx_v10_hqd_reset(struct amdgpu_device *adev,
+			       uint32_t pipe_id, uint32_t queue_id,
+			       uint32_t inst, unsigned int utimeout)
+{
+	return 0;
+}
+
+uint32_t kgd_gfx_v10_hqd_sdma_get_doorbell(struct amdgpu_device *adev,
+					   int engine, int queue)
+{
+	return 0;
+}
+
+const struct kfd2kgd_calls gfx_v10_kfd2kgd = {
+	.program_sh_mem_settings = kgd_program_sh_mem_settings,
+	.set_pasid_vmid_mapping = kgd_set_pasid_vmid_mapping,
+	.init_interrupts = kgd_init_interrupts,
+	.hqd_load = kgd_hqd_load,
+	.hiq_mqd_load = kgd_hiq_mqd_load,
+	.hqd_sdma_load = kgd_hqd_sdma_load,
+	.hqd_dump = kgd_hqd_dump,
+	.hqd_sdma_dump = kgd_hqd_sdma_dump,
+	.hqd_is_occupied = kgd_hqd_is_occupied,
+	.hqd_sdma_is_occupied = kgd_hqd_sdma_is_occupied,
+	.hqd_destroy = kgd_hqd_destroy,
+	.hqd_sdma_destroy = kgd_hqd_sdma_destroy,
+	.wave_control_execute = kgd_wave_control_execute,
+	.get_atc_vmid_pasid_mapping_info =
+			get_atc_vmid_pasid_mapping_info,
+	.set_vm_context_page_table_base = set_vm_context_page_table_base,
+	.enable_debug_trap = kgd_gfx_v10_enable_debug_trap,
+	.disable_debug_trap = kgd_gfx_v10_disable_debug_trap,
+	.validate_trap_override_request = kgd_gfx_v10_validate_trap_override_request,
+	.set_wave_launch_trap_override = kgd_gfx_v10_set_wave_launch_trap_override,
+	.set_wave_launch_mode = kgd_gfx_v10_set_wave_launch_mode,
+	.set_address_watch = kgd_gfx_v10_set_address_watch,
+	.clear_address_watch = kgd_gfx_v10_clear_address_watch,
+	.get_iq_wait_times = kgd_gfx_v10_get_iq_wait_times,
+	.build_dequeue_wait_counts_packet_info = kgd_gfx_v10_build_dequeue_wait_counts_packet_info,
+	.program_trap_handler_settings = program_trap_handler_settings,
+	.hqd_get_pq_addr = kgd_gfx_v10_hqd_get_pq_addr,
+	.hqd_reset = kgd_gfx_v10_hqd_reset,
+	.hqd_sdma_get_doorbell = kgd_gfx_v10_hqd_sdma_get_doorbell
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.h
new file mode 100644
index 000000000000..a4c607c88178
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.h
@@ -0,0 +1,70 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+uint32_t kgd_gfx_v10_enable_debug_trap(struct amdgpu_device *adev,
+				      bool restore_dbg_registers,
+				      uint32_t vmid);
+uint32_t kgd_gfx_v10_disable_debug_trap(struct amdgpu_device *adev,
+					bool keep_trap_enabled,
+					uint32_t vmid);
+int kgd_gfx_v10_validate_trap_override_request(struct amdgpu_device *adev,
+					     uint32_t trap_override,
+					     uint32_t *trap_mask_supported);
+uint32_t kgd_gfx_v10_set_wave_launch_trap_override(struct amdgpu_device *adev,
+					     uint32_t vmid,
+					     uint32_t trap_override,
+					     uint32_t trap_mask_bits,
+					     uint32_t trap_mask_request,
+					     uint32_t *trap_mask_prev,
+					     uint32_t kfd_dbg_trap_cntl_prev);
+uint32_t kgd_gfx_v10_set_wave_launch_mode(struct amdgpu_device *adev,
+					 uint8_t wave_launch_mode,
+					 uint32_t vmid);
+uint32_t kgd_gfx_v10_set_address_watch(struct amdgpu_device *adev,
+					uint64_t watch_address,
+					uint32_t watch_address_mask,
+					uint32_t watch_id,
+					uint32_t watch_mode,
+					uint32_t debug_vmid,
+					uint32_t inst);
+uint32_t kgd_gfx_v10_clear_address_watch(struct amdgpu_device *adev,
+					uint32_t watch_id);
+void kgd_gfx_v10_get_iq_wait_times(struct amdgpu_device *adev,
+				uint32_t *wait_times,
+				uint32_t inst);
+void kgd_gfx_v10_build_dequeue_wait_counts_packet_info(struct amdgpu_device *adev,
+					       uint32_t wait_times,
+					       uint32_t sch_wave,
+					       uint32_t que_sleep,
+					       uint32_t *reg_offset,
+					       uint32_t *reg_data);
+uint64_t kgd_gfx_v10_hqd_get_pq_addr(struct amdgpu_device *adev,
+				    uint32_t pipe_id,
+				    uint32_t queue_id,
+				    uint32_t inst);
+uint64_t kgd_gfx_v10_hqd_reset(struct amdgpu_device *adev,
+			      uint32_t pipe_id,
+			      uint32_t queue_id,
+			      uint32_t inst,
+			      unsigned int utimeout);
+uint32_t kgd_gfx_v10_hqd_sdma_get_doorbell(struct amdgpu_device *adev,
+					   int engine, int queue);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10_3.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10_3.c
new file mode 100644
index 000000000000..f2278a0937ff
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10_3.c
@@ -0,0 +1,687 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+#include <linux/mmu_context.h>
+#include "amdgpu.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_amdkfd_gfx_v10.h"
+#include "gc/gc_10_3_0_offset.h"
+#include "gc/gc_10_3_0_sh_mask.h"
+#include "oss/osssys_5_0_0_offset.h"
+#include "oss/osssys_5_0_0_sh_mask.h"
+#include "athub/athub_2_1_0_offset.h"
+#include "athub/athub_2_1_0_sh_mask.h"
+#include "soc15_common.h"
+#include "v10_structs.h"
+#include "nv.h"
+#include "nvd.h"
+
+enum hqd_dequeue_request_type {
+	NO_ACTION = 0,
+	DRAIN_PIPE,
+	RESET_WAVES,
+	SAVE_WAVES
+};
+
+static void lock_srbm(struct amdgpu_device *adev, uint32_t mec, uint32_t pipe,
+			uint32_t queue, uint32_t vmid)
+{
+	mutex_lock(&adev->srbm_mutex);
+	nv_grbm_select(adev, mec, pipe, queue, vmid);
+}
+
+static void unlock_srbm(struct amdgpu_device *adev)
+{
+	nv_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static void acquire_queue(struct amdgpu_device *adev, uint32_t pipe_id,
+				uint32_t queue_id)
+{
+	uint32_t mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	uint32_t pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+	lock_srbm(adev, mec, pipe, queue_id, 0);
+}
+
+static uint64_t get_queue_mask(struct amdgpu_device *adev,
+			       uint32_t pipe_id, uint32_t queue_id)
+{
+	unsigned int bit = pipe_id * adev->gfx.mec.num_queue_per_pipe +
+			queue_id;
+
+	return 1ull << bit;
+}
+
+static void release_queue(struct amdgpu_device *adev)
+{
+	unlock_srbm(adev);
+}
+
+static void program_sh_mem_settings_v10_3(struct amdgpu_device *adev, uint32_t vmid,
+					uint32_t sh_mem_config,
+					uint32_t sh_mem_ape1_base,
+					uint32_t sh_mem_ape1_limit,
+					uint32_t sh_mem_bases, uint32_t inst)
+{
+	lock_srbm(adev, 0, 0, 0, vmid);
+
+	WREG32_SOC15(GC, 0, mmSH_MEM_CONFIG, sh_mem_config);
+	WREG32_SOC15(GC, 0, mmSH_MEM_BASES, sh_mem_bases);
+	/* APE1 no longer exists on GFX9 */
+
+	unlock_srbm(adev);
+}
+
+/* ATC is defeatured on Sienna_Cichlid */
+static int set_pasid_vmid_mapping_v10_3(struct amdgpu_device *adev, unsigned int pasid,
+					unsigned int vmid, uint32_t inst)
+{
+	uint32_t value = pasid << IH_VMID_0_LUT__PASID__SHIFT;
+
+	/* Mapping vmid to pasid also for IH block */
+	pr_debug("mapping vmid %d -> pasid %d in IH block for GFX client\n",
+			vmid, pasid);
+	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_VMID_0_LUT) + vmid, value);
+
+	return 0;
+}
+
+static int init_interrupts_v10_3(struct amdgpu_device *adev, uint32_t pipe_id,
+				uint32_t inst)
+{
+	uint32_t mec;
+	uint32_t pipe;
+
+	mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+	lock_srbm(adev, mec, pipe, 0, 0);
+
+	WREG32_SOC15(GC, 0, mmCPC_INT_CNTL,
+		CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK |
+		CP_INT_CNTL_RING0__OPCODE_ERROR_INT_ENABLE_MASK);
+
+	unlock_srbm(adev);
+
+	return 0;
+}
+
+static uint32_t get_sdma_rlc_reg_offset(struct amdgpu_device *adev,
+				unsigned int engine_id,
+				unsigned int queue_id)
+{
+	uint32_t sdma_engine_reg_base = 0;
+	uint32_t sdma_rlc_reg_offset;
+
+	switch (engine_id) {
+	default:
+		dev_warn(adev->dev,
+			 "Invalid sdma engine id (%d), using engine id 0\n",
+			 engine_id);
+		fallthrough;
+	case 0:
+		sdma_engine_reg_base = SOC15_REG_OFFSET(SDMA0, 0,
+				mmSDMA0_RLC0_RB_CNTL) - mmSDMA0_RLC0_RB_CNTL;
+		break;
+	case 1:
+		sdma_engine_reg_base = SOC15_REG_OFFSET(SDMA0, 0,
+				mmSDMA1_RLC0_RB_CNTL) - mmSDMA0_RLC0_RB_CNTL;
+		break;
+	case 2:
+		sdma_engine_reg_base = SOC15_REG_OFFSET(SDMA0, 0,
+				mmSDMA2_RLC0_RB_CNTL) - mmSDMA0_RLC0_RB_CNTL;
+		break;
+	case 3:
+		sdma_engine_reg_base = SOC15_REG_OFFSET(SDMA0, 0,
+				mmSDMA3_RLC0_RB_CNTL) - mmSDMA0_RLC0_RB_CNTL;
+		break;
+	}
+
+	sdma_rlc_reg_offset = sdma_engine_reg_base
+		+ queue_id * (mmSDMA0_RLC1_RB_CNTL - mmSDMA0_RLC0_RB_CNTL);
+
+	pr_debug("RLC register offset for SDMA%d RLC%d: 0x%x\n", engine_id,
+			queue_id, sdma_rlc_reg_offset);
+
+	return sdma_rlc_reg_offset;
+}
+
+static inline struct v10_compute_mqd *get_mqd(void *mqd)
+{
+	return (struct v10_compute_mqd *)mqd;
+}
+
+static inline struct v10_sdma_mqd *get_sdma_mqd(void *mqd)
+{
+	return (struct v10_sdma_mqd *)mqd;
+}
+
+static int hqd_load_v10_3(struct amdgpu_device *adev, void *mqd,
+			uint32_t pipe_id, uint32_t queue_id,
+			uint32_t __user *wptr, uint32_t wptr_shift,
+			uint32_t wptr_mask, struct mm_struct *mm, uint32_t inst)
+{
+	struct v10_compute_mqd *m;
+	uint32_t *mqd_hqd;
+	uint32_t reg, hqd_base, data;
+
+	m = get_mqd(mqd);
+
+	pr_debug("Load hqd of pipe %d queue %d\n", pipe_id, queue_id);
+	acquire_queue(adev, pipe_id, queue_id);
+
+	/* HIQ is set during driver init period with vmid set to 0*/
+	if (m->cp_hqd_vmid == 0) {
+		uint32_t value, mec, pipe;
+
+		mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+		pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+		pr_debug("kfd: set HIQ, mec:%d, pipe:%d, queue:%d.\n",
+			mec, pipe, queue_id);
+		value = RREG32_SOC15(GC, 0, mmRLC_CP_SCHEDULERS);
+		value = REG_SET_FIELD(value, RLC_CP_SCHEDULERS, scheduler1,
+			((mec << 5) | (pipe << 3) | queue_id | 0x80));
+		WREG32_SOC15(GC, 0, mmRLC_CP_SCHEDULERS, value);
+	}
+
+	/* HQD registers extend from CP_MQD_BASE_ADDR to CP_HQD_EOP_WPTR_MEM. */
+	mqd_hqd = &m->cp_mqd_base_addr_lo;
+	hqd_base = SOC15_REG_OFFSET(GC, 0, mmCP_MQD_BASE_ADDR);
+
+	for (reg = hqd_base;
+	     reg <= SOC15_REG_OFFSET(GC, 0, mmCP_HQD_PQ_WPTR_HI); reg++)
+		WREG32_SOC15_IP(GC, reg, mqd_hqd[reg - hqd_base]);
+
+
+	/* Activate doorbell logic before triggering WPTR poll. */
+	data = REG_SET_FIELD(m->cp_hqd_pq_doorbell_control,
+			     CP_HQD_PQ_DOORBELL_CONTROL, DOORBELL_EN, 1);
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL, data);
+
+	if (wptr) {
+		/* Don't read wptr with get_user because the user
+		 * context may not be accessible (if this function
+		 * runs in a work queue). Instead trigger a one-shot
+		 * polling read from memory in the CP. This assumes
+		 * that wptr is GPU-accessible in the queue's VMID via
+		 * ATC or SVM. WPTR==RPTR before starting the poll so
+		 * the CP starts fetching new commands from the right
+		 * place.
+		 *
+		 * Guessing a 64-bit WPTR from a 32-bit RPTR is a bit
+		 * tricky. Assume that the queue didn't overflow. The
+		 * number of valid bits in the 32-bit RPTR depends on
+		 * the queue size. The remaining bits are taken from
+		 * the saved 64-bit WPTR. If the WPTR wrapped, add the
+		 * queue size.
+		 */
+		uint32_t queue_size =
+			2 << REG_GET_FIELD(m->cp_hqd_pq_control,
+					   CP_HQD_PQ_CONTROL, QUEUE_SIZE);
+		uint64_t guessed_wptr = m->cp_hqd_pq_rptr & (queue_size - 1);
+
+		if ((m->cp_hqd_pq_wptr_lo & (queue_size - 1)) < guessed_wptr)
+			guessed_wptr += queue_size;
+		guessed_wptr += m->cp_hqd_pq_wptr_lo & ~(queue_size - 1);
+		guessed_wptr += (uint64_t)m->cp_hqd_pq_wptr_hi << 32;
+
+		WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_LO,
+		       lower_32_bits(guessed_wptr));
+		WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_HI,
+		       upper_32_bits(guessed_wptr));
+		WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_POLL_ADDR,
+		       lower_32_bits((uint64_t)wptr));
+		WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_POLL_ADDR_HI,
+		       upper_32_bits((uint64_t)wptr));
+		pr_debug("%s setting CP_PQ_WPTR_POLL_CNTL1 to %x\n", __func__,
+			 (uint32_t)get_queue_mask(adev, pipe_id, queue_id));
+		WREG32_SOC15(GC, 0, mmCP_PQ_WPTR_POLL_CNTL1,
+		       (uint32_t)get_queue_mask(adev, pipe_id, queue_id));
+	}
+
+	/* Start the EOP fetcher */
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_EOP_RPTR),
+	       REG_SET_FIELD(m->cp_hqd_eop_rptr,
+			     CP_HQD_EOP_RPTR, INIT_FETCHER, 1));
+
+	data = REG_SET_FIELD(m->cp_hqd_active, CP_HQD_ACTIVE, ACTIVE, 1);
+	WREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE, data);
+
+	release_queue(adev);
+
+	return 0;
+}
+
+static int hiq_mqd_load_v10_3(struct amdgpu_device *adev, void *mqd,
+			    uint32_t pipe_id, uint32_t queue_id,
+			    uint32_t doorbell_off, uint32_t inst)
+{
+	struct amdgpu_ring *kiq_ring = &adev->gfx.kiq[0].ring;
+	struct v10_compute_mqd *m;
+	uint32_t mec, pipe;
+	int r;
+
+	m = get_mqd(mqd);
+
+	acquire_queue(adev, pipe_id, queue_id);
+
+	mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+	pr_debug("kfd: set HIQ, mec:%d, pipe:%d, queue:%d.\n",
+		 mec, pipe, queue_id);
+
+	spin_lock(&adev->gfx.kiq[0].ring_lock);
+	r = amdgpu_ring_alloc(kiq_ring, 7);
+	if (r) {
+		pr_err("Failed to alloc KIQ (%d).\n", r);
+		goto out_unlock;
+	}
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_MAP_QUEUES, 5));
+	amdgpu_ring_write(kiq_ring,
+			  PACKET3_MAP_QUEUES_QUEUE_SEL(0) | /* Queue_Sel */
+			  PACKET3_MAP_QUEUES_VMID(m->cp_hqd_vmid) | /* VMID */
+			  PACKET3_MAP_QUEUES_QUEUE(queue_id) |
+			  PACKET3_MAP_QUEUES_PIPE(pipe) |
+			  PACKET3_MAP_QUEUES_ME((mec - 1)) |
+			  PACKET3_MAP_QUEUES_QUEUE_TYPE(0) | /*queue_type: normal compute queue */
+			  PACKET3_MAP_QUEUES_ALLOC_FORMAT(0) | /* alloc format: all_on_one_pipe */
+			  PACKET3_MAP_QUEUES_ENGINE_SEL(1) | /* engine_sel: hiq */
+			  PACKET3_MAP_QUEUES_NUM_QUEUES(1)); /* num_queues: must be 1 */
+	amdgpu_ring_write(kiq_ring,
+			  PACKET3_MAP_QUEUES_DOORBELL_OFFSET(doorbell_off));
+	amdgpu_ring_write(kiq_ring, m->cp_mqd_base_addr_lo);
+	amdgpu_ring_write(kiq_ring, m->cp_mqd_base_addr_hi);
+	amdgpu_ring_write(kiq_ring, m->cp_hqd_pq_wptr_poll_addr_lo);
+	amdgpu_ring_write(kiq_ring, m->cp_hqd_pq_wptr_poll_addr_hi);
+	amdgpu_ring_commit(kiq_ring);
+
+out_unlock:
+	spin_unlock(&adev->gfx.kiq[0].ring_lock);
+	release_queue(adev);
+
+	return r;
+}
+
+static int hqd_dump_v10_3(struct amdgpu_device *adev,
+			uint32_t pipe_id, uint32_t queue_id,
+			uint32_t (**dump)[2], uint32_t *n_regs, uint32_t inst)
+{
+	uint32_t i = 0, reg;
+#define HQD_N_REGS 56
+#define DUMP_REG(addr) do {				\
+		if (WARN_ON_ONCE(i >= HQD_N_REGS))	\
+			break;				\
+		(*dump)[i][0] = (addr) << 2;		\
+		(*dump)[i++][1] = RREG32_SOC15_IP(GC, addr);		\
+	} while (0)
+
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
+	if (*dump == NULL)
+		return -ENOMEM;
+
+	acquire_queue(adev, pipe_id, queue_id);
+
+	for (reg = SOC15_REG_OFFSET(GC, 0, mmCP_MQD_BASE_ADDR);
+	     reg <= SOC15_REG_OFFSET(GC, 0, mmCP_HQD_PQ_WPTR_HI); reg++)
+		DUMP_REG(reg);
+
+	release_queue(adev);
+
+	WARN_ON_ONCE(i != HQD_N_REGS);
+	*n_regs = i;
+
+	return 0;
+}
+
+static int hqd_sdma_load_v10_3(struct amdgpu_device *adev, void *mqd,
+			     uint32_t __user *wptr, struct mm_struct *mm)
+{
+	struct v10_sdma_mqd *m;
+	uint32_t sdma_rlc_reg_offset;
+	unsigned long end_jiffies;
+	uint32_t data;
+	uint64_t data64;
+	uint64_t __user *wptr64 = (uint64_t __user *)wptr;
+
+	m = get_sdma_mqd(mqd);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev, m->sdma_engine_id,
+					    m->sdma_queue_id);
+
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL,
+		m->sdmax_rlcx_rb_cntl & (~SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK));
+
+	end_jiffies = msecs_to_jiffies(2000) + jiffies;
+	while (true) {
+		data = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_CONTEXT_STATUS);
+		if (data & SDMA0_RLC0_CONTEXT_STATUS__IDLE_MASK)
+			break;
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("SDMA RLC not idle in %s\n", __func__);
+			return -ETIME;
+		}
+		usleep_range(500, 1000);
+	}
+
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_DOORBELL_OFFSET,
+	       m->sdmax_rlcx_doorbell_offset);
+
+	data = REG_SET_FIELD(m->sdmax_rlcx_doorbell, SDMA0_RLC0_DOORBELL,
+			     ENABLE, 1);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_DOORBELL, data);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR,
+				m->sdmax_rlcx_rb_rptr);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR_HI,
+				m->sdmax_rlcx_rb_rptr_hi);
+
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_MINOR_PTR_UPDATE, 1);
+	if (read_user_wptr(mm, wptr64, data64)) {
+		WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_WPTR,
+		       lower_32_bits(data64));
+		WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_WPTR_HI,
+		       upper_32_bits(data64));
+	} else {
+		WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_WPTR,
+		       m->sdmax_rlcx_rb_rptr);
+		WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_WPTR_HI,
+		       m->sdmax_rlcx_rb_rptr_hi);
+	}
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_MINOR_PTR_UPDATE, 0);
+
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_BASE, m->sdmax_rlcx_rb_base);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_BASE_HI,
+			m->sdmax_rlcx_rb_base_hi);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR_ADDR_LO,
+			m->sdmax_rlcx_rb_rptr_addr_lo);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR_ADDR_HI,
+			m->sdmax_rlcx_rb_rptr_addr_hi);
+
+	data = REG_SET_FIELD(m->sdmax_rlcx_rb_cntl, SDMA0_RLC0_RB_CNTL,
+			     RB_ENABLE, 1);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL, data);
+
+	return 0;
+}
+
+static int hqd_sdma_dump_v10_3(struct amdgpu_device *adev,
+			     uint32_t engine_id, uint32_t queue_id,
+			     uint32_t (**dump)[2], uint32_t *n_regs)
+{
+	uint32_t sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev,
+			engine_id, queue_id);
+	uint32_t i = 0, reg;
+#undef HQD_N_REGS
+#define HQD_N_REGS (19+6+7+12)
+
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
+	if (*dump == NULL)
+		return -ENOMEM;
+
+	for (reg = mmSDMA0_RLC0_RB_CNTL; reg <= mmSDMA0_RLC0_DOORBELL; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+	for (reg = mmSDMA0_RLC0_STATUS; reg <= mmSDMA0_RLC0_CSA_ADDR_HI; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+	for (reg = mmSDMA0_RLC0_IB_SUB_REMAIN;
+	     reg <= mmSDMA0_RLC0_MINOR_PTR_UPDATE; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+	for (reg = mmSDMA0_RLC0_MIDCMD_DATA0;
+	     reg <= mmSDMA0_RLC0_MIDCMD_CNTL; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+
+	WARN_ON_ONCE(i != HQD_N_REGS);
+	*n_regs = i;
+
+	return 0;
+}
+
+static bool hqd_is_occupied_v10_3(struct amdgpu_device *adev,
+				uint64_t queue_address, uint32_t pipe_id,
+				uint32_t queue_id, uint32_t inst)
+{
+	uint32_t act;
+	bool retval = false;
+	uint32_t low, high;
+
+	acquire_queue(adev, pipe_id, queue_id);
+	act = RREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE);
+	if (act) {
+		low = lower_32_bits(queue_address >> 8);
+		high = upper_32_bits(queue_address >> 8);
+
+		if (low == RREG32_SOC15(GC, 0, mmCP_HQD_PQ_BASE) &&
+		   high == RREG32_SOC15(GC, 0, mmCP_HQD_PQ_BASE_HI))
+			retval = true;
+	}
+	release_queue(adev);
+	return retval;
+}
+
+static bool hqd_sdma_is_occupied_v10_3(struct amdgpu_device *adev,
+				void *mqd)
+{
+	struct v10_sdma_mqd *m;
+	uint32_t sdma_rlc_reg_offset;
+	uint32_t sdma_rlc_rb_cntl;
+
+	m = get_sdma_mqd(mqd);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev, m->sdma_engine_id,
+					    m->sdma_queue_id);
+
+	sdma_rlc_rb_cntl = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL);
+
+	if (sdma_rlc_rb_cntl & SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK)
+		return true;
+
+	return false;
+}
+
+static int hqd_destroy_v10_3(struct amdgpu_device *adev, void *mqd,
+				enum kfd_preempt_type reset_type,
+				unsigned int utimeout, uint32_t pipe_id,
+				uint32_t queue_id, uint32_t inst)
+{
+	enum hqd_dequeue_request_type type;
+	unsigned long end_jiffies;
+	uint32_t temp;
+	struct v10_compute_mqd *m = get_mqd(mqd);
+
+	acquire_queue(adev, pipe_id, queue_id);
+
+	if (m->cp_hqd_vmid == 0)
+		WREG32_FIELD15(GC, 0, RLC_CP_SCHEDULERS, scheduler1, 0);
+
+	switch (reset_type) {
+	case KFD_PREEMPT_TYPE_WAVEFRONT_DRAIN:
+		type = DRAIN_PIPE;
+		break;
+	case KFD_PREEMPT_TYPE_WAVEFRONT_RESET:
+		type = RESET_WAVES;
+		break;
+	case KFD_PREEMPT_TYPE_WAVEFRONT_SAVE:
+		type = SAVE_WAVES;
+		break;
+	default:
+		type = DRAIN_PIPE;
+		break;
+	}
+
+	WREG32_SOC15(GC, 0, mmCP_HQD_DEQUEUE_REQUEST, type);
+
+	end_jiffies = (utimeout * HZ / 1000) + jiffies;
+	while (true) {
+		temp = RREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE);
+		if (!(temp & CP_HQD_ACTIVE__ACTIVE_MASK))
+			break;
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("cp queue pipe %d queue %d preemption failed\n",
+					pipe_id, queue_id);
+			release_queue(adev);
+			return -ETIME;
+		}
+		usleep_range(500, 1000);
+	}
+
+	release_queue(adev);
+	return 0;
+}
+
+static int hqd_sdma_destroy_v10_3(struct amdgpu_device *adev, void *mqd,
+				unsigned int utimeout)
+{
+	struct v10_sdma_mqd *m;
+	uint32_t sdma_rlc_reg_offset;
+	uint32_t temp;
+	unsigned long end_jiffies = (utimeout * HZ / 1000) + jiffies;
+
+	m = get_sdma_mqd(mqd);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev, m->sdma_engine_id,
+					    m->sdma_queue_id);
+
+	temp = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL);
+	temp = temp & ~SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK;
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL, temp);
+
+	while (true) {
+		temp = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_CONTEXT_STATUS);
+		if (temp & SDMA0_RLC0_CONTEXT_STATUS__IDLE_MASK)
+			break;
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("SDMA RLC not idle in %s\n", __func__);
+			return -ETIME;
+		}
+		usleep_range(500, 1000);
+	}
+
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_DOORBELL, 0);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL,
+		RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL) |
+		SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK);
+
+	m->sdmax_rlcx_rb_rptr = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR);
+	m->sdmax_rlcx_rb_rptr_hi =
+		RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR_HI);
+
+	return 0;
+}
+
+static int wave_control_execute_v10_3(struct amdgpu_device *adev,
+					uint32_t gfx_index_val,
+					uint32_t sq_cmd, uint32_t inst)
+{
+	uint32_t data = 0;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	WREG32_SOC15(GC, 0, mmGRBM_GFX_INDEX, gfx_index_val);
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSQ_CMD), sq_cmd);
+
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		INSTANCE_BROADCAST_WRITES, 1);
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		SA_BROADCAST_WRITES, 1);
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		SE_BROADCAST_WRITES, 1);
+
+	WREG32_SOC15(GC, 0, mmGRBM_GFX_INDEX, data);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+static bool get_atc_vmid_pasid_mapping_info_v10_3(struct amdgpu_device *adev,
+					uint8_t vmid, uint16_t *p_pasid)
+{
+	uint32_t value;
+
+	value = RREG32(SOC15_REG_OFFSET(ATHUB, 0, mmATC_VMID0_PASID_MAPPING)
+		     + vmid);
+	*p_pasid = value & ATC_VMID0_PASID_MAPPING__PASID_MASK;
+
+	return !!(value & ATC_VMID0_PASID_MAPPING__VALID_MASK);
+}
+
+static void set_vm_context_page_table_base_v10_3(struct amdgpu_device *adev,
+		uint32_t vmid, uint64_t page_table_base)
+{
+	/* SDMA is on gfxhub as well for Navi1* series */
+	adev->gfxhub.funcs->setup_vm_pt_regs(adev, vmid, page_table_base);
+}
+
+static void program_trap_handler_settings_v10_3(struct amdgpu_device *adev,
+			uint32_t vmid, uint64_t tba_addr, uint64_t tma_addr,
+			uint32_t inst)
+{
+	lock_srbm(adev, 0, 0, 0, vmid);
+
+	/*
+	 * Program TBA registers
+	 */
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSQ_SHADER_TBA_LO),
+			lower_32_bits(tba_addr >> 8));
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSQ_SHADER_TBA_HI),
+			upper_32_bits(tba_addr >> 8) |
+			(1 << SQ_SHADER_TBA_HI__TRAP_EN__SHIFT));
+
+	/*
+	 * Program TMA registers
+	 */
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSQ_SHADER_TMA_LO),
+			lower_32_bits(tma_addr >> 8));
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSQ_SHADER_TMA_HI),
+			 upper_32_bits(tma_addr >> 8));
+
+	unlock_srbm(adev);
+}
+
+const struct kfd2kgd_calls gfx_v10_3_kfd2kgd = {
+	.program_sh_mem_settings = program_sh_mem_settings_v10_3,
+	.set_pasid_vmid_mapping = set_pasid_vmid_mapping_v10_3,
+	.init_interrupts = init_interrupts_v10_3,
+	.hqd_load = hqd_load_v10_3,
+	.hiq_mqd_load = hiq_mqd_load_v10_3,
+	.hqd_sdma_load = hqd_sdma_load_v10_3,
+	.hqd_dump = hqd_dump_v10_3,
+	.hqd_sdma_dump = hqd_sdma_dump_v10_3,
+	.hqd_is_occupied = hqd_is_occupied_v10_3,
+	.hqd_sdma_is_occupied = hqd_sdma_is_occupied_v10_3,
+	.hqd_destroy = hqd_destroy_v10_3,
+	.hqd_sdma_destroy = hqd_sdma_destroy_v10_3,
+	.wave_control_execute = wave_control_execute_v10_3,
+	.get_atc_vmid_pasid_mapping_info = get_atc_vmid_pasid_mapping_info_v10_3,
+	.set_vm_context_page_table_base = set_vm_context_page_table_base_v10_3,
+	.program_trap_handler_settings = program_trap_handler_settings_v10_3,
+	.get_iq_wait_times = kgd_gfx_v10_get_iq_wait_times,
+	.build_dequeue_wait_counts_packet_info = kgd_gfx_v10_build_dequeue_wait_counts_packet_info,
+	.enable_debug_trap = kgd_gfx_v10_enable_debug_trap,
+	.disable_debug_trap = kgd_gfx_v10_disable_debug_trap,
+	.validate_trap_override_request = kgd_gfx_v10_validate_trap_override_request,
+	.set_wave_launch_trap_override = kgd_gfx_v10_set_wave_launch_trap_override,
+	.set_wave_launch_mode = kgd_gfx_v10_set_wave_launch_mode,
+	.set_address_watch = kgd_gfx_v10_set_address_watch,
+	.clear_address_watch = kgd_gfx_v10_clear_address_watch,
+	.hqd_get_pq_addr = kgd_gfx_v10_hqd_get_pq_addr,
+	.hqd_reset = kgd_gfx_v10_hqd_reset,
+	.hqd_sdma_get_doorbell = kgd_gfx_v10_hqd_sdma_get_doorbell
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v11.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v11.c
new file mode 100644
index 000000000000..aaccf0b9947d
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v11.c
@@ -0,0 +1,835 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+#include <linux/mmu_context.h>
+#include "amdgpu.h"
+#include "amdgpu_amdkfd.h"
+#include "gc/gc_11_0_0_offset.h"
+#include "gc/gc_11_0_0_sh_mask.h"
+#include "oss/osssys_6_0_0_offset.h"
+#include "oss/osssys_6_0_0_sh_mask.h"
+#include "soc15_common.h"
+#include "soc15d.h"
+#include "v11_structs.h"
+#include "soc21.h"
+#include <uapi/linux/kfd_ioctl.h>
+
+enum hqd_dequeue_request_type {
+	NO_ACTION = 0,
+	DRAIN_PIPE,
+	RESET_WAVES,
+	SAVE_WAVES
+};
+
+static void lock_srbm(struct amdgpu_device *adev, uint32_t mec, uint32_t pipe,
+			uint32_t queue, uint32_t vmid)
+{
+	mutex_lock(&adev->srbm_mutex);
+	soc21_grbm_select(adev, mec, pipe, queue, vmid);
+}
+
+static void unlock_srbm(struct amdgpu_device *adev)
+{
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static void acquire_queue(struct amdgpu_device *adev, uint32_t pipe_id,
+				uint32_t queue_id)
+{
+	uint32_t mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	uint32_t pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+	lock_srbm(adev, mec, pipe, queue_id, 0);
+}
+
+static uint64_t get_queue_mask(struct amdgpu_device *adev,
+			       uint32_t pipe_id, uint32_t queue_id)
+{
+	unsigned int bit = pipe_id * adev->gfx.mec.num_queue_per_pipe +
+			queue_id;
+
+	return 1ull << bit;
+}
+
+static void release_queue(struct amdgpu_device *adev)
+{
+	unlock_srbm(adev);
+}
+
+static void program_sh_mem_settings_v11(struct amdgpu_device *adev, uint32_t vmid,
+					uint32_t sh_mem_config,
+					uint32_t sh_mem_ape1_base,
+					uint32_t sh_mem_ape1_limit,
+					uint32_t sh_mem_bases, uint32_t inst)
+{
+	lock_srbm(adev, 0, 0, 0, vmid);
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, regSH_MEM_CONFIG), sh_mem_config);
+	WREG32(SOC15_REG_OFFSET(GC, 0, regSH_MEM_BASES), sh_mem_bases);
+
+	unlock_srbm(adev);
+}
+
+static int set_pasid_vmid_mapping_v11(struct amdgpu_device *adev, unsigned int pasid,
+					unsigned int vmid, uint32_t inst)
+{
+	uint32_t value = pasid << IH_VMID_0_LUT__PASID__SHIFT;
+
+	/* Mapping vmid to pasid also for IH block */
+	pr_debug("mapping vmid %d -> pasid %d in IH block for GFX client\n",
+			vmid, pasid);
+	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, regIH_VMID_0_LUT) + vmid, value);
+
+	return 0;
+}
+
+static int init_interrupts_v11(struct amdgpu_device *adev, uint32_t pipe_id,
+				uint32_t inst)
+{
+	uint32_t mec;
+	uint32_t pipe;
+
+	mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+	lock_srbm(adev, mec, pipe, 0, 0);
+
+	WREG32_SOC15(GC, 0, regCPC_INT_CNTL,
+		CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK |
+		CP_INT_CNTL_RING0__OPCODE_ERROR_INT_ENABLE_MASK);
+
+	unlock_srbm(adev);
+
+	return 0;
+}
+
+static uint32_t get_sdma_rlc_reg_offset(struct amdgpu_device *adev,
+				unsigned int engine_id,
+				unsigned int queue_id)
+{
+	uint32_t sdma_engine_reg_base = 0;
+	uint32_t sdma_rlc_reg_offset;
+
+	switch (engine_id) {
+	case 0:
+		sdma_engine_reg_base = SOC15_REG_OFFSET(SDMA0, 0,
+				regSDMA0_QUEUE0_RB_CNTL) - regSDMA0_QUEUE0_RB_CNTL;
+		break;
+	case 1:
+		sdma_engine_reg_base = SOC15_REG_OFFSET(SDMA1, 0,
+				regSDMA1_QUEUE0_RB_CNTL) - regSDMA0_QUEUE0_RB_CNTL;
+		break;
+	default:
+		BUG();
+	}
+
+	sdma_rlc_reg_offset = sdma_engine_reg_base
+		+ queue_id * (regSDMA0_QUEUE1_RB_CNTL - regSDMA0_QUEUE0_RB_CNTL);
+
+	pr_debug("RLC register offset for SDMA%d RLC%d: 0x%x\n", engine_id,
+			queue_id, sdma_rlc_reg_offset);
+
+	return sdma_rlc_reg_offset;
+}
+
+static inline struct v11_compute_mqd *get_mqd(void *mqd)
+{
+	return (struct v11_compute_mqd *)mqd;
+}
+
+static inline struct v11_sdma_mqd *get_sdma_mqd(void *mqd)
+{
+	return (struct v11_sdma_mqd *)mqd;
+}
+
+static int hqd_load_v11(struct amdgpu_device *adev, void *mqd, uint32_t pipe_id,
+			uint32_t queue_id, uint32_t __user *wptr,
+			uint32_t wptr_shift, uint32_t wptr_mask,
+			struct mm_struct *mm, uint32_t inst)
+{
+	struct v11_compute_mqd *m;
+	uint32_t *mqd_hqd;
+	uint32_t reg, hqd_base, data;
+
+	m = get_mqd(mqd);
+
+	pr_debug("Load hqd of pipe %d queue %d\n", pipe_id, queue_id);
+	acquire_queue(adev, pipe_id, queue_id);
+
+	/* HIQ is set during driver init period with vmid set to 0*/
+	if (m->cp_hqd_vmid == 0) {
+		uint32_t value, mec, pipe;
+
+		mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+		pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+		pr_debug("kfd: set HIQ, mec:%d, pipe:%d, queue:%d.\n",
+			mec, pipe, queue_id);
+		value = RREG32(SOC15_REG_OFFSET(GC, 0, regRLC_CP_SCHEDULERS));
+		value = REG_SET_FIELD(value, RLC_CP_SCHEDULERS, scheduler1,
+			((mec << 5) | (pipe << 3) | queue_id | 0x80));
+		WREG32(SOC15_REG_OFFSET(GC, 0, regRLC_CP_SCHEDULERS), value);
+	}
+
+	/* HQD registers extend from CP_MQD_BASE_ADDR to CP_HQD_EOP_WPTR_MEM. */
+	mqd_hqd = &m->cp_mqd_base_addr_lo;
+	hqd_base = SOC15_REG_OFFSET(GC, 0, regCP_MQD_BASE_ADDR);
+
+	for (reg = hqd_base;
+	     reg <= SOC15_REG_OFFSET(GC, 0, regCP_HQD_PQ_WPTR_HI); reg++)
+		WREG32(reg, mqd_hqd[reg - hqd_base]);
+
+
+	/* Activate doorbell logic before triggering WPTR poll. */
+	data = REG_SET_FIELD(m->cp_hqd_pq_doorbell_control,
+			     CP_HQD_PQ_DOORBELL_CONTROL, DOORBELL_EN, 1);
+	WREG32(SOC15_REG_OFFSET(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL), data);
+
+	if (wptr) {
+		/* Don't read wptr with get_user because the user
+		 * context may not be accessible (if this function
+		 * runs in a work queue). Instead trigger a one-shot
+		 * polling read from memory in the CP. This assumes
+		 * that wptr is GPU-accessible in the queue's VMID via
+		 * ATC or SVM. WPTR==RPTR before starting the poll so
+		 * the CP starts fetching new commands from the right
+		 * place.
+		 *
+		 * Guessing a 64-bit WPTR from a 32-bit RPTR is a bit
+		 * tricky. Assume that the queue didn't overflow. The
+		 * number of valid bits in the 32-bit RPTR depends on
+		 * the queue size. The remaining bits are taken from
+		 * the saved 64-bit WPTR. If the WPTR wrapped, add the
+		 * queue size.
+		 */
+		uint32_t queue_size =
+			2 << REG_GET_FIELD(m->cp_hqd_pq_control,
+					   CP_HQD_PQ_CONTROL, QUEUE_SIZE);
+		uint64_t guessed_wptr = m->cp_hqd_pq_rptr & (queue_size - 1);
+
+		if ((m->cp_hqd_pq_wptr_lo & (queue_size - 1)) < guessed_wptr)
+			guessed_wptr += queue_size;
+		guessed_wptr += m->cp_hqd_pq_wptr_lo & ~(queue_size - 1);
+		guessed_wptr += (uint64_t)m->cp_hqd_pq_wptr_hi << 32;
+
+		WREG32(SOC15_REG_OFFSET(GC, 0, regCP_HQD_PQ_WPTR_LO),
+		       lower_32_bits(guessed_wptr));
+		WREG32(SOC15_REG_OFFSET(GC, 0, regCP_HQD_PQ_WPTR_HI),
+		       upper_32_bits(guessed_wptr));
+		WREG32(SOC15_REG_OFFSET(GC, 0, regCP_HQD_PQ_WPTR_POLL_ADDR),
+		       lower_32_bits((uint64_t)wptr));
+		WREG32(SOC15_REG_OFFSET(GC, 0, regCP_HQD_PQ_WPTR_POLL_ADDR_HI),
+		       upper_32_bits((uint64_t)wptr));
+		pr_debug("%s setting CP_PQ_WPTR_POLL_CNTL1 to %x\n", __func__,
+			 (uint32_t)get_queue_mask(adev, pipe_id, queue_id));
+		WREG32(SOC15_REG_OFFSET(GC, 0, regCP_PQ_WPTR_POLL_CNTL1),
+		       (uint32_t)get_queue_mask(adev, pipe_id, queue_id));
+	}
+
+	/* Start the EOP fetcher */
+	WREG32(SOC15_REG_OFFSET(GC, 0, regCP_HQD_EOP_RPTR),
+	       REG_SET_FIELD(m->cp_hqd_eop_rptr,
+			     CP_HQD_EOP_RPTR, INIT_FETCHER, 1));
+
+	data = REG_SET_FIELD(m->cp_hqd_active, CP_HQD_ACTIVE, ACTIVE, 1);
+	WREG32(SOC15_REG_OFFSET(GC, 0, regCP_HQD_ACTIVE), data);
+
+	release_queue(adev);
+
+	return 0;
+}
+
+static int hiq_mqd_load_v11(struct amdgpu_device *adev, void *mqd,
+			      uint32_t pipe_id, uint32_t queue_id,
+			      uint32_t doorbell_off, uint32_t inst)
+{
+	struct amdgpu_ring *kiq_ring = &adev->gfx.kiq[0].ring;
+	struct v11_compute_mqd *m;
+	uint32_t mec, pipe;
+	int r;
+
+	m = get_mqd(mqd);
+
+	acquire_queue(adev, pipe_id, queue_id);
+
+	mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+	pr_debug("kfd: set HIQ, mec:%d, pipe:%d, queue:%d.\n",
+		 mec, pipe, queue_id);
+
+	spin_lock(&adev->gfx.kiq[0].ring_lock);
+	r = amdgpu_ring_alloc(kiq_ring, 7);
+	if (r) {
+		pr_err("Failed to alloc KIQ (%d).\n", r);
+		goto out_unlock;
+	}
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_MAP_QUEUES, 5));
+	amdgpu_ring_write(kiq_ring,
+			  PACKET3_MAP_QUEUES_QUEUE_SEL(0) | /* Queue_Sel */
+			  PACKET3_MAP_QUEUES_VMID(m->cp_hqd_vmid) | /* VMID */
+			  PACKET3_MAP_QUEUES_QUEUE(queue_id) |
+			  PACKET3_MAP_QUEUES_PIPE(pipe) |
+			  PACKET3_MAP_QUEUES_ME((mec - 1)) |
+			  PACKET3_MAP_QUEUES_QUEUE_TYPE(0) | /*queue_type: normal compute queue */
+			  PACKET3_MAP_QUEUES_ALLOC_FORMAT(0) | /* alloc format: all_on_one_pipe */
+			  PACKET3_MAP_QUEUES_ENGINE_SEL(1) | /* engine_sel: hiq */
+			  PACKET3_MAP_QUEUES_NUM_QUEUES(1)); /* num_queues: must be 1 */
+	amdgpu_ring_write(kiq_ring,
+			PACKET3_MAP_QUEUES_DOORBELL_OFFSET(doorbell_off));
+	amdgpu_ring_write(kiq_ring, m->cp_mqd_base_addr_lo);
+	amdgpu_ring_write(kiq_ring, m->cp_mqd_base_addr_hi);
+	amdgpu_ring_write(kiq_ring, m->cp_hqd_pq_wptr_poll_addr_lo);
+	amdgpu_ring_write(kiq_ring, m->cp_hqd_pq_wptr_poll_addr_hi);
+	amdgpu_ring_commit(kiq_ring);
+
+out_unlock:
+	spin_unlock(&adev->gfx.kiq[0].ring_lock);
+	release_queue(adev);
+
+	return r;
+}
+
+static int hqd_dump_v11(struct amdgpu_device *adev,
+			uint32_t pipe_id, uint32_t queue_id,
+			uint32_t (**dump)[2], uint32_t *n_regs, uint32_t inst)
+{
+	uint32_t i = 0, reg;
+#define HQD_N_REGS 56
+#define DUMP_REG(addr) do {				\
+		if (WARN_ON_ONCE(i >= HQD_N_REGS))	\
+			break;				\
+		(*dump)[i][0] = (addr) << 2;		\
+		(*dump)[i++][1] = RREG32(addr);		\
+	} while (0)
+
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
+	if (*dump == NULL)
+		return -ENOMEM;
+
+	acquire_queue(adev, pipe_id, queue_id);
+
+	for (reg = SOC15_REG_OFFSET(GC, 0, regCP_MQD_BASE_ADDR);
+	     reg <= SOC15_REG_OFFSET(GC, 0, regCP_HQD_PQ_WPTR_HI); reg++)
+		DUMP_REG(reg);
+
+	release_queue(adev);
+
+	WARN_ON_ONCE(i != HQD_N_REGS);
+	*n_regs = i;
+
+	return 0;
+}
+
+static int hqd_sdma_load_v11(struct amdgpu_device *adev, void *mqd,
+			     uint32_t __user *wptr, struct mm_struct *mm)
+{
+	struct v11_sdma_mqd *m;
+	uint32_t sdma_rlc_reg_offset;
+	unsigned long end_jiffies;
+	uint32_t data;
+	uint64_t data64;
+	uint64_t __user *wptr64 = (uint64_t __user *)wptr;
+
+	m = get_sdma_mqd(mqd);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev, m->sdma_engine_id,
+					    m->sdma_queue_id);
+
+	WREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_RB_CNTL,
+		m->sdmax_rlcx_rb_cntl & (~SDMA0_QUEUE0_RB_CNTL__RB_ENABLE_MASK));
+
+	end_jiffies = msecs_to_jiffies(2000) + jiffies;
+	while (true) {
+		data = RREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_CONTEXT_STATUS);
+		if (data & SDMA0_QUEUE0_CONTEXT_STATUS__IDLE_MASK)
+			break;
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("SDMA RLC not idle in %s\n", __func__);
+			return -ETIME;
+		}
+		usleep_range(500, 1000);
+	}
+
+	WREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_DOORBELL_OFFSET,
+	       m->sdmax_rlcx_doorbell_offset);
+
+	data = REG_SET_FIELD(m->sdmax_rlcx_doorbell, SDMA0_QUEUE0_DOORBELL,
+			     ENABLE, 1);
+	WREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_DOORBELL, data);
+	WREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_RB_RPTR,
+				m->sdmax_rlcx_rb_rptr);
+	WREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_RB_RPTR_HI,
+				m->sdmax_rlcx_rb_rptr_hi);
+
+	WREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_MINOR_PTR_UPDATE, 1);
+	if (read_user_wptr(mm, wptr64, data64)) {
+		WREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_RB_WPTR,
+		       lower_32_bits(data64));
+		WREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_RB_WPTR_HI,
+		       upper_32_bits(data64));
+	} else {
+		WREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_RB_WPTR,
+		       m->sdmax_rlcx_rb_rptr);
+		WREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_RB_WPTR_HI,
+		       m->sdmax_rlcx_rb_rptr_hi);
+	}
+	WREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_MINOR_PTR_UPDATE, 0);
+
+	WREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_RB_BASE, m->sdmax_rlcx_rb_base);
+	WREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_RB_BASE_HI,
+			m->sdmax_rlcx_rb_base_hi);
+	WREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_RB_RPTR_ADDR_LO,
+			m->sdmax_rlcx_rb_rptr_addr_lo);
+	WREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_RB_RPTR_ADDR_HI,
+			m->sdmax_rlcx_rb_rptr_addr_hi);
+
+	data = REG_SET_FIELD(m->sdmax_rlcx_rb_cntl, SDMA0_QUEUE0_RB_CNTL,
+			     RB_ENABLE, 1);
+	WREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_RB_CNTL, data);
+
+	return 0;
+}
+
+static int hqd_sdma_dump_v11(struct amdgpu_device *adev,
+			     uint32_t engine_id, uint32_t queue_id,
+			     uint32_t (**dump)[2], uint32_t *n_regs)
+{
+	uint32_t sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev,
+			engine_id, queue_id);
+	uint32_t i = 0, reg;
+#undef HQD_N_REGS
+#define HQD_N_REGS (7+11+1+12+12)
+
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
+	if (*dump == NULL)
+		return -ENOMEM;
+
+	for (reg = regSDMA0_QUEUE0_RB_CNTL;
+	     reg <= regSDMA0_QUEUE0_RB_WPTR_HI; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+	for (reg = regSDMA0_QUEUE0_RB_RPTR_ADDR_HI;
+	     reg <= regSDMA0_QUEUE0_DOORBELL; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+	for (reg = regSDMA0_QUEUE0_DOORBELL_LOG;
+	     reg <= regSDMA0_QUEUE0_DOORBELL_LOG; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+	for (reg = regSDMA0_QUEUE0_DOORBELL_OFFSET;
+	     reg <= regSDMA0_QUEUE0_RB_PREEMPT; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+	for (reg = regSDMA0_QUEUE0_MIDCMD_DATA0;
+	     reg <= regSDMA0_QUEUE0_MIDCMD_CNTL; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+
+	WARN_ON_ONCE(i != HQD_N_REGS);
+	*n_regs = i;
+
+	return 0;
+}
+
+static bool hqd_is_occupied_v11(struct amdgpu_device *adev, uint64_t queue_address,
+				uint32_t pipe_id, uint32_t queue_id, uint32_t inst)
+{
+	uint32_t act;
+	bool retval = false;
+	uint32_t low, high;
+
+	acquire_queue(adev, pipe_id, queue_id);
+	act = RREG32(SOC15_REG_OFFSET(GC, 0, regCP_HQD_ACTIVE));
+	if (act) {
+		low = lower_32_bits(queue_address >> 8);
+		high = upper_32_bits(queue_address >> 8);
+
+		if (low == RREG32(SOC15_REG_OFFSET(GC, 0, regCP_HQD_PQ_BASE)) &&
+		   high == RREG32(SOC15_REG_OFFSET(GC, 0, regCP_HQD_PQ_BASE_HI)))
+			retval = true;
+	}
+	release_queue(adev);
+	return retval;
+}
+
+static bool hqd_sdma_is_occupied_v11(struct amdgpu_device *adev, void *mqd)
+{
+	struct v11_sdma_mqd *m;
+	uint32_t sdma_rlc_reg_offset;
+	uint32_t sdma_rlc_rb_cntl;
+
+	m = get_sdma_mqd(mqd);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev, m->sdma_engine_id,
+					    m->sdma_queue_id);
+
+	sdma_rlc_rb_cntl = RREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_RB_CNTL);
+
+	if (sdma_rlc_rb_cntl & SDMA0_QUEUE0_RB_CNTL__RB_ENABLE_MASK)
+		return true;
+
+	return false;
+}
+
+static int hqd_destroy_v11(struct amdgpu_device *adev, void *mqd,
+				enum kfd_preempt_type reset_type,
+				unsigned int utimeout, uint32_t pipe_id,
+				uint32_t queue_id, uint32_t inst)
+{
+	enum hqd_dequeue_request_type type;
+	unsigned long end_jiffies;
+	uint32_t temp;
+	struct v11_compute_mqd *m = get_mqd(mqd);
+
+	acquire_queue(adev, pipe_id, queue_id);
+
+	if (m->cp_hqd_vmid == 0)
+		WREG32_FIELD15_PREREG(GC, 0, RLC_CP_SCHEDULERS, scheduler1, 0);
+
+	switch (reset_type) {
+	case KFD_PREEMPT_TYPE_WAVEFRONT_DRAIN:
+		type = DRAIN_PIPE;
+		break;
+	case KFD_PREEMPT_TYPE_WAVEFRONT_RESET:
+		type = RESET_WAVES;
+		break;
+	default:
+		type = DRAIN_PIPE;
+		break;
+	}
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, regCP_HQD_DEQUEUE_REQUEST), type);
+
+	end_jiffies = (utimeout * HZ / 1000) + jiffies;
+	while (true) {
+		temp = RREG32(SOC15_REG_OFFSET(GC, 0, regCP_HQD_ACTIVE));
+		if (!(temp & CP_HQD_ACTIVE__ACTIVE_MASK))
+			break;
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("cp queue pipe %d queue %d preemption failed\n",
+					pipe_id, queue_id);
+			release_queue(adev);
+			return -ETIME;
+		}
+		usleep_range(500, 1000);
+	}
+
+	release_queue(adev);
+	return 0;
+}
+
+static int hqd_sdma_destroy_v11(struct amdgpu_device *adev, void *mqd,
+				unsigned int utimeout)
+{
+	struct v11_sdma_mqd *m;
+	uint32_t sdma_rlc_reg_offset;
+	uint32_t temp;
+	unsigned long end_jiffies = (utimeout * HZ / 1000) + jiffies;
+
+	m = get_sdma_mqd(mqd);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev, m->sdma_engine_id,
+					    m->sdma_queue_id);
+
+	temp = RREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_RB_CNTL);
+	temp = temp & ~SDMA0_QUEUE0_RB_CNTL__RB_ENABLE_MASK;
+	WREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_RB_CNTL, temp);
+
+	while (true) {
+		temp = RREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_CONTEXT_STATUS);
+		if (temp & SDMA0_QUEUE0_CONTEXT_STATUS__IDLE_MASK)
+			break;
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("SDMA RLC not idle in %s\n", __func__);
+			return -ETIME;
+		}
+		usleep_range(500, 1000);
+	}
+
+	WREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_DOORBELL, 0);
+	WREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_RB_CNTL,
+		RREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_RB_CNTL) |
+		SDMA0_QUEUE0_RB_CNTL__RB_ENABLE_MASK);
+
+	m->sdmax_rlcx_rb_rptr = RREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_RB_RPTR);
+	m->sdmax_rlcx_rb_rptr_hi =
+		RREG32(sdma_rlc_reg_offset + regSDMA0_QUEUE0_RB_RPTR_HI);
+
+	return 0;
+}
+
+static int wave_control_execute_v11(struct amdgpu_device *adev,
+					uint32_t gfx_index_val,
+					uint32_t sq_cmd, uint32_t inst)
+{
+	uint32_t data = 0;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, regGRBM_GFX_INDEX), gfx_index_val);
+	WREG32(SOC15_REG_OFFSET(GC, 0, regSQ_CMD), sq_cmd);
+
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		INSTANCE_BROADCAST_WRITES, 1);
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		SA_BROADCAST_WRITES, 1);
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		SE_BROADCAST_WRITES, 1);
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, regGRBM_GFX_INDEX), data);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+static void set_vm_context_page_table_base_v11(struct amdgpu_device *adev,
+		uint32_t vmid, uint64_t page_table_base)
+{
+	if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid)) {
+		pr_err("trying to set page table base for wrong VMID %u\n",
+		       vmid);
+		return;
+	}
+
+	/* SDMA is on gfxhub as well for gfx11 adapters */
+	adev->gfxhub.funcs->setup_vm_pt_regs(adev, vmid, page_table_base);
+}
+
+/*
+ * Returns TRAP_EN, EXCP_EN and EXCP_REPLACE.
+ *
+ * restore_dbg_registers is ignored here but is a general interface requirement
+ * for devices that support GFXOFF and where the RLC save/restore list
+ * does not support hw registers for debugging i.e. the driver has to manually
+ * initialize the debug mode registers after it has disabled GFX off during the
+ * debug session.
+ */
+static uint32_t kgd_gfx_v11_enable_debug_trap(struct amdgpu_device *adev,
+					    bool restore_dbg_registers,
+					    uint32_t vmid)
+{
+	uint32_t data = 0;
+
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, TRAP_EN, 1);
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_EN, 0);
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_REPLACE, 0);
+
+	return data;
+}
+
+/* Returns TRAP_EN, EXCP_EN and EXCP_REPLACE. */
+static uint32_t kgd_gfx_v11_disable_debug_trap(struct amdgpu_device *adev,
+						bool keep_trap_enabled,
+						uint32_t vmid)
+{
+	uint32_t data = 0;
+
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, TRAP_EN, 1);
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_EN, 0);
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_REPLACE, 0);
+
+	return data;
+}
+
+static int kgd_gfx_v11_validate_trap_override_request(struct amdgpu_device *adev,
+							uint32_t trap_override,
+							uint32_t *trap_mask_supported)
+{
+	*trap_mask_supported &= KFD_DBG_TRAP_MASK_FP_INVALID |
+				KFD_DBG_TRAP_MASK_FP_INPUT_DENORMAL |
+				KFD_DBG_TRAP_MASK_FP_DIVIDE_BY_ZERO |
+				KFD_DBG_TRAP_MASK_FP_OVERFLOW |
+				KFD_DBG_TRAP_MASK_FP_UNDERFLOW |
+				KFD_DBG_TRAP_MASK_FP_INEXACT |
+				KFD_DBG_TRAP_MASK_INT_DIVIDE_BY_ZERO |
+				KFD_DBG_TRAP_MASK_DBG_ADDRESS_WATCH |
+				KFD_DBG_TRAP_MASK_DBG_MEMORY_VIOLATION;
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(11, 0, 4))
+		*trap_mask_supported |= KFD_DBG_TRAP_MASK_TRAP_ON_WAVE_START |
+					KFD_DBG_TRAP_MASK_TRAP_ON_WAVE_END;
+
+	if (trap_override != KFD_DBG_TRAP_OVERRIDE_OR &&
+			trap_override != KFD_DBG_TRAP_OVERRIDE_REPLACE)
+		return -EPERM;
+
+	return 0;
+}
+
+static uint32_t trap_mask_map_sw_to_hw(uint32_t mask)
+{
+	uint32_t trap_on_start = (mask & KFD_DBG_TRAP_MASK_TRAP_ON_WAVE_START) ? 1 : 0;
+	uint32_t trap_on_end = (mask & KFD_DBG_TRAP_MASK_TRAP_ON_WAVE_END) ? 1 : 0;
+	uint32_t excp_en = mask & (KFD_DBG_TRAP_MASK_FP_INVALID |
+			KFD_DBG_TRAP_MASK_FP_INPUT_DENORMAL |
+			KFD_DBG_TRAP_MASK_FP_DIVIDE_BY_ZERO |
+			KFD_DBG_TRAP_MASK_FP_OVERFLOW |
+			KFD_DBG_TRAP_MASK_FP_UNDERFLOW |
+			KFD_DBG_TRAP_MASK_FP_INEXACT |
+			KFD_DBG_TRAP_MASK_INT_DIVIDE_BY_ZERO |
+			KFD_DBG_TRAP_MASK_DBG_ADDRESS_WATCH |
+			KFD_DBG_TRAP_MASK_DBG_MEMORY_VIOLATION);
+	uint32_t ret;
+
+	ret = REG_SET_FIELD(0, SPI_GDBG_PER_VMID_CNTL, EXCP_EN, excp_en);
+	ret = REG_SET_FIELD(ret, SPI_GDBG_PER_VMID_CNTL, TRAP_ON_START, trap_on_start);
+	ret = REG_SET_FIELD(ret, SPI_GDBG_PER_VMID_CNTL, TRAP_ON_END, trap_on_end);
+
+	return ret;
+}
+
+static uint32_t trap_mask_map_hw_to_sw(uint32_t mask)
+{
+	uint32_t ret = REG_GET_FIELD(mask, SPI_GDBG_PER_VMID_CNTL, EXCP_EN);
+
+	if (REG_GET_FIELD(mask, SPI_GDBG_PER_VMID_CNTL, TRAP_ON_START))
+		ret |= KFD_DBG_TRAP_MASK_TRAP_ON_WAVE_START;
+
+	if (REG_GET_FIELD(mask, SPI_GDBG_PER_VMID_CNTL, TRAP_ON_END))
+		ret |= KFD_DBG_TRAP_MASK_TRAP_ON_WAVE_END;
+
+	return ret;
+}
+
+/* Returns TRAP_EN, EXCP_EN and EXCP_REPLACE. */
+static uint32_t kgd_gfx_v11_set_wave_launch_trap_override(struct amdgpu_device *adev,
+					uint32_t vmid,
+					uint32_t trap_override,
+					uint32_t trap_mask_bits,
+					uint32_t trap_mask_request,
+					uint32_t *trap_mask_prev,
+					uint32_t kfd_dbg_trap_cntl_prev)
+{
+	uint32_t data = 0;
+
+	*trap_mask_prev = trap_mask_map_hw_to_sw(kfd_dbg_trap_cntl_prev);
+
+	data = (trap_mask_bits & trap_mask_request) | (*trap_mask_prev & ~trap_mask_request);
+	data = trap_mask_map_sw_to_hw(data);
+
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, TRAP_EN, 1);
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_REPLACE, trap_override);
+
+	return data;
+}
+
+static uint32_t kgd_gfx_v11_set_wave_launch_mode(struct amdgpu_device *adev,
+					uint8_t wave_launch_mode,
+					uint32_t vmid)
+{
+	uint32_t data = 0;
+
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, LAUNCH_MODE, wave_launch_mode);
+
+	return data;
+}
+
+#define TCP_WATCH_STRIDE (regTCP_WATCH1_ADDR_H - regTCP_WATCH0_ADDR_H)
+static uint32_t kgd_gfx_v11_set_address_watch(struct amdgpu_device *adev,
+					uint64_t watch_address,
+					uint32_t watch_address_mask,
+					uint32_t watch_id,
+					uint32_t watch_mode,
+					uint32_t debug_vmid,
+					uint32_t inst)
+{
+	uint32_t watch_address_high;
+	uint32_t watch_address_low;
+	uint32_t watch_address_cntl;
+
+	watch_address_cntl = 0;
+	watch_address_low = lower_32_bits(watch_address);
+	watch_address_high = upper_32_bits(watch_address) & 0xffff;
+
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			MODE,
+			watch_mode);
+
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			MASK,
+			watch_address_mask >> 7);
+
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			VALID,
+			1);
+
+	WREG32_RLC((SOC15_REG_OFFSET(GC, 0, regTCP_WATCH0_ADDR_H) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_high);
+
+	WREG32_RLC((SOC15_REG_OFFSET(GC, 0, regTCP_WATCH0_ADDR_L) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_low);
+
+	return watch_address_cntl;
+}
+
+static uint32_t kgd_gfx_v11_clear_address_watch(struct amdgpu_device *adev,
+						uint32_t watch_id)
+{
+	return 0;
+}
+
+static uint64_t kgd_gfx_v11_hqd_get_pq_addr(struct amdgpu_device *adev,
+					    uint32_t pipe_id, uint32_t queue_id,
+					    uint32_t inst)
+{
+	return 0;
+}
+
+static uint64_t kgd_gfx_v11_hqd_reset(struct amdgpu_device *adev,
+				      uint32_t pipe_id, uint32_t queue_id,
+				      uint32_t inst, unsigned int utimeout)
+{
+	return 0;
+}
+
+static uint32_t kgd_gfx_v11_hqd_sdma_get_doorbell(struct amdgpu_device *adev,
+						  int engine, int queue)
+{
+	return 0;
+}
+
+const struct kfd2kgd_calls gfx_v11_kfd2kgd = {
+	.program_sh_mem_settings = program_sh_mem_settings_v11,
+	.set_pasid_vmid_mapping = set_pasid_vmid_mapping_v11,
+	.init_interrupts = init_interrupts_v11,
+	.hqd_load = hqd_load_v11,
+	.hiq_mqd_load = hiq_mqd_load_v11,
+	.hqd_sdma_load = hqd_sdma_load_v11,
+	.hqd_dump = hqd_dump_v11,
+	.hqd_sdma_dump = hqd_sdma_dump_v11,
+	.hqd_is_occupied = hqd_is_occupied_v11,
+	.hqd_sdma_is_occupied = hqd_sdma_is_occupied_v11,
+	.hqd_destroy = hqd_destroy_v11,
+	.hqd_sdma_destroy = hqd_sdma_destroy_v11,
+	.wave_control_execute = wave_control_execute_v11,
+	.get_atc_vmid_pasid_mapping_info = NULL,
+	.set_vm_context_page_table_base = set_vm_context_page_table_base_v11,
+	.enable_debug_trap = kgd_gfx_v11_enable_debug_trap,
+	.disable_debug_trap = kgd_gfx_v11_disable_debug_trap,
+	.validate_trap_override_request = kgd_gfx_v11_validate_trap_override_request,
+	.set_wave_launch_trap_override = kgd_gfx_v11_set_wave_launch_trap_override,
+	.set_wave_launch_mode = kgd_gfx_v11_set_wave_launch_mode,
+	.set_address_watch = kgd_gfx_v11_set_address_watch,
+	.clear_address_watch = kgd_gfx_v11_clear_address_watch,
+	.hqd_get_pq_addr = kgd_gfx_v11_hqd_get_pq_addr,
+	.hqd_reset = kgd_gfx_v11_hqd_reset,
+	.hqd_sdma_get_doorbell = kgd_gfx_v11_hqd_sdma_get_doorbell
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v12.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v12.c
new file mode 100644
index 000000000000..e0ceab400b2d
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v12.c
@@ -0,0 +1,384 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include "amdgpu.h"
+#include "amdgpu_amdkfd.h"
+#include "gc/gc_12_0_0_offset.h"
+#include "gc/gc_12_0_0_sh_mask.h"
+#include "soc24.h"
+#include <uapi/linux/kfd_ioctl.h>
+
+static void lock_srbm(struct amdgpu_device *adev, uint32_t mec, uint32_t pipe,
+			uint32_t queue, uint32_t vmid)
+{
+	mutex_lock(&adev->srbm_mutex);
+	soc24_grbm_select(adev, mec, pipe, queue, vmid);
+}
+
+static void unlock_srbm(struct amdgpu_device *adev)
+{
+	soc24_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static void acquire_queue(struct amdgpu_device *adev, uint32_t pipe_id,
+				uint32_t queue_id)
+{
+	uint32_t mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	uint32_t pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+	lock_srbm(adev, mec, pipe, queue_id, 0);
+}
+
+static void release_queue(struct amdgpu_device *adev)
+{
+	unlock_srbm(adev);
+}
+
+static int init_interrupts_v12(struct amdgpu_device *adev, uint32_t pipe_id, uint32_t inst)
+{
+	uint32_t mec;
+	uint32_t pipe;
+
+	mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+	lock_srbm(adev, mec, pipe, 0, 0);
+
+	WREG32_SOC15(GC, 0, regCPC_INT_CNTL,
+		CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK |
+		CP_INT_CNTL_RING0__OPCODE_ERROR_INT_ENABLE_MASK);
+
+	unlock_srbm(adev);
+
+	return 0;
+}
+
+static uint32_t get_sdma_rlc_reg_offset(struct amdgpu_device *adev,
+				unsigned int engine_id,
+				unsigned int queue_id)
+{
+	uint32_t sdma_engine_reg_base = 0;
+	uint32_t sdma_rlc_reg_offset;
+
+	switch (engine_id) {
+	case 0:
+		sdma_engine_reg_base = SOC15_REG_OFFSET(SDMA0, 0,
+				regSDMA0_QUEUE0_RB_CNTL) - regSDMA0_QUEUE0_RB_CNTL;
+		break;
+	case 1:
+		sdma_engine_reg_base = SOC15_REG_OFFSET(SDMA1, 0,
+				regSDMA1_QUEUE0_RB_CNTL) - regSDMA0_QUEUE0_RB_CNTL;
+		break;
+	default:
+		BUG();
+	}
+
+	sdma_rlc_reg_offset = sdma_engine_reg_base
+		+ queue_id * (regSDMA0_QUEUE1_RB_CNTL - regSDMA0_QUEUE0_RB_CNTL);
+
+	pr_debug("RLC register offset for SDMA%d RLC%d: 0x%x\n", engine_id,
+			queue_id, sdma_rlc_reg_offset);
+
+	return sdma_rlc_reg_offset;
+}
+
+static int hqd_dump_v12(struct amdgpu_device *adev,
+			uint32_t pipe_id, uint32_t queue_id,
+			uint32_t (**dump)[2], uint32_t *n_regs, uint32_t inst)
+{
+	uint32_t i = 0, reg;
+#define HQD_N_REGS 56
+#define DUMP_REG(addr) do {				\
+		if (WARN_ON_ONCE(i >= HQD_N_REGS))	\
+			break;				\
+		(*dump)[i][0] = (addr) << 2;		\
+		(*dump)[i++][1] = RREG32(addr);		\
+	} while (0)
+
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
+	if (*dump == NULL)
+		return -ENOMEM;
+
+	acquire_queue(adev, pipe_id, queue_id);
+
+	for (reg = SOC15_REG_OFFSET(GC, 0, regCP_MQD_BASE_ADDR);
+	     reg <= SOC15_REG_OFFSET(GC, 0, regCP_HQD_PQ_WPTR_HI); reg++)
+		DUMP_REG(reg);
+
+	release_queue(adev);
+
+	WARN_ON_ONCE(i != HQD_N_REGS);
+	*n_regs = i;
+
+	return 0;
+}
+
+static int hqd_sdma_dump_v12(struct amdgpu_device *adev,
+			     uint32_t engine_id, uint32_t queue_id,
+			     uint32_t (**dump)[2], uint32_t *n_regs)
+{
+	uint32_t sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev,
+			engine_id, queue_id);
+	uint32_t i = 0, reg;
+
+	const uint32_t first_reg = regSDMA0_QUEUE0_RB_CNTL;
+	const uint32_t last_reg = regSDMA0_QUEUE0_CONTEXT_STATUS;
+#undef HQD_N_REGS
+#define HQD_N_REGS (last_reg - first_reg + 1)
+
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
+	if (*dump == NULL)
+		return -ENOMEM;
+
+	for (reg = first_reg;
+	     reg <= last_reg; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+
+	WARN_ON_ONCE(i != HQD_N_REGS);
+	*n_regs = i;
+
+	return 0;
+}
+
+static int wave_control_execute_v12(struct amdgpu_device *adev,
+					uint32_t gfx_index_val,
+					uint32_t sq_cmd, uint32_t inst)
+{
+	uint32_t data = 0;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, regGRBM_GFX_INDEX), gfx_index_val);
+	WREG32(SOC15_REG_OFFSET(GC, 0, regSQ_CMD), sq_cmd);
+
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		INSTANCE_BROADCAST_WRITES, 1);
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		SA_BROADCAST_WRITES, 1);
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		SE_BROADCAST_WRITES, 1);
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, regGRBM_GFX_INDEX), data);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+/* returns TRAP_EN, EXCP_EN and EXCP_REPLACE. */
+static uint32_t kgd_gfx_v12_enable_debug_trap(struct amdgpu_device *adev,
+					    bool restore_dbg_registers,
+					    uint32_t vmid)
+{
+	uint32_t data = 0;
+
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, TRAP_EN, 1);
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_EN, 0);
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_REPLACE, 0);
+
+	return data;
+}
+
+/* returns TRAP_EN, EXCP_EN and EXCP_REPLACE. */
+static uint32_t kgd_gfx_v12_disable_debug_trap(struct amdgpu_device *adev,
+						bool keep_trap_enabled,
+						uint32_t vmid)
+{
+	uint32_t data = 0;
+
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, TRAP_EN, 1);
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_EN, 0);
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_REPLACE, 0);
+
+	return data;
+}
+
+static int kgd_gfx_v12_validate_trap_override_request(struct amdgpu_device *adev,
+							uint32_t trap_override,
+							uint32_t *trap_mask_supported)
+{
+	*trap_mask_supported &= KFD_DBG_TRAP_MASK_FP_INVALID |
+				KFD_DBG_TRAP_MASK_FP_INPUT_DENORMAL |
+				KFD_DBG_TRAP_MASK_FP_DIVIDE_BY_ZERO |
+				KFD_DBG_TRAP_MASK_FP_OVERFLOW |
+				KFD_DBG_TRAP_MASK_FP_UNDERFLOW |
+				KFD_DBG_TRAP_MASK_FP_INEXACT |
+				KFD_DBG_TRAP_MASK_INT_DIVIDE_BY_ZERO |
+				KFD_DBG_TRAP_MASK_DBG_ADDRESS_WATCH |
+				KFD_DBG_TRAP_MASK_DBG_MEMORY_VIOLATION |
+				KFD_DBG_TRAP_MASK_TRAP_ON_WAVE_START |
+				KFD_DBG_TRAP_MASK_TRAP_ON_WAVE_END;
+
+
+	if (trap_override != KFD_DBG_TRAP_OVERRIDE_OR &&
+			trap_override != KFD_DBG_TRAP_OVERRIDE_REPLACE)
+		return -EPERM;
+
+	return 0;
+}
+
+static uint32_t trap_mask_map_sw_to_hw(uint32_t mask)
+{
+	uint32_t trap_on_start = (mask & KFD_DBG_TRAP_MASK_TRAP_ON_WAVE_START) ? 1 : 0;
+	uint32_t trap_on_end = (mask & KFD_DBG_TRAP_MASK_TRAP_ON_WAVE_END) ? 1 : 0;
+	uint32_t excp_en = mask & (KFD_DBG_TRAP_MASK_FP_INVALID |
+			KFD_DBG_TRAP_MASK_FP_INPUT_DENORMAL |
+			KFD_DBG_TRAP_MASK_FP_DIVIDE_BY_ZERO |
+			KFD_DBG_TRAP_MASK_FP_OVERFLOW |
+			KFD_DBG_TRAP_MASK_FP_UNDERFLOW |
+			KFD_DBG_TRAP_MASK_FP_INEXACT |
+			KFD_DBG_TRAP_MASK_INT_DIVIDE_BY_ZERO |
+			KFD_DBG_TRAP_MASK_DBG_ADDRESS_WATCH |
+			KFD_DBG_TRAP_MASK_DBG_MEMORY_VIOLATION);
+	uint32_t ret;
+
+	ret = REG_SET_FIELD(0, SPI_GDBG_PER_VMID_CNTL, EXCP_EN, excp_en);
+	ret = REG_SET_FIELD(ret, SPI_GDBG_PER_VMID_CNTL, TRAP_ON_START, trap_on_start);
+	ret = REG_SET_FIELD(ret, SPI_GDBG_PER_VMID_CNTL, TRAP_ON_END, trap_on_end);
+
+	return ret;
+}
+
+static uint32_t trap_mask_map_hw_to_sw(uint32_t mask)
+{
+	uint32_t ret = REG_GET_FIELD(mask, SPI_GDBG_PER_VMID_CNTL, EXCP_EN);
+
+	if (REG_GET_FIELD(mask, SPI_GDBG_PER_VMID_CNTL, TRAP_ON_START))
+		ret |= KFD_DBG_TRAP_MASK_TRAP_ON_WAVE_START;
+
+	if (REG_GET_FIELD(mask, SPI_GDBG_PER_VMID_CNTL, TRAP_ON_END))
+		ret |= KFD_DBG_TRAP_MASK_TRAP_ON_WAVE_END;
+
+	return ret;
+}
+
+/* returns TRAP_EN, EXCP_EN and EXCP_REPLACE. */
+static uint32_t kgd_gfx_v12_set_wave_launch_trap_override(struct amdgpu_device *adev,
+					uint32_t vmid,
+					uint32_t trap_override,
+					uint32_t trap_mask_bits,
+					uint32_t trap_mask_request,
+					uint32_t *trap_mask_prev,
+					uint32_t kfd_dbg_trap_cntl_prev)
+
+{
+	uint32_t data = 0;
+
+	*trap_mask_prev = trap_mask_map_hw_to_sw(kfd_dbg_trap_cntl_prev);
+
+	data = (trap_mask_bits & trap_mask_request) | (*trap_mask_prev & ~trap_mask_request);
+	data = trap_mask_map_sw_to_hw(data);
+
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, TRAP_EN, 1);
+	data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_REPLACE, trap_override);
+
+	return data;
+}
+
+/* returns STALL_VMID or LAUNCH_MODE. */
+static uint32_t kgd_gfx_v12_set_wave_launch_mode(struct amdgpu_device *adev,
+					uint8_t wave_launch_mode,
+					uint32_t vmid)
+{
+	uint32_t data = 0;
+	bool is_stall_mode = wave_launch_mode == 4;
+
+	if (is_stall_mode)
+		data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, STALL_VMID,
+									1);
+	else
+		data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, LAUNCH_MODE,
+							wave_launch_mode);
+
+	return data;
+}
+
+#define TCP_WATCH_STRIDE (regTCP_WATCH1_ADDR_H - regTCP_WATCH0_ADDR_H)
+static uint32_t kgd_gfx_v12_set_address_watch(struct amdgpu_device *adev,
+					uint64_t watch_address,
+					uint32_t watch_address_mask,
+					uint32_t watch_id,
+					uint32_t watch_mode,
+					uint32_t debug_vmid,
+					uint32_t inst)
+{
+	uint32_t watch_address_high;
+	uint32_t watch_address_low;
+	uint32_t watch_address_cntl;
+
+	watch_address_cntl = 0;
+	watch_address_low = lower_32_bits(watch_address);
+	watch_address_high = upper_32_bits(watch_address) & 0xffff;
+
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			MODE,
+			watch_mode);
+
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			MASK,
+			watch_address_mask >> 7);
+
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			VALID,
+			1);
+
+	WREG32_RLC((SOC15_REG_OFFSET(GC, 0, regTCP_WATCH0_ADDR_H) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_high);
+
+	WREG32_RLC((SOC15_REG_OFFSET(GC, 0, regTCP_WATCH0_ADDR_L) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_low);
+
+	return watch_address_cntl;
+}
+
+static uint32_t kgd_gfx_v12_clear_address_watch(struct amdgpu_device *adev,
+					uint32_t watch_id)
+{
+	return 0;
+}
+
+static uint32_t kgd_gfx_v12_hqd_sdma_get_doorbell(struct amdgpu_device *adev,
+						 int engine, int queue)
+{
+	return 0;
+}
+
+const struct kfd2kgd_calls gfx_v12_kfd2kgd = {
+	.init_interrupts = init_interrupts_v12,
+	.hqd_dump = hqd_dump_v12,
+	.hqd_sdma_dump = hqd_sdma_dump_v12,
+	.wave_control_execute = wave_control_execute_v12,
+	.get_atc_vmid_pasid_mapping_info = NULL,
+	.enable_debug_trap = kgd_gfx_v12_enable_debug_trap,
+	.disable_debug_trap = kgd_gfx_v12_disable_debug_trap,
+	.validate_trap_override_request = kgd_gfx_v12_validate_trap_override_request,
+	.set_wave_launch_trap_override = kgd_gfx_v12_set_wave_launch_trap_override,
+	.set_wave_launch_mode = kgd_gfx_v12_set_wave_launch_mode,
+	.set_address_watch = kgd_gfx_v12_set_address_watch,
+	.clear_address_watch = kgd_gfx_v12_clear_address_watch,
+	.hqd_sdma_get_doorbell = kgd_gfx_v12_hqd_sdma_get_doorbell
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
index b9dbbf9cb8b0..df77558e03ef 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
@@ -20,15 +20,10 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  */
 
-#include <linux/fdtable.h>
-#include <linux/uaccess.h>
-#include <linux/firmware.h>
-#include <drm/drmP.h>
 #include "amdgpu.h"
 #include "amdgpu_amdkfd.h"
 #include "cikd.h"
 #include "cik_sdma.h"
-#include "amdgpu_ucode.h"
 #include "gfx_v7_0.h"
 #include "gca/gfx_7_2_d.h"
 #include "gca/gfx_7_2_enum.h"
@@ -50,214 +45,54 @@ enum {
 	MAX_WATCH_ADDRESSES = 4
 };
 
-enum {
-	ADDRESS_WATCH_REG_ADDR_HI = 0,
-	ADDRESS_WATCH_REG_ADDR_LO,
-	ADDRESS_WATCH_REG_CNTL,
-	ADDRESS_WATCH_REG_MAX
-};
-
-/*  not defined in the CI/KV reg file  */
-enum {
-	ADDRESS_WATCH_REG_CNTL_ATC_BIT = 0x10000000UL,
-	ADDRESS_WATCH_REG_CNTL_DEFAULT_MASK = 0x00FFFFFF,
-	ADDRESS_WATCH_REG_ADDLOW_MASK_EXTENSION = 0x03000000,
-	/* extend the mask to 26 bits to match the low address field */
-	ADDRESS_WATCH_REG_ADDLOW_SHIFT = 6,
-	ADDRESS_WATCH_REG_ADDHIGH_MASK = 0xFFFF
-};
-
-static const uint32_t watchRegs[MAX_WATCH_ADDRESSES * ADDRESS_WATCH_REG_MAX] = {
-	mmTCP_WATCH0_ADDR_H, mmTCP_WATCH0_ADDR_L, mmTCP_WATCH0_CNTL,
-	mmTCP_WATCH1_ADDR_H, mmTCP_WATCH1_ADDR_L, mmTCP_WATCH1_CNTL,
-	mmTCP_WATCH2_ADDR_H, mmTCP_WATCH2_ADDR_L, mmTCP_WATCH2_CNTL,
-	mmTCP_WATCH3_ADDR_H, mmTCP_WATCH3_ADDR_L, mmTCP_WATCH3_CNTL
-};
-
-union TCP_WATCH_CNTL_BITS {
-	struct {
-		uint32_t mask:24;
-		uint32_t vmid:4;
-		uint32_t atc:1;
-		uint32_t mode:2;
-		uint32_t valid:1;
-	} bitfields, bits;
-	uint32_t u32All;
-	signed int i32All;
-	float f32All;
-};
-
-/*
- * Register access functions
- */
-
-static void kgd_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid,
-		uint32_t sh_mem_config,	uint32_t sh_mem_ape1_base,
-		uint32_t sh_mem_ape1_limit, uint32_t sh_mem_bases);
-
-static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid,
-					unsigned int vmid);
-
-static int kgd_init_pipeline(struct kgd_dev *kgd, uint32_t pipe_id,
-				uint32_t hpd_size, uint64_t hpd_gpu_addr);
-static int kgd_init_interrupts(struct kgd_dev *kgd, uint32_t pipe_id);
-static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
-			uint32_t queue_id, uint32_t __user *wptr,
-			uint32_t wptr_shift, uint32_t wptr_mask,
-			struct mm_struct *mm);
-static int kgd_hqd_sdma_load(struct kgd_dev *kgd, void *mqd);
-static bool kgd_hqd_is_occupied(struct kgd_dev *kgd, uint64_t queue_address,
-				uint32_t pipe_id, uint32_t queue_id);
-
-static int kgd_hqd_destroy(struct kgd_dev *kgd, void *mqd,
-				enum kfd_preempt_type reset_type,
-				unsigned int utimeout, uint32_t pipe_id,
-				uint32_t queue_id);
-static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd);
-static int kgd_hqd_sdma_destroy(struct kgd_dev *kgd, void *mqd,
-				unsigned int utimeout);
-static int kgd_address_watch_disable(struct kgd_dev *kgd);
-static int kgd_address_watch_execute(struct kgd_dev *kgd,
-					unsigned int watch_point_id,
-					uint32_t cntl_val,
-					uint32_t addr_hi,
-					uint32_t addr_lo);
-static int kgd_wave_control_execute(struct kgd_dev *kgd,
-					uint32_t gfx_index_val,
-					uint32_t sq_cmd);
-static uint32_t kgd_address_watch_get_offset(struct kgd_dev *kgd,
-					unsigned int watch_point_id,
-					unsigned int reg_offset);
-
-static bool get_atc_vmid_pasid_mapping_valid(struct kgd_dev *kgd, uint8_t vmid);
-static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
-							uint8_t vmid);
-static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid);
-
-static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type);
-static void set_scratch_backing_va(struct kgd_dev *kgd,
-					uint64_t va, uint32_t vmid);
-
-/* Because of REG_GET_FIELD() being used, we put this function in the
- * asic specific file.
- */
-static int get_tile_config(struct kgd_dev *kgd,
-		struct tile_config *config)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
-
-	config->gb_addr_config = adev->gfx.config.gb_addr_config;
-	config->num_banks = REG_GET_FIELD(adev->gfx.config.mc_arb_ramcfg,
-				MC_ARB_RAMCFG, NOOFBANK);
-	config->num_ranks = REG_GET_FIELD(adev->gfx.config.mc_arb_ramcfg,
-				MC_ARB_RAMCFG, NOOFRANKS);
-
-	config->tile_config_ptr = adev->gfx.config.tile_mode_array;
-	config->num_tile_configs =
-			ARRAY_SIZE(adev->gfx.config.tile_mode_array);
-	config->macro_tile_config_ptr =
-			adev->gfx.config.macrotile_mode_array;
-	config->num_macro_tile_configs =
-			ARRAY_SIZE(adev->gfx.config.macrotile_mode_array);
-
-	return 0;
-}
-
-static const struct kfd2kgd_calls kfd2kgd = {
-	.init_gtt_mem_allocation = alloc_gtt_mem,
-	.free_gtt_mem = free_gtt_mem,
-	.get_vmem_size = get_vmem_size,
-	.get_gpu_clock_counter = get_gpu_clock_counter,
-	.get_max_engine_clock_in_mhz = get_max_engine_clock_in_mhz,
-	.program_sh_mem_settings = kgd_program_sh_mem_settings,
-	.set_pasid_vmid_mapping = kgd_set_pasid_vmid_mapping,
-	.init_pipeline = kgd_init_pipeline,
-	.init_interrupts = kgd_init_interrupts,
-	.hqd_load = kgd_hqd_load,
-	.hqd_sdma_load = kgd_hqd_sdma_load,
-	.hqd_is_occupied = kgd_hqd_is_occupied,
-	.hqd_sdma_is_occupied = kgd_hqd_sdma_is_occupied,
-	.hqd_destroy = kgd_hqd_destroy,
-	.hqd_sdma_destroy = kgd_hqd_sdma_destroy,
-	.address_watch_disable = kgd_address_watch_disable,
-	.address_watch_execute = kgd_address_watch_execute,
-	.wave_control_execute = kgd_wave_control_execute,
-	.address_watch_get_offset = kgd_address_watch_get_offset,
-	.get_atc_vmid_pasid_mapping_pasid = get_atc_vmid_pasid_mapping_pasid,
-	.get_atc_vmid_pasid_mapping_valid = get_atc_vmid_pasid_mapping_valid,
-	.write_vmid_invalidate_request = write_vmid_invalidate_request,
-	.get_fw_version = get_fw_version,
-	.set_scratch_backing_va = set_scratch_backing_va,
-	.get_tile_config = get_tile_config,
-};
-
-struct kfd2kgd_calls *amdgpu_amdkfd_gfx_7_get_functions(void)
-{
-	return (struct kfd2kgd_calls *)&kfd2kgd;
-}
-
-static inline struct amdgpu_device *get_amdgpu_device(struct kgd_dev *kgd)
-{
-	return (struct amdgpu_device *)kgd;
-}
-
-static void lock_srbm(struct kgd_dev *kgd, uint32_t mec, uint32_t pipe,
+static void lock_srbm(struct amdgpu_device *adev, uint32_t mec, uint32_t pipe,
 			uint32_t queue, uint32_t vmid)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
 	uint32_t value = PIPEID(pipe) | MEID(mec) | VMID(vmid) | QUEUEID(queue);
 
 	mutex_lock(&adev->srbm_mutex);
 	WREG32(mmSRBM_GFX_CNTL, value);
 }
 
-static void unlock_srbm(struct kgd_dev *kgd)
+static void unlock_srbm(struct amdgpu_device *adev)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
-
 	WREG32(mmSRBM_GFX_CNTL, 0);
 	mutex_unlock(&adev->srbm_mutex);
 }
 
-static void acquire_queue(struct kgd_dev *kgd, uint32_t pipe_id,
+static void acquire_queue(struct amdgpu_device *adev, uint32_t pipe_id,
 				uint32_t queue_id)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
-
 	uint32_t mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
 	uint32_t pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
 
-	lock_srbm(kgd, mec, pipe, queue_id, 0);
+	lock_srbm(adev, mec, pipe, queue_id, 0);
 }
 
-static void release_queue(struct kgd_dev *kgd)
+static void release_queue(struct amdgpu_device *adev)
 {
-	unlock_srbm(kgd);
+	unlock_srbm(adev);
 }
 
-static void kgd_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid,
+static void kgd_program_sh_mem_settings(struct amdgpu_device *adev, uint32_t vmid,
 					uint32_t sh_mem_config,
 					uint32_t sh_mem_ape1_base,
 					uint32_t sh_mem_ape1_limit,
-					uint32_t sh_mem_bases)
+					uint32_t sh_mem_bases, uint32_t inst)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
-
-	lock_srbm(kgd, 0, 0, 0, vmid);
+	lock_srbm(adev, 0, 0, 0, vmid);
 
 	WREG32(mmSH_MEM_CONFIG, sh_mem_config);
 	WREG32(mmSH_MEM_APE1_BASE, sh_mem_ape1_base);
 	WREG32(mmSH_MEM_APE1_LIMIT, sh_mem_ape1_limit);
 	WREG32(mmSH_MEM_BASES, sh_mem_bases);
 
-	unlock_srbm(kgd);
+	unlock_srbm(adev);
 }
 
-static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid,
-					unsigned int vmid)
+static int kgd_set_pasid_vmid_mapping(struct amdgpu_device *adev, u32 pasid,
+					unsigned int vmid, uint32_t inst)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
-
 	/*
 	 * We have to assume that there is no outstanding mapping.
 	 * The ATC_VMID_PASID_MAPPING_UPDATE_STATUS bit could be 0 because
@@ -279,40 +114,34 @@ static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid,
 	return 0;
 }
 
-static int kgd_init_pipeline(struct kgd_dev *kgd, uint32_t pipe_id,
-				uint32_t hpd_size, uint64_t hpd_gpu_addr)
+static int kgd_init_interrupts(struct amdgpu_device *adev, uint32_t pipe_id,
+				uint32_t inst)
 {
-	/* amdgpu owns the per-pipe state */
-	return 0;
-}
-
-static int kgd_init_interrupts(struct kgd_dev *kgd, uint32_t pipe_id)
-{
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
 	uint32_t mec;
 	uint32_t pipe;
 
 	mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
 	pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
 
-	lock_srbm(kgd, mec, pipe, 0, 0);
+	lock_srbm(adev, mec, pipe, 0, 0);
 
 	WREG32(mmCPC_INT_CNTL, CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK |
 			CP_INT_CNTL_RING0__OPCODE_ERROR_INT_ENABLE_MASK);
 
-	unlock_srbm(kgd);
+	unlock_srbm(adev);
 
 	return 0;
 }
 
-static inline uint32_t get_sdma_base_addr(struct cik_sdma_rlc_registers *m)
+static inline uint32_t get_sdma_rlc_reg_offset(struct cik_sdma_rlc_registers *m)
 {
 	uint32_t retval;
 
 	retval = m->sdma_engine_id * SDMA1_REGISTER_OFFSET +
 			m->sdma_queue_id * KFD_CIK_SDMA_QUEUE_OFFSET;
 
-	pr_debug("kfd: sdma base address: 0x%x\n", retval);
+	pr_debug("RLC register offset for SDMA%d RLC%d: 0x%x\n",
+			m->sdma_engine_id, m->sdma_queue_id, retval);
 
 	return retval;
 }
@@ -327,19 +156,19 @@ static inline struct cik_sdma_rlc_registers *get_sdma_mqd(void *mqd)
 	return (struct cik_sdma_rlc_registers *)mqd;
 }
 
-static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
-			uint32_t queue_id, uint32_t __user *wptr,
-			uint32_t wptr_shift, uint32_t wptr_mask,
-			struct mm_struct *mm)
+static int kgd_hqd_load(struct amdgpu_device *adev, void *mqd,
+			uint32_t pipe_id, uint32_t queue_id,
+			uint32_t __user *wptr, uint32_t wptr_shift,
+			uint32_t wptr_mask, struct mm_struct *mm, uint32_t inst)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
 	struct cik_mqd *m;
 	uint32_t *mqd_hqd;
 	uint32_t reg, wptr_val, data;
+	bool valid_wptr = false;
 
 	m = get_mqd(mqd);
 
-	acquire_queue(kgd, pipe_id, queue_id);
+	acquire_queue(adev, pipe_id, queue_id);
 
 	/* HQD registers extend from CP_MQD_BASE_ADDR to CP_MQD_CONTROL. */
 	mqd_hqd = &m->cp_mqd_base_addr_lo;
@@ -354,59 +183,149 @@ static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
 			     CP_HQD_PQ_DOORBELL_CONTROL, DOORBELL_EN, 1);
 	WREG32(mmCP_HQD_PQ_DOORBELL_CONTROL, data);
 
-	if (read_user_wptr(mm, wptr, wptr_val))
+	/* read_user_ptr may take the mm->mmap_lock.
+	 * release srbm_mutex to avoid circular dependency between
+	 * srbm_mutex->mmap_lock->reservation_ww_class_mutex->srbm_mutex.
+	 */
+	release_queue(adev);
+	valid_wptr = read_user_wptr(mm, wptr, wptr_val);
+	acquire_queue(adev, pipe_id, queue_id);
+	if (valid_wptr)
 		WREG32(mmCP_HQD_PQ_WPTR, (wptr_val << wptr_shift) & wptr_mask);
 
 	data = REG_SET_FIELD(m->cp_hqd_active, CP_HQD_ACTIVE, ACTIVE, 1);
 	WREG32(mmCP_HQD_ACTIVE, data);
 
-	release_queue(kgd);
+	release_queue(adev);
+
+	return 0;
+}
+
+static int kgd_hqd_dump(struct amdgpu_device *adev,
+			uint32_t pipe_id, uint32_t queue_id,
+			uint32_t (**dump)[2], uint32_t *n_regs, uint32_t inst)
+{
+	uint32_t i = 0, reg;
+#define HQD_N_REGS (35+4)
+#define DUMP_REG(addr) do {				\
+		if (WARN_ON_ONCE(i >= HQD_N_REGS))	\
+			break;				\
+		(*dump)[i][0] = (addr) << 2;		\
+		(*dump)[i++][1] = RREG32(addr);		\
+	} while (0)
+
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
+	if (*dump == NULL)
+		return -ENOMEM;
+
+	acquire_queue(adev, pipe_id, queue_id);
+
+	DUMP_REG(mmCOMPUTE_STATIC_THREAD_MGMT_SE0);
+	DUMP_REG(mmCOMPUTE_STATIC_THREAD_MGMT_SE1);
+	DUMP_REG(mmCOMPUTE_STATIC_THREAD_MGMT_SE2);
+	DUMP_REG(mmCOMPUTE_STATIC_THREAD_MGMT_SE3);
+
+	for (reg = mmCP_MQD_BASE_ADDR; reg <= mmCP_MQD_CONTROL; reg++)
+		DUMP_REG(reg);
+
+	release_queue(adev);
+
+	WARN_ON_ONCE(i != HQD_N_REGS);
+	*n_regs = i;
 
 	return 0;
 }
 
-static int kgd_hqd_sdma_load(struct kgd_dev *kgd, void *mqd)
+static int kgd_hqd_sdma_load(struct amdgpu_device *adev, void *mqd,
+			     uint32_t __user *wptr, struct mm_struct *mm)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
 	struct cik_sdma_rlc_registers *m;
-	uint32_t sdma_base_addr;
+	unsigned long end_jiffies;
+	uint32_t sdma_rlc_reg_offset;
+	uint32_t data;
 
 	m = get_sdma_mqd(mqd);
-	sdma_base_addr = get_sdma_base_addr(m);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(m);
 
-	WREG32(sdma_base_addr + mmSDMA0_RLC0_VIRTUAL_ADDR,
-			m->sdma_rlc_virtual_addr);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL,
+		m->sdma_rlc_rb_cntl & (~SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK));
 
-	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_BASE,
-			m->sdma_rlc_rb_base);
+	end_jiffies = msecs_to_jiffies(2000) + jiffies;
+	while (true) {
+		data = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_CONTEXT_STATUS);
+		if (data & SDMA0_RLC0_CONTEXT_STATUS__IDLE_MASK)
+			break;
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("SDMA RLC not idle in %s\n", __func__);
+			return -ETIME;
+		}
+		usleep_range(500, 1000);
+	}
 
-	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_BASE_HI,
+	data = REG_SET_FIELD(m->sdma_rlc_doorbell, SDMA0_RLC0_DOORBELL,
+			     ENABLE, 1);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_DOORBELL, data);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR,
+				m->sdma_rlc_rb_rptr);
+
+	if (read_user_wptr(mm, wptr, data))
+		WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_WPTR, data);
+	else
+		WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_WPTR,
+		       m->sdma_rlc_rb_rptr);
+
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_VIRTUAL_ADDR,
+				m->sdma_rlc_virtual_addr);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_BASE, m->sdma_rlc_rb_base);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_BASE_HI,
 			m->sdma_rlc_rb_base_hi);
-
-	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR_ADDR_LO,
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR_ADDR_LO,
 			m->sdma_rlc_rb_rptr_addr_lo);
-
-	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR_ADDR_HI,
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR_ADDR_HI,
 			m->sdma_rlc_rb_rptr_addr_hi);
 
-	WREG32(sdma_base_addr + mmSDMA0_RLC0_DOORBELL,
-			m->sdma_rlc_doorbell);
+	data = REG_SET_FIELD(m->sdma_rlc_rb_cntl, SDMA0_RLC0_RB_CNTL,
+			     RB_ENABLE, 1);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL, data);
+
+	return 0;
+}
+
+static int kgd_hqd_sdma_dump(struct amdgpu_device *adev,
+			     uint32_t engine_id, uint32_t queue_id,
+			     uint32_t (**dump)[2], uint32_t *n_regs)
+{
+	uint32_t sdma_offset = engine_id * SDMA1_REGISTER_OFFSET +
+		queue_id * KFD_CIK_SDMA_QUEUE_OFFSET;
+	uint32_t i = 0, reg;
+#undef HQD_N_REGS
+#define HQD_N_REGS (19+4)
+
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
+	if (*dump == NULL)
+		return -ENOMEM;
 
-	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL,
-			m->sdma_rlc_rb_cntl);
+	for (reg = mmSDMA0_RLC0_RB_CNTL; reg <= mmSDMA0_RLC0_DOORBELL; reg++)
+		DUMP_REG(sdma_offset + reg);
+	for (reg = mmSDMA0_RLC0_VIRTUAL_ADDR; reg <= mmSDMA0_RLC0_WATERMARK;
+	     reg++)
+		DUMP_REG(sdma_offset + reg);
+
+	WARN_ON_ONCE(i != HQD_N_REGS);
+	*n_regs = i;
 
 	return 0;
 }
 
-static bool kgd_hqd_is_occupied(struct kgd_dev *kgd, uint64_t queue_address,
-				uint32_t pipe_id, uint32_t queue_id)
+static bool kgd_hqd_is_occupied(struct amdgpu_device *adev,
+				uint64_t queue_address, uint32_t pipe_id,
+				uint32_t queue_id, uint32_t inst)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
 	uint32_t act;
 	bool retval = false;
 	uint32_t low, high;
 
-	acquire_queue(kgd, pipe_id, queue_id);
+	acquire_queue(adev, pipe_id, queue_id);
 	act = RREG32(mmCP_HQD_ACTIVE);
 	if (act) {
 		low = lower_32_bits(queue_address >> 8);
@@ -416,21 +335,20 @@ static bool kgd_hqd_is_occupied(struct kgd_dev *kgd, uint64_t queue_address,
 				high == RREG32(mmCP_HQD_PQ_BASE_HI))
 			retval = true;
 	}
-	release_queue(kgd);
+	release_queue(adev);
 	return retval;
 }
 
-static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd)
+static bool kgd_hqd_sdma_is_occupied(struct amdgpu_device *adev, void *mqd)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
 	struct cik_sdma_rlc_registers *m;
-	uint32_t sdma_base_addr;
+	uint32_t sdma_rlc_reg_offset;
 	uint32_t sdma_rlc_rb_cntl;
 
 	m = get_sdma_mqd(mqd);
-	sdma_base_addr = get_sdma_base_addr(m);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(m);
 
-	sdma_rlc_rb_cntl = RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL);
+	sdma_rlc_rb_cntl = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL);
 
 	if (sdma_rlc_rb_cntl & SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK)
 		return true;
@@ -438,18 +356,20 @@ static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd)
 	return false;
 }
 
-static int kgd_hqd_destroy(struct kgd_dev *kgd, void *mqd,
+static int kgd_hqd_destroy(struct amdgpu_device *adev, void *mqd,
 				enum kfd_preempt_type reset_type,
 				unsigned int utimeout, uint32_t pipe_id,
-				uint32_t queue_id)
+				uint32_t queue_id, uint32_t inst)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
 	uint32_t temp;
 	enum hqd_dequeue_request_type type;
 	unsigned long flags, end_jiffies;
 	int retry;
 
-	acquire_queue(kgd, pipe_id, queue_id);
+	if (amdgpu_in_reset(adev))
+		return -EIO;
+
+	acquire_queue(adev, pipe_id, queue_id);
 	WREG32(mmCP_HQD_PQ_DOORBELL_CONTROL, 0);
 
 	switch (reset_type) {
@@ -527,106 +447,56 @@ loop:
 			break;
 		if (time_after(jiffies, end_jiffies)) {
 			pr_err("cp queue preemption time out\n");
-			release_queue(kgd);
+			release_queue(adev);
 			return -ETIME;
 		}
 		usleep_range(500, 1000);
 	}
 
-	release_queue(kgd);
+	release_queue(adev);
 	return 0;
 }
 
-static int kgd_hqd_sdma_destroy(struct kgd_dev *kgd, void *mqd,
+static int kgd_hqd_sdma_destroy(struct amdgpu_device *adev, void *mqd,
 				unsigned int utimeout)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
 	struct cik_sdma_rlc_registers *m;
-	uint32_t sdma_base_addr;
+	uint32_t sdma_rlc_reg_offset;
 	uint32_t temp;
-	int timeout = utimeout;
+	unsigned long end_jiffies = (utimeout * HZ / 1000) + jiffies;
 
 	m = get_sdma_mqd(mqd);
-	sdma_base_addr = get_sdma_base_addr(m);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(m);
 
-	temp = RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL);
+	temp = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL);
 	temp = temp & ~SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK;
-	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL, temp);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL, temp);
 
 	while (true) {
-		temp = RREG32(sdma_base_addr + mmSDMA0_RLC0_CONTEXT_STATUS);
-		if (temp & SDMA0_STATUS_REG__RB_CMD_IDLE__SHIFT)
+		temp = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_CONTEXT_STATUS);
+		if (temp & SDMA0_RLC0_CONTEXT_STATUS__IDLE_MASK)
 			break;
-		if (timeout <= 0)
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("SDMA RLC not idle in %s\n", __func__);
 			return -ETIME;
-		msleep(20);
-		timeout -= 20;
+		}
+		usleep_range(500, 1000);
 	}
 
-	WREG32(sdma_base_addr + mmSDMA0_RLC0_DOORBELL, 0);
-	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR, 0);
-	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_WPTR, 0);
-	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_BASE, 0);
-
-	return 0;
-}
-
-static int kgd_address_watch_disable(struct kgd_dev *kgd)
-{
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
-	union TCP_WATCH_CNTL_BITS cntl;
-	unsigned int i;
-
-	cntl.u32All = 0;
-
-	cntl.bitfields.valid = 0;
-	cntl.bitfields.mask = ADDRESS_WATCH_REG_CNTL_DEFAULT_MASK;
-	cntl.bitfields.atc = 1;
-
-	/* Turning off this address until we set all the registers */
-	for (i = 0; i < MAX_WATCH_ADDRESSES; i++)
-		WREG32(watchRegs[i * ADDRESS_WATCH_REG_MAX +
-			ADDRESS_WATCH_REG_CNTL], cntl.u32All);
-
-	return 0;
-}
-
-static int kgd_address_watch_execute(struct kgd_dev *kgd,
-					unsigned int watch_point_id,
-					uint32_t cntl_val,
-					uint32_t addr_hi,
-					uint32_t addr_lo)
-{
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
-	union TCP_WATCH_CNTL_BITS cntl;
-
-	cntl.u32All = cntl_val;
-
-	/* Turning off this watch point until we set all the registers */
-	cntl.bitfields.valid = 0;
-	WREG32(watchRegs[watch_point_id * ADDRESS_WATCH_REG_MAX +
-		ADDRESS_WATCH_REG_CNTL], cntl.u32All);
-
-	WREG32(watchRegs[watch_point_id * ADDRESS_WATCH_REG_MAX +
-		ADDRESS_WATCH_REG_ADDR_HI], addr_hi);
-
-	WREG32(watchRegs[watch_point_id * ADDRESS_WATCH_REG_MAX +
-		ADDRESS_WATCH_REG_ADDR_LO], addr_lo);
-
-	/* Enable the watch point */
-	cntl.bitfields.valid = 1;
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_DOORBELL, 0);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL,
+		RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL) |
+		SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK);
 
-	WREG32(watchRegs[watch_point_id * ADDRESS_WATCH_REG_MAX +
-		ADDRESS_WATCH_REG_CNTL], cntl.u32All);
+	m->sdma_rlc_rb_rptr = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR);
 
 	return 0;
 }
 
-static int kgd_wave_control_execute(struct kgd_dev *kgd,
+static int kgd_wave_control_execute(struct amdgpu_device *adev,
 					uint32_t gfx_index_val,
-					uint32_t sq_cmd)
+					uint32_t sq_cmd, uint32_t inst)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
 	uint32_t data;
 
 	mutex_lock(&adev->grbm_idx_mutex);
@@ -647,106 +517,73 @@ static int kgd_wave_control_execute(struct kgd_dev *kgd,
 	return 0;
 }
 
-static uint32_t kgd_address_watch_get_offset(struct kgd_dev *kgd,
-					unsigned int watch_point_id,
-					unsigned int reg_offset)
+static bool get_atc_vmid_pasid_mapping_info(struct amdgpu_device *adev,
+					uint8_t vmid, uint16_t *p_pasid)
 {
-	return watchRegs[watch_point_id * ADDRESS_WATCH_REG_MAX + reg_offset];
-}
+	uint32_t value;
 
-static bool get_atc_vmid_pasid_mapping_valid(struct kgd_dev *kgd,
-							uint8_t vmid)
-{
-	uint32_t reg;
-	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+	value = RREG32(mmATC_VMID0_PASID_MAPPING + vmid);
+	*p_pasid = value & ATC_VMID0_PASID_MAPPING__PASID_MASK;
 
-	reg = RREG32(mmATC_VMID0_PASID_MAPPING + vmid);
-	return reg & ATC_VMID0_PASID_MAPPING__VALID_MASK;
+	return !!(value & ATC_VMID0_PASID_MAPPING__VALID_MASK);
 }
 
-static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
-								uint8_t vmid)
+static void set_scratch_backing_va(struct amdgpu_device *adev,
+					uint64_t va, uint32_t vmid)
 {
-	uint32_t reg;
-	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
-
-	reg = RREG32(mmATC_VMID0_PASID_MAPPING + vmid);
-	return reg & ATC_VMID0_PASID_MAPPING__VALID_MASK;
+	lock_srbm(adev, 0, 0, 0, vmid);
+	WREG32(mmSH_HIDDEN_PRIVATE_BASE_VMID, va);
+	unlock_srbm(adev);
 }
 
-static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid)
+static void set_vm_context_page_table_base(struct amdgpu_device *adev,
+			uint32_t vmid, uint64_t page_table_base)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
-
-	WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
+	if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid)) {
+		pr_err("trying to set page table base for wrong VMID\n");
+		return;
+	}
+	WREG32(mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vmid - 8,
+		lower_32_bits(page_table_base));
 }
 
-static void set_scratch_backing_va(struct kgd_dev *kgd,
-					uint64_t va, uint32_t vmid)
+ /**
+  * read_vmid_from_vmfault_reg - read vmid from register
+  *
+  * adev: amdgpu_device pointer
+  * @vmid: vmid pointer
+  * read vmid from register (CIK).
+  */
+static uint32_t read_vmid_from_vmfault_reg(struct amdgpu_device *adev)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+	uint32_t status = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS);
 
-	lock_srbm(kgd, 0, 0, 0, vmid);
-	WREG32(mmSH_HIDDEN_PRIVATE_BASE_VMID, va);
-	unlock_srbm(kgd);
+	return REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS, VMID);
 }
 
-static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
-	const union amdgpu_firmware_header *hdr;
-
-	BUG_ON(kgd == NULL);
-
-	switch (type) {
-	case KGD_ENGINE_PFP:
-		hdr = (const union amdgpu_firmware_header *)
-						adev->gfx.pfp_fw->data;
-		break;
-
-	case KGD_ENGINE_ME:
-		hdr = (const union amdgpu_firmware_header *)
-						adev->gfx.me_fw->data;
-		break;
-
-	case KGD_ENGINE_CE:
-		hdr = (const union amdgpu_firmware_header *)
-						adev->gfx.ce_fw->data;
-		break;
-
-	case KGD_ENGINE_MEC1:
-		hdr = (const union amdgpu_firmware_header *)
-						adev->gfx.mec_fw->data;
-		break;
-
-	case KGD_ENGINE_MEC2:
-		hdr = (const union amdgpu_firmware_header *)
-						adev->gfx.mec2_fw->data;
-		break;
-
-	case KGD_ENGINE_RLC:
-		hdr = (const union amdgpu_firmware_header *)
-						adev->gfx.rlc_fw->data;
-		break;
-
-	case KGD_ENGINE_SDMA1:
-		hdr = (const union amdgpu_firmware_header *)
-						adev->sdma.instance[0].fw->data;
-		break;
-
-	case KGD_ENGINE_SDMA2:
-		hdr = (const union amdgpu_firmware_header *)
-						adev->sdma.instance[1].fw->data;
-		break;
+static uint32_t kgd_hqd_sdma_get_doorbell(struct amdgpu_device *adev,
+					  int engine, int queue)
 
-	default:
-		return 0;
-	}
-
-	if (hdr == NULL)
-		return 0;
-
-	/* Only 12 bit in use*/
-	return hdr->common.ucode_version;
+{
+	return 0;
 }
 
+const struct kfd2kgd_calls gfx_v7_kfd2kgd = {
+	.program_sh_mem_settings = kgd_program_sh_mem_settings,
+	.set_pasid_vmid_mapping = kgd_set_pasid_vmid_mapping,
+	.init_interrupts = kgd_init_interrupts,
+	.hqd_load = kgd_hqd_load,
+	.hqd_sdma_load = kgd_hqd_sdma_load,
+	.hqd_dump = kgd_hqd_dump,
+	.hqd_sdma_dump = kgd_hqd_sdma_dump,
+	.hqd_is_occupied = kgd_hqd_is_occupied,
+	.hqd_sdma_is_occupied = kgd_hqd_sdma_is_occupied,
+	.hqd_destroy = kgd_hqd_destroy,
+	.hqd_sdma_destroy = kgd_hqd_sdma_destroy,
+	.wave_control_execute = kgd_wave_control_execute,
+	.get_atc_vmid_pasid_mapping_info = get_atc_vmid_pasid_mapping_info,
+	.set_scratch_backing_va = set_scratch_backing_va,
+	.set_vm_context_page_table_base = set_vm_context_page_table_base,
+	.read_vmid_from_vmfault_reg = read_vmid_from_vmfault_reg,
+	.hqd_sdma_get_doorbell = kgd_hqd_sdma_get_doorbell,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
index 309f2419c6d8..e68c0fa8d751 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
@@ -20,14 +20,8 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  */
 
-#include <linux/module.h>
-#include <linux/fdtable.h>
-#include <linux/uaccess.h>
-#include <linux/firmware.h>
-#include <drm/drmP.h>
 #include "amdgpu.h"
 #include "amdgpu_amdkfd.h"
-#include "amdgpu_ucode.h"
 #include "gfx_v8_0.h"
 #include "gca/gfx_8_0_sh_mask.h"
 #include "gca/gfx_8_0_d.h"
@@ -45,180 +39,54 @@ enum hqd_dequeue_request_type {
 	RESET_WAVES
 };
 
-struct cik_sdma_rlc_registers;
-
-/*
- * Register access functions
- */
-
-static void kgd_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid,
-		uint32_t sh_mem_config,
-		uint32_t sh_mem_ape1_base, uint32_t sh_mem_ape1_limit,
-		uint32_t sh_mem_bases);
-static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid,
-		unsigned int vmid);
-static int kgd_init_pipeline(struct kgd_dev *kgd, uint32_t pipe_id,
-		uint32_t hpd_size, uint64_t hpd_gpu_addr);
-static int kgd_init_interrupts(struct kgd_dev *kgd, uint32_t pipe_id);
-static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
-			uint32_t queue_id, uint32_t __user *wptr,
-			uint32_t wptr_shift, uint32_t wptr_mask,
-			struct mm_struct *mm);
-static int kgd_hqd_sdma_load(struct kgd_dev *kgd, void *mqd);
-static bool kgd_hqd_is_occupied(struct kgd_dev *kgd, uint64_t queue_address,
-		uint32_t pipe_id, uint32_t queue_id);
-static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd);
-static int kgd_hqd_destroy(struct kgd_dev *kgd, void *mqd,
-				enum kfd_preempt_type reset_type,
-				unsigned int utimeout, uint32_t pipe_id,
-				uint32_t queue_id);
-static int kgd_hqd_sdma_destroy(struct kgd_dev *kgd, void *mqd,
-				unsigned int utimeout);
-static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid);
-static int kgd_address_watch_disable(struct kgd_dev *kgd);
-static int kgd_address_watch_execute(struct kgd_dev *kgd,
-					unsigned int watch_point_id,
-					uint32_t cntl_val,
-					uint32_t addr_hi,
-					uint32_t addr_lo);
-static int kgd_wave_control_execute(struct kgd_dev *kgd,
-					uint32_t gfx_index_val,
-					uint32_t sq_cmd);
-static uint32_t kgd_address_watch_get_offset(struct kgd_dev *kgd,
-					unsigned int watch_point_id,
-					unsigned int reg_offset);
-
-static bool get_atc_vmid_pasid_mapping_valid(struct kgd_dev *kgd,
-		uint8_t vmid);
-static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
-		uint8_t vmid);
-static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid);
-static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type);
-static void set_scratch_backing_va(struct kgd_dev *kgd,
-					uint64_t va, uint32_t vmid);
-
-/* Because of REG_GET_FIELD() being used, we put this function in the
- * asic specific file.
- */
-static int get_tile_config(struct kgd_dev *kgd,
-		struct tile_config *config)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
-
-	config->gb_addr_config = adev->gfx.config.gb_addr_config;
-	config->num_banks = REG_GET_FIELD(adev->gfx.config.mc_arb_ramcfg,
-				MC_ARB_RAMCFG, NOOFBANK);
-	config->num_ranks = REG_GET_FIELD(adev->gfx.config.mc_arb_ramcfg,
-				MC_ARB_RAMCFG, NOOFRANKS);
-
-	config->tile_config_ptr = adev->gfx.config.tile_mode_array;
-	config->num_tile_configs =
-			ARRAY_SIZE(adev->gfx.config.tile_mode_array);
-	config->macro_tile_config_ptr =
-			adev->gfx.config.macrotile_mode_array;
-	config->num_macro_tile_configs =
-			ARRAY_SIZE(adev->gfx.config.macrotile_mode_array);
-
-	return 0;
-}
-
-static const struct kfd2kgd_calls kfd2kgd = {
-	.init_gtt_mem_allocation = alloc_gtt_mem,
-	.free_gtt_mem = free_gtt_mem,
-	.get_vmem_size = get_vmem_size,
-	.get_gpu_clock_counter = get_gpu_clock_counter,
-	.get_max_engine_clock_in_mhz = get_max_engine_clock_in_mhz,
-	.program_sh_mem_settings = kgd_program_sh_mem_settings,
-	.set_pasid_vmid_mapping = kgd_set_pasid_vmid_mapping,
-	.init_pipeline = kgd_init_pipeline,
-	.init_interrupts = kgd_init_interrupts,
-	.hqd_load = kgd_hqd_load,
-	.hqd_sdma_load = kgd_hqd_sdma_load,
-	.hqd_is_occupied = kgd_hqd_is_occupied,
-	.hqd_sdma_is_occupied = kgd_hqd_sdma_is_occupied,
-	.hqd_destroy = kgd_hqd_destroy,
-	.hqd_sdma_destroy = kgd_hqd_sdma_destroy,
-	.address_watch_disable = kgd_address_watch_disable,
-	.address_watch_execute = kgd_address_watch_execute,
-	.wave_control_execute = kgd_wave_control_execute,
-	.address_watch_get_offset = kgd_address_watch_get_offset,
-	.get_atc_vmid_pasid_mapping_pasid =
-			get_atc_vmid_pasid_mapping_pasid,
-	.get_atc_vmid_pasid_mapping_valid =
-			get_atc_vmid_pasid_mapping_valid,
-	.write_vmid_invalidate_request = write_vmid_invalidate_request,
-	.get_fw_version = get_fw_version,
-	.set_scratch_backing_va = set_scratch_backing_va,
-	.get_tile_config = get_tile_config,
-};
-
-struct kfd2kgd_calls *amdgpu_amdkfd_gfx_8_0_get_functions(void)
-{
-	return (struct kfd2kgd_calls *)&kfd2kgd;
-}
-
-static inline struct amdgpu_device *get_amdgpu_device(struct kgd_dev *kgd)
-{
-	return (struct amdgpu_device *)kgd;
-}
-
-static void lock_srbm(struct kgd_dev *kgd, uint32_t mec, uint32_t pipe,
+static void lock_srbm(struct amdgpu_device *adev, uint32_t mec, uint32_t pipe,
 			uint32_t queue, uint32_t vmid)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
 	uint32_t value = PIPEID(pipe) | MEID(mec) | VMID(vmid) | QUEUEID(queue);
 
 	mutex_lock(&adev->srbm_mutex);
 	WREG32(mmSRBM_GFX_CNTL, value);
 }
 
-static void unlock_srbm(struct kgd_dev *kgd)
+static void unlock_srbm(struct amdgpu_device *adev)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
-
 	WREG32(mmSRBM_GFX_CNTL, 0);
 	mutex_unlock(&adev->srbm_mutex);
 }
 
-static void acquire_queue(struct kgd_dev *kgd, uint32_t pipe_id,
+static void acquire_queue(struct amdgpu_device *adev, uint32_t pipe_id,
 				uint32_t queue_id)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
-
 	uint32_t mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
 	uint32_t pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
 
-	lock_srbm(kgd, mec, pipe, queue_id, 0);
+	lock_srbm(adev, mec, pipe, queue_id, 0);
 }
 
-static void release_queue(struct kgd_dev *kgd)
+static void release_queue(struct amdgpu_device *adev)
 {
-	unlock_srbm(kgd);
+	unlock_srbm(adev);
 }
 
-static void kgd_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid,
+static void kgd_program_sh_mem_settings(struct amdgpu_device *adev, uint32_t vmid,
 					uint32_t sh_mem_config,
 					uint32_t sh_mem_ape1_base,
 					uint32_t sh_mem_ape1_limit,
-					uint32_t sh_mem_bases)
+					uint32_t sh_mem_bases, uint32_t inst)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
-
-	lock_srbm(kgd, 0, 0, 0, vmid);
+	lock_srbm(adev, 0, 0, 0, vmid);
 
 	WREG32(mmSH_MEM_CONFIG, sh_mem_config);
 	WREG32(mmSH_MEM_APE1_BASE, sh_mem_ape1_base);
 	WREG32(mmSH_MEM_APE1_LIMIT, sh_mem_ape1_limit);
 	WREG32(mmSH_MEM_BASES, sh_mem_bases);
 
-	unlock_srbm(kgd);
+	unlock_srbm(adev);
 }
 
-static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid,
-					unsigned int vmid)
+static int kgd_set_pasid_vmid_mapping(struct amdgpu_device *adev, u32 pasid,
+					unsigned int vmid, uint32_t inst)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
-
 	/*
 	 * We have to assume that there is no outstanding mapping.
 	 * The ATC_VMID_PASID_MAPPING_UPDATE_STATUS bit could be 0 because
@@ -241,34 +109,36 @@ static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid,
 	return 0;
 }
 
-static int kgd_init_pipeline(struct kgd_dev *kgd, uint32_t pipe_id,
-				uint32_t hpd_size, uint64_t hpd_gpu_addr)
+static int kgd_init_interrupts(struct amdgpu_device *adev, uint32_t pipe_id,
+				uint32_t inst)
 {
-	/* amdgpu owns the per-pipe state */
-	return 0;
-}
-
-static int kgd_init_interrupts(struct kgd_dev *kgd, uint32_t pipe_id)
-{
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
 	uint32_t mec;
 	uint32_t pipe;
 
 	mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
 	pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
 
-	lock_srbm(kgd, mec, pipe, 0, 0);
+	lock_srbm(adev, mec, pipe, 0, 0);
 
-	WREG32(mmCPC_INT_CNTL, CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK);
+	WREG32(mmCPC_INT_CNTL, CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK |
+			CP_INT_CNTL_RING0__OPCODE_ERROR_INT_ENABLE_MASK);
 
-	unlock_srbm(kgd);
+	unlock_srbm(adev);
 
 	return 0;
 }
 
-static inline uint32_t get_sdma_base_addr(struct cik_sdma_rlc_registers *m)
+static inline uint32_t get_sdma_rlc_reg_offset(struct vi_sdma_mqd *m)
 {
-	return 0;
+	uint32_t retval;
+
+	retval = m->sdma_engine_id * SDMA1_REGISTER_OFFSET +
+		m->sdma_queue_id * KFD_VI_SDMA_QUEUE_OFFSET;
+
+	pr_debug("RLC register offset for SDMA%d RLC%d: 0x%x\n",
+			m->sdma_engine_id, m->sdma_queue_id, retval);
+
+	return retval;
 }
 
 static inline struct vi_mqd *get_mqd(void *mqd)
@@ -276,24 +146,24 @@ static inline struct vi_mqd *get_mqd(void *mqd)
 	return (struct vi_mqd *)mqd;
 }
 
-static inline struct cik_sdma_rlc_registers *get_sdma_mqd(void *mqd)
+static inline struct vi_sdma_mqd *get_sdma_mqd(void *mqd)
 {
-	return (struct cik_sdma_rlc_registers *)mqd;
+	return (struct vi_sdma_mqd *)mqd;
 }
 
-static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
-			uint32_t queue_id, uint32_t __user *wptr,
-			uint32_t wptr_shift, uint32_t wptr_mask,
-			struct mm_struct *mm)
+static int kgd_hqd_load(struct amdgpu_device *adev, void *mqd,
+			uint32_t pipe_id, uint32_t queue_id,
+			uint32_t __user *wptr, uint32_t wptr_shift,
+			uint32_t wptr_mask, struct mm_struct *mm, uint32_t inst)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
 	struct vi_mqd *m;
 	uint32_t *mqd_hqd;
 	uint32_t reg, wptr_val, data;
+	bool valid_wptr = false;
 
 	m = get_mqd(mqd);
 
-	acquire_queue(kgd, pipe_id, queue_id);
+	acquire_queue(adev, pipe_id, queue_id);
 
 	/* HIQ is set during driver init period with vmid set to 0*/
 	if (m->cp_hqd_vmid == 0) {
@@ -321,7 +191,7 @@ static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
 	 * on ASICs that do not support context-save.
 	 * EOP writes/reads can start anywhere in the ring.
 	 */
-	if (get_amdgpu_device(kgd)->asic_type != CHIP_TONGA) {
+	if (adev->asic_type != CHIP_TONGA) {
 		WREG32(mmCP_HQD_EOP_RPTR, m->cp_hqd_eop_rptr);
 		WREG32(mmCP_HQD_EOP_WPTR, m->cp_hqd_eop_wptr);
 		WREG32(mmCP_HQD_EOP_WPTR_MEM, m->cp_hqd_eop_wptr_mem);
@@ -337,31 +207,157 @@ static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
 			     CP_HQD_PQ_DOORBELL_CONTROL, DOORBELL_EN, 1);
 	WREG32(mmCP_HQD_PQ_DOORBELL_CONTROL, data);
 
-	if (read_user_wptr(mm, wptr, wptr_val))
+	/* read_user_ptr may take the mm->mmap_lock.
+	 * release srbm_mutex to avoid circular dependency between
+	 * srbm_mutex->mmap_lock->reservation_ww_class_mutex->srbm_mutex.
+	 */
+	release_queue(adev);
+	valid_wptr = read_user_wptr(mm, wptr, wptr_val);
+	acquire_queue(adev, pipe_id, queue_id);
+	if (valid_wptr)
 		WREG32(mmCP_HQD_PQ_WPTR, (wptr_val << wptr_shift) & wptr_mask);
 
 	data = REG_SET_FIELD(m->cp_hqd_active, CP_HQD_ACTIVE, ACTIVE, 1);
 	WREG32(mmCP_HQD_ACTIVE, data);
 
-	release_queue(kgd);
+	release_queue(adev);
 
 	return 0;
 }
 
-static int kgd_hqd_sdma_load(struct kgd_dev *kgd, void *mqd)
+static int kgd_hqd_dump(struct amdgpu_device *adev,
+			uint32_t pipe_id, uint32_t queue_id,
+			uint32_t (**dump)[2], uint32_t *n_regs, uint32_t inst)
 {
+	uint32_t i = 0, reg;
+#define HQD_N_REGS (54+4)
+#define DUMP_REG(addr) do {				\
+		if (WARN_ON_ONCE(i >= HQD_N_REGS))	\
+			break;				\
+		(*dump)[i][0] = (addr) << 2;		\
+		(*dump)[i++][1] = RREG32(addr);		\
+	} while (0)
+
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
+	if (*dump == NULL)
+		return -ENOMEM;
+
+	acquire_queue(adev, pipe_id, queue_id);
+
+	DUMP_REG(mmCOMPUTE_STATIC_THREAD_MGMT_SE0);
+	DUMP_REG(mmCOMPUTE_STATIC_THREAD_MGMT_SE1);
+	DUMP_REG(mmCOMPUTE_STATIC_THREAD_MGMT_SE2);
+	DUMP_REG(mmCOMPUTE_STATIC_THREAD_MGMT_SE3);
+
+	for (reg = mmCP_MQD_BASE_ADDR; reg <= mmCP_HQD_EOP_DONES; reg++)
+		DUMP_REG(reg);
+
+	release_queue(adev);
+
+	WARN_ON_ONCE(i != HQD_N_REGS);
+	*n_regs = i;
+
 	return 0;
 }
 
-static bool kgd_hqd_is_occupied(struct kgd_dev *kgd, uint64_t queue_address,
-				uint32_t pipe_id, uint32_t queue_id)
+static int kgd_hqd_sdma_load(struct amdgpu_device *adev, void *mqd,
+			     uint32_t __user *wptr, struct mm_struct *mm)
+{
+	struct vi_sdma_mqd *m;
+	unsigned long end_jiffies;
+	uint32_t sdma_rlc_reg_offset;
+	uint32_t data;
+
+	m = get_sdma_mqd(mqd);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(m);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL,
+		m->sdmax_rlcx_rb_cntl & (~SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK));
+
+	end_jiffies = msecs_to_jiffies(2000) + jiffies;
+	while (true) {
+		data = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_CONTEXT_STATUS);
+		if (data & SDMA0_RLC0_CONTEXT_STATUS__IDLE_MASK)
+			break;
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("SDMA RLC not idle in %s\n", __func__);
+			return -ETIME;
+		}
+		usleep_range(500, 1000);
+	}
+
+	data = REG_SET_FIELD(m->sdmax_rlcx_doorbell, SDMA0_RLC0_DOORBELL,
+			     ENABLE, 1);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_DOORBELL, data);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR,
+				m->sdmax_rlcx_rb_rptr);
+
+	if (read_user_wptr(mm, wptr, data))
+		WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_WPTR, data);
+	else
+		WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_WPTR,
+		       m->sdmax_rlcx_rb_rptr);
+
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_VIRTUAL_ADDR,
+				m->sdmax_rlcx_virtual_addr);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_BASE, m->sdmax_rlcx_rb_base);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_BASE_HI,
+			m->sdmax_rlcx_rb_base_hi);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR_ADDR_LO,
+			m->sdmax_rlcx_rb_rptr_addr_lo);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR_ADDR_HI,
+			m->sdmax_rlcx_rb_rptr_addr_hi);
+
+	data = REG_SET_FIELD(m->sdmax_rlcx_rb_cntl, SDMA0_RLC0_RB_CNTL,
+			     RB_ENABLE, 1);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL, data);
+
+	return 0;
+}
+
+static int kgd_hqd_sdma_dump(struct amdgpu_device *adev,
+			     uint32_t engine_id, uint32_t queue_id,
+			     uint32_t (**dump)[2], uint32_t *n_regs)
+{
+	uint32_t sdma_offset = engine_id * SDMA1_REGISTER_OFFSET +
+		queue_id * KFD_VI_SDMA_QUEUE_OFFSET;
+	uint32_t i = 0, reg;
+#undef HQD_N_REGS
+#define HQD_N_REGS (19+4+2+3+7)
+
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
+	if (*dump == NULL)
+		return -ENOMEM;
+
+	for (reg = mmSDMA0_RLC0_RB_CNTL; reg <= mmSDMA0_RLC0_DOORBELL; reg++)
+		DUMP_REG(sdma_offset + reg);
+	for (reg = mmSDMA0_RLC0_VIRTUAL_ADDR; reg <= mmSDMA0_RLC0_WATERMARK;
+	     reg++)
+		DUMP_REG(sdma_offset + reg);
+	for (reg = mmSDMA0_RLC0_CSA_ADDR_LO; reg <= mmSDMA0_RLC0_CSA_ADDR_HI;
+	     reg++)
+		DUMP_REG(sdma_offset + reg);
+	for (reg = mmSDMA0_RLC0_IB_SUB_REMAIN; reg <= mmSDMA0_RLC0_DUMMY_REG;
+	     reg++)
+		DUMP_REG(sdma_offset + reg);
+	for (reg = mmSDMA0_RLC0_MIDCMD_DATA0; reg <= mmSDMA0_RLC0_MIDCMD_CNTL;
+	     reg++)
+		DUMP_REG(sdma_offset + reg);
+
+	WARN_ON_ONCE(i != HQD_N_REGS);
+	*n_regs = i;
+
+	return 0;
+}
+
+static bool kgd_hqd_is_occupied(struct amdgpu_device *adev,
+				uint64_t queue_address, uint32_t pipe_id,
+				uint32_t queue_id, uint32_t inst)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
 	uint32_t act;
 	bool retval = false;
 	uint32_t low, high;
 
-	acquire_queue(kgd, pipe_id, queue_id);
+	acquire_queue(adev, pipe_id, queue_id);
 	act = RREG32(mmCP_HQD_ACTIVE);
 	if (act) {
 		low = lower_32_bits(queue_address >> 8);
@@ -371,21 +367,20 @@ static bool kgd_hqd_is_occupied(struct kgd_dev *kgd, uint64_t queue_address,
 				high == RREG32(mmCP_HQD_PQ_BASE_HI))
 			retval = true;
 	}
-	release_queue(kgd);
+	release_queue(adev);
 	return retval;
 }
 
-static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd)
+static bool kgd_hqd_sdma_is_occupied(struct amdgpu_device *adev, void *mqd)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
-	struct cik_sdma_rlc_registers *m;
-	uint32_t sdma_base_addr;
+	struct vi_sdma_mqd *m;
+	uint32_t sdma_rlc_reg_offset;
 	uint32_t sdma_rlc_rb_cntl;
 
 	m = get_sdma_mqd(mqd);
-	sdma_base_addr = get_sdma_base_addr(m);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(m);
 
-	sdma_rlc_rb_cntl = RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL);
+	sdma_rlc_rb_cntl = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL);
 
 	if (sdma_rlc_rb_cntl & SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK)
 		return true;
@@ -393,19 +388,21 @@ static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd)
 	return false;
 }
 
-static int kgd_hqd_destroy(struct kgd_dev *kgd, void *mqd,
+static int kgd_hqd_destroy(struct amdgpu_device *adev, void *mqd,
 				enum kfd_preempt_type reset_type,
 				unsigned int utimeout, uint32_t pipe_id,
-				uint32_t queue_id)
+				uint32_t queue_id, uint32_t inst)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
 	uint32_t temp;
 	enum hqd_dequeue_request_type type;
 	unsigned long flags, end_jiffies;
 	int retry;
 	struct vi_mqd *m = get_mqd(mqd);
 
-	acquire_queue(kgd, pipe_id, queue_id);
+	if (amdgpu_in_reset(adev))
+		return -EIO;
+
+	acquire_queue(adev, pipe_id, queue_id);
 
 	if (m->cp_hqd_vmid == 0)
 		WREG32_FIELD(RLC_CP_SCHEDULERS, scheduler1, 0);
@@ -485,96 +482,67 @@ loop:
 			break;
 		if (time_after(jiffies, end_jiffies)) {
 			pr_err("cp queue preemption time out.\n");
-			release_queue(kgd);
+			release_queue(adev);
 			return -ETIME;
 		}
 		usleep_range(500, 1000);
 	}
 
-	release_queue(kgd);
+	release_queue(adev);
 	return 0;
 }
 
-static int kgd_hqd_sdma_destroy(struct kgd_dev *kgd, void *mqd,
+static int kgd_hqd_sdma_destroy(struct amdgpu_device *adev, void *mqd,
 				unsigned int utimeout)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
-	struct cik_sdma_rlc_registers *m;
-	uint32_t sdma_base_addr;
+	struct vi_sdma_mqd *m;
+	uint32_t sdma_rlc_reg_offset;
 	uint32_t temp;
-	int timeout = utimeout;
+	unsigned long end_jiffies = (utimeout * HZ / 1000) + jiffies;
 
 	m = get_sdma_mqd(mqd);
-	sdma_base_addr = get_sdma_base_addr(m);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(m);
 
-	temp = RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL);
+	temp = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL);
 	temp = temp & ~SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK;
-	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL, temp);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL, temp);
 
 	while (true) {
-		temp = RREG32(sdma_base_addr + mmSDMA0_RLC0_CONTEXT_STATUS);
-		if (temp & SDMA0_STATUS_REG__RB_CMD_IDLE__SHIFT)
+		temp = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_CONTEXT_STATUS);
+		if (temp & SDMA0_RLC0_CONTEXT_STATUS__IDLE_MASK)
 			break;
-		if (timeout <= 0)
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("SDMA RLC not idle in %s\n", __func__);
 			return -ETIME;
-		msleep(20);
-		timeout -= 20;
+		}
+		usleep_range(500, 1000);
 	}
 
-	WREG32(sdma_base_addr + mmSDMA0_RLC0_DOORBELL, 0);
-	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR, 0);
-	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_WPTR, 0);
-	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_BASE, 0);
-
-	return 0;
-}
-
-static bool get_atc_vmid_pasid_mapping_valid(struct kgd_dev *kgd,
-							uint8_t vmid)
-{
-	uint32_t reg;
-	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
-
-	reg = RREG32(mmATC_VMID0_PASID_MAPPING + vmid);
-	return reg & ATC_VMID0_PASID_MAPPING__VALID_MASK;
-}
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_DOORBELL, 0);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL,
+		RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL) |
+		SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK);
 
-static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
-								uint8_t vmid)
-{
-	uint32_t reg;
-	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+	m->sdmax_rlcx_rb_rptr = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR);
 
-	reg = RREG32(mmATC_VMID0_PASID_MAPPING + vmid);
-	return reg & ATC_VMID0_PASID_MAPPING__VALID_MASK;
+	return 0;
 }
 
-static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid)
+static bool get_atc_vmid_pasid_mapping_info(struct amdgpu_device *adev,
+					uint8_t vmid, uint16_t *p_pasid)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+	uint32_t value;
 
-	WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
-}
+	value = RREG32(mmATC_VMID0_PASID_MAPPING + vmid);
+	*p_pasid = value & ATC_VMID0_PASID_MAPPING__PASID_MASK;
 
-static int kgd_address_watch_disable(struct kgd_dev *kgd)
-{
-	return 0;
+	return !!(value & ATC_VMID0_PASID_MAPPING__VALID_MASK);
 }
 
-static int kgd_address_watch_execute(struct kgd_dev *kgd,
-					unsigned int watch_point_id,
-					uint32_t cntl_val,
-					uint32_t addr_hi,
-					uint32_t addr_lo)
-{
-	return 0;
-}
-
-static int kgd_wave_control_execute(struct kgd_dev *kgd,
+static int kgd_wave_control_execute(struct amdgpu_device *adev,
 					uint32_t gfx_index_val,
-					uint32_t sq_cmd)
+					uint32_t sq_cmd, uint32_t inst)
 {
-	struct amdgpu_device *adev = get_amdgpu_device(kgd);
 	uint32_t data = 0;
 
 	mutex_lock(&adev->grbm_idx_mutex);
@@ -595,78 +563,48 @@ static int kgd_wave_control_execute(struct kgd_dev *kgd,
 	return 0;
 }
 
-static uint32_t kgd_address_watch_get_offset(struct kgd_dev *kgd,
-					unsigned int watch_point_id,
-					unsigned int reg_offset)
-{
-	return 0;
-}
-
-static void set_scratch_backing_va(struct kgd_dev *kgd,
+static void set_scratch_backing_va(struct amdgpu_device *adev,
 					uint64_t va, uint32_t vmid)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
-
-	lock_srbm(kgd, 0, 0, 0, vmid);
+	lock_srbm(adev, 0, 0, 0, vmid);
 	WREG32(mmSH_HIDDEN_PRIVATE_BASE_VMID, va);
-	unlock_srbm(kgd);
+	unlock_srbm(adev);
 }
 
-static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type)
+static void set_vm_context_page_table_base(struct amdgpu_device *adev,
+		uint32_t vmid, uint64_t page_table_base)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
-	const union amdgpu_firmware_header *hdr;
-
-	BUG_ON(kgd == NULL);
-
-	switch (type) {
-	case KGD_ENGINE_PFP:
-		hdr = (const union amdgpu_firmware_header *)
-						adev->gfx.pfp_fw->data;
-		break;
-
-	case KGD_ENGINE_ME:
-		hdr = (const union amdgpu_firmware_header *)
-						adev->gfx.me_fw->data;
-		break;
-
-	case KGD_ENGINE_CE:
-		hdr = (const union amdgpu_firmware_header *)
-						adev->gfx.ce_fw->data;
-		break;
-
-	case KGD_ENGINE_MEC1:
-		hdr = (const union amdgpu_firmware_header *)
-						adev->gfx.mec_fw->data;
-		break;
-
-	case KGD_ENGINE_MEC2:
-		hdr = (const union amdgpu_firmware_header *)
-						adev->gfx.mec2_fw->data;
-		break;
-
-	case KGD_ENGINE_RLC:
-		hdr = (const union amdgpu_firmware_header *)
-						adev->gfx.rlc_fw->data;
-		break;
-
-	case KGD_ENGINE_SDMA1:
-		hdr = (const union amdgpu_firmware_header *)
-						adev->sdma.instance[0].fw->data;
-		break;
-
-	case KGD_ENGINE_SDMA2:
-		hdr = (const union amdgpu_firmware_header *)
-						adev->sdma.instance[1].fw->data;
-		break;
-
-	default:
-		return 0;
+	if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid)) {
+		pr_err("trying to set page table base for wrong VMID\n");
+		return;
 	}
+	WREG32(mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vmid - 8,
+			lower_32_bits(page_table_base));
+}
 
-	if (hdr == NULL)
-		return 0;
+static uint32_t kgd_hqd_sdma_get_doorbell(struct amdgpu_device *adev,
+					  int engine, int queue)
 
-	/* Only 12 bit in use*/
-	return hdr->common.ucode_version;
+{
+	return 0;
 }
+
+const struct kfd2kgd_calls gfx_v8_kfd2kgd = {
+	.program_sh_mem_settings = kgd_program_sh_mem_settings,
+	.set_pasid_vmid_mapping = kgd_set_pasid_vmid_mapping,
+	.init_interrupts = kgd_init_interrupts,
+	.hqd_load = kgd_hqd_load,
+	.hqd_sdma_load = kgd_hqd_sdma_load,
+	.hqd_dump = kgd_hqd_dump,
+	.hqd_sdma_dump = kgd_hqd_sdma_dump,
+	.hqd_is_occupied = kgd_hqd_is_occupied,
+	.hqd_sdma_is_occupied = kgd_hqd_sdma_is_occupied,
+	.hqd_destroy = kgd_hqd_destroy,
+	.hqd_sdma_destroy = kgd_hqd_sdma_destroy,
+	.wave_control_execute = kgd_wave_control_execute,
+	.get_atc_vmid_pasid_mapping_info =
+			get_atc_vmid_pasid_mapping_info,
+	.set_scratch_backing_va = set_scratch_backing_va,
+	.set_vm_context_page_table_base = set_vm_context_page_table_base,
+	.hqd_sdma_get_doorbell = kgd_hqd_sdma_get_doorbell,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c
new file mode 100644
index 000000000000..088d09cc7a72
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c
@@ -0,0 +1,1264 @@
+/*
+ * Copyright 2014-2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+#include "amdgpu.h"
+#include "amdgpu_amdkfd.h"
+#include "gc/gc_9_0_offset.h"
+#include "gc/gc_9_0_sh_mask.h"
+#include "vega10_enum.h"
+#include "sdma0/sdma0_4_0_offset.h"
+#include "sdma0/sdma0_4_0_sh_mask.h"
+#include "sdma1/sdma1_4_0_offset.h"
+#include "sdma1/sdma1_4_0_sh_mask.h"
+#include "athub/athub_1_0_offset.h"
+#include "athub/athub_1_0_sh_mask.h"
+#include "oss/osssys_4_0_offset.h"
+#include "oss/osssys_4_0_sh_mask.h"
+#include "soc15_common.h"
+#include "v9_structs.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "gfx_v9_0.h"
+#include "amdgpu_amdkfd_gfx_v9.h"
+#include <uapi/linux/kfd_ioctl.h>
+
+enum hqd_dequeue_request_type {
+	NO_ACTION = 0,
+	DRAIN_PIPE,
+	RESET_WAVES,
+	SAVE_WAVES
+};
+
+static void kgd_gfx_v9_lock_srbm(struct amdgpu_device *adev, uint32_t mec, uint32_t pipe,
+			uint32_t queue, uint32_t vmid, uint32_t inst)
+{
+	mutex_lock(&adev->srbm_mutex);
+	soc15_grbm_select(adev, mec, pipe, queue, vmid, GET_INST(GC, inst));
+}
+
+static void kgd_gfx_v9_unlock_srbm(struct amdgpu_device *adev, uint32_t inst)
+{
+	soc15_grbm_select(adev, 0, 0, 0, 0, GET_INST(GC, inst));
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+void kgd_gfx_v9_acquire_queue(struct amdgpu_device *adev, uint32_t pipe_id,
+				uint32_t queue_id, uint32_t inst)
+{
+	uint32_t mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	uint32_t pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+	kgd_gfx_v9_lock_srbm(adev, mec, pipe, queue_id, 0, inst);
+}
+
+uint64_t kgd_gfx_v9_get_queue_mask(struct amdgpu_device *adev,
+			       uint32_t pipe_id, uint32_t queue_id)
+{
+	unsigned int bit = pipe_id * adev->gfx.mec.num_queue_per_pipe +
+			queue_id;
+
+	return 1ull << bit;
+}
+
+void kgd_gfx_v9_release_queue(struct amdgpu_device *adev, uint32_t inst)
+{
+	kgd_gfx_v9_unlock_srbm(adev, inst);
+}
+
+void kgd_gfx_v9_program_sh_mem_settings(struct amdgpu_device *adev, uint32_t vmid,
+					uint32_t sh_mem_config,
+					uint32_t sh_mem_ape1_base,
+					uint32_t sh_mem_ape1_limit,
+					uint32_t sh_mem_bases, uint32_t inst)
+{
+	kgd_gfx_v9_lock_srbm(adev, 0, 0, 0, vmid, inst);
+
+	WREG32_SOC15_RLC(GC, GET_INST(GC, inst), mmSH_MEM_CONFIG, sh_mem_config);
+	WREG32_SOC15_RLC(GC, GET_INST(GC, inst), mmSH_MEM_BASES, sh_mem_bases);
+	/* APE1 no longer exists on GFX9 */
+
+	kgd_gfx_v9_unlock_srbm(adev, inst);
+}
+
+int kgd_gfx_v9_set_pasid_vmid_mapping(struct amdgpu_device *adev, u32 pasid,
+					unsigned int vmid, uint32_t inst)
+{
+	/*
+	 * We have to assume that there is no outstanding mapping.
+	 * The ATC_VMID_PASID_MAPPING_UPDATE_STATUS bit could be 0 because
+	 * a mapping is in progress or because a mapping finished
+	 * and the SW cleared it.
+	 * So the protocol is to always wait & clear.
+	 */
+	uint32_t pasid_mapping = (pasid == 0) ? 0 : (uint32_t)pasid |
+			ATC_VMID0_PASID_MAPPING__VALID_MASK;
+
+	/*
+	 * need to do this twice, once for gfx and once for mmhub
+	 * for ATC add 16 to VMID for mmhub, for IH different registers.
+	 * ATC_VMID0..15 registers are separate from ATC_VMID16..31.
+	 */
+
+	WREG32(SOC15_REG_OFFSET(ATHUB, 0, mmATC_VMID0_PASID_MAPPING) + vmid,
+	       pasid_mapping);
+
+	while (!(RREG32(SOC15_REG_OFFSET(
+				ATHUB, 0,
+				mmATC_VMID_PASID_MAPPING_UPDATE_STATUS)) &
+		 (1U << vmid)))
+		cpu_relax();
+
+	WREG32(SOC15_REG_OFFSET(ATHUB, 0,
+				mmATC_VMID_PASID_MAPPING_UPDATE_STATUS),
+	       1U << vmid);
+
+	/* Mapping vmid to pasid also for IH block */
+	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_VMID_0_LUT) + vmid,
+	       pasid_mapping);
+
+	WREG32(SOC15_REG_OFFSET(ATHUB, 0, mmATC_VMID16_PASID_MAPPING) + vmid,
+	       pasid_mapping);
+
+	while (!(RREG32(SOC15_REG_OFFSET(
+				ATHUB, 0,
+				mmATC_VMID_PASID_MAPPING_UPDATE_STATUS)) &
+		 (1U << (vmid + 16))))
+		cpu_relax();
+
+	WREG32(SOC15_REG_OFFSET(ATHUB, 0,
+				mmATC_VMID_PASID_MAPPING_UPDATE_STATUS),
+	       1U << (vmid + 16));
+
+	/* Mapping vmid to pasid also for IH block */
+	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_VMID_0_LUT_MM) + vmid,
+	       pasid_mapping);
+	return 0;
+}
+
+/* TODO - RING0 form of field is obsolete, seems to date back to SI
+ * but still works
+ */
+
+int kgd_gfx_v9_init_interrupts(struct amdgpu_device *adev, uint32_t pipe_id,
+				uint32_t inst)
+{
+	uint32_t mec;
+	uint32_t pipe;
+
+	mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+	kgd_gfx_v9_lock_srbm(adev, mec, pipe, 0, 0, inst);
+
+	WREG32_SOC15(GC, GET_INST(GC, inst), mmCPC_INT_CNTL,
+		CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK |
+		CP_INT_CNTL_RING0__OPCODE_ERROR_INT_ENABLE_MASK);
+
+	kgd_gfx_v9_unlock_srbm(adev, inst);
+
+	return 0;
+}
+
+static uint32_t get_sdma_rlc_reg_offset(struct amdgpu_device *adev,
+				unsigned int engine_id,
+				unsigned int queue_id)
+{
+	uint32_t sdma_engine_reg_base = 0;
+	uint32_t sdma_rlc_reg_offset;
+
+	switch (engine_id) {
+	default:
+		dev_warn(adev->dev,
+			 "Invalid sdma engine id (%d), using engine id 0\n",
+			 engine_id);
+		fallthrough;
+	case 0:
+		sdma_engine_reg_base = SOC15_REG_OFFSET(SDMA0, 0,
+				mmSDMA0_RLC0_RB_CNTL) - mmSDMA0_RLC0_RB_CNTL;
+		break;
+	case 1:
+		sdma_engine_reg_base = SOC15_REG_OFFSET(SDMA1, 0,
+				mmSDMA1_RLC0_RB_CNTL) - mmSDMA0_RLC0_RB_CNTL;
+		break;
+	}
+
+	sdma_rlc_reg_offset = sdma_engine_reg_base
+		+ queue_id * (mmSDMA0_RLC1_RB_CNTL - mmSDMA0_RLC0_RB_CNTL);
+
+	pr_debug("RLC register offset for SDMA%d RLC%d: 0x%x\n", engine_id,
+		 queue_id, sdma_rlc_reg_offset);
+
+	return sdma_rlc_reg_offset;
+}
+
+static inline struct v9_mqd *get_mqd(void *mqd)
+{
+	return (struct v9_mqd *)mqd;
+}
+
+static inline struct v9_sdma_mqd *get_sdma_mqd(void *mqd)
+{
+	return (struct v9_sdma_mqd *)mqd;
+}
+
+int kgd_gfx_v9_hqd_load(struct amdgpu_device *adev, void *mqd,
+			uint32_t pipe_id, uint32_t queue_id,
+			uint32_t __user *wptr, uint32_t wptr_shift,
+			uint32_t wptr_mask, struct mm_struct *mm,
+			uint32_t inst)
+{
+	struct v9_mqd *m;
+	uint32_t *mqd_hqd;
+	uint32_t reg, hqd_base, data;
+
+	m = get_mqd(mqd);
+
+	kgd_gfx_v9_acquire_queue(adev, pipe_id, queue_id, inst);
+
+	/* HQD registers extend from CP_MQD_BASE_ADDR to CP_HQD_EOP_WPTR_MEM. */
+	mqd_hqd = &m->cp_mqd_base_addr_lo;
+	hqd_base = SOC15_REG_OFFSET(GC, GET_INST(GC, inst), mmCP_MQD_BASE_ADDR);
+
+	for (reg = hqd_base;
+	     reg <= SOC15_REG_OFFSET(GC, GET_INST(GC, inst), mmCP_HQD_PQ_WPTR_HI); reg++)
+		WREG32_XCC(reg, mqd_hqd[reg - hqd_base], inst);
+
+
+	/* Activate doorbell logic before triggering WPTR poll. */
+	data = REG_SET_FIELD(m->cp_hqd_pq_doorbell_control,
+			     CP_HQD_PQ_DOORBELL_CONTROL, DOORBELL_EN, 1);
+	WREG32_SOC15_RLC(GC, GET_INST(GC, inst), mmCP_HQD_PQ_DOORBELL_CONTROL, data);
+
+	if (wptr) {
+		/* Don't read wptr with get_user because the user
+		 * context may not be accessible (if this function
+		 * runs in a work queue). Instead trigger a one-shot
+		 * polling read from memory in the CP. This assumes
+		 * that wptr is GPU-accessible in the queue's VMID via
+		 * ATC or SVM. WPTR==RPTR before starting the poll so
+		 * the CP starts fetching new commands from the right
+		 * place.
+		 *
+		 * Guessing a 64-bit WPTR from a 32-bit RPTR is a bit
+		 * tricky. Assume that the queue didn't overflow. The
+		 * number of valid bits in the 32-bit RPTR depends on
+		 * the queue size. The remaining bits are taken from
+		 * the saved 64-bit WPTR. If the WPTR wrapped, add the
+		 * queue size.
+		 */
+		uint32_t queue_size =
+			2 << REG_GET_FIELD(m->cp_hqd_pq_control,
+					   CP_HQD_PQ_CONTROL, QUEUE_SIZE);
+		uint64_t guessed_wptr = m->cp_hqd_pq_rptr & (queue_size - 1);
+
+		if ((m->cp_hqd_pq_wptr_lo & (queue_size - 1)) < guessed_wptr)
+			guessed_wptr += queue_size;
+		guessed_wptr += m->cp_hqd_pq_wptr_lo & ~(queue_size - 1);
+		guessed_wptr += (uint64_t)m->cp_hqd_pq_wptr_hi << 32;
+
+		WREG32_SOC15_RLC(GC, GET_INST(GC, inst), mmCP_HQD_PQ_WPTR_LO,
+			lower_32_bits(guessed_wptr));
+		WREG32_SOC15_RLC(GC, GET_INST(GC, inst), mmCP_HQD_PQ_WPTR_HI,
+			upper_32_bits(guessed_wptr));
+		WREG32_SOC15_RLC(GC, GET_INST(GC, inst), mmCP_HQD_PQ_WPTR_POLL_ADDR,
+			lower_32_bits((uintptr_t)wptr));
+		WREG32_SOC15_RLC(GC, GET_INST(GC, inst), mmCP_HQD_PQ_WPTR_POLL_ADDR_HI,
+			upper_32_bits((uintptr_t)wptr));
+		WREG32_SOC15_RLC(GC, GET_INST(GC, inst), mmCP_PQ_WPTR_POLL_CNTL1,
+			(uint32_t)kgd_gfx_v9_get_queue_mask(adev, pipe_id, queue_id));
+	}
+
+	/* Start the EOP fetcher */
+	WREG32_SOC15_RLC(GC, GET_INST(GC, inst), mmCP_HQD_EOP_RPTR,
+	       REG_SET_FIELD(m->cp_hqd_eop_rptr, CP_HQD_EOP_RPTR, INIT_FETCHER, 1));
+
+	data = REG_SET_FIELD(m->cp_hqd_active, CP_HQD_ACTIVE, ACTIVE, 1);
+	WREG32_SOC15_RLC(GC, GET_INST(GC, inst), mmCP_HQD_ACTIVE, data);
+
+	kgd_gfx_v9_release_queue(adev, inst);
+
+	return 0;
+}
+
+int kgd_gfx_v9_hiq_mqd_load(struct amdgpu_device *adev, void *mqd,
+			    uint32_t pipe_id, uint32_t queue_id,
+			    uint32_t doorbell_off, uint32_t inst)
+{
+	struct amdgpu_ring *kiq_ring = &adev->gfx.kiq[inst].ring;
+	struct v9_mqd *m;
+	uint32_t mec, pipe;
+	int r;
+
+	m = get_mqd(mqd);
+
+	kgd_gfx_v9_acquire_queue(adev, pipe_id, queue_id, inst);
+
+	mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+	pr_debug("kfd: set HIQ, mec:%d, pipe:%d, queue:%d.\n",
+		 mec, pipe, queue_id);
+
+	spin_lock(&adev->gfx.kiq[inst].ring_lock);
+	r = amdgpu_ring_alloc(kiq_ring, 7);
+	if (r) {
+		pr_err("Failed to alloc KIQ (%d).\n", r);
+		goto out_unlock;
+	}
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_MAP_QUEUES, 5));
+	amdgpu_ring_write(kiq_ring,
+			  PACKET3_MAP_QUEUES_QUEUE_SEL(0) | /* Queue_Sel */
+			  PACKET3_MAP_QUEUES_VMID(m->cp_hqd_vmid) | /* VMID */
+			  PACKET3_MAP_QUEUES_QUEUE(queue_id) |
+			  PACKET3_MAP_QUEUES_PIPE(pipe) |
+			  PACKET3_MAP_QUEUES_ME((mec - 1)) |
+			  PACKET3_MAP_QUEUES_QUEUE_TYPE(0) | /*queue_type: normal compute queue */
+			  PACKET3_MAP_QUEUES_ALLOC_FORMAT(0) | /* alloc format: all_on_one_pipe */
+			  PACKET3_MAP_QUEUES_ENGINE_SEL(1) | /* engine_sel: hiq */
+			  PACKET3_MAP_QUEUES_NUM_QUEUES(1)); /* num_queues: must be 1 */
+	amdgpu_ring_write(kiq_ring,
+			  PACKET3_MAP_QUEUES_DOORBELL_OFFSET(doorbell_off));
+	amdgpu_ring_write(kiq_ring, m->cp_mqd_base_addr_lo);
+	amdgpu_ring_write(kiq_ring, m->cp_mqd_base_addr_hi);
+	amdgpu_ring_write(kiq_ring, m->cp_hqd_pq_wptr_poll_addr_lo);
+	amdgpu_ring_write(kiq_ring, m->cp_hqd_pq_wptr_poll_addr_hi);
+	amdgpu_ring_commit(kiq_ring);
+
+out_unlock:
+	spin_unlock(&adev->gfx.kiq[inst].ring_lock);
+	kgd_gfx_v9_release_queue(adev, inst);
+
+	return r;
+}
+
+int kgd_gfx_v9_hqd_dump(struct amdgpu_device *adev,
+			uint32_t pipe_id, uint32_t queue_id,
+			uint32_t (**dump)[2], uint32_t *n_regs, uint32_t inst)
+{
+	uint32_t i = 0, reg;
+#define HQD_N_REGS 56
+#define DUMP_REG(addr) do {				\
+		if (WARN_ON_ONCE(i >= HQD_N_REGS))	\
+			break;				\
+		(*dump)[i][0] = (addr) << 2;		\
+		(*dump)[i++][1] = RREG32(addr);		\
+	} while (0)
+
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
+	if (*dump == NULL)
+		return -ENOMEM;
+
+	kgd_gfx_v9_acquire_queue(adev, pipe_id, queue_id, inst);
+
+	for (reg = SOC15_REG_OFFSET(GC, GET_INST(GC, inst), mmCP_MQD_BASE_ADDR);
+	     reg <= SOC15_REG_OFFSET(GC, GET_INST(GC, inst), mmCP_HQD_PQ_WPTR_HI); reg++)
+		DUMP_REG(reg);
+
+	kgd_gfx_v9_release_queue(adev, inst);
+
+	WARN_ON_ONCE(i != HQD_N_REGS);
+	*n_regs = i;
+
+	return 0;
+}
+
+static int kgd_hqd_sdma_load(struct amdgpu_device *adev, void *mqd,
+			     uint32_t __user *wptr, struct mm_struct *mm)
+{
+	struct v9_sdma_mqd *m;
+	uint32_t sdma_rlc_reg_offset;
+	unsigned long end_jiffies;
+	uint32_t data;
+	uint64_t data64;
+	uint64_t __user *wptr64 = (uint64_t __user *)wptr;
+
+	m = get_sdma_mqd(mqd);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev, m->sdma_engine_id,
+					    m->sdma_queue_id);
+
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL,
+		m->sdmax_rlcx_rb_cntl & (~SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK));
+
+	end_jiffies = msecs_to_jiffies(2000) + jiffies;
+	while (true) {
+		data = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_CONTEXT_STATUS);
+		if (data & SDMA0_RLC0_CONTEXT_STATUS__IDLE_MASK)
+			break;
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("SDMA RLC not idle in %s\n", __func__);
+			return -ETIME;
+		}
+		usleep_range(500, 1000);
+	}
+
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_DOORBELL_OFFSET,
+	       m->sdmax_rlcx_doorbell_offset);
+
+	data = REG_SET_FIELD(m->sdmax_rlcx_doorbell, SDMA0_RLC0_DOORBELL,
+			     ENABLE, 1);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_DOORBELL, data);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR,
+				m->sdmax_rlcx_rb_rptr);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR_HI,
+				m->sdmax_rlcx_rb_rptr_hi);
+
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_MINOR_PTR_UPDATE, 1);
+	if (read_user_wptr(mm, wptr64, data64)) {
+		WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_WPTR,
+		       lower_32_bits(data64));
+		WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_WPTR_HI,
+		       upper_32_bits(data64));
+	} else {
+		WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_WPTR,
+		       m->sdmax_rlcx_rb_rptr);
+		WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_WPTR_HI,
+		       m->sdmax_rlcx_rb_rptr_hi);
+	}
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_MINOR_PTR_UPDATE, 0);
+
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_BASE, m->sdmax_rlcx_rb_base);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_BASE_HI,
+			m->sdmax_rlcx_rb_base_hi);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR_ADDR_LO,
+			m->sdmax_rlcx_rb_rptr_addr_lo);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR_ADDR_HI,
+			m->sdmax_rlcx_rb_rptr_addr_hi);
+
+	data = REG_SET_FIELD(m->sdmax_rlcx_rb_cntl, SDMA0_RLC0_RB_CNTL,
+			     RB_ENABLE, 1);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL, data);
+
+	return 0;
+}
+
+static int kgd_hqd_sdma_dump(struct amdgpu_device *adev,
+			     uint32_t engine_id, uint32_t queue_id,
+			     uint32_t (**dump)[2], uint32_t *n_regs)
+{
+	uint32_t sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev,
+			engine_id, queue_id);
+	uint32_t i = 0, reg;
+#undef HQD_N_REGS
+#define HQD_N_REGS (19+6+7+10)
+
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
+	if (*dump == NULL)
+		return -ENOMEM;
+
+	for (reg = mmSDMA0_RLC0_RB_CNTL; reg <= mmSDMA0_RLC0_DOORBELL; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+	for (reg = mmSDMA0_RLC0_STATUS; reg <= mmSDMA0_RLC0_CSA_ADDR_HI; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+	for (reg = mmSDMA0_RLC0_IB_SUB_REMAIN;
+	     reg <= mmSDMA0_RLC0_MINOR_PTR_UPDATE; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+	for (reg = mmSDMA0_RLC0_MIDCMD_DATA0;
+	     reg <= mmSDMA0_RLC0_MIDCMD_CNTL; reg++)
+		DUMP_REG(sdma_rlc_reg_offset + reg);
+
+	WARN_ON_ONCE(i != HQD_N_REGS);
+	*n_regs = i;
+
+	return 0;
+}
+
+bool kgd_gfx_v9_hqd_is_occupied(struct amdgpu_device *adev,
+				uint64_t queue_address, uint32_t pipe_id,
+				uint32_t queue_id, uint32_t inst)
+{
+	uint32_t act;
+	bool retval = false;
+	uint32_t low, high;
+
+	kgd_gfx_v9_acquire_queue(adev, pipe_id, queue_id, inst);
+	act = RREG32_SOC15(GC, GET_INST(GC, inst), mmCP_HQD_ACTIVE);
+	if (act) {
+		low = lower_32_bits(queue_address >> 8);
+		high = upper_32_bits(queue_address >> 8);
+
+		if (low == RREG32_SOC15(GC, GET_INST(GC, inst), mmCP_HQD_PQ_BASE) &&
+		   high == RREG32_SOC15(GC, GET_INST(GC, inst), mmCP_HQD_PQ_BASE_HI))
+			retval = true;
+	}
+	kgd_gfx_v9_release_queue(adev, inst);
+	return retval;
+}
+
+static bool kgd_hqd_sdma_is_occupied(struct amdgpu_device *adev, void *mqd)
+{
+	struct v9_sdma_mqd *m;
+	uint32_t sdma_rlc_reg_offset;
+	uint32_t sdma_rlc_rb_cntl;
+
+	m = get_sdma_mqd(mqd);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev, m->sdma_engine_id,
+					    m->sdma_queue_id);
+
+	sdma_rlc_rb_cntl = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL);
+
+	if (sdma_rlc_rb_cntl & SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK)
+		return true;
+
+	return false;
+}
+
+int kgd_gfx_v9_hqd_destroy(struct amdgpu_device *adev, void *mqd,
+				enum kfd_preempt_type reset_type,
+				unsigned int utimeout, uint32_t pipe_id,
+				uint32_t queue_id, uint32_t inst)
+{
+	enum hqd_dequeue_request_type type;
+	unsigned long end_jiffies;
+	uint32_t temp;
+	struct v9_mqd *m = get_mqd(mqd);
+
+	if (amdgpu_in_reset(adev))
+		return -EIO;
+
+	kgd_gfx_v9_acquire_queue(adev, pipe_id, queue_id, inst);
+
+	if (m->cp_hqd_vmid == 0)
+		WREG32_FIELD15_RLC(GC, GET_INST(GC, inst), RLC_CP_SCHEDULERS, scheduler1, 0);
+
+	switch (reset_type) {
+	case KFD_PREEMPT_TYPE_WAVEFRONT_DRAIN:
+		type = DRAIN_PIPE;
+		break;
+	case KFD_PREEMPT_TYPE_WAVEFRONT_RESET:
+		type = RESET_WAVES;
+		break;
+	case KFD_PREEMPT_TYPE_WAVEFRONT_SAVE:
+		type = SAVE_WAVES;
+		break;
+	default:
+		type = DRAIN_PIPE;
+		break;
+	}
+
+	WREG32_SOC15_RLC(GC, GET_INST(GC, inst), mmCP_HQD_DEQUEUE_REQUEST, type);
+
+	end_jiffies = (utimeout * HZ / 1000) + jiffies;
+	while (true) {
+		temp = RREG32_SOC15(GC, GET_INST(GC, inst), mmCP_HQD_ACTIVE);
+		if (!(temp & CP_HQD_ACTIVE__ACTIVE_MASK))
+			break;
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("cp queue preemption time out.\n");
+			kgd_gfx_v9_release_queue(adev, inst);
+			return -ETIME;
+		}
+		usleep_range(500, 1000);
+	}
+
+	kgd_gfx_v9_release_queue(adev, inst);
+	return 0;
+}
+
+static int kgd_hqd_sdma_destroy(struct amdgpu_device *adev, void *mqd,
+				unsigned int utimeout)
+{
+	struct v9_sdma_mqd *m;
+	uint32_t sdma_rlc_reg_offset;
+	uint32_t temp;
+	unsigned long end_jiffies = (utimeout * HZ / 1000) + jiffies;
+
+	m = get_sdma_mqd(mqd);
+	sdma_rlc_reg_offset = get_sdma_rlc_reg_offset(adev, m->sdma_engine_id,
+					    m->sdma_queue_id);
+
+	temp = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL);
+	temp = temp & ~SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK;
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL, temp);
+
+	while (true) {
+		temp = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_CONTEXT_STATUS);
+		if (temp & SDMA0_RLC0_CONTEXT_STATUS__IDLE_MASK)
+			break;
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("SDMA RLC not idle in %s\n", __func__);
+			return -ETIME;
+		}
+		usleep_range(500, 1000);
+	}
+
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_DOORBELL, 0);
+	WREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL,
+		RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_CNTL) |
+		SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK);
+
+	m->sdmax_rlcx_rb_rptr = RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR);
+	m->sdmax_rlcx_rb_rptr_hi =
+		RREG32(sdma_rlc_reg_offset + mmSDMA0_RLC0_RB_RPTR_HI);
+
+	return 0;
+}
+
+bool kgd_gfx_v9_get_atc_vmid_pasid_mapping_info(struct amdgpu_device *adev,
+					uint8_t vmid, uint16_t *p_pasid)
+{
+	uint32_t value;
+
+	value = RREG32(SOC15_REG_OFFSET(ATHUB, 0, mmATC_VMID0_PASID_MAPPING)
+		     + vmid);
+	*p_pasid = value & ATC_VMID0_PASID_MAPPING__PASID_MASK;
+
+	return !!(value & ATC_VMID0_PASID_MAPPING__VALID_MASK);
+}
+
+int kgd_gfx_v9_wave_control_execute(struct amdgpu_device *adev,
+					uint32_t gfx_index_val,
+					uint32_t sq_cmd, uint32_t inst)
+{
+	uint32_t data = 0;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	WREG32_SOC15_RLC_SHADOW(GC, GET_INST(GC, inst), mmGRBM_GFX_INDEX, gfx_index_val);
+	WREG32_SOC15(GC, GET_INST(GC, inst), mmSQ_CMD, sq_cmd);
+
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		INSTANCE_BROADCAST_WRITES, 1);
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		SH_BROADCAST_WRITES, 1);
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		SE_BROADCAST_WRITES, 1);
+
+	WREG32_SOC15_RLC_SHADOW(GC, GET_INST(GC, inst), mmGRBM_GFX_INDEX, data);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+/*
+ * GFX9 helper for wave launch stall requirements on debug trap setting.
+ *
+ * vmid:
+ *   Target VMID to stall/unstall.
+ *
+ * stall:
+ *   0-unstall wave launch (enable), 1-stall wave launch (disable).
+ *   After wavefront launch has been stalled, allocated waves must drain from
+ *   SPI in order for debug trap settings to take effect on those waves.
+ *   This is roughly a ~96 clock cycle wait on SPI where a read on
+ *   SPI_GDBG_WAVE_CNTL translates to ~32 clock cycles.
+ *   KGD_GFX_V9_WAVE_LAUNCH_SPI_DRAIN_LATENCY indicates the number of reads required.
+ *
+ *   NOTE: We can afford to clear the entire STALL_VMID field on unstall
+ *   because GFX9.4.1 cannot support multi-process debugging due to trap
+ *   configuration and masking being limited to global scope.  Always assume
+ *   single process conditions.
+ */
+#define KGD_GFX_V9_WAVE_LAUNCH_SPI_DRAIN_LATENCY	3
+void kgd_gfx_v9_set_wave_launch_stall(struct amdgpu_device *adev,
+					uint32_t vmid,
+					bool stall)
+{
+	int i;
+	uint32_t data = RREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_WAVE_CNTL));
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 1))
+		data = REG_SET_FIELD(data, SPI_GDBG_WAVE_CNTL, STALL_VMID,
+							stall ? 1 << vmid : 0);
+	else
+		data = REG_SET_FIELD(data, SPI_GDBG_WAVE_CNTL, STALL_RA,
+							stall ? 1 : 0);
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_WAVE_CNTL), data);
+
+	if (!stall)
+		return;
+
+	for (i = 0; i < KGD_GFX_V9_WAVE_LAUNCH_SPI_DRAIN_LATENCY; i++)
+		RREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_WAVE_CNTL));
+}
+
+/*
+ * restore_dbg_registers is ignored here but is a general interface requirement
+ * for devices that support GFXOFF and where the RLC save/restore list
+ * does not support hw registers for debugging i.e. the driver has to manually
+ * initialize the debug mode registers after it has disabled GFX off during the
+ * debug session.
+ */
+uint32_t kgd_gfx_v9_enable_debug_trap(struct amdgpu_device *adev,
+				bool restore_dbg_registers,
+				uint32_t vmid)
+{
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	kgd_gfx_v9_set_wave_launch_stall(adev, vmid, true);
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_TRAP_MASK), 0);
+
+	kgd_gfx_v9_set_wave_launch_stall(adev, vmid, false);
+
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+/*
+ * keep_trap_enabled is ignored here but is a general interface requirement
+ * for devices that support multi-process debugging where the performance
+ * overhead from trap temporary setup needs to be bypassed when the debug
+ * session has ended.
+ */
+uint32_t kgd_gfx_v9_disable_debug_trap(struct amdgpu_device *adev,
+					bool keep_trap_enabled,
+					uint32_t vmid)
+{
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	kgd_gfx_v9_set_wave_launch_stall(adev, vmid, true);
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_TRAP_MASK), 0);
+
+	kgd_gfx_v9_set_wave_launch_stall(adev, vmid, false);
+
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+int kgd_gfx_v9_validate_trap_override_request(struct amdgpu_device *adev,
+					uint32_t trap_override,
+					uint32_t *trap_mask_supported)
+{
+	*trap_mask_supported &= KFD_DBG_TRAP_MASK_DBG_ADDRESS_WATCH;
+
+	/* The SPI_GDBG_TRAP_MASK register is global and affects all
+	 * processes. Only allow OR-ing the address-watch bit, since
+	 * this only affects processes under the debugger. Other bits
+	 * should stay 0 to avoid the debugger interfering with other
+	 * processes.
+	 */
+	if (trap_override != KFD_DBG_TRAP_OVERRIDE_OR)
+		return -EINVAL;
+
+	return 0;
+}
+
+uint32_t kgd_gfx_v9_set_wave_launch_trap_override(struct amdgpu_device *adev,
+					     uint32_t vmid,
+					     uint32_t trap_override,
+					     uint32_t trap_mask_bits,
+					     uint32_t trap_mask_request,
+					     uint32_t *trap_mask_prev,
+					     uint32_t kfd_dbg_cntl_prev)
+{
+	uint32_t data, wave_cntl_prev;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	wave_cntl_prev = RREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_WAVE_CNTL));
+
+	kgd_gfx_v9_set_wave_launch_stall(adev, vmid, true);
+
+	data = RREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_TRAP_MASK));
+	*trap_mask_prev = REG_GET_FIELD(data, SPI_GDBG_TRAP_MASK, EXCP_EN);
+
+	trap_mask_bits = (trap_mask_bits & trap_mask_request) |
+		(*trap_mask_prev & ~trap_mask_request);
+
+	data = REG_SET_FIELD(data, SPI_GDBG_TRAP_MASK, EXCP_EN, trap_mask_bits);
+	data = REG_SET_FIELD(data, SPI_GDBG_TRAP_MASK, REPLACE, trap_override);
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_TRAP_MASK), data);
+
+	/* We need to preserve wave launch mode stall settings. */
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_WAVE_CNTL), wave_cntl_prev);
+
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+uint32_t kgd_gfx_v9_set_wave_launch_mode(struct amdgpu_device *adev,
+					uint8_t wave_launch_mode,
+					uint32_t vmid)
+{
+	uint32_t data = 0;
+	bool is_mode_set = !!wave_launch_mode;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	kgd_gfx_v9_set_wave_launch_stall(adev, vmid, true);
+
+	data = REG_SET_FIELD(data, SPI_GDBG_WAVE_CNTL2,
+		VMID_MASK, is_mode_set ? 1 << vmid : 0);
+	data = REG_SET_FIELD(data, SPI_GDBG_WAVE_CNTL2,
+		MODE, is_mode_set ? wave_launch_mode : 0);
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_WAVE_CNTL2), data);
+
+	kgd_gfx_v9_set_wave_launch_stall(adev, vmid, false);
+
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+#define TCP_WATCH_STRIDE (mmTCP_WATCH1_ADDR_H - mmTCP_WATCH0_ADDR_H)
+uint32_t kgd_gfx_v9_set_address_watch(struct amdgpu_device *adev,
+					uint64_t watch_address,
+					uint32_t watch_address_mask,
+					uint32_t watch_id,
+					uint32_t watch_mode,
+					uint32_t debug_vmid,
+					uint32_t inst)
+{
+	uint32_t watch_address_high;
+	uint32_t watch_address_low;
+	uint32_t watch_address_cntl;
+
+	watch_address_cntl = 0;
+
+	watch_address_low = lower_32_bits(watch_address);
+	watch_address_high = upper_32_bits(watch_address) & 0xffff;
+
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			VMID,
+			debug_vmid);
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			MODE,
+			watch_mode);
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			MASK,
+			watch_address_mask >> 6);
+
+	/* Turning off this watch point until we set all the registers */
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			VALID,
+			0);
+
+	WREG32_RLC((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_CNTL) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_cntl);
+
+	WREG32_RLC((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_ADDR_H) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_high);
+
+	WREG32_RLC((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_ADDR_L) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_low);
+
+	/* Enable the watch point */
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			VALID,
+			1);
+
+	WREG32_RLC((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_CNTL) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_cntl);
+
+	return 0;
+}
+
+uint32_t kgd_gfx_v9_clear_address_watch(struct amdgpu_device *adev,
+					uint32_t watch_id)
+{
+	uint32_t watch_address_cntl;
+
+	watch_address_cntl = 0;
+
+	WREG32_RLC((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_CNTL) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_cntl);
+
+	return 0;
+}
+
+/* kgd_gfx_v9_get_iq_wait_times: Returns the mmCP_IQ_WAIT_TIME1/2 values
+ * The values read are:
+ *     ib_offload_wait_time     -- Wait Count for Indirect Buffer Offloads.
+ *     atomic_offload_wait_time -- Wait Count for L2 and GDS Atomics Offloads.
+ *     wrm_offload_wait_time    -- Wait Count for WAIT_REG_MEM Offloads.
+ *     gws_wait_time            -- Wait Count for Global Wave Syncs.
+ *     que_sleep_wait_time      -- Wait Count for Dequeue Retry.
+ *     sch_wave_wait_time       -- Wait Count for Scheduling Wave Message.
+ *     sem_rearm_wait_time      -- Wait Count for Semaphore re-arm.
+ *     deq_retry_wait_time      -- Wait Count for Global Wave Syncs.
+ */
+void kgd_gfx_v9_get_iq_wait_times(struct amdgpu_device *adev,
+					uint32_t *wait_times,
+					uint32_t inst)
+
+{
+	*wait_times = RREG32_SOC15_RLC(GC, GET_INST(GC, inst),
+			mmCP_IQ_WAIT_TIME2);
+}
+
+void kgd_gfx_v9_set_vm_context_page_table_base(struct amdgpu_device *adev,
+			uint32_t vmid, uint64_t page_table_base)
+{
+	if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid)) {
+		pr_err("trying to set page table base for wrong VMID %u\n",
+		       vmid);
+		return;
+	}
+
+	adev->mmhub.funcs->setup_vm_pt_regs(adev, vmid, page_table_base);
+
+	adev->gfxhub.funcs->setup_vm_pt_regs(adev, vmid, page_table_base);
+}
+
+static void lock_spi_csq_mutexes(struct amdgpu_device *adev)
+{
+	mutex_lock(&adev->srbm_mutex);
+	mutex_lock(&adev->grbm_idx_mutex);
+
+}
+
+static void unlock_spi_csq_mutexes(struct amdgpu_device *adev)
+{
+	mutex_unlock(&adev->grbm_idx_mutex);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+/**
+ * get_wave_count: Read device registers to get number of waves in flight for
+ * a particular queue. The method also returns the VMID associated with the
+ * queue.
+ *
+ * @adev: Handle of device whose registers are to be read
+ * @queue_idx: Index of queue in the queue-map bit-field
+ * @queue_cnt: Stores the wave count and doorbell offset for an active queue
+ * @inst: xcc's instance number on a multi-XCC setup
+ */
+static void get_wave_count(struct amdgpu_device *adev, int queue_idx,
+		struct kfd_cu_occupancy *queue_cnt, uint32_t inst)
+{
+	int pipe_idx;
+	int queue_slot;
+	unsigned int reg_val;
+	unsigned int wave_cnt;
+	/*
+	 * Program GRBM with appropriate MEID, PIPEID, QUEUEID and VMID
+	 * parameters to read out waves in flight. Get VMID if there are
+	 * non-zero waves in flight.
+	 */
+	pipe_idx = queue_idx / adev->gfx.mec.num_queue_per_pipe;
+	queue_slot = queue_idx % adev->gfx.mec.num_queue_per_pipe;
+	soc15_grbm_select(adev, 1, pipe_idx, queue_slot, 0, GET_INST(GC, inst));
+	reg_val = RREG32_SOC15_IP(GC, SOC15_REG_OFFSET(GC, GET_INST(GC, inst),
+				  mmSPI_CSQ_WF_ACTIVE_COUNT_0) + queue_slot);
+	wave_cnt = reg_val & SPI_CSQ_WF_ACTIVE_COUNT_0__COUNT_MASK;
+	if (wave_cnt != 0) {
+		queue_cnt->wave_cnt += wave_cnt;
+		queue_cnt->doorbell_off =
+			(RREG32_SOC15(GC, GET_INST(GC, inst), mmCP_HQD_PQ_DOORBELL_CONTROL) &
+			 CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_OFFSET_MASK) >>
+			 CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_OFFSET__SHIFT;
+	}
+}
+
+/**
+ * kgd_gfx_v9_get_cu_occupancy: Reads relevant registers associated with each
+ * shader engine and aggregates the number of waves that are in flight for the
+ * process whose pasid is provided as a parameter. The process could have ZERO
+ * or more queues running and submitting waves to compute units.
+ *
+ * @adev: Handle of device from which to get number of waves in flight
+ * @cu_occupancy: Array that gets filled with wave_cnt and doorbell offset
+ *		  for comparison later.
+ * @max_waves_per_cu: Output parameter updated with maximum number of waves
+ *                    possible per Compute Unit
+ * @inst: xcc's instance number on a multi-XCC setup
+ *
+ * Note: It's possible that the device has too many queues (oversubscription)
+ * in which case a VMID could be remapped to a different PASID. This could lead
+ * to an inaccurate wave count. Following is a high-level sequence:
+ *    Time T1: vmid = getVmid(); vmid is associated with Pasid P1
+ *    Time T2: passId = getPasId(vmid); vmid is associated with Pasid P2
+ * In the sequence above wave count obtained from time T1 will be incorrectly
+ * lost or added to total wave count.
+ *
+ * The registers that provide the waves in flight are:
+ *
+ *  SPI_CSQ_WF_ACTIVE_STATUS - bit-map of queues per pipe. The bit is ON if a
+ *  queue is slotted, OFF if there is no queue. A process could have ZERO or
+ *  more queues slotted and submitting waves to be run on compute units. Even
+ *  when there is a queue it is possible there could be zero wave fronts, this
+ *  can happen when queue is waiting on top-of-pipe events - e.g. waitRegMem
+ *  command
+ *
+ *  For each bit that is ON from above:
+ *
+ *    Read (SPI_CSQ_WF_ACTIVE_COUNT_0 + queue_idx) register. It provides the
+ *    number of waves that are in flight for the queue at specified index. The
+ *    index ranges from 0 to 7.
+ *
+ *    If non-zero waves are in flight, store the corresponding doorbell offset
+ *    of the queue, along with the wave count.
+ *
+ *    Determine if the queue belongs to the process by comparing the doorbell
+ *    offset against the process's queues. If it matches, aggregate the wave
+ *    count for the process.
+ *
+ *  Reading registers referenced above involves programming GRBM appropriately
+ */
+void kgd_gfx_v9_get_cu_occupancy(struct amdgpu_device *adev,
+				 struct kfd_cu_occupancy *cu_occupancy,
+				 int *max_waves_per_cu, uint32_t inst)
+{
+	int qidx;
+	int se_idx;
+	int se_cnt;
+	int queue_map;
+	int max_queue_cnt;
+	DECLARE_BITMAP(cp_queue_bitmap, AMDGPU_MAX_QUEUES);
+
+	lock_spi_csq_mutexes(adev);
+	soc15_grbm_select(adev, 1, 0, 0, 0, GET_INST(GC, inst));
+
+	/*
+	 * Iterate through the shader engines and arrays of the device
+	 * to get number of waves in flight
+	 */
+	bitmap_complement(cp_queue_bitmap, adev->gfx.mec_bitmap[0].queue_bitmap,
+			  AMDGPU_MAX_QUEUES);
+	max_queue_cnt = adev->gfx.mec.num_pipe_per_mec *
+			adev->gfx.mec.num_queue_per_pipe;
+	se_cnt = adev->gfx.config.max_shader_engines;
+	for (se_idx = 0; se_idx < se_cnt; se_idx++) {
+		amdgpu_gfx_select_se_sh(adev, se_idx, 0, 0xffffffff, inst);
+		queue_map = RREG32_SOC15(GC, GET_INST(GC, inst), mmSPI_CSQ_WF_ACTIVE_STATUS);
+
+		/*
+		 * Assumption: queue map encodes following schema: four
+		 * pipes per each micro-engine, with each pipe mapping
+		 * eight queues. This schema is true for GFX9 devices
+		 * and must be verified for newer device families
+		 */
+		for (qidx = 0; qidx < max_queue_cnt; qidx++) {
+			/* Skip qeueus that are not associated with
+			 * compute functions
+			 */
+			if (!test_bit(qidx, cp_queue_bitmap))
+				continue;
+
+			if (!(queue_map & (1 << qidx)))
+				continue;
+
+			/* Get number of waves in flight and aggregate them */
+			get_wave_count(adev, qidx, &cu_occupancy[qidx],
+					inst);
+		}
+	}
+
+	amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, inst);
+	soc15_grbm_select(adev, 0, 0, 0, 0, GET_INST(GC, inst));
+	unlock_spi_csq_mutexes(adev);
+
+	/* Update the output parameters and return */
+	*max_waves_per_cu = adev->gfx.cu_info.simd_per_cu *
+				adev->gfx.cu_info.max_waves_per_simd;
+}
+
+void kgd_gfx_v9_build_dequeue_wait_counts_packet_info(struct amdgpu_device *adev,
+		uint32_t wait_times,
+		uint32_t sch_wave,
+		uint32_t que_sleep,
+		uint32_t *reg_offset,
+		uint32_t *reg_data)
+{
+	*reg_data = wait_times;
+
+	if (sch_wave)
+		*reg_data = REG_SET_FIELD(*reg_data,
+				CP_IQ_WAIT_TIME2,
+				SCH_WAVE,
+				sch_wave);
+	if (que_sleep)
+		*reg_data = REG_SET_FIELD(*reg_data,
+				CP_IQ_WAIT_TIME2,
+				QUE_SLEEP,
+				que_sleep);
+
+	*reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_IQ_WAIT_TIME2);
+}
+
+void kgd_gfx_v9_program_trap_handler_settings(struct amdgpu_device *adev,
+		uint32_t vmid, uint64_t tba_addr, uint64_t tma_addr, uint32_t inst)
+{
+	kgd_gfx_v9_lock_srbm(adev, 0, 0, 0, vmid, inst);
+
+	/*
+	 * Program TBA registers
+	 */
+	WREG32_SOC15(GC, GET_INST(GC, inst), mmSQ_SHADER_TBA_LO,
+			lower_32_bits(tba_addr >> 8));
+	WREG32_SOC15(GC, GET_INST(GC, inst), mmSQ_SHADER_TBA_HI,
+			upper_32_bits(tba_addr >> 8));
+
+	/*
+	 * Program TMA registers
+	 */
+	WREG32_SOC15(GC, GET_INST(GC, inst), mmSQ_SHADER_TMA_LO,
+			lower_32_bits(tma_addr >> 8));
+	WREG32_SOC15(GC, GET_INST(GC, inst), mmSQ_SHADER_TMA_HI,
+			upper_32_bits(tma_addr >> 8));
+
+	kgd_gfx_v9_unlock_srbm(adev, inst);
+}
+
+uint64_t kgd_gfx_v9_hqd_get_pq_addr(struct amdgpu_device *adev,
+				    uint32_t pipe_id, uint32_t queue_id,
+				    uint32_t inst)
+{
+	uint32_t low, high;
+	uint64_t queue_addr = 0;
+
+	kgd_gfx_v9_acquire_queue(adev, pipe_id, queue_id, inst);
+	amdgpu_gfx_rlc_enter_safe_mode(adev, inst);
+
+	if (!RREG32_SOC15(GC, GET_INST(GC, inst), mmCP_HQD_ACTIVE))
+		goto unlock_out;
+
+	low = RREG32_SOC15(GC, GET_INST(GC, inst), mmCP_HQD_PQ_BASE);
+	high = RREG32_SOC15(GC, GET_INST(GC, inst), mmCP_HQD_PQ_BASE_HI);
+
+	/* only concerned with user queues. */
+	if (!high)
+		goto unlock_out;
+
+	queue_addr = (((queue_addr | high) << 32) | low) << 8;
+
+unlock_out:
+	amdgpu_gfx_rlc_exit_safe_mode(adev, inst);
+	kgd_gfx_v9_release_queue(adev, inst);
+
+	return queue_addr;
+}
+
+/* assume queue acquired  */
+static int kgd_gfx_v9_hqd_dequeue_wait(struct amdgpu_device *adev, uint32_t inst,
+				       unsigned int utimeout)
+{
+	unsigned long end_jiffies = (utimeout * HZ / 1000) + jiffies;
+
+	while (true) {
+		uint32_t temp = RREG32_SOC15(GC, GET_INST(GC, inst), mmCP_HQD_ACTIVE);
+
+		if (!(temp & CP_HQD_ACTIVE__ACTIVE_MASK))
+			return 0;
+
+		if (time_after(jiffies, end_jiffies))
+			return -ETIME;
+
+		usleep_range(500, 1000);
+	}
+}
+
+uint64_t kgd_gfx_v9_hqd_reset(struct amdgpu_device *adev,
+			      uint32_t pipe_id, uint32_t queue_id,
+			      uint32_t inst, unsigned int utimeout)
+{
+	uint32_t low, high, pipe_reset_data = 0;
+	uint64_t queue_addr = 0;
+
+	kgd_gfx_v9_acquire_queue(adev, pipe_id, queue_id, inst);
+	amdgpu_gfx_rlc_enter_safe_mode(adev, inst);
+
+	if (!RREG32_SOC15(GC, GET_INST(GC, inst), mmCP_HQD_ACTIVE))
+		goto unlock_out;
+
+	low = RREG32_SOC15(GC, GET_INST(GC, inst), mmCP_HQD_PQ_BASE);
+	high = RREG32_SOC15(GC, GET_INST(GC, inst), mmCP_HQD_PQ_BASE_HI);
+
+	/* only concerned with user queues. */
+	if (!high)
+		goto unlock_out;
+
+	queue_addr = (((queue_addr | high) << 32) | low) << 8;
+
+	pr_debug("Attempting queue reset on XCC %i pipe id %i queue id %i\n",
+		 inst, pipe_id, queue_id);
+
+	/* assume previous dequeue request issued will take affect after reset */
+	WREG32_SOC15(GC, GET_INST(GC, inst), mmSPI_COMPUTE_QUEUE_RESET, 0x1);
+
+	if (!kgd_gfx_v9_hqd_dequeue_wait(adev, inst, utimeout))
+		goto unlock_out;
+
+	pr_debug("Attempting pipe reset on XCC %i pipe id %i\n", inst, pipe_id);
+
+	pipe_reset_data = REG_SET_FIELD(pipe_reset_data, CP_MEC_CNTL, MEC_ME1_PIPE0_RESET, 1);
+	pipe_reset_data = pipe_reset_data << pipe_id;
+
+	WREG32_SOC15(GC, GET_INST(GC, inst), mmCP_MEC_CNTL, pipe_reset_data);
+	WREG32_SOC15(GC, GET_INST(GC, inst), mmCP_MEC_CNTL, 0);
+
+	if (kgd_gfx_v9_hqd_dequeue_wait(adev, inst, utimeout))
+		queue_addr = 0;
+
+unlock_out:
+	pr_debug("queue reset on XCC %i pipe id %i queue id %i %s\n",
+		 inst, pipe_id, queue_id, !!queue_addr ? "succeeded!" : "failed!");
+	amdgpu_gfx_rlc_exit_safe_mode(adev, inst);
+	kgd_gfx_v9_release_queue(adev, inst);
+
+	return queue_addr;
+}
+
+uint32_t kgd_gfx_v9_hqd_sdma_get_doorbell(struct amdgpu_device *adev,
+					  int engine, int queue)
+
+{
+	return 0;
+}
+
+const struct kfd2kgd_calls gfx_v9_kfd2kgd = {
+	.program_sh_mem_settings = kgd_gfx_v9_program_sh_mem_settings,
+	.set_pasid_vmid_mapping = kgd_gfx_v9_set_pasid_vmid_mapping,
+	.init_interrupts = kgd_gfx_v9_init_interrupts,
+	.hqd_load = kgd_gfx_v9_hqd_load,
+	.hiq_mqd_load = kgd_gfx_v9_hiq_mqd_load,
+	.hqd_sdma_load = kgd_hqd_sdma_load,
+	.hqd_dump = kgd_gfx_v9_hqd_dump,
+	.hqd_sdma_dump = kgd_hqd_sdma_dump,
+	.hqd_is_occupied = kgd_gfx_v9_hqd_is_occupied,
+	.hqd_sdma_is_occupied = kgd_hqd_sdma_is_occupied,
+	.hqd_destroy = kgd_gfx_v9_hqd_destroy,
+	.hqd_sdma_destroy = kgd_hqd_sdma_destroy,
+	.wave_control_execute = kgd_gfx_v9_wave_control_execute,
+	.get_atc_vmid_pasid_mapping_info =
+			kgd_gfx_v9_get_atc_vmid_pasid_mapping_info,
+	.set_vm_context_page_table_base = kgd_gfx_v9_set_vm_context_page_table_base,
+	.enable_debug_trap = kgd_gfx_v9_enable_debug_trap,
+	.disable_debug_trap = kgd_gfx_v9_disable_debug_trap,
+	.validate_trap_override_request = kgd_gfx_v9_validate_trap_override_request,
+	.set_wave_launch_trap_override = kgd_gfx_v9_set_wave_launch_trap_override,
+	.set_wave_launch_mode = kgd_gfx_v9_set_wave_launch_mode,
+	.set_address_watch = kgd_gfx_v9_set_address_watch,
+	.clear_address_watch = kgd_gfx_v9_clear_address_watch,
+	.get_iq_wait_times = kgd_gfx_v9_get_iq_wait_times,
+	.build_dequeue_wait_counts_packet_info = kgd_gfx_v9_build_dequeue_wait_counts_packet_info,
+	.get_cu_occupancy = kgd_gfx_v9_get_cu_occupancy,
+	.program_trap_handler_settings = kgd_gfx_v9_program_trap_handler_settings,
+	.hqd_get_pq_addr = kgd_gfx_v9_hqd_get_pq_addr,
+	.hqd_reset = kgd_gfx_v9_hqd_reset,
+	.hqd_sdma_get_doorbell = kgd_gfx_v9_hqd_sdma_get_doorbell
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.h
new file mode 100644
index 000000000000..704452ca62f8
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.h
@@ -0,0 +1,116 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+void kgd_gfx_v9_program_sh_mem_settings(struct amdgpu_device *adev, uint32_t vmid,
+		uint32_t sh_mem_config,
+		uint32_t sh_mem_ape1_base, uint32_t sh_mem_ape1_limit,
+		uint32_t sh_mem_bases, uint32_t inst);
+int kgd_gfx_v9_set_pasid_vmid_mapping(struct amdgpu_device *adev, u32 pasid,
+		unsigned int vmid, uint32_t inst);
+int kgd_gfx_v9_init_interrupts(struct amdgpu_device *adev, uint32_t pipe_id,
+				uint32_t inst);
+int kgd_gfx_v9_hqd_load(struct amdgpu_device *adev, void *mqd, uint32_t pipe_id,
+			uint32_t queue_id, uint32_t __user *wptr,
+			uint32_t wptr_shift, uint32_t wptr_mask,
+			struct mm_struct *mm, uint32_t inst);
+int kgd_gfx_v9_hiq_mqd_load(struct amdgpu_device *adev, void *mqd,
+			    uint32_t pipe_id, uint32_t queue_id,
+			    uint32_t doorbell_off, uint32_t inst);
+int kgd_gfx_v9_hqd_dump(struct amdgpu_device *adev,
+			uint32_t pipe_id, uint32_t queue_id,
+			uint32_t (**dump)[2], uint32_t *n_regs, uint32_t inst);
+bool kgd_gfx_v9_hqd_is_occupied(struct amdgpu_device *adev,
+			uint64_t queue_address, uint32_t pipe_id,
+			uint32_t queue_id, uint32_t inst);
+int kgd_gfx_v9_hqd_destroy(struct amdgpu_device *adev, void *mqd,
+				enum kfd_preempt_type reset_type,
+				unsigned int utimeout, uint32_t pipe_id,
+				uint32_t queue_id, uint32_t inst);
+int kgd_gfx_v9_wave_control_execute(struct amdgpu_device *adev,
+					uint32_t gfx_index_val,
+					uint32_t sq_cmd, uint32_t inst);
+bool kgd_gfx_v9_get_atc_vmid_pasid_mapping_info(struct amdgpu_device *adev,
+					uint8_t vmid, uint16_t *p_pasid);
+void kgd_gfx_v9_set_vm_context_page_table_base(struct amdgpu_device *adev,
+			uint32_t vmid, uint64_t page_table_base);
+void kgd_gfx_v9_get_cu_occupancy(struct amdgpu_device *adev,
+				 struct kfd_cu_occupancy *cu_occupancy,
+				 int *max_waves_per_cu, uint32_t inst);
+void kgd_gfx_v9_program_trap_handler_settings(struct amdgpu_device *adev,
+		uint32_t vmid, uint64_t tba_addr, uint64_t tma_addr,
+		uint32_t inst);
+void kgd_gfx_v9_acquire_queue(struct amdgpu_device *adev, uint32_t pipe_id,
+				uint32_t queue_id, uint32_t inst);
+uint64_t kgd_gfx_v9_get_queue_mask(struct amdgpu_device *adev,
+				uint32_t pipe_id, uint32_t queue_id);
+void kgd_gfx_v9_release_queue(struct amdgpu_device *adev, uint32_t inst);
+void kgd_gfx_v9_set_wave_launch_stall(struct amdgpu_device *adev,
+					uint32_t vmid,
+					bool stall);
+uint32_t kgd_gfx_v9_enable_debug_trap(struct amdgpu_device *adev,
+				      bool restore_dbg_registers,
+				      uint32_t vmid);
+uint32_t kgd_gfx_v9_disable_debug_trap(struct amdgpu_device *adev,
+					bool keep_trap_enabled,
+					uint32_t vmid);
+int kgd_gfx_v9_validate_trap_override_request(struct amdgpu_device *adev,
+					     uint32_t trap_override,
+					     uint32_t *trap_mask_supported);
+uint32_t kgd_gfx_v9_set_wave_launch_mode(struct amdgpu_device *adev,
+					uint8_t wave_launch_mode,
+					uint32_t vmid);
+uint32_t kgd_gfx_v9_set_wave_launch_trap_override(struct amdgpu_device *adev,
+					     uint32_t vmid,
+					     uint32_t trap_override,
+					     uint32_t trap_mask_bits,
+					     uint32_t trap_mask_request,
+					     uint32_t *trap_mask_prev,
+					     uint32_t kfd_dbg_trap_cntl_prev);
+uint32_t kgd_gfx_v9_set_address_watch(struct amdgpu_device *adev,
+					uint64_t watch_address,
+					uint32_t watch_address_mask,
+					uint32_t watch_id,
+					uint32_t watch_mode,
+					uint32_t debug_vmid,
+					uint32_t inst);
+uint32_t kgd_gfx_v9_clear_address_watch(struct amdgpu_device *adev,
+					uint32_t watch_id);
+void kgd_gfx_v9_get_iq_wait_times(struct amdgpu_device *adev,
+				uint32_t *wait_times,
+				uint32_t inst);
+void kgd_gfx_v9_build_dequeue_wait_counts_packet_info(struct amdgpu_device *adev,
+					       uint32_t wait_times,
+					       uint32_t sch_wave,
+					       uint32_t que_sleep,
+					       uint32_t *reg_offset,
+					       uint32_t *reg_data);
+uint64_t kgd_gfx_v9_hqd_get_pq_addr(struct amdgpu_device *adev,
+				    uint32_t pipe_id,
+				    uint32_t queue_id,
+				    uint32_t inst);
+uint64_t kgd_gfx_v9_hqd_reset(struct amdgpu_device *adev,
+			      uint32_t pipe_id,
+			      uint32_t queue_id,
+			      uint32_t inst,
+			      unsigned int utimeout);
+uint32_t kgd_gfx_v9_hqd_sdma_get_doorbell(struct amdgpu_device *adev,
+					  int engine, int queue);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
new file mode 100644
index 000000000000..83020963dfde
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
@@ -0,0 +1,3239 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2014-2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+#include <linux/dma-buf.h>
+#include <linux/list.h>
+#include <linux/pagemap.h>
+#include <linux/sched/mm.h>
+#include <linux/sched/task.h>
+#include <drm/ttm/ttm_tt.h>
+
+#include <drm/drm_exec.h>
+
+#include "amdgpu_object.h"
+#include "amdgpu_gem.h"
+#include "amdgpu_vm.h"
+#include "amdgpu_hmm.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_dma_buf.h"
+#include <uapi/linux/kfd_ioctl.h>
+#include "amdgpu_xgmi.h"
+#include "kfd_priv.h"
+#include "kfd_smi_events.h"
+
+/* Userptr restore delay, just long enough to allow consecutive VM
+ * changes to accumulate
+ */
+#define AMDGPU_USERPTR_RESTORE_DELAY_MS 1
+#define AMDGPU_RESERVE_MEM_LIMIT			(3UL << 29)
+
+/*
+ * Align VRAM availability to 2MB to avoid fragmentation caused by 4K allocations in the tail 2MB
+ * BO chunk
+ */
+#define VRAM_AVAILABLITY_ALIGN (1 << 21)
+
+/* Impose limit on how much memory KFD can use */
+static struct {
+	uint64_t max_system_mem_limit;
+	uint64_t max_ttm_mem_limit;
+	int64_t system_mem_used;
+	int64_t ttm_mem_used;
+	spinlock_t mem_limit_lock;
+} kfd_mem_limit;
+
+static const char * const domain_bit_to_string[] = {
+		"CPU",
+		"GTT",
+		"VRAM",
+		"GDS",
+		"GWS",
+		"OA"
+};
+
+#define domain_string(domain) domain_bit_to_string[ffs(domain)-1]
+
+static void amdgpu_amdkfd_restore_userptr_worker(struct work_struct *work);
+
+static bool kfd_mem_is_attached(struct amdgpu_vm *avm,
+		struct kgd_mem *mem)
+{
+	struct kfd_mem_attachment *entry;
+
+	list_for_each_entry(entry, &mem->attachments, list)
+		if (entry->bo_va->base.vm == avm)
+			return true;
+
+	return false;
+}
+
+/**
+ * reuse_dmamap() - Check whether adev can share the original
+ * userptr BO
+ *
+ * If both adev and bo_adev are in direct mapping or
+ * in the same iommu group, they can share the original BO.
+ *
+ * @adev: Device to which can or cannot share the original BO
+ * @bo_adev: Device to which allocated BO belongs to
+ *
+ * Return: returns true if adev can share original userptr BO,
+ * false otherwise.
+ */
+static bool reuse_dmamap(struct amdgpu_device *adev, struct amdgpu_device *bo_adev)
+{
+	return (adev->ram_is_direct_mapped && bo_adev->ram_is_direct_mapped) ||
+			(adev->dev->iommu_group == bo_adev->dev->iommu_group);
+}
+
+/* Set memory usage limits. Current, limits are
+ *  System (TTM + userptr) memory - 15/16th System RAM
+ *  TTM memory - 3/8th System RAM
+ */
+void amdgpu_amdkfd_gpuvm_init_mem_limits(void)
+{
+	struct sysinfo si;
+	uint64_t mem;
+
+	if (kfd_mem_limit.max_system_mem_limit)
+		return;
+
+	si_meminfo(&si);
+	mem = si.totalram - si.totalhigh;
+	mem *= si.mem_unit;
+
+	spin_lock_init(&kfd_mem_limit.mem_limit_lock);
+	kfd_mem_limit.max_system_mem_limit = mem - (mem >> 6);
+	if (kfd_mem_limit.max_system_mem_limit < 2 * AMDGPU_RESERVE_MEM_LIMIT)
+		kfd_mem_limit.max_system_mem_limit >>= 1;
+	else
+		kfd_mem_limit.max_system_mem_limit -= AMDGPU_RESERVE_MEM_LIMIT;
+
+	kfd_mem_limit.max_ttm_mem_limit = ttm_tt_pages_limit() << PAGE_SHIFT;
+	pr_debug("Kernel memory limit %lluM, TTM limit %lluM\n",
+		(kfd_mem_limit.max_system_mem_limit >> 20),
+		(kfd_mem_limit.max_ttm_mem_limit >> 20));
+}
+
+void amdgpu_amdkfd_reserve_system_mem(uint64_t size)
+{
+	kfd_mem_limit.system_mem_used += size;
+}
+
+/* Estimate page table size needed to represent a given memory size
+ *
+ * With 4KB pages, we need one 8 byte PTE for each 4KB of memory
+ * (factor 512, >> 9). With 2MB pages, we need one 8 byte PTE for 2MB
+ * of memory (factor 256K, >> 18). ROCm user mode tries to optimize
+ * for 2MB pages for TLB efficiency. However, small allocations and
+ * fragmented system memory still need some 4KB pages. We choose a
+ * compromise that should work in most cases without reserving too
+ * much memory for page tables unnecessarily (factor 16K, >> 14).
+ */
+
+#define ESTIMATE_PT_SIZE(mem_size) max(((mem_size) >> 14), AMDGPU_VM_RESERVED_VRAM)
+
+/**
+ * amdgpu_amdkfd_reserve_mem_limit() - Decrease available memory by size
+ * of buffer.
+ *
+ * @adev: Device to which allocated BO belongs to
+ * @size: Size of buffer, in bytes, encapsulated by B0. This should be
+ * equivalent to amdgpu_bo_size(BO)
+ * @alloc_flag: Flag used in allocating a BO as noted above
+ * @xcp_id: xcp_id is used to get xcp from xcp manager, one xcp is
+ * managed as one compute node in driver for app
+ *
+ * Return:
+ *	returns -ENOMEM in case of error, ZERO otherwise
+ */
+int amdgpu_amdkfd_reserve_mem_limit(struct amdgpu_device *adev,
+		uint64_t size, u32 alloc_flag, int8_t xcp_id)
+{
+	uint64_t reserved_for_pt =
+		ESTIMATE_PT_SIZE(amdgpu_amdkfd_total_mem_size);
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	uint64_t reserved_for_ras = (con ? con->reserved_pages_in_bytes : 0);
+	size_t system_mem_needed, ttm_mem_needed, vram_needed;
+	int ret = 0;
+	uint64_t vram_size = 0;
+
+	system_mem_needed = 0;
+	ttm_mem_needed = 0;
+	vram_needed = 0;
+	if (alloc_flag & KFD_IOC_ALLOC_MEM_FLAGS_GTT) {
+		system_mem_needed = size;
+		ttm_mem_needed = size;
+	} else if (alloc_flag & KFD_IOC_ALLOC_MEM_FLAGS_VRAM) {
+		/*
+		 * Conservatively round up the allocation requirement to 2 MB
+		 * to avoid fragmentation caused by 4K allocations in the tail
+		 * 2M BO chunk.
+		 */
+		vram_needed = size;
+		/*
+		 * For GFX 9.4.3, get the VRAM size from XCP structs
+		 */
+		if (WARN_ONCE(xcp_id < 0, "invalid XCP ID %d", xcp_id))
+			return -EINVAL;
+
+		vram_size = KFD_XCP_MEMORY_SIZE(adev, xcp_id);
+		if (adev->apu_prefer_gtt) {
+			system_mem_needed = size;
+			ttm_mem_needed = size;
+		}
+	} else if (alloc_flag & KFD_IOC_ALLOC_MEM_FLAGS_USERPTR) {
+		system_mem_needed = size;
+	} else if (!(alloc_flag &
+				(KFD_IOC_ALLOC_MEM_FLAGS_DOORBELL |
+				 KFD_IOC_ALLOC_MEM_FLAGS_MMIO_REMAP))) {
+		pr_err("%s: Invalid BO type %#x\n", __func__, alloc_flag);
+		return -ENOMEM;
+	}
+
+	spin_lock(&kfd_mem_limit.mem_limit_lock);
+
+	if (kfd_mem_limit.system_mem_used + system_mem_needed >
+	    kfd_mem_limit.max_system_mem_limit) {
+		pr_debug("Set no_system_mem_limit=1 if using shared memory\n");
+		if (!no_system_mem_limit) {
+			ret = -ENOMEM;
+			goto release;
+		}
+	}
+
+	if (kfd_mem_limit.ttm_mem_used + ttm_mem_needed >
+		kfd_mem_limit.max_ttm_mem_limit) {
+		ret = -ENOMEM;
+		goto release;
+	}
+
+	/*if is_app_apu is false and apu_prefer_gtt is true, it is an APU with
+	 * carve out < gtt. In that case, VRAM allocation will go to gtt domain, skip
+	 * VRAM check since ttm_mem_limit check already cover this allocation
+	 */
+
+	if (adev && xcp_id >= 0 && (!adev->apu_prefer_gtt || adev->gmc.is_app_apu)) {
+		uint64_t vram_available =
+			vram_size - reserved_for_pt - reserved_for_ras -
+			atomic64_read(&adev->vram_pin_size);
+		if (adev->kfd.vram_used[xcp_id] + vram_needed > vram_available) {
+			ret = -ENOMEM;
+			goto release;
+		}
+	}
+
+	/* Update memory accounting by decreasing available system
+	 * memory, TTM memory and GPU memory as computed above
+	 */
+	WARN_ONCE(vram_needed && !adev,
+		  "adev reference can't be null when vram is used");
+	if (adev && xcp_id >= 0) {
+		adev->kfd.vram_used[xcp_id] += vram_needed;
+		adev->kfd.vram_used_aligned[xcp_id] +=
+				adev->apu_prefer_gtt ?
+				vram_needed :
+				ALIGN(vram_needed, VRAM_AVAILABLITY_ALIGN);
+	}
+	kfd_mem_limit.system_mem_used += system_mem_needed;
+	kfd_mem_limit.ttm_mem_used += ttm_mem_needed;
+
+release:
+	spin_unlock(&kfd_mem_limit.mem_limit_lock);
+	return ret;
+}
+
+void amdgpu_amdkfd_unreserve_mem_limit(struct amdgpu_device *adev,
+		uint64_t size, u32 alloc_flag, int8_t xcp_id)
+{
+	spin_lock(&kfd_mem_limit.mem_limit_lock);
+
+	if (alloc_flag & KFD_IOC_ALLOC_MEM_FLAGS_GTT) {
+		kfd_mem_limit.system_mem_used -= size;
+		kfd_mem_limit.ttm_mem_used -= size;
+	} else if (alloc_flag & KFD_IOC_ALLOC_MEM_FLAGS_VRAM) {
+		WARN_ONCE(!adev,
+			  "adev reference can't be null when alloc mem flags vram is set");
+		if (WARN_ONCE(xcp_id < 0, "invalid XCP ID %d", xcp_id))
+			goto release;
+
+		if (adev) {
+			adev->kfd.vram_used[xcp_id] -= size;
+			if (adev->apu_prefer_gtt) {
+				adev->kfd.vram_used_aligned[xcp_id] -= size;
+				kfd_mem_limit.system_mem_used -= size;
+				kfd_mem_limit.ttm_mem_used -= size;
+			} else {
+				adev->kfd.vram_used_aligned[xcp_id] -=
+					ALIGN(size, VRAM_AVAILABLITY_ALIGN);
+			}
+		}
+	} else if (alloc_flag & KFD_IOC_ALLOC_MEM_FLAGS_USERPTR) {
+		kfd_mem_limit.system_mem_used -= size;
+	} else if (!(alloc_flag &
+				(KFD_IOC_ALLOC_MEM_FLAGS_DOORBELL |
+				 KFD_IOC_ALLOC_MEM_FLAGS_MMIO_REMAP))) {
+		pr_err("%s: Invalid BO type %#x\n", __func__, alloc_flag);
+		goto release;
+	}
+	WARN_ONCE(adev && xcp_id >= 0 && adev->kfd.vram_used[xcp_id] < 0,
+		  "KFD VRAM memory accounting unbalanced for xcp: %d", xcp_id);
+	WARN_ONCE(kfd_mem_limit.ttm_mem_used < 0,
+		  "KFD TTM memory accounting unbalanced");
+	WARN_ONCE(kfd_mem_limit.system_mem_used < 0,
+		  "KFD system memory accounting unbalanced");
+
+release:
+	spin_unlock(&kfd_mem_limit.mem_limit_lock);
+}
+
+void amdgpu_amdkfd_release_notify(struct amdgpu_bo *bo)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	u32 alloc_flags = bo->kfd_bo->alloc_flags;
+	u64 size = amdgpu_bo_size(bo);
+
+	amdgpu_amdkfd_unreserve_mem_limit(adev, size, alloc_flags,
+					  bo->xcp_id);
+
+	kfree(bo->kfd_bo);
+}
+
+/**
+ * create_dmamap_sg_bo() - Creates a amdgpu_bo object to reflect information
+ * about USERPTR or DOOREBELL or MMIO BO.
+ *
+ * @adev: Device for which dmamap BO is being created
+ * @mem: BO of peer device that is being DMA mapped. Provides parameters
+ *	 in building the dmamap BO
+ * @bo_out: Output parameter updated with handle of dmamap BO
+ */
+static int
+create_dmamap_sg_bo(struct amdgpu_device *adev,
+		 struct kgd_mem *mem, struct amdgpu_bo **bo_out)
+{
+	struct drm_gem_object *gem_obj;
+	int ret;
+	uint64_t flags = 0;
+
+	ret = amdgpu_bo_reserve(mem->bo, false);
+	if (ret)
+		return ret;
+
+	if (mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_USERPTR)
+		flags |= mem->bo->flags & (AMDGPU_GEM_CREATE_COHERENT |
+					AMDGPU_GEM_CREATE_UNCACHED);
+
+	ret = amdgpu_gem_object_create(adev, mem->bo->tbo.base.size, 1,
+			AMDGPU_GEM_DOMAIN_CPU, AMDGPU_GEM_CREATE_PREEMPTIBLE | flags,
+			ttm_bo_type_sg, mem->bo->tbo.base.resv, &gem_obj, 0);
+
+	amdgpu_bo_unreserve(mem->bo);
+
+	if (ret) {
+		pr_err("Error in creating DMA mappable SG BO on domain: %d\n", ret);
+		return -EINVAL;
+	}
+
+	*bo_out = gem_to_amdgpu_bo(gem_obj);
+	(*bo_out)->parent = amdgpu_bo_ref(mem->bo);
+	return ret;
+}
+
+/* amdgpu_amdkfd_remove_eviction_fence - Removes eviction fence from BO's
+ *  reservation object.
+ *
+ * @bo: [IN] Remove eviction fence(s) from this BO
+ * @ef: [IN] This eviction fence is removed if it
+ *  is present in the shared list.
+ *
+ * NOTE: Must be called with BO reserved i.e. bo->tbo.resv->lock held.
+ */
+static int amdgpu_amdkfd_remove_eviction_fence(struct amdgpu_bo *bo,
+					struct amdgpu_amdkfd_fence *ef)
+{
+	struct dma_fence *replacement;
+
+	if (!ef)
+		return -EINVAL;
+
+	/* TODO: Instead of block before we should use the fence of the page
+	 * table update and TLB flush here directly.
+	 */
+	replacement = dma_fence_get_stub();
+	dma_resv_replace_fences(bo->tbo.base.resv, ef->base.context,
+				replacement, DMA_RESV_USAGE_BOOKKEEP);
+	dma_fence_put(replacement);
+	return 0;
+}
+
+/**
+ * amdgpu_amdkfd_remove_all_eviction_fences - Remove all eviction fences
+ * @bo: the BO where to remove the evictions fences from.
+ *
+ * This functions should only be used on release when all references to the BO
+ * are already dropped. We remove the eviction fence from the private copy of
+ * the dma_resv object here since that is what is used during release to
+ * determine of the BO is idle or not.
+ */
+void amdgpu_amdkfd_remove_all_eviction_fences(struct amdgpu_bo *bo)
+{
+	struct dma_resv *resv = &bo->tbo.base._resv;
+	struct dma_fence *fence, *stub;
+	struct dma_resv_iter cursor;
+
+	dma_resv_assert_held(resv);
+
+	stub = dma_fence_get_stub();
+	dma_resv_for_each_fence(&cursor, resv, DMA_RESV_USAGE_BOOKKEEP, fence) {
+		if (!to_amdgpu_amdkfd_fence(fence))
+			continue;
+
+		dma_resv_replace_fences(resv, fence->context, stub,
+					DMA_RESV_USAGE_BOOKKEEP);
+	}
+	dma_fence_put(stub);
+}
+
+static int amdgpu_amdkfd_bo_validate(struct amdgpu_bo *bo, uint32_t domain,
+				     bool wait)
+{
+	struct ttm_operation_ctx ctx = { false, false };
+	int ret;
+
+	if (WARN(amdgpu_ttm_tt_get_usermm(bo->tbo.ttm),
+		 "Called with userptr BO"))
+		return -EINVAL;
+
+	/* bo has been pinned, not need validate it */
+	if (bo->tbo.pin_count)
+		return 0;
+
+	amdgpu_bo_placement_from_domain(bo, domain);
+
+	ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+	if (ret)
+		goto validate_fail;
+	if (wait)
+		amdgpu_bo_sync_wait(bo, AMDGPU_FENCE_OWNER_KFD, false);
+
+validate_fail:
+	return ret;
+}
+
+int amdgpu_amdkfd_bo_validate_and_fence(struct amdgpu_bo *bo,
+					uint32_t domain,
+					struct dma_fence *fence)
+{
+	int ret = amdgpu_bo_reserve(bo, false);
+
+	if (ret)
+		return ret;
+
+	ret = amdgpu_amdkfd_bo_validate(bo, domain, true);
+	if (ret)
+		goto unreserve_out;
+
+	ret = dma_resv_reserve_fences(bo->tbo.base.resv, 1);
+	if (ret)
+		goto unreserve_out;
+
+	dma_resv_add_fence(bo->tbo.base.resv, fence,
+			   DMA_RESV_USAGE_BOOKKEEP);
+
+unreserve_out:
+	amdgpu_bo_unreserve(bo);
+
+	return ret;
+}
+
+static int amdgpu_amdkfd_validate_vm_bo(void *_unused, struct amdgpu_bo *bo)
+{
+	return amdgpu_amdkfd_bo_validate(bo, bo->allowed_domains, false);
+}
+
+/* vm_validate_pt_pd_bos - Validate page table and directory BOs
+ *
+ * Page directories are not updated here because huge page handling
+ * during page table updates can invalidate page directory entries
+ * again. Page directories are only updated after updating page
+ * tables.
+ */
+static int vm_validate_pt_pd_bos(struct amdgpu_vm *vm,
+				 struct ww_acquire_ctx *ticket)
+{
+	struct amdgpu_bo *pd = vm->root.bo;
+	struct amdgpu_device *adev = amdgpu_ttm_adev(pd->tbo.bdev);
+	int ret;
+
+	ret = amdgpu_vm_validate(adev, vm, ticket,
+				 amdgpu_amdkfd_validate_vm_bo, NULL);
+	if (ret) {
+		pr_err("failed to validate PT BOs\n");
+		return ret;
+	}
+
+	vm->pd_phys_addr = amdgpu_gmc_pd_addr(vm->root.bo);
+
+	return 0;
+}
+
+static int vm_update_pds(struct amdgpu_vm *vm, struct amdgpu_sync *sync)
+{
+	struct amdgpu_bo *pd = vm->root.bo;
+	struct amdgpu_device *adev = amdgpu_ttm_adev(pd->tbo.bdev);
+	int ret;
+
+	ret = amdgpu_vm_update_pdes(adev, vm, false);
+	if (ret)
+		return ret;
+
+	return amdgpu_sync_fence(sync, vm->last_update, GFP_KERNEL);
+}
+
+static uint64_t get_pte_flags(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			      struct kgd_mem *mem)
+{
+	uint32_t mapping_flags = AMDGPU_VM_PAGE_READABLE |
+				 AMDGPU_VM_MTYPE_DEFAULT;
+
+	if (mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_WRITABLE)
+		mapping_flags |= AMDGPU_VM_PAGE_WRITEABLE;
+	if (mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_EXECUTABLE)
+		mapping_flags |= AMDGPU_VM_PAGE_EXECUTABLE;
+
+	return mapping_flags;
+}
+
+/**
+ * create_sg_table() - Create an sg_table for a contiguous DMA addr range
+ * @addr: The starting address to point to
+ * @size: Size of memory area in bytes being pointed to
+ *
+ * Allocates an instance of sg_table and initializes it to point to memory
+ * area specified by input parameters. The address used to build is assumed
+ * to be DMA mapped, if needed.
+ *
+ * DOORBELL or MMIO BOs use only one scatterlist node in their sg_table
+ * because they are physically contiguous.
+ *
+ * Return: Initialized instance of SG Table or NULL
+ */
+static struct sg_table *create_sg_table(uint64_t addr, uint32_t size)
+{
+	struct sg_table *sg = kmalloc(sizeof(*sg), GFP_KERNEL);
+
+	if (!sg)
+		return NULL;
+	if (sg_alloc_table(sg, 1, GFP_KERNEL)) {
+		kfree(sg);
+		return NULL;
+	}
+	sg_dma_address(sg->sgl) = addr;
+	sg->sgl->length = size;
+#ifdef CONFIG_NEED_SG_DMA_LENGTH
+	sg->sgl->dma_length = size;
+#endif
+	return sg;
+}
+
+static int
+kfd_mem_dmamap_userptr(struct kgd_mem *mem,
+		       struct kfd_mem_attachment *attachment)
+{
+	enum dma_data_direction direction =
+		mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_WRITABLE ?
+		DMA_BIDIRECTIONAL : DMA_TO_DEVICE;
+	struct ttm_operation_ctx ctx = {.interruptible = true};
+	struct amdgpu_bo *bo = attachment->bo_va->base.bo;
+	struct amdgpu_device *adev = attachment->adev;
+	struct ttm_tt *src_ttm = mem->bo->tbo.ttm;
+	struct ttm_tt *ttm = bo->tbo.ttm;
+	int ret;
+
+	if (WARN_ON(ttm->num_pages != src_ttm->num_pages))
+		return -EINVAL;
+
+	ttm->sg = kmalloc(sizeof(*ttm->sg), GFP_KERNEL);
+	if (unlikely(!ttm->sg))
+		return -ENOMEM;
+
+	/* Same sequence as in amdgpu_ttm_tt_pin_userptr */
+	ret = sg_alloc_table_from_pages(ttm->sg, src_ttm->pages,
+					ttm->num_pages, 0,
+					(u64)ttm->num_pages << PAGE_SHIFT,
+					GFP_KERNEL);
+	if (unlikely(ret))
+		goto free_sg;
+
+	ret = dma_map_sgtable(adev->dev, ttm->sg, direction, 0);
+	if (unlikely(ret))
+		goto release_sg;
+
+	amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_GTT);
+	ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+	if (ret)
+		goto unmap_sg;
+
+	return 0;
+
+unmap_sg:
+	dma_unmap_sgtable(adev->dev, ttm->sg, direction, 0);
+release_sg:
+	pr_err("DMA map userptr failed: %d\n", ret);
+	sg_free_table(ttm->sg);
+free_sg:
+	kfree(ttm->sg);
+	ttm->sg = NULL;
+	return ret;
+}
+
+static int
+kfd_mem_dmamap_dmabuf(struct kfd_mem_attachment *attachment)
+{
+	struct ttm_operation_ctx ctx = {.interruptible = true};
+	struct amdgpu_bo *bo = attachment->bo_va->base.bo;
+
+	amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_GTT);
+	return ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+}
+
+/**
+ * kfd_mem_dmamap_sg_bo() - Create DMA mapped sg_table to access DOORBELL or MMIO BO
+ * @mem: SG BO of the DOORBELL or MMIO resource on the owning device
+ * @attachment: Virtual address attachment of the BO on accessing device
+ *
+ * An access request from the device that owns DOORBELL does not require DMA mapping.
+ * This is because the request doesn't go through PCIe root complex i.e. it instead
+ * loops back. The need to DMA map arises only when accessing peer device's DOORBELL
+ *
+ * In contrast, all access requests for MMIO need to be DMA mapped without regard to
+ * device ownership. This is because access requests for MMIO go through PCIe root
+ * complex.
+ *
+ * This is accomplished in two steps:
+ *   - Obtain DMA mapped address of DOORBELL or MMIO memory that could be used
+ *         in updating requesting device's page table
+ *   - Signal TTM to mark memory pointed to by requesting device's BO as GPU
+ *         accessible. This allows an update of requesting device's page table
+ *         with entries associated with DOOREBELL or MMIO memory
+ *
+ * This method is invoked in the following contexts:
+ *   - Mapping of DOORBELL or MMIO BO of same or peer device
+ *   - Validating an evicted DOOREBELL or MMIO BO on device seeking access
+ *
+ * Return: ZERO if successful, NON-ZERO otherwise
+ */
+static int
+kfd_mem_dmamap_sg_bo(struct kgd_mem *mem,
+		     struct kfd_mem_attachment *attachment)
+{
+	struct ttm_operation_ctx ctx = {.interruptible = true};
+	struct amdgpu_bo *bo = attachment->bo_va->base.bo;
+	struct amdgpu_device *adev = attachment->adev;
+	struct ttm_tt *ttm = bo->tbo.ttm;
+	enum dma_data_direction dir;
+	dma_addr_t dma_addr;
+	bool mmio;
+	int ret;
+
+	/* Expect SG Table of dmapmap BO to be NULL */
+	mmio = (mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_MMIO_REMAP);
+	if (unlikely(ttm->sg)) {
+		pr_err("SG Table of %d BO for peer device is UNEXPECTEDLY NON-NULL", mmio);
+		return -EINVAL;
+	}
+
+	dir = mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_WRITABLE ?
+			DMA_BIDIRECTIONAL : DMA_TO_DEVICE;
+	dma_addr = mem->bo->tbo.sg->sgl->dma_address;
+	pr_debug("%d BO size: %d\n", mmio, mem->bo->tbo.sg->sgl->length);
+	pr_debug("%d BO address before DMA mapping: %llx\n", mmio, dma_addr);
+	dma_addr = dma_map_resource(adev->dev, dma_addr,
+			mem->bo->tbo.sg->sgl->length, dir, DMA_ATTR_SKIP_CPU_SYNC);
+	ret = dma_mapping_error(adev->dev, dma_addr);
+	if (unlikely(ret))
+		return ret;
+	pr_debug("%d BO address after DMA mapping: %llx\n", mmio, dma_addr);
+
+	ttm->sg = create_sg_table(dma_addr, mem->bo->tbo.sg->sgl->length);
+	if (unlikely(!ttm->sg)) {
+		ret = -ENOMEM;
+		goto unmap_sg;
+	}
+
+	amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_GTT);
+	ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+	if (unlikely(ret))
+		goto free_sg;
+
+	return ret;
+
+free_sg:
+	sg_free_table(ttm->sg);
+	kfree(ttm->sg);
+	ttm->sg = NULL;
+unmap_sg:
+	dma_unmap_resource(adev->dev, dma_addr, mem->bo->tbo.sg->sgl->length,
+			   dir, DMA_ATTR_SKIP_CPU_SYNC);
+	return ret;
+}
+
+static int
+kfd_mem_dmamap_attachment(struct kgd_mem *mem,
+			  struct kfd_mem_attachment *attachment)
+{
+	switch (attachment->type) {
+	case KFD_MEM_ATT_SHARED:
+		return 0;
+	case KFD_MEM_ATT_USERPTR:
+		return kfd_mem_dmamap_userptr(mem, attachment);
+	case KFD_MEM_ATT_DMABUF:
+		return kfd_mem_dmamap_dmabuf(attachment);
+	case KFD_MEM_ATT_SG:
+		return kfd_mem_dmamap_sg_bo(mem, attachment);
+	default:
+		WARN_ON_ONCE(1);
+	}
+	return -EINVAL;
+}
+
+static void
+kfd_mem_dmaunmap_userptr(struct kgd_mem *mem,
+			 struct kfd_mem_attachment *attachment)
+{
+	enum dma_data_direction direction =
+		mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_WRITABLE ?
+		DMA_BIDIRECTIONAL : DMA_TO_DEVICE;
+	struct ttm_operation_ctx ctx = {.interruptible = false};
+	struct amdgpu_bo *bo = attachment->bo_va->base.bo;
+	struct amdgpu_device *adev = attachment->adev;
+	struct ttm_tt *ttm = bo->tbo.ttm;
+
+	if (unlikely(!ttm->sg))
+		return;
+
+	amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_CPU);
+	(void)ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+
+	dma_unmap_sgtable(adev->dev, ttm->sg, direction, 0);
+	sg_free_table(ttm->sg);
+	kfree(ttm->sg);
+	ttm->sg = NULL;
+}
+
+static void
+kfd_mem_dmaunmap_dmabuf(struct kfd_mem_attachment *attachment)
+{
+	/* This is a no-op. We don't want to trigger eviction fences when
+	 * unmapping DMABufs. Therefore the invalidation (moving to system
+	 * domain) is done in kfd_mem_dmamap_dmabuf.
+	 */
+}
+
+/**
+ * kfd_mem_dmaunmap_sg_bo() - Free DMA mapped sg_table of DOORBELL or MMIO BO
+ * @mem: SG BO of the DOORBELL or MMIO resource on the owning device
+ * @attachment: Virtual address attachment of the BO on accessing device
+ *
+ * The method performs following steps:
+ *   - Signal TTM to mark memory pointed to by BO as GPU inaccessible
+ *   - Free SG Table that is used to encapsulate DMA mapped memory of
+ *          peer device's DOORBELL or MMIO memory
+ *
+ * This method is invoked in the following contexts:
+ *     UNMapping of DOORBELL or MMIO BO on a device having access to its memory
+ *     Eviction of DOOREBELL or MMIO BO on device having access to its memory
+ *
+ * Return: void
+ */
+static void
+kfd_mem_dmaunmap_sg_bo(struct kgd_mem *mem,
+		       struct kfd_mem_attachment *attachment)
+{
+	struct ttm_operation_ctx ctx = {.interruptible = true};
+	struct amdgpu_bo *bo = attachment->bo_va->base.bo;
+	struct amdgpu_device *adev = attachment->adev;
+	struct ttm_tt *ttm = bo->tbo.ttm;
+	enum dma_data_direction dir;
+
+	if (unlikely(!ttm->sg)) {
+		pr_debug("SG Table of BO is NULL");
+		return;
+	}
+
+	amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_CPU);
+	(void)ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+
+	dir = mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_WRITABLE ?
+				DMA_BIDIRECTIONAL : DMA_TO_DEVICE;
+	dma_unmap_resource(adev->dev, ttm->sg->sgl->dma_address,
+			ttm->sg->sgl->length, dir, DMA_ATTR_SKIP_CPU_SYNC);
+	sg_free_table(ttm->sg);
+	kfree(ttm->sg);
+	ttm->sg = NULL;
+	bo->tbo.sg = NULL;
+}
+
+static void
+kfd_mem_dmaunmap_attachment(struct kgd_mem *mem,
+			    struct kfd_mem_attachment *attachment)
+{
+	switch (attachment->type) {
+	case KFD_MEM_ATT_SHARED:
+		break;
+	case KFD_MEM_ATT_USERPTR:
+		kfd_mem_dmaunmap_userptr(mem, attachment);
+		break;
+	case KFD_MEM_ATT_DMABUF:
+		kfd_mem_dmaunmap_dmabuf(attachment);
+		break;
+	case KFD_MEM_ATT_SG:
+		kfd_mem_dmaunmap_sg_bo(mem, attachment);
+		break;
+	default:
+		WARN_ON_ONCE(1);
+	}
+}
+
+static int kfd_mem_export_dmabuf(struct kgd_mem *mem)
+{
+	if (!mem->dmabuf) {
+		struct amdgpu_device *bo_adev;
+		struct dma_buf *dmabuf;
+
+		bo_adev = amdgpu_ttm_adev(mem->bo->tbo.bdev);
+		dmabuf = drm_gem_prime_handle_to_dmabuf(&bo_adev->ddev, bo_adev->kfd.client.file,
+					       mem->gem_handle,
+			mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_WRITABLE ?
+					       DRM_RDWR : 0);
+		if (IS_ERR(dmabuf))
+			return PTR_ERR(dmabuf);
+		mem->dmabuf = dmabuf;
+	}
+
+	return 0;
+}
+
+static int
+kfd_mem_attach_dmabuf(struct amdgpu_device *adev, struct kgd_mem *mem,
+		      struct amdgpu_bo **bo)
+{
+	struct drm_gem_object *gobj;
+	int ret;
+
+	ret = kfd_mem_export_dmabuf(mem);
+	if (ret)
+		return ret;
+
+	gobj = amdgpu_gem_prime_import(adev_to_drm(adev), mem->dmabuf);
+	if (IS_ERR(gobj))
+		return PTR_ERR(gobj);
+
+	*bo = gem_to_amdgpu_bo(gobj);
+	(*bo)->flags |= AMDGPU_GEM_CREATE_PREEMPTIBLE;
+
+	return 0;
+}
+
+/* kfd_mem_attach - Add a BO to a VM
+ *
+ * Everything that needs to bo done only once when a BO is first added
+ * to a VM. It can later be mapped and unmapped many times without
+ * repeating these steps.
+ *
+ * 0. Create BO for DMA mapping, if needed
+ * 1. Allocate and initialize BO VA entry data structure
+ * 2. Add BO to the VM
+ * 3. Determine ASIC-specific PTE flags
+ * 4. Alloc page tables and directories if needed
+ * 4a.  Validate new page tables and directories
+ */
+static int kfd_mem_attach(struct amdgpu_device *adev, struct kgd_mem *mem,
+		struct amdgpu_vm *vm, bool is_aql)
+{
+	struct amdgpu_device *bo_adev = amdgpu_ttm_adev(mem->bo->tbo.bdev);
+	unsigned long bo_size = mem->bo->tbo.base.size;
+	uint64_t va = mem->va;
+	struct kfd_mem_attachment *attachment[2] = {NULL, NULL};
+	struct amdgpu_bo *bo[2] = {NULL, NULL};
+	struct amdgpu_bo_va *bo_va;
+	bool same_hive = false;
+	int i, ret;
+
+	if (!va) {
+		pr_err("Invalid VA when adding BO to VM\n");
+		return -EINVAL;
+	}
+
+	/* Determine access to VRAM, MMIO and DOORBELL BOs of peer devices
+	 *
+	 * The access path of MMIO and DOORBELL BOs of is always over PCIe.
+	 * In contrast the access path of VRAM BOs depens upon the type of
+	 * link that connects the peer device. Access over PCIe is allowed
+	 * if peer device has large BAR. In contrast, access over xGMI is
+	 * allowed for both small and large BAR configurations of peer device
+	 */
+	if ((adev != bo_adev && !adev->apu_prefer_gtt) &&
+	    ((mem->domain == AMDGPU_GEM_DOMAIN_VRAM) ||
+	     (mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_DOORBELL) ||
+	     (mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_MMIO_REMAP))) {
+		if (mem->domain == AMDGPU_GEM_DOMAIN_VRAM)
+			same_hive = amdgpu_xgmi_same_hive(adev, bo_adev);
+		if (!same_hive && !amdgpu_device_is_peer_accessible(bo_adev, adev))
+			return -EINVAL;
+	}
+
+	for (i = 0; i <= is_aql; i++) {
+		attachment[i] = kzalloc(sizeof(*attachment[i]), GFP_KERNEL);
+		if (unlikely(!attachment[i])) {
+			ret = -ENOMEM;
+			goto unwind;
+		}
+
+		pr_debug("\t add VA 0x%llx - 0x%llx to vm %p\n", va,
+			 va + bo_size, vm);
+
+		if ((adev == bo_adev && !(mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_MMIO_REMAP)) ||
+		    (amdgpu_ttm_tt_get_usermm(mem->bo->tbo.ttm) && reuse_dmamap(adev, bo_adev)) ||
+		    (mem->domain == AMDGPU_GEM_DOMAIN_GTT && reuse_dmamap(adev, bo_adev)) ||
+		    same_hive) {
+			/* Mappings on the local GPU, or VRAM mappings in the
+			 * local hive, or userptr, or GTT mapping can reuse dma map
+			 * address space share the original BO
+			 */
+			attachment[i]->type = KFD_MEM_ATT_SHARED;
+			bo[i] = mem->bo;
+			drm_gem_object_get(&bo[i]->tbo.base);
+		} else if (i > 0) {
+			/* Multiple mappings on the same GPU share the BO */
+			attachment[i]->type = KFD_MEM_ATT_SHARED;
+			bo[i] = bo[0];
+			drm_gem_object_get(&bo[i]->tbo.base);
+		} else if (amdgpu_ttm_tt_get_usermm(mem->bo->tbo.ttm)) {
+			/* Create an SG BO to DMA-map userptrs on other GPUs */
+			attachment[i]->type = KFD_MEM_ATT_USERPTR;
+			ret = create_dmamap_sg_bo(adev, mem, &bo[i]);
+			if (ret)
+				goto unwind;
+		/* Handle DOORBELL BOs of peer devices and MMIO BOs of local and peer devices */
+		} else if (mem->bo->tbo.type == ttm_bo_type_sg) {
+			WARN_ONCE(!(mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_DOORBELL ||
+				    mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_MMIO_REMAP),
+				  "Handing invalid SG BO in ATTACH request");
+			attachment[i]->type = KFD_MEM_ATT_SG;
+			ret = create_dmamap_sg_bo(adev, mem, &bo[i]);
+			if (ret)
+				goto unwind;
+		/* Enable acces to GTT and VRAM BOs of peer devices */
+		} else if (mem->domain == AMDGPU_GEM_DOMAIN_GTT ||
+			   mem->domain == AMDGPU_GEM_DOMAIN_VRAM) {
+			attachment[i]->type = KFD_MEM_ATT_DMABUF;
+			ret = kfd_mem_attach_dmabuf(adev, mem, &bo[i]);
+			if (ret)
+				goto unwind;
+			pr_debug("Employ DMABUF mechanism to enable peer GPU access\n");
+		} else {
+			WARN_ONCE(true, "Handling invalid ATTACH request");
+			ret = -EINVAL;
+			goto unwind;
+		}
+
+		/* Add BO to VM internal data structures */
+		ret = amdgpu_bo_reserve(bo[i], false);
+		if (ret) {
+			pr_debug("Unable to reserve BO during memory attach");
+			goto unwind;
+		}
+		bo_va = amdgpu_vm_bo_find(vm, bo[i]);
+		if (!bo_va)
+			bo_va = amdgpu_vm_bo_add(adev, vm, bo[i]);
+		else
+			++bo_va->ref_count;
+		attachment[i]->bo_va = bo_va;
+		amdgpu_bo_unreserve(bo[i]);
+		if (unlikely(!attachment[i]->bo_va)) {
+			ret = -ENOMEM;
+			pr_err("Failed to add BO object to VM. ret == %d\n",
+			       ret);
+			goto unwind;
+		}
+		attachment[i]->va = va;
+		attachment[i]->pte_flags = get_pte_flags(adev, vm, mem);
+		attachment[i]->adev = adev;
+		list_add(&attachment[i]->list, &mem->attachments);
+
+		va += bo_size;
+	}
+
+	return 0;
+
+unwind:
+	for (; i >= 0; i--) {
+		if (!attachment[i])
+			continue;
+		if (attachment[i]->bo_va) {
+			(void)amdgpu_bo_reserve(bo[i], true);
+			if (--attachment[i]->bo_va->ref_count == 0)
+				amdgpu_vm_bo_del(adev, attachment[i]->bo_va);
+			amdgpu_bo_unreserve(bo[i]);
+			list_del(&attachment[i]->list);
+		}
+		if (bo[i])
+			drm_gem_object_put(&bo[i]->tbo.base);
+		kfree(attachment[i]);
+	}
+	return ret;
+}
+
+static void kfd_mem_detach(struct kfd_mem_attachment *attachment)
+{
+	struct amdgpu_bo *bo = attachment->bo_va->base.bo;
+
+	pr_debug("\t remove VA 0x%llx in entry %p\n",
+			attachment->va, attachment);
+	if (--attachment->bo_va->ref_count == 0)
+		amdgpu_vm_bo_del(attachment->adev, attachment->bo_va);
+	drm_gem_object_put(&bo->tbo.base);
+	list_del(&attachment->list);
+	kfree(attachment);
+}
+
+static void add_kgd_mem_to_kfd_bo_list(struct kgd_mem *mem,
+				struct amdkfd_process_info *process_info,
+				bool userptr)
+{
+	mutex_lock(&process_info->lock);
+	if (userptr)
+		list_add_tail(&mem->validate_list,
+			      &process_info->userptr_valid_list);
+	else
+		list_add_tail(&mem->validate_list, &process_info->kfd_bo_list);
+	mutex_unlock(&process_info->lock);
+}
+
+static void remove_kgd_mem_from_kfd_bo_list(struct kgd_mem *mem,
+		struct amdkfd_process_info *process_info)
+{
+	mutex_lock(&process_info->lock);
+	list_del(&mem->validate_list);
+	mutex_unlock(&process_info->lock);
+}
+
+/* Initializes user pages. It registers the MMU notifier and validates
+ * the userptr BO in the GTT domain.
+ *
+ * The BO must already be on the userptr_valid_list. Otherwise an
+ * eviction and restore may happen that leaves the new BO unmapped
+ * with the user mode queues running.
+ *
+ * Takes the process_info->lock to protect against concurrent restore
+ * workers.
+ *
+ * Returns 0 for success, negative errno for errors.
+ */
+static int init_user_pages(struct kgd_mem *mem, uint64_t user_addr,
+			   bool criu_resume)
+{
+	struct amdkfd_process_info *process_info = mem->process_info;
+	struct amdgpu_bo *bo = mem->bo;
+	struct ttm_operation_ctx ctx = { true, false };
+	struct hmm_range *range;
+	int ret = 0;
+
+	mutex_lock(&process_info->lock);
+
+	ret = amdgpu_ttm_tt_set_userptr(&bo->tbo, user_addr, 0);
+	if (ret) {
+		pr_err("%s: Failed to set userptr: %d\n", __func__, ret);
+		goto out;
+	}
+
+	ret = amdgpu_hmm_register(bo, user_addr);
+	if (ret) {
+		pr_err("%s: Failed to register MMU notifier: %d\n",
+		       __func__, ret);
+		goto out;
+	}
+
+	if (criu_resume) {
+		/*
+		 * During a CRIU restore operation, the userptr buffer objects
+		 * will be validated in the restore_userptr_work worker at a
+		 * later stage when it is scheduled by another ioctl called by
+		 * CRIU master process for the target pid for restore.
+		 */
+		mutex_lock(&process_info->notifier_lock);
+		mem->invalid++;
+		mutex_unlock(&process_info->notifier_lock);
+		mutex_unlock(&process_info->lock);
+		return 0;
+	}
+
+	ret = amdgpu_ttm_tt_get_user_pages(bo, &range);
+	if (ret) {
+		if (ret == -EAGAIN)
+			pr_debug("Failed to get user pages, try again\n");
+		else
+			pr_err("%s: Failed to get user pages: %d\n", __func__, ret);
+		goto unregister_out;
+	}
+
+	ret = amdgpu_bo_reserve(bo, true);
+	if (ret) {
+		pr_err("%s: Failed to reserve BO\n", __func__);
+		goto release_out;
+	}
+
+	amdgpu_ttm_tt_set_user_pages(bo->tbo.ttm, range);
+
+	amdgpu_bo_placement_from_domain(bo, mem->domain);
+	ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+	if (ret)
+		pr_err("%s: failed to validate BO\n", __func__);
+	amdgpu_bo_unreserve(bo);
+
+release_out:
+	amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm, range);
+unregister_out:
+	if (ret)
+		amdgpu_hmm_unregister(bo);
+out:
+	mutex_unlock(&process_info->lock);
+	return ret;
+}
+
+/* Reserving a BO and its page table BOs must happen atomically to
+ * avoid deadlocks. Some operations update multiple VMs at once. Track
+ * all the reservation info in a context structure. Optionally a sync
+ * object can track VM updates.
+ */
+struct bo_vm_reservation_context {
+	/* DRM execution context for the reservation */
+	struct drm_exec exec;
+	/* Number of VMs reserved */
+	unsigned int n_vms;
+	/* Pointer to sync object */
+	struct amdgpu_sync *sync;
+};
+
+enum bo_vm_match {
+	BO_VM_NOT_MAPPED = 0,	/* Match VMs where a BO is not mapped */
+	BO_VM_MAPPED,		/* Match VMs where a BO is mapped     */
+	BO_VM_ALL,		/* Match all VMs a BO was added to    */
+};
+
+/**
+ * reserve_bo_and_vm - reserve a BO and a VM unconditionally.
+ * @mem: KFD BO structure.
+ * @vm: the VM to reserve.
+ * @ctx: the struct that will be used in unreserve_bo_and_vms().
+ */
+static int reserve_bo_and_vm(struct kgd_mem *mem,
+			      struct amdgpu_vm *vm,
+			      struct bo_vm_reservation_context *ctx)
+{
+	struct amdgpu_bo *bo = mem->bo;
+	int ret;
+
+	WARN_ON(!vm);
+
+	ctx->n_vms = 1;
+	ctx->sync = &mem->sync;
+	drm_exec_init(&ctx->exec, DRM_EXEC_INTERRUPTIBLE_WAIT, 0);
+	drm_exec_until_all_locked(&ctx->exec) {
+		ret = amdgpu_vm_lock_pd(vm, &ctx->exec, 2);
+		drm_exec_retry_on_contention(&ctx->exec);
+		if (unlikely(ret))
+			goto error;
+
+		ret = drm_exec_prepare_obj(&ctx->exec, &bo->tbo.base, 1);
+		drm_exec_retry_on_contention(&ctx->exec);
+		if (unlikely(ret))
+			goto error;
+	}
+	return 0;
+
+error:
+	pr_err("Failed to reserve buffers in ttm.\n");
+	drm_exec_fini(&ctx->exec);
+	return ret;
+}
+
+/**
+ * reserve_bo_and_cond_vms - reserve a BO and some VMs conditionally
+ * @mem: KFD BO structure.
+ * @vm: the VM to reserve. If NULL, then all VMs associated with the BO
+ * is used. Otherwise, a single VM associated with the BO.
+ * @map_type: the mapping status that will be used to filter the VMs.
+ * @ctx: the struct that will be used in unreserve_bo_and_vms().
+ *
+ * Returns 0 for success, negative for failure.
+ */
+static int reserve_bo_and_cond_vms(struct kgd_mem *mem,
+				struct amdgpu_vm *vm, enum bo_vm_match map_type,
+				struct bo_vm_reservation_context *ctx)
+{
+	struct kfd_mem_attachment *entry;
+	struct amdgpu_bo *bo = mem->bo;
+	int ret;
+
+	ctx->sync = &mem->sync;
+	drm_exec_init(&ctx->exec, DRM_EXEC_INTERRUPTIBLE_WAIT |
+		      DRM_EXEC_IGNORE_DUPLICATES, 0);
+	drm_exec_until_all_locked(&ctx->exec) {
+		ctx->n_vms = 0;
+		list_for_each_entry(entry, &mem->attachments, list) {
+			if ((vm && vm != entry->bo_va->base.vm) ||
+				(entry->is_mapped != map_type
+				&& map_type != BO_VM_ALL))
+				continue;
+
+			ret = amdgpu_vm_lock_pd(entry->bo_va->base.vm,
+						&ctx->exec, 2);
+			drm_exec_retry_on_contention(&ctx->exec);
+			if (unlikely(ret))
+				goto error;
+			++ctx->n_vms;
+		}
+
+		ret = drm_exec_prepare_obj(&ctx->exec, &bo->tbo.base, 1);
+		drm_exec_retry_on_contention(&ctx->exec);
+		if (unlikely(ret))
+			goto error;
+	}
+	return 0;
+
+error:
+	pr_err("Failed to reserve buffers in ttm.\n");
+	drm_exec_fini(&ctx->exec);
+	return ret;
+}
+
+/**
+ * unreserve_bo_and_vms - Unreserve BO and VMs from a reservation context
+ * @ctx: Reservation context to unreserve
+ * @wait: Optionally wait for a sync object representing pending VM updates
+ * @intr: Whether the wait is interruptible
+ *
+ * Also frees any resources allocated in
+ * reserve_bo_and_(cond_)vm(s). Returns the status from
+ * amdgpu_sync_wait.
+ */
+static int unreserve_bo_and_vms(struct bo_vm_reservation_context *ctx,
+				 bool wait, bool intr)
+{
+	int ret = 0;
+
+	if (wait)
+		ret = amdgpu_sync_wait(ctx->sync, intr);
+
+	drm_exec_fini(&ctx->exec);
+	ctx->sync = NULL;
+	return ret;
+}
+
+static int unmap_bo_from_gpuvm(struct kgd_mem *mem,
+				struct kfd_mem_attachment *entry,
+				struct amdgpu_sync *sync)
+{
+	struct amdgpu_bo_va *bo_va = entry->bo_va;
+	struct amdgpu_device *adev = entry->adev;
+	struct amdgpu_vm *vm = bo_va->base.vm;
+
+	if (bo_va->queue_refcount) {
+		pr_debug("bo_va->queue_refcount %d\n", bo_va->queue_refcount);
+		return -EBUSY;
+	}
+
+	(void)amdgpu_vm_bo_unmap(adev, bo_va, entry->va);
+
+	(void)amdgpu_vm_clear_freed(adev, vm, &bo_va->last_pt_update);
+
+	(void)amdgpu_sync_fence(sync, bo_va->last_pt_update, GFP_KERNEL);
+
+	return 0;
+}
+
+static int update_gpuvm_pte(struct kgd_mem *mem,
+			    struct kfd_mem_attachment *entry,
+			    struct amdgpu_sync *sync)
+{
+	struct amdgpu_bo_va *bo_va = entry->bo_va;
+	struct amdgpu_device *adev = entry->adev;
+	int ret;
+
+	ret = kfd_mem_dmamap_attachment(mem, entry);
+	if (ret)
+		return ret;
+
+	/* Update the page tables  */
+	ret = amdgpu_vm_bo_update(adev, bo_va, false);
+	if (ret) {
+		pr_err("amdgpu_vm_bo_update failed\n");
+		return ret;
+	}
+
+	return amdgpu_sync_fence(sync, bo_va->last_pt_update, GFP_KERNEL);
+}
+
+static int map_bo_to_gpuvm(struct kgd_mem *mem,
+			   struct kfd_mem_attachment *entry,
+			   struct amdgpu_sync *sync,
+			   bool no_update_pte)
+{
+	int ret;
+
+	/* Set virtual address for the allocation */
+	ret = amdgpu_vm_bo_map(entry->adev, entry->bo_va, entry->va, 0,
+			       amdgpu_bo_size(entry->bo_va->base.bo),
+			       entry->pte_flags);
+	if (ret) {
+		pr_err("Failed to map VA 0x%llx in vm. ret %d\n",
+				entry->va, ret);
+		return ret;
+	}
+
+	if (no_update_pte)
+		return 0;
+
+	ret = update_gpuvm_pte(mem, entry, sync);
+	if (ret) {
+		pr_err("update_gpuvm_pte() failed\n");
+		goto update_gpuvm_pte_failed;
+	}
+
+	return 0;
+
+update_gpuvm_pte_failed:
+	unmap_bo_from_gpuvm(mem, entry, sync);
+	kfd_mem_dmaunmap_attachment(mem, entry);
+	return ret;
+}
+
+static int process_validate_vms(struct amdkfd_process_info *process_info,
+				struct ww_acquire_ctx *ticket)
+{
+	struct amdgpu_vm *peer_vm;
+	int ret;
+
+	list_for_each_entry(peer_vm, &process_info->vm_list_head,
+			    vm_list_node) {
+		ret = vm_validate_pt_pd_bos(peer_vm, ticket);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int process_sync_pds_resv(struct amdkfd_process_info *process_info,
+				 struct amdgpu_sync *sync)
+{
+	struct amdgpu_vm *peer_vm;
+	int ret;
+
+	list_for_each_entry(peer_vm, &process_info->vm_list_head,
+			    vm_list_node) {
+		struct amdgpu_bo *pd = peer_vm->root.bo;
+
+		ret = amdgpu_sync_resv(NULL, sync, pd->tbo.base.resv,
+				       AMDGPU_SYNC_NE_OWNER,
+				       AMDGPU_FENCE_OWNER_KFD);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int process_update_pds(struct amdkfd_process_info *process_info,
+			      struct amdgpu_sync *sync)
+{
+	struct amdgpu_vm *peer_vm;
+	int ret;
+
+	list_for_each_entry(peer_vm, &process_info->vm_list_head,
+			    vm_list_node) {
+		ret = vm_update_pds(peer_vm, sync);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int init_kfd_vm(struct amdgpu_vm *vm, void **process_info,
+		       struct dma_fence **ef)
+{
+	struct amdkfd_process_info *info = NULL;
+	int ret;
+
+	if (!*process_info) {
+		info = kzalloc(sizeof(*info), GFP_KERNEL);
+		if (!info)
+			return -ENOMEM;
+
+		mutex_init(&info->lock);
+		mutex_init(&info->notifier_lock);
+		INIT_LIST_HEAD(&info->vm_list_head);
+		INIT_LIST_HEAD(&info->kfd_bo_list);
+		INIT_LIST_HEAD(&info->userptr_valid_list);
+		INIT_LIST_HEAD(&info->userptr_inval_list);
+
+		info->eviction_fence =
+			amdgpu_amdkfd_fence_create(dma_fence_context_alloc(1),
+						   current->mm,
+						   NULL);
+		if (!info->eviction_fence) {
+			pr_err("Failed to create eviction fence\n");
+			ret = -ENOMEM;
+			goto create_evict_fence_fail;
+		}
+
+		info->pid = get_task_pid(current->group_leader, PIDTYPE_PID);
+		INIT_DELAYED_WORK(&info->restore_userptr_work,
+				  amdgpu_amdkfd_restore_userptr_worker);
+
+		*process_info = info;
+	}
+
+	vm->process_info = *process_info;
+
+	/* Validate page directory and attach eviction fence */
+	ret = amdgpu_bo_reserve(vm->root.bo, true);
+	if (ret)
+		goto reserve_pd_fail;
+	ret = vm_validate_pt_pd_bos(vm, NULL);
+	if (ret) {
+		pr_err("validate_pt_pd_bos() failed\n");
+		goto validate_pd_fail;
+	}
+	ret = amdgpu_bo_sync_wait(vm->root.bo,
+				  AMDGPU_FENCE_OWNER_KFD, false);
+	if (ret)
+		goto wait_pd_fail;
+	ret = dma_resv_reserve_fences(vm->root.bo->tbo.base.resv, 1);
+	if (ret)
+		goto reserve_shared_fail;
+	dma_resv_add_fence(vm->root.bo->tbo.base.resv,
+			   &vm->process_info->eviction_fence->base,
+			   DMA_RESV_USAGE_BOOKKEEP);
+	amdgpu_bo_unreserve(vm->root.bo);
+
+	/* Update process info */
+	mutex_lock(&vm->process_info->lock);
+	list_add_tail(&vm->vm_list_node,
+			&(vm->process_info->vm_list_head));
+	vm->process_info->n_vms++;
+	if (ef)
+		*ef = dma_fence_get(&vm->process_info->eviction_fence->base);
+	mutex_unlock(&vm->process_info->lock);
+
+	return 0;
+
+reserve_shared_fail:
+wait_pd_fail:
+validate_pd_fail:
+	amdgpu_bo_unreserve(vm->root.bo);
+reserve_pd_fail:
+	vm->process_info = NULL;
+	if (info) {
+		dma_fence_put(&info->eviction_fence->base);
+		*process_info = NULL;
+		put_pid(info->pid);
+create_evict_fence_fail:
+		mutex_destroy(&info->lock);
+		mutex_destroy(&info->notifier_lock);
+		kfree(info);
+	}
+	return ret;
+}
+
+/**
+ * amdgpu_amdkfd_gpuvm_pin_bo() - Pins a BO using following criteria
+ * @bo: Handle of buffer object being pinned
+ * @domain: Domain into which BO should be pinned
+ *
+ *   - USERPTR BOs are UNPINNABLE and will return error
+ *   - All other BO types (GTT, VRAM, MMIO and DOORBELL) will have their
+ *     PIN count incremented. It is valid to PIN a BO multiple times
+ *
+ * Return: ZERO if successful in pinning, Non-Zero in case of error.
+ */
+static int amdgpu_amdkfd_gpuvm_pin_bo(struct amdgpu_bo *bo, u32 domain)
+{
+	int ret = 0;
+
+	ret = amdgpu_bo_reserve(bo, false);
+	if (unlikely(ret))
+		return ret;
+
+	if (bo->flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS) {
+		/*
+		 * If bo is not contiguous on VRAM, move to system memory first to ensure
+		 * we can get contiguous VRAM space after evicting other BOs.
+		 */
+		if (!(bo->tbo.resource->placement & TTM_PL_FLAG_CONTIGUOUS)) {
+			struct ttm_operation_ctx ctx = { true, false };
+
+			amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_GTT);
+			ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+			if (unlikely(ret)) {
+				pr_debug("validate bo 0x%p to GTT failed %d\n", &bo->tbo, ret);
+				goto out;
+			}
+		}
+	}
+
+	ret = amdgpu_bo_pin(bo, domain);
+	if (ret)
+		pr_err("Error in Pinning BO to domain: %d\n", domain);
+
+	amdgpu_bo_sync_wait(bo, AMDGPU_FENCE_OWNER_KFD, false);
+out:
+	amdgpu_bo_unreserve(bo);
+	return ret;
+}
+
+/**
+ * amdgpu_amdkfd_gpuvm_unpin_bo() - Unpins BO using following criteria
+ * @bo: Handle of buffer object being unpinned
+ *
+ *   - Is a illegal request for USERPTR BOs and is ignored
+ *   - All other BO types (GTT, VRAM, MMIO and DOORBELL) will have their
+ *     PIN count decremented. Calls to UNPIN must balance calls to PIN
+ */
+static void amdgpu_amdkfd_gpuvm_unpin_bo(struct amdgpu_bo *bo)
+{
+	int ret = 0;
+
+	ret = amdgpu_bo_reserve(bo, false);
+	if (unlikely(ret))
+		return;
+
+	amdgpu_bo_unpin(bo);
+	amdgpu_bo_unreserve(bo);
+}
+
+int amdgpu_amdkfd_gpuvm_acquire_process_vm(struct amdgpu_device *adev,
+					   struct amdgpu_vm *avm,
+					   void **process_info,
+					   struct dma_fence **ef)
+{
+	int ret;
+
+	/* Already a compute VM? */
+	if (avm->process_info)
+		return -EINVAL;
+
+	/* Convert VM into a compute VM */
+	ret = amdgpu_vm_make_compute(adev, avm);
+	if (ret)
+		return ret;
+
+	/* Initialize KFD part of the VM and process info */
+	ret = init_kfd_vm(avm, process_info, ef);
+	if (ret)
+		return ret;
+
+	amdgpu_vm_set_task_info(avm);
+
+	return 0;
+}
+
+void amdgpu_amdkfd_gpuvm_destroy_cb(struct amdgpu_device *adev,
+				    struct amdgpu_vm *vm)
+{
+	struct amdkfd_process_info *process_info = vm->process_info;
+
+	if (!process_info)
+		return;
+
+	/* Update process info */
+	mutex_lock(&process_info->lock);
+	process_info->n_vms--;
+	list_del(&vm->vm_list_node);
+	mutex_unlock(&process_info->lock);
+
+	vm->process_info = NULL;
+
+	/* Release per-process resources when last compute VM is destroyed */
+	if (!process_info->n_vms) {
+		WARN_ON(!list_empty(&process_info->kfd_bo_list));
+		WARN_ON(!list_empty(&process_info->userptr_valid_list));
+		WARN_ON(!list_empty(&process_info->userptr_inval_list));
+
+		dma_fence_put(&process_info->eviction_fence->base);
+		cancel_delayed_work_sync(&process_info->restore_userptr_work);
+		put_pid(process_info->pid);
+		mutex_destroy(&process_info->lock);
+		mutex_destroy(&process_info->notifier_lock);
+		kfree(process_info);
+	}
+}
+
+uint64_t amdgpu_amdkfd_gpuvm_get_process_page_dir(void *drm_priv)
+{
+	struct amdgpu_vm *avm = drm_priv_to_vm(drm_priv);
+	struct amdgpu_bo *pd = avm->root.bo;
+	struct amdgpu_device *adev = amdgpu_ttm_adev(pd->tbo.bdev);
+
+	if (adev->asic_type < CHIP_VEGA10)
+		return avm->pd_phys_addr >> AMDGPU_GPU_PAGE_SHIFT;
+	return avm->pd_phys_addr;
+}
+
+void amdgpu_amdkfd_block_mmu_notifications(void *p)
+{
+	struct amdkfd_process_info *pinfo = (struct amdkfd_process_info *)p;
+
+	mutex_lock(&pinfo->lock);
+	WRITE_ONCE(pinfo->block_mmu_notifications, true);
+	mutex_unlock(&pinfo->lock);
+}
+
+int amdgpu_amdkfd_criu_resume(void *p)
+{
+	int ret = 0;
+	struct amdkfd_process_info *pinfo = (struct amdkfd_process_info *)p;
+
+	mutex_lock(&pinfo->lock);
+	pr_debug("scheduling work\n");
+	mutex_lock(&pinfo->notifier_lock);
+	pinfo->evicted_bos++;
+	mutex_unlock(&pinfo->notifier_lock);
+	if (!READ_ONCE(pinfo->block_mmu_notifications)) {
+		ret = -EINVAL;
+		goto out_unlock;
+	}
+	WRITE_ONCE(pinfo->block_mmu_notifications, false);
+	queue_delayed_work(system_freezable_wq,
+			   &pinfo->restore_userptr_work, 0);
+
+out_unlock:
+	mutex_unlock(&pinfo->lock);
+	return ret;
+}
+
+size_t amdgpu_amdkfd_get_available_memory(struct amdgpu_device *adev,
+					  uint8_t xcp_id)
+{
+	uint64_t reserved_for_pt =
+		ESTIMATE_PT_SIZE(amdgpu_amdkfd_total_mem_size);
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	uint64_t reserved_for_ras = (con ? con->reserved_pages_in_bytes : 0);
+	ssize_t available;
+	uint64_t vram_available, system_mem_available, ttm_mem_available;
+
+	spin_lock(&kfd_mem_limit.mem_limit_lock);
+	if (adev->apu_prefer_gtt && !adev->gmc.is_app_apu)
+		vram_available = KFD_XCP_MEMORY_SIZE(adev, xcp_id)
+			- adev->kfd.vram_used_aligned[xcp_id];
+	else
+		vram_available = KFD_XCP_MEMORY_SIZE(adev, xcp_id)
+			- adev->kfd.vram_used_aligned[xcp_id]
+			- atomic64_read(&adev->vram_pin_size)
+			- reserved_for_pt
+			- reserved_for_ras;
+
+	if (adev->apu_prefer_gtt) {
+		system_mem_available = no_system_mem_limit ?
+					kfd_mem_limit.max_system_mem_limit :
+					kfd_mem_limit.max_system_mem_limit -
+					kfd_mem_limit.system_mem_used;
+
+		ttm_mem_available = kfd_mem_limit.max_ttm_mem_limit -
+				kfd_mem_limit.ttm_mem_used;
+
+		available = min3(system_mem_available, ttm_mem_available,
+				 vram_available);
+		available = ALIGN_DOWN(available, PAGE_SIZE);
+	} else {
+		available = ALIGN_DOWN(vram_available, VRAM_AVAILABLITY_ALIGN);
+	}
+
+	spin_unlock(&kfd_mem_limit.mem_limit_lock);
+
+	if (available < 0)
+		available = 0;
+
+	return available;
+}
+
+int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
+		struct amdgpu_device *adev, uint64_t va, uint64_t size,
+		void *drm_priv, struct kgd_mem **mem,
+		uint64_t *offset, uint32_t flags, bool criu_resume)
+{
+	struct amdgpu_vm *avm = drm_priv_to_vm(drm_priv);
+	struct amdgpu_fpriv *fpriv = container_of(avm, struct amdgpu_fpriv, vm);
+	enum ttm_bo_type bo_type = ttm_bo_type_device;
+	struct sg_table *sg = NULL;
+	uint64_t user_addr = 0;
+	struct amdgpu_bo *bo;
+	struct drm_gem_object *gobj = NULL;
+	u32 domain, alloc_domain;
+	uint64_t aligned_size;
+	int8_t xcp_id = -1;
+	u64 alloc_flags;
+	int ret;
+
+	/*
+	 * Check on which domain to allocate BO
+	 */
+	if (flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM) {
+		domain = alloc_domain = AMDGPU_GEM_DOMAIN_VRAM;
+
+		if (adev->apu_prefer_gtt) {
+			domain = AMDGPU_GEM_DOMAIN_GTT;
+			alloc_domain = AMDGPU_GEM_DOMAIN_GTT;
+			alloc_flags = 0;
+		} else {
+			alloc_flags = AMDGPU_GEM_CREATE_VRAM_WIPE_ON_RELEASE;
+			alloc_flags |= (flags & KFD_IOC_ALLOC_MEM_FLAGS_PUBLIC) ?
+			AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED : 0;
+
+			/* For contiguous VRAM allocation */
+			if (flags & KFD_IOC_ALLOC_MEM_FLAGS_CONTIGUOUS)
+				alloc_flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+		}
+		xcp_id = fpriv->xcp_id == AMDGPU_XCP_NO_PARTITION ?
+					0 : fpriv->xcp_id;
+	} else if (flags & KFD_IOC_ALLOC_MEM_FLAGS_GTT) {
+		domain = alloc_domain = AMDGPU_GEM_DOMAIN_GTT;
+		alloc_flags = 0;
+	} else {
+		domain = AMDGPU_GEM_DOMAIN_GTT;
+		alloc_domain = AMDGPU_GEM_DOMAIN_CPU;
+		alloc_flags = AMDGPU_GEM_CREATE_PREEMPTIBLE;
+
+		if (flags & KFD_IOC_ALLOC_MEM_FLAGS_USERPTR) {
+			if (!offset || !*offset)
+				return -EINVAL;
+			user_addr = untagged_addr(*offset);
+		} else if (flags & (KFD_IOC_ALLOC_MEM_FLAGS_DOORBELL |
+				    KFD_IOC_ALLOC_MEM_FLAGS_MMIO_REMAP)) {
+			bo_type = ttm_bo_type_sg;
+			if (size > UINT_MAX)
+				return -EINVAL;
+			sg = create_sg_table(*offset, size);
+			if (!sg)
+				return -ENOMEM;
+		} else {
+			return -EINVAL;
+		}
+	}
+
+	if (flags & KFD_IOC_ALLOC_MEM_FLAGS_COHERENT)
+		alloc_flags |= AMDGPU_GEM_CREATE_COHERENT;
+	if (flags & KFD_IOC_ALLOC_MEM_FLAGS_EXT_COHERENT)
+		alloc_flags |= AMDGPU_GEM_CREATE_EXT_COHERENT;
+	if (flags & KFD_IOC_ALLOC_MEM_FLAGS_UNCACHED)
+		alloc_flags |= AMDGPU_GEM_CREATE_UNCACHED;
+
+	*mem = kzalloc(sizeof(struct kgd_mem), GFP_KERNEL);
+	if (!*mem) {
+		ret = -ENOMEM;
+		goto err;
+	}
+	INIT_LIST_HEAD(&(*mem)->attachments);
+	mutex_init(&(*mem)->lock);
+	(*mem)->aql_queue = !!(flags & KFD_IOC_ALLOC_MEM_FLAGS_AQL_QUEUE_MEM);
+
+	/* Workaround for AQL queue wraparound bug. Map the same
+	 * memory twice. That means we only actually allocate half
+	 * the memory.
+	 */
+	if ((*mem)->aql_queue)
+		size >>= 1;
+	aligned_size = PAGE_ALIGN(size);
+
+	(*mem)->alloc_flags = flags;
+
+	amdgpu_sync_create(&(*mem)->sync);
+
+	ret = amdgpu_amdkfd_reserve_mem_limit(adev, aligned_size, flags,
+					      xcp_id);
+	if (ret) {
+		pr_debug("Insufficient memory\n");
+		goto err_reserve_limit;
+	}
+
+	pr_debug("\tcreate BO VA 0x%llx size 0x%llx domain %s xcp_id %d\n",
+		 va, (*mem)->aql_queue ? size << 1 : size,
+		 domain_string(alloc_domain), xcp_id);
+
+	ret = amdgpu_gem_object_create(adev, aligned_size, 1, alloc_domain, alloc_flags,
+				       bo_type, NULL, &gobj, xcp_id + 1);
+	if (ret) {
+		pr_debug("Failed to create BO on domain %s. ret %d\n",
+			 domain_string(alloc_domain), ret);
+		goto err_bo_create;
+	}
+	ret = drm_vma_node_allow(&gobj->vma_node, drm_priv);
+	if (ret) {
+		pr_debug("Failed to allow vma node access. ret %d\n", ret);
+		goto err_node_allow;
+	}
+	ret = drm_gem_handle_create(adev->kfd.client.file, gobj, &(*mem)->gem_handle);
+	if (ret)
+		goto err_gem_handle_create;
+	bo = gem_to_amdgpu_bo(gobj);
+	if (bo_type == ttm_bo_type_sg) {
+		bo->tbo.sg = sg;
+		bo->tbo.ttm->sg = sg;
+	}
+	bo->kfd_bo = *mem;
+	(*mem)->bo = bo;
+	if (user_addr)
+		bo->flags |= AMDGPU_AMDKFD_CREATE_USERPTR_BO;
+
+	(*mem)->va = va;
+	(*mem)->domain = domain;
+	(*mem)->mapped_to_gpu_memory = 0;
+	(*mem)->process_info = avm->process_info;
+
+	add_kgd_mem_to_kfd_bo_list(*mem, avm->process_info, user_addr);
+
+	if (user_addr) {
+		pr_debug("creating userptr BO for user_addr = %llx\n", user_addr);
+		ret = init_user_pages(*mem, user_addr, criu_resume);
+		if (ret)
+			goto allocate_init_user_pages_failed;
+	} else  if (flags & (KFD_IOC_ALLOC_MEM_FLAGS_DOORBELL |
+				KFD_IOC_ALLOC_MEM_FLAGS_MMIO_REMAP)) {
+		ret = amdgpu_amdkfd_gpuvm_pin_bo(bo, AMDGPU_GEM_DOMAIN_GTT);
+		if (ret) {
+			pr_err("Pinning MMIO/DOORBELL BO during ALLOC FAILED\n");
+			goto err_pin_bo;
+		}
+		bo->allowed_domains = AMDGPU_GEM_DOMAIN_GTT;
+		bo->preferred_domains = AMDGPU_GEM_DOMAIN_GTT;
+	} else {
+		mutex_lock(&avm->process_info->lock);
+		if (avm->process_info->eviction_fence &&
+		    !dma_fence_is_signaled(&avm->process_info->eviction_fence->base))
+			ret = amdgpu_amdkfd_bo_validate_and_fence(bo, domain,
+				&avm->process_info->eviction_fence->base);
+		mutex_unlock(&avm->process_info->lock);
+		if (ret)
+			goto err_validate_bo;
+	}
+
+	if (offset)
+		*offset = amdgpu_bo_mmap_offset(bo);
+
+	return 0;
+
+allocate_init_user_pages_failed:
+err_pin_bo:
+err_validate_bo:
+	remove_kgd_mem_from_kfd_bo_list(*mem, avm->process_info);
+	drm_gem_handle_delete(adev->kfd.client.file, (*mem)->gem_handle);
+err_gem_handle_create:
+	drm_vma_node_revoke(&gobj->vma_node, drm_priv);
+err_node_allow:
+	/* Don't unreserve system mem limit twice */
+	goto err_reserve_limit;
+err_bo_create:
+	amdgpu_amdkfd_unreserve_mem_limit(adev, aligned_size, flags, xcp_id);
+err_reserve_limit:
+	amdgpu_sync_free(&(*mem)->sync);
+	mutex_destroy(&(*mem)->lock);
+	if (gobj)
+		drm_gem_object_put(gobj);
+	else
+		kfree(*mem);
+err:
+	if (sg) {
+		sg_free_table(sg);
+		kfree(sg);
+	}
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_free_memory_of_gpu(
+		struct amdgpu_device *adev, struct kgd_mem *mem, void *drm_priv,
+		uint64_t *size)
+{
+	struct amdkfd_process_info *process_info = mem->process_info;
+	unsigned long bo_size = mem->bo->tbo.base.size;
+	bool use_release_notifier = (mem->bo->kfd_bo == mem);
+	struct kfd_mem_attachment *entry, *tmp;
+	struct bo_vm_reservation_context ctx;
+	unsigned int mapped_to_gpu_memory;
+	int ret;
+	bool is_imported = false;
+
+	mutex_lock(&mem->lock);
+
+	/* Unpin MMIO/DOORBELL BO's that were pinned during allocation */
+	if (mem->alloc_flags &
+	    (KFD_IOC_ALLOC_MEM_FLAGS_DOORBELL |
+	     KFD_IOC_ALLOC_MEM_FLAGS_MMIO_REMAP)) {
+		amdgpu_amdkfd_gpuvm_unpin_bo(mem->bo);
+	}
+
+	mapped_to_gpu_memory = mem->mapped_to_gpu_memory;
+	is_imported = mem->is_imported;
+	mutex_unlock(&mem->lock);
+	/* lock is not needed after this, since mem is unused and will
+	 * be freed anyway
+	 */
+
+	if (mapped_to_gpu_memory > 0) {
+		pr_debug("BO VA 0x%llx size 0x%lx is still mapped.\n",
+				mem->va, bo_size);
+		return -EBUSY;
+	}
+
+	/* Make sure restore workers don't access the BO any more */
+	mutex_lock(&process_info->lock);
+	list_del(&mem->validate_list);
+	mutex_unlock(&process_info->lock);
+
+	/* Cleanup user pages and MMU notifiers */
+	if (amdgpu_ttm_tt_get_usermm(mem->bo->tbo.ttm)) {
+		amdgpu_hmm_unregister(mem->bo);
+		mutex_lock(&process_info->notifier_lock);
+		amdgpu_ttm_tt_discard_user_pages(mem->bo->tbo.ttm, mem->range);
+		mutex_unlock(&process_info->notifier_lock);
+	}
+
+	ret = reserve_bo_and_cond_vms(mem, NULL, BO_VM_ALL, &ctx);
+	if (unlikely(ret))
+		return ret;
+
+	amdgpu_amdkfd_remove_eviction_fence(mem->bo,
+					process_info->eviction_fence);
+	pr_debug("Release VA 0x%llx - 0x%llx\n", mem->va,
+		mem->va + bo_size * (1 + mem->aql_queue));
+
+	/* Remove from VM internal data structures */
+	list_for_each_entry_safe(entry, tmp, &mem->attachments, list) {
+		kfd_mem_dmaunmap_attachment(mem, entry);
+		kfd_mem_detach(entry);
+	}
+
+	ret = unreserve_bo_and_vms(&ctx, false, false);
+
+	/* Free the sync object */
+	amdgpu_sync_free(&mem->sync);
+
+	/* If the SG is not NULL, it's one we created for a doorbell or mmio
+	 * remap BO. We need to free it.
+	 */
+	if (mem->bo->tbo.sg) {
+		sg_free_table(mem->bo->tbo.sg);
+		kfree(mem->bo->tbo.sg);
+	}
+
+	/* Update the size of the BO being freed if it was allocated from
+	 * VRAM and is not imported. For APP APU VRAM allocations are done
+	 * in GTT domain
+	 */
+	if (size) {
+		if (!is_imported &&
+		   (mem->bo->preferred_domains == AMDGPU_GEM_DOMAIN_VRAM ||
+		   (adev->apu_prefer_gtt &&
+		    mem->bo->preferred_domains == AMDGPU_GEM_DOMAIN_GTT)))
+			*size = bo_size;
+		else
+			*size = 0;
+	}
+
+	/* Free the BO*/
+	drm_vma_node_revoke(&mem->bo->tbo.base.vma_node, drm_priv);
+	drm_gem_handle_delete(adev->kfd.client.file, mem->gem_handle);
+	if (mem->dmabuf) {
+		dma_buf_put(mem->dmabuf);
+		mem->dmabuf = NULL;
+	}
+	mutex_destroy(&mem->lock);
+
+	/* If this releases the last reference, it will end up calling
+	 * amdgpu_amdkfd_release_notify and kfree the mem struct. That's why
+	 * this needs to be the last call here.
+	 */
+	drm_gem_object_put(&mem->bo->tbo.base);
+
+	/*
+	 * For kgd_mem allocated in amdgpu_amdkfd_gpuvm_import_dmabuf(),
+	 * explicitly free it here.
+	 */
+	if (!use_release_notifier)
+		kfree(mem);
+
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_map_memory_to_gpu(
+		struct amdgpu_device *adev, struct kgd_mem *mem,
+		void *drm_priv)
+{
+	struct amdgpu_vm *avm = drm_priv_to_vm(drm_priv);
+	int ret;
+	struct amdgpu_bo *bo;
+	uint32_t domain;
+	struct kfd_mem_attachment *entry;
+	struct bo_vm_reservation_context ctx;
+	unsigned long bo_size;
+	bool is_invalid_userptr = false;
+
+	bo = mem->bo;
+	if (!bo) {
+		pr_err("Invalid BO when mapping memory to GPU\n");
+		return -EINVAL;
+	}
+
+	/* Make sure restore is not running concurrently. Since we
+	 * don't map invalid userptr BOs, we rely on the next restore
+	 * worker to do the mapping
+	 */
+	mutex_lock(&mem->process_info->lock);
+
+	/* Lock notifier lock. If we find an invalid userptr BO, we can be
+	 * sure that the MMU notifier is no longer running
+	 * concurrently and the queues are actually stopped
+	 */
+	if (amdgpu_ttm_tt_get_usermm(bo->tbo.ttm)) {
+		mutex_lock(&mem->process_info->notifier_lock);
+		is_invalid_userptr = !!mem->invalid;
+		mutex_unlock(&mem->process_info->notifier_lock);
+	}
+
+	mutex_lock(&mem->lock);
+
+	domain = mem->domain;
+	bo_size = bo->tbo.base.size;
+
+	pr_debug("Map VA 0x%llx - 0x%llx to vm %p domain %s\n",
+			mem->va,
+			mem->va + bo_size * (1 + mem->aql_queue),
+			avm, domain_string(domain));
+
+	if (!kfd_mem_is_attached(avm, mem)) {
+		ret = kfd_mem_attach(adev, mem, avm, mem->aql_queue);
+		if (ret)
+			goto out;
+	}
+
+	ret = reserve_bo_and_vm(mem, avm, &ctx);
+	if (unlikely(ret))
+		goto out;
+
+	/* Userptr can be marked as "not invalid", but not actually be
+	 * validated yet (still in the system domain). In that case
+	 * the queues are still stopped and we can leave mapping for
+	 * the next restore worker
+	 */
+	if (amdgpu_ttm_tt_get_usermm(bo->tbo.ttm) &&
+	    bo->tbo.resource->mem_type == TTM_PL_SYSTEM)
+		is_invalid_userptr = true;
+
+	ret = vm_validate_pt_pd_bos(avm, NULL);
+	if (unlikely(ret))
+		goto out_unreserve;
+
+	list_for_each_entry(entry, &mem->attachments, list) {
+		if (entry->bo_va->base.vm != avm || entry->is_mapped)
+			continue;
+
+		pr_debug("\t map VA 0x%llx - 0x%llx in entry %p\n",
+			 entry->va, entry->va + bo_size, entry);
+
+		ret = map_bo_to_gpuvm(mem, entry, ctx.sync,
+				      is_invalid_userptr);
+		if (ret) {
+			pr_err("Failed to map bo to gpuvm\n");
+			goto out_unreserve;
+		}
+
+		ret = vm_update_pds(avm, ctx.sync);
+		if (ret) {
+			pr_err("Failed to update page directories\n");
+			goto out_unreserve;
+		}
+
+		entry->is_mapped = true;
+		mem->mapped_to_gpu_memory++;
+		pr_debug("\t INC mapping count %d\n",
+			 mem->mapped_to_gpu_memory);
+	}
+
+	ret = unreserve_bo_and_vms(&ctx, false, false);
+
+	goto out;
+
+out_unreserve:
+	unreserve_bo_and_vms(&ctx, false, false);
+out:
+	mutex_unlock(&mem->process_info->lock);
+	mutex_unlock(&mem->lock);
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_dmaunmap_mem(struct kgd_mem *mem, void *drm_priv)
+{
+	struct kfd_mem_attachment *entry;
+	struct amdgpu_vm *vm;
+	int ret;
+
+	vm = drm_priv_to_vm(drm_priv);
+
+	mutex_lock(&mem->lock);
+
+	ret = amdgpu_bo_reserve(mem->bo, true);
+	if (ret)
+		goto out;
+
+	list_for_each_entry(entry, &mem->attachments, list) {
+		if (entry->bo_va->base.vm != vm)
+			continue;
+		if (entry->bo_va->base.bo->tbo.ttm &&
+		    !entry->bo_va->base.bo->tbo.ttm->sg)
+			continue;
+
+		kfd_mem_dmaunmap_attachment(mem, entry);
+	}
+
+	amdgpu_bo_unreserve(mem->bo);
+out:
+	mutex_unlock(&mem->lock);
+
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu(
+		struct amdgpu_device *adev, struct kgd_mem *mem, void *drm_priv)
+{
+	struct amdgpu_vm *avm = drm_priv_to_vm(drm_priv);
+	unsigned long bo_size = mem->bo->tbo.base.size;
+	struct kfd_mem_attachment *entry;
+	struct bo_vm_reservation_context ctx;
+	int ret;
+
+	mutex_lock(&mem->lock);
+
+	ret = reserve_bo_and_cond_vms(mem, avm, BO_VM_MAPPED, &ctx);
+	if (unlikely(ret))
+		goto out;
+	/* If no VMs were reserved, it means the BO wasn't actually mapped */
+	if (ctx.n_vms == 0) {
+		ret = -EINVAL;
+		goto unreserve_out;
+	}
+
+	ret = vm_validate_pt_pd_bos(avm, NULL);
+	if (unlikely(ret))
+		goto unreserve_out;
+
+	pr_debug("Unmap VA 0x%llx - 0x%llx from vm %p\n",
+		mem->va,
+		mem->va + bo_size * (1 + mem->aql_queue),
+		avm);
+
+	list_for_each_entry(entry, &mem->attachments, list) {
+		if (entry->bo_va->base.vm != avm || !entry->is_mapped)
+			continue;
+
+		pr_debug("\t unmap VA 0x%llx - 0x%llx from entry %p\n",
+			 entry->va, entry->va + bo_size, entry);
+
+		ret = unmap_bo_from_gpuvm(mem, entry, ctx.sync);
+		if (ret)
+			goto unreserve_out;
+
+		entry->is_mapped = false;
+
+		mem->mapped_to_gpu_memory--;
+		pr_debug("\t DEC mapping count %d\n",
+			 mem->mapped_to_gpu_memory);
+	}
+
+unreserve_out:
+	unreserve_bo_and_vms(&ctx, false, false);
+out:
+	mutex_unlock(&mem->lock);
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_sync_memory(
+		struct amdgpu_device *adev, struct kgd_mem *mem, bool intr)
+{
+	struct amdgpu_sync sync;
+	int ret;
+
+	amdgpu_sync_create(&sync);
+
+	mutex_lock(&mem->lock);
+	amdgpu_sync_clone(&mem->sync, &sync);
+	mutex_unlock(&mem->lock);
+
+	ret = amdgpu_sync_wait(&sync, intr);
+	amdgpu_sync_free(&sync);
+	return ret;
+}
+
+/**
+ * amdgpu_amdkfd_map_gtt_bo_to_gart - Map BO to GART and increment reference count
+ * @bo: Buffer object to be mapped
+ * @bo_gart: Return bo reference
+ *
+ * Before return, bo reference count is incremented. To release the reference and unpin/
+ * unmap the BO, call amdgpu_amdkfd_free_gtt_mem.
+ */
+int amdgpu_amdkfd_map_gtt_bo_to_gart(struct amdgpu_bo *bo, struct amdgpu_bo **bo_gart)
+{
+	int ret;
+
+	ret = amdgpu_bo_reserve(bo, true);
+	if (ret) {
+		pr_err("Failed to reserve bo. ret %d\n", ret);
+		goto err_reserve_bo_failed;
+	}
+
+	ret = amdgpu_bo_pin(bo, AMDGPU_GEM_DOMAIN_GTT);
+	if (ret) {
+		pr_err("Failed to pin bo. ret %d\n", ret);
+		goto err_pin_bo_failed;
+	}
+
+	ret = amdgpu_ttm_alloc_gart(&bo->tbo);
+	if (ret) {
+		pr_err("Failed to bind bo to GART. ret %d\n", ret);
+		goto err_map_bo_gart_failed;
+	}
+
+	amdgpu_amdkfd_remove_eviction_fence(
+		bo, bo->vm_bo->vm->process_info->eviction_fence);
+
+	amdgpu_bo_unreserve(bo);
+
+	*bo_gart = amdgpu_bo_ref(bo);
+
+	return 0;
+
+err_map_bo_gart_failed:
+	amdgpu_bo_unpin(bo);
+err_pin_bo_failed:
+	amdgpu_bo_unreserve(bo);
+err_reserve_bo_failed:
+
+	return ret;
+}
+
+/** amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel() - Map a GTT BO for kernel CPU access
+ *
+ * @mem: Buffer object to be mapped for CPU access
+ * @kptr[out]: pointer in kernel CPU address space
+ * @size[out]: size of the buffer
+ *
+ * Pins the BO and maps it for kernel CPU access. The eviction fence is removed
+ * from the BO, since pinned BOs cannot be evicted. The bo must remain on the
+ * validate_list, so the GPU mapping can be restored after a page table was
+ * evicted.
+ *
+ * Return: 0 on success, error code on failure
+ */
+int amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel(struct kgd_mem *mem,
+					     void **kptr, uint64_t *size)
+{
+	int ret;
+	struct amdgpu_bo *bo = mem->bo;
+
+	if (amdgpu_ttm_tt_get_usermm(bo->tbo.ttm)) {
+		pr_err("userptr can't be mapped to kernel\n");
+		return -EINVAL;
+	}
+
+	mutex_lock(&mem->process_info->lock);
+
+	ret = amdgpu_bo_reserve(bo, true);
+	if (ret) {
+		pr_err("Failed to reserve bo. ret %d\n", ret);
+		goto bo_reserve_failed;
+	}
+
+	ret = amdgpu_bo_pin(bo, AMDGPU_GEM_DOMAIN_GTT);
+	if (ret) {
+		pr_err("Failed to pin bo. ret %d\n", ret);
+		goto pin_failed;
+	}
+
+	ret = amdgpu_bo_kmap(bo, kptr);
+	if (ret) {
+		pr_err("Failed to map bo to kernel. ret %d\n", ret);
+		goto kmap_failed;
+	}
+
+	amdgpu_amdkfd_remove_eviction_fence(
+		bo, mem->process_info->eviction_fence);
+
+	if (size)
+		*size = amdgpu_bo_size(bo);
+
+	amdgpu_bo_unreserve(bo);
+
+	mutex_unlock(&mem->process_info->lock);
+	return 0;
+
+kmap_failed:
+	amdgpu_bo_unpin(bo);
+pin_failed:
+	amdgpu_bo_unreserve(bo);
+bo_reserve_failed:
+	mutex_unlock(&mem->process_info->lock);
+
+	return ret;
+}
+
+/** amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel() - Unmap a GTT BO for kernel CPU access
+ *
+ * @mem: Buffer object to be unmapped for CPU access
+ *
+ * Removes the kernel CPU mapping and unpins the BO. It does not restore the
+ * eviction fence, so this function should only be used for cleanup before the
+ * BO is destroyed.
+ */
+void amdgpu_amdkfd_gpuvm_unmap_gtt_bo_from_kernel(struct kgd_mem *mem)
+{
+	struct amdgpu_bo *bo = mem->bo;
+
+	(void)amdgpu_bo_reserve(bo, true);
+	amdgpu_bo_kunmap(bo);
+	amdgpu_bo_unpin(bo);
+	amdgpu_bo_unreserve(bo);
+}
+
+int amdgpu_amdkfd_gpuvm_get_vm_fault_info(struct amdgpu_device *adev,
+					  struct kfd_vm_fault_info *mem)
+{
+	if (atomic_read(&adev->gmc.vm_fault_info_updated) == 1) {
+		*mem = *adev->gmc.vm_fault_info;
+		mb(); /* make sure read happened */
+		atomic_set(&adev->gmc.vm_fault_info_updated, 0);
+	}
+	return 0;
+}
+
+static int import_obj_create(struct amdgpu_device *adev,
+			     struct dma_buf *dma_buf,
+			     struct drm_gem_object *obj,
+			     uint64_t va, void *drm_priv,
+			     struct kgd_mem **mem, uint64_t *size,
+			     uint64_t *mmap_offset)
+{
+	struct amdgpu_vm *avm = drm_priv_to_vm(drm_priv);
+	struct amdgpu_bo *bo;
+	int ret;
+
+	bo = gem_to_amdgpu_bo(obj);
+	if (!(bo->preferred_domains & (AMDGPU_GEM_DOMAIN_VRAM |
+				    AMDGPU_GEM_DOMAIN_GTT)))
+		/* Only VRAM and GTT BOs are supported */
+		return -EINVAL;
+
+	*mem = kzalloc(sizeof(struct kgd_mem), GFP_KERNEL);
+	if (!*mem)
+		return -ENOMEM;
+
+	ret = drm_vma_node_allow(&obj->vma_node, drm_priv);
+	if (ret)
+		goto err_free_mem;
+
+	if (size)
+		*size = amdgpu_bo_size(bo);
+
+	if (mmap_offset)
+		*mmap_offset = amdgpu_bo_mmap_offset(bo);
+
+	INIT_LIST_HEAD(&(*mem)->attachments);
+	mutex_init(&(*mem)->lock);
+
+	(*mem)->alloc_flags =
+		((bo->preferred_domains & AMDGPU_GEM_DOMAIN_VRAM) ?
+		KFD_IOC_ALLOC_MEM_FLAGS_VRAM : KFD_IOC_ALLOC_MEM_FLAGS_GTT)
+		| KFD_IOC_ALLOC_MEM_FLAGS_WRITABLE
+		| KFD_IOC_ALLOC_MEM_FLAGS_EXECUTABLE;
+
+	get_dma_buf(dma_buf);
+	(*mem)->dmabuf = dma_buf;
+	(*mem)->bo = bo;
+	(*mem)->va = va;
+	(*mem)->domain = (bo->preferred_domains & AMDGPU_GEM_DOMAIN_VRAM) &&
+			 !adev->apu_prefer_gtt ?
+			 AMDGPU_GEM_DOMAIN_VRAM : AMDGPU_GEM_DOMAIN_GTT;
+
+	(*mem)->mapped_to_gpu_memory = 0;
+	(*mem)->process_info = avm->process_info;
+	add_kgd_mem_to_kfd_bo_list(*mem, avm->process_info, false);
+	amdgpu_sync_create(&(*mem)->sync);
+	(*mem)->is_imported = true;
+
+	mutex_lock(&avm->process_info->lock);
+	if (avm->process_info->eviction_fence &&
+	    !dma_fence_is_signaled(&avm->process_info->eviction_fence->base))
+		ret = amdgpu_amdkfd_bo_validate_and_fence(bo, (*mem)->domain,
+				&avm->process_info->eviction_fence->base);
+	mutex_unlock(&avm->process_info->lock);
+	if (ret)
+		goto err_remove_mem;
+
+	return 0;
+
+err_remove_mem:
+	remove_kgd_mem_from_kfd_bo_list(*mem, avm->process_info);
+	drm_vma_node_revoke(&obj->vma_node, drm_priv);
+err_free_mem:
+	kfree(*mem);
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_import_dmabuf_fd(struct amdgpu_device *adev, int fd,
+					 uint64_t va, void *drm_priv,
+					 struct kgd_mem **mem, uint64_t *size,
+					 uint64_t *mmap_offset)
+{
+	struct drm_gem_object *obj;
+	uint32_t handle;
+	int ret;
+
+	ret = drm_gem_prime_fd_to_handle(&adev->ddev, adev->kfd.client.file, fd,
+					 &handle);
+	if (ret)
+		return ret;
+	obj = drm_gem_object_lookup(adev->kfd.client.file, handle);
+	if (!obj) {
+		ret = -EINVAL;
+		goto err_release_handle;
+	}
+
+	ret = import_obj_create(adev, obj->dma_buf, obj, va, drm_priv, mem, size,
+				mmap_offset);
+	if (ret)
+		goto err_put_obj;
+
+	(*mem)->gem_handle = handle;
+
+	return 0;
+
+err_put_obj:
+	drm_gem_object_put(obj);
+err_release_handle:
+	drm_gem_handle_delete(adev->kfd.client.file, handle);
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_export_dmabuf(struct kgd_mem *mem,
+				      struct dma_buf **dma_buf)
+{
+	int ret;
+
+	mutex_lock(&mem->lock);
+	ret = kfd_mem_export_dmabuf(mem);
+	if (ret)
+		goto out;
+
+	get_dma_buf(mem->dmabuf);
+	*dma_buf = mem->dmabuf;
+out:
+	mutex_unlock(&mem->lock);
+	return ret;
+}
+
+/* Evict a userptr BO by stopping the queues if necessary
+ *
+ * Runs in MMU notifier, may be in RECLAIM_FS context. This means it
+ * cannot do any memory allocations, and cannot take any locks that
+ * are held elsewhere while allocating memory.
+ *
+ * It doesn't do anything to the BO itself. The real work happens in
+ * restore, where we get updated page addresses. This function only
+ * ensures that GPU access to the BO is stopped.
+ */
+int amdgpu_amdkfd_evict_userptr(struct mmu_interval_notifier *mni,
+				unsigned long cur_seq, struct kgd_mem *mem)
+{
+	struct amdkfd_process_info *process_info = mem->process_info;
+	int r = 0;
+
+	/* Do not process MMU notifications during CRIU restore until
+	 * KFD_CRIU_OP_RESUME IOCTL is received
+	 */
+	if (READ_ONCE(process_info->block_mmu_notifications))
+		return 0;
+
+	mutex_lock(&process_info->notifier_lock);
+	mmu_interval_set_seq(mni, cur_seq);
+
+	mem->invalid++;
+	if (++process_info->evicted_bos == 1) {
+		/* First eviction, stop the queues */
+		r = kgd2kfd_quiesce_mm(mni->mm,
+				       KFD_QUEUE_EVICTION_TRIGGER_USERPTR);
+
+		if (r && r != -ESRCH)
+			pr_err("Failed to quiesce KFD\n");
+
+		if (r != -ESRCH)
+			queue_delayed_work(system_freezable_wq,
+				&process_info->restore_userptr_work,
+				msecs_to_jiffies(AMDGPU_USERPTR_RESTORE_DELAY_MS));
+	}
+	mutex_unlock(&process_info->notifier_lock);
+
+	return r;
+}
+
+/* Update invalid userptr BOs
+ *
+ * Moves invalidated (evicted) userptr BOs from userptr_valid_list to
+ * userptr_inval_list and updates user pages for all BOs that have
+ * been invalidated since their last update.
+ */
+static int update_invalid_user_pages(struct amdkfd_process_info *process_info,
+				     struct mm_struct *mm)
+{
+	struct kgd_mem *mem, *tmp_mem;
+	struct amdgpu_bo *bo;
+	struct ttm_operation_ctx ctx = { false, false };
+	uint32_t invalid;
+	int ret = 0;
+
+	mutex_lock(&process_info->notifier_lock);
+
+	/* Move all invalidated BOs to the userptr_inval_list */
+	list_for_each_entry_safe(mem, tmp_mem,
+				 &process_info->userptr_valid_list,
+				 validate_list)
+		if (mem->invalid)
+			list_move_tail(&mem->validate_list,
+				       &process_info->userptr_inval_list);
+
+	/* Go through userptr_inval_list and update any invalid user_pages */
+	list_for_each_entry(mem, &process_info->userptr_inval_list,
+			    validate_list) {
+		invalid = mem->invalid;
+		if (!invalid)
+			/* BO hasn't been invalidated since the last
+			 * revalidation attempt. Keep its page list.
+			 */
+			continue;
+
+		bo = mem->bo;
+
+		amdgpu_ttm_tt_discard_user_pages(bo->tbo.ttm, mem->range);
+		mem->range = NULL;
+
+		/* BO reservations and getting user pages (hmm_range_fault)
+		 * must happen outside the notifier lock
+		 */
+		mutex_unlock(&process_info->notifier_lock);
+
+		/* Move the BO to system (CPU) domain if necessary to unmap
+		 * and free the SG table
+		 */
+		if (bo->tbo.resource->mem_type != TTM_PL_SYSTEM) {
+			if (amdgpu_bo_reserve(bo, true))
+				return -EAGAIN;
+			amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_CPU);
+			ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+			amdgpu_bo_unreserve(bo);
+			if (ret) {
+				pr_err("%s: Failed to invalidate userptr BO\n",
+				       __func__);
+				return -EAGAIN;
+			}
+		}
+
+		/* Get updated user pages */
+		ret = amdgpu_ttm_tt_get_user_pages(bo, &mem->range);
+		if (ret) {
+			pr_debug("Failed %d to get user pages\n", ret);
+
+			/* Return -EFAULT bad address error as success. It will
+			 * fail later with a VM fault if the GPU tries to access
+			 * it. Better than hanging indefinitely with stalled
+			 * user mode queues.
+			 *
+			 * Return other error -EBUSY or -ENOMEM to retry restore
+			 */
+			if (ret != -EFAULT)
+				return ret;
+
+			/* If applications unmap memory before destroying the userptr
+			 * from the KFD, trigger a segmentation fault in VM debug mode.
+			 */
+			if (amdgpu_ttm_adev(bo->tbo.bdev)->debug_vm_userptr) {
+				struct kfd_process *p;
+
+				pr_err("Pid %d unmapped memory before destroying userptr at GPU addr 0x%llx\n",
+								pid_nr(process_info->pid), mem->va);
+
+				// Send GPU VM fault to user space
+				p = kfd_lookup_process_by_pid(process_info->pid);
+				if (p) {
+					kfd_signal_vm_fault_event_with_userptr(p, mem->va);
+					kfd_unref_process(p);
+				}
+			}
+
+			ret = 0;
+		}
+
+		amdgpu_ttm_tt_set_user_pages(bo->tbo.ttm, mem->range);
+
+		mutex_lock(&process_info->notifier_lock);
+
+		/* Mark the BO as valid unless it was invalidated
+		 * again concurrently.
+		 */
+		if (mem->invalid != invalid) {
+			ret = -EAGAIN;
+			goto unlock_out;
+		}
+		 /* set mem valid if mem has hmm range associated */
+		if (mem->range)
+			mem->invalid = 0;
+	}
+
+unlock_out:
+	mutex_unlock(&process_info->notifier_lock);
+
+	return ret;
+}
+
+/* Validate invalid userptr BOs
+ *
+ * Validates BOs on the userptr_inval_list. Also updates GPUVM page tables
+ * with new page addresses and waits for the page table updates to complete.
+ */
+static int validate_invalid_user_pages(struct amdkfd_process_info *process_info)
+{
+	struct ttm_operation_ctx ctx = { false, false };
+	struct amdgpu_sync sync;
+	struct drm_exec exec;
+
+	struct amdgpu_vm *peer_vm;
+	struct kgd_mem *mem, *tmp_mem;
+	struct amdgpu_bo *bo;
+	int ret;
+
+	amdgpu_sync_create(&sync);
+
+	drm_exec_init(&exec, 0, 0);
+	/* Reserve all BOs and page tables for validation */
+	drm_exec_until_all_locked(&exec) {
+		/* Reserve all the page directories */
+		list_for_each_entry(peer_vm, &process_info->vm_list_head,
+				    vm_list_node) {
+			ret = amdgpu_vm_lock_pd(peer_vm, &exec, 2);
+			drm_exec_retry_on_contention(&exec);
+			if (unlikely(ret))
+				goto unreserve_out;
+		}
+
+		/* Reserve the userptr_inval_list entries to resv_list */
+		list_for_each_entry(mem, &process_info->userptr_inval_list,
+				    validate_list) {
+			struct drm_gem_object *gobj;
+
+			gobj = &mem->bo->tbo.base;
+			ret = drm_exec_prepare_obj(&exec, gobj, 1);
+			drm_exec_retry_on_contention(&exec);
+			if (unlikely(ret))
+				goto unreserve_out;
+		}
+	}
+
+	ret = process_validate_vms(process_info, NULL);
+	if (ret)
+		goto unreserve_out;
+
+	/* Validate BOs and update GPUVM page tables */
+	list_for_each_entry_safe(mem, tmp_mem,
+				 &process_info->userptr_inval_list,
+				 validate_list) {
+		struct kfd_mem_attachment *attachment;
+
+		bo = mem->bo;
+
+		/* Validate the BO if we got user pages */
+		if (bo->tbo.ttm->pages[0]) {
+			amdgpu_bo_placement_from_domain(bo, mem->domain);
+			ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+			if (ret) {
+				pr_err("%s: failed to validate BO\n", __func__);
+				goto unreserve_out;
+			}
+		}
+
+		/* Update mapping. If the BO was not validated
+		 * (because we couldn't get user pages), this will
+		 * clear the page table entries, which will result in
+		 * VM faults if the GPU tries to access the invalid
+		 * memory.
+		 */
+		list_for_each_entry(attachment, &mem->attachments, list) {
+			if (!attachment->is_mapped)
+				continue;
+
+			kfd_mem_dmaunmap_attachment(mem, attachment);
+			ret = update_gpuvm_pte(mem, attachment, &sync);
+			if (ret) {
+				pr_err("%s: update PTE failed\n", __func__);
+				/* make sure this gets validated again */
+				mutex_lock(&process_info->notifier_lock);
+				mem->invalid++;
+				mutex_unlock(&process_info->notifier_lock);
+				goto unreserve_out;
+			}
+		}
+	}
+
+	/* Update page directories */
+	ret = process_update_pds(process_info, &sync);
+
+unreserve_out:
+	drm_exec_fini(&exec);
+	amdgpu_sync_wait(&sync, false);
+	amdgpu_sync_free(&sync);
+
+	return ret;
+}
+
+/* Confirm that all user pages are valid while holding the notifier lock
+ *
+ * Moves valid BOs from the userptr_inval_list back to userptr_val_list.
+ */
+static int confirm_valid_user_pages_locked(struct amdkfd_process_info *process_info)
+{
+	struct kgd_mem *mem, *tmp_mem;
+	int ret = 0;
+
+	list_for_each_entry_safe(mem, tmp_mem,
+				 &process_info->userptr_inval_list,
+				 validate_list) {
+		bool valid;
+
+		/* keep mem without hmm range at userptr_inval_list */
+		if (!mem->range)
+			continue;
+
+		/* Only check mem with hmm range associated */
+		valid = amdgpu_ttm_tt_get_user_pages_done(
+					mem->bo->tbo.ttm, mem->range);
+
+		mem->range = NULL;
+		if (!valid) {
+			WARN(!mem->invalid, "Invalid BO not marked invalid");
+			ret = -EAGAIN;
+			continue;
+		}
+
+		if (mem->invalid) {
+			WARN(1, "Valid BO is marked invalid");
+			ret = -EAGAIN;
+			continue;
+		}
+
+		list_move_tail(&mem->validate_list,
+			       &process_info->userptr_valid_list);
+	}
+
+	return ret;
+}
+
+/* Worker callback to restore evicted userptr BOs
+ *
+ * Tries to update and validate all userptr BOs. If successful and no
+ * concurrent evictions happened, the queues are restarted. Otherwise,
+ * reschedule for another attempt later.
+ */
+static void amdgpu_amdkfd_restore_userptr_worker(struct work_struct *work)
+{
+	struct delayed_work *dwork = to_delayed_work(work);
+	struct amdkfd_process_info *process_info =
+		container_of(dwork, struct amdkfd_process_info,
+			     restore_userptr_work);
+	struct task_struct *usertask;
+	struct mm_struct *mm;
+	uint32_t evicted_bos;
+
+	mutex_lock(&process_info->notifier_lock);
+	evicted_bos = process_info->evicted_bos;
+	mutex_unlock(&process_info->notifier_lock);
+	if (!evicted_bos)
+		return;
+
+	/* Reference task and mm in case of concurrent process termination */
+	usertask = get_pid_task(process_info->pid, PIDTYPE_PID);
+	if (!usertask)
+		return;
+	mm = get_task_mm(usertask);
+	if (!mm) {
+		put_task_struct(usertask);
+		return;
+	}
+
+	mutex_lock(&process_info->lock);
+
+	if (update_invalid_user_pages(process_info, mm))
+		goto unlock_out;
+	/* userptr_inval_list can be empty if all evicted userptr BOs
+	 * have been freed. In that case there is nothing to validate
+	 * and we can just restart the queues.
+	 */
+	if (!list_empty(&process_info->userptr_inval_list)) {
+		if (validate_invalid_user_pages(process_info))
+			goto unlock_out;
+	}
+	/* Final check for concurrent evicton and atomic update. If
+	 * another eviction happens after successful update, it will
+	 * be a first eviction that calls quiesce_mm. The eviction
+	 * reference counting inside KFD will handle this case.
+	 */
+	mutex_lock(&process_info->notifier_lock);
+	if (process_info->evicted_bos != evicted_bos)
+		goto unlock_notifier_out;
+
+	if (confirm_valid_user_pages_locked(process_info)) {
+		WARN(1, "User pages unexpectedly invalid");
+		goto unlock_notifier_out;
+	}
+
+	process_info->evicted_bos = evicted_bos = 0;
+
+	if (kgd2kfd_resume_mm(mm)) {
+		pr_err("%s: Failed to resume KFD\n", __func__);
+		/* No recovery from this failure. Probably the CP is
+		 * hanging. No point trying again.
+		 */
+	}
+
+unlock_notifier_out:
+	mutex_unlock(&process_info->notifier_lock);
+unlock_out:
+	mutex_unlock(&process_info->lock);
+
+	/* If validation failed, reschedule another attempt */
+	if (evicted_bos) {
+		queue_delayed_work(system_freezable_wq,
+			&process_info->restore_userptr_work,
+			msecs_to_jiffies(AMDGPU_USERPTR_RESTORE_DELAY_MS));
+
+		kfd_smi_event_queue_restore_rescheduled(mm);
+	}
+	mmput(mm);
+	put_task_struct(usertask);
+}
+
+static void replace_eviction_fence(struct dma_fence __rcu **ef,
+				   struct dma_fence *new_ef)
+{
+	struct dma_fence *old_ef = rcu_replace_pointer(*ef, new_ef, true
+		/* protected by process_info->lock */);
+
+	/* If we're replacing an unsignaled eviction fence, that fence will
+	 * never be signaled, and if anyone is still waiting on that fence,
+	 * they will hang forever. This should never happen. We should only
+	 * replace the fence in restore_work that only gets scheduled after
+	 * eviction work signaled the fence.
+	 */
+	WARN_ONCE(!dma_fence_is_signaled(old_ef),
+		  "Replacing unsignaled eviction fence");
+	dma_fence_put(old_ef);
+}
+
+/** amdgpu_amdkfd_gpuvm_restore_process_bos - Restore all BOs for the given
+ *   KFD process identified by process_info
+ *
+ * @process_info: amdkfd_process_info of the KFD process
+ *
+ * After memory eviction, restore thread calls this function. The function
+ * should be called when the Process is still valid. BO restore involves -
+ *
+ * 1.  Release old eviction fence and create new one
+ * 2.  Get two copies of PD BO list from all the VMs. Keep one copy as pd_list.
+ * 3   Use the second PD list and kfd_bo_list to create a list (ctx.list) of
+ *     BOs that need to be reserved.
+ * 4.  Reserve all the BOs
+ * 5.  Validate of PD and PT BOs.
+ * 6.  Validate all KFD BOs using kfd_bo_list and Map them and add new fence
+ * 7.  Add fence to all PD and PT BOs.
+ * 8.  Unreserve all BOs
+ */
+int amdgpu_amdkfd_gpuvm_restore_process_bos(void *info, struct dma_fence __rcu **ef)
+{
+	struct amdkfd_process_info *process_info = info;
+	struct amdgpu_vm *peer_vm;
+	struct kgd_mem *mem;
+	struct list_head duplicate_save;
+	struct amdgpu_sync sync_obj;
+	unsigned long failed_size = 0;
+	unsigned long total_size = 0;
+	struct drm_exec exec;
+	int ret;
+
+	INIT_LIST_HEAD(&duplicate_save);
+
+	mutex_lock(&process_info->lock);
+
+	drm_exec_init(&exec, DRM_EXEC_IGNORE_DUPLICATES, 0);
+	drm_exec_until_all_locked(&exec) {
+		list_for_each_entry(peer_vm, &process_info->vm_list_head,
+				    vm_list_node) {
+			ret = amdgpu_vm_lock_pd(peer_vm, &exec, 2);
+			drm_exec_retry_on_contention(&exec);
+			if (unlikely(ret)) {
+				pr_err("Locking VM PD failed, ret: %d\n", ret);
+				goto ttm_reserve_fail;
+			}
+		}
+
+		/* Reserve all BOs and page tables/directory. Add all BOs from
+		 * kfd_bo_list to ctx.list
+		 */
+		list_for_each_entry(mem, &process_info->kfd_bo_list,
+				    validate_list) {
+			struct drm_gem_object *gobj;
+
+			gobj = &mem->bo->tbo.base;
+			ret = drm_exec_prepare_obj(&exec, gobj, 1);
+			drm_exec_retry_on_contention(&exec);
+			if (unlikely(ret)) {
+				pr_err("drm_exec_prepare_obj failed, ret: %d\n", ret);
+				goto ttm_reserve_fail;
+			}
+		}
+	}
+
+	amdgpu_sync_create(&sync_obj);
+
+	/* Validate BOs managed by KFD */
+	list_for_each_entry(mem, &process_info->kfd_bo_list,
+			    validate_list) {
+
+		struct amdgpu_bo *bo = mem->bo;
+		uint32_t domain = mem->domain;
+		struct dma_resv_iter cursor;
+		struct dma_fence *fence;
+
+		total_size += amdgpu_bo_size(bo);
+
+		ret = amdgpu_amdkfd_bo_validate(bo, domain, false);
+		if (ret) {
+			pr_debug("Memory eviction: Validate BOs failed\n");
+			failed_size += amdgpu_bo_size(bo);
+			ret = amdgpu_amdkfd_bo_validate(bo,
+						AMDGPU_GEM_DOMAIN_GTT, false);
+			if (ret) {
+				pr_debug("Memory eviction: Try again\n");
+				goto validate_map_fail;
+			}
+		}
+		dma_resv_for_each_fence(&cursor, bo->tbo.base.resv,
+					DMA_RESV_USAGE_KERNEL, fence) {
+			ret = amdgpu_sync_fence(&sync_obj, fence, GFP_KERNEL);
+			if (ret) {
+				pr_debug("Memory eviction: Sync BO fence failed. Try again\n");
+				goto validate_map_fail;
+			}
+		}
+	}
+
+	if (failed_size)
+		pr_debug("0x%lx/0x%lx in system\n", failed_size, total_size);
+
+	/* Validate PDs, PTs and evicted DMABuf imports last. Otherwise BO
+	 * validations above would invalidate DMABuf imports again.
+	 */
+	ret = process_validate_vms(process_info, &exec.ticket);
+	if (ret) {
+		pr_debug("Validating VMs failed, ret: %d\n", ret);
+		goto validate_map_fail;
+	}
+
+	/* Update mappings managed by KFD. */
+	list_for_each_entry(mem, &process_info->kfd_bo_list,
+			    validate_list) {
+		struct kfd_mem_attachment *attachment;
+
+		list_for_each_entry(attachment, &mem->attachments, list) {
+			if (!attachment->is_mapped)
+				continue;
+
+			kfd_mem_dmaunmap_attachment(mem, attachment);
+			ret = update_gpuvm_pte(mem, attachment, &sync_obj);
+			if (ret) {
+				pr_debug("Memory eviction: update PTE failed. Try again\n");
+				goto validate_map_fail;
+			}
+		}
+	}
+
+	/* Update mappings not managed by KFD */
+	list_for_each_entry(peer_vm, &process_info->vm_list_head,
+			vm_list_node) {
+		struct amdgpu_device *adev = amdgpu_ttm_adev(
+			peer_vm->root.bo->tbo.bdev);
+
+		struct amdgpu_fpriv *fpriv =
+			container_of(peer_vm, struct amdgpu_fpriv, vm);
+
+		ret = amdgpu_vm_bo_update(adev, fpriv->prt_va, false);
+		if (ret) {
+			dev_dbg(adev->dev,
+				"Memory eviction: handle PRT moved failed, pid %8d. Try again.\n",
+				pid_nr(process_info->pid));
+			goto validate_map_fail;
+		}
+
+		ret = amdgpu_vm_handle_moved(adev, peer_vm, &exec.ticket);
+		if (ret) {
+			dev_dbg(adev->dev,
+				"Memory eviction: handle moved failed, pid %8d. Try again.\n",
+				pid_nr(process_info->pid));
+			goto validate_map_fail;
+		}
+	}
+
+	/* Update page directories */
+	ret = process_update_pds(process_info, &sync_obj);
+	if (ret) {
+		pr_debug("Memory eviction: update PDs failed. Try again\n");
+		goto validate_map_fail;
+	}
+
+	/* Sync with fences on all the page tables. They implicitly depend on any
+	 * move fences from amdgpu_vm_handle_moved above.
+	 */
+	ret = process_sync_pds_resv(process_info, &sync_obj);
+	if (ret) {
+		pr_debug("Memory eviction: Failed to sync to PD BO moving fence. Try again\n");
+		goto validate_map_fail;
+	}
+
+	/* Wait for validate and PT updates to finish */
+	amdgpu_sync_wait(&sync_obj, false);
+
+	/* The old eviction fence may be unsignaled if restore happens
+	 * after a GPU reset or suspend/resume. Keep the old fence in that
+	 * case. Otherwise release the old eviction fence and create new
+	 * one, because fence only goes from unsignaled to signaled once
+	 * and cannot be reused. Use context and mm from the old fence.
+	 *
+	 * If an old eviction fence signals after this check, that's OK.
+	 * Anyone signaling an eviction fence must stop the queues first
+	 * and schedule another restore worker.
+	 */
+	if (dma_fence_is_signaled(&process_info->eviction_fence->base)) {
+		struct amdgpu_amdkfd_fence *new_fence =
+			amdgpu_amdkfd_fence_create(
+				process_info->eviction_fence->base.context,
+				process_info->eviction_fence->mm,
+				NULL);
+
+		if (!new_fence) {
+			pr_err("Failed to create eviction fence\n");
+			ret = -ENOMEM;
+			goto validate_map_fail;
+		}
+		dma_fence_put(&process_info->eviction_fence->base);
+		process_info->eviction_fence = new_fence;
+		replace_eviction_fence(ef, dma_fence_get(&new_fence->base));
+	} else {
+		WARN_ONCE(*ef != &process_info->eviction_fence->base,
+			  "KFD eviction fence doesn't match KGD process_info");
+	}
+
+	/* Attach new eviction fence to all BOs except pinned ones */
+	list_for_each_entry(mem, &process_info->kfd_bo_list, validate_list) {
+		if (mem->bo->tbo.pin_count)
+			continue;
+
+		dma_resv_add_fence(mem->bo->tbo.base.resv,
+				   &process_info->eviction_fence->base,
+				   DMA_RESV_USAGE_BOOKKEEP);
+	}
+	/* Attach eviction fence to PD / PT BOs and DMABuf imports */
+	list_for_each_entry(peer_vm, &process_info->vm_list_head,
+			    vm_list_node) {
+		struct amdgpu_bo *bo = peer_vm->root.bo;
+
+		dma_resv_add_fence(bo->tbo.base.resv,
+				   &process_info->eviction_fence->base,
+				   DMA_RESV_USAGE_BOOKKEEP);
+	}
+
+validate_map_fail:
+	amdgpu_sync_free(&sync_obj);
+ttm_reserve_fail:
+	drm_exec_fini(&exec);
+	mutex_unlock(&process_info->lock);
+	return ret;
+}
+
+int amdgpu_amdkfd_add_gws_to_process(void *info, void *gws, struct kgd_mem **mem)
+{
+	struct amdkfd_process_info *process_info = (struct amdkfd_process_info *)info;
+	struct amdgpu_bo *gws_bo = (struct amdgpu_bo *)gws;
+	int ret;
+
+	if (!info || !gws)
+		return -EINVAL;
+
+	*mem = kzalloc(sizeof(struct kgd_mem), GFP_KERNEL);
+	if (!*mem)
+		return -ENOMEM;
+
+	mutex_init(&(*mem)->lock);
+	INIT_LIST_HEAD(&(*mem)->attachments);
+	(*mem)->bo = amdgpu_bo_ref(gws_bo);
+	(*mem)->domain = AMDGPU_GEM_DOMAIN_GWS;
+	(*mem)->process_info = process_info;
+	add_kgd_mem_to_kfd_bo_list(*mem, process_info, false);
+	amdgpu_sync_create(&(*mem)->sync);
+
+
+	/* Validate gws bo the first time it is added to process */
+	mutex_lock(&(*mem)->process_info->lock);
+	ret = amdgpu_bo_reserve(gws_bo, false);
+	if (unlikely(ret)) {
+		pr_err("Reserve gws bo failed %d\n", ret);
+		goto bo_reservation_failure;
+	}
+
+	ret = amdgpu_amdkfd_bo_validate(gws_bo, AMDGPU_GEM_DOMAIN_GWS, true);
+	if (ret) {
+		pr_err("GWS BO validate failed %d\n", ret);
+		goto bo_validation_failure;
+	}
+	/* GWS resource is shared b/t amdgpu and amdkfd
+	 * Add process eviction fence to bo so they can
+	 * evict each other.
+	 */
+	ret = dma_resv_reserve_fences(gws_bo->tbo.base.resv, 1);
+	if (ret)
+		goto reserve_shared_fail;
+	dma_resv_add_fence(gws_bo->tbo.base.resv,
+			   &process_info->eviction_fence->base,
+			   DMA_RESV_USAGE_BOOKKEEP);
+	amdgpu_bo_unreserve(gws_bo);
+	mutex_unlock(&(*mem)->process_info->lock);
+
+	return ret;
+
+reserve_shared_fail:
+bo_validation_failure:
+	amdgpu_bo_unreserve(gws_bo);
+bo_reservation_failure:
+	mutex_unlock(&(*mem)->process_info->lock);
+	amdgpu_sync_free(&(*mem)->sync);
+	remove_kgd_mem_from_kfd_bo_list(*mem, process_info);
+	amdgpu_bo_unref(&gws_bo);
+	mutex_destroy(&(*mem)->lock);
+	kfree(*mem);
+	*mem = NULL;
+	return ret;
+}
+
+int amdgpu_amdkfd_remove_gws_from_process(void *info, void *mem)
+{
+	int ret;
+	struct amdkfd_process_info *process_info = (struct amdkfd_process_info *)info;
+	struct kgd_mem *kgd_mem = (struct kgd_mem *)mem;
+	struct amdgpu_bo *gws_bo = kgd_mem->bo;
+
+	/* Remove BO from process's validate list so restore worker won't touch
+	 * it anymore
+	 */
+	remove_kgd_mem_from_kfd_bo_list(kgd_mem, process_info);
+
+	ret = amdgpu_bo_reserve(gws_bo, false);
+	if (unlikely(ret)) {
+		pr_err("Reserve gws bo failed %d\n", ret);
+		//TODO add BO back to validate_list?
+		return ret;
+	}
+	amdgpu_amdkfd_remove_eviction_fence(gws_bo,
+			process_info->eviction_fence);
+	amdgpu_bo_unreserve(gws_bo);
+	amdgpu_sync_free(&kgd_mem->sync);
+	amdgpu_bo_unref(&gws_bo);
+	mutex_destroy(&kgd_mem->lock);
+	kfree(mem);
+	return 0;
+}
+
+/* Returns GPU-specific tiling mode information */
+int amdgpu_amdkfd_get_tile_config(struct amdgpu_device *adev,
+				struct tile_config *config)
+{
+	config->gb_addr_config = adev->gfx.config.gb_addr_config;
+	config->tile_config_ptr = adev->gfx.config.tile_mode_array;
+	config->num_tile_configs =
+			ARRAY_SIZE(adev->gfx.config.tile_mode_array);
+	config->macro_tile_config_ptr =
+			adev->gfx.config.macrotile_mode_array;
+	config->num_macro_tile_configs =
+			ARRAY_SIZE(adev->gfx.config.macrotile_mode_array);
+
+	/* Those values are not set from GFX9 onwards */
+	config->num_banks = adev->gfx.config.num_banks;
+	config->num_ranks = adev->gfx.config.num_ranks;
+
+	return 0;
+}
+
+bool amdgpu_amdkfd_bo_mapped_to_dev(void *drm_priv, struct kgd_mem *mem)
+{
+	struct amdgpu_vm *vm = drm_priv_to_vm(drm_priv);
+	struct kfd_mem_attachment *entry;
+
+	list_for_each_entry(entry, &mem->attachments, list) {
+		if (entry->is_mapped && entry->bo_va->base.vm == vm)
+			return true;
+	}
+	return false;
+}
+
+#if defined(CONFIG_DEBUG_FS)
+
+int kfd_debugfs_kfd_mem_limits(struct seq_file *m, void *data)
+{
+
+	spin_lock(&kfd_mem_limit.mem_limit_lock);
+	seq_printf(m, "System mem used %lldM out of %lluM\n",
+		  (kfd_mem_limit.system_mem_used >> 20),
+		  (kfd_mem_limit.max_system_mem_limit >> 20));
+	seq_printf(m, "TTM mem used %lldM out of %lluM\n",
+		  (kfd_mem_limit.ttm_mem_used >> 20),
+		  (kfd_mem_limit.max_ttm_mem_limit >> 20));
+	spin_unlock(&kfd_mem_limit.mem_limit_lock);
+
+	return 0;
+}
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c
index ce443586a0c7..763f2b8dcf13 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c
@@ -23,24 +23,19 @@
  * Authors: Dave Airlie
  *          Alex Deucher
  */
-#include <drm/drmP.h>
+
 #include <drm/amdgpu_drm.h>
 #include "amdgpu.h"
 #include "amdgpu_atombios.h"
+#include "amdgpu_atomfirmware.h"
 #include "amdgpu_i2c.h"
+#include "amdgpu_display.h"
 
 #include "atom.h"
 #include "atom-bits.h"
 #include "atombios_encoders.h"
 #include "bif/bif_4_1_d.h"
 
-static void amdgpu_atombios_lookup_i2c_gpio_quirks(struct amdgpu_device *adev,
-					  ATOM_GPIO_I2C_ASSIGMENT *gpio,
-					  u8 index)
-{
-
-}
-
 static struct amdgpu_i2c_bus_rec amdgpu_atombios_get_bus_rec_for_i2c_gpio(ATOM_GPIO_I2C_ASSIGMENT *gpio)
 {
 	struct amdgpu_i2c_bus_rec i2c;
@@ -106,9 +101,6 @@ struct amdgpu_i2c_bus_rec amdgpu_atombios_lookup_i2c_gpio(struct amdgpu_device *
 
 		gpio = &i2c_info->asGPIO_Info[0];
 		for (i = 0; i < num_indices; i++) {
-
-			amdgpu_atombios_lookup_i2c_gpio_quirks(adev, gpio, i);
-
 			if (gpio->sucI2cId.ucAccess == id) {
 				i2c = amdgpu_atombios_get_bus_rec_for_i2c_gpio(gpio);
 				break;
@@ -140,13 +132,43 @@ void amdgpu_atombios_i2c_init(struct amdgpu_device *adev)
 
 		gpio = &i2c_info->asGPIO_Info[0];
 		for (i = 0; i < num_indices; i++) {
-			amdgpu_atombios_lookup_i2c_gpio_quirks(adev, gpio, i);
-
 			i2c = amdgpu_atombios_get_bus_rec_for_i2c_gpio(gpio);
 
 			if (i2c.valid) {
 				sprintf(stmp, "0x%x", i2c.i2c_id);
-				adev->i2c_bus[i] = amdgpu_i2c_create(adev->ddev, &i2c, stmp);
+				adev->i2c_bus[i] = amdgpu_i2c_create(adev_to_drm(adev), &i2c, stmp);
+			}
+			gpio = (ATOM_GPIO_I2C_ASSIGMENT *)
+				((u8 *)gpio + sizeof(ATOM_GPIO_I2C_ASSIGMENT));
+		}
+	}
+}
+
+void amdgpu_atombios_oem_i2c_init(struct amdgpu_device *adev, u8 i2c_id)
+{
+	struct atom_context *ctx = adev->mode_info.atom_context;
+	ATOM_GPIO_I2C_ASSIGMENT *gpio;
+	struct amdgpu_i2c_bus_rec i2c;
+	int index = GetIndexIntoMasterTable(DATA, GPIO_I2C_Info);
+	struct _ATOM_GPIO_I2C_INFO *i2c_info;
+	uint16_t data_offset, size;
+	int i, num_indices;
+	char stmp[32];
+
+	if (amdgpu_atom_parse_data_header(ctx, index, &size, NULL, NULL, &data_offset)) {
+		i2c_info = (struct _ATOM_GPIO_I2C_INFO *)(ctx->bios + data_offset);
+
+		num_indices = (size - sizeof(ATOM_COMMON_TABLE_HEADER)) /
+			sizeof(ATOM_GPIO_I2C_ASSIGMENT);
+
+		gpio = &i2c_info->asGPIO_Info[0];
+		for (i = 0; i < num_indices; i++) {
+			i2c = amdgpu_atombios_get_bus_rec_for_i2c_gpio(gpio);
+
+			if (i2c.valid && i2c.i2c_id == i2c_id) {
+				sprintf(stmp, "OEM 0x%x", i2c.i2c_id);
+				adev->i2c_bus[i] = amdgpu_i2c_create(adev_to_drm(adev), &i2c, stmp);
+				break;
 			}
 			gpio = (ATOM_GPIO_I2C_ASSIGMENT *)
 				((u8 *)gpio + sizeof(ATOM_GPIO_I2C_ASSIGMENT));
@@ -336,17 +358,9 @@ bool amdgpu_atombios_get_connector_info_from_object_table(struct amdgpu_device *
 		path_size += le16_to_cpu(path->usSize);
 
 		if (device_support & le16_to_cpu(path->usDeviceTag)) {
-			uint8_t con_obj_id, con_obj_num, con_obj_type;
-
-			con_obj_id =
+			uint8_t con_obj_id =
 			    (le16_to_cpu(path->usConnObjectId) & OBJECT_ID_MASK)
 			    >> OBJECT_ID_SHIFT;
-			con_obj_num =
-			    (le16_to_cpu(path->usConnObjectId) & ENUM_ID_MASK)
-			    >> ENUM_ID_SHIFT;
-			con_obj_type =
-			    (le16_to_cpu(path->usConnObjectId) &
-			     OBJECT_TYPE_MASK) >> OBJECT_TYPE_SHIFT;
 
 			/* Skip TV/CV support */
 			if ((le16_to_cpu(path->usDeviceTag) ==
@@ -371,15 +385,7 @@ bool amdgpu_atombios_get_connector_info_from_object_table(struct amdgpu_device *
 			router.ddc_valid = false;
 			router.cd_valid = false;
 			for (j = 0; j < ((le16_to_cpu(path->usSize) - 8) / 2); j++) {
-				uint8_t grph_obj_id, grph_obj_num, grph_obj_type;
-
-				grph_obj_id =
-				    (le16_to_cpu(path->usGraphicObjIds[j]) &
-				     OBJECT_ID_MASK) >> OBJECT_ID_SHIFT;
-				grph_obj_num =
-				    (le16_to_cpu(path->usGraphicObjIds[j]) &
-				     ENUM_ID_MASK) >> ENUM_ID_SHIFT;
-				grph_obj_type =
+				uint8_t grph_obj_type =
 				    (le16_to_cpu(path->usGraphicObjIds[j]) &
 				     OBJECT_TYPE_MASK) >> OBJECT_TYPE_SHIFT;
 
@@ -555,7 +561,7 @@ bool amdgpu_atombios_get_connector_info_from_object_table(struct amdgpu_device *
 		}
 	}
 
-	amdgpu_link_encoder_connector(adev->ddev);
+	amdgpu_link_encoder_connector(adev_to_drm(adev));
 
 	return true;
 }
@@ -690,17 +696,16 @@ int amdgpu_atombios_get_clock_info(struct amdgpu_device *adev)
 			le32_to_cpu(firmware_info->info_21.ulDefaultDispEngineClkFreq);
 		/* set a reasonable default for DP */
 		if (adev->clock.default_dispclk < 53900) {
-			DRM_INFO("Changing default dispclk from %dMhz to 600Mhz\n",
-				 adev->clock.default_dispclk / 100);
+			DRM_DEBUG("Changing default dispclk from %dMhz to 600Mhz\n",
+				  adev->clock.default_dispclk / 100);
 			adev->clock.default_dispclk = 60000;
 		} else if (adev->clock.default_dispclk <= 60000) {
-			DRM_INFO("Changing default dispclk from %dMhz to 625Mhz\n",
-				 adev->clock.default_dispclk / 100);
+			DRM_DEBUG("Changing default dispclk from %dMhz to 625Mhz\n",
+				  adev->clock.default_dispclk / 100);
 			adev->clock.default_dispclk = 62500;
 		}
 		adev->clock.dp_extclk =
 			le16_to_cpu(firmware_info->info_21.usUniphyDPModeExtClkFreq);
-		adev->clock.current_dispclk = adev->clock.default_dispclk;
 
 		adev->clock.max_pixel_clock = le16_to_cpu(firmware_info->info.usMaxPixelClock);
 		if (adev->clock.max_pixel_clock == 0)
@@ -1032,7 +1037,9 @@ int amdgpu_atombios_get_clock_dividers(struct amdgpu_device *adev,
 		if (clock_type == COMPUTE_ENGINE_PLL_PARAM) {
 			args.v3.ulClockParams = cpu_to_le32((clock_type << 24) | clock);
 
-			amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+			if (amdgpu_atom_execute_table(adev->mode_info.atom_context,
+			    index, (uint32_t *)&args, sizeof(args)))
+				return -EINVAL;
 
 			dividers->post_div = args.v3.ucPostDiv;
 			dividers->enable_post_div = (args.v3.ucCntlFlag &
@@ -1052,7 +1059,9 @@ int amdgpu_atombios_get_clock_dividers(struct amdgpu_device *adev,
 			if (strobe_mode)
 				args.v5.ucInputFlag = ATOM_PLL_INPUT_FLAG_PLL_STROBE_MODE_EN;
 
-			amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+			if (amdgpu_atom_execute_table(adev->mode_info.atom_context,
+			    index, (uint32_t *)&args, sizeof(args)))
+				return -EINVAL;
 
 			dividers->post_div = args.v5.ucPostDiv;
 			dividers->enable_post_div = (args.v5.ucCntlFlag &
@@ -1070,7 +1079,9 @@ int amdgpu_atombios_get_clock_dividers(struct amdgpu_device *adev,
 		/* fusion */
 		args.v4.ulClock = cpu_to_le32(clock);	/* 10 khz */
 
-		amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+		if (amdgpu_atom_execute_table(adev->mode_info.atom_context,
+		    index, (uint32_t *)&args, sizeof(args)))
+			return -EINVAL;
 
 		dividers->post_divider = dividers->post_div = args.v4.ucPostDiv;
 		dividers->real_clock = le32_to_cpu(args.v4.ulClock);
@@ -1081,7 +1092,9 @@ int amdgpu_atombios_get_clock_dividers(struct amdgpu_device *adev,
 		args.v6_in.ulClock.ulComputeClockFlag = clock_type;
 		args.v6_in.ulClock.ulClockFreq = cpu_to_le32(clock);	/* 10 khz */
 
-		amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+		if (amdgpu_atom_execute_table(adev->mode_info.atom_context,
+		    index, (uint32_t *)&args, sizeof(args)))
+			return -EINVAL;
 
 		dividers->whole_fb_div = le16_to_cpu(args.v6_out.ulFbDiv.usFbDiv);
 		dividers->frac_fb_div = le16_to_cpu(args.v6_out.ulFbDiv.usFbDivFrac);
@@ -1097,6 +1110,7 @@ int amdgpu_atombios_get_clock_dividers(struct amdgpu_device *adev,
 	return 0;
 }
 
+#ifdef CONFIG_DRM_AMDGPU_SI
 int amdgpu_atombios_get_memory_pll_dividers(struct amdgpu_device *adev,
 					    u32 clock,
 					    bool strobe_mode,
@@ -1122,7 +1136,9 @@ int amdgpu_atombios_get_memory_pll_dividers(struct amdgpu_device *adev,
 			if (strobe_mode)
 				args.ucInputFlag |= MPLL_INPUT_FLAG_STROBE_MODE_EN;
 
-			amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+			if (amdgpu_atom_execute_table(adev->mode_info.atom_context,
+			    index, (uint32_t *)&args, sizeof(args)))
+				return -EINVAL;
 
 			mpll_param->clkfrac = le16_to_cpu(args.ulFbDiv.usFbDivFrac);
 			mpll_param->clkf = le16_to_cpu(args.ulFbDiv.usFbDiv);
@@ -1148,8 +1164,8 @@ int amdgpu_atombios_get_memory_pll_dividers(struct amdgpu_device *adev,
 	return 0;
 }
 
-void amdgpu_atombios_set_engine_dram_timings(struct amdgpu_device *adev,
-					     u32 eng_clock, u32 mem_clock)
+int amdgpu_atombios_set_engine_dram_timings(struct amdgpu_device *adev,
+					    u32 eng_clock, u32 mem_clock)
 {
 	SET_ENGINE_CLOCK_PS_ALLOCATION args;
 	int index = GetIndexIntoMasterTable(COMMAND, DynamicMemorySettings);
@@ -1164,7 +1180,8 @@ void amdgpu_atombios_set_engine_dram_timings(struct amdgpu_device *adev,
 	if (mem_clock)
 		args.sReserved.ulClock = cpu_to_le32(mem_clock & SET_CLOCK_FREQ_MASK);
 
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	return amdgpu_atom_execute_table(adev->mode_info.atom_context, index,
+					 (uint32_t *)&args, sizeof(args));
 }
 
 void amdgpu_atombios_get_default_voltages(struct amdgpu_device *adev,
@@ -1218,7 +1235,9 @@ int amdgpu_atombios_get_max_vddc(struct amdgpu_device *adev, u8 voltage_type,
 		args.v2.ucVoltageMode = 0;
 		args.v2.usVoltageLevel = 0;
 
-		amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+		if (amdgpu_atom_execute_table(adev->mode_info.atom_context,
+		    index, (uint32_t *)&args, sizeof(args)))
+			return -EINVAL;
 
 		*voltage = le16_to_cpu(args.v2.usVoltageLevel);
 		break;
@@ -1227,7 +1246,9 @@ int amdgpu_atombios_get_max_vddc(struct amdgpu_device *adev, u8 voltage_type,
 		args.v3.ucVoltageMode = ATOM_GET_VOLTAGE_LEVEL;
 		args.v3.usVoltageLevel = cpu_to_le16(voltage_id);
 
-		amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+		if (amdgpu_atom_execute_table(adev->mode_info.atom_context,
+		    index, (uint32_t *)&args, sizeof(args)))
+			return -EINVAL;
 
 		*voltage = le16_to_cpu(args.v3.usVoltageLevel);
 		break;
@@ -1246,157 +1267,6 @@ int amdgpu_atombios_get_leakage_vddc_based_on_leakage_idx(struct amdgpu_device *
 	return amdgpu_atombios_get_max_vddc(adev, VOLTAGE_TYPE_VDDC, leakage_idx, voltage);
 }
 
-int amdgpu_atombios_get_leakage_id_from_vbios(struct amdgpu_device *adev,
-					      u16 *leakage_id)
-{
-	union set_voltage args;
-	int index = GetIndexIntoMasterTable(COMMAND, SetVoltage);
-	u8 frev, crev;
-
-	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
-		return -EINVAL;
-
-	switch (crev) {
-	case 3:
-	case 4:
-		args.v3.ucVoltageType = 0;
-		args.v3.ucVoltageMode = ATOM_GET_LEAKAGE_ID;
-		args.v3.usVoltageLevel = 0;
-
-		amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
-
-		*leakage_id = le16_to_cpu(args.v3.usVoltageLevel);
-		break;
-	default:
-		DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
-		return -EINVAL;
-	}
-
-	return 0;
-}
-
-int amdgpu_atombios_get_leakage_vddc_based_on_leakage_params(struct amdgpu_device *adev,
-							     u16 *vddc, u16 *vddci,
-							     u16 virtual_voltage_id,
-							     u16 vbios_voltage_id)
-{
-	int index = GetIndexIntoMasterTable(DATA, ASIC_ProfilingInfo);
-	u8 frev, crev;
-	u16 data_offset, size;
-	int i, j;
-	ATOM_ASIC_PROFILING_INFO_V2_1 *profile;
-	u16 *leakage_bin, *vddc_id_buf, *vddc_buf, *vddci_id_buf, *vddci_buf;
-
-	*vddc = 0;
-	*vddci = 0;
-
-	if (!amdgpu_atom_parse_data_header(adev->mode_info.atom_context, index, &size,
-				    &frev, &crev, &data_offset))
-		return -EINVAL;
-
-	profile = (ATOM_ASIC_PROFILING_INFO_V2_1 *)
-		(adev->mode_info.atom_context->bios + data_offset);
-
-	switch (frev) {
-	case 1:
-		return -EINVAL;
-	case 2:
-		switch (crev) {
-		case 1:
-			if (size < sizeof(ATOM_ASIC_PROFILING_INFO_V2_1))
-				return -EINVAL;
-			leakage_bin = (u16 *)
-				(adev->mode_info.atom_context->bios + data_offset +
-				 le16_to_cpu(profile->usLeakageBinArrayOffset));
-			vddc_id_buf = (u16 *)
-				(adev->mode_info.atom_context->bios + data_offset +
-				 le16_to_cpu(profile->usElbVDDC_IdArrayOffset));
-			vddc_buf = (u16 *)
-				(adev->mode_info.atom_context->bios + data_offset +
-				 le16_to_cpu(profile->usElbVDDC_LevelArrayOffset));
-			vddci_id_buf = (u16 *)
-				(adev->mode_info.atom_context->bios + data_offset +
-				 le16_to_cpu(profile->usElbVDDCI_IdArrayOffset));
-			vddci_buf = (u16 *)
-				(adev->mode_info.atom_context->bios + data_offset +
-				 le16_to_cpu(profile->usElbVDDCI_LevelArrayOffset));
-
-			if (profile->ucElbVDDC_Num > 0) {
-				for (i = 0; i < profile->ucElbVDDC_Num; i++) {
-					if (vddc_id_buf[i] == virtual_voltage_id) {
-						for (j = 0; j < profile->ucLeakageBinNum; j++) {
-							if (vbios_voltage_id <= leakage_bin[j]) {
-								*vddc = vddc_buf[j * profile->ucElbVDDC_Num + i];
-								break;
-							}
-						}
-						break;
-					}
-				}
-			}
-			if (profile->ucElbVDDCI_Num > 0) {
-				for (i = 0; i < profile->ucElbVDDCI_Num; i++) {
-					if (vddci_id_buf[i] == virtual_voltage_id) {
-						for (j = 0; j < profile->ucLeakageBinNum; j++) {
-							if (vbios_voltage_id <= leakage_bin[j]) {
-								*vddci = vddci_buf[j * profile->ucElbVDDCI_Num + i];
-								break;
-							}
-						}
-						break;
-					}
-				}
-			}
-			break;
-		default:
-			DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
-			return -EINVAL;
-		}
-		break;
-	default:
-		DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
-		return -EINVAL;
-	}
-
-	return 0;
-}
-
-union get_voltage_info {
-	struct _GET_VOLTAGE_INFO_INPUT_PARAMETER_V1_2 in;
-	struct _GET_EVV_VOLTAGE_INFO_OUTPUT_PARAMETER_V1_2 evv_out;
-};
-
-int amdgpu_atombios_get_voltage_evv(struct amdgpu_device *adev,
-				    u16 virtual_voltage_id,
-				    u16 *voltage)
-{
-	int index = GetIndexIntoMasterTable(COMMAND, GetVoltageInfo);
-	u32 entry_id;
-	u32 count = adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.count;
-	union get_voltage_info args;
-
-	for (entry_id = 0; entry_id < count; entry_id++) {
-		if (adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[entry_id].v ==
-		    virtual_voltage_id)
-			break;
-	}
-
-	if (entry_id >= count)
-		return -EINVAL;
-
-	args.in.ucVoltageType = VOLTAGE_TYPE_VDDC;
-	args.in.ucVoltageMode = ATOM_GET_VOLTAGE_EVV_VOLTAGE;
-	args.in.usVoltageLevel = cpu_to_le16(virtual_voltage_id);
-	args.in.ulSCLKFreq =
-		cpu_to_le32(adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[entry_id].clk);
-
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
-
-	*voltage = le16_to_cpu(args.evv_out.usVoltageLevel);
-
-	return 0;
-}
-
 union voltage_object_info {
 	struct _ATOM_VOLTAGE_OBJECT_INFO v1;
 	struct _ATOM_VOLTAGE_OBJECT_INFO_V2 v2;
@@ -1415,7 +1285,7 @@ static ATOM_VOLTAGE_OBJECT_V3 *amdgpu_atombios_lookup_voltage_object_v3(ATOM_VOL
 {
 	u32 size = le16_to_cpu(v3->sHeader.usStructureSize);
 	u32 offset = offsetof(ATOM_VOLTAGE_OBJECT_INFO_V3_1, asVoltageObj[0]);
-	u8 *start = (u8*)v3;
+	u8 *start = (u8 *)v3;
 
 	while (offset < size) {
 		ATOM_VOLTAGE_OBJECT_V3 *vo = (ATOM_VOLTAGE_OBJECT_V3 *)(start + offset);
@@ -1640,6 +1510,8 @@ int amdgpu_atombios_init_mc_reg_table(struct amdgpu_device *adev,
 										(u32)le32_to_cpu(*((u32 *)reg_data + j));
 									j++;
 								} else if ((reg_table->mc_reg_address[i].pre_reg_data & LOW_NIBBLE_MASK) == DATA_EQU_PREV) {
+									if (i == 0)
+										continue;
 									reg_table->mc_reg_table_entry[num_ranges].mc_data[i] =
 										reg_table->mc_reg_table_entry[num_ranges].mc_data[i - 1];
 								}
@@ -1668,6 +1540,7 @@ int amdgpu_atombios_init_mc_reg_table(struct amdgpu_device *adev,
 	}
 	return -EINVAL;
 }
+#endif
 
 bool amdgpu_atombios_has_gpu_virtualization_table(struct amdgpu_device *adev)
 {
@@ -1699,7 +1572,7 @@ void amdgpu_atombios_scratch_regs_lock(struct amdgpu_device *adev, bool lock)
 	WREG32(adev->bios_scratch_reg_offset + 6, bios_6_scratch);
 }
 
-void amdgpu_atombios_scratch_regs_init(struct amdgpu_device *adev)
+static void amdgpu_atombios_scratch_regs_init(struct amdgpu_device *adev)
 {
 	uint32_t bios_2_scratch, bios_6_scratch;
 
@@ -1721,28 +1594,6 @@ void amdgpu_atombios_scratch_regs_init(struct amdgpu_device *adev)
 	WREG32(adev->bios_scratch_reg_offset + 6, bios_6_scratch);
 }
 
-void amdgpu_atombios_scratch_regs_save(struct amdgpu_device *adev)
-{
-	int i;
-
-	for (i = 0; i < AMDGPU_BIOS_NUM_SCRATCH; i++)
-		adev->bios_scratch[i] = RREG32(adev->bios_scratch_reg_offset + i);
-}
-
-void amdgpu_atombios_scratch_regs_restore(struct amdgpu_device *adev)
-{
-	int i;
-
-	/*
-	 * VBIOS will check ASIC_INIT_COMPLETE bit to decide if
-	 * execute ASIC_Init posting via driver
-	 */
-	adev->bios_scratch[7] &= ~ATOM_S7_ASIC_INIT_COMPLETE_MASK;
-
-	for (i = 0; i < AMDGPU_BIOS_NUM_SCRATCH; i++)
-		WREG32(adev->bios_scratch_reg_offset + i, adev->bios_scratch[i]);
-}
-
 void amdgpu_atombios_scratch_regs_engine_hung(struct amdgpu_device *adev,
 					      bool hung)
 {
@@ -1756,6 +1607,18 @@ void amdgpu_atombios_scratch_regs_engine_hung(struct amdgpu_device *adev,
 	WREG32(adev->bios_scratch_reg_offset + 3, tmp);
 }
 
+void amdgpu_atombios_scratch_regs_set_backlight_level(struct amdgpu_device *adev,
+						      u32 backlight_level)
+{
+	u32 tmp = RREG32(adev->bios_scratch_reg_offset + 2);
+
+	tmp &= ~ATOM_S2_CURRENT_BL_LEVEL_MASK;
+	tmp |= (backlight_level << ATOM_S2_CURRENT_BL_LEVEL_SHIFT) &
+		ATOM_S2_CURRENT_BL_LEVEL_MASK;
+
+	WREG32(adev->bios_scratch_reg_offset + 2, tmp);
+}
+
 bool amdgpu_atombios_scratch_need_asic_init(struct amdgpu_device *adev)
 {
 	u32 tmp = RREG32(adev->bios_scratch_reg_offset + 7);
@@ -1766,47 +1629,47 @@ bool amdgpu_atombios_scratch_need_asic_init(struct amdgpu_device *adev)
 		return true;
 }
 
-/* Atom needs data in little endian format
- * so swap as appropriate when copying data to
- * or from atom. Note that atom operates on
- * dw units.
+/* Atom needs data in little endian format so swap as appropriate when copying
+ * data to or from atom. Note that atom operates on dw units.
+ *
+ * Use to_le=true when sending data to atom and provide at least
+ * ALIGN(num_bytes,4) bytes in the dst buffer.
+ *
+ * Use to_le=false when receiving data from atom and provide ALIGN(num_bytes,4)
+ * byes in the src buffer.
  */
 void amdgpu_atombios_copy_swap(u8 *dst, u8 *src, u8 num_bytes, bool to_le)
 {
 #ifdef __BIG_ENDIAN
-	u8 src_tmp[20], dst_tmp[20]; /* used for byteswapping */
-	u32 *dst32, *src32;
+	u32 src_tmp[5], dst_tmp[5];
 	int i;
+	u8 align_num_bytes = ALIGN(num_bytes, 4);
 
-	memcpy(src_tmp, src, num_bytes);
-	src32 = (u32 *)src_tmp;
-	dst32 = (u32 *)dst_tmp;
 	if (to_le) {
-		for (i = 0; i < ((num_bytes + 3) / 4); i++)
-			dst32[i] = cpu_to_le32(src32[i]);
-		memcpy(dst, dst_tmp, num_bytes);
+		memcpy(src_tmp, src, num_bytes);
+		for (i = 0; i < align_num_bytes / 4; i++)
+			dst_tmp[i] = cpu_to_le32(src_tmp[i]);
+		memcpy(dst, dst_tmp, align_num_bytes);
 	} else {
-		u8 dws = num_bytes & ~3;
-		for (i = 0; i < ((num_bytes + 3) / 4); i++)
-			dst32[i] = le32_to_cpu(src32[i]);
-		memcpy(dst, dst_tmp, dws);
-		if (num_bytes % 4) {
-			for (i = 0; i < (num_bytes % 4); i++)
-				dst[dws+i] = dst_tmp[dws+i];
-		}
+		memcpy(src_tmp, src, align_num_bytes);
+		for (i = 0; i < align_num_bytes / 4; i++)
+			dst_tmp[i] = le32_to_cpu(src_tmp[i]);
+		memcpy(dst, dst_tmp, num_bytes);
 	}
 #else
 	memcpy(dst, src, num_bytes);
 #endif
 }
 
-int amdgpu_atombios_allocate_fb_scratch(struct amdgpu_device *adev)
+static int amdgpu_atombios_allocate_fb_scratch(struct amdgpu_device *adev)
 {
 	struct atom_context *ctx = adev->mode_info.atom_context;
 	int index = GetIndexIntoMasterTable(DATA, VRAM_UsageByFirmware);
 	uint16_t data_offset;
 	int usage_bytes = 0;
 	struct _ATOM_VRAM_USAGE_BY_FIRMWARE *firmware_usage;
+	u64 start_addr;
+	u64 size;
 
 	if (amdgpu_atom_parse_data_header(ctx, index, NULL, NULL, NULL, &data_offset)) {
 		firmware_usage = (struct _ATOM_VRAM_USAGE_BY_FIRMWARE *)(ctx->bios + data_offset);
@@ -1815,7 +1678,21 @@ int amdgpu_atombios_allocate_fb_scratch(struct amdgpu_device *adev)
 			  le32_to_cpu(firmware_usage->asFirmwareVramReserveInfo[0].ulStartAddrUsedByFirmware),
 			  le16_to_cpu(firmware_usage->asFirmwareVramReserveInfo[0].usFirmwareUseInKb));
 
-		usage_bytes = le16_to_cpu(firmware_usage->asFirmwareVramReserveInfo[0].usFirmwareUseInKb) * 1024;
+		start_addr = firmware_usage->asFirmwareVramReserveInfo[0].ulStartAddrUsedByFirmware;
+		size = firmware_usage->asFirmwareVramReserveInfo[0].usFirmwareUseInKb;
+
+		if ((uint32_t)(start_addr & ATOM_VRAM_OPERATION_FLAGS_MASK) ==
+			(uint32_t)(ATOM_VRAM_BLOCK_SRIOV_MSG_SHARE_RESERVATION <<
+			ATOM_VRAM_OPERATION_FLAGS_SHIFT)) {
+			/* Firmware request VRAM reservation for SR-IOV */
+			adev->mman.fw_vram_usage_start_offset = (start_addr &
+				(~ATOM_VRAM_OPERATION_FLAGS_MASK)) << 10;
+			adev->mman.fw_vram_usage_size = size << 10;
+			/* Use the default scratch size */
+			usage_bytes = 0;
+		} else {
+			usage_bytes = le16_to_cpu(firmware_usage->asFirmwareVramReserveInfo[0].usFirmwareUseInKb) * 1024;
+		}
 	}
 	ctx->scratch_size_bytes = 0;
 	if (usage_bytes == 0)
@@ -1827,3 +1704,248 @@ int amdgpu_atombios_allocate_fb_scratch(struct amdgpu_device *adev)
 	ctx->scratch_size_bytes = usage_bytes;
 	return 0;
 }
+
+/* ATOM accessor methods */
+/*
+ * ATOM is an interpreted byte code stored in tables in the vbios.  The
+ * driver registers callbacks to access registers and the interpreter
+ * in the driver parses the tables and executes then to program specific
+ * actions (set display modes, asic init, etc.).  See amdgpu_atombios.c,
+ * atombios.h, and atom.c
+ */
+
+/**
+ * cail_pll_read - read PLL register
+ *
+ * @info: atom card_info pointer
+ * @reg: PLL register offset
+ *
+ * Provides a PLL register accessor for the atom interpreter (r4xx+).
+ * Returns the value of the PLL register.
+ */
+static uint32_t cail_pll_read(struct card_info *info, uint32_t reg)
+{
+	return 0;
+}
+
+/**
+ * cail_pll_write - write PLL register
+ *
+ * @info: atom card_info pointer
+ * @reg: PLL register offset
+ * @val: value to write to the pll register
+ *
+ * Provides a PLL register accessor for the atom interpreter (r4xx+).
+ */
+static void cail_pll_write(struct card_info *info, uint32_t reg, uint32_t val)
+{
+
+}
+
+/**
+ * cail_mc_read - read MC (Memory Controller) register
+ *
+ * @info: atom card_info pointer
+ * @reg: MC register offset
+ *
+ * Provides an MC register accessor for the atom interpreter (r4xx+).
+ * Returns the value of the MC register.
+ */
+static uint32_t cail_mc_read(struct card_info *info, uint32_t reg)
+{
+	return 0;
+}
+
+/**
+ * cail_mc_write - write MC (Memory Controller) register
+ *
+ * @info: atom card_info pointer
+ * @reg: MC register offset
+ * @val: value to write to the pll register
+ *
+ * Provides a MC register accessor for the atom interpreter (r4xx+).
+ */
+static void cail_mc_write(struct card_info *info, uint32_t reg, uint32_t val)
+{
+
+}
+
+/**
+ * cail_reg_write - write MMIO register
+ *
+ * @info: atom card_info pointer
+ * @reg: MMIO register offset
+ * @val: value to write to the pll register
+ *
+ * Provides a MMIO register accessor for the atom interpreter (r4xx+).
+ */
+static void cail_reg_write(struct card_info *info, uint32_t reg, uint32_t val)
+{
+	struct amdgpu_device *adev = drm_to_adev(info->dev);
+
+	WREG32(reg, val);
+}
+
+/**
+ * cail_reg_read - read MMIO register
+ *
+ * @info: atom card_info pointer
+ * @reg: MMIO register offset
+ *
+ * Provides an MMIO register accessor for the atom interpreter (r4xx+).
+ * Returns the value of the MMIO register.
+ */
+static uint32_t cail_reg_read(struct card_info *info, uint32_t reg)
+{
+	struct amdgpu_device *adev = drm_to_adev(info->dev);
+	uint32_t r;
+
+	r = RREG32(reg);
+	return r;
+}
+
+static ssize_t amdgpu_atombios_get_vbios_version(struct device *dev,
+						 struct device_attribute *attr,
+						 char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	struct atom_context *ctx = adev->mode_info.atom_context;
+
+	return sysfs_emit(buf, "%s\n", ctx->vbios_pn);
+}
+
+static ssize_t amdgpu_atombios_get_vbios_build(struct device *dev,
+					       struct device_attribute *attr,
+					       char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	struct atom_context *ctx = adev->mode_info.atom_context;
+
+	return sysfs_emit(buf, "%s\n", ctx->build_num);
+}
+
+static DEVICE_ATTR(vbios_version, 0444, amdgpu_atombios_get_vbios_version,
+		   NULL);
+static DEVICE_ATTR(vbios_build, 0444, amdgpu_atombios_get_vbios_build, NULL);
+
+static struct attribute *amdgpu_vbios_version_attrs[] = {
+	&dev_attr_vbios_version.attr, &dev_attr_vbios_build.attr, NULL
+};
+
+static umode_t amdgpu_vbios_version_attrs_is_visible(struct kobject *kobj,
+						     struct attribute *attr,
+						     int index)
+{
+	struct device *dev = kobj_to_dev(kobj);
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	struct atom_context *ctx = adev->mode_info.atom_context;
+
+	if (attr == &dev_attr_vbios_build.attr && !strlen(ctx->build_num))
+		return 0;
+
+	return attr->mode;
+}
+
+const struct attribute_group amdgpu_vbios_version_attr_group = {
+	.attrs = amdgpu_vbios_version_attrs,
+	.is_visible = amdgpu_vbios_version_attrs_is_visible,
+};
+
+int amdgpu_atombios_sysfs_init(struct amdgpu_device *adev)
+{
+	if (adev->mode_info.atom_context)
+		return devm_device_add_group(adev->dev,
+					     &amdgpu_vbios_version_attr_group);
+
+	return 0;
+}
+
+/**
+ * amdgpu_atombios_fini - free the driver info and callbacks for atombios
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Frees the driver info and register access callbacks for the ATOM
+ * interpreter (r4xx+).
+ * Called at driver shutdown.
+ */
+void amdgpu_atombios_fini(struct amdgpu_device *adev)
+{
+	if (adev->mode_info.atom_context) {
+		kfree(adev->mode_info.atom_context->scratch);
+		kfree(adev->mode_info.atom_context->iio);
+	}
+	kfree(adev->mode_info.atom_context);
+	adev->mode_info.atom_context = NULL;
+	kfree(adev->mode_info.atom_card_info);
+	adev->mode_info.atom_card_info = NULL;
+}
+
+/**
+ * amdgpu_atombios_init - init the driver info and callbacks for atombios
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initializes the driver info and register access callbacks for the
+ * ATOM interpreter (r4xx+).
+ * Returns 0 on sucess, -ENOMEM on failure.
+ * Called at driver startup.
+ */
+int amdgpu_atombios_init(struct amdgpu_device *adev)
+{
+	struct card_info *atom_card_info =
+	    kzalloc(sizeof(struct card_info), GFP_KERNEL);
+
+	if (!atom_card_info)
+		return -ENOMEM;
+
+	adev->mode_info.atom_card_info = atom_card_info;
+	atom_card_info->dev = adev_to_drm(adev);
+	atom_card_info->reg_read = cail_reg_read;
+	atom_card_info->reg_write = cail_reg_write;
+	atom_card_info->mc_read = cail_mc_read;
+	atom_card_info->mc_write = cail_mc_write;
+	atom_card_info->pll_read = cail_pll_read;
+	atom_card_info->pll_write = cail_pll_write;
+
+	adev->mode_info.atom_context = amdgpu_atom_parse(atom_card_info, adev->bios);
+	if (!adev->mode_info.atom_context) {
+		amdgpu_atombios_fini(adev);
+		return -ENOMEM;
+	}
+
+	mutex_init(&adev->mode_info.atom_context->mutex);
+	if (adev->is_atom_fw) {
+		amdgpu_atomfirmware_scratch_regs_init(adev);
+		amdgpu_atomfirmware_allocate_fb_scratch(adev);
+		/* cached firmware_flags for further usage */
+		adev->mode_info.firmware_flags =
+			amdgpu_atomfirmware_query_firmware_capability(adev);
+	} else {
+		amdgpu_atombios_scratch_regs_init(adev);
+		amdgpu_atombios_allocate_fb_scratch(adev);
+	}
+
+	return 0;
+}
+
+int amdgpu_atombios_get_data_table(struct amdgpu_device *adev,
+				   uint32_t table,
+				   uint16_t *size,
+				   uint8_t *frev,
+				   uint8_t *crev,
+				   uint8_t **addr)
+{
+	uint16_t data_start;
+
+	if (!amdgpu_atom_parse_data_header(adev->mode_info.atom_context, table,
+					   size, frev, crev, &data_start))
+		return -EINVAL;
+
+	*addr = (uint8_t *)adev->mode_info.atom_context->bios + data_start;
+
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h
index b0d5d1d7fdba..867bc5c5ce67 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h
@@ -89,8 +89,7 @@ struct atom_memory_info {
 
 #define MAX_AC_TIMING_ENTRIES 16
 
-struct atom_memory_clock_range_table
-{
+struct atom_memory_clock_range_table {
 	u8 num_entries;
 	u8 rsv[3];
 	u32 mclk[MAX_AC_TIMING_ENTRIES];
@@ -118,14 +117,12 @@ struct atom_mc_reg_table {
 
 #define MAX_VOLTAGE_ENTRIES 32
 
-struct atom_voltage_table_entry
-{
+struct atom_voltage_table_entry {
 	u16 value;
 	u32 smio_low;
 };
 
-struct atom_voltage_table
-{
+struct atom_voltage_table {
 	u32 count;
 	u32 mask_low;
 	u32 phase_delay;
@@ -139,6 +136,7 @@ amdgpu_atombios_lookup_gpio(struct amdgpu_device *adev,
 struct amdgpu_i2c_bus_rec amdgpu_atombios_lookup_i2c_gpio(struct amdgpu_device *adev,
 							  uint8_t id);
 void amdgpu_atombios_i2c_init(struct amdgpu_device *adev);
+void amdgpu_atombios_oem_i2c_init(struct amdgpu_device *adev, u8 i2c_id);
 
 bool amdgpu_atombios_has_dce_engine_info(struct amdgpu_device *adev);
 
@@ -160,25 +158,14 @@ int amdgpu_atombios_get_clock_dividers(struct amdgpu_device *adev,
 				       bool strobe_mode,
 				       struct atom_clock_dividers *dividers);
 
+#ifdef CONFIG_DRM_AMDGPU_SI
 int amdgpu_atombios_get_memory_pll_dividers(struct amdgpu_device *adev,
 					    u32 clock,
 					    bool strobe_mode,
 					    struct atom_mpll_param *mpll_param);
 
-void amdgpu_atombios_set_engine_dram_timings(struct amdgpu_device *adev,
-					     u32 eng_clock, u32 mem_clock);
-
-int amdgpu_atombios_get_leakage_id_from_vbios(struct amdgpu_device *adev,
-					      u16 *leakage_id);
-
-int amdgpu_atombios_get_leakage_vddc_based_on_leakage_params(struct amdgpu_device *adev,
-							     u16 *vddc, u16 *vddci,
-							     u16 virtual_voltage_id,
-							     u16 vbios_voltage_id);
-
-int amdgpu_atombios_get_voltage_evv(struct amdgpu_device *adev,
-				    u16 virtual_voltage_id,
-				    u16 *voltage);
+int amdgpu_atombios_set_engine_dram_timings(struct amdgpu_device *adev,
+					    u32 eng_clock, u32 mem_clock);
 
 bool
 amdgpu_atombios_is_voltage_gpio(struct amdgpu_device *adev,
@@ -191,34 +178,43 @@ int amdgpu_atombios_get_voltage_table(struct amdgpu_device *adev,
 int amdgpu_atombios_init_mc_reg_table(struct amdgpu_device *adev,
 				      u8 module_index,
 				      struct atom_mc_reg_table *reg_table);
+int amdgpu_atombios_get_max_vddc(struct amdgpu_device *adev, u8 voltage_type,
+			     u16 voltage_id, u16 *voltage);
+int amdgpu_atombios_get_leakage_vddc_based_on_leakage_idx(struct amdgpu_device *adev,
+						      u16 *voltage,
+						      u16 leakage_idx);
+void amdgpu_atombios_get_default_voltages(struct amdgpu_device *adev,
+					  u16 *vddc, u16 *vddci, u16 *mvdd);
+int amdgpu_atombios_get_svi2_info(struct amdgpu_device *adev,
+			      u8 voltage_type,
+			      u8 *svd_gpio_id, u8 *svc_gpio_id);
+#endif
 
 bool amdgpu_atombios_has_gpu_virtualization_table(struct amdgpu_device *adev);
 
 void amdgpu_atombios_scratch_regs_lock(struct amdgpu_device *adev, bool lock);
-void amdgpu_atombios_scratch_regs_init(struct amdgpu_device *adev);
-void amdgpu_atombios_scratch_regs_save(struct amdgpu_device *adev);
-void amdgpu_atombios_scratch_regs_restore(struct amdgpu_device *adev);
 void amdgpu_atombios_scratch_regs_engine_hung(struct amdgpu_device *adev,
 					      bool hung);
+void amdgpu_atombios_scratch_regs_set_backlight_level(struct amdgpu_device *adev,
+						      u32 backlight_level);
 bool amdgpu_atombios_scratch_need_asic_init(struct amdgpu_device *adev);
 
 void amdgpu_atombios_copy_swap(u8 *dst, u8 *src, u8 num_bytes, bool to_le);
-int amdgpu_atombios_get_max_vddc(struct amdgpu_device *adev, u8 voltage_type,
-			     u16 voltage_id, u16 *voltage);
-int amdgpu_atombios_get_leakage_vddc_based_on_leakage_idx(struct amdgpu_device *adev,
-						      u16 *voltage,
-						      u16 leakage_idx);
-void amdgpu_atombios_get_default_voltages(struct amdgpu_device *adev,
-					  u16 *vddc, u16 *vddci, u16 *mvdd);
 int amdgpu_atombios_get_clock_dividers(struct amdgpu_device *adev,
 				       u8 clock_type,
 				       u32 clock,
 				       bool strobe_mode,
 				       struct atom_clock_dividers *dividers);
-int amdgpu_atombios_get_svi2_info(struct amdgpu_device *adev,
-			      u8 voltage_type,
-			      u8 *svd_gpio_id, u8 *svc_gpio_id);
 
-int amdgpu_atombios_allocate_fb_scratch(struct amdgpu_device *adev);
+int amdgpu_atombios_get_data_table(struct amdgpu_device *adev,
+				   uint32_t table,
+				   uint16_t *size,
+				   uint8_t *frev,
+				   uint8_t *crev,
+				   uint8_t **addr);
+
+void amdgpu_atombios_fini(struct amdgpu_device *adev);
+int amdgpu_atombios_init(struct amdgpu_device *adev);
+int amdgpu_atombios_sysfs_init(struct amdgpu_device *adev);
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c
index f9ffe8ef0cd6..c7d32fb216e4 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c
@@ -20,33 +20,69 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  *
  */
-#include <drm/drmP.h>
+
 #include <drm/amdgpu_drm.h>
 #include "amdgpu.h"
 #include "atomfirmware.h"
 #include "amdgpu_atomfirmware.h"
 #include "atom.h"
 #include "atombios.h"
+#include "soc15_hw_ip.h"
 
-#define get_index_into_master_table(master_table, table_name) (offsetof(struct master_table, table_name) / sizeof(uint16_t))
+union firmware_info {
+	struct atom_firmware_info_v3_1 v31;
+	struct atom_firmware_info_v3_2 v32;
+	struct atom_firmware_info_v3_3 v33;
+	struct atom_firmware_info_v3_4 v34;
+	struct atom_firmware_info_v3_5 v35;
+};
 
-bool amdgpu_atomfirmware_gpu_supports_virtualization(struct amdgpu_device *adev)
+/*
+ * Helper function to query firmware capability
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Return firmware_capability in firmwareinfo table on success or 0 if not
+ */
+uint32_t amdgpu_atomfirmware_query_firmware_capability(struct amdgpu_device *adev)
 {
-	int index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
-						firmwareinfo);
-	uint16_t data_offset;
-
-	if (amdgpu_atom_parse_data_header(adev->mode_info.atom_context, index, NULL,
-					  NULL, NULL, &data_offset)) {
-		struct atom_firmware_info_v3_1 *firmware_info =
-			(struct atom_firmware_info_v3_1 *)(adev->mode_info.atom_context->bios +
-							   data_offset);
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index;
+	u16 data_offset, size;
+	union firmware_info *firmware_info;
+	u8 frev, crev;
+	u32 fw_cap = 0;
 
-		if (le32_to_cpu(firmware_info->firmware_capability) &
-		    ATOM_FIRMWARE_CAP_GPU_VIRTUALIZATION)
-			return true;
+	index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
+			firmwareinfo);
+
+	if (amdgpu_atom_parse_data_header(adev->mode_info.atom_context,
+				index, &size, &frev, &crev, &data_offset)) {
+		/* support firmware_info 3.1 + */
+		if ((frev == 3 && crev >= 1) || (frev > 3)) {
+			firmware_info = (union firmware_info *)
+				(mode_info->atom_context->bios + data_offset);
+			fw_cap = le32_to_cpu(firmware_info->v31.firmware_capability);
+		}
 	}
-	return false;
+
+	return fw_cap;
+}
+
+/*
+ * Helper function to query gpu virtualizaiton capability
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Return true if gpu virtualization is supported or false if not
+ */
+bool amdgpu_atomfirmware_gpu_virtualization_supported(struct amdgpu_device *adev)
+{
+	u32 fw_cap;
+
+	fw_cap = adev->mode_info.firmware_flags;
+
+	return (fw_cap & ATOM_FIRMWARE_CAP_GPU_VIRTUALIZATION) ? true : false;
 }
 
 void amdgpu_atomfirmware_scratch_regs_init(struct amdgpu_device *adev)
@@ -66,25 +102,101 @@ void amdgpu_atomfirmware_scratch_regs_init(struct amdgpu_device *adev)
 	}
 }
 
+static int amdgpu_atomfirmware_allocate_fb_v2_1(struct amdgpu_device *adev,
+	struct vram_usagebyfirmware_v2_1 *fw_usage, int *usage_bytes)
+{
+	u32 start_addr, fw_size, drv_size;
+
+	start_addr = le32_to_cpu(fw_usage->start_address_in_kb);
+	fw_size = le16_to_cpu(fw_usage->used_by_firmware_in_kb);
+	drv_size = le16_to_cpu(fw_usage->used_by_driver_in_kb);
+
+	DRM_DEBUG("atom firmware v2_1 requested %08x %dkb fw %dkb drv\n",
+			  start_addr,
+			  fw_size,
+			  drv_size);
+
+	if ((start_addr & ATOM_VRAM_OPERATION_FLAGS_MASK) ==
+		(u32)(ATOM_VRAM_BLOCK_SRIOV_MSG_SHARE_RESERVATION <<
+		ATOM_VRAM_OPERATION_FLAGS_SHIFT)) {
+		/* Firmware request VRAM reservation for SR-IOV */
+		adev->mman.fw_vram_usage_start_offset = (start_addr &
+			(~ATOM_VRAM_OPERATION_FLAGS_MASK)) << 10;
+		adev->mman.fw_vram_usage_size = fw_size << 10;
+		/* Use the default scratch size */
+		*usage_bytes = 0;
+	} else {
+		*usage_bytes = drv_size << 10;
+	}
+	return 0;
+}
+
+static int amdgpu_atomfirmware_allocate_fb_v2_2(struct amdgpu_device *adev,
+		struct vram_usagebyfirmware_v2_2 *fw_usage, int *usage_bytes)
+{
+	u32 fw_start_addr, fw_size, drv_start_addr, drv_size;
+
+	fw_start_addr = le32_to_cpu(fw_usage->fw_region_start_address_in_kb);
+	fw_size = le16_to_cpu(fw_usage->used_by_firmware_in_kb);
+
+	drv_start_addr = le32_to_cpu(fw_usage->driver_region0_start_address_in_kb);
+	drv_size = le32_to_cpu(fw_usage->used_by_driver_region0_in_kb);
+
+	DRM_DEBUG("atom requested fw start at %08x %dkb and drv start at %08x %dkb\n",
+			  fw_start_addr,
+			  fw_size,
+			  drv_start_addr,
+			  drv_size);
+
+	if (amdgpu_sriov_vf(adev) &&
+	    ((fw_start_addr & (ATOM_VRAM_BLOCK_NEEDS_NO_RESERVATION <<
+		ATOM_VRAM_OPERATION_FLAGS_SHIFT)) == 0)) {
+		/* Firmware request VRAM reservation for SR-IOV */
+		adev->mman.fw_vram_usage_start_offset = (fw_start_addr &
+			(~ATOM_VRAM_OPERATION_FLAGS_MASK)) << 10;
+		adev->mman.fw_vram_usage_size = fw_size << 10;
+	}
+
+	if (amdgpu_sriov_vf(adev) &&
+	    ((drv_start_addr & (ATOM_VRAM_BLOCK_NEEDS_NO_RESERVATION <<
+		ATOM_VRAM_OPERATION_FLAGS_SHIFT)) == 0)) {
+		/* driver request VRAM reservation for SR-IOV */
+		adev->mman.drv_vram_usage_start_offset = (drv_start_addr &
+			(~ATOM_VRAM_OPERATION_FLAGS_MASK)) << 10;
+		adev->mman.drv_vram_usage_size = drv_size << 10;
+	}
+
+	*usage_bytes = 0;
+	return 0;
+}
+
 int amdgpu_atomfirmware_allocate_fb_scratch(struct amdgpu_device *adev)
 {
 	struct atom_context *ctx = adev->mode_info.atom_context;
 	int index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
 						vram_usagebyfirmware);
-	uint16_t data_offset;
+	struct vram_usagebyfirmware_v2_1 *fw_usage_v2_1;
+	struct vram_usagebyfirmware_v2_2 *fw_usage_v2_2;
+	u16 data_offset;
+	u8 frev, crev;
 	int usage_bytes = 0;
 
-	if (amdgpu_atom_parse_data_header(ctx, index, NULL, NULL, NULL, &data_offset)) {
-		struct vram_usagebyfirmware_v2_1 *firmware_usage =
-			(struct vram_usagebyfirmware_v2_1 *)(ctx->bios + data_offset);
-
-		DRM_DEBUG("atom firmware requested %08x %dkb fw %dkb drv\n",
-			  le32_to_cpu(firmware_usage->start_address_in_kb),
-			  le16_to_cpu(firmware_usage->used_by_firmware_in_kb),
-			  le16_to_cpu(firmware_usage->used_by_driver_in_kb));
-
-		usage_bytes = le16_to_cpu(firmware_usage->used_by_driver_in_kb) * 1024;
+	if (amdgpu_atom_parse_data_header(ctx, index, NULL, &frev, &crev, &data_offset)) {
+		if (frev == 2 && crev == 1) {
+			fw_usage_v2_1 =
+				(struct vram_usagebyfirmware_v2_1 *)(ctx->bios + data_offset);
+			amdgpu_atomfirmware_allocate_fb_v2_1(adev,
+					fw_usage_v2_1,
+					&usage_bytes);
+		} else if (frev >= 2 && crev >= 2) {
+			fw_usage_v2_2 =
+				(struct vram_usagebyfirmware_v2_2 *)(ctx->bios + data_offset);
+			amdgpu_atomfirmware_allocate_fb_v2_2(adev,
+					fw_usage_v2_2,
+					&usage_bytes);
+		}
 	}
+
 	ctx->scratch_size_bytes = 0;
 	if (usage_bytes == 0)
 		usage_bytes = 20 * 1024;
@@ -98,47 +210,505 @@ int amdgpu_atomfirmware_allocate_fb_scratch(struct amdgpu_device *adev)
 
 union igp_info {
 	struct atom_integrated_system_info_v1_11 v11;
+	struct atom_integrated_system_info_v1_12 v12;
+	struct atom_integrated_system_info_v2_1 v21;
+	struct atom_integrated_system_info_v2_3 v23;
 };
 
-/*
- * Return vram width from integrated system info table, if available,
- * or 0 if not.
- */
-int amdgpu_atomfirmware_get_vram_width(struct amdgpu_device *adev)
+union umc_info {
+	struct atom_umc_info_v3_1 v31;
+	struct atom_umc_info_v3_2 v32;
+	struct atom_umc_info_v3_3 v33;
+	struct atom_umc_info_v4_0 v40;
+};
+
+union vram_info {
+	struct atom_vram_info_header_v2_3 v23;
+	struct atom_vram_info_header_v2_4 v24;
+	struct atom_vram_info_header_v2_5 v25;
+	struct atom_vram_info_header_v2_6 v26;
+	struct atom_vram_info_header_v3_0 v30;
+};
+
+union vram_module {
+	struct atom_vram_module_v9 v9;
+	struct atom_vram_module_v10 v10;
+	struct atom_vram_module_v11 v11;
+	struct atom_vram_module_v3_0 v30;
+};
+
+static int convert_atom_mem_type_to_vram_type(struct amdgpu_device *adev,
+					      int atom_mem_type)
+{
+	int vram_type;
+
+	if (adev->flags & AMD_IS_APU) {
+		switch (atom_mem_type) {
+		case Ddr2MemType:
+		case LpDdr2MemType:
+			vram_type = AMDGPU_VRAM_TYPE_DDR2;
+			break;
+		case Ddr3MemType:
+		case LpDdr3MemType:
+			vram_type = AMDGPU_VRAM_TYPE_DDR3;
+			break;
+		case Ddr4MemType:
+			vram_type = AMDGPU_VRAM_TYPE_DDR4;
+			break;
+		case LpDdr4MemType:
+			vram_type = AMDGPU_VRAM_TYPE_LPDDR4;
+			break;
+		case Ddr5MemType:
+			vram_type = AMDGPU_VRAM_TYPE_DDR5;
+			break;
+		case LpDdr5MemType:
+			vram_type = AMDGPU_VRAM_TYPE_LPDDR5;
+			break;
+		default:
+			vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
+			break;
+		}
+	} else {
+		switch (atom_mem_type) {
+		case ATOM_DGPU_VRAM_TYPE_GDDR5:
+			vram_type = AMDGPU_VRAM_TYPE_GDDR5;
+			break;
+		case ATOM_DGPU_VRAM_TYPE_HBM2:
+		case ATOM_DGPU_VRAM_TYPE_HBM2E:
+		case ATOM_DGPU_VRAM_TYPE_HBM3:
+			vram_type = AMDGPU_VRAM_TYPE_HBM;
+			break;
+		case ATOM_DGPU_VRAM_TYPE_GDDR6:
+			vram_type = AMDGPU_VRAM_TYPE_GDDR6;
+			break;
+		case ATOM_DGPU_VRAM_TYPE_HBM3E:
+			vram_type = AMDGPU_VRAM_TYPE_HBM3E;
+			break;
+		default:
+			vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
+			break;
+		}
+	}
+
+	return vram_type;
+}
+
+int
+amdgpu_atomfirmware_get_vram_info(struct amdgpu_device *adev,
+				  int *vram_width, int *vram_type,
+				  int *vram_vendor)
 {
 	struct amdgpu_mode_info *mode_info = &adev->mode_info;
-	int index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
-						integratedsysteminfo);
+	int index, i = 0;
 	u16 data_offset, size;
 	union igp_info *igp_info;
+	union vram_info *vram_info;
+	union umc_info *umc_info;
+	union vram_module *vram_module;
 	u8 frev, crev;
-
-	/* get any igp specific overrides */
-	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, &size,
-				   &frev, &crev, &data_offset)) {
-		igp_info = (union igp_info *)
-			(mode_info->atom_context->bios + data_offset);
-		switch (crev) {
-		case 11:
-			return igp_info->v11.umachannelnumber * 64;
+	u8 mem_type;
+	u8 mem_vendor;
+	u32 mem_channel_number;
+	u32 mem_channel_width;
+	u32 module_id;
+
+	if (adev->flags & AMD_IS_APU)
+		index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
+						    integratedsysteminfo);
+	else {
+		switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+		case IP_VERSION(12, 0, 0):
+		case IP_VERSION(12, 0, 1):
+			index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1, umc_info);
+			break;
 		default:
-			return 0;
+			index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1, vram_info);
+		}
+	}
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context,
+					  index, &size,
+					  &frev, &crev, &data_offset)) {
+		if (adev->flags & AMD_IS_APU) {
+			igp_info = (union igp_info *)
+				(mode_info->atom_context->bios + data_offset);
+			switch (frev) {
+			case 1:
+				switch (crev) {
+				case 11:
+				case 12:
+					mem_channel_number = igp_info->v11.umachannelnumber;
+					if (!mem_channel_number)
+						mem_channel_number = 1;
+					mem_type = igp_info->v11.memorytype;
+					if (mem_type == LpDdr5MemType)
+						mem_channel_width = 32;
+					else
+						mem_channel_width = 64;
+					if (vram_width)
+						*vram_width = mem_channel_number * mem_channel_width;
+					if (vram_type)
+						*vram_type = convert_atom_mem_type_to_vram_type(adev, mem_type);
+					break;
+				default:
+					return -EINVAL;
+				}
+				break;
+			case 2:
+				switch (crev) {
+				case 1:
+				case 2:
+					mem_channel_number = igp_info->v21.umachannelnumber;
+					if (!mem_channel_number)
+						mem_channel_number = 1;
+					mem_type = igp_info->v21.memorytype;
+					if (mem_type == LpDdr5MemType)
+						mem_channel_width = 32;
+					else
+						mem_channel_width = 64;
+					if (vram_width)
+						*vram_width = mem_channel_number * mem_channel_width;
+					if (vram_type)
+						*vram_type = convert_atom_mem_type_to_vram_type(adev, mem_type);
+					break;
+				case 3:
+					mem_channel_number = igp_info->v23.umachannelnumber;
+					if (!mem_channel_number)
+						mem_channel_number = 1;
+					mem_type = igp_info->v23.memorytype;
+					if (mem_type == LpDdr5MemType)
+						mem_channel_width = 32;
+					else
+						mem_channel_width = 64;
+					if (vram_width)
+						*vram_width = mem_channel_number * mem_channel_width;
+					if (vram_type)
+						*vram_type = convert_atom_mem_type_to_vram_type(adev, mem_type);
+					break;
+				default:
+					return -EINVAL;
+				}
+				break;
+			default:
+				return -EINVAL;
+			}
+		} else {
+			switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+			case IP_VERSION(12, 0, 0):
+			case IP_VERSION(12, 0, 1):
+				umc_info = (union umc_info *)(mode_info->atom_context->bios + data_offset);
+
+				if (frev == 4) {
+					switch (crev) {
+					case 0:
+						mem_channel_number = le32_to_cpu(umc_info->v40.channel_num);
+						mem_type = le32_to_cpu(umc_info->v40.vram_type);
+						mem_channel_width = le32_to_cpu(umc_info->v40.channel_width);
+						mem_vendor = RREG32(adev->bios_scratch_reg_offset + 4) & 0xF;
+						if (vram_vendor)
+							*vram_vendor = mem_vendor;
+						if (vram_type)
+							*vram_type = convert_atom_mem_type_to_vram_type(adev, mem_type);
+						if (vram_width)
+							*vram_width = mem_channel_number * (1 << mem_channel_width);
+						break;
+					default:
+						return -EINVAL;
+					}
+				} else
+					return -EINVAL;
+				break;
+			default:
+				vram_info = (union vram_info *)
+					(mode_info->atom_context->bios + data_offset);
+
+				module_id = (RREG32(adev->bios_scratch_reg_offset + 4) & 0x00ff0000) >> 16;
+				if (frev == 3) {
+					switch (crev) {
+					/* v30 */
+					case 0:
+						vram_module = (union vram_module *)vram_info->v30.vram_module;
+						mem_vendor = (vram_module->v30.dram_vendor_id) & 0xF;
+						if (vram_vendor)
+							*vram_vendor = mem_vendor;
+						mem_type = vram_info->v30.memory_type;
+						if (vram_type)
+							*vram_type = convert_atom_mem_type_to_vram_type(adev, mem_type);
+						mem_channel_number = vram_info->v30.channel_num;
+						mem_channel_width = vram_info->v30.channel_width;
+						if (vram_width)
+							*vram_width = mem_channel_number * 16;
+						break;
+					default:
+						return -EINVAL;
+					}
+				} else if (frev == 2) {
+					switch (crev) {
+					/* v23 */
+					case 3:
+						if (module_id > vram_info->v23.vram_module_num)
+							module_id = 0;
+						vram_module = (union vram_module *)vram_info->v23.vram_module;
+						while (i < module_id) {
+							vram_module = (union vram_module *)
+								((u8 *)vram_module + vram_module->v9.vram_module_size);
+							i++;
+						}
+						mem_type = vram_module->v9.memory_type;
+						if (vram_type)
+							*vram_type = convert_atom_mem_type_to_vram_type(adev, mem_type);
+						mem_channel_number = vram_module->v9.channel_num;
+						mem_channel_width = vram_module->v9.channel_width;
+						if (vram_width)
+							*vram_width = mem_channel_number * (1 << mem_channel_width);
+						mem_vendor = (vram_module->v9.vender_rev_id) & 0xF;
+						if (vram_vendor)
+							*vram_vendor = mem_vendor;
+						break;
+					/* v24 */
+					case 4:
+						if (module_id > vram_info->v24.vram_module_num)
+							module_id = 0;
+						vram_module = (union vram_module *)vram_info->v24.vram_module;
+						while (i < module_id) {
+							vram_module = (union vram_module *)
+								((u8 *)vram_module + vram_module->v10.vram_module_size);
+							i++;
+						}
+						mem_type = vram_module->v10.memory_type;
+						if (vram_type)
+							*vram_type = convert_atom_mem_type_to_vram_type(adev, mem_type);
+						mem_channel_number = vram_module->v10.channel_num;
+						mem_channel_width = vram_module->v10.channel_width;
+						if (vram_width)
+							*vram_width = mem_channel_number * (1 << mem_channel_width);
+						mem_vendor = (vram_module->v10.vender_rev_id) & 0xF;
+						if (vram_vendor)
+							*vram_vendor = mem_vendor;
+						break;
+					/* v25 */
+					case 5:
+						if (module_id > vram_info->v25.vram_module_num)
+							module_id = 0;
+						vram_module = (union vram_module *)vram_info->v25.vram_module;
+						while (i < module_id) {
+							vram_module = (union vram_module *)
+								((u8 *)vram_module + vram_module->v11.vram_module_size);
+							i++;
+						}
+						mem_type = vram_module->v11.memory_type;
+						if (vram_type)
+							*vram_type = convert_atom_mem_type_to_vram_type(adev, mem_type);
+						mem_channel_number = vram_module->v11.channel_num;
+						mem_channel_width = vram_module->v11.channel_width;
+						if (vram_width)
+							*vram_width = mem_channel_number * (1 << mem_channel_width);
+						mem_vendor = (vram_module->v11.vender_rev_id) & 0xF;
+						if (vram_vendor)
+							*vram_vendor = mem_vendor;
+						break;
+					/* v26 */
+					case 6:
+						if (module_id > vram_info->v26.vram_module_num)
+							module_id = 0;
+						vram_module = (union vram_module *)vram_info->v26.vram_module;
+						while (i < module_id) {
+							vram_module = (union vram_module *)
+								((u8 *)vram_module + vram_module->v9.vram_module_size);
+							i++;
+						}
+						mem_type = vram_module->v9.memory_type;
+						if (vram_type)
+							*vram_type = convert_atom_mem_type_to_vram_type(adev, mem_type);
+						mem_channel_number = vram_module->v9.channel_num;
+						mem_channel_width = vram_module->v9.channel_width;
+						if (vram_width)
+							*vram_width = mem_channel_number * (1 << mem_channel_width);
+						mem_vendor = (vram_module->v9.vender_rev_id) & 0xF;
+						if (vram_vendor)
+							*vram_vendor = mem_vendor;
+						break;
+					default:
+						return -EINVAL;
+					}
+				} else {
+					/* invalid frev */
+					return -EINVAL;
+				}
+			}
 		}
 	}
 
 	return 0;
 }
 
-union firmware_info {
-	struct atom_firmware_info_v3_1 v31;
-};
+/*
+ * Return true if vbios enabled ecc by default, if umc info table is available
+ * or false if ecc is not enabled or umc info table is not available
+ */
+bool amdgpu_atomfirmware_mem_ecc_supported(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index;
+	u16 data_offset, size;
+	union umc_info *umc_info;
+	u8 frev, crev;
+	bool mem_ecc_enabled = false;
+	u8 umc_config;
+	u32 umc_config1;
+	adev->ras_default_ecc_enabled = false;
+
+	index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
+			umc_info);
+
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context,
+				index, &size, &frev, &crev, &data_offset)) {
+		umc_info = (union umc_info *)(mode_info->atom_context->bios + data_offset);
+		if (frev == 3) {
+			switch (crev) {
+			case 1:
+				umc_config = le32_to_cpu(umc_info->v31.umc_config);
+				mem_ecc_enabled =
+					(umc_config & UMC_CONFIG__DEFAULT_MEM_ECC_ENABLE) ? true : false;
+				break;
+			case 2:
+				umc_config = le32_to_cpu(umc_info->v32.umc_config);
+				mem_ecc_enabled =
+					(umc_config & UMC_CONFIG__DEFAULT_MEM_ECC_ENABLE) ? true : false;
+				break;
+			case 3:
+				umc_config = le32_to_cpu(umc_info->v33.umc_config);
+				umc_config1 = le32_to_cpu(umc_info->v33.umc_config1);
+				mem_ecc_enabled =
+					((umc_config & UMC_CONFIG__DEFAULT_MEM_ECC_ENABLE) ||
+					 (umc_config1 & UMC_CONFIG1__ENABLE_ECC_CAPABLE)) ? true : false;
+				adev->ras_default_ecc_enabled =
+					(umc_config & UMC_CONFIG__DEFAULT_MEM_ECC_ENABLE) ? true : false;
+				break;
+			default:
+				/* unsupported crev */
+				return false;
+			}
+		} else if (frev == 4) {
+			switch (crev) {
+			case 0:
+				umc_config = le32_to_cpu(umc_info->v40.umc_config);
+				umc_config1 = le32_to_cpu(umc_info->v40.umc_config1);
+				mem_ecc_enabled =
+					(umc_config1 & UMC_CONFIG1__ENABLE_ECC_CAPABLE) ? true : false;
+				adev->ras_default_ecc_enabled =
+					(umc_config & UMC_CONFIG__DEFAULT_MEM_ECC_ENABLE) ? true : false;
+				break;
+			default:
+				/* unsupported crev */
+				return false;
+			}
+		} else {
+			/* unsupported frev */
+			return false;
+		}
+	}
+
+	return mem_ecc_enabled;
+}
+
+/*
+ * Helper function to query sram ecc capablity
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Return true if vbios supports sram ecc or false if not
+ */
+bool amdgpu_atomfirmware_sram_ecc_supported(struct amdgpu_device *adev)
+{
+	u32 fw_cap;
+
+	fw_cap = adev->mode_info.firmware_flags;
+
+	return (fw_cap & ATOM_FIRMWARE_CAP_SRAM_ECC) ? true : false;
+}
+
+/*
+ * Helper function to query dynamic boot config capability
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Return true if vbios supports dynamic boot config or false if not
+ */
+bool amdgpu_atomfirmware_dynamic_boot_config_supported(struct amdgpu_device *adev)
+{
+	u32 fw_cap;
+
+	fw_cap = adev->mode_info.firmware_flags;
+
+	return (fw_cap & ATOM_FIRMWARE_CAP_DYNAMIC_BOOT_CFG_ENABLE) ? true : false;
+}
+
+/**
+ * amdgpu_atomfirmware_ras_rom_addr -- Get the RAS EEPROM addr from VBIOS
+ * @adev: amdgpu_device pointer
+ * @i2c_address: pointer to u8; if not NULL, will contain
+ *    the RAS EEPROM address if the function returns true
+ *
+ * Return true if VBIOS supports RAS EEPROM address reporting,
+ * else return false. If true and @i2c_address is not NULL,
+ * will contain the RAS ROM address.
+ */
+bool amdgpu_atomfirmware_ras_rom_addr(struct amdgpu_device *adev,
+				      u8 *i2c_address)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index;
+	u16 data_offset, size;
+	union firmware_info *firmware_info;
+	u8 frev, crev;
+
+	index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
+					    firmwareinfo);
+
+	if (amdgpu_atom_parse_data_header(adev->mode_info.atom_context,
+					  index, &size, &frev, &crev,
+					  &data_offset)) {
+		/* support firmware_info 3.4 + */
+		if ((frev == 3 && crev >= 4) || (frev > 3)) {
+			firmware_info = (union firmware_info *)
+				(mode_info->atom_context->bios + data_offset);
+			/* The ras_rom_i2c_slave_addr should ideally
+			 * be a 19-bit EEPROM address, which would be
+			 * used as is by the driver; see top of
+			 * amdgpu_eeprom.c.
+			 *
+			 * When this is the case, 0 is of course a
+			 * valid RAS EEPROM address, in which case,
+			 * we'll drop the first "if (firm...)" and only
+			 * leave the check for the pointer.
+			 *
+			 * The reason this works right now is because
+			 * ras_rom_i2c_slave_addr contains the EEPROM
+			 * device type qualifier 1010b in the top 4
+			 * bits.
+			 */
+			if (firmware_info->v34.ras_rom_i2c_slave_addr) {
+				if (i2c_address)
+					*i2c_address = firmware_info->v34.ras_rom_i2c_slave_addr;
+				return true;
+			}
+		}
+	}
+
+	return false;
+}
+
 
 union smu_info {
 	struct atom_smu_info_v3_1 v31;
+	struct atom_smu_info_v4_0 v40;
 };
 
-union umc_info {
-	struct atom_umc_info_v3_1 v31;
+union gfx_info {
+	struct atom_gfx_info_v2_2 v22;
+	struct atom_gfx_info_v2_4 v24;
+	struct atom_gfx_info_v2_7 v27;
+	struct atom_gfx_info_v3_0 v30;
 };
 
 int amdgpu_atomfirmware_get_clock_info(struct amdgpu_device *adev)
@@ -166,10 +736,6 @@ int amdgpu_atomfirmware_get_clock_info(struct amdgpu_device *adev)
 		adev->pm.current_sclk = adev->clock.default_sclk;
 		adev->pm.current_mclk = adev->clock.default_mclk;
 
-		/* not technically a clock, but... */
-		adev->mode_info.firmware_flags =
-			le32_to_cpu(firmware_info->v31.firmware_capability);
-
 		ret = 0;
 	}
 
@@ -182,7 +748,10 @@ int amdgpu_atomfirmware_get_clock_info(struct amdgpu_device *adev)
 					   data_offset);
 
 		/* system clock */
-		spll->reference_freq = le32_to_cpu(smu_info->v31.core_refclk_10khz);
+		if (frev == 3)
+			spll->reference_freq = le32_to_cpu(smu_info->v31.core_refclk_10khz);
+		else if (frev == 4)
+			spll->reference_freq = le32_to_cpu(smu_info->v40.core_refclk_10khz);
 
 		spll->reference_div = 0;
 		spll->min_post_div = 1;
@@ -219,5 +788,221 @@ int amdgpu_atomfirmware_get_clock_info(struct amdgpu_device *adev)
 		ret = 0;
 	}
 
+	/* if asic is Navi+, the rlc reference clock is used for system clock
+	 * from vbios gfx_info table */
+	if (adev->asic_type >= CHIP_NAVI10) {
+		index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
+						   gfx_info);
+		if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+					  &frev, &crev, &data_offset)) {
+			union gfx_info *gfx_info = (union gfx_info *)
+				(mode_info->atom_context->bios + data_offset);
+			if ((frev == 3) ||
+			    (frev == 2 && crev == 6)) {
+				spll->reference_freq = le32_to_cpu(gfx_info->v30.golden_tsc_count_lower_refclk);
+				ret = 0;
+			} else if ((frev == 2) &&
+				   (crev >= 2) &&
+				   (crev != 6)) {
+				spll->reference_freq = le32_to_cpu(gfx_info->v22.rlc_gpu_timer_refclk);
+				ret = 0;
+			} else {
+				BUG();
+			}
+		}
+	}
+
 	return ret;
 }
+
+int amdgpu_atomfirmware_get_gfx_info(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index;
+	uint8_t frev, crev;
+	uint16_t data_offset;
+
+	index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
+					    gfx_info);
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset)) {
+		union gfx_info *gfx_info = (union gfx_info *)
+			(mode_info->atom_context->bios + data_offset);
+		if (frev == 2) {
+			switch (crev) {
+			case 4:
+				adev->gfx.config.max_shader_engines = gfx_info->v24.max_shader_engines;
+				adev->gfx.config.max_cu_per_sh = gfx_info->v24.max_cu_per_sh;
+				adev->gfx.config.max_sh_per_se = gfx_info->v24.max_sh_per_se;
+				adev->gfx.config.max_backends_per_se = gfx_info->v24.max_backends_per_se;
+				adev->gfx.config.max_texture_channel_caches = gfx_info->v24.max_texture_channel_caches;
+				adev->gfx.config.max_gprs = le16_to_cpu(gfx_info->v24.gc_num_gprs);
+				adev->gfx.config.max_gs_threads = gfx_info->v24.gc_num_max_gs_thds;
+				adev->gfx.config.gs_vgt_table_depth = gfx_info->v24.gc_gs_table_depth;
+				adev->gfx.config.gs_prim_buffer_depth =
+					le16_to_cpu(gfx_info->v24.gc_gsprim_buff_depth);
+				adev->gfx.config.double_offchip_lds_buf =
+					gfx_info->v24.gc_double_offchip_lds_buffer;
+				adev->gfx.cu_info.wave_front_size = le16_to_cpu(gfx_info->v24.gc_wave_size);
+				adev->gfx.cu_info.max_waves_per_simd = le16_to_cpu(gfx_info->v24.gc_max_waves_per_simd);
+				adev->gfx.cu_info.max_scratch_slots_per_cu = gfx_info->v24.gc_max_scratch_slots_per_cu;
+				adev->gfx.cu_info.lds_size = le16_to_cpu(gfx_info->v24.gc_lds_size);
+				return 0;
+			case 7:
+				adev->gfx.config.max_shader_engines = gfx_info->v27.max_shader_engines;
+				adev->gfx.config.max_cu_per_sh = gfx_info->v27.max_cu_per_sh;
+				adev->gfx.config.max_sh_per_se = gfx_info->v27.max_sh_per_se;
+				adev->gfx.config.max_backends_per_se = gfx_info->v27.max_backends_per_se;
+				adev->gfx.config.max_texture_channel_caches = gfx_info->v27.max_texture_channel_caches;
+				adev->gfx.config.max_gprs = le16_to_cpu(gfx_info->v27.gc_num_gprs);
+				adev->gfx.config.max_gs_threads = gfx_info->v27.gc_num_max_gs_thds;
+				adev->gfx.config.gs_vgt_table_depth = gfx_info->v27.gc_gs_table_depth;
+				adev->gfx.config.gs_prim_buffer_depth = le16_to_cpu(gfx_info->v27.gc_gsprim_buff_depth);
+				adev->gfx.config.double_offchip_lds_buf = gfx_info->v27.gc_double_offchip_lds_buffer;
+				adev->gfx.cu_info.wave_front_size = le16_to_cpu(gfx_info->v27.gc_wave_size);
+				adev->gfx.cu_info.max_waves_per_simd = le16_to_cpu(gfx_info->v27.gc_max_waves_per_simd);
+				adev->gfx.cu_info.max_scratch_slots_per_cu = gfx_info->v27.gc_max_scratch_slots_per_cu;
+				adev->gfx.cu_info.lds_size = le16_to_cpu(gfx_info->v27.gc_lds_size);
+				return 0;
+			default:
+				return -EINVAL;
+			}
+		} else if (frev == 3) {
+			switch (crev) {
+			case 0:
+				adev->gfx.config.max_shader_engines = gfx_info->v30.max_shader_engines;
+				adev->gfx.config.max_cu_per_sh = gfx_info->v30.max_cu_per_sh;
+				adev->gfx.config.max_sh_per_se = gfx_info->v30.max_sh_per_se;
+				adev->gfx.config.max_backends_per_se = gfx_info->v30.max_backends_per_se;
+				adev->gfx.config.max_texture_channel_caches = gfx_info->v30.max_texture_channel_caches;
+				return 0;
+			default:
+				return -EINVAL;
+			}
+		} else {
+			return -EINVAL;
+		}
+
+	}
+	return -EINVAL;
+}
+
+/*
+ * Helper function to query two stage mem training capability
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Return true if two stage mem training is supported or false if not
+ */
+bool amdgpu_atomfirmware_mem_training_supported(struct amdgpu_device *adev)
+{
+	u32 fw_cap;
+
+	fw_cap = adev->mode_info.firmware_flags;
+
+	return (fw_cap & ATOM_FIRMWARE_CAP_ENABLE_2STAGE_BIST_TRAINING) ? true : false;
+}
+
+int amdgpu_atomfirmware_get_fw_reserved_fb_size(struct amdgpu_device *adev)
+{
+	struct atom_context *ctx = adev->mode_info.atom_context;
+	union firmware_info *firmware_info;
+	int index;
+	u16 data_offset, size;
+	u8 frev, crev;
+	int fw_reserved_fb_size;
+
+	index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
+			firmwareinfo);
+
+	if (!amdgpu_atom_parse_data_header(ctx, index, &size,
+				&frev, &crev, &data_offset))
+		/* fail to parse data_header */
+		return 0;
+
+	firmware_info = (union firmware_info *)(ctx->bios + data_offset);
+
+	if (frev != 3)
+		return -EINVAL;
+
+	switch (crev) {
+	case 4:
+		fw_reserved_fb_size =
+			(firmware_info->v34.fw_reserved_size_in_kb << 10);
+		break;
+	case 5:
+		fw_reserved_fb_size =
+			(firmware_info->v35.fw_reserved_size_in_kb << 10);
+		break;
+	default:
+		fw_reserved_fb_size = 0;
+		break;
+	}
+
+	return fw_reserved_fb_size;
+}
+
+/*
+ * Helper function to execute asic_init table
+ *
+ * @adev: amdgpu_device pointer
+ * @fb_reset: flag to indicate whether fb is reset or not
+ *
+ * Return 0 if succeed, otherwise failed
+ */
+int amdgpu_atomfirmware_asic_init(struct amdgpu_device *adev, bool fb_reset)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	struct atom_context *ctx;
+	uint8_t frev, crev;
+	uint16_t data_offset;
+	uint32_t bootup_sclk_in10khz, bootup_mclk_in10khz;
+	struct asic_init_ps_allocation_v2_1 asic_init_ps_v2_1;
+	int index;
+
+	if (!mode_info)
+		return -EINVAL;
+
+	ctx = mode_info->atom_context;
+	if (!ctx)
+		return -EINVAL;
+
+	/* query bootup sclk/mclk from firmware_info table */
+	index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
+					    firmwareinfo);
+	if (amdgpu_atom_parse_data_header(ctx, index, NULL,
+				&frev, &crev, &data_offset)) {
+		union firmware_info *firmware_info =
+			(union firmware_info *)(ctx->bios +
+						data_offset);
+
+		bootup_sclk_in10khz =
+			le32_to_cpu(firmware_info->v31.bootup_sclk_in10khz);
+		bootup_mclk_in10khz =
+			le32_to_cpu(firmware_info->v31.bootup_mclk_in10khz);
+	} else {
+		return -EINVAL;
+	}
+
+	index = get_index_into_master_table(atom_master_list_of_command_functions_v2_1,
+					asic_init);
+	if (amdgpu_atom_parse_cmd_header(mode_info->atom_context, index, &frev, &crev)) {
+		if (frev == 2 && crev >= 1) {
+			memset(&asic_init_ps_v2_1, 0, sizeof(asic_init_ps_v2_1));
+			asic_init_ps_v2_1.param.engineparam.sclkfreqin10khz = bootup_sclk_in10khz;
+			asic_init_ps_v2_1.param.memparam.mclkfreqin10khz = bootup_mclk_in10khz;
+			asic_init_ps_v2_1.param.engineparam.engineflag = b3NORMAL_ENGINE_INIT;
+			if (!fb_reset)
+				asic_init_ps_v2_1.param.memparam.memflag = b3DRAM_SELF_REFRESH_EXIT;
+			else
+				asic_init_ps_v2_1.param.memparam.memflag = 0;
+		} else {
+			return -EINVAL;
+		}
+	} else {
+		return -EINVAL;
+	}
+
+	return amdgpu_atom_execute_table(ctx, ATOM_CMD_INIT, (uint32_t *)&asic_init_ps_v2_1,
+		sizeof(asic_init_ps_v2_1));
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.h
index 288b97e54347..649b5530d8ae 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.h
@@ -24,10 +24,22 @@
 #ifndef __AMDGPU_ATOMFIRMWARE_H__
 #define __AMDGPU_ATOMFIRMWARE_H__
 
-bool amdgpu_atomfirmware_gpu_supports_virtualization(struct amdgpu_device *adev);
+#define get_index_into_master_table(master_table, table_name) (offsetof(struct master_table, table_name) / sizeof(uint16_t))
+
+uint32_t amdgpu_atomfirmware_query_firmware_capability(struct amdgpu_device *adev);
+bool amdgpu_atomfirmware_gpu_virtualization_supported(struct amdgpu_device *adev);
 void amdgpu_atomfirmware_scratch_regs_init(struct amdgpu_device *adev);
 int amdgpu_atomfirmware_allocate_fb_scratch(struct amdgpu_device *adev);
-int amdgpu_atomfirmware_get_vram_width(struct amdgpu_device *adev);
+int amdgpu_atomfirmware_get_vram_info(struct amdgpu_device *adev,
+	int *vram_width, int *vram_type, int *vram_vendor);
 int amdgpu_atomfirmware_get_clock_info(struct amdgpu_device *adev);
+int amdgpu_atomfirmware_get_gfx_info(struct amdgpu_device *adev);
+bool amdgpu_atomfirmware_mem_ecc_supported(struct amdgpu_device *adev);
+bool amdgpu_atomfirmware_sram_ecc_supported(struct amdgpu_device *adev);
+bool amdgpu_atomfirmware_ras_rom_addr(struct amdgpu_device *adev, uint8_t *i2c_address);
+bool amdgpu_atomfirmware_mem_training_supported(struct amdgpu_device *adev);
+bool amdgpu_atomfirmware_dynamic_boot_config_supported(struct amdgpu_device *adev);
+int amdgpu_atomfirmware_get_fw_reserved_fb_size(struct amdgpu_device *adev);
+int amdgpu_atomfirmware_asic_init(struct amdgpu_device *adev, bool fb_reset);
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c
index c13c51af0b68..3893e6fc2f03 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c
@@ -1,9 +1,8 @@
+// SPDX-License-Identifier: GPL-2.0-only
 /*
  * Copyright (c) 2010 Red Hat Inc.
  * Author : Dave Airlie <airlied@redhat.com>
  *
- * Licensed under GPLv2
- *
  * ATPX support for both Intel/ATI
  */
 #include <linux/vga_switcheroo.h>
@@ -12,8 +11,19 @@
 #include <linux/pci.h>
 #include <linux/delay.h>
 
+#include "amdgpu.h"
 #include "amd_acpi.h"
 
+#define AMDGPU_PX_QUIRK_FORCE_ATPX  (1 << 0)
+
+struct amdgpu_px_quirk {
+	u32 chip_vendor;
+	u32 chip_device;
+	u32 subsys_vendor;
+	u32 subsys_device;
+	u32 px_quirk_flags;
+};
+
 struct amdgpu_atpx_functions {
 	bool px_params;
 	bool power_cntl;
@@ -22,7 +32,7 @@ struct amdgpu_atpx_functions {
 	bool switch_start;
 	bool switch_end;
 	bool disp_connectors_mapping;
-	bool disp_detetion_ports;
+	bool disp_detection_ports;
 };
 
 struct amdgpu_atpx {
@@ -35,6 +45,7 @@ struct amdgpu_atpx {
 static struct amdgpu_atpx_priv {
 	bool atpx_detected;
 	bool bridge_pm_usable;
+	unsigned int quirks;
 	/* handle for device - and atpx */
 	acpi_handle dhandle;
 	acpi_handle other_handle;
@@ -63,22 +74,21 @@ struct atpx_mux {
 	u16 mux;
 } __packed;
 
-bool amdgpu_has_atpx(void) {
+bool amdgpu_has_atpx(void)
+{
 	return amdgpu_atpx_priv.atpx_detected;
 }
 
-bool amdgpu_has_atpx_dgpu_power_cntl(void) {
+bool amdgpu_has_atpx_dgpu_power_cntl(void)
+{
 	return amdgpu_atpx_priv.atpx.functions.power_cntl;
 }
 
-bool amdgpu_is_atpx_hybrid(void) {
+bool amdgpu_is_atpx_hybrid(void)
+{
 	return amdgpu_atpx_priv.atpx.is_hybrid;
 }
 
-bool amdgpu_atpx_dgpu_req_power_for_displays(void) {
-	return amdgpu_atpx_priv.atpx.dgpu_req_power_for_displays;
-}
-
 /**
  * amdgpu_atpx_call - call an ATPX method
  *
@@ -117,7 +127,7 @@ static union acpi_object *amdgpu_atpx_call(acpi_handle handle, int function,
 
 	/* Fail only if calling the method fails and ATPX is supported */
 	if (ACPI_FAILURE(status) && status != AE_NOT_FOUND) {
-		printk("failed to evaluate ATPX got %s\n",
+		pr_err("failed to evaluate ATPX got %s\n",
 		       acpi_format_exception(status));
 		kfree(buffer.pointer);
 		return NULL;
@@ -145,11 +155,11 @@ static void amdgpu_atpx_parse_functions(struct amdgpu_atpx_functions *f, u32 mas
 	f->switch_start = mask & ATPX_GRAPHICS_DEVICE_SWITCH_START_NOTIFICATION_SUPPORTED;
 	f->switch_end = mask & ATPX_GRAPHICS_DEVICE_SWITCH_END_NOTIFICATION_SUPPORTED;
 	f->disp_connectors_mapping = mask & ATPX_GET_DISPLAY_CONNECTORS_MAPPING_SUPPORTED;
-	f->disp_detetion_ports = mask & ATPX_GET_DISPLAY_DETECTION_PORTS_SUPPORTED;
+	f->disp_detection_ports = mask & ATPX_GET_DISPLAY_DETECTION_PORTS_SUPPORTED;
 }
 
 /**
- * amdgpu_atpx_validate_functions - validate ATPX functions
+ * amdgpu_atpx_validate - validate ATPX functions
  *
  * @atpx: amdgpu atpx struct
  *
@@ -173,7 +183,7 @@ static int amdgpu_atpx_validate(struct amdgpu_atpx *atpx)
 
 		size = *(u16 *) info->buffer.pointer;
 		if (size < 10) {
-			printk("ATPX buffer is too small: %zu\n", size);
+			pr_err("ATPX buffer is too small: %zu\n", size);
 			kfree(info);
 			return -EINVAL;
 		}
@@ -205,13 +215,19 @@ static int amdgpu_atpx_validate(struct amdgpu_atpx *atpx)
 
 	atpx->is_hybrid = false;
 	if (valid_bits & ATPX_MS_HYBRID_GFX_SUPPORTED) {
-		printk("ATPX Hybrid Graphics\n");
-		/*
-		 * Disable legacy PM methods only when pcie port PM is usable,
-		 * otherwise the device might fail to power off or power on.
-		 */
-		atpx->functions.power_cntl = !amdgpu_atpx_priv.bridge_pm_usable;
-		atpx->is_hybrid = true;
+		if (amdgpu_atpx_priv.quirks & AMDGPU_PX_QUIRK_FORCE_ATPX) {
+			pr_warn("ATPX Hybrid Graphics, forcing to ATPX\n");
+			atpx->functions.power_cntl = true;
+			atpx->is_hybrid = false;
+		} else {
+			pr_notice("ATPX Hybrid Graphics\n");
+			/*
+			 * Disable legacy PM methods only when pcie port PM is usable,
+			 * otherwise the device might fail to power off or power on.
+			 */
+			atpx->functions.power_cntl = !amdgpu_atpx_priv.bridge_pm_usable;
+			atpx->is_hybrid = true;
+		}
 	}
 
 	atpx->dgpu_req_power_for_displays = false;
@@ -246,7 +262,7 @@ static int amdgpu_atpx_verify_interface(struct amdgpu_atpx *atpx)
 
 	size = *(u16 *) info->buffer.pointer;
 	if (size < 8) {
-		printk("ATPX buffer is too small: %zu\n", size);
+		pr_err("ATPX buffer is too small: %zu\n", size);
 		err = -EINVAL;
 		goto out;
 	}
@@ -255,8 +271,8 @@ static int amdgpu_atpx_verify_interface(struct amdgpu_atpx *atpx)
 	memcpy(&output, info->buffer.pointer, size);
 
 	/* TODO: check version? */
-	printk("ATPX version %u, functions 0x%08x\n",
-	       output.version, output.function_bits);
+	pr_notice("ATPX version %u, functions 0x%08x\n",
+		  output.version, output.function_bits);
 
 	amdgpu_atpx_parse_functions(&atpx->functions, output.function_bits);
 
@@ -533,7 +549,7 @@ static int amdgpu_atpx_init(void)
  * look up whether we are the integrated or discrete GPU (all asics).
  * Returns the client id.
  */
-static int amdgpu_atpx_get_client_id(struct pci_dev *pdev)
+static enum vga_switcheroo_client_id amdgpu_atpx_get_client_id(struct pci_dev *pdev)
 {
 	if (amdgpu_atpx_priv.dhandle == ACPI_HANDLE(&pdev->dev))
 		return VGA_SWITCHEROO_IGD;
@@ -547,6 +563,34 @@ static const struct vga_switcheroo_handler amdgpu_atpx_handler = {
 	.get_client_id = amdgpu_atpx_get_client_id,
 };
 
+static const struct amdgpu_px_quirk amdgpu_px_quirk_list[] = {
+	/* HG _PR3 doesn't seem to work on this A+A weston board */
+	{ 0x1002, 0x6900, 0x1002, 0x0124, AMDGPU_PX_QUIRK_FORCE_ATPX },
+	{ 0x1002, 0x6900, 0x1028, 0x0812, AMDGPU_PX_QUIRK_FORCE_ATPX },
+	{ 0x1002, 0x6900, 0x1028, 0x0813, AMDGPU_PX_QUIRK_FORCE_ATPX },
+	{ 0x1002, 0x699f, 0x1028, 0x0814, AMDGPU_PX_QUIRK_FORCE_ATPX },
+	{ 0x1002, 0x6900, 0x1025, 0x125A, AMDGPU_PX_QUIRK_FORCE_ATPX },
+	{ 0x1002, 0x6900, 0x17AA, 0x3806, AMDGPU_PX_QUIRK_FORCE_ATPX },
+	{ 0, 0, 0, 0, 0 },
+};
+
+static void amdgpu_atpx_get_quirks(struct pci_dev *pdev)
+{
+	const struct amdgpu_px_quirk *p = amdgpu_px_quirk_list;
+
+	/* Apply PX quirks */
+	while (p && p->chip_device != 0) {
+		if (pdev->vendor == p->chip_vendor &&
+		    pdev->device == p->chip_device &&
+		    pdev->subsystem_vendor == p->subsys_vendor &&
+		    pdev->subsystem_device == p->subsys_device) {
+			amdgpu_atpx_priv.quirks |= p->px_quirk_flags;
+			break;
+		}
+		++p;
+	}
+}
+
 /**
  * amdgpu_atpx_detect - detect whether we have PX
  *
@@ -566,19 +610,21 @@ static bool amdgpu_atpx_detect(void)
 	while ((pdev = pci_get_class(PCI_CLASS_DISPLAY_VGA << 8, pdev)) != NULL) {
 		vga_count++;
 
-		has_atpx |= (amdgpu_atpx_pci_probe_handle(pdev) == true);
+		has_atpx |= amdgpu_atpx_pci_probe_handle(pdev);
 
 		parent_pdev = pci_upstream_bridge(pdev);
 		d3_supported |= parent_pdev && parent_pdev->bridge_d3;
+		amdgpu_atpx_get_quirks(pdev);
 	}
 
 	while ((pdev = pci_get_class(PCI_CLASS_DISPLAY_OTHER << 8, pdev)) != NULL) {
 		vga_count++;
 
-		has_atpx |= (amdgpu_atpx_pci_probe_handle(pdev) == true);
+		has_atpx |= amdgpu_atpx_pci_probe_handle(pdev);
 
 		parent_pdev = pci_upstream_bridge(pdev);
 		d3_supported |= parent_pdev && parent_pdev->bridge_d3;
+		amdgpu_atpx_get_quirks(pdev);
 	}
 
 	if (has_atpx && vga_count == 2) {
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c
index 63ec1e1bb6aa..199693369c7c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c
@@ -21,7 +21,7 @@
  *
  * Authors: Jerome Glisse
  */
-#include <drm/drmP.h>
+
 #include <drm/amdgpu_drm.h>
 #include "amdgpu.h"
 
@@ -29,120 +29,99 @@
 #define AMDGPU_BENCHMARK_COMMON_MODES_N 17
 
 static int amdgpu_benchmark_do_move(struct amdgpu_device *adev, unsigned size,
-				    uint64_t saddr, uint64_t daddr, int n)
+				    uint64_t saddr, uint64_t daddr, int n, s64 *time_ms)
 {
-	unsigned long start_jiffies;
-	unsigned long end_jiffies;
-	struct dma_fence *fence = NULL;
+	ktime_t stime, etime;
+	struct dma_fence *fence;
 	int i, r;
 
-	start_jiffies = jiffies;
+	stime = ktime_get();
 	for (i = 0; i < n; i++) {
 		struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
 		r = amdgpu_copy_buffer(ring, saddr, daddr, size, NULL, &fence,
-				       false, false);
+				       false, false, 0);
 		if (r)
 			goto exit_do_move;
 		r = dma_fence_wait(fence, false);
+		dma_fence_put(fence);
 		if (r)
 			goto exit_do_move;
-		dma_fence_put(fence);
 	}
-	end_jiffies = jiffies;
-	r = jiffies_to_msecs(end_jiffies - start_jiffies);
 
 exit_do_move:
-	if (fence)
-		dma_fence_put(fence);
+	etime = ktime_get();
+	*time_ms = ktime_ms_delta(etime, stime);
+
 	return r;
 }
 
 
-static void amdgpu_benchmark_log_results(int n, unsigned size,
-					 unsigned int time,
+static void amdgpu_benchmark_log_results(struct amdgpu_device *adev,
+					 int n, unsigned size,
+					 s64 time_ms,
 					 unsigned sdomain, unsigned ddomain,
 					 char *kind)
 {
-	unsigned int throughput = (n * (size >> 10)) / time;
-	DRM_INFO("amdgpu: %s %u bo moves of %u kB from"
-		 " %d to %d in %u ms, throughput: %u Mb/s or %u MB/s\n",
-		 kind, n, size >> 10, sdomain, ddomain, time,
+	s64 throughput = (n * (size >> 10));
+
+	throughput = div64_s64(throughput, time_ms);
+
+	dev_info(adev->dev, "amdgpu: %s %u bo moves of %u kB from"
+		 " %d to %d in %lld ms, throughput: %lld Mb/s or %lld MB/s\n",
+		 kind, n, size >> 10, sdomain, ddomain, time_ms,
 		 throughput * 8, throughput);
 }
 
-static void amdgpu_benchmark_move(struct amdgpu_device *adev, unsigned size,
-				  unsigned sdomain, unsigned ddomain)
+static int amdgpu_benchmark_move(struct amdgpu_device *adev, unsigned size,
+				 unsigned sdomain, unsigned ddomain)
 {
 	struct amdgpu_bo *dobj = NULL;
 	struct amdgpu_bo *sobj = NULL;
 	uint64_t saddr, daddr;
+	s64 time_ms;
 	int r, n;
-	int time;
 
 	n = AMDGPU_BENCHMARK_ITERATIONS;
-	r = amdgpu_bo_create(adev, size, PAGE_SIZE, true, sdomain, 0, NULL,
-			     NULL, 0, &sobj);
-	if (r) {
-		goto out_cleanup;
-	}
-	r = amdgpu_bo_reserve(sobj, false);
-	if (unlikely(r != 0))
-		goto out_cleanup;
-	r = amdgpu_bo_pin(sobj, sdomain, &saddr);
-	amdgpu_bo_unreserve(sobj);
-	if (r) {
-		goto out_cleanup;
-	}
-	r = amdgpu_bo_create(adev, size, PAGE_SIZE, true, ddomain, 0, NULL,
-			     NULL, 0, &dobj);
-	if (r) {
-		goto out_cleanup;
-	}
-	r = amdgpu_bo_reserve(dobj, false);
-	if (unlikely(r != 0))
+
+	r = amdgpu_bo_create_kernel(adev, size,
+				    PAGE_SIZE, sdomain,
+				    &sobj,
+				    &saddr,
+				    NULL);
+	if (r)
 		goto out_cleanup;
-	r = amdgpu_bo_pin(dobj, ddomain, &daddr);
-	amdgpu_bo_unreserve(dobj);
-	if (r) {
+	r = amdgpu_bo_create_kernel(adev, size,
+				    PAGE_SIZE, ddomain,
+				    &dobj,
+				    &daddr,
+				    NULL);
+	if (r)
 		goto out_cleanup;
-	}
 
 	if (adev->mman.buffer_funcs) {
-		time = amdgpu_benchmark_do_move(adev, size, saddr, daddr, n);
-		if (time < 0)
+		r = amdgpu_benchmark_do_move(adev, size, saddr, daddr, n, &time_ms);
+		if (r)
 			goto out_cleanup;
-		if (time > 0)
-			amdgpu_benchmark_log_results(n, size, time,
+		else
+			amdgpu_benchmark_log_results(adev, n, size, time_ms,
 						     sdomain, ddomain, "dma");
 	}
 
 out_cleanup:
 	/* Check error value now. The value can be overwritten when clean up.*/
-	if (r) {
-		DRM_ERROR("Error while benchmarking BO move.\n");
-	}
+	if (r < 0)
+		dev_info(adev->dev, "Error while benchmarking BO move.\n");
 
-	if (sobj) {
-		r = amdgpu_bo_reserve(sobj, true);
-		if (likely(r == 0)) {
-			amdgpu_bo_unpin(sobj);
-			amdgpu_bo_unreserve(sobj);
-		}
-		amdgpu_bo_unref(&sobj);
-	}
-	if (dobj) {
-		r = amdgpu_bo_reserve(dobj, true);
-		if (likely(r == 0)) {
-			amdgpu_bo_unpin(dobj);
-			amdgpu_bo_unreserve(dobj);
-		}
-		amdgpu_bo_unref(&dobj);
-	}
+	if (sobj)
+		amdgpu_bo_free_kernel(&sobj, &saddr, NULL);
+	if (dobj)
+		amdgpu_bo_free_kernel(&dobj, &daddr, NULL);
+	return r;
 }
 
-void amdgpu_benchmark(struct amdgpu_device *adev, int test_number)
+int amdgpu_benchmark(struct amdgpu_device *adev, int test_number)
 {
-	int i;
+	int i, r;
 	static const int common_modes[AMDGPU_BENCHMARK_COMMON_MODES_N] = {
 		640 * 480 * 4,
 		720 * 480 * 4,
@@ -163,63 +142,119 @@ void amdgpu_benchmark(struct amdgpu_device *adev, int test_number)
 		1920 * 1200 * 4
 	};
 
+	mutex_lock(&adev->benchmark_mutex);
 	switch (test_number) {
 	case 1:
+		dev_info(adev->dev,
+			 "benchmark test: %d (simple test, VRAM to GTT and GTT to VRAM)\n",
+			 test_number);
 		/* simple test, VRAM to GTT and GTT to VRAM */
-		amdgpu_benchmark_move(adev, 1024*1024, AMDGPU_GEM_DOMAIN_GTT,
-				      AMDGPU_GEM_DOMAIN_VRAM);
-		amdgpu_benchmark_move(adev, 1024*1024, AMDGPU_GEM_DOMAIN_VRAM,
-				      AMDGPU_GEM_DOMAIN_GTT);
+		r = amdgpu_benchmark_move(adev, 1024*1024, AMDGPU_GEM_DOMAIN_GTT,
+					  AMDGPU_GEM_DOMAIN_VRAM);
+		if (r)
+			goto done;
+		r = amdgpu_benchmark_move(adev, 1024*1024, AMDGPU_GEM_DOMAIN_VRAM,
+					  AMDGPU_GEM_DOMAIN_GTT);
+		if (r)
+			goto done;
 		break;
 	case 2:
+		dev_info(adev->dev,
+			 "benchmark test: %d (simple test, VRAM to VRAM)\n",
+			 test_number);
 		/* simple test, VRAM to VRAM */
-		amdgpu_benchmark_move(adev, 1024*1024, AMDGPU_GEM_DOMAIN_VRAM,
-				      AMDGPU_GEM_DOMAIN_VRAM);
+		r = amdgpu_benchmark_move(adev, 1024*1024, AMDGPU_GEM_DOMAIN_VRAM,
+					  AMDGPU_GEM_DOMAIN_VRAM);
+		if (r)
+			goto done;
 		break;
 	case 3:
+		dev_info(adev->dev,
+			 "benchmark test: %d (GTT to VRAM, buffer size sweep, powers of 2)\n",
+			 test_number);
 		/* GTT to VRAM, buffer size sweep, powers of 2 */
-		for (i = 1; i <= 16384; i <<= 1)
-			amdgpu_benchmark_move(adev, i * AMDGPU_GPU_PAGE_SIZE,
-					      AMDGPU_GEM_DOMAIN_GTT,
-					      AMDGPU_GEM_DOMAIN_VRAM);
+		for (i = 1; i <= 16384; i <<= 1) {
+			r = amdgpu_benchmark_move(adev, i * AMDGPU_GPU_PAGE_SIZE,
+						  AMDGPU_GEM_DOMAIN_GTT,
+						  AMDGPU_GEM_DOMAIN_VRAM);
+			if (r)
+				goto done;
+		}
 		break;
 	case 4:
+		dev_info(adev->dev,
+			 "benchmark test: %d (VRAM to GTT, buffer size sweep, powers of 2)\n",
+			 test_number);
 		/* VRAM to GTT, buffer size sweep, powers of 2 */
-		for (i = 1; i <= 16384; i <<= 1)
-			amdgpu_benchmark_move(adev, i * AMDGPU_GPU_PAGE_SIZE,
-					      AMDGPU_GEM_DOMAIN_VRAM,
-					      AMDGPU_GEM_DOMAIN_GTT);
+		for (i = 1; i <= 16384; i <<= 1) {
+			r = amdgpu_benchmark_move(adev, i * AMDGPU_GPU_PAGE_SIZE,
+						  AMDGPU_GEM_DOMAIN_VRAM,
+						  AMDGPU_GEM_DOMAIN_GTT);
+			if (r)
+				goto done;
+		}
 		break;
 	case 5:
+		dev_info(adev->dev,
+			 "benchmark test: %d (VRAM to VRAM, buffer size sweep, powers of 2)\n",
+			 test_number);
 		/* VRAM to VRAM, buffer size sweep, powers of 2 */
-		for (i = 1; i <= 16384; i <<= 1)
-			amdgpu_benchmark_move(adev, i * AMDGPU_GPU_PAGE_SIZE,
-					      AMDGPU_GEM_DOMAIN_VRAM,
-					      AMDGPU_GEM_DOMAIN_VRAM);
+		for (i = 1; i <= 16384; i <<= 1) {
+			r = amdgpu_benchmark_move(adev, i * AMDGPU_GPU_PAGE_SIZE,
+						  AMDGPU_GEM_DOMAIN_VRAM,
+						  AMDGPU_GEM_DOMAIN_VRAM);
+			if (r)
+				goto done;
+		}
 		break;
 	case 6:
+		dev_info(adev->dev,
+			 "benchmark test: %d (GTT to VRAM, buffer size sweep, common modes)\n",
+			 test_number);
 		/* GTT to VRAM, buffer size sweep, common modes */
-		for (i = 0; i < AMDGPU_BENCHMARK_COMMON_MODES_N; i++)
-			amdgpu_benchmark_move(adev, common_modes[i],
-					      AMDGPU_GEM_DOMAIN_GTT,
-					      AMDGPU_GEM_DOMAIN_VRAM);
+		for (i = 0; i < AMDGPU_BENCHMARK_COMMON_MODES_N; i++) {
+			r = amdgpu_benchmark_move(adev, common_modes[i],
+						  AMDGPU_GEM_DOMAIN_GTT,
+						  AMDGPU_GEM_DOMAIN_VRAM);
+			if (r)
+				goto done;
+		}
 		break;
 	case 7:
+		dev_info(adev->dev,
+			 "benchmark test: %d (VRAM to GTT, buffer size sweep, common modes)\n",
+			 test_number);
 		/* VRAM to GTT, buffer size sweep, common modes */
-		for (i = 0; i < AMDGPU_BENCHMARK_COMMON_MODES_N; i++)
-			amdgpu_benchmark_move(adev, common_modes[i],
-					      AMDGPU_GEM_DOMAIN_VRAM,
-					      AMDGPU_GEM_DOMAIN_GTT);
+		for (i = 0; i < AMDGPU_BENCHMARK_COMMON_MODES_N; i++) {
+			r = amdgpu_benchmark_move(adev, common_modes[i],
+						  AMDGPU_GEM_DOMAIN_VRAM,
+						  AMDGPU_GEM_DOMAIN_GTT);
+			if (r)
+				goto done;
+		}
 		break;
 	case 8:
+		dev_info(adev->dev,
+			 "benchmark test: %d (VRAM to VRAM, buffer size sweep, common modes)\n",
+			 test_number);
 		/* VRAM to VRAM, buffer size sweep, common modes */
-		for (i = 0; i < AMDGPU_BENCHMARK_COMMON_MODES_N; i++)
-			amdgpu_benchmark_move(adev, common_modes[i],
+		for (i = 0; i < AMDGPU_BENCHMARK_COMMON_MODES_N; i++) {
+			r = amdgpu_benchmark_move(adev, common_modes[i],
 					      AMDGPU_GEM_DOMAIN_VRAM,
 					      AMDGPU_GEM_DOMAIN_VRAM);
+			if (r)
+				goto done;
+		}
 		break;
 
 	default:
-		DRM_ERROR("Unknown benchmark\n");
+		dev_info(adev->dev, "Unknown benchmark %d\n", test_number);
+		r = -EINVAL;
+		break;
 	}
+
+done:
+	mutex_unlock(&adev->benchmark_mutex);
+
+	return r;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
index c21adf60a7f2..00e96419fcda 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
@@ -25,10 +25,12 @@
  *          Alex Deucher
  *          Jerome Glisse
  */
-#include <drm/drmP.h>
+
 #include "amdgpu.h"
 #include "atom.h"
 
+#include <linux/device.h>
+#include <linux/pci.h>
 #include <linux/slab.h>
 #include <linux/acpi.h>
 /*
@@ -45,69 +47,76 @@
 /* Check if current bios is an ATOM BIOS.
  * Return true if it is ATOM BIOS. Otherwise, return false.
  */
-static bool check_atom_bios(uint8_t *bios, size_t size)
+static bool check_atom_bios(struct amdgpu_device *adev, size_t size)
 {
 	uint16_t tmp, bios_header_start;
+	uint8_t *bios = adev->bios;
 
 	if (!bios || size < 0x49) {
-		DRM_INFO("vbios mem is null or mem size is wrong\n");
+		dev_dbg(adev->dev, "VBIOS mem is null or mem size is wrong\n");
 		return false;
 	}
 
 	if (!AMD_IS_VALID_VBIOS(bios)) {
-		DRM_INFO("BIOS signature incorrect %x %x\n", bios[0], bios[1]);
-		return false;
-	}
-
-	tmp = bios[0x18] | (bios[0x19] << 8);
-	if (bios[tmp + 0x14] != 0x0) {
-		DRM_INFO("Not an x86 BIOS ROM\n");
+		dev_dbg(adev->dev, "VBIOS signature incorrect %x %x\n", bios[0],
+			bios[1]);
 		return false;
 	}
 
 	bios_header_start = bios[0x48] | (bios[0x49] << 8);
 	if (!bios_header_start) {
-		DRM_INFO("Can't locate bios header\n");
+		dev_dbg(adev->dev, "Can't locate VBIOS header\n");
 		return false;
 	}
 
 	tmp = bios_header_start + 4;
 	if (size < tmp) {
-		DRM_INFO("BIOS header is broken\n");
+		dev_dbg(adev->dev, "VBIOS header is broken\n");
 		return false;
 	}
 
 	if (!memcmp(bios + tmp, "ATOM", 4) ||
 	    !memcmp(bios + tmp, "MOTA", 4)) {
-		DRM_DEBUG("ATOMBIOS detected\n");
+		dev_dbg(adev->dev, "ATOMBIOS detected\n");
 		return true;
 	}
 
 	return false;
 }
 
+void amdgpu_bios_release(struct amdgpu_device *adev)
+{
+	kfree(adev->bios);
+	adev->bios = NULL;
+	adev->bios_size = 0;
+}
+
 /* If you boot an IGP board with a discrete card as the primary,
  * the IGP rom is not accessible via the rom bar as the IGP rom is
  * part of the system bios.  On boot, the system bios puts a
  * copy of the igp rom at the start of vram if a discrete card is
  * present.
+ * For SR-IOV, the vbios image is also put in VRAM in the VF.
  */
-static bool igp_read_bios_from_vram(struct amdgpu_device *adev)
+static bool amdgpu_read_bios_from_vram(struct amdgpu_device *adev)
 {
 	uint8_t __iomem *bios;
 	resource_size_t vram_base;
 	resource_size_t size = 256 * 1024; /* ??? */
 
 	if (!(adev->flags & AMD_IS_APU))
-		if (amdgpu_need_post(adev))
+		if (amdgpu_device_need_post(adev))
 			return false;
 
+	/* FB BAR not enabled */
+	if (pci_resource_len(adev->pdev, 0) == 0)
+		return false;
+
 	adev->bios = NULL;
 	vram_base = pci_resource_start(adev->pdev, 0);
 	bios = ioremap_wc(vram_base, size);
-	if (!bios) {
+	if (!bios)
 		return false;
-	}
 
 	adev->bios = kmalloc(size, GFP_KERNEL);
 	if (!adev->bios) {
@@ -118,8 +127,8 @@ static bool igp_read_bios_from_vram(struct amdgpu_device *adev)
 	memcpy_fromio(adev->bios, bios, size);
 	iounmap(bios);
 
-	if (!check_atom_bios(adev->bios, size)) {
-		kfree(adev->bios);
+	if (!check_atom_bios(adev, size)) {
+		amdgpu_bios_release(adev);
 		return false;
 	}
 
@@ -134,9 +143,8 @@ bool amdgpu_read_bios(struct amdgpu_device *adev)
 	adev->bios = NULL;
 	/* XXX: some cards may return 0 for rom size? ddx has a workaround */
 	bios = pci_map_rom(adev->pdev, &size);
-	if (!bios) {
+	if (!bios)
 		return false;
-	}
 
 	adev->bios = kzalloc(size, GFP_KERNEL);
 	if (adev->bios == NULL) {
@@ -147,8 +155,8 @@ bool amdgpu_read_bios(struct amdgpu_device *adev)
 	memcpy_fromio(adev->bios, bios, size);
 	pci_unmap_rom(adev->pdev, bios);
 
-	if (!check_atom_bios(adev->bios, size)) {
-		kfree(adev->bios);
+	if (!check_atom_bios(adev, size)) {
+		amdgpu_bios_release(adev);
 		return false;
 	}
 
@@ -160,7 +168,7 @@ static bool amdgpu_read_bios_from_rom(struct amdgpu_device *adev)
 	u8 header[AMD_VBIOS_SIGNATURE_END+1] = {0};
 	int len;
 
-	if (!adev->asic_funcs->read_bios_from_rom)
+	if (!adev->asic_funcs || !adev->asic_funcs->read_bios_from_rom)
 		return false;
 
 	/* validate VBIOS signature */
@@ -169,9 +177,9 @@ static bool amdgpu_read_bios_from_rom(struct amdgpu_device *adev)
 	header[AMD_VBIOS_SIGNATURE_END] = 0;
 
 	if ((!AMD_IS_VALID_VBIOS(header)) ||
-	    0 != memcmp((char *)&header[AMD_VBIOS_SIGNATURE_OFFSET],
-			AMD_VBIOS_SIGNATURE,
-			strlen(AMD_VBIOS_SIGNATURE)))
+		memcmp((char *)&header[AMD_VBIOS_SIGNATURE_OFFSET],
+		       AMD_VBIOS_SIGNATURE,
+		       strlen(AMD_VBIOS_SIGNATURE)) != 0)
 		return false;
 
 	/* valid vbios, go on */
@@ -187,8 +195,8 @@ static bool amdgpu_read_bios_from_rom(struct amdgpu_device *adev)
 	/* read complete BIOS */
 	amdgpu_asic_read_bios_from_rom(adev, adev->bios, len);
 
-	if (!check_atom_bios(adev->bios, len)) {
-		kfree(adev->bios);
+	if (!check_atom_bios(adev, len)) {
+		amdgpu_bios_release(adev);
 		return false;
 	}
 
@@ -197,30 +205,36 @@ static bool amdgpu_read_bios_from_rom(struct amdgpu_device *adev)
 
 static bool amdgpu_read_platform_bios(struct amdgpu_device *adev)
 {
-	uint8_t __iomem *bios;
-	size_t size;
+	phys_addr_t rom = adev->pdev->rom;
+	size_t romlen = adev->pdev->romlen;
+	void __iomem *bios;
 
 	adev->bios = NULL;
 
-	bios = pci_platform_rom(adev->pdev, &size);
-	if (!bios) {
+	if (!rom || romlen == 0)
 		return false;
-	}
 
-	adev->bios = kzalloc(size, GFP_KERNEL);
-	if (adev->bios == NULL)
+	adev->bios = kzalloc(romlen, GFP_KERNEL);
+	if (!adev->bios)
 		return false;
 
-	memcpy_fromio(adev->bios, bios, size);
+	bios = ioremap(rom, romlen);
+	if (!bios)
+		goto free_bios;
 
-	if (!check_atom_bios(adev->bios, size)) {
-		kfree(adev->bios);
-		return false;
-	}
+	memcpy_fromio(adev->bios, bios, romlen);
+	iounmap(bios);
 
-	adev->bios_size = size;
+	if (!check_atom_bios(adev, romlen))
+		goto free_bios;
+
+	adev->bios_size = romlen;
 
 	return true;
+free_bios:
+	amdgpu_bios_release(adev);
+
+	return false;
 }
 
 #ifdef CONFIG_ACPI
@@ -260,7 +274,7 @@ static int amdgpu_atrm_call(acpi_handle atrm_handle, uint8_t *bios,
 
 	status = acpi_evaluate_object(atrm_handle, NULL, &atrm_arg, &buffer);
 	if (ACPI_FAILURE(status)) {
-		printk("failed to evaluate ATRM got %s\n", acpi_format_exception(status));
+		DRM_ERROR("failed to evaluate ATRM got %s\n", acpi_format_exception(status));
 		return -ENODEV;
 	}
 
@@ -281,42 +295,33 @@ static bool amdgpu_atrm_get_bios(struct amdgpu_device *adev)
 	acpi_status status;
 	bool found = false;
 
-	/* ATRM is for the discrete card only */
-	if (adev->flags & AMD_IS_APU)
+	/* ATRM is for on-platform devices only */
+	if (dev_is_removable(&adev->pdev->dev))
 		return false;
 
-	while ((pdev = pci_get_class(PCI_CLASS_DISPLAY_VGA << 8, pdev)) != NULL) {
+	while ((pdev = pci_get_base_class(PCI_BASE_CLASS_DISPLAY, pdev))) {
+		if ((pdev->class != PCI_CLASS_DISPLAY_VGA << 8) &&
+		    (pdev->class != PCI_CLASS_DISPLAY_OTHER << 8))
+			continue;
+
 		dhandle = ACPI_HANDLE(&pdev->dev);
 		if (!dhandle)
 			continue;
 
 		status = acpi_get_handle(dhandle, "ATRM", &atrm_handle);
-		if (!ACPI_FAILURE(status)) {
+		if (ACPI_SUCCESS(status)) {
 			found = true;
 			break;
 		}
 	}
 
-	if (!found) {
-		while ((pdev = pci_get_class(PCI_CLASS_DISPLAY_OTHER << 8, pdev)) != NULL) {
-			dhandle = ACPI_HANDLE(&pdev->dev);
-			if (!dhandle)
-				continue;
-
-			status = acpi_get_handle(dhandle, "ATRM", &atrm_handle);
-			if (!ACPI_FAILURE(status)) {
-				found = true;
-				break;
-			}
-		}
-	}
-
 	if (!found)
 		return false;
+	pci_dev_put(pdev);
 
 	adev->bios = kmalloc(size, GFP_KERNEL);
 	if (!adev->bios) {
-		DRM_ERROR("Unable to allocate bios\n");
+		dev_err(adev->dev, "Unable to allocate bios\n");
 		return false;
 	}
 
@@ -329,8 +334,8 @@ static bool amdgpu_atrm_get_bios(struct amdgpu_device *adev)
 			break;
 	}
 
-	if (!check_atom_bios(adev->bios, size)) {
-		kfree(adev->bios);
+	if (!check_atom_bios(adev, size)) {
+		amdgpu_bios_release(adev);
 		return false;
 	}
 	adev->bios_size = size;
@@ -345,10 +350,8 @@ static inline bool amdgpu_atrm_get_bios(struct amdgpu_device *adev)
 
 static bool amdgpu_read_disabled_bios(struct amdgpu_device *adev)
 {
-	if (adev->flags & AMD_IS_APU)
-		return igp_read_bios_from_vram(adev);
-	else
-		return amdgpu_asic_read_disabled_bios(adev);
+	return (!adev->asic_funcs || !adev->asic_funcs->read_disabled_bios) ?
+		false : amdgpu_asic_read_disabled_bios(adev);
 }
 
 #ifdef CONFIG_ACPI
@@ -357,13 +360,13 @@ static bool amdgpu_acpi_vfct_bios(struct amdgpu_device *adev)
 	struct acpi_table_header *hdr;
 	acpi_size tbl_size;
 	UEFI_ACPI_VFCT *vfct;
-	unsigned offset;
+	unsigned int offset;
 
 	if (!ACPI_SUCCESS(acpi_get_table("VFCT", 1, &hdr)))
 		return false;
 	tbl_size = hdr->length;
 	if (tbl_size < sizeof(UEFI_ACPI_VFCT)) {
-		DRM_ERROR("ACPI VFCT table present but broken (too short #1)\n");
+		dev_info(adev->dev, "ACPI VFCT table present but broken (too short #1),skipping\n");
 		return false;
 	}
 
@@ -376,13 +379,13 @@ static bool amdgpu_acpi_vfct_bios(struct amdgpu_device *adev)
 
 		offset += sizeof(VFCT_IMAGE_HEADER);
 		if (offset > tbl_size) {
-			DRM_ERROR("ACPI VFCT image header truncated\n");
+			dev_info(adev->dev, "ACPI VFCT image header truncated,skipping\n");
 			return false;
 		}
 
 		offset += vhdr->ImageLength;
 		if (offset > tbl_size) {
-			DRM_ERROR("ACPI VFCT image truncated\n");
+			dev_info(adev->dev, "ACPI VFCT image truncated,skipping\n");
 			return false;
 		}
 
@@ -396,8 +399,8 @@ static bool amdgpu_acpi_vfct_bios(struct amdgpu_device *adev)
 					     vhdr->ImageLength,
 					     GFP_KERNEL);
 
-			if (!check_atom_bios(adev->bios, vhdr->ImageLength)) {
-				kfree(adev->bios);
+			if (!check_atom_bios(adev, vhdr->ImageLength)) {
+				amdgpu_bios_release(adev);
 				return false;
 			}
 			adev->bios_size = vhdr->ImageLength;
@@ -405,7 +408,7 @@ static bool amdgpu_acpi_vfct_bios(struct amdgpu_device *adev)
 		}
 	}
 
-	DRM_ERROR("ACPI VFCT table present but broken (too short #2)\n");
+	dev_info(adev->dev, "ACPI VFCT table present but broken (too short #2),skipping\n");
 	return false;
 }
 #else
@@ -415,33 +418,158 @@ static inline bool amdgpu_acpi_vfct_bios(struct amdgpu_device *adev)
 }
 #endif
 
-bool amdgpu_get_bios(struct amdgpu_device *adev)
+static bool amdgpu_get_bios_apu(struct amdgpu_device *adev)
 {
-	if (amdgpu_atrm_get_bios(adev))
+	if (amdgpu_acpi_vfct_bios(adev)) {
+		dev_info(adev->dev, "Fetched VBIOS from VFCT\n");
 		goto success;
+	}
 
-	if (amdgpu_acpi_vfct_bios(adev))
+	if (amdgpu_read_bios_from_vram(adev)) {
+		dev_info(adev->dev, "Fetched VBIOS from VRAM BAR\n");
 		goto success;
+	}
 
-	if (igp_read_bios_from_vram(adev))
+	if (amdgpu_read_bios(adev)) {
+		dev_info(adev->dev, "Fetched VBIOS from ROM BAR\n");
 		goto success;
+	}
 
-	if (amdgpu_read_bios(adev))
+	if (amdgpu_read_platform_bios(adev)) {
+		dev_info(adev->dev, "Fetched VBIOS from platform\n");
 		goto success;
+	}
+
+	dev_err(adev->dev, "Unable to locate a BIOS ROM\n");
+	return false;
+
+success:
+	return true;
+}
+
+static bool amdgpu_prefer_rom_resource(struct amdgpu_device *adev)
+{
+	struct resource *res = &adev->pdev->resource[PCI_ROM_RESOURCE];
 
-	if (amdgpu_read_bios_from_rom(adev))
+	return (res->flags & IORESOURCE_ROM_SHADOW);
+}
+
+static bool amdgpu_get_bios_dgpu(struct amdgpu_device *adev)
+{
+	if (amdgpu_atrm_get_bios(adev)) {
+		dev_info(adev->dev, "Fetched VBIOS from ATRM\n");
 		goto success;
+	}
 
-	if (amdgpu_read_disabled_bios(adev))
+	if (amdgpu_acpi_vfct_bios(adev)) {
+		dev_info(adev->dev, "Fetched VBIOS from VFCT\n");
 		goto success;
+	}
 
-	if (amdgpu_read_platform_bios(adev))
+	/* this is required for SR-IOV */
+	if (amdgpu_read_bios_from_vram(adev)) {
+		dev_info(adev->dev, "Fetched VBIOS from VRAM BAR\n");
 		goto success;
+	}
+
+	if (amdgpu_prefer_rom_resource(adev)) {
+		if (amdgpu_read_bios(adev)) {
+			dev_info(adev->dev, "Fetched VBIOS from ROM BAR\n");
+			goto success;
+		}
+
+		if (amdgpu_read_platform_bios(adev)) {
+			dev_info(adev->dev, "Fetched VBIOS from platform\n");
+			goto success;
+		}
+
+	} else {
+		if (amdgpu_read_platform_bios(adev)) {
+			dev_info(adev->dev, "Fetched VBIOS from platform\n");
+			goto success;
+		}
 
-	DRM_ERROR("Unable to locate a BIOS ROM\n");
+		if (amdgpu_read_bios(adev)) {
+			dev_info(adev->dev, "Fetched VBIOS from ROM BAR\n");
+			goto success;
+		}
+	}
+
+	if (amdgpu_read_bios_from_rom(adev)) {
+		dev_info(adev->dev, "Fetched VBIOS from ROM\n");
+		goto success;
+	}
+
+	if (amdgpu_read_disabled_bios(adev)) {
+		dev_info(adev->dev, "Fetched VBIOS from disabled ROM BAR\n");
+		goto success;
+	}
+
+	dev_err(adev->dev, "Unable to locate a BIOS ROM\n");
 	return false;
 
 success:
-	adev->is_atom_fw = (adev->asic_type >= CHIP_VEGA10) ? true : false;
+	return true;
+}
+
+bool amdgpu_get_bios(struct amdgpu_device *adev)
+{
+	bool found;
+
+	if (adev->flags & AMD_IS_APU)
+		found = amdgpu_get_bios_apu(adev);
+	else
+		found = amdgpu_get_bios_dgpu(adev);
+
+	if (found)
+		adev->is_atom_fw = adev->asic_type >= CHIP_VEGA10;
+
+	return found;
+}
+
+/* helper function for soc15 and onwards to read bios from rom */
+bool amdgpu_soc15_read_bios_from_rom(struct amdgpu_device *adev,
+				     u8 *bios, u32 length_bytes)
+{
+	u32 *dw_ptr;
+	u32 i, length_dw;
+	u32 rom_offset;
+	u32 rom_index_offset;
+	u32 rom_data_offset;
+
+	if (bios == NULL)
+		return false;
+	if (length_bytes == 0)
+		return false;
+	/* APU vbios image is part of sbios image */
+	if (adev->flags & AMD_IS_APU)
+		return false;
+	if (!adev->smuio.funcs ||
+	    !adev->smuio.funcs->get_rom_index_offset ||
+	    !adev->smuio.funcs->get_rom_data_offset)
+		return false;
+
+	dw_ptr = (u32 *)bios;
+	length_dw = ALIGN(length_bytes, 4) / 4;
+
+	rom_index_offset =
+		adev->smuio.funcs->get_rom_index_offset(adev);
+	rom_data_offset =
+		adev->smuio.funcs->get_rom_data_offset(adev);
+
+	if (adev->nbio.funcs &&
+	    adev->nbio.funcs->get_rom_offset) {
+		rom_offset = adev->nbio.funcs->get_rom_offset(adev);
+		rom_offset = rom_offset << 17;
+	} else {
+		rom_offset = 0;
+	}
+
+	/* set rom index to rom_offset */
+	WREG32(rom_index_offset, rom_offset);
+	/* read out the rom data */
+	for (i = 0; i < length_dw; i++)
+		dw_ptr[i] = RREG32(rom_data_offset);
+
 	return true;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
index 59089e027f4d..66fb37b64388 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
@@ -28,100 +28,64 @@
  *    Christian König <deathsimple@vodafone.de>
  */
 
-#include <drm/drmP.h>
+#include <linux/sort.h>
+#include <linux/uaccess.h>
+
 #include "amdgpu.h"
 #include "amdgpu_trace.h"
 
 #define AMDGPU_BO_LIST_MAX_PRIORITY	32u
 #define AMDGPU_BO_LIST_NUM_BUCKETS	(AMDGPU_BO_LIST_MAX_PRIORITY + 1)
 
-static int amdgpu_bo_list_set(struct amdgpu_device *adev,
-				     struct drm_file *filp,
-				     struct amdgpu_bo_list *list,
-				     struct drm_amdgpu_bo_list_entry *info,
-				     unsigned num_entries);
+static void amdgpu_bo_list_free_rcu(struct rcu_head *rcu)
+{
+	struct amdgpu_bo_list *list = container_of(rcu, struct amdgpu_bo_list,
+						   rhead);
+	mutex_destroy(&list->bo_list_mutex);
+	kvfree(list);
+}
 
-static void amdgpu_bo_list_release_rcu(struct kref *ref)
+static void amdgpu_bo_list_free(struct kref *ref)
 {
-	unsigned i;
 	struct amdgpu_bo_list *list = container_of(ref, struct amdgpu_bo_list,
 						   refcount);
+	struct amdgpu_bo_list_entry *e;
 
-	for (i = 0; i < list->num_entries; ++i)
-		amdgpu_bo_unref(&list->array[i].robj);
-
-	mutex_destroy(&list->lock);
-	kvfree(list->array);
-	kfree_rcu(list, rhead);
+	amdgpu_bo_list_for_each_entry(e, list)
+		amdgpu_bo_unref(&e->bo);
+	call_rcu(&list->rhead, amdgpu_bo_list_free_rcu);
 }
 
-static int amdgpu_bo_list_create(struct amdgpu_device *adev,
-				 struct drm_file *filp,
-				 struct drm_amdgpu_bo_list_entry *info,
-				 unsigned num_entries,
-				 int *id)
+static int amdgpu_bo_list_entry_cmp(const void *_a, const void *_b)
 {
-	int r;
-	struct amdgpu_fpriv *fpriv = filp->driver_priv;
-	struct amdgpu_bo_list *list;
-
-	list = kzalloc(sizeof(struct amdgpu_bo_list), GFP_KERNEL);
-	if (!list)
-		return -ENOMEM;
-
-	/* initialize bo list*/
-	mutex_init(&list->lock);
-	kref_init(&list->refcount);
-	r = amdgpu_bo_list_set(adev, filp, list, info, num_entries);
-	if (r) {
-		kfree(list);
-		return r;
-	}
-
-	/* idr alloc should be called only after initialization of bo list. */
-	mutex_lock(&fpriv->bo_list_lock);
-	r = idr_alloc(&fpriv->bo_list_handles, list, 1, 0, GFP_KERNEL);
-	mutex_unlock(&fpriv->bo_list_lock);
-	if (r < 0) {
-		amdgpu_bo_list_free(list);
-		return r;
-	}
-	*id = r;
+	const struct amdgpu_bo_list_entry *a = _a, *b = _b;
 
+	if (a->priority > b->priority)
+		return 1;
+	if (a->priority < b->priority)
+		return -1;
 	return 0;
 }
 
-static void amdgpu_bo_list_destroy(struct amdgpu_fpriv *fpriv, int id)
+int amdgpu_bo_list_create(struct amdgpu_device *adev, struct drm_file *filp,
+			  struct drm_amdgpu_bo_list_entry *info,
+			  size_t num_entries, struct amdgpu_bo_list **result)
 {
-	struct amdgpu_bo_list *list;
-
-	mutex_lock(&fpriv->bo_list_lock);
-	list = idr_remove(&fpriv->bo_list_handles, id);
-	mutex_unlock(&fpriv->bo_list_lock);
-	if (list)
-		kref_put(&list->refcount, amdgpu_bo_list_release_rcu);
-}
-
-static int amdgpu_bo_list_set(struct amdgpu_device *adev,
-				     struct drm_file *filp,
-				     struct amdgpu_bo_list *list,
-				     struct drm_amdgpu_bo_list_entry *info,
-				     unsigned num_entries)
-{
-	struct amdgpu_bo_list_entry *array;
-	struct amdgpu_bo *gds_obj = adev->gds.gds_gfx_bo;
-	struct amdgpu_bo *gws_obj = adev->gds.gws_gfx_bo;
-	struct amdgpu_bo *oa_obj = adev->gds.oa_gfx_bo;
-
 	unsigned last_entry = 0, first_userptr = num_entries;
+	struct amdgpu_bo_list_entry *array;
+	struct amdgpu_bo_list *list;
+	uint64_t total_size = 0;
 	unsigned i;
 	int r;
-	unsigned long total_size = 0;
 
-	array = kvmalloc_array(num_entries, sizeof(struct amdgpu_bo_list_entry), GFP_KERNEL);
-	if (!array)
+	list = kvzalloc(struct_size(list, entries, num_entries), GFP_KERNEL);
+	if (!list)
 		return -ENOMEM;
-	memset(array, 0, num_entries * sizeof(struct amdgpu_bo_list_entry));
+
+	kref_init(&list->refcount);
+
+	list->num_entries = num_entries;
+	array = list->entries;
 
 	for (i = 0; i < num_entries; ++i) {
 		struct amdgpu_bo_list_entry *entry;
@@ -136,7 +100,7 @@ static int amdgpu_bo_list_set(struct amdgpu_device *adev,
 		}
 
 		bo = amdgpu_bo_ref(gem_to_amdgpu_bo(gobj));
-		drm_gem_object_put_unlocked(gobj);
+		drm_gem_object_put(gobj);
 
 		usermm = amdgpu_ttm_tt_get_usermm(bo->tbo.ttm);
 		if (usermm) {
@@ -150,166 +114,138 @@ static int amdgpu_bo_list_set(struct amdgpu_device *adev,
 			entry = &array[last_entry++];
 		}
 
-		entry->robj = bo;
 		entry->priority = min(info[i].bo_priority,
 				      AMDGPU_BO_LIST_MAX_PRIORITY);
-		entry->tv.bo = &entry->robj->tbo;
-		entry->tv.shared = !entry->robj->prime_shared_count;
-
-		if (entry->robj->preferred_domains == AMDGPU_GEM_DOMAIN_GDS)
-			gds_obj = entry->robj;
-		if (entry->robj->preferred_domains == AMDGPU_GEM_DOMAIN_GWS)
-			gws_obj = entry->robj;
-		if (entry->robj->preferred_domains == AMDGPU_GEM_DOMAIN_OA)
-			oa_obj = entry->robj;
-
-		total_size += amdgpu_bo_size(entry->robj);
-		trace_amdgpu_bo_list_set(list, entry->robj);
-	}
+		entry->bo = bo;
 
-	for (i = 0; i < list->num_entries; ++i)
-		amdgpu_bo_unref(&list->array[i].robj);
+		if (bo->preferred_domains == AMDGPU_GEM_DOMAIN_GDS)
+			list->gds_obj = bo;
+		if (bo->preferred_domains == AMDGPU_GEM_DOMAIN_GWS)
+			list->gws_obj = bo;
+		if (bo->preferred_domains == AMDGPU_GEM_DOMAIN_OA)
+			list->oa_obj = bo;
 
-	kvfree(list->array);
+		total_size += amdgpu_bo_size(bo);
+		trace_amdgpu_bo_list_set(list, bo);
+	}
 
-	list->gds_obj = gds_obj;
-	list->gws_obj = gws_obj;
-	list->oa_obj = oa_obj;
 	list->first_userptr = first_userptr;
-	list->array = array;
-	list->num_entries = num_entries;
+	sort(array, last_entry, sizeof(struct amdgpu_bo_list_entry),
+	     amdgpu_bo_list_entry_cmp, NULL);
 
 	trace_amdgpu_cs_bo_status(list->num_entries, total_size);
+
+	mutex_init(&list->bo_list_mutex);
+	*result = list;
 	return 0;
 
 error_free:
-	while (i--)
-		amdgpu_bo_unref(&array[i].robj);
-	kvfree(array);
+	for (i = 0; i < last_entry; ++i)
+		amdgpu_bo_unref(&array[i].bo);
+	for (i = first_userptr; i < num_entries; ++i)
+		amdgpu_bo_unref(&array[i].bo);
+	kvfree(list);
 	return r;
+
 }
 
-struct amdgpu_bo_list *
-amdgpu_bo_list_get(struct amdgpu_fpriv *fpriv, int id)
+static void amdgpu_bo_list_destroy(struct amdgpu_fpriv *fpriv, int id)
 {
-	struct amdgpu_bo_list *result;
-
-	rcu_read_lock();
-	result = idr_find(&fpriv->bo_list_handles, id);
-
-	if (result) {
-		if (kref_get_unless_zero(&result->refcount)) {
-			rcu_read_unlock();
-			mutex_lock(&result->lock);
-		} else {
-			rcu_read_unlock();
-			result = NULL;
-		}
-	} else {
-		rcu_read_unlock();
-	}
+	struct amdgpu_bo_list *list;
 
-	return result;
+	mutex_lock(&fpriv->bo_list_lock);
+	list = idr_remove(&fpriv->bo_list_handles, id);
+	mutex_unlock(&fpriv->bo_list_lock);
+	if (list)
+		kref_put(&list->refcount, amdgpu_bo_list_free);
 }
 
-void amdgpu_bo_list_get_list(struct amdgpu_bo_list *list,
-			     struct list_head *validated)
+int amdgpu_bo_list_get(struct amdgpu_fpriv *fpriv, int id,
+		       struct amdgpu_bo_list **result)
 {
-	/* This is based on the bucket sort with O(n) time complexity.
-	 * An item with priority "i" is added to bucket[i]. The lists are then
-	 * concatenated in descending order.
-	 */
-	struct list_head bucket[AMDGPU_BO_LIST_NUM_BUCKETS];
-	unsigned i;
-
-	for (i = 0; i < AMDGPU_BO_LIST_NUM_BUCKETS; i++)
-		INIT_LIST_HEAD(&bucket[i]);
-
-	/* Since buffers which appear sooner in the relocation list are
-	 * likely to be used more often than buffers which appear later
-	 * in the list, the sort mustn't change the ordering of buffers
-	 * with the same priority, i.e. it must be stable.
-	 */
-	for (i = 0; i < list->num_entries; i++) {
-		unsigned priority = list->array[i].priority;
+	rcu_read_lock();
+	*result = idr_find(&fpriv->bo_list_handles, id);
 
-		list_add_tail(&list->array[i].tv.head,
-			      &bucket[priority]);
-		list->array[i].user_pages = NULL;
+	if (*result && kref_get_unless_zero(&(*result)->refcount)) {
+		rcu_read_unlock();
+		return 0;
 	}
 
-	/* Connect the sorted buckets in the output list. */
-	for (i = 0; i < AMDGPU_BO_LIST_NUM_BUCKETS; i++)
-		list_splice(&bucket[i], validated);
+	rcu_read_unlock();
+	*result = NULL;
+	return -ENOENT;
 }
 
 void amdgpu_bo_list_put(struct amdgpu_bo_list *list)
 {
-	mutex_unlock(&list->lock);
-	kref_put(&list->refcount, amdgpu_bo_list_release_rcu);
+	kref_put(&list->refcount, amdgpu_bo_list_free);
 }
 
-void amdgpu_bo_list_free(struct amdgpu_bo_list *list)
+int amdgpu_bo_create_list_entry_array(struct drm_amdgpu_bo_list_in *in,
+				      struct drm_amdgpu_bo_list_entry **info_param)
 {
-	unsigned i;
+	const uint32_t info_size = sizeof(struct drm_amdgpu_bo_list_entry);
+	const void __user *uptr = u64_to_user_ptr(in->bo_info_ptr);
+	const uint32_t bo_info_size = in->bo_info_size;
+	const uint32_t bo_number = in->bo_number;
+	struct drm_amdgpu_bo_list_entry *info;
 
-	for (i = 0; i < list->num_entries; ++i)
-		amdgpu_bo_unref(&list->array[i].robj);
+	/* copy the handle array from userspace to a kernel buffer */
+	if (likely(info_size == bo_info_size)) {
+		info = vmemdup_array_user(uptr, bo_number, info_size);
+		if (IS_ERR(info))
+			return PTR_ERR(info);
+	} else {
+		const uint32_t bytes = min(bo_info_size, info_size);
+		unsigned i;
+
+		info = kvmalloc_array(bo_number, info_size, GFP_KERNEL);
+		if (!info)
+			return -ENOMEM;
 
-	mutex_destroy(&list->lock);
-	kvfree(list->array);
-	kfree(list);
+		memset(info, 0, bo_number * info_size);
+		for (i = 0; i < bo_number; ++i, uptr += bo_info_size) {
+			if (copy_from_user(&info[i], uptr, bytes)) {
+				kvfree(info);
+				return -EFAULT;
+			}
+		}
+	}
+
+	*info_param = info;
+	return 0;
 }
 
 int amdgpu_bo_list_ioctl(struct drm_device *dev, void *data,
 				struct drm_file *filp)
 {
-	const uint32_t info_size = sizeof(struct drm_amdgpu_bo_list_entry);
-
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_fpriv *fpriv = filp->driver_priv;
 	union drm_amdgpu_bo_list *args = data;
 	uint32_t handle = args->in.list_handle;
-	const void __user *uptr = u64_to_user_ptr(args->in.bo_info_ptr);
-
-	struct drm_amdgpu_bo_list_entry *info;
-	struct amdgpu_bo_list *list;
-
+	struct drm_amdgpu_bo_list_entry *info = NULL;
+	struct amdgpu_bo_list *list, *old;
 	int r;
 
-	info = kvmalloc_array(args->in.bo_number,
-			     sizeof(struct drm_amdgpu_bo_list_entry), GFP_KERNEL);
-	if (!info)
-		return -ENOMEM;
-
-	/* copy the handle array from userspace to a kernel buffer */
-	r = -EFAULT;
-	if (likely(info_size == args->in.bo_info_size)) {
-		unsigned long bytes = args->in.bo_number *
-			args->in.bo_info_size;
-
-		if (copy_from_user(info, uptr, bytes))
-			goto error_free;
-
-	} else {
-		unsigned long bytes = min(args->in.bo_info_size, info_size);
-		unsigned i;
-
-		memset(info, 0, args->in.bo_number * info_size);
-		for (i = 0; i < args->in.bo_number; ++i) {
-			if (copy_from_user(&info[i], uptr, bytes))
-				goto error_free;
-
-			uptr += args->in.bo_info_size;
-		}
-	}
+	r = amdgpu_bo_create_list_entry_array(&args->in, &info);
+	if (r)
+		return r;
 
 	switch (args->in.operation) {
 	case AMDGPU_BO_LIST_OP_CREATE:
 		r = amdgpu_bo_list_create(adev, filp, info, args->in.bo_number,
-					  &handle);
+					  &list);
 		if (r)
 			goto error_free;
+
+		mutex_lock(&fpriv->bo_list_lock);
+		r = idr_alloc(&fpriv->bo_list_handles, list, 1, 0, GFP_KERNEL);
+		mutex_unlock(&fpriv->bo_list_lock);
+		if (r < 0) {
+			goto error_put_list;
+		}
+
+		handle = r;
 		break;
 
 	case AMDGPU_BO_LIST_OP_DESTROY:
@@ -318,17 +254,21 @@ int amdgpu_bo_list_ioctl(struct drm_device *dev, void *data,
 		break;
 
 	case AMDGPU_BO_LIST_OP_UPDATE:
-		r = -ENOENT;
-		list = amdgpu_bo_list_get(fpriv, handle);
-		if (!list)
-			goto error_free;
-
-		r = amdgpu_bo_list_set(adev, filp, list, info,
-					      args->in.bo_number);
-		amdgpu_bo_list_put(list);
+		r = amdgpu_bo_list_create(adev, filp, info, args->in.bo_number,
+					  &list);
 		if (r)
 			goto error_free;
 
+		mutex_lock(&fpriv->bo_list_lock);
+		old = idr_replace(&fpriv->bo_list_handles, list, handle);
+		mutex_unlock(&fpriv->bo_list_lock);
+
+		if (IS_ERR(old)) {
+			r = PTR_ERR(old);
+			goto error_put_list;
+		}
+
+		amdgpu_bo_list_put(old);
 		break;
 
 	default:
@@ -342,6 +282,9 @@ int amdgpu_bo_list_ioctl(struct drm_device *dev, void *data,
 
 	return 0;
 
+error_put_list:
+	amdgpu_bo_list_put(list);
+
 error_free:
 	kvfree(info);
 	return r;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h
new file mode 100644
index 000000000000..a716c9886c74
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h
@@ -0,0 +1,83 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __AMDGPU_BO_LIST_H__
+#define __AMDGPU_BO_LIST_H__
+
+#include <drm/amdgpu_drm.h>
+
+struct hmm_range;
+
+struct drm_file;
+
+struct amdgpu_device;
+struct amdgpu_bo;
+struct amdgpu_bo_va;
+struct amdgpu_fpriv;
+
+struct amdgpu_bo_list_entry {
+	struct amdgpu_bo		*bo;
+	struct amdgpu_bo_va		*bo_va;
+	uint32_t			priority;
+	struct hmm_range		*range;
+	bool				user_invalidated;
+};
+
+struct amdgpu_bo_list {
+	struct rcu_head rhead;
+	struct kref refcount;
+	struct amdgpu_bo *gds_obj;
+	struct amdgpu_bo *gws_obj;
+	struct amdgpu_bo *oa_obj;
+	unsigned first_userptr;
+	unsigned num_entries;
+
+	/* Protect access during command submission.
+	 */
+	struct mutex bo_list_mutex;
+
+	struct amdgpu_bo_list_entry entries[] __counted_by(num_entries);
+};
+
+int amdgpu_bo_list_get(struct amdgpu_fpriv *fpriv, int id,
+		       struct amdgpu_bo_list **result);
+void amdgpu_bo_list_put(struct amdgpu_bo_list *list);
+int amdgpu_bo_create_list_entry_array(struct drm_amdgpu_bo_list_in *in,
+				      struct drm_amdgpu_bo_list_entry **info_param);
+
+int amdgpu_bo_list_create(struct amdgpu_device *adev,
+				 struct drm_file *filp,
+				 struct drm_amdgpu_bo_list_entry *info,
+				 size_t num_entries,
+				 struct amdgpu_bo_list **list);
+
+#define amdgpu_bo_list_for_each_entry(e, list) \
+	for (e = list->entries; \
+	     e != &list->entries[list->num_entries]; \
+	     ++e)
+
+#define amdgpu_bo_list_for_each_userptr_entry(e, list) \
+	for (e = &list->entries[list->first_userptr]; \
+	     e != &list->entries[list->num_entries]; \
+	     ++e)
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c
index fd435a96481c..004a6a9d6b9f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c
@@ -22,14 +22,12 @@
  *
  */
 #include <linux/list.h>
-#include <linux/slab.h>
 #include <linux/pci.h>
-#include <linux/acpi.h>
-#include <drm/drmP.h>
+#include <linux/slab.h>
+
 #include <linux/firmware.h>
 #include <drm/amdgpu_drm.h>
 #include "amdgpu.h"
-#include "cgs_linux.h"
 #include "atom.h"
 #include "amdgpu_ucode.h"
 
@@ -42,180 +40,14 @@ struct amdgpu_cgs_device {
 	struct amdgpu_device *adev =					\
 		((struct amdgpu_cgs_device *)cgs_device)->adev
 
-static int amdgpu_cgs_alloc_gpu_mem(struct cgs_device *cgs_device,
-				    enum cgs_gpu_mem_type type,
-				    uint64_t size, uint64_t align,
-				    uint64_t min_offset, uint64_t max_offset,
-				    cgs_handle_t *handle)
-{
-	CGS_FUNC_ADEV;
-	uint16_t flags = 0;
-	int ret = 0;
-	uint32_t domain = 0;
-	struct amdgpu_bo *obj;
-	struct ttm_placement placement;
-	struct ttm_place place;
-
-	if (min_offset > max_offset) {
-		BUG_ON(1);
-		return -EINVAL;
-	}
-
-	/* fail if the alignment is not a power of 2 */
-	if (((align != 1) && (align & (align - 1)))
-	    || size == 0 || align == 0)
-		return -EINVAL;
-
-
-	switch(type) {
-	case CGS_GPU_MEM_TYPE__VISIBLE_CONTIG_FB:
-	case CGS_GPU_MEM_TYPE__VISIBLE_FB:
-		flags = AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
-			AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
-		domain = AMDGPU_GEM_DOMAIN_VRAM;
-		if (max_offset > adev->mc.real_vram_size)
-			return -EINVAL;
-		place.fpfn = min_offset >> PAGE_SHIFT;
-		place.lpfn = max_offset >> PAGE_SHIFT;
-		place.flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_UNCACHED |
-			TTM_PL_FLAG_VRAM;
-		break;
-	case CGS_GPU_MEM_TYPE__INVISIBLE_CONTIG_FB:
-	case CGS_GPU_MEM_TYPE__INVISIBLE_FB:
-		flags = AMDGPU_GEM_CREATE_NO_CPU_ACCESS |
-			AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
-		domain = AMDGPU_GEM_DOMAIN_VRAM;
-		if (adev->mc.visible_vram_size < adev->mc.real_vram_size) {
-			place.fpfn =
-				max(min_offset, adev->mc.visible_vram_size) >> PAGE_SHIFT;
-			place.lpfn =
-				min(max_offset, adev->mc.real_vram_size) >> PAGE_SHIFT;
-			place.flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_UNCACHED |
-				TTM_PL_FLAG_VRAM;
-		}
-
-		break;
-	case CGS_GPU_MEM_TYPE__GART_CACHEABLE:
-		domain = AMDGPU_GEM_DOMAIN_GTT;
-		place.fpfn = min_offset >> PAGE_SHIFT;
-		place.lpfn = max_offset >> PAGE_SHIFT;
-		place.flags = TTM_PL_FLAG_CACHED | TTM_PL_FLAG_TT;
-		break;
-	case CGS_GPU_MEM_TYPE__GART_WRITECOMBINE:
-		flags = AMDGPU_GEM_CREATE_CPU_GTT_USWC;
-		domain = AMDGPU_GEM_DOMAIN_GTT;
-		place.fpfn = min_offset >> PAGE_SHIFT;
-		place.lpfn = max_offset >> PAGE_SHIFT;
-		place.flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_TT |
-			TTM_PL_FLAG_UNCACHED;
-		break;
-	default:
-		return -EINVAL;
-	}
-
-
-	*handle = 0;
-
-	placement.placement = &place;
-	placement.num_placement = 1;
-	placement.busy_placement = &place;
-	placement.num_busy_placement = 1;
-
-	ret = amdgpu_bo_create_restricted(adev, size, PAGE_SIZE,
-					  true, domain, flags,
-					  NULL, &placement, NULL,
-					  0, &obj);
-	if (ret) {
-		DRM_ERROR("(%d) bo create failed\n", ret);
-		return ret;
-	}
-	*handle = (cgs_handle_t)obj;
-
-	return ret;
-}
 
-static int amdgpu_cgs_free_gpu_mem(struct cgs_device *cgs_device, cgs_handle_t handle)
-{
-	struct amdgpu_bo *obj = (struct amdgpu_bo *)handle;
-
-	if (obj) {
-		int r = amdgpu_bo_reserve(obj, true);
-		if (likely(r == 0)) {
-			amdgpu_bo_kunmap(obj);
-			amdgpu_bo_unpin(obj);
-			amdgpu_bo_unreserve(obj);
-		}
-		amdgpu_bo_unref(&obj);
-
-	}
-	return 0;
-}
-
-static int amdgpu_cgs_gmap_gpu_mem(struct cgs_device *cgs_device, cgs_handle_t handle,
-				   uint64_t *mcaddr)
-{
-	int r;
-	u64 min_offset, max_offset;
-	struct amdgpu_bo *obj = (struct amdgpu_bo *)handle;
-
-	WARN_ON_ONCE(obj->placement.num_placement > 1);
-
-	min_offset = obj->placements[0].fpfn << PAGE_SHIFT;
-	max_offset = obj->placements[0].lpfn << PAGE_SHIFT;
-
-	r = amdgpu_bo_reserve(obj, true);
-	if (unlikely(r != 0))
-		return r;
-	r = amdgpu_bo_pin_restricted(obj, obj->preferred_domains,
-				     min_offset, max_offset, mcaddr);
-	amdgpu_bo_unreserve(obj);
-	return r;
-}
-
-static int amdgpu_cgs_gunmap_gpu_mem(struct cgs_device *cgs_device, cgs_handle_t handle)
-{
-	int r;
-	struct amdgpu_bo *obj = (struct amdgpu_bo *)handle;
-	r = amdgpu_bo_reserve(obj, true);
-	if (unlikely(r != 0))
-		return r;
-	r = amdgpu_bo_unpin(obj);
-	amdgpu_bo_unreserve(obj);
-	return r;
-}
-
-static int amdgpu_cgs_kmap_gpu_mem(struct cgs_device *cgs_device, cgs_handle_t handle,
-				   void **map)
-{
-	int r;
-	struct amdgpu_bo *obj = (struct amdgpu_bo *)handle;
-	r = amdgpu_bo_reserve(obj, true);
-	if (unlikely(r != 0))
-		return r;
-	r = amdgpu_bo_kmap(obj, map);
-	amdgpu_bo_unreserve(obj);
-	return r;
-}
-
-static int amdgpu_cgs_kunmap_gpu_mem(struct cgs_device *cgs_device, cgs_handle_t handle)
-{
-	int r;
-	struct amdgpu_bo *obj = (struct amdgpu_bo *)handle;
-	r = amdgpu_bo_reserve(obj, true);
-	if (unlikely(r != 0))
-		return r;
-	amdgpu_bo_kunmap(obj);
-	amdgpu_bo_unreserve(obj);
-	return r;
-}
-
-static uint32_t amdgpu_cgs_read_register(struct cgs_device *cgs_device, unsigned offset)
+static uint32_t amdgpu_cgs_read_register(struct cgs_device *cgs_device, unsigned int offset)
 {
 	CGS_FUNC_ADEV;
 	return RREG32(offset);
 }
 
-static void amdgpu_cgs_write_register(struct cgs_device *cgs_device, unsigned offset,
+static void amdgpu_cgs_write_register(struct cgs_device *cgs_device, unsigned int offset,
 				      uint32_t value)
 {
 	CGS_FUNC_ADEV;
@@ -224,12 +56,10 @@ static void amdgpu_cgs_write_register(struct cgs_device *cgs_device, unsigned of
 
 static uint32_t amdgpu_cgs_read_ind_register(struct cgs_device *cgs_device,
 					     enum cgs_ind_reg space,
-					     unsigned index)
+					     unsigned int index)
 {
 	CGS_FUNC_ADEV;
 	switch (space) {
-	case CGS_IND_REG__MMIO:
-		return RREG32_IDX(index);
 	case CGS_IND_REG__PCIE:
 		return RREG32_PCIE(index);
 	case CGS_IND_REG__SMC:
@@ -245,6 +75,8 @@ static uint32_t amdgpu_cgs_read_ind_register(struct cgs_device *cgs_device,
 	case CGS_IND_REG__AUDIO_ENDPT:
 		DRM_ERROR("audio endpt register access not implemented.\n");
 		return 0;
+	default:
+		BUG();
 	}
 	WARN(1, "Invalid indirect register space");
 	return 0;
@@ -252,12 +84,10 @@ static uint32_t amdgpu_cgs_read_ind_register(struct cgs_device *cgs_device,
 
 static void amdgpu_cgs_write_ind_register(struct cgs_device *cgs_device,
 					  enum cgs_ind_reg space,
-					  unsigned index, uint32_t value)
+					  unsigned int index, uint32_t value)
 {
 	CGS_FUNC_ADEV;
 	switch (space) {
-	case CGS_IND_REG__MMIO:
-		return WREG32_IDX(index, value);
 	case CGS_IND_REG__PCIE:
 		return WREG32_PCIE(index, value);
 	case CGS_IND_REG__SMC:
@@ -273,228 +103,12 @@ static void amdgpu_cgs_write_ind_register(struct cgs_device *cgs_device,
 	case CGS_IND_REG__AUDIO_ENDPT:
 		DRM_ERROR("audio endpt register access not implemented.\n");
 		return;
-	}
-	WARN(1, "Invalid indirect register space");
-}
-
-static int amdgpu_cgs_get_pci_resource(struct cgs_device *cgs_device,
-				       enum cgs_resource_type resource_type,
-				       uint64_t size,
-				       uint64_t offset,
-				       uint64_t *resource_base)
-{
-	CGS_FUNC_ADEV;
-
-	if (resource_base == NULL)
-		return -EINVAL;
-
-	switch (resource_type) {
-	case CGS_RESOURCE_TYPE_MMIO:
-		if (adev->rmmio_size == 0)
-			return -ENOENT;
-		if ((offset + size) > adev->rmmio_size)
-			return -EINVAL;
-		*resource_base = adev->rmmio_base;
-		return 0;
-	case CGS_RESOURCE_TYPE_DOORBELL:
-		if (adev->doorbell.size == 0)
-			return -ENOENT;
-		if ((offset + size) > adev->doorbell.size)
-			return -EINVAL;
-		*resource_base = adev->doorbell.base;
-		return 0;
-	case CGS_RESOURCE_TYPE_FB:
-	case CGS_RESOURCE_TYPE_IO:
-	case CGS_RESOURCE_TYPE_ROM:
 	default:
-		return -EINVAL;
+		BUG();
 	}
+	WARN(1, "Invalid indirect register space");
 }
 
-static const void *amdgpu_cgs_atom_get_data_table(struct cgs_device *cgs_device,
-						  unsigned table, uint16_t *size,
-						  uint8_t *frev, uint8_t *crev)
-{
-	CGS_FUNC_ADEV;
-	uint16_t data_start;
-
-	if (amdgpu_atom_parse_data_header(
-		    adev->mode_info.atom_context, table, size,
-		    frev, crev, &data_start))
-		return (uint8_t*)adev->mode_info.atom_context->bios +
-			data_start;
-
-	return NULL;
-}
-
-static int amdgpu_cgs_atom_get_cmd_table_revs(struct cgs_device *cgs_device, unsigned table,
-					      uint8_t *frev, uint8_t *crev)
-{
-	CGS_FUNC_ADEV;
-
-	if (amdgpu_atom_parse_cmd_header(
-		    adev->mode_info.atom_context, table,
-		    frev, crev))
-		return 0;
-
-	return -EINVAL;
-}
-
-static int amdgpu_cgs_atom_exec_cmd_table(struct cgs_device *cgs_device, unsigned table,
-					  void *args)
-{
-	CGS_FUNC_ADEV;
-
-	return amdgpu_atom_execute_table(
-		adev->mode_info.atom_context, table, args);
-}
-
-struct cgs_irq_params {
-	unsigned src_id;
-	cgs_irq_source_set_func_t set;
-	cgs_irq_handler_func_t handler;
-	void *private_data;
-};
-
-static int cgs_set_irq_state(struct amdgpu_device *adev,
-			     struct amdgpu_irq_src *src,
-			     unsigned type,
-			     enum amdgpu_interrupt_state state)
-{
-	struct cgs_irq_params *irq_params =
-		(struct cgs_irq_params *)src->data;
-	if (!irq_params)
-		return -EINVAL;
-	if (!irq_params->set)
-		return -EINVAL;
-	return irq_params->set(irq_params->private_data,
-			       irq_params->src_id,
-			       type,
-			       (int)state);
-}
-
-static int cgs_process_irq(struct amdgpu_device *adev,
-			   struct amdgpu_irq_src *source,
-			   struct amdgpu_iv_entry *entry)
-{
-	struct cgs_irq_params *irq_params =
-		(struct cgs_irq_params *)source->data;
-	if (!irq_params)
-		return -EINVAL;
-	if (!irq_params->handler)
-		return -EINVAL;
-	return irq_params->handler(irq_params->private_data,
-				   irq_params->src_id,
-				   entry->iv_entry);
-}
-
-static const struct amdgpu_irq_src_funcs cgs_irq_funcs = {
-	.set = cgs_set_irq_state,
-	.process = cgs_process_irq,
-};
-
-static int amdgpu_cgs_add_irq_source(void *cgs_device,
-				     unsigned client_id,
-				     unsigned src_id,
-				     unsigned num_types,
-				     cgs_irq_source_set_func_t set,
-				     cgs_irq_handler_func_t handler,
-				     void *private_data)
-{
-	CGS_FUNC_ADEV;
-	int ret = 0;
-	struct cgs_irq_params *irq_params;
-	struct amdgpu_irq_src *source =
-		kzalloc(sizeof(struct amdgpu_irq_src), GFP_KERNEL);
-	if (!source)
-		return -ENOMEM;
-	irq_params =
-		kzalloc(sizeof(struct cgs_irq_params), GFP_KERNEL);
-	if (!irq_params) {
-		kfree(source);
-		return -ENOMEM;
-	}
-	source->num_types = num_types;
-	source->funcs = &cgs_irq_funcs;
-	irq_params->src_id = src_id;
-	irq_params->set = set;
-	irq_params->handler = handler;
-	irq_params->private_data = private_data;
-	source->data = (void *)irq_params;
-	ret = amdgpu_irq_add_id(adev, client_id, src_id, source);
-	if (ret) {
-		kfree(irq_params);
-		kfree(source);
-	}
-
-	return ret;
-}
-
-static int amdgpu_cgs_irq_get(void *cgs_device, unsigned client_id,
-			      unsigned src_id, unsigned type)
-{
-	CGS_FUNC_ADEV;
-
-	if (!adev->irq.client[client_id].sources)
-		return -EINVAL;
-
-	return amdgpu_irq_get(adev, adev->irq.client[client_id].sources[src_id], type);
-}
-
-static int amdgpu_cgs_irq_put(void *cgs_device, unsigned client_id,
-			      unsigned src_id, unsigned type)
-{
-	CGS_FUNC_ADEV;
-
-	if (!adev->irq.client[client_id].sources)
-		return -EINVAL;
-
-	return amdgpu_irq_put(adev, adev->irq.client[client_id].sources[src_id], type);
-}
-
-static int amdgpu_cgs_set_clockgating_state(struct cgs_device *cgs_device,
-				  enum amd_ip_block_type block_type,
-				  enum amd_clockgating_state state)
-{
-	CGS_FUNC_ADEV;
-	int i, r = -1;
-
-	for (i = 0; i < adev->num_ip_blocks; i++) {
-		if (!adev->ip_blocks[i].status.valid)
-			continue;
-
-		if (adev->ip_blocks[i].version->type == block_type) {
-			r = adev->ip_blocks[i].version->funcs->set_clockgating_state(
-								(void *)adev,
-									state);
-			break;
-		}
-	}
-	return r;
-}
-
-static int amdgpu_cgs_set_powergating_state(struct cgs_device *cgs_device,
-				  enum amd_ip_block_type block_type,
-				  enum amd_powergating_state state)
-{
-	CGS_FUNC_ADEV;
-	int i, r = -1;
-
-	for (i = 0; i < adev->num_ip_blocks; i++) {
-		if (!adev->ip_blocks[i].status.valid)
-			continue;
-
-		if (adev->ip_blocks[i].version->type == block_type) {
-			r = adev->ip_blocks[i].version->funcs->set_powergating_state(
-								(void *)adev,
-									state);
-			break;
-		}
-	}
-	return r;
-}
-
-
 static uint32_t fw_type_convert(struct cgs_device *cgs_device, uint32_t fw_type)
 {
 	CGS_FUNC_ADEV;
@@ -542,18 +156,6 @@ static uint32_t fw_type_convert(struct cgs_device *cgs_device, uint32_t fw_type)
 	return result;
 }
 
-static int amdgpu_cgs_rel_firmware(struct cgs_device *cgs_device, enum cgs_ucode_id type)
-{
-	CGS_FUNC_ADEV;
-	if ((CGS_UCODE_ID_SMU == type) || (CGS_UCODE_ID_SMU_SK == type)) {
-		release_firmware(adev->pm.fw);
-		adev->pm.fw = NULL;
-		return 0;
-	}
-	/* cannot release other firmware because they are not created by cgs */
-	return -EINVAL;
-}
-
 static uint16_t amdgpu_get_firmware_version(struct cgs_device *cgs_device,
 					enum cgs_ucode_id type)
 {
@@ -561,77 +163,49 @@ static uint16_t amdgpu_get_firmware_version(struct cgs_device *cgs_device,
 	uint16_t fw_version = 0;
 
 	switch (type) {
-		case CGS_UCODE_ID_SDMA0:
-			fw_version = adev->sdma.instance[0].fw_version;
-			break;
-		case CGS_UCODE_ID_SDMA1:
-			fw_version = adev->sdma.instance[1].fw_version;
-			break;
-		case CGS_UCODE_ID_CP_CE:
-			fw_version = adev->gfx.ce_fw_version;
-			break;
-		case CGS_UCODE_ID_CP_PFP:
-			fw_version = adev->gfx.pfp_fw_version;
-			break;
-		case CGS_UCODE_ID_CP_ME:
-			fw_version = adev->gfx.me_fw_version;
-			break;
-		case CGS_UCODE_ID_CP_MEC:
-			fw_version = adev->gfx.mec_fw_version;
-			break;
-		case CGS_UCODE_ID_CP_MEC_JT1:
-			fw_version = adev->gfx.mec_fw_version;
-			break;
-		case CGS_UCODE_ID_CP_MEC_JT2:
-			fw_version = adev->gfx.mec_fw_version;
-			break;
-		case CGS_UCODE_ID_RLC_G:
-			fw_version = adev->gfx.rlc_fw_version;
-			break;
-		case CGS_UCODE_ID_STORAGE:
-			break;
-		default:
-			DRM_ERROR("firmware type %d do not have version\n", type);
-			break;
+	case CGS_UCODE_ID_SDMA0:
+		fw_version = adev->sdma.instance[0].fw_version;
+		break;
+	case CGS_UCODE_ID_SDMA1:
+		fw_version = adev->sdma.instance[1].fw_version;
+		break;
+	case CGS_UCODE_ID_CP_CE:
+		fw_version = adev->gfx.ce_fw_version;
+		break;
+	case CGS_UCODE_ID_CP_PFP:
+		fw_version = adev->gfx.pfp_fw_version;
+		break;
+	case CGS_UCODE_ID_CP_ME:
+		fw_version = adev->gfx.me_fw_version;
+		break;
+	case CGS_UCODE_ID_CP_MEC:
+		fw_version = adev->gfx.mec_fw_version;
+		break;
+	case CGS_UCODE_ID_CP_MEC_JT1:
+		fw_version = adev->gfx.mec_fw_version;
+		break;
+	case CGS_UCODE_ID_CP_MEC_JT2:
+		fw_version = adev->gfx.mec_fw_version;
+		break;
+	case CGS_UCODE_ID_RLC_G:
+		fw_version = adev->gfx.rlc_fw_version;
+		break;
+	case CGS_UCODE_ID_STORAGE:
+		break;
+	default:
+		DRM_ERROR("firmware type %d do not have version\n", type);
+		break;
 	}
 	return fw_version;
 }
 
-static int amdgpu_cgs_enter_safe_mode(struct cgs_device *cgs_device,
-					bool en)
-{
-	CGS_FUNC_ADEV;
-
-	if (adev->gfx.rlc.funcs->enter_safe_mode == NULL ||
-		adev->gfx.rlc.funcs->exit_safe_mode == NULL)
-		return 0;
-
-	if (en)
-		adev->gfx.rlc.funcs->enter_safe_mode(adev);
-	else
-		adev->gfx.rlc.funcs->exit_safe_mode(adev);
-
-	return 0;
-}
-
-static void amdgpu_cgs_lock_grbm_idx(struct cgs_device *cgs_device,
-					bool lock)
-{
-	CGS_FUNC_ADEV;
-
-	if (lock)
-		mutex_lock(&adev->grbm_idx_mutex);
-	else
-		mutex_unlock(&adev->grbm_idx_mutex);
-}
-
 static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device,
 					enum cgs_ucode_id type,
 					struct cgs_firmware_info *info)
 {
 	CGS_FUNC_ADEV;
 
-	if ((CGS_UCODE_ID_SMU != type) && (CGS_UCODE_ID_SMU_SK != type)) {
+	if (type != CGS_UCODE_ID_SMU && type != CGS_UCODE_ID_SMU_SK) {
 		uint64_t gpu_addr;
 		uint32_t data_size;
 		const struct gfx_firmware_header_v1_0 *header;
@@ -639,6 +213,9 @@ static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device,
 		struct amdgpu_firmware_info *ucode;
 
 		id = fw_type_convert(cgs_device, type);
+		if (id >= AMDGPU_UCODE_ID_MAXIMUM)
+			return -EINVAL;
+
 		ucode = &adev->firmware.ucode[id];
 		if (ucode->fw == NULL)
 			return -EINVAL;
@@ -658,7 +235,7 @@ static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device,
 		info->mc_addr = gpu_addr;
 		info->version = (uint16_t)le32_to_cpu(header->header.ucode_version);
 
-		if (CGS_UCODE_ID_CP_MEC == type)
+		if (type == CGS_UCODE_ID_CP_MEC)
 			info->image_size = le32_to_cpu(header->jt_offset) << 2;
 
 		info->fw_version = amdgpu_get_firmware_version(cgs_device, type);
@@ -675,89 +252,113 @@ static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device,
 
 		if (!adev->pm.fw) {
 			switch (adev->asic_type) {
+			case CHIP_BONAIRE:
+				if ((adev->pdev->revision == 0x80) ||
+					(adev->pdev->revision == 0x81) ||
+					(adev->pdev->device == 0x665f)) {
+					info->is_kicker = true;
+					strscpy(fw_name, "amdgpu/bonaire_k_smc.bin");
+				} else {
+					strscpy(fw_name, "amdgpu/bonaire_smc.bin");
+				}
+				break;
+			case CHIP_HAWAII:
+				if (adev->pdev->revision == 0x80) {
+					info->is_kicker = true;
+					strscpy(fw_name, "amdgpu/hawaii_k_smc.bin");
+				} else {
+					strscpy(fw_name, "amdgpu/hawaii_smc.bin");
+				}
+				break;
 			case CHIP_TOPAZ:
 				if (((adev->pdev->device == 0x6900) && (adev->pdev->revision == 0x81)) ||
 				    ((adev->pdev->device == 0x6900) && (adev->pdev->revision == 0x83)) ||
-				    ((adev->pdev->device == 0x6907) && (adev->pdev->revision == 0x87))) {
+				    ((adev->pdev->device == 0x6907) && (adev->pdev->revision == 0x87)) ||
+				    ((adev->pdev->device == 0x6900) && (adev->pdev->revision == 0xD1)) ||
+				    ((adev->pdev->device == 0x6900) && (adev->pdev->revision == 0xD3))) {
 					info->is_kicker = true;
-					strcpy(fw_name, "amdgpu/topaz_k_smc.bin");
+					strscpy(fw_name, "amdgpu/topaz_k_smc.bin");
 				} else
-					strcpy(fw_name, "amdgpu/topaz_smc.bin");
+					strscpy(fw_name, "amdgpu/topaz_smc.bin");
 				break;
 			case CHIP_TONGA:
 				if (((adev->pdev->device == 0x6939) && (adev->pdev->revision == 0xf1)) ||
 				    ((adev->pdev->device == 0x6938) && (adev->pdev->revision == 0xf1))) {
 					info->is_kicker = true;
-					strcpy(fw_name, "amdgpu/tonga_k_smc.bin");
+					strscpy(fw_name, "amdgpu/tonga_k_smc.bin");
 				} else
-					strcpy(fw_name, "amdgpu/tonga_smc.bin");
+					strscpy(fw_name, "amdgpu/tonga_smc.bin");
 				break;
 			case CHIP_FIJI:
-				strcpy(fw_name, "amdgpu/fiji_smc.bin");
+				strscpy(fw_name, "amdgpu/fiji_smc.bin");
 				break;
 			case CHIP_POLARIS11:
 				if (type == CGS_UCODE_ID_SMU) {
-					if (((adev->pdev->device == 0x67ef) &&
-					     ((adev->pdev->revision == 0xe0) ||
-					      (adev->pdev->revision == 0xe2) ||
-					      (adev->pdev->revision == 0xe5))) ||
-					    ((adev->pdev->device == 0x67ff) &&
-					     ((adev->pdev->revision == 0xcf) ||
-					      (adev->pdev->revision == 0xef) ||
-					      (adev->pdev->revision == 0xff)))) {
+					if (ASICID_IS_P21(adev->pdev->device, adev->pdev->revision)) {
+						info->is_kicker = true;
+						strscpy(fw_name, "amdgpu/polaris11_k_smc.bin");
+					} else if (ASICID_IS_P31(adev->pdev->device, adev->pdev->revision)) {
 						info->is_kicker = true;
-						strcpy(fw_name, "amdgpu/polaris11_k_smc.bin");
-					} else
-						strcpy(fw_name, "amdgpu/polaris11_smc.bin");
+						strscpy(fw_name, "amdgpu/polaris11_k2_smc.bin");
+					} else {
+						strscpy(fw_name, "amdgpu/polaris11_smc.bin");
+					}
 				} else if (type == CGS_UCODE_ID_SMU_SK) {
-					strcpy(fw_name, "amdgpu/polaris11_smc_sk.bin");
+					strscpy(fw_name, "amdgpu/polaris11_smc_sk.bin");
 				}
 				break;
 			case CHIP_POLARIS10:
 				if (type == CGS_UCODE_ID_SMU) {
-					if ((adev->pdev->device == 0x67df) &&
-					    ((adev->pdev->revision == 0xe0) ||
-					     (adev->pdev->revision == 0xe3) ||
-					     (adev->pdev->revision == 0xe4) ||
-					     (adev->pdev->revision == 0xe5) ||
-					     (adev->pdev->revision == 0xe7) ||
-					     (adev->pdev->revision == 0xef))) {
+					if (ASICID_IS_P20(adev->pdev->device, adev->pdev->revision)) {
 						info->is_kicker = true;
-						strcpy(fw_name, "amdgpu/polaris10_k_smc.bin");
-					} else
-						strcpy(fw_name, "amdgpu/polaris10_smc.bin");
+						strscpy(fw_name, "amdgpu/polaris10_k_smc.bin");
+					} else if (ASICID_IS_P30(adev->pdev->device, adev->pdev->revision)) {
+						info->is_kicker = true;
+						strscpy(fw_name, "amdgpu/polaris10_k2_smc.bin");
+					} else {
+						strscpy(fw_name, "amdgpu/polaris10_smc.bin");
+					}
 				} else if (type == CGS_UCODE_ID_SMU_SK) {
-					strcpy(fw_name, "amdgpu/polaris10_smc_sk.bin");
+					strscpy(fw_name, "amdgpu/polaris10_smc_sk.bin");
 				}
 				break;
 			case CHIP_POLARIS12:
-				strcpy(fw_name, "amdgpu/polaris12_smc.bin");
+				if (ASICID_IS_P23(adev->pdev->device, adev->pdev->revision)) {
+					info->is_kicker = true;
+					strscpy(fw_name, "amdgpu/polaris12_k_smc.bin");
+				} else {
+					strscpy(fw_name, "amdgpu/polaris12_smc.bin");
+				}
+				break;
+			case CHIP_VEGAM:
+				strscpy(fw_name, "amdgpu/vegam_smc.bin");
 				break;
 			case CHIP_VEGA10:
 				if ((adev->pdev->device == 0x687f) &&
 					((adev->pdev->revision == 0xc0) ||
 					(adev->pdev->revision == 0xc1) ||
 					(adev->pdev->revision == 0xc3)))
-					strcpy(fw_name, "amdgpu/vega10_acg_smc.bin");
+					strscpy(fw_name, "amdgpu/vega10_acg_smc.bin");
 				else
-					strcpy(fw_name, "amdgpu/vega10_smc.bin");
+					strscpy(fw_name, "amdgpu/vega10_smc.bin");
+				break;
+			case CHIP_VEGA12:
+				strscpy(fw_name, "amdgpu/vega12_smc.bin");
+				break;
+			case CHIP_VEGA20:
+				strscpy(fw_name, "amdgpu/vega20_smc.bin");
 				break;
 			default:
 				DRM_ERROR("SMC firmware not supported\n");
 				return -EINVAL;
 			}
 
-			err = request_firmware(&adev->pm.fw, fw_name, adev->dev);
-			if (err) {
-				DRM_ERROR("Failed to request firmware\n");
-				return err;
-			}
-
-			err = amdgpu_ucode_validate(adev->pm.fw);
+			err = amdgpu_ucode_request(adev, &adev->pm.fw,
+						   AMDGPU_UCODE_REQUIRED,
+						   "%s", fw_name);
 			if (err) {
 				DRM_ERROR("Failed to load firmware \"%s\"", fw_name);
-				release_firmware(adev->pm.fw);
-				adev->pm.fw = NULL;
+				amdgpu_ucode_release(&adev->pm.fw);
 				return err;
 			}
 
@@ -787,364 +388,12 @@ static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device,
 	return 0;
 }
 
-static int amdgpu_cgs_is_virtualization_enabled(void *cgs_device)
-{
-	CGS_FUNC_ADEV;
-	return amdgpu_sriov_vf(adev);
-}
-
-static int amdgpu_cgs_query_system_info(struct cgs_device *cgs_device,
-					struct cgs_system_info *sys_info)
-{
-	CGS_FUNC_ADEV;
-
-	if (NULL == sys_info)
-		return -ENODEV;
-
-	if (sizeof(struct cgs_system_info) != sys_info->size)
-		return -ENODEV;
-
-	switch (sys_info->info_id) {
-	case CGS_SYSTEM_INFO_ADAPTER_BDF_ID:
-		sys_info->value = adev->pdev->devfn | (adev->pdev->bus->number << 8);
-		break;
-	case CGS_SYSTEM_INFO_PCIE_GEN_INFO:
-		sys_info->value = adev->pm.pcie_gen_mask;
-		break;
-	case CGS_SYSTEM_INFO_PCIE_MLW:
-		sys_info->value = adev->pm.pcie_mlw_mask;
-		break;
-	case CGS_SYSTEM_INFO_PCIE_DEV:
-		sys_info->value = adev->pdev->device;
-		break;
-	case CGS_SYSTEM_INFO_PCIE_REV:
-		sys_info->value = adev->pdev->revision;
-		break;
-	case CGS_SYSTEM_INFO_CG_FLAGS:
-		sys_info->value = adev->cg_flags;
-		break;
-	case CGS_SYSTEM_INFO_PG_FLAGS:
-		sys_info->value = adev->pg_flags;
-		break;
-	case CGS_SYSTEM_INFO_GFX_CU_INFO:
-		sys_info->value = adev->gfx.cu_info.number;
-		break;
-	case CGS_SYSTEM_INFO_GFX_SE_INFO:
-		sys_info->value = adev->gfx.config.max_shader_engines;
-		break;
-	case CGS_SYSTEM_INFO_PCIE_SUB_SYS_ID:
-		sys_info->value = adev->pdev->subsystem_device;
-		break;
-	case CGS_SYSTEM_INFO_PCIE_SUB_SYS_VENDOR_ID:
-		sys_info->value = adev->pdev->subsystem_vendor;
-		break;
-	default:
-		return -ENODEV;
-	}
-
-	return 0;
-}
-
-static int amdgpu_cgs_get_active_displays_info(struct cgs_device *cgs_device,
-					  struct cgs_display_info *info)
-{
-	CGS_FUNC_ADEV;
-	struct amdgpu_crtc *amdgpu_crtc;
-	struct drm_device *ddev = adev->ddev;
-	struct drm_crtc *crtc;
-	uint32_t line_time_us, vblank_lines;
-	struct cgs_mode_info *mode_info;
-
-	if (info == NULL)
-		return -EINVAL;
-
-	mode_info = info->mode_info;
-	if (mode_info) {
-		/* if the displays are off, vblank time is max */
-		mode_info->vblank_time_us = 0xffffffff;
-		/* always set the reference clock */
-		mode_info->ref_clock = adev->clock.spll.reference_freq;
-	}
-
-	if (adev->mode_info.num_crtc && adev->mode_info.mode_config_initialized) {
-		list_for_each_entry(crtc,
-				&ddev->mode_config.crtc_list, head) {
-			amdgpu_crtc = to_amdgpu_crtc(crtc);
-			if (crtc->enabled) {
-				info->active_display_mask |= (1 << amdgpu_crtc->crtc_id);
-				info->display_count++;
-			}
-			if (mode_info != NULL &&
-				crtc->enabled && amdgpu_crtc->enabled &&
-				amdgpu_crtc->hw_mode.clock) {
-				line_time_us = (amdgpu_crtc->hw_mode.crtc_htotal * 1000) /
-							amdgpu_crtc->hw_mode.clock;
-				vblank_lines = amdgpu_crtc->hw_mode.crtc_vblank_end -
-							amdgpu_crtc->hw_mode.crtc_vdisplay +
-							(amdgpu_crtc->v_border * 2);
-				mode_info->vblank_time_us = vblank_lines * line_time_us;
-				mode_info->refresh_rate = drm_mode_vrefresh(&amdgpu_crtc->hw_mode);
-				mode_info->ref_clock = adev->clock.spll.reference_freq;
-				mode_info = NULL;
-			}
-		}
-	}
-
-	return 0;
-}
-
-
-static int amdgpu_cgs_notify_dpm_enabled(struct cgs_device *cgs_device, bool enabled)
-{
-	CGS_FUNC_ADEV;
-
-	adev->pm.dpm_enabled = enabled;
-
-	return 0;
-}
-
-/** \brief evaluate acpi namespace object, handle or pathname must be valid
- *  \param cgs_device
- *  \param info input/output arguments for the control method
- *  \return status
- */
-
-#if defined(CONFIG_ACPI)
-static int amdgpu_cgs_acpi_eval_object(struct cgs_device *cgs_device,
-				    struct cgs_acpi_method_info *info)
-{
-	CGS_FUNC_ADEV;
-	acpi_handle handle;
-	struct acpi_object_list input;
-	struct acpi_buffer output = { ACPI_ALLOCATE_BUFFER, NULL };
-	union acpi_object *params, *obj;
-	uint8_t name[5] = {'\0'};
-	struct cgs_acpi_method_argument *argument;
-	uint32_t i, count;
-	acpi_status status;
-	int result;
-
-	handle = ACPI_HANDLE(&adev->pdev->dev);
-	if (!handle)
-		return -ENODEV;
-
-	memset(&input, 0, sizeof(struct acpi_object_list));
-
-	/* validate input info */
-	if (info->size != sizeof(struct cgs_acpi_method_info))
-		return -EINVAL;
-
-	input.count = info->input_count;
-	if (info->input_count > 0) {
-		if (info->pinput_argument == NULL)
-			return -EINVAL;
-		argument = info->pinput_argument;
-		for (i = 0; i < info->input_count; i++) {
-			if (((argument->type == ACPI_TYPE_STRING) ||
-			     (argument->type == ACPI_TYPE_BUFFER)) &&
-			    (argument->pointer == NULL))
-				return -EINVAL;
-			argument++;
-		}
-	}
-
-	if (info->output_count > 0) {
-		if (info->poutput_argument == NULL)
-			return -EINVAL;
-		argument = info->poutput_argument;
-		for (i = 0; i < info->output_count; i++) {
-			if (((argument->type == ACPI_TYPE_STRING) ||
-				(argument->type == ACPI_TYPE_BUFFER))
-				&& (argument->pointer == NULL))
-				return -EINVAL;
-			argument++;
-		}
-	}
-
-	/* The path name passed to acpi_evaluate_object should be null terminated */
-	if ((info->field & CGS_ACPI_FIELD_METHOD_NAME) != 0) {
-		strncpy(name, (char *)&(info->name), sizeof(uint32_t));
-		name[4] = '\0';
-	}
-
-	/* parse input parameters */
-	if (input.count > 0) {
-		input.pointer = params =
-				kzalloc(sizeof(union acpi_object) * input.count, GFP_KERNEL);
-		if (params == NULL)
-			return -EINVAL;
-
-		argument = info->pinput_argument;
-
-		for (i = 0; i < input.count; i++) {
-			params->type = argument->type;
-			switch (params->type) {
-			case ACPI_TYPE_INTEGER:
-				params->integer.value = argument->value;
-				break;
-			case ACPI_TYPE_STRING:
-				params->string.length = argument->data_length;
-				params->string.pointer = argument->pointer;
-				break;
-			case ACPI_TYPE_BUFFER:
-				params->buffer.length = argument->data_length;
-				params->buffer.pointer = argument->pointer;
-				break;
-			default:
-				break;
-			}
-			params++;
-			argument++;
-		}
-	}
-
-	/* parse output info */
-	count = info->output_count;
-	argument = info->poutput_argument;
-
-	/* evaluate the acpi method */
-	status = acpi_evaluate_object(handle, name, &input, &output);
-
-	if (ACPI_FAILURE(status)) {
-		result = -EIO;
-		goto free_input;
-	}
-
-	/* return the output info */
-	obj = output.pointer;
-
-	if (count > 1) {
-		if ((obj->type != ACPI_TYPE_PACKAGE) ||
-			(obj->package.count != count)) {
-			result = -EIO;
-			goto free_obj;
-		}
-		params = obj->package.elements;
-	} else
-		params = obj;
-
-	if (params == NULL) {
-		result = -EIO;
-		goto free_obj;
-	}
-
-	for (i = 0; i < count; i++) {
-		if (argument->type != params->type) {
-			result = -EIO;
-			goto free_obj;
-		}
-		switch (params->type) {
-		case ACPI_TYPE_INTEGER:
-			argument->value = params->integer.value;
-			break;
-		case ACPI_TYPE_STRING:
-			if ((params->string.length != argument->data_length) ||
-				(params->string.pointer == NULL)) {
-				result = -EIO;
-				goto free_obj;
-			}
-			strncpy(argument->pointer,
-				params->string.pointer,
-				params->string.length);
-			break;
-		case ACPI_TYPE_BUFFER:
-			if (params->buffer.pointer == NULL) {
-				result = -EIO;
-				goto free_obj;
-			}
-			memcpy(argument->pointer,
-				params->buffer.pointer,
-				argument->data_length);
-			break;
-		default:
-			break;
-		}
-		argument++;
-		params++;
-	}
-
-	result = 0;
-free_obj:
-	kfree(obj);
-free_input:
-	kfree((void *)input.pointer);
-	return result;
-}
-#else
-static int amdgpu_cgs_acpi_eval_object(struct cgs_device *cgs_device,
-				struct cgs_acpi_method_info *info)
-{
-	return -EIO;
-}
-#endif
-
-static int amdgpu_cgs_call_acpi_method(struct cgs_device *cgs_device,
-					uint32_t acpi_method,
-					uint32_t acpi_function,
-					void *pinput, void *poutput,
-					uint32_t output_count,
-					uint32_t input_size,
-					uint32_t output_size)
-{
-	struct cgs_acpi_method_argument acpi_input[2] = { {0}, {0} };
-	struct cgs_acpi_method_argument acpi_output = {0};
-	struct cgs_acpi_method_info info = {0};
-
-	acpi_input[0].type = CGS_ACPI_TYPE_INTEGER;
-	acpi_input[0].data_length = sizeof(uint32_t);
-	acpi_input[0].value = acpi_function;
-
-	acpi_input[1].type = CGS_ACPI_TYPE_BUFFER;
-	acpi_input[1].data_length = input_size;
-	acpi_input[1].pointer = pinput;
-
-	acpi_output.type = CGS_ACPI_TYPE_BUFFER;
-	acpi_output.data_length = output_size;
-	acpi_output.pointer = poutput;
-
-	info.size = sizeof(struct cgs_acpi_method_info);
-	info.field = CGS_ACPI_FIELD_METHOD_NAME | CGS_ACPI_FIELD_INPUT_ARGUMENT_COUNT;
-	info.input_count = 2;
-	info.name = acpi_method;
-	info.pinput_argument = acpi_input;
-	info.output_count = output_count;
-	info.poutput_argument = &acpi_output;
-
-	return amdgpu_cgs_acpi_eval_object(cgs_device, &info);
-}
-
 static const struct cgs_ops amdgpu_cgs_ops = {
-	.alloc_gpu_mem = amdgpu_cgs_alloc_gpu_mem,
-	.free_gpu_mem = amdgpu_cgs_free_gpu_mem,
-	.gmap_gpu_mem = amdgpu_cgs_gmap_gpu_mem,
-	.gunmap_gpu_mem = amdgpu_cgs_gunmap_gpu_mem,
-	.kmap_gpu_mem = amdgpu_cgs_kmap_gpu_mem,
-	.kunmap_gpu_mem = amdgpu_cgs_kunmap_gpu_mem,
 	.read_register = amdgpu_cgs_read_register,
 	.write_register = amdgpu_cgs_write_register,
 	.read_ind_register = amdgpu_cgs_read_ind_register,
 	.write_ind_register = amdgpu_cgs_write_ind_register,
-	.get_pci_resource = amdgpu_cgs_get_pci_resource,
-	.atom_get_data_table = amdgpu_cgs_atom_get_data_table,
-	.atom_get_cmd_table_revs = amdgpu_cgs_atom_get_cmd_table_revs,
-	.atom_exec_cmd_table = amdgpu_cgs_atom_exec_cmd_table,
 	.get_firmware_info = amdgpu_cgs_get_firmware_info,
-	.rel_firmware = amdgpu_cgs_rel_firmware,
-	.set_powergating_state = amdgpu_cgs_set_powergating_state,
-	.set_clockgating_state = amdgpu_cgs_set_clockgating_state,
-	.get_active_displays_info = amdgpu_cgs_get_active_displays_info,
-	.notify_dpm_enabled = amdgpu_cgs_notify_dpm_enabled,
-	.call_acpi_method = amdgpu_cgs_call_acpi_method,
-	.query_system_info = amdgpu_cgs_query_system_info,
-	.is_virtualization_enabled = amdgpu_cgs_is_virtualization_enabled,
-	.enter_safe_mode = amdgpu_cgs_enter_safe_mode,
-	.lock_grbm_idx = amdgpu_cgs_lock_grbm_idx,
-};
-
-static const struct cgs_os_ops amdgpu_cgs_os_ops = {
-	.add_irq_source = amdgpu_cgs_add_irq_source,
-	.irq_get = amdgpu_cgs_irq_get,
-	.irq_put = amdgpu_cgs_irq_put
 };
 
 struct cgs_device *amdgpu_cgs_create_device(struct amdgpu_device *adev)
@@ -1158,7 +407,6 @@ struct cgs_device *amdgpu_cgs_create_device(struct amdgpu_device *adev)
 	}
 
 	cgs_device->base.ops = &amdgpu_cgs_ops;
-	cgs_device->base.os_ops = &amdgpu_cgs_os_ops;
 	cgs_device->adev = adev;
 
 	return (struct cgs_device *)cgs_device;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c
index 8d1cf2d3e663..47e9bfba0642 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c
@@ -23,10 +23,12 @@
  * Authors: Dave Airlie
  *          Alex Deucher
  */
-#include <drm/drmP.h>
-#include <drm/drm_edid.h>
+
+#include <drm/display/drm_dp_helper.h>
 #include <drm/drm_crtc_helper.h>
-#include <drm/drm_fb_helper.h>
+#include <drm/drm_edid.h>
+#include <drm/drm_modeset_helper_vtables.h>
+#include <drm/drm_probe_helper.h>
 #include <drm/amdgpu_drm.h>
 #include "amdgpu.h"
 #include "atom.h"
@@ -34,13 +36,14 @@
 #include "atombios_dp.h"
 #include "amdgpu_connectors.h"
 #include "amdgpu_i2c.h"
+#include "amdgpu_display.h"
 
 #include <linux/pm_runtime.h>
 
 void amdgpu_connector_hotplug(struct drm_connector *connector)
 {
 	struct drm_device *dev = connector->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 
 	/* bail if the connector does not have hpd pin, e.g.,
@@ -69,25 +72,18 @@ void amdgpu_connector_hotplug(struct drm_connector *connector)
 		/* don't do anything if sink is not display port, i.e.,
 		 * passive dp->(dvi|hdmi) adaptor
 		 */
-		if (dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) {
-			int saved_dpms = connector->dpms;
-			/* Only turn off the display if it's physically disconnected */
-			if (!amdgpu_display_hpd_sense(adev, amdgpu_connector->hpd.hpd)) {
-				drm_helper_connector_dpms(connector, DRM_MODE_DPMS_OFF);
-			} else if (amdgpu_atombios_dp_needs_link_train(amdgpu_connector)) {
-				/* Don't try to start link training before we
-				 * have the dpcd */
-				if (amdgpu_atombios_dp_get_dpcd(amdgpu_connector))
-					return;
-
-				/* set it to OFF so that drm_helper_connector_dpms()
-				 * won't return immediately since the current state
-				 * is ON at this point.
-				 */
-				connector->dpms = DRM_MODE_DPMS_OFF;
-				drm_helper_connector_dpms(connector, DRM_MODE_DPMS_ON);
-			}
-			connector->dpms = saved_dpms;
+		if (dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT &&
+		    amdgpu_display_hpd_sense(adev, amdgpu_connector->hpd.hpd) &&
+		    amdgpu_atombios_dp_needs_link_train(amdgpu_connector)) {
+			/* Don't start link training before we have the DPCD */
+			if (amdgpu_atombios_dp_get_dpcd(amdgpu_connector))
+				return;
+
+			/* Turn the connector off and back on immediately, which
+			 * will trigger link training
+			 */
+			drm_helper_connector_dpms(connector, DRM_MODE_DPMS_OFF);
+			drm_helper_connector_dpms(connector, DRM_MODE_DPMS_ON);
 		}
 	}
 }
@@ -107,13 +103,13 @@ int amdgpu_connector_get_monitor_bpc(struct drm_connector *connector)
 	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 	struct amdgpu_connector_atom_dig *dig_connector;
 	int bpc = 8;
-	unsigned mode_clock, max_tmds_clock;
+	unsigned int mode_clock, max_tmds_clock;
 
 	switch (connector->connector_type) {
 	case DRM_MODE_CONNECTOR_DVII:
 	case DRM_MODE_CONNECTOR_HDMIB:
 		if (amdgpu_connector->use_digital) {
-			if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+			if (connector->display_info.is_hdmi) {
 				if (connector->display_info.bpc)
 					bpc = connector->display_info.bpc;
 			}
@@ -121,7 +117,7 @@ int amdgpu_connector_get_monitor_bpc(struct drm_connector *connector)
 		break;
 	case DRM_MODE_CONNECTOR_DVID:
 	case DRM_MODE_CONNECTOR_HDMIA:
-		if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+		if (connector->display_info.is_hdmi) {
 			if (connector->display_info.bpc)
 				bpc = connector->display_info.bpc;
 		}
@@ -130,7 +126,7 @@ int amdgpu_connector_get_monitor_bpc(struct drm_connector *connector)
 		dig_connector = amdgpu_connector->con_priv;
 		if ((dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) ||
 		    (dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_eDP) ||
-		    drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+		    connector->display_info.is_hdmi) {
 			if (connector->display_info.bpc)
 				bpc = connector->display_info.bpc;
 		}
@@ -154,7 +150,7 @@ int amdgpu_connector_get_monitor_bpc(struct drm_connector *connector)
 		break;
 	}
 
-	if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+	if (connector->display_info.is_hdmi) {
 		/*
 		 * Pre DCE-8 hw can't handle > 12 bpc, and more than 12 bpc doesn't make
 		 * much sense without support for > 12 bpc framebuffers. RGB 4:4:4 at
@@ -180,7 +176,7 @@ int amdgpu_connector_get_monitor_bpc(struct drm_connector *connector)
 
 			/* Check if bpc is within clock limit. Try to degrade gracefully otherwise */
 			if ((bpc == 12) && (mode_clock * 3/2 > max_tmds_clock)) {
-				if ((connector->display_info.edid_hdmi_dc_modes & DRM_EDID_HDMI_DC_30) &&
+				if ((connector->display_info.edid_hdmi_rgb444_dc_modes & DRM_EDID_HDMI_DC_30) &&
 				    (mode_clock * 5/4 <= max_tmds_clock))
 					bpc = 10;
 				else
@@ -219,30 +215,20 @@ static void
 amdgpu_connector_update_scratch_regs(struct drm_connector *connector,
 				      enum drm_connector_status status)
 {
-	struct drm_encoder *best_encoder = NULL;
-	struct drm_encoder *encoder = NULL;
+	struct drm_encoder *best_encoder;
+	struct drm_encoder *encoder;
 	const struct drm_connector_helper_funcs *connector_funcs = connector->helper_private;
 	bool connected;
-	int i;
 
 	best_encoder = connector_funcs->best_encoder(connector);
 
-	for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) {
-		if (connector->encoder_ids[i] == 0)
-			break;
-
-		encoder = drm_encoder_find(connector->dev,
-					connector->encoder_ids[i]);
-		if (!encoder)
-			continue;
-
+	drm_connector_for_each_possible_encoder(connector, encoder) {
 		if ((encoder == best_encoder) && (status == connector_status_connected))
 			connected = true;
 		else
 			connected = false;
 
 		amdgpu_atombios_encoder_set_bios_scratch_regs(connector, encoder, connected);
-
 	}
 }
 
@@ -251,58 +237,25 @@ amdgpu_connector_find_encoder(struct drm_connector *connector,
 			       int encoder_type)
 {
 	struct drm_encoder *encoder;
-	int i;
-
-	for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) {
-		if (connector->encoder_ids[i] == 0)
-			break;
-		encoder = drm_encoder_find(connector->dev,
-					connector->encoder_ids[i]);
-		if (!encoder)
-			continue;
 
+	drm_connector_for_each_possible_encoder(connector, encoder) {
 		if (encoder->encoder_type == encoder_type)
 			return encoder;
 	}
-	return NULL;
-}
-
-struct edid *amdgpu_connector_edid(struct drm_connector *connector)
-{
-	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
-	struct drm_property_blob *edid_blob = connector->edid_blob_ptr;
 
-	if (amdgpu_connector->edid) {
-		return amdgpu_connector->edid;
-	} else if (edid_blob) {
-		struct edid *edid = kmemdup(edid_blob->data, edid_blob->length, GFP_KERNEL);
-		if (edid)
-			amdgpu_connector->edid = edid;
-	}
-	return amdgpu_connector->edid;
+	return NULL;
 }
 
 static struct edid *
 amdgpu_connector_get_hardcoded_edid(struct amdgpu_device *adev)
 {
-	struct edid *edid;
-
-	if (adev->mode_info.bios_hardcoded_edid) {
-		edid = kmalloc(adev->mode_info.bios_hardcoded_edid_size, GFP_KERNEL);
-		if (edid) {
-			memcpy((unsigned char *)edid,
-			       (unsigned char *)adev->mode_info.bios_hardcoded_edid,
-			       adev->mode_info.bios_hardcoded_edid_size);
-			return edid;
-		}
-	}
-	return NULL;
+	return drm_edid_duplicate(drm_edid_raw(adev->mode_info.bios_hardcoded_edid));
 }
 
 static void amdgpu_connector_get_edid(struct drm_connector *connector)
 {
 	struct drm_device *dev = connector->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 
 	if (amdgpu_connector->edid)
@@ -337,8 +290,10 @@ static void amdgpu_connector_get_edid(struct drm_connector *connector)
 	if (!amdgpu_connector->edid) {
 		/* some laptops provide a hardcoded edid in rom for LCDs */
 		if (((connector->connector_type == DRM_MODE_CONNECTOR_LVDS) ||
-		     (connector->connector_type == DRM_MODE_CONNECTOR_eDP)))
+		     (connector->connector_type == DRM_MODE_CONNECTOR_eDP))) {
 			amdgpu_connector->edid = amdgpu_connector_get_hardcoded_edid(adev);
+			drm_connector_update_edid_property(connector, amdgpu_connector->edid);
+		}
 	}
 }
 
@@ -346,10 +301,8 @@ static void amdgpu_connector_free_edid(struct drm_connector *connector)
 {
 	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 
-	if (amdgpu_connector->edid) {
-		kfree(amdgpu_connector->edid);
-		amdgpu_connector->edid = NULL;
-	}
+	kfree(amdgpu_connector->edid);
+	amdgpu_connector->edid = NULL;
 }
 
 static int amdgpu_connector_ddc_get_modes(struct drm_connector *connector)
@@ -358,23 +311,23 @@ static int amdgpu_connector_ddc_get_modes(struct drm_connector *connector)
 	int ret;
 
 	if (amdgpu_connector->edid) {
-		drm_mode_connector_update_edid_property(connector, amdgpu_connector->edid);
+		drm_connector_update_edid_property(connector, amdgpu_connector->edid);
 		ret = drm_add_edid_modes(connector, amdgpu_connector->edid);
-		drm_edid_to_eld(connector, amdgpu_connector->edid);
 		return ret;
 	}
-	drm_mode_connector_update_edid_property(connector, NULL);
+	drm_connector_update_edid_property(connector, NULL);
 	return 0;
 }
 
 static struct drm_encoder *
 amdgpu_connector_best_single_encoder(struct drm_connector *connector)
 {
-	int enc_id = connector->encoder_ids[0];
+	struct drm_encoder *encoder;
+
+	/* pick the first one */
+	drm_connector_for_each_possible_encoder(connector, encoder)
+		return encoder;
 
-	/* pick the encoder ids */
-	if (enc_id)
-		return drm_encoder_find(connector->dev, enc_id);
 	return NULL;
 }
 
@@ -411,6 +364,9 @@ amdgpu_connector_lcd_native_mode(struct drm_encoder *encoder)
 	    native_mode->vdisplay != 0 &&
 	    native_mode->clock != 0) {
 		mode = drm_mode_duplicate(dev, native_mode);
+		if (!mode)
+			return NULL;
+
 		mode->type = DRM_MODE_TYPE_PREFERRED | DRM_MODE_TYPE_DRIVER;
 		drm_mode_set_name(mode);
 
@@ -425,6 +381,9 @@ amdgpu_connector_lcd_native_mode(struct drm_encoder *encoder)
 		 * simpler.
 		 */
 		mode = drm_cvt_mode(dev, native_mode->hdisplay, native_mode->vdisplay, 60, true, false, false);
+		if (!mode)
+			return NULL;
+
 		mode->type = DRM_MODE_TYPE_PREFERRED | DRM_MODE_TYPE_DRIVER;
 		DRM_DEBUG_KMS("Adding cvt approximation of native panel mode %s\n", mode->name);
 	}
@@ -439,30 +398,28 @@ static void amdgpu_connector_add_common_modes(struct drm_encoder *encoder,
 	struct drm_display_mode *mode = NULL;
 	struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode;
 	int i;
-	static const struct mode_size {
+	int n;
+	struct mode_size {
+		char name[DRM_DISPLAY_MODE_LEN];
 		int w;
 		int h;
-	} common_modes[17] = {
-		{ 640,  480},
-		{ 720,  480},
-		{ 800,  600},
-		{ 848,  480},
-		{1024,  768},
-		{1152,  768},
-		{1280,  720},
-		{1280,  800},
-		{1280,  854},
-		{1280,  960},
-		{1280, 1024},
-		{1440,  900},
-		{1400, 1050},
-		{1680, 1050},
-		{1600, 1200},
-		{1920, 1080},
-		{1920, 1200}
+	} common_modes[] = {
+		{  "640x480",  640,  480},
+		{  "800x600",  800,  600},
+		{ "1024x768", 1024,  768},
+		{ "1280x720", 1280,  720},
+		{ "1280x800", 1280,  800},
+		{"1280x1024", 1280, 1024},
+		{ "1440x900", 1440,  900},
+		{"1680x1050", 1680, 1050},
+		{"1600x1200", 1600, 1200},
+		{"1920x1080", 1920, 1080},
+		{"1920x1200", 1920, 1200}
 	};
 
-	for (i = 0; i < 17; i++) {
+	n = ARRAY_SIZE(common_modes);
+
+	for (i = 0; i < n; i++) {
 		if (amdgpu_encoder->devices & (ATOM_DEVICE_TV_SUPPORT)) {
 			if (common_modes[i].w > 1024 ||
 			    common_modes[i].h > 768)
@@ -475,10 +432,12 @@ static void amdgpu_connector_add_common_modes(struct drm_encoder *encoder,
 			     common_modes[i].h == native_mode->vdisplay))
 				continue;
 		}
-		if (common_modes[i].w < 320 || common_modes[i].h < 200)
-			continue;
 
 		mode = drm_cvt_mode(dev, common_modes[i].w, common_modes[i].h, 60, false, false, false);
+		if (!mode)
+			return;
+		strscpy(mode->name, common_modes[i].name, DRM_DISPLAY_MODE_LEN);
+
 		drm_mode_probed_add(connector, mode);
 	}
 }
@@ -488,7 +447,7 @@ static int amdgpu_connector_set_property(struct drm_connector *connector,
 					  uint64_t val)
 {
 	struct drm_device *dev = connector->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct drm_encoder *encoder;
 	struct amdgpu_encoder *amdgpu_encoder;
 
@@ -603,16 +562,26 @@ static int amdgpu_connector_set_property(struct drm_connector *connector,
 			amdgpu_encoder = to_amdgpu_encoder(connector->encoder);
 		} else {
 			const struct drm_connector_helper_funcs *connector_funcs = connector->helper_private;
+
 			amdgpu_encoder = to_amdgpu_encoder(connector_funcs->best_encoder(connector));
 		}
 
 		switch (val) {
 		default:
-		case DRM_MODE_SCALE_NONE: rmx_type = RMX_OFF; break;
-		case DRM_MODE_SCALE_CENTER: rmx_type = RMX_CENTER; break;
-		case DRM_MODE_SCALE_ASPECT: rmx_type = RMX_ASPECT; break;
-		case DRM_MODE_SCALE_FULLSCREEN: rmx_type = RMX_FULL; break;
+		case DRM_MODE_SCALE_NONE:
+			rmx_type = RMX_OFF;
+			break;
+		case DRM_MODE_SCALE_CENTER:
+			rmx_type = RMX_CENTER;
+			break;
+		case DRM_MODE_SCALE_ASPECT:
+			rmx_type = RMX_ASPECT;
+			break;
+		case DRM_MODE_SCALE_FULLSCREEN:
+			rmx_type = RMX_FULL;
+			break;
 		}
+
 		if (amdgpu_encoder->rmx_type == rmx_type)
 			return 0;
 
@@ -641,7 +610,7 @@ amdgpu_connector_fixup_lcd_native_mode(struct drm_encoder *encoder,
 		if (mode->type & DRM_MODE_TYPE_PREFERRED) {
 			if (mode->hdisplay != native_mode->hdisplay ||
 			    mode->vdisplay != native_mode->vdisplay)
-				memcpy(native_mode, mode, sizeof(*mode));
+				drm_mode_copy(native_mode, mode);
 		}
 	}
 
@@ -650,7 +619,7 @@ amdgpu_connector_fixup_lcd_native_mode(struct drm_encoder *encoder,
 		list_for_each_entry_safe(mode, t, &connector->probed_modes, head) {
 			if (mode->hdisplay == native_mode->hdisplay &&
 			    mode->vdisplay == native_mode->vdisplay) {
-				*native_mode = *mode;
+				drm_mode_copy(native_mode, mode);
 				drm_mode_set_crtcinfo(native_mode, CRTC_INTERLACE_HALVE_V);
 				DRM_DEBUG_KMS("Determined LVDS native mode details from EDID\n");
 				break;
@@ -701,8 +670,8 @@ static int amdgpu_connector_lvds_get_modes(struct drm_connector *connector)
 	return ret;
 }
 
-static int amdgpu_connector_lvds_mode_valid(struct drm_connector *connector,
-					     struct drm_display_mode *mode)
+static enum drm_mode_status amdgpu_connector_lvds_mode_valid(struct drm_connector *connector,
+					     const struct drm_display_mode *mode)
 {
 	struct drm_encoder *encoder = amdgpu_connector_best_single_encoder(connector);
 
@@ -739,9 +708,13 @@ amdgpu_connector_lvds_detect(struct drm_connector *connector, bool force)
 	enum drm_connector_status ret = connector_status_disconnected;
 	int r;
 
-	r = pm_runtime_get_sync(connector->dev->dev);
-	if (r < 0)
-		return connector_status_disconnected;
+	if (!drm_kms_helper_is_poll_worker()) {
+		r = pm_runtime_get_sync(connector->dev->dev);
+		if (r < 0) {
+			pm_runtime_put_autosuspend(connector->dev->dev);
+			return connector_status_disconnected;
+		}
+	}
 
 	if (encoder) {
 		struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
@@ -760,8 +733,12 @@ amdgpu_connector_lvds_detect(struct drm_connector *connector, bool force)
 	/* check acpi lid status ??? */
 
 	amdgpu_connector_update_scratch_regs(connector, ret);
-	pm_runtime_mark_last_busy(connector->dev->dev);
-	pm_runtime_put_autosuspend(connector->dev->dev);
+
+	if (!drm_kms_helper_is_poll_worker()) {
+		pm_runtime_mark_last_busy(connector->dev->dev);
+		pm_runtime_put_autosuspend(connector->dev->dev);
+	}
+
 	return ret;
 }
 
@@ -802,16 +779,26 @@ static int amdgpu_connector_set_lcd_property(struct drm_connector *connector,
 		amdgpu_encoder = to_amdgpu_encoder(connector->encoder);
 	else {
 		const struct drm_connector_helper_funcs *connector_funcs = connector->helper_private;
+
 		amdgpu_encoder = to_amdgpu_encoder(connector_funcs->best_encoder(connector));
 	}
 
 	switch (value) {
-	case DRM_MODE_SCALE_NONE: rmx_type = RMX_OFF; break;
-	case DRM_MODE_SCALE_CENTER: rmx_type = RMX_CENTER; break;
-	case DRM_MODE_SCALE_ASPECT: rmx_type = RMX_ASPECT; break;
+	case DRM_MODE_SCALE_NONE:
+		rmx_type = RMX_OFF;
+		break;
+	case DRM_MODE_SCALE_CENTER:
+		rmx_type = RMX_CENTER;
+		break;
+	case DRM_MODE_SCALE_ASPECT:
+		rmx_type = RMX_ASPECT;
+		break;
 	default:
-	case DRM_MODE_SCALE_FULLSCREEN: rmx_type = RMX_FULL; break;
+	case DRM_MODE_SCALE_FULLSCREEN:
+		rmx_type = RMX_FULL;
+		break;
 	}
+
 	if (amdgpu_encoder->rmx_type == rmx_type)
 		return 0;
 
@@ -843,15 +830,16 @@ static int amdgpu_connector_vga_get_modes(struct drm_connector *connector)
 
 	amdgpu_connector_get_edid(connector);
 	ret = amdgpu_connector_ddc_get_modes(connector);
+	amdgpu_get_native_mode(connector);
 
 	return ret;
 }
 
-static int amdgpu_connector_vga_mode_valid(struct drm_connector *connector,
-					    struct drm_display_mode *mode)
+static enum drm_mode_status amdgpu_connector_vga_mode_valid(struct drm_connector *connector,
+					    const struct drm_display_mode *mode)
 {
 	struct drm_device *dev = connector->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 
 	/* XXX check mode bandwidth */
 
@@ -871,16 +859,20 @@ amdgpu_connector_vga_detect(struct drm_connector *connector, bool force)
 	enum drm_connector_status ret = connector_status_disconnected;
 	int r;
 
-	r = pm_runtime_get_sync(connector->dev->dev);
-	if (r < 0)
-		return connector_status_disconnected;
+	if (!drm_kms_helper_is_poll_worker()) {
+		r = pm_runtime_get_sync(connector->dev->dev);
+		if (r < 0) {
+			pm_runtime_put_autosuspend(connector->dev->dev);
+			return connector_status_disconnected;
+		}
+	}
 
 	encoder = amdgpu_connector_best_single_encoder(connector);
 	if (!encoder)
 		ret = connector_status_disconnected;
 
 	if (amdgpu_connector->ddc_bus)
-		dret = amdgpu_ddc_probe(amdgpu_connector, false);
+		dret = amdgpu_display_ddc_probe(amdgpu_connector, false);
 	if (dret) {
 		amdgpu_connector->detected_by_load = false;
 		amdgpu_connector_free_edid(connector);
@@ -927,8 +919,10 @@ amdgpu_connector_vga_detect(struct drm_connector *connector, bool force)
 	amdgpu_connector_update_scratch_regs(connector, ret);
 
 out:
-	pm_runtime_mark_last_busy(connector->dev->dev);
-	pm_runtime_put_autosuspend(connector->dev->dev);
+	if (!drm_kms_helper_is_poll_worker()) {
+		pm_runtime_mark_last_busy(connector->dev->dev);
+		pm_runtime_put_autosuspend(connector->dev->dev);
+	}
 
 	return ret;
 }
@@ -952,7 +946,7 @@ static bool
 amdgpu_connector_check_hpd_status_unchanged(struct drm_connector *connector)
 {
 	struct drm_device *dev = connector->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 	enum drm_connector_status status;
 
@@ -968,6 +962,41 @@ amdgpu_connector_check_hpd_status_unchanged(struct drm_connector *connector)
 	return false;
 }
 
+static void amdgpu_connector_shared_ddc(enum drm_connector_status *status,
+					struct drm_connector *connector,
+					struct amdgpu_connector *amdgpu_connector)
+{
+	struct drm_connector *list_connector;
+	struct drm_connector_list_iter iter;
+	struct amdgpu_connector *list_amdgpu_connector;
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+
+	if (amdgpu_connector->shared_ddc && *status == connector_status_connected) {
+		drm_connector_list_iter_begin(dev, &iter);
+		drm_for_each_connector_iter(list_connector,
+					    &iter) {
+			if (connector == list_connector)
+				continue;
+			list_amdgpu_connector = to_amdgpu_connector(list_connector);
+			if (list_amdgpu_connector->shared_ddc &&
+			    list_amdgpu_connector->ddc_bus->rec.i2c_id ==
+			     amdgpu_connector->ddc_bus->rec.i2c_id) {
+				/* cases where both connectors are digital */
+				if (list_connector->connector_type != DRM_MODE_CONNECTOR_VGA) {
+					/* hpd is our only option in this case */
+					if (!amdgpu_display_hpd_sense(adev,
+								      amdgpu_connector->hpd.hpd)) {
+						amdgpu_connector_free_edid(connector);
+						*status = connector_status_disconnected;
+					}
+				}
+			}
+		}
+		drm_connector_list_iter_end(&iter);
+	}
+}
+
 /*
  * DVI is complicated
  * Do a DDC probe, if DDC probe passes, get the full EDID so
@@ -983,25 +1012,48 @@ static enum drm_connector_status
 amdgpu_connector_dvi_detect(struct drm_connector *connector, bool force)
 {
 	struct drm_device *dev = connector->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
-	struct drm_encoder *encoder = NULL;
 	const struct drm_encoder_helper_funcs *encoder_funcs;
-	int i, r;
+	int r;
 	enum drm_connector_status ret = connector_status_disconnected;
 	bool dret = false, broken_edid = false;
 
-	r = pm_runtime_get_sync(connector->dev->dev);
-	if (r < 0)
-		return connector_status_disconnected;
+	if (!drm_kms_helper_is_poll_worker()) {
+		r = pm_runtime_get_sync(connector->dev->dev);
+		if (r < 0) {
+			pm_runtime_put_autosuspend(connector->dev->dev);
+			return connector_status_disconnected;
+		}
+	}
+
+	if (amdgpu_connector->detected_hpd_without_ddc) {
+		force = true;
+		amdgpu_connector->detected_hpd_without_ddc = false;
+	}
 
 	if (!force && amdgpu_connector_check_hpd_status_unchanged(connector)) {
 		ret = connector->status;
 		goto exit;
 	}
 
-	if (amdgpu_connector->ddc_bus)
-		dret = amdgpu_ddc_probe(amdgpu_connector, false);
+	if (amdgpu_connector->ddc_bus) {
+		dret = amdgpu_display_ddc_probe(amdgpu_connector, false);
+
+		/* Sometimes the pins required for the DDC probe on DVI
+		 * connectors don't make contact at the same time that the ones
+		 * for HPD do. If the DDC probe fails even though we had an HPD
+		 * signal, try again later
+		 */
+		if (!dret && !force &&
+		    amdgpu_display_hpd_sense(adev, amdgpu_connector->hpd.hpd)) {
+			DRM_DEBUG_KMS("hpd detected without ddc, retrying in 1 second\n");
+			amdgpu_connector->detected_hpd_without_ddc = true;
+			schedule_delayed_work(&adev->hotplug_work,
+					      msecs_to_jiffies(1000));
+			goto exit;
+		}
+	}
 	if (dret) {
 		amdgpu_connector->detected_by_load = false;
 		amdgpu_connector_free_edid(connector);
@@ -1031,27 +1083,7 @@ amdgpu_connector_dvi_detect(struct drm_connector *connector, bool force)
 			 * DDC line.  The latter is more complex because with DVI<->HDMI adapters
 			 * you don't really know what's connected to which port as both are digital.
 			 */
-			if (amdgpu_connector->shared_ddc && (ret == connector_status_connected)) {
-				struct drm_connector *list_connector;
-				struct amdgpu_connector *list_amdgpu_connector;
-				list_for_each_entry(list_connector, &dev->mode_config.connector_list, head) {
-					if (connector == list_connector)
-						continue;
-					list_amdgpu_connector = to_amdgpu_connector(list_connector);
-					if (list_amdgpu_connector->shared_ddc &&
-					    (list_amdgpu_connector->ddc_bus->rec.i2c_id ==
-					     amdgpu_connector->ddc_bus->rec.i2c_id)) {
-						/* cases where both connectors are digital */
-						if (list_connector->connector_type != DRM_MODE_CONNECTOR_VGA) {
-							/* hpd is our only option in this case */
-							if (!amdgpu_display_hpd_sense(adev, amdgpu_connector->hpd.hpd)) {
-								amdgpu_connector_free_edid(connector);
-								ret = connector_status_disconnected;
-							}
-						}
-					}
-				}
-			}
+			amdgpu_connector_shared_ddc(&ret, connector, amdgpu_connector);
 		}
 	}
 
@@ -1075,14 +1107,9 @@ amdgpu_connector_dvi_detect(struct drm_connector *connector, bool force)
 
 	/* find analog encoder */
 	if (amdgpu_connector->dac_load_detect) {
-		for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) {
-			if (connector->encoder_ids[i] == 0)
-				break;
-
-			encoder = drm_encoder_find(connector->dev, connector->encoder_ids[i]);
-			if (!encoder)
-				continue;
+		struct drm_encoder *encoder;
 
+		drm_connector_for_each_possible_encoder(connector, encoder) {
 			if (encoder->encoder_type != DRM_MODE_ENCODER_DAC &&
 			    encoder->encoder_type != DRM_MODE_ENCODER_TVDAC)
 				continue;
@@ -1104,7 +1131,8 @@ amdgpu_connector_dvi_detect(struct drm_connector *connector, bool force)
 					/* assume digital unless load detected otherwise */
 					amdgpu_connector->use_digital = true;
 					lret = encoder_funcs->detect(encoder, connector);
-					DRM_DEBUG_KMS("load_detect %x returned: %x\n",encoder->encoder_type,lret);
+					DRM_DEBUG_KMS("load_detect %x returned: %x\n",
+						      encoder->encoder_type, lret);
 					if (lret == connector_status_connected)
 						amdgpu_connector->use_digital = false;
 				}
@@ -1118,8 +1146,10 @@ out:
 	amdgpu_connector_update_scratch_regs(connector, ret);
 
 exit:
-	pm_runtime_mark_last_busy(connector->dev->dev);
-	pm_runtime_put_autosuspend(connector->dev->dev);
+	if (!drm_kms_helper_is_poll_worker()) {
+		pm_runtime_mark_last_busy(connector->dev->dev);
+		pm_runtime_put_autosuspend(connector->dev->dev);
+	}
 
 	return ret;
 }
@@ -1128,18 +1158,10 @@ exit:
 static struct drm_encoder *
 amdgpu_connector_dvi_encoder(struct drm_connector *connector)
 {
-	int enc_id = connector->encoder_ids[0];
 	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 	struct drm_encoder *encoder;
-	int i;
-	for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) {
-		if (connector->encoder_ids[i] == 0)
-			break;
-
-		encoder = drm_encoder_find(connector->dev, connector->encoder_ids[i]);
-		if (!encoder)
-			continue;
 
+	drm_connector_for_each_possible_encoder(connector, encoder) {
 		if (amdgpu_connector->use_digital == true) {
 			if (encoder->encoder_type == DRM_MODE_ENCODER_TMDS)
 				return encoder;
@@ -1154,43 +1176,85 @@ amdgpu_connector_dvi_encoder(struct drm_connector *connector)
 
 	/* then check use digitial */
 	/* pick the first one */
-	if (enc_id)
-		return drm_encoder_find(connector->dev, enc_id);
+	drm_connector_for_each_possible_encoder(connector, encoder)
+		return encoder;
+
 	return NULL;
 }
 
 static void amdgpu_connector_dvi_force(struct drm_connector *connector)
 {
 	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
 	if (connector->force == DRM_FORCE_ON)
 		amdgpu_connector->use_digital = false;
 	if (connector->force == DRM_FORCE_ON_DIGITAL)
 		amdgpu_connector->use_digital = true;
 }
 
-static int amdgpu_connector_dvi_mode_valid(struct drm_connector *connector,
-					    struct drm_display_mode *mode)
+/**
+ * amdgpu_max_hdmi_pixel_clock - Return max supported HDMI (TMDS) pixel clock
+ * @adev: pointer to amdgpu_device
+ *
+ * Return: maximum supported HDMI (TMDS) pixel clock in KHz.
+ */
+static int amdgpu_max_hdmi_pixel_clock(const struct amdgpu_device *adev)
+{
+	if (adev->asic_type >= CHIP_POLARIS10)
+		return 600000;
+	else if (adev->asic_type >= CHIP_TONGA)
+		return 300000;
+	else
+		return 297000;
+}
+
+/**
+ * amdgpu_connector_dvi_mode_valid - Validate a mode on DVI/HDMI connectors
+ * @connector: DRM connector to validate the mode on
+ * @mode: display mode to validate
+ *
+ * Validate the given display mode on DVI and HDMI connectors, including
+ * analog signals on DVI-I.
+ *
+ * Return: drm_mode_status indicating whether the mode is valid.
+ */
+static enum drm_mode_status amdgpu_connector_dvi_mode_valid(struct drm_connector *connector,
+					    const struct drm_display_mode *mode)
 {
 	struct drm_device *dev = connector->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	const int max_hdmi_pixel_clock = amdgpu_max_hdmi_pixel_clock(adev);
+	const int max_dvi_single_link_pixel_clock = 165000;
+	int max_digital_pixel_clock_khz;
 
 	/* XXX check mode bandwidth */
 
-	if (amdgpu_connector->use_digital && (mode->clock > 165000)) {
-		if ((amdgpu_connector->connector_object_id == CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I) ||
-		    (amdgpu_connector->connector_object_id == CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D) ||
-		    (amdgpu_connector->connector_object_id == CONNECTOR_OBJECT_ID_HDMI_TYPE_B)) {
-			return MODE_OK;
-		} else if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
-			/* HDMI 1.3+ supports max clock of 340 Mhz */
-			if (mode->clock > 340000)
-				return MODE_CLOCK_HIGH;
-			else
-				return MODE_OK;
-		} else {
-			return MODE_CLOCK_HIGH;
+	if (amdgpu_connector->use_digital) {
+		switch (amdgpu_connector->connector_object_id) {
+		case CONNECTOR_OBJECT_ID_HDMI_TYPE_A:
+			max_digital_pixel_clock_khz = max_hdmi_pixel_clock;
+			break;
+		case CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I:
+		case CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D:
+			max_digital_pixel_clock_khz = max_dvi_single_link_pixel_clock;
+			break;
+		case CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I:
+		case CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D:
+		case CONNECTOR_OBJECT_ID_HDMI_TYPE_B:
+			max_digital_pixel_clock_khz = max_dvi_single_link_pixel_clock * 2;
+			break;
 		}
+
+		/* When the display EDID claims that it's an HDMI display,
+		 * we use the HDMI encoder mode of the display HW,
+		 * so we should verify against the max HDMI clock here.
+		 */
+		if (connector->display_info.is_hdmi)
+			max_digital_pixel_clock_khz = max_hdmi_pixel_clock;
+
+		if (mode->clock > max_digital_pixel_clock_khz)
+			return MODE_CLOCK_HIGH;
 	}
 
 	/* check against the max pixel clock */
@@ -1290,17 +1354,8 @@ u16 amdgpu_connector_encoder_get_dp_bridge_encoder_id(struct drm_connector *conn
 {
 	struct drm_encoder *encoder;
 	struct amdgpu_encoder *amdgpu_encoder;
-	int i;
-
-	for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) {
-		if (connector->encoder_ids[i] == 0)
-			break;
-
-		encoder = drm_encoder_find(connector->dev,
-					connector->encoder_ids[i]);
-		if (!encoder)
-			continue;
 
+	drm_connector_for_each_possible_encoder(connector, encoder) {
 		amdgpu_encoder = to_amdgpu_encoder(encoder);
 
 		switch (amdgpu_encoder->encoder_id) {
@@ -1319,17 +1374,9 @@ static bool amdgpu_connector_encoder_is_hbr2(struct drm_connector *connector)
 {
 	struct drm_encoder *encoder;
 	struct amdgpu_encoder *amdgpu_encoder;
-	int i;
 	bool found = false;
 
-	for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) {
-		if (connector->encoder_ids[i] == 0)
-			break;
-		encoder = drm_encoder_find(connector->dev,
-					connector->encoder_ids[i]);
-		if (!encoder)
-			continue;
-
+	drm_connector_for_each_possible_encoder(connector, encoder) {
 		amdgpu_encoder = to_amdgpu_encoder(encoder);
 		if (amdgpu_encoder->caps & ATOM_ENCODER_CAP_RECORD_HBR2)
 			found = true;
@@ -1341,7 +1388,7 @@ static bool amdgpu_connector_encoder_is_hbr2(struct drm_connector *connector)
 bool amdgpu_connector_is_dp12_capable(struct drm_connector *connector)
 {
 	struct drm_device *dev = connector->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 
 	if ((adev->clock.default_dispclk >= 53900) &&
 	    amdgpu_connector_encoder_is_hbr2(connector)) {
@@ -1355,16 +1402,20 @@ static enum drm_connector_status
 amdgpu_connector_dp_detect(struct drm_connector *connector, bool force)
 {
 	struct drm_device *dev = connector->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 	enum drm_connector_status ret = connector_status_disconnected;
 	struct amdgpu_connector_atom_dig *amdgpu_dig_connector = amdgpu_connector->con_priv;
 	struct drm_encoder *encoder = amdgpu_connector_best_single_encoder(connector);
 	int r;
 
-	r = pm_runtime_get_sync(connector->dev->dev);
-	if (r < 0)
-		return connector_status_disconnected;
+	if (!drm_kms_helper_is_poll_worker()) {
+		r = pm_runtime_get_sync(connector->dev->dev);
+		if (r < 0) {
+			pm_runtime_put_autosuspend(connector->dev->dev);
+			return connector_status_disconnected;
+		}
+	}
 
 	if (!force && amdgpu_connector_check_hpd_status_unchanged(connector)) {
 		ret = connector->status;
@@ -1404,10 +1455,12 @@ amdgpu_connector_dp_detect(struct drm_connector *connector, bool force)
 			/* setup ddc on the bridge */
 			amdgpu_atombios_encoder_setup_ext_encoder_ddc(encoder);
 			/* bridge chips are always aux */
-			if (amdgpu_ddc_probe(amdgpu_connector, true)) /* try DDC */
+			/* try DDC */
+			if (amdgpu_display_ddc_probe(amdgpu_connector, true))
 				ret = connector_status_connected;
 			else if (amdgpu_connector->dac_load_detect) { /* try load detection */
 				const struct drm_encoder_helper_funcs *encoder_funcs = encoder->helper_private;
+
 				ret = encoder_funcs->detect(encoder, connector);
 			}
 		}
@@ -1424,7 +1477,8 @@ amdgpu_connector_dp_detect(struct drm_connector *connector, bool force)
 					ret = connector_status_connected;
 			} else {
 				/* try non-aux ddc (DP to DVI/HDMI/etc. adapter) */
-				if (amdgpu_ddc_probe(amdgpu_connector, false))
+				if (amdgpu_display_ddc_probe(amdgpu_connector,
+							     false))
 					ret = connector_status_connected;
 			}
 		}
@@ -1432,14 +1486,22 @@ amdgpu_connector_dp_detect(struct drm_connector *connector, bool force)
 
 	amdgpu_connector_update_scratch_regs(connector, ret);
 out:
-	pm_runtime_mark_last_busy(connector->dev->dev);
-	pm_runtime_put_autosuspend(connector->dev->dev);
+	if (!drm_kms_helper_is_poll_worker()) {
+		pm_runtime_mark_last_busy(connector->dev->dev);
+		pm_runtime_put_autosuspend(connector->dev->dev);
+	}
 
+	if (connector->connector_type == DRM_MODE_CONNECTOR_DisplayPort ||
+	    connector->connector_type == DRM_MODE_CONNECTOR_eDP)
+		drm_dp_set_subconnector_property(&amdgpu_connector->base,
+						 ret,
+						 amdgpu_dig_connector->dpcd,
+						 amdgpu_dig_connector->downstream_ports);
 	return ret;
 }
 
-static int amdgpu_connector_dp_mode_valid(struct drm_connector *connector,
-					   struct drm_display_mode *mode)
+static enum drm_mode_status amdgpu_connector_dp_mode_valid(struct drm_connector *connector,
+					   const struct drm_display_mode *mode)
 {
 	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 	struct amdgpu_connector_atom_dig *amdgpu_dig_connector = amdgpu_connector->con_priv;
@@ -1477,7 +1539,7 @@ static int amdgpu_connector_dp_mode_valid(struct drm_connector *connector,
 		    (amdgpu_dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_eDP)) {
 			return amdgpu_atombios_dp_mode_valid_helper(connector, mode);
 		} else {
-			if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+			if (connector->display_info.is_hdmi) {
 				/* HDMI 1.3+ supports max clock of 340 Mhz */
 				if (mode->clock > 340000)
 					return MODE_CLOCK_HIGH;
@@ -1491,6 +1553,20 @@ static int amdgpu_connector_dp_mode_valid(struct drm_connector *connector,
 	return MODE_OK;
 }
 
+static int
+amdgpu_connector_late_register(struct drm_connector *connector)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	int r = 0;
+
+	if (amdgpu_connector->ddc_bus->has_aux) {
+		amdgpu_connector->ddc_bus->aux.dev = amdgpu_connector->base.kdev;
+		r = drm_dp_aux_register(&amdgpu_connector->ddc_bus->aux);
+	}
+
+	return r;
+}
+
 static const struct drm_connector_helper_funcs amdgpu_connector_dp_helper_funcs = {
 	.get_modes = amdgpu_connector_dp_get_modes,
 	.mode_valid = amdgpu_connector_dp_mode_valid,
@@ -1505,6 +1581,7 @@ static const struct drm_connector_funcs amdgpu_connector_dp_funcs = {
 	.early_unregister = amdgpu_connector_unregister,
 	.destroy = amdgpu_connector_destroy,
 	.force = amdgpu_connector_dvi_force,
+	.late_register = amdgpu_connector_late_register,
 };
 
 static const struct drm_connector_funcs amdgpu_connector_edp_funcs = {
@@ -1515,6 +1592,7 @@ static const struct drm_connector_funcs amdgpu_connector_edp_funcs = {
 	.early_unregister = amdgpu_connector_unregister,
 	.destroy = amdgpu_connector_destroy,
 	.force = amdgpu_connector_dvi_force,
+	.late_register = amdgpu_connector_late_register,
 };
 
 void
@@ -1527,12 +1605,14 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 		      struct amdgpu_hpd *hpd,
 		      struct amdgpu_router *router)
 {
-	struct drm_device *dev = adev->ddev;
+	struct drm_device *dev = adev_to_drm(adev);
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector;
 	struct amdgpu_connector_atom_dig *amdgpu_dig_connector;
 	struct drm_encoder *encoder;
 	struct amdgpu_encoder *amdgpu_encoder;
+	struct i2c_adapter *ddc = NULL;
 	uint32_t subpixel_order = SubPixelNone;
 	bool shared_ddc = false;
 	bool is_dp_bridge = false;
@@ -1542,10 +1622,12 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 		return;
 
 	/* see if we already added it */
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		amdgpu_connector = to_amdgpu_connector(connector);
 		if (amdgpu_connector->connector_id == connector_id) {
 			amdgpu_connector->devices |= supported_device;
+			drm_connector_list_iter_end(&iter);
 			return;
 		}
 		if (amdgpu_connector->ddc_bus && i2c_bus->valid) {
@@ -1560,6 +1642,7 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 			}
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	/* check if it's a dp bridge */
 	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
@@ -1602,17 +1685,21 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 		amdgpu_connector->con_priv = amdgpu_dig_connector;
 		if (i2c_bus->valid) {
 			amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
-			if (amdgpu_connector->ddc_bus)
+			if (amdgpu_connector->ddc_bus) {
 				has_aux = true;
-			else
+				ddc = &amdgpu_connector->ddc_bus->adapter;
+			} else {
 				DRM_ERROR("DP: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+			}
 		}
 		switch (connector_type) {
 		case DRM_MODE_CONNECTOR_VGA:
 		case DRM_MODE_CONNECTOR_DVIA:
 		default:
-			drm_connector_init(dev, &amdgpu_connector->base,
-					   &amdgpu_connector_dp_funcs, connector_type);
+			drm_connector_init_with_ddc(dev, &amdgpu_connector->base,
+						    &amdgpu_connector_dp_funcs,
+						    connector_type,
+						    ddc);
 			drm_connector_helper_add(&amdgpu_connector->base,
 						 &amdgpu_connector_dp_helper_funcs);
 			connector->interlace_allowed = true;
@@ -1630,8 +1717,10 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 		case DRM_MODE_CONNECTOR_HDMIA:
 		case DRM_MODE_CONNECTOR_HDMIB:
 		case DRM_MODE_CONNECTOR_DisplayPort:
-			drm_connector_init(dev, &amdgpu_connector->base,
-					   &amdgpu_connector_dp_funcs, connector_type);
+			drm_connector_init_with_ddc(dev, &amdgpu_connector->base,
+						    &amdgpu_connector_dp_funcs,
+						    connector_type,
+						    ddc);
 			drm_connector_helper_add(&amdgpu_connector->base,
 						 &amdgpu_connector_dp_helper_funcs);
 			drm_object_attach_property(&amdgpu_connector->base.base,
@@ -1652,10 +1741,12 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 						   adev->mode_info.dither_property,
 						   AMDGPU_FMT_DITHER_DISABLE);
 
-			if (amdgpu_audio != 0)
+			if (amdgpu_audio != 0) {
 				drm_object_attach_property(&amdgpu_connector->base.base,
 							   adev->mode_info.audio_property,
 							   AMDGPU_AUDIO_AUTO);
+				amdgpu_connector->audio = AMDGPU_AUDIO_AUTO;
+			}
 
 			subpixel_order = SubPixelHorizontalRGB;
 			connector->interlace_allowed = true;
@@ -1672,8 +1763,10 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 			break;
 		case DRM_MODE_CONNECTOR_LVDS:
 		case DRM_MODE_CONNECTOR_eDP:
-			drm_connector_init(dev, &amdgpu_connector->base,
-					   &amdgpu_connector_edp_funcs, connector_type);
+			drm_connector_init_with_ddc(dev, &amdgpu_connector->base,
+						    &amdgpu_connector_edp_funcs,
+						    connector_type,
+						    ddc);
 			drm_connector_helper_add(&amdgpu_connector->base,
 						 &amdgpu_connector_dp_helper_funcs);
 			drm_object_attach_property(&amdgpu_connector->base.base,
@@ -1687,13 +1780,18 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 	} else {
 		switch (connector_type) {
 		case DRM_MODE_CONNECTOR_VGA:
-			drm_connector_init(dev, &amdgpu_connector->base, &amdgpu_connector_vga_funcs, connector_type);
-			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_vga_helper_funcs);
 			if (i2c_bus->valid) {
 				amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
 				if (!amdgpu_connector->ddc_bus)
 					DRM_ERROR("VGA: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+				else
+					ddc = &amdgpu_connector->ddc_bus->adapter;
 			}
+			drm_connector_init_with_ddc(dev, &amdgpu_connector->base,
+						    &amdgpu_connector_vga_funcs,
+						    connector_type,
+						    ddc);
+			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_vga_helper_funcs);
 			amdgpu_connector->dac_load_detect = true;
 			drm_object_attach_property(&amdgpu_connector->base.base,
 						      adev->mode_info.load_detect_property,
@@ -1707,13 +1805,18 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 			connector->doublescan_allowed = true;
 			break;
 		case DRM_MODE_CONNECTOR_DVIA:
-			drm_connector_init(dev, &amdgpu_connector->base, &amdgpu_connector_vga_funcs, connector_type);
-			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_vga_helper_funcs);
 			if (i2c_bus->valid) {
 				amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
 				if (!amdgpu_connector->ddc_bus)
 					DRM_ERROR("DVIA: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+				else
+					ddc = &amdgpu_connector->ddc_bus->adapter;
 			}
+			drm_connector_init_with_ddc(dev, &amdgpu_connector->base,
+						    &amdgpu_connector_vga_funcs,
+						    connector_type,
+						    ddc);
+			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_vga_helper_funcs);
 			amdgpu_connector->dac_load_detect = true;
 			drm_object_attach_property(&amdgpu_connector->base.base,
 						      adev->mode_info.load_detect_property,
@@ -1732,13 +1835,18 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 			if (!amdgpu_dig_connector)
 				goto failed;
 			amdgpu_connector->con_priv = amdgpu_dig_connector;
-			drm_connector_init(dev, &amdgpu_connector->base, &amdgpu_connector_dvi_funcs, connector_type);
-			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_dvi_helper_funcs);
 			if (i2c_bus->valid) {
 				amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
 				if (!amdgpu_connector->ddc_bus)
 					DRM_ERROR("DVI: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+				else
+					ddc = &amdgpu_connector->ddc_bus->adapter;
 			}
+			drm_connector_init_with_ddc(dev, &amdgpu_connector->base,
+						    &amdgpu_connector_dvi_funcs,
+						    connector_type,
+						    ddc);
+			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_dvi_helper_funcs);
 			subpixel_order = SubPixelHorizontalRGB;
 			drm_object_attach_property(&amdgpu_connector->base.base,
 						      adev->mode_info.coherent_mode_property,
@@ -1760,6 +1868,7 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 				drm_object_attach_property(&amdgpu_connector->base.base,
 							   adev->mode_info.audio_property,
 							   AMDGPU_AUDIO_AUTO);
+				amdgpu_connector->audio = AMDGPU_AUDIO_AUTO;
 			}
 			drm_object_attach_property(&amdgpu_connector->base.base,
 						   adev->mode_info.dither_property,
@@ -1782,13 +1891,18 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 			if (!amdgpu_dig_connector)
 				goto failed;
 			amdgpu_connector->con_priv = amdgpu_dig_connector;
-			drm_connector_init(dev, &amdgpu_connector->base, &amdgpu_connector_dvi_funcs, connector_type);
-			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_dvi_helper_funcs);
 			if (i2c_bus->valid) {
 				amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
 				if (!amdgpu_connector->ddc_bus)
 					DRM_ERROR("HDMI: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+				else
+					ddc = &amdgpu_connector->ddc_bus->adapter;
 			}
+			drm_connector_init_with_ddc(dev, &amdgpu_connector->base,
+						    &amdgpu_connector_dvi_funcs,
+						    connector_type,
+						    ddc);
+			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_dvi_helper_funcs);
 			drm_object_attach_property(&amdgpu_connector->base.base,
 						      adev->mode_info.coherent_mode_property,
 						      1);
@@ -1808,6 +1922,7 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 				drm_object_attach_property(&amdgpu_connector->base.base,
 							   adev->mode_info.audio_property,
 							   AMDGPU_AUDIO_AUTO);
+				amdgpu_connector->audio = AMDGPU_AUDIO_AUTO;
 			}
 			drm_object_attach_property(&amdgpu_connector->base.base,
 						   adev->mode_info.dither_property,
@@ -1824,15 +1939,20 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 			if (!amdgpu_dig_connector)
 				goto failed;
 			amdgpu_connector->con_priv = amdgpu_dig_connector;
-			drm_connector_init(dev, &amdgpu_connector->base, &amdgpu_connector_dp_funcs, connector_type);
-			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_dp_helper_funcs);
 			if (i2c_bus->valid) {
 				amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
-				if (amdgpu_connector->ddc_bus)
+				if (amdgpu_connector->ddc_bus) {
 					has_aux = true;
-				else
+					ddc = &amdgpu_connector->ddc_bus->adapter;
+				} else {
 					DRM_ERROR("DP: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+				}
 			}
+			drm_connector_init_with_ddc(dev, &amdgpu_connector->base,
+						    &amdgpu_connector_dp_funcs,
+						    connector_type,
+						    ddc);
+			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_dp_helper_funcs);
 			subpixel_order = SubPixelHorizontalRGB;
 			drm_object_attach_property(&amdgpu_connector->base.base,
 						      adev->mode_info.coherent_mode_property,
@@ -1853,6 +1973,7 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 				drm_object_attach_property(&amdgpu_connector->base.base,
 							   adev->mode_info.audio_property,
 							   AMDGPU_AUDIO_AUTO);
+				amdgpu_connector->audio = AMDGPU_AUDIO_AUTO;
 			}
 			drm_object_attach_property(&amdgpu_connector->base.base,
 						   adev->mode_info.dither_property,
@@ -1866,15 +1987,20 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 			if (!amdgpu_dig_connector)
 				goto failed;
 			amdgpu_connector->con_priv = amdgpu_dig_connector;
-			drm_connector_init(dev, &amdgpu_connector->base, &amdgpu_connector_edp_funcs, connector_type);
-			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_dp_helper_funcs);
 			if (i2c_bus->valid) {
 				amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
-				if (amdgpu_connector->ddc_bus)
+				if (amdgpu_connector->ddc_bus) {
 					has_aux = true;
-				else
+					ddc = &amdgpu_connector->ddc_bus->adapter;
+				} else {
 					DRM_ERROR("DP: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+				}
 			}
+			drm_connector_init_with_ddc(dev, &amdgpu_connector->base,
+						    &amdgpu_connector_edp_funcs,
+						    connector_type,
+						    ddc);
+			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_dp_helper_funcs);
 			drm_object_attach_property(&amdgpu_connector->base.base,
 						      dev->mode_config.scaling_mode_property,
 						      DRM_MODE_SCALE_FULLSCREEN);
@@ -1887,13 +2013,18 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 			if (!amdgpu_dig_connector)
 				goto failed;
 			amdgpu_connector->con_priv = amdgpu_dig_connector;
-			drm_connector_init(dev, &amdgpu_connector->base, &amdgpu_connector_lvds_funcs, connector_type);
-			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_lvds_helper_funcs);
 			if (i2c_bus->valid) {
 				amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
 				if (!amdgpu_connector->ddc_bus)
 					DRM_ERROR("LVDS: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+				else
+					ddc = &amdgpu_connector->ddc_bus->adapter;
 			}
+			drm_connector_init_with_ddc(dev, &amdgpu_connector->base,
+						    &amdgpu_connector_lvds_funcs,
+						    connector_type,
+						    ddc);
+			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_lvds_helper_funcs);
 			drm_object_attach_property(&amdgpu_connector->base.base,
 						      dev->mode_config.scaling_mode_property,
 						      DRM_MODE_SCALE_FULLSCREEN);
@@ -1907,17 +2038,21 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 	if (amdgpu_connector->hpd.hpd == AMDGPU_HPD_NONE) {
 		if (i2c_bus->valid) {
 			connector->polled = DRM_CONNECTOR_POLL_CONNECT |
-			                    DRM_CONNECTOR_POLL_DISCONNECT;
+						DRM_CONNECTOR_POLL_DISCONNECT;
 		}
 	} else
 		connector->polled = DRM_CONNECTOR_POLL_HPD;
 
 	connector->display_info.subpixel_order = subpixel_order;
-	drm_connector_register(connector);
 
 	if (has_aux)
 		amdgpu_atombios_dp_aux_init(amdgpu_connector);
 
+	if (connector_type == DRM_MODE_CONNECTOR_DisplayPort ||
+	    connector_type == DRM_MODE_CONNECTOR_eDP) {
+		drm_connector_attach_dp_subconnector_property(&amdgpu_connector->base);
+	}
+
 	return;
 
 failed:
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.h
index 61fcef15ad72..eff833b6ed31 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.h
@@ -24,7 +24,6 @@
 #ifndef __AMDGPU_CONNECTORS_H__
 #define __AMDGPU_CONNECTORS_H__
 
-struct edid *amdgpu_connector_edid(struct drm_connector *connector);
 void amdgpu_connector_hotplug(struct drm_connector *connector);
 int amdgpu_connector_get_monitor_bpc(struct drm_connector *connector);
 u16 amdgpu_connector_encoder_get_dp_bridge_encoder_id(struct drm_connector *connector);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c
new file mode 100644
index 000000000000..ef996493115f
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c
@@ -0,0 +1,591 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright 2025 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/list.h>
+#include "amdgpu.h"
+
+static const guid_t MCE			= CPER_NOTIFY_MCE;
+static const guid_t CMC			= CPER_NOTIFY_CMC;
+static const guid_t BOOT		= BOOT_TYPE;
+
+static const guid_t CRASHDUMP		= AMD_CRASHDUMP;
+static const guid_t RUNTIME		= AMD_GPU_NONSTANDARD_ERROR;
+
+static void __inc_entry_length(struct cper_hdr *hdr, uint32_t size)
+{
+	hdr->record_length += size;
+}
+
+static void amdgpu_cper_get_timestamp(struct cper_timestamp *timestamp)
+{
+	struct tm tm;
+	time64_t now = ktime_get_real_seconds();
+
+	time64_to_tm(now, 0, &tm);
+	timestamp->seconds = tm.tm_sec;
+	timestamp->minutes = tm.tm_min;
+	timestamp->hours = tm.tm_hour;
+	timestamp->flag = 0;
+	timestamp->day = tm.tm_mday;
+	timestamp->month = 1 + tm.tm_mon;
+	timestamp->year = (1900 + tm.tm_year) % 100;
+	timestamp->century = (1900 + tm.tm_year) / 100;
+}
+
+void amdgpu_cper_entry_fill_hdr(struct amdgpu_device *adev,
+				struct cper_hdr *hdr,
+				enum amdgpu_cper_type type,
+				enum cper_error_severity sev)
+{
+	char record_id[16];
+
+	hdr->signature[0]		= 'C';
+	hdr->signature[1]		= 'P';
+	hdr->signature[2]		= 'E';
+	hdr->signature[3]		= 'R';
+	hdr->revision			= CPER_HDR_REV_1;
+	hdr->signature_end		= 0xFFFFFFFF;
+	hdr->error_severity		= sev;
+
+	hdr->valid_bits.platform_id	= 1;
+	hdr->valid_bits.timestamp	= 1;
+
+	amdgpu_cper_get_timestamp(&hdr->timestamp);
+
+	snprintf(record_id, 9, "%d:%X",
+		 (adev->smuio.funcs && adev->smuio.funcs->get_socket_id) ?
+			 adev->smuio.funcs->get_socket_id(adev) :
+			 0,
+		 atomic_inc_return(&adev->cper.unique_id));
+	memcpy(hdr->record_id, record_id, 8);
+
+	snprintf(hdr->platform_id, 16, "0x%04X:0x%04X",
+		 adev->pdev->vendor, adev->pdev->device);
+	/* pmfw version should be part of creator_id according to CPER spec */
+	snprintf(hdr->creator_id, 16, "%s", CPER_CREATOR_ID_AMDGPU);
+
+	switch (type) {
+	case AMDGPU_CPER_TYPE_BOOT:
+		hdr->notify_type = BOOT;
+		break;
+	case AMDGPU_CPER_TYPE_FATAL:
+	case AMDGPU_CPER_TYPE_BP_THRESHOLD:
+		hdr->notify_type = MCE;
+		break;
+	case AMDGPU_CPER_TYPE_RUNTIME:
+		if (sev == CPER_SEV_NON_FATAL_CORRECTED)
+			hdr->notify_type = CMC;
+		else
+			hdr->notify_type = MCE;
+		break;
+	default:
+		dev_err(adev->dev, "Unknown CPER Type\n");
+		break;
+	}
+
+	__inc_entry_length(hdr, HDR_LEN);
+}
+
+static int amdgpu_cper_entry_fill_section_desc(struct amdgpu_device *adev,
+					       struct cper_sec_desc *section_desc,
+					       bool bp_threshold,
+					       bool poison,
+					       enum cper_error_severity sev,
+					       guid_t sec_type,
+					       uint32_t section_length,
+					       uint32_t section_offset)
+{
+	section_desc->revision_minor		= CPER_SEC_MINOR_REV_1;
+	section_desc->revision_major		= CPER_SEC_MAJOR_REV_22;
+	section_desc->sec_offset		= section_offset;
+	section_desc->sec_length		= section_length;
+	section_desc->valid_bits.fru_text	= 1;
+	section_desc->flag_bits.primary		= 1;
+	section_desc->severity			= sev;
+	section_desc->sec_type			= sec_type;
+
+	snprintf(section_desc->fru_text, 20, "OAM%d",
+		 (adev->smuio.funcs && adev->smuio.funcs->get_socket_id) ?
+			 adev->smuio.funcs->get_socket_id(adev) :
+			 0);
+
+	if (bp_threshold)
+		section_desc->flag_bits.exceed_err_threshold = 1;
+	if (poison)
+		section_desc->flag_bits.latent_err = 1;
+
+	return 0;
+}
+
+int amdgpu_cper_entry_fill_fatal_section(struct amdgpu_device *adev,
+					 struct cper_hdr *hdr,
+					 uint32_t idx,
+					 struct cper_sec_crashdump_reg_data reg_data)
+{
+	struct cper_sec_desc *section_desc;
+	struct cper_sec_crashdump_fatal *section;
+
+	section_desc = (struct cper_sec_desc *)((uint8_t *)hdr + SEC_DESC_OFFSET(idx));
+	section = (struct cper_sec_crashdump_fatal *)((uint8_t *)hdr +
+		   FATAL_SEC_OFFSET(hdr->sec_cnt, idx));
+
+	amdgpu_cper_entry_fill_section_desc(adev, section_desc, false, false,
+					    CPER_SEV_FATAL, CRASHDUMP, FATAL_SEC_LEN,
+					    FATAL_SEC_OFFSET(hdr->sec_cnt, idx));
+
+	section->body.reg_ctx_type = CPER_CTX_TYPE_CRASH;
+	section->body.reg_arr_size = sizeof(reg_data);
+	section->body.data = reg_data;
+
+	__inc_entry_length(hdr, SEC_DESC_LEN + FATAL_SEC_LEN);
+
+	return 0;
+}
+
+int amdgpu_cper_entry_fill_runtime_section(struct amdgpu_device *adev,
+					   struct cper_hdr *hdr,
+					   uint32_t idx,
+					   enum cper_error_severity sev,
+					   uint32_t *reg_dump,
+					   uint32_t reg_count)
+{
+	struct cper_sec_desc *section_desc;
+	struct cper_sec_nonstd_err *section;
+	bool poison;
+
+	poison = sev != CPER_SEV_NON_FATAL_CORRECTED;
+	section_desc = (struct cper_sec_desc *)((uint8_t *)hdr + SEC_DESC_OFFSET(idx));
+	section = (struct cper_sec_nonstd_err *)((uint8_t *)hdr +
+		   NONSTD_SEC_OFFSET(hdr->sec_cnt, idx));
+
+	amdgpu_cper_entry_fill_section_desc(adev, section_desc, false, poison,
+					    sev, RUNTIME, NONSTD_SEC_LEN,
+					    NONSTD_SEC_OFFSET(hdr->sec_cnt, idx));
+
+	reg_count = umin(reg_count, CPER_ACA_REG_COUNT);
+
+	section->hdr.valid_bits.err_info_cnt = 1;
+	section->hdr.valid_bits.err_context_cnt = 1;
+
+	section->info.error_type = RUNTIME;
+	section->info.ms_chk_bits.err_type_valid = 1;
+	section->ctx.reg_ctx_type = CPER_CTX_TYPE_CRASH;
+	section->ctx.reg_arr_size = sizeof(section->ctx.reg_dump);
+
+	memcpy(section->ctx.reg_dump, reg_dump, reg_count * sizeof(uint32_t));
+
+	__inc_entry_length(hdr, SEC_DESC_LEN + NONSTD_SEC_LEN);
+
+	return 0;
+}
+
+int amdgpu_cper_entry_fill_bad_page_threshold_section(struct amdgpu_device *adev,
+						      struct cper_hdr *hdr,
+						      uint32_t idx)
+{
+	struct cper_sec_desc *section_desc;
+	struct cper_sec_nonstd_err *section;
+	uint32_t socket_id;
+
+	section_desc = (struct cper_sec_desc *)((uint8_t *)hdr + SEC_DESC_OFFSET(idx));
+	section = (struct cper_sec_nonstd_err *)((uint8_t *)hdr +
+		   NONSTD_SEC_OFFSET(hdr->sec_cnt, idx));
+
+	amdgpu_cper_entry_fill_section_desc(adev, section_desc, true, false,
+					    CPER_SEV_FATAL, RUNTIME, NONSTD_SEC_LEN,
+					    NONSTD_SEC_OFFSET(hdr->sec_cnt, idx));
+
+	section->hdr.valid_bits.err_info_cnt = 1;
+	section->hdr.valid_bits.err_context_cnt = 1;
+
+	section->info.error_type = RUNTIME;
+	section->info.valid_bits.ms_chk = 1;
+	section->info.ms_chk_bits.err_type_valid = 1;
+	section->info.ms_chk_bits.err_type = 1;
+	section->info.ms_chk_bits.pcc = 1;
+	section->ctx.reg_ctx_type = CPER_CTX_TYPE_CRASH;
+	section->ctx.reg_arr_size = sizeof(section->ctx.reg_dump);
+
+	/* Hardcoded Reg dump for bad page threshold CPER */
+	socket_id = (adev->smuio.funcs && adev->smuio.funcs->get_socket_id) ?
+				adev->smuio.funcs->get_socket_id(adev) :
+				0;
+	section->ctx.reg_dump[CPER_ACA_REG_CTL_LO]    = 0x1;
+	section->ctx.reg_dump[CPER_ACA_REG_CTL_HI]    = 0x0;
+	section->ctx.reg_dump[CPER_ACA_REG_STATUS_LO] = 0x137;
+	section->ctx.reg_dump[CPER_ACA_REG_STATUS_HI] = 0xB0000000;
+	section->ctx.reg_dump[CPER_ACA_REG_ADDR_LO]   = 0x0;
+	section->ctx.reg_dump[CPER_ACA_REG_ADDR_HI]   = 0x0;
+	section->ctx.reg_dump[CPER_ACA_REG_MISC0_LO]  = 0x0;
+	section->ctx.reg_dump[CPER_ACA_REG_MISC0_HI]  = 0x0;
+	section->ctx.reg_dump[CPER_ACA_REG_CONFIG_LO] = 0x2;
+	section->ctx.reg_dump[CPER_ACA_REG_CONFIG_HI] = 0x1ff;
+	section->ctx.reg_dump[CPER_ACA_REG_IPID_LO]   = (socket_id / 4) & 0x01;
+	section->ctx.reg_dump[CPER_ACA_REG_IPID_HI]   = 0x096 | (((socket_id % 4) & 0x3) << 12);
+	section->ctx.reg_dump[CPER_ACA_REG_SYND_LO]   = 0x0;
+	section->ctx.reg_dump[CPER_ACA_REG_SYND_HI]   = 0x0;
+
+	__inc_entry_length(hdr, SEC_DESC_LEN + NONSTD_SEC_LEN);
+
+	return 0;
+}
+
+struct cper_hdr *amdgpu_cper_alloc_entry(struct amdgpu_device *adev,
+					 enum amdgpu_cper_type type,
+					 uint16_t section_count)
+{
+	struct cper_hdr *hdr;
+	uint32_t size = 0;
+
+	size += HDR_LEN;
+	size += (SEC_DESC_LEN * section_count);
+
+	switch (type) {
+	case AMDGPU_CPER_TYPE_RUNTIME:
+	case AMDGPU_CPER_TYPE_BP_THRESHOLD:
+		size += (NONSTD_SEC_LEN * section_count);
+		break;
+	case AMDGPU_CPER_TYPE_FATAL:
+		size += (FATAL_SEC_LEN * section_count);
+		break;
+	case AMDGPU_CPER_TYPE_BOOT:
+		size += (BOOT_SEC_LEN * section_count);
+		break;
+	default:
+		dev_err(adev->dev, "Unknown CPER Type!\n");
+		return NULL;
+	}
+
+	hdr = kzalloc(size, GFP_KERNEL);
+	if (!hdr)
+		return NULL;
+
+	/* Save this early */
+	hdr->sec_cnt = section_count;
+
+	return hdr;
+}
+
+int amdgpu_cper_generate_ue_record(struct amdgpu_device *adev,
+				   struct aca_bank *bank)
+{
+	struct cper_hdr *fatal = NULL;
+	struct cper_sec_crashdump_reg_data reg_data = { 0 };
+	struct amdgpu_ring *ring = &adev->cper.ring_buf;
+	int ret;
+
+	fatal = amdgpu_cper_alloc_entry(adev, AMDGPU_CPER_TYPE_FATAL, 1);
+	if (!fatal) {
+		dev_err(adev->dev, "fail to alloc cper entry for ue record\n");
+		return -ENOMEM;
+	}
+
+	reg_data.status_lo = lower_32_bits(bank->regs[ACA_REG_IDX_STATUS]);
+	reg_data.status_hi = upper_32_bits(bank->regs[ACA_REG_IDX_STATUS]);
+	reg_data.addr_lo   = lower_32_bits(bank->regs[ACA_REG_IDX_ADDR]);
+	reg_data.addr_hi   = upper_32_bits(bank->regs[ACA_REG_IDX_ADDR]);
+	reg_data.ipid_lo   = lower_32_bits(bank->regs[ACA_REG_IDX_IPID]);
+	reg_data.ipid_hi   = upper_32_bits(bank->regs[ACA_REG_IDX_IPID]);
+	reg_data.synd_lo   = lower_32_bits(bank->regs[ACA_REG_IDX_SYND]);
+	reg_data.synd_hi   = upper_32_bits(bank->regs[ACA_REG_IDX_SYND]);
+
+	amdgpu_cper_entry_fill_hdr(adev, fatal, AMDGPU_CPER_TYPE_FATAL, CPER_SEV_FATAL);
+	ret = amdgpu_cper_entry_fill_fatal_section(adev, fatal, 0, reg_data);
+	if (ret)
+		return ret;
+
+	amdgpu_cper_ring_write(ring, fatal, fatal->record_length);
+	kfree(fatal);
+
+	return 0;
+}
+
+int amdgpu_cper_generate_bp_threshold_record(struct amdgpu_device *adev)
+{
+	struct cper_hdr *bp_threshold = NULL;
+	struct amdgpu_ring *ring = &adev->cper.ring_buf;
+	int ret;
+
+	bp_threshold = amdgpu_cper_alloc_entry(adev, AMDGPU_CPER_TYPE_BP_THRESHOLD, 1);
+	if (!bp_threshold) {
+		dev_err(adev->dev, "fail to alloc cper entry for bad page threshold record\n");
+		return -ENOMEM;
+	}
+
+	amdgpu_cper_entry_fill_hdr(adev, bp_threshold,
+				   AMDGPU_CPER_TYPE_BP_THRESHOLD,
+				   CPER_SEV_FATAL);
+	ret = amdgpu_cper_entry_fill_bad_page_threshold_section(adev, bp_threshold, 0);
+	if (ret)
+		return ret;
+
+	amdgpu_cper_ring_write(ring, bp_threshold, bp_threshold->record_length);
+	kfree(bp_threshold);
+
+	return 0;
+}
+
+static enum cper_error_severity amdgpu_aca_err_type_to_cper_sev(struct amdgpu_device *adev,
+								enum aca_error_type aca_err_type)
+{
+	switch (aca_err_type) {
+	case ACA_ERROR_TYPE_UE:
+		return CPER_SEV_FATAL;
+	case ACA_ERROR_TYPE_CE:
+		return CPER_SEV_NON_FATAL_CORRECTED;
+	case ACA_ERROR_TYPE_DEFERRED:
+		return CPER_SEV_NON_FATAL_UNCORRECTED;
+	default:
+		dev_err(adev->dev, "Unknown ACA error type!\n");
+		return CPER_SEV_FATAL;
+	}
+}
+
+int amdgpu_cper_generate_ce_records(struct amdgpu_device *adev,
+				    struct aca_banks *banks,
+				    uint16_t bank_count)
+{
+	struct cper_hdr *corrected = NULL;
+	enum cper_error_severity sev = CPER_SEV_NON_FATAL_CORRECTED;
+	struct amdgpu_ring *ring = &adev->cper.ring_buf;
+	uint32_t reg_data[CPER_ACA_REG_COUNT] = { 0 };
+	struct aca_bank_node *node;
+	struct aca_bank *bank;
+	uint32_t i = 0;
+	int ret;
+
+	corrected = amdgpu_cper_alloc_entry(adev, AMDGPU_CPER_TYPE_RUNTIME, bank_count);
+	if (!corrected) {
+		dev_err(adev->dev, "fail to allocate cper entry for ce records\n");
+		return -ENOMEM;
+	}
+
+	/* Raise severity if any DE is detected in the ACA bank list */
+	list_for_each_entry(node, &banks->list, node) {
+		bank = &node->bank;
+		if (bank->aca_err_type == ACA_ERROR_TYPE_DEFERRED) {
+			sev = CPER_SEV_NON_FATAL_UNCORRECTED;
+			break;
+		}
+	}
+
+	amdgpu_cper_entry_fill_hdr(adev, corrected, AMDGPU_CPER_TYPE_RUNTIME, sev);
+
+	/* Combine CE and DE in cper record */
+	list_for_each_entry(node, &banks->list, node) {
+		bank = &node->bank;
+		reg_data[CPER_ACA_REG_CTL_LO]    = lower_32_bits(bank->regs[ACA_REG_IDX_CTL]);
+		reg_data[CPER_ACA_REG_CTL_HI]    = upper_32_bits(bank->regs[ACA_REG_IDX_CTL]);
+		reg_data[CPER_ACA_REG_STATUS_LO] = lower_32_bits(bank->regs[ACA_REG_IDX_STATUS]);
+		reg_data[CPER_ACA_REG_STATUS_HI] = upper_32_bits(bank->regs[ACA_REG_IDX_STATUS]);
+		reg_data[CPER_ACA_REG_ADDR_LO]   = lower_32_bits(bank->regs[ACA_REG_IDX_ADDR]);
+		reg_data[CPER_ACA_REG_ADDR_HI]   = upper_32_bits(bank->regs[ACA_REG_IDX_ADDR]);
+		reg_data[CPER_ACA_REG_MISC0_LO]  = lower_32_bits(bank->regs[ACA_REG_IDX_MISC0]);
+		reg_data[CPER_ACA_REG_MISC0_HI]  = upper_32_bits(bank->regs[ACA_REG_IDX_MISC0]);
+		reg_data[CPER_ACA_REG_CONFIG_LO] = lower_32_bits(bank->regs[ACA_REG_IDX_CONFIG]);
+		reg_data[CPER_ACA_REG_CONFIG_HI] = upper_32_bits(bank->regs[ACA_REG_IDX_CONFIG]);
+		reg_data[CPER_ACA_REG_IPID_LO]   = lower_32_bits(bank->regs[ACA_REG_IDX_IPID]);
+		reg_data[CPER_ACA_REG_IPID_HI]   = upper_32_bits(bank->regs[ACA_REG_IDX_IPID]);
+		reg_data[CPER_ACA_REG_SYND_LO]   = lower_32_bits(bank->regs[ACA_REG_IDX_SYND]);
+		reg_data[CPER_ACA_REG_SYND_HI]   = upper_32_bits(bank->regs[ACA_REG_IDX_SYND]);
+
+		ret = amdgpu_cper_entry_fill_runtime_section(adev, corrected, i++,
+				amdgpu_aca_err_type_to_cper_sev(adev, bank->aca_err_type),
+				reg_data, CPER_ACA_REG_COUNT);
+		if (ret)
+			return ret;
+	}
+
+	amdgpu_cper_ring_write(ring, corrected, corrected->record_length);
+	kfree(corrected);
+
+	return 0;
+}
+
+static bool amdgpu_cper_is_hdr(struct amdgpu_ring *ring, u64 pos)
+{
+	struct cper_hdr *chdr;
+
+	chdr = (struct cper_hdr *)&(ring->ring[pos]);
+	return strcmp(chdr->signature, "CPER") ? false : true;
+}
+
+static u32 amdgpu_cper_ring_get_ent_sz(struct amdgpu_ring *ring, u64 pos)
+{
+	struct cper_hdr *chdr;
+	u64 p;
+	u32 chunk, rec_len = 0;
+
+	chdr = (struct cper_hdr *)&(ring->ring[pos]);
+	chunk = ring->ring_size - (pos << 2);
+
+	if (!strcmp(chdr->signature, "CPER")) {
+		rec_len = chdr->record_length;
+		goto calc;
+	}
+
+	/* ring buffer is not full, no cper data after ring->wptr */
+	if (ring->count_dw)
+		goto calc;
+
+	for (p = pos + 1; p <= ring->buf_mask; p++) {
+		chdr = (struct cper_hdr *)&(ring->ring[p]);
+		if (!strcmp(chdr->signature, "CPER")) {
+			rec_len = (p - pos) << 2;
+			goto calc;
+		}
+	}
+
+calc:
+	if (!rec_len)
+		return chunk;
+	else
+		return umin(rec_len, chunk);
+}
+
+void amdgpu_cper_ring_write(struct amdgpu_ring *ring, void *src, int count)
+{
+	u64 pos, wptr_old, rptr;
+	int rec_cnt_dw = count >> 2;
+	u32 chunk, ent_sz;
+	u8 *s = (u8 *)src;
+
+	if (count >= ring->ring_size - 4) {
+		dev_err(ring->adev->dev,
+			"CPER data size(%d) is larger than ring size(%d)\n",
+			count, ring->ring_size - 4);
+
+		return;
+	}
+
+	mutex_lock(&ring->adev->cper.ring_lock);
+
+	wptr_old = ring->wptr;
+	rptr = *ring->rptr_cpu_addr & ring->ptr_mask;
+
+	while (count) {
+		ent_sz = amdgpu_cper_ring_get_ent_sz(ring, ring->wptr);
+		chunk = umin(ent_sz, count);
+
+		memcpy(&ring->ring[ring->wptr], s, chunk);
+
+		ring->wptr += (chunk >> 2);
+		ring->wptr &= ring->ptr_mask;
+		count -= chunk;
+		s += chunk;
+	}
+
+	if (ring->count_dw < rec_cnt_dw)
+		ring->count_dw = 0;
+
+	/* the buffer is overflow, adjust rptr */
+	if (((wptr_old < rptr) && (rptr <= ring->wptr)) ||
+	    ((ring->wptr < wptr_old) && (wptr_old < rptr)) ||
+	    ((rptr <= ring->wptr) && (ring->wptr < wptr_old))) {
+		pos = (ring->wptr + 1) & ring->ptr_mask;
+
+		do {
+			ent_sz = amdgpu_cper_ring_get_ent_sz(ring, pos);
+
+			rptr += (ent_sz >> 2);
+			rptr &= ring->ptr_mask;
+			*ring->rptr_cpu_addr = rptr;
+
+			pos = rptr;
+		} while (!amdgpu_cper_is_hdr(ring, rptr));
+	}
+
+	if (ring->count_dw >= rec_cnt_dw)
+		ring->count_dw -= rec_cnt_dw;
+	mutex_unlock(&ring->adev->cper.ring_lock);
+}
+
+static u64 amdgpu_cper_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	return *(ring->rptr_cpu_addr);
+}
+
+static u64 amdgpu_cper_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	return ring->wptr;
+}
+
+static const struct amdgpu_ring_funcs cper_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_CPER,
+	.align_mask = 0xff,
+	.support_64bit_ptrs = false,
+	.get_rptr = amdgpu_cper_ring_get_rptr,
+	.get_wptr = amdgpu_cper_ring_get_wptr,
+};
+
+static int amdgpu_cper_ring_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = &(adev->cper.ring_buf);
+
+	mutex_init(&adev->cper.ring_lock);
+
+	ring->adev = NULL;
+	ring->ring_obj = NULL;
+	ring->use_doorbell = false;
+	ring->no_scheduler = true;
+	ring->funcs = &cper_ring_funcs;
+
+	sprintf(ring->name, "cper");
+	return amdgpu_ring_init(adev, ring, CPER_MAX_RING_SIZE, NULL, 0,
+				AMDGPU_RING_PRIO_DEFAULT, NULL);
+}
+
+int amdgpu_cper_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (!amdgpu_aca_is_enabled(adev) && !amdgpu_sriov_ras_cper_en(adev))
+		return 0;
+
+	r = amdgpu_cper_ring_init(adev);
+	if (r) {
+		dev_err(adev->dev, "failed to initialize cper ring, r = %d\n", r);
+		return r;
+	}
+
+	mutex_init(&adev->cper.cper_lock);
+
+	adev->cper.enabled = true;
+	adev->cper.max_count = CPER_MAX_ALLOWED_COUNT;
+
+	return 0;
+}
+
+int amdgpu_cper_fini(struct amdgpu_device *adev)
+{
+	if (!amdgpu_aca_is_enabled(adev) && !amdgpu_sriov_ras_cper_en(adev))
+		return 0;
+
+	adev->cper.enabled = false;
+
+	amdgpu_ring_fini(&(adev->cper.ring_buf));
+	adev->cper.count = 0;
+	adev->cper.wptr = 0;
+
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.h
new file mode 100644
index 000000000000..bcb97d245673
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.h
@@ -0,0 +1,105 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright 2025 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_CPER_H__
+#define __AMDGPU_CPER_H__
+
+#include "amd_cper.h"
+#include "amdgpu_aca.h"
+
+#define CPER_MAX_ALLOWED_COUNT		0x1000
+#define CPER_MAX_RING_SIZE		0X100000
+#define HDR_LEN				(sizeof(struct cper_hdr))
+#define SEC_DESC_LEN			(sizeof(struct cper_sec_desc))
+
+#define BOOT_SEC_LEN			(sizeof(struct cper_sec_crashdump_boot))
+#define FATAL_SEC_LEN			(sizeof(struct cper_sec_crashdump_fatal))
+#define NONSTD_SEC_LEN			(sizeof(struct cper_sec_nonstd_err))
+
+#define SEC_DESC_OFFSET(idx)		(HDR_LEN + (SEC_DESC_LEN * idx))
+
+#define BOOT_SEC_OFFSET(count, idx)	(HDR_LEN + (SEC_DESC_LEN * count) + (BOOT_SEC_LEN * idx))
+#define FATAL_SEC_OFFSET(count, idx)	(HDR_LEN + (SEC_DESC_LEN * count) + (FATAL_SEC_LEN * idx))
+#define NONSTD_SEC_OFFSET(count, idx)	(HDR_LEN + (SEC_DESC_LEN * count) + (NONSTD_SEC_LEN * idx))
+
+enum amdgpu_cper_type {
+	AMDGPU_CPER_TYPE_RUNTIME,
+	AMDGPU_CPER_TYPE_FATAL,
+	AMDGPU_CPER_TYPE_BOOT,
+	AMDGPU_CPER_TYPE_BP_THRESHOLD,
+};
+
+struct amdgpu_cper {
+	bool enabled;
+
+	atomic_t unique_id;
+	struct mutex cper_lock;
+
+	/* Lifetime CPERs generated */
+	uint32_t count;
+	uint32_t max_count;
+
+	uint32_t wptr;
+
+	void *ring[CPER_MAX_ALLOWED_COUNT];
+	struct amdgpu_ring ring_buf;
+	struct mutex ring_lock;
+};
+
+void amdgpu_cper_entry_fill_hdr(struct amdgpu_device *adev,
+				struct cper_hdr *hdr,
+				enum amdgpu_cper_type type,
+				enum cper_error_severity sev);
+int amdgpu_cper_entry_fill_fatal_section(struct amdgpu_device *adev,
+					 struct cper_hdr *hdr,
+					 uint32_t idx,
+					 struct cper_sec_crashdump_reg_data reg_data);
+int amdgpu_cper_entry_fill_runtime_section(struct amdgpu_device *adev,
+					   struct cper_hdr *hdr,
+					   uint32_t idx,
+					   enum cper_error_severity sev,
+					   uint32_t *reg_dump,
+					   uint32_t reg_count);
+int amdgpu_cper_entry_fill_bad_page_threshold_section(struct amdgpu_device *adev,
+						      struct cper_hdr *hdr,
+						      uint32_t section_idx);
+
+struct cper_hdr *amdgpu_cper_alloc_entry(struct amdgpu_device *adev,
+					 enum amdgpu_cper_type type,
+					 uint16_t section_count);
+/* UE must be encoded into separated cper entries, 1 UE 1 cper */
+int amdgpu_cper_generate_ue_record(struct amdgpu_device *adev,
+				   struct aca_bank *bank);
+/* CEs and DEs are combined into 1 cper entry */
+int amdgpu_cper_generate_ce_records(struct amdgpu_device *adev,
+				    struct aca_banks *banks,
+				    uint16_t bank_count);
+/* Bad page threshold is encoded into separated cper entry */
+int amdgpu_cper_generate_bp_threshold_record(struct amdgpu_device *adev);
+void amdgpu_cper_ring_write(struct amdgpu_ring *ring,
+			void *src, int count);
+int amdgpu_cper_init(struct amdgpu_device *adev);
+int amdgpu_cper_fini(struct amdgpu_device *adev);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 60d8bedb694d..9cd7741d2254 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -24,16 +24,108 @@
  * Authors:
  *    Jerome Glisse <glisse@freedesktop.org>
  */
+
+#include <linux/file.h>
 #include <linux/pagemap.h>
-#include <drm/drmP.h>
+#include <linux/sync_file.h>
+#include <linux/dma-buf.h>
+#include <linux/hmm.h>
+
 #include <drm/amdgpu_drm.h>
 #include <drm/drm_syncobj.h>
+#include <drm/ttm/ttm_tt.h>
+
+#include "amdgpu_cs.h"
 #include "amdgpu.h"
 #include "amdgpu_trace.h"
+#include "amdgpu_gmc.h"
+#include "amdgpu_gem.h"
+#include "amdgpu_ras.h"
+
+static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p,
+				 struct amdgpu_device *adev,
+				 struct drm_file *filp,
+				 union drm_amdgpu_cs *cs)
+{
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+
+	if (cs->in.num_chunks == 0)
+		return -EINVAL;
+
+	memset(p, 0, sizeof(*p));
+	p->adev = adev;
+	p->filp = filp;
+
+	p->ctx = amdgpu_ctx_get(fpriv, cs->in.ctx_id);
+	if (!p->ctx)
+		return -EINVAL;
+
+	if (atomic_read(&p->ctx->guilty)) {
+		amdgpu_ctx_put(p->ctx);
+		return -ECANCELED;
+	}
+
+	amdgpu_sync_create(&p->sync);
+	drm_exec_init(&p->exec, DRM_EXEC_INTERRUPTIBLE_WAIT |
+		      DRM_EXEC_IGNORE_DUPLICATES, 0);
+	return 0;
+}
+
+static int amdgpu_cs_job_idx(struct amdgpu_cs_parser *p,
+			     struct drm_amdgpu_cs_chunk_ib *chunk_ib)
+{
+	struct drm_sched_entity *entity;
+	unsigned int i;
+	int r;
+
+	r = amdgpu_ctx_get_entity(p->ctx, chunk_ib->ip_type,
+				  chunk_ib->ip_instance,
+				  chunk_ib->ring, &entity);
+	if (r)
+		return r;
 
-static int amdgpu_cs_user_fence_chunk(struct amdgpu_cs_parser *p,
-				      struct drm_amdgpu_cs_chunk_fence *data,
-				      uint32_t *offset)
+	/*
+	 * Abort if there is no run queue associated with this entity.
+	 * Possibly because of disabled HW IP.
+	 */
+	if (entity->rq == NULL)
+		return -EINVAL;
+
+	/* Check if we can add this IB to some existing job */
+	for (i = 0; i < p->gang_size; ++i)
+		if (p->entities[i] == entity)
+			return i;
+
+	/* If not increase the gang size if possible */
+	if (i == AMDGPU_CS_GANG_SIZE)
+		return -EINVAL;
+
+	p->entities[i] = entity;
+	p->gang_size = i + 1;
+	return i;
+}
+
+static int amdgpu_cs_p1_ib(struct amdgpu_cs_parser *p,
+			   struct drm_amdgpu_cs_chunk_ib *chunk_ib,
+			   unsigned int *num_ibs)
+{
+	int r;
+
+	r = amdgpu_cs_job_idx(p, chunk_ib);
+	if (r < 0)
+		return r;
+
+	if (num_ibs[r] >= amdgpu_ring_max_ibs(chunk_ib->ip_type))
+		return -EINVAL;
+
+	++(num_ibs[r]);
+	p->gang_leader_idx = r;
+	return 0;
+}
+
+static int amdgpu_cs_p1_user_fence(struct amdgpu_cs_parser *p,
+				   struct drm_amdgpu_cs_chunk_fence *data,
+				   uint32_t *offset)
 {
 	struct drm_gem_object *gobj;
 	unsigned long size;
@@ -42,73 +134,74 @@ static int amdgpu_cs_user_fence_chunk(struct amdgpu_cs_parser *p,
 	if (gobj == NULL)
 		return -EINVAL;
 
-	p->uf_entry.robj = amdgpu_bo_ref(gem_to_amdgpu_bo(gobj));
-	p->uf_entry.priority = 0;
-	p->uf_entry.tv.bo = &p->uf_entry.robj->tbo;
-	p->uf_entry.tv.shared = true;
-	p->uf_entry.user_pages = NULL;
+	p->uf_bo = amdgpu_bo_ref(gem_to_amdgpu_bo(gobj));
+	drm_gem_object_put(gobj);
+
+	size = amdgpu_bo_size(p->uf_bo);
+	if (size != PAGE_SIZE || data->offset > (size - 8))
+		return -EINVAL;
 
-	size = amdgpu_bo_size(p->uf_entry.robj);
-	if (size != PAGE_SIZE || (data->offset + 8) > size)
+	if (amdgpu_ttm_tt_get_usermm(p->uf_bo->tbo.ttm))
 		return -EINVAL;
 
 	*offset = data->offset;
+	return 0;
+}
 
-	drm_gem_object_put_unlocked(gobj);
+static int amdgpu_cs_p1_bo_handles(struct amdgpu_cs_parser *p,
+				   struct drm_amdgpu_bo_list_in *data)
+{
+	struct drm_amdgpu_bo_list_entry *info;
+	int r;
 
-	if (amdgpu_ttm_tt_get_usermm(p->uf_entry.robj->tbo.ttm)) {
-		amdgpu_bo_unref(&p->uf_entry.robj);
-		return -EINVAL;
-	}
+	r = amdgpu_bo_create_list_entry_array(data, &info);
+	if (r)
+		return r;
+
+	r = amdgpu_bo_list_create(p->adev, p->filp, info, data->bo_number,
+				  &p->bo_list);
+	if (r)
+		goto error_free;
 
+	kvfree(info);
 	return 0;
+
+error_free:
+	kvfree(info);
+
+	return r;
 }
 
-static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
+/* Copy the data from userspace and go over it the first time */
+static int amdgpu_cs_pass1(struct amdgpu_cs_parser *p,
+			   union drm_amdgpu_cs *cs)
 {
 	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
+	unsigned int num_ibs[AMDGPU_CS_GANG_SIZE] = { };
 	struct amdgpu_vm *vm = &fpriv->vm;
-	union drm_amdgpu_cs *cs = data;
-	uint64_t *chunk_array_user;
 	uint64_t *chunk_array;
-	unsigned size, num_ibs = 0;
 	uint32_t uf_offset = 0;
-	int i;
+	size_t size;
 	int ret;
+	int i;
 
-	if (cs->in.num_chunks == 0)
-		return 0;
-
-	chunk_array = kmalloc_array(cs->in.num_chunks, sizeof(uint64_t), GFP_KERNEL);
-	if (!chunk_array)
-		return -ENOMEM;
-
-	p->ctx = amdgpu_ctx_get(fpriv, cs->in.ctx_id);
-	if (!p->ctx) {
-		ret = -EINVAL;
-		goto free_chunk;
-	}
-
-	/* get chunks */
-	chunk_array_user = u64_to_user_ptr(cs->in.chunks);
-	if (copy_from_user(chunk_array, chunk_array_user,
-			   sizeof(uint64_t)*cs->in.num_chunks)) {
-		ret = -EFAULT;
-		goto put_ctx;
-	}
+	chunk_array = memdup_array_user(u64_to_user_ptr(cs->in.chunks),
+					cs->in.num_chunks,
+					sizeof(uint64_t));
+	if (IS_ERR(chunk_array))
+		return PTR_ERR(chunk_array);
 
 	p->nchunks = cs->in.num_chunks;
-	p->chunks = kmalloc_array(p->nchunks, sizeof(struct amdgpu_cs_chunk),
+	p->chunks = kvmalloc_array(p->nchunks, sizeof(struct amdgpu_cs_chunk),
 			    GFP_KERNEL);
 	if (!p->chunks) {
 		ret = -ENOMEM;
-		goto put_ctx;
+		goto free_chunk;
 	}
 
 	for (i = 0; i < p->nchunks; i++) {
-		struct drm_amdgpu_cs_chunk __user **chunk_ptr = NULL;
+		struct drm_amdgpu_cs_chunk __user *chunk_ptr = NULL;
 		struct drm_amdgpu_cs_chunk user_chunk;
-		uint32_t __user *cdata;
 
 		chunk_ptr = u64_to_user_ptr(chunk_array[i]);
 		if (copy_from_user(&user_chunk, chunk_ptr,
@@ -121,57 +214,111 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
 		p->chunks[i].length_dw = user_chunk.length_dw;
 
 		size = p->chunks[i].length_dw;
-		cdata = u64_to_user_ptr(user_chunk.chunk_data);
 
-		p->chunks[i].kdata = kvmalloc_array(size, sizeof(uint32_t), GFP_KERNEL);
-		if (p->chunks[i].kdata == NULL) {
-			ret = -ENOMEM;
+		p->chunks[i].kdata = vmemdup_array_user(u64_to_user_ptr(user_chunk.chunk_data),
+							size,
+							sizeof(uint32_t));
+		if (IS_ERR(p->chunks[i].kdata)) {
+			ret = PTR_ERR(p->chunks[i].kdata);
 			i--;
 			goto free_partial_kdata;
 		}
 		size *= sizeof(uint32_t);
-		if (copy_from_user(p->chunks[i].kdata, cdata, size)) {
-			ret = -EFAULT;
-			goto free_partial_kdata;
-		}
 
+		/* Assume the worst on the following checks */
+		ret = -EINVAL;
 		switch (p->chunks[i].chunk_id) {
 		case AMDGPU_CHUNK_ID_IB:
-			++num_ibs;
+			if (size < sizeof(struct drm_amdgpu_cs_chunk_ib))
+				goto free_partial_kdata;
+
+			ret = amdgpu_cs_p1_ib(p, p->chunks[i].kdata, num_ibs);
+			if (ret)
+				goto free_partial_kdata;
 			break;
 
 		case AMDGPU_CHUNK_ID_FENCE:
-			size = sizeof(struct drm_amdgpu_cs_chunk_fence);
-			if (p->chunks[i].length_dw * sizeof(uint32_t) < size) {
-				ret = -EINVAL;
+			if (size < sizeof(struct drm_amdgpu_cs_chunk_fence))
 				goto free_partial_kdata;
-			}
 
-			ret = amdgpu_cs_user_fence_chunk(p, p->chunks[i].kdata,
-							 &uf_offset);
+			ret = amdgpu_cs_p1_user_fence(p, p->chunks[i].kdata,
+						      &uf_offset);
 			if (ret)
 				goto free_partial_kdata;
+			break;
+
+		case AMDGPU_CHUNK_ID_BO_HANDLES:
+			if (size < sizeof(struct drm_amdgpu_bo_list_in))
+				goto free_partial_kdata;
 
+			/* Only a single BO list is allowed to simplify handling. */
+			if (p->bo_list)
+				goto free_partial_kdata;
+
+			ret = amdgpu_cs_p1_bo_handles(p, p->chunks[i].kdata);
+			if (ret)
+				goto free_partial_kdata;
 			break;
 
 		case AMDGPU_CHUNK_ID_DEPENDENCIES:
 		case AMDGPU_CHUNK_ID_SYNCOBJ_IN:
 		case AMDGPU_CHUNK_ID_SYNCOBJ_OUT:
+		case AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES:
+		case AMDGPU_CHUNK_ID_SYNCOBJ_TIMELINE_WAIT:
+		case AMDGPU_CHUNK_ID_SYNCOBJ_TIMELINE_SIGNAL:
+		case AMDGPU_CHUNK_ID_CP_GFX_SHADOW:
 			break;
 
 		default:
-			ret = -EINVAL;
 			goto free_partial_kdata;
 		}
 	}
 
-	ret = amdgpu_job_alloc(p->adev, num_ibs, &p->job, vm);
-	if (ret)
+	if (!p->gang_size || (amdgpu_sriov_vf(p->adev) && p->gang_size > 1)) {
+		ret = -EINVAL;
+		goto free_all_kdata;
+	}
+
+	for (i = 0; i < p->gang_size; ++i) {
+		ret = amdgpu_job_alloc(p->adev, vm, p->entities[i], vm,
+				       num_ibs[i], &p->jobs[i],
+				       p->filp->client_id);
+		if (ret)
+			goto free_all_kdata;
+		switch (p->adev->enforce_isolation[fpriv->xcp_id]) {
+		case AMDGPU_ENFORCE_ISOLATION_DISABLE:
+		default:
+			p->jobs[i]->enforce_isolation = false;
+			p->jobs[i]->run_cleaner_shader = false;
+			break;
+		case AMDGPU_ENFORCE_ISOLATION_ENABLE:
+			p->jobs[i]->enforce_isolation = true;
+			p->jobs[i]->run_cleaner_shader = true;
+			break;
+		case AMDGPU_ENFORCE_ISOLATION_ENABLE_LEGACY:
+			p->jobs[i]->enforce_isolation = true;
+			p->jobs[i]->run_cleaner_shader = false;
+			break;
+		case AMDGPU_ENFORCE_ISOLATION_NO_CLEANER_SHADER:
+			p->jobs[i]->enforce_isolation = true;
+			p->jobs[i]->run_cleaner_shader = false;
+			break;
+		}
+	}
+	p->gang_leader = p->jobs[p->gang_leader_idx];
+
+	if (p->ctx->generation != p->gang_leader->generation) {
+		ret = -ECANCELED;
 		goto free_all_kdata;
+	}
+
+	if (p->uf_bo)
+		p->gang_leader->uf_addr = uf_offset;
+	kvfree(chunk_array);
+
+	/* Use this opportunity to fill in task info for the vm */
+	amdgpu_vm_set_task_info(vm);
 
-	if (p->uf_entry.robj)
-		p->job->uf_addr = uf_offset;
-	kfree(chunk_array);
 	return 0;
 
 free_all_kdata:
@@ -179,17 +326,334 @@ free_all_kdata:
 free_partial_kdata:
 	for (; i >= 0; i--)
 		kvfree(p->chunks[i].kdata);
-	kfree(p->chunks);
+	kvfree(p->chunks);
 	p->chunks = NULL;
 	p->nchunks = 0;
-put_ctx:
-	amdgpu_ctx_put(p->ctx);
 free_chunk:
-	kfree(chunk_array);
+	kvfree(chunk_array);
 
 	return ret;
 }
 
+static int amdgpu_cs_p2_ib(struct amdgpu_cs_parser *p,
+			   struct amdgpu_cs_chunk *chunk,
+			   unsigned int *ce_preempt,
+			   unsigned int *de_preempt)
+{
+	struct drm_amdgpu_cs_chunk_ib *chunk_ib = chunk->kdata;
+	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
+	struct amdgpu_vm *vm = &fpriv->vm;
+	struct amdgpu_ring *ring;
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	int r;
+
+	r = amdgpu_cs_job_idx(p, chunk_ib);
+	if (r < 0)
+		return r;
+
+	job = p->jobs[r];
+	ring = amdgpu_job_ring(job);
+	ib = &job->ibs[job->num_ibs++];
+
+	/* submissions to kernel queues are disabled */
+	if (ring->no_user_submission)
+		return -EINVAL;
+
+	/* MM engine doesn't support user fences */
+	if (p->uf_bo && ring->funcs->no_user_fence)
+		return -EINVAL;
+
+	if (chunk_ib->ip_type == AMDGPU_HW_IP_GFX &&
+	    chunk_ib->flags & AMDGPU_IB_FLAG_PREEMPT) {
+		if (chunk_ib->flags & AMDGPU_IB_FLAG_CE)
+			(*ce_preempt)++;
+		else
+			(*de_preempt)++;
+
+		/* Each GFX command submit allows only 1 IB max
+		 * preemptible for CE & DE */
+		if (*ce_preempt > 1 || *de_preempt > 1)
+			return -EINVAL;
+	}
+
+	if (chunk_ib->flags & AMDGPU_IB_FLAG_PREAMBLE)
+		job->preamble_status |= AMDGPU_PREAMBLE_IB_PRESENT;
+
+	r =  amdgpu_ib_get(p->adev, vm, ring->funcs->parse_cs ?
+			   chunk_ib->ib_bytes : 0,
+			   AMDGPU_IB_POOL_DELAYED, ib);
+	if (r) {
+		drm_err(adev_to_drm(p->adev), "Failed to get ib !\n");
+		return r;
+	}
+
+	ib->gpu_addr = chunk_ib->va_start;
+	ib->length_dw = chunk_ib->ib_bytes / 4;
+	ib->flags = chunk_ib->flags;
+	return 0;
+}
+
+static int amdgpu_cs_p2_dependencies(struct amdgpu_cs_parser *p,
+				     struct amdgpu_cs_chunk *chunk)
+{
+	struct drm_amdgpu_cs_chunk_dep *deps = chunk->kdata;
+	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
+	unsigned int num_deps;
+	int i, r;
+
+	num_deps = chunk->length_dw * 4 /
+		sizeof(struct drm_amdgpu_cs_chunk_dep);
+
+	for (i = 0; i < num_deps; ++i) {
+		struct amdgpu_ctx *ctx;
+		struct drm_sched_entity *entity;
+		struct dma_fence *fence;
+
+		ctx = amdgpu_ctx_get(fpriv, deps[i].ctx_id);
+		if (ctx == NULL)
+			return -EINVAL;
+
+		r = amdgpu_ctx_get_entity(ctx, deps[i].ip_type,
+					  deps[i].ip_instance,
+					  deps[i].ring, &entity);
+		if (r) {
+			amdgpu_ctx_put(ctx);
+			return r;
+		}
+
+		fence = amdgpu_ctx_get_fence(ctx, entity, deps[i].handle);
+		amdgpu_ctx_put(ctx);
+
+		if (IS_ERR(fence))
+			return PTR_ERR(fence);
+		else if (!fence)
+			continue;
+
+		if (chunk->chunk_id == AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES) {
+			struct drm_sched_fence *s_fence;
+			struct dma_fence *old = fence;
+
+			s_fence = to_drm_sched_fence(fence);
+			fence = dma_fence_get(&s_fence->scheduled);
+			dma_fence_put(old);
+		}
+
+		r = amdgpu_sync_fence(&p->sync, fence, GFP_KERNEL);
+		dma_fence_put(fence);
+		if (r)
+			return r;
+	}
+	return 0;
+}
+
+static int amdgpu_syncobj_lookup_and_add(struct amdgpu_cs_parser *p,
+					 uint32_t handle, u64 point,
+					 u64 flags)
+{
+	struct dma_fence *fence;
+	int r;
+
+	r = drm_syncobj_find_fence(p->filp, handle, point, flags, &fence);
+	if (r) {
+		drm_err(adev_to_drm(p->adev), "syncobj %u failed to find fence @ %llu (%d)!\n",
+			  handle, point, r);
+		return r;
+	}
+
+	r = amdgpu_sync_fence(&p->sync, fence, GFP_KERNEL);
+	dma_fence_put(fence);
+	return r;
+}
+
+static int amdgpu_cs_p2_syncobj_in(struct amdgpu_cs_parser *p,
+				   struct amdgpu_cs_chunk *chunk)
+{
+	struct drm_amdgpu_cs_chunk_sem *deps = chunk->kdata;
+	unsigned int num_deps;
+	int i, r;
+
+	num_deps = chunk->length_dw * 4 /
+		sizeof(struct drm_amdgpu_cs_chunk_sem);
+	for (i = 0; i < num_deps; ++i) {
+		r = amdgpu_syncobj_lookup_and_add(p, deps[i].handle, 0, 0);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+static int amdgpu_cs_p2_syncobj_timeline_wait(struct amdgpu_cs_parser *p,
+					      struct amdgpu_cs_chunk *chunk)
+{
+	struct drm_amdgpu_cs_chunk_syncobj *syncobj_deps = chunk->kdata;
+	unsigned int num_deps;
+	int i, r;
+
+	num_deps = chunk->length_dw * 4 /
+		sizeof(struct drm_amdgpu_cs_chunk_syncobj);
+	for (i = 0; i < num_deps; ++i) {
+		r = amdgpu_syncobj_lookup_and_add(p, syncobj_deps[i].handle,
+						  syncobj_deps[i].point,
+						  syncobj_deps[i].flags);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+static int amdgpu_cs_p2_syncobj_out(struct amdgpu_cs_parser *p,
+				    struct amdgpu_cs_chunk *chunk)
+{
+	struct drm_amdgpu_cs_chunk_sem *deps = chunk->kdata;
+	unsigned int num_deps;
+	int i;
+
+	num_deps = chunk->length_dw * 4 /
+		sizeof(struct drm_amdgpu_cs_chunk_sem);
+
+	if (p->post_deps)
+		return -EINVAL;
+
+	p->post_deps = kmalloc_array(num_deps, sizeof(*p->post_deps),
+				     GFP_KERNEL);
+	p->num_post_deps = 0;
+
+	if (!p->post_deps)
+		return -ENOMEM;
+
+
+	for (i = 0; i < num_deps; ++i) {
+		p->post_deps[i].syncobj =
+			drm_syncobj_find(p->filp, deps[i].handle);
+		if (!p->post_deps[i].syncobj)
+			return -EINVAL;
+		p->post_deps[i].chain = NULL;
+		p->post_deps[i].point = 0;
+		p->num_post_deps++;
+	}
+
+	return 0;
+}
+
+static int amdgpu_cs_p2_syncobj_timeline_signal(struct amdgpu_cs_parser *p,
+						struct amdgpu_cs_chunk *chunk)
+{
+	struct drm_amdgpu_cs_chunk_syncobj *syncobj_deps = chunk->kdata;
+	unsigned int num_deps;
+	int i;
+
+	num_deps = chunk->length_dw * 4 /
+		sizeof(struct drm_amdgpu_cs_chunk_syncobj);
+
+	if (p->post_deps)
+		return -EINVAL;
+
+	p->post_deps = kmalloc_array(num_deps, sizeof(*p->post_deps),
+				     GFP_KERNEL);
+	p->num_post_deps = 0;
+
+	if (!p->post_deps)
+		return -ENOMEM;
+
+	for (i = 0; i < num_deps; ++i) {
+		struct amdgpu_cs_post_dep *dep = &p->post_deps[i];
+
+		dep->chain = NULL;
+		if (syncobj_deps[i].point) {
+			dep->chain = dma_fence_chain_alloc();
+			if (!dep->chain)
+				return -ENOMEM;
+		}
+
+		dep->syncobj = drm_syncobj_find(p->filp,
+						syncobj_deps[i].handle);
+		if (!dep->syncobj) {
+			dma_fence_chain_free(dep->chain);
+			return -EINVAL;
+		}
+		dep->point = syncobj_deps[i].point;
+		p->num_post_deps++;
+	}
+
+	return 0;
+}
+
+static int amdgpu_cs_p2_shadow(struct amdgpu_cs_parser *p,
+			       struct amdgpu_cs_chunk *chunk)
+{
+	struct drm_amdgpu_cs_chunk_cp_gfx_shadow *shadow = chunk->kdata;
+	int i;
+
+	if (shadow->flags & ~AMDGPU_CS_CHUNK_CP_GFX_SHADOW_FLAGS_INIT_SHADOW)
+		return -EINVAL;
+
+	for (i = 0; i < p->gang_size; ++i) {
+		p->jobs[i]->shadow_va = shadow->shadow_va;
+		p->jobs[i]->csa_va = shadow->csa_va;
+		p->jobs[i]->gds_va = shadow->gds_va;
+		p->jobs[i]->init_shadow =
+			shadow->flags & AMDGPU_CS_CHUNK_CP_GFX_SHADOW_FLAGS_INIT_SHADOW;
+	}
+
+	return 0;
+}
+
+static int amdgpu_cs_pass2(struct amdgpu_cs_parser *p)
+{
+	unsigned int ce_preempt = 0, de_preempt = 0;
+	int i, r;
+
+	for (i = 0; i < p->nchunks; ++i) {
+		struct amdgpu_cs_chunk *chunk;
+
+		chunk = &p->chunks[i];
+
+		switch (chunk->chunk_id) {
+		case AMDGPU_CHUNK_ID_IB:
+			r = amdgpu_cs_p2_ib(p, chunk, &ce_preempt, &de_preempt);
+			if (r)
+				return r;
+			break;
+		case AMDGPU_CHUNK_ID_DEPENDENCIES:
+		case AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES:
+			r = amdgpu_cs_p2_dependencies(p, chunk);
+			if (r)
+				return r;
+			break;
+		case AMDGPU_CHUNK_ID_SYNCOBJ_IN:
+			r = amdgpu_cs_p2_syncobj_in(p, chunk);
+			if (r)
+				return r;
+			break;
+		case AMDGPU_CHUNK_ID_SYNCOBJ_OUT:
+			r = amdgpu_cs_p2_syncobj_out(p, chunk);
+			if (r)
+				return r;
+			break;
+		case AMDGPU_CHUNK_ID_SYNCOBJ_TIMELINE_WAIT:
+			r = amdgpu_cs_p2_syncobj_timeline_wait(p, chunk);
+			if (r)
+				return r;
+			break;
+		case AMDGPU_CHUNK_ID_SYNCOBJ_TIMELINE_SIGNAL:
+			r = amdgpu_cs_p2_syncobj_timeline_signal(p, chunk);
+			if (r)
+				return r;
+			break;
+		case AMDGPU_CHUNK_ID_CP_GFX_SHADOW:
+			r = amdgpu_cs_p2_shadow(p, chunk);
+			if (r)
+				return r;
+			break;
+		}
+	}
+
+	return 0;
+}
+
 /* Convert microseconds to bytes. */
 static u64 us_to_bytes(struct amdgpu_device *adev, s64 us)
 {
@@ -229,7 +693,6 @@ static void amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev,
 {
 	s64 time_us, increment_us;
 	u64 free_vram, total_vram, used_vram;
-
 	/* Allow a maximum of 200 accumulated ms. This is basically per-IB
 	 * throttling.
 	 *
@@ -245,8 +708,8 @@ static void amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev,
 		return;
 	}
 
-	total_vram = adev->mc.real_vram_size - adev->vram_pin_size;
-	used_vram = amdgpu_vram_mgr_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
+	total_vram = adev->gmc.real_vram_size - atomic64_read(&adev->vram_pin_size);
+	used_vram = ttm_resource_manager_usage(&adev->mman.vram_mgr.manager);
 	free_vram = used_vram >= total_vram ? 0 : total_vram - used_vram;
 
 	spin_lock(&adev->mm_stats.lock);
@@ -256,7 +719,7 @@ static void amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev,
 	increment_us = time_us - adev->mm_stats.last_update_us;
 	adev->mm_stats.last_update_us = time_us;
 	adev->mm_stats.accum_us = min(adev->mm_stats.accum_us + increment_us,
-                                      us_upper_bound);
+				      us_upper_bound);
 
 	/* This prevents the short period of low performance when the VRAM
 	 * usage is low and the driver is in debt or doesn't have enough
@@ -273,7 +736,7 @@ static void amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev,
 	if (free_vram >= 128 * 1024 * 1024 || free_vram >= total_vram / 8) {
 		s64 min_us;
 
-		/* Be more aggresive on dGPUs. Try to fill a portion of free
+		/* Be more aggressive on dGPUs. Try to fill a portion of free
 		 * VRAM now.
 		 */
 		if (!(adev->flags & AMD_IS_APU))
@@ -290,13 +753,14 @@ static void amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev,
 	*max_bytes = us_to_bytes(adev, adev->mm_stats.accum_us);
 
 	/* Do the same for visible VRAM if half of it is free */
-	if (adev->mc.visible_vram_size < adev->mc.real_vram_size) {
-		u64 total_vis_vram = adev->mc.visible_vram_size;
+	if (!amdgpu_gmc_vram_full_visible(&adev->gmc)) {
+		u64 total_vis_vram = adev->gmc.visible_vram_size;
 		u64 used_vis_vram =
-			amdgpu_vram_mgr_vis_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
+		  amdgpu_vram_mgr_vis_usage(&adev->mman.vram_mgr);
 
 		if (used_vis_vram < total_vis_vram) {
 			u64 free_vis_vram = total_vis_vram - used_vis_vram;
+
 			adev->mm_stats.accum_us_vis = min(adev->mm_stats.accum_us_vis +
 							  increment_us, us_upper_bound);
 
@@ -327,22 +791,28 @@ void amdgpu_cs_report_moved_bytes(struct amdgpu_device *adev, u64 num_bytes,
 	spin_unlock(&adev->mm_stats.lock);
 }
 
-static int amdgpu_cs_bo_validate(struct amdgpu_cs_parser *p,
-				 struct amdgpu_bo *bo)
+static int amdgpu_cs_bo_validate(void *param, struct amdgpu_bo *bo)
 {
 	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
-	u64 initial_bytes_moved, bytes_moved;
+	struct amdgpu_cs_parser *p = param;
+	struct ttm_operation_ctx ctx = {
+		.interruptible = true,
+		.no_wait_gpu = false,
+		.resv = bo->tbo.base.resv
+	};
 	uint32_t domain;
 	int r;
 
-	if (bo->pin_count)
+	if (bo->tbo.pin_count)
 		return 0;
 
 	/* Don't move this buffer if we have depleted our allowance
 	 * to move it. Don't move anything if the threshold is zero.
 	 */
-	if (p->bytes_moved < p->bytes_moved_threshold) {
-		if (adev->mc.visible_vram_size < adev->mc.real_vram_size &&
+	if (p->bytes_moved < p->bytes_moved_threshold &&
+	    (!bo->tbo.base.dma_buf ||
+	    list_empty(&bo->tbo.base.dma_buf->attachments))) {
+		if (!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
 		    (bo->flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED)) {
 			/* And don't move a CPU_ACCESS_REQUIRED BO to limited
 			 * visible VRAM if we've depleted our allowance to do
@@ -360,16 +830,13 @@ static int amdgpu_cs_bo_validate(struct amdgpu_cs_parser *p,
 	}
 
 retry:
-	amdgpu_ttm_placement_from_domain(bo, domain);
-	initial_bytes_moved = atomic64_read(&adev->num_bytes_moved);
-	r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false);
-	bytes_moved = atomic64_read(&adev->num_bytes_moved) -
-		      initial_bytes_moved;
-	p->bytes_moved += bytes_moved;
-	if (adev->mc.visible_vram_size < adev->mc.real_vram_size &&
-	    bo->tbo.mem.mem_type == TTM_PL_VRAM &&
-	    bo->tbo.mem.start < adev->mc.visible_vram_size >> PAGE_SHIFT)
-		p->bytes_moved_vis += bytes_moved;
+	amdgpu_bo_placement_from_domain(bo, domain);
+	r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+
+	p->bytes_moved += ctx.bytes_moved;
+	if (!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
+	    amdgpu_res_cpu_visible(adev, bo->tbo.resource))
+		p->bytes_moved_vis += ctx.bytes_moved;
 
 	if (unlikely(r == -ENOMEM) && domain != bo->allowed_domains) {
 		domain = bo->allowed_domains;
@@ -379,398 +846,269 @@ retry:
 	return r;
 }
 
-/* Last resort, try to evict something from the current working set */
-static bool amdgpu_cs_try_evict(struct amdgpu_cs_parser *p,
-				struct amdgpu_bo *validated)
-{
-	uint32_t domain = validated->allowed_domains;
-	int r;
-
-	if (!p->evictable)
-		return false;
-
-	for (;&p->evictable->tv.head != &p->validated;
-	     p->evictable = list_prev_entry(p->evictable, tv.head)) {
-
-		struct amdgpu_bo_list_entry *candidate = p->evictable;
-		struct amdgpu_bo *bo = candidate->robj;
-		struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
-		u64 initial_bytes_moved, bytes_moved;
-		bool update_bytes_moved_vis;
-		uint32_t other;
-
-		/* If we reached our current BO we can forget it */
-		if (candidate->robj == validated)
-			break;
-
-		other = amdgpu_mem_type_to_domain(bo->tbo.mem.mem_type);
-
-		/* Check if this BO is in one of the domains we need space for */
-		if (!(other & domain))
-			continue;
-
-		/* Check if we can move this BO somewhere else */
-		other = bo->allowed_domains & ~domain;
-		if (!other)
-			continue;
-
-		/* Good we can try to move this BO somewhere else */
-		amdgpu_ttm_placement_from_domain(bo, other);
-		update_bytes_moved_vis =
-			adev->mc.visible_vram_size < adev->mc.real_vram_size &&
-			bo->tbo.mem.mem_type == TTM_PL_VRAM &&
-			bo->tbo.mem.start < adev->mc.visible_vram_size >> PAGE_SHIFT;
-		initial_bytes_moved = atomic64_read(&adev->num_bytes_moved);
-		r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false);
-		bytes_moved = atomic64_read(&adev->num_bytes_moved) -
-			initial_bytes_moved;
-		p->bytes_moved += bytes_moved;
-		if (update_bytes_moved_vis)
-			p->bytes_moved_vis += bytes_moved;
-
-		if (unlikely(r))
-			break;
-
-		p->evictable = list_prev_entry(p->evictable, tv.head);
-		list_move(&candidate->tv.head, &p->validated);
-
-		return true;
-	}
-
-	return false;
-}
-
-static int amdgpu_cs_validate(void *param, struct amdgpu_bo *bo)
-{
-	struct amdgpu_cs_parser *p = param;
-	int r;
-
-	do {
-		r = amdgpu_cs_bo_validate(p, bo);
-	} while (r == -ENOMEM && amdgpu_cs_try_evict(p, bo));
-	if (r)
-		return r;
-
-	if (bo->shadow)
-		r = amdgpu_cs_bo_validate(p, bo->shadow);
-
-	return r;
-}
-
-static int amdgpu_cs_list_validate(struct amdgpu_cs_parser *p,
-			    struct list_head *validated)
-{
-	struct amdgpu_bo_list_entry *lobj;
-	int r;
-
-	list_for_each_entry(lobj, validated, tv.head) {
-		struct amdgpu_bo *bo = lobj->robj;
-		bool binding_userptr = false;
-		struct mm_struct *usermm;
-
-		usermm = amdgpu_ttm_tt_get_usermm(bo->tbo.ttm);
-		if (usermm && usermm != current->mm)
-			return -EPERM;
-
-		/* Check if we have user pages and nobody bound the BO already */
-		if (lobj->user_pages && bo->tbo.ttm->state != tt_bound) {
-			size_t size = sizeof(struct page *);
-
-			size *= bo->tbo.ttm->num_pages;
-			memcpy(bo->tbo.ttm->pages, lobj->user_pages, size);
-			binding_userptr = true;
-		}
-
-		if (p->evictable == lobj)
-			p->evictable = NULL;
-
-		r = amdgpu_cs_validate(p, bo);
-		if (r)
-			return r;
-
-		if (binding_userptr) {
-			kvfree(lobj->user_pages);
-			lobj->user_pages = NULL;
-		}
-	}
-	return 0;
-}
-
 static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 				union drm_amdgpu_cs *cs)
 {
 	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
+	struct ttm_operation_ctx ctx = { true, false };
+	struct amdgpu_vm *vm = &fpriv->vm;
 	struct amdgpu_bo_list_entry *e;
-	struct list_head duplicates;
-	bool need_mmap_lock = false;
-	unsigned i, tries = 10;
+	struct drm_gem_object *obj;
+	unsigned long index;
+	unsigned int i;
 	int r;
 
-	INIT_LIST_HEAD(&p->validated);
+	/* p->bo_list could already be assigned if AMDGPU_CHUNK_ID_BO_HANDLES is present */
+	if (cs->in.bo_list_handle) {
+		if (p->bo_list)
+			return -EINVAL;
 
-	p->bo_list = amdgpu_bo_list_get(fpriv, cs->in.bo_list_handle);
-	if (p->bo_list) {
-		need_mmap_lock = p->bo_list->first_userptr !=
-			p->bo_list->num_entries;
-		amdgpu_bo_list_get_list(p->bo_list, &p->validated);
+		r = amdgpu_bo_list_get(fpriv, cs->in.bo_list_handle,
+				       &p->bo_list);
+		if (r)
+			return r;
+	} else if (!p->bo_list) {
+		/* Create a empty bo_list when no handle is provided */
+		r = amdgpu_bo_list_create(p->adev, p->filp, NULL, 0,
+					  &p->bo_list);
+		if (r)
+			return r;
 	}
 
-	INIT_LIST_HEAD(&duplicates);
-	amdgpu_vm_get_pd_bo(&fpriv->vm, &p->validated, &p->vm_pd);
+	mutex_lock(&p->bo_list->bo_list_mutex);
 
-	if (p->uf_entry.robj)
-		list_add(&p->uf_entry.tv.head, &p->validated);
-
-	if (need_mmap_lock)
-		down_read(&current->mm->mmap_sem);
+	/* Get userptr backing pages. If pages are updated after registered
+	 * in amdgpu_gem_userptr_ioctl(), amdgpu_cs_list_validate() will do
+	 * amdgpu_ttm_backend_bind() to flush and invalidate new pages
+	 */
+	amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
+		bool userpage_invalidated = false;
+		struct amdgpu_bo *bo = e->bo;
 
-	while (1) {
-		struct list_head need_pages;
-		unsigned i;
+		r = amdgpu_ttm_tt_get_user_pages(bo, &e->range);
+		if (r)
+			goto out_free_user_pages;
 
-		r = ttm_eu_reserve_buffers(&p->ticket, &p->validated, true,
-					   &duplicates);
-		if (unlikely(r != 0)) {
-			if (r != -ERESTARTSYS)
-				DRM_ERROR("ttm_eu_reserve_buffers failed.\n");
-			goto error_free_pages;
+		for (i = 0; i < bo->tbo.ttm->num_pages; i++) {
+			if (bo->tbo.ttm->pages[i] != hmm_pfn_to_page(e->range->hmm_pfns[i])) {
+				userpage_invalidated = true;
+				break;
+			}
 		}
+		e->user_invalidated = userpage_invalidated;
+	}
 
-		/* Without a BO list we don't have userptr BOs */
-		if (!p->bo_list)
-			break;
-
-		INIT_LIST_HEAD(&need_pages);
-		for (i = p->bo_list->first_userptr;
-		     i < p->bo_list->num_entries; ++i) {
-
-			e = &p->bo_list->array[i];
-
-			if (amdgpu_ttm_tt_userptr_invalidated(e->robj->tbo.ttm,
-				 &e->user_invalidated) && e->user_pages) {
-
-				/* We acquired a page array, but somebody
-				 * invalidated it. Free it and try again
-				 */
-				release_pages(e->user_pages,
-					      e->robj->tbo.ttm->num_pages,
-					      false);
-				kvfree(e->user_pages);
-				e->user_pages = NULL;
-			}
+	drm_exec_until_all_locked(&p->exec) {
+		r = amdgpu_vm_lock_pd(&fpriv->vm, &p->exec, 1 + p->gang_size);
+		drm_exec_retry_on_contention(&p->exec);
+		if (unlikely(r))
+			goto out_free_user_pages;
 
-			if (e->robj->tbo.ttm->state != tt_bound &&
-			    !e->user_pages) {
-				list_del(&e->tv.head);
-				list_add(&e->tv.head, &need_pages);
+		amdgpu_bo_list_for_each_entry(e, p->bo_list) {
+			/* One fence for TTM and one for each CS job */
+			r = drm_exec_prepare_obj(&p->exec, &e->bo->tbo.base,
+						 1 + p->gang_size);
+			drm_exec_retry_on_contention(&p->exec);
+			if (unlikely(r))
+				goto out_free_user_pages;
 
-				amdgpu_bo_unreserve(e->robj);
-			}
+			e->bo_va = amdgpu_vm_bo_find(vm, e->bo);
 		}
 
-		if (list_empty(&need_pages))
-			break;
+		if (p->uf_bo) {
+			r = drm_exec_prepare_obj(&p->exec, &p->uf_bo->tbo.base,
+						 1 + p->gang_size);
+			drm_exec_retry_on_contention(&p->exec);
+			if (unlikely(r))
+				goto out_free_user_pages;
+		}
+	}
 
-		/* Unreserve everything again. */
-		ttm_eu_backoff_reservation(&p->ticket, &p->validated);
+	amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
+		struct mm_struct *usermm;
 
-		/* We tried too many times, just abort */
-		if (!--tries) {
-			r = -EDEADLK;
-			DRM_ERROR("deadlock in %s\n", __func__);
-			goto error_free_pages;
+		usermm = amdgpu_ttm_tt_get_usermm(e->bo->tbo.ttm);
+		if (usermm && usermm != current->mm) {
+			r = -EPERM;
+			goto out_free_user_pages;
 		}
 
-		/* Fill the page arrays for all userptrs. */
-		list_for_each_entry(e, &need_pages, tv.head) {
-			struct ttm_tt *ttm = e->robj->tbo.ttm;
-
-			e->user_pages = kvmalloc_array(ttm->num_pages,
-							 sizeof(struct page*),
-							 GFP_KERNEL | __GFP_ZERO);
-			if (!e->user_pages) {
-				r = -ENOMEM;
-				DRM_ERROR("calloc failure in %s\n", __func__);
-				goto error_free_pages;
-			}
+		if (amdgpu_ttm_tt_is_userptr(e->bo->tbo.ttm) &&
+		    e->user_invalidated) {
+			amdgpu_bo_placement_from_domain(e->bo,
+							AMDGPU_GEM_DOMAIN_CPU);
+			r = ttm_bo_validate(&e->bo->tbo, &e->bo->placement,
+					    &ctx);
+			if (r)
+				goto out_free_user_pages;
 
-			r = amdgpu_ttm_tt_get_user_pages(ttm, e->user_pages);
-			if (r) {
-				DRM_ERROR("amdgpu_ttm_tt_get_user_pages failed.\n");
-				kvfree(e->user_pages);
-				e->user_pages = NULL;
-				goto error_free_pages;
-			}
+			amdgpu_ttm_tt_set_user_pages(e->bo->tbo.ttm,
+						     e->range);
 		}
-
-		/* And try again. */
-		list_splice(&need_pages, &p->validated);
 	}
 
 	amdgpu_cs_get_threshold_for_moves(p->adev, &p->bytes_moved_threshold,
 					  &p->bytes_moved_vis_threshold);
 	p->bytes_moved = 0;
 	p->bytes_moved_vis = 0;
-	p->evictable = list_last_entry(&p->validated,
-				       struct amdgpu_bo_list_entry,
-				       tv.head);
 
-	r = amdgpu_vm_validate_pt_bos(p->adev, &fpriv->vm,
-				      amdgpu_cs_validate, p);
+	r = amdgpu_vm_validate(p->adev, &fpriv->vm, NULL,
+			       amdgpu_cs_bo_validate, p);
 	if (r) {
-		DRM_ERROR("amdgpu_vm_validate_pt_bos() failed.\n");
-		goto error_validate;
+		drm_err(adev_to_drm(p->adev), "amdgpu_vm_validate() failed.\n");
+		goto out_free_user_pages;
 	}
 
-	r = amdgpu_cs_list_validate(p, &duplicates);
-	if (r) {
-		DRM_ERROR("amdgpu_cs_list_validate(duplicates) failed.\n");
-		goto error_validate;
+	drm_exec_for_each_locked_object(&p->exec, index, obj) {
+		r = amdgpu_cs_bo_validate(p, gem_to_amdgpu_bo(obj));
+		if (unlikely(r))
+			goto out_free_user_pages;
 	}
 
-	r = amdgpu_cs_list_validate(p, &p->validated);
-	if (r) {
-		DRM_ERROR("amdgpu_cs_list_validate(validated) failed.\n");
-		goto error_validate;
+	if (p->uf_bo) {
+		r = amdgpu_ttm_alloc_gart(&p->uf_bo->tbo);
+		if (unlikely(r))
+			goto out_free_user_pages;
+
+		p->gang_leader->uf_addr += amdgpu_bo_gpu_offset(p->uf_bo);
 	}
 
 	amdgpu_cs_report_moved_bytes(p->adev, p->bytes_moved,
 				     p->bytes_moved_vis);
-	fpriv->vm.last_eviction_counter =
-		atomic64_read(&p->adev->num_evictions);
 
-	if (p->bo_list) {
-		struct amdgpu_bo *gds = p->bo_list->gds_obj;
-		struct amdgpu_bo *gws = p->bo_list->gws_obj;
-		struct amdgpu_bo *oa = p->bo_list->oa_obj;
-		struct amdgpu_vm *vm = &fpriv->vm;
-		unsigned i;
-
-		for (i = 0; i < p->bo_list->num_entries; i++) {
-			struct amdgpu_bo *bo = p->bo_list->array[i].robj;
+	for (i = 0; i < p->gang_size; ++i)
+		amdgpu_job_set_resources(p->jobs[i], p->bo_list->gds_obj,
+					 p->bo_list->gws_obj,
+					 p->bo_list->oa_obj);
+	return 0;
 
-			p->bo_list->array[i].bo_va = amdgpu_vm_bo_find(vm, bo);
-		}
+out_free_user_pages:
+	amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
+		struct amdgpu_bo *bo = e->bo;
 
-		if (gds) {
-			p->job->gds_base = amdgpu_bo_gpu_offset(gds);
-			p->job->gds_size = amdgpu_bo_size(gds);
-		}
-		if (gws) {
-			p->job->gws_base = amdgpu_bo_gpu_offset(gws);
-			p->job->gws_size = amdgpu_bo_size(gws);
-		}
-		if (oa) {
-			p->job->oa_base = amdgpu_bo_gpu_offset(oa);
-			p->job->oa_size = amdgpu_bo_size(oa);
-		}
+		amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm, e->range);
+		e->range = NULL;
 	}
+	mutex_unlock(&p->bo_list->bo_list_mutex);
+	return r;
+}
+
+static void trace_amdgpu_cs_ibs(struct amdgpu_cs_parser *p)
+{
+	int i, j;
+
+	if (!trace_amdgpu_cs_enabled())
+		return;
 
-	if (!r && p->uf_entry.robj) {
-		struct amdgpu_bo *uf = p->uf_entry.robj;
+	for (i = 0; i < p->gang_size; ++i) {
+		struct amdgpu_job *job = p->jobs[i];
 
-		r = amdgpu_ttm_bind(&uf->tbo, &uf->tbo.mem);
-		p->job->uf_addr += amdgpu_bo_gpu_offset(uf);
+		for (j = 0; j < job->num_ibs; ++j)
+			trace_amdgpu_cs(p, job, &job->ibs[j]);
 	}
+}
 
-error_validate:
-	if (r)
-		ttm_eu_backoff_reservation(&p->ticket, &p->validated);
+static int amdgpu_cs_patch_ibs(struct amdgpu_cs_parser *p,
+			       struct amdgpu_job *job)
+{
+	struct amdgpu_ring *ring = amdgpu_job_ring(job);
+	unsigned int i;
+	int r;
+
+	/* Only for UVD/VCE VM emulation */
+	if (!ring->funcs->parse_cs && !ring->funcs->patch_cs_in_place)
+		return 0;
 
-error_free_pages:
+	for (i = 0; i < job->num_ibs; ++i) {
+		struct amdgpu_ib *ib = &job->ibs[i];
+		struct amdgpu_bo_va_mapping *m;
+		struct amdgpu_bo *aobj;
+		uint64_t va_start;
+		uint8_t *kptr;
 
-	if (need_mmap_lock)
-		up_read(&current->mm->mmap_sem);
+		va_start = ib->gpu_addr & AMDGPU_GMC_HOLE_MASK;
+		r = amdgpu_cs_find_mapping(p, va_start, &aobj, &m);
+		if (r) {
+			drm_err(adev_to_drm(p->adev), "IB va_start is invalid\n");
+			return r;
+		}
 
-	if (p->bo_list) {
-		for (i = p->bo_list->first_userptr;
-		     i < p->bo_list->num_entries; ++i) {
-			e = &p->bo_list->array[i];
+		if ((va_start + ib->length_dw * 4) >
+		    (m->last + 1) * AMDGPU_GPU_PAGE_SIZE) {
+			drm_err(adev_to_drm(p->adev), "IB va_start+ib_bytes is invalid\n");
+			return -EINVAL;
+		}
 
-			if (!e->user_pages)
-				continue;
+		/* the IB should be reserved at this point */
+		r = amdgpu_bo_kmap(aobj, (void **)&kptr);
+		if (r)
+			return r;
+
+		kptr += va_start - (m->start * AMDGPU_GPU_PAGE_SIZE);
+
+		if (ring->funcs->parse_cs) {
+			memcpy(ib->ptr, kptr, ib->length_dw * 4);
+			amdgpu_bo_kunmap(aobj);
 
-			release_pages(e->user_pages,
-				      e->robj->tbo.ttm->num_pages,
-				      false);
-			kvfree(e->user_pages);
+			r = amdgpu_ring_parse_cs(ring, p, job, ib);
+			if (r)
+				return r;
+
+			if (ib->sa_bo)
+				ib->gpu_addr =  amdgpu_sa_bo_gpu_addr(ib->sa_bo);
+		} else {
+			ib->ptr = (uint32_t *)kptr;
+			r = amdgpu_ring_patch_cs_in_place(ring, p, job, ib);
+			amdgpu_bo_kunmap(aobj);
+			if (r)
+				return r;
 		}
 	}
 
-	return r;
+	return 0;
 }
 
-static int amdgpu_cs_sync_rings(struct amdgpu_cs_parser *p)
+static int amdgpu_cs_patch_jobs(struct amdgpu_cs_parser *p)
 {
-	struct amdgpu_bo_list_entry *e;
+	unsigned int i;
 	int r;
 
-	list_for_each_entry(e, &p->validated, tv.head) {
-		struct reservation_object *resv = e->robj->tbo.resv;
-		r = amdgpu_sync_resv(p->adev, &p->job->sync, resv, p->filp);
-
+	for (i = 0; i < p->gang_size; ++i) {
+		r = amdgpu_cs_patch_ibs(p, p->jobs[i]);
 		if (r)
 			return r;
 	}
 	return 0;
 }
 
-/**
- * cs_parser_fini() - clean parser states
- * @parser:	parser structure holding parsing context.
- * @error:	error number
- *
- * If error is set than unvalidate buffer, otherwise just free memory
- * used by parsing context.
- **/
-static void amdgpu_cs_parser_fini(struct amdgpu_cs_parser *parser, int error,
-				  bool backoff)
-{
-	unsigned i;
-
-	if (!error)
-		ttm_eu_fence_buffer_objects(&parser->ticket,
-					    &parser->validated,
-					    parser->fence);
-	else if (backoff)
-		ttm_eu_backoff_reservation(&parser->ticket,
-					   &parser->validated);
-
-	for (i = 0; i < parser->num_post_dep_syncobjs; i++)
-		drm_syncobj_put(parser->post_dep_syncobjs[i]);
-	kfree(parser->post_dep_syncobjs);
-
-	dma_fence_put(parser->fence);
-
-	if (parser->ctx)
-		amdgpu_ctx_put(parser->ctx);
-	if (parser->bo_list)
-		amdgpu_bo_list_put(parser->bo_list);
-
-	for (i = 0; i < parser->nchunks; i++)
-		kvfree(parser->chunks[i].kdata);
-	kfree(parser->chunks);
-	if (parser->job)
-		amdgpu_job_free(parser->job);
-	amdgpu_bo_unref(&parser->uf_entry.robj);
-}
-
-static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p)
+static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 {
-	struct amdgpu_device *adev = p->adev;
 	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
+	struct amdgpu_job *job = p->gang_leader;
+	struct amdgpu_device *adev = p->adev;
 	struct amdgpu_vm *vm = &fpriv->vm;
+	struct amdgpu_bo_list_entry *e;
 	struct amdgpu_bo_va *bo_va;
-	struct amdgpu_bo *bo;
-	int i, r;
+	unsigned int i;
+	int r;
 
-	r = amdgpu_vm_update_directories(adev, vm);
-	if (r)
-		return r;
+	/*
+	 * We can't use gang submit on with reserved VMIDs when the VM changes
+	 * can't be invalidated by more than one engine at the same time.
+	 */
+	if (p->gang_size > 1 && !adev->vm_manager.concurrent_flush) {
+		for (i = 0; i < p->gang_size; ++i) {
+			struct drm_sched_entity *entity = p->entities[i];
+			struct drm_gpu_scheduler *sched = entity->rq->sched;
+			struct amdgpu_ring *ring = to_amdgpu_ring(sched);
 
-	r = amdgpu_sync_fence(adev, &p->job->sync, vm->last_dir_update);
-	if (r)
-		return r;
+			if (amdgpu_vmid_uses_reserved(vm, ring->vm_hub))
+				return -EINVAL;
+		}
+	}
+
+	if (!amdgpu_vm_ready(vm))
+		return -EINVAL;
 
 	r = amdgpu_vm_clear_freed(adev, vm, NULL);
 	if (r)
@@ -780,439 +1118,354 @@ static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p)
 	if (r)
 		return r;
 
-	r = amdgpu_sync_fence(adev, &p->job->sync,
-			      fpriv->prt_va->last_pt_update);
+	r = amdgpu_sync_fence(&p->sync, fpriv->prt_va->last_pt_update,
+			      GFP_KERNEL);
 	if (r)
 		return r;
 
-	if (amdgpu_sriov_vf(adev)) {
-		struct dma_fence *f;
-
+	if (fpriv->csa_va) {
 		bo_va = fpriv->csa_va;
 		BUG_ON(!bo_va);
 		r = amdgpu_vm_bo_update(adev, bo_va, false);
 		if (r)
 			return r;
 
-		f = bo_va->last_pt_update;
-		r = amdgpu_sync_fence(adev, &p->job->sync, f);
+		r = amdgpu_sync_fence(&p->sync, bo_va->last_pt_update,
+				      GFP_KERNEL);
 		if (r)
 			return r;
 	}
 
-	if (p->bo_list) {
-		for (i = 0; i < p->bo_list->num_entries; i++) {
-			struct dma_fence *f;
+	/* FIXME: In theory this loop shouldn't be needed any more when
+	 * amdgpu_vm_handle_moved handles all moved BOs that are reserved
+	 * with p->ticket. But removing it caused test regressions, so I'm
+	 * leaving it here for now.
+	 */
+	amdgpu_bo_list_for_each_entry(e, p->bo_list) {
+		bo_va = e->bo_va;
+		if (bo_va == NULL)
+			continue;
 
-			/* ignore duplicates */
-			bo = p->bo_list->array[i].robj;
-			if (!bo)
-				continue;
+		r = amdgpu_vm_bo_update(adev, bo_va, false);
+		if (r)
+			return r;
 
-			bo_va = p->bo_list->array[i].bo_va;
-			if (bo_va == NULL)
-				continue;
+		r = amdgpu_sync_fence(&p->sync, bo_va->last_pt_update,
+				      GFP_KERNEL);
+		if (r)
+			return r;
+	}
 
-			r = amdgpu_vm_bo_update(adev, bo_va, false);
-			if (r)
-				return r;
+	r = amdgpu_vm_handle_moved(adev, vm, &p->exec.ticket);
+	if (r)
+		return r;
 
-			f = bo_va->last_pt_update;
-			r = amdgpu_sync_fence(adev, &p->job->sync, f);
-			if (r)
-				return r;
-		}
+	r = amdgpu_vm_update_pdes(adev, vm, false);
+	if (r)
+		return r;
 
-	}
+	r = amdgpu_sync_fence(&p->sync, vm->last_update, GFP_KERNEL);
+	if (r)
+		return r;
 
-	r = amdgpu_vm_clear_moved(adev, vm, &p->job->sync);
+	for (i = 0; i < p->gang_size; ++i) {
+		job = p->jobs[i];
 
-	if (amdgpu_vm_debug && p->bo_list) {
+		if (!job->vm)
+			continue;
+
+		job->vm_pd_addr = amdgpu_gmc_pd_addr(vm->root.bo);
+	}
+
+	if (adev->debug_vm) {
 		/* Invalidate all BOs to test for userspace bugs */
-		for (i = 0; i < p->bo_list->num_entries; i++) {
+		amdgpu_bo_list_for_each_entry(e, p->bo_list) {
+			struct amdgpu_bo *bo = e->bo;
+
 			/* ignore duplicates */
-			bo = p->bo_list->array[i].robj;
 			if (!bo)
 				continue;
 
-			amdgpu_vm_bo_invalidate(adev, bo);
+			amdgpu_vm_bo_invalidate(bo, false);
 		}
 	}
 
-	return r;
+	return 0;
 }
 
-static int amdgpu_cs_ib_vm_chunk(struct amdgpu_device *adev,
-				 struct amdgpu_cs_parser *p)
+static int amdgpu_cs_sync_rings(struct amdgpu_cs_parser *p)
 {
 	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
-	struct amdgpu_vm *vm = &fpriv->vm;
-	struct amdgpu_ring *ring = p->job->ring;
-	int i, r;
+	struct drm_gpu_scheduler *sched;
+	struct drm_gem_object *obj;
+	struct dma_fence *fence;
+	unsigned long index;
+	unsigned int i;
+	int r;
 
-	/* Only for UVD/VCE VM emulation */
-	if (ring->funcs->parse_cs) {
-		for (i = 0; i < p->job->num_ibs; i++) {
-			r = amdgpu_ring_parse_cs(ring, p, i);
-			if (r)
-				return r;
-		}
+	r = amdgpu_ctx_wait_prev_fence(p->ctx, p->entities[p->gang_leader_idx]);
+	if (r) {
+		if (r != -ERESTARTSYS)
+			drm_err(adev_to_drm(p->adev), "amdgpu_ctx_wait_prev_fence failed.\n");
+		return r;
 	}
 
-	if (p->job->vm) {
-		p->job->vm_pd_addr = amdgpu_bo_gpu_offset(vm->root.bo);
+	drm_exec_for_each_locked_object(&p->exec, index, obj) {
+		struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+
+		struct dma_resv *resv = bo->tbo.base.resv;
+		enum amdgpu_sync_mode sync_mode;
 
-		r = amdgpu_bo_vm_update_pte(p);
+		sync_mode = amdgpu_bo_explicit_sync(bo) ?
+			AMDGPU_SYNC_EXPLICIT : AMDGPU_SYNC_NE_OWNER;
+		r = amdgpu_sync_resv(p->adev, &p->sync, resv, sync_mode,
+				     &fpriv->vm);
 		if (r)
 			return r;
 	}
 
-	return amdgpu_cs_sync_rings(p);
-}
-
-static int amdgpu_cs_ib_fill(struct amdgpu_device *adev,
-			     struct amdgpu_cs_parser *parser)
-{
-	struct amdgpu_fpriv *fpriv = parser->filp->driver_priv;
-	struct amdgpu_vm *vm = &fpriv->vm;
-	int i, j;
-	int r, ce_preempt = 0, de_preempt = 0;
-
-	for (i = 0, j = 0; i < parser->nchunks && j < parser->job->num_ibs; i++) {
-		struct amdgpu_cs_chunk *chunk;
-		struct amdgpu_ib *ib;
-		struct drm_amdgpu_cs_chunk_ib *chunk_ib;
-		struct amdgpu_ring *ring;
+	for (i = 0; i < p->gang_size; ++i) {
+		r = amdgpu_sync_push_to_job(&p->sync, p->jobs[i]);
+		if (r)
+			return r;
+	}
 
-		chunk = &parser->chunks[i];
-		ib = &parser->job->ibs[j];
-		chunk_ib = (struct drm_amdgpu_cs_chunk_ib *)chunk->kdata;
+	sched = p->gang_leader->base.entity->rq->sched;
+	while ((fence = amdgpu_sync_get_fence(&p->sync))) {
+		struct drm_sched_fence *s_fence = to_drm_sched_fence(fence);
 
-		if (chunk->chunk_id != AMDGPU_CHUNK_ID_IB)
+		/*
+		 * When we have an dependency it might be necessary to insert a
+		 * pipeline sync to make sure that all caches etc are flushed and the
+		 * next job actually sees the results from the previous one
+		 * before we start executing on the same scheduler ring.
+		 */
+		if (!s_fence || s_fence->sched != sched) {
+			dma_fence_put(fence);
 			continue;
-
-		if (chunk_ib->ip_type == AMDGPU_HW_IP_GFX && amdgpu_sriov_vf(adev)) {
-			if (chunk_ib->flags & AMDGPU_IB_FLAG_PREEMPT) {
-				if (chunk_ib->flags & AMDGPU_IB_FLAG_CE)
-					ce_preempt++;
-				else
-					de_preempt++;
-			}
-
-			/* each GFX command submit allows 0 or 1 IB preemptible for CE & DE */
-			if (ce_preempt > 1 || de_preempt > 1)
-				return -EINVAL;
 		}
 
-		r = amdgpu_queue_mgr_map(adev, &parser->ctx->queue_mgr, chunk_ib->ip_type,
-					 chunk_ib->ip_instance, chunk_ib->ring, &ring);
+		r = amdgpu_sync_fence(&p->gang_leader->explicit_sync, fence,
+				      GFP_KERNEL);
+		dma_fence_put(fence);
 		if (r)
 			return r;
+	}
+	return 0;
+}
 
-		if (chunk_ib->flags & AMDGPU_IB_FLAG_PREAMBLE) {
-			parser->job->preamble_status |= AMDGPU_PREAMBLE_IB_PRESENT;
-			if (!parser->ctx->preamble_presented) {
-				parser->job->preamble_status |= AMDGPU_PREAMBLE_IB_PRESENT_FIRST;
-				parser->ctx->preamble_presented = true;
-			}
-		}
-
-		if (parser->job->ring && parser->job->ring != ring)
-			return -EINVAL;
-
-		parser->job->ring = ring;
-
-		if (ring->funcs->parse_cs) {
-			struct amdgpu_bo_va_mapping *m;
-			struct amdgpu_bo *aobj = NULL;
-			uint64_t offset;
-			uint8_t *kptr;
-
-			m = amdgpu_cs_find_mapping(parser, chunk_ib->va_start,
-						   &aobj);
-			if (!aobj) {
-				DRM_ERROR("IB va_start is invalid\n");
-				return -EINVAL;
-			}
-
-			if ((chunk_ib->va_start + chunk_ib->ib_bytes) >
-			    (m->last + 1) * AMDGPU_GPU_PAGE_SIZE) {
-				DRM_ERROR("IB va_start+ib_bytes is invalid\n");
-				return -EINVAL;
-			}
-
-			/* the IB should be reserved at this point */
-			r = amdgpu_bo_kmap(aobj, (void **)&kptr);
-			if (r) {
-				return r;
-			}
-
-			offset = m->start * AMDGPU_GPU_PAGE_SIZE;
-			kptr += chunk_ib->va_start - offset;
-
-			r =  amdgpu_ib_get(adev, vm, chunk_ib->ib_bytes, ib);
-			if (r) {
-				DRM_ERROR("Failed to get ib !\n");
-				return r;
-			}
+static void amdgpu_cs_post_dependencies(struct amdgpu_cs_parser *p)
+{
+	int i;
 
-			memcpy(ib->ptr, kptr, chunk_ib->ib_bytes);
-			amdgpu_bo_kunmap(aobj);
+	for (i = 0; i < p->num_post_deps; ++i) {
+		if (p->post_deps[i].chain && p->post_deps[i].point) {
+			drm_syncobj_add_point(p->post_deps[i].syncobj,
+					      p->post_deps[i].chain,
+					      p->fence, p->post_deps[i].point);
+			p->post_deps[i].chain = NULL;
 		} else {
-			r =  amdgpu_ib_get(adev, vm, 0, ib);
-			if (r) {
-				DRM_ERROR("Failed to get ib !\n");
-				return r;
-			}
-
+			drm_syncobj_replace_fence(p->post_deps[i].syncobj,
+						  p->fence);
 		}
-
-		ib->gpu_addr = chunk_ib->va_start;
-		ib->length_dw = chunk_ib->ib_bytes / 4;
-		ib->flags = chunk_ib->flags;
-		j++;
 	}
-
-	/* UVD & VCE fw doesn't support user fences */
-	if (parser->job->uf_addr && (
-	    parser->job->ring->funcs->type == AMDGPU_RING_TYPE_UVD ||
-	    parser->job->ring->funcs->type == AMDGPU_RING_TYPE_VCE))
-		return -EINVAL;
-
-	return 0;
 }
 
-static int amdgpu_cs_process_fence_dep(struct amdgpu_cs_parser *p,
-				       struct amdgpu_cs_chunk *chunk)
+static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
+			    union drm_amdgpu_cs *cs)
 {
 	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
-	unsigned num_deps;
-	int i, r;
-	struct drm_amdgpu_cs_chunk_dep *deps;
+	struct amdgpu_job *leader = p->gang_leader;
+	struct amdgpu_bo_list_entry *e;
+	struct drm_gem_object *gobj;
+	unsigned long index;
+	unsigned int i;
+	uint64_t seq;
+	int r;
 
-	deps = (struct drm_amdgpu_cs_chunk_dep *)chunk->kdata;
-	num_deps = chunk->length_dw * 4 /
-		sizeof(struct drm_amdgpu_cs_chunk_dep);
+	for (i = 0; i < p->gang_size; ++i)
+		drm_sched_job_arm(&p->jobs[i]->base);
 
-	for (i = 0; i < num_deps; ++i) {
-		struct amdgpu_ring *ring;
-		struct amdgpu_ctx *ctx;
+	for (i = 0; i < p->gang_size; ++i) {
 		struct dma_fence *fence;
 
-		ctx = amdgpu_ctx_get(fpriv, deps[i].ctx_id);
-		if (ctx == NULL)
-			return -EINVAL;
+		if (p->jobs[i] == leader)
+			continue;
 
-		r = amdgpu_queue_mgr_map(p->adev, &ctx->queue_mgr,
-					 deps[i].ip_type,
-					 deps[i].ip_instance,
-					 deps[i].ring, &ring);
+		fence = &p->jobs[i]->base.s_fence->scheduled;
+		dma_fence_get(fence);
+		r = drm_sched_job_add_dependency(&leader->base, fence);
 		if (r) {
-			amdgpu_ctx_put(ctx);
+			dma_fence_put(fence);
 			return r;
 		}
+	}
 
-		fence = amdgpu_ctx_get_fence(ctx, ring,
-					     deps[i].handle);
-		if (IS_ERR(fence)) {
-			r = PTR_ERR(fence);
-			amdgpu_ctx_put(ctx);
-			return r;
-		} else if (fence) {
-			r = amdgpu_sync_fence(p->adev, &p->job->sync,
-					      fence);
-			dma_fence_put(fence);
-			amdgpu_ctx_put(ctx);
-			if (r)
-				return r;
-		}
+	if (p->gang_size > 1) {
+		for (i = 0; i < p->gang_size; ++i)
+			amdgpu_job_set_gang_leader(p->jobs[i], leader);
 	}
-	return 0;
-}
 
-static int amdgpu_syncobj_lookup_and_add_to_sync(struct amdgpu_cs_parser *p,
-						 uint32_t handle)
-{
-	int r;
-	struct dma_fence *fence;
-	r = drm_syncobj_find_fence(p->filp, handle, &fence);
-	if (r)
+	/* No memory allocation is allowed while holding the notifier lock.
+	 * The lock is held until amdgpu_cs_submit is finished and fence is
+	 * added to BOs.
+	 */
+	mutex_lock(&p->adev->notifier_lock);
+
+	/* If userptr are invalidated after amdgpu_cs_parser_bos(), return
+	 * -EAGAIN, drmIoctl in libdrm will restart the amdgpu_cs_ioctl.
+	 */
+	r = 0;
+	amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
+		r |= !amdgpu_ttm_tt_get_user_pages_done(e->bo->tbo.ttm,
+							e->range);
+		e->range = NULL;
+	}
+	if (r) {
+		r = -EAGAIN;
+		mutex_unlock(&p->adev->notifier_lock);
 		return r;
+	}
 
-	r = amdgpu_sync_fence(p->adev, &p->job->sync, fence);
-	dma_fence_put(fence);
+	p->fence = dma_fence_get(&leader->base.s_fence->finished);
+	drm_exec_for_each_locked_object(&p->exec, index, gobj) {
 
-	return r;
-}
+		ttm_bo_move_to_lru_tail_unlocked(&gem_to_amdgpu_bo(gobj)->tbo);
 
-static int amdgpu_cs_process_syncobj_in_dep(struct amdgpu_cs_parser *p,
-					    struct amdgpu_cs_chunk *chunk)
-{
-	unsigned num_deps;
-	int i, r;
-	struct drm_amdgpu_cs_chunk_sem *deps;
+		/* Everybody except for the gang leader uses READ */
+		for (i = 0; i < p->gang_size; ++i) {
+			if (p->jobs[i] == leader)
+				continue;
 
-	deps = (struct drm_amdgpu_cs_chunk_sem *)chunk->kdata;
-	num_deps = chunk->length_dw * 4 /
-		sizeof(struct drm_amdgpu_cs_chunk_sem);
+			dma_resv_add_fence(gobj->resv,
+					   &p->jobs[i]->base.s_fence->finished,
+					   DMA_RESV_USAGE_READ);
+		}
 
-	for (i = 0; i < num_deps; ++i) {
-		r = amdgpu_syncobj_lookup_and_add_to_sync(p, deps[i].handle);
-		if (r)
-			return r;
+		/* The gang leader as remembered as writer */
+		dma_resv_add_fence(gobj->resv, p->fence, DMA_RESV_USAGE_WRITE);
 	}
-	return 0;
-}
 
-static int amdgpu_cs_process_syncobj_out_dep(struct amdgpu_cs_parser *p,
-					     struct amdgpu_cs_chunk *chunk)
-{
-	unsigned num_deps;
-	int i;
-	struct drm_amdgpu_cs_chunk_sem *deps;
-	deps = (struct drm_amdgpu_cs_chunk_sem *)chunk->kdata;
-	num_deps = chunk->length_dw * 4 /
-		sizeof(struct drm_amdgpu_cs_chunk_sem);
-
-	p->post_dep_syncobjs = kmalloc_array(num_deps,
-					     sizeof(struct drm_syncobj *),
-					     GFP_KERNEL);
-	p->num_post_dep_syncobjs = 0;
-
-	if (!p->post_dep_syncobjs)
-		return -ENOMEM;
+	seq = amdgpu_ctx_add_fence(p->ctx, p->entities[p->gang_leader_idx],
+				   p->fence);
+	amdgpu_cs_post_dependencies(p);
 
-	for (i = 0; i < num_deps; ++i) {
-		p->post_dep_syncobjs[i] = drm_syncobj_find(p->filp, deps[i].handle);
-		if (!p->post_dep_syncobjs[i])
-			return -EINVAL;
-		p->num_post_dep_syncobjs++;
+	if ((leader->preamble_status & AMDGPU_PREAMBLE_IB_PRESENT) &&
+	    !p->ctx->preamble_presented) {
+		leader->preamble_status |= AMDGPU_PREAMBLE_IB_PRESENT_FIRST;
+		p->ctx->preamble_presented = true;
 	}
-	return 0;
-}
-
-static int amdgpu_cs_dependencies(struct amdgpu_device *adev,
-				  struct amdgpu_cs_parser *p)
-{
-	int i, r;
-
-	for (i = 0; i < p->nchunks; ++i) {
-		struct amdgpu_cs_chunk *chunk;
 
-		chunk = &p->chunks[i];
+	cs->out.handle = seq;
+	leader->uf_sequence = seq;
 
-		if (chunk->chunk_id == AMDGPU_CHUNK_ID_DEPENDENCIES) {
-			r = amdgpu_cs_process_fence_dep(p, chunk);
-			if (r)
-				return r;
-		} else if (chunk->chunk_id == AMDGPU_CHUNK_ID_SYNCOBJ_IN) {
-			r = amdgpu_cs_process_syncobj_in_dep(p, chunk);
-			if (r)
-				return r;
-		} else if (chunk->chunk_id == AMDGPU_CHUNK_ID_SYNCOBJ_OUT) {
-			r = amdgpu_cs_process_syncobj_out_dep(p, chunk);
-			if (r)
-				return r;
-		}
+	amdgpu_vm_bo_trace_cs(&fpriv->vm, &p->exec.ticket);
+	for (i = 0; i < p->gang_size; ++i) {
+		amdgpu_job_free_resources(p->jobs[i]);
+		trace_amdgpu_cs_ioctl(p->jobs[i]);
+		drm_sched_entity_push_job(&p->jobs[i]->base);
+		p->jobs[i] = NULL;
 	}
 
-	return 0;
-}
+	amdgpu_vm_move_to_lru_tail(p->adev, &fpriv->vm);
 
-static void amdgpu_cs_post_dependencies(struct amdgpu_cs_parser *p)
-{
-	int i;
-
-	for (i = 0; i < p->num_post_dep_syncobjs; ++i)
-		drm_syncobj_replace_fence(p->post_dep_syncobjs[i], p->fence);
+	mutex_unlock(&p->adev->notifier_lock);
+	mutex_unlock(&p->bo_list->bo_list_mutex);
+	return 0;
 }
 
-static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
-			    union drm_amdgpu_cs *cs)
+/* Cleanup the parser structure */
+static void amdgpu_cs_parser_fini(struct amdgpu_cs_parser *parser)
 {
-	struct amdgpu_ring *ring = p->job->ring;
-	struct amd_sched_entity *entity = &p->ctx->rings[ring->idx].entity;
-	struct amdgpu_job *job;
-	int r;
+	unsigned int i;
 
-	job = p->job;
-	p->job = NULL;
+	amdgpu_sync_free(&parser->sync);
+	drm_exec_fini(&parser->exec);
 
-	r = amd_sched_job_init(&job->base, &ring->sched, entity, p->filp);
-	if (r) {
-		amdgpu_job_free(job);
-		return r;
+	for (i = 0; i < parser->num_post_deps; i++) {
+		drm_syncobj_put(parser->post_deps[i].syncobj);
+		kfree(parser->post_deps[i].chain);
 	}
+	kfree(parser->post_deps);
 
-	job->owner = p->filp;
-	job->fence_ctx = entity->fence_context;
-	p->fence = dma_fence_get(&job->base.s_fence->finished);
-
-	amdgpu_cs_post_dependencies(p);
+	dma_fence_put(parser->fence);
 
-	cs->out.handle = amdgpu_ctx_add_fence(p->ctx, ring, p->fence);
-	job->uf_sequence = cs->out.handle;
-	amdgpu_job_free_resources(job);
+	if (parser->ctx)
+		amdgpu_ctx_put(parser->ctx);
+	if (parser->bo_list)
+		amdgpu_bo_list_put(parser->bo_list);
 
-	trace_amdgpu_cs_ioctl(job);
-	amd_sched_entity_push_job(&job->base);
-	return 0;
+	for (i = 0; i < parser->nchunks; i++)
+		kvfree(parser->chunks[i].kdata);
+	kvfree(parser->chunks);
+	for (i = 0; i < parser->gang_size; ++i) {
+		if (parser->jobs[i])
+			amdgpu_job_free(parser->jobs[i]);
+	}
+	amdgpu_bo_unref(&parser->uf_bo);
 }
 
 int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 {
-	struct amdgpu_device *adev = dev->dev_private;
-	struct amdgpu_fpriv *fpriv = filp->driver_priv;
-	union drm_amdgpu_cs *cs = data;
-	struct amdgpu_cs_parser parser = {};
-	bool reserved_buffers = false;
-	int i, r;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	struct amdgpu_cs_parser parser;
+	int r;
+
+	if (amdgpu_ras_intr_triggered())
+		return -EHWPOISON;
 
 	if (!adev->accel_working)
 		return -EBUSY;
-	if (amdgpu_kms_vram_lost(adev, fpriv))
-		return -ENODEV;
-
-	parser.adev = adev;
-	parser.filp = filp;
 
-	r = amdgpu_cs_parser_init(&parser, data);
+	r = amdgpu_cs_parser_init(&parser, adev, filp, data);
 	if (r) {
-		DRM_ERROR("Failed to initialize parser !\n");
-		goto out;
+		drm_err_ratelimited(dev, "Failed to initialize parser %d!\n", r);
+		return r;
 	}
 
+	r = amdgpu_cs_pass1(&parser, data);
+	if (r)
+		goto error_fini;
+
+	r = amdgpu_cs_pass2(&parser);
+	if (r)
+		goto error_fini;
+
 	r = amdgpu_cs_parser_bos(&parser, data);
 	if (r) {
 		if (r == -ENOMEM)
-			DRM_ERROR("Not enough memory for command submission!\n");
-		else if (r != -ERESTARTSYS)
-			DRM_ERROR("Failed to process the buffer list %d!\n", r);
-		goto out;
+			drm_err(dev, "Not enough memory for command submission!\n");
+		else if (r != -ERESTARTSYS && r != -EAGAIN)
+			drm_dbg(dev, "Failed to process the buffer list %d!\n", r);
+		goto error_fini;
 	}
 
-	reserved_buffers = true;
-	r = amdgpu_cs_ib_fill(adev, &parser);
+	r = amdgpu_cs_patch_jobs(&parser);
 	if (r)
-		goto out;
+		goto error_backoff;
 
-	r = amdgpu_cs_dependencies(adev, &parser);
-	if (r) {
-		DRM_ERROR("Failed in the dependencies handling %d!\n", r);
-		goto out;
-	}
+	r = amdgpu_cs_vm_handling(&parser);
+	if (r)
+		goto error_backoff;
+
+	r = amdgpu_cs_sync_rings(&parser);
+	if (r)
+		goto error_backoff;
 
-	for (i = 0; i < parser.job->num_ibs; i++)
-		trace_amdgpu_cs(&parser, i);
+	trace_amdgpu_cs_ibs(&parser);
 
-	r = amdgpu_cs_ib_vm_chunk(adev, &parser);
+	r = amdgpu_cs_submit(&parser, data);
 	if (r)
-		goto out;
+		goto error_backoff;
 
-	r = amdgpu_cs_submit(&parser, cs);
+	amdgpu_cs_parser_fini(&parser);
+	return 0;
 
-out:
-	amdgpu_cs_parser_fini(&parser, r, reserved_buffers);
+error_backoff:
+	mutex_unlock(&parser.bo_list->bo_list_mutex);
+
+error_fini:
+	amdgpu_cs_parser_fini(&parser);
 	return r;
 }
 
@@ -1229,34 +1482,30 @@ int amdgpu_cs_wait_ioctl(struct drm_device *dev, void *data,
 			 struct drm_file *filp)
 {
 	union drm_amdgpu_wait_cs *wait = data;
-	struct amdgpu_device *adev = dev->dev_private;
-	struct amdgpu_fpriv *fpriv = filp->driver_priv;
 	unsigned long timeout = amdgpu_gem_timeout(wait->in.timeout);
-	struct amdgpu_ring *ring = NULL;
+	struct drm_sched_entity *entity;
 	struct amdgpu_ctx *ctx;
 	struct dma_fence *fence;
 	long r;
 
-	if (amdgpu_kms_vram_lost(adev, fpriv))
-		return -ENODEV;
-
 	ctx = amdgpu_ctx_get(filp->driver_priv, wait->in.ctx_id);
 	if (ctx == NULL)
 		return -EINVAL;
 
-	r = amdgpu_queue_mgr_map(adev, &ctx->queue_mgr,
-				 wait->in.ip_type, wait->in.ip_instance,
-				 wait->in.ring, &ring);
+	r = amdgpu_ctx_get_entity(ctx, wait->in.ip_type, wait->in.ip_instance,
+				  wait->in.ring, &entity);
 	if (r) {
 		amdgpu_ctx_put(ctx);
 		return r;
 	}
 
-	fence = amdgpu_ctx_get_fence(ctx, ring, wait->in.handle);
+	fence = amdgpu_ctx_get_fence(ctx, entity, wait->in.handle);
 	if (IS_ERR(fence))
 		r = PTR_ERR(fence);
 	else if (fence) {
 		r = dma_fence_wait_timeout(fence, true, timeout);
+		if (r > 0 && fence->error)
+			r = fence->error;
 		dma_fence_put(fence);
 	} else
 		r = 1;
@@ -1282,7 +1531,7 @@ static struct dma_fence *amdgpu_cs_get_fence(struct amdgpu_device *adev,
 					     struct drm_file *filp,
 					     struct drm_amdgpu_fence *user)
 {
-	struct amdgpu_ring *ring;
+	struct drm_sched_entity *entity;
 	struct amdgpu_ctx *ctx;
 	struct dma_fence *fence;
 	int r;
@@ -1291,21 +1540,81 @@ static struct dma_fence *amdgpu_cs_get_fence(struct amdgpu_device *adev,
 	if (ctx == NULL)
 		return ERR_PTR(-EINVAL);
 
-	r = amdgpu_queue_mgr_map(adev, &ctx->queue_mgr, user->ip_type,
-				 user->ip_instance, user->ring, &ring);
+	r = amdgpu_ctx_get_entity(ctx, user->ip_type, user->ip_instance,
+				  user->ring, &entity);
 	if (r) {
 		amdgpu_ctx_put(ctx);
 		return ERR_PTR(r);
 	}
 
-	fence = amdgpu_ctx_get_fence(ctx, ring, user->seq_no);
+	fence = amdgpu_ctx_get_fence(ctx, entity, user->seq_no);
 	amdgpu_ctx_put(ctx);
 
 	return fence;
 }
 
+int amdgpu_cs_fence_to_handle_ioctl(struct drm_device *dev, void *data,
+				    struct drm_file *filp)
+{
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	union drm_amdgpu_fence_to_handle *info = data;
+	struct dma_fence *fence;
+	struct drm_syncobj *syncobj;
+	struct sync_file *sync_file;
+	int fd, r;
+
+	fence = amdgpu_cs_get_fence(adev, filp, &info->in.fence);
+	if (IS_ERR(fence))
+		return PTR_ERR(fence);
+
+	if (!fence)
+		fence = dma_fence_get_stub();
+
+	switch (info->in.what) {
+	case AMDGPU_FENCE_TO_HANDLE_GET_SYNCOBJ:
+		r = drm_syncobj_create(&syncobj, 0, fence);
+		dma_fence_put(fence);
+		if (r)
+			return r;
+		r = drm_syncobj_get_handle(filp, syncobj, &info->out.handle);
+		drm_syncobj_put(syncobj);
+		return r;
+
+	case AMDGPU_FENCE_TO_HANDLE_GET_SYNCOBJ_FD:
+		r = drm_syncobj_create(&syncobj, 0, fence);
+		dma_fence_put(fence);
+		if (r)
+			return r;
+		r = drm_syncobj_get_fd(syncobj, (int *)&info->out.handle);
+		drm_syncobj_put(syncobj);
+		return r;
+
+	case AMDGPU_FENCE_TO_HANDLE_GET_SYNC_FILE_FD:
+		fd = get_unused_fd_flags(O_CLOEXEC);
+		if (fd < 0) {
+			dma_fence_put(fence);
+			return fd;
+		}
+
+		sync_file = sync_file_create(fence);
+		dma_fence_put(fence);
+		if (!sync_file) {
+			put_unused_fd(fd);
+			return -ENOMEM;
+		}
+
+		fd_install(fd, sync_file->file);
+		info->out.handle = fd;
+		return 0;
+
+	default:
+		dma_fence_put(fence);
+		return -EINVAL;
+	}
+}
+
 /**
- * amdgpu_cs_wait_all_fence - wait on all fences to signal
+ * amdgpu_cs_wait_all_fences - wait on all fences to signal
  *
  * @adev: amdgpu device
  * @filp: file private
@@ -1332,6 +1641,9 @@ static int amdgpu_cs_wait_all_fences(struct amdgpu_device *adev,
 			continue;
 
 		r = dma_fence_wait_timeout(fence, true, timeout);
+		if (r > 0 && fence->error)
+			r = fence->error;
+
 		dma_fence_put(fence);
 		if (r < 0)
 			return r;
@@ -1383,6 +1695,7 @@ static int amdgpu_cs_wait_any_fence(struct amdgpu_device *adev,
 			array[i] = fence;
 		} else { /* NULL, the fence has been already signaled */
 			r = 1;
+			first = i;
 			goto out;
 		}
 	}
@@ -1396,8 +1709,11 @@ out:
 	memset(wait, 0, sizeof(*wait));
 	wait->out.status = (r > 0);
 	wait->out.first_signaled = first;
-	/* set return value 0 to indicate success */
-	r = 0;
+
+	if (first < fence_count && array[first])
+		r = array[first]->error;
+	else
+		r = 0;
 
 err_free_fence_array:
 	for (i = 0; i < fence_count; i++)
@@ -1417,123 +1733,75 @@ err_free_fence_array:
 int amdgpu_cs_wait_fences_ioctl(struct drm_device *dev, void *data,
 				struct drm_file *filp)
 {
-	struct amdgpu_device *adev = dev->dev_private;
-	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	union drm_amdgpu_wait_fences *wait = data;
-	uint32_t fence_count = wait->in.fence_count;
-	struct drm_amdgpu_fence *fences_user;
 	struct drm_amdgpu_fence *fences;
 	int r;
 
-	if (amdgpu_kms_vram_lost(adev, fpriv))
-		return -ENODEV;
 	/* Get the fences from userspace */
-	fences = kmalloc_array(fence_count, sizeof(struct drm_amdgpu_fence),
-			GFP_KERNEL);
-	if (fences == NULL)
-		return -ENOMEM;
-
-	fences_user = u64_to_user_ptr(wait->in.fences);
-	if (copy_from_user(fences, fences_user,
-		sizeof(struct drm_amdgpu_fence) * fence_count)) {
-		r = -EFAULT;
-		goto err_free_fences;
-	}
+	fences = memdup_array_user(u64_to_user_ptr(wait->in.fences),
+				   wait->in.fence_count,
+				   sizeof(struct drm_amdgpu_fence));
+	if (IS_ERR(fences))
+		return PTR_ERR(fences);
 
 	if (wait->in.wait_all)
 		r = amdgpu_cs_wait_all_fences(adev, filp, wait, fences);
 	else
 		r = amdgpu_cs_wait_any_fence(adev, filp, wait, fences);
 
-err_free_fences:
 	kfree(fences);
 
 	return r;
 }
 
 /**
- * amdgpu_cs_find_bo_va - find bo_va for VM address
+ * amdgpu_cs_find_mapping - find bo_va for VM address
  *
  * @parser: command submission parser context
  * @addr: VM address
  * @bo: resulting BO of the mapping found
+ * @map: Placeholder to return found BO mapping
  *
  * Search the buffer objects in the command submission context for a certain
  * virtual memory address. Returns allocation structure when found, NULL
  * otherwise.
  */
-struct amdgpu_bo_va_mapping *
-amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser,
-		       uint64_t addr, struct amdgpu_bo **bo)
+int amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser,
+			   uint64_t addr, struct amdgpu_bo **bo,
+			   struct amdgpu_bo_va_mapping **map)
 {
+	struct amdgpu_fpriv *fpriv = parser->filp->driver_priv;
+	struct ttm_operation_ctx ctx = { false, false };
+	struct amdgpu_vm *vm = &fpriv->vm;
 	struct amdgpu_bo_va_mapping *mapping;
-	unsigned i;
-
-	if (!parser->bo_list)
-		return NULL;
+	int i, r;
 
 	addr /= AMDGPU_GPU_PAGE_SIZE;
 
-	for (i = 0; i < parser->bo_list->num_entries; i++) {
-		struct amdgpu_bo_list_entry *lobj;
-
-		lobj = &parser->bo_list->array[i];
-		if (!lobj->bo_va)
-			continue;
-
-		list_for_each_entry(mapping, &lobj->bo_va->valids, list) {
-			if (mapping->start > addr ||
-			    addr > mapping->last)
-				continue;
-
-			*bo = lobj->bo_va->base.bo;
-			return mapping;
-		}
-
-		list_for_each_entry(mapping, &lobj->bo_va->invalids, list) {
-			if (mapping->start > addr ||
-			    addr > mapping->last)
-				continue;
-
-			*bo = lobj->bo_va->base.bo;
-			return mapping;
-		}
-	}
-
-	return NULL;
-}
-
-/**
- * amdgpu_cs_sysvm_access_required - make BOs accessible by the system VM
- *
- * @parser: command submission parser context
- *
- * Helper for UVD/VCE VM emulation, make sure BOs are accessible by the system VM.
- */
-int amdgpu_cs_sysvm_access_required(struct amdgpu_cs_parser *parser)
-{
-	unsigned i;
-	int r;
-
-	if (!parser->bo_list)
-		return 0;
+	mapping = amdgpu_vm_bo_lookup_mapping(vm, addr);
+	if (!mapping || !mapping->bo_va || !mapping->bo_va->base.bo)
+		return -EINVAL;
 
-	for (i = 0; i < parser->bo_list->num_entries; i++) {
-		struct amdgpu_bo *bo = parser->bo_list->array[i].robj;
+	*bo = mapping->bo_va->base.bo;
+	*map = mapping;
 
-		r = amdgpu_ttm_bind(&bo->tbo, &bo->tbo.mem);
-		if (unlikely(r))
-			return r;
+	/* Double check that the BO is reserved by this CS */
+	if (dma_resv_locking_ctx((*bo)->tbo.base.resv) != &parser->exec.ticket)
+		return -EINVAL;
 
-		if (bo->flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS)
-			continue;
+	/* Make sure VRAM is allocated contigiously */
+	(*bo)->flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+	if ((*bo)->tbo.resource->mem_type == TTM_PL_VRAM &&
+	    !((*bo)->tbo.resource->placement & TTM_PL_FLAG_CONTIGUOUS)) {
 
-		bo->flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
-		amdgpu_ttm_placement_from_domain(bo, bo->allowed_domains);
-		r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false);
-		if (unlikely(r))
+		amdgpu_bo_placement_from_domain(*bo, (*bo)->allowed_domains);
+		for (i = 0; i < (*bo)->placement.num_placement; i++)
+			(*bo)->placements[i].flags |= TTM_PL_FLAG_CONTIGUOUS;
+		r = ttm_bo_validate(&(*bo)->tbo, &(*bo)->placement, &ctx);
+		if (r)
 			return r;
 	}
 
-	return 0;
+	return amdgpu_ttm_alloc_gart(&(*bo)->tbo);
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.h
new file mode 100644
index 000000000000..39c33ad100cb
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.h
@@ -0,0 +1,88 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __AMDGPU_CS_H__
+#define __AMDGPU_CS_H__
+
+#include <linux/ww_mutex.h>
+#include <drm/drm_exec.h>
+
+#include "amdgpu_job.h"
+#include "amdgpu_bo_list.h"
+#include "amdgpu_ring.h"
+
+#define AMDGPU_CS_GANG_SIZE	4
+
+struct amdgpu_bo_va_mapping;
+
+struct amdgpu_cs_chunk {
+	uint32_t		chunk_id;
+	uint32_t		length_dw;
+	void			*kdata;
+};
+
+struct amdgpu_cs_post_dep {
+	struct drm_syncobj *syncobj;
+	struct dma_fence_chain *chain;
+	u64 point;
+};
+
+struct amdgpu_cs_parser {
+	struct amdgpu_device	*adev;
+	struct drm_file		*filp;
+	struct amdgpu_ctx	*ctx;
+
+	/* chunks */
+	unsigned		nchunks;
+	struct amdgpu_cs_chunk	*chunks;
+
+	/* scheduler job objects */
+	unsigned int		gang_size;
+	unsigned int		gang_leader_idx;
+	struct drm_sched_entity	*entities[AMDGPU_CS_GANG_SIZE];
+	struct amdgpu_job	*jobs[AMDGPU_CS_GANG_SIZE];
+	struct amdgpu_job	*gang_leader;
+
+	/* buffer objects */
+	struct drm_exec			exec;
+	struct amdgpu_bo_list		*bo_list;
+	struct amdgpu_mn		*mn;
+	struct dma_fence		*fence;
+	uint64_t			bytes_moved_threshold;
+	uint64_t			bytes_moved_vis_threshold;
+	uint64_t			bytes_moved;
+	uint64_t			bytes_moved_vis;
+
+	/* user fence */
+	struct amdgpu_bo		*uf_bo;
+
+	unsigned			num_post_deps;
+	struct amdgpu_cs_post_dep	*post_deps;
+
+	struct amdgpu_sync		sync;
+};
+
+int amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser,
+			   uint64_t addr, struct amdgpu_bo **bo,
+			   struct amdgpu_bo_va_mapping **mapping);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c
new file mode 100644
index 000000000000..02138aa55793
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c
@@ -0,0 +1,135 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+
+ * * Author: Monk.liu@amd.com
+ */
+
+#include <drm/drm_exec.h>
+
+#include "amdgpu.h"
+
+uint64_t amdgpu_csa_vaddr(struct amdgpu_device *adev)
+{
+	uint64_t addr = AMDGPU_VA_RESERVED_CSA_START(adev);
+
+	addr = amdgpu_gmc_sign_extend(addr);
+
+	return addr;
+}
+
+int amdgpu_allocate_static_csa(struct amdgpu_device *adev, struct amdgpu_bo **bo,
+				u32 domain, uint32_t size)
+{
+	void *ptr;
+
+	amdgpu_bo_create_kernel(adev, size, PAGE_SIZE,
+				domain, bo,
+				NULL, &ptr);
+	if (!*bo)
+		return -ENOMEM;
+
+	memset(ptr, 0, size);
+	adev->virt.csa_cpu_addr = ptr;
+	return 0;
+}
+
+void amdgpu_free_static_csa(struct amdgpu_bo **bo)
+{
+	amdgpu_bo_free_kernel(bo, NULL, NULL);
+}
+
+/*
+ * amdgpu_map_static_csa should be called during amdgpu_vm_init
+ * it maps virtual address amdgpu_csa_vaddr() to this VM, and each command
+ * submission of GFX should use this virtual address within META_DATA init
+ * package to support SRIOV gfx preemption.
+ */
+int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			  struct amdgpu_bo *bo, struct amdgpu_bo_va **bo_va,
+			  uint64_t csa_addr, uint32_t size)
+{
+	struct drm_exec exec;
+	int r;
+
+	drm_exec_init(&exec, DRM_EXEC_INTERRUPTIBLE_WAIT, 0);
+	drm_exec_until_all_locked(&exec) {
+		r = amdgpu_vm_lock_pd(vm, &exec, 0);
+		if (likely(!r))
+			r = drm_exec_lock_obj(&exec, &bo->tbo.base);
+		drm_exec_retry_on_contention(&exec);
+		if (unlikely(r)) {
+			DRM_ERROR("failed to reserve CSA,PD BOs: err=%d\n", r);
+			goto error;
+		}
+	}
+
+	*bo_va = amdgpu_vm_bo_add(adev, vm, bo);
+	if (!*bo_va) {
+		r = -ENOMEM;
+		goto error;
+	}
+
+	r = amdgpu_vm_bo_map(adev, *bo_va, csa_addr, 0, size,
+			     AMDGPU_PTE_READABLE | AMDGPU_PTE_WRITEABLE |
+			     AMDGPU_PTE_EXECUTABLE);
+
+	if (r) {
+		DRM_ERROR("failed to do bo_map on static CSA, err=%d\n", r);
+		amdgpu_vm_bo_del(adev, *bo_va);
+		goto error;
+	}
+
+error:
+	drm_exec_fini(&exec);
+	return r;
+}
+
+int amdgpu_unmap_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			    struct amdgpu_bo *bo, struct amdgpu_bo_va *bo_va,
+			    uint64_t csa_addr)
+{
+	struct drm_exec exec;
+	int r;
+
+	drm_exec_init(&exec, 0, 0);
+	drm_exec_until_all_locked(&exec) {
+		r = amdgpu_vm_lock_pd(vm, &exec, 0);
+		if (likely(!r))
+			r = drm_exec_lock_obj(&exec, &bo->tbo.base);
+		drm_exec_retry_on_contention(&exec);
+		if (unlikely(r)) {
+			DRM_ERROR("failed to reserve CSA,PD BOs: err=%d\n", r);
+			goto error;
+		}
+	}
+
+	r = amdgpu_vm_bo_unmap(adev, bo_va, csa_addr);
+	if (r) {
+		DRM_ERROR("failed to do bo_unmap on static CSA, err=%d\n", r);
+		goto error;
+	}
+
+	amdgpu_vm_bo_del(adev, bo_va);
+
+error:
+	drm_exec_fini(&exec);
+	return r;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.h
new file mode 100644
index 000000000000..7dfc1f2012eb
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.h
@@ -0,0 +1,42 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Author: Monk.liu@amd.com
+ */
+
+#ifndef AMDGPU_CSA_MANAGER_H
+#define AMDGPU_CSA_MANAGER_H
+
+#define AMDGPU_CSA_SIZE		(128 * 1024)
+
+uint32_t amdgpu_get_total_csa_size(struct amdgpu_device *adev);
+uint64_t amdgpu_csa_vaddr(struct amdgpu_device *adev);
+int amdgpu_allocate_static_csa(struct amdgpu_device *adev, struct amdgpu_bo **bo,
+				u32 domain, uint32_t size);
+int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			  struct amdgpu_bo *bo, struct amdgpu_bo_va **bo_va,
+			  uint64_t csa_addr, uint32_t size);
+int amdgpu_unmap_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			    struct amdgpu_bo *bo, struct amdgpu_bo_va *bo_va,
+			    uint64_t csa_addr);
+void amdgpu_free_static_csa(struct amdgpu_bo **bo);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
index a11e44340b23..f5d5c45ddc0d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
@@ -22,84 +22,458 @@
  * Authors: monk liu <monk.liu@amd.com>
  */
 
-#include <drm/drmP.h>
+#include <drm/drm_auth.h>
+#include <drm/drm_drv.h>
 #include "amdgpu.h"
+#include "amdgpu_sched.h"
+#include "amdgpu_ras.h"
+#include <linux/nospec.h>
+
+#define to_amdgpu_ctx_entity(e)	\
+	container_of((e), struct amdgpu_ctx_entity, entity)
+
+const unsigned int amdgpu_ctx_num_entities[AMDGPU_HW_IP_NUM] = {
+	[AMDGPU_HW_IP_GFX]	=	1,
+	[AMDGPU_HW_IP_COMPUTE]	=	4,
+	[AMDGPU_HW_IP_DMA]	=	2,
+	[AMDGPU_HW_IP_UVD]	=	1,
+	[AMDGPU_HW_IP_VCE]	=	1,
+	[AMDGPU_HW_IP_UVD_ENC]	=	1,
+	[AMDGPU_HW_IP_VCN_DEC]	=	1,
+	[AMDGPU_HW_IP_VCN_ENC]	=	1,
+	[AMDGPU_HW_IP_VCN_JPEG]	=	1,
+	[AMDGPU_HW_IP_VPE]	=	1,
+};
+
+bool amdgpu_ctx_priority_is_valid(int32_t ctx_prio)
+{
+	switch (ctx_prio) {
+	case AMDGPU_CTX_PRIORITY_VERY_LOW:
+	case AMDGPU_CTX_PRIORITY_LOW:
+	case AMDGPU_CTX_PRIORITY_NORMAL:
+	case AMDGPU_CTX_PRIORITY_HIGH:
+	case AMDGPU_CTX_PRIORITY_VERY_HIGH:
+		return true;
+	default:
+	case AMDGPU_CTX_PRIORITY_UNSET:
+		/* UNSET priority is not valid and we don't carry that
+		 * around, but set it to NORMAL in the only place this
+		 * function is called, amdgpu_ctx_ioctl().
+		 */
+		return false;
+	}
+}
 
-static int amdgpu_ctx_init(struct amdgpu_device *adev, struct amdgpu_ctx *ctx)
+static enum drm_sched_priority
+amdgpu_ctx_to_drm_sched_prio(int32_t ctx_prio)
 {
-	unsigned i, j;
-	int r;
+	switch (ctx_prio) {
+	case AMDGPU_CTX_PRIORITY_UNSET:
+		pr_warn_once("AMD-->DRM context priority value UNSET-->NORMAL");
+		return DRM_SCHED_PRIORITY_NORMAL;
 
-	memset(ctx, 0, sizeof(*ctx));
-	ctx->adev = adev;
-	kref_init(&ctx->refcount);
-	spin_lock_init(&ctx->ring_lock);
-	ctx->fences = kcalloc(amdgpu_sched_jobs * AMDGPU_MAX_RINGS,
-			      sizeof(struct dma_fence*), GFP_KERNEL);
-	if (!ctx->fences)
-		return -ENOMEM;
+	case AMDGPU_CTX_PRIORITY_VERY_LOW:
+		return DRM_SCHED_PRIORITY_LOW;
+
+	case AMDGPU_CTX_PRIORITY_LOW:
+		return DRM_SCHED_PRIORITY_LOW;
+
+	case AMDGPU_CTX_PRIORITY_NORMAL:
+		return DRM_SCHED_PRIORITY_NORMAL;
+
+	case AMDGPU_CTX_PRIORITY_HIGH:
+		return DRM_SCHED_PRIORITY_HIGH;
+
+	case AMDGPU_CTX_PRIORITY_VERY_HIGH:
+		return DRM_SCHED_PRIORITY_HIGH;
 
-	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
-		ctx->rings[i].sequence = 1;
-		ctx->rings[i].fences = &ctx->fences[amdgpu_sched_jobs * i];
+	/* This should not happen as we sanitized userspace provided priority
+	 * already, WARN if this happens.
+	 */
+	default:
+		WARN(1, "Invalid context priority %d\n", ctx_prio);
+		return DRM_SCHED_PRIORITY_NORMAL;
 	}
 
-	ctx->reset_counter = atomic_read(&adev->gpu_reset_counter);
+}
 
-	/* create context entity for each ring */
-	for (i = 0; i < adev->num_rings; i++) {
-		struct amdgpu_ring *ring = adev->rings[i];
-		struct amd_sched_rq *rq;
+static int amdgpu_ctx_priority_permit(struct drm_file *filp,
+				      int32_t priority)
+{
+	/* NORMAL and below are accessible by everyone */
+	if (priority <= AMDGPU_CTX_PRIORITY_NORMAL)
+		return 0;
 
-		rq = &ring->sched.sched_rq[AMD_SCHED_PRIORITY_NORMAL];
+	if (capable(CAP_SYS_NICE))
+		return 0;
 
-		if (ring == &adev->gfx.kiq.ring)
-			continue;
+	if (drm_is_current_master(filp))
+		return 0;
 
-		r = amd_sched_entity_init(&ring->sched, &ctx->rings[i].entity,
-					  rq, amdgpu_sched_jobs);
+	return -EACCES;
+}
+
+static enum amdgpu_gfx_pipe_priority amdgpu_ctx_prio_to_gfx_pipe_prio(int32_t prio)
+{
+	switch (prio) {
+	case AMDGPU_CTX_PRIORITY_HIGH:
+	case AMDGPU_CTX_PRIORITY_VERY_HIGH:
+		return AMDGPU_GFX_PIPE_PRIO_HIGH;
+	default:
+		return AMDGPU_GFX_PIPE_PRIO_NORMAL;
+	}
+}
+
+static enum amdgpu_ring_priority_level amdgpu_ctx_sched_prio_to_ring_prio(int32_t prio)
+{
+	switch (prio) {
+	case AMDGPU_CTX_PRIORITY_HIGH:
+		return AMDGPU_RING_PRIO_1;
+	case AMDGPU_CTX_PRIORITY_VERY_HIGH:
+		return AMDGPU_RING_PRIO_2;
+	default:
+		return AMDGPU_RING_PRIO_0;
+	}
+}
+
+static unsigned int amdgpu_ctx_get_hw_prio(struct amdgpu_ctx *ctx, u32 hw_ip)
+{
+	struct amdgpu_device *adev = ctx->mgr->adev;
+	unsigned int hw_prio;
+	int32_t ctx_prio;
+
+	ctx_prio = (ctx->override_priority == AMDGPU_CTX_PRIORITY_UNSET) ?
+			ctx->init_priority : ctx->override_priority;
+
+	switch (hw_ip) {
+	case AMDGPU_HW_IP_GFX:
+	case AMDGPU_HW_IP_COMPUTE:
+		hw_prio = amdgpu_ctx_prio_to_gfx_pipe_prio(ctx_prio);
+		break;
+	case AMDGPU_HW_IP_VCE:
+	case AMDGPU_HW_IP_VCN_ENC:
+		hw_prio = amdgpu_ctx_sched_prio_to_ring_prio(ctx_prio);
+		break;
+	default:
+		hw_prio = AMDGPU_RING_PRIO_DEFAULT;
+		break;
+	}
+
+	hw_ip = array_index_nospec(hw_ip, AMDGPU_HW_IP_NUM);
+	if (adev->gpu_sched[hw_ip][hw_prio].num_scheds == 0)
+		hw_prio = AMDGPU_RING_PRIO_DEFAULT;
+
+	return hw_prio;
+}
+
+/* Calculate the time spend on the hw */
+static ktime_t amdgpu_ctx_fence_time(struct dma_fence *fence)
+{
+	struct drm_sched_fence *s_fence;
+
+	if (!fence)
+		return ns_to_ktime(0);
+
+	/* When the fence is not even scheduled it can't have spend time */
+	s_fence = to_drm_sched_fence(fence);
+	if (!test_bit(DMA_FENCE_FLAG_TIMESTAMP_BIT, &s_fence->scheduled.flags))
+		return ns_to_ktime(0);
+
+	/* When it is still running account how much already spend */
+	if (!test_bit(DMA_FENCE_FLAG_TIMESTAMP_BIT, &s_fence->finished.flags))
+		return ktime_sub(ktime_get(), s_fence->scheduled.timestamp);
+
+	return ktime_sub(s_fence->finished.timestamp,
+			 s_fence->scheduled.timestamp);
+}
+
+static ktime_t amdgpu_ctx_entity_time(struct amdgpu_ctx *ctx,
+				      struct amdgpu_ctx_entity *centity)
+{
+	ktime_t res = ns_to_ktime(0);
+	uint32_t i;
+
+	spin_lock(&ctx->ring_lock);
+	for (i = 0; i < amdgpu_sched_jobs; i++) {
+		res = ktime_add(res, amdgpu_ctx_fence_time(centity->fences[i]));
+	}
+	spin_unlock(&ctx->ring_lock);
+	return res;
+}
+
+static int amdgpu_ctx_init_entity(struct amdgpu_ctx *ctx, u32 hw_ip,
+				  const u32 ring)
+{
+	struct drm_gpu_scheduler **scheds = NULL, *sched = NULL;
+	struct amdgpu_device *adev = ctx->mgr->adev;
+	struct amdgpu_ctx_entity *entity;
+	enum drm_sched_priority drm_prio;
+	unsigned int hw_prio, num_scheds;
+	int32_t ctx_prio;
+	int r;
+
+	entity = kzalloc(struct_size(entity, fences, amdgpu_sched_jobs),
+			 GFP_KERNEL);
+	if (!entity)
+		return  -ENOMEM;
+
+	ctx_prio = (ctx->override_priority == AMDGPU_CTX_PRIORITY_UNSET) ?
+			ctx->init_priority : ctx->override_priority;
+	entity->hw_ip = hw_ip;
+	entity->sequence = 1;
+	hw_prio = amdgpu_ctx_get_hw_prio(ctx, hw_ip);
+	drm_prio = amdgpu_ctx_to_drm_sched_prio(ctx_prio);
+
+	hw_ip = array_index_nospec(hw_ip, AMDGPU_HW_IP_NUM);
+
+	if (!(adev)->xcp_mgr) {
+		scheds = adev->gpu_sched[hw_ip][hw_prio].sched;
+		num_scheds = adev->gpu_sched[hw_ip][hw_prio].num_scheds;
+	} else {
+		struct amdgpu_fpriv *fpriv;
+
+		fpriv = container_of(ctx->ctx_mgr, struct amdgpu_fpriv, ctx_mgr);
+		r = amdgpu_xcp_select_scheds(adev, hw_ip, hw_prio, fpriv,
+						&num_scheds, &scheds);
 		if (r)
-			goto failed;
+			goto cleanup_entity;
 	}
 
-	r = amdgpu_queue_mgr_init(adev, &ctx->queue_mgr);
+	/* disable load balance if the hw engine retains context among dependent jobs */
+	if (hw_ip == AMDGPU_HW_IP_VCN_ENC ||
+	    hw_ip == AMDGPU_HW_IP_VCN_DEC ||
+	    hw_ip == AMDGPU_HW_IP_UVD_ENC ||
+	    hw_ip == AMDGPU_HW_IP_UVD) {
+		sched = drm_sched_pick_best(scheds, num_scheds);
+		scheds = &sched;
+		num_scheds = 1;
+	}
+
+	r = drm_sched_entity_init(&entity->entity, drm_prio, scheds, num_scheds,
+				  &ctx->guilty);
 	if (r)
-		goto failed;
+		goto error_free_entity;
+
+	/* It's not an error if we fail to install the new entity */
+	if (cmpxchg(&ctx->entities[hw_ip][ring], NULL, entity))
+		goto cleanup_entity;
 
 	return 0;
 
-failed:
-	for (j = 0; j < i; j++)
-		amd_sched_entity_fini(&adev->rings[j]->sched,
-				      &ctx->rings[j].entity);
-	kfree(ctx->fences);
-	ctx->fences = NULL;
+cleanup_entity:
+	drm_sched_entity_fini(&entity->entity);
+
+error_free_entity:
+	kfree(entity);
+
 	return r;
 }
 
-static void amdgpu_ctx_fini(struct amdgpu_ctx *ctx)
+static ktime_t amdgpu_ctx_fini_entity(struct amdgpu_device *adev,
+				  struct amdgpu_ctx_entity *entity)
 {
-	struct amdgpu_device *adev = ctx->adev;
-	unsigned i, j;
+	ktime_t res = ns_to_ktime(0);
+	int i;
+
+	if (!entity)
+		return res;
+
+	for (i = 0; i < amdgpu_sched_jobs; ++i) {
+		res = ktime_add(res, amdgpu_ctx_fence_time(entity->fences[i]));
+		dma_fence_put(entity->fences[i]);
+	}
+
+	amdgpu_xcp_release_sched(adev, entity);
+
+	kfree(entity);
+	return res;
+}
+
+static int amdgpu_ctx_get_stable_pstate(struct amdgpu_ctx *ctx,
+					u32 *stable_pstate)
+{
+	struct amdgpu_device *adev = ctx->mgr->adev;
+	enum amd_dpm_forced_level current_level;
+
+	current_level = amdgpu_dpm_get_performance_level(adev);
+
+	switch (current_level) {
+	case AMD_DPM_FORCED_LEVEL_PROFILE_STANDARD:
+		*stable_pstate = AMDGPU_CTX_STABLE_PSTATE_STANDARD;
+		break;
+	case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK:
+		*stable_pstate = AMDGPU_CTX_STABLE_PSTATE_MIN_SCLK;
+		break;
+	case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK:
+		*stable_pstate = AMDGPU_CTX_STABLE_PSTATE_MIN_MCLK;
+		break;
+	case AMD_DPM_FORCED_LEVEL_PROFILE_PEAK:
+		*stable_pstate = AMDGPU_CTX_STABLE_PSTATE_PEAK;
+		break;
+	default:
+		*stable_pstate = AMDGPU_CTX_STABLE_PSTATE_NONE;
+		break;
+	}
+	return 0;
+}
+
+static int amdgpu_ctx_init(struct amdgpu_ctx_mgr *mgr, int32_t priority,
+			   struct drm_file *filp, struct amdgpu_ctx *ctx)
+{
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+	u32 current_stable_pstate;
+	int r;
+
+	r = amdgpu_ctx_priority_permit(filp, priority);
+	if (r)
+		return r;
+
+	memset(ctx, 0, sizeof(*ctx));
+
+	kref_init(&ctx->refcount);
+	ctx->mgr = mgr;
+	spin_lock_init(&ctx->ring_lock);
+
+	ctx->reset_counter = atomic_read(&mgr->adev->gpu_reset_counter);
+	ctx->reset_counter_query = ctx->reset_counter;
+	ctx->generation = amdgpu_vm_generation(mgr->adev, &fpriv->vm);
+	ctx->init_priority = priority;
+	ctx->override_priority = AMDGPU_CTX_PRIORITY_UNSET;
+
+	r = amdgpu_ctx_get_stable_pstate(ctx, &current_stable_pstate);
+	if (r)
+		return r;
+
+	if (mgr->adev->pm.stable_pstate_ctx)
+		ctx->stable_pstate = mgr->adev->pm.stable_pstate_ctx->stable_pstate;
+	else
+		ctx->stable_pstate = current_stable_pstate;
+
+	ctx->ctx_mgr = &(fpriv->ctx_mgr);
+	return 0;
+}
+
+static int amdgpu_ctx_set_stable_pstate(struct amdgpu_ctx *ctx,
+					u32 stable_pstate)
+{
+	struct amdgpu_device *adev = ctx->mgr->adev;
+	enum amd_dpm_forced_level level;
+	u32 current_stable_pstate;
+	int r;
+
+	mutex_lock(&adev->pm.stable_pstate_ctx_lock);
+	if (adev->pm.stable_pstate_ctx && adev->pm.stable_pstate_ctx != ctx) {
+		r = -EBUSY;
+		goto done;
+	}
+
+	r = amdgpu_ctx_get_stable_pstate(ctx, &current_stable_pstate);
+	if (r || (stable_pstate == current_stable_pstate))
+		goto done;
+
+	switch (stable_pstate) {
+	case AMDGPU_CTX_STABLE_PSTATE_NONE:
+		level = AMD_DPM_FORCED_LEVEL_AUTO;
+		break;
+	case AMDGPU_CTX_STABLE_PSTATE_STANDARD:
+		level = AMD_DPM_FORCED_LEVEL_PROFILE_STANDARD;
+		break;
+	case AMDGPU_CTX_STABLE_PSTATE_MIN_SCLK:
+		level = AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK;
+		break;
+	case AMDGPU_CTX_STABLE_PSTATE_MIN_MCLK:
+		level = AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK;
+		break;
+	case AMDGPU_CTX_STABLE_PSTATE_PEAK:
+		level = AMD_DPM_FORCED_LEVEL_PROFILE_PEAK;
+		break;
+	default:
+		r = -EINVAL;
+		goto done;
+	}
+
+	r = amdgpu_dpm_force_performance_level(adev, level);
+
+	if (level == AMD_DPM_FORCED_LEVEL_AUTO)
+		adev->pm.stable_pstate_ctx = NULL;
+	else
+		adev->pm.stable_pstate_ctx = ctx;
+done:
+	mutex_unlock(&adev->pm.stable_pstate_ctx_lock);
+
+	return r;
+}
+
+static void amdgpu_ctx_fini(struct kref *ref)
+{
+	struct amdgpu_ctx *ctx = container_of(ref, struct amdgpu_ctx, refcount);
+	struct amdgpu_ctx_mgr *mgr = ctx->mgr;
+	struct amdgpu_device *adev = mgr->adev;
+	unsigned i, j, idx;
 
 	if (!adev)
 		return;
 
-	for (i = 0; i < AMDGPU_MAX_RINGS; ++i)
-		for (j = 0; j < amdgpu_sched_jobs; ++j)
-			dma_fence_put(ctx->rings[i].fences[j]);
-	kfree(ctx->fences);
-	ctx->fences = NULL;
+	for (i = 0; i < AMDGPU_HW_IP_NUM; ++i) {
+		for (j = 0; j < AMDGPU_MAX_ENTITY_NUM; ++j) {
+			ktime_t spend;
 
-	for (i = 0; i < adev->num_rings; i++)
-		amd_sched_entity_fini(&adev->rings[i]->sched,
-				      &ctx->rings[i].entity);
+			spend = amdgpu_ctx_fini_entity(adev, ctx->entities[i][j]);
+			atomic64_add(ktime_to_ns(spend), &mgr->time_spend[i]);
+		}
+	}
 
-	amdgpu_queue_mgr_fini(adev, &ctx->queue_mgr);
+	if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+		amdgpu_ctx_set_stable_pstate(ctx, ctx->stable_pstate);
+		drm_dev_exit(idx);
+	}
+
+	kfree(ctx);
+}
+
+int amdgpu_ctx_get_entity(struct amdgpu_ctx *ctx, u32 hw_ip, u32 instance,
+			  u32 ring, struct drm_sched_entity **entity)
+{
+	int r;
+	struct drm_sched_entity *ctx_entity;
+
+	if (hw_ip >= AMDGPU_HW_IP_NUM) {
+		DRM_ERROR("unknown HW IP type: %d\n", hw_ip);
+		return -EINVAL;
+	}
+
+	/* Right now all IPs have only one instance - multiple rings. */
+	if (instance != 0) {
+		DRM_DEBUG("invalid ip instance: %d\n", instance);
+		return -EINVAL;
+	}
+
+	if (ring >= amdgpu_ctx_num_entities[hw_ip]) {
+		DRM_DEBUG("invalid ring: %d %d\n", hw_ip, ring);
+		return -EINVAL;
+	}
+
+	if (ctx->entities[hw_ip][ring] == NULL) {
+		r = amdgpu_ctx_init_entity(ctx, hw_ip, ring);
+		if (r)
+			return r;
+	}
+
+	ctx_entity = &ctx->entities[hw_ip][ring]->entity;
+	r = drm_sched_entity_error(ctx_entity);
+	if (r) {
+		DRM_DEBUG("error entity %p\n", ctx_entity);
+		return r;
+	}
+
+	*entity = ctx_entity;
+	return 0;
 }
 
 static int amdgpu_ctx_alloc(struct amdgpu_device *adev,
 			    struct amdgpu_fpriv *fpriv,
+			    struct drm_file *filp,
+			    int32_t priority,
 			    uint32_t *id)
 {
 	struct amdgpu_ctx_mgr *mgr = &fpriv->ctx_mgr;
@@ -111,14 +485,15 @@ static int amdgpu_ctx_alloc(struct amdgpu_device *adev,
 		return -ENOMEM;
 
 	mutex_lock(&mgr->lock);
-	r = idr_alloc(&mgr->ctx_handles, ctx, 1, 0, GFP_KERNEL);
+	r = idr_alloc(&mgr->ctx_handles, ctx, 1, AMDGPU_VM_MAX_NUM_CTX, GFP_KERNEL);
 	if (r < 0) {
 		mutex_unlock(&mgr->lock);
 		kfree(ctx);
 		return r;
 	}
+
 	*id = (uint32_t)r;
-	r = amdgpu_ctx_init(adev, ctx);
+	r = amdgpu_ctx_init(mgr, priority, filp, ctx);
 	if (r) {
 		idr_remove(&mgr->ctx_handles, *id);
 		*id = 0;
@@ -131,12 +506,19 @@ static int amdgpu_ctx_alloc(struct amdgpu_device *adev,
 static void amdgpu_ctx_do_release(struct kref *ref)
 {
 	struct amdgpu_ctx *ctx;
+	u32 i, j;
 
 	ctx = container_of(ref, struct amdgpu_ctx, refcount);
+	for (i = 0; i < AMDGPU_HW_IP_NUM; ++i) {
+		for (j = 0; j < amdgpu_ctx_num_entities[i]; ++j) {
+			if (!ctx->entities[i][j])
+				continue;
 
-	amdgpu_ctx_fini(ctx);
+			drm_sched_entity_destroy(&ctx->entities[i][j]->entity);
+		}
+	}
 
-	kfree(ctx);
+	amdgpu_ctx_fini(ref);
 }
 
 static int amdgpu_ctx_free(struct amdgpu_fpriv *fpriv, uint32_t id)
@@ -178,40 +560,166 @@ static int amdgpu_ctx_query(struct amdgpu_device *adev,
 	/* determine if a GPU reset has occured since the last call */
 	reset_counter = atomic_read(&adev->gpu_reset_counter);
 	/* TODO: this should ideally return NO, GUILTY, or INNOCENT. */
-	if (ctx->reset_counter == reset_counter)
+	if (ctx->reset_counter_query == reset_counter)
 		out->state.reset_status = AMDGPU_CTX_NO_RESET;
 	else
 		out->state.reset_status = AMDGPU_CTX_UNKNOWN_RESET;
-	ctx->reset_counter = reset_counter;
+	ctx->reset_counter_query = reset_counter;
+
+	mutex_unlock(&mgr->lock);
+	return 0;
+}
+
+#define AMDGPU_RAS_COUNTE_DELAY_MS 3000
+
+static int amdgpu_ctx_query2(struct amdgpu_device *adev,
+			     struct amdgpu_fpriv *fpriv, uint32_t id,
+			     union drm_amdgpu_ctx_out *out)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct amdgpu_ctx *ctx;
+	struct amdgpu_ctx_mgr *mgr;
+
+	if (!fpriv)
+		return -EINVAL;
+
+	mgr = &fpriv->ctx_mgr;
+	mutex_lock(&mgr->lock);
+	ctx = idr_find(&mgr->ctx_handles, id);
+	if (!ctx) {
+		mutex_unlock(&mgr->lock);
+		return -EINVAL;
+	}
+
+	out->state.flags = 0x0;
+	out->state.hangs = 0x0;
+
+	if (ctx->reset_counter != atomic_read(&adev->gpu_reset_counter))
+		out->state.flags |= AMDGPU_CTX_QUERY2_FLAGS_RESET;
+
+	if (ctx->generation != amdgpu_vm_generation(adev, &fpriv->vm))
+		out->state.flags |= AMDGPU_CTX_QUERY2_FLAGS_VRAMLOST;
+
+	if (atomic_read(&ctx->guilty))
+		out->state.flags |= AMDGPU_CTX_QUERY2_FLAGS_GUILTY;
+
+	if (amdgpu_in_reset(adev))
+		out->state.flags |= AMDGPU_CTX_QUERY2_FLAGS_RESET_IN_PROGRESS;
+
+	if (adev->ras_enabled && con) {
+		/* Return the cached values in O(1),
+		 * and schedule delayed work to cache
+		 * new vaues.
+		 */
+		int ce_count, ue_count;
+
+		ce_count = atomic_read(&con->ras_ce_count);
+		ue_count = atomic_read(&con->ras_ue_count);
+
+		if (ce_count != ctx->ras_counter_ce) {
+			ctx->ras_counter_ce = ce_count;
+			out->state.flags |= AMDGPU_CTX_QUERY2_FLAGS_RAS_CE;
+		}
+
+		if (ue_count != ctx->ras_counter_ue) {
+			ctx->ras_counter_ue = ue_count;
+			out->state.flags |= AMDGPU_CTX_QUERY2_FLAGS_RAS_UE;
+		}
+
+		schedule_delayed_work(&con->ras_counte_delay_work,
+				      msecs_to_jiffies(AMDGPU_RAS_COUNTE_DELAY_MS));
+	}
 
 	mutex_unlock(&mgr->lock);
 	return 0;
 }
 
+static int amdgpu_ctx_stable_pstate(struct amdgpu_device *adev,
+				    struct amdgpu_fpriv *fpriv, uint32_t id,
+				    bool set, u32 *stable_pstate)
+{
+	struct amdgpu_ctx *ctx;
+	struct amdgpu_ctx_mgr *mgr;
+	int r;
+
+	if (!fpriv)
+		return -EINVAL;
+
+	mgr = &fpriv->ctx_mgr;
+	mutex_lock(&mgr->lock);
+	ctx = idr_find(&mgr->ctx_handles, id);
+	if (!ctx) {
+		mutex_unlock(&mgr->lock);
+		return -EINVAL;
+	}
+
+	if (set)
+		r = amdgpu_ctx_set_stable_pstate(ctx, *stable_pstate);
+	else
+		r = amdgpu_ctx_get_stable_pstate(ctx, stable_pstate);
+
+	mutex_unlock(&mgr->lock);
+	return r;
+}
+
 int amdgpu_ctx_ioctl(struct drm_device *dev, void *data,
 		     struct drm_file *filp)
 {
 	int r;
-	uint32_t id;
+	uint32_t id, stable_pstate;
+	int32_t priority;
 
 	union drm_amdgpu_ctx *args = data;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_fpriv *fpriv = filp->driver_priv;
 
-	r = 0;
 	id = args->in.ctx_id;
+	priority = args->in.priority;
+
+	/* For backwards compatibility, we need to accept ioctls with garbage
+	 * in the priority field. Garbage values in the priority field, result
+	 * in the priority being set to NORMAL.
+	 */
+	if (!amdgpu_ctx_priority_is_valid(priority))
+		priority = AMDGPU_CTX_PRIORITY_NORMAL;
 
 	switch (args->in.op) {
 	case AMDGPU_CTX_OP_ALLOC_CTX:
-		r = amdgpu_ctx_alloc(adev, fpriv, &id);
+		if (args->in.flags)
+			return -EINVAL;
+		r = amdgpu_ctx_alloc(adev, fpriv, filp, priority, &id);
 		args->out.alloc.ctx_id = id;
 		break;
 	case AMDGPU_CTX_OP_FREE_CTX:
+		if (args->in.flags)
+			return -EINVAL;
 		r = amdgpu_ctx_free(fpriv, id);
 		break;
 	case AMDGPU_CTX_OP_QUERY_STATE:
+		if (args->in.flags)
+			return -EINVAL;
 		r = amdgpu_ctx_query(adev, fpriv, id, &args->out);
 		break;
+	case AMDGPU_CTX_OP_QUERY_STATE2:
+		if (args->in.flags)
+			return -EINVAL;
+		r = amdgpu_ctx_query2(adev, fpriv, id, &args->out);
+		break;
+	case AMDGPU_CTX_OP_GET_STABLE_PSTATE:
+		if (args->in.flags)
+			return -EINVAL;
+		r = amdgpu_ctx_stable_pstate(adev, fpriv, id, false, &stable_pstate);
+		if (!r)
+			args->out.pstate.flags = stable_pstate;
+		break;
+	case AMDGPU_CTX_OP_SET_STABLE_PSTATE:
+		if (args->in.flags & ~AMDGPU_CTX_STABLE_PSTATE_FLAGS_MASK)
+			return -EINVAL;
+		stable_pstate = args->in.flags & AMDGPU_CTX_STABLE_PSTATE_FLAGS_MASK;
+		if (stable_pstate > AMDGPU_CTX_STABLE_PSTATE_PEAK)
+			return -EINVAL;
+		r = amdgpu_ctx_stable_pstate(adev, fpriv, id, true, &stable_pstate);
+		break;
 	default:
 		return -EINVAL;
 	}
@@ -246,82 +754,239 @@ int amdgpu_ctx_put(struct amdgpu_ctx *ctx)
 	return 0;
 }
 
-uint64_t amdgpu_ctx_add_fence(struct amdgpu_ctx *ctx, struct amdgpu_ring *ring,
+uint64_t amdgpu_ctx_add_fence(struct amdgpu_ctx *ctx,
+			      struct drm_sched_entity *entity,
 			      struct dma_fence *fence)
 {
-	struct amdgpu_ctx_ring *cring = & ctx->rings[ring->idx];
-	uint64_t seq = cring->sequence;
-	unsigned idx = 0;
+	struct amdgpu_ctx_entity *centity = to_amdgpu_ctx_entity(entity);
+	uint64_t seq = centity->sequence;
 	struct dma_fence *other = NULL;
+	unsigned idx = 0;
 
 	idx = seq & (amdgpu_sched_jobs - 1);
-	other = cring->fences[idx];
-	if (other) {
-		signed long r;
-		r = dma_fence_wait_timeout(other, false, MAX_SCHEDULE_TIMEOUT);
-		if (r < 0)
-			DRM_ERROR("Error (%ld) waiting for fence!\n", r);
-	}
+	other = centity->fences[idx];
+	WARN_ON(other && !dma_fence_is_signaled(other));
 
 	dma_fence_get(fence);
 
 	spin_lock(&ctx->ring_lock);
-	cring->fences[idx] = fence;
-	cring->sequence++;
+	centity->fences[idx] = fence;
+	centity->sequence++;
 	spin_unlock(&ctx->ring_lock);
 
-	dma_fence_put(other);
+	atomic64_add(ktime_to_ns(amdgpu_ctx_fence_time(other)),
+		     &ctx->mgr->time_spend[centity->hw_ip]);
 
+	dma_fence_put(other);
 	return seq;
 }
 
 struct dma_fence *amdgpu_ctx_get_fence(struct amdgpu_ctx *ctx,
-				       struct amdgpu_ring *ring, uint64_t seq)
+				       struct drm_sched_entity *entity,
+				       uint64_t seq)
 {
-	struct amdgpu_ctx_ring *cring = & ctx->rings[ring->idx];
+	struct amdgpu_ctx_entity *centity = to_amdgpu_ctx_entity(entity);
 	struct dma_fence *fence;
 
 	spin_lock(&ctx->ring_lock);
 
 	if (seq == ~0ull)
-		seq = ctx->rings[ring->idx].sequence - 1;
+		seq = centity->sequence - 1;
 
-	if (seq >= cring->sequence) {
+	if (seq >= centity->sequence) {
 		spin_unlock(&ctx->ring_lock);
 		return ERR_PTR(-EINVAL);
 	}
 
 
-	if (seq + amdgpu_sched_jobs < cring->sequence) {
+	if (seq + amdgpu_sched_jobs < centity->sequence) {
 		spin_unlock(&ctx->ring_lock);
 		return NULL;
 	}
 
-	fence = dma_fence_get(cring->fences[seq & (amdgpu_sched_jobs - 1)]);
+	fence = dma_fence_get(centity->fences[seq & (amdgpu_sched_jobs - 1)]);
 	spin_unlock(&ctx->ring_lock);
 
 	return fence;
 }
 
-void amdgpu_ctx_mgr_init(struct amdgpu_ctx_mgr *mgr)
+static void amdgpu_ctx_set_entity_priority(struct amdgpu_ctx *ctx,
+					   struct amdgpu_ctx_entity *aentity,
+					   int hw_ip,
+					   int32_t priority)
+{
+	struct amdgpu_device *adev = ctx->mgr->adev;
+	unsigned int hw_prio;
+	struct drm_gpu_scheduler **scheds = NULL;
+	unsigned num_scheds;
+
+	/* set sw priority */
+	drm_sched_entity_set_priority(&aentity->entity,
+				      amdgpu_ctx_to_drm_sched_prio(priority));
+
+	/* set hw priority */
+	if (hw_ip == AMDGPU_HW_IP_COMPUTE || hw_ip == AMDGPU_HW_IP_GFX) {
+		hw_prio = amdgpu_ctx_get_hw_prio(ctx, hw_ip);
+		hw_prio = array_index_nospec(hw_prio, AMDGPU_RING_PRIO_MAX);
+		scheds = adev->gpu_sched[hw_ip][hw_prio].sched;
+		num_scheds = adev->gpu_sched[hw_ip][hw_prio].num_scheds;
+		drm_sched_entity_modify_sched(&aentity->entity, scheds,
+					      num_scheds);
+	}
+}
+
+void amdgpu_ctx_priority_override(struct amdgpu_ctx *ctx,
+				  int32_t priority)
 {
+	int32_t ctx_prio;
+	unsigned i, j;
+
+	ctx->override_priority = priority;
+
+	ctx_prio = (ctx->override_priority == AMDGPU_CTX_PRIORITY_UNSET) ?
+			ctx->init_priority : ctx->override_priority;
+	for (i = 0; i < AMDGPU_HW_IP_NUM; ++i) {
+		for (j = 0; j < amdgpu_ctx_num_entities[i]; ++j) {
+			if (!ctx->entities[i][j])
+				continue;
+
+			amdgpu_ctx_set_entity_priority(ctx, ctx->entities[i][j],
+						       i, ctx_prio);
+		}
+	}
+}
+
+int amdgpu_ctx_wait_prev_fence(struct amdgpu_ctx *ctx,
+			       struct drm_sched_entity *entity)
+{
+	struct amdgpu_ctx_entity *centity = to_amdgpu_ctx_entity(entity);
+	struct dma_fence *other;
+	unsigned idx;
+	long r;
+
+	spin_lock(&ctx->ring_lock);
+	idx = centity->sequence & (amdgpu_sched_jobs - 1);
+	other = dma_fence_get(centity->fences[idx]);
+	spin_unlock(&ctx->ring_lock);
+
+	if (!other)
+		return 0;
+
+	r = dma_fence_wait(other, true);
+	if (r < 0 && r != -ERESTARTSYS)
+		DRM_ERROR("Error (%ld) waiting for fence!\n", r);
+
+	dma_fence_put(other);
+	return r;
+}
+
+void amdgpu_ctx_mgr_init(struct amdgpu_ctx_mgr *mgr,
+			 struct amdgpu_device *adev)
+{
+	unsigned int i;
+
+	mgr->adev = adev;
 	mutex_init(&mgr->lock);
-	idr_init(&mgr->ctx_handles);
+	idr_init_base(&mgr->ctx_handles, 1);
+
+	for (i = 0; i < AMDGPU_HW_IP_NUM; ++i)
+		atomic64_set(&mgr->time_spend[i], 0);
 }
 
-void amdgpu_ctx_mgr_fini(struct amdgpu_ctx_mgr *mgr)
+long amdgpu_ctx_mgr_entity_flush(struct amdgpu_ctx_mgr *mgr, long timeout)
 {
 	struct amdgpu_ctx *ctx;
 	struct idr *idp;
-	uint32_t id;
+	uint32_t id, i, j;
+
+	idp = &mgr->ctx_handles;
+
+	mutex_lock(&mgr->lock);
+	idr_for_each_entry(idp, ctx, id) {
+		for (i = 0; i < AMDGPU_HW_IP_NUM; ++i) {
+			for (j = 0; j < amdgpu_ctx_num_entities[i]; ++j) {
+				struct drm_sched_entity *entity;
+
+				if (!ctx->entities[i][j])
+					continue;
+
+				entity = &ctx->entities[i][j]->entity;
+				timeout = drm_sched_entity_flush(entity, timeout);
+			}
+		}
+	}
+	mutex_unlock(&mgr->lock);
+	return timeout;
+}
+
+static void amdgpu_ctx_mgr_entity_fini(struct amdgpu_ctx_mgr *mgr)
+{
+	struct amdgpu_ctx *ctx;
+	struct idr *idp;
+	uint32_t id, i, j;
 
 	idp = &mgr->ctx_handles;
 
 	idr_for_each_entry(idp, ctx, id) {
-		if (kref_put(&ctx->refcount, amdgpu_ctx_do_release) != 1)
+		if (kref_read(&ctx->refcount) != 1) {
 			DRM_ERROR("ctx %p is still alive\n", ctx);
+			continue;
+		}
+
+		for (i = 0; i < AMDGPU_HW_IP_NUM; ++i) {
+			for (j = 0; j < amdgpu_ctx_num_entities[i]; ++j) {
+				struct drm_sched_entity *entity;
+
+				if (!ctx->entities[i][j])
+					continue;
+
+				entity = &ctx->entities[i][j]->entity;
+				drm_sched_entity_fini(entity);
+			}
+		}
+		kref_put(&ctx->refcount, amdgpu_ctx_fini);
 	}
+}
 
+void amdgpu_ctx_mgr_fini(struct amdgpu_ctx_mgr *mgr)
+{
+	amdgpu_ctx_mgr_entity_fini(mgr);
 	idr_destroy(&mgr->ctx_handles);
 	mutex_destroy(&mgr->lock);
 }
+
+void amdgpu_ctx_mgr_usage(struct amdgpu_ctx_mgr *mgr,
+			  ktime_t usage[AMDGPU_HW_IP_NUM])
+{
+	struct amdgpu_ctx *ctx;
+	unsigned int hw_ip, i;
+	uint32_t id;
+
+	/*
+	 * This is a little bit racy because it can be that a ctx or a fence are
+	 * destroyed just in the moment we try to account them. But that is ok
+	 * since exactly that case is explicitely allowed by the interface.
+	 */
+	mutex_lock(&mgr->lock);
+	for (hw_ip = 0; hw_ip < AMDGPU_HW_IP_NUM; ++hw_ip) {
+		uint64_t ns = atomic64_read(&mgr->time_spend[hw_ip]);
+
+		usage[hw_ip] = ns_to_ktime(ns);
+	}
+
+	idr_for_each_entry(&mgr->ctx_handles, ctx, id) {
+		for (hw_ip = 0; hw_ip < AMDGPU_HW_IP_NUM; ++hw_ip) {
+			for (i = 0; i < amdgpu_ctx_num_entities[hw_ip]; ++i) {
+				struct amdgpu_ctx_entity *centity;
+				ktime_t spend;
+
+				centity = ctx->entities[hw_ip][i];
+				if (!centity)
+					continue;
+				spend = amdgpu_ctx_entity_time(ctx, centity);
+				usage[hw_ip] = ktime_add(usage[hw_ip], spend);
+			}
+		}
+	}
+	mutex_unlock(&mgr->lock);
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h
new file mode 100644
index 000000000000..090dfe86f75b
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h
@@ -0,0 +1,100 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __AMDGPU_CTX_H__
+#define __AMDGPU_CTX_H__
+
+#include <linux/ktime.h>
+#include <linux/types.h>
+
+#include "amdgpu_ring.h"
+
+struct drm_device;
+struct drm_file;
+struct amdgpu_fpriv;
+struct amdgpu_ctx_mgr;
+
+#define AMDGPU_MAX_ENTITY_NUM 4
+
+struct amdgpu_ctx_entity {
+	uint32_t		hw_ip;
+	uint64_t		sequence;
+	struct drm_sched_entity	entity;
+	struct dma_fence	*fences[];
+};
+
+struct amdgpu_ctx {
+	struct kref			refcount;
+	struct amdgpu_ctx_mgr		*mgr;
+	unsigned			reset_counter;
+	unsigned			reset_counter_query;
+	uint64_t			generation;
+	spinlock_t			ring_lock;
+	struct amdgpu_ctx_entity	*entities[AMDGPU_HW_IP_NUM][AMDGPU_MAX_ENTITY_NUM];
+	bool				preamble_presented;
+	int32_t				init_priority;
+	int32_t				override_priority;
+	atomic_t			guilty;
+	unsigned long			ras_counter_ce;
+	unsigned long			ras_counter_ue;
+	uint32_t			stable_pstate;
+	struct amdgpu_ctx_mgr		*ctx_mgr;
+};
+
+struct amdgpu_ctx_mgr {
+	struct amdgpu_device	*adev;
+	struct mutex		lock;
+	/* protected by lock */
+	struct idr		ctx_handles;
+	atomic64_t		time_spend[AMDGPU_HW_IP_NUM];
+};
+
+extern const unsigned int amdgpu_ctx_num_entities[AMDGPU_HW_IP_NUM];
+
+struct amdgpu_ctx *amdgpu_ctx_get(struct amdgpu_fpriv *fpriv, uint32_t id);
+int amdgpu_ctx_put(struct amdgpu_ctx *ctx);
+
+int amdgpu_ctx_get_entity(struct amdgpu_ctx *ctx, u32 hw_ip, u32 instance,
+			  u32 ring, struct drm_sched_entity **entity);
+uint64_t amdgpu_ctx_add_fence(struct amdgpu_ctx *ctx,
+			      struct drm_sched_entity *entity,
+			      struct dma_fence *fence);
+struct dma_fence *amdgpu_ctx_get_fence(struct amdgpu_ctx *ctx,
+				       struct drm_sched_entity *entity,
+				       uint64_t seq);
+bool amdgpu_ctx_priority_is_valid(int32_t ctx_prio);
+void amdgpu_ctx_priority_override(struct amdgpu_ctx *ctx, int32_t ctx_prio);
+
+int amdgpu_ctx_ioctl(struct drm_device *dev, void *data,
+		     struct drm_file *filp);
+
+int amdgpu_ctx_wait_prev_fence(struct amdgpu_ctx *ctx,
+			       struct drm_sched_entity *entity);
+
+void amdgpu_ctx_mgr_init(struct amdgpu_ctx_mgr *mgr,
+			 struct amdgpu_device *adev);
+long amdgpu_ctx_mgr_entity_flush(struct amdgpu_ctx_mgr *mgr, long timeout);
+void amdgpu_ctx_mgr_fini(struct amdgpu_ctx_mgr *mgr);
+void amdgpu_ctx_mgr_usage(struct amdgpu_ctx_mgr *mgr,
+			  ktime_t usage[AMDGPU_HW_IP_NUM]);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c
new file mode 100644
index 000000000000..a70651050acf
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c
@@ -0,0 +1,2201 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/kthread.h>
+#include <linux/pci.h>
+#include <linux/uaccess.h>
+#include <linux/pm_runtime.h>
+
+#include "amdgpu.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_dm_debugfs.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_rap.h"
+#include "amdgpu_securedisplay.h"
+#include "amdgpu_fw_attestation.h"
+#include "amdgpu_umr.h"
+
+#include "amdgpu_reset.h"
+#include "amdgpu_psp_ta.h"
+
+#if defined(CONFIG_DEBUG_FS)
+
+/**
+ * amdgpu_debugfs_process_reg_op - Handle MMIO register reads/writes
+ *
+ * @read: True if reading
+ * @f: open file handle
+ * @buf: User buffer to write/read to
+ * @size: Number of bytes to write/read
+ * @pos:  Offset to seek to
+ *
+ * This debugfs entry has special meaning on the offset being sought.
+ * Various bits have different meanings:
+ *
+ * Bit 62:  Indicates a GRBM bank switch is needed
+ * Bit 61:  Indicates a SRBM bank switch is needed (implies bit 62 is
+ *	    zero)
+ * Bits 24..33: The SE or ME selector if needed
+ * Bits 34..43: The SH (or SA) or PIPE selector if needed
+ * Bits 44..53: The INSTANCE (or CU/WGP) or QUEUE selector if needed
+ *
+ * Bit 23:  Indicates that the PM power gating lock should be held
+ *	    This is necessary to read registers that might be
+ *	    unreliable during a power gating transistion.
+ *
+ * The lower bits are the BYTE offset of the register to read.  This
+ * allows reading multiple registers in a single call and having
+ * the returned size reflect that.
+ */
+static int  amdgpu_debugfs_process_reg_op(bool read, struct file *f,
+		char __user *buf, size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+	bool pm_pg_lock, use_bank, use_ring;
+	unsigned int instance_bank, sh_bank, se_bank, me, pipe, queue, vmid;
+
+	pm_pg_lock = use_bank = use_ring = false;
+	instance_bank = sh_bank = se_bank = me = pipe = queue = vmid = 0;
+
+	if (size & 0x3 || *pos & 0x3 ||
+			((*pos & (1ULL << 62)) && (*pos & (1ULL << 61))))
+		return -EINVAL;
+
+	/* are we reading registers for which a PG lock is necessary? */
+	pm_pg_lock = (*pos >> 23) & 1;
+
+	if (*pos & (1ULL << 62)) {
+		se_bank = (*pos & GENMASK_ULL(33, 24)) >> 24;
+		sh_bank = (*pos & GENMASK_ULL(43, 34)) >> 34;
+		instance_bank = (*pos & GENMASK_ULL(53, 44)) >> 44;
+
+		if (se_bank == 0x3FF)
+			se_bank = 0xFFFFFFFF;
+		if (sh_bank == 0x3FF)
+			sh_bank = 0xFFFFFFFF;
+		if (instance_bank == 0x3FF)
+			instance_bank = 0xFFFFFFFF;
+		use_bank = true;
+	} else if (*pos & (1ULL << 61)) {
+
+		me = (*pos & GENMASK_ULL(33, 24)) >> 24;
+		pipe = (*pos & GENMASK_ULL(43, 34)) >> 34;
+		queue = (*pos & GENMASK_ULL(53, 44)) >> 44;
+		vmid = (*pos & GENMASK_ULL(58, 54)) >> 54;
+
+		use_ring = true;
+	} else {
+		use_bank = use_ring = false;
+	}
+
+	*pos &= (1UL << 22) - 1;
+
+	r = pm_runtime_get_sync(adev_to_drm(adev)->dev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	r = amdgpu_virt_enable_access_debugfs(adev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	if (use_bank) {
+		if ((sh_bank != 0xFFFFFFFF && sh_bank >= adev->gfx.config.max_sh_per_se) ||
+		    (se_bank != 0xFFFFFFFF && se_bank >= adev->gfx.config.max_shader_engines)) {
+			pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+			pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+			amdgpu_virt_disable_access_debugfs(adev);
+			return -EINVAL;
+		}
+		mutex_lock(&adev->grbm_idx_mutex);
+		amdgpu_gfx_select_se_sh(adev, se_bank,
+					sh_bank, instance_bank, 0);
+	} else if (use_ring) {
+		mutex_lock(&adev->srbm_mutex);
+		amdgpu_gfx_select_me_pipe_q(adev, me, pipe, queue, vmid, 0);
+	}
+
+	if (pm_pg_lock)
+		mutex_lock(&adev->pm.mutex);
+
+	while (size) {
+		uint32_t value;
+
+		if (read) {
+			value = RREG32(*pos >> 2);
+			r = put_user(value, (uint32_t *)buf);
+		} else {
+			r = get_user(value, (uint32_t *)buf);
+			if (!r)
+				amdgpu_mm_wreg_mmio_rlc(adev, *pos >> 2, value, 0);
+		}
+		if (r) {
+			result = r;
+			goto end;
+		}
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+end:
+	if (use_bank) {
+		amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
+		mutex_unlock(&adev->grbm_idx_mutex);
+	} else if (use_ring) {
+		amdgpu_gfx_select_me_pipe_q(adev, 0, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+	}
+
+	if (pm_pg_lock)
+		mutex_unlock(&adev->pm.mutex);
+
+	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+
+	amdgpu_virt_disable_access_debugfs(adev);
+	return result;
+}
+
+/*
+ * amdgpu_debugfs_regs_read - Callback for reading MMIO registers
+ */
+static ssize_t amdgpu_debugfs_regs_read(struct file *f, char __user *buf,
+					size_t size, loff_t *pos)
+{
+	return amdgpu_debugfs_process_reg_op(true, f, buf, size, pos);
+}
+
+/*
+ * amdgpu_debugfs_regs_write - Callback for writing MMIO registers
+ */
+static ssize_t amdgpu_debugfs_regs_write(struct file *f, const char __user *buf,
+					 size_t size, loff_t *pos)
+{
+	return amdgpu_debugfs_process_reg_op(false, f, (char __user *)buf, size, pos);
+}
+
+static int amdgpu_debugfs_regs2_open(struct inode *inode, struct file *file)
+{
+	struct amdgpu_debugfs_regs2_data *rd;
+
+	rd = kzalloc(sizeof(*rd), GFP_KERNEL);
+	if (!rd)
+		return -ENOMEM;
+	rd->adev = file_inode(file)->i_private;
+	file->private_data = rd;
+	mutex_init(&rd->lock);
+
+	return 0;
+}
+
+static int amdgpu_debugfs_regs2_release(struct inode *inode, struct file *file)
+{
+	struct amdgpu_debugfs_regs2_data *rd = file->private_data;
+
+	mutex_destroy(&rd->lock);
+	kfree(file->private_data);
+	return 0;
+}
+
+static ssize_t amdgpu_debugfs_regs2_op(struct file *f, char __user *buf, u32 offset, size_t size, int write_en)
+{
+	struct amdgpu_debugfs_regs2_data *rd = f->private_data;
+	struct amdgpu_device *adev = rd->adev;
+	ssize_t result = 0;
+	int r;
+	uint32_t value;
+
+	if (size & 0x3 || offset & 0x3)
+		return -EINVAL;
+
+	r = pm_runtime_get_sync(adev_to_drm(adev)->dev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	r = amdgpu_virt_enable_access_debugfs(adev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	mutex_lock(&rd->lock);
+
+	if (rd->id.use_grbm) {
+		if ((rd->id.grbm.sh != 0xFFFFFFFF && rd->id.grbm.sh >= adev->gfx.config.max_sh_per_se) ||
+		    (rd->id.grbm.se != 0xFFFFFFFF && rd->id.grbm.se >= adev->gfx.config.max_shader_engines)) {
+			pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+			pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+			amdgpu_virt_disable_access_debugfs(adev);
+			mutex_unlock(&rd->lock);
+			return -EINVAL;
+		}
+		mutex_lock(&adev->grbm_idx_mutex);
+		amdgpu_gfx_select_se_sh(adev, rd->id.grbm.se,
+						  rd->id.grbm.sh,
+						  rd->id.grbm.instance, rd->id.xcc_id);
+	}
+
+	if (rd->id.use_srbm) {
+		mutex_lock(&adev->srbm_mutex);
+		amdgpu_gfx_select_me_pipe_q(adev, rd->id.srbm.me, rd->id.srbm.pipe,
+					    rd->id.srbm.queue, rd->id.srbm.vmid, rd->id.xcc_id);
+	}
+
+	if (rd->id.pg_lock)
+		mutex_lock(&adev->pm.mutex);
+
+	while (size) {
+		if (!write_en) {
+			value = RREG32(offset >> 2);
+			r = put_user(value, (uint32_t *)buf);
+		} else {
+			r = get_user(value, (uint32_t *)buf);
+			if (!r)
+				amdgpu_mm_wreg_mmio_rlc(adev, offset >> 2, value, rd->id.xcc_id);
+		}
+		if (r) {
+			result = r;
+			goto end;
+		}
+		offset += 4;
+		size -= 4;
+		result += 4;
+		buf += 4;
+	}
+end:
+	if (rd->id.use_grbm) {
+		amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, rd->id.xcc_id);
+		mutex_unlock(&adev->grbm_idx_mutex);
+	}
+
+	if (rd->id.use_srbm) {
+		amdgpu_gfx_select_me_pipe_q(adev, 0, 0, 0, 0, rd->id.xcc_id);
+		mutex_unlock(&adev->srbm_mutex);
+	}
+
+	if (rd->id.pg_lock)
+		mutex_unlock(&adev->pm.mutex);
+
+	mutex_unlock(&rd->lock);
+
+	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+
+	amdgpu_virt_disable_access_debugfs(adev);
+	return result;
+}
+
+static long amdgpu_debugfs_regs2_ioctl(struct file *f, unsigned int cmd, unsigned long data)
+{
+	struct amdgpu_debugfs_regs2_data *rd = f->private_data;
+	struct amdgpu_debugfs_regs2_iocdata v1_data;
+	int r;
+
+	mutex_lock(&rd->lock);
+
+	switch (cmd) {
+	case AMDGPU_DEBUGFS_REGS2_IOC_SET_STATE_V2:
+		r = copy_from_user(&rd->id, (struct amdgpu_debugfs_regs2_iocdata_v2 *)data,
+				   sizeof(rd->id));
+		if (r)
+			r = -EINVAL;
+		goto done;
+	case AMDGPU_DEBUGFS_REGS2_IOC_SET_STATE:
+		r = copy_from_user(&v1_data, (struct amdgpu_debugfs_regs2_iocdata *)data,
+				   sizeof(v1_data));
+		if (r) {
+			r = -EINVAL;
+			goto done;
+		}
+		goto v1_copy;
+	default:
+		r = -EINVAL;
+		goto done;
+	}
+
+v1_copy:
+	rd->id.use_srbm = v1_data.use_srbm;
+	rd->id.use_grbm = v1_data.use_grbm;
+	rd->id.pg_lock = v1_data.pg_lock;
+	rd->id.grbm.se = v1_data.grbm.se;
+	rd->id.grbm.sh = v1_data.grbm.sh;
+	rd->id.grbm.instance = v1_data.grbm.instance;
+	rd->id.srbm.me = v1_data.srbm.me;
+	rd->id.srbm.pipe = v1_data.srbm.pipe;
+	rd->id.srbm.queue = v1_data.srbm.queue;
+	rd->id.xcc_id = 0;
+done:
+	mutex_unlock(&rd->lock);
+	return r;
+}
+
+static ssize_t amdgpu_debugfs_regs2_read(struct file *f, char __user *buf, size_t size, loff_t *pos)
+{
+	return amdgpu_debugfs_regs2_op(f, buf, *pos, size, 0);
+}
+
+static ssize_t amdgpu_debugfs_regs2_write(struct file *f, const char __user *buf, size_t size, loff_t *pos)
+{
+	return amdgpu_debugfs_regs2_op(f, (char __user *)buf, *pos, size, 1);
+}
+
+static int amdgpu_debugfs_gprwave_open(struct inode *inode, struct file *file)
+{
+	struct amdgpu_debugfs_gprwave_data *rd;
+
+	rd = kzalloc(sizeof(*rd), GFP_KERNEL);
+	if (!rd)
+		return -ENOMEM;
+	rd->adev = file_inode(file)->i_private;
+	file->private_data = rd;
+	mutex_init(&rd->lock);
+
+	return 0;
+}
+
+static int amdgpu_debugfs_gprwave_release(struct inode *inode, struct file *file)
+{
+	struct amdgpu_debugfs_gprwave_data *rd = file->private_data;
+
+	mutex_destroy(&rd->lock);
+	kfree(file->private_data);
+	return 0;
+}
+
+static ssize_t amdgpu_debugfs_gprwave_read(struct file *f, char __user *buf, size_t size, loff_t *pos)
+{
+	struct amdgpu_debugfs_gprwave_data *rd = f->private_data;
+	struct amdgpu_device *adev = rd->adev;
+	ssize_t result = 0;
+	int r;
+	uint32_t *data, x;
+
+	if (size > 4096 || size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	r = pm_runtime_get_sync(adev_to_drm(adev)->dev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	r = amdgpu_virt_enable_access_debugfs(adev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	data = kcalloc(1024, sizeof(*data), GFP_KERNEL);
+	if (!data) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		amdgpu_virt_disable_access_debugfs(adev);
+		return -ENOMEM;
+	}
+
+	/* switch to the specific se/sh/cu */
+	mutex_lock(&adev->grbm_idx_mutex);
+	amdgpu_gfx_select_se_sh(adev, rd->id.se, rd->id.sh, rd->id.cu, rd->id.xcc_id);
+
+	if (!rd->id.gpr_or_wave) {
+		x = 0;
+		if (adev->gfx.funcs->read_wave_data)
+			adev->gfx.funcs->read_wave_data(adev, rd->id.xcc_id, rd->id.simd, rd->id.wave, data, &x);
+	} else {
+		x = size >> 2;
+		if (rd->id.gpr.vpgr_or_sgpr) {
+			if (adev->gfx.funcs->read_wave_vgprs)
+				adev->gfx.funcs->read_wave_vgprs(adev, rd->id.xcc_id, rd->id.simd, rd->id.wave, rd->id.gpr.thread, *pos, size>>2, data);
+		} else {
+			if (adev->gfx.funcs->read_wave_sgprs)
+				adev->gfx.funcs->read_wave_sgprs(adev, rd->id.xcc_id, rd->id.simd, rd->id.wave, *pos, size>>2, data);
+		}
+	}
+
+	amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, rd->id.xcc_id);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+
+	if (!x) {
+		result = -EINVAL;
+		goto done;
+	}
+
+	while (size && (*pos < x * 4)) {
+		uint32_t value;
+
+		value = data[*pos >> 2];
+		r = put_user(value, (uint32_t *)buf);
+		if (r) {
+			result = r;
+			goto done;
+		}
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+done:
+	amdgpu_virt_disable_access_debugfs(adev);
+	kfree(data);
+	return result;
+}
+
+static long amdgpu_debugfs_gprwave_ioctl(struct file *f, unsigned int cmd, unsigned long data)
+{
+	struct amdgpu_debugfs_gprwave_data *rd = f->private_data;
+	int r = 0;
+
+	mutex_lock(&rd->lock);
+
+	switch (cmd) {
+	case AMDGPU_DEBUGFS_GPRWAVE_IOC_SET_STATE:
+		if (copy_from_user(&rd->id,
+				   (struct amdgpu_debugfs_gprwave_iocdata *)data,
+				   sizeof(rd->id)))
+			r = -EFAULT;
+		goto done;
+	default:
+		r = -EINVAL;
+		goto done;
+	}
+
+done:
+	mutex_unlock(&rd->lock);
+	return r;
+}
+
+
+
+
+/**
+ * amdgpu_debugfs_regs_pcie_read - Read from a PCIE register
+ *
+ * @f: open file handle
+ * @buf: User buffer to store read data in
+ * @size: Number of bytes to read
+ * @pos:  Offset to seek to
+ *
+ * The lower bits are the BYTE offset of the register to read.  This
+ * allows reading multiple registers in a single call and having
+ * the returned size reflect that.
+ */
+static ssize_t amdgpu_debugfs_regs_pcie_read(struct file *f, char __user *buf,
+					size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	r = pm_runtime_get_sync(adev_to_drm(adev)->dev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	r = amdgpu_virt_enable_access_debugfs(adev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	while (size) {
+		uint32_t value;
+
+		if (upper_32_bits(*pos))
+			value = RREG32_PCIE_EXT(*pos);
+		else
+			value = RREG32_PCIE(*pos);
+
+		r = put_user(value, (uint32_t *)buf);
+		if (r)
+			goto out;
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	r = result;
+out:
+	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+	amdgpu_virt_disable_access_debugfs(adev);
+	return r;
+}
+
+/**
+ * amdgpu_debugfs_regs_pcie_write - Write to a PCIE register
+ *
+ * @f: open file handle
+ * @buf: User buffer to write data from
+ * @size: Number of bytes to write
+ * @pos:  Offset to seek to
+ *
+ * The lower bits are the BYTE offset of the register to write.  This
+ * allows writing multiple registers in a single call and having
+ * the returned size reflect that.
+ */
+static ssize_t amdgpu_debugfs_regs_pcie_write(struct file *f, const char __user *buf,
+					 size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	r = pm_runtime_get_sync(adev_to_drm(adev)->dev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	r = amdgpu_virt_enable_access_debugfs(adev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	while (size) {
+		uint32_t value;
+
+		r = get_user(value, (uint32_t *)buf);
+		if (r)
+			goto out;
+
+		if (upper_32_bits(*pos))
+			WREG32_PCIE_EXT(*pos, value);
+		else
+			WREG32_PCIE(*pos, value);
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	r = result;
+out:
+	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+	amdgpu_virt_disable_access_debugfs(adev);
+	return r;
+}
+
+/**
+ * amdgpu_debugfs_regs_didt_read - Read from a DIDT register
+ *
+ * @f: open file handle
+ * @buf: User buffer to store read data in
+ * @size: Number of bytes to read
+ * @pos:  Offset to seek to
+ *
+ * The lower bits are the BYTE offset of the register to read.  This
+ * allows reading multiple registers in a single call and having
+ * the returned size reflect that.
+ */
+static ssize_t amdgpu_debugfs_regs_didt_read(struct file *f, char __user *buf,
+					size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	if (!adev->didt_rreg)
+		return -EOPNOTSUPP;
+
+	r = pm_runtime_get_sync(adev_to_drm(adev)->dev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	r = amdgpu_virt_enable_access_debugfs(adev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	while (size) {
+		uint32_t value;
+
+		value = RREG32_DIDT(*pos >> 2);
+		r = put_user(value, (uint32_t *)buf);
+		if (r)
+			goto out;
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	r = result;
+out:
+	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+	amdgpu_virt_disable_access_debugfs(adev);
+	return r;
+}
+
+/**
+ * amdgpu_debugfs_regs_didt_write - Write to a DIDT register
+ *
+ * @f: open file handle
+ * @buf: User buffer to write data from
+ * @size: Number of bytes to write
+ * @pos:  Offset to seek to
+ *
+ * The lower bits are the BYTE offset of the register to write.  This
+ * allows writing multiple registers in a single call and having
+ * the returned size reflect that.
+ */
+static ssize_t amdgpu_debugfs_regs_didt_write(struct file *f, const char __user *buf,
+					 size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	if (!adev->didt_wreg)
+		return -EOPNOTSUPP;
+
+	r = pm_runtime_get_sync(adev_to_drm(adev)->dev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	r = amdgpu_virt_enable_access_debugfs(adev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	while (size) {
+		uint32_t value;
+
+		r = get_user(value, (uint32_t *)buf);
+		if (r)
+			goto out;
+
+		WREG32_DIDT(*pos >> 2, value);
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	r = result;
+out:
+	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+	amdgpu_virt_disable_access_debugfs(adev);
+	return r;
+}
+
+/**
+ * amdgpu_debugfs_regs_smc_read - Read from a SMC register
+ *
+ * @f: open file handle
+ * @buf: User buffer to store read data in
+ * @size: Number of bytes to read
+ * @pos:  Offset to seek to
+ *
+ * The lower bits are the BYTE offset of the register to read.  This
+ * allows reading multiple registers in a single call and having
+ * the returned size reflect that.
+ */
+static ssize_t amdgpu_debugfs_regs_smc_read(struct file *f, char __user *buf,
+					size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (!adev->smc_rreg)
+		return -EOPNOTSUPP;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	r = pm_runtime_get_sync(adev_to_drm(adev)->dev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	r = amdgpu_virt_enable_access_debugfs(adev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	while (size) {
+		uint32_t value;
+
+		value = RREG32_SMC(*pos);
+		r = put_user(value, (uint32_t *)buf);
+		if (r)
+			goto out;
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	r = result;
+out:
+	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+	amdgpu_virt_disable_access_debugfs(adev);
+	return r;
+}
+
+/**
+ * amdgpu_debugfs_regs_smc_write - Write to a SMC register
+ *
+ * @f: open file handle
+ * @buf: User buffer to write data from
+ * @size: Number of bytes to write
+ * @pos:  Offset to seek to
+ *
+ * The lower bits are the BYTE offset of the register to write.  This
+ * allows writing multiple registers in a single call and having
+ * the returned size reflect that.
+ */
+static ssize_t amdgpu_debugfs_regs_smc_write(struct file *f, const char __user *buf,
+					 size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (!adev->smc_wreg)
+		return -EOPNOTSUPP;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	r = pm_runtime_get_sync(adev_to_drm(adev)->dev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	r = amdgpu_virt_enable_access_debugfs(adev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	while (size) {
+		uint32_t value;
+
+		r = get_user(value, (uint32_t *)buf);
+		if (r)
+			goto out;
+
+		WREG32_SMC(*pos, value);
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	r = result;
+out:
+	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+	amdgpu_virt_disable_access_debugfs(adev);
+	return r;
+}
+
+/**
+ * amdgpu_debugfs_gca_config_read - Read from gfx config data
+ *
+ * @f: open file handle
+ * @buf: User buffer to store read data in
+ * @size: Number of bytes to read
+ * @pos:  Offset to seek to
+ *
+ * This file is used to access configuration data in a somewhat
+ * stable fashion.  The format is a series of DWORDs with the first
+ * indicating which revision it is.  New content is appended to the
+ * end so that older software can still read the data.
+ */
+
+static ssize_t amdgpu_debugfs_gca_config_read(struct file *f, char __user *buf,
+					size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+	uint32_t *config, no_regs = 0;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	config = kmalloc_array(256, sizeof(*config), GFP_KERNEL);
+	if (!config)
+		return -ENOMEM;
+
+	/* version, increment each time something is added */
+	config[no_regs++] = 5;
+	config[no_regs++] = adev->gfx.config.max_shader_engines;
+	config[no_regs++] = adev->gfx.config.max_tile_pipes;
+	config[no_regs++] = adev->gfx.config.max_cu_per_sh;
+	config[no_regs++] = adev->gfx.config.max_sh_per_se;
+	config[no_regs++] = adev->gfx.config.max_backends_per_se;
+	config[no_regs++] = adev->gfx.config.max_texture_channel_caches;
+	config[no_regs++] = adev->gfx.config.max_gprs;
+	config[no_regs++] = adev->gfx.config.max_gs_threads;
+	config[no_regs++] = adev->gfx.config.max_hw_contexts;
+	config[no_regs++] = adev->gfx.config.sc_prim_fifo_size_frontend;
+	config[no_regs++] = adev->gfx.config.sc_prim_fifo_size_backend;
+	config[no_regs++] = adev->gfx.config.sc_hiz_tile_fifo_size;
+	config[no_regs++] = adev->gfx.config.sc_earlyz_tile_fifo_size;
+	config[no_regs++] = adev->gfx.config.num_tile_pipes;
+	config[no_regs++] = adev->gfx.config.backend_enable_mask;
+	config[no_regs++] = adev->gfx.config.mem_max_burst_length_bytes;
+	config[no_regs++] = adev->gfx.config.mem_row_size_in_kb;
+	config[no_regs++] = adev->gfx.config.shader_engine_tile_size;
+	config[no_regs++] = adev->gfx.config.num_gpus;
+	config[no_regs++] = adev->gfx.config.multi_gpu_tile_size;
+	config[no_regs++] = adev->gfx.config.mc_arb_ramcfg;
+	config[no_regs++] = adev->gfx.config.gb_addr_config;
+	config[no_regs++] = adev->gfx.config.num_rbs;
+
+	/* rev==1 */
+	config[no_regs++] = adev->rev_id;
+	config[no_regs++] = adev->pg_flags;
+	config[no_regs++] = lower_32_bits(adev->cg_flags);
+
+	/* rev==2 */
+	config[no_regs++] = adev->family;
+	config[no_regs++] = adev->external_rev_id;
+
+	/* rev==3 */
+	config[no_regs++] = adev->pdev->device;
+	config[no_regs++] = adev->pdev->revision;
+	config[no_regs++] = adev->pdev->subsystem_device;
+	config[no_regs++] = adev->pdev->subsystem_vendor;
+
+	/* rev==4 APU flag */
+	config[no_regs++] = adev->flags & AMD_IS_APU ? 1 : 0;
+
+	/* rev==5 PG/CG flag upper 32bit */
+	config[no_regs++] = 0;
+	config[no_regs++] = upper_32_bits(adev->cg_flags);
+
+	while (size && (*pos < no_regs * 4)) {
+		uint32_t value;
+
+		value = config[*pos >> 2];
+		r = put_user(value, (uint32_t *)buf);
+		if (r) {
+			kfree(config);
+			return r;
+		}
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	kfree(config);
+	return result;
+}
+
+/**
+ * amdgpu_debugfs_sensor_read - Read from the powerplay sensors
+ *
+ * @f: open file handle
+ * @buf: User buffer to store read data in
+ * @size: Number of bytes to read
+ * @pos:  Offset to seek to
+ *
+ * The offset is treated as the BYTE address of one of the sensors
+ * enumerated in amd/include/kgd_pp_interface.h under the
+ * 'amd_pp_sensors' enumeration.  For instance to read the UVD VCLK
+ * you would use the offset 3 * 4 = 12.
+ */
+static ssize_t amdgpu_debugfs_sensor_read(struct file *f, char __user *buf,
+					size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	int idx, x, outsize, r, valuesize;
+	uint32_t values[16];
+
+	if (size & 3 || *pos & 0x3)
+		return -EINVAL;
+
+	if (!adev->pm.dpm_enabled)
+		return -EINVAL;
+
+	/* convert offset to sensor number */
+	idx = *pos >> 2;
+
+	valuesize = sizeof(values);
+
+	r = pm_runtime_get_sync(adev_to_drm(adev)->dev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	r = amdgpu_virt_enable_access_debugfs(adev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	r = amdgpu_dpm_read_sensor(adev, idx, &values[0], &valuesize);
+
+	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+
+	if (r) {
+		amdgpu_virt_disable_access_debugfs(adev);
+		return r;
+	}
+
+	if (size > valuesize) {
+		amdgpu_virt_disable_access_debugfs(adev);
+		return -EINVAL;
+	}
+
+	outsize = 0;
+	x = 0;
+	if (!r) {
+		while (size) {
+			r = put_user(values[x++], (int32_t *)buf);
+			buf += 4;
+			size -= 4;
+			outsize += 4;
+		}
+	}
+
+	amdgpu_virt_disable_access_debugfs(adev);
+	return !r ? outsize : r;
+}
+
+/** amdgpu_debugfs_wave_read - Read WAVE STATUS data
+ *
+ * @f: open file handle
+ * @buf: User buffer to store read data in
+ * @size: Number of bytes to read
+ * @pos:  Offset to seek to
+ *
+ * The offset being sought changes which wave that the status data
+ * will be returned for.  The bits are used as follows:
+ *
+ * Bits 0..6:	Byte offset into data
+ * Bits 7..14:	SE selector
+ * Bits 15..22:	SH/SA selector
+ * Bits 23..30: CU/{WGP+SIMD} selector
+ * Bits 31..36: WAVE ID selector
+ * Bits 37..44: SIMD ID selector
+ *
+ * The returned data begins with one DWORD of version information
+ * Followed by WAVE STATUS registers relevant to the GFX IP version
+ * being used.  See gfx_v8_0_read_wave_data() for an example output.
+ */
+static ssize_t amdgpu_debugfs_wave_read(struct file *f, char __user *buf,
+					size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = f->f_inode->i_private;
+	int r, x;
+	ssize_t result = 0;
+	uint32_t offset, se, sh, cu, wave, simd, data[32];
+
+	if (size & 3 || *pos & 3)
+		return -EINVAL;
+
+	/* decode offset */
+	offset = (*pos & GENMASK_ULL(6, 0));
+	se = (*pos & GENMASK_ULL(14, 7)) >> 7;
+	sh = (*pos & GENMASK_ULL(22, 15)) >> 15;
+	cu = (*pos & GENMASK_ULL(30, 23)) >> 23;
+	wave = (*pos & GENMASK_ULL(36, 31)) >> 31;
+	simd = (*pos & GENMASK_ULL(44, 37)) >> 37;
+
+	r = pm_runtime_get_sync(adev_to_drm(adev)->dev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	r = amdgpu_virt_enable_access_debugfs(adev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	/* switch to the specific se/sh/cu */
+	mutex_lock(&adev->grbm_idx_mutex);
+	amdgpu_gfx_select_se_sh(adev, se, sh, cu, 0);
+
+	x = 0;
+	if (adev->gfx.funcs->read_wave_data)
+		adev->gfx.funcs->read_wave_data(adev, 0, simd, wave, data, &x);
+
+	amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+
+	if (!x) {
+		amdgpu_virt_disable_access_debugfs(adev);
+		return -EINVAL;
+	}
+
+	while (size && (offset < x * 4)) {
+		uint32_t value;
+
+		value = data[offset >> 2];
+		r = put_user(value, (uint32_t *)buf);
+		if (r) {
+			amdgpu_virt_disable_access_debugfs(adev);
+			return r;
+		}
+
+		result += 4;
+		buf += 4;
+		offset += 4;
+		size -= 4;
+	}
+
+	amdgpu_virt_disable_access_debugfs(adev);
+	return result;
+}
+
+/** amdgpu_debugfs_gpr_read - Read wave gprs
+ *
+ * @f: open file handle
+ * @buf: User buffer to store read data in
+ * @size: Number of bytes to read
+ * @pos:  Offset to seek to
+ *
+ * The offset being sought changes which wave that the status data
+ * will be returned for.  The bits are used as follows:
+ *
+ * Bits 0..11:	Byte offset into data
+ * Bits 12..19:	SE selector
+ * Bits 20..27:	SH/SA selector
+ * Bits 28..35: CU/{WGP+SIMD} selector
+ * Bits 36..43: WAVE ID selector
+ * Bits 37..44: SIMD ID selector
+ * Bits 52..59: Thread selector
+ * Bits 60..61: Bank selector (VGPR=0,SGPR=1)
+ *
+ * The return data comes from the SGPR or VGPR register bank for
+ * the selected operational unit.
+ */
+static ssize_t amdgpu_debugfs_gpr_read(struct file *f, char __user *buf,
+					size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = f->f_inode->i_private;
+	int r;
+	ssize_t result = 0;
+	uint32_t offset, se, sh, cu, wave, simd, thread, bank, *data;
+
+	if (size > 4096 || size & 3 || *pos & 3)
+		return -EINVAL;
+
+	/* decode offset */
+	offset = (*pos & GENMASK_ULL(11, 0)) >> 2;
+	se = (*pos & GENMASK_ULL(19, 12)) >> 12;
+	sh = (*pos & GENMASK_ULL(27, 20)) >> 20;
+	cu = (*pos & GENMASK_ULL(35, 28)) >> 28;
+	wave = (*pos & GENMASK_ULL(43, 36)) >> 36;
+	simd = (*pos & GENMASK_ULL(51, 44)) >> 44;
+	thread = (*pos & GENMASK_ULL(59, 52)) >> 52;
+	bank = (*pos & GENMASK_ULL(61, 60)) >> 60;
+
+	data = kcalloc(1024, sizeof(*data), GFP_KERNEL);
+	if (!data)
+		return -ENOMEM;
+
+	r = pm_runtime_get_sync(adev_to_drm(adev)->dev);
+	if (r < 0)
+		goto err;
+
+	r = amdgpu_virt_enable_access_debugfs(adev);
+	if (r < 0)
+		goto err;
+
+	/* switch to the specific se/sh/cu */
+	mutex_lock(&adev->grbm_idx_mutex);
+	amdgpu_gfx_select_se_sh(adev, se, sh, cu, 0);
+
+	if (bank == 0) {
+		if (adev->gfx.funcs->read_wave_vgprs)
+			adev->gfx.funcs->read_wave_vgprs(adev, 0, simd, wave, thread, offset, size>>2, data);
+	} else {
+		if (adev->gfx.funcs->read_wave_sgprs)
+			adev->gfx.funcs->read_wave_sgprs(adev, 0, simd, wave, offset, size>>2, data);
+	}
+
+	amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+
+	while (size) {
+		uint32_t value;
+
+		value = data[result >> 2];
+		r = put_user(value, (uint32_t *)buf);
+		if (r) {
+			amdgpu_virt_disable_access_debugfs(adev);
+			goto err;
+		}
+
+		result += 4;
+		buf += 4;
+		size -= 4;
+	}
+
+	kfree(data);
+	amdgpu_virt_disable_access_debugfs(adev);
+	return result;
+
+err:
+	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+	kfree(data);
+	return r;
+}
+
+/**
+ * amdgpu_debugfs_gfxoff_residency_read - Read GFXOFF residency
+ *
+ * @f: open file handle
+ * @buf: User buffer to store read data in
+ * @size: Number of bytes to read
+ * @pos:  Offset to seek to
+ *
+ * Read the last residency value logged. It doesn't auto update, one needs to
+ * stop logging before getting the current value.
+ */
+static ssize_t amdgpu_debugfs_gfxoff_residency_read(struct file *f, char __user *buf,
+						    size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	r = pm_runtime_get_sync(adev_to_drm(adev)->dev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	while (size) {
+		uint32_t value;
+
+		r = amdgpu_get_gfx_off_residency(adev, &value);
+		if (r)
+			goto out;
+
+		r = put_user(value, (uint32_t *)buf);
+		if (r)
+			goto out;
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	r = result;
+out:
+	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+
+	return r;
+}
+
+/**
+ * amdgpu_debugfs_gfxoff_residency_write - Log GFXOFF Residency
+ *
+ * @f: open file handle
+ * @buf: User buffer to write data from
+ * @size: Number of bytes to write
+ * @pos:  Offset to seek to
+ *
+ * Write a 32-bit non-zero to start logging; write a 32-bit zero to stop
+ */
+static ssize_t amdgpu_debugfs_gfxoff_residency_write(struct file *f, const char __user *buf,
+						     size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	r = pm_runtime_get_sync(adev_to_drm(adev)->dev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	while (size) {
+		u32 value;
+
+		r = get_user(value, (uint32_t *)buf);
+		if (r)
+			goto out;
+
+		amdgpu_set_gfx_off_residency(adev, value ? true : false);
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	r = result;
+out:
+	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+
+	return r;
+}
+
+
+/**
+ * amdgpu_debugfs_gfxoff_count_read - Read GFXOFF entry count
+ *
+ * @f: open file handle
+ * @buf: User buffer to store read data in
+ * @size: Number of bytes to read
+ * @pos:  Offset to seek to
+ */
+static ssize_t amdgpu_debugfs_gfxoff_count_read(struct file *f, char __user *buf,
+						size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	r = pm_runtime_get_sync(adev_to_drm(adev)->dev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	while (size) {
+		u64 value = 0;
+
+		r = amdgpu_get_gfx_off_entrycount(adev, &value);
+		if (r)
+			goto out;
+
+		r = put_user(value, (u64 *)buf);
+		if (r)
+			goto out;
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	r = result;
+out:
+	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+
+	return r;
+}
+
+/**
+ * amdgpu_debugfs_gfxoff_write - Enable/disable GFXOFF
+ *
+ * @f: open file handle
+ * @buf: User buffer to write data from
+ * @size: Number of bytes to write
+ * @pos:  Offset to seek to
+ *
+ * Write a 32-bit zero to disable or a 32-bit non-zero to enable
+ */
+static ssize_t amdgpu_debugfs_gfxoff_write(struct file *f, const char __user *buf,
+					 size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	r = pm_runtime_get_sync(adev_to_drm(adev)->dev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	while (size) {
+		uint32_t value;
+
+		r = get_user(value, (uint32_t *)buf);
+		if (r)
+			goto out;
+
+		amdgpu_gfx_off_ctrl(adev, value ? true : false);
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	r = result;
+out:
+	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+
+	return r;
+}
+
+
+/**
+ * amdgpu_debugfs_gfxoff_read - read gfxoff status
+ *
+ * @f: open file handle
+ * @buf: User buffer to store read data in
+ * @size: Number of bytes to read
+ * @pos:  Offset to seek to
+ */
+static ssize_t amdgpu_debugfs_gfxoff_read(struct file *f, char __user *buf,
+					 size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	r = pm_runtime_get_sync(adev_to_drm(adev)->dev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	while (size) {
+		u32 value = adev->gfx.gfx_off_state;
+
+		r = put_user(value, (u32 *)buf);
+		if (r)
+			goto out;
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	r = result;
+out:
+	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+
+	return r;
+}
+
+static ssize_t amdgpu_debugfs_gfxoff_status_read(struct file *f, char __user *buf,
+						 size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	r = pm_runtime_get_sync(adev_to_drm(adev)->dev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	while (size) {
+		u32 value;
+
+		r = amdgpu_get_gfx_off_status(adev, &value);
+		if (r)
+			goto out;
+
+		r = put_user(value, (u32 *)buf);
+		if (r)
+			goto out;
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	r = result;
+out:
+	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+
+	return r;
+}
+
+static const struct file_operations amdgpu_debugfs_regs2_fops = {
+	.owner = THIS_MODULE,
+	.unlocked_ioctl = amdgpu_debugfs_regs2_ioctl,
+	.read = amdgpu_debugfs_regs2_read,
+	.write = amdgpu_debugfs_regs2_write,
+	.open = amdgpu_debugfs_regs2_open,
+	.release = amdgpu_debugfs_regs2_release,
+	.llseek = default_llseek
+};
+
+static const struct file_operations amdgpu_debugfs_gprwave_fops = {
+	.owner = THIS_MODULE,
+	.unlocked_ioctl = amdgpu_debugfs_gprwave_ioctl,
+	.read = amdgpu_debugfs_gprwave_read,
+	.open = amdgpu_debugfs_gprwave_open,
+	.release = amdgpu_debugfs_gprwave_release,
+	.llseek = default_llseek
+};
+
+static const struct file_operations amdgpu_debugfs_regs_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_regs_read,
+	.write = amdgpu_debugfs_regs_write,
+	.llseek = default_llseek
+};
+static const struct file_operations amdgpu_debugfs_regs_didt_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_regs_didt_read,
+	.write = amdgpu_debugfs_regs_didt_write,
+	.llseek = default_llseek
+};
+static const struct file_operations amdgpu_debugfs_regs_pcie_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_regs_pcie_read,
+	.write = amdgpu_debugfs_regs_pcie_write,
+	.llseek = default_llseek
+};
+static const struct file_operations amdgpu_debugfs_regs_smc_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_regs_smc_read,
+	.write = amdgpu_debugfs_regs_smc_write,
+	.llseek = default_llseek
+};
+
+static const struct file_operations amdgpu_debugfs_gca_config_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_gca_config_read,
+	.llseek = default_llseek
+};
+
+static const struct file_operations amdgpu_debugfs_sensors_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_sensor_read,
+	.llseek = default_llseek
+};
+
+static const struct file_operations amdgpu_debugfs_wave_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_wave_read,
+	.llseek = default_llseek
+};
+static const struct file_operations amdgpu_debugfs_gpr_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_gpr_read,
+	.llseek = default_llseek
+};
+
+static const struct file_operations amdgpu_debugfs_gfxoff_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_gfxoff_read,
+	.write = amdgpu_debugfs_gfxoff_write,
+	.llseek = default_llseek
+};
+
+static const struct file_operations amdgpu_debugfs_gfxoff_status_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_gfxoff_status_read,
+	.llseek = default_llseek
+};
+
+static const struct file_operations amdgpu_debugfs_gfxoff_count_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_gfxoff_count_read,
+	.llseek = default_llseek
+};
+
+static const struct file_operations amdgpu_debugfs_gfxoff_residency_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_gfxoff_residency_read,
+	.write = amdgpu_debugfs_gfxoff_residency_write,
+	.llseek = default_llseek
+};
+
+static const struct file_operations *debugfs_regs[] = {
+	&amdgpu_debugfs_regs_fops,
+	&amdgpu_debugfs_regs2_fops,
+	&amdgpu_debugfs_gprwave_fops,
+	&amdgpu_debugfs_regs_didt_fops,
+	&amdgpu_debugfs_regs_pcie_fops,
+	&amdgpu_debugfs_regs_smc_fops,
+	&amdgpu_debugfs_gca_config_fops,
+	&amdgpu_debugfs_sensors_fops,
+	&amdgpu_debugfs_wave_fops,
+	&amdgpu_debugfs_gpr_fops,
+	&amdgpu_debugfs_gfxoff_fops,
+	&amdgpu_debugfs_gfxoff_status_fops,
+	&amdgpu_debugfs_gfxoff_count_fops,
+	&amdgpu_debugfs_gfxoff_residency_fops,
+};
+
+static const char * const debugfs_regs_names[] = {
+	"amdgpu_regs",
+	"amdgpu_regs2",
+	"amdgpu_gprwave",
+	"amdgpu_regs_didt",
+	"amdgpu_regs_pcie",
+	"amdgpu_regs_smc",
+	"amdgpu_gca_config",
+	"amdgpu_sensors",
+	"amdgpu_wave",
+	"amdgpu_gpr",
+	"amdgpu_gfxoff",
+	"amdgpu_gfxoff_status",
+	"amdgpu_gfxoff_count",
+	"amdgpu_gfxoff_residency",
+};
+
+/**
+ * amdgpu_debugfs_regs_init -	Initialize debugfs entries that provide
+ *				register access.
+ *
+ * @adev: The device to attach the debugfs entries to
+ */
+int amdgpu_debugfs_regs_init(struct amdgpu_device *adev)
+{
+	struct drm_minor *minor = adev_to_drm(adev)->primary;
+	struct dentry *ent, *root = minor->debugfs_root;
+	unsigned int i;
+
+	for (i = 0; i < ARRAY_SIZE(debugfs_regs); i++) {
+		ent = debugfs_create_file(debugfs_regs_names[i],
+					  S_IFREG | 0400, root,
+					  adev, debugfs_regs[i]);
+		if (!i && !IS_ERR_OR_NULL(ent))
+			i_size_write(ent->d_inode, adev->rmmio_size);
+	}
+
+	return 0;
+}
+
+static int amdgpu_debugfs_test_ib_show(struct seq_file *m, void *unused)
+{
+	struct amdgpu_device *adev = m->private;
+	struct drm_device *dev = adev_to_drm(adev);
+	int r = 0, i;
+
+	r = pm_runtime_get_sync(dev->dev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(dev->dev);
+		return r;
+	}
+
+	/* Avoid accidently unparking the sched thread during GPU reset */
+	r = down_write_killable(&adev->reset_domain->sem);
+	if (r)
+		return r;
+
+	/* hold on the scheduler */
+	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
+		struct amdgpu_ring *ring = adev->rings[i];
+
+		if (!amdgpu_ring_sched_ready(ring))
+			continue;
+		drm_sched_wqueue_stop(&ring->sched);
+	}
+
+	seq_puts(m, "run ib test:\n");
+	r = amdgpu_ib_ring_tests(adev);
+	if (r)
+		seq_printf(m, "ib ring tests failed (%d).\n", r);
+	else
+		seq_puts(m, "ib ring tests passed.\n");
+
+	/* go on the scheduler */
+	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
+		struct amdgpu_ring *ring = adev->rings[i];
+
+		if (!amdgpu_ring_sched_ready(ring))
+			continue;
+		drm_sched_wqueue_start(&ring->sched);
+	}
+
+	up_write(&adev->reset_domain->sem);
+
+	pm_runtime_mark_last_busy(dev->dev);
+	pm_runtime_put_autosuspend(dev->dev);
+
+	return 0;
+}
+
+static int amdgpu_debugfs_evict_vram(void *data, u64 *val)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)data;
+	struct drm_device *dev = adev_to_drm(adev);
+	int r;
+
+	r = pm_runtime_get_sync(dev->dev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(dev->dev);
+		return r;
+	}
+
+	*val = amdgpu_ttm_evict_resources(adev, TTM_PL_VRAM);
+
+	pm_runtime_mark_last_busy(dev->dev);
+	pm_runtime_put_autosuspend(dev->dev);
+
+	return 0;
+}
+
+
+static int amdgpu_debugfs_evict_gtt(void *data, u64 *val)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)data;
+	struct drm_device *dev = adev_to_drm(adev);
+	int r;
+
+	r = pm_runtime_get_sync(dev->dev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(dev->dev);
+		return r;
+	}
+
+	*val = amdgpu_ttm_evict_resources(adev, TTM_PL_TT);
+
+	pm_runtime_mark_last_busy(dev->dev);
+	pm_runtime_put_autosuspend(dev->dev);
+
+	return 0;
+}
+
+static int amdgpu_debugfs_benchmark(void *data, u64 val)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)data;
+	struct drm_device *dev = adev_to_drm(adev);
+	int r;
+
+	r = pm_runtime_get_sync(dev->dev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(dev->dev);
+		return r;
+	}
+
+	r = amdgpu_benchmark(adev, val);
+
+	pm_runtime_mark_last_busy(dev->dev);
+	pm_runtime_put_autosuspend(dev->dev);
+
+	return r;
+}
+
+static int amdgpu_debugfs_vm_info_show(struct seq_file *m, void *unused)
+{
+	struct amdgpu_device *adev = m->private;
+	struct drm_device *dev = adev_to_drm(adev);
+	struct drm_file *file;
+	int r;
+
+	r = mutex_lock_interruptible(&dev->filelist_mutex);
+	if (r)
+		return r;
+
+	list_for_each_entry(file, &dev->filelist, lhead) {
+		struct amdgpu_fpriv *fpriv = file->driver_priv;
+		struct amdgpu_vm *vm = &fpriv->vm;
+		struct amdgpu_task_info *ti;
+
+		ti = amdgpu_vm_get_task_info_vm(vm);
+		if (ti) {
+			seq_printf(m, "pid:%d\tProcess:%s ----------\n", ti->task.pid, ti->process_name);
+			amdgpu_vm_put_task_info(ti);
+		}
+
+		r = amdgpu_bo_reserve(vm->root.bo, true);
+		if (r)
+			break;
+		amdgpu_debugfs_vm_bo_info(vm, m);
+		amdgpu_bo_unreserve(vm->root.bo);
+	}
+
+	mutex_unlock(&dev->filelist_mutex);
+
+	return r;
+}
+
+DEFINE_SHOW_ATTRIBUTE(amdgpu_debugfs_test_ib);
+DEFINE_SHOW_ATTRIBUTE(amdgpu_debugfs_vm_info);
+DEFINE_DEBUGFS_ATTRIBUTE(amdgpu_evict_vram_fops, amdgpu_debugfs_evict_vram,
+			 NULL, "%lld\n");
+DEFINE_DEBUGFS_ATTRIBUTE(amdgpu_evict_gtt_fops, amdgpu_debugfs_evict_gtt,
+			 NULL, "%lld\n");
+DEFINE_DEBUGFS_ATTRIBUTE(amdgpu_benchmark_fops, NULL, amdgpu_debugfs_benchmark,
+			 "%lld\n");
+
+static void amdgpu_ib_preempt_fences_swap(struct amdgpu_ring *ring,
+					  struct dma_fence **fences)
+{
+	struct amdgpu_fence_driver *drv = &ring->fence_drv;
+	uint32_t sync_seq, last_seq;
+
+	last_seq = atomic_read(&ring->fence_drv.last_seq);
+	sync_seq = ring->fence_drv.sync_seq;
+
+	last_seq &= drv->num_fences_mask;
+	sync_seq &= drv->num_fences_mask;
+
+	do {
+		struct dma_fence *fence, **ptr;
+
+		++last_seq;
+		last_seq &= drv->num_fences_mask;
+		ptr = &drv->fences[last_seq];
+
+		fence = rcu_dereference_protected(*ptr, 1);
+		RCU_INIT_POINTER(*ptr, NULL);
+
+		if (!fence)
+			continue;
+
+		fences[last_seq] = fence;
+
+	} while (last_seq != sync_seq);
+}
+
+static void amdgpu_ib_preempt_signal_fences(struct dma_fence **fences,
+					    int length)
+{
+	int i;
+	struct dma_fence *fence;
+
+	for (i = 0; i < length; i++) {
+		fence = fences[i];
+		if (!fence)
+			continue;
+		dma_fence_signal(fence);
+		dma_fence_put(fence);
+	}
+}
+
+static void amdgpu_ib_preempt_job_recovery(struct drm_gpu_scheduler *sched)
+{
+	struct drm_sched_job *s_job;
+	struct dma_fence *fence;
+
+	spin_lock(&sched->job_list_lock);
+	list_for_each_entry(s_job, &sched->pending_list, list) {
+		fence = sched->ops->run_job(s_job);
+		dma_fence_put(fence);
+	}
+	spin_unlock(&sched->job_list_lock);
+}
+
+static void amdgpu_ib_preempt_mark_partial_job(struct amdgpu_ring *ring)
+{
+	struct amdgpu_job *job;
+	struct drm_sched_job *s_job, *tmp;
+	uint32_t preempt_seq;
+	struct dma_fence *fence, **ptr;
+	struct amdgpu_fence_driver *drv = &ring->fence_drv;
+	struct drm_gpu_scheduler *sched = &ring->sched;
+	bool preempted = true;
+
+	if (ring->funcs->type != AMDGPU_RING_TYPE_GFX)
+		return;
+
+	preempt_seq = le32_to_cpu(*(drv->cpu_addr + 2));
+	if (preempt_seq <= atomic_read(&drv->last_seq)) {
+		preempted = false;
+		goto no_preempt;
+	}
+
+	preempt_seq &= drv->num_fences_mask;
+	ptr = &drv->fences[preempt_seq];
+	fence = rcu_dereference_protected(*ptr, 1);
+
+no_preempt:
+	spin_lock(&sched->job_list_lock);
+	list_for_each_entry_safe(s_job, tmp, &sched->pending_list, list) {
+		if (dma_fence_is_signaled(&s_job->s_fence->finished)) {
+			/* remove job from ring_mirror_list */
+			list_del_init(&s_job->list);
+			sched->ops->free_job(s_job);
+			continue;
+		}
+		job = to_amdgpu_job(s_job);
+		if (preempted && (&job->hw_fence.base) == fence)
+			/* mark the job as preempted */
+			job->preemption_status |= AMDGPU_IB_PREEMPTED;
+	}
+	spin_unlock(&sched->job_list_lock);
+}
+
+static int amdgpu_debugfs_ib_preempt(void *data, u64 val)
+{
+	int r, length;
+	struct amdgpu_ring *ring;
+	struct dma_fence **fences = NULL;
+	struct amdgpu_device *adev = (struct amdgpu_device *)data;
+
+	if (val >= AMDGPU_MAX_RINGS)
+		return -EINVAL;
+
+	ring = adev->rings[val];
+
+	if (!amdgpu_ring_sched_ready(ring) ||
+	    !ring->funcs->preempt_ib)
+		return -EINVAL;
+
+	/* the last preemption failed */
+	if (ring->trail_seq != le32_to_cpu(*ring->trail_fence_cpu_addr))
+		return -EBUSY;
+
+	length = ring->fence_drv.num_fences_mask + 1;
+	fences = kcalloc(length, sizeof(void *), GFP_KERNEL);
+	if (!fences)
+		return -ENOMEM;
+
+	/* Avoid accidently unparking the sched thread during GPU reset */
+	r = down_read_killable(&adev->reset_domain->sem);
+	if (r)
+		goto pro_end;
+
+	/* stop the scheduler */
+	drm_sched_wqueue_stop(&ring->sched);
+
+	/* preempt the IB */
+	r = amdgpu_ring_preempt_ib(ring);
+	if (r) {
+		DRM_WARN("failed to preempt ring %d\n", ring->idx);
+		goto failure;
+	}
+
+	amdgpu_fence_process(ring);
+
+	if (atomic_read(&ring->fence_drv.last_seq) !=
+	    ring->fence_drv.sync_seq) {
+		DRM_INFO("ring %d was preempted\n", ring->idx);
+
+		amdgpu_ib_preempt_mark_partial_job(ring);
+
+		/* swap out the old fences */
+		amdgpu_ib_preempt_fences_swap(ring, fences);
+
+		amdgpu_fence_driver_force_completion(ring);
+
+		/* resubmit unfinished jobs */
+		amdgpu_ib_preempt_job_recovery(&ring->sched);
+
+		/* wait for jobs finished */
+		amdgpu_fence_wait_empty(ring);
+
+		/* signal the old fences */
+		amdgpu_ib_preempt_signal_fences(fences, length);
+	}
+
+failure:
+	/* restart the scheduler */
+	drm_sched_wqueue_start(&ring->sched);
+
+	up_read(&adev->reset_domain->sem);
+
+pro_end:
+	kfree(fences);
+
+	return r;
+}
+
+static int amdgpu_debugfs_sclk_set(void *data, u64 val)
+{
+	int ret = 0;
+	uint32_t max_freq, min_freq;
+	struct amdgpu_device *adev = (struct amdgpu_device *)data;
+
+	if (amdgpu_sriov_multi_vf_mode(adev))
+		return -EINVAL;
+
+	ret = pm_runtime_get_sync(adev_to_drm(adev)->dev);
+	if (ret < 0) {
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return ret;
+	}
+
+	ret = amdgpu_dpm_get_dpm_freq_range(adev, PP_SCLK, &min_freq, &max_freq);
+	if (ret == -EOPNOTSUPP) {
+		ret = 0;
+		goto out;
+	}
+	if (ret || val > max_freq || val < min_freq) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	ret = amdgpu_dpm_set_soft_freq_range(adev, PP_SCLK, (uint32_t)val, (uint32_t)val);
+	if (ret)
+		ret = -EINVAL;
+
+out:
+	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+
+	return ret;
+}
+
+DEFINE_DEBUGFS_ATTRIBUTE(fops_ib_preempt, NULL,
+			amdgpu_debugfs_ib_preempt, "%llu\n");
+
+DEFINE_DEBUGFS_ATTRIBUTE(fops_sclk_set, NULL,
+			amdgpu_debugfs_sclk_set, "%llu\n");
+
+int amdgpu_debugfs_init(struct amdgpu_device *adev)
+{
+	struct dentry *root = adev_to_drm(adev)->primary->debugfs_root;
+	struct dentry *ent;
+	int r, i;
+
+	if (!debugfs_initialized())
+		return 0;
+
+	debugfs_create_x32("amdgpu_smu_debug", 0600, root,
+			   &adev->pm.smu_debug_mask);
+
+	ent = debugfs_create_file("amdgpu_preempt_ib", 0600, root, adev,
+				  &fops_ib_preempt);
+	if (IS_ERR(ent)) {
+		DRM_ERROR("unable to create amdgpu_preempt_ib debugsfs file\n");
+		return PTR_ERR(ent);
+	}
+
+	ent = debugfs_create_file("amdgpu_force_sclk", 0200, root, adev,
+				  &fops_sclk_set);
+	if (IS_ERR(ent)) {
+		DRM_ERROR("unable to create amdgpu_set_sclk debugsfs file\n");
+		return PTR_ERR(ent);
+	}
+
+	/* Register debugfs entries for amdgpu_ttm */
+	amdgpu_ttm_debugfs_init(adev);
+	amdgpu_debugfs_pm_init(adev);
+	amdgpu_debugfs_sa_init(adev);
+	amdgpu_debugfs_fence_init(adev);
+	amdgpu_debugfs_gem_init(adev);
+
+	r = amdgpu_debugfs_regs_init(adev);
+	if (r)
+		DRM_ERROR("registering register debugfs failed (%d).\n", r);
+
+	amdgpu_debugfs_firmware_init(adev);
+	amdgpu_ta_if_debugfs_init(adev);
+
+	amdgpu_debugfs_mes_event_log_init(adev);
+
+#if defined(CONFIG_DRM_AMD_DC)
+	if (adev->dc_enabled)
+		dtn_debugfs_init(adev);
+#endif
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		struct amdgpu_ring *ring = adev->rings[i];
+
+		if (!ring)
+			continue;
+
+		amdgpu_debugfs_ring_init(adev, ring);
+	}
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		if (!amdgpu_vcnfw_log)
+			break;
+
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		amdgpu_debugfs_vcn_fwlog_init(adev, i, &adev->vcn.inst[i]);
+	}
+
+	if (amdgpu_umsch_mm & amdgpu_umsch_mm_fwlog)
+		amdgpu_debugfs_umsch_fwlog_init(adev, &adev->umsch_mm);
+
+	amdgpu_debugfs_vcn_sched_mask_init(adev);
+	amdgpu_debugfs_jpeg_sched_mask_init(adev);
+	amdgpu_debugfs_gfx_sched_mask_init(adev);
+	amdgpu_debugfs_compute_sched_mask_init(adev);
+	amdgpu_debugfs_sdma_sched_mask_init(adev);
+
+	amdgpu_ras_debugfs_create_all(adev);
+	amdgpu_rap_debugfs_init(adev);
+	amdgpu_securedisplay_debugfs_init(adev);
+	amdgpu_fw_attestation_debugfs_init(adev);
+	amdgpu_psp_debugfs_init(adev);
+
+	debugfs_create_file("amdgpu_evict_vram", 0400, root, adev,
+			    &amdgpu_evict_vram_fops);
+	debugfs_create_file("amdgpu_evict_gtt", 0400, root, adev,
+			    &amdgpu_evict_gtt_fops);
+	debugfs_create_file("amdgpu_test_ib", 0400, root, adev,
+			    &amdgpu_debugfs_test_ib_fops);
+	debugfs_create_file("amdgpu_vm_info", 0444, root, adev,
+			    &amdgpu_debugfs_vm_info_fops);
+	debugfs_create_file("amdgpu_benchmark", 0200, root, adev,
+			    &amdgpu_benchmark_fops);
+
+	adev->debugfs_vbios_blob.data = adev->bios;
+	adev->debugfs_vbios_blob.size = adev->bios_size;
+	debugfs_create_blob("amdgpu_vbios", 0444, root,
+			    &adev->debugfs_vbios_blob);
+
+	adev->debugfs_discovery_blob.data = adev->mman.discovery_bin;
+	adev->debugfs_discovery_blob.size = adev->mman.discovery_tmr_size;
+	debugfs_create_blob("amdgpu_discovery", 0444, root,
+			    &adev->debugfs_discovery_blob);
+
+	return 0;
+}
+
+static int amdgpu_pt_info_read(struct seq_file *m, void *unused)
+{
+	struct drm_file *file;
+	struct amdgpu_fpriv *fpriv;
+	struct amdgpu_bo *root_bo;
+	struct amdgpu_device *adev;
+	int r;
+
+	file = m->private;
+	if (!file)
+		return -EINVAL;
+
+	adev = drm_to_adev(file->minor->dev);
+	fpriv = file->driver_priv;
+	if (!fpriv || !fpriv->vm.root.bo)
+		return -ENODEV;
+
+	root_bo = amdgpu_bo_ref(fpriv->vm.root.bo);
+	r = amdgpu_bo_reserve(root_bo, true);
+	if (r) {
+		amdgpu_bo_unref(&root_bo);
+		return -EINVAL;
+	}
+
+	seq_printf(m, "pd_address: 0x%llx\n", amdgpu_gmc_pd_addr(fpriv->vm.root.bo));
+	seq_printf(m, "max_pfn: 0x%llx\n", adev->vm_manager.max_pfn);
+	seq_printf(m, "num_level: 0x%x\n", adev->vm_manager.num_level);
+	seq_printf(m, "block_size: 0x%x\n", adev->vm_manager.block_size);
+	seq_printf(m, "fragment_size: 0x%x\n", adev->vm_manager.fragment_size);
+
+	amdgpu_bo_unreserve(root_bo);
+	amdgpu_bo_unref(&root_bo);
+
+	return 0;
+}
+
+static int amdgpu_pt_info_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, amdgpu_pt_info_read, inode->i_private);
+}
+
+static const struct file_operations amdgpu_pt_info_fops = {
+	.owner = THIS_MODULE,
+	.open = amdgpu_pt_info_open,
+	.read = seq_read,
+	.llseek = seq_lseek,
+	.release = single_release,
+};
+
+void amdgpu_debugfs_vm_init(struct drm_file *file)
+{
+	debugfs_create_file("vm_pagetable_info", 0444, file->debugfs_client, file,
+			    &amdgpu_pt_info_fops);
+}
+
+#else
+int amdgpu_debugfs_init(struct amdgpu_device *adev)
+{
+	return 0;
+}
+int amdgpu_debugfs_regs_init(struct amdgpu_device *adev)
+{
+	return 0;
+}
+void amdgpu_debugfs_vm_init(struct drm_file *file)
+{
+}
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.h
new file mode 100644
index 000000000000..e7b3c38e5186
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.h
@@ -0,0 +1,37 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+/*
+ * Debugfs
+ */
+
+int amdgpu_debugfs_regs_init(struct amdgpu_device *adev);
+int amdgpu_debugfs_init(struct amdgpu_device *adev);
+void amdgpu_debugfs_fini(struct amdgpu_device *adev);
+void amdgpu_debugfs_fence_init(struct amdgpu_device *adev);
+void amdgpu_debugfs_firmware_init(struct amdgpu_device *adev);
+void amdgpu_debugfs_gem_init(struct amdgpu_device *adev);
+void amdgpu_debugfs_mes_event_log_init(struct amdgpu_device *adev);
+void amdgpu_debugfs_vm_init(struct drm_file *file);
+
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c
new file mode 100644
index 000000000000..8a026bc9ea44
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c
@@ -0,0 +1,372 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <generated/utsrelease.h>
+#include <linux/devcoredump.h>
+#include "amdgpu_dev_coredump.h"
+#include "atom.h"
+
+#ifndef CONFIG_DEV_COREDUMP
+void amdgpu_coredump(struct amdgpu_device *adev, bool skip_vram_check,
+		     bool vram_lost, struct amdgpu_job *job)
+{
+}
+#else
+
+const char *hw_ip_names[MAX_HWIP] = {
+	[GC_HWIP]		= "GC",
+	[HDP_HWIP]		= "HDP",
+	[SDMA0_HWIP]		= "SDMA0",
+	[SDMA1_HWIP]		= "SDMA1",
+	[SDMA2_HWIP]		= "SDMA2",
+	[SDMA3_HWIP]		= "SDMA3",
+	[SDMA4_HWIP]		= "SDMA4",
+	[SDMA5_HWIP]		= "SDMA5",
+	[SDMA6_HWIP]		= "SDMA6",
+	[SDMA7_HWIP]		= "SDMA7",
+	[LSDMA_HWIP]		= "LSDMA",
+	[MMHUB_HWIP]		= "MMHUB",
+	[ATHUB_HWIP]		= "ATHUB",
+	[NBIO_HWIP]		= "NBIO",
+	[MP0_HWIP]		= "MP0",
+	[MP1_HWIP]		= "MP1",
+	[UVD_HWIP]		= "UVD/JPEG/VCN",
+	[VCN1_HWIP]		= "VCN1",
+	[VCE_HWIP]		= "VCE",
+	[VPE_HWIP]		= "VPE",
+	[DF_HWIP]		= "DF",
+	[DCE_HWIP]		= "DCE",
+	[OSSSYS_HWIP]		= "OSSSYS",
+	[SMUIO_HWIP]		= "SMUIO",
+	[PWR_HWIP]		= "PWR",
+	[NBIF_HWIP]		= "NBIF",
+	[THM_HWIP]		= "THM",
+	[CLK_HWIP]		= "CLK",
+	[UMC_HWIP]		= "UMC",
+	[RSMU_HWIP]		= "RSMU",
+	[XGMI_HWIP]		= "XGMI",
+	[DCI_HWIP]		= "DCI",
+	[PCIE_HWIP]		= "PCIE",
+};
+
+static void amdgpu_devcoredump_fw_info(struct amdgpu_device *adev,
+				       struct drm_printer *p)
+{
+	uint32_t version;
+	uint32_t feature;
+	uint8_t smu_program, smu_major, smu_minor, smu_debug;
+	struct atom_context *ctx = adev->mode_info.atom_context;
+
+	drm_printf(p, "VCE feature version: %u, fw version: 0x%08x\n",
+		   adev->vce.fb_version, adev->vce.fw_version);
+	drm_printf(p, "UVD feature version: %u, fw version: 0x%08x\n", 0,
+		   adev->uvd.fw_version);
+	drm_printf(p, "GMC feature version: %u, fw version: 0x%08x\n", 0,
+		   adev->gmc.fw_version);
+	drm_printf(p, "ME feature version: %u, fw version: 0x%08x\n",
+		   adev->gfx.me_feature_version, adev->gfx.me_fw_version);
+	drm_printf(p, "PFP feature version: %u, fw version: 0x%08x\n",
+		   adev->gfx.pfp_feature_version, adev->gfx.pfp_fw_version);
+	drm_printf(p, "CE feature version: %u, fw version: 0x%08x\n",
+		   adev->gfx.ce_feature_version, adev->gfx.ce_fw_version);
+	drm_printf(p, "RLC feature version: %u, fw version: 0x%08x\n",
+		   adev->gfx.rlc_feature_version, adev->gfx.rlc_fw_version);
+
+	drm_printf(p, "RLC SRLC feature version: %u, fw version: 0x%08x\n",
+		   adev->gfx.rlc_srlc_feature_version,
+		   adev->gfx.rlc_srlc_fw_version);
+	drm_printf(p, "RLC SRLG feature version: %u, fw version: 0x%08x\n",
+		   adev->gfx.rlc_srlg_feature_version,
+		   adev->gfx.rlc_srlg_fw_version);
+	drm_printf(p, "RLC SRLS feature version: %u, fw version: 0x%08x\n",
+		   adev->gfx.rlc_srls_feature_version,
+		   adev->gfx.rlc_srls_fw_version);
+	drm_printf(p, "RLCP feature version: %u, fw version: 0x%08x\n",
+		   adev->gfx.rlcp_ucode_feature_version,
+		   adev->gfx.rlcp_ucode_version);
+	drm_printf(p, "RLCV feature version: %u, fw version: 0x%08x\n",
+		   adev->gfx.rlcv_ucode_feature_version,
+		   adev->gfx.rlcv_ucode_version);
+	drm_printf(p, "MEC feature version: %u, fw version: 0x%08x\n",
+		   adev->gfx.mec_feature_version, adev->gfx.mec_fw_version);
+
+	if (adev->gfx.mec2_fw)
+		drm_printf(p, "MEC2 feature version: %u, fw version: 0x%08x\n",
+			   adev->gfx.mec2_feature_version,
+			   adev->gfx.mec2_fw_version);
+
+	drm_printf(p, "IMU feature version: %u, fw version: 0x%08x\n", 0,
+		   adev->gfx.imu_fw_version);
+	drm_printf(p, "PSP SOS feature version: %u, fw version: 0x%08x\n",
+		   adev->psp.sos.feature_version, adev->psp.sos.fw_version);
+	drm_printf(p, "PSP ASD feature version: %u, fw version: 0x%08x\n",
+		   adev->psp.asd_context.bin_desc.feature_version,
+		   adev->psp.asd_context.bin_desc.fw_version);
+
+	drm_printf(p, "TA XGMI feature version: 0x%08x, fw version: 0x%08x\n",
+		   adev->psp.xgmi_context.context.bin_desc.feature_version,
+		   adev->psp.xgmi_context.context.bin_desc.fw_version);
+	drm_printf(p, "TA RAS feature version: 0x%08x, fw version: 0x%08x\n",
+		   adev->psp.ras_context.context.bin_desc.feature_version,
+		   adev->psp.ras_context.context.bin_desc.fw_version);
+	drm_printf(p, "TA HDCP feature version: 0x%08x, fw version: 0x%08x\n",
+		   adev->psp.hdcp_context.context.bin_desc.feature_version,
+		   adev->psp.hdcp_context.context.bin_desc.fw_version);
+	drm_printf(p, "TA DTM feature version: 0x%08x, fw version: 0x%08x\n",
+		   adev->psp.dtm_context.context.bin_desc.feature_version,
+		   adev->psp.dtm_context.context.bin_desc.fw_version);
+	drm_printf(p, "TA RAP feature version: 0x%08x, fw version: 0x%08x\n",
+		   adev->psp.rap_context.context.bin_desc.feature_version,
+		   adev->psp.rap_context.context.bin_desc.fw_version);
+	drm_printf(p,
+		   "TA SECURE DISPLAY feature version: 0x%08x, fw version: 0x%08x\n",
+		   adev->psp.securedisplay_context.context.bin_desc.feature_version,
+		   adev->psp.securedisplay_context.context.bin_desc.fw_version);
+
+	/* SMC firmware */
+	version = adev->pm.fw_version;
+
+	smu_program = (version >> 24) & 0xff;
+	smu_major = (version >> 16) & 0xff;
+	smu_minor = (version >> 8) & 0xff;
+	smu_debug = (version >> 0) & 0xff;
+	drm_printf(p,
+		   "SMC feature version: %u, program: %d, fw version: 0x%08x (%d.%d.%d)\n",
+		   0, smu_program, version, smu_major, smu_minor, smu_debug);
+
+	/* SDMA firmware */
+	for (int i = 0; i < adev->sdma.num_instances; i++) {
+		drm_printf(p,
+			   "SDMA%d feature version: %u, firmware version: 0x%08x\n",
+			   i, adev->sdma.instance[i].feature_version,
+			   adev->sdma.instance[i].fw_version);
+	}
+
+	drm_printf(p, "VCN feature version: %u, fw version: 0x%08x\n", 0,
+		   adev->vcn.fw_version);
+	drm_printf(p, "DMCU feature version: %u, fw version: 0x%08x\n", 0,
+		   adev->dm.dmcu_fw_version);
+	drm_printf(p, "DMCUB feature version: %u, fw version: 0x%08x\n", 0,
+		   adev->dm.dmcub_fw_version);
+	drm_printf(p, "PSP TOC feature version: %u, fw version: 0x%08x\n",
+		   adev->psp.toc.feature_version, adev->psp.toc.fw_version);
+
+	version = adev->mes.kiq_version & AMDGPU_MES_VERSION_MASK;
+	feature = (adev->mes.kiq_version & AMDGPU_MES_FEAT_VERSION_MASK) >>
+		  AMDGPU_MES_FEAT_VERSION_SHIFT;
+	drm_printf(p, "MES_KIQ feature version: %u, fw version: 0x%08x\n",
+		   feature, version);
+
+	version = adev->mes.sched_version & AMDGPU_MES_VERSION_MASK;
+	feature = (adev->mes.sched_version & AMDGPU_MES_FEAT_VERSION_MASK) >>
+		  AMDGPU_MES_FEAT_VERSION_SHIFT;
+	drm_printf(p, "MES feature version: %u, fw version: 0x%08x\n", feature,
+		   version);
+
+	drm_printf(p, "VPE feature version: %u, fw version: 0x%08x\n",
+		   adev->vpe.feature_version, adev->vpe.fw_version);
+
+	drm_printf(p, "\nVBIOS Information\n");
+	drm_printf(p, "vbios name       : %s\n", ctx->name);
+	drm_printf(p, "vbios pn         : %s\n", ctx->vbios_pn);
+	drm_printf(p, "vbios version    : %d\n", ctx->version);
+	drm_printf(p, "vbios ver_str    : %s\n", ctx->vbios_ver_str);
+	drm_printf(p, "vbios date       : %s\n", ctx->date);
+}
+
+static ssize_t
+amdgpu_devcoredump_read(char *buffer, loff_t offset, size_t count,
+			void *data, size_t datalen)
+{
+	struct drm_printer p;
+	struct amdgpu_coredump_info *coredump = data;
+	struct drm_print_iterator iter;
+	struct amdgpu_vm_fault_info *fault_info;
+	struct amdgpu_ip_block *ip_block;
+	int ver;
+
+	iter.data = buffer;
+	iter.offset = 0;
+	iter.start = offset;
+	iter.remain = count;
+
+	p = drm_coredump_printer(&iter);
+
+	drm_printf(&p, "**** AMDGPU Device Coredump ****\n");
+	drm_printf(&p, "version: " AMDGPU_COREDUMP_VERSION "\n");
+	drm_printf(&p, "kernel: " UTS_RELEASE "\n");
+	drm_printf(&p, "module: " KBUILD_MODNAME "\n");
+	drm_printf(&p, "time: %lld.%09ld\n", coredump->reset_time.tv_sec,
+		   coredump->reset_time.tv_nsec);
+
+	if (coredump->reset_task_info.task.pid)
+		drm_printf(&p, "process_name: %s PID: %d\n",
+			   coredump->reset_task_info.process_name,
+			   coredump->reset_task_info.task.pid);
+
+	/* SOC Information */
+	drm_printf(&p, "\nSOC Information\n");
+	drm_printf(&p, "SOC Device id: %d\n", coredump->adev->pdev->device);
+	drm_printf(&p, "SOC PCI Revision id: %d\n", coredump->adev->pdev->revision);
+	drm_printf(&p, "SOC Family: %d\n", coredump->adev->family);
+	drm_printf(&p, "SOC Revision id: %d\n", coredump->adev->rev_id);
+	drm_printf(&p, "SOC External Revision id: %d\n", coredump->adev->external_rev_id);
+
+	/* Memory Information */
+	drm_printf(&p, "\nSOC Memory Information\n");
+	drm_printf(&p, "real vram size: %llu\n", coredump->adev->gmc.real_vram_size);
+	drm_printf(&p, "visible vram size: %llu\n", coredump->adev->gmc.visible_vram_size);
+	drm_printf(&p, "gtt size: %llu\n", coredump->adev->mman.gtt_mgr.manager.size);
+
+	/* GDS Config */
+	drm_printf(&p, "\nGDS Config\n");
+	drm_printf(&p, "gds: total size: %d\n", coredump->adev->gds.gds_size);
+	drm_printf(&p, "gds: compute partition size: %d\n", coredump->adev->gds.gds_size);
+	drm_printf(&p, "gds: gws per compute partition: %d\n", coredump->adev->gds.gws_size);
+	drm_printf(&p, "gds: os per compute partition: %d\n", coredump->adev->gds.oa_size);
+
+	/* HWIP Version Information */
+	drm_printf(&p, "\nHW IP Version Information\n");
+	for (int i = 1; i < MAX_HWIP; i++) {
+		for (int j = 0; j < HWIP_MAX_INSTANCE; j++) {
+			ver = coredump->adev->ip_versions[i][j];
+			if (ver)
+				drm_printf(&p, "HWIP: %s[%d][%d]: v%d.%d.%d.%d.%d\n",
+					   hw_ip_names[i], i, j,
+					   IP_VERSION_MAJ(ver),
+					   IP_VERSION_MIN(ver),
+					   IP_VERSION_REV(ver),
+					   IP_VERSION_VARIANT(ver),
+					   IP_VERSION_SUBREV(ver));
+		}
+	}
+
+	/* IP firmware information */
+	drm_printf(&p, "\nIP Firmwares\n");
+	amdgpu_devcoredump_fw_info(coredump->adev, &p);
+
+	if (coredump->ring) {
+		drm_printf(&p, "\nRing timed out details\n");
+		drm_printf(&p, "IP Type: %d Ring Name: %s\n",
+			   coredump->ring->funcs->type,
+			   coredump->ring->name);
+	}
+
+	/* Add page fault information */
+	fault_info = &coredump->adev->vm_manager.fault_info;
+	drm_printf(&p, "\n[%s] Page fault observed\n",
+		   fault_info->vmhub ? "mmhub" : "gfxhub");
+	drm_printf(&p, "Faulty page starting at address: 0x%016llx\n", fault_info->addr);
+	drm_printf(&p, "Protection fault status register: 0x%x\n\n", fault_info->status);
+
+	/* dump the ip state for each ip */
+	drm_printf(&p, "IP Dump\n");
+	for (int i = 0; i < coredump->adev->num_ip_blocks; i++) {
+		ip_block = &coredump->adev->ip_blocks[i];
+		if (ip_block->version->funcs->print_ip_state) {
+			drm_printf(&p, "IP: %s\n", ip_block->version->funcs->name);
+			ip_block->version->funcs->print_ip_state(ip_block, &p);
+			drm_printf(&p, "\n");
+		}
+	}
+
+	/* Add ring buffer information */
+	drm_printf(&p, "Ring buffer information\n");
+	for (int i = 0; i < coredump->adev->num_rings; i++) {
+		int j = 0;
+		struct amdgpu_ring *ring = coredump->adev->rings[i];
+
+		drm_printf(&p, "ring name: %s\n", ring->name);
+		drm_printf(&p, "Rptr: 0x%llx Wptr: 0x%llx RB mask: %x\n",
+			   amdgpu_ring_get_rptr(ring),
+			   amdgpu_ring_get_wptr(ring),
+			   ring->buf_mask);
+		drm_printf(&p, "Ring size in dwords: %d\n",
+			   ring->ring_size / 4);
+		drm_printf(&p, "Ring contents\n");
+		drm_printf(&p, "Offset \t Value\n");
+
+		while (j < ring->ring_size) {
+			drm_printf(&p, "0x%x \t 0x%x\n", j, ring->ring[j / 4]);
+			j += 4;
+		}
+	}
+
+	if (coredump->skip_vram_check)
+		drm_printf(&p, "VRAM lost check is skipped!\n");
+	else if (coredump->reset_vram_lost)
+		drm_printf(&p, "VRAM is lost due to GPU reset!\n");
+
+	return count - iter.remain;
+}
+
+static void amdgpu_devcoredump_free(void *data)
+{
+	kfree(data);
+}
+
+void amdgpu_coredump(struct amdgpu_device *adev, bool skip_vram_check,
+		     bool vram_lost, struct amdgpu_job *job)
+{
+	struct drm_device *dev = adev_to_drm(adev);
+	struct amdgpu_coredump_info *coredump;
+	struct drm_sched_job *s_job;
+
+	coredump = kzalloc(sizeof(*coredump), GFP_NOWAIT);
+
+	if (!coredump) {
+		DRM_ERROR("%s: failed to allocate memory for coredump\n", __func__);
+		return;
+	}
+
+	coredump->skip_vram_check = skip_vram_check;
+	coredump->reset_vram_lost = vram_lost;
+
+	if (job && job->pasid) {
+		struct amdgpu_task_info *ti;
+
+		ti = amdgpu_vm_get_task_info_pasid(adev, job->pasid);
+		if (ti) {
+			coredump->reset_task_info = *ti;
+			amdgpu_vm_put_task_info(ti);
+		}
+	}
+
+	if (job) {
+		s_job = &job->base;
+		coredump->ring = to_amdgpu_ring(s_job->sched);
+	}
+
+	coredump->adev = adev;
+
+	ktime_get_ts64(&coredump->reset_time);
+
+	dev_coredumpm(dev->dev, THIS_MODULE, coredump, 0, GFP_NOWAIT,
+		      amdgpu_devcoredump_read, amdgpu_devcoredump_free);
+
+	drm_info(dev, "AMDGPU device coredump file has been created\n");
+	drm_info(dev, "Check your /sys/class/drm/card%d/device/devcoredump/data\n",
+		 dev->primary->index);
+}
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.h
new file mode 100644
index 000000000000..ef9772c6bcc9
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.h
@@ -0,0 +1,46 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_DEV_COREDUMP_H__
+#define __AMDGPU_DEV_COREDUMP_H__
+
+#include "amdgpu.h"
+
+#ifdef CONFIG_DEV_COREDUMP
+
+#define AMDGPU_COREDUMP_VERSION "1"
+
+struct amdgpu_coredump_info {
+	struct amdgpu_device            *adev;
+	struct amdgpu_task_info         reset_task_info;
+	struct timespec64               reset_time;
+	bool                            skip_vram_check;
+	bool                            reset_vram_lost;
+	struct amdgpu_ring              *ring;
+};
+#endif
+
+void amdgpu_coredump(struct amdgpu_device *adev, bool skip_vram_check,
+		     bool vram_lost, struct amdgpu_job *job);
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index e630d918fefc..7a899fb4de29 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -25,13 +25,24 @@
  *          Alex Deucher
  *          Jerome Glisse
  */
+
+#include <linux/aperture.h>
+#include <linux/power_supply.h>
 #include <linux/kthread.h>
+#include <linux/module.h>
 #include <linux/console.h>
 #include <linux/slab.h>
-#include <linux/debugfs.h>
-#include <drm/drmP.h>
+#include <linux/iommu.h>
+#include <linux/pci.h>
+#include <linux/pci-p2pdma.h>
+#include <linux/apple-gmux.h>
+
+#include <drm/drm_atomic_helper.h>
+#include <drm/drm_client_event.h>
 #include <drm/drm_crtc_helper.h>
+#include <drm/drm_probe_helper.h>
 #include <drm/amdgpu_drm.h>
+#include <linux/device.h>
 #include <linux/vgaarb.h>
 #include <linux/vga_switcheroo.h>
 #include <linux/efi.h>
@@ -50,23 +61,55 @@
 #endif
 #include "vi.h"
 #include "soc15.h"
+#include "nv.h"
 #include "bif/bif_4_1_d.h"
-#include <linux/pci.h>
 #include <linux/firmware.h>
 #include "amdgpu_vf_error.h"
 
 #include "amdgpu_amdkfd.h"
+#include "amdgpu_pm.h"
+
+#include "amdgpu_xgmi.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_pmu.h"
+#include "amdgpu_fru_eeprom.h"
+#include "amdgpu_reset.h"
+#include "amdgpu_virt.h"
+#include "amdgpu_dev_coredump.h"
+
+#include <linux/suspend.h>
+#include <drm/task_barrier.h>
+#include <linux/pm_runtime.h>
+
+#include <drm/drm_drv.h>
+
+#if IS_ENABLED(CONFIG_X86)
+#include <asm/intel-family.h>
+#include <asm/cpu_device_id.h>
+#endif
 
 MODULE_FIRMWARE("amdgpu/vega10_gpu_info.bin");
+MODULE_FIRMWARE("amdgpu/vega12_gpu_info.bin");
 MODULE_FIRMWARE("amdgpu/raven_gpu_info.bin");
+MODULE_FIRMWARE("amdgpu/picasso_gpu_info.bin");
+MODULE_FIRMWARE("amdgpu/raven2_gpu_info.bin");
+MODULE_FIRMWARE("amdgpu/arcturus_gpu_info.bin");
+MODULE_FIRMWARE("amdgpu/navi12_gpu_info.bin");
+MODULE_FIRMWARE("amdgpu/cyan_skillfish_gpu_info.bin");
 
 #define AMDGPU_RESUME_MS		2000
+#define AMDGPU_MAX_RETRY_LIMIT		2
+#define AMDGPU_RETRY_SRIOV_RESET(r) ((r) == -EBUSY || (r) == -ETIMEDOUT || (r) == -EINVAL)
+#define AMDGPU_PCIE_INDEX_FALLBACK (0x38 >> 2)
+#define AMDGPU_PCIE_INDEX_HI_FALLBACK (0x44 >> 2)
+#define AMDGPU_PCIE_DATA_FALLBACK (0x3C >> 2)
 
-static int amdgpu_debugfs_regs_init(struct amdgpu_device *adev);
-static void amdgpu_debugfs_regs_cleanup(struct amdgpu_device *adev);
-static int amdgpu_debugfs_test_ib_ring_init(struct amdgpu_device *adev);
+#define AMDGPU_VBIOS_SKIP (1U << 0)
+#define AMDGPU_VBIOS_OPTIONAL (1U << 1)
 
-static const char *amdgpu_asic_name[] = {
+static const struct drm_driver amdgpu_kms_driver;
+
+const char *amdgpu_asic_name[] = {
 	"TAHITI",
 	"PITCAIRN",
 	"VERDE",
@@ -85,185 +128,1153 @@ static const char *amdgpu_asic_name[] = {
 	"POLARIS10",
 	"POLARIS11",
 	"POLARIS12",
+	"VEGAM",
 	"VEGA10",
+	"VEGA12",
+	"VEGA20",
 	"RAVEN",
+	"ARCTURUS",
+	"RENOIR",
+	"ALDEBARAN",
+	"NAVI10",
+	"CYAN_SKILLFISH",
+	"NAVI14",
+	"NAVI12",
+	"SIENNA_CICHLID",
+	"NAVY_FLOUNDER",
+	"VANGOGH",
+	"DIMGREY_CAVEFISH",
+	"BEIGE_GOBY",
+	"YELLOW_CARP",
+	"IP DISCOVERY",
 	"LAST",
 };
 
-bool amdgpu_device_is_px(struct drm_device *dev)
-{
-	struct amdgpu_device *adev = dev->dev_private;
+#define AMDGPU_IP_BLK_MASK_ALL GENMASK(AMD_IP_BLOCK_TYPE_NUM  - 1, 0)
+/*
+ * Default init level where all blocks are expected to be initialized. This is
+ * the level of initialization expected by default and also after a full reset
+ * of the device.
+ */
+struct amdgpu_init_level amdgpu_init_default = {
+	.level = AMDGPU_INIT_LEVEL_DEFAULT,
+	.hwini_ip_block_mask = AMDGPU_IP_BLK_MASK_ALL,
+};
 
-	if (adev->flags & AMD_IS_PX)
-		return true;
-	return false;
-}
+struct amdgpu_init_level amdgpu_init_recovery = {
+	.level = AMDGPU_INIT_LEVEL_RESET_RECOVERY,
+	.hwini_ip_block_mask = AMDGPU_IP_BLK_MASK_ALL,
+};
 
 /*
- * MMIO register access helper functions.
+ * Minimal blocks needed to be initialized before a XGMI hive can be reset. This
+ * is used for cases like reset on initialization where the entire hive needs to
+ * be reset before first use.
  */
-uint32_t amdgpu_mm_rreg(struct amdgpu_device *adev, uint32_t reg,
-			uint32_t acc_flags)
+struct amdgpu_init_level amdgpu_init_minimal_xgmi = {
+	.level = AMDGPU_INIT_LEVEL_MINIMAL_XGMI,
+	.hwini_ip_block_mask =
+		BIT(AMD_IP_BLOCK_TYPE_GMC) | BIT(AMD_IP_BLOCK_TYPE_SMC) |
+		BIT(AMD_IP_BLOCK_TYPE_COMMON) | BIT(AMD_IP_BLOCK_TYPE_IH) |
+		BIT(AMD_IP_BLOCK_TYPE_PSP)
+};
+
+static void amdgpu_device_load_switch_state(struct amdgpu_device *adev);
+
+static inline bool amdgpu_ip_member_of_hwini(struct amdgpu_device *adev,
+					     enum amd_ip_block_type block)
 {
-	uint32_t ret;
+	return (adev->init_lvl->hwini_ip_block_mask & (1U << block)) != 0;
+}
 
-	if (!(acc_flags & AMDGPU_REGS_NO_KIQ) && amdgpu_sriov_runtime(adev)) {
-		BUG_ON(in_interrupt());
-		return amdgpu_virt_kiq_rreg(adev, reg);
+void amdgpu_set_init_level(struct amdgpu_device *adev,
+			   enum amdgpu_init_lvl_id lvl)
+{
+	switch (lvl) {
+	case AMDGPU_INIT_LEVEL_MINIMAL_XGMI:
+		adev->init_lvl = &amdgpu_init_minimal_xgmi;
+		break;
+	case AMDGPU_INIT_LEVEL_RESET_RECOVERY:
+		adev->init_lvl = &amdgpu_init_recovery;
+		break;
+	case AMDGPU_INIT_LEVEL_DEFAULT:
+		fallthrough;
+	default:
+		adev->init_lvl = &amdgpu_init_default;
+		break;
 	}
+}
 
-	if ((reg * 4) < adev->rmmio_size && !(acc_flags & AMDGPU_REGS_IDX))
-		ret = readl(((void __iomem *)adev->rmmio) + (reg * 4));
-	else {
-		unsigned long flags;
+static inline void amdgpu_device_stop_pending_resets(struct amdgpu_device *adev);
+static int amdgpu_device_pm_notifier(struct notifier_block *nb, unsigned long mode,
+				     void *data);
 
-		spin_lock_irqsave(&adev->mmio_idx_lock, flags);
-		writel((reg * 4), ((void __iomem *)adev->rmmio) + (mmMM_INDEX * 4));
-		ret = readl(((void __iomem *)adev->rmmio) + (mmMM_DATA * 4));
-		spin_unlock_irqrestore(&adev->mmio_idx_lock, flags);
+/**
+ * DOC: pcie_replay_count
+ *
+ * The amdgpu driver provides a sysfs API for reporting the total number
+ * of PCIe replays (NAKs).
+ * The file pcie_replay_count is used for this and returns the total
+ * number of replays as a sum of the NAKs generated and NAKs received.
+ */
+
+static ssize_t amdgpu_device_get_pcie_replay_count(struct device *dev,
+		struct device_attribute *attr, char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	uint64_t cnt = amdgpu_asic_get_pcie_replay_count(adev);
+
+	return sysfs_emit(buf, "%llu\n", cnt);
+}
+
+static DEVICE_ATTR(pcie_replay_count, 0444,
+		amdgpu_device_get_pcie_replay_count, NULL);
+
+static int amdgpu_device_attr_sysfs_init(struct amdgpu_device *adev)
+{
+	int ret = 0;
+
+	if (amdgpu_nbio_is_replay_cnt_supported(adev))
+		ret = sysfs_create_file(&adev->dev->kobj,
+					&dev_attr_pcie_replay_count.attr);
+
+	return ret;
+}
+
+static void amdgpu_device_attr_sysfs_fini(struct amdgpu_device *adev)
+{
+	if (amdgpu_nbio_is_replay_cnt_supported(adev))
+		sysfs_remove_file(&adev->dev->kobj,
+				  &dev_attr_pcie_replay_count.attr);
+}
+
+static ssize_t amdgpu_sysfs_reg_state_get(struct file *f, struct kobject *kobj,
+					  const struct bin_attribute *attr, char *buf,
+					  loff_t ppos, size_t count)
+{
+	struct device *dev = kobj_to_dev(kobj);
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	ssize_t bytes_read;
+
+	switch (ppos) {
+	case AMDGPU_SYS_REG_STATE_XGMI:
+		bytes_read = amdgpu_asic_get_reg_state(
+			adev, AMDGPU_REG_STATE_TYPE_XGMI, buf, count);
+		break;
+	case AMDGPU_SYS_REG_STATE_WAFL:
+		bytes_read = amdgpu_asic_get_reg_state(
+			adev, AMDGPU_REG_STATE_TYPE_WAFL, buf, count);
+		break;
+	case AMDGPU_SYS_REG_STATE_PCIE:
+		bytes_read = amdgpu_asic_get_reg_state(
+			adev, AMDGPU_REG_STATE_TYPE_PCIE, buf, count);
+		break;
+	case AMDGPU_SYS_REG_STATE_USR:
+		bytes_read = amdgpu_asic_get_reg_state(
+			adev, AMDGPU_REG_STATE_TYPE_USR, buf, count);
+		break;
+	case AMDGPU_SYS_REG_STATE_USR_1:
+		bytes_read = amdgpu_asic_get_reg_state(
+			adev, AMDGPU_REG_STATE_TYPE_USR_1, buf, count);
+		break;
+	default:
+		return -EINVAL;
 	}
-	trace_amdgpu_mm_rreg(adev->pdev->device, reg, ret);
+
+	return bytes_read;
+}
+
+static const BIN_ATTR(reg_state, 0444, amdgpu_sysfs_reg_state_get, NULL,
+		      AMDGPU_SYS_REG_STATE_END);
+
+int amdgpu_reg_state_sysfs_init(struct amdgpu_device *adev)
+{
+	int ret;
+
+	if (!amdgpu_asic_get_reg_state_supported(adev))
+		return 0;
+
+	ret = sysfs_create_bin_file(&adev->dev->kobj, &bin_attr_reg_state);
+
 	return ret;
 }
 
-void amdgpu_mm_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v,
-		    uint32_t acc_flags)
+void amdgpu_reg_state_sysfs_fini(struct amdgpu_device *adev)
 {
-	trace_amdgpu_mm_wreg(adev->pdev->device, reg, v);
+	if (!amdgpu_asic_get_reg_state_supported(adev))
+		return;
+	sysfs_remove_bin_file(&adev->dev->kobj, &bin_attr_reg_state);
+}
 
-	if (adev->asic_type >= CHIP_VEGA10 && reg == 0) {
-		adev->last_mm_index = v;
+int amdgpu_ip_block_suspend(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	if (ip_block->version->funcs->suspend) {
+		r = ip_block->version->funcs->suspend(ip_block);
+		if (r) {
+			dev_err(ip_block->adev->dev,
+				"suspend of IP block <%s> failed %d\n",
+				ip_block->version->funcs->name, r);
+			return r;
+		}
 	}
 
-	if (!(acc_flags & AMDGPU_REGS_NO_KIQ) && amdgpu_sriov_runtime(adev)) {
-		BUG_ON(in_interrupt());
-		return amdgpu_virt_kiq_wreg(adev, reg, v);
+	ip_block->status.hw = false;
+	return 0;
+}
+
+int amdgpu_ip_block_resume(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	if (ip_block->version->funcs->resume) {
+		r = ip_block->version->funcs->resume(ip_block);
+		if (r) {
+			dev_err(ip_block->adev->dev,
+				"resume of IP block <%s> failed %d\n",
+				ip_block->version->funcs->name, r);
+			return r;
+		}
 	}
 
-	if ((reg * 4) < adev->rmmio_size && !(acc_flags & AMDGPU_REGS_IDX))
-		writel(v, ((void __iomem *)adev->rmmio) + (reg * 4));
-	else {
-		unsigned long flags;
+	ip_block->status.hw = true;
+	return 0;
+}
+
+/**
+ * DOC: board_info
+ *
+ * The amdgpu driver provides a sysfs API for giving board related information.
+ * It provides the form factor information in the format
+ *
+ *   type : form factor
+ *
+ * Possible form factor values
+ *
+ * - "cem"		- PCIE CEM card
+ * - "oam"		- Open Compute Accelerator Module
+ * - "unknown"	- Not known
+ *
+ */
+
+static ssize_t amdgpu_device_get_board_info(struct device *dev,
+					    struct device_attribute *attr,
+					    char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	enum amdgpu_pkg_type pkg_type = AMDGPU_PKG_TYPE_CEM;
+	const char *pkg;
 
-		spin_lock_irqsave(&adev->mmio_idx_lock, flags);
-		writel((reg * 4), ((void __iomem *)adev->rmmio) + (mmMM_INDEX * 4));
-		writel(v, ((void __iomem *)adev->rmmio) + (mmMM_DATA * 4));
-		spin_unlock_irqrestore(&adev->mmio_idx_lock, flags);
+	if (adev->smuio.funcs && adev->smuio.funcs->get_pkg_type)
+		pkg_type = adev->smuio.funcs->get_pkg_type(adev);
+
+	switch (pkg_type) {
+	case AMDGPU_PKG_TYPE_CEM:
+		pkg = "cem";
+		break;
+	case AMDGPU_PKG_TYPE_OAM:
+		pkg = "oam";
+		break;
+	default:
+		pkg = "unknown";
+		break;
 	}
 
-	if (adev->asic_type >= CHIP_VEGA10 && reg == 1 && adev->last_mm_index == 0x5702C) {
-		udelay(500);
+	return sysfs_emit(buf, "%s : %s\n", "type", pkg);
+}
+
+static DEVICE_ATTR(board_info, 0444, amdgpu_device_get_board_info, NULL);
+
+static struct attribute *amdgpu_board_attrs[] = {
+	&dev_attr_board_info.attr,
+	NULL,
+};
+
+static umode_t amdgpu_board_attrs_is_visible(struct kobject *kobj,
+					     struct attribute *attr, int n)
+{
+	struct device *dev = kobj_to_dev(kobj);
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	if (adev->flags & AMD_IS_APU)
+		return 0;
+
+	return attr->mode;
+}
+
+static const struct attribute_group amdgpu_board_attrs_group = {
+	.attrs = amdgpu_board_attrs,
+	.is_visible = amdgpu_board_attrs_is_visible
+};
+
+static void amdgpu_device_get_pcie_info(struct amdgpu_device *adev);
+
+/**
+ * amdgpu_device_supports_px - Is the device a dGPU with ATPX power control
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Returns true if the device is a dGPU with ATPX power control,
+ * otherwise return false.
+ */
+bool amdgpu_device_supports_px(struct amdgpu_device *adev)
+{
+	if ((adev->flags & AMD_IS_PX) && !amdgpu_is_atpx_hybrid())
+		return true;
+	return false;
+}
+
+/**
+ * amdgpu_device_supports_boco - Is the device a dGPU with ACPI power resources
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Returns true if the device is a dGPU with ACPI power control,
+ * otherwise return false.
+ */
+bool amdgpu_device_supports_boco(struct amdgpu_device *adev)
+{
+	if (!IS_ENABLED(CONFIG_HOTPLUG_PCI_PCIE))
+		return false;
+
+	if (adev->has_pr3 ||
+	    ((adev->flags & AMD_IS_PX) && amdgpu_is_atpx_hybrid()))
+		return true;
+	return false;
+}
+
+/**
+ * amdgpu_device_supports_baco - Does the device support BACO
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Return:
+ * 1 if the device supports BACO;
+ * 3 if the device supports MACO (only works if BACO is supported)
+ * otherwise return 0.
+ */
+int amdgpu_device_supports_baco(struct amdgpu_device *adev)
+{
+	return amdgpu_asic_supports_baco(adev);
+}
+
+void amdgpu_device_detect_runtime_pm_mode(struct amdgpu_device *adev)
+{
+	int bamaco_support;
+
+	adev->pm.rpm_mode = AMDGPU_RUNPM_NONE;
+	bamaco_support = amdgpu_device_supports_baco(adev);
+
+	switch (amdgpu_runtime_pm) {
+	case 2:
+		if (bamaco_support & MACO_SUPPORT) {
+			adev->pm.rpm_mode = AMDGPU_RUNPM_BAMACO;
+			dev_info(adev->dev, "Forcing BAMACO for runtime pm\n");
+		} else if (bamaco_support == BACO_SUPPORT) {
+			adev->pm.rpm_mode = AMDGPU_RUNPM_BACO;
+			dev_info(adev->dev, "Requested mode BAMACO not available,fallback to use BACO\n");
+		}
+		break;
+	case 1:
+		if (bamaco_support & BACO_SUPPORT) {
+			adev->pm.rpm_mode = AMDGPU_RUNPM_BACO;
+			dev_info(adev->dev, "Forcing BACO for runtime pm\n");
+		}
+		break;
+	case -1:
+	case -2:
+		if (amdgpu_device_supports_px(adev)) {
+			/* enable PX as runtime mode */
+			adev->pm.rpm_mode = AMDGPU_RUNPM_PX;
+			dev_info(adev->dev, "Using ATPX for runtime pm\n");
+		} else if (amdgpu_device_supports_boco(adev)) {
+			/* enable boco as runtime mode */
+			adev->pm.rpm_mode = AMDGPU_RUNPM_BOCO;
+			dev_info(adev->dev, "Using BOCO for runtime pm\n");
+		} else {
+			if (!bamaco_support)
+				goto no_runtime_pm;
+
+			switch (adev->asic_type) {
+			case CHIP_VEGA20:
+			case CHIP_ARCTURUS:
+				/* BACO are not supported on vega20 and arctrus */
+				break;
+			case CHIP_VEGA10:
+				/* enable BACO as runpm mode if noretry=0 */
+				if (!adev->gmc.noretry && !amdgpu_passthrough(adev))
+					adev->pm.rpm_mode = AMDGPU_RUNPM_BACO;
+				break;
+			default:
+				/* enable BACO as runpm mode on CI+ */
+				if (!amdgpu_passthrough(adev))
+					adev->pm.rpm_mode = AMDGPU_RUNPM_BACO;
+				break;
+			}
+
+			if (adev->pm.rpm_mode == AMDGPU_RUNPM_BACO) {
+				if (bamaco_support & MACO_SUPPORT) {
+					adev->pm.rpm_mode = AMDGPU_RUNPM_BAMACO;
+					dev_info(adev->dev, "Using BAMACO for runtime pm\n");
+				} else {
+					dev_info(adev->dev, "Using BACO for runtime pm\n");
+				}
+			}
+		}
+		break;
+	case 0:
+		dev_info(adev->dev, "runtime pm is manually disabled\n");
+		break;
+	default:
+		break;
 	}
+
+no_runtime_pm:
+	if (adev->pm.rpm_mode == AMDGPU_RUNPM_NONE)
+		dev_info(adev->dev, "Runtime PM not available\n");
+}
+/**
+ * amdgpu_device_supports_smart_shift - Is the device dGPU with
+ * smart shift support
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Returns true if the device is a dGPU with Smart Shift support,
+ * otherwise returns false.
+ */
+bool amdgpu_device_supports_smart_shift(struct amdgpu_device *adev)
+{
+	return (amdgpu_device_supports_boco(adev) &&
+		amdgpu_acpi_is_power_shift_control_supported());
 }
 
-u32 amdgpu_io_rreg(struct amdgpu_device *adev, u32 reg)
+/*
+ * VRAM access helper functions
+ */
+
+/**
+ * amdgpu_device_mm_access - access vram by MM_INDEX/MM_DATA
+ *
+ * @adev: amdgpu_device pointer
+ * @pos: offset of the buffer in vram
+ * @buf: virtual address of the buffer in system memory
+ * @size: read/write size, sizeof(@buf) must > @size
+ * @write: true - write to vram, otherwise - read from vram
+ */
+void amdgpu_device_mm_access(struct amdgpu_device *adev, loff_t pos,
+			     void *buf, size_t size, bool write)
 {
-	if ((reg * 4) < adev->rio_mem_size)
-		return ioread32(adev->rio_mem + (reg * 4));
-	else {
-		iowrite32((reg * 4), adev->rio_mem + (mmMM_INDEX * 4));
-		return ioread32(adev->rio_mem + (mmMM_DATA * 4));
+	unsigned long flags;
+	uint32_t hi = ~0, tmp = 0;
+	uint32_t *data = buf;
+	uint64_t last;
+	int idx;
+
+	if (!drm_dev_enter(adev_to_drm(adev), &idx))
+		return;
+
+	BUG_ON(!IS_ALIGNED(pos, 4) || !IS_ALIGNED(size, 4));
+
+	spin_lock_irqsave(&adev->mmio_idx_lock, flags);
+	for (last = pos + size; pos < last; pos += 4) {
+		tmp = pos >> 31;
+
+		WREG32_NO_KIQ(mmMM_INDEX, ((uint32_t)pos) | 0x80000000);
+		if (tmp != hi) {
+			WREG32_NO_KIQ(mmMM_INDEX_HI, tmp);
+			hi = tmp;
+		}
+		if (write)
+			WREG32_NO_KIQ(mmMM_DATA, *data++);
+		else
+			*data++ = RREG32_NO_KIQ(mmMM_DATA);
 	}
+
+	spin_unlock_irqrestore(&adev->mmio_idx_lock, flags);
+	drm_dev_exit(idx);
 }
 
-void amdgpu_io_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+/**
+ * amdgpu_device_aper_access - access vram by vram aperture
+ *
+ * @adev: amdgpu_device pointer
+ * @pos: offset of the buffer in vram
+ * @buf: virtual address of the buffer in system memory
+ * @size: read/write size, sizeof(@buf) must > @size
+ * @write: true - write to vram, otherwise - read from vram
+ *
+ * The return value means how many bytes have been transferred.
+ */
+size_t amdgpu_device_aper_access(struct amdgpu_device *adev, loff_t pos,
+				 void *buf, size_t size, bool write)
 {
-	if (adev->asic_type >= CHIP_VEGA10 && reg == 0) {
-		adev->last_mm_index = v;
+#ifdef CONFIG_64BIT
+	void __iomem *addr;
+	size_t count = 0;
+	uint64_t last;
+
+	if (!adev->mman.aper_base_kaddr)
+		return 0;
+
+	last = min(pos + size, adev->gmc.visible_vram_size);
+	if (last > pos) {
+		addr = adev->mman.aper_base_kaddr + pos;
+		count = last - pos;
+
+		if (write) {
+			memcpy_toio(addr, buf, count);
+			/* Make sure HDP write cache flush happens without any reordering
+			 * after the system memory contents are sent over PCIe device
+			 */
+			mb();
+			amdgpu_device_flush_hdp(adev, NULL);
+		} else {
+			amdgpu_device_invalidate_hdp(adev, NULL);
+			/* Make sure HDP read cache is invalidated before issuing a read
+			 * to the PCIe device
+			 */
+			mb();
+			memcpy_fromio(buf, addr, count);
+		}
+
 	}
 
-	if ((reg * 4) < adev->rio_mem_size)
-		iowrite32(v, adev->rio_mem + (reg * 4));
-	else {
-		iowrite32((reg * 4), adev->rio_mem + (mmMM_INDEX * 4));
-		iowrite32(v, adev->rio_mem + (mmMM_DATA * 4));
+	return count;
+#else
+	return 0;
+#endif
+}
+
+/**
+ * amdgpu_device_vram_access - read/write a buffer in vram
+ *
+ * @adev: amdgpu_device pointer
+ * @pos: offset of the buffer in vram
+ * @buf: virtual address of the buffer in system memory
+ * @size: read/write size, sizeof(@buf) must > @size
+ * @write: true - write to vram, otherwise - read from vram
+ */
+void amdgpu_device_vram_access(struct amdgpu_device *adev, loff_t pos,
+			       void *buf, size_t size, bool write)
+{
+	size_t count;
+
+	/* try to using vram apreature to access vram first */
+	count = amdgpu_device_aper_access(adev, pos, buf, size, write);
+	size -= count;
+	if (size) {
+		/* using MM to access rest vram */
+		pos += count;
+		buf += count;
+		amdgpu_device_mm_access(adev, pos, buf, size, write);
 	}
+}
+
+/*
+ * register access helper functions.
+ */
 
-	if (adev->asic_type >= CHIP_VEGA10 && reg == 1 && adev->last_mm_index == 0x5702C) {
-		udelay(500);
+/* Check if hw access should be skipped because of hotplug or device error */
+bool amdgpu_device_skip_hw_access(struct amdgpu_device *adev)
+{
+	if (adev->no_hw_access)
+		return true;
+
+#ifdef CONFIG_LOCKDEP
+	/*
+	 * This is a bit complicated to understand, so worth a comment. What we assert
+	 * here is that the GPU reset is not running on another thread in parallel.
+	 *
+	 * For this we trylock the read side of the reset semaphore, if that succeeds
+	 * we know that the reset is not running in parallel.
+	 *
+	 * If the trylock fails we assert that we are either already holding the read
+	 * side of the lock or are the reset thread itself and hold the write side of
+	 * the lock.
+	 */
+	if (in_task()) {
+		if (down_read_trylock(&adev->reset_domain->sem))
+			up_read(&adev->reset_domain->sem);
+		else
+			lockdep_assert_held(&adev->reset_domain->sem);
 	}
+#endif
+	return false;
 }
 
 /**
- * amdgpu_mm_rdoorbell - read a doorbell dword
+ * amdgpu_device_rreg - read a memory mapped IO or indirect register
  *
  * @adev: amdgpu_device pointer
- * @index: doorbell index
+ * @reg: dword aligned register offset
+ * @acc_flags: access flags which require special behavior
  *
- * Returns the value in the doorbell aperture at the
- * requested doorbell index (CIK).
+ * Returns the 32 bit value from the offset specified.
  */
-u32 amdgpu_mm_rdoorbell(struct amdgpu_device *adev, u32 index)
+uint32_t amdgpu_device_rreg(struct amdgpu_device *adev,
+			    uint32_t reg, uint32_t acc_flags)
 {
-	if (index < adev->doorbell.num_doorbells) {
-		return readl(adev->doorbell.ptr + index);
+	uint32_t ret;
+
+	if (amdgpu_device_skip_hw_access(adev))
+		return 0;
+
+	if ((reg * 4) < adev->rmmio_size) {
+		if (!(acc_flags & AMDGPU_REGS_NO_KIQ) &&
+		    amdgpu_sriov_runtime(adev) &&
+		    down_read_trylock(&adev->reset_domain->sem)) {
+			ret = amdgpu_kiq_rreg(adev, reg, 0);
+			up_read(&adev->reset_domain->sem);
+		} else {
+			ret = readl(((void __iomem *)adev->rmmio) + (reg * 4));
+		}
 	} else {
-		DRM_ERROR("reading beyond doorbell aperture: 0x%08x!\n", index);
+		ret = adev->pcie_rreg(adev, reg * 4);
+	}
+
+	trace_amdgpu_device_rreg(adev->pdev->device, reg, ret);
+
+	return ret;
+}
+
+/*
+ * MMIO register read with bytes helper functions
+ * @offset:bytes offset from MMIO start
+ */
+
+/**
+ * amdgpu_mm_rreg8 - read a memory mapped IO register
+ *
+ * @adev: amdgpu_device pointer
+ * @offset: byte aligned register offset
+ *
+ * Returns the 8 bit value from the offset specified.
+ */
+uint8_t amdgpu_mm_rreg8(struct amdgpu_device *adev, uint32_t offset)
+{
+	if (amdgpu_device_skip_hw_access(adev))
+		return 0;
+
+	if (offset < adev->rmmio_size)
+		return (readb(adev->rmmio + offset));
+	BUG();
+}
+
+
+/**
+ * amdgpu_device_xcc_rreg - read a memory mapped IO or indirect register with specific XCC
+ *
+ * @adev: amdgpu_device pointer
+ * @reg: dword aligned register offset
+ * @acc_flags: access flags which require special behavior
+ * @xcc_id: xcc accelerated compute core id
+ *
+ * Returns the 32 bit value from the offset specified.
+ */
+uint32_t amdgpu_device_xcc_rreg(struct amdgpu_device *adev,
+				uint32_t reg, uint32_t acc_flags,
+				uint32_t xcc_id)
+{
+	uint32_t ret, rlcg_flag;
+
+	if (amdgpu_device_skip_hw_access(adev))
 		return 0;
+
+	if ((reg * 4) < adev->rmmio_size) {
+		if (amdgpu_sriov_vf(adev) &&
+		    !amdgpu_sriov_runtime(adev) &&
+		    adev->gfx.rlc.rlcg_reg_access_supported &&
+		    amdgpu_virt_get_rlcg_reg_access_flag(adev, acc_flags,
+							 GC_HWIP, false,
+							 &rlcg_flag)) {
+			ret = amdgpu_virt_rlcg_reg_rw(adev, reg, 0, rlcg_flag, GET_INST(GC, xcc_id));
+		} else if (!(acc_flags & AMDGPU_REGS_NO_KIQ) &&
+		    amdgpu_sriov_runtime(adev) &&
+		    down_read_trylock(&adev->reset_domain->sem)) {
+			ret = amdgpu_kiq_rreg(adev, reg, xcc_id);
+			up_read(&adev->reset_domain->sem);
+		} else {
+			ret = readl(((void __iomem *)adev->rmmio) + (reg * 4));
+		}
+	} else {
+		ret = adev->pcie_rreg(adev, reg * 4);
 	}
+
+	return ret;
 }
 
+/*
+ * MMIO register write with bytes helper functions
+ * @offset:bytes offset from MMIO start
+ * @value: the value want to be written to the register
+ */
+
 /**
- * amdgpu_mm_wdoorbell - write a doorbell dword
+ * amdgpu_mm_wreg8 - read a memory mapped IO register
  *
  * @adev: amdgpu_device pointer
- * @index: doorbell index
+ * @offset: byte aligned register offset
+ * @value: 8 bit value to write
+ *
+ * Writes the value specified to the offset specified.
+ */
+void amdgpu_mm_wreg8(struct amdgpu_device *adev, uint32_t offset, uint8_t value)
+{
+	if (amdgpu_device_skip_hw_access(adev))
+		return;
+
+	if (offset < adev->rmmio_size)
+		writeb(value, adev->rmmio + offset);
+	else
+		BUG();
+}
+
+/**
+ * amdgpu_device_wreg - write to a memory mapped IO or indirect register
+ *
+ * @adev: amdgpu_device pointer
+ * @reg: dword aligned register offset
+ * @v: 32 bit value to write to the register
+ * @acc_flags: access flags which require special behavior
+ *
+ * Writes the value specified to the offset specified.
+ */
+void amdgpu_device_wreg(struct amdgpu_device *adev,
+			uint32_t reg, uint32_t v,
+			uint32_t acc_flags)
+{
+	if (amdgpu_device_skip_hw_access(adev))
+		return;
+
+	if ((reg * 4) < adev->rmmio_size) {
+		if (!(acc_flags & AMDGPU_REGS_NO_KIQ) &&
+		    amdgpu_sriov_runtime(adev) &&
+		    down_read_trylock(&adev->reset_domain->sem)) {
+			amdgpu_kiq_wreg(adev, reg, v, 0);
+			up_read(&adev->reset_domain->sem);
+		} else {
+			writel(v, ((void __iomem *)adev->rmmio) + (reg * 4));
+		}
+	} else {
+		adev->pcie_wreg(adev, reg * 4, v);
+	}
+
+	trace_amdgpu_device_wreg(adev->pdev->device, reg, v);
+}
+
+/**
+ * amdgpu_mm_wreg_mmio_rlc -  write register either with direct/indirect mmio or with RLC path if in range
+ *
+ * @adev: amdgpu_device pointer
+ * @reg: mmio/rlc register
  * @v: value to write
+ * @xcc_id: xcc accelerated compute core id
  *
- * Writes @v to the doorbell aperture at the
- * requested doorbell index (CIK).
+ * this function is invoked only for the debugfs register access
  */
-void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v)
+void amdgpu_mm_wreg_mmio_rlc(struct amdgpu_device *adev,
+			     uint32_t reg, uint32_t v,
+			     uint32_t xcc_id)
 {
-	if (index < adev->doorbell.num_doorbells) {
-		writel(v, adev->doorbell.ptr + index);
+	if (amdgpu_device_skip_hw_access(adev))
+		return;
+
+	if (amdgpu_sriov_fullaccess(adev) &&
+	    adev->gfx.rlc.funcs &&
+	    adev->gfx.rlc.funcs->is_rlcg_access_range) {
+		if (adev->gfx.rlc.funcs->is_rlcg_access_range(adev, reg))
+			return amdgpu_sriov_wreg(adev, reg, v, 0, 0, xcc_id);
+	} else if ((reg * 4) >= adev->rmmio_size) {
+		adev->pcie_wreg(adev, reg * 4, v);
 	} else {
-		DRM_ERROR("writing beyond doorbell aperture: 0x%08x!\n", index);
+		writel(v, ((void __iomem *)adev->rmmio) + (reg * 4));
 	}
 }
 
 /**
- * amdgpu_mm_rdoorbell64 - read a doorbell Qword
+ * amdgpu_device_xcc_wreg - write to a memory mapped IO or indirect register with specific XCC
  *
  * @adev: amdgpu_device pointer
- * @index: doorbell index
+ * @reg: dword aligned register offset
+ * @v: 32 bit value to write to the register
+ * @acc_flags: access flags which require special behavior
+ * @xcc_id: xcc accelerated compute core id
  *
- * Returns the value in the doorbell aperture at the
- * requested doorbell index (VEGA10+).
+ * Writes the value specified to the offset specified.
  */
-u64 amdgpu_mm_rdoorbell64(struct amdgpu_device *adev, u32 index)
+void amdgpu_device_xcc_wreg(struct amdgpu_device *adev,
+			uint32_t reg, uint32_t v,
+			uint32_t acc_flags, uint32_t xcc_id)
 {
-	if (index < adev->doorbell.num_doorbells) {
-		return atomic64_read((atomic64_t *)(adev->doorbell.ptr + index));
+	uint32_t rlcg_flag;
+
+	if (amdgpu_device_skip_hw_access(adev))
+		return;
+
+	if ((reg * 4) < adev->rmmio_size) {
+		if (amdgpu_sriov_vf(adev) &&
+		    !amdgpu_sriov_runtime(adev) &&
+		    adev->gfx.rlc.rlcg_reg_access_supported &&
+		    amdgpu_virt_get_rlcg_reg_access_flag(adev, acc_flags,
+							 GC_HWIP, true,
+							 &rlcg_flag)) {
+			amdgpu_virt_rlcg_reg_rw(adev, reg, v, rlcg_flag, GET_INST(GC, xcc_id));
+		} else if (!(acc_flags & AMDGPU_REGS_NO_KIQ) &&
+		    amdgpu_sriov_runtime(adev) &&
+		    down_read_trylock(&adev->reset_domain->sem)) {
+			amdgpu_kiq_wreg(adev, reg, v, xcc_id);
+			up_read(&adev->reset_domain->sem);
+		} else {
+			writel(v, ((void __iomem *)adev->rmmio) + (reg * 4));
+		}
 	} else {
-		DRM_ERROR("reading beyond doorbell aperture: 0x%08x!\n", index);
-		return 0;
+		adev->pcie_wreg(adev, reg * 4, v);
 	}
 }
 
 /**
- * amdgpu_mm_wdoorbell64 - write a doorbell Qword
+ * amdgpu_device_indirect_rreg - read an indirect register
  *
  * @adev: amdgpu_device pointer
- * @index: doorbell index
- * @v: value to write
+ * @reg_addr: indirect register address to read from
  *
- * Writes @v to the doorbell aperture at the
- * requested doorbell index (VEGA10+).
+ * Returns the value of indirect register @reg_addr
  */
-void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v)
+u32 amdgpu_device_indirect_rreg(struct amdgpu_device *adev,
+				u32 reg_addr)
 {
-	if (index < adev->doorbell.num_doorbells) {
-		atomic64_set((atomic64_t *)(adev->doorbell.ptr + index), v);
+	unsigned long flags, pcie_index, pcie_data;
+	void __iomem *pcie_index_offset;
+	void __iomem *pcie_data_offset;
+	u32 r;
+
+	pcie_index = adev->nbio.funcs->get_pcie_index_offset(adev);
+	pcie_data = adev->nbio.funcs->get_pcie_data_offset(adev);
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	pcie_index_offset = (void __iomem *)adev->rmmio + pcie_index * 4;
+	pcie_data_offset = (void __iomem *)adev->rmmio + pcie_data * 4;
+
+	writel(reg_addr, pcie_index_offset);
+	readl(pcie_index_offset);
+	r = readl(pcie_data_offset);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+
+	return r;
+}
+
+u32 amdgpu_device_indirect_rreg_ext(struct amdgpu_device *adev,
+				    u64 reg_addr)
+{
+	unsigned long flags, pcie_index, pcie_index_hi, pcie_data;
+	u32 r;
+	void __iomem *pcie_index_offset;
+	void __iomem *pcie_index_hi_offset;
+	void __iomem *pcie_data_offset;
+
+	if (unlikely(!adev->nbio.funcs)) {
+		pcie_index = AMDGPU_PCIE_INDEX_FALLBACK;
+		pcie_data = AMDGPU_PCIE_DATA_FALLBACK;
+	} else {
+		pcie_index = adev->nbio.funcs->get_pcie_index_offset(adev);
+		pcie_data = adev->nbio.funcs->get_pcie_data_offset(adev);
+	}
+
+	if (reg_addr >> 32) {
+		if (unlikely(!adev->nbio.funcs))
+			pcie_index_hi = AMDGPU_PCIE_INDEX_HI_FALLBACK;
+		else
+			pcie_index_hi = adev->nbio.funcs->get_pcie_index_hi_offset(adev);
 	} else {
-		DRM_ERROR("writing beyond doorbell aperture: 0x%08x!\n", index);
+		pcie_index_hi = 0;
 	}
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	pcie_index_offset = (void __iomem *)adev->rmmio + pcie_index * 4;
+	pcie_data_offset = (void __iomem *)adev->rmmio + pcie_data * 4;
+	if (pcie_index_hi != 0)
+		pcie_index_hi_offset = (void __iomem *)adev->rmmio +
+				pcie_index_hi * 4;
+
+	writel(reg_addr, pcie_index_offset);
+	readl(pcie_index_offset);
+	if (pcie_index_hi != 0) {
+		writel((reg_addr >> 32) & 0xff, pcie_index_hi_offset);
+		readl(pcie_index_hi_offset);
+	}
+	r = readl(pcie_data_offset);
+
+	/* clear the high bits */
+	if (pcie_index_hi != 0) {
+		writel(0, pcie_index_hi_offset);
+		readl(pcie_index_hi_offset);
+	}
+
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+
+	return r;
+}
+
+/**
+ * amdgpu_device_indirect_rreg64 - read a 64bits indirect register
+ *
+ * @adev: amdgpu_device pointer
+ * @reg_addr: indirect register address to read from
+ *
+ * Returns the value of indirect register @reg_addr
+ */
+u64 amdgpu_device_indirect_rreg64(struct amdgpu_device *adev,
+				  u32 reg_addr)
+{
+	unsigned long flags, pcie_index, pcie_data;
+	void __iomem *pcie_index_offset;
+	void __iomem *pcie_data_offset;
+	u64 r;
+
+	pcie_index = adev->nbio.funcs->get_pcie_index_offset(adev);
+	pcie_data = adev->nbio.funcs->get_pcie_data_offset(adev);
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	pcie_index_offset = (void __iomem *)adev->rmmio + pcie_index * 4;
+	pcie_data_offset = (void __iomem *)adev->rmmio + pcie_data * 4;
+
+	/* read low 32 bits */
+	writel(reg_addr, pcie_index_offset);
+	readl(pcie_index_offset);
+	r = readl(pcie_data_offset);
+	/* read high 32 bits */
+	writel(reg_addr + 4, pcie_index_offset);
+	readl(pcie_index_offset);
+	r |= ((u64)readl(pcie_data_offset) << 32);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+
+	return r;
+}
+
+u64 amdgpu_device_indirect_rreg64_ext(struct amdgpu_device *adev,
+				  u64 reg_addr)
+{
+	unsigned long flags, pcie_index, pcie_data;
+	unsigned long pcie_index_hi = 0;
+	void __iomem *pcie_index_offset;
+	void __iomem *pcie_index_hi_offset;
+	void __iomem *pcie_data_offset;
+	u64 r;
+
+	pcie_index = adev->nbio.funcs->get_pcie_index_offset(adev);
+	pcie_data = adev->nbio.funcs->get_pcie_data_offset(adev);
+	if ((reg_addr >> 32) && (adev->nbio.funcs->get_pcie_index_hi_offset))
+		pcie_index_hi = adev->nbio.funcs->get_pcie_index_hi_offset(adev);
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	pcie_index_offset = (void __iomem *)adev->rmmio + pcie_index * 4;
+	pcie_data_offset = (void __iomem *)adev->rmmio + pcie_data * 4;
+	if (pcie_index_hi != 0)
+		pcie_index_hi_offset = (void __iomem *)adev->rmmio +
+			pcie_index_hi * 4;
+
+	/* read low 32 bits */
+	writel(reg_addr, pcie_index_offset);
+	readl(pcie_index_offset);
+	if (pcie_index_hi != 0) {
+		writel((reg_addr >> 32) & 0xff, pcie_index_hi_offset);
+		readl(pcie_index_hi_offset);
+	}
+	r = readl(pcie_data_offset);
+	/* read high 32 bits */
+	writel(reg_addr + 4, pcie_index_offset);
+	readl(pcie_index_offset);
+	if (pcie_index_hi != 0) {
+		writel((reg_addr >> 32) & 0xff, pcie_index_hi_offset);
+		readl(pcie_index_hi_offset);
+	}
+	r |= ((u64)readl(pcie_data_offset) << 32);
+
+	/* clear the high bits */
+	if (pcie_index_hi != 0) {
+		writel(0, pcie_index_hi_offset);
+		readl(pcie_index_hi_offset);
+	}
+
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+
+	return r;
+}
+
+/**
+ * amdgpu_device_indirect_wreg - write an indirect register address
+ *
+ * @adev: amdgpu_device pointer
+ * @reg_addr: indirect register offset
+ * @reg_data: indirect register data
+ *
+ */
+void amdgpu_device_indirect_wreg(struct amdgpu_device *adev,
+				 u32 reg_addr, u32 reg_data)
+{
+	unsigned long flags, pcie_index, pcie_data;
+	void __iomem *pcie_index_offset;
+	void __iomem *pcie_data_offset;
+
+	pcie_index = adev->nbio.funcs->get_pcie_index_offset(adev);
+	pcie_data = adev->nbio.funcs->get_pcie_data_offset(adev);
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	pcie_index_offset = (void __iomem *)adev->rmmio + pcie_index * 4;
+	pcie_data_offset = (void __iomem *)adev->rmmio + pcie_data * 4;
+
+	writel(reg_addr, pcie_index_offset);
+	readl(pcie_index_offset);
+	writel(reg_data, pcie_data_offset);
+	readl(pcie_data_offset);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+}
+
+void amdgpu_device_indirect_wreg_ext(struct amdgpu_device *adev,
+				     u64 reg_addr, u32 reg_data)
+{
+	unsigned long flags, pcie_index, pcie_index_hi, pcie_data;
+	void __iomem *pcie_index_offset;
+	void __iomem *pcie_index_hi_offset;
+	void __iomem *pcie_data_offset;
+
+	pcie_index = adev->nbio.funcs->get_pcie_index_offset(adev);
+	pcie_data = adev->nbio.funcs->get_pcie_data_offset(adev);
+	if ((reg_addr >> 32) && (adev->nbio.funcs->get_pcie_index_hi_offset))
+		pcie_index_hi = adev->nbio.funcs->get_pcie_index_hi_offset(adev);
+	else
+		pcie_index_hi = 0;
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	pcie_index_offset = (void __iomem *)adev->rmmio + pcie_index * 4;
+	pcie_data_offset = (void __iomem *)adev->rmmio + pcie_data * 4;
+	if (pcie_index_hi != 0)
+		pcie_index_hi_offset = (void __iomem *)adev->rmmio +
+				pcie_index_hi * 4;
+
+	writel(reg_addr, pcie_index_offset);
+	readl(pcie_index_offset);
+	if (pcie_index_hi != 0) {
+		writel((reg_addr >> 32) & 0xff, pcie_index_hi_offset);
+		readl(pcie_index_hi_offset);
+	}
+	writel(reg_data, pcie_data_offset);
+	readl(pcie_data_offset);
+
+	/* clear the high bits */
+	if (pcie_index_hi != 0) {
+		writel(0, pcie_index_hi_offset);
+		readl(pcie_index_hi_offset);
+	}
+
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+}
+
+/**
+ * amdgpu_device_indirect_wreg64 - write a 64bits indirect register address
+ *
+ * @adev: amdgpu_device pointer
+ * @reg_addr: indirect register offset
+ * @reg_data: indirect register data
+ *
+ */
+void amdgpu_device_indirect_wreg64(struct amdgpu_device *adev,
+				   u32 reg_addr, u64 reg_data)
+{
+	unsigned long flags, pcie_index, pcie_data;
+	void __iomem *pcie_index_offset;
+	void __iomem *pcie_data_offset;
+
+	pcie_index = adev->nbio.funcs->get_pcie_index_offset(adev);
+	pcie_data = adev->nbio.funcs->get_pcie_data_offset(adev);
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	pcie_index_offset = (void __iomem *)adev->rmmio + pcie_index * 4;
+	pcie_data_offset = (void __iomem *)adev->rmmio + pcie_data * 4;
+
+	/* write low 32 bits */
+	writel(reg_addr, pcie_index_offset);
+	readl(pcie_index_offset);
+	writel((u32)(reg_data & 0xffffffffULL), pcie_data_offset);
+	readl(pcie_data_offset);
+	/* write high 32 bits */
+	writel(reg_addr + 4, pcie_index_offset);
+	readl(pcie_index_offset);
+	writel((u32)(reg_data >> 32), pcie_data_offset);
+	readl(pcie_data_offset);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+}
+
+void amdgpu_device_indirect_wreg64_ext(struct amdgpu_device *adev,
+				   u64 reg_addr, u64 reg_data)
+{
+	unsigned long flags, pcie_index, pcie_data;
+	unsigned long pcie_index_hi = 0;
+	void __iomem *pcie_index_offset;
+	void __iomem *pcie_index_hi_offset;
+	void __iomem *pcie_data_offset;
+
+	pcie_index = adev->nbio.funcs->get_pcie_index_offset(adev);
+	pcie_data = adev->nbio.funcs->get_pcie_data_offset(adev);
+	if ((reg_addr >> 32) && (adev->nbio.funcs->get_pcie_index_hi_offset))
+		pcie_index_hi = adev->nbio.funcs->get_pcie_index_hi_offset(adev);
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	pcie_index_offset = (void __iomem *)adev->rmmio + pcie_index * 4;
+	pcie_data_offset = (void __iomem *)adev->rmmio + pcie_data * 4;
+	if (pcie_index_hi != 0)
+		pcie_index_hi_offset = (void __iomem *)adev->rmmio +
+				pcie_index_hi * 4;
+
+	/* write low 32 bits */
+	writel(reg_addr, pcie_index_offset);
+	readl(pcie_index_offset);
+	if (pcie_index_hi != 0) {
+		writel((reg_addr >> 32) & 0xff, pcie_index_hi_offset);
+		readl(pcie_index_hi_offset);
+	}
+	writel((u32)(reg_data & 0xffffffffULL), pcie_data_offset);
+	readl(pcie_data_offset);
+	/* write high 32 bits */
+	writel(reg_addr + 4, pcie_index_offset);
+	readl(pcie_index_offset);
+	if (pcie_index_hi != 0) {
+		writel((reg_addr >> 32) & 0xff, pcie_index_hi_offset);
+		readl(pcie_index_hi_offset);
+	}
+	writel((u32)(reg_data >> 32), pcie_data_offset);
+	readl(pcie_data_offset);
+
+	/* clear the high bits */
+	if (pcie_index_hi != 0) {
+		writel(0, pcie_index_hi_offset);
+		readl(pcie_index_hi_offset);
+	}
+
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+}
+
+/**
+ * amdgpu_device_get_rev_id - query device rev_id
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Return device rev_id
+ */
+u32 amdgpu_device_get_rev_id(struct amdgpu_device *adev)
+{
+	return adev->nbio.funcs->get_rev_id(adev);
 }
 
 /**
  * amdgpu_invalid_rreg - dummy reg read function
  *
- * @adev: amdgpu device pointer
+ * @adev: amdgpu_device pointer
  * @reg: offset of register
  *
  * Dummy register read function.  Used for register blocks
@@ -272,7 +1283,14 @@ void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v)
  */
 static uint32_t amdgpu_invalid_rreg(struct amdgpu_device *adev, uint32_t reg)
 {
-	DRM_ERROR("Invalid callback to read register 0x%04X\n", reg);
+	dev_err(adev->dev, "Invalid callback to read register 0x%04X\n", reg);
+	BUG();
+	return 0;
+}
+
+static uint32_t amdgpu_invalid_rreg_ext(struct amdgpu_device *adev, uint64_t reg)
+{
+	dev_err(adev->dev, "Invalid callback to read register 0x%llX\n", reg);
 	BUG();
 	return 0;
 }
@@ -280,7 +1298,7 @@ static uint32_t amdgpu_invalid_rreg(struct amdgpu_device *adev, uint32_t reg)
 /**
  * amdgpu_invalid_wreg - dummy reg write function
  *
- * @adev: amdgpu device pointer
+ * @adev: amdgpu_device pointer
  * @reg: offset of register
  * @v: value to write to the register
  *
@@ -289,15 +1307,75 @@ static uint32_t amdgpu_invalid_rreg(struct amdgpu_device *adev, uint32_t reg)
  */
 static void amdgpu_invalid_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v)
 {
-	DRM_ERROR("Invalid callback to write register 0x%04X with 0x%08X\n",
-		  reg, v);
+	dev_err(adev->dev,
+		"Invalid callback to write register 0x%04X with 0x%08X\n", reg,
+		v);
+	BUG();
+}
+
+static void amdgpu_invalid_wreg_ext(struct amdgpu_device *adev, uint64_t reg, uint32_t v)
+{
+	dev_err(adev->dev,
+		"Invalid callback to write register 0x%llX with 0x%08X\n", reg,
+		v);
+	BUG();
+}
+
+/**
+ * amdgpu_invalid_rreg64 - dummy 64 bit reg read function
+ *
+ * @adev: amdgpu_device pointer
+ * @reg: offset of register
+ *
+ * Dummy register read function.  Used for register blocks
+ * that certain asics don't have (all asics).
+ * Returns the value in the register.
+ */
+static uint64_t amdgpu_invalid_rreg64(struct amdgpu_device *adev, uint32_t reg)
+{
+	dev_err(adev->dev, "Invalid callback to read 64 bit register 0x%04X\n",
+		reg);
+	BUG();
+	return 0;
+}
+
+static uint64_t amdgpu_invalid_rreg64_ext(struct amdgpu_device *adev, uint64_t reg)
+{
+	dev_err(adev->dev, "Invalid callback to read register 0x%llX\n", reg);
+	BUG();
+	return 0;
+}
+
+/**
+ * amdgpu_invalid_wreg64 - dummy reg write function
+ *
+ * @adev: amdgpu_device pointer
+ * @reg: offset of register
+ * @v: value to write to the register
+ *
+ * Dummy register read function.  Used for register blocks
+ * that certain asics don't have (all asics).
+ */
+static void amdgpu_invalid_wreg64(struct amdgpu_device *adev, uint32_t reg, uint64_t v)
+{
+	dev_err(adev->dev,
+		"Invalid callback to write 64 bit register 0x%04X with 0x%08llX\n",
+		reg, v);
+	BUG();
+}
+
+static void amdgpu_invalid_wreg64_ext(struct amdgpu_device *adev, uint64_t reg, uint64_t v)
+{
+	dev_err(adev->dev,
+		"Invalid callback to write 64 bit register 0x%llX with 0x%08llX\n",
+		reg, v);
 	BUG();
 }
 
 /**
  * amdgpu_block_invalid_rreg - dummy reg read function
  *
- * @adev: amdgpu device pointer
+ * @adev: amdgpu_device pointer
  * @block: offset of instance
  * @reg: offset of register
  *
@@ -308,8 +1386,9 @@ static void amdgpu_invalid_wreg(struct amdgpu_device *adev, uint32_t reg, uint32
 static uint32_t amdgpu_block_invalid_rreg(struct amdgpu_device *adev,
 					  uint32_t block, uint32_t reg)
 {
-	DRM_ERROR("Invalid callback to read register 0x%04X in block 0x%04X\n",
-		  reg, block);
+	dev_err(adev->dev,
+		"Invalid callback to read register 0x%04X in block 0x%04X\n",
+		reg, block);
 	BUG();
 	return 0;
 }
@@ -317,7 +1396,7 @@ static uint32_t amdgpu_block_invalid_rreg(struct amdgpu_device *adev,
 /**
  * amdgpu_block_invalid_wreg - dummy reg write function
  *
- * @adev: amdgpu device pointer
+ * @adev: amdgpu_device pointer
  * @block: offset of instance
  * @reg: offset of register
  * @v: value to write to the register
@@ -329,38 +1408,103 @@ static void amdgpu_block_invalid_wreg(struct amdgpu_device *adev,
 				      uint32_t block,
 				      uint32_t reg, uint32_t v)
 {
-	DRM_ERROR("Invalid block callback to write register 0x%04X in block 0x%04X with 0x%08X\n",
-		  reg, block, v);
+	dev_err(adev->dev,
+		"Invalid block callback to write register 0x%04X in block 0x%04X with 0x%08X\n",
+		reg, block, v);
 	BUG();
 }
 
-static int amdgpu_vram_scratch_init(struct amdgpu_device *adev)
+static uint32_t amdgpu_device_get_vbios_flags(struct amdgpu_device *adev)
 {
-	return amdgpu_bo_create_kernel(adev, AMDGPU_GPU_PAGE_SIZE,
-				       PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
-				       &adev->vram_scratch.robj,
-				       &adev->vram_scratch.gpu_addr,
-				       (void **)&adev->vram_scratch.ptr);
+	if (hweight32(adev->aid_mask) && (adev->flags & AMD_IS_APU))
+		return AMDGPU_VBIOS_SKIP;
+
+	if (hweight32(adev->aid_mask) && amdgpu_passthrough(adev))
+		return AMDGPU_VBIOS_OPTIONAL;
+
+	return 0;
 }
 
-static void amdgpu_vram_scratch_fini(struct amdgpu_device *adev)
+/**
+ * amdgpu_device_asic_init - Wrapper for atom asic_init
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Does any asic specific work and then calls atom asic init.
+ */
+static int amdgpu_device_asic_init(struct amdgpu_device *adev)
 {
-	amdgpu_bo_free_kernel(&adev->vram_scratch.robj, NULL, NULL);
+	uint32_t flags;
+	bool optional;
+	int ret;
+
+	amdgpu_asic_pre_asic_init(adev);
+	flags = amdgpu_device_get_vbios_flags(adev);
+	optional = !!(flags & (AMDGPU_VBIOS_OPTIONAL | AMDGPU_VBIOS_SKIP));
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(11, 0, 0)) {
+		amdgpu_psp_wait_for_bootloader(adev);
+		if (optional && !adev->bios)
+			return 0;
+
+		ret = amdgpu_atomfirmware_asic_init(adev, true);
+		return ret;
+	} else {
+		if (optional && !adev->bios)
+			return 0;
+
+		return amdgpu_atom_asic_init(adev->mode_info.atom_context);
+	}
+
+	return 0;
 }
 
 /**
- * amdgpu_program_register_sequence - program an array of registers.
+ * amdgpu_device_mem_scratch_init - allocate the VRAM scratch page
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocates a scratch page of VRAM for use by various things in the
+ * driver.
+ */
+static int amdgpu_device_mem_scratch_init(struct amdgpu_device *adev)
+{
+	return amdgpu_bo_create_kernel(adev, AMDGPU_GPU_PAGE_SIZE, PAGE_SIZE,
+				       AMDGPU_GEM_DOMAIN_VRAM |
+				       AMDGPU_GEM_DOMAIN_GTT,
+				       &adev->mem_scratch.robj,
+				       &adev->mem_scratch.gpu_addr,
+				       (void **)&adev->mem_scratch.ptr);
+}
+
+/**
+ * amdgpu_device_mem_scratch_fini - Free the VRAM scratch page
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Frees the VRAM scratch page.
+ */
+static void amdgpu_device_mem_scratch_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->mem_scratch.robj, NULL, NULL);
+}
+
+/**
+ * amdgpu_device_program_register_sequence - program an array of registers.
  *
  * @adev: amdgpu_device pointer
  * @registers: pointer to the register array
  * @array_size: size of the register array
  *
- * Programs an array or registers with and and or masks.
+ * Programs an array or registers with and or masks.
  * This is a helper for setting golden registers.
  */
-void amdgpu_program_register_sequence(struct amdgpu_device *adev,
-				      const u32 *registers,
-				      const u32 array_size)
+void amdgpu_device_program_register_sequence(struct amdgpu_device *adev,
+					     const u32 *registers,
+					     const u32 array_size)
 {
 	u32 tmp, reg, and_mask, or_mask;
 	int i;
@@ -368,7 +1512,7 @@ void amdgpu_program_register_sequence(struct amdgpu_device *adev,
 	if (array_size % 3)
 		return;
 
-	for (i = 0; i < array_size; i +=3) {
+	for (i = 0; i < array_size; i += 3) {
 		reg = registers[i + 0];
 		and_mask = registers[i + 1];
 		or_mask = registers[i + 2];
@@ -378,109 +1522,55 @@ void amdgpu_program_register_sequence(struct amdgpu_device *adev,
 		} else {
 			tmp = RREG32(reg);
 			tmp &= ~and_mask;
-			tmp |= or_mask;
+			if (adev->family >= AMDGPU_FAMILY_AI)
+				tmp |= (or_mask & and_mask);
+			else
+				tmp |= or_mask;
 		}
 		WREG32(reg, tmp);
 	}
 }
 
-void amdgpu_pci_config_reset(struct amdgpu_device *adev)
-{
-	pci_write_config_dword(adev->pdev, 0x7c, AMDGPU_ASIC_RESET_DATA);
-}
-
-/*
- * GPU doorbell aperture helpers function.
- */
 /**
- * amdgpu_doorbell_init - Init doorbell driver information.
+ * amdgpu_device_pci_config_reset - reset the GPU
  *
  * @adev: amdgpu_device pointer
  *
- * Init doorbell driver information (CIK)
- * Returns 0 on success, error on failure.
+ * Resets the GPU using the pci config reset sequence.
+ * Only applicable to asics prior to vega10.
  */
-static int amdgpu_doorbell_init(struct amdgpu_device *adev)
+void amdgpu_device_pci_config_reset(struct amdgpu_device *adev)
 {
-	/* doorbell bar mapping */
-	adev->doorbell.base = pci_resource_start(adev->pdev, 2);
-	adev->doorbell.size = pci_resource_len(adev->pdev, 2);
-
-	adev->doorbell.num_doorbells = min_t(u32, adev->doorbell.size / sizeof(u32),
-					     AMDGPU_DOORBELL_MAX_ASSIGNMENT+1);
-	if (adev->doorbell.num_doorbells == 0)
-		return -EINVAL;
-
-	adev->doorbell.ptr = ioremap(adev->doorbell.base,
-				     adev->doorbell.num_doorbells *
-				     sizeof(u32));
-	if (adev->doorbell.ptr == NULL)
-		return -ENOMEM;
-
-	return 0;
-}
-
-/**
- * amdgpu_doorbell_fini - Tear down doorbell driver information.
- *
- * @adev: amdgpu_device pointer
- *
- * Tear down doorbell driver information (CIK)
- */
-static void amdgpu_doorbell_fini(struct amdgpu_device *adev)
-{
-	iounmap(adev->doorbell.ptr);
-	adev->doorbell.ptr = NULL;
+	pci_write_config_dword(adev->pdev, 0x7c, AMDGPU_ASIC_RESET_DATA);
 }
 
 /**
- * amdgpu_doorbell_get_kfd_info - Report doorbell configuration required to
- *                                setup amdkfd
+ * amdgpu_device_pci_reset - reset the GPU using generic PCI means
  *
  * @adev: amdgpu_device pointer
- * @aperture_base: output returning doorbell aperture base physical address
- * @aperture_size: output returning doorbell aperture size in bytes
- * @start_offset: output returning # of doorbell bytes reserved for amdgpu.
  *
- * amdgpu and amdkfd share the doorbell aperture. amdgpu sets it up,
- * takes doorbells required for its own rings and reports the setup to amdkfd.
- * amdgpu reserved doorbells are at the start of the doorbell aperture.
+ * Resets the GPU using generic pci reset interfaces (FLR, SBR, etc.).
  */
-void amdgpu_doorbell_get_kfd_info(struct amdgpu_device *adev,
-				phys_addr_t *aperture_base,
-				size_t *aperture_size,
-				size_t *start_offset)
+int amdgpu_device_pci_reset(struct amdgpu_device *adev)
 {
-	/*
-	 * The first num_doorbells are used by amdgpu.
-	 * amdkfd takes whatever's left in the aperture.
-	 */
-	if (adev->doorbell.size > adev->doorbell.num_doorbells * sizeof(u32)) {
-		*aperture_base = adev->doorbell.base;
-		*aperture_size = adev->doorbell.size;
-		*start_offset = adev->doorbell.num_doorbells * sizeof(u32);
-	} else {
-		*aperture_base = 0;
-		*aperture_size = 0;
-		*start_offset = 0;
-	}
+	return pci_reset_function(adev->pdev);
 }
 
 /*
- * amdgpu_wb_*()
+ * amdgpu_device_wb_*()
  * Writeback is the method by which the GPU updates special pages in memory
  * with the status of certain GPU events (fences, ring pointers,etc.).
  */
 
 /**
- * amdgpu_wb_fini - Disable Writeback and free memory
+ * amdgpu_device_wb_fini - Disable Writeback and free memory
  *
  * @adev: amdgpu_device pointer
  *
  * Disables Writeback and frees the Writeback memory (all asics).
  * Used at driver shutdown.
  */
-static void amdgpu_wb_fini(struct amdgpu_device *adev)
+static void amdgpu_device_wb_fini(struct amdgpu_device *adev)
 {
 	if (adev->wb.wb_obj) {
 		amdgpu_bo_free_kernel(&adev->wb.wb_obj,
@@ -491,7 +1581,7 @@ static void amdgpu_wb_fini(struct amdgpu_device *adev)
 }
 
 /**
- * amdgpu_wb_init- Init Writeback driver info and allocate memory
+ * amdgpu_device_wb_init - Init Writeback driver info and allocate memory
  *
  * @adev: amdgpu_device pointer
  *
@@ -499,7 +1589,7 @@ static void amdgpu_wb_fini(struct amdgpu_device *adev)
  * Used at driver startup.
  * Returns 0 on success or an -error on failure.
  */
-static int amdgpu_wb_init(struct amdgpu_device *adev)
+static int amdgpu_device_wb_init(struct amdgpu_device *adev)
 {
 	int r;
 
@@ -518,14 +1608,14 @@ static int amdgpu_wb_init(struct amdgpu_device *adev)
 		memset(&adev->wb.used, 0, sizeof(adev->wb.used));
 
 		/* clear wb memory */
-		memset((char *)adev->wb.wb, 0, AMDGPU_MAX_WB * sizeof(uint32_t));
+		memset((char *)adev->wb.wb, 0, AMDGPU_MAX_WB * sizeof(uint32_t) * 8);
 	}
 
 	return 0;
 }
 
 /**
- * amdgpu_wb_get - Allocate a wb entry
+ * amdgpu_device_wb_get - Allocate a wb entry
  *
  * @adev: amdgpu_device pointer
  * @wb: wb index
@@ -533,124 +1623,145 @@ static int amdgpu_wb_init(struct amdgpu_device *adev)
  * Allocate a wb slot for use by the driver (all asics).
  * Returns 0 on success or -EINVAL on failure.
  */
-int amdgpu_wb_get(struct amdgpu_device *adev, u32 *wb)
+int amdgpu_device_wb_get(struct amdgpu_device *adev, u32 *wb)
 {
-	unsigned long offset = find_first_zero_bit(adev->wb.used, adev->wb.num_wb);
+	unsigned long flags, offset;
 
+	spin_lock_irqsave(&adev->wb.lock, flags);
+	offset = find_first_zero_bit(adev->wb.used, adev->wb.num_wb);
 	if (offset < adev->wb.num_wb) {
 		__set_bit(offset, adev->wb.used);
-		*wb = offset * 8; /* convert to dw offset */
+		spin_unlock_irqrestore(&adev->wb.lock, flags);
+		*wb = offset << 3; /* convert to dw offset */
 		return 0;
 	} else {
+		spin_unlock_irqrestore(&adev->wb.lock, flags);
 		return -EINVAL;
 	}
 }
 
 /**
- * amdgpu_wb_free - Free a wb entry
+ * amdgpu_device_wb_free - Free a wb entry
  *
  * @adev: amdgpu_device pointer
  * @wb: wb index
  *
  * Free a wb slot allocated for use by the driver (all asics)
  */
-void amdgpu_wb_free(struct amdgpu_device *adev, u32 wb)
+void amdgpu_device_wb_free(struct amdgpu_device *adev, u32 wb)
 {
+	unsigned long flags;
+
+	wb >>= 3;
+	spin_lock_irqsave(&adev->wb.lock, flags);
 	if (wb < adev->wb.num_wb)
 		__clear_bit(wb, adev->wb.used);
+	spin_unlock_irqrestore(&adev->wb.lock, flags);
 }
 
 /**
- * amdgpu_vram_location - try to find VRAM location
- * @adev: amdgpu device structure holding all necessary informations
- * @mc: memory controller structure holding memory informations
- * @base: base address at which to put VRAM
- *
- * Function will try to place VRAM at base address provided
- * as parameter (which is so far either PCI aperture address or
- * for IGP TOM base address).
- *
- * If there is not enough space to fit the unvisible VRAM in the 32bits
- * address space then we limit the VRAM size to the aperture.
+ * amdgpu_device_resize_fb_bar - try to resize FB BAR
  *
- * Note: We don't explicitly enforce VRAM start to be aligned on VRAM size,
- * this shouldn't be a problem as we are using the PCI aperture as a reference.
- * Otherwise this would be needed for rv280, all r3xx, and all r4xx, but
- * not IGP.
- *
- * Note: we use mc_vram_size as on some board we need to program the mc to
- * cover the whole aperture even if VRAM size is inferior to aperture size
- * Novell bug 204882 + along with lots of ubuntu ones
- *
- * Note: when limiting vram it's safe to overwritte real_vram_size because
- * we are not in case where real_vram_size is inferior to mc_vram_size (ie
- * note afected by bogus hw of Novell bug 204882 + along with lots of ubuntu
- * ones)
- *
- * Note: IGP TOM addr should be the same as the aperture addr, we don't
- * explicitly check for that though.
+ * @adev: amdgpu_device pointer
  *
- * FIXME: when reducing VRAM size align new size on power of 2.
+ * Try to resize FB BAR to make all VRAM CPU accessible. We try very hard not
+ * to fail, but if any of the BARs is not accessible after the size we abort
+ * driver loading by returning -ENODEV.
  */
-void amdgpu_vram_location(struct amdgpu_device *adev, struct amdgpu_mc *mc, u64 base)
+int amdgpu_device_resize_fb_bar(struct amdgpu_device *adev)
 {
-	uint64_t limit = (uint64_t)amdgpu_vram_limit << 20;
+	int rbar_size = pci_rebar_bytes_to_size(adev->gmc.real_vram_size);
+	struct pci_bus *root;
+	struct resource *res;
+	unsigned int i;
+	u16 cmd;
+	int r;
 
-	mc->vram_start = base;
-	if (mc->mc_vram_size > (adev->mc.mc_mask - base + 1)) {
-		dev_warn(adev->dev, "limiting VRAM to PCI aperture size\n");
-		mc->real_vram_size = mc->aper_size;
-		mc->mc_vram_size = mc->aper_size;
-	}
-	mc->vram_end = mc->vram_start + mc->mc_vram_size - 1;
-	if (limit && limit < mc->real_vram_size)
-		mc->real_vram_size = limit;
-	dev_info(adev->dev, "VRAM: %lluM 0x%016llX - 0x%016llX (%lluM used)\n",
-			mc->mc_vram_size >> 20, mc->vram_start,
-			mc->vram_end, mc->real_vram_size >> 20);
-}
+	if (!IS_ENABLED(CONFIG_PHYS_ADDR_T_64BIT))
+		return 0;
 
-/**
- * amdgpu_gart_location - try to find GTT location
- * @adev: amdgpu device structure holding all necessary informations
- * @mc: memory controller structure holding memory informations
- *
- * Function will place try to place GTT before or after VRAM.
- *
- * If GTT size is bigger than space left then we ajust GTT size.
- * Thus function will never fails.
- *
- * FIXME: when reducing GTT size align new size on power of 2.
- */
-void amdgpu_gart_location(struct amdgpu_device *adev, struct amdgpu_mc *mc)
-{
-	u64 size_af, size_bf;
+	/* Bypass for VF */
+	if (amdgpu_sriov_vf(adev))
+		return 0;
 
-	size_af = adev->mc.mc_mask - mc->vram_end;
-	size_bf = mc->vram_start;
-	if (size_bf > size_af) {
-		if (mc->gart_size > size_bf) {
-			dev_warn(adev->dev, "limiting GTT\n");
-			mc->gart_size = size_bf;
-		}
-		mc->gart_start = 0;
-	} else {
-		if (mc->gart_size > size_af) {
-			dev_warn(adev->dev, "limiting GTT\n");
-			mc->gart_size = size_af;
-		}
-		mc->gart_start = mc->vram_end + 1;
+	if (!amdgpu_rebar)
+		return 0;
+
+	/* resizing on Dell G5 SE platforms causes problems with runtime pm */
+	if ((amdgpu_runtime_pm != 0) &&
+	    adev->pdev->vendor == PCI_VENDOR_ID_ATI &&
+	    adev->pdev->device == 0x731f &&
+	    adev->pdev->subsystem_vendor == PCI_VENDOR_ID_DELL)
+		return 0;
+
+	/* PCI_EXT_CAP_ID_VNDR extended capability is located at 0x100 */
+	if (!pci_find_ext_capability(adev->pdev, PCI_EXT_CAP_ID_VNDR))
+		dev_warn(
+			adev->dev,
+			"System can't access extended configuration space, please check!!\n");
+
+	/* skip if the bios has already enabled large BAR */
+	if (adev->gmc.real_vram_size &&
+	    (pci_resource_len(adev->pdev, 0) >= adev->gmc.real_vram_size))
+		return 0;
+
+	/* Check if the root BUS has 64bit memory resources */
+	root = adev->pdev->bus;
+	while (root->parent)
+		root = root->parent;
+
+	pci_bus_for_each_resource(root, res, i) {
+		if (res && res->flags & (IORESOURCE_MEM | IORESOURCE_MEM_64) &&
+		    res->start > 0x100000000ull)
+			break;
 	}
-	mc->gart_end = mc->gart_start + mc->gart_size - 1;
-	dev_info(adev->dev, "GTT: %lluM 0x%016llX - 0x%016llX\n",
-			mc->gart_size >> 20, mc->gart_start, mc->gart_end);
+
+	/* Trying to resize is pointless without a root hub window above 4GB */
+	if (!res)
+		return 0;
+
+	/* Limit the BAR size to what is available */
+	rbar_size = min(fls(pci_rebar_get_possible_sizes(adev->pdev, 0)) - 1,
+			rbar_size);
+
+	/* Disable memory decoding while we change the BAR addresses and size */
+	pci_read_config_word(adev->pdev, PCI_COMMAND, &cmd);
+	pci_write_config_word(adev->pdev, PCI_COMMAND,
+			      cmd & ~PCI_COMMAND_MEMORY);
+
+	/* Free the VRAM and doorbell BAR, we most likely need to move both. */
+	amdgpu_doorbell_fini(adev);
+	if (adev->asic_type >= CHIP_BONAIRE)
+		pci_release_resource(adev->pdev, 2);
+
+	pci_release_resource(adev->pdev, 0);
+
+	r = pci_resize_resource(adev->pdev, 0, rbar_size);
+	if (r == -ENOSPC)
+		dev_info(adev->dev,
+			 "Not enough PCI address space for a large BAR.");
+	else if (r && r != -ENOTSUPP)
+		dev_err(adev->dev, "Problem resizing BAR0 (%d).", r);
+
+	pci_assign_unassigned_bus_resources(adev->pdev->bus);
+
+	/* When the doorbell or fb BAR isn't available we have no chance of
+	 * using the device.
+	 */
+	r = amdgpu_doorbell_init(adev);
+	if (r || (pci_resource_flags(adev->pdev, 0) & IORESOURCE_UNSET))
+		return -ENODEV;
+
+	pci_write_config_word(adev->pdev, PCI_COMMAND, cmd);
+
+	return 0;
 }
 
 /*
  * GPU helpers function.
  */
 /**
- * amdgpu_need_post - check if the hw need post or not
+ * amdgpu_device_need_post - check if the hw need post or not
  *
  * @adev: amdgpu_device pointer
  *
@@ -658,32 +1769,17 @@ void amdgpu_gart_location(struct amdgpu_device *adev, struct amdgpu_mc *mc)
  * or post is needed if  hw reset is performed.
  * Returns true if need or false if not.
  */
-bool amdgpu_need_post(struct amdgpu_device *adev)
+bool amdgpu_device_need_post(struct amdgpu_device *adev)
 {
-	uint32_t reg;
-
-	if (adev->has_hw_reset) {
-		adev->has_hw_reset = false;
-		return true;
-	}
+	uint32_t reg, flags;
 
-	/* bios scratch used on CIK+ */
-	if (adev->asic_type >= CHIP_BONAIRE)
-		return amdgpu_atombios_scratch_need_asic_init(adev);
-
-	/* check MEM_SIZE for older asics */
-	reg = amdgpu_asic_get_config_memsize(adev);
-
-	if ((reg != 0) && (reg != 0xffffffff))
+	if (amdgpu_sriov_vf(adev))
 		return false;
 
-	return true;
-
-}
-
-static bool amdgpu_vpost_needed(struct amdgpu_device *adev)
-{
-	if (amdgpu_sriov_vf(adev))
+	flags = amdgpu_device_get_vbios_flags(adev);
+	if (flags & AMDGPU_VBIOS_SKIP)
+		return false;
+	if ((flags & AMDGPU_VBIOS_OPTIONAL) && !adev->bios)
 		return false;
 
 	if (amdgpu_passthrough(adev)) {
@@ -695,285 +1791,168 @@ static bool amdgpu_vpost_needed(struct amdgpu_device *adev)
 		if (adev->asic_type == CHIP_FIJI) {
 			int err;
 			uint32_t fw_ver;
+
 			err = request_firmware(&adev->pm.fw, "amdgpu/fiji_smc.bin", adev->dev);
-			/* force vPost if error occured */
+			/* force vPost if error occurred */
 			if (err)
 				return true;
 
 			fw_ver = *((uint32_t *)adev->pm.fw->data + 69);
+			release_firmware(adev->pm.fw);
 			if (fw_ver < 0x00160e00)
 				return true;
 		}
 	}
-	return amdgpu_need_post(adev);
-}
-
-/**
- * amdgpu_dummy_page_init - init dummy page used by the driver
- *
- * @adev: amdgpu_device pointer
- *
- * Allocate the dummy page used by the driver (all asics).
- * This dummy page is used by the driver as a filler for gart entries
- * when pages are taken out of the GART
- * Returns 0 on sucess, -ENOMEM on failure.
- */
-int amdgpu_dummy_page_init(struct amdgpu_device *adev)
-{
-	if (adev->dummy_page.page)
-		return 0;
-	adev->dummy_page.page = alloc_page(GFP_DMA32 | GFP_KERNEL | __GFP_ZERO);
-	if (adev->dummy_page.page == NULL)
-		return -ENOMEM;
-	adev->dummy_page.addr = pci_map_page(adev->pdev, adev->dummy_page.page,
-					0, PAGE_SIZE, PCI_DMA_BIDIRECTIONAL);
-	if (pci_dma_mapping_error(adev->pdev, adev->dummy_page.addr)) {
-		dev_err(&adev->pdev->dev, "Failed to DMA MAP the dummy page\n");
-		__free_page(adev->dummy_page.page);
-		adev->dummy_page.page = NULL;
-		return -ENOMEM;
-	}
-	return 0;
-}
-
-/**
- * amdgpu_dummy_page_fini - free dummy page used by the driver
- *
- * @adev: amdgpu_device pointer
- *
- * Frees the dummy page used by the driver (all asics).
- */
-void amdgpu_dummy_page_fini(struct amdgpu_device *adev)
-{
-	if (adev->dummy_page.page == NULL)
-		return;
-	pci_unmap_page(adev->pdev, adev->dummy_page.addr,
-			PAGE_SIZE, PCI_DMA_BIDIRECTIONAL);
-	__free_page(adev->dummy_page.page);
-	adev->dummy_page.page = NULL;
-}
 
+	/* Don't post if we need to reset whole hive on init */
+	if (adev->init_lvl->level == AMDGPU_INIT_LEVEL_MINIMAL_XGMI)
+		return false;
 
-/* ATOM accessor methods */
-/*
- * ATOM is an interpreted byte code stored in tables in the vbios.  The
- * driver registers callbacks to access registers and the interpreter
- * in the driver parses the tables and executes then to program specific
- * actions (set display modes, asic init, etc.).  See amdgpu_atombios.c,
- * atombios.h, and atom.c
- */
+	if (adev->has_hw_reset) {
+		adev->has_hw_reset = false;
+		return true;
+	}
 
-/**
- * cail_pll_read - read PLL register
- *
- * @info: atom card_info pointer
- * @reg: PLL register offset
- *
- * Provides a PLL register accessor for the atom interpreter (r4xx+).
- * Returns the value of the PLL register.
- */
-static uint32_t cail_pll_read(struct card_info *info, uint32_t reg)
-{
-	return 0;
-}
+	/* bios scratch used on CIK+ */
+	if (adev->asic_type >= CHIP_BONAIRE)
+		return amdgpu_atombios_scratch_need_asic_init(adev);
 
-/**
- * cail_pll_write - write PLL register
- *
- * @info: atom card_info pointer
- * @reg: PLL register offset
- * @val: value to write to the pll register
- *
- * Provides a PLL register accessor for the atom interpreter (r4xx+).
- */
-static void cail_pll_write(struct card_info *info, uint32_t reg, uint32_t val)
-{
+	/* check MEM_SIZE for older asics */
+	reg = amdgpu_asic_get_config_memsize(adev);
 
-}
+	if ((reg != 0) && (reg != 0xffffffff))
+		return false;
 
-/**
- * cail_mc_read - read MC (Memory Controller) register
- *
- * @info: atom card_info pointer
- * @reg: MC register offset
- *
- * Provides an MC register accessor for the atom interpreter (r4xx+).
- * Returns the value of the MC register.
- */
-static uint32_t cail_mc_read(struct card_info *info, uint32_t reg)
-{
-	return 0;
+	return true;
 }
 
-/**
- * cail_mc_write - write MC (Memory Controller) register
- *
- * @info: atom card_info pointer
- * @reg: MC register offset
- * @val: value to write to the pll register
+/*
+ * Check whether seamless boot is supported.
  *
- * Provides a MC register accessor for the atom interpreter (r4xx+).
+ * So far we only support seamless boot on DCE 3.0 or later.
+ * If users report that it works on older ASICS as well, we may
+ * loosen this.
  */
-static void cail_mc_write(struct card_info *info, uint32_t reg, uint32_t val)
+bool amdgpu_device_seamless_boot_supported(struct amdgpu_device *adev)
 {
+	switch (amdgpu_seamless) {
+	case -1:
+		break;
+	case 1:
+		return true;
+	case 0:
+		return false;
+	default:
+		dev_err(adev->dev, "Invalid value for amdgpu.seamless: %d\n",
+			amdgpu_seamless);
+		return false;
+	}
 
-}
+	if (!(adev->flags & AMD_IS_APU))
+		return false;
 
-/**
- * cail_reg_write - write MMIO register
- *
- * @info: atom card_info pointer
- * @reg: MMIO register offset
- * @val: value to write to the pll register
- *
- * Provides a MMIO register accessor for the atom interpreter (r4xx+).
- */
-static void cail_reg_write(struct card_info *info, uint32_t reg, uint32_t val)
-{
-	struct amdgpu_device *adev = info->dev->dev_private;
+	if (adev->mman.keep_stolen_vga_memory)
+		return false;
 
-	WREG32(reg, val);
+	return amdgpu_ip_version(adev, DCE_HWIP, 0) >= IP_VERSION(3, 0, 0);
 }
 
-/**
- * cail_reg_read - read MMIO register
- *
- * @info: atom card_info pointer
- * @reg: MMIO register offset
+/*
+ * Intel hosts such as Rocket Lake, Alder Lake, Raptor Lake and Sapphire Rapids
+ * don't support dynamic speed switching. Until we have confirmation from Intel
+ * that a specific host supports it, it's safer that we keep it disabled for all.
  *
- * Provides an MMIO register accessor for the atom interpreter (r4xx+).
- * Returns the value of the MMIO register.
+ * https://edc.intel.com/content/www/us/en/design/products/platforms/details/raptor-lake-s/13th-generation-core-processors-datasheet-volume-1-of-2/005/pci-express-support/
+ * https://gitlab.freedesktop.org/drm/amd/-/issues/2663
  */
-static uint32_t cail_reg_read(struct card_info *info, uint32_t reg)
+static bool amdgpu_device_pcie_dynamic_switching_supported(struct amdgpu_device *adev)
 {
-	struct amdgpu_device *adev = info->dev->dev_private;
-	uint32_t r;
-
-	r = RREG32(reg);
-	return r;
-}
+#if IS_ENABLED(CONFIG_X86)
+	struct cpuinfo_x86 *c = &cpu_data(0);
 
-/**
- * cail_ioreg_write - write IO register
- *
- * @info: atom card_info pointer
- * @reg: IO register offset
- * @val: value to write to the pll register
- *
- * Provides a IO register accessor for the atom interpreter (r4xx+).
- */
-static void cail_ioreg_write(struct card_info *info, uint32_t reg, uint32_t val)
-{
-	struct amdgpu_device *adev = info->dev->dev_private;
+	/* eGPU change speeds based on USB4 fabric conditions */
+	if (dev_is_removable(adev->dev))
+		return true;
 
-	WREG32_IO(reg, val);
+	if (c->x86_vendor == X86_VENDOR_INTEL)
+		return false;
+#endif
+	return true;
 }
 
-/**
- * cail_ioreg_read - read IO register
- *
- * @info: atom card_info pointer
- * @reg: IO register offset
- *
- * Provides an IO register accessor for the atom interpreter (r4xx+).
- * Returns the value of the IO register.
- */
-static uint32_t cail_ioreg_read(struct card_info *info, uint32_t reg)
+static bool amdgpu_device_aspm_support_quirk(struct amdgpu_device *adev)
 {
-	struct amdgpu_device *adev = info->dev->dev_private;
-	uint32_t r;
+#if IS_ENABLED(CONFIG_X86)
+	struct cpuinfo_x86 *c = &cpu_data(0);
 
-	r = RREG32_IO(reg);
-	return r;
-}
+	if (!(amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(12, 0, 0) ||
+		  amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(12, 0, 1)))
+		return false;
 
-/**
- * amdgpu_atombios_fini - free the driver info and callbacks for atombios
- *
- * @adev: amdgpu_device pointer
- *
- * Frees the driver info and register access callbacks for the ATOM
- * interpreter (r4xx+).
- * Called at driver shutdown.
- */
-static void amdgpu_atombios_fini(struct amdgpu_device *adev)
-{
-	if (adev->mode_info.atom_context) {
-		kfree(adev->mode_info.atom_context->scratch);
-		kfree(adev->mode_info.atom_context->iio);
+	if (c->x86 == 6 &&
+		adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN5) {
+		switch (c->x86_model) {
+		case VFM_MODEL(INTEL_ALDERLAKE):
+		case VFM_MODEL(INTEL_ALDERLAKE_L):
+		case VFM_MODEL(INTEL_RAPTORLAKE):
+		case VFM_MODEL(INTEL_RAPTORLAKE_P):
+		case VFM_MODEL(INTEL_RAPTORLAKE_S):
+			return true;
+		default:
+			return false;
+		}
+	} else {
+		return false;
 	}
-	kfree(adev->mode_info.atom_context);
-	adev->mode_info.atom_context = NULL;
-	kfree(adev->mode_info.atom_card_info);
-	adev->mode_info.atom_card_info = NULL;
+#else
+	return false;
+#endif
 }
 
 /**
- * amdgpu_atombios_init - init the driver info and callbacks for atombios
+ * amdgpu_device_should_use_aspm - check if the device should program ASPM
  *
  * @adev: amdgpu_device pointer
  *
- * Initializes the driver info and register access callbacks for the
- * ATOM interpreter (r4xx+).
- * Returns 0 on sucess, -ENOMEM on failure.
- * Called at driver startup.
+ * Confirm whether the module parameter and pcie bridge agree that ASPM should
+ * be set for this device.
+ *
+ * Returns true if it should be used or false if not.
  */
-static int amdgpu_atombios_init(struct amdgpu_device *adev)
+bool amdgpu_device_should_use_aspm(struct amdgpu_device *adev)
 {
-	struct card_info *atom_card_info =
-	    kzalloc(sizeof(struct card_info), GFP_KERNEL);
-
-	if (!atom_card_info)
-		return -ENOMEM;
-
-	adev->mode_info.atom_card_info = atom_card_info;
-	atom_card_info->dev = adev->ddev;
-	atom_card_info->reg_read = cail_reg_read;
-	atom_card_info->reg_write = cail_reg_write;
-	/* needed for iio ops */
-	if (adev->rio_mem) {
-		atom_card_info->ioreg_read = cail_ioreg_read;
-		atom_card_info->ioreg_write = cail_ioreg_write;
-	} else {
-		DRM_INFO("PCI I/O BAR is not found. Using MMIO to access ATOM BIOS\n");
-		atom_card_info->ioreg_read = cail_reg_read;
-		atom_card_info->ioreg_write = cail_reg_write;
-	}
-	atom_card_info->mc_read = cail_mc_read;
-	atom_card_info->mc_write = cail_mc_write;
-	atom_card_info->pll_read = cail_pll_read;
-	atom_card_info->pll_write = cail_pll_write;
-
-	adev->mode_info.atom_context = amdgpu_atom_parse(atom_card_info, adev->bios);
-	if (!adev->mode_info.atom_context) {
-		amdgpu_atombios_fini(adev);
-		return -ENOMEM;
-	}
-
-	mutex_init(&adev->mode_info.atom_context->mutex);
-	if (adev->is_atom_fw) {
-		amdgpu_atomfirmware_scratch_regs_init(adev);
-		amdgpu_atomfirmware_allocate_fb_scratch(adev);
-	} else {
-		amdgpu_atombios_scratch_regs_init(adev);
-		amdgpu_atombios_allocate_fb_scratch(adev);
+	switch (amdgpu_aspm) {
+	case -1:
+		break;
+	case 0:
+		return false;
+	case 1:
+		return true;
+	default:
+		return false;
 	}
-	return 0;
+	if (adev->flags & AMD_IS_APU)
+		return false;
+	if (amdgpu_device_aspm_support_quirk(adev))
+		return false;
+	return pcie_aspm_enabled(adev->pdev);
 }
 
 /* if we get transitioned to only one device, take VGA back */
 /**
- * amdgpu_vga_set_decode - enable/disable vga decode
+ * amdgpu_device_vga_set_decode - enable/disable vga decode
  *
- * @cookie: amdgpu_device pointer
+ * @pdev: PCI device pointer
  * @state: enable/disable vga decode
  *
  * Enable/disable vga decode (all asics).
  * Returns VGA resource flags.
  */
-static unsigned int amdgpu_vga_set_decode(void *cookie, bool state)
+static unsigned int amdgpu_device_vga_set_decode(struct pci_dev *pdev,
+		bool state)
 {
-	struct amdgpu_device *adev = cookie;
+	struct amdgpu_device *adev = drm_to_adev(pci_get_drvdata(pdev));
+
 	amdgpu_asic_set_vga_state(adev, state);
 	if (state)
 		return VGA_RSRC_LEGACY_IO | VGA_RSRC_LEGACY_MEM |
@@ -982,81 +1961,147 @@ static unsigned int amdgpu_vga_set_decode(void *cookie, bool state)
 		return VGA_RSRC_NORMAL_IO | VGA_RSRC_NORMAL_MEM;
 }
 
-static void amdgpu_check_block_size(struct amdgpu_device *adev)
+/**
+ * amdgpu_device_check_block_size - validate the vm block size
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Validates the vm block size specified via module parameter.
+ * The vm block size defines number of bits in page table versus page directory,
+ * a page is 4KB so we have 12 bits offset, minimum 9 bits in the
+ * page table and the remaining bits are in the page directory.
+ */
+static void amdgpu_device_check_block_size(struct amdgpu_device *adev)
 {
 	/* defines number of bits in page table versus page directory,
 	 * a page is 4KB so we have 12 bits offset, minimum 9 bits in the
-	 * page table and the remaining bits are in the page directory */
+	 * page table and the remaining bits are in the page directory
+	 */
 	if (amdgpu_vm_block_size == -1)
 		return;
 
 	if (amdgpu_vm_block_size < 9) {
 		dev_warn(adev->dev, "VM page table size (%d) too small\n",
 			 amdgpu_vm_block_size);
-		goto def_value;
+		amdgpu_vm_block_size = -1;
 	}
-
-	if (amdgpu_vm_block_size > 24 ||
-	    (amdgpu_vm_size * 1024) < (1ull << amdgpu_vm_block_size)) {
-		dev_warn(adev->dev, "VM page table size (%d) too large\n",
-			 amdgpu_vm_block_size);
-		goto def_value;
-	}
-
-	return;
-
-def_value:
-	amdgpu_vm_block_size = -1;
 }
 
-static void amdgpu_check_vm_size(struct amdgpu_device *adev)
+/**
+ * amdgpu_device_check_vm_size - validate the vm size
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Validates the vm size in GB specified via module parameter.
+ * The VM size is the size of the GPU virtual memory space in GB.
+ */
+static void amdgpu_device_check_vm_size(struct amdgpu_device *adev)
 {
 	/* no need to check the default value */
 	if (amdgpu_vm_size == -1)
 		return;
 
-	if (!is_power_of_2(amdgpu_vm_size)) {
-		dev_warn(adev->dev, "VM size (%d) must be a power of 2\n",
-			 amdgpu_vm_size);
-		goto def_value;
-	}
-
 	if (amdgpu_vm_size < 1) {
 		dev_warn(adev->dev, "VM size (%d) too small, min is 1GB\n",
 			 amdgpu_vm_size);
-		goto def_value;
+		amdgpu_vm_size = -1;
 	}
+}
 
-	/*
-	 * Max GPUVM size for Cayman, SI, CI VI are 40 bits.
-	 */
-	if (amdgpu_vm_size > 1024) {
-		dev_warn(adev->dev, "VM size (%d) too large, max is 1TB\n",
-			 amdgpu_vm_size);
+static void amdgpu_device_check_smu_prv_buffer_size(struct amdgpu_device *adev)
+{
+	struct sysinfo si;
+	bool is_os_64 = (sizeof(void *) == 8);
+	uint64_t total_memory;
+	uint64_t dram_size_seven_GB = 0x1B8000000;
+	uint64_t dram_size_three_GB = 0xB8000000;
+
+	if (amdgpu_smu_memory_pool_size == 0)
+		return;
+
+	if (!is_os_64) {
+		dev_warn(adev->dev, "Not 64-bit OS, feature not supported\n");
+		goto def_value;
+	}
+	si_meminfo(&si);
+	total_memory = (uint64_t)si.totalram * si.mem_unit;
+
+	if ((amdgpu_smu_memory_pool_size == 1) ||
+		(amdgpu_smu_memory_pool_size == 2)) {
+		if (total_memory < dram_size_three_GB)
+			goto def_value1;
+	} else if ((amdgpu_smu_memory_pool_size == 4) ||
+		(amdgpu_smu_memory_pool_size == 8)) {
+		if (total_memory < dram_size_seven_GB)
+			goto def_value1;
+	} else {
+		dev_warn(adev->dev, "Smu memory pool size not supported\n");
 		goto def_value;
 	}
+	adev->pm.smu_prv_buffer_size = amdgpu_smu_memory_pool_size << 28;
 
 	return;
 
+def_value1:
+	dev_warn(adev->dev, "No enough system memory\n");
 def_value:
-	amdgpu_vm_size = -1;
+	adev->pm.smu_prv_buffer_size = 0;
+}
+
+static int amdgpu_device_init_apu_flags(struct amdgpu_device *adev)
+{
+	if (!(adev->flags & AMD_IS_APU) ||
+	    adev->asic_type < CHIP_RAVEN)
+		return 0;
+
+	switch (adev->asic_type) {
+	case CHIP_RAVEN:
+		if (adev->pdev->device == 0x15dd)
+			adev->apu_flags |= AMD_APU_IS_RAVEN;
+		if (adev->pdev->device == 0x15d8)
+			adev->apu_flags |= AMD_APU_IS_PICASSO;
+		break;
+	case CHIP_RENOIR:
+		if ((adev->pdev->device == 0x1636) ||
+		    (adev->pdev->device == 0x164c))
+			adev->apu_flags |= AMD_APU_IS_RENOIR;
+		else
+			adev->apu_flags |= AMD_APU_IS_GREEN_SARDINE;
+		break;
+	case CHIP_VANGOGH:
+		adev->apu_flags |= AMD_APU_IS_VANGOGH;
+		break;
+	case CHIP_YELLOW_CARP:
+		break;
+	case CHIP_CYAN_SKILLFISH:
+		if ((adev->pdev->device == 0x13FE) ||
+		    (adev->pdev->device == 0x143F))
+			adev->apu_flags |= AMD_APU_IS_CYAN_SKILLFISH2;
+		break;
+	default:
+		break;
+	}
+
+	return 0;
 }
 
 /**
- * amdgpu_check_arguments - validate module params
+ * amdgpu_device_check_arguments - validate module params
  *
  * @adev: amdgpu_device pointer
  *
  * Validates certain module parameters and updates
  * the associated values used by the driver (all asics).
  */
-static void amdgpu_check_arguments(struct amdgpu_device *adev)
+static int amdgpu_device_check_arguments(struct amdgpu_device *adev)
 {
+	int i;
+
 	if (amdgpu_sched_jobs < 4) {
 		dev_warn(adev->dev, "sched jobs (%d) must be at least 4\n",
 			 amdgpu_sched_jobs);
 		amdgpu_sched_jobs = 4;
-	} else if (!is_power_of_2(amdgpu_sched_jobs)){
+	} else if (!is_power_of_2(amdgpu_sched_jobs)) {
 		dev_warn(adev->dev, "sched jobs (%d) must be a power of 2\n",
 			 amdgpu_sched_jobs);
 		amdgpu_sched_jobs = roundup_pow_of_two(amdgpu_sched_jobs);
@@ -1083,16 +2128,56 @@ static void amdgpu_check_arguments(struct amdgpu_device *adev)
 		amdgpu_vm_fragment_size = -1;
 	}
 
-	amdgpu_check_vm_size(adev);
+	if (amdgpu_sched_hw_submission < 2) {
+		dev_warn(adev->dev, "sched hw submission jobs (%d) must be at least 2\n",
+			 amdgpu_sched_hw_submission);
+		amdgpu_sched_hw_submission = 2;
+	} else if (!is_power_of_2(amdgpu_sched_hw_submission)) {
+		dev_warn(adev->dev, "sched hw submission jobs (%d) must be a power of 2\n",
+			 amdgpu_sched_hw_submission);
+		amdgpu_sched_hw_submission = roundup_pow_of_two(amdgpu_sched_hw_submission);
+	}
 
-	amdgpu_check_block_size(adev);
+	if (amdgpu_reset_method < -1 || amdgpu_reset_method > 4) {
+		dev_warn(adev->dev, "invalid option for reset method, reverting to default\n");
+		amdgpu_reset_method = -1;
+	}
 
-	if (amdgpu_vram_page_split != -1 && (amdgpu_vram_page_split < 16 ||
-	    !is_power_of_2(amdgpu_vram_page_split))) {
-		dev_warn(adev->dev, "invalid VRAM page split (%d)\n",
-			 amdgpu_vram_page_split);
-		amdgpu_vram_page_split = 1024;
+	amdgpu_device_check_smu_prv_buffer_size(adev);
+
+	amdgpu_device_check_vm_size(adev);
+
+	amdgpu_device_check_block_size(adev);
+
+	adev->firmware.load_type = amdgpu_ucode_get_load_type(adev, amdgpu_fw_load_type);
+
+	for (i = 0; i < MAX_XCP; i++) {
+		switch (amdgpu_enforce_isolation) {
+		case -1:
+		case 0:
+		default:
+			/* disable */
+			adev->enforce_isolation[i] = AMDGPU_ENFORCE_ISOLATION_DISABLE;
+			break;
+		case 1:
+			/* enable */
+			adev->enforce_isolation[i] =
+				AMDGPU_ENFORCE_ISOLATION_ENABLE;
+			break;
+		case 2:
+			/* enable legacy mode */
+			adev->enforce_isolation[i] =
+				AMDGPU_ENFORCE_ISOLATION_ENABLE_LEGACY;
+			break;
+		case 3:
+			/* enable only process isolation without submitting cleaner shader */
+			adev->enforce_isolation[i] =
+				AMDGPU_ENFORCE_ISOLATION_NO_CLEANER_SHADER;
+			break;
+		}
 	}
+
+	return 0;
 }
 
 /**
@@ -1101,30 +2186,42 @@ static void amdgpu_check_arguments(struct amdgpu_device *adev)
  * @pdev: pci dev pointer
  * @state: vga_switcheroo state
  *
- * Callback for the switcheroo driver.  Suspends or resumes the
+ * Callback for the switcheroo driver.  Suspends or resumes
  * the asics before or after it is powered up using ACPI methods.
  */
-static void amdgpu_switcheroo_set_state(struct pci_dev *pdev, enum vga_switcheroo_state state)
+static void amdgpu_switcheroo_set_state(struct pci_dev *pdev,
+					enum vga_switcheroo_state state)
 {
 	struct drm_device *dev = pci_get_drvdata(pdev);
+	int r;
 
-	if (amdgpu_device_is_px(dev) && state == VGA_SWITCHEROO_OFF)
+	if (amdgpu_device_supports_px(drm_to_adev(dev)) &&
+	    state == VGA_SWITCHEROO_OFF)
 		return;
 
 	if (state == VGA_SWITCHEROO_ON) {
-		pr_info("amdgpu: switched on\n");
+		pr_info("switched on\n");
 		/* don't suspend or resume card normally */
 		dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
 
-		amdgpu_device_resume(dev, true, true);
+		pci_set_power_state(pdev, PCI_D0);
+		amdgpu_device_load_pci_state(pdev);
+		r = pci_enable_device(pdev);
+		if (r)
+			dev_warn(&pdev->dev, "pci_enable_device failed (%d)\n",
+				 r);
+		amdgpu_device_resume(dev, true);
 
 		dev->switch_power_state = DRM_SWITCH_POWER_ON;
-		drm_kms_helper_poll_enable(dev);
 	} else {
-		pr_info("amdgpu: switched off\n");
-		drm_kms_helper_poll_disable(dev);
+		dev_info(&pdev->dev, "switched off\n");
 		dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
-		amdgpu_device_suspend(dev, true, true);
+		amdgpu_device_prepare(dev);
+		amdgpu_device_suspend(dev, true);
+		amdgpu_device_cache_pci_state(pdev);
+		/* Shut down the device */
+		pci_disable_device(pdev);
+		pci_set_power_state(pdev, PCI_D3cold);
 		dev->switch_power_state = DRM_SWITCH_POWER_OFF;
 	}
 }
@@ -1142,12 +2239,12 @@ static bool amdgpu_switcheroo_can_switch(struct pci_dev *pdev)
 {
 	struct drm_device *dev = pci_get_drvdata(pdev);
 
-	/*
+       /*
 	* FIXME: open_count is protected by drm_global_mutex but that would lead to
 	* locking inversion with the driver load path. And the access here is
 	* completely racy anyway. So don't bother with locking for now.
 	*/
-	return dev->open_count == 0;
+	return atomic_read(&dev->open_count) == 0;
 }
 
 static const struct vga_switcheroo_client_ops amdgpu_switcheroo_ops = {
@@ -1156,10 +2253,22 @@ static const struct vga_switcheroo_client_ops amdgpu_switcheroo_ops = {
 	.can_switch = amdgpu_switcheroo_can_switch,
 };
 
-int amdgpu_set_clockgating_state(struct amdgpu_device *adev,
-				  enum amd_ip_block_type block_type,
-				  enum amd_clockgating_state state)
+/**
+ * amdgpu_device_ip_set_clockgating_state - set the CG state
+ *
+ * @dev: amdgpu_device pointer
+ * @block_type: Type of hardware IP (SMU, GFX, UVD, etc.)
+ * @state: clockgating state (gate or ungate)
+ *
+ * Sets the requested clockgating state for all instances of
+ * the hardware IP specified.
+ * Returns the error code from the last instance.
+ */
+int amdgpu_device_ip_set_clockgating_state(void *dev,
+					   enum amd_ip_block_type block_type,
+					   enum amd_clockgating_state state)
 {
+	struct amdgpu_device *adev = dev;
 	int i, r = 0;
 
 	for (i = 0; i < adev->num_ip_blocks; i++) {
@@ -1170,18 +2279,31 @@ int amdgpu_set_clockgating_state(struct amdgpu_device *adev,
 		if (!adev->ip_blocks[i].version->funcs->set_clockgating_state)
 			continue;
 		r = adev->ip_blocks[i].version->funcs->set_clockgating_state(
-			(void *)adev, state);
+			&adev->ip_blocks[i], state);
 		if (r)
-			DRM_ERROR("set_clockgating_state of IP block <%s> failed %d\n",
-				  adev->ip_blocks[i].version->funcs->name, r);
+			dev_err(adev->dev,
+				"set_clockgating_state of IP block <%s> failed %d\n",
+				adev->ip_blocks[i].version->funcs->name, r);
 	}
 	return r;
 }
 
-int amdgpu_set_powergating_state(struct amdgpu_device *adev,
-				  enum amd_ip_block_type block_type,
-				  enum amd_powergating_state state)
+/**
+ * amdgpu_device_ip_set_powergating_state - set the PG state
+ *
+ * @dev: amdgpu_device pointer
+ * @block_type: Type of hardware IP (SMU, GFX, UVD, etc.)
+ * @state: powergating state (gate or ungate)
+ *
+ * Sets the requested powergating state for all instances of
+ * the hardware IP specified.
+ * Returns the error code from the last instance.
+ */
+int amdgpu_device_ip_set_powergating_state(void *dev,
+					   enum amd_ip_block_type block_type,
+					   enum amd_powergating_state state)
 {
+	struct amdgpu_device *adev = dev;
 	int i, r = 0;
 
 	for (i = 0; i < adev->num_ip_blocks; i++) {
@@ -1192,15 +2314,28 @@ int amdgpu_set_powergating_state(struct amdgpu_device *adev,
 		if (!adev->ip_blocks[i].version->funcs->set_powergating_state)
 			continue;
 		r = adev->ip_blocks[i].version->funcs->set_powergating_state(
-			(void *)adev, state);
+			&adev->ip_blocks[i], state);
 		if (r)
-			DRM_ERROR("set_powergating_state of IP block <%s> failed %d\n",
-				  adev->ip_blocks[i].version->funcs->name, r);
+			dev_err(adev->dev,
+				"set_powergating_state of IP block <%s> failed %d\n",
+				adev->ip_blocks[i].version->funcs->name, r);
 	}
 	return r;
 }
 
-void amdgpu_get_clockgating_state(struct amdgpu_device *adev, u32 *flags)
+/**
+ * amdgpu_device_ip_get_clockgating_state - get the CG state
+ *
+ * @adev: amdgpu_device pointer
+ * @flags: clockgating feature flags
+ *
+ * Walks the list of IPs on the device and updates the clockgating
+ * flags for each IP.
+ * Updates @flags with the feature flags for each hardware IP where
+ * clockgating is enabled.
+ */
+void amdgpu_device_ip_get_clockgating_state(struct amdgpu_device *adev,
+					    u64 *flags)
 {
 	int i;
 
@@ -1208,12 +2343,22 @@ void amdgpu_get_clockgating_state(struct amdgpu_device *adev, u32 *flags)
 		if (!adev->ip_blocks[i].status.valid)
 			continue;
 		if (adev->ip_blocks[i].version->funcs->get_clockgating_state)
-			adev->ip_blocks[i].version->funcs->get_clockgating_state((void *)adev, flags);
+			adev->ip_blocks[i].version->funcs->get_clockgating_state(
+				&adev->ip_blocks[i], flags);
 	}
 }
 
-int amdgpu_wait_for_idle(struct amdgpu_device *adev,
-			 enum amd_ip_block_type block_type)
+/**
+ * amdgpu_device_ip_wait_for_idle - wait for idle
+ *
+ * @adev: amdgpu_device pointer
+ * @block_type: Type of hardware IP (SMU, GFX, UVD, etc.)
+ *
+ * Waits for the request hardware IP to be idle.
+ * Returns 0 for success or a negative error code on failure.
+ */
+int amdgpu_device_ip_wait_for_idle(struct amdgpu_device *adev,
+				   enum amd_ip_block_type block_type)
 {
 	int i, r;
 
@@ -1221,9 +2366,12 @@ int amdgpu_wait_for_idle(struct amdgpu_device *adev,
 		if (!adev->ip_blocks[i].status.valid)
 			continue;
 		if (adev->ip_blocks[i].version->type == block_type) {
-			r = adev->ip_blocks[i].version->funcs->wait_for_idle((void *)adev);
-			if (r)
-				return r;
+			if (adev->ip_blocks[i].version->funcs->wait_for_idle) {
+				r = adev->ip_blocks[i].version->funcs->wait_for_idle(
+								&adev->ip_blocks[i]);
+				if (r)
+					return r;
+			}
 			break;
 		}
 	}
@@ -1231,23 +2379,40 @@ int amdgpu_wait_for_idle(struct amdgpu_device *adev,
 
 }
 
-bool amdgpu_is_idle(struct amdgpu_device *adev,
-		    enum amd_ip_block_type block_type)
+/**
+ * amdgpu_device_ip_is_valid - is the hardware IP enabled
+ *
+ * @adev: amdgpu_device pointer
+ * @block_type: Type of hardware IP (SMU, GFX, UVD, etc.)
+ *
+ * Check if the hardware IP is enable or not.
+ * Returns true if it the IP is enable, false if not.
+ */
+bool amdgpu_device_ip_is_valid(struct amdgpu_device *adev,
+			       enum amd_ip_block_type block_type)
 {
 	int i;
 
 	for (i = 0; i < adev->num_ip_blocks; i++) {
-		if (!adev->ip_blocks[i].status.valid)
-			continue;
 		if (adev->ip_blocks[i].version->type == block_type)
-			return adev->ip_blocks[i].version->funcs->is_idle((void *)adev);
+			return adev->ip_blocks[i].status.valid;
 	}
-	return true;
+	return false;
 
 }
 
-struct amdgpu_ip_block * amdgpu_get_ip_block(struct amdgpu_device *adev,
-					     enum amd_ip_block_type type)
+/**
+ * amdgpu_device_ip_get_ip_block - get a hw IP pointer
+ *
+ * @adev: amdgpu_device pointer
+ * @type: Type of hardware IP (SMU, GFX, UVD, etc.)
+ *
+ * Returns a pointer to the hardware IP block structure
+ * if it exists for the asic, otherwise NULL.
+ */
+struct amdgpu_ip_block *
+amdgpu_device_ip_get_ip_block(struct amdgpu_device *adev,
+			      enum amd_ip_block_type type)
 {
 	int i;
 
@@ -1259,7 +2424,7 @@ struct amdgpu_ip_block * amdgpu_get_ip_block(struct amdgpu_device *adev,
 }
 
 /**
- * amdgpu_ip_block_version_cmp
+ * amdgpu_device_ip_block_version_cmp
  *
  * @adev: amdgpu_device pointer
  * @type: enum amd_ip_block_type
@@ -1269,11 +2434,11 @@ struct amdgpu_ip_block * amdgpu_get_ip_block(struct amdgpu_device *adev,
  * return 0 if equal or greater
  * return 1 if smaller or the ip_block doesn't exist
  */
-int amdgpu_ip_block_version_cmp(struct amdgpu_device *adev,
-				enum amd_ip_block_type type,
-				u32 major, u32 minor)
+int amdgpu_device_ip_block_version_cmp(struct amdgpu_device *adev,
+				       enum amd_ip_block_type type,
+				       u32 major, u32 minor)
 {
-	struct amdgpu_ip_block *ip_block = amdgpu_get_ip_block(adev, type);
+	struct amdgpu_ip_block *ip_block = amdgpu_device_ip_get_ip_block(adev, type);
 
 	if (ip_block && ((ip_block->version->major > major) ||
 			((ip_block->version->major == major) &&
@@ -1283,8 +2448,35 @@ int amdgpu_ip_block_version_cmp(struct amdgpu_device *adev,
 	return 1;
 }
 
+static const char *ip_block_names[] = {
+	[AMD_IP_BLOCK_TYPE_COMMON] = "common",
+	[AMD_IP_BLOCK_TYPE_GMC] = "gmc",
+	[AMD_IP_BLOCK_TYPE_IH] = "ih",
+	[AMD_IP_BLOCK_TYPE_SMC] = "smu",
+	[AMD_IP_BLOCK_TYPE_PSP] = "psp",
+	[AMD_IP_BLOCK_TYPE_DCE] = "dce",
+	[AMD_IP_BLOCK_TYPE_GFX] = "gfx",
+	[AMD_IP_BLOCK_TYPE_SDMA] = "sdma",
+	[AMD_IP_BLOCK_TYPE_UVD] = "uvd",
+	[AMD_IP_BLOCK_TYPE_VCE] = "vce",
+	[AMD_IP_BLOCK_TYPE_ACP] = "acp",
+	[AMD_IP_BLOCK_TYPE_VCN] = "vcn",
+	[AMD_IP_BLOCK_TYPE_MES] = "mes",
+	[AMD_IP_BLOCK_TYPE_JPEG] = "jpeg",
+	[AMD_IP_BLOCK_TYPE_VPE] = "vpe",
+	[AMD_IP_BLOCK_TYPE_UMSCH_MM] = "umsch_mm",
+	[AMD_IP_BLOCK_TYPE_ISP] = "isp",
+};
+
+static const char *ip_block_name(struct amdgpu_device *adev, enum amd_ip_block_type type)
+{
+	int idx = (int)type;
+
+	return idx < ARRAY_SIZE(ip_block_names) ? ip_block_names[idx] : "unknown";
+}
+
 /**
- * amdgpu_ip_block_add
+ * amdgpu_device_ip_block_add
  *
  * @adev: amdgpu_device pointer
  * @ip_block_version: pointer to the IP to add
@@ -1292,27 +2484,58 @@ int amdgpu_ip_block_version_cmp(struct amdgpu_device *adev,
  * Adds the IP block driver information to the collection of IPs
  * on the asic.
  */
-int amdgpu_ip_block_add(struct amdgpu_device *adev,
-			const struct amdgpu_ip_block_version *ip_block_version)
+int amdgpu_device_ip_block_add(struct amdgpu_device *adev,
+			       const struct amdgpu_ip_block_version *ip_block_version)
 {
 	if (!ip_block_version)
 		return -EINVAL;
 
-	DRM_DEBUG("add ip block number %d <%s>\n", adev->num_ip_blocks,
-		  ip_block_version->funcs->name);
+	switch (ip_block_version->type) {
+	case AMD_IP_BLOCK_TYPE_VCN:
+		if (adev->harvest_ip_mask & AMD_HARVEST_IP_VCN_MASK)
+			return 0;
+		break;
+	case AMD_IP_BLOCK_TYPE_JPEG:
+		if (adev->harvest_ip_mask & AMD_HARVEST_IP_JPEG_MASK)
+			return 0;
+		break;
+	default:
+		break;
+	}
+
+	dev_info(adev->dev, "detected ip block number %d <%s_v%d_%d_%d> (%s)\n",
+		 adev->num_ip_blocks,
+		 ip_block_name(adev, ip_block_version->type),
+		 ip_block_version->major,
+		 ip_block_version->minor,
+		 ip_block_version->rev,
+		 ip_block_version->funcs->name);
+
+	adev->ip_blocks[adev->num_ip_blocks].adev = adev;
 
 	adev->ip_blocks[adev->num_ip_blocks++].version = ip_block_version;
 
 	return 0;
 }
 
+/**
+ * amdgpu_device_enable_virtual_display - enable virtual display feature
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Enabled the virtual display feature if the user has enabled it via
+ * the module parameter virtual_display.  This feature provides a virtual
+ * display hardware on headless boards or in virtualized environments.
+ * This function parses and validates the configuration string specified by
+ * the user and configures the virtual display configuration (number of
+ * virtual connectors, crtcs, etc.) specified.
+ */
 static void amdgpu_device_enable_virtual_display(struct amdgpu_device *adev)
 {
 	adev->enable_virtual_display = false;
 
 	if (amdgpu_virtual_display) {
-		struct drm_device *ddev = adev->ddev;
-		const char *pci_address_name = pci_name(ddev->pdev);
+		const char *pci_address_name = pci_name(adev->pdev);
 		char *pciaddstr, *pciaddstr_tmp, *pciaddname_tmp, *pciaddname;
 
 		pciaddstr = kstrdup(amdgpu_virtual_display, GFP_KERNEL);
@@ -1343,69 +2566,82 @@ static void amdgpu_device_enable_virtual_display(struct amdgpu_device *adev)
 			}
 		}
 
-		DRM_INFO("virtual display string:%s, %s:virtual_display:%d, num_crtc:%d\n",
-			 amdgpu_virtual_display, pci_address_name,
-			 adev->enable_virtual_display, adev->mode_info.num_crtc);
+		dev_info(
+			adev->dev,
+			"virtual display string:%s, %s:virtual_display:%d, num_crtc:%d\n",
+			amdgpu_virtual_display, pci_address_name,
+			adev->enable_virtual_display, adev->mode_info.num_crtc);
 
 		kfree(pciaddstr);
 	}
 }
 
+void amdgpu_device_set_sriov_virtual_display(struct amdgpu_device *adev)
+{
+	if (amdgpu_sriov_vf(adev) && !adev->enable_virtual_display) {
+		adev->mode_info.num_crtc = 1;
+		adev->enable_virtual_display = true;
+		dev_info(adev->dev, "virtual_display:%d, num_crtc:%d\n",
+			 adev->enable_virtual_display,
+			 adev->mode_info.num_crtc);
+	}
+}
+
+/**
+ * amdgpu_device_parse_gpu_info_fw - parse gpu info firmware
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Parses the asic configuration parameters specified in the gpu info
+ * firmware and makes them available to the driver for use in configuring
+ * the asic.
+ * Returns 0 on success, -EINVAL on failure.
+ */
 static int amdgpu_device_parse_gpu_info_fw(struct amdgpu_device *adev)
 {
 	const char *chip_name;
-	char fw_name[30];
 	int err;
 	const struct gpu_info_firmware_header_v1_0 *hdr;
 
 	adev->firmware.gpu_info_fw = NULL;
 
 	switch (adev->asic_type) {
-	case CHIP_TOPAZ:
-	case CHIP_TONGA:
-	case CHIP_FIJI:
-	case CHIP_POLARIS11:
-	case CHIP_POLARIS10:
-	case CHIP_POLARIS12:
-	case CHIP_CARRIZO:
-	case CHIP_STONEY:
-#ifdef CONFIG_DRM_AMDGPU_SI
-	case CHIP_VERDE:
-	case CHIP_TAHITI:
-	case CHIP_PITCAIRN:
-	case CHIP_OLAND:
-	case CHIP_HAINAN:
-#endif
-#ifdef CONFIG_DRM_AMDGPU_CIK
-	case CHIP_BONAIRE:
-	case CHIP_HAWAII:
-	case CHIP_KAVERI:
-	case CHIP_KABINI:
-	case CHIP_MULLINS:
-#endif
 	default:
 		return 0;
 	case CHIP_VEGA10:
 		chip_name = "vega10";
 		break;
+	case CHIP_VEGA12:
+		chip_name = "vega12";
+		break;
 	case CHIP_RAVEN:
-		chip_name = "raven";
+		if (adev->apu_flags & AMD_APU_IS_RAVEN2)
+			chip_name = "raven2";
+		else if (adev->apu_flags & AMD_APU_IS_PICASSO)
+			chip_name = "picasso";
+		else
+			chip_name = "raven";
+		break;
+	case CHIP_ARCTURUS:
+		chip_name = "arcturus";
+		break;
+	case CHIP_NAVI12:
+		if (adev->mman.discovery_bin)
+			return 0;
+		chip_name = "navi12";
+		break;
+	case CHIP_CYAN_SKILLFISH:
+		chip_name = "cyan_skillfish";
 		break;
 	}
 
-	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_gpu_info.bin", chip_name);
-	err = request_firmware(&adev->firmware.gpu_info_fw, fw_name, adev->dev);
-	if (err) {
-		dev_err(adev->dev,
-			"Failed to load gpu_info firmware \"%s\"\n",
-			fw_name);
-		goto out;
-	}
-	err = amdgpu_ucode_validate(adev->firmware.gpu_info_fw);
+	err = amdgpu_ucode_request(adev, &adev->firmware.gpu_info_fw,
+				   AMDGPU_UCODE_OPTIONAL,
+				   "amdgpu/%s_gpu_info.bin", chip_name);
 	if (err) {
 		dev_err(adev->dev,
-			"Failed to validate gpu_info firmware \"%s\"\n",
-			fw_name);
+			"Failed to get gpu_info firmware \"%s_gpu_info.bin\"\n",
+			chip_name);
 		goto out;
 	}
 
@@ -1419,6 +2655,12 @@ static int amdgpu_device_parse_gpu_info_fw(struct amdgpu_device *adev)
 			(const struct gpu_info_firmware_v1_0 *)(adev->firmware.gpu_info_fw->data +
 								le32_to_cpu(hdr->header.ucode_array_offset_bytes));
 
+		/*
+		 * Should be dropped when DAL no longer needs it.
+		 */
+		if (adev->asic_type == CHIP_NAVI12)
+			goto parse_soc_bounding_box;
+
 		adev->gfx.config.max_shader_engines = le32_to_cpu(gpu_info_fw->gc_num_se);
 		adev->gfx.config.max_cu_per_sh = le32_to_cpu(gpu_info_fw->gc_num_cu_per_sh);
 		adev->gfx.config.max_sh_per_se = le32_to_cpu(gpu_info_fw->gc_num_sh_per_se);
@@ -1437,6 +2679,27 @@ static int amdgpu_device_parse_gpu_info_fw(struct amdgpu_device *adev)
 		adev->gfx.cu_info.max_scratch_slots_per_cu =
 			le32_to_cpu(gpu_info_fw->gc_max_scratch_slots_per_cu);
 		adev->gfx.cu_info.lds_size = le32_to_cpu(gpu_info_fw->gc_lds_size);
+		if (hdr->version_minor >= 1) {
+			const struct gpu_info_firmware_v1_1 *gpu_info_fw =
+				(const struct gpu_info_firmware_v1_1 *)(adev->firmware.gpu_info_fw->data +
+									le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			adev->gfx.config.num_sc_per_sh =
+				le32_to_cpu(gpu_info_fw->num_sc_per_sh);
+			adev->gfx.config.num_packer_per_sc =
+				le32_to_cpu(gpu_info_fw->num_packer_per_sc);
+		}
+
+parse_soc_bounding_box:
+		/*
+		 * soc bounding box info is not integrated in disocovery table,
+		 * we always need to parse it from gpu info firmware if needed.
+		 */
+		if (hdr->version_minor == 2) {
+			const struct gpu_info_firmware_v1_2 *gpu_info_fw =
+				(const struct gpu_info_firmware_v1_2 *)(adev->firmware.gpu_info_fw->data +
+									le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			adev->dm.soc_bounding_box = &gpu_info_fw->soc_bounding_box;
+		}
 		break;
 	}
 	default:
@@ -1449,30 +2712,51 @@ out:
 	return err;
 }
 
-static int amdgpu_early_init(struct amdgpu_device *adev)
+static void amdgpu_uid_init(struct amdgpu_device *adev)
 {
+	/* Initialize the UID for the device */
+	adev->uid_info = kzalloc(sizeof(struct amdgpu_uid), GFP_KERNEL);
+	if (!adev->uid_info) {
+		dev_warn(adev->dev, "Failed to allocate memory for UID\n");
+		return;
+	}
+	adev->uid_info->adev = adev;
+}
+
+static void amdgpu_uid_fini(struct amdgpu_device *adev)
+{
+	/* Free the UID memory */
+	kfree(adev->uid_info);
+	adev->uid_info = NULL;
+}
+
+/**
+ * amdgpu_device_ip_early_init - run early init for hardware IPs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Early initialization pass for hardware IPs.  The hardware IPs that make
+ * up each asic are discovered each IP's early_init callback is run.  This
+ * is the first stage in initializing the asic.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_early_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_ip_block *ip_block;
+	struct pci_dev *parent;
+	bool total, skip_bios;
+	uint32_t bios_flags;
 	int i, r;
 
 	amdgpu_device_enable_virtual_display(adev);
 
-	switch (adev->asic_type) {
-	case CHIP_TOPAZ:
-	case CHIP_TONGA:
-	case CHIP_FIJI:
-	case CHIP_POLARIS11:
-	case CHIP_POLARIS10:
-	case CHIP_POLARIS12:
-	case CHIP_CARRIZO:
-	case CHIP_STONEY:
-		if (adev->asic_type == CHIP_CARRIZO || adev->asic_type == CHIP_STONEY)
-			adev->family = AMDGPU_FAMILY_CZ;
-		else
-			adev->family = AMDGPU_FAMILY_VI;
-
-		r = vi_set_ip_blocks(adev);
+	if (amdgpu_sriov_vf(adev)) {
+		r = amdgpu_virt_request_full_gpu(adev, true);
 		if (r)
 			return r;
-		break;
+	}
+
+	switch (adev->asic_type) {
 #ifdef CONFIG_DRM_AMDGPU_SI
 	case CHIP_VERDE:
 	case CHIP_TAHITI:
@@ -1491,64 +2775,149 @@ static int amdgpu_early_init(struct amdgpu_device *adev)
 	case CHIP_KAVERI:
 	case CHIP_KABINI:
 	case CHIP_MULLINS:
-		if ((adev->asic_type == CHIP_BONAIRE) || (adev->asic_type == CHIP_HAWAII))
-			adev->family = AMDGPU_FAMILY_CI;
-		else
+		if (adev->flags & AMD_IS_APU)
 			adev->family = AMDGPU_FAMILY_KV;
+		else
+			adev->family = AMDGPU_FAMILY_CI;
 
 		r = cik_set_ip_blocks(adev);
 		if (r)
 			return r;
 		break;
 #endif
-	case  CHIP_VEGA10:
-	case  CHIP_RAVEN:
-		if (adev->asic_type == CHIP_RAVEN)
-			adev->family = AMDGPU_FAMILY_RV;
+	case CHIP_TOPAZ:
+	case CHIP_TONGA:
+	case CHIP_FIJI:
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+		if (adev->flags & AMD_IS_APU)
+			adev->family = AMDGPU_FAMILY_CZ;
 		else
-			adev->family = AMDGPU_FAMILY_AI;
+			adev->family = AMDGPU_FAMILY_VI;
 
-		r = soc15_set_ip_blocks(adev);
+		r = vi_set_ip_blocks(adev);
 		if (r)
 			return r;
 		break;
 	default:
-		/* FIXME: not supported yet */
-		return -EINVAL;
+		r = amdgpu_discovery_set_ip_blocks(adev);
+		if (r)
+			return r;
+		break;
 	}
 
-	r = amdgpu_device_parse_gpu_info_fw(adev);
-	if (r)
-		return r;
+	/* Check for IP version 9.4.3 with A0 hardware */
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) &&
+	    !amdgpu_device_get_rev_id(adev)) {
+		dev_err(adev->dev, "Unsupported A0 hardware\n");
+		return -ENODEV;	/* device unsupported - no device error */
+	}
+
+	if (amdgpu_has_atpx() &&
+	    (amdgpu_is_atpx_hybrid() ||
+	     amdgpu_has_atpx_dgpu_power_cntl()) &&
+	    ((adev->flags & AMD_IS_APU) == 0) &&
+	    !dev_is_removable(&adev->pdev->dev))
+		adev->flags |= AMD_IS_PX;
 
+	if (!(adev->flags & AMD_IS_APU)) {
+		parent = pcie_find_root_port(adev->pdev);
+		adev->has_pr3 = parent ? pci_pr3_present(parent) : false;
+	}
+
+	adev->pm.pp_feature = amdgpu_pp_feature_mask;
+	if (amdgpu_sriov_vf(adev) || sched_policy == KFD_SCHED_POLICY_NO_HWS)
+		adev->pm.pp_feature &= ~PP_GFXOFF_MASK;
+	if (amdgpu_sriov_vf(adev) && adev->asic_type == CHIP_SIENNA_CICHLID)
+		adev->pm.pp_feature &= ~PP_OVERDRIVE_MASK;
+	if (!amdgpu_device_pcie_dynamic_switching_supported(adev))
+		adev->pm.pp_feature &= ~PP_PCIE_DPM_MASK;
+
+	adev->virt.is_xgmi_node_migrate_enabled = false;
 	if (amdgpu_sriov_vf(adev)) {
-		r = amdgpu_virt_request_full_gpu(adev, true);
-		if (r)
-			return r;
+		adev->virt.is_xgmi_node_migrate_enabled =
+			amdgpu_ip_version((adev), GC_HWIP, 0) == IP_VERSION(9, 4, 4);
 	}
 
+	total = true;
 	for (i = 0; i < adev->num_ip_blocks; i++) {
+		ip_block = &adev->ip_blocks[i];
+
 		if ((amdgpu_ip_block_mask & (1 << i)) == 0) {
-			DRM_ERROR("disabled ip block: %d <%s>\n",
-				  i, adev->ip_blocks[i].version->funcs->name);
+			dev_warn(adev->dev, "disabled ip block: %d <%s>\n", i,
+				 adev->ip_blocks[i].version->funcs->name);
 			adev->ip_blocks[i].status.valid = false;
-		} else {
-			if (adev->ip_blocks[i].version->funcs->early_init) {
-				r = adev->ip_blocks[i].version->funcs->early_init((void *)adev);
-				if (r == -ENOENT) {
-					adev->ip_blocks[i].status.valid = false;
-				} else if (r) {
-					DRM_ERROR("early_init of IP block <%s> failed %d\n",
-						  adev->ip_blocks[i].version->funcs->name, r);
-					return r;
-				} else {
-					adev->ip_blocks[i].status.valid = true;
-				}
+		} else if (ip_block->version->funcs->early_init) {
+			r = ip_block->version->funcs->early_init(ip_block);
+			if (r == -ENOENT) {
+				adev->ip_blocks[i].status.valid = false;
+			} else if (r) {
+				dev_err(adev->dev,
+					"early_init of IP block <%s> failed %d\n",
+					adev->ip_blocks[i].version->funcs->name,
+					r);
+				total = false;
 			} else {
 				adev->ip_blocks[i].status.valid = true;
 			}
+		} else {
+			adev->ip_blocks[i].status.valid = true;
+		}
+		/* get the vbios after the asic_funcs are set up */
+		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_COMMON) {
+			r = amdgpu_device_parse_gpu_info_fw(adev);
+			if (r)
+				return r;
+
+			bios_flags = amdgpu_device_get_vbios_flags(adev);
+			skip_bios = !!(bios_flags & AMDGPU_VBIOS_SKIP);
+			/* Read BIOS */
+			if (!skip_bios) {
+				bool optional =
+					!!(bios_flags & AMDGPU_VBIOS_OPTIONAL);
+				if (!amdgpu_get_bios(adev) && !optional)
+					return -EINVAL;
+
+				if (optional && !adev->bios)
+					dev_info(
+						adev->dev,
+						"VBIOS image optional, proceeding without VBIOS image");
+
+				if (adev->bios) {
+					r = amdgpu_atombios_init(adev);
+					if (r) {
+						dev_err(adev->dev,
+							"amdgpu_atombios_init failed\n");
+						amdgpu_vf_error_put(
+							adev,
+							AMDGIM_ERROR_VF_ATOMBIOS_INIT_FAIL,
+							0, 0);
+						return r;
+					}
+				}
+			}
+
+			/*get pf2vf msg info at it's earliest time*/
+			if (amdgpu_sriov_vf(adev))
+				amdgpu_virt_init_data_exchange(adev);
+
 		}
 	}
+	if (!total)
+		return -ENODEV;
+
+	if (adev->gmc.xgmi.supported)
+		amdgpu_xgmi_early_init(adev);
+
+	if (amdgpu_is_multi_aid(adev))
+		amdgpu_uid_init(adev);
+	ip_block = amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_GFX);
+	if (ip_block->status.valid != false)
+		amdgpu_amdkfd_device_probe(adev);
 
 	adev->cg_flags &= amdgpu_cg_mask;
 	adev->pg_flags &= amdgpu_pg_mask;
@@ -1556,194 +2925,773 @@ static int amdgpu_early_init(struct amdgpu_device *adev)
 	return 0;
 }
 
-static int amdgpu_init(struct amdgpu_device *adev)
+static int amdgpu_device_ip_hw_init_phase1(struct amdgpu_device *adev)
 {
 	int i, r;
 
 	for (i = 0; i < adev->num_ip_blocks; i++) {
-		if (!adev->ip_blocks[i].status.valid)
+		if (!adev->ip_blocks[i].status.sw)
 			continue;
-		r = adev->ip_blocks[i].version->funcs->sw_init((void *)adev);
+		if (adev->ip_blocks[i].status.hw)
+			continue;
+		if (!amdgpu_ip_member_of_hwini(
+			    adev, adev->ip_blocks[i].version->type))
+			continue;
+		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_COMMON ||
+		    (amdgpu_sriov_vf(adev) && (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_PSP)) ||
+		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_IH) {
+			r = adev->ip_blocks[i].version->funcs->hw_init(&adev->ip_blocks[i]);
+			if (r) {
+				dev_err(adev->dev,
+					"hw_init of IP block <%s> failed %d\n",
+					adev->ip_blocks[i].version->funcs->name,
+					r);
+				return r;
+			}
+			adev->ip_blocks[i].status.hw = true;
+		}
+	}
+
+	return 0;
+}
+
+static int amdgpu_device_ip_hw_init_phase2(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.sw)
+			continue;
+		if (adev->ip_blocks[i].status.hw)
+			continue;
+		if (!amdgpu_ip_member_of_hwini(
+			    adev, adev->ip_blocks[i].version->type))
+			continue;
+		r = adev->ip_blocks[i].version->funcs->hw_init(&adev->ip_blocks[i]);
 		if (r) {
-			DRM_ERROR("sw_init of IP block <%s> failed %d\n",
-				  adev->ip_blocks[i].version->funcs->name, r);
+			dev_err(adev->dev,
+				"hw_init of IP block <%s> failed %d\n",
+				adev->ip_blocks[i].version->funcs->name, r);
 			return r;
 		}
+		adev->ip_blocks[i].status.hw = true;
+	}
+
+	return 0;
+}
+
+static int amdgpu_device_fw_loading(struct amdgpu_device *adev)
+{
+	int r = 0;
+	int i;
+	uint32_t smu_version;
+
+	if (adev->asic_type >= CHIP_VEGA10) {
+		for (i = 0; i < adev->num_ip_blocks; i++) {
+			if (adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_PSP)
+				continue;
+
+			if (!amdgpu_ip_member_of_hwini(adev,
+						       AMD_IP_BLOCK_TYPE_PSP))
+				break;
+
+			if (!adev->ip_blocks[i].status.sw)
+				continue;
+
+			/* no need to do the fw loading again if already done*/
+			if (adev->ip_blocks[i].status.hw == true)
+				break;
+
+			if (amdgpu_in_reset(adev) || adev->in_suspend) {
+				r = amdgpu_ip_block_resume(&adev->ip_blocks[i]);
+				if (r)
+					return r;
+			} else {
+				r = adev->ip_blocks[i].version->funcs->hw_init(&adev->ip_blocks[i]);
+				if (r) {
+					dev_err(adev->dev,
+						"hw_init of IP block <%s> failed %d\n",
+						adev->ip_blocks[i]
+							.version->funcs->name,
+						r);
+					return r;
+				}
+				adev->ip_blocks[i].status.hw = true;
+			}
+			break;
+		}
+	}
+
+	if (!amdgpu_sriov_vf(adev) || adev->asic_type == CHIP_TONGA)
+		r = amdgpu_pm_load_smu_firmware(adev, &smu_version);
+
+	return r;
+}
+
+static int amdgpu_device_init_schedulers(struct amdgpu_device *adev)
+{
+	struct drm_sched_init_args args = {
+		.ops = &amdgpu_sched_ops,
+		.num_rqs = DRM_SCHED_PRIORITY_COUNT,
+		.timeout_wq = adev->reset_domain->wq,
+		.dev = adev->dev,
+	};
+	long timeout;
+	int r, i;
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		struct amdgpu_ring *ring = adev->rings[i];
+
+		/* No need to setup the GPU scheduler for rings that don't need it */
+		if (!ring || ring->no_scheduler)
+			continue;
+
+		switch (ring->funcs->type) {
+		case AMDGPU_RING_TYPE_GFX:
+			timeout = adev->gfx_timeout;
+			break;
+		case AMDGPU_RING_TYPE_COMPUTE:
+			timeout = adev->compute_timeout;
+			break;
+		case AMDGPU_RING_TYPE_SDMA:
+			timeout = adev->sdma_timeout;
+			break;
+		default:
+			timeout = adev->video_timeout;
+			break;
+		}
+
+		args.timeout = timeout;
+		args.credit_limit = ring->num_hw_submission;
+		args.score = ring->sched_score;
+		args.name = ring->name;
+
+		r = drm_sched_init(&ring->sched, &args);
+		if (r) {
+			dev_err(adev->dev,
+				"Failed to create scheduler on ring %s.\n",
+				ring->name);
+			return r;
+		}
+		r = amdgpu_uvd_entity_init(adev, ring);
+		if (r) {
+			dev_err(adev->dev,
+				"Failed to create UVD scheduling entity on ring %s.\n",
+				ring->name);
+			return r;
+		}
+		r = amdgpu_vce_entity_init(adev, ring);
+		if (r) {
+			dev_err(adev->dev,
+				"Failed to create VCE scheduling entity on ring %s.\n",
+				ring->name);
+			return r;
+		}
+	}
+
+	if (adev->xcp_mgr)
+		amdgpu_xcp_update_partition_sched_list(adev);
+
+	return 0;
+}
+
+
+/**
+ * amdgpu_device_ip_init - run init for hardware IPs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Main initialization pass for hardware IPs.  The list of all the hardware
+ * IPs that make up the asic is walked and the sw_init and hw_init callbacks
+ * are run.  sw_init initializes the software state associated with each IP
+ * and hw_init initializes the hardware associated with each IP.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_init(struct amdgpu_device *adev)
+{
+	bool init_badpage;
+	int i, r;
+
+	r = amdgpu_ras_init(adev);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		if (adev->ip_blocks[i].version->funcs->sw_init) {
+			r = adev->ip_blocks[i].version->funcs->sw_init(&adev->ip_blocks[i]);
+			if (r) {
+				dev_err(adev->dev,
+					"sw_init of IP block <%s> failed %d\n",
+					adev->ip_blocks[i].version->funcs->name,
+					r);
+				goto init_failed;
+			}
+		}
 		adev->ip_blocks[i].status.sw = true;
-		/* need to do gmc hw init early so we can allocate gpu mem */
-		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GMC) {
-			r = amdgpu_vram_scratch_init(adev);
+
+		if (!amdgpu_ip_member_of_hwini(
+			    adev, adev->ip_blocks[i].version->type))
+			continue;
+
+		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_COMMON) {
+			/* need to do common hw init early so everything is set up for gmc */
+			r = adev->ip_blocks[i].version->funcs->hw_init(&adev->ip_blocks[i]);
 			if (r) {
-				DRM_ERROR("amdgpu_vram_scratch_init failed %d\n", r);
-				return r;
+				dev_err(adev->dev, "hw_init %d failed %d\n", i,
+					r);
+				goto init_failed;
 			}
-			r = adev->ip_blocks[i].version->funcs->hw_init((void *)adev);
+			adev->ip_blocks[i].status.hw = true;
+		} else if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GMC) {
+			/* need to do gmc hw init early so we can allocate gpu mem */
+			/* Try to reserve bad pages early */
+			if (amdgpu_sriov_vf(adev))
+				amdgpu_virt_exchange_data(adev);
+
+			r = amdgpu_device_mem_scratch_init(adev);
 			if (r) {
-				DRM_ERROR("hw_init %d failed %d\n", i, r);
-				return r;
+				dev_err(adev->dev,
+					"amdgpu_mem_scratch_init failed %d\n",
+					r);
+				goto init_failed;
 			}
-			r = amdgpu_wb_init(adev);
+			r = adev->ip_blocks[i].version->funcs->hw_init(&adev->ip_blocks[i]);
 			if (r) {
-				DRM_ERROR("amdgpu_wb_init failed %d\n", r);
-				return r;
+				dev_err(adev->dev, "hw_init %d failed %d\n", i,
+					r);
+				goto init_failed;
+			}
+			r = amdgpu_device_wb_init(adev);
+			if (r) {
+				dev_err(adev->dev,
+					"amdgpu_device_wb_init failed %d\n", r);
+				goto init_failed;
 			}
 			adev->ip_blocks[i].status.hw = true;
 
 			/* right after GMC hw init, we create CSA */
-			if (amdgpu_sriov_vf(adev)) {
-				r = amdgpu_allocate_static_csa(adev);
+			if (adev->gfx.mcbp) {
+				r = amdgpu_allocate_static_csa(adev, &adev->virt.csa_obj,
+							       AMDGPU_GEM_DOMAIN_VRAM |
+							       AMDGPU_GEM_DOMAIN_GTT,
+							       AMDGPU_CSA_SIZE);
 				if (r) {
-					DRM_ERROR("allocate CSA failed %d\n", r);
-					return r;
+					dev_err(adev->dev,
+						"allocate CSA failed %d\n", r);
+					goto init_failed;
 				}
 			}
+
+			r = amdgpu_seq64_init(adev);
+			if (r) {
+				dev_err(adev->dev, "allocate seq64 failed %d\n",
+					r);
+				goto init_failed;
+			}
 		}
 	}
 
-	for (i = 0; i < adev->num_ip_blocks; i++) {
-		if (!adev->ip_blocks[i].status.sw)
-			continue;
-		/* gmc hw init is done early */
-		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GMC)
-			continue;
-		r = adev->ip_blocks[i].version->funcs->hw_init((void *)adev);
-		if (r) {
-			DRM_ERROR("hw_init of IP block <%s> failed %d\n",
-				  adev->ip_blocks[i].version->funcs->name, r);
-			return r;
+	if (amdgpu_sriov_vf(adev))
+		amdgpu_virt_init_data_exchange(adev);
+
+	r = amdgpu_ib_pool_init(adev);
+	if (r) {
+		dev_err(adev->dev, "IB initialization failed (%d).\n", r);
+		amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_IB_INIT_FAIL, 0, r);
+		goto init_failed;
+	}
+
+	r = amdgpu_ucode_create_bo(adev); /* create ucode bo when sw_init complete*/
+	if (r)
+		goto init_failed;
+
+	r = amdgpu_device_ip_hw_init_phase1(adev);
+	if (r)
+		goto init_failed;
+
+	r = amdgpu_device_fw_loading(adev);
+	if (r)
+		goto init_failed;
+
+	r = amdgpu_device_ip_hw_init_phase2(adev);
+	if (r)
+		goto init_failed;
+
+	/*
+	 * retired pages will be loaded from eeprom and reserved here,
+	 * it should be called after amdgpu_device_ip_hw_init_phase2  since
+	 * for some ASICs the RAS EEPROM code relies on SMU fully functioning
+	 * for I2C communication which only true at this point.
+	 *
+	 * amdgpu_ras_recovery_init may fail, but the upper only cares the
+	 * failure from bad gpu situation and stop amdgpu init process
+	 * accordingly. For other failed cases, it will still release all
+	 * the resource and print error message, rather than returning one
+	 * negative value to upper level.
+	 *
+	 * Note: theoretically, this should be called before all vram allocations
+	 * to protect retired page from abusing
+	 */
+	init_badpage = (adev->init_lvl->level != AMDGPU_INIT_LEVEL_MINIMAL_XGMI);
+	r = amdgpu_ras_recovery_init(adev, init_badpage);
+	if (r)
+		goto init_failed;
+
+	/**
+	 * In case of XGMI grab extra reference for reset domain for this device
+	 */
+	if (adev->gmc.xgmi.num_physical_nodes > 1) {
+		if (amdgpu_xgmi_add_device(adev) == 0) {
+			if (!amdgpu_sriov_vf(adev)) {
+				struct amdgpu_hive_info *hive = amdgpu_get_xgmi_hive(adev);
+
+				if (WARN_ON(!hive)) {
+					r = -ENOENT;
+					goto init_failed;
+				}
+
+				if (!hive->reset_domain ||
+				    !amdgpu_reset_get_reset_domain(hive->reset_domain)) {
+					r = -ENOENT;
+					amdgpu_put_xgmi_hive(hive);
+					goto init_failed;
+				}
+
+				/* Drop the early temporary reset domain we created for device */
+				amdgpu_reset_put_reset_domain(adev->reset_domain);
+				adev->reset_domain = hive->reset_domain;
+				amdgpu_put_xgmi_hive(hive);
+			}
 		}
-		adev->ip_blocks[i].status.hw = true;
 	}
 
-	return 0;
+	r = amdgpu_device_init_schedulers(adev);
+	if (r)
+		goto init_failed;
+
+	if (adev->mman.buffer_funcs_ring->sched.ready)
+		amdgpu_ttm_set_buffer_funcs_status(adev, true);
+
+	/* Don't init kfd if whole hive need to be reset during init */
+	if (adev->init_lvl->level != AMDGPU_INIT_LEVEL_MINIMAL_XGMI) {
+		kgd2kfd_init_zone_device(adev);
+		amdgpu_amdkfd_device_init(adev);
+	}
+
+	amdgpu_fru_get_product_info(adev);
+
+	if (!amdgpu_sriov_vf(adev) || amdgpu_sriov_ras_cper_en(adev))
+		r = amdgpu_cper_init(adev);
+
+init_failed:
+
+	return r;
 }
 
-static void amdgpu_fill_reset_magic(struct amdgpu_device *adev)
+/**
+ * amdgpu_device_fill_reset_magic - writes reset magic to gart pointer
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Writes a reset magic value to the gart pointer in VRAM.  The driver calls
+ * this function before a GPU reset.  If the value is retained after a
+ * GPU reset, VRAM has not been lost. Some GPU resets may destroy VRAM contents.
+ */
+static void amdgpu_device_fill_reset_magic(struct amdgpu_device *adev)
 {
 	memcpy(adev->reset_magic, adev->gart.ptr, AMDGPU_RESET_MAGIC_NUM);
 }
 
-static bool amdgpu_check_vram_lost(struct amdgpu_device *adev)
+/**
+ * amdgpu_device_check_vram_lost - check if vram is valid
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Checks the reset magic value written to the gart pointer in VRAM.
+ * The driver calls this after a GPU reset to see if the contents of
+ * VRAM is lost or now.
+ * returns true if vram is lost, false if not.
+ */
+static bool amdgpu_device_check_vram_lost(struct amdgpu_device *adev)
 {
-	return !!memcmp(adev->gart.ptr, adev->reset_magic,
-			AMDGPU_RESET_MAGIC_NUM);
+	if (memcmp(adev->gart.ptr, adev->reset_magic,
+			AMDGPU_RESET_MAGIC_NUM))
+		return true;
+
+	if (!amdgpu_in_reset(adev))
+		return false;
+
+	/*
+	 * For all ASICs with baco/mode1 reset, the VRAM is
+	 * always assumed to be lost.
+	 */
+	switch (amdgpu_asic_reset_method(adev)) {
+	case AMD_RESET_METHOD_LEGACY:
+	case AMD_RESET_METHOD_LINK:
+	case AMD_RESET_METHOD_BACO:
+	case AMD_RESET_METHOD_MODE1:
+		return true;
+	default:
+		return false;
+	}
 }
 
-static int amdgpu_late_set_cg_state(struct amdgpu_device *adev)
+/**
+ * amdgpu_device_set_cg_state - set clockgating for amdgpu device
+ *
+ * @adev: amdgpu_device pointer
+ * @state: clockgating state (gate or ungate)
+ *
+ * The list of all the hardware IPs that make up the asic is walked and the
+ * set_clockgating_state callbacks are run.
+ * Late initialization pass enabling clockgating for hardware IPs.
+ * Fini or suspend, pass disabling clockgating for hardware IPs.
+ * Returns 0 on success, negative error code on failure.
+ */
+
+int amdgpu_device_set_cg_state(struct amdgpu_device *adev,
+			       enum amd_clockgating_state state)
 {
-	int i = 0, r;
+	int i, j, r;
 
-	for (i = 0; i < adev->num_ip_blocks; i++) {
-		if (!adev->ip_blocks[i].status.valid)
+	if (amdgpu_emu_mode == 1)
+		return 0;
+
+	for (j = 0; j < adev->num_ip_blocks; j++) {
+		i = state == AMD_CG_STATE_GATE ? j : adev->num_ip_blocks - j - 1;
+		if (!adev->ip_blocks[i].status.late_initialized)
+			continue;
+		/* skip CG for GFX, SDMA on S0ix */
+		if (adev->in_s0ix &&
+		    (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GFX ||
+		     adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SDMA))
 			continue;
 		/* skip CG for VCE/UVD, it's handled specially */
 		if (adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_UVD &&
-		    adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VCE) {
+		    adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VCE &&
+		    adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VCN &&
+		    adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_JPEG &&
+		    adev->ip_blocks[i].version->funcs->set_clockgating_state) {
 			/* enable clockgating to save power */
-			r = adev->ip_blocks[i].version->funcs->set_clockgating_state((void *)adev,
-										     AMD_CG_STATE_GATE);
+			r = adev->ip_blocks[i].version->funcs->set_clockgating_state(&adev->ip_blocks[i],
+										     state);
 			if (r) {
-				DRM_ERROR("set_clockgating_state(gate) of IP block <%s> failed %d\n",
-					  adev->ip_blocks[i].version->funcs->name, r);
+				dev_err(adev->dev,
+					"set_clockgating_state(gate) of IP block <%s> failed %d\n",
+					adev->ip_blocks[i].version->funcs->name,
+					r);
 				return r;
 			}
 		}
 	}
+
 	return 0;
 }
 
-static int amdgpu_late_init(struct amdgpu_device *adev)
+int amdgpu_device_set_pg_state(struct amdgpu_device *adev,
+			       enum amd_powergating_state state)
 {
-	int i = 0, r;
+	int i, j, r;
 
-	for (i = 0; i < adev->num_ip_blocks; i++) {
-		if (!adev->ip_blocks[i].status.valid)
+	if (amdgpu_emu_mode == 1)
+		return 0;
+
+	for (j = 0; j < adev->num_ip_blocks; j++) {
+		i = state == AMD_PG_STATE_GATE ? j : adev->num_ip_blocks - j - 1;
+		if (!adev->ip_blocks[i].status.late_initialized)
 			continue;
-		if (adev->ip_blocks[i].version->funcs->late_init) {
-			r = adev->ip_blocks[i].version->funcs->late_init((void *)adev);
+		/* skip PG for GFX, SDMA on S0ix */
+		if (adev->in_s0ix &&
+		    (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GFX ||
+		     adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SDMA))
+			continue;
+		/* skip CG for VCE/UVD, it's handled specially */
+		if (adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_UVD &&
+		    adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VCE &&
+		    adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VCN &&
+		    adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_JPEG &&
+		    adev->ip_blocks[i].version->funcs->set_powergating_state) {
+			/* enable powergating to save power */
+			r = adev->ip_blocks[i].version->funcs->set_powergating_state(&adev->ip_blocks[i],
+											state);
 			if (r) {
-				DRM_ERROR("late_init of IP block <%s> failed %d\n",
-					  adev->ip_blocks[i].version->funcs->name, r);
+				dev_err(adev->dev,
+					"set_powergating_state(gate) of IP block <%s> failed %d\n",
+					adev->ip_blocks[i].version->funcs->name,
+					r);
 				return r;
 			}
-			adev->ip_blocks[i].status.late_initialized = true;
 		}
 	}
+	return 0;
+}
 
-	mod_delayed_work(system_wq, &adev->late_init_work,
-			msecs_to_jiffies(AMDGPU_RESUME_MS));
+static int amdgpu_device_enable_mgpu_fan_boost(void)
+{
+	struct amdgpu_gpu_instance *gpu_ins;
+	struct amdgpu_device *adev;
+	int i, ret = 0;
 
-	amdgpu_fill_reset_magic(adev);
+	mutex_lock(&mgpu_info.mutex);
 
-	return 0;
+	/*
+	 * MGPU fan boost feature should be enabled
+	 * only when there are two or more dGPUs in
+	 * the system
+	 */
+	if (mgpu_info.num_dgpu < 2)
+		goto out;
+
+	for (i = 0; i < mgpu_info.num_dgpu; i++) {
+		gpu_ins = &(mgpu_info.gpu_ins[i]);
+		adev = gpu_ins->adev;
+		if (!(adev->flags & AMD_IS_APU || amdgpu_sriov_multi_vf_mode(adev)) &&
+		    !gpu_ins->mgpu_fan_enabled) {
+			ret = amdgpu_dpm_enable_mgpu_fan_boost(adev);
+			if (ret)
+				break;
+
+			gpu_ins->mgpu_fan_enabled = 1;
+		}
+	}
+
+out:
+	mutex_unlock(&mgpu_info.mutex);
+
+	return ret;
 }
 
-static int amdgpu_fini(struct amdgpu_device *adev)
+/**
+ * amdgpu_device_ip_late_init - run late init for hardware IPs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Late initialization pass for hardware IPs.  The list of all the hardware
+ * IPs that make up the asic is walked and the late_init callbacks are run.
+ * late_init covers any special initialization that an IP requires
+ * after all of the have been initialized or something that needs to happen
+ * late in the init process.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_late_init(struct amdgpu_device *adev)
 {
-	int i, r;
+	struct amdgpu_gpu_instance *gpu_instance;
+	int i = 0, r;
 
-	/* need to disable SMC first */
 	for (i = 0; i < adev->num_ip_blocks; i++) {
 		if (!adev->ip_blocks[i].status.hw)
 			continue;
-		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SMC) {
-			/* ungate blocks before hw fini so that we can shutdown the blocks safely */
-			r = adev->ip_blocks[i].version->funcs->set_clockgating_state((void *)adev,
-										     AMD_CG_STATE_UNGATE);
+		if (adev->ip_blocks[i].version->funcs->late_init) {
+			r = adev->ip_blocks[i].version->funcs->late_init(&adev->ip_blocks[i]);
 			if (r) {
-				DRM_ERROR("set_clockgating_state(ungate) of IP block <%s> failed %d\n",
-					  adev->ip_blocks[i].version->funcs->name, r);
+				dev_err(adev->dev,
+					"late_init of IP block <%s> failed %d\n",
+					adev->ip_blocks[i].version->funcs->name,
+					r);
 				return r;
 			}
-			r = adev->ip_blocks[i].version->funcs->hw_fini((void *)adev);
-			/* XXX handle errors */
-			if (r) {
-				DRM_DEBUG("hw_fini of IP block <%s> failed %d\n",
-					  adev->ip_blocks[i].version->funcs->name, r);
+		}
+		adev->ip_blocks[i].status.late_initialized = true;
+	}
+
+	r = amdgpu_ras_late_init(adev);
+	if (r) {
+		dev_err(adev->dev, "amdgpu_ras_late_init failed %d", r);
+		return r;
+	}
+
+	if (!amdgpu_reset_in_recovery(adev))
+		amdgpu_ras_set_error_query_ready(adev, true);
+
+	amdgpu_device_set_cg_state(adev, AMD_CG_STATE_GATE);
+	amdgpu_device_set_pg_state(adev, AMD_PG_STATE_GATE);
+
+	amdgpu_device_fill_reset_magic(adev);
+
+	r = amdgpu_device_enable_mgpu_fan_boost();
+	if (r)
+		dev_err(adev->dev, "enable mgpu fan boost failed (%d).\n", r);
+
+	/* For passthrough configuration on arcturus and aldebaran, enable special handling SBR */
+	if (amdgpu_passthrough(adev) &&
+	    ((adev->asic_type == CHIP_ARCTURUS && adev->gmc.xgmi.num_physical_nodes > 1) ||
+	     adev->asic_type == CHIP_ALDEBARAN))
+		amdgpu_dpm_handle_passthrough_sbr(adev, true);
+
+	if (adev->gmc.xgmi.num_physical_nodes > 1) {
+		mutex_lock(&mgpu_info.mutex);
+
+		/*
+		 * Reset device p-state to low as this was booted with high.
+		 *
+		 * This should be performed only after all devices from the same
+		 * hive get initialized.
+		 *
+		 * However, it's unknown how many device in the hive in advance.
+		 * As this is counted one by one during devices initializations.
+		 *
+		 * So, we wait for all XGMI interlinked devices initialized.
+		 * This may bring some delays as those devices may come from
+		 * different hives. But that should be OK.
+		 */
+		if (mgpu_info.num_dgpu == adev->gmc.xgmi.num_physical_nodes) {
+			for (i = 0; i < mgpu_info.num_gpu; i++) {
+				gpu_instance = &(mgpu_info.gpu_ins[i]);
+				if (gpu_instance->adev->flags & AMD_IS_APU)
+					continue;
+
+				r = amdgpu_xgmi_set_pstate(gpu_instance->adev,
+						AMDGPU_XGMI_PSTATE_MIN);
+				if (r) {
+					dev_err(adev->dev,
+						"pstate setting failed (%d).\n",
+						r);
+					break;
+				}
 			}
-			adev->ip_blocks[i].status.hw = false;
-			break;
 		}
+
+		mutex_unlock(&mgpu_info.mutex);
 	}
 
-	for (i = adev->num_ip_blocks - 1; i >= 0; i--) {
+	return 0;
+}
+
+static void amdgpu_ip_block_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	if (!ip_block->version->funcs->hw_fini) {
+		dev_err(adev->dev, "hw_fini of IP block <%s> not defined\n",
+			ip_block->version->funcs->name);
+	} else {
+		r = ip_block->version->funcs->hw_fini(ip_block);
+		/* XXX handle errors */
+		if (r) {
+			dev_dbg(adev->dev,
+				"hw_fini of IP block <%s> failed %d\n",
+				ip_block->version->funcs->name, r);
+		}
+	}
+
+	ip_block->status.hw = false;
+}
+
+/**
+ * amdgpu_device_smu_fini_early - smu hw_fini wrapper
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * For ASICs need to disable SMC first
+ */
+static void amdgpu_device_smu_fini_early(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) > IP_VERSION(9, 0, 0))
+		return;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
 		if (!adev->ip_blocks[i].status.hw)
 			continue;
-		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GMC) {
-			amdgpu_wb_fini(adev);
-			amdgpu_vram_scratch_fini(adev);
+		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SMC) {
+			amdgpu_ip_block_hw_fini(&adev->ip_blocks[i]);
+			break;
 		}
+	}
+}
 
-		if (adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_UVD &&
-			adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VCE) {
-			/* ungate blocks before hw fini so that we can shutdown the blocks safely */
-			r = adev->ip_blocks[i].version->funcs->set_clockgating_state((void *)adev,
-										     AMD_CG_STATE_UNGATE);
-			if (r) {
-				DRM_ERROR("set_clockgating_state(ungate) of IP block <%s> failed %d\n",
-					  adev->ip_blocks[i].version->funcs->name, r);
-				return r;
-			}
-		}
+static int amdgpu_device_ip_fini_early(struct amdgpu_device *adev)
+{
+	int i, r;
 
-		r = adev->ip_blocks[i].version->funcs->hw_fini((void *)adev);
-		/* XXX handle errors */
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].version->funcs->early_fini)
+			continue;
+
+		r = adev->ip_blocks[i].version->funcs->early_fini(&adev->ip_blocks[i]);
 		if (r) {
-			DRM_DEBUG("hw_fini of IP block <%s> failed %d\n",
-				  adev->ip_blocks[i].version->funcs->name, r);
+			dev_dbg(adev->dev,
+				"early_fini of IP block <%s> failed %d\n",
+				adev->ip_blocks[i].version->funcs->name, r);
 		}
+	}
 
-		adev->ip_blocks[i].status.hw = false;
+	amdgpu_device_set_pg_state(adev, AMD_PG_STATE_UNGATE);
+	amdgpu_device_set_cg_state(adev, AMD_CG_STATE_UNGATE);
+
+	amdgpu_amdkfd_suspend(adev, true);
+	amdgpu_userq_suspend(adev);
+
+	/* Workaround for ASICs need to disable SMC first */
+	amdgpu_device_smu_fini_early(adev);
+
+	for (i = adev->num_ip_blocks - 1; i >= 0; i--) {
+		if (!adev->ip_blocks[i].status.hw)
+			continue;
+
+		amdgpu_ip_block_hw_fini(&adev->ip_blocks[i]);
 	}
 
+	if (amdgpu_sriov_vf(adev)) {
+		if (amdgpu_virt_release_full_gpu(adev, false))
+			dev_err(adev->dev,
+				"failed to release exclusive mode on fini\n");
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_device_ip_fini - run fini for hardware IPs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Main teardown pass for hardware IPs.  The list of all the hardware
+ * IPs that make up the asic is walked and the hw_fini and sw_fini callbacks
+ * are run.  hw_fini tears down the hardware associated with each IP
+ * and sw_fini tears down any software state associated with each IP.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_fini(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	amdgpu_cper_fini(adev);
+
+	if (amdgpu_sriov_vf(adev) && adev->virt.ras_init_done)
+		amdgpu_virt_release_ras_err_handler_data(adev);
+
+	if (adev->gmc.xgmi.num_physical_nodes > 1)
+		amdgpu_xgmi_remove_device(adev);
+
+	amdgpu_amdkfd_device_fini_sw(adev);
+
 	for (i = adev->num_ip_blocks - 1; i >= 0; i--) {
 		if (!adev->ip_blocks[i].status.sw)
 			continue;
-		r = adev->ip_blocks[i].version->funcs->sw_fini((void *)adev);
-		/* XXX handle errors */
-		if (r) {
-			DRM_DEBUG("sw_fini of IP block <%s> failed %d\n",
-				  adev->ip_blocks[i].version->funcs->name, r);
+
+		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GMC) {
+			amdgpu_ucode_free_bo(adev);
+			amdgpu_free_static_csa(&adev->virt.csa_obj);
+			amdgpu_device_wb_fini(adev);
+			amdgpu_device_mem_scratch_fini(adev);
+			amdgpu_ib_pool_fini(adev);
+			amdgpu_seq64_fini(adev);
+			amdgpu_doorbell_fini(adev);
+		}
+		if (adev->ip_blocks[i].version->funcs->sw_fini) {
+			r = adev->ip_blocks[i].version->funcs->sw_fini(&adev->ip_blocks[i]);
+			/* XXX handle errors */
+			if (r) {
+				dev_dbg(adev->dev,
+					"sw_fini of IP block <%s> failed %d\n",
+					adev->ip_blocks[i].version->funcs->name,
+					r);
+			}
 		}
 		adev->ip_blocks[i].status.sw = false;
 		adev->ip_blocks[i].status.valid = false;
@@ -1753,201 +3701,693 @@ static int amdgpu_fini(struct amdgpu_device *adev)
 		if (!adev->ip_blocks[i].status.late_initialized)
 			continue;
 		if (adev->ip_blocks[i].version->funcs->late_fini)
-			adev->ip_blocks[i].version->funcs->late_fini((void *)adev);
+			adev->ip_blocks[i].version->funcs->late_fini(&adev->ip_blocks[i]);
 		adev->ip_blocks[i].status.late_initialized = false;
 	}
 
-	if (amdgpu_sriov_vf(adev)) {
-		amdgpu_bo_free_kernel(&adev->virt.csa_obj, &adev->virt.csa_vmid0_addr, NULL);
-		amdgpu_virt_release_full_gpu(adev, false);
-	}
+	amdgpu_ras_fini(adev);
+	amdgpu_uid_fini(adev);
 
 	return 0;
 }
 
-static void amdgpu_late_init_func_handler(struct work_struct *work)
+/**
+ * amdgpu_device_delayed_init_work_handler - work handler for IB tests
+ *
+ * @work: work_struct.
+ */
+static void amdgpu_device_delayed_init_work_handler(struct work_struct *work)
 {
 	struct amdgpu_device *adev =
-		container_of(work, struct amdgpu_device, late_init_work.work);
-	amdgpu_late_set_cg_state(adev);
+		container_of(work, struct amdgpu_device, delayed_init_work.work);
+	int r;
+
+	r = amdgpu_ib_ring_tests(adev);
+	if (r)
+		dev_err(adev->dev, "ib ring test failed (%d).\n", r);
+}
+
+static void amdgpu_device_delay_enable_gfx_off(struct work_struct *work)
+{
+	struct amdgpu_device *adev =
+		container_of(work, struct amdgpu_device, gfx.gfx_off_delay_work.work);
+
+	WARN_ON_ONCE(adev->gfx.gfx_off_state);
+	WARN_ON_ONCE(adev->gfx.gfx_off_req_count);
+
+	if (!amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_GFX, true, 0))
+		adev->gfx.gfx_off_state = true;
 }
 
-int amdgpu_suspend(struct amdgpu_device *adev)
+/**
+ * amdgpu_device_ip_suspend_phase1 - run suspend for hardware IPs (phase 1)
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Main suspend function for hardware IPs.  The list of all the hardware
+ * IPs that make up the asic is walked, clockgating is disabled and the
+ * suspend callbacks are run.  suspend puts the hardware and software state
+ * in each IP into a state suitable for suspend.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_suspend_phase1(struct amdgpu_device *adev)
 {
 	int i, r;
 
-	if (amdgpu_sriov_vf(adev))
-		amdgpu_virt_request_full_gpu(adev, false);
+	amdgpu_device_set_pg_state(adev, AMD_PG_STATE_UNGATE);
+	amdgpu_device_set_cg_state(adev, AMD_CG_STATE_UNGATE);
 
-	/* ungate SMC block first */
-	r = amdgpu_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_SMC,
-					 AMD_CG_STATE_UNGATE);
-	if (r) {
-		DRM_ERROR("set_clockgating_state(ungate) SMC failed %d\n",r);
+	/*
+	 * Per PMFW team's suggestion, driver needs to handle gfxoff
+	 * and df cstate features disablement for gpu reset(e.g. Mode1Reset)
+	 * scenario. Add the missing df cstate disablement here.
+	 */
+	if (amdgpu_dpm_set_df_cstate(adev, DF_CSTATE_DISALLOW))
+		dev_warn(adev->dev, "Failed to disallow df cstate");
+
+	for (i = adev->num_ip_blocks - 1; i >= 0; i--) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+
+		/* displays are handled separately */
+		if (adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_DCE)
+			continue;
+
+		/* XXX handle errors */
+		r = amdgpu_ip_block_suspend(&adev->ip_blocks[i]);
+		if (r)
+			return r;
 	}
 
+	return 0;
+}
+
+/**
+ * amdgpu_device_ip_suspend_phase2 - run suspend for hardware IPs (phase 2)
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Main suspend function for hardware IPs.  The list of all the hardware
+ * IPs that make up the asic is walked, clockgating is disabled and the
+ * suspend callbacks are run.  suspend puts the hardware and software state
+ * in each IP into a state suitable for suspend.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	if (adev->in_s0ix)
+		amdgpu_dpm_gfx_state_change(adev, sGpuChangeState_D3Entry);
+
 	for (i = adev->num_ip_blocks - 1; i >= 0; i--) {
 		if (!adev->ip_blocks[i].status.valid)
 			continue;
-		/* ungate blocks so that suspend can properly shut them down */
-		if (i != AMD_IP_BLOCK_TYPE_SMC) {
-			r = adev->ip_blocks[i].version->funcs->set_clockgating_state((void *)adev,
-										     AMD_CG_STATE_UNGATE);
-			if (r) {
-				DRM_ERROR("set_clockgating_state(ungate) of IP block <%s> failed %d\n",
-					  adev->ip_blocks[i].version->funcs->name, r);
-			}
+		/* displays are handled in phase1 */
+		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_DCE)
+			continue;
+		/* PSP lost connection when err_event_athub occurs */
+		if (amdgpu_ras_intr_triggered() &&
+		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_PSP) {
+			adev->ip_blocks[i].status.hw = false;
+			continue;
 		}
+
+		/* skip unnecessary suspend if we do not initialize them yet */
+		if (!amdgpu_ip_member_of_hwini(
+			    adev, adev->ip_blocks[i].version->type))
+			continue;
+
+		/* Since we skip suspend for S0i3, we need to cancel the delayed
+		 * idle work here as the suspend callback never gets called.
+		 */
+		if (adev->in_s0ix &&
+		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GFX &&
+		    amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(10, 0, 0))
+			cancel_delayed_work_sync(&adev->gfx.idle_work);
+		/* skip suspend of gfx/mes and psp for S0ix
+		 * gfx is in gfxoff state, so on resume it will exit gfxoff just
+		 * like at runtime. PSP is also part of the always on hardware
+		 * so no need to suspend it.
+		 */
+		if (adev->in_s0ix &&
+		    (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_PSP ||
+		     adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GFX ||
+		     adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_MES))
+			continue;
+
+		/* SDMA 5.x+ is part of GFX power domain so it's covered by GFXOFF */
+		if (adev->in_s0ix &&
+		    (amdgpu_ip_version(adev, SDMA0_HWIP, 0) >=
+		     IP_VERSION(5, 0, 0)) &&
+		    (adev->ip_blocks[i].version->type ==
+		     AMD_IP_BLOCK_TYPE_SDMA))
+			continue;
+
+		/* Once swPSP provides the IMU, RLC FW binaries to TOS during cold-boot.
+		 * These are in TMR, hence are expected to be reused by PSP-TOS to reload
+		 * from this location and RLC Autoload automatically also gets loaded
+		 * from here based on PMFW -> PSP message during re-init sequence.
+		 * Therefore, the psp suspend & resume should be skipped to avoid destroy
+		 * the TMR and reload FWs again for IMU enabled APU ASICs.
+		 */
+		if (amdgpu_in_reset(adev) &&
+		    (adev->flags & AMD_IS_APU) && adev->gfx.imu.funcs &&
+		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_PSP)
+			continue;
+
 		/* XXX handle errors */
-		r = adev->ip_blocks[i].version->funcs->suspend(adev);
-		/* XXX handle errors */
-		if (r) {
-			DRM_ERROR("suspend of IP block <%s> failed %d\n",
-				  adev->ip_blocks[i].version->funcs->name, r);
+		r = amdgpu_ip_block_suspend(&adev->ip_blocks[i]);
+		adev->ip_blocks[i].status.hw = false;
+
+		/* handle putting the SMC in the appropriate state */
+		if (!amdgpu_sriov_vf(adev)) {
+			if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SMC) {
+				r = amdgpu_dpm_set_mp1_state(adev, adev->mp1_state);
+				if (r) {
+					dev_err(adev->dev,
+						"SMC failed to set mp1 state %d, %d\n",
+						adev->mp1_state, r);
+					return r;
+				}
+			}
 		}
 	}
 
+	return 0;
+}
+
+/**
+ * amdgpu_device_ip_suspend - run suspend for hardware IPs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Main suspend function for hardware IPs.  The list of all the hardware
+ * IPs that make up the asic is walked, clockgating is disabled and the
+ * suspend callbacks are run.  suspend puts the hardware and software state
+ * in each IP into a state suitable for suspend.
+ * Returns 0 on success, negative error code on failure.
+ */
+int amdgpu_device_ip_suspend(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (amdgpu_sriov_vf(adev)) {
+		amdgpu_virt_fini_data_exchange(adev);
+		amdgpu_virt_request_full_gpu(adev, false);
+	}
+
+	amdgpu_ttm_set_buffer_funcs_status(adev, false);
+
+	r = amdgpu_device_ip_suspend_phase1(adev);
+	if (r)
+		return r;
+	r = amdgpu_device_ip_suspend_phase2(adev);
+
 	if (amdgpu_sriov_vf(adev))
 		amdgpu_virt_release_full_gpu(adev, false);
 
-	return 0;
+	return r;
 }
 
-static int amdgpu_sriov_reinit_early(struct amdgpu_device *adev)
+static int amdgpu_device_ip_reinit_early_sriov(struct amdgpu_device *adev)
 {
 	int i, r;
 
 	static enum amd_ip_block_type ip_order[] = {
-		AMD_IP_BLOCK_TYPE_GMC,
 		AMD_IP_BLOCK_TYPE_COMMON,
+		AMD_IP_BLOCK_TYPE_GMC,
+		AMD_IP_BLOCK_TYPE_PSP,
 		AMD_IP_BLOCK_TYPE_IH,
 	};
 
-	for (i = 0; i < ARRAY_SIZE(ip_order); i++) {
+	for (i = 0; i < adev->num_ip_blocks; i++) {
 		int j;
 		struct amdgpu_ip_block *block;
 
-		for (j = 0; j < adev->num_ip_blocks; j++) {
-			block = &adev->ip_blocks[j];
+		block = &adev->ip_blocks[i];
+		block->status.hw = false;
+
+		for (j = 0; j < ARRAY_SIZE(ip_order); j++) {
 
-			if (block->version->type != ip_order[i] ||
+			if (block->version->type != ip_order[j] ||
 				!block->status.valid)
 				continue;
 
-			r = block->version->funcs->hw_init(adev);
-			DRM_INFO("RE-INIT: %s %s\n", block->version->funcs->name, r?"failed":"successed");
+			r = block->version->funcs->hw_init(&adev->ip_blocks[i]);
+			if (r) {
+				dev_err(adev->dev, "RE-INIT-early: %s failed\n",
+					 block->version->funcs->name);
+				return r;
+			}
+			block->status.hw = true;
 		}
 	}
 
 	return 0;
 }
 
-static int amdgpu_sriov_reinit_late(struct amdgpu_device *adev)
+static int amdgpu_device_ip_reinit_late_sriov(struct amdgpu_device *adev)
 {
-	int i, r;
+	struct amdgpu_ip_block *block;
+	int i, r = 0;
 
 	static enum amd_ip_block_type ip_order[] = {
 		AMD_IP_BLOCK_TYPE_SMC,
 		AMD_IP_BLOCK_TYPE_DCE,
 		AMD_IP_BLOCK_TYPE_GFX,
 		AMD_IP_BLOCK_TYPE_SDMA,
+		AMD_IP_BLOCK_TYPE_MES,
 		AMD_IP_BLOCK_TYPE_UVD,
-		AMD_IP_BLOCK_TYPE_VCE
+		AMD_IP_BLOCK_TYPE_VCE,
+		AMD_IP_BLOCK_TYPE_VCN,
+		AMD_IP_BLOCK_TYPE_JPEG
 	};
 
 	for (i = 0; i < ARRAY_SIZE(ip_order); i++) {
-		int j;
-		struct amdgpu_ip_block *block;
+		block = amdgpu_device_ip_get_ip_block(adev, ip_order[i]);
 
-		for (j = 0; j < adev->num_ip_blocks; j++) {
-			block = &adev->ip_blocks[j];
+		if (!block)
+			continue;
 
-			if (block->version->type != ip_order[i] ||
-				!block->status.valid)
-				continue;
+		if (block->status.valid && !block->status.hw) {
+			if (block->version->type == AMD_IP_BLOCK_TYPE_SMC) {
+				r = amdgpu_ip_block_resume(block);
+			} else {
+				r = block->version->funcs->hw_init(block);
+			}
 
-			r = block->version->funcs->hw_init(adev);
-			DRM_INFO("RE-INIT: %s %s\n", block->version->funcs->name, r?"failed":"successed");
+			if (r) {
+				dev_err(adev->dev, "RE-INIT-late: %s failed\n",
+					 block->version->funcs->name);
+				break;
+			}
+			block->status.hw = true;
 		}
 	}
 
-	return 0;
+	return r;
 }
 
-static int amdgpu_resume_phase1(struct amdgpu_device *adev)
+/**
+ * amdgpu_device_ip_resume_phase1 - run resume for hardware IPs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * First resume function for hardware IPs.  The list of all the hardware
+ * IPs that make up the asic is walked and the resume callbacks are run for
+ * COMMON, GMC, and IH.  resume puts the hardware into a functional state
+ * after a suspend and updates the software state as necessary.  This
+ * function is also used for restoring the GPU after a GPU reset.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_resume_phase1(struct amdgpu_device *adev)
 {
 	int i, r;
 
 	for (i = 0; i < adev->num_ip_blocks; i++) {
-		if (!adev->ip_blocks[i].status.valid)
+		if (!adev->ip_blocks[i].status.valid || adev->ip_blocks[i].status.hw)
 			continue;
 		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_COMMON ||
-				adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GMC ||
-				adev->ip_blocks[i].version->type ==
-				AMD_IP_BLOCK_TYPE_IH) {
-			r = adev->ip_blocks[i].version->funcs->resume(adev);
-			if (r) {
-				DRM_ERROR("resume of IP block <%s> failed %d\n",
-					  adev->ip_blocks[i].version->funcs->name, r);
+		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GMC ||
+		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_IH ||
+		    (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_PSP && amdgpu_sriov_vf(adev))) {
+
+			r = amdgpu_ip_block_resume(&adev->ip_blocks[i]);
+			if (r)
 				return r;
-			}
 		}
 	}
 
 	return 0;
 }
 
-static int amdgpu_resume_phase2(struct amdgpu_device *adev)
+/**
+ * amdgpu_device_ip_resume_phase2 - run resume for hardware IPs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Second resume function for hardware IPs.  The list of all the hardware
+ * IPs that make up the asic is walked and the resume callbacks are run for
+ * all blocks except COMMON, GMC, and IH.  resume puts the hardware into a
+ * functional state after a suspend and updates the software state as
+ * necessary.  This function is also used for restoring the GPU after a GPU
+ * reset.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_resume_phase2(struct amdgpu_device *adev)
 {
 	int i, r;
 
 	for (i = 0; i < adev->num_ip_blocks; i++) {
-		if (!adev->ip_blocks[i].status.valid)
+		if (!adev->ip_blocks[i].status.valid || adev->ip_blocks[i].status.hw)
 			continue;
 		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_COMMON ||
-				adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GMC ||
-				adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_IH )
+		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GMC ||
+		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_IH ||
+		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_DCE ||
+		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_PSP)
 			continue;
-		r = adev->ip_blocks[i].version->funcs->resume(adev);
-		if (r) {
-			DRM_ERROR("resume of IP block <%s> failed %d\n",
-				  adev->ip_blocks[i].version->funcs->name, r);
+		r = amdgpu_ip_block_resume(&adev->ip_blocks[i]);
+		if (r)
 			return r;
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_device_ip_resume_phase3 - run resume for hardware IPs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Third resume function for hardware IPs.  The list of all the hardware
+ * IPs that make up the asic is walked and the resume callbacks are run for
+ * all DCE.  resume puts the hardware into a functional state after a suspend
+ * and updates the software state as necessary.  This function is also used
+ * for restoring the GPU after a GPU reset.
+ *
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_resume_phase3(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid || adev->ip_blocks[i].status.hw)
+			continue;
+		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_DCE) {
+			r = amdgpu_ip_block_resume(&adev->ip_blocks[i]);
+			if (r)
+				return r;
 		}
 	}
 
 	return 0;
 }
 
-static int amdgpu_resume(struct amdgpu_device *adev)
+/**
+ * amdgpu_device_ip_resume - run resume for hardware IPs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Main resume function for hardware IPs.  The hardware IPs
+ * are split into two resume functions because they are
+ * also used in recovering from a GPU reset and some additional
+ * steps need to be take between them.  In this case (S3/S4) they are
+ * run sequentially.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_resume(struct amdgpu_device *adev)
 {
 	int r;
 
-	r = amdgpu_resume_phase1(adev);
+	r = amdgpu_device_ip_resume_phase1(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_device_fw_loading(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_device_ip_resume_phase2(adev);
+
+	if (adev->mman.buffer_funcs_ring->sched.ready)
+		amdgpu_ttm_set_buffer_funcs_status(adev, true);
+
 	if (r)
 		return r;
-	r = amdgpu_resume_phase2(adev);
+
+	amdgpu_fence_driver_hw_init(adev);
+
+	r = amdgpu_device_ip_resume_phase3(adev);
 
 	return r;
 }
 
+/**
+ * amdgpu_device_detect_sriov_bios - determine if the board supports SR-IOV
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Query the VBIOS data tables to determine if the board supports SR-IOV.
+ */
 static void amdgpu_device_detect_sriov_bios(struct amdgpu_device *adev)
 {
-	if (adev->is_atom_fw) {
-		if (amdgpu_atomfirmware_gpu_supports_virtualization(adev))
-			adev->virt.caps |= AMDGPU_SRIOV_CAPS_SRIOV_VBIOS;
+	if (amdgpu_sriov_vf(adev)) {
+		if (adev->is_atom_fw) {
+			if (amdgpu_atomfirmware_gpu_virtualization_supported(adev))
+				adev->virt.caps |= AMDGPU_SRIOV_CAPS_SRIOV_VBIOS;
+		} else {
+			if (amdgpu_atombios_has_gpu_virtualization_table(adev))
+				adev->virt.caps |= AMDGPU_SRIOV_CAPS_SRIOV_VBIOS;
+		}
+
+		if (!(adev->virt.caps & AMDGPU_SRIOV_CAPS_SRIOV_VBIOS))
+			amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_NO_VBIOS, 0, 0);
+	}
+}
+
+/**
+ * amdgpu_device_asic_has_dc_support - determine if DC supports the asic
+ *
+ * @pdev : pci device context
+ * @asic_type: AMD asic type
+ *
+ * Check if there is DC (new modesetting infrastructre) support for an asic.
+ * returns true if DC has support, false if not.
+ */
+bool amdgpu_device_asic_has_dc_support(struct pci_dev *pdev,
+				       enum amd_asic_type asic_type)
+{
+	switch (asic_type) {
+#ifdef CONFIG_DRM_AMDGPU_SI
+	case CHIP_HAINAN:
+#endif
+	case CHIP_TOPAZ:
+		/* chips with no display hardware */
+		return false;
+#if defined(CONFIG_DRM_AMD_DC)
+	case CHIP_TAHITI:
+	case CHIP_PITCAIRN:
+	case CHIP_VERDE:
+	case CHIP_OLAND:
+		/*
+		 * We have systems in the wild with these ASICs that require
+		 * LVDS and VGA support which is not supported with DC.
+		 *
+		 * Fallback to the non-DC driver here by default so as not to
+		 * cause regressions.
+		 */
+#if defined(CONFIG_DRM_AMD_DC_SI)
+		return amdgpu_dc > 0;
+#else
+		return false;
+#endif
+	case CHIP_BONAIRE:
+	case CHIP_KAVERI:
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+		/*
+		 * We have systems in the wild with these ASICs that require
+		 * VGA support which is not supported with DC.
+		 *
+		 * Fallback to the non-DC driver here by default so as not to
+		 * cause regressions.
+		 */
+		return amdgpu_dc > 0;
+	default:
+		return amdgpu_dc != 0;
+#else
+	default:
+		if (amdgpu_dc > 0)
+			dev_info_once(
+				&pdev->dev,
+				"Display Core has been requested via kernel parameter but isn't supported by ASIC, ignoring\n");
+		return false;
+#endif
+	}
+}
+
+/**
+ * amdgpu_device_has_dc_support - check if dc is supported
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Returns true for supported, false for not supported
+ */
+bool amdgpu_device_has_dc_support(struct amdgpu_device *adev)
+{
+	if (adev->enable_virtual_display ||
+	    (adev->harvest_ip_mask & AMD_HARVEST_IP_DMU_MASK))
+		return false;
+
+	return amdgpu_device_asic_has_dc_support(adev->pdev, adev->asic_type);
+}
+
+static void amdgpu_device_xgmi_reset_func(struct work_struct *__work)
+{
+	struct amdgpu_device *adev =
+		container_of(__work, struct amdgpu_device, xgmi_reset_work);
+	struct amdgpu_hive_info *hive = amdgpu_get_xgmi_hive(adev);
+
+	/* It's a bug to not have a hive within this function */
+	if (WARN_ON(!hive))
+		return;
+
+	/*
+	 * Use task barrier to synchronize all xgmi reset works across the
+	 * hive. task_barrier_enter and task_barrier_exit will block
+	 * until all the threads running the xgmi reset works reach
+	 * those points. task_barrier_full will do both blocks.
+	 */
+	if (amdgpu_asic_reset_method(adev) == AMD_RESET_METHOD_BACO) {
+
+		task_barrier_enter(&hive->tb);
+		adev->asic_reset_res = amdgpu_device_baco_enter(adev);
+
+		if (adev->asic_reset_res)
+			goto fail;
+
+		task_barrier_exit(&hive->tb);
+		adev->asic_reset_res = amdgpu_device_baco_exit(adev);
+
+		if (adev->asic_reset_res)
+			goto fail;
+
+		amdgpu_ras_reset_error_count(adev, AMDGPU_RAS_BLOCK__MMHUB);
 	} else {
-		if (amdgpu_atombios_has_gpu_virtualization_table(adev))
-			adev->virt.caps |= AMDGPU_SRIOV_CAPS_SRIOV_VBIOS;
+
+		task_barrier_full(&hive->tb);
+		adev->asic_reset_res =  amdgpu_asic_reset(adev);
+	}
+
+fail:
+	if (adev->asic_reset_res)
+		dev_warn(adev->dev,
+			 "ASIC reset failed with error, %d for drm dev, %s",
+			 adev->asic_reset_res, adev_to_drm(adev)->unique);
+	amdgpu_put_xgmi_hive(hive);
+}
+
+static int amdgpu_device_get_job_timeout_settings(struct amdgpu_device *adev)
+{
+	char *input = amdgpu_lockup_timeout;
+	char *timeout_setting = NULL;
+	int index = 0;
+	long timeout;
+	int ret = 0;
+
+	/*
+	 * By default timeout for jobs is 10 sec
+	 */
+	adev->compute_timeout = adev->gfx_timeout = msecs_to_jiffies(10000);
+	adev->sdma_timeout = adev->video_timeout = adev->gfx_timeout;
+
+	if (strnlen(input, AMDGPU_MAX_TIMEOUT_PARAM_LENGTH)) {
+		while ((timeout_setting = strsep(&input, ",")) &&
+				strnlen(timeout_setting, AMDGPU_MAX_TIMEOUT_PARAM_LENGTH)) {
+			ret = kstrtol(timeout_setting, 0, &timeout);
+			if (ret)
+				return ret;
+
+			if (timeout == 0) {
+				index++;
+				continue;
+			} else if (timeout < 0) {
+				timeout = MAX_SCHEDULE_TIMEOUT;
+				dev_warn(adev->dev, "lockup timeout disabled");
+				add_taint(TAINT_SOFTLOCKUP, LOCKDEP_STILL_OK);
+			} else {
+				timeout = msecs_to_jiffies(timeout);
+			}
+
+			switch (index++) {
+			case 0:
+				adev->gfx_timeout = timeout;
+				break;
+			case 1:
+				adev->compute_timeout = timeout;
+				break;
+			case 2:
+				adev->sdma_timeout = timeout;
+				break;
+			case 3:
+				adev->video_timeout = timeout;
+				break;
+			default:
+				break;
+			}
+		}
+		/*
+		 * There is only one value specified and
+		 * it should apply to all non-compute jobs.
+		 */
+		if (index == 1) {
+			adev->sdma_timeout = adev->video_timeout = adev->gfx_timeout;
+			if (amdgpu_sriov_vf(adev) || amdgpu_passthrough(adev))
+				adev->compute_timeout = adev->gfx_timeout;
+		}
 	}
+
+	return ret;
+}
+
+/**
+ * amdgpu_device_check_iommu_direct_map - check if RAM direct mapped to GPU
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * RAM direct mapped to GPU if IOMMU is not enabled or is pass through mode
+ */
+static void amdgpu_device_check_iommu_direct_map(struct amdgpu_device *adev)
+{
+	struct iommu_domain *domain;
+
+	domain = iommu_get_domain_for_dev(adev->dev);
+	if (!domain || domain->type == IOMMU_DOMAIN_IDENTITY)
+		adev->ram_is_direct_mapped = true;
+}
+
+#if defined(CONFIG_HSA_AMD_P2P)
+/**
+ * amdgpu_device_check_iommu_remap - Check if DMA remapping is enabled.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * return if IOMMU remapping bar address
+ */
+static bool amdgpu_device_check_iommu_remap(struct amdgpu_device *adev)
+{
+	struct iommu_domain *domain;
+
+	domain = iommu_get_domain_for_dev(adev->dev);
+	if (domain && (domain->type == IOMMU_DOMAIN_DMA ||
+		domain->type ==	IOMMU_DOMAIN_DMA_FQ))
+		return true;
+
+	return false;
+}
+#endif
+
+static void amdgpu_device_set_mcbp(struct amdgpu_device *adev)
+{
+	if (amdgpu_mcbp == 1)
+		adev->gfx.mcbp = true;
+	else if (amdgpu_mcbp == 0)
+		adev->gfx.mcbp = false;
+
+	if (amdgpu_sriov_vf(adev))
+		adev->gfx.mcbp = true;
+
+	if (adev->gfx.mcbp)
+		dev_info(adev->dev, "MCBP is enabled\n");
 }
 
 /**
  * amdgpu_device_init - initialize the driver
  *
  * @adev: amdgpu_device pointer
- * @pdev: drm dev pointer
- * @pdev: pci dev pointer
  * @flags: driver flags
  *
  * Initializes the driver info and hw (all asics).
@@ -1955,37 +4395,50 @@ static void amdgpu_device_detect_sriov_bios(struct amdgpu_device *adev)
  * Called at driver startup.
  */
 int amdgpu_device_init(struct amdgpu_device *adev,
-		       struct drm_device *ddev,
-		       struct pci_dev *pdev,
 		       uint32_t flags)
 {
+	struct pci_dev *pdev = adev->pdev;
 	int r, i;
-	bool runtime = false;
+	bool px = false;
 	u32 max_MBps;
+	int tmp;
 
 	adev->shutdown = false;
-	adev->dev = &pdev->dev;
-	adev->ddev = ddev;
-	adev->pdev = pdev;
 	adev->flags = flags;
-	adev->asic_type = flags & AMD_ASIC_MASK;
+
+	if (amdgpu_force_asic_type >= 0 && amdgpu_force_asic_type < CHIP_LAST)
+		adev->asic_type = amdgpu_force_asic_type;
+	else
+		adev->asic_type = flags & AMD_ASIC_MASK;
+
 	adev->usec_timeout = AMDGPU_MAX_USEC_TIMEOUT;
-	adev->mc.gart_size = 512 * 1024 * 1024;
+	if (amdgpu_emu_mode == 1)
+		adev->usec_timeout *= 10;
+	adev->gmc.gart_size = 512 * 1024 * 1024;
 	adev->accel_working = false;
 	adev->num_rings = 0;
+	RCU_INIT_POINTER(adev->gang_submit, dma_fence_get_stub());
 	adev->mman.buffer_funcs = NULL;
 	adev->mman.buffer_funcs_ring = NULL;
 	adev->vm_manager.vm_pte_funcs = NULL;
-	adev->vm_manager.vm_pte_num_rings = 0;
-	adev->gart.gart_funcs = NULL;
+	adev->vm_manager.vm_pte_num_scheds = 0;
+	adev->gmc.gmc_funcs = NULL;
+	adev->harvest_ip_mask = 0x0;
 	adev->fence_context = dma_fence_context_alloc(AMDGPU_MAX_RINGS);
+	bitmap_zero(adev->gfx.pipe_reserve_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
 
 	adev->smc_rreg = &amdgpu_invalid_rreg;
 	adev->smc_wreg = &amdgpu_invalid_wreg;
 	adev->pcie_rreg = &amdgpu_invalid_rreg;
 	adev->pcie_wreg = &amdgpu_invalid_wreg;
+	adev->pcie_rreg_ext = &amdgpu_invalid_rreg_ext;
+	adev->pcie_wreg_ext = &amdgpu_invalid_wreg_ext;
 	adev->pciep_rreg = &amdgpu_invalid_rreg;
 	adev->pciep_wreg = &amdgpu_invalid_wreg;
+	adev->pcie_rreg64 = &amdgpu_invalid_rreg64;
+	adev->pcie_wreg64 = &amdgpu_invalid_wreg64;
+	adev->pcie_rreg64_ext = &amdgpu_invalid_rreg64_ext;
+	adev->pcie_wreg64_ext = &amdgpu_invalid_wreg64_ext;
 	adev->uvd_ctx_rreg = &amdgpu_invalid_rreg;
 	adev->uvd_ctx_wreg = &amdgpu_invalid_wreg;
 	adev->didt_rreg = &amdgpu_invalid_rreg;
@@ -1995,23 +4448,48 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	adev->audio_endpt_rreg = &amdgpu_block_invalid_rreg;
 	adev->audio_endpt_wreg = &amdgpu_block_invalid_wreg;
 
-
-	DRM_INFO("initializing kernel modesetting (%s 0x%04X:0x%04X 0x%04X:0x%04X 0x%02X).\n",
-		 amdgpu_asic_name[adev->asic_type], pdev->vendor, pdev->device,
-		 pdev->subsystem_vendor, pdev->subsystem_device, pdev->revision);
+	dev_info(
+		adev->dev,
+		"initializing kernel modesetting (%s 0x%04X:0x%04X 0x%04X:0x%04X 0x%02X).\n",
+		amdgpu_asic_name[adev->asic_type], pdev->vendor, pdev->device,
+		pdev->subsystem_vendor, pdev->subsystem_device, pdev->revision);
 
 	/* mutex initialization are all done here so we
-	 * can recall function without having locking issues */
-	atomic_set(&adev->irq.ih.lock, 0);
+	 * can recall function without having locking issues
+	 */
 	mutex_init(&adev->firmware.mutex);
 	mutex_init(&adev->pm.mutex);
 	mutex_init(&adev->gfx.gpu_clock_mutex);
 	mutex_init(&adev->srbm_mutex);
+	mutex_init(&adev->gfx.pipe_reserve_mutex);
+	mutex_init(&adev->gfx.gfx_off_mutex);
+	mutex_init(&adev->gfx.partition_mutex);
 	mutex_init(&adev->grbm_idx_mutex);
 	mutex_init(&adev->mn_lock);
+	mutex_init(&adev->virt.vf_errors.lock);
 	hash_init(adev->mn_hash);
-
-	amdgpu_check_arguments(adev);
+	mutex_init(&adev->psp.mutex);
+	mutex_init(&adev->notifier_lock);
+	mutex_init(&adev->pm.stable_pstate_ctx_lock);
+	mutex_init(&adev->benchmark_mutex);
+	mutex_init(&adev->gfx.reset_sem_mutex);
+	/* Initialize the mutex for cleaner shader isolation between GFX and compute processes */
+	mutex_init(&adev->enforce_isolation_mutex);
+	for (i = 0; i < MAX_XCP; ++i) {
+		adev->isolation[i].spearhead = dma_fence_get_stub();
+		amdgpu_sync_create(&adev->isolation[i].active);
+		amdgpu_sync_create(&adev->isolation[i].prev);
+	}
+	mutex_init(&adev->gfx.userq_sch_mutex);
+	mutex_init(&adev->gfx.workload_profile_mutex);
+	mutex_init(&adev->vcn.workload_profile_mutex);
+	mutex_init(&adev->userq_mutex);
+
+	amdgpu_device_init_apu_flags(adev);
+
+	r = amdgpu_device_check_arguments(adev);
+	if (r)
+		return r;
 
 	spin_lock_init(&adev->mmio_idx_lock);
 	spin_lock_init(&adev->smc_idx_lock);
@@ -2022,17 +4500,57 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	spin_lock_init(&adev->se_cac_idx_lock);
 	spin_lock_init(&adev->audio_endpt_idx_lock);
 	spin_lock_init(&adev->mm_stats.lock);
+	spin_lock_init(&adev->virt.rlcg_reg_lock);
+	spin_lock_init(&adev->wb.lock);
+
+	xa_init_flags(&adev->userq_xa, XA_FLAGS_LOCK_IRQ);
+
+	INIT_LIST_HEAD(&adev->reset_list);
 
-	INIT_LIST_HEAD(&adev->shadow_list);
-	mutex_init(&adev->shadow_list_lock);
+	INIT_LIST_HEAD(&adev->ras_list);
 
-	INIT_LIST_HEAD(&adev->gtt_list);
-	spin_lock_init(&adev->gtt_list_lock);
+	INIT_LIST_HEAD(&adev->pm.od_kobj_list);
+
+	INIT_LIST_HEAD(&adev->userq_mgr_list);
+
+	INIT_DELAYED_WORK(&adev->delayed_init_work,
+			  amdgpu_device_delayed_init_work_handler);
+	INIT_DELAYED_WORK(&adev->gfx.gfx_off_delay_work,
+			  amdgpu_device_delay_enable_gfx_off);
+	/*
+	 * Initialize the enforce_isolation work structures for each XCP
+	 * partition.  This work handler is responsible for enforcing shader
+	 * isolation on AMD GPUs.  It counts the number of emitted fences for
+	 * each GFX and compute ring.  If there are any fences, it schedules
+	 * the `enforce_isolation_work` to be run after a delay.  If there are
+	 * no fences, it signals the Kernel Fusion Driver (KFD) to resume the
+	 * runqueue.
+	 */
+	for (i = 0; i < MAX_XCP; i++) {
+		INIT_DELAYED_WORK(&adev->gfx.enforce_isolation[i].work,
+				  amdgpu_gfx_enforce_isolation_handler);
+		adev->gfx.enforce_isolation[i].adev = adev;
+		adev->gfx.enforce_isolation[i].xcp_id = i;
+	}
 
-	INIT_LIST_HEAD(&adev->ring_lru_list);
-	spin_lock_init(&adev->ring_lru_list_lock);
+	INIT_WORK(&adev->xgmi_reset_work, amdgpu_device_xgmi_reset_func);
 
-	INIT_DELAYED_WORK(&adev->late_init_work, amdgpu_late_init_func_handler);
+	adev->gfx.gfx_off_req_count = 1;
+	adev->gfx.gfx_off_residency = 0;
+	adev->gfx.gfx_off_entrycount = 0;
+	adev->pm.ac_power = power_supply_is_system_supplied() > 0;
+
+	atomic_set(&adev->throttling_logging_enabled, 1);
+	/*
+	 * If throttling continues, logging will be performed every minute
+	 * to avoid log flooding. "-1" is subtracted since the thermal
+	 * throttling interrupt comes every second. Thus, the total logging
+	 * interval is 59 seconds(retelimited printk interval) + 1(waiting
+	 * for throttling interrupt) = 60 seconds.
+	 */
+	ratelimit_state_init(&adev->throttling_logging_rs, (60 - 1) * HZ, 1);
+
+	ratelimit_set_flags(&adev->throttling_logging_rs, RATELIMIT_MSG_ON_RELEASE);
 
 	/* Registers mapping */
 	/* TODO: block userspace mapping of io register */
@@ -2044,122 +4562,211 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 		adev->rmmio_size = pci_resource_len(adev->pdev, 2);
 	}
 
+	for (i = 0; i < AMD_IP_BLOCK_TYPE_NUM; i++)
+		atomic_set(&adev->pm.pwr_state[i], POWER_STATE_UNKNOWN);
+
 	adev->rmmio = ioremap(adev->rmmio_base, adev->rmmio_size);
-	if (adev->rmmio == NULL) {
+	if (!adev->rmmio)
 		return -ENOMEM;
-	}
-	DRM_INFO("register mmio base: 0x%08X\n", (uint32_t)adev->rmmio_base);
-	DRM_INFO("register mmio size: %u\n", (unsigned)adev->rmmio_size);
 
-	if (adev->asic_type >= CHIP_BONAIRE)
-		/* doorbell bar mapping */
-		amdgpu_doorbell_init(adev);
-
-	/* io port mapping */
-	for (i = 0; i < DEVICE_COUNT_RESOURCE; i++) {
-		if (pci_resource_flags(adev->pdev, i) & IORESOURCE_IO) {
-			adev->rio_mem_size = pci_resource_len(adev->pdev, i);
-			adev->rio_mem = pci_iomap(adev->pdev, i, adev->rio_mem_size);
-			break;
-		}
+	dev_info(adev->dev, "register mmio base: 0x%08X\n",
+		 (uint32_t)adev->rmmio_base);
+	dev_info(adev->dev, "register mmio size: %u\n",
+		 (unsigned int)adev->rmmio_size);
+
+	/*
+	 * Reset domain needs to be present early, before XGMI hive discovered
+	 * (if any) and initialized to use reset sem and in_gpu reset flag
+	 * early on during init and before calling to RREG32.
+	 */
+	adev->reset_domain = amdgpu_reset_create_reset_domain(SINGLE_DEVICE, "amdgpu-reset-dev");
+	if (!adev->reset_domain)
+		return -ENOMEM;
+
+	/* detect hw virtualization here */
+	amdgpu_virt_init(adev);
+
+	amdgpu_device_get_pcie_info(adev);
+
+	r = amdgpu_device_get_job_timeout_settings(adev);
+	if (r) {
+		dev_err(adev->dev, "invalid lockup_timeout parameter syntax\n");
+		return r;
 	}
-	if (adev->rio_mem == NULL)
-		DRM_INFO("PCI I/O BAR is not found.\n");
 
+	amdgpu_device_set_mcbp(adev);
+
+	/*
+	 * By default, use default mode where all blocks are expected to be
+	 * initialized. At present a 'swinit' of blocks is required to be
+	 * completed before the need for a different level is detected.
+	 */
+	amdgpu_set_init_level(adev, AMDGPU_INIT_LEVEL_DEFAULT);
 	/* early init functions */
-	r = amdgpu_early_init(adev);
+	r = amdgpu_device_ip_early_init(adev);
 	if (r)
 		return r;
 
-	/* if we have > 1 VGA cards, then disable the amdgpu VGA resources */
-	/* this will fail for cards that aren't VGA class devices, just
-	 * ignore it */
-	vga_client_register(adev->pdev, adev, NULL, amdgpu_vga_set_decode);
-
-	if (amdgpu_runtime_pm == 1)
-		runtime = true;
-	if (amdgpu_device_is_px(ddev))
-		runtime = true;
-	if (!pci_is_thunderbolt_attached(adev->pdev))
-		vga_switcheroo_register_client(adev->pdev,
-					       &amdgpu_switcheroo_ops, runtime);
-	if (runtime)
-		vga_switcheroo_init_domain_pm_ops(adev->dev, &adev->vga_pm_domain);
+	/*
+	 * No need to remove conflicting FBs for non-display class devices.
+	 * This prevents the sysfb from being freed accidently.
+	 */
+	if ((pdev->class >> 8) == PCI_CLASS_DISPLAY_VGA ||
+	    (pdev->class >> 8) == PCI_CLASS_DISPLAY_OTHER) {
+		/* Get rid of things like offb */
+		r = aperture_remove_conflicting_pci_devices(adev->pdev, amdgpu_kms_driver.name);
+		if (r)
+			return r;
+	}
 
-	/* Read BIOS */
-	if (!amdgpu_get_bios(adev)) {
-		r = -EINVAL;
-		goto failed;
+	/* Enable TMZ based on IP_VERSION */
+	amdgpu_gmc_tmz_set(adev);
+
+	if (amdgpu_sriov_vf(adev) &&
+	    amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(10, 3, 0))
+		/* VF MMIO access (except mailbox range) from CPU
+		 * will be blocked during sriov runtime
+		 */
+		adev->virt.caps |= AMDGPU_VF_MMIO_ACCESS_PROTECT;
+
+	amdgpu_gmc_noretry_set(adev);
+	/* Need to get xgmi info early to decide the reset behavior*/
+	if (adev->gmc.xgmi.supported) {
+		r = adev->gfxhub.funcs->get_xgmi_info(adev);
+		if (r)
+			return r;
 	}
 
-	r = amdgpu_atombios_init(adev);
-	if (r) {
-		dev_err(adev->dev, "amdgpu_atombios_init failed\n");
-		amdgpu_vf_error_put(AMDGIM_ERROR_VF_ATOMBIOS_INIT_FAIL, 0, 0);
-		goto failed;
+	/* enable PCIE atomic ops */
+	if (amdgpu_sriov_vf(adev)) {
+		if (adev->virt.fw_reserve.p_pf2vf)
+			adev->have_atomics_support = ((struct amd_sriov_msg_pf2vf_info *)
+						      adev->virt.fw_reserve.p_pf2vf)->pcie_atomic_ops_support_flags ==
+				(PCI_EXP_DEVCAP2_ATOMIC_COMP32 | PCI_EXP_DEVCAP2_ATOMIC_COMP64);
+	/* APUs w/ gfx9 onwards doesn't reply on PCIe atomics, rather it is a
+	 * internal path natively support atomics, set have_atomics_support to true.
+	 */
+	} else if ((adev->flags & AMD_IS_APU) &&
+		   (amdgpu_ip_version(adev, GC_HWIP, 0) >
+		    IP_VERSION(9, 0, 0))) {
+		adev->have_atomics_support = true;
+	} else {
+		adev->have_atomics_support =
+			!pci_enable_atomic_ops_to_root(adev->pdev,
+					  PCI_EXP_DEVCAP2_ATOMIC_COMP32 |
+					  PCI_EXP_DEVCAP2_ATOMIC_COMP64);
+	}
+
+	if (!adev->have_atomics_support)
+		dev_info(adev->dev, "PCIE atomic ops is not supported\n");
+
+	/* doorbell bar mapping and doorbell index init*/
+	amdgpu_doorbell_init(adev);
+
+	if (amdgpu_emu_mode == 1) {
+		/* post the asic on emulation mode */
+		emu_soc_asic_init(adev);
+		goto fence_driver_init;
 	}
 
+	amdgpu_reset_init(adev);
+
 	/* detect if we are with an SRIOV vbios */
-	amdgpu_device_detect_sriov_bios(adev);
+	if (adev->bios)
+		amdgpu_device_detect_sriov_bios(adev);
+
+	/* check if we need to reset the asic
+	 *  E.g., driver was not cleanly unloaded previously, etc.
+	 */
+	if (!amdgpu_sriov_vf(adev) && amdgpu_asic_need_reset_on_init(adev)) {
+		if (adev->gmc.xgmi.num_physical_nodes) {
+			dev_info(adev->dev, "Pending hive reset.\n");
+			amdgpu_set_init_level(adev,
+					      AMDGPU_INIT_LEVEL_MINIMAL_XGMI);
+		} else if (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 10) &&
+				   !amdgpu_device_has_display_hardware(adev)) {
+					r = psp_gpu_reset(adev);
+		} else {
+				tmp = amdgpu_reset_method;
+				/* It should do a default reset when loading or reloading the driver,
+				 * regardless of the module parameter reset_method.
+				 */
+				amdgpu_reset_method = AMD_RESET_METHOD_NONE;
+				r = amdgpu_asic_reset(adev);
+				amdgpu_reset_method = tmp;
+		}
+
+		if (r) {
+		  dev_err(adev->dev, "asic reset on init failed\n");
+		  goto failed;
+		}
+	}
 
 	/* Post card if necessary */
-	if (amdgpu_vpost_needed(adev)) {
+	if (amdgpu_device_need_post(adev)) {
 		if (!adev->bios) {
 			dev_err(adev->dev, "no vBIOS found\n");
-			amdgpu_vf_error_put(AMDGIM_ERROR_VF_NO_VBIOS, 0, 0);
 			r = -EINVAL;
 			goto failed;
 		}
-		DRM_INFO("GPU posting now...\n");
-		r = amdgpu_atom_asic_init(adev->mode_info.atom_context);
+		dev_info(adev->dev, "GPU posting now...\n");
+		r = amdgpu_device_asic_init(adev);
 		if (r) {
 			dev_err(adev->dev, "gpu post error!\n");
-			amdgpu_vf_error_put(AMDGIM_ERROR_VF_GPU_POST_ERROR, 0, 0);
 			goto failed;
 		}
-	} else {
-		DRM_INFO("GPU post is not needed\n");
 	}
 
-	if (adev->is_atom_fw) {
-		/* Initialize clocks */
-		r = amdgpu_atomfirmware_get_clock_info(adev);
-		if (r) {
-			dev_err(adev->dev, "amdgpu_atomfirmware_get_clock_info failed\n");
-			amdgpu_vf_error_put(AMDGIM_ERROR_VF_ATOMBIOS_GET_CLOCK_FAIL, 0, 0);
-			goto failed;
-		}
-	} else {
-		/* Initialize clocks */
-		r = amdgpu_atombios_get_clock_info(adev);
-		if (r) {
-			dev_err(adev->dev, "amdgpu_atombios_get_clock_info failed\n");
-			amdgpu_vf_error_put(AMDGIM_ERROR_VF_ATOMBIOS_GET_CLOCK_FAIL, 0, 0);
-			goto failed;
+	if (adev->bios) {
+		if (adev->is_atom_fw) {
+			/* Initialize clocks */
+			r = amdgpu_atomfirmware_get_clock_info(adev);
+			if (r) {
+				dev_err(adev->dev, "amdgpu_atomfirmware_get_clock_info failed\n");
+				amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_ATOMBIOS_GET_CLOCK_FAIL, 0, 0);
+				goto failed;
+			}
+		} else {
+			/* Initialize clocks */
+			r = amdgpu_atombios_get_clock_info(adev);
+			if (r) {
+				dev_err(adev->dev, "amdgpu_atombios_get_clock_info failed\n");
+				amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_ATOMBIOS_GET_CLOCK_FAIL, 0, 0);
+				goto failed;
+			}
+			/* init i2c buses */
+			amdgpu_i2c_init(adev);
 		}
-		/* init i2c buses */
-		amdgpu_atombios_i2c_init(adev);
 	}
 
+fence_driver_init:
 	/* Fence driver */
-	r = amdgpu_fence_driver_init(adev);
+	r = amdgpu_fence_driver_sw_init(adev);
 	if (r) {
-		dev_err(adev->dev, "amdgpu_fence_driver_init failed\n");
-		amdgpu_vf_error_put(AMDGIM_ERROR_VF_FENCE_INIT_FAIL, 0, 0);
+		dev_err(adev->dev, "amdgpu_fence_driver_sw_init failed\n");
+		amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_FENCE_INIT_FAIL, 0, 0);
 		goto failed;
 	}
 
 	/* init the mode config */
-	drm_mode_config_init(adev->ddev);
+	drm_mode_config_init(adev_to_drm(adev));
 
-	r = amdgpu_init(adev);
+	r = amdgpu_device_ip_init(adev);
 	if (r) {
-		dev_err(adev->dev, "amdgpu_init failed\n");
-		amdgpu_vf_error_put(AMDGIM_ERROR_VF_AMDGPU_INIT_FAIL, 0, 0);
-		amdgpu_fini(adev);
-		goto failed;
+		dev_err(adev->dev, "amdgpu_device_ip_init failed\n");
+		amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_AMDGPU_INIT_FAIL, 0, 0);
+		goto release_ras_con;
 	}
 
+	amdgpu_fence_driver_hw_init(adev);
+
+	dev_info(adev->dev,
+		"SE %d, SH per SE %d, CU per SH %d, active_cu_number %d\n",
+			adev->gfx.config.max_shader_engines,
+			adev->gfx.config.max_sh_per_se,
+			adev->gfx.config.max_cu_per_sh,
+			adev->gfx.cu_info.number);
+
 	adev->accel_working = true;
 
 	amdgpu_vm_check_compute_bug(adev);
@@ -2172,353 +4779,656 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	/* Get a log2 for easy divisions. */
 	adev->mm_stats.log2_max_MBps = ilog2(max(1u, max_MBps));
 
-	r = amdgpu_ib_pool_init(adev);
-	if (r) {
-		dev_err(adev->dev, "IB initialization failed (%d).\n", r);
-		amdgpu_vf_error_put(AMDGIM_ERROR_VF_IB_INIT_FAIL, 0, r);
-		goto failed;
+	/*
+	 * Register gpu instance before amdgpu_device_enable_mgpu_fan_boost.
+	 * Otherwise the mgpu fan boost feature will be skipped due to the
+	 * gpu instance is counted less.
+	 */
+	amdgpu_register_gpu_instance(adev);
+
+	/* enable clockgating, etc. after ib tests, etc. since some blocks require
+	 * explicit gating rather than handling it automatically.
+	 */
+	if (adev->init_lvl->level != AMDGPU_INIT_LEVEL_MINIMAL_XGMI) {
+		r = amdgpu_device_ip_late_init(adev);
+		if (r) {
+			dev_err(adev->dev, "amdgpu_device_ip_late_init failed\n");
+			amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_AMDGPU_LATE_INIT_FAIL, 0, r);
+			goto release_ras_con;
+		}
+		/* must succeed. */
+		amdgpu_ras_resume(adev);
+		queue_delayed_work(system_wq, &adev->delayed_init_work,
+				   msecs_to_jiffies(AMDGPU_RESUME_MS));
 	}
 
-	r = amdgpu_ib_ring_tests(adev);
-	if (r)
-		DRM_ERROR("ib ring test failed (%d).\n", r);
+	if (amdgpu_sriov_vf(adev)) {
+		amdgpu_virt_release_full_gpu(adev, true);
+		flush_delayed_work(&adev->delayed_init_work);
+	}
 
-	amdgpu_fbdev_init(adev);
+	/*
+	 * Place those sysfs registering after `late_init`. As some of those
+	 * operations performed in `late_init` might affect the sysfs
+	 * interfaces creating.
+	 */
+	r = amdgpu_atombios_sysfs_init(adev);
+	if (r)
+		drm_err(&adev->ddev,
+			"registering atombios sysfs failed (%d).\n", r);
 
-	r = amdgpu_gem_debugfs_init(adev);
+	r = amdgpu_pm_sysfs_init(adev);
 	if (r)
-		DRM_ERROR("registering gem debugfs failed (%d).\n", r);
+		dev_err(adev->dev, "registering pm sysfs failed (%d).\n", r);
+
+	r = amdgpu_ucode_sysfs_init(adev);
+	if (r) {
+		adev->ucode_sysfs_en = false;
+		dev_err(adev->dev, "Creating firmware sysfs failed (%d).\n", r);
+	} else
+		adev->ucode_sysfs_en = true;
 
-	r = amdgpu_debugfs_regs_init(adev);
+	r = amdgpu_device_attr_sysfs_init(adev);
 	if (r)
-		DRM_ERROR("registering register debugfs failed (%d).\n", r);
+		dev_err(adev->dev, "Could not create amdgpu device attr\n");
 
-	r = amdgpu_debugfs_test_ib_ring_init(adev);
+	r = devm_device_add_group(adev->dev, &amdgpu_board_attrs_group);
 	if (r)
-		DRM_ERROR("registering register test ib ring debugfs failed (%d).\n", r);
+		dev_err(adev->dev,
+			"Could not create amdgpu board attributes\n");
+
+	amdgpu_fru_sysfs_init(adev);
+	amdgpu_reg_state_sysfs_init(adev);
+	amdgpu_xcp_sysfs_init(adev);
 
-	r = amdgpu_debugfs_firmware_init(adev);
+	if (IS_ENABLED(CONFIG_PERF_EVENTS))
+		r = amdgpu_pmu_init(adev);
 	if (r)
-		DRM_ERROR("registering firmware debugfs failed (%d).\n", r);
+		dev_err(adev->dev, "amdgpu_pmu_init failed\n");
 
-	if ((amdgpu_testing & 1)) {
-		if (adev->accel_working)
-			amdgpu_test_moves(adev);
-		else
-			DRM_INFO("amdgpu: acceleration disabled, skipping move tests\n");
-	}
-	if (amdgpu_benchmarking) {
-		if (adev->accel_working)
-			amdgpu_benchmark(adev, amdgpu_benchmarking);
-		else
-			DRM_INFO("amdgpu: acceleration disabled, skipping benchmarks\n");
-	}
+	/* Have stored pci confspace at hand for restore in sudden PCI error */
+	if (amdgpu_device_cache_pci_state(adev->pdev))
+		pci_restore_state(pdev);
 
-	/* enable clockgating, etc. after ib tests, etc. since some blocks require
-	 * explicit gating rather than handling it automatically.
+	/* if we have > 1 VGA cards, then disable the amdgpu VGA resources */
+	/* this will fail for cards that aren't VGA class devices, just
+	 * ignore it
 	 */
-	r = amdgpu_late_init(adev);
-	if (r) {
-		dev_err(adev->dev, "amdgpu_late_init failed\n");
-		amdgpu_vf_error_put(AMDGIM_ERROR_VF_AMDGPU_LATE_INIT_FAIL, 0, r);
+	if ((adev->pdev->class >> 8) == PCI_CLASS_DISPLAY_VGA)
+		vga_client_register(adev->pdev, amdgpu_device_vga_set_decode);
+
+	px = amdgpu_device_supports_px(adev);
+
+	if (px || (!dev_is_removable(&adev->pdev->dev) &&
+				apple_gmux_detect(NULL, NULL)))
+		vga_switcheroo_register_client(adev->pdev,
+					       &amdgpu_switcheroo_ops, px);
+
+	if (px)
+		vga_switcheroo_init_domain_pm_ops(adev->dev, &adev->vga_pm_domain);
+
+	if (adev->init_lvl->level == AMDGPU_INIT_LEVEL_MINIMAL_XGMI)
+		amdgpu_xgmi_reset_on_init(adev);
+
+	amdgpu_device_check_iommu_direct_map(adev);
+
+	adev->pm_nb.notifier_call = amdgpu_device_pm_notifier;
+	r = register_pm_notifier(&adev->pm_nb);
+	if (r)
 		goto failed;
-	}
 
 	return 0;
 
+release_ras_con:
+	if (amdgpu_sriov_vf(adev))
+		amdgpu_virt_release_full_gpu(adev, true);
+
+	/* failed in exclusive mode due to timeout */
+	if (amdgpu_sriov_vf(adev) &&
+		!amdgpu_sriov_runtime(adev) &&
+		amdgpu_virt_mmio_blocked(adev) &&
+		!amdgpu_virt_wait_reset(adev)) {
+		dev_err(adev->dev, "VF exclusive mode timeout\n");
+		/* Don't send request since VF is inactive. */
+		adev->virt.caps &= ~AMDGPU_SRIOV_CAPS_RUNTIME;
+		adev->virt.ops = NULL;
+		r = -EAGAIN;
+	}
+	amdgpu_release_ras_context(adev);
+
 failed:
 	amdgpu_vf_error_trans_all(adev);
-	if (runtime)
-		vga_switcheroo_fini_domain_pm_ops(adev->dev);
+
 	return r;
 }
 
+static void amdgpu_device_unmap_mmio(struct amdgpu_device *adev)
+{
+
+	/* Clear all CPU mappings pointing to this device */
+	unmap_mapping_range(adev->ddev.anon_inode->i_mapping, 0, 0, 1);
+
+	/* Unmap all mapped bars - Doorbell, registers and VRAM */
+	amdgpu_doorbell_fini(adev);
+
+	iounmap(adev->rmmio);
+	adev->rmmio = NULL;
+	if (adev->mman.aper_base_kaddr)
+		iounmap(adev->mman.aper_base_kaddr);
+	adev->mman.aper_base_kaddr = NULL;
+
+	/* Memory manager related */
+	if (!adev->gmc.xgmi.connected_to_cpu && !adev->gmc.is_app_apu) {
+		arch_phys_wc_del(adev->gmc.vram_mtrr);
+		arch_io_free_memtype_wc(adev->gmc.aper_base, adev->gmc.aper_size);
+	}
+}
+
 /**
- * amdgpu_device_fini - tear down the driver
+ * amdgpu_device_fini_hw - tear down the driver
  *
  * @adev: amdgpu_device pointer
  *
  * Tear down the driver info (all asics).
  * Called at driver shutdown.
  */
-void amdgpu_device_fini(struct amdgpu_device *adev)
+void amdgpu_device_fini_hw(struct amdgpu_device *adev)
 {
-	int r;
+	dev_info(adev->dev, "amdgpu: finishing device.\n");
+	flush_delayed_work(&adev->delayed_init_work);
 
-	DRM_INFO("amdgpu: finishing device.\n");
+	if (adev->mman.initialized)
+		drain_workqueue(adev->mman.bdev.wq);
 	adev->shutdown = true;
-	if (adev->mode_info.mode_config_initialized)
-		drm_crtc_force_disable_all(adev->ddev);
-	/* evict vram memory */
-	amdgpu_bo_evict_vram(adev);
-	amdgpu_ib_pool_fini(adev);
-	amdgpu_fence_driver_fini(adev);
-	amdgpu_fbdev_fini(adev);
-	r = amdgpu_fini(adev);
-	if (adev->firmware.gpu_info_fw) {
-		release_firmware(adev->firmware.gpu_info_fw);
-		adev->firmware.gpu_info_fw = NULL;
+
+	unregister_pm_notifier(&adev->pm_nb);
+
+	/* make sure IB test finished before entering exclusive mode
+	 * to avoid preemption on IB test
+	 */
+	if (amdgpu_sriov_vf(adev)) {
+		amdgpu_virt_request_full_gpu(adev, false);
+		amdgpu_virt_fini_data_exchange(adev);
 	}
+
+	/* disable all interrupts */
+	amdgpu_irq_disable_all(adev);
+	if (adev->mode_info.mode_config_initialized) {
+		if (!drm_drv_uses_atomic_modeset(adev_to_drm(adev)))
+			drm_helper_force_disable_all(adev_to_drm(adev));
+		else
+			drm_atomic_helper_shutdown(adev_to_drm(adev));
+	}
+	amdgpu_fence_driver_hw_fini(adev);
+
+	if (adev->pm.sysfs_initialized)
+		amdgpu_pm_sysfs_fini(adev);
+	if (adev->ucode_sysfs_en)
+		amdgpu_ucode_sysfs_fini(adev);
+	amdgpu_device_attr_sysfs_fini(adev);
+	amdgpu_fru_sysfs_fini(adev);
+
+	amdgpu_reg_state_sysfs_fini(adev);
+	amdgpu_xcp_sysfs_fini(adev);
+
+	/* disable ras feature must before hw fini */
+	amdgpu_ras_pre_fini(adev);
+
+	amdgpu_ttm_set_buffer_funcs_status(adev, false);
+
+	amdgpu_device_ip_fini_early(adev);
+
+	amdgpu_irq_fini_hw(adev);
+
+	if (adev->mman.initialized)
+		ttm_device_clear_dma_mappings(&adev->mman.bdev);
+
+	amdgpu_gart_dummy_page_fini(adev);
+
+	if (drm_dev_is_unplugged(adev_to_drm(adev)))
+		amdgpu_device_unmap_mmio(adev);
+
+}
+
+void amdgpu_device_fini_sw(struct amdgpu_device *adev)
+{
+	int i, idx;
+	bool px;
+
+	amdgpu_device_ip_fini(adev);
+	amdgpu_fence_driver_sw_fini(adev);
+	amdgpu_ucode_release(&adev->firmware.gpu_info_fw);
 	adev->accel_working = false;
-	cancel_delayed_work_sync(&adev->late_init_work);
+	dma_fence_put(rcu_dereference_protected(adev->gang_submit, true));
+	for (i = 0; i < MAX_XCP; ++i) {
+		dma_fence_put(adev->isolation[i].spearhead);
+		amdgpu_sync_free(&adev->isolation[i].active);
+		amdgpu_sync_free(&adev->isolation[i].prev);
+	}
+
+	amdgpu_reset_fini(adev);
+
 	/* free i2c buses */
 	amdgpu_i2c_fini(adev);
-	amdgpu_atombios_fini(adev);
-	kfree(adev->bios);
-	adev->bios = NULL;
-	if (!pci_is_thunderbolt_attached(adev->pdev))
+
+	if (adev->bios) {
+		if (amdgpu_emu_mode != 1)
+			amdgpu_atombios_fini(adev);
+		amdgpu_bios_release(adev);
+	}
+
+	kfree(adev->fru_info);
+	adev->fru_info = NULL;
+
+	kfree(adev->xcp_mgr);
+	adev->xcp_mgr = NULL;
+
+	px = amdgpu_device_supports_px(adev);
+
+	if (px || (!dev_is_removable(&adev->pdev->dev) &&
+				apple_gmux_detect(NULL, NULL)))
 		vga_switcheroo_unregister_client(adev->pdev);
-	if (adev->flags & AMD_IS_PX)
+
+	if (px)
 		vga_switcheroo_fini_domain_pm_ops(adev->dev);
-	vga_client_register(adev->pdev, NULL, NULL, NULL);
-	if (adev->rio_mem)
-		pci_iounmap(adev->pdev, adev->rio_mem);
-	adev->rio_mem = NULL;
-	iounmap(adev->rmmio);
-	adev->rmmio = NULL;
-	if (adev->asic_type >= CHIP_BONAIRE)
-		amdgpu_doorbell_fini(adev);
-	amdgpu_debugfs_regs_cleanup(adev);
+
+	if ((adev->pdev->class >> 8) == PCI_CLASS_DISPLAY_VGA)
+		vga_client_unregister(adev->pdev);
+
+	if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+
+		iounmap(adev->rmmio);
+		adev->rmmio = NULL;
+		drm_dev_exit(idx);
+	}
+
+	if (IS_ENABLED(CONFIG_PERF_EVENTS))
+		amdgpu_pmu_fini(adev);
+	if (adev->mman.discovery_bin)
+		amdgpu_discovery_fini(adev);
+
+	amdgpu_reset_put_reset_domain(adev->reset_domain);
+	adev->reset_domain = NULL;
+
+	kfree(adev->pci_state);
+	kfree(adev->pcie_reset_ctx.swds_pcistate);
+	kfree(adev->pcie_reset_ctx.swus_pcistate);
 }
 
+/**
+ * amdgpu_device_evict_resources - evict device resources
+ * @adev: amdgpu device object
+ *
+ * Evicts all ttm device resources(vram BOs, gart table) from the lru list
+ * of the vram memory type. Mainly used for evicting device resources
+ * at suspend time.
+ *
+ */
+static int amdgpu_device_evict_resources(struct amdgpu_device *adev)
+{
+	int ret;
+
+	/* No need to evict vram on APUs unless going to S4 */
+	if (!adev->in_s4 && (adev->flags & AMD_IS_APU))
+		return 0;
+
+	/* No need to evict when going to S5 through S4 callbacks */
+	if (system_state == SYSTEM_POWER_OFF)
+		return 0;
+
+	ret = amdgpu_ttm_evict_resources(adev, TTM_PL_VRAM);
+	if (ret) {
+		dev_warn(adev->dev, "evicting device resources failed\n");
+		return ret;
+	}
+
+	if (adev->in_s4) {
+		ret = ttm_device_prepare_hibernation(&adev->mman.bdev);
+		if (ret)
+			dev_err(adev->dev, "prepare hibernation failed, %d\n", ret);
+	}
+	return ret;
+}
 
 /*
  * Suspend & resume.
  */
 /**
- * amdgpu_device_suspend - initiate device suspend
+ * amdgpu_device_pm_notifier - Notification block for Suspend/Hibernate events
+ * @nb: notifier block
+ * @mode: suspend mode
+ * @data: data
  *
- * @pdev: drm dev pointer
- * @state: suspend state
+ * This function is called when the system is about to suspend or hibernate.
+ * It is used to set the appropriate flags so that eviction can be optimized
+ * in the pm prepare callback.
+ */
+static int amdgpu_device_pm_notifier(struct notifier_block *nb, unsigned long mode,
+				     void *data)
+{
+	struct amdgpu_device *adev = container_of(nb, struct amdgpu_device, pm_nb);
+
+	switch (mode) {
+	case PM_HIBERNATION_PREPARE:
+		adev->in_s4 = true;
+		break;
+	case PM_POST_HIBERNATION:
+		adev->in_s4 = false;
+		break;
+	}
+
+	return NOTIFY_DONE;
+}
+
+/**
+ * amdgpu_device_prepare - prepare for device suspend
  *
- * Puts the hw in the suspend state (all asics).
+ * @dev: drm dev pointer
+ *
+ * Prepare to put the hw in the suspend state (all asics).
  * Returns 0 for success or an error on failure.
  * Called at driver suspend.
  */
-int amdgpu_device_suspend(struct drm_device *dev, bool suspend, bool fbcon)
+int amdgpu_device_prepare(struct drm_device *dev)
 {
-	struct amdgpu_device *adev;
-	struct drm_crtc *crtc;
-	struct drm_connector *connector;
-	int r;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	int i, r;
 
-	if (dev == NULL || dev->dev_private == NULL) {
-		return -ENODEV;
+	if (dev->switch_power_state == DRM_SWITCH_POWER_OFF)
+		return 0;
+
+	/* Evict the majority of BOs before starting suspend sequence */
+	r = amdgpu_device_evict_resources(adev);
+	if (r)
+		return r;
+
+	flush_delayed_work(&adev->gfx.gfx_off_delay_work);
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		if (!adev->ip_blocks[i].version->funcs->prepare_suspend)
+			continue;
+		r = adev->ip_blocks[i].version->funcs->prepare_suspend(&adev->ip_blocks[i]);
+		if (r)
+			return r;
 	}
 
-	adev = dev->dev_private;
+	return 0;
+}
+
+/**
+ * amdgpu_device_complete - complete power state transition
+ *
+ * @dev: drm dev pointer
+ *
+ * Undo the changes from amdgpu_device_prepare. This will be
+ * called on all resume transitions, including those that failed.
+ */
+void amdgpu_device_complete(struct drm_device *dev)
+{
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	int i;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		if (!adev->ip_blocks[i].version->funcs->complete)
+			continue;
+		adev->ip_blocks[i].version->funcs->complete(&adev->ip_blocks[i]);
+	}
+}
+
+/**
+ * amdgpu_device_suspend - initiate device suspend
+ *
+ * @dev: drm dev pointer
+ * @notify_clients: notify in-kernel DRM clients
+ *
+ * Puts the hw in the suspend state (all asics).
+ * Returns 0 for success or an error on failure.
+ * Called at driver suspend.
+ */
+int amdgpu_device_suspend(struct drm_device *dev, bool notify_clients)
+{
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	int r = 0;
 
 	if (dev->switch_power_state == DRM_SWITCH_POWER_OFF)
 		return 0;
 
-	drm_kms_helper_poll_disable(dev);
+	adev->in_suspend = true;
 
-	/* turn off display hw */
-	drm_modeset_lock_all(dev);
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
-		drm_helper_connector_dpms(connector, DRM_MODE_DPMS_OFF);
+	if (amdgpu_sriov_vf(adev)) {
+		if (!adev->in_runpm)
+			amdgpu_amdkfd_suspend_process(adev);
+		amdgpu_virt_fini_data_exchange(adev);
+		r = amdgpu_virt_request_full_gpu(adev, false);
+		if (r)
+			return r;
 	}
-	drm_modeset_unlock_all(dev);
 
-	amdgpu_amdkfd_suspend(adev);
+	if (amdgpu_acpi_smart_shift_update(adev, AMDGPU_SS_DEV_D3))
+		dev_warn(adev->dev, "smart shift update failed\n");
 
-	/* unpin the front buffers and cursors */
-	list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
-		struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-		struct amdgpu_framebuffer *rfb = to_amdgpu_framebuffer(crtc->primary->fb);
-		struct amdgpu_bo *robj;
+	if (notify_clients)
+		drm_client_dev_suspend(adev_to_drm(adev), false);
 
-		if (amdgpu_crtc->cursor_bo) {
-			struct amdgpu_bo *aobj = gem_to_amdgpu_bo(amdgpu_crtc->cursor_bo);
-			r = amdgpu_bo_reserve(aobj, true);
-			if (r == 0) {
-				amdgpu_bo_unpin(aobj);
-				amdgpu_bo_unreserve(aobj);
-			}
-		}
+	cancel_delayed_work_sync(&adev->delayed_init_work);
 
-		if (rfb == NULL || rfb->obj == NULL) {
-			continue;
-		}
-		robj = gem_to_amdgpu_bo(rfb->obj);
-		/* don't unpin kernel fb objects */
-		if (!amdgpu_fbdev_robj_is_fb(adev, robj)) {
-			r = amdgpu_bo_reserve(robj, true);
-			if (r == 0) {
-				amdgpu_bo_unpin(robj);
-				amdgpu_bo_unreserve(robj);
-			}
-		}
-	}
-	/* evict vram memory */
-	amdgpu_bo_evict_vram(adev);
+	amdgpu_ras_suspend(adev);
+
+	amdgpu_device_ip_suspend_phase1(adev);
+
+	amdgpu_amdkfd_suspend(adev, !amdgpu_sriov_vf(adev) && !adev->in_runpm);
+	amdgpu_userq_suspend(adev);
+
+	r = amdgpu_device_evict_resources(adev);
+	if (r)
+		return r;
+
+	amdgpu_ttm_set_buffer_funcs_status(adev, false);
+
+	amdgpu_fence_driver_hw_fini(adev);
+
+	amdgpu_device_ip_suspend_phase2(adev);
+
+	if (amdgpu_sriov_vf(adev))
+		amdgpu_virt_release_full_gpu(adev, false);
+
+	r = amdgpu_dpm_notify_rlc_state(adev, false);
+	if (r)
+		return r;
 
-	amdgpu_fence_driver_suspend(adev);
+	return 0;
+}
 
-	r = amdgpu_suspend(adev);
+static inline int amdgpu_virt_resume(struct amdgpu_device *adev)
+{
+	int r;
+	unsigned int prev_physical_node_id = adev->gmc.xgmi.physical_node_id;
 
-	/* evict remaining vram memory
-	 * This second call to evict vram is to evict the gart page table
-	 * using the CPU.
+	/* During VM resume, QEMU programming of VF MSIX table (register GFXMSIX_VECT0_ADDR_LO)
+	 * may not work. The access could be blocked by nBIF protection as VF isn't in
+	 * exclusive access mode. Exclusive access is enabled now, disable/enable MSIX
+	 * so that QEMU reprograms MSIX table.
 	 */
-	amdgpu_bo_evict_vram(adev);
+	amdgpu_restore_msix(adev);
 
-	amdgpu_atombios_scratch_regs_save(adev);
-	pci_save_state(dev->pdev);
-	if (suspend) {
-		/* Shut down the device */
-		pci_disable_device(dev->pdev);
-		pci_set_power_state(dev->pdev, PCI_D3hot);
-	} else {
-		r = amdgpu_asic_reset(adev);
-		if (r)
-			DRM_ERROR("amdgpu asic reset failed\n");
-	}
+	r = adev->gfxhub.funcs->get_xgmi_info(adev);
+	if (r)
+		return r;
+
+	dev_info(adev->dev, "xgmi node, old id %d, new id %d\n",
+		prev_physical_node_id, adev->gmc.xgmi.physical_node_id);
+
+	adev->vm_manager.vram_base_offset = adev->gfxhub.funcs->get_mc_fb_offset(adev);
+	adev->vm_manager.vram_base_offset +=
+		adev->gmc.xgmi.physical_node_id * adev->gmc.xgmi.node_segment_size;
 
-	if (fbcon) {
-		console_lock();
-		amdgpu_fbdev_set_suspend(adev, 1);
-		console_unlock();
-	}
 	return 0;
 }
 
 /**
  * amdgpu_device_resume - initiate device resume
  *
- * @pdev: drm dev pointer
+ * @dev: drm dev pointer
+ * @notify_clients: notify in-kernel DRM clients
  *
  * Bring the hw back to operating state (all asics).
  * Returns 0 for success or an error on failure.
  * Called at driver resume.
  */
-int amdgpu_device_resume(struct drm_device *dev, bool resume, bool fbcon)
+int amdgpu_device_resume(struct drm_device *dev, bool notify_clients)
 {
-	struct drm_connector *connector;
-	struct amdgpu_device *adev = dev->dev_private;
-	struct drm_crtc *crtc;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	int r = 0;
 
-	if (dev->switch_power_state == DRM_SWITCH_POWER_OFF)
-		return 0;
-
-	if (fbcon)
-		console_lock();
+	if (amdgpu_sriov_vf(adev)) {
+		r = amdgpu_virt_request_full_gpu(adev, true);
+		if (r)
+			return r;
+	}
 
-	if (resume) {
-		pci_set_power_state(dev->pdev, PCI_D0);
-		pci_restore_state(dev->pdev);
-		r = pci_enable_device(dev->pdev);
+	if (amdgpu_virt_xgmi_migrate_enabled(adev)) {
+		r = amdgpu_virt_resume(adev);
 		if (r)
-			goto unlock;
+			goto exit;
 	}
-	amdgpu_atombios_scratch_regs_restore(adev);
+
+	if (dev->switch_power_state == DRM_SWITCH_POWER_OFF)
+		return 0;
+
+	if (adev->in_s0ix)
+		amdgpu_dpm_gfx_state_change(adev, sGpuChangeState_D0Entry);
 
 	/* post card */
-	if (amdgpu_need_post(adev)) {
-		r = amdgpu_atom_asic_init(adev->mode_info.atom_context);
+	if (amdgpu_device_need_post(adev)) {
+		r = amdgpu_device_asic_init(adev);
 		if (r)
-			DRM_ERROR("amdgpu asic init failed\n");
+			dev_err(adev->dev, "amdgpu asic init failed\n");
 	}
 
-	r = amdgpu_resume(adev);
+	r = amdgpu_device_ip_resume(adev);
+
 	if (r) {
-		DRM_ERROR("amdgpu_resume failed (%d).\n", r);
-		goto unlock;
+		dev_err(adev->dev, "amdgpu_device_ip_resume failed (%d).\n", r);
+		goto exit;
 	}
-	amdgpu_fence_driver_resume(adev);
 
-	if (resume) {
-		r = amdgpu_ib_ring_tests(adev);
-		if (r)
-			DRM_ERROR("ib ring test failed (%d).\n", r);
-	}
+	r = amdgpu_amdkfd_resume(adev, !amdgpu_sriov_vf(adev) && !adev->in_runpm);
+	if (r)
+		goto exit;
 
-	r = amdgpu_late_init(adev);
+	r = amdgpu_userq_resume(adev);
 	if (r)
-		goto unlock;
-
-	/* pin cursors */
-	list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
-		struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-
-		if (amdgpu_crtc->cursor_bo) {
-			struct amdgpu_bo *aobj = gem_to_amdgpu_bo(amdgpu_crtc->cursor_bo);
-			r = amdgpu_bo_reserve(aobj, true);
-			if (r == 0) {
-				r = amdgpu_bo_pin(aobj,
-						  AMDGPU_GEM_DOMAIN_VRAM,
-						  &amdgpu_crtc->cursor_addr);
-				if (r != 0)
-					DRM_ERROR("Failed to pin cursor BO (%d)\n", r);
-				amdgpu_bo_unreserve(aobj);
-			}
-		}
+		goto exit;
+
+	r = amdgpu_device_ip_late_init(adev);
+	if (r)
+		goto exit;
+
+	queue_delayed_work(system_wq, &adev->delayed_init_work,
+			   msecs_to_jiffies(AMDGPU_RESUME_MS));
+exit:
+	if (amdgpu_sriov_vf(adev)) {
+		amdgpu_virt_init_data_exchange(adev);
+		amdgpu_virt_release_full_gpu(adev, true);
+
+		if (!r && !adev->in_runpm)
+			r = amdgpu_amdkfd_resume_process(adev);
 	}
-	r = amdgpu_amdkfd_resume(adev);
+
 	if (r)
 		return r;
 
-	/* blat the mode back in */
-	if (fbcon) {
-		drm_helper_resume_force_mode(dev);
-		/* turn on display hw */
-		drm_modeset_lock_all(dev);
-		list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
-			drm_helper_connector_dpms(connector, DRM_MODE_DPMS_ON);
-		}
-		drm_modeset_unlock_all(dev);
-	}
+	/* Make sure IB tests flushed */
+	flush_delayed_work(&adev->delayed_init_work);
 
-	drm_kms_helper_poll_enable(dev);
+	if (notify_clients)
+		drm_client_dev_resume(adev_to_drm(adev), false);
 
-	/*
-	 * Most of the connector probing functions try to acquire runtime pm
-	 * refs to ensure that the GPU is powered on when connector polling is
-	 * performed. Since we're calling this from a runtime PM callback,
-	 * trying to acquire rpm refs will cause us to deadlock.
-	 *
-	 * Since we're guaranteed to be holding the rpm lock, it's safe to
-	 * temporarily disable the rpm helpers so this doesn't deadlock us.
-	 */
+	amdgpu_ras_resume(adev);
+
+	if (adev->mode_info.num_crtc) {
+		/*
+		 * Most of the connector probing functions try to acquire runtime pm
+		 * refs to ensure that the GPU is powered on when connector polling is
+		 * performed. Since we're calling this from a runtime PM callback,
+		 * trying to acquire rpm refs will cause us to deadlock.
+		 *
+		 * Since we're guaranteed to be holding the rpm lock, it's safe to
+		 * temporarily disable the rpm helpers so this doesn't deadlock us.
+		 */
 #ifdef CONFIG_PM
-	dev->dev->power.disable_depth++;
+		dev->dev->power.disable_depth++;
 #endif
-	drm_helper_hpd_irq_event(dev);
+		if (!adev->dc_enabled)
+			drm_helper_hpd_irq_event(dev);
+		else
+			drm_kms_helper_hotplug_event(dev);
 #ifdef CONFIG_PM
-	dev->dev->power.disable_depth--;
+		dev->dev->power.disable_depth--;
 #endif
+	}
 
-	if (fbcon)
-		amdgpu_fbdev_set_suspend(adev, 0);
+	amdgpu_vram_mgr_clear_reset_blocks(adev);
+	adev->in_suspend = false;
 
-unlock:
-	if (fbcon)
-		console_unlock();
+	if (amdgpu_acpi_smart_shift_update(adev, AMDGPU_SS_DEV_D0))
+		dev_warn(adev->dev, "smart shift update failed\n");
 
-	return r;
+	return 0;
 }
 
-static bool amdgpu_check_soft_reset(struct amdgpu_device *adev)
+/**
+ * amdgpu_device_ip_check_soft_reset - did soft reset succeed
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * The list of all the hardware IPs that make up the asic is walked and
+ * the check_soft_reset callbacks are run.  check_soft_reset determines
+ * if the asic is still hung or not.
+ * Returns true if any of the IPs are still in a hung state, false if not.
+ */
+static bool amdgpu_device_ip_check_soft_reset(struct amdgpu_device *adev)
 {
 	int i;
 	bool asic_hang = false;
 
+	if (amdgpu_sriov_vf(adev))
+		return true;
+
+	if (amdgpu_asic_need_full_reset(adev))
+		return true;
+
 	for (i = 0; i < adev->num_ip_blocks; i++) {
 		if (!adev->ip_blocks[i].status.valid)
 			continue;
 		if (adev->ip_blocks[i].version->funcs->check_soft_reset)
 			adev->ip_blocks[i].status.hang =
-				adev->ip_blocks[i].version->funcs->check_soft_reset(adev);
+				adev->ip_blocks[i].version->funcs->check_soft_reset(
+					&adev->ip_blocks[i]);
 		if (adev->ip_blocks[i].status.hang) {
-			DRM_INFO("IP block:%s is hung!\n", adev->ip_blocks[i].version->funcs->name);
+			dev_info(adev->dev, "IP block:%s is hung!\n", adev->ip_blocks[i].version->funcs->name);
 			asic_hang = true;
 		}
 	}
 	return asic_hang;
 }
 
-static int amdgpu_pre_soft_reset(struct amdgpu_device *adev)
+/**
+ * amdgpu_device_ip_pre_soft_reset - prepare for soft reset
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * The list of all the hardware IPs that make up the asic is walked and the
+ * pre_soft_reset callbacks are run if the block is hung.  pre_soft_reset
+ * handles any IP specific hardware or software state changes that are
+ * necessary for a soft reset to succeed.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_pre_soft_reset(struct amdgpu_device *adev)
 {
 	int i, r = 0;
 
@@ -2527,7 +5437,7 @@ static int amdgpu_pre_soft_reset(struct amdgpu_device *adev)
 			continue;
 		if (adev->ip_blocks[i].status.hang &&
 		    adev->ip_blocks[i].version->funcs->pre_soft_reset) {
-			r = adev->ip_blocks[i].version->funcs->pre_soft_reset(adev);
+			r = adev->ip_blocks[i].version->funcs->pre_soft_reset(&adev->ip_blocks[i]);
 			if (r)
 				return r;
 		}
@@ -2536,19 +5446,32 @@ static int amdgpu_pre_soft_reset(struct amdgpu_device *adev)
 	return 0;
 }
 
-static bool amdgpu_need_full_reset(struct amdgpu_device *adev)
+/**
+ * amdgpu_device_ip_need_full_reset - check if a full asic reset is needed
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Some hardware IPs cannot be soft reset.  If they are hung, a full gpu
+ * reset is necessary to recover.
+ * Returns true if a full asic reset is required, false if not.
+ */
+static bool amdgpu_device_ip_need_full_reset(struct amdgpu_device *adev)
 {
 	int i;
 
+	if (amdgpu_asic_need_full_reset(adev))
+		return true;
+
 	for (i = 0; i < adev->num_ip_blocks; i++) {
 		if (!adev->ip_blocks[i].status.valid)
 			continue;
 		if ((adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GMC) ||
 		    (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SMC) ||
 		    (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_ACP) ||
-		    (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_DCE)) {
+		    (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_DCE) ||
+		     adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_PSP) {
 			if (adev->ip_blocks[i].status.hang) {
-				DRM_INFO("Some block need full reset!\n");
+				dev_info(adev->dev, "Some block need full reset!\n");
 				return true;
 			}
 		}
@@ -2556,7 +5479,18 @@ static bool amdgpu_need_full_reset(struct amdgpu_device *adev)
 	return false;
 }
 
-static int amdgpu_soft_reset(struct amdgpu_device *adev)
+/**
+ * amdgpu_device_ip_soft_reset - do a soft reset
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * The list of all the hardware IPs that make up the asic is walked and the
+ * soft_reset callbacks are run if the block is hung.  soft_reset handles any
+ * IP specific hardware or software state changes that are necessary to soft
+ * reset the IP.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_soft_reset(struct amdgpu_device *adev)
 {
 	int i, r = 0;
 
@@ -2565,7 +5499,7 @@ static int amdgpu_soft_reset(struct amdgpu_device *adev)
 			continue;
 		if (adev->ip_blocks[i].status.hang &&
 		    adev->ip_blocks[i].version->funcs->soft_reset) {
-			r = adev->ip_blocks[i].version->funcs->soft_reset(adev);
+			r = adev->ip_blocks[i].version->funcs->soft_reset(&adev->ip_blocks[i]);
 			if (r)
 				return r;
 		}
@@ -2574,7 +5508,18 @@ static int amdgpu_soft_reset(struct amdgpu_device *adev)
 	return 0;
 }
 
-static int amdgpu_post_soft_reset(struct amdgpu_device *adev)
+/**
+ * amdgpu_device_ip_post_soft_reset - clean up from soft reset
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * The list of all the hardware IPs that make up the asic is walked and the
+ * post_soft_reset callbacks are run if the asic was hung.  post_soft_reset
+ * handles any IP specific hardware or software state changes that are
+ * necessary after the IP has been soft reset.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_post_soft_reset(struct amdgpu_device *adev)
 {
 	int i, r = 0;
 
@@ -2583,7 +5528,7 @@ static int amdgpu_post_soft_reset(struct amdgpu_device *adev)
 			continue;
 		if (adev->ip_blocks[i].status.hang &&
 		    adev->ip_blocks[i].version->funcs->post_soft_reset)
-			r = adev->ip_blocks[i].version->funcs->post_soft_reset(adev);
+			r = adev->ip_blocks[i].version->funcs->post_soft_reset(&adev->ip_blocks[i]);
 		if (r)
 			return r;
 	}
@@ -2591,345 +5536,1137 @@ static int amdgpu_post_soft_reset(struct amdgpu_device *adev)
 	return 0;
 }
 
-bool amdgpu_need_backup(struct amdgpu_device *adev)
+/**
+ * amdgpu_device_reset_sriov - reset ASIC for SR-IOV vf
+ *
+ * @adev: amdgpu_device pointer
+ * @reset_context: amdgpu reset context pointer
+ *
+ * do VF FLR and reinitialize Asic
+ * return 0 means succeeded otherwise failed
+ */
+static int amdgpu_device_reset_sriov(struct amdgpu_device *adev,
+				     struct amdgpu_reset_context *reset_context)
 {
-	if (adev->flags & AMD_IS_APU)
-		return false;
+	int r;
+	struct amdgpu_hive_info *hive = NULL;
 
-	return amdgpu_lockup_timeout > 0 ? true : false;
-}
+	if (test_bit(AMDGPU_HOST_FLR, &reset_context->flags)) {
+		if (!amdgpu_ras_get_fed_status(adev))
+			amdgpu_virt_ready_to_reset(adev);
+		amdgpu_virt_wait_reset(adev);
+		clear_bit(AMDGPU_HOST_FLR, &reset_context->flags);
+		r = amdgpu_virt_request_full_gpu(adev, true);
+	} else {
+		r = amdgpu_virt_reset_gpu(adev);
+	}
+	if (r)
+		return r;
 
-static int amdgpu_recover_vram_from_shadow(struct amdgpu_device *adev,
-					   struct amdgpu_ring *ring,
-					   struct amdgpu_bo *bo,
-					   struct dma_fence **fence)
-{
-	uint32_t domain;
-	int r;
+	amdgpu_ras_clear_err_state(adev);
+	amdgpu_irq_gpu_reset_resume_helper(adev);
 
-	if (!bo->shadow)
-		return 0;
+	/* some sw clean up VF needs to do before recover */
+	amdgpu_virt_post_reset(adev);
 
-	r = amdgpu_bo_reserve(bo, true);
+	/* Resume IP prior to SMC */
+	r = amdgpu_device_ip_reinit_early_sriov(adev);
 	if (r)
 		return r;
-	domain = amdgpu_mem_type_to_domain(bo->tbo.mem.mem_type);
-	/* if bo has been evicted, then no need to recover */
-	if (domain == AMDGPU_GEM_DOMAIN_VRAM) {
-		r = amdgpu_bo_validate(bo->shadow);
-		if (r) {
-			DRM_ERROR("bo validate failed!\n");
-			goto err;
-		}
 
-		r = amdgpu_bo_restore_from_shadow(adev, ring, bo,
-						 NULL, fence, true);
-		if (r) {
-			DRM_ERROR("recover page table failed!\n");
-			goto err;
-		}
-	}
-err:
-	amdgpu_bo_unreserve(bo);
-	return r;
+	amdgpu_virt_init_data_exchange(adev);
+
+	r = amdgpu_device_fw_loading(adev);
+	if (r)
+		return r;
+
+	/* now we are okay to resume SMC/CP/SDMA */
+	r = amdgpu_device_ip_reinit_late_sriov(adev);
+	if (r)
+		return r;
+
+	hive = amdgpu_get_xgmi_hive(adev);
+	/* Update PSP FW topology after reset */
+	if (hive && adev->gmc.xgmi.num_physical_nodes > 1)
+		r = amdgpu_xgmi_update_topology(hive, adev);
+	if (hive)
+		amdgpu_put_xgmi_hive(hive);
+	if (r)
+		return r;
+
+	r = amdgpu_ib_ring_tests(adev);
+	if (r)
+		return r;
+
+	if (adev->virt.gim_feature & AMDGIM_FEATURE_GIM_FLR_VRAMLOST)
+		amdgpu_inc_vram_lost(adev);
+
+	/* need to be called during full access so we can't do it later like
+	 * bare-metal does.
+	 */
+	amdgpu_amdkfd_post_reset(adev);
+	amdgpu_virt_release_full_gpu(adev, true);
+
+	/* Aldebaran and gfx_11_0_3 support ras in SRIOV, so need resume ras during reset */
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(11, 0, 3))
+		amdgpu_ras_resume(adev);
+
+	amdgpu_virt_ras_telemetry_post_reset(adev);
+
+	return 0;
 }
 
 /**
- * amdgpu_sriov_gpu_reset - reset the asic
+ * amdgpu_device_has_job_running - check if there is any unfinished job
  *
- * @adev: amdgpu device pointer
- * @job: which job trigger hang
+ * @adev: amdgpu_device pointer
  *
- * Attempt the reset the GPU if it has hung (all asics).
- * for SRIOV case.
- * Returns 0 for success or an error on failure.
+ * check if there is any job running on the device when guest driver receives
+ * FLR notification from host driver. If there are still jobs running, then
+ * the guest driver will not respond the FLR reset. Instead, let the job hit
+ * the timeout and guest driver then issue the reset request.
  */
-int amdgpu_sriov_gpu_reset(struct amdgpu_device *adev, struct amdgpu_job *job)
+bool amdgpu_device_has_job_running(struct amdgpu_device *adev)
 {
-	int i, j, r = 0;
-	int resched;
-	struct amdgpu_bo *bo, *tmp;
-	struct amdgpu_ring *ring;
-	struct dma_fence *fence = NULL, *next = NULL;
+	int i;
 
-	mutex_lock(&adev->virt.lock_reset);
-	atomic_inc(&adev->gpu_reset_counter);
-	adev->gfx.in_reset = true;
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		struct amdgpu_ring *ring = adev->rings[i];
 
-	/* block TTM */
-	resched = ttm_bo_lock_delayed_workqueue(&adev->mman.bdev);
+		if (!amdgpu_ring_sched_ready(ring))
+			continue;
 
-	/* we start from the ring trigger GPU hang */
-	j = job ? job->ring->idx : 0;
+		if (amdgpu_fence_count_emitted(ring))
+			return true;
+	}
+	return false;
+}
 
-	/* block scheduler */
-	for (i = j; i < j + AMDGPU_MAX_RINGS; ++i) {
-		ring = adev->rings[i % AMDGPU_MAX_RINGS];
-		if (!ring || !ring->sched.thread)
-			continue;
+/**
+ * amdgpu_device_should_recover_gpu - check if we should try GPU recovery
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Check amdgpu_gpu_recovery and SRIOV status to see if we should try to recover
+ * a hung GPU.
+ */
+bool amdgpu_device_should_recover_gpu(struct amdgpu_device *adev)
+{
 
-		kthread_park(ring->sched.thread);
+	if (amdgpu_gpu_recovery == 0)
+		goto disabled;
 
-		if (job && j != i)
-			continue;
+	/* Skip soft reset check in fatal error mode */
+	if (!amdgpu_ras_is_poison_mode_supported(adev))
+		return true;
+
+	if (amdgpu_sriov_vf(adev))
+		return true;
 
-		/* here give the last chance to check if job removed from mirror-list
-		 * since we already pay some time on kthread_park */
-		if (job && list_empty(&job->base.node)) {
-			kthread_unpark(ring->sched.thread);
-			goto give_up_reset;
+	if (amdgpu_gpu_recovery == -1) {
+		switch (adev->asic_type) {
+#ifdef CONFIG_DRM_AMDGPU_SI
+		case CHIP_VERDE:
+		case CHIP_TAHITI:
+		case CHIP_PITCAIRN:
+		case CHIP_OLAND:
+		case CHIP_HAINAN:
+#endif
+#ifdef CONFIG_DRM_AMDGPU_CIK
+		case CHIP_KAVERI:
+		case CHIP_KABINI:
+		case CHIP_MULLINS:
+#endif
+		case CHIP_CARRIZO:
+		case CHIP_STONEY:
+		case CHIP_CYAN_SKILLFISH:
+			goto disabled;
+		default:
+			break;
 		}
+	}
+
+	return true;
+
+disabled:
+		dev_info(adev->dev, "GPU recovery disabled.\n");
+		return false;
+}
+
+int amdgpu_device_mode1_reset(struct amdgpu_device *adev)
+{
+	u32 i;
+	int ret = 0;
+
+	if (adev->bios)
+		amdgpu_atombios_scratch_regs_engine_hung(adev, true);
+
+	dev_info(adev->dev, "GPU mode1 reset\n");
+
+	/* Cache the state before bus master disable. The saved config space
+	 * values are used in other cases like restore after mode-2 reset.
+	 */
+	amdgpu_device_cache_pci_state(adev->pdev);
+
+	/* disable BM */
+	pci_clear_master(adev->pdev);
+
+	if (amdgpu_dpm_is_mode1_reset_supported(adev)) {
+		dev_info(adev->dev, "GPU smu mode1 reset\n");
+		ret = amdgpu_dpm_mode1_reset(adev);
+	} else {
+		dev_info(adev->dev, "GPU psp mode1 reset\n");
+		ret = psp_gpu_reset(adev);
+	}
+
+	if (ret)
+		goto mode1_reset_failed;
+
+	amdgpu_device_load_pci_state(adev->pdev);
+	ret = amdgpu_psp_wait_for_bootloader(adev);
+	if (ret)
+		goto mode1_reset_failed;
+
+	/* wait for asic to come out of reset */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		u32 memsize = adev->nbio.funcs->get_memsize(adev);
+
+		if (memsize != 0xffffffff)
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout) {
+		ret = -ETIMEDOUT;
+		goto mode1_reset_failed;
+	}
+
+	if (adev->bios)
+		amdgpu_atombios_scratch_regs_engine_hung(adev, false);
+
+	return 0;
+
+mode1_reset_failed:
+	dev_err(adev->dev, "GPU mode1 reset failed\n");
+	return ret;
+}
+
+int amdgpu_device_link_reset(struct amdgpu_device *adev)
+{
+	int ret = 0;
+
+	dev_info(adev->dev, "GPU link reset\n");
 
-		if (amd_sched_invalidate_job(&job->base, amdgpu_job_hang_limit))
-			amd_sched_job_kickout(&job->base);
+	if (!amdgpu_reset_in_dpc(adev))
+		ret = amdgpu_dpm_link_reset(adev);
 
-		/* only do job_reset on the hang ring if @job not NULL */
-		amd_sched_hw_job_reset(&ring->sched);
+	if (ret)
+		goto link_reset_failed;
+
+	ret = amdgpu_psp_wait_for_bootloader(adev);
+	if (ret)
+		goto link_reset_failed;
+
+	return 0;
+
+link_reset_failed:
+	dev_err(adev->dev, "GPU link reset failed\n");
+	return ret;
+}
+
+int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev,
+				 struct amdgpu_reset_context *reset_context)
+{
+	int i, r = 0;
+	struct amdgpu_job *job = NULL;
+	struct amdgpu_device *tmp_adev = reset_context->reset_req_dev;
+	bool need_full_reset =
+		test_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags);
+
+	if (reset_context->reset_req_dev == adev)
+		job = reset_context->job;
+
+	if (amdgpu_sriov_vf(adev))
+		amdgpu_virt_pre_reset(adev);
+
+	amdgpu_fence_driver_isr_toggle(adev, true);
+
+	/* block all schedulers and reset given job's ring */
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		struct amdgpu_ring *ring = adev->rings[i];
+
+		if (!amdgpu_ring_sched_ready(ring))
+			continue;
+
+		/* Clear job fence from fence drv to avoid force_completion
+		 * leave NULL and vm flush fence in fence drv
+		 */
+		amdgpu_fence_driver_clear_job_fences(ring);
 
 		/* after all hw jobs are reset, hw fence is meaningless, so force_completion */
-		amdgpu_fence_driver_force_completion_ring(ring);
+		amdgpu_fence_driver_force_completion(ring);
 	}
 
-	/* request to take full control of GPU before re-initialization  */
-	if (job)
-		amdgpu_virt_reset_gpu(adev);
+	amdgpu_fence_driver_isr_toggle(adev, false);
+
+	if (job && job->vm)
+		drm_sched_increase_karma(&job->base);
+
+	r = amdgpu_reset_prepare_hwcontext(adev, reset_context);
+	/* If reset handler not implemented, continue; otherwise return */
+	if (r == -EOPNOTSUPP)
+		r = 0;
 	else
-		amdgpu_virt_request_full_gpu(adev, true);
+		return r;
 
+	/* Don't suspend on bare metal if we are not going to HW reset the ASIC */
+	if (!amdgpu_sriov_vf(adev)) {
 
-	/* Resume IP prior to SMC */
-	amdgpu_sriov_reinit_early(adev);
+		if (!need_full_reset)
+			need_full_reset = amdgpu_device_ip_need_full_reset(adev);
 
-	/* we need recover gart prior to run SMC/CP/SDMA resume */
-	amdgpu_ttm_recover_gart(adev);
+		if (!need_full_reset && amdgpu_gpu_recovery &&
+		    amdgpu_device_ip_check_soft_reset(adev)) {
+			amdgpu_device_ip_pre_soft_reset(adev);
+			r = amdgpu_device_ip_soft_reset(adev);
+			amdgpu_device_ip_post_soft_reset(adev);
+			if (r || amdgpu_device_ip_check_soft_reset(adev)) {
+				dev_info(adev->dev, "soft reset failed, will fallback to full reset!\n");
+				need_full_reset = true;
+			}
+		}
 
-	/* now we are okay to resume SMC/CP/SDMA */
-	amdgpu_sriov_reinit_late(adev);
+		if (!test_bit(AMDGPU_SKIP_COREDUMP, &reset_context->flags)) {
+			dev_info(tmp_adev->dev, "Dumping IP State\n");
+			/* Trigger ip dump before we reset the asic */
+			for (i = 0; i < tmp_adev->num_ip_blocks; i++)
+				if (tmp_adev->ip_blocks[i].version->funcs->dump_ip_state)
+					tmp_adev->ip_blocks[i].version->funcs
+						->dump_ip_state((void *)&tmp_adev->ip_blocks[i]);
+			dev_info(tmp_adev->dev, "Dumping IP State Completed\n");
+		}
 
-	amdgpu_irq_gpu_reset_resume_helper(adev);
+		if (need_full_reset)
+			r = amdgpu_device_ip_suspend(adev);
+		if (need_full_reset)
+			set_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags);
+		else
+			clear_bit(AMDGPU_NEED_FULL_RESET,
+				  &reset_context->flags);
+	}
 
-	if (amdgpu_ib_ring_tests(adev))
-		dev_err(adev->dev, "[GPU_RESET] ib ring test failed (%d).\n", r);
+	return r;
+}
 
-	/* release full control of GPU after ib test */
-	amdgpu_virt_release_full_gpu(adev, true);
+int amdgpu_device_reinit_after_reset(struct amdgpu_reset_context *reset_context)
+{
+	struct list_head *device_list_handle;
+	bool full_reset, vram_lost = false;
+	struct amdgpu_device *tmp_adev;
+	int r, init_level;
+
+	device_list_handle = reset_context->reset_device_list;
+
+	if (!device_list_handle)
+		return -EINVAL;
 
-	DRM_INFO("recover vram bo from shadow\n");
+	full_reset = test_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags);
 
-	ring = adev->mman.buffer_funcs_ring;
-	mutex_lock(&adev->shadow_list_lock);
-	list_for_each_entry_safe(bo, tmp, &adev->shadow_list, shadow_list) {
-		next = NULL;
-		amdgpu_recover_vram_from_shadow(adev, ring, bo, &next);
-		if (fence) {
-			r = dma_fence_wait(fence, false);
+	/**
+	 * If it's reset on init, it's default init level, otherwise keep level
+	 * as recovery level.
+	 */
+	if (reset_context->method == AMD_RESET_METHOD_ON_INIT)
+			init_level = AMDGPU_INIT_LEVEL_DEFAULT;
+	else
+			init_level = AMDGPU_INIT_LEVEL_RESET_RECOVERY;
+
+	r = 0;
+	list_for_each_entry(tmp_adev, device_list_handle, reset_list) {
+		amdgpu_set_init_level(tmp_adev, init_level);
+		if (full_reset) {
+			/* post card */
+			amdgpu_reset_set_dpc_status(tmp_adev, false);
+			amdgpu_ras_clear_err_state(tmp_adev);
+			r = amdgpu_device_asic_init(tmp_adev);
 			if (r) {
-				WARN(r, "recovery from shadow isn't completed\n");
-				break;
+				dev_warn(tmp_adev->dev, "asic atom init failed!");
+			} else {
+				dev_info(tmp_adev->dev, "GPU reset succeeded, trying to resume\n");
+
+				r = amdgpu_device_ip_resume_phase1(tmp_adev);
+				if (r)
+					goto out;
+
+				vram_lost = amdgpu_device_check_vram_lost(tmp_adev);
+
+				if (!test_bit(AMDGPU_SKIP_COREDUMP, &reset_context->flags))
+					amdgpu_coredump(tmp_adev, false, vram_lost, reset_context->job);
+
+				if (vram_lost) {
+					dev_info(
+						tmp_adev->dev,
+						"VRAM is lost due to GPU reset!\n");
+					amdgpu_inc_vram_lost(tmp_adev);
+				}
+
+				r = amdgpu_device_fw_loading(tmp_adev);
+				if (r)
+					return r;
+
+				r = amdgpu_xcp_restore_partition_mode(
+					tmp_adev->xcp_mgr);
+				if (r)
+					goto out;
+
+				r = amdgpu_device_ip_resume_phase2(tmp_adev);
+				if (r)
+					goto out;
+
+				if (tmp_adev->mman.buffer_funcs_ring->sched.ready)
+					amdgpu_ttm_set_buffer_funcs_status(tmp_adev, true);
+
+				r = amdgpu_device_ip_resume_phase3(tmp_adev);
+				if (r)
+					goto out;
+
+				if (vram_lost)
+					amdgpu_device_fill_reset_magic(tmp_adev);
+
+				/*
+				 * Add this ASIC as tracked as reset was already
+				 * complete successfully.
+				 */
+				amdgpu_register_gpu_instance(tmp_adev);
+
+				if (!reset_context->hive &&
+				    tmp_adev->gmc.xgmi.num_physical_nodes > 1)
+					amdgpu_xgmi_add_device(tmp_adev);
+
+				r = amdgpu_device_ip_late_init(tmp_adev);
+				if (r)
+					goto out;
+
+				drm_client_dev_resume(adev_to_drm(tmp_adev), false);
+
+				/*
+				 * The GPU enters bad state once faulty pages
+				 * by ECC has reached the threshold, and ras
+				 * recovery is scheduled next. So add one check
+				 * here to break recovery if it indeed exceeds
+				 * bad page threshold, and remind user to
+				 * retire this GPU or setting one bigger
+				 * bad_page_threshold value to fix this once
+				 * probing driver again.
+				 */
+				if (!amdgpu_ras_is_rma(tmp_adev)) {
+					/* must succeed. */
+					amdgpu_ras_resume(tmp_adev);
+				} else {
+					r = -EINVAL;
+					goto out;
+				}
+
+				/* Update PSP FW topology after reset */
+				if (reset_context->hive &&
+				    tmp_adev->gmc.xgmi.num_physical_nodes > 1)
+					r = amdgpu_xgmi_update_topology(
+						reset_context->hive, tmp_adev);
 			}
 		}
 
-		dma_fence_put(fence);
-		fence = next;
-	}
-	mutex_unlock(&adev->shadow_list_lock);
+out:
+		if (!r) {
+			/* IP init is complete now, set level as default */
+			amdgpu_set_init_level(tmp_adev,
+					      AMDGPU_INIT_LEVEL_DEFAULT);
+			amdgpu_irq_gpu_reset_resume_helper(tmp_adev);
+			r = amdgpu_ib_ring_tests(tmp_adev);
+			if (r) {
+				dev_err(tmp_adev->dev, "ib ring test failed (%d).\n", r);
+				r = -EAGAIN;
+				goto end;
+			}
+		}
 
-	if (fence) {
-		r = dma_fence_wait(fence, false);
 		if (r)
-			WARN(r, "recovery from shadow isn't completed\n");
+			tmp_adev->asic_reset_res = r;
 	}
-	dma_fence_put(fence);
 
-	for (i = j; i < j + AMDGPU_MAX_RINGS; ++i) {
-		ring = adev->rings[i % AMDGPU_MAX_RINGS];
-		if (!ring || !ring->sched.thread)
-			continue;
+end:
+	return r;
+}
 
-		if (job && j != i) {
-			kthread_unpark(ring->sched.thread);
-			continue;
+int amdgpu_do_asic_reset(struct list_head *device_list_handle,
+			 struct amdgpu_reset_context *reset_context)
+{
+	struct amdgpu_device *tmp_adev = NULL;
+	bool need_full_reset, skip_hw_reset;
+	int r = 0;
+
+	/* Try reset handler method first */
+	tmp_adev = list_first_entry(device_list_handle, struct amdgpu_device,
+				    reset_list);
+
+	reset_context->reset_device_list = device_list_handle;
+	r = amdgpu_reset_perform_reset(tmp_adev, reset_context);
+	/* If reset handler not implemented, continue; otherwise return */
+	if (r == -EOPNOTSUPP)
+		r = 0;
+	else
+		return r;
+
+	/* Reset handler not implemented, use the default method */
+	need_full_reset =
+		test_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags);
+	skip_hw_reset = test_bit(AMDGPU_SKIP_HW_RESET, &reset_context->flags);
+
+	/*
+	 * ASIC reset has to be done on all XGMI hive nodes ASAP
+	 * to allow proper links negotiation in FW (within 1 sec)
+	 */
+	if (!skip_hw_reset && need_full_reset) {
+		list_for_each_entry(tmp_adev, device_list_handle, reset_list) {
+			/* For XGMI run all resets in parallel to speed up the process */
+			if (tmp_adev->gmc.xgmi.num_physical_nodes > 1) {
+				if (!queue_work(system_unbound_wq,
+						&tmp_adev->xgmi_reset_work))
+					r = -EALREADY;
+			} else
+				r = amdgpu_asic_reset(tmp_adev);
+
+			if (r) {
+				dev_err(tmp_adev->dev,
+					"ASIC reset failed with error, %d for drm dev, %s",
+					r, adev_to_drm(tmp_adev)->unique);
+				goto out;
+			}
 		}
 
-		amd_sched_job_recovery(&ring->sched);
-		kthread_unpark(ring->sched.thread);
+		/* For XGMI wait for all resets to complete before proceed */
+		if (!r) {
+			list_for_each_entry(tmp_adev, device_list_handle,
+					    reset_list) {
+				if (tmp_adev->gmc.xgmi.num_physical_nodes > 1) {
+					flush_work(&tmp_adev->xgmi_reset_work);
+					r = tmp_adev->asic_reset_res;
+					if (r)
+						break;
+				}
+			}
+		}
 	}
 
-	drm_helper_resume_force_mode(adev->ddev);
-give_up_reset:
-	ttm_bo_unlock_delayed_workqueue(&adev->mman.bdev, resched);
-	if (r) {
-		/* bad news, how to tell it to userspace ? */
-		dev_info(adev->dev, "GPU reset failed\n");
-	} else {
-		dev_info(adev->dev, "GPU reset successed!\n");
+	if (!r && amdgpu_ras_intr_triggered()) {
+		list_for_each_entry(tmp_adev, device_list_handle, reset_list) {
+			amdgpu_ras_reset_error_count(tmp_adev,
+						     AMDGPU_RAS_BLOCK__MMHUB);
+		}
+
+		amdgpu_ras_intr_cleared();
 	}
 
-	adev->gfx.in_reset = false;
-	mutex_unlock(&adev->virt.lock_reset);
+	r = amdgpu_device_reinit_after_reset(reset_context);
+	if (r == -EAGAIN)
+		set_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags);
+	else
+		clear_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags);
+
+out:
 	return r;
 }
 
-/**
- * amdgpu_gpu_reset - reset the asic
- *
- * @adev: amdgpu device pointer
- *
- * Attempt the reset the GPU if it has hung (all asics).
- * Returns 0 for success or an error on failure.
- */
-int amdgpu_gpu_reset(struct amdgpu_device *adev)
+static void amdgpu_device_set_mp1_state(struct amdgpu_device *adev)
 {
-	int i, r;
-	int resched;
-	bool need_full_reset, vram_lost = false;
 
-	if (!amdgpu_check_soft_reset(adev)) {
-		DRM_INFO("No hardware hang detected. Did some blocks stall?\n");
-		return 0;
+	switch (amdgpu_asic_reset_method(adev)) {
+	case AMD_RESET_METHOD_MODE1:
+	case AMD_RESET_METHOD_LINK:
+		adev->mp1_state = PP_MP1_STATE_SHUTDOWN;
+		break;
+	case AMD_RESET_METHOD_MODE2:
+		adev->mp1_state = PP_MP1_STATE_RESET;
+		break;
+	default:
+		adev->mp1_state = PP_MP1_STATE_NONE;
+		break;
 	}
+}
 
-	atomic_inc(&adev->gpu_reset_counter);
+static void amdgpu_device_unset_mp1_state(struct amdgpu_device *adev)
+{
+	amdgpu_vf_error_trans_all(adev);
+	adev->mp1_state = PP_MP1_STATE_NONE;
+}
 
-	/* block TTM */
-	resched = ttm_bo_lock_delayed_workqueue(&adev->mman.bdev);
+static void amdgpu_device_resume_display_audio(struct amdgpu_device *adev)
+{
+	struct pci_dev *p = NULL;
 
-	/* block scheduler */
-	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
-		struct amdgpu_ring *ring = adev->rings[i];
+	p = pci_get_domain_bus_and_slot(pci_domain_nr(adev->pdev->bus),
+			adev->pdev->bus->number, 1);
+	if (p) {
+		pm_runtime_enable(&(p->dev));
+		pm_runtime_resume(&(p->dev));
+	}
 
-		if (!ring || !ring->sched.thread)
-			continue;
-		kthread_park(ring->sched.thread);
-		amd_sched_hw_job_reset(&ring->sched);
+	pci_dev_put(p);
+}
+
+static int amdgpu_device_suspend_display_audio(struct amdgpu_device *adev)
+{
+	enum amd_reset_method reset_method;
+	struct pci_dev *p = NULL;
+	u64 expires;
+
+	/*
+	 * For now, only BACO and mode1 reset are confirmed
+	 * to suffer the audio issue without proper suspended.
+	 */
+	reset_method = amdgpu_asic_reset_method(adev);
+	if ((reset_method != AMD_RESET_METHOD_BACO) &&
+	     (reset_method != AMD_RESET_METHOD_MODE1))
+		return -EINVAL;
+
+	p = pci_get_domain_bus_and_slot(pci_domain_nr(adev->pdev->bus),
+			adev->pdev->bus->number, 1);
+	if (!p)
+		return -ENODEV;
+
+	expires = pm_runtime_autosuspend_expiration(&(p->dev));
+	if (!expires)
+		/*
+		 * If we cannot get the audio device autosuspend delay,
+		 * a fixed 4S interval will be used. Considering 3S is
+		 * the audio controller default autosuspend delay setting.
+		 * 4S used here is guaranteed to cover that.
+		 */
+		expires = ktime_get_mono_fast_ns() + NSEC_PER_SEC * 4ULL;
+
+	while (!pm_runtime_status_suspended(&(p->dev))) {
+		if (!pm_runtime_suspend(&(p->dev)))
+			break;
+
+		if (expires < ktime_get_mono_fast_ns()) {
+			dev_warn(adev->dev, "failed to suspend display audio\n");
+			pci_dev_put(p);
+			/* TODO: abort the succeeding gpu reset? */
+			return -ETIMEDOUT;
+		}
+	}
+
+	pm_runtime_disable(&(p->dev));
+
+	pci_dev_put(p);
+	return 0;
+}
+
+static inline void amdgpu_device_stop_pending_resets(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+
+#if defined(CONFIG_DEBUG_FS)
+	if (!amdgpu_sriov_vf(adev))
+		cancel_work(&adev->reset_work);
+#endif
+
+	if (adev->kfd.dev)
+		cancel_work(&adev->kfd.reset_work);
+
+	if (amdgpu_sriov_vf(adev))
+		cancel_work(&adev->virt.flr_work);
+
+	if (con && adev->ras_enabled)
+		cancel_work(&con->recovery_work);
+
+}
+
+static int amdgpu_device_health_check(struct list_head *device_list_handle)
+{
+	struct amdgpu_device *tmp_adev;
+	int ret = 0;
+
+	list_for_each_entry(tmp_adev, device_list_handle, reset_list) {
+		ret |= amdgpu_device_bus_status_check(tmp_adev);
 	}
-	/* after all hw jobs are reset, hw fence is meaningless, so force_completion */
-	amdgpu_fence_driver_force_completion(adev);
 
-	need_full_reset = amdgpu_need_full_reset(adev);
+	return ret;
+}
 
-	if (!need_full_reset) {
-		amdgpu_pre_soft_reset(adev);
-		r = amdgpu_soft_reset(adev);
-		amdgpu_post_soft_reset(adev);
-		if (r || amdgpu_check_soft_reset(adev)) {
-			DRM_INFO("soft reset failed, will fallback to full reset!\n");
-			need_full_reset = true;
+static void amdgpu_device_recovery_prepare(struct amdgpu_device *adev,
+					  struct list_head *device_list,
+					  struct amdgpu_hive_info *hive)
+{
+	struct amdgpu_device *tmp_adev = NULL;
+
+	/*
+	 * Build list of devices to reset.
+	 * In case we are in XGMI hive mode, resort the device list
+	 * to put adev in the 1st position.
+	 */
+	if (!amdgpu_sriov_vf(adev) && (adev->gmc.xgmi.num_physical_nodes > 1) && hive) {
+		list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) {
+			list_add_tail(&tmp_adev->reset_list, device_list);
+			if (adev->shutdown)
+				tmp_adev->shutdown = true;
+			if (amdgpu_reset_in_dpc(adev))
+				tmp_adev->pcie_reset_ctx.in_link_reset = true;
 		}
+		if (!list_is_first(&adev->reset_list, device_list))
+			list_rotate_to_front(&adev->reset_list, device_list);
+	} else {
+		list_add_tail(&adev->reset_list, device_list);
 	}
+}
 
-	if (need_full_reset) {
-		r = amdgpu_suspend(adev);
+static void amdgpu_device_recovery_get_reset_lock(struct amdgpu_device *adev,
+						  struct list_head *device_list)
+{
+	struct amdgpu_device *tmp_adev = NULL;
 
-retry:
-		amdgpu_atombios_scratch_regs_save(adev);
-		r = amdgpu_asic_reset(adev);
-		amdgpu_atombios_scratch_regs_restore(adev);
-		/* post card */
-		amdgpu_atom_asic_init(adev->mode_info.atom_context);
+	if (list_empty(device_list))
+		return;
+	tmp_adev =
+		list_first_entry(device_list, struct amdgpu_device, reset_list);
+	amdgpu_device_lock_reset_domain(tmp_adev->reset_domain);
+}
 
-		if (!r) {
-			dev_info(adev->dev, "GPU reset succeeded, trying to resume\n");
-			r = amdgpu_resume_phase1(adev);
-			if (r)
-				goto out;
-			vram_lost = amdgpu_check_vram_lost(adev);
-			if (vram_lost) {
-				DRM_ERROR("VRAM is lost!\n");
-				atomic_inc(&adev->vram_lost_counter);
-			}
-			r = amdgpu_ttm_recover_gart(adev);
-			if (r)
-				goto out;
-			r = amdgpu_resume_phase2(adev);
-			if (r)
-				goto out;
-			if (vram_lost)
-				amdgpu_fill_reset_magic(adev);
+static void amdgpu_device_recovery_put_reset_lock(struct amdgpu_device *adev,
+						  struct list_head *device_list)
+{
+	struct amdgpu_device *tmp_adev = NULL;
+
+	if (list_empty(device_list))
+		return;
+	tmp_adev =
+		list_first_entry(device_list, struct amdgpu_device, reset_list);
+	amdgpu_device_unlock_reset_domain(tmp_adev->reset_domain);
+}
+
+static void amdgpu_device_halt_activities(struct amdgpu_device *adev,
+					  struct amdgpu_job *job,
+					  struct amdgpu_reset_context *reset_context,
+					  struct list_head *device_list,
+					  struct amdgpu_hive_info *hive,
+					  bool need_emergency_restart)
+{
+	struct amdgpu_device *tmp_adev = NULL;
+	int i;
+
+	/* block all schedulers and reset given job's ring */
+	list_for_each_entry(tmp_adev, device_list, reset_list) {
+		amdgpu_device_set_mp1_state(tmp_adev);
+
+		/*
+		 * Try to put the audio codec into suspend state
+		 * before gpu reset started.
+		 *
+		 * Due to the power domain of the graphics device
+		 * is shared with AZ power domain. Without this,
+		 * we may change the audio hardware from behind
+		 * the audio driver's back. That will trigger
+		 * some audio codec errors.
+		 */
+		if (!amdgpu_device_suspend_display_audio(tmp_adev))
+			tmp_adev->pcie_reset_ctx.audio_suspended = true;
+
+		amdgpu_ras_set_error_query_ready(tmp_adev, false);
+
+		cancel_delayed_work_sync(&tmp_adev->delayed_init_work);
+
+		amdgpu_amdkfd_pre_reset(tmp_adev, reset_context);
+
+		/*
+		 * Mark these ASICs to be reset as untracked first
+		 * And add them back after reset completed
+		 */
+		amdgpu_unregister_gpu_instance(tmp_adev);
+
+		drm_client_dev_suspend(adev_to_drm(tmp_adev), false);
+
+		/* disable ras on ALL IPs */
+		if (!need_emergency_restart && !amdgpu_reset_in_dpc(adev) &&
+		    amdgpu_device_ip_need_full_reset(tmp_adev))
+			amdgpu_ras_suspend(tmp_adev);
+
+		for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+			struct amdgpu_ring *ring = tmp_adev->rings[i];
+
+			if (!amdgpu_ring_sched_ready(ring))
+				continue;
+
+			drm_sched_stop(&ring->sched, job ? &job->base : NULL);
+
+			if (need_emergency_restart)
+				amdgpu_job_stop_all_jobs_on_sched(&ring->sched);
 		}
+		atomic_inc(&tmp_adev->gpu_reset_counter);
 	}
-out:
-	if (!r) {
-		amdgpu_irq_gpu_reset_resume_helper(adev);
-		r = amdgpu_ib_ring_tests(adev);
+}
+
+static int amdgpu_device_asic_reset(struct amdgpu_device *adev,
+			      struct list_head *device_list,
+			      struct amdgpu_reset_context *reset_context)
+{
+	struct amdgpu_device *tmp_adev = NULL;
+	int retry_limit = AMDGPU_MAX_RETRY_LIMIT;
+	int r = 0;
+
+retry:	/* Rest of adevs pre asic reset from XGMI hive. */
+	list_for_each_entry(tmp_adev, device_list, reset_list) {
+		r = amdgpu_device_pre_asic_reset(tmp_adev, reset_context);
+		/*TODO Should we stop ?*/
 		if (r) {
-			dev_err(adev->dev, "ib ring test failed (%d).\n", r);
-			r = amdgpu_suspend(adev);
-			need_full_reset = true;
+			dev_err(tmp_adev->dev, "GPU pre asic reset failed with err, %d for drm dev, %s ",
+				  r, adev_to_drm(tmp_adev)->unique);
+			tmp_adev->asic_reset_res = r;
+		}
+	}
+
+	/* Actual ASIC resets if needed.*/
+	/* Host driver will handle XGMI hive reset for SRIOV */
+	if (amdgpu_sriov_vf(adev)) {
+
+		/* Bail out of reset early */
+		if (amdgpu_ras_is_rma(adev))
+			return -ENODEV;
+
+		if (amdgpu_ras_get_fed_status(adev) || amdgpu_virt_rcvd_ras_interrupt(adev)) {
+			dev_dbg(adev->dev, "Detected RAS error, wait for FLR completion\n");
+			amdgpu_ras_set_fed(adev, true);
+			set_bit(AMDGPU_HOST_FLR, &reset_context->flags);
+		}
+
+		r = amdgpu_device_reset_sriov(adev, reset_context);
+		if (AMDGPU_RETRY_SRIOV_RESET(r) && (retry_limit--) > 0) {
+			amdgpu_virt_release_full_gpu(adev, true);
 			goto retry;
 		}
-		/**
-		 * recovery vm page tables, since we cannot depend on VRAM is
-		 * consistent after gpu full reset.
+		if (r)
+			adev->asic_reset_res = r;
+	} else {
+		r = amdgpu_do_asic_reset(device_list, reset_context);
+		if (r && r == -EAGAIN)
+			goto retry;
+	}
+
+	list_for_each_entry(tmp_adev, device_list, reset_list) {
+		/*
+		 * Drop any pending non scheduler resets queued before reset is done.
+		 * Any reset scheduled after this point would be valid. Scheduler resets
+		 * were already dropped during drm_sched_stop and no new ones can come
+		 * in before drm_sched_start.
 		 */
-		if (need_full_reset && amdgpu_need_backup(adev)) {
-			struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
-			struct amdgpu_bo *bo, *tmp;
-			struct dma_fence *fence = NULL, *next = NULL;
-
-			DRM_INFO("recover vram bo from shadow\n");
-			mutex_lock(&adev->shadow_list_lock);
-			list_for_each_entry_safe(bo, tmp, &adev->shadow_list, shadow_list) {
-				next = NULL;
-				amdgpu_recover_vram_from_shadow(adev, ring, bo, &next);
-				if (fence) {
-					r = dma_fence_wait(fence, false);
-					if (r) {
-						WARN(r, "recovery from shadow isn't completed\n");
-						break;
-					}
-				}
+		amdgpu_device_stop_pending_resets(tmp_adev);
+	}
+
+	return r;
+}
+
+static int amdgpu_device_sched_resume(struct list_head *device_list,
+			      struct amdgpu_reset_context *reset_context,
+			      bool   job_signaled)
+{
+	struct amdgpu_device *tmp_adev = NULL;
+	int i, r = 0;
+
+	/* Post ASIC reset for all devs .*/
+	list_for_each_entry(tmp_adev, device_list, reset_list) {
 
-				dma_fence_put(fence);
-				fence = next;
-			}
-			mutex_unlock(&adev->shadow_list_lock);
-			if (fence) {
-				r = dma_fence_wait(fence, false);
-				if (r)
-					WARN(r, "recovery from shadow isn't completed\n");
-			}
-			dma_fence_put(fence);
-		}
 		for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
-			struct amdgpu_ring *ring = adev->rings[i];
+			struct amdgpu_ring *ring = tmp_adev->rings[i];
 
-			if (!ring || !ring->sched.thread)
+			if (!amdgpu_ring_sched_ready(ring))
 				continue;
 
-			amd_sched_job_recovery(&ring->sched);
-			kthread_unpark(ring->sched.thread);
+			drm_sched_start(&ring->sched, 0);
 		}
-	} else {
-		dev_err(adev->dev, "asic resume failed (%d).\n", r);
-		amdgpu_vf_error_put(AMDGIM_ERROR_VF_ASIC_RESUME_FAIL, 0, r);
-		for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
-			if (adev->rings[i] && adev->rings[i]->sched.thread) {
-				kthread_unpark(adev->rings[i]->sched.thread);
-			}
+
+		if (!drm_drv_uses_atomic_modeset(adev_to_drm(tmp_adev)) && !job_signaled)
+			drm_helper_resume_force_mode(adev_to_drm(tmp_adev));
+
+		if (tmp_adev->asic_reset_res) {
+			/* bad news, how to tell it to userspace ?
+			 * for ras error, we should report GPU bad status instead of
+			 * reset failure
+			 */
+			if (reset_context->src != AMDGPU_RESET_SRC_RAS ||
+			    !amdgpu_ras_eeprom_check_err_threshold(tmp_adev))
+				dev_info(
+					tmp_adev->dev,
+					"GPU reset(%d) failed with error %d \n",
+					atomic_read(
+						&tmp_adev->gpu_reset_counter),
+					tmp_adev->asic_reset_res);
+			amdgpu_vf_error_put(tmp_adev,
+					    AMDGIM_ERROR_VF_GPU_RESET_FAIL, 0,
+					    tmp_adev->asic_reset_res);
+			if (!r)
+				r = tmp_adev->asic_reset_res;
+			tmp_adev->asic_reset_res = 0;
+		} else {
+			dev_info(tmp_adev->dev, "GPU reset(%d) succeeded!\n",
+				 atomic_read(&tmp_adev->gpu_reset_counter));
+			if (amdgpu_acpi_smart_shift_update(tmp_adev,
+							   AMDGPU_SS_DEV_D0))
+				dev_warn(tmp_adev->dev,
+					 "smart shift update failed\n");
 		}
 	}
 
-	drm_helper_resume_force_mode(adev->ddev);
+	return r;
+}
 
-	ttm_bo_unlock_delayed_workqueue(&adev->mman.bdev, resched);
-	if (r) {
-		/* bad news, how to tell it to userspace ? */
-		dev_info(adev->dev, "GPU reset failed\n");
-		amdgpu_vf_error_put(AMDGIM_ERROR_VF_GPU_RESET_FAIL, 0, r);
+static void amdgpu_device_gpu_resume(struct amdgpu_device *adev,
+			      struct list_head *device_list,
+			      bool   need_emergency_restart)
+{
+	struct amdgpu_device *tmp_adev = NULL;
+
+	list_for_each_entry(tmp_adev, device_list, reset_list) {
+		/* unlock kfd: SRIOV would do it separately */
+		if (!need_emergency_restart && !amdgpu_sriov_vf(tmp_adev))
+			amdgpu_amdkfd_post_reset(tmp_adev);
+
+		/* kfd_post_reset will do nothing if kfd device is not initialized,
+		 * need to bring up kfd here if it's not be initialized before
+		 */
+		if (!adev->kfd.init_complete)
+			amdgpu_amdkfd_device_init(adev);
+
+		if (tmp_adev->pcie_reset_ctx.audio_suspended)
+			amdgpu_device_resume_display_audio(tmp_adev);
+
+		amdgpu_device_unset_mp1_state(tmp_adev);
+
+		amdgpu_ras_set_error_query_ready(tmp_adev, true);
+
+	}
+}
+
+
+/**
+ * amdgpu_device_gpu_recover - reset the asic and recover scheduler
+ *
+ * @adev: amdgpu_device pointer
+ * @job: which job trigger hang
+ * @reset_context: amdgpu reset context pointer
+ *
+ * Attempt to reset the GPU if it has hung (all asics).
+ * Attempt to do soft-reset or full-reset and reinitialize Asic
+ * Returns 0 for success or an error on failure.
+ */
+
+int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
+			      struct amdgpu_job *job,
+			      struct amdgpu_reset_context *reset_context)
+{
+	struct list_head device_list;
+	bool job_signaled = false;
+	struct amdgpu_hive_info *hive = NULL;
+	int r = 0;
+	bool need_emergency_restart = false;
+
+	/*
+	 * If it reaches here because of hang/timeout and a RAS error is
+	 * detected at the same time, let RAS recovery take care of it.
+	 */
+	if (amdgpu_ras_is_err_state(adev, AMDGPU_RAS_BLOCK__ANY) &&
+	    !amdgpu_sriov_vf(adev) &&
+	    reset_context->src != AMDGPU_RESET_SRC_RAS) {
+		dev_dbg(adev->dev,
+			"Gpu recovery from source: %d yielding to RAS error recovery handling",
+			reset_context->src);
+		return 0;
 	}
-	else {
-		dev_info(adev->dev, "GPU reset successed!\n");
+
+	/*
+	 * Special case: RAS triggered and full reset isn't supported
+	 */
+	need_emergency_restart = amdgpu_ras_need_emergency_restart(adev);
+
+	/*
+	 * Flush RAM to disk so that after reboot
+	 * the user can read log and see why the system rebooted.
+	 */
+	if (need_emergency_restart && amdgpu_ras_get_context(adev) &&
+		amdgpu_ras_get_context(adev)->reboot) {
+		dev_warn(adev->dev, "Emergency reboot.");
+
+		ksys_sync_helper();
+		emergency_restart();
+	}
+
+	dev_info(adev->dev, "GPU %s begin!. Source:  %d\n",
+		 need_emergency_restart ? "jobs stop" : "reset",
+		 reset_context->src);
+
+	if (!amdgpu_sriov_vf(adev))
+		hive = amdgpu_get_xgmi_hive(adev);
+	if (hive)
+		mutex_lock(&hive->hive_lock);
+
+	reset_context->job = job;
+	reset_context->hive = hive;
+	INIT_LIST_HEAD(&device_list);
+
+	amdgpu_device_recovery_prepare(adev, &device_list, hive);
+
+	if (!amdgpu_sriov_vf(adev)) {
+		r = amdgpu_device_health_check(&device_list);
+		if (r)
+			goto end_reset;
+	}
+
+	/* We need to lock reset domain only once both for XGMI and single device */
+	amdgpu_device_recovery_get_reset_lock(adev, &device_list);
+
+	amdgpu_device_halt_activities(adev, job, reset_context, &device_list,
+				      hive, need_emergency_restart);
+	if (need_emergency_restart)
+		goto skip_sched_resume;
+	/*
+	 * Must check guilty signal here since after this point all old
+	 * HW fences are force signaled.
+	 *
+	 * job->base holds a reference to parent fence
+	 */
+	if (job && dma_fence_is_signaled(&job->hw_fence.base)) {
+		job_signaled = true;
+		dev_info(adev->dev, "Guilty job already signaled, skipping HW reset");
+		goto skip_hw_reset;
+	}
+
+	r = amdgpu_device_asic_reset(adev, &device_list, reset_context);
+	if (r)
+		goto reset_unlock;
+skip_hw_reset:
+	r = amdgpu_device_sched_resume(&device_list, reset_context, job_signaled);
+	if (r)
+		goto reset_unlock;
+skip_sched_resume:
+	amdgpu_device_gpu_resume(adev, &device_list, need_emergency_restart);
+reset_unlock:
+	amdgpu_device_recovery_put_reset_lock(adev, &device_list);
+end_reset:
+	if (hive) {
+		mutex_unlock(&hive->hive_lock);
+		amdgpu_put_xgmi_hive(hive);
+	}
+
+	if (r)
+		dev_info(adev->dev, "GPU reset end with ret = %d\n", r);
+
+	atomic_set(&adev->reset_domain->reset_res, r);
+
+	if (!r) {
+		struct amdgpu_task_info *ti = NULL;
+
+		if (job)
+			ti = amdgpu_vm_get_task_info_pasid(adev, job->pasid);
+
+		drm_dev_wedged_event(adev_to_drm(adev), DRM_WEDGE_RECOVERY_NONE,
+				     ti ? &ti->task : NULL);
+
+		amdgpu_vm_put_task_info(ti);
 	}
 
-	amdgpu_vf_error_trans_all(adev);
 	return r;
 }
 
-void amdgpu_get_pcie_info(struct amdgpu_device *adev)
+/**
+ * amdgpu_device_partner_bandwidth - find the bandwidth of appropriate partner
+ *
+ * @adev: amdgpu_device pointer
+ * @speed: pointer to the speed of the link
+ * @width: pointer to the width of the link
+ *
+ * Evaluate the hierarchy to find the speed and bandwidth capabilities of the
+ * first physical partner to an AMD dGPU.
+ * This will exclude any virtual switches and links.
+ */
+static void amdgpu_device_partner_bandwidth(struct amdgpu_device *adev,
+					    enum pci_bus_speed *speed,
+					    enum pcie_link_width *width)
 {
-	u32 mask;
-	int ret;
+	struct pci_dev *parent = adev->pdev;
+
+	if (!speed || !width)
+		return;
+
+	*speed = PCI_SPEED_UNKNOWN;
+	*width = PCIE_LNK_WIDTH_UNKNOWN;
+
+	if (amdgpu_device_pcie_dynamic_switching_supported(adev)) {
+		while ((parent = pci_upstream_bridge(parent))) {
+			/* skip upstream/downstream switches internal to dGPU*/
+			if (parent->vendor == PCI_VENDOR_ID_ATI)
+				continue;
+			*speed = pcie_get_speed_cap(parent);
+			*width = pcie_get_width_cap(parent);
+			break;
+		}
+	} else {
+		/* use the current speeds rather than max if switching is not supported */
+		pcie_bandwidth_available(adev->pdev, NULL, speed, width);
+	}
+}
+
+/**
+ * amdgpu_device_gpu_bandwidth - find the bandwidth of the GPU
+ *
+ * @adev: amdgpu_device pointer
+ * @speed: pointer to the speed of the link
+ * @width: pointer to the width of the link
+ *
+ * Evaluate the hierarchy to find the speed and bandwidth capabilities of the
+ * AMD dGPU which may be a virtual upstream bridge.
+ */
+static void amdgpu_device_gpu_bandwidth(struct amdgpu_device *adev,
+					enum pci_bus_speed *speed,
+					enum pcie_link_width *width)
+{
+	struct pci_dev *parent = adev->pdev;
+
+	if (!speed || !width)
+		return;
+
+	parent = pci_upstream_bridge(parent);
+	if (parent && parent->vendor == PCI_VENDOR_ID_ATI) {
+		/* use the upstream/downstream switches internal to dGPU */
+		*speed = pcie_get_speed_cap(parent);
+		*width = pcie_get_width_cap(parent);
+		while ((parent = pci_upstream_bridge(parent))) {
+			if (parent->vendor == PCI_VENDOR_ID_ATI) {
+				/* use the upstream/downstream switches internal to dGPU */
+				*speed = pcie_get_speed_cap(parent);
+				*width = pcie_get_width_cap(parent);
+			}
+		}
+	} else {
+		/* use the device itself */
+		*speed = pcie_get_speed_cap(adev->pdev);
+		*width = pcie_get_width_cap(adev->pdev);
+	}
+}
+
+/**
+ * amdgpu_device_get_pcie_info - fence pcie info about the PCIE slot
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Fetches and stores in the driver the PCIE capabilities (gen speed
+ * and lanes) of the slot the device is in. Handles APUs and
+ * virtualized environments where PCIE config space may not be available.
+ */
+static void amdgpu_device_get_pcie_info(struct amdgpu_device *adev)
+{
+	enum pci_bus_speed speed_cap, platform_speed_cap;
+	enum pcie_link_width platform_link_width, link_width;
 
 	if (amdgpu_pcie_gen_cap)
 		adev->pm.pcie_gen_mask = amdgpu_pcie_gen_cap;
@@ -2938,7 +6675,7 @@ void amdgpu_get_pcie_info(struct amdgpu_device *adev)
 		adev->pm.pcie_mlw_mask = amdgpu_pcie_lane_cap;
 
 	/* covers APUs as well */
-	if (pci_is_root_bus(adev->pdev->bus)) {
+	if (pci_is_root_bus(adev->pdev->bus) && !amdgpu_passthrough(adev)) {
 		if (adev->pm.pcie_gen_mask == 0)
 			adev->pm.pcie_gen_mask = AMDGPU_DEFAULT_PCIE_GEN_MASK;
 		if (adev->pm.pcie_mlw_mask == 0)
@@ -2946,822 +6683,1001 @@ void amdgpu_get_pcie_info(struct amdgpu_device *adev)
 		return;
 	}
 
+	if (adev->pm.pcie_gen_mask && adev->pm.pcie_mlw_mask)
+		return;
+
+	amdgpu_device_partner_bandwidth(adev, &platform_speed_cap,
+					&platform_link_width);
+	amdgpu_device_gpu_bandwidth(adev, &speed_cap, &link_width);
+
 	if (adev->pm.pcie_gen_mask == 0) {
-		ret = drm_pcie_get_speed_cap_mask(adev->ddev, &mask);
-		if (!ret) {
-			adev->pm.pcie_gen_mask = (CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN1 |
+		/* asic caps */
+		if (speed_cap == PCI_SPEED_UNKNOWN) {
+			adev->pm.pcie_gen_mask |= (CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN1 |
 						  CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN2 |
 						  CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN3);
-
-			if (mask & DRM_PCIE_SPEED_25)
-				adev->pm.pcie_gen_mask |= CAIL_PCIE_LINK_SPEED_SUPPORT_GEN1;
-			if (mask & DRM_PCIE_SPEED_50)
-				adev->pm.pcie_gen_mask |= CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2;
-			if (mask & DRM_PCIE_SPEED_80)
-				adev->pm.pcie_gen_mask |= CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3;
 		} else {
-			adev->pm.pcie_gen_mask = AMDGPU_DEFAULT_PCIE_GEN_MASK;
+			if (speed_cap == PCIE_SPEED_32_0GT)
+				adev->pm.pcie_gen_mask |= (CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN1 |
+							  CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN2 |
+							  CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN3 |
+							  CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN4 |
+							  CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN5);
+			else if (speed_cap == PCIE_SPEED_16_0GT)
+				adev->pm.pcie_gen_mask |= (CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN1 |
+							  CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN2 |
+							  CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN3 |
+							  CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN4);
+			else if (speed_cap == PCIE_SPEED_8_0GT)
+				adev->pm.pcie_gen_mask |= (CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN1 |
+							  CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN2 |
+							  CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN3);
+			else if (speed_cap == PCIE_SPEED_5_0GT)
+				adev->pm.pcie_gen_mask |= (CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN1 |
+							  CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN2);
+			else
+				adev->pm.pcie_gen_mask |= CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN1;
+		}
+		/* platform caps */
+		if (platform_speed_cap == PCI_SPEED_UNKNOWN) {
+			adev->pm.pcie_gen_mask |= (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN1 |
+						   CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2);
+		} else {
+			if (platform_speed_cap == PCIE_SPEED_32_0GT)
+				adev->pm.pcie_gen_mask |= (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN1 |
+							   CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2 |
+							   CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3 |
+							   CAIL_PCIE_LINK_SPEED_SUPPORT_GEN4 |
+							   CAIL_PCIE_LINK_SPEED_SUPPORT_GEN5);
+			else if (platform_speed_cap == PCIE_SPEED_16_0GT)
+				adev->pm.pcie_gen_mask |= (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN1 |
+							   CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2 |
+							   CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3 |
+							   CAIL_PCIE_LINK_SPEED_SUPPORT_GEN4);
+			else if (platform_speed_cap == PCIE_SPEED_8_0GT)
+				adev->pm.pcie_gen_mask |= (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN1 |
+							   CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2 |
+							   CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3);
+			else if (platform_speed_cap == PCIE_SPEED_5_0GT)
+				adev->pm.pcie_gen_mask |= (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN1 |
+							   CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2);
+			else
+				adev->pm.pcie_gen_mask |= CAIL_PCIE_LINK_SPEED_SUPPORT_GEN1;
+
 		}
 	}
 	if (adev->pm.pcie_mlw_mask == 0) {
-		ret = drm_pcie_get_max_link_width(adev->ddev, &mask);
-		if (!ret) {
-			switch (mask) {
-			case 32:
-				adev->pm.pcie_mlw_mask = (CAIL_PCIE_LINK_WIDTH_SUPPORT_X32 |
-							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X16 |
-							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X12 |
-							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X8 |
-							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X4 |
-							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 |
-							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X1);
+		/* asic caps */
+		if (link_width == PCIE_LNK_WIDTH_UNKNOWN) {
+			adev->pm.pcie_mlw_mask |= AMDGPU_DEFAULT_ASIC_PCIE_MLW_MASK;
+		} else {
+			switch (link_width) {
+			case PCIE_LNK_X32:
+				adev->pm.pcie_mlw_mask |= (CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X32 |
+							   CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X16 |
+							   CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X12 |
+							   CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X8 |
+							   CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X4 |
+							   CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X2 |
+							   CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X1);
 				break;
-			case 16:
-				adev->pm.pcie_mlw_mask = (CAIL_PCIE_LINK_WIDTH_SUPPORT_X16 |
-							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X12 |
-							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X8 |
-							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X4 |
-							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 |
-							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X1);
+			case PCIE_LNK_X16:
+				adev->pm.pcie_mlw_mask |= (CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X16 |
+							   CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X12 |
+							   CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X8 |
+							   CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X4 |
+							   CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X2 |
+							   CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X1);
 				break;
-			case 12:
-				adev->pm.pcie_mlw_mask = (CAIL_PCIE_LINK_WIDTH_SUPPORT_X12 |
-							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X8 |
-							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X4 |
-							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 |
-							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X1);
+			case PCIE_LNK_X12:
+				adev->pm.pcie_mlw_mask |= (CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X12 |
+							   CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X8 |
+							   CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X4 |
+							   CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X2 |
+							   CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X1);
 				break;
-			case 8:
-				adev->pm.pcie_mlw_mask = (CAIL_PCIE_LINK_WIDTH_SUPPORT_X8 |
-							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X4 |
-							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 |
-							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X1);
+			case PCIE_LNK_X8:
+				adev->pm.pcie_mlw_mask |= (CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X8 |
+							   CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X4 |
+							   CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X2 |
+							   CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X1);
 				break;
-			case 4:
-				adev->pm.pcie_mlw_mask = (CAIL_PCIE_LINK_WIDTH_SUPPORT_X4 |
-							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 |
-							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X1);
+			case PCIE_LNK_X4:
+				adev->pm.pcie_mlw_mask |= (CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X4 |
+							   CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X2 |
+							   CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X1);
 				break;
-			case 2:
-				adev->pm.pcie_mlw_mask = (CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 |
-							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X1);
+			case PCIE_LNK_X2:
+				adev->pm.pcie_mlw_mask |= (CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X2 |
+							   CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X1);
 				break;
-			case 1:
-				adev->pm.pcie_mlw_mask = CAIL_PCIE_LINK_WIDTH_SUPPORT_X1;
+			case PCIE_LNK_X1:
+				adev->pm.pcie_mlw_mask |= CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X1;
 				break;
 			default:
 				break;
 			}
+		}
+		/* platform caps */
+		if (platform_link_width == PCIE_LNK_WIDTH_UNKNOWN) {
+			adev->pm.pcie_mlw_mask |= AMDGPU_DEFAULT_PCIE_MLW_MASK;
 		} else {
-			adev->pm.pcie_mlw_mask = AMDGPU_DEFAULT_PCIE_MLW_MASK;
+			switch (platform_link_width) {
+			case PCIE_LNK_X32:
+				adev->pm.pcie_mlw_mask |= (CAIL_PCIE_LINK_WIDTH_SUPPORT_X32 |
+							   CAIL_PCIE_LINK_WIDTH_SUPPORT_X16 |
+							   CAIL_PCIE_LINK_WIDTH_SUPPORT_X12 |
+							   CAIL_PCIE_LINK_WIDTH_SUPPORT_X8 |
+							   CAIL_PCIE_LINK_WIDTH_SUPPORT_X4 |
+							   CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 |
+							   CAIL_PCIE_LINK_WIDTH_SUPPORT_X1);
+				break;
+			case PCIE_LNK_X16:
+				adev->pm.pcie_mlw_mask |= (CAIL_PCIE_LINK_WIDTH_SUPPORT_X16 |
+							   CAIL_PCIE_LINK_WIDTH_SUPPORT_X12 |
+							   CAIL_PCIE_LINK_WIDTH_SUPPORT_X8 |
+							   CAIL_PCIE_LINK_WIDTH_SUPPORT_X4 |
+							   CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 |
+							   CAIL_PCIE_LINK_WIDTH_SUPPORT_X1);
+				break;
+			case PCIE_LNK_X12:
+				adev->pm.pcie_mlw_mask |= (CAIL_PCIE_LINK_WIDTH_SUPPORT_X12 |
+							   CAIL_PCIE_LINK_WIDTH_SUPPORT_X8 |
+							   CAIL_PCIE_LINK_WIDTH_SUPPORT_X4 |
+							   CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 |
+							   CAIL_PCIE_LINK_WIDTH_SUPPORT_X1);
+				break;
+			case PCIE_LNK_X8:
+				adev->pm.pcie_mlw_mask |= (CAIL_PCIE_LINK_WIDTH_SUPPORT_X8 |
+							   CAIL_PCIE_LINK_WIDTH_SUPPORT_X4 |
+							   CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 |
+							   CAIL_PCIE_LINK_WIDTH_SUPPORT_X1);
+				break;
+			case PCIE_LNK_X4:
+				adev->pm.pcie_mlw_mask |= (CAIL_PCIE_LINK_WIDTH_SUPPORT_X4 |
+							   CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 |
+							   CAIL_PCIE_LINK_WIDTH_SUPPORT_X1);
+				break;
+			case PCIE_LNK_X2:
+				adev->pm.pcie_mlw_mask |= (CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 |
+							   CAIL_PCIE_LINK_WIDTH_SUPPORT_X1);
+				break;
+			case PCIE_LNK_X1:
+				adev->pm.pcie_mlw_mask |= CAIL_PCIE_LINK_WIDTH_SUPPORT_X1;
+				break;
+			default:
+				break;
+			}
 		}
 	}
 }
 
-/*
- * Debugfs
+/**
+ * amdgpu_device_is_peer_accessible - Check peer access through PCIe BAR
+ *
+ * @adev: amdgpu_device pointer
+ * @peer_adev: amdgpu_device pointer for peer device trying to access @adev
+ *
+ * Return true if @peer_adev can access (DMA) @adev through the PCIe
+ * BAR, i.e. @adev is "large BAR" and the BAR matches the DMA mask of
+ * @peer_adev.
  */
-int amdgpu_debugfs_add_files(struct amdgpu_device *adev,
-			     const struct drm_info_list *files,
-			     unsigned nfiles)
+bool amdgpu_device_is_peer_accessible(struct amdgpu_device *adev,
+				      struct amdgpu_device *peer_adev)
 {
-	unsigned i;
-
-	for (i = 0; i < adev->debugfs_count; i++) {
-		if (adev->debugfs[i].files == files) {
-			/* Already registered */
-			return 0;
-		}
-	}
-
-	i = adev->debugfs_count + 1;
-	if (i > AMDGPU_DEBUGFS_MAX_COMPONENTS) {
-		DRM_ERROR("Reached maximum number of debugfs components.\n");
-		DRM_ERROR("Report so we increase "
-			  "AMDGPU_DEBUGFS_MAX_COMPONENTS.\n");
-		return -EINVAL;
-	}
-	adev->debugfs[adev->debugfs_count].files = files;
-	adev->debugfs[adev->debugfs_count].num_files = nfiles;
-	adev->debugfs_count = i;
-#if defined(CONFIG_DEBUG_FS)
-	drm_debugfs_create_files(files, nfiles,
-				 adev->ddev->primary->debugfs_root,
-				 adev->ddev->primary);
+#ifdef CONFIG_HSA_AMD_P2P
+	bool p2p_access =
+		!adev->gmc.xgmi.connected_to_cpu &&
+		!(pci_p2pdma_distance(adev->pdev, peer_adev->dev, false) < 0);
+	if (!p2p_access)
+		dev_info(adev->dev, "PCIe P2P access from peer device %s is not supported by the chipset\n",
+			pci_name(peer_adev->pdev));
+
+	bool is_large_bar = adev->gmc.visible_vram_size &&
+		adev->gmc.real_vram_size == adev->gmc.visible_vram_size;
+	bool p2p_addressable = amdgpu_device_check_iommu_remap(peer_adev);
+
+	if (!p2p_addressable) {
+		uint64_t address_mask = peer_adev->dev->dma_mask ?
+			~*peer_adev->dev->dma_mask : ~((1ULL << 32) - 1);
+		resource_size_t aper_limit =
+			adev->gmc.aper_base + adev->gmc.aper_size - 1;
+
+		p2p_addressable = !(adev->gmc.aper_base & address_mask ||
+				     aper_limit & address_mask);
+	}
+	return pcie_p2p && is_large_bar && p2p_access && p2p_addressable;
+#else
+	return false;
 #endif
-	return 0;
 }
 
-#if defined(CONFIG_DEBUG_FS)
-
-static ssize_t amdgpu_debugfs_regs_read(struct file *f, char __user *buf,
-					size_t size, loff_t *pos)
+int amdgpu_device_baco_enter(struct amdgpu_device *adev)
 {
-	struct amdgpu_device *adev = file_inode(f)->i_private;
-	ssize_t result = 0;
-	int r;
-	bool pm_pg_lock, use_bank;
-	unsigned instance_bank, sh_bank, se_bank;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
 
-	if (size & 0x3 || *pos & 0x3)
-		return -EINVAL;
+	if (!amdgpu_device_supports_baco(adev))
+		return -ENOTSUPP;
 
-	/* are we reading registers for which a PG lock is necessary? */
-	pm_pg_lock = (*pos >> 23) & 1;
-
-	if (*pos & (1ULL << 62)) {
-		se_bank = (*pos >> 24) & 0x3FF;
-		sh_bank = (*pos >> 34) & 0x3FF;
-		instance_bank = (*pos >> 44) & 0x3FF;
-
-		if (se_bank == 0x3FF)
-			se_bank = 0xFFFFFFFF;
-		if (sh_bank == 0x3FF)
-			sh_bank = 0xFFFFFFFF;
-		if (instance_bank == 0x3FF)
-			instance_bank = 0xFFFFFFFF;
-		use_bank = 1;
-	} else {
-		use_bank = 0;
-	}
+	if (ras && adev->ras_enabled &&
+	    adev->nbio.funcs->enable_doorbell_interrupt)
+		adev->nbio.funcs->enable_doorbell_interrupt(adev, false);
 
-	*pos &= (1UL << 22) - 1;
+	return amdgpu_dpm_baco_enter(adev);
+}
 
-	if (use_bank) {
-		if ((sh_bank != 0xFFFFFFFF && sh_bank >= adev->gfx.config.max_sh_per_se) ||
-		    (se_bank != 0xFFFFFFFF && se_bank >= adev->gfx.config.max_shader_engines))
-			return -EINVAL;
-		mutex_lock(&adev->grbm_idx_mutex);
-		amdgpu_gfx_select_se_sh(adev, se_bank,
-					sh_bank, instance_bank);
-	}
+int amdgpu_device_baco_exit(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+	int ret = 0;
 
-	if (pm_pg_lock)
-		mutex_lock(&adev->pm.mutex);
+	if (!amdgpu_device_supports_baco(adev))
+		return -ENOTSUPP;
 
-	while (size) {
-		uint32_t value;
+	ret = amdgpu_dpm_baco_exit(adev);
+	if (ret)
+		return ret;
 
-		if (*pos > adev->rmmio_size)
-			goto end;
+	if (ras && adev->ras_enabled &&
+	    adev->nbio.funcs->enable_doorbell_interrupt)
+		adev->nbio.funcs->enable_doorbell_interrupt(adev, true);
 
-		value = RREG32(*pos >> 2);
-		r = put_user(value, (uint32_t *)buf);
-		if (r) {
-			result = r;
-			goto end;
-		}
+	if (amdgpu_passthrough(adev) && adev->nbio.funcs &&
+	    adev->nbio.funcs->clear_doorbell_interrupt)
+		adev->nbio.funcs->clear_doorbell_interrupt(adev);
 
-		result += 4;
-		buf += 4;
-		*pos += 4;
-		size -= 4;
-	}
-
-end:
-	if (use_bank) {
-		amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
-		mutex_unlock(&adev->grbm_idx_mutex);
-	}
+	return 0;
+}
 
-	if (pm_pg_lock)
-		mutex_unlock(&adev->pm.mutex);
+/**
+ * amdgpu_pci_error_detected - Called when a PCI error is detected.
+ * @pdev: PCI device struct
+ * @state: PCI channel state
+ *
+ * Description: Called when a PCI error is detected.
+ *
+ * Return: PCI_ERS_RESULT_NEED_RESET or PCI_ERS_RESULT_DISCONNECT.
+ */
+pci_ers_result_t amdgpu_pci_error_detected(struct pci_dev *pdev, pci_channel_state_t state)
+{
+	struct drm_device *dev = pci_get_drvdata(pdev);
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	struct amdgpu_hive_info *hive __free(xgmi_put_hive) =
+		amdgpu_get_xgmi_hive(adev);
+	struct amdgpu_reset_context reset_context;
+	struct list_head device_list;
+
+	dev_info(adev->dev, "PCI error: detected callback!!\n");
+
+	adev->pci_channel_state = state;
+
+	switch (state) {
+	case pci_channel_io_normal:
+		dev_info(adev->dev, "pci_channel_io_normal: state(%d)!!\n", state);
+		return PCI_ERS_RESULT_CAN_RECOVER;
+	case pci_channel_io_frozen:
+		/* Fatal error, prepare for slot reset */
+		dev_info(adev->dev, "pci_channel_io_frozen: state(%d)!!\n", state);
+		if (hive) {
+			/* Hive devices should be able to support FW based
+			 * link reset on other devices, if not return.
+			 */
+			if (!amdgpu_dpm_is_link_reset_supported(adev)) {
+				dev_warn(adev->dev,
+					 "No support for XGMI hive yet...\n");
+				return PCI_ERS_RESULT_DISCONNECT;
+			}
+			/* Set dpc status only if device is part of hive
+			 * Non-hive devices should be able to recover after
+			 * link reset.
+			 */
+			amdgpu_reset_set_dpc_status(adev, true);
 
-	return result;
+			mutex_lock(&hive->hive_lock);
+		}
+		memset(&reset_context, 0, sizeof(reset_context));
+		INIT_LIST_HEAD(&device_list);
+
+		amdgpu_device_recovery_prepare(adev, &device_list, hive);
+		amdgpu_device_recovery_get_reset_lock(adev, &device_list);
+		amdgpu_device_halt_activities(adev, NULL, &reset_context, &device_list,
+					      hive, false);
+		if (hive)
+			mutex_unlock(&hive->hive_lock);
+		return PCI_ERS_RESULT_NEED_RESET;
+	case pci_channel_io_perm_failure:
+		/* Permanent error, prepare for device removal */
+		dev_info(adev->dev, "pci_channel_io_perm_failure: state(%d)!!\n", state);
+		return PCI_ERS_RESULT_DISCONNECT;
+	}
+
+	return PCI_ERS_RESULT_NEED_RESET;
 }
 
-static ssize_t amdgpu_debugfs_regs_write(struct file *f, const char __user *buf,
-					 size_t size, loff_t *pos)
+/**
+ * amdgpu_pci_mmio_enabled - Enable MMIO and dump debug registers
+ * @pdev: pointer to PCI device
+ */
+pci_ers_result_t amdgpu_pci_mmio_enabled(struct pci_dev *pdev)
 {
-	struct amdgpu_device *adev = file_inode(f)->i_private;
-	ssize_t result = 0;
-	int r;
-	bool pm_pg_lock, use_bank;
-	unsigned instance_bank, sh_bank, se_bank;
+	struct drm_device *dev = pci_get_drvdata(pdev);
+	struct amdgpu_device *adev = drm_to_adev(dev);
 
-	if (size & 0x3 || *pos & 0x3)
-		return -EINVAL;
+	dev_info(adev->dev, "PCI error: mmio enabled callback!!\n");
 
-	/* are we reading registers for which a PG lock is necessary? */
-	pm_pg_lock = (*pos >> 23) & 1;
-
-	if (*pos & (1ULL << 62)) {
-		se_bank = (*pos >> 24) & 0x3FF;
-		sh_bank = (*pos >> 34) & 0x3FF;
-		instance_bank = (*pos >> 44) & 0x3FF;
-
-		if (se_bank == 0x3FF)
-			se_bank = 0xFFFFFFFF;
-		if (sh_bank == 0x3FF)
-			sh_bank = 0xFFFFFFFF;
-		if (instance_bank == 0x3FF)
-			instance_bank = 0xFFFFFFFF;
-		use_bank = 1;
-	} else {
-		use_bank = 0;
-	}
+	/* TODO - dump whatever for debugging purposes */
 
-	*pos &= (1UL << 22) - 1;
+	/* This called only if amdgpu_pci_error_detected returns
+	 * PCI_ERS_RESULT_CAN_RECOVER. Read/write to the device still
+	 * works, no need to reset slot.
+	 */
 
-	if (use_bank) {
-		if ((sh_bank != 0xFFFFFFFF && sh_bank >= adev->gfx.config.max_sh_per_se) ||
-		    (se_bank != 0xFFFFFFFF && se_bank >= adev->gfx.config.max_shader_engines))
-			return -EINVAL;
-		mutex_lock(&adev->grbm_idx_mutex);
-		amdgpu_gfx_select_se_sh(adev, se_bank,
-					sh_bank, instance_bank);
-	}
+	return PCI_ERS_RESULT_RECOVERED;
+}
 
-	if (pm_pg_lock)
-		mutex_lock(&adev->pm.mutex);
+/**
+ * amdgpu_pci_slot_reset - Called when PCI slot has been reset.
+ * @pdev: PCI device struct
+ *
+ * Description: This routine is called by the pci error recovery
+ * code after the PCI slot has been reset, just before we
+ * should resume normal operations.
+ */
+pci_ers_result_t amdgpu_pci_slot_reset(struct pci_dev *pdev)
+{
+	struct drm_device *dev = pci_get_drvdata(pdev);
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	struct amdgpu_reset_context reset_context;
+	struct amdgpu_device *tmp_adev;
+	struct amdgpu_hive_info *hive;
+	struct list_head device_list;
+	struct pci_dev *link_dev;
+	int r = 0, i, timeout;
+	u32 memsize;
+	u16 status;
+
+	dev_info(adev->dev, "PCI error: slot reset callback!!\n");
+
+	memset(&reset_context, 0, sizeof(reset_context));
+
+	if (adev->pcie_reset_ctx.swus)
+		link_dev = adev->pcie_reset_ctx.swus;
+	else
+		link_dev = adev->pdev;
+	/* wait for asic to come out of reset, timeout = 10s */
+	timeout = 10000;
+	do {
+		usleep_range(10000, 10500);
+		r = pci_read_config_word(link_dev, PCI_VENDOR_ID, &status);
+		timeout -= 10;
+	} while (timeout > 0 && (status != PCI_VENDOR_ID_ATI) &&
+		 (status != PCI_VENDOR_ID_AMD));
+
+	if ((status != PCI_VENDOR_ID_ATI) && (status != PCI_VENDOR_ID_AMD)) {
+		r = -ETIME;
+		goto out;
+	}
 
-	while (size) {
-		uint32_t value;
+	amdgpu_device_load_switch_state(adev);
+	/* Restore PCI confspace */
+	amdgpu_device_load_pci_state(pdev);
 
-		if (*pos > adev->rmmio_size)
-			return result;
+	/* confirm  ASIC came out of reset */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		memsize = amdgpu_asic_get_config_memsize(adev);
 
-		r = get_user(value, (uint32_t *)buf);
-		if (r)
-			return r;
+		if (memsize != 0xffffffff)
+			break;
+		udelay(1);
+	}
+	if (memsize == 0xffffffff) {
+		r = -ETIME;
+		goto out;
+	}
 
-		WREG32(*pos >> 2, value);
+	reset_context.method = AMD_RESET_METHOD_NONE;
+	reset_context.reset_req_dev = adev;
+	set_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
+	set_bit(AMDGPU_SKIP_COREDUMP, &reset_context.flags);
+	INIT_LIST_HEAD(&device_list);
 
-		result += 4;
-		buf += 4;
-		*pos += 4;
-		size -= 4;
+	hive = amdgpu_get_xgmi_hive(adev);
+	if (hive) {
+		mutex_lock(&hive->hive_lock);
+		reset_context.hive = hive;
+		list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) {
+			tmp_adev->pcie_reset_ctx.in_link_reset = true;
+			list_add_tail(&tmp_adev->reset_list, &device_list);
+		}
+	} else {
+		set_bit(AMDGPU_SKIP_HW_RESET, &reset_context.flags);
+		list_add_tail(&adev->reset_list, &device_list);
 	}
 
-	if (use_bank) {
-		amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
-		mutex_unlock(&adev->grbm_idx_mutex);
+	r = amdgpu_device_asic_reset(adev, &device_list, &reset_context);
+out:
+	if (!r) {
+		if (amdgpu_device_cache_pci_state(adev->pdev))
+			pci_restore_state(adev->pdev);
+		dev_info(adev->dev, "PCIe error recovery succeeded\n");
+	} else {
+		dev_err(adev->dev, "PCIe error recovery failed, err:%d\n", r);
+		if (hive) {
+			list_for_each_entry(tmp_adev, &device_list, reset_list)
+				amdgpu_device_unset_mp1_state(tmp_adev);
+		}
+		amdgpu_device_recovery_put_reset_lock(adev, &device_list);
 	}
 
-	if (pm_pg_lock)
-		mutex_unlock(&adev->pm.mutex);
+	if (hive) {
+		mutex_unlock(&hive->hive_lock);
+		amdgpu_put_xgmi_hive(hive);
+	}
 
-	return result;
+	return r ? PCI_ERS_RESULT_DISCONNECT : PCI_ERS_RESULT_RECOVERED;
 }
 
-static ssize_t amdgpu_debugfs_regs_pcie_read(struct file *f, char __user *buf,
-					size_t size, loff_t *pos)
+/**
+ * amdgpu_pci_resume() - resume normal ops after PCI reset
+ * @pdev: pointer to PCI device
+ *
+ * Called when the error recovery driver tells us that its
+ * OK to resume normal operation.
+ */
+void amdgpu_pci_resume(struct pci_dev *pdev)
 {
-	struct amdgpu_device *adev = file_inode(f)->i_private;
-	ssize_t result = 0;
-	int r;
+	struct drm_device *dev = pci_get_drvdata(pdev);
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	struct list_head device_list;
+	struct amdgpu_hive_info *hive = NULL;
+	struct amdgpu_device *tmp_adev = NULL;
 
-	if (size & 0x3 || *pos & 0x3)
-		return -EINVAL;
+	dev_info(adev->dev, "PCI error: resume callback!!\n");
 
-	while (size) {
-		uint32_t value;
+	/* Only continue execution for the case of pci_channel_io_frozen */
+	if (adev->pci_channel_state != pci_channel_io_frozen)
+		return;
 
-		value = RREG32_PCIE(*pos >> 2);
-		r = put_user(value, (uint32_t *)buf);
-		if (r)
-			return r;
+	INIT_LIST_HEAD(&device_list);
 
-		result += 4;
-		buf += 4;
-		*pos += 4;
-		size -= 4;
-	}
+	hive = amdgpu_get_xgmi_hive(adev);
+	if (hive) {
+		mutex_lock(&hive->hive_lock);
+		list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) {
+			tmp_adev->pcie_reset_ctx.in_link_reset = false;
+			list_add_tail(&tmp_adev->reset_list, &device_list);
+		}
+	} else
+		list_add_tail(&adev->reset_list, &device_list);
+
+	amdgpu_device_sched_resume(&device_list, NULL, NULL);
+	amdgpu_device_gpu_resume(adev, &device_list, false);
+	amdgpu_device_recovery_put_reset_lock(adev, &device_list);
 
-	return result;
+	if (hive) {
+		mutex_unlock(&hive->hive_lock);
+		amdgpu_put_xgmi_hive(hive);
+	}
 }
 
-static ssize_t amdgpu_debugfs_regs_pcie_write(struct file *f, const char __user *buf,
-					 size_t size, loff_t *pos)
+static void amdgpu_device_cache_switch_state(struct amdgpu_device *adev)
 {
-	struct amdgpu_device *adev = file_inode(f)->i_private;
-	ssize_t result = 0;
+	struct pci_dev *swus, *swds;
 	int r;
 
-	if (size & 0x3 || *pos & 0x3)
-		return -EINVAL;
-
-	while (size) {
-		uint32_t value;
-
-		r = get_user(value, (uint32_t *)buf);
-		if (r)
-			return r;
+	swds = pci_upstream_bridge(adev->pdev);
+	if (!swds || swds->vendor != PCI_VENDOR_ID_ATI ||
+	    pci_pcie_type(swds) != PCI_EXP_TYPE_DOWNSTREAM)
+		return;
+	swus = pci_upstream_bridge(swds);
+	if (!swus ||
+	    (swus->vendor != PCI_VENDOR_ID_ATI &&
+	     swus->vendor != PCI_VENDOR_ID_AMD) ||
+	    pci_pcie_type(swus) != PCI_EXP_TYPE_UPSTREAM)
+		return;
 
-		WREG32_PCIE(*pos >> 2, value);
+	/* If already saved, return */
+	if (adev->pcie_reset_ctx.swus)
+		return;
+	/* Upstream bridge is ATI, assume it's SWUS/DS architecture */
+	r = pci_save_state(swds);
+	if (r)
+		return;
+	adev->pcie_reset_ctx.swds_pcistate = pci_store_saved_state(swds);
 
-		result += 4;
-		buf += 4;
-		*pos += 4;
-		size -= 4;
-	}
+	r = pci_save_state(swus);
+	if (r)
+		return;
+	adev->pcie_reset_ctx.swus_pcistate = pci_store_saved_state(swus);
 
-	return result;
+	adev->pcie_reset_ctx.swus = swus;
 }
 
-static ssize_t amdgpu_debugfs_regs_didt_read(struct file *f, char __user *buf,
-					size_t size, loff_t *pos)
+static void amdgpu_device_load_switch_state(struct amdgpu_device *adev)
 {
-	struct amdgpu_device *adev = file_inode(f)->i_private;
-	ssize_t result = 0;
+	struct pci_dev *pdev;
 	int r;
 
-	if (size & 0x3 || *pos & 0x3)
-		return -EINVAL;
-
-	while (size) {
-		uint32_t value;
-
-		value = RREG32_DIDT(*pos >> 2);
-		r = put_user(value, (uint32_t *)buf);
-		if (r)
-			return r;
+	if (!adev->pcie_reset_ctx.swds_pcistate ||
+	    !adev->pcie_reset_ctx.swus_pcistate)
+		return;
 
-		result += 4;
-		buf += 4;
-		*pos += 4;
-		size -= 4;
+	pdev = adev->pcie_reset_ctx.swus;
+	r = pci_load_saved_state(pdev, adev->pcie_reset_ctx.swus_pcistate);
+	if (!r) {
+		pci_restore_state(pdev);
+	} else {
+		dev_warn(adev->dev, "Failed to load SWUS state, err:%d\n", r);
+		return;
 	}
 
-	return result;
+	pdev = pci_upstream_bridge(adev->pdev);
+	r = pci_load_saved_state(pdev, adev->pcie_reset_ctx.swds_pcistate);
+	if (!r)
+		pci_restore_state(pdev);
+	else
+		dev_warn(adev->dev, "Failed to load SWDS state, err:%d\n", r);
 }
 
-static ssize_t amdgpu_debugfs_regs_didt_write(struct file *f, const char __user *buf,
-					 size_t size, loff_t *pos)
+bool amdgpu_device_cache_pci_state(struct pci_dev *pdev)
 {
-	struct amdgpu_device *adev = file_inode(f)->i_private;
-	ssize_t result = 0;
+	struct drm_device *dev = pci_get_drvdata(pdev);
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	int r;
 
-	if (size & 0x3 || *pos & 0x3)
-		return -EINVAL;
-
-	while (size) {
-		uint32_t value;
+	if (amdgpu_sriov_vf(adev))
+		return false;
 
-		r = get_user(value, (uint32_t *)buf);
-		if (r)
-			return r;
+	r = pci_save_state(pdev);
+	if (!r) {
+		kfree(adev->pci_state);
 
-		WREG32_DIDT(*pos >> 2, value);
+		adev->pci_state = pci_store_saved_state(pdev);
 
-		result += 4;
-		buf += 4;
-		*pos += 4;
-		size -= 4;
+		if (!adev->pci_state) {
+			dev_err(adev->dev, "Failed to store PCI saved state");
+			return false;
+		}
+	} else {
+		dev_warn(adev->dev, "Failed to save PCI state, err:%d\n", r);
+		return false;
 	}
 
-	return result;
+	amdgpu_device_cache_switch_state(adev);
+
+	return true;
 }
 
-static ssize_t amdgpu_debugfs_regs_smc_read(struct file *f, char __user *buf,
-					size_t size, loff_t *pos)
+bool amdgpu_device_load_pci_state(struct pci_dev *pdev)
 {
-	struct amdgpu_device *adev = file_inode(f)->i_private;
-	ssize_t result = 0;
+	struct drm_device *dev = pci_get_drvdata(pdev);
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	int r;
 
-	if (size & 0x3 || *pos & 0x3)
-		return -EINVAL;
-
-	while (size) {
-		uint32_t value;
+	if (!adev->pci_state)
+		return false;
 
-		value = RREG32_SMC(*pos);
-		r = put_user(value, (uint32_t *)buf);
-		if (r)
-			return r;
+	r = pci_load_saved_state(pdev, adev->pci_state);
 
-		result += 4;
-		buf += 4;
-		*pos += 4;
-		size -= 4;
+	if (!r) {
+		pci_restore_state(pdev);
+	} else {
+		dev_warn(adev->dev, "Failed to load PCI state, err:%d\n", r);
+		return false;
 	}
 
-	return result;
+	return true;
 }
 
-static ssize_t amdgpu_debugfs_regs_smc_write(struct file *f, const char __user *buf,
-					 size_t size, loff_t *pos)
+void amdgpu_device_flush_hdp(struct amdgpu_device *adev,
+		struct amdgpu_ring *ring)
 {
-	struct amdgpu_device *adev = file_inode(f)->i_private;
-	ssize_t result = 0;
-	int r;
-
-	if (size & 0x3 || *pos & 0x3)
-		return -EINVAL;
-
-	while (size) {
-		uint32_t value;
+#ifdef CONFIG_X86_64
+	if ((adev->flags & AMD_IS_APU) && !amdgpu_passthrough(adev))
+		return;
+#endif
+	if (adev->gmc.xgmi.connected_to_cpu)
+		return;
 
-		r = get_user(value, (uint32_t *)buf);
-		if (r)
-			return r;
+	if (ring && ring->funcs->emit_hdp_flush)
+		amdgpu_ring_emit_hdp_flush(ring);
+	else
+		amdgpu_asic_flush_hdp(adev, ring);
+}
 
-		WREG32_SMC(*pos, value);
+void amdgpu_device_invalidate_hdp(struct amdgpu_device *adev,
+		struct amdgpu_ring *ring)
+{
+#ifdef CONFIG_X86_64
+	if ((adev->flags & AMD_IS_APU) && !amdgpu_passthrough(adev))
+		return;
+#endif
+	if (adev->gmc.xgmi.connected_to_cpu)
+		return;
 
-		result += 4;
-		buf += 4;
-		*pos += 4;
-		size -= 4;
-	}
+	amdgpu_asic_invalidate_hdp(adev, ring);
+}
 
-	return result;
+int amdgpu_in_reset(struct amdgpu_device *adev)
+{
+	return atomic_read(&adev->reset_domain->in_gpu_reset);
 }
 
-static ssize_t amdgpu_debugfs_gca_config_read(struct file *f, char __user *buf,
-					size_t size, loff_t *pos)
+/**
+ * amdgpu_device_halt() - bring hardware to some kind of halt state
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Bring hardware to some kind of halt state so that no one can touch it
+ * any more. It will help to maintain error context when error occurred.
+ * Compare to a simple hang, the system will keep stable at least for SSH
+ * access. Then it should be trivial to inspect the hardware state and
+ * see what's going on. Implemented as following:
+ *
+ * 1. drm_dev_unplug() makes device inaccessible to user space(IOCTLs, etc),
+ *    clears all CPU mappings to device, disallows remappings through page faults
+ * 2. amdgpu_irq_disable_all() disables all interrupts
+ * 3. amdgpu_fence_driver_hw_fini() signals all HW fences
+ * 4. set adev->no_hw_access to avoid potential crashes after setp 5
+ * 5. amdgpu_device_unmap_mmio() clears all MMIO mappings
+ * 6. pci_disable_device() and pci_wait_for_pending_transaction()
+ *    flush any in flight DMA operations
+ */
+void amdgpu_device_halt(struct amdgpu_device *adev)
 {
-	struct amdgpu_device *adev = file_inode(f)->i_private;
-	ssize_t result = 0;
-	int r;
-	uint32_t *config, no_regs = 0;
+	struct pci_dev *pdev = adev->pdev;
+	struct drm_device *ddev = adev_to_drm(adev);
 
-	if (size & 0x3 || *pos & 0x3)
-		return -EINVAL;
+	amdgpu_xcp_dev_unplug(adev);
+	drm_dev_unplug(ddev);
 
-	config = kmalloc_array(256, sizeof(*config), GFP_KERNEL);
-	if (!config)
-		return -ENOMEM;
+	amdgpu_irq_disable_all(adev);
 
-	/* version, increment each time something is added */
-	config[no_regs++] = 3;
-	config[no_regs++] = adev->gfx.config.max_shader_engines;
-	config[no_regs++] = adev->gfx.config.max_tile_pipes;
-	config[no_regs++] = adev->gfx.config.max_cu_per_sh;
-	config[no_regs++] = adev->gfx.config.max_sh_per_se;
-	config[no_regs++] = adev->gfx.config.max_backends_per_se;
-	config[no_regs++] = adev->gfx.config.max_texture_channel_caches;
-	config[no_regs++] = adev->gfx.config.max_gprs;
-	config[no_regs++] = adev->gfx.config.max_gs_threads;
-	config[no_regs++] = adev->gfx.config.max_hw_contexts;
-	config[no_regs++] = adev->gfx.config.sc_prim_fifo_size_frontend;
-	config[no_regs++] = adev->gfx.config.sc_prim_fifo_size_backend;
-	config[no_regs++] = adev->gfx.config.sc_hiz_tile_fifo_size;
-	config[no_regs++] = adev->gfx.config.sc_earlyz_tile_fifo_size;
-	config[no_regs++] = adev->gfx.config.num_tile_pipes;
-	config[no_regs++] = adev->gfx.config.backend_enable_mask;
-	config[no_regs++] = adev->gfx.config.mem_max_burst_length_bytes;
-	config[no_regs++] = adev->gfx.config.mem_row_size_in_kb;
-	config[no_regs++] = adev->gfx.config.shader_engine_tile_size;
-	config[no_regs++] = adev->gfx.config.num_gpus;
-	config[no_regs++] = adev->gfx.config.multi_gpu_tile_size;
-	config[no_regs++] = adev->gfx.config.mc_arb_ramcfg;
-	config[no_regs++] = adev->gfx.config.gb_addr_config;
-	config[no_regs++] = adev->gfx.config.num_rbs;
-
-	/* rev==1 */
-	config[no_regs++] = adev->rev_id;
-	config[no_regs++] = adev->pg_flags;
-	config[no_regs++] = adev->cg_flags;
-
-	/* rev==2 */
-	config[no_regs++] = adev->family;
-	config[no_regs++] = adev->external_rev_id;
-
-	/* rev==3 */
-	config[no_regs++] = adev->pdev->device;
-	config[no_regs++] = adev->pdev->revision;
-	config[no_regs++] = adev->pdev->subsystem_device;
-	config[no_regs++] = adev->pdev->subsystem_vendor;
-
-	while (size && (*pos < no_regs * 4)) {
-		uint32_t value;
-
-		value = config[*pos >> 2];
-		r = put_user(value, (uint32_t *)buf);
-		if (r) {
-			kfree(config);
-			return r;
-		}
+	amdgpu_fence_driver_hw_fini(adev);
 
-		result += 4;
-		buf += 4;
-		*pos += 4;
-		size -= 4;
-	}
+	adev->no_hw_access = true;
+
+	amdgpu_device_unmap_mmio(adev);
 
-	kfree(config);
-	return result;
+	pci_disable_device(pdev);
+	pci_wait_for_pending_transaction(pdev);
 }
 
-static ssize_t amdgpu_debugfs_sensor_read(struct file *f, char __user *buf,
-					size_t size, loff_t *pos)
+u32 amdgpu_device_pcie_port_rreg(struct amdgpu_device *adev,
+				u32 reg)
 {
-	struct amdgpu_device *adev = file_inode(f)->i_private;
-	int idx, x, outsize, r, valuesize;
-	uint32_t values[16];
+	unsigned long flags, address, data;
+	u32 r;
 
-	if (size & 3 || *pos & 0x3)
-		return -EINVAL;
-
-	if (amdgpu_dpm == 0)
-		return -EINVAL;
+	address = adev->nbio.funcs->get_pcie_port_index_offset(adev);
+	data = adev->nbio.funcs->get_pcie_port_data_offset(adev);
 
-	/* convert offset to sensor number */
-	idx = *pos >> 2;
-
-	valuesize = sizeof(values);
-	if (adev->powerplay.pp_funcs && adev->powerplay.pp_funcs->read_sensor)
-		r = adev->powerplay.pp_funcs->read_sensor(adev->powerplay.pp_handle, idx, &values[0], &valuesize);
-	else if (adev->pm.funcs && adev->pm.funcs->read_sensor)
-		r = adev->pm.funcs->read_sensor(adev, idx, &values[0],
-						&valuesize);
-	else
-		return -EINVAL;
-
-	if (size > valuesize)
-		return -EINVAL;
-
-	outsize = 0;
-	x = 0;
-	if (!r) {
-		while (size) {
-			r = put_user(values[x++], (int32_t *)buf);
-			buf += 4;
-			size -= 4;
-			outsize += 4;
-		}
-	}
-
-	return !r ? outsize : r;
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(address, reg * 4);
+	(void)RREG32(address);
+	r = RREG32(data);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+	return r;
 }
 
-static ssize_t amdgpu_debugfs_wave_read(struct file *f, char __user *buf,
-					size_t size, loff_t *pos)
+void amdgpu_device_pcie_port_wreg(struct amdgpu_device *adev,
+				u32 reg, u32 v)
 {
-	struct amdgpu_device *adev = f->f_inode->i_private;
-	int r, x;
-	ssize_t result=0;
-	uint32_t offset, se, sh, cu, wave, simd, data[32];
+	unsigned long flags, address, data;
 
-	if (size & 3 || *pos & 3)
-		return -EINVAL;
-
-	/* decode offset */
-	offset = (*pos & 0x7F);
-	se = ((*pos >> 7) & 0xFF);
-	sh = ((*pos >> 15) & 0xFF);
-	cu = ((*pos >> 23) & 0xFF);
-	wave = ((*pos >> 31) & 0xFF);
-	simd = ((*pos >> 37) & 0xFF);
+	address = adev->nbio.funcs->get_pcie_port_index_offset(adev);
+	data = adev->nbio.funcs->get_pcie_port_data_offset(adev);
 
-	/* switch to the specific se/sh/cu */
-	mutex_lock(&adev->grbm_idx_mutex);
-	amdgpu_gfx_select_se_sh(adev, se, sh, cu);
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(address, reg * 4);
+	(void)RREG32(address);
+	WREG32(data, v);
+	(void)RREG32(data);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+}
 
-	x = 0;
-	if (adev->gfx.funcs->read_wave_data)
-		adev->gfx.funcs->read_wave_data(adev, simd, wave, data, &x);
+/**
+ * amdgpu_device_get_gang - return a reference to the current gang
+ * @adev: amdgpu_device pointer
+ *
+ * Returns: A new reference to the current gang leader.
+ */
+struct dma_fence *amdgpu_device_get_gang(struct amdgpu_device *adev)
+{
+	struct dma_fence *fence;
 
-	amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
-	mutex_unlock(&adev->grbm_idx_mutex);
+	rcu_read_lock();
+	fence = dma_fence_get_rcu_safe(&adev->gang_submit);
+	rcu_read_unlock();
+	return fence;
+}
 
-	if (!x)
-		return -EINVAL;
+/**
+ * amdgpu_device_switch_gang - switch to a new gang
+ * @adev: amdgpu_device pointer
+ * @gang: the gang to switch to
+ *
+ * Try to switch to a new gang.
+ * Returns: NULL if we switched to the new gang or a reference to the current
+ * gang leader.
+ */
+struct dma_fence *amdgpu_device_switch_gang(struct amdgpu_device *adev,
+					    struct dma_fence *gang)
+{
+	struct dma_fence *old = NULL;
 
-	while (size && (offset < x * 4)) {
-		uint32_t value;
+	dma_fence_get(gang);
+	do {
+		dma_fence_put(old);
+		old = amdgpu_device_get_gang(adev);
+		if (old == gang)
+			break;
 
-		value = data[offset >> 2];
-		r = put_user(value, (uint32_t *)buf);
-		if (r)
-			return r;
+		if (!dma_fence_is_signaled(old)) {
+			dma_fence_put(gang);
+			return old;
+		}
 
-		result += 4;
-		buf += 4;
-		offset += 4;
-		size -= 4;
-	}
+	} while (cmpxchg((struct dma_fence __force **)&adev->gang_submit,
+			 old, gang) != old);
 
-	return result;
+	/*
+	 * Drop it once for the exchanged reference in adev and once for the
+	 * thread local reference acquired in amdgpu_device_get_gang().
+	 */
+	dma_fence_put(old);
+	dma_fence_put(old);
+	return NULL;
 }
 
-static ssize_t amdgpu_debugfs_gpr_read(struct file *f, char __user *buf,
-					size_t size, loff_t *pos)
+/**
+ * amdgpu_device_enforce_isolation - enforce HW isolation
+ * @adev: the amdgpu device pointer
+ * @ring: the HW ring the job is supposed to run on
+ * @job: the job which is about to be pushed to the HW ring
+ *
+ * Makes sure that only one client at a time can use the GFX block.
+ * Returns: The dependency to wait on before the job can be pushed to the HW.
+ * The function is called multiple times until NULL is returned.
+ */
+struct dma_fence *amdgpu_device_enforce_isolation(struct amdgpu_device *adev,
+						  struct amdgpu_ring *ring,
+						  struct amdgpu_job *job)
 {
-	struct amdgpu_device *adev = f->f_inode->i_private;
+	struct amdgpu_isolation *isolation = &adev->isolation[ring->xcp_id];
+	struct drm_sched_fence *f = job->base.s_fence;
+	struct dma_fence *dep;
+	void *owner;
 	int r;
-	ssize_t result = 0;
-	uint32_t offset, se, sh, cu, wave, simd, thread, bank, *data;
 
-	if (size & 3 || *pos & 3)
-		return -EINVAL;
+	/*
+	 * For now enforce isolation only for the GFX block since we only need
+	 * the cleaner shader on those rings.
+	 */
+	if (ring->funcs->type != AMDGPU_RING_TYPE_GFX &&
+	    ring->funcs->type != AMDGPU_RING_TYPE_COMPUTE)
+		return NULL;
 
-	/* decode offset */
-	offset = (*pos & 0xFFF);       /* in dwords */
-	se = ((*pos >> 12) & 0xFF);
-	sh = ((*pos >> 20) & 0xFF);
-	cu = ((*pos >> 28) & 0xFF);
-	wave = ((*pos >> 36) & 0xFF);
-	simd = ((*pos >> 44) & 0xFF);
-	thread = ((*pos >> 52) & 0xFF);
-	bank = ((*pos >> 60) & 1);
-
-	data = kmalloc_array(1024, sizeof(*data), GFP_KERNEL);
-	if (!data)
-		return -ENOMEM;
+	/*
+	 * All submissions where enforce isolation is false are handled as if
+	 * they come from a single client. Use ~0l as the owner to distinct it
+	 * from kernel submissions where the owner is NULL.
+	 */
+	owner = job->enforce_isolation ? f->owner : (void *)~0l;
 
-	/* switch to the specific se/sh/cu */
-	mutex_lock(&adev->grbm_idx_mutex);
-	amdgpu_gfx_select_se_sh(adev, se, sh, cu);
+	mutex_lock(&adev->enforce_isolation_mutex);
 
-	if (bank == 0) {
-		if (adev->gfx.funcs->read_wave_vgprs)
-			adev->gfx.funcs->read_wave_vgprs(adev, simd, wave, thread, offset, size>>2, data);
-	} else {
-		if (adev->gfx.funcs->read_wave_sgprs)
-			adev->gfx.funcs->read_wave_sgprs(adev, simd, wave, offset, size>>2, data);
+	/*
+	 * The "spearhead" submission is the first one which changes the
+	 * ownership to its client. We always need to wait for it to be
+	 * pushed to the HW before proceeding with anything.
+	 */
+	if (&f->scheduled != isolation->spearhead &&
+	    !dma_fence_is_signaled(isolation->spearhead)) {
+		dep = isolation->spearhead;
+		goto out_grab_ref;
 	}
 
-	amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
-	mutex_unlock(&adev->grbm_idx_mutex);
-
-	while (size) {
-		uint32_t value;
+	if (isolation->owner != owner) {
 
-		value = data[offset++];
-		r = put_user(value, (uint32_t *)buf);
-		if (r) {
-			result = r;
-			goto err;
+		/*
+		 * Wait for any gang to be assembled before switching to a
+		 * different owner or otherwise we could deadlock the
+		 * submissions.
+		 */
+		if (!job->gang_submit) {
+			dep = amdgpu_device_get_gang(adev);
+			if (!dma_fence_is_signaled(dep))
+				goto out_return_dep;
+			dma_fence_put(dep);
 		}
 
-		result += 4;
-		buf += 4;
-		size -= 4;
+		dma_fence_put(isolation->spearhead);
+		isolation->spearhead = dma_fence_get(&f->scheduled);
+		amdgpu_sync_move(&isolation->active, &isolation->prev);
+		trace_amdgpu_isolation(isolation->owner, owner);
+		isolation->owner = owner;
 	}
 
-err:
-	kfree(data);
-	return result;
+	/*
+	 * Specifying the ring here helps to pipeline submissions even when
+	 * isolation is enabled. If that is not desired for testing NULL can be
+	 * used instead of the ring to enforce a CPU round trip while switching
+	 * between clients.
+	 */
+	dep = amdgpu_sync_peek_fence(&isolation->prev, ring);
+	r = amdgpu_sync_fence(&isolation->active, &f->finished, GFP_NOWAIT);
+	if (r)
+		dev_warn(adev->dev, "OOM tracking isolation\n");
+
+out_grab_ref:
+	dma_fence_get(dep);
+out_return_dep:
+	mutex_unlock(&adev->enforce_isolation_mutex);
+	return dep;
 }
 
-static const struct file_operations amdgpu_debugfs_regs_fops = {
-	.owner = THIS_MODULE,
-	.read = amdgpu_debugfs_regs_read,
-	.write = amdgpu_debugfs_regs_write,
-	.llseek = default_llseek
-};
-static const struct file_operations amdgpu_debugfs_regs_didt_fops = {
-	.owner = THIS_MODULE,
-	.read = amdgpu_debugfs_regs_didt_read,
-	.write = amdgpu_debugfs_regs_didt_write,
-	.llseek = default_llseek
-};
-static const struct file_operations amdgpu_debugfs_regs_pcie_fops = {
-	.owner = THIS_MODULE,
-	.read = amdgpu_debugfs_regs_pcie_read,
-	.write = amdgpu_debugfs_regs_pcie_write,
-	.llseek = default_llseek
-};
-static const struct file_operations amdgpu_debugfs_regs_smc_fops = {
-	.owner = THIS_MODULE,
-	.read = amdgpu_debugfs_regs_smc_read,
-	.write = amdgpu_debugfs_regs_smc_write,
-	.llseek = default_llseek
-};
+bool amdgpu_device_has_display_hardware(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+#ifdef CONFIG_DRM_AMDGPU_SI
+	case CHIP_HAINAN:
+#endif
+	case CHIP_TOPAZ:
+		/* chips with no display hardware */
+		return false;
+#ifdef CONFIG_DRM_AMDGPU_SI
+	case CHIP_TAHITI:
+	case CHIP_PITCAIRN:
+	case CHIP_VERDE:
+	case CHIP_OLAND:
+#endif
+#ifdef CONFIG_DRM_AMDGPU_CIK
+	case CHIP_BONAIRE:
+	case CHIP_HAWAII:
+	case CHIP_KAVERI:
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+#endif
+	case CHIP_TONGA:
+	case CHIP_FIJI:
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+		/* chips with display hardware */
+		return true;
+	default:
+		/* IP discovery */
+		if (!amdgpu_ip_version(adev, DCE_HWIP, 0) ||
+		    (adev->harvest_ip_mask & AMD_HARVEST_IP_DMU_MASK))
+			return false;
+		return true;
+	}
+}
 
-static const struct file_operations amdgpu_debugfs_gca_config_fops = {
-	.owner = THIS_MODULE,
-	.read = amdgpu_debugfs_gca_config_read,
-	.llseek = default_llseek
-};
+uint32_t amdgpu_device_wait_on_rreg(struct amdgpu_device *adev,
+		uint32_t inst, uint32_t reg_addr, char reg_name[],
+		uint32_t expected_value, uint32_t mask)
+{
+	uint32_t ret = 0;
+	uint32_t old_ = 0;
+	uint32_t tmp_ = RREG32(reg_addr);
+	uint32_t loop = adev->usec_timeout;
+
+	while ((tmp_ & (mask)) != (expected_value)) {
+		if (old_ != tmp_) {
+			loop = adev->usec_timeout;
+			old_ = tmp_;
+		} else
+			udelay(1);
+		tmp_ = RREG32(reg_addr);
+		loop--;
+		if (!loop) {
+			dev_warn(
+				adev->dev,
+				"Register(%d) [%s] failed to reach value 0x%08x != 0x%08xn",
+				inst, reg_name, (uint32_t)expected_value,
+				(uint32_t)(tmp_ & (mask)));
+			ret = -ETIMEDOUT;
+			break;
+		}
+	}
+	return ret;
+}
 
-static const struct file_operations amdgpu_debugfs_sensors_fops = {
-	.owner = THIS_MODULE,
-	.read = amdgpu_debugfs_sensor_read,
-	.llseek = default_llseek
-};
+ssize_t amdgpu_get_soft_full_reset_mask(struct amdgpu_ring *ring)
+{
+	ssize_t size = 0;
 
-static const struct file_operations amdgpu_debugfs_wave_fops = {
-	.owner = THIS_MODULE,
-	.read = amdgpu_debugfs_wave_read,
-	.llseek = default_llseek
-};
-static const struct file_operations amdgpu_debugfs_gpr_fops = {
-	.owner = THIS_MODULE,
-	.read = amdgpu_debugfs_gpr_read,
-	.llseek = default_llseek
-};
+	if (!ring || !ring->adev)
+		return size;
 
-static const struct file_operations *debugfs_regs[] = {
-	&amdgpu_debugfs_regs_fops,
-	&amdgpu_debugfs_regs_didt_fops,
-	&amdgpu_debugfs_regs_pcie_fops,
-	&amdgpu_debugfs_regs_smc_fops,
-	&amdgpu_debugfs_gca_config_fops,
-	&amdgpu_debugfs_sensors_fops,
-	&amdgpu_debugfs_wave_fops,
-	&amdgpu_debugfs_gpr_fops,
-};
+	if (amdgpu_device_should_recover_gpu(ring->adev))
+		size |= AMDGPU_RESET_TYPE_FULL;
 
-static const char *debugfs_regs_names[] = {
-	"amdgpu_regs",
-	"amdgpu_regs_didt",
-	"amdgpu_regs_pcie",
-	"amdgpu_regs_smc",
-	"amdgpu_gca_config",
-	"amdgpu_sensors",
-	"amdgpu_wave",
-	"amdgpu_gpr",
-};
+	if (unlikely(!ring->adev->debug_disable_soft_recovery) &&
+	    !amdgpu_sriov_vf(ring->adev) && ring->funcs->soft_recovery)
+		size |= AMDGPU_RESET_TYPE_SOFT_RESET;
 
-static int amdgpu_debugfs_regs_init(struct amdgpu_device *adev)
+	return size;
+}
+
+ssize_t amdgpu_show_reset_mask(char *buf, uint32_t supported_reset)
 {
-	struct drm_minor *minor = adev->ddev->primary;
-	struct dentry *ent, *root = minor->debugfs_root;
-	unsigned i, j;
+	ssize_t size = 0;
 
-	for (i = 0; i < ARRAY_SIZE(debugfs_regs); i++) {
-		ent = debugfs_create_file(debugfs_regs_names[i],
-					  S_IFREG | S_IRUGO, root,
-					  adev, debugfs_regs[i]);
-		if (IS_ERR(ent)) {
-			for (j = 0; j < i; j++) {
-				debugfs_remove(adev->debugfs_regs[i]);
-				adev->debugfs_regs[i] = NULL;
-			}
-			return PTR_ERR(ent);
-		}
+	if (supported_reset == 0) {
+		size += sysfs_emit_at(buf, size, "unsupported");
+		size += sysfs_emit_at(buf, size, "\n");
+		return size;
 
-		if (!i)
-			i_size_write(ent->d_inode, adev->rmmio_size);
-		adev->debugfs_regs[i] = ent;
 	}
 
-	return 0;
-}
+	if (supported_reset & AMDGPU_RESET_TYPE_SOFT_RESET)
+		size += sysfs_emit_at(buf, size, "soft ");
 
-static void amdgpu_debugfs_regs_cleanup(struct amdgpu_device *adev)
-{
-	unsigned i;
+	if (supported_reset & AMDGPU_RESET_TYPE_PER_QUEUE)
+		size += sysfs_emit_at(buf, size, "queue ");
 
-	for (i = 0; i < ARRAY_SIZE(debugfs_regs); i++) {
-		if (adev->debugfs_regs[i]) {
-			debugfs_remove(adev->debugfs_regs[i]);
-			adev->debugfs_regs[i] = NULL;
-		}
-	}
+	if (supported_reset & AMDGPU_RESET_TYPE_PER_PIPE)
+		size += sysfs_emit_at(buf, size, "pipe ");
+
+	if (supported_reset & AMDGPU_RESET_TYPE_FULL)
+		size += sysfs_emit_at(buf, size, "full ");
+
+	size += sysfs_emit_at(buf, size, "\n");
+	return size;
 }
 
-static int amdgpu_debugfs_test_ib(struct seq_file *m, void *data)
+void amdgpu_device_set_uid(struct amdgpu_uid *uid_info,
+			   enum amdgpu_uid_type type, uint8_t inst,
+			   uint64_t uid)
 {
-	struct drm_info_node *node = (struct drm_info_node *) m->private;
-	struct drm_device *dev = node->minor->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-	int r = 0, i;
-
-	/* hold on the scheduler */
-	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
-		struct amdgpu_ring *ring = adev->rings[i];
+	if (!uid_info)
+		return;
 
-		if (!ring || !ring->sched.thread)
-			continue;
-		kthread_park(ring->sched.thread);
+	if (type >= AMDGPU_UID_TYPE_MAX) {
+		dev_err_once(uid_info->adev->dev, "Invalid UID type %d\n",
+			     type);
+		return;
 	}
 
-	seq_printf(m, "run ib test:\n");
-	r = amdgpu_ib_ring_tests(adev);
-	if (r)
-		seq_printf(m, "ib ring tests failed (%d).\n", r);
-	else
-		seq_printf(m, "ib ring tests passed.\n");
-
-	/* go on the scheduler */
-	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
-		struct amdgpu_ring *ring = adev->rings[i];
+	if (inst >= AMDGPU_UID_INST_MAX) {
+		dev_err_once(uid_info->adev->dev, "Invalid UID instance %d\n",
+			     inst);
+		return;
+	}
 
-		if (!ring || !ring->sched.thread)
-			continue;
-		kthread_unpark(ring->sched.thread);
+	if (uid_info->uid[type][inst] != 0) {
+		dev_warn_once(
+			uid_info->adev->dev,
+			"Overwriting existing UID %llu for type %d instance %d\n",
+			uid_info->uid[type][inst], type, inst);
 	}
 
-	return 0;
+	uid_info->uid[type][inst] = uid;
 }
 
-static const struct drm_info_list amdgpu_debugfs_test_ib_ring_list[] = {
-	{"amdgpu_test_ib", &amdgpu_debugfs_test_ib}
-};
-
-static int amdgpu_debugfs_test_ib_ring_init(struct amdgpu_device *adev)
+u64 amdgpu_device_get_uid(struct amdgpu_uid *uid_info,
+			  enum amdgpu_uid_type type, uint8_t inst)
 {
-	return amdgpu_debugfs_add_files(adev,
-					amdgpu_debugfs_test_ib_ring_list, 1);
-}
+	if (!uid_info)
+		return 0;
 
-int amdgpu_debugfs_init(struct drm_minor *minor)
-{
-	return 0;
-}
-#else
-static int amdgpu_debugfs_test_ib_ring_init(struct amdgpu_device *adev)
-{
-	return 0;
-}
-static int amdgpu_debugfs_regs_init(struct amdgpu_device *adev)
-{
-	return 0;
+	if (type >= AMDGPU_UID_TYPE_MAX) {
+		dev_err_once(uid_info->adev->dev, "Invalid UID type %d\n",
+			     type);
+		return 0;
+	}
+
+	if (inst >= AMDGPU_UID_INST_MAX) {
+		dev_err_once(uid_info->adev->dev, "Invalid UID instance %d\n",
+			     inst);
+		return 0;
+	}
+
+	return uid_info->uid[type][inst];
 }
-static void amdgpu_debugfs_regs_cleanup(struct amdgpu_device *adev) { }
-#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_df.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_df.h
new file mode 100644
index 000000000000..eb605e79ae0e
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_df.h
@@ -0,0 +1,64 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_DF_H__
+#define __AMDGPU_DF_H__
+
+struct amdgpu_df_hash_status {
+	bool hash_64k;
+	bool hash_2m;
+	bool hash_1g;
+};
+
+struct amdgpu_df_funcs {
+	void (*sw_init)(struct amdgpu_device *adev);
+	void (*sw_fini)(struct amdgpu_device *adev);
+	void (*hw_init)(struct amdgpu_device *adev);
+	void (*enable_broadcast_mode)(struct amdgpu_device *adev,
+				      bool enable);
+	u32 (*get_fb_channel_number)(struct amdgpu_device *adev);
+	u32 (*get_hbm_channel_number)(struct amdgpu_device *adev);
+	void (*update_medium_grain_clock_gating)(struct amdgpu_device *adev,
+						 bool enable);
+	void (*get_clockgating_state)(struct amdgpu_device *adev,
+				      u64 *flags);
+	void (*enable_ecc_force_par_wr_rmw)(struct amdgpu_device *adev,
+					    bool enable);
+	int (*pmc_start)(struct amdgpu_device *adev, uint64_t config,
+					 int counter_idx, int is_add);
+	int (*pmc_stop)(struct amdgpu_device *adev, uint64_t config,
+					 int counter_idx, int is_remove);
+	void (*pmc_get_count)(struct amdgpu_device *adev, uint64_t config,
+					 int counter_idx, uint64_t *count);
+	uint64_t (*get_fica)(struct amdgpu_device *adev, uint32_t ficaa_val);
+	void (*set_fica)(struct amdgpu_device *adev, uint32_t ficaa_val,
+			 uint32_t ficadl_val, uint32_t ficadh_val);
+	bool (*query_ras_poison_mode)(struct amdgpu_device *adev);
+};
+
+struct amdgpu_df {
+	struct amdgpu_df_hash_status	hash_status;
+	const struct amdgpu_df_funcs	*funcs;
+};
+
+#endif /* __AMDGPU_DF_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
new file mode 100644
index 000000000000..73401f0aeb34
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
@@ -0,0 +1,3157 @@
+/*
+ * Copyright 2018-2024 Advanced Micro Devices, Inc. All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+
+#include "amdgpu.h"
+#include "amdgpu_discovery.h"
+#include "soc15_hw_ip.h"
+#include "discovery.h"
+#include "amdgpu_ras.h"
+
+#include "soc15.h"
+#include "gfx_v9_0.h"
+#include "gfx_v9_4_3.h"
+#include "gmc_v9_0.h"
+#include "df_v1_7.h"
+#include "df_v3_6.h"
+#include "df_v4_3.h"
+#include "df_v4_6_2.h"
+#include "df_v4_15.h"
+#include "nbio_v6_1.h"
+#include "nbio_v7_0.h"
+#include "nbio_v7_4.h"
+#include "nbio_v7_9.h"
+#include "nbio_v7_11.h"
+#include "hdp_v4_0.h"
+#include "vega10_ih.h"
+#include "vega20_ih.h"
+#include "sdma_v4_0.h"
+#include "sdma_v4_4_2.h"
+#include "uvd_v7_0.h"
+#include "vce_v4_0.h"
+#include "vcn_v1_0.h"
+#include "vcn_v2_5.h"
+#include "jpeg_v2_5.h"
+#include "smuio_v9_0.h"
+#include "gmc_v10_0.h"
+#include "gmc_v11_0.h"
+#include "gmc_v12_0.h"
+#include "gfxhub_v2_0.h"
+#include "mmhub_v2_0.h"
+#include "nbio_v2_3.h"
+#include "nbio_v4_3.h"
+#include "nbio_v7_2.h"
+#include "nbio_v7_7.h"
+#include "nbif_v6_3_1.h"
+#include "hdp_v5_0.h"
+#include "hdp_v5_2.h"
+#include "hdp_v6_0.h"
+#include "hdp_v7_0.h"
+#include "nv.h"
+#include "soc21.h"
+#include "soc24.h"
+#include "navi10_ih.h"
+#include "ih_v6_0.h"
+#include "ih_v6_1.h"
+#include "ih_v7_0.h"
+#include "gfx_v10_0.h"
+#include "gfx_v11_0.h"
+#include "gfx_v12_0.h"
+#include "sdma_v5_0.h"
+#include "sdma_v5_2.h"
+#include "sdma_v6_0.h"
+#include "sdma_v7_0.h"
+#include "lsdma_v6_0.h"
+#include "lsdma_v7_0.h"
+#include "vcn_v2_0.h"
+#include "jpeg_v2_0.h"
+#include "vcn_v3_0.h"
+#include "jpeg_v3_0.h"
+#include "vcn_v4_0.h"
+#include "jpeg_v4_0.h"
+#include "vcn_v4_0_3.h"
+#include "jpeg_v4_0_3.h"
+#include "vcn_v4_0_5.h"
+#include "jpeg_v4_0_5.h"
+#include "amdgpu_vkms.h"
+#include "mes_v11_0.h"
+#include "mes_v12_0.h"
+#include "smuio_v11_0.h"
+#include "smuio_v11_0_6.h"
+#include "smuio_v13_0.h"
+#include "smuio_v13_0_3.h"
+#include "smuio_v13_0_6.h"
+#include "smuio_v14_0_2.h"
+#include "vcn_v5_0_0.h"
+#include "vcn_v5_0_1.h"
+#include "jpeg_v5_0_0.h"
+#include "jpeg_v5_0_1.h"
+
+#include "amdgpu_vpe.h"
+#if defined(CONFIG_DRM_AMD_ISP)
+#include "amdgpu_isp.h"
+#endif
+
+MODULE_FIRMWARE("amdgpu/ip_discovery.bin");
+MODULE_FIRMWARE("amdgpu/vega10_ip_discovery.bin");
+MODULE_FIRMWARE("amdgpu/vega12_ip_discovery.bin");
+MODULE_FIRMWARE("amdgpu/vega20_ip_discovery.bin");
+MODULE_FIRMWARE("amdgpu/raven_ip_discovery.bin");
+MODULE_FIRMWARE("amdgpu/raven2_ip_discovery.bin");
+MODULE_FIRMWARE("amdgpu/picasso_ip_discovery.bin");
+MODULE_FIRMWARE("amdgpu/arcturus_ip_discovery.bin");
+MODULE_FIRMWARE("amdgpu/aldebaran_ip_discovery.bin");
+
+#define mmIP_DISCOVERY_VERSION  0x16A00
+#define mmRCC_CONFIG_MEMSIZE	0xde3
+#define mmMP0_SMN_C2PMSG_33	0x16061
+#define mmMM_INDEX		0x0
+#define mmMM_INDEX_HI		0x6
+#define mmMM_DATA		0x1
+
+static const char *hw_id_names[HW_ID_MAX] = {
+	[MP1_HWID]		= "MP1",
+	[MP2_HWID]		= "MP2",
+	[THM_HWID]		= "THM",
+	[SMUIO_HWID]		= "SMUIO",
+	[FUSE_HWID]		= "FUSE",
+	[CLKA_HWID]		= "CLKA",
+	[PWR_HWID]		= "PWR",
+	[GC_HWID]		= "GC",
+	[UVD_HWID]		= "UVD",
+	[AUDIO_AZ_HWID]		= "AUDIO_AZ",
+	[ACP_HWID]		= "ACP",
+	[DCI_HWID]		= "DCI",
+	[DMU_HWID]		= "DMU",
+	[DCO_HWID]		= "DCO",
+	[DIO_HWID]		= "DIO",
+	[XDMA_HWID]		= "XDMA",
+	[DCEAZ_HWID]		= "DCEAZ",
+	[DAZ_HWID]		= "DAZ",
+	[SDPMUX_HWID]		= "SDPMUX",
+	[NTB_HWID]		= "NTB",
+	[IOHC_HWID]		= "IOHC",
+	[L2IMU_HWID]		= "L2IMU",
+	[VCE_HWID]		= "VCE",
+	[MMHUB_HWID]		= "MMHUB",
+	[ATHUB_HWID]		= "ATHUB",
+	[DBGU_NBIO_HWID]	= "DBGU_NBIO",
+	[DFX_HWID]		= "DFX",
+	[DBGU0_HWID]		= "DBGU0",
+	[DBGU1_HWID]		= "DBGU1",
+	[OSSSYS_HWID]		= "OSSSYS",
+	[HDP_HWID]		= "HDP",
+	[SDMA0_HWID]		= "SDMA0",
+	[SDMA1_HWID]		= "SDMA1",
+	[SDMA2_HWID]		= "SDMA2",
+	[SDMA3_HWID]		= "SDMA3",
+	[LSDMA_HWID]		= "LSDMA",
+	[ISP_HWID]		= "ISP",
+	[DBGU_IO_HWID]		= "DBGU_IO",
+	[DF_HWID]		= "DF",
+	[CLKB_HWID]		= "CLKB",
+	[FCH_HWID]		= "FCH",
+	[DFX_DAP_HWID]		= "DFX_DAP",
+	[L1IMU_PCIE_HWID]	= "L1IMU_PCIE",
+	[L1IMU_NBIF_HWID]	= "L1IMU_NBIF",
+	[L1IMU_IOAGR_HWID]	= "L1IMU_IOAGR",
+	[L1IMU3_HWID]		= "L1IMU3",
+	[L1IMU4_HWID]		= "L1IMU4",
+	[L1IMU5_HWID]		= "L1IMU5",
+	[L1IMU6_HWID]		= "L1IMU6",
+	[L1IMU7_HWID]		= "L1IMU7",
+	[L1IMU8_HWID]		= "L1IMU8",
+	[L1IMU9_HWID]		= "L1IMU9",
+	[L1IMU10_HWID]		= "L1IMU10",
+	[L1IMU11_HWID]		= "L1IMU11",
+	[L1IMU12_HWID]		= "L1IMU12",
+	[L1IMU13_HWID]		= "L1IMU13",
+	[L1IMU14_HWID]		= "L1IMU14",
+	[L1IMU15_HWID]		= "L1IMU15",
+	[WAFLC_HWID]		= "WAFLC",
+	[FCH_USB_PD_HWID]	= "FCH_USB_PD",
+	[PCIE_HWID]		= "PCIE",
+	[PCS_HWID]		= "PCS",
+	[DDCL_HWID]		= "DDCL",
+	[SST_HWID]		= "SST",
+	[IOAGR_HWID]		= "IOAGR",
+	[NBIF_HWID]		= "NBIF",
+	[IOAPIC_HWID]		= "IOAPIC",
+	[SYSTEMHUB_HWID]	= "SYSTEMHUB",
+	[NTBCCP_HWID]		= "NTBCCP",
+	[UMC_HWID]		= "UMC",
+	[SATA_HWID]		= "SATA",
+	[USB_HWID]		= "USB",
+	[CCXSEC_HWID]		= "CCXSEC",
+	[XGMI_HWID]		= "XGMI",
+	[XGBE_HWID]		= "XGBE",
+	[MP0_HWID]		= "MP0",
+	[VPE_HWID]		= "VPE",
+};
+
+static int hw_id_map[MAX_HWIP] = {
+	[GC_HWIP]	= GC_HWID,
+	[HDP_HWIP]	= HDP_HWID,
+	[SDMA0_HWIP]	= SDMA0_HWID,
+	[SDMA1_HWIP]	= SDMA1_HWID,
+	[SDMA2_HWIP]    = SDMA2_HWID,
+	[SDMA3_HWIP]    = SDMA3_HWID,
+	[LSDMA_HWIP]    = LSDMA_HWID,
+	[MMHUB_HWIP]	= MMHUB_HWID,
+	[ATHUB_HWIP]	= ATHUB_HWID,
+	[NBIO_HWIP]	= NBIF_HWID,
+	[MP0_HWIP]	= MP0_HWID,
+	[MP1_HWIP]	= MP1_HWID,
+	[UVD_HWIP]	= UVD_HWID,
+	[VCE_HWIP]	= VCE_HWID,
+	[DF_HWIP]	= DF_HWID,
+	[DCE_HWIP]	= DMU_HWID,
+	[OSSSYS_HWIP]	= OSSSYS_HWID,
+	[SMUIO_HWIP]	= SMUIO_HWID,
+	[PWR_HWIP]	= PWR_HWID,
+	[NBIF_HWIP]	= NBIF_HWID,
+	[THM_HWIP]	= THM_HWID,
+	[CLK_HWIP]	= CLKA_HWID,
+	[UMC_HWIP]	= UMC_HWID,
+	[XGMI_HWIP]	= XGMI_HWID,
+	[DCI_HWIP]	= DCI_HWID,
+	[PCIE_HWIP]	= PCIE_HWID,
+	[VPE_HWIP]	= VPE_HWID,
+	[ISP_HWIP]	= ISP_HWID,
+};
+
+static int amdgpu_discovery_read_binary_from_sysmem(struct amdgpu_device *adev, uint8_t *binary)
+{
+	u64 tmr_offset, tmr_size, pos;
+	void *discv_regn;
+	int ret;
+
+	ret = amdgpu_acpi_get_tmr_info(adev, &tmr_offset, &tmr_size);
+	if (ret)
+		return ret;
+
+	pos = tmr_offset + tmr_size - DISCOVERY_TMR_OFFSET;
+
+	/* This region is read-only and reserved from system use */
+	discv_regn = memremap(pos, adev->mman.discovery_tmr_size, MEMREMAP_WC);
+	if (discv_regn) {
+		memcpy(binary, discv_regn, adev->mman.discovery_tmr_size);
+		memunmap(discv_regn);
+		return 0;
+	}
+
+	return -ENOENT;
+}
+
+#define IP_DISCOVERY_V2		2
+#define IP_DISCOVERY_V4		4
+
+static int amdgpu_discovery_read_binary_from_mem(struct amdgpu_device *adev,
+						 uint8_t *binary)
+{
+	bool sz_valid = true;
+	uint64_t vram_size;
+	int i, ret = 0;
+	u32 msg;
+
+	if (!amdgpu_sriov_vf(adev)) {
+		/* It can take up to two second for IFWI init to complete on some dGPUs,
+		 * but generally it should be in the 60-100ms range.  Normally this starts
+		 * as soon as the device gets power so by the time the OS loads this has long
+		 * completed.  However, when a card is hotplugged via e.g., USB4, we need to
+		 * wait for this to complete.  Once the C2PMSG is updated, we can
+		 * continue.
+		 */
+
+		for (i = 0; i < 2000; i++) {
+			msg = RREG32(mmMP0_SMN_C2PMSG_33);
+			if (msg & 0x80000000)
+				break;
+			msleep(1);
+		}
+	}
+
+	vram_size = RREG32(mmRCC_CONFIG_MEMSIZE);
+	if (!vram_size || vram_size == U32_MAX)
+		sz_valid = false;
+	else
+		vram_size <<= 20;
+
+	if (sz_valid) {
+		uint64_t pos = vram_size - DISCOVERY_TMR_OFFSET;
+		amdgpu_device_vram_access(adev, pos, (uint32_t *)binary,
+					  adev->mman.discovery_tmr_size, false);
+	} else {
+		ret = amdgpu_discovery_read_binary_from_sysmem(adev, binary);
+	}
+
+	if (ret)
+		dev_err(adev->dev,
+			"failed to read discovery info from memory, vram size read: %llx",
+			vram_size);
+
+	return ret;
+}
+
+static int amdgpu_discovery_read_binary_from_file(struct amdgpu_device *adev,
+							uint8_t *binary,
+							const char *fw_name)
+{
+	const struct firmware *fw;
+	int r;
+
+	r = firmware_request_nowarn(&fw, fw_name, adev->dev);
+	if (r) {
+		if (amdgpu_discovery == 2)
+			dev_err(adev->dev, "can't load firmware \"%s\"\n", fw_name);
+		else
+			drm_info(&adev->ddev, "Optional firmware \"%s\" was not found\n", fw_name);
+		return r;
+	}
+
+	memcpy((u8 *)binary, (u8 *)fw->data, fw->size);
+	release_firmware(fw);
+
+	return 0;
+}
+
+static uint16_t amdgpu_discovery_calculate_checksum(uint8_t *data, uint32_t size)
+{
+	uint16_t checksum = 0;
+	int i;
+
+	for (i = 0; i < size; i++)
+		checksum += data[i];
+
+	return checksum;
+}
+
+static inline bool amdgpu_discovery_verify_checksum(uint8_t *data, uint32_t size,
+						    uint16_t expected)
+{
+	return !!(amdgpu_discovery_calculate_checksum(data, size) == expected);
+}
+
+static inline bool amdgpu_discovery_verify_binary_signature(uint8_t *binary)
+{
+	struct binary_header *bhdr;
+	bhdr = (struct binary_header *)binary;
+
+	return (le32_to_cpu(bhdr->binary_signature) == BINARY_SIGNATURE);
+}
+
+static void amdgpu_discovery_harvest_config_quirk(struct amdgpu_device *adev)
+{
+	/*
+	 * So far, apply this quirk only on those Navy Flounder boards which
+	 * have a bad harvest table of VCN config.
+	 */
+	if ((amdgpu_ip_version(adev, UVD_HWIP, 1) == IP_VERSION(3, 0, 1)) &&
+	    (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 3, 2))) {
+		switch (adev->pdev->revision) {
+		case 0xC1:
+		case 0xC2:
+		case 0xC3:
+		case 0xC5:
+		case 0xC7:
+		case 0xCF:
+		case 0xDF:
+			adev->vcn.harvest_config |= AMDGPU_VCN_HARVEST_VCN1;
+			adev->vcn.inst_mask &= ~AMDGPU_VCN_HARVEST_VCN1;
+			break;
+		default:
+			break;
+		}
+	}
+}
+
+static int amdgpu_discovery_verify_npsinfo(struct amdgpu_device *adev,
+					   struct binary_header *bhdr)
+{
+	struct table_info *info;
+	uint16_t checksum;
+	uint16_t offset;
+
+	info = &bhdr->table_list[NPS_INFO];
+	offset = le16_to_cpu(info->offset);
+	checksum = le16_to_cpu(info->checksum);
+
+	struct nps_info_header *nhdr =
+		(struct nps_info_header *)(adev->mman.discovery_bin + offset);
+
+	if (le32_to_cpu(nhdr->table_id) != NPS_INFO_TABLE_ID) {
+		dev_dbg(adev->dev, "invalid ip discovery nps info table id\n");
+		return -EINVAL;
+	}
+
+	if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset,
+					      le32_to_cpu(nhdr->size_bytes),
+					      checksum)) {
+		dev_dbg(adev->dev, "invalid nps info data table checksum\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static const char *amdgpu_discovery_get_fw_name(struct amdgpu_device *adev)
+{
+	if (amdgpu_discovery == 2)
+		return "amdgpu/ip_discovery.bin";
+
+	switch (adev->asic_type) {
+	case CHIP_VEGA10:
+		return "amdgpu/vega10_ip_discovery.bin";
+	case CHIP_VEGA12:
+		return "amdgpu/vega12_ip_discovery.bin";
+	case CHIP_RAVEN:
+		if (adev->apu_flags & AMD_APU_IS_RAVEN2)
+			return "amdgpu/raven2_ip_discovery.bin";
+		else if (adev->apu_flags & AMD_APU_IS_PICASSO)
+			return "amdgpu/picasso_ip_discovery.bin";
+		else
+			return "amdgpu/raven_ip_discovery.bin";
+	case CHIP_VEGA20:
+		return "amdgpu/vega20_ip_discovery.bin";
+	case CHIP_ARCTURUS:
+		return "amdgpu/arcturus_ip_discovery.bin";
+	case CHIP_ALDEBARAN:
+		return "amdgpu/aldebaran_ip_discovery.bin";
+	default:
+		return NULL;
+	}
+}
+
+static int amdgpu_discovery_init(struct amdgpu_device *adev)
+{
+	struct table_info *info;
+	struct binary_header *bhdr;
+	const char *fw_name;
+	uint16_t offset;
+	uint16_t size;
+	uint16_t checksum;
+	int r;
+
+	adev->mman.discovery_tmr_size = DISCOVERY_TMR_SIZE;
+	adev->mman.discovery_bin = kzalloc(adev->mman.discovery_tmr_size, GFP_KERNEL);
+	if (!adev->mman.discovery_bin)
+		return -ENOMEM;
+
+	/* Read from file if it is the preferred option */
+	fw_name = amdgpu_discovery_get_fw_name(adev);
+	if (fw_name != NULL) {
+		drm_dbg(&adev->ddev, "use ip discovery information from file");
+		r = amdgpu_discovery_read_binary_from_file(adev, adev->mman.discovery_bin, fw_name);
+		if (r)
+			goto out;
+	} else {
+		drm_dbg(&adev->ddev, "use ip discovery information from memory");
+		r = amdgpu_discovery_read_binary_from_mem(
+			adev, adev->mman.discovery_bin);
+		if (r)
+			goto out;
+	}
+
+	/* check the ip discovery binary signature */
+	if (!amdgpu_discovery_verify_binary_signature(adev->mman.discovery_bin)) {
+		dev_err(adev->dev,
+			"get invalid ip discovery binary signature\n");
+		r = -EINVAL;
+		goto out;
+	}
+
+	bhdr = (struct binary_header *)adev->mman.discovery_bin;
+
+	offset = offsetof(struct binary_header, binary_checksum) +
+		sizeof(bhdr->binary_checksum);
+	size = le16_to_cpu(bhdr->binary_size) - offset;
+	checksum = le16_to_cpu(bhdr->binary_checksum);
+
+	if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset,
+					      size, checksum)) {
+		dev_err(adev->dev, "invalid ip discovery binary checksum\n");
+		r = -EINVAL;
+		goto out;
+	}
+
+	info = &bhdr->table_list[IP_DISCOVERY];
+	offset = le16_to_cpu(info->offset);
+	checksum = le16_to_cpu(info->checksum);
+
+	if (offset) {
+		struct ip_discovery_header *ihdr =
+			(struct ip_discovery_header *)(adev->mman.discovery_bin + offset);
+		if (le32_to_cpu(ihdr->signature) != DISCOVERY_TABLE_SIGNATURE) {
+			dev_err(adev->dev, "invalid ip discovery data table signature\n");
+			r = -EINVAL;
+			goto out;
+		}
+
+		if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset,
+						      le16_to_cpu(ihdr->size), checksum)) {
+			dev_err(adev->dev, "invalid ip discovery data table checksum\n");
+			r = -EINVAL;
+			goto out;
+		}
+	}
+
+	info = &bhdr->table_list[GC];
+	offset = le16_to_cpu(info->offset);
+	checksum = le16_to_cpu(info->checksum);
+
+	if (offset) {
+		struct gpu_info_header *ghdr =
+			(struct gpu_info_header *)(adev->mman.discovery_bin + offset);
+
+		if (le32_to_cpu(ghdr->table_id) != GC_TABLE_ID) {
+			dev_err(adev->dev, "invalid ip discovery gc table id\n");
+			r = -EINVAL;
+			goto out;
+		}
+
+		if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset,
+						      le32_to_cpu(ghdr->size), checksum)) {
+			dev_err(adev->dev, "invalid gc data table checksum\n");
+			r = -EINVAL;
+			goto out;
+		}
+	}
+
+	info = &bhdr->table_list[HARVEST_INFO];
+	offset = le16_to_cpu(info->offset);
+	checksum = le16_to_cpu(info->checksum);
+
+	if (offset) {
+		struct harvest_info_header *hhdr =
+			(struct harvest_info_header *)(adev->mman.discovery_bin + offset);
+
+		if (le32_to_cpu(hhdr->signature) != HARVEST_TABLE_SIGNATURE) {
+			dev_err(adev->dev, "invalid ip discovery harvest table signature\n");
+			r = -EINVAL;
+			goto out;
+		}
+
+		if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset,
+						      sizeof(struct harvest_table), checksum)) {
+			dev_err(adev->dev, "invalid harvest data table checksum\n");
+			r = -EINVAL;
+			goto out;
+		}
+	}
+
+	info = &bhdr->table_list[VCN_INFO];
+	offset = le16_to_cpu(info->offset);
+	checksum = le16_to_cpu(info->checksum);
+
+	if (offset) {
+		struct vcn_info_header *vhdr =
+			(struct vcn_info_header *)(adev->mman.discovery_bin + offset);
+
+		if (le32_to_cpu(vhdr->table_id) != VCN_INFO_TABLE_ID) {
+			dev_err(adev->dev, "invalid ip discovery vcn table id\n");
+			r = -EINVAL;
+			goto out;
+		}
+
+		if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset,
+						      le32_to_cpu(vhdr->size_bytes), checksum)) {
+			dev_err(adev->dev, "invalid vcn data table checksum\n");
+			r = -EINVAL;
+			goto out;
+		}
+	}
+
+	info = &bhdr->table_list[MALL_INFO];
+	offset = le16_to_cpu(info->offset);
+	checksum = le16_to_cpu(info->checksum);
+
+	if (0 && offset) {
+		struct mall_info_header *mhdr =
+			(struct mall_info_header *)(adev->mman.discovery_bin + offset);
+
+		if (le32_to_cpu(mhdr->table_id) != MALL_INFO_TABLE_ID) {
+			dev_err(adev->dev, "invalid ip discovery mall table id\n");
+			r = -EINVAL;
+			goto out;
+		}
+
+		if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset,
+						      le32_to_cpu(mhdr->size_bytes), checksum)) {
+			dev_err(adev->dev, "invalid mall data table checksum\n");
+			r = -EINVAL;
+			goto out;
+		}
+	}
+
+	return 0;
+
+out:
+	kfree(adev->mman.discovery_bin);
+	adev->mman.discovery_bin = NULL;
+	if ((amdgpu_discovery != 2) &&
+	    (RREG32(mmIP_DISCOVERY_VERSION) == 4))
+		amdgpu_ras_query_boot_status(adev, 4);
+	return r;
+}
+
+static void amdgpu_discovery_sysfs_fini(struct amdgpu_device *adev);
+
+void amdgpu_discovery_fini(struct amdgpu_device *adev)
+{
+	amdgpu_discovery_sysfs_fini(adev);
+	kfree(adev->mman.discovery_bin);
+	adev->mman.discovery_bin = NULL;
+}
+
+static int amdgpu_discovery_validate_ip(struct amdgpu_device *adev,
+					uint8_t instance, uint16_t hw_id)
+{
+	if (instance >= HWIP_MAX_INSTANCE) {
+		dev_err(adev->dev,
+			"Unexpected instance_number (%d) from ip discovery blob\n",
+			instance);
+		return -EINVAL;
+	}
+	if (hw_id >= HW_ID_MAX) {
+		dev_err(adev->dev,
+			"Unexpected hw_id (%d) from ip discovery blob\n",
+			hw_id);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static void amdgpu_discovery_read_harvest_bit_per_ip(struct amdgpu_device *adev,
+						uint32_t *vcn_harvest_count)
+{
+	struct binary_header *bhdr;
+	struct ip_discovery_header *ihdr;
+	struct die_header *dhdr;
+	struct ip *ip;
+	uint16_t die_offset, ip_offset, num_dies, num_ips;
+	uint16_t hw_id;
+	uint8_t inst;
+	int i, j;
+
+	bhdr = (struct binary_header *)adev->mman.discovery_bin;
+	ihdr = (struct ip_discovery_header *)(adev->mman.discovery_bin +
+			le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset));
+	num_dies = le16_to_cpu(ihdr->num_dies);
+
+	/* scan harvest bit of all IP data structures */
+	for (i = 0; i < num_dies; i++) {
+		die_offset = le16_to_cpu(ihdr->die_info[i].die_offset);
+		dhdr = (struct die_header *)(adev->mman.discovery_bin + die_offset);
+		num_ips = le16_to_cpu(dhdr->num_ips);
+		ip_offset = die_offset + sizeof(*dhdr);
+
+		for (j = 0; j < num_ips; j++) {
+			ip = (struct ip *)(adev->mman.discovery_bin +
+					   ip_offset);
+			inst = ip->number_instance;
+			hw_id = le16_to_cpu(ip->hw_id);
+			if (amdgpu_discovery_validate_ip(adev, inst, hw_id))
+				goto next_ip;
+
+			if (ip->harvest == 1) {
+				switch (hw_id) {
+				case VCN_HWID:
+					(*vcn_harvest_count)++;
+					if (inst == 0) {
+						adev->vcn.harvest_config |= AMDGPU_VCN_HARVEST_VCN0;
+						adev->vcn.inst_mask &=
+							~AMDGPU_VCN_HARVEST_VCN0;
+						adev->jpeg.inst_mask &=
+							~AMDGPU_VCN_HARVEST_VCN0;
+					} else {
+						adev->vcn.harvest_config |= AMDGPU_VCN_HARVEST_VCN1;
+						adev->vcn.inst_mask &=
+							~AMDGPU_VCN_HARVEST_VCN1;
+						adev->jpeg.inst_mask &=
+							~AMDGPU_VCN_HARVEST_VCN1;
+					}
+					break;
+				case DMU_HWID:
+					adev->harvest_ip_mask |= AMD_HARVEST_IP_DMU_MASK;
+					break;
+				default:
+					break;
+				}
+			}
+next_ip:
+			ip_offset += struct_size(ip, base_address,
+						 ip->num_base_address);
+		}
+	}
+}
+
+static void amdgpu_discovery_read_from_harvest_table(struct amdgpu_device *adev,
+						     uint32_t *vcn_harvest_count,
+						     uint32_t *umc_harvest_count)
+{
+	struct binary_header *bhdr;
+	struct harvest_table *harvest_info;
+	u16 offset;
+	int i;
+	uint32_t umc_harvest_config = 0;
+
+	bhdr = (struct binary_header *)adev->mman.discovery_bin;
+	offset = le16_to_cpu(bhdr->table_list[HARVEST_INFO].offset);
+
+	if (!offset) {
+		dev_err(adev->dev, "invalid harvest table offset\n");
+		return;
+	}
+
+	harvest_info = (struct harvest_table *)(adev->mman.discovery_bin + offset);
+
+	for (i = 0; i < 32; i++) {
+		if (le16_to_cpu(harvest_info->list[i].hw_id) == 0)
+			break;
+
+		switch (le16_to_cpu(harvest_info->list[i].hw_id)) {
+		case VCN_HWID:
+			(*vcn_harvest_count)++;
+			adev->vcn.harvest_config |=
+				(1 << harvest_info->list[i].number_instance);
+			adev->jpeg.harvest_config |=
+				(1 << harvest_info->list[i].number_instance);
+
+			adev->vcn.inst_mask &=
+				~(1U << harvest_info->list[i].number_instance);
+			adev->jpeg.inst_mask &=
+				~(1U << harvest_info->list[i].number_instance);
+			break;
+		case DMU_HWID:
+			adev->harvest_ip_mask |= AMD_HARVEST_IP_DMU_MASK;
+			break;
+		case UMC_HWID:
+			umc_harvest_config |=
+				1 << (le16_to_cpu(harvest_info->list[i].number_instance));
+			(*umc_harvest_count)++;
+			break;
+		case GC_HWID:
+			adev->gfx.xcc_mask &=
+				~(1U << harvest_info->list[i].number_instance);
+			break;
+		case SDMA0_HWID:
+			adev->sdma.sdma_mask &=
+				~(1U << harvest_info->list[i].number_instance);
+			break;
+#if defined(CONFIG_DRM_AMD_ISP)
+		case ISP_HWID:
+			adev->isp.harvest_config |=
+				~(1U << harvest_info->list[i].number_instance);
+			break;
+#endif
+		default:
+			break;
+		}
+	}
+
+	adev->umc.active_mask = ((1 << adev->umc.node_inst_num) - 1) &
+				~umc_harvest_config;
+}
+
+/* ================================================== */
+
+struct ip_hw_instance {
+	struct kobject kobj; /* ip_discovery/die/#die/#hw_id/#instance/<attrs...> */
+
+	int hw_id;
+	u8  num_instance;
+	u8  major, minor, revision;
+	u8  harvest;
+
+	int num_base_addresses;
+	u32 base_addr[] __counted_by(num_base_addresses);
+};
+
+struct ip_hw_id {
+	struct kset hw_id_kset;  /* ip_discovery/die/#die/#hw_id/, contains ip_hw_instance */
+	int hw_id;
+};
+
+struct ip_die_entry {
+	struct kset ip_kset;     /* ip_discovery/die/#die/, contains ip_hw_id  */
+	u16 num_ips;
+};
+
+/* -------------------------------------------------- */
+
+struct ip_hw_instance_attr {
+	struct attribute attr;
+	ssize_t (*show)(struct ip_hw_instance *ip_hw_instance, char *buf);
+};
+
+static ssize_t hw_id_show(struct ip_hw_instance *ip_hw_instance, char *buf)
+{
+	return sysfs_emit(buf, "%d\n", ip_hw_instance->hw_id);
+}
+
+static ssize_t num_instance_show(struct ip_hw_instance *ip_hw_instance, char *buf)
+{
+	return sysfs_emit(buf, "%d\n", ip_hw_instance->num_instance);
+}
+
+static ssize_t major_show(struct ip_hw_instance *ip_hw_instance, char *buf)
+{
+	return sysfs_emit(buf, "%d\n", ip_hw_instance->major);
+}
+
+static ssize_t minor_show(struct ip_hw_instance *ip_hw_instance, char *buf)
+{
+	return sysfs_emit(buf, "%d\n", ip_hw_instance->minor);
+}
+
+static ssize_t revision_show(struct ip_hw_instance *ip_hw_instance, char *buf)
+{
+	return sysfs_emit(buf, "%d\n", ip_hw_instance->revision);
+}
+
+static ssize_t harvest_show(struct ip_hw_instance *ip_hw_instance, char *buf)
+{
+	return sysfs_emit(buf, "0x%01X\n", ip_hw_instance->harvest);
+}
+
+static ssize_t num_base_addresses_show(struct ip_hw_instance *ip_hw_instance, char *buf)
+{
+	return sysfs_emit(buf, "%d\n", ip_hw_instance->num_base_addresses);
+}
+
+static ssize_t base_addr_show(struct ip_hw_instance *ip_hw_instance, char *buf)
+{
+	ssize_t res, at;
+	int ii;
+
+	for (res = at = ii = 0; ii < ip_hw_instance->num_base_addresses; ii++) {
+		/* Here we satisfy the condition that, at + size <= PAGE_SIZE.
+		 */
+		if (at + 12 > PAGE_SIZE)
+			break;
+		res = sysfs_emit_at(buf, at, "0x%08X\n",
+				    ip_hw_instance->base_addr[ii]);
+		if (res <= 0)
+			break;
+		at += res;
+	}
+
+	return res < 0 ? res : at;
+}
+
+static struct ip_hw_instance_attr ip_hw_attr[] = {
+	__ATTR_RO(hw_id),
+	__ATTR_RO(num_instance),
+	__ATTR_RO(major),
+	__ATTR_RO(minor),
+	__ATTR_RO(revision),
+	__ATTR_RO(harvest),
+	__ATTR_RO(num_base_addresses),
+	__ATTR_RO(base_addr),
+};
+
+static struct attribute *ip_hw_instance_attrs[ARRAY_SIZE(ip_hw_attr) + 1];
+ATTRIBUTE_GROUPS(ip_hw_instance);
+
+#define to_ip_hw_instance(x) container_of(x, struct ip_hw_instance, kobj)
+#define to_ip_hw_instance_attr(x) container_of(x, struct ip_hw_instance_attr, attr)
+
+static ssize_t ip_hw_instance_attr_show(struct kobject *kobj,
+					struct attribute *attr,
+					char *buf)
+{
+	struct ip_hw_instance *ip_hw_instance = to_ip_hw_instance(kobj);
+	struct ip_hw_instance_attr *ip_hw_attr = to_ip_hw_instance_attr(attr);
+
+	if (!ip_hw_attr->show)
+		return -EIO;
+
+	return ip_hw_attr->show(ip_hw_instance, buf);
+}
+
+static const struct sysfs_ops ip_hw_instance_sysfs_ops = {
+	.show = ip_hw_instance_attr_show,
+};
+
+static void ip_hw_instance_release(struct kobject *kobj)
+{
+	struct ip_hw_instance *ip_hw_instance = to_ip_hw_instance(kobj);
+
+	kfree(ip_hw_instance);
+}
+
+static const struct kobj_type ip_hw_instance_ktype = {
+	.release = ip_hw_instance_release,
+	.sysfs_ops = &ip_hw_instance_sysfs_ops,
+	.default_groups = ip_hw_instance_groups,
+};
+
+/* -------------------------------------------------- */
+
+#define to_ip_hw_id(x)  container_of(to_kset(x), struct ip_hw_id, hw_id_kset)
+
+static void ip_hw_id_release(struct kobject *kobj)
+{
+	struct ip_hw_id *ip_hw_id = to_ip_hw_id(kobj);
+
+	if (!list_empty(&ip_hw_id->hw_id_kset.list))
+		DRM_ERROR("ip_hw_id->hw_id_kset is not empty");
+	kfree(ip_hw_id);
+}
+
+static const struct kobj_type ip_hw_id_ktype = {
+	.release = ip_hw_id_release,
+	.sysfs_ops = &kobj_sysfs_ops,
+};
+
+/* -------------------------------------------------- */
+
+static void die_kobj_release(struct kobject *kobj);
+static void ip_disc_release(struct kobject *kobj);
+
+struct ip_die_entry_attribute {
+	struct attribute attr;
+	ssize_t (*show)(struct ip_die_entry *ip_die_entry, char *buf);
+};
+
+#define to_ip_die_entry_attr(x)  container_of(x, struct ip_die_entry_attribute, attr)
+
+static ssize_t num_ips_show(struct ip_die_entry *ip_die_entry, char *buf)
+{
+	return sysfs_emit(buf, "%d\n", ip_die_entry->num_ips);
+}
+
+/* If there are more ip_die_entry attrs, other than the number of IPs,
+ * we can make this intro an array of attrs, and then initialize
+ * ip_die_entry_attrs in a loop.
+ */
+static struct ip_die_entry_attribute num_ips_attr =
+	__ATTR_RO(num_ips);
+
+static struct attribute *ip_die_entry_attrs[] = {
+	&num_ips_attr.attr,
+	NULL,
+};
+ATTRIBUTE_GROUPS(ip_die_entry); /* ip_die_entry_groups */
+
+#define to_ip_die_entry(x) container_of(to_kset(x), struct ip_die_entry, ip_kset)
+
+static ssize_t ip_die_entry_attr_show(struct kobject *kobj,
+				      struct attribute *attr,
+				      char *buf)
+{
+	struct ip_die_entry_attribute *ip_die_entry_attr = to_ip_die_entry_attr(attr);
+	struct ip_die_entry *ip_die_entry = to_ip_die_entry(kobj);
+
+	if (!ip_die_entry_attr->show)
+		return -EIO;
+
+	return ip_die_entry_attr->show(ip_die_entry, buf);
+}
+
+static void ip_die_entry_release(struct kobject *kobj)
+{
+	struct ip_die_entry *ip_die_entry = to_ip_die_entry(kobj);
+
+	if (!list_empty(&ip_die_entry->ip_kset.list))
+		DRM_ERROR("ip_die_entry->ip_kset is not empty");
+	kfree(ip_die_entry);
+}
+
+static const struct sysfs_ops ip_die_entry_sysfs_ops = {
+	.show = ip_die_entry_attr_show,
+};
+
+static const struct kobj_type ip_die_entry_ktype = {
+	.release = ip_die_entry_release,
+	.sysfs_ops = &ip_die_entry_sysfs_ops,
+	.default_groups = ip_die_entry_groups,
+};
+
+static const struct kobj_type die_kobj_ktype = {
+	.release = die_kobj_release,
+	.sysfs_ops = &kobj_sysfs_ops,
+};
+
+static const struct kobj_type ip_discovery_ktype = {
+	.release = ip_disc_release,
+	.sysfs_ops = &kobj_sysfs_ops,
+};
+
+struct ip_discovery_top {
+	struct kobject kobj;    /* ip_discovery/ */
+	struct kset die_kset;   /* ip_discovery/die/, contains ip_die_entry */
+	struct amdgpu_device *adev;
+};
+
+static void die_kobj_release(struct kobject *kobj)
+{
+	struct ip_discovery_top *ip_top = container_of(to_kset(kobj),
+						       struct ip_discovery_top,
+						       die_kset);
+	if (!list_empty(&ip_top->die_kset.list))
+		DRM_ERROR("ip_top->die_kset is not empty");
+}
+
+static void ip_disc_release(struct kobject *kobj)
+{
+	struct ip_discovery_top *ip_top = container_of(kobj, struct ip_discovery_top,
+						       kobj);
+	struct amdgpu_device *adev = ip_top->adev;
+
+	adev->ip_top = NULL;
+	kfree(ip_top);
+}
+
+static uint8_t amdgpu_discovery_get_harvest_info(struct amdgpu_device *adev,
+						 uint16_t hw_id, uint8_t inst)
+{
+	uint8_t harvest = 0;
+
+	/* Until a uniform way is figured, get mask based on hwid */
+	switch (hw_id) {
+	case VCN_HWID:
+		harvest = ((1 << inst) & adev->vcn.inst_mask) == 0;
+		break;
+	case DMU_HWID:
+		if (adev->harvest_ip_mask & AMD_HARVEST_IP_DMU_MASK)
+			harvest = 0x1;
+		break;
+	case UMC_HWID:
+		/* TODO: It needs another parsing; for now, ignore.*/
+		break;
+	case GC_HWID:
+		harvest = ((1 << inst) & adev->gfx.xcc_mask) == 0;
+		break;
+	case SDMA0_HWID:
+		harvest = ((1 << inst) & adev->sdma.sdma_mask) == 0;
+		break;
+	default:
+		break;
+	}
+
+	return harvest;
+}
+
+static int amdgpu_discovery_sysfs_ips(struct amdgpu_device *adev,
+				      struct ip_die_entry *ip_die_entry,
+				      const size_t _ip_offset, const int num_ips,
+				      bool reg_base_64)
+{
+	int ii, jj, kk, res;
+	uint16_t hw_id;
+	uint8_t inst;
+
+	DRM_DEBUG("num_ips:%d", num_ips);
+
+	/* Find all IPs of a given HW ID, and add their instance to
+	 * #die/#hw_id/#instance/<attributes>
+	 */
+	for (ii = 0; ii < HW_ID_MAX; ii++) {
+		struct ip_hw_id *ip_hw_id = NULL;
+		size_t ip_offset = _ip_offset;
+
+		for (jj = 0; jj < num_ips; jj++) {
+			struct ip_v4 *ip;
+			struct ip_hw_instance *ip_hw_instance;
+
+			ip = (struct ip_v4 *)(adev->mman.discovery_bin + ip_offset);
+			inst = ip->instance_number;
+			hw_id = le16_to_cpu(ip->hw_id);
+			if (amdgpu_discovery_validate_ip(adev, inst, hw_id) ||
+			    hw_id != ii)
+				goto next_ip;
+
+			DRM_DEBUG("match:%d @ ip_offset:%zu", ii, ip_offset);
+
+			/* We have a hw_id match; register the hw
+			 * block if not yet registered.
+			 */
+			if (!ip_hw_id) {
+				ip_hw_id = kzalloc(sizeof(*ip_hw_id), GFP_KERNEL);
+				if (!ip_hw_id)
+					return -ENOMEM;
+				ip_hw_id->hw_id = ii;
+
+				kobject_set_name(&ip_hw_id->hw_id_kset.kobj, "%d", ii);
+				ip_hw_id->hw_id_kset.kobj.kset = &ip_die_entry->ip_kset;
+				ip_hw_id->hw_id_kset.kobj.ktype = &ip_hw_id_ktype;
+				res = kset_register(&ip_hw_id->hw_id_kset);
+				if (res) {
+					DRM_ERROR("Couldn't register ip_hw_id kset");
+					kfree(ip_hw_id);
+					return res;
+				}
+				if (hw_id_names[ii]) {
+					res = sysfs_create_link(&ip_die_entry->ip_kset.kobj,
+								&ip_hw_id->hw_id_kset.kobj,
+								hw_id_names[ii]);
+					if (res) {
+						DRM_ERROR("Couldn't create IP link %s in IP Die:%s\n",
+							  hw_id_names[ii],
+							  kobject_name(&ip_die_entry->ip_kset.kobj));
+					}
+				}
+			}
+
+			/* Now register its instance.
+			 */
+			ip_hw_instance = kzalloc(struct_size(ip_hw_instance,
+							     base_addr,
+							     ip->num_base_address),
+						 GFP_KERNEL);
+			if (!ip_hw_instance) {
+				DRM_ERROR("no memory for ip_hw_instance");
+				return -ENOMEM;
+			}
+			ip_hw_instance->hw_id = le16_to_cpu(ip->hw_id); /* == ii */
+			ip_hw_instance->num_instance = ip->instance_number;
+			ip_hw_instance->major = ip->major;
+			ip_hw_instance->minor = ip->minor;
+			ip_hw_instance->revision = ip->revision;
+			ip_hw_instance->harvest =
+				amdgpu_discovery_get_harvest_info(
+					adev, ip_hw_instance->hw_id,
+					ip_hw_instance->num_instance);
+			ip_hw_instance->num_base_addresses = ip->num_base_address;
+
+			for (kk = 0; kk < ip_hw_instance->num_base_addresses; kk++) {
+				if (reg_base_64)
+					ip_hw_instance->base_addr[kk] =
+						lower_32_bits(le64_to_cpu(ip->base_address_64[kk])) & 0x3FFFFFFF;
+				else
+					ip_hw_instance->base_addr[kk] = ip->base_address[kk];
+			}
+
+			kobject_init(&ip_hw_instance->kobj, &ip_hw_instance_ktype);
+			ip_hw_instance->kobj.kset = &ip_hw_id->hw_id_kset;
+			res = kobject_add(&ip_hw_instance->kobj, NULL,
+					  "%d", ip_hw_instance->num_instance);
+next_ip:
+			if (reg_base_64)
+				ip_offset += struct_size(ip, base_address_64,
+							 ip->num_base_address);
+			else
+				ip_offset += struct_size(ip, base_address,
+							 ip->num_base_address);
+		}
+	}
+
+	return 0;
+}
+
+static int amdgpu_discovery_sysfs_recurse(struct amdgpu_device *adev)
+{
+	struct binary_header *bhdr;
+	struct ip_discovery_header *ihdr;
+	struct die_header *dhdr;
+	struct kset *die_kset = &adev->ip_top->die_kset;
+	u16 num_dies, die_offset, num_ips;
+	size_t ip_offset;
+	int ii, res;
+
+	bhdr = (struct binary_header *)adev->mman.discovery_bin;
+	ihdr = (struct ip_discovery_header *)(adev->mman.discovery_bin +
+					      le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset));
+	num_dies = le16_to_cpu(ihdr->num_dies);
+
+	DRM_DEBUG("number of dies: %d\n", num_dies);
+
+	for (ii = 0; ii < num_dies; ii++) {
+		struct ip_die_entry *ip_die_entry;
+
+		die_offset = le16_to_cpu(ihdr->die_info[ii].die_offset);
+		dhdr = (struct die_header *)(adev->mman.discovery_bin + die_offset);
+		num_ips = le16_to_cpu(dhdr->num_ips);
+		ip_offset = die_offset + sizeof(*dhdr);
+
+		/* Add the die to the kset.
+		 *
+		 * dhdr->die_id == ii, which was checked in
+		 * amdgpu_discovery_reg_base_init().
+		 */
+
+		ip_die_entry = kzalloc(sizeof(*ip_die_entry), GFP_KERNEL);
+		if (!ip_die_entry)
+			return -ENOMEM;
+
+		ip_die_entry->num_ips = num_ips;
+
+		kobject_set_name(&ip_die_entry->ip_kset.kobj, "%d", le16_to_cpu(dhdr->die_id));
+		ip_die_entry->ip_kset.kobj.kset = die_kset;
+		ip_die_entry->ip_kset.kobj.ktype = &ip_die_entry_ktype;
+		res = kset_register(&ip_die_entry->ip_kset);
+		if (res) {
+			DRM_ERROR("Couldn't register ip_die_entry kset");
+			kfree(ip_die_entry);
+			return res;
+		}
+
+		amdgpu_discovery_sysfs_ips(adev, ip_die_entry, ip_offset, num_ips, !!ihdr->base_addr_64_bit);
+	}
+
+	return 0;
+}
+
+static int amdgpu_discovery_sysfs_init(struct amdgpu_device *adev)
+{
+	struct kset *die_kset;
+	int res, ii;
+
+	if (!adev->mman.discovery_bin)
+		return -EINVAL;
+
+	adev->ip_top = kzalloc(sizeof(*adev->ip_top), GFP_KERNEL);
+	if (!adev->ip_top)
+		return -ENOMEM;
+
+	adev->ip_top->adev = adev;
+
+	res = kobject_init_and_add(&adev->ip_top->kobj, &ip_discovery_ktype,
+				   &adev->dev->kobj, "ip_discovery");
+	if (res) {
+		DRM_ERROR("Couldn't init and add ip_discovery/");
+		goto Err;
+	}
+
+	die_kset = &adev->ip_top->die_kset;
+	kobject_set_name(&die_kset->kobj, "%s", "die");
+	die_kset->kobj.parent = &adev->ip_top->kobj;
+	die_kset->kobj.ktype = &die_kobj_ktype;
+	res = kset_register(&adev->ip_top->die_kset);
+	if (res) {
+		DRM_ERROR("Couldn't register die_kset");
+		goto Err;
+	}
+
+	for (ii = 0; ii < ARRAY_SIZE(ip_hw_attr); ii++)
+		ip_hw_instance_attrs[ii] = &ip_hw_attr[ii].attr;
+	ip_hw_instance_attrs[ii] = NULL;
+
+	res = amdgpu_discovery_sysfs_recurse(adev);
+
+	return res;
+Err:
+	kobject_put(&adev->ip_top->kobj);
+	return res;
+}
+
+/* -------------------------------------------------- */
+
+#define list_to_kobj(el) container_of(el, struct kobject, entry)
+
+static void amdgpu_discovery_sysfs_ip_hw_free(struct ip_hw_id *ip_hw_id)
+{
+	struct list_head *el, *tmp;
+	struct kset *hw_id_kset;
+
+	hw_id_kset = &ip_hw_id->hw_id_kset;
+	spin_lock(&hw_id_kset->list_lock);
+	list_for_each_prev_safe(el, tmp, &hw_id_kset->list) {
+		list_del_init(el);
+		spin_unlock(&hw_id_kset->list_lock);
+		/* kobject is embedded in ip_hw_instance */
+		kobject_put(list_to_kobj(el));
+		spin_lock(&hw_id_kset->list_lock);
+	}
+	spin_unlock(&hw_id_kset->list_lock);
+	kobject_put(&ip_hw_id->hw_id_kset.kobj);
+}
+
+static void amdgpu_discovery_sysfs_die_free(struct ip_die_entry *ip_die_entry)
+{
+	struct list_head *el, *tmp;
+	struct kset *ip_kset;
+
+	ip_kset = &ip_die_entry->ip_kset;
+	spin_lock(&ip_kset->list_lock);
+	list_for_each_prev_safe(el, tmp, &ip_kset->list) {
+		list_del_init(el);
+		spin_unlock(&ip_kset->list_lock);
+		amdgpu_discovery_sysfs_ip_hw_free(to_ip_hw_id(list_to_kobj(el)));
+		spin_lock(&ip_kset->list_lock);
+	}
+	spin_unlock(&ip_kset->list_lock);
+	kobject_put(&ip_die_entry->ip_kset.kobj);
+}
+
+static void amdgpu_discovery_sysfs_fini(struct amdgpu_device *adev)
+{
+	struct list_head *el, *tmp;
+	struct kset *die_kset;
+
+	die_kset = &adev->ip_top->die_kset;
+	spin_lock(&die_kset->list_lock);
+	list_for_each_prev_safe(el, tmp, &die_kset->list) {
+		list_del_init(el);
+		spin_unlock(&die_kset->list_lock);
+		amdgpu_discovery_sysfs_die_free(to_ip_die_entry(list_to_kobj(el)));
+		spin_lock(&die_kset->list_lock);
+	}
+	spin_unlock(&die_kset->list_lock);
+	kobject_put(&adev->ip_top->die_kset.kobj);
+	kobject_put(&adev->ip_top->kobj);
+}
+
+/* ================================================== */
+
+static int amdgpu_discovery_reg_base_init(struct amdgpu_device *adev)
+{
+	uint8_t num_base_address, subrev, variant;
+	struct binary_header *bhdr;
+	struct ip_discovery_header *ihdr;
+	struct die_header *dhdr;
+	struct ip_v4 *ip;
+	uint16_t die_offset;
+	uint16_t ip_offset;
+	uint16_t num_dies;
+	uint32_t wafl_ver;
+	uint16_t num_ips;
+	uint16_t hw_id;
+	uint8_t inst;
+	int hw_ip;
+	int i, j, k;
+	int r;
+
+	r = amdgpu_discovery_init(adev);
+	if (r)
+		return r;
+
+	wafl_ver = 0;
+	adev->gfx.xcc_mask = 0;
+	adev->sdma.sdma_mask = 0;
+	adev->vcn.inst_mask = 0;
+	adev->jpeg.inst_mask = 0;
+	bhdr = (struct binary_header *)adev->mman.discovery_bin;
+	ihdr = (struct ip_discovery_header *)(adev->mman.discovery_bin +
+			le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset));
+	num_dies = le16_to_cpu(ihdr->num_dies);
+
+	DRM_DEBUG("number of dies: %d\n", num_dies);
+
+	for (i = 0; i < num_dies; i++) {
+		die_offset = le16_to_cpu(ihdr->die_info[i].die_offset);
+		dhdr = (struct die_header *)(adev->mman.discovery_bin + die_offset);
+		num_ips = le16_to_cpu(dhdr->num_ips);
+		ip_offset = die_offset + sizeof(*dhdr);
+
+		if (le16_to_cpu(dhdr->die_id) != i) {
+			DRM_ERROR("invalid die id %d, expected %d\n",
+					le16_to_cpu(dhdr->die_id), i);
+			return -EINVAL;
+		}
+
+		DRM_DEBUG("number of hardware IPs on die%d: %d\n",
+				le16_to_cpu(dhdr->die_id), num_ips);
+
+		for (j = 0; j < num_ips; j++) {
+			ip = (struct ip_v4 *)(adev->mman.discovery_bin + ip_offset);
+
+			inst = ip->instance_number;
+			hw_id = le16_to_cpu(ip->hw_id);
+			if (amdgpu_discovery_validate_ip(adev, inst, hw_id))
+				goto next_ip;
+
+			num_base_address = ip->num_base_address;
+
+			DRM_DEBUG("%s(%d) #%d v%d.%d.%d:\n",
+				  hw_id_names[le16_to_cpu(ip->hw_id)],
+				  le16_to_cpu(ip->hw_id),
+				  ip->instance_number,
+				  ip->major, ip->minor,
+				  ip->revision);
+
+			if (le16_to_cpu(ip->hw_id) == VCN_HWID) {
+				/* Bit [5:0]: original revision value
+				 * Bit [7:6]: en/decode capability:
+				 *     0b00 : VCN function normally
+				 *     0b10 : encode is disabled
+				 *     0b01 : decode is disabled
+				 */
+				if (adev->vcn.num_vcn_inst <
+				    AMDGPU_MAX_VCN_INSTANCES) {
+					adev->vcn.inst[adev->vcn.num_vcn_inst].vcn_config =
+						ip->revision & 0xc0;
+					adev->vcn.num_vcn_inst++;
+					adev->vcn.inst_mask |=
+						(1U << ip->instance_number);
+					adev->jpeg.inst_mask |=
+						(1U << ip->instance_number);
+				} else {
+					dev_err(adev->dev, "Too many VCN instances: %d vs %d\n",
+						adev->vcn.num_vcn_inst + 1,
+						AMDGPU_MAX_VCN_INSTANCES);
+				}
+				ip->revision &= ~0xc0;
+			}
+			if (le16_to_cpu(ip->hw_id) == SDMA0_HWID ||
+			    le16_to_cpu(ip->hw_id) == SDMA1_HWID ||
+			    le16_to_cpu(ip->hw_id) == SDMA2_HWID ||
+			    le16_to_cpu(ip->hw_id) == SDMA3_HWID) {
+				if (adev->sdma.num_instances <
+				    AMDGPU_MAX_SDMA_INSTANCES) {
+					adev->sdma.num_instances++;
+					adev->sdma.sdma_mask |=
+						(1U << ip->instance_number);
+				} else {
+					dev_err(adev->dev, "Too many SDMA instances: %d vs %d\n",
+						adev->sdma.num_instances + 1,
+						AMDGPU_MAX_SDMA_INSTANCES);
+				}
+			}
+
+			if (le16_to_cpu(ip->hw_id) == VPE_HWID) {
+				if (adev->vpe.num_instances < AMDGPU_MAX_VPE_INSTANCES)
+					adev->vpe.num_instances++;
+				else
+					dev_err(adev->dev, "Too many VPE instances: %d vs %d\n",
+						adev->vpe.num_instances + 1,
+						AMDGPU_MAX_VPE_INSTANCES);
+			}
+
+			if (le16_to_cpu(ip->hw_id) == UMC_HWID) {
+				adev->gmc.num_umc++;
+				adev->umc.node_inst_num++;
+			}
+
+			if (le16_to_cpu(ip->hw_id) == GC_HWID)
+				adev->gfx.xcc_mask |=
+					(1U << ip->instance_number);
+
+			if (!wafl_ver && le16_to_cpu(ip->hw_id) == WAFLC_HWID)
+				wafl_ver = IP_VERSION_FULL(ip->major, ip->minor,
+							   ip->revision, 0, 0);
+
+			for (k = 0; k < num_base_address; k++) {
+				/*
+				 * convert the endianness of base addresses in place,
+				 * so that we don't need to convert them when accessing adev->reg_offset.
+				 */
+				if (ihdr->base_addr_64_bit)
+					/* Truncate the 64bit base address from ip discovery
+					 * and only store lower 32bit ip base in reg_offset[].
+					 * Bits > 32 follows ASIC specific format, thus just
+					 * discard them and handle it within specific ASIC.
+					 * By this way reg_offset[] and related helpers can
+					 * stay unchanged.
+					 * The base address is in dwords, thus clear the
+					 * highest 2 bits to store.
+					 */
+					ip->base_address[k] =
+						lower_32_bits(le64_to_cpu(ip->base_address_64[k])) & 0x3FFFFFFF;
+				else
+					ip->base_address[k] = le32_to_cpu(ip->base_address[k]);
+				DRM_DEBUG("\t0x%08x\n", ip->base_address[k]);
+			}
+
+			for (hw_ip = 0; hw_ip < MAX_HWIP; hw_ip++) {
+				if (hw_id_map[hw_ip] == le16_to_cpu(ip->hw_id) &&
+				    hw_id_map[hw_ip] != 0) {
+					DRM_DEBUG("set register base offset for %s\n",
+							hw_id_names[le16_to_cpu(ip->hw_id)]);
+					adev->reg_offset[hw_ip][ip->instance_number] =
+						ip->base_address;
+					/* Instance support is somewhat inconsistent.
+					 * SDMA is a good example.  Sienna cichlid has 4 total
+					 * SDMA instances, each enumerated separately (HWIDs
+					 * 42, 43, 68, 69).  Arcturus has 8 total SDMA instances,
+					 * but they are enumerated as multiple instances of the
+					 * same HWIDs (4x HWID 42, 4x HWID 43).  UMC is another
+					 * example.  On most chips there are multiple instances
+					 * with the same HWID.
+					 */
+
+					if (ihdr->version < 3) {
+						subrev = 0;
+						variant = 0;
+					} else {
+						subrev = ip->sub_revision;
+						variant = ip->variant;
+					}
+
+					adev->ip_versions[hw_ip]
+							 [ip->instance_number] =
+						IP_VERSION_FULL(ip->major,
+								ip->minor,
+								ip->revision,
+								variant,
+								subrev);
+				}
+			}
+
+next_ip:
+			if (ihdr->base_addr_64_bit)
+				ip_offset += struct_size(ip, base_address_64, ip->num_base_address);
+			else
+				ip_offset += struct_size(ip, base_address, ip->num_base_address);
+		}
+	}
+
+	if (wafl_ver && !adev->ip_versions[XGMI_HWIP][0])
+		adev->ip_versions[XGMI_HWIP][0] = wafl_ver;
+
+	return 0;
+}
+
+static void amdgpu_discovery_harvest_ip(struct amdgpu_device *adev)
+{
+	struct ip_discovery_header *ihdr;
+	struct binary_header *bhdr;
+	int vcn_harvest_count = 0;
+	int umc_harvest_count = 0;
+	uint16_t offset, ihdr_ver;
+
+	bhdr = (struct binary_header *)adev->mman.discovery_bin;
+	offset = le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset);
+	ihdr = (struct ip_discovery_header *)(adev->mman.discovery_bin +
+					      offset);
+	ihdr_ver = le16_to_cpu(ihdr->version);
+	/*
+	 * Harvest table does not fit Navi1x and legacy GPUs,
+	 * so read harvest bit per IP data structure to set
+	 * harvest configuration.
+	 */
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(10, 2, 0) &&
+	    ihdr_ver <= 2) {
+		if ((adev->pdev->device == 0x731E &&
+			(adev->pdev->revision == 0xC6 ||
+			 adev->pdev->revision == 0xC7)) ||
+			(adev->pdev->device == 0x7340 &&
+			 adev->pdev->revision == 0xC9) ||
+			(adev->pdev->device == 0x7360 &&
+			 adev->pdev->revision == 0xC7))
+			amdgpu_discovery_read_harvest_bit_per_ip(adev,
+				&vcn_harvest_count);
+	} else {
+		amdgpu_discovery_read_from_harvest_table(adev,
+							 &vcn_harvest_count,
+							 &umc_harvest_count);
+	}
+
+	amdgpu_discovery_harvest_config_quirk(adev);
+
+	if (vcn_harvest_count == adev->vcn.num_vcn_inst) {
+		adev->harvest_ip_mask |= AMD_HARVEST_IP_VCN_MASK;
+		adev->harvest_ip_mask |= AMD_HARVEST_IP_JPEG_MASK;
+	}
+
+	if (umc_harvest_count < adev->gmc.num_umc) {
+		adev->gmc.num_umc -= umc_harvest_count;
+	}
+}
+
+union gc_info {
+	struct gc_info_v1_0 v1;
+	struct gc_info_v1_1 v1_1;
+	struct gc_info_v1_2 v1_2;
+	struct gc_info_v1_3 v1_3;
+	struct gc_info_v2_0 v2;
+	struct gc_info_v2_1 v2_1;
+};
+
+static int amdgpu_discovery_get_gfx_info(struct amdgpu_device *adev)
+{
+	struct binary_header *bhdr;
+	union gc_info *gc_info;
+	u16 offset;
+
+	if (!adev->mman.discovery_bin) {
+		DRM_ERROR("ip discovery uninitialized\n");
+		return -EINVAL;
+	}
+
+	bhdr = (struct binary_header *)adev->mman.discovery_bin;
+	offset = le16_to_cpu(bhdr->table_list[GC].offset);
+
+	if (!offset)
+		return 0;
+
+	gc_info = (union gc_info *)(adev->mman.discovery_bin + offset);
+
+	switch (le16_to_cpu(gc_info->v1.header.version_major)) {
+	case 1:
+		adev->gfx.config.max_shader_engines = le32_to_cpu(gc_info->v1.gc_num_se);
+		adev->gfx.config.max_cu_per_sh = 2 * (le32_to_cpu(gc_info->v1.gc_num_wgp0_per_sa) +
+						      le32_to_cpu(gc_info->v1.gc_num_wgp1_per_sa));
+		adev->gfx.config.max_sh_per_se = le32_to_cpu(gc_info->v1.gc_num_sa_per_se);
+		adev->gfx.config.max_backends_per_se = le32_to_cpu(gc_info->v1.gc_num_rb_per_se);
+		adev->gfx.config.max_texture_channel_caches = le32_to_cpu(gc_info->v1.gc_num_gl2c);
+		adev->gfx.config.max_gprs = le32_to_cpu(gc_info->v1.gc_num_gprs);
+		adev->gfx.config.max_gs_threads = le32_to_cpu(gc_info->v1.gc_num_max_gs_thds);
+		adev->gfx.config.gs_vgt_table_depth = le32_to_cpu(gc_info->v1.gc_gs_table_depth);
+		adev->gfx.config.gs_prim_buffer_depth = le32_to_cpu(gc_info->v1.gc_gsprim_buff_depth);
+		adev->gfx.config.double_offchip_lds_buf = le32_to_cpu(gc_info->v1.gc_double_offchip_lds_buffer);
+		adev->gfx.cu_info.wave_front_size = le32_to_cpu(gc_info->v1.gc_wave_size);
+		adev->gfx.cu_info.max_waves_per_simd = le32_to_cpu(gc_info->v1.gc_max_waves_per_simd);
+		adev->gfx.cu_info.max_scratch_slots_per_cu = le32_to_cpu(gc_info->v1.gc_max_scratch_slots_per_cu);
+		adev->gfx.cu_info.lds_size = le32_to_cpu(gc_info->v1.gc_lds_size);
+		adev->gfx.config.num_sc_per_sh = le32_to_cpu(gc_info->v1.gc_num_sc_per_se) /
+			le32_to_cpu(gc_info->v1.gc_num_sa_per_se);
+		adev->gfx.config.num_packer_per_sc = le32_to_cpu(gc_info->v1.gc_num_packer_per_sc);
+		if (le16_to_cpu(gc_info->v1.header.version_minor) >= 1) {
+			adev->gfx.config.gc_num_tcp_per_sa = le32_to_cpu(gc_info->v1_1.gc_num_tcp_per_sa);
+			adev->gfx.config.gc_num_sdp_interface = le32_to_cpu(gc_info->v1_1.gc_num_sdp_interface);
+			adev->gfx.config.gc_num_tcps = le32_to_cpu(gc_info->v1_1.gc_num_tcps);
+		}
+		if (le16_to_cpu(gc_info->v1.header.version_minor) >= 2) {
+			adev->gfx.config.gc_num_tcp_per_wpg = le32_to_cpu(gc_info->v1_2.gc_num_tcp_per_wpg);
+			adev->gfx.config.gc_tcp_l1_size = le32_to_cpu(gc_info->v1_2.gc_tcp_l1_size);
+			adev->gfx.config.gc_num_sqc_per_wgp = le32_to_cpu(gc_info->v1_2.gc_num_sqc_per_wgp);
+			adev->gfx.config.gc_l1_instruction_cache_size_per_sqc = le32_to_cpu(gc_info->v1_2.gc_l1_instruction_cache_size_per_sqc);
+			adev->gfx.config.gc_l1_data_cache_size_per_sqc = le32_to_cpu(gc_info->v1_2.gc_l1_data_cache_size_per_sqc);
+			adev->gfx.config.gc_gl1c_per_sa = le32_to_cpu(gc_info->v1_2.gc_gl1c_per_sa);
+			adev->gfx.config.gc_gl1c_size_per_instance = le32_to_cpu(gc_info->v1_2.gc_gl1c_size_per_instance);
+			adev->gfx.config.gc_gl2c_per_gpu = le32_to_cpu(gc_info->v1_2.gc_gl2c_per_gpu);
+		}
+		if (le16_to_cpu(gc_info->v1.header.version_minor) >= 3) {
+			adev->gfx.config.gc_tcp_size_per_cu = le32_to_cpu(gc_info->v1_3.gc_tcp_size_per_cu);
+			adev->gfx.config.gc_tcp_cache_line_size = le32_to_cpu(gc_info->v1_3.gc_tcp_cache_line_size);
+			adev->gfx.config.gc_instruction_cache_size_per_sqc = le32_to_cpu(gc_info->v1_3.gc_instruction_cache_size_per_sqc);
+			adev->gfx.config.gc_instruction_cache_line_size = le32_to_cpu(gc_info->v1_3.gc_instruction_cache_line_size);
+			adev->gfx.config.gc_scalar_data_cache_size_per_sqc = le32_to_cpu(gc_info->v1_3.gc_scalar_data_cache_size_per_sqc);
+			adev->gfx.config.gc_scalar_data_cache_line_size = le32_to_cpu(gc_info->v1_3.gc_scalar_data_cache_line_size);
+			adev->gfx.config.gc_tcc_size = le32_to_cpu(gc_info->v1_3.gc_tcc_size);
+			adev->gfx.config.gc_tcc_cache_line_size = le32_to_cpu(gc_info->v1_3.gc_tcc_cache_line_size);
+		}
+		break;
+	case 2:
+		adev->gfx.config.max_shader_engines = le32_to_cpu(gc_info->v2.gc_num_se);
+		adev->gfx.config.max_cu_per_sh = le32_to_cpu(gc_info->v2.gc_num_cu_per_sh);
+		adev->gfx.config.max_sh_per_se = le32_to_cpu(gc_info->v2.gc_num_sh_per_se);
+		adev->gfx.config.max_backends_per_se = le32_to_cpu(gc_info->v2.gc_num_rb_per_se);
+		adev->gfx.config.max_texture_channel_caches = le32_to_cpu(gc_info->v2.gc_num_tccs);
+		adev->gfx.config.max_gprs = le32_to_cpu(gc_info->v2.gc_num_gprs);
+		adev->gfx.config.max_gs_threads = le32_to_cpu(gc_info->v2.gc_num_max_gs_thds);
+		adev->gfx.config.gs_vgt_table_depth = le32_to_cpu(gc_info->v2.gc_gs_table_depth);
+		adev->gfx.config.gs_prim_buffer_depth = le32_to_cpu(gc_info->v2.gc_gsprim_buff_depth);
+		adev->gfx.config.double_offchip_lds_buf = le32_to_cpu(gc_info->v2.gc_double_offchip_lds_buffer);
+		adev->gfx.cu_info.wave_front_size = le32_to_cpu(gc_info->v2.gc_wave_size);
+		adev->gfx.cu_info.max_waves_per_simd = le32_to_cpu(gc_info->v2.gc_max_waves_per_simd);
+		adev->gfx.cu_info.max_scratch_slots_per_cu = le32_to_cpu(gc_info->v2.gc_max_scratch_slots_per_cu);
+		adev->gfx.cu_info.lds_size = le32_to_cpu(gc_info->v2.gc_lds_size);
+		adev->gfx.config.num_sc_per_sh = le32_to_cpu(gc_info->v2.gc_num_sc_per_se) /
+			le32_to_cpu(gc_info->v2.gc_num_sh_per_se);
+		adev->gfx.config.num_packer_per_sc = le32_to_cpu(gc_info->v2.gc_num_packer_per_sc);
+		if (le16_to_cpu(gc_info->v2.header.version_minor) == 1) {
+			adev->gfx.config.gc_num_tcp_per_sa = le32_to_cpu(gc_info->v2_1.gc_num_tcp_per_sh);
+			adev->gfx.config.gc_tcp_size_per_cu = le32_to_cpu(gc_info->v2_1.gc_tcp_size_per_cu);
+			adev->gfx.config.gc_num_sdp_interface = le32_to_cpu(gc_info->v2_1.gc_num_sdp_interface); /* per XCD */
+			adev->gfx.config.gc_num_cu_per_sqc = le32_to_cpu(gc_info->v2_1.gc_num_cu_per_sqc);
+			adev->gfx.config.gc_l1_instruction_cache_size_per_sqc = le32_to_cpu(gc_info->v2_1.gc_instruction_cache_size_per_sqc);
+			adev->gfx.config.gc_l1_data_cache_size_per_sqc = le32_to_cpu(gc_info->v2_1.gc_scalar_data_cache_size_per_sqc);
+			adev->gfx.config.gc_tcc_size = le32_to_cpu(gc_info->v2_1.gc_tcc_size); /* per XCD */
+		}
+		break;
+	default:
+		dev_err(adev->dev,
+			"Unhandled GC info table %d.%d\n",
+			le16_to_cpu(gc_info->v1.header.version_major),
+			le16_to_cpu(gc_info->v1.header.version_minor));
+		return -EINVAL;
+	}
+	return 0;
+}
+
+union mall_info {
+	struct mall_info_v1_0 v1;
+	struct mall_info_v2_0 v2;
+};
+
+static int amdgpu_discovery_get_mall_info(struct amdgpu_device *adev)
+{
+	struct binary_header *bhdr;
+	union mall_info *mall_info;
+	u32 u, mall_size_per_umc, m_s_present, half_use;
+	u64 mall_size;
+	u16 offset;
+
+	if (!adev->mman.discovery_bin) {
+		DRM_ERROR("ip discovery uninitialized\n");
+		return -EINVAL;
+	}
+
+	bhdr = (struct binary_header *)adev->mman.discovery_bin;
+	offset = le16_to_cpu(bhdr->table_list[MALL_INFO].offset);
+
+	if (!offset)
+		return 0;
+
+	mall_info = (union mall_info *)(adev->mman.discovery_bin + offset);
+
+	switch (le16_to_cpu(mall_info->v1.header.version_major)) {
+	case 1:
+		mall_size = 0;
+		mall_size_per_umc = le32_to_cpu(mall_info->v1.mall_size_per_m);
+		m_s_present = le32_to_cpu(mall_info->v1.m_s_present);
+		half_use = le32_to_cpu(mall_info->v1.m_half_use);
+		for (u = 0; u < adev->gmc.num_umc; u++) {
+			if (m_s_present & (1 << u))
+				mall_size += mall_size_per_umc * 2;
+			else if (half_use & (1 << u))
+				mall_size += mall_size_per_umc / 2;
+			else
+				mall_size += mall_size_per_umc;
+		}
+		adev->gmc.mall_size = mall_size;
+		adev->gmc.m_half_use = half_use;
+		break;
+	case 2:
+		mall_size_per_umc = le32_to_cpu(mall_info->v2.mall_size_per_umc);
+		adev->gmc.mall_size = (uint64_t)mall_size_per_umc * adev->gmc.num_umc;
+		break;
+	default:
+		dev_err(adev->dev,
+			"Unhandled MALL info table %d.%d\n",
+			le16_to_cpu(mall_info->v1.header.version_major),
+			le16_to_cpu(mall_info->v1.header.version_minor));
+		return -EINVAL;
+	}
+	return 0;
+}
+
+union vcn_info {
+	struct vcn_info_v1_0 v1;
+};
+
+static int amdgpu_discovery_get_vcn_info(struct amdgpu_device *adev)
+{
+	struct binary_header *bhdr;
+	union vcn_info *vcn_info;
+	u16 offset;
+	int v;
+
+	if (!adev->mman.discovery_bin) {
+		DRM_ERROR("ip discovery uninitialized\n");
+		return -EINVAL;
+	}
+
+	/* num_vcn_inst is currently limited to AMDGPU_MAX_VCN_INSTANCES
+	 * which is smaller than VCN_INFO_TABLE_MAX_NUM_INSTANCES
+	 * but that may change in the future with new GPUs so keep this
+	 * check for defensive purposes.
+	 */
+	if (adev->vcn.num_vcn_inst > VCN_INFO_TABLE_MAX_NUM_INSTANCES) {
+		dev_err(adev->dev, "invalid vcn instances\n");
+		return -EINVAL;
+	}
+
+	bhdr = (struct binary_header *)adev->mman.discovery_bin;
+	offset = le16_to_cpu(bhdr->table_list[VCN_INFO].offset);
+
+	if (!offset)
+		return 0;
+
+	vcn_info = (union vcn_info *)(adev->mman.discovery_bin + offset);
+
+	switch (le16_to_cpu(vcn_info->v1.header.version_major)) {
+	case 1:
+		/* num_vcn_inst is currently limited to AMDGPU_MAX_VCN_INSTANCES
+		 * so this won't overflow.
+		 */
+		for (v = 0; v < adev->vcn.num_vcn_inst; v++) {
+			adev->vcn.inst[v].vcn_codec_disable_mask =
+				le32_to_cpu(vcn_info->v1.instance_info[v].fuse_data.all_bits);
+		}
+		break;
+	default:
+		dev_err(adev->dev,
+			"Unhandled VCN info table %d.%d\n",
+			le16_to_cpu(vcn_info->v1.header.version_major),
+			le16_to_cpu(vcn_info->v1.header.version_minor));
+		return -EINVAL;
+	}
+	return 0;
+}
+
+union nps_info {
+	struct nps_info_v1_0 v1;
+};
+
+static int amdgpu_discovery_refresh_nps_info(struct amdgpu_device *adev,
+					     union nps_info *nps_data)
+{
+	uint64_t vram_size, pos, offset;
+	struct nps_info_header *nhdr;
+	struct binary_header bhdr;
+	uint16_t checksum;
+
+	vram_size = (uint64_t)RREG32(mmRCC_CONFIG_MEMSIZE) << 20;
+	pos = vram_size - DISCOVERY_TMR_OFFSET;
+	amdgpu_device_vram_access(adev, pos, &bhdr, sizeof(bhdr), false);
+
+	offset = le16_to_cpu(bhdr.table_list[NPS_INFO].offset);
+	checksum = le16_to_cpu(bhdr.table_list[NPS_INFO].checksum);
+
+	amdgpu_device_vram_access(adev, (pos + offset), nps_data,
+				  sizeof(*nps_data), false);
+
+	nhdr = (struct nps_info_header *)(nps_data);
+	if (!amdgpu_discovery_verify_checksum((uint8_t *)nps_data,
+					      le32_to_cpu(nhdr->size_bytes),
+					      checksum)) {
+		dev_err(adev->dev, "nps data refresh, checksum mismatch\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+int amdgpu_discovery_get_nps_info(struct amdgpu_device *adev,
+				  uint32_t *nps_type,
+				  struct amdgpu_gmc_memrange **ranges,
+				  int *range_cnt, bool refresh)
+{
+	struct amdgpu_gmc_memrange *mem_ranges;
+	struct binary_header *bhdr;
+	union nps_info *nps_info;
+	union nps_info nps_data;
+	u16 offset;
+	int i, r;
+
+	if (!nps_type || !range_cnt || !ranges)
+		return -EINVAL;
+
+	if (refresh) {
+		r = amdgpu_discovery_refresh_nps_info(adev, &nps_data);
+		if (r)
+			return r;
+		nps_info = &nps_data;
+	} else {
+		if (!adev->mman.discovery_bin) {
+			dev_err(adev->dev,
+				"fetch mem range failed, ip discovery uninitialized\n");
+			return -EINVAL;
+		}
+
+		bhdr = (struct binary_header *)adev->mman.discovery_bin;
+		offset = le16_to_cpu(bhdr->table_list[NPS_INFO].offset);
+
+		if (!offset)
+			return -ENOENT;
+
+		/* If verification fails, return as if NPS table doesn't exist */
+		if (amdgpu_discovery_verify_npsinfo(adev, bhdr))
+			return -ENOENT;
+
+		nps_info =
+			(union nps_info *)(adev->mman.discovery_bin + offset);
+	}
+
+	switch (le16_to_cpu(nps_info->v1.header.version_major)) {
+	case 1:
+		mem_ranges = kvcalloc(nps_info->v1.count,
+				      sizeof(*mem_ranges),
+				      GFP_KERNEL);
+		if (!mem_ranges)
+			return -ENOMEM;
+		*nps_type = nps_info->v1.nps_type;
+		*range_cnt = nps_info->v1.count;
+		for (i = 0; i < *range_cnt; i++) {
+			mem_ranges[i].base_address =
+				nps_info->v1.instance_info[i].base_address;
+			mem_ranges[i].limit_address =
+				nps_info->v1.instance_info[i].limit_address;
+			mem_ranges[i].nid_mask = -1;
+			mem_ranges[i].flags = 0;
+		}
+		*ranges = mem_ranges;
+		break;
+	default:
+		dev_err(adev->dev, "Unhandled NPS info table %d.%d\n",
+			le16_to_cpu(nps_info->v1.header.version_major),
+			le16_to_cpu(nps_info->v1.header.version_minor));
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int amdgpu_discovery_set_common_ip_blocks(struct amdgpu_device *adev)
+{
+	/* what IP to use for this? */
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 0, 1):
+	case IP_VERSION(9, 1, 0):
+	case IP_VERSION(9, 2, 1):
+	case IP_VERSION(9, 2, 2):
+	case IP_VERSION(9, 3, 0):
+	case IP_VERSION(9, 4, 0):
+	case IP_VERSION(9, 4, 1):
+	case IP_VERSION(9, 4, 2):
+	case IP_VERSION(9, 4, 3):
+	case IP_VERSION(9, 4, 4):
+	case IP_VERSION(9, 5, 0):
+		amdgpu_device_ip_block_add(adev, &vega10_common_ip_block);
+		break;
+	case IP_VERSION(10, 1, 10):
+	case IP_VERSION(10, 1, 1):
+	case IP_VERSION(10, 1, 2):
+	case IP_VERSION(10, 1, 3):
+	case IP_VERSION(10, 1, 4):
+	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 3):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+	case IP_VERSION(10, 3, 6):
+	case IP_VERSION(10, 3, 7):
+		amdgpu_device_ip_block_add(adev, &nv_common_ip_block);
+		break;
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 1):
+	case IP_VERSION(11, 0, 2):
+	case IP_VERSION(11, 0, 3):
+	case IP_VERSION(11, 0, 4):
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 1):
+	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
+		amdgpu_device_ip_block_add(adev, &soc21_common_ip_block);
+		break;
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 0, 1):
+		amdgpu_device_ip_block_add(adev, &soc24_common_ip_block);
+		break;
+	default:
+		dev_err(adev->dev,
+			"Failed to add common ip block(GC_HWIP:0x%x)\n",
+			amdgpu_ip_version(adev, GC_HWIP, 0));
+		return -EINVAL;
+	}
+	return 0;
+}
+
+static int amdgpu_discovery_set_gmc_ip_blocks(struct amdgpu_device *adev)
+{
+	/* use GC or MMHUB IP version */
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 0, 1):
+	case IP_VERSION(9, 1, 0):
+	case IP_VERSION(9, 2, 1):
+	case IP_VERSION(9, 2, 2):
+	case IP_VERSION(9, 3, 0):
+	case IP_VERSION(9, 4, 0):
+	case IP_VERSION(9, 4, 1):
+	case IP_VERSION(9, 4, 2):
+	case IP_VERSION(9, 4, 3):
+	case IP_VERSION(9, 4, 4):
+	case IP_VERSION(9, 5, 0):
+		amdgpu_device_ip_block_add(adev, &gmc_v9_0_ip_block);
+		break;
+	case IP_VERSION(10, 1, 10):
+	case IP_VERSION(10, 1, 1):
+	case IP_VERSION(10, 1, 2):
+	case IP_VERSION(10, 1, 3):
+	case IP_VERSION(10, 1, 4):
+	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 3):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+	case IP_VERSION(10, 3, 6):
+	case IP_VERSION(10, 3, 7):
+		amdgpu_device_ip_block_add(adev, &gmc_v10_0_ip_block);
+		break;
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 1):
+	case IP_VERSION(11, 0, 2):
+	case IP_VERSION(11, 0, 3):
+	case IP_VERSION(11, 0, 4):
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 1):
+	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
+		amdgpu_device_ip_block_add(adev, &gmc_v11_0_ip_block);
+		break;
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 0, 1):
+		amdgpu_device_ip_block_add(adev, &gmc_v12_0_ip_block);
+		break;
+	default:
+		dev_err(adev->dev, "Failed to add gmc ip block(GC_HWIP:0x%x)\n",
+			amdgpu_ip_version(adev, GC_HWIP, 0));
+		return -EINVAL;
+	}
+	return 0;
+}
+
+static int amdgpu_discovery_set_ih_ip_blocks(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, OSSSYS_HWIP, 0)) {
+	case IP_VERSION(4, 0, 0):
+	case IP_VERSION(4, 0, 1):
+	case IP_VERSION(4, 1, 0):
+	case IP_VERSION(4, 1, 1):
+	case IP_VERSION(4, 3, 0):
+		amdgpu_device_ip_block_add(adev, &vega10_ih_ip_block);
+		break;
+	case IP_VERSION(4, 2, 0):
+	case IP_VERSION(4, 2, 1):
+	case IP_VERSION(4, 4, 0):
+	case IP_VERSION(4, 4, 2):
+	case IP_VERSION(4, 4, 5):
+		amdgpu_device_ip_block_add(adev, &vega20_ih_ip_block);
+		break;
+	case IP_VERSION(5, 0, 0):
+	case IP_VERSION(5, 0, 1):
+	case IP_VERSION(5, 0, 2):
+	case IP_VERSION(5, 0, 3):
+	case IP_VERSION(5, 2, 0):
+	case IP_VERSION(5, 2, 1):
+		amdgpu_device_ip_block_add(adev, &navi10_ih_ip_block);
+		break;
+	case IP_VERSION(6, 0, 0):
+	case IP_VERSION(6, 0, 1):
+	case IP_VERSION(6, 0, 2):
+		amdgpu_device_ip_block_add(adev, &ih_v6_0_ip_block);
+		break;
+	case IP_VERSION(6, 1, 0):
+		amdgpu_device_ip_block_add(adev, &ih_v6_1_ip_block);
+		break;
+	case IP_VERSION(7, 0, 0):
+		amdgpu_device_ip_block_add(adev, &ih_v7_0_ip_block);
+		break;
+	default:
+		dev_err(adev->dev,
+			"Failed to add ih ip block(OSSSYS_HWIP:0x%x)\n",
+			amdgpu_ip_version(adev, OSSSYS_HWIP, 0));
+		return -EINVAL;
+	}
+	return 0;
+}
+
+static int amdgpu_discovery_set_psp_ip_blocks(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) {
+	case IP_VERSION(9, 0, 0):
+		amdgpu_device_ip_block_add(adev, &psp_v3_1_ip_block);
+		break;
+	case IP_VERSION(10, 0, 0):
+	case IP_VERSION(10, 0, 1):
+		amdgpu_device_ip_block_add(adev, &psp_v10_0_ip_block);
+		break;
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 2):
+	case IP_VERSION(11, 0, 4):
+	case IP_VERSION(11, 0, 5):
+	case IP_VERSION(11, 0, 9):
+	case IP_VERSION(11, 0, 7):
+	case IP_VERSION(11, 0, 11):
+	case IP_VERSION(11, 0, 12):
+	case IP_VERSION(11, 0, 13):
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 2):
+		amdgpu_device_ip_block_add(adev, &psp_v11_0_ip_block);
+		break;
+	case IP_VERSION(11, 0, 8):
+		amdgpu_device_ip_block_add(adev, &psp_v11_0_8_ip_block);
+		break;
+	case IP_VERSION(11, 0, 3):
+	case IP_VERSION(12, 0, 1):
+		amdgpu_device_ip_block_add(adev, &psp_v12_0_ip_block);
+		break;
+	case IP_VERSION(13, 0, 0):
+	case IP_VERSION(13, 0, 1):
+	case IP_VERSION(13, 0, 2):
+	case IP_VERSION(13, 0, 3):
+	case IP_VERSION(13, 0, 5):
+	case IP_VERSION(13, 0, 6):
+	case IP_VERSION(13, 0, 7):
+	case IP_VERSION(13, 0, 8):
+	case IP_VERSION(13, 0, 10):
+	case IP_VERSION(13, 0, 11):
+	case IP_VERSION(13, 0, 12):
+	case IP_VERSION(13, 0, 14):
+	case IP_VERSION(14, 0, 0):
+	case IP_VERSION(14, 0, 1):
+	case IP_VERSION(14, 0, 4):
+		amdgpu_device_ip_block_add(adev, &psp_v13_0_ip_block);
+		break;
+	case IP_VERSION(13, 0, 4):
+		amdgpu_device_ip_block_add(adev, &psp_v13_0_4_ip_block);
+		break;
+	case IP_VERSION(14, 0, 2):
+	case IP_VERSION(14, 0, 3):
+	case IP_VERSION(14, 0, 5):
+		amdgpu_device_ip_block_add(adev, &psp_v14_0_ip_block);
+		break;
+	default:
+		dev_err(adev->dev,
+			"Failed to add psp ip block(MP0_HWIP:0x%x)\n",
+			amdgpu_ip_version(adev, MP0_HWIP, 0));
+		return -EINVAL;
+	}
+	return 0;
+}
+
+static int amdgpu_discovery_set_smu_ip_blocks(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) {
+	case IP_VERSION(9, 0, 0):
+	case IP_VERSION(10, 0, 0):
+	case IP_VERSION(10, 0, 1):
+	case IP_VERSION(11, 0, 2):
+		if (adev->asic_type == CHIP_ARCTURUS)
+			amdgpu_device_ip_block_add(adev, &smu_v11_0_ip_block);
+		else
+			amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
+		break;
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 5):
+	case IP_VERSION(11, 0, 9):
+	case IP_VERSION(11, 0, 7):
+	case IP_VERSION(11, 0, 11):
+	case IP_VERSION(11, 0, 12):
+	case IP_VERSION(11, 0, 13):
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 2):
+		amdgpu_device_ip_block_add(adev, &smu_v11_0_ip_block);
+		break;
+	case IP_VERSION(11, 0, 8):
+		if (adev->apu_flags & AMD_APU_IS_CYAN_SKILLFISH2)
+			amdgpu_device_ip_block_add(adev, &smu_v11_0_ip_block);
+		break;
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 0, 1):
+		amdgpu_device_ip_block_add(adev, &smu_v12_0_ip_block);
+		break;
+	case IP_VERSION(13, 0, 0):
+	case IP_VERSION(13, 0, 1):
+	case IP_VERSION(13, 0, 2):
+	case IP_VERSION(13, 0, 3):
+	case IP_VERSION(13, 0, 4):
+	case IP_VERSION(13, 0, 5):
+	case IP_VERSION(13, 0, 6):
+	case IP_VERSION(13, 0, 7):
+	case IP_VERSION(13, 0, 8):
+	case IP_VERSION(13, 0, 10):
+	case IP_VERSION(13, 0, 11):
+	case IP_VERSION(13, 0, 14):
+	case IP_VERSION(13, 0, 12):
+		amdgpu_device_ip_block_add(adev, &smu_v13_0_ip_block);
+		break;
+	case IP_VERSION(14, 0, 0):
+	case IP_VERSION(14, 0, 1):
+	case IP_VERSION(14, 0, 2):
+	case IP_VERSION(14, 0, 3):
+	case IP_VERSION(14, 0, 4):
+	case IP_VERSION(14, 0, 5):
+		amdgpu_device_ip_block_add(adev, &smu_v14_0_ip_block);
+		break;
+	default:
+		dev_err(adev->dev,
+			"Failed to add smu ip block(MP1_HWIP:0x%x)\n",
+			amdgpu_ip_version(adev, MP1_HWIP, 0));
+		return -EINVAL;
+	}
+	return 0;
+}
+
+#if defined(CONFIG_DRM_AMD_DC)
+static void amdgpu_discovery_set_sriov_display(struct amdgpu_device *adev)
+{
+	amdgpu_device_set_sriov_virtual_display(adev);
+	amdgpu_device_ip_block_add(adev, &amdgpu_vkms_ip_block);
+}
+#endif
+
+static int amdgpu_discovery_set_display_ip_blocks(struct amdgpu_device *adev)
+{
+	if (adev->enable_virtual_display) {
+		amdgpu_device_ip_block_add(adev, &amdgpu_vkms_ip_block);
+		return 0;
+	}
+
+	if (!amdgpu_device_has_dc_support(adev))
+		return 0;
+
+#if defined(CONFIG_DRM_AMD_DC)
+	if (amdgpu_ip_version(adev, DCE_HWIP, 0)) {
+		switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) {
+		case IP_VERSION(1, 0, 0):
+		case IP_VERSION(1, 0, 1):
+		case IP_VERSION(2, 0, 2):
+		case IP_VERSION(2, 0, 0):
+		case IP_VERSION(2, 0, 3):
+		case IP_VERSION(2, 1, 0):
+		case IP_VERSION(3, 0, 0):
+		case IP_VERSION(3, 0, 2):
+		case IP_VERSION(3, 0, 3):
+		case IP_VERSION(3, 0, 1):
+		case IP_VERSION(3, 1, 2):
+		case IP_VERSION(3, 1, 3):
+		case IP_VERSION(3, 1, 4):
+		case IP_VERSION(3, 1, 5):
+		case IP_VERSION(3, 1, 6):
+		case IP_VERSION(3, 2, 0):
+		case IP_VERSION(3, 2, 1):
+		case IP_VERSION(3, 5, 0):
+		case IP_VERSION(3, 5, 1):
+		case IP_VERSION(3, 6, 0):
+		case IP_VERSION(4, 1, 0):
+			/* TODO: Fix IP version. DC code expects version 4.0.1 */
+			if (adev->ip_versions[DCE_HWIP][0] == IP_VERSION(4, 1, 0))
+				adev->ip_versions[DCE_HWIP][0] = IP_VERSION(4, 0, 1);
+
+			if (amdgpu_sriov_vf(adev))
+				amdgpu_discovery_set_sriov_display(adev);
+			else
+				amdgpu_device_ip_block_add(adev, &dm_ip_block);
+			break;
+		default:
+			dev_err(adev->dev,
+				"Failed to add dm ip block(DCE_HWIP:0x%x)\n",
+				amdgpu_ip_version(adev, DCE_HWIP, 0));
+			return -EINVAL;
+		}
+	} else if (amdgpu_ip_version(adev, DCI_HWIP, 0)) {
+		switch (amdgpu_ip_version(adev, DCI_HWIP, 0)) {
+		case IP_VERSION(12, 0, 0):
+		case IP_VERSION(12, 0, 1):
+		case IP_VERSION(12, 1, 0):
+			if (amdgpu_sriov_vf(adev))
+				amdgpu_discovery_set_sriov_display(adev);
+			else
+				amdgpu_device_ip_block_add(adev, &dm_ip_block);
+			break;
+		default:
+			dev_err(adev->dev,
+				"Failed to add dm ip block(DCI_HWIP:0x%x)\n",
+				amdgpu_ip_version(adev, DCI_HWIP, 0));
+			return -EINVAL;
+		}
+	}
+#endif
+	return 0;
+}
+
+static int amdgpu_discovery_set_gc_ip_blocks(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 0, 1):
+	case IP_VERSION(9, 1, 0):
+	case IP_VERSION(9, 2, 1):
+	case IP_VERSION(9, 2, 2):
+	case IP_VERSION(9, 3, 0):
+	case IP_VERSION(9, 4, 0):
+	case IP_VERSION(9, 4, 1):
+	case IP_VERSION(9, 4, 2):
+		amdgpu_device_ip_block_add(adev, &gfx_v9_0_ip_block);
+		break;
+	case IP_VERSION(9, 4, 3):
+	case IP_VERSION(9, 4, 4):
+	case IP_VERSION(9, 5, 0):
+		amdgpu_device_ip_block_add(adev, &gfx_v9_4_3_ip_block);
+		break;
+	case IP_VERSION(10, 1, 10):
+	case IP_VERSION(10, 1, 2):
+	case IP_VERSION(10, 1, 1):
+	case IP_VERSION(10, 1, 3):
+	case IP_VERSION(10, 1, 4):
+	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+	case IP_VERSION(10, 3, 6):
+	case IP_VERSION(10, 3, 3):
+	case IP_VERSION(10, 3, 7):
+		amdgpu_device_ip_block_add(adev, &gfx_v10_0_ip_block);
+		break;
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 1):
+	case IP_VERSION(11, 0, 2):
+	case IP_VERSION(11, 0, 3):
+	case IP_VERSION(11, 0, 4):
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 1):
+	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
+		amdgpu_device_ip_block_add(adev, &gfx_v11_0_ip_block);
+		break;
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 0, 1):
+		amdgpu_device_ip_block_add(adev, &gfx_v12_0_ip_block);
+		break;
+	default:
+		dev_err(adev->dev, "Failed to add gfx ip block(GC_HWIP:0x%x)\n",
+			amdgpu_ip_version(adev, GC_HWIP, 0));
+		return -EINVAL;
+	}
+	return 0;
+}
+
+static int amdgpu_discovery_set_sdma_ip_blocks(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) {
+	case IP_VERSION(4, 0, 0):
+	case IP_VERSION(4, 0, 1):
+	case IP_VERSION(4, 1, 0):
+	case IP_VERSION(4, 1, 1):
+	case IP_VERSION(4, 1, 2):
+	case IP_VERSION(4, 2, 0):
+	case IP_VERSION(4, 2, 2):
+	case IP_VERSION(4, 4, 0):
+		amdgpu_device_ip_block_add(adev, &sdma_v4_0_ip_block);
+		break;
+	case IP_VERSION(4, 4, 2):
+	case IP_VERSION(4, 4, 5):
+	case IP_VERSION(4, 4, 4):
+		amdgpu_device_ip_block_add(adev, &sdma_v4_4_2_ip_block);
+		break;
+	case IP_VERSION(5, 0, 0):
+	case IP_VERSION(5, 0, 1):
+	case IP_VERSION(5, 0, 2):
+	case IP_VERSION(5, 0, 5):
+		amdgpu_device_ip_block_add(adev, &sdma_v5_0_ip_block);
+		break;
+	case IP_VERSION(5, 2, 0):
+	case IP_VERSION(5, 2, 2):
+	case IP_VERSION(5, 2, 4):
+	case IP_VERSION(5, 2, 5):
+	case IP_VERSION(5, 2, 6):
+	case IP_VERSION(5, 2, 3):
+	case IP_VERSION(5, 2, 1):
+	case IP_VERSION(5, 2, 7):
+		amdgpu_device_ip_block_add(adev, &sdma_v5_2_ip_block);
+		break;
+	case IP_VERSION(6, 0, 0):
+	case IP_VERSION(6, 0, 1):
+	case IP_VERSION(6, 0, 2):
+	case IP_VERSION(6, 0, 3):
+	case IP_VERSION(6, 1, 0):
+	case IP_VERSION(6, 1, 1):
+	case IP_VERSION(6, 1, 2):
+	case IP_VERSION(6, 1, 3):
+		amdgpu_device_ip_block_add(adev, &sdma_v6_0_ip_block);
+		break;
+	case IP_VERSION(7, 0, 0):
+	case IP_VERSION(7, 0, 1):
+		amdgpu_device_ip_block_add(adev, &sdma_v7_0_ip_block);
+		break;
+	default:
+		dev_err(adev->dev,
+			"Failed to add sdma ip block(SDMA0_HWIP:0x%x)\n",
+			amdgpu_ip_version(adev, SDMA0_HWIP, 0));
+		return -EINVAL;
+	}
+	return 0;
+}
+
+static int amdgpu_discovery_set_mm_ip_blocks(struct amdgpu_device *adev)
+{
+	if (amdgpu_ip_version(adev, VCE_HWIP, 0)) {
+		switch (amdgpu_ip_version(adev, UVD_HWIP, 0)) {
+		case IP_VERSION(7, 0, 0):
+		case IP_VERSION(7, 2, 0):
+			/* UVD is not supported on vega20 SR-IOV */
+			if (!(adev->asic_type == CHIP_VEGA20 && amdgpu_sriov_vf(adev)))
+				amdgpu_device_ip_block_add(adev, &uvd_v7_0_ip_block);
+			break;
+		default:
+			dev_err(adev->dev,
+				"Failed to add uvd v7 ip block(UVD_HWIP:0x%x)\n",
+				amdgpu_ip_version(adev, UVD_HWIP, 0));
+			return -EINVAL;
+		}
+		switch (amdgpu_ip_version(adev, VCE_HWIP, 0)) {
+		case IP_VERSION(4, 0, 0):
+		case IP_VERSION(4, 1, 0):
+			/* VCE is not supported on vega20 SR-IOV */
+			if (!(adev->asic_type == CHIP_VEGA20 && amdgpu_sriov_vf(adev)))
+				amdgpu_device_ip_block_add(adev, &vce_v4_0_ip_block);
+			break;
+		default:
+			dev_err(adev->dev,
+				"Failed to add VCE v4 ip block(VCE_HWIP:0x%x)\n",
+				amdgpu_ip_version(adev, VCE_HWIP, 0));
+			return -EINVAL;
+		}
+	} else {
+		switch (amdgpu_ip_version(adev, UVD_HWIP, 0)) {
+		case IP_VERSION(1, 0, 0):
+		case IP_VERSION(1, 0, 1):
+			amdgpu_device_ip_block_add(adev, &vcn_v1_0_ip_block);
+			break;
+		case IP_VERSION(2, 0, 0):
+		case IP_VERSION(2, 0, 2):
+		case IP_VERSION(2, 2, 0):
+			amdgpu_device_ip_block_add(adev, &vcn_v2_0_ip_block);
+			if (!amdgpu_sriov_vf(adev))
+				amdgpu_device_ip_block_add(adev, &jpeg_v2_0_ip_block);
+			break;
+		case IP_VERSION(2, 0, 3):
+			break;
+		case IP_VERSION(2, 5, 0):
+			amdgpu_device_ip_block_add(adev, &vcn_v2_5_ip_block);
+			amdgpu_device_ip_block_add(adev, &jpeg_v2_5_ip_block);
+			break;
+		case IP_VERSION(2, 6, 0):
+			amdgpu_device_ip_block_add(adev, &vcn_v2_6_ip_block);
+			amdgpu_device_ip_block_add(adev, &jpeg_v2_6_ip_block);
+			break;
+		case IP_VERSION(3, 0, 0):
+		case IP_VERSION(3, 0, 16):
+		case IP_VERSION(3, 1, 1):
+		case IP_VERSION(3, 1, 2):
+		case IP_VERSION(3, 0, 2):
+			amdgpu_device_ip_block_add(adev, &vcn_v3_0_ip_block);
+			if (!amdgpu_sriov_vf(adev))
+				amdgpu_device_ip_block_add(adev, &jpeg_v3_0_ip_block);
+			break;
+		case IP_VERSION(3, 0, 33):
+			amdgpu_device_ip_block_add(adev, &vcn_v3_0_ip_block);
+			break;
+		case IP_VERSION(4, 0, 0):
+		case IP_VERSION(4, 0, 2):
+		case IP_VERSION(4, 0, 4):
+			amdgpu_device_ip_block_add(adev, &vcn_v4_0_ip_block);
+			amdgpu_device_ip_block_add(adev, &jpeg_v4_0_ip_block);
+			break;
+		case IP_VERSION(4, 0, 3):
+			amdgpu_device_ip_block_add(adev, &vcn_v4_0_3_ip_block);
+			amdgpu_device_ip_block_add(adev, &jpeg_v4_0_3_ip_block);
+			break;
+		case IP_VERSION(4, 0, 5):
+		case IP_VERSION(4, 0, 6):
+			amdgpu_device_ip_block_add(adev, &vcn_v4_0_5_ip_block);
+			amdgpu_device_ip_block_add(adev, &jpeg_v4_0_5_ip_block);
+			break;
+		case IP_VERSION(5, 0, 0):
+			amdgpu_device_ip_block_add(adev, &vcn_v5_0_0_ip_block);
+			amdgpu_device_ip_block_add(adev, &jpeg_v5_0_0_ip_block);
+			break;
+		case IP_VERSION(5, 0, 1):
+			amdgpu_device_ip_block_add(adev, &vcn_v5_0_1_ip_block);
+			amdgpu_device_ip_block_add(adev, &jpeg_v5_0_1_ip_block);
+			break;
+		default:
+			dev_err(adev->dev,
+				"Failed to add vcn/jpeg ip block(UVD_HWIP:0x%x)\n",
+				amdgpu_ip_version(adev, UVD_HWIP, 0));
+			return -EINVAL;
+		}
+	}
+	return 0;
+}
+
+static int amdgpu_discovery_set_mes_ip_blocks(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 1):
+	case IP_VERSION(11, 0, 2):
+	case IP_VERSION(11, 0, 3):
+	case IP_VERSION(11, 0, 4):
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 1):
+	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
+		amdgpu_device_ip_block_add(adev, &mes_v11_0_ip_block);
+		adev->enable_mes = true;
+		adev->enable_mes_kiq = true;
+		break;
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 0, 1):
+		amdgpu_device_ip_block_add(adev, &mes_v12_0_ip_block);
+		adev->enable_mes = true;
+		adev->enable_mes_kiq = true;
+		if (amdgpu_uni_mes)
+			adev->enable_uni_mes = true;
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static void amdgpu_discovery_init_soc_config(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 4, 3):
+	case IP_VERSION(9, 4, 4):
+	case IP_VERSION(9, 5, 0):
+		aqua_vanjaram_init_soc_config(adev);
+		break;
+	default:
+		break;
+	}
+}
+
+static int amdgpu_discovery_set_vpe_ip_blocks(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, VPE_HWIP, 0)) {
+	case IP_VERSION(6, 1, 0):
+	case IP_VERSION(6, 1, 1):
+	case IP_VERSION(6, 1, 3):
+		amdgpu_device_ip_block_add(adev, &vpe_v6_1_ip_block);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int amdgpu_discovery_set_umsch_mm_ip_blocks(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, VCN_HWIP, 0)) {
+	case IP_VERSION(4, 0, 5):
+	case IP_VERSION(4, 0, 6):
+		if (amdgpu_umsch_mm & 0x1) {
+			amdgpu_device_ip_block_add(adev, &umsch_mm_v4_0_ip_block);
+			adev->enable_umsch_mm = true;
+		}
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int amdgpu_discovery_set_isp_ip_blocks(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DRM_AMD_ISP)
+	switch (amdgpu_ip_version(adev, ISP_HWIP, 0)) {
+	case IP_VERSION(4, 1, 0):
+		amdgpu_device_ip_block_add(adev, &isp_v4_1_0_ip_block);
+		break;
+	case IP_VERSION(4, 1, 1):
+		amdgpu_device_ip_block_add(adev, &isp_v4_1_1_ip_block);
+		break;
+	default:
+		break;
+	}
+#endif
+
+	return 0;
+}
+
+int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev)
+{
+	int r;
+
+	switch (adev->asic_type) {
+	case CHIP_VEGA10:
+		/* This is not fatal.  We only need the discovery
+		 * binary for sysfs.  We don't need it for a
+		 * functional system.
+		 */
+		amdgpu_discovery_init(adev);
+		vega10_reg_base_init(adev);
+		adev->sdma.num_instances = 2;
+		adev->gmc.num_umc = 4;
+		adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 0, 0);
+		adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 0, 0);
+		adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(4, 0, 0);
+		adev->ip_versions[HDP_HWIP][0] = IP_VERSION(4, 0, 0);
+		adev->ip_versions[SDMA0_HWIP][0] = IP_VERSION(4, 0, 0);
+		adev->ip_versions[SDMA1_HWIP][0] = IP_VERSION(4, 0, 0);
+		adev->ip_versions[DF_HWIP][0] = IP_VERSION(2, 1, 0);
+		adev->ip_versions[NBIO_HWIP][0] = IP_VERSION(6, 1, 0);
+		adev->ip_versions[UMC_HWIP][0] = IP_VERSION(6, 0, 0);
+		adev->ip_versions[MP0_HWIP][0] = IP_VERSION(9, 0, 0);
+		adev->ip_versions[MP1_HWIP][0] = IP_VERSION(9, 0, 0);
+		adev->ip_versions[THM_HWIP][0] = IP_VERSION(9, 0, 0);
+		adev->ip_versions[SMUIO_HWIP][0] = IP_VERSION(9, 0, 0);
+		adev->ip_versions[GC_HWIP][0] = IP_VERSION(9, 0, 1);
+		adev->ip_versions[UVD_HWIP][0] = IP_VERSION(7, 0, 0);
+		adev->ip_versions[VCE_HWIP][0] = IP_VERSION(4, 0, 0);
+		adev->ip_versions[DCI_HWIP][0] = IP_VERSION(12, 0, 0);
+		break;
+	case CHIP_VEGA12:
+		/* This is not fatal.  We only need the discovery
+		 * binary for sysfs.  We don't need it for a
+		 * functional system.
+		 */
+		amdgpu_discovery_init(adev);
+		vega10_reg_base_init(adev);
+		adev->sdma.num_instances = 2;
+		adev->gmc.num_umc = 4;
+		adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 3, 0);
+		adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 3, 0);
+		adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(4, 0, 1);
+		adev->ip_versions[HDP_HWIP][0] = IP_VERSION(4, 0, 1);
+		adev->ip_versions[SDMA0_HWIP][0] = IP_VERSION(4, 0, 1);
+		adev->ip_versions[SDMA1_HWIP][0] = IP_VERSION(4, 0, 1);
+		adev->ip_versions[DF_HWIP][0] = IP_VERSION(2, 5, 0);
+		adev->ip_versions[NBIO_HWIP][0] = IP_VERSION(6, 2, 0);
+		adev->ip_versions[UMC_HWIP][0] = IP_VERSION(6, 1, 0);
+		adev->ip_versions[MP0_HWIP][0] = IP_VERSION(9, 0, 0);
+		adev->ip_versions[MP1_HWIP][0] = IP_VERSION(9, 0, 0);
+		adev->ip_versions[THM_HWIP][0] = IP_VERSION(9, 0, 0);
+		adev->ip_versions[SMUIO_HWIP][0] = IP_VERSION(9, 0, 1);
+		adev->ip_versions[GC_HWIP][0] = IP_VERSION(9, 2, 1);
+		adev->ip_versions[UVD_HWIP][0] = IP_VERSION(7, 0, 0);
+		adev->ip_versions[VCE_HWIP][0] = IP_VERSION(4, 0, 0);
+		adev->ip_versions[DCI_HWIP][0] = IP_VERSION(12, 0, 1);
+		break;
+	case CHIP_RAVEN:
+		/* This is not fatal.  We only need the discovery
+		 * binary for sysfs.  We don't need it for a
+		 * functional system.
+		 */
+		amdgpu_discovery_init(adev);
+		vega10_reg_base_init(adev);
+		adev->sdma.num_instances = 1;
+		adev->vcn.num_vcn_inst = 1;
+		adev->gmc.num_umc = 2;
+		if (adev->apu_flags & AMD_APU_IS_RAVEN2) {
+			adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 2, 0);
+			adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 2, 0);
+			adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(4, 1, 1);
+			adev->ip_versions[HDP_HWIP][0] = IP_VERSION(4, 1, 1);
+			adev->ip_versions[SDMA0_HWIP][0] = IP_VERSION(4, 1, 1);
+			adev->ip_versions[DF_HWIP][0] = IP_VERSION(2, 1, 1);
+			adev->ip_versions[NBIO_HWIP][0] = IP_VERSION(7, 0, 1);
+			adev->ip_versions[UMC_HWIP][0] = IP_VERSION(7, 5, 0);
+			adev->ip_versions[MP0_HWIP][0] = IP_VERSION(10, 0, 1);
+			adev->ip_versions[MP1_HWIP][0] = IP_VERSION(10, 0, 1);
+			adev->ip_versions[THM_HWIP][0] = IP_VERSION(10, 1, 0);
+			adev->ip_versions[SMUIO_HWIP][0] = IP_VERSION(10, 0, 1);
+			adev->ip_versions[GC_HWIP][0] = IP_VERSION(9, 2, 2);
+			adev->ip_versions[UVD_HWIP][0] = IP_VERSION(1, 0, 1);
+			adev->ip_versions[DCE_HWIP][0] = IP_VERSION(1, 0, 1);
+			adev->ip_versions[ISP_HWIP][0] = IP_VERSION(2, 0, 0);
+		} else {
+			adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 1, 0);
+			adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 1, 0);
+			adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(4, 1, 0);
+			adev->ip_versions[HDP_HWIP][0] = IP_VERSION(4, 1, 0);
+			adev->ip_versions[SDMA0_HWIP][0] = IP_VERSION(4, 1, 0);
+			adev->ip_versions[DF_HWIP][0] = IP_VERSION(2, 1, 0);
+			adev->ip_versions[NBIO_HWIP][0] = IP_VERSION(7, 0, 0);
+			adev->ip_versions[UMC_HWIP][0] = IP_VERSION(7, 0, 0);
+			adev->ip_versions[MP0_HWIP][0] = IP_VERSION(10, 0, 0);
+			adev->ip_versions[MP1_HWIP][0] = IP_VERSION(10, 0, 0);
+			adev->ip_versions[THM_HWIP][0] = IP_VERSION(10, 0, 0);
+			adev->ip_versions[SMUIO_HWIP][0] = IP_VERSION(10, 0, 0);
+			adev->ip_versions[GC_HWIP][0] = IP_VERSION(9, 1, 0);
+			adev->ip_versions[UVD_HWIP][0] = IP_VERSION(1, 0, 0);
+			adev->ip_versions[DCE_HWIP][0] = IP_VERSION(1, 0, 0);
+			adev->ip_versions[ISP_HWIP][0] = IP_VERSION(2, 0, 0);
+		}
+		break;
+	case CHIP_VEGA20:
+		/* This is not fatal.  We only need the discovery
+		 * binary for sysfs.  We don't need it for a
+		 * functional system.
+		 */
+		amdgpu_discovery_init(adev);
+		vega20_reg_base_init(adev);
+		adev->sdma.num_instances = 2;
+		adev->gmc.num_umc = 8;
+		adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 4, 0);
+		adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 4, 0);
+		adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(4, 2, 0);
+		adev->ip_versions[HDP_HWIP][0] = IP_VERSION(4, 2, 0);
+		adev->ip_versions[SDMA0_HWIP][0] = IP_VERSION(4, 2, 0);
+		adev->ip_versions[SDMA1_HWIP][0] = IP_VERSION(4, 2, 0);
+		adev->ip_versions[DF_HWIP][0] = IP_VERSION(3, 6, 0);
+		adev->ip_versions[NBIO_HWIP][0] = IP_VERSION(7, 4, 0);
+		adev->ip_versions[UMC_HWIP][0] = IP_VERSION(6, 1, 1);
+		adev->ip_versions[MP0_HWIP][0] = IP_VERSION(11, 0, 2);
+		adev->ip_versions[MP1_HWIP][0] = IP_VERSION(11, 0, 2);
+		adev->ip_versions[THM_HWIP][0] = IP_VERSION(11, 0, 2);
+		adev->ip_versions[SMUIO_HWIP][0] = IP_VERSION(11, 0, 2);
+		adev->ip_versions[GC_HWIP][0] = IP_VERSION(9, 4, 0);
+		adev->ip_versions[UVD_HWIP][0] = IP_VERSION(7, 2, 0);
+		adev->ip_versions[UVD_HWIP][1] = IP_VERSION(7, 2, 0);
+		adev->ip_versions[VCE_HWIP][0] = IP_VERSION(4, 1, 0);
+		adev->ip_versions[DCI_HWIP][0] = IP_VERSION(12, 1, 0);
+		break;
+	case CHIP_ARCTURUS:
+		/* This is not fatal.  We only need the discovery
+		 * binary for sysfs.  We don't need it for a
+		 * functional system.
+		 */
+		amdgpu_discovery_init(adev);
+		arct_reg_base_init(adev);
+		adev->sdma.num_instances = 8;
+		adev->vcn.num_vcn_inst = 2;
+		adev->gmc.num_umc = 8;
+		adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 4, 1);
+		adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 4, 1);
+		adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(4, 2, 1);
+		adev->ip_versions[HDP_HWIP][0] = IP_VERSION(4, 2, 1);
+		adev->ip_versions[SDMA0_HWIP][0] = IP_VERSION(4, 2, 2);
+		adev->ip_versions[SDMA1_HWIP][0] = IP_VERSION(4, 2, 2);
+		adev->ip_versions[SDMA1_HWIP][1] = IP_VERSION(4, 2, 2);
+		adev->ip_versions[SDMA1_HWIP][2] = IP_VERSION(4, 2, 2);
+		adev->ip_versions[SDMA1_HWIP][3] = IP_VERSION(4, 2, 2);
+		adev->ip_versions[SDMA1_HWIP][4] = IP_VERSION(4, 2, 2);
+		adev->ip_versions[SDMA1_HWIP][5] = IP_VERSION(4, 2, 2);
+		adev->ip_versions[SDMA1_HWIP][6] = IP_VERSION(4, 2, 2);
+		adev->ip_versions[DF_HWIP][0] = IP_VERSION(3, 6, 1);
+		adev->ip_versions[NBIO_HWIP][0] = IP_VERSION(7, 4, 1);
+		adev->ip_versions[UMC_HWIP][0] = IP_VERSION(6, 1, 2);
+		adev->ip_versions[MP0_HWIP][0] = IP_VERSION(11, 0, 4);
+		adev->ip_versions[MP1_HWIP][0] = IP_VERSION(11, 0, 2);
+		adev->ip_versions[THM_HWIP][0] = IP_VERSION(11, 0, 3);
+		adev->ip_versions[SMUIO_HWIP][0] = IP_VERSION(11, 0, 3);
+		adev->ip_versions[GC_HWIP][0] = IP_VERSION(9, 4, 1);
+		adev->ip_versions[UVD_HWIP][0] = IP_VERSION(2, 5, 0);
+		adev->ip_versions[UVD_HWIP][1] = IP_VERSION(2, 5, 0);
+		break;
+	case CHIP_ALDEBARAN:
+		/* This is not fatal.  We only need the discovery
+		 * binary for sysfs.  We don't need it for a
+		 * functional system.
+		 */
+		amdgpu_discovery_init(adev);
+		aldebaran_reg_base_init(adev);
+		adev->sdma.num_instances = 5;
+		adev->vcn.num_vcn_inst = 2;
+		adev->gmc.num_umc = 4;
+		adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 4, 2);
+		adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 4, 2);
+		adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(4, 4, 0);
+		adev->ip_versions[HDP_HWIP][0] = IP_VERSION(4, 4, 0);
+		adev->ip_versions[SDMA0_HWIP][0] = IP_VERSION(4, 4, 0);
+		adev->ip_versions[SDMA0_HWIP][1] = IP_VERSION(4, 4, 0);
+		adev->ip_versions[SDMA0_HWIP][2] = IP_VERSION(4, 4, 0);
+		adev->ip_versions[SDMA0_HWIP][3] = IP_VERSION(4, 4, 0);
+		adev->ip_versions[SDMA0_HWIP][4] = IP_VERSION(4, 4, 0);
+		adev->ip_versions[DF_HWIP][0] = IP_VERSION(3, 6, 2);
+		adev->ip_versions[NBIO_HWIP][0] = IP_VERSION(7, 4, 4);
+		adev->ip_versions[UMC_HWIP][0] = IP_VERSION(6, 7, 0);
+		adev->ip_versions[MP0_HWIP][0] = IP_VERSION(13, 0, 2);
+		adev->ip_versions[MP1_HWIP][0] = IP_VERSION(13, 0, 2);
+		adev->ip_versions[THM_HWIP][0] = IP_VERSION(13, 0, 2);
+		adev->ip_versions[SMUIO_HWIP][0] = IP_VERSION(13, 0, 2);
+		adev->ip_versions[GC_HWIP][0] = IP_VERSION(9, 4, 2);
+		adev->ip_versions[UVD_HWIP][0] = IP_VERSION(2, 6, 0);
+		adev->ip_versions[UVD_HWIP][1] = IP_VERSION(2, 6, 0);
+		adev->ip_versions[XGMI_HWIP][0] = IP_VERSION(6, 1, 0);
+		break;
+	case CHIP_CYAN_SKILLFISH:
+		if (adev->apu_flags & AMD_APU_IS_CYAN_SKILLFISH2) {
+			r = amdgpu_discovery_reg_base_init(adev);
+			if (r)
+				return -EINVAL;
+
+			amdgpu_discovery_harvest_ip(adev);
+			amdgpu_discovery_get_gfx_info(adev);
+			amdgpu_discovery_get_mall_info(adev);
+			amdgpu_discovery_get_vcn_info(adev);
+		} else {
+			cyan_skillfish_reg_base_init(adev);
+			adev->sdma.num_instances = 2;
+			adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(2, 0, 3);
+			adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(2, 0, 3);
+			adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(5, 0, 1);
+			adev->ip_versions[HDP_HWIP][0] = IP_VERSION(5, 0, 1);
+			adev->ip_versions[SDMA0_HWIP][0] = IP_VERSION(5, 0, 1);
+			adev->ip_versions[SDMA1_HWIP][1] = IP_VERSION(5, 0, 1);
+			adev->ip_versions[DF_HWIP][0] = IP_VERSION(3, 5, 0);
+			adev->ip_versions[NBIO_HWIP][0] = IP_VERSION(2, 1, 1);
+			adev->ip_versions[UMC_HWIP][0] = IP_VERSION(8, 1, 1);
+			adev->ip_versions[MP0_HWIP][0] = IP_VERSION(11, 0, 8);
+			adev->ip_versions[MP1_HWIP][0] = IP_VERSION(11, 0, 8);
+			adev->ip_versions[THM_HWIP][0] = IP_VERSION(11, 0, 1);
+			adev->ip_versions[SMUIO_HWIP][0] = IP_VERSION(11, 0, 8);
+			adev->ip_versions[GC_HWIP][0] = IP_VERSION(10, 1, 3);
+			adev->ip_versions[UVD_HWIP][0] = IP_VERSION(2, 0, 3);
+		}
+		break;
+	default:
+		r = amdgpu_discovery_reg_base_init(adev);
+		if (r) {
+			drm_err(&adev->ddev, "discovery failed: %d\n", r);
+			return r;
+		}
+
+		amdgpu_discovery_harvest_ip(adev);
+		amdgpu_discovery_get_gfx_info(adev);
+		amdgpu_discovery_get_mall_info(adev);
+		amdgpu_discovery_get_vcn_info(adev);
+		break;
+	}
+
+	amdgpu_discovery_init_soc_config(adev);
+	amdgpu_discovery_sysfs_init(adev);
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 0, 1):
+	case IP_VERSION(9, 2, 1):
+	case IP_VERSION(9, 4, 0):
+	case IP_VERSION(9, 4, 1):
+	case IP_VERSION(9, 4, 2):
+	case IP_VERSION(9, 4, 3):
+	case IP_VERSION(9, 4, 4):
+	case IP_VERSION(9, 5, 0):
+		adev->family = AMDGPU_FAMILY_AI;
+		break;
+	case IP_VERSION(9, 1, 0):
+	case IP_VERSION(9, 2, 2):
+	case IP_VERSION(9, 3, 0):
+		adev->family = AMDGPU_FAMILY_RV;
+		break;
+	case IP_VERSION(10, 1, 10):
+	case IP_VERSION(10, 1, 1):
+	case IP_VERSION(10, 1, 2):
+	case IP_VERSION(10, 1, 3):
+	case IP_VERSION(10, 1, 4):
+	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+		adev->family = AMDGPU_FAMILY_NV;
+		break;
+	case IP_VERSION(10, 3, 1):
+		adev->family = AMDGPU_FAMILY_VGH;
+		adev->apu_flags |= AMD_APU_IS_VANGOGH;
+		break;
+	case IP_VERSION(10, 3, 3):
+		adev->family = AMDGPU_FAMILY_YC;
+		break;
+	case IP_VERSION(10, 3, 6):
+		adev->family = AMDGPU_FAMILY_GC_10_3_6;
+		break;
+	case IP_VERSION(10, 3, 7):
+		adev->family = AMDGPU_FAMILY_GC_10_3_7;
+		break;
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 2):
+	case IP_VERSION(11, 0, 3):
+		adev->family = AMDGPU_FAMILY_GC_11_0_0;
+		break;
+	case IP_VERSION(11, 0, 1):
+	case IP_VERSION(11, 0, 4):
+		adev->family = AMDGPU_FAMILY_GC_11_0_1;
+		break;
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 1):
+	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
+		adev->family = AMDGPU_FAMILY_GC_11_5_0;
+		break;
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 0, 1):
+		adev->family = AMDGPU_FAMILY_GC_12_0_0;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 1, 0):
+	case IP_VERSION(9, 2, 2):
+	case IP_VERSION(9, 3, 0):
+	case IP_VERSION(10, 1, 3):
+	case IP_VERSION(10, 1, 4):
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 3):
+	case IP_VERSION(10, 3, 6):
+	case IP_VERSION(10, 3, 7):
+	case IP_VERSION(11, 0, 1):
+	case IP_VERSION(11, 0, 4):
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 1):
+	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
+		adev->flags |= AMD_IS_APU;
+		break;
+	default:
+		break;
+	}
+
+	/* set NBIO version */
+	switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) {
+	case IP_VERSION(6, 1, 0):
+	case IP_VERSION(6, 2, 0):
+		adev->nbio.funcs = &nbio_v6_1_funcs;
+		adev->nbio.hdp_flush_reg = &nbio_v6_1_hdp_flush_reg;
+		break;
+	case IP_VERSION(7, 0, 0):
+	case IP_VERSION(7, 0, 1):
+	case IP_VERSION(2, 5, 0):
+		adev->nbio.funcs = &nbio_v7_0_funcs;
+		adev->nbio.hdp_flush_reg = &nbio_v7_0_hdp_flush_reg;
+		break;
+	case IP_VERSION(7, 4, 0):
+	case IP_VERSION(7, 4, 1):
+	case IP_VERSION(7, 4, 4):
+		adev->nbio.funcs = &nbio_v7_4_funcs;
+		adev->nbio.hdp_flush_reg = &nbio_v7_4_hdp_flush_reg;
+		break;
+	case IP_VERSION(7, 9, 0):
+	case IP_VERSION(7, 9, 1):
+		adev->nbio.funcs = &nbio_v7_9_funcs;
+		adev->nbio.hdp_flush_reg = &nbio_v7_9_hdp_flush_reg;
+		break;
+	case IP_VERSION(7, 11, 0):
+	case IP_VERSION(7, 11, 1):
+	case IP_VERSION(7, 11, 2):
+	case IP_VERSION(7, 11, 3):
+		adev->nbio.funcs = &nbio_v7_11_funcs;
+		adev->nbio.hdp_flush_reg = &nbio_v7_11_hdp_flush_reg;
+		break;
+	case IP_VERSION(7, 2, 0):
+	case IP_VERSION(7, 2, 1):
+	case IP_VERSION(7, 3, 0):
+	case IP_VERSION(7, 5, 0):
+	case IP_VERSION(7, 5, 1):
+		adev->nbio.funcs = &nbio_v7_2_funcs;
+		adev->nbio.hdp_flush_reg = &nbio_v7_2_hdp_flush_reg;
+		break;
+	case IP_VERSION(2, 1, 1):
+	case IP_VERSION(2, 3, 0):
+	case IP_VERSION(2, 3, 1):
+	case IP_VERSION(2, 3, 2):
+	case IP_VERSION(3, 3, 0):
+	case IP_VERSION(3, 3, 1):
+	case IP_VERSION(3, 3, 2):
+	case IP_VERSION(3, 3, 3):
+		adev->nbio.funcs = &nbio_v2_3_funcs;
+		adev->nbio.hdp_flush_reg = &nbio_v2_3_hdp_flush_reg;
+		break;
+	case IP_VERSION(4, 3, 0):
+	case IP_VERSION(4, 3, 1):
+		if (amdgpu_sriov_vf(adev))
+			adev->nbio.funcs = &nbio_v4_3_sriov_funcs;
+		else
+			adev->nbio.funcs = &nbio_v4_3_funcs;
+		adev->nbio.hdp_flush_reg = &nbio_v4_3_hdp_flush_reg;
+		break;
+	case IP_VERSION(7, 7, 0):
+	case IP_VERSION(7, 7, 1):
+		adev->nbio.funcs = &nbio_v7_7_funcs;
+		adev->nbio.hdp_flush_reg = &nbio_v7_7_hdp_flush_reg;
+		break;
+	case IP_VERSION(6, 3, 1):
+		adev->nbio.funcs = &nbif_v6_3_1_funcs;
+		adev->nbio.hdp_flush_reg = &nbif_v6_3_1_hdp_flush_reg;
+		break;
+	default:
+		break;
+	}
+
+	switch (amdgpu_ip_version(adev, HDP_HWIP, 0)) {
+	case IP_VERSION(4, 0, 0):
+	case IP_VERSION(4, 0, 1):
+	case IP_VERSION(4, 1, 0):
+	case IP_VERSION(4, 1, 1):
+	case IP_VERSION(4, 1, 2):
+	case IP_VERSION(4, 2, 0):
+	case IP_VERSION(4, 2, 1):
+	case IP_VERSION(4, 4, 0):
+	case IP_VERSION(4, 4, 2):
+	case IP_VERSION(4, 4, 5):
+		adev->hdp.funcs = &hdp_v4_0_funcs;
+		break;
+	case IP_VERSION(5, 0, 0):
+	case IP_VERSION(5, 0, 1):
+	case IP_VERSION(5, 0, 2):
+	case IP_VERSION(5, 0, 3):
+	case IP_VERSION(5, 0, 4):
+	case IP_VERSION(5, 2, 0):
+		adev->hdp.funcs = &hdp_v5_0_funcs;
+		break;
+	case IP_VERSION(5, 2, 1):
+		adev->hdp.funcs = &hdp_v5_2_funcs;
+		break;
+	case IP_VERSION(6, 0, 0):
+	case IP_VERSION(6, 0, 1):
+	case IP_VERSION(6, 1, 0):
+		adev->hdp.funcs = &hdp_v6_0_funcs;
+		break;
+	case IP_VERSION(7, 0, 0):
+		adev->hdp.funcs = &hdp_v7_0_funcs;
+		break;
+	default:
+		break;
+	}
+
+	switch (amdgpu_ip_version(adev, DF_HWIP, 0)) {
+	case IP_VERSION(3, 6, 0):
+	case IP_VERSION(3, 6, 1):
+	case IP_VERSION(3, 6, 2):
+		adev->df.funcs = &df_v3_6_funcs;
+		break;
+	case IP_VERSION(2, 1, 0):
+	case IP_VERSION(2, 1, 1):
+	case IP_VERSION(2, 5, 0):
+	case IP_VERSION(3, 5, 1):
+	case IP_VERSION(3, 5, 2):
+		adev->df.funcs = &df_v1_7_funcs;
+		break;
+	case IP_VERSION(4, 3, 0):
+		adev->df.funcs = &df_v4_3_funcs;
+		break;
+	case IP_VERSION(4, 6, 2):
+		adev->df.funcs = &df_v4_6_2_funcs;
+		break;
+	case IP_VERSION(4, 15, 0):
+	case IP_VERSION(4, 15, 1):
+		adev->df.funcs = &df_v4_15_funcs;
+		break;
+	default:
+		break;
+	}
+
+	switch (amdgpu_ip_version(adev, SMUIO_HWIP, 0)) {
+	case IP_VERSION(9, 0, 0):
+	case IP_VERSION(9, 0, 1):
+	case IP_VERSION(10, 0, 0):
+	case IP_VERSION(10, 0, 1):
+	case IP_VERSION(10, 0, 2):
+		adev->smuio.funcs = &smuio_v9_0_funcs;
+		break;
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 2):
+	case IP_VERSION(11, 0, 3):
+	case IP_VERSION(11, 0, 4):
+	case IP_VERSION(11, 0, 7):
+	case IP_VERSION(11, 0, 8):
+		adev->smuio.funcs = &smuio_v11_0_funcs;
+		break;
+	case IP_VERSION(11, 0, 6):
+	case IP_VERSION(11, 0, 10):
+	case IP_VERSION(11, 0, 11):
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(13, 0, 1):
+	case IP_VERSION(13, 0, 9):
+	case IP_VERSION(13, 0, 10):
+		adev->smuio.funcs = &smuio_v11_0_6_funcs;
+		break;
+	case IP_VERSION(13, 0, 2):
+		adev->smuio.funcs = &smuio_v13_0_funcs;
+		break;
+	case IP_VERSION(13, 0, 3):
+	case IP_VERSION(13, 0, 11):
+		adev->smuio.funcs = &smuio_v13_0_3_funcs;
+		if (adev->smuio.funcs->get_pkg_type(adev) == AMDGPU_PKG_TYPE_APU) {
+			adev->flags |= AMD_IS_APU;
+		}
+		break;
+	case IP_VERSION(13, 0, 6):
+	case IP_VERSION(13, 0, 8):
+	case IP_VERSION(14, 0, 0):
+	case IP_VERSION(14, 0, 1):
+		adev->smuio.funcs = &smuio_v13_0_6_funcs;
+		break;
+	case IP_VERSION(14, 0, 2):
+		adev->smuio.funcs = &smuio_v14_0_2_funcs;
+		break;
+	default:
+		break;
+	}
+
+	switch (amdgpu_ip_version(adev, LSDMA_HWIP, 0)) {
+	case IP_VERSION(6, 0, 0):
+	case IP_VERSION(6, 0, 1):
+	case IP_VERSION(6, 0, 2):
+	case IP_VERSION(6, 0, 3):
+		adev->lsdma.funcs = &lsdma_v6_0_funcs;
+		break;
+	case IP_VERSION(7, 0, 0):
+	case IP_VERSION(7, 0, 1):
+		adev->lsdma.funcs = &lsdma_v7_0_funcs;
+		break;
+	default:
+		break;
+	}
+
+	r = amdgpu_discovery_set_common_ip_blocks(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_discovery_set_gmc_ip_blocks(adev);
+	if (r)
+		return r;
+
+	/* For SR-IOV, PSP needs to be initialized before IH */
+	if (amdgpu_sriov_vf(adev)) {
+		r = amdgpu_discovery_set_psp_ip_blocks(adev);
+		if (r)
+			return r;
+		r = amdgpu_discovery_set_ih_ip_blocks(adev);
+		if (r)
+			return r;
+	} else {
+		r = amdgpu_discovery_set_ih_ip_blocks(adev);
+		if (r)
+			return r;
+
+		if (likely(adev->firmware.load_type == AMDGPU_FW_LOAD_PSP)) {
+			r = amdgpu_discovery_set_psp_ip_blocks(adev);
+			if (r)
+				return r;
+		}
+	}
+
+	if (likely(adev->firmware.load_type == AMDGPU_FW_LOAD_PSP)) {
+		r = amdgpu_discovery_set_smu_ip_blocks(adev);
+		if (r)
+			return r;
+	}
+
+	r = amdgpu_discovery_set_display_ip_blocks(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_discovery_set_gc_ip_blocks(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_discovery_set_sdma_ip_blocks(adev);
+	if (r)
+		return r;
+
+	if ((adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT &&
+	     !amdgpu_sriov_vf(adev)) ||
+	    (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO && amdgpu_dpm == 1)) {
+		r = amdgpu_discovery_set_smu_ip_blocks(adev);
+		if (r)
+			return r;
+	}
+
+	r = amdgpu_discovery_set_mm_ip_blocks(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_discovery_set_mes_ip_blocks(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_discovery_set_vpe_ip_blocks(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_discovery_set_umsch_mm_ip_blocks(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_discovery_set_isp_ip_blocks(adev);
+	if (r)
+		return r;
+	return 0;
+}
+
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.h
new file mode 100644
index 000000000000..b44d56465c5b
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.h
@@ -0,0 +1,38 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_DISCOVERY__
+#define __AMDGPU_DISCOVERY__
+
+#define DISCOVERY_TMR_SIZE      (10 << 10)
+#define DISCOVERY_TMR_OFFSET    (64 << 10)
+
+void amdgpu_discovery_fini(struct amdgpu_device *adev);
+int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev);
+
+int amdgpu_discovery_get_nps_info(struct amdgpu_device *adev,
+				  uint32_t *nps_type,
+				  struct amdgpu_gmc_memrange **ranges,
+				  int *range_cnt, bool refresh);
+
+#endif /* __AMDGPU_DISCOVERY__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
index 6ad243293a78..51bab32fd8c6 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
@@ -23,19 +23,73 @@
  * Authors: Dave Airlie
  *          Alex Deucher
  */
-#include <drm/drmP.h>
+
 #include <drm/amdgpu_drm.h>
 #include "amdgpu.h"
 #include "amdgpu_i2c.h"
 #include "atom.h"
 #include "amdgpu_connectors.h"
+#include "amdgpu_display.h"
+#include "soc15_common.h"
+#include "gc/gc_11_0_0_offset.h"
+#include "gc/gc_11_0_0_sh_mask.h"
+#include "bif/bif_4_1_d.h"
 #include <asm/div64.h>
 
+#include <linux/pci.h>
 #include <linux/pm_runtime.h>
 #include <drm/drm_crtc_helper.h>
+#include <drm/drm_damage_helper.h>
+#include <drm/drm_drv.h>
 #include <drm/drm_edid.h>
+#include <drm/drm_fb_helper.h>
+#include <drm/drm_gem_framebuffer_helper.h>
+#include <drm/drm_fourcc.h>
+#include <drm/drm_modeset_helper.h>
+#include <drm/drm_vblank.h>
 
-static void amdgpu_flip_callback(struct dma_fence *f, struct dma_fence_cb *cb)
+/**
+ * amdgpu_display_hotplug_work_func - work handler for display hotplug event
+ *
+ * @work: work struct pointer
+ *
+ * This is the hotplug event work handler (all ASICs).
+ * The work gets scheduled from the IRQ handler if there
+ * was a hotplug interrupt.  It walks through the connector table
+ * and calls hotplug handler for each connector. After this, it sends
+ * a DRM hotplug event to alert userspace.
+ *
+ * This design approach is required in order to defer hotplug event handling
+ * from the IRQ handler to a work handler because hotplug handler has to use
+ * mutexes which cannot be locked in an IRQ handler (since &mutex_lock may
+ * sleep).
+ */
+void amdgpu_display_hotplug_work_func(struct work_struct *work)
+{
+	struct amdgpu_device *adev = container_of(work, struct amdgpu_device,
+						  hotplug_work.work);
+	struct drm_device *dev = adev_to_drm(adev);
+	struct drm_mode_config *mode_config = &dev->mode_config;
+	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
+
+	mutex_lock(&mode_config->mutex);
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter)
+		amdgpu_connector_hotplug(connector);
+	drm_connector_list_iter_end(&iter);
+	mutex_unlock(&mode_config->mutex);
+	/* Just fire off a uevent and let userspace tell us what to do */
+	drm_helper_hpd_irq_event(dev);
+}
+
+static int amdgpu_display_framebuffer_init(struct drm_device *dev,
+					   struct amdgpu_framebuffer *rfb,
+					   const struct drm_mode_fb_cmd2 *mode_cmd,
+					   struct drm_gem_object *obj);
+
+static void amdgpu_display_flip_callback(struct dma_fence *f,
+					 struct dma_fence_cb *cb)
 {
 	struct amdgpu_flip_work *work =
 		container_of(cb, struct amdgpu_flip_work, cb);
@@ -44,24 +98,25 @@ static void amdgpu_flip_callback(struct dma_fence *f, struct dma_fence_cb *cb)
 	schedule_work(&work->flip_work.work);
 }
 
-static bool amdgpu_flip_handle_fence(struct amdgpu_flip_work *work,
-				     struct dma_fence **f)
+static bool amdgpu_display_flip_handle_fence(struct amdgpu_flip_work *work,
+					     struct dma_fence **f)
 {
-	struct dma_fence *fence= *f;
+	struct dma_fence *fence = *f;
 
 	if (fence == NULL)
 		return false;
 
 	*f = NULL;
 
-	if (!dma_fence_add_callback(fence, &work->cb, amdgpu_flip_callback))
+	if (!dma_fence_add_callback(fence, &work->cb,
+				    amdgpu_display_flip_callback))
 		return true;
 
 	dma_fence_put(fence);
 	return false;
 }
 
-static void amdgpu_flip_work_func(struct work_struct *__work)
+static void amdgpu_display_flip_work_func(struct work_struct *__work)
 {
 	struct delayed_work *delayed_work =
 		container_of(__work, struct delayed_work, work);
@@ -72,27 +127,24 @@ static void amdgpu_flip_work_func(struct work_struct *__work)
 
 	struct drm_crtc *crtc = &amdgpu_crtc->base;
 	unsigned long flags;
-	unsigned i;
+	unsigned int i;
 	int vpos, hpos;
 
-	if (amdgpu_flip_handle_fence(work, &work->excl))
-		return;
-
 	for (i = 0; i < work->shared_count; ++i)
-		if (amdgpu_flip_handle_fence(work, &work->shared[i]))
+		if (amdgpu_display_flip_handle_fence(work, &work->shared[i]))
 			return;
 
 	/* Wait until we're out of the vertical blank period before the one
 	 * targeted by the flip
 	 */
 	if (amdgpu_crtc->enabled &&
-	    (amdgpu_get_crtc_scanoutpos(adev->ddev, work->crtc_id, 0,
-					&vpos, &hpos, NULL, NULL,
-					&crtc->hwmode)
+	    (amdgpu_display_get_crtc_scanoutpos(adev_to_drm(adev), work->crtc_id, 0,
+						&vpos, &hpos, NULL, NULL,
+						&crtc->hwmode)
 	     & (DRM_SCANOUTPOS_VALID | DRM_SCANOUTPOS_IN_VBLANK)) ==
 	    (DRM_SCANOUTPOS_VALID | DRM_SCANOUTPOS_IN_VBLANK) &&
 	    (int)(work->target_vblank -
-		  amdgpu_get_vblank_counter_kms(adev->ddev, amdgpu_crtc->crtc_id)) > 0) {
+		  amdgpu_get_vblank_counter_kms(crtc)) > 0) {
 		schedule_delayed_work(&work->flip_work, usecs_to_jiffies(1000));
 		return;
 	}
@@ -108,15 +160,16 @@ static void amdgpu_flip_work_func(struct work_struct *__work)
 	spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
 
 
-	DRM_DEBUG_DRIVER("crtc:%d[%p], pflip_stat:AMDGPU_FLIP_SUBMITTED, work: %p,\n",
-					 amdgpu_crtc->crtc_id, amdgpu_crtc, work);
+	drm_dbg_vbl(adev_to_drm(adev),
+		    "crtc:%d[%p], pflip_stat:AMDGPU_FLIP_SUBMITTED, work: %p,\n",
+		    amdgpu_crtc->crtc_id, amdgpu_crtc, work);
 
 }
 
 /*
  * Handle unpin events outside the interrupt handler proper.
  */
-static void amdgpu_unpin_work_func(struct work_struct *__work)
+static void amdgpu_display_unpin_work_func(struct work_struct *__work)
 {
 	struct amdgpu_flip_work *work =
 		container_of(__work, struct amdgpu_flip_work, unpin_work);
@@ -125,10 +178,7 @@ static void amdgpu_unpin_work_func(struct work_struct *__work)
 	/* unpin of the old buffer */
 	r = amdgpu_bo_reserve(work->old_abo, true);
 	if (likely(r == 0)) {
-		r = amdgpu_bo_unpin(work->old_abo);
-		if (unlikely(r != 0)) {
-			DRM_ERROR("failed to unpin buffer after flip\n");
-		}
+		amdgpu_bo_unpin(work->old_abo);
 		amdgpu_bo_unreserve(work->old_abo);
 	} else
 		DRM_ERROR("failed to reserve buffer after flip\n");
@@ -138,31 +188,28 @@ static void amdgpu_unpin_work_func(struct work_struct *__work)
 	kfree(work);
 }
 
-int amdgpu_crtc_page_flip_target(struct drm_crtc *crtc,
-				 struct drm_framebuffer *fb,
-				 struct drm_pending_vblank_event *event,
-				 uint32_t page_flip_flags, uint32_t target,
-				 struct drm_modeset_acquire_ctx *ctx)
+int amdgpu_display_crtc_page_flip_target(struct drm_crtc *crtc,
+				struct drm_framebuffer *fb,
+				struct drm_pending_vblank_event *event,
+				uint32_t page_flip_flags, uint32_t target,
+				struct drm_modeset_acquire_ctx *ctx)
 {
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct amdgpu_framebuffer *old_amdgpu_fb;
-	struct amdgpu_framebuffer *new_amdgpu_fb;
 	struct drm_gem_object *obj;
 	struct amdgpu_flip_work *work;
 	struct amdgpu_bo *new_abo;
 	unsigned long flags;
 	u64 tiling_flags;
-	u64 base;
 	int i, r;
 
-	work = kzalloc(sizeof *work, GFP_KERNEL);
+	work = kzalloc(sizeof(*work), GFP_KERNEL);
 	if (work == NULL)
 		return -ENOMEM;
 
-	INIT_DELAYED_WORK(&work->flip_work, amdgpu_flip_work_func);
-	INIT_WORK(&work->unpin_work, amdgpu_unpin_work_func);
+	INIT_DELAYED_WORK(&work->flip_work, amdgpu_display_flip_work_func);
+	INIT_WORK(&work->unpin_work, amdgpu_display_unpin_work_func);
 
 	work->event = event;
 	work->adev = adev;
@@ -170,15 +217,13 @@ int amdgpu_crtc_page_flip_target(struct drm_crtc *crtc,
 	work->async = (page_flip_flags & DRM_MODE_PAGE_FLIP_ASYNC) != 0;
 
 	/* schedule unpin of the old buffer */
-	old_amdgpu_fb = to_amdgpu_framebuffer(crtc->primary->fb);
-	obj = old_amdgpu_fb->obj;
+	obj = crtc->primary->fb->obj[0];
 
 	/* take a reference to the old object */
 	work->old_abo = gem_to_amdgpu_bo(obj);
 	amdgpu_bo_ref(work->old_abo);
 
-	new_amdgpu_fb = to_amdgpu_framebuffer(fb);
-	obj = new_amdgpu_fb->obj;
+	obj = fb->obj[0];
 	new_abo = gem_to_amdgpu_bo(obj);
 
 	/* pin the new buffer */
@@ -188,15 +233,25 @@ int amdgpu_crtc_page_flip_target(struct drm_crtc *crtc,
 		goto cleanup;
 	}
 
-	r = amdgpu_bo_pin(new_abo, AMDGPU_GEM_DOMAIN_VRAM, &base);
+	if (!adev->enable_virtual_display) {
+		new_abo->flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+		r = amdgpu_bo_pin(new_abo,
+				  amdgpu_display_supported_domains(adev, new_abo->flags));
+		if (unlikely(r != 0)) {
+			DRM_ERROR("failed to pin new abo buffer before flip\n");
+			goto unreserve;
+		}
+	}
+
+	r = amdgpu_ttm_alloc_gart(&new_abo->tbo);
 	if (unlikely(r != 0)) {
-		DRM_ERROR("failed to pin new abo buffer before flip\n");
-		goto unreserve;
+		DRM_ERROR("%p bind failed\n", new_abo);
+		goto unpin;
 	}
 
-	r = reservation_object_get_fences_rcu(new_abo->tbo.resv, &work->excl,
-					      &work->shared_count,
-					      &work->shared);
+	r = dma_resv_get_fences(new_abo->tbo.base.resv, DMA_RESV_USAGE_WRITE,
+				&work->shared_count,
+				&work->shared);
 	if (unlikely(r != 0)) {
 		DRM_ERROR("failed to get fences for buffer\n");
 		goto unpin;
@@ -205,9 +260,10 @@ int amdgpu_crtc_page_flip_target(struct drm_crtc *crtc,
 	amdgpu_bo_get_tiling_flags(new_abo, &tiling_flags);
 	amdgpu_bo_unreserve(new_abo);
 
-	work->base = base;
-	work->target_vblank = target - drm_crtc_vblank_count(crtc) +
-		amdgpu_get_vblank_counter_kms(dev, work->crtc_id);
+	if (!adev->enable_virtual_display)
+		work->base = amdgpu_bo_gpu_offset(new_abo);
+	work->target_vblank = target - (uint32_t)drm_crtc_vblank_count(crtc) +
+		amdgpu_get_vblank_counter_kms(crtc);
 
 	/* we borrow the event spin lock for protecting flip_wrok */
 	spin_lock_irqsave(&crtc->dev->event_lock, flags);
@@ -227,7 +283,7 @@ int amdgpu_crtc_page_flip_target(struct drm_crtc *crtc,
 	/* update crtc fb */
 	crtc->primary->fb = fb;
 	spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
-	amdgpu_flip_work_func(&work->flip_work.work);
+	amdgpu_display_flip_work_func(&work->flip_work.work);
 	return 0;
 
 pflip_cleanup:
@@ -236,15 +292,14 @@ pflip_cleanup:
 		goto cleanup;
 	}
 unpin:
-	if (unlikely(amdgpu_bo_unpin(new_abo) != 0)) {
-		DRM_ERROR("failed to unpin new abo in error path\n");
-	}
+	if (!adev->enable_virtual_display)
+		amdgpu_bo_unpin(new_abo);
+
 unreserve:
 	amdgpu_bo_unreserve(new_abo);
 
 cleanup:
 	amdgpu_bo_unref(&work->old_abo);
-	dma_fence_put(work->excl);
 	for (i = 0; i < work->shared_count; ++i)
 		dma_fence_put(work->shared[i]);
 	kfree(work->shared);
@@ -253,8 +308,8 @@ cleanup:
 	return r;
 }
 
-int amdgpu_crtc_set_config(struct drm_mode_set *set,
-			   struct drm_modeset_acquire_ctx *ctx)
+int amdgpu_display_crtc_set_config(struct drm_mode_set *set,
+				   struct drm_modeset_acquire_ctx *ctx)
 {
 	struct drm_device *dev;
 	struct amdgpu_device *adev;
@@ -269,7 +324,7 @@ int amdgpu_crtc_set_config(struct drm_mode_set *set,
 
 	ret = pm_runtime_get_sync(dev->dev);
 	if (ret < 0)
-		return ret;
+		goto out;
 
 	ret = drm_crtc_helper_set_config(set, ctx);
 
@@ -279,20 +334,20 @@ int amdgpu_crtc_set_config(struct drm_mode_set *set,
 
 	pm_runtime_mark_last_busy(dev->dev);
 
-	adev = dev->dev_private;
+	adev = drm_to_adev(dev);
 	/* if we have active crtcs and we don't have a power ref,
-	   take the current one */
+	 * take the current one
+	 */
 	if (active && !adev->have_disp_power_ref) {
 		adev->have_disp_power_ref = true;
 		return ret;
 	}
-	/* if we have no active crtcs, then drop the power ref
-	   we got before */
-	if (!active && adev->have_disp_power_ref) {
-		pm_runtime_put_autosuspend(dev->dev);
+	/* if we have no active crtcs, then go to
+	 * drop the power ref we got before
+	 */
+	if (!active && adev->have_disp_power_ref)
 		adev->have_disp_power_ref = false;
-	}
-
+out:
 	/* drop the power reference we got coming in here */
 	pm_runtime_put_autosuspend(dev->dev);
 	return ret;
@@ -351,17 +406,19 @@ static const char *hpd_names[6] = {
 	"HPD6",
 };
 
-void amdgpu_print_display_setup(struct drm_device *dev)
+void amdgpu_display_print_display_setup(struct drm_device *dev)
 {
 	struct drm_connector *connector;
 	struct amdgpu_connector *amdgpu_connector;
 	struct drm_encoder *encoder;
 	struct amdgpu_encoder *amdgpu_encoder;
+	struct drm_connector_list_iter iter;
 	uint32_t devices;
 	int i = 0;
 
+	drm_connector_list_iter_begin(dev, &iter);
 	DRM_INFO("AMDGPU Display Connectors\n");
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_for_each_connector_iter(connector, &iter) {
 		amdgpu_connector = to_amdgpu_connector(connector);
 		DRM_INFO("Connector %d:\n", i);
 		DRM_INFO("  %s\n", connector->name);
@@ -425,14 +482,11 @@ void amdgpu_print_display_setup(struct drm_device *dev)
 		}
 		i++;
 	}
+	drm_connector_list_iter_end(&iter);
 }
 
-/**
- * amdgpu_ddc_probe
- *
- */
-bool amdgpu_ddc_probe(struct amdgpu_connector *amdgpu_connector,
-		       bool use_aux)
+bool amdgpu_display_ddc_probe(struct amdgpu_connector *amdgpu_connector,
+			      bool use_aux)
 {
 	u8 out = 0x0;
 	u8 buf[8];
@@ -456,11 +510,10 @@ bool amdgpu_ddc_probe(struct amdgpu_connector *amdgpu_connector,
 	if (amdgpu_connector->router.ddc_valid)
 		amdgpu_i2c_router_select_ddc_port(amdgpu_connector);
 
-	if (use_aux) {
+	if (use_aux)
 		ret = i2c_transfer(&amdgpu_connector->ddc_bus->aux.ddc, msgs, 2);
-	} else {
+	else
 		ret = i2c_transfer(&amdgpu_connector->ddc_bus->adapter, msgs, 2);
-	}
 
 	if (ret != 2)
 		/* Couldn't find an accessible DDC on this connector */
@@ -469,172 +522,899 @@ bool amdgpu_ddc_probe(struct amdgpu_connector *amdgpu_connector,
 	 * EDID header starts with:
 	 * 0x00,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0x00.
 	 * Only the first 6 bytes must be valid as
-	 * drm_edid_block_valid() can fix the last 2 bytes */
+	 * drm_edid_block_valid() can fix the last 2 bytes
+	 */
 	if (drm_edid_header_is_valid(buf) < 6) {
 		/* Couldn't find an accessible EDID on this
-		 * connector */
+		 * connector
+		 */
 		return false;
 	}
 	return true;
 }
 
-static void amdgpu_user_framebuffer_destroy(struct drm_framebuffer *fb)
+static int amdgpu_dirtyfb(struct drm_framebuffer *fb, struct drm_file *file,
+			  unsigned int flags, unsigned int color,
+			  struct drm_clip_rect *clips, unsigned int num_clips)
 {
-	struct amdgpu_framebuffer *amdgpu_fb = to_amdgpu_framebuffer(fb);
 
-	drm_gem_object_put_unlocked(amdgpu_fb->obj);
-	drm_framebuffer_cleanup(fb);
-	kfree(amdgpu_fb);
+	if (file)
+		return -ENOSYS;
+
+	return drm_atomic_helper_dirtyfb(fb, file, flags, color, clips,
+					 num_clips);
 }
 
-static int amdgpu_user_framebuffer_create_handle(struct drm_framebuffer *fb,
-						  struct drm_file *file_priv,
-						  unsigned int *handle)
-{
-	struct amdgpu_framebuffer *amdgpu_fb = to_amdgpu_framebuffer(fb);
+static const struct drm_framebuffer_funcs amdgpu_fb_funcs = {
+	.destroy = drm_gem_fb_destroy,
+	.create_handle = drm_gem_fb_create_handle,
+};
 
-	return drm_gem_handle_create(file_priv, amdgpu_fb->obj, handle);
+static const struct drm_framebuffer_funcs amdgpu_fb_funcs_atomic = {
+	.destroy = drm_gem_fb_destroy,
+	.create_handle = drm_gem_fb_create_handle,
+	.dirty = amdgpu_dirtyfb
+};
+
+uint32_t amdgpu_display_supported_domains(struct amdgpu_device *adev,
+					  uint64_t bo_flags)
+{
+	uint32_t domain = AMDGPU_GEM_DOMAIN_VRAM;
+
+#if defined(CONFIG_DRM_AMD_DC)
+	/*
+	 * if amdgpu_bo_support_uswc returns false it means that USWC mappings
+	 * is not supported for this board. But this mapping is required
+	 * to avoid hang caused by placement of scanout BO in GTT on certain
+	 * APUs. So force the BO placement to VRAM in case this architecture
+	 * will not allow USWC mappings.
+	 * Also, don't allow GTT domain if the BO doesn't have USWC flag set.
+	 */
+	if ((bo_flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC) &&
+	    amdgpu_bo_support_uswc(bo_flags) &&
+	    adev->dc_enabled &&
+	    adev->mode_info.gpu_vm_support)
+		domain |= AMDGPU_GEM_DOMAIN_GTT;
+#endif
+
+	return domain;
 }
 
-static const struct drm_framebuffer_funcs amdgpu_fb_funcs = {
-	.destroy = amdgpu_user_framebuffer_destroy,
-	.create_handle = amdgpu_user_framebuffer_create_handle,
+static const struct drm_format_info dcc_formats[] = {
+	{ .format = DRM_FORMAT_XRGB8888, .depth = 24, .num_planes = 2,
+	  .cpp = { 4, 0, }, .block_w = {1, 1, 1}, .block_h = {1, 1, 1}, .hsub = 1, .vsub = 1, },
+	 { .format = DRM_FORMAT_XBGR8888, .depth = 24, .num_planes = 2,
+	  .cpp = { 4, 0, }, .block_w = {1, 1, 1}, .block_h = {1, 1, 1}, .hsub = 1, .vsub = 1, },
+	{ .format = DRM_FORMAT_ARGB8888, .depth = 32, .num_planes = 2,
+	  .cpp = { 4, 0, }, .block_w = {1, 1, 1}, .block_h = {1, 1, 1}, .hsub = 1, .vsub = 1,
+	   .has_alpha = true, },
+	{ .format = DRM_FORMAT_ABGR8888, .depth = 32, .num_planes = 2,
+	  .cpp = { 4, 0, }, .block_w = {1, 1, 1}, .block_h = {1, 1, 1}, .hsub = 1, .vsub = 1,
+	  .has_alpha = true, },
+	{ .format = DRM_FORMAT_BGRA8888, .depth = 32, .num_planes = 2,
+	  .cpp = { 4, 0, }, .block_w = {1, 1, 1}, .block_h = {1, 1, 1}, .hsub = 1, .vsub = 1,
+	  .has_alpha = true, },
+	{ .format = DRM_FORMAT_XRGB2101010, .depth = 30, .num_planes = 2,
+	  .cpp = { 4, 0, }, .block_w = {1, 1, 1}, .block_h = {1, 1, 1}, .hsub = 1, .vsub = 1, },
+	{ .format = DRM_FORMAT_XBGR2101010, .depth = 30, .num_planes = 2,
+	  .cpp = { 4, 0, }, .block_w = {1, 1, 1}, .block_h = {1, 1, 1}, .hsub = 1, .vsub = 1, },
+	{ .format = DRM_FORMAT_ARGB2101010, .depth = 30, .num_planes = 2,
+	  .cpp = { 4, 0, }, .block_w = {1, 1, 1}, .block_h = {1, 1, 1}, .hsub = 1, .vsub = 1,
+	  .has_alpha = true, },
+	{ .format = DRM_FORMAT_ABGR2101010, .depth = 30, .num_planes = 2,
+	  .cpp = { 4, 0, }, .block_w = {1, 1, 1}, .block_h = {1, 1, 1}, .hsub = 1, .vsub = 1,
+	  .has_alpha = true, },
+	{ .format = DRM_FORMAT_RGB565, .depth = 16, .num_planes = 2,
+	  .cpp = { 2, 0, }, .block_w = {1, 1, 1}, .block_h = {1, 1, 1}, .hsub = 1, .vsub = 1, },
+};
+
+static const struct drm_format_info dcc_retile_formats[] = {
+	{ .format = DRM_FORMAT_XRGB8888, .depth = 24, .num_planes = 3,
+	  .cpp = { 4, 0, 0 }, .block_w = {1, 1, 1}, .block_h = {1, 1, 1}, .hsub = 1, .vsub = 1, },
+	 { .format = DRM_FORMAT_XBGR8888, .depth = 24, .num_planes = 3,
+	  .cpp = { 4, 0, 0 }, .block_w = {1, 1, 1}, .block_h = {1, 1, 1}, .hsub = 1, .vsub = 1, },
+	{ .format = DRM_FORMAT_ARGB8888, .depth = 32, .num_planes = 3,
+	  .cpp = { 4, 0, 0 }, .block_w = {1, 1, 1}, .block_h = {1, 1, 1}, .hsub = 1, .vsub = 1,
+	   .has_alpha = true, },
+	{ .format = DRM_FORMAT_ABGR8888, .depth = 32, .num_planes = 3,
+	  .cpp = { 4, 0, 0 }, .block_w = {1, 1, 1}, .block_h = {1, 1, 1}, .hsub = 1, .vsub = 1,
+	  .has_alpha = true, },
+	{ .format = DRM_FORMAT_BGRA8888, .depth = 32, .num_planes = 3,
+	  .cpp = { 4, 0, 0 }, .block_w = {1, 1, 1}, .block_h = {1, 1, 1}, .hsub = 1, .vsub = 1,
+	  .has_alpha = true, },
+	{ .format = DRM_FORMAT_XRGB2101010, .depth = 30, .num_planes = 3,
+	  .cpp = { 4, 0, 0 }, .block_w = {1, 1, 1}, .block_h = {1, 1, 1}, .hsub = 1, .vsub = 1, },
+	{ .format = DRM_FORMAT_XBGR2101010, .depth = 30, .num_planes = 3,
+	  .cpp = { 4, 0, 0 }, .block_w = {1, 1, 1}, .block_h = {1, 1, 1}, .hsub = 1, .vsub = 1, },
+	{ .format = DRM_FORMAT_ARGB2101010, .depth = 30, .num_planes = 3,
+	  .cpp = { 4, 0, 0 }, .block_w = {1, 1, 1}, .block_h = {1, 1, 1}, .hsub = 1, .vsub = 1,
+	  .has_alpha = true, },
+	{ .format = DRM_FORMAT_ABGR2101010, .depth = 30, .num_planes = 3,
+	  .cpp = { 4, 0, 0 }, .block_w = {1, 1, 1}, .block_h = {1, 1, 1}, .hsub = 1, .vsub = 1,
+	  .has_alpha = true, },
+	{ .format = DRM_FORMAT_RGB565, .depth = 16, .num_planes = 3,
+	  .cpp = { 2, 0, 0 }, .block_w = {1, 1, 1}, .block_h = {1, 1, 1}, .hsub = 1, .vsub = 1, },
 };
 
-int
-amdgpu_framebuffer_init(struct drm_device *dev,
-			struct amdgpu_framebuffer *rfb,
-			const struct drm_mode_fb_cmd2 *mode_cmd,
-			struct drm_gem_object *obj)
+static const struct drm_format_info *
+lookup_format_info(const struct drm_format_info formats[],
+		  int num_formats, u32 format)
+{
+	int i;
+
+	for (i = 0; i < num_formats; i++) {
+		if (formats[i].format == format)
+			return &formats[i];
+	}
+
+	return NULL;
+}
+
+const struct drm_format_info *
+amdgpu_lookup_format_info(u32 format, uint64_t modifier)
+{
+	if (!IS_AMD_FMT_MOD(modifier))
+		return NULL;
+
+	if (AMD_FMT_MOD_GET(TILE_VERSION, modifier) < AMD_FMT_MOD_TILE_VER_GFX9 ||
+	    AMD_FMT_MOD_GET(TILE_VERSION, modifier) >= AMD_FMT_MOD_TILE_VER_GFX12)
+		return NULL;
+
+	if (AMD_FMT_MOD_GET(DCC_RETILE, modifier))
+		return lookup_format_info(dcc_retile_formats,
+					  ARRAY_SIZE(dcc_retile_formats),
+					  format);
+
+	if (AMD_FMT_MOD_GET(DCC, modifier))
+		return lookup_format_info(dcc_formats, ARRAY_SIZE(dcc_formats),
+					  format);
+
+	/* returning NULL will cause the default format structs to be used. */
+	return NULL;
+}
+
+
+/*
+ * Tries to extract the renderable DCC offset from the opaque metadata attached
+ * to the buffer.
+ */
+static int
+extract_render_dcc_offset(struct amdgpu_device *adev,
+			  struct drm_gem_object *obj,
+			  uint64_t *offset)
+{
+	struct amdgpu_bo *rbo;
+	int r = 0;
+	uint32_t metadata[10]; /* Something that fits a descriptor + header. */
+	uint32_t size;
+
+	rbo = gem_to_amdgpu_bo(obj);
+	r = amdgpu_bo_reserve(rbo, false);
+
+	if (unlikely(r)) {
+		/* Don't show error message when returning -ERESTARTSYS */
+		if (r != -ERESTARTSYS)
+			DRM_ERROR("Unable to reserve buffer: %d\n", r);
+		return r;
+	}
+
+	r = amdgpu_bo_get_metadata(rbo, metadata, sizeof(metadata), &size, NULL);
+	amdgpu_bo_unreserve(rbo);
+
+	if (r)
+		return r;
+
+	/*
+	 * The first word is the metadata version, and we need space for at least
+	 * the version + pci vendor+device id + 8 words for a descriptor.
+	 */
+	if (size < 40  || metadata[0] != 1)
+		return -EINVAL;
+
+	if (adev->family >= AMDGPU_FAMILY_NV) {
+		/* resource word 6/7 META_DATA_ADDRESS{_LO} */
+		*offset = ((u64)metadata[9] << 16u) |
+			  ((metadata[8] & 0xFF000000u) >> 16);
+	} else {
+		/* resource word 5/7 META_DATA_ADDRESS */
+		*offset = ((u64)metadata[9] << 8u) |
+			  ((u64)(metadata[7] & 0x1FE0000u) << 23);
+	}
+
+	return 0;
+}
+
+static int convert_tiling_flags_to_modifier_gfx12(struct amdgpu_framebuffer *afb)
+{
+	u64 modifier = 0;
+	int swizzle_mode = AMDGPU_TILING_GET(afb->tiling_flags, GFX12_SWIZZLE_MODE);
+
+	if (!swizzle_mode) {
+		modifier = DRM_FORMAT_MOD_LINEAR;
+	} else {
+		int max_comp_block =
+			AMDGPU_TILING_GET(afb->tiling_flags, GFX12_DCC_MAX_COMPRESSED_BLOCK);
+
+		modifier =
+			AMD_FMT_MOD |
+			AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX12) |
+			AMD_FMT_MOD_SET(TILE, swizzle_mode) |
+			AMD_FMT_MOD_SET(DCC, afb->gfx12_dcc) |
+			AMD_FMT_MOD_SET(DCC_MAX_COMPRESSED_BLOCK, max_comp_block);
+	}
+
+	afb->base.modifier = modifier;
+	afb->base.flags |= DRM_MODE_FB_MODIFIERS;
+	return 0;
+}
+
+static int convert_tiling_flags_to_modifier(struct amdgpu_framebuffer *afb)
+{
+	struct amdgpu_device *adev = drm_to_adev(afb->base.dev);
+	uint64_t modifier = 0;
+	int num_pipes = 0;
+	int num_pkrs = 0;
+
+	num_pkrs = adev->gfx.config.gb_addr_config_fields.num_pkrs;
+	num_pipes = adev->gfx.config.gb_addr_config_fields.num_pipes;
+
+	if (!afb->tiling_flags || !AMDGPU_TILING_GET(afb->tiling_flags, SWIZZLE_MODE)) {
+		modifier = DRM_FORMAT_MOD_LINEAR;
+	} else {
+		int swizzle = AMDGPU_TILING_GET(afb->tiling_flags, SWIZZLE_MODE);
+		bool has_xor = swizzle >= 16;
+		int block_size_bits;
+		int version;
+		int pipe_xor_bits = 0;
+		int bank_xor_bits = 0;
+		int packers = 0;
+		int rb = 0;
+		int pipes = ilog2(num_pipes);
+		uint32_t dcc_offset = AMDGPU_TILING_GET(afb->tiling_flags, DCC_OFFSET_256B);
+
+		switch (swizzle >> 2) {
+		case 0: /* 256B */
+			block_size_bits = 8;
+			break;
+		case 1: /* 4KiB */
+		case 5: /* 4KiB _X */
+			block_size_bits = 12;
+			break;
+		case 2: /* 64KiB */
+		case 4: /* 64 KiB _T */
+		case 6: /* 64 KiB _X */
+			block_size_bits = 16;
+			break;
+		case 7: /* 256 KiB */
+			block_size_bits = 18;
+			break;
+		default:
+			/* RESERVED or VAR */
+			return -EINVAL;
+		}
+
+		if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(11, 0, 0))
+			version = AMD_FMT_MOD_TILE_VER_GFX11;
+		else if (amdgpu_ip_version(adev, GC_HWIP, 0) >=
+			 IP_VERSION(10, 3, 0))
+			version = AMD_FMT_MOD_TILE_VER_GFX10_RBPLUS;
+		else if (amdgpu_ip_version(adev, GC_HWIP, 0) >=
+			 IP_VERSION(10, 0, 0))
+			version = AMD_FMT_MOD_TILE_VER_GFX10;
+		else
+			version = AMD_FMT_MOD_TILE_VER_GFX9;
+
+		switch (swizzle & 3) {
+		case 0: /* Z microtiling */
+			return -EINVAL;
+		case 1: /* S microtiling */
+			if (amdgpu_ip_version(adev, GC_HWIP, 0) <
+			    IP_VERSION(11, 0, 0)) {
+				if (!has_xor)
+					version = AMD_FMT_MOD_TILE_VER_GFX9;
+			}
+			break;
+		case 2:
+			if (amdgpu_ip_version(adev, GC_HWIP, 0) <
+			    IP_VERSION(11, 0, 0)) {
+				if (!has_xor && afb->base.format->cpp[0] != 4)
+					version = AMD_FMT_MOD_TILE_VER_GFX9;
+			}
+			break;
+		case 3:
+			break;
+		}
+
+		if (has_xor) {
+			if (num_pipes == num_pkrs && num_pkrs == 0) {
+				DRM_ERROR("invalid number of pipes and packers\n");
+				return -EINVAL;
+			}
+
+			switch (version) {
+			case AMD_FMT_MOD_TILE_VER_GFX11:
+				pipe_xor_bits = min(block_size_bits - 8, pipes);
+				packers = ilog2(adev->gfx.config.gb_addr_config_fields.num_pkrs);
+				break;
+			case AMD_FMT_MOD_TILE_VER_GFX10_RBPLUS:
+				pipe_xor_bits = min(block_size_bits - 8, pipes);
+				packers = min(block_size_bits - 8 - pipe_xor_bits,
+					      ilog2(adev->gfx.config.gb_addr_config_fields.num_pkrs));
+				break;
+			case AMD_FMT_MOD_TILE_VER_GFX10:
+				pipe_xor_bits = min(block_size_bits - 8, pipes);
+				break;
+			case AMD_FMT_MOD_TILE_VER_GFX9:
+				rb = ilog2(adev->gfx.config.gb_addr_config_fields.num_se) +
+				     ilog2(adev->gfx.config.gb_addr_config_fields.num_rb_per_se);
+				pipe_xor_bits = min(block_size_bits - 8, pipes +
+						    ilog2(adev->gfx.config.gb_addr_config_fields.num_se));
+				bank_xor_bits = min(block_size_bits - 8 - pipe_xor_bits,
+						    ilog2(adev->gfx.config.gb_addr_config_fields.num_banks));
+				break;
+			}
+		}
+
+		modifier = AMD_FMT_MOD |
+			   AMD_FMT_MOD_SET(TILE, AMDGPU_TILING_GET(afb->tiling_flags, SWIZZLE_MODE)) |
+			   AMD_FMT_MOD_SET(TILE_VERSION, version) |
+			   AMD_FMT_MOD_SET(PIPE_XOR_BITS, pipe_xor_bits) |
+			   AMD_FMT_MOD_SET(BANK_XOR_BITS, bank_xor_bits) |
+			   AMD_FMT_MOD_SET(PACKERS, packers);
+
+		if (dcc_offset != 0) {
+			bool dcc_i64b = AMDGPU_TILING_GET(afb->tiling_flags, DCC_INDEPENDENT_64B) != 0;
+			bool dcc_i128b = version >= AMD_FMT_MOD_TILE_VER_GFX10_RBPLUS;
+			const struct drm_format_info *format_info;
+			u64 render_dcc_offset;
+
+			/* Enable constant encode on RAVEN2 and later. */
+			bool dcc_constant_encode =
+				(adev->asic_type > CHIP_RAVEN ||
+				 (adev->asic_type == CHIP_RAVEN &&
+				  adev->external_rev_id >= 0x81)) &&
+				amdgpu_ip_version(adev, GC_HWIP, 0) <
+					IP_VERSION(11, 0, 0);
+
+			int max_cblock_size = dcc_i64b ? AMD_FMT_MOD_DCC_BLOCK_64B :
+					      dcc_i128b ? AMD_FMT_MOD_DCC_BLOCK_128B :
+					      AMD_FMT_MOD_DCC_BLOCK_256B;
+
+			modifier |= AMD_FMT_MOD_SET(DCC, 1) |
+				    AMD_FMT_MOD_SET(DCC_CONSTANT_ENCODE, dcc_constant_encode) |
+				    AMD_FMT_MOD_SET(DCC_INDEPENDENT_64B, dcc_i64b) |
+				    AMD_FMT_MOD_SET(DCC_INDEPENDENT_128B, dcc_i128b) |
+				    AMD_FMT_MOD_SET(DCC_MAX_COMPRESSED_BLOCK, max_cblock_size);
+
+			afb->base.offsets[1] = dcc_offset * 256 + afb->base.offsets[0];
+			afb->base.pitches[1] =
+				AMDGPU_TILING_GET(afb->tiling_flags, DCC_PITCH_MAX) + 1;
+
+			/*
+			 * If the userspace driver uses retiling the tiling flags do not contain
+			 * info on the renderable DCC buffer. Luckily the opaque metadata contains
+			 * the info so we can try to extract it. The kernel does not use this info
+			 * but we should convert it to a modifier plane for getfb2, so the
+			 * userspace driver that gets it doesn't have to juggle around another DCC
+			 * plane internally.
+			 */
+			if (extract_render_dcc_offset(adev, afb->base.obj[0],
+						      &render_dcc_offset) == 0 &&
+			    render_dcc_offset != 0 &&
+			    render_dcc_offset != afb->base.offsets[1] &&
+			    render_dcc_offset < UINT_MAX) {
+				uint32_t dcc_block_bits;  /* of base surface data */
+
+				modifier |= AMD_FMT_MOD_SET(DCC_RETILE, 1);
+				afb->base.offsets[2] = render_dcc_offset;
+
+				if (adev->family >= AMDGPU_FAMILY_NV) {
+					int extra_pipe = 0;
+
+					if ((amdgpu_ip_version(adev, GC_HWIP,
+							       0) >=
+					     IP_VERSION(10, 3, 0)) &&
+					    pipes == packers && pipes > 1)
+						extra_pipe = 1;
+
+					dcc_block_bits = max(20, 16 + pipes + extra_pipe);
+				} else {
+					modifier |= AMD_FMT_MOD_SET(RB, rb) |
+						    AMD_FMT_MOD_SET(PIPE, pipes);
+					dcc_block_bits = max(20, 18 + rb);
+				}
+
+				dcc_block_bits -= ilog2(afb->base.format->cpp[0]);
+				afb->base.pitches[2] = ALIGN(afb->base.width,
+							     1u << ((dcc_block_bits + 1) / 2));
+			}
+			format_info = amdgpu_lookup_format_info(afb->base.format->format,
+								modifier);
+			if (!format_info)
+				return -EINVAL;
+
+			afb->base.format = format_info;
+		}
+	}
+
+	afb->base.modifier = modifier;
+	afb->base.flags |= DRM_MODE_FB_MODIFIERS;
+	return 0;
+}
+
+/* Mirrors the is_displayable check in radeonsi's gfx6_compute_surface */
+static int check_tiling_flags_gfx6(struct amdgpu_framebuffer *afb)
+{
+	u64 micro_tile_mode;
+
+	if (AMDGPU_TILING_GET(afb->tiling_flags, ARRAY_MODE) == 1) /* LINEAR_ALIGNED */
+		return 0;
+
+	micro_tile_mode = AMDGPU_TILING_GET(afb->tiling_flags, MICRO_TILE_MODE);
+	switch (micro_tile_mode) {
+	case 0: /* DISPLAY */
+	case 3: /* RENDER */
+		return 0;
+	default:
+		drm_dbg_kms(afb->base.dev,
+			    "Micro tile mode %llu not supported for scanout\n",
+			    micro_tile_mode);
+		return -EINVAL;
+	}
+}
+
+static void get_block_dimensions(unsigned int block_log2, unsigned int cpp,
+				 unsigned int *width, unsigned int *height)
+{
+	unsigned int cpp_log2 = ilog2(cpp);
+	unsigned int pixel_log2 = block_log2 - cpp_log2;
+	unsigned int width_log2 = (pixel_log2 + 1) / 2;
+	unsigned int height_log2 = pixel_log2 - width_log2;
+
+	*width = 1 << width_log2;
+	*height = 1 << height_log2;
+}
+
+static unsigned int get_dcc_block_size(uint64_t modifier, bool rb_aligned,
+				       bool pipe_aligned)
+{
+	unsigned int ver = AMD_FMT_MOD_GET(TILE_VERSION, modifier);
+
+	switch (ver) {
+	case AMD_FMT_MOD_TILE_VER_GFX9: {
+		/*
+		 * TODO: for pipe aligned we may need to check the alignment of the
+		 * total size of the surface, which may need to be bigger than the
+		 * natural alignment due to some HW workarounds
+		 */
+		return max(10 + (rb_aligned ? (int)AMD_FMT_MOD_GET(RB, modifier) : 0), 12);
+	}
+	case AMD_FMT_MOD_TILE_VER_GFX10:
+	case AMD_FMT_MOD_TILE_VER_GFX10_RBPLUS:
+	case AMD_FMT_MOD_TILE_VER_GFX11: {
+		int pipes_log2 = AMD_FMT_MOD_GET(PIPE_XOR_BITS, modifier);
+
+		if (ver >= AMD_FMT_MOD_TILE_VER_GFX10_RBPLUS && pipes_log2 > 1 &&
+		    AMD_FMT_MOD_GET(PACKERS, modifier) == pipes_log2)
+			++pipes_log2;
+
+		return max(8 + (pipe_aligned ? pipes_log2 : 0), 12);
+	}
+	default:
+		return 0;
+	}
+}
+
+static int amdgpu_display_verify_plane(struct amdgpu_framebuffer *rfb, int plane,
+				       const struct drm_format_info *format,
+				       unsigned int block_width, unsigned int block_height,
+				       unsigned int block_size_log2)
+{
+	unsigned int width = rfb->base.width /
+		((plane && plane < format->num_planes) ? format->hsub : 1);
+	unsigned int height = rfb->base.height /
+		((plane && plane < format->num_planes) ? format->vsub : 1);
+	unsigned int cpp = plane < format->num_planes ? format->cpp[plane] : 1;
+	unsigned int block_pitch = block_width * cpp;
+	unsigned int min_pitch = ALIGN(width * cpp, block_pitch);
+	unsigned int block_size = 1 << block_size_log2;
+	uint64_t size;
+
+	if (rfb->base.pitches[plane] % block_pitch) {
+		drm_dbg_kms(rfb->base.dev,
+			    "pitch %d for plane %d is not a multiple of block pitch %d\n",
+			    rfb->base.pitches[plane], plane, block_pitch);
+		return -EINVAL;
+	}
+	if (rfb->base.pitches[plane] < min_pitch) {
+		drm_dbg_kms(rfb->base.dev,
+			    "pitch %d for plane %d is less than minimum pitch %d\n",
+			    rfb->base.pitches[plane], plane, min_pitch);
+		return -EINVAL;
+	}
+
+	/* Force at least natural alignment. */
+	if (rfb->base.offsets[plane] % block_size) {
+		drm_dbg_kms(rfb->base.dev,
+			    "offset 0x%x for plane %d is not a multiple of block pitch 0x%x\n",
+			    rfb->base.offsets[plane], plane, block_size);
+		return -EINVAL;
+	}
+
+	size = rfb->base.offsets[plane] +
+		(uint64_t)rfb->base.pitches[plane] / block_pitch *
+		block_size * DIV_ROUND_UP(height, block_height);
+
+	if (rfb->base.obj[0]->size < size) {
+		drm_dbg_kms(rfb->base.dev,
+			    "BO size 0x%zx is less than 0x%llx required for plane %d\n",
+			    rfb->base.obj[0]->size, size, plane);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+
+static int amdgpu_display_verify_sizes(struct amdgpu_framebuffer *rfb)
 {
+	const struct drm_format_info *format_info = drm_format_info(rfb->base.format->format);
+	uint64_t modifier = rfb->base.modifier;
 	int ret;
-	rfb->obj = obj;
-	drm_helper_mode_fill_fb_struct(dev, &rfb->base, mode_cmd);
-	ret = drm_framebuffer_init(dev, &rfb->base, &amdgpu_fb_funcs);
-	if (ret) {
-		rfb->obj = NULL;
+	unsigned int i, block_width, block_height, block_size_log2;
+
+	if (rfb->base.dev->mode_config.fb_modifiers_not_supported)
+		return 0;
+
+	for (i = 0; i < format_info->num_planes; ++i) {
+		if (modifier == DRM_FORMAT_MOD_LINEAR) {
+			block_width = 256 / format_info->cpp[i];
+			block_height = 1;
+			block_size_log2 = 8;
+		} else if (AMD_FMT_MOD_GET(TILE_VERSION, modifier) >= AMD_FMT_MOD_TILE_VER_GFX12) {
+			int swizzle = AMD_FMT_MOD_GET(TILE, modifier);
+
+			switch (swizzle) {
+			case AMD_FMT_MOD_TILE_GFX12_256B_2D:
+				block_size_log2 = 8;
+				break;
+			case AMD_FMT_MOD_TILE_GFX12_4K_2D:
+				block_size_log2 = 12;
+				break;
+			case AMD_FMT_MOD_TILE_GFX12_64K_2D:
+				block_size_log2 = 16;
+				break;
+			case AMD_FMT_MOD_TILE_GFX12_256K_2D:
+				block_size_log2 = 18;
+				break;
+			default:
+				drm_dbg_kms(rfb->base.dev,
+					    "Gfx12 swizzle mode with unknown block size: %d\n", swizzle);
+				return -EINVAL;
+			}
+
+			get_block_dimensions(block_size_log2, format_info->cpp[i],
+					     &block_width, &block_height);
+		} else {
+			int swizzle = AMD_FMT_MOD_GET(TILE, modifier);
+
+			switch ((swizzle & ~3) + 1) {
+			case DC_SW_256B_S:
+				block_size_log2 = 8;
+				break;
+			case DC_SW_4KB_S:
+			case DC_SW_4KB_S_X:
+				block_size_log2 = 12;
+				break;
+			case DC_SW_64KB_S:
+			case DC_SW_64KB_S_T:
+			case DC_SW_64KB_S_X:
+				block_size_log2 = 16;
+				break;
+			case DC_SW_VAR_S_X:
+				block_size_log2 = 18;
+				break;
+			default:
+				drm_dbg_kms(rfb->base.dev,
+					    "Swizzle mode with unknown block size: %d\n", swizzle);
+				return -EINVAL;
+			}
+
+			get_block_dimensions(block_size_log2, format_info->cpp[i],
+					     &block_width, &block_height);
+		}
+
+		ret = amdgpu_display_verify_plane(rfb, i, format_info,
+						  block_width, block_height, block_size_log2);
+		if (ret)
+			return ret;
+	}
+
+	if (AMD_FMT_MOD_GET(TILE_VERSION, modifier) <= AMD_FMT_MOD_TILE_VER_GFX11 &&
+	    AMD_FMT_MOD_GET(DCC, modifier)) {
+		if (AMD_FMT_MOD_GET(DCC_RETILE, modifier)) {
+			block_size_log2 = get_dcc_block_size(modifier, false, false);
+			get_block_dimensions(block_size_log2 + 8, format_info->cpp[0],
+					     &block_width, &block_height);
+			ret = amdgpu_display_verify_plane(rfb, i, format_info,
+							  block_width, block_height,
+							  block_size_log2);
+			if (ret)
+				return ret;
+
+			++i;
+			block_size_log2 = get_dcc_block_size(modifier, true, true);
+		} else {
+			bool pipe_aligned = AMD_FMT_MOD_GET(DCC_PIPE_ALIGN, modifier);
+
+			block_size_log2 = get_dcc_block_size(modifier, true, pipe_aligned);
+		}
+		get_block_dimensions(block_size_log2 + 8, format_info->cpp[0],
+				     &block_width, &block_height);
+		ret = amdgpu_display_verify_plane(rfb, i, format_info,
+						  block_width, block_height, block_size_log2);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int amdgpu_display_get_fb_info(const struct amdgpu_framebuffer *amdgpu_fb,
+				      uint64_t *tiling_flags, bool *tmz_surface,
+				      bool *gfx12_dcc)
+{
+	struct amdgpu_bo *rbo;
+	int r;
+
+	if (!amdgpu_fb) {
+		*tiling_flags = 0;
+		*tmz_surface = false;
+		*gfx12_dcc = false;
+		return 0;
+	}
+
+	rbo = gem_to_amdgpu_bo(amdgpu_fb->base.obj[0]);
+	r = amdgpu_bo_reserve(rbo, false);
+
+	if (unlikely(r)) {
+		/* Don't show error message when returning -ERESTARTSYS */
+		if (r != -ERESTARTSYS)
+			DRM_ERROR("Unable to reserve buffer: %d\n", r);
+		return r;
+	}
+
+	amdgpu_bo_get_tiling_flags(rbo, tiling_flags);
+	*tmz_surface = amdgpu_bo_encrypted(rbo);
+	*gfx12_dcc = rbo->flags & AMDGPU_GEM_CREATE_GFX12_DCC;
+
+	amdgpu_bo_unreserve(rbo);
+
+	return r;
+}
+
+static int amdgpu_display_gem_fb_verify_and_init(struct drm_device *dev,
+						 struct amdgpu_framebuffer *rfb,
+						 struct drm_file *file_priv,
+						 const struct drm_format_info *info,
+						 const struct drm_mode_fb_cmd2 *mode_cmd,
+						 struct drm_gem_object *obj)
+{
+	int ret;
+
+	rfb->base.obj[0] = obj;
+	drm_helper_mode_fill_fb_struct(dev, &rfb->base, info, mode_cmd);
+	/* Verify that the modifier is supported. */
+	if (!drm_any_plane_has_format(dev, mode_cmd->pixel_format,
+				      mode_cmd->modifier[0])) {
+		drm_dbg_kms(dev,
+			    "unsupported pixel format %p4cc / modifier 0x%llx\n",
+			    &mode_cmd->pixel_format, mode_cmd->modifier[0]);
+
+		ret = -EINVAL;
+		goto err;
+	}
+
+	ret = amdgpu_display_framebuffer_init(dev, rfb, mode_cmd, obj);
+	if (ret)
+		goto err;
+
+	if (drm_drv_uses_atomic_modeset(dev))
+		ret = drm_framebuffer_init(dev, &rfb->base,
+					   &amdgpu_fb_funcs_atomic);
+	else
+		ret = drm_framebuffer_init(dev, &rfb->base, &amdgpu_fb_funcs);
+
+	if (ret)
+		goto err;
+
+	return 0;
+err:
+	drm_dbg_kms(dev, "Failed to verify and init gem fb: %d\n", ret);
+	rfb->base.obj[0] = NULL;
+	return ret;
+}
+
+static int amdgpu_display_framebuffer_init(struct drm_device *dev,
+					   struct amdgpu_framebuffer *rfb,
+					   const struct drm_mode_fb_cmd2 *mode_cmd,
+					   struct drm_gem_object *obj)
+{
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	int ret, i;
+
+	/*
+	 * This needs to happen before modifier conversion as that might change
+	 * the number of planes.
+	 */
+	for (i = 1; i < rfb->base.format->num_planes; ++i) {
+		if (mode_cmd->handles[i] != mode_cmd->handles[0]) {
+			drm_dbg_kms(dev, "Plane 0 and %d have different BOs: %u vs. %u\n",
+				    i, mode_cmd->handles[0], mode_cmd->handles[i]);
+			ret = -EINVAL;
+			return ret;
+		}
+	}
+
+	ret = amdgpu_display_get_fb_info(rfb, &rfb->tiling_flags, &rfb->tmz_surface,
+					 &rfb->gfx12_dcc);
+	if (ret)
+		return ret;
+
+	if (dev->mode_config.fb_modifiers_not_supported && !adev->enable_virtual_display) {
+		drm_WARN_ONCE(dev, adev->family >= AMDGPU_FAMILY_AI,
+			      "GFX9+ requires FB check based on format modifier\n");
+		ret = check_tiling_flags_gfx6(rfb);
+		if (ret)
+			return ret;
+	}
+
+	if (!dev->mode_config.fb_modifiers_not_supported &&
+	    !(rfb->base.flags & DRM_MODE_FB_MODIFIERS)) {
+		if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(12, 0, 0))
+			ret = convert_tiling_flags_to_modifier_gfx12(rfb);
+		else
+			ret = convert_tiling_flags_to_modifier(rfb);
+
+		if (ret) {
+			drm_dbg_kms(dev, "Failed to convert tiling flags 0x%llX to a modifier",
+				    rfb->tiling_flags);
+			return ret;
+		}
+	}
+
+	ret = amdgpu_display_verify_sizes(rfb);
+	if (ret)
 		return ret;
+
+	for (i = 0; i < rfb->base.format->num_planes; ++i) {
+		drm_gem_object_get(rfb->base.obj[0]);
+		rfb->base.obj[i] = rfb->base.obj[0];
 	}
+
 	return 0;
 }
 
-static struct drm_framebuffer *
-amdgpu_user_framebuffer_create(struct drm_device *dev,
-			       struct drm_file *file_priv,
-			       const struct drm_mode_fb_cmd2 *mode_cmd)
+struct drm_framebuffer *
+amdgpu_display_user_framebuffer_create(struct drm_device *dev,
+				       struct drm_file *file_priv,
+				       const struct drm_format_info *info,
+				       const struct drm_mode_fb_cmd2 *mode_cmd)
 {
-	struct drm_gem_object *obj;
 	struct amdgpu_framebuffer *amdgpu_fb;
+	struct drm_gem_object *obj;
+	struct amdgpu_bo *bo;
+	uint32_t domains;
 	int ret;
 
 	obj = drm_gem_object_lookup(file_priv, mode_cmd->handles[0]);
 	if (obj ==  NULL) {
-		dev_err(&dev->pdev->dev, "No GEM object associated to handle 0x%08X, "
-			"can't create framebuffer\n", mode_cmd->handles[0]);
+		drm_dbg_kms(dev,
+			    "No GEM object associated to handle 0x%08X, can't create framebuffer\n",
+			    mode_cmd->handles[0]);
+
 		return ERR_PTR(-ENOENT);
 	}
 
 	/* Handle is imported dma-buf, so cannot be migrated to VRAM for scanout */
-	if (obj->import_attach) {
-		DRM_DEBUG_KMS("Cannot create framebuffer from imported dma_buf\n");
+	bo = gem_to_amdgpu_bo(obj);
+	domains = amdgpu_display_supported_domains(drm_to_adev(dev), bo->flags);
+	if (drm_gem_is_imported(obj) && !(domains & AMDGPU_GEM_DOMAIN_GTT)) {
+		drm_dbg_kms(dev, "Cannot create framebuffer from imported dma_buf\n");
+		drm_gem_object_put(obj);
 		return ERR_PTR(-EINVAL);
 	}
 
 	amdgpu_fb = kzalloc(sizeof(*amdgpu_fb), GFP_KERNEL);
 	if (amdgpu_fb == NULL) {
-		drm_gem_object_put_unlocked(obj);
+		drm_gem_object_put(obj);
 		return ERR_PTR(-ENOMEM);
 	}
 
-	ret = amdgpu_framebuffer_init(dev, amdgpu_fb, mode_cmd, obj);
+	ret = amdgpu_display_gem_fb_verify_and_init(dev, amdgpu_fb, file_priv,
+						    info, mode_cmd, obj);
 	if (ret) {
 		kfree(amdgpu_fb);
-		drm_gem_object_put_unlocked(obj);
+		drm_gem_object_put(obj);
 		return ERR_PTR(ret);
 	}
 
+	drm_gem_object_put(obj);
 	return &amdgpu_fb->base;
 }
 
-static void amdgpu_output_poll_changed(struct drm_device *dev)
-{
-	struct amdgpu_device *adev = dev->dev_private;
-	amdgpu_fb_output_poll_changed(adev);
-}
-
 const struct drm_mode_config_funcs amdgpu_mode_funcs = {
-	.fb_create = amdgpu_user_framebuffer_create,
-	.output_poll_changed = amdgpu_output_poll_changed
+	.fb_create = amdgpu_display_user_framebuffer_create,
 };
 
-static const struct drm_prop_enum_list amdgpu_underscan_enum_list[] =
-{	{ UNDERSCAN_OFF, "off" },
+static const struct drm_prop_enum_list amdgpu_underscan_enum_list[] = {
+	{ UNDERSCAN_OFF, "off" },
 	{ UNDERSCAN_ON, "on" },
 	{ UNDERSCAN_AUTO, "auto" },
 };
 
-static const struct drm_prop_enum_list amdgpu_audio_enum_list[] =
-{	{ AMDGPU_AUDIO_DISABLE, "off" },
+static const struct drm_prop_enum_list amdgpu_audio_enum_list[] = {
+	{ AMDGPU_AUDIO_DISABLE, "off" },
 	{ AMDGPU_AUDIO_ENABLE, "on" },
 	{ AMDGPU_AUDIO_AUTO, "auto" },
 };
 
 /* XXX support different dither options? spatial, temporal, both, etc. */
-static const struct drm_prop_enum_list amdgpu_dither_enum_list[] =
-{	{ AMDGPU_FMT_DITHER_DISABLE, "off" },
+static const struct drm_prop_enum_list amdgpu_dither_enum_list[] = {
+	{ AMDGPU_FMT_DITHER_DISABLE, "off" },
 	{ AMDGPU_FMT_DITHER_ENABLE, "on" },
 };
 
-int amdgpu_modeset_create_props(struct amdgpu_device *adev)
+int amdgpu_display_modeset_create_props(struct amdgpu_device *adev)
 {
 	int sz;
 
 	adev->mode_info.coherent_mode_property =
-		drm_property_create_range(adev->ddev, 0 , "coherent", 0, 1);
+		drm_property_create_range(adev_to_drm(adev), 0, "coherent", 0, 1);
 	if (!adev->mode_info.coherent_mode_property)
 		return -ENOMEM;
 
 	adev->mode_info.load_detect_property =
-		drm_property_create_range(adev->ddev, 0, "load detection", 0, 1);
+		drm_property_create_range(adev_to_drm(adev), 0, "load detection", 0, 1);
 	if (!adev->mode_info.load_detect_property)
 		return -ENOMEM;
 
-	drm_mode_create_scaling_mode_property(adev->ddev);
+	drm_mode_create_scaling_mode_property(adev_to_drm(adev));
 
 	sz = ARRAY_SIZE(amdgpu_underscan_enum_list);
 	adev->mode_info.underscan_property =
-		drm_property_create_enum(adev->ddev, 0,
-				    "underscan",
-				    amdgpu_underscan_enum_list, sz);
+		drm_property_create_enum(adev_to_drm(adev), 0,
+					 "underscan",
+					 amdgpu_underscan_enum_list, sz);
 
 	adev->mode_info.underscan_hborder_property =
-		drm_property_create_range(adev->ddev, 0,
-					"underscan hborder", 0, 128);
+		drm_property_create_range(adev_to_drm(adev), 0,
+					  "underscan hborder", 0, 128);
 	if (!adev->mode_info.underscan_hborder_property)
 		return -ENOMEM;
 
 	adev->mode_info.underscan_vborder_property =
-		drm_property_create_range(adev->ddev, 0,
-					"underscan vborder", 0, 128);
+		drm_property_create_range(adev_to_drm(adev), 0,
+					  "underscan vborder", 0, 128);
 	if (!adev->mode_info.underscan_vborder_property)
 		return -ENOMEM;
 
 	sz = ARRAY_SIZE(amdgpu_audio_enum_list);
 	adev->mode_info.audio_property =
-		drm_property_create_enum(adev->ddev, 0,
+		drm_property_create_enum(adev_to_drm(adev), 0,
 					 "audio",
 					 amdgpu_audio_enum_list, sz);
 
 	sz = ARRAY_SIZE(amdgpu_dither_enum_list);
 	adev->mode_info.dither_property =
-		drm_property_create_enum(adev->ddev, 0,
+		drm_property_create_enum(adev_to_drm(adev), 0,
 					 "dither",
 					 amdgpu_dither_enum_list, sz);
 
 	return 0;
 }
 
-void amdgpu_update_display_priority(struct amdgpu_device *adev)
+void amdgpu_display_update_priority(struct amdgpu_device *adev)
 {
 	/* adjustment options for the display watermarks */
 	if ((amdgpu_disp_priority == 0) || (amdgpu_disp_priority > 2))
@@ -644,7 +1424,7 @@ void amdgpu_update_display_priority(struct amdgpu_device *adev)
 
 }
 
-static bool is_hdtv_mode(const struct drm_display_mode *mode)
+static bool amdgpu_display_is_hdtv_mode(const struct drm_display_mode *mode)
 {
 	/* try and guess if this is a tv or a monitor */
 	if ((mode->vdisplay == 480 && mode->hdisplay == 720) || /* 480p */
@@ -656,16 +1436,15 @@ static bool is_hdtv_mode(const struct drm_display_mode *mode)
 		return false;
 }
 
-bool amdgpu_crtc_scaling_mode_fixup(struct drm_crtc *crtc,
-				    const struct drm_display_mode *mode,
-				    struct drm_display_mode *adjusted_mode)
+bool amdgpu_display_crtc_scaling_mode_fixup(struct drm_crtc *crtc,
+					const struct drm_display_mode *mode,
+					struct drm_display_mode *adjusted_mode)
 {
 	struct drm_device *dev = crtc->dev;
 	struct drm_encoder *encoder;
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct amdgpu_encoder *amdgpu_encoder;
 	struct drm_connector *connector;
-	struct amdgpu_connector *amdgpu_connector;
 	u32 src_v = 1, dst_v = 1;
 	u32 src_h = 1, dst_h = 1;
 
@@ -677,7 +1456,6 @@ bool amdgpu_crtc_scaling_mode_fixup(struct drm_crtc *crtc,
 			continue;
 		amdgpu_encoder = to_amdgpu_encoder(encoder);
 		connector = amdgpu_get_connector_for_encoder(encoder);
-		amdgpu_connector = to_amdgpu_connector(connector);
 
 		/* set scaling */
 		if (amdgpu_encoder->rmx_type == RMX_OFF)
@@ -700,8 +1478,8 @@ bool amdgpu_crtc_scaling_mode_fixup(struct drm_crtc *crtc,
 		if ((!(mode->flags & DRM_MODE_FLAG_INTERLACE)) &&
 		    ((amdgpu_encoder->underscan_type == UNDERSCAN_ON) ||
 		     ((amdgpu_encoder->underscan_type == UNDERSCAN_AUTO) &&
-		      drm_detect_hdmi_monitor(amdgpu_connector_edid(connector)) &&
-		      is_hdtv_mode(mode)))) {
+		      connector && connector->display_info.is_hdmi &&
+		      amdgpu_display_is_hdtv_mode(mode)))) {
 			if (amdgpu_encoder->underscan_hborder != 0)
 				amdgpu_crtc->h_border = amdgpu_encoder->underscan_hborder;
 			else
@@ -719,6 +1497,7 @@ bool amdgpu_crtc_scaling_mode_fixup(struct drm_crtc *crtc,
 	}
 	if (amdgpu_crtc->rmx_type != RMX_OFF) {
 		fixed20_12 a, b;
+
 		a.full = dfixed_const(src_v);
 		b.full = dfixed_const(dst_v);
 		amdgpu_crtc->vsc.full = dfixed_div(a, b);
@@ -738,7 +1517,7 @@ bool amdgpu_crtc_scaling_mode_fixup(struct drm_crtc *crtc,
  *
  * \param dev Device to query.
  * \param pipe Crtc to query.
- * \param flags Flags from caller (DRM_CALLED_FROM_VBLIRQ or 0).
+ * \param flags from caller (DRM_CALLED_FROM_VBLIRQ or 0).
  *              For driver internal use only also supports these flags:
  *
  *              USE_REAL_VBLANKSTART to use the real start of vblank instead
@@ -769,16 +1548,16 @@ bool amdgpu_crtc_scaling_mode_fixup(struct drm_crtc *crtc,
  * unknown small number of scanlines wrt. real scanout position.
  *
  */
-int amdgpu_get_crtc_scanoutpos(struct drm_device *dev, unsigned int pipe,
-			       unsigned int flags, int *vpos, int *hpos,
-			       ktime_t *stime, ktime_t *etime,
-			       const struct drm_display_mode *mode)
+int amdgpu_display_get_crtc_scanoutpos(struct drm_device *dev,
+			unsigned int pipe, unsigned int flags, int *vpos,
+			int *hpos, ktime_t *stime, ktime_t *etime,
+			const struct drm_display_mode *mode)
 {
 	u32 vbl = 0, position = 0;
 	int vbl_start, vbl_end, vtotal, ret = 0;
 	bool in_vbl = true;
 
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 
 	/* preempt_disable_rt() should go right here in PREEMPT_RT patchset. */
 
@@ -805,8 +1584,7 @@ int amdgpu_get_crtc_scanoutpos(struct drm_device *dev, unsigned int pipe,
 		ret |= DRM_SCANOUTPOS_ACCURATE;
 		vbl_start = vbl & 0x1fff;
 		vbl_end = (vbl >> 16) & 0x1fff;
-	}
-	else {
+	} else {
 		/* No: Fake something reasonable which gives at least ok results. */
 		vbl_start = mode->crtc_vdisplay;
 		vbl_end = 0;
@@ -814,8 +1592,8 @@ int amdgpu_get_crtc_scanoutpos(struct drm_device *dev, unsigned int pipe,
 
 	/* Called from driver internal vblank counter query code? */
 	if (flags & GET_DISTANCE_TO_VBLANKSTART) {
-	    /* Caller wants distance from real vbl_start in *hpos */
-	    *hpos = *vpos - vbl_start;
+		/* Caller wants distance from real vbl_start in *hpos */
+		*hpos = *vpos - vbl_start;
 	}
 
 	/* Fudge vblank to start a few scanlines earlier to handle the
@@ -837,7 +1615,7 @@ int amdgpu_get_crtc_scanoutpos(struct drm_device *dev, unsigned int pipe,
 
 	/* In vblank? */
 	if (in_vbl)
-	    ret |= DRM_SCANOUTPOS_IN_VBLANK;
+		ret |= DRM_SCANOUTPOS_IN_VBLANK;
 
 	/* Called from driver internal vblank counter query code? */
 	if (flags & GET_DISTANCE_TO_VBLANKSTART) {
@@ -855,7 +1633,12 @@ int amdgpu_get_crtc_scanoutpos(struct drm_device *dev, unsigned int pipe,
 	/* Inside "upper part" of vblank area? Apply corrective offset if so: */
 	if (in_vbl && (*vpos >= vbl_start)) {
 		vtotal = mode->crtc_vtotal;
-		*vpos = *vpos - vtotal;
+
+		/* With variable refresh rate displays the vpos can exceed
+		 * the vtotal value. Clamp to 0 to return -vbl_end instead
+		 * of guessing the remaining number of lines until scanout.
+		 */
+		*vpos = (*vpos < vtotal) ? (*vpos - vtotal) : 0;
 	}
 
 	/* Correct for shifted end of vbl at vbl_end. */
@@ -864,7 +1647,7 @@ int amdgpu_get_crtc_scanoutpos(struct drm_device *dev, unsigned int pipe,
 	return ret;
 }
 
-int amdgpu_crtc_idx_to_irq_type(struct amdgpu_device *adev, int crtc)
+int amdgpu_display_crtc_idx_to_irq_type(struct amdgpu_device *adev, int crtc)
 {
 	if (crtc < 0 || crtc >= adev->mode_info.num_crtc)
 		return AMDGPU_CRTC_IRQ_NONE;
@@ -886,3 +1669,204 @@ int amdgpu_crtc_idx_to_irq_type(struct amdgpu_device *adev, int crtc)
 		return AMDGPU_CRTC_IRQ_NONE;
 	}
 }
+
+bool amdgpu_crtc_get_scanout_position(struct drm_crtc *crtc,
+			bool in_vblank_irq, int *vpos,
+			int *hpos, ktime_t *stime, ktime_t *etime,
+			const struct drm_display_mode *mode)
+{
+	struct drm_device *dev = crtc->dev;
+	unsigned int pipe = crtc->index;
+
+	return amdgpu_display_get_crtc_scanoutpos(dev, pipe, 0, vpos, hpos,
+						  stime, etime, mode);
+}
+
+static bool
+amdgpu_display_robj_is_fb(struct amdgpu_device *adev, struct amdgpu_bo *robj)
+{
+	struct drm_device *dev = adev_to_drm(adev);
+	struct drm_fb_helper *fb_helper = dev->fb_helper;
+
+	if (!fb_helper || !fb_helper->buffer)
+		return false;
+
+	if (gem_to_amdgpu_bo(fb_helper->buffer->gem) != robj)
+		return false;
+
+	return true;
+}
+
+int amdgpu_display_suspend_helper(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev_to_drm(adev);
+	struct drm_crtc *crtc;
+	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
+	int r;
+
+	drm_kms_helper_poll_disable(dev);
+
+	/* turn off display hw */
+	drm_modeset_lock_all(dev);
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter)
+		drm_helper_connector_dpms(connector,
+					  DRM_MODE_DPMS_OFF);
+	drm_connector_list_iter_end(&iter);
+	drm_modeset_unlock_all(dev);
+	/* unpin the front buffers and cursors */
+	list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
+		struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+		struct drm_framebuffer *fb = crtc->primary->fb;
+		struct amdgpu_bo *robj;
+
+		if (amdgpu_crtc->cursor_bo && !adev->enable_virtual_display) {
+			struct amdgpu_bo *aobj = gem_to_amdgpu_bo(amdgpu_crtc->cursor_bo);
+
+			r = amdgpu_bo_reserve(aobj, true);
+			if (r == 0) {
+				amdgpu_bo_unpin(aobj);
+				amdgpu_bo_unreserve(aobj);
+			}
+		}
+
+		if (!fb || !fb->obj[0])
+			continue;
+
+		robj = gem_to_amdgpu_bo(fb->obj[0]);
+		if (!amdgpu_display_robj_is_fb(adev, robj)) {
+			r = amdgpu_bo_reserve(robj, true);
+			if (r == 0) {
+				amdgpu_bo_unpin(robj);
+				amdgpu_bo_unreserve(robj);
+			}
+		}
+	}
+	return 0;
+}
+
+int amdgpu_display_resume_helper(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev_to_drm(adev);
+	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
+	struct drm_crtc *crtc;
+	int r;
+
+	/* pin cursors */
+	list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
+		struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+		if (amdgpu_crtc->cursor_bo && !adev->enable_virtual_display) {
+			struct amdgpu_bo *aobj = gem_to_amdgpu_bo(amdgpu_crtc->cursor_bo);
+
+			r = amdgpu_bo_reserve(aobj, true);
+			if (r == 0) {
+				aobj->flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+				r = amdgpu_bo_pin(aobj, AMDGPU_GEM_DOMAIN_VRAM);
+				if (r != 0)
+					dev_err(adev->dev, "Failed to pin cursor BO (%d)\n", r);
+				amdgpu_crtc->cursor_addr = amdgpu_bo_gpu_offset(aobj);
+				amdgpu_bo_unreserve(aobj);
+			}
+		}
+	}
+
+	drm_helper_resume_force_mode(dev);
+
+	/* turn on display hw */
+	drm_modeset_lock_all(dev);
+
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter)
+		drm_helper_connector_dpms(connector,
+					  DRM_MODE_DPMS_ON);
+	drm_connector_list_iter_end(&iter);
+
+	drm_modeset_unlock_all(dev);
+
+	drm_kms_helper_poll_enable(dev);
+
+	return 0;
+}
+
+/* panic_bo is set in amdgpu_dm_plane_get_scanout_buffer() and only used in amdgpu_dm_set_pixel()
+ * they are called from the panic handler, and protected by the drm_panic spinlock.
+ */
+static struct amdgpu_bo *panic_abo;
+
+/* Use the indirect MMIO to write each pixel to the GPU VRAM,
+ * This is a simplified version of amdgpu_device_mm_access()
+ */
+static void amdgpu_display_set_pixel(struct drm_scanout_buffer *sb,
+				     unsigned int x,
+				     unsigned int y,
+				     u32 color)
+{
+	struct amdgpu_res_cursor cursor;
+	unsigned long offset;
+	struct amdgpu_bo *abo = panic_abo;
+	struct amdgpu_device *adev = amdgpu_ttm_adev(abo->tbo.bdev);
+	uint32_t tmp;
+
+	offset = x * 4 + y * sb->pitch[0];
+	amdgpu_res_first(abo->tbo.resource, offset, 4, &cursor);
+
+	tmp = cursor.start >> 31;
+	WREG32_NO_KIQ(mmMM_INDEX, ((uint32_t) cursor.start) | 0x80000000);
+	if (tmp != 0xffffffff)
+		WREG32_NO_KIQ(mmMM_INDEX_HI, tmp);
+	WREG32_NO_KIQ(mmMM_DATA, color);
+}
+
+int amdgpu_display_get_scanout_buffer(struct drm_plane *plane,
+				      struct drm_scanout_buffer *sb)
+{
+	struct amdgpu_bo *abo;
+	struct drm_framebuffer *fb = plane->state->fb;
+
+	if (!fb)
+		return -EINVAL;
+
+	DRM_DEBUG_KMS("Framebuffer %dx%d %p4cc\n", fb->width, fb->height, &fb->format->format);
+
+	abo = gem_to_amdgpu_bo(fb->obj[0]);
+	if (!abo)
+		return -EINVAL;
+
+	sb->width = fb->width;
+	sb->height = fb->height;
+	/* Use the generic linear format, because tiling will be disabled in panic_flush() */
+	sb->format = drm_format_info(fb->format->format);
+	if (!sb->format)
+		return -EINVAL;
+
+	sb->pitch[0] = fb->pitches[0];
+
+	if (abo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS) {
+		if (abo->tbo.resource->mem_type != TTM_PL_VRAM) {
+			drm_warn(plane->dev, "amdgpu panic, framebuffer not in VRAM\n");
+			return -EINVAL;
+		}
+		/* Only handle 32bits format, to simplify mmio access */
+		if (fb->format->cpp[0] != 4) {
+			drm_warn(plane->dev, "amdgpu panic, pixel format is not 32bits\n");
+			return -EINVAL;
+		}
+		sb->set_pixel = amdgpu_display_set_pixel;
+		panic_abo = abo;
+		return 0;
+	}
+	if (!abo->kmap.virtual &&
+	    ttm_bo_kmap(&abo->tbo, 0, PFN_UP(abo->tbo.base.size), &abo->kmap)) {
+		drm_warn(plane->dev, "amdgpu bo map failed, panic won't be displayed\n");
+		return -ENOMEM;
+	}
+	if (abo->kmap.bo_kmap_type & TTM_BO_MAP_IOMEM_MASK)
+		iosys_map_set_vaddr_iomem(&sb->map[0], abo->kmap.virtual);
+	else
+		iosys_map_set_vaddr(&sb->map[0], abo->kmap.virtual);
+
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h
new file mode 100644
index 000000000000..930c171473b4
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h
@@ -0,0 +1,58 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __AMDGPU_DISPLAY_H__
+#define __AMDGPU_DISPLAY_H__
+
+#include <drm/drm_panic.h>
+
+#define amdgpu_display_vblank_get_counter(adev, crtc) (adev)->mode_info.funcs->vblank_get_counter((adev), (crtc))
+#define amdgpu_display_backlight_set_level(adev, e, l) (adev)->mode_info.funcs->backlight_set_level((e), (l))
+#define amdgpu_display_backlight_get_level(adev, e) (adev)->mode_info.funcs->backlight_get_level((e))
+#define amdgpu_display_hpd_sense(adev, h) (adev)->mode_info.funcs->hpd_sense((adev), (h))
+#define amdgpu_display_hpd_set_polarity(adev, h) (adev)->mode_info.funcs->hpd_set_polarity((adev), (h))
+#define amdgpu_display_hpd_get_gpio_reg(adev) (adev)->mode_info.funcs->hpd_get_gpio_reg((adev))
+#define amdgpu_display_bandwidth_update(adev) (adev)->mode_info.funcs->bandwidth_update((adev))
+#define amdgpu_display_page_flip(adev, crtc, base, async) (adev)->mode_info.funcs->page_flip((adev), (crtc), (base), (async))
+#define amdgpu_display_page_flip_get_scanoutpos(adev, crtc, vbl, pos) (adev)->mode_info.funcs->page_flip_get_scanoutpos((adev), (crtc), (vbl), (pos))
+#define amdgpu_display_add_encoder(adev, e, s, c) (adev)->mode_info.funcs->add_encoder((adev), (e), (s), (c))
+#define amdgpu_display_add_connector(adev, ci, sd, ct, ib, coi, h, r) (adev)->mode_info.funcs->add_connector((adev), (ci), (sd), (ct), (ib), (coi), (h), (r))
+
+void amdgpu_display_hotplug_work_func(struct work_struct *work);
+void amdgpu_display_update_priority(struct amdgpu_device *adev);
+uint32_t amdgpu_display_supported_domains(struct amdgpu_device *adev,
+					  uint64_t bo_flags);
+struct drm_framebuffer *
+amdgpu_display_user_framebuffer_create(struct drm_device *dev,
+				       struct drm_file *file_priv,
+				       const struct drm_format_info *info,
+				       const struct drm_mode_fb_cmd2 *mode_cmd);
+const struct drm_format_info *
+amdgpu_lookup_format_info(u32 format, uint64_t modifier);
+
+int amdgpu_display_suspend_helper(struct amdgpu_device *adev);
+int amdgpu_display_resume_helper(struct amdgpu_device *adev);
+
+int amdgpu_display_get_scanout_buffer(struct drm_plane *plane,
+				      struct drm_scanout_buffer *sb);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c
new file mode 100644
index 000000000000..8561ad7f6180
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c
@@ -0,0 +1,574 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * based on nouveau_prime.c
+ *
+ * Authors: Alex Deucher
+ */
+
+/**
+ * DOC: PRIME Buffer Sharing
+ *
+ * The following callback implementations are used for :ref:`sharing GEM buffer
+ * objects between different devices via PRIME <prime_buffer_sharing>`.
+ */
+
+#include "amdgpu.h"
+#include "amdgpu_display.h"
+#include "amdgpu_gem.h"
+#include "amdgpu_dma_buf.h"
+#include "amdgpu_xgmi.h"
+#include "amdgpu_vm.h"
+#include <drm/amdgpu_drm.h>
+#include <drm/ttm/ttm_tt.h>
+#include <linux/dma-buf.h>
+#include <linux/dma-fence-array.h>
+#include <linux/pci-p2pdma.h>
+
+static const struct dma_buf_attach_ops amdgpu_dma_buf_attach_ops;
+
+/**
+ * dma_buf_attach_adev - Helper to get adev of an attachment
+ *
+ * @attach: attachment
+ *
+ * Returns:
+ * A struct amdgpu_device * if the attaching device is an amdgpu device or
+ * partition, NULL otherwise.
+ */
+static struct amdgpu_device *dma_buf_attach_adev(struct dma_buf_attachment *attach)
+{
+	if (attach->importer_ops == &amdgpu_dma_buf_attach_ops) {
+		struct drm_gem_object *obj = attach->importer_priv;
+		struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+
+		return amdgpu_ttm_adev(bo->tbo.bdev);
+	}
+
+	return NULL;
+}
+
+/**
+ * amdgpu_dma_buf_attach - &dma_buf_ops.attach implementation
+ *
+ * @dmabuf: DMA-buf where we attach to
+ * @attach: attachment to add
+ *
+ * Add the attachment as user to the exported DMA-buf.
+ */
+static int amdgpu_dma_buf_attach(struct dma_buf *dmabuf,
+				 struct dma_buf_attachment *attach)
+{
+	struct amdgpu_device *attach_adev = dma_buf_attach_adev(attach);
+	struct drm_gem_object *obj = dmabuf->priv;
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+
+	if (!amdgpu_dmabuf_is_xgmi_accessible(attach_adev, bo) &&
+	    pci_p2pdma_distance(adev->pdev, attach->dev, false) < 0)
+		attach->peer2peer = false;
+
+	amdgpu_vm_bo_update_shared(bo);
+
+	return 0;
+}
+
+/**
+ * amdgpu_dma_buf_pin - &dma_buf_ops.pin implementation
+ *
+ * @attach: attachment to pin down
+ *
+ * Pin the BO which is backing the DMA-buf so that it can't move any more.
+ */
+static int amdgpu_dma_buf_pin(struct dma_buf_attachment *attach)
+{
+	struct dma_buf *dmabuf = attach->dmabuf;
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(dmabuf->priv);
+	u32 domains = bo->allowed_domains;
+
+	dma_resv_assert_held(dmabuf->resv);
+
+	/* Try pinning into VRAM to allow P2P with RDMA NICs without ODP
+	 * support if all attachments can do P2P. If any attachment can't do
+	 * P2P just pin into GTT instead.
+	 *
+	 * To avoid with conflicting pinnings between GPUs and RDMA when move
+	 * notifiers are disabled, only allow pinning in VRAM when move
+	 * notiers are enabled.
+	 */
+	if (!IS_ENABLED(CONFIG_DMABUF_MOVE_NOTIFY)) {
+		domains &= ~AMDGPU_GEM_DOMAIN_VRAM;
+	} else {
+		list_for_each_entry(attach, &dmabuf->attachments, node)
+			if (!attach->peer2peer)
+				domains &= ~AMDGPU_GEM_DOMAIN_VRAM;
+	}
+
+	if (domains & AMDGPU_GEM_DOMAIN_VRAM)
+		bo->flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
+
+	if (WARN_ON(!domains))
+		return -EINVAL;
+
+	return amdgpu_bo_pin(bo, domains);
+}
+
+/**
+ * amdgpu_dma_buf_unpin - &dma_buf_ops.unpin implementation
+ *
+ * @attach: attachment to unpin
+ *
+ * Unpin a previously pinned BO to make it movable again.
+ */
+static void amdgpu_dma_buf_unpin(struct dma_buf_attachment *attach)
+{
+	struct drm_gem_object *obj = attach->dmabuf->priv;
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+
+	amdgpu_bo_unpin(bo);
+}
+
+/**
+ * amdgpu_dma_buf_map - &dma_buf_ops.map_dma_buf implementation
+ * @attach: DMA-buf attachment
+ * @dir: DMA direction
+ *
+ * Makes sure that the shared DMA buffer can be accessed by the target device.
+ * For now, simply pins it to the GTT domain, where it should be accessible by
+ * all DMA devices.
+ *
+ * Returns:
+ * sg_table filled with the DMA addresses to use or ERR_PRT with negative error
+ * code.
+ */
+static struct sg_table *amdgpu_dma_buf_map(struct dma_buf_attachment *attach,
+					   enum dma_data_direction dir)
+{
+	struct dma_buf *dma_buf = attach->dmabuf;
+	struct drm_gem_object *obj = dma_buf->priv;
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	struct sg_table *sgt;
+	long r;
+
+	if (!bo->tbo.pin_count) {
+		/* move buffer into GTT or VRAM */
+		struct ttm_operation_ctx ctx = { false, false };
+		unsigned int domains = AMDGPU_GEM_DOMAIN_GTT;
+
+		if (bo->preferred_domains & AMDGPU_GEM_DOMAIN_VRAM &&
+		    attach->peer2peer) {
+			bo->flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
+			domains |= AMDGPU_GEM_DOMAIN_VRAM;
+		}
+		amdgpu_bo_placement_from_domain(bo, domains);
+		r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+		if (r)
+			return ERR_PTR(r);
+	}
+
+	switch (bo->tbo.resource->mem_type) {
+	case TTM_PL_TT:
+		sgt = drm_prime_pages_to_sg(obj->dev,
+					    bo->tbo.ttm->pages,
+					    bo->tbo.ttm->num_pages);
+		if (IS_ERR(sgt))
+			return sgt;
+
+		if (dma_map_sgtable(attach->dev, sgt, dir,
+				    DMA_ATTR_SKIP_CPU_SYNC))
+			goto error_free;
+		break;
+
+	case TTM_PL_VRAM:
+		/* XGMI-accessible memory should never be DMA-mapped */
+		if (WARN_ON(amdgpu_dmabuf_is_xgmi_accessible(
+				dma_buf_attach_adev(attach), bo)))
+			return ERR_PTR(-EINVAL);
+
+		r = amdgpu_vram_mgr_alloc_sgt(adev, bo->tbo.resource, 0,
+					      bo->tbo.base.size, attach->dev,
+					      dir, &sgt);
+		if (r)
+			return ERR_PTR(r);
+		break;
+	default:
+		return ERR_PTR(-EINVAL);
+	}
+
+	return sgt;
+
+error_free:
+	sg_free_table(sgt);
+	kfree(sgt);
+	return ERR_PTR(-EBUSY);
+}
+
+/**
+ * amdgpu_dma_buf_unmap - &dma_buf_ops.unmap_dma_buf implementation
+ * @attach: DMA-buf attachment
+ * @sgt: sg_table to unmap
+ * @dir: DMA direction
+ *
+ * This is called when a shared DMA buffer no longer needs to be accessible by
+ * another device. For now, simply unpins the buffer from GTT.
+ */
+static void amdgpu_dma_buf_unmap(struct dma_buf_attachment *attach,
+				 struct sg_table *sgt,
+				 enum dma_data_direction dir)
+{
+	if (sg_page(sgt->sgl)) {
+		dma_unmap_sgtable(attach->dev, sgt, dir, 0);
+		sg_free_table(sgt);
+		kfree(sgt);
+	} else {
+		amdgpu_vram_mgr_free_sgt(attach->dev, dir, sgt);
+	}
+}
+
+/**
+ * amdgpu_dma_buf_begin_cpu_access - &dma_buf_ops.begin_cpu_access implementation
+ * @dma_buf: Shared DMA buffer
+ * @direction: Direction of DMA transfer
+ *
+ * This is called before CPU access to the shared DMA buffer's memory. If it's
+ * a read access, the buffer is moved to the GTT domain if possible, for optimal
+ * CPU read performance.
+ *
+ * Returns:
+ * 0 on success or a negative error code on failure.
+ */
+static int amdgpu_dma_buf_begin_cpu_access(struct dma_buf *dma_buf,
+					   enum dma_data_direction direction)
+{
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(dma_buf->priv);
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	struct ttm_operation_ctx ctx = { true, false };
+	u32 domain = amdgpu_display_supported_domains(adev, bo->flags);
+	int ret;
+	bool reads = (direction == DMA_BIDIRECTIONAL ||
+		      direction == DMA_FROM_DEVICE);
+
+	if (!reads || !(domain & AMDGPU_GEM_DOMAIN_GTT))
+		return 0;
+
+	/* move to gtt */
+	ret = amdgpu_bo_reserve(bo, false);
+	if (unlikely(ret != 0))
+		return ret;
+
+	if (!bo->tbo.pin_count &&
+	    (bo->allowed_domains & AMDGPU_GEM_DOMAIN_GTT)) {
+		amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_GTT);
+		ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+	}
+
+	amdgpu_bo_unreserve(bo);
+	return ret;
+}
+
+static int amdgpu_dma_buf_vmap(struct dma_buf *dma_buf, struct iosys_map *map)
+{
+	struct drm_gem_object *obj = dma_buf->priv;
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+	int ret;
+
+	/*
+	 * Pin to keep buffer in place while it's vmap'ed. The actual
+	 * domain is not that important as long as it's mapable. Using
+	 * GTT and VRAM should be compatible with most use cases.
+	 */
+	ret = amdgpu_bo_pin(bo, AMDGPU_GEM_DOMAIN_GTT | AMDGPU_GEM_DOMAIN_VRAM);
+	if (ret)
+		return ret;
+	ret = drm_gem_dmabuf_vmap(dma_buf, map);
+	if (ret)
+		amdgpu_bo_unpin(bo);
+
+	return ret;
+}
+
+static void amdgpu_dma_buf_vunmap(struct dma_buf *dma_buf, struct iosys_map *map)
+{
+	struct drm_gem_object *obj = dma_buf->priv;
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+
+	drm_gem_dmabuf_vunmap(dma_buf, map);
+	amdgpu_bo_unpin(bo);
+}
+
+const struct dma_buf_ops amdgpu_dmabuf_ops = {
+	.attach = amdgpu_dma_buf_attach,
+	.pin = amdgpu_dma_buf_pin,
+	.unpin = amdgpu_dma_buf_unpin,
+	.map_dma_buf = amdgpu_dma_buf_map,
+	.unmap_dma_buf = amdgpu_dma_buf_unmap,
+	.release = drm_gem_dmabuf_release,
+	.begin_cpu_access = amdgpu_dma_buf_begin_cpu_access,
+	.mmap = drm_gem_dmabuf_mmap,
+	.vmap = amdgpu_dma_buf_vmap,
+	.vunmap = amdgpu_dma_buf_vunmap,
+};
+
+/**
+ * amdgpu_gem_prime_export - &drm_driver.gem_prime_export implementation
+ * @gobj: GEM BO
+ * @flags: Flags such as DRM_CLOEXEC and DRM_RDWR.
+ *
+ * The main work is done by the &drm_gem_prime_export helper.
+ *
+ * Returns:
+ * Shared DMA buffer representing the GEM BO from the given device.
+ */
+struct dma_buf *amdgpu_gem_prime_export(struct drm_gem_object *gobj,
+					int flags)
+{
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(gobj);
+	struct dma_buf *buf;
+	struct ttm_operation_ctx ctx = {
+		.interruptible = true,
+		.no_wait_gpu = true,
+		/* We opt to avoid OOM on system pages allocations */
+		.gfp_retry_mayfail = true,
+		.allow_res_evict = false,
+	};
+	int ret;
+
+	if (amdgpu_ttm_tt_get_usermm(bo->tbo.ttm) ||
+	    bo->flags & AMDGPU_GEM_CREATE_VM_ALWAYS_VALID)
+		return ERR_PTR(-EPERM);
+
+	ret = ttm_bo_setup_export(&bo->tbo, &ctx);
+	if (ret)
+		return ERR_PTR(ret);
+
+	buf = drm_gem_prime_export(gobj, flags);
+	if (!IS_ERR(buf))
+		buf->ops = &amdgpu_dmabuf_ops;
+
+	return buf;
+}
+
+/**
+ * amdgpu_dma_buf_create_obj - create BO for DMA-buf import
+ *
+ * @dev: DRM device
+ * @dma_buf: DMA-buf
+ *
+ * Creates an empty SG BO for DMA-buf import.
+ *
+ * Returns:
+ * A new GEM BO of the given DRM device, representing the memory
+ * described by the given DMA-buf attachment and scatter/gather table.
+ */
+static struct drm_gem_object *
+amdgpu_dma_buf_create_obj(struct drm_device *dev, struct dma_buf *dma_buf)
+{
+	struct dma_resv *resv = dma_buf->resv;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	struct drm_gem_object *gobj;
+	struct amdgpu_bo *bo;
+	uint64_t flags = 0;
+	int ret;
+
+	dma_resv_lock(resv, NULL);
+
+	if (dma_buf->ops == &amdgpu_dmabuf_ops) {
+		struct amdgpu_bo *other = gem_to_amdgpu_bo(dma_buf->priv);
+
+		flags |= other->flags & (AMDGPU_GEM_CREATE_CPU_GTT_USWC |
+					 AMDGPU_GEM_CREATE_COHERENT |
+					 AMDGPU_GEM_CREATE_EXT_COHERENT |
+					 AMDGPU_GEM_CREATE_UNCACHED);
+	}
+
+	ret = amdgpu_gem_object_create(adev, dma_buf->size, PAGE_SIZE,
+				       AMDGPU_GEM_DOMAIN_CPU, flags,
+				       ttm_bo_type_sg, resv, &gobj, 0);
+	if (ret)
+		goto error;
+
+	bo = gem_to_amdgpu_bo(gobj);
+	bo->allowed_domains = AMDGPU_GEM_DOMAIN_GTT;
+	bo->preferred_domains = AMDGPU_GEM_DOMAIN_GTT;
+
+	dma_resv_unlock(resv);
+	return gobj;
+
+error:
+	dma_resv_unlock(resv);
+	return ERR_PTR(ret);
+}
+
+/**
+ * amdgpu_dma_buf_move_notify - &attach.move_notify implementation
+ *
+ * @attach: the DMA-buf attachment
+ *
+ * Invalidate the DMA-buf attachment, making sure that the we re-create the
+ * mapping before the next use.
+ */
+static void
+amdgpu_dma_buf_move_notify(struct dma_buf_attachment *attach)
+{
+	struct drm_gem_object *obj = attach->importer_priv;
+	struct ww_acquire_ctx *ticket = dma_resv_locking_ctx(obj->resv);
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	struct ttm_operation_ctx ctx = { false, false };
+	struct ttm_placement placement = {};
+	struct amdgpu_vm_bo_base *bo_base;
+	int r;
+
+	/* FIXME: This should be after the "if", but needs a fix to make sure
+	 * DMABuf imports are initialized in the right VM list.
+	 */
+	amdgpu_vm_bo_invalidate(bo, false);
+	if (!bo->tbo.resource || bo->tbo.resource->mem_type == TTM_PL_SYSTEM)
+		return;
+
+	r = ttm_bo_validate(&bo->tbo, &placement, &ctx);
+	if (r) {
+		DRM_ERROR("Failed to invalidate DMA-buf import (%d))\n", r);
+		return;
+	}
+
+	for (bo_base = bo->vm_bo; bo_base; bo_base = bo_base->next) {
+		struct amdgpu_vm *vm = bo_base->vm;
+		struct dma_resv *resv = vm->root.bo->tbo.base.resv;
+
+		if (ticket) {
+			/* When we get an error here it means that somebody
+			 * else is holding the VM lock and updating page tables
+			 * So we can just continue here.
+			 */
+			r = dma_resv_lock(resv, ticket);
+			if (r)
+				continue;
+
+		} else {
+			/* TODO: This is more problematic and we actually need
+			 * to allow page tables updates without holding the
+			 * lock.
+			 */
+			if (!dma_resv_trylock(resv))
+				continue;
+		}
+
+		/* Reserve fences for two SDMA page table updates */
+		r = dma_resv_reserve_fences(resv, 2);
+		if (!r)
+			r = amdgpu_vm_clear_freed(adev, vm, NULL);
+		if (!r)
+			r = amdgpu_vm_handle_moved(adev, vm, ticket);
+
+		if (r && r != -EBUSY)
+			DRM_ERROR("Failed to invalidate VM page tables (%d))\n",
+				  r);
+
+		dma_resv_unlock(resv);
+	}
+}
+
+static const struct dma_buf_attach_ops amdgpu_dma_buf_attach_ops = {
+	.allow_peer2peer = true,
+	.move_notify = amdgpu_dma_buf_move_notify
+};
+
+/**
+ * amdgpu_gem_prime_import - &drm_driver.gem_prime_import implementation
+ * @dev: DRM device
+ * @dma_buf: Shared DMA buffer
+ *
+ * Import a dma_buf into a the driver and potentially create a new GEM object.
+ *
+ * Returns:
+ * GEM BO representing the shared DMA buffer for the given device.
+ */
+struct drm_gem_object *amdgpu_gem_prime_import(struct drm_device *dev,
+					       struct dma_buf *dma_buf)
+{
+	struct dma_buf_attachment *attach;
+	struct drm_gem_object *obj;
+
+	if (dma_buf->ops == &amdgpu_dmabuf_ops) {
+		obj = dma_buf->priv;
+		if (obj->dev == dev) {
+			/*
+			 * Importing dmabuf exported from out own gem increases
+			 * refcount on gem itself instead of f_count of dmabuf.
+			 */
+			drm_gem_object_get(obj);
+			return obj;
+		}
+	}
+
+	obj = amdgpu_dma_buf_create_obj(dev, dma_buf);
+	if (IS_ERR(obj))
+		return obj;
+
+	attach = dma_buf_dynamic_attach(dma_buf, dev->dev,
+					&amdgpu_dma_buf_attach_ops, obj);
+	if (IS_ERR(attach)) {
+		drm_gem_object_put(obj);
+		return ERR_CAST(attach);
+	}
+
+	get_dma_buf(dma_buf);
+	obj->import_attach = attach;
+	return obj;
+}
+
+/**
+ * amdgpu_dmabuf_is_xgmi_accessible - Check if xgmi available for P2P transfer
+ *
+ * @adev: amdgpu_device pointer of the importer
+ * @bo: amdgpu buffer object
+ *
+ * Returns:
+ * True if dmabuf accessible over xgmi, false otherwise.
+ */
+bool amdgpu_dmabuf_is_xgmi_accessible(struct amdgpu_device *adev,
+				      struct amdgpu_bo *bo)
+{
+	struct drm_gem_object *obj = &bo->tbo.base;
+	struct drm_gem_object *gobj;
+
+	if (!adev)
+		return false;
+
+	if (drm_gem_is_imported(obj)) {
+		struct dma_buf *dma_buf = obj->import_attach->dmabuf;
+
+		if (dma_buf->ops != &amdgpu_dmabuf_ops)
+			/* No XGMI with non AMD GPUs */
+			return false;
+
+		gobj = dma_buf->priv;
+		bo = gem_to_amdgpu_bo(gobj);
+	}
+
+	if (amdgpu_xgmi_same_hive(adev, amdgpu_ttm_adev(bo->tbo.bdev)) &&
+			(bo->preferred_domains & AMDGPU_GEM_DOMAIN_VRAM))
+		return true;
+
+	return false;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.h
new file mode 100644
index 000000000000..3e93b9b407a9
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.h
@@ -0,0 +1,37 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __AMDGPU_DMA_BUF_H__
+#define __AMDGPU_DMA_BUF_H__
+
+#include <drm/drm_gem.h>
+
+struct dma_buf *amdgpu_gem_prime_export(struct drm_gem_object *gobj,
+					int flags);
+struct drm_gem_object *amdgpu_gem_prime_import(struct drm_device *dev,
+					    struct dma_buf *dma_buf);
+bool amdgpu_dmabuf_is_xgmi_accessible(struct amdgpu_device *adev,
+				      struct amdgpu_bo *bo);
+
+extern const struct dma_buf_ops amdgpu_dmabuf_ops;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell.h
new file mode 100644
index 000000000000..2675689ef70f
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell.h
@@ -0,0 +1,372 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef AMDGPU_DOORBELL_H
+#define AMDGPU_DOORBELL_H
+
+/*
+ * GPU doorbell structures, functions & helpers
+ */
+struct amdgpu_doorbell {
+	/* doorbell mmio */
+	resource_size_t		base;
+	resource_size_t		size;
+
+	/* Number of doorbells reserved for amdgpu kernel driver */
+	u32 num_kernel_doorbells;
+
+	/* Kernel doorbells */
+	struct amdgpu_bo *kernel_doorbells;
+
+	/* For CPU access of doorbells */
+	uint32_t *cpu_addr;
+};
+
+/* Reserved doorbells for amdgpu (including multimedia).
+ * KFD can use all the rest in the 2M doorbell bar.
+ * For asic before vega10, doorbell is 32-bit, so the
+ * index/offset is in dword. For vega10 and after, doorbell
+ * can be 64-bit, so the index defined is in qword.
+ */
+struct amdgpu_doorbell_index {
+	uint32_t kiq;
+	uint32_t mec_ring0;
+	uint32_t mec_ring1;
+	uint32_t mec_ring2;
+	uint32_t mec_ring3;
+	uint32_t mec_ring4;
+	uint32_t mec_ring5;
+	uint32_t mec_ring6;
+	uint32_t mec_ring7;
+	uint32_t userqueue_start;
+	uint32_t userqueue_end;
+	uint32_t gfx_ring0;
+	uint32_t gfx_ring1;
+	uint32_t gfx_userqueue_start;
+	uint32_t gfx_userqueue_end;
+	uint32_t sdma_engine[16];
+	uint32_t mes_ring0;
+	uint32_t mes_ring1;
+	uint32_t ih;
+	union {
+		struct {
+			uint32_t vcn_ring0_1;
+			uint32_t vcn_ring2_3;
+			uint32_t vcn_ring4_5;
+			uint32_t vcn_ring6_7;
+		} vcn;
+		struct {
+			uint32_t uvd_ring0_1;
+			uint32_t uvd_ring2_3;
+			uint32_t uvd_ring4_5;
+			uint32_t uvd_ring6_7;
+			uint32_t vce_ring0_1;
+			uint32_t vce_ring2_3;
+			uint32_t vce_ring4_5;
+			uint32_t vce_ring6_7;
+		} uvd_vce;
+	};
+	uint32_t vpe_ring;
+	uint32_t first_non_cp;
+	uint32_t last_non_cp;
+	uint32_t max_assignment;
+	/* Per engine SDMA doorbell size in dword */
+	uint32_t sdma_doorbell_range;
+	/* Per xcc doorbell size for KIQ/KCQ */
+	uint32_t xcc_doorbell_range;
+};
+
+enum AMDGPU_DOORBELL_ASSIGNMENT {
+	AMDGPU_DOORBELL_KIQ                     = 0x000,
+	AMDGPU_DOORBELL_HIQ                     = 0x001,
+	AMDGPU_DOORBELL_DIQ                     = 0x002,
+	AMDGPU_DOORBELL_MEC_RING0               = 0x010,
+	AMDGPU_DOORBELL_MEC_RING1               = 0x011,
+	AMDGPU_DOORBELL_MEC_RING2               = 0x012,
+	AMDGPU_DOORBELL_MEC_RING3               = 0x013,
+	AMDGPU_DOORBELL_MEC_RING4               = 0x014,
+	AMDGPU_DOORBELL_MEC_RING5               = 0x015,
+	AMDGPU_DOORBELL_MEC_RING6               = 0x016,
+	AMDGPU_DOORBELL_MEC_RING7               = 0x017,
+	AMDGPU_DOORBELL_GFX_RING0               = 0x020,
+	AMDGPU_DOORBELL_sDMA_ENGINE0            = 0x1E0,
+	AMDGPU_DOORBELL_sDMA_ENGINE1            = 0x1E1,
+	AMDGPU_DOORBELL_IH                      = 0x1E8,
+	AMDGPU_DOORBELL_MAX_ASSIGNMENT          = 0x3FF,
+	AMDGPU_DOORBELL_INVALID                 = 0xFFFF
+};
+
+enum AMDGPU_VEGA20_DOORBELL_ASSIGNMENT {
+
+	/* Compute + GFX: 0~255 */
+	AMDGPU_VEGA20_DOORBELL_KIQ                     = 0x000,
+	AMDGPU_VEGA20_DOORBELL_HIQ                     = 0x001,
+	AMDGPU_VEGA20_DOORBELL_DIQ                     = 0x002,
+	AMDGPU_VEGA20_DOORBELL_MEC_RING0               = 0x003,
+	AMDGPU_VEGA20_DOORBELL_MEC_RING1               = 0x004,
+	AMDGPU_VEGA20_DOORBELL_MEC_RING2               = 0x005,
+	AMDGPU_VEGA20_DOORBELL_MEC_RING3               = 0x006,
+	AMDGPU_VEGA20_DOORBELL_MEC_RING4               = 0x007,
+	AMDGPU_VEGA20_DOORBELL_MEC_RING5               = 0x008,
+	AMDGPU_VEGA20_DOORBELL_MEC_RING6               = 0x009,
+	AMDGPU_VEGA20_DOORBELL_MEC_RING7               = 0x00A,
+	AMDGPU_VEGA20_DOORBELL_USERQUEUE_START	       = 0x00B,
+	AMDGPU_VEGA20_DOORBELL_USERQUEUE_END	       = 0x08A,
+	AMDGPU_VEGA20_DOORBELL_GFX_RING0               = 0x08B,
+	/* SDMA:256~335*/
+	AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE0            = 0x100,
+	AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE1            = 0x10A,
+	AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE2            = 0x114,
+	AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE3            = 0x11E,
+	AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE4            = 0x128,
+	AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE5            = 0x132,
+	AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE6            = 0x13C,
+	AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE7            = 0x146,
+	/* IH: 376~391 */
+	AMDGPU_VEGA20_DOORBELL_IH                      = 0x178,
+	/* MMSCH: 392~407
+	 * overlap the doorbell assignment with VCN as they are  mutually exclusive
+	 * VCN engine's doorbell is 32 bit and two VCN ring share one QWORD
+	 */
+	AMDGPU_VEGA20_DOORBELL64_VCN0_1                  = 0x188, /* VNC0 */
+	AMDGPU_VEGA20_DOORBELL64_VCN2_3                  = 0x189,
+	AMDGPU_VEGA20_DOORBELL64_VCN4_5                  = 0x18A,
+	AMDGPU_VEGA20_DOORBELL64_VCN6_7                  = 0x18B,
+
+	AMDGPU_VEGA20_DOORBELL64_VCN8_9                  = 0x18C, /* VNC1 */
+	AMDGPU_VEGA20_DOORBELL64_VCNa_b                  = 0x18D,
+	AMDGPU_VEGA20_DOORBELL64_VCNc_d                  = 0x18E,
+	AMDGPU_VEGA20_DOORBELL64_VCNe_f                  = 0x18F,
+
+	AMDGPU_VEGA20_DOORBELL64_UVD_RING0_1             = 0x188,
+	AMDGPU_VEGA20_DOORBELL64_UVD_RING2_3             = 0x189,
+	AMDGPU_VEGA20_DOORBELL64_UVD_RING4_5             = 0x18A,
+	AMDGPU_VEGA20_DOORBELL64_UVD_RING6_7             = 0x18B,
+
+	AMDGPU_VEGA20_DOORBELL64_VCE_RING0_1             = 0x18C,
+	AMDGPU_VEGA20_DOORBELL64_VCE_RING2_3             = 0x18D,
+	AMDGPU_VEGA20_DOORBELL64_VCE_RING4_5             = 0x18E,
+	AMDGPU_VEGA20_DOORBELL64_VCE_RING6_7             = 0x18F,
+
+	AMDGPU_VEGA20_DOORBELL64_FIRST_NON_CP            = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE0,
+	AMDGPU_VEGA20_DOORBELL64_LAST_NON_CP             = AMDGPU_VEGA20_DOORBELL64_VCE_RING6_7,
+
+	/* kiq/kcq from second XCD. Max 8 XCDs */
+	AMDGPU_VEGA20_DOORBELL_XCC1_KIQ_START             = 0x190,
+	/* 8 compute rings per GC. Max to 0x1CE */
+	AMDGPU_VEGA20_DOORBELL_XCC1_MEC_RING0_START       = 0x197,
+
+	/* AID1 SDMA: 0x1D0 ~ 0x1F7 */
+	AMDGPU_VEGA20_DOORBELL_AID1_sDMA_START           = 0x1D0,
+
+	AMDGPU_VEGA20_DOORBELL_MAX_ASSIGNMENT            = 0x1F7,
+	AMDGPU_VEGA20_DOORBELL_INVALID                   = 0xFFFF
+};
+
+enum AMDGPU_NAVI10_DOORBELL_ASSIGNMENT {
+
+	/* Compute + GFX: 0~255 */
+	AMDGPU_NAVI10_DOORBELL_KIQ			= 0x000,
+	AMDGPU_NAVI10_DOORBELL_HIQ			= 0x001,
+	AMDGPU_NAVI10_DOORBELL_DIQ			= 0x002,
+	AMDGPU_NAVI10_DOORBELL_MEC_RING0		= 0x003,
+	AMDGPU_NAVI10_DOORBELL_MEC_RING1		= 0x004,
+	AMDGPU_NAVI10_DOORBELL_MEC_RING2		= 0x005,
+	AMDGPU_NAVI10_DOORBELL_MEC_RING3		= 0x006,
+	AMDGPU_NAVI10_DOORBELL_MEC_RING4		= 0x007,
+	AMDGPU_NAVI10_DOORBELL_MEC_RING5		= 0x008,
+	AMDGPU_NAVI10_DOORBELL_MEC_RING6		= 0x009,
+	AMDGPU_NAVI10_DOORBELL_MEC_RING7		= 0x00A,
+	AMDGPU_NAVI10_DOORBELL_MES_RING0	        = 0x00B,
+	AMDGPU_NAVI10_DOORBELL_MES_RING1		= 0x00C,
+	AMDGPU_NAVI10_DOORBELL_USERQUEUE_START		= 0x00D,
+	AMDGPU_NAVI10_DOORBELL_USERQUEUE_END		= 0x08A,
+	AMDGPU_NAVI10_DOORBELL_GFX_RING0		= 0x08B,
+	AMDGPU_NAVI10_DOORBELL_GFX_RING1		= 0x08C,
+	AMDGPU_NAVI10_DOORBELL_GFX_USERQUEUE_START	= 0x08D,
+	AMDGPU_NAVI10_DOORBELL_GFX_USERQUEUE_END	= 0x0FF,
+
+	/* SDMA:256~335*/
+	AMDGPU_NAVI10_DOORBELL_sDMA_ENGINE0		= 0x100,
+	AMDGPU_NAVI10_DOORBELL_sDMA_ENGINE1		= 0x10A,
+	AMDGPU_NAVI10_DOORBELL_sDMA_ENGINE2		= 0x114,
+	AMDGPU_NAVI10_DOORBELL_sDMA_ENGINE3		= 0x11E,
+	/* IH: 376~391 */
+	AMDGPU_NAVI10_DOORBELL_IH			= 0x178,
+	/* MMSCH: 392~407
+	 * overlap the doorbell assignment with VCN as they are  mutually exclusive
+	 * VCE engine's doorbell is 32 bit and two VCE ring share one QWORD
+	 */
+	AMDGPU_NAVI10_DOORBELL64_VCN0_1			= 0x188, /* lower 32 bits for VNC0 and upper 32 bits for VNC1 */
+	AMDGPU_NAVI10_DOORBELL64_VCN2_3			= 0x189,
+	AMDGPU_NAVI10_DOORBELL64_VCN4_5			= 0x18A,
+	AMDGPU_NAVI10_DOORBELL64_VCN6_7			= 0x18B,
+
+	AMDGPU_NAVI10_DOORBELL64_VCN8_9			= 0x18C,
+	AMDGPU_NAVI10_DOORBELL64_VCNa_b			= 0x18D,
+	AMDGPU_NAVI10_DOORBELL64_VCNc_d			= 0x18E,
+	AMDGPU_NAVI10_DOORBELL64_VCNe_f			= 0x18F,
+
+	AMDGPU_NAVI10_DOORBELL64_VPE			= 0x190,
+
+	AMDGPU_NAVI10_DOORBELL64_FIRST_NON_CP		= AMDGPU_NAVI10_DOORBELL_sDMA_ENGINE0,
+	AMDGPU_NAVI10_DOORBELL64_LAST_NON_CP		= AMDGPU_NAVI10_DOORBELL64_VPE,
+
+	AMDGPU_NAVI10_DOORBELL_MAX_ASSIGNMENT		= AMDGPU_NAVI10_DOORBELL64_VPE,
+	AMDGPU_NAVI10_DOORBELL_INVALID			= 0xFFFF
+};
+
+/*
+ * 64bit doorbell, offset are in QWORD, occupy 2KB doorbell space
+ */
+enum AMDGPU_DOORBELL64_ASSIGNMENT {
+	/*
+	 * All compute related doorbells: kiq, hiq, diq, traditional compute queue, user queue, should locate in
+	 * a continues range so that programming CP_MEC_DOORBELL_RANGE_LOWER/UPPER can cover this range.
+	 *  Compute related doorbells are allocated from 0x00 to 0x8a
+	 */
+
+
+	/* kernel scheduling */
+	AMDGPU_DOORBELL64_KIQ                     = 0x00,
+
+	/* HSA interface queue and debug queue */
+	AMDGPU_DOORBELL64_HIQ                     = 0x01,
+	AMDGPU_DOORBELL64_DIQ                     = 0x02,
+
+	/* Compute engines */
+	AMDGPU_DOORBELL64_MEC_RING0               = 0x03,
+	AMDGPU_DOORBELL64_MEC_RING1               = 0x04,
+	AMDGPU_DOORBELL64_MEC_RING2               = 0x05,
+	AMDGPU_DOORBELL64_MEC_RING3               = 0x06,
+	AMDGPU_DOORBELL64_MEC_RING4               = 0x07,
+	AMDGPU_DOORBELL64_MEC_RING5               = 0x08,
+	AMDGPU_DOORBELL64_MEC_RING6               = 0x09,
+	AMDGPU_DOORBELL64_MEC_RING7               = 0x0a,
+
+	/* User queue doorbell range (128 doorbells) */
+	AMDGPU_DOORBELL64_USERQUEUE_START         = 0x0b,
+	AMDGPU_DOORBELL64_USERQUEUE_END           = 0x8a,
+
+	/* Graphics engine */
+	AMDGPU_DOORBELL64_GFX_RING0               = 0x8b,
+
+	/*
+	 * Other graphics doorbells can be allocated here: from 0x8c to 0xdf
+	 * Graphics voltage island aperture 1
+	 * default non-graphics QWORD index is 0xe0 - 0xFF inclusive
+	 */
+
+	/* For vega10 sriov, the sdma doorbell must be fixed as follow
+	 * to keep the same setting with host driver, or it will
+	 * happen conflicts
+	 */
+	AMDGPU_DOORBELL64_sDMA_ENGINE0            = 0xF0,
+	AMDGPU_DOORBELL64_sDMA_HI_PRI_ENGINE0     = 0xF1,
+	AMDGPU_DOORBELL64_sDMA_ENGINE1            = 0xF2,
+	AMDGPU_DOORBELL64_sDMA_HI_PRI_ENGINE1     = 0xF3,
+
+	/* Interrupt handler */
+	AMDGPU_DOORBELL64_IH                      = 0xF4,  /* For legacy interrupt ring buffer */
+	AMDGPU_DOORBELL64_IH_RING1                = 0xF5,  /* For page migration request log */
+	AMDGPU_DOORBELL64_IH_RING2                = 0xF6,  /* For page migration translation/invalidation log */
+
+	/* VCN engine use 32 bits doorbell  */
+	AMDGPU_DOORBELL64_VCN0_1                  = 0xF8, /* lower 32 bits for VNC0 and upper 32 bits for VNC1 */
+	AMDGPU_DOORBELL64_VCN2_3                  = 0xF9,
+	AMDGPU_DOORBELL64_VCN4_5                  = 0xFA,
+	AMDGPU_DOORBELL64_VCN6_7                  = 0xFB,
+
+	/* overlap the doorbell assignment with VCN as they are  mutually exclusive
+	 * VCE engine's doorbell is 32 bit and two VCE ring share one QWORD
+	 */
+	AMDGPU_DOORBELL64_UVD_RING0_1             = 0xF8,
+	AMDGPU_DOORBELL64_UVD_RING2_3             = 0xF9,
+	AMDGPU_DOORBELL64_UVD_RING4_5             = 0xFA,
+	AMDGPU_DOORBELL64_UVD_RING6_7             = 0xFB,
+
+	AMDGPU_DOORBELL64_VCE_RING0_1             = 0xFC,
+	AMDGPU_DOORBELL64_VCE_RING2_3             = 0xFD,
+	AMDGPU_DOORBELL64_VCE_RING4_5             = 0xFE,
+	AMDGPU_DOORBELL64_VCE_RING6_7             = 0xFF,
+
+	AMDGPU_DOORBELL64_FIRST_NON_CP            = AMDGPU_DOORBELL64_sDMA_ENGINE0,
+	AMDGPU_DOORBELL64_LAST_NON_CP             = AMDGPU_DOORBELL64_VCE_RING6_7,
+
+	AMDGPU_DOORBELL64_MAX_ASSIGNMENT          = 0xFF,
+	AMDGPU_DOORBELL64_INVALID                 = 0xFFFF
+};
+
+enum AMDGPU_DOORBELL_ASSIGNMENT_LAYOUT1 {
+
+	/* XCC0: 0x00 ~20, XCC1: 20 ~ 2F ... */
+
+	/* KIQ/HIQ/DIQ */
+	AMDGPU_DOORBELL_LAYOUT1_KIQ_START		= 0x000,
+	AMDGPU_DOORBELL_LAYOUT1_HIQ			= 0x001,
+	AMDGPU_DOORBELL_LAYOUT1_DIQ			= 0x002,
+	/* Compute: 0x08 ~ 0x20  */
+	AMDGPU_DOORBELL_LAYOUT1_MEC_RING_START		= 0x008,
+	AMDGPU_DOORBELL_LAYOUT1_MEC_RING_END		= 0x00F,
+	AMDGPU_DOORBELL_LAYOUT1_USERQUEUE_START		= 0x010,
+	AMDGPU_DOORBELL_LAYOUT1_USERQUEUE_END		= 0x01F,
+	AMDGPU_DOORBELL_LAYOUT1_XCC_RANGE		= 0x020,
+
+	/* SDMA: 0x100 ~ 0x19F */
+	AMDGPU_DOORBELL_LAYOUT1_sDMA_ENGINE_START	= 0x100,
+	AMDGPU_DOORBELL_LAYOUT1_sDMA_ENGINE_END		= 0x19F,
+	/* IH: 0x1A0 ~ 0x1AF */
+	AMDGPU_DOORBELL_LAYOUT1_IH                      = 0x1A0,
+	/* VCN: 0x1B0 ~ 0x1E8 */
+	AMDGPU_DOORBELL_LAYOUT1_VCN_START               = 0x1B0,
+	AMDGPU_DOORBELL_LAYOUT1_VCN_END                 = 0x1E8,
+
+	AMDGPU_DOORBELL_LAYOUT1_FIRST_NON_CP		= AMDGPU_DOORBELL_LAYOUT1_sDMA_ENGINE_START,
+	AMDGPU_DOORBELL_LAYOUT1_LAST_NON_CP		= AMDGPU_DOORBELL_LAYOUT1_VCN_END,
+
+	AMDGPU_DOORBELL_LAYOUT1_MAX_ASSIGNMENT          = 0x1E8,
+	AMDGPU_DOORBELL_LAYOUT1_INVALID                 = 0xFFFF
+};
+
+u32 amdgpu_mm_rdoorbell(struct amdgpu_device *adev, u32 index);
+void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v);
+u64 amdgpu_mm_rdoorbell64(struct amdgpu_device *adev, u32 index);
+void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v);
+
+/*
+ * GPU doorbell aperture helpers function.
+ */
+int amdgpu_doorbell_init(struct amdgpu_device *adev);
+void amdgpu_doorbell_fini(struct amdgpu_device *adev);
+int amdgpu_doorbell_create_kernel_doorbells(struct amdgpu_device *adev);
+uint32_t amdgpu_doorbell_index_on_bar(struct amdgpu_device *adev,
+				      struct amdgpu_bo *db_bo,
+				      uint32_t doorbell_index,
+				      uint32_t db_size);
+
+#define RDOORBELL32(index) amdgpu_mm_rdoorbell(adev, (index))
+#define WDOORBELL32(index, v) amdgpu_mm_wdoorbell(adev, (index), (v))
+#define RDOORBELL64(index) amdgpu_mm_rdoorbell64(adev, (index))
+#define WDOORBELL64(index, v) amdgpu_mm_wdoorbell64(adev, (index), (v))
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell_mgr.c
new file mode 100644
index 000000000000..3040437d99c2
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell_mgr.c
@@ -0,0 +1,244 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+
+/**
+ * amdgpu_mm_rdoorbell - read a doorbell dword
+ *
+ * @adev: amdgpu_device pointer
+ * @index: doorbell index
+ *
+ * Returns the value in the doorbell aperture at the
+ * requested doorbell index (CIK).
+ */
+u32 amdgpu_mm_rdoorbell(struct amdgpu_device *adev, u32 index)
+{
+	if (amdgpu_device_skip_hw_access(adev))
+		return 0;
+
+	if (index < adev->doorbell.num_kernel_doorbells)
+		return readl(adev->doorbell.cpu_addr + index);
+
+	dev_err(adev->dev, "reading beyond doorbell aperture: 0x%08x!\n",
+		index);
+	return 0;
+}
+
+/**
+ * amdgpu_mm_wdoorbell - write a doorbell dword
+ *
+ * @adev: amdgpu_device pointer
+ * @index: doorbell index
+ * @v: value to write
+ *
+ * Writes @v to the doorbell aperture at the
+ * requested doorbell index (CIK).
+ */
+void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v)
+{
+	if (amdgpu_device_skip_hw_access(adev))
+		return;
+
+	if (index < adev->doorbell.num_kernel_doorbells)
+		writel(v, adev->doorbell.cpu_addr + index);
+	else
+		dev_err(adev->dev,
+			"writing beyond doorbell aperture: 0x%08x!\n", index);
+}
+
+/**
+ * amdgpu_mm_rdoorbell64 - read a doorbell Qword
+ *
+ * @adev: amdgpu_device pointer
+ * @index: doorbell index
+ *
+ * Returns the value in the doorbell aperture at the
+ * requested doorbell index (VEGA10+).
+ */
+u64 amdgpu_mm_rdoorbell64(struct amdgpu_device *adev, u32 index)
+{
+	if (amdgpu_device_skip_hw_access(adev))
+		return 0;
+
+	if (index < adev->doorbell.num_kernel_doorbells)
+		return atomic64_read((atomic64_t *)(adev->doorbell.cpu_addr + index));
+
+	dev_err(adev->dev, "reading beyond doorbell aperture: 0x%08x!\n",
+		index);
+	return 0;
+}
+
+/**
+ * amdgpu_mm_wdoorbell64 - write a doorbell Qword
+ *
+ * @adev: amdgpu_device pointer
+ * @index: doorbell index
+ * @v: value to write
+ *
+ * Writes @v to the doorbell aperture at the
+ * requested doorbell index (VEGA10+).
+ */
+void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v)
+{
+	if (amdgpu_device_skip_hw_access(adev))
+		return;
+
+	if (index < adev->doorbell.num_kernel_doorbells)
+		atomic64_set((atomic64_t *)(adev->doorbell.cpu_addr + index), v);
+	else
+		dev_err(adev->dev,
+			"writing beyond doorbell aperture: 0x%08x!\n", index);
+}
+
+/**
+ * amdgpu_doorbell_index_on_bar - Find doorbell's absolute offset in BAR
+ *
+ * @adev: amdgpu_device pointer
+ * @db_bo: doorbell object's bo
+ * @doorbell_index: doorbell relative index in this doorbell object
+ * @db_size: doorbell size is in byte
+ *
+ * returns doorbell's absolute index in BAR
+ */
+uint32_t amdgpu_doorbell_index_on_bar(struct amdgpu_device *adev,
+				      struct amdgpu_bo *db_bo,
+				      uint32_t doorbell_index,
+				      uint32_t db_size)
+{
+	int db_bo_offset;
+
+	db_bo_offset = amdgpu_bo_gpu_offset_no_check(db_bo);
+
+	/* doorbell index is 32 bit but doorbell's size can be 32 bit
+	 * or 64 bit, so *db_size(in byte)/4 for alignment.
+	 */
+	return db_bo_offset / sizeof(u32) + doorbell_index *
+	       DIV_ROUND_UP(db_size, 4);
+}
+
+/**
+ * amdgpu_doorbell_create_kernel_doorbells - Create kernel doorbells for graphics
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Creates doorbells for graphics driver usages.
+ * returns 0 on success, error otherwise.
+ */
+int amdgpu_doorbell_create_kernel_doorbells(struct amdgpu_device *adev)
+{
+	int r;
+	int size;
+
+	/* SI HW does not have doorbells, skip allocation */
+	if (adev->doorbell.num_kernel_doorbells == 0)
+		return 0;
+
+	/* Reserve first num_kernel_doorbells (page-aligned) for kernel ops */
+	size = ALIGN(adev->doorbell.num_kernel_doorbells * sizeof(u32), PAGE_SIZE);
+
+	/* Allocate an extra page for MES kernel usages (ring test) */
+	adev->mes.db_start_dw_offset = size / sizeof(u32);
+	size += PAGE_SIZE;
+
+	r = amdgpu_bo_create_kernel(adev,
+				    size,
+				    PAGE_SIZE,
+				    AMDGPU_GEM_DOMAIN_DOORBELL,
+				    &adev->doorbell.kernel_doorbells,
+				    NULL,
+				    (void **)&adev->doorbell.cpu_addr);
+	if (r) {
+		dev_err(adev->dev,
+			"Failed to allocate kernel doorbells, err=%d\n", r);
+		return r;
+	}
+
+	adev->doorbell.num_kernel_doorbells = size / sizeof(u32);
+	return 0;
+}
+
+/*
+ * GPU doorbell aperture helpers function.
+ */
+/**
+ * amdgpu_doorbell_init - Init doorbell driver information.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Init doorbell driver information (CIK)
+ * Returns 0 on success, error on failure.
+ */
+int amdgpu_doorbell_init(struct amdgpu_device *adev)
+{
+
+	/* No doorbell on SI hardware generation */
+	if (adev->asic_type < CHIP_BONAIRE) {
+		adev->doorbell.base = 0;
+		adev->doorbell.size = 0;
+		adev->doorbell.num_kernel_doorbells = 0;
+		return 0;
+	}
+
+	if (pci_resource_flags(adev->pdev, 2) & IORESOURCE_UNSET)
+		return -EINVAL;
+
+	amdgpu_asic_init_doorbell_index(adev);
+
+	/* doorbell bar mapping */
+	adev->doorbell.base = pci_resource_start(adev->pdev, 2);
+	adev->doorbell.size = pci_resource_len(adev->pdev, 2);
+
+	adev->doorbell.num_kernel_doorbells =
+		min_t(u32, adev->doorbell.size / sizeof(u32),
+		      adev->doorbell_index.max_assignment + 1);
+	if (adev->doorbell.num_kernel_doorbells == 0)
+		return -EINVAL;
+
+	/*
+	 * For Vega, reserve and map two pages on doorbell BAR since SDMA
+	 * paging queue doorbell use the second page. The
+	 * AMDGPU_DOORBELL64_MAX_ASSIGNMENT definition assumes all the
+	 * doorbells are in the first page. So with paging queue enabled,
+	 * the max num_kernel_doorbells should + 1 page (0x400 in dword)
+	 */
+	if (adev->asic_type >= CHIP_VEGA10)
+		adev->doorbell.num_kernel_doorbells += 0x400;
+
+	return 0;
+}
+
+/**
+ * amdgpu_doorbell_fini - Tear down doorbell driver information.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tear down doorbell driver information (CIK)
+ */
+void amdgpu_doorbell_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->doorbell.kernel_doorbells,
+			      NULL,
+			      (void **)&adev->doorbell.cpu_addr);
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.c
deleted file mode 100644
index 1cb52fd19060..000000000000
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.c
+++ /dev/null
@@ -1,969 +0,0 @@
-/*
- * Copyright 2011 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- * Authors: Alex Deucher
- */
-
-#include <drm/drmP.h>
-#include "amdgpu.h"
-#include "amdgpu_atombios.h"
-#include "amdgpu_i2c.h"
-#include "amdgpu_dpm.h"
-#include "atom.h"
-
-void amdgpu_dpm_print_class_info(u32 class, u32 class2)
-{
-	const char *s;
-
-	switch (class & ATOM_PPLIB_CLASSIFICATION_UI_MASK) {
-	case ATOM_PPLIB_CLASSIFICATION_UI_NONE:
-	default:
-		s = "none";
-		break;
-	case ATOM_PPLIB_CLASSIFICATION_UI_BATTERY:
-		s = "battery";
-		break;
-	case ATOM_PPLIB_CLASSIFICATION_UI_BALANCED:
-		s = "balanced";
-		break;
-	case ATOM_PPLIB_CLASSIFICATION_UI_PERFORMANCE:
-		s = "performance";
-		break;
-	}
-	printk("\tui class: %s\n", s);
-	printk("\tinternal class:");
-	if (((class & ~ATOM_PPLIB_CLASSIFICATION_UI_MASK) == 0) &&
-	    (class2 == 0))
-		pr_cont(" none");
-	else {
-		if (class & ATOM_PPLIB_CLASSIFICATION_BOOT)
-			pr_cont(" boot");
-		if (class & ATOM_PPLIB_CLASSIFICATION_THERMAL)
-			pr_cont(" thermal");
-		if (class & ATOM_PPLIB_CLASSIFICATION_LIMITEDPOWERSOURCE)
-			pr_cont(" limited_pwr");
-		if (class & ATOM_PPLIB_CLASSIFICATION_REST)
-			pr_cont(" rest");
-		if (class & ATOM_PPLIB_CLASSIFICATION_FORCED)
-			pr_cont(" forced");
-		if (class & ATOM_PPLIB_CLASSIFICATION_3DPERFORMANCE)
-			pr_cont(" 3d_perf");
-		if (class & ATOM_PPLIB_CLASSIFICATION_OVERDRIVETEMPLATE)
-			pr_cont(" ovrdrv");
-		if (class & ATOM_PPLIB_CLASSIFICATION_UVDSTATE)
-			pr_cont(" uvd");
-		if (class & ATOM_PPLIB_CLASSIFICATION_3DLOW)
-			pr_cont(" 3d_low");
-		if (class & ATOM_PPLIB_CLASSIFICATION_ACPI)
-			pr_cont(" acpi");
-		if (class & ATOM_PPLIB_CLASSIFICATION_HD2STATE)
-			pr_cont(" uvd_hd2");
-		if (class & ATOM_PPLIB_CLASSIFICATION_HDSTATE)
-			pr_cont(" uvd_hd");
-		if (class & ATOM_PPLIB_CLASSIFICATION_SDSTATE)
-			pr_cont(" uvd_sd");
-		if (class2 & ATOM_PPLIB_CLASSIFICATION2_LIMITEDPOWERSOURCE_2)
-			pr_cont(" limited_pwr2");
-		if (class2 & ATOM_PPLIB_CLASSIFICATION2_ULV)
-			pr_cont(" ulv");
-		if (class2 & ATOM_PPLIB_CLASSIFICATION2_MVC)
-			pr_cont(" uvd_mvc");
-	}
-	pr_cont("\n");
-}
-
-void amdgpu_dpm_print_cap_info(u32 caps)
-{
-	printk("\tcaps:");
-	if (caps & ATOM_PPLIB_SINGLE_DISPLAY_ONLY)
-		pr_cont(" single_disp");
-	if (caps & ATOM_PPLIB_SUPPORTS_VIDEO_PLAYBACK)
-		pr_cont(" video");
-	if (caps & ATOM_PPLIB_DISALLOW_ON_DC)
-		pr_cont(" no_dc");
-	pr_cont("\n");
-}
-
-void amdgpu_dpm_print_ps_status(struct amdgpu_device *adev,
-				struct amdgpu_ps *rps)
-{
-	printk("\tstatus:");
-	if (rps == adev->pm.dpm.current_ps)
-		pr_cont(" c");
-	if (rps == adev->pm.dpm.requested_ps)
-		pr_cont(" r");
-	if (rps == adev->pm.dpm.boot_ps)
-		pr_cont(" b");
-	pr_cont("\n");
-}
-
-
-u32 amdgpu_dpm_get_vblank_time(struct amdgpu_device *adev)
-{
-	struct drm_device *dev = adev->ddev;
-	struct drm_crtc *crtc;
-	struct amdgpu_crtc *amdgpu_crtc;
-	u32 vblank_in_pixels;
-	u32 vblank_time_us = 0xffffffff; /* if the displays are off, vblank time is max */
-
-	if (adev->mode_info.num_crtc && adev->mode_info.mode_config_initialized) {
-		list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
-			amdgpu_crtc = to_amdgpu_crtc(crtc);
-			if (crtc->enabled && amdgpu_crtc->enabled && amdgpu_crtc->hw_mode.clock) {
-				vblank_in_pixels =
-					amdgpu_crtc->hw_mode.crtc_htotal *
-					(amdgpu_crtc->hw_mode.crtc_vblank_end -
-					amdgpu_crtc->hw_mode.crtc_vdisplay +
-					(amdgpu_crtc->v_border * 2));
-
-				vblank_time_us = vblank_in_pixels * 1000 / amdgpu_crtc->hw_mode.clock;
-				break;
-			}
-		}
-	}
-
-	return vblank_time_us;
-}
-
-u32 amdgpu_dpm_get_vrefresh(struct amdgpu_device *adev)
-{
-	struct drm_device *dev = adev->ddev;
-	struct drm_crtc *crtc;
-	struct amdgpu_crtc *amdgpu_crtc;
-	u32 vrefresh = 0;
-
-	if (adev->mode_info.num_crtc && adev->mode_info.mode_config_initialized) {
-		list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
-			amdgpu_crtc = to_amdgpu_crtc(crtc);
-			if (crtc->enabled && amdgpu_crtc->enabled && amdgpu_crtc->hw_mode.clock) {
-				vrefresh = drm_mode_vrefresh(&amdgpu_crtc->hw_mode);
-				break;
-			}
-		}
-	}
-
-	return vrefresh;
-}
-
-void amdgpu_calculate_u_and_p(u32 i, u32 r_c, u32 p_b,
-			      u32 *p, u32 *u)
-{
-	u32 b_c = 0;
-	u32 i_c;
-	u32 tmp;
-
-	i_c = (i * r_c) / 100;
-	tmp = i_c >> p_b;
-
-	while (tmp) {
-		b_c++;
-		tmp >>= 1;
-	}
-
-	*u = (b_c + 1) / 2;
-	*p = i_c / (1 << (2 * (*u)));
-}
-
-int amdgpu_calculate_at(u32 t, u32 h, u32 fh, u32 fl, u32 *tl, u32 *th)
-{
-	u32 k, a, ah, al;
-	u32 t1;
-
-	if ((fl == 0) || (fh == 0) || (fl > fh))
-		return -EINVAL;
-
-	k = (100 * fh) / fl;
-	t1 = (t * (k - 100));
-	a = (1000 * (100 * h + t1)) / (10000 + (t1 / 100));
-	a = (a + 5) / 10;
-	ah = ((a * t) + 5000) / 10000;
-	al = a - ah;
-
-	*th = t - ah;
-	*tl = t + al;
-
-	return 0;
-}
-
-bool amdgpu_is_uvd_state(u32 class, u32 class2)
-{
-	if (class & ATOM_PPLIB_CLASSIFICATION_UVDSTATE)
-		return true;
-	if (class & ATOM_PPLIB_CLASSIFICATION_HD2STATE)
-		return true;
-	if (class & ATOM_PPLIB_CLASSIFICATION_HDSTATE)
-		return true;
-	if (class & ATOM_PPLIB_CLASSIFICATION_SDSTATE)
-		return true;
-	if (class2 & ATOM_PPLIB_CLASSIFICATION2_MVC)
-		return true;
-	return false;
-}
-
-bool amdgpu_is_internal_thermal_sensor(enum amdgpu_int_thermal_type sensor)
-{
-	switch (sensor) {
-	case THERMAL_TYPE_RV6XX:
-	case THERMAL_TYPE_RV770:
-	case THERMAL_TYPE_EVERGREEN:
-	case THERMAL_TYPE_SUMO:
-	case THERMAL_TYPE_NI:
-	case THERMAL_TYPE_SI:
-	case THERMAL_TYPE_CI:
-	case THERMAL_TYPE_KV:
-		return true;
-	case THERMAL_TYPE_ADT7473_WITH_INTERNAL:
-	case THERMAL_TYPE_EMC2103_WITH_INTERNAL:
-		return false; /* need special handling */
-	case THERMAL_TYPE_NONE:
-	case THERMAL_TYPE_EXTERNAL:
-	case THERMAL_TYPE_EXTERNAL_GPIO:
-	default:
-		return false;
-	}
-}
-
-union power_info {
-	struct _ATOM_POWERPLAY_INFO info;
-	struct _ATOM_POWERPLAY_INFO_V2 info_2;
-	struct _ATOM_POWERPLAY_INFO_V3 info_3;
-	struct _ATOM_PPLIB_POWERPLAYTABLE pplib;
-	struct _ATOM_PPLIB_POWERPLAYTABLE2 pplib2;
-	struct _ATOM_PPLIB_POWERPLAYTABLE3 pplib3;
-	struct _ATOM_PPLIB_POWERPLAYTABLE4 pplib4;
-	struct _ATOM_PPLIB_POWERPLAYTABLE5 pplib5;
-};
-
-union fan_info {
-	struct _ATOM_PPLIB_FANTABLE fan;
-	struct _ATOM_PPLIB_FANTABLE2 fan2;
-	struct _ATOM_PPLIB_FANTABLE3 fan3;
-};
-
-static int amdgpu_parse_clk_voltage_dep_table(struct amdgpu_clock_voltage_dependency_table *amdgpu_table,
-					      ATOM_PPLIB_Clock_Voltage_Dependency_Table *atom_table)
-{
-	u32 size = atom_table->ucNumEntries *
-		sizeof(struct amdgpu_clock_voltage_dependency_entry);
-	int i;
-	ATOM_PPLIB_Clock_Voltage_Dependency_Record *entry;
-
-	amdgpu_table->entries = kzalloc(size, GFP_KERNEL);
-	if (!amdgpu_table->entries)
-		return -ENOMEM;
-
-	entry = &atom_table->entries[0];
-	for (i = 0; i < atom_table->ucNumEntries; i++) {
-		amdgpu_table->entries[i].clk = le16_to_cpu(entry->usClockLow) |
-			(entry->ucClockHigh << 16);
-		amdgpu_table->entries[i].v = le16_to_cpu(entry->usVoltage);
-		entry = (ATOM_PPLIB_Clock_Voltage_Dependency_Record *)
-			((u8 *)entry + sizeof(ATOM_PPLIB_Clock_Voltage_Dependency_Record));
-	}
-	amdgpu_table->count = atom_table->ucNumEntries;
-
-	return 0;
-}
-
-int amdgpu_get_platform_caps(struct amdgpu_device *adev)
-{
-	struct amdgpu_mode_info *mode_info = &adev->mode_info;
-	union power_info *power_info;
-	int index = GetIndexIntoMasterTable(DATA, PowerPlayInfo);
-	u16 data_offset;
-	u8 frev, crev;
-
-	if (!amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
-				   &frev, &crev, &data_offset))
-		return -EINVAL;
-	power_info = (union power_info *)(mode_info->atom_context->bios + data_offset);
-
-	adev->pm.dpm.platform_caps = le32_to_cpu(power_info->pplib.ulPlatformCaps);
-	adev->pm.dpm.backbias_response_time = le16_to_cpu(power_info->pplib.usBackbiasTime);
-	adev->pm.dpm.voltage_response_time = le16_to_cpu(power_info->pplib.usVoltageTime);
-
-	return 0;
-}
-
-/* sizeof(ATOM_PPLIB_EXTENDEDHEADER) */
-#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V2 12
-#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V3 14
-#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V4 16
-#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V5 18
-#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V6 20
-#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V7 22
-#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V8 24
-#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V9 26
-
-int amdgpu_parse_extended_power_table(struct amdgpu_device *adev)
-{
-	struct amdgpu_mode_info *mode_info = &adev->mode_info;
-	union power_info *power_info;
-	union fan_info *fan_info;
-	ATOM_PPLIB_Clock_Voltage_Dependency_Table *dep_table;
-	int index = GetIndexIntoMasterTable(DATA, PowerPlayInfo);
-	u16 data_offset;
-	u8 frev, crev;
-	int ret, i;
-
-	if (!amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
-				   &frev, &crev, &data_offset))
-		return -EINVAL;
-	power_info = (union power_info *)(mode_info->atom_context->bios + data_offset);
-
-	/* fan table */
-	if (le16_to_cpu(power_info->pplib.usTableSize) >=
-	    sizeof(struct _ATOM_PPLIB_POWERPLAYTABLE3)) {
-		if (power_info->pplib3.usFanTableOffset) {
-			fan_info = (union fan_info *)(mode_info->atom_context->bios + data_offset +
-						      le16_to_cpu(power_info->pplib3.usFanTableOffset));
-			adev->pm.dpm.fan.t_hyst = fan_info->fan.ucTHyst;
-			adev->pm.dpm.fan.t_min = le16_to_cpu(fan_info->fan.usTMin);
-			adev->pm.dpm.fan.t_med = le16_to_cpu(fan_info->fan.usTMed);
-			adev->pm.dpm.fan.t_high = le16_to_cpu(fan_info->fan.usTHigh);
-			adev->pm.dpm.fan.pwm_min = le16_to_cpu(fan_info->fan.usPWMMin);
-			adev->pm.dpm.fan.pwm_med = le16_to_cpu(fan_info->fan.usPWMMed);
-			adev->pm.dpm.fan.pwm_high = le16_to_cpu(fan_info->fan.usPWMHigh);
-			if (fan_info->fan.ucFanTableFormat >= 2)
-				adev->pm.dpm.fan.t_max = le16_to_cpu(fan_info->fan2.usTMax);
-			else
-				adev->pm.dpm.fan.t_max = 10900;
-			adev->pm.dpm.fan.cycle_delay = 100000;
-			if (fan_info->fan.ucFanTableFormat >= 3) {
-				adev->pm.dpm.fan.control_mode = fan_info->fan3.ucFanControlMode;
-				adev->pm.dpm.fan.default_max_fan_pwm =
-					le16_to_cpu(fan_info->fan3.usFanPWMMax);
-				adev->pm.dpm.fan.default_fan_output_sensitivity = 4836;
-				adev->pm.dpm.fan.fan_output_sensitivity =
-					le16_to_cpu(fan_info->fan3.usFanOutputSensitivity);
-			}
-			adev->pm.dpm.fan.ucode_fan_control = true;
-		}
-	}
-
-	/* clock dependancy tables, shedding tables */
-	if (le16_to_cpu(power_info->pplib.usTableSize) >=
-	    sizeof(struct _ATOM_PPLIB_POWERPLAYTABLE4)) {
-		if (power_info->pplib4.usVddcDependencyOnSCLKOffset) {
-			dep_table = (ATOM_PPLIB_Clock_Voltage_Dependency_Table *)
-				(mode_info->atom_context->bios + data_offset +
-				 le16_to_cpu(power_info->pplib4.usVddcDependencyOnSCLKOffset));
-			ret = amdgpu_parse_clk_voltage_dep_table(&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk,
-								 dep_table);
-			if (ret) {
-				amdgpu_free_extended_power_table(adev);
-				return ret;
-			}
-		}
-		if (power_info->pplib4.usVddciDependencyOnMCLKOffset) {
-			dep_table = (ATOM_PPLIB_Clock_Voltage_Dependency_Table *)
-				(mode_info->atom_context->bios + data_offset +
-				 le16_to_cpu(power_info->pplib4.usVddciDependencyOnMCLKOffset));
-			ret = amdgpu_parse_clk_voltage_dep_table(&adev->pm.dpm.dyn_state.vddci_dependency_on_mclk,
-								 dep_table);
-			if (ret) {
-				amdgpu_free_extended_power_table(adev);
-				return ret;
-			}
-		}
-		if (power_info->pplib4.usVddcDependencyOnMCLKOffset) {
-			dep_table = (ATOM_PPLIB_Clock_Voltage_Dependency_Table *)
-				(mode_info->atom_context->bios + data_offset +
-				 le16_to_cpu(power_info->pplib4.usVddcDependencyOnMCLKOffset));
-			ret = amdgpu_parse_clk_voltage_dep_table(&adev->pm.dpm.dyn_state.vddc_dependency_on_mclk,
-								 dep_table);
-			if (ret) {
-				amdgpu_free_extended_power_table(adev);
-				return ret;
-			}
-		}
-		if (power_info->pplib4.usMvddDependencyOnMCLKOffset) {
-			dep_table = (ATOM_PPLIB_Clock_Voltage_Dependency_Table *)
-				(mode_info->atom_context->bios + data_offset +
-				 le16_to_cpu(power_info->pplib4.usMvddDependencyOnMCLKOffset));
-			ret = amdgpu_parse_clk_voltage_dep_table(&adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk,
-								 dep_table);
-			if (ret) {
-				amdgpu_free_extended_power_table(adev);
-				return ret;
-			}
-		}
-		if (power_info->pplib4.usMaxClockVoltageOnDCOffset) {
-			ATOM_PPLIB_Clock_Voltage_Limit_Table *clk_v =
-				(ATOM_PPLIB_Clock_Voltage_Limit_Table *)
-				(mode_info->atom_context->bios + data_offset +
-				 le16_to_cpu(power_info->pplib4.usMaxClockVoltageOnDCOffset));
-			if (clk_v->ucNumEntries) {
-				adev->pm.dpm.dyn_state.max_clock_voltage_on_dc.sclk =
-					le16_to_cpu(clk_v->entries[0].usSclkLow) |
-					(clk_v->entries[0].ucSclkHigh << 16);
-				adev->pm.dpm.dyn_state.max_clock_voltage_on_dc.mclk =
-					le16_to_cpu(clk_v->entries[0].usMclkLow) |
-					(clk_v->entries[0].ucMclkHigh << 16);
-				adev->pm.dpm.dyn_state.max_clock_voltage_on_dc.vddc =
-					le16_to_cpu(clk_v->entries[0].usVddc);
-				adev->pm.dpm.dyn_state.max_clock_voltage_on_dc.vddci =
-					le16_to_cpu(clk_v->entries[0].usVddci);
-			}
-		}
-		if (power_info->pplib4.usVddcPhaseShedLimitsTableOffset) {
-			ATOM_PPLIB_PhaseSheddingLimits_Table *psl =
-				(ATOM_PPLIB_PhaseSheddingLimits_Table *)
-				(mode_info->atom_context->bios + data_offset +
-				 le16_to_cpu(power_info->pplib4.usVddcPhaseShedLimitsTableOffset));
-			ATOM_PPLIB_PhaseSheddingLimits_Record *entry;
-
-			adev->pm.dpm.dyn_state.phase_shedding_limits_table.entries =
-				kzalloc(psl->ucNumEntries *
-					sizeof(struct amdgpu_phase_shedding_limits_entry),
-					GFP_KERNEL);
-			if (!adev->pm.dpm.dyn_state.phase_shedding_limits_table.entries) {
-				amdgpu_free_extended_power_table(adev);
-				return -ENOMEM;
-			}
-
-			entry = &psl->entries[0];
-			for (i = 0; i < psl->ucNumEntries; i++) {
-				adev->pm.dpm.dyn_state.phase_shedding_limits_table.entries[i].sclk =
-					le16_to_cpu(entry->usSclkLow) | (entry->ucSclkHigh << 16);
-				adev->pm.dpm.dyn_state.phase_shedding_limits_table.entries[i].mclk =
-					le16_to_cpu(entry->usMclkLow) | (entry->ucMclkHigh << 16);
-				adev->pm.dpm.dyn_state.phase_shedding_limits_table.entries[i].voltage =
-					le16_to_cpu(entry->usVoltage);
-				entry = (ATOM_PPLIB_PhaseSheddingLimits_Record *)
-					((u8 *)entry + sizeof(ATOM_PPLIB_PhaseSheddingLimits_Record));
-			}
-			adev->pm.dpm.dyn_state.phase_shedding_limits_table.count =
-				psl->ucNumEntries;
-		}
-	}
-
-	/* cac data */
-	if (le16_to_cpu(power_info->pplib.usTableSize) >=
-	    sizeof(struct _ATOM_PPLIB_POWERPLAYTABLE5)) {
-		adev->pm.dpm.tdp_limit = le32_to_cpu(power_info->pplib5.ulTDPLimit);
-		adev->pm.dpm.near_tdp_limit = le32_to_cpu(power_info->pplib5.ulNearTDPLimit);
-		adev->pm.dpm.near_tdp_limit_adjusted = adev->pm.dpm.near_tdp_limit;
-		adev->pm.dpm.tdp_od_limit = le16_to_cpu(power_info->pplib5.usTDPODLimit);
-		if (adev->pm.dpm.tdp_od_limit)
-			adev->pm.dpm.power_control = true;
-		else
-			adev->pm.dpm.power_control = false;
-		adev->pm.dpm.tdp_adjustment = 0;
-		adev->pm.dpm.sq_ramping_threshold = le32_to_cpu(power_info->pplib5.ulSQRampingThreshold);
-		adev->pm.dpm.cac_leakage = le32_to_cpu(power_info->pplib5.ulCACLeakage);
-		adev->pm.dpm.load_line_slope = le16_to_cpu(power_info->pplib5.usLoadLineSlope);
-		if (power_info->pplib5.usCACLeakageTableOffset) {
-			ATOM_PPLIB_CAC_Leakage_Table *cac_table =
-				(ATOM_PPLIB_CAC_Leakage_Table *)
-				(mode_info->atom_context->bios + data_offset +
-				 le16_to_cpu(power_info->pplib5.usCACLeakageTableOffset));
-			ATOM_PPLIB_CAC_Leakage_Record *entry;
-			u32 size = cac_table->ucNumEntries * sizeof(struct amdgpu_cac_leakage_table);
-			adev->pm.dpm.dyn_state.cac_leakage_table.entries = kzalloc(size, GFP_KERNEL);
-			if (!adev->pm.dpm.dyn_state.cac_leakage_table.entries) {
-				amdgpu_free_extended_power_table(adev);
-				return -ENOMEM;
-			}
-			entry = &cac_table->entries[0];
-			for (i = 0; i < cac_table->ucNumEntries; i++) {
-				if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_EVV) {
-					adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc1 =
-						le16_to_cpu(entry->usVddc1);
-					adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc2 =
-						le16_to_cpu(entry->usVddc2);
-					adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc3 =
-						le16_to_cpu(entry->usVddc3);
-				} else {
-					adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc =
-						le16_to_cpu(entry->usVddc);
-					adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].leakage =
-						le32_to_cpu(entry->ulLeakageValue);
-				}
-				entry = (ATOM_PPLIB_CAC_Leakage_Record *)
-					((u8 *)entry + sizeof(ATOM_PPLIB_CAC_Leakage_Record));
-			}
-			adev->pm.dpm.dyn_state.cac_leakage_table.count = cac_table->ucNumEntries;
-		}
-	}
-
-	/* ext tables */
-	if (le16_to_cpu(power_info->pplib.usTableSize) >=
-	    sizeof(struct _ATOM_PPLIB_POWERPLAYTABLE3)) {
-		ATOM_PPLIB_EXTENDEDHEADER *ext_hdr = (ATOM_PPLIB_EXTENDEDHEADER *)
-			(mode_info->atom_context->bios + data_offset +
-			 le16_to_cpu(power_info->pplib3.usExtendendedHeaderOffset));
-		if ((le16_to_cpu(ext_hdr->usSize) >= SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V2) &&
-			ext_hdr->usVCETableOffset) {
-			VCEClockInfoArray *array = (VCEClockInfoArray *)
-				(mode_info->atom_context->bios + data_offset +
-				 le16_to_cpu(ext_hdr->usVCETableOffset) + 1);
-			ATOM_PPLIB_VCE_Clock_Voltage_Limit_Table *limits =
-				(ATOM_PPLIB_VCE_Clock_Voltage_Limit_Table *)
-				(mode_info->atom_context->bios + data_offset +
-				 le16_to_cpu(ext_hdr->usVCETableOffset) + 1 +
-				 1 + array->ucNumEntries * sizeof(VCEClockInfo));
-			ATOM_PPLIB_VCE_State_Table *states =
-				(ATOM_PPLIB_VCE_State_Table *)
-				(mode_info->atom_context->bios + data_offset +
-				 le16_to_cpu(ext_hdr->usVCETableOffset) + 1 +
-				 1 + (array->ucNumEntries * sizeof (VCEClockInfo)) +
-				 1 + (limits->numEntries * sizeof(ATOM_PPLIB_VCE_Clock_Voltage_Limit_Record)));
-			ATOM_PPLIB_VCE_Clock_Voltage_Limit_Record *entry;
-			ATOM_PPLIB_VCE_State_Record *state_entry;
-			VCEClockInfo *vce_clk;
-			u32 size = limits->numEntries *
-				sizeof(struct amdgpu_vce_clock_voltage_dependency_entry);
-			adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries =
-				kzalloc(size, GFP_KERNEL);
-			if (!adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries) {
-				amdgpu_free_extended_power_table(adev);
-				return -ENOMEM;
-			}
-			adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.count =
-				limits->numEntries;
-			entry = &limits->entries[0];
-			state_entry = &states->entries[0];
-			for (i = 0; i < limits->numEntries; i++) {
-				vce_clk = (VCEClockInfo *)
-					((u8 *)&array->entries[0] +
-					 (entry->ucVCEClockInfoIndex * sizeof(VCEClockInfo)));
-				adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries[i].evclk =
-					le16_to_cpu(vce_clk->usEVClkLow) | (vce_clk->ucEVClkHigh << 16);
-				adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries[i].ecclk =
-					le16_to_cpu(vce_clk->usECClkLow) | (vce_clk->ucECClkHigh << 16);
-				adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries[i].v =
-					le16_to_cpu(entry->usVoltage);
-				entry = (ATOM_PPLIB_VCE_Clock_Voltage_Limit_Record *)
-					((u8 *)entry + sizeof(ATOM_PPLIB_VCE_Clock_Voltage_Limit_Record));
-			}
-			adev->pm.dpm.num_of_vce_states =
-					states->numEntries > AMD_MAX_VCE_LEVELS ?
-					AMD_MAX_VCE_LEVELS : states->numEntries;
-			for (i = 0; i < adev->pm.dpm.num_of_vce_states; i++) {
-				vce_clk = (VCEClockInfo *)
-					((u8 *)&array->entries[0] +
-					 (state_entry->ucVCEClockInfoIndex * sizeof(VCEClockInfo)));
-				adev->pm.dpm.vce_states[i].evclk =
-					le16_to_cpu(vce_clk->usEVClkLow) | (vce_clk->ucEVClkHigh << 16);
-				adev->pm.dpm.vce_states[i].ecclk =
-					le16_to_cpu(vce_clk->usECClkLow) | (vce_clk->ucECClkHigh << 16);
-				adev->pm.dpm.vce_states[i].clk_idx =
-					state_entry->ucClockInfoIndex & 0x3f;
-				adev->pm.dpm.vce_states[i].pstate =
-					(state_entry->ucClockInfoIndex & 0xc0) >> 6;
-				state_entry = (ATOM_PPLIB_VCE_State_Record *)
-					((u8 *)state_entry + sizeof(ATOM_PPLIB_VCE_State_Record));
-			}
-		}
-		if ((le16_to_cpu(ext_hdr->usSize) >= SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V3) &&
-			ext_hdr->usUVDTableOffset) {
-			UVDClockInfoArray *array = (UVDClockInfoArray *)
-				(mode_info->atom_context->bios + data_offset +
-				 le16_to_cpu(ext_hdr->usUVDTableOffset) + 1);
-			ATOM_PPLIB_UVD_Clock_Voltage_Limit_Table *limits =
-				(ATOM_PPLIB_UVD_Clock_Voltage_Limit_Table *)
-				(mode_info->atom_context->bios + data_offset +
-				 le16_to_cpu(ext_hdr->usUVDTableOffset) + 1 +
-				 1 + (array->ucNumEntries * sizeof (UVDClockInfo)));
-			ATOM_PPLIB_UVD_Clock_Voltage_Limit_Record *entry;
-			u32 size = limits->numEntries *
-				sizeof(struct amdgpu_uvd_clock_voltage_dependency_entry);
-			adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries =
-				kzalloc(size, GFP_KERNEL);
-			if (!adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries) {
-				amdgpu_free_extended_power_table(adev);
-				return -ENOMEM;
-			}
-			adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.count =
-				limits->numEntries;
-			entry = &limits->entries[0];
-			for (i = 0; i < limits->numEntries; i++) {
-				UVDClockInfo *uvd_clk = (UVDClockInfo *)
-					((u8 *)&array->entries[0] +
-					 (entry->ucUVDClockInfoIndex * sizeof(UVDClockInfo)));
-				adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries[i].vclk =
-					le16_to_cpu(uvd_clk->usVClkLow) | (uvd_clk->ucVClkHigh << 16);
-				adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries[i].dclk =
-					le16_to_cpu(uvd_clk->usDClkLow) | (uvd_clk->ucDClkHigh << 16);
-				adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries[i].v =
-					le16_to_cpu(entry->usVoltage);
-				entry = (ATOM_PPLIB_UVD_Clock_Voltage_Limit_Record *)
-					((u8 *)entry + sizeof(ATOM_PPLIB_UVD_Clock_Voltage_Limit_Record));
-			}
-		}
-		if ((le16_to_cpu(ext_hdr->usSize) >= SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V4) &&
-			ext_hdr->usSAMUTableOffset) {
-			ATOM_PPLIB_SAMClk_Voltage_Limit_Table *limits =
-				(ATOM_PPLIB_SAMClk_Voltage_Limit_Table *)
-				(mode_info->atom_context->bios + data_offset +
-				 le16_to_cpu(ext_hdr->usSAMUTableOffset) + 1);
-			ATOM_PPLIB_SAMClk_Voltage_Limit_Record *entry;
-			u32 size = limits->numEntries *
-				sizeof(struct amdgpu_clock_voltage_dependency_entry);
-			adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.entries =
-				kzalloc(size, GFP_KERNEL);
-			if (!adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.entries) {
-				amdgpu_free_extended_power_table(adev);
-				return -ENOMEM;
-			}
-			adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.count =
-				limits->numEntries;
-			entry = &limits->entries[0];
-			for (i = 0; i < limits->numEntries; i++) {
-				adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.entries[i].clk =
-					le16_to_cpu(entry->usSAMClockLow) | (entry->ucSAMClockHigh << 16);
-				adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.entries[i].v =
-					le16_to_cpu(entry->usVoltage);
-				entry = (ATOM_PPLIB_SAMClk_Voltage_Limit_Record *)
-					((u8 *)entry + sizeof(ATOM_PPLIB_SAMClk_Voltage_Limit_Record));
-			}
-		}
-		if ((le16_to_cpu(ext_hdr->usSize) >= SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V5) &&
-		    ext_hdr->usPPMTableOffset) {
-			ATOM_PPLIB_PPM_Table *ppm = (ATOM_PPLIB_PPM_Table *)
-				(mode_info->atom_context->bios + data_offset +
-				 le16_to_cpu(ext_hdr->usPPMTableOffset));
-			adev->pm.dpm.dyn_state.ppm_table =
-				kzalloc(sizeof(struct amdgpu_ppm_table), GFP_KERNEL);
-			if (!adev->pm.dpm.dyn_state.ppm_table) {
-				amdgpu_free_extended_power_table(adev);
-				return -ENOMEM;
-			}
-			adev->pm.dpm.dyn_state.ppm_table->ppm_design = ppm->ucPpmDesign;
-			adev->pm.dpm.dyn_state.ppm_table->cpu_core_number =
-				le16_to_cpu(ppm->usCpuCoreNumber);
-			adev->pm.dpm.dyn_state.ppm_table->platform_tdp =
-				le32_to_cpu(ppm->ulPlatformTDP);
-			adev->pm.dpm.dyn_state.ppm_table->small_ac_platform_tdp =
-				le32_to_cpu(ppm->ulSmallACPlatformTDP);
-			adev->pm.dpm.dyn_state.ppm_table->platform_tdc =
-				le32_to_cpu(ppm->ulPlatformTDC);
-			adev->pm.dpm.dyn_state.ppm_table->small_ac_platform_tdc =
-				le32_to_cpu(ppm->ulSmallACPlatformTDC);
-			adev->pm.dpm.dyn_state.ppm_table->apu_tdp =
-				le32_to_cpu(ppm->ulApuTDP);
-			adev->pm.dpm.dyn_state.ppm_table->dgpu_tdp =
-				le32_to_cpu(ppm->ulDGpuTDP);
-			adev->pm.dpm.dyn_state.ppm_table->dgpu_ulv_power =
-				le32_to_cpu(ppm->ulDGpuUlvPower);
-			adev->pm.dpm.dyn_state.ppm_table->tj_max =
-				le32_to_cpu(ppm->ulTjmax);
-		}
-		if ((le16_to_cpu(ext_hdr->usSize) >= SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V6) &&
-			ext_hdr->usACPTableOffset) {
-			ATOM_PPLIB_ACPClk_Voltage_Limit_Table *limits =
-				(ATOM_PPLIB_ACPClk_Voltage_Limit_Table *)
-				(mode_info->atom_context->bios + data_offset +
-				 le16_to_cpu(ext_hdr->usACPTableOffset) + 1);
-			ATOM_PPLIB_ACPClk_Voltage_Limit_Record *entry;
-			u32 size = limits->numEntries *
-				sizeof(struct amdgpu_clock_voltage_dependency_entry);
-			adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.entries =
-				kzalloc(size, GFP_KERNEL);
-			if (!adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.entries) {
-				amdgpu_free_extended_power_table(adev);
-				return -ENOMEM;
-			}
-			adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.count =
-				limits->numEntries;
-			entry = &limits->entries[0];
-			for (i = 0; i < limits->numEntries; i++) {
-				adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.entries[i].clk =
-					le16_to_cpu(entry->usACPClockLow) | (entry->ucACPClockHigh << 16);
-				adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.entries[i].v =
-					le16_to_cpu(entry->usVoltage);
-				entry = (ATOM_PPLIB_ACPClk_Voltage_Limit_Record *)
-					((u8 *)entry + sizeof(ATOM_PPLIB_ACPClk_Voltage_Limit_Record));
-			}
-		}
-		if ((le16_to_cpu(ext_hdr->usSize) >= SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V7) &&
-			ext_hdr->usPowerTuneTableOffset) {
-			u8 rev = *(u8 *)(mode_info->atom_context->bios + data_offset +
-					 le16_to_cpu(ext_hdr->usPowerTuneTableOffset));
-			ATOM_PowerTune_Table *pt;
-			adev->pm.dpm.dyn_state.cac_tdp_table =
-				kzalloc(sizeof(struct amdgpu_cac_tdp_table), GFP_KERNEL);
-			if (!adev->pm.dpm.dyn_state.cac_tdp_table) {
-				amdgpu_free_extended_power_table(adev);
-				return -ENOMEM;
-			}
-			if (rev > 0) {
-				ATOM_PPLIB_POWERTUNE_Table_V1 *ppt = (ATOM_PPLIB_POWERTUNE_Table_V1 *)
-					(mode_info->atom_context->bios + data_offset +
-					 le16_to_cpu(ext_hdr->usPowerTuneTableOffset));
-				adev->pm.dpm.dyn_state.cac_tdp_table->maximum_power_delivery_limit =
-					ppt->usMaximumPowerDeliveryLimit;
-				pt = &ppt->power_tune_table;
-			} else {
-				ATOM_PPLIB_POWERTUNE_Table *ppt = (ATOM_PPLIB_POWERTUNE_Table *)
-					(mode_info->atom_context->bios + data_offset +
-					 le16_to_cpu(ext_hdr->usPowerTuneTableOffset));
-				adev->pm.dpm.dyn_state.cac_tdp_table->maximum_power_delivery_limit = 255;
-				pt = &ppt->power_tune_table;
-			}
-			adev->pm.dpm.dyn_state.cac_tdp_table->tdp = le16_to_cpu(pt->usTDP);
-			adev->pm.dpm.dyn_state.cac_tdp_table->configurable_tdp =
-				le16_to_cpu(pt->usConfigurableTDP);
-			adev->pm.dpm.dyn_state.cac_tdp_table->tdc = le16_to_cpu(pt->usTDC);
-			adev->pm.dpm.dyn_state.cac_tdp_table->battery_power_limit =
-				le16_to_cpu(pt->usBatteryPowerLimit);
-			adev->pm.dpm.dyn_state.cac_tdp_table->small_power_limit =
-				le16_to_cpu(pt->usSmallPowerLimit);
-			adev->pm.dpm.dyn_state.cac_tdp_table->low_cac_leakage =
-				le16_to_cpu(pt->usLowCACLeakage);
-			adev->pm.dpm.dyn_state.cac_tdp_table->high_cac_leakage =
-				le16_to_cpu(pt->usHighCACLeakage);
-		}
-		if ((le16_to_cpu(ext_hdr->usSize) >= SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V8) &&
-				ext_hdr->usSclkVddgfxTableOffset) {
-			dep_table = (ATOM_PPLIB_Clock_Voltage_Dependency_Table *)
-				(mode_info->atom_context->bios + data_offset +
-				 le16_to_cpu(ext_hdr->usSclkVddgfxTableOffset));
-			ret = amdgpu_parse_clk_voltage_dep_table(
-					&adev->pm.dpm.dyn_state.vddgfx_dependency_on_sclk,
-					dep_table);
-			if (ret) {
-				kfree(adev->pm.dpm.dyn_state.vddgfx_dependency_on_sclk.entries);
-				return ret;
-			}
-		}
-	}
-
-	return 0;
-}
-
-void amdgpu_free_extended_power_table(struct amdgpu_device *adev)
-{
-	struct amdgpu_dpm_dynamic_state *dyn_state = &adev->pm.dpm.dyn_state;
-
-	kfree(dyn_state->vddc_dependency_on_sclk.entries);
-	kfree(dyn_state->vddci_dependency_on_mclk.entries);
-	kfree(dyn_state->vddc_dependency_on_mclk.entries);
-	kfree(dyn_state->mvdd_dependency_on_mclk.entries);
-	kfree(dyn_state->cac_leakage_table.entries);
-	kfree(dyn_state->phase_shedding_limits_table.entries);
-	kfree(dyn_state->ppm_table);
-	kfree(dyn_state->cac_tdp_table);
-	kfree(dyn_state->vce_clock_voltage_dependency_table.entries);
-	kfree(dyn_state->uvd_clock_voltage_dependency_table.entries);
-	kfree(dyn_state->samu_clock_voltage_dependency_table.entries);
-	kfree(dyn_state->acp_clock_voltage_dependency_table.entries);
-	kfree(dyn_state->vddgfx_dependency_on_sclk.entries);
-}
-
-static const char *pp_lib_thermal_controller_names[] = {
-	"NONE",
-	"lm63",
-	"adm1032",
-	"adm1030",
-	"max6649",
-	"lm64",
-	"f75375",
-	"RV6xx",
-	"RV770",
-	"adt7473",
-	"NONE",
-	"External GPIO",
-	"Evergreen",
-	"emc2103",
-	"Sumo",
-	"Northern Islands",
-	"Southern Islands",
-	"lm96163",
-	"Sea Islands",
-	"Kaveri/Kabini",
-};
-
-void amdgpu_add_thermal_controller(struct amdgpu_device *adev)
-{
-	struct amdgpu_mode_info *mode_info = &adev->mode_info;
-	ATOM_PPLIB_POWERPLAYTABLE *power_table;
-	int index = GetIndexIntoMasterTable(DATA, PowerPlayInfo);
-	ATOM_PPLIB_THERMALCONTROLLER *controller;
-	struct amdgpu_i2c_bus_rec i2c_bus;
-	u16 data_offset;
-	u8 frev, crev;
-
-	if (!amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
-				   &frev, &crev, &data_offset))
-		return;
-	power_table = (ATOM_PPLIB_POWERPLAYTABLE *)
-		(mode_info->atom_context->bios + data_offset);
-	controller = &power_table->sThermalController;
-
-	/* add the i2c bus for thermal/fan chip */
-	if (controller->ucType > 0) {
-		if (controller->ucFanParameters & ATOM_PP_FANPARAMETERS_NOFAN)
-			adev->pm.no_fan = true;
-		adev->pm.fan_pulses_per_revolution =
-			controller->ucFanParameters & ATOM_PP_FANPARAMETERS_TACHOMETER_PULSES_PER_REVOLUTION_MASK;
-		if (adev->pm.fan_pulses_per_revolution) {
-			adev->pm.fan_min_rpm = controller->ucFanMinRPM;
-			adev->pm.fan_max_rpm = controller->ucFanMaxRPM;
-		}
-		if (controller->ucType == ATOM_PP_THERMALCONTROLLER_RV6xx) {
-			DRM_INFO("Internal thermal controller %s fan control\n",
-				 (controller->ucFanParameters &
-				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
-			adev->pm.int_thermal_type = THERMAL_TYPE_RV6XX;
-		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_RV770) {
-			DRM_INFO("Internal thermal controller %s fan control\n",
-				 (controller->ucFanParameters &
-				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
-			adev->pm.int_thermal_type = THERMAL_TYPE_RV770;
-		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_EVERGREEN) {
-			DRM_INFO("Internal thermal controller %s fan control\n",
-				 (controller->ucFanParameters &
-				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
-			adev->pm.int_thermal_type = THERMAL_TYPE_EVERGREEN;
-		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_SUMO) {
-			DRM_INFO("Internal thermal controller %s fan control\n",
-				 (controller->ucFanParameters &
-				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
-			adev->pm.int_thermal_type = THERMAL_TYPE_SUMO;
-		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_NISLANDS) {
-			DRM_INFO("Internal thermal controller %s fan control\n",
-				 (controller->ucFanParameters &
-				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
-			adev->pm.int_thermal_type = THERMAL_TYPE_NI;
-		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_SISLANDS) {
-			DRM_INFO("Internal thermal controller %s fan control\n",
-				 (controller->ucFanParameters &
-				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
-			adev->pm.int_thermal_type = THERMAL_TYPE_SI;
-		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_CISLANDS) {
-			DRM_INFO("Internal thermal controller %s fan control\n",
-				 (controller->ucFanParameters &
-				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
-			adev->pm.int_thermal_type = THERMAL_TYPE_CI;
-		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_KAVERI) {
-			DRM_INFO("Internal thermal controller %s fan control\n",
-				 (controller->ucFanParameters &
-				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
-			adev->pm.int_thermal_type = THERMAL_TYPE_KV;
-		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_EXTERNAL_GPIO) {
-			DRM_INFO("External GPIO thermal controller %s fan control\n",
-				 (controller->ucFanParameters &
-				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
-			adev->pm.int_thermal_type = THERMAL_TYPE_EXTERNAL_GPIO;
-		} else if (controller->ucType ==
-			   ATOM_PP_THERMALCONTROLLER_ADT7473_WITH_INTERNAL) {
-			DRM_INFO("ADT7473 with internal thermal controller %s fan control\n",
-				 (controller->ucFanParameters &
-				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
-			adev->pm.int_thermal_type = THERMAL_TYPE_ADT7473_WITH_INTERNAL;
-		} else if (controller->ucType ==
-			   ATOM_PP_THERMALCONTROLLER_EMC2103_WITH_INTERNAL) {
-			DRM_INFO("EMC2103 with internal thermal controller %s fan control\n",
-				 (controller->ucFanParameters &
-				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
-			adev->pm.int_thermal_type = THERMAL_TYPE_EMC2103_WITH_INTERNAL;
-		} else if (controller->ucType < ARRAY_SIZE(pp_lib_thermal_controller_names)) {
-			DRM_INFO("Possible %s thermal controller at 0x%02x %s fan control\n",
-				 pp_lib_thermal_controller_names[controller->ucType],
-				 controller->ucI2cAddress >> 1,
-				 (controller->ucFanParameters &
-				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
-			adev->pm.int_thermal_type = THERMAL_TYPE_EXTERNAL;
-			i2c_bus = amdgpu_atombios_lookup_i2c_gpio(adev, controller->ucI2cLine);
-			adev->pm.i2c_bus = amdgpu_i2c_lookup(adev, &i2c_bus);
-			if (adev->pm.i2c_bus) {
-				struct i2c_board_info info = { };
-				const char *name = pp_lib_thermal_controller_names[controller->ucType];
-				info.addr = controller->ucI2cAddress >> 1;
-				strlcpy(info.type, name, sizeof(info.type));
-				i2c_new_device(&adev->pm.i2c_bus->adapter, &info);
-			}
-		} else {
-			DRM_INFO("Unknown thermal controller type %d at 0x%02x %s fan control\n",
-				 controller->ucType,
-				 controller->ucI2cAddress >> 1,
-				 (controller->ucFanParameters &
-				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
-		}
-	}
-}
-
-enum amdgpu_pcie_gen amdgpu_get_pcie_gen_support(struct amdgpu_device *adev,
-						 u32 sys_mask,
-						 enum amdgpu_pcie_gen asic_gen,
-						 enum amdgpu_pcie_gen default_gen)
-{
-	switch (asic_gen) {
-	case AMDGPU_PCIE_GEN1:
-		return AMDGPU_PCIE_GEN1;
-	case AMDGPU_PCIE_GEN2:
-		return AMDGPU_PCIE_GEN2;
-	case AMDGPU_PCIE_GEN3:
-		return AMDGPU_PCIE_GEN3;
-	default:
-		if ((sys_mask & DRM_PCIE_SPEED_80) && (default_gen == AMDGPU_PCIE_GEN3))
-			return AMDGPU_PCIE_GEN3;
-		else if ((sys_mask & DRM_PCIE_SPEED_50) && (default_gen == AMDGPU_PCIE_GEN2))
-			return AMDGPU_PCIE_GEN2;
-		else
-			return AMDGPU_PCIE_GEN1;
-	}
-	return AMDGPU_PCIE_GEN1;
-}
-
-u16 amdgpu_get_pcie_lane_support(struct amdgpu_device *adev,
-				 u16 asic_lanes,
-				 u16 default_lanes)
-{
-	switch (asic_lanes) {
-	case 0:
-	default:
-		return default_lanes;
-	case 1:
-		return 1;
-	case 2:
-		return 2;
-	case 4:
-		return 4;
-	case 8:
-		return 8;
-	case 12:
-		return 12;
-	case 16:
-		return 16;
-	}
-}
-
-u8 amdgpu_encode_pci_lane_width(u32 lanes)
-{
-	u8 encoded_lanes[] = { 0, 1, 2, 0, 3, 0, 0, 0, 4, 0, 0, 0, 5, 0, 0, 0, 6 };
-
-	if (lanes > 16)
-		return 0;
-
-	return encoded_lanes[lanes];
-}
-
-struct amd_vce_state*
-amdgpu_get_vce_clock_state(struct amdgpu_device *adev, unsigned idx)
-{
-	if (idx < adev->pm.dpm.num_of_vce_states)
-		return &adev->pm.dpm.vce_states[idx];
-
-	return NULL;
-}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h
deleted file mode 100644
index 8c96a4caa715..000000000000
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h
+++ /dev/null
@@ -1,556 +0,0 @@
-/*
- * Copyright 2014 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-#ifndef __AMDGPU_DPM_H__
-#define __AMDGPU_DPM_H__
-
-enum amdgpu_int_thermal_type {
-	THERMAL_TYPE_NONE,
-	THERMAL_TYPE_EXTERNAL,
-	THERMAL_TYPE_EXTERNAL_GPIO,
-	THERMAL_TYPE_RV6XX,
-	THERMAL_TYPE_RV770,
-	THERMAL_TYPE_ADT7473_WITH_INTERNAL,
-	THERMAL_TYPE_EVERGREEN,
-	THERMAL_TYPE_SUMO,
-	THERMAL_TYPE_NI,
-	THERMAL_TYPE_SI,
-	THERMAL_TYPE_EMC2103_WITH_INTERNAL,
-	THERMAL_TYPE_CI,
-	THERMAL_TYPE_KV,
-};
-
-enum amdgpu_dpm_auto_throttle_src {
-	AMDGPU_DPM_AUTO_THROTTLE_SRC_THERMAL,
-	AMDGPU_DPM_AUTO_THROTTLE_SRC_EXTERNAL
-};
-
-enum amdgpu_dpm_event_src {
-	AMDGPU_DPM_EVENT_SRC_ANALOG = 0,
-	AMDGPU_DPM_EVENT_SRC_EXTERNAL = 1,
-	AMDGPU_DPM_EVENT_SRC_DIGITAL = 2,
-	AMDGPU_DPM_EVENT_SRC_ANALOG_OR_EXTERNAL = 3,
-	AMDGPU_DPM_EVENT_SRC_DIGIAL_OR_EXTERNAL = 4
-};
-
-#define SCLK_DEEP_SLEEP_MASK 0x8
-
-struct amdgpu_ps {
-	u32 caps; /* vbios flags */
-	u32 class; /* vbios flags */
-	u32 class2; /* vbios flags */
-	/* UVD clocks */
-	u32 vclk;
-	u32 dclk;
-	/* VCE clocks */
-	u32 evclk;
-	u32 ecclk;
-	bool vce_active;
-	enum amd_vce_level vce_level;
-	/* asic priv */
-	void *ps_priv;
-};
-
-struct amdgpu_dpm_thermal {
-	/* thermal interrupt work */
-	struct work_struct work;
-	/* low temperature threshold */
-	int                min_temp;
-	/* high temperature threshold */
-	int                max_temp;
-	/* was last interrupt low to high or high to low */
-	bool               high_to_low;
-	/* interrupt source */
-	struct amdgpu_irq_src	irq;
-};
-
-enum amdgpu_clk_action
-{
-	AMDGPU_SCLK_UP = 1,
-	AMDGPU_SCLK_DOWN
-};
-
-struct amdgpu_blacklist_clocks
-{
-	u32 sclk;
-	u32 mclk;
-	enum amdgpu_clk_action action;
-};
-
-struct amdgpu_clock_and_voltage_limits {
-	u32 sclk;
-	u32 mclk;
-	u16 vddc;
-	u16 vddci;
-};
-
-struct amdgpu_clock_array {
-	u32 count;
-	u32 *values;
-};
-
-struct amdgpu_clock_voltage_dependency_entry {
-	u32 clk;
-	u16 v;
-};
-
-struct amdgpu_clock_voltage_dependency_table {
-	u32 count;
-	struct amdgpu_clock_voltage_dependency_entry *entries;
-};
-
-union amdgpu_cac_leakage_entry {
-	struct {
-		u16 vddc;
-		u32 leakage;
-	};
-	struct {
-		u16 vddc1;
-		u16 vddc2;
-		u16 vddc3;
-	};
-};
-
-struct amdgpu_cac_leakage_table {
-	u32 count;
-	union amdgpu_cac_leakage_entry *entries;
-};
-
-struct amdgpu_phase_shedding_limits_entry {
-	u16 voltage;
-	u32 sclk;
-	u32 mclk;
-};
-
-struct amdgpu_phase_shedding_limits_table {
-	u32 count;
-	struct amdgpu_phase_shedding_limits_entry *entries;
-};
-
-struct amdgpu_uvd_clock_voltage_dependency_entry {
-	u32 vclk;
-	u32 dclk;
-	u16 v;
-};
-
-struct amdgpu_uvd_clock_voltage_dependency_table {
-	u8 count;
-	struct amdgpu_uvd_clock_voltage_dependency_entry *entries;
-};
-
-struct amdgpu_vce_clock_voltage_dependency_entry {
-	u32 ecclk;
-	u32 evclk;
-	u16 v;
-};
-
-struct amdgpu_vce_clock_voltage_dependency_table {
-	u8 count;
-	struct amdgpu_vce_clock_voltage_dependency_entry *entries;
-};
-
-struct amdgpu_ppm_table {
-	u8 ppm_design;
-	u16 cpu_core_number;
-	u32 platform_tdp;
-	u32 small_ac_platform_tdp;
-	u32 platform_tdc;
-	u32 small_ac_platform_tdc;
-	u32 apu_tdp;
-	u32 dgpu_tdp;
-	u32 dgpu_ulv_power;
-	u32 tj_max;
-};
-
-struct amdgpu_cac_tdp_table {
-	u16 tdp;
-	u16 configurable_tdp;
-	u16 tdc;
-	u16 battery_power_limit;
-	u16 small_power_limit;
-	u16 low_cac_leakage;
-	u16 high_cac_leakage;
-	u16 maximum_power_delivery_limit;
-};
-
-struct amdgpu_dpm_dynamic_state {
-	struct amdgpu_clock_voltage_dependency_table vddc_dependency_on_sclk;
-	struct amdgpu_clock_voltage_dependency_table vddci_dependency_on_mclk;
-	struct amdgpu_clock_voltage_dependency_table vddc_dependency_on_mclk;
-	struct amdgpu_clock_voltage_dependency_table mvdd_dependency_on_mclk;
-	struct amdgpu_clock_voltage_dependency_table vddc_dependency_on_dispclk;
-	struct amdgpu_uvd_clock_voltage_dependency_table uvd_clock_voltage_dependency_table;
-	struct amdgpu_vce_clock_voltage_dependency_table vce_clock_voltage_dependency_table;
-	struct amdgpu_clock_voltage_dependency_table samu_clock_voltage_dependency_table;
-	struct amdgpu_clock_voltage_dependency_table acp_clock_voltage_dependency_table;
-	struct amdgpu_clock_voltage_dependency_table vddgfx_dependency_on_sclk;
-	struct amdgpu_clock_array valid_sclk_values;
-	struct amdgpu_clock_array valid_mclk_values;
-	struct amdgpu_clock_and_voltage_limits max_clock_voltage_on_dc;
-	struct amdgpu_clock_and_voltage_limits max_clock_voltage_on_ac;
-	u32 mclk_sclk_ratio;
-	u32 sclk_mclk_delta;
-	u16 vddc_vddci_delta;
-	u16 min_vddc_for_pcie_gen2;
-	struct amdgpu_cac_leakage_table cac_leakage_table;
-	struct amdgpu_phase_shedding_limits_table phase_shedding_limits_table;
-	struct amdgpu_ppm_table *ppm_table;
-	struct amdgpu_cac_tdp_table *cac_tdp_table;
-};
-
-struct amdgpu_dpm_fan {
-	u16 t_min;
-	u16 t_med;
-	u16 t_high;
-	u16 pwm_min;
-	u16 pwm_med;
-	u16 pwm_high;
-	u8 t_hyst;
-	u32 cycle_delay;
-	u16 t_max;
-	u8 control_mode;
-	u16 default_max_fan_pwm;
-	u16 default_fan_output_sensitivity;
-	u16 fan_output_sensitivity;
-	bool ucode_fan_control;
-};
-
-enum amdgpu_pcie_gen {
-	AMDGPU_PCIE_GEN1 = 0,
-	AMDGPU_PCIE_GEN2 = 1,
-	AMDGPU_PCIE_GEN3 = 2,
-	AMDGPU_PCIE_GEN_INVALID = 0xffff
-};
-
-struct amdgpu_dpm_funcs {
-	int (*get_temperature)(struct amdgpu_device *adev);
-	int (*pre_set_power_state)(struct amdgpu_device *adev);
-	int (*set_power_state)(struct amdgpu_device *adev);
-	void (*post_set_power_state)(struct amdgpu_device *adev);
-	void (*display_configuration_changed)(struct amdgpu_device *adev);
-	u32 (*get_sclk)(struct amdgpu_device *adev, bool low);
-	u32 (*get_mclk)(struct amdgpu_device *adev, bool low);
-	void (*print_power_state)(struct amdgpu_device *adev, struct amdgpu_ps *ps);
-	void (*debugfs_print_current_performance_level)(struct amdgpu_device *adev, struct seq_file *m);
-	int (*force_performance_level)(struct amdgpu_device *adev, enum amd_dpm_forced_level level);
-	bool (*vblank_too_short)(struct amdgpu_device *adev);
-	void (*powergate_uvd)(struct amdgpu_device *adev, bool gate);
-	void (*powergate_vce)(struct amdgpu_device *adev, bool gate);
-	void (*enable_bapm)(struct amdgpu_device *adev, bool enable);
-	void (*set_fan_control_mode)(struct amdgpu_device *adev, u32 mode);
-	u32 (*get_fan_control_mode)(struct amdgpu_device *adev);
-	int (*set_fan_speed_percent)(struct amdgpu_device *adev, u32 speed);
-	int (*get_fan_speed_percent)(struct amdgpu_device *adev, u32 *speed);
-	int (*force_clock_level)(struct amdgpu_device *adev, enum pp_clock_type type, uint32_t mask);
-	int (*print_clock_levels)(struct amdgpu_device *adev, enum pp_clock_type type, char *buf);
-	int (*get_sclk_od)(struct amdgpu_device *adev);
-	int (*set_sclk_od)(struct amdgpu_device *adev, uint32_t value);
-	int (*get_mclk_od)(struct amdgpu_device *adev);
-	int (*set_mclk_od)(struct amdgpu_device *adev, uint32_t value);
-	int (*check_state_equal)(struct amdgpu_device *adev,
-				struct amdgpu_ps *cps,
-				struct amdgpu_ps *rps,
-				bool *equal);
-	int (*read_sensor)(struct amdgpu_device *adev, int idx, void *value,
-			   int *size);
-
-	struct amd_vce_state* (*get_vce_clock_state)(struct amdgpu_device *adev, unsigned idx);
-	int (*reset_power_profile_state)(struct amdgpu_device *adev,
-			struct amd_pp_profile *request);
-	int (*get_power_profile_state)(struct amdgpu_device *adev,
-			struct amd_pp_profile *query);
-	int (*set_power_profile_state)(struct amdgpu_device *adev,
-			struct amd_pp_profile *request);
-	int (*switch_power_profile)(struct amdgpu_device *adev,
-			enum amd_pp_profile_type type);
-};
-
-#define amdgpu_dpm_pre_set_power_state(adev) (adev)->pm.funcs->pre_set_power_state((adev))
-#define amdgpu_dpm_set_power_state(adev) (adev)->pm.funcs->set_power_state((adev))
-#define amdgpu_dpm_post_set_power_state(adev) (adev)->pm.funcs->post_set_power_state((adev))
-#define amdgpu_dpm_display_configuration_changed(adev) (adev)->pm.funcs->display_configuration_changed((adev))
-#define amdgpu_dpm_print_power_state(adev, ps) (adev)->pm.funcs->print_power_state((adev), (ps))
-#define amdgpu_dpm_vblank_too_short(adev) (adev)->pm.funcs->vblank_too_short((adev))
-#define amdgpu_dpm_enable_bapm(adev, e) (adev)->pm.funcs->enable_bapm((adev), (e))
-
-#define amdgpu_dpm_read_sensor(adev, idx, value, size) \
-	((adev)->pp_enabled ? \
-		(adev)->powerplay.pp_funcs->read_sensor(adev->powerplay.pp_handle, (idx), (value), (size)) : \
-		(adev)->pm.funcs->read_sensor((adev), (idx), (value), (size)))
-
-#define amdgpu_dpm_get_temperature(adev) \
-	((adev)->pp_enabled ?						\
-	      (adev)->powerplay.pp_funcs->get_temperature((adev)->powerplay.pp_handle) : \
-	      (adev)->pm.funcs->get_temperature((adev)))
-
-#define amdgpu_dpm_set_fan_control_mode(adev, m) \
-	((adev)->pp_enabled ?						\
-	      (adev)->powerplay.pp_funcs->set_fan_control_mode((adev)->powerplay.pp_handle, (m)) : \
-	      (adev)->pm.funcs->set_fan_control_mode((adev), (m)))
-
-#define amdgpu_dpm_get_fan_control_mode(adev) \
-	((adev)->pp_enabled ?						\
-	      (adev)->powerplay.pp_funcs->get_fan_control_mode((adev)->powerplay.pp_handle) : \
-	      (adev)->pm.funcs->get_fan_control_mode((adev)))
-
-#define amdgpu_dpm_set_fan_speed_percent(adev, s) \
-	((adev)->pp_enabled ?						\
-	      (adev)->powerplay.pp_funcs->set_fan_speed_percent((adev)->powerplay.pp_handle, (s)) : \
-	      (adev)->pm.funcs->set_fan_speed_percent((adev), (s)))
-
-#define amdgpu_dpm_get_fan_speed_percent(adev, s) \
-	((adev)->pp_enabled ?						\
-	      (adev)->powerplay.pp_funcs->get_fan_speed_percent((adev)->powerplay.pp_handle, (s)) : \
-	      (adev)->pm.funcs->get_fan_speed_percent((adev), (s)))
-
-#define amdgpu_dpm_get_fan_speed_rpm(adev, s) \
-	((adev)->pp_enabled ?						\
-	      (adev)->powerplay.pp_funcs->get_fan_speed_rpm((adev)->powerplay.pp_handle, (s)) : \
-	      -EINVAL)
-
-#define amdgpu_dpm_get_sclk(adev, l) \
-	((adev)->pp_enabled ?						\
-	      (adev)->powerplay.pp_funcs->get_sclk((adev)->powerplay.pp_handle, (l)) : \
-		(adev)->pm.funcs->get_sclk((adev), (l)))
-
-#define amdgpu_dpm_get_mclk(adev, l)  \
-	((adev)->pp_enabled ?						\
-	      (adev)->powerplay.pp_funcs->get_mclk((adev)->powerplay.pp_handle, (l)) : \
-	      (adev)->pm.funcs->get_mclk((adev), (l)))
-
-
-#define amdgpu_dpm_force_performance_level(adev, l) \
-	((adev)->pp_enabled ?						\
-	      (adev)->powerplay.pp_funcs->force_performance_level((adev)->powerplay.pp_handle, (l)) : \
-	      (adev)->pm.funcs->force_performance_level((adev), (l)))
-
-#define amdgpu_dpm_powergate_uvd(adev, g) \
-	((adev)->pp_enabled ?						\
-	      (adev)->powerplay.pp_funcs->powergate_uvd((adev)->powerplay.pp_handle, (g)) : \
-	      (adev)->pm.funcs->powergate_uvd((adev), (g)))
-
-#define amdgpu_dpm_powergate_vce(adev, g) \
-	((adev)->pp_enabled ?						\
-	      (adev)->powerplay.pp_funcs->powergate_vce((adev)->powerplay.pp_handle, (g)) : \
-	      (adev)->pm.funcs->powergate_vce((adev), (g)))
-
-#define amdgpu_dpm_get_current_power_state(adev) \
-	(adev)->powerplay.pp_funcs->get_current_power_state((adev)->powerplay.pp_handle)
-
-#define amdgpu_dpm_get_pp_num_states(adev, data) \
-	(adev)->powerplay.pp_funcs->get_pp_num_states((adev)->powerplay.pp_handle, data)
-
-#define amdgpu_dpm_get_pp_table(adev, table) \
-	(adev)->powerplay.pp_funcs->get_pp_table((adev)->powerplay.pp_handle, table)
-
-#define amdgpu_dpm_set_pp_table(adev, buf, size) \
-	(adev)->powerplay.pp_funcs->set_pp_table((adev)->powerplay.pp_handle, buf, size)
-
-#define amdgpu_dpm_print_clock_levels(adev, type, buf) \
-	(adev)->powerplay.pp_funcs->print_clock_levels((adev)->powerplay.pp_handle, type, buf)
-
-#define amdgpu_dpm_force_clock_level(adev, type, level) \
-		(adev)->powerplay.pp_funcs->force_clock_level((adev)->powerplay.pp_handle, type, level)
-
-#define amdgpu_dpm_get_sclk_od(adev) \
-	(adev)->powerplay.pp_funcs->get_sclk_od((adev)->powerplay.pp_handle)
-
-#define amdgpu_dpm_set_sclk_od(adev, value) \
-	(adev)->powerplay.pp_funcs->set_sclk_od((adev)->powerplay.pp_handle, value)
-
-#define amdgpu_dpm_get_mclk_od(adev) \
-	((adev)->powerplay.pp_funcs->get_mclk_od((adev)->powerplay.pp_handle))
-
-#define amdgpu_dpm_set_mclk_od(adev, value) \
-	((adev)->powerplay.pp_funcs->set_mclk_od((adev)->powerplay.pp_handle, value))
-
-#define amdgpu_dpm_dispatch_task(adev, event_id, input, output)		\
-	(adev)->powerplay.pp_funcs->dispatch_tasks((adev)->powerplay.pp_handle, (event_id), (input), (output))
-
-#define amgdpu_dpm_check_state_equal(adev, cps, rps, equal) (adev)->pm.funcs->check_state_equal((adev), (cps),(rps),(equal))
-
-#define amdgpu_dpm_get_vce_clock_state(adev, i)				\
-	((adev)->pp_enabled ?						\
-	 (adev)->powerplay.pp_funcs->get_vce_clock_state((adev)->powerplay.pp_handle, (i)) : \
-	 (adev)->pm.funcs->get_vce_clock_state((adev), (i)))
-
-#define amdgpu_dpm_get_performance_level(adev) \
-	((adev)->pp_enabled ?						\
-	(adev)->powerplay.pp_funcs->get_performance_level((adev)->powerplay.pp_handle) : \
-	(adev)->pm.dpm.forced_level)
-
-#define amdgpu_dpm_reset_power_profile_state(adev, request) \
-	((adev)->powerplay.pp_funcs->reset_power_profile_state(\
-			(adev)->powerplay.pp_handle, request))
-
-#define amdgpu_dpm_get_power_profile_state(adev, query) \
-	((adev)->powerplay.pp_funcs->get_power_profile_state(\
-			(adev)->powerplay.pp_handle, query))
-
-#define amdgpu_dpm_set_power_profile_state(adev, request) \
-	((adev)->powerplay.pp_funcs->set_power_profile_state(\
-			(adev)->powerplay.pp_handle, request))
-
-#define amdgpu_dpm_switch_power_profile(adev, type) \
-	((adev)->powerplay.pp_funcs->switch_power_profile(\
-			(adev)->powerplay.pp_handle, type))
-
-struct amdgpu_dpm {
-	struct amdgpu_ps        *ps;
-	/* number of valid power states */
-	int                     num_ps;
-	/* current power state that is active */
-	struct amdgpu_ps        *current_ps;
-	/* requested power state */
-	struct amdgpu_ps        *requested_ps;
-	/* boot up power state */
-	struct amdgpu_ps        *boot_ps;
-	/* default uvd power state */
-	struct amdgpu_ps        *uvd_ps;
-	/* vce requirements */
-	u32                  num_of_vce_states;
-	struct amd_vce_state vce_states[AMD_MAX_VCE_LEVELS];
-	enum amd_vce_level vce_level;
-	enum amd_pm_state_type state;
-	enum amd_pm_state_type user_state;
-	enum amd_pm_state_type last_state;
-	enum amd_pm_state_type last_user_state;
-	u32                     platform_caps;
-	u32                     voltage_response_time;
-	u32                     backbias_response_time;
-	void                    *priv;
-	u32			new_active_crtcs;
-	int			new_active_crtc_count;
-	u32			current_active_crtcs;
-	int			current_active_crtc_count;
-	struct amdgpu_dpm_dynamic_state dyn_state;
-	struct amdgpu_dpm_fan fan;
-	u32 tdp_limit;
-	u32 near_tdp_limit;
-	u32 near_tdp_limit_adjusted;
-	u32 sq_ramping_threshold;
-	u32 cac_leakage;
-	u16 tdp_od_limit;
-	u32 tdp_adjustment;
-	u16 load_line_slope;
-	bool power_control;
-	bool ac_power;
-	/* special states active */
-	bool                    thermal_active;
-	bool                    uvd_active;
-	bool                    vce_active;
-	/* thermal handling */
-	struct amdgpu_dpm_thermal thermal;
-	/* forced levels */
-	enum amd_dpm_forced_level forced_level;
-};
-
-struct amdgpu_pm {
-	struct mutex		mutex;
-	u32                     current_sclk;
-	u32                     current_mclk;
-	u32                     default_sclk;
-	u32                     default_mclk;
-	struct amdgpu_i2c_chan *i2c_bus;
-	/* internal thermal controller on rv6xx+ */
-	enum amdgpu_int_thermal_type int_thermal_type;
-	struct device	        *int_hwmon_dev;
-	/* fan control parameters */
-	bool                    no_fan;
-	u8                      fan_pulses_per_revolution;
-	u8                      fan_min_rpm;
-	u8                      fan_max_rpm;
-	/* dpm */
-	bool                    dpm_enabled;
-	bool                    sysfs_initialized;
-	struct amdgpu_dpm       dpm;
-	const struct firmware	*fw;	/* SMC firmware */
-	uint32_t                fw_version;
-	const struct amdgpu_dpm_funcs *funcs;
-	uint32_t                pcie_gen_mask;
-	uint32_t                pcie_mlw_mask;
-	struct amd_pp_display_configuration pm_display_cfg;/* set by DAL */
-};
-
-#define R600_SSTU_DFLT                               0
-#define R600_SST_DFLT                                0x00C8
-
-/* XXX are these ok? */
-#define R600_TEMP_RANGE_MIN (90 * 1000)
-#define R600_TEMP_RANGE_MAX (120 * 1000)
-
-#define FDO_PWM_MODE_STATIC  1
-#define FDO_PWM_MODE_STATIC_RPM 5
-
-enum amdgpu_td {
-	AMDGPU_TD_AUTO,
-	AMDGPU_TD_UP,
-	AMDGPU_TD_DOWN,
-};
-
-enum amdgpu_display_watermark {
-	AMDGPU_DISPLAY_WATERMARK_LOW = 0,
-	AMDGPU_DISPLAY_WATERMARK_HIGH = 1,
-};
-
-enum amdgpu_display_gap
-{
-    AMDGPU_PM_DISPLAY_GAP_VBLANK_OR_WM = 0,
-    AMDGPU_PM_DISPLAY_GAP_VBLANK       = 1,
-    AMDGPU_PM_DISPLAY_GAP_WATERMARK    = 2,
-    AMDGPU_PM_DISPLAY_GAP_IGNORE       = 3,
-};
-
-void amdgpu_dpm_print_class_info(u32 class, u32 class2);
-void amdgpu_dpm_print_cap_info(u32 caps);
-void amdgpu_dpm_print_ps_status(struct amdgpu_device *adev,
-				struct amdgpu_ps *rps);
-u32 amdgpu_dpm_get_vblank_time(struct amdgpu_device *adev);
-u32 amdgpu_dpm_get_vrefresh(struct amdgpu_device *adev);
-bool amdgpu_is_uvd_state(u32 class, u32 class2);
-void amdgpu_calculate_u_and_p(u32 i, u32 r_c, u32 p_b,
-			      u32 *p, u32 *u);
-int amdgpu_calculate_at(u32 t, u32 h, u32 fh, u32 fl, u32 *tl, u32 *th);
-
-bool amdgpu_is_internal_thermal_sensor(enum amdgpu_int_thermal_type sensor);
-
-int amdgpu_get_platform_caps(struct amdgpu_device *adev);
-
-int amdgpu_parse_extended_power_table(struct amdgpu_device *adev);
-void amdgpu_free_extended_power_table(struct amdgpu_device *adev);
-
-void amdgpu_add_thermal_controller(struct amdgpu_device *adev);
-
-enum amdgpu_pcie_gen amdgpu_get_pcie_gen_support(struct amdgpu_device *adev,
-						 u32 sys_mask,
-						 enum amdgpu_pcie_gen asic_gen,
-						 enum amdgpu_pcie_gen default_gen);
-
-u16 amdgpu_get_pcie_lane_support(struct amdgpu_device *adev,
-				 u16 asic_lanes,
-				 u16 default_lanes);
-u8 amdgpu_encode_pci_lane_width(u32 lanes);
-
-struct amd_vce_state*
-amdgpu_get_vce_clock_state(struct amdgpu_device *adev, unsigned idx);
-
-#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
index 0f16986ec5bc..bff25ef3e2d0 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
@@ -1,10 +1,3 @@
-/**
- * \file amdgpu_drv.c
- * AMD Amdgpu driver
- *
- * \author Gareth Hughes <gareth@valinux.com>
- */
-
 /*
  * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California.
  * All Rights Reserved.
@@ -29,22 +22,38 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  */
 
-#include <drm/drmP.h>
 #include <drm/amdgpu_drm.h>
+#include <drm/clients/drm_client_setup.h>
+#include <drm/drm_drv.h>
+#include <drm/drm_fbdev_ttm.h>
 #include <drm/drm_gem.h>
-#include "amdgpu_drv.h"
-
+#include <drm/drm_managed.h>
 #include <drm/drm_pciids.h>
-#include <linux/console.h>
+#include <drm/drm_probe_helper.h>
+#include <drm/drm_vblank.h>
+
+#include <linux/cc_platform.h>
+#include <linux/dynamic_debug.h>
 #include <linux/module.h>
+#include <linux/mmu_notifier.h>
 #include <linux/pm_runtime.h>
+#include <linux/suspend.h>
 #include <linux/vga_switcheroo.h>
-#include <drm/drm_crtc_helper.h>
 
 #include "amdgpu.h"
-#include "amdgpu_irq.h"
-
 #include "amdgpu_amdkfd.h"
+#include "amdgpu_dma_buf.h"
+#include "amdgpu_drv.h"
+#include "amdgpu_fdinfo.h"
+#include "amdgpu_irq.h"
+#include "amdgpu_psp.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_reset.h"
+#include "amdgpu_sched.h"
+#include "amdgpu_xgmi.h"
+#include "amdgpu_userq.h"
+#include "amdgpu_userq_fence.h"
+#include "../amdxcp/amdgpu_xcp_drv.h"
 
 /*
  * KMS wrapper.
@@ -69,205 +78,559 @@
  * - 3.17.0 - Add AMDGPU_NUM_VRAM_CPU_PAGE_FAULTS.
  * - 3.18.0 - Export gpu always on cu bitmap
  * - 3.19.0 - Add support for UVD MJPEG decode
+ * - 3.20.0 - Add support for local BOs
+ * - 3.21.0 - Add DRM_AMDGPU_FENCE_TO_HANDLE ioctl
+ * - 3.22.0 - Add DRM_AMDGPU_SCHED ioctl
+ * - 3.23.0 - Add query for VRAM lost counter
+ * - 3.24.0 - Add high priority compute support for gfx9
+ * - 3.25.0 - Add support for sensor query info (stable pstate sclk/mclk).
+ * - 3.26.0 - GFX9: Process AMDGPU_IB_FLAG_TC_WB_NOT_INVALIDATE.
+ * - 3.27.0 - Add new chunk to AMDGPU_CS to enable BO_LIST creation.
+ * - 3.28.0 - Add AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES
+ * - 3.29.0 - Add AMDGPU_IB_FLAG_RESET_GDS_MAX_WAVE_ID
+ * - 3.30.0 - Add AMDGPU_SCHED_OP_CONTEXT_PRIORITY_OVERRIDE.
+ * - 3.31.0 - Add support for per-flip tiling attribute changes with DC
+ * - 3.32.0 - Add syncobj timeline support to AMDGPU_CS.
+ * - 3.33.0 - Fixes for GDS ENOMEM failures in AMDGPU_CS.
+ * - 3.34.0 - Non-DC can flip correctly between buffers with different pitches
+ * - 3.35.0 - Add drm_amdgpu_info_device::tcc_disabled_mask
+ * - 3.36.0 - Allow reading more status registers on si/cik
+ * - 3.37.0 - L2 is invalidated before SDMA IBs, needed for correctness
+ * - 3.38.0 - Add AMDGPU_IB_FLAG_EMIT_MEM_SYNC
+ * - 3.39.0 - DMABUF implicit sync does a full pipeline sync
+ * - 3.40.0 - Add AMDGPU_IDS_FLAGS_TMZ
+ * - 3.41.0 - Add video codec query
+ * - 3.42.0 - Add 16bpc fixed point display support
+ * - 3.43.0 - Add device hot plug/unplug support
+ * - 3.44.0 - DCN3 supports DCC independent block settings: !64B && 128B, 64B && 128B
+ * - 3.45.0 - Add context ioctl stable pstate interface
+ * - 3.46.0 - To enable hot plug amdgpu tests in libdrm
+ * - 3.47.0 - Add AMDGPU_GEM_CREATE_DISCARDABLE and AMDGPU_VM_NOALLOC flags
+ * - 3.48.0 - Add IP discovery version info to HW INFO
+ * - 3.49.0 - Add gang submit into CS IOCTL
+ * - 3.50.0 - Update AMDGPU_INFO_DEV_INFO IOCTL for minimum engine and memory clock
+ *            Update AMDGPU_INFO_SENSOR IOCTL for PEAK_PSTATE engine and memory clock
+ *   3.51.0 - Return the PCIe gen and lanes from the INFO ioctl
+ *   3.52.0 - Add AMDGPU_IDS_FLAGS_CONFORMANT_TRUNC_COORD, add device_info fields:
+ *            tcp_cache_size, num_sqc_per_wgp, sqc_data_cache_size, sqc_inst_cache_size,
+ *            gl1c_cache_size, gl2c_cache_size, mall_size, enabled_rb_pipes_mask_hi
+ *   3.53.0 - Support for GFX11 CP GFX shadowing
+ *   3.54.0 - Add AMDGPU_CTX_QUERY2_FLAGS_RESET_IN_PROGRESS support
+ * - 3.55.0 - Add AMDGPU_INFO_GPUVM_FAULT query
+ * - 3.56.0 - Update IB start address and size alignment for decode and encode
+ * - 3.57.0 - Compute tunneling on GFX10+
+ * - 3.58.0 - Add GFX12 DCC support
+ * - 3.59.0 - Cleared VRAM
+ * - 3.60.0 - Add AMDGPU_TILING_GFX12_DCC_WRITE_COMPRESS_DISABLE (Vulkan requirement)
+ * - 3.61.0 - Contains fix for RV/PCO compute queues
+ * - 3.62.0 - Add AMDGPU_IDS_FLAGS_MODE_PF, AMDGPU_IDS_FLAGS_MODE_VF & AMDGPU_IDS_FLAGS_MODE_PT
+ * - 3.63.0 - GFX12 display DCC supports 256B max compressed block size
+ * - 3.64.0 - Userq IP support query
  */
 #define KMS_DRIVER_MAJOR	3
-#define KMS_DRIVER_MINOR	19
+#define KMS_DRIVER_MINOR	64
 #define KMS_DRIVER_PATCHLEVEL	0
 
-int amdgpu_vram_limit = 0;
-int amdgpu_vis_vram_limit = 0;
+/*
+ * amdgpu.debug module options. Are all disabled by default
+ */
+enum AMDGPU_DEBUG_MASK {
+	AMDGPU_DEBUG_VM = BIT(0),
+	AMDGPU_DEBUG_LARGEBAR = BIT(1),
+	AMDGPU_DEBUG_DISABLE_GPU_SOFT_RECOVERY = BIT(2),
+	AMDGPU_DEBUG_USE_VRAM_FW_BUF = BIT(3),
+	AMDGPU_DEBUG_ENABLE_RAS_ACA = BIT(4),
+	AMDGPU_DEBUG_ENABLE_EXP_RESETS = BIT(5),
+	AMDGPU_DEBUG_DISABLE_GPU_RING_RESET = BIT(6),
+	AMDGPU_DEBUG_SMU_POOL = BIT(7),
+	AMDGPU_DEBUG_VM_USERPTR = BIT(8),
+	AMDGPU_DEBUG_DISABLE_RAS_CE_LOG = BIT(9)
+};
+
+unsigned int amdgpu_vram_limit = UINT_MAX;
+int amdgpu_vis_vram_limit;
 int amdgpu_gart_size = -1; /* auto */
 int amdgpu_gtt_size = -1; /* auto */
 int amdgpu_moverate = -1; /* auto */
-int amdgpu_benchmarking = 0;
-int amdgpu_testing = 0;
 int amdgpu_audio = -1;
-int amdgpu_disp_priority = 0;
-int amdgpu_hw_i2c = 0;
+int amdgpu_disp_priority;
+int amdgpu_hw_i2c;
 int amdgpu_pcie_gen2 = -1;
 int amdgpu_msi = -1;
-int amdgpu_lockup_timeout = 0;
+char amdgpu_lockup_timeout[AMDGPU_MAX_TIMEOUT_PARAM_LENGTH];
 int amdgpu_dpm = -1;
 int amdgpu_fw_load_type = -1;
 int amdgpu_aspm = -1;
 int amdgpu_runtime_pm = -1;
-unsigned amdgpu_ip_block_mask = 0xffffffff;
+uint amdgpu_ip_block_mask = 0xffffffff;
 int amdgpu_bapm = -1;
-int amdgpu_deep_color = 0;
+int amdgpu_deep_color;
 int amdgpu_vm_size = -1;
 int amdgpu_vm_fragment_size = -1;
 int amdgpu_vm_block_size = -1;
-int amdgpu_vm_fault_stop = 0;
-int amdgpu_vm_debug = 0;
-int amdgpu_vram_page_split = 512;
+int amdgpu_vm_fault_stop;
 int amdgpu_vm_update_mode = -1;
-int amdgpu_exp_hw_support = 0;
+int amdgpu_exp_hw_support;
+int amdgpu_dc = -1;
 int amdgpu_sched_jobs = 32;
 int amdgpu_sched_hw_submission = 2;
-int amdgpu_no_evict = 0;
-int amdgpu_direct_gma_size = 0;
-unsigned amdgpu_pcie_gen_cap = 0;
-unsigned amdgpu_pcie_lane_cap = 0;
-unsigned amdgpu_cg_mask = 0xffffffff;
-unsigned amdgpu_pg_mask = 0xffffffff;
-unsigned amdgpu_sdma_phase_quantum = 32;
-char *amdgpu_disable_cu = NULL;
-char *amdgpu_virtual_display = NULL;
-unsigned amdgpu_pp_feature_mask = 0xffffffff;
-int amdgpu_ngg = 0;
-int amdgpu_prim_buf_per_se = 0;
-int amdgpu_pos_buf_per_se = 0;
-int amdgpu_cntl_sb_buf_per_se = 0;
-int amdgpu_param_buf_per_se = 0;
-int amdgpu_job_hang_limit = 0;
+uint amdgpu_pcie_gen_cap;
+uint amdgpu_pcie_lane_cap;
+u64 amdgpu_cg_mask = 0xffffffffffffffff;
+uint amdgpu_pg_mask = 0xffffffff;
+uint amdgpu_sdma_phase_quantum = 32;
+char *amdgpu_disable_cu;
+char *amdgpu_virtual_display;
+int amdgpu_enforce_isolation = -1;
+int amdgpu_modeset = -1;
+
+/* Specifies the default granularity for SVM, used in buffer
+ * migration and restoration of backing memory when handling
+ * recoverable page faults.
+ *
+ * The value is given as log(numPages(buffer)); for a 2 MiB
+ * buffer it computes to be 9
+ */
+uint amdgpu_svm_default_granularity = 9;
+
+/*
+ * OverDrive(bit 14) disabled by default
+ * GFX DCS(bit 19) disabled by default
+ */
+uint amdgpu_pp_feature_mask = 0xfff7bfff;
+uint amdgpu_force_long_training;
 int amdgpu_lbpw = -1;
+int amdgpu_compute_multipipe = -1;
+int amdgpu_gpu_recovery = -1; /* auto */
+int amdgpu_emu_mode;
+uint amdgpu_smu_memory_pool_size;
+int amdgpu_smu_pptable_id = -1;
+/*
+ * FBC (bit 0) disabled by default
+ * MULTI_MON_PP_MCLK_SWITCH (bit 1) enabled by default
+ *   - With this, for multiple monitors in sync(e.g. with the same model),
+ *     mclk switching will be allowed. And the mclk will be not foced to the
+ *     highest. That helps saving some idle power.
+ * DISABLE_FRACTIONAL_PWM (bit 2) disabled by default
+ * PSR (bit 3) disabled by default
+ * EDP NO POWER SEQUENCING (bit 4) disabled by default
+ */
+uint amdgpu_dc_feature_mask = 2;
+uint amdgpu_dc_debug_mask;
+uint amdgpu_dc_visual_confirm;
+int amdgpu_async_gfx_ring = 1;
+int amdgpu_mcbp = -1;
+int amdgpu_discovery = -1;
+int amdgpu_mes;
+int amdgpu_mes_log_enable = 0;
+int amdgpu_mes_kiq;
+int amdgpu_uni_mes = 1;
+int amdgpu_noretry = -1;
+int amdgpu_force_asic_type = -1;
+int amdgpu_tmz = -1; /* auto */
+uint amdgpu_freesync_vid_mode;
+int amdgpu_reset_method = -1; /* auto */
+int amdgpu_num_kcq = -1;
+int amdgpu_smartshift_bias;
+int amdgpu_use_xgmi_p2p = 1;
+int amdgpu_vcnfw_log;
+int amdgpu_sg_display = -1; /* auto */
+int amdgpu_user_partt_mode = AMDGPU_AUTO_COMPUTE_PARTITION_MODE;
+int amdgpu_umsch_mm;
+int amdgpu_seamless = -1; /* auto */
+uint amdgpu_debug_mask;
+int amdgpu_agp = -1; /* auto */
+int amdgpu_wbrf = -1;
+int amdgpu_damage_clips = -1; /* auto */
+int amdgpu_umsch_mm_fwlog;
+int amdgpu_rebar = -1; /* auto */
+int amdgpu_user_queue = -1;
+
+DECLARE_DYNDBG_CLASSMAP(drm_debug_classes, DD_CLASS_TYPE_DISJOINT_BITS, 0,
+			"DRM_UT_CORE",
+			"DRM_UT_DRIVER",
+			"DRM_UT_KMS",
+			"DRM_UT_PRIME",
+			"DRM_UT_ATOMIC",
+			"DRM_UT_VBL",
+			"DRM_UT_STATE",
+			"DRM_UT_LEASE",
+			"DRM_UT_DP",
+			"DRM_UT_DRMRES");
+
+struct amdgpu_mgpu_info mgpu_info = {
+	.mutex = __MUTEX_INITIALIZER(mgpu_info.mutex),
+};
+int amdgpu_ras_enable = -1;
+uint amdgpu_ras_mask = 0xffffffff;
+int amdgpu_bad_page_threshold = -1;
+struct amdgpu_watchdog_timer amdgpu_watchdog_timer = {
+	.timeout_fatal_disable = false,
+	.period = 0x0, /* default to 0x0 (timeout disable) */
+};
 
+/**
+ * DOC: vramlimit (int)
+ * Restrict the total amount of VRAM in MiB for testing.  The default is 0 (Use full VRAM).
+ */
 MODULE_PARM_DESC(vramlimit, "Restrict VRAM for testing, in megabytes");
 module_param_named(vramlimit, amdgpu_vram_limit, int, 0600);
 
+/**
+ * DOC: vis_vramlimit (int)
+ * Restrict the amount of CPU visible VRAM in MiB for testing.  The default is 0 (Use full CPU visible VRAM).
+ */
 MODULE_PARM_DESC(vis_vramlimit, "Restrict visible VRAM for testing, in megabytes");
 module_param_named(vis_vramlimit, amdgpu_vis_vram_limit, int, 0444);
 
-MODULE_PARM_DESC(gartsize, "Size of GART to setup in megabytes (32, 64, etc., -1=auto)");
+/**
+ * DOC: gartsize (uint)
+ * Restrict the size of GART (for kernel use) in Mib (32, 64, etc.) for testing.
+ * The default is -1 (The size depends on asic).
+ */
+MODULE_PARM_DESC(gartsize, "Size of kernel GART to setup in megabytes (32, 64, etc., -1=auto)");
 module_param_named(gartsize, amdgpu_gart_size, uint, 0600);
 
-MODULE_PARM_DESC(gttsize, "Size of the GTT domain in megabytes (-1 = auto)");
+/**
+ * DOC: gttsize (int)
+ * Restrict the size of GTT domain (for userspace use) in MiB for testing.
+ * The default is -1 (Use value specified by TTM).
+ * This parameter is deprecated and will be removed in the future.
+ */
+MODULE_PARM_DESC(gttsize, "Size of the GTT userspace domain in megabytes (-1 = auto)");
 module_param_named(gttsize, amdgpu_gtt_size, int, 0600);
 
+/**
+ * DOC: moverate (int)
+ * Set maximum buffer migration rate in MB/s. The default is -1 (8 MB/s).
+ */
 MODULE_PARM_DESC(moverate, "Maximum buffer migration rate in MB/s. (32, 64, etc., -1=auto, 0=1=disabled)");
 module_param_named(moverate, amdgpu_moverate, int, 0600);
 
-MODULE_PARM_DESC(benchmark, "Run benchmark");
-module_param_named(benchmark, amdgpu_benchmarking, int, 0444);
-
-MODULE_PARM_DESC(test, "Run tests");
-module_param_named(test, amdgpu_testing, int, 0444);
-
+/**
+ * DOC: audio (int)
+ * Set HDMI/DPAudio. Only affects non-DC display handling. The default is -1 (Enabled), set 0 to disabled it.
+ */
 MODULE_PARM_DESC(audio, "Audio enable (-1 = auto, 0 = disable, 1 = enable)");
 module_param_named(audio, amdgpu_audio, int, 0444);
 
+/**
+ * DOC: disp_priority (int)
+ * Set display Priority (1 = normal, 2 = high). Only affects non-DC display handling. The default is 0 (auto).
+ */
 MODULE_PARM_DESC(disp_priority, "Display Priority (0 = auto, 1 = normal, 2 = high)");
 module_param_named(disp_priority, amdgpu_disp_priority, int, 0444);
 
+/**
+ * DOC: hw_i2c (int)
+ * To enable hw i2c engine. Only affects non-DC display handling. The default is 0 (Disabled).
+ */
 MODULE_PARM_DESC(hw_i2c, "hw i2c engine enable (0 = disable)");
 module_param_named(hw_i2c, amdgpu_hw_i2c, int, 0444);
 
+/**
+ * DOC: pcie_gen2 (int)
+ * To disable PCIE Gen2/3 mode (0 = disable, 1 = enable). The default is -1 (auto, enabled).
+ */
 MODULE_PARM_DESC(pcie_gen2, "PCIE Gen2 mode (-1 = auto, 0 = disable, 1 = enable)");
 module_param_named(pcie_gen2, amdgpu_pcie_gen2, int, 0444);
 
+/**
+ * DOC: msi (int)
+ * To disable Message Signaled Interrupts (MSI) functionality (1 = enable, 0 = disable). The default is -1 (auto, enabled).
+ */
 MODULE_PARM_DESC(msi, "MSI support (1 = enable, 0 = disable, -1 = auto)");
 module_param_named(msi, amdgpu_msi, int, 0444);
 
-MODULE_PARM_DESC(lockup_timeout, "GPU lockup timeout in ms (default 0 = disable)");
-module_param_named(lockup_timeout, amdgpu_lockup_timeout, int, 0444);
+/**
+ * DOC: svm_default_granularity (uint)
+ * Used in buffer migration and handling of recoverable page faults
+ */
+MODULE_PARM_DESC(svm_default_granularity, "SVM's default granularity in log(2^Pages), default 9 = 2^9 = 2 MiB");
+module_param_named(svm_default_granularity, amdgpu_svm_default_granularity, uint, 0644);
+
+/**
+ * DOC: lockup_timeout (string)
+ * Set GPU scheduler timeout value in ms.
+ *
+ * The format can be [Non-Compute] or [GFX,Compute,SDMA,Video]. That is there can be one or
+ * multiple values specified. 0 and negative values are invalidated. They will be adjusted
+ * to the default timeout.
+ *
+ * - With one value specified, the setting will apply to all non-compute jobs.
+ * - With multiple values specified, the first one will be for GFX.
+ *   The second one is for Compute. The third and fourth ones are
+ *   for SDMA and Video.
+ *
+ * By default(with no lockup_timeout settings), the timeout for all jobs is 10000.
+ */
+MODULE_PARM_DESC(lockup_timeout,
+		 "GPU lockup timeout in ms (default: 10000 for all jobs. "
+		 "0: keep default value. negative: infinity timeout), format: for bare metal [Non-Compute] or [GFX,Compute,SDMA,Video]; "
+		 "for passthrough or sriov [all jobs] or [GFX,Compute,SDMA,Video].");
+module_param_string(lockup_timeout, amdgpu_lockup_timeout, sizeof(amdgpu_lockup_timeout), 0444);
 
+/**
+ * DOC: dpm (int)
+ * Override for dynamic power management setting
+ * (0 = disable, 1 = enable)
+ * The default is -1 (auto).
+ */
 MODULE_PARM_DESC(dpm, "DPM support (1 = enable, 0 = disable, -1 = auto)");
 module_param_named(dpm, amdgpu_dpm, int, 0444);
 
-MODULE_PARM_DESC(fw_load_type, "firmware loading type (0 = direct, 1 = SMU, 2 = PSP, -1 = auto)");
+/**
+ * DOC: fw_load_type (int)
+ * Set different firmware loading type for debugging, if supported.
+ * Set to 0 to force direct loading if supported by the ASIC.  Set
+ * to -1 to select the default loading mode for the ASIC, as defined
+ * by the driver.  The default is -1 (auto).
+ */
+MODULE_PARM_DESC(fw_load_type, "firmware loading type (3 = rlc backdoor autoload if supported, 2 = smu load if supported, 1 = psp load, 0 = force direct if supported, -1 = auto)");
 module_param_named(fw_load_type, amdgpu_fw_load_type, int, 0444);
 
+/**
+ * DOC: aspm (int)
+ * To disable ASPM (1 = enable, 0 = disable). The default is -1 (auto, enabled).
+ */
 MODULE_PARM_DESC(aspm, "ASPM support (1 = enable, 0 = disable, -1 = auto)");
 module_param_named(aspm, amdgpu_aspm, int, 0444);
 
-MODULE_PARM_DESC(runpm, "PX runtime pm (1 = force enable, 0 = disable, -1 = PX only default)");
+/**
+ * DOC: runpm (int)
+ * Override for runtime power management control for dGPUs. The amdgpu driver can dynamically power down
+ * the dGPUs when they are idle if supported. The default is -1 (auto enable).
+ * Setting the value to 0 disables this functionality.
+ * Setting the value to -2 is auto enabled with power down when displays are attached.
+ */
+MODULE_PARM_DESC(runpm, "PX runtime pm (2 = force enable with BAMACO, 1 = force enable with BACO, 0 = disable, -1 = auto, -2 = auto with displays)");
 module_param_named(runpm, amdgpu_runtime_pm, int, 0444);
 
+/**
+ * DOC: ip_block_mask (uint)
+ * Override what IP blocks are enabled on the GPU. Each GPU is a collection of IP blocks (gfx, display, video, etc.).
+ * Use this parameter to disable specific blocks. Note that the IP blocks do not have a fixed index. Some asics may not have
+ * some IPs or may include multiple instances of an IP so the ordering various from asic to asic. See the driver output in
+ * the kernel log for the list of IPs on the asic. The default is 0xffffffff (enable all blocks on a device).
+ */
 MODULE_PARM_DESC(ip_block_mask, "IP Block Mask (all blocks enabled (default))");
-module_param_named(ip_block_mask, amdgpu_ip_block_mask, uint, 0444);
+module_param_named_unsafe(ip_block_mask, amdgpu_ip_block_mask, uint, 0444);
 
+/**
+ * DOC: bapm (int)
+ * Bidirectional Application Power Management (BAPM) used to dynamically share TDP between CPU and GPU. Set value 0 to disable it.
+ * The default -1 (auto, enabled)
+ */
 MODULE_PARM_DESC(bapm, "BAPM support (1 = enable, 0 = disable, -1 = auto)");
 module_param_named(bapm, amdgpu_bapm, int, 0444);
 
+/**
+ * DOC: deep_color (int)
+ * Set 1 to enable Deep Color support. Only affects non-DC display handling. The default is 0 (disabled).
+ */
 MODULE_PARM_DESC(deep_color, "Deep Color support (1 = enable, 0 = disable (default))");
 module_param_named(deep_color, amdgpu_deep_color, int, 0444);
 
+/**
+ * DOC: vm_size (int)
+ * Override the size of the GPU's per client virtual address space in GiB.  The default is -1 (automatic for each asic).
+ */
 MODULE_PARM_DESC(vm_size, "VM address space size in gigabytes (default 64GB)");
 module_param_named(vm_size, amdgpu_vm_size, int, 0444);
 
+/**
+ * DOC: vm_fragment_size (int)
+ * Override VM fragment size in bits (4, 5, etc. 4 = 64K, 9 = 2M). The default is -1 (automatic for each asic).
+ */
 MODULE_PARM_DESC(vm_fragment_size, "VM fragment size in bits (4, 5, etc. 4 = 64K (default), Max 9 = 2M)");
 module_param_named(vm_fragment_size, amdgpu_vm_fragment_size, int, 0444);
 
+/**
+ * DOC: vm_block_size (int)
+ * Override VM page table size in bits (default depending on vm_size and hw setup). The default is -1 (automatic for each asic).
+ */
 MODULE_PARM_DESC(vm_block_size, "VM page table size in bits (default depending on vm_size)");
 module_param_named(vm_block_size, amdgpu_vm_block_size, int, 0444);
 
+/**
+ * DOC: vm_fault_stop (int)
+ * Stop on VM fault for debugging (0 = never, 1 = print first, 2 = always). The default is 0 (No stop).
+ */
 MODULE_PARM_DESC(vm_fault_stop, "Stop on VM fault (0 = never (default), 1 = print first, 2 = always)");
 module_param_named(vm_fault_stop, amdgpu_vm_fault_stop, int, 0444);
 
-MODULE_PARM_DESC(vm_debug, "Debug VM handling (0 = disabled (default), 1 = enabled)");
-module_param_named(vm_debug, amdgpu_vm_debug, int, 0644);
-
+/**
+ * DOC: vm_update_mode (int)
+ * Override VM update mode. VM updated by using CPU (0 = never, 1 = Graphics only, 2 = Compute only, 3 = Both). The default
+ * is -1 (Only in large BAR(LB) systems Compute VM tables will be updated by CPU, otherwise 0, never).
+ */
 MODULE_PARM_DESC(vm_update_mode, "VM update using CPU (0 = never (default except for large BAR(LB)), 1 = Graphics only, 2 = Compute only (default for LB), 3 = Both");
 module_param_named(vm_update_mode, amdgpu_vm_update_mode, int, 0444);
 
-MODULE_PARM_DESC(vram_page_split, "Number of pages after we split VRAM allocations (default 512, -1 = disable)");
-module_param_named(vram_page_split, amdgpu_vram_page_split, int, 0444);
-
+/**
+ * DOC: exp_hw_support (int)
+ * Enable experimental hw support (1 = enable). The default is 0 (disabled).
+ */
 MODULE_PARM_DESC(exp_hw_support, "experimental hw support (1 = enable, 0 = disable (default))");
-module_param_named(exp_hw_support, amdgpu_exp_hw_support, int, 0444);
+module_param_named_unsafe(exp_hw_support, amdgpu_exp_hw_support, int, 0444);
 
+/**
+ * DOC: dc (int)
+ * Disable/Enable Display Core driver for debugging (1 = enable, 0 = disable). The default is -1 (automatic for each asic).
+ */
+MODULE_PARM_DESC(dc, "Display Core driver (1 = enable, 0 = disable, -1 = auto (default))");
+module_param_named(dc, amdgpu_dc, int, 0444);
+
+/**
+ * DOC: sched_jobs (int)
+ * Override the max number of jobs supported in the sw queue. The default is 32.
+ */
 MODULE_PARM_DESC(sched_jobs, "the max number of jobs supported in the sw queue (default 32)");
 module_param_named(sched_jobs, amdgpu_sched_jobs, int, 0444);
 
+/**
+ * DOC: sched_hw_submission (int)
+ * Override the max number of HW submissions. The default is 2.
+ */
 MODULE_PARM_DESC(sched_hw_submission, "the max number of HW submissions (default 2)");
 module_param_named(sched_hw_submission, amdgpu_sched_hw_submission, int, 0444);
 
+/**
+ * DOC: ppfeaturemask (hexint)
+ * Override power features enabled. See enum PP_FEATURE_MASK in drivers/gpu/drm/amd/include/amd_shared.h.
+ * The default is the current set of stable power features.
+ */
 MODULE_PARM_DESC(ppfeaturemask, "all power features enabled (default))");
-module_param_named(ppfeaturemask, amdgpu_pp_feature_mask, uint, 0444);
-
-MODULE_PARM_DESC(no_evict, "Support pinning request from user space (1 = enable, 0 = disable (default))");
-module_param_named(no_evict, amdgpu_no_evict, int, 0444);
+module_param_named(ppfeaturemask, amdgpu_pp_feature_mask, hexint, 0444);
 
-MODULE_PARM_DESC(direct_gma_size, "Direct GMA size in megabytes (max 96MB)");
-module_param_named(direct_gma_size, amdgpu_direct_gma_size, int, 0444);
+/**
+ * DOC: forcelongtraining (uint)
+ * Force long memory training in resume.
+ * The default is zero, indicates short training in resume.
+ */
+MODULE_PARM_DESC(forcelongtraining, "force memory long training");
+module_param_named(forcelongtraining, amdgpu_force_long_training, uint, 0444);
 
+/**
+ * DOC: pcie_gen_cap (uint)
+ * Override PCIE gen speed capabilities. See the CAIL flags in drivers/gpu/drm/amd/include/amd_pcie.h.
+ * The default is 0 (automatic for each asic).
+ */
 MODULE_PARM_DESC(pcie_gen_cap, "PCIE Gen Caps (0: autodetect (default))");
 module_param_named(pcie_gen_cap, amdgpu_pcie_gen_cap, uint, 0444);
 
+/**
+ * DOC: pcie_lane_cap (uint)
+ * Override PCIE lanes capabilities. See the CAIL flags in drivers/gpu/drm/amd/include/amd_pcie.h.
+ * The default is 0 (automatic for each asic).
+ */
 MODULE_PARM_DESC(pcie_lane_cap, "PCIE Lane Caps (0: autodetect (default))");
 module_param_named(pcie_lane_cap, amdgpu_pcie_lane_cap, uint, 0444);
 
+/**
+ * DOC: cg_mask (ullong)
+ * Override Clockgating features enabled on GPU (0 = disable clock gating). See the AMD_CG_SUPPORT flags in
+ * drivers/gpu/drm/amd/include/amd_shared.h. The default is 0xffffffffffffffff (all enabled).
+ */
 MODULE_PARM_DESC(cg_mask, "Clockgating flags mask (0 = disable clock gating)");
-module_param_named(cg_mask, amdgpu_cg_mask, uint, 0444);
+module_param_named(cg_mask, amdgpu_cg_mask, ullong, 0444);
 
+/**
+ * DOC: pg_mask (uint)
+ * Override Powergating features enabled on GPU (0 = disable power gating). See the AMD_PG_SUPPORT flags in
+ * drivers/gpu/drm/amd/include/amd_shared.h. The default is 0xffffffff (all enabled).
+ */
 MODULE_PARM_DESC(pg_mask, "Powergating flags mask (0 = disable power gating)");
 module_param_named(pg_mask, amdgpu_pg_mask, uint, 0444);
 
+/**
+ * DOC: sdma_phase_quantum (uint)
+ * Override SDMA context switch phase quantum (x 1K GPU clock cycles, 0 = no change). The default is 32.
+ */
 MODULE_PARM_DESC(sdma_phase_quantum, "SDMA context switch phase quantum (x 1K GPU clock cycles, 0 = no change (default 32))");
 module_param_named(sdma_phase_quantum, amdgpu_sdma_phase_quantum, uint, 0444);
 
+/**
+ * DOC: disable_cu (charp)
+ * Set to disable CUs (It's set like se.sh.cu,...). The default is NULL.
+ */
 MODULE_PARM_DESC(disable_cu, "Disable CUs (se.sh.cu,...)");
 module_param_named(disable_cu, amdgpu_disable_cu, charp, 0444);
 
+/**
+ * DOC: virtual_display (charp)
+ * Set to enable virtual display feature. This feature provides a virtual display hardware on headless boards
+ * or in virtualized environments. It will be set like xxxx:xx:xx.x,x;xxxx:xx:xx.x,x. It's the pci address of
+ * the device, plus the number of crtcs to expose. E.g., 0000:26:00.0,4 would enable 4 virtual crtcs on the pci
+ * device at 26:00.0. The default is NULL.
+ */
 MODULE_PARM_DESC(virtual_display,
 		 "Enable virtual display feature (the virtual_display will be set like xxxx:xx:xx.x,x;xxxx:xx:xx.x,x)");
 module_param_named(virtual_display, amdgpu_virtual_display, charp, 0444);
 
-MODULE_PARM_DESC(ngg, "Next Generation Graphics (1 = enable, 0 = disable(default depending on gfx))");
-module_param_named(ngg, amdgpu_ngg, int, 0444);
+/**
+ * DOC: lbpw (int)
+ * Override Load Balancing Per Watt (LBPW) support (1 = enable, 0 = disable). The default is -1 (auto, enabled).
+ */
+MODULE_PARM_DESC(lbpw, "Load Balancing Per Watt (LBPW) support (1 = enable, 0 = disable, -1 = auto)");
+module_param_named(lbpw, amdgpu_lbpw, int, 0444);
+
+MODULE_PARM_DESC(compute_multipipe, "Force compute queues to be spread across pipes (1 = enable, 0 = disable, -1 = auto)");
+module_param_named(compute_multipipe, amdgpu_compute_multipipe, int, 0444);
 
-MODULE_PARM_DESC(prim_buf_per_se, "the size of Primitive Buffer per Shader Engine (default depending on gfx)");
-module_param_named(prim_buf_per_se, amdgpu_prim_buf_per_se, int, 0444);
+/**
+ * DOC: gpu_recovery (int)
+ * Set to enable GPU recovery mechanism (1 = enable, 0 = disable). The default is -1 (auto, disabled except SRIOV).
+ */
+MODULE_PARM_DESC(gpu_recovery, "Enable GPU recovery mechanism, (1 = enable, 0 = disable, -1 = auto)");
+module_param_named_unsafe(gpu_recovery, amdgpu_gpu_recovery, int, 0444);
 
-MODULE_PARM_DESC(pos_buf_per_se, "the size of Position Buffer per Shader Engine (default depending on gfx)");
-module_param_named(pos_buf_per_se, amdgpu_pos_buf_per_se, int, 0444);
+/**
+ * DOC: emu_mode (int)
+ * Set value 1 to enable emulation mode. This is only needed when running on an emulator. The default is 0 (disabled).
+ */
+MODULE_PARM_DESC(emu_mode, "Emulation mode, (1 = enable, 0 = disable)");
+module_param_named_unsafe(emu_mode, amdgpu_emu_mode, int, 0444);
 
-MODULE_PARM_DESC(cntl_sb_buf_per_se, "the size of Control Sideband per Shader Engine (default depending on gfx)");
-module_param_named(cntl_sb_buf_per_se, amdgpu_cntl_sb_buf_per_se, int, 0444);
+/**
+ * DOC: ras_enable (int)
+ * Enable RAS features on the GPU (0 = disable, 1 = enable, -1 = auto (default))
+ */
+MODULE_PARM_DESC(ras_enable, "Enable RAS features on the GPU (0 = disable, 1 = enable, -1 = auto (default))");
+module_param_named(ras_enable, amdgpu_ras_enable, int, 0444);
 
-MODULE_PARM_DESC(param_buf_per_se, "the size of Off-Chip Pramater Cache per Shader Engine (default depending on gfx)");
-module_param_named(param_buf_per_se, amdgpu_param_buf_per_se, int, 0444);
+/**
+ * DOC: ras_mask (uint)
+ * Mask of RAS features to enable (default 0xffffffff), only valid when ras_enable == 1
+ * See the flags in drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
+ */
+MODULE_PARM_DESC(ras_mask, "Mask of RAS features to enable (default 0xffffffff), only valid when ras_enable == 1");
+module_param_named(ras_mask, amdgpu_ras_mask, uint, 0444);
 
-MODULE_PARM_DESC(job_hang_limit, "how much time allow a job hang and not drop it (default 0)");
-module_param_named(job_hang_limit, amdgpu_job_hang_limit, int ,0444);
+/**
+ * DOC: timeout_fatal_disable (bool)
+ * Disable Watchdog timeout fatal error event
+ */
+MODULE_PARM_DESC(timeout_fatal_disable, "disable watchdog timeout fatal error (false = default)");
+module_param_named(timeout_fatal_disable, amdgpu_watchdog_timer.timeout_fatal_disable, bool, 0644);
 
-MODULE_PARM_DESC(lbpw, "Load Balancing Per Watt (LBPW) support (1 = enable, 0 = disable, -1 = auto)");
-module_param_named(lbpw, amdgpu_lbpw, int, 0444);
+/**
+ * DOC: timeout_period (uint)
+ * Modify the watchdog timeout max_cycles as (1 << period)
+ */
+MODULE_PARM_DESC(timeout_period, "watchdog timeout period (0 = timeout disabled, 1 ~ 0x23 = timeout maxcycles = (1 << period)");
+module_param_named(timeout_period, amdgpu_watchdog_timer.period, uint, 0644);
 
+/**
+ * DOC: si_support (int)
+ * Set SI support driver. This parameter works after set config CONFIG_DRM_AMDGPU_SI. For SI asic, when radeon driver is enabled,
+ * set value 0 to use radeon driver, while set value 1 to use amdgpu driver. The default is using radeon driver when it available,
+ * otherwise using amdgpu driver.
+ */
 #ifdef CONFIG_DRM_AMDGPU_SI
 
-#if defined(CONFIG_DRM_RADEON) || defined(CONFIG_DRM_RADEON_MODULE)
-int amdgpu_si_support = 0;
+#if IS_ENABLED(CONFIG_DRM_RADEON) || IS_ENABLED(CONFIG_DRM_RADEON_MODULE)
+int amdgpu_si_support;
 MODULE_PARM_DESC(si_support, "SI support (1 = enabled, 0 = disabled (default))");
 #else
 int amdgpu_si_support = 1;
@@ -277,10 +640,16 @@ MODULE_PARM_DESC(si_support, "SI support (1 = enabled (default), 0 = disabled)")
 module_param_named(si_support, amdgpu_si_support, int, 0444);
 #endif
 
+/**
+ * DOC: cik_support (int)
+ * Set CIK support driver. This parameter works after set config CONFIG_DRM_AMDGPU_CIK. For CIK asic, when radeon driver is enabled,
+ * set value 0 to use radeon driver, while set value 1 to use amdgpu driver. The default is using radeon driver when it available,
+ * otherwise using amdgpu driver.
+ */
 #ifdef CONFIG_DRM_AMDGPU_CIK
 
-#if defined(CONFIG_DRM_RADEON) || defined(CONFIG_DRM_RADEON_MODULE)
-int amdgpu_cik_support = 0;
+#if IS_ENABLED(CONFIG_DRM_RADEON) || IS_ENABLED(CONFIG_DRM_RADEON_MODULE)
+int amdgpu_cik_support;
 MODULE_PARM_DESC(cik_support, "CIK support (1 = enabled, 0 = disabled (default))");
 #else
 int amdgpu_cik_support = 1;
@@ -290,9 +659,1187 @@ MODULE_PARM_DESC(cik_support, "CIK support (1 = enabled (default), 0 = disabled)
 module_param_named(cik_support, amdgpu_cik_support, int, 0444);
 #endif
 
+/**
+ * DOC: smu_memory_pool_size (uint)
+ * It is used to reserve gtt for smu debug usage, setting value 0 to disable it. The actual size is value * 256MiB.
+ * E.g. 0x1 = 256Mbyte, 0x2 = 512Mbyte, 0x4 = 1 Gbyte, 0x8 = 2GByte. The default is 0 (disabled).
+ */
+MODULE_PARM_DESC(smu_memory_pool_size,
+	"reserve gtt for smu debug usage, 0 = disable,0x1 = 256Mbyte, 0x2 = 512Mbyte, 0x4 = 1 Gbyte, 0x8 = 2GByte");
+module_param_named(smu_memory_pool_size, amdgpu_smu_memory_pool_size, uint, 0444);
+
+/**
+ * DOC: async_gfx_ring (int)
+ * It is used to enable gfx rings that could be configured with different prioritites or equal priorities
+ */
+MODULE_PARM_DESC(async_gfx_ring,
+	"Asynchronous GFX rings that could be configured with either different priorities (HP3D ring and LP3D ring), or equal priorities (0 = disabled, 1 = enabled (default))");
+module_param_named(async_gfx_ring, amdgpu_async_gfx_ring, int, 0444);
+
+/**
+ * DOC: mcbp (int)
+ * It is used to enable mid command buffer preemption. (0 = disabled, 1 = enabled, -1 auto (default))
+ */
+MODULE_PARM_DESC(mcbp,
+	"Enable Mid-command buffer preemption (0 = disabled, 1 = enabled), -1 = auto (default)");
+module_param_named(mcbp, amdgpu_mcbp, int, 0444);
+
+/**
+ * DOC: discovery (int)
+ * Allow driver to discover hardware IP information from IP Discovery table at the top of VRAM.
+ * (-1 = auto (default), 0 = disabled, 1 = enabled, 2 = use ip_discovery table from file)
+ */
+MODULE_PARM_DESC(discovery,
+	"Allow driver to discover hardware IPs from IP Discovery table at the top of VRAM");
+module_param_named(discovery, amdgpu_discovery, int, 0444);
+
+/**
+ * DOC: mes (int)
+ * Enable Micro Engine Scheduler. This is a new hw scheduling engine for gfx, sdma, and compute.
+ * (0 = disabled (default), 1 = enabled)
+ */
+MODULE_PARM_DESC(mes,
+	"Enable Micro Engine Scheduler (0 = disabled (default), 1 = enabled)");
+module_param_named(mes, amdgpu_mes, int, 0444);
+
+/**
+ * DOC: mes_log_enable (int)
+ * Enable Micro Engine Scheduler log. This is used to enable/disable MES internal log.
+ * (0 = disabled (default), 1 = enabled)
+ */
+MODULE_PARM_DESC(mes_log_enable,
+	"Enable Micro Engine Scheduler log (0 = disabled (default), 1 = enabled)");
+module_param_named(mes_log_enable, amdgpu_mes_log_enable, int, 0444);
+
+/**
+ * DOC: mes_kiq (int)
+ * Enable Micro Engine Scheduler KIQ. This is a new engine pipe for kiq.
+ * (0 = disabled (default), 1 = enabled)
+ */
+MODULE_PARM_DESC(mes_kiq,
+	"Enable Micro Engine Scheduler KIQ (0 = disabled (default), 1 = enabled)");
+module_param_named(mes_kiq, amdgpu_mes_kiq, int, 0444);
+
+/**
+ * DOC: uni_mes (int)
+ * Enable Unified Micro Engine Scheduler. This is a new engine pipe for unified scheduler.
+ * (0 = disabled (default), 1 = enabled)
+ */
+MODULE_PARM_DESC(uni_mes,
+	"Enable Unified Micro Engine Scheduler (0 = disabled, 1 = enabled(default)");
+module_param_named(uni_mes, amdgpu_uni_mes, int, 0444);
+
+/**
+ * DOC: noretry (int)
+ * Disable XNACK retry in the SQ by default on GFXv9 hardware. On ASICs that
+ * do not support per-process XNACK this also disables retry page faults.
+ * (0 = retry enabled, 1 = retry disabled, -1 auto (default))
+ */
+MODULE_PARM_DESC(noretry,
+	"Disable retry faults (0 = retry enabled, 1 = retry disabled, -1 auto (default))");
+module_param_named(noretry, amdgpu_noretry, int, 0644);
+
+/**
+ * DOC: force_asic_type (int)
+ * A non negative value used to specify the asic type for all supported GPUs.
+ */
+MODULE_PARM_DESC(force_asic_type,
+	"A non negative value used to specify the asic type for all supported GPUs");
+module_param_named_unsafe(force_asic_type, amdgpu_force_asic_type, int, 0444);
+
+/**
+ * DOC: use_xgmi_p2p (int)
+ * Enables/disables XGMI P2P interface (0 = disable, 1 = enable).
+ */
+MODULE_PARM_DESC(use_xgmi_p2p,
+	"Enable XGMI P2P interface (0 = disable; 1 = enable (default))");
+module_param_named(use_xgmi_p2p, amdgpu_use_xgmi_p2p, int, 0444);
+
+
+#ifdef CONFIG_HSA_AMD
+/**
+ * DOC: sched_policy (int)
+ * Set scheduling policy. Default is HWS(hardware scheduling) with over-subscription.
+ * Setting 1 disables over-subscription. Setting 2 disables HWS and statically
+ * assigns queues to HQDs.
+ */
+int sched_policy = KFD_SCHED_POLICY_HWS;
+module_param_unsafe(sched_policy, int, 0444);
+MODULE_PARM_DESC(sched_policy,
+	"Scheduling policy (0 = HWS (Default), 1 = HWS without over-subscription, 2 = Non-HWS (Used for debugging only)");
+
+/**
+ * DOC: hws_max_conc_proc (int)
+ * Maximum number of processes that HWS can schedule concurrently. The maximum is the
+ * number of VMIDs assigned to the HWS, which is also the default.
+ */
+int hws_max_conc_proc = -1;
+module_param(hws_max_conc_proc, int, 0444);
+MODULE_PARM_DESC(hws_max_conc_proc,
+	"Max # processes HWS can execute concurrently when sched_policy=0 (0 = no concurrency, #VMIDs for KFD = Maximum(default))");
+
+/**
+ * DOC: cwsr_enable (int)
+ * CWSR(compute wave store and resume) allows the GPU to preempt shader execution in
+ * the middle of a compute wave. Default is 1 to enable this feature. Setting 0
+ * disables it.
+ */
+int cwsr_enable = 1;
+module_param(cwsr_enable, int, 0444);
+MODULE_PARM_DESC(cwsr_enable, "CWSR enable (0 = Off, 1 = On (Default))");
+
+/**
+ * DOC: max_num_of_queues_per_device (int)
+ * Maximum number of queues per device. Valid setting is between 1 and 4096. Default
+ * is 4096.
+ */
+int max_num_of_queues_per_device = KFD_MAX_NUM_OF_QUEUES_PER_DEVICE_DEFAULT;
+module_param(max_num_of_queues_per_device, int, 0444);
+MODULE_PARM_DESC(max_num_of_queues_per_device,
+	"Maximum number of supported queues per device (1 = Minimum, 4096 = default)");
+
+/**
+ * DOC: send_sigterm (int)
+ * Send sigterm to HSA process on unhandled exceptions. Default is not to send sigterm
+ * but just print errors on dmesg. Setting 1 enables sending sigterm.
+ */
+int send_sigterm;
+module_param(send_sigterm, int, 0444);
+MODULE_PARM_DESC(send_sigterm,
+	"Send sigterm to HSA process on unhandled exception (0 = disable, 1 = enable)");
+
+/**
+ * DOC: halt_if_hws_hang (int)
+ * Halt if HWS hang is detected. Default value, 0, disables the halt on hang.
+ * Setting 1 enables halt on hang.
+ */
+int halt_if_hws_hang;
+module_param_unsafe(halt_if_hws_hang, int, 0644);
+MODULE_PARM_DESC(halt_if_hws_hang, "Halt if HWS hang is detected (0 = off (default), 1 = on)");
+
+/**
+ * DOC: hws_gws_support(bool)
+ * Assume that HWS supports GWS barriers regardless of what firmware version
+ * check says. Default value: false (rely on MEC2 firmware version check).
+ */
+bool hws_gws_support;
+module_param_unsafe(hws_gws_support, bool, 0444);
+MODULE_PARM_DESC(hws_gws_support, "Assume MEC2 FW supports GWS barriers (false = rely on FW version check (Default), true = force supported)");
+
+/**
+ * DOC: queue_preemption_timeout_ms (int)
+ * queue preemption timeout in ms (1 = Minimum, 9000 = default)
+ */
+int queue_preemption_timeout_ms = 9000;
+module_param(queue_preemption_timeout_ms, int, 0644);
+MODULE_PARM_DESC(queue_preemption_timeout_ms, "queue preemption timeout in ms (1 = Minimum, 9000 = default)");
+
+/**
+ * DOC: debug_evictions(bool)
+ * Enable extra debug messages to help determine the cause of evictions
+ */
+bool debug_evictions;
+module_param(debug_evictions, bool, 0644);
+MODULE_PARM_DESC(debug_evictions, "enable eviction debug messages (false = default)");
+
+/**
+ * DOC: no_system_mem_limit(bool)
+ * Disable system memory limit, to support multiple process shared memory
+ */
+bool no_system_mem_limit;
+module_param(no_system_mem_limit, bool, 0644);
+MODULE_PARM_DESC(no_system_mem_limit, "disable system memory limit (false = default)");
+
+/**
+ * DOC: no_queue_eviction_on_vm_fault (int)
+ * If set, process queues will not be evicted on gpuvm fault. This is to keep the wavefront context for debugging (0 = queue eviction, 1 = no queue eviction). The default is 0 (queue eviction).
+ */
+int amdgpu_no_queue_eviction_on_vm_fault;
+MODULE_PARM_DESC(no_queue_eviction_on_vm_fault, "No queue eviction on VM fault (0 = queue eviction, 1 = no queue eviction)");
+module_param_named_unsafe(no_queue_eviction_on_vm_fault, amdgpu_no_queue_eviction_on_vm_fault, int, 0444);
+#endif
+
+/**
+ * DOC: mtype_local (int)
+ */
+int amdgpu_mtype_local;
+MODULE_PARM_DESC(mtype_local, "MTYPE for local memory (0 = MTYPE_RW (default), 1 = MTYPE_NC, 2 = MTYPE_CC)");
+module_param_named_unsafe(mtype_local, amdgpu_mtype_local, int, 0444);
+
+/**
+ * DOC: pcie_p2p (bool)
+ * Enable PCIe P2P (requires large-BAR). Default value: true (on)
+ */
+#ifdef CONFIG_HSA_AMD_P2P
+bool pcie_p2p = true;
+module_param(pcie_p2p, bool, 0444);
+MODULE_PARM_DESC(pcie_p2p, "Enable PCIe P2P (requires large-BAR). (N = off, Y = on(default))");
+#endif
+
+/**
+ * DOC: dcfeaturemask (uint)
+ * Override display features enabled. See enum DC_FEATURE_MASK in drivers/gpu/drm/amd/include/amd_shared.h.
+ * The default is the current set of stable display features.
+ */
+MODULE_PARM_DESC(dcfeaturemask, "all stable DC features enabled (default))");
+module_param_named(dcfeaturemask, amdgpu_dc_feature_mask, uint, 0444);
+
+/**
+ * DOC: dcdebugmask (uint)
+ * Display debug options. See enum DC_DEBUG_MASK in drivers/gpu/drm/amd/include/amd_shared.h.
+ */
+MODULE_PARM_DESC(dcdebugmask, "all debug options disabled (default))");
+module_param_named(dcdebugmask, amdgpu_dc_debug_mask, uint, 0444);
+
+MODULE_PARM_DESC(visualconfirm, "Visual confirm (0 = off (default), 1 = MPO, 5 = PSR)");
+module_param_named(visualconfirm, amdgpu_dc_visual_confirm, uint, 0444);
+
+/**
+ * DOC: abmlevel (uint)
+ * Override the default ABM (Adaptive Backlight Management) level used for DC
+ * enabled hardware. Requires DMCU to be supported and loaded.
+ * Valid levels are 0-4. A value of 0 indicates that ABM should be disabled by
+ * default. Values 1-4 control the maximum allowable brightness reduction via
+ * the ABM algorithm, with 1 being the least reduction and 4 being the most
+ * reduction.
+ *
+ * Defaults to -1, or auto. Userspace can only override this level after
+ * boot if it's set to auto.
+ */
+int amdgpu_dm_abm_level = -1;
+MODULE_PARM_DESC(abmlevel,
+		 "ABM level (0 = off, 1-4 = backlight reduction level, -1 auto (default))");
+module_param_named(abmlevel, amdgpu_dm_abm_level, int, 0444);
+
+int amdgpu_backlight = -1;
+MODULE_PARM_DESC(backlight, "Backlight control (0 = pwm, 1 = aux, -1 auto (default))");
+module_param_named(backlight, amdgpu_backlight, bint, 0444);
+
+/**
+ * DOC: damageclips (int)
+ * Enable or disable damage clips support. If damage clips support is disabled,
+ * we will force full frame updates, irrespective of what user space sends to
+ * us.
+ *
+ * Defaults to -1 (where it is enabled unless a PSR-SU display is detected).
+ */
+MODULE_PARM_DESC(damageclips,
+		 "Damage clips support (0 = disable, 1 = enable, -1 auto (default))");
+module_param_named(damageclips, amdgpu_damage_clips, int, 0444);
+
+/**
+ * DOC: tmz (int)
+ * Trusted Memory Zone (TMZ) is a method to protect data being written
+ * to or read from memory.
+ *
+ * The default value: 0 (off).  TODO: change to auto till it is completed.
+ */
+MODULE_PARM_DESC(tmz, "Enable TMZ feature (-1 = auto (default), 0 = off, 1 = on)");
+module_param_named(tmz, amdgpu_tmz, int, 0444);
+
+/**
+ * DOC: freesync_video (uint)
+ * Enable the optimization to adjust front porch timing to achieve seamless
+ * mode change experience when setting a freesync supported mode for which full
+ * modeset is not needed.
+ *
+ * The Display Core will add a set of modes derived from the base FreeSync
+ * video mode into the corresponding connector's mode list based on commonly
+ * used refresh rates and VRR range of the connected display, when users enable
+ * this feature. From the userspace perspective, they can see a seamless mode
+ * change experience when the change between different refresh rates under the
+ * same resolution. Additionally, userspace applications such as Video playback
+ * can read this modeset list and change the refresh rate based on the video
+ * frame rate. Finally, the userspace can also derive an appropriate mode for a
+ * particular refresh rate based on the FreeSync Mode and add it to the
+ * connector's mode list.
+ *
+ * Note: This is an experimental feature.
+ *
+ * The default value: 0 (off).
+ */
+MODULE_PARM_DESC(
+	freesync_video,
+	"Adds additional modes via VRR for refresh changes without a full modeset (0 = off (default), 1 = on)");
+module_param_named(freesync_video, amdgpu_freesync_vid_mode, uint, 0444);
+
+/**
+ * DOC: reset_method (int)
+ * GPU reset method (-1 = auto (default), 0 = legacy, 1 = mode0, 2 = mode1, 3 = mode2, 4 = baco)
+ */
+MODULE_PARM_DESC(reset_method, "GPU reset method (-1 = auto (default), 0 = legacy, 1 = mode0, 2 = mode1, 3 = mode2, 4 = baco/bamaco)");
+module_param_named_unsafe(reset_method, amdgpu_reset_method, int, 0644);
+
+/**
+ * DOC: bad_page_threshold (int) Bad page threshold is specifies the
+ * threshold value of faulty pages detected by RAS ECC, which may
+ * result in the GPU entering bad status when the number of total
+ * faulty pages by ECC exceeds the threshold value.
+ */
+MODULE_PARM_DESC(bad_page_threshold, "Bad page threshold(-1 = ignore threshold (default value), 0 = disable bad page retirement, -2 = threshold determined by a formula, 0 < threshold < max records, user-defined threshold)");
+module_param_named(bad_page_threshold, amdgpu_bad_page_threshold, int, 0444);
+
+MODULE_PARM_DESC(num_kcq, "number of kernel compute queue user want to setup (8 if set to greater than 8 or less than 0, only affect gfx 8+)");
+module_param_named(num_kcq, amdgpu_num_kcq, int, 0444);
+
+/**
+ * DOC: vcnfw_log (int)
+ * Enable vcnfw log output for debugging, the default is disabled.
+ */
+MODULE_PARM_DESC(vcnfw_log, "Enable vcnfw log(0 = disable (default value), 1 = enable)");
+module_param_named(vcnfw_log, amdgpu_vcnfw_log, int, 0444);
+
+/**
+ * DOC: sg_display (int)
+ * Disable S/G (scatter/gather) display (i.e., display from system memory).
+ * This option is only relevant on APUs.  Set this option to 0 to disable
+ * S/G display if you experience flickering or other issues under memory
+ * pressure and report the issue.
+ */
+MODULE_PARM_DESC(sg_display, "S/G Display (-1 = auto (default), 0 = disable)");
+module_param_named(sg_display, amdgpu_sg_display, int, 0444);
+
+/**
+ * DOC: umsch_mm (int)
+ * Enable Multi Media User Mode Scheduler. This is a HW scheduling engine for VCN and VPE.
+ * (0 = disabled (default), 1 = enabled)
+ */
+MODULE_PARM_DESC(umsch_mm,
+	"Enable Multi Media User Mode Scheduler (0 = disabled (default), 1 = enabled)");
+module_param_named(umsch_mm, amdgpu_umsch_mm, int, 0444);
+
+/**
+ * DOC: umsch_mm_fwlog (int)
+ * Enable umschfw log output for debugging, the default is disabled.
+ */
+MODULE_PARM_DESC(umsch_mm_fwlog, "Enable umschfw log(0 = disable (default value), 1 = enable)");
+module_param_named(umsch_mm_fwlog, amdgpu_umsch_mm_fwlog, int, 0444);
+
+/**
+ * DOC: smu_pptable_id (int)
+ * Used to override pptable id. id = 0 use VBIOS pptable.
+ * id > 0 use the soft pptable with specicfied id.
+ */
+MODULE_PARM_DESC(smu_pptable_id,
+	"specify pptable id to be used (-1 = auto(default) value, 0 = use pptable from vbios, > 0 = soft pptable id)");
+module_param_named(smu_pptable_id, amdgpu_smu_pptable_id, int, 0444);
+
+/**
+ * DOC: partition_mode (int)
+ * Used to override the default SPX mode.
+ */
+MODULE_PARM_DESC(
+	user_partt_mode,
+	"specify partition mode to be used (-2 = AMDGPU_AUTO_COMPUTE_PARTITION_MODE(default value) \
+						0 = AMDGPU_SPX_PARTITION_MODE, \
+						1 = AMDGPU_DPX_PARTITION_MODE, \
+						2 = AMDGPU_TPX_PARTITION_MODE, \
+						3 = AMDGPU_QPX_PARTITION_MODE, \
+						4 = AMDGPU_CPX_PARTITION_MODE)");
+module_param_named(user_partt_mode, amdgpu_user_partt_mode, uint, 0444);
+
+
+/**
+ * DOC: enforce_isolation (int)
+ * enforce process isolation between graphics and compute.
+ * (-1 = auto, 0 = disable, 1 = enable, 2 = enable legacy mode, 3 = enable without cleaner shader)
+ */
+module_param_named(enforce_isolation, amdgpu_enforce_isolation, int, 0444);
+MODULE_PARM_DESC(enforce_isolation,
+"enforce process isolation between graphics and compute. (-1 = auto, 0 = disable, 1 = enable, 2 = enable legacy mode, 3 = enable without cleaner shader)");
+
+/**
+ * DOC: modeset (int)
+ * Override nomodeset (1 = override, -1 = auto). The default is -1 (auto).
+ */
+MODULE_PARM_DESC(modeset, "Override nomodeset (1 = enable, -1 = auto)");
+module_param_named(modeset, amdgpu_modeset, int, 0444);
+
+/**
+ * DOC: seamless (int)
+ * Seamless boot will keep the image on the screen during the boot process.
+ */
+MODULE_PARM_DESC(seamless, "Seamless boot (-1 = auto (default), 0 = disable, 1 = enable)");
+module_param_named(seamless, amdgpu_seamless, int, 0444);
+
+/**
+ * DOC: debug_mask (uint)
+ * Debug options for amdgpu, work as a binary mask with the following options:
+ *
+ * - 0x1: Debug VM handling
+ * - 0x2: Enable simulating large-bar capability on non-large bar system. This
+ *   limits the VRAM size reported to ROCm applications to the visible
+ *   size, usually 256MB.
+ * - 0x4: Disable GPU soft recovery, always do a full reset
+ * - 0x8: Use VRAM for firmware loading
+ * - 0x10: Enable ACA based RAS logging
+ * - 0x20: Enable experimental resets
+ * - 0x40: Disable ring resets
+ * - 0x80: Use VRAM for SMU pool
+ */
+MODULE_PARM_DESC(debug_mask, "debug options for amdgpu, disabled by default");
+module_param_named_unsafe(debug_mask, amdgpu_debug_mask, uint, 0444);
+
+/**
+ * DOC: agp (int)
+ * Enable the AGP aperture.  This provides an aperture in the GPU's internal
+ * address space for direct access to system memory.  Note that these accesses
+ * are non-snooped, so they are only used for access to uncached memory.
+ */
+MODULE_PARM_DESC(agp, "AGP (-1 = auto (default), 0 = disable, 1 = enable)");
+module_param_named(agp, amdgpu_agp, int, 0444);
+
+/**
+ * DOC: wbrf (int)
+ * Enable Wifi RFI interference mitigation feature.
+ * Due to electrical and mechanical constraints there may be likely interference of
+ * relatively high-powered harmonics of the (G-)DDR memory clocks with local radio
+ * module frequency bands used by Wifi 6/6e/7. To mitigate the possible RFI interference,
+ * with this feature enabled, PMFW will use either “shadowed P-State” or “P-State” based
+ * on active list of frequencies in-use (to be avoided) as part of initial setting or
+ * P-state transition. However, there may be potential performance impact with this
+ * feature enabled.
+ * (0 = disabled, 1 = enabled, -1 = auto (default setting, will be enabled if supported))
+ */
+MODULE_PARM_DESC(wbrf,
+	"Enable Wifi RFI interference mitigation (0 = disabled, 1 = enabled, -1 = auto(default)");
+module_param_named(wbrf, amdgpu_wbrf, int, 0444);
+
+/**
+ * DOC: rebar (int)
+ * Allow BAR resizing.  Disable this to prevent the driver from attempting
+ * to resize the BAR if the GPU supports it and there is available MMIO space.
+ * Note that this just prevents the driver from resizing the BAR.  The BIOS
+ * may have already resized the BAR at boot time.
+ */
+MODULE_PARM_DESC(rebar, "Resizable BAR (-1 = auto (default), 0 = disable, 1 = enable)");
+module_param_named(rebar, amdgpu_rebar, int, 0444);
+
+/**
+ * DOC: user_queue (int)
+ * Enable user queues on systems that support user queues. Possible values:
+ *
+ * - -1 = auto (ASIC specific default)
+ * -  0 = user queues disabled
+ * -  1 = user queues enabled and kernel queues enabled (if supported)
+ * -  2 = user queues enabled and kernel queues disabled
+ */
+MODULE_PARM_DESC(user_queue, "Enable user queues (-1 = auto (default), 0 = disable, 1 = enable, 2 = enable UQs and disable KQs)");
+module_param_named(user_queue, amdgpu_user_queue, int, 0444);
+
+/* These devices are not supported by amdgpu.
+ * They are supported by the mach64, r128, radeon drivers
+ */
+static const u16 amdgpu_unsupported_pciidlist[] = {
+	/* mach64 */
+	0x4354,
+	0x4358,
+	0x4554,
+	0x4742,
+	0x4744,
+	0x4749,
+	0x474C,
+	0x474D,
+	0x474E,
+	0x474F,
+	0x4750,
+	0x4751,
+	0x4752,
+	0x4753,
+	0x4754,
+	0x4755,
+	0x4756,
+	0x4757,
+	0x4758,
+	0x4759,
+	0x475A,
+	0x4C42,
+	0x4C44,
+	0x4C47,
+	0x4C49,
+	0x4C4D,
+	0x4C4E,
+	0x4C50,
+	0x4C51,
+	0x4C52,
+	0x4C53,
+	0x5654,
+	0x5655,
+	0x5656,
+	/* r128 */
+	0x4c45,
+	0x4c46,
+	0x4d46,
+	0x4d4c,
+	0x5041,
+	0x5042,
+	0x5043,
+	0x5044,
+	0x5045,
+	0x5046,
+	0x5047,
+	0x5048,
+	0x5049,
+	0x504A,
+	0x504B,
+	0x504C,
+	0x504D,
+	0x504E,
+	0x504F,
+	0x5050,
+	0x5051,
+	0x5052,
+	0x5053,
+	0x5054,
+	0x5055,
+	0x5056,
+	0x5057,
+	0x5058,
+	0x5245,
+	0x5246,
+	0x5247,
+	0x524b,
+	0x524c,
+	0x534d,
+	0x5446,
+	0x544C,
+	0x5452,
+	/* radeon */
+	0x3150,
+	0x3151,
+	0x3152,
+	0x3154,
+	0x3155,
+	0x3E50,
+	0x3E54,
+	0x4136,
+	0x4137,
+	0x4144,
+	0x4145,
+	0x4146,
+	0x4147,
+	0x4148,
+	0x4149,
+	0x414A,
+	0x414B,
+	0x4150,
+	0x4151,
+	0x4152,
+	0x4153,
+	0x4154,
+	0x4155,
+	0x4156,
+	0x4237,
+	0x4242,
+	0x4336,
+	0x4337,
+	0x4437,
+	0x4966,
+	0x4967,
+	0x4A48,
+	0x4A49,
+	0x4A4A,
+	0x4A4B,
+	0x4A4C,
+	0x4A4D,
+	0x4A4E,
+	0x4A4F,
+	0x4A50,
+	0x4A54,
+	0x4B48,
+	0x4B49,
+	0x4B4A,
+	0x4B4B,
+	0x4B4C,
+	0x4C57,
+	0x4C58,
+	0x4C59,
+	0x4C5A,
+	0x4C64,
+	0x4C66,
+	0x4C67,
+	0x4E44,
+	0x4E45,
+	0x4E46,
+	0x4E47,
+	0x4E48,
+	0x4E49,
+	0x4E4A,
+	0x4E4B,
+	0x4E50,
+	0x4E51,
+	0x4E52,
+	0x4E53,
+	0x4E54,
+	0x4E56,
+	0x5144,
+	0x5145,
+	0x5146,
+	0x5147,
+	0x5148,
+	0x514C,
+	0x514D,
+	0x5157,
+	0x5158,
+	0x5159,
+	0x515A,
+	0x515E,
+	0x5460,
+	0x5462,
+	0x5464,
+	0x5548,
+	0x5549,
+	0x554A,
+	0x554B,
+	0x554C,
+	0x554D,
+	0x554E,
+	0x554F,
+	0x5550,
+	0x5551,
+	0x5552,
+	0x5554,
+	0x564A,
+	0x564B,
+	0x564F,
+	0x5652,
+	0x5653,
+	0x5657,
+	0x5834,
+	0x5835,
+	0x5954,
+	0x5955,
+	0x5974,
+	0x5975,
+	0x5960,
+	0x5961,
+	0x5962,
+	0x5964,
+	0x5965,
+	0x5969,
+	0x5a41,
+	0x5a42,
+	0x5a61,
+	0x5a62,
+	0x5b60,
+	0x5b62,
+	0x5b63,
+	0x5b64,
+	0x5b65,
+	0x5c61,
+	0x5c63,
+	0x5d48,
+	0x5d49,
+	0x5d4a,
+	0x5d4c,
+	0x5d4d,
+	0x5d4e,
+	0x5d4f,
+	0x5d50,
+	0x5d52,
+	0x5d57,
+	0x5e48,
+	0x5e4a,
+	0x5e4b,
+	0x5e4c,
+	0x5e4d,
+	0x5e4f,
+	0x6700,
+	0x6701,
+	0x6702,
+	0x6703,
+	0x6704,
+	0x6705,
+	0x6706,
+	0x6707,
+	0x6708,
+	0x6709,
+	0x6718,
+	0x6719,
+	0x671c,
+	0x671d,
+	0x671f,
+	0x6720,
+	0x6721,
+	0x6722,
+	0x6723,
+	0x6724,
+	0x6725,
+	0x6726,
+	0x6727,
+	0x6728,
+	0x6729,
+	0x6738,
+	0x6739,
+	0x673e,
+	0x6740,
+	0x6741,
+	0x6742,
+	0x6743,
+	0x6744,
+	0x6745,
+	0x6746,
+	0x6747,
+	0x6748,
+	0x6749,
+	0x674A,
+	0x6750,
+	0x6751,
+	0x6758,
+	0x6759,
+	0x675B,
+	0x675D,
+	0x675F,
+	0x6760,
+	0x6761,
+	0x6762,
+	0x6763,
+	0x6764,
+	0x6765,
+	0x6766,
+	0x6767,
+	0x6768,
+	0x6770,
+	0x6771,
+	0x6772,
+	0x6778,
+	0x6779,
+	0x677B,
+	0x6840,
+	0x6841,
+	0x6842,
+	0x6843,
+	0x6849,
+	0x684C,
+	0x6850,
+	0x6858,
+	0x6859,
+	0x6880,
+	0x6888,
+	0x6889,
+	0x688A,
+	0x688C,
+	0x688D,
+	0x6898,
+	0x6899,
+	0x689b,
+	0x689c,
+	0x689d,
+	0x689e,
+	0x68a0,
+	0x68a1,
+	0x68a8,
+	0x68a9,
+	0x68b0,
+	0x68b8,
+	0x68b9,
+	0x68ba,
+	0x68be,
+	0x68bf,
+	0x68c0,
+	0x68c1,
+	0x68c7,
+	0x68c8,
+	0x68c9,
+	0x68d8,
+	0x68d9,
+	0x68da,
+	0x68de,
+	0x68e0,
+	0x68e1,
+	0x68e4,
+	0x68e5,
+	0x68e8,
+	0x68e9,
+	0x68f1,
+	0x68f2,
+	0x68f8,
+	0x68f9,
+	0x68fa,
+	0x68fe,
+	0x7100,
+	0x7101,
+	0x7102,
+	0x7103,
+	0x7104,
+	0x7105,
+	0x7106,
+	0x7108,
+	0x7109,
+	0x710A,
+	0x710B,
+	0x710C,
+	0x710E,
+	0x710F,
+	0x7140,
+	0x7141,
+	0x7142,
+	0x7143,
+	0x7144,
+	0x7145,
+	0x7146,
+	0x7147,
+	0x7149,
+	0x714A,
+	0x714B,
+	0x714C,
+	0x714D,
+	0x714E,
+	0x714F,
+	0x7151,
+	0x7152,
+	0x7153,
+	0x715E,
+	0x715F,
+	0x7180,
+	0x7181,
+	0x7183,
+	0x7186,
+	0x7187,
+	0x7188,
+	0x718A,
+	0x718B,
+	0x718C,
+	0x718D,
+	0x718F,
+	0x7193,
+	0x7196,
+	0x719B,
+	0x719F,
+	0x71C0,
+	0x71C1,
+	0x71C2,
+	0x71C3,
+	0x71C4,
+	0x71C5,
+	0x71C6,
+	0x71C7,
+	0x71CD,
+	0x71CE,
+	0x71D2,
+	0x71D4,
+	0x71D5,
+	0x71D6,
+	0x71DA,
+	0x71DE,
+	0x7200,
+	0x7210,
+	0x7211,
+	0x7240,
+	0x7243,
+	0x7244,
+	0x7245,
+	0x7246,
+	0x7247,
+	0x7248,
+	0x7249,
+	0x724A,
+	0x724B,
+	0x724C,
+	0x724D,
+	0x724E,
+	0x724F,
+	0x7280,
+	0x7281,
+	0x7283,
+	0x7284,
+	0x7287,
+	0x7288,
+	0x7289,
+	0x728B,
+	0x728C,
+	0x7290,
+	0x7291,
+	0x7293,
+	0x7297,
+	0x7834,
+	0x7835,
+	0x791e,
+	0x791f,
+	0x793f,
+	0x7941,
+	0x7942,
+	0x796c,
+	0x796d,
+	0x796e,
+	0x796f,
+	0x9400,
+	0x9401,
+	0x9402,
+	0x9403,
+	0x9405,
+	0x940A,
+	0x940B,
+	0x940F,
+	0x94A0,
+	0x94A1,
+	0x94A3,
+	0x94B1,
+	0x94B3,
+	0x94B4,
+	0x94B5,
+	0x94B9,
+	0x9440,
+	0x9441,
+	0x9442,
+	0x9443,
+	0x9444,
+	0x9446,
+	0x944A,
+	0x944B,
+	0x944C,
+	0x944E,
+	0x9450,
+	0x9452,
+	0x9456,
+	0x945A,
+	0x945B,
+	0x945E,
+	0x9460,
+	0x9462,
+	0x946A,
+	0x946B,
+	0x947A,
+	0x947B,
+	0x9480,
+	0x9487,
+	0x9488,
+	0x9489,
+	0x948A,
+	0x948F,
+	0x9490,
+	0x9491,
+	0x9495,
+	0x9498,
+	0x949C,
+	0x949E,
+	0x949F,
+	0x94C0,
+	0x94C1,
+	0x94C3,
+	0x94C4,
+	0x94C5,
+	0x94C6,
+	0x94C7,
+	0x94C8,
+	0x94C9,
+	0x94CB,
+	0x94CC,
+	0x94CD,
+	0x9500,
+	0x9501,
+	0x9504,
+	0x9505,
+	0x9506,
+	0x9507,
+	0x9508,
+	0x9509,
+	0x950F,
+	0x9511,
+	0x9515,
+	0x9517,
+	0x9519,
+	0x9540,
+	0x9541,
+	0x9542,
+	0x954E,
+	0x954F,
+	0x9552,
+	0x9553,
+	0x9555,
+	0x9557,
+	0x955f,
+	0x9580,
+	0x9581,
+	0x9583,
+	0x9586,
+	0x9587,
+	0x9588,
+	0x9589,
+	0x958A,
+	0x958B,
+	0x958C,
+	0x958D,
+	0x958E,
+	0x958F,
+	0x9590,
+	0x9591,
+	0x9593,
+	0x9595,
+	0x9596,
+	0x9597,
+	0x9598,
+	0x9599,
+	0x959B,
+	0x95C0,
+	0x95C2,
+	0x95C4,
+	0x95C5,
+	0x95C6,
+	0x95C7,
+	0x95C9,
+	0x95CC,
+	0x95CD,
+	0x95CE,
+	0x95CF,
+	0x9610,
+	0x9611,
+	0x9612,
+	0x9613,
+	0x9614,
+	0x9615,
+	0x9616,
+	0x9640,
+	0x9641,
+	0x9642,
+	0x9643,
+	0x9644,
+	0x9645,
+	0x9647,
+	0x9648,
+	0x9649,
+	0x964a,
+	0x964b,
+	0x964c,
+	0x964e,
+	0x964f,
+	0x9710,
+	0x9711,
+	0x9712,
+	0x9713,
+	0x9714,
+	0x9715,
+	0x9802,
+	0x9803,
+	0x9804,
+	0x9805,
+	0x9806,
+	0x9807,
+	0x9808,
+	0x9809,
+	0x980A,
+	0x9900,
+	0x9901,
+	0x9903,
+	0x9904,
+	0x9905,
+	0x9906,
+	0x9907,
+	0x9908,
+	0x9909,
+	0x990A,
+	0x990B,
+	0x990C,
+	0x990D,
+	0x990E,
+	0x990F,
+	0x9910,
+	0x9913,
+	0x9917,
+	0x9918,
+	0x9919,
+	0x9990,
+	0x9991,
+	0x9992,
+	0x9993,
+	0x9994,
+	0x9995,
+	0x9996,
+	0x9997,
+	0x9998,
+	0x9999,
+	0x999A,
+	0x999B,
+	0x999C,
+	0x999D,
+	0x99A0,
+	0x99A2,
+	0x99A4,
+	/* radeon secondary ids */
+	0x3171,
+	0x3e70,
+	0x4164,
+	0x4165,
+	0x4166,
+	0x4168,
+	0x4170,
+	0x4171,
+	0x4172,
+	0x4173,
+	0x496e,
+	0x4a69,
+	0x4a6a,
+	0x4a6b,
+	0x4a70,
+	0x4a74,
+	0x4b69,
+	0x4b6b,
+	0x4b6c,
+	0x4c6e,
+	0x4e64,
+	0x4e65,
+	0x4e66,
+	0x4e67,
+	0x4e68,
+	0x4e69,
+	0x4e6a,
+	0x4e71,
+	0x4f73,
+	0x5569,
+	0x556b,
+	0x556d,
+	0x556f,
+	0x5571,
+	0x5854,
+	0x5874,
+	0x5940,
+	0x5941,
+	0x5b70,
+	0x5b72,
+	0x5b73,
+	0x5b74,
+	0x5b75,
+	0x5d44,
+	0x5d45,
+	0x5d6d,
+	0x5d6f,
+	0x5d72,
+	0x5d77,
+	0x5e6b,
+	0x5e6d,
+	0x7120,
+	0x7124,
+	0x7129,
+	0x712e,
+	0x712f,
+	0x7162,
+	0x7163,
+	0x7166,
+	0x7167,
+	0x7172,
+	0x7173,
+	0x71a0,
+	0x71a1,
+	0x71a3,
+	0x71a7,
+	0x71bb,
+	0x71e0,
+	0x71e1,
+	0x71e2,
+	0x71e6,
+	0x71e7,
+	0x71f2,
+	0x7269,
+	0x726b,
+	0x726e,
+	0x72a0,
+	0x72a8,
+	0x72b1,
+	0x72b3,
+	0x793f,
+};
 
 static const struct pci_device_id pciidlist[] = {
-#ifdef  CONFIG_DRM_AMDGPU_SI
 	{0x1002, 0x6780, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
 	{0x1002, 0x6784, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
 	{0x1002, 0x6788, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
@@ -365,8 +1912,6 @@ static const struct pci_device_id pciidlist[] = {
 	{0x1002, 0x6665, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAINAN|AMD_IS_MOBILITY},
 	{0x1002, 0x6667, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAINAN|AMD_IS_MOBILITY},
 	{0x1002, 0x666F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAINAN|AMD_IS_MOBILITY},
-#endif
-#ifdef CONFIG_DRM_AMDGPU_CIK
 	/* Kaveri */
 	{0x1002, 0x1304, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
 	{0x1002, 0x1305, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
@@ -449,7 +1994,6 @@ static const struct pci_device_id pciidlist[] = {
 	{0x1002, 0x985D, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
 	{0x1002, 0x985E, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
 	{0x1002, 0x985F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
-#endif
 	/* topaz */
 	{0x1002, 0x6900, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TOPAZ},
 	{0x1002, 0x6901, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TOPAZ},
@@ -500,6 +2044,7 @@ static const struct pci_device_id pciidlist[] = {
 	{0x1002, 0x67CA, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},
 	{0x1002, 0x67CC, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},
 	{0x1002, 0x67CF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},
+	{0x1002, 0x6FDF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},
 	/* Polaris12 */
 	{0x1002, 0x6980, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS12},
 	{0x1002, 0x6981, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS12},
@@ -509,53 +2054,286 @@ static const struct pci_device_id pciidlist[] = {
 	{0x1002, 0x6995, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS12},
 	{0x1002, 0x6997, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS12},
 	{0x1002, 0x699F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS12},
+	/* VEGAM */
+	{0x1002, 0x694C, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGAM},
+	{0x1002, 0x694E, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGAM},
+	{0x1002, 0x694F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGAM},
 	/* Vega 10 */
-	{0x1002, 0x6860, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10|AMD_EXP_HW_SUPPORT},
-	{0x1002, 0x6861, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10|AMD_EXP_HW_SUPPORT},
-	{0x1002, 0x6862, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10|AMD_EXP_HW_SUPPORT},
-	{0x1002, 0x6863, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10|AMD_EXP_HW_SUPPORT},
-	{0x1002, 0x6864, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10|AMD_EXP_HW_SUPPORT},
-	{0x1002, 0x6867, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10|AMD_EXP_HW_SUPPORT},
-	{0x1002, 0x6868, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10|AMD_EXP_HW_SUPPORT},
-	{0x1002, 0x686c, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10|AMD_EXP_HW_SUPPORT},
-	{0x1002, 0x687f, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10|AMD_EXP_HW_SUPPORT},
+	{0x1002, 0x6860, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x6861, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x6862, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x6863, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x6864, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x6867, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x6868, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x6869, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x686a, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x686b, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x686c, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x686d, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x686e, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x686f, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x687f, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	/* Vega 12 */
+	{0x1002, 0x69A0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA12},
+	{0x1002, 0x69A1, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA12},
+	{0x1002, 0x69A2, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA12},
+	{0x1002, 0x69A3, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA12},
+	{0x1002, 0x69AF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA12},
+	/* Vega 20 */
+	{0x1002, 0x66A0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA20},
+	{0x1002, 0x66A1, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA20},
+	{0x1002, 0x66A2, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA20},
+	{0x1002, 0x66A3, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA20},
+	{0x1002, 0x66A4, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA20},
+	{0x1002, 0x66A7, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA20},
+	{0x1002, 0x66AF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA20},
 	/* Raven */
-	{0x1002, 0x15dd, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_RAVEN|AMD_IS_APU|AMD_EXP_HW_SUPPORT},
+	{0x1002, 0x15dd, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_RAVEN|AMD_IS_APU},
+	{0x1002, 0x15d8, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_RAVEN|AMD_IS_APU},
+	/* Arcturus */
+	{0x1002, 0x738C, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ARCTURUS},
+	{0x1002, 0x7388, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ARCTURUS},
+	{0x1002, 0x738E, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ARCTURUS},
+	{0x1002, 0x7390, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ARCTURUS},
+	/* Navi10 */
+	{0x1002, 0x7310, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI10},
+	{0x1002, 0x7312, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI10},
+	{0x1002, 0x7318, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI10},
+	{0x1002, 0x7319, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI10},
+	{0x1002, 0x731A, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI10},
+	{0x1002, 0x731B, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI10},
+	{0x1002, 0x731E, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI10},
+	{0x1002, 0x731F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI10},
+	/* Navi14 */
+	{0x1002, 0x7340, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI14},
+	{0x1002, 0x7341, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI14},
+	{0x1002, 0x7347, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI14},
+	{0x1002, 0x734F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI14},
+
+	/* Renoir */
+	{0x1002, 0x15E7, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_RENOIR|AMD_IS_APU},
+	{0x1002, 0x1636, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_RENOIR|AMD_IS_APU},
+	{0x1002, 0x1638, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_RENOIR|AMD_IS_APU},
+	{0x1002, 0x164C, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_RENOIR|AMD_IS_APU},
+
+	/* Navi12 */
+	{0x1002, 0x7360, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI12},
+	{0x1002, 0x7362, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI12},
+
+	/* Sienna_Cichlid */
+	{0x1002, 0x73A0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID},
+	{0x1002, 0x73A1, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID},
+	{0x1002, 0x73A2, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID},
+	{0x1002, 0x73A3, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID},
+	{0x1002, 0x73A5, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID},
+	{0x1002, 0x73A8, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID},
+	{0x1002, 0x73A9, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID},
+	{0x1002, 0x73AB, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID},
+	{0x1002, 0x73AC, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID},
+	{0x1002, 0x73AD, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID},
+	{0x1002, 0x73AE, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID},
+	{0x1002, 0x73AF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID},
+	{0x1002, 0x73BF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID},
+
+	/* Yellow Carp */
+	{0x1002, 0x164D, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_YELLOW_CARP|AMD_IS_APU},
+	{0x1002, 0x1681, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_YELLOW_CARP|AMD_IS_APU},
+
+	/* Navy_Flounder */
+	{0x1002, 0x73C0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVY_FLOUNDER},
+	{0x1002, 0x73C1, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVY_FLOUNDER},
+	{0x1002, 0x73C3, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVY_FLOUNDER},
+	{0x1002, 0x73DA, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVY_FLOUNDER},
+	{0x1002, 0x73DB, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVY_FLOUNDER},
+	{0x1002, 0x73DC, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVY_FLOUNDER},
+	{0x1002, 0x73DD, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVY_FLOUNDER},
+	{0x1002, 0x73DE, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVY_FLOUNDER},
+	{0x1002, 0x73DF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVY_FLOUNDER},
+
+	/* DIMGREY_CAVEFISH */
+	{0x1002, 0x73E0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_DIMGREY_CAVEFISH},
+	{0x1002, 0x73E1, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_DIMGREY_CAVEFISH},
+	{0x1002, 0x73E2, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_DIMGREY_CAVEFISH},
+	{0x1002, 0x73E3, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_DIMGREY_CAVEFISH},
+	{0x1002, 0x73E8, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_DIMGREY_CAVEFISH},
+	{0x1002, 0x73E9, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_DIMGREY_CAVEFISH},
+	{0x1002, 0x73EA, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_DIMGREY_CAVEFISH},
+	{0x1002, 0x73EB, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_DIMGREY_CAVEFISH},
+	{0x1002, 0x73EC, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_DIMGREY_CAVEFISH},
+	{0x1002, 0x73ED, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_DIMGREY_CAVEFISH},
+	{0x1002, 0x73EF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_DIMGREY_CAVEFISH},
+	{0x1002, 0x73FF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_DIMGREY_CAVEFISH},
+
+	/* Aldebaran */
+	{0x1002, 0x7408, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ALDEBARAN},
+	{0x1002, 0x740C, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ALDEBARAN},
+	{0x1002, 0x740F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ALDEBARAN},
+	{0x1002, 0x7410, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ALDEBARAN},
+
+	/* CYAN_SKILLFISH */
+	{0x1002, 0x13DB, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU},
+	{0x1002, 0x13F9, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU},
+	{0x1002, 0x13FA, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU},
+	{0x1002, 0x13FB, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU},
+	{0x1002, 0x13FC, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU},
+	{0x1002, 0x13FE, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU},
+	{0x1002, 0x143F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU},
+
+	/* BEIGE_GOBY */
+	{0x1002, 0x7420, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BEIGE_GOBY},
+	{0x1002, 0x7421, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BEIGE_GOBY},
+	{0x1002, 0x7422, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BEIGE_GOBY},
+	{0x1002, 0x7423, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BEIGE_GOBY},
+	{0x1002, 0x7424, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BEIGE_GOBY},
+	{0x1002, 0x743F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BEIGE_GOBY},
+
+	{ PCI_DEVICE(0x1002, PCI_ANY_ID),
+	  .class = PCI_CLASS_DISPLAY_VGA << 8,
+	  .class_mask = 0xffffff,
+	  .driver_data = CHIP_IP_DISCOVERY },
+
+	{ PCI_DEVICE(0x1002, PCI_ANY_ID),
+	  .class = PCI_CLASS_DISPLAY_OTHER << 8,
+	  .class_mask = 0xffffff,
+	  .driver_data = CHIP_IP_DISCOVERY },
+
+	{ PCI_DEVICE(0x1002, PCI_ANY_ID),
+	  .class = PCI_CLASS_ACCELERATOR_PROCESSING << 8,
+	  .class_mask = 0xffffff,
+	  .driver_data = CHIP_IP_DISCOVERY },
 
 	{0, 0, 0}
 };
 
 MODULE_DEVICE_TABLE(pci, pciidlist);
 
-static struct drm_driver kms_driver;
+static const struct amdgpu_asic_type_quirk asic_type_quirks[] = {
+	/* differentiate between P10 and P11 asics with the same DID */
+	{0x67FF, 0xE3, CHIP_POLARIS10},
+	{0x67FF, 0xE7, CHIP_POLARIS10},
+	{0x67FF, 0xF3, CHIP_POLARIS10},
+	{0x67FF, 0xF7, CHIP_POLARIS10},
+};
+
+static const struct drm_driver amdgpu_kms_driver;
 
-static int amdgpu_kick_out_firmware_fb(struct pci_dev *pdev)
+static void amdgpu_get_secondary_funcs(struct amdgpu_device *adev)
 {
-	struct apertures_struct *ap;
-	bool primary = false;
+	struct pci_dev *p = NULL;
+	int i;
 
-	ap = alloc_apertures(1);
-	if (!ap)
-		return -ENOMEM;
+	/* 0 - GPU
+	 * 1 - audio
+	 * 2 - USB
+	 * 3 - UCSI
+	 */
+	for (i = 1; i < 4; i++) {
+		p = pci_get_domain_bus_and_slot(pci_domain_nr(adev->pdev->bus),
+						adev->pdev->bus->number, i);
+		if (p) {
+			pm_runtime_get_sync(&p->dev);
+			pm_runtime_mark_last_busy(&p->dev);
+			pm_runtime_put_autosuspend(&p->dev);
+			pci_dev_put(p);
+		}
+	}
+}
 
-	ap->ranges[0].base = pci_resource_start(pdev, 0);
-	ap->ranges[0].size = pci_resource_len(pdev, 0);
+static void amdgpu_init_debug_options(struct amdgpu_device *adev)
+{
+	if (amdgpu_debug_mask & AMDGPU_DEBUG_VM) {
+		pr_info("debug: VM handling debug enabled\n");
+		adev->debug_vm = true;
+	}
 
-#ifdef CONFIG_X86
-	primary = pdev->resource[PCI_ROM_RESOURCE].flags & IORESOURCE_ROM_SHADOW;
-#endif
-	drm_fb_helper_remove_conflicting_framebuffers(ap, "amdgpudrmfb", primary);
-	kfree(ap);
+	if (amdgpu_debug_mask & AMDGPU_DEBUG_LARGEBAR) {
+		pr_info("debug: enabled simulating large-bar capability on non-large bar system\n");
+		adev->debug_largebar = true;
+	}
 
-	return 0;
+	if (amdgpu_debug_mask & AMDGPU_DEBUG_DISABLE_GPU_SOFT_RECOVERY) {
+		pr_info("debug: soft reset for GPU recovery disabled\n");
+		adev->debug_disable_soft_recovery = true;
+	}
+
+	if (amdgpu_debug_mask & AMDGPU_DEBUG_USE_VRAM_FW_BUF) {
+		pr_info("debug: place fw in vram for frontdoor loading\n");
+		adev->debug_use_vram_fw_buf = true;
+	}
+
+	if (amdgpu_debug_mask & AMDGPU_DEBUG_ENABLE_RAS_ACA) {
+		pr_info("debug: enable RAS ACA\n");
+		adev->debug_enable_ras_aca = true;
+	}
+
+	if (amdgpu_debug_mask & AMDGPU_DEBUG_ENABLE_EXP_RESETS) {
+		pr_info("debug: enable experimental reset features\n");
+		adev->debug_exp_resets = true;
+	}
+
+	if (amdgpu_debug_mask & AMDGPU_DEBUG_DISABLE_GPU_RING_RESET) {
+		pr_info("debug: ring reset disabled\n");
+		adev->debug_disable_gpu_ring_reset = true;
+	}
+	if (amdgpu_debug_mask & AMDGPU_DEBUG_SMU_POOL) {
+		pr_info("debug: use vram for smu pool\n");
+		adev->pm.smu_debug_mask |= SMU_DEBUG_POOL_USE_VRAM;
+	}
+	if (amdgpu_debug_mask & AMDGPU_DEBUG_VM_USERPTR) {
+		pr_info("debug: VM mode debug for userptr is enabled\n");
+		adev->debug_vm_userptr = true;
+	}
+
+	if (amdgpu_debug_mask & AMDGPU_DEBUG_DISABLE_RAS_CE_LOG) {
+		pr_info("debug: disable kernel logs of correctable errors\n");
+		adev->debug_disable_ce_logs = true;
+	}
+}
+
+static unsigned long amdgpu_fix_asic_type(struct pci_dev *pdev, unsigned long flags)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(asic_type_quirks); i++) {
+		if (pdev->device == asic_type_quirks[i].device &&
+			pdev->revision == asic_type_quirks[i].revision) {
+				flags &= ~AMD_ASIC_MASK;
+				flags |= asic_type_quirks[i].type;
+				break;
+			}
+	}
+
+	return flags;
 }
 
 static int amdgpu_pci_probe(struct pci_dev *pdev,
 			    const struct pci_device_id *ent)
 {
-	struct drm_device *dev;
+	struct drm_device *ddev;
+	struct amdgpu_device *adev;
 	unsigned long flags = ent->driver_data;
-	int ret;
+	int ret, retry = 0, i;
+	bool supports_atomic = false;
+
+	if ((pdev->class >> 8) == PCI_CLASS_DISPLAY_VGA ||
+	    (pdev->class >> 8) == PCI_CLASS_DISPLAY_OTHER) {
+		if (drm_firmware_drivers_only() && amdgpu_modeset == -1)
+			return -EINVAL;
+	}
+
+	/* skip devices which are owned by radeon */
+	for (i = 0; i < ARRAY_SIZE(amdgpu_unsupported_pciidlist); i++) {
+		if (amdgpu_unsupported_pciidlist[i] == pdev->device)
+			return -ENODEV;
+	}
+
+	if (amdgpu_aspm == -1 && !pcie_aspm_enabled(pdev))
+		amdgpu_aspm = 0;
+
+	if (amdgpu_virtual_display ||
+	    amdgpu_device_asic_has_dc_support(pdev, flags & AMD_ASIC_MASK))
+		supports_atomic = true;
 
 	if ((flags & AMD_EXP_HW_SUPPORT) && !amdgpu_exp_hw_support) {
 		DRM_INFO("This hardware requires experimental hardware support.\n"
@@ -563,41 +2341,173 @@ static int amdgpu_pci_probe(struct pci_dev *pdev,
 		return -ENODEV;
 	}
 
-	/*
-	 * Initialize amdkfd before starting radeon. If it was not loaded yet,
-	 * defer radeon probing
+	flags = amdgpu_fix_asic_type(pdev, flags);
+
+	/* Due to hardware bugs, S/G Display on raven requires a 1:1 IOMMU mapping,
+	 * however, SME requires an indirect IOMMU mapping because the encryption
+	 * bit is beyond the DMA mask of the chip.
 	 */
-	ret = amdgpu_amdkfd_init();
-	if (ret == -EPROBE_DEFER)
-		return ret;
+	if (cc_platform_has(CC_ATTR_MEM_ENCRYPT) &&
+	    ((flags & AMD_ASIC_MASK) == CHIP_RAVEN)) {
+		dev_info(&pdev->dev,
+			 "SME is not compatible with RAVEN\n");
+		return -ENOTSUPP;
+	}
 
-	/* Get rid of things like offb */
-	ret = amdgpu_kick_out_firmware_fb(pdev);
+	switch (flags & AMD_ASIC_MASK) {
+	case CHIP_TAHITI:
+	case CHIP_PITCAIRN:
+	case CHIP_VERDE:
+	case CHIP_OLAND:
+	case CHIP_HAINAN:
+#ifdef CONFIG_DRM_AMDGPU_SI
+		if (!amdgpu_si_support) {
+			dev_info(&pdev->dev,
+				 "SI support provided by radeon.\n");
+			dev_info(&pdev->dev,
+				 "Use radeon.si_support=0 amdgpu.si_support=1 to override.\n"
+				);
+			return -ENODEV;
+		}
+		break;
+#else
+		dev_info(&pdev->dev, "amdgpu is built without SI support.\n");
+		return -ENODEV;
+#endif
+	case CHIP_KAVERI:
+	case CHIP_BONAIRE:
+	case CHIP_HAWAII:
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+#ifdef CONFIG_DRM_AMDGPU_CIK
+		if (!amdgpu_cik_support) {
+			dev_info(&pdev->dev,
+				 "CIK support provided by radeon.\n");
+			dev_info(&pdev->dev,
+				 "Use radeon.cik_support=0 amdgpu.cik_support=1 to override.\n"
+				);
+			return -ENODEV;
+		}
+		break;
+#else
+		dev_info(&pdev->dev, "amdgpu is built without CIK support.\n");
+		return -ENODEV;
+#endif
+	default:
+		break;
+	}
+
+	adev = devm_drm_dev_alloc(&pdev->dev, &amdgpu_kms_driver, typeof(*adev), ddev);
+	if (IS_ERR(adev))
+		return PTR_ERR(adev);
+
+	adev->dev  = &pdev->dev;
+	adev->pdev = pdev;
+	ddev = adev_to_drm(adev);
+
+	if (!supports_atomic)
+		ddev->driver_features &= ~DRIVER_ATOMIC;
+
+	ret = pci_enable_device(pdev);
 	if (ret)
 		return ret;
 
-	dev = drm_dev_alloc(&kms_driver, &pdev->dev);
-	if (IS_ERR(dev))
-		return PTR_ERR(dev);
+	pci_set_drvdata(pdev, ddev);
 
-	ret = pci_enable_device(pdev);
+	amdgpu_init_debug_options(adev);
+
+	ret = amdgpu_driver_load_kms(adev, flags);
 	if (ret)
-		goto err_free;
+		goto err_pci;
 
-	dev->pdev = pdev;
+retry_init:
+	ret = drm_dev_register(ddev, flags);
+	if (ret == -EAGAIN && ++retry <= 3) {
+		DRM_INFO("retry init %d\n", retry);
+		/* Don't request EX mode too frequently which is attacking */
+		msleep(5000);
+		goto retry_init;
+	} else if (ret) {
+		goto err_pci;
+	}
 
-	pci_set_drvdata(pdev, dev);
+	ret = amdgpu_xcp_dev_register(adev, ent);
+	if (ret)
+		goto err_pci;
 
-	ret = drm_dev_register(dev, ent->driver_data);
+	ret = amdgpu_amdkfd_drm_client_create(adev);
 	if (ret)
 		goto err_pci;
 
+	/*
+	 * 1. don't init fbdev on hw without DCE
+	 * 2. don't init fbdev if there are no connectors
+	 */
+	if (adev->mode_info.mode_config_initialized &&
+	    !list_empty(&adev_to_drm(adev)->mode_config.connector_list)) {
+		const struct drm_format_info *format;
+
+		/* select 8 bpp console on low vram cards */
+		if (adev->gmc.real_vram_size <= (32*1024*1024))
+			format = drm_format_info(DRM_FORMAT_C8);
+		else
+			format = NULL;
+
+		drm_client_setup(adev_to_drm(adev), format);
+	}
+
+	ret = amdgpu_debugfs_init(adev);
+	if (ret)
+		DRM_ERROR("Creating debugfs files failed (%d).\n", ret);
+
+	if (adev->pm.rpm_mode != AMDGPU_RUNPM_NONE) {
+		/* only need to skip on ATPX */
+		if (amdgpu_device_supports_px(adev))
+			dev_pm_set_driver_flags(ddev->dev, DPM_FLAG_NO_DIRECT_COMPLETE);
+		/* we want direct complete for BOCO */
+		if (amdgpu_device_supports_boco(adev))
+			dev_pm_set_driver_flags(ddev->dev, DPM_FLAG_SMART_PREPARE |
+						DPM_FLAG_SMART_SUSPEND |
+						DPM_FLAG_MAY_SKIP_RESUME);
+		pm_runtime_use_autosuspend(ddev->dev);
+		pm_runtime_set_autosuspend_delay(ddev->dev, 5000);
+
+		pm_runtime_allow(ddev->dev);
+
+		pm_runtime_mark_last_busy(ddev->dev);
+		pm_runtime_put_autosuspend(ddev->dev);
+
+		pci_wake_from_d3(pdev, TRUE);
+
+		/*
+		 * For runpm implemented via BACO, PMFW will handle the
+		 * timing for BACO in and out:
+		 *   - put ASIC into BACO state only when both video and
+		 *     audio functions are in D3 state.
+		 *   - pull ASIC out of BACO state when either video or
+		 *     audio function is in D0 state.
+		 * Also, at startup, PMFW assumes both functions are in
+		 * D0 state.
+		 *
+		 * So if snd driver was loaded prior to amdgpu driver
+		 * and audio function was put into D3 state, there will
+		 * be no PMFW-aware D-state transition(D0->D3) on runpm
+		 * suspend. Thus the BACO will be not correctly kicked in.
+		 *
+		 * Via amdgpu_get_secondary_funcs(), the audio dev is put
+		 * into D0 state. Then there will be a PMFW-aware D-state
+		 * transition(D0->D3) on runpm suspend.
+		 */
+		if (amdgpu_device_supports_baco(adev) &&
+		    !(adev->flags & AMD_IS_APU) &&
+		    adev->asic_type >= CHIP_NAVI10)
+			amdgpu_get_secondary_funcs(adev);
+	}
+
 	return 0;
 
 err_pci:
 	pci_disable_device(pdev);
-err_free:
-	drm_dev_unref(dev);
 	return ret;
 }
 
@@ -605,104 +2515,355 @@ static void
 amdgpu_pci_remove(struct pci_dev *pdev)
 {
 	struct drm_device *dev = pci_get_drvdata(pdev);
+	struct amdgpu_device *adev = drm_to_adev(dev);
+
+	amdgpu_ras_eeprom_check_and_recover(adev);
+	amdgpu_xcp_dev_unplug(adev);
+	amdgpu_gmc_prepare_nps_mode_change(adev);
+	drm_dev_unplug(dev);
+
+	if (adev->pm.rpm_mode != AMDGPU_RUNPM_NONE) {
+		pm_runtime_get_sync(dev->dev);
+		pm_runtime_forbid(dev->dev);
+	}
 
-	drm_dev_unregister(dev);
-	drm_dev_unref(dev);
+	amdgpu_driver_unload_kms(dev);
+
+	/*
+	 * Flush any in flight DMA operations from device.
+	 * Clear the Bus Master Enable bit and then wait on the PCIe Device
+	 * StatusTransactions Pending bit.
+	 */
+	pci_disable_device(pdev);
+	pci_wait_for_pending_transaction(pdev);
 }
 
 static void
 amdgpu_pci_shutdown(struct pci_dev *pdev)
 {
 	struct drm_device *dev = pci_get_drvdata(pdev);
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+
+	if (amdgpu_ras_intr_triggered())
+		return;
+
+	/* device maybe not resumed here, return immediately in this case */
+	if (adev->in_s4 && adev->in_suspend)
+		return;
 
 	/* if we are running in a VM, make sure the device
 	 * torn down properly on reboot/shutdown.
 	 * unfortunately we can't detect certain
 	 * hypervisors so just do this all the time.
 	 */
-	amdgpu_suspend(adev);
+	if (!amdgpu_passthrough(adev))
+		adev->mp1_state = PP_MP1_STATE_UNLOAD;
+	amdgpu_device_ip_suspend(adev);
+	adev->mp1_state = PP_MP1_STATE_NONE;
+}
+
+static int amdgpu_pmops_prepare(struct device *dev)
+{
+	struct drm_device *drm_dev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(drm_dev);
+
+	/* device maybe not resumed here, return immediately in this case */
+	if (adev->in_s4 && adev->in_suspend)
+		return 0;
+
+	/* Return a positive number here so
+	 * DPM_FLAG_SMART_SUSPEND works properly
+	 */
+	if (amdgpu_device_supports_boco(adev) && pm_runtime_suspended(dev))
+		return 1;
+
+	/* if we will not support s3 or s2i for the device
+	 *  then skip suspend
+	 */
+	if (!amdgpu_acpi_is_s0ix_active(adev) &&
+	    !amdgpu_acpi_is_s3_active(adev))
+		return 1;
+
+	return amdgpu_device_prepare(drm_dev);
+}
+
+static void amdgpu_pmops_complete(struct device *dev)
+{
+	amdgpu_device_complete(dev_get_drvdata(dev));
 }
 
 static int amdgpu_pmops_suspend(struct device *dev)
 {
-	struct pci_dev *pdev = to_pci_dev(dev);
+	struct drm_device *drm_dev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(drm_dev);
+
+	if (amdgpu_acpi_is_s0ix_active(adev))
+		adev->in_s0ix = true;
+	else if (amdgpu_acpi_is_s3_active(adev))
+		adev->in_s3 = true;
+	if (!adev->in_s0ix && !adev->in_s3) {
+#if IS_ENABLED(CONFIG_SUSPEND)
+		/* don't allow going deep first time followed by s2idle the next time */
+		if (adev->last_suspend_state != PM_SUSPEND_ON &&
+		    adev->last_suspend_state != pm_suspend_target_state) {
+			drm_err_once(drm_dev, "Unsupported suspend state %d\n",
+				     pm_suspend_target_state);
+			return -EINVAL;
+		}
+#endif
+		return 0;
+	}
 
-	struct drm_device *drm_dev = pci_get_drvdata(pdev);
-	return amdgpu_device_suspend(drm_dev, true, true);
+#if IS_ENABLED(CONFIG_SUSPEND)
+	/* cache the state last used for suspend */
+	adev->last_suspend_state = pm_suspend_target_state;
+#endif
+
+	return amdgpu_device_suspend(drm_dev, true);
+}
+
+static int amdgpu_pmops_suspend_noirq(struct device *dev)
+{
+	struct drm_device *drm_dev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(drm_dev);
+
+	if (amdgpu_acpi_should_gpu_reset(adev))
+		return amdgpu_asic_reset(adev);
+
+	return 0;
 }
 
 static int amdgpu_pmops_resume(struct device *dev)
 {
-	struct pci_dev *pdev = to_pci_dev(dev);
-	struct drm_device *drm_dev = pci_get_drvdata(pdev);
+	struct drm_device *drm_dev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(drm_dev);
+	int r;
 
-	/* GPU comes up enabled by the bios on resume */
-	if (amdgpu_device_is_px(drm_dev)) {
-		pm_runtime_disable(dev);
-		pm_runtime_set_active(dev);
-		pm_runtime_enable(dev);
-	}
+	if (!adev->in_s0ix && !adev->in_s3)
+		return 0;
 
-	return amdgpu_device_resume(drm_dev, true, true);
+	/* Avoids registers access if device is physically gone */
+	if (!pci_device_is_present(adev->pdev))
+		adev->no_hw_access = true;
+
+	r = amdgpu_device_resume(drm_dev, true);
+	if (amdgpu_acpi_is_s0ix_active(adev))
+		adev->in_s0ix = false;
+	else
+		adev->in_s3 = false;
+	return r;
 }
 
 static int amdgpu_pmops_freeze(struct device *dev)
 {
-	struct pci_dev *pdev = to_pci_dev(dev);
+	struct drm_device *drm_dev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(drm_dev);
+	int r;
 
-	struct drm_device *drm_dev = pci_get_drvdata(pdev);
-	return amdgpu_device_suspend(drm_dev, false, true);
+	r = amdgpu_device_suspend(drm_dev, true);
+	if (r)
+		return r;
+
+	if (amdgpu_acpi_should_gpu_reset(adev))
+		return amdgpu_asic_reset(adev);
+	return 0;
 }
 
 static int amdgpu_pmops_thaw(struct device *dev)
 {
-	struct pci_dev *pdev = to_pci_dev(dev);
+	struct drm_device *drm_dev = dev_get_drvdata(dev);
 
-	struct drm_device *drm_dev = pci_get_drvdata(pdev);
-	return amdgpu_device_resume(drm_dev, false, true);
+	/* do not resume device if it's normal hibernation */
+	if (!pm_hibernate_is_recovering() && !pm_hibernation_mode_is_suspend())
+		return 0;
+
+	return amdgpu_device_resume(drm_dev, true);
 }
 
 static int amdgpu_pmops_poweroff(struct device *dev)
 {
-	struct pci_dev *pdev = to_pci_dev(dev);
+	struct drm_device *drm_dev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(drm_dev);
 
-	struct drm_device *drm_dev = pci_get_drvdata(pdev);
-	return amdgpu_device_suspend(drm_dev, true, true);
+	/* device maybe not resumed here, return immediately in this case */
+	if (adev->in_s4 && adev->in_suspend)
+		return 0;
+
+	return amdgpu_device_suspend(drm_dev, true);
 }
 
 static int amdgpu_pmops_restore(struct device *dev)
 {
+	struct drm_device *drm_dev = dev_get_drvdata(dev);
+
+	return amdgpu_device_resume(drm_dev, true);
+}
+
+static int amdgpu_runtime_idle_check_display(struct device *dev)
+{
 	struct pci_dev *pdev = to_pci_dev(dev);
+	struct drm_device *drm_dev = pci_get_drvdata(pdev);
+	struct amdgpu_device *adev = drm_to_adev(drm_dev);
+
+	if (adev->mode_info.num_crtc) {
+		struct drm_connector *list_connector;
+		struct drm_connector_list_iter iter;
+		int ret = 0;
+
+		if (amdgpu_runtime_pm != -2) {
+			/* XXX: Return busy if any displays are connected to avoid
+			 * possible display wakeups after runtime resume due to
+			 * hotplug events in case any displays were connected while
+			 * the GPU was in suspend.  Remove this once that is fixed.
+			 */
+			mutex_lock(&drm_dev->mode_config.mutex);
+			drm_connector_list_iter_begin(drm_dev, &iter);
+			drm_for_each_connector_iter(list_connector, &iter) {
+				if (list_connector->status == connector_status_connected) {
+					ret = -EBUSY;
+					break;
+				}
+			}
+			drm_connector_list_iter_end(&iter);
+			mutex_unlock(&drm_dev->mode_config.mutex);
+
+			if (ret)
+				return ret;
+		}
 
+		if (adev->dc_enabled) {
+			struct drm_crtc *crtc;
+
+			drm_for_each_crtc(crtc, drm_dev) {
+				drm_modeset_lock(&crtc->mutex, NULL);
+				if (crtc->state->active)
+					ret = -EBUSY;
+				drm_modeset_unlock(&crtc->mutex);
+				if (ret < 0)
+					break;
+			}
+		} else {
+			mutex_lock(&drm_dev->mode_config.mutex);
+			drm_modeset_lock(&drm_dev->mode_config.connection_mutex, NULL);
+
+			drm_connector_list_iter_begin(drm_dev, &iter);
+			drm_for_each_connector_iter(list_connector, &iter) {
+				if (list_connector->dpms ==  DRM_MODE_DPMS_ON) {
+					ret = -EBUSY;
+					break;
+				}
+			}
+
+			drm_connector_list_iter_end(&iter);
+
+			drm_modeset_unlock(&drm_dev->mode_config.connection_mutex);
+			mutex_unlock(&drm_dev->mode_config.mutex);
+		}
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int amdgpu_runtime_idle_check_userq(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
 	struct drm_device *drm_dev = pci_get_drvdata(pdev);
-	return amdgpu_device_resume(drm_dev, false, true);
+	struct amdgpu_device *adev = drm_to_adev(drm_dev);
+	struct amdgpu_usermode_queue *queue;
+	struct amdgpu_userq_mgr *uqm, *tmp;
+	int queue_id;
+	int ret = 0;
+
+	mutex_lock(&adev->userq_mutex);
+	list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) {
+		idr_for_each_entry(&uqm->userq_idr, queue, queue_id) {
+			ret = -EBUSY;
+			goto done;
+		}
+	}
+done:
+	mutex_unlock(&adev->userq_mutex);
+
+	return ret;
 }
 
 static int amdgpu_pmops_runtime_suspend(struct device *dev)
 {
 	struct pci_dev *pdev = to_pci_dev(dev);
 	struct drm_device *drm_dev = pci_get_drvdata(pdev);
-	int ret;
+	struct amdgpu_device *adev = drm_to_adev(drm_dev);
+	int ret, i;
 
-	if (!amdgpu_device_is_px(drm_dev)) {
+	if (adev->pm.rpm_mode == AMDGPU_RUNPM_NONE) {
 		pm_runtime_forbid(dev);
 		return -EBUSY;
 	}
 
-	drm_dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
-	drm_kms_helper_poll_disable(drm_dev);
-	vga_switcheroo_set_dynamic_switch(pdev, VGA_SWITCHEROO_OFF);
+	ret = amdgpu_runtime_idle_check_display(dev);
+	if (ret)
+		return ret;
+	ret = amdgpu_runtime_idle_check_userq(dev);
+	if (ret)
+		return ret;
 
-	ret = amdgpu_device_suspend(drm_dev, false, false);
-	pci_save_state(pdev);
-	pci_disable_device(pdev);
-	pci_ignore_hotplug(pdev);
-	if (amdgpu_is_atpx_hybrid())
+	/* wait for all rings to drain before suspending */
+	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
+		struct amdgpu_ring *ring = adev->rings[i];
+
+		if (ring && ring->sched.ready) {
+			ret = amdgpu_fence_wait_empty(ring);
+			if (ret)
+				return -EBUSY;
+		}
+	}
+
+	adev->in_runpm = true;
+	if (adev->pm.rpm_mode == AMDGPU_RUNPM_PX)
+		drm_dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
+
+	/*
+	 * By setting mp1_state as PP_MP1_STATE_UNLOAD, MP1 will do some
+	 * proper cleanups and put itself into a state ready for PNP. That
+	 * can address some random resuming failure observed on BOCO capable
+	 * platforms.
+	 * TODO: this may be also needed for PX capable platform.
+	 */
+	if (adev->pm.rpm_mode == AMDGPU_RUNPM_BOCO)
+		adev->mp1_state = PP_MP1_STATE_UNLOAD;
+
+	ret = amdgpu_device_prepare(drm_dev);
+	if (ret)
+		return ret;
+	ret = amdgpu_device_suspend(drm_dev, false);
+	if (ret) {
+		adev->in_runpm = false;
+		if (adev->pm.rpm_mode == AMDGPU_RUNPM_BOCO)
+			adev->mp1_state = PP_MP1_STATE_NONE;
+		return ret;
+	}
+
+	if (adev->pm.rpm_mode == AMDGPU_RUNPM_BOCO)
+		adev->mp1_state = PP_MP1_STATE_NONE;
+
+	if (adev->pm.rpm_mode == AMDGPU_RUNPM_PX) {
+		/* Only need to handle PCI state in the driver for ATPX
+		 * PCI core handles it for _PR3.
+		 */
+		amdgpu_device_cache_pci_state(pdev);
+		pci_disable_device(pdev);
+		pci_ignore_hotplug(pdev);
 		pci_set_power_state(pdev, PCI_D3cold);
-	else if (!amdgpu_has_atpx_dgpu_power_cntl())
-		pci_set_power_state(pdev, PCI_D3hot);
-	drm_dev->switch_power_state = DRM_SWITCH_POWER_DYNAMIC_OFF;
+		drm_dev->switch_power_state = DRM_SWITCH_POWER_DYNAMIC_OFF;
+	} else if (adev->pm.rpm_mode == AMDGPU_RUNPM_BOCO) {
+		/* nothing to do */
+	} else if ((adev->pm.rpm_mode == AMDGPU_RUNPM_BACO) ||
+			(adev->pm.rpm_mode == AMDGPU_RUNPM_BAMACO)) {
+		amdgpu_device_baco_enter(adev);
+	}
+
+	dev_dbg(&pdev->dev, "asic/device is runtime suspended\n");
 
 	return 0;
 }
@@ -711,51 +2872,87 @@ static int amdgpu_pmops_runtime_resume(struct device *dev)
 {
 	struct pci_dev *pdev = to_pci_dev(dev);
 	struct drm_device *drm_dev = pci_get_drvdata(pdev);
+	struct amdgpu_device *adev = drm_to_adev(drm_dev);
 	int ret;
 
-	if (!amdgpu_device_is_px(drm_dev))
+	if (adev->pm.rpm_mode == AMDGPU_RUNPM_NONE)
 		return -EINVAL;
 
-	drm_dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
+	/* Avoids registers access if device is physically gone */
+	if (!pci_device_is_present(adev->pdev))
+		adev->no_hw_access = true;
 
-	if (amdgpu_is_atpx_hybrid() ||
-	    !amdgpu_has_atpx_dgpu_power_cntl())
+	if (adev->pm.rpm_mode == AMDGPU_RUNPM_PX) {
+		drm_dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
+
+		/* Only need to handle PCI state in the driver for ATPX
+		 * PCI core handles it for _PR3.
+		 */
 		pci_set_power_state(pdev, PCI_D0);
-	pci_restore_state(pdev);
-	ret = pci_enable_device(pdev);
-	if (ret)
+		amdgpu_device_load_pci_state(pdev);
+		ret = pci_enable_device(pdev);
+		if (ret)
+			return ret;
+		pci_set_master(pdev);
+	} else if (adev->pm.rpm_mode == AMDGPU_RUNPM_BOCO) {
+		/* Only need to handle PCI state in the driver for ATPX
+		 * PCI core handles it for _PR3.
+		 */
+		pci_set_master(pdev);
+	} else if ((adev->pm.rpm_mode == AMDGPU_RUNPM_BACO) ||
+			(adev->pm.rpm_mode == AMDGPU_RUNPM_BAMACO)) {
+		amdgpu_device_baco_exit(adev);
+	}
+	ret = amdgpu_device_resume(drm_dev, false);
+	if (ret) {
+		if (adev->pm.rpm_mode == AMDGPU_RUNPM_PX)
+			pci_disable_device(pdev);
 		return ret;
-	pci_set_master(pdev);
+	}
 
-	ret = amdgpu_device_resume(drm_dev, false, false);
-	drm_kms_helper_poll_enable(drm_dev);
-	vga_switcheroo_set_dynamic_switch(pdev, VGA_SWITCHEROO_ON);
-	drm_dev->switch_power_state = DRM_SWITCH_POWER_ON;
+	if (adev->pm.rpm_mode == AMDGPU_RUNPM_PX)
+		drm_dev->switch_power_state = DRM_SWITCH_POWER_ON;
+	adev->in_runpm = false;
 	return 0;
 }
 
 static int amdgpu_pmops_runtime_idle(struct device *dev)
 {
-	struct pci_dev *pdev = to_pci_dev(dev);
-	struct drm_device *drm_dev = pci_get_drvdata(pdev);
-	struct drm_crtc *crtc;
+	struct drm_device *drm_dev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(drm_dev);
+	int ret;
 
-	if (!amdgpu_device_is_px(drm_dev)) {
+	if (adev->pm.rpm_mode == AMDGPU_RUNPM_NONE) {
 		pm_runtime_forbid(dev);
 		return -EBUSY;
 	}
 
-	list_for_each_entry(crtc, &drm_dev->mode_config.crtc_list, head) {
-		if (crtc->enabled) {
-			DRM_DEBUG_DRIVER("failing to power off - crtc active\n");
-			return -EBUSY;
-		}
-	}
+	ret = amdgpu_runtime_idle_check_display(dev);
+	if (ret)
+		goto done;
 
+	ret = amdgpu_runtime_idle_check_userq(dev);
+done:
 	pm_runtime_mark_last_busy(dev);
 	pm_runtime_autosuspend(dev);
-	/* we don't want the main rpm_idle to call suspend - we want to autosuspend */
-	return 1;
+	return ret;
+}
+
+static int amdgpu_drm_release(struct inode *inode, struct file *filp)
+{
+	struct drm_file *file_priv = filp->private_data;
+	struct amdgpu_fpriv *fpriv = file_priv->driver_priv;
+	struct drm_device *dev = file_priv->minor->dev;
+	int idx;
+
+	if (fpriv && drm_dev_enter(dev, &idx)) {
+		fpriv->evf_mgr.fd_closing = true;
+		amdgpu_eviction_fence_destroy(&fpriv->evf_mgr);
+		amdgpu_userq_mgr_fini(&fpriv->userq_mgr);
+		drm_dev_exit(idx);
+	}
+
+	return drm_release(inode, filp);
 }
 
 long amdgpu_drm_ioctl(struct file *filp,
@@ -764,105 +2961,168 @@ long amdgpu_drm_ioctl(struct file *filp,
 	struct drm_file *file_priv = filp->private_data;
 	struct drm_device *dev;
 	long ret;
+
 	dev = file_priv->minor->dev;
 	ret = pm_runtime_get_sync(dev->dev);
 	if (ret < 0)
-		return ret;
+		goto out;
 
 	ret = drm_ioctl(filp, cmd, arg);
 
 	pm_runtime_mark_last_busy(dev->dev);
+out:
 	pm_runtime_put_autosuspend(dev->dev);
 	return ret;
 }
 
 static const struct dev_pm_ops amdgpu_pm_ops = {
-	.suspend = amdgpu_pmops_suspend,
-	.resume = amdgpu_pmops_resume,
-	.freeze = amdgpu_pmops_freeze,
-	.thaw = amdgpu_pmops_thaw,
-	.poweroff = amdgpu_pmops_poweroff,
-	.restore = amdgpu_pmops_restore,
+	.prepare = pm_sleep_ptr(amdgpu_pmops_prepare),
+	.complete = pm_sleep_ptr(amdgpu_pmops_complete),
+	.suspend = pm_sleep_ptr(amdgpu_pmops_suspend),
+	.suspend_noirq = pm_sleep_ptr(amdgpu_pmops_suspend_noirq),
+	.resume = pm_sleep_ptr(amdgpu_pmops_resume),
+	.freeze = pm_sleep_ptr(amdgpu_pmops_freeze),
+	.thaw = pm_sleep_ptr(amdgpu_pmops_thaw),
+	.poweroff = pm_sleep_ptr(amdgpu_pmops_poweroff),
+	.restore = pm_sleep_ptr(amdgpu_pmops_restore),
 	.runtime_suspend = amdgpu_pmops_runtime_suspend,
 	.runtime_resume = amdgpu_pmops_runtime_resume,
 	.runtime_idle = amdgpu_pmops_runtime_idle,
 };
 
+static int amdgpu_flush(struct file *f, fl_owner_t id)
+{
+	struct drm_file *file_priv = f->private_data;
+	struct amdgpu_fpriv *fpriv = file_priv->driver_priv;
+	long timeout = MAX_WAIT_SCHED_ENTITY_Q_EMPTY;
+
+	timeout = amdgpu_ctx_mgr_entity_flush(&fpriv->ctx_mgr, timeout);
+	timeout = amdgpu_vm_wait_idle(&fpriv->vm, timeout);
+
+	return timeout >= 0 ? 0 : timeout;
+}
+
 static const struct file_operations amdgpu_driver_kms_fops = {
 	.owner = THIS_MODULE,
 	.open = drm_open,
-	.release = drm_release,
+	.flush = amdgpu_flush,
+	.release = amdgpu_drm_release,
 	.unlocked_ioctl = amdgpu_drm_ioctl,
-	.mmap = amdgpu_mmap,
+	.mmap = drm_gem_mmap,
 	.poll = drm_poll,
 	.read = drm_read,
 #ifdef CONFIG_COMPAT
 	.compat_ioctl = amdgpu_kms_compat_ioctl,
 #endif
+#ifdef CONFIG_PROC_FS
+	.show_fdinfo = drm_show_fdinfo,
+#endif
+	.fop_flags = FOP_UNSIGNED_OFFSET,
 };
 
-static bool
-amdgpu_get_crtc_scanout_position(struct drm_device *dev, unsigned int pipe,
-				 bool in_vblank_irq, int *vpos, int *hpos,
-				 ktime_t *stime, ktime_t *etime,
-				 const struct drm_display_mode *mode)
+int amdgpu_file_to_fpriv(struct file *filp, struct amdgpu_fpriv **fpriv)
 {
-	return amdgpu_get_crtc_scanoutpos(dev, pipe, 0, vpos, hpos,
-					  stime, etime, mode);
+	struct drm_file *file;
+
+	if (!filp)
+		return -EINVAL;
+
+	if (filp->f_op != &amdgpu_driver_kms_fops)
+		return -EINVAL;
+
+	file = filp->private_data;
+	*fpriv = file->driver_priv;
+	return 0;
 }
 
-static struct drm_driver kms_driver = {
+const struct drm_ioctl_desc amdgpu_ioctls_kms[] = {
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_CREATE, amdgpu_gem_create_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_CTX, amdgpu_ctx_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_VM, amdgpu_vm_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_SCHED, amdgpu_sched_ioctl, DRM_MASTER),
+	DRM_IOCTL_DEF_DRV(AMDGPU_BO_LIST, amdgpu_bo_list_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_FENCE_TO_HANDLE, amdgpu_cs_fence_to_handle_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	/* KMS */
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_MMAP, amdgpu_gem_mmap_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_WAIT_IDLE, amdgpu_gem_wait_idle_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_CS, amdgpu_cs_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_INFO, amdgpu_info_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_WAIT_CS, amdgpu_cs_wait_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_WAIT_FENCES, amdgpu_cs_wait_fences_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_METADATA, amdgpu_gem_metadata_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_VA, amdgpu_gem_va_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_OP, amdgpu_gem_op_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_USERPTR, amdgpu_gem_userptr_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_USERQ, amdgpu_userq_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_USERQ_SIGNAL, amdgpu_userq_signal_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_USERQ_WAIT, amdgpu_userq_wait_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_LIST_HANDLES, amdgpu_gem_list_handles_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+};
+
+static const struct drm_driver amdgpu_kms_driver = {
 	.driver_features =
-	    DRIVER_USE_AGP |
-	    DRIVER_HAVE_IRQ | DRIVER_IRQ_SHARED | DRIVER_GEM |
-	    DRIVER_PRIME | DRIVER_RENDER | DRIVER_MODESET | DRIVER_SYNCOBJ,
-	.load = amdgpu_driver_load_kms,
+	    DRIVER_ATOMIC |
+	    DRIVER_GEM |
+	    DRIVER_RENDER | DRIVER_MODESET | DRIVER_SYNCOBJ |
+	    DRIVER_SYNCOBJ_TIMELINE,
 	.open = amdgpu_driver_open_kms,
 	.postclose = amdgpu_driver_postclose_kms,
-	.lastclose = amdgpu_driver_lastclose_kms,
-	.unload = amdgpu_driver_unload_kms,
-	.get_vblank_counter = amdgpu_get_vblank_counter_kms,
-	.enable_vblank = amdgpu_enable_vblank_kms,
-	.disable_vblank = amdgpu_disable_vblank_kms,
-	.get_vblank_timestamp = drm_calc_vbltimestamp_from_scanoutpos,
-	.get_scanout_position = amdgpu_get_crtc_scanout_position,
-#if defined(CONFIG_DEBUG_FS)
-	.debugfs_init = amdgpu_debugfs_init,
+	.ioctls = amdgpu_ioctls_kms,
+	.num_ioctls = ARRAY_SIZE(amdgpu_ioctls_kms),
+	.dumb_create = amdgpu_mode_dumb_create,
+	.dumb_map_offset = amdgpu_mode_dumb_mmap,
+	DRM_FBDEV_TTM_DRIVER_OPS,
+	.fops = &amdgpu_driver_kms_fops,
+	.release = &amdgpu_driver_release_kms,
+#ifdef CONFIG_PROC_FS
+	.show_fdinfo = amdgpu_show_fdinfo,
 #endif
-	.irq_preinstall = amdgpu_irq_preinstall,
-	.irq_postinstall = amdgpu_irq_postinstall,
-	.irq_uninstall = amdgpu_irq_uninstall,
-	.irq_handler = amdgpu_irq_handler,
+
+	.gem_prime_import = amdgpu_gem_prime_import,
+
+	.name = DRIVER_NAME,
+	.desc = DRIVER_DESC,
+	.major = KMS_DRIVER_MAJOR,
+	.minor = KMS_DRIVER_MINOR,
+	.patchlevel = KMS_DRIVER_PATCHLEVEL,
+};
+
+const struct drm_driver amdgpu_partition_driver = {
+	.driver_features =
+	    DRIVER_GEM | DRIVER_RENDER | DRIVER_SYNCOBJ |
+	    DRIVER_SYNCOBJ_TIMELINE,
+	.open = amdgpu_driver_open_kms,
+	.postclose = amdgpu_driver_postclose_kms,
 	.ioctls = amdgpu_ioctls_kms,
-	.gem_free_object_unlocked = amdgpu_gem_object_free,
-	.gem_open_object = amdgpu_gem_object_open,
-	.gem_close_object = amdgpu_gem_object_close,
+	.num_ioctls = ARRAY_SIZE(amdgpu_ioctls_kms),
 	.dumb_create = amdgpu_mode_dumb_create,
 	.dumb_map_offset = amdgpu_mode_dumb_mmap,
+	DRM_FBDEV_TTM_DRIVER_OPS,
 	.fops = &amdgpu_driver_kms_fops,
+	.release = &amdgpu_driver_release_kms,
 
-	.prime_handle_to_fd = drm_gem_prime_handle_to_fd,
-	.prime_fd_to_handle = drm_gem_prime_fd_to_handle,
-	.gem_prime_export = amdgpu_gem_prime_export,
-	.gem_prime_import = drm_gem_prime_import,
-	.gem_prime_pin = amdgpu_gem_prime_pin,
-	.gem_prime_unpin = amdgpu_gem_prime_unpin,
-	.gem_prime_res_obj = amdgpu_gem_prime_res_obj,
-	.gem_prime_get_sg_table = amdgpu_gem_prime_get_sg_table,
-	.gem_prime_import_sg_table = amdgpu_gem_prime_import_sg_table,
-	.gem_prime_vmap = amdgpu_gem_prime_vmap,
-	.gem_prime_vunmap = amdgpu_gem_prime_vunmap,
+	.gem_prime_import = amdgpu_gem_prime_import,
 
 	.name = DRIVER_NAME,
 	.desc = DRIVER_DESC,
-	.date = DRIVER_DATE,
 	.major = KMS_DRIVER_MAJOR,
 	.minor = KMS_DRIVER_MINOR,
 	.patchlevel = KMS_DRIVER_PATCHLEVEL,
 };
 
-static struct drm_driver *driver;
-static struct pci_driver *pdriver;
+static struct pci_error_handlers amdgpu_pci_err_handler = {
+	.error_detected	= amdgpu_pci_error_detected,
+	.mmio_enabled	= amdgpu_pci_mmio_enabled,
+	.slot_reset	= amdgpu_pci_slot_reset,
+	.resume		= amdgpu_pci_resume,
+};
+
+static const struct attribute_group *amdgpu_sysfs_groups[] = {
+	&amdgpu_vram_mgr_attr_group,
+	&amdgpu_gtt_mgr_attr_group,
+	&amdgpu_flash_attr_group,
+	NULL,
+};
 
 static struct pci_driver amdgpu_kms_pci_driver = {
 	.name = DRIVER_NAME,
@@ -870,11 +3130,11 @@ static struct pci_driver amdgpu_kms_pci_driver = {
 	.probe = amdgpu_pci_probe,
 	.remove = amdgpu_pci_remove,
 	.shutdown = amdgpu_pci_shutdown,
-	.driver.pm = &amdgpu_pm_ops,
+	.driver.pm = pm_ptr(&amdgpu_pm_ops),
+	.err_handler = &amdgpu_pci_err_handler,
+	.dev_groups = amdgpu_sysfs_groups,
 };
 
-
-
 static int __init amdgpu_init(void)
 {
 	int r;
@@ -883,28 +3143,25 @@ static int __init amdgpu_init(void)
 	if (r)
 		goto error_sync;
 
-	r = amdgpu_fence_slab_init();
+	r = amdgpu_userq_fence_slab_init();
 	if (r)
 		goto error_fence;
 
-	r = amd_sched_fence_slab_init();
-	if (r)
-		goto error_sched;
-
-	if (vgacon_text_force()) {
-		DRM_ERROR("VGACON disables amdgpu kernel modesetting.\n");
-		return -EINVAL;
-	}
 	DRM_INFO("amdgpu kernel modesetting enabled.\n");
-	driver = &kms_driver;
-	pdriver = &amdgpu_kms_pci_driver;
-	driver->num_ioctls = amdgpu_max_kms_ioctl;
 	amdgpu_register_atpx_handler();
-	/* let modprobe override vga console setting */
-	return pci_register_driver(pdriver);
+	amdgpu_acpi_detect();
 
-error_sched:
-	amdgpu_fence_slab_fini();
+	/* Ignore KFD init failures. Normal when CONFIG_HSA_AMD is not set. */
+	amdgpu_amdkfd_init();
+
+	if (amdgpu_pp_feature_mask & PP_OVERDRIVE_MASK) {
+		add_taint(TAINT_CPU_OUT_OF_SPEC, LOCKDEP_STILL_OK);
+		pr_crit("Overdrive is enabled, please disable it before "
+			"reporting any bugs unrelated to overdrive.\n");
+	}
+
+	/* let modprobe override vga console setting */
+	return pci_register_driver(&amdgpu_kms_pci_driver);
 
 error_fence:
 	amdgpu_sync_fini();
@@ -916,11 +3173,13 @@ error_sync:
 static void __exit amdgpu_exit(void)
 {
 	amdgpu_amdkfd_fini();
-	pci_unregister_driver(pdriver);
+	pci_unregister_driver(&amdgpu_kms_pci_driver);
 	amdgpu_unregister_atpx_handler();
+	amdgpu_acpi_release();
 	amdgpu_sync_fini();
-	amd_sched_fence_slab_fini();
-	amdgpu_fence_slab_fini();
+	amdgpu_userq_fence_slab_fini();
+	mmu_notifier_synchronize();
+	amdgpu_xcp_drv_release();
 }
 
 module_init(amdgpu_init);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.h
index e3a4f7048042..2d86cc6f7f4d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.h
@@ -40,9 +40,13 @@
 
 #define DRIVER_NAME		"amdgpu"
 #define DRIVER_DESC		"AMD GPU"
-#define DRIVER_DATE		"20150101"
+
+extern const struct drm_driver amdgpu_partition_driver;
 
 long amdgpu_drm_ioctl(struct file *filp,
 		      unsigned int cmd, unsigned long arg);
 
+long amdgpu_kms_compat_ioctl(struct file *filp,
+			     unsigned int cmd, unsigned long arg);
+
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_eeprom.c
new file mode 100644
index 000000000000..8cd69836dd99
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_eeprom.c
@@ -0,0 +1,240 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu_eeprom.h"
+#include "amdgpu.h"
+
+/* AT24CM02 and M24M02-R have a 256-byte write page size.
+ */
+#define EEPROM_PAGE_BITS   8
+#define EEPROM_PAGE_SIZE   (1U << EEPROM_PAGE_BITS)
+#define EEPROM_PAGE_MASK   (EEPROM_PAGE_SIZE - 1)
+
+#define EEPROM_OFFSET_SIZE 2
+
+/* EEPROM memory addresses are 19-bits long, which can
+ * be partitioned into 3, 8, 8 bits, for a total of 19.
+ * The upper 3 bits are sent as part of the 7-bit
+ * "Device Type Identifier"--an I2C concept, which for EEPROM devices
+ * is hard-coded as 1010b, indicating that it is an EEPROM
+ * device--this is the wire format, followed by the upper
+ * 3 bits of the 19-bit address, followed by the direction,
+ * followed by two bytes holding the rest of the 16-bits of
+ * the EEPROM memory address. The format on the wire for EEPROM
+ * devices is: 1010XYZD, A15:A8, A7:A0,
+ * Where D is the direction and sequenced out by the hardware.
+ * Bits XYZ are memory address bits 18, 17 and 16.
+ * These bits are compared to how pins 1-3 of the part are connected,
+ * depending on the size of the part, more on that later.
+ *
+ * Note that of this wire format, a client is in control
+ * of, and needs to specify only XYZ, A15:A8, A7:0, bits,
+ * which is exactly the EEPROM memory address, or offset,
+ * in order to address up to 8 EEPROM devices on the I2C bus.
+ *
+ * For instance, a 2-Mbit I2C EEPROM part, addresses all its bytes,
+ * using an 18-bit address, bit 17 to 0 and thus would use all but one bit of
+ * the 19 bits previously mentioned. The designer would then not connect
+ * pins 1 and 2, and pin 3 usually named "A_2" or "E2", would be connected to
+ * either Vcc or GND. This would allow for up to two 2-Mbit parts on
+ * the same bus, where one would be addressable with bit 18 as 1, and
+ * the other with bit 18 of the address as 0.
+ *
+ * For a 2-Mbit part, bit 18 is usually known as the "Chip Enable" or
+ * "Hardware Address Bit". This bit is compared to the load on pin 3
+ * of the device, described above, and if there is a match, then this
+ * device responds to the command. This way, you can connect two
+ * 2-Mbit EEPROM devices on the same bus, but see one contiguous
+ * memory from 0 to 7FFFFh, where address 0 to 3FFFF is in the device
+ * whose pin 3 is connected to GND, and address 40000 to 7FFFFh is in
+ * the 2nd device, whose pin 3 is connected to Vcc.
+ *
+ * This addressing you encode in the 32-bit "eeprom_addr" below,
+ * namely the 19-bits "XYZ,A15:A0", as a single 19-bit address. For
+ * instance, eeprom_addr = 0x6DA01, is 110_1101_1010_0000_0001, where
+ * XYZ=110b, and A15:A0=DA01h. The XYZ bits become part of the device
+ * address, and the rest of the address bits are sent as the memory
+ * address bytes.
+ *
+ * That is, for an I2C EEPROM driver everything is controlled by
+ * the "eeprom_addr".
+ *
+ * See also top of amdgpu_ras_eeprom.c.
+ *
+ * P.S. If you need to write, lock and read the Identification Page,
+ * (M24M02-DR device only, which we do not use), change the "7" to
+ * "0xF" in the macro below, and let the client set bit 20 to 1 in
+ * "eeprom_addr", and set A10 to 0 to write into it, and A10 and A1 to
+ * 1 to lock it permanently.
+ */
+#define MAKE_I2C_ADDR(_aa) ((0xA << 3) | (((_aa) >> 16) & 0xF))
+
+static int __amdgpu_eeprom_xfer(struct i2c_adapter *i2c_adap, u32 eeprom_addr,
+				u8 *eeprom_buf, u32 buf_size, bool read)
+{
+	u8 eeprom_offset_buf[EEPROM_OFFSET_SIZE];
+	struct i2c_msg msgs[] = {
+		{
+			.flags = 0,
+			.len = EEPROM_OFFSET_SIZE,
+			.buf = eeprom_offset_buf,
+		},
+		{
+			.flags = read ? I2C_M_RD : 0,
+		},
+	};
+	const u8 *p = eeprom_buf;
+	int r;
+	u16 len;
+
+	for (r = 0; buf_size > 0;
+	      buf_size -= len, eeprom_addr += len, eeprom_buf += len) {
+		/* Set the EEPROM address we want to write to/read from.
+		 */
+		msgs[0].addr = MAKE_I2C_ADDR(eeprom_addr);
+		msgs[1].addr = msgs[0].addr;
+		msgs[0].buf[0] = (eeprom_addr >> 8) & 0xff;
+		msgs[0].buf[1] = eeprom_addr & 0xff;
+
+		if (!read) {
+			/* Write the maximum amount of data, without
+			 * crossing the device's page boundary, as per
+			 * its spec. Partial page writes are allowed,
+			 * starting at any location within the page,
+			 * so long as the page boundary isn't crossed
+			 * over (actually the page pointer rolls
+			 * over).
+			 *
+			 * As per the AT24CM02 EEPROM spec, after
+			 * writing into a page, the I2C driver should
+			 * terminate the transfer, i.e. in
+			 * "i2c_transfer()" below, with a STOP
+			 * condition, so that the self-timed write
+			 * cycle begins. This is implied for the
+			 * "i2c_transfer()" abstraction.
+			 */
+			len = min(EEPROM_PAGE_SIZE - (eeprom_addr & EEPROM_PAGE_MASK),
+					buf_size);
+		} else {
+			/* Reading from the EEPROM has no limitation
+			 * on the number of bytes read from the EEPROM
+			 * device--they are simply sequenced out.
+			 * Keep in mind that i2c_msg.len is u16 type.
+			 */
+			len = min(U16_MAX, buf_size);
+		}
+		msgs[1].len = len;
+		msgs[1].buf = eeprom_buf;
+
+		/* This constitutes a START-STOP transaction.
+		 */
+		r = i2c_transfer(i2c_adap, msgs, ARRAY_SIZE(msgs));
+		if (r != ARRAY_SIZE(msgs))
+			break;
+
+		if (!read) {
+			/* According to EEPROM specs the length of the
+			 * self-writing cycle, tWR (tW), is 10 ms.
+			 *
+			 * TODO: Use polling on ACK, aka Acknowledge
+			 * Polling, to minimize waiting for the
+			 * internal write cycle to complete, as it is
+			 * usually smaller than tWR (tW).
+			 */
+			msleep(10);
+		}
+	}
+
+	return r < 0 ? r : eeprom_buf - p;
+}
+
+/**
+ * amdgpu_eeprom_xfer -- Read/write from/to an I2C EEPROM device
+ * @i2c_adap: pointer to the I2C adapter to use
+ * @eeprom_addr: EEPROM address from which to read/write
+ * @eeprom_buf: pointer to data buffer to read into/write from
+ * @buf_size: the size of @eeprom_buf
+ * @read: True if reading from the EEPROM, false if writing
+ *
+ * Returns the number of bytes read/written; -errno on error.
+ */
+static int amdgpu_eeprom_xfer(struct i2c_adapter *i2c_adap, u32 eeprom_addr,
+			      u8 *eeprom_buf, u32 buf_size, bool read)
+{
+	const struct i2c_adapter_quirks *quirks = i2c_adap->quirks;
+	u16 limit;
+	u16 ps; /* Partial size */
+	int res = 0, r;
+
+	if (!quirks)
+		limit = 0;
+	else if (read)
+		limit = quirks->max_read_len;
+	else
+		limit = quirks->max_write_len;
+
+	if (limit == 0) {
+		return __amdgpu_eeprom_xfer(i2c_adap, eeprom_addr,
+					    eeprom_buf, buf_size, read);
+	} else if (limit <= EEPROM_OFFSET_SIZE) {
+		dev_err_ratelimited(&i2c_adap->dev,
+				    "maddr:0x%04X size:0x%02X:quirk max_%s_len must be > %d",
+				    eeprom_addr, buf_size,
+				    str_read_write(read), EEPROM_OFFSET_SIZE);
+		return -EINVAL;
+	}
+
+	/* The "limit" includes all data bytes sent/received,
+	 * which would include the EEPROM_OFFSET_SIZE bytes.
+	 * Account for them here.
+	 */
+	limit -= EEPROM_OFFSET_SIZE;
+	for ( ; buf_size > 0;
+	      buf_size -= ps, eeprom_addr += ps, eeprom_buf += ps) {
+		ps = min(limit, buf_size);
+
+		r = __amdgpu_eeprom_xfer(i2c_adap, eeprom_addr,
+					 eeprom_buf, ps, read);
+		if (r < 0)
+			return r;
+		res += r;
+	}
+
+	return res;
+}
+
+int amdgpu_eeprom_read(struct i2c_adapter *i2c_adap,
+		       u32 eeprom_addr, u8 *eeprom_buf,
+		       u32 bytes)
+{
+	return amdgpu_eeprom_xfer(i2c_adap, eeprom_addr, eeprom_buf, bytes,
+				  true);
+}
+
+int amdgpu_eeprom_write(struct i2c_adapter *i2c_adap,
+			u32 eeprom_addr, u8 *eeprom_buf,
+			u32 bytes)
+{
+	return amdgpu_eeprom_xfer(i2c_adap, eeprom_addr, eeprom_buf, bytes,
+				  false);
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_eeprom.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_eeprom.h
new file mode 100644
index 000000000000..8083b8253ef4
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_eeprom.h
@@ -0,0 +1,37 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef _AMDGPU_EEPROM_H
+#define _AMDGPU_EEPROM_H
+
+#include <linux/i2c.h>
+
+int amdgpu_eeprom_read(struct i2c_adapter *i2c_adap,
+		       u32 eeprom_addr, u8 *eeprom_buf,
+		       u32 bytes);
+
+int amdgpu_eeprom_write(struct i2c_adapter *i2c_adap,
+			u32 eeprom_addr, u8 *eeprom_buf,
+			u32 bytes);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_encoders.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_encoders.c
index 94138abe093b..3aaeed2d3562 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_encoders.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_encoders.c
@@ -23,30 +23,32 @@
  * Authors: Dave Airlie
  *          Alex Deucher
  */
-#include <drm/drmP.h>
-#include <drm/drm_crtc_helper.h>
+
 #include <drm/amdgpu_drm.h>
 #include "amdgpu.h"
 #include "amdgpu_connectors.h"
+#include "amdgpu_display.h"
 #include "atom.h"
 #include "atombios_encoders.h"
 
 void
 amdgpu_link_encoder_connector(struct drm_device *dev)
 {
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector;
 	struct drm_encoder *encoder;
 	struct amdgpu_encoder *amdgpu_encoder;
 
+	drm_connector_list_iter_begin(dev, &iter);
 	/* walk the list and link encoders to connectors */
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_for_each_connector_iter(connector, &iter) {
 		amdgpu_connector = to_amdgpu_connector(connector);
 		list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
 			amdgpu_encoder = to_amdgpu_encoder(encoder);
 			if (amdgpu_encoder->devices & amdgpu_connector->devices) {
-				drm_mode_connector_attach_encoder(connector, encoder);
+				drm_connector_attach_encoder(connector, encoder);
 				if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) {
 					amdgpu_atombios_encoder_init_backlight(amdgpu_encoder, connector);
 					adev->mode_info.bl_encoder = amdgpu_encoder;
@@ -54,6 +56,7 @@ amdgpu_link_encoder_connector(struct drm_device *dev)
 			}
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 }
 
 void amdgpu_encoder_set_active_device(struct drm_encoder *encoder)
@@ -61,16 +64,20 @@ void amdgpu_encoder_set_active_device(struct drm_encoder *encoder)
 	struct drm_device *dev = encoder->dev;
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
 			amdgpu_encoder->active_device = amdgpu_encoder->devices & amdgpu_connector->devices;
 			DRM_DEBUG_KMS("setting active device to %08x from %08x %08x for encoder %d\n",
 				  amdgpu_encoder->active_device, amdgpu_encoder->devices,
 				  amdgpu_connector->devices, encoder->encoder_type);
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 }
 
 struct drm_connector *
@@ -78,15 +85,20 @@ amdgpu_get_connector_for_encoder(struct drm_encoder *encoder)
 {
 	struct drm_device *dev = encoder->dev;
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct drm_connector *connector;
+	struct drm_connector *connector, *found = NULL;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector;
 
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		amdgpu_connector = to_amdgpu_connector(connector);
-		if (amdgpu_encoder->active_device & amdgpu_connector->devices)
-			return connector;
+		if (amdgpu_encoder->active_device & amdgpu_connector->devices) {
+			found = connector;
+			break;
+		}
 	}
-	return NULL;
+	drm_connector_list_iter_end(&iter);
+	return found;
 }
 
 struct drm_connector *
@@ -94,15 +106,20 @@ amdgpu_get_connector_for_encoder_init(struct drm_encoder *encoder)
 {
 	struct drm_device *dev = encoder->dev;
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct drm_connector *connector;
+	struct drm_connector *connector, *found = NULL;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector;
 
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		amdgpu_connector = to_amdgpu_connector(connector);
-		if (amdgpu_encoder->devices & amdgpu_connector->devices)
-			return connector;
+		if (amdgpu_encoder->devices & amdgpu_connector->devices) {
+			found = connector;
+			break;
+		}
 	}
-	return NULL;
+	drm_connector_list_iter_end(&iter);
+	return found;
 }
 
 struct drm_encoder *amdgpu_get_external_encoder(struct drm_encoder *encoder)
@@ -149,12 +166,12 @@ void amdgpu_panel_mode_fixup(struct drm_encoder *encoder,
 {
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode;
-	unsigned hblank = native_mode->htotal - native_mode->hdisplay;
-	unsigned vblank = native_mode->vtotal - native_mode->vdisplay;
-	unsigned hover = native_mode->hsync_start - native_mode->hdisplay;
-	unsigned vover = native_mode->vsync_start - native_mode->vdisplay;
-	unsigned hsync_width = native_mode->hsync_end - native_mode->hsync_start;
-	unsigned vsync_width = native_mode->vsync_end - native_mode->vsync_start;
+	unsigned int hblank = native_mode->htotal - native_mode->hdisplay;
+	unsigned int vblank = native_mode->vtotal - native_mode->vdisplay;
+	unsigned int hover = native_mode->hsync_start - native_mode->hdisplay;
+	unsigned int vover = native_mode->vsync_start - native_mode->vdisplay;
+	unsigned int hsync_width = native_mode->hsync_end - native_mode->hsync_start;
+	unsigned int vsync_width = native_mode->vsync_end - native_mode->vsync_start;
 
 	adjusted_mode->clock = native_mode->clock;
 	adjusted_mode->flags = native_mode->flags;
@@ -205,7 +222,7 @@ bool amdgpu_dig_monitor_is_duallink(struct drm_encoder *encoder,
 	case DRM_MODE_CONNECTOR_HDMIB:
 		if (amdgpu_connector->use_digital) {
 			/* HDMI 1.3 supports up to 340 Mhz over single link */
-			if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+			if (connector->display_info.is_hdmi) {
 				if (pixel_clock > 340000)
 					return true;
 				else
@@ -227,7 +244,7 @@ bool amdgpu_dig_monitor_is_duallink(struct drm_encoder *encoder,
 			return false;
 		else {
 			/* HDMI 1.3 supports up to 340 Mhz over single link */
-			if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+			if (connector->display_info.is_hdmi) {
 				if (pixel_clock > 340000)
 					return true;
 				else
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c
new file mode 100644
index 000000000000..23d7d0b0d625
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c
@@ -0,0 +1,241 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/sched.h>
+#include <drm/drm_exec.h>
+#include "amdgpu.h"
+
+#define work_to_evf_mgr(w, name) container_of(w, struct amdgpu_eviction_fence_mgr, name)
+#define evf_mgr_to_fpriv(e) container_of(e, struct amdgpu_fpriv, evf_mgr)
+
+static const char *
+amdgpu_eviction_fence_get_driver_name(struct dma_fence *fence)
+{
+	return "amdgpu_eviction_fence";
+}
+
+static const char *
+amdgpu_eviction_fence_get_timeline_name(struct dma_fence *f)
+{
+	struct amdgpu_eviction_fence *ef;
+
+	ef = container_of(f, struct amdgpu_eviction_fence, base);
+	return ef->timeline_name;
+}
+
+int
+amdgpu_eviction_fence_replace_fence(struct amdgpu_eviction_fence_mgr *evf_mgr,
+				    struct drm_exec *exec)
+{
+	struct amdgpu_eviction_fence *old_ef, *new_ef;
+	struct drm_gem_object *obj;
+	unsigned long index;
+	int ret;
+
+	if (evf_mgr->ev_fence &&
+	    !dma_fence_is_signaled(&evf_mgr->ev_fence->base))
+		return 0;
+	/*
+	 * Steps to replace eviction fence:
+	 * * lock all objects in exec (caller)
+	 * * create a new eviction fence
+	 * * update new eviction fence in evf_mgr
+	 * * attach the new eviction fence to BOs
+	 * * release the old fence
+	 * * unlock the objects (caller)
+	 */
+	new_ef = amdgpu_eviction_fence_create(evf_mgr);
+	if (!new_ef) {
+		DRM_ERROR("Failed to create new eviction fence\n");
+		return -ENOMEM;
+	}
+
+	/* Update the eviction fence now */
+	spin_lock(&evf_mgr->ev_fence_lock);
+	old_ef = evf_mgr->ev_fence;
+	evf_mgr->ev_fence = new_ef;
+	spin_unlock(&evf_mgr->ev_fence_lock);
+
+	/* Attach the new fence */
+	drm_exec_for_each_locked_object(exec, index, obj) {
+		struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+
+		if (!bo)
+			continue;
+		ret = amdgpu_eviction_fence_attach(evf_mgr, bo);
+		if (ret) {
+			DRM_ERROR("Failed to attch new eviction fence\n");
+			goto free_err;
+		}
+	}
+
+	/* Free old fence */
+	if (old_ef)
+		dma_fence_put(&old_ef->base);
+	return 0;
+
+free_err:
+	kfree(new_ef);
+	return ret;
+}
+
+static void
+amdgpu_eviction_fence_suspend_worker(struct work_struct *work)
+{
+	struct amdgpu_eviction_fence_mgr *evf_mgr = work_to_evf_mgr(work, suspend_work.work);
+	struct amdgpu_fpriv *fpriv = evf_mgr_to_fpriv(evf_mgr);
+	struct amdgpu_userq_mgr *uq_mgr = &fpriv->userq_mgr;
+	struct amdgpu_eviction_fence *ev_fence;
+
+	mutex_lock(&uq_mgr->userq_mutex);
+	spin_lock(&evf_mgr->ev_fence_lock);
+	ev_fence = evf_mgr->ev_fence;
+	if (ev_fence)
+		dma_fence_get(&ev_fence->base);
+	else
+		goto unlock;
+	spin_unlock(&evf_mgr->ev_fence_lock);
+
+	amdgpu_userq_evict(uq_mgr, ev_fence);
+
+	mutex_unlock(&uq_mgr->userq_mutex);
+	dma_fence_put(&ev_fence->base);
+	return;
+
+unlock:
+	spin_unlock(&evf_mgr->ev_fence_lock);
+	mutex_unlock(&uq_mgr->userq_mutex);
+}
+
+static bool amdgpu_eviction_fence_enable_signaling(struct dma_fence *f)
+{
+	struct amdgpu_eviction_fence_mgr *evf_mgr;
+	struct amdgpu_eviction_fence *ev_fence;
+
+	if (!f)
+		return true;
+
+	ev_fence = to_ev_fence(f);
+	evf_mgr = ev_fence->evf_mgr;
+
+	schedule_delayed_work(&evf_mgr->suspend_work, 0);
+	return true;
+}
+
+static const struct dma_fence_ops amdgpu_eviction_fence_ops = {
+	.get_driver_name = amdgpu_eviction_fence_get_driver_name,
+	.get_timeline_name = amdgpu_eviction_fence_get_timeline_name,
+	.enable_signaling = amdgpu_eviction_fence_enable_signaling,
+};
+
+void amdgpu_eviction_fence_signal(struct amdgpu_eviction_fence_mgr *evf_mgr,
+				  struct amdgpu_eviction_fence *ev_fence)
+{
+	spin_lock(&evf_mgr->ev_fence_lock);
+	dma_fence_signal(&ev_fence->base);
+	spin_unlock(&evf_mgr->ev_fence_lock);
+}
+
+struct amdgpu_eviction_fence *
+amdgpu_eviction_fence_create(struct amdgpu_eviction_fence_mgr *evf_mgr)
+{
+	struct amdgpu_eviction_fence *ev_fence;
+
+	ev_fence = kzalloc(sizeof(*ev_fence), GFP_KERNEL);
+	if (!ev_fence)
+		return NULL;
+
+	ev_fence->evf_mgr = evf_mgr;
+	get_task_comm(ev_fence->timeline_name, current);
+	spin_lock_init(&ev_fence->lock);
+	dma_fence_init64(&ev_fence->base, &amdgpu_eviction_fence_ops,
+			 &ev_fence->lock, evf_mgr->ev_fence_ctx,
+			 atomic_inc_return(&evf_mgr->ev_fence_seq));
+	return ev_fence;
+}
+
+void amdgpu_eviction_fence_destroy(struct amdgpu_eviction_fence_mgr *evf_mgr)
+{
+	struct amdgpu_eviction_fence *ev_fence;
+
+	/* Wait for any pending work to execute */
+	flush_delayed_work(&evf_mgr->suspend_work);
+
+	spin_lock(&evf_mgr->ev_fence_lock);
+	ev_fence = evf_mgr->ev_fence;
+	spin_unlock(&evf_mgr->ev_fence_lock);
+
+	if (!ev_fence)
+		return;
+
+	dma_fence_wait(&ev_fence->base, false);
+
+	/* Last unref of ev_fence */
+	dma_fence_put(&ev_fence->base);
+}
+
+int amdgpu_eviction_fence_attach(struct amdgpu_eviction_fence_mgr *evf_mgr,
+				 struct amdgpu_bo *bo)
+{
+	struct amdgpu_eviction_fence *ev_fence;
+	struct dma_resv *resv = bo->tbo.base.resv;
+	int ret;
+
+	if (!resv)
+		return 0;
+
+	ret = dma_resv_reserve_fences(resv, 1);
+	if (ret) {
+		DRM_DEBUG_DRIVER("Failed to resv fence space\n");
+		return ret;
+	}
+
+	spin_lock(&evf_mgr->ev_fence_lock);
+	ev_fence = evf_mgr->ev_fence;
+	if (ev_fence)
+		dma_resv_add_fence(resv, &ev_fence->base, DMA_RESV_USAGE_BOOKKEEP);
+	spin_unlock(&evf_mgr->ev_fence_lock);
+
+	return 0;
+}
+
+void amdgpu_eviction_fence_detach(struct amdgpu_eviction_fence_mgr *evf_mgr,
+				  struct amdgpu_bo *bo)
+{
+	struct dma_fence *stub = dma_fence_get_stub();
+
+	dma_resv_replace_fences(bo->tbo.base.resv, evf_mgr->ev_fence_ctx,
+				stub, DMA_RESV_USAGE_BOOKKEEP);
+	dma_fence_put(stub);
+}
+
+int amdgpu_eviction_fence_init(struct amdgpu_eviction_fence_mgr *evf_mgr)
+{
+	/* This needs to be done one time per open */
+	atomic_set(&evf_mgr->ev_fence_seq, 0);
+	evf_mgr->ev_fence_ctx = dma_fence_context_alloc(1);
+	spin_lock_init(&evf_mgr->ev_fence_lock);
+
+	INIT_DELAYED_WORK(&evf_mgr->suspend_work, amdgpu_eviction_fence_suspend_worker);
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.h
new file mode 100644
index 000000000000..fcd867b7147d
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.h
@@ -0,0 +1,69 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef AMDGPU_EV_FENCE_H_
+#define AMDGPU_EV_FENCE_H_
+
+struct amdgpu_eviction_fence {
+	struct dma_fence base;
+	spinlock_t	 lock;
+	char		 timeline_name[TASK_COMM_LEN];
+	struct amdgpu_eviction_fence_mgr *evf_mgr;
+};
+
+struct amdgpu_eviction_fence_mgr {
+	u64			ev_fence_ctx;
+	atomic_t		ev_fence_seq;
+	spinlock_t		ev_fence_lock;
+	struct amdgpu_eviction_fence *ev_fence;
+	struct delayed_work	suspend_work;
+	uint8_t fd_closing;
+};
+
+/* Eviction fence helper functions */
+struct amdgpu_eviction_fence *
+amdgpu_eviction_fence_create(struct amdgpu_eviction_fence_mgr *evf_mgr);
+
+void
+amdgpu_eviction_fence_destroy(struct amdgpu_eviction_fence_mgr *evf_mgr);
+
+int
+amdgpu_eviction_fence_attach(struct amdgpu_eviction_fence_mgr *evf_mgr,
+			     struct amdgpu_bo *bo);
+
+void
+amdgpu_eviction_fence_detach(struct amdgpu_eviction_fence_mgr *evf_mgr,
+			     struct amdgpu_bo *bo);
+
+int
+amdgpu_eviction_fence_init(struct amdgpu_eviction_fence_mgr *evf_mgr);
+
+void
+amdgpu_eviction_fence_signal(struct amdgpu_eviction_fence_mgr *evf_mgr,
+			     struct amdgpu_eviction_fence *ev_fence);
+
+int
+amdgpu_eviction_fence_replace_fence(struct amdgpu_eviction_fence_mgr *evf_mgr,
+				    struct drm_exec *exec);
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c
deleted file mode 100644
index 9afa9c097e1f..000000000000
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c
+++ /dev/null
@@ -1,420 +0,0 @@
-/*
- * Copyright © 2007 David Airlie
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice (including the next
- * paragraph) shall be included in all copies or substantial portions of the
- * Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
- * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
- * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
- * DEALINGS IN THE SOFTWARE.
- *
- * Authors:
- *     David Airlie
- */
-#include <linux/module.h>
-#include <linux/slab.h>
-#include <linux/pm_runtime.h>
-
-#include <drm/drmP.h>
-#include <drm/drm_crtc.h>
-#include <drm/drm_crtc_helper.h>
-#include <drm/amdgpu_drm.h>
-#include "amdgpu.h"
-#include "cikd.h"
-
-#include <drm/drm_fb_helper.h>
-
-#include <linux/vga_switcheroo.h>
-
-/* object hierarchy -
-   this contains a helper + a amdgpu fb
-   the helper contains a pointer to amdgpu framebuffer baseclass.
-*/
-struct amdgpu_fbdev {
-	struct drm_fb_helper helper;
-	struct amdgpu_framebuffer rfb;
-	struct amdgpu_device *adev;
-};
-
-static int
-amdgpufb_open(struct fb_info *info, int user)
-{
-	struct amdgpu_fbdev *rfbdev = info->par;
-	struct amdgpu_device *adev = rfbdev->adev;
-	int ret = pm_runtime_get_sync(adev->ddev->dev);
-	if (ret < 0 && ret != -EACCES) {
-		pm_runtime_mark_last_busy(adev->ddev->dev);
-		pm_runtime_put_autosuspend(adev->ddev->dev);
-		return ret;
-	}
-	return 0;
-}
-
-static int
-amdgpufb_release(struct fb_info *info, int user)
-{
-	struct amdgpu_fbdev *rfbdev = info->par;
-	struct amdgpu_device *adev = rfbdev->adev;
-
-	pm_runtime_mark_last_busy(adev->ddev->dev);
-	pm_runtime_put_autosuspend(adev->ddev->dev);
-	return 0;
-}
-
-static struct fb_ops amdgpufb_ops = {
-	.owner = THIS_MODULE,
-	DRM_FB_HELPER_DEFAULT_OPS,
-	.fb_open = amdgpufb_open,
-	.fb_release = amdgpufb_release,
-	.fb_fillrect = drm_fb_helper_cfb_fillrect,
-	.fb_copyarea = drm_fb_helper_cfb_copyarea,
-	.fb_imageblit = drm_fb_helper_cfb_imageblit,
-};
-
-
-int amdgpu_align_pitch(struct amdgpu_device *adev, int width, int cpp, bool tiled)
-{
-	int aligned = width;
-	int pitch_mask = 0;
-
-	switch (cpp) {
-	case 1:
-		pitch_mask = 255;
-		break;
-	case 2:
-		pitch_mask = 127;
-		break;
-	case 3:
-	case 4:
-		pitch_mask = 63;
-		break;
-	}
-
-	aligned += pitch_mask;
-	aligned &= ~pitch_mask;
-	return aligned * cpp;
-}
-
-static void amdgpufb_destroy_pinned_object(struct drm_gem_object *gobj)
-{
-	struct amdgpu_bo *abo = gem_to_amdgpu_bo(gobj);
-	int ret;
-
-	ret = amdgpu_bo_reserve(abo, true);
-	if (likely(ret == 0)) {
-		amdgpu_bo_kunmap(abo);
-		amdgpu_bo_unpin(abo);
-		amdgpu_bo_unreserve(abo);
-	}
-	drm_gem_object_put_unlocked(gobj);
-}
-
-static int amdgpufb_create_pinned_object(struct amdgpu_fbdev *rfbdev,
-					 struct drm_mode_fb_cmd2 *mode_cmd,
-					 struct drm_gem_object **gobj_p)
-{
-	struct amdgpu_device *adev = rfbdev->adev;
-	struct drm_gem_object *gobj = NULL;
-	struct amdgpu_bo *abo = NULL;
-	bool fb_tiled = false; /* useful for testing */
-	u32 tiling_flags = 0;
-	int ret;
-	int aligned_size, size;
-	int height = mode_cmd->height;
-	u32 cpp;
-
-	cpp = drm_format_plane_cpp(mode_cmd->pixel_format, 0);
-
-	/* need to align pitch with crtc limits */
-	mode_cmd->pitches[0] = amdgpu_align_pitch(adev, mode_cmd->width, cpp,
-						  fb_tiled);
-
-	height = ALIGN(mode_cmd->height, 8);
-	size = mode_cmd->pitches[0] * height;
-	aligned_size = ALIGN(size, PAGE_SIZE);
-	ret = amdgpu_gem_object_create(adev, aligned_size, 0,
-				       AMDGPU_GEM_DOMAIN_VRAM,
-				       AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
-				       AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS |
-				       AMDGPU_GEM_CREATE_VRAM_CLEARED,
-				       true, &gobj);
-	if (ret) {
-		pr_err("failed to allocate framebuffer (%d)\n", aligned_size);
-		return -ENOMEM;
-	}
-	abo = gem_to_amdgpu_bo(gobj);
-
-	if (fb_tiled)
-		tiling_flags = AMDGPU_TILING_SET(ARRAY_MODE, GRPH_ARRAY_2D_TILED_THIN1);
-
-	ret = amdgpu_bo_reserve(abo, false);
-	if (unlikely(ret != 0))
-		goto out_unref;
-
-	if (tiling_flags) {
-		ret = amdgpu_bo_set_tiling_flags(abo,
-						 tiling_flags);
-		if (ret)
-			dev_err(adev->dev, "FB failed to set tiling flags\n");
-	}
-
-
-	ret = amdgpu_bo_pin(abo, AMDGPU_GEM_DOMAIN_VRAM, NULL);
-	if (ret) {
-		amdgpu_bo_unreserve(abo);
-		goto out_unref;
-	}
-	ret = amdgpu_bo_kmap(abo, NULL);
-	amdgpu_bo_unreserve(abo);
-	if (ret) {
-		goto out_unref;
-	}
-
-	*gobj_p = gobj;
-	return 0;
-out_unref:
-	amdgpufb_destroy_pinned_object(gobj);
-	*gobj_p = NULL;
-	return ret;
-}
-
-static int amdgpufb_create(struct drm_fb_helper *helper,
-			   struct drm_fb_helper_surface_size *sizes)
-{
-	struct amdgpu_fbdev *rfbdev = (struct amdgpu_fbdev *)helper;
-	struct amdgpu_device *adev = rfbdev->adev;
-	struct fb_info *info;
-	struct drm_framebuffer *fb = NULL;
-	struct drm_mode_fb_cmd2 mode_cmd;
-	struct drm_gem_object *gobj = NULL;
-	struct amdgpu_bo *abo = NULL;
-	int ret;
-	unsigned long tmp;
-
-	mode_cmd.width = sizes->surface_width;
-	mode_cmd.height = sizes->surface_height;
-
-	if (sizes->surface_bpp == 24)
-		sizes->surface_bpp = 32;
-
-	mode_cmd.pixel_format = drm_mode_legacy_fb_format(sizes->surface_bpp,
-							  sizes->surface_depth);
-
-	ret = amdgpufb_create_pinned_object(rfbdev, &mode_cmd, &gobj);
-	if (ret) {
-		DRM_ERROR("failed to create fbcon object %d\n", ret);
-		return ret;
-	}
-
-	abo = gem_to_amdgpu_bo(gobj);
-
-	/* okay we have an object now allocate the framebuffer */
-	info = drm_fb_helper_alloc_fbi(helper);
-	if (IS_ERR(info)) {
-		ret = PTR_ERR(info);
-		goto out;
-	}
-
-	info->par = rfbdev;
-	info->skip_vt_switch = true;
-
-	ret = amdgpu_framebuffer_init(adev->ddev, &rfbdev->rfb, &mode_cmd, gobj);
-	if (ret) {
-		DRM_ERROR("failed to initialize framebuffer %d\n", ret);
-		goto out;
-	}
-
-	fb = &rfbdev->rfb.base;
-
-	/* setup helper */
-	rfbdev->helper.fb = fb;
-
-	strcpy(info->fix.id, "amdgpudrmfb");
-
-	drm_fb_helper_fill_fix(info, fb->pitches[0], fb->format->depth);
-
-	info->fbops = &amdgpufb_ops;
-
-	tmp = amdgpu_bo_gpu_offset(abo) - adev->mc.vram_start;
-	info->fix.smem_start = adev->mc.aper_base + tmp;
-	info->fix.smem_len = amdgpu_bo_size(abo);
-	info->screen_base = amdgpu_bo_kptr(abo);
-	info->screen_size = amdgpu_bo_size(abo);
-
-	drm_fb_helper_fill_var(info, &rfbdev->helper, sizes->fb_width, sizes->fb_height);
-
-	/* setup aperture base/size for vesafb takeover */
-	info->apertures->ranges[0].base = adev->ddev->mode_config.fb_base;
-	info->apertures->ranges[0].size = adev->mc.aper_size;
-
-	/* Use default scratch pixmap (info->pixmap.flags = FB_PIXMAP_SYSTEM) */
-
-	if (info->screen_base == NULL) {
-		ret = -ENOSPC;
-		goto out;
-	}
-
-	DRM_INFO("fb mappable at 0x%lX\n",  info->fix.smem_start);
-	DRM_INFO("vram apper at 0x%lX\n",  (unsigned long)adev->mc.aper_base);
-	DRM_INFO("size %lu\n", (unsigned long)amdgpu_bo_size(abo));
-	DRM_INFO("fb depth is %d\n", fb->format->depth);
-	DRM_INFO("   pitch is %d\n", fb->pitches[0]);
-
-	vga_switcheroo_client_fb_set(adev->ddev->pdev, info);
-	return 0;
-
-out:
-	if (abo) {
-
-	}
-	if (fb && ret) {
-		drm_gem_object_put_unlocked(gobj);
-		drm_framebuffer_unregister_private(fb);
-		drm_framebuffer_cleanup(fb);
-		kfree(fb);
-	}
-	return ret;
-}
-
-void amdgpu_fb_output_poll_changed(struct amdgpu_device *adev)
-{
-	if (adev->mode_info.rfbdev)
-		drm_fb_helper_hotplug_event(&adev->mode_info.rfbdev->helper);
-}
-
-static int amdgpu_fbdev_destroy(struct drm_device *dev, struct amdgpu_fbdev *rfbdev)
-{
-	struct amdgpu_framebuffer *rfb = &rfbdev->rfb;
-
-	drm_fb_helper_unregister_fbi(&rfbdev->helper);
-
-	if (rfb->obj) {
-		amdgpufb_destroy_pinned_object(rfb->obj);
-		rfb->obj = NULL;
-	}
-	drm_fb_helper_fini(&rfbdev->helper);
-	drm_framebuffer_unregister_private(&rfb->base);
-	drm_framebuffer_cleanup(&rfb->base);
-
-	return 0;
-}
-
-static const struct drm_fb_helper_funcs amdgpu_fb_helper_funcs = {
-	.fb_probe = amdgpufb_create,
-};
-
-int amdgpu_fbdev_init(struct amdgpu_device *adev)
-{
-	struct amdgpu_fbdev *rfbdev;
-	int bpp_sel = 32;
-	int ret;
-
-	/* don't init fbdev on hw without DCE */
-	if (!adev->mode_info.mode_config_initialized)
-		return 0;
-
-	/* don't init fbdev if there are no connectors */
-	if (list_empty(&adev->ddev->mode_config.connector_list))
-		return 0;
-
-	/* select 8 bpp console on low vram cards */
-	if (adev->mc.real_vram_size <= (32*1024*1024))
-		bpp_sel = 8;
-
-	rfbdev = kzalloc(sizeof(struct amdgpu_fbdev), GFP_KERNEL);
-	if (!rfbdev)
-		return -ENOMEM;
-
-	rfbdev->adev = adev;
-	adev->mode_info.rfbdev = rfbdev;
-
-	drm_fb_helper_prepare(adev->ddev, &rfbdev->helper,
-			&amdgpu_fb_helper_funcs);
-
-	ret = drm_fb_helper_init(adev->ddev, &rfbdev->helper,
-				 AMDGPUFB_CONN_LIMIT);
-	if (ret) {
-		kfree(rfbdev);
-		return ret;
-	}
-
-	drm_fb_helper_single_add_all_connectors(&rfbdev->helper);
-
-	/* disable all the possible outputs/crtcs before entering KMS mode */
-	drm_helper_disable_unused_functions(adev->ddev);
-
-	drm_fb_helper_initial_config(&rfbdev->helper, bpp_sel);
-	return 0;
-}
-
-void amdgpu_fbdev_fini(struct amdgpu_device *adev)
-{
-	if (!adev->mode_info.rfbdev)
-		return;
-
-	amdgpu_fbdev_destroy(adev->ddev, adev->mode_info.rfbdev);
-	kfree(adev->mode_info.rfbdev);
-	adev->mode_info.rfbdev = NULL;
-}
-
-void amdgpu_fbdev_set_suspend(struct amdgpu_device *adev, int state)
-{
-	if (adev->mode_info.rfbdev)
-		drm_fb_helper_set_suspend(&adev->mode_info.rfbdev->helper,
-			state);
-}
-
-int amdgpu_fbdev_total_size(struct amdgpu_device *adev)
-{
-	struct amdgpu_bo *robj;
-	int size = 0;
-
-	if (!adev->mode_info.rfbdev)
-		return 0;
-
-	robj = gem_to_amdgpu_bo(adev->mode_info.rfbdev->rfb.obj);
-	size += amdgpu_bo_size(robj);
-	return size;
-}
-
-bool amdgpu_fbdev_robj_is_fb(struct amdgpu_device *adev, struct amdgpu_bo *robj)
-{
-	if (!adev->mode_info.rfbdev)
-		return false;
-	if (robj == gem_to_amdgpu_bo(adev->mode_info.rfbdev->rfb.obj))
-		return true;
-	return false;
-}
-
-void amdgpu_fbdev_restore_mode(struct amdgpu_device *adev)
-{
-	struct amdgpu_fbdev *afbdev;
-	struct drm_fb_helper *fb_helper;
-	int ret;
-
-	if (!adev)
-		return;
-
-	afbdev = adev->mode_info.rfbdev;
-
-	if (!afbdev)
-		return;
-
-	fb_helper = &afbdev->helper;
-
-	ret = drm_fb_helper_restore_fbdev_mode_unlocked(fb_helper);
-	if (ret)
-		DRM_DEBUG("failed to restore crtc mode\n");
-}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fdinfo.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fdinfo.c
new file mode 100644
index 000000000000..b349bb3676d5
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fdinfo.c
@@ -0,0 +1,124 @@
+// SPDX-License-Identifier: MIT
+/* Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: David Nieto
+ *          Roy Sun
+ */
+
+#include <linux/debugfs.h>
+#include <linux/list.h>
+#include <linux/module.h>
+#include <linux/uaccess.h>
+#include <linux/reboot.h>
+#include <linux/syscalls.h>
+
+#include <drm/amdgpu_drm.h>
+#include <drm/drm_debugfs.h>
+#include <drm/drm_drv.h>
+#include <drm/drm_file.h>
+
+#include "amdgpu.h"
+#include "amdgpu_vm.h"
+#include "amdgpu_gem.h"
+#include "amdgpu_ctx.h"
+#include "amdgpu_fdinfo.h"
+
+
+static const char *amdgpu_ip_name[AMDGPU_HW_IP_NUM] = {
+	[AMDGPU_HW_IP_GFX]	=	"gfx",
+	[AMDGPU_HW_IP_COMPUTE]	=	"compute",
+	[AMDGPU_HW_IP_DMA]	=	"dma",
+	[AMDGPU_HW_IP_UVD]	=	"dec",
+	[AMDGPU_HW_IP_VCE]	=	"enc",
+	[AMDGPU_HW_IP_UVD_ENC]	=	"enc_1",
+	[AMDGPU_HW_IP_VCN_DEC]	=	"dec",
+	[AMDGPU_HW_IP_VCN_ENC]	=	"enc",
+	[AMDGPU_HW_IP_VCN_JPEG]	=	"jpeg",
+	[AMDGPU_HW_IP_VPE]	=	"vpe",
+};
+
+void amdgpu_show_fdinfo(struct drm_printer *p, struct drm_file *file)
+{
+	struct amdgpu_fpriv *fpriv = file->driver_priv;
+	struct amdgpu_vm *vm = &fpriv->vm;
+
+	struct amdgpu_mem_stats stats[__AMDGPU_PL_NUM];
+	ktime_t usage[AMDGPU_HW_IP_NUM];
+	const char *pl_name[] = {
+		[TTM_PL_VRAM] = "vram",
+		[TTM_PL_TT] = "gtt",
+		[TTM_PL_SYSTEM] = "cpu",
+		[AMDGPU_PL_GDS] = "gds",
+		[AMDGPU_PL_GWS] = "gws",
+		[AMDGPU_PL_OA] = "oa",
+		[AMDGPU_PL_DOORBELL] = "doorbell",
+		[AMDGPU_PL_MMIO_REMAP] = "mmioremap",
+	};
+	unsigned int hw_ip, i;
+
+	amdgpu_vm_get_memory(vm, stats);
+	amdgpu_ctx_mgr_usage(&fpriv->ctx_mgr, usage);
+
+	/*
+	 * ******************************************************************
+	 * For text output format description please see drm-usage-stats.rst!
+	 * ******************************************************************
+	 */
+
+	drm_printf(p, "pasid:\t%u\n", fpriv->vm.pasid);
+
+	for (i = 0; i < ARRAY_SIZE(pl_name); i++) {
+		if (!pl_name[i])
+			continue;
+
+		drm_print_memory_stats(p,
+				       &stats[i].drm,
+				       DRM_GEM_OBJECT_RESIDENT |
+				       DRM_GEM_OBJECT_PURGEABLE,
+				       pl_name[i]);
+	}
+
+	/* Legacy amdgpu keys, alias to drm-resident-memory-: */
+	drm_printf(p, "drm-memory-vram:\t%llu KiB\n",
+		   stats[TTM_PL_VRAM].drm.resident/1024UL);
+	drm_printf(p, "drm-memory-gtt: \t%llu KiB\n",
+		   stats[TTM_PL_TT].drm.resident/1024UL);
+	drm_printf(p, "drm-memory-cpu: \t%llu KiB\n",
+		   stats[TTM_PL_SYSTEM].drm.resident/1024UL);
+
+	/* Amdgpu specific memory accounting keys: */
+	drm_printf(p, "amd-evicted-vram:\t%llu KiB\n",
+		   stats[TTM_PL_VRAM].evicted/1024UL);
+	drm_printf(p, "amd-requested-vram:\t%llu KiB\n",
+		   (stats[TTM_PL_VRAM].drm.shared +
+		    stats[TTM_PL_VRAM].drm.private) / 1024UL);
+	drm_printf(p, "amd-requested-gtt:\t%llu KiB\n",
+		   (stats[TTM_PL_TT].drm.shared +
+		    stats[TTM_PL_TT].drm.private) / 1024UL);
+
+	for (hw_ip = 0; hw_ip < AMDGPU_HW_IP_NUM; ++hw_ip) {
+		if (!usage[hw_ip])
+			continue;
+
+		drm_printf(p, "drm-engine-%s:\t%lld ns\n", amdgpu_ip_name[hw_ip],
+			   ktime_to_ns(usage[hw_ip]));
+	}
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fdinfo.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_fdinfo.h
new file mode 100644
index 000000000000..0398f5a159ef
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fdinfo.h
@@ -0,0 +1,42 @@
+/* SPDX-License-Identifier: MIT
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: David Nieto
+ *          Roy Sun
+ */
+#ifndef __AMDGPU_SMI_H__
+#define __AMDGPU_SMI_H__
+
+#include <linux/idr.h>
+#include <linux/kfifo.h>
+#include <linux/rbtree.h>
+#include <drm/gpu_scheduler.h>
+#include <drm/drm_file.h>
+#include <linux/sched/mm.h>
+
+#include "amdgpu_sync.h"
+#include "amdgpu_ring.h"
+#include "amdgpu_ids.h"
+
+uint32_t amdgpu_get_ip_count(struct amdgpu_device *adev, int id);
+void amdgpu_show_fdinfo(struct drm_printer *p, struct drm_file *file);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
index 333bad749067..fd8cca241da6 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
@@ -34,52 +34,24 @@
 #include <linux/kref.h>
 #include <linux/slab.h>
 #include <linux/firmware.h>
-#include <drm/drmP.h>
+#include <linux/pm_runtime.h>
+
+#include <drm/drm_drv.h>
 #include "amdgpu.h"
 #include "amdgpu_trace.h"
+#include "amdgpu_reset.h"
 
 /*
- * Fences
- * Fences mark an event in the GPUs pipeline and are used
- * for GPU/CPU synchronization.  When the fence is written,
- * it is expected that all buffers associated with that fence
- * are no longer in use by the associated ring on the GPU and
- * that the the relevant GPU caches have been flushed.
- */
-
-struct amdgpu_fence {
-	struct dma_fence base;
-
-	/* RB, DMA, etc. */
-	struct amdgpu_ring		*ring;
-};
-
-static struct kmem_cache *amdgpu_fence_slab;
-
-int amdgpu_fence_slab_init(void)
-{
-	amdgpu_fence_slab = kmem_cache_create(
-		"amdgpu_fence", sizeof(struct amdgpu_fence), 0,
-		SLAB_HWCACHE_ALIGN, NULL);
-	if (!amdgpu_fence_slab)
-		return -ENOMEM;
-	return 0;
-}
-
-void amdgpu_fence_slab_fini(void)
-{
-	rcu_barrier();
-	kmem_cache_destroy(amdgpu_fence_slab);
-}
-/*
  * Cast helper
  */
 static const struct dma_fence_ops amdgpu_fence_ops;
+static const struct dma_fence_ops amdgpu_job_fence_ops;
 static inline struct amdgpu_fence *to_amdgpu_fence(struct dma_fence *f)
 {
 	struct amdgpu_fence *__f = container_of(f, struct amdgpu_fence, base);
 
-	if (__f->base.ops == &amdgpu_fence_ops)
+	if (__f->base.ops == &amdgpu_fence_ops ||
+	    __f->base.ops == &amdgpu_job_fence_ops)
 		return __f;
 
 	return NULL;
@@ -127,43 +99,110 @@ static u32 amdgpu_fence_read(struct amdgpu_ring *ring)
  *
  * @ring: ring the fence is associated with
  * @f: resulting fence object
+ * @af: amdgpu fence input
+ * @flags: flags to pass into the subordinate .emit_fence() call
  *
  * Emits a fence command on the requested ring (all asics).
  * Returns 0 on success, -ENOMEM on failure.
  */
-int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **f)
+int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **f,
+		      struct amdgpu_fence *af, unsigned int flags)
 {
 	struct amdgpu_device *adev = ring->adev;
-	struct amdgpu_fence *fence;
-	struct dma_fence *old, **ptr;
+	struct dma_fence *fence;
+	struct amdgpu_fence *am_fence;
+	struct dma_fence __rcu **ptr;
 	uint32_t seq;
+	int r;
 
-	fence = kmem_cache_alloc(amdgpu_fence_slab, GFP_KERNEL);
-	if (fence == NULL)
-		return -ENOMEM;
+	if (!af) {
+		/* create a separate hw fence */
+		am_fence = kzalloc(sizeof(*am_fence), GFP_KERNEL);
+		if (!am_fence)
+			return -ENOMEM;
+	} else {
+		am_fence = af;
+	}
+	fence = &am_fence->base;
+	am_fence->ring = ring;
 
 	seq = ++ring->fence_drv.sync_seq;
-	fence->ring = ring;
-	dma_fence_init(&fence->base, &amdgpu_fence_ops,
-		       &ring->fence_drv.lock,
-		       adev->fence_context + ring->idx,
-		       seq);
-	amdgpu_ring_emit_fence(ring, ring->fence_drv.gpu_addr,
-			       seq, AMDGPU_FENCE_FLAG_INT);
+	am_fence->seq = seq;
+	if (af) {
+		dma_fence_init(fence, &amdgpu_job_fence_ops,
+			       &ring->fence_drv.lock,
+			       adev->fence_context + ring->idx, seq);
+		/* Against remove in amdgpu_job_{free, free_cb} */
+		dma_fence_get(fence);
+	} else {
+		dma_fence_init(fence, &amdgpu_fence_ops,
+			       &ring->fence_drv.lock,
+			       adev->fence_context + ring->idx, seq);
+	}
 
+	amdgpu_ring_emit_fence(ring, ring->fence_drv.gpu_addr,
+			       seq, flags | AMDGPU_FENCE_FLAG_INT);
+	amdgpu_fence_save_wptr(fence);
+	pm_runtime_get_noresume(adev_to_drm(adev)->dev);
 	ptr = &ring->fence_drv.fences[seq & ring->fence_drv.num_fences_mask];
+	if (unlikely(rcu_dereference_protected(*ptr, 1))) {
+		struct dma_fence *old;
+
+		rcu_read_lock();
+		old = dma_fence_get_rcu_safe(ptr);
+		rcu_read_unlock();
+
+		if (old) {
+			r = dma_fence_wait(old, false);
+			dma_fence_put(old);
+			if (r)
+				return r;
+		}
+	}
+
+	to_amdgpu_fence(fence)->start_timestamp = ktime_get();
+
 	/* This function can't be called concurrently anyway, otherwise
 	 * emitting the fence would mess up the hardware ring buffer.
 	 */
-	old = rcu_dereference_protected(*ptr, 1);
-	if (old && !dma_fence_is_signaled(old)) {
-		DRM_INFO("rcu slot is busy\n");
-		dma_fence_wait(old, false);
-	}
+	rcu_assign_pointer(*ptr, dma_fence_get(fence));
 
-	rcu_assign_pointer(*ptr, dma_fence_get(&fence->base));
+	*f = fence;
 
-	*f = &fence->base;
+	return 0;
+}
+
+/**
+ * amdgpu_fence_emit_polling - emit a fence on the requeste ring
+ *
+ * @ring: ring the fence is associated with
+ * @s: resulting sequence number
+ * @timeout: the timeout for waiting in usecs
+ *
+ * Emits a fence command on the requested ring (all asics).
+ * Used For polling fence.
+ * Returns 0 on success, -ENOMEM on failure.
+ */
+int amdgpu_fence_emit_polling(struct amdgpu_ring *ring, uint32_t *s,
+			      uint32_t timeout)
+{
+	uint32_t seq;
+	signed long r;
+
+	if (!s)
+		return -EINVAL;
+
+	seq = ++ring->fence_drv.sync_seq;
+	r = amdgpu_fence_wait_polling(ring,
+				      seq - ring->fence_drv.num_fences_mask,
+				      timeout);
+	if (r < 1)
+		return -ETIMEDOUT;
+
+	amdgpu_ring_emit_fence(ring, ring->fence_drv.gpu_addr,
+			       seq, 0);
+
+	*s = seq;
 
 	return 0;
 }
@@ -189,12 +228,14 @@ static void amdgpu_fence_schedule_fallback(struct amdgpu_ring *ring)
  * Checks the current fence value and calculates the last
  * signalled fence value. Wakes the fence queue if the
  * sequence number has increased.
+ *
+ * Returns true if fence was processed
  */
-void amdgpu_fence_process(struct amdgpu_ring *ring)
+bool amdgpu_fence_process(struct amdgpu_ring *ring)
 {
 	struct amdgpu_fence_driver *drv = &ring->fence_drv;
+	struct amdgpu_device *adev = ring->adev;
 	uint32_t seq, last_seq;
-	int r;
 
 	do {
 		last_seq = atomic_read(&ring->fence_drv.last_seq);
@@ -202,17 +243,19 @@ void amdgpu_fence_process(struct amdgpu_ring *ring)
 
 	} while (atomic_cmpxchg(&drv->last_seq, last_seq, seq) != last_seq);
 
-	if (seq != ring->fence_drv.sync_seq)
+	if (timer_delete(&ring->fence_drv.fallback_timer) &&
+	    seq != ring->fence_drv.sync_seq)
 		amdgpu_fence_schedule_fallback(ring);
 
 	if (unlikely(seq == last_seq))
-		return;
+		return false;
 
 	last_seq &= drv->num_fences_mask;
 	seq &= drv->num_fences_mask;
 
 	do {
 		struct dma_fence *fence, **ptr;
+		struct amdgpu_fence *am_fence;
 
 		++last_seq;
 		last_seq &= drv->num_fences_mask;
@@ -225,34 +268,42 @@ void amdgpu_fence_process(struct amdgpu_ring *ring)
 		if (!fence)
 			continue;
 
-		r = dma_fence_signal(fence);
-		if (!r)
-			DMA_FENCE_TRACE(fence, "signaled from irq context\n");
-		else
-			BUG();
-
+		/* Save the wptr in the fence driver so we know what the last processed
+		 * wptr was.  This is required for re-emitting the ring state for
+		 * queues that are reset but are not guilty and thus have no guilty fence.
+		 */
+		am_fence = container_of(fence, struct amdgpu_fence, base);
+		drv->signalled_wptr = am_fence->wptr;
+		dma_fence_signal(fence);
 		dma_fence_put(fence);
+		pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 	} while (last_seq != seq);
+
+	return true;
 }
 
 /**
  * amdgpu_fence_fallback - fallback for hardware interrupts
  *
- * @work: delayed work item
+ * @t: timer context used to obtain the pointer to ring structure
  *
  * Checks for fence activity.
  */
-static void amdgpu_fence_fallback(unsigned long arg)
+static void amdgpu_fence_fallback(struct timer_list *t)
 {
-	struct amdgpu_ring *ring = (void *)arg;
+	struct amdgpu_ring *ring = timer_container_of(ring, t,
+						      fence_drv.fallback_timer);
 
-	amdgpu_fence_process(ring);
+	if (amdgpu_fence_process(ring))
+		dev_warn(ring->adev->dev,
+			 "Fence fallback timer expired on ring %s\n",
+			 ring->name);
 }
 
 /**
  * amdgpu_fence_wait_empty - wait for all fences to signal
  *
- * @adev: amdgpu device pointer
  * @ring: ring index the fence is associated with
  *
  * Wait for all fences on the requested ring to signal (all asics).
@@ -260,7 +311,7 @@ static void amdgpu_fence_fallback(unsigned long arg)
  */
 int amdgpu_fence_wait_empty(struct amdgpu_ring *ring)
 {
-	uint64_t seq = ACCESS_ONCE(ring->fence_drv.sync_seq);
+	uint64_t seq = READ_ONCE(ring->fence_drv.sync_seq);
 	struct dma_fence *fence, **ptr;
 	int r;
 
@@ -282,6 +333,27 @@ int amdgpu_fence_wait_empty(struct amdgpu_ring *ring)
 }
 
 /**
+ * amdgpu_fence_wait_polling - busy wait for givn sequence number
+ *
+ * @ring: ring index the fence is associated with
+ * @wait_seq: sequence number to wait
+ * @timeout: the timeout for waiting in usecs
+ *
+ * Wait for all fences on the requested ring to signal (all asics).
+ * Returns left time if no timeout, 0 or minus if timeout.
+ */
+signed long amdgpu_fence_wait_polling(struct amdgpu_ring *ring,
+				      uint32_t wait_seq,
+				      signed long timeout)
+{
+
+	while ((int32_t)(wait_seq - amdgpu_fence_read(ring)) > 0 && timeout > 0) {
+		udelay(2);
+		timeout -= 2;
+	}
+	return timeout > 0 ? timeout : 0;
+}
+/**
  * amdgpu_fence_count_emitted - get the count of emitted fences
  *
  * @ring: ring the fence is associated with
@@ -290,21 +362,71 @@ int amdgpu_fence_wait_empty(struct amdgpu_ring *ring)
  * Returns the number of emitted fences on the ring.  Used by the
  * dynpm code to ring track activity.
  */
-unsigned amdgpu_fence_count_emitted(struct amdgpu_ring *ring)
+unsigned int amdgpu_fence_count_emitted(struct amdgpu_ring *ring)
 {
 	uint64_t emitted;
 
 	/* We are not protected by ring lock when reading the last sequence
 	 * but it's ok to report slightly wrong fence count here.
 	 */
-	amdgpu_fence_process(ring);
 	emitted = 0x100000000ull;
 	emitted -= atomic_read(&ring->fence_drv.last_seq);
-	emitted += ACCESS_ONCE(ring->fence_drv.sync_seq);
+	emitted += READ_ONCE(ring->fence_drv.sync_seq);
 	return lower_32_bits(emitted);
 }
 
 /**
+ * amdgpu_fence_last_unsignaled_time_us - the time fence emitted until now
+ * @ring: ring the fence is associated with
+ *
+ * Find the earliest fence unsignaled until now, calculate the time delta
+ * between the time fence emitted and now.
+ */
+u64 amdgpu_fence_last_unsignaled_time_us(struct amdgpu_ring *ring)
+{
+	struct amdgpu_fence_driver *drv = &ring->fence_drv;
+	struct dma_fence *fence;
+	uint32_t last_seq, sync_seq;
+
+	last_seq = atomic_read(&ring->fence_drv.last_seq);
+	sync_seq = READ_ONCE(ring->fence_drv.sync_seq);
+	if (last_seq == sync_seq)
+		return 0;
+
+	++last_seq;
+	last_seq &= drv->num_fences_mask;
+	fence = drv->fences[last_seq];
+	if (!fence)
+		return 0;
+
+	return ktime_us_delta(ktime_get(),
+		to_amdgpu_fence(fence)->start_timestamp);
+}
+
+/**
+ * amdgpu_fence_update_start_timestamp - update the timestamp of the fence
+ * @ring: ring the fence is associated with
+ * @seq: the fence seq number to update.
+ * @timestamp: the start timestamp to update.
+ *
+ * The function called at the time the fence and related ib is about to
+ * resubmit to gpu in MCBP scenario. Thus we do not consider race condition
+ * with amdgpu_fence_process to modify the same fence.
+ */
+void amdgpu_fence_update_start_timestamp(struct amdgpu_ring *ring, uint32_t seq, ktime_t timestamp)
+{
+	struct amdgpu_fence_driver *drv = &ring->fence_drv;
+	struct dma_fence *fence;
+
+	seq &= drv->num_fences_mask;
+	fence = drv->fences[seq];
+	if (!fence)
+		return;
+
+	to_amdgpu_fence(fence)->start_timestamp = timestamp;
+}
+
+/**
  * amdgpu_fence_driver_start_ring - make the fence driver
  * ready for use on the requested ring.
  *
@@ -319,30 +441,28 @@ unsigned amdgpu_fence_count_emitted(struct amdgpu_ring *ring)
  */
 int amdgpu_fence_driver_start_ring(struct amdgpu_ring *ring,
 				   struct amdgpu_irq_src *irq_src,
-				   unsigned irq_type)
+				   unsigned int irq_type)
 {
 	struct amdgpu_device *adev = ring->adev;
 	uint64_t index;
 
-	if (ring != &adev->uvd.ring) {
-		ring->fence_drv.cpu_addr = &adev->wb.wb[ring->fence_offs];
-		ring->fence_drv.gpu_addr = adev->wb.gpu_addr + (ring->fence_offs * 4);
+	if (ring->funcs->type != AMDGPU_RING_TYPE_UVD) {
+		ring->fence_drv.cpu_addr = ring->fence_cpu_addr;
+		ring->fence_drv.gpu_addr = ring->fence_gpu_addr;
 	} else {
 		/* put fence directly behind firmware */
 		index = ALIGN(adev->uvd.fw->size, 8);
-		ring->fence_drv.cpu_addr = adev->uvd.cpu_addr + index;
-		ring->fence_drv.gpu_addr = adev->uvd.gpu_addr + index;
+		ring->fence_drv.cpu_addr = adev->uvd.inst[ring->me].cpu_addr + index;
+		ring->fence_drv.gpu_addr = adev->uvd.inst[ring->me].gpu_addr + index;
 	}
 	amdgpu_fence_write(ring, atomic_read(&ring->fence_drv.last_seq));
-	amdgpu_irq_get(adev, irq_src, irq_type);
 
 	ring->fence_drv.irq_src = irq_src;
 	ring->fence_drv.irq_type = irq_type;
 	ring->fence_drv.initialized = true;
 
-	dev_info(adev->dev, "fence driver on ring %d use gpu addr 0x%016llx, "
-		 "cpu addr 0x%p\n", ring->idx,
-		 ring->fence_drv.gpu_addr, ring->fence_drv.cpu_addr);
+	DRM_DEV_DEBUG(adev->dev, "fence driver on ring %s use gpu addr 0x%016llx\n",
+		      ring->name, ring->fence_drv.gpu_addr);
 	return 0;
 }
 
@@ -351,19 +471,18 @@ int amdgpu_fence_driver_start_ring(struct amdgpu_ring *ring,
  * for the requested ring.
  *
  * @ring: ring to init the fence driver on
- * @num_hw_submission: number of entries on the hardware queue
  *
  * Init the fence driver for the requested ring (all asics).
  * Helper function for amdgpu_fence_driver_init().
  */
-int amdgpu_fence_driver_init_ring(struct amdgpu_ring *ring,
-				  unsigned num_hw_submission)
+int amdgpu_fence_driver_init_ring(struct amdgpu_ring *ring)
 {
-	long timeout;
-	int r;
+	struct amdgpu_device *adev = ring->adev;
+
+	if (!adev)
+		return -EINVAL;
 
-	/* Check that num_hw_submission is a power of two */
-	if ((num_hw_submission & (num_hw_submission - 1)) != 0)
+	if (!is_power_of_2(ring->num_hw_submission))
 		return -EINVAL;
 
 	ring->fence_drv.cpu_addr = NULL;
@@ -372,44 +491,21 @@ int amdgpu_fence_driver_init_ring(struct amdgpu_ring *ring,
 	atomic_set(&ring->fence_drv.last_seq, 0);
 	ring->fence_drv.initialized = false;
 
-	setup_timer(&ring->fence_drv.fallback_timer, amdgpu_fence_fallback,
-		    (unsigned long)ring);
+	timer_setup(&ring->fence_drv.fallback_timer, amdgpu_fence_fallback, 0);
 
-	ring->fence_drv.num_fences_mask = num_hw_submission * 2 - 1;
+	ring->fence_drv.num_fences_mask = ring->num_hw_submission * 2 - 1;
 	spin_lock_init(&ring->fence_drv.lock);
-	ring->fence_drv.fences = kcalloc(num_hw_submission * 2, sizeof(void *),
+	ring->fence_drv.fences = kcalloc(ring->num_hw_submission * 2, sizeof(void *),
 					 GFP_KERNEL);
+
 	if (!ring->fence_drv.fences)
 		return -ENOMEM;
 
-	/* No need to setup the GPU scheduler for KIQ ring */
-	if (ring->funcs->type != AMDGPU_RING_TYPE_KIQ) {
-		timeout = msecs_to_jiffies(amdgpu_lockup_timeout);
-		if (timeout == 0) {
-			/*
-			 * FIXME:
-			 * Delayed workqueue cannot use it directly,
-			 * so the scheduler will not use delayed workqueue if
-			 * MAX_SCHEDULE_TIMEOUT is set.
-			 * Currently keep it simple and silly.
-			 */
-			timeout = MAX_SCHEDULE_TIMEOUT;
-		}
-		r = amd_sched_init(&ring->sched, &amdgpu_sched_ops,
-				   num_hw_submission,
-				   timeout, ring->name);
-		if (r) {
-			DRM_ERROR("Failed to create scheduler on ring %s.\n",
-				  ring->name);
-			return r;
-		}
-	}
-
 	return 0;
 }
 
 /**
- * amdgpu_fence_driver_init - init the fence driver
+ * amdgpu_fence_driver_sw_init - init the fence driver
  * for all possible rings.
  *
  * @adev: amdgpu device pointer
@@ -420,131 +516,304 @@ int amdgpu_fence_driver_init_ring(struct amdgpu_ring *ring,
  * amdgpu_fence_driver_start_ring().
  * Returns 0 for success.
  */
-int amdgpu_fence_driver_init(struct amdgpu_device *adev)
+int amdgpu_fence_driver_sw_init(struct amdgpu_device *adev)
 {
-	if (amdgpu_debugfs_fence_init(adev))
-		dev_err(adev->dev, "fence debugfs file creation failed\n");
-
 	return 0;
 }
 
 /**
- * amdgpu_fence_driver_fini - tear down the fence driver
+ * amdgpu_fence_need_ring_interrupt_restore - helper function to check whether
+ * fence driver interrupts need to be restored.
+ *
+ * @ring: ring that to be checked
+ *
+ * Interrupts for rings that belong to GFX IP don't need to be restored
+ * when the target power state is s0ix.
+ *
+ * Return true if need to restore interrupts, false otherwise.
+ */
+static bool amdgpu_fence_need_ring_interrupt_restore(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	bool is_gfx_power_domain = false;
+
+	switch (ring->funcs->type) {
+	case AMDGPU_RING_TYPE_SDMA:
+	/* SDMA 5.x+ is part of GFX power domain so it's covered by GFXOFF */
+		if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) >=
+		    IP_VERSION(5, 0, 0))
+			is_gfx_power_domain = true;
+		break;
+	case AMDGPU_RING_TYPE_GFX:
+	case AMDGPU_RING_TYPE_COMPUTE:
+	case AMDGPU_RING_TYPE_KIQ:
+	case AMDGPU_RING_TYPE_MES:
+		is_gfx_power_domain = true;
+		break;
+	default:
+		break;
+	}
+
+	return !(adev->in_s0ix && is_gfx_power_domain);
+}
+
+/**
+ * amdgpu_fence_driver_hw_fini - tear down the fence driver
  * for all possible rings.
  *
  * @adev: amdgpu device pointer
  *
  * Tear down the fence driver for all possible rings (all asics).
  */
-void amdgpu_fence_driver_fini(struct amdgpu_device *adev)
+void amdgpu_fence_driver_hw_fini(struct amdgpu_device *adev)
 {
-	unsigned i, j;
-	int r;
+	int i, r;
 
 	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
 		struct amdgpu_ring *ring = adev->rings[i];
 
 		if (!ring || !ring->fence_drv.initialized)
 			continue;
-		r = amdgpu_fence_wait_empty(ring);
-		if (r) {
-			/* no need to trigger GPU reset as we are unloading */
-			amdgpu_fence_driver_force_completion(adev);
-		}
-		amdgpu_irq_put(adev, ring->fence_drv.irq_src,
-			       ring->fence_drv.irq_type);
-		amd_sched_fini(&ring->sched);
-		del_timer_sync(&ring->fence_drv.fallback_timer);
-		for (j = 0; j <= ring->fence_drv.num_fences_mask; ++j)
-			dma_fence_put(ring->fence_drv.fences[j]);
-		kfree(ring->fence_drv.fences);
-		ring->fence_drv.fences = NULL;
-		ring->fence_drv.initialized = false;
+
+		/* You can't wait for HW to signal if it's gone */
+		if (!drm_dev_is_unplugged(adev_to_drm(adev)))
+			r = amdgpu_fence_wait_empty(ring);
+		else
+			r = -ENODEV;
+		/* no need to trigger GPU reset as we are unloading */
+		if (r)
+			amdgpu_fence_driver_force_completion(ring);
+
+		if (!drm_dev_is_unplugged(adev_to_drm(adev)) &&
+		    ring->fence_drv.irq_src &&
+		    amdgpu_fence_need_ring_interrupt_restore(ring))
+			amdgpu_irq_put(adev, ring->fence_drv.irq_src,
+				       ring->fence_drv.irq_type);
+
+		timer_delete_sync(&ring->fence_drv.fallback_timer);
 	}
 }
 
-/**
- * amdgpu_fence_driver_suspend - suspend the fence driver
- * for all possible rings.
- *
- * @adev: amdgpu device pointer
- *
- * Suspend the fence driver for all possible rings (all asics).
- */
-void amdgpu_fence_driver_suspend(struct amdgpu_device *adev)
+/* Will either stop and flush handlers for amdgpu interrupt or reanble it */
+void amdgpu_fence_driver_isr_toggle(struct amdgpu_device *adev, bool stop)
 {
-	int i, r;
+	int i;
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
+		struct amdgpu_ring *ring = adev->rings[i];
+
+		if (!ring || !ring->fence_drv.initialized || !ring->fence_drv.irq_src)
+			continue;
+
+		if (stop)
+			disable_irq(adev->irq.irq);
+		else
+			enable_irq(adev->irq.irq);
+	}
+}
+
+void amdgpu_fence_driver_sw_fini(struct amdgpu_device *adev)
+{
+	unsigned int i, j;
 
 	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
 		struct amdgpu_ring *ring = adev->rings[i];
+
 		if (!ring || !ring->fence_drv.initialized)
 			continue;
 
-		/* wait for gpu to finish processing current batch */
-		r = amdgpu_fence_wait_empty(ring);
-		if (r) {
-			/* delay GPU reset to resume */
-			amdgpu_fence_driver_force_completion(adev);
-		}
+		/*
+		 * Notice we check for sched.ops since there's some
+		 * override on the meaning of sched.ready by amdgpu.
+		 * The natural check would be sched.ready, which is
+		 * set as drm_sched_init() finishes...
+		 */
+		if (ring->sched.ops)
+			drm_sched_fini(&ring->sched);
 
-		/* disable the interrupt */
-		amdgpu_irq_put(adev, ring->fence_drv.irq_src,
-			       ring->fence_drv.irq_type);
+		for (j = 0; j <= ring->fence_drv.num_fences_mask; ++j)
+			dma_fence_put(ring->fence_drv.fences[j]);
+		kfree(ring->fence_drv.fences);
+		ring->fence_drv.fences = NULL;
+		ring->fence_drv.initialized = false;
 	}
 }
 
 /**
- * amdgpu_fence_driver_resume - resume the fence driver
+ * amdgpu_fence_driver_hw_init - enable the fence driver
  * for all possible rings.
  *
  * @adev: amdgpu device pointer
  *
- * Resume the fence driver for all possible rings (all asics).
+ * Enable the fence driver for all possible rings (all asics).
  * Not all asics have all rings, so each asic will only
  * start the fence driver on the rings it has using
  * amdgpu_fence_driver_start_ring().
  * Returns 0 for success.
  */
-void amdgpu_fence_driver_resume(struct amdgpu_device *adev)
+void amdgpu_fence_driver_hw_init(struct amdgpu_device *adev)
 {
 	int i;
 
 	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
 		struct amdgpu_ring *ring = adev->rings[i];
+
 		if (!ring || !ring->fence_drv.initialized)
 			continue;
 
 		/* enable the interrupt */
-		amdgpu_irq_get(adev, ring->fence_drv.irq_src,
-			       ring->fence_drv.irq_type);
+		if (ring->fence_drv.irq_src &&
+		    amdgpu_fence_need_ring_interrupt_restore(ring))
+			amdgpu_irq_get(adev, ring->fence_drv.irq_src,
+				       ring->fence_drv.irq_type);
 	}
 }
 
 /**
- * amdgpu_fence_driver_force_completion - force all fence waiter to complete
+ * amdgpu_fence_driver_clear_job_fences - clear job embedded fences of ring
  *
- * @adev: amdgpu device pointer
+ * @ring: fence of the ring to be cleared
  *
- * In case of GPU reset failure make sure no process keep waiting on fence
- * that will never complete.
  */
-void amdgpu_fence_driver_force_completion(struct amdgpu_device *adev)
+void amdgpu_fence_driver_clear_job_fences(struct amdgpu_ring *ring)
 {
 	int i;
+	struct dma_fence *old, **ptr;
 
-	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
-		struct amdgpu_ring *ring = adev->rings[i];
-		if (!ring || !ring->fence_drv.initialized)
-			continue;
+	for (i = 0; i <= ring->fence_drv.num_fences_mask; i++) {
+		ptr = &ring->fence_drv.fences[i];
+		old = rcu_dereference_protected(*ptr, 1);
+		if (old && old->ops == &amdgpu_job_fence_ops) {
+			struct amdgpu_job *job;
 
-		amdgpu_fence_write(ring, ring->fence_drv.sync_seq);
+			/* For non-scheduler bad job, i.e. failed ib test, we need to signal
+			 * it right here or we won't be able to track them in fence_drv
+			 * and they will remain unsignaled during sa_bo free.
+			 */
+			job = container_of(old, struct amdgpu_job, hw_fence.base);
+			if (!job->base.s_fence && !dma_fence_is_signaled(old))
+				dma_fence_signal(old);
+			RCU_INIT_POINTER(*ptr, NULL);
+			dma_fence_put(old);
+		}
 	}
 }
 
-void amdgpu_fence_driver_force_completion_ring(struct amdgpu_ring *ring)
+/**
+ * amdgpu_fence_driver_set_error - set error code on fences
+ * @ring: the ring which contains the fences
+ * @error: the error code to set
+ *
+ * Set an error code to all the fences pending on the ring.
+ */
+void amdgpu_fence_driver_set_error(struct amdgpu_ring *ring, int error)
 {
-	if (ring)
-		amdgpu_fence_write(ring, ring->fence_drv.sync_seq);
+	struct amdgpu_fence_driver *drv = &ring->fence_drv;
+	unsigned long flags;
+
+	spin_lock_irqsave(&drv->lock, flags);
+	for (unsigned int i = 0; i <= drv->num_fences_mask; ++i) {
+		struct dma_fence *fence;
+
+		fence = rcu_dereference_protected(drv->fences[i],
+						  lockdep_is_held(&drv->lock));
+		if (fence && !dma_fence_is_signaled_locked(fence))
+			dma_fence_set_error(fence, error);
+	}
+	spin_unlock_irqrestore(&drv->lock, flags);
+}
+
+/**
+ * amdgpu_fence_driver_force_completion - force signal latest fence of ring
+ *
+ * @ring: fence of the ring to signal
+ *
+ */
+void amdgpu_fence_driver_force_completion(struct amdgpu_ring *ring)
+{
+	amdgpu_fence_driver_set_error(ring, -ECANCELED);
+	amdgpu_fence_write(ring, ring->fence_drv.sync_seq);
+	amdgpu_fence_process(ring);
+}
+
+
+/*
+ * Kernel queue reset handling
+ *
+ * The driver can reset individual queues for most engines, but those queues
+ * may contain work from multiple contexts.  Resetting the queue will reset
+ * lose all of that state.  In order to minimize the collateral damage, the
+ * driver will save the ring contents which are not associated with the guilty
+ * context prior to resetting the queue.  After resetting the queue the queue
+ * contents from the other contexts is re-emitted to the rings so that it can
+ * be processed by the engine.  To handle this, we save the queue's write
+ * pointer (wptr) in the fences associated with each context.  If we get a
+ * queue timeout, we can then use the wptrs from the fences to determine
+ * which data needs to be saved out of the queue's ring buffer.
+ */
+
+/**
+ * amdgpu_fence_driver_guilty_force_completion - force signal of specified sequence
+ *
+ * @fence: fence of the ring to signal
+ *
+ */
+void amdgpu_fence_driver_guilty_force_completion(struct amdgpu_fence *fence)
+{
+	dma_fence_set_error(&fence->base, -ETIME);
+	amdgpu_fence_write(fence->ring, fence->seq);
+	amdgpu_fence_process(fence->ring);
+}
+
+void amdgpu_fence_save_wptr(struct dma_fence *fence)
+{
+	struct amdgpu_fence *am_fence = container_of(fence, struct amdgpu_fence, base);
+
+	am_fence->wptr = am_fence->ring->wptr;
+}
+
+static void amdgpu_ring_backup_unprocessed_command(struct amdgpu_ring *ring,
+						   u64 start_wptr, u32 end_wptr)
+{
+	unsigned int first_idx = start_wptr & ring->buf_mask;
+	unsigned int last_idx = end_wptr & ring->buf_mask;
+	unsigned int i;
+
+	/* Backup the contents of the ring buffer. */
+	for (i = first_idx; i != last_idx; ++i, i &= ring->buf_mask)
+		ring->ring_backup[ring->ring_backup_entries_to_copy++] = ring->ring[i];
+}
+
+void amdgpu_ring_backup_unprocessed_commands(struct amdgpu_ring *ring,
+					     struct amdgpu_fence *guilty_fence)
+{
+	struct dma_fence *unprocessed;
+	struct dma_fence __rcu **ptr;
+	struct amdgpu_fence *fence;
+	u64 wptr, i, seqno;
+
+	seqno = amdgpu_fence_read(ring);
+	wptr = ring->fence_drv.signalled_wptr;
+	ring->ring_backup_entries_to_copy = 0;
+
+	for (i = seqno + 1; i <= ring->fence_drv.sync_seq; ++i) {
+		ptr = &ring->fence_drv.fences[i & ring->fence_drv.num_fences_mask];
+		rcu_read_lock();
+		unprocessed = rcu_dereference(*ptr);
+
+		if (unprocessed && !dma_fence_is_signaled(unprocessed)) {
+			fence = container_of(unprocessed, struct amdgpu_fence, base);
+
+			/* save everything if the ring is not guilty, otherwise
+			 * just save the content from other contexts.
+			 */
+			if (!guilty_fence || (fence->context != guilty_fence->context))
+				amdgpu_ring_backup_unprocessed_command(ring, wptr,
+								       fence->wptr);
+			wptr = fence->wptr;
+		}
+		rcu_read_unlock();
+	}
 }
 
 /*
@@ -558,13 +827,19 @@ static const char *amdgpu_fence_get_driver_name(struct dma_fence *fence)
 
 static const char *amdgpu_fence_get_timeline_name(struct dma_fence *f)
 {
-	struct amdgpu_fence *fence = to_amdgpu_fence(f);
-	return (const char *)fence->ring->name;
+	return (const char *)to_amdgpu_fence(f)->ring->name;
+}
+
+static const char *amdgpu_job_fence_get_timeline_name(struct dma_fence *f)
+{
+	struct amdgpu_job *job = container_of(f, struct amdgpu_job, hw_fence.base);
+
+	return (const char *)to_amdgpu_ring(job->base.sched)->name;
 }
 
 /**
  * amdgpu_fence_enable_signaling - enable signalling on fence
- * @fence: fence
+ * @f: fence
  *
  * This function is called with fence_queue lock held, and adds a callback
  * to fence_queue that checks if this fence is signaled, and if so it
@@ -572,13 +847,25 @@ static const char *amdgpu_fence_get_timeline_name(struct dma_fence *f)
  */
 static bool amdgpu_fence_enable_signaling(struct dma_fence *f)
 {
-	struct amdgpu_fence *fence = to_amdgpu_fence(f);
-	struct amdgpu_ring *ring = fence->ring;
+	if (!timer_pending(&to_amdgpu_fence(f)->ring->fence_drv.fallback_timer))
+		amdgpu_fence_schedule_fallback(to_amdgpu_fence(f)->ring);
 
-	if (!timer_pending(&ring->fence_drv.fallback_timer))
-		amdgpu_fence_schedule_fallback(ring);
+	return true;
+}
+
+/**
+ * amdgpu_job_fence_enable_signaling - enable signalling on job fence
+ * @f: fence
+ *
+ * This is the simliar function with amdgpu_fence_enable_signaling above, it
+ * only handles the job embedded fence.
+ */
+static bool amdgpu_job_fence_enable_signaling(struct dma_fence *f)
+{
+	struct amdgpu_job *job = container_of(f, struct amdgpu_job, hw_fence.base);
 
-	DMA_FENCE_TRACE(&fence->base, "armed on ring %i!\n", ring->idx);
+	if (!timer_pending(&to_amdgpu_ring(job->base.sched)->fence_drv.fallback_timer))
+		amdgpu_fence_schedule_fallback(to_amdgpu_ring(job->base.sched));
 
 	return true;
 }
@@ -593,14 +880,30 @@ static bool amdgpu_fence_enable_signaling(struct dma_fence *f)
 static void amdgpu_fence_free(struct rcu_head *rcu)
 {
 	struct dma_fence *f = container_of(rcu, struct dma_fence, rcu);
-	struct amdgpu_fence *fence = to_amdgpu_fence(f);
-	kmem_cache_free(amdgpu_fence_slab, fence);
+
+	/* free fence_slab if it's separated fence*/
+	kfree(to_amdgpu_fence(f));
+}
+
+/**
+ * amdgpu_job_fence_free - free up the job with embedded fence
+ *
+ * @rcu: RCU callback head
+ *
+ * Free up the job with embedded fence after the RCU grace period.
+ */
+static void amdgpu_job_fence_free(struct rcu_head *rcu)
+{
+	struct dma_fence *f = container_of(rcu, struct dma_fence, rcu);
+
+	/* free job if fence has a parent job */
+	kfree(container_of(f, struct amdgpu_job, hw_fence.base));
 }
 
 /**
  * amdgpu_fence_release - callback that fence can be freed
  *
- * @fence: fence
+ * @f: fence
  *
  * This function is called when the reference count becomes zero.
  * It just RCU schedules freeing up the fence.
@@ -610,76 +913,147 @@ static void amdgpu_fence_release(struct dma_fence *f)
 	call_rcu(&f->rcu, amdgpu_fence_free);
 }
 
+/**
+ * amdgpu_job_fence_release - callback that job embedded fence can be freed
+ *
+ * @f: fence
+ *
+ * This is the simliar function with amdgpu_fence_release above, it
+ * only handles the job embedded fence.
+ */
+static void amdgpu_job_fence_release(struct dma_fence *f)
+{
+	call_rcu(&f->rcu, amdgpu_job_fence_free);
+}
+
 static const struct dma_fence_ops amdgpu_fence_ops = {
 	.get_driver_name = amdgpu_fence_get_driver_name,
 	.get_timeline_name = amdgpu_fence_get_timeline_name,
 	.enable_signaling = amdgpu_fence_enable_signaling,
-	.wait = dma_fence_default_wait,
 	.release = amdgpu_fence_release,
 };
 
+static const struct dma_fence_ops amdgpu_job_fence_ops = {
+	.get_driver_name = amdgpu_fence_get_driver_name,
+	.get_timeline_name = amdgpu_job_fence_get_timeline_name,
+	.enable_signaling = amdgpu_job_fence_enable_signaling,
+	.release = amdgpu_job_fence_release,
+};
+
 /*
  * Fence debugfs
  */
 #if defined(CONFIG_DEBUG_FS)
-static int amdgpu_debugfs_fence_info(struct seq_file *m, void *data)
+static int amdgpu_debugfs_fence_info_show(struct seq_file *m, void *unused)
 {
-	struct drm_info_node *node = (struct drm_info_node *)m->private;
-	struct drm_device *dev = node->minor->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = m->private;
 	int i;
 
 	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
 		struct amdgpu_ring *ring = adev->rings[i];
+
 		if (!ring || !ring->fence_drv.initialized)
 			continue;
 
 		amdgpu_fence_process(ring);
 
 		seq_printf(m, "--- ring %d (%s) ---\n", i, ring->name);
-		seq_printf(m, "Last signaled fence 0x%08x\n",
+		seq_printf(m, "Last signaled fence          0x%08x\n",
 			   atomic_read(&ring->fence_drv.last_seq));
-		seq_printf(m, "Last emitted        0x%08x\n",
+		seq_printf(m, "Last emitted                 0x%08x\n",
 			   ring->fence_drv.sync_seq);
+
+		if (ring->funcs->type == AMDGPU_RING_TYPE_GFX ||
+		    ring->funcs->type == AMDGPU_RING_TYPE_SDMA) {
+			seq_printf(m, "Last signaled trailing fence 0x%08x\n",
+				   le32_to_cpu(*ring->trail_fence_cpu_addr));
+			seq_printf(m, "Last emitted                 0x%08x\n",
+				   ring->trail_seq);
+		}
+
+		if (ring->funcs->type != AMDGPU_RING_TYPE_GFX)
+			continue;
+
+		/* set in CP_VMID_PREEMPT and preemption occurred */
+		seq_printf(m, "Last preempted               0x%08x\n",
+			   le32_to_cpu(*(ring->fence_drv.cpu_addr + 2)));
+		/* set in CP_VMID_RESET and reset occurred */
+		seq_printf(m, "Last reset                   0x%08x\n",
+			   le32_to_cpu(*(ring->fence_drv.cpu_addr + 4)));
+		/* Both preemption and reset occurred */
+		seq_printf(m, "Last both                    0x%08x\n",
+			   le32_to_cpu(*(ring->fence_drv.cpu_addr + 6)));
 	}
 	return 0;
 }
 
-/**
- * amdgpu_debugfs_gpu_reset - manually trigger a gpu reset
+/*
+ * amdgpu_debugfs_gpu_recover - manually trigger a gpu reset & recover
  *
  * Manually trigger a gpu reset at the next fence wait.
  */
-static int amdgpu_debugfs_gpu_reset(struct seq_file *m, void *data)
+static int gpu_recover_get(void *data, u64 *val)
 {
-	struct drm_info_node *node = (struct drm_info_node *) m->private;
-	struct drm_device *dev = node->minor->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = (struct amdgpu_device *)data;
+	struct drm_device *dev = adev_to_drm(adev);
+	int r;
+
+	r = pm_runtime_get_sync(dev->dev);
+	if (r < 0) {
+		pm_runtime_put_autosuspend(dev->dev);
+		return 0;
+	}
+
+	if (amdgpu_reset_domain_schedule(adev->reset_domain, &adev->reset_work))
+		flush_work(&adev->reset_work);
+
+	*val = atomic_read(&adev->reset_domain->reset_res);
 
-	seq_printf(m, "gpu reset\n");
-	amdgpu_gpu_reset(adev);
+	pm_runtime_mark_last_busy(dev->dev);
+	pm_runtime_put_autosuspend(dev->dev);
 
 	return 0;
 }
 
-static const struct drm_info_list amdgpu_debugfs_fence_list[] = {
-	{"amdgpu_fence_info", &amdgpu_debugfs_fence_info, 0, NULL},
-	{"amdgpu_gpu_reset", &amdgpu_debugfs_gpu_reset, 0, NULL}
-};
+DEFINE_SHOW_ATTRIBUTE(amdgpu_debugfs_fence_info);
+DEFINE_DEBUGFS_ATTRIBUTE(amdgpu_debugfs_gpu_recover_fops, gpu_recover_get, NULL,
+			 "%lld\n");
+
+static void amdgpu_debugfs_reset_work(struct work_struct *work)
+{
+	struct amdgpu_device *adev = container_of(work, struct amdgpu_device,
+						  reset_work);
+
+	struct amdgpu_reset_context reset_context;
+
+	memset(&reset_context, 0, sizeof(reset_context));
+
+	reset_context.method = AMD_RESET_METHOD_NONE;
+	reset_context.reset_req_dev = adev;
+	reset_context.src = AMDGPU_RESET_SRC_USER;
+	set_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
+	set_bit(AMDGPU_SKIP_COREDUMP, &reset_context.flags);
+
+	amdgpu_device_gpu_recover(adev, NULL, &reset_context);
+}
 
-static const struct drm_info_list amdgpu_debugfs_fence_list_sriov[] = {
-	{"amdgpu_fence_info", &amdgpu_debugfs_fence_info, 0, NULL},
-};
 #endif
 
-int amdgpu_debugfs_fence_init(struct amdgpu_device *adev)
+void amdgpu_debugfs_fence_init(struct amdgpu_device *adev)
 {
 #if defined(CONFIG_DEBUG_FS)
-	if (amdgpu_sriov_vf(adev))
-		return amdgpu_debugfs_add_files(adev, amdgpu_debugfs_fence_list_sriov, 1);
-	return amdgpu_debugfs_add_files(adev, amdgpu_debugfs_fence_list, 2);
-#else
-	return 0;
+	struct drm_minor *minor = adev_to_drm(adev)->primary;
+	struct dentry *root = minor->debugfs_root;
+
+	debugfs_create_file("amdgpu_fence_info", 0444, root, adev,
+			    &amdgpu_debugfs_fence_info_fops);
+
+	if (!amdgpu_sriov_vf(adev)) {
+
+		INIT_WORK(&adev->reset_work, amdgpu_debugfs_reset_work);
+		debugfs_create_file("amdgpu_gpu_recover", 0444, root, adev,
+				    &amdgpu_debugfs_gpu_recover_fops);
+	}
 #endif
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c
new file mode 100644
index 000000000000..b0082aa7f3c6
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c
@@ -0,0 +1,408 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/pci.h>
+
+#include "amdgpu.h"
+#include "amdgpu_i2c.h"
+#include "smu_v11_0_i2c.h"
+#include "atom.h"
+#include "amdgpu_fru_eeprom.h"
+#include "amdgpu_eeprom.h"
+
+#define FRU_EEPROM_MADDR_6      0x60000
+#define FRU_EEPROM_MADDR_8      0x80000
+#define FRU_EEPROM_MADDR_INV    0xFFFFF
+
+static bool is_fru_eeprom_supported(struct amdgpu_device *adev, u32 *fru_addr)
+{
+	/* Only server cards have the FRU EEPROM
+	 * TODO: See if we can figure this out dynamically instead of
+	 * having to parse VBIOS versions.
+	 */
+	struct atom_context *atom_ctx = adev->mode_info.atom_context;
+
+	/* The i2c access is blocked on VF
+	 * TODO: Need other way to get the info
+	 * Also, FRU not valid for APU devices.
+	 */
+	if (amdgpu_sriov_vf(adev) || (adev->flags & AMD_IS_APU))
+		return false;
+
+	/* The default I2C EEPROM address of the FRU.
+	 */
+	if (fru_addr)
+		*fru_addr = FRU_EEPROM_MADDR_8;
+
+	/* VBIOS is of the format ###-DXXXYYYY-##. For SKU identification,
+	 * we can use just the "DXXX" portion. If there were more models, we
+	 * could convert the 3 characters to a hex integer and use a switch
+	 * for ease/speed/readability. For now, 2 string comparisons are
+	 * reasonable and not too expensive
+	 */
+	switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) {
+	case IP_VERSION(11, 0, 2):
+		switch (adev->asic_type) {
+		case CHIP_VEGA20:
+			/* D161 and D163 are the VG20 server SKUs */
+			if (atom_ctx && (strnstr(atom_ctx->vbios_pn, "D161",
+						 sizeof(atom_ctx->vbios_pn)) ||
+					 strnstr(atom_ctx->vbios_pn, "D163",
+						 sizeof(atom_ctx->vbios_pn)))) {
+				if (fru_addr)
+					*fru_addr = FRU_EEPROM_MADDR_6;
+				return true;
+			} else {
+				return false;
+			}
+		case CHIP_ARCTURUS:
+		default:
+			return false;
+		}
+	case IP_VERSION(11, 0, 7):
+		if (atom_ctx && strnstr(atom_ctx->vbios_pn, "D603",
+					sizeof(atom_ctx->vbios_pn))) {
+			if (strnstr(atom_ctx->vbios_pn, "D603GLXE",
+				    sizeof(atom_ctx->vbios_pn))) {
+				return false;
+			}
+
+			if (fru_addr)
+				*fru_addr = FRU_EEPROM_MADDR_6;
+			return true;
+
+		} else {
+			return false;
+		}
+	case IP_VERSION(13, 0, 2):
+		/* All Aldebaran SKUs have an FRU */
+		if (atom_ctx && !strnstr(atom_ctx->vbios_pn, "D673",
+					 sizeof(atom_ctx->vbios_pn)))
+			if (fru_addr)
+				*fru_addr = FRU_EEPROM_MADDR_6;
+		return true;
+	case IP_VERSION(13, 0, 6):
+	case IP_VERSION(13, 0, 14):
+			if (fru_addr)
+				*fru_addr = FRU_EEPROM_MADDR_8;
+			return true;
+	case IP_VERSION(13, 0, 12):
+			if (fru_addr)
+				*fru_addr = FRU_EEPROM_MADDR_INV;
+			return true;
+	default:
+		return false;
+	}
+}
+
+int amdgpu_fru_get_product_info(struct amdgpu_device *adev)
+{
+	struct amdgpu_fru_info *fru_info;
+	unsigned char buf[8], *pia;
+	u32 addr, fru_addr;
+	int size, len;
+	u8 csum;
+
+	if (!is_fru_eeprom_supported(adev, &fru_addr))
+		return 0;
+
+	/* FRU data avaialble, but no direct EEPROM access */
+	if (fru_addr == FRU_EEPROM_MADDR_INV)
+		return 0;
+
+	if (!adev->fru_info) {
+		adev->fru_info = kzalloc(sizeof(*adev->fru_info), GFP_KERNEL);
+		if (!adev->fru_info)
+			return -ENOMEM;
+	}
+
+	fru_info = adev->fru_info;
+	/* For Arcturus-and-later, default value of serial_number is unique_id
+	 * so convert it to a 16-digit HEX string for convenience and
+	 * backwards-compatibility.
+	 */
+	sprintf(fru_info->serial, "%llx", adev->unique_id);
+
+	/* If algo exists, it means that the i2c_adapter's initialized */
+	if (!adev->pm.fru_eeprom_i2c_bus || !adev->pm.fru_eeprom_i2c_bus->algo) {
+		dev_warn(adev->dev,
+			 "Cannot access FRU, EEPROM accessor not initialized");
+		return -ENODEV;
+	}
+
+	/* Read the IPMI Common header */
+	len = amdgpu_eeprom_read(adev->pm.fru_eeprom_i2c_bus, fru_addr, buf,
+				 sizeof(buf));
+	if (len != 8) {
+		dev_err(adev->dev, "Couldn't read the IPMI Common Header: %d",
+			len);
+		return len < 0 ? len : -EIO;
+	}
+
+	if (buf[0] != 1) {
+		dev_err(adev->dev, "Bad IPMI Common Header version: 0x%02x",
+			buf[0]);
+		return -EIO;
+	}
+
+	for (csum = 0; len > 0; len--)
+		csum += buf[len - 1];
+	if (csum) {
+		dev_err(adev->dev, "Bad IPMI Common Header checksum: 0x%02x",
+			csum);
+		return -EIO;
+	}
+
+	/* Get the offset to the Product Info Area (PIA). */
+	addr = buf[4] * 8;
+	if (!addr)
+		return 0;
+
+	/* Get the absolute address to the PIA. */
+	addr += fru_addr;
+
+	/* Read the header of the PIA. */
+	len = amdgpu_eeprom_read(adev->pm.fru_eeprom_i2c_bus, addr, buf, 3);
+	if (len != 3) {
+		dev_err(adev->dev,
+			"Couldn't read the Product Info Area header: %d", len);
+		return len < 0 ? len : -EIO;
+	}
+
+	if (buf[0] != 1) {
+		dev_err(adev->dev, "Bad IPMI Product Info Area version: 0x%02x",
+			buf[0]);
+		return -EIO;
+	}
+
+	size = buf[1] * 8;
+	pia = kzalloc(size, GFP_KERNEL);
+	if (!pia)
+		return -ENOMEM;
+
+	/* Read the whole PIA. */
+	len = amdgpu_eeprom_read(adev->pm.fru_eeprom_i2c_bus, addr, pia, size);
+	if (len != size) {
+		kfree(pia);
+		dev_err(adev->dev, "Couldn't read the Product Info Area: %d",
+			len);
+		return len < 0 ? len : -EIO;
+	}
+
+	for (csum = 0; size > 0; size--)
+		csum += pia[size - 1];
+	if (csum) {
+		dev_err(adev->dev, "Bad Product Info Area checksum: 0x%02x",
+			csum);
+		kfree(pia);
+		return -EIO;
+	}
+
+	/* Now extract useful information from the PIA.
+	 *
+	 * Read Manufacturer Name field whose length is [3].
+	 */
+	addr = 3;
+	if (addr + 1 >= len)
+		goto Out;
+	memcpy(fru_info->manufacturer_name, pia + addr + 1,
+	       min_t(size_t, sizeof(fru_info->manufacturer_name),
+		     pia[addr] & 0x3F));
+	fru_info->manufacturer_name[sizeof(fru_info->manufacturer_name) - 1] =
+		'\0';
+
+	/* Read Product Name field. */
+	addr += 1 + (pia[addr] & 0x3F);
+	if (addr + 1 >= len)
+		goto Out;
+	memcpy(fru_info->product_name, pia + addr + 1,
+	       min_t(size_t, sizeof(fru_info->product_name), pia[addr] & 0x3F));
+	fru_info->product_name[sizeof(fru_info->product_name) - 1] = '\0';
+
+	/* Go to the Product Part/Model Number field. */
+	addr += 1 + (pia[addr] & 0x3F);
+	if (addr + 1 >= len)
+		goto Out;
+	memcpy(fru_info->product_number, pia + addr + 1,
+	       min_t(size_t, sizeof(fru_info->product_number),
+		     pia[addr] & 0x3F));
+	fru_info->product_number[sizeof(fru_info->product_number) - 1] = '\0';
+
+	/* Go to the Product Version field. */
+	addr += 1 + (pia[addr] & 0x3F);
+
+	/* Go to the Product Serial Number field. */
+	addr += 1 + (pia[addr] & 0x3F);
+	if (addr + 1 >= len)
+		goto Out;
+	memcpy(fru_info->serial, pia + addr + 1,
+	       min_t(size_t, sizeof(fru_info->serial), pia[addr] & 0x3F));
+	fru_info->serial[sizeof(fru_info->serial) - 1] = '\0';
+
+	/* Asset Tag field */
+	addr += 1 + (pia[addr] & 0x3F);
+
+	/* FRU File Id field. This could be 'null'. */
+	addr += 1 + (pia[addr] & 0x3F);
+	if ((addr + 1 >= len) || !(pia[addr] & 0x3F))
+		goto Out;
+	memcpy(fru_info->fru_id, pia + addr + 1,
+	       min_t(size_t, sizeof(fru_info->fru_id), pia[addr] & 0x3F));
+	fru_info->fru_id[sizeof(fru_info->fru_id) - 1] = '\0';
+
+Out:
+	kfree(pia);
+	return 0;
+}
+
+/**
+ * DOC: product_name
+ *
+ * The amdgpu driver provides a sysfs API for reporting the product name
+ * for the device
+ * The file product_name is used for this and returns the product name
+ * as returned from the FRU.
+ * NOTE: This is only available for certain server cards
+ */
+
+static ssize_t amdgpu_fru_product_name_show(struct device *dev,
+					    struct device_attribute *attr,
+					    char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	return sysfs_emit(buf, "%s\n", adev->fru_info->product_name);
+}
+
+static DEVICE_ATTR(product_name, 0444, amdgpu_fru_product_name_show, NULL);
+
+/**
+ * DOC: product_number
+ *
+ * The amdgpu driver provides a sysfs API for reporting the part number
+ * for the device
+ * The file product_number is used for this and returns the part number
+ * as returned from the FRU.
+ * NOTE: This is only available for certain server cards
+ */
+
+static ssize_t amdgpu_fru_product_number_show(struct device *dev,
+					      struct device_attribute *attr,
+					      char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	return sysfs_emit(buf, "%s\n", adev->fru_info->product_number);
+}
+
+static DEVICE_ATTR(product_number, 0444, amdgpu_fru_product_number_show, NULL);
+
+/**
+ * DOC: serial_number
+ *
+ * The amdgpu driver provides a sysfs API for reporting the serial number
+ * for the device
+ * The file serial_number is used for this and returns the serial number
+ * as returned from the FRU.
+ * NOTE: This is only available for certain server cards
+ */
+
+static ssize_t amdgpu_fru_serial_number_show(struct device *dev,
+					     struct device_attribute *attr,
+					     char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	return sysfs_emit(buf, "%s\n", adev->fru_info->serial);
+}
+
+static DEVICE_ATTR(serial_number, 0444, amdgpu_fru_serial_number_show, NULL);
+
+/**
+ * DOC: fru_id
+ *
+ * The amdgpu driver provides a sysfs API for reporting FRU File Id
+ * for the device.
+ * The file fru_id is used for this and returns the File Id value
+ * as returned from the FRU.
+ * NOTE: This is only available for certain server cards
+ */
+
+static ssize_t amdgpu_fru_id_show(struct device *dev,
+				  struct device_attribute *attr, char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	return sysfs_emit(buf, "%s\n", adev->fru_info->fru_id);
+}
+
+static DEVICE_ATTR(fru_id, 0444, amdgpu_fru_id_show, NULL);
+
+/**
+ * DOC: manufacturer
+ *
+ * The amdgpu driver provides a sysfs API for reporting manufacturer name from
+ * FRU information.
+ * The file manufacturer returns the value as returned from the FRU.
+ * NOTE: This is only available for certain server cards
+ */
+
+static ssize_t amdgpu_fru_manufacturer_name_show(struct device *dev,
+						 struct device_attribute *attr,
+						 char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	return sysfs_emit(buf, "%s\n", adev->fru_info->manufacturer_name);
+}
+
+static DEVICE_ATTR(manufacturer, 0444, amdgpu_fru_manufacturer_name_show, NULL);
+
+static const struct attribute *amdgpu_fru_attributes[] = {
+	&dev_attr_product_name.attr,
+	&dev_attr_product_number.attr,
+	&dev_attr_serial_number.attr,
+	&dev_attr_fru_id.attr,
+	&dev_attr_manufacturer.attr,
+	NULL
+};
+
+int amdgpu_fru_sysfs_init(struct amdgpu_device *adev)
+{
+	if (!is_fru_eeprom_supported(adev, NULL) || !adev->fru_info)
+		return 0;
+
+	return sysfs_create_files(&adev->dev->kobj, amdgpu_fru_attributes);
+}
+
+void amdgpu_fru_sysfs_fini(struct amdgpu_device *adev)
+{
+	if (!adev->fru_info)
+		return;
+
+	sysfs_remove_files(&adev->dev->kobj, amdgpu_fru_attributes);
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.h
new file mode 100644
index 000000000000..98f3196599ef
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.h
@@ -0,0 +1,42 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_FRU_EEPROM_H__
+#define __AMDGPU_FRU_EEPROM_H__
+
+#define AMDGPU_PRODUCT_NAME_LEN 64
+
+/* FRU product information */
+struct amdgpu_fru_info {
+	char				product_number[20];
+	char				product_name[AMDGPU_PRODUCT_NAME_LEN];
+	char				serial[20];
+	char				manufacturer_name[32];
+	char				fru_id[50];
+};
+
+int amdgpu_fru_get_product_info(struct amdgpu_device *adev);
+int amdgpu_fru_sysfs_init(struct amdgpu_device *adev);
+void amdgpu_fru_sysfs_fini(struct amdgpu_device *adev);
+
+#endif  // __AMDGPU_FRU_EEPROM_H__
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fw_attestation.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fw_attestation.c
new file mode 100644
index 000000000000..328a1b963548
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fw_attestation.c
@@ -0,0 +1,145 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/debugfs.h>
+#include <linux/firmware.h>
+#include <linux/dma-mapping.h>
+
+#include "amdgpu.h"
+#include "amdgpu_fw_attestation.h"
+#include "amdgpu_psp.h"
+#include "amdgpu_ucode.h"
+#include "soc15_common.h"
+
+#define FW_ATTESTATION_DB_COOKIE        0x143b6a37
+#define FW_ATTESTATION_RECORD_VALID	1
+#define FW_ATTESTATION_MAX_SIZE		4096
+
+struct FW_ATT_DB_HEADER {
+	uint32_t AttDbVersion;           /* version of the fwar feature */
+	uint32_t AttDbCookie;            /* cookie as an extra check for corrupt data */
+};
+
+struct FW_ATT_RECORD {
+	uint16_t AttFwIdV1;              /* Legacy FW Type field */
+	uint16_t AttFwIdV2;              /* V2 FW ID field */
+	uint32_t AttFWVersion;           /* FW Version */
+	uint16_t AttFWActiveFunctionID;  /* The VF ID (only in VF Attestation Table) */
+	uint8_t  AttSource;              /* FW source indicator */
+	uint8_t  RecordValid;            /* Indicates whether the record is a valid entry */
+	uint32_t AttFwTaId;              /* Ta ID (only in TA Attestation Table) */
+};
+
+static ssize_t amdgpu_fw_attestation_debugfs_read(struct file *f,
+						  char __user *buf,
+						  size_t size,
+						  loff_t *pos)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)file_inode(f)->i_private;
+	uint64_t records_addr = 0;
+	uint64_t vram_pos = 0;
+	struct FW_ATT_DB_HEADER fw_att_hdr = {0};
+	struct FW_ATT_RECORD fw_att_record = {0};
+
+	if (size < sizeof(struct FW_ATT_RECORD)) {
+		DRM_WARN("FW attestation input buffer not enough memory");
+		return -EINVAL;
+	}
+
+	if ((*pos + sizeof(struct FW_ATT_DB_HEADER)) >= FW_ATTESTATION_MAX_SIZE) {
+		DRM_WARN("FW attestation out of bounds");
+		return 0;
+	}
+
+	if (psp_get_fw_attestation_records_addr(&adev->psp, &records_addr)) {
+		DRM_WARN("Failed to get FW attestation record address");
+		return -EINVAL;
+	}
+
+	vram_pos =  records_addr - adev->gmc.vram_start;
+
+	if (*pos == 0) {
+		amdgpu_device_vram_access(adev,
+					  vram_pos,
+					  (uint32_t *)&fw_att_hdr,
+					  sizeof(struct FW_ATT_DB_HEADER),
+					  false);
+
+		if (fw_att_hdr.AttDbCookie != FW_ATTESTATION_DB_COOKIE) {
+			DRM_WARN("Invalid FW attestation cookie");
+			return -EINVAL;
+		}
+
+		DRM_INFO("FW attestation version = 0x%X", fw_att_hdr.AttDbVersion);
+	}
+
+	amdgpu_device_vram_access(adev,
+				  vram_pos + sizeof(struct FW_ATT_DB_HEADER) + *pos,
+				  (uint32_t *)&fw_att_record,
+				  sizeof(struct FW_ATT_RECORD),
+				  false);
+
+	if (fw_att_record.RecordValid != FW_ATTESTATION_RECORD_VALID)
+		return 0;
+
+	if (copy_to_user(buf, (void *)&fw_att_record, sizeof(struct FW_ATT_RECORD)))
+		return -EINVAL;
+
+	*pos += sizeof(struct FW_ATT_RECORD);
+
+	return sizeof(struct FW_ATT_RECORD);
+}
+
+static const struct file_operations amdgpu_fw_attestation_debugfs_ops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_fw_attestation_debugfs_read,
+	.write = NULL,
+	.llseek = default_llseek
+};
+
+static int amdgpu_is_fw_attestation_supported(struct amdgpu_device *adev)
+{
+	if (adev->flags & AMD_IS_APU)
+		return 0;
+
+	if (amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(14, 0, 2) ||
+	    amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(14, 0, 3))
+		return 0;
+
+	if (adev->asic_type >= CHIP_SIENNA_CICHLID)
+		return 1;
+
+	return 0;
+}
+
+void amdgpu_fw_attestation_debugfs_init(struct amdgpu_device *adev)
+{
+	if (!amdgpu_is_fw_attestation_supported(adev))
+		return;
+
+	debugfs_create_file("amdgpu_fw_attestation",
+			    0400,
+			    adev_to_drm(adev)->primary->debugfs_root,
+			    adev,
+			    &amdgpu_fw_attestation_debugfs_ops);
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fw_attestation.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_fw_attestation.h
new file mode 100644
index 000000000000..90af4fe58c99
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fw_attestation.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ *
+ */
+#ifndef _AMDGPU_FW_ATTESTATION_H
+#define _AMDGPU_FW_ATTESTATION_H
+
+#include "amdgpu.h"
+
+void amdgpu_fw_attestation_debugfs_init(struct amdgpu_device *adev);
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c
index f4370081f6e6..b2033f8352f5 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c
@@ -25,12 +25,18 @@
  *          Alex Deucher
  *          Jerome Glisse
  */
-#include <drm/drmP.h>
+
+#include <linux/pci.h>
+#include <linux/vmalloc.h>
+
 #include <drm/amdgpu_drm.h>
 #ifdef CONFIG_X86
 #include <asm/set_memory.h>
 #endif
 #include "amdgpu.h"
+#include "amdgpu_reset.h"
+#include <drm/drm_drv.h>
+#include <drm/ttm/ttm_tt.h>
 
 /*
  * GART
@@ -57,143 +63,211 @@
  */
 
 /**
- * amdgpu_gart_table_ram_alloc - allocate system ram for gart page table
+ * amdgpu_gart_dummy_page_init - init dummy page used by the driver
  *
  * @adev: amdgpu_device pointer
  *
- * Allocate system memory for GART page table
- * (r1xx-r3xx, non-pcie r4xx, rs400).  These asics require the
- * gart table to be in system memory.
- * Returns 0 for success, -ENOMEM for failure.
+ * Allocate the dummy page used by the driver (all asics).
+ * This dummy page is used by the driver as a filler for gart entries
+ * when pages are taken out of the GART
+ * Returns 0 on sucess, -ENOMEM on failure.
  */
-int amdgpu_gart_table_ram_alloc(struct amdgpu_device *adev)
+static int amdgpu_gart_dummy_page_init(struct amdgpu_device *adev)
 {
-	void *ptr;
+	struct page *dummy_page = ttm_glob.dummy_read_page;
 
-	ptr = pci_alloc_consistent(adev->pdev, adev->gart.table_size,
-				   &adev->gart.table_addr);
-	if (ptr == NULL) {
+	if (adev->dummy_page_addr)
+		return 0;
+	adev->dummy_page_addr = dma_map_page_attrs(&adev->pdev->dev, dummy_page, 0,
+							PAGE_SIZE, DMA_BIDIRECTIONAL,
+							DMA_ATTR_SKIP_CPU_SYNC);
+	if (dma_mapping_error(&adev->pdev->dev, adev->dummy_page_addr)) {
+		dev_err(&adev->pdev->dev, "Failed to DMA MAP the dummy page\n");
+		adev->dummy_page_addr = 0;
 		return -ENOMEM;
 	}
-#ifdef CONFIG_X86
-	if (0) {
-		set_memory_uc((unsigned long)ptr,
-			      adev->gart.table_size >> PAGE_SHIFT);
-	}
-#endif
-	adev->gart.ptr = ptr;
-	memset((void *)adev->gart.ptr, 0, adev->gart.table_size);
 	return 0;
 }
 
 /**
- * amdgpu_gart_table_ram_free - free system ram for gart page table
+ * amdgpu_gart_dummy_page_fini - free dummy page used by the driver
  *
  * @adev: amdgpu_device pointer
  *
- * Free system memory for GART page table
- * (r1xx-r3xx, non-pcie r4xx, rs400).  These asics require the
- * gart table to be in system memory.
+ * Frees the dummy page used by the driver (all asics).
  */
-void amdgpu_gart_table_ram_free(struct amdgpu_device *adev)
+void amdgpu_gart_dummy_page_fini(struct amdgpu_device *adev)
 {
-	if (adev->gart.ptr == NULL) {
+	if (!adev->dummy_page_addr)
 		return;
-	}
-#ifdef CONFIG_X86
-	if (0) {
-		set_memory_wb((unsigned long)adev->gart.ptr,
-			      adev->gart.table_size >> PAGE_SHIFT);
-	}
-#endif
-	pci_free_consistent(adev->pdev, adev->gart.table_size,
-			    (void *)adev->gart.ptr,
-			    adev->gart.table_addr);
-	adev->gart.ptr = NULL;
-	adev->gart.table_addr = 0;
+	dma_unmap_page_attrs(&adev->pdev->dev, adev->dummy_page_addr, PAGE_SIZE,
+				DMA_BIDIRECTIONAL,
+				DMA_ATTR_SKIP_CPU_SYNC);
+	adev->dummy_page_addr = 0;
 }
 
 /**
- * amdgpu_gart_table_vram_alloc - allocate vram for gart page table
+ * amdgpu_gart_table_ram_alloc - allocate system ram for gart page table
  *
  * @adev: amdgpu_device pointer
  *
- * Allocate video memory for GART page table
- * (pcie r4xx, r5xx+).  These asics require the
- * gart table to be in video memory.
+ * Allocate system memory for GART page table for ASICs that don't have
+ * dedicated VRAM.
  * Returns 0 for success, error for failure.
  */
-int amdgpu_gart_table_vram_alloc(struct amdgpu_device *adev)
+int amdgpu_gart_table_ram_alloc(struct amdgpu_device *adev)
 {
-	int r;
+	unsigned int order = get_order(adev->gart.table_size);
+	gfp_t gfp_flags = GFP_KERNEL | __GFP_ZERO;
+	struct amdgpu_bo *bo = NULL;
+	struct sg_table *sg = NULL;
+	struct amdgpu_bo_param bp;
+	dma_addr_t dma_addr;
+	struct page *p;
+	unsigned long x;
+	int ret;
 
-	if (adev->gart.robj == NULL) {
-		r = amdgpu_bo_create(adev, adev->gart.table_size,
-				     PAGE_SIZE, true, AMDGPU_GEM_DOMAIN_VRAM,
-				     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
-				     AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
-				     NULL, NULL, 0, &adev->gart.robj);
-		if (r) {
-			return r;
-		}
+	if (adev->gart.bo != NULL)
+		return 0;
+
+	p = alloc_pages(gfp_flags, order);
+	if (!p)
+		return -ENOMEM;
+
+	/* assign pages to this device */
+	for (x = 0; x < (1UL << order); x++)
+		p[x].mapping = adev->mman.bdev.dev_mapping;
+
+	/* If the hardware does not support UTCL2 snooping of the CPU caches
+	 * then set_memory_wc() could be used as a workaround to mark the pages
+	 * as write combine memory.
+	 */
+	dma_addr = dma_map_page(&adev->pdev->dev, p, 0, adev->gart.table_size,
+				DMA_BIDIRECTIONAL);
+	if (dma_mapping_error(&adev->pdev->dev, dma_addr)) {
+		dev_err(&adev->pdev->dev, "Failed to DMA MAP the GART BO page\n");
+		__free_pages(p, order);
+		p = NULL;
+		return -EFAULT;
+	}
+
+	dev_info(adev->dev, "%s dma_addr:%pad\n", __func__, &dma_addr);
+	/* Create SG table */
+	sg = kmalloc(sizeof(*sg), GFP_KERNEL);
+	if (!sg) {
+		ret = -ENOMEM;
+		goto error;
+	}
+	ret = sg_alloc_table(sg, 1, GFP_KERNEL);
+	if (ret)
+		goto error;
+
+	sg_dma_address(sg->sgl) = dma_addr;
+	sg->sgl->length = adev->gart.table_size;
+#ifdef CONFIG_NEED_SG_DMA_LENGTH
+	sg->sgl->dma_length = adev->gart.table_size;
+#endif
+	/* Create SG BO */
+	memset(&bp, 0, sizeof(bp));
+	bp.size = adev->gart.table_size;
+	bp.byte_align = PAGE_SIZE;
+	bp.domain = AMDGPU_GEM_DOMAIN_CPU;
+	bp.type = ttm_bo_type_sg;
+	bp.resv = NULL;
+	bp.bo_ptr_size = sizeof(struct amdgpu_bo);
+	bp.flags = 0;
+	ret = amdgpu_bo_create(adev, &bp, &bo);
+	if (ret)
+		goto error;
+
+	bo->tbo.sg = sg;
+	bo->tbo.ttm->sg = sg;
+	bo->allowed_domains = AMDGPU_GEM_DOMAIN_GTT;
+	bo->preferred_domains = AMDGPU_GEM_DOMAIN_GTT;
+
+	ret = amdgpu_bo_reserve(bo, true);
+	if (ret) {
+		dev_err(adev->dev, "(%d) failed to reserve bo for GART system bo\n", ret);
+		goto error;
 	}
+
+	ret = amdgpu_bo_pin(bo, AMDGPU_GEM_DOMAIN_GTT);
+	WARN(ret, "Pinning the GART table failed");
+	if (ret)
+		goto error_resv;
+
+	adev->gart.bo = bo;
+	adev->gart.ptr = page_to_virt(p);
+	/* Make GART table accessible in VMID0 */
+	ret = amdgpu_ttm_alloc_gart(&adev->gart.bo->tbo);
+	if (ret)
+		amdgpu_gart_table_ram_free(adev);
+	amdgpu_bo_unreserve(bo);
+
 	return 0;
+
+error_resv:
+	amdgpu_bo_unreserve(bo);
+error:
+	amdgpu_bo_unref(&bo);
+	if (sg) {
+		sg_free_table(sg);
+		kfree(sg);
+	}
+	__free_pages(p, order);
+	return ret;
 }
 
 /**
- * amdgpu_gart_table_vram_pin - pin gart page table in vram
+ * amdgpu_gart_table_ram_free - free gart page table system ram
  *
  * @adev: amdgpu_device pointer
  *
- * Pin the GART page table in vram so it will not be moved
- * by the memory manager (pcie r4xx, r5xx+).  These asics require the
- * gart table to be in video memory.
- * Returns 0 for success, error for failure.
+ * Free the system memory used for the GART page tableon ASICs that don't
+ * have dedicated VRAM.
  */
-int amdgpu_gart_table_vram_pin(struct amdgpu_device *adev)
+void amdgpu_gart_table_ram_free(struct amdgpu_device *adev)
 {
-	uint64_t gpu_addr;
-	int r;
+	unsigned int order = get_order(adev->gart.table_size);
+	struct sg_table *sg = adev->gart.bo->tbo.sg;
+	struct page *p;
+	unsigned long x;
+	int ret;
 
-	r = amdgpu_bo_reserve(adev->gart.robj, false);
-	if (unlikely(r != 0))
-		return r;
-	r = amdgpu_bo_pin(adev->gart.robj,
-				AMDGPU_GEM_DOMAIN_VRAM, &gpu_addr);
-	if (r) {
-		amdgpu_bo_unreserve(adev->gart.robj);
-		return r;
+	ret = amdgpu_bo_reserve(adev->gart.bo, false);
+	if (!ret) {
+		amdgpu_bo_unpin(adev->gart.bo);
+		amdgpu_bo_unreserve(adev->gart.bo);
 	}
-	r = amdgpu_bo_kmap(adev->gart.robj, &adev->gart.ptr);
-	if (r)
-		amdgpu_bo_unpin(adev->gart.robj);
-	amdgpu_bo_unreserve(adev->gart.robj);
-	adev->gart.table_addr = gpu_addr;
-	return r;
+	amdgpu_bo_unref(&adev->gart.bo);
+	sg_free_table(sg);
+	kfree(sg);
+	p = virt_to_page(adev->gart.ptr);
+	for (x = 0; x < (1UL << order); x++)
+		p[x].mapping = NULL;
+	__free_pages(p, order);
+
+	adev->gart.ptr = NULL;
 }
 
 /**
- * amdgpu_gart_table_vram_unpin - unpin gart page table in vram
+ * amdgpu_gart_table_vram_alloc - allocate vram for gart page table
  *
  * @adev: amdgpu_device pointer
  *
- * Unpin the GART page table in vram (pcie r4xx, r5xx+).
- * These asics require the gart table to be in video memory.
+ * Allocate video memory for GART page table
+ * (pcie r4xx, r5xx+).  These asics require the
+ * gart table to be in video memory.
+ * Returns 0 for success, error for failure.
  */
-void amdgpu_gart_table_vram_unpin(struct amdgpu_device *adev)
+int amdgpu_gart_table_vram_alloc(struct amdgpu_device *adev)
 {
-	int r;
+	if (adev->gart.bo != NULL)
+		return 0;
 
-	if (adev->gart.robj == NULL) {
-		return;
-	}
-	r = amdgpu_bo_reserve(adev->gart.robj, true);
-	if (likely(r == 0)) {
-		amdgpu_bo_kunmap(adev->gart.robj);
-		amdgpu_bo_unpin(adev->gart.robj);
-		amdgpu_bo_unreserve(adev->gart.robj);
-		adev->gart.ptr = NULL;
-	}
+	return amdgpu_bo_create_kernel(adev,  adev->gart.table_size, PAGE_SIZE,
+				       AMDGPU_GEM_DOMAIN_VRAM, &adev->gart.bo,
+				       NULL, (void *)&adev->gart.ptr);
 }
 
 /**
@@ -207,10 +281,7 @@ void amdgpu_gart_table_vram_unpin(struct amdgpu_device *adev)
  */
 void amdgpu_gart_table_vram_free(struct amdgpu_device *adev)
 {
-	if (adev->gart.robj == NULL) {
-		return;
-	}
-	amdgpu_bo_unref(&adev->gart.robj);
+	amdgpu_bo_free_kernel(&adev->gart.bo, NULL, (void *)&adev->gart.ptr);
 }
 
 /*
@@ -227,7 +298,7 @@ void amdgpu_gart_table_vram_free(struct amdgpu_device *adev)
  * replaces them with the dummy page (all asics).
  * Returns 0 for success, -EINVAL for failure.
  */
-int amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset,
+void amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset,
 			int pages)
 {
 	unsigned t;
@@ -236,31 +307,30 @@ int amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset,
 	u64 page_base;
 	/* Starting from VEGA10, system bit must be 0 to mean invalid. */
 	uint64_t flags = 0;
+	int idx;
 
-	if (!adev->gart.ready) {
-		WARN(1, "trying to unbind memory from uninitialized GART !\n");
-		return -EINVAL;
-	}
+	if (!adev->gart.ptr)
+		return;
+
+	if (!drm_dev_enter(adev_to_drm(adev), &idx))
+		return;
 
 	t = offset / AMDGPU_GPU_PAGE_SIZE;
-	p = t / (PAGE_SIZE / AMDGPU_GPU_PAGE_SIZE);
+	p = t / AMDGPU_GPU_PAGES_IN_CPU_PAGE;
 	for (i = 0; i < pages; i++, p++) {
-#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS
-		adev->gart.pages[p] = NULL;
-#endif
-		page_base = adev->dummy_page.addr;
+		page_base = adev->dummy_page_addr;
 		if (!adev->gart.ptr)
 			continue;
 
-		for (j = 0; j < (PAGE_SIZE / AMDGPU_GPU_PAGE_SIZE); j++, t++) {
-			amdgpu_gart_set_pte_pde(adev, adev->gart.ptr,
-						t, page_base, flags);
+		for (j = 0; j < AMDGPU_GPU_PAGES_IN_CPU_PAGE; j++, t++) {
+			amdgpu_gmc_set_pte_pde(adev, adev->gart.ptr,
+					       t, page_base, flags);
 			page_base += AMDGPU_GPU_PAGE_SIZE;
 		}
 	}
-	mb();
-	amdgpu_gart_flush_gpu_tlb(adev, 0);
-	return 0;
+	amdgpu_gart_invalidate_tlb(adev);
+
+	drm_dev_exit(idx);
 }
 
 /**
@@ -270,32 +340,33 @@ int amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset,
  * @offset: offset into the GPU's gart aperture
  * @pages: number of pages to bind
  * @dma_addr: DMA addresses of pages
+ * @flags: page table entry flags
+ * @dst: CPU address of the gart table
  *
  * Map the dma_addresses into GART entries (all asics).
  * Returns 0 for success, -EINVAL for failure.
  */
-int amdgpu_gart_map(struct amdgpu_device *adev, uint64_t offset,
+void amdgpu_gart_map(struct amdgpu_device *adev, uint64_t offset,
 		    int pages, dma_addr_t *dma_addr, uint64_t flags,
 		    void *dst)
 {
 	uint64_t page_base;
 	unsigned i, j, t;
+	int idx;
 
-	if (!adev->gart.ready) {
-		WARN(1, "trying to bind memory to uninitialized GART !\n");
-		return -EINVAL;
-	}
+	if (!drm_dev_enter(adev_to_drm(adev), &idx))
+		return;
 
 	t = offset / AMDGPU_GPU_PAGE_SIZE;
 
 	for (i = 0; i < pages; i++) {
 		page_base = dma_addr[i];
-		for (j = 0; j < (PAGE_SIZE / AMDGPU_GPU_PAGE_SIZE); j++, t++) {
-			amdgpu_gart_set_pte_pde(adev, dst, t, page_base, flags);
+		for (j = 0; j < AMDGPU_GPU_PAGES_IN_CPU_PAGE; j++, t++) {
+			amdgpu_gmc_set_pte_pde(adev, dst, t, page_base, flags);
 			page_base += AMDGPU_GPU_PAGE_SIZE;
 		}
 	}
-	return 0;
+	drm_dev_exit(idx);
 }
 
 /**
@@ -304,44 +375,45 @@ int amdgpu_gart_map(struct amdgpu_device *adev, uint64_t offset,
  * @adev: amdgpu_device pointer
  * @offset: offset into the GPU's gart aperture
  * @pages: number of pages to bind
- * @pagelist: pages to bind
  * @dma_addr: DMA addresses of pages
+ * @flags: page table entry flags
  *
  * Binds the requested pages to the gart page table
  * (all asics).
  * Returns 0 for success, -EINVAL for failure.
  */
-int amdgpu_gart_bind(struct amdgpu_device *adev, uint64_t offset,
-		     int pages, struct page **pagelist, dma_addr_t *dma_addr,
+void amdgpu_gart_bind(struct amdgpu_device *adev, uint64_t offset,
+		     int pages, dma_addr_t *dma_addr,
 		     uint64_t flags)
 {
-#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS
-	unsigned i,t,p;
-#endif
-	int r;
+	if (!adev->gart.ptr)
+		return;
 
-	if (!adev->gart.ready) {
-		WARN(1, "trying to bind memory to uninitialized GART !\n");
-		return -EINVAL;
-	}
+	amdgpu_gart_map(adev, offset, pages, dma_addr, flags, adev->gart.ptr);
+}
 
-#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS
-	t = offset / AMDGPU_GPU_PAGE_SIZE;
-	p = t / (PAGE_SIZE / AMDGPU_GPU_PAGE_SIZE);
-	for (i = 0; i < pages; i++, p++)
-		adev->gart.pages[p] = pagelist[i];
-#endif
+/**
+ * amdgpu_gart_invalidate_tlb - invalidate gart TLB
+ *
+ * @adev: amdgpu device driver pointer
+ *
+ * Invalidate gart TLB which can be use as a way to flush gart changes
+ *
+ */
+void amdgpu_gart_invalidate_tlb(struct amdgpu_device *adev)
+{
+	int i;
 
-	if (adev->gart.ptr) {
-		r = amdgpu_gart_map(adev, offset, pages, dma_addr, flags,
-			    adev->gart.ptr);
-		if (r)
-			return r;
-	}
+	if (!adev->gart.ptr)
+		return;
 
 	mb();
-	amdgpu_gart_flush_gpu_tlb(adev, 0);
-	return 0;
+	if (down_read_trylock(&adev->reset_domain->sem)) {
+		amdgpu_device_flush_hdp(adev, NULL);
+		up_read(&adev->reset_domain->sem);
+	}
+	for_each_set_bit(i, adev->vmhubs_mask, AMDGPU_MAX_VMHUBS)
+		amdgpu_gmc_flush_gpu_tlb(adev, 0, i, 0);
 }
 
 /**
@@ -356,7 +428,7 @@ int amdgpu_gart_init(struct amdgpu_device *adev)
 {
 	int r;
 
-	if (adev->dummy_page.page)
+	if (adev->dummy_page_addr)
 		return 0;
 
 	/* We need PAGE_SIZE >= AMDGPU_GPU_PAGE_SIZE */
@@ -364,44 +436,14 @@ int amdgpu_gart_init(struct amdgpu_device *adev)
 		DRM_ERROR("Page size is smaller than GPU page size!\n");
 		return -EINVAL;
 	}
-	r = amdgpu_dummy_page_init(adev);
+	r = amdgpu_gart_dummy_page_init(adev);
 	if (r)
 		return r;
 	/* Compute table size */
-	adev->gart.num_cpu_pages = adev->mc.gart_size / PAGE_SIZE;
-	adev->gart.num_gpu_pages = adev->mc.gart_size / AMDGPU_GPU_PAGE_SIZE;
+	adev->gart.num_cpu_pages = adev->gmc.gart_size / PAGE_SIZE;
+	adev->gart.num_gpu_pages = adev->gmc.gart_size / AMDGPU_GPU_PAGE_SIZE;
 	DRM_INFO("GART: num cpu pages %u, num gpu pages %u\n",
 		 adev->gart.num_cpu_pages, adev->gart.num_gpu_pages);
 
-#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS
-	/* Allocate pages table */
-	adev->gart.pages = vzalloc(sizeof(void *) * adev->gart.num_cpu_pages);
-	if (adev->gart.pages == NULL) {
-		amdgpu_gart_fini(adev);
-		return -ENOMEM;
-	}
-#endif
-
 	return 0;
 }
-
-/**
- * amdgpu_gart_fini - tear down the driver info for managing the gart
- *
- * @adev: amdgpu_device pointer
- *
- * Tear down the gart driver info and free the dummy page (all asics).
- */
-void amdgpu_gart_fini(struct amdgpu_device *adev)
-{
-	if (adev->gart.ready) {
-		/* unbind pages */
-		amdgpu_gart_unbind(adev, 0, adev->gart.num_cpu_pages);
-	}
-	adev->gart.ready = false;
-#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS
-	vfree(adev->gart.pages);
-	adev->gart.pages = NULL;
-#endif
-	amdgpu_dummy_page_fini(adev);
-}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h
index afbe803b1a13..7cc980bf4725 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h
@@ -31,46 +31,38 @@
  */
 struct amdgpu_device;
 struct amdgpu_bo;
-struct amdgpu_gart_funcs;
 
 #define AMDGPU_GPU_PAGE_SIZE 4096
 #define AMDGPU_GPU_PAGE_MASK (AMDGPU_GPU_PAGE_SIZE - 1)
 #define AMDGPU_GPU_PAGE_SHIFT 12
 #define AMDGPU_GPU_PAGE_ALIGN(a) (((a) + AMDGPU_GPU_PAGE_MASK) & ~AMDGPU_GPU_PAGE_MASK)
 
+#define AMDGPU_GPU_PAGES_IN_CPU_PAGE (PAGE_SIZE / AMDGPU_GPU_PAGE_SIZE)
+
 struct amdgpu_gart {
-	dma_addr_t			table_addr;
-	struct amdgpu_bo		*robj;
+	struct amdgpu_bo		*bo;
+	/* CPU kmapped address of gart table */
 	void				*ptr;
 	unsigned			num_gpu_pages;
 	unsigned			num_cpu_pages;
 	unsigned			table_size;
-#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS
-	struct page			**pages;
-#endif
-	bool				ready;
 
 	/* Asic default pte flags */
 	uint64_t			gart_pte_flags;
-
-	const struct amdgpu_gart_funcs *gart_funcs;
 };
 
 int amdgpu_gart_table_ram_alloc(struct amdgpu_device *adev);
 void amdgpu_gart_table_ram_free(struct amdgpu_device *adev);
 int amdgpu_gart_table_vram_alloc(struct amdgpu_device *adev);
 void amdgpu_gart_table_vram_free(struct amdgpu_device *adev);
-int amdgpu_gart_table_vram_pin(struct amdgpu_device *adev);
-void amdgpu_gart_table_vram_unpin(struct amdgpu_device *adev);
 int amdgpu_gart_init(struct amdgpu_device *adev);
-void amdgpu_gart_fini(struct amdgpu_device *adev);
-int amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset,
-		       int pages);
-int amdgpu_gart_map(struct amdgpu_device *adev, uint64_t offset,
-		    int pages, dma_addr_t *dma_addr, uint64_t flags,
-		    void *dst);
-int amdgpu_gart_bind(struct amdgpu_device *adev, uint64_t offset,
-		     int pages, struct page **pagelist,
-		     dma_addr_t *dma_addr, uint64_t flags);
-
+void amdgpu_gart_dummy_page_fini(struct amdgpu_device *adev);
+void amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset,
+			int pages);
+void amdgpu_gart_map(struct amdgpu_device *adev, uint64_t offset,
+		     int pages, dma_addr_t *dma_addr, uint64_t flags,
+		     void *dst);
+void amdgpu_gart_bind(struct amdgpu_device *adev, uint64_t offset,
+		      int pages, dma_addr_t *dma_addr, uint64_t flags);
+void amdgpu_gart_invalidate_tlb(struct amdgpu_device *adev);
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gds.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gds.h
index e73728d90388..f6ac1e9548f2 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gds.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gds.h
@@ -24,34 +24,14 @@
 #ifndef __AMDGPU_GDS_H__
 #define __AMDGPU_GDS_H__
 
-/* Because TTM request that alloacted buffer should be PAGE_SIZE aligned,
- * we should report GDS/GWS/OA size as PAGE_SIZE aligned
- * */
-#define AMDGPU_GDS_SHIFT	2
-#define AMDGPU_GWS_SHIFT	PAGE_SHIFT
-#define AMDGPU_OA_SHIFT		PAGE_SHIFT
-
 struct amdgpu_ring;
 struct amdgpu_bo;
 
-struct amdgpu_gds_asic_info {
-	uint32_t	total_size;
-	uint32_t	gfx_partition_size;
-	uint32_t	cs_partition_size;
-};
-
 struct amdgpu_gds {
-	struct amdgpu_gds_asic_info	mem;
-	struct amdgpu_gds_asic_info	gws;
-	struct amdgpu_gds_asic_info	oa;
-	/* At present, GDS, GWS and OA resources for gfx (graphics)
-	 * is always pre-allocated and available for graphics operation.
-	 * Such resource is shared between all gfx clients.
-	 * TODO: move this operation to user space
-	 * */
-	struct amdgpu_bo*		gds_gfx_bo;
-	struct amdgpu_bo*		gws_gfx_bo;
-	struct amdgpu_bo*		oa_gfx_bo;
+	uint32_t gds_size;
+	uint32_t gws_size;
+	uint32_t oa_size;
+	uint32_t gds_compute_max_wave_id;
 };
 
 struct amdgpu_gds_reg_offset {
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index 7171968f261e..b7ebae289bea 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -26,59 +26,219 @@
  *          Jerome Glisse
  */
 #include <linux/ktime.h>
+#include <linux/module.h>
 #include <linux/pagemap.h>
-#include <drm/drmP.h>
+#include <linux/pci.h>
+#include <linux/dma-buf.h>
+
 #include <drm/amdgpu_drm.h>
+#include <drm/drm_drv.h>
+#include <drm/drm_exec.h>
+#include <drm/drm_gem_ttm_helper.h>
+#include <drm/ttm/ttm_tt.h>
+#include <drm/drm_syncobj.h>
+
 #include "amdgpu.h"
+#include "amdgpu_display.h"
+#include "amdgpu_dma_buf.h"
+#include "amdgpu_hmm.h"
+#include "amdgpu_xgmi.h"
+#include "amdgpu_vm.h"
+
+static int
+amdgpu_gem_add_input_fence(struct drm_file *filp,
+			   uint64_t syncobj_handles_array,
+			   uint32_t num_syncobj_handles)
+{
+	struct dma_fence *fence;
+	uint32_t *syncobj_handles;
+	int ret, i;
+
+	if (!num_syncobj_handles)
+		return 0;
+
+	syncobj_handles = memdup_user(u64_to_user_ptr(syncobj_handles_array),
+				      size_mul(sizeof(uint32_t), num_syncobj_handles));
+	if (IS_ERR(syncobj_handles))
+		return PTR_ERR(syncobj_handles);
+
+	for (i = 0; i < num_syncobj_handles; i++) {
+
+		if (!syncobj_handles[i]) {
+			ret = -EINVAL;
+			goto free_memdup;
+		}
+
+		ret = drm_syncobj_find_fence(filp, syncobj_handles[i], 0, 0, &fence);
+		if (ret)
+			goto free_memdup;
 
-void amdgpu_gem_object_free(struct drm_gem_object *gobj)
+		dma_fence_wait(fence, false);
+
+		/* TODO: optimize async handling */
+		dma_fence_put(fence);
+	}
+
+free_memdup:
+	kfree(syncobj_handles);
+	return ret;
+}
+
+static int
+amdgpu_gem_update_timeline_node(struct drm_file *filp,
+				uint32_t syncobj_handle,
+				uint64_t point,
+				struct drm_syncobj **syncobj,
+				struct dma_fence_chain **chain)
 {
-	struct amdgpu_bo *robj = gem_to_amdgpu_bo(gobj);
+	if (!syncobj_handle)
+		return 0;
+
+	/* Find the sync object */
+	*syncobj = drm_syncobj_find(filp, syncobj_handle);
+	if (!*syncobj)
+		return -ENOENT;
 
-	if (robj) {
-		if (robj->gem_base.import_attach)
-			drm_prime_gem_destroy(&robj->gem_base, robj->tbo.sg);
-		amdgpu_mn_unregister(robj);
-		amdgpu_bo_unref(&robj);
+	if (!point)
+		return 0;
+
+	/* Allocate the chain node */
+	*chain = dma_fence_chain_alloc();
+	if (!*chain) {
+		drm_syncobj_put(*syncobj);
+		return -ENOMEM;
 	}
+
+	return 0;
 }
 
-int amdgpu_gem_object_create(struct amdgpu_device *adev, unsigned long size,
-				int alignment, u32 initial_domain,
-				u64 flags, bool kernel,
-				struct drm_gem_object **obj)
+static void
+amdgpu_gem_update_bo_mapping(struct drm_file *filp,
+			     struct amdgpu_bo_va *bo_va,
+			     uint32_t operation,
+			     uint64_t point,
+			     struct dma_fence *fence,
+			     struct drm_syncobj *syncobj,
+			     struct dma_fence_chain *chain)
 {
-	struct amdgpu_bo *robj;
-	int r;
+	struct amdgpu_bo *bo = bo_va ? bo_va->base.bo : NULL;
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+	struct amdgpu_vm *vm = &fpriv->vm;
+	struct dma_fence *last_update;
 
-	*obj = NULL;
-	/* At least align on page size */
-	if (alignment < PAGE_SIZE) {
-		alignment = PAGE_SIZE;
+	if (!syncobj)
+		return;
+
+	/* Find the last update fence */
+	switch (operation) {
+	case AMDGPU_VA_OP_MAP:
+	case AMDGPU_VA_OP_REPLACE:
+		if (bo && (bo->tbo.base.resv == vm->root.bo->tbo.base.resv))
+			last_update = vm->last_update;
+		else
+			last_update = bo_va->last_pt_update;
+		break;
+	case AMDGPU_VA_OP_UNMAP:
+	case AMDGPU_VA_OP_CLEAR:
+		last_update = fence;
+		break;
+	default:
+		return;
 	}
 
-retry:
-	r = amdgpu_bo_create(adev, size, alignment, kernel, initial_domain,
-			     flags, NULL, NULL, 0, &robj);
-	if (r) {
-		if (r != -ERESTARTSYS) {
-			if (initial_domain == AMDGPU_GEM_DOMAIN_VRAM) {
-				initial_domain |= AMDGPU_GEM_DOMAIN_GTT;
-				goto retry;
-			}
-			DRM_ERROR("Failed to allocate GEM object (%ld, %d, %u, %d)\n",
-				  size, initial_domain, alignment, r);
+	/* Add fence to timeline */
+	if (!point)
+		drm_syncobj_replace_fence(syncobj, last_update);
+	else
+		drm_syncobj_add_point(syncobj, chain, last_update, point);
+}
+
+static vm_fault_t amdgpu_gem_fault(struct vm_fault *vmf)
+{
+	struct ttm_buffer_object *bo = vmf->vma->vm_private_data;
+	struct drm_device *ddev = bo->base.dev;
+	vm_fault_t ret;
+	int idx;
+
+	ret = ttm_bo_vm_reserve(bo, vmf);
+	if (ret)
+		return ret;
+
+	if (drm_dev_enter(ddev, &idx)) {
+		ret = amdgpu_bo_fault_reserve_notify(bo);
+		if (ret) {
+			drm_dev_exit(idx);
+			goto unlock;
 		}
-		return r;
+
+		ret = ttm_bo_vm_fault_reserved(vmf, vmf->vma->vm_page_prot,
+					       TTM_BO_VM_NUM_PREFAULT);
+
+		drm_dev_exit(idx);
+	} else {
+		ret = ttm_bo_vm_dummy_page(vmf, vmf->vma->vm_page_prot);
 	}
-	*obj = &robj->gem_base;
+	if (ret == VM_FAULT_RETRY && !(vmf->flags & FAULT_FLAG_RETRY_NOWAIT))
+		return ret;
+
+unlock:
+	dma_resv_unlock(bo->base.resv);
+	return ret;
+}
+
+static const struct vm_operations_struct amdgpu_gem_vm_ops = {
+	.fault = amdgpu_gem_fault,
+	.open = ttm_bo_vm_open,
+	.close = ttm_bo_vm_close,
+	.access = ttm_bo_vm_access
+};
+
+static void amdgpu_gem_object_free(struct drm_gem_object *gobj)
+{
+	struct amdgpu_bo *aobj = gem_to_amdgpu_bo(gobj);
+
+	amdgpu_hmm_unregister(aobj);
+	ttm_bo_put(&aobj->tbo);
+}
+
+int amdgpu_gem_object_create(struct amdgpu_device *adev, unsigned long size,
+			     int alignment, u32 initial_domain,
+			     u64 flags, enum ttm_bo_type type,
+			     struct dma_resv *resv,
+			     struct drm_gem_object **obj, int8_t xcp_id_plus1)
+{
+	struct amdgpu_bo *bo;
+	struct amdgpu_bo_user *ubo;
+	struct amdgpu_bo_param bp;
+	int r;
+
+	memset(&bp, 0, sizeof(bp));
+	*obj = NULL;
+	flags |= AMDGPU_GEM_CREATE_VRAM_WIPE_ON_RELEASE;
+
+	bp.size = size;
+	bp.byte_align = alignment;
+	bp.type = type;
+	bp.resv = resv;
+	bp.preferred_domain = initial_domain;
+	bp.flags = flags;
+	bp.domain = initial_domain;
+	bp.bo_ptr_size = sizeof(struct amdgpu_bo);
+	bp.xcp_id_plus1 = xcp_id_plus1;
+
+	r = amdgpu_bo_create_user(adev, &bp, &ubo);
+	if (r)
+		return r;
+
+	bo = &ubo->bo;
+	*obj = &bo->tbo.base;
 
 	return 0;
 }
 
 void amdgpu_gem_force_release(struct amdgpu_device *adev)
 {
-	struct drm_device *ddev = adev->ddev;
+	struct drm_device *ddev = adev_to_drm(adev);
 	struct drm_file *file;
 
 	mutex_lock(&ddev->filelist_mutex);
@@ -91,7 +251,7 @@ void amdgpu_gem_force_release(struct amdgpu_device *adev)
 		spin_lock(&file->table_lock);
 		idr_for_each_entry(&file->object_idr, gobj, handle) {
 			WARN_ONCE(1, "And also active allocations!\n");
-			drm_gem_object_put_unlocked(gobj);
+			drm_gem_object_put(gobj);
 		}
 		idr_destroy(&file->object_idr);
 		spin_unlock(&file->table_lock);
@@ -104,164 +264,257 @@ void amdgpu_gem_force_release(struct amdgpu_device *adev)
  * Call from drm_gem_handle_create which appear in both new and open ioctl
  * case.
  */
-int amdgpu_gem_object_open(struct drm_gem_object *obj,
-			   struct drm_file *file_priv)
+static int amdgpu_gem_object_open(struct drm_gem_object *obj,
+				  struct drm_file *file_priv)
 {
 	struct amdgpu_bo *abo = gem_to_amdgpu_bo(obj);
 	struct amdgpu_device *adev = amdgpu_ttm_adev(abo->tbo.bdev);
 	struct amdgpu_fpriv *fpriv = file_priv->driver_priv;
 	struct amdgpu_vm *vm = &fpriv->vm;
 	struct amdgpu_bo_va *bo_va;
+	struct mm_struct *mm;
 	int r;
+
+	mm = amdgpu_ttm_tt_get_usermm(abo->tbo.ttm);
+	if (mm && mm != current->mm)
+		return -EPERM;
+
+	if (abo->flags & AMDGPU_GEM_CREATE_VM_ALWAYS_VALID &&
+	    !amdgpu_vm_is_bo_always_valid(vm, abo))
+		return -EPERM;
+
 	r = amdgpu_bo_reserve(abo, false);
 	if (r)
 		return r;
 
+	amdgpu_vm_bo_update_shared(abo);
 	bo_va = amdgpu_vm_bo_find(vm, abo);
-	if (!bo_va) {
+	if (!bo_va)
 		bo_va = amdgpu_vm_bo_add(adev, vm, abo);
-	} else {
+	else
 		++bo_va->ref_count;
-	}
-	amdgpu_bo_unreserve(abo);
-	return 0;
-}
 
-static int amdgpu_gem_vm_check(void *param, struct amdgpu_bo *bo)
-{
-	/* if anything is swapped out don't swap it in here,
-	   just abort and wait for the next CS */
-	if (!amdgpu_bo_gpu_accessible(bo))
-		return -ERESTARTSYS;
-
-	if (bo->shadow && !amdgpu_bo_gpu_accessible(bo->shadow))
-		return -ERESTARTSYS;
-
-	return 0;
-}
+	/* attach gfx eviction fence */
+	r = amdgpu_eviction_fence_attach(&fpriv->evf_mgr, abo);
+	if (r) {
+		DRM_DEBUG_DRIVER("Failed to attach eviction fence to BO\n");
+		amdgpu_bo_unreserve(abo);
+		return r;
+	}
 
-static bool amdgpu_gem_vm_ready(struct amdgpu_device *adev,
-				struct amdgpu_vm *vm,
-				struct list_head *list)
-{
-	struct ttm_validate_buffer *entry;
+	amdgpu_bo_unreserve(abo);
 
-	list_for_each_entry(entry, list, head) {
-		struct amdgpu_bo *bo =
-			container_of(entry->bo, struct amdgpu_bo, tbo);
-		if (amdgpu_gem_vm_check(NULL, bo))
-			return false;
+	/* Validate and add eviction fence to DMABuf imports with dynamic
+	 * attachment in compute VMs. Re-validation will be done by
+	 * amdgpu_vm_validate. Fences are on the reservation shared with the
+	 * export, which is currently required to be validated and fenced
+	 * already by amdgpu_amdkfd_gpuvm_restore_process_bos.
+	 *
+	 * Nested locking below for the case that a GEM object is opened in
+	 * kfd_mem_export_dmabuf. Since the lock below is only taken for imports,
+	 * but not for export, this is a different lock class that cannot lead to
+	 * circular lock dependencies.
+	 */
+	if (!vm->is_compute_context || !vm->process_info)
+		return 0;
+	if (!drm_gem_is_imported(obj) ||
+	    !dma_buf_is_dynamic(obj->import_attach->dmabuf))
+		return 0;
+	mutex_lock_nested(&vm->process_info->lock, 1);
+	if (!WARN_ON(!vm->process_info->eviction_fence)) {
+		r = amdgpu_amdkfd_bo_validate_and_fence(abo, AMDGPU_GEM_DOMAIN_GTT,
+							&vm->process_info->eviction_fence->base);
+		if (r) {
+			struct amdgpu_task_info *ti = amdgpu_vm_get_task_info_vm(vm);
+
+			dev_warn(adev->dev, "validate_and_fence failed: %d\n", r);
+			if (ti) {
+				dev_warn(adev->dev, "pid %d\n", ti->task.pid);
+				amdgpu_vm_put_task_info(ti);
+			}
+		}
 	}
+	mutex_unlock(&vm->process_info->lock);
 
-	return !amdgpu_vm_validate_pt_bos(adev, vm, amdgpu_gem_vm_check, NULL);
+	return r;
 }
 
-void amdgpu_gem_object_close(struct drm_gem_object *obj,
-			     struct drm_file *file_priv)
+static void amdgpu_gem_object_close(struct drm_gem_object *obj,
+				    struct drm_file *file_priv)
 {
 	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
 	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
 	struct amdgpu_fpriv *fpriv = file_priv->driver_priv;
 	struct amdgpu_vm *vm = &fpriv->vm;
 
-	struct amdgpu_bo_list_entry vm_pd;
-	struct list_head list;
-	struct ttm_validate_buffer tv;
-	struct ww_acquire_ctx ticket;
+	struct dma_fence *fence = NULL;
 	struct amdgpu_bo_va *bo_va;
-	int r;
+	struct drm_exec exec;
+	long r;
+
+	drm_exec_init(&exec, DRM_EXEC_IGNORE_DUPLICATES, 0);
+	drm_exec_until_all_locked(&exec) {
+		r = drm_exec_prepare_obj(&exec, &bo->tbo.base, 1);
+		drm_exec_retry_on_contention(&exec);
+		if (unlikely(r))
+			goto out_unlock;
+
+		r = amdgpu_vm_lock_pd(vm, &exec, 0);
+		drm_exec_retry_on_contention(&exec);
+		if (unlikely(r))
+			goto out_unlock;
+	}
 
-	INIT_LIST_HEAD(&list);
+	if (!amdgpu_vm_is_bo_always_valid(vm, bo))
+		amdgpu_eviction_fence_detach(&fpriv->evf_mgr, bo);
 
-	tv.bo = &bo->tbo;
-	tv.shared = true;
-	list_add(&tv.head, &list);
+	bo_va = amdgpu_vm_bo_find(vm, bo);
+	if (!bo_va || --bo_va->ref_count)
+		goto out_unlock;
 
-	amdgpu_vm_get_pd_bo(vm, &list, &vm_pd);
+	amdgpu_vm_bo_del(adev, bo_va);
+	amdgpu_vm_bo_update_shared(bo);
+	if (!amdgpu_vm_ready(vm))
+		goto out_unlock;
 
-	r = ttm_eu_reserve_buffers(&ticket, &list, false, NULL);
-	if (r) {
-		dev_err(adev->dev, "leaking bo va because "
-			"we fail to reserve bo (%d)\n", r);
-		return;
-	}
-	bo_va = amdgpu_vm_bo_find(vm, bo);
-	if (bo_va && --bo_va->ref_count == 0) {
-		amdgpu_vm_bo_rmv(adev, bo_va);
+	r = amdgpu_vm_clear_freed(adev, vm, &fence);
+	if (unlikely(r < 0))
+		dev_err(adev->dev, "failed to clear page "
+			"tables on GEM object close (%ld)\n", r);
+	if (r || !fence)
+		goto out_unlock;
 
-		if (amdgpu_gem_vm_ready(adev, vm, &list)) {
-			struct dma_fence *fence = NULL;
+	amdgpu_bo_fence(bo, fence, true);
+	dma_fence_put(fence);
 
-			r = amdgpu_vm_clear_freed(adev, vm, &fence);
-			if (unlikely(r)) {
-				dev_err(adev->dev, "failed to clear page "
-					"tables on GEM object close (%d)\n", r);
-			}
+out_unlock:
+	if (r)
+		dev_err(adev->dev, "leaking bo va (%ld)\n", r);
+	drm_exec_fini(&exec);
+}
 
-			if (fence) {
-				amdgpu_bo_fence(bo, fence, true);
-				dma_fence_put(fence);
-			}
-		}
-	}
-	ttm_eu_backoff_reservation(&ticket, &list);
+static int amdgpu_gem_object_mmap(struct drm_gem_object *obj, struct vm_area_struct *vma)
+{
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+
+	if (amdgpu_ttm_tt_get_usermm(bo->tbo.ttm))
+		return -EPERM;
+	if (bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS)
+		return -EPERM;
+
+	/* Workaround for Thunk bug creating PROT_NONE,MAP_PRIVATE mappings
+	 * for debugger access to invisible VRAM. Should have used MAP_SHARED
+	 * instead. Clearing VM_MAYWRITE prevents the mapping from ever
+	 * becoming writable and makes is_cow_mapping(vm_flags) false.
+	 */
+	if (is_cow_mapping(vma->vm_flags) &&
+	    !(vma->vm_flags & VM_ACCESS_FLAGS))
+		vm_flags_clear(vma, VM_MAYWRITE);
+
+	return drm_gem_ttm_mmap(obj, vma);
 }
 
+const struct drm_gem_object_funcs amdgpu_gem_object_funcs = {
+	.free = amdgpu_gem_object_free,
+	.open = amdgpu_gem_object_open,
+	.close = amdgpu_gem_object_close,
+	.export = amdgpu_gem_prime_export,
+	.vmap = drm_gem_ttm_vmap,
+	.vunmap = drm_gem_ttm_vunmap,
+	.mmap = amdgpu_gem_object_mmap,
+	.vm_ops = &amdgpu_gem_vm_ops,
+};
+
 /*
  * GEM ioctls.
  */
 int amdgpu_gem_create_ioctl(struct drm_device *dev, void *data,
 			    struct drm_file *filp)
 {
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+	struct amdgpu_vm *vm = &fpriv->vm;
 	union drm_amdgpu_gem_create *args = data;
+	uint64_t flags = args->in.domain_flags;
 	uint64_t size = args->in.bo_size;
+	struct dma_resv *resv = NULL;
 	struct drm_gem_object *gobj;
-	uint32_t handle;
-	bool kernel = false;
+	uint32_t handle, initial_domain;
 	int r;
 
 	/* reject invalid gem flags */
-	if (args->in.domain_flags & ~(AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
-				      AMDGPU_GEM_CREATE_NO_CPU_ACCESS |
-				      AMDGPU_GEM_CREATE_CPU_GTT_USWC |
-				      AMDGPU_GEM_CREATE_VRAM_CLEARED))
+	if (flags & ~AMDGPU_GEM_CREATE_SETTABLE_MASK)
 		return -EINVAL;
 
 	/* reject invalid gem domains */
-	if (args->in.domains & ~(AMDGPU_GEM_DOMAIN_CPU |
-				 AMDGPU_GEM_DOMAIN_GTT |
-				 AMDGPU_GEM_DOMAIN_VRAM |
-				 AMDGPU_GEM_DOMAIN_GDS |
-				 AMDGPU_GEM_DOMAIN_GWS |
-				 AMDGPU_GEM_DOMAIN_OA))
+	if (args->in.domains & ~AMDGPU_GEM_DOMAIN_MASK)
+		return -EINVAL;
+
+	if (!amdgpu_is_tmz(adev) && (flags & AMDGPU_GEM_CREATE_ENCRYPTED)) {
+		DRM_NOTE_ONCE("Cannot allocate secure buffer since TMZ is disabled\n");
+		return -EINVAL;
+	}
+
+	/* always clear VRAM */
+	flags |= AMDGPU_GEM_CREATE_VRAM_CLEARED;
+
+	if (args->in.domains & AMDGPU_GEM_DOMAIN_MMIO_REMAP)
 		return -EINVAL;
 
 	/* create a gem object to contain this object in */
 	if (args->in.domains & (AMDGPU_GEM_DOMAIN_GDS |
 	    AMDGPU_GEM_DOMAIN_GWS | AMDGPU_GEM_DOMAIN_OA)) {
-		kernel = true;
-		if (args->in.domains == AMDGPU_GEM_DOMAIN_GDS)
-			size = size << AMDGPU_GDS_SHIFT;
-		else if (args->in.domains == AMDGPU_GEM_DOMAIN_GWS)
-			size = size << AMDGPU_GWS_SHIFT;
-		else if (args->in.domains == AMDGPU_GEM_DOMAIN_OA)
-			size = size << AMDGPU_OA_SHIFT;
-		else
+		if (flags & AMDGPU_GEM_CREATE_VM_ALWAYS_VALID) {
+			/* if gds bo is created from user space, it must be
+			 * passed to bo list
+			 */
+			DRM_ERROR("GDS bo cannot be per-vm-bo\n");
 			return -EINVAL;
+		}
+		flags |= AMDGPU_GEM_CREATE_NO_CPU_ACCESS;
 	}
-	size = roundup(size, PAGE_SIZE);
 
+	if (flags & AMDGPU_GEM_CREATE_VM_ALWAYS_VALID) {
+		r = amdgpu_bo_reserve(vm->root.bo, false);
+		if (r)
+			return r;
+
+		resv = vm->root.bo->tbo.base.resv;
+	}
+
+	initial_domain = (u32)(0xffffffff & args->in.domains);
+retry:
 	r = amdgpu_gem_object_create(adev, size, args->in.alignment,
-				     (u32)(0xffffffff & args->in.domains),
-				     args->in.domain_flags,
-				     kernel, &gobj);
+				     initial_domain,
+				     flags, ttm_bo_type_device, resv, &gobj, fpriv->xcp_id + 1);
+	if (r && r != -ERESTARTSYS) {
+		if (flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED) {
+			flags &= ~AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
+			goto retry;
+		}
+
+		if (initial_domain == AMDGPU_GEM_DOMAIN_VRAM) {
+			initial_domain |= AMDGPU_GEM_DOMAIN_GTT;
+			goto retry;
+		}
+		DRM_DEBUG("Failed to allocate GEM object (%llu, %d, %llu, %d)\n",
+				size, initial_domain, args->in.alignment, r);
+	}
+
+	if (flags & AMDGPU_GEM_CREATE_VM_ALWAYS_VALID) {
+		if (!r) {
+			struct amdgpu_bo *abo = gem_to_amdgpu_bo(gobj);
+
+			abo->parent = amdgpu_bo_ref(vm->root.bo);
+		}
+		amdgpu_bo_unreserve(vm->root.bo);
+	}
 	if (r)
 		return r;
 
 	r = drm_gem_handle_create(filp, gobj, &handle);
 	/* drop reference from allocate - handle holds it now */
-	drm_gem_object_put_unlocked(gobj);
+	drm_gem_object_put(gobj);
 	if (r)
 		return r;
 
@@ -273,13 +526,18 @@ int amdgpu_gem_create_ioctl(struct drm_device *dev, void *data,
 int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
 			     struct drm_file *filp)
 {
-	struct amdgpu_device *adev = dev->dev_private;
+	struct ttm_operation_ctx ctx = { true, false };
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct drm_amdgpu_gem_userptr *args = data;
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
 	struct drm_gem_object *gobj;
+	struct hmm_range *range;
 	struct amdgpu_bo *bo;
 	uint32_t handle;
 	int r;
 
+	args->addr = untagged_addr(args->addr);
+
 	if (offset_in_page(args->addr | args->size))
 		return -EINVAL;
 
@@ -297,63 +555,52 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
 	}
 
 	/* create a gem object to contain this object in */
-	r = amdgpu_gem_object_create(adev, args->size, 0,
-				     AMDGPU_GEM_DOMAIN_CPU, 0,
-				     0, &gobj);
+	r = amdgpu_gem_object_create(adev, args->size, 0, AMDGPU_GEM_DOMAIN_CPU,
+				     0, ttm_bo_type_device, NULL, &gobj, fpriv->xcp_id + 1);
 	if (r)
 		return r;
 
 	bo = gem_to_amdgpu_bo(gobj);
 	bo->preferred_domains = AMDGPU_GEM_DOMAIN_GTT;
 	bo->allowed_domains = AMDGPU_GEM_DOMAIN_GTT;
-	r = amdgpu_ttm_tt_set_userptr(bo->tbo.ttm, args->addr, args->flags);
+	r = amdgpu_ttm_tt_set_userptr(&bo->tbo, args->addr, args->flags);
 	if (r)
 		goto release_object;
 
-	if (args->flags & AMDGPU_GEM_USERPTR_REGISTER) {
-		r = amdgpu_mn_register(bo, args->addr);
-		if (r)
-			goto release_object;
-	}
+	r = amdgpu_hmm_register(bo, args->addr);
+	if (r)
+		goto release_object;
 
 	if (args->flags & AMDGPU_GEM_USERPTR_VALIDATE) {
-		down_read(&current->mm->mmap_sem);
-
-		r = amdgpu_ttm_tt_get_user_pages(bo->tbo.ttm,
-						 bo->tbo.ttm->pages);
+		r = amdgpu_ttm_tt_get_user_pages(bo, &range);
 		if (r)
-			goto unlock_mmap_sem;
+			goto release_object;
 
 		r = amdgpu_bo_reserve(bo, true);
 		if (r)
-			goto free_pages;
+			goto user_pages_done;
+
+		amdgpu_ttm_tt_set_user_pages(bo->tbo.ttm, range);
 
-		amdgpu_ttm_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_GTT);
-		r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false);
+		amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_GTT);
+		r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
 		amdgpu_bo_unreserve(bo);
 		if (r)
-			goto free_pages;
-
-		up_read(&current->mm->mmap_sem);
+			goto user_pages_done;
 	}
 
 	r = drm_gem_handle_create(filp, gobj, &handle);
-	/* drop reference from allocate - handle holds it now */
-	drm_gem_object_put_unlocked(gobj);
 	if (r)
-		return r;
+		goto user_pages_done;
 
 	args->handle = handle;
-	return 0;
-
-free_pages:
-	release_pages(bo->tbo.ttm->pages, bo->tbo.ttm->num_pages, false);
 
-unlock_mmap_sem:
-	up_read(&current->mm->mmap_sem);
+user_pages_done:
+	if (args->flags & AMDGPU_GEM_USERPTR_VALIDATE)
+		amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm, range);
 
 release_object:
-	drm_gem_object_put_unlocked(gobj);
+	drm_gem_object_put(gobj);
 
 	return r;
 }
@@ -366,17 +613,17 @@ int amdgpu_mode_dumb_mmap(struct drm_file *filp,
 	struct amdgpu_bo *robj;
 
 	gobj = drm_gem_object_lookup(filp, handle);
-	if (gobj == NULL) {
+	if (!gobj)
 		return -ENOENT;
-	}
+
 	robj = gem_to_amdgpu_bo(gobj);
 	if (amdgpu_ttm_tt_get_usermm(robj->tbo.ttm) ||
 	    (robj->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS)) {
-		drm_gem_object_put_unlocked(gobj);
+		drm_gem_object_put(gobj);
 		return -EPERM;
 	}
 	*offset_p = amdgpu_bo_mmap_offset(robj);
-	drm_gem_object_put_unlocked(gobj);
+	drm_gem_object_put(gobj);
 	return 0;
 }
 
@@ -385,6 +632,7 @@ int amdgpu_gem_mmap_ioctl(struct drm_device *dev, void *data,
 {
 	union drm_amdgpu_gem_mmap *args = data;
 	uint32_t handle = args->in.handle;
+
 	memset(args, 0, sizeof(*args));
 	return amdgpu_mode_dumb_mmap(filp, dev, handle, &args->out.addr_ptr);
 }
@@ -411,7 +659,7 @@ unsigned long amdgpu_gem_timeout(uint64_t timeout_ns)
 
 	timeout_jiffies = nsecs_to_jiffies(ktime_to_ns(timeout));
 	/*  clamp timeout to avoid unsigned-> signed overflow */
-	if (timeout_jiffies > MAX_SCHEDULE_TIMEOUT )
+	if (timeout_jiffies > MAX_SCHEDULE_TIMEOUT)
 		return MAX_SCHEDULE_TIMEOUT - 1;
 
 	return timeout_jiffies;
@@ -429,12 +677,12 @@ int amdgpu_gem_wait_idle_ioctl(struct drm_device *dev, void *data,
 	long ret;
 
 	gobj = drm_gem_object_lookup(filp, handle);
-	if (gobj == NULL) {
+	if (!gobj)
 		return -ENOENT;
-	}
+
 	robj = gem_to_amdgpu_bo(gobj);
-	ret = reservation_object_wait_timeout_rcu(robj->tbo.resv, true, true,
-						  timeout);
+	ret = dma_resv_wait_timeout(robj->tbo.base.resv, DMA_RESV_USAGE_READ,
+				    true, timeout);
 
 	/* ret == 0 means not signaled,
 	 * ret > 0 means signaled
@@ -446,7 +694,7 @@ int amdgpu_gem_wait_idle_ioctl(struct drm_device *dev, void *data,
 	} else
 		r = ret;
 
-	drm_gem_object_put_unlocked(gobj);
+	drm_gem_object_put(gobj);
 	return r;
 }
 
@@ -458,7 +706,7 @@ int amdgpu_gem_metadata_ioctl(struct drm_device *dev, void *data,
 	struct amdgpu_bo *robj;
 	int r = -1;
 
-	DRM_DEBUG("%d \n", args->handle);
+	DRM_DEBUG("%d\n", args->handle);
 	gobj = drm_gem_object_lookup(filp, args->handle);
 	if (gobj == NULL)
 		return -ENOENT;
@@ -489,7 +737,7 @@ int amdgpu_gem_metadata_ioctl(struct drm_device *dev, void *data,
 unreserve:
 	amdgpu_bo_unreserve(robj);
 out:
-	drm_gem_object_put_unlocked(gobj);
+	drm_gem_object_put(gobj);
 	return r;
 }
 
@@ -499,38 +747,44 @@ out:
  * @adev: amdgpu_device pointer
  * @vm: vm to update
  * @bo_va: bo_va to update
- * @list: validation list
  * @operation: map, unmap or clear
  *
  * Update the bo_va directly after setting its address. Errors are not
  * vital here, so they are not reported back to userspace.
+ *
+ * Returns resulting fence if freed BO(s) got cleared from the PT.
+ * otherwise stub fence in case of error.
  */
-static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev,
-				    struct amdgpu_vm *vm,
-				    struct amdgpu_bo_va *bo_va,
-				    struct list_head *list,
-				    uint32_t operation)
+static struct dma_fence *
+amdgpu_gem_va_update_vm(struct amdgpu_device *adev,
+			struct amdgpu_vm *vm,
+			struct amdgpu_bo_va *bo_va,
+			uint32_t operation)
 {
-	int r = -ERESTARTSYS;
+	struct dma_fence *fence = dma_fence_get_stub();
+	int r;
 
-	if (!amdgpu_gem_vm_ready(adev, vm, list))
-		goto error;
+	if (!amdgpu_vm_ready(vm))
+		return fence;
 
-	r = amdgpu_vm_update_directories(adev, vm);
-	if (r)
-		goto error;
-
-	r = amdgpu_vm_clear_freed(adev, vm, NULL);
+	r = amdgpu_vm_clear_freed(adev, vm, &fence);
 	if (r)
 		goto error;
 
 	if (operation == AMDGPU_VA_OP_MAP ||
-	    operation == AMDGPU_VA_OP_REPLACE)
+	    operation == AMDGPU_VA_OP_REPLACE) {
 		r = amdgpu_vm_bo_update(adev, bo_va, false);
+		if (r)
+			goto error;
+	}
+
+	r = amdgpu_vm_update_pdes(adev, vm, false);
 
 error:
 	if (r && r != -ERESTARTSYS)
 		DRM_ERROR("Couldn't update BO_VA (%d)\n", r);
+
+	return fence;
 }
 
 int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
@@ -538,33 +792,53 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 {
 	const uint32_t valid_flags = AMDGPU_VM_DELAY_UPDATE |
 		AMDGPU_VM_PAGE_READABLE | AMDGPU_VM_PAGE_WRITEABLE |
-		AMDGPU_VM_PAGE_EXECUTABLE | AMDGPU_VM_MTYPE_MASK;
+		AMDGPU_VM_PAGE_EXECUTABLE | AMDGPU_VM_MTYPE_MASK |
+		AMDGPU_VM_PAGE_NOALLOC;
 	const uint32_t prt_flags = AMDGPU_VM_DELAY_UPDATE |
 		AMDGPU_VM_PAGE_PRT;
 
 	struct drm_amdgpu_gem_va *args = data;
 	struct drm_gem_object *gobj;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_fpriv *fpriv = filp->driver_priv;
 	struct amdgpu_bo *abo;
 	struct amdgpu_bo_va *bo_va;
-	struct amdgpu_bo_list_entry vm_pd;
-	struct ttm_validate_buffer tv;
-	struct ww_acquire_ctx ticket;
-	struct list_head list;
-	uint64_t va_flags;
+	struct drm_syncobj *timeline_syncobj = NULL;
+	struct dma_fence_chain *timeline_chain = NULL;
+	struct dma_fence *fence;
+	struct drm_exec exec;
+	uint64_t vm_size;
 	int r = 0;
 
-	if (args->va_address < AMDGPU_VA_RESERVED_SIZE) {
-		dev_err(&dev->pdev->dev,
-			"va_address 0x%lX is in reserved area 0x%X\n",
-			(unsigned long)args->va_address,
-			AMDGPU_VA_RESERVED_SIZE);
+	if (args->va_address < AMDGPU_VA_RESERVED_BOTTOM) {
+		dev_dbg(dev->dev,
+			"va_address 0x%llx is in reserved area 0x%llx\n",
+			args->va_address, AMDGPU_VA_RESERVED_BOTTOM);
+		return -EINVAL;
+	}
+
+	if (args->va_address >= AMDGPU_GMC_HOLE_START &&
+	    args->va_address < AMDGPU_GMC_HOLE_END) {
+		dev_dbg(dev->dev,
+			"va_address 0x%llx is in VA hole 0x%llx-0x%llx\n",
+			args->va_address, AMDGPU_GMC_HOLE_START,
+			AMDGPU_GMC_HOLE_END);
+		return -EINVAL;
+	}
+
+	args->va_address &= AMDGPU_GMC_HOLE_MASK;
+
+	vm_size = adev->vm_manager.max_pfn * AMDGPU_GPU_PAGE_SIZE;
+	vm_size -= AMDGPU_VA_RESERVED_TOP;
+	if (args->va_address + args->map_size > vm_size) {
+		dev_dbg(dev->dev,
+			"va_address 0x%llx is in top reserved area 0x%llx\n",
+			args->va_address + args->map_size, vm_size);
 		return -EINVAL;
 	}
 
 	if ((args->flags & ~valid_flags) && (args->flags & ~prt_flags)) {
-		dev_err(&dev->pdev->dev, "invalid flags combination 0x%08X\n",
+		dev_dbg(dev->dev, "invalid flags combination 0x%08X\n",
 			args->flags);
 		return -EINVAL;
 	}
@@ -576,42 +850,49 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 	case AMDGPU_VA_OP_REPLACE:
 		break;
 	default:
-		dev_err(&dev->pdev->dev, "unsupported operation %d\n",
+		dev_dbg(dev->dev, "unsupported operation %d\n",
 			args->operation);
 		return -EINVAL;
 	}
-	if ((args->operation == AMDGPU_VA_OP_MAP) ||
-	    (args->operation == AMDGPU_VA_OP_REPLACE)) {
-		if (amdgpu_kms_vram_lost(adev, fpriv))
-			return -ENODEV;
-	}
 
-	INIT_LIST_HEAD(&list);
 	if ((args->operation != AMDGPU_VA_OP_CLEAR) &&
 	    !(args->flags & AMDGPU_VM_PAGE_PRT)) {
 		gobj = drm_gem_object_lookup(filp, args->handle);
 		if (gobj == NULL)
 			return -ENOENT;
 		abo = gem_to_amdgpu_bo(gobj);
-		tv.bo = &abo->tbo;
-		tv.shared = false;
-		list_add(&tv.head, &list);
 	} else {
 		gobj = NULL;
 		abo = NULL;
 	}
 
-	amdgpu_vm_get_pd_bo(&fpriv->vm, &list, &vm_pd);
-
-	r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL);
+	r = amdgpu_gem_add_input_fence(filp,
+				       args->input_fence_syncobj_handles,
+				       args->num_syncobj_handles);
 	if (r)
-		goto error_unref;
+		goto error_put_gobj;
+
+	drm_exec_init(&exec, DRM_EXEC_INTERRUPTIBLE_WAIT |
+		      DRM_EXEC_IGNORE_DUPLICATES, 0);
+	drm_exec_until_all_locked(&exec) {
+		if (gobj) {
+			r = drm_exec_lock_obj(&exec, gobj);
+			drm_exec_retry_on_contention(&exec);
+			if (unlikely(r))
+				goto error;
+		}
+
+		r = amdgpu_vm_lock_pd(&fpriv->vm, &exec, 2);
+		drm_exec_retry_on_contention(&exec);
+		if (unlikely(r))
+			goto error;
+	}
 
 	if (abo) {
 		bo_va = amdgpu_vm_bo_find(&fpriv->vm, abo);
 		if (!bo_va) {
 			r = -ENOENT;
-			goto error_backoff;
+			goto error;
 		}
 	} else if (args->operation != AMDGPU_VA_OP_CLEAR) {
 		bo_va = fpriv->prt_va;
@@ -619,17 +900,19 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 		bo_va = NULL;
 	}
 
+	r = amdgpu_gem_update_timeline_node(filp,
+					    args->vm_timeline_syncobj_out,
+					    args->vm_timeline_point,
+					    &timeline_syncobj,
+					    &timeline_chain);
+	if (r)
+		goto error;
+
 	switch (args->operation) {
 	case AMDGPU_VA_OP_MAP:
-		r = amdgpu_vm_alloc_pts(adev, bo_va->base.vm, args->va_address,
-					args->map_size);
-		if (r)
-			goto error_backoff;
-
-		va_flags = amdgpu_vm_get_pte_flags(adev, args->flags);
 		r = amdgpu_vm_bo_map(adev, bo_va, args->va_address,
 				     args->offset_in_bo, args->map_size,
-				     va_flags);
+				     args->flags);
 		break;
 	case AMDGPU_VA_OP_UNMAP:
 		r = amdgpu_vm_bo_unmap(adev, bo_va, args->va_address);
@@ -641,28 +924,32 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 						args->map_size);
 		break;
 	case AMDGPU_VA_OP_REPLACE:
-		r = amdgpu_vm_alloc_pts(adev, bo_va->base.vm, args->va_address,
-					args->map_size);
-		if (r)
-			goto error_backoff;
-
-		va_flags = amdgpu_vm_get_pte_flags(adev, args->flags);
 		r = amdgpu_vm_bo_replace_map(adev, bo_va, args->va_address,
 					     args->offset_in_bo, args->map_size,
-					     va_flags);
+					     args->flags);
 		break;
 	default:
 		break;
 	}
-	if (!r && !(args->flags & AMDGPU_VM_DELAY_UPDATE) && !amdgpu_vm_debug)
-		amdgpu_gem_va_update_vm(adev, &fpriv->vm, bo_va, &list,
-					args->operation);
+	if (!r && !(args->flags & AMDGPU_VM_DELAY_UPDATE) && !adev->debug_vm) {
+		fence = amdgpu_gem_va_update_vm(adev, &fpriv->vm, bo_va,
+						args->operation);
+
+		if (timeline_syncobj)
+			amdgpu_gem_update_bo_mapping(filp, bo_va,
+					     args->operation,
+					     args->vm_timeline_point,
+					     fence, timeline_syncobj,
+					     timeline_chain);
+		else
+			dma_fence_put(fence);
 
-error_backoff:
-	ttm_eu_backoff_reservation(&ticket, &list);
+	}
 
-error_unref:
-	drm_gem_object_put_unlocked(gobj);
+error:
+	drm_exec_fini(&exec);
+error_put_gobj:
+	drm_gem_object_put(gobj);
 	return r;
 }
 
@@ -671,44 +958,69 @@ int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,
 {
 	struct drm_amdgpu_gem_op *args = data;
 	struct drm_gem_object *gobj;
+	struct amdgpu_vm_bo_base *base;
 	struct amdgpu_bo *robj;
+	struct drm_exec exec;
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
 	int r;
 
+	if (args->padding)
+		return -EINVAL;
+
 	gobj = drm_gem_object_lookup(filp, args->handle);
-	if (gobj == NULL) {
+	if (!gobj)
 		return -ENOENT;
-	}
+
 	robj = gem_to_amdgpu_bo(gobj);
 
-	r = amdgpu_bo_reserve(robj, false);
-	if (unlikely(r))
-		goto out;
+	drm_exec_init(&exec, DRM_EXEC_INTERRUPTIBLE_WAIT |
+			  DRM_EXEC_IGNORE_DUPLICATES, 0);
+	drm_exec_until_all_locked(&exec) {
+		r = drm_exec_lock_obj(&exec, gobj);
+		drm_exec_retry_on_contention(&exec);
+		if (r)
+			goto out_exec;
+
+		if (args->op == AMDGPU_GEM_OP_GET_MAPPING_INFO) {
+			r = amdgpu_vm_lock_pd(&fpriv->vm, &exec, 0);
+			drm_exec_retry_on_contention(&exec);
+			if (r)
+				goto out_exec;
+		}
+	}
 
 	switch (args->op) {
 	case AMDGPU_GEM_OP_GET_GEM_CREATE_INFO: {
 		struct drm_amdgpu_gem_create_in info;
 		void __user *out = u64_to_user_ptr(args->value);
 
-		info.bo_size = robj->gem_base.size;
-		info.alignment = robj->tbo.mem.page_alignment << PAGE_SHIFT;
+		info.bo_size = robj->tbo.base.size;
+		info.alignment = robj->tbo.page_alignment << PAGE_SHIFT;
 		info.domains = robj->preferred_domains;
 		info.domain_flags = robj->flags;
-		amdgpu_bo_unreserve(robj);
+		drm_exec_fini(&exec);
 		if (copy_to_user(out, &info, sizeof(info)))
 			r = -EFAULT;
 		break;
 	}
 	case AMDGPU_GEM_OP_SET_PLACEMENT:
-		if (robj->prime_shared_count && (args->value & AMDGPU_GEM_DOMAIN_VRAM)) {
+		if (drm_gem_is_imported(&robj->tbo.base) &&
+		    args->value & AMDGPU_GEM_DOMAIN_VRAM) {
 			r = -EINVAL;
-			amdgpu_bo_unreserve(robj);
-			break;
+			goto out_exec;
 		}
 		if (amdgpu_ttm_tt_get_usermm(robj->tbo.ttm)) {
 			r = -EPERM;
-			amdgpu_bo_unreserve(robj);
-			break;
+			goto out_exec;
 		}
+		for (base = robj->vm_bo; base; base = base->next)
+			if (amdgpu_xgmi_same_hive(amdgpu_ttm_adev(robj->tbo.bdev),
+				amdgpu_ttm_adev(base->vm->root.bo->tbo.bdev))) {
+				r = -EINVAL;
+				goto out_exec;
+			}
+
+
 		robj->preferred_domains = args->value & (AMDGPU_GEM_DOMAIN_VRAM |
 							AMDGPU_GEM_DOMAIN_GTT |
 							AMDGPU_GEM_DOMAIN_CPU);
@@ -716,94 +1028,224 @@ int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,
 		if (robj->allowed_domains == AMDGPU_GEM_DOMAIN_VRAM)
 			robj->allowed_domains |= AMDGPU_GEM_DOMAIN_GTT;
 
-		amdgpu_bo_unreserve(robj);
+		if (robj->flags & AMDGPU_GEM_CREATE_VM_ALWAYS_VALID)
+			amdgpu_vm_bo_invalidate(robj, true);
+		drm_exec_fini(&exec);
 		break;
+	case AMDGPU_GEM_OP_GET_MAPPING_INFO: {
+		struct amdgpu_bo_va *bo_va = amdgpu_vm_bo_find(&fpriv->vm, robj);
+		struct drm_amdgpu_gem_vm_entry *vm_entries;
+		struct amdgpu_bo_va_mapping *mapping;
+		int num_mappings = 0;
+		/*
+		 * num_entries is set as an input to the size of the user-allocated array of
+		 * drm_amdgpu_gem_vm_entry stored at args->value.
+		 * num_entries is sent back as output as the number of mappings the bo has.
+		 * If that number is larger than the size of the array, the ioctl must
+		 * be retried.
+		 */
+		vm_entries = kvcalloc(args->num_entries, sizeof(*vm_entries), GFP_KERNEL);
+		if (!vm_entries)
+			return -ENOMEM;
+
+		amdgpu_vm_bo_va_for_each_valid_mapping(bo_va, mapping) {
+			if (num_mappings < args->num_entries) {
+				vm_entries[num_mappings].addr = mapping->start * AMDGPU_GPU_PAGE_SIZE;
+				vm_entries[num_mappings].size = (mapping->last - mapping->start + 1) * AMDGPU_GPU_PAGE_SIZE;
+				vm_entries[num_mappings].offset = mapping->offset;
+				vm_entries[num_mappings].flags = mapping->flags;
+			}
+			num_mappings += 1;
+		}
+
+		amdgpu_vm_bo_va_for_each_invalid_mapping(bo_va, mapping) {
+			if (num_mappings < args->num_entries) {
+				vm_entries[num_mappings].addr = mapping->start * AMDGPU_GPU_PAGE_SIZE;
+				vm_entries[num_mappings].size = (mapping->last - mapping->start + 1) * AMDGPU_GPU_PAGE_SIZE;
+				vm_entries[num_mappings].offset = mapping->offset;
+				vm_entries[num_mappings].flags = mapping->flags;
+			}
+			num_mappings += 1;
+		}
+
+		drm_exec_fini(&exec);
+
+		if (num_mappings > 0 && num_mappings <= args->num_entries)
+			if (copy_to_user(u64_to_user_ptr(args->value), vm_entries, num_mappings * sizeof(*vm_entries)))
+				r = -EFAULT;
+
+		args->num_entries = num_mappings;
+
+		kvfree(vm_entries);
+		break;
+	}
 	default:
-		amdgpu_bo_unreserve(robj);
+		drm_exec_fini(&exec);
 		r = -EINVAL;
 	}
 
-out:
-	drm_gem_object_put_unlocked(gobj);
+	drm_gem_object_put(gobj);
+	return r;
+out_exec:
+	drm_exec_fini(&exec);
+	drm_gem_object_put(gobj);
 	return r;
 }
 
+/**
+ * amdgpu_gem_list_handles_ioctl - get information about a process' buffer objects
+ *
+ * @dev: drm device pointer
+ * @data: drm_amdgpu_gem_list_handles
+ * @filp: drm file pointer
+ *
+ * num_entries is set as an input to the size of the entries array.
+ * num_entries is sent back as output as the number of bos in the process.
+ * If that number is larger than the size of the array, the ioctl must
+ * be retried.
+ *
+ * Returns:
+ * 0 for success, -errno for errors.
+ */
+int amdgpu_gem_list_handles_ioctl(struct drm_device *dev, void *data,
+				  struct drm_file *filp)
+{
+	struct drm_amdgpu_gem_list_handles *args = data;
+	struct drm_amdgpu_gem_list_handles_entry *bo_entries;
+	struct drm_gem_object *gobj;
+	int id, ret = 0;
+	int bo_index = 0;
+	int num_bos = 0;
+
+	spin_lock(&filp->table_lock);
+	idr_for_each_entry(&filp->object_idr, gobj, id)
+		num_bos += 1;
+	spin_unlock(&filp->table_lock);
+
+	if (args->num_entries < num_bos) {
+		args->num_entries = num_bos;
+		return 0;
+	}
+
+	if (num_bos == 0) {
+		args->num_entries = 0;
+		return 0;
+	}
+
+	bo_entries = kvcalloc(num_bos, sizeof(*bo_entries), GFP_KERNEL);
+	if (!bo_entries)
+		return -ENOMEM;
+
+	spin_lock(&filp->table_lock);
+	idr_for_each_entry(&filp->object_idr, gobj, id) {
+		struct amdgpu_bo *bo = gem_to_amdgpu_bo(gobj);
+		struct drm_amdgpu_gem_list_handles_entry *bo_entry;
+
+		if (bo_index >= num_bos) {
+			ret = -EAGAIN;
+			break;
+		}
+
+		bo_entry = &bo_entries[bo_index];
+
+		bo_entry->size = amdgpu_bo_size(bo);
+		bo_entry->alloc_flags = bo->flags & AMDGPU_GEM_CREATE_SETTABLE_MASK;
+		bo_entry->preferred_domains = bo->preferred_domains;
+		bo_entry->gem_handle = id;
+		bo_entry->alignment = bo->tbo.page_alignment;
+
+		if (bo->tbo.base.import_attach)
+			bo_entry->flags |= AMDGPU_GEM_LIST_HANDLES_FLAG_IS_IMPORT;
+
+		bo_index += 1;
+	}
+	spin_unlock(&filp->table_lock);
+
+	args->num_entries = bo_index;
+
+	if (!ret)
+		if (copy_to_user(u64_to_user_ptr(args->entries), bo_entries, num_bos * sizeof(*bo_entries)))
+			ret = -EFAULT;
+
+	kvfree(bo_entries);
+
+	return ret;
+}
+
+static int amdgpu_gem_align_pitch(struct amdgpu_device *adev,
+				  int width,
+				  int cpp,
+				  bool tiled)
+{
+	int aligned = width;
+	int pitch_mask = 0;
+
+	switch (cpp) {
+	case 1:
+		pitch_mask = 255;
+		break;
+	case 2:
+		pitch_mask = 127;
+		break;
+	case 3:
+	case 4:
+		pitch_mask = 63;
+		break;
+	}
+
+	aligned += pitch_mask;
+	aligned &= ~pitch_mask;
+	return aligned * cpp;
+}
+
 int amdgpu_mode_dumb_create(struct drm_file *file_priv,
 			    struct drm_device *dev,
 			    struct drm_mode_create_dumb *args)
 {
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	struct amdgpu_fpriv *fpriv = file_priv->driver_priv;
 	struct drm_gem_object *gobj;
 	uint32_t handle;
+	u64 flags = AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
+		    AMDGPU_GEM_CREATE_CPU_GTT_USWC |
+		    AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+	u32 domain;
 	int r;
 
-	args->pitch = amdgpu_align_pitch(adev, args->width,
-					 DIV_ROUND_UP(args->bpp, 8), 0);
+	/*
+	 * The buffer returned from this function should be cleared, but
+	 * it can only be done if the ring is enabled or we'll fail to
+	 * create the buffer.
+	 */
+	if (adev->mman.buffer_funcs_enabled)
+		flags |= AMDGPU_GEM_CREATE_VRAM_CLEARED;
+
+	args->pitch = amdgpu_gem_align_pitch(adev, args->width,
+					     DIV_ROUND_UP(args->bpp, 8), 0);
 	args->size = (u64)args->pitch * args->height;
 	args->size = ALIGN(args->size, PAGE_SIZE);
-
-	r = amdgpu_gem_object_create(adev, args->size, 0,
-				     AMDGPU_GEM_DOMAIN_VRAM,
-				     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
-				     ttm_bo_type_device,
-				     &gobj);
+	domain = amdgpu_bo_get_preferred_domain(adev,
+				amdgpu_display_supported_domains(adev, flags));
+	r = amdgpu_gem_object_create(adev, args->size, 0, domain, flags,
+				     ttm_bo_type_device, NULL, &gobj, fpriv->xcp_id + 1);
 	if (r)
 		return -ENOMEM;
 
 	r = drm_gem_handle_create(file_priv, gobj, &handle);
 	/* drop reference from allocate - handle holds it now */
-	drm_gem_object_put_unlocked(gobj);
-	if (r) {
+	drm_gem_object_put(gobj);
+	if (r)
 		return r;
-	}
+
 	args->handle = handle;
 	return 0;
 }
 
 #if defined(CONFIG_DEBUG_FS)
-static int amdgpu_debugfs_gem_bo_info(int id, void *ptr, void *data)
-{
-	struct drm_gem_object *gobj = ptr;
-	struct amdgpu_bo *bo = gem_to_amdgpu_bo(gobj);
-	struct seq_file *m = data;
-
-	unsigned domain;
-	const char *placement;
-	unsigned pin_count;
-	uint64_t offset;
-
-	domain = amdgpu_mem_type_to_domain(bo->tbo.mem.mem_type);
-	switch (domain) {
-	case AMDGPU_GEM_DOMAIN_VRAM:
-		placement = "VRAM";
-		break;
-	case AMDGPU_GEM_DOMAIN_GTT:
-		placement = " GTT";
-		break;
-	case AMDGPU_GEM_DOMAIN_CPU:
-	default:
-		placement = " CPU";
-		break;
-	}
-	seq_printf(m, "\t0x%08x: %12ld byte %s",
-		   id, amdgpu_bo_size(bo), placement);
-
-	offset = ACCESS_ONCE(bo->tbo.mem.start);
-	if (offset != AMDGPU_BO_INVALID_OFFSET)
-		seq_printf(m, " @ 0x%010Lx", offset);
-
-	pin_count = ACCESS_ONCE(bo->pin_count);
-	if (pin_count)
-		seq_printf(m, " pin count %d", pin_count);
-	seq_printf(m, "\n");
-
-	return 0;
-}
-
-static int amdgpu_debugfs_gem_info(struct seq_file *m, void *data)
+static int amdgpu_debugfs_gem_info_show(struct seq_file *m, void *unused)
 {
-	struct drm_info_node *node = (struct drm_info_node *)m->private;
-	struct drm_device *dev = node->minor->dev;
+	struct amdgpu_device *adev = m->private;
+	struct drm_device *dev = adev_to_drm(adev);
 	struct drm_file *file;
 	int r;
 
@@ -813,6 +1255,9 @@ static int amdgpu_debugfs_gem_info(struct seq_file *m, void *data)
 
 	list_for_each_entry(file, &dev->filelist, lhead) {
 		struct task_struct *task;
+		struct drm_gem_object *gobj;
+		struct pid *pid;
+		int id;
 
 		/*
 		 * Although we have a valid reference on file->pid, that does
@@ -821,13 +1266,18 @@ static int amdgpu_debugfs_gem_info(struct seq_file *m, void *data)
 		 * Therefore, we need to protect this ->comm access using RCU.
 		 */
 		rcu_read_lock();
-		task = pid_task(file->pid, PIDTYPE_PID);
-		seq_printf(m, "pid %8d command %s:\n", pid_nr(file->pid),
+		pid = rcu_dereference(file->pid);
+		task = pid_task(pid, PIDTYPE_TGID);
+		seq_printf(m, "pid %8d command %s:\n", pid_nr(pid),
 			   task ? task->comm : "<unknown>");
 		rcu_read_unlock();
 
 		spin_lock(&file->table_lock);
-		idr_for_each(&file->object_idr, amdgpu_debugfs_gem_bo_info, m);
+		idr_for_each_entry(&file->object_idr, gobj, id) {
+			struct amdgpu_bo *bo = gem_to_amdgpu_bo(gobj);
+
+			amdgpu_bo_print_info(id, bo, m);
+		}
 		spin_unlock(&file->table_lock);
 	}
 
@@ -835,15 +1285,17 @@ static int amdgpu_debugfs_gem_info(struct seq_file *m, void *data)
 	return 0;
 }
 
-static const struct drm_info_list amdgpu_debugfs_gem_list[] = {
-	{"amdgpu_gem_info", &amdgpu_debugfs_gem_info, 0, NULL},
-};
+DEFINE_SHOW_ATTRIBUTE(amdgpu_debugfs_gem_info);
+
 #endif
 
-int amdgpu_gem_debugfs_init(struct amdgpu_device *adev)
+void amdgpu_debugfs_gem_init(struct amdgpu_device *adev)
 {
 #if defined(CONFIG_DEBUG_FS)
-	return amdgpu_debugfs_add_files(adev, amdgpu_debugfs_gem_list, 1);
+	struct drm_minor *minor = adev_to_drm(adev)->primary;
+	struct dentry *root = minor->debugfs_root;
+
+	debugfs_create_file("amdgpu_gem_info", 0444, root, adev,
+			    &amdgpu_debugfs_gem_info_fops);
 #endif
-	return 0;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.h
new file mode 100644
index 000000000000..b558336bc4c6
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.h
@@ -0,0 +1,90 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __AMDGPU_GEM_H__
+#define __AMDGPU_GEM_H__
+
+#include <drm/amdgpu_drm.h>
+#include <drm/drm_gem.h>
+
+/*
+ * GEM.
+ */
+
+#define AMDGPU_GEM_DOMAIN_MAX		0x3
+#define gem_to_amdgpu_bo(gobj) container_of((gobj), struct amdgpu_bo, tbo.base)
+
+extern const struct drm_gem_object_funcs amdgpu_gem_object_funcs;
+
+unsigned long amdgpu_gem_timeout(uint64_t timeout_ns);
+
+/*
+ * GEM objects.
+ */
+void amdgpu_gem_force_release(struct amdgpu_device *adev);
+int amdgpu_gem_object_create(struct amdgpu_device *adev, unsigned long size,
+			     int alignment, u32 initial_domain,
+			     u64 flags, enum ttm_bo_type type,
+			     struct dma_resv *resv,
+			     struct drm_gem_object **obj, int8_t xcp_id_plus1);
+int amdgpu_mode_dumb_create(struct drm_file *file_priv,
+			    struct drm_device *dev,
+			    struct drm_mode_create_dumb *args);
+int amdgpu_mode_dumb_mmap(struct drm_file *filp,
+			  struct drm_device *dev,
+			  uint32_t handle, uint64_t *offset_p);
+
+int amdgpu_gem_create_ioctl(struct drm_device *dev, void *data,
+			    struct drm_file *filp);
+int amdgpu_gem_info_ioctl(struct drm_device *dev, void *data,
+			  struct drm_file *filp);
+int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
+			struct drm_file *filp);
+int amdgpu_gem_mmap_ioctl(struct drm_device *dev, void *data,
+			  struct drm_file *filp);
+int amdgpu_gem_wait_idle_ioctl(struct drm_device *dev, void *data,
+			      struct drm_file *filp);
+int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
+			  struct drm_file *filp);
+int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,
+			struct drm_file *filp);
+int amdgpu_gem_list_handles_ioctl(struct drm_device *dev, void *data,
+				  struct drm_file *filp);
+
+int amdgpu_gem_metadata_ioctl(struct drm_device *dev, void *data,
+				struct drm_file *filp);
+
+#define AMDGPU_GEM_CREATE_SETTABLE_MASK	(AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED | \
+	AMDGPU_GEM_CREATE_NO_CPU_ACCESS | \
+	AMDGPU_GEM_CREATE_CPU_GTT_USWC | \
+	AMDGPU_GEM_CREATE_VRAM_CLEARED | \
+	AMDGPU_GEM_CREATE_VM_ALWAYS_VALID | \
+	AMDGPU_GEM_CREATE_EXPLICIT_SYNC | \
+	AMDGPU_GEM_CREATE_VRAM_WIPE_ON_RELEASE | \
+	AMDGPU_GEM_CREATE_ENCRYPTED | \
+	AMDGPU_GEM_CREATE_GFX12_DCC | \
+	AMDGPU_GEM_CREATE_DISCARDABLE | \
+	AMDGPU_GEM_CREATE_COHERENT | \
+	AMDGPU_GEM_CREATE_UNCACHED | \
+	AMDGPU_GEM_CREATE_EXT_COHERENT)
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
index 4f6c68fc1dd9..ebe2b4c68b0f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
@@ -22,47 +22,78 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  *
  */
-#include <drm/drmP.h>
+
+#include <linux/firmware.h>
+#include <linux/pm_runtime.h>
+
 #include "amdgpu.h"
 #include "amdgpu_gfx.h"
+#include "amdgpu_rlc.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_reset.h"
+#include "amdgpu_xcp.h"
+#include "amdgpu_xgmi.h"
+#include "nvd.h"
+
+/* delay 0.1 second to enable gfx off feature */
+#define GFX_OFF_DELAY_ENABLE         msecs_to_jiffies(100)
+
+#define GFX_OFF_NO_DELAY 0
 
 /*
- * GPU scratch registers helpers function.
+ * GPU GFX IP block helpers function.
  */
-/**
- * amdgpu_gfx_scratch_get - Allocate a scratch register
- *
- * @adev: amdgpu_device pointer
- * @reg: scratch register mmio offset
- *
- * Allocate a CP scratch register for use by the driver (all asics).
- * Returns 0 on success or -EINVAL on failure.
- */
-int amdgpu_gfx_scratch_get(struct amdgpu_device *adev, uint32_t *reg)
+
+int amdgpu_gfx_mec_queue_to_bit(struct amdgpu_device *adev, int mec,
+				int pipe, int queue)
 {
-	int i;
+	int bit = 0;
 
-	i = ffs(adev->gfx.scratch.free_mask);
-	if (i != 0 && i <= adev->gfx.scratch.num_reg) {
-		i--;
-		adev->gfx.scratch.free_mask &= ~(1u << i);
-		*reg = adev->gfx.scratch.reg_base + i;
-		return 0;
-	}
-	return -EINVAL;
+	bit += mec * adev->gfx.mec.num_pipe_per_mec
+		* adev->gfx.mec.num_queue_per_pipe;
+	bit += pipe * adev->gfx.mec.num_queue_per_pipe;
+	bit += queue;
+
+	return bit;
 }
 
-/**
- * amdgpu_gfx_scratch_free - Free a scratch register
- *
- * @adev: amdgpu_device pointer
- * @reg: scratch register mmio offset
- *
- * Free a CP scratch register allocated for use by the driver (all asics)
- */
-void amdgpu_gfx_scratch_free(struct amdgpu_device *adev, uint32_t reg)
+void amdgpu_queue_mask_bit_to_mec_queue(struct amdgpu_device *adev, int bit,
+				 int *mec, int *pipe, int *queue)
+{
+	*queue = bit % adev->gfx.mec.num_queue_per_pipe;
+	*pipe = (bit / adev->gfx.mec.num_queue_per_pipe)
+		% adev->gfx.mec.num_pipe_per_mec;
+	*mec = (bit / adev->gfx.mec.num_queue_per_pipe)
+	       / adev->gfx.mec.num_pipe_per_mec;
+
+}
+
+bool amdgpu_gfx_is_mec_queue_enabled(struct amdgpu_device *adev,
+				     int xcc_id, int mec, int pipe, int queue)
 {
-	adev->gfx.scratch.free_mask |= 1u << (reg - adev->gfx.scratch.reg_base);
+	return test_bit(amdgpu_gfx_mec_queue_to_bit(adev, mec, pipe, queue),
+			adev->gfx.mec_bitmap[xcc_id].queue_bitmap);
+}
+
+static int amdgpu_gfx_me_queue_to_bit(struct amdgpu_device *adev,
+				      int me, int pipe, int queue)
+{
+	int num_queue_per_pipe = 1; /* we only enable 1 KGQ per pipe */
+	int bit = 0;
+
+	bit += me * adev->gfx.me.num_pipe_per_me
+		* num_queue_per_pipe;
+	bit += pipe * num_queue_per_pipe;
+	bit += queue;
+
+	return bit;
+}
+
+bool amdgpu_gfx_is_me_queue_enabled(struct amdgpu_device *adev,
+				    int me, int pipe, int queue)
+{
+	return test_bit(amdgpu_gfx_me_queue_to_bit(adev, me, pipe, queue),
+			adev->gfx.me.queue_bitmap);
 }
 
 /**
@@ -75,9 +106,9 @@ void amdgpu_gfx_scratch_free(struct amdgpu_device *adev, uint32_t reg)
  * The bitmask of CUs to be disabled in the shader array determined by se and
  * sh is stored in mask[se * max_sh + sh].
  */
-void amdgpu_gfx_parse_disable_cu(unsigned *mask, unsigned max_se, unsigned max_sh)
+void amdgpu_gfx_parse_disable_cu(unsigned int *mask, unsigned int max_se, unsigned int max_sh)
 {
-	unsigned se, sh, cu;
+	unsigned int se, sh, cu;
 	const char *p;
 
 	memset(mask, 0, sizeof(*mask) * max_se * max_sh);
@@ -89,6 +120,7 @@ void amdgpu_gfx_parse_disable_cu(unsigned *mask, unsigned max_se, unsigned max_s
 	for (;;) {
 		char *next;
 		int ret = sscanf(p, "%u.%u.%u", &se, &sh, &cu);
+
 		if (ret < 3) {
 			DRM_ERROR("amdgpu: could not parse disable_cu\n");
 			return;
@@ -109,46 +141,132 @@ void amdgpu_gfx_parse_disable_cu(unsigned *mask, unsigned max_se, unsigned max_s
 	}
 }
 
+static bool amdgpu_gfx_is_graphics_multipipe_capable(struct amdgpu_device *adev)
+{
+	return amdgpu_async_gfx_ring && adev->gfx.me.num_pipe_per_me > 1;
+}
+
+static bool amdgpu_gfx_is_compute_multipipe_capable(struct amdgpu_device *adev)
+{
+	if (amdgpu_compute_multipipe != -1) {
+		dev_info(adev->dev, "amdgpu: forcing compute pipe policy %d\n",
+			 amdgpu_compute_multipipe);
+		return amdgpu_compute_multipipe == 1;
+	}
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) > IP_VERSION(9, 0, 0))
+		return true;
+
+	/* FIXME: spreading the queues across pipes causes perf regressions
+	 * on POLARIS11 compute workloads */
+	if (adev->asic_type == CHIP_POLARIS11)
+		return false;
+
+	return adev->gfx.mec.num_mec > 1;
+}
+
+bool amdgpu_gfx_is_high_priority_graphics_queue(struct amdgpu_device *adev,
+						struct amdgpu_ring *ring)
+{
+	int queue = ring->queue;
+	int pipe = ring->pipe;
+
+	/* Policy: use pipe1 queue0 as high priority graphics queue if we
+	 * have more than one gfx pipe.
+	 */
+	if (amdgpu_gfx_is_graphics_multipipe_capable(adev) &&
+	    adev->gfx.num_gfx_rings > 1 && pipe == 1 && queue == 0) {
+		int me = ring->me;
+		int bit;
+
+		bit = amdgpu_gfx_me_queue_to_bit(adev, me, pipe, queue);
+		if (ring == &adev->gfx.gfx_ring[bit])
+			return true;
+	}
+
+	return false;
+}
+
+bool amdgpu_gfx_is_high_priority_compute_queue(struct amdgpu_device *adev,
+					       struct amdgpu_ring *ring)
+{
+	/* Policy: use 1st queue as high priority compute queue if we
+	 * have more than one compute queue.
+	 */
+	if (adev->gfx.num_compute_rings > 1 &&
+	    ring == &adev->gfx.compute_ring[0])
+		return true;
+
+	return false;
+}
+
 void amdgpu_gfx_compute_queue_acquire(struct amdgpu_device *adev)
 {
-	int i, queue, pipe, mec;
+	int i, j, queue, pipe;
+	bool multipipe_policy = amdgpu_gfx_is_compute_multipipe_capable(adev);
+	int max_queues_per_mec = min(adev->gfx.mec.num_pipe_per_mec *
+				     adev->gfx.mec.num_queue_per_pipe,
+				     adev->gfx.num_compute_rings);
+	int num_xcc = adev->gfx.xcc_mask ? NUM_XCC(adev->gfx.xcc_mask) : 1;
 
-	/* policy for amdgpu compute queue ownership */
-	for (i = 0; i < AMDGPU_MAX_COMPUTE_QUEUES; ++i) {
-		queue = i % adev->gfx.mec.num_queue_per_pipe;
-		pipe = (i / adev->gfx.mec.num_queue_per_pipe)
-			% adev->gfx.mec.num_pipe_per_mec;
-		mec = (i / adev->gfx.mec.num_queue_per_pipe)
-			/ adev->gfx.mec.num_pipe_per_mec;
-
-		/* we've run out of HW */
-		if (mec >= adev->gfx.mec.num_mec)
-			break;
+	if (multipipe_policy) {
+		/* policy: make queues evenly cross all pipes on MEC1 only
+		 * for multiple xcc, just use the original policy for simplicity */
+		for (j = 0; j < num_xcc; j++) {
+			for (i = 0; i < max_queues_per_mec; i++) {
+				pipe = i % adev->gfx.mec.num_pipe_per_mec;
+				queue = (i / adev->gfx.mec.num_pipe_per_mec) %
+					 adev->gfx.mec.num_queue_per_pipe;
 
-		/* FIXME: spreading the queues across pipes causes perf regressions */
-		if (0) {
-			/* policy: amdgpu owns the first two queues of the first MEC */
-			if (mec == 0 && queue < 2)
-				set_bit(i, adev->gfx.mec.queue_bitmap);
-		} else {
-			/* policy: amdgpu owns all queues in the first pipe */
-			if (mec == 0 && pipe == 0)
-				set_bit(i, adev->gfx.mec.queue_bitmap);
+				set_bit(pipe * adev->gfx.mec.num_queue_per_pipe + queue,
+					adev->gfx.mec_bitmap[j].queue_bitmap);
+			}
+		}
+	} else {
+		/* policy: amdgpu owns all queues in the given pipe */
+		for (j = 0; j < num_xcc; j++) {
+			for (i = 0; i < max_queues_per_mec; ++i)
+				set_bit(i, adev->gfx.mec_bitmap[j].queue_bitmap);
 		}
 	}
 
-	/* update the number of active compute rings */
-	adev->gfx.num_compute_rings =
-		bitmap_weight(adev->gfx.mec.queue_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
+	for (j = 0; j < num_xcc; j++) {
+		dev_dbg(adev->dev, "mec queue bitmap weight=%d\n",
+			bitmap_weight(adev->gfx.mec_bitmap[j].queue_bitmap, AMDGPU_MAX_COMPUTE_QUEUES));
+	}
+}
 
-	/* If you hit this case and edited the policy, you probably just
-	 * need to increase AMDGPU_MAX_COMPUTE_RINGS */
-	if (WARN_ON(adev->gfx.num_compute_rings > AMDGPU_MAX_COMPUTE_RINGS))
-		adev->gfx.num_compute_rings = AMDGPU_MAX_COMPUTE_RINGS;
+void amdgpu_gfx_graphics_queue_acquire(struct amdgpu_device *adev)
+{
+	int i, queue, pipe;
+	bool multipipe_policy = amdgpu_gfx_is_graphics_multipipe_capable(adev);
+	int num_queue_per_pipe = 1; /* we only enable 1 KGQ per pipe */
+	int max_queues_per_me = adev->gfx.me.num_pipe_per_me * num_queue_per_pipe;
+
+	if (multipipe_policy) {
+		/* policy: amdgpu owns the first queue per pipe at this stage
+		 * will extend to mulitple queues per pipe later */
+		for (i = 0; i < max_queues_per_me; i++) {
+			pipe = i % adev->gfx.me.num_pipe_per_me;
+			queue = (i / adev->gfx.me.num_pipe_per_me) %
+				num_queue_per_pipe;
+
+			set_bit(pipe * num_queue_per_pipe + queue,
+				adev->gfx.me.queue_bitmap);
+		}
+	} else {
+		for (i = 0; i < max_queues_per_me; ++i)
+			set_bit(i, adev->gfx.me.queue_bitmap);
+	}
+
+	/* update the number of active graphics rings */
+	if (adev->gfx.num_gfx_rings)
+		adev->gfx.num_gfx_rings =
+			bitmap_weight(adev->gfx.me.queue_bitmap, AMDGPU_MAX_GFX_QUEUES);
 }
 
 static int amdgpu_gfx_kiq_acquire(struct amdgpu_device *adev,
-				  struct amdgpu_ring *ring)
+				  struct amdgpu_ring *ring, int xcc_id)
 {
 	int queue_bit;
 	int mec, pipe, queue;
@@ -157,14 +275,18 @@ static int amdgpu_gfx_kiq_acquire(struct amdgpu_device *adev,
 		    * adev->gfx.mec.num_pipe_per_mec
 		    * adev->gfx.mec.num_queue_per_pipe;
 
-	while (queue_bit-- >= 0) {
-		if (test_bit(queue_bit, adev->gfx.mec.queue_bitmap))
+	while (--queue_bit >= 0) {
+		if (test_bit(queue_bit, adev->gfx.mec_bitmap[xcc_id].queue_bitmap))
 			continue;
 
-		amdgpu_gfx_bit_to_queue(adev, queue_bit, &mec, &pipe, &queue);
+		amdgpu_queue_mask_bit_to_mec_queue(adev, queue_bit, &mec, &pipe, &queue);
 
-		/* Using pipes 2/3 from MEC 2 seems cause problems */
-		if (mec == 1 && pipe > 1)
+		/*
+		 * 1. Using pipes 2/3 from MEC 2 seems cause problems.
+		 * 2. It must use queue id 0, because CGPG_IDLE/SAVE/LOAD/RUN
+		 * only can be issued on queue 0.
+		 */
+		if ((mec == 1 && pipe > 1) || queue != 0)
 			continue;
 
 		ring->me = mec + 1;
@@ -178,58 +300,60 @@ static int amdgpu_gfx_kiq_acquire(struct amdgpu_device *adev,
 	return -EINVAL;
 }
 
-int amdgpu_gfx_kiq_init_ring(struct amdgpu_device *adev,
-			     struct amdgpu_ring *ring,
-			     struct amdgpu_irq_src *irq)
+int amdgpu_gfx_kiq_init_ring(struct amdgpu_device *adev, int xcc_id)
 {
-	struct amdgpu_kiq *kiq = &adev->gfx.kiq;
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[xcc_id];
+	struct amdgpu_irq_src *irq = &kiq->irq;
+	struct amdgpu_ring *ring = &kiq->ring;
 	int r = 0;
 
-	mutex_init(&kiq->ring_mutex);
-
-	r = amdgpu_wb_get(adev, &adev->virt.reg_val_offs);
-	if (r)
-		return r;
+	spin_lock_init(&kiq->ring_lock);
 
 	ring->adev = NULL;
 	ring->ring_obj = NULL;
 	ring->use_doorbell = true;
-	ring->doorbell_index = AMDGPU_DOORBELL_KIQ;
+	ring->xcc_id = xcc_id;
+	ring->vm_hub = AMDGPU_GFXHUB(xcc_id);
+	ring->doorbell_index =
+		(adev->doorbell_index.kiq +
+		 xcc_id * adev->doorbell_index.xcc_doorbell_range)
+		<< 1;
 
-	r = amdgpu_gfx_kiq_acquire(adev, ring);
+	r = amdgpu_gfx_kiq_acquire(adev, ring, xcc_id);
 	if (r)
 		return r;
 
 	ring->eop_gpu_addr = kiq->eop_gpu_addr;
-	sprintf(ring->name, "kiq_%d.%d.%d", ring->me, ring->pipe, ring->queue);
-	r = amdgpu_ring_init(adev, ring, 1024,
-			     irq, AMDGPU_CP_KIQ_IRQ_DRIVER0);
+	ring->no_scheduler = true;
+	snprintf(ring->name, sizeof(ring->name), "kiq_%hhu.%hhu.%hhu.%hhu",
+		 (unsigned char)xcc_id, (unsigned char)ring->me,
+		 (unsigned char)ring->pipe, (unsigned char)ring->queue);
+	r = amdgpu_ring_init(adev, ring, 1024, irq, AMDGPU_CP_KIQ_IRQ_DRIVER0,
+			     AMDGPU_RING_PRIO_DEFAULT, NULL);
 	if (r)
 		dev_warn(adev->dev, "(%d) failed to init kiq ring\n", r);
 
 	return r;
 }
 
-void amdgpu_gfx_kiq_free_ring(struct amdgpu_ring *ring,
-			      struct amdgpu_irq_src *irq)
+void amdgpu_gfx_kiq_free_ring(struct amdgpu_ring *ring)
 {
-	amdgpu_wb_free(ring->adev, ring->adev->virt.reg_val_offs);
 	amdgpu_ring_fini(ring);
 }
 
-void amdgpu_gfx_kiq_fini(struct amdgpu_device *adev)
+void amdgpu_gfx_kiq_fini(struct amdgpu_device *adev, int xcc_id)
 {
-	struct amdgpu_kiq *kiq = &adev->gfx.kiq;
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[xcc_id];
 
 	amdgpu_bo_free_kernel(&kiq->eop_obj, &kiq->eop_gpu_addr, NULL);
 }
 
 int amdgpu_gfx_kiq_init(struct amdgpu_device *adev,
-			unsigned hpd_size)
+			unsigned int hpd_size, int xcc_id)
 {
 	int r;
 	u32 *hpd;
-	struct amdgpu_kiq *kiq = &adev->gfx.kiq;
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[xcc_id];
 
 	r = amdgpu_bo_create_kernel(adev, hpd_size, PAGE_SIZE,
 				    AMDGPU_GEM_DOMAIN_GTT, &kiq->eop_obj,
@@ -250,68 +374,2114 @@ int amdgpu_gfx_kiq_init(struct amdgpu_device *adev,
 	return 0;
 }
 
-/* create MQD for each compute queue */
-int amdgpu_gfx_compute_mqd_sw_init(struct amdgpu_device *adev,
-				   unsigned mqd_size)
+/* create MQD for each compute/gfx queue */
+int amdgpu_gfx_mqd_sw_init(struct amdgpu_device *adev,
+			   unsigned int mqd_size, int xcc_id)
 {
-	struct amdgpu_ring *ring = NULL;
-	int r, i;
+	int r, i, j;
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[xcc_id];
+	struct amdgpu_ring *ring = &kiq->ring;
+	u32 domain = AMDGPU_GEM_DOMAIN_GTT;
+
+#if !defined(CONFIG_ARM) && !defined(CONFIG_ARM64)
+	/* Only enable on gfx10 and 11 for now to avoid changing behavior on older chips */
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(10, 0, 0))
+		domain |= AMDGPU_GEM_DOMAIN_VRAM;
+#endif
 
 	/* create MQD for KIQ */
-	ring = &adev->gfx.kiq.ring;
-	if (!ring->mqd_obj) {
+	if (!adev->enable_mes_kiq && !ring->mqd_obj) {
+		/* originaly the KIQ MQD is put in GTT domain, but for SRIOV VRAM domain is a must
+		 * otherwise hypervisor trigger SAVE_VF fail after driver unloaded which mean MQD
+		 * deallocated and gart_unbind, to strict diverage we decide to use VRAM domain for
+		 * KIQ MQD no matter SRIOV or Bare-metal
+		 */
 		r = amdgpu_bo_create_kernel(adev, mqd_size, PAGE_SIZE,
-					    AMDGPU_GEM_DOMAIN_GTT, &ring->mqd_obj,
-					    &ring->mqd_gpu_addr, &ring->mqd_ptr);
+					    AMDGPU_GEM_DOMAIN_VRAM |
+					    AMDGPU_GEM_DOMAIN_GTT,
+					    &ring->mqd_obj,
+					    &ring->mqd_gpu_addr,
+					    &ring->mqd_ptr);
 		if (r) {
 			dev_warn(adev->dev, "failed to create ring mqd ob (%d)", r);
 			return r;
 		}
 
 		/* prepare MQD backup */
-		adev->gfx.mec.mqd_backup[AMDGPU_MAX_COMPUTE_RINGS] = kmalloc(mqd_size, GFP_KERNEL);
-		if (!adev->gfx.mec.mqd_backup[AMDGPU_MAX_COMPUTE_RINGS])
-				dev_warn(adev->dev, "no memory to create MQD backup for ring %s\n", ring->name);
+		kiq->mqd_backup = kzalloc(mqd_size, GFP_KERNEL);
+		if (!kiq->mqd_backup) {
+			dev_warn(adev->dev,
+				 "no memory to create MQD backup for ring %s\n", ring->name);
+			return -ENOMEM;
+		}
+	}
+
+	if (adev->asic_type >= CHIP_NAVI10 && amdgpu_async_gfx_ring) {
+		/* create MQD for each KGQ */
+		for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+			ring = &adev->gfx.gfx_ring[i];
+			if (!ring->mqd_obj) {
+				r = amdgpu_bo_create_kernel(adev, mqd_size, PAGE_SIZE,
+							    domain, &ring->mqd_obj,
+							    &ring->mqd_gpu_addr, &ring->mqd_ptr);
+				if (r) {
+					dev_warn(adev->dev, "failed to create ring mqd bo (%d)", r);
+					return r;
+				}
+
+				ring->mqd_size = mqd_size;
+				/* prepare MQD backup */
+				adev->gfx.me.mqd_backup[i] = kzalloc(mqd_size, GFP_KERNEL);
+				if (!adev->gfx.me.mqd_backup[i]) {
+					dev_warn(adev->dev, "no memory to create MQD backup for ring %s\n", ring->name);
+					return -ENOMEM;
+				}
+			}
+		}
 	}
 
 	/* create MQD for each KCQ */
 	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
-		ring = &adev->gfx.compute_ring[i];
+		j = i + xcc_id * adev->gfx.num_compute_rings;
+		ring = &adev->gfx.compute_ring[j];
 		if (!ring->mqd_obj) {
 			r = amdgpu_bo_create_kernel(adev, mqd_size, PAGE_SIZE,
-						    AMDGPU_GEM_DOMAIN_GTT, &ring->mqd_obj,
+						    domain, &ring->mqd_obj,
 						    &ring->mqd_gpu_addr, &ring->mqd_ptr);
 			if (r) {
-				dev_warn(adev->dev, "failed to create ring mqd ob (%d)", r);
+				dev_warn(adev->dev, "failed to create ring mqd bo (%d)", r);
 				return r;
 			}
 
+			ring->mqd_size = mqd_size;
 			/* prepare MQD backup */
-			adev->gfx.mec.mqd_backup[i] = kmalloc(mqd_size, GFP_KERNEL);
-			if (!adev->gfx.mec.mqd_backup[i])
+			adev->gfx.mec.mqd_backup[j] = kzalloc(mqd_size, GFP_KERNEL);
+			if (!adev->gfx.mec.mqd_backup[j]) {
 				dev_warn(adev->dev, "no memory to create MQD backup for ring %s\n", ring->name);
+				return -ENOMEM;
+			}
 		}
 	}
 
 	return 0;
 }
 
-void amdgpu_gfx_compute_mqd_sw_fini(struct amdgpu_device *adev)
+void amdgpu_gfx_mqd_sw_fini(struct amdgpu_device *adev, int xcc_id)
 {
 	struct amdgpu_ring *ring = NULL;
-	int i;
+	int i, j;
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[xcc_id];
+
+	if (adev->asic_type >= CHIP_NAVI10 && amdgpu_async_gfx_ring) {
+		for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+			ring = &adev->gfx.gfx_ring[i];
+			kfree(adev->gfx.me.mqd_backup[i]);
+			amdgpu_bo_free_kernel(&ring->mqd_obj,
+					      &ring->mqd_gpu_addr,
+					      &ring->mqd_ptr);
+		}
+	}
 
 	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
-		ring = &adev->gfx.compute_ring[i];
-		kfree(adev->gfx.mec.mqd_backup[i]);
+		j = i + xcc_id * adev->gfx.num_compute_rings;
+		ring = &adev->gfx.compute_ring[j];
+		kfree(adev->gfx.mec.mqd_backup[j]);
 		amdgpu_bo_free_kernel(&ring->mqd_obj,
 				      &ring->mqd_gpu_addr,
 				      &ring->mqd_ptr);
 	}
 
-	ring = &adev->gfx.kiq.ring;
-	kfree(adev->gfx.mec.mqd_backup[AMDGPU_MAX_COMPUTE_RINGS]);
+	ring = &kiq->ring;
+	kfree(kiq->mqd_backup);
 	amdgpu_bo_free_kernel(&ring->mqd_obj,
 			      &ring->mqd_gpu_addr,
 			      &ring->mqd_ptr);
 }
+
+int amdgpu_gfx_disable_kcq(struct amdgpu_device *adev, int xcc_id)
+{
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[xcc_id];
+	struct amdgpu_ring *kiq_ring = &kiq->ring;
+	int i, r = 0;
+	int j;
+
+	if (adev->enable_mes) {
+		for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+			j = i + xcc_id * adev->gfx.num_compute_rings;
+			amdgpu_mes_unmap_legacy_queue(adev,
+						   &adev->gfx.compute_ring[j],
+						   RESET_QUEUES, 0, 0);
+		}
+		return 0;
+	}
+
+	if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
+		return -EINVAL;
+
+	if (!kiq_ring->sched.ready || amdgpu_in_reset(adev))
+		return 0;
+
+	spin_lock(&kiq->ring_lock);
+	if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->unmap_queues_size *
+					adev->gfx.num_compute_rings)) {
+		spin_unlock(&kiq->ring_lock);
+		return -ENOMEM;
+	}
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		j = i + xcc_id * adev->gfx.num_compute_rings;
+		kiq->pmf->kiq_unmap_queues(kiq_ring,
+					   &adev->gfx.compute_ring[j],
+					   RESET_QUEUES, 0, 0);
+	}
+	/* Submit unmap queue packet */
+	amdgpu_ring_commit(kiq_ring);
+	/*
+	 * Ring test will do a basic scratch register change check. Just run
+	 * this to ensure that unmap queues that is submitted before got
+	 * processed successfully before returning.
+	 */
+	r = amdgpu_ring_test_helper(kiq_ring);
+
+	spin_unlock(&kiq->ring_lock);
+
+	return r;
+}
+
+int amdgpu_gfx_disable_kgq(struct amdgpu_device *adev, int xcc_id)
+{
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[xcc_id];
+	struct amdgpu_ring *kiq_ring = &kiq->ring;
+	int i, r = 0;
+	int j;
+
+	if (adev->enable_mes) {
+		if (amdgpu_gfx_is_master_xcc(adev, xcc_id)) {
+			for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+				j = i + xcc_id * adev->gfx.num_gfx_rings;
+				amdgpu_mes_unmap_legacy_queue(adev,
+						      &adev->gfx.gfx_ring[j],
+						      PREEMPT_QUEUES, 0, 0);
+			}
+		}
+		return 0;
+	}
+
+	if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
+		return -EINVAL;
+
+	if (!adev->gfx.kiq[0].ring.sched.ready || amdgpu_in_reset(adev))
+		return 0;
+
+	if (amdgpu_gfx_is_master_xcc(adev, xcc_id)) {
+		spin_lock(&kiq->ring_lock);
+		if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->unmap_queues_size *
+						adev->gfx.num_gfx_rings)) {
+			spin_unlock(&kiq->ring_lock);
+			return -ENOMEM;
+		}
+
+		for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+			j = i + xcc_id * adev->gfx.num_gfx_rings;
+			kiq->pmf->kiq_unmap_queues(kiq_ring,
+						   &adev->gfx.gfx_ring[j],
+						   PREEMPT_QUEUES, 0, 0);
+		}
+		/* Submit unmap queue packet */
+		amdgpu_ring_commit(kiq_ring);
+
+		/*
+		 * Ring test will do a basic scratch register change check.
+		 * Just run this to ensure that unmap queues that is submitted
+		 * before got processed successfully before returning.
+		 */
+		r = amdgpu_ring_test_helper(kiq_ring);
+		spin_unlock(&kiq->ring_lock);
+	}
+
+	return r;
+}
+
+int amdgpu_queue_mask_bit_to_set_resource_bit(struct amdgpu_device *adev,
+					int queue_bit)
+{
+	int mec, pipe, queue;
+	int set_resource_bit = 0;
+
+	amdgpu_queue_mask_bit_to_mec_queue(adev, queue_bit, &mec, &pipe, &queue);
+
+	set_resource_bit = mec * 4 * 8 + pipe * 8 + queue;
+
+	return set_resource_bit;
+}
+
+static int amdgpu_gfx_mes_enable_kcq(struct amdgpu_device *adev, int xcc_id)
+{
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[xcc_id];
+	struct amdgpu_ring *kiq_ring = &kiq->ring;
+	uint64_t queue_mask = ~0ULL;
+	int r, i, j;
+
+	amdgpu_device_flush_hdp(adev, NULL);
+
+	if (!adev->enable_uni_mes) {
+		spin_lock(&kiq->ring_lock);
+		r = amdgpu_ring_alloc(kiq_ring, kiq->pmf->set_resources_size);
+		if (r) {
+			dev_err(adev->dev, "Failed to lock KIQ (%d).\n", r);
+			spin_unlock(&kiq->ring_lock);
+			return r;
+		}
+
+		kiq->pmf->kiq_set_resources(kiq_ring, queue_mask);
+		r = amdgpu_ring_test_helper(kiq_ring);
+		spin_unlock(&kiq->ring_lock);
+		if (r)
+			dev_err(adev->dev, "KIQ failed to set resources\n");
+	}
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		j = i + xcc_id * adev->gfx.num_compute_rings;
+		r = amdgpu_mes_map_legacy_queue(adev,
+						&adev->gfx.compute_ring[j]);
+		if (r) {
+			dev_err(adev->dev, "failed to map compute queue\n");
+			return r;
+		}
+	}
+
+	return 0;
+}
+
+int amdgpu_gfx_enable_kcq(struct amdgpu_device *adev, int xcc_id)
+{
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[xcc_id];
+	struct amdgpu_ring *kiq_ring = &kiq->ring;
+	uint64_t queue_mask = 0;
+	int r, i, j;
+
+	if (adev->mes.enable_legacy_queue_map)
+		return amdgpu_gfx_mes_enable_kcq(adev, xcc_id);
+
+	if (!kiq->pmf || !kiq->pmf->kiq_map_queues || !kiq->pmf->kiq_set_resources)
+		return -EINVAL;
+
+	for (i = 0; i < AMDGPU_MAX_COMPUTE_QUEUES; ++i) {
+		if (!test_bit(i, adev->gfx.mec_bitmap[xcc_id].queue_bitmap))
+			continue;
+
+		/* This situation may be hit in the future if a new HW
+		 * generation exposes more than 64 queues. If so, the
+		 * definition of queue_mask needs updating */
+		if (WARN_ON(i > (sizeof(queue_mask)*8))) {
+			dev_err(adev->dev, "Invalid KCQ enabled: %d\n", i);
+			break;
+		}
+
+		queue_mask |= (1ull << amdgpu_queue_mask_bit_to_set_resource_bit(adev, i));
+	}
+
+	amdgpu_device_flush_hdp(adev, NULL);
+
+	dev_info(adev->dev, "kiq ring mec %d pipe %d q %d\n", kiq_ring->me,
+		 kiq_ring->pipe, kiq_ring->queue);
+
+	spin_lock(&kiq->ring_lock);
+	r = amdgpu_ring_alloc(kiq_ring, kiq->pmf->map_queues_size *
+					adev->gfx.num_compute_rings +
+					kiq->pmf->set_resources_size);
+	if (r) {
+		dev_err(adev->dev, "Failed to lock KIQ (%d).\n", r);
+		spin_unlock(&kiq->ring_lock);
+		return r;
+	}
+
+	kiq->pmf->kiq_set_resources(kiq_ring, queue_mask);
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		j = i + xcc_id * adev->gfx.num_compute_rings;
+		kiq->pmf->kiq_map_queues(kiq_ring,
+					 &adev->gfx.compute_ring[j]);
+	}
+	/* Submit map queue packet */
+	amdgpu_ring_commit(kiq_ring);
+	/*
+	 * Ring test will do a basic scratch register change check. Just run
+	 * this to ensure that map queues that is submitted before got
+	 * processed successfully before returning.
+	 */
+	r = amdgpu_ring_test_helper(kiq_ring);
+	spin_unlock(&kiq->ring_lock);
+	if (r)
+		dev_err(adev->dev, "KCQ enable failed\n");
+
+	return r;
+}
+
+int amdgpu_gfx_enable_kgq(struct amdgpu_device *adev, int xcc_id)
+{
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[xcc_id];
+	struct amdgpu_ring *kiq_ring = &kiq->ring;
+	int r, i, j;
+
+	if (!kiq->pmf || !kiq->pmf->kiq_map_queues)
+		return -EINVAL;
+
+	amdgpu_device_flush_hdp(adev, NULL);
+
+	if (adev->mes.enable_legacy_queue_map) {
+		for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+			j = i + xcc_id * adev->gfx.num_gfx_rings;
+			r = amdgpu_mes_map_legacy_queue(adev,
+							&adev->gfx.gfx_ring[j]);
+			if (r) {
+				dev_err(adev->dev, "failed to map gfx queue\n");
+				return r;
+			}
+		}
+
+		return 0;
+	}
+
+	spin_lock(&kiq->ring_lock);
+	/* No need to map kcq on the slave */
+	if (amdgpu_gfx_is_master_xcc(adev, xcc_id)) {
+		r = amdgpu_ring_alloc(kiq_ring, kiq->pmf->map_queues_size *
+						adev->gfx.num_gfx_rings);
+		if (r) {
+			dev_err(adev->dev, "Failed to lock KIQ (%d).\n", r);
+			spin_unlock(&kiq->ring_lock);
+			return r;
+		}
+
+		for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+			j = i + xcc_id * adev->gfx.num_gfx_rings;
+			kiq->pmf->kiq_map_queues(kiq_ring,
+						 &adev->gfx.gfx_ring[j]);
+		}
+	}
+	/* Submit map queue packet */
+	amdgpu_ring_commit(kiq_ring);
+	/*
+	 * Ring test will do a basic scratch register change check. Just run
+	 * this to ensure that map queues that is submitted before got
+	 * processed successfully before returning.
+	 */
+	r = amdgpu_ring_test_helper(kiq_ring);
+	spin_unlock(&kiq->ring_lock);
+	if (r)
+		dev_err(adev->dev, "KGQ enable failed\n");
+
+	return r;
+}
+
+static void amdgpu_gfx_do_off_ctrl(struct amdgpu_device *adev, bool enable,
+				   bool no_delay)
+{
+	unsigned long delay = GFX_OFF_DELAY_ENABLE;
+
+	if (!(adev->pm.pp_feature & PP_GFXOFF_MASK))
+		return;
+
+	mutex_lock(&adev->gfx.gfx_off_mutex);
+
+	if (enable) {
+		/* If the count is already 0, it means there's an imbalance bug somewhere.
+		 * Note that the bug may be in a different caller than the one which triggers the
+		 * WARN_ON_ONCE.
+		 */
+		if (WARN_ON_ONCE(adev->gfx.gfx_off_req_count == 0))
+			goto unlock;
+
+		adev->gfx.gfx_off_req_count--;
+
+		if (adev->gfx.gfx_off_req_count == 0 &&
+		    !adev->gfx.gfx_off_state) {
+			/* If going to s2idle, no need to wait */
+			if (no_delay) {
+				if (!amdgpu_dpm_set_powergating_by_smu(adev,
+						AMD_IP_BLOCK_TYPE_GFX, true, 0))
+					adev->gfx.gfx_off_state = true;
+			} else {
+				schedule_delayed_work(&adev->gfx.gfx_off_delay_work,
+					      delay);
+			}
+		}
+	} else {
+		if (adev->gfx.gfx_off_req_count == 0) {
+			cancel_delayed_work_sync(&adev->gfx.gfx_off_delay_work);
+
+			if (adev->gfx.gfx_off_state &&
+			    !amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_GFX, false, 0)) {
+				adev->gfx.gfx_off_state = false;
+
+				if (adev->gfx.funcs->init_spm_golden) {
+					dev_dbg(adev->dev,
+						"GFXOFF is disabled, re-init SPM golden settings\n");
+					amdgpu_gfx_init_spm_golden(adev);
+				}
+			}
+		}
+
+		adev->gfx.gfx_off_req_count++;
+	}
+
+unlock:
+	mutex_unlock(&adev->gfx.gfx_off_mutex);
+}
+
+/* amdgpu_gfx_off_ctrl - Handle gfx off feature enable/disable
+ *
+ * @adev: amdgpu_device pointer
+ * @bool enable true: enable gfx off feature, false: disable gfx off feature
+ *
+ * 1. gfx off feature will be enabled by gfx ip after gfx cg pg enabled.
+ * 2. other client can send request to disable gfx off feature, the request should be honored.
+ * 3. other client can cancel their request of disable gfx off feature
+ * 4. other client should not send request to enable gfx off feature before disable gfx off feature.
+ *
+ * gfx off allow will be delayed by GFX_OFF_DELAY_ENABLE ms.
+ */
+void amdgpu_gfx_off_ctrl(struct amdgpu_device *adev, bool enable)
+{
+	/* If going to s2idle, no need to wait */
+	bool no_delay = adev->in_s0ix ? true : false;
+
+	amdgpu_gfx_do_off_ctrl(adev, enable, no_delay);
+}
+
+/* amdgpu_gfx_off_ctrl_immediate - Handle gfx off feature enable/disable
+ *
+ * @adev: amdgpu_device pointer
+ * @bool enable true: enable gfx off feature, false: disable gfx off feature
+ *
+ * 1. gfx off feature will be enabled by gfx ip after gfx cg pg enabled.
+ * 2. other client can send request to disable gfx off feature, the request should be honored.
+ * 3. other client can cancel their request of disable gfx off feature
+ * 4. other client should not send request to enable gfx off feature before disable gfx off feature.
+ *
+ * gfx off allow will be issued immediately.
+ */
+void amdgpu_gfx_off_ctrl_immediate(struct amdgpu_device *adev, bool enable)
+{
+	amdgpu_gfx_do_off_ctrl(adev, enable, true);
+}
+
+int amdgpu_set_gfx_off_residency(struct amdgpu_device *adev, bool value)
+{
+	int r = 0;
+
+	mutex_lock(&adev->gfx.gfx_off_mutex);
+
+	r = amdgpu_dpm_set_residency_gfxoff(adev, value);
+
+	mutex_unlock(&adev->gfx.gfx_off_mutex);
+
+	return r;
+}
+
+int amdgpu_get_gfx_off_residency(struct amdgpu_device *adev, u32 *value)
+{
+	int r = 0;
+
+	mutex_lock(&adev->gfx.gfx_off_mutex);
+
+	r = amdgpu_dpm_get_residency_gfxoff(adev, value);
+
+	mutex_unlock(&adev->gfx.gfx_off_mutex);
+
+	return r;
+}
+
+int amdgpu_get_gfx_off_entrycount(struct amdgpu_device *adev, u64 *value)
+{
+	int r = 0;
+
+	mutex_lock(&adev->gfx.gfx_off_mutex);
+
+	r = amdgpu_dpm_get_entrycount_gfxoff(adev, value);
+
+	mutex_unlock(&adev->gfx.gfx_off_mutex);
+
+	return r;
+}
+
+int amdgpu_get_gfx_off_status(struct amdgpu_device *adev, uint32_t *value)
+{
+
+	int r = 0;
+
+	mutex_lock(&adev->gfx.gfx_off_mutex);
+
+	r = amdgpu_dpm_get_status_gfxoff(adev, value);
+
+	mutex_unlock(&adev->gfx.gfx_off_mutex);
+
+	return r;
+}
+
+int amdgpu_gfx_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
+{
+	int r;
+
+	if (amdgpu_ras_is_supported(adev, ras_block->block)) {
+		if (!amdgpu_persistent_edc_harvesting_supported(adev)) {
+			r = amdgpu_ras_reset_error_status(adev, AMDGPU_RAS_BLOCK__GFX);
+			if (r)
+				return r;
+		}
+
+		r = amdgpu_ras_block_late_init(adev, ras_block);
+		if (r)
+			return r;
+
+		if (amdgpu_sriov_vf(adev))
+			return r;
+
+		if (adev->gfx.cp_ecc_error_irq.funcs) {
+			r = amdgpu_irq_get(adev, &adev->gfx.cp_ecc_error_irq, 0);
+			if (r)
+				goto late_fini;
+		}
+	} else {
+		amdgpu_ras_feature_enable_on_boot(adev, ras_block, 0);
+	}
+
+	return 0;
+late_fini:
+	amdgpu_ras_block_late_fini(adev, ras_block);
+	return r;
+}
+
+int amdgpu_gfx_ras_sw_init(struct amdgpu_device *adev)
+{
+	int err = 0;
+	struct amdgpu_gfx_ras *ras = NULL;
+
+	/* adev->gfx.ras is NULL, which means gfx does not
+	 * support ras function, then do nothing here.
+	 */
+	if (!adev->gfx.ras)
+		return 0;
+
+	ras = adev->gfx.ras;
+
+	err = amdgpu_ras_register_ras_block(adev, &ras->ras_block);
+	if (err) {
+		dev_err(adev->dev, "Failed to register gfx ras block!\n");
+		return err;
+	}
+
+	strcpy(ras->ras_block.ras_comm.name, "gfx");
+	ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__GFX;
+	ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
+	adev->gfx.ras_if = &ras->ras_block.ras_comm;
+
+	/* If not define special ras_late_init function, use gfx default ras_late_init */
+	if (!ras->ras_block.ras_late_init)
+		ras->ras_block.ras_late_init = amdgpu_gfx_ras_late_init;
+
+	/* If not defined special ras_cb function, use default ras_cb */
+	if (!ras->ras_block.ras_cb)
+		ras->ras_block.ras_cb = amdgpu_gfx_process_ras_data_cb;
+
+	return 0;
+}
+
+int amdgpu_gfx_poison_consumption_handler(struct amdgpu_device *adev,
+						struct amdgpu_iv_entry *entry)
+{
+	if (adev->gfx.ras && adev->gfx.ras->poison_consumption_handler)
+		return adev->gfx.ras->poison_consumption_handler(adev, entry);
+
+	return 0;
+}
+
+int amdgpu_gfx_process_ras_data_cb(struct amdgpu_device *adev,
+		void *err_data,
+		struct amdgpu_iv_entry *entry)
+{
+	/* TODO ue will trigger an interrupt.
+	 *
+	 * When “Full RAS” is enabled, the per-IP interrupt sources should
+	 * be disabled and the driver should only look for the aggregated
+	 * interrupt via sync flood
+	 */
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX)) {
+		kgd2kfd_set_sram_ecc_flag(adev->kfd.dev);
+		if (adev->gfx.ras && adev->gfx.ras->ras_block.hw_ops &&
+		    adev->gfx.ras->ras_block.hw_ops->query_ras_error_count)
+			adev->gfx.ras->ras_block.hw_ops->query_ras_error_count(adev, err_data);
+		amdgpu_ras_reset_gpu(adev);
+	}
+	return AMDGPU_RAS_SUCCESS;
+}
+
+int amdgpu_gfx_cp_ecc_error_irq(struct amdgpu_device *adev,
+				  struct amdgpu_irq_src *source,
+				  struct amdgpu_iv_entry *entry)
+{
+	struct ras_common_if *ras_if = adev->gfx.ras_if;
+	struct ras_dispatch_if ih_data = {
+		.entry = entry,
+	};
+
+	if (!ras_if)
+		return 0;
+
+	ih_data.head = *ras_if;
+
+	dev_err(adev->dev, "CP ECC ERROR IRQ\n");
+	amdgpu_ras_interrupt_dispatch(adev, &ih_data);
+	return 0;
+}
+
+void amdgpu_gfx_ras_error_func(struct amdgpu_device *adev,
+		void *ras_error_status,
+		void (*func)(struct amdgpu_device *adev, void *ras_error_status,
+				int xcc_id))
+{
+	int i;
+	int num_xcc = adev->gfx.xcc_mask ? NUM_XCC(adev->gfx.xcc_mask) : 1;
+	uint32_t xcc_mask = GENMASK(num_xcc - 1, 0);
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+
+	if (err_data) {
+		err_data->ue_count = 0;
+		err_data->ce_count = 0;
+	}
+
+	for_each_inst(i, xcc_mask)
+		func(adev, ras_error_status, i);
+}
+
+uint32_t amdgpu_kiq_rreg(struct amdgpu_device *adev, uint32_t reg, uint32_t xcc_id)
+{
+	signed long r, cnt = 0;
+	unsigned long flags;
+	uint32_t seq, reg_val_offs = 0, value = 0;
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[xcc_id];
+	struct amdgpu_ring *ring = &kiq->ring;
+
+	if (amdgpu_device_skip_hw_access(adev))
+		return 0;
+
+	if (adev->mes.ring[0].sched.ready)
+		return amdgpu_mes_rreg(adev, reg);
+
+	BUG_ON(!ring->funcs->emit_rreg);
+
+	spin_lock_irqsave(&kiq->ring_lock, flags);
+	if (amdgpu_device_wb_get(adev, &reg_val_offs)) {
+		pr_err("critical bug! too many kiq readers\n");
+		goto failed_unlock;
+	}
+	r = amdgpu_ring_alloc(ring, 32);
+	if (r)
+		goto failed_unlock;
+
+	amdgpu_ring_emit_rreg(ring, reg, reg_val_offs);
+	r = amdgpu_fence_emit_polling(ring, &seq, MAX_KIQ_REG_WAIT);
+	if (r)
+		goto failed_undo;
+
+	amdgpu_ring_commit(ring);
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+
+	r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
+
+	/* don't wait anymore for gpu reset case because this way may
+	 * block gpu_recover() routine forever, e.g. this virt_kiq_rreg
+	 * is triggered in TTM and ttm_bo_lock_delayed_workqueue() will
+	 * never return if we keep waiting in virt_kiq_rreg, which cause
+	 * gpu_recover() hang there.
+	 *
+	 * also don't wait anymore for IRQ context
+	 * */
+	if (r < 1 && (amdgpu_in_reset(adev) || in_interrupt()))
+		goto failed_kiq_read;
+
+	might_sleep();
+	while (r < 1 && cnt++ < MAX_KIQ_REG_TRY) {
+		if (amdgpu_in_reset(adev))
+			goto failed_kiq_read;
+
+		msleep(MAX_KIQ_REG_BAILOUT_INTERVAL);
+		r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
+	}
+
+	if (cnt > MAX_KIQ_REG_TRY)
+		goto failed_kiq_read;
+
+	mb();
+	value = adev->wb.wb[reg_val_offs];
+	amdgpu_device_wb_free(adev, reg_val_offs);
+	return value;
+
+failed_undo:
+	amdgpu_ring_undo(ring);
+failed_unlock:
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+failed_kiq_read:
+	if (reg_val_offs)
+		amdgpu_device_wb_free(adev, reg_val_offs);
+	dev_err(adev->dev, "failed to read reg:%x\n", reg);
+	return ~0;
+}
+
+void amdgpu_kiq_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v, uint32_t xcc_id)
+{
+	signed long r, cnt = 0;
+	unsigned long flags;
+	uint32_t seq;
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[xcc_id];
+	struct amdgpu_ring *ring = &kiq->ring;
+
+	BUG_ON(!ring->funcs->emit_wreg);
+
+	if (amdgpu_device_skip_hw_access(adev))
+		return;
+
+	if (adev->mes.ring[0].sched.ready) {
+		amdgpu_mes_wreg(adev, reg, v);
+		return;
+	}
+
+	spin_lock_irqsave(&kiq->ring_lock, flags);
+	r = amdgpu_ring_alloc(ring, 32);
+	if (r)
+		goto failed_unlock;
+
+	amdgpu_ring_emit_wreg(ring, reg, v);
+	r = amdgpu_fence_emit_polling(ring, &seq, MAX_KIQ_REG_WAIT);
+	if (r)
+		goto failed_undo;
+
+	amdgpu_ring_commit(ring);
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+
+	r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
+
+	/* don't wait anymore for gpu reset case because this way may
+	 * block gpu_recover() routine forever, e.g. this virt_kiq_rreg
+	 * is triggered in TTM and ttm_bo_lock_delayed_workqueue() will
+	 * never return if we keep waiting in virt_kiq_rreg, which cause
+	 * gpu_recover() hang there.
+	 *
+	 * also don't wait anymore for IRQ context
+	 * */
+	if (r < 1 && (amdgpu_in_reset(adev) || in_interrupt()))
+		goto failed_kiq_write;
+
+	might_sleep();
+	while (r < 1 && cnt++ < MAX_KIQ_REG_TRY) {
+		if (amdgpu_in_reset(adev))
+			goto failed_kiq_write;
+
+		msleep(MAX_KIQ_REG_BAILOUT_INTERVAL);
+		r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
+	}
+
+	if (cnt > MAX_KIQ_REG_TRY)
+		goto failed_kiq_write;
+
+	return;
+
+failed_undo:
+	amdgpu_ring_undo(ring);
+failed_unlock:
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+failed_kiq_write:
+	dev_err(adev->dev, "failed to write reg:%x\n", reg);
+}
+
+int amdgpu_gfx_get_num_kcq(struct amdgpu_device *adev)
+{
+	if (amdgpu_num_kcq == -1) {
+		return 8;
+	} else if (amdgpu_num_kcq > 8 || amdgpu_num_kcq < 0) {
+		dev_warn(adev->dev, "set kernel compute queue number to 8 due to invalid parameter provided by user\n");
+		return 8;
+	}
+	return amdgpu_num_kcq;
+}
+
+void amdgpu_gfx_cp_init_microcode(struct amdgpu_device *adev,
+				  uint32_t ucode_id)
+{
+	const struct gfx_firmware_header_v1_0 *cp_hdr;
+	const struct gfx_firmware_header_v2_0 *cp_hdr_v2_0;
+	struct amdgpu_firmware_info *info = NULL;
+	const struct firmware *ucode_fw;
+	unsigned int fw_size;
+
+	switch (ucode_id) {
+	case AMDGPU_UCODE_ID_CP_PFP:
+		cp_hdr = (const struct gfx_firmware_header_v1_0 *)
+			adev->gfx.pfp_fw->data;
+		adev->gfx.pfp_fw_version =
+			le32_to_cpu(cp_hdr->header.ucode_version);
+		adev->gfx.pfp_feature_version =
+			le32_to_cpu(cp_hdr->ucode_feature_version);
+		ucode_fw = adev->gfx.pfp_fw;
+		fw_size = le32_to_cpu(cp_hdr->header.ucode_size_bytes);
+		break;
+	case AMDGPU_UCODE_ID_CP_RS64_PFP:
+		cp_hdr_v2_0 = (const struct gfx_firmware_header_v2_0 *)
+			adev->gfx.pfp_fw->data;
+		adev->gfx.pfp_fw_version =
+			le32_to_cpu(cp_hdr_v2_0->header.ucode_version);
+		adev->gfx.pfp_feature_version =
+			le32_to_cpu(cp_hdr_v2_0->ucode_feature_version);
+		ucode_fw = adev->gfx.pfp_fw;
+		fw_size = le32_to_cpu(cp_hdr_v2_0->ucode_size_bytes);
+		break;
+	case AMDGPU_UCODE_ID_CP_RS64_PFP_P0_STACK:
+	case AMDGPU_UCODE_ID_CP_RS64_PFP_P1_STACK:
+		cp_hdr_v2_0 = (const struct gfx_firmware_header_v2_0 *)
+			adev->gfx.pfp_fw->data;
+		ucode_fw = adev->gfx.pfp_fw;
+		fw_size = le32_to_cpu(cp_hdr_v2_0->data_size_bytes);
+		break;
+	case AMDGPU_UCODE_ID_CP_ME:
+		cp_hdr = (const struct gfx_firmware_header_v1_0 *)
+			adev->gfx.me_fw->data;
+		adev->gfx.me_fw_version =
+			le32_to_cpu(cp_hdr->header.ucode_version);
+		adev->gfx.me_feature_version =
+			le32_to_cpu(cp_hdr->ucode_feature_version);
+		ucode_fw = adev->gfx.me_fw;
+		fw_size = le32_to_cpu(cp_hdr->header.ucode_size_bytes);
+		break;
+	case AMDGPU_UCODE_ID_CP_RS64_ME:
+		cp_hdr_v2_0 = (const struct gfx_firmware_header_v2_0 *)
+			adev->gfx.me_fw->data;
+		adev->gfx.me_fw_version =
+			le32_to_cpu(cp_hdr_v2_0->header.ucode_version);
+		adev->gfx.me_feature_version =
+			le32_to_cpu(cp_hdr_v2_0->ucode_feature_version);
+		ucode_fw = adev->gfx.me_fw;
+		fw_size = le32_to_cpu(cp_hdr_v2_0->ucode_size_bytes);
+		break;
+	case AMDGPU_UCODE_ID_CP_RS64_ME_P0_STACK:
+	case AMDGPU_UCODE_ID_CP_RS64_ME_P1_STACK:
+		cp_hdr_v2_0 = (const struct gfx_firmware_header_v2_0 *)
+			adev->gfx.me_fw->data;
+		ucode_fw = adev->gfx.me_fw;
+		fw_size = le32_to_cpu(cp_hdr_v2_0->data_size_bytes);
+		break;
+	case AMDGPU_UCODE_ID_CP_CE:
+		cp_hdr = (const struct gfx_firmware_header_v1_0 *)
+			adev->gfx.ce_fw->data;
+		adev->gfx.ce_fw_version =
+			le32_to_cpu(cp_hdr->header.ucode_version);
+		adev->gfx.ce_feature_version =
+			le32_to_cpu(cp_hdr->ucode_feature_version);
+		ucode_fw = adev->gfx.ce_fw;
+		fw_size = le32_to_cpu(cp_hdr->header.ucode_size_bytes);
+		break;
+	case AMDGPU_UCODE_ID_CP_MEC1:
+		cp_hdr = (const struct gfx_firmware_header_v1_0 *)
+			adev->gfx.mec_fw->data;
+		adev->gfx.mec_fw_version =
+			le32_to_cpu(cp_hdr->header.ucode_version);
+		adev->gfx.mec_feature_version =
+			le32_to_cpu(cp_hdr->ucode_feature_version);
+		ucode_fw = adev->gfx.mec_fw;
+		fw_size = le32_to_cpu(cp_hdr->header.ucode_size_bytes) -
+			  le32_to_cpu(cp_hdr->jt_size) * 4;
+		break;
+	case AMDGPU_UCODE_ID_CP_MEC1_JT:
+		cp_hdr = (const struct gfx_firmware_header_v1_0 *)
+			adev->gfx.mec_fw->data;
+		ucode_fw = adev->gfx.mec_fw;
+		fw_size = le32_to_cpu(cp_hdr->jt_size) * 4;
+		break;
+	case AMDGPU_UCODE_ID_CP_MEC2:
+		cp_hdr = (const struct gfx_firmware_header_v1_0 *)
+			adev->gfx.mec2_fw->data;
+		adev->gfx.mec2_fw_version =
+			le32_to_cpu(cp_hdr->header.ucode_version);
+		adev->gfx.mec2_feature_version =
+			le32_to_cpu(cp_hdr->ucode_feature_version);
+		ucode_fw = adev->gfx.mec2_fw;
+		fw_size = le32_to_cpu(cp_hdr->header.ucode_size_bytes) -
+			  le32_to_cpu(cp_hdr->jt_size) * 4;
+		break;
+	case AMDGPU_UCODE_ID_CP_MEC2_JT:
+		cp_hdr = (const struct gfx_firmware_header_v1_0 *)
+			adev->gfx.mec2_fw->data;
+		ucode_fw = adev->gfx.mec2_fw;
+		fw_size = le32_to_cpu(cp_hdr->jt_size) * 4;
+		break;
+	case AMDGPU_UCODE_ID_CP_RS64_MEC:
+		cp_hdr_v2_0 = (const struct gfx_firmware_header_v2_0 *)
+			adev->gfx.mec_fw->data;
+		adev->gfx.mec_fw_version =
+			le32_to_cpu(cp_hdr_v2_0->header.ucode_version);
+		adev->gfx.mec_feature_version =
+			le32_to_cpu(cp_hdr_v2_0->ucode_feature_version);
+		ucode_fw = adev->gfx.mec_fw;
+		fw_size = le32_to_cpu(cp_hdr_v2_0->ucode_size_bytes);
+		break;
+	case AMDGPU_UCODE_ID_CP_RS64_MEC_P0_STACK:
+	case AMDGPU_UCODE_ID_CP_RS64_MEC_P1_STACK:
+	case AMDGPU_UCODE_ID_CP_RS64_MEC_P2_STACK:
+	case AMDGPU_UCODE_ID_CP_RS64_MEC_P3_STACK:
+		cp_hdr_v2_0 = (const struct gfx_firmware_header_v2_0 *)
+			adev->gfx.mec_fw->data;
+		ucode_fw = adev->gfx.mec_fw;
+		fw_size = le32_to_cpu(cp_hdr_v2_0->data_size_bytes);
+		break;
+	default:
+		dev_err(adev->dev, "Invalid ucode id %u\n", ucode_id);
+		return;
+	}
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		info = &adev->firmware.ucode[ucode_id];
+		info->ucode_id = ucode_id;
+		info->fw = ucode_fw;
+		adev->firmware.fw_size += ALIGN(fw_size, PAGE_SIZE);
+	}
+}
+
+bool amdgpu_gfx_is_master_xcc(struct amdgpu_device *adev, int xcc_id)
+{
+	return !(xcc_id % (adev->gfx.num_xcc_per_xcp ?
+			adev->gfx.num_xcc_per_xcp : 1));
+}
+
+static ssize_t amdgpu_gfx_get_current_compute_partition(struct device *dev,
+						struct device_attribute *addr,
+						char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	int mode;
+
+	/* Only minimal precaution taken to reject requests while in reset.*/
+	if (amdgpu_in_reset(adev))
+		return -EPERM;
+
+	mode = amdgpu_xcp_query_partition_mode(adev->xcp_mgr,
+					       AMDGPU_XCP_FL_NONE);
+
+	return sysfs_emit(buf, "%s\n", amdgpu_gfx_compute_mode_desc(mode));
+}
+
+static ssize_t amdgpu_gfx_set_compute_partition(struct device *dev,
+						struct device_attribute *addr,
+						const char *buf, size_t count)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	enum amdgpu_gfx_partition mode;
+	int ret = 0, num_xcc;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	if (num_xcc % 2 != 0)
+		return -EINVAL;
+
+	if (!strncasecmp("SPX", buf, strlen("SPX"))) {
+		mode = AMDGPU_SPX_PARTITION_MODE;
+	} else if (!strncasecmp("DPX", buf, strlen("DPX"))) {
+		/*
+		 * DPX mode needs AIDs to be in multiple of 2.
+		 * Each AID connects 2 XCCs.
+		 */
+		if (num_xcc%4)
+			return -EINVAL;
+		mode = AMDGPU_DPX_PARTITION_MODE;
+	} else if (!strncasecmp("TPX", buf, strlen("TPX"))) {
+		if (num_xcc != 6)
+			return -EINVAL;
+		mode = AMDGPU_TPX_PARTITION_MODE;
+	} else if (!strncasecmp("QPX", buf, strlen("QPX"))) {
+		if (num_xcc != 8)
+			return -EINVAL;
+		mode = AMDGPU_QPX_PARTITION_MODE;
+	} else if (!strncasecmp("CPX", buf, strlen("CPX"))) {
+		mode = AMDGPU_CPX_PARTITION_MODE;
+	} else {
+		return -EINVAL;
+	}
+
+	/* Don't allow a switch while under reset */
+	if (!down_read_trylock(&adev->reset_domain->sem))
+		return -EPERM;
+
+	ret = amdgpu_xcp_switch_partition_mode(adev->xcp_mgr, mode);
+
+	up_read(&adev->reset_domain->sem);
+
+	if (ret)
+		return ret;
+
+	return count;
+}
+
+static const char *xcp_desc[] = {
+	[AMDGPU_SPX_PARTITION_MODE] = "SPX",
+	[AMDGPU_DPX_PARTITION_MODE] = "DPX",
+	[AMDGPU_TPX_PARTITION_MODE] = "TPX",
+	[AMDGPU_QPX_PARTITION_MODE] = "QPX",
+	[AMDGPU_CPX_PARTITION_MODE] = "CPX",
+};
+
+static ssize_t amdgpu_gfx_get_available_compute_partition(struct device *dev,
+						struct device_attribute *addr,
+						char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	struct amdgpu_xcp_mgr *xcp_mgr = adev->xcp_mgr;
+	int size = 0, mode;
+	char *sep = "";
+
+	if (!xcp_mgr || !xcp_mgr->avail_xcp_modes)
+		return sysfs_emit(buf, "Not supported\n");
+
+	for_each_inst(mode, xcp_mgr->avail_xcp_modes) {
+		size += sysfs_emit_at(buf, size, "%s%s", sep, xcp_desc[mode]);
+		sep = ", ";
+	}
+
+	size += sysfs_emit_at(buf, size, "\n");
+
+	return size;
+}
+
+static int amdgpu_gfx_run_cleaner_shader_job(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct drm_gpu_scheduler *sched = &ring->sched;
+	struct drm_sched_entity entity;
+	static atomic_t counter;
+	struct dma_fence *f;
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	void *owner;
+	int i, r;
+
+	/* Initialize the scheduler entity */
+	r = drm_sched_entity_init(&entity, DRM_SCHED_PRIORITY_NORMAL,
+				  &sched, 1, NULL);
+	if (r) {
+		dev_err(adev->dev, "Failed setting up GFX kernel entity.\n");
+		goto err;
+	}
+
+	/*
+	 * Use some unique dummy value as the owner to make sure we execute
+	 * the cleaner shader on each submission. The value just need to change
+	 * for each submission and is otherwise meaningless.
+	 */
+	owner = (void *)(unsigned long)atomic_inc_return(&counter);
+
+	r = amdgpu_job_alloc_with_ib(ring->adev, &entity, owner,
+				     64, 0, &job,
+				     AMDGPU_KERNEL_JOB_ID_CLEANER_SHADER);
+	if (r)
+		goto err;
+
+	job->enforce_isolation = true;
+	/* always run the cleaner shader */
+	job->run_cleaner_shader = true;
+
+	ib = &job->ibs[0];
+	for (i = 0; i <= ring->funcs->align_mask; ++i)
+		ib->ptr[i] = ring->funcs->nop;
+	ib->length_dw = ring->funcs->align_mask + 1;
+
+	f = amdgpu_job_submit(job);
+
+	r = dma_fence_wait(f, false);
+	if (r)
+		goto err;
+
+	dma_fence_put(f);
+
+	/* Clean up the scheduler entity */
+	drm_sched_entity_destroy(&entity);
+	return 0;
+
+err:
+	return r;
+}
+
+static int amdgpu_gfx_run_cleaner_shader(struct amdgpu_device *adev, int xcp_id)
+{
+	int num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	struct amdgpu_ring *ring;
+	int num_xcc_to_clear;
+	int i, r, xcc_id;
+
+	if (adev->gfx.num_xcc_per_xcp)
+		num_xcc_to_clear = adev->gfx.num_xcc_per_xcp;
+	else
+		num_xcc_to_clear = 1;
+
+	for (xcc_id = 0; xcc_id < num_xcc; xcc_id++) {
+		for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+			ring = &adev->gfx.compute_ring[i + xcc_id * adev->gfx.num_compute_rings];
+			if ((ring->xcp_id == xcp_id) && ring->sched.ready) {
+				r = amdgpu_gfx_run_cleaner_shader_job(ring);
+				if (r)
+					return r;
+				num_xcc_to_clear--;
+				break;
+			}
+		}
+	}
+
+	if (num_xcc_to_clear)
+		return -ENOENT;
+
+	return 0;
+}
+
+/**
+ * amdgpu_gfx_set_run_cleaner_shader - Execute the AMDGPU GFX Cleaner Shader
+ * @dev: The device structure
+ * @attr: The device attribute structure
+ * @buf: The buffer containing the input data
+ * @count: The size of the input data
+ *
+ * Provides the sysfs interface to manually run a cleaner shader, which is
+ * used to clear the GPU state between different tasks. Writing a value to the
+ * 'run_cleaner_shader' sysfs file triggers the cleaner shader execution.
+ * The value written corresponds to the partition index on multi-partition
+ * devices. On single-partition devices, the value should be '0'.
+ *
+ * The cleaner shader clears the Local Data Store (LDS) and General Purpose
+ * Registers (GPRs) to ensure data isolation between GPU workloads.
+ *
+ * Return: The number of bytes written to the sysfs file.
+ */
+static ssize_t amdgpu_gfx_set_run_cleaner_shader(struct device *dev,
+						 struct device_attribute *attr,
+						 const char *buf,
+						 size_t count)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	int ret;
+	long value;
+
+	if (amdgpu_in_reset(adev))
+		return -EPERM;
+	if (adev->in_suspend && !adev->in_runpm)
+		return -EPERM;
+
+	if (adev->gfx.disable_kq)
+		return -EPERM;
+
+	ret = kstrtol(buf, 0, &value);
+
+	if (ret)
+		return -EINVAL;
+
+	if (value < 0)
+		return -EINVAL;
+
+	if (adev->xcp_mgr) {
+		if (value >= adev->xcp_mgr->num_xcps)
+			return -EINVAL;
+	} else {
+		if (value > 1)
+			return -EINVAL;
+	}
+
+	ret = pm_runtime_get_sync(ddev->dev);
+	if (ret < 0) {
+		pm_runtime_put_autosuspend(ddev->dev);
+		return ret;
+	}
+
+	ret = amdgpu_gfx_run_cleaner_shader(adev, value);
+
+	pm_runtime_mark_last_busy(ddev->dev);
+	pm_runtime_put_autosuspend(ddev->dev);
+
+	if (ret)
+		return ret;
+
+	return count;
+}
+
+/**
+ * amdgpu_gfx_get_enforce_isolation - Query AMDGPU GFX Enforce Isolation Settings
+ * @dev: The device structure
+ * @attr: The device attribute structure
+ * @buf: The buffer to store the output data
+ *
+ * Provides the sysfs read interface to get the current settings of the 'enforce_isolation'
+ * feature for each GPU partition. Reading from the 'enforce_isolation'
+ * sysfs file returns the isolation settings for all partitions, where '0'
+ * indicates disabled, '1' indicates enabled, and '2' indicates enabled in legacy mode,
+ * and '3' indicates enabled without cleaner shader.
+ *
+ * Return: The number of bytes read from the sysfs file.
+ */
+static ssize_t amdgpu_gfx_get_enforce_isolation(struct device *dev,
+						struct device_attribute *attr,
+						char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	int i;
+	ssize_t size = 0;
+
+	if (adev->xcp_mgr) {
+		for (i = 0; i < adev->xcp_mgr->num_xcps; i++) {
+			size += sysfs_emit_at(buf, size, "%u", adev->enforce_isolation[i]);
+			if (i < (adev->xcp_mgr->num_xcps - 1))
+				size += sysfs_emit_at(buf, size, " ");
+		}
+		buf[size++] = '\n';
+	} else {
+		size = sysfs_emit_at(buf, 0, "%u\n", adev->enforce_isolation[0]);
+	}
+
+	return size;
+}
+
+/**
+ * amdgpu_gfx_set_enforce_isolation - Control AMDGPU GFX Enforce Isolation
+ * @dev: The device structure
+ * @attr: The device attribute structure
+ * @buf: The buffer containing the input data
+ * @count: The size of the input data
+ *
+ * This function allows control over the 'enforce_isolation' feature, which
+ * serializes access to the graphics engine. Writing '0' to disable, '1' to
+ * enable isolation with cleaner shader, '2' to enable legacy isolation without
+ * cleaner shader, or '3' to enable process isolation without submitting the
+ * cleaner shader to the 'enforce_isolation' sysfs file sets the isolation mode
+ * for each partition. The input should specify the setting for all
+ * partitions.
+ *
+ * Return: The number of bytes written to the sysfs file.
+ */
+static ssize_t amdgpu_gfx_set_enforce_isolation(struct device *dev,
+						struct device_attribute *attr,
+						const char *buf, size_t count)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	long partition_values[MAX_XCP] = {0};
+	int ret, i, num_partitions;
+	const char *input_buf = buf;
+
+	for (i = 0; i < (adev->xcp_mgr ? adev->xcp_mgr->num_xcps : 1); i++) {
+		ret = sscanf(input_buf, "%ld", &partition_values[i]);
+		if (ret <= 0)
+			break;
+
+		/* Move the pointer to the next value in the string */
+		input_buf = strchr(input_buf, ' ');
+		if (input_buf) {
+			input_buf++;
+		} else {
+			i++;
+			break;
+		}
+	}
+	num_partitions = i;
+
+	if (adev->xcp_mgr && num_partitions != adev->xcp_mgr->num_xcps)
+		return -EINVAL;
+
+	if (!adev->xcp_mgr && num_partitions != 1)
+		return -EINVAL;
+
+	for (i = 0; i < num_partitions; i++) {
+		if (partition_values[i] != 0 &&
+		    partition_values[i] != 1 &&
+		    partition_values[i] != 2 &&
+		    partition_values[i] != 3)
+			return -EINVAL;
+	}
+
+	mutex_lock(&adev->enforce_isolation_mutex);
+	for (i = 0; i < num_partitions; i++) {
+		switch (partition_values[i]) {
+		case 0:
+		default:
+			adev->enforce_isolation[i] = AMDGPU_ENFORCE_ISOLATION_DISABLE;
+			break;
+		case 1:
+			adev->enforce_isolation[i] =
+				AMDGPU_ENFORCE_ISOLATION_ENABLE;
+			break;
+		case 2:
+			adev->enforce_isolation[i] =
+				AMDGPU_ENFORCE_ISOLATION_ENABLE_LEGACY;
+			break;
+		case 3:
+			adev->enforce_isolation[i] =
+				AMDGPU_ENFORCE_ISOLATION_NO_CLEANER_SHADER;
+			break;
+		}
+	}
+	mutex_unlock(&adev->enforce_isolation_mutex);
+
+	amdgpu_mes_update_enforce_isolation(adev);
+
+	return count;
+}
+
+static ssize_t amdgpu_gfx_get_gfx_reset_mask(struct device *dev,
+						struct device_attribute *attr,
+						char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	if (!adev)
+		return -ENODEV;
+
+	return amdgpu_show_reset_mask(buf, adev->gfx.gfx_supported_reset);
+}
+
+static ssize_t amdgpu_gfx_get_compute_reset_mask(struct device *dev,
+						struct device_attribute *attr,
+						char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	if (!adev)
+		return -ENODEV;
+
+	return amdgpu_show_reset_mask(buf, adev->gfx.compute_supported_reset);
+}
+
+static DEVICE_ATTR(run_cleaner_shader, 0200,
+		   NULL, amdgpu_gfx_set_run_cleaner_shader);
+
+static DEVICE_ATTR(enforce_isolation, 0644,
+		   amdgpu_gfx_get_enforce_isolation,
+		   amdgpu_gfx_set_enforce_isolation);
+
+static DEVICE_ATTR(current_compute_partition, 0644,
+		   amdgpu_gfx_get_current_compute_partition,
+		   amdgpu_gfx_set_compute_partition);
+
+static DEVICE_ATTR(available_compute_partition, 0444,
+		   amdgpu_gfx_get_available_compute_partition, NULL);
+static DEVICE_ATTR(gfx_reset_mask, 0444,
+		   amdgpu_gfx_get_gfx_reset_mask, NULL);
+
+static DEVICE_ATTR(compute_reset_mask, 0444,
+		   amdgpu_gfx_get_compute_reset_mask, NULL);
+
+static int amdgpu_gfx_sysfs_xcp_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_xcp_mgr *xcp_mgr = adev->xcp_mgr;
+	bool xcp_switch_supported;
+	int r;
+
+	if (!xcp_mgr)
+		return 0;
+
+	xcp_switch_supported =
+		(xcp_mgr->funcs && xcp_mgr->funcs->switch_partition_mode);
+
+	if (!xcp_switch_supported)
+		dev_attr_current_compute_partition.attr.mode &=
+			~(S_IWUSR | S_IWGRP | S_IWOTH);
+
+	r = device_create_file(adev->dev, &dev_attr_current_compute_partition);
+	if (r)
+		return r;
+
+	if (xcp_switch_supported)
+		r = device_create_file(adev->dev,
+				       &dev_attr_available_compute_partition);
+
+	return r;
+}
+
+static void amdgpu_gfx_sysfs_xcp_fini(struct amdgpu_device *adev)
+{
+	struct amdgpu_xcp_mgr *xcp_mgr = adev->xcp_mgr;
+	bool xcp_switch_supported;
+
+	if (!xcp_mgr)
+		return;
+
+	xcp_switch_supported =
+		(xcp_mgr->funcs && xcp_mgr->funcs->switch_partition_mode);
+	device_remove_file(adev->dev, &dev_attr_current_compute_partition);
+
+	if (xcp_switch_supported)
+		device_remove_file(adev->dev,
+				   &dev_attr_available_compute_partition);
+}
+
+static int amdgpu_gfx_sysfs_isolation_shader_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	r = device_create_file(adev->dev, &dev_attr_enforce_isolation);
+	if (r)
+		return r;
+	if (adev->gfx.enable_cleaner_shader)
+		r = device_create_file(adev->dev, &dev_attr_run_cleaner_shader);
+
+	return r;
+}
+
+static void amdgpu_gfx_sysfs_isolation_shader_fini(struct amdgpu_device *adev)
+{
+	device_remove_file(adev->dev, &dev_attr_enforce_isolation);
+	if (adev->gfx.enable_cleaner_shader)
+		device_remove_file(adev->dev, &dev_attr_run_cleaner_shader);
+}
+
+static int amdgpu_gfx_sysfs_reset_mask_init(struct amdgpu_device *adev)
+{
+	int r = 0;
+
+	if (!amdgpu_gpu_recovery)
+		return r;
+
+	if (adev->gfx.num_gfx_rings) {
+		r = device_create_file(adev->dev, &dev_attr_gfx_reset_mask);
+		if (r)
+			return r;
+	}
+
+	if (adev->gfx.num_compute_rings) {
+		r = device_create_file(adev->dev, &dev_attr_compute_reset_mask);
+		if (r)
+			return r;
+	}
+
+	return r;
+}
+
+static void amdgpu_gfx_sysfs_reset_mask_fini(struct amdgpu_device *adev)
+{
+	if (!amdgpu_gpu_recovery)
+		return;
+
+	if (adev->gfx.num_gfx_rings)
+		device_remove_file(adev->dev, &dev_attr_gfx_reset_mask);
+
+	if (adev->gfx.num_compute_rings)
+		device_remove_file(adev->dev, &dev_attr_compute_reset_mask);
+}
+
+int amdgpu_gfx_sysfs_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	r = amdgpu_gfx_sysfs_xcp_init(adev);
+	if (r) {
+		dev_err(adev->dev, "failed to create xcp sysfs files");
+		return r;
+	}
+
+	r = amdgpu_gfx_sysfs_isolation_shader_init(adev);
+	if (r)
+		dev_err(adev->dev, "failed to create isolation sysfs files");
+
+	r = amdgpu_gfx_sysfs_reset_mask_init(adev);
+	if (r)
+		dev_err(adev->dev, "failed to create reset mask sysfs files");
+
+	return r;
+}
+
+void amdgpu_gfx_sysfs_fini(struct amdgpu_device *adev)
+{
+	if (adev->dev->kobj.sd) {
+		amdgpu_gfx_sysfs_xcp_fini(adev);
+		amdgpu_gfx_sysfs_isolation_shader_fini(adev);
+		amdgpu_gfx_sysfs_reset_mask_fini(adev);
+	}
+}
+
+int amdgpu_gfx_cleaner_shader_sw_init(struct amdgpu_device *adev,
+				      unsigned int cleaner_shader_size)
+{
+	if (!adev->gfx.enable_cleaner_shader)
+		return -EOPNOTSUPP;
+
+	return amdgpu_bo_create_kernel(adev, cleaner_shader_size, PAGE_SIZE,
+				       AMDGPU_GEM_DOMAIN_VRAM | AMDGPU_GEM_DOMAIN_GTT,
+				       &adev->gfx.cleaner_shader_obj,
+				       &adev->gfx.cleaner_shader_gpu_addr,
+				       (void **)&adev->gfx.cleaner_shader_cpu_ptr);
+}
+
+void amdgpu_gfx_cleaner_shader_sw_fini(struct amdgpu_device *adev)
+{
+	if (!adev->gfx.enable_cleaner_shader)
+		return;
+
+	amdgpu_bo_free_kernel(&adev->gfx.cleaner_shader_obj,
+			      &adev->gfx.cleaner_shader_gpu_addr,
+			      (void **)&adev->gfx.cleaner_shader_cpu_ptr);
+}
+
+void amdgpu_gfx_cleaner_shader_init(struct amdgpu_device *adev,
+				    unsigned int cleaner_shader_size,
+				    const void *cleaner_shader_ptr)
+{
+	if (!adev->gfx.enable_cleaner_shader)
+		return;
+
+	if (adev->gfx.cleaner_shader_cpu_ptr && cleaner_shader_ptr)
+		memcpy_toio(adev->gfx.cleaner_shader_cpu_ptr, cleaner_shader_ptr,
+			    cleaner_shader_size);
+}
+
+/**
+ * amdgpu_gfx_kfd_sch_ctrl - Control the KFD scheduler from the KGD (Graphics Driver)
+ * @adev: amdgpu_device pointer
+ * @idx: Index of the scheduler to control
+ * @enable: Whether to enable or disable the KFD scheduler
+ *
+ * This function is used to control the KFD (Kernel Fusion Driver) scheduler
+ * from the KGD. It is part of the cleaner shader feature. This function plays
+ * a key role in enforcing process isolation on the GPU.
+ *
+ * The function uses a reference count mechanism (kfd_sch_req_count) to keep
+ * track of the number of requests to enable the KFD scheduler. When a request
+ * to enable the KFD scheduler is made, the reference count is decremented.
+ * When the reference count reaches zero, a delayed work is scheduled to
+ * enforce isolation after a delay of GFX_SLICE_PERIOD.
+ *
+ * When a request to disable the KFD scheduler is made, the function first
+ * checks if the reference count is zero. If it is, it cancels the delayed work
+ * for enforcing isolation and checks if the KFD scheduler is active. If the
+ * KFD scheduler is active, it sends a request to stop the KFD scheduler and
+ * sets the KFD scheduler state to inactive. Then, it increments the reference
+ * count.
+ *
+ * The function is synchronized using the kfd_sch_mutex to ensure that the KFD
+ * scheduler state and reference count are updated atomically.
+ *
+ * Note: If the reference count is already zero when a request to enable the
+ * KFD scheduler is made, it means there's an imbalance bug somewhere. The
+ * function triggers a warning in this case.
+ */
+static void amdgpu_gfx_kfd_sch_ctrl(struct amdgpu_device *adev, u32 idx,
+				    bool enable)
+{
+	mutex_lock(&adev->gfx.userq_sch_mutex);
+
+	if (enable) {
+		/* If the count is already 0, it means there's an imbalance bug somewhere.
+		 * Note that the bug may be in a different caller than the one which triggers the
+		 * WARN_ON_ONCE.
+		 */
+		if (WARN_ON_ONCE(adev->gfx.userq_sch_req_count[idx] == 0)) {
+			dev_err(adev->dev, "Attempted to enable KFD scheduler when reference count is already zero\n");
+			goto unlock;
+		}
+
+		adev->gfx.userq_sch_req_count[idx]--;
+
+		if (adev->gfx.userq_sch_req_count[idx] == 0 &&
+		    adev->gfx.userq_sch_inactive[idx]) {
+			schedule_delayed_work(&adev->gfx.enforce_isolation[idx].work,
+					      msecs_to_jiffies(adev->gfx.enforce_isolation_time[idx]));
+		}
+	} else {
+		if (adev->gfx.userq_sch_req_count[idx] == 0) {
+			cancel_delayed_work_sync(&adev->gfx.enforce_isolation[idx].work);
+			if (!adev->gfx.userq_sch_inactive[idx]) {
+				amdgpu_userq_stop_sched_for_enforce_isolation(adev, idx);
+				if (adev->kfd.init_complete)
+					amdgpu_amdkfd_stop_sched(adev, idx);
+				adev->gfx.userq_sch_inactive[idx] = true;
+			}
+		}
+
+		adev->gfx.userq_sch_req_count[idx]++;
+	}
+
+unlock:
+	mutex_unlock(&adev->gfx.userq_sch_mutex);
+}
+
+/**
+ * amdgpu_gfx_enforce_isolation_handler - work handler for enforcing shader isolation
+ *
+ * @work: work_struct.
+ *
+ * This function is the work handler for enforcing shader isolation on AMD GPUs.
+ * It counts the number of emitted fences for each GFX and compute ring. If there
+ * are any fences, it schedules the `enforce_isolation_work` to be run after a
+ * delay of `GFX_SLICE_PERIOD`. If there are no fences, it signals the Kernel Fusion
+ * Driver (KFD) to resume the runqueue. The function is synchronized using the
+ * `enforce_isolation_mutex`.
+ */
+void amdgpu_gfx_enforce_isolation_handler(struct work_struct *work)
+{
+	struct amdgpu_isolation_work *isolation_work =
+		container_of(work, struct amdgpu_isolation_work, work.work);
+	struct amdgpu_device *adev = isolation_work->adev;
+	u32 i, idx, fences = 0;
+
+	if (isolation_work->xcp_id == AMDGPU_XCP_NO_PARTITION)
+		idx = 0;
+	else
+		idx = isolation_work->xcp_id;
+
+	if (idx >= MAX_XCP)
+		return;
+
+	mutex_lock(&adev->enforce_isolation_mutex);
+	for (i = 0; i < AMDGPU_MAX_GFX_RINGS; ++i) {
+		if (isolation_work->xcp_id == adev->gfx.gfx_ring[i].xcp_id)
+			fences += amdgpu_fence_count_emitted(&adev->gfx.gfx_ring[i]);
+	}
+	for (i = 0; i < (AMDGPU_MAX_COMPUTE_RINGS * AMDGPU_MAX_GC_INSTANCES); ++i) {
+		if (isolation_work->xcp_id == adev->gfx.compute_ring[i].xcp_id)
+			fences += amdgpu_fence_count_emitted(&adev->gfx.compute_ring[i]);
+	}
+	if (fences) {
+		/* we've already had our timeslice, so let's wrap this up */
+		schedule_delayed_work(&adev->gfx.enforce_isolation[idx].work,
+				      msecs_to_jiffies(1));
+	} else {
+		/* Tell KFD to resume the runqueue */
+		WARN_ON_ONCE(!adev->gfx.userq_sch_inactive[idx]);
+		WARN_ON_ONCE(adev->gfx.userq_sch_req_count[idx]);
+
+		amdgpu_userq_start_sched_for_enforce_isolation(adev, idx);
+		if (adev->kfd.init_complete)
+			amdgpu_amdkfd_start_sched(adev, idx);
+		adev->gfx.userq_sch_inactive[idx] = false;
+	}
+	mutex_unlock(&adev->enforce_isolation_mutex);
+}
+
+/**
+ * amdgpu_gfx_enforce_isolation_wait_for_kfd - Manage KFD wait period for process isolation
+ * @adev: amdgpu_device pointer
+ * @idx: Index of the GPU partition
+ *
+ * When kernel submissions come in, the jobs are given a time slice and once
+ * that time slice is up, if there are KFD user queues active, kernel
+ * submissions are blocked until KFD has had its time slice. Once the KFD time
+ * slice is up, KFD user queues are preempted and kernel submissions are
+ * unblocked and allowed to run again.
+ */
+static void
+amdgpu_gfx_enforce_isolation_wait_for_kfd(struct amdgpu_device *adev,
+					  u32 idx)
+{
+	unsigned long cjiffies;
+	bool wait = false;
+
+	mutex_lock(&adev->enforce_isolation_mutex);
+	if (adev->enforce_isolation[idx] == AMDGPU_ENFORCE_ISOLATION_ENABLE) {
+		/* set the initial values if nothing is set */
+		if (!adev->gfx.enforce_isolation_jiffies[idx]) {
+			adev->gfx.enforce_isolation_jiffies[idx] = jiffies;
+			adev->gfx.enforce_isolation_time[idx] =	GFX_SLICE_PERIOD_MS;
+		}
+		/* Make sure KFD gets a chance to run */
+		if (amdgpu_amdkfd_compute_active(adev, idx)) {
+			cjiffies = jiffies;
+			if (time_after(cjiffies, adev->gfx.enforce_isolation_jiffies[idx])) {
+				cjiffies -= adev->gfx.enforce_isolation_jiffies[idx];
+				if ((jiffies_to_msecs(cjiffies) >= GFX_SLICE_PERIOD_MS)) {
+					/* if our time is up, let KGD work drain before scheduling more */
+					wait = true;
+					/* reset the timer period */
+					adev->gfx.enforce_isolation_time[idx] =	GFX_SLICE_PERIOD_MS;
+				} else {
+					/* set the timer period to what's left in our time slice */
+					adev->gfx.enforce_isolation_time[idx] =
+						GFX_SLICE_PERIOD_MS - jiffies_to_msecs(cjiffies);
+				}
+			} else {
+				/* if jiffies wrap around we will just wait a little longer */
+				adev->gfx.enforce_isolation_jiffies[idx] = jiffies;
+			}
+		} else {
+			/* if there is no KFD work, then set the full slice period */
+			adev->gfx.enforce_isolation_jiffies[idx] = jiffies;
+			adev->gfx.enforce_isolation_time[idx] = GFX_SLICE_PERIOD_MS;
+		}
+	}
+	mutex_unlock(&adev->enforce_isolation_mutex);
+
+	if (wait)
+		msleep(GFX_SLICE_PERIOD_MS);
+}
+
+/**
+ * amdgpu_gfx_enforce_isolation_ring_begin_use - Begin use of a ring with enforced isolation
+ * @ring: Pointer to the amdgpu_ring structure
+ *
+ * Ring begin_use helper implementation for gfx which serializes access to the
+ * gfx IP between kernel submission IOCTLs and KFD user queues when isolation
+ * enforcement is enabled. The kernel submission IOCTLs and KFD user queues
+ * each get a time slice when both are active.
+ */
+void amdgpu_gfx_enforce_isolation_ring_begin_use(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 idx;
+	bool sched_work = false;
+
+	if (!adev->gfx.enable_cleaner_shader)
+		return;
+
+	if (ring->xcp_id == AMDGPU_XCP_NO_PARTITION)
+		idx = 0;
+	else
+		idx = ring->xcp_id;
+
+	if (idx >= MAX_XCP)
+		return;
+
+	/* Don't submit more work until KFD has had some time */
+	amdgpu_gfx_enforce_isolation_wait_for_kfd(adev, idx);
+
+	mutex_lock(&adev->enforce_isolation_mutex);
+	if (adev->enforce_isolation[idx] == AMDGPU_ENFORCE_ISOLATION_ENABLE) {
+		if (adev->kfd.init_complete)
+			sched_work = true;
+	}
+	mutex_unlock(&adev->enforce_isolation_mutex);
+
+	if (sched_work)
+		amdgpu_gfx_kfd_sch_ctrl(adev, idx, false);
+}
+
+/**
+ * amdgpu_gfx_enforce_isolation_ring_end_use - End use of a ring with enforced isolation
+ * @ring: Pointer to the amdgpu_ring structure
+ *
+ * Ring end_use helper implementation for gfx which serializes access to the
+ * gfx IP between kernel submission IOCTLs and KFD user queues when isolation
+ * enforcement is enabled. The kernel submission IOCTLs and KFD user queues
+ * each get a time slice when both are active.
+ */
+void amdgpu_gfx_enforce_isolation_ring_end_use(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 idx;
+	bool sched_work = false;
+
+	if (!adev->gfx.enable_cleaner_shader)
+		return;
+
+	if (ring->xcp_id == AMDGPU_XCP_NO_PARTITION)
+		idx = 0;
+	else
+		idx = ring->xcp_id;
+
+	if (idx >= MAX_XCP)
+		return;
+
+	mutex_lock(&adev->enforce_isolation_mutex);
+	if (adev->enforce_isolation[idx] == AMDGPU_ENFORCE_ISOLATION_ENABLE) {
+		if (adev->kfd.init_complete)
+			sched_work = true;
+	}
+	mutex_unlock(&adev->enforce_isolation_mutex);
+
+	if (sched_work)
+		amdgpu_gfx_kfd_sch_ctrl(adev, idx, true);
+}
+
+void amdgpu_gfx_profile_idle_work_handler(struct work_struct *work)
+{
+	struct amdgpu_device *adev =
+		container_of(work, struct amdgpu_device, gfx.idle_work.work);
+	enum PP_SMC_POWER_PROFILE profile;
+	u32 i, fences = 0;
+	int r;
+
+	if (adev->gfx.num_gfx_rings)
+		profile = PP_SMC_POWER_PROFILE_FULLSCREEN3D;
+	else
+		profile = PP_SMC_POWER_PROFILE_COMPUTE;
+
+	for (i = 0; i < AMDGPU_MAX_GFX_RINGS; ++i)
+		fences += amdgpu_fence_count_emitted(&adev->gfx.gfx_ring[i]);
+	for (i = 0; i < (AMDGPU_MAX_COMPUTE_RINGS * AMDGPU_MAX_GC_INSTANCES); ++i)
+		fences += amdgpu_fence_count_emitted(&adev->gfx.compute_ring[i]);
+	if (!fences && !atomic_read(&adev->gfx.total_submission_cnt)) {
+		mutex_lock(&adev->gfx.workload_profile_mutex);
+		if (adev->gfx.workload_profile_active) {
+			r = amdgpu_dpm_switch_power_profile(adev, profile, false);
+			if (r)
+				dev_warn(adev->dev, "(%d) failed to disable %s power profile mode\n", r,
+					 profile == PP_SMC_POWER_PROFILE_FULLSCREEN3D ?
+					 "fullscreen 3D" : "compute");
+			adev->gfx.workload_profile_active = false;
+		}
+		mutex_unlock(&adev->gfx.workload_profile_mutex);
+	} else {
+		schedule_delayed_work(&adev->gfx.idle_work, GFX_PROFILE_IDLE_TIMEOUT);
+	}
+}
+
+void amdgpu_gfx_profile_ring_begin_use(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	enum PP_SMC_POWER_PROFILE profile;
+	int r;
+
+	if (amdgpu_dpm_is_overdrive_enabled(adev))
+		return;
+
+	if (adev->gfx.num_gfx_rings)
+		profile = PP_SMC_POWER_PROFILE_FULLSCREEN3D;
+	else
+		profile = PP_SMC_POWER_PROFILE_COMPUTE;
+
+	atomic_inc(&adev->gfx.total_submission_cnt);
+
+	cancel_delayed_work_sync(&adev->gfx.idle_work);
+
+	/* We can safely return early here because we've cancelled the
+	 * the delayed work so there is no one else to set it to false
+	 * and we don't care if someone else sets it to true.
+	 */
+	if (adev->gfx.workload_profile_active)
+		return;
+
+	mutex_lock(&adev->gfx.workload_profile_mutex);
+	if (!adev->gfx.workload_profile_active) {
+		r = amdgpu_dpm_switch_power_profile(adev, profile, true);
+		if (r)
+			dev_warn(adev->dev, "(%d) failed to disable %s power profile mode\n", r,
+				 profile == PP_SMC_POWER_PROFILE_FULLSCREEN3D ?
+				 "fullscreen 3D" : "compute");
+		adev->gfx.workload_profile_active = true;
+	}
+	mutex_unlock(&adev->gfx.workload_profile_mutex);
+}
+
+void amdgpu_gfx_profile_ring_end_use(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (amdgpu_dpm_is_overdrive_enabled(adev))
+		return;
+
+	atomic_dec(&ring->adev->gfx.total_submission_cnt);
+
+	schedule_delayed_work(&ring->adev->gfx.idle_work, GFX_PROFILE_IDLE_TIMEOUT);
+}
+
+/**
+ * amdgpu_gfx_csb_preamble_start - Set CSB preamble start
+ *
+ * @buffer: This is an output variable that gets the PACKET3 preamble setup.
+ *
+ * Return:
+ * return the latest index.
+ */
+u32 amdgpu_gfx_csb_preamble_start(u32 *buffer)
+{
+	u32 count = 0;
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	buffer[count++] = cpu_to_le32(0x80000000);
+	buffer[count++] = cpu_to_le32(0x80000000);
+
+	return count;
+}
+
+/**
+ * amdgpu_gfx_csb_data_parser - Parser CS data
+ *
+ * @adev: amdgpu_device pointer used to get the CS data and other gfx info.
+ * @buffer: This is an output variable that gets the PACKET3 preamble end.
+ * @count: Index to start set the preemble end.
+ *
+ * Return:
+ * return the latest index.
+ */
+u32 amdgpu_gfx_csb_data_parser(struct amdgpu_device *adev, u32 *buffer, u32 count)
+{
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+	u32 i;
+
+	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT) {
+				buffer[count++] = cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, ext->reg_count));
+				buffer[count++] = cpu_to_le32(ext->reg_index - PACKET3_SET_CONTEXT_REG_START);
+
+				for (i = 0; i < ext->reg_count; i++)
+					buffer[count++] = cpu_to_le32(ext->extent[i]);
+			}
+		}
+	}
+
+	return count;
+}
+
+/**
+ * amdgpu_gfx_csb_preamble_end - Set CSB preamble end
+ *
+ * @buffer: This is an output variable that gets the PACKET3 preamble end.
+ * @count: Index to start set the preemble end.
+ */
+void amdgpu_gfx_csb_preamble_end(u32 *buffer, u32 count)
+{
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_END_CLEAR_STATE);
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CLEAR_STATE, 0));
+	buffer[count++] = cpu_to_le32(0);
+}
+
+/*
+ * debugfs for to enable/disable gfx job submission to specific core.
+ */
+#if defined(CONFIG_DEBUG_FS)
+static int amdgpu_debugfs_gfx_sched_mask_set(void *data, u64 val)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)data;
+	u32 i;
+	u64 mask = 0;
+	struct amdgpu_ring *ring;
+
+	if (!adev)
+		return -ENODEV;
+
+	mask = (1ULL << adev->gfx.num_gfx_rings) - 1;
+	if ((val & mask) == 0)
+		return -EINVAL;
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; ++i) {
+		ring = &adev->gfx.gfx_ring[i];
+		if (val & (1 << i))
+			ring->sched.ready = true;
+		else
+			ring->sched.ready = false;
+	}
+	/* publish sched.ready flag update effective immediately across smp */
+	smp_rmb();
+	return 0;
+}
+
+static int amdgpu_debugfs_gfx_sched_mask_get(void *data, u64 *val)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)data;
+	u32 i;
+	u64 mask = 0;
+	struct amdgpu_ring *ring;
+
+	if (!adev)
+		return -ENODEV;
+	for (i = 0; i < adev->gfx.num_gfx_rings; ++i) {
+		ring = &adev->gfx.gfx_ring[i];
+		if (ring->sched.ready)
+			mask |= 1ULL << i;
+	}
+
+	*val = mask;
+	return 0;
+}
+
+DEFINE_DEBUGFS_ATTRIBUTE(amdgpu_debugfs_gfx_sched_mask_fops,
+			 amdgpu_debugfs_gfx_sched_mask_get,
+			 amdgpu_debugfs_gfx_sched_mask_set, "%llx\n");
+
+#endif
+
+void amdgpu_debugfs_gfx_sched_mask_init(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	struct drm_minor *minor = adev_to_drm(adev)->primary;
+	struct dentry *root = minor->debugfs_root;
+	char name[32];
+
+	if (!(adev->gfx.num_gfx_rings > 1))
+		return;
+	sprintf(name, "amdgpu_gfx_sched_mask");
+	debugfs_create_file(name, 0600, root, adev,
+			    &amdgpu_debugfs_gfx_sched_mask_fops);
+#endif
+}
+
+/*
+ * debugfs for to enable/disable compute job submission to specific core.
+ */
+#if defined(CONFIG_DEBUG_FS)
+static int amdgpu_debugfs_compute_sched_mask_set(void *data, u64 val)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)data;
+	u32 i;
+	u64 mask = 0;
+	struct amdgpu_ring *ring;
+
+	if (!adev)
+		return -ENODEV;
+
+	mask = (1ULL << adev->gfx.num_compute_rings) - 1;
+	if ((val & mask) == 0)
+		return -EINVAL;
+
+	for (i = 0; i < adev->gfx.num_compute_rings; ++i) {
+		ring = &adev->gfx.compute_ring[i];
+		if (val & (1 << i))
+			ring->sched.ready = true;
+		else
+			ring->sched.ready = false;
+	}
+
+	/* publish sched.ready flag update effective immediately across smp */
+	smp_rmb();
+	return 0;
+}
+
+static int amdgpu_debugfs_compute_sched_mask_get(void *data, u64 *val)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)data;
+	u32 i;
+	u64 mask = 0;
+	struct amdgpu_ring *ring;
+
+	if (!adev)
+		return -ENODEV;
+	for (i = 0; i < adev->gfx.num_compute_rings; ++i) {
+		ring = &adev->gfx.compute_ring[i];
+		if (ring->sched.ready)
+			mask |= 1ULL << i;
+	}
+
+	*val = mask;
+	return 0;
+}
+
+DEFINE_DEBUGFS_ATTRIBUTE(amdgpu_debugfs_compute_sched_mask_fops,
+			 amdgpu_debugfs_compute_sched_mask_get,
+			 amdgpu_debugfs_compute_sched_mask_set, "%llx\n");
+
+#endif
+
+void amdgpu_debugfs_compute_sched_mask_init(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	struct drm_minor *minor = adev_to_drm(adev)->primary;
+	struct dentry *root = minor->debugfs_root;
+	char name[32];
+
+	if (!(adev->gfx.num_compute_rings > 1))
+		return;
+	sprintf(name, "amdgpu_compute_sched_mask");
+	debugfs_create_file(name, 0600, root, adev,
+			    &amdgpu_debugfs_compute_sched_mask_fops);
+#endif
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
index 1f279050d334..fb5f7a0ee029 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
@@ -24,28 +24,533 @@
 #ifndef __AMDGPU_GFX_H__
 #define __AMDGPU_GFX_H__
 
-int amdgpu_gfx_scratch_get(struct amdgpu_device *adev, uint32_t *reg);
-void amdgpu_gfx_scratch_free(struct amdgpu_device *adev, uint32_t reg);
+/*
+ * GFX stuff
+ */
+#include "clearstate_defs.h"
+#include "amdgpu_ring.h"
+#include "amdgpu_rlc.h"
+#include "amdgpu_imu.h"
+#include "soc15.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_ring_mux.h"
+#include "amdgpu_xcp.h"
 
-void amdgpu_gfx_parse_disable_cu(unsigned *mask, unsigned max_se,
-		unsigned max_sh);
+/* GFX current status */
+#define AMDGPU_GFX_NORMAL_MODE			0x00000000L
+#define AMDGPU_GFX_SAFE_MODE			0x00000001L
+#define AMDGPU_GFX_PG_DISABLED_MODE		0x00000002L
+#define AMDGPU_GFX_CG_DISABLED_MODE		0x00000004L
+#define AMDGPU_GFX_LBPW_DISABLED_MODE		0x00000008L
 
-void amdgpu_gfx_compute_queue_acquire(struct amdgpu_device *adev);
+#define AMDGPU_MAX_GC_INSTANCES		8
+#define AMDGPU_MAX_QUEUES		128
 
-int amdgpu_gfx_kiq_init_ring(struct amdgpu_device *adev,
-			     struct amdgpu_ring *ring,
-			     struct amdgpu_irq_src *irq);
+#define AMDGPU_MAX_GFX_QUEUES AMDGPU_MAX_QUEUES
+#define AMDGPU_MAX_COMPUTE_QUEUES AMDGPU_MAX_QUEUES
 
-void amdgpu_gfx_kiq_free_ring(struct amdgpu_ring *ring,
-			      struct amdgpu_irq_src *irq);
+enum amdgpu_gfx_pipe_priority {
+	AMDGPU_GFX_PIPE_PRIO_NORMAL = AMDGPU_RING_PRIO_1,
+	AMDGPU_GFX_PIPE_PRIO_HIGH = AMDGPU_RING_PRIO_2
+};
 
-void amdgpu_gfx_kiq_fini(struct amdgpu_device *adev);
-int amdgpu_gfx_kiq_init(struct amdgpu_device *adev,
-			unsigned hpd_size);
+#define AMDGPU_GFX_QUEUE_PRIORITY_MINIMUM  0
+#define AMDGPU_GFX_QUEUE_PRIORITY_MAXIMUM  15
+
+/* 1 second timeout */
+#define GFX_PROFILE_IDLE_TIMEOUT	msecs_to_jiffies(1000)
+
+enum amdgpu_gfx_partition {
+	AMDGPU_SPX_PARTITION_MODE = 0,
+	AMDGPU_DPX_PARTITION_MODE = 1,
+	AMDGPU_TPX_PARTITION_MODE = 2,
+	AMDGPU_QPX_PARTITION_MODE = 3,
+	AMDGPU_CPX_PARTITION_MODE = 4,
+	AMDGPU_UNKNOWN_COMPUTE_PARTITION_MODE = -1,
+	/* Automatically choose the right mode */
+	AMDGPU_AUTO_COMPUTE_PARTITION_MODE = -2,
+};
+
+#define NUM_XCC(x) hweight16(x)
+
+enum amdgpu_gfx_ras_mem_id_type {
+	AMDGPU_GFX_CP_MEM = 0,
+	AMDGPU_GFX_GCEA_MEM,
+	AMDGPU_GFX_GC_CANE_MEM,
+	AMDGPU_GFX_GCUTCL2_MEM,
+	AMDGPU_GFX_GDS_MEM,
+	AMDGPU_GFX_LDS_MEM,
+	AMDGPU_GFX_RLC_MEM,
+	AMDGPU_GFX_SP_MEM,
+	AMDGPU_GFX_SPI_MEM,
+	AMDGPU_GFX_SQC_MEM,
+	AMDGPU_GFX_SQ_MEM,
+	AMDGPU_GFX_TA_MEM,
+	AMDGPU_GFX_TCC_MEM,
+	AMDGPU_GFX_TCA_MEM,
+	AMDGPU_GFX_TCI_MEM,
+	AMDGPU_GFX_TCP_MEM,
+	AMDGPU_GFX_TD_MEM,
+	AMDGPU_GFX_TCX_MEM,
+	AMDGPU_GFX_ATC_L2_MEM,
+	AMDGPU_GFX_UTCL2_MEM,
+	AMDGPU_GFX_VML2_MEM,
+	AMDGPU_GFX_VML2_WALKER_MEM,
+	AMDGPU_GFX_MEM_TYPE_NUM
+};
+
+struct amdgpu_mec {
+	struct amdgpu_bo	*hpd_eop_obj;
+	u64			hpd_eop_gpu_addr;
+	struct amdgpu_bo	*mec_fw_obj;
+	u64			mec_fw_gpu_addr;
+	struct amdgpu_bo	*mec_fw_data_obj;
+	u64			mec_fw_data_gpu_addr;
+
+	u32 num_mec;
+	u32 num_pipe_per_mec;
+	u32 num_queue_per_pipe;
+	void			*mqd_backup[AMDGPU_MAX_COMPUTE_RINGS * AMDGPU_MAX_GC_INSTANCES];
+};
+
+struct amdgpu_mec_bitmap {
+	/* These are the resources for which amdgpu takes ownership */
+	DECLARE_BITMAP(queue_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
+};
+
+enum amdgpu_unmap_queues_action {
+	PREEMPT_QUEUES = 0,
+	RESET_QUEUES,
+	DISABLE_PROCESS_QUEUES,
+	PREEMPT_QUEUES_NO_UNMAP,
+};
+
+struct kiq_pm4_funcs {
+	/* Support ASIC-specific kiq pm4 packets*/
+	void (*kiq_set_resources)(struct amdgpu_ring *kiq_ring,
+					uint64_t queue_mask);
+	void (*kiq_map_queues)(struct amdgpu_ring *kiq_ring,
+					struct amdgpu_ring *ring);
+	void (*kiq_unmap_queues)(struct amdgpu_ring *kiq_ring,
+				 struct amdgpu_ring *ring,
+				 enum amdgpu_unmap_queues_action action,
+				 u64 gpu_addr, u64 seq);
+	void (*kiq_query_status)(struct amdgpu_ring *kiq_ring,
+					struct amdgpu_ring *ring,
+					u64 addr,
+					u64 seq);
+	void (*kiq_invalidate_tlbs)(struct amdgpu_ring *kiq_ring,
+				uint16_t pasid, uint32_t flush_type,
+				bool all_hub);
+	void (*kiq_reset_hw_queue)(struct amdgpu_ring *kiq_ring,
+				   uint32_t queue_type, uint32_t me_id,
+				   uint32_t pipe_id, uint32_t queue_id,
+				   uint32_t xcc_id, uint32_t vmid);
+	/* Packet sizes */
+	int set_resources_size;
+	int map_queues_size;
+	int unmap_queues_size;
+	int query_status_size;
+	int invalidate_tlbs_size;
+};
+
+struct amdgpu_kiq {
+	u64			eop_gpu_addr;
+	struct amdgpu_bo	*eop_obj;
+	spinlock_t              ring_lock;
+	struct amdgpu_ring	ring;
+	struct amdgpu_irq_src	irq;
+	const struct kiq_pm4_funcs *pmf;
+	void			*mqd_backup;
+};
+
+/*
+ * GFX configurations
+ */
+#define AMDGPU_GFX_MAX_SE 4
+#define AMDGPU_GFX_MAX_SH_PER_SE 2
+
+/**
+ * amdgpu_rb_config - Configure a single Render Backend (RB)
+ *
+ * Bad RBs are fused off and there is a harvest register the driver reads to
+ * determine which RB(s) are fused off so that the driver can configure the
+ * hardware state so that nothing gets sent to them. There are also user
+ * harvest registers that the driver can program to disable additional RBs,
+ * etc., for testing purposes.
+ */
+struct amdgpu_rb_config {
+	/**
+	 * @rb_backend_disable:
+	 *
+	 * The value captured from register RB_BACKEND_DISABLE indicates if the
+	 * RB backend is disabled or not.
+	 */
+	uint32_t rb_backend_disable;
+
+	/**
+	 * @user_rb_backend_disable:
+	 *
+	 * The value captured from register USER_RB_BACKEND_DISABLE indicates
+	 * if the User RB backend is disabled or not.
+	 */
+	uint32_t user_rb_backend_disable;
+
+	/**
+	 * @raster_config:
+	 *
+	 * To set up all of the states, it is necessary to have two registers
+	 * to keep all of the states. This field holds the first register.
+	 */
+	uint32_t raster_config;
+
+	/**
+	 * @raster_config_1:
+	 *
+	 * To set up all of the states, it is necessary to have two registers
+	 * to keep all of the states. This field holds the second register.
+	 */
+	uint32_t raster_config_1;
+};
+
+struct gb_addr_config {
+	uint16_t pipe_interleave_size;
+	uint8_t num_pipes;
+	uint8_t max_compress_frags;
+	uint8_t num_banks;
+	uint8_t num_se;
+	uint8_t num_rb_per_se;
+	uint8_t num_pkrs;
+};
+
+struct amdgpu_gfx_config {
+	unsigned max_shader_engines;
+	unsigned max_tile_pipes;
+	unsigned max_cu_per_sh;
+	unsigned max_sh_per_se;
+	unsigned max_backends_per_se;
+	unsigned max_texture_channel_caches;
+	unsigned max_gprs;
+	unsigned max_gs_threads;
+	unsigned max_hw_contexts;
+	unsigned sc_prim_fifo_size_frontend;
+	unsigned sc_prim_fifo_size_backend;
+	unsigned sc_hiz_tile_fifo_size;
+	unsigned sc_earlyz_tile_fifo_size;
+
+	unsigned num_tile_pipes;
+	unsigned backend_enable_mask;
+	unsigned mem_max_burst_length_bytes;
+	unsigned mem_row_size_in_kb;
+	unsigned shader_engine_tile_size;
+	unsigned num_gpus;
+	unsigned multi_gpu_tile_size;
+	unsigned mc_arb_ramcfg;
+	unsigned num_banks;
+	unsigned num_ranks;
+	unsigned gb_addr_config;
+	unsigned num_rbs;
+	unsigned gs_vgt_table_depth;
+	unsigned gs_prim_buffer_depth;
+
+	uint32_t tile_mode_array[32];
+	uint32_t macrotile_mode_array[16];
+
+	struct gb_addr_config gb_addr_config_fields;
+
+	/**
+	 * @rb_config:
+	 *
+	 * Matrix that keeps all the Render Backend (color and depth buffer
+	 * handling) configuration on the 3D engine.
+	 */
+	struct amdgpu_rb_config rb_config[AMDGPU_GFX_MAX_SE][AMDGPU_GFX_MAX_SH_PER_SE];
+
+	/* gfx configure feature */
+	uint32_t double_offchip_lds_buf;
+	/* cached value of DB_DEBUG2 */
+	uint32_t db_debug2;
+	/* gfx10 specific config */
+	uint32_t num_sc_per_sh;
+	uint32_t num_packer_per_sc;
+	uint32_t pa_sc_tile_steering_override;
+	/* Whether texture coordinate truncation is conformant. */
+	bool ta_cntl2_truncate_coord_mode;
+	uint64_t tcc_disabled_mask;
+	uint32_t gc_num_tcp_per_sa;
+	uint32_t gc_num_sdp_interface;
+	uint32_t gc_num_tcps;
+	uint32_t gc_num_tcp_per_wpg;
+	uint32_t gc_tcp_l1_size;
+	uint32_t gc_num_sqc_per_wgp;
+	uint32_t gc_l1_instruction_cache_size_per_sqc;
+	uint32_t gc_l1_data_cache_size_per_sqc;
+	uint32_t gc_gl1c_per_sa;
+	uint32_t gc_gl1c_size_per_instance;
+	uint32_t gc_gl2c_per_gpu;
+	uint32_t gc_tcp_size_per_cu;
+	uint32_t gc_num_cu_per_sqc;
+	uint32_t gc_tcc_size;
+	uint32_t gc_tcp_cache_line_size;
+	uint32_t gc_instruction_cache_size_per_sqc;
+	uint32_t gc_instruction_cache_line_size;
+	uint32_t gc_scalar_data_cache_size_per_sqc;
+	uint32_t gc_scalar_data_cache_line_size;
+	uint32_t gc_tcc_cache_line_size;
+};
+
+struct amdgpu_cu_info {
+	uint32_t simd_per_cu;
+	uint32_t max_waves_per_simd;
+	uint32_t wave_front_size;
+	uint32_t max_scratch_slots_per_cu;
+	uint32_t lds_size;
+
+	/* total active CU number */
+	uint32_t number;
+	uint32_t ao_cu_mask;
+	uint32_t ao_cu_bitmap[4][4];
+	uint32_t bitmap[AMDGPU_MAX_GC_INSTANCES][4][4];
+};
+
+struct amdgpu_gfx_ras {
+	struct amdgpu_ras_block_object  ras_block;
+	void (*enable_watchdog_timer)(struct amdgpu_device *adev);
+	int (*rlc_gc_fed_irq)(struct amdgpu_device *adev,
+				struct amdgpu_irq_src *source,
+				struct amdgpu_iv_entry *entry);
+	int (*poison_consumption_handler)(struct amdgpu_device *adev,
+						struct amdgpu_iv_entry *entry);
+};
+
+struct amdgpu_gfx_shadow_info {
+	u32 shadow_size;
+	u32 shadow_alignment;
+	u32 csa_size;
+	u32 csa_alignment;
+};
+
+struct amdgpu_gfx_funcs {
+	/* get the gpu clock counter */
+	uint64_t (*get_gpu_clock_counter)(struct amdgpu_device *adev);
+	void (*select_se_sh)(struct amdgpu_device *adev, u32 se_num,
+			     u32 sh_num, u32 instance, int xcc_id);
+	void (*read_wave_data)(struct amdgpu_device *adev, uint32_t xcc_id, uint32_t simd,
+			       uint32_t wave, uint32_t *dst, int *no_fields);
+	void (*read_wave_vgprs)(struct amdgpu_device *adev, uint32_t xcc_id, uint32_t simd,
+				uint32_t wave, uint32_t thread, uint32_t start,
+				uint32_t size, uint32_t *dst);
+	void (*read_wave_sgprs)(struct amdgpu_device *adev, uint32_t xcc_id, uint32_t simd,
+				uint32_t wave, uint32_t start, uint32_t size,
+				uint32_t *dst);
+	void (*select_me_pipe_q)(struct amdgpu_device *adev, u32 me, u32 pipe,
+				 u32 queue, u32 vmid, u32 xcc_id);
+	void (*init_spm_golden)(struct amdgpu_device *adev);
+	void (*update_perfmon_mgcg)(struct amdgpu_device *adev, bool enable);
+	int (*get_gfx_shadow_info)(struct amdgpu_device *adev,
+				   struct amdgpu_gfx_shadow_info *shadow_info,
+				   bool skip_check);
+	enum amdgpu_gfx_partition
+			(*query_partition_mode)(struct amdgpu_device *adev);
+	int (*switch_partition_mode)(struct amdgpu_device *adev,
+				     int num_xccs_per_xcp);
+	int (*ih_node_to_logical_xcc)(struct amdgpu_device *adev, int ih_node);
+	int (*get_xccs_per_xcp)(struct amdgpu_device *adev);
+};
+
+struct sq_work {
+	struct work_struct	work;
+	unsigned ih_data;
+};
+
+struct amdgpu_pfp {
+	struct amdgpu_bo		*pfp_fw_obj;
+	uint64_t			pfp_fw_gpu_addr;
+	uint32_t			*pfp_fw_ptr;
 
-int amdgpu_gfx_compute_mqd_sw_init(struct amdgpu_device *adev,
-				   unsigned mqd_size);
-void amdgpu_gfx_compute_mqd_sw_fini(struct amdgpu_device *adev);
+	struct amdgpu_bo		*pfp_fw_data_obj;
+	uint64_t			pfp_fw_data_gpu_addr;
+	uint32_t			*pfp_fw_data_ptr;
+};
+
+struct amdgpu_ce {
+	struct amdgpu_bo		*ce_fw_obj;
+	uint64_t			ce_fw_gpu_addr;
+	uint32_t			*ce_fw_ptr;
+};
+
+struct amdgpu_me {
+	struct amdgpu_bo		*me_fw_obj;
+	uint64_t			me_fw_gpu_addr;
+	uint32_t			*me_fw_ptr;
+
+	struct amdgpu_bo		*me_fw_data_obj;
+	uint64_t			me_fw_data_gpu_addr;
+	uint32_t			*me_fw_data_ptr;
+
+	uint32_t			num_me;
+	uint32_t			num_pipe_per_me;
+	uint32_t			num_queue_per_pipe;
+	void				*mqd_backup[AMDGPU_MAX_GFX_RINGS];
+
+	/* These are the resources for which amdgpu takes ownership */
+	DECLARE_BITMAP(queue_bitmap, AMDGPU_MAX_GFX_QUEUES);
+};
+
+struct amdgpu_isolation_work {
+	struct amdgpu_device		*adev;
+	u32				xcp_id;
+	struct delayed_work		work;
+};
+
+struct amdgpu_gfx {
+	struct mutex			gpu_clock_mutex;
+	struct amdgpu_gfx_config	config;
+	struct amdgpu_rlc		rlc;
+	struct amdgpu_pfp		pfp;
+	struct amdgpu_ce		ce;
+	struct amdgpu_me		me;
+	struct amdgpu_mec		mec;
+	struct amdgpu_mec_bitmap	mec_bitmap[AMDGPU_MAX_GC_INSTANCES];
+	struct amdgpu_kiq		kiq[AMDGPU_MAX_GC_INSTANCES];
+	struct amdgpu_imu		imu;
+	bool				rs64_enable; /* firmware format */
+	const struct firmware		*me_fw;	/* ME firmware */
+	uint32_t			me_fw_version;
+	const struct firmware		*pfp_fw; /* PFP firmware */
+	uint32_t			pfp_fw_version;
+	const struct firmware		*ce_fw;	/* CE firmware */
+	uint32_t			ce_fw_version;
+	const struct firmware		*rlc_fw; /* RLC firmware */
+	uint32_t			rlc_fw_version;
+	const struct firmware		*mec_fw; /* MEC firmware */
+	uint32_t			mec_fw_version;
+	const struct firmware		*mec2_fw; /* MEC2 firmware */
+	uint32_t			mec2_fw_version;
+	const struct firmware		*imu_fw; /* IMU firmware */
+	uint32_t			imu_fw_version;
+	uint32_t			me_feature_version;
+	uint32_t			ce_feature_version;
+	uint32_t			pfp_feature_version;
+	uint32_t			rlc_feature_version;
+	uint32_t			rlc_srlc_fw_version;
+	uint32_t			rlc_srlc_feature_version;
+	uint32_t			rlc_srlg_fw_version;
+	uint32_t			rlc_srlg_feature_version;
+	uint32_t			rlc_srls_fw_version;
+	uint32_t			rlc_srls_feature_version;
+	uint32_t			rlcp_ucode_version;
+	uint32_t			rlcp_ucode_feature_version;
+	uint32_t			rlcv_ucode_version;
+	uint32_t			rlcv_ucode_feature_version;
+	uint32_t			mec_feature_version;
+	uint32_t			mec2_feature_version;
+	bool				mec_fw_write_wait;
+	bool				me_fw_write_wait;
+	bool				cp_fw_write_wait;
+	struct amdgpu_ring		gfx_ring[AMDGPU_MAX_GFX_RINGS];
+	unsigned			num_gfx_rings;
+	struct amdgpu_ring		compute_ring[AMDGPU_MAX_COMPUTE_RINGS * AMDGPU_MAX_GC_INSTANCES];
+	unsigned			num_compute_rings;
+	struct amdgpu_irq_src		eop_irq;
+	struct amdgpu_irq_src		priv_reg_irq;
+	struct amdgpu_irq_src		priv_inst_irq;
+	struct amdgpu_irq_src		bad_op_irq;
+	struct amdgpu_irq_src		cp_ecc_error_irq;
+	struct amdgpu_irq_src		sq_irq;
+	struct amdgpu_irq_src		rlc_gc_fed_irq;
+	struct sq_work			sq_work;
+
+	/* gfx status */
+	uint32_t			gfx_current_status;
+	/* ce ram size*/
+	unsigned			ce_ram_size;
+	struct amdgpu_cu_info		cu_info;
+	const struct amdgpu_gfx_funcs	*funcs;
+
+	/* reset mask */
+	uint32_t                        grbm_soft_reset;
+	uint32_t                        srbm_soft_reset;
+	uint32_t 			gfx_supported_reset;
+	uint32_t 			compute_supported_reset;
+
+	/* gfx off */
+	bool                            gfx_off_state;      /* true: enabled, false: disabled */
+	struct mutex                    gfx_off_mutex;      /* mutex to change gfxoff state */
+	uint32_t                        gfx_off_req_count;  /* default 1, enable gfx off: dec 1, disable gfx off: add 1 */
+	struct delayed_work             gfx_off_delay_work; /* async work to set gfx block off */
+	uint32_t                        gfx_off_residency;  /* last logged residency */
+	uint64_t                        gfx_off_entrycount; /* count of times GPU has get into GFXOFF state */
+
+	/* pipe reservation */
+	struct mutex			pipe_reserve_mutex;
+	DECLARE_BITMAP			(pipe_reserve_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
+
+	/*ras */
+	struct ras_common_if		*ras_if;
+	struct amdgpu_gfx_ras		*ras;
+
+	bool				is_poweron;
+
+	struct amdgpu_ring		sw_gfx_ring[AMDGPU_MAX_SW_GFX_RINGS];
+	struct amdgpu_ring_mux          muxer;
+
+	bool				cp_gfx_shadow; /* for gfx11 */
+
+	uint16_t 			xcc_mask;
+	uint32_t			num_xcc_per_xcp;
+	struct mutex			partition_mutex;
+	bool				mcbp; /* mid command buffer preemption */
+
+	/* IP reg dump */
+	uint32_t			*ip_dump_core;
+	uint32_t			*ip_dump_compute_queues;
+	uint32_t			*ip_dump_gfx_queues;
+
+	struct mutex			reset_sem_mutex;
+
+	/* cleaner shader */
+	struct amdgpu_bo		*cleaner_shader_obj;
+	unsigned int                    cleaner_shader_size;
+	u64				cleaner_shader_gpu_addr;
+	void				*cleaner_shader_cpu_ptr;
+	const void			*cleaner_shader_ptr;
+	bool				enable_cleaner_shader;
+	struct amdgpu_isolation_work	enforce_isolation[MAX_XCP];
+	/* Mutex for synchronizing KFD scheduler operations */
+	struct mutex                    userq_sch_mutex;
+	u64				userq_sch_req_count[MAX_XCP];
+	bool				userq_sch_inactive[MAX_XCP];
+	unsigned long			enforce_isolation_jiffies[MAX_XCP];
+	unsigned long			enforce_isolation_time[MAX_XCP];
+
+	atomic_t			total_submission_cnt;
+	struct delayed_work		idle_work;
+	bool				workload_profile_active;
+	struct mutex                    workload_profile_mutex;
+
+	bool				disable_kq;
+	bool				disable_uq;
+};
+
+struct amdgpu_gfx_ras_reg_entry {
+	struct amdgpu_ras_err_status_reg_entry reg_entry;
+	enum amdgpu_gfx_ras_mem_id_type mem_id_type;
+	uint32_t se_num;
+};
+
+struct amdgpu_gfx_ras_mem_id_entry {
+	const struct amdgpu_ras_memory_id_entry *mem_id_ent;
+	uint32_t size;
+};
+
+#define AMDGPU_GFX_MEMID_ENT(x) {(x), ARRAY_SIZE(x)},
+
+#define amdgpu_gfx_get_gpu_clock_counter(adev) (adev)->gfx.funcs->get_gpu_clock_counter((adev))
+#define amdgpu_gfx_select_se_sh(adev, se, sh, instance, xcc_id) ((adev)->gfx.funcs->select_se_sh((adev), (se), (sh), (instance), (xcc_id)))
+#define amdgpu_gfx_select_me_pipe_q(adev, me, pipe, q, vmid, xcc_id) ((adev)->gfx.funcs->select_me_pipe_q((adev), (me), (pipe), (q), (vmid), (xcc_id)))
+#define amdgpu_gfx_init_spm_golden(adev) (adev)->gfx.funcs->init_spm_golden((adev))
+#define amdgpu_gfx_get_gfx_shadow_info(adev, si) ((adev)->gfx.funcs->get_gfx_shadow_info((adev), (si), false))
 
 /**
  * amdgpu_gfx_create_bitmask - create a bitmask
@@ -60,34 +565,106 @@ static inline u32 amdgpu_gfx_create_bitmask(u32 bit_width)
 	return (u32)((1ULL << bit_width) - 1);
 }
 
-static inline int amdgpu_gfx_queue_to_bit(struct amdgpu_device *adev,
-					  int mec, int pipe, int queue)
-{
-	int bit = 0;
+void amdgpu_gfx_parse_disable_cu(unsigned *mask, unsigned max_se,
+				 unsigned max_sh);
 
-	bit += mec * adev->gfx.mec.num_pipe_per_mec
-		* adev->gfx.mec.num_queue_per_pipe;
-	bit += pipe * adev->gfx.mec.num_queue_per_pipe;
-	bit += queue;
+int amdgpu_gfx_kiq_init_ring(struct amdgpu_device *adev, int xcc_id);
 
-	return bit;
-}
+void amdgpu_gfx_kiq_free_ring(struct amdgpu_ring *ring);
 
-static inline void amdgpu_gfx_bit_to_queue(struct amdgpu_device *adev, int bit,
-					   int *mec, int *pipe, int *queue)
-{
-	*queue = bit % adev->gfx.mec.num_queue_per_pipe;
-	*pipe = (bit / adev->gfx.mec.num_queue_per_pipe)
-		% adev->gfx.mec.num_pipe_per_mec;
-	*mec = (bit / adev->gfx.mec.num_queue_per_pipe)
-	       / adev->gfx.mec.num_pipe_per_mec;
+void amdgpu_gfx_kiq_fini(struct amdgpu_device *adev, int xcc_id);
+int amdgpu_gfx_kiq_init(struct amdgpu_device *adev,
+			unsigned hpd_size, int xcc_id);
 
-}
-static inline bool amdgpu_gfx_is_mec_queue_enabled(struct amdgpu_device *adev,
-						   int mec, int pipe, int queue)
+int amdgpu_gfx_mqd_sw_init(struct amdgpu_device *adev,
+			   unsigned mqd_size, int xcc_id);
+void amdgpu_gfx_mqd_sw_fini(struct amdgpu_device *adev, int xcc_id);
+int amdgpu_gfx_disable_kcq(struct amdgpu_device *adev, int xcc_id);
+int amdgpu_gfx_enable_kcq(struct amdgpu_device *adev, int xcc_id);
+int amdgpu_gfx_disable_kgq(struct amdgpu_device *adev, int xcc_id);
+int amdgpu_gfx_enable_kgq(struct amdgpu_device *adev, int xcc_id);
+
+void amdgpu_gfx_compute_queue_acquire(struct amdgpu_device *adev);
+void amdgpu_gfx_graphics_queue_acquire(struct amdgpu_device *adev);
+
+int amdgpu_gfx_mec_queue_to_bit(struct amdgpu_device *adev, int mec,
+				int pipe, int queue);
+void amdgpu_queue_mask_bit_to_mec_queue(struct amdgpu_device *adev, int bit,
+				 int *mec, int *pipe, int *queue);
+bool amdgpu_gfx_is_mec_queue_enabled(struct amdgpu_device *adev, int xcc_id,
+				     int mec, int pipe, int queue);
+bool amdgpu_gfx_is_high_priority_compute_queue(struct amdgpu_device *adev,
+					       struct amdgpu_ring *ring);
+bool amdgpu_gfx_is_high_priority_graphics_queue(struct amdgpu_device *adev,
+						struct amdgpu_ring *ring);
+bool amdgpu_gfx_is_me_queue_enabled(struct amdgpu_device *adev, int me,
+				    int pipe, int queue);
+void amdgpu_gfx_off_ctrl(struct amdgpu_device *adev, bool enable);
+void amdgpu_gfx_off_ctrl_immediate(struct amdgpu_device *adev, bool enable);
+int amdgpu_get_gfx_off_status(struct amdgpu_device *adev, uint32_t *value);
+int amdgpu_gfx_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block);
+void amdgpu_gfx_ras_fini(struct amdgpu_device *adev);
+int amdgpu_get_gfx_off_entrycount(struct amdgpu_device *adev, u64 *value);
+int amdgpu_get_gfx_off_residency(struct amdgpu_device *adev, u32 *residency);
+int amdgpu_set_gfx_off_residency(struct amdgpu_device *adev, bool value);
+int amdgpu_gfx_process_ras_data_cb(struct amdgpu_device *adev,
+		void *err_data,
+		struct amdgpu_iv_entry *entry);
+int amdgpu_gfx_cp_ecc_error_irq(struct amdgpu_device *adev,
+				  struct amdgpu_irq_src *source,
+				  struct amdgpu_iv_entry *entry);
+uint32_t amdgpu_kiq_rreg(struct amdgpu_device *adev, uint32_t reg, uint32_t xcc_id);
+void amdgpu_kiq_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v, uint32_t xcc_id);
+int amdgpu_gfx_get_num_kcq(struct amdgpu_device *adev);
+void amdgpu_gfx_cp_init_microcode(struct amdgpu_device *adev, uint32_t ucode_id);
+
+int amdgpu_gfx_ras_sw_init(struct amdgpu_device *adev);
+int amdgpu_gfx_poison_consumption_handler(struct amdgpu_device *adev,
+						struct amdgpu_iv_entry *entry);
+
+bool amdgpu_gfx_is_master_xcc(struct amdgpu_device *adev, int xcc_id);
+int amdgpu_gfx_sysfs_init(struct amdgpu_device *adev);
+void amdgpu_gfx_sysfs_fini(struct amdgpu_device *adev);
+void amdgpu_gfx_ras_error_func(struct amdgpu_device *adev,
+		void *ras_error_status,
+		void (*func)(struct amdgpu_device *adev, void *ras_error_status,
+				int xcc_id));
+int amdgpu_gfx_cleaner_shader_sw_init(struct amdgpu_device *adev,
+				      unsigned int cleaner_shader_size);
+void amdgpu_gfx_cleaner_shader_sw_fini(struct amdgpu_device *adev);
+void amdgpu_gfx_cleaner_shader_init(struct amdgpu_device *adev,
+				    unsigned int cleaner_shader_size,
+				    const void *cleaner_shader_ptr);
+void amdgpu_gfx_enforce_isolation_handler(struct work_struct *work);
+void amdgpu_gfx_enforce_isolation_ring_begin_use(struct amdgpu_ring *ring);
+void amdgpu_gfx_enforce_isolation_ring_end_use(struct amdgpu_ring *ring);
+
+void amdgpu_gfx_profile_idle_work_handler(struct work_struct *work);
+void amdgpu_gfx_profile_ring_begin_use(struct amdgpu_ring *ring);
+void amdgpu_gfx_profile_ring_end_use(struct amdgpu_ring *ring);
+u32 amdgpu_gfx_csb_preamble_start(u32 *buffer);
+u32 amdgpu_gfx_csb_data_parser(struct amdgpu_device *adev, u32 *buffer, u32 count);
+void amdgpu_gfx_csb_preamble_end(u32 *buffer, u32 count);
+
+void amdgpu_debugfs_gfx_sched_mask_init(struct amdgpu_device *adev);
+void amdgpu_debugfs_compute_sched_mask_init(struct amdgpu_device *adev);
+
+static inline const char *amdgpu_gfx_compute_mode_desc(int mode)
 {
-	return test_bit(amdgpu_gfx_queue_to_bit(adev, mec, pipe, queue),
-			adev->gfx.mec.queue_bitmap);
+	switch (mode) {
+	case AMDGPU_SPX_PARTITION_MODE:
+		return "SPX";
+	case AMDGPU_DPX_PARTITION_MODE:
+		return "DPX";
+	case AMDGPU_TPX_PARTITION_MODE:
+		return "TPX";
+	case AMDGPU_QPX_PARTITION_MODE:
+		return "QPX";
+	case AMDGPU_CPX_PARTITION_MODE:
+		return "CPX";
+	default:
+		return "UNKNOWN";
+	}
 }
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfxhub.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfxhub.h
new file mode 100644
index 000000000000..c7b44aeb671b
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfxhub.h
@@ -0,0 +1,47 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __AMDGPU_GFXHUB_H__
+#define __AMDGPU_GFXHUB_H__
+
+struct amdgpu_gfxhub_funcs {
+	u64 (*get_fb_location)(struct amdgpu_device *adev);
+	u64 (*get_mc_fb_offset)(struct amdgpu_device *adev);
+	void (*setup_vm_pt_regs)(struct amdgpu_device *adev, uint32_t vmid,
+			uint64_t page_table_base);
+	int (*gart_enable)(struct amdgpu_device *adev);
+
+	void (*gart_disable)(struct amdgpu_device *adev);
+	void (*set_fault_enable_default)(struct amdgpu_device *adev, bool value);
+	void (*init)(struct amdgpu_device *adev);
+	int (*get_xgmi_info)(struct amdgpu_device *adev);
+	void (*utcl2_harvest)(struct amdgpu_device *adev);
+	void (*mode2_save_regs)(struct amdgpu_device *adev);
+	void (*mode2_restore_regs)(struct amdgpu_device *adev);
+	void (*halt)(struct amdgpu_device *adev);
+};
+
+struct amdgpu_gfxhub {
+	const struct amdgpu_gfxhub_funcs *funcs;
+};
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
new file mode 100644
index 000000000000..9dcf51991b5b
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
@@ -0,0 +1,1679 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+
+#include <linux/io-64-nonatomic-lo-hi.h>
+#ifdef CONFIG_X86
+#include <asm/hypervisor.h>
+#endif
+
+#include "amdgpu.h"
+#include "amdgpu_gmc.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_reset.h"
+#include "amdgpu_xgmi.h"
+
+#include <drm/drm_drv.h>
+#include <drm/ttm/ttm_tt.h>
+
+static const u64 four_gb = 0x100000000ULL;
+
+bool amdgpu_gmc_is_pdb0_enabled(struct amdgpu_device *adev)
+{
+	return adev->gmc.xgmi.connected_to_cpu || amdgpu_virt_xgmi_migrate_enabled(adev);
+}
+
+/**
+ * amdgpu_gmc_pdb0_alloc - allocate vram for pdb0
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate video memory for pdb0 and map it for CPU access
+ * Returns 0 for success, error for failure.
+ */
+int amdgpu_gmc_pdb0_alloc(struct amdgpu_device *adev)
+{
+	int r;
+	struct amdgpu_bo_param bp;
+	u64 vram_size = adev->gmc.xgmi.node_segment_size * adev->gmc.xgmi.num_physical_nodes;
+	uint32_t pde0_page_shift = adev->gmc.vmid0_page_table_block_size + 21;
+	uint32_t npdes = (vram_size + (1ULL << pde0_page_shift) - 1) >> pde0_page_shift;
+
+	memset(&bp, 0, sizeof(bp));
+	bp.size = PAGE_ALIGN((npdes + 1) * 8);
+	bp.byte_align = PAGE_SIZE;
+	bp.domain = AMDGPU_GEM_DOMAIN_VRAM;
+	bp.flags = AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
+		AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+	bp.type = ttm_bo_type_kernel;
+	bp.resv = NULL;
+	bp.bo_ptr_size = sizeof(struct amdgpu_bo);
+
+	r = amdgpu_bo_create(adev, &bp, &adev->gmc.pdb0_bo);
+	if (r)
+		return r;
+
+	r = amdgpu_bo_reserve(adev->gmc.pdb0_bo, false);
+	if (unlikely(r != 0))
+		goto bo_reserve_failure;
+
+	r = amdgpu_bo_pin(adev->gmc.pdb0_bo, AMDGPU_GEM_DOMAIN_VRAM);
+	if (r)
+		goto bo_pin_failure;
+	r = amdgpu_bo_kmap(adev->gmc.pdb0_bo, &adev->gmc.ptr_pdb0);
+	if (r)
+		goto bo_kmap_failure;
+
+	amdgpu_bo_unreserve(adev->gmc.pdb0_bo);
+	return 0;
+
+bo_kmap_failure:
+	amdgpu_bo_unpin(adev->gmc.pdb0_bo);
+bo_pin_failure:
+	amdgpu_bo_unreserve(adev->gmc.pdb0_bo);
+bo_reserve_failure:
+	amdgpu_bo_unref(&adev->gmc.pdb0_bo);
+	return r;
+}
+
+/**
+ * amdgpu_gmc_get_pde_for_bo - get the PDE for a BO
+ *
+ * @bo: the BO to get the PDE for
+ * @level: the level in the PD hirarchy
+ * @addr: resulting addr
+ * @flags: resulting flags
+ *
+ * Get the address and flags to be used for a PDE (Page Directory Entry).
+ */
+void amdgpu_gmc_get_pde_for_bo(struct amdgpu_bo *bo, int level,
+			       uint64_t *addr, uint64_t *flags)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+
+	switch (bo->tbo.resource->mem_type) {
+	case TTM_PL_TT:
+		*addr = bo->tbo.ttm->dma_address[0];
+		break;
+	case TTM_PL_VRAM:
+		*addr = amdgpu_bo_gpu_offset(bo);
+		break;
+	default:
+		*addr = 0;
+		break;
+	}
+	*flags = amdgpu_ttm_tt_pde_flags(bo->tbo.ttm, bo->tbo.resource);
+	amdgpu_gmc_get_vm_pde(adev, level, addr, flags);
+}
+
+/*
+ * amdgpu_gmc_pd_addr - return the address of the root directory
+ */
+uint64_t amdgpu_gmc_pd_addr(struct amdgpu_bo *bo)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	uint64_t pd_addr;
+
+	/* TODO: move that into ASIC specific code */
+	if (adev->asic_type >= CHIP_VEGA10) {
+		uint64_t flags = AMDGPU_PTE_VALID;
+
+		amdgpu_gmc_get_pde_for_bo(bo, -1, &pd_addr, &flags);
+		pd_addr |= flags;
+	} else {
+		pd_addr = amdgpu_bo_gpu_offset(bo);
+	}
+	return pd_addr;
+}
+
+/**
+ * amdgpu_gmc_set_pte_pde - update the page tables using CPU
+ *
+ * @adev: amdgpu_device pointer
+ * @cpu_pt_addr: cpu address of the page table
+ * @gpu_page_idx: entry in the page table to update
+ * @addr: dst addr to write into pte/pde
+ * @flags: access flags
+ *
+ * Update the page tables using CPU.
+ */
+int amdgpu_gmc_set_pte_pde(struct amdgpu_device *adev, void *cpu_pt_addr,
+				uint32_t gpu_page_idx, uint64_t addr,
+				uint64_t flags)
+{
+	void __iomem *ptr = (void *)cpu_pt_addr;
+	uint64_t value;
+
+	/*
+	 * The following is for PTE only. GART does not have PDEs.
+	*/
+	value = addr & 0x0000FFFFFFFFF000ULL;
+	value |= flags;
+	writeq(value, ptr + (gpu_page_idx * 8));
+
+	return 0;
+}
+
+/**
+ * amdgpu_gmc_agp_addr - return the address in the AGP address space
+ *
+ * @bo: TTM BO which needs the address, must be in GTT domain
+ *
+ * Tries to figure out how to access the BO through the AGP aperture. Returns
+ * AMDGPU_BO_INVALID_OFFSET if that is not possible.
+ */
+uint64_t amdgpu_gmc_agp_addr(struct ttm_buffer_object *bo)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev);
+
+	if (!bo->ttm)
+		return AMDGPU_BO_INVALID_OFFSET;
+
+	if (bo->ttm->num_pages != 1 || bo->ttm->caching == ttm_cached)
+		return AMDGPU_BO_INVALID_OFFSET;
+
+	if (bo->ttm->dma_address[0] + PAGE_SIZE >= adev->gmc.agp_size)
+		return AMDGPU_BO_INVALID_OFFSET;
+
+	return adev->gmc.agp_start + bo->ttm->dma_address[0];
+}
+
+/**
+ * amdgpu_gmc_vram_location - try to find VRAM location
+ *
+ * @adev: amdgpu device structure holding all necessary information
+ * @mc: memory controller structure holding memory information
+ * @base: base address at which to put VRAM
+ *
+ * Function will try to place VRAM at base address provided
+ * as parameter.
+ */
+void amdgpu_gmc_vram_location(struct amdgpu_device *adev, struct amdgpu_gmc *mc,
+			      u64 base)
+{
+	uint64_t vis_limit = (uint64_t)amdgpu_vis_vram_limit << 20;
+	uint64_t limit = (uint64_t)amdgpu_vram_limit << 20;
+
+	mc->vram_start = base;
+	mc->vram_end = mc->vram_start + mc->mc_vram_size - 1;
+	if (limit < mc->real_vram_size)
+		mc->real_vram_size = limit;
+
+	if (vis_limit && vis_limit < mc->visible_vram_size)
+		mc->visible_vram_size = vis_limit;
+
+	if (mc->real_vram_size < mc->visible_vram_size)
+		mc->visible_vram_size = mc->real_vram_size;
+
+	if (mc->xgmi.num_physical_nodes == 0) {
+		mc->fb_start = mc->vram_start;
+		mc->fb_end = mc->vram_end;
+	}
+	dev_info(adev->dev, "VRAM: %lluM 0x%016llX - 0x%016llX (%lluM used)\n",
+			mc->mc_vram_size >> 20, mc->vram_start,
+			mc->vram_end, mc->real_vram_size >> 20);
+}
+
+/** amdgpu_gmc_sysvm_location - place vram and gart in sysvm aperture
+ *
+ * @adev: amdgpu device structure holding all necessary information
+ * @mc: memory controller structure holding memory information
+ *
+ * This function is only used if use GART for FB translation. In such
+ * case, we use sysvm aperture (vmid0 page tables) for both vram
+ * and gart (aka system memory) access.
+ *
+ * GPUVM (and our organization of vmid0 page tables) require sysvm
+ * aperture to be placed at a location aligned with 8 times of native
+ * page size. For example, if vm_context0_cntl.page_table_block_size
+ * is 12, then native page size is 8G (2M*2^12), sysvm should start
+ * with a 64G aligned address. For simplicity, we just put sysvm at
+ * address 0. So vram start at address 0 and gart is right after vram.
+ */
+void amdgpu_gmc_sysvm_location(struct amdgpu_device *adev, struct amdgpu_gmc *mc)
+{
+	u64 hive_vram_start = 0;
+	u64 hive_vram_end = mc->xgmi.node_segment_size * mc->xgmi.num_physical_nodes - 1;
+	mc->vram_start = mc->xgmi.node_segment_size * mc->xgmi.physical_node_id;
+	mc->vram_end = mc->vram_start + mc->xgmi.node_segment_size - 1;
+	/* node_segment_size may not 4GB aligned on SRIOV, align up is needed. */
+	mc->gart_start = ALIGN(hive_vram_end + 1, four_gb);
+	mc->gart_end = mc->gart_start + mc->gart_size - 1;
+	if (amdgpu_virt_xgmi_migrate_enabled(adev)) {
+		/* set mc->vram_start to 0 to switch the returned GPU address of
+		 * amdgpu_bo_create_reserved() from FB aperture to GART aperture.
+		 */
+		mc->vram_start = 0;
+		mc->vram_end = mc->vram_start + mc->mc_vram_size - 1;
+		mc->visible_vram_size = min(mc->visible_vram_size, mc->real_vram_size);
+	} else {
+		mc->fb_start = hive_vram_start;
+		mc->fb_end = hive_vram_end;
+	}
+	dev_info(adev->dev, "VRAM: %lluM 0x%016llX - 0x%016llX (%lluM used)\n",
+			mc->mc_vram_size >> 20, mc->vram_start,
+			mc->vram_end, mc->real_vram_size >> 20);
+	dev_info(adev->dev, "GART: %lluM 0x%016llX - 0x%016llX\n",
+			mc->gart_size >> 20, mc->gart_start, mc->gart_end);
+}
+
+/**
+ * amdgpu_gmc_gart_location - try to find GART location
+ *
+ * @adev: amdgpu device structure holding all necessary information
+ * @mc: memory controller structure holding memory information
+ * @gart_placement: GART placement policy with respect to VRAM
+ *
+ * Function will try to place GART before or after VRAM.
+ * If GART size is bigger than space left then we ajust GART size.
+ * Thus function will never fails.
+ */
+void amdgpu_gmc_gart_location(struct amdgpu_device *adev, struct amdgpu_gmc *mc,
+			      enum amdgpu_gart_placement gart_placement)
+{
+	u64 size_af, size_bf;
+	/*To avoid the hole, limit the max mc address to AMDGPU_GMC_HOLE_START*/
+	u64 max_mc_address = min(adev->gmc.mc_mask, AMDGPU_GMC_HOLE_START - 1);
+
+	/* VCE doesn't like it when BOs cross a 4GB segment, so align
+	 * the GART base on a 4GB boundary as well.
+	 */
+	size_bf = mc->fb_start;
+	size_af = max_mc_address + 1 - ALIGN(mc->fb_end + 1, four_gb);
+
+	if (mc->gart_size > max(size_bf, size_af)) {
+		dev_warn(adev->dev, "limiting GART\n");
+		mc->gart_size = max(size_bf, size_af);
+	}
+
+	switch (gart_placement) {
+	case AMDGPU_GART_PLACEMENT_HIGH:
+		mc->gart_start = max_mc_address - mc->gart_size + 1;
+		break;
+	case AMDGPU_GART_PLACEMENT_LOW:
+		mc->gart_start = 0;
+		break;
+	case AMDGPU_GART_PLACEMENT_BEST_FIT:
+	default:
+		if ((size_bf >= mc->gart_size && size_bf < size_af) ||
+		    (size_af < mc->gart_size))
+			mc->gart_start = 0;
+		else
+			mc->gart_start = max_mc_address - mc->gart_size + 1;
+		break;
+	}
+
+	mc->gart_start &= ~(four_gb - 1);
+	mc->gart_end = mc->gart_start + mc->gart_size - 1;
+	dev_info(adev->dev, "GART: %lluM 0x%016llX - 0x%016llX\n",
+			mc->gart_size >> 20, mc->gart_start, mc->gart_end);
+}
+
+/**
+ * amdgpu_gmc_agp_location - try to find AGP location
+ * @adev: amdgpu device structure holding all necessary information
+ * @mc: memory controller structure holding memory information
+ *
+ * Function will place try to find a place for the AGP BAR in the MC address
+ * space.
+ *
+ * AGP BAR will be assigned the largest available hole in the address space.
+ * Should be called after VRAM and GART locations are setup.
+ */
+void amdgpu_gmc_agp_location(struct amdgpu_device *adev, struct amdgpu_gmc *mc)
+{
+	const uint64_t sixteen_gb = 1ULL << 34;
+	const uint64_t sixteen_gb_mask = ~(sixteen_gb - 1);
+	u64 size_af, size_bf;
+
+	if (mc->fb_start > mc->gart_start) {
+		size_bf = (mc->fb_start & sixteen_gb_mask) -
+			ALIGN(mc->gart_end + 1, sixteen_gb);
+		size_af = mc->mc_mask + 1 - ALIGN(mc->fb_end + 1, sixteen_gb);
+	} else {
+		size_bf = mc->fb_start & sixteen_gb_mask;
+		size_af = (mc->gart_start & sixteen_gb_mask) -
+			ALIGN(mc->fb_end + 1, sixteen_gb);
+	}
+
+	if (size_bf > size_af) {
+		mc->agp_start = (mc->fb_start - size_bf) & sixteen_gb_mask;
+		mc->agp_size = size_bf;
+	} else {
+		mc->agp_start = ALIGN(mc->fb_end + 1, sixteen_gb);
+		mc->agp_size = size_af;
+	}
+
+	mc->agp_end = mc->agp_start + mc->agp_size - 1;
+	dev_info(adev->dev, "AGP: %lluM 0x%016llX - 0x%016llX\n",
+			mc->agp_size >> 20, mc->agp_start, mc->agp_end);
+}
+
+/**
+ * amdgpu_gmc_set_agp_default - Set the default AGP aperture value.
+ * @adev: amdgpu device structure holding all necessary information
+ * @mc: memory controller structure holding memory information
+ *
+ * To disable the AGP aperture, you need to set the start to a larger
+ * value than the end.  This function sets the default value which
+ * can then be overridden using amdgpu_gmc_agp_location() if you want
+ * to enable the AGP aperture on a specific chip.
+ *
+ */
+void amdgpu_gmc_set_agp_default(struct amdgpu_device *adev,
+				struct amdgpu_gmc *mc)
+{
+	mc->agp_start = 0xffffffffffff;
+	mc->agp_end = 0;
+	mc->agp_size = 0;
+}
+
+/**
+ * amdgpu_gmc_fault_key - get hask key from vm fault address and pasid
+ *
+ * @addr: 48 bit physical address, page aligned (36 significant bits)
+ * @pasid: 16 bit process address space identifier
+ */
+static inline uint64_t amdgpu_gmc_fault_key(uint64_t addr, uint16_t pasid)
+{
+	return addr << 4 | pasid;
+}
+
+/**
+ * amdgpu_gmc_filter_faults - filter VM faults
+ *
+ * @adev: amdgpu device structure
+ * @ih: interrupt ring that the fault received from
+ * @addr: address of the VM fault
+ * @pasid: PASID of the process causing the fault
+ * @timestamp: timestamp of the fault
+ *
+ * Returns:
+ * True if the fault was filtered and should not be processed further.
+ * False if the fault is a new one and needs to be handled.
+ */
+bool amdgpu_gmc_filter_faults(struct amdgpu_device *adev,
+			      struct amdgpu_ih_ring *ih, uint64_t addr,
+			      uint16_t pasid, uint64_t timestamp)
+{
+	struct amdgpu_gmc *gmc = &adev->gmc;
+	uint64_t stamp, key = amdgpu_gmc_fault_key(addr, pasid);
+	struct amdgpu_gmc_fault *fault;
+	uint32_t hash;
+
+	/* Stale retry fault if timestamp goes backward */
+	if (amdgpu_ih_ts_after(timestamp, ih->processed_timestamp))
+		return true;
+
+	/* If we don't have space left in the ring buffer return immediately */
+	stamp = max(timestamp, AMDGPU_GMC_FAULT_TIMEOUT + 1) -
+		AMDGPU_GMC_FAULT_TIMEOUT;
+	if (gmc->fault_ring[gmc->last_fault].timestamp >= stamp)
+		return true;
+
+	/* Try to find the fault in the hash */
+	hash = hash_64(key, AMDGPU_GMC_FAULT_HASH_ORDER);
+	fault = &gmc->fault_ring[gmc->fault_hash[hash].idx];
+	while (fault->timestamp >= stamp) {
+		uint64_t tmp;
+
+		if (atomic64_read(&fault->key) == key) {
+			/*
+			 * if we get a fault which is already present in
+			 * the fault_ring and the timestamp of
+			 * the fault is after the expired timestamp,
+			 * then this is a new fault that needs to be added
+			 * into the fault ring.
+			 */
+			if (fault->timestamp_expiry != 0 &&
+			    amdgpu_ih_ts_after(fault->timestamp_expiry,
+					       timestamp))
+				break;
+			else
+				return true;
+		}
+
+		tmp = fault->timestamp;
+		fault = &gmc->fault_ring[fault->next];
+
+		/* Check if the entry was reused */
+		if (fault->timestamp >= tmp)
+			break;
+	}
+
+	/* Add the fault to the ring */
+	fault = &gmc->fault_ring[gmc->last_fault];
+	atomic64_set(&fault->key, key);
+	fault->timestamp = timestamp;
+
+	/* And update the hash */
+	fault->next = gmc->fault_hash[hash].idx;
+	gmc->fault_hash[hash].idx = gmc->last_fault++;
+	return false;
+}
+
+/**
+ * amdgpu_gmc_filter_faults_remove - remove address from VM faults filter
+ *
+ * @adev: amdgpu device structure
+ * @addr: address of the VM fault
+ * @pasid: PASID of the process causing the fault
+ *
+ * Remove the address from fault filter, then future vm fault on this address
+ * will pass to retry fault handler to recover.
+ */
+void amdgpu_gmc_filter_faults_remove(struct amdgpu_device *adev, uint64_t addr,
+				     uint16_t pasid)
+{
+	struct amdgpu_gmc *gmc = &adev->gmc;
+	uint64_t key = amdgpu_gmc_fault_key(addr, pasid);
+	struct amdgpu_ih_ring *ih;
+	struct amdgpu_gmc_fault *fault;
+	uint32_t last_wptr;
+	uint64_t last_ts;
+	uint32_t hash;
+	uint64_t tmp;
+
+	if (adev->irq.retry_cam_enabled)
+		return;
+
+	ih = &adev->irq.ih1;
+	/* Get the WPTR of the last entry in IH ring */
+	last_wptr = amdgpu_ih_get_wptr(adev, ih);
+	/* Order wptr with ring data. */
+	rmb();
+	/* Get the timetamp of the last entry in IH ring */
+	last_ts = amdgpu_ih_decode_iv_ts(adev, ih, last_wptr, -1);
+
+	hash = hash_64(key, AMDGPU_GMC_FAULT_HASH_ORDER);
+	fault = &gmc->fault_ring[gmc->fault_hash[hash].idx];
+	do {
+		if (atomic64_read(&fault->key) == key) {
+			/*
+			 * Update the timestamp when this fault
+			 * expired.
+			 */
+			fault->timestamp_expiry = last_ts;
+			break;
+		}
+
+		tmp = fault->timestamp;
+		fault = &gmc->fault_ring[fault->next];
+	} while (fault->timestamp < tmp);
+}
+
+int amdgpu_gmc_ras_sw_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	/* umc ras block */
+	r = amdgpu_umc_ras_sw_init(adev);
+	if (r)
+		return r;
+
+	/* mmhub ras block */
+	r = amdgpu_mmhub_ras_sw_init(adev);
+	if (r)
+		return r;
+
+	/* hdp ras block */
+	r = amdgpu_hdp_ras_sw_init(adev);
+	if (r)
+		return r;
+
+	/* mca.x ras block */
+	r = amdgpu_mca_mp0_ras_sw_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_mca_mp1_ras_sw_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_mca_mpio_ras_sw_init(adev);
+	if (r)
+		return r;
+
+	/* xgmi ras block */
+	r = amdgpu_xgmi_ras_sw_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+int amdgpu_gmc_ras_late_init(struct amdgpu_device *adev)
+{
+	return 0;
+}
+
+void amdgpu_gmc_ras_fini(struct amdgpu_device *adev)
+{
+
+}
+
+	/*
+	 * The latest engine allocation on gfx9/10 is:
+	 * Engine 2, 3: firmware
+	 * Engine 0, 1, 4~16: amdgpu ring,
+	 *                    subject to change when ring number changes
+	 * Engine 17: Gart flushes
+	 */
+#define AMDGPU_VMHUB_INV_ENG_BITMAP		0x1FFF3
+
+int amdgpu_gmc_allocate_vm_inv_eng(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	unsigned vm_inv_engs[AMDGPU_MAX_VMHUBS] = {0};
+	unsigned i;
+	unsigned vmhub, inv_eng;
+	struct amdgpu_ring *shared_ring;
+
+	/* init the vm inv eng for all vmhubs */
+	for_each_set_bit(i, adev->vmhubs_mask, AMDGPU_MAX_VMHUBS) {
+		vm_inv_engs[i] = AMDGPU_VMHUB_INV_ENG_BITMAP;
+		/* reserve engine 5 for firmware */
+		if (adev->enable_mes)
+			vm_inv_engs[i] &= ~(1 << 5);
+		/* reserve mmhub engine 3 for firmware */
+		if (adev->enable_umsch_mm)
+			vm_inv_engs[i] &= ~(1 << 3);
+	}
+
+	for (i = 0; i < adev->num_rings; ++i) {
+		ring = adev->rings[i];
+		vmhub = ring->vm_hub;
+
+		if (ring == &adev->mes.ring[0] ||
+		    ring == &adev->mes.ring[1] ||
+		    ring == &adev->umsch_mm.ring ||
+		    ring == &adev->cper.ring_buf)
+			continue;
+
+		/* Skip if the ring is a shared ring */
+		if (amdgpu_sdma_is_shared_inv_eng(adev, ring))
+			continue;
+
+		inv_eng = ffs(vm_inv_engs[vmhub]);
+		if (!inv_eng) {
+			dev_err(adev->dev, "no VM inv eng for ring %s\n",
+				ring->name);
+			return -EINVAL;
+		}
+
+		ring->vm_inv_eng = inv_eng - 1;
+		vm_inv_engs[vmhub] &= ~(1 << ring->vm_inv_eng);
+
+		dev_info(adev->dev, "ring %s uses VM inv eng %u on hub %u\n",
+			 ring->name, ring->vm_inv_eng, ring->vm_hub);
+		/* SDMA has a special packet which allows it to use the same
+		 * invalidation engine for all the rings in one instance.
+		 * Therefore, we do not allocate a separate VM invalidation engine
+		 * for SDMA page rings. Instead, they share the VM invalidation
+		 * engine with the SDMA gfx ring. This change ensures efficient
+		 * resource management and avoids the issue of insufficient VM
+		 * invalidation engines.
+		 */
+		shared_ring = amdgpu_sdma_get_shared_ring(adev, ring);
+		if (shared_ring) {
+			shared_ring->vm_inv_eng = ring->vm_inv_eng;
+			dev_info(adev->dev, "ring %s shares VM invalidation engine %u with ring %s on hub %u\n",
+					ring->name, ring->vm_inv_eng, shared_ring->name, ring->vm_hub);
+			continue;
+		}
+	}
+
+	return 0;
+}
+
+void amdgpu_gmc_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid,
+			      uint32_t vmhub, uint32_t flush_type)
+{
+	struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
+	struct amdgpu_vmhub *hub = &adev->vmhub[vmhub];
+	struct dma_fence *fence;
+	struct amdgpu_job *job;
+	int r;
+
+	if (!hub->sdma_invalidation_workaround || vmid ||
+	    !adev->mman.buffer_funcs_enabled || !adev->ib_pool_ready ||
+	    !ring->sched.ready) {
+		/*
+		 * A GPU reset should flush all TLBs anyway, so no need to do
+		 * this while one is ongoing.
+		 */
+		if (!down_read_trylock(&adev->reset_domain->sem))
+			return;
+
+		if (adev->gmc.flush_tlb_needs_extra_type_2)
+			adev->gmc.gmc_funcs->flush_gpu_tlb(adev, vmid,
+							   vmhub, 2);
+
+		if (adev->gmc.flush_tlb_needs_extra_type_0 && flush_type == 2)
+			adev->gmc.gmc_funcs->flush_gpu_tlb(adev, vmid,
+							   vmhub, 0);
+
+		adev->gmc.gmc_funcs->flush_gpu_tlb(adev, vmid, vmhub,
+						   flush_type);
+		up_read(&adev->reset_domain->sem);
+		return;
+	}
+
+	/* The SDMA on Navi 1x has a bug which can theoretically result in memory
+	 * corruption if an invalidation happens at the same time as an VA
+	 * translation. Avoid this by doing the invalidation from the SDMA
+	 * itself at least for GART.
+	 */
+	mutex_lock(&adev->mman.gtt_window_lock);
+	r = amdgpu_job_alloc_with_ib(ring->adev, &adev->mman.high_pr,
+				     AMDGPU_FENCE_OWNER_UNDEFINED,
+				     16 * 4, AMDGPU_IB_POOL_IMMEDIATE,
+				     &job, AMDGPU_KERNEL_JOB_ID_FLUSH_GPU_TLB);
+	if (r)
+		goto error_alloc;
+
+	job->vm_pd_addr = amdgpu_gmc_pd_addr(adev->gart.bo);
+	job->vm_needs_flush = true;
+	job->ibs->ptr[job->ibs->length_dw++] = ring->funcs->nop;
+	amdgpu_ring_pad_ib(ring, &job->ibs[0]);
+	fence = amdgpu_job_submit(job);
+	mutex_unlock(&adev->mman.gtt_window_lock);
+
+	dma_fence_wait(fence, false);
+	dma_fence_put(fence);
+
+	return;
+
+error_alloc:
+	mutex_unlock(&adev->mman.gtt_window_lock);
+	dev_err(adev->dev, "Error flushing GPU TLB using the SDMA (%d)!\n", r);
+}
+
+int amdgpu_gmc_flush_gpu_tlb_pasid(struct amdgpu_device *adev, uint16_t pasid,
+				   uint32_t flush_type, bool all_hub,
+				   uint32_t inst)
+{
+	struct amdgpu_ring *ring = &adev->gfx.kiq[inst].ring;
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[inst];
+	unsigned int ndw;
+	int r, cnt = 0;
+	uint32_t seq;
+
+	/*
+	 * A GPU reset should flush all TLBs anyway, so no need to do
+	 * this while one is ongoing.
+	 */
+	if (!down_read_trylock(&adev->reset_domain->sem))
+		return 0;
+
+	if (!adev->gmc.flush_pasid_uses_kiq || !ring->sched.ready) {
+		if (adev->gmc.flush_tlb_needs_extra_type_2)
+			adev->gmc.gmc_funcs->flush_gpu_tlb_pasid(adev, pasid,
+								 2, all_hub,
+								 inst);
+
+		if (adev->gmc.flush_tlb_needs_extra_type_0 && flush_type == 2)
+			adev->gmc.gmc_funcs->flush_gpu_tlb_pasid(adev, pasid,
+								 0, all_hub,
+								 inst);
+
+		adev->gmc.gmc_funcs->flush_gpu_tlb_pasid(adev, pasid,
+							 flush_type, all_hub,
+							 inst);
+		r = 0;
+	} else {
+		/* 2 dwords flush + 8 dwords fence */
+		ndw = kiq->pmf->invalidate_tlbs_size + 8;
+
+		if (adev->gmc.flush_tlb_needs_extra_type_2)
+			ndw += kiq->pmf->invalidate_tlbs_size;
+
+		if (adev->gmc.flush_tlb_needs_extra_type_0)
+			ndw += kiq->pmf->invalidate_tlbs_size;
+
+		spin_lock(&adev->gfx.kiq[inst].ring_lock);
+		r = amdgpu_ring_alloc(ring, ndw);
+		if (r) {
+			spin_unlock(&adev->gfx.kiq[inst].ring_lock);
+			goto error_unlock_reset;
+		}
+		if (adev->gmc.flush_tlb_needs_extra_type_2)
+			kiq->pmf->kiq_invalidate_tlbs(ring, pasid, 2, all_hub);
+
+		if (flush_type == 2 && adev->gmc.flush_tlb_needs_extra_type_0)
+			kiq->pmf->kiq_invalidate_tlbs(ring, pasid, 0, all_hub);
+
+		kiq->pmf->kiq_invalidate_tlbs(ring, pasid, flush_type, all_hub);
+		r = amdgpu_fence_emit_polling(ring, &seq, MAX_KIQ_REG_WAIT);
+		if (r) {
+			amdgpu_ring_undo(ring);
+			spin_unlock(&adev->gfx.kiq[inst].ring_lock);
+			goto error_unlock_reset;
+		}
+
+		amdgpu_ring_commit(ring);
+		spin_unlock(&adev->gfx.kiq[inst].ring_lock);
+
+		r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
+
+		might_sleep();
+		while (r < 1 && cnt++ < MAX_KIQ_REG_TRY &&
+		       !amdgpu_reset_pending(adev->reset_domain)) {
+			msleep(MAX_KIQ_REG_BAILOUT_INTERVAL);
+			r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
+		}
+
+		if (cnt > MAX_KIQ_REG_TRY) {
+			dev_err(adev->dev, "timeout waiting for kiq fence\n");
+			r = -ETIME;
+		} else
+			r = 0;
+	}
+
+error_unlock_reset:
+	up_read(&adev->reset_domain->sem);
+	return r;
+}
+
+void amdgpu_gmc_fw_reg_write_reg_wait(struct amdgpu_device *adev,
+				      uint32_t reg0, uint32_t reg1,
+				      uint32_t ref, uint32_t mask,
+				      uint32_t xcc_inst)
+{
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[xcc_inst];
+	struct amdgpu_ring *ring = &kiq->ring;
+	signed long r, cnt = 0;
+	unsigned long flags;
+	uint32_t seq;
+
+	if (adev->mes.ring[0].sched.ready) {
+		amdgpu_mes_reg_write_reg_wait(adev, reg0, reg1,
+					      ref, mask);
+		return;
+	}
+
+	spin_lock_irqsave(&kiq->ring_lock, flags);
+	amdgpu_ring_alloc(ring, 32);
+	amdgpu_ring_emit_reg_write_reg_wait(ring, reg0, reg1,
+					    ref, mask);
+	r = amdgpu_fence_emit_polling(ring, &seq, MAX_KIQ_REG_WAIT);
+	if (r)
+		goto failed_undo;
+
+	amdgpu_ring_commit(ring);
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+
+	r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
+
+	/* don't wait anymore for IRQ context */
+	if (r < 1 && in_interrupt())
+		goto failed_kiq;
+
+	might_sleep();
+	while (r < 1 && cnt++ < MAX_KIQ_REG_TRY &&
+	       !amdgpu_reset_pending(adev->reset_domain)) {
+
+		msleep(MAX_KIQ_REG_BAILOUT_INTERVAL);
+		r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
+	}
+
+	if (cnt > MAX_KIQ_REG_TRY)
+		goto failed_kiq;
+
+	return;
+
+failed_undo:
+	amdgpu_ring_undo(ring);
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+failed_kiq:
+	dev_err(adev->dev, "failed to write reg %x wait reg %x\n", reg0, reg1);
+}
+
+/**
+ * amdgpu_gmc_tmz_set -- check and set if a device supports TMZ
+ * @adev: amdgpu_device pointer
+ *
+ * Check and set if an the device @adev supports Trusted Memory
+ * Zones (TMZ).
+ */
+void amdgpu_gmc_tmz_set(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	/* RAVEN */
+	case IP_VERSION(9, 2, 2):
+	case IP_VERSION(9, 1, 0):
+	/* RENOIR looks like RAVEN */
+	case IP_VERSION(9, 3, 0):
+	/* GC 10.3.7 */
+	case IP_VERSION(10, 3, 7):
+	/* GC 11.0.1 */
+	case IP_VERSION(11, 0, 1):
+		if (amdgpu_tmz == 0) {
+			adev->gmc.tmz_enabled = false;
+			dev_info(adev->dev,
+				 "Trusted Memory Zone (TMZ) feature disabled (cmd line)\n");
+		} else {
+			adev->gmc.tmz_enabled = true;
+			dev_info(adev->dev,
+				 "Trusted Memory Zone (TMZ) feature enabled\n");
+		}
+		break;
+	case IP_VERSION(10, 1, 10):
+	case IP_VERSION(10, 1, 1):
+	case IP_VERSION(10, 1, 2):
+	case IP_VERSION(10, 1, 3):
+	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+	case IP_VERSION(10, 3, 6):
+	/* VANGOGH */
+	case IP_VERSION(10, 3, 1):
+	/* YELLOW_CARP*/
+	case IP_VERSION(10, 3, 3):
+	case IP_VERSION(11, 0, 4):
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 1):
+	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
+		/* Don't enable it by default yet.
+		 */
+		if (amdgpu_tmz < 1) {
+			adev->gmc.tmz_enabled = false;
+			dev_info(adev->dev,
+				 "Trusted Memory Zone (TMZ) feature disabled as experimental (default)\n");
+		} else {
+			adev->gmc.tmz_enabled = true;
+			dev_info(adev->dev,
+				 "Trusted Memory Zone (TMZ) feature enabled as experimental (cmd line)\n");
+		}
+		break;
+	default:
+		adev->gmc.tmz_enabled = false;
+		dev_info(adev->dev,
+			 "Trusted Memory Zone (TMZ) feature not supported\n");
+		break;
+	}
+}
+
+/**
+ * amdgpu_gmc_noretry_set -- set per asic noretry defaults
+ * @adev: amdgpu_device pointer
+ *
+ * Set a per asic default for the no-retry parameter.
+ *
+ */
+void amdgpu_gmc_noretry_set(struct amdgpu_device *adev)
+{
+	struct amdgpu_gmc *gmc = &adev->gmc;
+	uint32_t gc_ver = amdgpu_ip_version(adev, GC_HWIP, 0);
+	bool noretry_default = (gc_ver == IP_VERSION(9, 0, 1) ||
+				gc_ver == IP_VERSION(9, 4, 0) ||
+				gc_ver == IP_VERSION(9, 4, 1) ||
+				gc_ver == IP_VERSION(9, 4, 2) ||
+				gc_ver == IP_VERSION(9, 4, 3) ||
+				gc_ver == IP_VERSION(9, 4, 4) ||
+				gc_ver == IP_VERSION(9, 5, 0) ||
+				gc_ver >= IP_VERSION(10, 3, 0));
+
+	if (!amdgpu_sriov_xnack_support(adev))
+		gmc->noretry = 1;
+	else
+		gmc->noretry = (amdgpu_noretry == -1) ? noretry_default : amdgpu_noretry;
+}
+
+void amdgpu_gmc_set_vm_fault_masks(struct amdgpu_device *adev, int hub_type,
+				   bool enable)
+{
+	struct amdgpu_vmhub *hub;
+	u32 tmp, reg, i;
+
+	hub = &adev->vmhub[hub_type];
+	for (i = 0; i < 16; i++) {
+		reg = hub->vm_context0_cntl + hub->ctx_distance * i;
+
+		tmp = (hub_type == AMDGPU_GFXHUB(0)) ?
+			RREG32_SOC15_IP(GC, reg) :
+			RREG32_SOC15_IP(MMHUB, reg);
+
+		if (enable)
+			tmp |= hub->vm_cntx_cntl_vm_fault;
+		else
+			tmp &= ~hub->vm_cntx_cntl_vm_fault;
+
+		(hub_type == AMDGPU_GFXHUB(0)) ?
+			WREG32_SOC15_IP(GC, reg, tmp) :
+			WREG32_SOC15_IP(MMHUB, reg, tmp);
+	}
+}
+
+void amdgpu_gmc_get_vbios_allocations(struct amdgpu_device *adev)
+{
+	unsigned size;
+
+	/*
+	 * Some ASICs need to reserve a region of video memory to avoid access
+	 * from driver
+	 */
+	adev->mman.stolen_reserved_offset = 0;
+	adev->mman.stolen_reserved_size = 0;
+
+	/*
+	 * TODO:
+	 * Currently there is a bug where some memory client outside
+	 * of the driver writes to first 8M of VRAM on S3 resume,
+	 * this overrides GART which by default gets placed in first 8M and
+	 * causes VM_FAULTS once GTT is accessed.
+	 * Keep the stolen memory reservation until the while this is not solved.
+	 */
+	switch (adev->asic_type) {
+	case CHIP_VEGA10:
+		adev->mman.keep_stolen_vga_memory = true;
+		/*
+		 * VEGA10 SRIOV VF with MS_HYPERV host needs some firmware reserved area.
+		 */
+#ifdef CONFIG_X86
+		if (amdgpu_sriov_vf(adev) && hypervisor_is_type(X86_HYPER_MS_HYPERV)) {
+			adev->mman.stolen_reserved_offset = 0x500000;
+			adev->mman.stolen_reserved_size = 0x200000;
+		}
+#endif
+		break;
+	case CHIP_RAVEN:
+	case CHIP_RENOIR:
+		adev->mman.keep_stolen_vga_memory = true;
+		break;
+	default:
+		adev->mman.keep_stolen_vga_memory = false;
+		break;
+	}
+
+	if (amdgpu_sriov_vf(adev) ||
+	    !amdgpu_device_has_display_hardware(adev)) {
+		size = 0;
+	} else {
+		size = amdgpu_gmc_get_vbios_fb_size(adev);
+
+		if (adev->mman.keep_stolen_vga_memory)
+			size = max(size, (unsigned)AMDGPU_VBIOS_VGA_ALLOCATION);
+	}
+
+	/* set to 0 if the pre-OS buffer uses up most of vram */
+	if ((adev->gmc.real_vram_size - size) < (8 * 1024 * 1024))
+		size = 0;
+
+	if (size > AMDGPU_VBIOS_VGA_ALLOCATION) {
+		adev->mman.stolen_vga_size = AMDGPU_VBIOS_VGA_ALLOCATION;
+		adev->mman.stolen_extended_size = size - adev->mman.stolen_vga_size;
+	} else {
+		adev->mman.stolen_vga_size = size;
+		adev->mman.stolen_extended_size = 0;
+	}
+}
+
+/**
+ * amdgpu_gmc_init_pdb0 - initialize PDB0
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * This function is only used when GART page table is used
+ * for FB address translatioin. In such a case, we construct
+ * a 2-level system VM page table: PDB0->PTB, to cover both
+ * VRAM of the hive and system memory.
+ *
+ * PDB0 is static, initialized once on driver initialization.
+ * The first n entries of PDB0 are used as PTE by setting
+ * P bit to 1, pointing to VRAM. The n+1'th entry points
+ * to a big PTB covering system memory.
+ *
+ */
+void amdgpu_gmc_init_pdb0(struct amdgpu_device *adev)
+{
+	int i;
+	uint64_t flags = adev->gart.gart_pte_flags; //TODO it is UC. explore NC/RW?
+	/* Each PDE0 (used as PTE) covers (2^vmid0_page_table_block_size)*2M
+	 */
+	u64 vram_size = adev->gmc.xgmi.node_segment_size * adev->gmc.xgmi.num_physical_nodes;
+	u64 pde0_page_size = (1ULL<<adev->gmc.vmid0_page_table_block_size)<<21;
+	u64 vram_addr, vram_end;
+	u64 gart_ptb_gpu_pa = amdgpu_gmc_vram_pa(adev, adev->gart.bo);
+	int idx;
+
+	if (!drm_dev_enter(adev_to_drm(adev), &idx))
+		return;
+
+	flags |= AMDGPU_PTE_VALID | AMDGPU_PTE_READABLE;
+	flags |= AMDGPU_PTE_WRITEABLE;
+	flags |= AMDGPU_PTE_SNOOPED;
+	flags |= AMDGPU_PTE_FRAG((adev->gmc.vmid0_page_table_block_size + 9*1));
+	flags |= AMDGPU_PDE_PTE_FLAG(adev);
+
+	vram_addr = adev->vm_manager.vram_base_offset;
+	if (!amdgpu_virt_xgmi_migrate_enabled(adev))
+		vram_addr -= adev->gmc.xgmi.physical_node_id * adev->gmc.xgmi.node_segment_size;
+	vram_end = vram_addr + vram_size;
+
+	/* The first n PDE0 entries are used as PTE,
+	 * pointing to vram
+	 */
+	for (i = 0; vram_addr < vram_end; i++, vram_addr += pde0_page_size)
+		amdgpu_gmc_set_pte_pde(adev, adev->gmc.ptr_pdb0, i, vram_addr, flags);
+
+	/* The n+1'th PDE0 entry points to a huge
+	 * PTB who has more than 512 entries each
+	 * pointing to a 4K system page
+	 */
+	flags = AMDGPU_PTE_VALID;
+	flags |= AMDGPU_PTE_SNOOPED | AMDGPU_PDE_BFS_FLAG(adev, 0);
+	/* Requires gart_ptb_gpu_pa to be 4K aligned */
+	amdgpu_gmc_set_pte_pde(adev, adev->gmc.ptr_pdb0, i, gart_ptb_gpu_pa, flags);
+	drm_dev_exit(idx);
+}
+
+/**
+ * amdgpu_gmc_vram_mc2pa - calculate vram buffer's physical address from MC
+ * address
+ *
+ * @adev: amdgpu_device pointer
+ * @mc_addr: MC address of buffer
+ */
+uint64_t amdgpu_gmc_vram_mc2pa(struct amdgpu_device *adev, uint64_t mc_addr)
+{
+	return mc_addr - adev->gmc.vram_start + adev->vm_manager.vram_base_offset;
+}
+
+/**
+ * amdgpu_gmc_vram_pa - calculate vram buffer object's physical address from
+ * GPU's view
+ *
+ * @adev: amdgpu_device pointer
+ * @bo: amdgpu buffer object
+ */
+uint64_t amdgpu_gmc_vram_pa(struct amdgpu_device *adev, struct amdgpu_bo *bo)
+{
+	return amdgpu_gmc_vram_mc2pa(adev, amdgpu_bo_gpu_offset(bo));
+}
+
+int amdgpu_gmc_vram_checking(struct amdgpu_device *adev)
+{
+	struct amdgpu_bo *vram_bo = NULL;
+	uint64_t vram_gpu = 0;
+	void *vram_ptr = NULL;
+
+	int ret, size = 0x100000;
+	uint8_t cptr[10];
+
+	ret = amdgpu_bo_create_kernel(adev, size, PAGE_SIZE,
+				AMDGPU_GEM_DOMAIN_VRAM,
+				&vram_bo,
+				&vram_gpu,
+				&vram_ptr);
+	if (ret)
+		return ret;
+
+	memset(vram_ptr, 0x86, size);
+	memset(cptr, 0x86, 10);
+
+	/**
+	 * Check the start, the mid, and the end of the memory if the content of
+	 * each byte is the pattern "0x86". If yes, we suppose the vram bo is
+	 * workable.
+	 *
+	 * Note: If check the each byte of whole 1M bo, it will cost too many
+	 * seconds, so here, we just pick up three parts for emulation.
+	 */
+	ret = memcmp(vram_ptr, cptr, 10);
+	if (ret) {
+		ret = -EIO;
+		goto release_buffer;
+	}
+
+	ret = memcmp(vram_ptr + (size / 2), cptr, 10);
+	if (ret) {
+		ret = -EIO;
+		goto release_buffer;
+	}
+
+	ret = memcmp(vram_ptr + size - 10, cptr, 10);
+	if (ret) {
+		ret = -EIO;
+		goto release_buffer;
+	}
+
+release_buffer:
+	amdgpu_bo_free_kernel(&vram_bo, &vram_gpu,
+			&vram_ptr);
+
+	return ret;
+}
+
+static const char *nps_desc[] = {
+	[AMDGPU_NPS1_PARTITION_MODE] = "NPS1",
+	[AMDGPU_NPS2_PARTITION_MODE] = "NPS2",
+	[AMDGPU_NPS3_PARTITION_MODE] = "NPS3",
+	[AMDGPU_NPS4_PARTITION_MODE] = "NPS4",
+	[AMDGPU_NPS6_PARTITION_MODE] = "NPS6",
+	[AMDGPU_NPS8_PARTITION_MODE] = "NPS8",
+};
+
+static ssize_t available_memory_partition_show(struct device *dev,
+					       struct device_attribute *addr,
+					       char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	int size = 0, mode;
+	char *sep = "";
+
+	for_each_inst(mode, adev->gmc.supported_nps_modes) {
+		size += sysfs_emit_at(buf, size, "%s%s", sep, nps_desc[mode]);
+		sep = ", ";
+	}
+	size += sysfs_emit_at(buf, size, "\n");
+
+	return size;
+}
+
+static ssize_t current_memory_partition_store(struct device *dev,
+					      struct device_attribute *attr,
+					      const char *buf, size_t count)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	enum amdgpu_memory_partition mode;
+	struct amdgpu_hive_info *hive;
+	int i;
+
+	mode = UNKNOWN_MEMORY_PARTITION_MODE;
+	for_each_inst(i, adev->gmc.supported_nps_modes) {
+		if (!strncasecmp(nps_desc[i], buf, strlen(nps_desc[i]))) {
+			mode = i;
+			break;
+		}
+	}
+
+	if (mode == UNKNOWN_MEMORY_PARTITION_MODE)
+		return -EINVAL;
+
+	if (mode == adev->gmc.gmc_funcs->query_mem_partition_mode(adev)) {
+		dev_info(
+			adev->dev,
+			"requested NPS mode is same as current NPS mode, skipping\n");
+		return count;
+	}
+
+	/* If device is part of hive, all devices in the hive should request the
+	 * same mode. Hence store the requested mode in hive.
+	 */
+	hive = amdgpu_get_xgmi_hive(adev);
+	if (hive) {
+		atomic_set(&hive->requested_nps_mode, mode);
+		amdgpu_put_xgmi_hive(hive);
+	} else {
+		adev->gmc.requested_nps_mode = mode;
+	}
+
+	dev_info(
+		adev->dev,
+		"NPS mode change requested, please remove and reload the driver\n");
+
+	return count;
+}
+
+static ssize_t current_memory_partition_show(
+	struct device *dev, struct device_attribute *addr, char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	enum amdgpu_memory_partition mode;
+
+	/* Only minimal precaution taken to reject requests while in reset */
+	if (amdgpu_in_reset(adev))
+		return -EPERM;
+
+	mode = adev->gmc.gmc_funcs->query_mem_partition_mode(adev);
+	if ((mode >= ARRAY_SIZE(nps_desc)) ||
+	    (BIT(mode) & AMDGPU_ALL_NPS_MASK) != BIT(mode))
+		return sysfs_emit(buf, "UNKNOWN\n");
+
+	return sysfs_emit(buf, "%s\n", nps_desc[mode]);
+}
+
+static DEVICE_ATTR_RW(current_memory_partition);
+static DEVICE_ATTR_RO(available_memory_partition);
+
+int amdgpu_gmc_sysfs_init(struct amdgpu_device *adev)
+{
+	bool nps_switch_support;
+	int r = 0;
+
+	if (!adev->gmc.gmc_funcs->query_mem_partition_mode)
+		return 0;
+
+	nps_switch_support = (hweight32(adev->gmc.supported_nps_modes &
+					AMDGPU_ALL_NPS_MASK) > 1);
+	if (!nps_switch_support)
+		dev_attr_current_memory_partition.attr.mode &=
+			~(S_IWUSR | S_IWGRP | S_IWOTH);
+	else
+		r = device_create_file(adev->dev,
+				       &dev_attr_available_memory_partition);
+
+	if (r)
+		return r;
+
+	return device_create_file(adev->dev,
+				  &dev_attr_current_memory_partition);
+}
+
+void amdgpu_gmc_sysfs_fini(struct amdgpu_device *adev)
+{
+	if (!adev->gmc.gmc_funcs->query_mem_partition_mode)
+		return;
+
+	device_remove_file(adev->dev, &dev_attr_current_memory_partition);
+	device_remove_file(adev->dev, &dev_attr_available_memory_partition);
+}
+
+int amdgpu_gmc_get_nps_memranges(struct amdgpu_device *adev,
+				 struct amdgpu_mem_partition_info *mem_ranges,
+				 uint8_t *exp_ranges)
+{
+	struct amdgpu_gmc_memrange *ranges;
+	int range_cnt, ret, i, j;
+	uint32_t nps_type;
+	bool refresh;
+
+	if (!mem_ranges || !exp_ranges)
+		return -EINVAL;
+
+	refresh = (adev->init_lvl->level != AMDGPU_INIT_LEVEL_MINIMAL_XGMI) &&
+		  (adev->gmc.reset_flags & AMDGPU_GMC_INIT_RESET_NPS);
+	ret = amdgpu_discovery_get_nps_info(adev, &nps_type, &ranges,
+					    &range_cnt, refresh);
+
+	if (ret)
+		return ret;
+
+	/* TODO: For now, expect ranges and partition count to be the same.
+	 * Adjust if there are holes expected in any NPS domain.
+	 */
+	if (*exp_ranges && (range_cnt != *exp_ranges)) {
+		dev_warn(
+			adev->dev,
+			"NPS config mismatch - expected ranges: %d discovery - nps mode: %d, nps ranges: %d",
+			*exp_ranges, nps_type, range_cnt);
+		ret = -EINVAL;
+		goto err;
+	}
+
+	for (i = 0; i < range_cnt; ++i) {
+		if (ranges[i].base_address >= ranges[i].limit_address) {
+			dev_warn(
+				adev->dev,
+				"Invalid NPS range - nps mode: %d, range[%d]: base: %llx limit: %llx",
+				nps_type, i, ranges[i].base_address,
+				ranges[i].limit_address);
+			ret = -EINVAL;
+			goto err;
+		}
+
+		/* Check for overlaps, not expecting any now */
+		for (j = i - 1; j >= 0; j--) {
+			if (max(ranges[j].base_address,
+				ranges[i].base_address) <=
+			    min(ranges[j].limit_address,
+				ranges[i].limit_address)) {
+				dev_warn(
+					adev->dev,
+					"overlapping ranges detected [ %llx - %llx ] | [%llx - %llx]",
+					ranges[j].base_address,
+					ranges[j].limit_address,
+					ranges[i].base_address,
+					ranges[i].limit_address);
+				ret = -EINVAL;
+				goto err;
+			}
+		}
+
+		mem_ranges[i].range.fpfn =
+			(ranges[i].base_address -
+			 adev->vm_manager.vram_base_offset) >>
+			AMDGPU_GPU_PAGE_SHIFT;
+		mem_ranges[i].range.lpfn =
+			(ranges[i].limit_address -
+			 adev->vm_manager.vram_base_offset) >>
+			AMDGPU_GPU_PAGE_SHIFT;
+		mem_ranges[i].size =
+			ranges[i].limit_address - ranges[i].base_address + 1;
+	}
+
+	if (!*exp_ranges)
+		*exp_ranges = range_cnt;
+err:
+	kfree(ranges);
+
+	return ret;
+}
+
+int amdgpu_gmc_request_memory_partition(struct amdgpu_device *adev,
+					int nps_mode)
+{
+	/* Not supported on VF devices and APUs */
+	if (amdgpu_sriov_vf(adev) || (adev->flags & AMD_IS_APU))
+		return -EOPNOTSUPP;
+
+	if (!adev->psp.funcs) {
+		dev_err(adev->dev,
+			"PSP interface not available for nps mode change request");
+		return -EINVAL;
+	}
+
+	return psp_memory_partition(&adev->psp, nps_mode);
+}
+
+static inline bool amdgpu_gmc_need_nps_switch_req(struct amdgpu_device *adev,
+						  int req_nps_mode,
+						  int cur_nps_mode)
+{
+	return (((BIT(req_nps_mode) & adev->gmc.supported_nps_modes) ==
+			BIT(req_nps_mode)) &&
+		req_nps_mode != cur_nps_mode);
+}
+
+void amdgpu_gmc_prepare_nps_mode_change(struct amdgpu_device *adev)
+{
+	int req_nps_mode, cur_nps_mode, r;
+	struct amdgpu_hive_info *hive;
+
+	if (amdgpu_sriov_vf(adev) || !adev->gmc.supported_nps_modes ||
+	    !adev->gmc.gmc_funcs->request_mem_partition_mode)
+		return;
+
+	cur_nps_mode = adev->gmc.gmc_funcs->query_mem_partition_mode(adev);
+	hive = amdgpu_get_xgmi_hive(adev);
+	if (hive) {
+		req_nps_mode = atomic_read(&hive->requested_nps_mode);
+		if (!amdgpu_gmc_need_nps_switch_req(adev, req_nps_mode,
+						    cur_nps_mode)) {
+			amdgpu_put_xgmi_hive(hive);
+			return;
+		}
+		r = amdgpu_xgmi_request_nps_change(adev, hive, req_nps_mode);
+		amdgpu_put_xgmi_hive(hive);
+		goto out;
+	}
+
+	req_nps_mode = adev->gmc.requested_nps_mode;
+	if (!amdgpu_gmc_need_nps_switch_req(adev, req_nps_mode, cur_nps_mode))
+		return;
+
+	/* even if this fails, we should let driver unload w/o blocking */
+	r = adev->gmc.gmc_funcs->request_mem_partition_mode(adev, req_nps_mode);
+out:
+	if (r)
+		dev_err(adev->dev, "NPS mode change request failed\n");
+	else
+		dev_info(
+			adev->dev,
+			"NPS mode change request done, reload driver to complete the change\n");
+}
+
+bool amdgpu_gmc_need_reset_on_init(struct amdgpu_device *adev)
+{
+	if (adev->gmc.gmc_funcs->need_reset_on_init)
+		return adev->gmc.gmc_funcs->need_reset_on_init(adev);
+
+	return false;
+}
+
+enum amdgpu_memory_partition
+amdgpu_gmc_get_vf_memory_partition(struct amdgpu_device *adev)
+{
+	switch (adev->gmc.num_mem_partitions) {
+	case 0:
+		return UNKNOWN_MEMORY_PARTITION_MODE;
+	case 1:
+		return AMDGPU_NPS1_PARTITION_MODE;
+	case 2:
+		return AMDGPU_NPS2_PARTITION_MODE;
+	case 4:
+		return AMDGPU_NPS4_PARTITION_MODE;
+	case 8:
+		return AMDGPU_NPS8_PARTITION_MODE;
+	default:
+		return AMDGPU_NPS1_PARTITION_MODE;
+	}
+}
+
+enum amdgpu_memory_partition
+amdgpu_gmc_get_memory_partition(struct amdgpu_device *adev, u32 *supp_modes)
+{
+	enum amdgpu_memory_partition mode = UNKNOWN_MEMORY_PARTITION_MODE;
+
+	if (adev->nbio.funcs &&
+	    adev->nbio.funcs->get_memory_partition_mode)
+		mode = adev->nbio.funcs->get_memory_partition_mode(adev,
+								   supp_modes);
+	else
+		dev_warn(adev->dev, "memory partition mode query is not supported\n");
+
+	return mode;
+}
+
+enum amdgpu_memory_partition
+amdgpu_gmc_query_memory_partition(struct amdgpu_device *adev)
+{
+	if (amdgpu_sriov_vf(adev))
+		return amdgpu_gmc_get_vf_memory_partition(adev);
+	else
+		return amdgpu_gmc_get_memory_partition(adev, NULL);
+}
+
+static bool amdgpu_gmc_validate_partition_info(struct amdgpu_device *adev)
+{
+	enum amdgpu_memory_partition mode;
+	u32 supp_modes;
+	bool valid;
+
+	mode = amdgpu_gmc_get_memory_partition(adev, &supp_modes);
+
+	/* Mode detected by hardware not present in supported modes */
+	if ((mode != UNKNOWN_MEMORY_PARTITION_MODE) &&
+	    !(BIT(mode - 1) & supp_modes))
+		return false;
+
+	switch (mode) {
+	case UNKNOWN_MEMORY_PARTITION_MODE:
+	case AMDGPU_NPS1_PARTITION_MODE:
+		valid = (adev->gmc.num_mem_partitions == 1);
+		break;
+	case AMDGPU_NPS2_PARTITION_MODE:
+		valid = (adev->gmc.num_mem_partitions == 2);
+		break;
+	case AMDGPU_NPS4_PARTITION_MODE:
+		valid = (adev->gmc.num_mem_partitions == 3 ||
+			 adev->gmc.num_mem_partitions == 4);
+		break;
+	case AMDGPU_NPS8_PARTITION_MODE:
+		valid = (adev->gmc.num_mem_partitions == 8);
+		break;
+	default:
+		valid = false;
+	}
+
+	return valid;
+}
+
+static bool amdgpu_gmc_is_node_present(int *node_ids, int num_ids, int nid)
+{
+	int i;
+
+	/* Check if node with id 'nid' is present in 'node_ids' array */
+	for (i = 0; i < num_ids; ++i)
+		if (node_ids[i] == nid)
+			return true;
+
+	return false;
+}
+
+static void
+amdgpu_gmc_init_acpi_mem_ranges(struct amdgpu_device *adev,
+				struct amdgpu_mem_partition_info *mem_ranges)
+{
+	struct amdgpu_numa_info numa_info;
+	int node_ids[AMDGPU_MAX_MEM_RANGES];
+	int num_ranges = 0, ret;
+	int num_xcc, xcc_id;
+	uint32_t xcc_mask;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	xcc_mask = (1U << num_xcc) - 1;
+
+	for_each_inst(xcc_id, xcc_mask)	{
+		ret = amdgpu_acpi_get_mem_info(adev, xcc_id, &numa_info);
+		if (ret)
+			continue;
+
+		if (numa_info.nid == NUMA_NO_NODE) {
+			mem_ranges[0].size = numa_info.size;
+			mem_ranges[0].numa.node = numa_info.nid;
+			num_ranges = 1;
+			break;
+		}
+
+		if (amdgpu_gmc_is_node_present(node_ids, num_ranges,
+					     numa_info.nid))
+			continue;
+
+		node_ids[num_ranges] = numa_info.nid;
+		mem_ranges[num_ranges].numa.node = numa_info.nid;
+		mem_ranges[num_ranges].size = numa_info.size;
+		++num_ranges;
+	}
+
+	adev->gmc.num_mem_partitions = num_ranges;
+}
+
+void amdgpu_gmc_init_sw_mem_ranges(struct amdgpu_device *adev,
+				   struct amdgpu_mem_partition_info *mem_ranges)
+{
+	enum amdgpu_memory_partition mode;
+	u32 start_addr = 0, size;
+	int i, r, l;
+
+	mode = amdgpu_gmc_query_memory_partition(adev);
+
+	switch (mode) {
+	case UNKNOWN_MEMORY_PARTITION_MODE:
+		adev->gmc.num_mem_partitions = 0;
+		break;
+	case AMDGPU_NPS1_PARTITION_MODE:
+		adev->gmc.num_mem_partitions = 1;
+		break;
+	case AMDGPU_NPS2_PARTITION_MODE:
+		adev->gmc.num_mem_partitions = 2;
+		break;
+	case AMDGPU_NPS4_PARTITION_MODE:
+		if (adev->flags & AMD_IS_APU)
+			adev->gmc.num_mem_partitions = 3;
+		else
+			adev->gmc.num_mem_partitions = 4;
+		break;
+	case AMDGPU_NPS8_PARTITION_MODE:
+		adev->gmc.num_mem_partitions = 8;
+		break;
+	default:
+		adev->gmc.num_mem_partitions = 1;
+		break;
+	}
+
+	/* Use NPS range info, if populated */
+	r = amdgpu_gmc_get_nps_memranges(adev, mem_ranges,
+					 &adev->gmc.num_mem_partitions);
+	if (!r) {
+		l = 0;
+		for (i = 1; i < adev->gmc.num_mem_partitions; ++i) {
+			if (mem_ranges[i].range.lpfn >
+			    mem_ranges[i - 1].range.lpfn)
+				l = i;
+		}
+
+	} else {
+		if (!adev->gmc.num_mem_partitions) {
+			dev_warn(adev->dev,
+				 "Not able to detect NPS mode, fall back to NPS1\n");
+			adev->gmc.num_mem_partitions = 1;
+		}
+		/* Fallback to sw based calculation */
+		size = (adev->gmc.real_vram_size + SZ_16M) >> AMDGPU_GPU_PAGE_SHIFT;
+		size /= adev->gmc.num_mem_partitions;
+
+		for (i = 0; i < adev->gmc.num_mem_partitions; ++i) {
+			mem_ranges[i].range.fpfn = start_addr;
+			mem_ranges[i].size =
+				((u64)size << AMDGPU_GPU_PAGE_SHIFT);
+			mem_ranges[i].range.lpfn = start_addr + size - 1;
+			start_addr += size;
+		}
+
+		l = adev->gmc.num_mem_partitions - 1;
+	}
+
+	/* Adjust the last one */
+	mem_ranges[l].range.lpfn =
+		(adev->gmc.real_vram_size >> AMDGPU_GPU_PAGE_SHIFT) - 1;
+	mem_ranges[l].size =
+		adev->gmc.real_vram_size -
+		((u64)mem_ranges[l].range.fpfn << AMDGPU_GPU_PAGE_SHIFT);
+}
+
+int amdgpu_gmc_init_mem_ranges(struct amdgpu_device *adev)
+{
+	bool valid;
+
+	adev->gmc.mem_partitions = kcalloc(AMDGPU_MAX_MEM_RANGES,
+					   sizeof(struct amdgpu_mem_partition_info),
+					   GFP_KERNEL);
+	if (!adev->gmc.mem_partitions)
+		return -ENOMEM;
+
+	if (adev->gmc.is_app_apu)
+		amdgpu_gmc_init_acpi_mem_ranges(adev, adev->gmc.mem_partitions);
+	else
+		amdgpu_gmc_init_sw_mem_ranges(adev, adev->gmc.mem_partitions);
+
+	if (amdgpu_sriov_vf(adev))
+		valid = true;
+	else
+		valid = amdgpu_gmc_validate_partition_info(adev);
+	if (!valid) {
+		/* TODO: handle invalid case */
+		dev_warn(adev->dev,
+			 "Mem ranges not matching with hardware config\n");
+	}
+
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
new file mode 100644
index 000000000000..55097ca10738
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
@@ -0,0 +1,471 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+#ifndef __AMDGPU_GMC_H__
+#define __AMDGPU_GMC_H__
+
+#include <linux/types.h>
+
+#include "amdgpu_irq.h"
+#include "amdgpu_xgmi.h"
+#include "amdgpu_ras.h"
+
+/* VA hole for 48bit addresses on Vega10 */
+#define AMDGPU_GMC_HOLE_START	0x0000800000000000ULL
+#define AMDGPU_GMC_HOLE_END	0xffff800000000000ULL
+
+/*
+ * Hardware is programmed as if the hole doesn't exists with start and end
+ * address values.
+ *
+ * This mask is used to remove the upper 16bits of the VA and so come up with
+ * the linear addr value.
+ */
+#define AMDGPU_GMC_HOLE_MASK	0x0000ffffffffffffULL
+
+/*
+ * Ring size as power of two for the log of recent faults.
+ */
+#define AMDGPU_GMC_FAULT_RING_ORDER	8
+#define AMDGPU_GMC_FAULT_RING_SIZE	(1 << AMDGPU_GMC_FAULT_RING_ORDER)
+
+/*
+ * Hash size as power of two for the log of recent faults
+ */
+#define AMDGPU_GMC_FAULT_HASH_ORDER	8
+#define AMDGPU_GMC_FAULT_HASH_SIZE	(1 << AMDGPU_GMC_FAULT_HASH_ORDER)
+
+/*
+ * Number of IH timestamp ticks until a fault is considered handled
+ */
+#define AMDGPU_GMC_FAULT_TIMEOUT	5000ULL
+
+/* XNACK flags */
+#define AMDGPU_GMC_XNACK_FLAG_CHAIN BIT(0)
+
+struct firmware;
+
+enum amdgpu_memory_partition {
+	UNKNOWN_MEMORY_PARTITION_MODE = 0,
+	AMDGPU_NPS1_PARTITION_MODE = 1,
+	AMDGPU_NPS2_PARTITION_MODE = 2,
+	AMDGPU_NPS3_PARTITION_MODE = 3,
+	AMDGPU_NPS4_PARTITION_MODE = 4,
+	AMDGPU_NPS6_PARTITION_MODE = 6,
+	AMDGPU_NPS8_PARTITION_MODE = 8,
+};
+
+#define AMDGPU_ALL_NPS_MASK                                                  \
+	(BIT(AMDGPU_NPS1_PARTITION_MODE) | BIT(AMDGPU_NPS2_PARTITION_MODE) | \
+	 BIT(AMDGPU_NPS3_PARTITION_MODE) | BIT(AMDGPU_NPS4_PARTITION_MODE) | \
+	 BIT(AMDGPU_NPS6_PARTITION_MODE) | BIT(AMDGPU_NPS8_PARTITION_MODE))
+
+#define AMDGPU_GMC_INIT_RESET_NPS  BIT(0)
+
+#define AMDGPU_MAX_MEM_RANGES 8
+
+/*
+ * GMC page fault information
+ */
+struct amdgpu_gmc_fault {
+	uint64_t	timestamp:48;
+	uint64_t	next:AMDGPU_GMC_FAULT_RING_ORDER;
+	atomic64_t	key;
+	uint64_t	timestamp_expiry:48;
+};
+
+/*
+ * VMHUB structures, functions & helpers
+ */
+struct amdgpu_vmhub_funcs {
+	void (*print_l2_protection_fault_status)(struct amdgpu_device *adev,
+						 uint32_t status);
+	uint32_t (*get_invalidate_req)(unsigned int vmid, uint32_t flush_type);
+};
+
+struct amdgpu_vmhub {
+	uint32_t	ctx0_ptb_addr_lo32;
+	uint32_t	ctx0_ptb_addr_hi32;
+	uint32_t	vm_inv_eng0_sem;
+	uint32_t	vm_inv_eng0_req;
+	uint32_t	vm_inv_eng0_ack;
+	uint32_t	vm_context0_cntl;
+	uint32_t	vm_l2_pro_fault_status;
+	uint32_t	vm_l2_pro_fault_cntl;
+
+	/*
+	 * store the register distances between two continuous context domain
+	 * and invalidation engine.
+	 */
+	uint32_t	ctx_distance;
+	uint32_t	ctx_addr_distance; /* include LO32/HI32 */
+	uint32_t	eng_distance;
+	uint32_t	eng_addr_distance; /* include LO32/HI32 */
+
+	uint32_t        vm_cntx_cntl;
+	uint32_t	vm_cntx_cntl_vm_fault;
+	uint32_t	vm_l2_bank_select_reserved_cid2;
+
+	uint32_t	vm_contexts_disable;
+
+	bool		sdma_invalidation_workaround;
+
+	const struct amdgpu_vmhub_funcs *vmhub_funcs;
+};
+
+/*
+ * GPU MC structures, functions & helpers
+ */
+struct amdgpu_gmc_funcs {
+	/* flush the vm tlb via mmio */
+	void (*flush_gpu_tlb)(struct amdgpu_device *adev, uint32_t vmid,
+				uint32_t vmhub, uint32_t flush_type);
+	/* flush the vm tlb via pasid */
+	void (*flush_gpu_tlb_pasid)(struct amdgpu_device *adev, uint16_t pasid,
+				    uint32_t flush_type, bool all_hub,
+				    uint32_t inst);
+	/* flush the vm tlb via ring */
+	uint64_t (*emit_flush_gpu_tlb)(struct amdgpu_ring *ring, unsigned vmid,
+				       uint64_t pd_addr);
+	/* Change the VMID -> PASID mapping */
+	void (*emit_pasid_mapping)(struct amdgpu_ring *ring, unsigned vmid,
+				   unsigned pasid);
+	/* enable/disable PRT support */
+	void (*set_prt)(struct amdgpu_device *adev, bool enable);
+	/* get the pde for a given mc addr */
+	void (*get_vm_pde)(struct amdgpu_device *adev, int level,
+			   u64 *dst, u64 *flags);
+	/* get the pte flags to use for PTEs */
+	void (*get_vm_pte)(struct amdgpu_device *adev,
+			   struct amdgpu_vm *vm,
+			   struct amdgpu_bo *bo,
+			   uint32_t vm_flags,
+			   uint64_t *pte_flags);
+	/* override per-page pte flags */
+	void (*override_vm_pte_flags)(struct amdgpu_device *dev,
+				      struct amdgpu_vm *vm,
+				      uint64_t addr, uint64_t *flags);
+	/* get the amount of memory used by the vbios for pre-OS console */
+	unsigned int (*get_vbios_fb_size)(struct amdgpu_device *adev);
+	/* get the DCC buffer alignment */
+	unsigned int (*get_dcc_alignment)(struct amdgpu_device *adev);
+
+	enum amdgpu_memory_partition (*query_mem_partition_mode)(
+		struct amdgpu_device *adev);
+	/* Request NPS mode */
+	int (*request_mem_partition_mode)(struct amdgpu_device *adev,
+					  int nps_mode);
+	bool (*need_reset_on_init)(struct amdgpu_device *adev);
+};
+
+struct amdgpu_mem_partition_info {
+	union {
+		struct {
+			uint32_t fpfn;
+			uint32_t lpfn;
+		} range;
+		struct {
+			int node;
+		} numa;
+	};
+	uint64_t size;
+};
+
+#define INVALID_PFN    -1
+
+struct amdgpu_gmc_memrange {
+	uint64_t base_address;
+	uint64_t limit_address;
+	uint32_t flags;
+	int nid_mask;
+};
+
+enum amdgpu_gart_placement {
+	AMDGPU_GART_PLACEMENT_BEST_FIT = 0,
+	AMDGPU_GART_PLACEMENT_HIGH,
+	AMDGPU_GART_PLACEMENT_LOW,
+};
+
+struct amdgpu_gmc {
+	/* FB's physical address in MMIO space (for CPU to
+	 * map FB). This is different compared to the agp/
+	 * gart/vram_start/end field as the later is from
+	 * GPU's view and aper_base is from CPU's view.
+	 */
+	resource_size_t		aper_size;
+	resource_size_t		aper_base;
+	/* for some chips with <= 32MB we need to lie
+	 * about vram size near mc fb location */
+	u64			mc_vram_size;
+	u64			visible_vram_size;
+	/* AGP aperture start and end in MC address space
+	 * Driver find a hole in the MC address space
+	 * to place AGP by setting MC_VM_AGP_BOT/TOP registers
+	 * Under VMID0, logical address == MC address. AGP
+	 * aperture maps to physical bus or IOVA addressed.
+	 * AGP aperture is used to simulate FB in ZFB case.
+	 * AGP aperture is also used for page table in system
+	 * memory (mainly for APU).
+	 *
+	 */
+	u64			agp_size;
+	u64			agp_start;
+	u64			agp_end;
+	/* GART aperture start and end in MC address space
+	 * Driver find a hole in the MC address space
+	 * to place GART by setting VM_CONTEXT0_PAGE_TABLE_START/END_ADDR
+	 * registers
+	 * Under VMID0, logical address inside GART aperture will
+	 * be translated through gpuvm gart page table to access
+	 * paged system memory
+	 */
+	u64			gart_size;
+	u64			gart_start;
+	u64			gart_end;
+	/* Frame buffer aperture of this GPU device. Different from
+	 * fb_start (see below), this only covers the local GPU device.
+	 * If driver uses FB aperture to access FB, driver get fb_start from
+	 * MC_VM_FB_LOCATION_BASE (set by vbios) and calculate vram_start
+	 * of this local device by adding an offset inside the XGMI hive.
+	 * If driver uses GART table for VMID0 FB access, driver finds a hole in
+	 * VMID0's virtual address space to place the SYSVM aperture inside
+	 * which the first part is vram and the second part is gart (covering
+	 * system ram).
+	 */
+	u64			vram_start;
+	u64			vram_end;
+	/* FB region , it's same as local vram region in single GPU, in XGMI
+	 * configuration, this region covers all GPUs in the same hive ,
+	 * each GPU in the hive has the same view of this FB region .
+	 * GPU0's vram starts at offset (0 * segment size) ,
+	 * GPU1 starts at offset (1 * segment size), etc.
+	 */
+	u64			fb_start;
+	u64			fb_end;
+	unsigned		vram_width;
+	u64			real_vram_size;
+	int			vram_mtrr;
+	u64                     mc_mask;
+	const struct firmware   *fw;	/* MC firmware */
+	uint32_t                fw_version;
+	struct amdgpu_irq_src	vm_fault;
+	uint32_t		vram_type;
+	uint8_t			vram_vendor;
+	uint32_t                srbm_soft_reset;
+	bool			prt_warning;
+	uint32_t		sdpif_register;
+	/* apertures */
+	u64			shared_aperture_start;
+	u64			shared_aperture_end;
+	u64			private_aperture_start;
+	u64			private_aperture_end;
+	/* protects concurrent invalidation */
+	spinlock_t		invalidate_lock;
+	bool			translate_further;
+	struct kfd_vm_fault_info *vm_fault_info;
+	atomic_t		vm_fault_info_updated;
+
+	struct amdgpu_gmc_fault	fault_ring[AMDGPU_GMC_FAULT_RING_SIZE];
+	struct {
+		uint64_t	idx:AMDGPU_GMC_FAULT_RING_ORDER;
+	} fault_hash[AMDGPU_GMC_FAULT_HASH_SIZE];
+	uint64_t		last_fault:AMDGPU_GMC_FAULT_RING_ORDER;
+
+	bool tmz_enabled;
+	bool is_app_apu;
+
+	struct amdgpu_mem_partition_info *mem_partitions;
+	uint8_t num_mem_partitions;
+	const struct amdgpu_gmc_funcs	*gmc_funcs;
+	enum amdgpu_memory_partition	requested_nps_mode;
+	uint32_t supported_nps_modes;
+	uint32_t reset_flags;
+
+	struct amdgpu_xgmi xgmi;
+	struct amdgpu_irq_src	ecc_irq;
+	int noretry;
+	uint32_t xnack_flags;
+
+	uint32_t	vmid0_page_table_block_size;
+	uint32_t	vmid0_page_table_depth;
+	struct amdgpu_bo		*pdb0_bo;
+	/* CPU kmapped address of pdb0*/
+	void				*ptr_pdb0;
+
+	/* MALL size */
+	u64 mall_size;
+	uint32_t m_half_use;
+
+	/* number of UMC instances */
+	int num_umc;
+	/* mode2 save restore */
+	u64 VM_L2_CNTL;
+	u64 VM_L2_CNTL2;
+	u64 VM_DUMMY_PAGE_FAULT_CNTL;
+	u64 VM_DUMMY_PAGE_FAULT_ADDR_LO32;
+	u64 VM_DUMMY_PAGE_FAULT_ADDR_HI32;
+	u64 VM_L2_PROTECTION_FAULT_CNTL;
+	u64 VM_L2_PROTECTION_FAULT_CNTL2;
+	u64 VM_L2_PROTECTION_FAULT_MM_CNTL3;
+	u64 VM_L2_PROTECTION_FAULT_MM_CNTL4;
+	u64 VM_L2_PROTECTION_FAULT_ADDR_LO32;
+	u64 VM_L2_PROTECTION_FAULT_ADDR_HI32;
+	u64 VM_DEBUG;
+	u64 VM_L2_MM_GROUP_RT_CLASSES;
+	u64 VM_L2_BANK_SELECT_RESERVED_CID;
+	u64 VM_L2_BANK_SELECT_RESERVED_CID2;
+	u64 VM_L2_CACHE_PARITY_CNTL;
+	u64 VM_L2_IH_LOG_CNTL;
+	u64 VM_CONTEXT_CNTL[16];
+	u64 VM_CONTEXT_PAGE_TABLE_BASE_ADDR_LO32[16];
+	u64 VM_CONTEXT_PAGE_TABLE_BASE_ADDR_HI32[16];
+	u64 VM_CONTEXT_PAGE_TABLE_START_ADDR_LO32[16];
+	u64 VM_CONTEXT_PAGE_TABLE_START_ADDR_HI32[16];
+	u64 VM_CONTEXT_PAGE_TABLE_END_ADDR_LO32[16];
+	u64 VM_CONTEXT_PAGE_TABLE_END_ADDR_HI32[16];
+	u64 MC_VM_MX_L1_TLB_CNTL;
+
+	u64 noretry_flags;
+
+	bool flush_tlb_needs_extra_type_0;
+	bool flush_tlb_needs_extra_type_2;
+	bool flush_pasid_uses_kiq;
+};
+
+#define amdgpu_gmc_emit_flush_gpu_tlb(r, vmid, addr) (r)->adev->gmc.gmc_funcs->emit_flush_gpu_tlb((r), (vmid), (addr))
+#define amdgpu_gmc_emit_pasid_mapping(r, vmid, pasid) (r)->adev->gmc.gmc_funcs->emit_pasid_mapping((r), (vmid), (pasid))
+#define amdgpu_gmc_get_vm_pde(adev, level, dst, flags) (adev)->gmc.gmc_funcs->get_vm_pde((adev), (level), (dst), (flags))
+#define amdgpu_gmc_get_vm_pte(adev, vm, bo, vm_flags, pte_flags) \
+	((adev)->gmc.gmc_funcs->get_vm_pte((adev), (vm), (bo), (vm_flags), \
+					   (pte_flags)))
+#define amdgpu_gmc_override_vm_pte_flags(adev, vm, addr, pte_flags)	\
+	(adev)->gmc.gmc_funcs->override_vm_pte_flags			\
+		((adev), (vm), (addr), (pte_flags))
+#define amdgpu_gmc_get_vbios_fb_size(adev) (adev)->gmc.gmc_funcs->get_vbios_fb_size((adev))
+#define amdgpu_gmc_get_dcc_alignment(adev) ({			\
+	typeof(adev) _adev = (adev);				\
+	_adev->gmc.gmc_funcs->get_dcc_alignment(_adev);		\
+})
+
+/**
+ * amdgpu_gmc_vram_full_visible - Check if full VRAM is visible through the BAR
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Returns:
+ * True if full VRAM is visible through the BAR
+ */
+static inline bool amdgpu_gmc_vram_full_visible(struct amdgpu_gmc *gmc)
+{
+	WARN_ON(gmc->real_vram_size < gmc->visible_vram_size);
+
+	return (gmc->real_vram_size == gmc->visible_vram_size);
+}
+
+/**
+ * amdgpu_gmc_sign_extend - sign extend the given gmc address
+ *
+ * @addr: address to extend
+ */
+static inline uint64_t amdgpu_gmc_sign_extend(uint64_t addr)
+{
+	if (addr >= AMDGPU_GMC_HOLE_START)
+		addr |= AMDGPU_GMC_HOLE_END;
+
+	return addr;
+}
+
+bool amdgpu_gmc_is_pdb0_enabled(struct amdgpu_device *adev);
+int amdgpu_gmc_pdb0_alloc(struct amdgpu_device *adev);
+void amdgpu_gmc_get_pde_for_bo(struct amdgpu_bo *bo, int level,
+			       uint64_t *addr, uint64_t *flags);
+int amdgpu_gmc_set_pte_pde(struct amdgpu_device *adev, void *cpu_pt_addr,
+				uint32_t gpu_page_idx, uint64_t addr,
+				uint64_t flags);
+uint64_t amdgpu_gmc_pd_addr(struct amdgpu_bo *bo);
+uint64_t amdgpu_gmc_agp_addr(struct ttm_buffer_object *bo);
+void amdgpu_gmc_sysvm_location(struct amdgpu_device *adev, struct amdgpu_gmc *mc);
+void amdgpu_gmc_vram_location(struct amdgpu_device *adev, struct amdgpu_gmc *mc,
+			      u64 base);
+void amdgpu_gmc_gart_location(struct amdgpu_device *adev,
+			      struct amdgpu_gmc *mc,
+			      enum amdgpu_gart_placement gart_placement);
+void amdgpu_gmc_agp_location(struct amdgpu_device *adev,
+			     struct amdgpu_gmc *mc);
+void amdgpu_gmc_set_agp_default(struct amdgpu_device *adev,
+				struct amdgpu_gmc *mc);
+bool amdgpu_gmc_filter_faults(struct amdgpu_device *adev,
+			      struct amdgpu_ih_ring *ih, uint64_t addr,
+			      uint16_t pasid, uint64_t timestamp);
+void amdgpu_gmc_filter_faults_remove(struct amdgpu_device *adev, uint64_t addr,
+				     uint16_t pasid);
+int amdgpu_gmc_ras_sw_init(struct amdgpu_device *adev);
+int amdgpu_gmc_ras_late_init(struct amdgpu_device *adev);
+void amdgpu_gmc_ras_fini(struct amdgpu_device *adev);
+int amdgpu_gmc_allocate_vm_inv_eng(struct amdgpu_device *adev);
+void amdgpu_gmc_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid,
+			      uint32_t vmhub, uint32_t flush_type);
+int amdgpu_gmc_flush_gpu_tlb_pasid(struct amdgpu_device *adev, uint16_t pasid,
+				   uint32_t flush_type, bool all_hub,
+				   uint32_t inst);
+void amdgpu_gmc_fw_reg_write_reg_wait(struct amdgpu_device *adev,
+				      uint32_t reg0, uint32_t reg1,
+				      uint32_t ref, uint32_t mask,
+				      uint32_t xcc_inst);
+
+extern void amdgpu_gmc_tmz_set(struct amdgpu_device *adev);
+extern void amdgpu_gmc_noretry_set(struct amdgpu_device *adev);
+
+extern void
+amdgpu_gmc_set_vm_fault_masks(struct amdgpu_device *adev, int hub_type,
+			      bool enable);
+
+void amdgpu_gmc_get_vbios_allocations(struct amdgpu_device *adev);
+
+void amdgpu_gmc_init_pdb0(struct amdgpu_device *adev);
+uint64_t amdgpu_gmc_vram_mc2pa(struct amdgpu_device *adev, uint64_t mc_addr);
+uint64_t amdgpu_gmc_vram_pa(struct amdgpu_device *adev, struct amdgpu_bo *bo);
+int amdgpu_gmc_vram_checking(struct amdgpu_device *adev);
+int amdgpu_gmc_sysfs_init(struct amdgpu_device *adev);
+void amdgpu_gmc_sysfs_fini(struct amdgpu_device *adev);
+
+int amdgpu_gmc_get_nps_memranges(struct amdgpu_device *adev,
+				 struct amdgpu_mem_partition_info *mem_ranges,
+				 uint8_t *exp_ranges);
+
+int amdgpu_gmc_request_memory_partition(struct amdgpu_device *adev,
+					int nps_mode);
+void amdgpu_gmc_prepare_nps_mode_change(struct amdgpu_device *adev);
+bool amdgpu_gmc_need_reset_on_init(struct amdgpu_device *adev);
+enum amdgpu_memory_partition
+amdgpu_gmc_get_vf_memory_partition(struct amdgpu_device *adev);
+enum amdgpu_memory_partition
+amdgpu_gmc_get_memory_partition(struct amdgpu_device *adev, u32 *supp_modes);
+enum amdgpu_memory_partition
+amdgpu_gmc_query_memory_partition(struct amdgpu_device *adev);
+int amdgpu_gmc_init_mem_ranges(struct amdgpu_device *adev);
+void amdgpu_gmc_init_sw_mem_ranges(struct amdgpu_device *adev,
+				   struct amdgpu_mem_partition_info *mem_ranges);
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c
index 0d15eb7d31d7..0760e70402ec 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c
@@ -22,131 +22,82 @@
  * Authors: Christian König
  */
 
-#include <drm/drmP.h>
-#include "amdgpu.h"
+#include <drm/ttm/ttm_range_manager.h>
 
-struct amdgpu_gtt_mgr {
-	struct drm_mm mm;
-	spinlock_t lock;
-	atomic64_t available;
-};
+#include "amdgpu.h"
 
-/**
- * amdgpu_gtt_mgr_init - init GTT manager and DRM MM
- *
- * @man: TTM memory type manager
- * @p_size: maximum size of GTT
- *
- * Allocate and initialize the GTT manager.
- */
-static int amdgpu_gtt_mgr_init(struct ttm_mem_type_manager *man,
-			       unsigned long p_size)
+static inline struct amdgpu_gtt_mgr *
+to_gtt_mgr(struct ttm_resource_manager *man)
 {
-	struct amdgpu_device *adev = amdgpu_ttm_adev(man->bdev);
-	struct amdgpu_gtt_mgr *mgr;
-	uint64_t start, size;
-
-	mgr = kzalloc(sizeof(*mgr), GFP_KERNEL);
-	if (!mgr)
-		return -ENOMEM;
-
-	start = AMDGPU_GTT_MAX_TRANSFER_SIZE * AMDGPU_GTT_NUM_TRANSFER_WINDOWS;
-	size = (adev->mc.gart_size >> PAGE_SHIFT) - start;
-	drm_mm_init(&mgr->mm, start, size);
-	spin_lock_init(&mgr->lock);
-	atomic64_set(&mgr->available, p_size);
-	man->priv = mgr;
-	return 0;
+	return container_of(man, struct amdgpu_gtt_mgr, manager);
 }
 
 /**
- * amdgpu_gtt_mgr_fini - free and destroy GTT manager
+ * DOC: mem_info_gtt_total
  *
- * @man: TTM memory type manager
- *
- * Destroy and free the GTT manager, returns -EBUSY if ranges are still
- * allocated inside it.
+ * The amdgpu driver provides a sysfs API for reporting current total size of
+ * the GTT.
+ * The file mem_info_gtt_total is used for this, and returns the total size of
+ * the GTT block, in bytes
  */
-static int amdgpu_gtt_mgr_fini(struct ttm_mem_type_manager *man)
+static ssize_t amdgpu_mem_info_gtt_total_show(struct device *dev,
+					      struct device_attribute *attr,
+					      char *buf)
 {
-	struct amdgpu_gtt_mgr *mgr = man->priv;
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	struct ttm_resource_manager *man;
 
-	spin_lock(&mgr->lock);
-	if (!drm_mm_clean(&mgr->mm)) {
-		spin_unlock(&mgr->lock);
-		return -EBUSY;
-	}
-
-	drm_mm_takedown(&mgr->mm);
-	spin_unlock(&mgr->lock);
-	kfree(mgr);
-	man->priv = NULL;
-	return 0;
+	man = ttm_manager_type(&adev->mman.bdev, TTM_PL_TT);
+	return sysfs_emit(buf, "%llu\n", man->size);
 }
 
 /**
- * amdgpu_gtt_mgr_is_allocated - Check if mem has address space
+ * DOC: mem_info_gtt_used
  *
- * @mem: the mem object to check
- *
- * Check if a mem object has already address space allocated.
+ * The amdgpu driver provides a sysfs API for reporting current total amount of
+ * used GTT.
+ * The file mem_info_gtt_used is used for this, and returns the current used
+ * size of the GTT block, in bytes
  */
-bool amdgpu_gtt_mgr_is_allocated(struct ttm_mem_reg *mem)
+static ssize_t amdgpu_mem_info_gtt_used_show(struct device *dev,
+					     struct device_attribute *attr,
+					     char *buf)
 {
-	struct drm_mm_node *node = mem->mm_node;
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	struct ttm_resource_manager *man = &adev->mman.gtt_mgr.manager;
 
-	return (node->start != AMDGPU_BO_INVALID_OFFSET);
+	return sysfs_emit(buf, "%llu\n", ttm_resource_manager_usage(man));
 }
 
+static DEVICE_ATTR(mem_info_gtt_total, S_IRUGO,
+	           amdgpu_mem_info_gtt_total_show, NULL);
+static DEVICE_ATTR(mem_info_gtt_used, S_IRUGO,
+	           amdgpu_mem_info_gtt_used_show, NULL);
+
+static struct attribute *amdgpu_gtt_mgr_attributes[] = {
+	&dev_attr_mem_info_gtt_total.attr,
+	&dev_attr_mem_info_gtt_used.attr,
+	NULL
+};
+
+const struct attribute_group amdgpu_gtt_mgr_attr_group = {
+	.attrs = amdgpu_gtt_mgr_attributes
+};
+
 /**
- * amdgpu_gtt_mgr_alloc - allocate new ranges
+ * amdgpu_gtt_mgr_has_gart_addr - Check if mem has address space
  *
- * @man: TTM memory type manager
- * @tbo: TTM BO we need this range for
- * @place: placement flags and restrictions
- * @mem: the resulting mem object
+ * @res: the mem object to check
  *
- * Allocate the address space for a node.
+ * Check if a mem object has already address space allocated.
  */
-static int amdgpu_gtt_mgr_alloc(struct ttm_mem_type_manager *man,
-				struct ttm_buffer_object *tbo,
-				const struct ttm_place *place,
-				struct ttm_mem_reg *mem)
+bool amdgpu_gtt_mgr_has_gart_addr(struct ttm_resource *res)
 {
-	struct amdgpu_device *adev = amdgpu_ttm_adev(man->bdev);
-	struct amdgpu_gtt_mgr *mgr = man->priv;
-	struct drm_mm_node *node = mem->mm_node;
-	enum drm_mm_insert_mode mode;
-	unsigned long fpfn, lpfn;
-	int r;
-
-	if (amdgpu_gtt_mgr_is_allocated(mem))
-		return 0;
-
-	if (place)
-		fpfn = place->fpfn;
-	else
-		fpfn = 0;
-
-	if (place && place->lpfn)
-		lpfn = place->lpfn;
-	else
-		lpfn = adev->gart.num_cpu_pages;
+	struct ttm_range_mgr_node *node = to_ttm_range_mgr_node(res);
 
-	mode = DRM_MM_INSERT_BEST;
-	if (place && place->flags & TTM_PL_FLAG_TOPDOWN)
-		mode = DRM_MM_INSERT_HIGH;
-
-	spin_lock(&mgr->lock);
-	r = drm_mm_insert_node_in_range(&mgr->mm, node,
-					mem->num_pages, mem->page_alignment, 0,
-					fpfn, lpfn, mode);
-	spin_unlock(&mgr->lock);
-
-	if (!r)
-		mem->start = node->start;
-
-	return r;
+	return drm_mm_node_allocated(&node->mm_nodes[0]);
 }
 
 /**
@@ -155,53 +106,54 @@ static int amdgpu_gtt_mgr_alloc(struct ttm_mem_type_manager *man,
  * @man: TTM memory type manager
  * @tbo: TTM BO we need this range for
  * @place: placement flags and restrictions
- * @mem: the resulting mem object
+ * @res: the resulting mem object
  *
  * Dummy, allocate the node but no space for it yet.
  */
-static int amdgpu_gtt_mgr_new(struct ttm_mem_type_manager *man,
+static int amdgpu_gtt_mgr_new(struct ttm_resource_manager *man,
 			      struct ttm_buffer_object *tbo,
 			      const struct ttm_place *place,
-			      struct ttm_mem_reg *mem)
+			      struct ttm_resource **res)
 {
-	struct amdgpu_gtt_mgr *mgr = man->priv;
-	struct drm_mm_node *node;
+	struct amdgpu_gtt_mgr *mgr = to_gtt_mgr(man);
+	uint32_t num_pages = PFN_UP(tbo->base.size);
+	struct ttm_range_mgr_node *node;
 	int r;
 
-	spin_lock(&mgr->lock);
-	if (atomic64_read(&mgr->available) < mem->num_pages) {
-		spin_unlock(&mgr->lock);
-		return 0;
-	}
-	atomic64_sub(mem->num_pages, &mgr->available);
-	spin_unlock(&mgr->lock);
+	node = kzalloc(struct_size(node, mm_nodes, 1), GFP_KERNEL);
+	if (!node)
+		return -ENOMEM;
 
-	node = kzalloc(sizeof(*node), GFP_KERNEL);
-	if (!node) {
-		r = -ENOMEM;
-		goto err_out;
+	ttm_resource_init(tbo, place, &node->base);
+	if (!(place->flags & TTM_PL_FLAG_TEMPORARY) &&
+	    ttm_resource_manager_usage(man) > man->size) {
+		r = -ENOSPC;
+		goto err_free;
 	}
 
-	node->start = AMDGPU_BO_INVALID_OFFSET;
-	node->size = mem->num_pages;
-	mem->mm_node = node;
-
-	if (place->fpfn || place->lpfn || place->flags & TTM_PL_FLAG_TOPDOWN) {
-		r = amdgpu_gtt_mgr_alloc(man, tbo, place, mem);
-		if (unlikely(r)) {
-			kfree(node);
-			mem->mm_node = NULL;
-			r = 0;
-			goto err_out;
-		}
+	if (place->lpfn) {
+		spin_lock(&mgr->lock);
+		r = drm_mm_insert_node_in_range(&mgr->mm, &node->mm_nodes[0],
+						num_pages, tbo->page_alignment,
+						0, place->fpfn, place->lpfn,
+						DRM_MM_INSERT_BEST);
+		spin_unlock(&mgr->lock);
+		if (unlikely(r))
+			goto err_free;
+
+		node->base.start = node->mm_nodes[0].start;
 	} else {
-		mem->start = node->start;
+		node->mm_nodes[0].start = 0;
+		node->mm_nodes[0].size = PFN_UP(node->base.size);
+		node->base.start = AMDGPU_BO_INVALID_OFFSET;
 	}
 
+	*res = &node->base;
 	return 0;
-err_out:
-	atomic64_add(mem->num_pages, &mgr->available);
 
+err_free:
+	ttm_resource_fini(man, &node->base);
+	kfree(node);
 	return r;
 }
 
@@ -209,43 +161,81 @@ err_out:
  * amdgpu_gtt_mgr_del - free ranges
  *
  * @man: TTM memory type manager
- * @tbo: TTM BO we need this range for
- * @place: placement flags and restrictions
- * @mem: TTM memory object
+ * @res: TTM memory object
  *
  * Free the allocated GTT again.
  */
-static void amdgpu_gtt_mgr_del(struct ttm_mem_type_manager *man,
-			       struct ttm_mem_reg *mem)
+static void amdgpu_gtt_mgr_del(struct ttm_resource_manager *man,
+			       struct ttm_resource *res)
 {
-	struct amdgpu_gtt_mgr *mgr = man->priv;
-	struct drm_mm_node *node = mem->mm_node;
-
-	if (!node)
-		return;
+	struct ttm_range_mgr_node *node = to_ttm_range_mgr_node(res);
+	struct amdgpu_gtt_mgr *mgr = to_gtt_mgr(man);
 
 	spin_lock(&mgr->lock);
-	if (node->start != AMDGPU_BO_INVALID_OFFSET)
-		drm_mm_remove_node(node);
+	if (drm_mm_node_allocated(&node->mm_nodes[0]))
+		drm_mm_remove_node(&node->mm_nodes[0]);
 	spin_unlock(&mgr->lock);
-	atomic64_add(mem->num_pages, &mgr->available);
 
+	ttm_resource_fini(man, res);
 	kfree(node);
-	mem->mm_node = NULL;
 }
 
 /**
- * amdgpu_gtt_mgr_usage - return usage of GTT domain
+ * amdgpu_gtt_mgr_recover - re-init gart
  *
- * @man: TTM memory type manager
+ * @mgr: amdgpu_gtt_mgr pointer
+ *
+ * Re-init the gart for each known BO in the GTT.
+ */
+void amdgpu_gtt_mgr_recover(struct amdgpu_gtt_mgr *mgr)
+{
+	struct ttm_range_mgr_node *node;
+	struct drm_mm_node *mm_node;
+	struct amdgpu_device *adev;
+
+	adev = container_of(mgr, typeof(*adev), mman.gtt_mgr);
+	spin_lock(&mgr->lock);
+	drm_mm_for_each_node(mm_node, &mgr->mm) {
+		node = container_of(mm_node, typeof(*node), mm_nodes[0]);
+		amdgpu_ttm_recover_gart(node->base.bo);
+	}
+	spin_unlock(&mgr->lock);
+}
+
+/**
+ * amdgpu_gtt_mgr_intersects - test for intersection
+ *
+ * @man: Our manager object
+ * @res: The resource to test
+ * @place: The place for the new allocation
+ * @size: The size of the new allocation
  *
- * Return how many bytes are used in the GTT domain
+ * Simplified intersection test, only interesting if we need GART or not.
  */
-uint64_t amdgpu_gtt_mgr_usage(struct ttm_mem_type_manager *man)
+static bool amdgpu_gtt_mgr_intersects(struct ttm_resource_manager *man,
+				      struct ttm_resource *res,
+				      const struct ttm_place *place,
+				      size_t size)
 {
-	struct amdgpu_gtt_mgr *mgr = man->priv;
+	return !place->lpfn || amdgpu_gtt_mgr_has_gart_addr(res);
+}
 
-	return (u64)(man->size - atomic64_read(&mgr->available)) * PAGE_SIZE;
+/**
+ * amdgpu_gtt_mgr_compatible - test for compatibility
+ *
+ * @man: Our manager object
+ * @res: The resource to test
+ * @place: The place for the new allocation
+ * @size: The size of the new allocation
+ *
+ * Simplified compatibility test.
+ */
+static bool amdgpu_gtt_mgr_compatible(struct ttm_resource_manager *man,
+				      struct ttm_resource *res,
+				      const struct ttm_place *place,
+				      size_t size)
+{
+	return !place->lpfn || amdgpu_gtt_mgr_has_gart_addr(res);
 }
 
 /**
@@ -256,24 +246,77 @@ uint64_t amdgpu_gtt_mgr_usage(struct ttm_mem_type_manager *man)
  *
  * Dump the table content using printk.
  */
-static void amdgpu_gtt_mgr_debug(struct ttm_mem_type_manager *man,
+static void amdgpu_gtt_mgr_debug(struct ttm_resource_manager *man,
 				 struct drm_printer *printer)
 {
-	struct amdgpu_gtt_mgr *mgr = man->priv;
+	struct amdgpu_gtt_mgr *mgr = to_gtt_mgr(man);
 
 	spin_lock(&mgr->lock);
 	drm_mm_print(&mgr->mm, printer);
 	spin_unlock(&mgr->lock);
-
-	drm_printf(printer, "man size:%llu pages, gtt available:%llu pages, usage:%lluMB\n",
-		   man->size, (u64)atomic64_read(&mgr->available),
-		   amdgpu_gtt_mgr_usage(man) >> 20);
 }
 
-const struct ttm_mem_type_manager_func amdgpu_gtt_mgr_func = {
-	.init = amdgpu_gtt_mgr_init,
-	.takedown = amdgpu_gtt_mgr_fini,
-	.get_node = amdgpu_gtt_mgr_new,
-	.put_node = amdgpu_gtt_mgr_del,
+static const struct ttm_resource_manager_func amdgpu_gtt_mgr_func = {
+	.alloc = amdgpu_gtt_mgr_new,
+	.free = amdgpu_gtt_mgr_del,
+	.intersects = amdgpu_gtt_mgr_intersects,
+	.compatible = amdgpu_gtt_mgr_compatible,
 	.debug = amdgpu_gtt_mgr_debug
 };
+
+/**
+ * amdgpu_gtt_mgr_init - init GTT manager and DRM MM
+ *
+ * @adev: amdgpu_device pointer
+ * @gtt_size: maximum size of GTT
+ *
+ * Allocate and initialize the GTT manager.
+ */
+int amdgpu_gtt_mgr_init(struct amdgpu_device *adev, uint64_t gtt_size)
+{
+	struct amdgpu_gtt_mgr *mgr = &adev->mman.gtt_mgr;
+	struct ttm_resource_manager *man = &mgr->manager;
+	uint64_t start, size;
+
+	man->use_tt = true;
+	man->func = &amdgpu_gtt_mgr_func;
+
+	ttm_resource_manager_init(man, &adev->mman.bdev, gtt_size);
+
+	start = AMDGPU_GTT_MAX_TRANSFER_SIZE * AMDGPU_GTT_NUM_TRANSFER_WINDOWS;
+	size = (adev->gmc.gart_size >> PAGE_SHIFT) - start;
+	drm_mm_init(&mgr->mm, start, size);
+	spin_lock_init(&mgr->lock);
+
+	ttm_set_driver_manager(&adev->mman.bdev, TTM_PL_TT, &mgr->manager);
+	ttm_resource_manager_set_used(man, true);
+	return 0;
+}
+
+/**
+ * amdgpu_gtt_mgr_fini - free and destroy GTT manager
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Destroy and free the GTT manager, returns -EBUSY if ranges are still
+ * allocated inside it.
+ */
+void amdgpu_gtt_mgr_fini(struct amdgpu_device *adev)
+{
+	struct amdgpu_gtt_mgr *mgr = &adev->mman.gtt_mgr;
+	struct ttm_resource_manager *man = &mgr->manager;
+	int ret;
+
+	ttm_resource_manager_set_used(man, false);
+
+	ret = ttm_resource_manager_evict_all(&adev->mman.bdev, man);
+	if (ret)
+		return;
+
+	spin_lock(&mgr->lock);
+	drm_mm_takedown(&mgr->mm);
+	spin_unlock(&mgr->lock);
+
+	ttm_resource_manager_cleanup(man);
+	ttm_set_driver_manager(&adev->mman.bdev, TTM_PL_TT, NULL);
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c
new file mode 100644
index 000000000000..6e02fb9ac2f6
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c
@@ -0,0 +1,68 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "amdgpu_ras.h"
+#include <uapi/linux/kfd_ioctl.h>
+
+int amdgpu_hdp_ras_sw_init(struct amdgpu_device *adev)
+{
+	int err;
+	struct amdgpu_hdp_ras *ras;
+
+	if (!adev->hdp.ras)
+		return 0;
+
+	ras = adev->hdp.ras;
+	err = amdgpu_ras_register_ras_block(adev, &ras->ras_block);
+	if (err) {
+		dev_err(adev->dev, "Failed to register hdp ras block!\n");
+		return err;
+	}
+
+	strcpy(ras->ras_block.ras_comm.name, "hdp");
+	ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__HDP;
+	ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
+	adev->hdp.ras_if = &ras->ras_block.ras_comm;
+
+	/* hdp ras follows amdgpu_ras_block_late_init_default for late init */
+	return 0;
+}
+
+void amdgpu_hdp_generic_flush(struct amdgpu_device *adev,
+			      struct amdgpu_ring *ring)
+{
+	if (!ring || !ring->funcs->emit_wreg) {
+		WREG32((adev->rmmio_remap.reg_offset +
+			KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >>
+			       2,
+		       0);
+		if (adev->nbio.funcs->get_memsize)
+			adev->nbio.funcs->get_memsize(adev);
+	} else {
+		amdgpu_ring_emit_wreg(ring,
+				      (adev->rmmio_remap.reg_offset +
+				       KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >>
+					      2,
+				      0);
+	}
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.h
new file mode 100644
index 000000000000..4cfd932b7e91
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.h
@@ -0,0 +1,49 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __AMDGPU_HDP_H__
+#define __AMDGPU_HDP_H__
+#include "amdgpu_ras.h"
+
+struct amdgpu_hdp_ras {
+	struct amdgpu_ras_block_object ras_block;
+};
+
+struct amdgpu_hdp_funcs {
+	void (*flush_hdp)(struct amdgpu_device *adev, struct amdgpu_ring *ring);
+	void (*invalidate_hdp)(struct amdgpu_device *adev,
+			       struct amdgpu_ring *ring);
+	void (*update_clock_gating)(struct amdgpu_device *adev, bool enable);
+	void (*get_clock_gating_state)(struct amdgpu_device *adev, u64 *flags);
+	void (*init_registers)(struct amdgpu_device *adev);
+};
+
+struct amdgpu_hdp {
+	struct ras_common_if			*ras_if;
+	const struct amdgpu_hdp_funcs		*funcs;
+	struct amdgpu_hdp_ras	*ras;
+};
+
+int amdgpu_hdp_ras_sw_init(struct amdgpu_device *adev);
+void amdgpu_hdp_generic_flush(struct amdgpu_device *adev,
+			      struct amdgpu_ring *ring);
+#endif /* __AMDGPU_HDP_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c
new file mode 100644
index 000000000000..2c6a6b858112
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c
@@ -0,0 +1,248 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+/*
+ * Authors:
+ *    Christian König <christian.koenig@amd.com>
+ */
+
+/**
+ * DOC: MMU Notifier
+ *
+ * For coherent userptr handling registers an MMU notifier to inform the driver
+ * about updates on the page tables of a process.
+ *
+ * When somebody tries to invalidate the page tables we block the update until
+ * all operations on the pages in question are completed, then those pages are
+ * marked as accessed and also dirty if it wasn't a read only access.
+ *
+ * New command submissions using the userptrs in question are delayed until all
+ * page table invalidation are completed and we once more see a coherent process
+ * address space.
+ */
+
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include <drm/drm.h>
+
+#include "amdgpu.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_hmm.h"
+
+#define MAX_WALK_BYTE	(2UL << 30)
+
+/**
+ * amdgpu_hmm_invalidate_gfx - callback to notify about mm change
+ *
+ * @mni: the range (mm) is about to update
+ * @range: details on the invalidation
+ * @cur_seq: Value to pass to mmu_interval_set_seq()
+ *
+ * Block for operations on BOs to finish and mark pages as accessed and
+ * potentially dirty.
+ */
+static bool amdgpu_hmm_invalidate_gfx(struct mmu_interval_notifier *mni,
+				      const struct mmu_notifier_range *range,
+				      unsigned long cur_seq)
+{
+	struct amdgpu_bo *bo = container_of(mni, struct amdgpu_bo, notifier);
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	long r;
+
+	if (!mmu_notifier_range_blockable(range))
+		return false;
+
+	mutex_lock(&adev->notifier_lock);
+
+	mmu_interval_set_seq(mni, cur_seq);
+
+	r = dma_resv_wait_timeout(bo->tbo.base.resv, DMA_RESV_USAGE_BOOKKEEP,
+				  false, MAX_SCHEDULE_TIMEOUT);
+	mutex_unlock(&adev->notifier_lock);
+	if (r <= 0)
+		DRM_ERROR("(%ld) failed to wait for user bo\n", r);
+	return true;
+}
+
+static const struct mmu_interval_notifier_ops amdgpu_hmm_gfx_ops = {
+	.invalidate = amdgpu_hmm_invalidate_gfx,
+};
+
+/**
+ * amdgpu_hmm_invalidate_hsa - callback to notify about mm change
+ *
+ * @mni: the range (mm) is about to update
+ * @range: details on the invalidation
+ * @cur_seq: Value to pass to mmu_interval_set_seq()
+ *
+ * We temporarily evict the BO attached to this range. This necessitates
+ * evicting all user-mode queues of the process.
+ */
+static bool amdgpu_hmm_invalidate_hsa(struct mmu_interval_notifier *mni,
+				      const struct mmu_notifier_range *range,
+				      unsigned long cur_seq)
+{
+	struct amdgpu_bo *bo = container_of(mni, struct amdgpu_bo, notifier);
+
+	if (!mmu_notifier_range_blockable(range))
+		return false;
+
+	amdgpu_amdkfd_evict_userptr(mni, cur_seq, bo->kfd_bo);
+
+	return true;
+}
+
+static const struct mmu_interval_notifier_ops amdgpu_hmm_hsa_ops = {
+	.invalidate = amdgpu_hmm_invalidate_hsa,
+};
+
+/**
+ * amdgpu_hmm_register - register a BO for notifier updates
+ *
+ * @bo: amdgpu buffer object
+ * @addr: userptr addr we should monitor
+ *
+ * Registers a mmu_notifier for the given BO at the specified address.
+ * Returns 0 on success, -ERRNO if anything goes wrong.
+ */
+int amdgpu_hmm_register(struct amdgpu_bo *bo, unsigned long addr)
+{
+	int r;
+
+	if (bo->kfd_bo)
+		r = mmu_interval_notifier_insert(&bo->notifier, current->mm,
+						    addr, amdgpu_bo_size(bo),
+						    &amdgpu_hmm_hsa_ops);
+	else
+		r = mmu_interval_notifier_insert(&bo->notifier, current->mm, addr,
+							amdgpu_bo_size(bo),
+							&amdgpu_hmm_gfx_ops);
+	if (r)
+		/*
+		 * Make sure amdgpu_hmm_unregister() doesn't call
+		 * mmu_interval_notifier_remove() when the notifier isn't properly
+		 * initialized.
+		 */
+		bo->notifier.mm = NULL;
+
+	return r;
+}
+
+/**
+ * amdgpu_hmm_unregister - unregister a BO for notifier updates
+ *
+ * @bo: amdgpu buffer object
+ *
+ * Remove any registration of mmu notifier updates from the buffer object.
+ */
+void amdgpu_hmm_unregister(struct amdgpu_bo *bo)
+{
+	if (!bo->notifier.mm)
+		return;
+	mmu_interval_notifier_remove(&bo->notifier);
+	bo->notifier.mm = NULL;
+}
+
+int amdgpu_hmm_range_get_pages(struct mmu_interval_notifier *notifier,
+			       uint64_t start, uint64_t npages, bool readonly,
+			       void *owner,
+			       struct hmm_range **phmm_range)
+{
+	struct hmm_range *hmm_range;
+	unsigned long end;
+	unsigned long timeout;
+	unsigned long *pfns;
+	int r = 0;
+
+	hmm_range = kzalloc(sizeof(*hmm_range), GFP_KERNEL);
+	if (unlikely(!hmm_range))
+		return -ENOMEM;
+
+	pfns = kvmalloc_array(npages, sizeof(*pfns), GFP_KERNEL);
+	if (unlikely(!pfns)) {
+		r = -ENOMEM;
+		goto out_free_range;
+	}
+
+	hmm_range->notifier = notifier;
+	hmm_range->default_flags = HMM_PFN_REQ_FAULT;
+	if (!readonly)
+		hmm_range->default_flags |= HMM_PFN_REQ_WRITE;
+	hmm_range->hmm_pfns = pfns;
+	hmm_range->start = start;
+	end = start + npages * PAGE_SIZE;
+	hmm_range->dev_private_owner = owner;
+
+	do {
+		hmm_range->end = min(hmm_range->start + MAX_WALK_BYTE, end);
+
+		pr_debug("hmm range: start = 0x%lx, end = 0x%lx",
+			hmm_range->start, hmm_range->end);
+
+		timeout = jiffies + msecs_to_jiffies(HMM_RANGE_DEFAULT_TIMEOUT);
+
+retry:
+		hmm_range->notifier_seq = mmu_interval_read_begin(notifier);
+		r = hmm_range_fault(hmm_range);
+		if (unlikely(r)) {
+			if (r == -EBUSY && !time_after(jiffies, timeout))
+				goto retry;
+			goto out_free_pfns;
+		}
+
+		if (hmm_range->end == end)
+			break;
+		hmm_range->hmm_pfns += MAX_WALK_BYTE >> PAGE_SHIFT;
+		hmm_range->start = hmm_range->end;
+	} while (hmm_range->end < end);
+
+	hmm_range->start = start;
+	hmm_range->hmm_pfns = pfns;
+
+	*phmm_range = hmm_range;
+
+	return 0;
+
+out_free_pfns:
+	kvfree(pfns);
+out_free_range:
+	kfree(hmm_range);
+
+	if (r == -EBUSY)
+		r = -EAGAIN;
+	return r;
+}
+
+bool amdgpu_hmm_range_get_pages_done(struct hmm_range *hmm_range)
+{
+	bool r;
+
+	r = mmu_interval_read_retry(hmm_range->notifier,
+				    hmm_range->notifier_seq);
+	kvfree(hmm_range->hmm_pfns);
+	kfree(hmm_range);
+
+	return r;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.h
new file mode 100644
index 000000000000..953e1d06de20
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.h
@@ -0,0 +1,53 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Christian König
+ */
+#ifndef __AMDGPU_MN_H__
+#define __AMDGPU_MN_H__
+
+#include <linux/types.h>
+#include <linux/hmm.h>
+#include <linux/rwsem.h>
+#include <linux/workqueue.h>
+#include <linux/interval_tree.h>
+#include <linux/mmu_notifier.h>
+
+int amdgpu_hmm_range_get_pages(struct mmu_interval_notifier *notifier,
+			       uint64_t start, uint64_t npages, bool readonly,
+			       void *owner,
+			       struct hmm_range **phmm_range);
+bool amdgpu_hmm_range_get_pages_done(struct hmm_range *hmm_range);
+
+#if defined(CONFIG_HMM_MIRROR)
+int amdgpu_hmm_register(struct amdgpu_bo *bo, unsigned long addr);
+void amdgpu_hmm_unregister(struct amdgpu_bo *bo);
+#else
+static inline int amdgpu_hmm_register(struct amdgpu_bo *bo, unsigned long addr)
+{
+	DRM_WARN_ONCE("HMM_MIRROR kernel config option is not enabled, "
+		      "add CONFIG_ZONE_DEVICE=y in config file to fix this\n");
+	return -ENODEV;
+}
+static inline void amdgpu_hmm_unregister(struct amdgpu_bo *bo) {}
+#endif
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c
index f2739995c335..9cb72f0c5277 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c
@@ -23,9 +23,9 @@
  * Authors: Dave Airlie
  *          Alex Deucher
  */
-#include <linux/export.h>
 
-#include <drm/drmP.h>
+#include <linux/pci.h>
+
 #include <drm/drm_edid.h>
 #include <drm/amdgpu_drm.h>
 #include "amdgpu.h"
@@ -39,7 +39,7 @@
 static int amdgpu_i2c_pre_xfer(struct i2c_adapter *i2c_adap)
 {
 	struct amdgpu_i2c_chan *i2c = i2c_get_adapdata(i2c_adap);
-	struct amdgpu_device *adev = i2c->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(i2c->dev);
 	struct amdgpu_i2c_bus_rec *rec = &i2c->rec;
 	uint32_t temp;
 
@@ -81,7 +81,7 @@ static int amdgpu_i2c_pre_xfer(struct i2c_adapter *i2c_adap)
 static void amdgpu_i2c_post_xfer(struct i2c_adapter *i2c_adap)
 {
 	struct amdgpu_i2c_chan *i2c = i2c_get_adapdata(i2c_adap);
-	struct amdgpu_device *adev = i2c->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(i2c->dev);
 	struct amdgpu_i2c_bus_rec *rec = &i2c->rec;
 	uint32_t temp;
 
@@ -100,7 +100,7 @@ static void amdgpu_i2c_post_xfer(struct i2c_adapter *i2c_adap)
 static int amdgpu_i2c_get_clock(void *i2c_priv)
 {
 	struct amdgpu_i2c_chan *i2c = i2c_priv;
-	struct amdgpu_device *adev = i2c->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(i2c->dev);
 	struct amdgpu_i2c_bus_rec *rec = &i2c->rec;
 	uint32_t val;
 
@@ -115,7 +115,7 @@ static int amdgpu_i2c_get_clock(void *i2c_priv)
 static int amdgpu_i2c_get_data(void *i2c_priv)
 {
 	struct amdgpu_i2c_chan *i2c = i2c_priv;
-	struct amdgpu_device *adev = i2c->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(i2c->dev);
 	struct amdgpu_i2c_bus_rec *rec = &i2c->rec;
 	uint32_t val;
 
@@ -129,7 +129,7 @@ static int amdgpu_i2c_get_data(void *i2c_priv)
 static void amdgpu_i2c_set_clock(void *i2c_priv, int clock)
 {
 	struct amdgpu_i2c_chan *i2c = i2c_priv;
-	struct amdgpu_device *adev = i2c->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(i2c->dev);
 	struct amdgpu_i2c_bus_rec *rec = &i2c->rec;
 	uint32_t val;
 
@@ -142,7 +142,7 @@ static void amdgpu_i2c_set_clock(void *i2c_priv, int clock)
 static void amdgpu_i2c_set_data(void *i2c_priv, int data)
 {
 	struct amdgpu_i2c_chan *i2c = i2c_priv;
-	struct amdgpu_device *adev = i2c->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(i2c->dev);
 	struct amdgpu_i2c_bus_rec *rec = &i2c->rec;
 	uint32_t val;
 
@@ -174,8 +174,7 @@ struct amdgpu_i2c_chan *amdgpu_i2c_create(struct drm_device *dev,
 
 	i2c->rec = *rec;
 	i2c->adapter.owner = THIS_MODULE;
-	i2c->adapter.class = I2C_CLASS_DDC;
-	i2c->adapter.dev.parent = &dev->pdev->dev;
+	i2c->adapter.dev.parent = dev->dev;
 	i2c->dev = dev;
 	i2c_set_adapdata(&i2c->adapter, i2c);
 	mutex_init(&i2c->mutex);
@@ -185,7 +184,7 @@ struct amdgpu_i2c_chan *amdgpu_i2c_create(struct drm_device *dev,
 		snprintf(i2c->adapter.name, sizeof(i2c->adapter.name),
 			 "AMDGPU i2c hw bus %s", name);
 		i2c->adapter.algo = &amdgpu_atombios_i2c_algo;
-		ret = i2c_add_adapter(&i2c->adapter);
+		ret = devm_i2c_add_adapter(dev->dev, &i2c->adapter);
 		if (ret)
 			goto out_free;
 	} else {
@@ -216,22 +215,23 @@ out_free:
 
 }
 
-void amdgpu_i2c_destroy(struct amdgpu_i2c_chan *i2c)
-{
-	if (!i2c)
-		return;
-	WARN_ON(i2c->has_aux);
-	i2c_del_adapter(&i2c->adapter);
-	kfree(i2c);
-}
-
-/* Add the default buses */
 void amdgpu_i2c_init(struct amdgpu_device *adev)
 {
-	if (amdgpu_hw_i2c)
-		DRM_INFO("hw_i2c forced on, you may experience display detection problems!\n");
-
-	amdgpu_atombios_i2c_init(adev);
+	if (!adev->is_atom_fw) {
+		if (!amdgpu_device_has_dc_support(adev)) {
+			amdgpu_atombios_i2c_init(adev);
+		} else {
+			switch (adev->asic_type) {
+			case CHIP_POLARIS10:
+			case CHIP_POLARIS11:
+			case CHIP_POLARIS12:
+				amdgpu_atombios_oem_i2c_init(adev, 0x97);
+				break;
+			default:
+				break;
+			}
+		}
+	}
 }
 
 /* remove all the buses */
@@ -239,28 +239,9 @@ void amdgpu_i2c_fini(struct amdgpu_device *adev)
 {
 	int i;
 
-	for (i = 0; i < AMDGPU_MAX_I2C_BUS; i++) {
-		if (adev->i2c_bus[i]) {
-			amdgpu_i2c_destroy(adev->i2c_bus[i]);
+	for (i = 0; i < AMDGPU_MAX_I2C_BUS; i++)
+		if (adev->i2c_bus[i])
 			adev->i2c_bus[i] = NULL;
-		}
-	}
-}
-
-/* Add additional buses */
-void amdgpu_i2c_add(struct amdgpu_device *adev,
-		    const struct amdgpu_i2c_bus_rec *rec,
-		    const char *name)
-{
-	struct drm_device *dev = adev->ddev;
-	int i;
-
-	for (i = 0; i < AMDGPU_MAX_I2C_BUS; i++) {
-		if (!adev->i2c_bus[i]) {
-			adev->i2c_bus[i] = amdgpu_i2c_create(dev, rec, name);
-			return;
-		}
-	}
 }
 
 /* looks up bus based on id */
@@ -279,7 +260,7 @@ amdgpu_i2c_lookup(struct amdgpu_device *adev,
 	return NULL;
 }
 
-static void amdgpu_i2c_get_byte(struct amdgpu_i2c_chan *i2c_bus,
+static int amdgpu_i2c_get_byte(struct amdgpu_i2c_chan *i2c_bus,
 				 u8 slave_addr,
 				 u8 addr,
 				 u8 *val)
@@ -304,16 +285,18 @@ static void amdgpu_i2c_get_byte(struct amdgpu_i2c_chan *i2c_bus,
 	out_buf[0] = addr;
 	out_buf[1] = 0;
 
-	if (i2c_transfer(&i2c_bus->adapter, msgs, 2) == 2) {
-		*val = in_buf[0];
-		DRM_DEBUG("val = 0x%02x\n", *val);
-	} else {
-		DRM_DEBUG("i2c 0x%02x 0x%02x read failed\n",
-			  addr, *val);
+	if (i2c_transfer(&i2c_bus->adapter, msgs, 2) != 2) {
+		DRM_DEBUG("i2c 0x%02x read failed\n", addr);
+		return -EIO;
 	}
+
+	*val = in_buf[0];
+	DRM_DEBUG("val = 0x%02x\n", *val);
+
+	return 0;
 }
 
-static void amdgpu_i2c_put_byte(struct amdgpu_i2c_chan *i2c_bus,
+static int amdgpu_i2c_put_byte(struct amdgpu_i2c_chan *i2c_bus,
 				 u8 slave_addr,
 				 u8 addr,
 				 u8 val)
@@ -329,16 +312,19 @@ static void amdgpu_i2c_put_byte(struct amdgpu_i2c_chan *i2c_bus,
 	out_buf[0] = addr;
 	out_buf[1] = val;
 
-	if (i2c_transfer(&i2c_bus->adapter, &msg, 1) != 1)
-		DRM_DEBUG("i2c 0x%02x 0x%02x write failed\n",
-			  addr, val);
+	if (i2c_transfer(&i2c_bus->adapter, &msg, 1) != 1) {
+		DRM_DEBUG("i2c 0x%02x 0x%02x write failed\n", addr, val);
+		return -EIO;
+	}
+
+	return 0;
 }
 
 /* ddc router switching */
 void
 amdgpu_i2c_router_select_ddc_port(const struct amdgpu_connector *amdgpu_connector)
 {
-	u8 val;
+	u8 val = 0;
 
 	if (!amdgpu_connector->router.ddc_valid)
 		return;
@@ -346,16 +332,18 @@ amdgpu_i2c_router_select_ddc_port(const struct amdgpu_connector *amdgpu_connecto
 	if (!amdgpu_connector->router_bus)
 		return;
 
-	amdgpu_i2c_get_byte(amdgpu_connector->router_bus,
+	if (amdgpu_i2c_get_byte(amdgpu_connector->router_bus,
 			    amdgpu_connector->router.i2c_addr,
-			    0x3, &val);
+			    0x3, &val))
+		return;
 	val &= ~amdgpu_connector->router.ddc_mux_control_pin;
 	amdgpu_i2c_put_byte(amdgpu_connector->router_bus,
 			    amdgpu_connector->router.i2c_addr,
 			    0x3, val);
-	amdgpu_i2c_get_byte(amdgpu_connector->router_bus,
+	if (amdgpu_i2c_get_byte(amdgpu_connector->router_bus,
 			    amdgpu_connector->router.i2c_addr,
-			    0x1, &val);
+			    0x1, &val))
+		return;
 	val &= ~amdgpu_connector->router.ddc_mux_control_pin;
 	val |= amdgpu_connector->router.ddc_mux_state;
 	amdgpu_i2c_put_byte(amdgpu_connector->router_bus,
@@ -375,16 +363,18 @@ amdgpu_i2c_router_select_cd_port(const struct amdgpu_connector *amdgpu_connector
 	if (!amdgpu_connector->router_bus)
 		return;
 
-	amdgpu_i2c_get_byte(amdgpu_connector->router_bus,
+	if (amdgpu_i2c_get_byte(amdgpu_connector->router_bus,
 			    amdgpu_connector->router.i2c_addr,
-			    0x3, &val);
+			    0x3, &val))
+		return;
 	val &= ~amdgpu_connector->router.cd_mux_control_pin;
 	amdgpu_i2c_put_byte(amdgpu_connector->router_bus,
 			    amdgpu_connector->router.i2c_addr,
 			    0x3, val);
-	amdgpu_i2c_get_byte(amdgpu_connector->router_bus,
+	if (amdgpu_i2c_get_byte(amdgpu_connector->router_bus,
 			    amdgpu_connector->router.i2c_addr,
-			    0x1, &val);
+			    0x1, &val))
+		return;
 	val &= ~amdgpu_connector->router.cd_mux_control_pin;
 	val |= amdgpu_connector->router.cd_mux_state;
 	amdgpu_i2c_put_byte(amdgpu_connector->router_bus,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.h
index 63c2ff7499e1..1d3d3806e0dd 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.h
@@ -30,9 +30,6 @@ struct amdgpu_i2c_chan *amdgpu_i2c_create(struct drm_device *dev,
 void amdgpu_i2c_destroy(struct amdgpu_i2c_chan *i2c);
 void amdgpu_i2c_init(struct amdgpu_device *adev);
 void amdgpu_i2c_fini(struct amdgpu_device *adev);
-void amdgpu_i2c_add(struct amdgpu_device *adev,
-		    const struct amdgpu_i2c_bus_rec *rec,
-		    const char *name);
 struct amdgpu_i2c_chan *
 amdgpu_i2c_lookup(struct amdgpu_device *adev,
 		  const struct amdgpu_i2c_bus_rec *i2c_bus);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
index 659997bfff30..7d9bcb72e8dd 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
@@ -28,12 +28,15 @@
  */
 #include <linux/seq_file.h>
 #include <linux/slab.h>
-#include <drm/drmP.h>
+
 #include <drm/amdgpu_drm.h>
+
 #include "amdgpu.h"
 #include "atom.h"
+#include "amdgpu_trace.h"
 
 #define AMDGPU_IB_TEST_TIMEOUT	msecs_to_jiffies(1000)
+#define AMDGPU_IB_TEST_GFX_XGMI_TIMEOUT	msecs_to_jiffies(2000)
 
 /*
  * IB
@@ -44,13 +47,14 @@
  * produce command buffers which are send to the kernel and
  * put in IBs for execution by the requested ring.
  */
-static int amdgpu_debugfs_sa_init(struct amdgpu_device *adev);
 
 /**
  * amdgpu_ib_get - request an IB (Indirect Buffer)
  *
- * @ring: ring index the IB is associated with
+ * @adev: amdgpu_device pointer
+ * @vm: amdgpu_vm pointer
  * @size: requested IB size
+ * @pool_type: IB pool type (delayed, immediate, direct)
  * @ib: IB object returned
  *
  * Request an IB (all asics).  IBs are allocated using the
@@ -58,19 +62,22 @@ static int amdgpu_debugfs_sa_init(struct amdgpu_device *adev);
  * Returns 0 on success, error on failure.
  */
 int amdgpu_ib_get(struct amdgpu_device *adev, struct amdgpu_vm *vm,
-		  unsigned size, struct amdgpu_ib *ib)
+		  unsigned int size, enum amdgpu_ib_pool_type pool_type,
+		  struct amdgpu_ib *ib)
 {
 	int r;
 
 	if (size) {
-		r = amdgpu_sa_bo_new(&adev->ring_tmp_bo,
-				      &ib->sa_bo, size, 256);
+		r = amdgpu_sa_bo_new(&adev->ib_pools[pool_type],
+				     &ib->sa_bo, size);
 		if (r) {
 			dev_err(adev->dev, "failed to get a new IB (%d)\n", r);
 			return r;
 		}
 
 		ib->ptr = amdgpu_sa_bo_cpu_addr(ib->sa_bo);
+		/* flush the cache before commit the IB */
+		ib->flags = AMDGPU_IB_FLAG_EMIT_MEM_SYNC;
 
 		if (!vm)
 			ib->gpu_addr = amdgpu_sa_bo_gpu_addr(ib->sa_bo);
@@ -82,24 +89,23 @@ int amdgpu_ib_get(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 /**
  * amdgpu_ib_free - free an IB (Indirect Buffer)
  *
- * @adev: amdgpu_device pointer
  * @ib: IB object to free
  * @f: the fence SA bo need wait on for the ib alloation
  *
  * Free an IB (all asics).
  */
-void amdgpu_ib_free(struct amdgpu_device *adev, struct amdgpu_ib *ib,
-		    struct dma_fence *f)
+void amdgpu_ib_free(struct amdgpu_ib *ib, struct dma_fence *f)
 {
-	amdgpu_sa_bo_free(adev, &ib->sa_bo, f);
+	amdgpu_sa_bo_free(&ib->sa_bo, f);
 }
 
 /**
  * amdgpu_ib_schedule - schedule an IB (Indirect Buffer) on the ring
  *
- * @adev: amdgpu_device pointer
+ * @ring: ring index the IB is associated with
  * @num_ibs: number of IBs to schedule
  * @ibs: IB objects to schedule
+ * @job: job to schedule
  * @f: fence created during this submission
  *
  * Schedule an IB on the associated ring (all asics).
@@ -115,22 +121,26 @@ void amdgpu_ib_free(struct amdgpu_device *adev, struct amdgpu_ib *ib,
  * a CONST_IB), it will be put on the ring prior to the DE IB.  Prior
  * to SI there was just a DE IB.
  */
-int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
+int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs,
 		       struct amdgpu_ib *ibs, struct amdgpu_job *job,
 		       struct dma_fence **f)
 {
 	struct amdgpu_device *adev = ring->adev;
 	struct amdgpu_ib *ib = &ibs[0];
 	struct dma_fence *tmp = NULL;
-	bool skip_preamble, need_ctx_switch;
-	unsigned patch_offset = ~0;
+	struct amdgpu_fence *af;
+	bool need_ctx_switch;
 	struct amdgpu_vm *vm;
 	uint64_t fence_ctx;
 	uint32_t status = 0, alloc_size;
-
-	unsigned i;
-	int r = 0;
+	unsigned int fence_flags = 0;
+	bool secure, init_shadow;
+	u64 shadow_va, csa_va, gds_va;
+	int vmid = AMDGPU_JOB_GET_VMID(job);
 	bool need_pipe_sync = false;
+	unsigned int cond_exec;
+	unsigned int i;
+	int r = 0;
 
 	if (num_ibs == 0)
 		return -EINVAL;
@@ -138,22 +148,44 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 	/* ring tests don't use a job */
 	if (job) {
 		vm = job->vm;
-		fence_ctx = job->fence_ctx;
+		fence_ctx = job->base.s_fence ?
+			job->base.s_fence->scheduled.context : 0;
+		shadow_va = job->shadow_va;
+		csa_va = job->csa_va;
+		gds_va = job->gds_va;
+		init_shadow = job->init_shadow;
+		af = &job->hw_fence;
+		/* Save the context of the job for reset handling.
+		 * The driver needs this so it can skip the ring
+		 * contents for guilty contexts.
+		 */
+		af->context = job->base.s_fence ? job->base.s_fence->finished.context : 0;
 	} else {
 		vm = NULL;
 		fence_ctx = 0;
+		shadow_va = 0;
+		csa_va = 0;
+		gds_va = 0;
+		init_shadow = false;
+		af = NULL;
 	}
 
-	if (!ring->ready) {
+	if (!ring->sched.ready) {
 		dev_err(adev->dev, "couldn't schedule ib on ring <%s>\n", ring->name);
 		return -EINVAL;
 	}
 
-	if (vm && !job->vm_id) {
+	if (vm && !job->vmid) {
 		dev_err(adev->dev, "VM IB without ID\n");
 		return -EINVAL;
 	}
 
+	if ((ib->flags & AMDGPU_IB_FLAGS_SECURE) &&
+	    (!ring->funcs->secure_submission_supported)) {
+		dev_err(adev->dev, "secure submissions not supported on ring <%s>\n", ring->name);
+		return -EINVAL;
+	}
+
 	alloc_size = ring->funcs->emit_frame_size + num_ibs *
 		ring->funcs->emit_ib_size;
 
@@ -163,13 +195,26 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 		return r;
 	}
 
+	need_ctx_switch = ring->current_ctx != fence_ctx;
 	if (ring->funcs->emit_pipeline_sync && job &&
-	    ((tmp = amdgpu_sync_get_fence(&job->sched_sync)) ||
-	     amdgpu_vm_need_pipeline_sync(ring, job))) {
+	    ((tmp = amdgpu_sync_get_fence(&job->explicit_sync)) ||
+	     need_ctx_switch || amdgpu_vm_need_pipeline_sync(ring, job))) {
+
 		need_pipe_sync = true;
+
+		if (tmp)
+			trace_amdgpu_ib_pipe_sync(job, tmp);
+
 		dma_fence_put(tmp);
 	}
 
+	if ((ib->flags & AMDGPU_IB_FLAG_EMIT_MEM_SYNC) && ring->funcs->emit_mem_sync)
+		ring->funcs->emit_mem_sync(ring);
+
+	if (ring->funcs->emit_wave_limit &&
+	    ring->hw_prio == AMDGPU_GFX_PIPE_PRIO_HIGH)
+		ring->funcs->emit_wave_limit(ring, true);
+
 	if (ring->funcs->insert_start)
 		ring->funcs->insert_start(ring);
 
@@ -181,57 +226,74 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 		}
 	}
 
+	amdgpu_ring_ib_begin(ring);
+
+	if (ring->funcs->emit_gfx_shadow)
+		amdgpu_ring_emit_gfx_shadow(ring, shadow_va, csa_va, gds_va,
+					    init_shadow, vmid);
+
 	if (ring->funcs->init_cond_exec)
-		patch_offset = amdgpu_ring_init_cond_exec(ring);
+		cond_exec = amdgpu_ring_init_cond_exec(ring,
+						       ring->cond_exe_gpu_addr);
 
-	if (ring->funcs->emit_hdp_flush
-#ifdef CONFIG_X86_64
-	    && !(adev->flags & AMD_IS_APU)
-#endif
-	   )
-		amdgpu_ring_emit_hdp_flush(ring);
+	amdgpu_device_flush_hdp(adev, ring);
+
+	if (need_ctx_switch)
+		status |= AMDGPU_HAVE_CTX_SWITCH;
 
-	skip_preamble = ring->current_ctx == fence_ctx;
-	need_ctx_switch = ring->current_ctx != fence_ctx;
 	if (job && ring->funcs->emit_cntxcntl) {
-		if (need_ctx_switch)
-			status |= AMDGPU_HAVE_CTX_SWITCH;
 		status |= job->preamble_status;
-
+		status |= job->preemption_status;
 		amdgpu_ring_emit_cntxcntl(ring, status);
 	}
 
+	/* Setup initial TMZiness and send it off.
+	 */
+	secure = false;
+	if (job && ring->funcs->emit_frame_cntl) {
+		secure = ib->flags & AMDGPU_IB_FLAGS_SECURE;
+		amdgpu_ring_emit_frame_cntl(ring, true, secure);
+	}
+
 	for (i = 0; i < num_ibs; ++i) {
 		ib = &ibs[i];
 
-		/* drop preamble IBs if we don't have a context switch */
-		if ((ib->flags & AMDGPU_IB_FLAG_PREAMBLE) &&
-			skip_preamble &&
-			!(status & AMDGPU_PREAMBLE_IB_PRESENT_FIRST) &&
-			!amdgpu_sriov_vf(adev)) /* for SRIOV preemption, Preamble CE ib must be inserted anyway */
-			continue;
+		if (job && ring->funcs->emit_frame_cntl) {
+			if (secure != !!(ib->flags & AMDGPU_IB_FLAGS_SECURE)) {
+				amdgpu_ring_emit_frame_cntl(ring, false, secure);
+				secure = !secure;
+				amdgpu_ring_emit_frame_cntl(ring, true, secure);
+			}
+		}
 
-		amdgpu_ring_emit_ib(ring, ib, job ? job->vm_id : 0,
-				    need_ctx_switch);
-		need_ctx_switch = false;
+		amdgpu_ring_emit_ib(ring, job, ib, status);
+		status &= ~AMDGPU_HAVE_CTX_SWITCH;
 	}
 
-	if (ring->funcs->emit_tmz)
-		amdgpu_ring_emit_tmz(ring, false);
+	if (job && ring->funcs->emit_frame_cntl)
+		amdgpu_ring_emit_frame_cntl(ring, false, secure);
 
-	if (ring->funcs->emit_hdp_invalidate
-#ifdef CONFIG_X86_64
-	    && !(adev->flags & AMD_IS_APU)
-#endif
-	   )
-		amdgpu_ring_emit_hdp_invalidate(ring);
+	amdgpu_device_invalidate_hdp(adev, ring);
+
+	if (ib->flags & AMDGPU_IB_FLAG_TC_WB_NOT_INVALIDATE)
+		fence_flags |= AMDGPU_FENCE_FLAG_TC_WB_ONLY;
+
+	/* wrap the last IB with fence */
+	if (job && job->uf_addr) {
+		amdgpu_ring_emit_fence(ring, job->uf_addr, job->uf_sequence,
+				       fence_flags | AMDGPU_FENCE_FLAG_64BIT);
+	}
 
-	r = amdgpu_fence_emit(ring, f);
+	if (ring->funcs->emit_gfx_shadow && ring->funcs->init_cond_exec) {
+		amdgpu_ring_emit_gfx_shadow(ring, 0, 0, 0, false, 0);
+		amdgpu_ring_init_cond_exec(ring, ring->cond_exe_gpu_addr);
+	}
+
+	r = amdgpu_fence_emit(ring, f, af, fence_flags);
 	if (r) {
 		dev_err(adev->dev, "failed to emit fence (%d)\n", r);
-		if (job && job->vm_id)
-			amdgpu_vm_reset_id(adev, ring->funcs->vmhub,
-					   job->vm_id);
+		if (job && job->vmid)
+			amdgpu_vmid_reset(adev, ring->vm_hub, job->vmid);
 		amdgpu_ring_undo(ring);
 		return r;
 	}
@@ -239,19 +301,27 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 	if (ring->funcs->insert_end)
 		ring->funcs->insert_end(ring);
 
-	/* wrap the last IB with fence */
-	if (job && job->uf_addr) {
-		amdgpu_ring_emit_fence(ring, job->uf_addr, job->uf_sequence,
-				       AMDGPU_FENCE_FLAG_64BIT);
-	}
-
-	if (patch_offset != ~0 && ring->funcs->patch_cond_exec)
-		amdgpu_ring_patch_cond_exec(ring, patch_offset);
+	amdgpu_ring_patch_cond_exec(ring, cond_exec);
 
 	ring->current_ctx = fence_ctx;
-	if (vm && ring->funcs->emit_switch_buffer)
+	if (job && ring->funcs->emit_switch_buffer)
 		amdgpu_ring_emit_switch_buffer(ring);
+
+	if (ring->funcs->emit_wave_limit &&
+	    ring->hw_prio == AMDGPU_GFX_PIPE_PRIO_HIGH)
+		ring->funcs->emit_wave_limit(ring, false);
+
+	/* Save the wptr associated with this fence.
+	 * This must be last for resets to work properly
+	 * as we need to save the wptr associated with this
+	 * fence so we know what rings contents to backup
+	 * after we reset the queue.
+	 */
+	amdgpu_fence_save_wptr(*f);
+
+	amdgpu_ring_ib_end(ring);
 	amdgpu_ring_commit(ring);
+
 	return 0;
 }
 
@@ -266,29 +336,26 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
  */
 int amdgpu_ib_pool_init(struct amdgpu_device *adev)
 {
-	int r;
+	int r, i;
 
-	if (adev->ib_pool_ready) {
+	if (adev->ib_pool_ready)
 		return 0;
-	}
-	r = amdgpu_sa_bo_manager_init(adev, &adev->ring_tmp_bo,
-				      AMDGPU_IB_POOL_SIZE*64*1024,
-				      AMDGPU_GPU_PAGE_SIZE,
-				      AMDGPU_GEM_DOMAIN_GTT);
-	if (r) {
-		return r;
-	}
 
-	r = amdgpu_sa_bo_manager_start(adev, &adev->ring_tmp_bo);
-	if (r) {
-		return r;
+	for (i = 0; i < AMDGPU_IB_POOL_MAX; i++) {
+		r = amdgpu_sa_bo_manager_init(adev, &adev->ib_pools[i],
+					      AMDGPU_IB_POOL_SIZE, 256,
+					      AMDGPU_GEM_DOMAIN_GTT);
+		if (r)
+			goto error;
 	}
-
 	adev->ib_pool_ready = true;
-	if (amdgpu_debugfs_sa_init(adev)) {
-		dev_err(adev->dev, "failed to register debugfs file for SA\n");
-	}
+
 	return 0;
+
+error:
+	while (i--)
+		amdgpu_sa_bo_manager_fini(adev, &adev->ib_pools[i]);
+	return r;
 }
 
 /**
@@ -301,11 +368,14 @@ int amdgpu_ib_pool_init(struct amdgpu_device *adev)
  */
 void amdgpu_ib_pool_fini(struct amdgpu_device *adev)
 {
-	if (adev->ib_pool_ready) {
-		amdgpu_sa_bo_manager_suspend(adev, &adev->ring_tmp_bo);
-		amdgpu_sa_bo_manager_fini(adev, &adev->ring_tmp_bo);
-		adev->ib_pool_ready = false;
-	}
+	int i;
+
+	if (!adev->ib_pool_ready)
+		return;
+
+	for (i = 0; i < AMDGPU_IB_POOL_MAX; i++)
+		amdgpu_sa_bo_manager_fini(adev, &adev->ib_pools[i]);
+	adev->ib_pool_ready = false;
 }
 
 /**
@@ -320,30 +390,74 @@ void amdgpu_ib_pool_fini(struct amdgpu_device *adev)
  */
 int amdgpu_ib_ring_tests(struct amdgpu_device *adev)
 {
-	unsigned i;
+	long tmo_gfx, tmo_mm;
 	int r, ret = 0;
+	unsigned int i;
+
+	tmo_mm = tmo_gfx = AMDGPU_IB_TEST_TIMEOUT;
+	if (amdgpu_sriov_vf(adev)) {
+		/* for MM engines in hypervisor side they are not scheduled together
+		 * with CP and SDMA engines, so even in exclusive mode MM engine could
+		 * still running on other VF thus the IB TEST TIMEOUT for MM engines
+		 * under SR-IOV should be set to a long time. 8 sec should be enough
+		 * for the MM comes back to this VF.
+		 */
+		tmo_mm = 8 * AMDGPU_IB_TEST_TIMEOUT;
+	}
+
+	if (amdgpu_sriov_runtime(adev)) {
+		/* for CP & SDMA engines since they are scheduled together so
+		 * need to make the timeout width enough to cover the time
+		 * cost waiting for it coming back under RUNTIME only
+		 */
+		tmo_gfx = 8 * AMDGPU_IB_TEST_TIMEOUT;
+	} else if (adev->gmc.xgmi.hive_id) {
+		tmo_gfx = AMDGPU_IB_TEST_GFX_XGMI_TIMEOUT;
+	}
 
-	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+	for (i = 0; i < adev->num_rings; ++i) {
 		struct amdgpu_ring *ring = adev->rings[i];
+		long tmo;
 
-		if (!ring || !ring->ready)
+		/* KIQ rings don't have an IB test because we never submit IBs
+		 * to them and they have no interrupt support.
+		 */
+		if (!ring->sched.ready || !ring->funcs->test_ib)
 			continue;
 
-		r = amdgpu_ring_test_ib(ring, AMDGPU_IB_TEST_TIMEOUT);
-		if (r) {
-			ring->ready = false;
-
-			if (ring == &adev->gfx.gfx_ring[0]) {
-				/* oh, oh, that's really bad */
-				DRM_ERROR("amdgpu: failed testing IB on GFX ring (%d).\n", r);
-				adev->accel_working = false;
-				return r;
-
-			} else {
-				/* still not good, but we can live with it */
-				DRM_ERROR("amdgpu: failed testing IB on ring %d (%d).\n", i, r);
-				ret = r;
-			}
+		if (adev->enable_mes &&
+		    ring->funcs->type == AMDGPU_RING_TYPE_KIQ)
+			continue;
+
+		/* MM engine need more time */
+		if (ring->funcs->type == AMDGPU_RING_TYPE_UVD ||
+			ring->funcs->type == AMDGPU_RING_TYPE_VCE ||
+			ring->funcs->type == AMDGPU_RING_TYPE_UVD_ENC ||
+			ring->funcs->type == AMDGPU_RING_TYPE_VCN_DEC ||
+			ring->funcs->type == AMDGPU_RING_TYPE_VCN_ENC ||
+			ring->funcs->type == AMDGPU_RING_TYPE_VCN_JPEG)
+			tmo = tmo_mm;
+		else
+			tmo = tmo_gfx;
+
+		r = amdgpu_ring_test_ib(ring, tmo);
+		if (!r) {
+			DRM_DEV_DEBUG(adev->dev, "ib test on %s succeeded\n",
+				      ring->name);
+			continue;
+		}
+
+		ring->sched.ready = false;
+		DRM_DEV_ERROR(adev->dev, "IB test failed on %s (%d).\n",
+			  ring->name, r);
+
+		if (ring == &adev->gfx.gfx_ring[0]) {
+			/* oh, oh, that's really bad */
+			adev->accel_working = false;
+			return r;
+
+		} else {
+			ret = r;
 		}
 	}
 	return ret;
@@ -354,29 +468,34 @@ int amdgpu_ib_ring_tests(struct amdgpu_device *adev)
  */
 #if defined(CONFIG_DEBUG_FS)
 
-static int amdgpu_debugfs_sa_info(struct seq_file *m, void *data)
+static int amdgpu_debugfs_sa_info_show(struct seq_file *m, void *unused)
 {
-	struct drm_info_node *node = (struct drm_info_node *) m->private;
-	struct drm_device *dev = node->minor->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = m->private;
 
-	amdgpu_sa_bo_dump_debug_info(&adev->ring_tmp_bo, m);
+	seq_puts(m, "--------------------- DELAYED ---------------------\n");
+	amdgpu_sa_bo_dump_debug_info(&adev->ib_pools[AMDGPU_IB_POOL_DELAYED],
+				     m);
+	seq_puts(m, "-------------------- IMMEDIATE --------------------\n");
+	amdgpu_sa_bo_dump_debug_info(&adev->ib_pools[AMDGPU_IB_POOL_IMMEDIATE],
+				     m);
+	seq_puts(m, "--------------------- DIRECT ----------------------\n");
+	amdgpu_sa_bo_dump_debug_info(&adev->ib_pools[AMDGPU_IB_POOL_DIRECT], m);
 
 	return 0;
-
 }
 
-static const struct drm_info_list amdgpu_debugfs_sa_list[] = {
-	{"amdgpu_sa_info", &amdgpu_debugfs_sa_info, 0, NULL},
-};
+DEFINE_SHOW_ATTRIBUTE(amdgpu_debugfs_sa_info);
 
 #endif
 
-static int amdgpu_debugfs_sa_init(struct amdgpu_device *adev)
+void amdgpu_debugfs_sa_init(struct amdgpu_device *adev)
 {
 #if defined(CONFIG_DEBUG_FS)
-	return amdgpu_debugfs_add_files(adev, amdgpu_debugfs_sa_list, 1);
-#else
-	return 0;
+	struct drm_minor *minor = adev_to_drm(adev)->primary;
+	struct dentry *root = minor->debugfs_root;
+
+	debugfs_create_file("amdgpu_sa_info", 0444, root, adev,
+			    &amdgpu_debugfs_sa_info_fops);
+
 #endif
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c
new file mode 100644
index 000000000000..3ef5bc95642c
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c
@@ -0,0 +1,642 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu_ids.h"
+
+#include <linux/idr.h>
+#include <linux/dma-fence-array.h>
+
+
+#include "amdgpu.h"
+#include "amdgpu_trace.h"
+
+/*
+ * PASID manager
+ *
+ * PASIDs are global address space identifiers that can be shared
+ * between the GPU, an IOMMU and the driver. VMs on different devices
+ * may use the same PASID if they share the same address
+ * space. Therefore PASIDs are allocated using a global IDA. VMs are
+ * looked up from the PASID per amdgpu_device.
+ */
+static DEFINE_IDA(amdgpu_pasid_ida);
+
+/* Helper to free pasid from a fence callback */
+struct amdgpu_pasid_cb {
+	struct dma_fence_cb cb;
+	u32 pasid;
+};
+
+/**
+ * amdgpu_pasid_alloc - Allocate a PASID
+ * @bits: Maximum width of the PASID in bits, must be at least 1
+ *
+ * Allocates a PASID of the given width while keeping smaller PASIDs
+ * available if possible.
+ *
+ * Returns a positive integer on success. Returns %-EINVAL if bits==0.
+ * Returns %-ENOSPC if no PASID was available. Returns %-ENOMEM on
+ * memory allocation failure.
+ */
+int amdgpu_pasid_alloc(unsigned int bits)
+{
+	int pasid = -EINVAL;
+
+	for (bits = min(bits, 31U); bits > 0; bits--) {
+		pasid = ida_alloc_range(&amdgpu_pasid_ida, 1U << (bits - 1),
+					(1U << bits) - 1, GFP_KERNEL);
+		if (pasid != -ENOSPC)
+			break;
+	}
+
+	if (pasid >= 0)
+		trace_amdgpu_pasid_allocated(pasid);
+
+	return pasid;
+}
+
+/**
+ * amdgpu_pasid_free - Free a PASID
+ * @pasid: PASID to free
+ */
+void amdgpu_pasid_free(u32 pasid)
+{
+	trace_amdgpu_pasid_freed(pasid);
+	ida_free(&amdgpu_pasid_ida, pasid);
+}
+
+static void amdgpu_pasid_free_cb(struct dma_fence *fence,
+				 struct dma_fence_cb *_cb)
+{
+	struct amdgpu_pasid_cb *cb =
+		container_of(_cb, struct amdgpu_pasid_cb, cb);
+
+	amdgpu_pasid_free(cb->pasid);
+	dma_fence_put(fence);
+	kfree(cb);
+}
+
+/**
+ * amdgpu_pasid_free_delayed - free pasid when fences signal
+ *
+ * @resv: reservation object with the fences to wait for
+ * @pasid: pasid to free
+ *
+ * Free the pasid only after all the fences in resv are signaled.
+ */
+void amdgpu_pasid_free_delayed(struct dma_resv *resv,
+			       u32 pasid)
+{
+	struct amdgpu_pasid_cb *cb;
+	struct dma_fence *fence;
+	int r;
+
+	r = dma_resv_get_singleton(resv, DMA_RESV_USAGE_BOOKKEEP, &fence);
+	if (r)
+		goto fallback;
+
+	if (!fence) {
+		amdgpu_pasid_free(pasid);
+		return;
+	}
+
+	cb = kmalloc(sizeof(*cb), GFP_KERNEL);
+	if (!cb) {
+		/* Last resort when we are OOM */
+		dma_fence_wait(fence, false);
+		dma_fence_put(fence);
+		amdgpu_pasid_free(pasid);
+	} else {
+		cb->pasid = pasid;
+		if (dma_fence_add_callback(fence, &cb->cb,
+					   amdgpu_pasid_free_cb))
+			amdgpu_pasid_free_cb(fence, &cb->cb);
+	}
+
+	return;
+
+fallback:
+	/* Not enough memory for the delayed delete, as last resort
+	 * block for all the fences to complete.
+	 */
+	dma_resv_wait_timeout(resv, DMA_RESV_USAGE_BOOKKEEP,
+			      false, MAX_SCHEDULE_TIMEOUT);
+	amdgpu_pasid_free(pasid);
+}
+
+/*
+ * VMID manager
+ *
+ * VMIDs are a per VMHUB identifier for page tables handling.
+ */
+
+/**
+ * amdgpu_vmid_had_gpu_reset - check if reset occured since last use
+ *
+ * @adev: amdgpu_device pointer
+ * @id: VMID structure
+ *
+ * Check if GPU reset occured since last use of the VMID.
+ */
+bool amdgpu_vmid_had_gpu_reset(struct amdgpu_device *adev,
+			       struct amdgpu_vmid *id)
+{
+	return id->current_gpu_reset_count !=
+		atomic_read(&adev->gpu_reset_counter);
+}
+
+/* Check if we need to switch to another set of resources */
+static bool amdgpu_vmid_gds_switch_needed(struct amdgpu_vmid *id,
+					  struct amdgpu_job *job)
+{
+	return id->gds_base != job->gds_base ||
+		id->gds_size != job->gds_size ||
+		id->gws_base != job->gws_base ||
+		id->gws_size != job->gws_size ||
+		id->oa_base != job->oa_base ||
+		id->oa_size != job->oa_size;
+}
+
+/* Check if the id is compatible with the job */
+static bool amdgpu_vmid_compatible(struct amdgpu_vmid *id,
+				   struct amdgpu_job *job)
+{
+	return  id->pd_gpu_addr == job->vm_pd_addr &&
+		!amdgpu_vmid_gds_switch_needed(id, job);
+}
+
+/**
+ * amdgpu_vmid_grab_idle - grab idle VMID
+ *
+ * @ring: ring we want to submit job to
+ * @idle: resulting idle VMID
+ * @fence: fence to wait for if no id could be grabbed
+ *
+ * Try to find an idle VMID, if none is idle add a fence to wait to the sync
+ * object. Returns -ENOMEM when we are out of memory.
+ */
+static int amdgpu_vmid_grab_idle(struct amdgpu_ring *ring,
+				 struct amdgpu_vmid **idle,
+				 struct dma_fence **fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned vmhub = ring->vm_hub;
+	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
+	struct dma_fence **fences;
+	unsigned i;
+
+	if (!dma_fence_is_signaled(ring->vmid_wait)) {
+		*fence = dma_fence_get(ring->vmid_wait);
+		return 0;
+	}
+
+	fences = kmalloc_array(id_mgr->num_ids, sizeof(void *), GFP_NOWAIT);
+	if (!fences)
+		return -ENOMEM;
+
+	/* Check if we have an idle VMID */
+	i = 0;
+	list_for_each_entry((*idle), &id_mgr->ids_lru, list) {
+		/* Don't use per engine and per process VMID at the same time */
+		struct amdgpu_ring *r = adev->vm_manager.concurrent_flush ?
+			NULL : ring;
+
+		fences[i] = amdgpu_sync_peek_fence(&(*idle)->active, r);
+		if (!fences[i])
+			break;
+		++i;
+	}
+
+	/* If we can't find a idle VMID to use, wait till one becomes available */
+	if (&(*idle)->list == &id_mgr->ids_lru) {
+		u64 fence_context = adev->vm_manager.fence_context + ring->idx;
+		unsigned seqno = ++adev->vm_manager.seqno[ring->idx];
+		struct dma_fence_array *array;
+		unsigned j;
+
+		*idle = NULL;
+		for (j = 0; j < i; ++j)
+			dma_fence_get(fences[j]);
+
+		array = dma_fence_array_create(i, fences, fence_context,
+					       seqno, true);
+		if (!array) {
+			for (j = 0; j < i; ++j)
+				dma_fence_put(fences[j]);
+			kfree(fences);
+			return -ENOMEM;
+		}
+
+		*fence = dma_fence_get(&array->base);
+		dma_fence_put(ring->vmid_wait);
+		ring->vmid_wait = &array->base;
+		return 0;
+	}
+	kfree(fences);
+
+	return 0;
+}
+
+/**
+ * amdgpu_vmid_grab_reserved - try to assign reserved VMID
+ *
+ * @vm: vm to allocate id for
+ * @ring: ring we want to submit job to
+ * @job: job who wants to use the VMID
+ * @id: resulting VMID
+ * @fence: fence to wait for if no id could be grabbed
+ *
+ * Try to assign a reserved VMID.
+ */
+static int amdgpu_vmid_grab_reserved(struct amdgpu_vm *vm,
+				     struct amdgpu_ring *ring,
+				     struct amdgpu_job *job,
+				     struct amdgpu_vmid **id,
+				     struct dma_fence **fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned vmhub = ring->vm_hub;
+	uint64_t fence_context = adev->fence_context + ring->idx;
+	bool needs_flush = vm->use_cpu_for_update;
+	uint64_t updates = amdgpu_vm_tlb_seq(vm);
+	int r;
+
+	*id = vm->reserved_vmid[vmhub];
+	if ((*id)->owner != vm->immediate.fence_context ||
+	    !amdgpu_vmid_compatible(*id, job) ||
+	    (*id)->flushed_updates < updates ||
+	    !(*id)->last_flush ||
+	    ((*id)->last_flush->context != fence_context &&
+	     !dma_fence_is_signaled((*id)->last_flush)))
+		needs_flush = true;
+
+	if ((*id)->owner != vm->immediate.fence_context ||
+	    (!adev->vm_manager.concurrent_flush && needs_flush)) {
+		struct dma_fence *tmp;
+
+		/* Don't use per engine and per process VMID at the
+		 * same time
+		 */
+		if (adev->vm_manager.concurrent_flush)
+			ring = NULL;
+
+		/* to prevent one context starved by another context */
+		(*id)->pd_gpu_addr = 0;
+		tmp = amdgpu_sync_peek_fence(&(*id)->active, ring);
+		if (tmp) {
+			*id = NULL;
+			*fence = dma_fence_get(tmp);
+			return 0;
+		}
+	}
+
+	/* Good we can use this VMID. Remember this submission as
+	* user of the VMID.
+	*/
+	r = amdgpu_sync_fence(&(*id)->active, &job->base.s_fence->finished,
+			      GFP_NOWAIT);
+	if (r)
+		return r;
+
+	job->vm_needs_flush = needs_flush;
+	job->spm_update_needed = true;
+	return 0;
+}
+
+/**
+ * amdgpu_vmid_grab_used - try to reuse a VMID
+ *
+ * @vm: vm to allocate id for
+ * @ring: ring we want to submit job to
+ * @job: job who wants to use the VMID
+ * @id: resulting VMID
+ *
+ * Try to reuse a VMID for this submission.
+ */
+static int amdgpu_vmid_grab_used(struct amdgpu_vm *vm,
+				 struct amdgpu_ring *ring,
+				 struct amdgpu_job *job,
+				 struct amdgpu_vmid **id)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned vmhub = ring->vm_hub;
+	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
+	uint64_t fence_context = adev->fence_context + ring->idx;
+	uint64_t updates = amdgpu_vm_tlb_seq(vm);
+	int r;
+
+	job->vm_needs_flush = vm->use_cpu_for_update;
+
+	/* Check if we can use a VMID already assigned to this VM */
+	list_for_each_entry_reverse((*id), &id_mgr->ids_lru, list) {
+		bool needs_flush = vm->use_cpu_for_update;
+
+		/* Check all the prerequisites to using this VMID */
+		if ((*id)->owner != vm->immediate.fence_context)
+			continue;
+
+		if (!amdgpu_vmid_compatible(*id, job))
+			continue;
+
+		if (!(*id)->last_flush ||
+		    ((*id)->last_flush->context != fence_context &&
+		     !dma_fence_is_signaled((*id)->last_flush)))
+			needs_flush = true;
+
+		if ((*id)->flushed_updates < updates)
+			needs_flush = true;
+
+		if (needs_flush && !adev->vm_manager.concurrent_flush)
+			continue;
+
+		/* Good, we can use this VMID. Remember this submission as
+		 * user of the VMID.
+		 */
+		r = amdgpu_sync_fence(&(*id)->active,
+				      &job->base.s_fence->finished,
+				      GFP_NOWAIT);
+		if (r)
+			return r;
+
+		job->vm_needs_flush |= needs_flush;
+		return 0;
+	}
+
+	*id = NULL;
+	return 0;
+}
+
+/**
+ * amdgpu_vmid_grab - allocate the next free VMID
+ *
+ * @vm: vm to allocate id for
+ * @ring: ring we want to submit job to
+ * @job: job who wants to use the VMID
+ * @fence: fence to wait for if no id could be grabbed
+ *
+ * Allocate an id for the vm, adding fences to the sync obj as necessary.
+ */
+int amdgpu_vmid_grab(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
+		     struct amdgpu_job *job, struct dma_fence **fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned vmhub = ring->vm_hub;
+	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
+	struct amdgpu_vmid *idle = NULL;
+	struct amdgpu_vmid *id = NULL;
+	int r = 0;
+
+	mutex_lock(&id_mgr->lock);
+	r = amdgpu_vmid_grab_idle(ring, &idle, fence);
+	if (r || !idle)
+		goto error;
+
+	if (amdgpu_vmid_uses_reserved(vm, vmhub)) {
+		r = amdgpu_vmid_grab_reserved(vm, ring, job, &id, fence);
+		if (r || !id)
+			goto error;
+	} else {
+		r = amdgpu_vmid_grab_used(vm, ring, job, &id);
+		if (r)
+			goto error;
+
+		if (!id) {
+			/* Still no ID to use? Then use the idle one found earlier */
+			id = idle;
+
+			/* Remember this submission as user of the VMID */
+			r = amdgpu_sync_fence(&id->active,
+					      &job->base.s_fence->finished,
+					      GFP_NOWAIT);
+			if (r)
+				goto error;
+
+			job->vm_needs_flush = true;
+		}
+
+		list_move_tail(&id->list, &id_mgr->ids_lru);
+	}
+
+	job->gds_switch_needed = amdgpu_vmid_gds_switch_needed(id, job);
+	if (job->vm_needs_flush) {
+		id->flushed_updates = amdgpu_vm_tlb_seq(vm);
+		dma_fence_put(id->last_flush);
+		id->last_flush = NULL;
+	}
+	job->vmid = id - id_mgr->ids;
+	job->pasid = vm->pasid;
+
+	id->gds_base = job->gds_base;
+	id->gds_size = job->gds_size;
+	id->gws_base = job->gws_base;
+	id->gws_size = job->gws_size;
+	id->oa_base = job->oa_base;
+	id->oa_size = job->oa_size;
+	id->pd_gpu_addr = job->vm_pd_addr;
+	id->owner = vm->immediate.fence_context;
+
+	trace_amdgpu_vm_grab_id(vm, ring, job);
+
+error:
+	mutex_unlock(&id_mgr->lock);
+	return r;
+}
+
+/*
+ * amdgpu_vmid_uses_reserved - check if a VM will use a reserved VMID
+ * @vm: the VM to check
+ * @vmhub: the VMHUB which will be used
+ *
+ * Returns: True if the VM will use a reserved VMID.
+ */
+bool amdgpu_vmid_uses_reserved(struct amdgpu_vm *vm, unsigned int vmhub)
+{
+	return vm->reserved_vmid[vmhub];
+}
+
+/*
+ * amdgpu_vmid_alloc_reserved - reserve a specific VMID for this vm
+ * @adev: amdgpu device structure
+ * @vm: the VM to reserve an ID for
+ * @vmhub: the VMHUB which should be used
+ *
+ * Mostly used to have a reserved VMID for debugging and SPM.
+ *
+ * Returns: 0 for success, -ENOENT if an ID is already reserved.
+ */
+int amdgpu_vmid_alloc_reserved(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			       unsigned vmhub)
+{
+	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
+	struct amdgpu_vmid *id;
+	int r = 0;
+
+	mutex_lock(&id_mgr->lock);
+	if (vm->reserved_vmid[vmhub])
+		goto unlock;
+	if (id_mgr->reserved_vmid) {
+		r = -ENOENT;
+		goto unlock;
+	}
+	/* Remove from normal round robin handling */
+	id = list_first_entry(&id_mgr->ids_lru, struct amdgpu_vmid, list);
+	list_del_init(&id->list);
+	vm->reserved_vmid[vmhub] = id;
+	id_mgr->reserved_vmid = true;
+	mutex_unlock(&id_mgr->lock);
+
+	return 0;
+unlock:
+	mutex_unlock(&id_mgr->lock);
+	return r;
+}
+
+/*
+ * amdgpu_vmid_free_reserved - free up a reserved VMID again
+ * @adev: amdgpu device structure
+ * @vm: the VM with the reserved ID
+ * @vmhub: the VMHUB which should be used
+ */
+void amdgpu_vmid_free_reserved(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			       unsigned vmhub)
+{
+	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
+
+	mutex_lock(&id_mgr->lock);
+	if (vm->reserved_vmid[vmhub]) {
+		list_add(&vm->reserved_vmid[vmhub]->list,
+			&id_mgr->ids_lru);
+		vm->reserved_vmid[vmhub] = NULL;
+		id_mgr->reserved_vmid = false;
+	}
+	mutex_unlock(&id_mgr->lock);
+}
+
+/**
+ * amdgpu_vmid_reset - reset VMID to zero
+ *
+ * @adev: amdgpu device structure
+ * @vmhub: vmhub type
+ * @vmid: vmid number to use
+ *
+ * Reset saved GDW, GWS and OA to force switch on next flush.
+ */
+void amdgpu_vmid_reset(struct amdgpu_device *adev, unsigned vmhub,
+		       unsigned vmid)
+{
+	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
+	struct amdgpu_vmid *id = &id_mgr->ids[vmid];
+
+	mutex_lock(&id_mgr->lock);
+	id->owner = 0;
+	id->gds_base = 0;
+	id->gds_size = 0;
+	id->gws_base = 0;
+	id->gws_size = 0;
+	id->oa_base = 0;
+	id->oa_size = 0;
+	mutex_unlock(&id_mgr->lock);
+}
+
+/**
+ * amdgpu_vmid_reset_all - reset VMID to zero
+ *
+ * @adev: amdgpu device structure
+ *
+ * Reset VMID to force flush on next use
+ */
+void amdgpu_vmid_reset_all(struct amdgpu_device *adev)
+{
+	unsigned i, j;
+
+	for (i = 0; i < AMDGPU_MAX_VMHUBS; ++i) {
+		struct amdgpu_vmid_mgr *id_mgr =
+			&adev->vm_manager.id_mgr[i];
+
+		for (j = 1; j < id_mgr->num_ids; ++j)
+			amdgpu_vmid_reset(adev, i, j);
+	}
+}
+
+/**
+ * amdgpu_vmid_mgr_init - init the VMID manager
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialize the VM manager structures
+ */
+void amdgpu_vmid_mgr_init(struct amdgpu_device *adev)
+{
+	unsigned i, j;
+
+	for (i = 0; i < AMDGPU_MAX_VMHUBS; ++i) {
+		struct amdgpu_vmid_mgr *id_mgr =
+			&adev->vm_manager.id_mgr[i];
+
+		mutex_init(&id_mgr->lock);
+		INIT_LIST_HEAD(&id_mgr->ids_lru);
+
+		/* for GC <10, SDMA uses MMHUB so use first_kfd_vmid for both GC and MM */
+		if (amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(10, 0, 0))
+			/* manage only VMIDs not used by KFD */
+			id_mgr->num_ids = adev->vm_manager.first_kfd_vmid;
+		else if (AMDGPU_IS_MMHUB0(i) ||
+			 AMDGPU_IS_MMHUB1(i))
+			id_mgr->num_ids = 16;
+		else
+			/* manage only VMIDs not used by KFD */
+			id_mgr->num_ids = adev->vm_manager.first_kfd_vmid;
+
+		/* skip over VMID 0, since it is the system VM */
+		for (j = 1; j < id_mgr->num_ids; ++j) {
+			amdgpu_vmid_reset(adev, i, j);
+			amdgpu_sync_create(&id_mgr->ids[j].active);
+			list_add_tail(&id_mgr->ids[j].list, &id_mgr->ids_lru);
+		}
+	}
+}
+
+/**
+ * amdgpu_vmid_mgr_fini - cleanup VM manager
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Cleanup the VM manager and free resources.
+ */
+void amdgpu_vmid_mgr_fini(struct amdgpu_device *adev)
+{
+	unsigned i, j;
+
+	for (i = 0; i < AMDGPU_MAX_VMHUBS; ++i) {
+		struct amdgpu_vmid_mgr *id_mgr =
+			&adev->vm_manager.id_mgr[i];
+
+		mutex_destroy(&id_mgr->lock);
+		for (j = 0; j < AMDGPU_NUM_VMID; ++j) {
+			struct amdgpu_vmid *id = &id_mgr->ids[j];
+
+			amdgpu_sync_free(&id->active);
+			dma_fence_put(id->last_flush);
+			dma_fence_put(id->pasid_mapping);
+		}
+	}
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.h
new file mode 100644
index 000000000000..b3649cd3af56
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.h
@@ -0,0 +1,94 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __AMDGPU_IDS_H__
+#define __AMDGPU_IDS_H__
+
+#include <linux/types.h>
+#include <linux/mutex.h>
+#include <linux/list.h>
+#include <linux/dma-fence.h>
+
+#include "amdgpu_sync.h"
+
+/* maximum number of VMIDs */
+#define AMDGPU_NUM_VMID	16
+
+struct amdgpu_device;
+struct amdgpu_vm;
+struct amdgpu_ring;
+struct amdgpu_sync;
+struct amdgpu_job;
+
+struct amdgpu_vmid {
+	struct list_head	list;
+	struct amdgpu_sync	active;
+	struct dma_fence	*last_flush;
+	uint64_t		owner;
+
+	uint64_t		pd_gpu_addr;
+	/* last flushed PD/PT update */
+	uint64_t		flushed_updates;
+
+	uint32_t                current_gpu_reset_count;
+
+	uint32_t		gds_base;
+	uint32_t		gds_size;
+	uint32_t		gws_base;
+	uint32_t		gws_size;
+	uint32_t		oa_base;
+	uint32_t		oa_size;
+
+	unsigned		pasid;
+	struct dma_fence	*pasid_mapping;
+};
+
+struct amdgpu_vmid_mgr {
+	struct mutex		lock;
+	unsigned		num_ids;
+	struct list_head	ids_lru;
+	struct amdgpu_vmid	ids[AMDGPU_NUM_VMID];
+	bool			reserved_vmid;
+};
+
+int amdgpu_pasid_alloc(unsigned int bits);
+void amdgpu_pasid_free(u32 pasid);
+void amdgpu_pasid_free_delayed(struct dma_resv *resv,
+			       u32 pasid);
+
+bool amdgpu_vmid_had_gpu_reset(struct amdgpu_device *adev,
+			       struct amdgpu_vmid *id);
+bool amdgpu_vmid_uses_reserved(struct amdgpu_vm *vm, unsigned int vmhub);
+int amdgpu_vmid_alloc_reserved(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			       unsigned vmhub);
+void amdgpu_vmid_free_reserved(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			       unsigned vmhub);
+int amdgpu_vmid_grab(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
+		     struct amdgpu_job *job, struct dma_fence **fence);
+void amdgpu_vmid_reset(struct amdgpu_device *adev, unsigned vmhub,
+		       unsigned vmid);
+void amdgpu_vmid_reset_all(struct amdgpu_device *adev);
+
+void amdgpu_vmid_mgr_init(struct amdgpu_device *adev);
+void amdgpu_vmid_mgr_fini(struct amdgpu_device *adev);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c
index 3ab4c65ecc8b..a6419246e9c2 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c
@@ -21,49 +21,26 @@
  *
  */
 
-#include <drm/drmP.h>
+#include <linux/dma-mapping.h>
+
 #include "amdgpu.h"
 #include "amdgpu_ih.h"
-#include "amdgpu_amdkfd.h"
-
-/**
- * amdgpu_ih_ring_alloc - allocate memory for the IH ring
- *
- * @adev: amdgpu_device pointer
- *
- * Allocate a ring buffer for the interrupt controller.
- * Returns 0 for success, errors for failure.
- */
-static int amdgpu_ih_ring_alloc(struct amdgpu_device *adev)
-{
-	int r;
-
-	/* Allocate ring buffer */
-	if (adev->irq.ih.ring_obj == NULL) {
-		r = amdgpu_bo_create_kernel(adev, adev->irq.ih.ring_size,
-					    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
-					    &adev->irq.ih.ring_obj,
-					    &adev->irq.ih.gpu_addr,
-					    (void **)&adev->irq.ih.ring);
-		if (r) {
-			DRM_ERROR("amdgpu: failed to create ih ring buffer (%d).\n", r);
-			return r;
-		}
-	}
-	return 0;
-}
+#include "amdgpu_reset.h"
 
 /**
  * amdgpu_ih_ring_init - initialize the IH state
  *
  * @adev: amdgpu_device pointer
+ * @ih: ih ring to initialize
+ * @ring_size: ring size to allocate
+ * @use_bus_addr: true when we can use dma_alloc_coherent
  *
  * Initializes the IH state and allocates a buffer
  * for the IH ring buffer.
  * Returns 0 for success, errors for failure.
  */
-int amdgpu_ih_ring_init(struct amdgpu_device *adev, unsigned ring_size,
-			bool use_bus_addr)
+int amdgpu_ih_ring_init(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih,
+			unsigned ring_size, bool use_bus_addr)
 {
 	u32 rb_bufsz;
 	int r;
@@ -71,122 +48,270 @@ int amdgpu_ih_ring_init(struct amdgpu_device *adev, unsigned ring_size,
 	/* Align ring size */
 	rb_bufsz = order_base_2(ring_size / 4);
 	ring_size = (1 << rb_bufsz) * 4;
-	adev->irq.ih.ring_size = ring_size;
-	adev->irq.ih.ptr_mask = adev->irq.ih.ring_size - 1;
-	adev->irq.ih.rptr = 0;
-	adev->irq.ih.use_bus_addr = use_bus_addr;
-
-	if (adev->irq.ih.use_bus_addr) {
-		if (!adev->irq.ih.ring) {
-			/* add 8 bytes for the rptr/wptr shadows and
-			 * add them to the end of the ring allocation.
-			 */
-			adev->irq.ih.ring = pci_alloc_consistent(adev->pdev,
-								 adev->irq.ih.ring_size + 8,
-								 &adev->irq.ih.rb_dma_addr);
-			if (adev->irq.ih.ring == NULL)
-				return -ENOMEM;
-			memset((void *)adev->irq.ih.ring, 0, adev->irq.ih.ring_size + 8);
-			adev->irq.ih.wptr_offs = (adev->irq.ih.ring_size / 4) + 0;
-			adev->irq.ih.rptr_offs = (adev->irq.ih.ring_size / 4) + 1;
-		}
-		return 0;
+	ih->ring_size = ring_size;
+	ih->ptr_mask = ih->ring_size - 1;
+	ih->rptr = 0;
+	ih->use_bus_addr = use_bus_addr;
+
+	if (use_bus_addr) {
+		dma_addr_t dma_addr;
+
+		if (ih->ring)
+			return 0;
+
+		/* add 8 bytes for the rptr/wptr shadows and
+		 * add them to the end of the ring allocation.
+		 */
+		ih->ring = dma_alloc_coherent(adev->dev, ih->ring_size + 8,
+					      &dma_addr, GFP_KERNEL);
+		if (ih->ring == NULL)
+			return -ENOMEM;
+
+		ih->gpu_addr = dma_addr;
+		ih->wptr_addr = dma_addr + ih->ring_size;
+		ih->wptr_cpu = &ih->ring[ih->ring_size / 4];
+		ih->rptr_addr = dma_addr + ih->ring_size + 4;
+		ih->rptr_cpu = &ih->ring[(ih->ring_size / 4) + 1];
 	} else {
-		r = amdgpu_wb_get(adev, &adev->irq.ih.wptr_offs);
+		unsigned wptr_offs, rptr_offs;
+
+		r = amdgpu_device_wb_get(adev, &wptr_offs);
+		if (r)
+			return r;
+
+		r = amdgpu_device_wb_get(adev, &rptr_offs);
 		if (r) {
-			dev_err(adev->dev, "(%d) ih wptr_offs wb alloc failed\n", r);
+			amdgpu_device_wb_free(adev, wptr_offs);
 			return r;
 		}
 
-		r = amdgpu_wb_get(adev, &adev->irq.ih.rptr_offs);
+		r = amdgpu_bo_create_kernel(adev, ih->ring_size, PAGE_SIZE,
+					    AMDGPU_GEM_DOMAIN_GTT,
+					    &ih->ring_obj, &ih->gpu_addr,
+					    (void **)&ih->ring);
 		if (r) {
-			amdgpu_wb_free(adev, adev->irq.ih.wptr_offs);
-			dev_err(adev->dev, "(%d) ih rptr_offs wb alloc failed\n", r);
+			amdgpu_device_wb_free(adev, rptr_offs);
+			amdgpu_device_wb_free(adev, wptr_offs);
 			return r;
 		}
 
-		return amdgpu_ih_ring_alloc(adev);
+		ih->wptr_addr = adev->wb.gpu_addr + wptr_offs * 4;
+		ih->wptr_cpu = &adev->wb.wb[wptr_offs];
+		ih->rptr_addr = adev->wb.gpu_addr + rptr_offs * 4;
+		ih->rptr_cpu = &adev->wb.wb[rptr_offs];
 	}
+
+	init_waitqueue_head(&ih->wait_process);
+	return 0;
 }
 
 /**
  * amdgpu_ih_ring_fini - tear down the IH state
  *
  * @adev: amdgpu_device pointer
+ * @ih: ih ring to tear down
  *
  * Tears down the IH state and frees buffer
  * used for the IH ring buffer.
  */
-void amdgpu_ih_ring_fini(struct amdgpu_device *adev)
+void amdgpu_ih_ring_fini(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih)
 {
-	if (adev->irq.ih.use_bus_addr) {
-		if (adev->irq.ih.ring) {
-			/* add 8 bytes for the rptr/wptr shadows and
-			 * add them to the end of the ring allocation.
-			 */
-			pci_free_consistent(adev->pdev, adev->irq.ih.ring_size + 8,
-					    (void *)adev->irq.ih.ring,
-					    adev->irq.ih.rb_dma_addr);
-			adev->irq.ih.ring = NULL;
-		}
+
+	if (!ih->ring)
+		return;
+
+	if (ih->use_bus_addr) {
+
+		/* add 8 bytes for the rptr/wptr shadows and
+		 * add them to the end of the ring allocation.
+		 */
+		dma_free_coherent(adev->dev, ih->ring_size + 8,
+				  (void *)ih->ring, ih->gpu_addr);
+		ih->ring = NULL;
 	} else {
-		amdgpu_bo_free_kernel(&adev->irq.ih.ring_obj,
-				      &adev->irq.ih.gpu_addr,
-				      (void **)&adev->irq.ih.ring);
-		amdgpu_wb_free(adev, adev->irq.ih.wptr_offs);
-		amdgpu_wb_free(adev, adev->irq.ih.rptr_offs);
+		amdgpu_bo_free_kernel(&ih->ring_obj, &ih->gpu_addr,
+				      (void **)&ih->ring);
+		amdgpu_device_wb_free(adev, (ih->wptr_addr - ih->gpu_addr) / 4);
+		amdgpu_device_wb_free(adev, (ih->rptr_addr - ih->gpu_addr) / 4);
+	}
+}
+
+/**
+ * amdgpu_ih_ring_write - write IV to the ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: ih ring to write to
+ * @iv: the iv to write
+ * @num_dw: size of the iv in dw
+ *
+ * Writes an IV to the ring buffer using the CPU and increment the wptr.
+ * Used for testing and delegating IVs to a software ring.
+ */
+void amdgpu_ih_ring_write(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih,
+			  const uint32_t *iv, unsigned int num_dw)
+{
+	uint32_t wptr = le32_to_cpu(*ih->wptr_cpu) >> 2;
+	unsigned int i;
+
+	for (i = 0; i < num_dw; ++i)
+	        ih->ring[wptr++] = cpu_to_le32(iv[i]);
+
+	wptr <<= 2;
+	wptr &= ih->ptr_mask;
+
+	/* Only commit the new wptr if we don't overflow */
+	if (wptr != READ_ONCE(ih->rptr)) {
+		wmb();
+		WRITE_ONCE(*ih->wptr_cpu, cpu_to_le32(wptr));
+	} else if (adev->irq.retry_cam_enabled) {
+		dev_warn_once(adev->dev, "IH soft ring buffer overflow 0x%X, 0x%X\n",
+			      wptr, ih->rptr);
 	}
 }
 
 /**
+ * amdgpu_ih_wait_on_checkpoint_process_ts - wait to process IVs up to checkpoint
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: ih ring to process
+ *
+ * Used to ensure ring has processed IVs up to the checkpoint write pointer.
+ */
+int amdgpu_ih_wait_on_checkpoint_process_ts(struct amdgpu_device *adev,
+					struct amdgpu_ih_ring *ih)
+{
+	uint32_t checkpoint_wptr;
+	uint64_t checkpoint_ts;
+	long timeout = HZ;
+
+	if (!ih->enabled || adev->shutdown)
+		return -ENODEV;
+
+	checkpoint_wptr = amdgpu_ih_get_wptr(adev, ih);
+	/* Order wptr with ring data. */
+	rmb();
+	checkpoint_ts = amdgpu_ih_decode_iv_ts(adev, ih, checkpoint_wptr, -1);
+
+	return wait_event_interruptible_timeout(ih->wait_process,
+		    amdgpu_ih_ts_after(checkpoint_ts, ih->processed_timestamp) ||
+		    ih->rptr == amdgpu_ih_get_wptr(adev, ih), timeout);
+}
+
+/**
  * amdgpu_ih_process - interrupt handler
  *
  * @adev: amdgpu_device pointer
+ * @ih: ih ring to process
  *
  * Interrupt hander (VI), walk the IH ring.
  * Returns irq process return code.
  */
-int amdgpu_ih_process(struct amdgpu_device *adev)
+int amdgpu_ih_process(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih)
 {
-	struct amdgpu_iv_entry entry;
+	unsigned int count;
 	u32 wptr;
 
-	if (!adev->irq.ih.enabled || adev->shutdown)
+	if (!ih->enabled || adev->shutdown)
 		return IRQ_NONE;
 
-	wptr = amdgpu_ih_get_wptr(adev);
+	wptr = amdgpu_ih_get_wptr(adev, ih);
 
 restart_ih:
-	/* is somebody else already processing irqs? */
-	if (atomic_xchg(&adev->irq.ih.lock, 1))
-		return IRQ_NONE;
-
-	DRM_DEBUG("%s: rptr %d, wptr %d\n", __func__, adev->irq.ih.rptr, wptr);
+	count  = AMDGPU_IH_MAX_NUM_IVS;
+	dev_dbg(adev->dev, "%s: rptr %d, wptr %d\n", __func__, ih->rptr, wptr);
 
 	/* Order reading of wptr vs. reading of IH ring data */
 	rmb();
 
-	while (adev->irq.ih.rptr != wptr) {
-		u32 ring_index = adev->irq.ih.rptr >> 2;
-
-		/* Before dispatching irq to IP blocks, send it to amdkfd */
-		amdgpu_amdkfd_interrupt(adev,
-				(const void *) &adev->irq.ih.ring[ring_index]);
+	while (ih->rptr != wptr && --count) {
+		amdgpu_irq_dispatch(adev, ih);
+		ih->rptr &= ih->ptr_mask;
+	}
 
-		entry.iv_entry = (const uint32_t *)
-			&adev->irq.ih.ring[ring_index];
-		amdgpu_ih_decode_iv(adev, &entry);
-		adev->irq.ih.rptr &= adev->irq.ih.ptr_mask;
+	if (!ih->overflow)
+		amdgpu_ih_set_rptr(adev, ih);
 
-		amdgpu_irq_dispatch(adev, &entry);
-	}
-	amdgpu_ih_set_rptr(adev);
-	atomic_set(&adev->irq.ih.lock, 0);
+	wake_up_all(&ih->wait_process);
 
 	/* make sure wptr hasn't changed while processing */
-	wptr = amdgpu_ih_get_wptr(adev);
-	if (wptr != adev->irq.ih.rptr)
-		goto restart_ih;
+	wptr = amdgpu_ih_get_wptr(adev, ih);
+	if (wptr != ih->rptr)
+		if (!ih->overflow)
+			goto restart_ih;
+
+	if (ih->overflow)
+		if (amdgpu_sriov_runtime(adev))
+			WARN_ONCE(!amdgpu_reset_domain_schedule(adev->reset_domain,
+				   &adev->virt.flr_work),
+				  "Failed to queue work! at %s",
+				  __func__);
 
 	return IRQ_HANDLED;
 }
+
+/**
+ * amdgpu_ih_decode_iv_helper - decode an interrupt vector
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: ih ring to process
+ * @entry: IV entry
+ *
+ * Decodes the interrupt vector at the current rptr
+ * position and also advance the position for Vega10
+ * and later GPUs.
+ */
+void amdgpu_ih_decode_iv_helper(struct amdgpu_device *adev,
+				struct amdgpu_ih_ring *ih,
+				struct amdgpu_iv_entry *entry)
+{
+	/* wptr/rptr are in bytes! */
+	u32 ring_index = ih->rptr >> 2;
+	uint32_t dw[8];
+
+	dw[0] = le32_to_cpu(ih->ring[ring_index + 0]);
+	dw[1] = le32_to_cpu(ih->ring[ring_index + 1]);
+	dw[2] = le32_to_cpu(ih->ring[ring_index + 2]);
+	dw[3] = le32_to_cpu(ih->ring[ring_index + 3]);
+	dw[4] = le32_to_cpu(ih->ring[ring_index + 4]);
+	dw[5] = le32_to_cpu(ih->ring[ring_index + 5]);
+	dw[6] = le32_to_cpu(ih->ring[ring_index + 6]);
+	dw[7] = le32_to_cpu(ih->ring[ring_index + 7]);
+
+	entry->client_id = dw[0] & 0xff;
+	entry->src_id = (dw[0] >> 8) & 0xff;
+	entry->ring_id = (dw[0] >> 16) & 0xff;
+	entry->vmid = (dw[0] >> 24) & 0xf;
+	entry->vmid_src = (dw[0] >> 31);
+	entry->timestamp = dw[1] | ((u64)(dw[2] & 0xffff) << 32);
+	entry->timestamp_src = dw[2] >> 31;
+	entry->pasid = dw[3] & 0xffff;
+	entry->node_id = (dw[3] >> 16) & 0xff;
+	entry->src_data[0] = dw[4];
+	entry->src_data[1] = dw[5];
+	entry->src_data[2] = dw[6];
+	entry->src_data[3] = dw[7];
+
+	/* wptr/rptr are in bytes! */
+	ih->rptr += 32;
+}
+
+uint64_t amdgpu_ih_decode_iv_ts_helper(struct amdgpu_ih_ring *ih, u32 rptr,
+				       signed int offset)
+{
+	uint32_t iv_size = 32;
+	uint32_t ring_index;
+	uint32_t dw1, dw2;
+
+	rptr += iv_size * offset;
+	ring_index = (rptr & ih->ptr_mask) >> 2;
+
+	dw1 = le32_to_cpu(ih->ring[ring_index + 1]);
+	dw2 = le32_to_cpu(ih->ring[ring_index + 2]);
+	return dw1 | ((u64)(dw2 & 0xffff) << 32);
+}
+
+const char *amdgpu_ih_ring_name(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih)
+{
+	return ih == &adev->irq.ih ? "ih" : ih == &adev->irq.ih_soft ? "sw ih" :
+	       ih == &adev->irq.ih1 ? "ih1" : ih == &adev->irq.ih2 ? "ih2" : "unknown";
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h
index 3de8e74e5b3a..f58b6be7fccc 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h
@@ -24,90 +24,95 @@
 #ifndef __AMDGPU_IH_H__
 #define __AMDGPU_IH_H__
 
+/* Maximum number of IVs processed at once */
+#define AMDGPU_IH_MAX_NUM_IVS	32
+
+#define IH_RING_SIZE	(256 * 1024)
+#define IH_SW_RING_SIZE	(16 * 1024)	/* enough for 512 CAM entries */
+
 struct amdgpu_device;
- /*
-  * vega10+ IH clients
- */
-enum amdgpu_ih_clientid
-{
-    AMDGPU_IH_CLIENTID_IH	    = 0x00,
-    AMDGPU_IH_CLIENTID_ACP	    = 0x01,
-    AMDGPU_IH_CLIENTID_ATHUB	    = 0x02,
-    AMDGPU_IH_CLIENTID_BIF	    = 0x03,
-    AMDGPU_IH_CLIENTID_DCE	    = 0x04,
-    AMDGPU_IH_CLIENTID_ISP	    = 0x05,
-    AMDGPU_IH_CLIENTID_PCIE0	    = 0x06,
-    AMDGPU_IH_CLIENTID_RLC	    = 0x07,
-    AMDGPU_IH_CLIENTID_SDMA0	    = 0x08,
-    AMDGPU_IH_CLIENTID_SDMA1	    = 0x09,
-    AMDGPU_IH_CLIENTID_SE0SH	    = 0x0a,
-    AMDGPU_IH_CLIENTID_SE1SH	    = 0x0b,
-    AMDGPU_IH_CLIENTID_SE2SH	    = 0x0c,
-    AMDGPU_IH_CLIENTID_SE3SH	    = 0x0d,
-    AMDGPU_IH_CLIENTID_SYSHUB	    = 0x0e,
-    AMDGPU_IH_CLIENTID_THM	    = 0x0f,
-    AMDGPU_IH_CLIENTID_UVD	    = 0x10,
-    AMDGPU_IH_CLIENTID_VCE0	    = 0x11,
-    AMDGPU_IH_CLIENTID_VMC	    = 0x12,
-    AMDGPU_IH_CLIENTID_XDMA	    = 0x13,
-    AMDGPU_IH_CLIENTID_GRBM_CP	    = 0x14,
-    AMDGPU_IH_CLIENTID_ATS	    = 0x15,
-    AMDGPU_IH_CLIENTID_ROM_SMUIO    = 0x16,
-    AMDGPU_IH_CLIENTID_DF	    = 0x17,
-    AMDGPU_IH_CLIENTID_VCE1	    = 0x18,
-    AMDGPU_IH_CLIENTID_PWR	    = 0x19,
-    AMDGPU_IH_CLIENTID_UTCL2	    = 0x1b,
-    AMDGPU_IH_CLIENTID_EA	    = 0x1c,
-    AMDGPU_IH_CLIENTID_UTCL2LOG	    = 0x1d,
-    AMDGPU_IH_CLIENTID_MP0	    = 0x1e,
-    AMDGPU_IH_CLIENTID_MP1	    = 0x1f,
-
-    AMDGPU_IH_CLIENTID_MAX,
-
-    AMDGPU_IH_CLIENTID_VCN	    = AMDGPU_IH_CLIENTID_UVD
-};
+struct amdgpu_iv_entry;
 
-#define AMDGPU_IH_CLIENTID_LEGACY 0
+struct amdgpu_ih_regs {
+	uint32_t ih_rb_base;
+	uint32_t ih_rb_base_hi;
+	uint32_t ih_rb_cntl;
+	uint32_t ih_rb_wptr;
+	uint32_t ih_rb_rptr;
+	uint32_t ih_doorbell_rptr;
+	uint32_t ih_rb_wptr_addr_lo;
+	uint32_t ih_rb_wptr_addr_hi;
+	uint32_t psp_reg_id;
+};
 
 /*
  * R6xx+ IH ring
  */
 struct amdgpu_ih_ring {
-	struct amdgpu_bo	*ring_obj;
-	volatile uint32_t	*ring;
-	unsigned		rptr;
 	unsigned		ring_size;
-	uint64_t		gpu_addr;
 	uint32_t		ptr_mask;
-	atomic_t		lock;
-	bool                    enabled;
-	unsigned		wptr_offs;
-	unsigned		rptr_offs;
 	u32			doorbell_index;
 	bool			use_doorbell;
 	bool			use_bus_addr;
-	dma_addr_t		rb_dma_addr; /* only used when use_bus_addr = true */
+
+	struct amdgpu_bo	*ring_obj;
+	uint32_t		*ring;
+	uint64_t		gpu_addr;
+
+	uint64_t		wptr_addr;
+	uint32_t		*wptr_cpu;
+
+	uint64_t		rptr_addr;
+	uint32_t		*rptr_cpu;
+
+	bool                    enabled;
+	unsigned		rptr;
+	struct amdgpu_ih_regs	ih_regs;
+
+	/* For waiting on IH processing at checkpoint. */
+	wait_queue_head_t wait_process;
+	uint64_t		processed_timestamp;
+	bool overflow;
 };
 
-#define AMDGPU_IH_SRC_DATA_MAX_SIZE_DW 4
-
-struct amdgpu_iv_entry {
-	unsigned client_id;
-	unsigned src_id;
-	unsigned ring_id;
-	unsigned vm_id;
-	unsigned vm_id_src;
-	uint64_t timestamp;
-	unsigned timestamp_src;
-	unsigned pas_id;
-	unsigned pasid_src;
-	unsigned src_data[AMDGPU_IH_SRC_DATA_MAX_SIZE_DW];
-	const uint32_t *iv_entry;
+/* return true if time stamp t2 is after t1 with 48bit wrap around */
+#define amdgpu_ih_ts_after(t1, t2) \
+		(((int64_t)((t2) << 16) - (int64_t)((t1) << 16)) > 0LL)
+
+#define amdgpu_ih_ts_after_or_equal(t1, t2) \
+		(((int64_t)((t2) << 16) - (int64_t)((t1) << 16)) >= 0LL)
+
+/* provided by the ih block */
+struct amdgpu_ih_funcs {
+	/* ring read/write ptr handling, called from interrupt context */
+	u32 (*get_wptr)(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih);
+	void (*decode_iv)(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih,
+			  struct amdgpu_iv_entry *entry);
+	uint64_t (*decode_iv_ts)(struct amdgpu_ih_ring *ih, u32 rptr,
+				 signed int offset);
+	void (*set_rptr)(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih);
 };
 
-int amdgpu_ih_ring_init(struct amdgpu_device *adev, unsigned ring_size,
-			bool use_bus_addr);
-void amdgpu_ih_ring_fini(struct amdgpu_device *adev);
-int amdgpu_ih_process(struct amdgpu_device *adev);
+#define amdgpu_ih_get_wptr(adev, ih) (adev)->irq.ih_funcs->get_wptr((adev), (ih))
+#define amdgpu_ih_decode_iv(adev, iv) \
+	(adev)->irq.ih_funcs->decode_iv((adev), (ih), (iv))
+#define amdgpu_ih_decode_iv_ts(adev, ih, rptr, offset) \
+	(WARN_ON_ONCE(!(adev)->irq.ih_funcs->decode_iv_ts) ? 0 : \
+	(adev)->irq.ih_funcs->decode_iv_ts((ih), (rptr), (offset)))
+#define amdgpu_ih_set_rptr(adev, ih) (adev)->irq.ih_funcs->set_rptr((adev), (ih))
 
+int amdgpu_ih_ring_init(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih,
+			unsigned ring_size, bool use_bus_addr);
+void amdgpu_ih_ring_fini(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih);
+void amdgpu_ih_ring_write(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih,
+			  const uint32_t *iv, unsigned int num_dw);
+int amdgpu_ih_wait_on_checkpoint_process_ts(struct amdgpu_device *adev,
+					    struct amdgpu_ih_ring *ih);
+int amdgpu_ih_process(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih);
+void amdgpu_ih_decode_iv_helper(struct amdgpu_device *adev,
+				struct amdgpu_ih_ring *ih,
+				struct amdgpu_iv_entry *entry);
+uint64_t amdgpu_ih_decode_iv_ts_helper(struct amdgpu_ih_ring *ih, u32 rptr,
+				       signed int offset);
+const char *amdgpu_ih_ring_name(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih);
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_imu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_imu.h
new file mode 100644
index 000000000000..484e936812e4
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_imu.h
@@ -0,0 +1,58 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_IMU_H__
+#define __AMDGPU_IMU_H__
+
+enum imu_work_mode {
+	DEBUG_MODE,
+	MISSION_MODE
+};
+
+struct amdgpu_imu_funcs {
+    int (*init_microcode)(struct amdgpu_device *adev);
+    int (*load_microcode)(struct amdgpu_device *adev);
+    void (*setup_imu)(struct amdgpu_device *adev);
+    int (*start_imu)(struct amdgpu_device *adev);
+    void (*program_rlc_ram)(struct amdgpu_device *adev);
+    int (*wait_for_reset_status)(struct amdgpu_device *adev);
+};
+
+struct imu_rlc_ram_golden {
+    u32 hwip;
+    u32 instance;
+    u32 segment;
+    u32 reg;
+    u32 data;
+    u32 addr_mask;
+};
+
+#define IMU_RLC_RAM_GOLDEN_VALUE(ip, inst, reg, data, addr_mask) \
+    { ip##_HWIP, inst, reg##_BASE_IDX, reg, data, addr_mask }
+
+struct amdgpu_imu {
+    const struct amdgpu_imu_funcs *funcs;
+    enum imu_work_mode mode;
+};
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ioc32.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ioc32.c
index 26482914dc4b..a1cbd7c3deb2 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ioc32.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ioc32.c
@@ -1,4 +1,4 @@
-/**
+/*
  * \file amdgpu_ioc32.c
  *
  * 32-bit ioctl compatibility routines for the AMDGPU DRM.
@@ -29,19 +29,17 @@
  */
 #include <linux/compat.h>
 
-#include <drm/drmP.h>
 #include <drm/amdgpu_drm.h>
+#include <drm/drm_ioctl.h>
+
 #include "amdgpu_drv.h"
 
 long amdgpu_kms_compat_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
 {
 	unsigned int nr = DRM_IOCTL_NR(cmd);
-	int ret;
 
 	if (nr < DRM_COMMAND_BASE)
 		return drm_compat_ioctl(filp, cmd, arg);
 
-	ret = amdgpu_drm_ioctl(filp, cmd, arg);
-
-	return ret;
+	return amdgpu_drm_ioctl(filp, cmd, arg);
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ip.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ip.c
new file mode 100644
index 000000000000..99e1cf4fc955
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ip.c
@@ -0,0 +1,96 @@
+/*
+ * Copyright 2025 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "amdgpu_ip.h"
+
+static int8_t amdgpu_logical_to_dev_inst(struct amdgpu_device *adev,
+					 enum amd_hw_ip_block_type block,
+					 int8_t inst)
+{
+	int8_t dev_inst;
+
+	switch (block) {
+	case GC_HWIP:
+	case SDMA0_HWIP:
+	/* Both JPEG and VCN as JPEG is only alias of VCN */
+	case VCN_HWIP:
+		dev_inst = adev->ip_map.dev_inst[block][inst];
+		break;
+	default:
+		/* For rest of the IPs, no look up required.
+		 * Assume 'logical instance == physical instance' for all configs. */
+		dev_inst = inst;
+		break;
+	}
+
+	return dev_inst;
+}
+
+static uint32_t amdgpu_logical_to_dev_mask(struct amdgpu_device *adev,
+					   enum amd_hw_ip_block_type block,
+					   uint32_t mask)
+{
+	uint32_t dev_mask = 0;
+	int8_t log_inst, dev_inst;
+
+	while (mask) {
+		log_inst = ffs(mask) - 1;
+		dev_inst = amdgpu_logical_to_dev_inst(adev, block, log_inst);
+		dev_mask |= (1 << dev_inst);
+		mask &= ~(1 << log_inst);
+	}
+
+	return dev_mask;
+}
+
+static void amdgpu_populate_ip_map(struct amdgpu_device *adev,
+				   enum amd_hw_ip_block_type ip_block,
+				   uint32_t inst_mask)
+{
+	int l = 0, i;
+
+	while (inst_mask) {
+		i = ffs(inst_mask) - 1;
+		adev->ip_map.dev_inst[ip_block][l++] = i;
+		inst_mask &= ~(1 << i);
+	}
+	for (; l < HWIP_MAX_INSTANCE; l++)
+		adev->ip_map.dev_inst[ip_block][l] = -1;
+}
+
+void amdgpu_ip_map_init(struct amdgpu_device *adev)
+{
+	u32 ip_map[][2] = {
+		{ GC_HWIP, adev->gfx.xcc_mask },
+		{ SDMA0_HWIP, adev->sdma.sdma_mask },
+		{ VCN_HWIP, adev->vcn.inst_mask },
+	};
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(ip_map); ++i)
+		amdgpu_populate_ip_map(adev, ip_map[i][0], ip_map[i][1]);
+
+	adev->ip_map.logical_to_dev_inst = amdgpu_logical_to_dev_inst;
+	adev->ip_map.logical_to_dev_mask = amdgpu_logical_to_dev_mask;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ip.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ip.h
new file mode 100644
index 000000000000..2490fd322aec
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ip.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2025 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_IP_H__
+#define __AMDGPU_IP_H__
+
+void amdgpu_ip_map_init(struct amdgpu_device *adev);
+
+#endif /* __AMDGPU_IP_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
index 538e5f27d120..8112ffc85995 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
@@ -25,77 +25,110 @@
  *          Alex Deucher
  *          Jerome Glisse
  */
+
+/**
+ * DOC: Interrupt Handling
+ *
+ * Interrupts generated within GPU hardware raise interrupt requests that are
+ * passed to amdgpu IRQ handler which is responsible for detecting source and
+ * type of the interrupt and dispatching matching handlers. If handling an
+ * interrupt requires calling kernel functions that may sleep processing is
+ * dispatched to work handlers.
+ *
+ * If MSI functionality is not disabled by module parameter then MSI
+ * support will be enabled.
+ *
+ * For GPU interrupt sources that may be driven by another driver, IRQ domain
+ * support is used (with mapping between virtual and hardware IRQs).
+ */
+
 #include <linux/irq.h>
-#include <drm/drmP.h>
-#include <drm/drm_crtc_helper.h>
+#include <linux/pci.h>
+
+#include <drm/drm_vblank.h>
 #include <drm/amdgpu_drm.h>
+#include <drm/drm_drv.h>
 #include "amdgpu.h"
 #include "amdgpu_ih.h"
 #include "atom.h"
 #include "amdgpu_connectors.h"
 #include "amdgpu_trace.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_ras.h"
 
 #include <linux/pm_runtime.h>
 
+#ifdef CONFIG_DRM_AMD_DC
+#include "amdgpu_dm_irq.h"
+#endif
+
 #define AMDGPU_WAIT_IDLE_TIMEOUT 200
 
-/*
- * Handle hotplug events outside the interrupt handler proper.
- */
-/**
- * amdgpu_hotplug_work_func - display hotplug work handler
- *
- * @work: work struct
- *
- * This is the hot plug event work handler (all asics).
- * The work gets scheduled from the irq handler if there
- * was a hot plug interrupt.  It walks the connector table
- * and calls the hotplug handler for each one, then sends
- * a drm hotplug event to alert userspace.
- */
-static void amdgpu_hotplug_work_func(struct work_struct *work)
-{
-	struct amdgpu_device *adev = container_of(work, struct amdgpu_device,
-						  hotplug_work);
-	struct drm_device *dev = adev->ddev;
-	struct drm_mode_config *mode_config = &dev->mode_config;
-	struct drm_connector *connector;
-
-	mutex_lock(&mode_config->mutex);
-	list_for_each_entry(connector, &mode_config->connector_list, head)
-		amdgpu_connector_hotplug(connector);
-	mutex_unlock(&mode_config->mutex);
-	/* Just fire off a uevent and let userspace tell us what to do */
-	drm_helper_hpd_irq_event(dev);
-}
+const char *soc15_ih_clientid_name[] = {
+	"IH",
+	"SDMA2 or ACP",
+	"ATHUB",
+	"BIF",
+	"SDMA3 or DCE",
+	"SDMA4 or ISP",
+	"VMC1 or PCIE0",
+	"RLC",
+	"SDMA0",
+	"SDMA1",
+	"SE0SH",
+	"SE1SH",
+	"SE2SH",
+	"SE3SH",
+	"VCN1 or UVD1",
+	"THM",
+	"VCN or UVD",
+	"SDMA5 or VCE0",
+	"VMC",
+	"SDMA6 or XDMA",
+	"GRBM_CP",
+	"ATS",
+	"ROM_SMUIO",
+	"DF",
+	"SDMA7 or VCE1",
+	"PWR",
+	"reserved",
+	"UTCL2",
+	"EA",
+	"UTCL2LOG",
+	"MP0",
+	"MP1"
+};
+
+const int node_id_to_phys_map[NODEID_MAX] = {
+	[AID0_NODEID] = 0,
+	[XCD0_NODEID] = 0,
+	[XCD1_NODEID] = 1,
+	[AID1_NODEID] = 1,
+	[XCD2_NODEID] = 2,
+	[XCD3_NODEID] = 3,
+	[AID2_NODEID] = 2,
+	[XCD4_NODEID] = 4,
+	[XCD5_NODEID] = 5,
+	[AID3_NODEID] = 3,
+	[XCD6_NODEID] = 6,
+	[XCD7_NODEID] = 7,
+};
 
 /**
- * amdgpu_irq_reset_work_func - execute gpu reset
+ * amdgpu_irq_disable_all - disable *all* interrupts
  *
- * @work: work struct
+ * @adev: amdgpu device pointer
  *
- * Execute scheduled gpu reset (cayman+).
- * This function is called when the irq handler
- * thinks we need a gpu reset.
+ * Disable all types of interrupts from all sources.
  */
-static void amdgpu_irq_reset_work_func(struct work_struct *work)
-{
-	struct amdgpu_device *adev = container_of(work, struct amdgpu_device,
-						  reset_work);
-
-	if (!amdgpu_sriov_vf(adev))
-		amdgpu_gpu_reset(adev);
-}
-
-/* Disable *all* interrupts */
-static void amdgpu_irq_disable_all(struct amdgpu_device *adev)
+void amdgpu_irq_disable_all(struct amdgpu_device *adev)
 {
 	unsigned long irqflags;
-	unsigned i, j, k;
+	unsigned int i, j, k;
 	int r;
 
 	spin_lock_irqsave(&adev->irq.lock, irqflags);
-	for (i = 0; i < AMDGPU_IH_CLIENTID_MAX; ++i) {
+	for (i = 0; i < AMDGPU_IRQ_CLIENTID_MAX; ++i) {
 		if (!adev->irq.client[i].sources)
 			continue;
 
@@ -106,12 +139,12 @@ static void amdgpu_irq_disable_all(struct amdgpu_device *adev)
 				continue;
 
 			for (k = 0; k < src->num_types; ++k) {
-				atomic_set(&src->enabled_types[k], 0);
 				r = src->funcs->set(adev, src, k,
 						    AMDGPU_IRQ_STATE_DISABLE);
 				if (r)
-					DRM_ERROR("error disabling interrupt (%d)\n",
-						  r);
+					dev_err(adev->dev,
+						"error disabling interrupt (%d)\n",
+						r);
 			}
 		}
 	}
@@ -119,86 +152,89 @@ static void amdgpu_irq_disable_all(struct amdgpu_device *adev)
 }
 
 /**
- * amdgpu_irq_preinstall - drm irq preinstall callback
+ * amdgpu_irq_handler - IRQ handler
  *
- * @dev: drm dev pointer
+ * @irq: IRQ number (unused)
+ * @arg: pointer to DRM device
  *
- * Gets the hw ready to enable irqs (all asics).
- * This function disables all interrupt sources on the GPU.
+ * IRQ handler for amdgpu driver (all ASICs).
+ *
+ * Returns:
+ * result of handling the IRQ, as defined by &irqreturn_t
  */
-void amdgpu_irq_preinstall(struct drm_device *dev)
+static irqreturn_t amdgpu_irq_handler(int irq, void *arg)
 {
-	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_device *dev = (struct drm_device *) arg;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	irqreturn_t ret;
+
+	ret = amdgpu_ih_process(adev, &adev->irq.ih);
+	if (ret == IRQ_HANDLED)
+		pm_runtime_mark_last_busy(dev->dev);
 
-	/* Disable *all* interrupts */
-	amdgpu_irq_disable_all(adev);
-	/* Clear bits */
-	amdgpu_ih_process(adev);
+	amdgpu_ras_interrupt_fatal_error_handler(adev);
+
+	return ret;
 }
 
 /**
- * amdgpu_irq_postinstall - drm irq preinstall callback
+ * amdgpu_irq_handle_ih1 - kick of processing for IH1
  *
- * @dev: drm dev pointer
+ * @work: work structure in struct amdgpu_irq
  *
- * Handles stuff to be done after enabling irqs (all asics).
- * Returns 0 on success.
+ * Kick of processing IH ring 1.
  */
-int amdgpu_irq_postinstall(struct drm_device *dev)
+static void amdgpu_irq_handle_ih1(struct work_struct *work)
 {
-	dev->max_vblank_count = 0x00ffffff;
-	return 0;
+	struct amdgpu_device *adev = container_of(work, struct amdgpu_device,
+						  irq.ih1_work);
+
+	amdgpu_ih_process(adev, &adev->irq.ih1);
 }
 
 /**
- * amdgpu_irq_uninstall - drm irq uninstall callback
+ * amdgpu_irq_handle_ih2 - kick of processing for IH2
  *
- * @dev: drm dev pointer
+ * @work: work structure in struct amdgpu_irq
  *
- * This function disables all interrupt sources on the GPU (all asics).
+ * Kick of processing IH ring 2.
  */
-void amdgpu_irq_uninstall(struct drm_device *dev)
+static void amdgpu_irq_handle_ih2(struct work_struct *work)
 {
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = container_of(work, struct amdgpu_device,
+						  irq.ih2_work);
 
-	if (adev == NULL) {
-		return;
-	}
-	amdgpu_irq_disable_all(adev);
+	amdgpu_ih_process(adev, &adev->irq.ih2);
 }
 
 /**
- * amdgpu_irq_handler - irq handler
+ * amdgpu_irq_handle_ih_soft - kick of processing for ih_soft
  *
- * @int irq, void *arg: args
+ * @work: work structure in struct amdgpu_irq
  *
- * This is the irq handler for the amdgpu driver (all asics).
+ * Kick of processing IH soft ring.
  */
-irqreturn_t amdgpu_irq_handler(int irq, void *arg)
+static void amdgpu_irq_handle_ih_soft(struct work_struct *work)
 {
-	struct drm_device *dev = (struct drm_device *) arg;
-	struct amdgpu_device *adev = dev->dev_private;
-	irqreturn_t ret;
+	struct amdgpu_device *adev = container_of(work, struct amdgpu_device,
+						  irq.ih_soft_work);
 
-	ret = amdgpu_ih_process(adev);
-	if (ret == IRQ_HANDLED)
-		pm_runtime_mark_last_busy(dev->dev);
-	return ret;
+	amdgpu_ih_process(adev, &adev->irq.ih_soft);
 }
 
 /**
- * amdgpu_msi_ok - asic specific msi checks
+ * amdgpu_msi_ok - check whether MSI functionality is enabled
  *
- * @adev: amdgpu device pointer
+ * @adev: amdgpu device pointer (unused)
+ *
+ * Checks whether MSI functionality has been disabled via module parameter
+ * (all ASICs).
  *
- * Handles asic specific MSI checks to determine if
- * MSIs should be enabled on a particular chip (all asics).
- * Returns true if MSIs should be enabled, false if MSIs
- * should not be enabled.
+ * Returns:
+ * *true* if MSIs are allowed to be enabled or *false* otherwise
  */
 static bool amdgpu_msi_ok(struct amdgpu_device *adev)
 {
-	/* force MSI on */
 	if (amdgpu_msi == 1)
 		return true;
 	else if (amdgpu_msi == 0)
@@ -207,77 +243,119 @@ static bool amdgpu_msi_ok(struct amdgpu_device *adev)
 	return true;
 }
 
+void amdgpu_restore_msix(struct amdgpu_device *adev)
+{
+	u16 ctrl;
+
+	pci_read_config_word(adev->pdev, adev->pdev->msix_cap + PCI_MSIX_FLAGS, &ctrl);
+	if (!(ctrl & PCI_MSIX_FLAGS_ENABLE))
+		return;
+
+	/* VF FLR */
+	ctrl &= ~PCI_MSIX_FLAGS_ENABLE;
+	pci_write_config_word(adev->pdev, adev->pdev->msix_cap + PCI_MSIX_FLAGS, ctrl);
+	ctrl |= PCI_MSIX_FLAGS_ENABLE;
+	pci_write_config_word(adev->pdev, adev->pdev->msix_cap + PCI_MSIX_FLAGS, ctrl);
+}
+
 /**
- * amdgpu_irq_init - init driver interrupt info
+ * amdgpu_irq_init - initialize interrupt handling
  *
  * @adev: amdgpu device pointer
  *
- * Sets up the work irq handlers, vblank init, MSIs, etc. (all asics).
- * Returns 0 for success, error for failure.
+ * Sets up work functions for hotplug and reset interrupts, enables MSI
+ * functionality, initializes vblank, hotplug and reset interrupt handling.
+ *
+ * Returns:
+ * 0 on success or error code on failure
  */
 int amdgpu_irq_init(struct amdgpu_device *adev)
 {
-	int r = 0;
+	unsigned int irq, flags;
+	int r;
 
 	spin_lock_init(&adev->irq.lock);
 
-	if (!adev->enable_virtual_display)
-		/* Disable vblank irqs aggressively for power-saving */
-		adev->ddev->vblank_disable_immediate = true;
+	/* Enable MSI if not disabled by module parameter */
+	adev->irq.msi_enabled = false;
+
+	if (!amdgpu_msi_ok(adev))
+		flags = PCI_IRQ_INTX;
+	else
+		flags = PCI_IRQ_ALL_TYPES;
 
-	r = drm_vblank_init(adev->ddev, adev->mode_info.num_crtc);
-	if (r) {
+	/* we only need one vector */
+	r = pci_alloc_irq_vectors(adev->pdev, 1, 1, flags);
+	if (r < 0) {
+		dev_err(adev->dev, "Failed to alloc msi vectors\n");
 		return r;
 	}
 
-	/* enable msi */
-	adev->irq.msi_enabled = false;
-
 	if (amdgpu_msi_ok(adev)) {
-		int ret = pci_enable_msi(adev->pdev);
-		if (!ret) {
-			adev->irq.msi_enabled = true;
-			dev_info(adev->dev, "amdgpu: using MSI.\n");
-		}
+		adev->irq.msi_enabled = true;
+		dev_dbg(adev->dev, "using MSI/MSI-X.\n");
 	}
 
-	INIT_WORK(&adev->hotplug_work, amdgpu_hotplug_work_func);
-	INIT_WORK(&adev->reset_work, amdgpu_irq_reset_work_func);
+	INIT_WORK(&adev->irq.ih1_work, amdgpu_irq_handle_ih1);
+	INIT_WORK(&adev->irq.ih2_work, amdgpu_irq_handle_ih2);
+	INIT_WORK(&adev->irq.ih_soft_work, amdgpu_irq_handle_ih_soft);
+
+	/* Use vector 0 for MSI-X. */
+	r = pci_irq_vector(adev->pdev, 0);
+	if (r < 0)
+		goto free_vectors;
+	irq = r;
+
+	/* PCI devices require shared interrupts. */
+	r = request_irq(irq, amdgpu_irq_handler, IRQF_SHARED, adev_to_drm(adev)->driver->name,
+			adev_to_drm(adev));
+	if (r)
+		goto free_vectors;
 
 	adev->irq.installed = true;
-	r = drm_irq_install(adev->ddev, adev->ddev->pdev->irq);
-	if (r) {
+	adev->irq.irq = irq;
+	adev_to_drm(adev)->max_vblank_count = 0x00ffffff;
+
+	dev_dbg(adev->dev, "amdgpu: irq initialized.\n");
+	return 0;
+
+free_vectors:
+	if (adev->irq.msi_enabled)
+		pci_free_irq_vectors(adev->pdev);
+
+	adev->irq.msi_enabled = false;
+	return r;
+}
+
+void amdgpu_irq_fini_hw(struct amdgpu_device *adev)
+{
+	if (adev->irq.installed) {
+		free_irq(adev->irq.irq, adev_to_drm(adev));
 		adev->irq.installed = false;
-		flush_work(&adev->hotplug_work);
-		cancel_work_sync(&adev->reset_work);
-		return r;
+		if (adev->irq.msi_enabled)
+			pci_free_irq_vectors(adev->pdev);
 	}
 
-	DRM_INFO("amdgpu: irq initialized.\n");
-	return 0;
+	amdgpu_ih_ring_fini(adev, &adev->irq.ih_soft);
+	amdgpu_ih_ring_fini(adev, &adev->irq.ih);
+	amdgpu_ih_ring_fini(adev, &adev->irq.ih1);
+	amdgpu_ih_ring_fini(adev, &adev->irq.ih2);
 }
 
 /**
- * amdgpu_irq_fini - tear down driver interrupt info
+ * amdgpu_irq_fini_sw - shut down interrupt handling
  *
  * @adev: amdgpu device pointer
  *
- * Tears down the work irq handlers, vblank handlers, MSIs, etc. (all asics).
+ * Tears down work functions for hotplug and reset interrupts, disables MSI
+ * functionality, shuts down vblank, hotplug and reset interrupt handling,
+ * turns off interrupts from all sources (all ASICs).
  */
-void amdgpu_irq_fini(struct amdgpu_device *adev)
+void amdgpu_irq_fini_sw(struct amdgpu_device *adev)
 {
-	unsigned i, j;
-
-	if (adev->irq.installed) {
-		drm_irq_uninstall(adev->ddev);
-		adev->irq.installed = false;
-		if (adev->irq.msi_enabled)
-			pci_disable_msi(adev->pdev);
-		flush_work(&adev->hotplug_work);
-		cancel_work_sync(&adev->reset_work);
-	}
+	unsigned int i, j;
 
-	for (i = 0; i < AMDGPU_IH_CLIENTID_MAX; ++i) {
+	for (i = 0; i < AMDGPU_IRQ_CLIENTID_MAX; ++i) {
 		if (!adev->irq.client[i].sources)
 			continue;
 
@@ -289,29 +367,30 @@ void amdgpu_irq_fini(struct amdgpu_device *adev)
 
 			kfree(src->enabled_types);
 			src->enabled_types = NULL;
-			if (src->data) {
-				kfree(src->data);
-				kfree(src);
-				adev->irq.client[i].sources[j] = NULL;
-			}
 		}
 		kfree(adev->irq.client[i].sources);
+		adev->irq.client[i].sources = NULL;
 	}
 }
 
 /**
- * amdgpu_irq_add_id - register irq source
+ * amdgpu_irq_add_id - register IRQ source
  *
  * @adev: amdgpu device pointer
- * @src_id: source id for this source
- * @source: irq source
+ * @client_id: client id
+ * @src_id: source id
+ * @source: IRQ source pointer
+ *
+ * Registers IRQ source on a client.
  *
+ * Returns:
+ * 0 on success or error code otherwise
  */
 int amdgpu_irq_add_id(struct amdgpu_device *adev,
-		      unsigned client_id, unsigned src_id,
+		      unsigned int client_id, unsigned int src_id,
 		      struct amdgpu_irq_src *source)
 {
-	if (client_id >= AMDGPU_IH_CLIENTID_MAX)
+	if (client_id >= AMDGPU_IRQ_CLIENTID_MAX)
 		return -EINVAL;
 
 	if (src_id >= AMDGPU_MAX_IRQ_SRC_ID)
@@ -348,65 +427,107 @@ int amdgpu_irq_add_id(struct amdgpu_device *adev,
 }
 
 /**
- * amdgpu_irq_dispatch - dispatch irq to IP blocks
+ * amdgpu_irq_dispatch - dispatch IRQ to IP blocks
  *
  * @adev: amdgpu device pointer
- * @entry: interrupt vector
+ * @ih: interrupt ring instance
  *
- * Dispatches the irq to the different IP blocks
+ * Dispatches IRQ to IP blocks.
  */
 void amdgpu_irq_dispatch(struct amdgpu_device *adev,
-			 struct amdgpu_iv_entry *entry)
+			 struct amdgpu_ih_ring *ih)
 {
-	unsigned client_id = entry->client_id;
-	unsigned src_id = entry->src_id;
+	u32 ring_index = ih->rptr >> 2;
+	struct amdgpu_iv_entry entry;
+	unsigned int client_id, src_id;
 	struct amdgpu_irq_src *src;
+	bool handled = false;
 	int r;
 
-	trace_amdgpu_iv(entry);
+	entry.ih = ih;
+	entry.iv_entry = (const uint32_t *)&ih->ring[ring_index];
 
-	if (client_id >= AMDGPU_IH_CLIENTID_MAX) {
-		DRM_DEBUG("Invalid client_id in IV: %d\n", client_id);
-		return;
-	}
+	/*
+	 * timestamp is not supported on some legacy SOCs (cik, cz, iceland,
+	 * si and tonga), so initialize timestamp and timestamp_src to 0
+	 */
+	entry.timestamp = 0;
+	entry.timestamp_src = 0;
 
-	if (src_id >= AMDGPU_MAX_IRQ_SRC_ID) {
-		DRM_DEBUG("Invalid src_id in IV: %d\n", src_id);
-		return;
-	}
+	amdgpu_ih_decode_iv(adev, &entry);
 
-	if (adev->irq.virq[src_id]) {
-		generic_handle_irq(irq_find_mapping(adev->irq.domain, src_id));
-	} else {
-		if (!adev->irq.client[client_id].sources) {
-			DRM_DEBUG("Unregistered interrupt client_id: %d src_id: %d\n",
-				  client_id, src_id);
-			return;
-		}
+	trace_amdgpu_iv(ih - &adev->irq.ih, &entry);
 
-		src = adev->irq.client[client_id].sources[src_id];
-		if (!src) {
-			DRM_DEBUG("Unhandled interrupt src_id: %d\n", src_id);
-			return;
-		}
+	client_id = entry.client_id;
+	src_id = entry.src_id;
+
+	if (client_id >= AMDGPU_IRQ_CLIENTID_MAX) {
+		dev_dbg(adev->dev, "Invalid client_id in IV: %d\n", client_id);
+
+	} else	if (src_id >= AMDGPU_MAX_IRQ_SRC_ID) {
+		dev_dbg(adev->dev, "Invalid src_id in IV: %d\n", src_id);
+
+	} else if (((client_id == AMDGPU_IRQ_CLIENTID_LEGACY) ||
+		    (client_id == SOC15_IH_CLIENTID_ISP)) &&
+		   adev->irq.virq[src_id]) {
+		generic_handle_domain_irq(adev->irq.domain, src_id);
+
+	} else if (!adev->irq.client[client_id].sources) {
+		dev_dbg(adev->dev,
+			"Unregistered interrupt client_id: %d src_id: %d\n",
+			client_id, src_id);
+
+	} else if ((src = adev->irq.client[client_id].sources[src_id])) {
+		r = src->funcs->process(adev, src, &entry);
+		if (r < 0)
+			dev_err(adev->dev, "error processing interrupt (%d)\n",
+				r);
+		else if (r)
+			handled = true;
 
-		r = src->funcs->process(adev, src, entry);
-		if (r)
-			DRM_ERROR("error processing interrupt (%d)\n", r);
+	} else {
+		dev_dbg(adev->dev,
+			"Unregistered interrupt src_id: %d of client_id:%d\n",
+			src_id, client_id);
 	}
+
+	/* Send it to amdkfd as well if it isn't already handled */
+	if (!handled)
+		amdgpu_amdkfd_interrupt(adev, entry.iv_entry);
+
+	if (amdgpu_ih_ts_after(ih->processed_timestamp, entry.timestamp))
+		ih->processed_timestamp = entry.timestamp;
+}
+
+/**
+ * amdgpu_irq_delegate - delegate IV to soft IH ring
+ *
+ * @adev: amdgpu device pointer
+ * @entry: IV entry
+ * @num_dw: size of IV
+ *
+ * Delegate the IV to the soft IH ring and schedule processing of it. Used
+ * if the hardware delegation to IH1 or IH2 doesn't work for some reason.
+ */
+void amdgpu_irq_delegate(struct amdgpu_device *adev,
+			 struct amdgpu_iv_entry *entry,
+			 unsigned int num_dw)
+{
+	amdgpu_ih_ring_write(adev, &adev->irq.ih_soft, entry->iv_entry, num_dw);
+	schedule_work(&adev->irq.ih_soft_work);
 }
 
 /**
- * amdgpu_irq_update - update hw interrupt state
+ * amdgpu_irq_update - update hardware interrupt state
  *
  * @adev: amdgpu device pointer
- * @src: interrupt src you want to enable
- * @type: type of interrupt you want to update
+ * @src: interrupt source pointer
+ * @type: type of interrupt
  *
- * Updates the interrupt state for a specific src (all asics).
+ * Updates interrupt state for the specific source (all ASICs).
  */
 int amdgpu_irq_update(struct amdgpu_device *adev,
-			     struct amdgpu_irq_src *src, unsigned type)
+			     struct amdgpu_irq_src *src, unsigned int type)
 {
 	unsigned long irqflags;
 	enum amdgpu_interrupt_state state;
@@ -414,8 +535,9 @@ int amdgpu_irq_update(struct amdgpu_device *adev,
 
 	spin_lock_irqsave(&adev->irq.lock, irqflags);
 
-	/* we need to determine after taking the lock, otherwise
-	   we might disable just enabled interrupts again */
+	/* We need to determine after taking the lock, otherwise
+	 * we might disable just enabled interrupts again
+	 */
 	if (amdgpu_irq_enabled(adev, src, type))
 		state = AMDGPU_IRQ_STATE_ENABLE;
 	else
@@ -426,18 +548,29 @@ int amdgpu_irq_update(struct amdgpu_device *adev,
 	return r;
 }
 
+/**
+ * amdgpu_irq_gpu_reset_resume_helper - update interrupt states on all sources
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Updates state of all types of interrupts on all sources on resume after
+ * reset.
+ */
 void amdgpu_irq_gpu_reset_resume_helper(struct amdgpu_device *adev)
 {
 	int i, j, k;
 
-	for (i = 0; i < AMDGPU_IH_CLIENTID_MAX; ++i) {
+	if (amdgpu_sriov_vf(adev) || amdgpu_passthrough(adev))
+		amdgpu_restore_msix(adev);
+
+	for (i = 0; i < AMDGPU_IRQ_CLIENTID_MAX; ++i) {
 		if (!adev->irq.client[i].sources)
 			continue;
 
 		for (j = 0; j < AMDGPU_MAX_IRQ_SRC_ID; ++j) {
 			struct amdgpu_irq_src *src = adev->irq.client[i].sources[j];
 
-			if (!src)
+			if (!src || !src->funcs || !src->funcs->set)
 				continue;
 			for (k = 0; k < src->num_types; k++)
 				amdgpu_irq_update(adev, src, k);
@@ -449,15 +582,18 @@ void amdgpu_irq_gpu_reset_resume_helper(struct amdgpu_device *adev)
  * amdgpu_irq_get - enable interrupt
  *
  * @adev: amdgpu device pointer
- * @src: interrupt src you want to enable
- * @type: type of interrupt you want to enable
+ * @src: interrupt source pointer
+ * @type: type of interrupt
  *
- * Enables the interrupt type for a specific src (all asics).
+ * Enables specified type of interrupt on the specified source (all ASICs).
+ *
+ * Returns:
+ * 0 on success or error code otherwise
  */
 int amdgpu_irq_get(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
-		   unsigned type)
+		   unsigned int type)
 {
-	if (!adev->ddev->irq_enabled)
+	if (!adev->irq.installed)
 		return -ENOENT;
 
 	if (type >= src->num_types)
@@ -476,15 +612,22 @@ int amdgpu_irq_get(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
  * amdgpu_irq_put - disable interrupt
  *
  * @adev: amdgpu device pointer
- * @src: interrupt src you want to disable
- * @type: type of interrupt you want to disable
+ * @src: interrupt source pointer
+ * @type: type of interrupt
+ *
+ * Enables specified type of interrupt on the specified source (all ASICs).
  *
- * Disables the interrupt type for a specific src (all asics).
+ * Returns:
+ * 0 on success or error code otherwise
  */
 int amdgpu_irq_put(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
-		   unsigned type)
+		   unsigned int type)
 {
-	if (!adev->ddev->irq_enabled)
+	/* When the threshold is reached,the interrupt source may not be enabled.return -EINVAL */
+	if (amdgpu_ras_is_rma(adev) && !amdgpu_irq_enabled(adev, src, type))
+		return -EINVAL;
+
+	if (!adev->irq.installed)
 		return -ENOENT;
 
 	if (type >= src->num_types)
@@ -493,6 +636,9 @@ int amdgpu_irq_put(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
 	if (!src->enabled_types || !src->funcs->set)
 		return -EINVAL;
 
+	if (WARN_ON(!amdgpu_irq_enabled(adev, src, type)))
+		return -EINVAL;
+
 	if (atomic_dec_and_test(&src->enabled_types[type]))
 		return amdgpu_irq_update(adev, src, type);
 
@@ -500,17 +646,22 @@ int amdgpu_irq_put(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
 }
 
 /**
- * amdgpu_irq_enabled - test if irq is enabled or not
+ * amdgpu_irq_enabled - check whether interrupt is enabled or not
  *
  * @adev: amdgpu device pointer
- * @idx: interrupt src you want to test
+ * @src: interrupt source pointer
+ * @type: type of interrupt
  *
- * Tests if the given interrupt source is enabled or not
+ * Checks whether the given type of interrupt is enabled on the given source.
+ *
+ * Returns:
+ * *true* if interrupt is enabled, *false* if interrupt is disabled or on
+ * invalid parameters
  */
 bool amdgpu_irq_enabled(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
-			unsigned type)
+			unsigned int type)
 {
-	if (!adev->ddev->irq_enabled)
+	if (!adev->irq.installed)
 		return false;
 
 	if (type >= src->num_types)
@@ -522,7 +673,7 @@ bool amdgpu_irq_enabled(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
 	return !!atomic_read(&src->enabled_types[type]);
 }
 
-/* gen irq */
+/* XXX: Generic IRQ handling */
 static void amdgpu_irq_mask(struct irq_data *irqd)
 {
 	/* XXX */
@@ -533,12 +684,26 @@ static void amdgpu_irq_unmask(struct irq_data *irqd)
 	/* XXX */
 }
 
+/* amdgpu hardware interrupt chip descriptor */
 static struct irq_chip amdgpu_irq_chip = {
 	.name = "amdgpu-ih",
 	.irq_mask = amdgpu_irq_mask,
 	.irq_unmask = amdgpu_irq_unmask,
 };
 
+/**
+ * amdgpu_irqdomain_map - create mapping between virtual and hardware IRQ numbers
+ *
+ * @d: amdgpu IRQ domain pointer (unused)
+ * @irq: virtual IRQ number
+ * @hwirq: hardware irq number
+ *
+ * Current implementation assigns simple interrupt handler to the given virtual
+ * IRQ.
+ *
+ * Returns:
+ * 0 on success or error code otherwise
+ */
 static int amdgpu_irqdomain_map(struct irq_domain *d,
 				unsigned int irq, irq_hw_number_t hwirq)
 {
@@ -550,24 +715,28 @@ static int amdgpu_irqdomain_map(struct irq_domain *d,
 	return 0;
 }
 
+/* Implementation of methods for amdgpu IRQ domain */
 static const struct irq_domain_ops amdgpu_hw_irqdomain_ops = {
 	.map = amdgpu_irqdomain_map,
 };
 
 /**
- * amdgpu_irq_add_domain - create a linear irq domain
+ * amdgpu_irq_add_domain - create a linear IRQ domain
  *
  * @adev: amdgpu device pointer
  *
- * Create an irq domain for GPU interrupt sources
+ * Creates an IRQ domain for GPU interrupt sources
  * that may be driven by another driver (e.g., ACP).
+ *
+ * Returns:
+ * 0 on success or error code otherwise
  */
 int amdgpu_irq_add_domain(struct amdgpu_device *adev)
 {
-	adev->irq.domain = irq_domain_add_linear(NULL, AMDGPU_MAX_IRQ_SRC_ID,
-						 &amdgpu_hw_irqdomain_ops, adev);
+	adev->irq.domain = irq_domain_create_linear(NULL, AMDGPU_MAX_IRQ_SRC_ID,
+						    &amdgpu_hw_irqdomain_ops, adev);
 	if (!adev->irq.domain) {
-		DRM_ERROR("GPU irq add domain failed\n");
+		dev_err(adev->dev, "GPU irq add domain failed\n");
 		return -ENODEV;
 	}
 
@@ -575,11 +744,11 @@ int amdgpu_irq_add_domain(struct amdgpu_device *adev)
 }
 
 /**
- * amdgpu_irq_remove_domain - remove the irq domain
+ * amdgpu_irq_remove_domain - remove the IRQ domain
  *
  * @adev: amdgpu device pointer
  *
- * Remove the irq domain for GPU interrupt sources
+ * Removes the IRQ domain for GPU interrupt sources
  * that may be driven by another driver (e.g., ACP).
  */
 void amdgpu_irq_remove_domain(struct amdgpu_device *adev)
@@ -591,18 +760,19 @@ void amdgpu_irq_remove_domain(struct amdgpu_device *adev)
 }
 
 /**
- * amdgpu_irq_create_mapping - create a mapping between a domain irq and a
- *                             Linux irq
+ * amdgpu_irq_create_mapping - create mapping between domain Linux IRQs
  *
  * @adev: amdgpu device pointer
  * @src_id: IH source id
  *
- * Create a mapping between a domain irq (GPU IH src id) and a Linux irq
+ * Creates mapping between a domain IRQ (GPU IH src id) and a Linux IRQ
  * Use this for components that generate a GPU interrupt, but are driven
  * by a different driver (e.g., ACP).
- * Returns the Linux irq.
+ *
+ * Returns:
+ * Linux IRQ
  */
-unsigned amdgpu_irq_create_mapping(struct amdgpu_device *adev, unsigned src_id)
+unsigned int amdgpu_irq_create_mapping(struct amdgpu_device *adev, unsigned int src_id)
 {
 	adev->irq.virq[src_id] = irq_create_mapping(adev->irq.domain, src_id);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h
index 0610cc4a9788..9f0417456abd 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h
@@ -25,24 +25,43 @@
 #define __AMDGPU_IRQ_H__
 
 #include <linux/irqdomain.h>
+#include "soc15_ih_clientid.h"
 #include "amdgpu_ih.h"
 
-#define AMDGPU_MAX_IRQ_SRC_ID	0x100
+#define AMDGPU_MAX_IRQ_SRC_ID		0x100
 #define AMDGPU_MAX_IRQ_CLIENT_ID	0x100
 
+#define AMDGPU_IRQ_CLIENTID_LEGACY	0
+#define AMDGPU_IRQ_CLIENTID_MAX		SOC15_IH_CLIENTID_MAX
+
+#define AMDGPU_IRQ_SRC_DATA_MAX_SIZE_DW	4
+
 struct amdgpu_device;
-struct amdgpu_iv_entry;
 
 enum amdgpu_interrupt_state {
 	AMDGPU_IRQ_STATE_DISABLE,
 	AMDGPU_IRQ_STATE_ENABLE,
 };
 
+struct amdgpu_iv_entry {
+	struct amdgpu_ih_ring *ih;
+	unsigned client_id;
+	unsigned src_id;
+	unsigned ring_id;
+	unsigned vmid;
+	unsigned vmid_src;
+	uint64_t timestamp;
+	unsigned timestamp_src;
+	unsigned pasid;
+	unsigned node_id;
+	unsigned src_data[AMDGPU_IRQ_SRC_DATA_MAX_SIZE_DW];
+	const uint32_t *iv_entry;
+};
+
 struct amdgpu_irq_src {
 	unsigned				num_types;
 	atomic_t				*enabled_types;
 	const struct amdgpu_irq_src_funcs	*funcs;
-	void *data;
 };
 
 struct amdgpu_irq_client {
@@ -61,35 +80,59 @@ struct amdgpu_irq_src_funcs {
 
 struct amdgpu_irq {
 	bool				installed;
+	unsigned int			irq;
 	spinlock_t			lock;
 	/* interrupt sources */
-	struct amdgpu_irq_client	client[AMDGPU_IH_CLIENTID_MAX];
+	struct amdgpu_irq_client	client[AMDGPU_IRQ_CLIENTID_MAX];
 
 	/* status, etc. */
 	bool				msi_enabled; /* msi enabled */
 
-	/* interrupt ring */
-	struct amdgpu_ih_ring		ih;
-	const struct amdgpu_ih_funcs	*ih_funcs;
+	/* interrupt rings */
+	struct amdgpu_ih_ring		ih, ih1, ih2, ih_soft;
+	const struct amdgpu_ih_funcs    *ih_funcs;
+	struct work_struct		ih1_work, ih2_work, ih_soft_work;
+	struct amdgpu_irq_src		self_irq;
 
 	/* gen irq stuff */
 	struct irq_domain		*domain; /* GPU irq controller domain */
 	unsigned			virq[AMDGPU_MAX_IRQ_SRC_ID];
 	uint32_t                        srbm_soft_reset;
+	u32                             retry_cam_doorbell_index;
+	bool                            retry_cam_enabled;
 };
 
-void amdgpu_irq_preinstall(struct drm_device *dev);
-int amdgpu_irq_postinstall(struct drm_device *dev);
-void amdgpu_irq_uninstall(struct drm_device *dev);
-irqreturn_t amdgpu_irq_handler(int irq, void *arg);
+enum interrupt_node_id_per_aid {
+	AID0_NODEID = 0,
+	XCD0_NODEID = 1,
+	XCD1_NODEID = 2,
+	AID1_NODEID = 4,
+	XCD2_NODEID = 5,
+	XCD3_NODEID = 6,
+	AID2_NODEID = 8,
+	XCD4_NODEID = 9,
+	XCD5_NODEID = 10,
+	AID3_NODEID = 12,
+	XCD6_NODEID = 13,
+	XCD7_NODEID = 14,
+	NODEID_MAX,
+};
+
+extern const int node_id_to_phys_map[NODEID_MAX];
+
+void amdgpu_irq_disable_all(struct amdgpu_device *adev);
 
 int amdgpu_irq_init(struct amdgpu_device *adev);
-void amdgpu_irq_fini(struct amdgpu_device *adev);
+void amdgpu_irq_fini_sw(struct amdgpu_device *adev);
+void amdgpu_irq_fini_hw(struct amdgpu_device *adev);
 int amdgpu_irq_add_id(struct amdgpu_device *adev,
 		      unsigned client_id, unsigned src_id,
 		      struct amdgpu_irq_src *source);
 void amdgpu_irq_dispatch(struct amdgpu_device *adev,
-			 struct amdgpu_iv_entry *entry);
+			 struct amdgpu_ih_ring *ih);
+void amdgpu_irq_delegate(struct amdgpu_device *adev,
+			 struct amdgpu_iv_entry *entry,
+			 unsigned int num_dw);
 int amdgpu_irq_update(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
 		      unsigned type);
 int amdgpu_irq_get(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
@@ -103,5 +146,6 @@ void amdgpu_irq_gpu_reset_resume_helper(struct amdgpu_device *adev);
 int amdgpu_irq_add_domain(struct amdgpu_device *adev);
 void amdgpu_irq_remove_domain(struct amdgpu_device *adev);
 unsigned amdgpu_irq_create_mapping(struct amdgpu_device *adev, unsigned src_id);
+void amdgpu_restore_msix(struct amdgpu_device *adev);
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.c
new file mode 100644
index 000000000000..9cddbf50442a
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.c
@@ -0,0 +1,343 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright (C) 2024 Advanced Micro Devices, Inc. All rights reserved.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+
+#include <linux/firmware.h>
+#include <linux/mfd/core.h>
+
+#include "amdgpu.h"
+#include "amdgpu_isp.h"
+#include "isp_v4_1_0.h"
+#include "isp_v4_1_1.h"
+
+#define ISP_MC_ADDR_ALIGN (1024 * 32)
+
+/**
+ * isp_hw_init - start and test isp block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ */
+static int isp_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_isp *isp = &adev->isp;
+
+	if (isp->funcs->hw_init != NULL)
+		return isp->funcs->hw_init(isp);
+
+	return -ENODEV;
+}
+
+/**
+ * isp_hw_fini - stop the hardware block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ */
+static int isp_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_isp *isp = &ip_block->adev->isp;
+
+	if (isp->funcs->hw_fini != NULL)
+		return isp->funcs->hw_fini(isp);
+
+	return -ENODEV;
+}
+
+static int isp_load_fw_by_psp(struct amdgpu_device *adev)
+{
+	const struct common_firmware_header *hdr;
+	char ucode_prefix[10];
+	int r = 0;
+
+	/* get isp fw binary name and path */
+	amdgpu_ucode_ip_version_decode(adev, ISP_HWIP, ucode_prefix,
+				       sizeof(ucode_prefix));
+
+	/* read isp fw */
+	r = amdgpu_ucode_request(adev, &adev->isp.fw, AMDGPU_UCODE_OPTIONAL,
+				"amdgpu/%s.bin", ucode_prefix);
+	if (r) {
+		amdgpu_ucode_release(&adev->isp.fw);
+		return r;
+	}
+
+	hdr = (const struct common_firmware_header *)adev->isp.fw->data;
+
+	adev->firmware.ucode[AMDGPU_UCODE_ID_ISP].ucode_id =
+		AMDGPU_UCODE_ID_ISP;
+	adev->firmware.ucode[AMDGPU_UCODE_ID_ISP].fw = adev->isp.fw;
+
+	adev->firmware.fw_size +=
+		ALIGN(le32_to_cpu(hdr->ucode_size_bytes), PAGE_SIZE);
+
+	return r;
+}
+
+static int isp_early_init(struct amdgpu_ip_block *ip_block)
+{
+
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_isp *isp = &adev->isp;
+
+	switch (amdgpu_ip_version(adev, ISP_HWIP, 0)) {
+	case IP_VERSION(4, 1, 0):
+		isp_v4_1_0_set_isp_funcs(isp);
+		break;
+	case IP_VERSION(4, 1, 1):
+		isp_v4_1_1_set_isp_funcs(isp);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	isp->adev = adev;
+	isp->parent = adev->dev;
+
+	if (isp_load_fw_by_psp(adev)) {
+		DRM_DEBUG_DRIVER("%s: isp fw load failed\n", __func__);
+		return -ENOENT;
+	}
+
+	return 0;
+}
+
+static bool isp_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	return true;
+}
+
+static int isp_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+				     enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int isp_set_powergating_state(struct amdgpu_ip_block *ip_block,
+				     enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static int is_valid_isp_device(struct device *isp_parent, struct device *amdgpu_dev)
+{
+	if (isp_parent != amdgpu_dev)
+		return -EINVAL;
+
+	return 0;
+}
+
+/**
+ * isp_user_buffer_alloc - create user buffer object (BO) for isp
+ *
+ * @dev: isp device handle
+ * @dmabuf: DMABUF handle for isp buffer allocated in system memory
+ * @buf_obj: GPU buffer object handle to initialize
+ * @buf_addr: GPU addr of the pinned BO to initialize
+ *
+ * Imports isp DMABUF to allocate and pin a user BO for isp internal use. It does
+ * GART alloc to generate GPU addr for BO to make it accessible through the
+ * GART aperture for ISP HW.
+ *
+ * This function is exported to allow the V4L2 isp device external to drm device
+ * to create and access the isp user BO.
+ *
+ * Returns:
+ * 0 on success, negative error code otherwise.
+ */
+int isp_user_buffer_alloc(struct device *dev, void *dmabuf,
+			  void **buf_obj, u64 *buf_addr)
+{
+	struct platform_device *ispdev = to_platform_device(dev);
+	const struct isp_platform_data *isp_pdata;
+	struct amdgpu_device *adev;
+	struct mfd_cell *mfd_cell;
+	struct amdgpu_bo *bo;
+	u64 gpu_addr;
+	int ret;
+
+	if (WARN_ON(!ispdev))
+		return -ENODEV;
+
+	if (WARN_ON(!buf_obj))
+		return -EINVAL;
+
+	if (WARN_ON(!buf_addr))
+		return -EINVAL;
+
+	mfd_cell = &ispdev->mfd_cell[0];
+	if (!mfd_cell)
+		return -ENODEV;
+
+	isp_pdata = mfd_cell->platform_data;
+	adev = isp_pdata->adev;
+
+	ret = is_valid_isp_device(ispdev->dev.parent, adev->dev);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_bo_create_isp_user(adev, dmabuf,
+					AMDGPU_GEM_DOMAIN_GTT, &bo, &gpu_addr);
+	if (ret) {
+		drm_err(&adev->ddev, "failed to alloc gart user buffer (%d)", ret);
+		return ret;
+	}
+
+	*buf_obj = (void *)bo;
+	*buf_addr = gpu_addr;
+
+	return 0;
+}
+EXPORT_SYMBOL(isp_user_buffer_alloc);
+
+/**
+ * isp_user_buffer_free - free isp user buffer object (BO)
+ *
+ * @buf_obj: amdgpu isp user BO to free
+ *
+ * unpin and unref BO for isp internal use.
+ *
+ * This function is exported to allow the V4L2 isp device
+ * external to drm device to free the isp user BO.
+ */
+void isp_user_buffer_free(void *buf_obj)
+{
+	amdgpu_bo_free_isp_user(buf_obj);
+}
+EXPORT_SYMBOL(isp_user_buffer_free);
+
+/**
+ * isp_kernel_buffer_alloc - create kernel buffer object (BO) for isp
+ *
+ * @dev: isp device handle
+ * @size: size for the new BO
+ * @buf_obj: GPU BO handle to initialize
+ * @gpu_addr: GPU addr of the pinned BO
+ * @cpu_addr: CPU address mapping of BO
+ *
+ * Allocates and pins a kernel BO for internal isp firmware use.
+ *
+ * This function is exported to allow the V4L2 isp device
+ * external to drm device to create and access the kernel BO.
+ *
+ * Returns:
+ * 0 on success, negative error code otherwise.
+ */
+int isp_kernel_buffer_alloc(struct device *dev, u64 size,
+			    void **buf_obj, u64 *gpu_addr, void **cpu_addr)
+{
+	struct platform_device *ispdev = to_platform_device(dev);
+	struct amdgpu_bo **bo = (struct amdgpu_bo **)buf_obj;
+	const struct isp_platform_data *isp_pdata;
+	struct amdgpu_device *adev;
+	struct mfd_cell *mfd_cell;
+	int ret;
+
+	if (WARN_ON(!ispdev))
+		return -ENODEV;
+
+	if (WARN_ON(!buf_obj))
+		return -EINVAL;
+
+	if (WARN_ON(!gpu_addr))
+		return -EINVAL;
+
+	if (WARN_ON(!cpu_addr))
+		return -EINVAL;
+
+	mfd_cell = &ispdev->mfd_cell[0];
+	if (!mfd_cell)
+		return -ENODEV;
+
+	isp_pdata = mfd_cell->platform_data;
+	adev = isp_pdata->adev;
+
+	ret = is_valid_isp_device(ispdev->dev.parent, adev->dev);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_bo_create_kernel(adev,
+				      size,
+				      ISP_MC_ADDR_ALIGN,
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      bo,
+				      gpu_addr,
+				      cpu_addr);
+	if (!cpu_addr || ret) {
+		drm_err(&adev->ddev, "failed to alloc gart kernel buffer (%d)", ret);
+		return ret;
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL(isp_kernel_buffer_alloc);
+
+/**
+ * isp_kernel_buffer_free - free isp kernel buffer object (BO)
+ *
+ * @buf_obj: amdgpu isp user BO to free
+ * @gpu_addr: GPU addr of isp kernel BO
+ * @cpu_addr: CPU addr of isp kernel BO
+ *
+ * unmaps and unpin a isp kernel BO.
+ *
+ * This function is exported to allow the V4L2 isp device
+ * external to drm device to free the kernel BO.
+ */
+void isp_kernel_buffer_free(void **buf_obj, u64 *gpu_addr, void **cpu_addr)
+{
+	struct amdgpu_bo **bo = (struct amdgpu_bo **)buf_obj;
+
+	amdgpu_bo_free_kernel(bo, gpu_addr, cpu_addr);
+}
+EXPORT_SYMBOL(isp_kernel_buffer_free);
+
+static const struct amd_ip_funcs isp_ip_funcs = {
+	.name = "isp_ip",
+	.early_init = isp_early_init,
+	.hw_init = isp_hw_init,
+	.hw_fini = isp_hw_fini,
+	.is_idle = isp_is_idle,
+	.set_clockgating_state = isp_set_clockgating_state,
+	.set_powergating_state = isp_set_powergating_state,
+};
+
+const struct amdgpu_ip_block_version isp_v4_1_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_ISP,
+	.major = 4,
+	.minor = 1,
+	.rev = 0,
+	.funcs = &isp_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version isp_v4_1_1_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_ISP,
+	.major = 4,
+	.minor = 1,
+	.rev = 1,
+	.funcs = &isp_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.h
new file mode 100644
index 000000000000..d6f4ffa4c97c
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.h
@@ -0,0 +1,60 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright (C) 2024 Advanced Micro Devices, Inc. All rights reserved.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+
+#ifndef __AMDGPU_ISP_H__
+#define __AMDGPU_ISP_H__
+
+#include <drm/amd/isp.h>
+#include <linux/pm_domain.h>
+
+#define ISP_REGS_OFFSET_END 0x629A4
+
+struct amdgpu_isp;
+
+struct isp_funcs {
+	int (*hw_init)(struct amdgpu_isp *isp);
+	int (*hw_fini)(struct amdgpu_isp *isp);
+};
+
+struct amdgpu_isp {
+	struct device *parent;
+	struct amdgpu_device	*adev;
+	const struct isp_funcs	*funcs;
+	struct mfd_cell *isp_cell;
+	struct resource *isp_res;
+	struct resource *isp_i2c_res;
+	struct resource *isp_gpio_res;
+	struct isp_platform_data *isp_pdata;
+	unsigned int harvest_config;
+	const struct firmware	*fw;
+	struct generic_pm_domain ispgpd;
+};
+
+extern const struct amdgpu_ip_block_version isp_v4_1_0_ip_block;
+extern const struct amdgpu_ip_block_version isp_v4_1_1_ip_block;
+
+#endif /* __AMDGPU_ISP_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
index 4510627ae83e..d020a890a0ea 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
@@ -24,191 +24,447 @@
 #include <linux/kthread.h>
 #include <linux/wait.h>
 #include <linux/sched.h>
-#include <drm/drmP.h>
+
+#include <drm/drm_drv.h>
+
 #include "amdgpu.h"
 #include "amdgpu_trace.h"
+#include "amdgpu_reset.h"
+#include "amdgpu_dev_coredump.h"
+#include "amdgpu_xgmi.h"
 
-static void amdgpu_job_timedout(struct amd_sched_job *s_job)
+static void amdgpu_job_do_core_dump(struct amdgpu_device *adev,
+				    struct amdgpu_job *job)
 {
-	struct amdgpu_job *job = container_of(s_job, struct amdgpu_job, base);
+	int i;
 
-	DRM_ERROR("ring %s timeout, last signaled seq=%u, last emitted seq=%u\n",
-		  job->base.sched->name,
-		  atomic_read(&job->ring->fence_drv.last_seq),
-		  job->ring->fence_drv.sync_seq);
+	dev_info(adev->dev, "Dumping IP State\n");
+	for (i = 0; i < adev->num_ip_blocks; i++)
+		if (adev->ip_blocks[i].version->funcs->dump_ip_state)
+			adev->ip_blocks[i].version->funcs
+				->dump_ip_state((void *)&adev->ip_blocks[i]);
+	dev_info(adev->dev, "Dumping IP State Completed\n");
 
-	if (amdgpu_sriov_vf(job->adev))
-		amdgpu_sriov_gpu_reset(job->adev, job);
-	else
-		amdgpu_gpu_reset(job->adev);
+	amdgpu_coredump(adev, true, false, job);
+}
+
+static void amdgpu_job_core_dump(struct amdgpu_device *adev,
+				 struct amdgpu_job *job)
+{
+	struct list_head device_list, *device_list_handle =  NULL;
+	struct amdgpu_device *tmp_adev = NULL;
+	struct amdgpu_hive_info *hive = NULL;
+
+	if (!amdgpu_sriov_vf(adev))
+		hive = amdgpu_get_xgmi_hive(adev);
+	if (hive)
+		mutex_lock(&hive->hive_lock);
+	/*
+	 * Reuse the logic in amdgpu_device_gpu_recover() to build list of
+	 * devices for code dump
+	 */
+	INIT_LIST_HEAD(&device_list);
+	if (!amdgpu_sriov_vf(adev) && (adev->gmc.xgmi.num_physical_nodes > 1) && hive) {
+		list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head)
+			list_add_tail(&tmp_adev->reset_list, &device_list);
+		if (!list_is_first(&adev->reset_list, &device_list))
+			list_rotate_to_front(&adev->reset_list, &device_list);
+		device_list_handle = &device_list;
+	} else {
+		list_add_tail(&adev->reset_list, &device_list);
+		device_list_handle = &device_list;
+	}
+
+	/* Do the coredump for each device */
+	list_for_each_entry(tmp_adev, device_list_handle, reset_list)
+		amdgpu_job_do_core_dump(tmp_adev, job);
+
+	if (hive) {
+		mutex_unlock(&hive->hive_lock);
+		amdgpu_put_xgmi_hive(hive);
+	}
 }
 
-int amdgpu_job_alloc(struct amdgpu_device *adev, unsigned num_ibs,
-		     struct amdgpu_job **job, struct amdgpu_vm *vm)
+static enum drm_gpu_sched_stat amdgpu_job_timedout(struct drm_sched_job *s_job)
 {
-	size_t size = sizeof(struct amdgpu_job);
+	struct amdgpu_ring *ring = to_amdgpu_ring(s_job->sched);
+	struct amdgpu_job *job = to_amdgpu_job(s_job);
+	struct drm_wedge_task_info *info = NULL;
+	struct amdgpu_task_info *ti = NULL;
+	struct amdgpu_device *adev = ring->adev;
+	int idx, r;
+
+	if (!drm_dev_enter(adev_to_drm(adev), &idx)) {
+		dev_info(adev->dev, "%s - device unplugged skipping recovery on scheduler:%s",
+			 __func__, s_job->sched->name);
+
+		/* Effectively the job is aborted as the device is gone */
+		return DRM_GPU_SCHED_STAT_ENODEV;
+	}
+
+	/*
+	 * Do the coredump immediately after a job timeout to get a very
+	 * close dump/snapshot/representation of GPU's current error status
+	 * Skip it for SRIOV, since VF FLR will be triggered by host driver
+	 * before job timeout
+	 */
+	if (!amdgpu_sriov_vf(adev))
+		amdgpu_job_core_dump(adev, job);
+
+	if (amdgpu_gpu_recovery &&
+	    amdgpu_ring_is_reset_type_supported(ring, AMDGPU_RESET_TYPE_SOFT_RESET) &&
+	    amdgpu_ring_soft_recovery(ring, job->vmid, s_job->s_fence->parent)) {
+		dev_err(adev->dev, "ring %s timeout, but soft recovered\n",
+			s_job->sched->name);
+		goto exit;
+	}
+
+	dev_err(adev->dev, "ring %s timeout, signaled seq=%u, emitted seq=%u\n",
+		job->base.sched->name, atomic_read(&ring->fence_drv.last_seq),
+		ring->fence_drv.sync_seq);
+
+	ti = amdgpu_vm_get_task_info_pasid(ring->adev, job->pasid);
+	if (ti) {
+		amdgpu_vm_print_task_info(adev, ti);
+		info = &ti->task;
+	}
+
+	/* attempt a per ring reset */
+	if (unlikely(adev->debug_disable_gpu_ring_reset)) {
+		dev_err(adev->dev, "Ring reset disabled by debug mask\n");
+	} else if (amdgpu_gpu_recovery &&
+		   amdgpu_ring_is_reset_type_supported(ring, AMDGPU_RESET_TYPE_PER_QUEUE) &&
+		   ring->funcs->reset) {
+		dev_err(adev->dev, "Starting %s ring reset\n",
+			s_job->sched->name);
+		r = amdgpu_ring_reset(ring, job->vmid, &job->hw_fence);
+		if (!r) {
+			atomic_inc(&ring->adev->gpu_reset_counter);
+			dev_err(adev->dev, "Ring %s reset succeeded\n",
+				ring->sched.name);
+			drm_dev_wedged_event(adev_to_drm(adev),
+					     DRM_WEDGE_RECOVERY_NONE, info);
+			goto exit;
+		}
+		dev_err(adev->dev, "Ring %s reset failed\n", ring->sched.name);
+	}
+
+	dma_fence_set_error(&s_job->s_fence->finished, -ETIME);
+
+	if (amdgpu_device_should_recover_gpu(ring->adev)) {
+		struct amdgpu_reset_context reset_context;
+		memset(&reset_context, 0, sizeof(reset_context));
+
+		reset_context.method = AMD_RESET_METHOD_NONE;
+		reset_context.reset_req_dev = adev;
+		reset_context.src = AMDGPU_RESET_SRC_JOB;
+		clear_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
+
+		/*
+		 * To avoid an unnecessary extra coredump, as we have already
+		 * got the very close representation of GPU's error status
+		 */
+		set_bit(AMDGPU_SKIP_COREDUMP, &reset_context.flags);
+
+		r = amdgpu_device_gpu_recover(ring->adev, job, &reset_context);
+		if (r)
+			dev_err(adev->dev, "GPU Recovery Failed: %d\n", r);
+	} else {
+		drm_sched_suspend_timeout(&ring->sched);
+		if (amdgpu_sriov_vf(adev))
+			adev->virt.tdr_debug = true;
+	}
+
+exit:
+	amdgpu_vm_put_task_info(ti);
+	drm_dev_exit(idx);
+	return DRM_GPU_SCHED_STAT_RESET;
+}
 
+int amdgpu_job_alloc(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+		     struct drm_sched_entity *entity, void *owner,
+		     unsigned int num_ibs, struct amdgpu_job **job,
+		     u64 drm_client_id)
+{
 	if (num_ibs == 0)
 		return -EINVAL;
 
-	size += sizeof(struct amdgpu_ib) * num_ibs;
-
-	*job = kzalloc(size, GFP_KERNEL);
+	*job = kzalloc(struct_size(*job, ibs, num_ibs), GFP_KERNEL);
 	if (!*job)
 		return -ENOMEM;
 
-	(*job)->adev = adev;
 	(*job)->vm = vm;
-	(*job)->ibs = (void *)&(*job)[1];
-	(*job)->num_ibs = num_ibs;
 
-	amdgpu_sync_create(&(*job)->sync);
-	amdgpu_sync_create(&(*job)->dep_sync);
-	amdgpu_sync_create(&(*job)->sched_sync);
+	amdgpu_sync_create(&(*job)->explicit_sync);
+	(*job)->generation = amdgpu_vm_generation(adev, vm);
+	(*job)->vm_pd_addr = AMDGPU_BO_INVALID_OFFSET;
 
-	return 0;
+	if (!entity)
+		return 0;
+
+	return drm_sched_job_init(&(*job)->base, entity, 1, owner,
+				  drm_client_id);
 }
 
-int amdgpu_job_alloc_with_ib(struct amdgpu_device *adev, unsigned size,
-			     struct amdgpu_job **job)
+int amdgpu_job_alloc_with_ib(struct amdgpu_device *adev,
+			     struct drm_sched_entity *entity, void *owner,
+			     size_t size, enum amdgpu_ib_pool_type pool_type,
+			     struct amdgpu_job **job, u64 k_job_id)
 {
 	int r;
 
-	r = amdgpu_job_alloc(adev, 1, job, NULL);
+	r = amdgpu_job_alloc(adev, NULL, entity, owner, 1, job,
+			     k_job_id);
 	if (r)
 		return r;
 
-	r = amdgpu_ib_get(adev, NULL, size, &(*job)->ibs[0]);
-	if (r)
+	(*job)->num_ibs = 1;
+	r = amdgpu_ib_get(adev, NULL, size, pool_type, &(*job)->ibs[0]);
+	if (r) {
+		if (entity)
+			drm_sched_job_cleanup(&(*job)->base);
 		kfree(*job);
-	else
-		(*job)->vm_pd_addr = adev->gart.table_addr;
+	}
 
 	return r;
 }
 
+void amdgpu_job_set_resources(struct amdgpu_job *job, struct amdgpu_bo *gds,
+			      struct amdgpu_bo *gws, struct amdgpu_bo *oa)
+{
+	if (gds) {
+		job->gds_base = amdgpu_bo_gpu_offset(gds) >> PAGE_SHIFT;
+		job->gds_size = amdgpu_bo_size(gds) >> PAGE_SHIFT;
+	}
+	if (gws) {
+		job->gws_base = amdgpu_bo_gpu_offset(gws) >> PAGE_SHIFT;
+		job->gws_size = amdgpu_bo_size(gws) >> PAGE_SHIFT;
+	}
+	if (oa) {
+		job->oa_base = amdgpu_bo_gpu_offset(oa) >> PAGE_SHIFT;
+		job->oa_size = amdgpu_bo_size(oa) >> PAGE_SHIFT;
+	}
+}
+
 void amdgpu_job_free_resources(struct amdgpu_job *job)
 {
 	struct dma_fence *f;
 	unsigned i;
 
-	/* use sched fence if available */
-	f = job->base.s_fence ? &job->base.s_fence->finished : job->fence;
+	/* Check if any fences where initialized */
+	if (job->base.s_fence && job->base.s_fence->finished.ops)
+		f = &job->base.s_fence->finished;
+	else if (job->hw_fence.base.ops)
+		f = &job->hw_fence.base;
+	else
+		f = NULL;
 
 	for (i = 0; i < job->num_ibs; ++i)
-		amdgpu_ib_free(job->adev, &job->ibs[i], f);
+		amdgpu_ib_free(&job->ibs[i], f);
+}
+
+static void amdgpu_job_free_cb(struct drm_sched_job *s_job)
+{
+	struct amdgpu_job *job = to_amdgpu_job(s_job);
+
+	drm_sched_job_cleanup(s_job);
+
+	amdgpu_sync_free(&job->explicit_sync);
+
+	/* only put the hw fence if has embedded fence */
+	if (!job->hw_fence.base.ops)
+		kfree(job);
+	else
+		dma_fence_put(&job->hw_fence.base);
 }
 
-static void amdgpu_job_free_cb(struct amd_sched_job *s_job)
+void amdgpu_job_set_gang_leader(struct amdgpu_job *job,
+				struct amdgpu_job *leader)
 {
-	struct amdgpu_job *job = container_of(s_job, struct amdgpu_job, base);
+	struct dma_fence *fence = &leader->base.s_fence->scheduled;
+
+	WARN_ON(job->gang_submit);
 
-	dma_fence_put(job->fence);
-	amdgpu_sync_free(&job->sync);
-	amdgpu_sync_free(&job->dep_sync);
-	amdgpu_sync_free(&job->sched_sync);
-	kfree(job);
+	/*
+	 * Don't add a reference when we are the gang leader to avoid circle
+	 * dependency.
+	 */
+	if (job != leader)
+		dma_fence_get(fence);
+	job->gang_submit = fence;
 }
 
 void amdgpu_job_free(struct amdgpu_job *job)
 {
+	if (job->base.entity)
+		drm_sched_job_cleanup(&job->base);
+
+	amdgpu_job_free_resources(job);
+	amdgpu_sync_free(&job->explicit_sync);
+	if (job->gang_submit != &job->base.s_fence->scheduled)
+		dma_fence_put(job->gang_submit);
+
+	if (!job->hw_fence.base.ops)
+		kfree(job);
+	else
+		dma_fence_put(&job->hw_fence.base);
+}
+
+struct dma_fence *amdgpu_job_submit(struct amdgpu_job *job)
+{
+	struct dma_fence *f;
+
+	drm_sched_job_arm(&job->base);
+	f = dma_fence_get(&job->base.s_fence->finished);
 	amdgpu_job_free_resources(job);
+	drm_sched_entity_push_job(&job->base);
 
-	dma_fence_put(job->fence);
-	amdgpu_sync_free(&job->sync);
-	amdgpu_sync_free(&job->dep_sync);
-	amdgpu_sync_free(&job->sched_sync);
-	kfree(job);
+	return f;
 }
 
-int amdgpu_job_submit(struct amdgpu_job *job, struct amdgpu_ring *ring,
-		      struct amd_sched_entity *entity, void *owner,
-		      struct dma_fence **f)
+int amdgpu_job_submit_direct(struct amdgpu_job *job, struct amdgpu_ring *ring,
+			     struct dma_fence **fence)
 {
 	int r;
-	job->ring = ring;
 
-	if (!f)
-		return -EINVAL;
+	job->base.sched = &ring->sched;
+	r = amdgpu_ib_schedule(ring, job->num_ibs, job->ibs, job, fence);
 
-	r = amd_sched_job_init(&job->base, &ring->sched, entity, owner);
 	if (r)
 		return r;
 
-	job->owner = owner;
-	job->fence_ctx = entity->fence_context;
-	*f = dma_fence_get(&job->base.s_fence->finished);
-	amdgpu_job_free_resources(job);
-	amd_sched_entity_push_job(&job->base);
-
+	amdgpu_job_free(job);
 	return 0;
 }
 
-static struct dma_fence *amdgpu_job_dependency(struct amd_sched_job *sched_job)
+static struct dma_fence *
+amdgpu_job_prepare_job(struct drm_sched_job *sched_job,
+		      struct drm_sched_entity *s_entity)
 {
+	struct amdgpu_ring *ring = to_amdgpu_ring(s_entity->rq->sched);
 	struct amdgpu_job *job = to_amdgpu_job(sched_job);
-	struct amdgpu_vm *vm = job->vm;
-
-	struct dma_fence *fence = amdgpu_sync_get_fence(&job->dep_sync);
+	struct dma_fence *fence;
 	int r;
 
-	if (amd_sched_dependency_optimized(fence, sched_job->s_entity)) {
-		r = amdgpu_sync_fence(job->adev, &job->sched_sync, fence);
-		if (r)
-			DRM_ERROR("Error adding fence to sync (%d)\n", r);
-	}
-	if (!fence)
-		fence = amdgpu_sync_get_fence(&job->sync);
-	while (fence == NULL && vm && !job->vm_id) {
-		struct amdgpu_ring *ring = job->ring;
+	r = drm_sched_entity_error(s_entity);
+	if (r)
+		goto error;
 
-		r = amdgpu_vm_grab_id(vm, ring, &job->sync,
-				      &job->base.s_fence->finished,
-				      job);
-		if (r)
-			DRM_ERROR("Error getting VM ID (%d)\n", r);
+	if (job->gang_submit) {
+		fence = amdgpu_device_switch_gang(ring->adev, job->gang_submit);
+		if (fence)
+			return fence;
+	}
 
-		fence = amdgpu_sync_get_fence(&job->sync);
+	fence = amdgpu_device_enforce_isolation(ring->adev, ring, job);
+	if (fence)
+		return fence;
+
+	if (job->vm && !job->vmid) {
+		r = amdgpu_vmid_grab(job->vm, ring, job, &fence);
+		if (r) {
+			dev_err(ring->adev->dev, "Error getting VM ID (%d)\n", r);
+			goto error;
+		}
+		return fence;
 	}
 
-	return fence;
+	return NULL;
+
+error:
+	dma_fence_set_error(&job->base.s_fence->finished, r);
+	return NULL;
 }
 
-static struct dma_fence *amdgpu_job_run(struct amd_sched_job *sched_job)
+static struct dma_fence *amdgpu_job_run(struct drm_sched_job *sched_job)
 {
-	struct dma_fence *fence = NULL;
+	struct amdgpu_ring *ring = to_amdgpu_ring(sched_job->sched);
+	struct amdgpu_device *adev = ring->adev;
+	struct dma_fence *fence = NULL, *finished;
 	struct amdgpu_job *job;
-	struct amdgpu_fpriv *fpriv = NULL;
-	int r;
+	int r = 0;
 
-	if (!sched_job) {
-		DRM_ERROR("job is null\n");
-		return NULL;
-	}
 	job = to_amdgpu_job(sched_job);
-
-	BUG_ON(amdgpu_sync_peek_fence(&job->sync, NULL));
+	finished = &job->base.s_fence->finished;
 
 	trace_amdgpu_sched_run_job(job);
-	if (job->vm)
-		fpriv = container_of(job->vm, struct amdgpu_fpriv, vm);
-	/* skip ib schedule when vram is lost */
-	if (fpriv && amdgpu_kms_vram_lost(job->adev, fpriv))
-		DRM_ERROR("Skip scheduling IBs!\n");
-	else {
-		r = amdgpu_ib_schedule(job->ring, job->num_ibs, job->ibs, job, &fence);
+
+	/* Skip job if VRAM is lost and never resubmit gangs */
+	if (job->generation != amdgpu_vm_generation(adev, job->vm) ||
+	    (job->job_run_counter && job->gang_submit))
+		dma_fence_set_error(finished, -ECANCELED);
+
+	if (finished->error < 0) {
+		dev_dbg(adev->dev, "Skip scheduling IBs in ring(%s)",
+			ring->name);
+	} else {
+		r = amdgpu_ib_schedule(ring, job->num_ibs, job->ibs, job,
+				       &fence);
 		if (r)
-			DRM_ERROR("Error scheduling IBs (%d)\n", r);
+			dev_err(adev->dev,
+				"Error scheduling IBs (%d) in ring(%s)", r,
+				ring->name);
 	}
-	/* if gpu reset, hw fence will be replaced here */
-	dma_fence_put(job->fence);
-	job->fence = dma_fence_get(fence);
+
+	job->job_run_counter++;
 	amdgpu_job_free_resources(job);
+
+	fence = r ? ERR_PTR(r) : fence;
 	return fence;
 }
 
-const struct amd_sched_backend_ops amdgpu_sched_ops = {
-	.dependency = amdgpu_job_dependency,
+/*
+ * This is a duplicate function from DRM scheduler sched_internal.h.
+ * Plan is to remove it when amdgpu_job_stop_all_jobs_on_sched is removed, due
+ * latter being incorrect and racy.
+ *
+ * See https://lore.kernel.org/amd-gfx/44edde63-7181-44fb-a4f7-94e50514f539@amd.com/
+ */
+static struct drm_sched_job *
+drm_sched_entity_queue_pop(struct drm_sched_entity *entity)
+{
+	struct spsc_node *node;
+
+	node = spsc_queue_pop(&entity->job_queue);
+	if (!node)
+		return NULL;
+
+	return container_of(node, struct drm_sched_job, queue_node);
+}
+
+void amdgpu_job_stop_all_jobs_on_sched(struct drm_gpu_scheduler *sched)
+{
+	struct drm_sched_job *s_job;
+	struct drm_sched_entity *s_entity = NULL;
+	int i;
+
+	/* Signal all jobs not yet scheduled */
+	for (i = DRM_SCHED_PRIORITY_KERNEL; i < sched->num_rqs; i++) {
+		struct drm_sched_rq *rq = sched->sched_rq[i];
+		spin_lock(&rq->lock);
+		list_for_each_entry(s_entity, &rq->entities, list) {
+			while ((s_job = drm_sched_entity_queue_pop(s_entity))) {
+				struct drm_sched_fence *s_fence = s_job->s_fence;
+
+				dma_fence_signal(&s_fence->scheduled);
+				dma_fence_set_error(&s_fence->finished, -EHWPOISON);
+				dma_fence_signal(&s_fence->finished);
+			}
+		}
+		spin_unlock(&rq->lock);
+	}
+
+	/* Signal all jobs already scheduled to HW */
+	list_for_each_entry(s_job, &sched->pending_list, list) {
+		struct drm_sched_fence *s_fence = s_job->s_fence;
+
+		dma_fence_set_error(&s_fence->finished, -EHWPOISON);
+		dma_fence_signal(&s_fence->finished);
+	}
+}
+
+const struct drm_sched_backend_ops amdgpu_sched_ops = {
+	.prepare_job = amdgpu_job_prepare_job,
 	.run_job = amdgpu_job_run,
 	.timedout_job = amdgpu_job_timedout,
 	.free_job = amdgpu_job_free_cb
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h
new file mode 100644
index 000000000000..4a6487eb6cb5
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h
@@ -0,0 +1,129 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __AMDGPU_JOB_H__
+#define __AMDGPU_JOB_H__
+
+#include <drm/gpu_scheduler.h>
+#include "amdgpu_sync.h"
+#include "amdgpu_ring.h"
+
+/* bit set means command submit involves a preamble IB */
+#define AMDGPU_PREAMBLE_IB_PRESENT          (1 << 0)
+/* bit set means preamble IB is first presented in belonging context */
+#define AMDGPU_PREAMBLE_IB_PRESENT_FIRST    (1 << 1)
+/* bit set means context switch occured */
+#define AMDGPU_HAVE_CTX_SWITCH              (1 << 2)
+/* bit set means IB is preempted */
+#define AMDGPU_IB_PREEMPTED                 (1 << 3)
+
+#define to_amdgpu_job(sched_job)		\
+		container_of((sched_job), struct amdgpu_job, base)
+
+#define AMDGPU_JOB_GET_VMID(job) ((job) ? (job)->vmid : 0)
+
+struct amdgpu_fence;
+enum amdgpu_ib_pool_type;
+
+/* Internal kernel job ids. (decreasing values, starting from U64_MAX). */
+#define AMDGPU_KERNEL_JOB_ID_VM_UPDATE              (18446744073709551615ULL)
+#define AMDGPU_KERNEL_JOB_ID_VM_UPDATE_PDES         (18446744073709551614ULL)
+#define AMDGPU_KERNEL_JOB_ID_VM_UPDATE_RANGE        (18446744073709551613ULL)
+#define AMDGPU_KERNEL_JOB_ID_VM_PT_CLEAR            (18446744073709551612ULL)
+#define AMDGPU_KERNEL_JOB_ID_TTM_MAP_BUFFER         (18446744073709551611ULL)
+#define AMDGPU_KERNEL_JOB_ID_TTM_ACCESS_MEMORY_SDMA (18446744073709551610ULL)
+#define AMDGPU_KERNEL_JOB_ID_TTM_COPY_BUFFER        (18446744073709551609ULL)
+#define AMDGPU_KERNEL_JOB_ID_CLEAR_ON_RELEASE       (18446744073709551608ULL)
+#define AMDGPU_KERNEL_JOB_ID_MOVE_BLIT              (18446744073709551607ULL)
+#define AMDGPU_KERNEL_JOB_ID_TTM_CLEAR_BUFFER       (18446744073709551606ULL)
+#define AMDGPU_KERNEL_JOB_ID_CLEANER_SHADER         (18446744073709551605ULL)
+#define AMDGPU_KERNEL_JOB_ID_FLUSH_GPU_TLB          (18446744073709551604ULL)
+#define AMDGPU_KERNEL_JOB_ID_KFD_GART_MAP           (18446744073709551603ULL)
+#define AMDGPU_KERNEL_JOB_ID_VCN_RING_TEST          (18446744073709551602ULL)
+
+struct amdgpu_job {
+	struct drm_sched_job    base;
+	struct amdgpu_vm	*vm;
+	struct amdgpu_sync	explicit_sync;
+	struct amdgpu_fence	hw_fence;
+	struct dma_fence	*gang_submit;
+	uint32_t		preamble_status;
+	uint32_t                preemption_status;
+	bool                    vm_needs_flush;
+	bool			gds_switch_needed;
+	bool			spm_update_needed;
+	uint64_t		vm_pd_addr;
+	unsigned		vmid;
+	unsigned		pasid;
+	uint32_t		gds_base, gds_size;
+	uint32_t		gws_base, gws_size;
+	uint32_t		oa_base, oa_size;
+	uint64_t		generation;
+
+	/* user fence handling */
+	uint64_t		uf_addr;
+	uint64_t		uf_sequence;
+
+	/* virtual addresses for shadow/GDS/CSA */
+	uint64_t		shadow_va;
+	uint64_t		csa_va;
+	uint64_t		gds_va;
+	bool			init_shadow;
+
+	/* job_run_counter >= 1 means a resubmit job */
+	uint32_t		job_run_counter;
+
+	/* enforce isolation */
+	bool			enforce_isolation;
+	bool			run_cleaner_shader;
+
+	uint32_t		num_ibs;
+	struct amdgpu_ib	ibs[];
+};
+
+static inline struct amdgpu_ring *amdgpu_job_ring(struct amdgpu_job *job)
+{
+	return to_amdgpu_ring(job->base.entity->rq->sched);
+}
+
+int amdgpu_job_alloc(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+		     struct drm_sched_entity *entity, void *owner,
+		     unsigned int num_ibs, struct amdgpu_job **job,
+		     u64 drm_client_id);
+int amdgpu_job_alloc_with_ib(struct amdgpu_device *adev,
+			     struct drm_sched_entity *entity, void *owner,
+			     size_t size, enum amdgpu_ib_pool_type pool_type,
+			     struct amdgpu_job **job,
+			     u64 k_job_id);
+void amdgpu_job_set_resources(struct amdgpu_job *job, struct amdgpu_bo *gds,
+			      struct amdgpu_bo *gws, struct amdgpu_bo *oa);
+void amdgpu_job_free_resources(struct amdgpu_job *job);
+void amdgpu_job_set_gang_leader(struct amdgpu_job *job,
+				struct amdgpu_job *leader);
+void amdgpu_job_free(struct amdgpu_job *job);
+struct dma_fence *amdgpu_job_submit(struct amdgpu_job *job);
+int amdgpu_job_submit_direct(struct amdgpu_job *job, struct amdgpu_ring *ring,
+			     struct dma_fence **fence);
+
+void amdgpu_job_stop_all_jobs_on_sched(struct drm_gpu_scheduler *sched);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c
new file mode 100644
index 000000000000..6b7d66b6d4cc
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c
@@ -0,0 +1,607 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+
+#include "amdgpu.h"
+#include "amdgpu_jpeg.h"
+#include "amdgpu_pm.h"
+#include "soc15d.h"
+#include "soc15_common.h"
+
+#define JPEG_IDLE_TIMEOUT	msecs_to_jiffies(1000)
+
+static void amdgpu_jpeg_idle_work_handler(struct work_struct *work);
+static void amdgpu_jpeg_reg_dump_fini(struct amdgpu_device *adev);
+
+int amdgpu_jpeg_sw_init(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	INIT_DELAYED_WORK(&adev->jpeg.idle_work, amdgpu_jpeg_idle_work_handler);
+	mutex_init(&adev->jpeg.jpeg_pg_lock);
+	atomic_set(&adev->jpeg.total_submission_cnt, 0);
+
+	if ((adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) &&
+	    (adev->pg_flags & AMD_PG_SUPPORT_JPEG_DPG))
+		adev->jpeg.indirect_sram = true;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; i++) {
+		if (adev->jpeg.harvest_config & (1U << i))
+			continue;
+
+		if (adev->jpeg.indirect_sram) {
+			r = amdgpu_bo_create_kernel(adev, 64 * 2 * 4, PAGE_SIZE,
+					AMDGPU_GEM_DOMAIN_VRAM |
+					AMDGPU_GEM_DOMAIN_GTT,
+					&adev->jpeg.inst[i].dpg_sram_bo,
+					&adev->jpeg.inst[i].dpg_sram_gpu_addr,
+					&adev->jpeg.inst[i].dpg_sram_cpu_addr);
+			if (r) {
+				dev_err(adev->dev,
+				"JPEG %d (%d) failed to allocate DPG bo\n", i, r);
+				return r;
+			}
+		}
+	}
+
+	return 0;
+}
+
+int amdgpu_jpeg_sw_fini(struct amdgpu_device *adev)
+{
+	int i, j;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1U << i))
+			continue;
+
+		amdgpu_bo_free_kernel(
+			&adev->jpeg.inst[i].dpg_sram_bo,
+			&adev->jpeg.inst[i].dpg_sram_gpu_addr,
+			(void **)&adev->jpeg.inst[i].dpg_sram_cpu_addr);
+
+		for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j)
+			amdgpu_ring_fini(&adev->jpeg.inst[i].ring_dec[j]);
+	}
+
+	if (adev->jpeg.reg_list)
+		amdgpu_jpeg_reg_dump_fini(adev);
+
+	mutex_destroy(&adev->jpeg.jpeg_pg_lock);
+
+	return 0;
+}
+
+int amdgpu_jpeg_suspend(struct amdgpu_device *adev)
+{
+	cancel_delayed_work_sync(&adev->jpeg.idle_work);
+
+	return 0;
+}
+
+int amdgpu_jpeg_resume(struct amdgpu_device *adev)
+{
+	return 0;
+}
+
+static void amdgpu_jpeg_idle_work_handler(struct work_struct *work)
+{
+	struct amdgpu_device *adev =
+		container_of(work, struct amdgpu_device, jpeg.idle_work.work);
+	unsigned int fences = 0;
+	unsigned int i, j;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1U << i))
+			continue;
+
+		for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j)
+			fences += amdgpu_fence_count_emitted(&adev->jpeg.inst[i].ring_dec[j]);
+	}
+
+	if (!fences && !atomic_read(&adev->jpeg.total_submission_cnt)) {
+		mutex_lock(&adev->jpeg.jpeg_pg_lock);
+		amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_JPEG,
+						       AMD_PG_STATE_GATE);
+		mutex_unlock(&adev->jpeg.jpeg_pg_lock);
+	} else
+		schedule_delayed_work(&adev->jpeg.idle_work, JPEG_IDLE_TIMEOUT);
+}
+
+void amdgpu_jpeg_ring_begin_use(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	atomic_inc(&adev->jpeg.total_submission_cnt);
+	cancel_delayed_work_sync(&adev->jpeg.idle_work);
+
+	mutex_lock(&adev->jpeg.jpeg_pg_lock);
+	amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_JPEG,
+						       AMD_PG_STATE_UNGATE);
+	mutex_unlock(&adev->jpeg.jpeg_pg_lock);
+}
+
+void amdgpu_jpeg_ring_end_use(struct amdgpu_ring *ring)
+{
+	atomic_dec(&ring->adev->jpeg.total_submission_cnt);
+	schedule_delayed_work(&ring->adev->jpeg.idle_work, JPEG_IDLE_TIMEOUT);
+}
+
+int amdgpu_jpeg_dec_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	/* JPEG in SRIOV does not support direct register read/write */
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	r = amdgpu_ring_alloc(ring, 3);
+	if (r)
+		return r;
+
+	WREG32(adev->jpeg.inst[ring->me].external.jpeg_pitch[ring->pipe], 0xCAFEDEAD);
+	/* Add a read register to make sure the write register is executed. */
+	RREG32(adev->jpeg.inst[ring->me].external.jpeg_pitch[ring->pipe]);
+
+	amdgpu_ring_write(ring, PACKET0(adev->jpeg.internal.jpeg_pitch[ring->pipe], 0));
+	amdgpu_ring_write(ring, 0xABADCAFE);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(adev->jpeg.inst[ring->me].external.jpeg_pitch[ring->pipe]);
+		if (tmp == 0xABADCAFE)
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+
+	return r;
+}
+
+static int amdgpu_jpeg_dec_set_reg(struct amdgpu_ring *ring, uint32_t handle,
+		struct dma_fence **fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	struct dma_fence *f = NULL;
+	const unsigned ib_size_dw = 16;
+	int i, r;
+
+	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL, ib_size_dw * 4,
+				     AMDGPU_IB_POOL_DIRECT, &job,
+				     AMDGPU_KERNEL_JOB_ID_VCN_RING_TEST);
+	if (r)
+		return r;
+
+	ib = &job->ibs[0];
+
+	ib->ptr[0] = PACKETJ(adev->jpeg.internal.jpeg_pitch[ring->pipe], 0, 0, PACKETJ_TYPE0);
+	ib->ptr[1] = 0xDEADBEEF;
+	for (i = 2; i < 16; i += 2) {
+		ib->ptr[i] = PACKETJ(0, 0, 0, PACKETJ_TYPE6);
+		ib->ptr[i+1] = 0;
+	}
+	ib->length_dw = 16;
+
+	r = amdgpu_job_submit_direct(job, ring, &f);
+	if (r)
+		goto err;
+
+	if (fence)
+		*fence = dma_fence_get(f);
+	dma_fence_put(f);
+
+	return 0;
+
+err:
+	amdgpu_job_free(job);
+	return r;
+}
+
+int amdgpu_jpeg_dec_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t tmp = 0;
+	unsigned i;
+	struct dma_fence *fence = NULL;
+	long r = 0;
+
+	r = amdgpu_jpeg_dec_set_reg(ring, 1, &fence);
+	if (r)
+		goto error;
+
+	r = dma_fence_wait_timeout(fence, false, timeout);
+	if (r == 0) {
+		r = -ETIMEDOUT;
+		goto error;
+	} else if (r < 0) {
+		goto error;
+	} else {
+		r = 0;
+	}
+
+	if (!amdgpu_sriov_vf(adev)) {
+		for (i = 0; i < adev->usec_timeout; i++) {
+			tmp = RREG32(adev->jpeg.inst[ring->me].external.jpeg_pitch[ring->pipe]);
+			if (tmp == 0xDEADBEEF)
+				break;
+			udelay(1);
+			if (amdgpu_emu_mode == 1)
+				udelay(10);
+		}
+
+		if (i >= adev->usec_timeout)
+			r = -ETIMEDOUT;
+	}
+
+	dma_fence_put(fence);
+error:
+	return r;
+}
+
+int amdgpu_jpeg_process_poison_irq(struct amdgpu_device *adev,
+				struct amdgpu_irq_src *source,
+				struct amdgpu_iv_entry *entry)
+{
+	struct ras_common_if *ras_if = adev->jpeg.ras_if;
+	struct ras_dispatch_if ih_data = {
+		.entry = entry,
+	};
+
+	if (!ras_if)
+		return 0;
+
+	ih_data.head = *ras_if;
+	amdgpu_ras_interrupt_dispatch(adev, &ih_data);
+
+	return 0;
+}
+
+int amdgpu_jpeg_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
+{
+	int r, i;
+
+	r = amdgpu_ras_block_late_init(adev, ras_block);
+	if (r)
+		return r;
+
+	if (amdgpu_ras_is_supported(adev, ras_block->block)) {
+		for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+			if (adev->jpeg.harvest_config & (1 << i) ||
+			    !adev->jpeg.inst[i].ras_poison_irq.funcs)
+				continue;
+
+			r = amdgpu_irq_get(adev, &adev->jpeg.inst[i].ras_poison_irq, 0);
+			if (r)
+				goto late_fini;
+		}
+	}
+	return 0;
+
+late_fini:
+	amdgpu_ras_block_late_fini(adev, ras_block);
+	return r;
+}
+
+int amdgpu_jpeg_ras_sw_init(struct amdgpu_device *adev)
+{
+	int err;
+	struct amdgpu_jpeg_ras *ras;
+
+	if (!adev->jpeg.ras)
+		return 0;
+
+	ras = adev->jpeg.ras;
+	err = amdgpu_ras_register_ras_block(adev, &ras->ras_block);
+	if (err) {
+		dev_err(adev->dev, "Failed to register jpeg ras block!\n");
+		return err;
+	}
+
+	strcpy(ras->ras_block.ras_comm.name, "jpeg");
+	ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__JPEG;
+	ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__POISON;
+	adev->jpeg.ras_if = &ras->ras_block.ras_comm;
+
+	if (!ras->ras_block.ras_late_init)
+		ras->ras_block.ras_late_init = amdgpu_jpeg_ras_late_init;
+
+	return 0;
+}
+
+int amdgpu_jpeg_psp_update_sram(struct amdgpu_device *adev, int inst_idx,
+			       enum AMDGPU_UCODE_ID ucode_id)
+{
+	struct amdgpu_firmware_info ucode = {
+		.ucode_id = AMDGPU_UCODE_ID_JPEG_RAM,
+		.mc_addr = adev->jpeg.inst[inst_idx].dpg_sram_gpu_addr,
+		.ucode_size = ((uintptr_t)adev->jpeg.inst[inst_idx].dpg_sram_curr_addr -
+			      (uintptr_t)adev->jpeg.inst[inst_idx].dpg_sram_cpu_addr),
+	};
+
+	return psp_execute_ip_fw_load(&adev->psp, &ucode);
+}
+
+/*
+ * debugfs for to enable/disable jpeg job submission to specific core.
+ */
+#if defined(CONFIG_DEBUG_FS)
+static int amdgpu_debugfs_jpeg_sched_mask_set(void *data, u64 val)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)data;
+	u32 i, j;
+	u64 mask = 0;
+	struct amdgpu_ring *ring;
+
+	if (!adev)
+		return -ENODEV;
+
+	mask = (1ULL << (adev->jpeg.num_jpeg_inst * adev->jpeg.num_jpeg_rings)) - 1;
+	if ((val & mask) == 0)
+		return -EINVAL;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
+			ring = &adev->jpeg.inst[i].ring_dec[j];
+			if (val & (BIT_ULL(1) << ((i * adev->jpeg.num_jpeg_rings) + j)))
+				ring->sched.ready = true;
+			else
+				ring->sched.ready = false;
+		}
+	}
+	/* publish sched.ready flag update effective immediately across smp */
+	smp_rmb();
+	return 0;
+}
+
+static int amdgpu_debugfs_jpeg_sched_mask_get(void *data, u64 *val)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)data;
+	u32 i, j;
+	u64 mask = 0;
+	struct amdgpu_ring *ring;
+
+	if (!adev)
+		return -ENODEV;
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
+			ring = &adev->jpeg.inst[i].ring_dec[j];
+			if (ring->sched.ready)
+				mask |= 1ULL << ((i * adev->jpeg.num_jpeg_rings) + j);
+		}
+	}
+	*val = mask;
+	return 0;
+}
+
+DEFINE_DEBUGFS_ATTRIBUTE(amdgpu_debugfs_jpeg_sched_mask_fops,
+			 amdgpu_debugfs_jpeg_sched_mask_get,
+			 amdgpu_debugfs_jpeg_sched_mask_set, "%llx\n");
+
+#endif
+
+void amdgpu_debugfs_jpeg_sched_mask_init(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	struct drm_minor *minor = adev_to_drm(adev)->primary;
+	struct dentry *root = minor->debugfs_root;
+	char name[32];
+
+	if (!(adev->jpeg.num_jpeg_inst > 1) && !(adev->jpeg.num_jpeg_rings > 1))
+		return;
+	sprintf(name, "amdgpu_jpeg_sched_mask");
+	debugfs_create_file(name, 0600, root, adev,
+			    &amdgpu_debugfs_jpeg_sched_mask_fops);
+#endif
+}
+
+static ssize_t amdgpu_get_jpeg_reset_mask(struct device *dev,
+						struct device_attribute *attr,
+						char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	if (!adev)
+		return -ENODEV;
+
+	return amdgpu_show_reset_mask(buf, adev->jpeg.supported_reset);
+}
+
+static DEVICE_ATTR(jpeg_reset_mask, 0444,
+		   amdgpu_get_jpeg_reset_mask, NULL);
+
+int amdgpu_jpeg_sysfs_reset_mask_init(struct amdgpu_device *adev)
+{
+	int r = 0;
+
+	if (adev->jpeg.num_jpeg_inst) {
+		r = device_create_file(adev->dev, &dev_attr_jpeg_reset_mask);
+		if (r)
+			return r;
+	}
+
+	return r;
+}
+
+void amdgpu_jpeg_sysfs_reset_mask_fini(struct amdgpu_device *adev)
+{
+	if (adev->dev->kobj.sd) {
+		if (adev->jpeg.num_jpeg_inst)
+			device_remove_file(adev->dev, &dev_attr_jpeg_reset_mask);
+	}
+}
+
+int amdgpu_jpeg_reg_dump_init(struct amdgpu_device *adev,
+			       const struct amdgpu_hwip_reg_entry *reg, u32 count)
+{
+	adev->jpeg.ip_dump = kcalloc(adev->jpeg.num_jpeg_inst * count,
+				     sizeof(uint32_t), GFP_KERNEL);
+	if (!adev->jpeg.ip_dump) {
+		dev_err(adev->dev,
+			"Failed to allocate memory for JPEG IP Dump\n");
+		return -ENOMEM;
+	}
+	adev->jpeg.reg_list = reg;
+	adev->jpeg.reg_count = count;
+
+	return 0;
+}
+
+static void amdgpu_jpeg_reg_dump_fini(struct amdgpu_device *adev)
+{
+	kfree(adev->jpeg.ip_dump);
+	adev->jpeg.reg_list = NULL;
+	adev->jpeg.reg_count = 0;
+}
+
+void amdgpu_jpeg_dump_ip_state(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	u32 inst_off, inst_id, is_powered;
+	int i, j;
+
+	if (!adev->jpeg.ip_dump)
+		return;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; i++) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+
+		inst_id = GET_INST(JPEG, i);
+		inst_off = i * adev->jpeg.reg_count;
+		/* check power status from UVD_JPEG_POWER_STATUS */
+		adev->jpeg.ip_dump[inst_off] =
+			RREG32(SOC15_REG_ENTRY_OFFSET_INST(adev->jpeg.reg_list[0],
+							   inst_id));
+		is_powered = ((adev->jpeg.ip_dump[inst_off] & 0x1) != 1);
+
+		if (is_powered)
+			for (j = 1; j < adev->jpeg.reg_count; j++)
+				adev->jpeg.ip_dump[inst_off + j] =
+					RREG32(SOC15_REG_ENTRY_OFFSET_INST(adev->jpeg.reg_list[j],
+									   inst_id));
+	}
+}
+
+void amdgpu_jpeg_print_ip_state(struct amdgpu_ip_block *ip_block, struct drm_printer *p)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	u32 inst_off, is_powered;
+	int i, j;
+
+	if (!adev->jpeg.ip_dump)
+		return;
+
+	drm_printf(p, "num_instances:%d\n", adev->jpeg.num_jpeg_inst);
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; i++) {
+		if (adev->jpeg.harvest_config & (1 << i)) {
+			drm_printf(p, "\nHarvested Instance:JPEG%d Skipping dump\n", i);
+			continue;
+		}
+
+		inst_off = i * adev->jpeg.reg_count;
+		is_powered = ((adev->jpeg.ip_dump[inst_off] & 0x1) != 1);
+
+		if (is_powered) {
+			drm_printf(p, "Active Instance:JPEG%d\n", i);
+			for (j = 0; j < adev->jpeg.reg_count; j++)
+				drm_printf(p, "%-50s \t 0x%08x\n", adev->jpeg.reg_list[j].reg_name,
+					   adev->jpeg.ip_dump[inst_off + j]);
+		} else
+			drm_printf(p, "\nInactive Instance:JPEG%d\n", i);
+	}
+}
+
+static inline bool amdgpu_jpeg_reg_valid(u32 reg)
+{
+	if (reg < JPEG_REG_RANGE_START || reg > JPEG_REG_RANGE_END ||
+	    (reg >= JPEG_ATOMIC_RANGE_START && reg <= JPEG_ATOMIC_RANGE_END))
+		return false;
+	else
+		return true;
+}
+
+/**
+ * amdgpu_jpeg_dec_parse_cs - command submission parser
+ *
+ * @parser: Command submission parser context
+ * @job: the job to parse
+ * @ib: the IB to parse
+ *
+ * Parse the command stream, return -EINVAL for invalid packet,
+ * 0 otherwise
+ */
+
+int amdgpu_jpeg_dec_parse_cs(struct amdgpu_cs_parser *parser,
+			      struct amdgpu_job *job,
+			      struct amdgpu_ib *ib)
+{
+	u32 i, reg, res, cond, type;
+	struct amdgpu_device *adev = parser->adev;
+
+	for (i = 0; i < ib->length_dw ; i += 2) {
+		reg  = CP_PACKETJ_GET_REG(ib->ptr[i]);
+		res  = CP_PACKETJ_GET_RES(ib->ptr[i]);
+		cond = CP_PACKETJ_GET_COND(ib->ptr[i]);
+		type = CP_PACKETJ_GET_TYPE(ib->ptr[i]);
+
+		if (res) /* only support 0 at the moment */
+			return -EINVAL;
+
+		switch (type) {
+		case PACKETJ_TYPE0:
+			if (cond != PACKETJ_CONDITION_CHECK0 ||
+			    !amdgpu_jpeg_reg_valid(reg)) {
+				dev_err(adev->dev, "Invalid packet [0x%08x]!\n", ib->ptr[i]);
+				return -EINVAL;
+			}
+			break;
+		case PACKETJ_TYPE3:
+			if (cond != PACKETJ_CONDITION_CHECK3 ||
+			    !amdgpu_jpeg_reg_valid(reg)) {
+				dev_err(adev->dev, "Invalid packet [0x%08x]!\n", ib->ptr[i]);
+				return -EINVAL;
+			}
+			break;
+		case PACKETJ_TYPE6:
+			if (ib->ptr[i] == CP_PACKETJ_NOP)
+				continue;
+			dev_err(adev->dev, "Invalid packet [0x%08x]!\n", ib->ptr[i]);
+			return -EINVAL;
+		default:
+			dev_err(adev->dev, "Unknown packet type %d !\n", type);
+			return -EINVAL;
+		}
+	}
+
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h
new file mode 100644
index 000000000000..346ae0ab09d3
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h
@@ -0,0 +1,184 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_JPEG_H__
+#define __AMDGPU_JPEG_H__
+
+#include "amdgpu_ras.h"
+#include "amdgpu_cs.h"
+
+#define AMDGPU_MAX_JPEG_INSTANCES	4
+#define AMDGPU_MAX_JPEG_RINGS           10
+#define AMDGPU_MAX_JPEG_RINGS_4_0_3     8
+
+#define JPEG_REG_RANGE_START            0x4000
+#define JPEG_REG_RANGE_END              0x41c2
+#define JPEG_ATOMIC_RANGE_START         0x4120
+#define JPEG_ATOMIC_RANGE_END           0x412A
+
+
+#define AMDGPU_JPEG_HARVEST_JPEG0 (1 << 0)
+#define AMDGPU_JPEG_HARVEST_JPEG1 (1 << 1)
+
+#define WREG32_SOC15_JPEG_DPG_MODE(inst_idx, offset, value, indirect)			\
+	do {										\
+		if (!indirect) {							\
+			WREG32_SOC15(JPEG, GET_INST(JPEG, inst_idx),			\
+				     mmUVD_DPG_LMA_DATA, value);			\
+			WREG32_SOC15(							\
+				JPEG, GET_INST(JPEG, inst_idx),				\
+				mmUVD_DPG_LMA_CTL,					\
+				(UVD_DPG_LMA_CTL__READ_WRITE_MASK |			\
+				 offset << UVD_DPG_LMA_CTL__READ_WRITE_ADDR__SHIFT |	\
+				 indirect << UVD_DPG_LMA_CTL__SRAM_SEL__SHIFT));	\
+		} else {								\
+			*adev->jpeg.inst[inst_idx].dpg_sram_curr_addr++ =		\
+				offset;							\
+			*adev->jpeg.inst[inst_idx].dpg_sram_curr_addr++ =		\
+				value;							\
+		}									\
+	} while (0)
+
+#define RREG32_SOC15_JPEG_DPG_MODE(inst_idx, offset, mask_en)					\
+	({											\
+		WREG32_SOC15(JPEG, inst_idx, mmUVD_DPG_LMA_CTL,					\
+			(0x0 << UVD_DPG_LMA_CTL__READ_WRITE__SHIFT |				\
+			mask_en << UVD_DPG_LMA_CTL__MASK_EN__SHIFT |				\
+			offset << UVD_DPG_LMA_CTL__READ_WRITE_ADDR__SHIFT));			\
+		RREG32_SOC15(JPEG, inst_idx, mmUVD_DPG_LMA_DATA);				\
+	})
+
+#define WREG32_SOC24_JPEG_DPG_MODE(inst_idx, offset, value, indirect)		\
+	do {									\
+		WREG32_SOC15(JPEG, GET_INST(JPEG, inst_idx),			\
+			     regUVD_DPG_LMA_DATA, value);			\
+		WREG32_SOC15(JPEG, GET_INST(JPEG, inst_idx),			\
+			     regUVD_DPG_LMA_MASK, 0xFFFFFFFF);			\
+		WREG32_SOC15(							\
+			JPEG, GET_INST(JPEG, inst_idx),				\
+			regUVD_DPG_LMA_CTL,					\
+			(UVD_DPG_LMA_CTL__READ_WRITE_MASK |			\
+			 offset << UVD_DPG_LMA_CTL__READ_WRITE_ADDR__SHIFT |	\
+			 indirect << UVD_DPG_LMA_CTL__SRAM_SEL__SHIFT));	\
+	} while (0)
+
+#define RREG32_SOC24_JPEG_DPG_MODE(inst_idx, offset, mask_en)			\
+	do {									\
+		WREG32_SOC15(JPEG, GET_INST(JPEG, inst_idx),			\
+			regUVD_DPG_LMA_MASK, 0xFFFFFFFF);			\
+		WREG32_SOC15(JPEG, GET_INST(JPEG, inst_idx),			\
+			regUVD_DPG_LMA_CTL,					\
+			(UVD_DPG_LMA_CTL__MASK_EN_MASK |			\
+			offset << UVD_DPG_LMA_CTL__READ_WRITE_ADDR__SHIFT));	\
+		RREG32_SOC15(JPEG, inst_idx, regUVD_DPG_LMA_DATA);		\
+	} while (0)
+
+#define ADD_SOC24_JPEG_TO_DPG_SRAM(inst_idx, offset, value, indirect)		\
+	do {									\
+		*adev->jpeg.inst[inst_idx].dpg_sram_curr_addr++ = offset;	\
+		*adev->jpeg.inst[inst_idx].dpg_sram_curr_addr++ = value;	\
+	} while (0)
+
+struct amdgpu_hwip_reg_entry;
+
+enum amdgpu_jpeg_caps {
+	AMDGPU_JPEG_RRMT_ENABLED,
+};
+
+#define AMDGPU_JPEG_CAPS(caps) BIT(AMDGPU_JPEG_##caps)
+
+struct amdgpu_jpeg_reg{
+	unsigned jpeg_pitch[AMDGPU_MAX_JPEG_RINGS];
+};
+
+struct amdgpu_jpeg_inst {
+	struct amdgpu_ring ring_dec[AMDGPU_MAX_JPEG_RINGS];
+	struct amdgpu_irq_src irq;
+	struct amdgpu_irq_src ras_poison_irq;
+	struct amdgpu_jpeg_reg external;
+	struct amdgpu_bo	*dpg_sram_bo;
+	struct dpg_pause_state	pause_state;
+	void			*dpg_sram_cpu_addr;
+	uint64_t		dpg_sram_gpu_addr;
+	uint32_t		*dpg_sram_curr_addr;
+	uint8_t aid_id;
+};
+
+struct amdgpu_jpeg_ras {
+	struct amdgpu_ras_block_object ras_block;
+};
+
+struct amdgpu_jpeg {
+	uint8_t	num_jpeg_inst;
+	struct amdgpu_jpeg_inst inst[AMDGPU_MAX_JPEG_INSTANCES];
+	unsigned num_jpeg_rings;
+	struct amdgpu_jpeg_reg internal;
+	unsigned harvest_config;
+	struct delayed_work idle_work;
+	enum amd_powergating_state cur_state;
+	struct mutex jpeg_pg_lock;
+	atomic_t total_submission_cnt;
+	struct ras_common_if	*ras_if;
+	struct amdgpu_jpeg_ras	*ras;
+
+	uint16_t inst_mask;
+	uint8_t num_inst_per_aid;
+	bool	indirect_sram;
+	uint32_t supported_reset;
+	uint32_t caps;
+	u32 *ip_dump;
+	u32 reg_count;
+	const struct amdgpu_hwip_reg_entry *reg_list;
+};
+
+int amdgpu_jpeg_sw_init(struct amdgpu_device *adev);
+int amdgpu_jpeg_sw_fini(struct amdgpu_device *adev);
+int amdgpu_jpeg_suspend(struct amdgpu_device *adev);
+int amdgpu_jpeg_resume(struct amdgpu_device *adev);
+
+void amdgpu_jpeg_ring_begin_use(struct amdgpu_ring *ring);
+void amdgpu_jpeg_ring_end_use(struct amdgpu_ring *ring);
+
+int amdgpu_jpeg_dec_ring_test_ring(struct amdgpu_ring *ring);
+int amdgpu_jpeg_dec_ring_test_ib(struct amdgpu_ring *ring, long timeout);
+
+int amdgpu_jpeg_process_poison_irq(struct amdgpu_device *adev,
+				struct amdgpu_irq_src *source,
+				struct amdgpu_iv_entry *entry);
+int amdgpu_jpeg_ras_late_init(struct amdgpu_device *adev,
+				struct ras_common_if *ras_block);
+int amdgpu_jpeg_ras_sw_init(struct amdgpu_device *adev);
+int amdgpu_jpeg_psp_update_sram(struct amdgpu_device *adev, int inst_idx,
+			       enum AMDGPU_UCODE_ID ucode_id);
+void amdgpu_debugfs_jpeg_sched_mask_init(struct amdgpu_device *adev);
+int amdgpu_jpeg_sysfs_reset_mask_init(struct amdgpu_device *adev);
+void amdgpu_jpeg_sysfs_reset_mask_fini(struct amdgpu_device *adev);
+int amdgpu_jpeg_reg_dump_init(struct amdgpu_device *adev,
+			       const struct amdgpu_hwip_reg_entry *reg, u32 count);
+void amdgpu_jpeg_dump_ip_state(struct amdgpu_ip_block *ip_block);
+void amdgpu_jpeg_print_ip_state(struct amdgpu_ip_block *ip_block, struct drm_printer *p);
+int amdgpu_jpeg_dec_parse_cs(struct amdgpu_cs_parser *parser,
+			     struct amdgpu_job *job,
+			     struct amdgpu_ib *ib);
+
+#endif /*__AMDGPU_JPEG_H__*/
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
index e16229000a98..a9327472c651 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
@@ -25,16 +25,51 @@
  *          Alex Deucher
  *          Jerome Glisse
  */
-#include <drm/drmP.h>
+
 #include "amdgpu.h"
 #include <drm/amdgpu_drm.h>
+#include <drm/drm_drv.h>
+#include <drm/drm_fb_helper.h>
 #include "amdgpu_uvd.h"
 #include "amdgpu_vce.h"
+#include "atom.h"
 
 #include <linux/vga_switcheroo.h>
 #include <linux/slab.h>
+#include <linux/uaccess.h>
+#include <linux/pci.h>
 #include <linux/pm_runtime.h>
 #include "amdgpu_amdkfd.h"
+#include "amdgpu_gem.h"
+#include "amdgpu_display.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_reset.h"
+#include "amd_pcie.h"
+#include "amdgpu_userq.h"
+
+void amdgpu_unregister_gpu_instance(struct amdgpu_device *adev)
+{
+	struct amdgpu_gpu_instance *gpu_instance;
+	int i;
+
+	mutex_lock(&mgpu_info.mutex);
+
+	for (i = 0; i < mgpu_info.num_gpu; i++) {
+		gpu_instance = &(mgpu_info.gpu_ins[i]);
+		if (gpu_instance->adev == adev) {
+			mgpu_info.gpu_ins[i] =
+				mgpu_info.gpu_ins[mgpu_info.num_gpu - 1];
+			mgpu_info.num_gpu--;
+			if (adev->flags & AMD_IS_APU)
+				mgpu_info.num_apu--;
+			else
+				mgpu_info.num_dgpu--;
+			break;
+		}
+	}
+
+	mutex_unlock(&mgpu_info.mutex);
+}
 
 /**
  * amdgpu_driver_unload_kms - Main unload function for KMS.
@@ -46,95 +81,63 @@
  */
 void amdgpu_driver_unload_kms(struct drm_device *dev)
 {
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 
 	if (adev == NULL)
 		return;
 
+	amdgpu_unregister_gpu_instance(adev);
+
 	if (adev->rmmio == NULL)
-		goto done_free;
+		return;
 
-	if (amdgpu_sriov_vf(adev))
-		amdgpu_virt_request_full_gpu(adev, false);
+	if (amdgpu_acpi_smart_shift_update(adev, AMDGPU_SS_DRV_UNLOAD))
+		DRM_WARN("smart shift update failed\n");
 
-	if (amdgpu_device_is_px(dev)) {
-		pm_runtime_get_sync(dev->dev);
-		pm_runtime_forbid(dev->dev);
-	}
+	amdgpu_acpi_fini(adev);
+	amdgpu_device_fini_hw(adev);
+}
+
+void amdgpu_register_gpu_instance(struct amdgpu_device *adev)
+{
+	struct amdgpu_gpu_instance *gpu_instance;
 
-	amdgpu_amdkfd_device_fini(adev);
+	mutex_lock(&mgpu_info.mutex);
 
-	amdgpu_acpi_fini(adev);
+	if (mgpu_info.num_gpu >= MAX_GPU_INSTANCE) {
+		DRM_ERROR("Cannot register more gpu instance\n");
+		mutex_unlock(&mgpu_info.mutex);
+		return;
+	}
 
-	amdgpu_device_fini(adev);
+	gpu_instance = &(mgpu_info.gpu_ins[mgpu_info.num_gpu]);
+	gpu_instance->adev = adev;
+	gpu_instance->mgpu_fan_enabled = 0;
 
-done_free:
-	kfree(adev);
-	dev->dev_private = NULL;
+	mgpu_info.num_gpu++;
+	if (adev->flags & AMD_IS_APU)
+		mgpu_info.num_apu++;
+	else
+		mgpu_info.num_dgpu++;
+
+	mutex_unlock(&mgpu_info.mutex);
 }
 
 /**
  * amdgpu_driver_load_kms - Main load function for KMS.
  *
- * @dev: drm dev pointer
+ * @adev: pointer to struct amdgpu_device
  * @flags: device flags
  *
  * This is the main load function for KMS (all asics).
  * Returns 0 on success, error on failure.
  */
-int amdgpu_driver_load_kms(struct drm_device *dev, unsigned long flags)
+int amdgpu_driver_load_kms(struct amdgpu_device *adev, unsigned long flags)
 {
-	struct amdgpu_device *adev;
+	struct drm_device *dev;
 	int r, acpi_status;
 
-#ifdef CONFIG_DRM_AMDGPU_SI
-	if (!amdgpu_si_support) {
-		switch (flags & AMD_ASIC_MASK) {
-		case CHIP_TAHITI:
-		case CHIP_PITCAIRN:
-		case CHIP_VERDE:
-		case CHIP_OLAND:
-		case CHIP_HAINAN:
-			dev_info(dev->dev,
-				 "SI support provided by radeon.\n");
-			dev_info(dev->dev,
-				 "Use radeon.si_support=0 amdgpu.si_support=1 to override.\n"
-				);
-			return -ENODEV;
-		}
-	}
-#endif
-#ifdef CONFIG_DRM_AMDGPU_CIK
-	if (!amdgpu_cik_support) {
-		switch (flags & AMD_ASIC_MASK) {
-		case CHIP_KAVERI:
-		case CHIP_BONAIRE:
-		case CHIP_HAWAII:
-		case CHIP_KABINI:
-		case CHIP_MULLINS:
-			dev_info(dev->dev,
-				 "CIK support provided by radeon.\n");
-			dev_info(dev->dev,
-				 "Use radeon.cik_support=0 amdgpu.cik_support=1 to override.\n"
-				);
-			return -ENODEV;
-		}
-	}
-#endif
-
-	adev = kzalloc(sizeof(struct amdgpu_device), GFP_KERNEL);
-	if (adev == NULL) {
-		return -ENOMEM;
-	}
-	dev->dev_private = (void *)adev;
-
-	if ((amdgpu_runtime_pm != 0) &&
-	    amdgpu_has_atpx() &&
-	    (amdgpu_is_atpx_hybrid() ||
-	     amdgpu_has_atpx_dgpu_power_cntl()) &&
-	    ((flags & AMD_IS_APU) == 0) &&
-	    !pci_is_thunderbolt_attached(dev->pdev))
-		flags |= AMD_IS_PX;
+	dev = adev_to_drm(adev);
 
 	/* amdgpu_device_init should report only fatal error
 	 * like memory allocation failure or iomapping failure,
@@ -142,48 +145,70 @@ int amdgpu_driver_load_kms(struct drm_device *dev, unsigned long flags)
 	 * properly initialize the GPU MC controller and permit
 	 * VRAM allocation
 	 */
-	r = amdgpu_device_init(adev, dev, dev->pdev, flags);
+	r = amdgpu_device_init(adev, flags);
 	if (r) {
-		dev_err(&dev->pdev->dev, "Fatal error during GPU init\n");
+		dev_err(dev->dev, "Fatal error during GPU init\n");
 		goto out;
 	}
 
+	amdgpu_device_detect_runtime_pm_mode(adev);
+
 	/* Call ACPI methods: require modeset init
 	 * but failure is not fatal
 	 */
-	if (!r) {
-		acpi_status = amdgpu_acpi_init(adev);
-		if (acpi_status)
-		dev_dbg(&dev->pdev->dev,
-				"Error during ACPI methods call\n");
-	}
 
-	amdgpu_amdkfd_device_probe(adev);
-	amdgpu_amdkfd_device_init(adev);
+	acpi_status = amdgpu_acpi_init(adev);
+	if (acpi_status)
+		dev_dbg(dev->dev, "Error during ACPI methods call\n");
 
-	if (amdgpu_device_is_px(dev)) {
-		pm_runtime_use_autosuspend(dev->dev);
-		pm_runtime_set_autosuspend_delay(dev->dev, 5000);
-		pm_runtime_set_active(dev->dev);
-		pm_runtime_allow(dev->dev);
-		pm_runtime_mark_last_busy(dev->dev);
-		pm_runtime_put_autosuspend(dev->dev);
-	}
-
-	if (amdgpu_sriov_vf(adev))
-		amdgpu_virt_release_full_gpu(adev, true);
+	if (amdgpu_acpi_smart_shift_update(adev, AMDGPU_SS_DRV_LOAD))
+		DRM_WARN("smart shift update failed\n");
 
 out:
-	if (r) {
-		/* balance pm_runtime_get_sync in amdgpu_driver_unload_kms */
-		if (adev->rmmio && amdgpu_device_is_px(dev))
-			pm_runtime_put_noidle(dev->dev);
+	if (r)
 		amdgpu_driver_unload_kms(dev);
-	}
 
 	return r;
 }
 
+static enum amd_ip_block_type
+	amdgpu_ip_get_block_type(struct amdgpu_device *adev, uint32_t ip)
+{
+	enum amd_ip_block_type type;
+
+	switch (ip) {
+	case AMDGPU_HW_IP_GFX:
+		type = AMD_IP_BLOCK_TYPE_GFX;
+		break;
+	case AMDGPU_HW_IP_COMPUTE:
+		type = AMD_IP_BLOCK_TYPE_GFX;
+		break;
+	case AMDGPU_HW_IP_DMA:
+		type = AMD_IP_BLOCK_TYPE_SDMA;
+		break;
+	case AMDGPU_HW_IP_UVD:
+	case AMDGPU_HW_IP_UVD_ENC:
+		type = AMD_IP_BLOCK_TYPE_UVD;
+		break;
+	case AMDGPU_HW_IP_VCE:
+		type = AMD_IP_BLOCK_TYPE_VCE;
+		break;
+	case AMDGPU_HW_IP_VCN_DEC:
+	case AMDGPU_HW_IP_VCN_ENC:
+		type = AMD_IP_BLOCK_TYPE_VCN;
+		break;
+	case AMDGPU_HW_IP_VCN_JPEG:
+		type = (amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_JPEG)) ?
+				   AMD_IP_BLOCK_TYPE_JPEG : AMD_IP_BLOCK_TYPE_VCN;
+		break;
+	default:
+		type = AMD_IP_BLOCK_TYPE_NUM;
+		break;
+	}
+
+	return type;
+}
+
 static int amdgpu_firmware_info(struct drm_amdgpu_info_firmware *fw_info,
 				struct drm_amdgpu_query_fw *query_fw,
 				struct amdgpu_device *adev)
@@ -197,8 +222,12 @@ static int amdgpu_firmware_info(struct drm_amdgpu_info_firmware *fw_info,
 		fw_info->ver = adev->uvd.fw_version;
 		fw_info->feature = 0;
 		break;
+	case AMDGPU_INFO_FW_VCN:
+		fw_info->ver = adev->vcn.fw_version;
+		fw_info->feature = 0;
+		break;
 	case AMDGPU_INFO_FW_GMC:
-		fw_info->ver = adev->mc.fw_version;
+		fw_info->ver = adev->gmc.fw_version;
 		fw_info->feature = 0;
 		break;
 	case AMDGPU_INFO_FW_GFX_ME:
@@ -217,6 +246,26 @@ static int amdgpu_firmware_info(struct drm_amdgpu_info_firmware *fw_info,
 		fw_info->ver = adev->gfx.rlc_fw_version;
 		fw_info->feature = adev->gfx.rlc_feature_version;
 		break;
+	case AMDGPU_INFO_FW_GFX_RLC_RESTORE_LIST_CNTL:
+		fw_info->ver = adev->gfx.rlc_srlc_fw_version;
+		fw_info->feature = adev->gfx.rlc_srlc_feature_version;
+		break;
+	case AMDGPU_INFO_FW_GFX_RLC_RESTORE_LIST_GPM_MEM:
+		fw_info->ver = adev->gfx.rlc_srlg_fw_version;
+		fw_info->feature = adev->gfx.rlc_srlg_feature_version;
+		break;
+	case AMDGPU_INFO_FW_GFX_RLC_RESTORE_LIST_SRM_MEM:
+		fw_info->ver = adev->gfx.rlc_srls_fw_version;
+		fw_info->feature = adev->gfx.rlc_srls_feature_version;
+		break;
+	case AMDGPU_INFO_FW_GFX_RLCP:
+		fw_info->ver = adev->gfx.rlcp_ucode_version;
+		fw_info->feature = adev->gfx.rlcp_ucode_feature_version;
+		break;
+	case AMDGPU_INFO_FW_GFX_RLCV:
+		fw_info->ver = adev->gfx.rlcv_ucode_version;
+		fw_info->feature = adev->gfx.rlcv_ucode_feature_version;
+		break;
 	case AMDGPU_INFO_FW_GFX_MEC:
 		if (query_fw->index == 0) {
 			fw_info->ver = adev->gfx.mec_fw_version;
@@ -231,6 +280,43 @@ static int amdgpu_firmware_info(struct drm_amdgpu_info_firmware *fw_info,
 		fw_info->ver = adev->pm.fw_version;
 		fw_info->feature = 0;
 		break;
+	case AMDGPU_INFO_FW_TA:
+		switch (query_fw->index) {
+		case TA_FW_TYPE_PSP_XGMI:
+			fw_info->ver = adev->psp.xgmi_context.context.bin_desc.fw_version;
+			fw_info->feature = adev->psp.xgmi_context.context
+						   .bin_desc.feature_version;
+			break;
+		case TA_FW_TYPE_PSP_RAS:
+			fw_info->ver = adev->psp.ras_context.context.bin_desc.fw_version;
+			fw_info->feature = adev->psp.ras_context.context
+						   .bin_desc.feature_version;
+			break;
+		case TA_FW_TYPE_PSP_HDCP:
+			fw_info->ver = adev->psp.hdcp_context.context.bin_desc.fw_version;
+			fw_info->feature = adev->psp.hdcp_context.context
+						   .bin_desc.feature_version;
+			break;
+		case TA_FW_TYPE_PSP_DTM:
+			fw_info->ver = adev->psp.dtm_context.context.bin_desc.fw_version;
+			fw_info->feature = adev->psp.dtm_context.context
+						   .bin_desc.feature_version;
+			break;
+		case TA_FW_TYPE_PSP_RAP:
+			fw_info->ver = adev->psp.rap_context.context.bin_desc.fw_version;
+			fw_info->feature = adev->psp.rap_context.context
+						   .bin_desc.feature_version;
+			break;
+		case TA_FW_TYPE_PSP_SECUREDISPLAY:
+			fw_info->ver = adev->psp.securedisplay_context.context.bin_desc.fw_version;
+			fw_info->feature =
+				adev->psp.securedisplay_context.context.bin_desc
+					.feature_version;
+			break;
+		default:
+			return -EINVAL;
+		}
+		break;
 	case AMDGPU_INFO_FW_SDMA:
 		if (query_fw->index >= adev->sdma.num_instances)
 			return -EINVAL;
@@ -238,12 +324,46 @@ static int amdgpu_firmware_info(struct drm_amdgpu_info_firmware *fw_info,
 		fw_info->feature = adev->sdma.instance[query_fw->index].feature_version;
 		break;
 	case AMDGPU_INFO_FW_SOS:
-		fw_info->ver = adev->psp.sos_fw_version;
-		fw_info->feature = adev->psp.sos_feature_version;
+		fw_info->ver = adev->psp.sos.fw_version;
+		fw_info->feature = adev->psp.sos.feature_version;
 		break;
 	case AMDGPU_INFO_FW_ASD:
-		fw_info->ver = adev->psp.asd_fw_version;
-		fw_info->feature = adev->psp.asd_feature_version;
+		fw_info->ver = adev->psp.asd_context.bin_desc.fw_version;
+		fw_info->feature = adev->psp.asd_context.bin_desc.feature_version;
+		break;
+	case AMDGPU_INFO_FW_DMCU:
+		fw_info->ver = adev->dm.dmcu_fw_version;
+		fw_info->feature = 0;
+		break;
+	case AMDGPU_INFO_FW_DMCUB:
+		fw_info->ver = adev->dm.dmcub_fw_version;
+		fw_info->feature = 0;
+		break;
+	case AMDGPU_INFO_FW_TOC:
+		fw_info->ver = adev->psp.toc.fw_version;
+		fw_info->feature = adev->psp.toc.feature_version;
+		break;
+	case AMDGPU_INFO_FW_CAP:
+		fw_info->ver = adev->psp.cap_fw_version;
+		fw_info->feature = adev->psp.cap_feature_version;
+		break;
+	case AMDGPU_INFO_FW_MES_KIQ:
+		fw_info->ver = adev->mes.kiq_version & AMDGPU_MES_VERSION_MASK;
+		fw_info->feature = (adev->mes.kiq_version & AMDGPU_MES_FEAT_VERSION_MASK)
+					>> AMDGPU_MES_FEAT_VERSION_SHIFT;
+		break;
+	case AMDGPU_INFO_FW_MES:
+		fw_info->ver = adev->mes.sched_version & AMDGPU_MES_VERSION_MASK;
+		fw_info->feature = (adev->mes.sched_version & AMDGPU_MES_FEAT_VERSION_MASK)
+					>> AMDGPU_MES_FEAT_VERSION_SHIFT;
+		break;
+	case AMDGPU_INFO_FW_IMU:
+		fw_info->ver = adev->gfx.imu_fw_version;
+		fw_info->feature = 0;
+		break;
+	case AMDGPU_INFO_FW_VPE:
+		fw_info->ver = adev->vpe.fw_version;
+		fw_info->feature = adev->vpe.feature_version;
 		break;
 	default:
 		return -EINVAL;
@@ -251,13 +371,237 @@ static int amdgpu_firmware_info(struct drm_amdgpu_info_firmware *fw_info,
 	return 0;
 }
 
+static int amdgpu_userq_metadata_info_gfx(struct amdgpu_device *adev,
+					  struct drm_amdgpu_info *info,
+					  struct drm_amdgpu_info_uq_metadata_gfx *meta)
+{
+	int ret = -EOPNOTSUPP;
+
+	if (adev->gfx.funcs->get_gfx_shadow_info) {
+		struct amdgpu_gfx_shadow_info shadow = {};
+
+		adev->gfx.funcs->get_gfx_shadow_info(adev, &shadow, true);
+		meta->shadow_size = shadow.shadow_size;
+		meta->shadow_alignment = shadow.shadow_alignment;
+		meta->csa_size = shadow.csa_size;
+		meta->csa_alignment = shadow.csa_alignment;
+		ret = 0;
+	}
+
+	return ret;
+}
+
+static int amdgpu_hw_ip_info(struct amdgpu_device *adev,
+			     struct drm_amdgpu_info *info,
+			     struct drm_amdgpu_info_hw_ip *result)
+{
+	uint32_t ib_start_alignment = 0;
+	uint32_t ib_size_alignment = 0;
+	enum amd_ip_block_type type;
+	unsigned int num_rings = 0;
+	uint32_t num_slots = 0;
+	unsigned int i, j;
+
+	if (info->query_hw_ip.ip_instance >= AMDGPU_HW_IP_INSTANCE_MAX_COUNT)
+		return -EINVAL;
+
+	switch (info->query_hw_ip.type) {
+	case AMDGPU_HW_IP_GFX:
+		type = AMD_IP_BLOCK_TYPE_GFX;
+		for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+			if (adev->gfx.gfx_ring[i].sched.ready &&
+			    !adev->gfx.gfx_ring[i].no_user_submission)
+				++num_rings;
+
+		if (!adev->gfx.disable_uq) {
+			for (i = 0; i < AMDGPU_MES_MAX_GFX_PIPES; i++)
+				num_slots += hweight32(adev->mes.gfx_hqd_mask[i]);
+		}
+
+		ib_start_alignment = 32;
+		ib_size_alignment = 32;
+		break;
+	case AMDGPU_HW_IP_COMPUTE:
+		type = AMD_IP_BLOCK_TYPE_GFX;
+		for (i = 0; i < adev->gfx.num_compute_rings; i++)
+			if (adev->gfx.compute_ring[i].sched.ready &&
+			    !adev->gfx.compute_ring[i].no_user_submission)
+				++num_rings;
+
+		if (!adev->sdma.disable_uq) {
+			for (i = 0; i < AMDGPU_MES_MAX_COMPUTE_PIPES; i++)
+				num_slots += hweight32(adev->mes.compute_hqd_mask[i]);
+		}
+
+		ib_start_alignment = 32;
+		ib_size_alignment = 32;
+		break;
+	case AMDGPU_HW_IP_DMA:
+		type = AMD_IP_BLOCK_TYPE_SDMA;
+		for (i = 0; i < adev->sdma.num_instances; i++)
+			if (adev->sdma.instance[i].ring.sched.ready &&
+			    !adev->sdma.instance[i].ring.no_user_submission)
+				++num_rings;
+
+		if (!adev->gfx.disable_uq) {
+			for (i = 0; i < AMDGPU_MES_MAX_SDMA_PIPES; i++)
+				num_slots += hweight32(adev->mes.sdma_hqd_mask[i]);
+		}
+
+		ib_start_alignment = 256;
+		ib_size_alignment = 4;
+		break;
+	case AMDGPU_HW_IP_UVD:
+		type = AMD_IP_BLOCK_TYPE_UVD;
+		for (i = 0; i < adev->uvd.num_uvd_inst; i++) {
+			if (adev->uvd.harvest_config & (1 << i))
+				continue;
+
+			if (adev->uvd.inst[i].ring.sched.ready &&
+			    !adev->uvd.inst[i].ring.no_user_submission)
+				++num_rings;
+		}
+		ib_start_alignment = 256;
+		ib_size_alignment = 64;
+		break;
+	case AMDGPU_HW_IP_VCE:
+		type = AMD_IP_BLOCK_TYPE_VCE;
+		for (i = 0; i < adev->vce.num_rings; i++)
+			if (adev->vce.ring[i].sched.ready &&
+			    !adev->vce.ring[i].no_user_submission)
+				++num_rings;
+		ib_start_alignment = 256;
+		ib_size_alignment = 4;
+		break;
+	case AMDGPU_HW_IP_UVD_ENC:
+		type = AMD_IP_BLOCK_TYPE_UVD;
+		for (i = 0; i < adev->uvd.num_uvd_inst; i++) {
+			if (adev->uvd.harvest_config & (1 << i))
+				continue;
+
+			for (j = 0; j < adev->uvd.num_enc_rings; j++)
+				if (adev->uvd.inst[i].ring_enc[j].sched.ready &&
+				    !adev->uvd.inst[i].ring_enc[j].no_user_submission)
+					++num_rings;
+		}
+		ib_start_alignment = 256;
+		ib_size_alignment = 4;
+		break;
+	case AMDGPU_HW_IP_VCN_DEC:
+		type = AMD_IP_BLOCK_TYPE_VCN;
+		for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+			if (adev->vcn.harvest_config & (1 << i))
+				continue;
+
+			if (adev->vcn.inst[i].ring_dec.sched.ready &&
+			    !adev->vcn.inst[i].ring_dec.no_user_submission)
+				++num_rings;
+		}
+		ib_start_alignment = 256;
+		ib_size_alignment = 64;
+		break;
+	case AMDGPU_HW_IP_VCN_ENC:
+		type = AMD_IP_BLOCK_TYPE_VCN;
+		for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+			if (adev->vcn.harvest_config & (1 << i))
+				continue;
+
+			for (j = 0; j < adev->vcn.inst[i].num_enc_rings; j++)
+				if (adev->vcn.inst[i].ring_enc[j].sched.ready &&
+				    !adev->vcn.inst[i].ring_enc[j].no_user_submission)
+					++num_rings;
+		}
+		ib_start_alignment = 256;
+		ib_size_alignment = 4;
+		break;
+	case AMDGPU_HW_IP_VCN_JPEG:
+		type = (amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_JPEG)) ?
+			AMD_IP_BLOCK_TYPE_JPEG : AMD_IP_BLOCK_TYPE_VCN;
+
+		for (i = 0; i < adev->jpeg.num_jpeg_inst; i++) {
+			if (adev->jpeg.harvest_config & (1 << i))
+				continue;
+
+			for (j = 0; j < adev->jpeg.num_jpeg_rings; j++)
+				if (adev->jpeg.inst[i].ring_dec[j].sched.ready &&
+				    !adev->jpeg.inst[i].ring_dec[j].no_user_submission)
+					++num_rings;
+		}
+		ib_start_alignment = 256;
+		ib_size_alignment = 64;
+		break;
+	case AMDGPU_HW_IP_VPE:
+		type = AMD_IP_BLOCK_TYPE_VPE;
+		if (adev->vpe.ring.sched.ready &&
+		    !adev->vpe.ring.no_user_submission)
+			++num_rings;
+		ib_start_alignment = 256;
+		ib_size_alignment = 4;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	for (i = 0; i < adev->num_ip_blocks; i++)
+		if (adev->ip_blocks[i].version->type == type &&
+		    adev->ip_blocks[i].status.valid)
+			break;
+
+	if (i == adev->num_ip_blocks)
+		return 0;
+
+	num_rings = min(amdgpu_ctx_num_entities[info->query_hw_ip.type],
+			num_rings);
+
+	result->hw_ip_version_major = adev->ip_blocks[i].version->major;
+	result->hw_ip_version_minor = adev->ip_blocks[i].version->minor;
+
+	if (adev->asic_type >= CHIP_VEGA10) {
+		switch (type) {
+		case AMD_IP_BLOCK_TYPE_GFX:
+			result->ip_discovery_version =
+				IP_VERSION_MAJ_MIN_REV(amdgpu_ip_version(adev, GC_HWIP, 0));
+			break;
+		case AMD_IP_BLOCK_TYPE_SDMA:
+			result->ip_discovery_version =
+				IP_VERSION_MAJ_MIN_REV(amdgpu_ip_version(adev, SDMA0_HWIP, 0));
+			break;
+		case AMD_IP_BLOCK_TYPE_UVD:
+		case AMD_IP_BLOCK_TYPE_VCN:
+		case AMD_IP_BLOCK_TYPE_JPEG:
+			result->ip_discovery_version =
+				IP_VERSION_MAJ_MIN_REV(amdgpu_ip_version(adev, UVD_HWIP, 0));
+			break;
+		case AMD_IP_BLOCK_TYPE_VCE:
+			result->ip_discovery_version =
+				IP_VERSION_MAJ_MIN_REV(amdgpu_ip_version(adev, VCE_HWIP, 0));
+			break;
+		case AMD_IP_BLOCK_TYPE_VPE:
+			result->ip_discovery_version =
+				IP_VERSION_MAJ_MIN_REV(amdgpu_ip_version(adev, VPE_HWIP, 0));
+			break;
+		default:
+			result->ip_discovery_version = 0;
+			break;
+		}
+	} else {
+		result->ip_discovery_version = 0;
+	}
+	result->capabilities_flags = 0;
+	result->available_rings = (1 << num_rings) - 1;
+	result->userq_num_slots = num_slots;
+	result->ib_start_alignment = ib_start_alignment;
+	result->ib_size_alignment = ib_size_alignment;
+	return 0;
+}
+
 /*
  * Userspace get information ioctl
  */
 /**
  * amdgpu_info_ioctl - answer a device specific request.
  *
- * @adev: amdgpu device pointer
+ * @dev: drm device pointer
  * @data: request object
  * @filp: drm filp
  *
@@ -266,24 +610,26 @@ static int amdgpu_firmware_info(struct drm_amdgpu_info_firmware *fw_info,
  * etc. (all asics).
  * Returns 0 on success, -EINVAL on failure.
  */
-static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
+int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 {
-	struct amdgpu_device *adev = dev->dev_private;
-	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct drm_amdgpu_info *info = data;
 	struct amdgpu_mode_info *minfo = &adev->mode_info;
 	void __user *out = (void __user *)(uintptr_t)info->return_pointer;
+	struct amdgpu_fpriv *fpriv;
+	struct amdgpu_ip_block *ip_block;
+	enum amd_ip_block_type type;
+	struct amdgpu_xcp *xcp;
+	u32 count, inst_mask;
 	uint32_t size = info->return_size;
 	struct drm_crtc *crtc;
 	uint32_t ui32 = 0;
 	uint64_t ui64 = 0;
-	int i, found;
+	int i, found, ret;
 	int ui32_size = sizeof(ui32);
 
 	if (!info->return_size || !info->return_pointer)
 		return -EINVAL;
-	if (amdgpu_kms_vram_lost(adev, fpriv))
-		return -ENODEV;
 
 	switch (info->query) {
 	case AMDGPU_INFO_ACCEL_WORKING:
@@ -294,6 +640,7 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 			crtc = (struct drm_crtc *)minfo->crtcs[i];
 			if (crtc && crtc->base.id == info->mode_crtc.id) {
 				struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
 				ui32 = amdgpu_crtc->crtc_id;
 				found = 1;
 				break;
@@ -306,125 +653,82 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 		return copy_to_user(out, &ui32, min(size, 4u)) ? -EFAULT : 0;
 	case AMDGPU_INFO_HW_IP_INFO: {
 		struct drm_amdgpu_info_hw_ip ip = {};
-		enum amd_ip_block_type type;
-		uint32_t ring_mask = 0;
-		uint32_t ib_start_alignment = 0;
-		uint32_t ib_size_alignment = 0;
 
-		if (info->query_hw_ip.ip_instance >= AMDGPU_HW_IP_INSTANCE_MAX_COUNT)
-			return -EINVAL;
+		ret = amdgpu_hw_ip_info(adev, info, &ip);
+		if (ret)
+			return ret;
 
-		switch (info->query_hw_ip.type) {
-		case AMDGPU_HW_IP_GFX:
-			type = AMD_IP_BLOCK_TYPE_GFX;
-			for (i = 0; i < adev->gfx.num_gfx_rings; i++)
-				ring_mask |= ((adev->gfx.gfx_ring[i].ready ? 1 : 0) << i);
-			ib_start_alignment = AMDGPU_GPU_PAGE_SIZE;
-			ib_size_alignment = 8;
-			break;
-		case AMDGPU_HW_IP_COMPUTE:
-			type = AMD_IP_BLOCK_TYPE_GFX;
-			for (i = 0; i < adev->gfx.num_compute_rings; i++)
-				ring_mask |= ((adev->gfx.compute_ring[i].ready ? 1 : 0) << i);
-			ib_start_alignment = AMDGPU_GPU_PAGE_SIZE;
-			ib_size_alignment = 8;
-			break;
-		case AMDGPU_HW_IP_DMA:
-			type = AMD_IP_BLOCK_TYPE_SDMA;
-			for (i = 0; i < adev->sdma.num_instances; i++)
-				ring_mask |= ((adev->sdma.instance[i].ring.ready ? 1 : 0) << i);
-			ib_start_alignment = AMDGPU_GPU_PAGE_SIZE;
-			ib_size_alignment = 1;
-			break;
-		case AMDGPU_HW_IP_UVD:
-			type = AMD_IP_BLOCK_TYPE_UVD;
-			ring_mask = adev->uvd.ring.ready ? 1 : 0;
-			ib_start_alignment = AMDGPU_GPU_PAGE_SIZE;
-			ib_size_alignment = 16;
-			break;
-		case AMDGPU_HW_IP_VCE:
-			type = AMD_IP_BLOCK_TYPE_VCE;
-			for (i = 0; i < adev->vce.num_rings; i++)
-				ring_mask |= ((adev->vce.ring[i].ready ? 1 : 0) << i);
-			ib_start_alignment = AMDGPU_GPU_PAGE_SIZE;
-			ib_size_alignment = 1;
-			break;
-		case AMDGPU_HW_IP_UVD_ENC:
-			type = AMD_IP_BLOCK_TYPE_UVD;
-			for (i = 0; i < adev->uvd.num_enc_rings; i++)
-				ring_mask |= ((adev->uvd.ring_enc[i].ready ? 1 : 0) << i);
-			ib_start_alignment = AMDGPU_GPU_PAGE_SIZE;
-			ib_size_alignment = 1;
-			break;
-		case AMDGPU_HW_IP_VCN_DEC:
-			type = AMD_IP_BLOCK_TYPE_VCN;
-			ring_mask = adev->vcn.ring_dec.ready ? 1 : 0;
-			ib_start_alignment = AMDGPU_GPU_PAGE_SIZE;
-			ib_size_alignment = 16;
-			break;
-		case AMDGPU_HW_IP_VCN_ENC:
-			type = AMD_IP_BLOCK_TYPE_VCN;
-			for (i = 0; i < adev->vcn.num_enc_rings; i++)
-				ring_mask |= ((adev->vcn.ring_enc[i].ready ? 1 : 0) << i);
-			ib_start_alignment = AMDGPU_GPU_PAGE_SIZE;
-			ib_size_alignment = 1;
-			break;
-		default:
+		ret = copy_to_user(out, &ip, min_t(size_t, size, sizeof(ip)));
+		return ret ? -EFAULT : 0;
+	}
+	case AMDGPU_INFO_HW_IP_COUNT: {
+		fpriv = (struct amdgpu_fpriv *)filp->driver_priv;
+		type = amdgpu_ip_get_block_type(adev, info->query_hw_ip.type);
+		ip_block = amdgpu_device_ip_get_ip_block(adev, type);
+
+		if (!ip_block || !ip_block->status.valid)
 			return -EINVAL;
-		}
 
-		for (i = 0; i < adev->num_ip_blocks; i++) {
-			if (adev->ip_blocks[i].version->type == type &&
-			    adev->ip_blocks[i].status.valid) {
-				ip.hw_ip_version_major = adev->ip_blocks[i].version->major;
-				ip.hw_ip_version_minor = adev->ip_blocks[i].version->minor;
-				ip.capabilities_flags = 0;
-				ip.available_rings = ring_mask;
-				ip.ib_start_alignment = ib_start_alignment;
-				ip.ib_size_alignment = ib_size_alignment;
+		if (adev->xcp_mgr && adev->xcp_mgr->num_xcps > 0 &&
+		    fpriv->xcp_id < adev->xcp_mgr->num_xcps) {
+			xcp = &adev->xcp_mgr->xcp[fpriv->xcp_id];
+			switch (type) {
+			case AMD_IP_BLOCK_TYPE_GFX:
+				ret = amdgpu_xcp_get_inst_details(xcp, AMDGPU_XCP_GFX, &inst_mask);
+				if (ret)
+					return ret;
+				count = hweight32(inst_mask);
 				break;
+			case AMD_IP_BLOCK_TYPE_SDMA:
+				ret = amdgpu_xcp_get_inst_details(xcp, AMDGPU_XCP_SDMA, &inst_mask);
+				if (ret)
+					return ret;
+				count = hweight32(inst_mask);
+				break;
+			case AMD_IP_BLOCK_TYPE_JPEG:
+				ret = amdgpu_xcp_get_inst_details(xcp, AMDGPU_XCP_VCN, &inst_mask);
+				if (ret)
+					return ret;
+				count = hweight32(inst_mask) * adev->jpeg.num_jpeg_rings;
+				break;
+			case AMD_IP_BLOCK_TYPE_VCN:
+				ret = amdgpu_xcp_get_inst_details(xcp, AMDGPU_XCP_VCN, &inst_mask);
+				if (ret)
+					return ret;
+				count = hweight32(inst_mask);
+				break;
+			default:
+				return -EINVAL;
 			}
+
+			return copy_to_user(out, &count, min(size, 4u)) ? -EFAULT : 0;
 		}
-		return copy_to_user(out, &ip,
-				    min((size_t)size, sizeof(ip))) ? -EFAULT : 0;
-	}
-	case AMDGPU_INFO_HW_IP_COUNT: {
-		enum amd_ip_block_type type;
-		uint32_t count = 0;
 
-		switch (info->query_hw_ip.type) {
-		case AMDGPU_HW_IP_GFX:
-			type = AMD_IP_BLOCK_TYPE_GFX;
-			break;
-		case AMDGPU_HW_IP_COMPUTE:
-			type = AMD_IP_BLOCK_TYPE_GFX;
+		switch (type) {
+		case AMD_IP_BLOCK_TYPE_GFX:
+		case AMD_IP_BLOCK_TYPE_VCE:
+			count = 1;
 			break;
-		case AMDGPU_HW_IP_DMA:
-			type = AMD_IP_BLOCK_TYPE_SDMA;
+		case AMD_IP_BLOCK_TYPE_SDMA:
+			count = adev->sdma.num_instances;
 			break;
-		case AMDGPU_HW_IP_UVD:
-			type = AMD_IP_BLOCK_TYPE_UVD;
-			break;
-		case AMDGPU_HW_IP_VCE:
-			type = AMD_IP_BLOCK_TYPE_VCE;
+		case AMD_IP_BLOCK_TYPE_JPEG:
+			count = adev->jpeg.num_jpeg_inst * adev->jpeg.num_jpeg_rings;
 			break;
-		case AMDGPU_HW_IP_UVD_ENC:
-			type = AMD_IP_BLOCK_TYPE_UVD;
+		case AMD_IP_BLOCK_TYPE_VCN:
+			count = adev->vcn.num_vcn_inst;
 			break;
-		case AMDGPU_HW_IP_VCN_DEC:
-		case AMDGPU_HW_IP_VCN_ENC:
-			type = AMD_IP_BLOCK_TYPE_VCN;
+		case AMD_IP_BLOCK_TYPE_UVD:
+			count = adev->uvd.num_uvd_inst;
 			break;
+		/* For all other IP block types not listed in the switch statement
+		 * the ip status is valid here and the instance count is one.
+		 */
 		default:
-			return -EINVAL;
+			count = 1;
+			break;
 		}
 
-		for (i = 0; i < adev->num_ip_blocks; i++)
-			if (adev->ip_blocks[i].version->type == type &&
-			    adev->ip_blocks[i].status.valid &&
-			    count < AMDGPU_HW_IP_INSTANCE_MAX_COUNT)
-				count++;
-
 		return copy_to_user(out, &count, min(size, 4u)) ? -EFAULT : 0;
 	}
 	case AMDGPU_INFO_TIMESTAMP:
@@ -432,7 +736,6 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
 	case AMDGPU_INFO_FW_VERSION: {
 		struct drm_amdgpu_info_firmware fw_info;
-		int ret;
 
 		/* We only support one instance of each IP block right now. */
 		if (info->query_fw.ip_instance != 0)
@@ -455,68 +758,71 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 		ui64 = atomic64_read(&adev->num_vram_cpu_page_faults);
 		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
 	case AMDGPU_INFO_VRAM_USAGE:
-		ui64 = amdgpu_vram_mgr_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
+		ui64 = ttm_resource_manager_usage(&adev->mman.vram_mgr.manager);
 		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
 	case AMDGPU_INFO_VIS_VRAM_USAGE:
-		ui64 = amdgpu_vram_mgr_vis_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
+		ui64 = amdgpu_vram_mgr_vis_usage(&adev->mman.vram_mgr);
 		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
 	case AMDGPU_INFO_GTT_USAGE:
-		ui64 = amdgpu_gtt_mgr_usage(&adev->mman.bdev.man[TTM_PL_TT]);
+		ui64 = ttm_resource_manager_usage(&adev->mman.gtt_mgr.manager);
 		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
 	case AMDGPU_INFO_GDS_CONFIG: {
 		struct drm_amdgpu_info_gds gds_info;
 
 		memset(&gds_info, 0, sizeof(gds_info));
-		gds_info.gds_gfx_partition_size = adev->gds.mem.gfx_partition_size >> AMDGPU_GDS_SHIFT;
-		gds_info.compute_partition_size = adev->gds.mem.cs_partition_size >> AMDGPU_GDS_SHIFT;
-		gds_info.gds_total_size = adev->gds.mem.total_size >> AMDGPU_GDS_SHIFT;
-		gds_info.gws_per_gfx_partition = adev->gds.gws.gfx_partition_size >> AMDGPU_GWS_SHIFT;
-		gds_info.gws_per_compute_partition = adev->gds.gws.cs_partition_size >> AMDGPU_GWS_SHIFT;
-		gds_info.oa_per_gfx_partition = adev->gds.oa.gfx_partition_size >> AMDGPU_OA_SHIFT;
-		gds_info.oa_per_compute_partition = adev->gds.oa.cs_partition_size >> AMDGPU_OA_SHIFT;
+		gds_info.compute_partition_size = adev->gds.gds_size;
+		gds_info.gds_total_size = adev->gds.gds_size;
+		gds_info.gws_per_compute_partition = adev->gds.gws_size;
+		gds_info.oa_per_compute_partition = adev->gds.oa_size;
 		return copy_to_user(out, &gds_info,
 				    min((size_t)size, sizeof(gds_info))) ? -EFAULT : 0;
 	}
 	case AMDGPU_INFO_VRAM_GTT: {
 		struct drm_amdgpu_info_vram_gtt vram_gtt;
 
-		vram_gtt.vram_size = adev->mc.real_vram_size;
-		vram_gtt.vram_size -= adev->vram_pin_size;
-		vram_gtt.vram_cpu_accessible_size = adev->mc.visible_vram_size;
-		vram_gtt.vram_cpu_accessible_size -= (adev->vram_pin_size - adev->invisible_pin_size);
-		vram_gtt.gtt_size = adev->mman.bdev.man[TTM_PL_TT].size;
-		vram_gtt.gtt_size *= PAGE_SIZE;
-		vram_gtt.gtt_size -= adev->gart_pin_size;
+		vram_gtt.vram_size = adev->gmc.real_vram_size -
+			atomic64_read(&adev->vram_pin_size) -
+			AMDGPU_VM_RESERVED_VRAM;
+		vram_gtt.vram_cpu_accessible_size =
+			min(adev->gmc.visible_vram_size -
+			    atomic64_read(&adev->visible_pin_size),
+			    vram_gtt.vram_size);
+		vram_gtt.gtt_size = ttm_manager_type(&adev->mman.bdev, TTM_PL_TT)->size;
+		vram_gtt.gtt_size -= atomic64_read(&adev->gart_pin_size);
 		return copy_to_user(out, &vram_gtt,
 				    min((size_t)size, sizeof(vram_gtt))) ? -EFAULT : 0;
 	}
 	case AMDGPU_INFO_MEMORY: {
 		struct drm_amdgpu_memory_info mem;
+		struct ttm_resource_manager *gtt_man =
+			&adev->mman.gtt_mgr.manager;
+		struct ttm_resource_manager *vram_man =
+			&adev->mman.vram_mgr.manager;
 
 		memset(&mem, 0, sizeof(mem));
-		mem.vram.total_heap_size = adev->mc.real_vram_size;
-		mem.vram.usable_heap_size =
-			adev->mc.real_vram_size - adev->vram_pin_size;
+		mem.vram.total_heap_size = adev->gmc.real_vram_size;
+		mem.vram.usable_heap_size = adev->gmc.real_vram_size -
+			atomic64_read(&adev->vram_pin_size) -
+			AMDGPU_VM_RESERVED_VRAM;
 		mem.vram.heap_usage =
-			amdgpu_vram_mgr_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
+			ttm_resource_manager_usage(vram_man);
 		mem.vram.max_allocation = mem.vram.usable_heap_size * 3 / 4;
 
 		mem.cpu_accessible_vram.total_heap_size =
-			adev->mc.visible_vram_size;
+			adev->gmc.visible_vram_size;
 		mem.cpu_accessible_vram.usable_heap_size =
-			adev->mc.visible_vram_size -
-			(adev->vram_pin_size - adev->invisible_pin_size);
+			min(adev->gmc.visible_vram_size -
+			    atomic64_read(&adev->visible_pin_size),
+			    mem.vram.usable_heap_size);
 		mem.cpu_accessible_vram.heap_usage =
-			amdgpu_vram_mgr_vis_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
+			amdgpu_vram_mgr_vis_usage(&adev->mman.vram_mgr);
 		mem.cpu_accessible_vram.max_allocation =
 			mem.cpu_accessible_vram.usable_heap_size * 3 / 4;
 
-		mem.gtt.total_heap_size = adev->mman.bdev.man[TTM_PL_TT].size;
-		mem.gtt.total_heap_size *= PAGE_SIZE;
-		mem.gtt.usable_heap_size = mem.gtt.total_heap_size
-			- adev->gart_pin_size;
-		mem.gtt.heap_usage =
-			amdgpu_gtt_mgr_usage(&adev->mman.bdev.man[TTM_PL_TT]);
+		mem.gtt.total_heap_size = gtt_man->size;
+		mem.gtt.usable_heap_size = mem.gtt.total_heap_size -
+			atomic64_read(&adev->gart_pin_size);
+		mem.gtt.heap_usage = ttm_resource_manager_usage(gtt_man);
 		mem.gtt.max_allocation = mem.gtt.usable_heap_size * 3 / 4;
 
 		return copy_to_user(out, &mem,
@@ -524,110 +830,218 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 				    ? -EFAULT : 0;
 	}
 	case AMDGPU_INFO_READ_MMR_REG: {
-		unsigned n, alloc_size;
+		int ret = 0;
+		unsigned int n, alloc_size;
 		uint32_t *regs;
-		unsigned se_num = (info->read_mmr_reg.instance >>
+		unsigned int se_num = (info->read_mmr_reg.instance >>
 				   AMDGPU_INFO_MMR_SE_INDEX_SHIFT) &
 				  AMDGPU_INFO_MMR_SE_INDEX_MASK;
-		unsigned sh_num = (info->read_mmr_reg.instance >>
+		unsigned int sh_num = (info->read_mmr_reg.instance >>
 				   AMDGPU_INFO_MMR_SH_INDEX_SHIFT) &
 				  AMDGPU_INFO_MMR_SH_INDEX_MASK;
 
+		if (!down_read_trylock(&adev->reset_domain->sem))
+			return -ENOENT;
+
 		/* set full masks if the userspace set all bits
-		 * in the bitfields */
-		if (se_num == AMDGPU_INFO_MMR_SE_INDEX_MASK)
+		 * in the bitfields
+		 */
+		if (se_num == AMDGPU_INFO_MMR_SE_INDEX_MASK) {
 			se_num = 0xffffffff;
-		if (sh_num == AMDGPU_INFO_MMR_SH_INDEX_MASK)
+		} else if (se_num >= AMDGPU_GFX_MAX_SE) {
+			ret = -EINVAL;
+			goto out;
+		}
+
+		if (sh_num == AMDGPU_INFO_MMR_SH_INDEX_MASK) {
 			sh_num = 0xffffffff;
+		} else if (sh_num >= AMDGPU_GFX_MAX_SH_PER_SE) {
+			ret = -EINVAL;
+			goto out;
+		}
+
+		if (info->read_mmr_reg.count > 128) {
+			ret = -EINVAL;
+			goto out;
+		}
 
 		regs = kmalloc_array(info->read_mmr_reg.count, sizeof(*regs), GFP_KERNEL);
-		if (!regs)
-			return -ENOMEM;
+		if (!regs) {
+			ret = -ENOMEM;
+			goto out;
+		}
+
 		alloc_size = info->read_mmr_reg.count * sizeof(*regs);
 
-		for (i = 0; i < info->read_mmr_reg.count; i++)
+		amdgpu_gfx_off_ctrl(adev, false);
+		for (i = 0; i < info->read_mmr_reg.count; i++) {
 			if (amdgpu_asic_read_register(adev, se_num, sh_num,
 						      info->read_mmr_reg.dword_offset + i,
 						      &regs[i])) {
 				DRM_DEBUG_KMS("unallowed offset %#x\n",
 					      info->read_mmr_reg.dword_offset + i);
 				kfree(regs);
-				return -EFAULT;
+				amdgpu_gfx_off_ctrl(adev, true);
+				ret = -EFAULT;
+				goto out;
 			}
+		}
+		amdgpu_gfx_off_ctrl(adev, true);
 		n = copy_to_user(out, regs, min(size, alloc_size));
 		kfree(regs);
-		return n ? -EFAULT : 0;
+		ret = (n ? -EFAULT : 0);
+out:
+		up_read(&adev->reset_domain->sem);
+		return ret;
 	}
 	case AMDGPU_INFO_DEV_INFO: {
-		struct drm_amdgpu_info_device dev_info = {};
-
-		dev_info.device_id = dev->pdev->device;
-		dev_info.chip_rev = adev->rev_id;
-		dev_info.external_rev = adev->external_rev_id;
-		dev_info.pci_rev = dev->pdev->revision;
-		dev_info.family = adev->family;
-		dev_info.num_shader_engines = adev->gfx.config.max_shader_engines;
-		dev_info.num_shader_arrays_per_engine = adev->gfx.config.max_sh_per_se;
+		struct drm_amdgpu_info_device *dev_info;
+		uint64_t vm_size;
+		uint32_t pcie_gen_mask, pcie_width_mask;
+
+		dev_info = kzalloc(sizeof(*dev_info), GFP_KERNEL);
+		if (!dev_info)
+			return -ENOMEM;
+
+		dev_info->device_id = adev->pdev->device;
+		dev_info->chip_rev = adev->rev_id;
+		dev_info->external_rev = adev->external_rev_id;
+		dev_info->pci_rev = adev->pdev->revision;
+		dev_info->family = adev->family;
+		dev_info->num_shader_engines = adev->gfx.config.max_shader_engines;
+		dev_info->num_shader_arrays_per_engine = adev->gfx.config.max_sh_per_se;
 		/* return all clocks in KHz */
-		dev_info.gpu_counter_freq = amdgpu_asic_get_xclk(adev) * 10;
+		dev_info->gpu_counter_freq = amdgpu_asic_get_xclk(adev) * 10;
 		if (adev->pm.dpm_enabled) {
-			dev_info.max_engine_clock = amdgpu_dpm_get_sclk(adev, false) * 10;
-			dev_info.max_memory_clock = amdgpu_dpm_get_mclk(adev, false) * 10;
+			dev_info->max_engine_clock = amdgpu_dpm_get_sclk(adev, false) * 10;
+			dev_info->max_memory_clock = amdgpu_dpm_get_mclk(adev, false) * 10;
+			dev_info->min_engine_clock = amdgpu_dpm_get_sclk(adev, true) * 10;
+			dev_info->min_memory_clock = amdgpu_dpm_get_mclk(adev, true) * 10;
 		} else {
-			dev_info.max_engine_clock = adev->clock.default_sclk * 10;
-			dev_info.max_memory_clock = adev->clock.default_mclk * 10;
+			dev_info->max_engine_clock =
+				dev_info->min_engine_clock =
+					adev->clock.default_sclk * 10;
+			dev_info->max_memory_clock =
+				dev_info->min_memory_clock =
+					adev->clock.default_mclk * 10;
 		}
-		dev_info.enabled_rb_pipes_mask = adev->gfx.config.backend_enable_mask;
-		dev_info.num_rb_pipes = adev->gfx.config.max_backends_per_se *
+		dev_info->enabled_rb_pipes_mask = adev->gfx.config.backend_enable_mask;
+		dev_info->num_rb_pipes = adev->gfx.config.max_backends_per_se *
 			adev->gfx.config.max_shader_engines;
-		dev_info.num_hw_gfx_contexts = adev->gfx.config.max_hw_contexts;
-		dev_info._pad = 0;
-		dev_info.ids_flags = 0;
+		dev_info->num_hw_gfx_contexts = adev->gfx.config.max_hw_contexts;
+		dev_info->ids_flags = 0;
 		if (adev->flags & AMD_IS_APU)
-			dev_info.ids_flags |= AMDGPU_IDS_FLAGS_FUSION;
-		if (amdgpu_sriov_vf(adev))
-			dev_info.ids_flags |= AMDGPU_IDS_FLAGS_PREEMPTION;
-		dev_info.virtual_address_offset = AMDGPU_VA_RESERVED_SIZE;
-		dev_info.virtual_address_max = (uint64_t)adev->vm_manager.max_pfn * AMDGPU_GPU_PAGE_SIZE;
-		dev_info.virtual_address_alignment = max((int)PAGE_SIZE, AMDGPU_GPU_PAGE_SIZE);
-		dev_info.pte_fragment_size = (1 << adev->vm_manager.fragment_size) * AMDGPU_GPU_PAGE_SIZE;
-		dev_info.gart_page_size = AMDGPU_GPU_PAGE_SIZE;
-		dev_info.cu_active_number = adev->gfx.cu_info.number;
-		dev_info.cu_ao_mask = adev->gfx.cu_info.ao_cu_mask;
-		dev_info.ce_ram_size = adev->gfx.ce_ram_size;
-		memcpy(&dev_info.cu_ao_bitmap[0], &adev->gfx.cu_info.ao_cu_bitmap[0],
+			dev_info->ids_flags |= AMDGPU_IDS_FLAGS_FUSION;
+		if (adev->gfx.mcbp)
+			dev_info->ids_flags |= AMDGPU_IDS_FLAGS_PREEMPTION;
+		if (amdgpu_is_tmz(adev))
+			dev_info->ids_flags |= AMDGPU_IDS_FLAGS_TMZ;
+		if (adev->gfx.config.ta_cntl2_truncate_coord_mode)
+			dev_info->ids_flags |= AMDGPU_IDS_FLAGS_CONFORMANT_TRUNC_COORD;
+
+		/* Gang submit is not supported under SRIOV currently */
+		if (!amdgpu_sriov_vf(adev))
+			dev_info->ids_flags |= AMDGPU_IDS_FLAGS_GANG_SUBMIT;
+
+		if (amdgpu_passthrough(adev))
+			dev_info->ids_flags |= (AMDGPU_IDS_FLAGS_MODE_PT <<
+						AMDGPU_IDS_FLAGS_MODE_SHIFT) &
+						AMDGPU_IDS_FLAGS_MODE_MASK;
+		else if (amdgpu_sriov_vf(adev))
+			dev_info->ids_flags |= (AMDGPU_IDS_FLAGS_MODE_VF <<
+						AMDGPU_IDS_FLAGS_MODE_SHIFT) &
+						AMDGPU_IDS_FLAGS_MODE_MASK;
+
+		vm_size = adev->vm_manager.max_pfn * AMDGPU_GPU_PAGE_SIZE;
+		vm_size -= AMDGPU_VA_RESERVED_TOP;
+
+		/* Older VCE FW versions are buggy and can handle only 40bits */
+		if (adev->vce.fw_version &&
+		    adev->vce.fw_version < AMDGPU_VCE_FW_53_45)
+			vm_size = min(vm_size, 1ULL << 40);
+
+		dev_info->virtual_address_offset = AMDGPU_VA_RESERVED_BOTTOM;
+		dev_info->virtual_address_max =
+			min(vm_size, AMDGPU_GMC_HOLE_START);
+
+		if (vm_size > AMDGPU_GMC_HOLE_START) {
+			dev_info->high_va_offset = AMDGPU_GMC_HOLE_END;
+			dev_info->high_va_max = AMDGPU_GMC_HOLE_END | vm_size;
+		}
+		dev_info->virtual_address_alignment = max_t(u32, PAGE_SIZE, AMDGPU_GPU_PAGE_SIZE);
+		dev_info->pte_fragment_size = (1 << adev->vm_manager.fragment_size) * AMDGPU_GPU_PAGE_SIZE;
+		dev_info->gart_page_size = max_t(u32, PAGE_SIZE, AMDGPU_GPU_PAGE_SIZE);
+		dev_info->cu_active_number = adev->gfx.cu_info.number;
+		dev_info->cu_ao_mask = adev->gfx.cu_info.ao_cu_mask;
+		dev_info->ce_ram_size = adev->gfx.ce_ram_size;
+		memcpy(&dev_info->cu_ao_bitmap[0], &adev->gfx.cu_info.ao_cu_bitmap[0],
 		       sizeof(adev->gfx.cu_info.ao_cu_bitmap));
-		memcpy(&dev_info.cu_bitmap[0], &adev->gfx.cu_info.bitmap[0],
-		       sizeof(adev->gfx.cu_info.bitmap));
-		dev_info.vram_type = adev->mc.vram_type;
-		dev_info.vram_bit_width = adev->mc.vram_width;
-		dev_info.vce_harvest_config = adev->vce.harvest_config;
-		dev_info.gc_double_offchip_lds_buf =
+		memcpy(&dev_info->cu_bitmap[0], &adev->gfx.cu_info.bitmap[0],
+		       sizeof(dev_info->cu_bitmap));
+		dev_info->vram_type = adev->gmc.vram_type;
+		dev_info->vram_bit_width = adev->gmc.vram_width;
+		dev_info->vce_harvest_config = adev->vce.harvest_config;
+		dev_info->gc_double_offchip_lds_buf =
 			adev->gfx.config.double_offchip_lds_buf;
-
-		if (amdgpu_ngg) {
-			dev_info.prim_buf_gpu_addr = adev->gfx.ngg.buf[NGG_PRIM].gpu_addr;
-			dev_info.prim_buf_size = adev->gfx.ngg.buf[NGG_PRIM].size;
-			dev_info.pos_buf_gpu_addr = adev->gfx.ngg.buf[NGG_POS].gpu_addr;
-			dev_info.pos_buf_size = adev->gfx.ngg.buf[NGG_POS].size;
-			dev_info.cntl_sb_buf_gpu_addr = adev->gfx.ngg.buf[NGG_CNTL].gpu_addr;
-			dev_info.cntl_sb_buf_size = adev->gfx.ngg.buf[NGG_CNTL].size;
-			dev_info.param_buf_gpu_addr = adev->gfx.ngg.buf[NGG_PARAM].gpu_addr;
-			dev_info.param_buf_size = adev->gfx.ngg.buf[NGG_PARAM].size;
+		dev_info->wave_front_size = adev->gfx.cu_info.wave_front_size;
+		dev_info->num_shader_visible_vgprs = adev->gfx.config.max_gprs;
+		dev_info->num_cu_per_sh = adev->gfx.config.max_cu_per_sh;
+		dev_info->num_tcc_blocks = adev->gfx.config.max_texture_channel_caches;
+		dev_info->gs_vgt_table_depth = adev->gfx.config.gs_vgt_table_depth;
+		dev_info->gs_prim_buffer_depth = adev->gfx.config.gs_prim_buffer_depth;
+		dev_info->max_gs_waves_per_vgt = adev->gfx.config.max_gs_threads;
+
+		if (adev->family >= AMDGPU_FAMILY_NV)
+			dev_info->pa_sc_tile_steering_override =
+				adev->gfx.config.pa_sc_tile_steering_override;
+
+		dev_info->tcc_disabled_mask = adev->gfx.config.tcc_disabled_mask;
+
+		/* Combine the chip gen mask with the platform (CPU/mobo) mask. */
+		pcie_gen_mask = adev->pm.pcie_gen_mask &
+			(adev->pm.pcie_gen_mask >> CAIL_PCIE_LINK_SPEED_SUPPORT_SHIFT);
+		pcie_width_mask = adev->pm.pcie_mlw_mask &
+			(adev->pm.pcie_mlw_mask >> CAIL_PCIE_LINK_WIDTH_SUPPORT_SHIFT);
+		dev_info->pcie_gen = fls(pcie_gen_mask);
+		dev_info->pcie_num_lanes =
+			pcie_width_mask & CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X32 ? 32 :
+			pcie_width_mask & CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X16 ? 16 :
+			pcie_width_mask & CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X12 ? 12 :
+			pcie_width_mask & CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X8 ? 8 :
+			pcie_width_mask & CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X4 ? 4 :
+			pcie_width_mask & CAIL_ASIC_PCIE_LINK_WIDTH_SUPPORT_X2 ? 2 : 1;
+
+		dev_info->tcp_cache_size = adev->gfx.config.gc_tcp_l1_size;
+		dev_info->num_sqc_per_wgp = adev->gfx.config.gc_num_sqc_per_wgp;
+		dev_info->sqc_data_cache_size = adev->gfx.config.gc_l1_data_cache_size_per_sqc;
+		dev_info->sqc_inst_cache_size = adev->gfx.config.gc_l1_instruction_cache_size_per_sqc;
+		dev_info->gl1c_cache_size = adev->gfx.config.gc_gl1c_size_per_instance *
+					    adev->gfx.config.gc_gl1c_per_sa;
+		dev_info->gl2c_cache_size = adev->gfx.config.gc_gl2c_per_gpu;
+		dev_info->mall_size = adev->gmc.mall_size;
+
+
+		if (adev->gfx.funcs->get_gfx_shadow_info) {
+			struct amdgpu_gfx_shadow_info shadow_info;
+
+			ret = amdgpu_gfx_get_gfx_shadow_info(adev, &shadow_info);
+			if (!ret) {
+				dev_info->shadow_size = shadow_info.shadow_size;
+				dev_info->shadow_alignment = shadow_info.shadow_alignment;
+				dev_info->csa_size = shadow_info.csa_size;
+				dev_info->csa_alignment = shadow_info.csa_alignment;
+			}
 		}
-		dev_info.wave_front_size = adev->gfx.cu_info.wave_front_size;
-		dev_info.num_shader_visible_vgprs = adev->gfx.config.max_gprs;
-		dev_info.num_cu_per_sh = adev->gfx.config.max_cu_per_sh;
-		dev_info.num_tcc_blocks = adev->gfx.config.max_texture_channel_caches;
-		dev_info.gs_vgt_table_depth = adev->gfx.config.gs_vgt_table_depth;
-		dev_info.gs_prim_buffer_depth = adev->gfx.config.gs_prim_buffer_depth;
-		dev_info.max_gs_waves_per_vgt = adev->gfx.config.max_gs_threads;
-
-		return copy_to_user(out, &dev_info,
-				    min((size_t)size, sizeof(dev_info))) ? -EFAULT : 0;
+
+		dev_info->userq_ip_mask = amdgpu_userq_get_supported_ip_mask(adev);
+
+		ret = copy_to_user(out, dev_info,
+				   min((size_t)size, sizeof(*dev_info))) ? -EFAULT : 0;
+		kfree(dev_info);
+		return ret;
 	}
 	case AMDGPU_INFO_VCE_CLOCK_TABLE: {
-		unsigned i;
+		unsigned int i;
 		struct drm_amdgpu_info_vce_clock_table vce_clk_table = {};
 		struct amd_vce_state *vce_state;
 
@@ -664,6 +1078,26 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 					    min((size_t)size, (size_t)(bios_size - bios_offset)))
 					? -EFAULT : 0;
 		}
+		case AMDGPU_INFO_VBIOS_INFO: {
+			struct drm_amdgpu_info_vbios vbios_info = {};
+			struct atom_context *atom_context;
+
+			atom_context = adev->mode_info.atom_context;
+			if (atom_context) {
+				memcpy(vbios_info.name, atom_context->name,
+				       sizeof(atom_context->name));
+				memcpy(vbios_info.vbios_pn, atom_context->vbios_pn,
+				       sizeof(atom_context->vbios_pn));
+				vbios_info.version = atom_context->version;
+				memcpy(vbios_info.vbios_ver_str, atom_context->vbios_ver_str,
+				       sizeof(atom_context->vbios_ver_str));
+				memcpy(vbios_info.date, atom_context->date,
+				       sizeof(atom_context->date));
+			}
+
+			return copy_to_user(out, &vbios_info,
+						min((size_t)size, sizeof(vbios_info))) ? -EFAULT : 0;
+		}
 		default:
 			DRM_DEBUG_KMS("Invalid request %d\n",
 					info->vbios_info.type);
@@ -692,10 +1126,7 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 		}
 	}
 	case AMDGPU_INFO_SENSOR: {
-		struct pp_gpu_power query = {0};
-		int query_size = sizeof(query);
-
-		if (amdgpu_dpm == 0)
+		if (!adev->pm.dpm_enabled)
 			return -ENOENT;
 
 		switch (info->sensor_info.type) {
@@ -736,11 +1167,25 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 		case AMDGPU_INFO_SENSOR_GPU_AVG_POWER:
 			/* get average GPU power */
 			if (amdgpu_dpm_read_sensor(adev,
-						   AMDGPU_PP_SENSOR_GPU_POWER,
-						   (void *)&query, &query_size)) {
+						   AMDGPU_PP_SENSOR_GPU_AVG_POWER,
+						   (void *)&ui32, &ui32_size)) {
+				/* fall back to input power for backwards compat */
+				if (amdgpu_dpm_read_sensor(adev,
+							   AMDGPU_PP_SENSOR_GPU_INPUT_POWER,
+							   (void *)&ui32, &ui32_size)) {
+					return -EINVAL;
+				}
+			}
+			ui32 >>= 8;
+			break;
+		case AMDGPU_INFO_SENSOR_GPU_INPUT_POWER:
+			/* get input GPU power */
+			if (amdgpu_dpm_read_sensor(adev,
+						   AMDGPU_PP_SENSOR_GPU_INPUT_POWER,
+						   (void *)&ui32, &ui32_size)) {
 				return -EINVAL;
 			}
-			ui32 = query.average_gpu_power >> 8;
+			ui32 >>= 8;
 			break;
 		case AMDGPU_INFO_SENSOR_VDDNB:
 			/* get VDDNB in millivolts */
@@ -758,6 +1203,42 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 				return -EINVAL;
 			}
 			break;
+		case AMDGPU_INFO_SENSOR_STABLE_PSTATE_GFX_SCLK:
+			/* get stable pstate sclk in Mhz */
+			if (amdgpu_dpm_read_sensor(adev,
+						   AMDGPU_PP_SENSOR_STABLE_PSTATE_SCLK,
+						   (void *)&ui32, &ui32_size)) {
+				return -EINVAL;
+			}
+			ui32 /= 100;
+			break;
+		case AMDGPU_INFO_SENSOR_STABLE_PSTATE_GFX_MCLK:
+			/* get stable pstate mclk in Mhz */
+			if (amdgpu_dpm_read_sensor(adev,
+						   AMDGPU_PP_SENSOR_STABLE_PSTATE_MCLK,
+						   (void *)&ui32, &ui32_size)) {
+				return -EINVAL;
+			}
+			ui32 /= 100;
+			break;
+		case AMDGPU_INFO_SENSOR_PEAK_PSTATE_GFX_SCLK:
+			/* get peak pstate sclk in Mhz */
+			if (amdgpu_dpm_read_sensor(adev,
+						   AMDGPU_PP_SENSOR_PEAK_PSTATE_SCLK,
+						   (void *)&ui32, &ui32_size)) {
+				return -EINVAL;
+			}
+			ui32 /= 100;
+			break;
+		case AMDGPU_INFO_SENSOR_PEAK_PSTATE_GFX_MCLK:
+			/* get peak pstate mclk in Mhz */
+			if (amdgpu_dpm_read_sensor(adev,
+						   AMDGPU_PP_SENSOR_PEAK_PSTATE_MCLK,
+						   (void *)&ui32, &ui32_size)) {
+				return -EINVAL;
+			}
+			ui32 /= 100;
+			break;
 		default:
 			DRM_DEBUG_KMS("Invalid request %d\n",
 				      info->sensor_info.type);
@@ -765,36 +1246,131 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 		}
 		return copy_to_user(out, &ui32, min(size, 4u)) ? -EFAULT : 0;
 	}
-	default:
-		DRM_DEBUG_KMS("Invalid request %d\n", info->query);
-		return -EINVAL;
+	case AMDGPU_INFO_VRAM_LOST_COUNTER:
+		ui32 = atomic_read(&adev->vram_lost_counter);
+		return copy_to_user(out, &ui32, min(size, 4u)) ? -EFAULT : 0;
+	case AMDGPU_INFO_RAS_ENABLED_FEATURES: {
+		struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+		uint64_t ras_mask;
+
+		if (!ras)
+			return -EINVAL;
+		ras_mask = (uint64_t)adev->ras_enabled << 32 | ras->features;
+
+		return copy_to_user(out, &ras_mask,
+				min_t(u64, size, sizeof(ras_mask))) ?
+			-EFAULT : 0;
 	}
-	return 0;
-}
+	case AMDGPU_INFO_VIDEO_CAPS: {
+		const struct amdgpu_video_codecs *codecs;
+		struct drm_amdgpu_info_video_caps *caps;
+		int r;
 
+		if (!adev->asic_funcs->query_video_codecs)
+			return -EINVAL;
 
-/*
- * Outdated mess for old drm with Xorg being in charge (void function now).
- */
-/**
- * amdgpu_driver_lastclose_kms - drm callback for last close
- *
- * @dev: drm dev pointer
- *
- * Switch vga_switcheroo state after last close (all asics).
- */
-void amdgpu_driver_lastclose_kms(struct drm_device *dev)
-{
-	struct amdgpu_device *adev = dev->dev_private;
+		switch (info->video_cap.type) {
+		case AMDGPU_INFO_VIDEO_CAPS_DECODE:
+			r = amdgpu_asic_query_video_codecs(adev, false, &codecs);
+			if (r)
+				return -EINVAL;
+			break;
+		case AMDGPU_INFO_VIDEO_CAPS_ENCODE:
+			r = amdgpu_asic_query_video_codecs(adev, true, &codecs);
+			if (r)
+				return -EINVAL;
+			break;
+		default:
+			DRM_DEBUG_KMS("Invalid request %d\n",
+				      info->video_cap.type);
+			return -EINVAL;
+		}
 
-	amdgpu_fbdev_restore_mode(adev);
-	vga_switcheroo_process_delayed_switch();
-}
+		caps = kzalloc(sizeof(*caps), GFP_KERNEL);
+		if (!caps)
+			return -ENOMEM;
 
-bool amdgpu_kms_vram_lost(struct amdgpu_device *adev,
-			  struct amdgpu_fpriv *fpriv)
-{
-	return fpriv->vram_lost_counter != atomic_read(&adev->vram_lost_counter);
+		for (i = 0; i < codecs->codec_count; i++) {
+			int idx = codecs->codec_array[i].codec_type;
+
+			switch (idx) {
+			case AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2:
+			case AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4:
+			case AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1:
+			case AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC:
+			case AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC:
+			case AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG:
+			case AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9:
+			case AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1:
+				caps->codec_info[idx].valid = 1;
+				caps->codec_info[idx].max_width =
+					codecs->codec_array[i].max_width;
+				caps->codec_info[idx].max_height =
+					codecs->codec_array[i].max_height;
+				caps->codec_info[idx].max_pixels_per_frame =
+					codecs->codec_array[i].max_pixels_per_frame;
+				caps->codec_info[idx].max_level =
+					codecs->codec_array[i].max_level;
+				break;
+			default:
+				break;
+			}
+		}
+		r = copy_to_user(out, caps,
+				 min((size_t)size, sizeof(*caps))) ? -EFAULT : 0;
+		kfree(caps);
+		return r;
+	}
+	case AMDGPU_INFO_MAX_IBS: {
+		uint32_t max_ibs[AMDGPU_HW_IP_NUM];
+
+		for (i = 0; i < AMDGPU_HW_IP_NUM; ++i)
+			max_ibs[i] = amdgpu_ring_max_ibs(i);
+
+		return copy_to_user(out, max_ibs,
+				    min((size_t)size, sizeof(max_ibs))) ? -EFAULT : 0;
+	}
+	case AMDGPU_INFO_GPUVM_FAULT: {
+		struct amdgpu_fpriv *fpriv = filp->driver_priv;
+		struct amdgpu_vm *vm = &fpriv->vm;
+		struct drm_amdgpu_info_gpuvm_fault gpuvm_fault;
+		unsigned long flags;
+
+		if (!vm)
+			return -EINVAL;
+
+		memset(&gpuvm_fault, 0, sizeof(gpuvm_fault));
+
+		xa_lock_irqsave(&adev->vm_manager.pasids, flags);
+		gpuvm_fault.addr = vm->fault_info.addr;
+		gpuvm_fault.status = vm->fault_info.status;
+		gpuvm_fault.vmhub = vm->fault_info.vmhub;
+		xa_unlock_irqrestore(&adev->vm_manager.pasids, flags);
+
+		return copy_to_user(out, &gpuvm_fault,
+				    min((size_t)size, sizeof(gpuvm_fault))) ? -EFAULT : 0;
+	}
+	case AMDGPU_INFO_UQ_FW_AREAS: {
+		struct drm_amdgpu_info_uq_metadata meta_info = {};
+
+		switch (info->query_hw_ip.type) {
+		case AMDGPU_HW_IP_GFX:
+			ret = amdgpu_userq_metadata_info_gfx(adev, info, &meta_info.gfx);
+			if (ret)
+				return ret;
+
+			ret = copy_to_user(out, &meta_info,
+						min((size_t)size, sizeof(meta_info))) ? -EFAULT : 0;
+			return 0;
+		default:
+			return -EINVAL;
+		}
+	}
+	default:
+		DRM_DEBUG_KMS("Invalid request %d\n", info->query);
+		return -EINVAL;
+	}
+	return 0;
 }
 
 /**
@@ -808,15 +1384,24 @@ bool amdgpu_kms_vram_lost(struct amdgpu_device *adev,
  */
 int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv)
 {
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_fpriv *fpriv;
-	int r;
+	int r, pasid;
+
+	/* Ensure IB tests are run on ring */
+	flush_delayed_work(&adev->delayed_init_work);
+
+
+	if (amdgpu_ras_intr_triggered()) {
+		DRM_ERROR("RAS Intr triggered, device disabled!!");
+		return -EHWPOISON;
+	}
 
 	file_priv->driver_priv = NULL;
 
 	r = pm_runtime_get_sync(dev->dev);
 	if (r < 0)
-		return r;
+		goto pm_put;
 
 	fpriv = kzalloc(sizeof(*fpriv), GFP_KERNEL);
 	if (unlikely(!fpriv)) {
@@ -824,37 +1409,69 @@ int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv)
 		goto out_suspend;
 	}
 
-	r = amdgpu_vm_init(adev, &fpriv->vm,
-			   AMDGPU_VM_CONTEXT_GFX);
-	if (r) {
-		kfree(fpriv);
-		goto out_suspend;
+	pasid = amdgpu_pasid_alloc(16);
+	if (pasid < 0) {
+		dev_warn(adev->dev, "No more PASIDs available!");
+		pasid = 0;
 	}
 
+	r = amdgpu_xcp_open_device(adev, fpriv, file_priv);
+	if (r)
+		goto error_pasid;
+
+	amdgpu_debugfs_vm_init(file_priv);
+
+	r = amdgpu_vm_init(adev, &fpriv->vm, fpriv->xcp_id, pasid);
+	if (r)
+		goto error_pasid;
+
 	fpriv->prt_va = amdgpu_vm_bo_add(adev, &fpriv->vm, NULL);
 	if (!fpriv->prt_va) {
 		r = -ENOMEM;
-		amdgpu_vm_fini(adev, &fpriv->vm);
-		kfree(fpriv);
-		goto out_suspend;
+		goto error_vm;
 	}
 
-	if (amdgpu_sriov_vf(adev)) {
-		r = amdgpu_map_static_csa(adev, &fpriv->vm, &fpriv->csa_va);
+	if (adev->gfx.mcbp) {
+		uint64_t csa_addr = amdgpu_csa_vaddr(adev) & AMDGPU_GMC_HOLE_MASK;
+
+		r = amdgpu_map_static_csa(adev, &fpriv->vm, adev->virt.csa_obj,
+						&fpriv->csa_va, csa_addr, AMDGPU_CSA_SIZE);
 		if (r)
-			goto out_suspend;
+			goto error_vm;
 	}
 
+	r = amdgpu_seq64_map(adev, &fpriv->vm, &fpriv->seq64_va);
+	if (r)
+		goto error_vm;
+
 	mutex_init(&fpriv->bo_list_lock);
-	idr_init(&fpriv->bo_list_handles);
+	idr_init_base(&fpriv->bo_list_handles, 1);
+
+	r = amdgpu_userq_mgr_init(&fpriv->userq_mgr, file_priv, adev);
+	if (r)
+		DRM_WARN("Can't setup usermode queues, use legacy workload submission only\n");
+
+	r = amdgpu_eviction_fence_init(&fpriv->evf_mgr);
+	if (r)
+		goto error_vm;
 
-	amdgpu_ctx_mgr_init(&fpriv->ctx_mgr);
+	amdgpu_ctx_mgr_init(&fpriv->ctx_mgr, adev);
 
-	fpriv->vram_lost_counter = atomic_read(&adev->vram_lost_counter);
 	file_priv->driver_priv = fpriv;
+	goto out_suspend;
+
+error_vm:
+	amdgpu_vm_fini(adev, &fpriv->vm);
+
+error_pasid:
+	if (pasid)
+		amdgpu_pasid_free(pasid);
+
+	kfree(fpriv);
 
 out_suspend:
 	pm_runtime_mark_last_busy(dev->dev);
+pm_put:
 	pm_runtime_put_autosuspend(dev->dev);
 
 	return r;
@@ -871,9 +1488,11 @@ out_suspend:
 void amdgpu_driver_postclose_kms(struct drm_device *dev,
 				 struct drm_file *file_priv)
 {
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_fpriv *fpriv = file_priv->driver_priv;
 	struct amdgpu_bo_list *list;
+	struct amdgpu_bo *pd;
+	u32 pasid;
 	int handle;
 
 	if (!fpriv)
@@ -881,27 +1500,37 @@ void amdgpu_driver_postclose_kms(struct drm_device *dev,
 
 	pm_runtime_get_sync(dev->dev);
 
-	amdgpu_ctx_mgr_fini(&fpriv->ctx_mgr);
-
-	if (adev->asic_type != CHIP_RAVEN) {
+	if (amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_UVD) != NULL)
 		amdgpu_uvd_free_handles(adev, file_priv);
+	if (amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_VCE) != NULL)
 		amdgpu_vce_free_handles(adev, file_priv);
-	}
 
-	amdgpu_vm_bo_rmv(adev, fpriv->prt_va);
+	if (fpriv->csa_va) {
+		uint64_t csa_addr = amdgpu_csa_vaddr(adev) & AMDGPU_GMC_HOLE_MASK;
 
-	if (amdgpu_sriov_vf(adev)) {
-		/* TODO: how to handle reserve failure */
-		BUG_ON(amdgpu_bo_reserve(adev->virt.csa_obj, true));
-		amdgpu_vm_bo_rmv(adev, fpriv->csa_va);
+		WARN_ON(amdgpu_unmap_static_csa(adev, &fpriv->vm, adev->virt.csa_obj,
+						fpriv->csa_va, csa_addr));
 		fpriv->csa_va = NULL;
-		amdgpu_bo_unreserve(adev->virt.csa_obj);
 	}
 
+	amdgpu_seq64_unmap(adev, fpriv);
+
+	pasid = fpriv->vm.pasid;
+	pd = amdgpu_bo_ref(fpriv->vm.root.bo);
+	if (!WARN_ON(amdgpu_bo_reserve(pd, true))) {
+		amdgpu_vm_bo_del(adev, fpriv->prt_va);
+		amdgpu_bo_unreserve(pd);
+	}
+
+	amdgpu_ctx_mgr_fini(&fpriv->ctx_mgr);
 	amdgpu_vm_fini(adev, &fpriv->vm);
 
+	if (pasid)
+		amdgpu_pasid_free_delayed(pd->tbo.base.resv, pasid);
+	amdgpu_bo_unref(&pd);
+
 	idr_for_each_entry(&fpriv->bo_list_handles, list, handle)
-		amdgpu_bo_list_free(list);
+		amdgpu_bo_list_put(list);
 
 	idr_destroy(&fpriv->bo_list_handles);
 	mutex_destroy(&fpriv->bo_list_lock);
@@ -913,21 +1542,31 @@ void amdgpu_driver_postclose_kms(struct drm_device *dev,
 	pm_runtime_put_autosuspend(dev->dev);
 }
 
+
+void amdgpu_driver_release_kms(struct drm_device *dev)
+{
+	struct amdgpu_device *adev = drm_to_adev(dev);
+
+	amdgpu_device_fini_sw(adev);
+	pci_set_drvdata(adev->pdev, NULL);
+}
+
 /*
  * VBlank related functions.
  */
 /**
  * amdgpu_get_vblank_counter_kms - get frame count
  *
- * @dev: drm dev pointer
- * @pipe: crtc to get the frame count from
+ * @crtc: crtc to get the frame count from
  *
  * Gets the frame count on the requested crtc (all asics).
  * Returns frame count on success, -EINVAL on failure.
  */
-u32 amdgpu_get_vblank_counter_kms(struct drm_device *dev, unsigned int pipe)
+u32 amdgpu_get_vblank_counter_kms(struct drm_crtc *crtc)
 {
-	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_device *dev = crtc->dev;
+	unsigned int pipe = crtc->index;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	int vpos, hpos, stat;
 	u32 count;
 
@@ -950,11 +1589,11 @@ u32 amdgpu_get_vblank_counter_kms(struct drm_device *dev, unsigned int pipe)
 		 */
 		do {
 			count = amdgpu_display_vblank_get_counter(adev, pipe);
-			/* Ask amdgpu_get_crtc_scanoutpos to return vpos as
-			 * distance to start of vblank, instead of regular
-			 * vertical scanout pos.
+			/* Ask amdgpu_display_get_crtc_scanoutpos to return
+			 * vpos as distance to start of vblank, instead of
+			 * regular vertical scanout pos.
 			 */
-			stat = amdgpu_get_crtc_scanoutpos(
+			stat = amdgpu_display_get_crtc_scanoutpos(
 				dev, pipe, GET_DISTANCE_TO_VBLANKSTART,
 				&vpos, &hpos, NULL, NULL,
 				&adev->mode_info.crtcs[pipe]->base.hwmode);
@@ -986,16 +1625,17 @@ u32 amdgpu_get_vblank_counter_kms(struct drm_device *dev, unsigned int pipe)
 /**
  * amdgpu_enable_vblank_kms - enable vblank interrupt
  *
- * @dev: drm dev pointer
- * @pipe: crtc to enable vblank interrupt for
+ * @crtc: crtc to enable vblank interrupt for
  *
  * Enable the interrupt on the requested crtc (all asics).
  * Returns 0 on success, -EINVAL on failure.
  */
-int amdgpu_enable_vblank_kms(struct drm_device *dev, unsigned int pipe)
+int amdgpu_enable_vblank_kms(struct drm_crtc *crtc)
 {
-	struct amdgpu_device *adev = dev->dev_private;
-	int idx = amdgpu_crtc_idx_to_irq_type(adev, pipe);
+	struct drm_device *dev = crtc->dev;
+	unsigned int pipe = crtc->index;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	int idx = amdgpu_display_crtc_idx_to_irq_type(adev, pipe);
 
 	return amdgpu_irq_get(adev, &adev->crtc_irq, idx);
 }
@@ -1003,52 +1643,45 @@ int amdgpu_enable_vblank_kms(struct drm_device *dev, unsigned int pipe)
 /**
  * amdgpu_disable_vblank_kms - disable vblank interrupt
  *
- * @dev: drm dev pointer
- * @pipe: crtc to disable vblank interrupt for
+ * @crtc: crtc to disable vblank interrupt for
  *
  * Disable the interrupt on the requested crtc (all asics).
  */
-void amdgpu_disable_vblank_kms(struct drm_device *dev, unsigned int pipe)
+void amdgpu_disable_vblank_kms(struct drm_crtc *crtc)
 {
-	struct amdgpu_device *adev = dev->dev_private;
-	int idx = amdgpu_crtc_idx_to_irq_type(adev, pipe);
+	struct drm_device *dev = crtc->dev;
+	unsigned int pipe = crtc->index;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	int idx = amdgpu_display_crtc_idx_to_irq_type(adev, pipe);
 
 	amdgpu_irq_put(adev, &adev->crtc_irq, idx);
 }
 
-const struct drm_ioctl_desc amdgpu_ioctls_kms[] = {
-	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_CREATE, amdgpu_gem_create_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
-	DRM_IOCTL_DEF_DRV(AMDGPU_CTX, amdgpu_ctx_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
-	DRM_IOCTL_DEF_DRV(AMDGPU_VM, amdgpu_vm_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
-	DRM_IOCTL_DEF_DRV(AMDGPU_BO_LIST, amdgpu_bo_list_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
-	/* KMS */
-	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_MMAP, amdgpu_gem_mmap_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
-	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_WAIT_IDLE, amdgpu_gem_wait_idle_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
-	DRM_IOCTL_DEF_DRV(AMDGPU_CS, amdgpu_cs_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
-	DRM_IOCTL_DEF_DRV(AMDGPU_INFO, amdgpu_info_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
-	DRM_IOCTL_DEF_DRV(AMDGPU_WAIT_CS, amdgpu_cs_wait_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
-	DRM_IOCTL_DEF_DRV(AMDGPU_WAIT_FENCES, amdgpu_cs_wait_fences_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
-	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_METADATA, amdgpu_gem_metadata_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
-	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_VA, amdgpu_gem_va_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
-	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_OP, amdgpu_gem_op_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
-	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_USERPTR, amdgpu_gem_userptr_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
-};
-const int amdgpu_max_kms_ioctl = ARRAY_SIZE(amdgpu_ioctls_kms);
-
 /*
  * Debugfs info
  */
 #if defined(CONFIG_DEBUG_FS)
 
-static int amdgpu_debugfs_firmware_info(struct seq_file *m, void *data)
+static int amdgpu_debugfs_firmware_info_show(struct seq_file *m, void *unused)
 {
-	struct drm_info_node *node = (struct drm_info_node *) m->private;
-	struct drm_device *dev = node->minor->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = m->private;
 	struct drm_amdgpu_info_firmware fw_info;
 	struct drm_amdgpu_query_fw query_fw;
+	struct atom_context *ctx = adev->mode_info.atom_context;
+	uint8_t smu_program, smu_major, smu_minor, smu_debug;
 	int ret, i;
 
+	static const char *ta_fw_name[TA_FW_TYPE_MAX_INDEX] = {
+#define TA_FW_NAME(type)[TA_FW_TYPE_PSP_##type] = #type
+		TA_FW_NAME(XGMI),
+		TA_FW_NAME(RAS),
+		TA_FW_NAME(HDCP),
+		TA_FW_NAME(DTM),
+		TA_FW_NAME(RAP),
+		TA_FW_NAME(SECUREDISPLAY),
+#undef TA_FW_NAME
+	};
+
 	/* VCE */
 	query_fw.fw_type = AMDGPU_INFO_FW_VCE;
 	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
@@ -1105,6 +1738,46 @@ static int amdgpu_debugfs_firmware_info(struct seq_file *m, void *data)
 	seq_printf(m, "RLC feature version: %u, firmware version: 0x%08x\n",
 		   fw_info.feature, fw_info.ver);
 
+	/* RLC SAVE RESTORE LIST CNTL */
+	query_fw.fw_type = AMDGPU_INFO_FW_GFX_RLC_RESTORE_LIST_CNTL;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "RLC SRLC feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* RLC SAVE RESTORE LIST GPM MEM */
+	query_fw.fw_type = AMDGPU_INFO_FW_GFX_RLC_RESTORE_LIST_GPM_MEM;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "RLC SRLG feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* RLC SAVE RESTORE LIST SRM MEM */
+	query_fw.fw_type = AMDGPU_INFO_FW_GFX_RLC_RESTORE_LIST_SRM_MEM;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "RLC SRLS feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* RLCP */
+	query_fw.fw_type = AMDGPU_INFO_FW_GFX_RLCP;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "RLCP feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* RLCV */
+	query_fw.fw_type = AMDGPU_INFO_FW_GFX_RLCV;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "RLCV feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
 	/* MEC */
 	query_fw.fw_type = AMDGPU_INFO_FW_GFX_MEC;
 	query_fw.index = 0;
@@ -1115,8 +1788,7 @@ static int amdgpu_debugfs_firmware_info(struct seq_file *m, void *data)
 		   fw_info.feature, fw_info.ver);
 
 	/* MEC2 */
-	if (adev->asic_type == CHIP_KAVERI ||
-	    (adev->asic_type > CHIP_TOPAZ && adev->asic_type != CHIP_STONEY)) {
+	if (adev->gfx.mec2_fw) {
 		query_fw.index = 1;
 		ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
 		if (ret)
@@ -1125,6 +1797,15 @@ static int amdgpu_debugfs_firmware_info(struct seq_file *m, void *data)
 			   fw_info.feature, fw_info.ver);
 	}
 
+	/* IMU */
+	query_fw.fw_type = AMDGPU_INFO_FW_IMU;
+	query_fw.index = 0;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "IMU feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
 	/* PSP SOS */
 	query_fw.fw_type = AMDGPU_INFO_FW_SOS;
 	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
@@ -1142,13 +1823,28 @@ static int amdgpu_debugfs_firmware_info(struct seq_file *m, void *data)
 	seq_printf(m, "ASD feature version: %u, firmware version: 0x%08x\n",
 		   fw_info.feature, fw_info.ver);
 
+	query_fw.fw_type = AMDGPU_INFO_FW_TA;
+	for (i = TA_FW_TYPE_PSP_XGMI; i < TA_FW_TYPE_MAX_INDEX; i++) {
+		query_fw.index = i;
+		ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+		if (ret)
+			continue;
+
+		seq_printf(m, "TA %s feature version: 0x%08x, firmware version: 0x%08x\n",
+			   ta_fw_name[i], fw_info.feature, fw_info.ver);
+	}
+
 	/* SMC */
 	query_fw.fw_type = AMDGPU_INFO_FW_SMC;
 	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
 	if (ret)
 		return ret;
-	seq_printf(m, "SMC feature version: %u, firmware version: 0x%08x\n",
-		   fw_info.feature, fw_info.ver);
+	smu_program = (fw_info.ver >> 24) & 0xff;
+	smu_major = (fw_info.ver >> 16) & 0xff;
+	smu_minor = (fw_info.ver >> 8) & 0xff;
+	smu_debug = (fw_info.ver >> 0) & 0xff;
+	seq_printf(m, "SMC feature version: %u, program: %d, firmware version: 0x%08x (%d.%d.%d)\n",
+		   fw_info.feature, smu_program, fw_info.ver, smu_major, smu_minor, smu_debug);
 
 	/* SDMA */
 	query_fw.fw_type = AMDGPU_INFO_FW_SDMA;
@@ -1161,20 +1857,89 @@ static int amdgpu_debugfs_firmware_info(struct seq_file *m, void *data)
 			   i, fw_info.feature, fw_info.ver);
 	}
 
+	/* VCN */
+	query_fw.fw_type = AMDGPU_INFO_FW_VCN;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "VCN feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* DMCU */
+	query_fw.fw_type = AMDGPU_INFO_FW_DMCU;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "DMCU feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* DMCUB */
+	query_fw.fw_type = AMDGPU_INFO_FW_DMCUB;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "DMCUB feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* TOC */
+	query_fw.fw_type = AMDGPU_INFO_FW_TOC;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "TOC feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* CAP */
+	if (adev->psp.cap_fw) {
+		query_fw.fw_type = AMDGPU_INFO_FW_CAP;
+		ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+		if (ret)
+			return ret;
+		seq_printf(m, "CAP feature version: %u, firmware version: 0x%08x\n",
+				fw_info.feature, fw_info.ver);
+	}
+
+	/* MES_KIQ */
+	query_fw.fw_type = AMDGPU_INFO_FW_MES_KIQ;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "MES_KIQ feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* MES */
+	query_fw.fw_type = AMDGPU_INFO_FW_MES;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "MES feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* VPE */
+	query_fw.fw_type = AMDGPU_INFO_FW_VPE;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "VPE feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	seq_printf(m, "VBIOS version: %s\n", ctx->vbios_pn);
+
 	return 0;
 }
 
-static const struct drm_info_list amdgpu_firmware_info_list[] = {
-	{"amdgpu_firmware_info", amdgpu_debugfs_firmware_info, 0, NULL},
-};
+DEFINE_SHOW_ATTRIBUTE(amdgpu_debugfs_firmware_info);
+
 #endif
 
-int amdgpu_debugfs_firmware_init(struct amdgpu_device *adev)
+void amdgpu_debugfs_firmware_init(struct amdgpu_device *adev)
 {
 #if defined(CONFIG_DEBUG_FS)
-	return amdgpu_debugfs_add_files(adev, amdgpu_firmware_info_list,
-					ARRAY_SIZE(amdgpu_firmware_info_list));
-#else
-	return 0;
+	struct drm_minor *minor = adev_to_drm(adev)->primary;
+	struct dentry *root = minor->debugfs_root;
+
+	debugfs_create_file("amdgpu_firmware_info", 0444, root,
+			    adev, &amdgpu_debugfs_firmware_info_fops);
+
 #endif
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_lsdma.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_lsdma.c
new file mode 100644
index 000000000000..4d1d4994ea3f
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_lsdma.c
@@ -0,0 +1,91 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "amdgpu_lsdma.h"
+
+#define AMDGPU_LSDMA_MAX_SIZE	0x2000000ULL
+
+int amdgpu_lsdma_wait_for(struct amdgpu_device *adev,
+			  uint32_t reg_index, uint32_t reg_val,
+			  uint32_t mask)
+{
+	uint32_t val;
+	int i;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		val = RREG32(reg_index);
+		if ((val & mask) == reg_val)
+			return 0;
+		udelay(1);
+	}
+
+	return -ETIME;
+}
+
+int amdgpu_lsdma_copy_mem(struct amdgpu_device *adev,
+			  uint64_t src_addr,
+			  uint64_t dst_addr,
+			  uint64_t mem_size)
+{
+	int ret;
+
+	if (mem_size == 0)
+		return -EINVAL;
+
+	while (mem_size > 0) {
+		uint64_t current_copy_size = min(mem_size, AMDGPU_LSDMA_MAX_SIZE);
+
+		ret = adev->lsdma.funcs->copy_mem(adev, src_addr, dst_addr, current_copy_size);
+		if (ret)
+			return ret;
+		src_addr += current_copy_size;
+		dst_addr += current_copy_size;
+		mem_size -= current_copy_size;
+	}
+
+	return 0;
+}
+
+int amdgpu_lsdma_fill_mem(struct amdgpu_device *adev,
+			  uint64_t dst_addr,
+			  uint32_t data,
+			  uint64_t mem_size)
+{
+	int ret;
+
+	if (mem_size == 0)
+		return -EINVAL;
+
+	while (mem_size > 0) {
+		uint64_t current_fill_size = min(mem_size, AMDGPU_LSDMA_MAX_SIZE);
+
+		ret = adev->lsdma.funcs->fill_mem(adev, dst_addr, data, current_fill_size);
+		if (ret)
+			return ret;
+		dst_addr += current_fill_size;
+		mem_size -= current_fill_size;
+	}
+
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_lsdma.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_lsdma.h
new file mode 100644
index 000000000000..c61ba58c5ee0
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_lsdma.h
@@ -0,0 +1,46 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_LSDMA_H__
+#define __AMDGPU_LSDMA_H__
+
+struct amdgpu_lsdma {
+	const struct amdgpu_lsdma_funcs      *funcs;
+};
+
+struct amdgpu_lsdma_funcs {
+	int (*copy_mem)(struct amdgpu_device *adev, uint64_t src_addr,
+			uint64_t dst_addr, uint64_t size);
+	int (*fill_mem)(struct amdgpu_device *adev, uint64_t dst_addr,
+			uint32_t data, uint64_t size);
+	void (*update_memory_power_gating)(struct amdgpu_device *adev, bool enable);
+};
+
+int amdgpu_lsdma_copy_mem(struct amdgpu_device *adev, uint64_t src_addr,
+			  uint64_t dst_addr, uint64_t mem_size);
+int amdgpu_lsdma_fill_mem(struct amdgpu_device *adev, uint64_t dst_addr,
+			  uint32_t data, uint64_t mem_size);
+int amdgpu_lsdma_wait_for(struct amdgpu_device *adev, uint32_t reg_index,
+			  uint32_t reg_val, uint32_t mask);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c
new file mode 100644
index 000000000000..3ca03b5e0f91
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c
@@ -0,0 +1,630 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu_ras.h"
+#include "amdgpu.h"
+#include "amdgpu_mca.h"
+
+#include "umc/umc_6_7_0_offset.h"
+#include "umc/umc_6_7_0_sh_mask.h"
+
+static bool amdgpu_mca_is_deferred_error(struct amdgpu_device *adev,
+					uint64_t mc_status)
+{
+	if (adev->umc.ras->check_ecc_err_status)
+		return adev->umc.ras->check_ecc_err_status(adev,
+				AMDGPU_MCA_ERROR_TYPE_DE, &mc_status);
+
+	return false;
+}
+
+void amdgpu_mca_query_correctable_error_count(struct amdgpu_device *adev,
+					      uint64_t mc_status_addr,
+					      unsigned long *error_count)
+{
+	uint64_t mc_status = RREG64_PCIE(mc_status_addr);
+
+	if (REG_GET_FIELD(mc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
+	    REG_GET_FIELD(mc_status, MCA_UMC_UMC0_MCUMC_STATUST0, CECC) == 1)
+		*error_count += 1;
+}
+
+void amdgpu_mca_query_uncorrectable_error_count(struct amdgpu_device *adev,
+						uint64_t mc_status_addr,
+						unsigned long *error_count)
+{
+	uint64_t mc_status = RREG64_PCIE(mc_status_addr);
+
+	if ((REG_GET_FIELD(mc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1) &&
+	    (REG_GET_FIELD(mc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1 ||
+	    REG_GET_FIELD(mc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1 ||
+	    REG_GET_FIELD(mc_status, MCA_UMC_UMC0_MCUMC_STATUST0, PCC) == 1 ||
+	    REG_GET_FIELD(mc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UC) == 1 ||
+	    REG_GET_FIELD(mc_status, MCA_UMC_UMC0_MCUMC_STATUST0, TCC) == 1))
+		*error_count += 1;
+}
+
+void amdgpu_mca_reset_error_count(struct amdgpu_device *adev,
+				  uint64_t mc_status_addr)
+{
+	WREG64_PCIE(mc_status_addr, 0x0ULL);
+}
+
+void amdgpu_mca_query_ras_error_count(struct amdgpu_device *adev,
+				      uint64_t mc_status_addr,
+				      void *ras_error_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+
+	amdgpu_mca_query_correctable_error_count(adev, mc_status_addr, &(err_data->ce_count));
+	amdgpu_mca_query_uncorrectable_error_count(adev, mc_status_addr, &(err_data->ue_count));
+
+	amdgpu_mca_reset_error_count(adev, mc_status_addr);
+}
+
+int amdgpu_mca_mp0_ras_sw_init(struct amdgpu_device *adev)
+{
+	int err;
+	struct amdgpu_mca_ras_block *ras;
+
+	if (!adev->mca.mp0.ras)
+		return 0;
+
+	ras = adev->mca.mp0.ras;
+
+	err = amdgpu_ras_register_ras_block(adev, &ras->ras_block);
+	if (err) {
+		dev_err(adev->dev, "Failed to register mca.mp0 ras block!\n");
+		return err;
+	}
+
+	strcpy(ras->ras_block.ras_comm.name, "mca.mp0");
+	ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__MCA;
+	ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
+	adev->mca.mp0.ras_if = &ras->ras_block.ras_comm;
+
+	return 0;
+}
+
+int amdgpu_mca_mp1_ras_sw_init(struct amdgpu_device *adev)
+{
+	int err;
+	struct amdgpu_mca_ras_block *ras;
+
+	if (!adev->mca.mp1.ras)
+		return 0;
+
+	ras = adev->mca.mp1.ras;
+
+	err = amdgpu_ras_register_ras_block(adev, &ras->ras_block);
+	if (err) {
+		dev_err(adev->dev, "Failed to register mca.mp1 ras block!\n");
+		return err;
+	}
+
+	strcpy(ras->ras_block.ras_comm.name, "mca.mp1");
+	ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__MCA;
+	ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
+	adev->mca.mp1.ras_if = &ras->ras_block.ras_comm;
+
+	return 0;
+}
+
+int amdgpu_mca_mpio_ras_sw_init(struct amdgpu_device *adev)
+{
+	int err;
+	struct amdgpu_mca_ras_block *ras;
+
+	if (!adev->mca.mpio.ras)
+		return 0;
+
+	ras = adev->mca.mpio.ras;
+
+	err = amdgpu_ras_register_ras_block(adev, &ras->ras_block);
+	if (err) {
+		dev_err(adev->dev, "Failed to register mca.mpio ras block!\n");
+		return err;
+	}
+
+	strcpy(ras->ras_block.ras_comm.name, "mca.mpio");
+	ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__MCA;
+	ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
+	adev->mca.mpio.ras_if = &ras->ras_block.ras_comm;
+
+	return 0;
+}
+
+static void amdgpu_mca_bank_set_init(struct mca_bank_set *mca_set)
+{
+	if (!mca_set)
+		return;
+
+	memset(mca_set, 0, sizeof(*mca_set));
+	INIT_LIST_HEAD(&mca_set->list);
+}
+
+static int amdgpu_mca_bank_set_add_entry(struct mca_bank_set *mca_set, struct mca_bank_entry *entry)
+{
+	struct mca_bank_node *node;
+
+	if (!entry)
+		return -EINVAL;
+
+	node = kvzalloc(sizeof(*node), GFP_KERNEL);
+	if (!node)
+		return -ENOMEM;
+
+	memcpy(&node->entry, entry, sizeof(*entry));
+
+	INIT_LIST_HEAD(&node->node);
+	list_add_tail(&node->node, &mca_set->list);
+
+	mca_set->nr_entries++;
+
+	return 0;
+}
+
+static int amdgpu_mca_bank_set_merge(struct mca_bank_set *mca_set, struct mca_bank_set *new)
+{
+	struct mca_bank_node *node;
+
+	list_for_each_entry(node, &new->list, node)
+		amdgpu_mca_bank_set_add_entry(mca_set, &node->entry);
+
+	return 0;
+}
+
+static void amdgpu_mca_bank_set_remove_node(struct mca_bank_set *mca_set, struct mca_bank_node *node)
+{
+	if (!node)
+		return;
+
+	list_del(&node->node);
+	kvfree(node);
+
+	mca_set->nr_entries--;
+}
+
+static void amdgpu_mca_bank_set_release(struct mca_bank_set *mca_set)
+{
+	struct mca_bank_node *node, *tmp;
+
+	if (list_empty(&mca_set->list))
+		return;
+
+	list_for_each_entry_safe(node, tmp, &mca_set->list, node)
+		amdgpu_mca_bank_set_remove_node(mca_set, node);
+}
+
+void amdgpu_mca_smu_init_funcs(struct amdgpu_device *adev, const struct amdgpu_mca_smu_funcs *mca_funcs)
+{
+	struct amdgpu_mca *mca = &adev->mca;
+
+	mca->mca_funcs = mca_funcs;
+}
+
+int amdgpu_mca_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_mca *mca = &adev->mca;
+	struct mca_bank_cache *mca_cache;
+	int i;
+
+	atomic_set(&mca->ue_update_flag, 0);
+
+	for (i = 0; i < ARRAY_SIZE(mca->mca_caches); i++) {
+		mca_cache = &mca->mca_caches[i];
+		mutex_init(&mca_cache->lock);
+		amdgpu_mca_bank_set_init(&mca_cache->mca_set);
+	}
+
+	return 0;
+}
+
+void amdgpu_mca_fini(struct amdgpu_device *adev)
+{
+	struct amdgpu_mca *mca = &adev->mca;
+	struct mca_bank_cache *mca_cache;
+	int i;
+
+	atomic_set(&mca->ue_update_flag, 0);
+
+	for (i = 0; i < ARRAY_SIZE(mca->mca_caches); i++) {
+		mca_cache = &mca->mca_caches[i];
+		amdgpu_mca_bank_set_release(&mca_cache->mca_set);
+		mutex_destroy(&mca_cache->lock);
+	}
+}
+
+int amdgpu_mca_reset(struct amdgpu_device *adev)
+{
+	amdgpu_mca_fini(adev);
+
+	return amdgpu_mca_init(adev);
+}
+
+int amdgpu_mca_smu_set_debug_mode(struct amdgpu_device *adev, bool enable)
+{
+	const struct amdgpu_mca_smu_funcs *mca_funcs = adev->mca.mca_funcs;
+
+	if (mca_funcs && mca_funcs->mca_set_debug_mode)
+		return mca_funcs->mca_set_debug_mode(adev, enable);
+
+	return -EOPNOTSUPP;
+}
+
+static void amdgpu_mca_smu_mca_bank_dump(struct amdgpu_device *adev, int idx, struct mca_bank_entry *entry,
+					 struct ras_query_context *qctx)
+{
+	u64 event_id = qctx ? qctx->evid.event_id : RAS_EVENT_INVALID_ID;
+
+	RAS_EVENT_LOG(adev, event_id, HW_ERR "Accelerator Check Architecture events logged\n");
+	RAS_EVENT_LOG(adev, event_id, HW_ERR "aca entry[%02d].STATUS=0x%016llx\n",
+		      idx, entry->regs[MCA_REG_IDX_STATUS]);
+	RAS_EVENT_LOG(adev, event_id, HW_ERR "aca entry[%02d].ADDR=0x%016llx\n",
+		      idx, entry->regs[MCA_REG_IDX_ADDR]);
+	RAS_EVENT_LOG(adev, event_id, HW_ERR "aca entry[%02d].MISC0=0x%016llx\n",
+		      idx, entry->regs[MCA_REG_IDX_MISC0]);
+	RAS_EVENT_LOG(adev, event_id, HW_ERR "aca entry[%02d].IPID=0x%016llx\n",
+		      idx, entry->regs[MCA_REG_IDX_IPID]);
+	RAS_EVENT_LOG(adev, event_id, HW_ERR "aca entry[%02d].SYND=0x%016llx\n",
+		      idx, entry->regs[MCA_REG_IDX_SYND]);
+}
+
+static int amdgpu_mca_smu_get_valid_mca_count(struct amdgpu_device *adev, enum amdgpu_mca_error_type type, uint32_t *count)
+{
+	const struct amdgpu_mca_smu_funcs *mca_funcs = adev->mca.mca_funcs;
+
+	if (!count)
+		return -EINVAL;
+
+	if (mca_funcs && mca_funcs->mca_get_valid_mca_count)
+		return mca_funcs->mca_get_valid_mca_count(adev, type, count);
+
+	return -EOPNOTSUPP;
+}
+
+static int amdgpu_mca_smu_get_mca_entry(struct amdgpu_device *adev, enum amdgpu_mca_error_type type,
+					int idx, struct mca_bank_entry *entry)
+{
+	const struct amdgpu_mca_smu_funcs *mca_funcs = adev->mca.mca_funcs;
+	int count;
+
+	if (!mca_funcs || !mca_funcs->mca_get_mca_entry)
+		return -EOPNOTSUPP;
+
+	switch (type) {
+	case AMDGPU_MCA_ERROR_TYPE_UE:
+		count = mca_funcs->max_ue_count;
+		break;
+	case AMDGPU_MCA_ERROR_TYPE_CE:
+		count = mca_funcs->max_ce_count;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	if (idx >= count)
+		return -EINVAL;
+
+	return mca_funcs->mca_get_mca_entry(adev, type, idx, entry);
+}
+
+static bool amdgpu_mca_bank_should_update(struct amdgpu_device *adev, enum amdgpu_mca_error_type type)
+{
+	struct amdgpu_mca *mca = &adev->mca;
+	bool ret = true;
+
+	/*
+	 * Because the UE Valid MCA count will only be cleared after reset,
+	 * in order to avoid repeated counting of the error count,
+	 * the aca bank is only updated once during the gpu recovery stage.
+	 */
+	if (type == AMDGPU_MCA_ERROR_TYPE_UE) {
+		if (amdgpu_ras_intr_triggered())
+			ret = atomic_cmpxchg(&mca->ue_update_flag, 0, 1) == 0;
+		else
+			atomic_set(&mca->ue_update_flag, 0);
+	}
+
+	return ret;
+}
+
+static bool amdgpu_mca_bank_should_dump(struct amdgpu_device *adev, enum amdgpu_mca_error_type type,
+					struct mca_bank_entry *entry)
+{
+	bool ret;
+
+	switch (type) {
+	case AMDGPU_MCA_ERROR_TYPE_CE:
+		ret = amdgpu_mca_is_deferred_error(adev, entry->regs[MCA_REG_IDX_STATUS]);
+		break;
+	case AMDGPU_MCA_ERROR_TYPE_UE:
+	default:
+		ret = true;
+		break;
+	}
+
+	return ret;
+}
+
+static int amdgpu_mca_smu_get_mca_set(struct amdgpu_device *adev, enum amdgpu_mca_error_type type, struct mca_bank_set *mca_set,
+				      struct ras_query_context *qctx)
+{
+	struct mca_bank_entry entry;
+	uint32_t count = 0, i;
+	int ret;
+
+	if (!mca_set)
+		return -EINVAL;
+
+	if (!amdgpu_mca_bank_should_update(adev, type))
+		return 0;
+
+	ret = amdgpu_mca_smu_get_valid_mca_count(adev, type, &count);
+	if (ret)
+		return ret;
+
+	for (i = 0; i < count; i++) {
+		memset(&entry, 0, sizeof(entry));
+		ret = amdgpu_mca_smu_get_mca_entry(adev, type, i, &entry);
+		if (ret)
+			return ret;
+
+		amdgpu_mca_bank_set_add_entry(mca_set, &entry);
+
+		if (amdgpu_mca_bank_should_dump(adev, type, &entry))
+			amdgpu_mca_smu_mca_bank_dump(adev, i, &entry, qctx);
+	}
+
+	return 0;
+}
+
+static int amdgpu_mca_smu_parse_mca_error_count(struct amdgpu_device *adev, enum amdgpu_ras_block blk,
+						enum amdgpu_mca_error_type type, struct mca_bank_entry *entry, uint32_t *count)
+{
+	const struct amdgpu_mca_smu_funcs *mca_funcs = adev->mca.mca_funcs;
+
+	if (!count || !entry)
+		return -EINVAL;
+
+	if (!mca_funcs || !mca_funcs->mca_parse_mca_error_count)
+		return -EOPNOTSUPP;
+
+	return mca_funcs->mca_parse_mca_error_count(adev, blk, type, entry, count);
+}
+
+static int amdgpu_mca_dispatch_mca_set(struct amdgpu_device *adev, enum amdgpu_ras_block blk, enum amdgpu_mca_error_type type,
+				       struct mca_bank_set *mca_set, struct ras_err_data *err_data)
+{
+	struct amdgpu_smuio_mcm_config_info mcm_info;
+	struct mca_bank_node *node, *tmp;
+	struct mca_bank_entry *entry;
+	uint32_t count;
+	int ret;
+
+	if (!mca_set)
+		return -EINVAL;
+
+	if (!mca_set->nr_entries)
+		return 0;
+
+	list_for_each_entry_safe(node, tmp, &mca_set->list, node) {
+		entry = &node->entry;
+
+		count = 0;
+		ret = amdgpu_mca_smu_parse_mca_error_count(adev, blk, type, entry, &count);
+		if (ret && ret != -EOPNOTSUPP)
+			return ret;
+
+		if (!count)
+			continue;
+
+		memset(&mcm_info, 0, sizeof(mcm_info));
+
+		mcm_info.socket_id = entry->info.socket_id;
+		mcm_info.die_id = entry->info.aid;
+
+		if (type == AMDGPU_MCA_ERROR_TYPE_UE) {
+			amdgpu_ras_error_statistic_ue_count(err_data,
+							    &mcm_info, (uint64_t)count);
+		} else {
+			if (amdgpu_mca_is_deferred_error(adev, entry->regs[MCA_REG_IDX_STATUS]))
+				amdgpu_ras_error_statistic_de_count(err_data,
+								    &mcm_info, (uint64_t)count);
+			else
+				amdgpu_ras_error_statistic_ce_count(err_data,
+								    &mcm_info, (uint64_t)count);
+		}
+
+		amdgpu_mca_bank_set_remove_node(mca_set, node);
+	}
+
+	return 0;
+}
+
+static int amdgpu_mca_add_mca_set_to_cache(struct amdgpu_device *adev, enum amdgpu_mca_error_type type, struct mca_bank_set *new)
+{
+	struct mca_bank_cache *mca_cache = &adev->mca.mca_caches[type];
+	int ret;
+
+	mutex_lock(&mca_cache->lock);
+	ret = amdgpu_mca_bank_set_merge(&mca_cache->mca_set, new);
+	mutex_unlock(&mca_cache->lock);
+
+	return ret;
+}
+
+int amdgpu_mca_smu_log_ras_error(struct amdgpu_device *adev, enum amdgpu_ras_block blk, enum amdgpu_mca_error_type type,
+				 struct ras_err_data *err_data, struct ras_query_context *qctx)
+{
+	struct mca_bank_set mca_set;
+	struct mca_bank_cache *mca_cache = &adev->mca.mca_caches[type];
+	int ret;
+
+	amdgpu_mca_bank_set_init(&mca_set);
+
+	ret = amdgpu_mca_smu_get_mca_set(adev, type, &mca_set, qctx);
+	if (ret)
+		goto out_mca_release;
+
+	ret = amdgpu_mca_dispatch_mca_set(adev, blk, type, &mca_set, err_data);
+	if (ret)
+		goto out_mca_release;
+
+	/* add remain mca bank to mca cache */
+	if (mca_set.nr_entries) {
+		ret = amdgpu_mca_add_mca_set_to_cache(adev, type, &mca_set);
+		if (ret)
+			goto out_mca_release;
+	}
+
+	/* dispatch mca set again if mca cache has valid data */
+	mutex_lock(&mca_cache->lock);
+	if (mca_cache->mca_set.nr_entries)
+		ret = amdgpu_mca_dispatch_mca_set(adev, blk, type, &mca_cache->mca_set, err_data);
+	mutex_unlock(&mca_cache->lock);
+
+out_mca_release:
+	amdgpu_mca_bank_set_release(&mca_set);
+
+	return ret;
+}
+
+#if defined(CONFIG_DEBUG_FS)
+static int amdgpu_mca_smu_debug_mode_set(void *data, u64 val)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)data;
+	int ret;
+
+	ret = amdgpu_ras_set_mca_debug_mode(adev, val ? true : false);
+	if (ret)
+		return ret;
+
+	dev_info(adev->dev, "amdgpu set smu mca debug mode %s success\n", val ? "on" : "off");
+
+	return 0;
+}
+
+static void mca_dump_entry(struct seq_file *m, struct mca_bank_entry *entry)
+{
+	int i, idx = entry->idx;
+	int reg_idx_array[] = {
+		MCA_REG_IDX_STATUS,
+		MCA_REG_IDX_ADDR,
+		MCA_REG_IDX_MISC0,
+		MCA_REG_IDX_IPID,
+		MCA_REG_IDX_SYND,
+	};
+
+	seq_printf(m, "mca entry[%d].type: %s\n", idx, entry->type == AMDGPU_MCA_ERROR_TYPE_UE ? "UE" : "CE");
+	seq_printf(m, "mca entry[%d].ip: %d\n", idx, entry->ip);
+	seq_printf(m, "mca entry[%d].info: socketid:%d aid:%d hwid:0x%03x mcatype:0x%04x\n",
+		   idx, entry->info.socket_id, entry->info.aid, entry->info.hwid, entry->info.mcatype);
+
+	for (i = 0; i < ARRAY_SIZE(reg_idx_array); i++)
+		seq_printf(m, "mca entry[%d].regs[%d]: 0x%016llx\n", idx, reg_idx_array[i], entry->regs[reg_idx_array[i]]);
+}
+
+static int mca_dump_show(struct seq_file *m, enum amdgpu_mca_error_type type)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)m->private;
+	struct mca_bank_node *node;
+	struct mca_bank_set mca_set;
+	struct ras_query_context qctx;
+	int ret;
+
+	amdgpu_mca_bank_set_init(&mca_set);
+
+	qctx.evid.event_id = RAS_EVENT_INVALID_ID;
+	ret = amdgpu_mca_smu_get_mca_set(adev, type, &mca_set, &qctx);
+	if (ret)
+		goto err_free_mca_set;
+
+	seq_printf(m, "amdgpu smu %s valid mca count: %d\n",
+		   type == AMDGPU_MCA_ERROR_TYPE_UE ? "UE" : "CE", mca_set.nr_entries);
+
+	if (!mca_set.nr_entries)
+		goto err_free_mca_set;
+
+	list_for_each_entry(node, &mca_set.list, node)
+		mca_dump_entry(m, &node->entry);
+
+	/* add mca bank to mca bank cache */
+	ret = amdgpu_mca_add_mca_set_to_cache(adev, type, &mca_set);
+
+err_free_mca_set:
+	amdgpu_mca_bank_set_release(&mca_set);
+
+	return ret;
+}
+
+static int mca_dump_ce_show(struct seq_file *m, void *unused)
+{
+	return mca_dump_show(m, AMDGPU_MCA_ERROR_TYPE_CE);
+}
+
+static int mca_dump_ce_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, mca_dump_ce_show, inode->i_private);
+}
+
+static const struct file_operations mca_ce_dump_debug_fops = {
+	.owner = THIS_MODULE,
+	.open = mca_dump_ce_open,
+	.read = seq_read,
+	.llseek = seq_lseek,
+	.release = single_release,
+};
+
+static int mca_dump_ue_show(struct seq_file *m, void *unused)
+{
+	return mca_dump_show(m, AMDGPU_MCA_ERROR_TYPE_UE);
+}
+
+static int mca_dump_ue_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, mca_dump_ue_show, inode->i_private);
+}
+
+static const struct file_operations mca_ue_dump_debug_fops = {
+	.owner = THIS_MODULE,
+	.open = mca_dump_ue_open,
+	.read = seq_read,
+	.llseek = seq_lseek,
+	.release = single_release,
+};
+
+DEFINE_DEBUGFS_ATTRIBUTE(mca_debug_mode_fops, NULL, amdgpu_mca_smu_debug_mode_set, "%llu\n");
+#endif
+
+void amdgpu_mca_smu_debugfs_init(struct amdgpu_device *adev, struct dentry *root)
+{
+#if defined(CONFIG_DEBUG_FS)
+	if (!root)
+		return;
+
+	debugfs_create_file("mca_debug_mode", 0200, root, adev, &mca_debug_mode_fops);
+	debugfs_create_file("mca_ue_dump", 0400, root, adev, &mca_ue_dump_debug_fops);
+	debugfs_create_file("mca_ce_dump", 0400, root, adev, &mca_ce_dump_debug_fops);
+#endif
+}
+
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h
new file mode 100644
index 000000000000..e80323ff90c1
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h
@@ -0,0 +1,169 @@
+/*
+ * Copyright (C) 2021  Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+#ifndef __AMDGPU_MCA_H__
+#define __AMDGPU_MCA_H__
+
+#include "amdgpu_ras.h"
+
+#define MCA_MAX_REGS_COUNT	(16)
+
+#define MCA_REG_FIELD(x, h, l)			(((x) & GENMASK_ULL(h, l)) >> l)
+#define MCA_REG__STATUS__VAL(x)			MCA_REG_FIELD(x, 63, 63)
+#define MCA_REG__STATUS__OVERFLOW(x)		MCA_REG_FIELD(x, 62, 62)
+#define MCA_REG__STATUS__UC(x)			MCA_REG_FIELD(x, 61, 61)
+#define MCA_REG__STATUS__EN(x)			MCA_REG_FIELD(x, 60, 60)
+#define MCA_REG__STATUS__MISCV(x)		MCA_REG_FIELD(x, 59, 59)
+#define MCA_REG__STATUS__ADDRV(x)		MCA_REG_FIELD(x, 58, 58)
+#define MCA_REG__STATUS__PCC(x)			MCA_REG_FIELD(x, 57, 57)
+#define MCA_REG__STATUS__ERRCOREIDVAL(x)	MCA_REG_FIELD(x, 56, 56)
+#define MCA_REG__STATUS__TCC(x)			MCA_REG_FIELD(x, 55, 55)
+#define MCA_REG__STATUS__SYNDV(x)		MCA_REG_FIELD(x, 53, 53)
+#define MCA_REG__STATUS__CECC(x)		MCA_REG_FIELD(x, 46, 46)
+#define MCA_REG__STATUS__UECC(x)		MCA_REG_FIELD(x, 45, 45)
+#define MCA_REG__STATUS__DEFERRED(x)		MCA_REG_FIELD(x, 44, 44)
+#define MCA_REG__STATUS__POISON(x)		MCA_REG_FIELD(x, 43, 43)
+#define MCA_REG__STATUS__SCRUB(x)		MCA_REG_FIELD(x, 40, 40)
+#define MCA_REG__STATUS__ERRCOREID(x)		MCA_REG_FIELD(x, 37, 32)
+#define MCA_REG__STATUS__ADDRLSB(x)		MCA_REG_FIELD(x, 29, 24)
+#define MCA_REG__STATUS__ERRORCODEEXT(x)	MCA_REG_FIELD(x, 21, 16)
+#define MCA_REG__STATUS__ERRORCODE(x)		MCA_REG_FIELD(x, 15, 0)
+
+#define MCA_REG__MISC0__ERRCNT(x)		MCA_REG_FIELD(x, 43, 32)
+
+#define MCA_REG__SYND__ERRORINFORMATION(x)	MCA_REG_FIELD(x, 17, 0)
+
+enum amdgpu_mca_ip {
+	AMDGPU_MCA_IP_UNKNOW = -1,
+	AMDGPU_MCA_IP_PSP = 0,
+	AMDGPU_MCA_IP_SDMA,
+	AMDGPU_MCA_IP_GC,
+	AMDGPU_MCA_IP_SMU,
+	AMDGPU_MCA_IP_MP5,
+	AMDGPU_MCA_IP_UMC,
+	AMDGPU_MCA_IP_PCS_XGMI,
+	AMDGPU_MCA_IP_COUNT,
+};
+
+enum amdgpu_mca_error_type {
+	AMDGPU_MCA_ERROR_TYPE_UE = 0,
+	AMDGPU_MCA_ERROR_TYPE_CE,
+	AMDGPU_MCA_ERROR_TYPE_DE,
+};
+
+struct amdgpu_mca_ras_block {
+	struct amdgpu_ras_block_object ras_block;
+};
+
+struct amdgpu_mca_ras {
+	struct ras_common_if *ras_if;
+	struct amdgpu_mca_ras_block *ras;
+};
+
+struct mca_bank_set {
+	int nr_entries;
+	struct list_head list;
+};
+
+struct mca_bank_cache {
+	struct mca_bank_set mca_set;
+	struct mutex lock;
+};
+
+struct amdgpu_mca {
+	struct amdgpu_mca_ras mp0;
+	struct amdgpu_mca_ras mp1;
+	struct amdgpu_mca_ras mpio;
+	const struct amdgpu_mca_smu_funcs *mca_funcs;
+	struct mca_bank_cache mca_caches[AMDGPU_MCA_ERROR_TYPE_DE];
+	atomic_t ue_update_flag;
+};
+
+enum mca_reg_idx {
+	MCA_REG_IDX_STATUS		= 1,
+	MCA_REG_IDX_ADDR		= 2,
+	MCA_REG_IDX_MISC0		= 3,
+	MCA_REG_IDX_IPID		= 5,
+	MCA_REG_IDX_SYND		= 6,
+	MCA_REG_IDX_COUNT		= 16,
+};
+
+struct mca_bank_info {
+	int socket_id;
+	int aid;
+	int hwid;
+	int mcatype;
+};
+
+struct mca_bank_entry {
+	int idx;
+	enum amdgpu_mca_error_type type;
+	enum amdgpu_mca_ip ip;
+	struct mca_bank_info info;
+	uint64_t regs[MCA_MAX_REGS_COUNT];
+};
+
+struct mca_bank_node {
+	struct mca_bank_entry entry;
+	struct list_head node;
+};
+
+struct amdgpu_mca_smu_funcs {
+	int max_ue_count;
+	int max_ce_count;
+	int (*mca_set_debug_mode)(struct amdgpu_device *adev, bool enable);
+	int (*mca_parse_mca_error_count)(struct amdgpu_device *adev, enum amdgpu_ras_block blk, enum amdgpu_mca_error_type type,
+					 struct mca_bank_entry *entry, uint32_t *count);
+	int (*mca_get_valid_mca_count)(struct amdgpu_device *adev, enum amdgpu_mca_error_type type,
+				       uint32_t *count);
+	int (*mca_get_mca_entry)(struct amdgpu_device *adev, enum amdgpu_mca_error_type type,
+				 int idx, struct mca_bank_entry *entry);
+};
+
+void amdgpu_mca_query_correctable_error_count(struct amdgpu_device *adev,
+					      uint64_t mc_status_addr,
+					      unsigned long *error_count);
+
+void amdgpu_mca_query_uncorrectable_error_count(struct amdgpu_device *adev,
+						uint64_t mc_status_addr,
+						unsigned long *error_count);
+
+void amdgpu_mca_reset_error_count(struct amdgpu_device *adev,
+				  uint64_t mc_status_addr);
+
+void amdgpu_mca_query_ras_error_count(struct amdgpu_device *adev,
+				      uint64_t mc_status_addr,
+				      void *ras_error_status);
+int amdgpu_mca_mp0_ras_sw_init(struct amdgpu_device *adev);
+int amdgpu_mca_mp1_ras_sw_init(struct amdgpu_device *adev);
+int amdgpu_mca_mpio_ras_sw_init(struct amdgpu_device *adev);
+
+void amdgpu_mca_smu_init_funcs(struct amdgpu_device *adev, const struct amdgpu_mca_smu_funcs *mca_funcs);
+int amdgpu_mca_init(struct amdgpu_device *adev);
+void amdgpu_mca_fini(struct amdgpu_device *adev);
+int amdgpu_mca_reset(struct amdgpu_device *adev);
+int amdgpu_mca_smu_set_debug_mode(struct amdgpu_device *adev, bool enable);
+int amdgpu_mca_smu_get_mca_set_error_count(struct amdgpu_device *adev, enum amdgpu_ras_block blk,
+					   enum amdgpu_mca_error_type type, uint32_t *total);
+void amdgpu_mca_smu_debugfs_init(struct amdgpu_device *adev, struct dentry *root);
+int amdgpu_mca_smu_log_ras_error(struct amdgpu_device *adev, enum amdgpu_ras_block blk, enum amdgpu_mca_error_type type,
+				 struct ras_err_data *err_data, struct ras_query_context *qctx);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c
new file mode 100644
index 000000000000..5bf9be073cdd
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c
@@ -0,0 +1,770 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include <drm/drm_exec.h>
+
+#include "amdgpu_mes.h"
+#include "amdgpu.h"
+#include "soc15_common.h"
+#include "amdgpu_mes_ctx.h"
+
+#define AMDGPU_MES_MAX_NUM_OF_QUEUES_PER_PROCESS 1024
+#define AMDGPU_ONE_DOORBELL_SIZE 8
+
+int amdgpu_mes_doorbell_process_slice(struct amdgpu_device *adev)
+{
+	return roundup(AMDGPU_ONE_DOORBELL_SIZE *
+		       AMDGPU_MES_MAX_NUM_OF_QUEUES_PER_PROCESS,
+		       PAGE_SIZE);
+}
+
+static int amdgpu_mes_doorbell_init(struct amdgpu_device *adev)
+{
+	int i;
+	struct amdgpu_mes *mes = &adev->mes;
+
+	/* Bitmap for dynamic allocation of kernel doorbells */
+	mes->doorbell_bitmap = bitmap_zalloc(PAGE_SIZE / sizeof(u32), GFP_KERNEL);
+	if (!mes->doorbell_bitmap) {
+		dev_err(adev->dev, "Failed to allocate MES doorbell bitmap\n");
+		return -ENOMEM;
+	}
+
+	mes->num_mes_dbs = PAGE_SIZE / AMDGPU_ONE_DOORBELL_SIZE;
+	for (i = 0; i < AMDGPU_MES_PRIORITY_NUM_LEVELS; i++) {
+		adev->mes.aggregated_doorbells[i] = mes->db_start_dw_offset + i * 2;
+		set_bit(i, mes->doorbell_bitmap);
+	}
+
+	return 0;
+}
+
+static int amdgpu_mes_event_log_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (!amdgpu_mes_log_enable)
+		return 0;
+
+	r = amdgpu_bo_create_kernel(adev, adev->mes.event_log_size, PAGE_SIZE,
+				    AMDGPU_GEM_DOMAIN_VRAM,
+				    &adev->mes.event_log_gpu_obj,
+				    &adev->mes.event_log_gpu_addr,
+				    &adev->mes.event_log_cpu_addr);
+	if (r) {
+		dev_warn(adev->dev, "failed to create MES event log buffer (%d)", r);
+		return r;
+	}
+
+	memset(adev->mes.event_log_cpu_addr, 0, adev->mes.event_log_size);
+
+	return  0;
+
+}
+
+static void amdgpu_mes_doorbell_free(struct amdgpu_device *adev)
+{
+	bitmap_free(adev->mes.doorbell_bitmap);
+}
+
+int amdgpu_mes_init(struct amdgpu_device *adev)
+{
+	int i, r, num_pipes;
+
+	adev->mes.adev = adev;
+
+	idr_init(&adev->mes.pasid_idr);
+	idr_init(&adev->mes.gang_id_idr);
+	idr_init(&adev->mes.queue_id_idr);
+	ida_init(&adev->mes.doorbell_ida);
+	spin_lock_init(&adev->mes.queue_id_lock);
+	mutex_init(&adev->mes.mutex_hidden);
+
+	for (i = 0; i < AMDGPU_MAX_MES_PIPES; i++)
+		spin_lock_init(&adev->mes.ring_lock[i]);
+
+	adev->mes.total_max_queue = AMDGPU_FENCE_MES_QUEUE_ID_MASK;
+	adev->mes.vmid_mask_mmhub = 0xffffff00;
+	adev->mes.vmid_mask_gfxhub = adev->gfx.disable_kq ? 0xfffffffe : 0xffffff00;
+
+	num_pipes = adev->gfx.me.num_pipe_per_me * adev->gfx.me.num_me;
+	if (num_pipes > AMDGPU_MES_MAX_GFX_PIPES)
+		dev_warn(adev->dev, "more gfx pipes than supported by MES! (%d vs %d)\n",
+			 num_pipes, AMDGPU_MES_MAX_GFX_PIPES);
+
+	for (i = 0; i < AMDGPU_MES_MAX_GFX_PIPES; i++) {
+		if (i >= num_pipes)
+			break;
+		if (amdgpu_ip_version(adev, GC_HWIP, 0) >=
+		    IP_VERSION(12, 0, 0))
+			/*
+			 * GFX V12 has only one GFX pipe, but 8 queues in it.
+			 * GFX pipe 0 queue 0 is being used by Kernel queue.
+			 * Set GFX pipe 0 queue 1-7 for MES scheduling
+			 * mask = 1111 1110b
+			 */
+			adev->mes.gfx_hqd_mask[i] = adev->gfx.disable_kq ? 0xFF : 0xFE;
+		else
+			/*
+			 * GFX pipe 0 queue 0 is being used by Kernel queue.
+			 * Set GFX pipe 0 queue 1 for MES scheduling
+			 * mask = 10b
+			 */
+			adev->mes.gfx_hqd_mask[i] = adev->gfx.disable_kq ? 0x3 : 0x2;
+	}
+
+	num_pipes = adev->gfx.mec.num_pipe_per_mec * adev->gfx.mec.num_mec;
+	if (num_pipes > AMDGPU_MES_MAX_COMPUTE_PIPES)
+		dev_warn(adev->dev, "more compute pipes than supported by MES! (%d vs %d)\n",
+			 num_pipes, AMDGPU_MES_MAX_COMPUTE_PIPES);
+
+	for (i = 0; i < AMDGPU_MES_MAX_COMPUTE_PIPES; i++) {
+		if (i >= num_pipes)
+			break;
+		adev->mes.compute_hqd_mask[i] = adev->gfx.disable_kq ? 0xF : 0xC;
+	}
+
+	num_pipes = adev->sdma.num_instances;
+	if (num_pipes > AMDGPU_MES_MAX_SDMA_PIPES)
+		dev_warn(adev->dev, "more SDMA pipes than supported by MES! (%d vs %d)\n",
+			 num_pipes, AMDGPU_MES_MAX_SDMA_PIPES);
+
+	for (i = 0; i < AMDGPU_MES_MAX_SDMA_PIPES; i++) {
+		if (i >= num_pipes)
+			break;
+		adev->mes.sdma_hqd_mask[i] = 0xfc;
+	}
+
+	for (i = 0; i < AMDGPU_MAX_MES_PIPES; i++) {
+		r = amdgpu_device_wb_get(adev, &adev->mes.sch_ctx_offs[i]);
+		if (r) {
+			dev_err(adev->dev,
+				"(%d) ring trail_fence_offs wb alloc failed\n",
+				r);
+			goto error;
+		}
+		adev->mes.sch_ctx_gpu_addr[i] =
+			adev->wb.gpu_addr + (adev->mes.sch_ctx_offs[i] * 4);
+		adev->mes.sch_ctx_ptr[i] =
+			(uint64_t *)&adev->wb.wb[adev->mes.sch_ctx_offs[i]];
+
+		r = amdgpu_device_wb_get(adev,
+				 &adev->mes.query_status_fence_offs[i]);
+		if (r) {
+			dev_err(adev->dev,
+			      "(%d) query_status_fence_offs wb alloc failed\n",
+			      r);
+			goto error;
+		}
+		adev->mes.query_status_fence_gpu_addr[i] = adev->wb.gpu_addr +
+			(adev->mes.query_status_fence_offs[i] * 4);
+		adev->mes.query_status_fence_ptr[i] =
+			(uint64_t *)&adev->wb.wb[adev->mes.query_status_fence_offs[i]];
+	}
+
+	r = amdgpu_mes_doorbell_init(adev);
+	if (r)
+		goto error;
+
+	r = amdgpu_mes_event_log_init(adev);
+	if (r)
+		goto error_doorbell;
+
+	if (adev->mes.hung_queue_db_array_size) {
+		r = amdgpu_bo_create_kernel(adev,
+					    adev->mes.hung_queue_db_array_size * sizeof(u32),
+					    PAGE_SIZE,
+					    AMDGPU_GEM_DOMAIN_GTT,
+					    &adev->mes.hung_queue_db_array_gpu_obj,
+					    &adev->mes.hung_queue_db_array_gpu_addr,
+					    &adev->mes.hung_queue_db_array_cpu_addr);
+		if (r) {
+			dev_warn(adev->dev, "failed to create MES hung db array buffer (%d)", r);
+			goto error_doorbell;
+		}
+	}
+
+	return 0;
+
+error_doorbell:
+	amdgpu_mes_doorbell_free(adev);
+error:
+	for (i = 0; i < AMDGPU_MAX_MES_PIPES; i++) {
+		if (adev->mes.sch_ctx_ptr[i])
+			amdgpu_device_wb_free(adev, adev->mes.sch_ctx_offs[i]);
+		if (adev->mes.query_status_fence_ptr[i])
+			amdgpu_device_wb_free(adev,
+				      adev->mes.query_status_fence_offs[i]);
+	}
+
+	idr_destroy(&adev->mes.pasid_idr);
+	idr_destroy(&adev->mes.gang_id_idr);
+	idr_destroy(&adev->mes.queue_id_idr);
+	ida_destroy(&adev->mes.doorbell_ida);
+	mutex_destroy(&adev->mes.mutex_hidden);
+	return r;
+}
+
+void amdgpu_mes_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	amdgpu_bo_free_kernel(&adev->mes.hung_queue_db_array_gpu_obj,
+			      &adev->mes.hung_queue_db_array_gpu_addr,
+			      &adev->mes.hung_queue_db_array_cpu_addr);
+
+	amdgpu_bo_free_kernel(&adev->mes.event_log_gpu_obj,
+			      &adev->mes.event_log_gpu_addr,
+			      &adev->mes.event_log_cpu_addr);
+
+	for (i = 0; i < AMDGPU_MAX_MES_PIPES; i++) {
+		if (adev->mes.sch_ctx_ptr[i])
+			amdgpu_device_wb_free(adev, adev->mes.sch_ctx_offs[i]);
+		if (adev->mes.query_status_fence_ptr[i])
+			amdgpu_device_wb_free(adev,
+				      adev->mes.query_status_fence_offs[i]);
+	}
+
+	amdgpu_mes_doorbell_free(adev);
+
+	idr_destroy(&adev->mes.pasid_idr);
+	idr_destroy(&adev->mes.gang_id_idr);
+	idr_destroy(&adev->mes.queue_id_idr);
+	ida_destroy(&adev->mes.doorbell_ida);
+	mutex_destroy(&adev->mes.mutex_hidden);
+}
+
+int amdgpu_mes_suspend(struct amdgpu_device *adev)
+{
+	struct mes_suspend_gang_input input;
+	int r;
+
+	if (!amdgpu_mes_suspend_resume_all_supported(adev))
+		return 0;
+
+	memset(&input, 0x0, sizeof(struct mes_suspend_gang_input));
+	input.suspend_all_gangs = 1;
+
+	/*
+	 * Avoid taking any other locks under MES lock to avoid circular
+	 * lock dependencies.
+	 */
+	amdgpu_mes_lock(&adev->mes);
+	r = adev->mes.funcs->suspend_gang(&adev->mes, &input);
+	amdgpu_mes_unlock(&adev->mes);
+	if (r)
+		dev_err(adev->dev, "failed to suspend all gangs");
+
+	return r;
+}
+
+int amdgpu_mes_resume(struct amdgpu_device *adev)
+{
+	struct mes_resume_gang_input input;
+	int r;
+
+	if (!amdgpu_mes_suspend_resume_all_supported(adev))
+		return 0;
+
+	memset(&input, 0x0, sizeof(struct mes_resume_gang_input));
+	input.resume_all_gangs = 1;
+
+	/*
+	 * Avoid taking any other locks under MES lock to avoid circular
+	 * lock dependencies.
+	 */
+	amdgpu_mes_lock(&adev->mes);
+	r = adev->mes.funcs->resume_gang(&adev->mes, &input);
+	amdgpu_mes_unlock(&adev->mes);
+	if (r)
+		dev_err(adev->dev, "failed to resume all gangs");
+
+	return r;
+}
+
+int amdgpu_mes_map_legacy_queue(struct amdgpu_device *adev,
+				struct amdgpu_ring *ring)
+{
+	struct mes_map_legacy_queue_input queue_input;
+	int r;
+
+	memset(&queue_input, 0, sizeof(queue_input));
+
+	queue_input.queue_type = ring->funcs->type;
+	queue_input.doorbell_offset = ring->doorbell_index;
+	queue_input.pipe_id = ring->pipe;
+	queue_input.queue_id = ring->queue;
+	queue_input.mqd_addr = amdgpu_bo_gpu_offset(ring->mqd_obj);
+	queue_input.wptr_addr = ring->wptr_gpu_addr;
+
+	amdgpu_mes_lock(&adev->mes);
+	r = adev->mes.funcs->map_legacy_queue(&adev->mes, &queue_input);
+	amdgpu_mes_unlock(&adev->mes);
+	if (r)
+		dev_err(adev->dev, "failed to map legacy queue\n");
+
+	return r;
+}
+
+int amdgpu_mes_unmap_legacy_queue(struct amdgpu_device *adev,
+				  struct amdgpu_ring *ring,
+				  enum amdgpu_unmap_queues_action action,
+				  u64 gpu_addr, u64 seq)
+{
+	struct mes_unmap_legacy_queue_input queue_input;
+	int r;
+
+	queue_input.action = action;
+	queue_input.queue_type = ring->funcs->type;
+	queue_input.doorbell_offset = ring->doorbell_index;
+	queue_input.pipe_id = ring->pipe;
+	queue_input.queue_id = ring->queue;
+	queue_input.trail_fence_addr = gpu_addr;
+	queue_input.trail_fence_data = seq;
+
+	amdgpu_mes_lock(&adev->mes);
+	r = adev->mes.funcs->unmap_legacy_queue(&adev->mes, &queue_input);
+	amdgpu_mes_unlock(&adev->mes);
+	if (r)
+		dev_err(adev->dev, "failed to unmap legacy queue\n");
+
+	return r;
+}
+
+int amdgpu_mes_reset_legacy_queue(struct amdgpu_device *adev,
+				  struct amdgpu_ring *ring,
+				  unsigned int vmid,
+				  bool use_mmio)
+{
+	struct mes_reset_queue_input queue_input;
+	int r;
+
+	memset(&queue_input, 0, sizeof(queue_input));
+
+	queue_input.queue_type = ring->funcs->type;
+	queue_input.doorbell_offset = ring->doorbell_index;
+	queue_input.me_id = ring->me;
+	queue_input.pipe_id = ring->pipe;
+	queue_input.queue_id = ring->queue;
+	queue_input.mqd_addr = ring->mqd_obj ? amdgpu_bo_gpu_offset(ring->mqd_obj) : 0;
+	queue_input.wptr_addr = ring->wptr_gpu_addr;
+	queue_input.vmid = vmid;
+	queue_input.use_mmio = use_mmio;
+	queue_input.is_kq = true;
+	if (ring->funcs->type == AMDGPU_RING_TYPE_GFX)
+		queue_input.legacy_gfx = true;
+
+	amdgpu_mes_lock(&adev->mes);
+	r = adev->mes.funcs->reset_hw_queue(&adev->mes, &queue_input);
+	amdgpu_mes_unlock(&adev->mes);
+	if (r)
+		dev_err(adev->dev, "failed to reset legacy queue\n");
+
+	return r;
+}
+
+int amdgpu_mes_get_hung_queue_db_array_size(struct amdgpu_device *adev)
+{
+	return adev->mes.hung_queue_db_array_size;
+}
+
+int amdgpu_mes_detect_and_reset_hung_queues(struct amdgpu_device *adev,
+					    int queue_type,
+					    bool detect_only,
+					    unsigned int *hung_db_num,
+					    u32 *hung_db_array)
+
+{
+	struct mes_detect_and_reset_queue_input input;
+	u32 *db_array = adev->mes.hung_queue_db_array_cpu_addr;
+	int r, i;
+
+	if (!hung_db_num || !hung_db_array)
+		return -EINVAL;
+
+	if ((queue_type != AMDGPU_RING_TYPE_GFX) &&
+	    (queue_type != AMDGPU_RING_TYPE_COMPUTE) &&
+	    (queue_type != AMDGPU_RING_TYPE_SDMA))
+		return -EINVAL;
+
+	/* Clear the doorbell array before detection */
+	memset(adev->mes.hung_queue_db_array_cpu_addr, 0,
+		adev->mes.hung_queue_db_array_size * sizeof(u32));
+	input.queue_type = queue_type;
+	input.detect_only = detect_only;
+
+	r = adev->mes.funcs->detect_and_reset_hung_queues(&adev->mes,
+							  &input);
+	if (r) {
+		dev_err(adev->dev, "failed to detect and reset\n");
+	} else {
+		*hung_db_num = 0;
+		for (i = 0; i < adev->mes.hung_queue_db_array_size; i++) {
+			if (db_array[i] != AMDGPU_MES_INVALID_DB_OFFSET) {
+				hung_db_array[i] = db_array[i];
+				*hung_db_num += 1;
+			}
+		}
+	}
+
+	return r;
+}
+
+uint32_t amdgpu_mes_rreg(struct amdgpu_device *adev, uint32_t reg)
+{
+	struct mes_misc_op_input op_input;
+	int r, val = 0;
+	uint32_t addr_offset = 0;
+	uint64_t read_val_gpu_addr;
+	uint32_t *read_val_ptr;
+
+	if (amdgpu_device_wb_get(adev, &addr_offset)) {
+		dev_err(adev->dev, "critical bug! too many mes readers\n");
+		goto error;
+	}
+	read_val_gpu_addr = adev->wb.gpu_addr + (addr_offset * 4);
+	read_val_ptr = (uint32_t *)&adev->wb.wb[addr_offset];
+	op_input.op = MES_MISC_OP_READ_REG;
+	op_input.read_reg.reg_offset = reg;
+	op_input.read_reg.buffer_addr = read_val_gpu_addr;
+
+	if (!adev->mes.funcs->misc_op) {
+		dev_err(adev->dev, "mes rreg is not supported!\n");
+		goto error;
+	}
+
+	amdgpu_mes_lock(&adev->mes);
+	r = adev->mes.funcs->misc_op(&adev->mes, &op_input);
+	amdgpu_mes_unlock(&adev->mes);
+	if (r)
+		dev_err(adev->dev, "failed to read reg (0x%x)\n", reg);
+	else
+		val = *(read_val_ptr);
+
+error:
+	if (addr_offset)
+		amdgpu_device_wb_free(adev, addr_offset);
+	return val;
+}
+
+int amdgpu_mes_wreg(struct amdgpu_device *adev,
+		    uint32_t reg, uint32_t val)
+{
+	struct mes_misc_op_input op_input;
+	int r;
+
+	op_input.op = MES_MISC_OP_WRITE_REG;
+	op_input.write_reg.reg_offset = reg;
+	op_input.write_reg.reg_value = val;
+
+	if (!adev->mes.funcs->misc_op) {
+		dev_err(adev->dev, "mes wreg is not supported!\n");
+		r = -EINVAL;
+		goto error;
+	}
+
+	amdgpu_mes_lock(&adev->mes);
+	r = adev->mes.funcs->misc_op(&adev->mes, &op_input);
+	amdgpu_mes_unlock(&adev->mes);
+	if (r)
+		dev_err(adev->dev, "failed to write reg (0x%x)\n", reg);
+
+error:
+	return r;
+}
+
+int amdgpu_mes_reg_write_reg_wait(struct amdgpu_device *adev,
+				  uint32_t reg0, uint32_t reg1,
+				  uint32_t ref, uint32_t mask)
+{
+	struct mes_misc_op_input op_input;
+	int r;
+
+	op_input.op = MES_MISC_OP_WRM_REG_WR_WAIT;
+	op_input.wrm_reg.reg0 = reg0;
+	op_input.wrm_reg.reg1 = reg1;
+	op_input.wrm_reg.ref = ref;
+	op_input.wrm_reg.mask = mask;
+
+	if (!adev->mes.funcs->misc_op) {
+		dev_err(adev->dev, "mes reg_write_reg_wait is not supported!\n");
+		r = -EINVAL;
+		goto error;
+	}
+
+	amdgpu_mes_lock(&adev->mes);
+	r = adev->mes.funcs->misc_op(&adev->mes, &op_input);
+	amdgpu_mes_unlock(&adev->mes);
+	if (r)
+		dev_err(adev->dev, "failed to reg_write_reg_wait\n");
+
+error:
+	return r;
+}
+
+int amdgpu_mes_set_shader_debugger(struct amdgpu_device *adev,
+				uint64_t process_context_addr,
+				uint32_t spi_gdbg_per_vmid_cntl,
+				const uint32_t *tcp_watch_cntl,
+				uint32_t flags,
+				bool trap_en)
+{
+	struct mes_misc_op_input op_input = {0};
+	int r;
+
+	if (!adev->mes.funcs->misc_op) {
+		dev_err(adev->dev,
+			"mes set shader debugger is not supported!\n");
+		return -EINVAL;
+	}
+
+	op_input.op = MES_MISC_OP_SET_SHADER_DEBUGGER;
+	op_input.set_shader_debugger.process_context_addr = process_context_addr;
+	op_input.set_shader_debugger.flags.u32all = flags;
+
+	/* use amdgpu mes_flush_shader_debugger instead */
+	if (op_input.set_shader_debugger.flags.process_ctx_flush)
+		return -EINVAL;
+
+	op_input.set_shader_debugger.spi_gdbg_per_vmid_cntl = spi_gdbg_per_vmid_cntl;
+	memcpy(op_input.set_shader_debugger.tcp_watch_cntl, tcp_watch_cntl,
+			sizeof(op_input.set_shader_debugger.tcp_watch_cntl));
+
+	if (((adev->mes.sched_version & AMDGPU_MES_API_VERSION_MASK) >>
+			AMDGPU_MES_API_VERSION_SHIFT) >= 14)
+		op_input.set_shader_debugger.trap_en = trap_en;
+
+	amdgpu_mes_lock(&adev->mes);
+
+	r = adev->mes.funcs->misc_op(&adev->mes, &op_input);
+	if (r)
+		dev_err(adev->dev, "failed to set_shader_debugger\n");
+
+	amdgpu_mes_unlock(&adev->mes);
+
+	return r;
+}
+
+int amdgpu_mes_flush_shader_debugger(struct amdgpu_device *adev,
+				     uint64_t process_context_addr)
+{
+	struct mes_misc_op_input op_input = {0};
+	int r;
+
+	if (!adev->mes.funcs->misc_op) {
+		dev_err(adev->dev,
+			"mes flush shader debugger is not supported!\n");
+		return -EINVAL;
+	}
+
+	op_input.op = MES_MISC_OP_SET_SHADER_DEBUGGER;
+	op_input.set_shader_debugger.process_context_addr = process_context_addr;
+	op_input.set_shader_debugger.flags.process_ctx_flush = true;
+
+	amdgpu_mes_lock(&adev->mes);
+
+	r = adev->mes.funcs->misc_op(&adev->mes, &op_input);
+	if (r)
+		dev_err(adev->dev, "failed to set_shader_debugger\n");
+
+	amdgpu_mes_unlock(&adev->mes);
+
+	return r;
+}
+
+uint32_t amdgpu_mes_get_aggregated_doorbell_index(struct amdgpu_device *adev,
+						   enum amdgpu_mes_priority_level prio)
+{
+	return adev->mes.aggregated_doorbells[prio];
+}
+
+int amdgpu_mes_init_microcode(struct amdgpu_device *adev, int pipe)
+{
+	const struct mes_firmware_header_v1_0 *mes_hdr;
+	struct amdgpu_firmware_info *info;
+	char ucode_prefix[30];
+	char fw_name[50];
+	bool need_retry = false;
+	u32 *ucode_ptr;
+	int r;
+
+	amdgpu_ucode_ip_version_decode(adev, GC_HWIP, ucode_prefix,
+				       sizeof(ucode_prefix));
+	if (adev->enable_uni_mes) {
+		snprintf(fw_name, sizeof(fw_name),
+			 "amdgpu/%s_uni_mes.bin", ucode_prefix);
+	} else if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(11, 0, 0) &&
+	    amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(12, 0, 0)) {
+		snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mes%s.bin",
+			 ucode_prefix,
+			 pipe == AMDGPU_MES_SCHED_PIPE ? "_2" : "1");
+		need_retry = true;
+	} else {
+		snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mes%s.bin",
+			 ucode_prefix,
+			 pipe == AMDGPU_MES_SCHED_PIPE ? "" : "1");
+	}
+
+	r = amdgpu_ucode_request(adev, &adev->mes.fw[pipe], AMDGPU_UCODE_REQUIRED,
+				 "%s", fw_name);
+	if (r && need_retry && pipe == AMDGPU_MES_SCHED_PIPE) {
+		dev_info(adev->dev, "try to fall back to %s_mes.bin\n", ucode_prefix);
+		r = amdgpu_ucode_request(adev, &adev->mes.fw[pipe],
+					 AMDGPU_UCODE_REQUIRED,
+					 "amdgpu/%s_mes.bin", ucode_prefix);
+	}
+
+	if (r)
+		goto out;
+
+	mes_hdr = (const struct mes_firmware_header_v1_0 *)
+		adev->mes.fw[pipe]->data;
+	adev->mes.uc_start_addr[pipe] =
+		le32_to_cpu(mes_hdr->mes_uc_start_addr_lo) |
+		((uint64_t)(le32_to_cpu(mes_hdr->mes_uc_start_addr_hi)) << 32);
+	adev->mes.data_start_addr[pipe] =
+		le32_to_cpu(mes_hdr->mes_data_start_addr_lo) |
+		((uint64_t)(le32_to_cpu(mes_hdr->mes_data_start_addr_hi)) << 32);
+	ucode_ptr = (u32 *)(adev->mes.fw[pipe]->data +
+			  sizeof(union amdgpu_firmware_header));
+	adev->mes.fw_version[pipe] =
+		le32_to_cpu(ucode_ptr[24]) & AMDGPU_MES_VERSION_MASK;
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		int ucode, ucode_data;
+
+		if (pipe == AMDGPU_MES_SCHED_PIPE) {
+			ucode = AMDGPU_UCODE_ID_CP_MES;
+			ucode_data = AMDGPU_UCODE_ID_CP_MES_DATA;
+		} else {
+			ucode = AMDGPU_UCODE_ID_CP_MES1;
+			ucode_data = AMDGPU_UCODE_ID_CP_MES1_DATA;
+		}
+
+		info = &adev->firmware.ucode[ucode];
+		info->ucode_id = ucode;
+		info->fw = adev->mes.fw[pipe];
+		adev->firmware.fw_size +=
+			ALIGN(le32_to_cpu(mes_hdr->mes_ucode_size_bytes),
+			      PAGE_SIZE);
+
+		info = &adev->firmware.ucode[ucode_data];
+		info->ucode_id = ucode_data;
+		info->fw = adev->mes.fw[pipe];
+		adev->firmware.fw_size +=
+			ALIGN(le32_to_cpu(mes_hdr->mes_ucode_data_size_bytes),
+			      PAGE_SIZE);
+	}
+
+	return 0;
+out:
+	amdgpu_ucode_release(&adev->mes.fw[pipe]);
+	return r;
+}
+
+bool amdgpu_mes_suspend_resume_all_supported(struct amdgpu_device *adev)
+{
+	uint32_t mes_rev = adev->mes.sched_version & AMDGPU_MES_VERSION_MASK;
+	bool is_supported = false;
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(11, 0, 0) &&
+	    amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(12, 0, 0) &&
+	    mes_rev >= 0x63)
+		is_supported = true;
+
+	return is_supported;
+}
+
+/* Fix me -- node_id is used to identify the correct MES instances in the future */
+static int amdgpu_mes_set_enforce_isolation(struct amdgpu_device *adev,
+					    uint32_t node_id, bool enable)
+{
+	struct mes_misc_op_input op_input = {0};
+	int r;
+
+	op_input.op = MES_MISC_OP_CHANGE_CONFIG;
+	op_input.change_config.option.limit_single_process = enable ? 1 : 0;
+
+	if (!adev->mes.funcs->misc_op) {
+		dev_err(adev->dev, "mes change config is not supported!\n");
+		r = -EINVAL;
+		goto error;
+	}
+
+	amdgpu_mes_lock(&adev->mes);
+	r = adev->mes.funcs->misc_op(&adev->mes, &op_input);
+	amdgpu_mes_unlock(&adev->mes);
+	if (r)
+		dev_err(adev->dev, "failed to change_config.\n");
+
+error:
+	return r;
+}
+
+int amdgpu_mes_update_enforce_isolation(struct amdgpu_device *adev)
+{
+	int i, r = 0;
+
+	if (adev->enable_mes && adev->gfx.enable_cleaner_shader) {
+		mutex_lock(&adev->enforce_isolation_mutex);
+		for (i = 0; i < (adev->xcp_mgr ? adev->xcp_mgr->num_xcps : 1); i++) {
+			if (adev->enforce_isolation[i] == AMDGPU_ENFORCE_ISOLATION_ENABLE)
+				r |= amdgpu_mes_set_enforce_isolation(adev, i, true);
+			else
+				r |= amdgpu_mes_set_enforce_isolation(adev, i, false);
+		}
+		mutex_unlock(&adev->enforce_isolation_mutex);
+	}
+	return r;
+}
+
+#if defined(CONFIG_DEBUG_FS)
+
+static int amdgpu_debugfs_mes_event_log_show(struct seq_file *m, void *unused)
+{
+	struct amdgpu_device *adev = m->private;
+	uint32_t *mem = (uint32_t *)(adev->mes.event_log_cpu_addr);
+
+	seq_hex_dump(m, "", DUMP_PREFIX_OFFSET, 32, 4,
+		     mem, adev->mes.event_log_size, false);
+
+	return 0;
+}
+
+DEFINE_SHOW_ATTRIBUTE(amdgpu_debugfs_mes_event_log);
+
+#endif
+
+void amdgpu_debugfs_mes_event_log_init(struct amdgpu_device *adev)
+{
+
+#if defined(CONFIG_DEBUG_FS)
+	struct drm_minor *minor = adev_to_drm(adev)->primary;
+	struct dentry *root = minor->debugfs_root;
+	if (adev->enable_mes && amdgpu_mes_log_enable)
+		debugfs_create_file("amdgpu_mes_event_log", 0444, root,
+				    adev, &amdgpu_debugfs_mes_event_log_fops);
+
+#endif
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h
new file mode 100644
index 000000000000..6b506fc72f58
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h
@@ -0,0 +1,506 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_MES_H__
+#define __AMDGPU_MES_H__
+
+#include "amdgpu_irq.h"
+#include "kgd_kfd_interface.h"
+#include "amdgpu_gfx.h"
+#include "amdgpu_doorbell.h"
+#include <linux/sched/mm.h>
+
+#define AMDGPU_MES_MAX_COMPUTE_PIPES        8
+#define AMDGPU_MES_MAX_GFX_PIPES            2
+#define AMDGPU_MES_MAX_SDMA_PIPES           2
+
+#define AMDGPU_MES_API_VERSION_SHIFT	12
+#define AMDGPU_MES_FEAT_VERSION_SHIFT	24
+
+#define AMDGPU_MES_VERSION_MASK		0x00000fff
+#define AMDGPU_MES_API_VERSION_MASK	0x00fff000
+#define AMDGPU_MES_FEAT_VERSION_MASK	0xff000000
+#define AMDGPU_MES_MSCRATCH_SIZE	0x40000
+#define AMDGPU_MES_INVALID_DB_OFFSET	0xffffffff
+
+enum amdgpu_mes_priority_level {
+	AMDGPU_MES_PRIORITY_LEVEL_LOW       = 0,
+	AMDGPU_MES_PRIORITY_LEVEL_NORMAL    = 1,
+	AMDGPU_MES_PRIORITY_LEVEL_MEDIUM    = 2,
+	AMDGPU_MES_PRIORITY_LEVEL_HIGH      = 3,
+	AMDGPU_MES_PRIORITY_LEVEL_REALTIME  = 4,
+	AMDGPU_MES_PRIORITY_NUM_LEVELS
+};
+
+#define AMDGPU_MES_PROC_CTX_SIZE 0x1000 /* one page area */
+#define AMDGPU_MES_GANG_CTX_SIZE 0x1000 /* one page area */
+
+struct amdgpu_mes_funcs;
+
+enum amdgpu_mes_pipe {
+	AMDGPU_MES_SCHED_PIPE = 0,
+	AMDGPU_MES_KIQ_PIPE,
+	AMDGPU_MAX_MES_PIPES = 2,
+};
+
+struct amdgpu_mes {
+	struct amdgpu_device            *adev;
+
+	struct mutex                    mutex_hidden;
+
+	struct idr                      pasid_idr;
+	struct idr                      gang_id_idr;
+	struct idr                      queue_id_idr;
+	struct ida                      doorbell_ida;
+
+	spinlock_t                      queue_id_lock;
+
+	uint32_t			sched_version;
+	uint32_t			kiq_version;
+	uint32_t			fw_version[AMDGPU_MAX_MES_PIPES];
+	bool                            enable_legacy_queue_map;
+
+	uint32_t                        total_max_queue;
+	uint32_t                        max_doorbell_slices;
+
+	uint64_t                        default_process_quantum;
+	uint64_t                        default_gang_quantum;
+
+	struct amdgpu_ring              ring[AMDGPU_MAX_MES_PIPES];
+	spinlock_t                      ring_lock[AMDGPU_MAX_MES_PIPES];
+
+	const struct firmware           *fw[AMDGPU_MAX_MES_PIPES];
+
+	/* mes ucode */
+	struct amdgpu_bo		*ucode_fw_obj[AMDGPU_MAX_MES_PIPES];
+	uint64_t			ucode_fw_gpu_addr[AMDGPU_MAX_MES_PIPES];
+	uint32_t			*ucode_fw_ptr[AMDGPU_MAX_MES_PIPES];
+	uint64_t                        uc_start_addr[AMDGPU_MAX_MES_PIPES];
+
+	/* mes ucode data */
+	struct amdgpu_bo		*data_fw_obj[AMDGPU_MAX_MES_PIPES];
+	uint64_t			data_fw_gpu_addr[AMDGPU_MAX_MES_PIPES];
+	uint32_t			*data_fw_ptr[AMDGPU_MAX_MES_PIPES];
+	uint64_t                        data_start_addr[AMDGPU_MAX_MES_PIPES];
+
+	/* eop gpu obj */
+	struct amdgpu_bo		*eop_gpu_obj[AMDGPU_MAX_MES_PIPES];
+	uint64_t                        eop_gpu_addr[AMDGPU_MAX_MES_PIPES];
+
+	void                            *mqd_backup[AMDGPU_MAX_MES_PIPES];
+	struct amdgpu_irq_src	        irq[AMDGPU_MAX_MES_PIPES];
+
+	uint32_t                        vmid_mask_gfxhub;
+	uint32_t                        vmid_mask_mmhub;
+	uint32_t                        gfx_hqd_mask[AMDGPU_MES_MAX_GFX_PIPES];
+	uint32_t                        compute_hqd_mask[AMDGPU_MES_MAX_COMPUTE_PIPES];
+	uint32_t                        sdma_hqd_mask[AMDGPU_MES_MAX_SDMA_PIPES];
+	uint32_t                        aggregated_doorbells[AMDGPU_MES_PRIORITY_NUM_LEVELS];
+	uint32_t                        sch_ctx_offs[AMDGPU_MAX_MES_PIPES];
+	uint64_t			sch_ctx_gpu_addr[AMDGPU_MAX_MES_PIPES];
+	uint64_t			*sch_ctx_ptr[AMDGPU_MAX_MES_PIPES];
+	uint32_t			query_status_fence_offs[AMDGPU_MAX_MES_PIPES];
+	uint64_t			query_status_fence_gpu_addr[AMDGPU_MAX_MES_PIPES];
+	uint64_t			*query_status_fence_ptr[AMDGPU_MAX_MES_PIPES];
+
+	uint32_t			saved_flags;
+
+	/* initialize kiq pipe */
+	int                             (*kiq_hw_init)(struct amdgpu_device *adev);
+	int                             (*kiq_hw_fini)(struct amdgpu_device *adev);
+
+	/* MES doorbells */
+	uint32_t			db_start_dw_offset;
+	uint32_t			num_mes_dbs;
+	unsigned long			*doorbell_bitmap;
+
+	/* MES event log buffer */
+	uint32_t			event_log_size;
+	struct amdgpu_bo	*event_log_gpu_obj;
+	uint64_t			event_log_gpu_addr;
+	void				*event_log_cpu_addr;
+
+	/* ip specific functions */
+	const struct amdgpu_mes_funcs   *funcs;
+
+	/* mes resource_1 bo*/
+	struct amdgpu_bo    *resource_1[AMDGPU_MAX_MES_PIPES];
+	uint64_t            resource_1_gpu_addr[AMDGPU_MAX_MES_PIPES];
+	void                *resource_1_addr[AMDGPU_MAX_MES_PIPES];
+
+	int				hung_queue_db_array_size;
+	struct amdgpu_bo		*hung_queue_db_array_gpu_obj;
+	uint64_t			hung_queue_db_array_gpu_addr;
+	void				*hung_queue_db_array_cpu_addr;
+};
+
+struct amdgpu_mes_gang {
+	int 				gang_id;
+	int 				priority;
+	int 				inprocess_gang_priority;
+	int 				global_priority_level;
+	struct list_head 		list;
+	struct amdgpu_mes_process 	*process;
+	struct amdgpu_bo 		*gang_ctx_bo;
+	uint64_t 			gang_ctx_gpu_addr;
+	void 				*gang_ctx_cpu_ptr;
+	uint64_t 			gang_quantum;
+	struct list_head 		queue_list;
+};
+
+struct amdgpu_mes_queue {
+	struct list_head 		list;
+	struct amdgpu_mes_gang 		*gang;
+	int 				queue_id;
+	uint64_t 			doorbell_off;
+	struct amdgpu_bo		*mqd_obj;
+	void				*mqd_cpu_ptr;
+	uint64_t 			mqd_gpu_addr;
+	uint64_t 			wptr_gpu_addr;
+	int 				queue_type;
+	int 				paging;
+	struct amdgpu_ring 		*ring;
+};
+
+struct amdgpu_mes_queue_properties {
+	int 			queue_type;
+	uint64_t                hqd_base_gpu_addr;
+	uint64_t                rptr_gpu_addr;
+	uint64_t                wptr_gpu_addr;
+	uint64_t                wptr_mc_addr;
+	uint32_t                queue_size;
+	uint64_t                eop_gpu_addr;
+	uint32_t                hqd_pipe_priority;
+	uint32_t                hqd_queue_priority;
+	bool 			paging;
+	struct amdgpu_ring 	*ring;
+	/* out */
+	uint64_t       		doorbell_off;
+};
+
+struct amdgpu_mes_gang_properties {
+	uint32_t 	priority;
+	uint32_t 	gang_quantum;
+	uint32_t 	inprocess_gang_priority;
+	uint32_t 	priority_level;
+	int 		global_priority_level;
+};
+
+struct mes_add_queue_input {
+	uint32_t	process_id;
+	uint64_t	page_table_base_addr;
+	uint64_t	process_va_start;
+	uint64_t	process_va_end;
+	uint64_t	process_quantum;
+	uint64_t	process_context_addr;
+	uint64_t	gang_quantum;
+	uint64_t	gang_context_addr;
+	uint32_t	inprocess_gang_priority;
+	uint32_t	gang_global_priority_level;
+	uint32_t	doorbell_offset;
+	uint64_t	mqd_addr;
+	uint64_t	wptr_addr;
+	uint64_t	wptr_mc_addr;
+	uint32_t	queue_type;
+	uint32_t	paging;
+	uint32_t        gws_base;
+	uint32_t        gws_size;
+	uint64_t	tba_addr;
+	uint64_t	tma_addr;
+	uint32_t	trap_en;
+	uint32_t	skip_process_ctx_clear;
+	uint32_t	is_kfd_process;
+	uint32_t	is_aql_queue;
+	uint32_t	queue_size;
+	uint32_t	exclusively_scheduled;
+};
+
+struct mes_remove_queue_input {
+	uint32_t	doorbell_offset;
+	uint64_t	gang_context_addr;
+};
+
+struct mes_map_legacy_queue_input {
+	uint32_t                           queue_type;
+	uint32_t                           doorbell_offset;
+	uint32_t                           pipe_id;
+	uint32_t                           queue_id;
+	uint64_t                           mqd_addr;
+	uint64_t                           wptr_addr;
+};
+
+struct mes_unmap_legacy_queue_input {
+	enum amdgpu_unmap_queues_action    action;
+	uint32_t                           queue_type;
+	uint32_t                           doorbell_offset;
+	uint32_t                           pipe_id;
+	uint32_t                           queue_id;
+	uint64_t                           trail_fence_addr;
+	uint64_t                           trail_fence_data;
+};
+
+struct mes_suspend_gang_input {
+	bool		suspend_all_gangs;
+	uint64_t	gang_context_addr;
+	uint64_t	suspend_fence_addr;
+	uint32_t	suspend_fence_value;
+};
+
+struct mes_resume_gang_input {
+	bool		resume_all_gangs;
+	uint64_t	gang_context_addr;
+};
+
+struct mes_reset_queue_input {
+	uint32_t                           queue_type;
+	uint32_t                           doorbell_offset;
+	bool                               use_mmio;
+	uint32_t                           me_id;
+	uint32_t                           pipe_id;
+	uint32_t                           queue_id;
+	uint64_t                           mqd_addr;
+	uint64_t                           wptr_addr;
+	uint32_t                           vmid;
+	bool                               legacy_gfx;
+	bool                               is_kq;
+};
+
+struct mes_detect_and_reset_queue_input {
+	uint32_t                           queue_type;
+	bool                               detect_only;
+};
+
+struct mes_inv_tlbs_pasid_input {
+	uint32_t        xcc_id;
+	uint16_t        pasid;
+	uint8_t         hub_id;
+	uint8_t         flush_type;
+};
+
+enum mes_misc_opcode {
+	MES_MISC_OP_WRITE_REG,
+	MES_MISC_OP_READ_REG,
+	MES_MISC_OP_WRM_REG_WAIT,
+	MES_MISC_OP_WRM_REG_WR_WAIT,
+	MES_MISC_OP_SET_SHADER_DEBUGGER,
+	MES_MISC_OP_CHANGE_CONFIG,
+};
+
+struct mes_misc_op_input {
+	enum mes_misc_opcode op;
+
+	union {
+		struct {
+			uint32_t                  reg_offset;
+			uint64_t                  buffer_addr;
+		} read_reg;
+
+		struct {
+			uint32_t                  reg_offset;
+			uint32_t                  reg_value;
+		} write_reg;
+
+		struct {
+			uint32_t                   ref;
+			uint32_t                   mask;
+			uint32_t                   reg0;
+			uint32_t                   reg1;
+		} wrm_reg;
+
+		struct {
+			uint64_t process_context_addr;
+			union {
+				struct {
+					uint32_t single_memop : 1;
+					uint32_t single_alu_op : 1;
+					uint32_t reserved: 29;
+					uint32_t process_ctx_flush: 1;
+				};
+				uint32_t u32all;
+			} flags;
+			uint32_t spi_gdbg_per_vmid_cntl;
+			uint32_t tcp_watch_cntl[4];
+			uint32_t trap_en;
+		} set_shader_debugger;
+
+		struct {
+			union {
+				struct {
+					uint32_t limit_single_process : 1;
+					uint32_t enable_hws_logging_buffer : 1;
+					uint32_t reserved : 30;
+				};
+				uint32_t all;
+			} option;
+			struct {
+				uint32_t tdr_level;
+				uint32_t tdr_delay;
+			} tdr_config;
+		} change_config;
+	};
+};
+
+struct amdgpu_mes_funcs {
+	int (*add_hw_queue)(struct amdgpu_mes *mes,
+			    struct mes_add_queue_input *input);
+
+	int (*remove_hw_queue)(struct amdgpu_mes *mes,
+			       struct mes_remove_queue_input *input);
+
+	int (*map_legacy_queue)(struct amdgpu_mes *mes,
+				struct mes_map_legacy_queue_input *input);
+
+	int (*unmap_legacy_queue)(struct amdgpu_mes *mes,
+				  struct mes_unmap_legacy_queue_input *input);
+
+	int (*suspend_gang)(struct amdgpu_mes *mes,
+			    struct mes_suspend_gang_input *input);
+
+	int (*resume_gang)(struct amdgpu_mes *mes,
+			   struct mes_resume_gang_input *input);
+
+	int (*misc_op)(struct amdgpu_mes *mes,
+		       struct mes_misc_op_input *input);
+
+	int (*reset_hw_queue)(struct amdgpu_mes *mes,
+			      struct mes_reset_queue_input *input);
+
+	int (*detect_and_reset_hung_queues)(struct amdgpu_mes *mes,
+			      struct mes_detect_and_reset_queue_input *input);
+
+
+	int (*invalidate_tlbs_pasid)(struct amdgpu_mes *mes,
+			      struct mes_inv_tlbs_pasid_input *input);
+};
+
+#define amdgpu_mes_kiq_hw_init(adev) (adev)->mes.kiq_hw_init((adev))
+#define amdgpu_mes_kiq_hw_fini(adev) (adev)->mes.kiq_hw_fini((adev))
+
+int amdgpu_mes_init_microcode(struct amdgpu_device *adev, int pipe);
+int amdgpu_mes_init(struct amdgpu_device *adev);
+void amdgpu_mes_fini(struct amdgpu_device *adev);
+
+int amdgpu_mes_suspend(struct amdgpu_device *adev);
+int amdgpu_mes_resume(struct amdgpu_device *adev);
+
+int amdgpu_mes_map_legacy_queue(struct amdgpu_device *adev,
+				struct amdgpu_ring *ring);
+int amdgpu_mes_unmap_legacy_queue(struct amdgpu_device *adev,
+				  struct amdgpu_ring *ring,
+				  enum amdgpu_unmap_queues_action action,
+				  u64 gpu_addr, u64 seq);
+int amdgpu_mes_reset_legacy_queue(struct amdgpu_device *adev,
+				  struct amdgpu_ring *ring,
+				  unsigned int vmid,
+				  bool use_mmio);
+
+int amdgpu_mes_get_hung_queue_db_array_size(struct amdgpu_device *adev);
+int amdgpu_mes_detect_and_reset_hung_queues(struct amdgpu_device *adev,
+					    int queue_type,
+					    bool detect_only,
+					    unsigned int *hung_db_num,
+					    u32 *hung_db_array);
+
+uint32_t amdgpu_mes_rreg(struct amdgpu_device *adev, uint32_t reg);
+int amdgpu_mes_wreg(struct amdgpu_device *adev,
+		    uint32_t reg, uint32_t val);
+int amdgpu_mes_reg_write_reg_wait(struct amdgpu_device *adev,
+				  uint32_t reg0, uint32_t reg1,
+				  uint32_t ref, uint32_t mask);
+int amdgpu_mes_set_shader_debugger(struct amdgpu_device *adev,
+				uint64_t process_context_addr,
+				uint32_t spi_gdbg_per_vmid_cntl,
+				const uint32_t *tcp_watch_cntl,
+				uint32_t flags,
+				bool trap_en);
+int amdgpu_mes_flush_shader_debugger(struct amdgpu_device *adev,
+				uint64_t process_context_addr);
+
+uint32_t amdgpu_mes_get_aggregated_doorbell_index(struct amdgpu_device *adev,
+						   enum amdgpu_mes_priority_level prio);
+
+int amdgpu_mes_doorbell_process_slice(struct amdgpu_device *adev);
+
+/*
+ * MES lock can be taken in MMU notifiers.
+ *
+ * A bit more detail about why to set no-FS reclaim with MES lock:
+ *
+ * The purpose of the MMU notifier is to stop GPU access to memory so
+ * that the Linux VM subsystem can move pages around safely. This is
+ * done by preempting user mode queues for the affected process. When
+ * MES is used, MES lock needs to be taken to preempt the queues.
+ *
+ * The MMU notifier callback entry point in the driver is
+ * amdgpu_mn_invalidate_range_start_hsa. The relevant call chain from
+ * there is:
+ * amdgpu_amdkfd_evict_userptr -> kgd2kfd_quiesce_mm ->
+ * kfd_process_evict_queues -> pdd->dev->dqm->ops.evict_process_queues
+ *
+ * The last part of the chain is a function pointer where we take the
+ * MES lock.
+ *
+ * The problem with taking locks in the MMU notifier is, that MMU
+ * notifiers can be called in reclaim-FS context. That's where the
+ * kernel frees up pages to make room for new page allocations under
+ * memory pressure. While we are running in reclaim-FS context, we must
+ * not trigger another memory reclaim operation because that would
+ * recursively reenter the reclaim code and cause a deadlock. The
+ * memalloc_nofs_save/restore calls guarantee that.
+ *
+ * In addition we also need to avoid lock dependencies on other locks taken
+ * under the MES lock, for example reservation locks. Here is a possible
+ * scenario of a deadlock:
+ * Thread A: takes and holds reservation lock | triggers reclaim-FS |
+ * MMU notifier | blocks trying to take MES lock
+ * Thread B: takes and holds MES lock | blocks trying to take reservation lock
+ *
+ * In this scenario Thread B gets involved in a deadlock even without
+ * triggering a reclaim-FS operation itself.
+ * To fix this and break the lock dependency chain you'd need to either:
+ * 1. protect reservation locks with memalloc_nofs_save/restore, or
+ * 2. avoid taking reservation locks under the MES lock.
+ *
+ * Reservation locks are taken all over the kernel in different subsystems, we
+ * have no control over them and their lock dependencies.So the only workable
+ * solution is to avoid taking other locks under the MES lock.
+ * As a result, make sure no reclaim-FS happens while holding this lock anywhere
+ * to prevent deadlocks when an MMU notifier runs in reclaim-FS context.
+ */
+static inline void amdgpu_mes_lock(struct amdgpu_mes *mes)
+{
+	mutex_lock(&mes->mutex_hidden);
+	mes->saved_flags = memalloc_noreclaim_save();
+}
+
+static inline void amdgpu_mes_unlock(struct amdgpu_mes *mes)
+{
+	memalloc_noreclaim_restore(mes->saved_flags);
+	mutex_unlock(&mes->mutex_hidden);
+}
+
+bool amdgpu_mes_suspend_resume_all_supported(struct amdgpu_device *adev);
+
+int amdgpu_mes_update_enforce_isolation(struct amdgpu_device *adev);
+
+#endif /* __AMDGPU_MES_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes_ctx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes_ctx.h
new file mode 100644
index 000000000000..912a5be2ece6
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes_ctx.h
@@ -0,0 +1,122 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_MES_CTX_H__
+#define __AMDGPU_MES_CTX_H__
+
+#include "v10_structs.h"
+
+enum {
+	AMDGPU_MES_CTX_RPTR_OFFS = 0,
+	AMDGPU_MES_CTX_WPTR_OFFS,
+	AMDGPU_MES_CTX_FENCE_OFFS,
+	AMDGPU_MES_CTX_COND_EXE_OFFS,
+	AMDGPU_MES_CTX_TRAIL_FENCE_OFFS,
+	AMDGPU_MES_CTX_MAX_OFFS,
+};
+
+enum {
+	AMDGPU_MES_CTX_RING_OFFS = AMDGPU_MES_CTX_MAX_OFFS,
+	AMDGPU_MES_CTX_IB_OFFS,
+	AMDGPU_MES_CTX_PADDING_OFFS,
+};
+
+#define AMDGPU_MES_CTX_MAX_GFX_RINGS            1
+#define AMDGPU_MES_CTX_MAX_COMPUTE_RINGS        4
+#define AMDGPU_MES_CTX_MAX_SDMA_RINGS           2
+#define AMDGPU_MES_CTX_MAX_RINGS					\
+	(AMDGPU_MES_CTX_MAX_GFX_RINGS +					\
+	 AMDGPU_MES_CTX_MAX_COMPUTE_RINGS +				\
+	 AMDGPU_MES_CTX_MAX_SDMA_RINGS)
+
+#define AMDGPU_CSA_SDMA_SIZE    64
+#define GFX10_MEC_HPD_SIZE	2048
+
+struct amdgpu_wb_slot {
+	uint32_t data[8];
+};
+
+struct amdgpu_mes_ctx_meta_data {
+	struct {
+		uint8_t ring[PAGE_SIZE * 4];
+
+		/* gfx csa */
+		struct v10_gfx_meta_data gfx_meta_data;
+
+		uint8_t gds_backup[64 * 1024];
+
+		struct amdgpu_wb_slot slots[AMDGPU_MES_CTX_MAX_OFFS];
+
+		/* only for ib test */
+		uint32_t ib[256] __aligned(256);
+
+		uint32_t padding[64];
+
+	} __aligned(PAGE_SIZE) gfx[AMDGPU_MES_CTX_MAX_GFX_RINGS];
+
+	struct {
+		uint8_t ring[PAGE_SIZE * 4];
+
+		uint8_t mec_hpd[GFX10_MEC_HPD_SIZE];
+
+		struct amdgpu_wb_slot slots[AMDGPU_MES_CTX_MAX_OFFS];
+
+		/* only for ib test */
+		uint32_t ib[256] __aligned(256);
+
+		uint32_t padding[64];
+
+	} __aligned(PAGE_SIZE) compute[AMDGPU_MES_CTX_MAX_COMPUTE_RINGS];
+
+	struct {
+		uint8_t ring[PAGE_SIZE * 4];
+
+		/* sdma csa for mcbp */
+		uint8_t sdma_meta_data[AMDGPU_CSA_SDMA_SIZE];
+
+		struct amdgpu_wb_slot slots[AMDGPU_MES_CTX_MAX_OFFS];
+
+		/* only for ib test */
+		uint32_t ib[256] __aligned(256);
+
+		uint32_t padding[64];
+
+	} __aligned(PAGE_SIZE) sdma[AMDGPU_MES_CTX_MAX_SDMA_RINGS];
+};
+
+struct amdgpu_mes_ctx_data {
+	struct amdgpu_bo	*meta_data_obj;
+	uint64_t                meta_data_gpu_addr;
+	uint64_t                meta_data_mc_addr;
+	struct amdgpu_bo_va	*meta_data_va;
+	void                    *meta_data_ptr;
+	uint32_t                gang_ids[AMDGPU_HW_IP_DMA+1];
+};
+
+#define AMDGPU_FENCE_MES_QUEUE_FLAG     0x1000000u
+#define AMDGPU_FENCE_MES_QUEUE_ID_MASK  (AMDGPU_FENCE_MES_QUEUE_FLAG - 1)
+
+#define AMDGPU_FENCE_MES_QUEUE_FLAG     0x1000000u
+#define AMDGPU_FENCE_MES_QUEUE_ID_MASK  (AMDGPU_FENCE_MES_QUEUE_FLAG - 1)
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.c
new file mode 100644
index 000000000000..0f6b1021fef3
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.c
@@ -0,0 +1,46 @@
+/*
+ * Copyright (C) 2023  Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+#include "amdgpu.h"
+#include "amdgpu_ras.h"
+
+int amdgpu_mmhub_ras_sw_init(struct amdgpu_device *adev)
+{
+	int err;
+	struct amdgpu_mmhub_ras *ras;
+
+	if (!adev->mmhub.ras)
+		return 0;
+
+	ras = adev->mmhub.ras;
+	err = amdgpu_ras_register_ras_block(adev, &ras->ras_block);
+	if (err) {
+		dev_err(adev->dev, "Failed to register mmhub ras block!\n");
+		return err;
+	}
+
+	strcpy(ras->ras_block.ras_comm.name, "mmhub");
+	ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__MMHUB;
+	ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
+	adev->mmhub.ras_if = &ras->ras_block.ras_comm;
+
+	/* mmhub ras follows amdgpu_ras_block_late_init_default for late init */
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h
new file mode 100644
index 000000000000..1ca9d4ed8063
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h
@@ -0,0 +1,77 @@
+/*
+ * Copyright (C) 2019  Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+#ifndef __AMDGPU_MMHUB_H__
+#define __AMDGPU_MMHUB_H__
+
+enum amdgpu_mmhub_ras_memory_id {
+	AMDGPU_MMHUB_WGMI_PAGEMEM = 0,
+	AMDGPU_MMHUB_RGMI_PAGEMEM = 1,
+	AMDGPU_MMHUB_WDRAM_PAGEMEM = 2,
+	AMDGPU_MMHUB_RDRAM_PAGEMEM = 3,
+	AMDGPU_MMHUB_WIO_CMDMEM = 4,
+	AMDGPU_MMHUB_RIO_CMDMEM = 5,
+	AMDGPU_MMHUB_WGMI_CMDMEM = 6,
+	AMDGPU_MMHUB_RGMI_CMDMEM = 7,
+	AMDGPU_MMHUB_WDRAM_CMDMEM = 8,
+	AMDGPU_MMHUB_RDRAM_CMDMEM = 9,
+	AMDGPU_MMHUB_MAM_DMEM0 = 10,
+	AMDGPU_MMHUB_MAM_DMEM1 = 11,
+	AMDGPU_MMHUB_MAM_DMEM2 = 12,
+	AMDGPU_MMHUB_MAM_DMEM3 = 13,
+	AMDGPU_MMHUB_WRET_TAGMEM = 19,
+	AMDGPU_MMHUB_RRET_TAGMEM = 20,
+	AMDGPU_MMHUB_WIO_DATAMEM = 21,
+	AMDGPU_MMHUB_WGMI_DATAMEM = 22,
+	AMDGPU_MMHUB_WDRAM_DATAMEM = 23,
+	AMDGPU_MMHUB_MEMORY_BLOCK_LAST,
+};
+
+struct amdgpu_mmhub_ras {
+	struct amdgpu_ras_block_object ras_block;
+};
+
+struct amdgpu_mmhub_funcs {
+	u64 (*get_fb_location)(struct amdgpu_device *adev);
+	u64 (*get_mc_fb_offset)(struct amdgpu_device *adev);
+	void (*init)(struct amdgpu_device *adev);
+	int (*gart_enable)(struct amdgpu_device *adev);
+	void (*set_fault_enable_default)(struct amdgpu_device *adev,
+			bool value);
+	void (*gart_disable)(struct amdgpu_device *adev);
+	int (*set_clockgating)(struct amdgpu_device *adev,
+			       enum amd_clockgating_state state);
+	void (*get_clockgating)(struct amdgpu_device *adev, u64 *flags);
+	void (*setup_vm_pt_regs)(struct amdgpu_device *adev, uint32_t vmid,
+				uint64_t page_table_base);
+	void (*update_power_gating)(struct amdgpu_device *adev,
+                                bool enable);
+};
+
+struct amdgpu_mmhub {
+	struct ras_common_if *ras_if;
+	const struct amdgpu_mmhub_funcs *funcs;
+	struct amdgpu_mmhub_ras  *ras;
+};
+
+int amdgpu_mmhub_ras_sw_init(struct amdgpu_device *adev);
+
+#endif
+
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
deleted file mode 100644
index 3b0f2ec6eec7..000000000000
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
+++ /dev/null
@@ -1,342 +0,0 @@
-/*
- * Copyright 2014 Advanced Micro Devices, Inc.
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
- * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
- * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
- * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
- * USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- */
-/*
- * Authors:
- *    Christian König <christian.koenig@amd.com>
- */
-
-#include <linux/firmware.h>
-#include <linux/module.h>
-#include <linux/mmu_notifier.h>
-#include <linux/interval_tree.h>
-#include <drm/drmP.h>
-#include <drm/drm.h>
-
-#include "amdgpu.h"
-
-struct amdgpu_mn {
-	/* constant after initialisation */
-	struct amdgpu_device	*adev;
-	struct mm_struct	*mm;
-	struct mmu_notifier	mn;
-
-	/* only used on destruction */
-	struct work_struct	work;
-
-	/* protected by adev->mn_lock */
-	struct hlist_node	node;
-
-	/* objects protected by lock */
-	struct mutex		lock;
-	struct rb_root_cached	objects;
-};
-
-struct amdgpu_mn_node {
-	struct interval_tree_node	it;
-	struct list_head		bos;
-};
-
-/**
- * amdgpu_mn_destroy - destroy the rmn
- *
- * @work: previously sheduled work item
- *
- * Lazy destroys the notifier from a work item
- */
-static void amdgpu_mn_destroy(struct work_struct *work)
-{
-	struct amdgpu_mn *rmn = container_of(work, struct amdgpu_mn, work);
-	struct amdgpu_device *adev = rmn->adev;
-	struct amdgpu_mn_node *node, *next_node;
-	struct amdgpu_bo *bo, *next_bo;
-
-	mutex_lock(&adev->mn_lock);
-	mutex_lock(&rmn->lock);
-	hash_del(&rmn->node);
-	rbtree_postorder_for_each_entry_safe(node, next_node,
-					     &rmn->objects.rb_root, it.rb) {
-		list_for_each_entry_safe(bo, next_bo, &node->bos, mn_list) {
-			bo->mn = NULL;
-			list_del_init(&bo->mn_list);
-		}
-		kfree(node);
-	}
-	mutex_unlock(&rmn->lock);
-	mutex_unlock(&adev->mn_lock);
-	mmu_notifier_unregister_no_release(&rmn->mn, rmn->mm);
-	kfree(rmn);
-}
-
-/**
- * amdgpu_mn_release - callback to notify about mm destruction
- *
- * @mn: our notifier
- * @mn: the mm this callback is about
- *
- * Shedule a work item to lazy destroy our notifier.
- */
-static void amdgpu_mn_release(struct mmu_notifier *mn,
-			      struct mm_struct *mm)
-{
-	struct amdgpu_mn *rmn = container_of(mn, struct amdgpu_mn, mn);
-	INIT_WORK(&rmn->work, amdgpu_mn_destroy);
-	schedule_work(&rmn->work);
-}
-
-/**
- * amdgpu_mn_invalidate_node - unmap all BOs of a node
- *
- * @node: the node with the BOs to unmap
- *
- * We block for all BOs and unmap them by move them
- * into system domain again.
- */
-static void amdgpu_mn_invalidate_node(struct amdgpu_mn_node *node,
-				      unsigned long start,
-				      unsigned long end)
-{
-	struct amdgpu_bo *bo;
-	long r;
-
-	list_for_each_entry(bo, &node->bos, mn_list) {
-
-		if (!amdgpu_ttm_tt_affect_userptr(bo->tbo.ttm, start, end))
-			continue;
-
-		r = amdgpu_bo_reserve(bo, true);
-		if (r) {
-			DRM_ERROR("(%ld) failed to reserve user bo\n", r);
-			continue;
-		}
-
-		r = reservation_object_wait_timeout_rcu(bo->tbo.resv,
-			true, false, MAX_SCHEDULE_TIMEOUT);
-		if (r <= 0)
-			DRM_ERROR("(%ld) failed to wait for user bo\n", r);
-
-		amdgpu_ttm_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_CPU);
-		r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false);
-		if (r)
-			DRM_ERROR("(%ld) failed to validate user bo\n", r);
-
-		amdgpu_bo_unreserve(bo);
-	}
-}
-
-/**
- * amdgpu_mn_invalidate_range_start - callback to notify about mm change
- *
- * @mn: our notifier
- * @mn: the mm this callback is about
- * @start: start of updated range
- * @end: end of updated range
- *
- * We block for all BOs between start and end to be idle and
- * unmap them by move them into system domain again.
- */
-static void amdgpu_mn_invalidate_range_start(struct mmu_notifier *mn,
-					     struct mm_struct *mm,
-					     unsigned long start,
-					     unsigned long end)
-{
-	struct amdgpu_mn *rmn = container_of(mn, struct amdgpu_mn, mn);
-	struct interval_tree_node *it;
-
-	/* notification is exclusive, but interval is inclusive */
-	end -= 1;
-
-	mutex_lock(&rmn->lock);
-
-	it = interval_tree_iter_first(&rmn->objects, start, end);
-	while (it) {
-		struct amdgpu_mn_node *node;
-
-		node = container_of(it, struct amdgpu_mn_node, it);
-		it = interval_tree_iter_next(it, start, end);
-
-		amdgpu_mn_invalidate_node(node, start, end);
-	}
-
-	mutex_unlock(&rmn->lock);
-}
-
-static const struct mmu_notifier_ops amdgpu_mn_ops = {
-	.release = amdgpu_mn_release,
-	.invalidate_range_start = amdgpu_mn_invalidate_range_start,
-};
-
-/**
- * amdgpu_mn_get - create notifier context
- *
- * @adev: amdgpu device pointer
- *
- * Creates a notifier context for current->mm.
- */
-static struct amdgpu_mn *amdgpu_mn_get(struct amdgpu_device *adev)
-{
-	struct mm_struct *mm = current->mm;
-	struct amdgpu_mn *rmn;
-	int r;
-
-	mutex_lock(&adev->mn_lock);
-	if (down_write_killable(&mm->mmap_sem)) {
-		mutex_unlock(&adev->mn_lock);
-		return ERR_PTR(-EINTR);
-	}
-
-	hash_for_each_possible(adev->mn_hash, rmn, node, (unsigned long)mm)
-		if (rmn->mm == mm)
-			goto release_locks;
-
-	rmn = kzalloc(sizeof(*rmn), GFP_KERNEL);
-	if (!rmn) {
-		rmn = ERR_PTR(-ENOMEM);
-		goto release_locks;
-	}
-
-	rmn->adev = adev;
-	rmn->mm = mm;
-	rmn->mn.ops = &amdgpu_mn_ops;
-	mutex_init(&rmn->lock);
-	rmn->objects = RB_ROOT_CACHED;
-
-	r = __mmu_notifier_register(&rmn->mn, mm);
-	if (r)
-		goto free_rmn;
-
-	hash_add(adev->mn_hash, &rmn->node, (unsigned long)mm);
-
-release_locks:
-	up_write(&mm->mmap_sem);
-	mutex_unlock(&adev->mn_lock);
-
-	return rmn;
-
-free_rmn:
-	up_write(&mm->mmap_sem);
-	mutex_unlock(&adev->mn_lock);
-	kfree(rmn);
-
-	return ERR_PTR(r);
-}
-
-/**
- * amdgpu_mn_register - register a BO for notifier updates
- *
- * @bo: amdgpu buffer object
- * @addr: userptr addr we should monitor
- *
- * Registers an MMU notifier for the given BO at the specified address.
- * Returns 0 on success, -ERRNO if anything goes wrong.
- */
-int amdgpu_mn_register(struct amdgpu_bo *bo, unsigned long addr)
-{
-	unsigned long end = addr + amdgpu_bo_size(bo) - 1;
-	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
-	struct amdgpu_mn *rmn;
-	struct amdgpu_mn_node *node = NULL;
-	struct list_head bos;
-	struct interval_tree_node *it;
-
-	rmn = amdgpu_mn_get(adev);
-	if (IS_ERR(rmn))
-		return PTR_ERR(rmn);
-
-	INIT_LIST_HEAD(&bos);
-
-	mutex_lock(&rmn->lock);
-
-	while ((it = interval_tree_iter_first(&rmn->objects, addr, end))) {
-		kfree(node);
-		node = container_of(it, struct amdgpu_mn_node, it);
-		interval_tree_remove(&node->it, &rmn->objects);
-		addr = min(it->start, addr);
-		end = max(it->last, end);
-		list_splice(&node->bos, &bos);
-	}
-
-	if (!node) {
-		node = kmalloc(sizeof(struct amdgpu_mn_node), GFP_KERNEL);
-		if (!node) {
-			mutex_unlock(&rmn->lock);
-			return -ENOMEM;
-		}
-	}
-
-	bo->mn = rmn;
-
-	node->it.start = addr;
-	node->it.last = end;
-	INIT_LIST_HEAD(&node->bos);
-	list_splice(&bos, &node->bos);
-	list_add(&bo->mn_list, &node->bos);
-
-	interval_tree_insert(&node->it, &rmn->objects);
-
-	mutex_unlock(&rmn->lock);
-
-	return 0;
-}
-
-/**
- * amdgpu_mn_unregister - unregister a BO for notifier updates
- *
- * @bo: amdgpu buffer object
- *
- * Remove any registration of MMU notifier updates from the buffer object.
- */
-void amdgpu_mn_unregister(struct amdgpu_bo *bo)
-{
-	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
-	struct amdgpu_mn *rmn;
-	struct list_head *head;
-
-	mutex_lock(&adev->mn_lock);
-
-	rmn = bo->mn;
-	if (rmn == NULL) {
-		mutex_unlock(&adev->mn_lock);
-		return;
-	}
-
-	mutex_lock(&rmn->lock);
-
-	/* save the next list entry for later */
-	head = bo->mn_list.next;
-
-	bo->mn = NULL;
-	list_del_init(&bo->mn_list);
-
-	if (list_empty(head)) {
-		struct amdgpu_mn_node *node;
-		node = container_of(head, struct amdgpu_mn_node, bos);
-		interval_tree_remove(&node->it, &rmn->objects);
-		kfree(node);
-	}
-
-	mutex_unlock(&rmn->lock);
-	mutex_unlock(&adev->mn_lock);
-}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h
index 2af2678ddaf6..20460cfd09bc 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h
@@ -30,32 +30,39 @@
 #ifndef AMDGPU_MODE_H
 #define AMDGPU_MODE_H
 
+#include <drm/display/drm_dp_helper.h>
 #include <drm/drm_crtc.h>
-#include <drm/drm_edid.h>
 #include <drm/drm_encoder.h>
-#include <drm/drm_dp_helper.h>
 #include <drm/drm_fixed.h>
-#include <drm/drm_crtc_helper.h>
-#include <drm/drm_fb_helper.h>
-#include <drm/drm_plane_helper.h>
+#include <drm/drm_framebuffer.h>
+#include <drm/drm_probe_helper.h>
 #include <linux/i2c.h>
 #include <linux/i2c-algo-bit.h>
 #include <linux/hrtimer.h>
 #include "amdgpu_irq.h"
 
+#include <drm/display/drm_dp_mst_helper.h>
+#include "modules/inc/mod_freesync.h"
+#include "amdgpu_dm_irq_params.h"
+
 struct amdgpu_bo;
 struct amdgpu_device;
 struct amdgpu_encoder;
 struct amdgpu_router;
 struct amdgpu_hpd;
+struct edid;
+struct drm_edid;
 
 #define to_amdgpu_crtc(x) container_of(x, struct amdgpu_crtc, base)
 #define to_amdgpu_connector(x) container_of(x, struct amdgpu_connector, base)
 #define to_amdgpu_encoder(x) container_of(x, struct amdgpu_encoder, base)
 #define to_amdgpu_framebuffer(x) container_of(x, struct amdgpu_framebuffer, base)
 
+#define to_dm_plane_state(x)	container_of(x, struct dm_plane_state, base)
+
 #define AMDGPU_MAX_HPD_PINS 6
 #define AMDGPU_MAX_CRTCS 6
+#define AMDGPU_MAX_PLANES 6
 #define AMDGPU_MAX_AFMT_BLOCKS 9
 
 enum amdgpu_rmx_type {
@@ -81,7 +88,6 @@ enum amdgpu_hpd_id {
 	AMDGPU_HPD_4,
 	AMDGPU_HPD_5,
 	AMDGPU_HPD_6,
-	AMDGPU_HPD_LAST,
 	AMDGPU_HPD_NONE = 0xff,
 };
 
@@ -98,7 +104,6 @@ enum amdgpu_crtc_irq {
 	AMDGPU_CRTC_IRQ_VLINE4,
 	AMDGPU_CRTC_IRQ_VLINE5,
 	AMDGPU_CRTC_IRQ_VLINE6,
-	AMDGPU_CRTC_IRQ_LAST,
 	AMDGPU_CRTC_IRQ_NONE = 0xff
 };
 
@@ -109,7 +114,6 @@ enum amdgpu_pageflip_irq {
 	AMDGPU_PAGEFLIP_IRQ_D4,
 	AMDGPU_PAGEFLIP_IRQ_D5,
 	AMDGPU_PAGEFLIP_IRQ_D6,
-	AMDGPU_PAGEFLIP_IRQ_LAST,
 	AMDGPU_PAGEFLIP_IRQ_NONE = 0xff
 };
 
@@ -227,8 +231,6 @@ struct amdgpu_i2c_chan {
 	struct mutex mutex;
 };
 
-struct amdgpu_fbdev;
-
 struct amdgpu_afmt {
 	bool enabled;
 	int offset;
@@ -262,8 +264,6 @@ struct amdgpu_display_funcs {
 	void (*bandwidth_update)(struct amdgpu_device *adev);
 	/* get frame count */
 	u32 (*vblank_get_counter)(struct amdgpu_device *adev, int crtc);
-	/* wait for vblank */
-	void (*vblank_wait)(struct amdgpu_device *adev, int crtc);
 	/* set backlight level */
 	void (*backlight_set_level)(struct amdgpu_encoder *amdgpu_encoder,
 				    u8 level);
@@ -292,6 +292,19 @@ struct amdgpu_display_funcs {
 			      uint16_t connector_object_id,
 			      struct amdgpu_hpd *hpd,
 			      struct amdgpu_router *router);
+
+
+};
+
+struct amdgpu_framebuffer {
+	struct drm_framebuffer base;
+
+	uint64_t tiling_flags;
+	bool tmz_surface;
+	bool gfx12_dcc;
+
+	/* caching for later use */
+	uint64_t address;
 };
 
 struct amdgpu_mode_info {
@@ -299,6 +312,7 @@ struct amdgpu_mode_info {
 	struct card_info *atom_card_info;
 	bool mode_config_initialized;
 	struct amdgpu_crtc *crtcs[AMDGPU_MAX_CRTCS];
+	struct drm_plane *planes[AMDGPU_MAX_PLANES];
 	struct amdgpu_afmt *afmt[AMDGPU_MAX_AFMT_BLOCKS];
 	/* DVI-I properties */
 	struct drm_property *coherent_mode_property;
@@ -313,34 +327,121 @@ struct amdgpu_mode_info {
 	/* FMT dithering */
 	struct drm_property *dither_property;
 	/* hardcoded DFP edid from BIOS */
-	struct edid *bios_hardcoded_edid;
-	int bios_hardcoded_edid_size;
+	const struct drm_edid *bios_hardcoded_edid;
 
-	/* pointer to fbdev info structure */
-	struct amdgpu_fbdev *rfbdev;
 	/* firmware flags */
-	u16 firmware_flags;
+	u32 firmware_flags;
 	/* pointer to backlight encoder */
 	struct amdgpu_encoder *bl_encoder;
+	u8 bl_level; /* saved backlight level */
 	struct amdgpu_audio	audio; /* audio stuff */
 	int			num_crtc; /* number of crtcs */
 	int			num_hpd; /* number of hpd pins */
 	int			num_dig; /* number of dig blocks */
+	bool			gpu_vm_support; /* supports display from GTT */
 	int			disp_priority;
 	const struct amdgpu_display_funcs *funcs;
+	const enum drm_plane_type *plane_type;
+
+	/* Driver-private color mgmt props */
+
+	/* @plane_degamma_lut_property: Plane property to set a degamma LUT to
+	 * convert encoded values to light linear values before sampling or
+	 * blending.
+	 */
+	struct drm_property *plane_degamma_lut_property;
+	/* @plane_degamma_lut_size_property: Plane property to define the max
+	 * size of degamma LUT as supported by the driver (read-only).
+	 */
+	struct drm_property *plane_degamma_lut_size_property;
+	/**
+	 * @plane_degamma_tf_property: Plane pre-defined transfer function to
+	 * to go from scanout/encoded values to linear values.
+	 */
+	struct drm_property *plane_degamma_tf_property;
+	/**
+	 * @plane_hdr_mult_property:
+	 */
+	struct drm_property *plane_hdr_mult_property;
+
+	struct drm_property *plane_ctm_property;
+	/**
+	 * @shaper_lut_property: Plane property to set pre-blending shaper LUT
+	 * that converts color content before 3D LUT. If
+	 * plane_shaper_tf_property != Identity TF, AMD color module will
+	 * combine the user LUT values with pre-defined TF into the LUT
+	 * parameters to be programmed.
+	 */
+	struct drm_property *plane_shaper_lut_property;
+	/**
+	 * @shaper_lut_size_property: Plane property for the size of
+	 * pre-blending shaper LUT as supported by the driver (read-only).
+	 */
+	struct drm_property *plane_shaper_lut_size_property;
+	/**
+	 * @plane_shaper_tf_property: Plane property to set a predefined
+	 * transfer function for pre-blending shaper (before applying 3D LUT)
+	 * with or without LUT. There is no shaper ROM, but we can use AMD
+	 * color modules to program LUT parameters from predefined TF (or
+	 * from a combination of pre-defined TF and the custom 1D LUT).
+	 */
+	struct drm_property *plane_shaper_tf_property;
+	/**
+	 * @plane_lut3d_property: Plane property for color transformation using
+	 * a 3D LUT (pre-blending), a three-dimensional array where each
+	 * element is an RGB triplet. Each dimension has the size of
+	 * lut3d_size. The array contains samples from the approximated
+	 * function. On AMD, values between samples are estimated by
+	 * tetrahedral interpolation. The array is accessed with three indices,
+	 * one for each input dimension (color channel), blue being the
+	 * outermost dimension, red the innermost.
+	 */
+	struct drm_property *plane_lut3d_property;
+	/**
+	 * @plane_degamma_lut_size_property: Plane property to define the max
+	 * size of 3D LUT as supported by the driver (read-only). The max size
+	 * is the max size of one dimension and, therefore, the max number of
+	 * entries for 3D LUT array is the 3D LUT size cubed;
+	 */
+	struct drm_property *plane_lut3d_size_property;
+	/**
+	 * @plane_blend_lut_property: Plane property for output gamma before
+	 * blending. Userspace set a blend LUT to convert colors after 3D LUT
+	 * conversion. It works as a post-3DLUT 1D LUT. With shaper LUT, they
+	 * are sandwiching 3D LUT with two 1D LUT. If plane_blend_tf_property
+	 * != Identity TF, AMD color module will combine the user LUT values
+	 * with pre-defined TF into the LUT parameters to be programmed.
+	 */
+	struct drm_property *plane_blend_lut_property;
+	/**
+	 * @plane_blend_lut_size_property: Plane property to define the max
+	 * size of blend LUT as supported by the driver (read-only).
+	 */
+	struct drm_property *plane_blend_lut_size_property;
+	/**
+	 * @plane_blend_tf_property: Plane property to set a predefined
+	 * transfer function for pre-blending blend/out_gamma (after applying
+	 * 3D LUT) with or without LUT. There is no blend ROM, but we can use
+	 * AMD color modules to program LUT parameters from predefined TF (or
+	 * from a combination of pre-defined TF and the custom 1D LUT).
+	 */
+	struct drm_property *plane_blend_tf_property;
+	/* @regamma_tf_property: Transfer function for CRTC regamma
+	 * (post-blending). Possible values are defined by `enum
+	 * amdgpu_transfer_function`. There is no regamma ROM, but we can use
+	 * AMD color modules to program LUT parameters from predefined TF (or
+	 * from a combination of pre-defined TF and the custom 1D LUT).
+	 */
+	struct drm_property *regamma_tf_property;
 };
 
 #define AMDGPU_MAX_BL_LEVEL 0xFF
 
-#if defined(CONFIG_BACKLIGHT_CLASS_DEVICE) || defined(CONFIG_BACKLIGHT_CLASS_DEVICE_MODULE)
-
 struct amdgpu_backlight_privdata {
 	struct amdgpu_encoder *encoder;
 	uint8_t negative;
 };
 
-#endif
-
 struct amdgpu_atom_ss {
 	uint16_t percentage;
 	uint16_t percentage_divider;
@@ -381,6 +482,8 @@ struct amdgpu_crtc {
 	struct amdgpu_flip_work *pflip_works;
 	enum amdgpu_flip_status pflip_status;
 	int deferred_flip_completion;
+	/* parameters access from DM IRQ handler */
+	struct dm_irq_params dm_irq_params;
 	/* pll sharing */
 	struct amdgpu_atom_ss ss;
 	bool ss_enabled;
@@ -393,13 +496,18 @@ struct amdgpu_crtc {
 	struct drm_connector *connector;
 	/* for dpm */
 	u32 line_time;
-	u32 wm_low;
-	u32 wm_high;
 	u32 lb_vblank_lead_lines;
 	struct drm_display_mode hw_mode;
 	/* for virtual dce */
 	struct hrtimer vblank_timer;
 	enum amdgpu_interrupt_state vsync_timer_enabled;
+
+	int otg_inst;
+	struct drm_pending_vblank_event *event;
+
+	bool wb_pending;
+	bool wb_enabled;
+	struct drm_writeback_connector *wb_conn;
 };
 
 struct amdgpu_encoder_atom_dig {
@@ -442,6 +550,7 @@ struct amdgpu_encoder {
 struct amdgpu_connector_atom_dig {
 	/* displayport */
 	u8 dpcd[DP_RECEIVER_CAP_SIZE];
+	u8 downstream_ports[DP_MAX_DOWNSTREAM_PORTS];
 	u8 dp_sink_type;
 	int dp_clock;
 	int dp_lane_count;
@@ -489,6 +598,20 @@ enum amdgpu_connector_dither {
 	AMDGPU_FMT_DITHER_ENABLE = 1,
 };
 
+struct amdgpu_dm_dp_aux {
+	struct drm_dp_aux aux;
+	struct ddc_service *ddc_service;
+};
+
+struct amdgpu_i2c_adapter {
+	struct i2c_adapter base;
+
+	struct ddc_service *ddc_service;
+	bool oem;
+};
+
+#define TO_DM_AUX(x) container_of((x), struct amdgpu_dm_dp_aux, aux)
+
 struct amdgpu_connector {
 	struct drm_connector base;
 	uint32_t connector_id;
@@ -503,6 +626,7 @@ struct amdgpu_connector {
 	void *con_priv;
 	bool dac_load_detect;
 	bool detected_by_load; /* if the connection status was determined by load */
+	bool detected_hpd_without_ddc; /* if an HPD signal was detected on DVI, but ddc probing failed */
 	uint16_t connector_object_id;
 	struct amdgpu_hpd hpd;
 	struct amdgpu_router router;
@@ -512,15 +636,22 @@ struct amdgpu_connector {
 	unsigned pixelclock_for_modeset;
 };
 
-struct amdgpu_framebuffer {
-	struct drm_framebuffer base;
-	struct drm_gem_object *obj;
+/* TODO: start to use this struct and remove same field from base one */
+struct amdgpu_mst_connector {
+	struct amdgpu_connector base;
+
+	struct drm_dp_mst_topology_mgr mst_mgr;
+	struct amdgpu_dm_dp_aux dm_dp_aux;
+	struct drm_dp_mst_port *mst_output_port;
+	struct amdgpu_connector *mst_root;
+	bool is_mst_connector;
+	struct amdgpu_encoder *mst_encoder;
 };
 
 #define ENCODER_MODE_IS_DP(em) (((em) == ATOM_ENCODER_MODE_DP) || \
 				((em) == ATOM_ENCODER_MODE_DP_MST))
 
-/* Driver internal use only flags of amdgpu_get_crtc_scanoutpos() */
+/* Driver internal use only flags of amdgpu_display_get_crtc_scanoutpos() */
 #define DRM_SCANOUTPOS_VALID        (1 << 0)
 #define DRM_SCANOUTPOS_IN_VBLANK    (1 << 1)
 #define DRM_SCANOUTPOS_ACCURATE     (1 << 2)
@@ -539,54 +670,42 @@ bool amdgpu_dig_monitor_is_duallink(struct drm_encoder *encoder,
 u16 amdgpu_encoder_get_dp_bridge_encoder_id(struct drm_encoder *encoder);
 struct drm_encoder *amdgpu_get_external_encoder(struct drm_encoder *encoder);
 
-bool amdgpu_ddc_probe(struct amdgpu_connector *amdgpu_connector, bool use_aux);
+bool amdgpu_display_ddc_probe(struct amdgpu_connector *amdgpu_connector,
+			      bool use_aux);
 
 void amdgpu_encoder_set_active_device(struct drm_encoder *encoder);
 
-int amdgpu_get_crtc_scanoutpos(struct drm_device *dev, unsigned int pipe,
-			       unsigned int flags, int *vpos, int *hpos,
-			       ktime_t *stime, ktime_t *etime,
-			       const struct drm_display_mode *mode);
-
-int amdgpu_framebuffer_init(struct drm_device *dev,
-			     struct amdgpu_framebuffer *rfb,
-			     const struct drm_mode_fb_cmd2 *mode_cmd,
-			     struct drm_gem_object *obj);
+int amdgpu_display_get_crtc_scanoutpos(struct drm_device *dev,
+			unsigned int pipe, unsigned int flags, int *vpos,
+			int *hpos, ktime_t *stime, ktime_t *etime,
+			const struct drm_display_mode *mode);
 
 int amdgpufb_remove(struct drm_device *dev, struct drm_framebuffer *fb);
 
 void amdgpu_enc_destroy(struct drm_encoder *encoder);
 void amdgpu_copy_fb(struct drm_device *dev, struct drm_gem_object *dst_obj);
-bool amdgpu_crtc_scaling_mode_fixup(struct drm_crtc *crtc,
-					const struct drm_display_mode *mode,
-					struct drm_display_mode *adjusted_mode);
+bool amdgpu_display_crtc_scaling_mode_fixup(struct drm_crtc *crtc,
+				const struct drm_display_mode *mode,
+				struct drm_display_mode *adjusted_mode);
 void amdgpu_panel_mode_fixup(struct drm_encoder *encoder,
 			     struct drm_display_mode *adjusted_mode);
-int amdgpu_crtc_idx_to_irq_type(struct amdgpu_device *adev, int crtc);
-
-/* fbdev layer */
-int amdgpu_fbdev_init(struct amdgpu_device *adev);
-void amdgpu_fbdev_fini(struct amdgpu_device *adev);
-void amdgpu_fbdev_set_suspend(struct amdgpu_device *adev, int state);
-int amdgpu_fbdev_total_size(struct amdgpu_device *adev);
-bool amdgpu_fbdev_robj_is_fb(struct amdgpu_device *adev, struct amdgpu_bo *robj);
-void amdgpu_fbdev_restore_mode(struct amdgpu_device *adev);
-
-void amdgpu_fb_output_poll_changed(struct amdgpu_device *adev);
-
+int amdgpu_display_crtc_idx_to_irq_type(struct amdgpu_device *adev, int crtc);
 
-int amdgpu_align_pitch(struct amdgpu_device *adev, int width, int bpp, bool tiled);
+bool amdgpu_crtc_get_scanout_position(struct drm_crtc *crtc,
+			bool in_vblank_irq, int *vpos,
+			int *hpos, ktime_t *stime, ktime_t *etime,
+			const struct drm_display_mode *mode);
 
 /* amdgpu_display.c */
-void amdgpu_print_display_setup(struct drm_device *dev);
-int amdgpu_modeset_create_props(struct amdgpu_device *adev);
-int amdgpu_crtc_set_config(struct drm_mode_set *set,
-			   struct drm_modeset_acquire_ctx *ctx);
-int amdgpu_crtc_page_flip_target(struct drm_crtc *crtc,
-				 struct drm_framebuffer *fb,
-				 struct drm_pending_vblank_event *event,
-				 uint32_t page_flip_flags, uint32_t target,
-				 struct drm_modeset_acquire_ctx *ctx);
+void amdgpu_display_print_display_setup(struct drm_device *dev);
+int amdgpu_display_modeset_create_props(struct amdgpu_device *adev);
+int amdgpu_display_crtc_set_config(struct drm_mode_set *set,
+				   struct drm_modeset_acquire_ctx *ctx);
+int amdgpu_display_crtc_page_flip_target(struct drm_crtc *crtc,
+				struct drm_framebuffer *fb,
+				struct drm_pending_vblank_event *event,
+				uint32_t page_flip_flags, uint32_t target,
+				struct drm_modeset_acquire_ctx *ctx);
 extern const struct drm_mode_config_funcs amdgpu_mode_funcs;
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.c
new file mode 100644
index 000000000000..a974265837f0
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.c
@@ -0,0 +1,86 @@
+/*
+ * Copyright (C) 2019  Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include "amdgpu.h"
+#include "amdgpu_ras.h"
+
+int amdgpu_nbio_ras_sw_init(struct amdgpu_device *adev)
+{
+	int err;
+	struct amdgpu_nbio_ras *ras;
+
+	if (!adev->nbio.ras)
+		return 0;
+
+	ras = adev->nbio.ras;
+	err = amdgpu_ras_register_ras_block(adev, &ras->ras_block);
+	if (err) {
+		dev_err(adev->dev, "Failed to register pcie_bif ras block!\n");
+		return err;
+	}
+
+	strcpy(ras->ras_block.ras_comm.name, "pcie_bif");
+	ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__PCIE_BIF;
+	ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
+	adev->nbio.ras_if = &ras->ras_block.ras_comm;
+
+	return 0;
+}
+
+u64 amdgpu_nbio_get_pcie_replay_count(struct amdgpu_device *adev)
+{
+	if (adev->nbio.funcs && adev->nbio.funcs->get_pcie_replay_count)
+		return adev->nbio.funcs->get_pcie_replay_count(adev);
+
+	return 0;
+}
+
+bool amdgpu_nbio_is_replay_cnt_supported(struct amdgpu_device *adev)
+{
+	if (amdgpu_sriov_vf(adev) || !adev->asic_funcs ||
+	    !adev->asic_funcs->get_pcie_replay_count ||
+	    (!adev->nbio.funcs || !adev->nbio.funcs->get_pcie_replay_count))
+		return false;
+
+	return true;
+}
+
+int amdgpu_nbio_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
+{
+	int r;
+	r = amdgpu_ras_block_late_init(adev, ras_block);
+	if (r)
+		return r;
+
+	if (amdgpu_ras_is_supported(adev, ras_block->block)) {
+		r = amdgpu_irq_get(adev, &adev->nbio.ras_controller_irq, 0);
+		if (r)
+			goto late_fini;
+		r = amdgpu_irq_get(adev, &adev->nbio.ras_err_event_athub_irq, 0);
+		if (r)
+			goto late_fini;
+	}
+
+	return 0;
+late_fini:
+	amdgpu_ras_block_late_fini(adev, ras_block);
+	return r;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h
new file mode 100644
index 000000000000..b528de6a01f6
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h
@@ -0,0 +1,124 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __AMDGPU_NBIO_H__
+#define __AMDGPU_NBIO_H__
+
+/*
+ * amdgpu nbio functions
+ */
+struct nbio_hdp_flush_reg {
+	u32 ref_and_mask_cp0;
+	u32 ref_and_mask_cp1;
+	u32 ref_and_mask_cp2;
+	u32 ref_and_mask_cp3;
+	u32 ref_and_mask_cp4;
+	u32 ref_and_mask_cp5;
+	u32 ref_and_mask_cp6;
+	u32 ref_and_mask_cp7;
+	u32 ref_and_mask_cp8;
+	u32 ref_and_mask_cp9;
+	u32 ref_and_mask_sdma0;
+	u32 ref_and_mask_sdma1;
+	u32 ref_and_mask_sdma2;
+	u32 ref_and_mask_sdma3;
+	u32 ref_and_mask_sdma4;
+	u32 ref_and_mask_sdma5;
+	u32 ref_and_mask_sdma6;
+	u32 ref_and_mask_sdma7;
+};
+
+struct amdgpu_nbio_ras {
+	struct amdgpu_ras_block_object ras_block;
+	void (*handle_ras_controller_intr_no_bifring)(struct amdgpu_device *adev);
+	void (*handle_ras_err_event_athub_intr_no_bifring)(struct amdgpu_device *adev);
+	int (*init_ras_controller_interrupt)(struct amdgpu_device *adev);
+	int (*init_ras_err_event_athub_interrupt)(struct amdgpu_device *adev);
+};
+
+struct amdgpu_nbio_funcs {
+	const struct nbio_hdp_flush_reg *hdp_flush_reg;
+	u32 (*get_hdp_flush_req_offset)(struct amdgpu_device *adev);
+	u32 (*get_hdp_flush_done_offset)(struct amdgpu_device *adev);
+	u32 (*get_pcie_index_offset)(struct amdgpu_device *adev);
+	u32 (*get_pcie_data_offset)(struct amdgpu_device *adev);
+	u32 (*get_pcie_index_hi_offset)(struct amdgpu_device *adev);
+	u32 (*get_pcie_port_index_offset)(struct amdgpu_device *adev);
+	u32 (*get_pcie_port_data_offset)(struct amdgpu_device *adev);
+	u32 (*get_rev_id)(struct amdgpu_device *adev);
+	void (*mc_access_enable)(struct amdgpu_device *adev, bool enable);
+	u32 (*get_memsize)(struct amdgpu_device *adev);
+	void (*sdma_doorbell_range)(struct amdgpu_device *adev, int instance,
+			bool use_doorbell, int doorbell_index, int doorbell_size);
+	void (*vpe_doorbell_range)(struct amdgpu_device *adev, int instance,
+			bool use_doorbell, int doorbell_index, int doorbell_size);
+	void (*vcn_doorbell_range)(struct amdgpu_device *adev, bool use_doorbell,
+				   int doorbell_index, int instance);
+	void (*gc_doorbell_init)(struct amdgpu_device *adev);
+	void (*enable_doorbell_aperture)(struct amdgpu_device *adev,
+					 bool enable);
+	void (*enable_doorbell_selfring_aperture)(struct amdgpu_device *adev,
+						  bool enable);
+	void (*ih_doorbell_range)(struct amdgpu_device *adev,
+				  bool use_doorbell, int doorbell_index);
+	void (*enable_doorbell_interrupt)(struct amdgpu_device *adev,
+					  bool enable);
+	void (*update_medium_grain_clock_gating)(struct amdgpu_device *adev,
+						 bool enable);
+	void (*update_medium_grain_light_sleep)(struct amdgpu_device *adev,
+						bool enable);
+	void (*get_clockgating_state)(struct amdgpu_device *adev,
+				      u64 *flags);
+	void (*ih_control)(struct amdgpu_device *adev);
+	void (*init_registers)(struct amdgpu_device *adev);
+	void (*remap_hdp_registers)(struct amdgpu_device *adev);
+	void (*enable_aspm)(struct amdgpu_device *adev,
+			    bool enable);
+	void (*program_aspm)(struct amdgpu_device *adev);
+	void (*apply_lc_spc_mode_wa)(struct amdgpu_device *adev);
+	void (*apply_l1_link_width_reconfig_wa)(struct amdgpu_device *adev);
+	void (*clear_doorbell_interrupt)(struct amdgpu_device *adev);
+	u32 (*get_rom_offset)(struct amdgpu_device *adev);
+	int (*get_compute_partition_mode)(struct amdgpu_device *adev);
+	u32 (*get_memory_partition_mode)(struct amdgpu_device *adev,
+					 u32 *supp_modes);
+	bool (*is_nps_switch_requested)(struct amdgpu_device *adev);
+	u64 (*get_pcie_replay_count)(struct amdgpu_device *adev);
+	void (*set_reg_remap)(struct amdgpu_device *adev);
+};
+
+struct amdgpu_nbio {
+	const struct nbio_hdp_flush_reg *hdp_flush_reg;
+	struct amdgpu_irq_src ras_controller_irq;
+	struct amdgpu_irq_src ras_err_event_athub_irq;
+	struct ras_common_if *ras_if;
+	const struct amdgpu_nbio_funcs *funcs;
+	struct amdgpu_nbio_ras  *ras;
+};
+
+int amdgpu_nbio_ras_sw_init(struct amdgpu_device *adev);
+int amdgpu_nbio_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block);
+u64 amdgpu_nbio_get_pcie_replay_count(struct amdgpu_device *adev);
+
+bool amdgpu_nbio_is_replay_cnt_supported(struct amdgpu_device *adev);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
index 9e495da0bb03..e08f58de4b17 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
@@ -31,145 +31,197 @@
  */
 #include <linux/list.h>
 #include <linux/slab.h>
-#include <drm/drmP.h>
+#include <linux/dma-buf.h>
+#include <linux/export.h>
+
+#include <drm/drm_drv.h>
 #include <drm/amdgpu_drm.h>
 #include <drm/drm_cache.h>
 #include "amdgpu.h"
 #include "amdgpu_trace.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_vram_mgr.h"
+#include "amdgpu_vm.h"
+#include "amdgpu_dma_buf.h"
 
-static void amdgpu_ttm_bo_destroy(struct ttm_buffer_object *tbo)
-{
-	struct amdgpu_device *adev = amdgpu_ttm_adev(tbo->bdev);
-	struct amdgpu_bo *bo;
+/**
+ * DOC: amdgpu_object
+ *
+ * This defines the interfaces to operate on an &amdgpu_bo buffer object which
+ * represents memory used by driver (VRAM, system memory, etc.). The driver
+ * provides DRM/GEM APIs to userspace. DRM/GEM APIs then use these interfaces
+ * to create/destroy/set buffer object which are then managed by the kernel TTM
+ * memory manager.
+ * The interfaces are also used internally by kernel clients, including gfx,
+ * uvd, etc. for kernel managed allocations used by the GPU.
+ *
+ */
 
-	bo = container_of(tbo, struct amdgpu_bo, tbo);
+static void amdgpu_bo_destroy(struct ttm_buffer_object *tbo)
+{
+	struct amdgpu_bo *bo = ttm_to_amdgpu_bo(tbo);
 
 	amdgpu_bo_kunmap(bo);
 
-	drm_gem_object_release(&bo->gem_base);
+	if (drm_gem_is_imported(&bo->tbo.base))
+		drm_prime_gem_destroy(&bo->tbo.base, bo->tbo.sg);
+	drm_gem_object_release(&bo->tbo.base);
 	amdgpu_bo_unref(&bo->parent);
-	if (!list_empty(&bo->shadow_list)) {
-		mutex_lock(&adev->shadow_list_lock);
-		list_del_init(&bo->shadow_list);
-		mutex_unlock(&adev->shadow_list_lock);
-	}
-	kfree(bo->metadata);
-	kfree(bo);
+	kvfree(bo);
+}
+
+static void amdgpu_bo_user_destroy(struct ttm_buffer_object *tbo)
+{
+	struct amdgpu_bo *bo = ttm_to_amdgpu_bo(tbo);
+	struct amdgpu_bo_user *ubo;
+
+	ubo = to_amdgpu_bo_user(bo);
+	kfree(ubo->metadata);
+	amdgpu_bo_destroy(tbo);
 }
 
-bool amdgpu_ttm_bo_is_amdgpu_bo(struct ttm_buffer_object *bo)
+/**
+ * amdgpu_bo_is_amdgpu_bo - check if the buffer object is an &amdgpu_bo
+ * @bo: buffer object to be checked
+ *
+ * Uses destroy function associated with the object to determine if this is
+ * an &amdgpu_bo.
+ *
+ * Returns:
+ * true if the object belongs to &amdgpu_bo, false if not.
+ */
+bool amdgpu_bo_is_amdgpu_bo(struct ttm_buffer_object *bo)
 {
-	if (bo->destroy == &amdgpu_ttm_bo_destroy)
+	if (bo->destroy == &amdgpu_bo_destroy ||
+	    bo->destroy == &amdgpu_bo_user_destroy)
 		return true;
+
 	return false;
 }
 
-static void amdgpu_ttm_placement_init(struct amdgpu_device *adev,
-				      struct ttm_placement *placement,
-				      struct ttm_place *places,
-				      u32 domain, u64 flags)
+/**
+ * amdgpu_bo_placement_from_domain - set buffer's placement
+ * @abo: &amdgpu_bo buffer object whose placement is to be set
+ * @domain: requested domain
+ *
+ * Sets buffer's placement according to requested domain and the buffer's
+ * flags.
+ */
+void amdgpu_bo_placement_from_domain(struct amdgpu_bo *abo, u32 domain)
 {
+	struct amdgpu_device *adev = amdgpu_ttm_adev(abo->tbo.bdev);
+	struct ttm_placement *placement = &abo->placement;
+	struct ttm_place *places = abo->placements;
+	u64 flags = abo->flags;
 	u32 c = 0;
 
 	if (domain & AMDGPU_GEM_DOMAIN_VRAM) {
-		unsigned visible_pfn = adev->mc.visible_vram_size >> PAGE_SHIFT;
-
-		places[c].fpfn = 0;
-		places[c].lpfn = 0;
-		places[c].flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_UNCACHED |
-			TTM_PL_FLAG_VRAM;
+		unsigned int visible_pfn = adev->gmc.visible_vram_size >> PAGE_SHIFT;
+		int8_t mem_id = KFD_XCP_MEM_ID(adev, abo->xcp_id);
+
+		if (adev->gmc.mem_partitions && mem_id >= 0) {
+			places[c].fpfn = adev->gmc.mem_partitions[mem_id].range.fpfn;
+			/*
+			 * memory partition range lpfn is inclusive start + size - 1
+			 * TTM place lpfn is exclusive start + size
+			 */
+			places[c].lpfn = adev->gmc.mem_partitions[mem_id].range.lpfn + 1;
+		} else {
+			places[c].fpfn = 0;
+			places[c].lpfn = 0;
+		}
+		places[c].mem_type = TTM_PL_VRAM;
+		places[c].flags = 0;
 
 		if (flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED)
-			places[c].lpfn = visible_pfn;
+			places[c].lpfn = min_not_zero(places[c].lpfn, visible_pfn);
 		else
 			places[c].flags |= TTM_PL_FLAG_TOPDOWN;
 
-		if (flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS)
+		if (abo->tbo.type == ttm_bo_type_kernel &&
+		    flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS)
 			places[c].flags |= TTM_PL_FLAG_CONTIGUOUS;
+
+		c++;
+	}
+
+	if (domain & AMDGPU_GEM_DOMAIN_DOORBELL) {
+		places[c].fpfn = 0;
+		places[c].lpfn = 0;
+		places[c].mem_type = AMDGPU_PL_DOORBELL;
+		places[c].flags = 0;
+		c++;
+	}
+
+	if (domain & AMDGPU_GEM_DOMAIN_MMIO_REMAP) {
+		places[c].fpfn = 0;
+		places[c].lpfn = 0;
+		places[c].mem_type = AMDGPU_PL_MMIO_REMAP;
+		places[c].flags = 0;
 		c++;
 	}
 
 	if (domain & AMDGPU_GEM_DOMAIN_GTT) {
 		places[c].fpfn = 0;
-		if (flags & AMDGPU_GEM_CREATE_SHADOW)
-			places[c].lpfn = adev->mc.gart_size >> PAGE_SHIFT;
-		else
-			places[c].lpfn = 0;
-		places[c].flags = TTM_PL_FLAG_TT;
-		if (flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC)
-			places[c].flags |= TTM_PL_FLAG_WC |
-				TTM_PL_FLAG_UNCACHED;
-		else
-			places[c].flags |= TTM_PL_FLAG_CACHED;
+		places[c].lpfn = 0;
+		places[c].mem_type =
+			abo->flags & AMDGPU_GEM_CREATE_PREEMPTIBLE ?
+			AMDGPU_PL_PREEMPT : TTM_PL_TT;
+		places[c].flags = 0;
+		/*
+		 * When GTT is just an alternative to VRAM make sure that we
+		 * only use it as fallback and still try to fill up VRAM first.
+		 */
+		if (abo->tbo.resource && !(adev->flags & AMD_IS_APU) &&
+		    domain & abo->preferred_domains & AMDGPU_GEM_DOMAIN_VRAM)
+			places[c].flags |= TTM_PL_FLAG_FALLBACK;
 		c++;
 	}
 
 	if (domain & AMDGPU_GEM_DOMAIN_CPU) {
 		places[c].fpfn = 0;
 		places[c].lpfn = 0;
-		places[c].flags = TTM_PL_FLAG_SYSTEM;
-		if (flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC)
-			places[c].flags |= TTM_PL_FLAG_WC |
-				TTM_PL_FLAG_UNCACHED;
-		else
-			places[c].flags |= TTM_PL_FLAG_CACHED;
+		places[c].mem_type = TTM_PL_SYSTEM;
+		places[c].flags = 0;
 		c++;
 	}
 
 	if (domain & AMDGPU_GEM_DOMAIN_GDS) {
 		places[c].fpfn = 0;
 		places[c].lpfn = 0;
-		places[c].flags = TTM_PL_FLAG_UNCACHED | AMDGPU_PL_FLAG_GDS;
+		places[c].mem_type = AMDGPU_PL_GDS;
+		places[c].flags = 0;
 		c++;
 	}
 
 	if (domain & AMDGPU_GEM_DOMAIN_GWS) {
 		places[c].fpfn = 0;
 		places[c].lpfn = 0;
-		places[c].flags = TTM_PL_FLAG_UNCACHED | AMDGPU_PL_FLAG_GWS;
+		places[c].mem_type = AMDGPU_PL_GWS;
+		places[c].flags = 0;
 		c++;
 	}
 
 	if (domain & AMDGPU_GEM_DOMAIN_OA) {
 		places[c].fpfn = 0;
 		places[c].lpfn = 0;
-		places[c].flags = TTM_PL_FLAG_UNCACHED | AMDGPU_PL_FLAG_OA;
+		places[c].mem_type = AMDGPU_PL_OA;
+		places[c].flags = 0;
 		c++;
 	}
 
 	if (!c) {
 		places[c].fpfn = 0;
 		places[c].lpfn = 0;
-		places[c].flags = TTM_PL_MASK_CACHING | TTM_PL_FLAG_SYSTEM;
+		places[c].mem_type = TTM_PL_SYSTEM;
+		places[c].flags = 0;
 		c++;
 	}
 
+	BUG_ON(c > AMDGPU_BO_MAX_PLACEMENTS);
+
 	placement->num_placement = c;
 	placement->placement = places;
-
-	placement->num_busy_placement = c;
-	placement->busy_placement = places;
-}
-
-void amdgpu_ttm_placement_from_domain(struct amdgpu_bo *abo, u32 domain)
-{
-	struct amdgpu_device *adev = amdgpu_ttm_adev(abo->tbo.bdev);
-
-	amdgpu_ttm_placement_init(adev, &abo->placement, abo->placements,
-				  domain, abo->flags);
-}
-
-static void amdgpu_fill_placement_to_bo(struct amdgpu_bo *bo,
-					struct ttm_placement *placement)
-{
-	BUG_ON(placement->num_placement > (AMDGPU_GEM_DOMAIN_MAX + 1));
-
-	memcpy(bo->placements, placement->placement,
-	       placement->num_placement * sizeof(struct ttm_place));
-	bo->placement.num_placement = placement->num_placement;
-	bo->placement.num_busy_placement = placement->num_busy_placement;
-	bo->placement.placement = bo->placements;
-	bo->placement.busy_placement = bo->placements;
 }
 
 /**
@@ -179,28 +231,45 @@ static void amdgpu_fill_placement_to_bo(struct amdgpu_bo *bo,
  * @size: size for the new BO
  * @align: alignment for the new BO
  * @domain: where to place it
- * @bo_ptr: resulting BO
+ * @bo_ptr: used to initialize BOs in structures
  * @gpu_addr: GPU addr of the pinned BO
  * @cpu_addr: optional CPU address mapping
  *
  * Allocates and pins a BO for kernel internal use, and returns it still
  * reserved.
  *
- * Returns 0 on success, negative error code otherwise.
+ * Note: For bo_ptr new BO is only created if bo_ptr points to NULL.
+ *
+ * Returns:
+ * 0 on success, negative error code otherwise.
  */
 int amdgpu_bo_create_reserved(struct amdgpu_device *adev,
 			      unsigned long size, int align,
 			      u32 domain, struct amdgpu_bo **bo_ptr,
 			      u64 *gpu_addr, void **cpu_addr)
 {
+	struct amdgpu_bo_param bp;
 	bool free = false;
 	int r;
 
+	if (!size) {
+		amdgpu_bo_unref(bo_ptr);
+		return 0;
+	}
+
+	memset(&bp, 0, sizeof(bp));
+	bp.size = size;
+	bp.byte_align = align;
+	bp.domain = domain;
+	bp.flags = cpu_addr ? AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED
+		: AMDGPU_GEM_CREATE_NO_CPU_ACCESS;
+	bp.flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+	bp.type = ttm_bo_type_kernel;
+	bp.resv = NULL;
+	bp.bo_ptr_size = sizeof(struct amdgpu_bo);
+
 	if (!*bo_ptr) {
-		r = amdgpu_bo_create(adev, size, align, true, domain,
-				     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
-				     AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
-				     NULL, NULL, 0, bo_ptr);
+		r = amdgpu_bo_create(adev, &bp, bo_ptr);
 		if (r) {
 			dev_err(adev->dev, "(%d) failed to allocate kernel bo\n",
 				r);
@@ -215,22 +284,33 @@ int amdgpu_bo_create_reserved(struct amdgpu_device *adev,
 		goto error_free;
 	}
 
-	r = amdgpu_bo_pin(*bo_ptr, domain, gpu_addr);
+	r = amdgpu_bo_pin(*bo_ptr, domain);
 	if (r) {
 		dev_err(adev->dev, "(%d) kernel bo pin failed\n", r);
 		goto error_unreserve;
 	}
 
+	r = amdgpu_ttm_alloc_gart(&(*bo_ptr)->tbo);
+	if (r) {
+		dev_err(adev->dev, "%p bind failed\n", *bo_ptr);
+		goto error_unpin;
+	}
+
+	if (gpu_addr)
+		*gpu_addr = amdgpu_bo_gpu_offset(*bo_ptr);
+
 	if (cpu_addr) {
 		r = amdgpu_bo_kmap(*bo_ptr, cpu_addr);
 		if (r) {
 			dev_err(adev->dev, "(%d) kernel bo map failed\n", r);
-			goto error_unreserve;
+			goto error_unpin;
 		}
 	}
 
 	return 0;
 
+error_unpin:
+	amdgpu_bo_unpin(*bo_ptr);
 error_unreserve:
 	amdgpu_bo_unreserve(*bo_ptr);
 
@@ -248,13 +328,19 @@ error_free:
  * @size: size for the new BO
  * @align: alignment for the new BO
  * @domain: where to place it
- * @bo_ptr: resulting BO
+ * @bo_ptr:  used to initialize BOs in structures
  * @gpu_addr: GPU addr of the pinned BO
  * @cpu_addr: optional CPU address mapping
  *
  * Allocates and pins a BO for kernel internal use.
  *
- * Returns 0 on success, negative error code otherwise.
+ * This function is exported to allow the V4L2 isp device
+ * external to drm device to create and access the kernel BO.
+ *
+ * Note: For bo_ptr new BO is only created if bo_ptr points to NULL.
+ *
+ * Returns:
+ * 0 on success, negative error code otherwise.
  */
 int amdgpu_bo_create_kernel(struct amdgpu_device *adev,
 			    unsigned long size, int align,
@@ -269,17 +355,158 @@ int amdgpu_bo_create_kernel(struct amdgpu_device *adev,
 	if (r)
 		return r;
 
-	amdgpu_bo_unreserve(*bo_ptr);
+	if (*bo_ptr)
+		amdgpu_bo_unreserve(*bo_ptr);
 
 	return 0;
 }
 
 /**
+ * amdgpu_bo_create_isp_user - create user BO for isp
+ *
+ * @adev: amdgpu device object
+ * @dma_buf: DMABUF handle for isp buffer
+ * @domain: where to place it
+ * @bo:  used to initialize BOs in structures
+ * @gpu_addr: GPU addr of the pinned BO
+ *
+ * Imports isp DMABUF to allocate and pin a user BO for isp internal use. It does
+ * GART alloc to generate gpu_addr for BO to make it accessible through the
+ * GART aperture for ISP HW.
+ *
+ * This function is exported to allow the V4L2 isp device external to drm device
+ * to create and access the isp user BO.
+ *
+ * Returns:
+ * 0 on success, negative error code otherwise.
+ */
+int amdgpu_bo_create_isp_user(struct amdgpu_device *adev,
+			   struct dma_buf *dma_buf, u32 domain, struct amdgpu_bo **bo,
+			   u64 *gpu_addr)
+
+{
+	struct drm_gem_object *gem_obj;
+	int r;
+
+	gem_obj = amdgpu_gem_prime_import(&adev->ddev, dma_buf);
+	*bo = gem_to_amdgpu_bo(gem_obj);
+	if (!(*bo)) {
+		dev_err(adev->dev, "failed to get valid isp user bo\n");
+		return -EINVAL;
+	}
+
+	r = amdgpu_bo_reserve(*bo, false);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to reserve isp user bo\n", r);
+		return r;
+	}
+
+	r = amdgpu_bo_pin(*bo, domain);
+	if (r) {
+		dev_err(adev->dev, "(%d) isp user bo pin failed\n", r);
+		goto error_unreserve;
+	}
+
+	r = amdgpu_ttm_alloc_gart(&(*bo)->tbo);
+	if (r) {
+		dev_err(adev->dev, "%p bind failed\n", *bo);
+		goto error_unpin;
+	}
+
+	if (!WARN_ON(!gpu_addr))
+		*gpu_addr = amdgpu_bo_gpu_offset(*bo);
+
+	amdgpu_bo_unreserve(*bo);
+
+	return 0;
+
+error_unpin:
+	amdgpu_bo_unpin(*bo);
+error_unreserve:
+	amdgpu_bo_unreserve(*bo);
+	amdgpu_bo_unref(bo);
+
+	return r;
+}
+
+/**
+ * amdgpu_bo_create_kernel_at - create BO for kernel use at specific location
+ *
+ * @adev: amdgpu device object
+ * @offset: offset of the BO
+ * @size: size of the BO
+ * @bo_ptr:  used to initialize BOs in structures
+ * @cpu_addr: optional CPU address mapping
+ *
+ * Creates a kernel BO at a specific offset in VRAM.
+ *
+ * Returns:
+ * 0 on success, negative error code otherwise.
+ */
+int amdgpu_bo_create_kernel_at(struct amdgpu_device *adev,
+			       uint64_t offset, uint64_t size,
+			       struct amdgpu_bo **bo_ptr, void **cpu_addr)
+{
+	struct ttm_operation_ctx ctx = { false, false };
+	unsigned int i;
+	int r;
+
+	offset &= PAGE_MASK;
+	size = ALIGN(size, PAGE_SIZE);
+
+	r = amdgpu_bo_create_reserved(adev, size, PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_VRAM, bo_ptr, NULL,
+				      cpu_addr);
+	if (r)
+		return r;
+
+	if ((*bo_ptr) == NULL)
+		return 0;
+
+	/*
+	 * Remove the original mem node and create a new one at the request
+	 * position.
+	 */
+	if (cpu_addr)
+		amdgpu_bo_kunmap(*bo_ptr);
+
+	ttm_resource_free(&(*bo_ptr)->tbo, &(*bo_ptr)->tbo.resource);
+
+	for (i = 0; i < (*bo_ptr)->placement.num_placement; ++i) {
+		(*bo_ptr)->placements[i].fpfn = offset >> PAGE_SHIFT;
+		(*bo_ptr)->placements[i].lpfn = (offset + size) >> PAGE_SHIFT;
+	}
+	r = ttm_bo_mem_space(&(*bo_ptr)->tbo, &(*bo_ptr)->placement,
+			     &(*bo_ptr)->tbo.resource, &ctx);
+	if (r)
+		goto error;
+
+	if (cpu_addr) {
+		r = amdgpu_bo_kmap(*bo_ptr, cpu_addr);
+		if (r)
+			goto error;
+	}
+
+	amdgpu_bo_unreserve(*bo_ptr);
+	return 0;
+
+error:
+	amdgpu_bo_unreserve(*bo_ptr);
+	amdgpu_bo_unref(bo_ptr);
+	return r;
+}
+
+/**
  * amdgpu_bo_free_kernel - free BO for kernel use
  *
  * @bo: amdgpu BO to free
+ * @gpu_addr: pointer to where the BO's GPU memory space address was stored
+ * @cpu_addr: pointer to where the BO's CPU memory space address was stored
  *
  * unmaps and unpin a BO for kernel internal use.
+ *
+ * This function is exported to allow the V4L2 isp device
+ * external to drm device to free the kernel BO.
  */
 void amdgpu_bo_free_kernel(struct amdgpu_bo **bo, u64 *gpu_addr,
 			   void **cpu_addr)
@@ -287,6 +514,8 @@ void amdgpu_bo_free_kernel(struct amdgpu_bo **bo, u64 *gpu_addr,
 	if (*bo == NULL)
 		return;
 
+	WARN_ON(amdgpu_ttm_adev((*bo)->tbo.bdev)->in_suspend);
+
 	if (likely(amdgpu_bo_reserve(*bo, true) == 0)) {
 		if (cpu_addr)
 			amdgpu_bo_kunmap(*bo);
@@ -303,64 +532,67 @@ void amdgpu_bo_free_kernel(struct amdgpu_bo **bo, u64 *gpu_addr,
 		*cpu_addr = NULL;
 }
 
-int amdgpu_bo_create_restricted(struct amdgpu_device *adev,
-				unsigned long size, int byte_align,
-				bool kernel, u32 domain, u64 flags,
-				struct sg_table *sg,
-				struct ttm_placement *placement,
-				struct reservation_object *resv,
-				uint64_t init_value,
-				struct amdgpu_bo **bo_ptr)
+/**
+ * amdgpu_bo_free_isp_user - free BO for isp use
+ *
+ * @bo: amdgpu isp user BO to free
+ *
+ * unpin and unref BO for isp internal use.
+ *
+ * This function is exported to allow the V4L2 isp device
+ * external to drm device to free the isp user BO.
+ */
+void amdgpu_bo_free_isp_user(struct amdgpu_bo *bo)
 {
-	struct amdgpu_bo *bo;
-	enum ttm_bo_type type;
-	unsigned long page_align;
-	u64 initial_bytes_moved, bytes_moved;
-	size_t acc_size;
-	int r;
-
-	page_align = roundup(byte_align, PAGE_SIZE) >> PAGE_SHIFT;
-	size = ALIGN(size, PAGE_SIZE);
+	if (bo == NULL)
+		return;
 
-	if (kernel) {
-		type = ttm_bo_type_kernel;
-	} else if (sg) {
-		type = ttm_bo_type_sg;
-	} else {
-		type = ttm_bo_type_device;
+	if (amdgpu_bo_reserve(bo, true) == 0) {
+		amdgpu_bo_unpin(bo);
+		amdgpu_bo_unreserve(bo);
 	}
-	*bo_ptr = NULL;
+	amdgpu_bo_unref(&bo);
+}
 
-	acc_size = ttm_bo_dma_acc_size(&adev->mman.bdev, size,
-				       sizeof(struct amdgpu_bo));
+/* Validate bo size is bit bigger than the request domain */
+static bool amdgpu_bo_validate_size(struct amdgpu_device *adev,
+					  unsigned long size, u32 domain)
+{
+	struct ttm_resource_manager *man = NULL;
 
-	bo = kzalloc(sizeof(struct amdgpu_bo), GFP_KERNEL);
-	if (bo == NULL)
-		return -ENOMEM;
-	r = drm_gem_object_init(adev->ddev, &bo->gem_base, size);
-	if (unlikely(r)) {
-		kfree(bo);
-		return r;
+	/*
+	 * If GTT is part of requested domains the check must succeed to
+	 * allow fall back to GTT.
+	 */
+	if (domain & AMDGPU_GEM_DOMAIN_GTT)
+		man = ttm_manager_type(&adev->mman.bdev, TTM_PL_TT);
+	else if (domain & AMDGPU_GEM_DOMAIN_VRAM)
+		man = ttm_manager_type(&adev->mman.bdev, TTM_PL_VRAM);
+	else
+		return true;
+
+	if (!man) {
+		if (domain & AMDGPU_GEM_DOMAIN_GTT)
+			WARN_ON_ONCE("GTT domain requested but GTT mem manager uninitialized");
+		return false;
 	}
-	INIT_LIST_HEAD(&bo->shadow_list);
-	INIT_LIST_HEAD(&bo->va);
-	bo->preferred_domains = domain & (AMDGPU_GEM_DOMAIN_VRAM |
-					 AMDGPU_GEM_DOMAIN_GTT |
-					 AMDGPU_GEM_DOMAIN_CPU |
-					 AMDGPU_GEM_DOMAIN_GDS |
-					 AMDGPU_GEM_DOMAIN_GWS |
-					 AMDGPU_GEM_DOMAIN_OA);
-	bo->allowed_domains = bo->preferred_domains;
-	if (!kernel && bo->allowed_domains == AMDGPU_GEM_DOMAIN_VRAM)
-		bo->allowed_domains |= AMDGPU_GEM_DOMAIN_GTT;
 
-	bo->flags = flags;
+	/* TODO add more domains checks, such as AMDGPU_GEM_DOMAIN_CPU, _DOMAIN_DOORBELL */
+	if (size < man->size)
+		return true;
+
+	DRM_DEBUG("BO size %lu > total memory in domain: %llu\n", size, man->size);
+	return false;
+}
+
+bool amdgpu_bo_support_uswc(u64 bo_flags)
+{
 
 #ifdef CONFIG_X86_32
 	/* XXX: Write-combined CPU mappings of GTT seem broken on 32-bit
 	 * See https://bugs.freedesktop.org/show_bug.cgi?id=84627
 	 */
-	bo->flags &= ~AMDGPU_GEM_CREATE_CPU_GTT_USWC;
+	return false;
 #elif defined(CONFIG_X86) && !defined(CONFIG_X86_PAT)
 	/* Don't try to enable write-combining when it can't work, or things
 	 * may be slow
@@ -372,235 +604,225 @@ int amdgpu_bo_create_restricted(struct amdgpu_device *adev,
 	 thanks to write-combining
 #endif
 
-	if (bo->flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC)
+	if (bo_flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC)
 		DRM_INFO_ONCE("Please enable CONFIG_MTRR and CONFIG_X86_PAT for "
 			      "better performance thanks to write-combining\n");
-	bo->flags &= ~AMDGPU_GEM_CREATE_CPU_GTT_USWC;
+	return false;
 #else
 	/* For architectures that don't support WC memory,
 	 * mask out the WC flag from the BO
 	 */
 	if (!drm_arch_can_wc_memory())
-		bo->flags &= ~AMDGPU_GEM_CREATE_CPU_GTT_USWC;
+		return false;
+
+	return true;
 #endif
+}
+
+/**
+ * amdgpu_bo_create - create an &amdgpu_bo buffer object
+ * @adev: amdgpu device object
+ * @bp: parameters to be used for the buffer object
+ * @bo_ptr: pointer to the buffer object pointer
+ *
+ * Creates an &amdgpu_bo buffer object.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_bo_create(struct amdgpu_device *adev,
+			       struct amdgpu_bo_param *bp,
+			       struct amdgpu_bo **bo_ptr)
+{
+	struct ttm_operation_ctx ctx = {
+		.interruptible = (bp->type != ttm_bo_type_kernel),
+		.no_wait_gpu = bp->no_wait_gpu,
+		/* We opt to avoid OOM on system pages allocations */
+		.gfp_retry_mayfail = true,
+		.allow_res_evict = bp->type != ttm_bo_type_kernel,
+		.resv = bp->resv
+	};
+	struct amdgpu_bo *bo;
+	unsigned long page_align, size = bp->size;
+	int r;
+
+	/* Note that GDS/GWS/OA allocates 1 page per byte/resource. */
+	if (bp->domain & (AMDGPU_GEM_DOMAIN_GWS | AMDGPU_GEM_DOMAIN_OA)) {
+		/* GWS and OA don't need any alignment. */
+		page_align = bp->byte_align;
+		size <<= PAGE_SHIFT;
+
+	} else if (bp->domain & AMDGPU_GEM_DOMAIN_GDS) {
+		/* Both size and alignment must be a multiple of 4. */
+		page_align = ALIGN(bp->byte_align, 4);
+		size = ALIGN(size, 4) << PAGE_SHIFT;
+	} else {
+		/* Memory should be aligned at least to a page size. */
+		page_align = ALIGN(bp->byte_align, PAGE_SIZE) >> PAGE_SHIFT;
+		size = ALIGN(size, PAGE_SIZE);
+	}
+
+	if (!amdgpu_bo_validate_size(adev, size, bp->domain))
+		return -ENOMEM;
+
+	BUG_ON(bp->bo_ptr_size < sizeof(struct amdgpu_bo));
+
+	*bo_ptr = NULL;
+	bo = kvzalloc(bp->bo_ptr_size, GFP_KERNEL);
+	if (bo == NULL)
+		return -ENOMEM;
+	drm_gem_private_object_init(adev_to_drm(adev), &bo->tbo.base, size);
+	bo->tbo.base.funcs = &amdgpu_gem_object_funcs;
+	bo->vm_bo = NULL;
+	bo->preferred_domains = bp->preferred_domain ? bp->preferred_domain :
+		bp->domain;
+	bo->allowed_domains = bo->preferred_domains;
+	if (bp->type != ttm_bo_type_kernel &&
+	    !(bp->flags & AMDGPU_GEM_CREATE_DISCARDABLE) &&
+	    bo->allowed_domains == AMDGPU_GEM_DOMAIN_VRAM)
+		bo->allowed_domains |= AMDGPU_GEM_DOMAIN_GTT;
+
+	bo->flags = bp->flags;
+
+	if (adev->gmc.mem_partitions)
+		/* For GPUs with spatial partitioning, bo->xcp_id=-1 means any partition */
+		bo->xcp_id = bp->xcp_id_plus1 - 1;
+	else
+		/* For GPUs without spatial partitioning */
+		bo->xcp_id = 0;
 
-	amdgpu_fill_placement_to_bo(bo, placement);
-	/* Kernel allocation are uninterruptible */
-
-	initial_bytes_moved = atomic64_read(&adev->num_bytes_moved);
-	r = ttm_bo_init_reserved(&adev->mman.bdev, &bo->tbo, size, type,
-				 &bo->placement, page_align, !kernel, NULL,
-				 acc_size, sg, resv, &amdgpu_ttm_bo_destroy);
-	bytes_moved = atomic64_read(&adev->num_bytes_moved) -
-		      initial_bytes_moved;
-	if (adev->mc.visible_vram_size < adev->mc.real_vram_size &&
-	    bo->tbo.mem.mem_type == TTM_PL_VRAM &&
-	    bo->tbo.mem.start < adev->mc.visible_vram_size >> PAGE_SHIFT)
-		amdgpu_cs_report_moved_bytes(adev, bytes_moved, bytes_moved);
+	if (!amdgpu_bo_support_uswc(bo->flags))
+		bo->flags &= ~AMDGPU_GEM_CREATE_CPU_GTT_USWC;
+
+	bo->tbo.bdev = &adev->mman.bdev;
+	if (bp->domain & (AMDGPU_GEM_DOMAIN_GWS | AMDGPU_GEM_DOMAIN_OA |
+			  AMDGPU_GEM_DOMAIN_GDS))
+		amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_CPU);
 	else
-		amdgpu_cs_report_moved_bytes(adev, bytes_moved, 0);
+		amdgpu_bo_placement_from_domain(bo, bp->domain);
+	if (bp->type == ttm_bo_type_kernel)
+		bo->tbo.priority = 2;
+	else if (!(bp->flags & AMDGPU_GEM_CREATE_DISCARDABLE))
+		bo->tbo.priority = 1;
+
+	if (!bp->destroy)
+		bp->destroy = &amdgpu_bo_destroy;
 
+	r = ttm_bo_init_reserved(&adev->mman.bdev, &bo->tbo, bp->type,
+				 &bo->placement, page_align, &ctx,  NULL,
+				 bp->resv, bp->destroy);
 	if (unlikely(r != 0))
 		return r;
 
-	if (kernel)
-		bo->tbo.priority = 1;
+	if (!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
+	    amdgpu_res_cpu_visible(adev, bo->tbo.resource))
+		amdgpu_cs_report_moved_bytes(adev, ctx.bytes_moved,
+					     ctx.bytes_moved);
+	else
+		amdgpu_cs_report_moved_bytes(adev, ctx.bytes_moved, 0);
 
-	if (flags & AMDGPU_GEM_CREATE_VRAM_CLEARED &&
-	    bo->tbo.mem.placement & TTM_PL_FLAG_VRAM) {
+	if (bp->flags & AMDGPU_GEM_CREATE_VRAM_CLEARED &&
+	    bo->tbo.resource->mem_type == TTM_PL_VRAM) {
 		struct dma_fence *fence;
 
-		r = amdgpu_fill_buffer(bo, init_value, bo->tbo.resv, &fence);
+		r = amdgpu_ttm_clear_buffer(bo, bo->tbo.base.resv, &fence);
 		if (unlikely(r))
 			goto fail_unreserve;
 
-		amdgpu_bo_fence(bo, fence, false);
-		dma_fence_put(bo->tbo.moving);
-		bo->tbo.moving = dma_fence_get(fence);
+		dma_resv_add_fence(bo->tbo.base.resv, fence,
+				   DMA_RESV_USAGE_KERNEL);
 		dma_fence_put(fence);
 	}
-	if (!resv)
+	if (!bp->resv)
 		amdgpu_bo_unreserve(bo);
 	*bo_ptr = bo;
 
 	trace_amdgpu_bo_create(bo);
 
 	/* Treat CPU_ACCESS_REQUIRED only as a hint if given by UMD */
-	if (type == ttm_bo_type_device)
+	if (bp->type == ttm_bo_type_device)
 		bo->flags &= ~AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
 
 	return 0;
 
 fail_unreserve:
-	if (!resv)
-		ww_mutex_unlock(&bo->tbo.resv->lock);
+	if (!bp->resv)
+		dma_resv_unlock(bo->tbo.base.resv);
 	amdgpu_bo_unref(&bo);
 	return r;
 }
 
-static int amdgpu_bo_create_shadow(struct amdgpu_device *adev,
-				   unsigned long size, int byte_align,
-				   struct amdgpu_bo *bo)
-{
-	struct ttm_placement placement = {0};
-	struct ttm_place placements[AMDGPU_GEM_DOMAIN_MAX + 1];
-	int r;
-
-	if (bo->shadow)
-		return 0;
-
-	memset(&placements, 0, sizeof(placements));
-	amdgpu_ttm_placement_init(adev, &placement, placements,
-				  AMDGPU_GEM_DOMAIN_GTT,
-				  AMDGPU_GEM_CREATE_CPU_GTT_USWC |
-				  AMDGPU_GEM_CREATE_SHADOW);
-
-	r = amdgpu_bo_create_restricted(adev, size, byte_align, true,
-					AMDGPU_GEM_DOMAIN_GTT,
-					AMDGPU_GEM_CREATE_CPU_GTT_USWC |
-					AMDGPU_GEM_CREATE_SHADOW,
-					NULL, &placement,
-					bo->tbo.resv,
-					0,
-					&bo->shadow);
-	if (!r) {
-		bo->shadow->parent = amdgpu_bo_ref(bo);
-		mutex_lock(&adev->shadow_list_lock);
-		list_add_tail(&bo->shadow_list, &adev->shadow_list);
-		mutex_unlock(&adev->shadow_list_lock);
-	}
-
-	return r;
-}
-
-/* init_value will only take effect when flags contains
- * AMDGPU_GEM_CREATE_VRAM_CLEARED.
+/**
+ * amdgpu_bo_create_user - create an &amdgpu_bo_user buffer object
+ * @adev: amdgpu device object
+ * @bp: parameters to be used for the buffer object
+ * @ubo_ptr: pointer to the buffer object pointer
+ *
+ * Create a BO to be used by user application;
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
  */
-int amdgpu_bo_create(struct amdgpu_device *adev,
-		     unsigned long size, int byte_align,
-		     bool kernel, u32 domain, u64 flags,
-		     struct sg_table *sg,
-		     struct reservation_object *resv,
-		     uint64_t init_value,
-		     struct amdgpu_bo **bo_ptr)
-{
-	struct ttm_placement placement = {0};
-	struct ttm_place placements[AMDGPU_GEM_DOMAIN_MAX + 1];
-	uint64_t parent_flags = flags & ~AMDGPU_GEM_CREATE_SHADOW;
-	int r;
-
-	memset(&placements, 0, sizeof(placements));
-	amdgpu_ttm_placement_init(adev, &placement, placements,
-				  domain, parent_flags);
-
-	r = amdgpu_bo_create_restricted(adev, size, byte_align, kernel, domain,
-					parent_flags, sg, &placement, resv,
-					init_value, bo_ptr);
-	if (r)
-		return r;
-
-	if ((flags & AMDGPU_GEM_CREATE_SHADOW) && amdgpu_need_backup(adev)) {
-		if (!resv)
-			WARN_ON(reservation_object_lock((*bo_ptr)->tbo.resv,
-							NULL));
-
-		r = amdgpu_bo_create_shadow(adev, size, byte_align, (*bo_ptr));
-
-		if (!resv)
-			reservation_object_unlock((*bo_ptr)->tbo.resv);
-
-		if (r)
-			amdgpu_bo_unref(bo_ptr);
-	}
-
-	return r;
-}
-
-int amdgpu_bo_backup_to_shadow(struct amdgpu_device *adev,
-			       struct amdgpu_ring *ring,
-			       struct amdgpu_bo *bo,
-			       struct reservation_object *resv,
-			       struct dma_fence **fence,
-			       bool direct)
 
+int amdgpu_bo_create_user(struct amdgpu_device *adev,
+			  struct amdgpu_bo_param *bp,
+			  struct amdgpu_bo_user **ubo_ptr)
 {
-	struct amdgpu_bo *shadow = bo->shadow;
-	uint64_t bo_addr, shadow_addr;
+	struct amdgpu_bo *bo_ptr;
 	int r;
 
-	if (!shadow)
-		return -EINVAL;
-
-	bo_addr = amdgpu_bo_gpu_offset(bo);
-	shadow_addr = amdgpu_bo_gpu_offset(bo->shadow);
-
-	r = reservation_object_reserve_shared(bo->tbo.resv);
+	bp->bo_ptr_size = sizeof(struct amdgpu_bo_user);
+	bp->destroy = &amdgpu_bo_user_destroy;
+	r = amdgpu_bo_create(adev, bp, &bo_ptr);
 	if (r)
-		goto err;
-
-	r = amdgpu_copy_buffer(ring, bo_addr, shadow_addr,
-			       amdgpu_bo_size(bo), resv, fence,
-			       direct, false);
-	if (!r)
-		amdgpu_bo_fence(bo, *fence, true);
-
-err:
-	return r;
-}
-
-int amdgpu_bo_validate(struct amdgpu_bo *bo)
-{
-	uint32_t domain;
-	int r;
-
-	if (bo->pin_count)
-		return 0;
-
-	domain = bo->preferred_domains;
-
-retry:
-	amdgpu_ttm_placement_from_domain(bo, domain);
-	r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false);
-	if (unlikely(r == -ENOMEM) && domain != bo->allowed_domains) {
-		domain = bo->allowed_domains;
-		goto retry;
-	}
+		return r;
 
+	*ubo_ptr = to_amdgpu_bo_user(bo_ptr);
 	return r;
 }
 
-int amdgpu_bo_restore_from_shadow(struct amdgpu_device *adev,
-				  struct amdgpu_ring *ring,
-				  struct amdgpu_bo *bo,
-				  struct reservation_object *resv,
-				  struct dma_fence **fence,
-				  bool direct)
+/**
+ * amdgpu_bo_create_vm - create an &amdgpu_bo_vm buffer object
+ * @adev: amdgpu device object
+ * @bp: parameters to be used for the buffer object
+ * @vmbo_ptr: pointer to the buffer object pointer
+ *
+ * Create a BO to be for GPUVM.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
 
+int amdgpu_bo_create_vm(struct amdgpu_device *adev,
+			struct amdgpu_bo_param *bp,
+			struct amdgpu_bo_vm **vmbo_ptr)
 {
-	struct amdgpu_bo *shadow = bo->shadow;
-	uint64_t bo_addr, shadow_addr;
+	struct amdgpu_bo *bo_ptr;
 	int r;
 
-	if (!shadow)
-		return -EINVAL;
-
-	bo_addr = amdgpu_bo_gpu_offset(bo);
-	shadow_addr = amdgpu_bo_gpu_offset(bo->shadow);
-
-	r = reservation_object_reserve_shared(bo->tbo.resv);
+	/* bo_ptr_size will be determined by the caller and it depends on
+	 * num of amdgpu_vm_pt entries.
+	 */
+	BUG_ON(bp->bo_ptr_size < sizeof(struct amdgpu_bo_vm));
+	r = amdgpu_bo_create(adev, bp, &bo_ptr);
 	if (r)
-		goto err;
-
-	r = amdgpu_copy_buffer(ring, shadow_addr, bo_addr,
-			       amdgpu_bo_size(bo), resv, fence,
-			       direct, false);
-	if (!r)
-		amdgpu_bo_fence(bo, *fence, true);
+		return r;
 
-err:
+	*vmbo_ptr = to_amdgpu_bo_vm(bo_ptr);
 	return r;
 }
 
+/**
+ * amdgpu_bo_kmap - map an &amdgpu_bo buffer object
+ * @bo: &amdgpu_bo buffer object to be mapped
+ * @ptr: kernel virtual address to be returned
+ *
+ * Calls ttm_bo_kmap() to set up the kernel virtual mapping; calls
+ * amdgpu_bo_kptr() to get the kernel virtual address.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
 int amdgpu_bo_kmap(struct amdgpu_bo *bo, void **ptr)
 {
 	void *kptr;
@@ -609,6 +831,11 @@ int amdgpu_bo_kmap(struct amdgpu_bo *bo, void **ptr)
 	if (bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS)
 		return -EPERM;
 
+	r = dma_resv_wait_timeout(bo->tbo.base.resv, DMA_RESV_USAGE_KERNEL,
+				  false, MAX_SCHEDULE_TIMEOUT);
+	if (r < 0)
+		return r;
+
 	kptr = amdgpu_bo_kptr(bo);
 	if (kptr) {
 		if (ptr)
@@ -616,12 +843,7 @@ int amdgpu_bo_kmap(struct amdgpu_bo *bo, void **ptr)
 		return 0;
 	}
 
-	r = reservation_object_wait_timeout_rcu(bo->tbo.resv, false, false,
-						MAX_SCHEDULE_TIMEOUT);
-	if (r < 0)
-		return r;
-
-	r = ttm_bo_kmap(&bo->tbo, 0, bo->tbo.num_pages, &bo->kmap);
+	r = ttm_bo_kmap(&bo->tbo, 0, PFN_UP(bo->tbo.base.size), &bo->kmap);
 	if (r)
 		return r;
 
@@ -631,6 +853,15 @@ int amdgpu_bo_kmap(struct amdgpu_bo *bo, void **ptr)
 	return 0;
 }
 
+/**
+ * amdgpu_bo_kptr - returns a kernel virtual address of the buffer object
+ * @bo: &amdgpu_bo buffer object
+ *
+ * Calls ttm_kmap_obj_virtual() to get the kernel virtual address
+ *
+ * Returns:
+ * the virtual address of a buffer object area.
+ */
 void *amdgpu_bo_kptr(struct amdgpu_bo *bo)
 {
 	bool is_iomem;
@@ -638,173 +869,174 @@ void *amdgpu_bo_kptr(struct amdgpu_bo *bo)
 	return ttm_kmap_obj_virtual(&bo->kmap, &is_iomem);
 }
 
+/**
+ * amdgpu_bo_kunmap - unmap an &amdgpu_bo buffer object
+ * @bo: &amdgpu_bo buffer object to be unmapped
+ *
+ * Unmaps a kernel map set up by amdgpu_bo_kmap().
+ */
 void amdgpu_bo_kunmap(struct amdgpu_bo *bo)
 {
 	if (bo->kmap.bo)
 		ttm_bo_kunmap(&bo->kmap);
 }
 
+/**
+ * amdgpu_bo_ref - reference an &amdgpu_bo buffer object
+ * @bo: &amdgpu_bo buffer object
+ *
+ * References the contained &ttm_buffer_object.
+ *
+ * Returns:
+ * a refcounted pointer to the &amdgpu_bo buffer object.
+ */
 struct amdgpu_bo *amdgpu_bo_ref(struct amdgpu_bo *bo)
 {
 	if (bo == NULL)
 		return NULL;
 
-	ttm_bo_reference(&bo->tbo);
+	drm_gem_object_get(&bo->tbo.base);
 	return bo;
 }
 
+/**
+ * amdgpu_bo_unref - unreference an &amdgpu_bo buffer object
+ * @bo: &amdgpu_bo buffer object
+ *
+ * Unreferences the contained &ttm_buffer_object and clear the pointer
+ */
 void amdgpu_bo_unref(struct amdgpu_bo **bo)
 {
-	struct ttm_buffer_object *tbo;
-
 	if ((*bo) == NULL)
 		return;
 
-	tbo = &((*bo)->tbo);
-	ttm_bo_unref(&tbo);
-	if (tbo == NULL)
-		*bo = NULL;
+	drm_gem_object_put(&(*bo)->tbo.base);
+	*bo = NULL;
 }
 
-int amdgpu_bo_pin_restricted(struct amdgpu_bo *bo, u32 domain,
-			     u64 min_offset, u64 max_offset,
-			     u64 *gpu_addr)
+/**
+ * amdgpu_bo_pin - pin an &amdgpu_bo buffer object
+ * @bo: &amdgpu_bo buffer object to be pinned
+ * @domain: domain to be pinned to
+ *
+ * Pins the buffer object according to requested domain. If the memory is
+ * unbound gart memory, binds the pages into gart table. Adjusts pin_count and
+ * pin_size accordingly.
+ *
+ * Pinning means to lock pages in memory along with keeping them at a fixed
+ * offset. It is required when a buffer can not be moved, for example, when
+ * a display buffer is being scanned out.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_bo_pin(struct amdgpu_bo *bo, u32 domain)
 {
 	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	struct ttm_operation_ctx ctx = { false, false };
 	int r, i;
-	unsigned fpfn, lpfn;
 
 	if (amdgpu_ttm_tt_get_usermm(bo->tbo.ttm))
 		return -EPERM;
 
-	if (WARN_ON_ONCE(min_offset > max_offset))
-		return -EINVAL;
+	/* Check domain to be pinned to against preferred domains */
+	if (bo->preferred_domains & domain)
+		domain = bo->preferred_domains & domain;
 
 	/* A shared bo cannot be migrated to VRAM */
-	if (bo->prime_shared_count && (domain == AMDGPU_GEM_DOMAIN_VRAM))
-		return -EINVAL;
+	if (drm_gem_is_imported(&bo->tbo.base)) {
+		if (domain & AMDGPU_GEM_DOMAIN_GTT)
+			domain = AMDGPU_GEM_DOMAIN_GTT;
+		else
+			return -EINVAL;
+	}
 
-	if (bo->pin_count) {
-		uint32_t mem_type = bo->tbo.mem.mem_type;
+	if (bo->tbo.pin_count) {
+		uint32_t mem_type = bo->tbo.resource->mem_type;
+		uint32_t mem_flags = bo->tbo.resource->placement;
 
-		if (domain != amdgpu_mem_type_to_domain(mem_type))
+		if (!(domain & amdgpu_mem_type_to_domain(mem_type)))
 			return -EINVAL;
 
-		bo->pin_count++;
-		if (gpu_addr)
-			*gpu_addr = amdgpu_bo_gpu_offset(bo);
-
-		if (max_offset != 0) {
-			u64 domain_start = bo->tbo.bdev->man[mem_type].gpu_offset;
-			WARN_ON_ONCE(max_offset <
-				     (amdgpu_bo_gpu_offset(bo) - domain_start));
-		}
+		if ((mem_type == TTM_PL_VRAM) &&
+		    (bo->flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS) &&
+		    !(mem_flags & TTM_PL_FLAG_CONTIGUOUS))
+			return -EINVAL;
 
+		ttm_bo_pin(&bo->tbo);
 		return 0;
 	}
 
-	bo->flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
-	amdgpu_ttm_placement_from_domain(bo, domain);
+	/* This assumes only APU display buffers are pinned with (VRAM|GTT).
+	 * See function amdgpu_display_supported_domains()
+	 */
+	domain = amdgpu_bo_get_preferred_domain(adev, domain);
+
+	if (drm_gem_is_imported(&bo->tbo.base))
+		dma_buf_pin(bo->tbo.base.import_attach);
+
+	/* force to pin into visible video ram */
+	if (!(bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS))
+		bo->flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
+	amdgpu_bo_placement_from_domain(bo, domain);
 	for (i = 0; i < bo->placement.num_placement; i++) {
-		/* force to pin into visible video ram */
-		if ((bo->placements[i].flags & TTM_PL_FLAG_VRAM) &&
-		    !(bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS) &&
-		    (!max_offset || max_offset >
-		     adev->mc.visible_vram_size)) {
-			if (WARN_ON_ONCE(min_offset >
-					 adev->mc.visible_vram_size))
-				return -EINVAL;
-			fpfn = min_offset >> PAGE_SHIFT;
-			lpfn = adev->mc.visible_vram_size >> PAGE_SHIFT;
-		} else {
-			fpfn = min_offset >> PAGE_SHIFT;
-			lpfn = max_offset >> PAGE_SHIFT;
-		}
-		if (fpfn > bo->placements[i].fpfn)
-			bo->placements[i].fpfn = fpfn;
-		if (!bo->placements[i].lpfn ||
-		    (lpfn && lpfn < bo->placements[i].lpfn))
-			bo->placements[i].lpfn = lpfn;
-		bo->placements[i].flags |= TTM_PL_FLAG_NO_EVICT;
+		if (bo->flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS &&
+		    bo->placements[i].mem_type == TTM_PL_VRAM)
+			bo->placements[i].flags |= TTM_PL_FLAG_CONTIGUOUS;
 	}
 
-	r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false);
+	r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
 	if (unlikely(r)) {
 		dev_err(adev->dev, "%p pin failed\n", bo);
 		goto error;
 	}
 
-	bo->pin_count = 1;
-	if (gpu_addr != NULL) {
-		r = amdgpu_ttm_bind(&bo->tbo, &bo->tbo.mem);
-		if (unlikely(r)) {
-			dev_err(adev->dev, "%p bind failed\n", bo);
-			goto error;
-		}
-		*gpu_addr = amdgpu_bo_gpu_offset(bo);
-	}
-	if (domain == AMDGPU_GEM_DOMAIN_VRAM) {
-		adev->vram_pin_size += amdgpu_bo_size(bo);
-		if (bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS)
-			adev->invisible_pin_size += amdgpu_bo_size(bo);
-	} else if (domain == AMDGPU_GEM_DOMAIN_GTT) {
-		adev->gart_pin_size += amdgpu_bo_size(bo);
+	ttm_bo_pin(&bo->tbo);
+
+	if (bo->tbo.resource->mem_type == TTM_PL_VRAM) {
+		atomic64_add(amdgpu_bo_size(bo), &adev->vram_pin_size);
+		atomic64_add(amdgpu_vram_mgr_bo_visible_size(bo),
+			     &adev->visible_pin_size);
+	} else if (bo->tbo.resource->mem_type == TTM_PL_TT) {
+		atomic64_add(amdgpu_bo_size(bo), &adev->gart_pin_size);
 	}
 
 error:
 	return r;
 }
 
-int amdgpu_bo_pin(struct amdgpu_bo *bo, u32 domain, u64 *gpu_addr)
-{
-	return amdgpu_bo_pin_restricted(bo, domain, 0, 0, gpu_addr);
-}
-
-int amdgpu_bo_unpin(struct amdgpu_bo *bo)
+/**
+ * amdgpu_bo_unpin - unpin an &amdgpu_bo buffer object
+ * @bo: &amdgpu_bo buffer object to be unpinned
+ *
+ * Decreases the pin_count, and clears the flags if pin_count reaches 0.
+ * Changes placement and pin size accordingly.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+void amdgpu_bo_unpin(struct amdgpu_bo *bo)
 {
 	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
-	int r, i;
-
-	if (!bo->pin_count) {
-		dev_warn(adev->dev, "%p unpin not necessary\n", bo);
-		return 0;
-	}
-	bo->pin_count--;
-	if (bo->pin_count)
-		return 0;
-	for (i = 0; i < bo->placement.num_placement; i++) {
-		bo->placements[i].lpfn = 0;
-		bo->placements[i].flags &= ~TTM_PL_FLAG_NO_EVICT;
-	}
-	r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false);
-	if (unlikely(r)) {
-		dev_err(adev->dev, "%p validate failed for unpin\n", bo);
-		goto error;
-	}
 
-	if (bo->tbo.mem.mem_type == TTM_PL_VRAM) {
-		adev->vram_pin_size -= amdgpu_bo_size(bo);
-		if (bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS)
-			adev->invisible_pin_size -= amdgpu_bo_size(bo);
-	} else if (bo->tbo.mem.mem_type == TTM_PL_TT) {
-		adev->gart_pin_size -= amdgpu_bo_size(bo);
-	}
+	ttm_bo_unpin(&bo->tbo);
+	if (bo->tbo.pin_count)
+		return;
 
-error:
-	return r;
-}
+	if (drm_gem_is_imported(&bo->tbo.base))
+		dma_buf_unpin(bo->tbo.base.import_attach);
 
-int amdgpu_bo_evict_vram(struct amdgpu_device *adev)
-{
-	/* late 2.6.33 fix IGP hibernate - we need pm ops to do this correct */
-	if (0 && (adev->flags & AMD_IS_APU)) {
-		/* Useless to evict on IGP chips */
-		return 0;
+	if (bo->tbo.resource->mem_type == TTM_PL_VRAM) {
+		atomic64_sub(amdgpu_bo_size(bo), &adev->vram_pin_size);
+		atomic64_sub(amdgpu_vram_mgr_bo_visible_size(bo),
+			     &adev->visible_pin_size);
+	} else if (bo->tbo.resource->mem_type == TTM_PL_TT) {
+		atomic64_sub(amdgpu_bo_size(bo), &adev->gart_pin_size);
 	}
-	return ttm_bo_evict_mm(&adev->mman.bdev, TTM_PL_VRAM);
+
 }
 
-static const char *amdgpu_vram_names[] = {
+static const char * const amdgpu_vram_names[] = {
 	"UNKNOWN",
 	"GDDR1",
 	"DDR2",
@@ -812,69 +1044,143 @@ static const char *amdgpu_vram_names[] = {
 	"GDDR4",
 	"GDDR5",
 	"HBM",
-	"DDR3"
+	"DDR3",
+	"DDR4",
+	"GDDR6",
+	"DDR5",
+	"LPDDR4",
+	"LPDDR5",
+	"HBM3E"
 };
 
+/**
+ * amdgpu_bo_init - initialize memory manager
+ * @adev: amdgpu device object
+ *
+ * Calls amdgpu_ttm_init() to initialize amdgpu memory manager.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
 int amdgpu_bo_init(struct amdgpu_device *adev)
 {
-	/* reserve PAT memory space to WC for VRAM */
-	arch_io_reserve_memtype_wc(adev->mc.aper_base,
-				   adev->mc.aper_size);
+	/* On A+A platform, VRAM can be mapped as WB */
+	if (!adev->gmc.xgmi.connected_to_cpu && !adev->gmc.is_app_apu) {
+		/* reserve PAT memory space to WC for VRAM */
+		int r = arch_io_reserve_memtype_wc(adev->gmc.aper_base,
+				adev->gmc.aper_size);
+
+		if (r) {
+			DRM_ERROR("Unable to set WC memtype for the aperture base\n");
+			return r;
+		}
+
+		/* Add an MTRR for the VRAM */
+		adev->gmc.vram_mtrr = arch_phys_wc_add(adev->gmc.aper_base,
+				adev->gmc.aper_size);
+	}
 
-	/* Add an MTRR for the VRAM */
-	adev->mc.vram_mtrr = arch_phys_wc_add(adev->mc.aper_base,
-					      adev->mc.aper_size);
 	DRM_INFO("Detected VRAM RAM=%lluM, BAR=%lluM\n",
-		adev->mc.mc_vram_size >> 20,
-		(unsigned long long)adev->mc.aper_size >> 20);
+		 adev->gmc.mc_vram_size >> 20,
+		 (unsigned long long)adev->gmc.aper_size >> 20);
 	DRM_INFO("RAM width %dbits %s\n",
-		 adev->mc.vram_width, amdgpu_vram_names[adev->mc.vram_type]);
+		 adev->gmc.vram_width, amdgpu_vram_names[adev->gmc.vram_type]);
 	return amdgpu_ttm_init(adev);
 }
 
+/**
+ * amdgpu_bo_fini - tear down memory manager
+ * @adev: amdgpu device object
+ *
+ * Reverses amdgpu_bo_init() to tear down memory manager.
+ */
 void amdgpu_bo_fini(struct amdgpu_device *adev)
 {
+	int idx;
+
 	amdgpu_ttm_fini(adev);
-	arch_phys_wc_del(adev->mc.vram_mtrr);
-	arch_io_free_memtype_wc(adev->mc.aper_base, adev->mc.aper_size);
-}
 
-int amdgpu_bo_fbdev_mmap(struct amdgpu_bo *bo,
-			     struct vm_area_struct *vma)
-{
-	return ttm_fbdev_mmap(vma, &bo->tbo);
+	if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+		if (!adev->gmc.xgmi.connected_to_cpu && !adev->gmc.is_app_apu) {
+			arch_phys_wc_del(adev->gmc.vram_mtrr);
+			arch_io_free_memtype_wc(adev->gmc.aper_base, adev->gmc.aper_size);
+		}
+		drm_dev_exit(idx);
+	}
 }
 
+/**
+ * amdgpu_bo_set_tiling_flags - set tiling flags
+ * @bo: &amdgpu_bo buffer object
+ * @tiling_flags: new flags
+ *
+ * Sets buffer object's tiling flags with the new one. Used by GEM ioctl or
+ * kernel driver to set the tiling flags on a buffer.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
 int amdgpu_bo_set_tiling_flags(struct amdgpu_bo *bo, u64 tiling_flags)
 {
 	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	struct amdgpu_bo_user *ubo;
 
+	BUG_ON(bo->tbo.type == ttm_bo_type_kernel);
 	if (adev->family <= AMDGPU_FAMILY_CZ &&
 	    AMDGPU_TILING_GET(tiling_flags, TILE_SPLIT) > 6)
 		return -EINVAL;
 
-	bo->tiling_flags = tiling_flags;
+	ubo = to_amdgpu_bo_user(bo);
+	ubo->tiling_flags = tiling_flags;
 	return 0;
 }
 
+/**
+ * amdgpu_bo_get_tiling_flags - get tiling flags
+ * @bo: &amdgpu_bo buffer object
+ * @tiling_flags: returned flags
+ *
+ * Gets buffer object's tiling flags. Used by GEM ioctl or kernel driver to
+ * set the tiling flags on a buffer.
+ */
 void amdgpu_bo_get_tiling_flags(struct amdgpu_bo *bo, u64 *tiling_flags)
 {
-	lockdep_assert_held(&bo->tbo.resv->lock.base);
+	struct amdgpu_bo_user *ubo;
+
+	BUG_ON(bo->tbo.type == ttm_bo_type_kernel);
+	dma_resv_assert_held(bo->tbo.base.resv);
+	ubo = to_amdgpu_bo_user(bo);
 
 	if (tiling_flags)
-		*tiling_flags = bo->tiling_flags;
+		*tiling_flags = ubo->tiling_flags;
 }
 
-int amdgpu_bo_set_metadata (struct amdgpu_bo *bo, void *metadata,
-			    uint32_t metadata_size, uint64_t flags)
+/**
+ * amdgpu_bo_set_metadata - set metadata
+ * @bo: &amdgpu_bo buffer object
+ * @metadata: new metadata
+ * @metadata_size: size of the new metadata
+ * @flags: flags of the new metadata
+ *
+ * Sets buffer object's metadata, its size and flags.
+ * Used via GEM ioctl.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_bo_set_metadata(struct amdgpu_bo *bo, void *metadata,
+			   u32 metadata_size, uint64_t flags)
 {
+	struct amdgpu_bo_user *ubo;
 	void *buffer;
 
+	BUG_ON(bo->tbo.type == ttm_bo_type_kernel);
+	ubo = to_amdgpu_bo_user(bo);
 	if (!metadata_size) {
-		if (bo->metadata_size) {
-			kfree(bo->metadata);
-			bo->metadata = NULL;
-			bo->metadata_size = 0;
+		if (ubo->metadata_size) {
+			kfree(ubo->metadata);
+			ubo->metadata = NULL;
+			ubo->metadata_size = 0;
 		}
 		return 0;
 	}
@@ -886,111 +1192,196 @@ int amdgpu_bo_set_metadata (struct amdgpu_bo *bo, void *metadata,
 	if (buffer == NULL)
 		return -ENOMEM;
 
-	kfree(bo->metadata);
-	bo->metadata_flags = flags;
-	bo->metadata = buffer;
-	bo->metadata_size = metadata_size;
+	kfree(ubo->metadata);
+	ubo->metadata_flags = flags;
+	ubo->metadata = buffer;
+	ubo->metadata_size = metadata_size;
 
 	return 0;
 }
 
+/**
+ * amdgpu_bo_get_metadata - get metadata
+ * @bo: &amdgpu_bo buffer object
+ * @buffer: returned metadata
+ * @buffer_size: size of the buffer
+ * @metadata_size: size of the returned metadata
+ * @flags: flags of the returned metadata
+ *
+ * Gets buffer object's metadata, its size and flags. buffer_size shall not be
+ * less than metadata_size.
+ * Used via GEM ioctl.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
 int amdgpu_bo_get_metadata(struct amdgpu_bo *bo, void *buffer,
 			   size_t buffer_size, uint32_t *metadata_size,
 			   uint64_t *flags)
 {
+	struct amdgpu_bo_user *ubo;
+
 	if (!buffer && !metadata_size)
 		return -EINVAL;
 
+	BUG_ON(bo->tbo.type == ttm_bo_type_kernel);
+	ubo = to_amdgpu_bo_user(bo);
+	if (metadata_size)
+		*metadata_size = ubo->metadata_size;
+
 	if (buffer) {
-		if (buffer_size < bo->metadata_size)
+		if (buffer_size < ubo->metadata_size)
 			return -EINVAL;
 
-		if (bo->metadata_size)
-			memcpy(buffer, bo->metadata, bo->metadata_size);
+		if (ubo->metadata_size)
+			memcpy(buffer, ubo->metadata, ubo->metadata_size);
 	}
 
-	if (metadata_size)
-		*metadata_size = bo->metadata_size;
 	if (flags)
-		*flags = bo->metadata_flags;
+		*flags = ubo->metadata_flags;
 
 	return 0;
 }
 
+/**
+ * amdgpu_bo_move_notify - notification about a memory move
+ * @bo: pointer to a buffer object
+ * @evict: if this move is evicting the buffer from the graphics address space
+ * @new_mem: new resource for backing the BO
+ *
+ * Marks the corresponding &amdgpu_bo buffer object as invalid, also performs
+ * bookkeeping.
+ * TTM driver callback which is called when ttm moves a buffer.
+ */
 void amdgpu_bo_move_notify(struct ttm_buffer_object *bo,
 			   bool evict,
-			   struct ttm_mem_reg *new_mem)
+			   struct ttm_resource *new_mem)
 {
-	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev);
+	struct ttm_resource *old_mem = bo->resource;
 	struct amdgpu_bo *abo;
-	struct ttm_mem_reg *old_mem = &bo->mem;
 
-	if (!amdgpu_ttm_bo_is_amdgpu_bo(bo))
+	if (!amdgpu_bo_is_amdgpu_bo(bo))
 		return;
 
-	abo = container_of(bo, struct amdgpu_bo, tbo);
-	amdgpu_vm_bo_invalidate(adev, abo);
+	abo = ttm_to_amdgpu_bo(bo);
+	amdgpu_vm_bo_move(abo, new_mem, evict);
 
 	amdgpu_bo_kunmap(abo);
 
-	/* remember the eviction */
-	if (evict)
-		atomic64_inc(&adev->num_evictions);
-
-	/* update statistics */
-	if (!new_mem)
-		return;
+	if (abo->tbo.base.dma_buf && !drm_gem_is_imported(&abo->tbo.base) &&
+	    old_mem && old_mem->mem_type != TTM_PL_SYSTEM)
+		dma_buf_move_notify(abo->tbo.base.dma_buf);
 
 	/* move_notify is called before move happens */
-	trace_amdgpu_ttm_bo_move(abo, new_mem->mem_type, old_mem->mem_type);
+	trace_amdgpu_bo_move(abo, new_mem ? new_mem->mem_type : -1,
+			     old_mem ? old_mem->mem_type : -1);
 }
 
-int amdgpu_bo_fault_reserve_notify(struct ttm_buffer_object *bo)
+/**
+ * amdgpu_bo_release_notify - notification about a BO being released
+ * @bo: pointer to a buffer object
+ *
+ * Wipes VRAM buffers whose contents should not be leaked before the
+ * memory is released.
+ */
+void amdgpu_bo_release_notify(struct ttm_buffer_object *bo)
 {
 	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev);
+	struct dma_fence *fence = NULL;
 	struct amdgpu_bo *abo;
-	unsigned long offset, size;
 	int r;
 
-	if (!amdgpu_ttm_bo_is_amdgpu_bo(bo))
-		return 0;
+	if (!amdgpu_bo_is_amdgpu_bo(bo))
+		return;
+
+	abo = ttm_to_amdgpu_bo(bo);
+
+	WARN_ON(abo->vm_bo);
+
+	if (abo->kfd_bo)
+		amdgpu_amdkfd_release_notify(abo);
+
+	/*
+	 * We lock the private dma_resv object here and since the BO is about to
+	 * be released nobody else should have a pointer to it.
+	 * So when this locking here fails something is wrong with the reference
+	 * counting.
+	 */
+	if (WARN_ON_ONCE(!dma_resv_trylock(&bo->base._resv)))
+		return;
 
-	abo = container_of(bo, struct amdgpu_bo, tbo);
+	amdgpu_amdkfd_remove_all_eviction_fences(abo);
+
+	if (!bo->resource || bo->resource->mem_type != TTM_PL_VRAM ||
+	    !(abo->flags & AMDGPU_GEM_CREATE_VRAM_WIPE_ON_RELEASE) ||
+	    adev->in_suspend || drm_dev_is_unplugged(adev_to_drm(adev)))
+		goto out;
+
+	r = dma_resv_reserve_fences(&bo->base._resv, 1);
+	if (r)
+		goto out;
+
+	r = amdgpu_fill_buffer(abo, 0, &bo->base._resv, &fence, true,
+			       AMDGPU_KERNEL_JOB_ID_CLEAR_ON_RELEASE);
+	if (WARN_ON(r))
+		goto out;
+
+	amdgpu_vram_mgr_set_cleared(bo->resource);
+	dma_resv_add_fence(&bo->base._resv, fence, DMA_RESV_USAGE_KERNEL);
+	dma_fence_put(fence);
+
+out:
+	dma_resv_unlock(&bo->base._resv);
+}
+
+/**
+ * amdgpu_bo_fault_reserve_notify - notification about a memory fault
+ * @bo: pointer to a buffer object
+ *
+ * Notifies the driver we are taking a fault on this BO and have reserved it,
+ * also performs bookkeeping.
+ * TTM driver callback for dealing with vm faults.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+vm_fault_t amdgpu_bo_fault_reserve_notify(struct ttm_buffer_object *bo)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev);
+	struct ttm_operation_ctx ctx = { false, false };
+	struct amdgpu_bo *abo = ttm_to_amdgpu_bo(bo);
+	int r;
 
 	/* Remember that this BO was accessed by the CPU */
 	abo->flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
 
-	if (bo->mem.mem_type != TTM_PL_VRAM)
-		return 0;
-
-	size = bo->mem.num_pages << PAGE_SHIFT;
-	offset = bo->mem.start << PAGE_SHIFT;
-	if ((offset + size) <= adev->mc.visible_vram_size)
+	if (amdgpu_res_cpu_visible(adev, bo->resource))
 		return 0;
 
 	/* Can't move a pinned BO to visible VRAM */
-	if (abo->pin_count > 0)
-		return -EINVAL;
+	if (abo->tbo.pin_count > 0)
+		return VM_FAULT_SIGBUS;
 
 	/* hurrah the memory is not visible ! */
 	atomic64_inc(&adev->num_vram_cpu_page_faults);
-	amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_VRAM |
-					 AMDGPU_GEM_DOMAIN_GTT);
+	amdgpu_bo_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_VRAM |
+					AMDGPU_GEM_DOMAIN_GTT);
 
 	/* Avoid costly evictions; only set GTT as a busy placement */
-	abo->placement.num_busy_placement = 1;
-	abo->placement.busy_placement = &abo->placements[1];
+	abo->placements[0].flags |= TTM_PL_FLAG_DESIRED;
 
-	r = ttm_bo_validate(bo, &abo->placement, false, false);
-	if (unlikely(r != 0))
-		return r;
+	r = ttm_bo_validate(bo, &abo->placement, &ctx);
+	if (unlikely(r == -EBUSY || r == -ERESTARTSYS))
+		return VM_FAULT_NOPAGE;
+	else if (unlikely(r))
+		return VM_FAULT_SIGBUS;
 
-	offset = bo->mem.start << PAGE_SHIFT;
 	/* this should never happen */
-	if (bo->mem.mem_type == TTM_PL_VRAM &&
-	    (offset + size) > adev->mc.visible_vram_size)
-		return -EINVAL;
+	if (bo->resource->mem_type == TTM_PL_VRAM &&
+	    !amdgpu_res_cpu_visible(adev, bo->resource))
+		return VM_FAULT_SIGBUS;
 
+	ttm_bo_move_to_lru_tail_unlocked(bo);
 	return 0;
 }
 
@@ -1005,33 +1396,293 @@ int amdgpu_bo_fault_reserve_notify(struct ttm_buffer_object *bo)
 void amdgpu_bo_fence(struct amdgpu_bo *bo, struct dma_fence *fence,
 		     bool shared)
 {
-	struct reservation_object *resv = bo->tbo.resv;
+	struct dma_resv *resv = bo->tbo.base.resv;
+	int r;
 
-	if (shared)
-		reservation_object_add_shared_fence(resv, fence);
-	else
-		reservation_object_add_excl_fence(resv, fence);
+	r = dma_resv_reserve_fences(resv, 1);
+	if (r) {
+		/* As last resort on OOM we block for the fence */
+		dma_fence_wait(fence, false);
+		return;
+	}
+
+	dma_resv_add_fence(resv, fence, shared ? DMA_RESV_USAGE_READ :
+			   DMA_RESV_USAGE_WRITE);
+}
+
+/**
+ * amdgpu_bo_sync_wait_resv - Wait for BO reservation fences
+ *
+ * @adev: amdgpu device pointer
+ * @resv: reservation object to sync to
+ * @sync_mode: synchronization mode
+ * @owner: fence owner
+ * @intr: Whether the wait is interruptible
+ *
+ * Extract the fences from the reservation object and waits for them to finish.
+ *
+ * Returns:
+ * 0 on success, errno otherwise.
+ */
+int amdgpu_bo_sync_wait_resv(struct amdgpu_device *adev, struct dma_resv *resv,
+			     enum amdgpu_sync_mode sync_mode, void *owner,
+			     bool intr)
+{
+	struct amdgpu_sync sync;
+	int r;
+
+	amdgpu_sync_create(&sync);
+	amdgpu_sync_resv(adev, &sync, resv, sync_mode, owner);
+	r = amdgpu_sync_wait(&sync, intr);
+	amdgpu_sync_free(&sync);
+	return r;
+}
+
+/**
+ * amdgpu_bo_sync_wait - Wrapper for amdgpu_bo_sync_wait_resv
+ * @bo: buffer object to wait for
+ * @owner: fence owner
+ * @intr: Whether the wait is interruptible
+ *
+ * Wrapper to wait for fences in a BO.
+ * Returns:
+ * 0 on success, errno otherwise.
+ */
+int amdgpu_bo_sync_wait(struct amdgpu_bo *bo, void *owner, bool intr)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+
+	return amdgpu_bo_sync_wait_resv(adev, bo->tbo.base.resv,
+					AMDGPU_SYNC_NE_OWNER, owner, intr);
 }
 
 /**
  * amdgpu_bo_gpu_offset - return GPU offset of bo
  * @bo:	amdgpu object for which we query the offset
  *
- * Returns current GPU offset of the object.
- *
  * Note: object should either be pinned or reserved when calling this
  * function, it might be useful to add check for this for debugging.
+ *
+ * Returns:
+ * current GPU offset of the object.
  */
 u64 amdgpu_bo_gpu_offset(struct amdgpu_bo *bo)
 {
-	WARN_ON_ONCE(bo->tbo.mem.mem_type == TTM_PL_SYSTEM);
-	WARN_ON_ONCE(bo->tbo.mem.mem_type == TTM_PL_TT &&
-		     !amdgpu_ttm_is_bound(bo->tbo.ttm));
-	WARN_ON_ONCE(!ww_mutex_is_locked(&bo->tbo.resv->lock) &&
-		     !bo->pin_count);
-	WARN_ON_ONCE(bo->tbo.mem.start == AMDGPU_BO_INVALID_OFFSET);
-	WARN_ON_ONCE(bo->tbo.mem.mem_type == TTM_PL_VRAM &&
+	WARN_ON_ONCE(bo->tbo.resource->mem_type == TTM_PL_SYSTEM);
+	WARN_ON_ONCE(!dma_resv_is_locked(bo->tbo.base.resv) &&
+		     !bo->tbo.pin_count && bo->tbo.type != ttm_bo_type_kernel);
+	WARN_ON_ONCE(bo->tbo.resource->start == AMDGPU_BO_INVALID_OFFSET);
+	WARN_ON_ONCE(bo->tbo.resource->mem_type == TTM_PL_VRAM &&
 		     !(bo->flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS));
 
-	return bo->tbo.offset;
+	return amdgpu_bo_gpu_offset_no_check(bo);
+}
+
+/**
+ * amdgpu_bo_fb_aper_addr - return FB aperture GPU offset of the VRAM bo
+ * @bo:	amdgpu VRAM buffer object for which we query the offset
+ *
+ * Returns:
+ * current FB aperture GPU offset of the object.
+ */
+u64 amdgpu_bo_fb_aper_addr(struct amdgpu_bo *bo)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	uint64_t offset, fb_base;
+
+	WARN_ON_ONCE(bo->tbo.resource->mem_type != TTM_PL_VRAM);
+
+	fb_base = adev->gmc.fb_start;
+	fb_base += adev->gmc.xgmi.physical_node_id * adev->gmc.xgmi.node_segment_size;
+	offset = (bo->tbo.resource->start << PAGE_SHIFT) + fb_base;
+	return amdgpu_gmc_sign_extend(offset);
+}
+
+/**
+ * amdgpu_bo_gpu_offset_no_check - return GPU offset of bo
+ * @bo:	amdgpu object for which we query the offset
+ *
+ * Returns:
+ * current GPU offset of the object without raising warnings.
+ */
+u64 amdgpu_bo_gpu_offset_no_check(struct amdgpu_bo *bo)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	uint64_t offset = AMDGPU_BO_INVALID_OFFSET;
+
+	if (bo->tbo.resource->mem_type == TTM_PL_TT)
+		offset = amdgpu_gmc_agp_addr(&bo->tbo);
+
+	if (offset == AMDGPU_BO_INVALID_OFFSET)
+		offset = (bo->tbo.resource->start << PAGE_SHIFT) +
+			amdgpu_ttm_domain_start(adev, bo->tbo.resource->mem_type);
+
+	return amdgpu_gmc_sign_extend(offset);
 }
+
+/**
+ * amdgpu_bo_mem_stats_placement - bo placement for memory accounting
+ * @bo:	the buffer object we should look at
+ *
+ * BO can have multiple preferred placements, to avoid double counting we want
+ * to file it under a single placement for memory stats.
+ * Luckily, if we take the highest set bit in preferred_domains the result is
+ * quite sensible.
+ *
+ * Returns:
+ * Which of the placements should the BO be accounted under.
+ */
+uint32_t amdgpu_bo_mem_stats_placement(struct amdgpu_bo *bo)
+{
+	uint32_t domain = bo->preferred_domains & AMDGPU_GEM_DOMAIN_MASK;
+
+	if (!domain)
+		return TTM_PL_SYSTEM;
+
+	switch (rounddown_pow_of_two(domain)) {
+	case AMDGPU_GEM_DOMAIN_CPU:
+		return TTM_PL_SYSTEM;
+	case AMDGPU_GEM_DOMAIN_GTT:
+		return TTM_PL_TT;
+	case AMDGPU_GEM_DOMAIN_VRAM:
+		return TTM_PL_VRAM;
+	case AMDGPU_GEM_DOMAIN_GDS:
+		return AMDGPU_PL_GDS;
+	case AMDGPU_GEM_DOMAIN_GWS:
+		return AMDGPU_PL_GWS;
+	case AMDGPU_GEM_DOMAIN_OA:
+		return AMDGPU_PL_OA;
+	case AMDGPU_GEM_DOMAIN_DOORBELL:
+		return AMDGPU_PL_DOORBELL;
+	case AMDGPU_GEM_DOMAIN_MMIO_REMAP:
+		return AMDGPU_PL_MMIO_REMAP;
+	default:
+		return TTM_PL_SYSTEM;
+	}
+}
+
+/**
+ * amdgpu_bo_get_preferred_domain - get preferred domain
+ * @adev: amdgpu device object
+ * @domain: allowed :ref:`memory domains <amdgpu_memory_domains>`
+ *
+ * Returns:
+ * Which of the allowed domains is preferred for allocating the BO.
+ */
+uint32_t amdgpu_bo_get_preferred_domain(struct amdgpu_device *adev,
+					    uint32_t domain)
+{
+	if ((domain == (AMDGPU_GEM_DOMAIN_VRAM | AMDGPU_GEM_DOMAIN_GTT)) &&
+	    ((adev->asic_type == CHIP_CARRIZO) || (adev->asic_type == CHIP_STONEY))) {
+		domain = AMDGPU_GEM_DOMAIN_VRAM;
+		if (adev->gmc.real_vram_size <= AMDGPU_SG_THRESHOLD)
+			domain = AMDGPU_GEM_DOMAIN_GTT;
+	}
+	return domain;
+}
+
+#if defined(CONFIG_DEBUG_FS)
+#define amdgpu_bo_print_flag(m, bo, flag)		        \
+	do {							\
+		if (bo->flags & (AMDGPU_GEM_CREATE_ ## flag)) {	\
+			seq_printf((m), " " #flag);		\
+		}						\
+	} while (0)
+
+/**
+ * amdgpu_bo_print_info - print BO info in debugfs file
+ *
+ * @id: Index or Id of the BO
+ * @bo: Requested BO for printing info
+ * @m: debugfs file
+ *
+ * Print BO information in debugfs file
+ *
+ * Returns:
+ * Size of the BO in bytes.
+ */
+u64 amdgpu_bo_print_info(int id, struct amdgpu_bo *bo, struct seq_file *m)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	struct dma_buf_attachment *attachment;
+	struct dma_buf *dma_buf;
+	const char *placement;
+	unsigned int pin_count;
+	u64 size;
+
+	if (dma_resv_trylock(bo->tbo.base.resv)) {
+		if (!bo->tbo.resource) {
+			placement = "NONE";
+		} else {
+			switch (bo->tbo.resource->mem_type) {
+			case TTM_PL_VRAM:
+				if (amdgpu_res_cpu_visible(adev, bo->tbo.resource))
+					placement = "VRAM VISIBLE";
+				else
+					placement = "VRAM";
+				break;
+			case TTM_PL_TT:
+				placement = "GTT";
+				break;
+			case AMDGPU_PL_GDS:
+				placement = "GDS";
+				break;
+			case AMDGPU_PL_GWS:
+				placement = "GWS";
+				break;
+			case AMDGPU_PL_OA:
+				placement = "OA";
+				break;
+			case AMDGPU_PL_PREEMPT:
+				placement = "PREEMPTIBLE";
+				break;
+			case AMDGPU_PL_DOORBELL:
+				placement = "DOORBELL";
+				break;
+			case AMDGPU_PL_MMIO_REMAP:
+				placement = "MMIO REMAP";
+				break;
+			case TTM_PL_SYSTEM:
+			default:
+				placement = "CPU";
+				break;
+			}
+		}
+		dma_resv_unlock(bo->tbo.base.resv);
+	} else {
+		placement = "UNKNOWN";
+	}
+
+	size = amdgpu_bo_size(bo);
+	seq_printf(m, "\t\t0x%08x: %12lld byte %s",
+			id, size, placement);
+
+	pin_count = READ_ONCE(bo->tbo.pin_count);
+	if (pin_count)
+		seq_printf(m, " pin count %d", pin_count);
+
+	dma_buf = READ_ONCE(bo->tbo.base.dma_buf);
+	attachment = READ_ONCE(bo->tbo.base.import_attach);
+
+	if (attachment)
+		seq_printf(m, " imported from ino:%lu", file_inode(dma_buf->file)->i_ino);
+	else if (dma_buf)
+		seq_printf(m, " exported as ino:%lu", file_inode(dma_buf->file)->i_ino);
+
+	amdgpu_bo_print_flag(m, bo, CPU_ACCESS_REQUIRED);
+	amdgpu_bo_print_flag(m, bo, NO_CPU_ACCESS);
+	amdgpu_bo_print_flag(m, bo, CPU_GTT_USWC);
+	amdgpu_bo_print_flag(m, bo, VRAM_CLEARED);
+	amdgpu_bo_print_flag(m, bo, VRAM_CONTIGUOUS);
+	amdgpu_bo_print_flag(m, bo, VM_ALWAYS_VALID);
+	amdgpu_bo_print_flag(m, bo, EXPLICIT_SYNC);
+	/* Add the gem obj resv fence dump*/
+	if (dma_resv_trylock(bo->tbo.base.resv)) {
+		dma_resv_describe(bo->tbo.base.resv, m);
+		dma_resv_unlock(bo->tbo.base.resv);
+	}
+	seq_puts(m, "\n");
+
+	return size;
+}
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
index a288fa6d72c8..656b8a931dae 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
@@ -30,18 +30,46 @@
 
 #include <drm/amdgpu_drm.h>
 #include "amdgpu.h"
+#include "amdgpu_res_cursor.h"
+
+#ifdef CONFIG_MMU_NOTIFIER
+#include <linux/mmu_notifier.h>
+#endif
 
 #define AMDGPU_BO_INVALID_OFFSET	LONG_MAX
+#define AMDGPU_BO_MAX_PLACEMENTS	3
+
+/* BO flag to indicate a KFD userptr BO */
+#define AMDGPU_AMDKFD_CREATE_USERPTR_BO	(1ULL << 63)
+
+#define to_amdgpu_bo_user(abo) container_of((abo), struct amdgpu_bo_user, bo)
+#define to_amdgpu_bo_vm(abo) container_of((abo), struct amdgpu_bo_vm, bo)
+
+struct amdgpu_bo_param {
+	unsigned long			size;
+	int				byte_align;
+	u32				bo_ptr_size;
+	u32				domain;
+	u32				preferred_domain;
+	u64				flags;
+	enum ttm_bo_type		type;
+	bool				no_wait_gpu;
+	struct dma_resv			*resv;
+	void				(*destroy)(struct ttm_buffer_object *bo);
+	/* xcp partition number plus 1, 0 means any partition */
+	int8_t				xcp_id_plus1;
+};
 
 /* bo virtual addresses in a vm */
 struct amdgpu_bo_va_mapping {
+	struct amdgpu_bo_va		*bo_va;
 	struct list_head		list;
 	struct rb_node			rb;
 	uint64_t			start;
 	uint64_t			last;
 	uint64_t			__subtree_last;
 	uint64_t			offset;
-	uint64_t			flags;
+	uint32_t			flags;
 };
 
 /* User space allocated BO in a VM */
@@ -49,45 +77,73 @@ struct amdgpu_bo_va {
 	struct amdgpu_vm_bo_base	base;
 
 	/* protected by bo being reserved */
-	struct dma_fence	        *last_pt_update;
 	unsigned			ref_count;
 
+	/* all other members protected by the VM PD being reserved */
+	struct dma_fence	        *last_pt_update;
+
 	/* mappings for this bo_va */
 	struct list_head		invalids;
 	struct list_head		valids;
+
+	/* If the mappings are cleared or filled */
+	bool				cleared;
+
+	bool				is_xgmi;
+
+	/*
+	 * protected by vm reservation lock
+	 * if non-zero, cannot unmap from GPU because user queues may still access it
+	 */
+	unsigned int			queue_refcount;
 };
 
 struct amdgpu_bo {
 	/* Protected by tbo.reserved */
 	u32				preferred_domains;
 	u32				allowed_domains;
-	struct ttm_place		placements[AMDGPU_GEM_DOMAIN_MAX + 1];
+	struct ttm_place		placements[AMDGPU_BO_MAX_PLACEMENTS];
 	struct ttm_placement		placement;
 	struct ttm_buffer_object	tbo;
 	struct ttm_bo_kmap_obj		kmap;
 	u64				flags;
-	unsigned			pin_count;
+	/* per VM structure for page tables and with virtual addresses */
+	struct amdgpu_vm_bo_base	*vm_bo;
+	/* Constant after initialization */
+	struct amdgpu_bo		*parent;
+
+#ifdef CONFIG_MMU_NOTIFIER
+	struct mmu_interval_notifier	notifier;
+#endif
+	struct kgd_mem                  *kfd_bo;
+
+	/*
+	 * For GPUs with spatial partitioning, xcp partition number, -1 means
+	 * any partition. For other ASICs without spatial partition, always 0
+	 * for memory accounting.
+	 */
+	int8_t				xcp_id;
+};
+
+struct amdgpu_bo_user {
+	struct amdgpu_bo		bo;
 	u64				tiling_flags;
 	u64				metadata_flags;
 	void				*metadata;
 	u32				metadata_size;
-	unsigned			prime_shared_count;
-	/* list of all virtual address to which this bo is associated to */
-	struct list_head		va;
-	/* Constant after initialization */
-	struct drm_gem_object		gem_base;
-	struct amdgpu_bo		*parent;
-	struct amdgpu_bo		*shadow;
 
-	struct ttm_bo_kmap_obj		dma_buf_vmap;
-	struct amdgpu_mn		*mn;
+};
 
-	union {
-		struct list_head	mn_list;
-		struct list_head	shadow_list;
-	};
+struct amdgpu_bo_vm {
+	struct amdgpu_bo		bo;
+	struct amdgpu_vm_bo_base        entries[];
 };
 
+static inline struct amdgpu_bo *ttm_to_amdgpu_bo(struct ttm_buffer_object *tbo)
+{
+	return container_of(tbo, struct amdgpu_bo, tbo);
+}
+
 /**
  * amdgpu_mem_type_to_domain - return domain corresponding to mem_type
  * @mem_type:	ttm memory type
@@ -109,6 +165,10 @@ static inline unsigned amdgpu_mem_type_to_domain(u32 mem_type)
 		return AMDGPU_GEM_DOMAIN_GWS;
 	case AMDGPU_PL_OA:
 		return AMDGPU_GEM_DOMAIN_OA;
+	case AMDGPU_PL_DOORBELL:
+		return AMDGPU_GEM_DOMAIN_DOORBELL;
+	case AMDGPU_PL_MMIO_REMAP:
+		return AMDGPU_GEM_DOMAIN_MMIO_REMAP;
 	default:
 		break;
 	}
@@ -145,17 +205,17 @@ static inline void amdgpu_bo_unreserve(struct amdgpu_bo *bo)
 
 static inline unsigned long amdgpu_bo_size(struct amdgpu_bo *bo)
 {
-	return bo->tbo.num_pages << PAGE_SHIFT;
+	return bo->tbo.base.size;
 }
 
 static inline unsigned amdgpu_bo_ngpu_pages(struct amdgpu_bo *bo)
 {
-	return (bo->tbo.num_pages << PAGE_SHIFT) / AMDGPU_GPU_PAGE_SIZE;
+	return bo->tbo.base.size / AMDGPU_GPU_PAGE_SIZE;
 }
 
 static inline unsigned amdgpu_bo_gpu_page_alignment(struct amdgpu_bo *bo)
 {
-	return (bo->tbo.mem.page_alignment << PAGE_SHIFT) / AMDGPU_GPU_PAGE_SIZE;
+	return (bo->tbo.page_alignment << PAGE_SHIFT) / AMDGPU_GPU_PAGE_SIZE;
 }
 
 /**
@@ -166,37 +226,34 @@ static inline unsigned amdgpu_bo_gpu_page_alignment(struct amdgpu_bo *bo)
  */
 static inline u64 amdgpu_bo_mmap_offset(struct amdgpu_bo *bo)
 {
-	return drm_vma_node_offset_addr(&bo->tbo.vma_node);
+	return drm_vma_node_offset_addr(&bo->tbo.base.vma_node);
 }
 
 /**
- * amdgpu_bo_gpu_accessible - return whether the bo is currently in memory that
- * is accessible to the GPU.
+ * amdgpu_bo_explicit_sync - return whether the bo is explicitly synced
  */
-static inline bool amdgpu_bo_gpu_accessible(struct amdgpu_bo *bo)
+static inline bool amdgpu_bo_explicit_sync(struct amdgpu_bo *bo)
 {
-	switch (bo->tbo.mem.mem_type) {
-	case TTM_PL_TT: return amdgpu_ttm_is_bound(bo->tbo.ttm);
-	case TTM_PL_VRAM: return true;
-	default: return false;
-	}
+	return bo->flags & AMDGPU_GEM_CREATE_EXPLICIT_SYNC;
 }
 
+/**
+ * amdgpu_bo_encrypted - test if the BO is encrypted
+ * @bo: pointer to a buffer object
+ *
+ * Return true if the buffer object is encrypted, false otherwise.
+ */
+static inline bool amdgpu_bo_encrypted(struct amdgpu_bo *bo)
+{
+	return bo->flags & AMDGPU_GEM_CREATE_ENCRYPTED;
+}
+
+bool amdgpu_bo_is_amdgpu_bo(struct ttm_buffer_object *bo);
+void amdgpu_bo_placement_from_domain(struct amdgpu_bo *abo, u32 domain);
+
 int amdgpu_bo_create(struct amdgpu_device *adev,
-			    unsigned long size, int byte_align,
-			    bool kernel, u32 domain, u64 flags,
-			    struct sg_table *sg,
-			    struct reservation_object *resv,
-			    uint64_t init_value,
-			    struct amdgpu_bo **bo_ptr);
-int amdgpu_bo_create_restricted(struct amdgpu_device *adev,
-				unsigned long size, int byte_align,
-				bool kernel, u32 domain, u64 flags,
-				struct sg_table *sg,
-				struct ttm_placement *placement,
-			        struct reservation_object *resv,
-				uint64_t init_value,
-				struct amdgpu_bo **bo_ptr);
+		     struct amdgpu_bo_param *bp,
+		     struct amdgpu_bo **bo_ptr);
 int amdgpu_bo_create_reserved(struct amdgpu_device *adev,
 			      unsigned long size, int align,
 			      u32 domain, struct amdgpu_bo **bo_ptr,
@@ -205,23 +262,31 @@ int amdgpu_bo_create_kernel(struct amdgpu_device *adev,
 			    unsigned long size, int align,
 			    u32 domain, struct amdgpu_bo **bo_ptr,
 			    u64 *gpu_addr, void **cpu_addr);
+int amdgpu_bo_create_isp_user(struct amdgpu_device *adev,
+			   struct dma_buf *dbuf, u32 domain,
+			   struct amdgpu_bo **bo,
+			   u64 *gpu_addr);
+int amdgpu_bo_create_kernel_at(struct amdgpu_device *adev,
+			       uint64_t offset, uint64_t size,
+			       struct amdgpu_bo **bo_ptr, void **cpu_addr);
+int amdgpu_bo_create_user(struct amdgpu_device *adev,
+			  struct amdgpu_bo_param *bp,
+			  struct amdgpu_bo_user **ubo_ptr);
+int amdgpu_bo_create_vm(struct amdgpu_device *adev,
+			struct amdgpu_bo_param *bp,
+			struct amdgpu_bo_vm **ubo_ptr);
 void amdgpu_bo_free_kernel(struct amdgpu_bo **bo, u64 *gpu_addr,
 			   void **cpu_addr);
+void amdgpu_bo_free_isp_user(struct amdgpu_bo *bo);
 int amdgpu_bo_kmap(struct amdgpu_bo *bo, void **ptr);
 void *amdgpu_bo_kptr(struct amdgpu_bo *bo);
 void amdgpu_bo_kunmap(struct amdgpu_bo *bo);
 struct amdgpu_bo *amdgpu_bo_ref(struct amdgpu_bo *bo);
 void amdgpu_bo_unref(struct amdgpu_bo **bo);
-int amdgpu_bo_pin(struct amdgpu_bo *bo, u32 domain, u64 *gpu_addr);
-int amdgpu_bo_pin_restricted(struct amdgpu_bo *bo, u32 domain,
-			     u64 min_offset, u64 max_offset,
-			     u64 *gpu_addr);
-int amdgpu_bo_unpin(struct amdgpu_bo *bo);
-int amdgpu_bo_evict_vram(struct amdgpu_device *adev);
+int amdgpu_bo_pin(struct amdgpu_bo *bo, u32 domain);
+void amdgpu_bo_unpin(struct amdgpu_bo *bo);
 int amdgpu_bo_init(struct amdgpu_device *adev);
 void amdgpu_bo_fini(struct amdgpu_device *adev);
-int amdgpu_bo_fbdev_mmap(struct amdgpu_bo *bo,
-				struct vm_area_struct *vma);
 int amdgpu_bo_set_tiling_flags(struct amdgpu_bo *bo, u64 tiling_flags);
 void amdgpu_bo_get_tiling_flags(struct amdgpu_bo *bo, u64 *tiling_flags);
 int amdgpu_bo_set_metadata (struct amdgpu_bo *bo, void *metadata,
@@ -231,37 +296,41 @@ int amdgpu_bo_get_metadata(struct amdgpu_bo *bo, void *buffer,
 			   uint64_t *flags);
 void amdgpu_bo_move_notify(struct ttm_buffer_object *bo,
 			   bool evict,
-			   struct ttm_mem_reg *new_mem);
-int amdgpu_bo_fault_reserve_notify(struct ttm_buffer_object *bo);
+			   struct ttm_resource *new_mem);
+void amdgpu_bo_release_notify(struct ttm_buffer_object *bo);
+vm_fault_t amdgpu_bo_fault_reserve_notify(struct ttm_buffer_object *bo);
 void amdgpu_bo_fence(struct amdgpu_bo *bo, struct dma_fence *fence,
 		     bool shared);
+int amdgpu_bo_sync_wait_resv(struct amdgpu_device *adev, struct dma_resv *resv,
+			     enum amdgpu_sync_mode sync_mode, void *owner,
+			     bool intr);
+int amdgpu_bo_sync_wait(struct amdgpu_bo *bo, void *owner, bool intr);
 u64 amdgpu_bo_gpu_offset(struct amdgpu_bo *bo);
-int amdgpu_bo_backup_to_shadow(struct amdgpu_device *adev,
-			       struct amdgpu_ring *ring,
-			       struct amdgpu_bo *bo,
-			       struct reservation_object *resv,
-			       struct dma_fence **fence, bool direct);
-int amdgpu_bo_validate(struct amdgpu_bo *bo);
-int amdgpu_bo_restore_from_shadow(struct amdgpu_device *adev,
-				  struct amdgpu_ring *ring,
-				  struct amdgpu_bo *bo,
-				  struct reservation_object *resv,
-				  struct dma_fence **fence,
-				  bool direct);
-
+u64 amdgpu_bo_fb_aper_addr(struct amdgpu_bo *bo);
+u64 amdgpu_bo_gpu_offset_no_check(struct amdgpu_bo *bo);
+uint32_t amdgpu_bo_mem_stats_placement(struct amdgpu_bo *bo);
+uint32_t amdgpu_bo_get_preferred_domain(struct amdgpu_device *adev,
+					    uint32_t domain);
 
 /*
  * sub allocation
  */
+static inline struct amdgpu_sa_manager *
+to_amdgpu_sa_manager(struct drm_suballoc_manager *manager)
+{
+	return container_of(manager, struct amdgpu_sa_manager, base);
+}
 
-static inline uint64_t amdgpu_sa_bo_gpu_addr(struct amdgpu_sa_bo *sa_bo)
+static inline uint64_t amdgpu_sa_bo_gpu_addr(struct drm_suballoc *sa_bo)
 {
-	return sa_bo->manager->gpu_addr + sa_bo->soffset;
+	return to_amdgpu_sa_manager(sa_bo->manager)->gpu_addr +
+		drm_suballoc_soffset(sa_bo);
 }
 
-static inline void * amdgpu_sa_bo_cpu_addr(struct amdgpu_sa_bo *sa_bo)
+static inline void *amdgpu_sa_bo_cpu_addr(struct drm_suballoc *sa_bo)
 {
-	return sa_bo->manager->cpu_ptr + sa_bo->soffset;
+	return to_amdgpu_sa_manager(sa_bo->manager)->cpu_ptr +
+		drm_suballoc_soffset(sa_bo);
 }
 
 int amdgpu_sa_bo_manager_init(struct amdgpu_device *adev,
@@ -271,18 +340,19 @@ void amdgpu_sa_bo_manager_fini(struct amdgpu_device *adev,
 				      struct amdgpu_sa_manager *sa_manager);
 int amdgpu_sa_bo_manager_start(struct amdgpu_device *adev,
 				      struct amdgpu_sa_manager *sa_manager);
-int amdgpu_sa_bo_manager_suspend(struct amdgpu_device *adev,
-					struct amdgpu_sa_manager *sa_manager);
 int amdgpu_sa_bo_new(struct amdgpu_sa_manager *sa_manager,
-		     struct amdgpu_sa_bo **sa_bo,
-		     unsigned size, unsigned align);
-void amdgpu_sa_bo_free(struct amdgpu_device *adev,
-			      struct amdgpu_sa_bo **sa_bo,
-			      struct dma_fence *fence);
+		     struct drm_suballoc **sa_bo,
+		     unsigned int size);
+void amdgpu_sa_bo_free(struct drm_suballoc **sa_bo,
+		       struct dma_fence *fence);
 #if defined(CONFIG_DEBUG_FS)
 void amdgpu_sa_bo_dump_debug_info(struct amdgpu_sa_manager *sa_manager,
 					 struct seq_file *m);
+u64 amdgpu_bo_print_info(int id, struct amdgpu_bo *bo, struct seq_file *m);
 #endif
+void amdgpu_debugfs_sa_init(struct amdgpu_device *adev);
+
+bool amdgpu_bo_support_uswc(u64 bo_flags);
 
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.c
index 8e67c1210d7c..675aa138ea11 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.c
@@ -20,7 +20,7 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  *
  */
-#include <drm/drmP.h>
+
 #include <drm/amdgpu_drm.h>
 #include "amdgpu.h"
 #include "atom.h"
@@ -69,6 +69,7 @@ static void amdgpu_pll_reduce_ratio(unsigned *nom, unsigned *den,
 /**
  * amdgpu_pll_get_fb_ref_div - feedback and ref divider calculation
  *
+ * @adev: amdgpu_device pointer
  * @nom: nominator
  * @den: denominator
  * @post_div: post divider
@@ -80,15 +81,20 @@ static void amdgpu_pll_reduce_ratio(unsigned *nom, unsigned *den,
  * Calculate feedback and reference divider for a given post divider. Makes
  * sure we stay within the limits.
  */
-static void amdgpu_pll_get_fb_ref_div(unsigned nom, unsigned den, unsigned post_div,
-				      unsigned fb_div_max, unsigned ref_div_max,
-				      unsigned *fb_div, unsigned *ref_div)
+static void amdgpu_pll_get_fb_ref_div(struct amdgpu_device *adev, unsigned int nom,
+				      unsigned int den, unsigned int post_div,
+				      unsigned int fb_div_max, unsigned int ref_div_max,
+				      unsigned int *fb_div, unsigned int *ref_div)
 {
+
 	/* limit reference * post divider to a maximum */
-	ref_div_max = min(128 / post_div, ref_div_max);
+	if (adev->family == AMDGPU_FAMILY_SI)
+		ref_div_max = min(100 / post_div, ref_div_max);
+	else
+		ref_div_max = min(128 / post_div, ref_div_max);
 
 	/* get matching reference and feedback divider */
-	*ref_div = min(max(DIV_ROUND_CLOSEST(den, post_div), 1u), ref_div_max);
+	*ref_div = clamp(DIV_ROUND_CLOSEST(den, post_div), 1u, ref_div_max);
 	*fb_div = DIV_ROUND_CLOSEST(nom * *ref_div * post_div, den);
 
 	/* limit fb divider to its maximum */
@@ -101,17 +107,20 @@ static void amdgpu_pll_get_fb_ref_div(unsigned nom, unsigned den, unsigned post_
 /**
  * amdgpu_pll_compute - compute PLL paramaters
  *
+ * @adev: amdgpu_device pointer
  * @pll: information about the PLL
+ * @freq: requested frequency
  * @dot_clock_p: resulting pixel clock
- * fb_div_p: resulting feedback divider
- * frac_fb_div_p: fractional part of the feedback divider
- * ref_div_p: resulting reference divider
- * post_div_p: resulting reference divider
+ * @fb_div_p: resulting feedback divider
+ * @frac_fb_div_p: fractional part of the feedback divider
+ * @ref_div_p: resulting reference divider
+ * @post_div_p: resulting reference divider
  *
  * Try to calculate the PLL parameters to generate the given frequency:
  * dot_clock = (ref_freq * feedback_div) / (ref_div * post_div)
  */
-void amdgpu_pll_compute(struct amdgpu_pll *pll,
+void amdgpu_pll_compute(struct amdgpu_device *adev,
+			struct amdgpu_pll *pll,
 			u32 freq,
 			u32 *dot_clock_p,
 			u32 *fb_div_p,
@@ -198,7 +207,7 @@ void amdgpu_pll_compute(struct amdgpu_pll *pll,
 
 	for (post_div = post_div_min; post_div <= post_div_max; ++post_div) {
 		unsigned diff;
-		amdgpu_pll_get_fb_ref_div(nom, den, post_div, fb_div_max,
+		amdgpu_pll_get_fb_ref_div(adev, nom, den, post_div, fb_div_max,
 					  ref_div_max, &fb_div, &ref_div);
 		diff = abs(target_clock - (pll->reference_freq * fb_div) /
 			(ref_div * post_div));
@@ -213,7 +222,7 @@ void amdgpu_pll_compute(struct amdgpu_pll *pll,
 	post_div = post_div_best;
 
 	/* get the feedback and reference divider for the optimal value */
-	amdgpu_pll_get_fb_ref_div(nom, den, post_div, fb_div_max, ref_div_max,
+	amdgpu_pll_get_fb_ref_div(adev, nom, den, post_div, fb_div_max, ref_div_max,
 				  &fb_div, &ref_div);
 
 	/* reduce the numbers to a simpler ratio once more */
@@ -308,7 +317,6 @@ int amdgpu_pll_get_shared_dp_ppll(struct drm_crtc *crtc)
  * amdgpu_pll_get_shared_nondp_ppll - return the PPLL used by another non-DP crtc
  *
  * @crtc: drm crtc
- * @encoder: drm encoder
  *
  * Returns the PPLL (Pixel PLL) used by another non-DP crtc/encoder which can
  * be shared (i.e., same clock).
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.h
index db6136f68b82..44a583d6c9b4 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.h
@@ -24,7 +24,8 @@
 #ifndef __AMDGPU_PLL_H__
 #define __AMDGPU_PLL_H__
 
-void amdgpu_pll_compute(struct amdgpu_pll *pll,
+void amdgpu_pll_compute(struct amdgpu_device *adev,
+			 struct amdgpu_pll *pll,
 			 u32 freq,
 			 u32 *dot_clock_p,
 			 u32 *fb_div_p,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c
deleted file mode 100644
index 7df503aedb69..000000000000
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c
+++ /dev/null
@@ -1,1659 +0,0 @@
-/*
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- * Authors: Rafał Miłecki <zajec5@gmail.com>
- *          Alex Deucher <alexdeucher@gmail.com>
- */
-#include <drm/drmP.h>
-#include "amdgpu.h"
-#include "amdgpu_drv.h"
-#include "amdgpu_pm.h"
-#include "amdgpu_dpm.h"
-#include "atom.h"
-#include <linux/power_supply.h>
-#include <linux/hwmon.h>
-#include <linux/hwmon-sysfs.h>
-
-#include "amd_powerplay.h"
-
-static int amdgpu_debugfs_pm_init(struct amdgpu_device *adev);
-
-static const struct cg_flag_name clocks[] = {
-	{AMD_CG_SUPPORT_GFX_MGCG, "Graphics Medium Grain Clock Gating"},
-	{AMD_CG_SUPPORT_GFX_MGLS, "Graphics Medium Grain memory Light Sleep"},
-	{AMD_CG_SUPPORT_GFX_CGCG, "Graphics Coarse Grain Clock Gating"},
-	{AMD_CG_SUPPORT_GFX_CGLS, "Graphics Coarse Grain memory Light Sleep"},
-	{AMD_CG_SUPPORT_GFX_CGTS, "Graphics Coarse Grain Tree Shader Clock Gating"},
-	{AMD_CG_SUPPORT_GFX_CGTS_LS, "Graphics Coarse Grain Tree Shader Light Sleep"},
-	{AMD_CG_SUPPORT_GFX_CP_LS, "Graphics Command Processor Light Sleep"},
-	{AMD_CG_SUPPORT_GFX_RLC_LS, "Graphics Run List Controller Light Sleep"},
-	{AMD_CG_SUPPORT_GFX_3D_CGCG, "Graphics 3D Coarse Grain Clock Gating"},
-	{AMD_CG_SUPPORT_GFX_3D_CGLS, "Graphics 3D Coarse Grain memory Light Sleep"},
-	{AMD_CG_SUPPORT_MC_LS, "Memory Controller Light Sleep"},
-	{AMD_CG_SUPPORT_MC_MGCG, "Memory Controller Medium Grain Clock Gating"},
-	{AMD_CG_SUPPORT_SDMA_LS, "System Direct Memory Access Light Sleep"},
-	{AMD_CG_SUPPORT_SDMA_MGCG, "System Direct Memory Access Medium Grain Clock Gating"},
-	{AMD_CG_SUPPORT_BIF_MGCG, "Bus Interface Medium Grain Clock Gating"},
-	{AMD_CG_SUPPORT_BIF_LS, "Bus Interface Light Sleep"},
-	{AMD_CG_SUPPORT_UVD_MGCG, "Unified Video Decoder Medium Grain Clock Gating"},
-	{AMD_CG_SUPPORT_VCE_MGCG, "Video Compression Engine Medium Grain Clock Gating"},
-	{AMD_CG_SUPPORT_HDP_LS, "Host Data Path Light Sleep"},
-	{AMD_CG_SUPPORT_HDP_MGCG, "Host Data Path Medium Grain Clock Gating"},
-	{AMD_CG_SUPPORT_DRM_MGCG, "Digital Right Management Medium Grain Clock Gating"},
-	{AMD_CG_SUPPORT_DRM_LS, "Digital Right Management Light Sleep"},
-	{AMD_CG_SUPPORT_ROM_MGCG, "Rom Medium Grain Clock Gating"},
-	{AMD_CG_SUPPORT_DF_MGCG, "Data Fabric Medium Grain Clock Gating"},
-	{0, NULL},
-};
-
-void amdgpu_pm_acpi_event_handler(struct amdgpu_device *adev)
-{
-	if (adev->pp_enabled)
-		/* TODO */
-		return;
-
-	if (adev->pm.dpm_enabled) {
-		mutex_lock(&adev->pm.mutex);
-		if (power_supply_is_system_supplied() > 0)
-			adev->pm.dpm.ac_power = true;
-		else
-			adev->pm.dpm.ac_power = false;
-		if (adev->pm.funcs->enable_bapm)
-			amdgpu_dpm_enable_bapm(adev, adev->pm.dpm.ac_power);
-		mutex_unlock(&adev->pm.mutex);
-	}
-}
-
-static ssize_t amdgpu_get_dpm_state(struct device *dev,
-				    struct device_attribute *attr,
-				    char *buf)
-{
-	struct drm_device *ddev = dev_get_drvdata(dev);
-	struct amdgpu_device *adev = ddev->dev_private;
-	enum amd_pm_state_type pm;
-
-	if (adev->pp_enabled) {
-		pm = amdgpu_dpm_get_current_power_state(adev);
-	} else
-		pm = adev->pm.dpm.user_state;
-
-	return snprintf(buf, PAGE_SIZE, "%s\n",
-			(pm == POWER_STATE_TYPE_BATTERY) ? "battery" :
-			(pm == POWER_STATE_TYPE_BALANCED) ? "balanced" : "performance");
-}
-
-static ssize_t amdgpu_set_dpm_state(struct device *dev,
-				    struct device_attribute *attr,
-				    const char *buf,
-				    size_t count)
-{
-	struct drm_device *ddev = dev_get_drvdata(dev);
-	struct amdgpu_device *adev = ddev->dev_private;
-	enum amd_pm_state_type  state;
-
-	if (strncmp("battery", buf, strlen("battery")) == 0)
-		state = POWER_STATE_TYPE_BATTERY;
-	else if (strncmp("balanced", buf, strlen("balanced")) == 0)
-		state = POWER_STATE_TYPE_BALANCED;
-	else if (strncmp("performance", buf, strlen("performance")) == 0)
-		state = POWER_STATE_TYPE_PERFORMANCE;
-	else {
-		count = -EINVAL;
-		goto fail;
-	}
-
-	if (adev->pp_enabled) {
-		amdgpu_dpm_dispatch_task(adev, AMD_PP_EVENT_ENABLE_USER_STATE, &state, NULL);
-	} else {
-		mutex_lock(&adev->pm.mutex);
-		adev->pm.dpm.user_state = state;
-		mutex_unlock(&adev->pm.mutex);
-
-		/* Can't set dpm state when the card is off */
-		if (!(adev->flags & AMD_IS_PX) ||
-		    (ddev->switch_power_state == DRM_SWITCH_POWER_ON))
-			amdgpu_pm_compute_clocks(adev);
-	}
-fail:
-	return count;
-}
-
-static ssize_t amdgpu_get_dpm_forced_performance_level(struct device *dev,
-						struct device_attribute *attr,
-								char *buf)
-{
-	struct drm_device *ddev = dev_get_drvdata(dev);
-	struct amdgpu_device *adev = ddev->dev_private;
-	enum amd_dpm_forced_level level;
-
-	if  ((adev->flags & AMD_IS_PX) &&
-	     (ddev->switch_power_state != DRM_SWITCH_POWER_ON))
-		return snprintf(buf, PAGE_SIZE, "off\n");
-
-	level = amdgpu_dpm_get_performance_level(adev);
-	return snprintf(buf, PAGE_SIZE, "%s\n",
-			(level == AMD_DPM_FORCED_LEVEL_AUTO) ? "auto" :
-			(level == AMD_DPM_FORCED_LEVEL_LOW) ? "low" :
-			(level == AMD_DPM_FORCED_LEVEL_HIGH) ? "high" :
-			(level == AMD_DPM_FORCED_LEVEL_MANUAL) ? "manual" :
-			(level == AMD_DPM_FORCED_LEVEL_PROFILE_STANDARD) ? "profile_standard" :
-			(level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK) ? "profile_min_sclk" :
-			(level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK) ? "profile_min_mclk" :
-			(level == AMD_DPM_FORCED_LEVEL_PROFILE_PEAK) ? "profile_peak" :
-			"unknown");
-}
-
-static ssize_t amdgpu_set_dpm_forced_performance_level(struct device *dev,
-						       struct device_attribute *attr,
-						       const char *buf,
-						       size_t count)
-{
-	struct drm_device *ddev = dev_get_drvdata(dev);
-	struct amdgpu_device *adev = ddev->dev_private;
-	enum amd_dpm_forced_level level;
-	enum amd_dpm_forced_level current_level;
-	int ret = 0;
-
-	/* Can't force performance level when the card is off */
-	if  ((adev->flags & AMD_IS_PX) &&
-	     (ddev->switch_power_state != DRM_SWITCH_POWER_ON))
-		return -EINVAL;
-
-	current_level = amdgpu_dpm_get_performance_level(adev);
-
-	if (strncmp("low", buf, strlen("low")) == 0) {
-		level = AMD_DPM_FORCED_LEVEL_LOW;
-	} else if (strncmp("high", buf, strlen("high")) == 0) {
-		level = AMD_DPM_FORCED_LEVEL_HIGH;
-	} else if (strncmp("auto", buf, strlen("auto")) == 0) {
-		level = AMD_DPM_FORCED_LEVEL_AUTO;
-	} else if (strncmp("manual", buf, strlen("manual")) == 0) {
-		level = AMD_DPM_FORCED_LEVEL_MANUAL;
-	} else if (strncmp("profile_exit", buf, strlen("profile_exit")) == 0) {
-		level = AMD_DPM_FORCED_LEVEL_PROFILE_EXIT;
-	} else if (strncmp("profile_standard", buf, strlen("profile_standard")) == 0) {
-		level = AMD_DPM_FORCED_LEVEL_PROFILE_STANDARD;
-	} else if (strncmp("profile_min_sclk", buf, strlen("profile_min_sclk")) == 0) {
-		level = AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK;
-	} else if (strncmp("profile_min_mclk", buf, strlen("profile_min_mclk")) == 0) {
-		level = AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK;
-	} else if (strncmp("profile_peak", buf, strlen("profile_peak")) == 0) {
-		level = AMD_DPM_FORCED_LEVEL_PROFILE_PEAK;
-	}  else {
-		count = -EINVAL;
-		goto fail;
-	}
-
-	if (current_level == level)
-		return count;
-
-	if (adev->pp_enabled)
-		amdgpu_dpm_force_performance_level(adev, level);
-	else {
-		mutex_lock(&adev->pm.mutex);
-		if (adev->pm.dpm.thermal_active) {
-			count = -EINVAL;
-			mutex_unlock(&adev->pm.mutex);
-			goto fail;
-		}
-		ret = amdgpu_dpm_force_performance_level(adev, level);
-		if (ret)
-			count = -EINVAL;
-		else
-			adev->pm.dpm.forced_level = level;
-		mutex_unlock(&adev->pm.mutex);
-	}
-
-fail:
-	return count;
-}
-
-static ssize_t amdgpu_get_pp_num_states(struct device *dev,
-		struct device_attribute *attr,
-		char *buf)
-{
-	struct drm_device *ddev = dev_get_drvdata(dev);
-	struct amdgpu_device *adev = ddev->dev_private;
-	struct pp_states_info data;
-	int i, buf_len;
-
-	if (adev->pp_enabled)
-		amdgpu_dpm_get_pp_num_states(adev, &data);
-
-	buf_len = snprintf(buf, PAGE_SIZE, "states: %d\n", data.nums);
-	for (i = 0; i < data.nums; i++)
-		buf_len += snprintf(buf + buf_len, PAGE_SIZE, "%d %s\n", i,
-				(data.states[i] == POWER_STATE_TYPE_INTERNAL_BOOT) ? "boot" :
-				(data.states[i] == POWER_STATE_TYPE_BATTERY) ? "battery" :
-				(data.states[i] == POWER_STATE_TYPE_BALANCED) ? "balanced" :
-				(data.states[i] == POWER_STATE_TYPE_PERFORMANCE) ? "performance" : "default");
-
-	return buf_len;
-}
-
-static ssize_t amdgpu_get_pp_cur_state(struct device *dev,
-		struct device_attribute *attr,
-		char *buf)
-{
-	struct drm_device *ddev = dev_get_drvdata(dev);
-	struct amdgpu_device *adev = ddev->dev_private;
-	struct pp_states_info data;
-	enum amd_pm_state_type pm = 0;
-	int i = 0;
-
-	if (adev->pp_enabled) {
-
-		pm = amdgpu_dpm_get_current_power_state(adev);
-		amdgpu_dpm_get_pp_num_states(adev, &data);
-
-		for (i = 0; i < data.nums; i++) {
-			if (pm == data.states[i])
-				break;
-		}
-
-		if (i == data.nums)
-			i = -EINVAL;
-	}
-
-	return snprintf(buf, PAGE_SIZE, "%d\n", i);
-}
-
-static ssize_t amdgpu_get_pp_force_state(struct device *dev,
-		struct device_attribute *attr,
-		char *buf)
-{
-	struct drm_device *ddev = dev_get_drvdata(dev);
-	struct amdgpu_device *adev = ddev->dev_private;
-	struct pp_states_info data;
-	enum amd_pm_state_type pm = 0;
-	int i;
-
-	if (adev->pp_force_state_enabled && adev->pp_enabled) {
-		pm = amdgpu_dpm_get_current_power_state(adev);
-		amdgpu_dpm_get_pp_num_states(adev, &data);
-
-		for (i = 0; i < data.nums; i++) {
-			if (pm == data.states[i])
-				break;
-		}
-
-		if (i == data.nums)
-			i = -EINVAL;
-
-		return snprintf(buf, PAGE_SIZE, "%d\n", i);
-
-	} else
-		return snprintf(buf, PAGE_SIZE, "\n");
-}
-
-static ssize_t amdgpu_set_pp_force_state(struct device *dev,
-		struct device_attribute *attr,
-		const char *buf,
-		size_t count)
-{
-	struct drm_device *ddev = dev_get_drvdata(dev);
-	struct amdgpu_device *adev = ddev->dev_private;
-	enum amd_pm_state_type state = 0;
-	unsigned long idx;
-	int ret;
-
-	if (strlen(buf) == 1)
-		adev->pp_force_state_enabled = false;
-	else if (adev->pp_enabled) {
-		struct pp_states_info data;
-
-		ret = kstrtoul(buf, 0, &idx);
-		if (ret || idx >= ARRAY_SIZE(data.states)) {
-			count = -EINVAL;
-			goto fail;
-		}
-
-		amdgpu_dpm_get_pp_num_states(adev, &data);
-		state = data.states[idx];
-		/* only set user selected power states */
-		if (state != POWER_STATE_TYPE_INTERNAL_BOOT &&
-		    state != POWER_STATE_TYPE_DEFAULT) {
-			amdgpu_dpm_dispatch_task(adev,
-					AMD_PP_EVENT_ENABLE_USER_STATE, &state, NULL);
-			adev->pp_force_state_enabled = true;
-		}
-	}
-fail:
-	return count;
-}
-
-static ssize_t amdgpu_get_pp_table(struct device *dev,
-		struct device_attribute *attr,
-		char *buf)
-{
-	struct drm_device *ddev = dev_get_drvdata(dev);
-	struct amdgpu_device *adev = ddev->dev_private;
-	char *table = NULL;
-	int size;
-
-	if (adev->pp_enabled)
-		size = amdgpu_dpm_get_pp_table(adev, &table);
-	else
-		return 0;
-
-	if (size >= PAGE_SIZE)
-		size = PAGE_SIZE - 1;
-
-	memcpy(buf, table, size);
-
-	return size;
-}
-
-static ssize_t amdgpu_set_pp_table(struct device *dev,
-		struct device_attribute *attr,
-		const char *buf,
-		size_t count)
-{
-	struct drm_device *ddev = dev_get_drvdata(dev);
-	struct amdgpu_device *adev = ddev->dev_private;
-
-	if (adev->pp_enabled)
-		amdgpu_dpm_set_pp_table(adev, buf, count);
-
-	return count;
-}
-
-static ssize_t amdgpu_get_pp_dpm_sclk(struct device *dev,
-		struct device_attribute *attr,
-		char *buf)
-{
-	struct drm_device *ddev = dev_get_drvdata(dev);
-	struct amdgpu_device *adev = ddev->dev_private;
-	ssize_t size = 0;
-
-	if (adev->pp_enabled)
-		size = amdgpu_dpm_print_clock_levels(adev, PP_SCLK, buf);
-	else if (adev->pm.funcs->print_clock_levels)
-		size = adev->pm.funcs->print_clock_levels(adev, PP_SCLK, buf);
-
-	return size;
-}
-
-static ssize_t amdgpu_set_pp_dpm_sclk(struct device *dev,
-		struct device_attribute *attr,
-		const char *buf,
-		size_t count)
-{
-	struct drm_device *ddev = dev_get_drvdata(dev);
-	struct amdgpu_device *adev = ddev->dev_private;
-	int ret;
-	long level;
-	uint32_t i, mask = 0;
-	char sub_str[2];
-
-	for (i = 0; i < strlen(buf); i++) {
-		if (*(buf + i) == '\n')
-			continue;
-		sub_str[0] = *(buf + i);
-		sub_str[1] = '\0';
-		ret = kstrtol(sub_str, 0, &level);
-
-		if (ret) {
-			count = -EINVAL;
-			goto fail;
-		}
-		mask |= 1 << level;
-	}
-
-	if (adev->pp_enabled)
-		amdgpu_dpm_force_clock_level(adev, PP_SCLK, mask);
-	else if (adev->pm.funcs->force_clock_level)
-		adev->pm.funcs->force_clock_level(adev, PP_SCLK, mask);
-fail:
-	return count;
-}
-
-static ssize_t amdgpu_get_pp_dpm_mclk(struct device *dev,
-		struct device_attribute *attr,
-		char *buf)
-{
-	struct drm_device *ddev = dev_get_drvdata(dev);
-	struct amdgpu_device *adev = ddev->dev_private;
-	ssize_t size = 0;
-
-	if (adev->pp_enabled)
-		size = amdgpu_dpm_print_clock_levels(adev, PP_MCLK, buf);
-	else if (adev->pm.funcs->print_clock_levels)
-		size = adev->pm.funcs->print_clock_levels(adev, PP_MCLK, buf);
-
-	return size;
-}
-
-static ssize_t amdgpu_set_pp_dpm_mclk(struct device *dev,
-		struct device_attribute *attr,
-		const char *buf,
-		size_t count)
-{
-	struct drm_device *ddev = dev_get_drvdata(dev);
-	struct amdgpu_device *adev = ddev->dev_private;
-	int ret;
-	long level;
-	uint32_t i, mask = 0;
-	char sub_str[2];
-
-	for (i = 0; i < strlen(buf); i++) {
-		if (*(buf + i) == '\n')
-			continue;
-		sub_str[0] = *(buf + i);
-		sub_str[1] = '\0';
-		ret = kstrtol(sub_str, 0, &level);
-
-		if (ret) {
-			count = -EINVAL;
-			goto fail;
-		}
-		mask |= 1 << level;
-	}
-
-	if (adev->pp_enabled)
-		amdgpu_dpm_force_clock_level(adev, PP_MCLK, mask);
-	else if (adev->pm.funcs->force_clock_level)
-		adev->pm.funcs->force_clock_level(adev, PP_MCLK, mask);
-fail:
-	return count;
-}
-
-static ssize_t amdgpu_get_pp_dpm_pcie(struct device *dev,
-		struct device_attribute *attr,
-		char *buf)
-{
-	struct drm_device *ddev = dev_get_drvdata(dev);
-	struct amdgpu_device *adev = ddev->dev_private;
-	ssize_t size = 0;
-
-	if (adev->pp_enabled)
-		size = amdgpu_dpm_print_clock_levels(adev, PP_PCIE, buf);
-	else if (adev->pm.funcs->print_clock_levels)
-		size = adev->pm.funcs->print_clock_levels(adev, PP_PCIE, buf);
-
-	return size;
-}
-
-static ssize_t amdgpu_set_pp_dpm_pcie(struct device *dev,
-		struct device_attribute *attr,
-		const char *buf,
-		size_t count)
-{
-	struct drm_device *ddev = dev_get_drvdata(dev);
-	struct amdgpu_device *adev = ddev->dev_private;
-	int ret;
-	long level;
-	uint32_t i, mask = 0;
-	char sub_str[2];
-
-	for (i = 0; i < strlen(buf); i++) {
-		if (*(buf + i) == '\n')
-			continue;
-		sub_str[0] = *(buf + i);
-		sub_str[1] = '\0';
-		ret = kstrtol(sub_str, 0, &level);
-
-		if (ret) {
-			count = -EINVAL;
-			goto fail;
-		}
-		mask |= 1 << level;
-	}
-
-	if (adev->pp_enabled)
-		amdgpu_dpm_force_clock_level(adev, PP_PCIE, mask);
-	else if (adev->pm.funcs->force_clock_level)
-		adev->pm.funcs->force_clock_level(adev, PP_PCIE, mask);
-fail:
-	return count;
-}
-
-static ssize_t amdgpu_get_pp_sclk_od(struct device *dev,
-		struct device_attribute *attr,
-		char *buf)
-{
-	struct drm_device *ddev = dev_get_drvdata(dev);
-	struct amdgpu_device *adev = ddev->dev_private;
-	uint32_t value = 0;
-
-	if (adev->pp_enabled)
-		value = amdgpu_dpm_get_sclk_od(adev);
-	else if (adev->pm.funcs->get_sclk_od)
-		value = adev->pm.funcs->get_sclk_od(adev);
-
-	return snprintf(buf, PAGE_SIZE, "%d\n", value);
-}
-
-static ssize_t amdgpu_set_pp_sclk_od(struct device *dev,
-		struct device_attribute *attr,
-		const char *buf,
-		size_t count)
-{
-	struct drm_device *ddev = dev_get_drvdata(dev);
-	struct amdgpu_device *adev = ddev->dev_private;
-	int ret;
-	long int value;
-
-	ret = kstrtol(buf, 0, &value);
-
-	if (ret) {
-		count = -EINVAL;
-		goto fail;
-	}
-
-	if (adev->pp_enabled) {
-		amdgpu_dpm_set_sclk_od(adev, (uint32_t)value);
-		amdgpu_dpm_dispatch_task(adev, AMD_PP_EVENT_READJUST_POWER_STATE, NULL, NULL);
-	} else if (adev->pm.funcs->set_sclk_od) {
-		adev->pm.funcs->set_sclk_od(adev, (uint32_t)value);
-		adev->pm.dpm.current_ps = adev->pm.dpm.boot_ps;
-		amdgpu_pm_compute_clocks(adev);
-	}
-
-fail:
-	return count;
-}
-
-static ssize_t amdgpu_get_pp_mclk_od(struct device *dev,
-		struct device_attribute *attr,
-		char *buf)
-{
-	struct drm_device *ddev = dev_get_drvdata(dev);
-	struct amdgpu_device *adev = ddev->dev_private;
-	uint32_t value = 0;
-
-	if (adev->pp_enabled)
-		value = amdgpu_dpm_get_mclk_od(adev);
-	else if (adev->pm.funcs->get_mclk_od)
-		value = adev->pm.funcs->get_mclk_od(adev);
-
-	return snprintf(buf, PAGE_SIZE, "%d\n", value);
-}
-
-static ssize_t amdgpu_set_pp_mclk_od(struct device *dev,
-		struct device_attribute *attr,
-		const char *buf,
-		size_t count)
-{
-	struct drm_device *ddev = dev_get_drvdata(dev);
-	struct amdgpu_device *adev = ddev->dev_private;
-	int ret;
-	long int value;
-
-	ret = kstrtol(buf, 0, &value);
-
-	if (ret) {
-		count = -EINVAL;
-		goto fail;
-	}
-
-	if (adev->pp_enabled) {
-		amdgpu_dpm_set_mclk_od(adev, (uint32_t)value);
-		amdgpu_dpm_dispatch_task(adev, AMD_PP_EVENT_READJUST_POWER_STATE, NULL, NULL);
-	} else if (adev->pm.funcs->set_mclk_od) {
-		adev->pm.funcs->set_mclk_od(adev, (uint32_t)value);
-		adev->pm.dpm.current_ps = adev->pm.dpm.boot_ps;
-		amdgpu_pm_compute_clocks(adev);
-	}
-
-fail:
-	return count;
-}
-
-static ssize_t amdgpu_get_pp_power_profile(struct device *dev,
-		char *buf, struct amd_pp_profile *query)
-{
-	struct drm_device *ddev = dev_get_drvdata(dev);
-	struct amdgpu_device *adev = ddev->dev_private;
-	int ret = 0;
-
-	if (adev->pp_enabled)
-		ret = amdgpu_dpm_get_power_profile_state(
-				adev, query);
-	else if (adev->pm.funcs->get_power_profile_state)
-		ret = adev->pm.funcs->get_power_profile_state(
-				adev, query);
-
-	if (ret)
-		return ret;
-
-	return snprintf(buf, PAGE_SIZE,
-			"%d %d %d %d %d\n",
-			query->min_sclk / 100,
-			query->min_mclk / 100,
-			query->activity_threshold,
-			query->up_hyst,
-			query->down_hyst);
-}
-
-static ssize_t amdgpu_get_pp_gfx_power_profile(struct device *dev,
-		struct device_attribute *attr,
-		char *buf)
-{
-	struct amd_pp_profile query = {0};
-
-	query.type = AMD_PP_GFX_PROFILE;
-
-	return amdgpu_get_pp_power_profile(dev, buf, &query);
-}
-
-static ssize_t amdgpu_get_pp_compute_power_profile(struct device *dev,
-		struct device_attribute *attr,
-		char *buf)
-{
-	struct amd_pp_profile query = {0};
-
-	query.type = AMD_PP_COMPUTE_PROFILE;
-
-	return amdgpu_get_pp_power_profile(dev, buf, &query);
-}
-
-static ssize_t amdgpu_set_pp_power_profile(struct device *dev,
-		const char *buf,
-		size_t count,
-		struct amd_pp_profile *request)
-{
-	struct drm_device *ddev = dev_get_drvdata(dev);
-	struct amdgpu_device *adev = ddev->dev_private;
-	uint32_t loop = 0;
-	char *sub_str, buf_cpy[128], *tmp_str;
-	const char delimiter[3] = {' ', '\n', '\0'};
-	long int value;
-	int ret = 0;
-
-	if (strncmp("reset", buf, strlen("reset")) == 0) {
-		if (adev->pp_enabled)
-			ret = amdgpu_dpm_reset_power_profile_state(
-					adev, request);
-		else if (adev->pm.funcs->reset_power_profile_state)
-			ret = adev->pm.funcs->reset_power_profile_state(
-					adev, request);
-		if (ret) {
-			count = -EINVAL;
-			goto fail;
-		}
-		return count;
-	}
-
-	if (strncmp("set", buf, strlen("set")) == 0) {
-		if (adev->pp_enabled)
-			ret = amdgpu_dpm_set_power_profile_state(
-					adev, request);
-		else if (adev->pm.funcs->set_power_profile_state)
-			ret = adev->pm.funcs->set_power_profile_state(
-					adev, request);
-		if (ret) {
-			count = -EINVAL;
-			goto fail;
-		}
-		return count;
-	}
-
-	if (count + 1 >= 128) {
-		count = -EINVAL;
-		goto fail;
-	}
-
-	memcpy(buf_cpy, buf, count + 1);
-	tmp_str = buf_cpy;
-
-	while (tmp_str[0]) {
-		sub_str = strsep(&tmp_str, delimiter);
-		ret = kstrtol(sub_str, 0, &value);
-		if (ret) {
-			count = -EINVAL;
-			goto fail;
-		}
-
-		switch (loop) {
-		case 0:
-			/* input unit MHz convert to dpm table unit 10KHz*/
-			request->min_sclk = (uint32_t)value * 100;
-			break;
-		case 1:
-			/* input unit MHz convert to dpm table unit 10KHz*/
-			request->min_mclk = (uint32_t)value * 100;
-			break;
-		case 2:
-			request->activity_threshold = (uint16_t)value;
-			break;
-		case 3:
-			request->up_hyst = (uint8_t)value;
-			break;
-		case 4:
-			request->down_hyst = (uint8_t)value;
-			break;
-		default:
-			break;
-		}
-
-		loop++;
-	}
-
-	if (adev->pp_enabled)
-		ret = amdgpu_dpm_set_power_profile_state(
-				adev, request);
-	else if (adev->pm.funcs->set_power_profile_state)
-		ret = adev->pm.funcs->set_power_profile_state(
-				adev, request);
-
-	if (ret)
-		count = -EINVAL;
-
-fail:
-	return count;
-}
-
-static ssize_t amdgpu_set_pp_gfx_power_profile(struct device *dev,
-		struct device_attribute *attr,
-		const char *buf,
-		size_t count)
-{
-	struct amd_pp_profile request = {0};
-
-	request.type = AMD_PP_GFX_PROFILE;
-
-	return amdgpu_set_pp_power_profile(dev, buf, count, &request);
-}
-
-static ssize_t amdgpu_set_pp_compute_power_profile(struct device *dev,
-		struct device_attribute *attr,
-		const char *buf,
-		size_t count)
-{
-	struct amd_pp_profile request = {0};
-
-	request.type = AMD_PP_COMPUTE_PROFILE;
-
-	return amdgpu_set_pp_power_profile(dev, buf, count, &request);
-}
-
-static DEVICE_ATTR(power_dpm_state, S_IRUGO | S_IWUSR, amdgpu_get_dpm_state, amdgpu_set_dpm_state);
-static DEVICE_ATTR(power_dpm_force_performance_level, S_IRUGO | S_IWUSR,
-		   amdgpu_get_dpm_forced_performance_level,
-		   amdgpu_set_dpm_forced_performance_level);
-static DEVICE_ATTR(pp_num_states, S_IRUGO, amdgpu_get_pp_num_states, NULL);
-static DEVICE_ATTR(pp_cur_state, S_IRUGO, amdgpu_get_pp_cur_state, NULL);
-static DEVICE_ATTR(pp_force_state, S_IRUGO | S_IWUSR,
-		amdgpu_get_pp_force_state,
-		amdgpu_set_pp_force_state);
-static DEVICE_ATTR(pp_table, S_IRUGO | S_IWUSR,
-		amdgpu_get_pp_table,
-		amdgpu_set_pp_table);
-static DEVICE_ATTR(pp_dpm_sclk, S_IRUGO | S_IWUSR,
-		amdgpu_get_pp_dpm_sclk,
-		amdgpu_set_pp_dpm_sclk);
-static DEVICE_ATTR(pp_dpm_mclk, S_IRUGO | S_IWUSR,
-		amdgpu_get_pp_dpm_mclk,
-		amdgpu_set_pp_dpm_mclk);
-static DEVICE_ATTR(pp_dpm_pcie, S_IRUGO | S_IWUSR,
-		amdgpu_get_pp_dpm_pcie,
-		amdgpu_set_pp_dpm_pcie);
-static DEVICE_ATTR(pp_sclk_od, S_IRUGO | S_IWUSR,
-		amdgpu_get_pp_sclk_od,
-		amdgpu_set_pp_sclk_od);
-static DEVICE_ATTR(pp_mclk_od, S_IRUGO | S_IWUSR,
-		amdgpu_get_pp_mclk_od,
-		amdgpu_set_pp_mclk_od);
-static DEVICE_ATTR(pp_gfx_power_profile, S_IRUGO | S_IWUSR,
-		amdgpu_get_pp_gfx_power_profile,
-		amdgpu_set_pp_gfx_power_profile);
-static DEVICE_ATTR(pp_compute_power_profile, S_IRUGO | S_IWUSR,
-		amdgpu_get_pp_compute_power_profile,
-		amdgpu_set_pp_compute_power_profile);
-
-static ssize_t amdgpu_hwmon_show_temp(struct device *dev,
-				      struct device_attribute *attr,
-				      char *buf)
-{
-	struct amdgpu_device *adev = dev_get_drvdata(dev);
-	struct drm_device *ddev = adev->ddev;
-	int temp;
-
-	/* Can't get temperature when the card is off */
-	if  ((adev->flags & AMD_IS_PX) &&
-	     (ddev->switch_power_state != DRM_SWITCH_POWER_ON))
-		return -EINVAL;
-
-	if (!adev->pp_enabled && !adev->pm.funcs->get_temperature)
-		temp = 0;
-	else
-		temp = amdgpu_dpm_get_temperature(adev);
-
-	return snprintf(buf, PAGE_SIZE, "%d\n", temp);
-}
-
-static ssize_t amdgpu_hwmon_show_temp_thresh(struct device *dev,
-					     struct device_attribute *attr,
-					     char *buf)
-{
-	struct amdgpu_device *adev = dev_get_drvdata(dev);
-	int hyst = to_sensor_dev_attr(attr)->index;
-	int temp;
-
-	if (hyst)
-		temp = adev->pm.dpm.thermal.min_temp;
-	else
-		temp = adev->pm.dpm.thermal.max_temp;
-
-	return snprintf(buf, PAGE_SIZE, "%d\n", temp);
-}
-
-static ssize_t amdgpu_hwmon_get_pwm1_enable(struct device *dev,
-					    struct device_attribute *attr,
-					    char *buf)
-{
-	struct amdgpu_device *adev = dev_get_drvdata(dev);
-	u32 pwm_mode = 0;
-
-	if (!adev->pp_enabled && !adev->pm.funcs->get_fan_control_mode)
-		return -EINVAL;
-
-	pwm_mode = amdgpu_dpm_get_fan_control_mode(adev);
-
-	return sprintf(buf, "%i\n", pwm_mode);
-}
-
-static ssize_t amdgpu_hwmon_set_pwm1_enable(struct device *dev,
-					    struct device_attribute *attr,
-					    const char *buf,
-					    size_t count)
-{
-	struct amdgpu_device *adev = dev_get_drvdata(dev);
-	int err;
-	int value;
-
-	if (!adev->pp_enabled && !adev->pm.funcs->set_fan_control_mode)
-		return -EINVAL;
-
-	err = kstrtoint(buf, 10, &value);
-	if (err)
-		return err;
-
-	amdgpu_dpm_set_fan_control_mode(adev, value);
-
-	return count;
-}
-
-static ssize_t amdgpu_hwmon_get_pwm1_min(struct device *dev,
-					 struct device_attribute *attr,
-					 char *buf)
-{
-	return sprintf(buf, "%i\n", 0);
-}
-
-static ssize_t amdgpu_hwmon_get_pwm1_max(struct device *dev,
-					 struct device_attribute *attr,
-					 char *buf)
-{
-	return sprintf(buf, "%i\n", 255);
-}
-
-static ssize_t amdgpu_hwmon_set_pwm1(struct device *dev,
-				     struct device_attribute *attr,
-				     const char *buf, size_t count)
-{
-	struct amdgpu_device *adev = dev_get_drvdata(dev);
-	int err;
-	u32 value;
-
-	err = kstrtou32(buf, 10, &value);
-	if (err)
-		return err;
-
-	value = (value * 100) / 255;
-
-	err = amdgpu_dpm_set_fan_speed_percent(adev, value);
-	if (err)
-		return err;
-
-	return count;
-}
-
-static ssize_t amdgpu_hwmon_get_pwm1(struct device *dev,
-				     struct device_attribute *attr,
-				     char *buf)
-{
-	struct amdgpu_device *adev = dev_get_drvdata(dev);
-	int err;
-	u32 speed;
-
-	err = amdgpu_dpm_get_fan_speed_percent(adev, &speed);
-	if (err)
-		return err;
-
-	speed = (speed * 255) / 100;
-
-	return sprintf(buf, "%i\n", speed);
-}
-
-static ssize_t amdgpu_hwmon_get_fan1_input(struct device *dev,
-					   struct device_attribute *attr,
-					   char *buf)
-{
-	struct amdgpu_device *adev = dev_get_drvdata(dev);
-	int err;
-	u32 speed;
-
-	err = amdgpu_dpm_get_fan_speed_rpm(adev, &speed);
-	if (err)
-		return err;
-
-	return sprintf(buf, "%i\n", speed);
-}
-
-static SENSOR_DEVICE_ATTR(temp1_input, S_IRUGO, amdgpu_hwmon_show_temp, NULL, 0);
-static SENSOR_DEVICE_ATTR(temp1_crit, S_IRUGO, amdgpu_hwmon_show_temp_thresh, NULL, 0);
-static SENSOR_DEVICE_ATTR(temp1_crit_hyst, S_IRUGO, amdgpu_hwmon_show_temp_thresh, NULL, 1);
-static SENSOR_DEVICE_ATTR(pwm1, S_IRUGO | S_IWUSR, amdgpu_hwmon_get_pwm1, amdgpu_hwmon_set_pwm1, 0);
-static SENSOR_DEVICE_ATTR(pwm1_enable, S_IRUGO | S_IWUSR, amdgpu_hwmon_get_pwm1_enable, amdgpu_hwmon_set_pwm1_enable, 0);
-static SENSOR_DEVICE_ATTR(pwm1_min, S_IRUGO, amdgpu_hwmon_get_pwm1_min, NULL, 0);
-static SENSOR_DEVICE_ATTR(pwm1_max, S_IRUGO, amdgpu_hwmon_get_pwm1_max, NULL, 0);
-static SENSOR_DEVICE_ATTR(fan1_input, S_IRUGO, amdgpu_hwmon_get_fan1_input, NULL, 0);
-
-static struct attribute *hwmon_attributes[] = {
-	&sensor_dev_attr_temp1_input.dev_attr.attr,
-	&sensor_dev_attr_temp1_crit.dev_attr.attr,
-	&sensor_dev_attr_temp1_crit_hyst.dev_attr.attr,
-	&sensor_dev_attr_pwm1.dev_attr.attr,
-	&sensor_dev_attr_pwm1_enable.dev_attr.attr,
-	&sensor_dev_attr_pwm1_min.dev_attr.attr,
-	&sensor_dev_attr_pwm1_max.dev_attr.attr,
-	&sensor_dev_attr_fan1_input.dev_attr.attr,
-	NULL
-};
-
-static umode_t hwmon_attributes_visible(struct kobject *kobj,
-					struct attribute *attr, int index)
-{
-	struct device *dev = kobj_to_dev(kobj);
-	struct amdgpu_device *adev = dev_get_drvdata(dev);
-	umode_t effective_mode = attr->mode;
-
-	/* Skip limit attributes if DPM is not enabled */
-	if (!adev->pm.dpm_enabled &&
-	    (attr == &sensor_dev_attr_temp1_crit.dev_attr.attr ||
-	     attr == &sensor_dev_attr_temp1_crit_hyst.dev_attr.attr ||
-	     attr == &sensor_dev_attr_pwm1.dev_attr.attr ||
-	     attr == &sensor_dev_attr_pwm1_enable.dev_attr.attr ||
-	     attr == &sensor_dev_attr_pwm1_max.dev_attr.attr ||
-	     attr == &sensor_dev_attr_pwm1_min.dev_attr.attr))
-		return 0;
-
-	if (adev->pp_enabled)
-		return effective_mode;
-
-	/* Skip fan attributes if fan is not present */
-	if (adev->pm.no_fan &&
-	    (attr == &sensor_dev_attr_pwm1.dev_attr.attr ||
-	     attr == &sensor_dev_attr_pwm1_enable.dev_attr.attr ||
-	     attr == &sensor_dev_attr_pwm1_max.dev_attr.attr ||
-	     attr == &sensor_dev_attr_pwm1_min.dev_attr.attr))
-		return 0;
-
-	/* mask fan attributes if we have no bindings for this asic to expose */
-	if ((!adev->pm.funcs->get_fan_speed_percent &&
-	     attr == &sensor_dev_attr_pwm1.dev_attr.attr) || /* can't query fan */
-	    (!adev->pm.funcs->get_fan_control_mode &&
-	     attr == &sensor_dev_attr_pwm1_enable.dev_attr.attr)) /* can't query state */
-		effective_mode &= ~S_IRUGO;
-
-	if ((!adev->pm.funcs->set_fan_speed_percent &&
-	     attr == &sensor_dev_attr_pwm1.dev_attr.attr) || /* can't manage fan */
-	    (!adev->pm.funcs->set_fan_control_mode &&
-	     attr == &sensor_dev_attr_pwm1_enable.dev_attr.attr)) /* can't manage state */
-		effective_mode &= ~S_IWUSR;
-
-	/* hide max/min values if we can't both query and manage the fan */
-	if ((!adev->pm.funcs->set_fan_speed_percent &&
-	     !adev->pm.funcs->get_fan_speed_percent) &&
-	    (attr == &sensor_dev_attr_pwm1_max.dev_attr.attr ||
-	     attr == &sensor_dev_attr_pwm1_min.dev_attr.attr))
-		return 0;
-
-	/* requires powerplay */
-	if (attr == &sensor_dev_attr_fan1_input.dev_attr.attr)
-		return 0;
-
-	return effective_mode;
-}
-
-static const struct attribute_group hwmon_attrgroup = {
-	.attrs = hwmon_attributes,
-	.is_visible = hwmon_attributes_visible,
-};
-
-static const struct attribute_group *hwmon_groups[] = {
-	&hwmon_attrgroup,
-	NULL
-};
-
-void amdgpu_dpm_thermal_work_handler(struct work_struct *work)
-{
-	struct amdgpu_device *adev =
-		container_of(work, struct amdgpu_device,
-			     pm.dpm.thermal.work);
-	/* switch to the thermal state */
-	enum amd_pm_state_type dpm_state = POWER_STATE_TYPE_INTERNAL_THERMAL;
-
-	if (!adev->pm.dpm_enabled)
-		return;
-
-	if (adev->pm.funcs->get_temperature) {
-		int temp = amdgpu_dpm_get_temperature(adev);
-
-		if (temp < adev->pm.dpm.thermal.min_temp)
-			/* switch back the user state */
-			dpm_state = adev->pm.dpm.user_state;
-	} else {
-		if (adev->pm.dpm.thermal.high_to_low)
-			/* switch back the user state */
-			dpm_state = adev->pm.dpm.user_state;
-	}
-	mutex_lock(&adev->pm.mutex);
-	if (dpm_state == POWER_STATE_TYPE_INTERNAL_THERMAL)
-		adev->pm.dpm.thermal_active = true;
-	else
-		adev->pm.dpm.thermal_active = false;
-	adev->pm.dpm.state = dpm_state;
-	mutex_unlock(&adev->pm.mutex);
-
-	amdgpu_pm_compute_clocks(adev);
-}
-
-static struct amdgpu_ps *amdgpu_dpm_pick_power_state(struct amdgpu_device *adev,
-						     enum amd_pm_state_type dpm_state)
-{
-	int i;
-	struct amdgpu_ps *ps;
-	u32 ui_class;
-	bool single_display = (adev->pm.dpm.new_active_crtc_count < 2) ?
-		true : false;
-
-	/* check if the vblank period is too short to adjust the mclk */
-	if (single_display && adev->pm.funcs->vblank_too_short) {
-		if (amdgpu_dpm_vblank_too_short(adev))
-			single_display = false;
-	}
-
-	/* certain older asics have a separare 3D performance state,
-	 * so try that first if the user selected performance
-	 */
-	if (dpm_state == POWER_STATE_TYPE_PERFORMANCE)
-		dpm_state = POWER_STATE_TYPE_INTERNAL_3DPERF;
-	/* balanced states don't exist at the moment */
-	if (dpm_state == POWER_STATE_TYPE_BALANCED)
-		dpm_state = POWER_STATE_TYPE_PERFORMANCE;
-
-restart_search:
-	/* Pick the best power state based on current conditions */
-	for (i = 0; i < adev->pm.dpm.num_ps; i++) {
-		ps = &adev->pm.dpm.ps[i];
-		ui_class = ps->class & ATOM_PPLIB_CLASSIFICATION_UI_MASK;
-		switch (dpm_state) {
-		/* user states */
-		case POWER_STATE_TYPE_BATTERY:
-			if (ui_class == ATOM_PPLIB_CLASSIFICATION_UI_BATTERY) {
-				if (ps->caps & ATOM_PPLIB_SINGLE_DISPLAY_ONLY) {
-					if (single_display)
-						return ps;
-				} else
-					return ps;
-			}
-			break;
-		case POWER_STATE_TYPE_BALANCED:
-			if (ui_class == ATOM_PPLIB_CLASSIFICATION_UI_BALANCED) {
-				if (ps->caps & ATOM_PPLIB_SINGLE_DISPLAY_ONLY) {
-					if (single_display)
-						return ps;
-				} else
-					return ps;
-			}
-			break;
-		case POWER_STATE_TYPE_PERFORMANCE:
-			if (ui_class == ATOM_PPLIB_CLASSIFICATION_UI_PERFORMANCE) {
-				if (ps->caps & ATOM_PPLIB_SINGLE_DISPLAY_ONLY) {
-					if (single_display)
-						return ps;
-				} else
-					return ps;
-			}
-			break;
-		/* internal states */
-		case POWER_STATE_TYPE_INTERNAL_UVD:
-			if (adev->pm.dpm.uvd_ps)
-				return adev->pm.dpm.uvd_ps;
-			else
-				break;
-		case POWER_STATE_TYPE_INTERNAL_UVD_SD:
-			if (ps->class & ATOM_PPLIB_CLASSIFICATION_SDSTATE)
-				return ps;
-			break;
-		case POWER_STATE_TYPE_INTERNAL_UVD_HD:
-			if (ps->class & ATOM_PPLIB_CLASSIFICATION_HDSTATE)
-				return ps;
-			break;
-		case POWER_STATE_TYPE_INTERNAL_UVD_HD2:
-			if (ps->class & ATOM_PPLIB_CLASSIFICATION_HD2STATE)
-				return ps;
-			break;
-		case POWER_STATE_TYPE_INTERNAL_UVD_MVC:
-			if (ps->class2 & ATOM_PPLIB_CLASSIFICATION2_MVC)
-				return ps;
-			break;
-		case POWER_STATE_TYPE_INTERNAL_BOOT:
-			return adev->pm.dpm.boot_ps;
-		case POWER_STATE_TYPE_INTERNAL_THERMAL:
-			if (ps->class & ATOM_PPLIB_CLASSIFICATION_THERMAL)
-				return ps;
-			break;
-		case POWER_STATE_TYPE_INTERNAL_ACPI:
-			if (ps->class & ATOM_PPLIB_CLASSIFICATION_ACPI)
-				return ps;
-			break;
-		case POWER_STATE_TYPE_INTERNAL_ULV:
-			if (ps->class2 & ATOM_PPLIB_CLASSIFICATION2_ULV)
-				return ps;
-			break;
-		case POWER_STATE_TYPE_INTERNAL_3DPERF:
-			if (ps->class & ATOM_PPLIB_CLASSIFICATION_3DPERFORMANCE)
-				return ps;
-			break;
-		default:
-			break;
-		}
-	}
-	/* use a fallback state if we didn't match */
-	switch (dpm_state) {
-	case POWER_STATE_TYPE_INTERNAL_UVD_SD:
-		dpm_state = POWER_STATE_TYPE_INTERNAL_UVD_HD;
-		goto restart_search;
-	case POWER_STATE_TYPE_INTERNAL_UVD_HD:
-	case POWER_STATE_TYPE_INTERNAL_UVD_HD2:
-	case POWER_STATE_TYPE_INTERNAL_UVD_MVC:
-		if (adev->pm.dpm.uvd_ps) {
-			return adev->pm.dpm.uvd_ps;
-		} else {
-			dpm_state = POWER_STATE_TYPE_PERFORMANCE;
-			goto restart_search;
-		}
-	case POWER_STATE_TYPE_INTERNAL_THERMAL:
-		dpm_state = POWER_STATE_TYPE_INTERNAL_ACPI;
-		goto restart_search;
-	case POWER_STATE_TYPE_INTERNAL_ACPI:
-		dpm_state = POWER_STATE_TYPE_BATTERY;
-		goto restart_search;
-	case POWER_STATE_TYPE_BATTERY:
-	case POWER_STATE_TYPE_BALANCED:
-	case POWER_STATE_TYPE_INTERNAL_3DPERF:
-		dpm_state = POWER_STATE_TYPE_PERFORMANCE;
-		goto restart_search;
-	default:
-		break;
-	}
-
-	return NULL;
-}
-
-static void amdgpu_dpm_change_power_state_locked(struct amdgpu_device *adev)
-{
-	struct amdgpu_ps *ps;
-	enum amd_pm_state_type dpm_state;
-	int ret;
-	bool equal;
-
-	/* if dpm init failed */
-	if (!adev->pm.dpm_enabled)
-		return;
-
-	if (adev->pm.dpm.user_state != adev->pm.dpm.state) {
-		/* add other state override checks here */
-		if ((!adev->pm.dpm.thermal_active) &&
-		    (!adev->pm.dpm.uvd_active))
-			adev->pm.dpm.state = adev->pm.dpm.user_state;
-	}
-	dpm_state = adev->pm.dpm.state;
-
-	ps = amdgpu_dpm_pick_power_state(adev, dpm_state);
-	if (ps)
-		adev->pm.dpm.requested_ps = ps;
-	else
-		return;
-
-	if (amdgpu_dpm == 1) {
-		printk("switching from power state:\n");
-		amdgpu_dpm_print_power_state(adev, adev->pm.dpm.current_ps);
-		printk("switching to power state:\n");
-		amdgpu_dpm_print_power_state(adev, adev->pm.dpm.requested_ps);
-	}
-
-	/* update whether vce is active */
-	ps->vce_active = adev->pm.dpm.vce_active;
-
-	amdgpu_dpm_display_configuration_changed(adev);
-
-	ret = amdgpu_dpm_pre_set_power_state(adev);
-	if (ret)
-		return;
-
-	if ((0 != amgdpu_dpm_check_state_equal(adev, adev->pm.dpm.current_ps, adev->pm.dpm.requested_ps, &equal)))
-		equal = false;
-
-	if (equal)
-		return;
-
-	amdgpu_dpm_set_power_state(adev);
-	amdgpu_dpm_post_set_power_state(adev);
-
-	adev->pm.dpm.current_active_crtcs = adev->pm.dpm.new_active_crtcs;
-	adev->pm.dpm.current_active_crtc_count = adev->pm.dpm.new_active_crtc_count;
-
-	if (adev->pm.funcs->force_performance_level) {
-		if (adev->pm.dpm.thermal_active) {
-			enum amd_dpm_forced_level level = adev->pm.dpm.forced_level;
-			/* force low perf level for thermal */
-			amdgpu_dpm_force_performance_level(adev, AMD_DPM_FORCED_LEVEL_LOW);
-			/* save the user's level */
-			adev->pm.dpm.forced_level = level;
-		} else {
-			/* otherwise, user selected level */
-			amdgpu_dpm_force_performance_level(adev, adev->pm.dpm.forced_level);
-		}
-	}
-}
-
-void amdgpu_dpm_enable_uvd(struct amdgpu_device *adev, bool enable)
-{
-	if (adev->pp_enabled || adev->pm.funcs->powergate_uvd) {
-		/* enable/disable UVD */
-		mutex_lock(&adev->pm.mutex);
-		amdgpu_dpm_powergate_uvd(adev, !enable);
-		mutex_unlock(&adev->pm.mutex);
-	} else {
-		if (enable) {
-			mutex_lock(&adev->pm.mutex);
-			adev->pm.dpm.uvd_active = true;
-			adev->pm.dpm.state = POWER_STATE_TYPE_INTERNAL_UVD;
-			mutex_unlock(&adev->pm.mutex);
-		} else {
-			mutex_lock(&adev->pm.mutex);
-			adev->pm.dpm.uvd_active = false;
-			mutex_unlock(&adev->pm.mutex);
-		}
-		amdgpu_pm_compute_clocks(adev);
-	}
-}
-
-void amdgpu_dpm_enable_vce(struct amdgpu_device *adev, bool enable)
-{
-	if (adev->pp_enabled || adev->pm.funcs->powergate_vce) {
-		/* enable/disable VCE */
-		mutex_lock(&adev->pm.mutex);
-		amdgpu_dpm_powergate_vce(adev, !enable);
-		mutex_unlock(&adev->pm.mutex);
-	} else {
-		if (enable) {
-			mutex_lock(&adev->pm.mutex);
-			adev->pm.dpm.vce_active = true;
-			/* XXX select vce level based on ring/task */
-			adev->pm.dpm.vce_level = AMD_VCE_LEVEL_AC_ALL;
-			mutex_unlock(&adev->pm.mutex);
-			amdgpu_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
-							AMD_CG_STATE_UNGATE);
-			amdgpu_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
-							AMD_PG_STATE_UNGATE);
-			amdgpu_pm_compute_clocks(adev);
-		} else {
-			amdgpu_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
-							AMD_PG_STATE_GATE);
-			amdgpu_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
-							AMD_CG_STATE_GATE);
-			mutex_lock(&adev->pm.mutex);
-			adev->pm.dpm.vce_active = false;
-			mutex_unlock(&adev->pm.mutex);
-			amdgpu_pm_compute_clocks(adev);
-		}
-
-	}
-}
-
-void amdgpu_pm_print_power_states(struct amdgpu_device *adev)
-{
-	int i;
-
-	if (adev->pp_enabled)
-		/* TO DO */
-		return;
-
-	for (i = 0; i < adev->pm.dpm.num_ps; i++)
-		amdgpu_dpm_print_power_state(adev, &adev->pm.dpm.ps[i]);
-
-}
-
-int amdgpu_pm_sysfs_init(struct amdgpu_device *adev)
-{
-	int ret;
-
-	if (adev->pm.sysfs_initialized)
-		return 0;
-
-	if (!adev->pp_enabled) {
-		if (adev->pm.funcs->get_temperature == NULL)
-			return 0;
-	}
-
-	adev->pm.int_hwmon_dev = hwmon_device_register_with_groups(adev->dev,
-								   DRIVER_NAME, adev,
-								   hwmon_groups);
-	if (IS_ERR(adev->pm.int_hwmon_dev)) {
-		ret = PTR_ERR(adev->pm.int_hwmon_dev);
-		dev_err(adev->dev,
-			"Unable to register hwmon device: %d\n", ret);
-		return ret;
-	}
-
-	ret = device_create_file(adev->dev, &dev_attr_power_dpm_state);
-	if (ret) {
-		DRM_ERROR("failed to create device file for dpm state\n");
-		return ret;
-	}
-	ret = device_create_file(adev->dev, &dev_attr_power_dpm_force_performance_level);
-	if (ret) {
-		DRM_ERROR("failed to create device file for dpm state\n");
-		return ret;
-	}
-
-	if (adev->pp_enabled) {
-		ret = device_create_file(adev->dev, &dev_attr_pp_num_states);
-		if (ret) {
-			DRM_ERROR("failed to create device file pp_num_states\n");
-			return ret;
-		}
-		ret = device_create_file(adev->dev, &dev_attr_pp_cur_state);
-		if (ret) {
-			DRM_ERROR("failed to create device file pp_cur_state\n");
-			return ret;
-		}
-		ret = device_create_file(adev->dev, &dev_attr_pp_force_state);
-		if (ret) {
-			DRM_ERROR("failed to create device file pp_force_state\n");
-			return ret;
-		}
-		ret = device_create_file(adev->dev, &dev_attr_pp_table);
-		if (ret) {
-			DRM_ERROR("failed to create device file pp_table\n");
-			return ret;
-		}
-	}
-
-	ret = device_create_file(adev->dev, &dev_attr_pp_dpm_sclk);
-	if (ret) {
-		DRM_ERROR("failed to create device file pp_dpm_sclk\n");
-		return ret;
-	}
-	ret = device_create_file(adev->dev, &dev_attr_pp_dpm_mclk);
-	if (ret) {
-		DRM_ERROR("failed to create device file pp_dpm_mclk\n");
-		return ret;
-	}
-	ret = device_create_file(adev->dev, &dev_attr_pp_dpm_pcie);
-	if (ret) {
-		DRM_ERROR("failed to create device file pp_dpm_pcie\n");
-		return ret;
-	}
-	ret = device_create_file(adev->dev, &dev_attr_pp_sclk_od);
-	if (ret) {
-		DRM_ERROR("failed to create device file pp_sclk_od\n");
-		return ret;
-	}
-	ret = device_create_file(adev->dev, &dev_attr_pp_mclk_od);
-	if (ret) {
-		DRM_ERROR("failed to create device file pp_mclk_od\n");
-		return ret;
-	}
-	ret = device_create_file(adev->dev,
-			&dev_attr_pp_gfx_power_profile);
-	if (ret) {
-		DRM_ERROR("failed to create device file	"
-				"pp_gfx_power_profile\n");
-		return ret;
-	}
-	ret = device_create_file(adev->dev,
-			&dev_attr_pp_compute_power_profile);
-	if (ret) {
-		DRM_ERROR("failed to create device file	"
-				"pp_compute_power_profile\n");
-		return ret;
-	}
-
-	ret = amdgpu_debugfs_pm_init(adev);
-	if (ret) {
-		DRM_ERROR("Failed to register debugfs file for dpm!\n");
-		return ret;
-	}
-
-	adev->pm.sysfs_initialized = true;
-
-	return 0;
-}
-
-void amdgpu_pm_sysfs_fini(struct amdgpu_device *adev)
-{
-	if (adev->pm.int_hwmon_dev)
-		hwmon_device_unregister(adev->pm.int_hwmon_dev);
-	device_remove_file(adev->dev, &dev_attr_power_dpm_state);
-	device_remove_file(adev->dev, &dev_attr_power_dpm_force_performance_level);
-	if (adev->pp_enabled) {
-		device_remove_file(adev->dev, &dev_attr_pp_num_states);
-		device_remove_file(adev->dev, &dev_attr_pp_cur_state);
-		device_remove_file(adev->dev, &dev_attr_pp_force_state);
-		device_remove_file(adev->dev, &dev_attr_pp_table);
-	}
-	device_remove_file(adev->dev, &dev_attr_pp_dpm_sclk);
-	device_remove_file(adev->dev, &dev_attr_pp_dpm_mclk);
-	device_remove_file(adev->dev, &dev_attr_pp_dpm_pcie);
-	device_remove_file(adev->dev, &dev_attr_pp_sclk_od);
-	device_remove_file(adev->dev, &dev_attr_pp_mclk_od);
-	device_remove_file(adev->dev,
-			&dev_attr_pp_gfx_power_profile);
-	device_remove_file(adev->dev,
-			&dev_attr_pp_compute_power_profile);
-}
-
-void amdgpu_pm_compute_clocks(struct amdgpu_device *adev)
-{
-	struct drm_device *ddev = adev->ddev;
-	struct drm_crtc *crtc;
-	struct amdgpu_crtc *amdgpu_crtc;
-	int i = 0;
-
-	if (!adev->pm.dpm_enabled)
-		return;
-
-	if (adev->mode_info.num_crtc)
-		amdgpu_display_bandwidth_update(adev);
-
-	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
-		struct amdgpu_ring *ring = adev->rings[i];
-		if (ring && ring->ready)
-			amdgpu_fence_wait_empty(ring);
-	}
-
-	if (adev->pp_enabled) {
-		amdgpu_dpm_dispatch_task(adev, AMD_PP_EVENT_DISPLAY_CONFIG_CHANGE, NULL, NULL);
-	} else {
-		mutex_lock(&adev->pm.mutex);
-		adev->pm.dpm.new_active_crtcs = 0;
-		adev->pm.dpm.new_active_crtc_count = 0;
-		if (adev->mode_info.num_crtc && adev->mode_info.mode_config_initialized) {
-			list_for_each_entry(crtc,
-					    &ddev->mode_config.crtc_list, head) {
-				amdgpu_crtc = to_amdgpu_crtc(crtc);
-				if (crtc->enabled) {
-					adev->pm.dpm.new_active_crtcs |= (1 << amdgpu_crtc->crtc_id);
-					adev->pm.dpm.new_active_crtc_count++;
-				}
-			}
-		}
-		/* update battery/ac status */
-		if (power_supply_is_system_supplied() > 0)
-			adev->pm.dpm.ac_power = true;
-		else
-			adev->pm.dpm.ac_power = false;
-
-		amdgpu_dpm_change_power_state_locked(adev);
-
-		mutex_unlock(&adev->pm.mutex);
-	}
-}
-
-/*
- * Debugfs info
- */
-#if defined(CONFIG_DEBUG_FS)
-
-static int amdgpu_debugfs_pm_info_pp(struct seq_file *m, struct amdgpu_device *adev)
-{
-	uint32_t value;
-	struct pp_gpu_power query = {0};
-	int size;
-
-	/* sanity check PP is enabled */
-	if (!(adev->powerplay.pp_funcs &&
-	      adev->powerplay.pp_funcs->read_sensor))
-	      return -EINVAL;
-
-	/* GPU Clocks */
-	size = sizeof(value);
-	seq_printf(m, "GFX Clocks and Power:\n");
-	if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_GFX_MCLK, (void *)&value, &size))
-		seq_printf(m, "\t%u MHz (MCLK)\n", value/100);
-	if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_GFX_SCLK, (void *)&value, &size))
-		seq_printf(m, "\t%u MHz (SCLK)\n", value/100);
-	if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_VDDGFX, (void *)&value, &size))
-		seq_printf(m, "\t%u mV (VDDGFX)\n", value);
-	if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_VDDNB, (void *)&value, &size))
-		seq_printf(m, "\t%u mV (VDDNB)\n", value);
-	size = sizeof(query);
-	if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_GPU_POWER, (void *)&query, &size)) {
-		seq_printf(m, "\t%u.%u W (VDDC)\n", query.vddc_power >> 8,
-				query.vddc_power & 0xff);
-		seq_printf(m, "\t%u.%u W (VDDCI)\n", query.vddci_power >> 8,
-				query.vddci_power & 0xff);
-		seq_printf(m, "\t%u.%u W (max GPU)\n", query.max_gpu_power >> 8,
-				query.max_gpu_power & 0xff);
-		seq_printf(m, "\t%u.%u W (average GPU)\n", query.average_gpu_power >> 8,
-				query.average_gpu_power & 0xff);
-	}
-	size = sizeof(value);
-	seq_printf(m, "\n");
-
-	/* GPU Temp */
-	if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_GPU_TEMP, (void *)&value, &size))
-		seq_printf(m, "GPU Temperature: %u C\n", value/1000);
-
-	/* GPU Load */
-	if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_GPU_LOAD, (void *)&value, &size))
-		seq_printf(m, "GPU Load: %u %%\n", value);
-	seq_printf(m, "\n");
-
-	/* UVD clocks */
-	if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_UVD_POWER, (void *)&value, &size)) {
-		if (!value) {
-			seq_printf(m, "UVD: Disabled\n");
-		} else {
-			seq_printf(m, "UVD: Enabled\n");
-			if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_UVD_DCLK, (void *)&value, &size))
-				seq_printf(m, "\t%u MHz (DCLK)\n", value/100);
-			if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_UVD_VCLK, (void *)&value, &size))
-				seq_printf(m, "\t%u MHz (VCLK)\n", value/100);
-		}
-	}
-	seq_printf(m, "\n");
-
-	/* VCE clocks */
-	if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_VCE_POWER, (void *)&value, &size)) {
-		if (!value) {
-			seq_printf(m, "VCE: Disabled\n");
-		} else {
-			seq_printf(m, "VCE: Enabled\n");
-			if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_VCE_ECCLK, (void *)&value, &size))
-				seq_printf(m, "\t%u MHz (ECCLK)\n", value/100);
-		}
-	}
-
-	return 0;
-}
-
-static void amdgpu_parse_cg_state(struct seq_file *m, u32 flags)
-{
-	int i;
-
-	for (i = 0; clocks[i].flag; i++)
-		seq_printf(m, "\t%s: %s\n", clocks[i].name,
-			   (flags & clocks[i].flag) ? "On" : "Off");
-}
-
-static int amdgpu_debugfs_pm_info(struct seq_file *m, void *data)
-{
-	struct drm_info_node *node = (struct drm_info_node *) m->private;
-	struct drm_device *dev = node->minor->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-	struct drm_device *ddev = adev->ddev;
-	u32 flags = 0;
-
-	amdgpu_get_clockgating_state(adev, &flags);
-	seq_printf(m, "Clock Gating Flags Mask: 0x%x\n", flags);
-	amdgpu_parse_cg_state(m, flags);
-	seq_printf(m, "\n");
-
-	if (!adev->pm.dpm_enabled) {
-		seq_printf(m, "dpm not enabled\n");
-		return 0;
-	}
-	if  ((adev->flags & AMD_IS_PX) &&
-	     (ddev->switch_power_state != DRM_SWITCH_POWER_ON)) {
-		seq_printf(m, "PX asic powered off\n");
-	} else if (adev->pp_enabled) {
-		return amdgpu_debugfs_pm_info_pp(m, adev);
-	} else {
-		mutex_lock(&adev->pm.mutex);
-		if (adev->pm.funcs->debugfs_print_current_performance_level)
-			adev->pm.funcs->debugfs_print_current_performance_level(adev, m);
-		else
-			seq_printf(m, "Debugfs support not implemented for this asic\n");
-		mutex_unlock(&adev->pm.mutex);
-	}
-
-	return 0;
-}
-
-static const struct drm_info_list amdgpu_pm_info_list[] = {
-	{"amdgpu_pm_info", amdgpu_debugfs_pm_info, 0, NULL},
-};
-#endif
-
-static int amdgpu_debugfs_pm_init(struct amdgpu_device *adev)
-{
-#if defined(CONFIG_DEBUG_FS)
-	return amdgpu_debugfs_add_files(adev, amdgpu_pm_info_list, ARRAY_SIZE(amdgpu_pm_info_list));
-#else
-	return 0;
-#endif
-}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.h
deleted file mode 100644
index f21a7716b90e..000000000000
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.h
+++ /dev/null
@@ -1,42 +0,0 @@
-/*
- * Copyright 2014 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-
-#ifndef __AMDGPU_PM_H__
-#define __AMDGPU_PM_H__
-
-struct cg_flag_name
-{
-	u32 flag;
-	const char *name;
-};
-
-void amdgpu_pm_acpi_event_handler(struct amdgpu_device *adev);
-int amdgpu_pm_sysfs_init(struct amdgpu_device *adev);
-void amdgpu_pm_sysfs_fini(struct amdgpu_device *adev);
-void amdgpu_pm_print_power_states(struct amdgpu_device *adev);
-void amdgpu_pm_compute_clocks(struct amdgpu_device *adev);
-void amdgpu_dpm_thermal_work_handler(struct work_struct *work);
-void amdgpu_dpm_enable_uvd(struct amdgpu_device *adev, bool enable);
-void amdgpu_dpm_enable_vce(struct amdgpu_device *adev, bool enable);
-
-#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c
new file mode 100644
index 000000000000..6e91ea1de5aa
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c
@@ -0,0 +1,680 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/perf_event.h>
+#include <linux/init.h>
+#include "amdgpu.h"
+#include "amdgpu_pmu.h"
+
+#define PMU_NAME_SIZE 32
+#define NUM_FORMATS_AMDGPU_PMU		4
+#define NUM_FORMATS_DF_VEGA20		3
+#define NUM_EVENTS_DF_VEGA20		8
+#define NUM_EVENT_TYPES_VEGA20		1
+#define NUM_EVENTS_VEGA20_XGMI		2
+#define NUM_EVENTS_VEGA20_MAX		NUM_EVENTS_VEGA20_XGMI
+#define NUM_EVENT_TYPES_ARCTURUS	1
+#define NUM_EVENTS_ARCTURUS_XGMI	6
+#define NUM_EVENTS_ARCTURUS_MAX		NUM_EVENTS_ARCTURUS_XGMI
+
+struct amdgpu_pmu_event_attribute {
+	struct device_attribute attr;
+	const char *event_str;
+	unsigned int type;
+};
+
+/* record to keep track of pmu entry per pmu type per device */
+struct amdgpu_pmu_entry {
+	struct list_head entry;
+	struct amdgpu_device *adev;
+	struct pmu pmu;
+	unsigned int pmu_perf_type;
+	char *pmu_type_name;
+	char *pmu_file_prefix;
+	struct attribute_group fmt_attr_group;
+	struct amdgpu_pmu_event_attribute *fmt_attr;
+	struct attribute_group evt_attr_group;
+	struct amdgpu_pmu_event_attribute *evt_attr;
+};
+
+static ssize_t amdgpu_pmu_event_show(struct device *dev,
+				struct device_attribute *attr, char *buf)
+{
+	struct amdgpu_pmu_event_attribute *amdgpu_pmu_attr;
+
+	amdgpu_pmu_attr = container_of(attr, struct amdgpu_pmu_event_attribute,
+									attr);
+
+	if (!amdgpu_pmu_attr->type)
+		return sprintf(buf, "%s\n", amdgpu_pmu_attr->event_str);
+
+	return sprintf(buf, "%s,type=0x%x\n",
+			amdgpu_pmu_attr->event_str, amdgpu_pmu_attr->type);
+}
+
+static LIST_HEAD(amdgpu_pmu_list);
+
+
+struct amdgpu_pmu_attr {
+	const char *name;
+	const char *config;
+};
+
+struct amdgpu_pmu_type {
+	const unsigned int type;
+	const unsigned int num_of_type;
+};
+
+struct amdgpu_pmu_config {
+	struct amdgpu_pmu_attr *formats;
+	unsigned int num_formats;
+	struct amdgpu_pmu_attr *events;
+	unsigned int num_events;
+	struct amdgpu_pmu_type *types;
+	unsigned int num_types;
+};
+
+/*
+ * Events fall under two categories:
+ *  - PMU typed
+ *    Events in /sys/bus/event_source/devices/amdgpu_<pmu_type>_<dev_num> have
+ *    performance counter operations handled by one IP <pmu_type>.  Formats and
+ *    events should be defined by <pmu_type>_<asic_type>_formats and
+ *    <pmu_type>_<asic_type>_events respectively.
+ *
+ *  - Event config typed
+ *    Events in /sys/bus/event_source/devices/amdgpu_<dev_num> have performance
+ *    counter operations that can be handled by multiple IPs dictated by their
+ *    "type" format field.  Formats and events should be defined by
+ *    amdgpu_pmu_formats and <asic_type>_events respectively.  Format field
+ *    "type" is generated in amdgpu_pmu_event_show and defined in
+ *    <asic_type>_event_config_types.
+ */
+
+static struct amdgpu_pmu_attr amdgpu_pmu_formats[NUM_FORMATS_AMDGPU_PMU] = {
+	{ .name = "event", .config = "config:0-7" },
+	{ .name = "instance", .config = "config:8-15" },
+	{ .name = "umask", .config = "config:16-23"},
+	{ .name = "type", .config = "config:56-63"}
+};
+
+/* Vega20 events */
+static struct amdgpu_pmu_attr vega20_events[NUM_EVENTS_VEGA20_MAX] = {
+	{ .name = "xgmi_link0_data_outbound",
+			.config = "event=0x7,instance=0x46,umask=0x2" },
+	{ .name = "xgmi_link1_data_outbound",
+			.config = "event=0x7,instance=0x47,umask=0x2" }
+};
+
+static struct amdgpu_pmu_type vega20_types[NUM_EVENT_TYPES_VEGA20] = {
+	{ .type = AMDGPU_PMU_EVENT_CONFIG_TYPE_XGMI,
+					.num_of_type = NUM_EVENTS_VEGA20_XGMI }
+};
+
+static struct amdgpu_pmu_config vega20_config = {
+	.formats = amdgpu_pmu_formats,
+	.num_formats = ARRAY_SIZE(amdgpu_pmu_formats),
+	.events = vega20_events,
+	.num_events = ARRAY_SIZE(vega20_events),
+	.types = vega20_types,
+	.num_types = ARRAY_SIZE(vega20_types)
+};
+
+/* Vega20 data fabric (DF) events */
+static struct amdgpu_pmu_attr df_vega20_formats[NUM_FORMATS_DF_VEGA20] = {
+	{ .name = "event", .config = "config:0-7" },
+	{ .name = "instance", .config = "config:8-15" },
+	{ .name = "umask", .config = "config:16-23"}
+};
+
+static struct amdgpu_pmu_attr df_vega20_events[NUM_EVENTS_DF_VEGA20] = {
+	{ .name = "cake0_pcsout_txdata",
+			.config = "event=0x7,instance=0x46,umask=0x2" },
+	{ .name = "cake1_pcsout_txdata",
+			.config = "event=0x7,instance=0x47,umask=0x2" },
+	{ .name = "cake0_pcsout_txmeta",
+			.config = "event=0x7,instance=0x46,umask=0x4" },
+	{ .name = "cake1_pcsout_txmeta",
+			.config = "event=0x7,instance=0x47,umask=0x4" },
+	{ .name = "cake0_ftiinstat_reqalloc",
+			.config = "event=0xb,instance=0x46,umask=0x4" },
+	{ .name = "cake1_ftiinstat_reqalloc",
+			.config = "event=0xb,instance=0x47,umask=0x4" },
+	{ .name = "cake0_ftiinstat_rspalloc",
+			.config = "event=0xb,instance=0x46,umask=0x8" },
+	{ .name = "cake1_ftiinstat_rspalloc",
+			.config = "event=0xb,instance=0x47,umask=0x8" }
+};
+
+static struct amdgpu_pmu_config df_vega20_config = {
+	.formats = df_vega20_formats,
+	.num_formats = ARRAY_SIZE(df_vega20_formats),
+	.events = df_vega20_events,
+	.num_events = ARRAY_SIZE(df_vega20_events),
+	.types = NULL,
+	.num_types = 0
+};
+
+/* Arcturus events */
+static struct amdgpu_pmu_attr arcturus_events[NUM_EVENTS_ARCTURUS_MAX] = {
+	{ .name = "xgmi_link0_data_outbound",
+			.config = "event=0x7,instance=0x4b,umask=0x2" },
+	{ .name = "xgmi_link1_data_outbound",
+			.config = "event=0x7,instance=0x4c,umask=0x2" },
+	{ .name = "xgmi_link2_data_outbound",
+			.config = "event=0x7,instance=0x4d,umask=0x2" },
+	{ .name = "xgmi_link3_data_outbound",
+			.config = "event=0x7,instance=0x4e,umask=0x2" },
+	{ .name = "xgmi_link4_data_outbound",
+			.config = "event=0x7,instance=0x4f,umask=0x2" },
+	{ .name = "xgmi_link5_data_outbound",
+			.config = "event=0x7,instance=0x50,umask=0x2" }
+};
+
+static struct amdgpu_pmu_type arcturus_types[NUM_EVENT_TYPES_ARCTURUS] = {
+	{ .type = AMDGPU_PMU_EVENT_CONFIG_TYPE_XGMI,
+				.num_of_type = NUM_EVENTS_ARCTURUS_XGMI }
+};
+
+static struct amdgpu_pmu_config arcturus_config = {
+	.formats = amdgpu_pmu_formats,
+	.num_formats = ARRAY_SIZE(amdgpu_pmu_formats),
+	.events = arcturus_events,
+	.num_events = ARRAY_SIZE(arcturus_events),
+	.types = arcturus_types,
+	.num_types = ARRAY_SIZE(arcturus_types)
+};
+
+/* initialize perf counter */
+static int amdgpu_perf_event_init(struct perf_event *event)
+{
+	struct hw_perf_event *hwc = &event->hw;
+
+	/* test the event attr type check for PMU enumeration */
+	if (event->attr.type != event->pmu->type)
+		return -ENOENT;
+
+	/* update the hw_perf_event struct with config data */
+	hwc->config = event->attr.config;
+	hwc->config_base = AMDGPU_PMU_PERF_TYPE_NONE;
+
+	return 0;
+}
+
+/* start perf counter */
+static void amdgpu_perf_start(struct perf_event *event, int flags)
+{
+	struct hw_perf_event *hwc = &event->hw;
+	struct amdgpu_pmu_entry *pe = container_of(event->pmu,
+						  struct amdgpu_pmu_entry,
+						  pmu);
+	int target_cntr = 0;
+
+	if (WARN_ON_ONCE(!(hwc->state & PERF_HES_STOPPED)))
+		return;
+
+	if ((!pe->adev->df.funcs) ||
+	    (!pe->adev->df.funcs->pmc_start))
+		return;
+
+	WARN_ON_ONCE(!(hwc->state & PERF_HES_UPTODATE));
+	hwc->state = 0;
+
+	switch (hwc->config_base) {
+	case AMDGPU_PMU_EVENT_CONFIG_TYPE_DF:
+	case AMDGPU_PMU_EVENT_CONFIG_TYPE_XGMI:
+		if (!(flags & PERF_EF_RELOAD)) {
+			target_cntr = pe->adev->df.funcs->pmc_start(pe->adev,
+						hwc->config, 0 /* unused */,
+						1 /* add counter */);
+			if (target_cntr < 0)
+				break;
+
+			hwc->idx = target_cntr;
+		}
+
+		pe->adev->df.funcs->pmc_start(pe->adev, hwc->config,
+								hwc->idx, 0);
+		break;
+	default:
+		break;
+	}
+
+	perf_event_update_userpage(event);
+}
+
+/* read perf counter */
+static void amdgpu_perf_read(struct perf_event *event)
+{
+	struct hw_perf_event *hwc = &event->hw;
+	struct amdgpu_pmu_entry *pe = container_of(event->pmu,
+						  struct amdgpu_pmu_entry,
+						  pmu);
+	u64 count, prev;
+
+	if ((!pe->adev->df.funcs) ||
+	    (!pe->adev->df.funcs->pmc_get_count))
+		return;
+
+	prev = local64_read(&hwc->prev_count);
+	do {
+		switch (hwc->config_base) {
+		case AMDGPU_PMU_EVENT_CONFIG_TYPE_DF:
+		case AMDGPU_PMU_EVENT_CONFIG_TYPE_XGMI:
+			pe->adev->df.funcs->pmc_get_count(pe->adev,
+						hwc->config, hwc->idx, &count);
+			break;
+		default:
+			count = 0;
+			break;
+		}
+	} while (!local64_try_cmpxchg(&hwc->prev_count, &prev, count));
+
+	local64_add(count - prev, &event->count);
+}
+
+/* stop perf counter */
+static void amdgpu_perf_stop(struct perf_event *event, int flags)
+{
+	struct hw_perf_event *hwc = &event->hw;
+	struct amdgpu_pmu_entry *pe = container_of(event->pmu,
+						  struct amdgpu_pmu_entry,
+						  pmu);
+
+	if (hwc->state & PERF_HES_UPTODATE)
+		return;
+
+	if ((!pe->adev->df.funcs) ||
+	    (!pe->adev->df.funcs->pmc_stop))
+		return;
+
+	switch (hwc->config_base) {
+	case AMDGPU_PMU_EVENT_CONFIG_TYPE_DF:
+	case AMDGPU_PMU_EVENT_CONFIG_TYPE_XGMI:
+		pe->adev->df.funcs->pmc_stop(pe->adev, hwc->config, hwc->idx,
+									0);
+		break;
+	default:
+		break;
+	}
+
+	WARN_ON_ONCE(hwc->state & PERF_HES_STOPPED);
+	hwc->state |= PERF_HES_STOPPED;
+
+	if (hwc->state & PERF_HES_UPTODATE)
+		return;
+
+	amdgpu_perf_read(event);
+	hwc->state |= PERF_HES_UPTODATE;
+}
+
+/* add perf counter */
+static int amdgpu_perf_add(struct perf_event *event, int flags)
+{
+	struct hw_perf_event *hwc = &event->hw;
+	int retval = 0, target_cntr;
+	struct amdgpu_pmu_entry *pe = container_of(event->pmu,
+						  struct amdgpu_pmu_entry,
+						  pmu);
+
+	if ((!pe->adev->df.funcs) ||
+	    (!pe->adev->df.funcs->pmc_start))
+		return -EINVAL;
+
+	switch (pe->pmu_perf_type) {
+	case AMDGPU_PMU_PERF_TYPE_DF:
+		hwc->config_base = AMDGPU_PMU_EVENT_CONFIG_TYPE_DF;
+		break;
+	case AMDGPU_PMU_PERF_TYPE_ALL:
+		hwc->config_base = (hwc->config >>
+					AMDGPU_PMU_EVENT_CONFIG_TYPE_SHIFT) &
+					AMDGPU_PMU_EVENT_CONFIG_TYPE_MASK;
+		break;
+	}
+
+	event->hw.state = PERF_HES_UPTODATE | PERF_HES_STOPPED;
+
+	switch (hwc->config_base) {
+	case AMDGPU_PMU_EVENT_CONFIG_TYPE_DF:
+	case AMDGPU_PMU_EVENT_CONFIG_TYPE_XGMI:
+		target_cntr = pe->adev->df.funcs->pmc_start(pe->adev,
+						hwc->config, 0 /* unused */,
+						1 /* add counter */);
+		if (target_cntr < 0)
+			retval = target_cntr;
+		else
+			hwc->idx = target_cntr;
+
+		break;
+	default:
+		return 0;
+	}
+
+	if (retval)
+		return retval;
+
+	if (flags & PERF_EF_START)
+		amdgpu_perf_start(event, PERF_EF_RELOAD);
+
+	return retval;
+}
+
+/* delete perf counter  */
+static void amdgpu_perf_del(struct perf_event *event, int flags)
+{
+	struct hw_perf_event *hwc = &event->hw;
+	struct amdgpu_pmu_entry *pe = container_of(event->pmu,
+						  struct amdgpu_pmu_entry,
+						  pmu);
+	if ((!pe->adev->df.funcs) ||
+	    (!pe->adev->df.funcs->pmc_stop))
+		return;
+
+	amdgpu_perf_stop(event, PERF_EF_UPDATE);
+
+	switch (hwc->config_base) {
+	case AMDGPU_PMU_EVENT_CONFIG_TYPE_DF:
+	case AMDGPU_PMU_EVENT_CONFIG_TYPE_XGMI:
+		pe->adev->df.funcs->pmc_stop(pe->adev, hwc->config, hwc->idx,
+									1);
+		break;
+	default:
+		break;
+	}
+
+	perf_event_update_userpage(event);
+}
+
+static void amdgpu_pmu_create_event_attrs_by_type(
+				struct attribute_group *attr_group,
+				struct amdgpu_pmu_event_attribute *pmu_attr,
+				struct amdgpu_pmu_attr events[],
+				int s_offset,
+				int e_offset,
+				unsigned int type)
+{
+	int i;
+
+	pmu_attr += s_offset;
+
+	for (i = s_offset; i < e_offset; i++) {
+		attr_group->attrs[i] = &pmu_attr->attr.attr;
+		sysfs_attr_init(&pmu_attr->attr.attr);
+		pmu_attr->attr.attr.name = events[i].name;
+		pmu_attr->attr.attr.mode = 0444;
+		pmu_attr->attr.show = amdgpu_pmu_event_show;
+		pmu_attr->event_str = events[i].config;
+		pmu_attr->type = type;
+		pmu_attr++;
+	}
+}
+
+static void amdgpu_pmu_create_attrs(struct attribute_group *attr_group,
+				struct amdgpu_pmu_event_attribute *pmu_attr,
+				struct amdgpu_pmu_attr events[],
+				int num_events)
+{
+	amdgpu_pmu_create_event_attrs_by_type(attr_group, pmu_attr, events, 0,
+				num_events, AMDGPU_PMU_EVENT_CONFIG_TYPE_NONE);
+}
+
+
+static int amdgpu_pmu_alloc_pmu_attrs(
+				struct attribute_group *fmt_attr_group,
+				struct amdgpu_pmu_event_attribute **fmt_attr,
+				struct attribute_group *evt_attr_group,
+				struct amdgpu_pmu_event_attribute **evt_attr,
+				struct amdgpu_pmu_config *config)
+{
+	*fmt_attr = kcalloc(config->num_formats, sizeof(**fmt_attr),
+								GFP_KERNEL);
+
+	if (!(*fmt_attr))
+		return -ENOMEM;
+
+	fmt_attr_group->attrs = kcalloc(config->num_formats + 1,
+				sizeof(*fmt_attr_group->attrs), GFP_KERNEL);
+
+	if (!fmt_attr_group->attrs)
+		goto err_fmt_attr_grp;
+
+	*evt_attr = kcalloc(config->num_events, sizeof(**evt_attr), GFP_KERNEL);
+
+	if (!(*evt_attr))
+		goto err_evt_attr;
+
+	evt_attr_group->attrs = kcalloc(config->num_events + 1,
+				sizeof(*evt_attr_group->attrs), GFP_KERNEL);
+
+	if (!evt_attr_group->attrs)
+		goto err_evt_attr_grp;
+
+	return 0;
+err_evt_attr_grp:
+	kfree(*evt_attr);
+err_evt_attr:
+	kfree(fmt_attr_group->attrs);
+err_fmt_attr_grp:
+	kfree(*fmt_attr);
+	return -ENOMEM;
+}
+
+/* init pmu tracking per pmu type */
+static int init_pmu_entry_by_type_and_add(struct amdgpu_pmu_entry *pmu_entry,
+			struct amdgpu_pmu_config *config)
+{
+	const struct attribute_group *attr_groups[] = {
+		&pmu_entry->fmt_attr_group,
+		&pmu_entry->evt_attr_group,
+		NULL
+	};
+	char pmu_name[PMU_NAME_SIZE];
+	int ret = 0, total_num_events = 0;
+
+	pmu_entry->pmu = (struct pmu){
+		.event_init = amdgpu_perf_event_init,
+		.add = amdgpu_perf_add,
+		.del = amdgpu_perf_del,
+		.start = amdgpu_perf_start,
+		.stop = amdgpu_perf_stop,
+		.read = amdgpu_perf_read,
+		.task_ctx_nr = perf_invalid_context,
+	};
+
+	ret = amdgpu_pmu_alloc_pmu_attrs(&pmu_entry->fmt_attr_group,
+					&pmu_entry->fmt_attr,
+					&pmu_entry->evt_attr_group,
+					&pmu_entry->evt_attr,
+					config);
+
+	if (ret)
+		goto err_out;
+
+	amdgpu_pmu_create_attrs(&pmu_entry->fmt_attr_group, pmu_entry->fmt_attr,
+					config->formats, config->num_formats);
+
+	if (pmu_entry->pmu_perf_type == AMDGPU_PMU_PERF_TYPE_ALL) {
+		int i;
+
+		for (i = 0; i < config->num_types; i++) {
+			amdgpu_pmu_create_event_attrs_by_type(
+					&pmu_entry->evt_attr_group,
+					pmu_entry->evt_attr,
+					config->events,
+					total_num_events,
+					total_num_events +
+						config->types[i].num_of_type,
+					config->types[i].type);
+			total_num_events += config->types[i].num_of_type;
+		}
+	} else {
+		amdgpu_pmu_create_attrs(&pmu_entry->evt_attr_group,
+					pmu_entry->evt_attr,
+					config->events, config->num_events);
+		total_num_events = config->num_events;
+	}
+
+	pmu_entry->pmu.attr_groups = kmemdup(attr_groups, sizeof(attr_groups),
+								GFP_KERNEL);
+
+	if (!pmu_entry->pmu.attr_groups) {
+		ret = -ENOMEM;
+		goto err_attr_group;
+	}
+
+	snprintf(pmu_name, PMU_NAME_SIZE, "%s_%d", pmu_entry->pmu_file_prefix,
+				adev_to_drm(pmu_entry->adev)->primary->index);
+
+	ret = perf_pmu_register(&pmu_entry->pmu, pmu_name, -1);
+
+	if (ret)
+		goto err_register;
+
+	if (pmu_entry->pmu_perf_type != AMDGPU_PMU_PERF_TYPE_ALL)
+		pr_info("Detected AMDGPU %s Counters. # of Counters = %d.\n",
+				pmu_entry->pmu_type_name, total_num_events);
+	else
+		pr_info("Detected AMDGPU %d Perf Events.\n", total_num_events);
+
+
+	list_add_tail(&pmu_entry->entry, &amdgpu_pmu_list);
+
+	return 0;
+err_register:
+	kfree(pmu_entry->pmu.attr_groups);
+err_attr_group:
+	kfree(pmu_entry->fmt_attr_group.attrs);
+	kfree(pmu_entry->fmt_attr);
+	kfree(pmu_entry->evt_attr_group.attrs);
+	kfree(pmu_entry->evt_attr);
+err_out:
+	pr_warn("Error initializing AMDGPU %s PMUs.\n",
+						pmu_entry->pmu_type_name);
+	return ret;
+}
+
+/* destroy all pmu data associated with target device */
+void amdgpu_pmu_fini(struct amdgpu_device *adev)
+{
+	struct amdgpu_pmu_entry *pe, *temp;
+
+	list_for_each_entry_safe(pe, temp, &amdgpu_pmu_list, entry) {
+		if (pe->adev != adev)
+			continue;
+		list_del(&pe->entry);
+		perf_pmu_unregister(&pe->pmu);
+		kfree(pe->pmu.attr_groups);
+		kfree(pe->fmt_attr_group.attrs);
+		kfree(pe->fmt_attr);
+		kfree(pe->evt_attr_group.attrs);
+		kfree(pe->evt_attr);
+		kfree(pe);
+	}
+}
+
+static struct amdgpu_pmu_entry *create_pmu_entry(struct amdgpu_device *adev,
+						unsigned int pmu_type,
+						char *pmu_type_name,
+						char *pmu_file_prefix)
+{
+	struct amdgpu_pmu_entry *pmu_entry;
+
+	pmu_entry = kzalloc(sizeof(struct amdgpu_pmu_entry), GFP_KERNEL);
+
+	if (!pmu_entry)
+		return pmu_entry;
+
+	pmu_entry->adev = adev;
+	pmu_entry->fmt_attr_group.name = "format";
+	pmu_entry->fmt_attr_group.attrs = NULL;
+	pmu_entry->evt_attr_group.name = "events";
+	pmu_entry->evt_attr_group.attrs = NULL;
+	pmu_entry->pmu_perf_type = pmu_type;
+	pmu_entry->pmu_type_name = pmu_type_name;
+	pmu_entry->pmu_file_prefix = pmu_file_prefix;
+
+	return pmu_entry;
+}
+
+/* init amdgpu_pmu */
+int amdgpu_pmu_init(struct amdgpu_device *adev)
+{
+	int ret = 0;
+	struct amdgpu_pmu_entry *pmu_entry, *pmu_entry_df;
+
+	switch (adev->asic_type) {
+	case CHIP_VEGA20:
+		pmu_entry_df = create_pmu_entry(adev, AMDGPU_PMU_PERF_TYPE_DF,
+						"DF", "amdgpu_df");
+
+		if (!pmu_entry_df)
+			return -ENOMEM;
+
+		ret = init_pmu_entry_by_type_and_add(pmu_entry_df,
+							&df_vega20_config);
+
+		if (ret) {
+			kfree(pmu_entry_df);
+			return ret;
+		}
+
+		pmu_entry = create_pmu_entry(adev, AMDGPU_PMU_PERF_TYPE_ALL,
+						"", "amdgpu");
+
+		if (!pmu_entry) {
+			amdgpu_pmu_fini(adev);
+			return -ENOMEM;
+		}
+
+		ret = init_pmu_entry_by_type_and_add(pmu_entry,
+							&vega20_config);
+
+		if (ret) {
+			kfree(pmu_entry);
+			amdgpu_pmu_fini(adev);
+			return ret;
+		}
+
+		break;
+	case CHIP_ARCTURUS:
+		pmu_entry = create_pmu_entry(adev, AMDGPU_PMU_PERF_TYPE_ALL,
+						"", "amdgpu");
+		if (!pmu_entry)
+			return -ENOMEM;
+
+		ret = init_pmu_entry_by_type_and_add(pmu_entry,
+							&arcturus_config);
+
+		if (ret) {
+			kfree(pmu_entry);
+			return -ENOMEM;
+		}
+
+		break;
+
+	default:
+		return 0;
+	}
+
+	return ret;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.h
new file mode 100644
index 000000000000..6882dc48c5d8
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.h
@@ -0,0 +1,57 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef _AMDGPU_PMU_H_
+#define _AMDGPU_PMU_H_
+
+/* PMU types. */
+enum amdgpu_pmu_perf_type {
+	AMDGPU_PMU_PERF_TYPE_NONE = 0,
+	AMDGPU_PMU_PERF_TYPE_DF,
+	AMDGPU_PMU_PERF_TYPE_ALL
+};
+
+/*
+ * PMU type AMDGPU_PMU_PERF_TYPE_ALL can hold events of different "type"
+ * configurations.  Event config types are parsed from the 64-bit raw
+ * config (See EVENT_CONFIG_TYPE_SHIFT and EVENT_CONFIG_TYPE_MASK) and
+ * are registered into the HW perf events config_base.
+ *
+ * PMU types with only a single event configuration type
+ * (non-AMDGPU_PMU_PERF_TYPE_ALL) have their event config type auto generated
+ * when the performance counter is added.
+ */
+enum amdgpu_pmu_event_config_type {
+	AMDGPU_PMU_EVENT_CONFIG_TYPE_NONE = 0,
+	AMDGPU_PMU_EVENT_CONFIG_TYPE_DF,
+	AMDGPU_PMU_EVENT_CONFIG_TYPE_XGMI,
+	AMDGPU_PMU_EVENT_CONFIG_TYPE_MAX
+};
+
+#define AMDGPU_PMU_EVENT_CONFIG_TYPE_SHIFT	56
+#define AMDGPU_PMU_EVENT_CONFIG_TYPE_MASK	0xff
+
+int amdgpu_pmu_init(struct amdgpu_device *adev);
+void amdgpu_pmu_fini(struct amdgpu_device *adev);
+
+#endif /* _AMDGPU_PMU_H_ */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c
deleted file mode 100644
index b7e1c026c0c8..000000000000
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c
+++ /dev/null
@@ -1,322 +0,0 @@
-/*
- * Copyright 2015 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- * Authors: AMD
- *
- */
-#include "atom.h"
-#include "amdgpu.h"
-#include "amd_shared.h"
-#include <linux/module.h>
-#include <linux/moduleparam.h>
-#include "amdgpu_pm.h"
-#include <drm/amdgpu_drm.h>
-#include "amdgpu_powerplay.h"
-#include "si_dpm.h"
-#include "cik_dpm.h"
-#include "vi_dpm.h"
-
-static int amdgpu_create_pp_handle(struct amdgpu_device *adev)
-{
-	struct amd_pp_init pp_init;
-	struct amd_powerplay *amd_pp;
-	int ret;
-
-	amd_pp = &(adev->powerplay);
-	pp_init.chip_family = adev->family;
-	pp_init.chip_id = adev->asic_type;
-	pp_init.pm_en = (amdgpu_dpm != 0 && !amdgpu_sriov_vf(adev)) ? true : false;
-	pp_init.feature_mask = amdgpu_pp_feature_mask;
-	pp_init.device = amdgpu_cgs_create_device(adev);
-	ret = amd_powerplay_create(&pp_init, &(amd_pp->pp_handle));
-	if (ret)
-		return -EINVAL;
-	return 0;
-}
-
-static int amdgpu_pp_early_init(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	struct amd_powerplay *amd_pp;
-	int ret = 0;
-
-	amd_pp = &(adev->powerplay);
-	adev->pp_enabled = false;
-	amd_pp->pp_handle = (void *)adev;
-
-	switch (adev->asic_type) {
-	case CHIP_POLARIS11:
-	case CHIP_POLARIS10:
-	case CHIP_POLARIS12:
-	case CHIP_TONGA:
-	case CHIP_FIJI:
-	case CHIP_TOPAZ:
-	case CHIP_CARRIZO:
-	case CHIP_STONEY:
-	case CHIP_VEGA10:
-	case CHIP_RAVEN:
-		adev->pp_enabled = true;
-		if (amdgpu_create_pp_handle(adev))
-			return -EINVAL;
-		amd_pp->ip_funcs = &pp_ip_funcs;
-		amd_pp->pp_funcs = &pp_dpm_funcs;
-		break;
-	/* These chips don't have powerplay implemenations */
-#ifdef CONFIG_DRM_AMDGPU_SI
-	case CHIP_TAHITI:
-	case CHIP_PITCAIRN:
-	case CHIP_VERDE:
-	case CHIP_OLAND:
-	case CHIP_HAINAN:
-		amd_pp->ip_funcs = &si_dpm_ip_funcs;
-	break;
-#endif
-#ifdef CONFIG_DRM_AMDGPU_CIK
-	case CHIP_BONAIRE:
-	case CHIP_HAWAII:
-		amd_pp->ip_funcs = &ci_dpm_ip_funcs;
-		break;
-	case CHIP_KABINI:
-	case CHIP_MULLINS:
-	case CHIP_KAVERI:
-		amd_pp->ip_funcs = &kv_dpm_ip_funcs;
-		break;
-#endif
-	default:
-		ret = -EINVAL;
-		break;
-	}
-
-	if (adev->powerplay.ip_funcs->early_init)
-		ret = adev->powerplay.ip_funcs->early_init(
-					adev->powerplay.pp_handle);
-
-	if (ret == PP_DPM_DISABLED) {
-		adev->pm.dpm_enabled = false;
-		return 0;
-	}
-	return ret;
-}
-
-
-static int amdgpu_pp_late_init(void *handle)
-{
-	int ret = 0;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (adev->powerplay.ip_funcs->late_init)
-		ret = adev->powerplay.ip_funcs->late_init(
-					adev->powerplay.pp_handle);
-
-	if (adev->pp_enabled && adev->pm.dpm_enabled) {
-		amdgpu_pm_sysfs_init(adev);
-		amdgpu_dpm_dispatch_task(adev, AMD_PP_EVENT_COMPLETE_INIT, NULL, NULL);
-	}
-
-	return ret;
-}
-
-static int amdgpu_pp_sw_init(void *handle)
-{
-	int ret = 0;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (adev->powerplay.ip_funcs->sw_init)
-		ret = adev->powerplay.ip_funcs->sw_init(
-					adev->powerplay.pp_handle);
-
-	return ret;
-}
-
-static int amdgpu_pp_sw_fini(void *handle)
-{
-	int ret = 0;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (adev->powerplay.ip_funcs->sw_fini)
-		ret = adev->powerplay.ip_funcs->sw_fini(
-					adev->powerplay.pp_handle);
-	if (ret)
-		return ret;
-
-	return ret;
-}
-
-static int amdgpu_pp_hw_init(void *handle)
-{
-	int ret = 0;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (adev->pp_enabled && adev->firmware.load_type == AMDGPU_FW_LOAD_SMU)
-		amdgpu_ucode_init_bo(adev);
-
-	if (adev->powerplay.ip_funcs->hw_init)
-		ret = adev->powerplay.ip_funcs->hw_init(
-					adev->powerplay.pp_handle);
-
-	if (ret == PP_DPM_DISABLED) {
-		adev->pm.dpm_enabled = false;
-		return 0;
-	}
-
-	if ((amdgpu_dpm != 0) && !amdgpu_sriov_vf(adev))
-		adev->pm.dpm_enabled = true;
-
-	return ret;
-}
-
-static int amdgpu_pp_hw_fini(void *handle)
-{
-	int ret = 0;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (adev->pp_enabled && adev->pm.dpm_enabled)
-		amdgpu_pm_sysfs_fini(adev);
-
-	if (adev->powerplay.ip_funcs->hw_fini)
-		ret = adev->powerplay.ip_funcs->hw_fini(
-					adev->powerplay.pp_handle);
-
-	if (adev->pp_enabled && adev->firmware.load_type == AMDGPU_FW_LOAD_SMU)
-		amdgpu_ucode_fini_bo(adev);
-
-	return ret;
-}
-
-static void amdgpu_pp_late_fini(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (adev->powerplay.ip_funcs->late_fini)
-		adev->powerplay.ip_funcs->late_fini(
-			  adev->powerplay.pp_handle);
-
-
-	if (adev->pp_enabled)
-		amd_powerplay_destroy(adev->powerplay.pp_handle);
-}
-
-static int amdgpu_pp_suspend(void *handle)
-{
-	int ret = 0;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (adev->powerplay.ip_funcs->suspend)
-		ret = adev->powerplay.ip_funcs->suspend(
-					 adev->powerplay.pp_handle);
-	return ret;
-}
-
-static int amdgpu_pp_resume(void *handle)
-{
-	int ret = 0;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (adev->powerplay.ip_funcs->resume)
-		ret = adev->powerplay.ip_funcs->resume(
-					adev->powerplay.pp_handle);
-	return ret;
-}
-
-static int amdgpu_pp_set_clockgating_state(void *handle,
-					enum amd_clockgating_state state)
-{
-	int ret = 0;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (adev->powerplay.ip_funcs->set_clockgating_state)
-		ret = adev->powerplay.ip_funcs->set_clockgating_state(
-				adev->powerplay.pp_handle, state);
-	return ret;
-}
-
-static int amdgpu_pp_set_powergating_state(void *handle,
-					enum amd_powergating_state state)
-{
-	int ret = 0;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (adev->powerplay.ip_funcs->set_powergating_state)
-		ret = adev->powerplay.ip_funcs->set_powergating_state(
-				 adev->powerplay.pp_handle, state);
-	return ret;
-}
-
-
-static bool amdgpu_pp_is_idle(void *handle)
-{
-	bool ret = true;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (adev->powerplay.ip_funcs->is_idle)
-		ret = adev->powerplay.ip_funcs->is_idle(
-					adev->powerplay.pp_handle);
-	return ret;
-}
-
-static int amdgpu_pp_wait_for_idle(void *handle)
-{
-	int ret = 0;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (adev->powerplay.ip_funcs->wait_for_idle)
-		ret = adev->powerplay.ip_funcs->wait_for_idle(
-					adev->powerplay.pp_handle);
-	return ret;
-}
-
-static int amdgpu_pp_soft_reset(void *handle)
-{
-	int ret = 0;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (adev->powerplay.ip_funcs->soft_reset)
-		ret = adev->powerplay.ip_funcs->soft_reset(
-					adev->powerplay.pp_handle);
-	return ret;
-}
-
-static const struct amd_ip_funcs amdgpu_pp_ip_funcs = {
-	.name = "amdgpu_powerplay",
-	.early_init = amdgpu_pp_early_init,
-	.late_init = amdgpu_pp_late_init,
-	.sw_init = amdgpu_pp_sw_init,
-	.sw_fini = amdgpu_pp_sw_fini,
-	.hw_init = amdgpu_pp_hw_init,
-	.hw_fini = amdgpu_pp_hw_fini,
-	.late_fini = amdgpu_pp_late_fini,
-	.suspend = amdgpu_pp_suspend,
-	.resume = amdgpu_pp_resume,
-	.is_idle = amdgpu_pp_is_idle,
-	.wait_for_idle = amdgpu_pp_wait_for_idle,
-	.soft_reset = amdgpu_pp_soft_reset,
-	.set_clockgating_state = amdgpu_pp_set_clockgating_state,
-	.set_powergating_state = amdgpu_pp_set_powergating_state,
-};
-
-const struct amdgpu_ip_block_version amdgpu_pp_ip_block =
-{
-	.type = AMD_IP_BLOCK_TYPE_SMC,
-	.major = 1,
-	.minor = 0,
-	.rev = 0,
-	.funcs = &amdgpu_pp_ip_funcs,
-};
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.h
deleted file mode 100644
index c0c4bfdcdb14..000000000000
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.h
+++ /dev/null
@@ -1,33 +0,0 @@
-/*
- *  Copyright 2015 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- * Authors: AMD
- *
- */
-
-#ifndef __AMDGPU_POWERPLAY_H__
-#define __AMDGPU_POWERPLAY_H__
-
-#include "amd_shared.h"
-
-extern const struct amdgpu_ip_block_version amdgpu_pp_ip_block;
-
-#endif /* __AMDGPU_POWERPLAY_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_preempt_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_preempt_mgr.c
new file mode 100644
index 000000000000..34b5e22b44e5
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_preempt_mgr.c
@@ -0,0 +1,145 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
+/*
+ * Copyright 2016-2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Christian König, Felix Kuehling
+ */
+
+#include "amdgpu.h"
+
+/**
+ * DOC: mem_info_preempt_used
+ *
+ * The amdgpu driver provides a sysfs API for reporting current total amount of
+ * used preemptible memory.
+ * The file mem_info_preempt_used is used for this, and returns the current
+ * used size of the preemptible block, in bytes
+ */
+static ssize_t mem_info_preempt_used_show(struct device *dev,
+					  struct device_attribute *attr,
+					  char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	struct ttm_resource_manager *man = &adev->mman.preempt_mgr;
+
+	return sysfs_emit(buf, "%llu\n", ttm_resource_manager_usage(man));
+}
+
+static DEVICE_ATTR_RO(mem_info_preempt_used);
+
+/**
+ * amdgpu_preempt_mgr_new - allocate a new node
+ *
+ * @man: TTM memory type manager
+ * @tbo: TTM BO we need this range for
+ * @place: placement flags and restrictions
+ * @res: TTM memory object
+ *
+ * Dummy, just count the space used without allocating resources or any limit.
+ */
+static int amdgpu_preempt_mgr_new(struct ttm_resource_manager *man,
+				  struct ttm_buffer_object *tbo,
+				  const struct ttm_place *place,
+				  struct ttm_resource **res)
+{
+	*res = kzalloc(sizeof(**res), GFP_KERNEL);
+	if (!*res)
+		return -ENOMEM;
+
+	ttm_resource_init(tbo, place, *res);
+	(*res)->start = AMDGPU_BO_INVALID_OFFSET;
+	return 0;
+}
+
+/**
+ * amdgpu_preempt_mgr_del - free ranges
+ *
+ * @man: TTM memory type manager
+ * @res: TTM memory object
+ *
+ * Free the allocated GTT again.
+ */
+static void amdgpu_preempt_mgr_del(struct ttm_resource_manager *man,
+				   struct ttm_resource *res)
+{
+	ttm_resource_fini(man, res);
+	kfree(res);
+}
+
+static const struct ttm_resource_manager_func amdgpu_preempt_mgr_func = {
+	.alloc = amdgpu_preempt_mgr_new,
+	.free = amdgpu_preempt_mgr_del,
+};
+
+/**
+ * amdgpu_preempt_mgr_init - init PREEMPT manager and DRM MM
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate and initialize the GTT manager.
+ */
+int amdgpu_preempt_mgr_init(struct amdgpu_device *adev)
+{
+	struct ttm_resource_manager *man = &adev->mman.preempt_mgr;
+	int ret;
+
+	man->use_tt = true;
+	man->func = &amdgpu_preempt_mgr_func;
+
+	ttm_resource_manager_init(man, &adev->mman.bdev, (1 << 30));
+
+	ret = device_create_file(adev->dev, &dev_attr_mem_info_preempt_used);
+	if (ret) {
+		DRM_ERROR("Failed to create device file mem_info_preempt_used\n");
+		return ret;
+	}
+
+	ttm_set_driver_manager(&adev->mman.bdev, AMDGPU_PL_PREEMPT, man);
+	ttm_resource_manager_set_used(man, true);
+	return 0;
+}
+
+/**
+ * amdgpu_preempt_mgr_fini - free and destroy GTT manager
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Destroy and free the GTT manager, returns -EBUSY if ranges are still
+ * allocated inside it.
+ */
+void amdgpu_preempt_mgr_fini(struct amdgpu_device *adev)
+{
+	struct ttm_resource_manager *man = &adev->mman.preempt_mgr;
+	int ret;
+
+	ttm_resource_manager_set_used(man, false);
+
+	ret = ttm_resource_manager_evict_all(&adev->mman.bdev, man);
+	if (ret)
+		return;
+
+	if (adev->dev->kobj.sd)
+		device_remove_file(adev->dev, &dev_attr_mem_info_preempt_used);
+
+	ttm_resource_manager_cleanup(man);
+	ttm_set_driver_manager(&adev->mman.bdev, AMDGPU_PL_PREEMPT, NULL);
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c
deleted file mode 100644
index 5b3f92891f89..000000000000
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c
+++ /dev/null
@@ -1,143 +0,0 @@
-/*
- * Copyright 2012 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- * based on nouveau_prime.c
- *
- * Authors: Alex Deucher
- */
-#include <drm/drmP.h>
-
-#include "amdgpu.h"
-#include <drm/amdgpu_drm.h>
-#include <linux/dma-buf.h>
-
-struct sg_table *amdgpu_gem_prime_get_sg_table(struct drm_gem_object *obj)
-{
-	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
-	int npages = bo->tbo.num_pages;
-
-	return drm_prime_pages_to_sg(bo->tbo.ttm->pages, npages);
-}
-
-void *amdgpu_gem_prime_vmap(struct drm_gem_object *obj)
-{
-	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
-	int ret;
-
-	ret = ttm_bo_kmap(&bo->tbo, 0, bo->tbo.num_pages,
-			  &bo->dma_buf_vmap);
-	if (ret)
-		return ERR_PTR(ret);
-
-	return bo->dma_buf_vmap.virtual;
-}
-
-void amdgpu_gem_prime_vunmap(struct drm_gem_object *obj, void *vaddr)
-{
-	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
-
-	ttm_bo_kunmap(&bo->dma_buf_vmap);
-}
-
-struct drm_gem_object *
-amdgpu_gem_prime_import_sg_table(struct drm_device *dev,
-				 struct dma_buf_attachment *attach,
-				 struct sg_table *sg)
-{
-	struct reservation_object *resv = attach->dmabuf->resv;
-	struct amdgpu_device *adev = dev->dev_private;
-	struct amdgpu_bo *bo;
-	int ret;
-
-	ww_mutex_lock(&resv->lock, NULL);
-	ret = amdgpu_bo_create(adev, attach->dmabuf->size, PAGE_SIZE, false,
-			       AMDGPU_GEM_DOMAIN_GTT, 0, sg, resv, 0, &bo);
-	ww_mutex_unlock(&resv->lock);
-	if (ret)
-		return ERR_PTR(ret);
-
-	bo->prime_shared_count = 1;
-	return &bo->gem_base;
-}
-
-int amdgpu_gem_prime_pin(struct drm_gem_object *obj)
-{
-	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
-	long ret = 0;
-
-	ret = amdgpu_bo_reserve(bo, false);
-	if (unlikely(ret != 0))
-		return ret;
-
-	/*
-	 * Wait for all shared fences to complete before we switch to future
-	 * use of exclusive fence on this prime shared bo.
-	 */
-	ret = reservation_object_wait_timeout_rcu(bo->tbo.resv, true, false,
-						  MAX_SCHEDULE_TIMEOUT);
-	if (unlikely(ret < 0)) {
-		DRM_DEBUG_PRIME("Fence wait failed: %li\n", ret);
-		amdgpu_bo_unreserve(bo);
-		return ret;
-	}
-
-	/* pin buffer into GTT */
-	ret = amdgpu_bo_pin(bo, AMDGPU_GEM_DOMAIN_GTT, NULL);
-	if (likely(ret == 0))
-		bo->prime_shared_count++;
-
-	amdgpu_bo_unreserve(bo);
-	return ret;
-}
-
-void amdgpu_gem_prime_unpin(struct drm_gem_object *obj)
-{
-	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
-	int ret = 0;
-
-	ret = amdgpu_bo_reserve(bo, true);
-	if (unlikely(ret != 0))
-		return;
-
-	amdgpu_bo_unpin(bo);
-	if (bo->prime_shared_count)
-		bo->prime_shared_count--;
-	amdgpu_bo_unreserve(bo);
-}
-
-struct reservation_object *amdgpu_gem_prime_res_obj(struct drm_gem_object *obj)
-{
-	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
-
-	return bo->tbo.resv;
-}
-
-struct dma_buf *amdgpu_gem_prime_export(struct drm_device *dev,
-					struct drm_gem_object *gobj,
-					int flags)
-{
-	struct amdgpu_bo *bo = gem_to_amdgpu_bo(gobj);
-
-	if (amdgpu_ttm_tt_get_usermm(bo->tbo.ttm))
-		return ERR_PTR(-EPERM);
-
-	return drm_gem_prime_export(dev, gobj, flags);
-}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
index 8c2204c7b384..8c0e5d03de50 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
@@ -24,54 +24,240 @@
  */
 
 #include <linux/firmware.h>
-#include <drm/drmP.h>
+#include <drm/drm_drv.h>
+
 #include "amdgpu.h"
 #include "amdgpu_psp.h"
 #include "amdgpu_ucode.h"
+#include "amdgpu_xgmi.h"
 #include "soc15_common.h"
 #include "psp_v3_1.h"
 #include "psp_v10_0.h"
+#include "psp_v11_0.h"
+#include "psp_v11_0_8.h"
+#include "psp_v12_0.h"
+#include "psp_v13_0.h"
+#include "psp_v13_0_4.h"
+#include "psp_v14_0.h"
+
+#include "amdgpu_ras.h"
+#include "amdgpu_securedisplay.h"
+#include "amdgpu_atomfirmware.h"
 
-static void psp_set_funcs(struct amdgpu_device *adev);
+#define AMD_VBIOS_FILE_MAX_SIZE_B      (1024*1024*16)
 
-static int psp_early_init(void *handle)
+static int psp_load_smu_fw(struct psp_context *psp);
+static int psp_rap_terminate(struct psp_context *psp);
+static int psp_securedisplay_terminate(struct psp_context *psp);
+
+static int psp_ring_init(struct psp_context *psp,
+			 enum psp_ring_type ring_type)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int ret = 0;
+	struct psp_ring *ring;
+	struct amdgpu_device *adev = psp->adev;
 
-	psp_set_funcs(adev);
+	ring = &psp->km_ring;
+
+	ring->ring_type = ring_type;
+
+	/* allocate 4k Page of Local Frame Buffer memory for ring */
+	ring->ring_size = 0x1000;
+	ret = amdgpu_bo_create_kernel(adev, ring->ring_size, PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_VRAM |
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->firmware.rbuf,
+				      &ring->ring_mem_mc_addr,
+				      (void **)&ring->ring_mem);
+	if (ret) {
+		ring->ring_size = 0;
+		return ret;
+	}
 
 	return 0;
 }
 
-static int psp_sw_init(void *handle)
+/*
+ * Due to DF Cstate management centralized to PMFW, the firmware
+ * loading sequence will be updated as below:
+ *   - Load KDB
+ *   - Load SYS_DRV
+ *   - Load tOS
+ *   - Load PMFW
+ *   - Setup TMR
+ *   - Load other non-psp fw
+ *   - Load ASD
+ *   - Load XGMI/RAS/HDCP/DTM TA if any
+ *
+ * This new sequence is required for
+ *   - Arcturus and onwards
+ */
+static void psp_check_pmfw_centralized_cstate_management(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_sriov_vf(adev)) {
+		psp->pmfw_centralized_cstate_management = false;
+		return;
+	}
+
+	switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) {
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 4):
+	case IP_VERSION(11, 0, 5):
+	case IP_VERSION(11, 0, 7):
+	case IP_VERSION(11, 0, 9):
+	case IP_VERSION(11, 0, 11):
+	case IP_VERSION(11, 0, 12):
+	case IP_VERSION(11, 0, 13):
+	case IP_VERSION(13, 0, 0):
+	case IP_VERSION(13, 0, 2):
+	case IP_VERSION(13, 0, 7):
+		psp->pmfw_centralized_cstate_management = true;
+		break;
+	default:
+		psp->pmfw_centralized_cstate_management = false;
+		break;
+	}
+}
+
+static int psp_init_sriov_microcode(struct psp_context *psp)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = psp->adev;
+	char ucode_prefix[30];
+	int ret = 0;
+
+	amdgpu_ucode_ip_version_decode(adev, MP0_HWIP, ucode_prefix, sizeof(ucode_prefix));
+
+	switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) {
+	case IP_VERSION(9, 0, 0):
+	case IP_VERSION(11, 0, 7):
+	case IP_VERSION(11, 0, 9):
+		adev->virt.autoload_ucode_id = AMDGPU_UCODE_ID_CP_MEC2;
+		ret = psp_init_cap_microcode(psp, ucode_prefix);
+		break;
+	case IP_VERSION(13, 0, 2):
+		adev->virt.autoload_ucode_id = AMDGPU_UCODE_ID_CP_MEC2;
+		ret = psp_init_cap_microcode(psp, ucode_prefix);
+		ret &= psp_init_ta_microcode(psp, ucode_prefix);
+		break;
+	case IP_VERSION(13, 0, 0):
+		adev->virt.autoload_ucode_id = 0;
+		break;
+	case IP_VERSION(13, 0, 6):
+	case IP_VERSION(13, 0, 14):
+		ret = psp_init_cap_microcode(psp, ucode_prefix);
+		ret &= psp_init_ta_microcode(psp, ucode_prefix);
+		break;
+	case IP_VERSION(13, 0, 10):
+		adev->virt.autoload_ucode_id = AMDGPU_UCODE_ID_CP_MES1_DATA;
+		ret = psp_init_cap_microcode(psp, ucode_prefix);
+		break;
+	case IP_VERSION(13, 0, 12):
+		ret = psp_init_ta_microcode(psp, ucode_prefix);
+		break;
+	default:
+		return -EINVAL;
+	}
+	return ret;
+}
+
+static int psp_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
 	struct psp_context *psp = &adev->psp;
-	int ret;
 
-	switch (adev->asic_type) {
-	case CHIP_VEGA10:
-		psp->init_microcode = psp_v3_1_init_microcode;
-		psp->bootloader_load_sysdrv = psp_v3_1_bootloader_load_sysdrv;
-		psp->bootloader_load_sos = psp_v3_1_bootloader_load_sos;
-		psp->prep_cmd_buf = psp_v3_1_prep_cmd_buf;
-		psp->ring_init = psp_v3_1_ring_init;
-		psp->ring_create = psp_v3_1_ring_create;
-		psp->ring_destroy = psp_v3_1_ring_destroy;
-		psp->cmd_submit = psp_v3_1_cmd_submit;
-		psp->compare_sram_data = psp_v3_1_compare_sram_data;
-		psp->smu_reload_quirk = psp_v3_1_smu_reload_quirk;
-		break;
-	case CHIP_RAVEN:
-#if 0
-		psp->init_microcode = psp_v10_0_init_microcode;
-#endif
-		psp->prep_cmd_buf = psp_v10_0_prep_cmd_buf;
-		psp->ring_init = psp_v10_0_ring_init;
-		psp->ring_create = psp_v10_0_ring_create;
-		psp->ring_destroy = psp_v10_0_ring_destroy;
-		psp->cmd_submit = psp_v10_0_cmd_submit;
-		psp->compare_sram_data = psp_v10_0_compare_sram_data;
+	psp->autoload_supported = true;
+	psp->boot_time_tmr = true;
+
+	switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) {
+	case IP_VERSION(9, 0, 0):
+		psp_v3_1_set_psp_funcs(psp);
+		psp->autoload_supported = false;
+		psp->boot_time_tmr = false;
+		break;
+	case IP_VERSION(10, 0, 0):
+	case IP_VERSION(10, 0, 1):
+		psp_v10_0_set_psp_funcs(psp);
+		psp->autoload_supported = false;
+		psp->boot_time_tmr = false;
+		break;
+	case IP_VERSION(11, 0, 2):
+	case IP_VERSION(11, 0, 4):
+		psp_v11_0_set_psp_funcs(psp);
+		psp->autoload_supported = false;
+		psp->boot_time_tmr = false;
+		break;
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 7):
+		adev->psp.sup_pd_fw_up = !amdgpu_sriov_vf(adev);
+		fallthrough;
+	case IP_VERSION(11, 0, 5):
+	case IP_VERSION(11, 0, 9):
+	case IP_VERSION(11, 0, 11):
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 0, 12):
+	case IP_VERSION(11, 0, 13):
+		psp_v11_0_set_psp_funcs(psp);
+		psp->boot_time_tmr = false;
+		break;
+	case IP_VERSION(11, 0, 3):
+	case IP_VERSION(12, 0, 1):
+		psp_v12_0_set_psp_funcs(psp);
+		psp->autoload_supported = false;
+		psp->boot_time_tmr = false;
+		break;
+	case IP_VERSION(13, 0, 2):
+		psp->boot_time_tmr = false;
+		fallthrough;
+	case IP_VERSION(13, 0, 6):
+	case IP_VERSION(13, 0, 14):
+		psp_v13_0_set_psp_funcs(psp);
+		psp->autoload_supported = false;
+		break;
+	case IP_VERSION(13, 0, 12):
+		psp_v13_0_set_psp_funcs(psp);
+		psp->autoload_supported = false;
+		adev->psp.sup_ifwi_up = !amdgpu_sriov_vf(adev);
+		break;
+	case IP_VERSION(13, 0, 1):
+	case IP_VERSION(13, 0, 3):
+	case IP_VERSION(13, 0, 5):
+	case IP_VERSION(13, 0, 8):
+	case IP_VERSION(13, 0, 11):
+	case IP_VERSION(14, 0, 0):
+	case IP_VERSION(14, 0, 1):
+	case IP_VERSION(14, 0, 4):
+		psp_v13_0_set_psp_funcs(psp);
+		psp->boot_time_tmr = false;
+		break;
+	case IP_VERSION(11, 0, 8):
+		if (adev->apu_flags & AMD_APU_IS_CYAN_SKILLFISH2) {
+			psp_v11_0_8_set_psp_funcs(psp);
+		}
+		psp->autoload_supported = false;
+		psp->boot_time_tmr = false;
+		break;
+	case IP_VERSION(13, 0, 0):
+	case IP_VERSION(13, 0, 7):
+	case IP_VERSION(13, 0, 10):
+		psp_v13_0_set_psp_funcs(psp);
+		adev->psp.sup_ifwi_up = !amdgpu_sriov_vf(adev);
+		psp->boot_time_tmr = false;
+		break;
+	case IP_VERSION(13, 0, 4):
+		psp_v13_0_4_set_psp_funcs(psp);
+		psp->boot_time_tmr = false;
+		break;
+	case IP_VERSION(14, 0, 2):
+	case IP_VERSION(14, 0, 3):
+		adev->psp.sup_ifwi_up = !amdgpu_sriov_vf(adev);
+		psp_v14_0_set_psp_funcs(psp);
+		break;
+	case IP_VERSION(14, 0, 5):
+		psp_v14_0_set_psp_funcs(psp);
+		psp->boot_time_tmr = false;
 		break;
 	default:
 		return -EINVAL;
@@ -79,27 +265,328 @@ static int psp_sw_init(void *handle)
 
 	psp->adev = adev;
 
-	ret = psp_init_microcode(psp);
-	if (ret) {
-		DRM_ERROR("Failed to load psp firmware!\n");
-		return ret;
+	adev->psp_timeout = 20000;
+
+	psp_check_pmfw_centralized_cstate_management(psp);
+
+	if (amdgpu_sriov_vf(adev))
+		return psp_init_sriov_microcode(psp);
+	else
+		return psp_init_microcode(psp);
+}
+
+void psp_ta_free_shared_buf(struct ta_mem_context *mem_ctx)
+{
+	amdgpu_bo_free_kernel(&mem_ctx->shared_bo, &mem_ctx->shared_mc_addr,
+			      &mem_ctx->shared_buf);
+	mem_ctx->shared_bo = NULL;
+}
+
+static void psp_free_shared_bufs(struct psp_context *psp)
+{
+	void *tmr_buf;
+	void **pptr;
+
+	/* free TMR memory buffer */
+	pptr = amdgpu_sriov_vf(psp->adev) ? &tmr_buf : NULL;
+	amdgpu_bo_free_kernel(&psp->tmr_bo, &psp->tmr_mc_addr, pptr);
+	psp->tmr_bo = NULL;
+
+	/* free xgmi shared memory */
+	psp_ta_free_shared_buf(&psp->xgmi_context.context.mem_context);
+
+	/* free ras shared memory */
+	psp_ta_free_shared_buf(&psp->ras_context.context.mem_context);
+
+	/* free hdcp shared memory */
+	psp_ta_free_shared_buf(&psp->hdcp_context.context.mem_context);
+
+	/* free dtm shared memory */
+	psp_ta_free_shared_buf(&psp->dtm_context.context.mem_context);
+
+	/* free rap shared memory */
+	psp_ta_free_shared_buf(&psp->rap_context.context.mem_context);
+
+	/* free securedisplay shared memory */
+	psp_ta_free_shared_buf(&psp->securedisplay_context.context.mem_context);
+
+
+}
+
+static void psp_memory_training_fini(struct psp_context *psp)
+{
+	struct psp_memory_training_context *ctx = &psp->mem_train_ctx;
+
+	ctx->init = PSP_MEM_TRAIN_NOT_SUPPORT;
+	kfree(ctx->sys_cache);
+	ctx->sys_cache = NULL;
+}
+
+static int psp_memory_training_init(struct psp_context *psp)
+{
+	int ret;
+	struct psp_memory_training_context *ctx = &psp->mem_train_ctx;
+
+	if (ctx->init != PSP_MEM_TRAIN_RESERVE_SUCCESS) {
+		dev_dbg(psp->adev->dev, "memory training is not supported!\n");
+		return 0;
+	}
+
+	ctx->sys_cache = kzalloc(ctx->train_data_size, GFP_KERNEL);
+	if (ctx->sys_cache == NULL) {
+		dev_err(psp->adev->dev, "alloc mem_train_ctx.sys_cache failed!\n");
+		ret = -ENOMEM;
+		goto Err_out;
 	}
 
+	dev_dbg(psp->adev->dev,
+		"train_data_size:%llx,p2c_train_data_offset:%llx,c2p_train_data_offset:%llx.\n",
+		ctx->train_data_size,
+		ctx->p2c_train_data_offset,
+		ctx->c2p_train_data_offset);
+	ctx->init = PSP_MEM_TRAIN_INIT_SUCCESS;
 	return 0;
+
+Err_out:
+	psp_memory_training_fini(psp);
+	return ret;
 }
 
-static int psp_sw_fini(void *handle)
+/*
+ * Helper funciton to query psp runtime database entry
+ *
+ * @adev: amdgpu_device pointer
+ * @entry_type: the type of psp runtime database entry
+ * @db_entry: runtime database entry pointer
+ *
+ * Return false if runtime database doesn't exit or entry is invalid
+ * or true if the specific database entry is found, and copy to @db_entry
+ */
+static bool psp_get_runtime_db_entry(struct amdgpu_device *adev,
+				     enum psp_runtime_entry_type entry_type,
+				     void *db_entry)
 {
+	uint64_t db_header_pos, db_dir_pos;
+	struct psp_runtime_data_header db_header = {0};
+	struct psp_runtime_data_directory db_dir = {0};
+	bool ret = false;
+	int i;
+
+	if (amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 6) ||
+	    amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 12) ||
+	    amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 14))
+		return false;
+
+	db_header_pos = adev->gmc.mc_vram_size - PSP_RUNTIME_DB_OFFSET;
+	db_dir_pos = db_header_pos + sizeof(struct psp_runtime_data_header);
+
+	/* read runtime db header from vram */
+	amdgpu_device_vram_access(adev, db_header_pos, (uint32_t *)&db_header,
+			sizeof(struct psp_runtime_data_header), false);
+
+	if (db_header.cookie != PSP_RUNTIME_DB_COOKIE_ID) {
+		/* runtime db doesn't exist, exit */
+		dev_dbg(adev->dev, "PSP runtime database doesn't exist\n");
+		return false;
+	}
+
+	/* read runtime database entry from vram */
+	amdgpu_device_vram_access(adev, db_dir_pos, (uint32_t *)&db_dir,
+			sizeof(struct psp_runtime_data_directory), false);
+
+	if (db_dir.entry_count >= PSP_RUNTIME_DB_DIAG_ENTRY_MAX_COUNT) {
+		/* invalid db entry count, exit */
+		dev_warn(adev->dev, "Invalid PSP runtime database entry count\n");
+		return false;
+	}
+
+	/* look up for requested entry type */
+	for (i = 0; i < db_dir.entry_count && !ret; i++) {
+		if (db_dir.entry_list[i].entry_type == entry_type) {
+			switch (entry_type) {
+			case PSP_RUNTIME_ENTRY_TYPE_BOOT_CONFIG:
+				if (db_dir.entry_list[i].size < sizeof(struct psp_runtime_boot_cfg_entry)) {
+					/* invalid db entry size */
+					dev_warn(adev->dev, "Invalid PSP runtime database boot cfg entry size\n");
+					return false;
+				}
+				/* read runtime database entry */
+				amdgpu_device_vram_access(adev, db_header_pos + db_dir.entry_list[i].offset,
+							  (uint32_t *)db_entry, sizeof(struct psp_runtime_boot_cfg_entry), false);
+				ret = true;
+				break;
+			case PSP_RUNTIME_ENTRY_TYPE_PPTABLE_ERR_STATUS:
+				if (db_dir.entry_list[i].size < sizeof(struct psp_runtime_scpm_entry)) {
+					/* invalid db entry size */
+					dev_warn(adev->dev, "Invalid PSP runtime database scpm entry size\n");
+					return false;
+				}
+				/* read runtime database entry */
+				amdgpu_device_vram_access(adev, db_header_pos + db_dir.entry_list[i].offset,
+							  (uint32_t *)db_entry, sizeof(struct psp_runtime_scpm_entry), false);
+				ret = true;
+				break;
+			default:
+				ret = false;
+				break;
+			}
+		}
+	}
+
+	return ret;
+}
+
+static int psp_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct psp_context *psp = &adev->psp;
+	int ret;
+	struct psp_runtime_boot_cfg_entry boot_cfg_entry;
+	struct psp_memory_training_context *mem_training_ctx = &psp->mem_train_ctx;
+	struct psp_runtime_scpm_entry scpm_entry;
+
+	psp->cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL);
+	if (!psp->cmd) {
+		dev_err(adev->dev, "Failed to allocate memory to command buffer!\n");
+		return -ENOMEM;
+	}
+
+	adev->psp.xgmi_context.supports_extended_data =
+		!adev->gmc.xgmi.connected_to_cpu &&
+		amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 2);
+
+	memset(&scpm_entry, 0, sizeof(scpm_entry));
+	if ((psp_get_runtime_db_entry(adev,
+				PSP_RUNTIME_ENTRY_TYPE_PPTABLE_ERR_STATUS,
+				&scpm_entry)) &&
+	    (scpm_entry.scpm_status != SCPM_DISABLE)) {
+		adev->scpm_enabled = true;
+		adev->scpm_status = scpm_entry.scpm_status;
+	} else {
+		adev->scpm_enabled = false;
+		adev->scpm_status = SCPM_DISABLE;
+	}
+
+	/* TODO: stop gpu driver services and print alarm if scpm is enabled with error status */
+
+	memset(&boot_cfg_entry, 0, sizeof(boot_cfg_entry));
+	if (psp_get_runtime_db_entry(adev,
+				PSP_RUNTIME_ENTRY_TYPE_BOOT_CONFIG,
+				&boot_cfg_entry)) {
+		psp->boot_cfg_bitmask = boot_cfg_entry.boot_cfg_bitmask;
+		if ((psp->boot_cfg_bitmask) &
+		    BOOT_CFG_FEATURE_TWO_STAGE_DRAM_TRAINING) {
+			/* If psp runtime database exists, then
+			 * only enable two stage memory training
+			 * when TWO_STAGE_DRAM_TRAINING bit is set
+			 * in runtime database
+			 */
+			mem_training_ctx->enable_mem_training = true;
+		}
+
+	} else {
+		/* If psp runtime database doesn't exist or is
+		 * invalid, force enable two stage memory training
+		 */
+		mem_training_ctx->enable_mem_training = true;
+	}
+
+	if (mem_training_ctx->enable_mem_training) {
+		ret = psp_memory_training_init(psp);
+		if (ret) {
+			dev_err(adev->dev, "Failed to initialize memory training!\n");
+			return ret;
+		}
+
+		ret = psp_mem_training(psp, PSP_MEM_TRAIN_COLD_BOOT);
+		if (ret) {
+			dev_err(adev->dev, "Failed to process memory training!\n");
+			return ret;
+		}
+	}
+
+	ret = amdgpu_bo_create_kernel(adev, PSP_1_MEG, PSP_1_MEG,
+				      (amdgpu_sriov_vf(adev) || adev->debug_use_vram_fw_buf) ?
+				      AMDGPU_GEM_DOMAIN_VRAM : AMDGPU_GEM_DOMAIN_GTT,
+				      &psp->fw_pri_bo,
+				      &psp->fw_pri_mc_addr,
+				      &psp->fw_pri_buf);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_bo_create_kernel(adev, PSP_FENCE_BUFFER_SIZE, PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_VRAM |
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &psp->fence_buf_bo,
+				      &psp->fence_buf_mc_addr,
+				      &psp->fence_buf);
+	if (ret)
+		goto failed1;
+
+	ret = amdgpu_bo_create_kernel(adev, PSP_CMD_BUFFER_SIZE, PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_VRAM |
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &psp->cmd_buf_bo, &psp->cmd_buf_mc_addr,
+				      (void **)&psp->cmd_buf_mem);
+	if (ret)
+		goto failed2;
+
 	return 0;
+
+failed2:
+	amdgpu_bo_free_kernel(&psp->fence_buf_bo,
+			      &psp->fence_buf_mc_addr, &psp->fence_buf);
+failed1:
+	amdgpu_bo_free_kernel(&psp->fw_pri_bo,
+			      &psp->fw_pri_mc_addr, &psp->fw_pri_buf);
+	return ret;
 }
 
-int psp_wait_for(struct psp_context *psp, uint32_t reg_index,
-		 uint32_t reg_val, uint32_t mask, bool check_changed)
+static int psp_sw_fini(struct amdgpu_ip_block *ip_block)
 {
+	struct amdgpu_device *adev = ip_block->adev;
+	struct psp_context *psp = &adev->psp;
+
+	psp_memory_training_fini(psp);
+
+	amdgpu_ucode_release(&psp->sos_fw);
+	amdgpu_ucode_release(&psp->asd_fw);
+	amdgpu_ucode_release(&psp->ta_fw);
+	amdgpu_ucode_release(&psp->cap_fw);
+	amdgpu_ucode_release(&psp->toc_fw);
+
+	kfree(psp->cmd);
+	psp->cmd = NULL;
+
+	psp_free_shared_bufs(psp);
+
+	if (psp->km_ring.ring_mem)
+		amdgpu_bo_free_kernel(&adev->firmware.rbuf,
+				      &psp->km_ring.ring_mem_mc_addr,
+				      (void **)&psp->km_ring.ring_mem);
+
+	amdgpu_bo_free_kernel(&psp->fw_pri_bo,
+			      &psp->fw_pri_mc_addr, &psp->fw_pri_buf);
+	amdgpu_bo_free_kernel(&psp->fence_buf_bo,
+			      &psp->fence_buf_mc_addr, &psp->fence_buf);
+	amdgpu_bo_free_kernel(&psp->cmd_buf_bo, &psp->cmd_buf_mc_addr,
+			      (void **)&psp->cmd_buf_mem);
+
+	return 0;
+}
+
+int psp_wait_for(struct psp_context *psp, uint32_t reg_index, uint32_t reg_val,
+		 uint32_t mask, uint32_t flags)
+{
+	bool check_changed = flags & PSP_WAITREG_CHANGED;
+	bool verbose = !(flags & PSP_WAITREG_NOVERBOSE);
 	uint32_t val;
 	int i;
 	struct amdgpu_device *adev = psp->adev;
 
+	if (psp->adev->no_hw_access)
+		return 0;
+
 	for (i = 0; i < adev->usec_timeout; i++) {
 		val = RREG32(reg_index);
 		if (check_changed) {
@@ -112,207 +599,2496 @@ int psp_wait_for(struct psp_context *psp, uint32_t reg_index,
 		udelay(1);
 	}
 
+	if (verbose)
+		dev_err(adev->dev,
+			"psp reg (0x%x) wait timed out, mask: %x, read: %x exp: %x",
+			reg_index, mask, val, reg_val);
+
+	return -ETIME;
+}
+
+int psp_wait_for_spirom_update(struct psp_context *psp, uint32_t reg_index,
+			       uint32_t reg_val, uint32_t mask, uint32_t msec_timeout)
+{
+	uint32_t val;
+	int i;
+	struct amdgpu_device *adev = psp->adev;
+
+	if (psp->adev->no_hw_access)
+		return 0;
+
+	for (i = 0; i < msec_timeout; i++) {
+		val = RREG32(reg_index);
+		if ((val & mask) == reg_val)
+			return 0;
+		msleep(1);
+	}
+
 	return -ETIME;
 }
 
+static const char *psp_gfx_cmd_name(enum psp_gfx_cmd_id cmd_id)
+{
+	switch (cmd_id) {
+	case GFX_CMD_ID_LOAD_TA:
+		return "LOAD_TA";
+	case GFX_CMD_ID_UNLOAD_TA:
+		return "UNLOAD_TA";
+	case GFX_CMD_ID_INVOKE_CMD:
+		return "INVOKE_CMD";
+	case GFX_CMD_ID_LOAD_ASD:
+		return "LOAD_ASD";
+	case GFX_CMD_ID_SETUP_TMR:
+		return "SETUP_TMR";
+	case GFX_CMD_ID_LOAD_IP_FW:
+		return "LOAD_IP_FW";
+	case GFX_CMD_ID_DESTROY_TMR:
+		return "DESTROY_TMR";
+	case GFX_CMD_ID_SAVE_RESTORE:
+		return "SAVE_RESTORE_IP_FW";
+	case GFX_CMD_ID_SETUP_VMR:
+		return "SETUP_VMR";
+	case GFX_CMD_ID_DESTROY_VMR:
+		return "DESTROY_VMR";
+	case GFX_CMD_ID_PROG_REG:
+		return "PROG_REG";
+	case GFX_CMD_ID_GET_FW_ATTESTATION:
+		return "GET_FW_ATTESTATION";
+	case GFX_CMD_ID_LOAD_TOC:
+		return "ID_LOAD_TOC";
+	case GFX_CMD_ID_AUTOLOAD_RLC:
+		return "AUTOLOAD_RLC";
+	case GFX_CMD_ID_BOOT_CFG:
+		return "BOOT_CFG";
+	case GFX_CMD_ID_CONFIG_SQ_PERFMON:
+		return "CONFIG_SQ_PERFMON";
+	case GFX_CMD_ID_FB_FW_RESERV_ADDR:
+		return "FB_FW_RESERV_ADDR";
+	case GFX_CMD_ID_FB_FW_RESERV_EXT_ADDR:
+		return "FB_FW_RESERV_EXT_ADDR";
+	case GFX_CMD_ID_SRIOV_SPATIAL_PART:
+		return "SPATIAL_PARTITION";
+	case GFX_CMD_ID_FB_NPS_MODE:
+		return "NPS_MODE_CHANGE";
+	default:
+		return "UNKNOWN CMD";
+	}
+}
+
+static bool psp_err_warn(struct psp_context *psp)
+{
+	struct psp_gfx_cmd_resp *cmd = psp->cmd_buf_mem;
+
+	/* This response indicates reg list is already loaded */
+	if (amdgpu_ip_version(psp->adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 2) &&
+	    cmd->cmd_id == GFX_CMD_ID_LOAD_IP_FW &&
+	    cmd->cmd.cmd_load_ip_fw.fw_type == GFX_FW_TYPE_REG_LIST &&
+	    cmd->resp.status == TEE_ERROR_CANCEL)
+		return false;
+
+	return true;
+}
+
 static int
 psp_cmd_submit_buf(struct psp_context *psp,
 		   struct amdgpu_firmware_info *ucode,
-		   struct psp_gfx_cmd_resp *cmd, uint64_t fence_mc_addr,
-		   int index)
+		   struct psp_gfx_cmd_resp *cmd, uint64_t fence_mc_addr)
 {
 	int ret;
+	int index;
+	int timeout = psp->adev->psp_timeout;
+	bool ras_intr = false;
+	bool skip_unsupport = false;
+
+	if (psp->adev->no_hw_access)
+		return 0;
 
 	memset(psp->cmd_buf_mem, 0, PSP_CMD_BUFFER_SIZE);
 
 	memcpy(psp->cmd_buf_mem, cmd, sizeof(struct psp_gfx_cmd_resp));
 
-	ret = psp_cmd_submit(psp, ucode, psp->cmd_buf_mc_addr,
-			     fence_mc_addr, index);
+	index = atomic_inc_return(&psp->fence_value);
+	ret = psp_ring_cmd_submit(psp, psp->cmd_buf_mc_addr, fence_mc_addr, index);
+	if (ret) {
+		atomic_dec(&psp->fence_value);
+		goto exit;
+	}
 
+	amdgpu_device_invalidate_hdp(psp->adev, NULL);
 	while (*((unsigned int *)psp->fence_buf) != index) {
-		msleep(1);
+		if (--timeout == 0)
+			break;
+		/*
+		 * Shouldn't wait for timeout when err_event_athub occurs,
+		 * because gpu reset thread triggered and lock resource should
+		 * be released for psp resume sequence.
+		 */
+		ras_intr = amdgpu_ras_intr_triggered();
+		if (ras_intr)
+			break;
+		usleep_range(10, 100);
+		amdgpu_device_invalidate_hdp(psp->adev, NULL);
+	}
+
+	/* We allow TEE_ERROR_NOT_SUPPORTED for VMR command and PSP_ERR_UNKNOWN_COMMAND in SRIOV */
+	skip_unsupport = (psp->cmd_buf_mem->resp.status == TEE_ERROR_NOT_SUPPORTED ||
+		psp->cmd_buf_mem->resp.status == PSP_ERR_UNKNOWN_COMMAND) && amdgpu_sriov_vf(psp->adev);
+
+	memcpy(&cmd->resp, &psp->cmd_buf_mem->resp, sizeof(struct psp_gfx_resp));
+
+	/* In some cases, psp response status is not 0 even there is no
+	 * problem while the command is submitted. Some version of PSP FW
+	 * doesn't write 0 to that field.
+	 * So here we would like to only print a warning instead of an error
+	 * during psp initialization to avoid breaking hw_init and it doesn't
+	 * return -EINVAL.
+	 */
+	if (!skip_unsupport && (psp->cmd_buf_mem->resp.status || !timeout) && !ras_intr) {
+		if (ucode)
+			dev_warn(psp->adev->dev,
+				 "failed to load ucode %s(0x%X) ",
+				 amdgpu_ucode_name(ucode->ucode_id), ucode->ucode_id);
+		if (psp_err_warn(psp))
+			dev_warn(
+				psp->adev->dev,
+				"psp gfx command %s(0x%X) failed and response status is (0x%X)\n",
+				psp_gfx_cmd_name(psp->cmd_buf_mem->cmd_id),
+				psp->cmd_buf_mem->cmd_id,
+				psp->cmd_buf_mem->resp.status);
+		/* If any firmware (including CAP) load fails under SRIOV, it should
+		 * return failure to stop the VF from initializing.
+		 * Also return failure in case of timeout
+		 */
+		if ((ucode && amdgpu_sriov_vf(psp->adev)) || !timeout) {
+			ret = -EINVAL;
+			goto exit;
+		}
+	}
+
+	if (ucode) {
+		ucode->tmr_mc_addr_lo = psp->cmd_buf_mem->resp.fw_addr_lo;
+		ucode->tmr_mc_addr_hi = psp->cmd_buf_mem->resp.fw_addr_hi;
 	}
 
+exit:
 	return ret;
 }
 
-static void psp_prep_tmr_cmd_buf(struct psp_gfx_cmd_resp *cmd,
-				 uint64_t tmr_mc, uint32_t size)
+static struct psp_gfx_cmd_resp *acquire_psp_cmd_buf(struct psp_context *psp)
+{
+	struct psp_gfx_cmd_resp *cmd = psp->cmd;
+
+	mutex_lock(&psp->mutex);
+
+	memset(cmd, 0, sizeof(struct psp_gfx_cmd_resp));
+
+	return cmd;
+}
+
+static void release_psp_cmd_buf(struct psp_context *psp)
 {
-	cmd->cmd_id = GFX_CMD_ID_SETUP_TMR;
+	mutex_unlock(&psp->mutex);
+}
+
+static void psp_prep_tmr_cmd_buf(struct psp_context *psp,
+				 struct psp_gfx_cmd_resp *cmd,
+				 uint64_t tmr_mc, struct amdgpu_bo *tmr_bo)
+{
+	struct amdgpu_device *adev = psp->adev;
+	uint32_t size = 0;
+	uint64_t tmr_pa = 0;
+
+	if (tmr_bo) {
+		size = amdgpu_bo_size(tmr_bo);
+		tmr_pa = amdgpu_gmc_vram_pa(adev, tmr_bo);
+	}
+
+	if (amdgpu_sriov_vf(psp->adev))
+		cmd->cmd_id = GFX_CMD_ID_SETUP_VMR;
+	else
+		cmd->cmd_id = GFX_CMD_ID_SETUP_TMR;
 	cmd->cmd.cmd_setup_tmr.buf_phy_addr_lo = lower_32_bits(tmr_mc);
 	cmd->cmd.cmd_setup_tmr.buf_phy_addr_hi = upper_32_bits(tmr_mc);
 	cmd->cmd.cmd_setup_tmr.buf_size = size;
+	cmd->cmd.cmd_setup_tmr.bitfield.virt_phy_addr = 1;
+	cmd->cmd.cmd_setup_tmr.system_phy_addr_lo = lower_32_bits(tmr_pa);
+	cmd->cmd.cmd_setup_tmr.system_phy_addr_hi = upper_32_bits(tmr_pa);
+}
+
+static void psp_prep_load_toc_cmd_buf(struct psp_gfx_cmd_resp *cmd,
+				      uint64_t pri_buf_mc, uint32_t size)
+{
+	cmd->cmd_id = GFX_CMD_ID_LOAD_TOC;
+	cmd->cmd.cmd_load_toc.toc_phy_addr_lo = lower_32_bits(pri_buf_mc);
+	cmd->cmd.cmd_load_toc.toc_phy_addr_hi = upper_32_bits(pri_buf_mc);
+	cmd->cmd.cmd_load_toc.toc_size = size;
+}
+
+/* Issue LOAD TOC cmd to PSP to part toc and calculate tmr size needed */
+static int psp_load_toc(struct psp_context *psp,
+			uint32_t *tmr_size)
+{
+	int ret;
+	struct psp_gfx_cmd_resp *cmd = acquire_psp_cmd_buf(psp);
+
+	/* Copy toc to psp firmware private buffer */
+	psp_copy_fw(psp, psp->toc.start_addr, psp->toc.size_bytes);
+
+	psp_prep_load_toc_cmd_buf(cmd, psp->fw_pri_mc_addr, psp->toc.size_bytes);
+
+	ret = psp_cmd_submit_buf(psp, NULL, cmd,
+				 psp->fence_buf_mc_addr);
+	if (!ret)
+		*tmr_size = psp->cmd_buf_mem->resp.tmr_size;
+
+	release_psp_cmd_buf(psp);
+
+	return ret;
 }
 
 /* Set up Trusted Memory Region */
 static int psp_tmr_init(struct psp_context *psp)
 {
-	int ret;
+	int ret = 0;
+	int tmr_size;
+	void *tmr_buf;
+	void **pptr;
 
 	/*
-	 * Allocate 3M memory aligned to 1M from Frame Buffer (local
-	 * physical).
+	 * According to HW engineer, they prefer the TMR address be "naturally
+	 * aligned" , e.g. the start address be an integer divide of TMR size.
 	 *
 	 * Note: this memory need be reserved till the driver
 	 * uninitializes.
 	 */
-	ret = amdgpu_bo_create_kernel(psp->adev, 0x300000, 0x100000,
-				      AMDGPU_GEM_DOMAIN_VRAM,
-				      &psp->tmr_bo, &psp->tmr_mc_addr, &psp->tmr_buf);
+	tmr_size = PSP_TMR_SIZE(psp->adev);
+
+	/* For ASICs support RLC autoload, psp will parse the toc
+	 * and calculate the total size of TMR needed
+	 */
+	if (!amdgpu_sriov_vf(psp->adev) &&
+	    psp->toc.start_addr &&
+	    psp->toc.size_bytes &&
+	    psp->fw_pri_buf) {
+		ret = psp_load_toc(psp, &tmr_size);
+		if (ret) {
+			dev_err(psp->adev->dev, "Failed to load toc\n");
+			return ret;
+		}
+	}
+
+	if (!psp->tmr_bo && !psp->boot_time_tmr) {
+		pptr = amdgpu_sriov_vf(psp->adev) ? &tmr_buf : NULL;
+		ret = amdgpu_bo_create_kernel(psp->adev, tmr_size,
+					      PSP_TMR_ALIGNMENT,
+					      AMDGPU_GEM_DOMAIN_GTT | AMDGPU_GEM_DOMAIN_VRAM,
+					      &psp->tmr_bo, &psp->tmr_mc_addr,
+					      pptr);
+	}
+	if (amdgpu_virt_xgmi_migrate_enabled(psp->adev) && psp->tmr_bo)
+		psp->tmr_mc_addr = amdgpu_bo_fb_aper_addr(psp->tmr_bo);
 
 	return ret;
 }
 
+static bool psp_skip_tmr(struct psp_context *psp)
+{
+	switch (amdgpu_ip_version(psp->adev, MP0_HWIP, 0)) {
+	case IP_VERSION(11, 0, 9):
+	case IP_VERSION(11, 0, 7):
+	case IP_VERSION(13, 0, 2):
+	case IP_VERSION(13, 0, 6):
+	case IP_VERSION(13, 0, 10):
+	case IP_VERSION(13, 0, 12):
+	case IP_VERSION(13, 0, 14):
+		return true;
+	default:
+		return false;
+	}
+}
+
 static int psp_tmr_load(struct psp_context *psp)
 {
 	int ret;
 	struct psp_gfx_cmd_resp *cmd;
 
-	cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL);
-	if (!cmd)
-		return -ENOMEM;
+	/* For Navi12 and CHIP_SIENNA_CICHLID SRIOV, do not set up TMR.
+	 * Already set up by host driver.
+	 */
+	if (amdgpu_sriov_vf(psp->adev) && psp_skip_tmr(psp))
+		return 0;
+
+	cmd = acquire_psp_cmd_buf(psp);
+
+	psp_prep_tmr_cmd_buf(psp, cmd, psp->tmr_mc_addr, psp->tmr_bo);
+	if (psp->tmr_bo)
+		dev_info(psp->adev->dev, "reserve 0x%lx from 0x%llx for PSP TMR\n",
+			 amdgpu_bo_size(psp->tmr_bo), psp->tmr_mc_addr);
+
+	ret = psp_cmd_submit_buf(psp, NULL, cmd,
+				 psp->fence_buf_mc_addr);
+
+	release_psp_cmd_buf(psp);
+
+	return ret;
+}
+
+static void psp_prep_tmr_unload_cmd_buf(struct psp_context *psp,
+					struct psp_gfx_cmd_resp *cmd)
+{
+	if (amdgpu_sriov_vf(psp->adev))
+		cmd->cmd_id = GFX_CMD_ID_DESTROY_VMR;
+	else
+		cmd->cmd_id = GFX_CMD_ID_DESTROY_TMR;
+}
+
+static int psp_tmr_unload(struct psp_context *psp)
+{
+	int ret;
+	struct psp_gfx_cmd_resp *cmd;
+
+	/* skip TMR unload for Navi12 and CHIP_SIENNA_CICHLID SRIOV,
+	 * as TMR is not loaded at all
+	 */
+	if (amdgpu_sriov_vf(psp->adev) && psp_skip_tmr(psp))
+		return 0;
+
+	cmd = acquire_psp_cmd_buf(psp);
 
-	psp_prep_tmr_cmd_buf(cmd, psp->tmr_mc_addr, 0x300000);
+	psp_prep_tmr_unload_cmd_buf(psp, cmd);
+	dev_dbg(psp->adev->dev, "free PSP TMR buffer\n");
 
 	ret = psp_cmd_submit_buf(psp, NULL, cmd,
-				 psp->fence_buf_mc_addr, 1);
+				 psp->fence_buf_mc_addr);
+
+	release_psp_cmd_buf(psp);
+
+	return ret;
+}
+
+static int psp_tmr_terminate(struct psp_context *psp)
+{
+	return psp_tmr_unload(psp);
+}
+
+int psp_get_fw_attestation_records_addr(struct psp_context *psp,
+					uint64_t *output_ptr)
+{
+	int ret;
+	struct psp_gfx_cmd_resp *cmd;
+
+	if (!output_ptr)
+		return -EINVAL;
+
+	if (amdgpu_sriov_vf(psp->adev))
+		return 0;
+
+	cmd = acquire_psp_cmd_buf(psp);
+
+	cmd->cmd_id = GFX_CMD_ID_GET_FW_ATTESTATION;
+
+	ret = psp_cmd_submit_buf(psp, NULL, cmd,
+				 psp->fence_buf_mc_addr);
+
+	if (!ret) {
+		*output_ptr = ((uint64_t)cmd->resp.uresp.fwar_db_info.fwar_db_addr_lo) +
+			      ((uint64_t)cmd->resp.uresp.fwar_db_info.fwar_db_addr_hi << 32);
+	}
+
+	release_psp_cmd_buf(psp);
+
+	return ret;
+}
+
+static int psp_get_fw_reservation_info(struct psp_context *psp,
+						   uint32_t cmd_id,
+						   uint64_t *addr,
+						   uint32_t *size)
+{
+	int ret;
+	uint32_t status;
+	struct psp_gfx_cmd_resp *cmd;
+
+	cmd = acquire_psp_cmd_buf(psp);
+
+	cmd->cmd_id = cmd_id;
+
+	ret = psp_cmd_submit_buf(psp, NULL, cmd,
+				 psp->fence_buf_mc_addr);
+	if (ret) {
+		release_psp_cmd_buf(psp);
+		return ret;
+	}
+
+	status = cmd->resp.status;
+	if (status == PSP_ERR_UNKNOWN_COMMAND) {
+		release_psp_cmd_buf(psp);
+		*addr = 0;
+		*size = 0;
+		return 0;
+	}
+
+	*addr = (uint64_t)cmd->resp.uresp.fw_reserve_info.reserve_base_address_hi << 32 |
+		cmd->resp.uresp.fw_reserve_info.reserve_base_address_lo;
+	*size = cmd->resp.uresp.fw_reserve_info.reserve_size;
+
+	release_psp_cmd_buf(psp);
+
+	return 0;
+}
+
+int psp_update_fw_reservation(struct psp_context *psp)
+{
+	int ret;
+	uint64_t reserv_addr, reserv_addr_ext;
+	uint32_t reserv_size, reserv_size_ext, mp0_ip_ver;
+	struct amdgpu_device *adev = psp->adev;
+
+	mp0_ip_ver = amdgpu_ip_version(adev, MP0_HWIP, 0);
+
+	if (amdgpu_sriov_vf(psp->adev))
+		return 0;
+
+	switch (mp0_ip_ver) {
+	case IP_VERSION(14, 0, 2):
+		if (adev->psp.sos.fw_version < 0x3b0e0d)
+			return 0;
+		break;
+
+	case IP_VERSION(14, 0, 3):
+		if (adev->psp.sos.fw_version < 0x3a0e14)
+			return 0;
+		break;
+
+	default:
+		return 0;
+	}
+
+	ret = psp_get_fw_reservation_info(psp, GFX_CMD_ID_FB_FW_RESERV_ADDR, &reserv_addr, &reserv_size);
 	if (ret)
-		goto failed;
+		return ret;
+	ret = psp_get_fw_reservation_info(psp, GFX_CMD_ID_FB_FW_RESERV_EXT_ADDR, &reserv_addr_ext, &reserv_size_ext);
+	if (ret)
+		return ret;
+
+	if (reserv_addr != adev->gmc.real_vram_size - reserv_size) {
+		dev_warn(adev->dev, "reserve fw region is not valid!\n");
+		return 0;
+	}
 
-	kfree(cmd);
+	amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory, NULL, NULL);
+
+	reserv_size = roundup(reserv_size, SZ_1M);
+
+	ret = amdgpu_bo_create_kernel_at(adev, reserv_addr, reserv_size, &adev->mman.fw_reserved_memory, NULL);
+	if (ret) {
+		dev_err(adev->dev, "reserve fw region failed(%d)!\n", ret);
+		amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory, NULL, NULL);
+		return ret;
+	}
+
+	reserv_size_ext = roundup(reserv_size_ext, SZ_1M);
+
+	ret = amdgpu_bo_create_kernel_at(adev, reserv_addr_ext, reserv_size_ext,
+					 &adev->mman.fw_reserved_memory_extend, NULL);
+	if (ret) {
+		dev_err(adev->dev, "reserve extend fw region failed(%d)!\n", ret);
+		amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory_extend, NULL, NULL);
+		return ret;
+	}
 
 	return 0;
+}
+
+static int psp_boot_config_get(struct amdgpu_device *adev, uint32_t *boot_cfg)
+{
+	struct psp_context *psp = &adev->psp;
+	struct psp_gfx_cmd_resp *cmd;
+	int ret;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	cmd = acquire_psp_cmd_buf(psp);
+
+	cmd->cmd_id = GFX_CMD_ID_BOOT_CFG;
+	cmd->cmd.boot_cfg.sub_cmd = BOOTCFG_CMD_GET;
+
+	ret = psp_cmd_submit_buf(psp, NULL, cmd, psp->fence_buf_mc_addr);
+	if (!ret) {
+		*boot_cfg =
+			(cmd->resp.uresp.boot_cfg.boot_cfg & BOOT_CONFIG_GECC) ? 1 : 0;
+	}
+
+	release_psp_cmd_buf(psp);
 
-failed:
-	kfree(cmd);
 	return ret;
 }
 
-static void psp_prep_asd_cmd_buf(struct psp_gfx_cmd_resp *cmd,
-				 uint64_t asd_mc, uint64_t asd_mc_shared,
-				 uint32_t size, uint32_t shared_size)
+static int psp_boot_config_set(struct amdgpu_device *adev, uint32_t boot_cfg)
 {
-	cmd->cmd_id = GFX_CMD_ID_LOAD_ASD;
-	cmd->cmd.cmd_load_ta.app_phy_addr_lo = lower_32_bits(asd_mc);
-	cmd->cmd.cmd_load_ta.app_phy_addr_hi = upper_32_bits(asd_mc);
-	cmd->cmd.cmd_load_ta.app_len = size;
+	int ret;
+	struct psp_context *psp = &adev->psp;
+	struct psp_gfx_cmd_resp *cmd;
 
-	cmd->cmd.cmd_load_ta.cmd_buf_phy_addr_lo = lower_32_bits(asd_mc_shared);
-	cmd->cmd.cmd_load_ta.cmd_buf_phy_addr_hi = upper_32_bits(asd_mc_shared);
-	cmd->cmd.cmd_load_ta.cmd_buf_len = shared_size;
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	cmd = acquire_psp_cmd_buf(psp);
+
+	cmd->cmd_id = GFX_CMD_ID_BOOT_CFG;
+	cmd->cmd.boot_cfg.sub_cmd = BOOTCFG_CMD_SET;
+	cmd->cmd.boot_cfg.boot_config = boot_cfg;
+	cmd->cmd.boot_cfg.boot_config_valid = boot_cfg;
+
+	ret = psp_cmd_submit_buf(psp, NULL, cmd, psp->fence_buf_mc_addr);
+
+	release_psp_cmd_buf(psp);
+
+	return ret;
 }
 
-static int psp_asd_init(struct psp_context *psp)
+static int psp_rl_load(struct amdgpu_device *adev)
 {
 	int ret;
+	struct psp_context *psp = &adev->psp;
+	struct psp_gfx_cmd_resp *cmd;
 
-	/*
-	 * Allocate 16k memory aligned to 4k from Frame Buffer (local
-	 * physical) for shared ASD <-> Driver
-	 */
-	ret = amdgpu_bo_create_kernel(psp->adev, PSP_ASD_SHARED_MEM_SIZE,
-				      PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
-				      &psp->asd_shared_bo,
-				      &psp->asd_shared_mc_addr,
-				      &psp->asd_shared_buf);
+	if (!is_psp_fw_valid(psp->rl))
+		return 0;
+
+	cmd = acquire_psp_cmd_buf(psp);
+
+	memset(psp->fw_pri_buf, 0, PSP_1_MEG);
+	memcpy(psp->fw_pri_buf, psp->rl.start_addr, psp->rl.size_bytes);
+
+	cmd->cmd_id = GFX_CMD_ID_LOAD_IP_FW;
+	cmd->cmd.cmd_load_ip_fw.fw_phy_addr_lo = lower_32_bits(psp->fw_pri_mc_addr);
+	cmd->cmd.cmd_load_ip_fw.fw_phy_addr_hi = upper_32_bits(psp->fw_pri_mc_addr);
+	cmd->cmd.cmd_load_ip_fw.fw_size = psp->rl.size_bytes;
+	cmd->cmd.cmd_load_ip_fw.fw_type = GFX_FW_TYPE_REG_LIST;
+
+	ret = psp_cmd_submit_buf(psp, NULL, cmd, psp->fence_buf_mc_addr);
+
+	release_psp_cmd_buf(psp);
 
 	return ret;
 }
 
-static int psp_asd_load(struct psp_context *psp)
+int psp_memory_partition(struct psp_context *psp, int mode)
 {
+	struct psp_gfx_cmd_resp *cmd;
 	int ret;
+
+	if (amdgpu_sriov_vf(psp->adev))
+		return 0;
+
+	cmd = acquire_psp_cmd_buf(psp);
+
+	cmd->cmd_id = GFX_CMD_ID_FB_NPS_MODE;
+	cmd->cmd.cmd_memory_part.mode = mode;
+
+	dev_info(psp->adev->dev,
+		 "Requesting %d memory partition change through PSP", mode);
+	ret = psp_cmd_submit_buf(psp, NULL, cmd, psp->fence_buf_mc_addr);
+	if (ret)
+		dev_err(psp->adev->dev,
+			"PSP request failed to change to NPS%d mode\n", mode);
+
+	release_psp_cmd_buf(psp);
+
+	return ret;
+}
+
+int psp_spatial_partition(struct psp_context *psp, int mode)
+{
 	struct psp_gfx_cmd_resp *cmd;
+	int ret;
+
+	if (amdgpu_sriov_vf(psp->adev))
+		return 0;
+
+	cmd = acquire_psp_cmd_buf(psp);
+
+	cmd->cmd_id = GFX_CMD_ID_SRIOV_SPATIAL_PART;
+	cmd->cmd.cmd_spatial_part.mode = mode;
+
+	dev_info(psp->adev->dev, "Requesting %d partitions through PSP", mode);
+	ret = psp_cmd_submit_buf(psp, NULL, cmd, psp->fence_buf_mc_addr);
+
+	release_psp_cmd_buf(psp);
+
+	return ret;
+}
+
+static int psp_asd_initialize(struct psp_context *psp)
+{
+	int ret;
 
 	/* If PSP version doesn't match ASD version, asd loading will be failed.
 	 * add workaround to bypass it for sriov now.
 	 * TODO: add version check to make it common
 	 */
+	if (amdgpu_sriov_vf(psp->adev) || !psp->asd_context.bin_desc.size_bytes)
+		return 0;
+
+	/* bypass asd if display hardware is not available */
+	if (!amdgpu_device_has_display_hardware(psp->adev) &&
+	    amdgpu_ip_version(psp->adev, MP0_HWIP, 0) >= IP_VERSION(13, 0, 10))
+		return 0;
+
+	psp->asd_context.mem_context.shared_mc_addr  = 0;
+	psp->asd_context.mem_context.shared_mem_size = PSP_ASD_SHARED_MEM_SIZE;
+	psp->asd_context.ta_load_type                = GFX_CMD_ID_LOAD_ASD;
+
+	ret = psp_ta_load(psp, &psp->asd_context);
+	if (!ret)
+		psp->asd_context.initialized = true;
+
+	return ret;
+}
+
+static void psp_prep_ta_unload_cmd_buf(struct psp_gfx_cmd_resp *cmd,
+				       uint32_t session_id)
+{
+	cmd->cmd_id = GFX_CMD_ID_UNLOAD_TA;
+	cmd->cmd.cmd_unload_ta.session_id = session_id;
+}
+
+int psp_ta_unload(struct psp_context *psp, struct ta_context *context)
+{
+	int ret;
+	struct psp_gfx_cmd_resp *cmd = acquire_psp_cmd_buf(psp);
+
+	psp_prep_ta_unload_cmd_buf(cmd, context->session_id);
+
+	ret = psp_cmd_submit_buf(psp, NULL, cmd, psp->fence_buf_mc_addr);
+
+	context->resp_status = cmd->resp.status;
+
+	release_psp_cmd_buf(psp);
+
+	return ret;
+}
+
+static int psp_asd_terminate(struct psp_context *psp)
+{
+	int ret;
+
 	if (amdgpu_sriov_vf(psp->adev))
 		return 0;
 
-	cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL);
-	if (!cmd)
-		return -ENOMEM;
+	if (!psp->asd_context.initialized)
+		return 0;
 
-	memset(psp->fw_pri_buf, 0, PSP_1_MEG);
-	memcpy(psp->fw_pri_buf, psp->asd_start_addr, psp->asd_ucode_size);
+	ret = psp_ta_unload(psp, &psp->asd_context);
+	if (!ret)
+		psp->asd_context.initialized = false;
+
+	return ret;
+}
+
+static void psp_prep_reg_prog_cmd_buf(struct psp_gfx_cmd_resp *cmd,
+		uint32_t id, uint32_t value)
+{
+	cmd->cmd_id = GFX_CMD_ID_PROG_REG;
+	cmd->cmd.cmd_setup_reg_prog.reg_value = value;
+	cmd->cmd.cmd_setup_reg_prog.reg_id = id;
+}
+
+int psp_reg_program(struct psp_context *psp, enum psp_reg_prog_id reg,
+		uint32_t value)
+{
+	struct psp_gfx_cmd_resp *cmd;
+	int ret = 0;
+
+	if (reg >= PSP_REG_LAST)
+		return -EINVAL;
+
+	cmd = acquire_psp_cmd_buf(psp);
+
+	psp_prep_reg_prog_cmd_buf(cmd, reg, value);
+	ret = psp_cmd_submit_buf(psp, NULL, cmd, psp->fence_buf_mc_addr);
+	if (ret)
+		dev_err(psp->adev->dev, "PSP failed to program reg id %d\n", reg);
+
+	release_psp_cmd_buf(psp);
+
+	return ret;
+}
+
+static void psp_prep_ta_load_cmd_buf(struct psp_gfx_cmd_resp *cmd,
+				     uint64_t ta_bin_mc,
+				     struct ta_context *context)
+{
+	cmd->cmd_id				= context->ta_load_type;
+	cmd->cmd.cmd_load_ta.app_phy_addr_lo	= lower_32_bits(ta_bin_mc);
+	cmd->cmd.cmd_load_ta.app_phy_addr_hi	= upper_32_bits(ta_bin_mc);
+	cmd->cmd.cmd_load_ta.app_len		= context->bin_desc.size_bytes;
+
+	cmd->cmd.cmd_load_ta.cmd_buf_phy_addr_lo =
+		lower_32_bits(context->mem_context.shared_mc_addr);
+	cmd->cmd.cmd_load_ta.cmd_buf_phy_addr_hi =
+		upper_32_bits(context->mem_context.shared_mc_addr);
+	cmd->cmd.cmd_load_ta.cmd_buf_len = context->mem_context.shared_mem_size;
+}
+
+int psp_ta_init_shared_buf(struct psp_context *psp,
+				  struct ta_mem_context *mem_ctx)
+{
+	/*
+	 * Allocate 16k memory aligned to 4k from Frame Buffer (local
+	 * physical) for ta to host memory
+	 */
+	return amdgpu_bo_create_kernel(psp->adev, mem_ctx->shared_mem_size,
+				      PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM |
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &mem_ctx->shared_bo,
+				      &mem_ctx->shared_mc_addr,
+				      &mem_ctx->shared_buf);
+}
+
+static void psp_prep_ta_invoke_cmd_buf(struct psp_gfx_cmd_resp *cmd,
+				       uint32_t ta_cmd_id,
+				       uint32_t session_id)
+{
+	cmd->cmd_id				= GFX_CMD_ID_INVOKE_CMD;
+	cmd->cmd.cmd_invoke_cmd.session_id	= session_id;
+	cmd->cmd.cmd_invoke_cmd.ta_cmd_id	= ta_cmd_id;
+}
+
+int psp_ta_invoke(struct psp_context *psp,
+		  uint32_t ta_cmd_id,
+		  struct ta_context *context)
+{
+	int ret;
+	struct psp_gfx_cmd_resp *cmd = acquire_psp_cmd_buf(psp);
 
-	psp_prep_asd_cmd_buf(cmd, psp->fw_pri_mc_addr, psp->asd_shared_mc_addr,
-			     psp->asd_ucode_size, PSP_ASD_SHARED_MEM_SIZE);
+	psp_prep_ta_invoke_cmd_buf(cmd, ta_cmd_id, context->session_id);
 
 	ret = psp_cmd_submit_buf(psp, NULL, cmd,
-				 psp->fence_buf_mc_addr, 2);
+				 psp->fence_buf_mc_addr);
+
+	context->resp_status = cmd->resp.status;
 
-	kfree(cmd);
+	release_psp_cmd_buf(psp);
 
 	return ret;
 }
 
-static int psp_hw_start(struct psp_context *psp)
+int psp_ta_load(struct psp_context *psp, struct ta_context *context)
+{
+	int ret;
+	struct psp_gfx_cmd_resp *cmd;
+
+	cmd = acquire_psp_cmd_buf(psp);
+
+	psp_copy_fw(psp, context->bin_desc.start_addr,
+		    context->bin_desc.size_bytes);
+
+	if (amdgpu_virt_xgmi_migrate_enabled(psp->adev) &&
+		context->mem_context.shared_bo)
+		context->mem_context.shared_mc_addr =
+			amdgpu_bo_fb_aper_addr(context->mem_context.shared_bo);
+
+	psp_prep_ta_load_cmd_buf(cmd, psp->fw_pri_mc_addr, context);
+
+	ret = psp_cmd_submit_buf(psp, NULL, cmd,
+				 psp->fence_buf_mc_addr);
+
+	context->resp_status = cmd->resp.status;
+
+	if (!ret)
+		context->session_id = cmd->resp.session_id;
+
+	release_psp_cmd_buf(psp);
+
+	return ret;
+}
+
+int psp_xgmi_invoke(struct psp_context *psp, uint32_t ta_cmd_id)
+{
+	return psp_ta_invoke(psp, ta_cmd_id, &psp->xgmi_context.context);
+}
+
+int psp_xgmi_terminate(struct psp_context *psp)
 {
 	int ret;
+	struct amdgpu_device *adev = psp->adev;
+
+	/* XGMI TA unload currently is not supported on Arcturus/Aldebaran A+A */
+	if (amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(11, 0, 4) ||
+	    (amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 2) &&
+	     adev->gmc.xgmi.connected_to_cpu))
+		return 0;
+
+	if (!psp->xgmi_context.context.initialized)
+		return 0;
+
+	ret = psp_ta_unload(psp, &psp->xgmi_context.context);
 
-	ret = psp_bootloader_load_sysdrv(psp);
+	psp->xgmi_context.context.initialized = false;
+
+	return ret;
+}
+
+int psp_xgmi_initialize(struct psp_context *psp, bool set_extended_data, bool load_ta)
+{
+	struct ta_xgmi_shared_memory *xgmi_cmd;
+	int ret;
+
+	if (!psp->ta_fw ||
+	    !psp->xgmi_context.context.bin_desc.size_bytes ||
+	    !psp->xgmi_context.context.bin_desc.start_addr)
+		return -ENOENT;
+
+	if (!load_ta)
+		goto invoke;
+
+	psp->xgmi_context.context.mem_context.shared_mem_size = PSP_XGMI_SHARED_MEM_SIZE;
+	psp->xgmi_context.context.ta_load_type = GFX_CMD_ID_LOAD_TA;
+
+	if (!psp->xgmi_context.context.mem_context.shared_buf) {
+		ret = psp_ta_init_shared_buf(psp, &psp->xgmi_context.context.mem_context);
+		if (ret)
+			return ret;
+	}
+
+	/* Load XGMI TA */
+	ret = psp_ta_load(psp, &psp->xgmi_context.context);
+	if (!ret)
+		psp->xgmi_context.context.initialized = true;
+	else
+		return ret;
+
+invoke:
+	/* Initialize XGMI session */
+	xgmi_cmd = (struct ta_xgmi_shared_memory *)(psp->xgmi_context.context.mem_context.shared_buf);
+	memset(xgmi_cmd, 0, sizeof(struct ta_xgmi_shared_memory));
+	xgmi_cmd->flag_extend_link_record = set_extended_data;
+	xgmi_cmd->cmd_id = TA_COMMAND_XGMI__INITIALIZE;
+
+	ret = psp_xgmi_invoke(psp, xgmi_cmd->cmd_id);
+	/* note down the capbility flag for XGMI TA */
+	psp->xgmi_context.xgmi_ta_caps = xgmi_cmd->caps_flag;
+
+	return ret;
+}
+
+int psp_xgmi_get_hive_id(struct psp_context *psp, uint64_t *hive_id)
+{
+	struct ta_xgmi_shared_memory *xgmi_cmd;
+	int ret;
+
+	xgmi_cmd = (struct ta_xgmi_shared_memory *)psp->xgmi_context.context.mem_context.shared_buf;
+	memset(xgmi_cmd, 0, sizeof(struct ta_xgmi_shared_memory));
+
+	xgmi_cmd->cmd_id = TA_COMMAND_XGMI__GET_HIVE_ID;
+
+	/* Invoke xgmi ta to get hive id */
+	ret = psp_xgmi_invoke(psp, xgmi_cmd->cmd_id);
 	if (ret)
 		return ret;
 
-	ret = psp_bootloader_load_sos(psp);
+	*hive_id = xgmi_cmd->xgmi_out_message.get_hive_id.hive_id;
+
+	return 0;
+}
+
+int psp_xgmi_get_node_id(struct psp_context *psp, uint64_t *node_id)
+{
+	struct ta_xgmi_shared_memory *xgmi_cmd;
+	int ret;
+
+	xgmi_cmd = (struct ta_xgmi_shared_memory *)psp->xgmi_context.context.mem_context.shared_buf;
+	memset(xgmi_cmd, 0, sizeof(struct ta_xgmi_shared_memory));
+
+	xgmi_cmd->cmd_id = TA_COMMAND_XGMI__GET_NODE_ID;
+
+	/* Invoke xgmi ta to get the node id */
+	ret = psp_xgmi_invoke(psp, xgmi_cmd->cmd_id);
 	if (ret)
 		return ret;
 
-	ret = psp_ring_create(psp, PSP_RING_TYPE__KM);
+	*node_id = xgmi_cmd->xgmi_out_message.get_node_id.node_id;
+
+	return 0;
+}
+
+static bool psp_xgmi_peer_link_info_supported(struct psp_context *psp)
+{
+	return (amdgpu_ip_version(psp->adev, MP0_HWIP, 0) ==
+			IP_VERSION(13, 0, 2) &&
+		psp->xgmi_context.context.bin_desc.fw_version >= 0x2000000b) ||
+	       amdgpu_ip_version(psp->adev, MP0_HWIP, 0) >=
+		       IP_VERSION(13, 0, 6);
+}
+
+/*
+ * Chips that support extended topology information require the driver to
+ * reflect topology information in the opposite direction.  This is
+ * because the TA has already exceeded its link record limit and if the
+ * TA holds bi-directional information, the driver would have to do
+ * multiple fetches instead of just two.
+ */
+static void psp_xgmi_reflect_topology_info(struct psp_context *psp,
+					struct psp_xgmi_node_info node_info)
+{
+	struct amdgpu_device *mirror_adev;
+	struct amdgpu_hive_info *hive;
+	uint64_t src_node_id = psp->adev->gmc.xgmi.node_id;
+	uint64_t dst_node_id = node_info.node_id;
+	uint8_t dst_num_hops = node_info.num_hops;
+	uint8_t dst_num_links = node_info.num_links;
+
+	hive = amdgpu_get_xgmi_hive(psp->adev);
+	if (WARN_ON(!hive))
+		return;
+
+	list_for_each_entry(mirror_adev, &hive->device_list, gmc.xgmi.head) {
+		struct psp_xgmi_topology_info *mirror_top_info;
+		int j;
+
+		if (mirror_adev->gmc.xgmi.node_id != dst_node_id)
+			continue;
+
+		mirror_top_info = &mirror_adev->psp.xgmi_context.top_info;
+		for (j = 0; j < mirror_top_info->num_nodes; j++) {
+			if (mirror_top_info->nodes[j].node_id != src_node_id)
+				continue;
+
+			mirror_top_info->nodes[j].num_hops = dst_num_hops;
+			/*
+			 * prevent 0 num_links value re-reflection since reflection
+			 * criteria is based on num_hops (direct or indirect).
+			 *
+			 */
+			if (dst_num_links)
+				mirror_top_info->nodes[j].num_links = dst_num_links;
+
+			break;
+		}
+
+		break;
+	}
+
+	amdgpu_put_xgmi_hive(hive);
+}
+
+int psp_xgmi_get_topology_info(struct psp_context *psp,
+			       int number_devices,
+			       struct psp_xgmi_topology_info *topology,
+			       bool get_extended_data)
+{
+	struct ta_xgmi_shared_memory *xgmi_cmd;
+	struct ta_xgmi_cmd_get_topology_info_input *topology_info_input;
+	struct ta_xgmi_cmd_get_topology_info_output *topology_info_output;
+	int i;
+	int ret;
+
+	if (!topology || topology->num_nodes > TA_XGMI__MAX_CONNECTED_NODES)
+		return -EINVAL;
+
+	xgmi_cmd = (struct ta_xgmi_shared_memory *)psp->xgmi_context.context.mem_context.shared_buf;
+	memset(xgmi_cmd, 0, sizeof(struct ta_xgmi_shared_memory));
+	xgmi_cmd->flag_extend_link_record = get_extended_data;
+
+	/* Fill in the shared memory with topology information as input */
+	topology_info_input = &xgmi_cmd->xgmi_in_message.get_topology_info;
+	xgmi_cmd->cmd_id = TA_COMMAND_XGMI__GET_TOPOLOGY_INFO;
+	topology_info_input->num_nodes = number_devices;
+
+	for (i = 0; i < topology_info_input->num_nodes; i++) {
+		topology_info_input->nodes[i].node_id = topology->nodes[i].node_id;
+		topology_info_input->nodes[i].num_hops = topology->nodes[i].num_hops;
+		topology_info_input->nodes[i].is_sharing_enabled = topology->nodes[i].is_sharing_enabled;
+		topology_info_input->nodes[i].sdma_engine = topology->nodes[i].sdma_engine;
+	}
+
+	/* Invoke xgmi ta to get the topology information */
+	ret = psp_xgmi_invoke(psp, TA_COMMAND_XGMI__GET_TOPOLOGY_INFO);
 	if (ret)
 		return ret;
 
-	ret = psp_tmr_load(psp);
+	/* Read the output topology information from the shared memory */
+	topology_info_output = &xgmi_cmd->xgmi_out_message.get_topology_info;
+	topology->num_nodes = xgmi_cmd->xgmi_out_message.get_topology_info.num_nodes;
+	for (i = 0; i < topology->num_nodes; i++) {
+		/* extended data will either be 0 or equal to non-extended data */
+		if (topology_info_output->nodes[i].num_hops)
+			topology->nodes[i].num_hops = topology_info_output->nodes[i].num_hops;
+
+		/* non-extended data gets everything here so no need to update */
+		if (!get_extended_data) {
+			topology->nodes[i].node_id = topology_info_output->nodes[i].node_id;
+			topology->nodes[i].is_sharing_enabled =
+					topology_info_output->nodes[i].is_sharing_enabled;
+			topology->nodes[i].sdma_engine =
+					topology_info_output->nodes[i].sdma_engine;
+		}
+
+	}
+
+	/* Invoke xgmi ta again to get the link information */
+	if (psp_xgmi_peer_link_info_supported(psp)) {
+		struct ta_xgmi_cmd_get_peer_link_info *link_info_output;
+		struct ta_xgmi_cmd_get_extend_peer_link_info *link_extend_info_output;
+		bool requires_reflection =
+			(psp->xgmi_context.supports_extended_data &&
+			 get_extended_data) ||
+			amdgpu_ip_version(psp->adev, MP0_HWIP, 0) ==
+				IP_VERSION(13, 0, 6) ||
+			amdgpu_ip_version(psp->adev, MP0_HWIP, 0) ==
+				IP_VERSION(13, 0, 14);
+		bool ta_port_num_support = amdgpu_sriov_vf(psp->adev) ? 0 :
+				psp->xgmi_context.xgmi_ta_caps & EXTEND_PEER_LINK_INFO_CMD_FLAG;
+
+		/* popluate the shared output buffer rather than the cmd input buffer
+		 * with node_ids as the input for GET_PEER_LINKS command execution.
+		 * This is required for GET_PEER_LINKS per xgmi ta implementation.
+		 * The same requirement for GET_EXTEND_PEER_LINKS command.
+		 */
+		if (ta_port_num_support) {
+			link_extend_info_output = &xgmi_cmd->xgmi_out_message.get_extend_link_info;
+
+			for (i = 0; i < topology->num_nodes; i++)
+				link_extend_info_output->nodes[i].node_id = topology->nodes[i].node_id;
+
+			link_extend_info_output->num_nodes = topology->num_nodes;
+			xgmi_cmd->cmd_id = TA_COMMAND_XGMI__GET_EXTEND_PEER_LINKS;
+		} else {
+			link_info_output = &xgmi_cmd->xgmi_out_message.get_link_info;
+
+			for (i = 0; i < topology->num_nodes; i++)
+				link_info_output->nodes[i].node_id = topology->nodes[i].node_id;
+
+			link_info_output->num_nodes = topology->num_nodes;
+			xgmi_cmd->cmd_id = TA_COMMAND_XGMI__GET_PEER_LINKS;
+		}
+
+		ret = psp_xgmi_invoke(psp, xgmi_cmd->cmd_id);
+		if (ret)
+			return ret;
+
+		for (i = 0; i < topology->num_nodes; i++) {
+			uint8_t node_num_links = ta_port_num_support ?
+				link_extend_info_output->nodes[i].num_links : link_info_output->nodes[i].num_links;
+			/* accumulate num_links on extended data */
+			if (get_extended_data) {
+				topology->nodes[i].num_links = topology->nodes[i].num_links + node_num_links;
+			} else {
+				topology->nodes[i].num_links = (requires_reflection && topology->nodes[i].num_links) ?
+								topology->nodes[i].num_links : node_num_links;
+			}
+			/* popluate the connected port num info if supported and available */
+			if (ta_port_num_support && topology->nodes[i].num_links) {
+				memcpy(topology->nodes[i].port_num, link_extend_info_output->nodes[i].port_num,
+				       sizeof(struct xgmi_connected_port_num) * TA_XGMI__MAX_PORT_NUM);
+			}
+
+			/* reflect the topology information for bi-directionality */
+			if (requires_reflection && topology->nodes[i].num_hops)
+				psp_xgmi_reflect_topology_info(psp, topology->nodes[i]);
+		}
+	}
+
+	return 0;
+}
+
+int psp_xgmi_set_topology_info(struct psp_context *psp,
+			       int number_devices,
+			       struct psp_xgmi_topology_info *topology)
+{
+	struct ta_xgmi_shared_memory *xgmi_cmd;
+	struct ta_xgmi_cmd_get_topology_info_input *topology_info_input;
+	int i;
+
+	if (!topology || topology->num_nodes > TA_XGMI__MAX_CONNECTED_NODES)
+		return -EINVAL;
+
+	xgmi_cmd = (struct ta_xgmi_shared_memory *)psp->xgmi_context.context.mem_context.shared_buf;
+	memset(xgmi_cmd, 0, sizeof(struct ta_xgmi_shared_memory));
+
+	topology_info_input = &xgmi_cmd->xgmi_in_message.get_topology_info;
+	xgmi_cmd->cmd_id = TA_COMMAND_XGMI__SET_TOPOLOGY_INFO;
+	topology_info_input->num_nodes = number_devices;
+
+	for (i = 0; i < topology_info_input->num_nodes; i++) {
+		topology_info_input->nodes[i].node_id = topology->nodes[i].node_id;
+		topology_info_input->nodes[i].num_hops = topology->nodes[i].num_hops;
+		topology_info_input->nodes[i].is_sharing_enabled = 1;
+		topology_info_input->nodes[i].sdma_engine = topology->nodes[i].sdma_engine;
+	}
+
+	/* Invoke xgmi ta to set topology information */
+	return psp_xgmi_invoke(psp, TA_COMMAND_XGMI__SET_TOPOLOGY_INFO);
+}
+
+// ras begin
+static void psp_ras_ta_check_status(struct psp_context *psp)
+{
+	struct ta_ras_shared_memory *ras_cmd =
+		(struct ta_ras_shared_memory *)psp->ras_context.context.mem_context.shared_buf;
+
+	switch (ras_cmd->ras_status) {
+	case TA_RAS_STATUS__ERROR_UNSUPPORTED_IP:
+		dev_warn(psp->adev->dev,
+			 "RAS WARNING: cmd failed due to unsupported ip\n");
+		break;
+	case TA_RAS_STATUS__ERROR_UNSUPPORTED_ERROR_INJ:
+		dev_warn(psp->adev->dev,
+			 "RAS WARNING: cmd failed due to unsupported error injection\n");
+		break;
+	case TA_RAS_STATUS__SUCCESS:
+		break;
+	case TA_RAS_STATUS__TEE_ERROR_ACCESS_DENIED:
+		if (ras_cmd->cmd_id == TA_RAS_COMMAND__TRIGGER_ERROR)
+			dev_warn(psp->adev->dev,
+				 "RAS WARNING: Inject error to critical region is not allowed\n");
+		break;
+	default:
+		dev_warn(psp->adev->dev,
+			 "RAS WARNING: ras status = 0x%X\n", ras_cmd->ras_status);
+		break;
+	}
+}
+
+static int psp_ras_send_cmd(struct psp_context *psp,
+		enum ras_command cmd_id, void *in, void *out)
+{
+	struct ta_ras_shared_memory *ras_cmd;
+	uint32_t cmd = cmd_id;
+	int ret = 0;
+
+	if (!in)
+		return -EINVAL;
+
+	mutex_lock(&psp->ras_context.mutex);
+	ras_cmd = (struct ta_ras_shared_memory *)psp->ras_context.context.mem_context.shared_buf;
+	memset(ras_cmd, 0, sizeof(struct ta_ras_shared_memory));
+
+	switch (cmd) {
+	case TA_RAS_COMMAND__ENABLE_FEATURES:
+	case TA_RAS_COMMAND__DISABLE_FEATURES:
+		memcpy(&ras_cmd->ras_in_message,
+			in, sizeof(ras_cmd->ras_in_message));
+		break;
+	case TA_RAS_COMMAND__TRIGGER_ERROR:
+		memcpy(&ras_cmd->ras_in_message.trigger_error,
+			in, sizeof(ras_cmd->ras_in_message.trigger_error));
+		break;
+	case TA_RAS_COMMAND__QUERY_ADDRESS:
+		memcpy(&ras_cmd->ras_in_message.address,
+			in, sizeof(ras_cmd->ras_in_message.address));
+		break;
+	default:
+		dev_err(psp->adev->dev, "Invalid ras cmd id: %u\n", cmd);
+		ret = -EINVAL;
+		goto err_out;
+	}
+
+	ras_cmd->cmd_id = cmd;
+	ret = psp_ras_invoke(psp, ras_cmd->cmd_id);
+
+	switch (cmd) {
+	case TA_RAS_COMMAND__TRIGGER_ERROR:
+		if (!ret && out)
+			memcpy(out, &ras_cmd->ras_status, sizeof(ras_cmd->ras_status));
+		break;
+	case TA_RAS_COMMAND__QUERY_ADDRESS:
+		if (ret || ras_cmd->ras_status || psp->cmd_buf_mem->resp.status)
+			ret = -EINVAL;
+		else if (out)
+			memcpy(out,
+				&ras_cmd->ras_out_message.address,
+				sizeof(ras_cmd->ras_out_message.address));
+		break;
+	default:
+		break;
+	}
+
+err_out:
+	mutex_unlock(&psp->ras_context.mutex);
+
+	return ret;
+}
+
+int psp_ras_invoke(struct psp_context *psp, uint32_t ta_cmd_id)
+{
+	struct ta_ras_shared_memory *ras_cmd;
+	int ret;
+
+	ras_cmd = (struct ta_ras_shared_memory *)psp->ras_context.context.mem_context.shared_buf;
+
+	/*
+	 * TODO: bypass the loading in sriov for now
+	 */
+	if (amdgpu_sriov_vf(psp->adev))
+		return 0;
+
+	ret = psp_ta_invoke(psp, ta_cmd_id, &psp->ras_context.context);
+
+	if (amdgpu_ras_intr_triggered())
+		return ret;
+
+	if (ras_cmd->if_version > RAS_TA_HOST_IF_VER) {
+		dev_warn(psp->adev->dev, "RAS: Unsupported Interface\n");
+		return -EINVAL;
+	}
+
+	if (!ret) {
+		if (ras_cmd->ras_out_message.flags.err_inject_switch_disable_flag) {
+			dev_warn(psp->adev->dev, "ECC switch disabled\n");
+
+			ras_cmd->ras_status = TA_RAS_STATUS__ERROR_RAS_NOT_AVAILABLE;
+		} else if (ras_cmd->ras_out_message.flags.reg_access_failure_flag)
+			dev_warn(psp->adev->dev,
+				 "RAS internal register access blocked\n");
+
+		psp_ras_ta_check_status(psp);
+	}
+
+	return ret;
+}
+
+int psp_ras_enable_features(struct psp_context *psp,
+		union ta_ras_cmd_input *info, bool enable)
+{
+	enum ras_command cmd_id;
+	int ret;
+
+	if (!psp->ras_context.context.initialized || !info)
+		return -EINVAL;
+
+	cmd_id = enable ?
+		TA_RAS_COMMAND__ENABLE_FEATURES : TA_RAS_COMMAND__DISABLE_FEATURES;
+	ret = psp_ras_send_cmd(psp, cmd_id, info, NULL);
 	if (ret)
+		return -EINVAL;
+
+	return 0;
+}
+
+int psp_ras_terminate(struct psp_context *psp)
+{
+	int ret;
+
+	/*
+	 * TODO: bypass the terminate in sriov for now
+	 */
+	if (amdgpu_sriov_vf(psp->adev))
+		return 0;
+
+	if (!psp->ras_context.context.initialized)
+		return 0;
+
+	ret = psp_ta_unload(psp, &psp->ras_context.context);
+
+	psp->ras_context.context.initialized = false;
+
+	mutex_destroy(&psp->ras_context.mutex);
+
+	return ret;
+}
+
+int psp_ras_initialize(struct psp_context *psp)
+{
+	int ret;
+	uint32_t boot_cfg = 0xFF;
+	struct amdgpu_device *adev = psp->adev;
+	struct ta_ras_shared_memory *ras_cmd;
+
+	/*
+	 * TODO: bypass the initialize in sriov for now
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	if (!adev->psp.ras_context.context.bin_desc.size_bytes ||
+	    !adev->psp.ras_context.context.bin_desc.start_addr) {
+		dev_info(adev->dev, "RAS: optional ras ta ucode is not available\n");
+		return 0;
+	}
+
+	if (amdgpu_atomfirmware_dynamic_boot_config_supported(adev)) {
+		/* query GECC enablement status from boot config
+		 * boot_cfg: 1: GECC is enabled or 0: GECC is disabled
+		 */
+		ret = psp_boot_config_get(adev, &boot_cfg);
+		if (ret)
+			dev_warn(adev->dev, "PSP get boot config failed\n");
+
+		if (boot_cfg == 1 && !adev->ras_default_ecc_enabled &&
+		    amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__UMC)) {
+			dev_warn(adev->dev, "GECC is currently enabled, which may affect performance\n");
+			dev_warn(adev->dev,
+				"To disable GECC, please reboot the system and load the amdgpu driver with the parameter amdgpu_ras_enable=0\n");
+		} else {
+			if ((adev->ras_default_ecc_enabled || amdgpu_ras_enable == 1) &&
+				amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__UMC)) {
+				if (boot_cfg == 1) {
+					dev_info(adev->dev, "GECC is enabled\n");
+				} else {
+					/* enable GECC in next boot cycle if it is disabled
+					 * in boot config, or force enable GECC if failed to
+					 * get boot configuration
+					 */
+					ret = psp_boot_config_set(adev, BOOT_CONFIG_GECC);
+					if (ret)
+						dev_warn(adev->dev, "PSP set boot config failed\n");
+					else
+						dev_warn(adev->dev, "GECC will be enabled in next boot cycle\n");
+				}
+			} else {
+				if (!boot_cfg) {
+					if (!adev->ras_default_ecc_enabled &&
+					    amdgpu_ras_enable != 1 &&
+					    amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__UMC))
+						dev_warn(adev->dev, "GECC is disabled, set amdgpu_ras_enable=1 to enable GECC in next boot cycle if needed\n");
+					else
+						dev_info(adev->dev, "GECC is disabled\n");
+				} else {
+					/* disable GECC in next boot cycle if ras is
+					 * disabled by module parameter amdgpu_ras_enable
+					 * and/or amdgpu_ras_mask, or boot_config_get call
+					 * is failed
+					 */
+					ret = psp_boot_config_set(adev, 0);
+					if (ret)
+						dev_warn(adev->dev, "PSP set boot config failed\n");
+					else
+						dev_warn(adev->dev, "GECC will be disabled in next boot cycle if set amdgpu_ras_enable and/or amdgpu_ras_mask to 0x0\n");
+				}
+			}
+		}
+	}
+
+	psp->ras_context.context.mem_context.shared_mem_size = PSP_RAS_SHARED_MEM_SIZE;
+	psp->ras_context.context.ta_load_type = GFX_CMD_ID_LOAD_TA;
+
+	if (!psp->ras_context.context.mem_context.shared_buf) {
+		ret = psp_ta_init_shared_buf(psp, &psp->ras_context.context.mem_context);
+		if (ret)
+			return ret;
+	}
+
+	ras_cmd = (struct ta_ras_shared_memory *)psp->ras_context.context.mem_context.shared_buf;
+	memset(ras_cmd, 0, sizeof(struct ta_ras_shared_memory));
+
+	if (amdgpu_ras_is_poison_mode_supported(adev))
+		ras_cmd->ras_in_message.init_flags.poison_mode_en = 1;
+	if (!adev->gmc.xgmi.connected_to_cpu && !adev->gmc.is_app_apu)
+		ras_cmd->ras_in_message.init_flags.dgpu_mode = 1;
+	ras_cmd->ras_in_message.init_flags.xcc_mask =
+		adev->gfx.xcc_mask;
+	ras_cmd->ras_in_message.init_flags.channel_dis_num = hweight32(adev->gmc.m_half_use) * 2;
+	if (adev->gmc.gmc_funcs->query_mem_partition_mode)
+		ras_cmd->ras_in_message.init_flags.nps_mode =
+			adev->gmc.gmc_funcs->query_mem_partition_mode(adev);
+	ras_cmd->ras_in_message.init_flags.active_umc_mask = adev->umc.active_mask;
+
+	ret = psp_ta_load(psp, &psp->ras_context.context);
+
+	if (!ret && !ras_cmd->ras_status) {
+		psp->ras_context.context.initialized = true;
+		mutex_init(&psp->ras_context.mutex);
+	} else {
+		if (ras_cmd->ras_status)
+			dev_warn(adev->dev, "RAS Init Status: 0x%X\n", ras_cmd->ras_status);
+
+		/* fail to load RAS TA */
+		psp->ras_context.context.initialized = false;
+	}
+
+	return ret;
+}
+
+int psp_ras_trigger_error(struct psp_context *psp,
+			  struct ta_ras_trigger_error_input *info, uint32_t instance_mask)
+{
+	struct amdgpu_device *adev = psp->adev;
+	int ret;
+	uint32_t dev_mask;
+	uint32_t ras_status = 0;
+
+	if (!psp->ras_context.context.initialized || !info)
+		return -EINVAL;
+
+	switch (info->block_id) {
+	case TA_RAS_BLOCK__GFX:
+		dev_mask = GET_MASK(GC, instance_mask);
+		break;
+	case TA_RAS_BLOCK__SDMA:
+		dev_mask = GET_MASK(SDMA0, instance_mask);
+		break;
+	case TA_RAS_BLOCK__VCN:
+	case TA_RAS_BLOCK__JPEG:
+		dev_mask = GET_MASK(VCN, instance_mask);
+		break;
+	default:
+		dev_mask = instance_mask;
+		break;
+	}
+
+	/* reuse sub_block_index for backward compatibility */
+	dev_mask <<= AMDGPU_RAS_INST_SHIFT;
+	dev_mask &= AMDGPU_RAS_INST_MASK;
+	info->sub_block_index |= dev_mask;
+
+	ret = psp_ras_send_cmd(psp,
+			TA_RAS_COMMAND__TRIGGER_ERROR, info, &ras_status);
+	if (ret)
+		return -EINVAL;
+
+	/* If err_event_athub occurs error inject was successful, however
+	 *  return status from TA is no long reliable
+	 */
+	if (amdgpu_ras_intr_triggered())
+		return 0;
+
+	if (ras_status == TA_RAS_STATUS__TEE_ERROR_ACCESS_DENIED)
+		return -EACCES;
+	else if (ras_status)
+		return -EINVAL;
+
+	return 0;
+}
+
+int psp_ras_query_address(struct psp_context *psp,
+			  struct ta_ras_query_address_input *addr_in,
+			  struct ta_ras_query_address_output *addr_out)
+{
+	int ret;
+
+	if (!psp->ras_context.context.initialized ||
+		!addr_in || !addr_out)
+		return -EINVAL;
+
+	ret = psp_ras_send_cmd(psp,
+			TA_RAS_COMMAND__QUERY_ADDRESS, addr_in, addr_out);
+
+	return ret;
+}
+// ras end
+
+// HDCP start
+static int psp_hdcp_initialize(struct psp_context *psp)
+{
+	int ret;
+
+	/*
+	 * TODO: bypass the initialize in sriov for now
+	 */
+	if (amdgpu_sriov_vf(psp->adev))
+		return 0;
+
+	/* bypass hdcp initialization if dmu is harvested */
+	if (!amdgpu_device_has_display_hardware(psp->adev))
+		return 0;
+
+	if (!psp->hdcp_context.context.bin_desc.size_bytes ||
+	    !psp->hdcp_context.context.bin_desc.start_addr) {
+		dev_info(psp->adev->dev, "HDCP: optional hdcp ta ucode is not available\n");
+		return 0;
+	}
+
+	psp->hdcp_context.context.mem_context.shared_mem_size = PSP_HDCP_SHARED_MEM_SIZE;
+	psp->hdcp_context.context.ta_load_type = GFX_CMD_ID_LOAD_TA;
+
+	if (!psp->hdcp_context.context.mem_context.shared_buf) {
+		ret = psp_ta_init_shared_buf(psp, &psp->hdcp_context.context.mem_context);
+		if (ret)
+			return ret;
+	}
+
+	ret = psp_ta_load(psp, &psp->hdcp_context.context);
+	if (!ret) {
+		psp->hdcp_context.context.initialized = true;
+		mutex_init(&psp->hdcp_context.mutex);
+	}
+
+	return ret;
+}
+
+int psp_hdcp_invoke(struct psp_context *psp, uint32_t ta_cmd_id)
+{
+	/*
+	 * TODO: bypass the loading in sriov for now
+	 */
+	if (amdgpu_sriov_vf(psp->adev))
+		return 0;
+
+	if (!psp->hdcp_context.context.initialized)
+		return 0;
+
+	return psp_ta_invoke(psp, ta_cmd_id, &psp->hdcp_context.context);
+}
+
+static int psp_hdcp_terminate(struct psp_context *psp)
+{
+	int ret;
+
+	/*
+	 * TODO: bypass the terminate in sriov for now
+	 */
+	if (amdgpu_sriov_vf(psp->adev))
+		return 0;
+
+	if (!psp->hdcp_context.context.initialized)
+		return 0;
+
+	ret = psp_ta_unload(psp, &psp->hdcp_context.context);
+
+	psp->hdcp_context.context.initialized = false;
+
+	return ret;
+}
+// HDCP end
+
+// DTM start
+static int psp_dtm_initialize(struct psp_context *psp)
+{
+	int ret;
+
+	/*
+	 * TODO: bypass the initialize in sriov for now
+	 */
+	if (amdgpu_sriov_vf(psp->adev))
+		return 0;
+
+	/* bypass dtm initialization if dmu is harvested */
+	if (!amdgpu_device_has_display_hardware(psp->adev))
+		return 0;
+
+	if (!psp->dtm_context.context.bin_desc.size_bytes ||
+	    !psp->dtm_context.context.bin_desc.start_addr) {
+		dev_info(psp->adev->dev, "DTM: optional dtm ta ucode is not available\n");
+		return 0;
+	}
+
+	psp->dtm_context.context.mem_context.shared_mem_size = PSP_DTM_SHARED_MEM_SIZE;
+	psp->dtm_context.context.ta_load_type = GFX_CMD_ID_LOAD_TA;
+
+	if (!psp->dtm_context.context.mem_context.shared_buf) {
+		ret = psp_ta_init_shared_buf(psp, &psp->dtm_context.context.mem_context);
+		if (ret)
+			return ret;
+	}
+
+	ret = psp_ta_load(psp, &psp->dtm_context.context);
+	if (!ret) {
+		psp->dtm_context.context.initialized = true;
+		mutex_init(&psp->dtm_context.mutex);
+	}
+
+	return ret;
+}
+
+int psp_dtm_invoke(struct psp_context *psp, uint32_t ta_cmd_id)
+{
+	/*
+	 * TODO: bypass the loading in sriov for now
+	 */
+	if (amdgpu_sriov_vf(psp->adev))
+		return 0;
+
+	if (!psp->dtm_context.context.initialized)
+		return 0;
+
+	return psp_ta_invoke(psp, ta_cmd_id, &psp->dtm_context.context);
+}
+
+static int psp_dtm_terminate(struct psp_context *psp)
+{
+	int ret;
+
+	/*
+	 * TODO: bypass the terminate in sriov for now
+	 */
+	if (amdgpu_sriov_vf(psp->adev))
+		return 0;
+
+	if (!psp->dtm_context.context.initialized)
+		return 0;
+
+	ret = psp_ta_unload(psp, &psp->dtm_context.context);
+
+	psp->dtm_context.context.initialized = false;
+
+	return ret;
+}
+// DTM end
+
+// RAP start
+static int psp_rap_initialize(struct psp_context *psp)
+{
+	int ret;
+	enum ta_rap_status status = TA_RAP_STATUS__SUCCESS;
+
+	/*
+	 * TODO: bypass the initialize in sriov for now
+	 */
+	if (amdgpu_sriov_vf(psp->adev))
+		return 0;
+
+	if (!psp->rap_context.context.bin_desc.size_bytes ||
+	    !psp->rap_context.context.bin_desc.start_addr) {
+		dev_info(psp->adev->dev, "RAP: optional rap ta ucode is not available\n");
+		return 0;
+	}
+
+	psp->rap_context.context.mem_context.shared_mem_size = PSP_RAP_SHARED_MEM_SIZE;
+	psp->rap_context.context.ta_load_type = GFX_CMD_ID_LOAD_TA;
+
+	if (!psp->rap_context.context.mem_context.shared_buf) {
+		ret = psp_ta_init_shared_buf(psp, &psp->rap_context.context.mem_context);
+		if (ret)
+			return ret;
+	}
+
+	ret = psp_ta_load(psp, &psp->rap_context.context);
+	if (!ret) {
+		psp->rap_context.context.initialized = true;
+		mutex_init(&psp->rap_context.mutex);
+	} else
 		return ret;
 
-	ret = psp_asd_load(psp);
+	ret = psp_rap_invoke(psp, TA_CMD_RAP__INITIALIZE, &status);
+	if (ret || status != TA_RAP_STATUS__SUCCESS) {
+		psp_rap_terminate(psp);
+		/* free rap shared memory */
+		psp_ta_free_shared_buf(&psp->rap_context.context.mem_context);
+
+		dev_warn(psp->adev->dev, "RAP TA initialize fail (%d) status %d.\n",
+			 ret, status);
+
+		return ret;
+	}
+
+	return 0;
+}
+
+static int psp_rap_terminate(struct psp_context *psp)
+{
+	int ret;
+
+	if (!psp->rap_context.context.initialized)
+		return 0;
+
+	ret = psp_ta_unload(psp, &psp->rap_context.context);
+
+	psp->rap_context.context.initialized = false;
+
+	return ret;
+}
+
+int psp_rap_invoke(struct psp_context *psp, uint32_t ta_cmd_id, enum ta_rap_status *status)
+{
+	struct ta_rap_shared_memory *rap_cmd;
+	int ret = 0;
+
+	if (!psp->rap_context.context.initialized)
+		return 0;
+
+	if (ta_cmd_id != TA_CMD_RAP__INITIALIZE &&
+	    ta_cmd_id != TA_CMD_RAP__VALIDATE_L0)
+		return -EINVAL;
+
+	mutex_lock(&psp->rap_context.mutex);
+
+	rap_cmd = (struct ta_rap_shared_memory *)
+		  psp->rap_context.context.mem_context.shared_buf;
+	memset(rap_cmd, 0, sizeof(struct ta_rap_shared_memory));
+
+	rap_cmd->cmd_id = ta_cmd_id;
+	rap_cmd->validation_method_id = METHOD_A;
+
+	ret = psp_ta_invoke(psp, rap_cmd->cmd_id, &psp->rap_context.context);
 	if (ret)
+		goto out_unlock;
+
+	if (status)
+		*status = rap_cmd->rap_status;
+
+out_unlock:
+	mutex_unlock(&psp->rap_context.mutex);
+
+	return ret;
+}
+// RAP end
+
+/* securedisplay start */
+static int psp_securedisplay_initialize(struct psp_context *psp)
+{
+	int ret;
+	struct ta_securedisplay_cmd *securedisplay_cmd;
+
+	/*
+	 * TODO: bypass the initialize in sriov for now
+	 */
+	if (amdgpu_sriov_vf(psp->adev))
+		return 0;
+
+	/* bypass securedisplay initialization if dmu is harvested */
+	if (!amdgpu_device_has_display_hardware(psp->adev))
+		return 0;
+
+	if (!psp->securedisplay_context.context.bin_desc.size_bytes ||
+	    !psp->securedisplay_context.context.bin_desc.start_addr) {
+		dev_info(psp->adev->dev,
+			 "SECUREDISPLAY: optional securedisplay ta ucode is not available\n");
+		return 0;
+	}
+
+	psp->securedisplay_context.context.mem_context.shared_mem_size =
+		PSP_SECUREDISPLAY_SHARED_MEM_SIZE;
+	psp->securedisplay_context.context.ta_load_type = GFX_CMD_ID_LOAD_TA;
+
+	if (!psp->securedisplay_context.context.initialized) {
+		ret = psp_ta_init_shared_buf(psp,
+					     &psp->securedisplay_context.context.mem_context);
+		if (ret)
+			return ret;
+	}
+
+	ret = psp_ta_load(psp, &psp->securedisplay_context.context);
+	if (!ret && !psp->securedisplay_context.context.resp_status) {
+		psp->securedisplay_context.context.initialized = true;
+		mutex_init(&psp->securedisplay_context.mutex);
+	} else
 		return ret;
 
+	mutex_lock(&psp->securedisplay_context.mutex);
+
+	psp_prep_securedisplay_cmd_buf(psp, &securedisplay_cmd,
+			TA_SECUREDISPLAY_COMMAND__QUERY_TA);
+
+	ret = psp_securedisplay_invoke(psp, TA_SECUREDISPLAY_COMMAND__QUERY_TA);
+
+	mutex_unlock(&psp->securedisplay_context.mutex);
+
+	if (ret) {
+		psp_securedisplay_terminate(psp);
+		/* free securedisplay shared memory */
+		psp_ta_free_shared_buf(&psp->securedisplay_context.context.mem_context);
+		dev_err(psp->adev->dev, "SECUREDISPLAY TA initialize fail.\n");
+		return -EINVAL;
+	}
+
+	if (securedisplay_cmd->status != TA_SECUREDISPLAY_STATUS__SUCCESS) {
+		psp_securedisplay_parse_resp_status(psp, securedisplay_cmd->status);
+		dev_err(psp->adev->dev, "SECUREDISPLAY: query securedisplay TA failed. ret 0x%x\n",
+			securedisplay_cmd->securedisplay_out_message.query_ta.query_cmd_ret);
+		/* don't try again */
+		psp->securedisplay_context.context.bin_desc.size_bytes = 0;
+	}
+
 	return 0;
 }
 
-static int psp_np_fw_load(struct psp_context *psp)
+static int psp_securedisplay_terminate(struct psp_context *psp)
+{
+	int ret;
+
+	/*
+	 * TODO:bypass the terminate in sriov for now
+	 */
+	if (amdgpu_sriov_vf(psp->adev))
+		return 0;
+
+	if (!psp->securedisplay_context.context.initialized)
+		return 0;
+
+	ret = psp_ta_unload(psp, &psp->securedisplay_context.context);
+
+	psp->securedisplay_context.context.initialized = false;
+
+	return ret;
+}
+
+int psp_securedisplay_invoke(struct psp_context *psp, uint32_t ta_cmd_id)
+{
+	int ret;
+
+	if (!psp->securedisplay_context.context.initialized)
+		return -EINVAL;
+
+	if (ta_cmd_id != TA_SECUREDISPLAY_COMMAND__QUERY_TA &&
+	    ta_cmd_id != TA_SECUREDISPLAY_COMMAND__SEND_ROI_CRC &&
+	    ta_cmd_id != TA_SECUREDISPLAY_COMMAND__SEND_ROI_CRC_V2)
+		return -EINVAL;
+
+	ret = psp_ta_invoke(psp, ta_cmd_id, &psp->securedisplay_context.context);
+
+	return ret;
+}
+/* SECUREDISPLAY end */
+
+int amdgpu_psp_wait_for_bootloader(struct amdgpu_device *adev)
+{
+	struct psp_context *psp = &adev->psp;
+	int ret = 0;
+
+	if (!amdgpu_sriov_vf(adev) && psp->funcs && psp->funcs->wait_for_bootloader != NULL)
+		ret = psp->funcs->wait_for_bootloader(psp);
+
+	return ret;
+}
+
+bool amdgpu_psp_get_ras_capability(struct psp_context *psp)
+{
+	if (psp->funcs &&
+	    psp->funcs->get_ras_capability) {
+		return psp->funcs->get_ras_capability(psp);
+	} else {
+		return false;
+	}
+}
+
+bool amdgpu_psp_tos_reload_needed(struct amdgpu_device *adev)
+{
+	struct psp_context *psp = &adev->psp;
+
+	if (amdgpu_sriov_vf(adev) || (adev->flags & AMD_IS_APU))
+		return false;
+
+	if (psp->funcs && psp->funcs->is_reload_needed)
+		return psp->funcs->is_reload_needed(psp);
+
+	return false;
+}
+
+static void psp_update_gpu_addresses(struct amdgpu_device *adev)
+{
+	struct psp_context *psp = &adev->psp;
+
+	if (psp->cmd_buf_bo && psp->cmd_buf_mem) {
+		psp->fw_pri_mc_addr = amdgpu_bo_fb_aper_addr(psp->fw_pri_bo);
+		psp->fence_buf_mc_addr = amdgpu_bo_fb_aper_addr(psp->fence_buf_bo);
+		psp->cmd_buf_mc_addr = amdgpu_bo_fb_aper_addr(psp->cmd_buf_bo);
+	}
+	if (adev->firmware.rbuf && psp->km_ring.ring_mem)
+		psp->km_ring.ring_mem_mc_addr = amdgpu_bo_fb_aper_addr(adev->firmware.rbuf);
+}
+
+static int psp_hw_start(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+	int ret;
+
+	if (amdgpu_virt_xgmi_migrate_enabled(adev))
+		psp_update_gpu_addresses(adev);
+
+	if (!amdgpu_sriov_vf(adev)) {
+		if ((is_psp_fw_valid(psp->kdb)) &&
+		    (psp->funcs->bootloader_load_kdb != NULL)) {
+			ret = psp_bootloader_load_kdb(psp);
+			if (ret) {
+				dev_err(adev->dev, "PSP load kdb failed!\n");
+				return ret;
+			}
+		}
+
+		if ((is_psp_fw_valid(psp->spl)) &&
+		    (psp->funcs->bootloader_load_spl != NULL)) {
+			ret = psp_bootloader_load_spl(psp);
+			if (ret) {
+				dev_err(adev->dev, "PSP load spl failed!\n");
+				return ret;
+			}
+		}
+
+		if ((is_psp_fw_valid(psp->sys)) &&
+		    (psp->funcs->bootloader_load_sysdrv != NULL)) {
+			ret = psp_bootloader_load_sysdrv(psp);
+			if (ret) {
+				dev_err(adev->dev, "PSP load sys drv failed!\n");
+				return ret;
+			}
+		}
+
+		if ((is_psp_fw_valid(psp->soc_drv)) &&
+		    (psp->funcs->bootloader_load_soc_drv != NULL)) {
+			ret = psp_bootloader_load_soc_drv(psp);
+			if (ret) {
+				dev_err(adev->dev, "PSP load soc drv failed!\n");
+				return ret;
+			}
+		}
+
+		if ((is_psp_fw_valid(psp->intf_drv)) &&
+		    (psp->funcs->bootloader_load_intf_drv != NULL)) {
+			ret = psp_bootloader_load_intf_drv(psp);
+			if (ret) {
+				dev_err(adev->dev, "PSP load intf drv failed!\n");
+				return ret;
+			}
+		}
+
+		if ((is_psp_fw_valid(psp->dbg_drv)) &&
+		    (psp->funcs->bootloader_load_dbg_drv != NULL)) {
+			ret = psp_bootloader_load_dbg_drv(psp);
+			if (ret) {
+				dev_err(adev->dev, "PSP load dbg drv failed!\n");
+				return ret;
+			}
+		}
+
+		if ((is_psp_fw_valid(psp->ras_drv)) &&
+		    (psp->funcs->bootloader_load_ras_drv != NULL)) {
+			ret = psp_bootloader_load_ras_drv(psp);
+			if (ret) {
+				dev_err(adev->dev, "PSP load ras_drv failed!\n");
+				return ret;
+			}
+		}
+
+		if ((is_psp_fw_valid(psp->ipkeymgr_drv)) &&
+		    (psp->funcs->bootloader_load_ipkeymgr_drv != NULL)) {
+			ret = psp_bootloader_load_ipkeymgr_drv(psp);
+			if (ret) {
+				dev_err(adev->dev, "PSP load ipkeymgr_drv failed!\n");
+				return ret;
+			}
+		}
+
+		if ((is_psp_fw_valid(psp->spdm_drv)) &&
+		    (psp->funcs->bootloader_load_spdm_drv != NULL)) {
+			ret = psp_bootloader_load_spdm_drv(psp);
+			if (ret) {
+				dev_err(adev->dev, "PSP load spdm_drv failed!\n");
+				return ret;
+			}
+		}
+
+		if ((is_psp_fw_valid(psp->sos)) &&
+		    (psp->funcs->bootloader_load_sos != NULL)) {
+			ret = psp_bootloader_load_sos(psp);
+			if (ret) {
+				dev_err(adev->dev, "PSP load sos failed!\n");
+				return ret;
+			}
+		}
+	}
+
+	ret = psp_ring_create(psp, PSP_RING_TYPE__KM);
+	if (ret) {
+		dev_err(adev->dev, "PSP create ring failed!\n");
+		return ret;
+	}
+
+	if (!amdgpu_in_reset(adev) && !adev->in_suspend) {
+		ret = psp_update_fw_reservation(psp);
+		if (ret) {
+			dev_err(adev->dev, "update fw reservation failed!\n");
+			return ret;
+		}
+	}
+
+	if (amdgpu_sriov_vf(adev) && amdgpu_in_reset(adev))
+		goto skip_pin_bo;
+
+	if (!psp->boot_time_tmr || psp->autoload_supported) {
+		ret = psp_tmr_init(psp);
+		if (ret) {
+			dev_err(adev->dev, "PSP tmr init failed!\n");
+			return ret;
+		}
+	}
+
+skip_pin_bo:
+	/*
+	 * For ASICs with DF Cstate management centralized
+	 * to PMFW, TMR setup should be performed after PMFW
+	 * loaded and before other non-psp firmware loaded.
+	 */
+	if (psp->pmfw_centralized_cstate_management) {
+		ret = psp_load_smu_fw(psp);
+		if (ret)
+			return ret;
+	}
+
+	if (!psp->boot_time_tmr || !psp->autoload_supported) {
+		ret = psp_tmr_load(psp);
+		if (ret) {
+			dev_err(adev->dev, "PSP load tmr failed!\n");
+			return ret;
+		}
+	}
+
+	return 0;
+}
+
+static int psp_get_fw_type(struct amdgpu_firmware_info *ucode,
+			   enum psp_gfx_fw_type *type)
+{
+	switch (ucode->ucode_id) {
+	case AMDGPU_UCODE_ID_CAP:
+		*type = GFX_FW_TYPE_CAP;
+		break;
+	case AMDGPU_UCODE_ID_SDMA0:
+		*type = GFX_FW_TYPE_SDMA0;
+		break;
+	case AMDGPU_UCODE_ID_SDMA1:
+		*type = GFX_FW_TYPE_SDMA1;
+		break;
+	case AMDGPU_UCODE_ID_SDMA2:
+		*type = GFX_FW_TYPE_SDMA2;
+		break;
+	case AMDGPU_UCODE_ID_SDMA3:
+		*type = GFX_FW_TYPE_SDMA3;
+		break;
+	case AMDGPU_UCODE_ID_SDMA4:
+		*type = GFX_FW_TYPE_SDMA4;
+		break;
+	case AMDGPU_UCODE_ID_SDMA5:
+		*type = GFX_FW_TYPE_SDMA5;
+		break;
+	case AMDGPU_UCODE_ID_SDMA6:
+		*type = GFX_FW_TYPE_SDMA6;
+		break;
+	case AMDGPU_UCODE_ID_SDMA7:
+		*type = GFX_FW_TYPE_SDMA7;
+		break;
+	case AMDGPU_UCODE_ID_CP_MES:
+		*type = GFX_FW_TYPE_CP_MES;
+		break;
+	case AMDGPU_UCODE_ID_CP_MES_DATA:
+		*type = GFX_FW_TYPE_MES_STACK;
+		break;
+	case AMDGPU_UCODE_ID_CP_MES1:
+		*type = GFX_FW_TYPE_CP_MES_KIQ;
+		break;
+	case AMDGPU_UCODE_ID_CP_MES1_DATA:
+		*type = GFX_FW_TYPE_MES_KIQ_STACK;
+		break;
+	case AMDGPU_UCODE_ID_CP_CE:
+		*type = GFX_FW_TYPE_CP_CE;
+		break;
+	case AMDGPU_UCODE_ID_CP_PFP:
+		*type = GFX_FW_TYPE_CP_PFP;
+		break;
+	case AMDGPU_UCODE_ID_CP_ME:
+		*type = GFX_FW_TYPE_CP_ME;
+		break;
+	case AMDGPU_UCODE_ID_CP_MEC1:
+		*type = GFX_FW_TYPE_CP_MEC;
+		break;
+	case AMDGPU_UCODE_ID_CP_MEC1_JT:
+		*type = GFX_FW_TYPE_CP_MEC_ME1;
+		break;
+	case AMDGPU_UCODE_ID_CP_MEC2:
+		*type = GFX_FW_TYPE_CP_MEC;
+		break;
+	case AMDGPU_UCODE_ID_CP_MEC2_JT:
+		*type = GFX_FW_TYPE_CP_MEC_ME2;
+		break;
+	case AMDGPU_UCODE_ID_RLC_P:
+		*type = GFX_FW_TYPE_RLC_P;
+		break;
+	case AMDGPU_UCODE_ID_RLC_V:
+		*type = GFX_FW_TYPE_RLC_V;
+		break;
+	case AMDGPU_UCODE_ID_RLC_G:
+		*type = GFX_FW_TYPE_RLC_G;
+		break;
+	case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL:
+		*type = GFX_FW_TYPE_RLC_RESTORE_LIST_SRM_CNTL;
+		break;
+	case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM:
+		*type = GFX_FW_TYPE_RLC_RESTORE_LIST_GPM_MEM;
+		break;
+	case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM:
+		*type = GFX_FW_TYPE_RLC_RESTORE_LIST_SRM_MEM;
+		break;
+	case AMDGPU_UCODE_ID_RLC_IRAM:
+		*type = GFX_FW_TYPE_RLC_IRAM;
+		break;
+	case AMDGPU_UCODE_ID_RLC_DRAM:
+		*type = GFX_FW_TYPE_RLC_DRAM_BOOT;
+		break;
+	case AMDGPU_UCODE_ID_GLOBAL_TAP_DELAYS:
+		*type = GFX_FW_TYPE_GLOBAL_TAP_DELAYS;
+		break;
+	case AMDGPU_UCODE_ID_SE0_TAP_DELAYS:
+		*type = GFX_FW_TYPE_SE0_TAP_DELAYS;
+		break;
+	case AMDGPU_UCODE_ID_SE1_TAP_DELAYS:
+		*type = GFX_FW_TYPE_SE1_TAP_DELAYS;
+		break;
+	case AMDGPU_UCODE_ID_SE2_TAP_DELAYS:
+		*type = GFX_FW_TYPE_SE2_TAP_DELAYS;
+		break;
+	case AMDGPU_UCODE_ID_SE3_TAP_DELAYS:
+		*type = GFX_FW_TYPE_SE3_TAP_DELAYS;
+		break;
+	case AMDGPU_UCODE_ID_SMC:
+		*type = GFX_FW_TYPE_SMU;
+		break;
+	case AMDGPU_UCODE_ID_PPTABLE:
+		*type = GFX_FW_TYPE_PPTABLE;
+		break;
+	case AMDGPU_UCODE_ID_UVD:
+		*type = GFX_FW_TYPE_UVD;
+		break;
+	case AMDGPU_UCODE_ID_UVD1:
+		*type = GFX_FW_TYPE_UVD1;
+		break;
+	case AMDGPU_UCODE_ID_VCE:
+		*type = GFX_FW_TYPE_VCE;
+		break;
+	case AMDGPU_UCODE_ID_VCN:
+		*type = GFX_FW_TYPE_VCN;
+		break;
+	case AMDGPU_UCODE_ID_VCN1:
+		*type = GFX_FW_TYPE_VCN1;
+		break;
+	case AMDGPU_UCODE_ID_DMCU_ERAM:
+		*type = GFX_FW_TYPE_DMCU_ERAM;
+		break;
+	case AMDGPU_UCODE_ID_DMCU_INTV:
+		*type = GFX_FW_TYPE_DMCU_ISR;
+		break;
+	case AMDGPU_UCODE_ID_VCN0_RAM:
+		*type = GFX_FW_TYPE_VCN0_RAM;
+		break;
+	case AMDGPU_UCODE_ID_VCN1_RAM:
+		*type = GFX_FW_TYPE_VCN1_RAM;
+		break;
+	case AMDGPU_UCODE_ID_DMCUB:
+		*type = GFX_FW_TYPE_DMUB;
+		break;
+	case AMDGPU_UCODE_ID_SDMA_UCODE_TH0:
+	case AMDGPU_UCODE_ID_SDMA_RS64:
+		*type = GFX_FW_TYPE_SDMA_UCODE_TH0;
+		break;
+	case AMDGPU_UCODE_ID_SDMA_UCODE_TH1:
+		*type = GFX_FW_TYPE_SDMA_UCODE_TH1;
+		break;
+	case AMDGPU_UCODE_ID_IMU_I:
+		*type = GFX_FW_TYPE_IMU_I;
+		break;
+	case AMDGPU_UCODE_ID_IMU_D:
+		*type = GFX_FW_TYPE_IMU_D;
+		break;
+	case AMDGPU_UCODE_ID_CP_RS64_PFP:
+		*type = GFX_FW_TYPE_RS64_PFP;
+		break;
+	case AMDGPU_UCODE_ID_CP_RS64_ME:
+		*type = GFX_FW_TYPE_RS64_ME;
+		break;
+	case AMDGPU_UCODE_ID_CP_RS64_MEC:
+		*type = GFX_FW_TYPE_RS64_MEC;
+		break;
+	case AMDGPU_UCODE_ID_CP_RS64_PFP_P0_STACK:
+		*type = GFX_FW_TYPE_RS64_PFP_P0_STACK;
+		break;
+	case AMDGPU_UCODE_ID_CP_RS64_PFP_P1_STACK:
+		*type = GFX_FW_TYPE_RS64_PFP_P1_STACK;
+		break;
+	case AMDGPU_UCODE_ID_CP_RS64_ME_P0_STACK:
+		*type = GFX_FW_TYPE_RS64_ME_P0_STACK;
+		break;
+	case AMDGPU_UCODE_ID_CP_RS64_ME_P1_STACK:
+		*type = GFX_FW_TYPE_RS64_ME_P1_STACK;
+		break;
+	case AMDGPU_UCODE_ID_CP_RS64_MEC_P0_STACK:
+		*type = GFX_FW_TYPE_RS64_MEC_P0_STACK;
+		break;
+	case AMDGPU_UCODE_ID_CP_RS64_MEC_P1_STACK:
+		*type = GFX_FW_TYPE_RS64_MEC_P1_STACK;
+		break;
+	case AMDGPU_UCODE_ID_CP_RS64_MEC_P2_STACK:
+		*type = GFX_FW_TYPE_RS64_MEC_P2_STACK;
+		break;
+	case AMDGPU_UCODE_ID_CP_RS64_MEC_P3_STACK:
+		*type = GFX_FW_TYPE_RS64_MEC_P3_STACK;
+		break;
+	case AMDGPU_UCODE_ID_VPE_CTX:
+		*type = GFX_FW_TYPE_VPEC_FW1;
+		break;
+	case AMDGPU_UCODE_ID_VPE_CTL:
+		*type = GFX_FW_TYPE_VPEC_FW2;
+		break;
+	case AMDGPU_UCODE_ID_VPE:
+		*type = GFX_FW_TYPE_VPE;
+		break;
+	case AMDGPU_UCODE_ID_UMSCH_MM_UCODE:
+		*type = GFX_FW_TYPE_UMSCH_UCODE;
+		break;
+	case AMDGPU_UCODE_ID_UMSCH_MM_DATA:
+		*type = GFX_FW_TYPE_UMSCH_DATA;
+		break;
+	case AMDGPU_UCODE_ID_UMSCH_MM_CMD_BUFFER:
+		*type = GFX_FW_TYPE_UMSCH_CMD_BUFFER;
+		break;
+	case AMDGPU_UCODE_ID_P2S_TABLE:
+		*type = GFX_FW_TYPE_P2S_TABLE;
+		break;
+	case AMDGPU_UCODE_ID_JPEG_RAM:
+		*type = GFX_FW_TYPE_JPEG_RAM;
+		break;
+	case AMDGPU_UCODE_ID_ISP:
+		*type = GFX_FW_TYPE_ISP;
+		break;
+	case AMDGPU_UCODE_ID_MAXIMUM:
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static void psp_print_fw_hdr(struct psp_context *psp,
+			     struct amdgpu_firmware_info *ucode)
+{
+	struct amdgpu_device *adev = psp->adev;
+	struct common_firmware_header *hdr;
+
+	switch (ucode->ucode_id) {
+	case AMDGPU_UCODE_ID_SDMA0:
+	case AMDGPU_UCODE_ID_SDMA1:
+	case AMDGPU_UCODE_ID_SDMA2:
+	case AMDGPU_UCODE_ID_SDMA3:
+	case AMDGPU_UCODE_ID_SDMA4:
+	case AMDGPU_UCODE_ID_SDMA5:
+	case AMDGPU_UCODE_ID_SDMA6:
+	case AMDGPU_UCODE_ID_SDMA7:
+		hdr = (struct common_firmware_header *)
+			adev->sdma.instance[ucode->ucode_id - AMDGPU_UCODE_ID_SDMA0].fw->data;
+		amdgpu_ucode_print_sdma_hdr(hdr);
+		break;
+	case AMDGPU_UCODE_ID_CP_CE:
+		hdr = (struct common_firmware_header *)adev->gfx.ce_fw->data;
+		amdgpu_ucode_print_gfx_hdr(hdr);
+		break;
+	case AMDGPU_UCODE_ID_CP_PFP:
+		hdr = (struct common_firmware_header *)adev->gfx.pfp_fw->data;
+		amdgpu_ucode_print_gfx_hdr(hdr);
+		break;
+	case AMDGPU_UCODE_ID_CP_ME:
+		hdr = (struct common_firmware_header *)adev->gfx.me_fw->data;
+		amdgpu_ucode_print_gfx_hdr(hdr);
+		break;
+	case AMDGPU_UCODE_ID_CP_MEC1:
+		hdr = (struct common_firmware_header *)adev->gfx.mec_fw->data;
+		amdgpu_ucode_print_gfx_hdr(hdr);
+		break;
+	case AMDGPU_UCODE_ID_RLC_G:
+		hdr = (struct common_firmware_header *)adev->gfx.rlc_fw->data;
+		amdgpu_ucode_print_rlc_hdr(hdr);
+		break;
+	case AMDGPU_UCODE_ID_SMC:
+		hdr = (struct common_firmware_header *)adev->pm.fw->data;
+		amdgpu_ucode_print_smc_hdr(hdr);
+		break;
+	default:
+		break;
+	}
+}
+
+static int psp_prep_load_ip_fw_cmd_buf(struct psp_context *psp,
+				       struct amdgpu_firmware_info *ucode,
+				       struct psp_gfx_cmd_resp *cmd)
+{
+	int ret;
+	uint64_t fw_mem_mc_addr = ucode->mc_addr;
+
+	cmd->cmd_id = GFX_CMD_ID_LOAD_IP_FW;
+	cmd->cmd.cmd_load_ip_fw.fw_phy_addr_lo = lower_32_bits(fw_mem_mc_addr);
+	cmd->cmd.cmd_load_ip_fw.fw_phy_addr_hi = upper_32_bits(fw_mem_mc_addr);
+	cmd->cmd.cmd_load_ip_fw.fw_size = ucode->ucode_size;
+
+	ret = psp_get_fw_type(ucode, &cmd->cmd.cmd_load_ip_fw.fw_type);
+	if (ret)
+		dev_err(psp->adev->dev, "Unknown firmware type\n");
+
+	return ret;
+}
+
+int psp_execute_ip_fw_load(struct psp_context *psp,
+			   struct amdgpu_firmware_info *ucode)
+{
+	int ret = 0;
+	struct psp_gfx_cmd_resp *cmd = acquire_psp_cmd_buf(psp);
+
+	ret = psp_prep_load_ip_fw_cmd_buf(psp, ucode, cmd);
+	if (!ret) {
+		ret = psp_cmd_submit_buf(psp, ucode, cmd,
+					 psp->fence_buf_mc_addr);
+	}
+
+	release_psp_cmd_buf(psp);
+
+	return ret;
+}
+
+static int psp_load_p2s_table(struct psp_context *psp)
+{
+	int ret;
+	struct amdgpu_device *adev = psp->adev;
+	struct amdgpu_firmware_info *ucode =
+		&adev->firmware.ucode[AMDGPU_UCODE_ID_P2S_TABLE];
+
+	if (adev->in_runpm && ((adev->pm.rpm_mode == AMDGPU_RUNPM_BACO) ||
+				(adev->pm.rpm_mode == AMDGPU_RUNPM_BAMACO)))
+		return 0;
+
+	if (amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 6) ||
+	    amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 14)) {
+		uint32_t supp_vers = adev->flags & AMD_IS_APU ? 0x0036013D :
+								0x0036003C;
+		if (psp->sos.fw_version < supp_vers)
+			return 0;
+	}
+
+	if (!ucode->fw || amdgpu_sriov_vf(psp->adev))
+		return 0;
+
+	ret = psp_execute_ip_fw_load(psp, ucode);
+
+	return ret;
+}
+
+static int psp_load_smu_fw(struct psp_context *psp)
+{
+	int ret;
+	struct amdgpu_device *adev = psp->adev;
+	struct amdgpu_firmware_info *ucode =
+			&adev->firmware.ucode[AMDGPU_UCODE_ID_SMC];
+	struct amdgpu_ras *ras = psp->ras_context.ras;
+
+	/*
+	 * Skip SMU FW reloading in case of using BACO for runpm only,
+	 * as SMU is always alive.
+	 */
+	if (adev->in_runpm && ((adev->pm.rpm_mode == AMDGPU_RUNPM_BACO) ||
+				(adev->pm.rpm_mode == AMDGPU_RUNPM_BAMACO)))
+		return 0;
+
+	if (!ucode->fw || amdgpu_sriov_vf(psp->adev))
+		return 0;
+
+	if ((amdgpu_in_reset(adev) && ras && adev->ras_enabled &&
+	     (amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(11, 0, 4) ||
+	      amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(11, 0, 2)))) {
+		ret = amdgpu_dpm_set_mp1_state(adev, PP_MP1_STATE_UNLOAD);
+		if (ret)
+			dev_err(adev->dev, "Failed to set MP1 state prepare for reload\n");
+	}
+
+	ret = psp_execute_ip_fw_load(psp, ucode);
+
+	if (ret)
+		dev_err(adev->dev, "PSP load smu failed!\n");
+
+	return ret;
+}
+
+static bool fw_load_skip_check(struct psp_context *psp,
+			       struct amdgpu_firmware_info *ucode)
+{
+	if (!ucode->fw || !ucode->ucode_size)
+		return true;
+
+	if (ucode->ucode_id == AMDGPU_UCODE_ID_P2S_TABLE)
+		return true;
+
+	if (ucode->ucode_id == AMDGPU_UCODE_ID_SMC &&
+	    (psp_smu_reload_quirk(psp) ||
+	     psp->autoload_supported ||
+	     psp->pmfw_centralized_cstate_management))
+		return true;
+
+	if (amdgpu_sriov_vf(psp->adev) &&
+	    amdgpu_virt_fw_load_skip_check(psp->adev, ucode->ucode_id))
+		return true;
+
+	if (psp->autoload_supported &&
+	    (ucode->ucode_id == AMDGPU_UCODE_ID_CP_MEC1_JT ||
+	     ucode->ucode_id == AMDGPU_UCODE_ID_CP_MEC2_JT))
+		/* skip mec JT when autoload is enabled */
+		return true;
+
+	return false;
+}
+
+int psp_load_fw_list(struct psp_context *psp,
+		     struct amdgpu_firmware_info **ucode_list, int ucode_count)
+{
+	int ret = 0, i;
+	struct amdgpu_firmware_info *ucode;
+
+	for (i = 0; i < ucode_count; ++i) {
+		ucode = ucode_list[i];
+		psp_print_fw_hdr(psp, ucode);
+		ret = psp_execute_ip_fw_load(psp, ucode);
+		if (ret)
+			return ret;
+	}
+	return ret;
+}
+
+static int psp_load_non_psp_fw(struct psp_context *psp)
 {
 	int i, ret;
 	struct amdgpu_firmware_info *ucode;
-	struct amdgpu_device* adev = psp->adev;
+	struct amdgpu_device *adev = psp->adev;
+
+	if (psp->autoload_supported &&
+	    !psp->pmfw_centralized_cstate_management) {
+		ret = psp_load_smu_fw(psp);
+		if (ret)
+			return ret;
+	}
+
+	/* Load P2S table first if it's available */
+	psp_load_p2s_table(psp);
 
 	for (i = 0; i < adev->firmware.max_ucodes; i++) {
 		ucode = &adev->firmware.ucode[i];
-		if (!ucode->fw)
-			continue;
 
 		if (ucode->ucode_id == AMDGPU_UCODE_ID_SMC &&
-		    psp_smu_reload_quirk(psp))
+		    !fw_load_skip_check(psp, ucode)) {
+			ret = psp_load_smu_fw(psp);
+			if (ret)
+				return ret;
+			continue;
+		}
+
+		if (fw_load_skip_check(psp, ucode))
 			continue;
-		if (amdgpu_sriov_vf(adev) &&
-		   (ucode->ucode_id == AMDGPU_UCODE_ID_SDMA0
-		    || ucode->ucode_id == AMDGPU_UCODE_ID_SDMA1
-		    || ucode->ucode_id == AMDGPU_UCODE_ID_RLC_G))
-			/*skip ucode loading in SRIOV VF */
+
+		if (psp->autoload_supported &&
+		    (amdgpu_ip_version(adev, MP0_HWIP, 0) ==
+			     IP_VERSION(11, 0, 7) ||
+		     amdgpu_ip_version(adev, MP0_HWIP, 0) ==
+			     IP_VERSION(11, 0, 11) ||
+		     amdgpu_ip_version(adev, MP0_HWIP, 0) ==
+			     IP_VERSION(11, 0, 12)) &&
+		    (ucode->ucode_id == AMDGPU_UCODE_ID_SDMA1 ||
+		     ucode->ucode_id == AMDGPU_UCODE_ID_SDMA2 ||
+		     ucode->ucode_id == AMDGPU_UCODE_ID_SDMA3))
+			/* PSP only receive one SDMA fw for sienna_cichlid,
+			 * as all four sdma fw are same
+			 */
 			continue;
 
-		ret = psp_prep_cmd_buf(ucode, psp->cmd);
-		if (ret)
-			return ret;
+		psp_print_fw_hdr(psp, ucode);
 
-		ret = psp_cmd_submit_buf(psp, ucode, psp->cmd,
-					 psp->fence_buf_mc_addr, i + 3);
+		ret = psp_execute_ip_fw_load(psp, ucode);
 		if (ret)
 			return ret;
 
-#if 0
-		/* check if firmware loaded sucessfully */
-		if (!amdgpu_psp_check_fw_loading_status(adev, i))
-			return -EINVAL;
-#endif
+		/* Start rlc autoload after psp received all the gfx firmware */
+		if (psp->autoload_supported && ucode->ucode_id == (amdgpu_sriov_vf(adev) ?
+		    adev->virt.autoload_ucode_id : AMDGPU_UCODE_ID_RLC_G)) {
+			ret = psp_rlc_autoload_start(psp);
+			if (ret) {
+				dev_err(adev->dev, "Failed to start rlc autoload\n");
+				return ret;
+			}
+		}
 	}
 
 	return 0;
@@ -323,94 +3099,106 @@ static int psp_load_fw(struct amdgpu_device *adev)
 	int ret;
 	struct psp_context *psp = &adev->psp;
 
-	psp->cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL);
-	if (!psp->cmd)
-		return -ENOMEM;
+	if (amdgpu_sriov_vf(adev) && amdgpu_in_reset(adev)) {
+		/* should not destroy ring, only stop */
+		psp_ring_stop(psp, PSP_RING_TYPE__KM);
+	} else {
+		memset(psp->fence_buf, 0, PSP_FENCE_BUFFER_SIZE);
 
-	ret = amdgpu_bo_create_kernel(adev, PSP_1_MEG, PSP_1_MEG,
-				      AMDGPU_GEM_DOMAIN_GTT,
-				      &psp->fw_pri_bo,
-				      &psp->fw_pri_mc_addr,
-				      &psp->fw_pri_buf);
+		ret = psp_ring_init(psp, PSP_RING_TYPE__KM);
+		if (ret) {
+			dev_err(adev->dev, "PSP ring init failed!\n");
+			goto failed;
+		}
+	}
+
+	ret = psp_hw_start(psp);
 	if (ret)
 		goto failed;
 
-	ret = amdgpu_bo_create_kernel(adev, PSP_FENCE_BUFFER_SIZE, PAGE_SIZE,
-				      AMDGPU_GEM_DOMAIN_VRAM,
-				      &psp->fence_buf_bo,
-				      &psp->fence_buf_mc_addr,
-				      &psp->fence_buf);
+	ret = psp_load_non_psp_fw(psp);
 	if (ret)
-		goto failed_mem2;
+		goto failed1;
 
-	ret = amdgpu_bo_create_kernel(adev, PSP_CMD_BUFFER_SIZE, PAGE_SIZE,
-				      AMDGPU_GEM_DOMAIN_VRAM,
-				      &psp->cmd_buf_bo, &psp->cmd_buf_mc_addr,
-				      (void **)&psp->cmd_buf_mem);
-	if (ret)
-		goto failed_mem1;
+	ret = psp_asd_initialize(psp);
+	if (ret) {
+		dev_err(adev->dev, "PSP load asd failed!\n");
+		goto failed1;
+	}
 
-	memset(psp->fence_buf, 0, PSP_FENCE_BUFFER_SIZE);
+	ret = psp_rl_load(adev);
+	if (ret) {
+		dev_err(adev->dev, "PSP load RL failed!\n");
+		goto failed1;
+	}
 
-	ret = psp_ring_init(psp, PSP_RING_TYPE__KM);
-	if (ret)
-		goto failed_mem;
+	if (amdgpu_sriov_vf(adev) && amdgpu_in_reset(adev)) {
+		if (adev->gmc.xgmi.num_physical_nodes > 1) {
+			ret = psp_xgmi_initialize(psp, false, true);
+			/* Warning the XGMI seesion initialize failure
+			 * Instead of stop driver initialization
+			 */
+			if (ret)
+				dev_err(psp->adev->dev,
+					"XGMI: Failed to initialize XGMI session\n");
+		}
+	}
 
-	ret = psp_tmr_init(psp);
-	if (ret)
-		goto failed_mem;
+	if (psp->ta_fw) {
+		ret = psp_ras_initialize(psp);
+		if (ret)
+			dev_err(psp->adev->dev,
+				"RAS: Failed to initialize RAS\n");
 
-	ret = psp_asd_init(psp);
-	if (ret)
-		goto failed_mem;
+		ret = psp_hdcp_initialize(psp);
+		if (ret)
+			dev_err(psp->adev->dev,
+				"HDCP: Failed to initialize HDCP\n");
 
-	ret = psp_hw_start(psp);
-	if (ret)
-		goto failed_mem;
+		ret = psp_dtm_initialize(psp);
+		if (ret)
+			dev_err(psp->adev->dev,
+				"DTM: Failed to initialize DTM\n");
 
-	ret = psp_np_fw_load(psp);
-	if (ret)
-		goto failed_mem;
+		ret = psp_rap_initialize(psp);
+		if (ret)
+			dev_err(psp->adev->dev,
+				"RAP: Failed to initialize RAP\n");
+
+		ret = psp_securedisplay_initialize(psp);
+		if (ret)
+			dev_err(psp->adev->dev,
+				"SECUREDISPLAY: Failed to initialize SECUREDISPLAY\n");
+	}
 
 	return 0;
 
-failed_mem:
-	amdgpu_bo_free_kernel(&psp->cmd_buf_bo,
-			      &psp->cmd_buf_mc_addr,
-			      (void **)&psp->cmd_buf_mem);
-failed_mem1:
-	amdgpu_bo_free_kernel(&psp->fence_buf_bo,
-			      &psp->fence_buf_mc_addr, &psp->fence_buf);
-failed_mem2:
-	amdgpu_bo_free_kernel(&psp->fw_pri_bo,
-			      &psp->fw_pri_mc_addr, &psp->fw_pri_buf);
+failed1:
+	psp_free_shared_bufs(psp);
 failed:
-	kfree(psp->cmd);
-	psp->cmd = NULL;
+	/*
+	 * all cleanup jobs (xgmi terminate, ras terminate,
+	 * ring destroy, cmd/fence/fw buffers destory,
+	 * psp->cmd destory) are delayed to psp_hw_fini
+	 */
+	psp_ring_destroy(psp, PSP_RING_TYPE__KM);
 	return ret;
 }
 
-static int psp_hw_init(void *handle)
+static int psp_hw_init(struct amdgpu_ip_block *ip_block)
 {
 	int ret;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-
-	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
-		return 0;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	mutex_lock(&adev->firmware.mutex);
-	/*
-	 * This sequence is just used on hw_init only once, no need on
-	 * resume.
-	 */
+
 	ret = amdgpu_ucode_init_bo(adev);
 	if (ret)
 		goto failed;
 
 	ret = psp_load_fw(adev);
 	if (ret) {
-		DRM_ERROR("PSP firmware loading failed\n");
+		dev_err(adev->dev, "PSP firmware loading failed\n");
 		goto failed;
 	}
 
@@ -423,131 +3211,1252 @@ failed:
 	return -EINVAL;
 }
 
-static int psp_hw_fini(void *handle)
+static int psp_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	struct psp_context *psp = &adev->psp;
 
-	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
-		return 0;
+	if (psp->ta_fw) {
+		psp_ras_terminate(psp);
+		psp_securedisplay_terminate(psp);
+		psp_rap_terminate(psp);
+		psp_dtm_terminate(psp);
+		psp_hdcp_terminate(psp);
 
-	amdgpu_ucode_fini_bo(adev);
-
-	psp_ring_destroy(psp, PSP_RING_TYPE__KM);
+		if (adev->gmc.xgmi.num_physical_nodes > 1)
+			psp_xgmi_terminate(psp);
+	}
 
-	amdgpu_bo_free_kernel(&psp->tmr_bo, &psp->tmr_mc_addr, &psp->tmr_buf);
-	amdgpu_bo_free_kernel(&psp->fw_pri_bo,
-			      &psp->fw_pri_mc_addr, &psp->fw_pri_buf);
-	amdgpu_bo_free_kernel(&psp->fence_buf_bo,
-			      &psp->fence_buf_mc_addr, &psp->fence_buf);
-	amdgpu_bo_free_kernel(&psp->asd_shared_bo, &psp->asd_shared_mc_addr,
-			      &psp->asd_shared_buf);
-	amdgpu_bo_free_kernel(&psp->cmd_buf_bo, &psp->cmd_buf_mc_addr,
-			      (void **)&psp->cmd_buf_mem);
+	psp_asd_terminate(psp);
+	psp_tmr_terminate(psp);
 
-	kfree(psp->cmd);
-	psp->cmd = NULL;
+	psp_ring_destroy(psp, PSP_RING_TYPE__KM);
 
 	return 0;
 }
 
-static int psp_suspend(void *handle)
+static int psp_suspend(struct amdgpu_ip_block *ip_block)
 {
-	return 0;
+	int ret = 0;
+	struct amdgpu_device *adev = ip_block->adev;
+	struct psp_context *psp = &adev->psp;
+
+	if (adev->gmc.xgmi.num_physical_nodes > 1 &&
+	    psp->xgmi_context.context.initialized) {
+		ret = psp_xgmi_terminate(psp);
+		if (ret) {
+			dev_err(adev->dev, "Failed to terminate xgmi ta\n");
+			goto out;
+		}
+	}
+
+	if (psp->ta_fw) {
+		ret = psp_ras_terminate(psp);
+		if (ret) {
+			dev_err(adev->dev, "Failed to terminate ras ta\n");
+			goto out;
+		}
+		ret = psp_hdcp_terminate(psp);
+		if (ret) {
+			dev_err(adev->dev, "Failed to terminate hdcp ta\n");
+			goto out;
+		}
+		ret = psp_dtm_terminate(psp);
+		if (ret) {
+			dev_err(adev->dev, "Failed to terminate dtm ta\n");
+			goto out;
+		}
+		ret = psp_rap_terminate(psp);
+		if (ret) {
+			dev_err(adev->dev, "Failed to terminate rap ta\n");
+			goto out;
+		}
+		ret = psp_securedisplay_terminate(psp);
+		if (ret) {
+			dev_err(adev->dev, "Failed to terminate securedisplay ta\n");
+			goto out;
+		}
+	}
+
+	ret = psp_asd_terminate(psp);
+	if (ret) {
+		dev_err(adev->dev, "Failed to terminate asd\n");
+		goto out;
+	}
+
+	ret = psp_tmr_terminate(psp);
+	if (ret) {
+		dev_err(adev->dev, "Failed to terminate tmr\n");
+		goto out;
+	}
+
+	ret = psp_ring_stop(psp, PSP_RING_TYPE__KM);
+	if (ret)
+		dev_err(adev->dev, "PSP ring stop failed\n");
+
+out:
+	return ret;
 }
 
-static int psp_resume(void *handle)
+static int psp_resume(struct amdgpu_ip_block *ip_block)
 {
 	int ret;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	struct psp_context *psp = &adev->psp;
 
-	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
-		return 0;
+	dev_info(adev->dev, "PSP is resuming...\n");
 
-	DRM_INFO("PSP is resuming...\n");
+	if (psp->mem_train_ctx.enable_mem_training) {
+		ret = psp_mem_training(psp, PSP_MEM_TRAIN_RESUME);
+		if (ret) {
+			dev_err(adev->dev, "Failed to process memory training!\n");
+			return ret;
+		}
+	}
 
 	mutex_lock(&adev->firmware.mutex);
 
+	ret = amdgpu_ucode_init_bo(adev);
+	if (ret)
+		goto failed;
+
 	ret = psp_hw_start(psp);
 	if (ret)
 		goto failed;
 
-	ret = psp_np_fw_load(psp);
+	ret = psp_load_non_psp_fw(psp);
 	if (ret)
 		goto failed;
 
+	ret = psp_asd_initialize(psp);
+	if (ret) {
+		dev_err(adev->dev, "PSP load asd failed!\n");
+		goto failed;
+	}
+
+	ret = psp_rl_load(adev);
+	if (ret) {
+		dev_err(adev->dev, "PSP load RL failed!\n");
+		goto failed;
+	}
+
+	if (adev->gmc.xgmi.num_physical_nodes > 1) {
+		ret = psp_xgmi_initialize(psp, false, true);
+		/* Warning the XGMI seesion initialize failure
+		 * Instead of stop driver initialization
+		 */
+		if (ret)
+			dev_err(psp->adev->dev,
+				"XGMI: Failed to initialize XGMI session\n");
+	}
+
+	if (psp->ta_fw) {
+		ret = psp_ras_initialize(psp);
+		if (ret)
+			dev_err(psp->adev->dev,
+				"RAS: Failed to initialize RAS\n");
+
+		ret = psp_hdcp_initialize(psp);
+		if (ret)
+			dev_err(psp->adev->dev,
+				"HDCP: Failed to initialize HDCP\n");
+
+		ret = psp_dtm_initialize(psp);
+		if (ret)
+			dev_err(psp->adev->dev,
+				"DTM: Failed to initialize DTM\n");
+
+		ret = psp_rap_initialize(psp);
+		if (ret)
+			dev_err(psp->adev->dev,
+				"RAP: Failed to initialize RAP\n");
+
+		ret = psp_securedisplay_initialize(psp);
+		if (ret)
+			dev_err(psp->adev->dev,
+				"SECUREDISPLAY: Failed to initialize SECUREDISPLAY\n");
+	}
+
 	mutex_unlock(&adev->firmware.mutex);
 
 	return 0;
 
 failed:
-	DRM_ERROR("PSP resume failed\n");
+	dev_err(adev->dev, "PSP resume failed\n");
 	mutex_unlock(&adev->firmware.mutex);
 	return ret;
 }
 
-static bool psp_check_fw_loading_status(struct amdgpu_device *adev,
-					enum AMDGPU_UCODE_ID ucode_type)
+int psp_gpu_reset(struct amdgpu_device *adev)
 {
-	struct amdgpu_firmware_info *ucode = NULL;
+	int ret;
 
-	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
-		DRM_INFO("firmware is not loaded by PSP\n");
-		return true;
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
+		return 0;
+
+	mutex_lock(&adev->psp.mutex);
+	ret = psp_mode1_reset(&adev->psp);
+	mutex_unlock(&adev->psp.mutex);
+
+	return ret;
+}
+
+int psp_rlc_autoload_start(struct psp_context *psp)
+{
+	int ret;
+	struct psp_gfx_cmd_resp *cmd = acquire_psp_cmd_buf(psp);
+
+	cmd->cmd_id = GFX_CMD_ID_AUTOLOAD_RLC;
+
+	ret = psp_cmd_submit_buf(psp, NULL, cmd,
+				 psp->fence_buf_mc_addr);
+
+	release_psp_cmd_buf(psp);
+
+	return ret;
+}
+
+int psp_ring_cmd_submit(struct psp_context *psp,
+			uint64_t cmd_buf_mc_addr,
+			uint64_t fence_mc_addr,
+			int index)
+{
+	unsigned int psp_write_ptr_reg = 0;
+	struct psp_gfx_rb_frame *write_frame;
+	struct psp_ring *ring = &psp->km_ring;
+	struct psp_gfx_rb_frame *ring_buffer_start = ring->ring_mem;
+	struct psp_gfx_rb_frame *ring_buffer_end = ring_buffer_start +
+		ring->ring_size / sizeof(struct psp_gfx_rb_frame) - 1;
+	struct amdgpu_device *adev = psp->adev;
+	uint32_t ring_size_dw = ring->ring_size / 4;
+	uint32_t rb_frame_size_dw = sizeof(struct psp_gfx_rb_frame) / 4;
+
+	/* KM (GPCOM) prepare write pointer */
+	psp_write_ptr_reg = psp_ring_get_wptr(psp);
+
+	/* Update KM RB frame pointer to new frame */
+	/* write_frame ptr increments by size of rb_frame in bytes */
+	/* psp_write_ptr_reg increments by size of rb_frame in DWORDs */
+	if ((psp_write_ptr_reg % ring_size_dw) == 0)
+		write_frame = ring_buffer_start;
+	else
+		write_frame = ring_buffer_start + (psp_write_ptr_reg / rb_frame_size_dw);
+	/* Check invalid write_frame ptr address */
+	if ((write_frame < ring_buffer_start) || (ring_buffer_end < write_frame)) {
+		dev_err(adev->dev,
+			"ring_buffer_start = %p; ring_buffer_end = %p; write_frame = %p\n",
+			ring_buffer_start, ring_buffer_end, write_frame);
+		dev_err(adev->dev,
+			"write_frame is pointing to address out of bounds\n");
+		return -EINVAL;
 	}
 
-	if (!adev->firmware.fw_size)
-		return false;
+	/* Initialize KM RB frame */
+	memset(write_frame, 0, sizeof(struct psp_gfx_rb_frame));
 
-	ucode = &adev->firmware.ucode[ucode_type];
-	if (!ucode->fw || !ucode->ucode_size)
-		return false;
+	/* Update KM RB frame */
+	write_frame->cmd_buf_addr_hi = upper_32_bits(cmd_buf_mc_addr);
+	write_frame->cmd_buf_addr_lo = lower_32_bits(cmd_buf_mc_addr);
+	write_frame->fence_addr_hi = upper_32_bits(fence_mc_addr);
+	write_frame->fence_addr_lo = lower_32_bits(fence_mc_addr);
+	write_frame->fence_value = index;
+	amdgpu_device_flush_hdp(adev, NULL);
+
+	/* Update the write Pointer in DWORDs */
+	psp_write_ptr_reg = (psp_write_ptr_reg + rb_frame_size_dw) % ring_size_dw;
+	psp_ring_set_wptr(psp, psp_write_ptr_reg);
+	return 0;
+}
+
+int psp_init_asd_microcode(struct psp_context *psp, const char *chip_name)
+{
+	struct amdgpu_device *adev = psp->adev;
+	const struct psp_firmware_header_v1_0 *asd_hdr;
+	int err = 0;
 
-	return psp_compare_sram_data(&adev->psp, ucode, ucode_type);
+	err = amdgpu_ucode_request(adev, &adev->psp.asd_fw, AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_asd.bin", chip_name);
+	if (err)
+		goto out;
+
+	asd_hdr = (const struct psp_firmware_header_v1_0 *)adev->psp.asd_fw->data;
+	adev->psp.asd_context.bin_desc.fw_version = le32_to_cpu(asd_hdr->header.ucode_version);
+	adev->psp.asd_context.bin_desc.feature_version = le32_to_cpu(asd_hdr->sos.fw_version);
+	adev->psp.asd_context.bin_desc.size_bytes = le32_to_cpu(asd_hdr->header.ucode_size_bytes);
+	adev->psp.asd_context.bin_desc.start_addr = (uint8_t *)asd_hdr +
+				le32_to_cpu(asd_hdr->header.ucode_array_offset_bytes);
+	return 0;
+out:
+	amdgpu_ucode_release(&adev->psp.asd_fw);
+	return err;
 }
 
-static int psp_set_clockgating_state(void *handle,
-				     enum amd_clockgating_state state)
+int psp_init_toc_microcode(struct psp_context *psp, const char *chip_name)
 {
+	struct amdgpu_device *adev = psp->adev;
+	const struct psp_firmware_header_v1_0 *toc_hdr;
+	int err = 0;
+
+	err = amdgpu_ucode_request(adev, &adev->psp.toc_fw, AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_toc.bin", chip_name);
+	if (err)
+		goto out;
+
+	toc_hdr = (const struct psp_firmware_header_v1_0 *)adev->psp.toc_fw->data;
+	adev->psp.toc.fw_version = le32_to_cpu(toc_hdr->header.ucode_version);
+	adev->psp.toc.feature_version = le32_to_cpu(toc_hdr->sos.fw_version);
+	adev->psp.toc.size_bytes = le32_to_cpu(toc_hdr->header.ucode_size_bytes);
+	adev->psp.toc.start_addr = (uint8_t *)toc_hdr +
+				le32_to_cpu(toc_hdr->header.ucode_array_offset_bytes);
 	return 0;
+out:
+	amdgpu_ucode_release(&adev->psp.toc_fw);
+	return err;
 }
 
-static int psp_set_powergating_state(void *handle,
+static int parse_sos_bin_descriptor(struct psp_context *psp,
+				   const struct psp_fw_bin_desc *desc,
+				   const struct psp_firmware_header_v2_0 *sos_hdr)
+{
+	uint8_t *ucode_start_addr  = NULL;
+
+	if (!psp || !desc || !sos_hdr)
+		return -EINVAL;
+
+	ucode_start_addr  = (uint8_t *)sos_hdr +
+			    le32_to_cpu(desc->offset_bytes) +
+			    le32_to_cpu(sos_hdr->header.ucode_array_offset_bytes);
+
+	switch (desc->fw_type) {
+	case PSP_FW_TYPE_PSP_SOS:
+		psp->sos.fw_version        = le32_to_cpu(desc->fw_version);
+		psp->sos.feature_version   = le32_to_cpu(desc->fw_version);
+		psp->sos.size_bytes        = le32_to_cpu(desc->size_bytes);
+		psp->sos.start_addr	   = ucode_start_addr;
+		break;
+	case PSP_FW_TYPE_PSP_SYS_DRV:
+		psp->sys.fw_version        = le32_to_cpu(desc->fw_version);
+		psp->sys.feature_version   = le32_to_cpu(desc->fw_version);
+		psp->sys.size_bytes        = le32_to_cpu(desc->size_bytes);
+		psp->sys.start_addr        = ucode_start_addr;
+		break;
+	case PSP_FW_TYPE_PSP_KDB:
+		psp->kdb.fw_version        = le32_to_cpu(desc->fw_version);
+		psp->kdb.feature_version   = le32_to_cpu(desc->fw_version);
+		psp->kdb.size_bytes        = le32_to_cpu(desc->size_bytes);
+		psp->kdb.start_addr        = ucode_start_addr;
+		break;
+	case PSP_FW_TYPE_PSP_TOC:
+		psp->toc.fw_version        = le32_to_cpu(desc->fw_version);
+		psp->toc.feature_version   = le32_to_cpu(desc->fw_version);
+		psp->toc.size_bytes        = le32_to_cpu(desc->size_bytes);
+		psp->toc.start_addr        = ucode_start_addr;
+		break;
+	case PSP_FW_TYPE_PSP_SPL:
+		psp->spl.fw_version        = le32_to_cpu(desc->fw_version);
+		psp->spl.feature_version   = le32_to_cpu(desc->fw_version);
+		psp->spl.size_bytes        = le32_to_cpu(desc->size_bytes);
+		psp->spl.start_addr        = ucode_start_addr;
+		break;
+	case PSP_FW_TYPE_PSP_RL:
+		psp->rl.fw_version         = le32_to_cpu(desc->fw_version);
+		psp->rl.feature_version    = le32_to_cpu(desc->fw_version);
+		psp->rl.size_bytes         = le32_to_cpu(desc->size_bytes);
+		psp->rl.start_addr         = ucode_start_addr;
+		break;
+	case PSP_FW_TYPE_PSP_SOC_DRV:
+		psp->soc_drv.fw_version         = le32_to_cpu(desc->fw_version);
+		psp->soc_drv.feature_version    = le32_to_cpu(desc->fw_version);
+		psp->soc_drv.size_bytes         = le32_to_cpu(desc->size_bytes);
+		psp->soc_drv.start_addr         = ucode_start_addr;
+		break;
+	case PSP_FW_TYPE_PSP_INTF_DRV:
+		psp->intf_drv.fw_version        = le32_to_cpu(desc->fw_version);
+		psp->intf_drv.feature_version   = le32_to_cpu(desc->fw_version);
+		psp->intf_drv.size_bytes        = le32_to_cpu(desc->size_bytes);
+		psp->intf_drv.start_addr        = ucode_start_addr;
+		break;
+	case PSP_FW_TYPE_PSP_DBG_DRV:
+		psp->dbg_drv.fw_version         = le32_to_cpu(desc->fw_version);
+		psp->dbg_drv.feature_version    = le32_to_cpu(desc->fw_version);
+		psp->dbg_drv.size_bytes         = le32_to_cpu(desc->size_bytes);
+		psp->dbg_drv.start_addr         = ucode_start_addr;
+		break;
+	case PSP_FW_TYPE_PSP_RAS_DRV:
+		psp->ras_drv.fw_version         = le32_to_cpu(desc->fw_version);
+		psp->ras_drv.feature_version    = le32_to_cpu(desc->fw_version);
+		psp->ras_drv.size_bytes         = le32_to_cpu(desc->size_bytes);
+		psp->ras_drv.start_addr         = ucode_start_addr;
+		break;
+	case PSP_FW_TYPE_PSP_IPKEYMGR_DRV:
+		psp->ipkeymgr_drv.fw_version         = le32_to_cpu(desc->fw_version);
+		psp->ipkeymgr_drv.feature_version    = le32_to_cpu(desc->fw_version);
+		psp->ipkeymgr_drv.size_bytes         = le32_to_cpu(desc->size_bytes);
+		psp->ipkeymgr_drv.start_addr         = ucode_start_addr;
+		break;
+	case PSP_FW_TYPE_PSP_SPDM_DRV:
+		psp->spdm_drv.fw_version	= le32_to_cpu(desc->fw_version);
+		psp->spdm_drv.feature_version	= le32_to_cpu(desc->fw_version);
+		psp->spdm_drv.size_bytes	= le32_to_cpu(desc->size_bytes);
+		psp->spdm_drv.start_addr	= ucode_start_addr;
+		break;
+	default:
+		dev_warn(psp->adev->dev, "Unsupported PSP FW type: %d\n", desc->fw_type);
+		break;
+	}
+
+	return 0;
+}
+
+static int psp_init_sos_base_fw(struct amdgpu_device *adev)
+{
+	const struct psp_firmware_header_v1_0 *sos_hdr;
+	const struct psp_firmware_header_v1_3 *sos_hdr_v1_3;
+	uint8_t *ucode_array_start_addr;
+
+	sos_hdr = (const struct psp_firmware_header_v1_0 *)adev->psp.sos_fw->data;
+	ucode_array_start_addr = (uint8_t *)sos_hdr +
+		le32_to_cpu(sos_hdr->header.ucode_array_offset_bytes);
+
+	if (adev->gmc.xgmi.connected_to_cpu ||
+	    (amdgpu_ip_version(adev, MP0_HWIP, 0) != IP_VERSION(13, 0, 2))) {
+		adev->psp.sos.fw_version = le32_to_cpu(sos_hdr->header.ucode_version);
+		adev->psp.sos.feature_version = le32_to_cpu(sos_hdr->sos.fw_version);
+
+		adev->psp.sys.size_bytes = le32_to_cpu(sos_hdr->sos.offset_bytes);
+		adev->psp.sys.start_addr = ucode_array_start_addr;
+
+		adev->psp.sos.size_bytes = le32_to_cpu(sos_hdr->sos.size_bytes);
+		adev->psp.sos.start_addr = ucode_array_start_addr +
+				le32_to_cpu(sos_hdr->sos.offset_bytes);
+	} else {
+		/* Load alternate PSP SOS FW */
+		sos_hdr_v1_3 = (const struct psp_firmware_header_v1_3 *)adev->psp.sos_fw->data;
+
+		adev->psp.sos.fw_version = le32_to_cpu(sos_hdr_v1_3->sos_aux.fw_version);
+		adev->psp.sos.feature_version = le32_to_cpu(sos_hdr_v1_3->sos_aux.fw_version);
+
+		adev->psp.sys.size_bytes = le32_to_cpu(sos_hdr_v1_3->sys_drv_aux.size_bytes);
+		adev->psp.sys.start_addr = ucode_array_start_addr +
+			le32_to_cpu(sos_hdr_v1_3->sys_drv_aux.offset_bytes);
+
+		adev->psp.sos.size_bytes = le32_to_cpu(sos_hdr_v1_3->sos_aux.size_bytes);
+		adev->psp.sos.start_addr = ucode_array_start_addr +
+			le32_to_cpu(sos_hdr_v1_3->sos_aux.offset_bytes);
+	}
+
+	if ((adev->psp.sys.size_bytes == 0) || (adev->psp.sos.size_bytes == 0)) {
+		dev_warn(adev->dev, "PSP SOS FW not available");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+int psp_init_sos_microcode(struct psp_context *psp, const char *chip_name)
+{
+	struct amdgpu_device *adev = psp->adev;
+	const struct psp_firmware_header_v1_0 *sos_hdr;
+	const struct psp_firmware_header_v1_1 *sos_hdr_v1_1;
+	const struct psp_firmware_header_v1_2 *sos_hdr_v1_2;
+	const struct psp_firmware_header_v1_3 *sos_hdr_v1_3;
+	const struct psp_firmware_header_v2_0 *sos_hdr_v2_0;
+	const struct psp_firmware_header_v2_1 *sos_hdr_v2_1;
+	int fw_index, fw_bin_count, start_index = 0;
+	const struct psp_fw_bin_desc *fw_bin;
+	uint8_t *ucode_array_start_addr;
+	int err = 0;
+
+	if (amdgpu_is_kicker_fw(adev))
+		err = amdgpu_ucode_request(adev, &adev->psp.sos_fw, AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_sos_kicker.bin", chip_name);
+	else
+		err = amdgpu_ucode_request(adev, &adev->psp.sos_fw, AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_sos.bin", chip_name);
+	if (err)
+		goto out;
+
+	sos_hdr = (const struct psp_firmware_header_v1_0 *)adev->psp.sos_fw->data;
+	ucode_array_start_addr = (uint8_t *)sos_hdr +
+		le32_to_cpu(sos_hdr->header.ucode_array_offset_bytes);
+	amdgpu_ucode_print_psp_hdr(&sos_hdr->header);
+
+	switch (sos_hdr->header.header_version_major) {
+	case 1:
+		err = psp_init_sos_base_fw(adev);
+		if (err)
+			goto out;
+
+		if (sos_hdr->header.header_version_minor == 1) {
+			sos_hdr_v1_1 = (const struct psp_firmware_header_v1_1 *)adev->psp.sos_fw->data;
+			adev->psp.toc.size_bytes = le32_to_cpu(sos_hdr_v1_1->toc.size_bytes);
+			adev->psp.toc.start_addr = (uint8_t *)adev->psp.sys.start_addr +
+					le32_to_cpu(sos_hdr_v1_1->toc.offset_bytes);
+			adev->psp.kdb.size_bytes = le32_to_cpu(sos_hdr_v1_1->kdb.size_bytes);
+			adev->psp.kdb.start_addr = (uint8_t *)adev->psp.sys.start_addr +
+					le32_to_cpu(sos_hdr_v1_1->kdb.offset_bytes);
+		}
+		if (sos_hdr->header.header_version_minor == 2) {
+			sos_hdr_v1_2 = (const struct psp_firmware_header_v1_2 *)adev->psp.sos_fw->data;
+			adev->psp.kdb.size_bytes = le32_to_cpu(sos_hdr_v1_2->kdb.size_bytes);
+			adev->psp.kdb.start_addr = (uint8_t *)adev->psp.sys.start_addr +
+						    le32_to_cpu(sos_hdr_v1_2->kdb.offset_bytes);
+		}
+		if (sos_hdr->header.header_version_minor == 3) {
+			sos_hdr_v1_3 = (const struct psp_firmware_header_v1_3 *)adev->psp.sos_fw->data;
+			adev->psp.toc.size_bytes = le32_to_cpu(sos_hdr_v1_3->v1_1.toc.size_bytes);
+			adev->psp.toc.start_addr = ucode_array_start_addr +
+				le32_to_cpu(sos_hdr_v1_3->v1_1.toc.offset_bytes);
+			adev->psp.kdb.size_bytes = le32_to_cpu(sos_hdr_v1_3->v1_1.kdb.size_bytes);
+			adev->psp.kdb.start_addr = ucode_array_start_addr +
+				le32_to_cpu(sos_hdr_v1_3->v1_1.kdb.offset_bytes);
+			adev->psp.spl.size_bytes = le32_to_cpu(sos_hdr_v1_3->spl.size_bytes);
+			adev->psp.spl.start_addr = ucode_array_start_addr +
+				le32_to_cpu(sos_hdr_v1_3->spl.offset_bytes);
+			adev->psp.rl.size_bytes = le32_to_cpu(sos_hdr_v1_3->rl.size_bytes);
+			adev->psp.rl.start_addr = ucode_array_start_addr +
+				le32_to_cpu(sos_hdr_v1_3->rl.offset_bytes);
+		}
+		break;
+	case 2:
+		sos_hdr_v2_0 = (const struct psp_firmware_header_v2_0 *)adev->psp.sos_fw->data;
+
+		fw_bin_count = le32_to_cpu(sos_hdr_v2_0->psp_fw_bin_count);
+
+		if (fw_bin_count >= UCODE_MAX_PSP_PACKAGING) {
+			dev_err(adev->dev, "packed SOS count exceeds maximum limit\n");
+			err = -EINVAL;
+			goto out;
+		}
+
+		if (sos_hdr_v2_0->header.header_version_minor == 1) {
+			sos_hdr_v2_1 = (const struct psp_firmware_header_v2_1 *)adev->psp.sos_fw->data;
+
+			fw_bin = sos_hdr_v2_1->psp_fw_bin;
+
+			if (psp_is_aux_sos_load_required(psp))
+				start_index = le32_to_cpu(sos_hdr_v2_1->psp_aux_fw_bin_index);
+			else
+				fw_bin_count -= le32_to_cpu(sos_hdr_v2_1->psp_aux_fw_bin_index);
+
+		} else {
+			fw_bin = sos_hdr_v2_0->psp_fw_bin;
+		}
+
+		for (fw_index = start_index; fw_index < fw_bin_count; fw_index++) {
+			err = parse_sos_bin_descriptor(psp, fw_bin + fw_index,
+						       sos_hdr_v2_0);
+			if (err)
+				goto out;
+		}
+		break;
+	default:
+		dev_err(adev->dev,
+			"unsupported psp sos firmware\n");
+		err = -EINVAL;
+		goto out;
+	}
+
+	return 0;
+out:
+	amdgpu_ucode_release(&adev->psp.sos_fw);
+
+	return err;
+}
+
+static bool is_ta_fw_applicable(struct psp_context *psp,
+			     const struct psp_fw_bin_desc *desc)
+{
+	struct amdgpu_device *adev = psp->adev;
+	uint32_t fw_version;
+
+	switch (desc->fw_type) {
+	case TA_FW_TYPE_PSP_XGMI:
+	case TA_FW_TYPE_PSP_XGMI_AUX:
+		/* for now, AUX TA only exists on 13.0.6 ta bin,
+		 * from v20.00.0x.14
+		 */
+		if (amdgpu_ip_version(adev, MP0_HWIP, 0) ==
+		    IP_VERSION(13, 0, 6)) {
+			fw_version = le32_to_cpu(desc->fw_version);
+
+			if (adev->flags & AMD_IS_APU &&
+			    (fw_version & 0xff) >= 0x14)
+				return desc->fw_type == TA_FW_TYPE_PSP_XGMI_AUX;
+			else
+				return desc->fw_type == TA_FW_TYPE_PSP_XGMI;
+		}
+		break;
+	default:
+		break;
+	}
+
+	return true;
+}
+
+static int parse_ta_bin_descriptor(struct psp_context *psp,
+				   const struct psp_fw_bin_desc *desc,
+				   const struct ta_firmware_header_v2_0 *ta_hdr)
+{
+	uint8_t *ucode_start_addr  = NULL;
+
+	if (!psp || !desc || !ta_hdr)
+		return -EINVAL;
+
+	if (!is_ta_fw_applicable(psp, desc))
+		return 0;
+
+	ucode_start_addr  = (uint8_t *)ta_hdr +
+			    le32_to_cpu(desc->offset_bytes) +
+			    le32_to_cpu(ta_hdr->header.ucode_array_offset_bytes);
+
+	switch (desc->fw_type) {
+	case TA_FW_TYPE_PSP_ASD:
+		psp->asd_context.bin_desc.fw_version        = le32_to_cpu(desc->fw_version);
+		psp->asd_context.bin_desc.feature_version   = le32_to_cpu(desc->fw_version);
+		psp->asd_context.bin_desc.size_bytes        = le32_to_cpu(desc->size_bytes);
+		psp->asd_context.bin_desc.start_addr        = ucode_start_addr;
+		break;
+	case TA_FW_TYPE_PSP_XGMI:
+	case TA_FW_TYPE_PSP_XGMI_AUX:
+		psp->xgmi_context.context.bin_desc.fw_version       = le32_to_cpu(desc->fw_version);
+		psp->xgmi_context.context.bin_desc.size_bytes       = le32_to_cpu(desc->size_bytes);
+		psp->xgmi_context.context.bin_desc.start_addr       = ucode_start_addr;
+		break;
+	case TA_FW_TYPE_PSP_RAS:
+		psp->ras_context.context.bin_desc.fw_version        = le32_to_cpu(desc->fw_version);
+		psp->ras_context.context.bin_desc.size_bytes        = le32_to_cpu(desc->size_bytes);
+		psp->ras_context.context.bin_desc.start_addr        = ucode_start_addr;
+		break;
+	case TA_FW_TYPE_PSP_HDCP:
+		psp->hdcp_context.context.bin_desc.fw_version       = le32_to_cpu(desc->fw_version);
+		psp->hdcp_context.context.bin_desc.size_bytes       = le32_to_cpu(desc->size_bytes);
+		psp->hdcp_context.context.bin_desc.start_addr       = ucode_start_addr;
+		break;
+	case TA_FW_TYPE_PSP_DTM:
+		psp->dtm_context.context.bin_desc.fw_version       = le32_to_cpu(desc->fw_version);
+		psp->dtm_context.context.bin_desc.size_bytes       = le32_to_cpu(desc->size_bytes);
+		psp->dtm_context.context.bin_desc.start_addr       = ucode_start_addr;
+		break;
+	case TA_FW_TYPE_PSP_RAP:
+		psp->rap_context.context.bin_desc.fw_version       = le32_to_cpu(desc->fw_version);
+		psp->rap_context.context.bin_desc.size_bytes       = le32_to_cpu(desc->size_bytes);
+		psp->rap_context.context.bin_desc.start_addr       = ucode_start_addr;
+		break;
+	case TA_FW_TYPE_PSP_SECUREDISPLAY:
+		psp->securedisplay_context.context.bin_desc.fw_version =
+			le32_to_cpu(desc->fw_version);
+		psp->securedisplay_context.context.bin_desc.size_bytes =
+			le32_to_cpu(desc->size_bytes);
+		psp->securedisplay_context.context.bin_desc.start_addr =
+			ucode_start_addr;
+		break;
+	default:
+		dev_warn(psp->adev->dev, "Unsupported TA type: %d\n", desc->fw_type);
+		break;
+	}
+
+	return 0;
+}
+
+static int parse_ta_v1_microcode(struct psp_context *psp)
+{
+	const struct ta_firmware_header_v1_0 *ta_hdr;
+	struct amdgpu_device *adev = psp->adev;
+
+	ta_hdr = (const struct ta_firmware_header_v1_0 *) adev->psp.ta_fw->data;
+
+	if (le16_to_cpu(ta_hdr->header.header_version_major) != 1)
+		return -EINVAL;
+
+	adev->psp.xgmi_context.context.bin_desc.fw_version =
+		le32_to_cpu(ta_hdr->xgmi.fw_version);
+	adev->psp.xgmi_context.context.bin_desc.size_bytes =
+		le32_to_cpu(ta_hdr->xgmi.size_bytes);
+	adev->psp.xgmi_context.context.bin_desc.start_addr =
+		(uint8_t *)ta_hdr +
+		le32_to_cpu(ta_hdr->header.ucode_array_offset_bytes);
+
+	adev->psp.ras_context.context.bin_desc.fw_version =
+		le32_to_cpu(ta_hdr->ras.fw_version);
+	adev->psp.ras_context.context.bin_desc.size_bytes =
+		le32_to_cpu(ta_hdr->ras.size_bytes);
+	adev->psp.ras_context.context.bin_desc.start_addr =
+		(uint8_t *)adev->psp.xgmi_context.context.bin_desc.start_addr +
+		le32_to_cpu(ta_hdr->ras.offset_bytes);
+
+	adev->psp.hdcp_context.context.bin_desc.fw_version =
+		le32_to_cpu(ta_hdr->hdcp.fw_version);
+	adev->psp.hdcp_context.context.bin_desc.size_bytes =
+		le32_to_cpu(ta_hdr->hdcp.size_bytes);
+	adev->psp.hdcp_context.context.bin_desc.start_addr =
+		(uint8_t *)ta_hdr +
+		le32_to_cpu(ta_hdr->header.ucode_array_offset_bytes);
+
+	adev->psp.dtm_context.context.bin_desc.fw_version =
+		le32_to_cpu(ta_hdr->dtm.fw_version);
+	adev->psp.dtm_context.context.bin_desc.size_bytes =
+		le32_to_cpu(ta_hdr->dtm.size_bytes);
+	adev->psp.dtm_context.context.bin_desc.start_addr =
+		(uint8_t *)adev->psp.hdcp_context.context.bin_desc.start_addr +
+		le32_to_cpu(ta_hdr->dtm.offset_bytes);
+
+	adev->psp.securedisplay_context.context.bin_desc.fw_version =
+		le32_to_cpu(ta_hdr->securedisplay.fw_version);
+	adev->psp.securedisplay_context.context.bin_desc.size_bytes =
+		le32_to_cpu(ta_hdr->securedisplay.size_bytes);
+	adev->psp.securedisplay_context.context.bin_desc.start_addr =
+		(uint8_t *)adev->psp.hdcp_context.context.bin_desc.start_addr +
+		le32_to_cpu(ta_hdr->securedisplay.offset_bytes);
+
+	adev->psp.ta_fw_version = le32_to_cpu(ta_hdr->header.ucode_version);
+
+	return 0;
+}
+
+static int parse_ta_v2_microcode(struct psp_context *psp)
+{
+	const struct ta_firmware_header_v2_0 *ta_hdr;
+	struct amdgpu_device *adev = psp->adev;
+	int err = 0;
+	int ta_index = 0;
+
+	ta_hdr = (const struct ta_firmware_header_v2_0 *)adev->psp.ta_fw->data;
+
+	if (le16_to_cpu(ta_hdr->header.header_version_major) != 2)
+		return -EINVAL;
+
+	if (le32_to_cpu(ta_hdr->ta_fw_bin_count) >= UCODE_MAX_PSP_PACKAGING) {
+		dev_err(adev->dev, "packed TA count exceeds maximum limit\n");
+		return -EINVAL;
+	}
+
+	for (ta_index = 0; ta_index < le32_to_cpu(ta_hdr->ta_fw_bin_count); ta_index++) {
+		err = parse_ta_bin_descriptor(psp,
+					      &ta_hdr->ta_fw_bin[ta_index],
+					      ta_hdr);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+int psp_init_ta_microcode(struct psp_context *psp, const char *chip_name)
+{
+	const struct common_firmware_header *hdr;
+	struct amdgpu_device *adev = psp->adev;
+	int err;
+
+	if (amdgpu_is_kicker_fw(adev))
+		err = amdgpu_ucode_request(adev, &adev->psp.ta_fw, AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_ta_kicker.bin", chip_name);
+	else
+		err = amdgpu_ucode_request(adev, &adev->psp.ta_fw, AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_ta.bin", chip_name);
+	if (err)
+		return err;
+
+	hdr = (const struct common_firmware_header *)adev->psp.ta_fw->data;
+	switch (le16_to_cpu(hdr->header_version_major)) {
+	case 1:
+		err = parse_ta_v1_microcode(psp);
+		break;
+	case 2:
+		err = parse_ta_v2_microcode(psp);
+		break;
+	default:
+		dev_err(adev->dev, "unsupported TA header version\n");
+		err = -EINVAL;
+	}
+
+	if (err)
+		amdgpu_ucode_release(&adev->psp.ta_fw);
+
+	return err;
+}
+
+int psp_init_cap_microcode(struct psp_context *psp, const char *chip_name)
+{
+	struct amdgpu_device *adev = psp->adev;
+	const struct psp_firmware_header_v1_0 *cap_hdr_v1_0;
+	struct amdgpu_firmware_info *info = NULL;
+	int err = 0;
+
+	if (!amdgpu_sriov_vf(adev)) {
+		dev_err(adev->dev, "cap microcode should only be loaded under SRIOV\n");
+		return -EINVAL;
+	}
+
+	err = amdgpu_ucode_request(adev, &adev->psp.cap_fw, AMDGPU_UCODE_OPTIONAL,
+				   "amdgpu/%s_cap.bin", chip_name);
+	if (err) {
+		if (err == -ENODEV) {
+			dev_warn(adev->dev, "cap microcode does not exist, skip\n");
+			err = 0;
+		} else {
+			dev_err(adev->dev, "fail to initialize cap microcode\n");
+		}
+		goto out;
+	}
+
+	info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CAP];
+	info->ucode_id = AMDGPU_UCODE_ID_CAP;
+	info->fw = adev->psp.cap_fw;
+	cap_hdr_v1_0 = (const struct psp_firmware_header_v1_0 *)
+		adev->psp.cap_fw->data;
+	adev->firmware.fw_size += ALIGN(
+			le32_to_cpu(cap_hdr_v1_0->header.ucode_size_bytes), PAGE_SIZE);
+	adev->psp.cap_fw_version = le32_to_cpu(cap_hdr_v1_0->header.ucode_version);
+	adev->psp.cap_feature_version = le32_to_cpu(cap_hdr_v1_0->sos.fw_version);
+	adev->psp.cap_ucode_size = le32_to_cpu(cap_hdr_v1_0->header.ucode_size_bytes);
+
+	return 0;
+
+out:
+	amdgpu_ucode_release(&adev->psp.cap_fw);
+	return err;
+}
+
+int psp_config_sq_perfmon(struct psp_context *psp,
+		uint32_t xcp_id, bool core_override_enable,
+		bool reg_override_enable, bool perfmon_override_enable)
+{
+	int ret;
+
+	if (amdgpu_sriov_vf(psp->adev))
+		return 0;
+
+	if (xcp_id > MAX_XCP) {
+		dev_err(psp->adev->dev, "invalid xcp_id %d\n", xcp_id);
+		return -EINVAL;
+	}
+
+	if (amdgpu_ip_version(psp->adev, MP0_HWIP, 0) != IP_VERSION(13, 0, 6)) {
+		dev_err(psp->adev->dev, "Unsupported MP0 version 0x%x for CONFIG_SQ_PERFMON command\n",
+			amdgpu_ip_version(psp->adev, MP0_HWIP, 0));
+		return -EINVAL;
+	}
+	struct psp_gfx_cmd_resp *cmd = acquire_psp_cmd_buf(psp);
+
+	cmd->cmd_id	=	GFX_CMD_ID_CONFIG_SQ_PERFMON;
+	cmd->cmd.config_sq_perfmon.gfx_xcp_mask	=	BIT_MASK(xcp_id);
+	cmd->cmd.config_sq_perfmon.core_override	=	core_override_enable;
+	cmd->cmd.config_sq_perfmon.reg_override	=	reg_override_enable;
+	cmd->cmd.config_sq_perfmon.perfmon_override = perfmon_override_enable;
+
+	ret = psp_cmd_submit_buf(psp, NULL, cmd, psp->fence_buf_mc_addr);
+	if (ret)
+		dev_warn(psp->adev->dev, "PSP failed to config sq: xcp%d core%d reg%d perfmon%d\n",
+			xcp_id, core_override_enable, reg_override_enable, perfmon_override_enable);
+
+	release_psp_cmd_buf(psp);
+	return ret;
+}
+
+static int psp_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int psp_set_powergating_state(struct amdgpu_ip_block *ip_block,
 				     enum amd_powergating_state state)
 {
 	return 0;
 }
 
+static ssize_t psp_usbc_pd_fw_sysfs_read(struct device *dev,
+					 struct device_attribute *attr,
+					 char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	struct amdgpu_ip_block *ip_block;
+	uint32_t fw_ver;
+	int ret;
+
+	ip_block = amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_PSP);
+	if (!ip_block || !ip_block->status.late_initialized) {
+		dev_info(adev->dev, "PSP block is not ready yet\n.");
+		return -EBUSY;
+	}
+
+	mutex_lock(&adev->psp.mutex);
+	ret = psp_read_usbc_pd_fw(&adev->psp, &fw_ver);
+	mutex_unlock(&adev->psp.mutex);
+
+	if (ret) {
+		dev_err(adev->dev, "Failed to read USBC PD FW, err = %d\n", ret);
+		return ret;
+	}
+
+	return sysfs_emit(buf, "%x\n", fw_ver);
+}
+
+static ssize_t psp_usbc_pd_fw_sysfs_write(struct device *dev,
+						       struct device_attribute *attr,
+						       const char *buf,
+						       size_t count)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	int ret, idx;
+	const struct firmware *usbc_pd_fw;
+	struct amdgpu_bo *fw_buf_bo = NULL;
+	uint64_t fw_pri_mc_addr;
+	void *fw_pri_cpu_addr;
+	struct amdgpu_ip_block *ip_block;
+
+	ip_block = amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_PSP);
+	if (!ip_block || !ip_block->status.late_initialized) {
+		dev_err(adev->dev, "PSP block is not ready yet.");
+		return -EBUSY;
+	}
+
+	if (!drm_dev_enter(ddev, &idx))
+		return -ENODEV;
+
+	ret = amdgpu_ucode_request(adev, &usbc_pd_fw, AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s", buf);
+	if (ret)
+		goto fail;
+
+	/* LFB address which is aligned to 1MB boundary per PSP request */
+	ret = amdgpu_bo_create_kernel(adev, usbc_pd_fw->size, 0x100000,
+				      AMDGPU_GEM_DOMAIN_VRAM |
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &fw_buf_bo, &fw_pri_mc_addr,
+				      &fw_pri_cpu_addr);
+	if (ret)
+		goto rel_buf;
+
+	memcpy_toio(fw_pri_cpu_addr, usbc_pd_fw->data, usbc_pd_fw->size);
+
+	mutex_lock(&adev->psp.mutex);
+	ret = psp_load_usbc_pd_fw(&adev->psp, fw_pri_mc_addr);
+	mutex_unlock(&adev->psp.mutex);
+
+	amdgpu_bo_free_kernel(&fw_buf_bo, &fw_pri_mc_addr, &fw_pri_cpu_addr);
+
+rel_buf:
+	amdgpu_ucode_release(&usbc_pd_fw);
+fail:
+	if (ret) {
+		dev_err(adev->dev, "Failed to load USBC PD FW, err = %d", ret);
+		count = ret;
+	}
+
+	drm_dev_exit(idx);
+	return count;
+}
+
+void psp_copy_fw(struct psp_context *psp, uint8_t *start_addr, uint32_t bin_size)
+{
+	int idx;
+
+	if (!drm_dev_enter(adev_to_drm(psp->adev), &idx))
+		return;
+
+	memset(psp->fw_pri_buf, 0, PSP_1_MEG);
+	memcpy(psp->fw_pri_buf, start_addr, bin_size);
+
+	drm_dev_exit(idx);
+}
+
+/**
+ * DOC: usbc_pd_fw
+ * Reading from this file will retrieve the USB-C PD firmware version. Writing to
+ * this file will trigger the update process.
+ */
+static DEVICE_ATTR(usbc_pd_fw, 0644,
+		   psp_usbc_pd_fw_sysfs_read,
+		   psp_usbc_pd_fw_sysfs_write);
+
+int is_psp_fw_valid(struct psp_bin_desc bin)
+{
+	return bin.size_bytes;
+}
+
+static ssize_t amdgpu_psp_vbflash_write(struct file *filp, struct kobject *kobj,
+					const struct bin_attribute *bin_attr,
+					char *buffer, loff_t pos, size_t count)
+{
+	struct device *dev = kobj_to_dev(kobj);
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	adev->psp.vbflash_done = false;
+
+	/* Safeguard against memory drain */
+	if (adev->psp.vbflash_image_size > AMD_VBIOS_FILE_MAX_SIZE_B) {
+		dev_err(adev->dev, "File size cannot exceed %u\n", AMD_VBIOS_FILE_MAX_SIZE_B);
+		kvfree(adev->psp.vbflash_tmp_buf);
+		adev->psp.vbflash_tmp_buf = NULL;
+		adev->psp.vbflash_image_size = 0;
+		return -ENOMEM;
+	}
+
+	/* TODO Just allocate max for now and optimize to realloc later if needed */
+	if (!adev->psp.vbflash_tmp_buf) {
+		adev->psp.vbflash_tmp_buf = kvmalloc(AMD_VBIOS_FILE_MAX_SIZE_B, GFP_KERNEL);
+		if (!adev->psp.vbflash_tmp_buf)
+			return -ENOMEM;
+	}
+
+	mutex_lock(&adev->psp.mutex);
+	memcpy(adev->psp.vbflash_tmp_buf + pos, buffer, count);
+	adev->psp.vbflash_image_size += count;
+	mutex_unlock(&adev->psp.mutex);
+
+	dev_dbg(adev->dev, "IFWI staged for update\n");
+
+	return count;
+}
+
+static ssize_t amdgpu_psp_vbflash_read(struct file *filp, struct kobject *kobj,
+				       const struct bin_attribute *bin_attr, char *buffer,
+				       loff_t pos, size_t count)
+{
+	struct device *dev = kobj_to_dev(kobj);
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	struct amdgpu_bo *fw_buf_bo = NULL;
+	uint64_t fw_pri_mc_addr;
+	void *fw_pri_cpu_addr;
+	int ret;
+
+	if (adev->psp.vbflash_image_size == 0)
+		return -EINVAL;
+
+	dev_dbg(adev->dev, "PSP IFWI flash process initiated\n");
+
+	ret = amdgpu_bo_create_kernel(adev, adev->psp.vbflash_image_size,
+					AMDGPU_GPU_PAGE_SIZE,
+					AMDGPU_GEM_DOMAIN_VRAM,
+					&fw_buf_bo,
+					&fw_pri_mc_addr,
+					&fw_pri_cpu_addr);
+	if (ret)
+		goto rel_buf;
+
+	memcpy_toio(fw_pri_cpu_addr, adev->psp.vbflash_tmp_buf, adev->psp.vbflash_image_size);
+
+	mutex_lock(&adev->psp.mutex);
+	ret = psp_update_spirom(&adev->psp, fw_pri_mc_addr);
+	mutex_unlock(&adev->psp.mutex);
+
+	amdgpu_bo_free_kernel(&fw_buf_bo, &fw_pri_mc_addr, &fw_pri_cpu_addr);
+
+rel_buf:
+	kvfree(adev->psp.vbflash_tmp_buf);
+	adev->psp.vbflash_tmp_buf = NULL;
+	adev->psp.vbflash_image_size = 0;
+
+	if (ret) {
+		dev_err(adev->dev, "Failed to load IFWI, err = %d\n", ret);
+		return ret;
+	}
+
+	dev_dbg(adev->dev, "PSP IFWI flash process done\n");
+	return 0;
+}
+
+/**
+ * DOC: psp_vbflash
+ * Writing to this file will stage an IFWI for update. Reading from this file
+ * will trigger the update process.
+ */
+static const struct bin_attribute psp_vbflash_bin_attr = {
+	.attr = {.name = "psp_vbflash", .mode = 0660},
+	.size = 0,
+	.write = amdgpu_psp_vbflash_write,
+	.read = amdgpu_psp_vbflash_read,
+};
+
+/**
+ * DOC: psp_vbflash_status
+ * The status of the flash process.
+ * 0: IFWI flash not complete.
+ * 1: IFWI flash complete.
+ */
+static ssize_t amdgpu_psp_vbflash_status(struct device *dev,
+					 struct device_attribute *attr,
+					 char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	uint32_t vbflash_status;
+
+	vbflash_status = psp_vbflash_status(&adev->psp);
+	if (!adev->psp.vbflash_done)
+		vbflash_status = 0;
+	else if (adev->psp.vbflash_done && !(vbflash_status & 0x80000000))
+		vbflash_status = 1;
+
+	return sysfs_emit(buf, "0x%x\n", vbflash_status);
+}
+static DEVICE_ATTR(psp_vbflash_status, 0440, amdgpu_psp_vbflash_status, NULL);
+
+static const struct bin_attribute *const bin_flash_attrs[] = {
+	&psp_vbflash_bin_attr,
+	NULL
+};
+
+static struct attribute *flash_attrs[] = {
+	&dev_attr_psp_vbflash_status.attr,
+	&dev_attr_usbc_pd_fw.attr,
+	NULL
+};
+
+static umode_t amdgpu_flash_attr_is_visible(struct kobject *kobj, struct attribute *attr, int idx)
+{
+	struct device *dev = kobj_to_dev(kobj);
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	if (attr == &dev_attr_usbc_pd_fw.attr)
+		return adev->psp.sup_pd_fw_up ? 0660 : 0;
+
+	return adev->psp.sup_ifwi_up ? 0440 : 0;
+}
+
+static umode_t amdgpu_bin_flash_attr_is_visible(struct kobject *kobj,
+						const struct bin_attribute *attr,
+						int idx)
+{
+	struct device *dev = kobj_to_dev(kobj);
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	return adev->psp.sup_ifwi_up ? 0660 : 0;
+}
+
+const struct attribute_group amdgpu_flash_attr_group = {
+	.attrs = flash_attrs,
+	.bin_attrs = bin_flash_attrs,
+	.is_bin_visible = amdgpu_bin_flash_attr_is_visible,
+	.is_visible = amdgpu_flash_attr_is_visible,
+};
+
+#if defined(CONFIG_DEBUG_FS)
+static int psp_read_spirom_debugfs_open(struct inode *inode, struct file *filp)
+{
+	struct amdgpu_device *adev = filp->f_inode->i_private;
+	struct spirom_bo *bo_triplet;
+	int ret;
+
+	/* serialize the open() file calling */
+	if (!mutex_trylock(&adev->psp.mutex))
+		return -EBUSY;
+
+	/*
+	 * make sure only one userpace process is alive for dumping so that
+	 * only one memory buffer of AMD_VBIOS_FILE_MAX_SIZE * 2 is consumed.
+	 * let's say the case where one process try opening the file while
+	 * another one has proceeded to read or release. In this way, eliminate
+	 * the use of mutex for read() or release() callback as well.
+	 */
+	if (adev->psp.spirom_dump_trip) {
+		mutex_unlock(&adev->psp.mutex);
+		return -EBUSY;
+	}
+
+	bo_triplet = kzalloc(sizeof(struct spirom_bo), GFP_KERNEL);
+	if (!bo_triplet) {
+		mutex_unlock(&adev->psp.mutex);
+		return -ENOMEM;
+	}
+
+	ret = amdgpu_bo_create_kernel(adev, AMD_VBIOS_FILE_MAX_SIZE_B * 2,
+				      AMDGPU_GPU_PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &bo_triplet->bo,
+				      &bo_triplet->mc_addr,
+				      &bo_triplet->cpu_addr);
+	if (ret)
+		goto rel_trip;
+
+	ret = psp_dump_spirom(&adev->psp, bo_triplet->mc_addr);
+	if (ret)
+		goto rel_bo;
+
+	adev->psp.spirom_dump_trip = bo_triplet;
+	mutex_unlock(&adev->psp.mutex);
+	return 0;
+rel_bo:
+	amdgpu_bo_free_kernel(&bo_triplet->bo, &bo_triplet->mc_addr,
+			      &bo_triplet->cpu_addr);
+rel_trip:
+	kfree(bo_triplet);
+	mutex_unlock(&adev->psp.mutex);
+	dev_err(adev->dev, "Trying IFWI dump fails, err = %d\n", ret);
+	return ret;
+}
+
+static ssize_t psp_read_spirom_debugfs_read(struct file *filp, char __user *buf, size_t size,
+					    loff_t *pos)
+{
+	struct amdgpu_device *adev = filp->f_inode->i_private;
+	struct spirom_bo *bo_triplet = adev->psp.spirom_dump_trip;
+
+	if (!bo_triplet)
+		return -EINVAL;
+
+	return simple_read_from_buffer(buf,
+				       size,
+				       pos, bo_triplet->cpu_addr,
+				       AMD_VBIOS_FILE_MAX_SIZE_B * 2);
+}
+
+static int psp_read_spirom_debugfs_release(struct inode *inode, struct file *filp)
+{
+	struct amdgpu_device *adev = filp->f_inode->i_private;
+	struct spirom_bo *bo_triplet = adev->psp.spirom_dump_trip;
+
+	if (bo_triplet) {
+		amdgpu_bo_free_kernel(&bo_triplet->bo, &bo_triplet->mc_addr,
+				      &bo_triplet->cpu_addr);
+		kfree(bo_triplet);
+	}
+
+	adev->psp.spirom_dump_trip = NULL;
+	return 0;
+}
+
+static const struct file_operations psp_dump_spirom_debugfs_ops = {
+	.owner = THIS_MODULE,
+	.open = psp_read_spirom_debugfs_open,
+	.read = psp_read_spirom_debugfs_read,
+	.release = psp_read_spirom_debugfs_release,
+	.llseek = default_llseek,
+};
+#endif
+
+void amdgpu_psp_debugfs_init(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	struct drm_minor *minor = adev_to_drm(adev)->primary;
+
+	debugfs_create_file_size("psp_spirom_dump", 0444, minor->debugfs_root,
+				 adev, &psp_dump_spirom_debugfs_ops, AMD_VBIOS_FILE_MAX_SIZE_B * 2);
+#endif
+}
+
 const struct amd_ip_funcs psp_ip_funcs = {
 	.name = "psp",
 	.early_init = psp_early_init,
-	.late_init = NULL,
 	.sw_init = psp_sw_init,
 	.sw_fini = psp_sw_fini,
 	.hw_init = psp_hw_init,
 	.hw_fini = psp_hw_fini,
 	.suspend = psp_suspend,
 	.resume = psp_resume,
-	.is_idle = NULL,
-	.wait_for_idle = NULL,
-	.soft_reset = NULL,
 	.set_clockgating_state = psp_set_clockgating_state,
 	.set_powergating_state = psp_set_powergating_state,
 };
 
-static const struct amdgpu_psp_funcs psp_funcs = {
-	.check_fw_loading_status = psp_check_fw_loading_status,
-};
-
-static void psp_set_funcs(struct amdgpu_device *adev)
-{
-	if (NULL == adev->firmware.funcs)
-		adev->firmware.funcs = &psp_funcs;
-}
-
-const struct amdgpu_ip_block_version psp_v3_1_ip_block =
-{
+const struct amdgpu_ip_block_version psp_v3_1_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_PSP,
 	.major = 3,
 	.minor = 1,
@@ -555,11 +4464,58 @@ const struct amdgpu_ip_block_version psp_v3_1_ip_block =
 	.funcs = &psp_ip_funcs,
 };
 
-const struct amdgpu_ip_block_version psp_v10_0_ip_block =
-{
+const struct amdgpu_ip_block_version psp_v10_0_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_PSP,
 	.major = 10,
 	.minor = 0,
 	.rev = 0,
 	.funcs = &psp_ip_funcs,
 };
+
+const struct amdgpu_ip_block_version psp_v11_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_PSP,
+	.major = 11,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &psp_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version psp_v11_0_8_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_PSP,
+	.major = 11,
+	.minor = 0,
+	.rev = 8,
+	.funcs = &psp_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version psp_v12_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_PSP,
+	.major = 12,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &psp_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version psp_v13_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_PSP,
+	.major = 13,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &psp_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version psp_v13_0_4_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_PSP,
+	.major = 13,
+	.minor = 0,
+	.rev = 4,
+	.funcs = &psp_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version psp_v14_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_PSP,
+	.major = 14,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &psp_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
index 538fa9dbfb21..237b624aa51c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
@@ -27,14 +27,86 @@
 
 #include "amdgpu.h"
 #include "psp_gfx_if.h"
+#include "ta_xgmi_if.h"
+#include "ta_ras_if.h"
+#include "ta_rap_if.h"
+#include "ta_secureDisplay_if.h"
 
 #define PSP_FENCE_BUFFER_SIZE	0x1000
 #define PSP_CMD_BUFFER_SIZE	0x1000
-#define PSP_ASD_SHARED_MEM_SIZE	0x4000
 #define PSP_1_MEG		0x100000
+#define PSP_TMR_SIZE(adev)	((adev)->asic_type == CHIP_ALDEBARAN ? 0x800000 : 0x400000)
+#define PSP_TMR_ALIGNMENT	0x100000
+#define PSP_FW_NAME_LEN		0x24
 
-enum psp_ring_type
-{
+/* VBIOS gfl defines */
+#define MBOX_READY_MASK 0x80000000
+#define MBOX_STATUS_MASK 0x0000FFFF
+#define MBOX_COMMAND_MASK 0x00FF0000
+#define MBOX_READY_FLAG 0x80000000
+#define C2PMSG_CMD_SPI_UPDATE_ROM_IMAGE_ADDR_LO 0x2
+#define C2PMSG_CMD_SPI_UPDATE_ROM_IMAGE_ADDR_HI 0x3
+#define C2PMSG_CMD_SPI_UPDATE_FLASH_IMAGE 0x4
+#define C2PMSG_CMD_SPI_GET_ROM_IMAGE_ADDR_LO 0xf
+#define C2PMSG_CMD_SPI_GET_ROM_IMAGE_ADDR_HI 0x10
+#define C2PMSG_CMD_SPI_GET_FLASH_IMAGE 0x11
+
+/* Command register bit 31 set to indicate readiness */
+#define MBOX_TOS_READY_FLAG (GFX_FLAG_RESPONSE)
+#define MBOX_TOS_READY_MASK (GFX_CMD_RESPONSE_MASK | GFX_CMD_STATUS_MASK)
+
+/* Values to check for a successful GFX_CMD response wait. Check against
+ * both status bits and response state - helps to detect a command failure
+ * or other unexpected cases like a device drop reading all 0xFFs
+ */
+#define MBOX_TOS_RESP_FLAG (GFX_FLAG_RESPONSE)
+#define MBOX_TOS_RESP_MASK (GFX_CMD_RESPONSE_MASK | GFX_CMD_STATUS_MASK)
+
+extern const struct attribute_group amdgpu_flash_attr_group;
+
+enum psp_shared_mem_size {
+	PSP_ASD_SHARED_MEM_SIZE				= 0x0,
+	PSP_XGMI_SHARED_MEM_SIZE			= 0x4000,
+	PSP_RAS_SHARED_MEM_SIZE				= 0x4000,
+	PSP_HDCP_SHARED_MEM_SIZE			= 0x4000,
+	PSP_DTM_SHARED_MEM_SIZE				= 0x4000,
+	PSP_RAP_SHARED_MEM_SIZE				= 0x4000,
+	PSP_SECUREDISPLAY_SHARED_MEM_SIZE	= 0x4000,
+};
+
+enum ta_type_id {
+	TA_TYPE_XGMI = 1,
+	TA_TYPE_RAS,
+	TA_TYPE_HDCP,
+	TA_TYPE_DTM,
+	TA_TYPE_RAP,
+	TA_TYPE_SECUREDISPLAY,
+
+	TA_TYPE_MAX_INDEX,
+};
+
+struct psp_context;
+struct psp_xgmi_node_info;
+struct psp_xgmi_topology_info;
+struct psp_bin_desc;
+
+enum psp_bootloader_cmd {
+	PSP_BL__LOAD_SYSDRV		= 0x10000,
+	PSP_BL__LOAD_SOSDRV		= 0x20000,
+	PSP_BL__LOAD_KEY_DATABASE	= 0x80000,
+	PSP_BL__LOAD_SOCDRV             = 0xB0000,
+	PSP_BL__LOAD_DBGDRV             = 0xC0000,
+	PSP_BL__LOAD_HADDRV		= PSP_BL__LOAD_DBGDRV,
+	PSP_BL__LOAD_INTFDRV		= 0xD0000,
+	PSP_BL__LOAD_RASDRV		= 0xE0000,
+	PSP_BL__LOAD_IPKEYMGRDRV	= 0xF0000,
+	PSP_BL__DRAM_LONG_TRAIN		= 0x100000,
+	PSP_BL__DRAM_SHORT_TRAIN	= 0x200000,
+	PSP_BL__LOAD_TOS_SPL_TABLE	= 0x10000000,
+	PSP_BL__LOAD_SPDMDRV		= 0x20000000,
+};
+
+enum psp_ring_type {
 	PSP_RING_TYPE__INVALID = 0,
 	/*
 	 * These values map to the way the PSP kernel identifies the
@@ -44,75 +116,337 @@ enum psp_ring_type
 	PSP_RING_TYPE__KM = 2  /* Kernel mode ring (formerly called GPCOM) */
 };
 
-struct psp_ring
-{
+struct psp_ring {
 	enum psp_ring_type		ring_type;
 	struct psp_gfx_rb_frame		*ring_mem;
 	uint64_t			ring_mem_mc_addr;
 	void				*ring_mem_handle;
 	uint32_t			ring_size;
+	uint32_t			ring_wptr;
 };
 
-struct psp_context
-{
-	struct amdgpu_device            *adev;
-	struct psp_ring                 km_ring;
-	struct psp_gfx_cmd_resp		*cmd;
+/* More registers may will be supported */
+enum psp_reg_prog_id {
+	PSP_REG_IH_RB_CNTL        = 0,  /* register IH_RB_CNTL */
+	PSP_REG_IH_RB_CNTL_RING1  = 1,  /* register IH_RB_CNTL_RING1 */
+	PSP_REG_IH_RB_CNTL_RING2  = 2,  /* register IH_RB_CNTL_RING2 */
+	PSP_REG_MMHUB_L1_TLB_CNTL = 25,
+	PSP_REG_LAST
+};
+
+#define PSP_WAITREG_CHANGED BIT(0) /* check if the value has changed */
+#define PSP_WAITREG_NOVERBOSE BIT(1) /* No error verbose */
 
+struct psp_funcs {
 	int (*init_microcode)(struct psp_context *psp);
+	int (*wait_for_bootloader)(struct psp_context *psp);
+	int (*bootloader_load_kdb)(struct psp_context *psp);
+	int (*bootloader_load_spl)(struct psp_context *psp);
 	int (*bootloader_load_sysdrv)(struct psp_context *psp);
+	int (*bootloader_load_soc_drv)(struct psp_context *psp);
+	int (*bootloader_load_intf_drv)(struct psp_context *psp);
+	int (*bootloader_load_dbg_drv)(struct psp_context *psp);
+	int (*bootloader_load_ras_drv)(struct psp_context *psp);
+	int (*bootloader_load_ipkeymgr_drv)(struct psp_context *psp);
+	int (*bootloader_load_spdm_drv)(struct psp_context *psp);
 	int (*bootloader_load_sos)(struct psp_context *psp);
-	int (*prep_cmd_buf)(struct amdgpu_firmware_info *ucode,
-			    struct psp_gfx_cmd_resp *cmd);
-	int (*ring_init)(struct psp_context *psp, enum psp_ring_type ring_type);
-	int (*ring_create)(struct psp_context *psp, enum psp_ring_type ring_type);
+	int (*ring_create)(struct psp_context *psp,
+			   enum psp_ring_type ring_type);
+	int (*ring_stop)(struct psp_context *psp,
+			    enum psp_ring_type ring_type);
 	int (*ring_destroy)(struct psp_context *psp,
 			    enum psp_ring_type ring_type);
-	int (*cmd_submit)(struct psp_context *psp, struct amdgpu_firmware_info *ucode,
-			  uint64_t cmd_buf_mc_addr, uint64_t fence_mc_addr, int index);
-	bool (*compare_sram_data)(struct psp_context *psp,
-				  struct amdgpu_firmware_info *ucode,
-				  enum AMDGPU_UCODE_ID ucode_type);
 	bool (*smu_reload_quirk)(struct psp_context *psp);
+	int (*mode1_reset)(struct psp_context *psp);
+	int (*mem_training)(struct psp_context *psp, uint32_t ops);
+	uint32_t (*ring_get_wptr)(struct psp_context *psp);
+	void (*ring_set_wptr)(struct psp_context *psp, uint32_t value);
+	int (*load_usbc_pd_fw)(struct psp_context *psp, uint64_t fw_pri_mc_addr);
+	int (*read_usbc_pd_fw)(struct psp_context *psp, uint32_t *fw_ver);
+	int (*update_spirom)(struct psp_context *psp, uint64_t fw_pri_mc_addr);
+	int (*dump_spirom)(struct psp_context *psp, uint64_t fw_pri_mc_addr);
+	int (*vbflash_stat)(struct psp_context *psp);
+	int (*fatal_error_recovery_quirk)(struct psp_context *psp);
+	bool (*get_ras_capability)(struct psp_context *psp);
+	bool (*is_aux_sos_load_required)(struct psp_context *psp);
+	bool (*is_reload_needed)(struct psp_context *psp);
+	int (*reg_program_no_ring)(struct psp_context *psp, uint32_t val,
+				   enum psp_reg_prog_id id);
+};
 
-	/* fence buffer */
-	struct amdgpu_bo 		*fw_pri_bo;
-	uint64_t 			fw_pri_mc_addr;
+struct ta_funcs {
+	int (*fn_ta_initialize)(struct psp_context *psp);
+	int (*fn_ta_invoke)(struct psp_context *psp, uint32_t ta_cmd_id);
+	int (*fn_ta_terminate)(struct psp_context *psp);
+};
+
+#define AMDGPU_XGMI_MAX_CONNECTED_NODES		64
+struct psp_xgmi_node_info {
+	uint64_t				node_id;
+	uint8_t					num_hops;
+	uint8_t					is_sharing_enabled;
+	enum ta_xgmi_assigned_sdma_engine	sdma_engine;
+	uint8_t					num_links;
+	struct xgmi_connected_port_num		port_num[TA_XGMI__MAX_PORT_NUM];
+};
+
+struct psp_xgmi_topology_info {
+	uint32_t			num_nodes;
+	struct psp_xgmi_node_info	nodes[AMDGPU_XGMI_MAX_CONNECTED_NODES];
+};
+
+struct psp_bin_desc {
+	uint32_t fw_version;
+	uint32_t feature_version;
+	uint32_t size_bytes;
+	uint8_t *start_addr;
+};
+
+struct ta_mem_context {
+	struct amdgpu_bo		*shared_bo;
+	uint64_t		shared_mc_addr;
+	void			*shared_buf;
+	enum psp_shared_mem_size	shared_mem_size;
+};
+
+struct ta_context {
+	bool			initialized;
+	uint32_t		session_id;
+	uint32_t		resp_status;
+	struct ta_mem_context	mem_context;
+	struct psp_bin_desc		bin_desc;
+	enum psp_gfx_cmd_id		ta_load_type;
+	enum ta_type_id		ta_type;
+};
+
+struct ta_cp_context {
+	struct ta_context		context;
+	struct mutex			mutex;
+};
+
+struct psp_xgmi_context {
+	struct ta_context		context;
+	struct psp_xgmi_topology_info	top_info;
+	bool				supports_extended_data;
+	uint8_t				xgmi_ta_caps;
+};
+
+struct psp_ras_context {
+	struct ta_context		context;
+	struct amdgpu_ras		*ras;
+	struct mutex			mutex;
+};
+
+#define MEM_TRAIN_SYSTEM_SIGNATURE		0x54534942
+#define GDDR6_MEM_TRAINING_DATA_SIZE_IN_BYTES	0x1000
+#define GDDR6_MEM_TRAINING_OFFSET		0x8000
+/*Define the VRAM size that will be encroached by BIST training.*/
+#define BIST_MEM_TRAINING_ENCROACHED_SIZE	0x2000000
+
+enum psp_memory_training_init_flag {
+	PSP_MEM_TRAIN_NOT_SUPPORT	= 0x0,
+	PSP_MEM_TRAIN_SUPPORT		= 0x1,
+	PSP_MEM_TRAIN_INIT_FAILED	= 0x2,
+	PSP_MEM_TRAIN_RESERVE_SUCCESS	= 0x4,
+	PSP_MEM_TRAIN_INIT_SUCCESS	= 0x8,
+};
+
+enum psp_memory_training_ops {
+	PSP_MEM_TRAIN_SEND_LONG_MSG	= 0x1,
+	PSP_MEM_TRAIN_SAVE		= 0x2,
+	PSP_MEM_TRAIN_RESTORE		= 0x4,
+	PSP_MEM_TRAIN_SEND_SHORT_MSG	= 0x8,
+	PSP_MEM_TRAIN_COLD_BOOT		= PSP_MEM_TRAIN_SEND_LONG_MSG,
+	PSP_MEM_TRAIN_RESUME		= PSP_MEM_TRAIN_SEND_SHORT_MSG,
+};
+
+struct psp_memory_training_context {
+	/*training data size*/
+	u64 train_data_size;
+	/*
+	 * sys_cache
+	 * cpu virtual address
+	 * system memory buffer that used to store the training data.
+	 */
+	void *sys_cache;
+
+	/*vram offset of the p2c training data*/
+	u64 p2c_train_data_offset;
+
+	/*vram offset of the c2p training data*/
+	u64 c2p_train_data_offset;
+	struct amdgpu_bo *c2p_bo;
+
+	enum psp_memory_training_init_flag init;
+	u32 training_cnt;
+	bool enable_mem_training;
+};
+
+/** PSP runtime DB **/
+#define PSP_RUNTIME_DB_SIZE_IN_BYTES		0x10000
+#define PSP_RUNTIME_DB_OFFSET			0x100000
+#define PSP_RUNTIME_DB_COOKIE_ID		0x0ed5
+#define PSP_RUNTIME_DB_VER_1			0x0100
+#define PSP_RUNTIME_DB_DIAG_ENTRY_MAX_COUNT	0x40
+
+enum psp_runtime_entry_type {
+	PSP_RUNTIME_ENTRY_TYPE_INVALID		= 0x0,
+	PSP_RUNTIME_ENTRY_TYPE_TEST		= 0x1,
+	PSP_RUNTIME_ENTRY_TYPE_MGPU_COMMON	= 0x2,  /* Common mGPU runtime data */
+	PSP_RUNTIME_ENTRY_TYPE_MGPU_WAFL	= 0x3,  /* WAFL runtime data */
+	PSP_RUNTIME_ENTRY_TYPE_MGPU_XGMI	= 0x4,  /* XGMI runtime data */
+	PSP_RUNTIME_ENTRY_TYPE_BOOT_CONFIG	= 0x5,  /* Boot Config runtime data */
+	PSP_RUNTIME_ENTRY_TYPE_PPTABLE_ERR_STATUS = 0x6, /* SCPM validation data */
+};
+
+/* PSP runtime DB header */
+struct psp_runtime_data_header {
+	/* determine the existence of runtime db */
+	uint16_t cookie;
+	/* version of runtime db */
+	uint16_t version;
+};
+
+/* PSP runtime DB entry */
+struct psp_runtime_entry {
+	/* type of runtime db entry */
+	uint32_t entry_type;
+	/* offset of entry in bytes */
+	uint16_t offset;
+	/* size of entry in bytes */
+	uint16_t size;
+};
+
+/* PSP runtime DB directory */
+struct psp_runtime_data_directory {
+	/* number of valid entries */
+	uint16_t			entry_count;
+	/* db entries*/
+	struct psp_runtime_entry	entry_list[PSP_RUNTIME_DB_DIAG_ENTRY_MAX_COUNT];
+};
+
+/* PSP runtime DB boot config feature bitmask */
+enum psp_runtime_boot_cfg_feature {
+	BOOT_CFG_FEATURE_GECC                       = 0x1,
+	BOOT_CFG_FEATURE_TWO_STAGE_DRAM_TRAINING    = 0x2,
+};
+
+/* PSP run time DB SCPM authentication defines */
+enum psp_runtime_scpm_authentication {
+	SCPM_DISABLE                     = 0x0,
+	SCPM_ENABLE                      = 0x1,
+	SCPM_ENABLE_WITH_SCPM_ERR        = 0x2,
+};
+
+/* PSP runtime DB boot config entry */
+struct psp_runtime_boot_cfg_entry {
+	uint32_t boot_cfg_bitmask;
+	uint32_t reserved;
+};
+
+/* PSP runtime DB SCPM entry */
+struct psp_runtime_scpm_entry {
+	enum psp_runtime_scpm_authentication scpm_status;
+};
+
+#if defined(CONFIG_DEBUG_FS)
+struct spirom_bo {
+	struct amdgpu_bo *bo;
+	uint64_t mc_addr;
+	void *cpu_addr;
+};
+#endif
+
+struct psp_context {
+	struct amdgpu_device		*adev;
+	struct psp_ring			km_ring;
+	struct psp_gfx_cmd_resp		*cmd;
+
+	const struct psp_funcs		*funcs;
+	const struct ta_funcs		*ta_funcs;
+
+	/* firmware buffer */
+	struct amdgpu_bo		*fw_pri_bo;
+	uint64_t			fw_pri_mc_addr;
 	void				*fw_pri_buf;
 
 	/* sos firmware */
 	const struct firmware		*sos_fw;
-	uint32_t			sos_fw_version;
-	uint32_t			sos_feature_version;
-	uint32_t			sys_bin_size;
-	uint32_t			sos_bin_size;
-	uint8_t				*sys_start_addr;
-	uint8_t				*sos_start_addr;
+	struct psp_bin_desc		sys;
+	struct psp_bin_desc		sos;
+	struct psp_bin_desc		toc;
+	struct psp_bin_desc		kdb;
+	struct psp_bin_desc		spl;
+	struct psp_bin_desc		rl;
+	struct psp_bin_desc		soc_drv;
+	struct psp_bin_desc		intf_drv;
+	struct psp_bin_desc		dbg_drv;
+	struct psp_bin_desc		ras_drv;
+	struct psp_bin_desc		ipkeymgr_drv;
+	struct psp_bin_desc		spdm_drv;
 
 	/* tmr buffer */
-	struct amdgpu_bo 		*tmr_bo;
-	uint64_t 			tmr_mc_addr;
-	void				*tmr_buf;
+	struct amdgpu_bo		*tmr_bo;
+	uint64_t			tmr_mc_addr;
 
-	/* asd firmware and buffer */
+	/* asd firmware */
 	const struct firmware		*asd_fw;
-	uint32_t			asd_fw_version;
-	uint32_t			asd_feature_version;
-	uint32_t			asd_ucode_size;
-	uint8_t				*asd_start_addr;
-	struct amdgpu_bo 		*asd_shared_bo;
-	uint64_t 			asd_shared_mc_addr;
-	void				*asd_shared_buf;
+
+	/* toc firmware */
+	const struct firmware		*toc_fw;
+
+	/* cap firmware */
+	const struct firmware		*cap_fw;
 
 	/* fence buffer */
-	struct amdgpu_bo 		*fence_buf_bo;
-	uint64_t 			fence_buf_mc_addr;
+	struct amdgpu_bo		*fence_buf_bo;
+	uint64_t			fence_buf_mc_addr;
 	void				*fence_buf;
 
 	/* cmd buffer */
 	struct amdgpu_bo		*cmd_buf_bo;
 	uint64_t			cmd_buf_mc_addr;
 	struct psp_gfx_cmd_resp		*cmd_buf_mem;
+
+	/* fence value associated with cmd buffer */
+	atomic_t			fence_value;
+	/* flag to mark whether gfx fw autoload is supported or not */
+	bool				autoload_supported;
+	/* flag to mark whether psp use runtime TMR or boottime TMR */
+	bool				boot_time_tmr;
+	/* flag to mark whether df cstate management centralized to PMFW */
+	bool				pmfw_centralized_cstate_management;
+
+	/* xgmi ta firmware and buffer */
+	const struct firmware		*ta_fw;
+	uint32_t			ta_fw_version;
+
+	uint32_t			cap_fw_version;
+	uint32_t			cap_feature_version;
+	uint32_t			cap_ucode_size;
+
+	struct ta_context		asd_context;
+	struct psp_xgmi_context		xgmi_context;
+	struct psp_ras_context		ras_context;
+	struct ta_cp_context		hdcp_context;
+	struct ta_cp_context		dtm_context;
+	struct ta_cp_context		rap_context;
+	struct ta_cp_context		securedisplay_context;
+	struct mutex			mutex;
+	struct psp_memory_training_context mem_train_ctx;
+
+	uint32_t			boot_cfg_bitmask;
+
+	/* firmware upgrades supported */
+	bool				sup_pd_fw_up;
+	bool				sup_ifwi_up;
+
+	char				*vbflash_tmp_buf;
+	size_t				vbflash_image_size;
+	bool				vbflash_done;
+#if defined(CONFIG_DEBUG_FS)
+	struct spirom_bo *spirom_dump_trip;
+#endif
 };
 
 struct amdgpu_psp_funcs {
@@ -120,29 +454,173 @@ struct amdgpu_psp_funcs {
 					enum AMDGPU_UCODE_ID);
 };
 
-#define psp_prep_cmd_buf(ucode, type) (psp)->prep_cmd_buf((ucode), (type))
-#define psp_ring_init(psp, type) (psp)->ring_init((psp), (type))
-#define psp_ring_create(psp, type) (psp)->ring_create((psp), (type))
-#define psp_ring_destroy(psp, type) ((psp)->ring_destroy((psp), (type)))
-#define psp_cmd_submit(psp, ucode, cmd_mc, fence_mc, index) \
-		(psp)->cmd_submit((psp), (ucode), (cmd_mc), (fence_mc), (index))
-#define psp_compare_sram_data(psp, ucode, type) \
-		(psp)->compare_sram_data((psp), (ucode), (type))
+
+#define psp_ring_create(psp, type) (psp)->funcs->ring_create((psp), (type))
+#define psp_ring_stop(psp, type) (psp)->funcs->ring_stop((psp), (type))
+#define psp_ring_destroy(psp, type) ((psp)->funcs->ring_destroy((psp), (type)))
 #define psp_init_microcode(psp) \
-		((psp)->init_microcode ? (psp)->init_microcode((psp)) : 0)
+		((psp)->funcs->init_microcode ? (psp)->funcs->init_microcode((psp)) : 0)
+#define psp_bootloader_load_kdb(psp) \
+		((psp)->funcs->bootloader_load_kdb ? (psp)->funcs->bootloader_load_kdb((psp)) : 0)
+#define psp_bootloader_load_spl(psp) \
+		((psp)->funcs->bootloader_load_spl ? (psp)->funcs->bootloader_load_spl((psp)) : 0)
 #define psp_bootloader_load_sysdrv(psp) \
-		((psp)->bootloader_load_sysdrv ? (psp)->bootloader_load_sysdrv((psp)) : 0)
+		((psp)->funcs->bootloader_load_sysdrv ? (psp)->funcs->bootloader_load_sysdrv((psp)) : 0)
+#define psp_bootloader_load_soc_drv(psp) \
+		((psp)->funcs->bootloader_load_soc_drv ? (psp)->funcs->bootloader_load_soc_drv((psp)) : 0)
+#define psp_bootloader_load_intf_drv(psp) \
+		((psp)->funcs->bootloader_load_intf_drv ? (psp)->funcs->bootloader_load_intf_drv((psp)) : 0)
+#define psp_bootloader_load_dbg_drv(psp) \
+		((psp)->funcs->bootloader_load_dbg_drv ? (psp)->funcs->bootloader_load_dbg_drv((psp)) : 0)
+#define psp_bootloader_load_ras_drv(psp) \
+		((psp)->funcs->bootloader_load_ras_drv ? \
+		(psp)->funcs->bootloader_load_ras_drv((psp)) : 0)
+#define psp_bootloader_load_ipkeymgr_drv(psp) \
+		((psp)->funcs->bootloader_load_ipkeymgr_drv ? \
+		 (psp)->funcs->bootloader_load_ipkeymgr_drv((psp)) : 0)
+#define psp_bootloader_load_spdm_drv(psp) \
+		((psp)->funcs->bootloader_load_spdm_drv ? \
+		 (psp)->funcs->bootloader_load_spdm_drv((psp)) : 0)
 #define psp_bootloader_load_sos(psp) \
-		((psp)->bootloader_load_sos ? (psp)->bootloader_load_sos((psp)) : 0)
+		((psp)->funcs->bootloader_load_sos ? (psp)->funcs->bootloader_load_sos((psp)) : 0)
 #define psp_smu_reload_quirk(psp) \
-		((psp)->smu_reload_quirk ? (psp)->smu_reload_quirk((psp)) : false)
+		((psp)->funcs->smu_reload_quirk ? (psp)->funcs->smu_reload_quirk((psp)) : false)
+#define psp_mode1_reset(psp) \
+		((psp)->funcs->mode1_reset ? (psp)->funcs->mode1_reset((psp)) : false)
+#define psp_mem_training(psp, ops) \
+	((psp)->funcs->mem_training ? (psp)->funcs->mem_training((psp), (ops)) : 0)
+
+#define psp_ring_get_wptr(psp) (psp)->funcs->ring_get_wptr((psp))
+#define psp_ring_set_wptr(psp, value) (psp)->funcs->ring_set_wptr((psp), (value))
+
+#define psp_load_usbc_pd_fw(psp, fw_pri_mc_addr) \
+	((psp)->funcs->load_usbc_pd_fw ? \
+	(psp)->funcs->load_usbc_pd_fw((psp), (fw_pri_mc_addr)) : -EINVAL)
+
+#define psp_read_usbc_pd_fw(psp, fw_ver) \
+	((psp)->funcs->read_usbc_pd_fw ? \
+	(psp)->funcs->read_usbc_pd_fw((psp), fw_ver) : -EINVAL)
+
+#define psp_update_spirom(psp, fw_pri_mc_addr) \
+	((psp)->funcs->update_spirom ? \
+	(psp)->funcs->update_spirom((psp), fw_pri_mc_addr) : -EINVAL)
+
+#define psp_dump_spirom(psp, fw_pri_mc_addr) \
+	((psp)->funcs->dump_spirom ? \
+	(psp)->funcs->dump_spirom((psp), fw_pri_mc_addr) : -EINVAL)
+
+#define psp_vbflash_status(psp) \
+	((psp)->funcs->vbflash_stat ? \
+	(psp)->funcs->vbflash_stat((psp)) : -EINVAL)
+
+#define psp_fatal_error_recovery_quirk(psp) \
+	((psp)->funcs->fatal_error_recovery_quirk ? \
+	(psp)->funcs->fatal_error_recovery_quirk((psp)) : 0)
+
+#define psp_is_aux_sos_load_required(psp) \
+	((psp)->funcs->is_aux_sos_load_required ? (psp)->funcs->is_aux_sos_load_required((psp)) : 0)
+
+#define psp_reg_program_no_ring(psp, val, id) \
+	((psp)->funcs->reg_program_no_ring ? \
+	(psp)->funcs->reg_program_no_ring((psp), val, id) : -EINVAL)
 
 extern const struct amd_ip_funcs psp_ip_funcs;
 
 extern const struct amdgpu_ip_block_version psp_v3_1_ip_block;
-extern int psp_wait_for(struct psp_context *psp, uint32_t reg_index,
-			uint32_t field_val, uint32_t mask, bool check_changed);
-
 extern const struct amdgpu_ip_block_version psp_v10_0_ip_block;
+extern const struct amdgpu_ip_block_version psp_v11_0_ip_block;
+extern const struct amdgpu_ip_block_version psp_v11_0_8_ip_block;
+extern const struct amdgpu_ip_block_version psp_v12_0_ip_block;
+extern const struct amdgpu_ip_block_version psp_v13_0_ip_block;
+extern const struct amdgpu_ip_block_version psp_v13_0_4_ip_block;
+extern const struct amdgpu_ip_block_version psp_v14_0_ip_block;
+
+int psp_wait_for(struct psp_context *psp, uint32_t reg_index,
+		 uint32_t field_val, uint32_t mask, uint32_t flags);
+extern int psp_wait_for_spirom_update(struct psp_context *psp, uint32_t reg_index,
+			uint32_t field_val, uint32_t mask, uint32_t msec_timeout);
+
+int psp_execute_ip_fw_load(struct psp_context *psp,
+			   struct amdgpu_firmware_info *ucode);
+
+int psp_gpu_reset(struct amdgpu_device *adev);
+
+int psp_ta_init_shared_buf(struct psp_context *psp,
+				  struct ta_mem_context *mem_ctx);
+void psp_ta_free_shared_buf(struct ta_mem_context *mem_ctx);
+int psp_ta_unload(struct psp_context *psp, struct ta_context *context);
+int psp_ta_load(struct psp_context *psp, struct ta_context *context);
+int psp_ta_invoke(struct psp_context *psp,
+			uint32_t ta_cmd_id,
+			struct ta_context *context);
+
+int psp_xgmi_initialize(struct psp_context *psp, bool set_extended_data, bool load_ta);
+int psp_xgmi_terminate(struct psp_context *psp);
+int psp_xgmi_invoke(struct psp_context *psp, uint32_t ta_cmd_id);
+int psp_xgmi_get_hive_id(struct psp_context *psp, uint64_t *hive_id);
+int psp_xgmi_get_node_id(struct psp_context *psp, uint64_t *node_id);
+int psp_xgmi_get_topology_info(struct psp_context *psp,
+			       int number_devices,
+			       struct psp_xgmi_topology_info *topology,
+			       bool get_extended_data);
+int psp_xgmi_set_topology_info(struct psp_context *psp,
+			       int number_devices,
+			       struct psp_xgmi_topology_info *topology);
+int psp_ras_initialize(struct psp_context *psp);
+int psp_ras_invoke(struct psp_context *psp, uint32_t ta_cmd_id);
+int psp_ras_enable_features(struct psp_context *psp,
+		union ta_ras_cmd_input *info, bool enable);
+int psp_ras_trigger_error(struct psp_context *psp,
+			  struct ta_ras_trigger_error_input *info, uint32_t instance_mask);
+int psp_ras_terminate(struct psp_context *psp);
+int psp_ras_query_address(struct psp_context *psp,
+			  struct ta_ras_query_address_input *addr_in,
+			  struct ta_ras_query_address_output *addr_out);
+
+int psp_hdcp_invoke(struct psp_context *psp, uint32_t ta_cmd_id);
+int psp_dtm_invoke(struct psp_context *psp, uint32_t ta_cmd_id);
+int psp_rap_invoke(struct psp_context *psp, uint32_t ta_cmd_id, enum ta_rap_status *status);
+int psp_securedisplay_invoke(struct psp_context *psp, uint32_t ta_cmd_id);
+
+int psp_rlc_autoload_start(struct psp_context *psp);
+
+int psp_reg_program(struct psp_context *psp, enum psp_reg_prog_id reg,
+		uint32_t value);
+int psp_ring_cmd_submit(struct psp_context *psp,
+			uint64_t cmd_buf_mc_addr,
+			uint64_t fence_mc_addr,
+			int index);
+int psp_init_asd_microcode(struct psp_context *psp,
+			   const char *chip_name);
+int psp_init_toc_microcode(struct psp_context *psp,
+			   const char *chip_name);
+int psp_init_sos_microcode(struct psp_context *psp,
+			   const char *chip_name);
+int psp_init_ta_microcode(struct psp_context *psp,
+			  const char *chip_name);
+int psp_init_cap_microcode(struct psp_context *psp,
+			  const char *chip_name);
+int psp_get_fw_attestation_records_addr(struct psp_context *psp,
+					uint64_t *output_ptr);
+int psp_update_fw_reservation(struct psp_context *psp);
+int psp_load_fw_list(struct psp_context *psp,
+		     struct amdgpu_firmware_info **ucode_list, int ucode_count);
+void psp_copy_fw(struct psp_context *psp, uint8_t *start_addr, uint32_t bin_size);
+
+int psp_spatial_partition(struct psp_context *psp, int mode);
+int psp_memory_partition(struct psp_context *psp, int mode);
+
+int is_psp_fw_valid(struct psp_bin_desc bin);
+
+int amdgpu_psp_wait_for_bootloader(struct amdgpu_device *adev);
+bool amdgpu_psp_get_ras_capability(struct psp_context *psp);
+
+int psp_config_sq_perfmon(struct psp_context *psp, uint32_t xcp_id,
+	bool core_override_enable, bool reg_override_enable, bool perfmon_override_enable);
+bool amdgpu_psp_tos_reload_needed(struct amdgpu_device *adev);
+int amdgpu_psp_reg_program_no_ring(struct psp_context *psp, uint32_t val,
+				   enum psp_reg_prog_id id);
+void amdgpu_psp_debugfs_init(struct amdgpu_device *adev);
+
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.c
new file mode 100644
index 000000000000..6e8aad91bcd3
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.c
@@ -0,0 +1,392 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "amdgpu_psp_ta.h"
+
+#if defined(CONFIG_DEBUG_FS)
+
+static ssize_t ta_if_load_debugfs_write(struct file *fp, const char *buf,
+					    size_t len, loff_t *off);
+static ssize_t ta_if_unload_debugfs_write(struct file *fp, const char *buf,
+					    size_t len, loff_t *off);
+static ssize_t ta_if_invoke_debugfs_write(struct file *fp, const char *buf,
+					    size_t len, loff_t *off);
+
+static uint32_t get_bin_version(const uint8_t *bin)
+{
+	const struct common_firmware_header *hdr =
+			     (const struct common_firmware_header *)bin;
+
+	return hdr->ucode_version;
+}
+
+static int prep_ta_mem_context(struct ta_mem_context *mem_context,
+					     uint8_t *shared_buf,
+					     uint32_t shared_buf_len)
+{
+	if (mem_context->shared_mem_size < shared_buf_len)
+		return -EINVAL;
+	memset(mem_context->shared_buf, 0, mem_context->shared_mem_size);
+	memcpy((void *)mem_context->shared_buf, shared_buf, shared_buf_len);
+
+	return 0;
+}
+
+static bool is_ta_type_valid(enum ta_type_id ta_type)
+{
+	switch (ta_type) {
+	case TA_TYPE_RAS:
+		return true;
+	default:
+		return false;
+	}
+}
+
+static const struct ta_funcs ras_ta_funcs = {
+	.fn_ta_initialize = psp_ras_initialize,
+	.fn_ta_invoke    = psp_ras_invoke,
+	.fn_ta_terminate = psp_ras_terminate
+};
+
+static void set_ta_context_funcs(struct psp_context *psp,
+						      enum ta_type_id ta_type,
+						      struct ta_context **pcontext)
+{
+	switch (ta_type) {
+	case TA_TYPE_RAS:
+		*pcontext = &psp->ras_context.context;
+		psp->ta_funcs = &ras_ta_funcs;
+		break;
+	default:
+		break;
+	}
+}
+
+static const struct file_operations ta_load_debugfs_fops = {
+	.write  = ta_if_load_debugfs_write,
+	.llseek = default_llseek,
+	.owner  = THIS_MODULE
+};
+
+static const struct file_operations ta_unload_debugfs_fops = {
+	.write  = ta_if_unload_debugfs_write,
+	.llseek = default_llseek,
+	.owner  = THIS_MODULE
+};
+
+static const struct file_operations ta_invoke_debugfs_fops = {
+	.write  = ta_if_invoke_debugfs_write,
+	.llseek = default_llseek,
+	.owner  = THIS_MODULE
+};
+
+/*
+ * DOC: AMDGPU TA debugfs interfaces
+ *
+ * Three debugfs interfaces can be opened by a program to
+ * load/invoke/unload TA,
+ *
+ * - /sys/kernel/debug/dri/<N>/ta_if/ta_load
+ * - /sys/kernel/debug/dri/<N>/ta_if/ta_invoke
+ * - /sys/kernel/debug/dri/<N>/ta_if/ta_unload
+ *
+ * How to use the interfaces in a program?
+ *
+ * A program needs to provide transmit buffer to the interfaces
+ * and will receive buffer from the interfaces below,
+ *
+ * - For TA load debugfs interface:
+ *   Transmit buffer:
+ *    - TA type (4bytes)
+ *    - TA bin length (4bytes)
+ *    - TA bin
+ *   Receive buffer:
+ *    - TA ID (4bytes)
+ *
+ * - For TA invoke debugfs interface:
+ *   Transmit buffer:
+ *    - TA type (4bytes)
+ *    - TA ID (4bytes)
+ *    - TA CMD ID (4bytes)
+ *    - TA shard buf length
+ *      (4bytes, value not beyond TA shared memory size)
+ *    - TA shared buf
+ *   Receive buffer:
+ *    - TA shared buf
+ *
+ * - For TA unload debugfs interface:
+ *   Transmit buffer:
+ *    - TA type (4bytes)
+ *    - TA ID (4bytes)
+ */
+
+static ssize_t ta_if_load_debugfs_write(struct file *fp, const char *buf, size_t len, loff_t *off)
+{
+	uint32_t ta_type    = 0;
+	uint32_t ta_bin_len = 0;
+	uint8_t  *ta_bin    = NULL;
+	uint32_t copy_pos   = 0;
+	int      ret        = 0;
+
+	struct amdgpu_device *adev    = (struct amdgpu_device *)file_inode(fp)->i_private;
+	struct psp_context   *psp     = &adev->psp;
+	struct ta_context    *context = NULL;
+
+	if (!buf)
+		return -EINVAL;
+
+	ret = copy_from_user((void *)&ta_type, &buf[copy_pos], sizeof(uint32_t));
+	if (ret || (!is_ta_type_valid(ta_type)))
+		return -EFAULT;
+
+	copy_pos += sizeof(uint32_t);
+
+	ret = copy_from_user((void *)&ta_bin_len, &buf[copy_pos], sizeof(uint32_t));
+	if (ret)
+		return -EFAULT;
+
+	if (ta_bin_len > PSP_1_MEG)
+		return -EINVAL;
+
+	copy_pos += sizeof(uint32_t);
+
+	ta_bin = memdup_user(&buf[copy_pos], ta_bin_len);
+	if (IS_ERR(ta_bin))
+		return PTR_ERR(ta_bin);
+
+	/* Set TA context and functions */
+	set_ta_context_funcs(psp, ta_type, &context);
+
+	if (!psp->ta_funcs || !psp->ta_funcs->fn_ta_terminate) {
+		dev_err(adev->dev, "Unsupported function to terminate TA\n");
+		ret = -EOPNOTSUPP;
+		goto err_free_bin;
+	}
+
+	/*
+	 * Allocate TA shared buf in case shared buf was freed
+	 * due to loading TA failed before.
+	 */
+	if (!context->mem_context.shared_buf) {
+		ret = psp_ta_init_shared_buf(psp, &context->mem_context);
+		if (ret) {
+			ret = -ENOMEM;
+			goto err_free_bin;
+		}
+	}
+
+	ret = psp_fn_ta_terminate(psp);
+	if (ret || context->resp_status) {
+		dev_err(adev->dev,
+			"Failed to unload embedded TA (%d) and status (0x%X)\n",
+			ret, context->resp_status);
+		if (!ret)
+			ret = -EINVAL;
+		goto err_free_ta_shared_buf;
+	}
+
+	/* Prepare TA context for TA initialization */
+	context->ta_type                     = ta_type;
+	context->bin_desc.fw_version         = get_bin_version(ta_bin);
+	context->bin_desc.size_bytes         = ta_bin_len;
+	context->bin_desc.start_addr         = ta_bin;
+
+	if (!psp->ta_funcs->fn_ta_initialize) {
+		dev_err(adev->dev, "Unsupported function to initialize TA\n");
+		ret = -EOPNOTSUPP;
+		goto err_free_ta_shared_buf;
+	}
+
+	ret = psp_fn_ta_initialize(psp);
+	if (ret || context->resp_status) {
+		dev_err(adev->dev, "Failed to load TA via debugfs (%d) and status (0x%X)\n",
+			ret, context->resp_status);
+		if (!ret)
+			ret = -EINVAL;
+		goto err_free_ta_shared_buf;
+	}
+
+	if (copy_to_user((char *)buf, (void *)&context->session_id, sizeof(uint32_t)))
+		ret = -EFAULT;
+
+err_free_ta_shared_buf:
+	/* Only free TA shared buf when returns error code */
+	if (ret && context->mem_context.shared_buf)
+		psp_ta_free_shared_buf(&context->mem_context);
+err_free_bin:
+	kfree(ta_bin);
+
+	return ret;
+}
+
+static ssize_t ta_if_unload_debugfs_write(struct file *fp, const char *buf, size_t len, loff_t *off)
+{
+	uint32_t ta_type    = 0;
+	uint32_t ta_id      = 0;
+	uint32_t copy_pos   = 0;
+	int      ret        = 0;
+
+	struct amdgpu_device *adev    = (struct amdgpu_device *)file_inode(fp)->i_private;
+	struct psp_context   *psp     = &adev->psp;
+	struct ta_context    *context = NULL;
+
+	if (!buf)
+		return -EINVAL;
+
+	ret = copy_from_user((void *)&ta_type, &buf[copy_pos], sizeof(uint32_t));
+	if (ret || (!is_ta_type_valid(ta_type)))
+		return -EFAULT;
+
+	copy_pos += sizeof(uint32_t);
+
+	ret = copy_from_user((void *)&ta_id, &buf[copy_pos], sizeof(uint32_t));
+	if (ret)
+		return -EFAULT;
+
+	set_ta_context_funcs(psp, ta_type, &context);
+	context->session_id = ta_id;
+
+	if (!psp->ta_funcs || !psp->ta_funcs->fn_ta_terminate) {
+		dev_err(adev->dev, "Unsupported function to terminate TA\n");
+		return -EOPNOTSUPP;
+	}
+
+	ret = psp_fn_ta_terminate(psp);
+	if (ret || context->resp_status) {
+		dev_err(adev->dev, "Failed to unload TA via debugfs (%d) and status (0x%X)\n",
+			ret, context->resp_status);
+		if (!ret)
+			ret = -EINVAL;
+	}
+
+	if (context->mem_context.shared_buf)
+		psp_ta_free_shared_buf(&context->mem_context);
+
+	return ret;
+}
+
+static ssize_t ta_if_invoke_debugfs_write(struct file *fp, const char *buf, size_t len, loff_t *off)
+{
+	uint32_t ta_type        = 0;
+	uint32_t ta_id          = 0;
+	uint32_t cmd_id         = 0;
+	uint32_t shared_buf_len = 0;
+	uint8_t *shared_buf     = NULL;
+	uint32_t copy_pos       = 0;
+	int      ret            = 0;
+
+	struct amdgpu_device *adev    = (struct amdgpu_device *)file_inode(fp)->i_private;
+	struct psp_context   *psp     = &adev->psp;
+	struct ta_context    *context = NULL;
+
+	if (!buf)
+		return -EINVAL;
+
+	ret = copy_from_user((void *)&ta_type, &buf[copy_pos], sizeof(uint32_t));
+	if (ret)
+		return -EFAULT;
+	copy_pos += sizeof(uint32_t);
+
+	ret = copy_from_user((void *)&ta_id, &buf[copy_pos], sizeof(uint32_t));
+	if (ret)
+		return -EFAULT;
+	copy_pos += sizeof(uint32_t);
+
+	ret = copy_from_user((void *)&cmd_id, &buf[copy_pos], sizeof(uint32_t));
+	if (ret)
+		return -EFAULT;
+	copy_pos += sizeof(uint32_t);
+
+	ret = copy_from_user((void *)&shared_buf_len, &buf[copy_pos], sizeof(uint32_t));
+	if (ret)
+		return -EFAULT;
+	copy_pos += sizeof(uint32_t);
+
+	shared_buf = memdup_user(&buf[copy_pos], shared_buf_len);
+	if (IS_ERR(shared_buf))
+		return PTR_ERR(shared_buf);
+
+	set_ta_context_funcs(psp, ta_type, &context);
+
+	if (!context || !context->initialized) {
+		dev_err(adev->dev, "TA is not initialized\n");
+		ret = -EINVAL;
+		goto err_free_shared_buf;
+	}
+
+	if (!psp->ta_funcs || !psp->ta_funcs->fn_ta_invoke) {
+		dev_err(adev->dev, "Unsupported function to invoke TA\n");
+		ret = -EOPNOTSUPP;
+		goto err_free_shared_buf;
+	}
+
+	context->session_id = ta_id;
+
+	mutex_lock(&psp->ras_context.mutex);
+	ret = prep_ta_mem_context(&context->mem_context, shared_buf, shared_buf_len);
+	if (ret)
+		goto err_free_shared_buf;
+
+	ret = psp_fn_ta_invoke(psp, cmd_id);
+	if (ret || context->resp_status) {
+		dev_err(adev->dev, "Failed to invoke TA via debugfs (%d) and status (0x%X)\n",
+			ret, context->resp_status);
+		if (!ret) {
+			ret = -EINVAL;
+			goto err_free_shared_buf;
+		}
+	}
+
+	if (copy_to_user((char *)&buf[copy_pos], context->mem_context.shared_buf, shared_buf_len))
+		ret = -EFAULT;
+
+err_free_shared_buf:
+	mutex_unlock(&psp->ras_context.mutex);
+	kfree(shared_buf);
+
+	return ret;
+}
+
+void amdgpu_ta_if_debugfs_init(struct amdgpu_device *adev)
+{
+	struct drm_minor *minor = adev_to_drm(adev)->primary;
+
+	struct dentry *dir = debugfs_create_dir("ta_if", minor->debugfs_root);
+
+	debugfs_create_file("ta_load", 0200, dir, adev,
+				     &ta_load_debugfs_fops);
+
+	debugfs_create_file("ta_unload", 0200, dir,
+				     adev, &ta_unload_debugfs_fops);
+
+	debugfs_create_file("ta_invoke", 0200, dir,
+				     adev, &ta_invoke_debugfs_fops);
+}
+
+#else
+void amdgpu_ta_if_debugfs_init(struct amdgpu_device *adev)
+{
+
+}
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.h
new file mode 100644
index 000000000000..14cd1c81c3e6
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.h
@@ -0,0 +1,34 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_PSP_TA_H__
+#define __AMDGPU_PSP_TA_H__
+
+/* Calling set_ta_context_funcs is required before using the following macros */
+#define psp_fn_ta_initialize(psp) ((psp)->ta_funcs->fn_ta_initialize((psp)))
+#define psp_fn_ta_invoke(psp, ta_cmd_id) ((psp)->ta_funcs->fn_ta_invoke((psp), (ta_cmd_id)))
+#define psp_fn_ta_terminate(psp) ((psp)->ta_funcs->fn_ta_terminate((psp)))
+
+void amdgpu_ta_if_debugfs_init(struct amdgpu_device *adev);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_queue_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_queue_mgr.c
deleted file mode 100644
index befc09b68543..000000000000
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_queue_mgr.c
+++ /dev/null
@@ -1,299 +0,0 @@
-/*
- * Copyright 2017 Valve Corporation
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- * Authors: Andres Rodriguez
- */
-
-#include "amdgpu.h"
-#include "amdgpu_ring.h"
-
-static int amdgpu_queue_mapper_init(struct amdgpu_queue_mapper *mapper,
-				    int hw_ip)
-{
-	if (!mapper)
-		return -EINVAL;
-
-	if (hw_ip > AMDGPU_MAX_IP_NUM)
-		return -EINVAL;
-
-	mapper->hw_ip = hw_ip;
-	mutex_init(&mapper->lock);
-
-	memset(mapper->queue_map, 0, sizeof(mapper->queue_map));
-
-	return 0;
-}
-
-static struct amdgpu_ring *amdgpu_get_cached_map(struct amdgpu_queue_mapper *mapper,
-					  int ring)
-{
-	return mapper->queue_map[ring];
-}
-
-static int amdgpu_update_cached_map(struct amdgpu_queue_mapper *mapper,
-			     int ring, struct amdgpu_ring *pring)
-{
-	if (WARN_ON(mapper->queue_map[ring])) {
-		DRM_ERROR("Un-expected ring re-map\n");
-		return -EINVAL;
-	}
-
-	mapper->queue_map[ring] = pring;
-
-	return 0;
-}
-
-static int amdgpu_identity_map(struct amdgpu_device *adev,
-			       struct amdgpu_queue_mapper *mapper,
-			       int ring,
-			       struct amdgpu_ring **out_ring)
-{
-	switch (mapper->hw_ip) {
-	case AMDGPU_HW_IP_GFX:
-		*out_ring = &adev->gfx.gfx_ring[ring];
-		break;
-	case AMDGPU_HW_IP_COMPUTE:
-		*out_ring = &adev->gfx.compute_ring[ring];
-		break;
-	case AMDGPU_HW_IP_DMA:
-		*out_ring = &adev->sdma.instance[ring].ring;
-		break;
-	case AMDGPU_HW_IP_UVD:
-		*out_ring = &adev->uvd.ring;
-		break;
-	case AMDGPU_HW_IP_VCE:
-		*out_ring = &adev->vce.ring[ring];
-		break;
-	case AMDGPU_HW_IP_UVD_ENC:
-		*out_ring = &adev->uvd.ring_enc[ring];
-		break;
-	case AMDGPU_HW_IP_VCN_DEC:
-		*out_ring = &adev->vcn.ring_dec;
-		break;
-	case AMDGPU_HW_IP_VCN_ENC:
-		*out_ring = &adev->vcn.ring_enc[ring];
-		break;
-	default:
-		*out_ring = NULL;
-		DRM_ERROR("unknown HW IP type: %d\n", mapper->hw_ip);
-		return -EINVAL;
-	}
-
-	return amdgpu_update_cached_map(mapper, ring, *out_ring);
-}
-
-static enum amdgpu_ring_type amdgpu_hw_ip_to_ring_type(int hw_ip)
-{
-	switch (hw_ip) {
-	case AMDGPU_HW_IP_GFX:
-		return AMDGPU_RING_TYPE_GFX;
-	case AMDGPU_HW_IP_COMPUTE:
-		return AMDGPU_RING_TYPE_COMPUTE;
-	case AMDGPU_HW_IP_DMA:
-		return AMDGPU_RING_TYPE_SDMA;
-	case AMDGPU_HW_IP_UVD:
-		return AMDGPU_RING_TYPE_UVD;
-	case AMDGPU_HW_IP_VCE:
-		return AMDGPU_RING_TYPE_VCE;
-	default:
-		DRM_ERROR("Invalid HW IP specified %d\n", hw_ip);
-		return -1;
-	}
-}
-
-static int amdgpu_lru_map(struct amdgpu_device *adev,
-			  struct amdgpu_queue_mapper *mapper,
-			  int user_ring,
-			  struct amdgpu_ring **out_ring)
-{
-	int r, i, j;
-	int ring_type = amdgpu_hw_ip_to_ring_type(mapper->hw_ip);
-	int ring_blacklist[AMDGPU_MAX_RINGS];
-	struct amdgpu_ring *ring;
-
-	/* 0 is a valid ring index, so initialize to -1 */
-	memset(ring_blacklist, 0xff, sizeof(ring_blacklist));
-
-	for (i = 0, j = 0; i < AMDGPU_MAX_RINGS; i++) {
-		ring = mapper->queue_map[i];
-		if (ring)
-			ring_blacklist[j++] = ring->idx;
-	}
-
-	r = amdgpu_ring_lru_get(adev, ring_type, ring_blacklist,
-				j, out_ring);
-	if (r)
-		return r;
-
-	return amdgpu_update_cached_map(mapper, user_ring, *out_ring);
-}
-
-/**
- * amdgpu_queue_mgr_init - init an amdgpu_queue_mgr struct
- *
- * @adev: amdgpu_device pointer
- * @mgr: amdgpu_queue_mgr structure holding queue information
- *
- * Initialize the the selected @mgr (all asics).
- *
- * Returns 0 on success, error on failure.
- */
-int amdgpu_queue_mgr_init(struct amdgpu_device *adev,
-			  struct amdgpu_queue_mgr *mgr)
-{
-	int i, r;
-
-	if (!adev || !mgr)
-		return -EINVAL;
-
-	memset(mgr, 0, sizeof(*mgr));
-
-	for (i = 0; i < AMDGPU_MAX_IP_NUM; ++i) {
-		r = amdgpu_queue_mapper_init(&mgr->mapper[i], i);
-		if (r)
-			return r;
-	}
-
-	return 0;
-}
-
-/**
- * amdgpu_queue_mgr_fini - de-initialize an amdgpu_queue_mgr struct
- *
- * @adev: amdgpu_device pointer
- * @mgr: amdgpu_queue_mgr structure holding queue information
- *
- * De-initialize the the selected @mgr (all asics).
- *
- * Returns 0 on success, error on failure.
- */
-int amdgpu_queue_mgr_fini(struct amdgpu_device *adev,
-			  struct amdgpu_queue_mgr *mgr)
-{
-	return 0;
-}
-
-/**
- * amdgpu_queue_mgr_map - Map a userspace ring id to an amdgpu_ring
- *
- * @adev: amdgpu_device pointer
- * @mgr: amdgpu_queue_mgr structure holding queue information
- * @hw_ip: HW IP enum
- * @instance: HW instance
- * @ring: user ring id
- * @our_ring: pointer to mapped amdgpu_ring
- *
- * Map a userspace ring id to an appropriate kernel ring. Different
- * policies are configurable at a HW IP level.
- *
- * Returns 0 on success, error on failure.
- */
-int amdgpu_queue_mgr_map(struct amdgpu_device *adev,
-			 struct amdgpu_queue_mgr *mgr,
-			 int hw_ip, int instance, int ring,
-			 struct amdgpu_ring **out_ring)
-{
-	int r, ip_num_rings;
-	struct amdgpu_queue_mapper *mapper = &mgr->mapper[hw_ip];
-
-	if (!adev || !mgr || !out_ring)
-		return -EINVAL;
-
-	if (hw_ip >= AMDGPU_MAX_IP_NUM)
-		return -EINVAL;
-
-	if (ring >= AMDGPU_MAX_RINGS)
-		return -EINVAL;
-
-	/* Right now all IPs have only one instance - multiple rings. */
-	if (instance != 0) {
-		DRM_ERROR("invalid ip instance: %d\n", instance);
-		return -EINVAL;
-	}
-
-	switch (hw_ip) {
-	case AMDGPU_HW_IP_GFX:
-		ip_num_rings = adev->gfx.num_gfx_rings;
-		break;
-	case AMDGPU_HW_IP_COMPUTE:
-		ip_num_rings = adev->gfx.num_compute_rings;
-		break;
-	case AMDGPU_HW_IP_DMA:
-		ip_num_rings = adev->sdma.num_instances;
-		break;
-	case AMDGPU_HW_IP_UVD:
-		ip_num_rings = 1;
-		break;
-	case AMDGPU_HW_IP_VCE:
-		ip_num_rings = adev->vce.num_rings;
-		break;
-	case AMDGPU_HW_IP_UVD_ENC:
-		ip_num_rings = adev->uvd.num_enc_rings;
-		break;
-	case AMDGPU_HW_IP_VCN_DEC:
-		ip_num_rings = 1;
-		break;
-	case AMDGPU_HW_IP_VCN_ENC:
-		ip_num_rings = adev->vcn.num_enc_rings;
-		break;
-	default:
-		DRM_ERROR("unknown ip type: %d\n", hw_ip);
-		return -EINVAL;
-	}
-
-	if (ring >= ip_num_rings) {
-		DRM_ERROR("Ring index:%d exceeds maximum:%d for ip:%d\n",
-				ring, ip_num_rings, hw_ip);
-		return -EINVAL;
-	}
-
-	mutex_lock(&mapper->lock);
-
-	*out_ring = amdgpu_get_cached_map(mapper, ring);
-	if (*out_ring) {
-		/* cache hit */
-		r = 0;
-		goto out_unlock;
-	}
-
-	switch (mapper->hw_ip) {
-	case AMDGPU_HW_IP_GFX:
-	case AMDGPU_HW_IP_UVD:
-	case AMDGPU_HW_IP_VCE:
-	case AMDGPU_HW_IP_UVD_ENC:
-	case AMDGPU_HW_IP_VCN_DEC:
-	case AMDGPU_HW_IP_VCN_ENC:
-		r = amdgpu_identity_map(adev, mapper, ring, out_ring);
-		break;
-	case AMDGPU_HW_IP_DMA:
-	case AMDGPU_HW_IP_COMPUTE:
-		r = amdgpu_lru_map(adev, mapper, ring, out_ring);
-		break;
-	default:
-		*out_ring = NULL;
-		r = -EINVAL;
-		DRM_ERROR("unknown HW IP type: %d\n", mapper->hw_ip);
-	}
-
-out_unlock:
-	mutex_unlock(&mapper->lock);
-	return r;
-}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_rap.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_rap.c
new file mode 100644
index 000000000000..123bcf5c2bb1
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_rap.c
@@ -0,0 +1,126 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ *
+ */
+#include <linux/debugfs.h>
+#include <linux/pm_runtime.h>
+
+#include "amdgpu.h"
+#include "amdgpu_rap.h"
+
+/**
+ * DOC: AMDGPU RAP debugfs test interface
+ *
+ * how to use?
+ * echo opcode > <debugfs_dir>/dri/xxx/rap_test
+ *
+ * opcode:
+ * currently, only 2 is supported by Linux host driver,
+ * opcode 2 stands for TA_CMD_RAP__VALIDATE_L0, used to
+ * trigger L0 policy validation, you can refer more detail
+ * from header file ta_rap_if.h
+ *
+ */
+static ssize_t amdgpu_rap_debugfs_write(struct file *f, const char __user *buf,
+		size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)file_inode(f)->i_private;
+	struct ta_rap_shared_memory *rap_shared_mem;
+	struct ta_rap_cmd_output_data *rap_cmd_output;
+	struct drm_device *dev = adev_to_drm(adev);
+	uint32_t op;
+	enum ta_rap_status status;
+	int ret;
+
+	if (*pos || size != 2)
+		return -EINVAL;
+
+	ret = kstrtouint_from_user(buf, size, *pos, &op);
+	if (ret)
+		return ret;
+
+	ret = pm_runtime_get_sync(dev->dev);
+	if (ret < 0) {
+		pm_runtime_put_autosuspend(dev->dev);
+		return ret;
+	}
+
+	/* make sure gfx core is on, RAP TA cann't handle
+	 * GFX OFF case currently.
+	 */
+	amdgpu_gfx_off_ctrl(adev, false);
+
+	switch (op) {
+	case 2:
+		ret = psp_rap_invoke(&adev->psp, op, &status);
+		if (!ret && status == TA_RAP_STATUS__SUCCESS) {
+			dev_info(adev->dev, "RAP L0 validate test success.\n");
+		} else {
+			rap_shared_mem = (struct ta_rap_shared_memory *)
+					 adev->psp.rap_context.context.mem_context.shared_buf;
+			rap_cmd_output = &(rap_shared_mem->rap_out_message.output);
+
+			dev_info(adev->dev, "RAP test failed, the output is:\n");
+			dev_info(adev->dev, "\tlast_subsection: 0x%08x.\n",
+				 rap_cmd_output->last_subsection);
+			dev_info(adev->dev, "\tnum_total_validate: 0x%08x.\n",
+				 rap_cmd_output->num_total_validate);
+			dev_info(adev->dev, "\tnum_valid: 0x%08x.\n",
+				 rap_cmd_output->num_valid);
+			dev_info(adev->dev, "\tlast_validate_addr: 0x%08x.\n",
+				 rap_cmd_output->last_validate_addr);
+			dev_info(adev->dev, "\tlast_validate_val: 0x%08x.\n",
+				 rap_cmd_output->last_validate_val);
+			dev_info(adev->dev, "\tlast_validate_val_exptd: 0x%08x.\n",
+				 rap_cmd_output->last_validate_val_exptd);
+		}
+		break;
+	default:
+		dev_info(adev->dev, "Unsupported op id: %d, ", op);
+		dev_info(adev->dev, "Only support op 2(L0 validate test).\n");
+		break;
+	}
+
+	amdgpu_gfx_off_ctrl(adev, true);
+	pm_runtime_mark_last_busy(dev->dev);
+	pm_runtime_put_autosuspend(dev->dev);
+
+	return size;
+}
+
+static const struct file_operations amdgpu_rap_debugfs_ops = {
+	.owner = THIS_MODULE,
+	.read = NULL,
+	.write = amdgpu_rap_debugfs_write,
+	.llseek = default_llseek
+};
+
+void amdgpu_rap_debugfs_init(struct amdgpu_device *adev)
+{
+	struct drm_minor *minor = adev_to_drm(adev)->primary;
+
+	if (!adev->psp.rap_context.context.initialized)
+		return;
+
+	debugfs_create_file("rap_test", S_IWUSR, minor->debugfs_root,
+				adev, &amdgpu_rap_debugfs_ops);
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_rap.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_rap.h
new file mode 100644
index 000000000000..ec6d7632d3a0
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_rap.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ *
+ */
+#ifndef _AMDGPU_RAP_H
+#define _AMDGPU_RAP_H
+
+#include "amdgpu.h"
+
+void amdgpu_rap_debugfs_init(struct amdgpu_device *adev);
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
new file mode 100644
index 000000000000..e0ee21150860
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
@@ -0,0 +1,5492 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ *
+ */
+#include <linux/debugfs.h>
+#include <linux/list.h>
+#include <linux/module.h>
+#include <linux/uaccess.h>
+#include <linux/reboot.h>
+#include <linux/syscalls.h>
+#include <linux/pm_runtime.h>
+#include <linux/list_sort.h>
+
+#include "amdgpu.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_atomfirmware.h"
+#include "amdgpu_xgmi.h"
+#include "ivsrcid/nbio/irqsrcs_nbif_7_4.h"
+#include "nbio_v4_3.h"
+#include "nbif_v6_3_1.h"
+#include "nbio_v7_9.h"
+#include "atom.h"
+#include "amdgpu_reset.h"
+#include "amdgpu_psp.h"
+
+#ifdef CONFIG_X86_MCE_AMD
+#include <asm/mce.h>
+
+static bool notifier_registered;
+#endif
+static const char *RAS_FS_NAME = "ras";
+
+const char *ras_error_string[] = {
+	"none",
+	"parity",
+	"single_correctable",
+	"multi_uncorrectable",
+	"poison",
+};
+
+const char *ras_block_string[] = {
+	"umc",
+	"sdma",
+	"gfx",
+	"mmhub",
+	"athub",
+	"pcie_bif",
+	"hdp",
+	"xgmi_wafl",
+	"df",
+	"smn",
+	"sem",
+	"mp0",
+	"mp1",
+	"fuse",
+	"mca",
+	"vcn",
+	"jpeg",
+	"ih",
+	"mpio",
+	"mmsch",
+};
+
+const char *ras_mca_block_string[] = {
+	"mca_mp0",
+	"mca_mp1",
+	"mca_mpio",
+	"mca_iohc",
+};
+
+struct amdgpu_ras_block_list {
+	/* ras block link */
+	struct list_head node;
+
+	struct amdgpu_ras_block_object *ras_obj;
+};
+
+const char *get_ras_block_str(struct ras_common_if *ras_block)
+{
+	if (!ras_block)
+		return "NULL";
+
+	if (ras_block->block >= AMDGPU_RAS_BLOCK_COUNT ||
+	    ras_block->block >= ARRAY_SIZE(ras_block_string))
+		return "OUT OF RANGE";
+
+	if (ras_block->block == AMDGPU_RAS_BLOCK__MCA)
+		return ras_mca_block_string[ras_block->sub_block_index];
+
+	return ras_block_string[ras_block->block];
+}
+
+#define ras_block_str(_BLOCK_) \
+	(((_BLOCK_) < ARRAY_SIZE(ras_block_string)) ? ras_block_string[_BLOCK_] : "Out Of Range")
+
+#define ras_err_str(i) (ras_error_string[ffs(i)])
+
+#define RAS_DEFAULT_FLAGS (AMDGPU_RAS_FLAG_INIT_BY_VBIOS)
+
+/* inject address is 52 bits */
+#define	RAS_UMC_INJECT_ADDR_LIMIT	(0x1ULL << 52)
+
+/* typical ECC bad page rate is 1 bad page per 100MB VRAM */
+#define RAS_BAD_PAGE_COVER              (100 * 1024 * 1024ULL)
+
+#define MAX_UMC_POISON_POLLING_TIME_ASYNC  10
+
+#define AMDGPU_RAS_RETIRE_PAGE_INTERVAL 100  //ms
+
+#define MAX_FLUSH_RETIRE_DWORK_TIMES  100
+
+#define BYPASS_ALLOCATED_ADDRESS        0x0
+#define BYPASS_INITIALIZATION_ADDRESS   0x1
+
+enum amdgpu_ras_retire_page_reservation {
+	AMDGPU_RAS_RETIRE_PAGE_RESERVED,
+	AMDGPU_RAS_RETIRE_PAGE_PENDING,
+	AMDGPU_RAS_RETIRE_PAGE_FAULT,
+};
+
+atomic_t amdgpu_ras_in_intr = ATOMIC_INIT(0);
+
+static int amdgpu_ras_check_bad_page_unlock(struct amdgpu_ras *con,
+				uint64_t addr);
+static int amdgpu_ras_check_bad_page(struct amdgpu_device *adev,
+				uint64_t addr);
+
+static void amdgpu_ras_critical_region_init(struct amdgpu_device *adev);
+static void amdgpu_ras_critical_region_fini(struct amdgpu_device *adev);
+
+#ifdef CONFIG_X86_MCE_AMD
+static void amdgpu_register_bad_pages_mca_notifier(struct amdgpu_device *adev);
+struct mce_notifier_adev_list {
+	struct amdgpu_device *devs[MAX_GPU_INSTANCE];
+	int num_gpu;
+};
+static struct mce_notifier_adev_list mce_adev_list;
+#endif
+
+void amdgpu_ras_set_error_query_ready(struct amdgpu_device *adev, bool ready)
+{
+	if (adev && amdgpu_ras_get_context(adev))
+		amdgpu_ras_get_context(adev)->error_query_ready = ready;
+}
+
+static bool amdgpu_ras_get_error_query_ready(struct amdgpu_device *adev)
+{
+	if (adev && amdgpu_ras_get_context(adev))
+		return amdgpu_ras_get_context(adev)->error_query_ready;
+
+	return false;
+}
+
+static int amdgpu_reserve_page_direct(struct amdgpu_device *adev, uint64_t address)
+{
+	struct ras_err_data err_data;
+	struct eeprom_table_record err_rec;
+	int ret;
+
+	ret = amdgpu_ras_check_bad_page(adev, address);
+	if (ret == -EINVAL) {
+		dev_warn(adev->dev,
+			"RAS WARN: input address 0x%llx is invalid.\n",
+			address);
+		return -EINVAL;
+	} else if (ret == 1) {
+		dev_warn(adev->dev,
+			"RAS WARN: 0x%llx has already been marked as bad page!\n",
+			address);
+		return 0;
+	}
+
+	ret = amdgpu_ras_error_data_init(&err_data);
+	if (ret)
+		return ret;
+
+	memset(&err_rec, 0x0, sizeof(struct eeprom_table_record));
+	err_data.err_addr = &err_rec;
+	amdgpu_umc_fill_error_record(&err_data, address, address, 0, 0);
+
+	if (amdgpu_bad_page_threshold != 0) {
+		amdgpu_ras_add_bad_pages(adev, err_data.err_addr,
+					 err_data.err_addr_cnt, false);
+		amdgpu_ras_save_bad_pages(adev, NULL);
+	}
+
+	amdgpu_ras_error_data_fini(&err_data);
+
+	dev_warn(adev->dev, "WARNING: THIS IS ONLY FOR TEST PURPOSES AND WILL CORRUPT RAS EEPROM\n");
+	dev_warn(adev->dev, "Clear EEPROM:\n");
+	dev_warn(adev->dev, "    echo 1 > /sys/kernel/debug/dri/0/ras/ras_eeprom_reset\n");
+
+	return 0;
+}
+
+static int amdgpu_check_address_validity(struct amdgpu_device *adev,
+			uint64_t address, uint64_t flags)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct amdgpu_vram_block_info blk_info;
+	uint64_t page_pfns[32] = {0};
+	int i, ret, count;
+	bool hit = false;
+
+	if (amdgpu_ip_version(adev, UMC_HWIP, 0) < IP_VERSION(12, 0, 0))
+		return 0;
+
+	if (amdgpu_sriov_vf(adev)) {
+		if (amdgpu_virt_check_vf_critical_region(adev, address, &hit))
+			return -EPERM;
+		return hit ? -EACCES : 0;
+	}
+
+	if ((address >= adev->gmc.mc_vram_size) ||
+	    (address >= RAS_UMC_INJECT_ADDR_LIMIT))
+		return -EFAULT;
+
+	count = amdgpu_umc_lookup_bad_pages_in_a_row(adev,
+				address, page_pfns, ARRAY_SIZE(page_pfns));
+	if (count <= 0)
+		return -EPERM;
+
+	for (i = 0; i < count; i++) {
+		memset(&blk_info, 0, sizeof(blk_info));
+		ret = amdgpu_vram_mgr_query_address_block_info(&adev->mman.vram_mgr,
+					page_pfns[i] << AMDGPU_GPU_PAGE_SHIFT, &blk_info);
+		if (!ret) {
+			/* The input address that needs to be checked is allocated by
+			 * current calling process, so it is necessary to exclude
+			 * the calling process.
+			 */
+			if ((flags == BYPASS_ALLOCATED_ADDRESS) &&
+			    ((blk_info.task.pid != task_pid_nr(current)) ||
+				strncmp(blk_info.task.comm, current->comm, TASK_COMM_LEN)))
+				return -EACCES;
+			else if ((flags == BYPASS_INITIALIZATION_ADDRESS) &&
+				(blk_info.task.pid == con->init_task_pid) &&
+				!strncmp(blk_info.task.comm, con->init_task_comm, TASK_COMM_LEN))
+				return -EACCES;
+		}
+	}
+
+	return 0;
+}
+
+static ssize_t amdgpu_ras_debugfs_read(struct file *f, char __user *buf,
+					size_t size, loff_t *pos)
+{
+	struct ras_manager *obj = (struct ras_manager *)file_inode(f)->i_private;
+	struct ras_query_if info = {
+		.head = obj->head,
+	};
+	ssize_t s;
+	char val[128];
+
+	if (amdgpu_ras_query_error_status(obj->adev, &info))
+		return -EINVAL;
+
+	/* Hardware counter will be reset automatically after the query on Vega20 and Arcturus */
+	if (amdgpu_ip_version(obj->adev, MP0_HWIP, 0) != IP_VERSION(11, 0, 2) &&
+	    amdgpu_ip_version(obj->adev, MP0_HWIP, 0) != IP_VERSION(11, 0, 4)) {
+		if (amdgpu_ras_reset_error_status(obj->adev, info.head.block))
+			dev_warn(obj->adev->dev, "Failed to reset error counter and error status");
+	}
+
+	s = snprintf(val, sizeof(val), "%s: %lu\n%s: %lu\n",
+			"ue", info.ue_count,
+			"ce", info.ce_count);
+	if (*pos >= s)
+		return 0;
+
+	s -= *pos;
+	s = min_t(u64, s, size);
+
+
+	if (copy_to_user(buf, &val[*pos], s))
+		return -EINVAL;
+
+	*pos += s;
+
+	return s;
+}
+
+static const struct file_operations amdgpu_ras_debugfs_ops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_ras_debugfs_read,
+	.write = NULL,
+	.llseek = default_llseek
+};
+
+static int amdgpu_ras_find_block_id_by_name(const char *name, int *block_id)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(ras_block_string); i++) {
+		*block_id = i;
+		if (strcmp(name, ras_block_string[i]) == 0)
+			return 0;
+	}
+	return -EINVAL;
+}
+
+static int amdgpu_ras_debugfs_ctrl_parse_data(struct file *f,
+		const char __user *buf, size_t size,
+		loff_t *pos, struct ras_debug_if *data)
+{
+	ssize_t s = min_t(u64, 64, size);
+	char str[65];
+	char block_name[33];
+	char err[9] = "ue";
+	int op = -1;
+	int block_id;
+	uint32_t sub_block;
+	u64 address, value;
+	/* default value is 0 if the mask is not set by user */
+	u32 instance_mask = 0;
+
+	if (*pos)
+		return -EINVAL;
+	*pos = size;
+
+	memset(str, 0, sizeof(str));
+	memset(data, 0, sizeof(*data));
+
+	if (copy_from_user(str, buf, s))
+		return -EINVAL;
+
+	if (sscanf(str, "disable %32s", block_name) == 1)
+		op = 0;
+	else if (sscanf(str, "enable %32s %8s", block_name, err) == 2)
+		op = 1;
+	else if (sscanf(str, "inject %32s %8s", block_name, err) == 2)
+		op = 2;
+	else if (strstr(str, "retire_page") != NULL)
+		op = 3;
+	else if (strstr(str, "check_address") != NULL)
+		op = 4;
+	else if (str[0] && str[1] && str[2] && str[3])
+		/* ascii string, but commands are not matched. */
+		return -EINVAL;
+
+	if (op != -1) {
+		if (op == 3) {
+			if (sscanf(str, "%*s 0x%llx", &address) != 1 &&
+			    sscanf(str, "%*s %llu", &address) != 1)
+				return -EINVAL;
+
+			data->op = op;
+			data->inject.address = address;
+
+			return 0;
+		} else if (op == 4) {
+			if (sscanf(str, "%*s 0x%llx 0x%llx", &address, &value) != 2 &&
+			    sscanf(str, "%*s %llu %llu", &address, &value) != 2)
+				return -EINVAL;
+
+			data->op = op;
+			data->inject.address = address;
+			data->inject.value = value;
+			return 0;
+		}
+
+		if (amdgpu_ras_find_block_id_by_name(block_name, &block_id))
+			return -EINVAL;
+
+		data->head.block = block_id;
+		/* only ue, ce and poison errors are supported */
+		if (!memcmp("ue", err, 2))
+			data->head.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
+		else if (!memcmp("ce", err, 2))
+			data->head.type = AMDGPU_RAS_ERROR__SINGLE_CORRECTABLE;
+		else if (!memcmp("poison", err, 6))
+			data->head.type = AMDGPU_RAS_ERROR__POISON;
+		else
+			return -EINVAL;
+
+		data->op = op;
+
+		if (op == 2) {
+			if (sscanf(str, "%*s %*s %*s 0x%x 0x%llx 0x%llx 0x%x",
+				   &sub_block, &address, &value, &instance_mask) != 4 &&
+			    sscanf(str, "%*s %*s %*s %u %llu %llu %u",
+				   &sub_block, &address, &value, &instance_mask) != 4 &&
+				sscanf(str, "%*s %*s %*s 0x%x 0x%llx 0x%llx",
+				   &sub_block, &address, &value) != 3 &&
+			    sscanf(str, "%*s %*s %*s %u %llu %llu",
+				   &sub_block, &address, &value) != 3)
+				return -EINVAL;
+			data->head.sub_block_index = sub_block;
+			data->inject.address = address;
+			data->inject.value = value;
+			data->inject.instance_mask = instance_mask;
+		}
+	} else {
+		if (size < sizeof(*data))
+			return -EINVAL;
+
+		if (copy_from_user(data, buf, sizeof(*data)))
+			return -EINVAL;
+	}
+
+	return 0;
+}
+
+static void amdgpu_ras_instance_mask_check(struct amdgpu_device *adev,
+				struct ras_debug_if *data)
+{
+	int num_xcc = adev->gfx.xcc_mask ? NUM_XCC(adev->gfx.xcc_mask) : 1;
+	uint32_t mask, inst_mask = data->inject.instance_mask;
+
+	/* no need to set instance mask if there is only one instance */
+	if (num_xcc <= 1 && inst_mask) {
+		data->inject.instance_mask = 0;
+		dev_dbg(adev->dev,
+			"RAS inject mask(0x%x) isn't supported and force it to 0.\n",
+			inst_mask);
+
+		return;
+	}
+
+	switch (data->head.block) {
+	case AMDGPU_RAS_BLOCK__GFX:
+		mask = GENMASK(num_xcc - 1, 0);
+		break;
+	case AMDGPU_RAS_BLOCK__SDMA:
+		mask = GENMASK(adev->sdma.num_instances - 1, 0);
+		break;
+	case AMDGPU_RAS_BLOCK__VCN:
+	case AMDGPU_RAS_BLOCK__JPEG:
+		mask = GENMASK(adev->vcn.num_vcn_inst - 1, 0);
+		break;
+	default:
+		mask = inst_mask;
+		break;
+	}
+
+	/* remove invalid bits in instance mask */
+	data->inject.instance_mask &= mask;
+	if (inst_mask != data->inject.instance_mask)
+		dev_dbg(adev->dev,
+			"Adjust RAS inject mask 0x%x to 0x%x\n",
+			inst_mask, data->inject.instance_mask);
+}
+
+/**
+ * DOC: AMDGPU RAS debugfs control interface
+ *
+ * The control interface accepts struct ras_debug_if which has two members.
+ *
+ * First member: ras_debug_if::head or ras_debug_if::inject.
+ *
+ * head is used to indicate which IP block will be under control.
+ *
+ * head has four members, they are block, type, sub_block_index, name.
+ * block: which IP will be under control.
+ * type: what kind of error will be enabled/disabled/injected.
+ * sub_block_index: some IPs have subcomponets. say, GFX, sDMA.
+ * name: the name of IP.
+ *
+ * inject has three more members than head, they are address, value and mask.
+ * As their names indicate, inject operation will write the
+ * value to the address.
+ *
+ * The second member: struct ras_debug_if::op.
+ * It has three kinds of operations.
+ *
+ * - 0: disable RAS on the block. Take ::head as its data.
+ * - 1: enable RAS on the block. Take ::head as its data.
+ * - 2: inject errors on the block. Take ::inject as its data.
+ *
+ * How to use the interface?
+ *
+ * In a program
+ *
+ * Copy the struct ras_debug_if in your code and initialize it.
+ * Write the struct to the control interface.
+ *
+ * From shell
+ *
+ * .. code-block:: bash
+ *
+ *	echo "disable <block>" > /sys/kernel/debug/dri/<N>/ras/ras_ctrl
+ *	echo "enable  <block> <error>" > /sys/kernel/debug/dri/<N>/ras/ras_ctrl
+ *	echo "inject  <block> <error> <sub-block> <address> <value> <mask>" > /sys/kernel/debug/dri/<N>/ras/ras_ctrl
+ *
+ * Where N, is the card which you want to affect.
+ *
+ * "disable" requires only the block.
+ * "enable" requires the block and error type.
+ * "inject" requires the block, error type, address, and value.
+ *
+ * The block is one of: umc, sdma, gfx, etc.
+ *	see ras_block_string[] for details
+ *
+ * The error type is one of: ue, ce and poison where,
+ *	ue is multi-uncorrectable
+ *	ce is single-correctable
+ *	poison is poison
+ *
+ * The sub-block is a the sub-block index, pass 0 if there is no sub-block.
+ * The address and value are hexadecimal numbers, leading 0x is optional.
+ * The mask means instance mask, is optional, default value is 0x1.
+ *
+ * For instance,
+ *
+ * .. code-block:: bash
+ *
+ *	echo inject umc ue 0x0 0x0 0x0 > /sys/kernel/debug/dri/0/ras/ras_ctrl
+ *	echo inject umc ce 0 0 0 3 > /sys/kernel/debug/dri/0/ras/ras_ctrl
+ *	echo disable umc > /sys/kernel/debug/dri/0/ras/ras_ctrl
+ *
+ * How to check the result of the operation?
+ *
+ * To check disable/enable, see "ras" features at,
+ * /sys/class/drm/card[0/1/2...]/device/ras/features
+ *
+ * To check inject, see the corresponding error count at,
+ * /sys/class/drm/card[0/1/2...]/device/ras/[gfx|sdma|umc|...]_err_count
+ *
+ * .. note::
+ *	Operations are only allowed on blocks which are supported.
+ *	Check the "ras" mask at /sys/module/amdgpu/parameters/ras_mask
+ *	to see which blocks support RAS on a particular asic.
+ *
+ */
+static ssize_t amdgpu_ras_debugfs_ctrl_write(struct file *f,
+					     const char __user *buf,
+					     size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)file_inode(f)->i_private;
+	struct ras_debug_if data;
+	int ret = 0;
+
+	if (!amdgpu_ras_get_error_query_ready(adev)) {
+		dev_warn(adev->dev, "RAS WARN: error injection "
+				"currently inaccessible\n");
+		return size;
+	}
+
+	ret = amdgpu_ras_debugfs_ctrl_parse_data(f, buf, size, pos, &data);
+	if (ret)
+		return ret;
+
+	if (data.op == 3) {
+		ret = amdgpu_reserve_page_direct(adev, data.inject.address);
+		if (!ret)
+			return size;
+		else
+			return ret;
+	} else if (data.op == 4) {
+		ret = amdgpu_check_address_validity(adev, data.inject.address, data.inject.value);
+		return ret ? ret : size;
+	}
+
+	if (!amdgpu_ras_is_supported(adev, data.head.block))
+		return -EINVAL;
+
+	switch (data.op) {
+	case 0:
+		ret = amdgpu_ras_feature_enable(adev, &data.head, 0);
+		break;
+	case 1:
+		ret = amdgpu_ras_feature_enable(adev, &data.head, 1);
+		break;
+	case 2:
+		/* umc ce/ue error injection for a bad page is not allowed */
+		if (data.head.block == AMDGPU_RAS_BLOCK__UMC)
+			ret = amdgpu_ras_check_bad_page(adev, data.inject.address);
+		if (ret == -EINVAL) {
+			dev_warn(adev->dev, "RAS WARN: input address 0x%llx is invalid.",
+					data.inject.address);
+			break;
+		} else if (ret == 1) {
+			dev_warn(adev->dev, "RAS WARN: inject: 0x%llx has already been marked as bad!\n",
+					data.inject.address);
+			break;
+		}
+
+		amdgpu_ras_instance_mask_check(adev, &data);
+
+		/* data.inject.address is offset instead of absolute gpu address */
+		ret = amdgpu_ras_error_inject(adev, &data.inject);
+		break;
+	default:
+		ret = -EINVAL;
+		break;
+	}
+
+	if (ret)
+		return ret;
+
+	return size;
+}
+
+/**
+ * DOC: AMDGPU RAS debugfs EEPROM table reset interface
+ *
+ * Some boards contain an EEPROM which is used to persistently store a list of
+ * bad pages which experiences ECC errors in vram.  This interface provides
+ * a way to reset the EEPROM, e.g., after testing error injection.
+ *
+ * Usage:
+ *
+ * .. code-block:: bash
+ *
+ *	echo 1 > ../ras/ras_eeprom_reset
+ *
+ * will reset EEPROM table to 0 entries.
+ *
+ */
+static ssize_t amdgpu_ras_debugfs_eeprom_write(struct file *f,
+					       const char __user *buf,
+					       size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev =
+		(struct amdgpu_device *)file_inode(f)->i_private;
+	int ret;
+
+	ret = amdgpu_ras_eeprom_reset_table(
+		&(amdgpu_ras_get_context(adev)->eeprom_control));
+
+	if (!ret) {
+		/* Something was written to EEPROM.
+		 */
+		amdgpu_ras_get_context(adev)->flags = RAS_DEFAULT_FLAGS;
+		return size;
+	} else {
+		return ret;
+	}
+}
+
+static const struct file_operations amdgpu_ras_debugfs_ctrl_ops = {
+	.owner = THIS_MODULE,
+	.read = NULL,
+	.write = amdgpu_ras_debugfs_ctrl_write,
+	.llseek = default_llseek
+};
+
+static const struct file_operations amdgpu_ras_debugfs_eeprom_ops = {
+	.owner = THIS_MODULE,
+	.read = NULL,
+	.write = amdgpu_ras_debugfs_eeprom_write,
+	.llseek = default_llseek
+};
+
+/**
+ * DOC: AMDGPU RAS sysfs Error Count Interface
+ *
+ * It allows the user to read the error count for each IP block on the gpu through
+ * /sys/class/drm/card[0/1/2...]/device/ras/[gfx/sdma/...]_err_count
+ *
+ * It outputs the multiple lines which report the uncorrected (ue) and corrected
+ * (ce) error counts.
+ *
+ * The format of one line is below,
+ *
+ * [ce|ue]: count
+ *
+ * Example:
+ *
+ * .. code-block:: bash
+ *
+ *	ue: 0
+ *	ce: 1
+ *
+ */
+static ssize_t amdgpu_ras_sysfs_read(struct device *dev,
+		struct device_attribute *attr, char *buf)
+{
+	struct ras_manager *obj = container_of(attr, struct ras_manager, sysfs_attr);
+	struct ras_query_if info = {
+		.head = obj->head,
+	};
+
+	if (!amdgpu_ras_get_error_query_ready(obj->adev))
+		return sysfs_emit(buf, "Query currently inaccessible\n");
+
+	if (amdgpu_ras_query_error_status(obj->adev, &info))
+		return -EINVAL;
+
+	if (amdgpu_ip_version(obj->adev, MP0_HWIP, 0) != IP_VERSION(11, 0, 2) &&
+	    amdgpu_ip_version(obj->adev, MP0_HWIP, 0) != IP_VERSION(11, 0, 4)) {
+		if (amdgpu_ras_reset_error_status(obj->adev, info.head.block))
+			dev_warn(obj->adev->dev, "Failed to reset error counter and error status");
+	}
+
+	if (info.head.block == AMDGPU_RAS_BLOCK__UMC)
+		return sysfs_emit(buf, "%s: %lu\n%s: %lu\n%s: %lu\n", "ue", info.ue_count,
+				"ce", info.ce_count, "de", info.de_count);
+	else
+		return sysfs_emit(buf, "%s: %lu\n%s: %lu\n", "ue", info.ue_count,
+				"ce", info.ce_count);
+}
+
+/* obj begin */
+
+#define get_obj(obj) do { (obj)->use++; } while (0)
+#define alive_obj(obj) ((obj)->use)
+
+static inline void put_obj(struct ras_manager *obj)
+{
+	if (obj && (--obj->use == 0)) {
+		list_del(&obj->node);
+		amdgpu_ras_error_data_fini(&obj->err_data);
+	}
+
+	if (obj && (obj->use < 0))
+		DRM_ERROR("RAS ERROR: Unbalance obj(%s) use\n", get_ras_block_str(&obj->head));
+}
+
+/* make one obj and return it. */
+static struct ras_manager *amdgpu_ras_create_obj(struct amdgpu_device *adev,
+		struct ras_common_if *head)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_manager *obj;
+
+	if (!adev->ras_enabled || !con)
+		return NULL;
+
+	if (head->block >= AMDGPU_RAS_BLOCK_COUNT)
+		return NULL;
+
+	if (head->block == AMDGPU_RAS_BLOCK__MCA) {
+		if (head->sub_block_index >= AMDGPU_RAS_MCA_BLOCK__LAST)
+			return NULL;
+
+		obj = &con->objs[AMDGPU_RAS_BLOCK__LAST + head->sub_block_index];
+	} else
+		obj = &con->objs[head->block];
+
+	/* already exist. return obj? */
+	if (alive_obj(obj))
+		return NULL;
+
+	if (amdgpu_ras_error_data_init(&obj->err_data))
+		return NULL;
+
+	obj->head = *head;
+	obj->adev = adev;
+	list_add(&obj->node, &con->head);
+	get_obj(obj);
+
+	return obj;
+}
+
+/* return an obj equal to head, or the first when head is NULL */
+struct ras_manager *amdgpu_ras_find_obj(struct amdgpu_device *adev,
+		struct ras_common_if *head)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_manager *obj;
+	int i;
+
+	if (!adev->ras_enabled || !con)
+		return NULL;
+
+	if (head) {
+		if (head->block >= AMDGPU_RAS_BLOCK_COUNT)
+			return NULL;
+
+		if (head->block == AMDGPU_RAS_BLOCK__MCA) {
+			if (head->sub_block_index >= AMDGPU_RAS_MCA_BLOCK__LAST)
+				return NULL;
+
+			obj = &con->objs[AMDGPU_RAS_BLOCK__LAST + head->sub_block_index];
+		} else
+			obj = &con->objs[head->block];
+
+		if (alive_obj(obj))
+			return obj;
+	} else {
+		for (i = 0; i < AMDGPU_RAS_BLOCK_COUNT + AMDGPU_RAS_MCA_BLOCK_COUNT; i++) {
+			obj = &con->objs[i];
+			if (alive_obj(obj))
+				return obj;
+		}
+	}
+
+	return NULL;
+}
+/* obj end */
+
+/* feature ctl begin */
+static int amdgpu_ras_is_feature_allowed(struct amdgpu_device *adev,
+					 struct ras_common_if *head)
+{
+	return adev->ras_hw_enabled & BIT(head->block);
+}
+
+static int amdgpu_ras_is_feature_enabled(struct amdgpu_device *adev,
+		struct ras_common_if *head)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+
+	return con->features & BIT(head->block);
+}
+
+/*
+ * if obj is not created, then create one.
+ * set feature enable flag.
+ */
+static int __amdgpu_ras_feature_enable(struct amdgpu_device *adev,
+		struct ras_common_if *head, int enable)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_manager *obj = amdgpu_ras_find_obj(adev, head);
+
+	/* If hardware does not support ras, then do not create obj.
+	 * But if hardware support ras, we can create the obj.
+	 * Ras framework checks con->hw_supported to see if it need do
+	 * corresponding initialization.
+	 * IP checks con->support to see if it need disable ras.
+	 */
+	if (!amdgpu_ras_is_feature_allowed(adev, head))
+		return 0;
+
+	if (enable) {
+		if (!obj) {
+			obj = amdgpu_ras_create_obj(adev, head);
+			if (!obj)
+				return -EINVAL;
+		} else {
+			/* In case we create obj somewhere else */
+			get_obj(obj);
+		}
+		con->features |= BIT(head->block);
+	} else {
+		if (obj && amdgpu_ras_is_feature_enabled(adev, head)) {
+			con->features &= ~BIT(head->block);
+			put_obj(obj);
+		}
+	}
+
+	return 0;
+}
+
+/* wrapper of psp_ras_enable_features */
+int amdgpu_ras_feature_enable(struct amdgpu_device *adev,
+		struct ras_common_if *head, bool enable)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	union ta_ras_cmd_input *info;
+	int ret;
+
+	if (!con)
+		return -EINVAL;
+
+	/* For non-gfx ip, do not enable ras feature if it is not allowed */
+	/* For gfx ip, regardless of feature support status, */
+	/* Force issue enable or disable ras feature commands */
+	if (head->block != AMDGPU_RAS_BLOCK__GFX &&
+	    !amdgpu_ras_is_feature_allowed(adev, head))
+		return 0;
+
+	/* Only enable gfx ras feature from host side */
+	if (head->block == AMDGPU_RAS_BLOCK__GFX &&
+	    !amdgpu_sriov_vf(adev) &&
+	    !amdgpu_ras_intr_triggered()) {
+		info = kzalloc(sizeof(union ta_ras_cmd_input), GFP_KERNEL);
+		if (!info)
+			return -ENOMEM;
+
+		if (!enable) {
+			info->disable_features = (struct ta_ras_disable_features_input) {
+				.block_id =  amdgpu_ras_block_to_ta(head->block),
+				.error_type = amdgpu_ras_error_to_ta(head->type),
+			};
+		} else {
+			info->enable_features = (struct ta_ras_enable_features_input) {
+				.block_id =  amdgpu_ras_block_to_ta(head->block),
+				.error_type = amdgpu_ras_error_to_ta(head->type),
+			};
+		}
+
+		ret = psp_ras_enable_features(&adev->psp, info, enable);
+		if (ret) {
+			dev_err(adev->dev, "ras %s %s failed poison:%d ret:%d\n",
+				enable ? "enable":"disable",
+				get_ras_block_str(head),
+				amdgpu_ras_is_poison_mode_supported(adev), ret);
+			kfree(info);
+			return ret;
+		}
+
+		kfree(info);
+	}
+
+	/* setup the obj */
+	__amdgpu_ras_feature_enable(adev, head, enable);
+
+	return 0;
+}
+
+/* Only used in device probe stage and called only once. */
+int amdgpu_ras_feature_enable_on_boot(struct amdgpu_device *adev,
+		struct ras_common_if *head, bool enable)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	int ret;
+
+	if (!con)
+		return -EINVAL;
+
+	if (con->flags & AMDGPU_RAS_FLAG_INIT_BY_VBIOS) {
+		if (enable) {
+			/* There is no harm to issue a ras TA cmd regardless of
+			 * the currecnt ras state.
+			 * If current state == target state, it will do nothing
+			 * But sometimes it requests driver to reset and repost
+			 * with error code -EAGAIN.
+			 */
+			ret = amdgpu_ras_feature_enable(adev, head, 1);
+			/* With old ras TA, we might fail to enable ras.
+			 * Log it and just setup the object.
+			 * TODO need remove this WA in the future.
+			 */
+			if (ret == -EINVAL) {
+				ret = __amdgpu_ras_feature_enable(adev, head, 1);
+				if (!ret)
+					dev_info(adev->dev,
+						"RAS INFO: %s setup object\n",
+						get_ras_block_str(head));
+			}
+		} else {
+			/* setup the object then issue a ras TA disable cmd.*/
+			ret = __amdgpu_ras_feature_enable(adev, head, 1);
+			if (ret)
+				return ret;
+
+			/* gfx block ras disable cmd must send to ras-ta */
+			if (head->block == AMDGPU_RAS_BLOCK__GFX)
+				con->features |= BIT(head->block);
+
+			ret = amdgpu_ras_feature_enable(adev, head, 0);
+
+			/* clean gfx block ras features flag */
+			if (adev->ras_enabled && head->block == AMDGPU_RAS_BLOCK__GFX)
+				con->features &= ~BIT(head->block);
+		}
+	} else
+		ret = amdgpu_ras_feature_enable(adev, head, enable);
+
+	return ret;
+}
+
+static int amdgpu_ras_disable_all_features(struct amdgpu_device *adev,
+		bool bypass)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_manager *obj, *tmp;
+
+	list_for_each_entry_safe(obj, tmp, &con->head, node) {
+		/* bypass psp.
+		 * aka just release the obj and corresponding flags
+		 */
+		if (bypass) {
+			if (__amdgpu_ras_feature_enable(adev, &obj->head, 0))
+				break;
+		} else {
+			if (amdgpu_ras_feature_enable(adev, &obj->head, 0))
+				break;
+		}
+	}
+
+	return con->features;
+}
+
+static int amdgpu_ras_enable_all_features(struct amdgpu_device *adev,
+		bool bypass)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	int i;
+	const enum amdgpu_ras_error_type default_ras_type = AMDGPU_RAS_ERROR__NONE;
+
+	for (i = 0; i < AMDGPU_RAS_BLOCK_COUNT; i++) {
+		struct ras_common_if head = {
+			.block = i,
+			.type = default_ras_type,
+			.sub_block_index = 0,
+		};
+
+		if (i == AMDGPU_RAS_BLOCK__MCA)
+			continue;
+
+		if (bypass) {
+			/*
+			 * bypass psp. vbios enable ras for us.
+			 * so just create the obj
+			 */
+			if (__amdgpu_ras_feature_enable(adev, &head, 1))
+				break;
+		} else {
+			if (amdgpu_ras_feature_enable(adev, &head, 1))
+				break;
+		}
+	}
+
+	for (i = 0; i < AMDGPU_RAS_MCA_BLOCK_COUNT; i++) {
+		struct ras_common_if head = {
+			.block = AMDGPU_RAS_BLOCK__MCA,
+			.type = default_ras_type,
+			.sub_block_index = i,
+		};
+
+		if (bypass) {
+			/*
+			 * bypass psp. vbios enable ras for us.
+			 * so just create the obj
+			 */
+			if (__amdgpu_ras_feature_enable(adev, &head, 1))
+				break;
+		} else {
+			if (amdgpu_ras_feature_enable(adev, &head, 1))
+				break;
+		}
+	}
+
+	return con->features;
+}
+/* feature ctl end */
+
+static int amdgpu_ras_block_match_default(struct amdgpu_ras_block_object *block_obj,
+		enum amdgpu_ras_block block)
+{
+	if (!block_obj)
+		return -EINVAL;
+
+	if (block_obj->ras_comm.block == block)
+		return 0;
+
+	return -EINVAL;
+}
+
+static struct amdgpu_ras_block_object *amdgpu_ras_get_ras_block(struct amdgpu_device *adev,
+					enum amdgpu_ras_block block, uint32_t sub_block_index)
+{
+	struct amdgpu_ras_block_list *node, *tmp;
+	struct amdgpu_ras_block_object *obj;
+
+	if (block >= AMDGPU_RAS_BLOCK__LAST)
+		return NULL;
+
+	list_for_each_entry_safe(node, tmp, &adev->ras_list, node) {
+		if (!node->ras_obj) {
+			dev_warn(adev->dev, "Warning: abnormal ras list node.\n");
+			continue;
+		}
+
+		obj = node->ras_obj;
+		if (obj->ras_block_match) {
+			if (obj->ras_block_match(obj, block, sub_block_index) == 0)
+				return obj;
+		} else {
+			if (amdgpu_ras_block_match_default(obj, block) == 0)
+				return obj;
+		}
+	}
+
+	return NULL;
+}
+
+static void amdgpu_ras_get_ecc_info(struct amdgpu_device *adev, struct ras_err_data *err_data)
+{
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+	int ret = 0;
+
+	/*
+	 * choosing right query method according to
+	 * whether smu support query error information
+	 */
+	ret = amdgpu_dpm_get_ecc_info(adev, (void *)&(ras->umc_ecc));
+	if (ret == -EOPNOTSUPP) {
+		if (adev->umc.ras && adev->umc.ras->ras_block.hw_ops &&
+			adev->umc.ras->ras_block.hw_ops->query_ras_error_count)
+			adev->umc.ras->ras_block.hw_ops->query_ras_error_count(adev, err_data);
+
+		/* umc query_ras_error_address is also responsible for clearing
+		 * error status
+		 */
+		if (adev->umc.ras && adev->umc.ras->ras_block.hw_ops &&
+		    adev->umc.ras->ras_block.hw_ops->query_ras_error_address)
+			adev->umc.ras->ras_block.hw_ops->query_ras_error_address(adev, err_data);
+	} else if (!ret) {
+		if (adev->umc.ras &&
+			adev->umc.ras->ecc_info_query_ras_error_count)
+			adev->umc.ras->ecc_info_query_ras_error_count(adev, err_data);
+
+		if (adev->umc.ras &&
+			adev->umc.ras->ecc_info_query_ras_error_address)
+			adev->umc.ras->ecc_info_query_ras_error_address(adev, err_data);
+	}
+}
+
+static void amdgpu_ras_error_print_error_data(struct amdgpu_device *adev,
+					      struct ras_manager *ras_mgr,
+					      struct ras_err_data *err_data,
+					      struct ras_query_context *qctx,
+					      const char *blk_name,
+					      bool is_ue,
+					      bool is_de)
+{
+	struct amdgpu_smuio_mcm_config_info *mcm_info;
+	struct ras_err_node *err_node;
+	struct ras_err_info *err_info;
+	u64 event_id = qctx->evid.event_id;
+
+	if (is_ue) {
+		for_each_ras_error(err_node, err_data) {
+			err_info = &err_node->err_info;
+			mcm_info = &err_info->mcm_info;
+			if (err_info->ue_count) {
+				RAS_EVENT_LOG(adev, event_id, "socket: %d, die: %d, "
+					      "%lld new uncorrectable hardware errors detected in %s block\n",
+					      mcm_info->socket_id,
+					      mcm_info->die_id,
+					      err_info->ue_count,
+					      blk_name);
+			}
+		}
+
+		for_each_ras_error(err_node, &ras_mgr->err_data) {
+			err_info = &err_node->err_info;
+			mcm_info = &err_info->mcm_info;
+			RAS_EVENT_LOG(adev, event_id, "socket: %d, die: %d, "
+				      "%lld uncorrectable hardware errors detected in total in %s block\n",
+				      mcm_info->socket_id, mcm_info->die_id, err_info->ue_count, blk_name);
+		}
+
+	} else {
+		if (is_de) {
+			for_each_ras_error(err_node, err_data) {
+				err_info = &err_node->err_info;
+				mcm_info = &err_info->mcm_info;
+				if (err_info->de_count) {
+					RAS_EVENT_LOG(adev, event_id, "socket: %d, die: %d, "
+						      "%lld new deferred hardware errors detected in %s block\n",
+						      mcm_info->socket_id,
+						      mcm_info->die_id,
+						      err_info->de_count,
+						      blk_name);
+				}
+			}
+
+			for_each_ras_error(err_node, &ras_mgr->err_data) {
+				err_info = &err_node->err_info;
+				mcm_info = &err_info->mcm_info;
+				RAS_EVENT_LOG(adev, event_id, "socket: %d, die: %d, "
+					      "%lld deferred hardware errors detected in total in %s block\n",
+					      mcm_info->socket_id, mcm_info->die_id,
+					      err_info->de_count, blk_name);
+			}
+		} else {
+			if (adev->debug_disable_ce_logs)
+				return;
+
+			for_each_ras_error(err_node, err_data) {
+				err_info = &err_node->err_info;
+				mcm_info = &err_info->mcm_info;
+				if (err_info->ce_count) {
+					RAS_EVENT_LOG(adev, event_id, "socket: %d, die: %d, "
+						      "%lld new correctable hardware errors detected in %s block\n",
+						      mcm_info->socket_id,
+						      mcm_info->die_id,
+						      err_info->ce_count,
+						      blk_name);
+				}
+			}
+
+			for_each_ras_error(err_node, &ras_mgr->err_data) {
+				err_info = &err_node->err_info;
+				mcm_info = &err_info->mcm_info;
+				RAS_EVENT_LOG(adev, event_id, "socket: %d, die: %d, "
+					      "%lld correctable hardware errors detected in total in %s block\n",
+					      mcm_info->socket_id, mcm_info->die_id,
+					      err_info->ce_count, blk_name);
+			}
+		}
+	}
+}
+
+static inline bool err_data_has_source_info(struct ras_err_data *data)
+{
+	return !list_empty(&data->err_node_list);
+}
+
+static void amdgpu_ras_error_generate_report(struct amdgpu_device *adev,
+					     struct ras_query_if *query_if,
+					     struct ras_err_data *err_data,
+					     struct ras_query_context *qctx)
+{
+	struct ras_manager *ras_mgr = amdgpu_ras_find_obj(adev, &query_if->head);
+	const char *blk_name = get_ras_block_str(&query_if->head);
+	u64 event_id = qctx->evid.event_id;
+
+	if (err_data->ce_count) {
+		if (err_data_has_source_info(err_data)) {
+			amdgpu_ras_error_print_error_data(adev, ras_mgr, err_data, qctx,
+							  blk_name, false, false);
+		} else if (!adev->aid_mask &&
+			   adev->smuio.funcs &&
+			   adev->smuio.funcs->get_socket_id &&
+			   adev->smuio.funcs->get_die_id) {
+			RAS_EVENT_LOG(adev, event_id, "socket: %d, die: %d "
+				      "%ld correctable hardware errors "
+				      "detected in %s block\n",
+				      adev->smuio.funcs->get_socket_id(adev),
+				      adev->smuio.funcs->get_die_id(adev),
+				      ras_mgr->err_data.ce_count,
+				      blk_name);
+		} else {
+			RAS_EVENT_LOG(adev, event_id, "%ld correctable hardware errors "
+				      "detected in %s block\n",
+				      ras_mgr->err_data.ce_count,
+				      blk_name);
+		}
+	}
+
+	if (err_data->ue_count) {
+		if (err_data_has_source_info(err_data)) {
+			amdgpu_ras_error_print_error_data(adev, ras_mgr, err_data, qctx,
+							  blk_name, true, false);
+		} else if (!adev->aid_mask &&
+			   adev->smuio.funcs &&
+			   adev->smuio.funcs->get_socket_id &&
+			   adev->smuio.funcs->get_die_id) {
+			RAS_EVENT_LOG(adev, event_id, "socket: %d, die: %d "
+				      "%ld uncorrectable hardware errors "
+				      "detected in %s block\n",
+				      adev->smuio.funcs->get_socket_id(adev),
+				      adev->smuio.funcs->get_die_id(adev),
+				      ras_mgr->err_data.ue_count,
+				      blk_name);
+		} else {
+			RAS_EVENT_LOG(adev, event_id, "%ld uncorrectable hardware errors "
+				      "detected in %s block\n",
+				      ras_mgr->err_data.ue_count,
+				      blk_name);
+		}
+	}
+
+	if (err_data->de_count) {
+		if (err_data_has_source_info(err_data)) {
+			amdgpu_ras_error_print_error_data(adev, ras_mgr, err_data, qctx,
+							  blk_name, false, true);
+		} else if (!adev->aid_mask &&
+			   adev->smuio.funcs &&
+			   adev->smuio.funcs->get_socket_id &&
+			   adev->smuio.funcs->get_die_id) {
+			RAS_EVENT_LOG(adev, event_id, "socket: %d, die: %d "
+				      "%ld deferred hardware errors "
+				      "detected in %s block\n",
+				      adev->smuio.funcs->get_socket_id(adev),
+				      adev->smuio.funcs->get_die_id(adev),
+				      ras_mgr->err_data.de_count,
+				      blk_name);
+		} else {
+			RAS_EVENT_LOG(adev, event_id, "%ld deferred hardware errors "
+				      "detected in %s block\n",
+				      ras_mgr->err_data.de_count,
+				      blk_name);
+		}
+	}
+}
+
+static void amdgpu_ras_virt_error_generate_report(struct amdgpu_device *adev,
+						  struct ras_query_if *query_if,
+						  struct ras_err_data *err_data,
+						  struct ras_query_context *qctx)
+{
+	unsigned long new_ue, new_ce, new_de;
+	struct ras_manager *obj = amdgpu_ras_find_obj(adev, &query_if->head);
+	const char *blk_name = get_ras_block_str(&query_if->head);
+	u64 event_id = qctx->evid.event_id;
+
+	new_ce = err_data->ce_count - obj->err_data.ce_count;
+	new_ue = err_data->ue_count - obj->err_data.ue_count;
+	new_de = err_data->de_count - obj->err_data.de_count;
+
+	if (new_ce) {
+		RAS_EVENT_LOG(adev, event_id, "%lu correctable hardware errors "
+			      "detected in %s block\n",
+			      new_ce,
+			      blk_name);
+	}
+
+	if (new_ue) {
+		RAS_EVENT_LOG(adev, event_id, "%lu uncorrectable hardware errors "
+			      "detected in %s block\n",
+			      new_ue,
+			      blk_name);
+	}
+
+	if (new_de) {
+		RAS_EVENT_LOG(adev, event_id, "%lu deferred hardware errors "
+			      "detected in %s block\n",
+			      new_de,
+			      blk_name);
+	}
+}
+
+static void amdgpu_rasmgr_error_data_statistic_update(struct ras_manager *obj, struct ras_err_data *err_data)
+{
+	struct ras_err_node *err_node;
+	struct ras_err_info *err_info;
+
+	if (err_data_has_source_info(err_data)) {
+		for_each_ras_error(err_node, err_data) {
+			err_info = &err_node->err_info;
+			amdgpu_ras_error_statistic_de_count(&obj->err_data,
+					&err_info->mcm_info, err_info->de_count);
+			amdgpu_ras_error_statistic_ce_count(&obj->err_data,
+					&err_info->mcm_info, err_info->ce_count);
+			amdgpu_ras_error_statistic_ue_count(&obj->err_data,
+					&err_info->mcm_info, err_info->ue_count);
+		}
+	} else {
+		/* for legacy asic path which doesn't has error source info */
+		obj->err_data.ue_count += err_data->ue_count;
+		obj->err_data.ce_count += err_data->ce_count;
+		obj->err_data.de_count += err_data->de_count;
+	}
+}
+
+static void amdgpu_ras_mgr_virt_error_data_statistics_update(struct ras_manager *obj,
+							     struct ras_err_data *err_data)
+{
+	/* Host reports absolute counts */
+	obj->err_data.ue_count = err_data->ue_count;
+	obj->err_data.ce_count = err_data->ce_count;
+	obj->err_data.de_count = err_data->de_count;
+}
+
+static struct ras_manager *get_ras_manager(struct amdgpu_device *adev, enum amdgpu_ras_block blk)
+{
+	struct ras_common_if head;
+
+	memset(&head, 0, sizeof(head));
+	head.block = blk;
+
+	return amdgpu_ras_find_obj(adev, &head);
+}
+
+int amdgpu_ras_bind_aca(struct amdgpu_device *adev, enum amdgpu_ras_block blk,
+			const struct aca_info *aca_info, void *data)
+{
+	struct ras_manager *obj;
+
+	/* in resume phase, no need to create aca fs node */
+	if (adev->in_suspend || amdgpu_reset_in_recovery(adev))
+		return 0;
+
+	obj = get_ras_manager(adev, blk);
+	if (!obj)
+		return -EINVAL;
+
+	return amdgpu_aca_add_handle(adev, &obj->aca_handle, ras_block_str(blk), aca_info, data);
+}
+
+int amdgpu_ras_unbind_aca(struct amdgpu_device *adev, enum amdgpu_ras_block blk)
+{
+	struct ras_manager *obj;
+
+	obj = get_ras_manager(adev, blk);
+	if (!obj)
+		return -EINVAL;
+
+	amdgpu_aca_remove_handle(&obj->aca_handle);
+
+	return 0;
+}
+
+static int amdgpu_aca_log_ras_error_data(struct amdgpu_device *adev, enum amdgpu_ras_block blk,
+					 enum aca_error_type type, struct ras_err_data *err_data,
+					 struct ras_query_context *qctx)
+{
+	struct ras_manager *obj;
+
+	obj = get_ras_manager(adev, blk);
+	if (!obj)
+		return -EINVAL;
+
+	return amdgpu_aca_get_error_data(adev, &obj->aca_handle, type, err_data, qctx);
+}
+
+ssize_t amdgpu_ras_aca_sysfs_read(struct device *dev, struct device_attribute *attr,
+				  struct aca_handle *handle, char *buf, void *data)
+{
+	struct ras_manager *obj = container_of(handle, struct ras_manager, aca_handle);
+	struct ras_query_if info = {
+		.head = obj->head,
+	};
+
+	if (!amdgpu_ras_get_error_query_ready(obj->adev))
+		return sysfs_emit(buf, "Query currently inaccessible\n");
+
+	if (amdgpu_ras_query_error_status(obj->adev, &info))
+		return -EINVAL;
+
+	return sysfs_emit(buf, "%s: %lu\n%s: %lu\n%s: %lu\n", "ue", info.ue_count,
+			  "ce", info.ce_count, "de", info.de_count);
+}
+
+static int amdgpu_ras_query_error_status_helper(struct amdgpu_device *adev,
+						struct ras_query_if *info,
+						struct ras_err_data *err_data,
+						struct ras_query_context *qctx,
+						unsigned int error_query_mode)
+{
+	enum amdgpu_ras_block blk = info ? info->head.block : AMDGPU_RAS_BLOCK_COUNT;
+	struct amdgpu_ras_block_object *block_obj = NULL;
+	int ret;
+
+	if (blk == AMDGPU_RAS_BLOCK_COUNT)
+		return -EINVAL;
+
+	if (error_query_mode == AMDGPU_RAS_INVALID_ERROR_QUERY)
+		return -EINVAL;
+
+	if (error_query_mode == AMDGPU_RAS_VIRT_ERROR_COUNT_QUERY) {
+		return amdgpu_virt_req_ras_err_count(adev, blk, err_data);
+	} else if (error_query_mode == AMDGPU_RAS_DIRECT_ERROR_QUERY) {
+		if (info->head.block == AMDGPU_RAS_BLOCK__UMC) {
+			amdgpu_ras_get_ecc_info(adev, err_data);
+		} else {
+			block_obj = amdgpu_ras_get_ras_block(adev, info->head.block, 0);
+			if (!block_obj || !block_obj->hw_ops) {
+				dev_dbg_once(adev->dev, "%s doesn't config RAS function\n",
+					     get_ras_block_str(&info->head));
+				return -EINVAL;
+			}
+
+			if (block_obj->hw_ops->query_ras_error_count)
+				block_obj->hw_ops->query_ras_error_count(adev, err_data);
+
+			if ((info->head.block == AMDGPU_RAS_BLOCK__SDMA) ||
+			    (info->head.block == AMDGPU_RAS_BLOCK__GFX) ||
+			    (info->head.block == AMDGPU_RAS_BLOCK__MMHUB)) {
+				if (block_obj->hw_ops->query_ras_error_status)
+					block_obj->hw_ops->query_ras_error_status(adev);
+			}
+		}
+	} else {
+		if (amdgpu_aca_is_enabled(adev)) {
+			ret = amdgpu_aca_log_ras_error_data(adev, blk, ACA_ERROR_TYPE_UE, err_data, qctx);
+			if (ret)
+				return ret;
+
+			ret = amdgpu_aca_log_ras_error_data(adev, blk, ACA_ERROR_TYPE_CE, err_data, qctx);
+			if (ret)
+				return ret;
+
+			ret = amdgpu_aca_log_ras_error_data(adev, blk, ACA_ERROR_TYPE_DEFERRED, err_data, qctx);
+			if (ret)
+				return ret;
+		} else {
+			/* FIXME: add code to check return value later */
+			amdgpu_mca_smu_log_ras_error(adev, blk, AMDGPU_MCA_ERROR_TYPE_UE, err_data, qctx);
+			amdgpu_mca_smu_log_ras_error(adev, blk, AMDGPU_MCA_ERROR_TYPE_CE, err_data, qctx);
+		}
+	}
+
+	return 0;
+}
+
+/* query/inject/cure begin */
+static int amdgpu_ras_query_error_status_with_event(struct amdgpu_device *adev,
+						    struct ras_query_if *info,
+						    enum ras_event_type type)
+{
+	struct ras_manager *obj = amdgpu_ras_find_obj(adev, &info->head);
+	struct ras_err_data err_data;
+	struct ras_query_context qctx;
+	unsigned int error_query_mode;
+	int ret;
+
+	if (!obj)
+		return -EINVAL;
+
+	ret = amdgpu_ras_error_data_init(&err_data);
+	if (ret)
+		return ret;
+
+	if (!amdgpu_ras_get_error_query_mode(adev, &error_query_mode))
+		return -EINVAL;
+
+	memset(&qctx, 0, sizeof(qctx));
+	qctx.evid.type = type;
+	qctx.evid.event_id = amdgpu_ras_acquire_event_id(adev, type);
+
+	if (!down_read_trylock(&adev->reset_domain->sem)) {
+		ret = -EIO;
+		goto out_fini_err_data;
+	}
+
+	ret = amdgpu_ras_query_error_status_helper(adev, info,
+						   &err_data,
+						   &qctx,
+						   error_query_mode);
+	up_read(&adev->reset_domain->sem);
+	if (ret)
+		goto out_fini_err_data;
+
+	if (error_query_mode != AMDGPU_RAS_VIRT_ERROR_COUNT_QUERY) {
+		amdgpu_rasmgr_error_data_statistic_update(obj, &err_data);
+		amdgpu_ras_error_generate_report(adev, info, &err_data, &qctx);
+	} else {
+		/* Host provides absolute error counts. First generate the report
+		 * using the previous VF internal count against new host count.
+		 * Then Update VF internal count.
+		 */
+		amdgpu_ras_virt_error_generate_report(adev, info, &err_data, &qctx);
+		amdgpu_ras_mgr_virt_error_data_statistics_update(obj, &err_data);
+	}
+
+	info->ue_count = obj->err_data.ue_count;
+	info->ce_count = obj->err_data.ce_count;
+	info->de_count = obj->err_data.de_count;
+
+out_fini_err_data:
+	amdgpu_ras_error_data_fini(&err_data);
+
+	return ret;
+}
+
+int amdgpu_ras_query_error_status(struct amdgpu_device *adev, struct ras_query_if *info)
+{
+	return amdgpu_ras_query_error_status_with_event(adev, info, RAS_EVENT_TYPE_INVALID);
+}
+
+int amdgpu_ras_reset_error_count(struct amdgpu_device *adev,
+		enum amdgpu_ras_block block)
+{
+	struct amdgpu_ras_block_object *block_obj = amdgpu_ras_get_ras_block(adev, block, 0);
+	const struct amdgpu_mca_smu_funcs *mca_funcs = adev->mca.mca_funcs;
+	const struct aca_smu_funcs *smu_funcs = adev->aca.smu_funcs;
+
+	if (!block_obj || !block_obj->hw_ops) {
+		dev_dbg_once(adev->dev, "%s doesn't config RAS function\n",
+				ras_block_str(block));
+		return -EOPNOTSUPP;
+	}
+
+	if (!amdgpu_ras_is_supported(adev, block) ||
+	    !amdgpu_ras_get_aca_debug_mode(adev))
+		return -EOPNOTSUPP;
+
+	if (amdgpu_sriov_vf(adev))
+		return -EOPNOTSUPP;
+
+	/* skip ras error reset in gpu reset */
+	if ((amdgpu_in_reset(adev) || amdgpu_ras_in_recovery(adev)) &&
+	    ((smu_funcs && smu_funcs->set_debug_mode) ||
+	     (mca_funcs && mca_funcs->mca_set_debug_mode)))
+		return -EOPNOTSUPP;
+
+	if (block_obj->hw_ops->reset_ras_error_count)
+		block_obj->hw_ops->reset_ras_error_count(adev);
+
+	return 0;
+}
+
+int amdgpu_ras_reset_error_status(struct amdgpu_device *adev,
+		enum amdgpu_ras_block block)
+{
+	struct amdgpu_ras_block_object *block_obj = amdgpu_ras_get_ras_block(adev, block, 0);
+
+	if (amdgpu_ras_reset_error_count(adev, block) == -EOPNOTSUPP)
+		return 0;
+
+	if ((block == AMDGPU_RAS_BLOCK__GFX) ||
+	    (block == AMDGPU_RAS_BLOCK__MMHUB)) {
+		if (block_obj->hw_ops->reset_ras_error_status)
+			block_obj->hw_ops->reset_ras_error_status(adev);
+	}
+
+	return 0;
+}
+
+/* wrapper of psp_ras_trigger_error */
+int amdgpu_ras_error_inject(struct amdgpu_device *adev,
+		struct ras_inject_if *info)
+{
+	struct ras_manager *obj = amdgpu_ras_find_obj(adev, &info->head);
+	struct ta_ras_trigger_error_input block_info = {
+		.block_id =  amdgpu_ras_block_to_ta(info->head.block),
+		.inject_error_type = amdgpu_ras_error_to_ta(info->head.type),
+		.sub_block_index = info->head.sub_block_index,
+		.address = info->address,
+		.value = info->value,
+	};
+	int ret = -EINVAL;
+	struct amdgpu_ras_block_object *block_obj = amdgpu_ras_get_ras_block(adev,
+							info->head.block,
+							info->head.sub_block_index);
+
+	/* inject on guest isn't allowed, return success directly */
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	if (!obj)
+		return -EINVAL;
+
+	if (!block_obj || !block_obj->hw_ops)	{
+		dev_dbg_once(adev->dev, "%s doesn't config RAS function\n",
+			     get_ras_block_str(&info->head));
+		return -EINVAL;
+	}
+
+	/* Calculate XGMI relative offset */
+	if (adev->gmc.xgmi.num_physical_nodes > 1 &&
+	    info->head.block != AMDGPU_RAS_BLOCK__GFX) {
+		block_info.address =
+			amdgpu_xgmi_get_relative_phy_addr(adev,
+							  block_info.address);
+	}
+
+	if (block_obj->hw_ops->ras_error_inject) {
+		if (info->head.block == AMDGPU_RAS_BLOCK__GFX)
+			ret = block_obj->hw_ops->ras_error_inject(adev, info, info->instance_mask);
+		else /* Special ras_error_inject is defined (e.g: xgmi) */
+			ret = block_obj->hw_ops->ras_error_inject(adev, &block_info,
+						info->instance_mask);
+	} else {
+		/* default path */
+		ret = psp_ras_trigger_error(&adev->psp, &block_info, info->instance_mask);
+	}
+
+	if (ret)
+		dev_err(adev->dev, "ras inject %s failed %d\n",
+			get_ras_block_str(&info->head), ret);
+
+	return ret;
+}
+
+/**
+ * amdgpu_ras_query_error_count_helper -- Get error counter for specific IP
+ * @adev: pointer to AMD GPU device
+ * @ce_count: pointer to an integer to be set to the count of correctible errors.
+ * @ue_count: pointer to an integer to be set to the count of uncorrectible errors.
+ * @query_info: pointer to ras_query_if
+ *
+ * Return 0 for query success or do nothing, otherwise return an error
+ * on failures
+ */
+static int amdgpu_ras_query_error_count_helper(struct amdgpu_device *adev,
+					       unsigned long *ce_count,
+					       unsigned long *ue_count,
+					       struct ras_query_if *query_info)
+{
+	int ret;
+
+	if (!query_info)
+		/* do nothing if query_info is not specified */
+		return 0;
+
+	ret = amdgpu_ras_query_error_status(adev, query_info);
+	if (ret)
+		return ret;
+
+	*ce_count += query_info->ce_count;
+	*ue_count += query_info->ue_count;
+
+	/* some hardware/IP supports read to clear
+	 * no need to explictly reset the err status after the query call */
+	if (amdgpu_ip_version(adev, MP0_HWIP, 0) != IP_VERSION(11, 0, 2) &&
+	    amdgpu_ip_version(adev, MP0_HWIP, 0) != IP_VERSION(11, 0, 4)) {
+		if (amdgpu_ras_reset_error_status(adev, query_info->head.block))
+			dev_warn(adev->dev,
+				 "Failed to reset error counter and error status\n");
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_ras_query_error_count -- Get error counts of all IPs or specific IP
+ * @adev: pointer to AMD GPU device
+ * @ce_count: pointer to an integer to be set to the count of correctible errors.
+ * @ue_count: pointer to an integer to be set to the count of uncorrectible
+ * errors.
+ * @query_info: pointer to ras_query_if if the query request is only for
+ * specific ip block; if info is NULL, then the qurey request is for
+ * all the ip blocks that support query ras error counters/status
+ *
+ * If set, @ce_count or @ue_count, count and return the corresponding
+ * error counts in those integer pointers. Return 0 if the device
+ * supports RAS. Return -EOPNOTSUPP if the device doesn't support RAS.
+ */
+int amdgpu_ras_query_error_count(struct amdgpu_device *adev,
+				 unsigned long *ce_count,
+				 unsigned long *ue_count,
+				 struct ras_query_if *query_info)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_manager *obj;
+	unsigned long ce, ue;
+	int ret;
+
+	if (!adev->ras_enabled || !con)
+		return -EOPNOTSUPP;
+
+	/* Don't count since no reporting.
+	 */
+	if (!ce_count && !ue_count)
+		return 0;
+
+	ce = 0;
+	ue = 0;
+	if (!query_info) {
+		/* query all the ip blocks that support ras query interface */
+		list_for_each_entry(obj, &con->head, node) {
+			struct ras_query_if info = {
+				.head = obj->head,
+			};
+
+			ret = amdgpu_ras_query_error_count_helper(adev, &ce, &ue, &info);
+		}
+	} else {
+		/* query specific ip block */
+		ret = amdgpu_ras_query_error_count_helper(adev, &ce, &ue, query_info);
+	}
+
+	if (ret)
+		return ret;
+
+	if (ce_count)
+		*ce_count = ce;
+
+	if (ue_count)
+		*ue_count = ue;
+
+	return 0;
+}
+/* query/inject/cure end */
+
+
+/* sysfs begin */
+
+static int amdgpu_ras_badpages_read(struct amdgpu_device *adev,
+		struct ras_badpage **bps, unsigned int *count);
+
+static char *amdgpu_ras_badpage_flags_str(unsigned int flags)
+{
+	switch (flags) {
+	case AMDGPU_RAS_RETIRE_PAGE_RESERVED:
+		return "R";
+	case AMDGPU_RAS_RETIRE_PAGE_PENDING:
+		return "P";
+	case AMDGPU_RAS_RETIRE_PAGE_FAULT:
+	default:
+		return "F";
+	}
+}
+
+/**
+ * DOC: AMDGPU RAS sysfs gpu_vram_bad_pages Interface
+ *
+ * It allows user to read the bad pages of vram on the gpu through
+ * /sys/class/drm/card[0/1/2...]/device/ras/gpu_vram_bad_pages
+ *
+ * It outputs multiple lines, and each line stands for one gpu page.
+ *
+ * The format of one line is below,
+ * gpu pfn : gpu page size : flags
+ *
+ * gpu pfn and gpu page size are printed in hex format.
+ * flags can be one of below character,
+ *
+ * R: reserved, this gpu page is reserved and not able to use.
+ *
+ * P: pending for reserve, this gpu page is marked as bad, will be reserved
+ * in next window of page_reserve.
+ *
+ * F: unable to reserve. this gpu page can't be reserved due to some reasons.
+ *
+ * Examples:
+ *
+ * .. code-block:: bash
+ *
+ *	0x00000001 : 0x00001000 : R
+ *	0x00000002 : 0x00001000 : P
+ *
+ */
+
+static ssize_t amdgpu_ras_sysfs_badpages_read(struct file *f,
+		struct kobject *kobj, const struct bin_attribute *attr,
+		char *buf, loff_t ppos, size_t count)
+{
+	struct amdgpu_ras *con =
+		container_of(attr, struct amdgpu_ras, badpages_attr);
+	struct amdgpu_device *adev = con->adev;
+	const unsigned int element_size =
+		sizeof("0xabcdabcd : 0x12345678 : R\n") - 1;
+	unsigned int start = div64_ul(ppos + element_size - 1, element_size);
+	unsigned int end = div64_ul(ppos + count - 1, element_size);
+	ssize_t s = 0;
+	struct ras_badpage *bps = NULL;
+	unsigned int bps_count = 0;
+
+	memset(buf, 0, count);
+
+	if (amdgpu_ras_badpages_read(adev, &bps, &bps_count))
+		return 0;
+
+	for (; start < end && start < bps_count; start++)
+		s += scnprintf(&buf[s], element_size + 1,
+				"0x%08x : 0x%08x : %1s\n",
+				bps[start].bp,
+				bps[start].size,
+				amdgpu_ras_badpage_flags_str(bps[start].flags));
+
+	kfree(bps);
+
+	return s;
+}
+
+static ssize_t amdgpu_ras_sysfs_features_read(struct device *dev,
+		struct device_attribute *attr, char *buf)
+{
+	struct amdgpu_ras *con =
+		container_of(attr, struct amdgpu_ras, features_attr);
+
+	return sysfs_emit(buf, "feature mask: 0x%x\n", con->features);
+}
+
+static ssize_t amdgpu_ras_sysfs_version_show(struct device *dev,
+		struct device_attribute *attr, char *buf)
+{
+	struct amdgpu_ras *con =
+		container_of(attr, struct amdgpu_ras, version_attr);
+	return sysfs_emit(buf, "table version: 0x%x\n", con->eeprom_control.tbl_hdr.version);
+}
+
+static ssize_t amdgpu_ras_sysfs_schema_show(struct device *dev,
+		struct device_attribute *attr, char *buf)
+{
+	struct amdgpu_ras *con =
+		container_of(attr, struct amdgpu_ras, schema_attr);
+	return sysfs_emit(buf, "schema: 0x%x\n", con->schema);
+}
+
+static struct {
+	enum ras_event_type type;
+	const char *name;
+} dump_event[] = {
+	{RAS_EVENT_TYPE_FATAL, "Fatal Error"},
+	{RAS_EVENT_TYPE_POISON_CREATION, "Poison Creation"},
+	{RAS_EVENT_TYPE_POISON_CONSUMPTION, "Poison Consumption"},
+};
+
+static ssize_t amdgpu_ras_sysfs_event_state_show(struct device *dev,
+						 struct device_attribute *attr, char *buf)
+{
+	struct amdgpu_ras *con =
+		container_of(attr, struct amdgpu_ras, event_state_attr);
+	struct ras_event_manager *event_mgr = con->event_mgr;
+	struct ras_event_state *event_state;
+	int i, size = 0;
+
+	if (!event_mgr)
+		return -EINVAL;
+
+	size += sysfs_emit_at(buf, size, "current seqno: %llu\n", atomic64_read(&event_mgr->seqno));
+	for (i = 0; i < ARRAY_SIZE(dump_event); i++) {
+		event_state = &event_mgr->event_state[dump_event[i].type];
+		size += sysfs_emit_at(buf, size, "%s: count:%llu, last_seqno:%llu\n",
+				      dump_event[i].name,
+				      atomic64_read(&event_state->count),
+				      event_state->last_seqno);
+	}
+
+	return (ssize_t)size;
+}
+
+static void amdgpu_ras_sysfs_remove_bad_page_node(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+
+	if (adev->dev->kobj.sd)
+		sysfs_remove_file_from_group(&adev->dev->kobj,
+				&con->badpages_attr.attr,
+				RAS_FS_NAME);
+}
+
+static int amdgpu_ras_sysfs_remove_dev_attr_node(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct attribute *attrs[] = {
+		&con->features_attr.attr,
+		&con->version_attr.attr,
+		&con->schema_attr.attr,
+		&con->event_state_attr.attr,
+		NULL
+	};
+	struct attribute_group group = {
+		.name = RAS_FS_NAME,
+		.attrs = attrs,
+	};
+
+	if (adev->dev->kobj.sd)
+		sysfs_remove_group(&adev->dev->kobj, &group);
+
+	return 0;
+}
+
+int amdgpu_ras_sysfs_create(struct amdgpu_device *adev,
+		struct ras_common_if *head)
+{
+	struct ras_manager *obj = amdgpu_ras_find_obj(adev, head);
+
+	if (amdgpu_aca_is_enabled(adev))
+		return 0;
+
+	if (!obj || obj->attr_inuse)
+		return -EINVAL;
+
+	if (amdgpu_sriov_vf(adev) && !amdgpu_virt_ras_telemetry_block_en(adev, head->block))
+		return 0;
+
+	get_obj(obj);
+
+	snprintf(obj->fs_data.sysfs_name, sizeof(obj->fs_data.sysfs_name),
+		"%s_err_count", head->name);
+
+	obj->sysfs_attr = (struct device_attribute){
+		.attr = {
+			.name = obj->fs_data.sysfs_name,
+			.mode = S_IRUGO,
+		},
+			.show = amdgpu_ras_sysfs_read,
+	};
+	sysfs_attr_init(&obj->sysfs_attr.attr);
+
+	if (sysfs_add_file_to_group(&adev->dev->kobj,
+				&obj->sysfs_attr.attr,
+				RAS_FS_NAME)) {
+		put_obj(obj);
+		return -EINVAL;
+	}
+
+	obj->attr_inuse = 1;
+
+	return 0;
+}
+
+int amdgpu_ras_sysfs_remove(struct amdgpu_device *adev,
+		struct ras_common_if *head)
+{
+	struct ras_manager *obj = amdgpu_ras_find_obj(adev, head);
+
+	if (amdgpu_aca_is_enabled(adev))
+		return 0;
+
+	if (!obj || !obj->attr_inuse)
+		return -EINVAL;
+
+	if (adev->dev->kobj.sd)
+		sysfs_remove_file_from_group(&adev->dev->kobj,
+				&obj->sysfs_attr.attr,
+				RAS_FS_NAME);
+	obj->attr_inuse = 0;
+	put_obj(obj);
+
+	return 0;
+}
+
+static int amdgpu_ras_sysfs_remove_all(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_manager *obj, *tmp;
+
+	list_for_each_entry_safe(obj, tmp, &con->head, node) {
+		amdgpu_ras_sysfs_remove(adev, &obj->head);
+	}
+
+	if (amdgpu_bad_page_threshold != 0)
+		amdgpu_ras_sysfs_remove_bad_page_node(adev);
+
+	amdgpu_ras_sysfs_remove_dev_attr_node(adev);
+
+	return 0;
+}
+/* sysfs end */
+
+/**
+ * DOC: AMDGPU RAS Reboot Behavior for Unrecoverable Errors
+ *
+ * Normally when there is an uncorrectable error, the driver will reset
+ * the GPU to recover.  However, in the event of an unrecoverable error,
+ * the driver provides an interface to reboot the system automatically
+ * in that event.
+ *
+ * The following file in debugfs provides that interface:
+ * /sys/kernel/debug/dri/[0/1/2...]/ras/auto_reboot
+ *
+ * Usage:
+ *
+ * .. code-block:: bash
+ *
+ *	echo true > .../ras/auto_reboot
+ *
+ */
+/* debugfs begin */
+static struct dentry *amdgpu_ras_debugfs_create_ctrl_node(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct amdgpu_ras_eeprom_control *eeprom = &con->eeprom_control;
+	struct drm_minor  *minor = adev_to_drm(adev)->primary;
+	struct dentry     *dir;
+
+	dir = debugfs_create_dir(RAS_FS_NAME, minor->debugfs_root);
+	debugfs_create_file("ras_ctrl", S_IWUGO | S_IRUGO, dir, adev,
+			    &amdgpu_ras_debugfs_ctrl_ops);
+	debugfs_create_file("ras_eeprom_reset", S_IWUGO | S_IRUGO, dir, adev,
+			    &amdgpu_ras_debugfs_eeprom_ops);
+	debugfs_create_u32("bad_page_cnt_threshold", 0444, dir,
+			   &con->bad_page_cnt_threshold);
+	debugfs_create_u32("ras_num_recs", 0444, dir, &eeprom->ras_num_recs);
+	debugfs_create_x32("ras_hw_enabled", 0444, dir, &adev->ras_hw_enabled);
+	debugfs_create_x32("ras_enabled", 0444, dir, &adev->ras_enabled);
+	debugfs_create_file("ras_eeprom_size", S_IRUGO, dir, adev,
+			    &amdgpu_ras_debugfs_eeprom_size_ops);
+	con->de_ras_eeprom_table = debugfs_create_file("ras_eeprom_table",
+						       S_IRUGO, dir, adev,
+						       &amdgpu_ras_debugfs_eeprom_table_ops);
+	amdgpu_ras_debugfs_set_ret_size(&con->eeprom_control);
+
+	/*
+	 * After one uncorrectable error happens, usually GPU recovery will
+	 * be scheduled. But due to the known problem in GPU recovery failing
+	 * to bring GPU back, below interface provides one direct way to
+	 * user to reboot system automatically in such case within
+	 * ERREVENT_ATHUB_INTERRUPT generated. Normal GPU recovery routine
+	 * will never be called.
+	 */
+	debugfs_create_bool("auto_reboot", S_IWUGO | S_IRUGO, dir, &con->reboot);
+
+	/*
+	 * User could set this not to clean up hardware's error count register
+	 * of RAS IPs during ras recovery.
+	 */
+	debugfs_create_bool("disable_ras_err_cnt_harvest", 0644, dir,
+			    &con->disable_ras_err_cnt_harvest);
+	return dir;
+}
+
+static void amdgpu_ras_debugfs_create(struct amdgpu_device *adev,
+				      struct ras_fs_if *head,
+				      struct dentry *dir)
+{
+	struct ras_manager *obj = amdgpu_ras_find_obj(adev, &head->head);
+
+	if (!obj || !dir)
+		return;
+
+	get_obj(obj);
+
+	memcpy(obj->fs_data.debugfs_name,
+			head->debugfs_name,
+			sizeof(obj->fs_data.debugfs_name));
+
+	debugfs_create_file(obj->fs_data.debugfs_name, S_IWUGO | S_IRUGO, dir,
+			    obj, &amdgpu_ras_debugfs_ops);
+}
+
+static bool amdgpu_ras_aca_is_supported(struct amdgpu_device *adev)
+{
+	bool ret;
+
+	switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) {
+	case IP_VERSION(13, 0, 6):
+	case IP_VERSION(13, 0, 12):
+	case IP_VERSION(13, 0, 14):
+		ret = true;
+		break;
+	default:
+		ret = false;
+		break;
+	}
+
+	return ret;
+}
+
+void amdgpu_ras_debugfs_create_all(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct dentry *dir;
+	struct ras_manager *obj;
+	struct ras_fs_if fs_info;
+
+	/*
+	 * it won't be called in resume path, no need to check
+	 * suspend and gpu reset status
+	 */
+	if (!IS_ENABLED(CONFIG_DEBUG_FS) || !con)
+		return;
+
+	dir = amdgpu_ras_debugfs_create_ctrl_node(adev);
+
+	list_for_each_entry(obj, &con->head, node) {
+		if (amdgpu_ras_is_supported(adev, obj->head.block) &&
+			(obj->attr_inuse == 1)) {
+			sprintf(fs_info.debugfs_name, "%s_err_inject",
+					get_ras_block_str(&obj->head));
+			fs_info.head = obj->head;
+			amdgpu_ras_debugfs_create(adev, &fs_info, dir);
+		}
+	}
+
+	if (amdgpu_ras_aca_is_supported(adev)) {
+		if (amdgpu_aca_is_enabled(adev))
+			amdgpu_aca_smu_debugfs_init(adev, dir);
+		else
+			amdgpu_mca_smu_debugfs_init(adev, dir);
+	}
+}
+
+/* debugfs end */
+
+/* ras fs */
+static const BIN_ATTR(gpu_vram_bad_pages, S_IRUGO,
+		      amdgpu_ras_sysfs_badpages_read, NULL, 0);
+static DEVICE_ATTR(features, S_IRUGO,
+		amdgpu_ras_sysfs_features_read, NULL);
+static DEVICE_ATTR(version, 0444,
+		amdgpu_ras_sysfs_version_show, NULL);
+static DEVICE_ATTR(schema, 0444,
+		amdgpu_ras_sysfs_schema_show, NULL);
+static DEVICE_ATTR(event_state, 0444,
+		   amdgpu_ras_sysfs_event_state_show, NULL);
+static int amdgpu_ras_fs_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct attribute_group group = {
+		.name = RAS_FS_NAME,
+	};
+	struct attribute *attrs[] = {
+		&con->features_attr.attr,
+		&con->version_attr.attr,
+		&con->schema_attr.attr,
+		&con->event_state_attr.attr,
+		NULL
+	};
+	const struct bin_attribute *bin_attrs[] = {
+		NULL,
+		NULL,
+	};
+	int r;
+
+	group.attrs = attrs;
+
+	/* add features entry */
+	con->features_attr = dev_attr_features;
+	sysfs_attr_init(attrs[0]);
+
+	/* add version entry */
+	con->version_attr = dev_attr_version;
+	sysfs_attr_init(attrs[1]);
+
+	/* add schema entry */
+	con->schema_attr = dev_attr_schema;
+	sysfs_attr_init(attrs[2]);
+
+	/* add event_state entry */
+	con->event_state_attr = dev_attr_event_state;
+	sysfs_attr_init(attrs[3]);
+
+	if (amdgpu_bad_page_threshold != 0) {
+		/* add bad_page_features entry */
+		con->badpages_attr = bin_attr_gpu_vram_bad_pages;
+		sysfs_bin_attr_init(&con->badpages_attr);
+		bin_attrs[0] = &con->badpages_attr;
+		group.bin_attrs = bin_attrs;
+	}
+
+	r = sysfs_create_group(&adev->dev->kobj, &group);
+	if (r)
+		dev_err(adev->dev, "Failed to create RAS sysfs group!");
+
+	return 0;
+}
+
+static int amdgpu_ras_fs_fini(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_manager *con_obj, *ip_obj, *tmp;
+
+	if (IS_ENABLED(CONFIG_DEBUG_FS)) {
+		list_for_each_entry_safe(con_obj, tmp, &con->head, node) {
+			ip_obj = amdgpu_ras_find_obj(adev, &con_obj->head);
+			if (ip_obj)
+				put_obj(ip_obj);
+		}
+	}
+
+	amdgpu_ras_sysfs_remove_all(adev);
+	return 0;
+}
+/* ras fs end */
+
+/* ih begin */
+
+/* For the hardware that cannot enable bif ring for both ras_controller_irq
+ * and ras_err_evnet_athub_irq ih cookies, the driver has to poll status
+ * register to check whether the interrupt is triggered or not, and properly
+ * ack the interrupt if it is there
+ */
+void amdgpu_ras_interrupt_fatal_error_handler(struct amdgpu_device *adev)
+{
+	/* Fatal error events are handled on host side */
+	if (amdgpu_sriov_vf(adev))
+		return;
+	/*
+	 * If the current interrupt is caused by a non-fatal RAS error, skip
+	 * check for fatal error. For fatal errors, FED status of all devices
+	 * in XGMI hive gets set when the first device gets fatal error
+	 * interrupt. The error gets propagated to other devices as well, so
+	 * make sure to ack the interrupt regardless of FED status.
+	 */
+	if (!amdgpu_ras_get_fed_status(adev) &&
+	    amdgpu_ras_is_err_state(adev, AMDGPU_RAS_BLOCK__ANY))
+		return;
+
+	if (adev->nbio.ras &&
+	    adev->nbio.ras->handle_ras_controller_intr_no_bifring)
+		adev->nbio.ras->handle_ras_controller_intr_no_bifring(adev);
+
+	if (adev->nbio.ras &&
+	    adev->nbio.ras->handle_ras_err_event_athub_intr_no_bifring)
+		adev->nbio.ras->handle_ras_err_event_athub_intr_no_bifring(adev);
+}
+
+static void amdgpu_ras_interrupt_poison_consumption_handler(struct ras_manager *obj,
+				struct amdgpu_iv_entry *entry)
+{
+	bool poison_stat = false;
+	struct amdgpu_device *adev = obj->adev;
+	struct amdgpu_ras_block_object *block_obj =
+		amdgpu_ras_get_ras_block(adev, obj->head.block, 0);
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	enum ras_event_type type = RAS_EVENT_TYPE_POISON_CONSUMPTION;
+	u64 event_id;
+	int ret;
+
+	if (!block_obj || !con)
+		return;
+
+	ret = amdgpu_ras_mark_ras_event(adev, type);
+	if (ret)
+		return;
+
+	amdgpu_ras_set_err_poison(adev, block_obj->ras_comm.block);
+	/* both query_poison_status and handle_poison_consumption are optional,
+	 * but at least one of them should be implemented if we need poison
+	 * consumption handler
+	 */
+	if (block_obj->hw_ops && block_obj->hw_ops->query_poison_status) {
+		poison_stat = block_obj->hw_ops->query_poison_status(adev);
+		if (!poison_stat) {
+			/* Not poison consumption interrupt, no need to handle it */
+			dev_info(adev->dev, "No RAS poison status in %s poison IH.\n",
+					block_obj->ras_comm.name);
+
+			return;
+		}
+	}
+
+	amdgpu_umc_poison_handler(adev, obj->head.block, 0);
+
+	if (block_obj->hw_ops && block_obj->hw_ops->handle_poison_consumption)
+		poison_stat = block_obj->hw_ops->handle_poison_consumption(adev);
+
+	/* gpu reset is fallback for failed and default cases.
+	 * For RMA case, amdgpu_umc_poison_handler will handle gpu reset.
+	 */
+	if (poison_stat && !amdgpu_ras_is_rma(adev)) {
+		event_id = amdgpu_ras_acquire_event_id(adev, type);
+		RAS_EVENT_LOG(adev, event_id,
+			      "GPU reset for %s RAS poison consumption is issued!\n",
+			      block_obj->ras_comm.name);
+		amdgpu_ras_reset_gpu(adev);
+	}
+
+	if (!poison_stat)
+		amdgpu_gfx_poison_consumption_handler(adev, entry);
+}
+
+static void amdgpu_ras_interrupt_poison_creation_handler(struct ras_manager *obj,
+				struct amdgpu_iv_entry *entry)
+{
+	struct amdgpu_device *adev = obj->adev;
+	enum ras_event_type type = RAS_EVENT_TYPE_POISON_CREATION;
+	u64 event_id;
+	int ret;
+
+	ret = amdgpu_ras_mark_ras_event(adev, type);
+	if (ret)
+		return;
+
+	event_id = amdgpu_ras_acquire_event_id(adev, type);
+	RAS_EVENT_LOG(adev, event_id, "Poison is created\n");
+
+	if (amdgpu_ip_version(obj->adev, UMC_HWIP, 0) >= IP_VERSION(12, 0, 0)) {
+		struct amdgpu_ras *con = amdgpu_ras_get_context(obj->adev);
+
+		atomic_inc(&con->page_retirement_req_cnt);
+		atomic_inc(&con->poison_creation_count);
+
+		wake_up(&con->page_retirement_wq);
+	}
+}
+
+static void amdgpu_ras_interrupt_umc_handler(struct ras_manager *obj,
+				struct amdgpu_iv_entry *entry)
+{
+	struct ras_ih_data *data = &obj->ih_data;
+	struct ras_err_data err_data;
+	int ret;
+
+	if (!data->cb)
+		return;
+
+	ret = amdgpu_ras_error_data_init(&err_data);
+	if (ret)
+		return;
+
+	/* Let IP handle its data, maybe we need get the output
+	 * from the callback to update the error type/count, etc
+	 */
+	amdgpu_ras_set_fed(obj->adev, true);
+	ret = data->cb(obj->adev, &err_data, entry);
+	/* ue will trigger an interrupt, and in that case
+	 * we need do a reset to recovery the whole system.
+	 * But leave IP do that recovery, here we just dispatch
+	 * the error.
+	 */
+	if (ret == AMDGPU_RAS_SUCCESS) {
+		/* these counts could be left as 0 if
+		 * some blocks do not count error number
+		 */
+		obj->err_data.ue_count += err_data.ue_count;
+		obj->err_data.ce_count += err_data.ce_count;
+		obj->err_data.de_count += err_data.de_count;
+	}
+
+	amdgpu_ras_error_data_fini(&err_data);
+}
+
+static void amdgpu_ras_interrupt_handler(struct ras_manager *obj)
+{
+	struct ras_ih_data *data = &obj->ih_data;
+	struct amdgpu_iv_entry entry;
+
+	while (data->rptr != data->wptr) {
+		rmb();
+		memcpy(&entry, &data->ring[data->rptr],
+				data->element_size);
+
+		wmb();
+		data->rptr = (data->aligned_element_size +
+				data->rptr) % data->ring_size;
+
+		if (amdgpu_ras_is_poison_mode_supported(obj->adev)) {
+			if (obj->head.block == AMDGPU_RAS_BLOCK__UMC)
+				amdgpu_ras_interrupt_poison_creation_handler(obj, &entry);
+			else
+				amdgpu_ras_interrupt_poison_consumption_handler(obj, &entry);
+		} else {
+			if (obj->head.block == AMDGPU_RAS_BLOCK__UMC)
+				amdgpu_ras_interrupt_umc_handler(obj, &entry);
+			else
+				dev_warn(obj->adev->dev,
+					"No RAS interrupt handler for non-UMC block with poison disabled.\n");
+		}
+	}
+}
+
+static void amdgpu_ras_interrupt_process_handler(struct work_struct *work)
+{
+	struct ras_ih_data *data =
+		container_of(work, struct ras_ih_data, ih_work);
+	struct ras_manager *obj =
+		container_of(data, struct ras_manager, ih_data);
+
+	amdgpu_ras_interrupt_handler(obj);
+}
+
+int amdgpu_ras_interrupt_dispatch(struct amdgpu_device *adev,
+		struct ras_dispatch_if *info)
+{
+	struct ras_manager *obj;
+	struct ras_ih_data *data;
+
+	obj = amdgpu_ras_find_obj(adev, &info->head);
+	if (!obj)
+		return -EINVAL;
+
+	data = &obj->ih_data;
+
+	if (data->inuse == 0)
+		return 0;
+
+	/* Might be overflow... */
+	memcpy(&data->ring[data->wptr], info->entry,
+			data->element_size);
+
+	wmb();
+	data->wptr = (data->aligned_element_size +
+			data->wptr) % data->ring_size;
+
+	schedule_work(&data->ih_work);
+
+	return 0;
+}
+
+int amdgpu_ras_interrupt_remove_handler(struct amdgpu_device *adev,
+		struct ras_common_if *head)
+{
+	struct ras_manager *obj = amdgpu_ras_find_obj(adev, head);
+	struct ras_ih_data *data;
+
+	if (!obj)
+		return -EINVAL;
+
+	data = &obj->ih_data;
+	if (data->inuse == 0)
+		return 0;
+
+	cancel_work_sync(&data->ih_work);
+
+	kfree(data->ring);
+	memset(data, 0, sizeof(*data));
+	put_obj(obj);
+
+	return 0;
+}
+
+int amdgpu_ras_interrupt_add_handler(struct amdgpu_device *adev,
+		struct ras_common_if *head)
+{
+	struct ras_manager *obj = amdgpu_ras_find_obj(adev, head);
+	struct ras_ih_data *data;
+	struct amdgpu_ras_block_object *ras_obj;
+
+	if (!obj) {
+		/* in case we registe the IH before enable ras feature */
+		obj = amdgpu_ras_create_obj(adev, head);
+		if (!obj)
+			return -EINVAL;
+	} else
+		get_obj(obj);
+
+	ras_obj = container_of(head, struct amdgpu_ras_block_object, ras_comm);
+
+	data = &obj->ih_data;
+	/* add the callback.etc */
+	*data = (struct ras_ih_data) {
+		.inuse = 0,
+		.cb = ras_obj->ras_cb,
+		.element_size = sizeof(struct amdgpu_iv_entry),
+		.rptr = 0,
+		.wptr = 0,
+	};
+
+	INIT_WORK(&data->ih_work, amdgpu_ras_interrupt_process_handler);
+
+	data->aligned_element_size = ALIGN(data->element_size, 8);
+	/* the ring can store 64 iv entries. */
+	data->ring_size = 64 * data->aligned_element_size;
+	data->ring = kmalloc(data->ring_size, GFP_KERNEL);
+	if (!data->ring) {
+		put_obj(obj);
+		return -ENOMEM;
+	}
+
+	/* IH is ready */
+	data->inuse = 1;
+
+	return 0;
+}
+
+static int amdgpu_ras_interrupt_remove_all(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_manager *obj, *tmp;
+
+	list_for_each_entry_safe(obj, tmp, &con->head, node) {
+		amdgpu_ras_interrupt_remove_handler(adev, &obj->head);
+	}
+
+	return 0;
+}
+/* ih end */
+
+/* traversal all IPs except NBIO to query error counter */
+static void amdgpu_ras_log_on_err_counter(struct amdgpu_device *adev, enum ras_event_type type)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_manager *obj;
+
+	if (!adev->ras_enabled || !con)
+		return;
+
+	list_for_each_entry(obj, &con->head, node) {
+		struct ras_query_if info = {
+			.head = obj->head,
+		};
+
+		/*
+		 * PCIE_BIF IP has one different isr by ras controller
+		 * interrupt, the specific ras counter query will be
+		 * done in that isr. So skip such block from common
+		 * sync flood interrupt isr calling.
+		 */
+		if (info.head.block == AMDGPU_RAS_BLOCK__PCIE_BIF)
+			continue;
+
+		/*
+		 * this is a workaround for aldebaran, skip send msg to
+		 * smu to get ecc_info table due to smu handle get ecc
+		 * info table failed temporarily.
+		 * should be removed until smu fix handle ecc_info table.
+		 */
+		if ((info.head.block == AMDGPU_RAS_BLOCK__UMC) &&
+		    (amdgpu_ip_version(adev, MP1_HWIP, 0) ==
+		     IP_VERSION(13, 0, 2)))
+			continue;
+
+		amdgpu_ras_query_error_status_with_event(adev, &info, type);
+
+		if (amdgpu_ip_version(adev, MP0_HWIP, 0) !=
+			    IP_VERSION(11, 0, 2) &&
+		    amdgpu_ip_version(adev, MP0_HWIP, 0) !=
+			    IP_VERSION(11, 0, 4) &&
+		    amdgpu_ip_version(adev, MP0_HWIP, 0) !=
+			    IP_VERSION(13, 0, 0)) {
+			if (amdgpu_ras_reset_error_status(adev, info.head.block))
+				dev_warn(adev->dev, "Failed to reset error counter and error status");
+		}
+	}
+}
+
+/* Parse RdRspStatus and WrRspStatus */
+static void amdgpu_ras_error_status_query(struct amdgpu_device *adev,
+					  struct ras_query_if *info)
+{
+	struct amdgpu_ras_block_object *block_obj;
+	/*
+	 * Only two block need to query read/write
+	 * RspStatus at current state
+	 */
+	if ((info->head.block != AMDGPU_RAS_BLOCK__GFX) &&
+		(info->head.block != AMDGPU_RAS_BLOCK__MMHUB))
+		return;
+
+	block_obj = amdgpu_ras_get_ras_block(adev,
+					info->head.block,
+					info->head.sub_block_index);
+
+	if (!block_obj || !block_obj->hw_ops) {
+		dev_dbg_once(adev->dev, "%s doesn't config RAS function\n",
+			     get_ras_block_str(&info->head));
+		return;
+	}
+
+	if (block_obj->hw_ops->query_ras_error_status)
+		block_obj->hw_ops->query_ras_error_status(adev);
+
+}
+
+static void amdgpu_ras_query_err_status(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_manager *obj;
+
+	if (!adev->ras_enabled || !con)
+		return;
+
+	list_for_each_entry(obj, &con->head, node) {
+		struct ras_query_if info = {
+			.head = obj->head,
+		};
+
+		amdgpu_ras_error_status_query(adev, &info);
+	}
+}
+
+/* recovery begin */
+
+/* return 0 on success.
+ * caller need free bps.
+ */
+static int amdgpu_ras_badpages_read(struct amdgpu_device *adev,
+		struct ras_badpage **bps, unsigned int *count)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_err_handler_data *data;
+	int i = 0;
+	int ret = 0, status;
+
+	if (!con || !con->eh_data || !bps || !count)
+		return -EINVAL;
+
+	mutex_lock(&con->recovery_lock);
+	data = con->eh_data;
+	if (!data || data->count == 0) {
+		*bps = NULL;
+		ret = -EINVAL;
+		goto out;
+	}
+
+	*bps = kmalloc_array(data->count, sizeof(struct ras_badpage), GFP_KERNEL);
+	if (!*bps) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	for (; i < data->count; i++) {
+		if (!data->bps[i].ts)
+			continue;
+
+		(*bps)[i] = (struct ras_badpage){
+			.bp = data->bps[i].retired_page,
+			.size = AMDGPU_GPU_PAGE_SIZE,
+			.flags = AMDGPU_RAS_RETIRE_PAGE_RESERVED,
+		};
+
+		if (amdgpu_ras_check_critical_address(adev,
+			data->bps[i].retired_page << AMDGPU_GPU_PAGE_SHIFT))
+			continue;
+
+		status = amdgpu_vram_mgr_query_page_status(&adev->mman.vram_mgr,
+				data->bps[i].retired_page << AMDGPU_GPU_PAGE_SHIFT);
+		if (status == -EBUSY)
+			(*bps)[i].flags = AMDGPU_RAS_RETIRE_PAGE_PENDING;
+		else if (status == -ENOENT)
+			(*bps)[i].flags = AMDGPU_RAS_RETIRE_PAGE_FAULT;
+	}
+
+	*count = con->bad_page_num;
+out:
+	mutex_unlock(&con->recovery_lock);
+	return ret;
+}
+
+static void amdgpu_ras_set_fed_all(struct amdgpu_device *adev,
+				   struct amdgpu_hive_info *hive, bool status)
+{
+	struct amdgpu_device *tmp_adev;
+
+	if (hive) {
+		list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head)
+			amdgpu_ras_set_fed(tmp_adev, status);
+	} else {
+		amdgpu_ras_set_fed(adev, status);
+	}
+}
+
+bool amdgpu_ras_in_recovery(struct amdgpu_device *adev)
+{
+	struct amdgpu_hive_info *hive = amdgpu_get_xgmi_hive(adev);
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+	int hive_ras_recovery = 0;
+
+	if (hive) {
+		hive_ras_recovery = atomic_read(&hive->ras_recovery);
+		amdgpu_put_xgmi_hive(hive);
+	}
+
+	if (ras && (atomic_read(&ras->in_recovery) || hive_ras_recovery))
+		return true;
+
+	return false;
+}
+
+static enum ras_event_type amdgpu_ras_get_fatal_error_event(struct amdgpu_device *adev)
+{
+	if (amdgpu_ras_intr_triggered())
+		return RAS_EVENT_TYPE_FATAL;
+	else
+		return RAS_EVENT_TYPE_POISON_CONSUMPTION;
+}
+
+static void amdgpu_ras_do_recovery(struct work_struct *work)
+{
+	struct amdgpu_ras *ras =
+		container_of(work, struct amdgpu_ras, recovery_work);
+	struct amdgpu_device *remote_adev = NULL;
+	struct amdgpu_device *adev = ras->adev;
+	struct list_head device_list, *device_list_handle =  NULL;
+	struct amdgpu_hive_info *hive = amdgpu_get_xgmi_hive(adev);
+	unsigned int error_query_mode;
+	enum ras_event_type type;
+
+	if (hive) {
+		atomic_set(&hive->ras_recovery, 1);
+
+		/* If any device which is part of the hive received RAS fatal
+		 * error interrupt, set fatal error status on all. This
+		 * condition will need a recovery, and flag will be cleared
+		 * as part of recovery.
+		 */
+		list_for_each_entry(remote_adev, &hive->device_list,
+				    gmc.xgmi.head)
+			if (amdgpu_ras_get_fed_status(remote_adev)) {
+				amdgpu_ras_set_fed_all(adev, hive, true);
+				break;
+			}
+	}
+	if (!ras->disable_ras_err_cnt_harvest) {
+
+		/* Build list of devices to query RAS related errors */
+		if  (hive && adev->gmc.xgmi.num_physical_nodes > 1) {
+			device_list_handle = &hive->device_list;
+		} else {
+			INIT_LIST_HEAD(&device_list);
+			list_add_tail(&adev->gmc.xgmi.head, &device_list);
+			device_list_handle = &device_list;
+		}
+
+		if (amdgpu_ras_get_error_query_mode(adev, &error_query_mode)) {
+			if (error_query_mode == AMDGPU_RAS_FIRMWARE_ERROR_QUERY) {
+				/* wait 500ms to ensure pmfw polling mca bank info done */
+				msleep(500);
+			}
+		}
+
+		type = amdgpu_ras_get_fatal_error_event(adev);
+		list_for_each_entry(remote_adev,
+				device_list_handle, gmc.xgmi.head) {
+			amdgpu_ras_query_err_status(remote_adev);
+			amdgpu_ras_log_on_err_counter(remote_adev, type);
+		}
+
+	}
+
+	if (amdgpu_device_should_recover_gpu(ras->adev)) {
+		struct amdgpu_reset_context reset_context;
+		memset(&reset_context, 0, sizeof(reset_context));
+
+		reset_context.method = AMD_RESET_METHOD_NONE;
+		reset_context.reset_req_dev = adev;
+		reset_context.src = AMDGPU_RESET_SRC_RAS;
+		set_bit(AMDGPU_SKIP_COREDUMP, &reset_context.flags);
+
+		/* Perform full reset in fatal error mode */
+		if (!amdgpu_ras_is_poison_mode_supported(ras->adev))
+			set_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
+		else {
+			clear_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
+
+			if (ras->gpu_reset_flags & AMDGPU_RAS_GPU_RESET_MODE2_RESET) {
+				ras->gpu_reset_flags &= ~AMDGPU_RAS_GPU_RESET_MODE2_RESET;
+				reset_context.method = AMD_RESET_METHOD_MODE2;
+			}
+
+			/* Fatal error occurs in poison mode, mode1 reset is used to
+			 * recover gpu.
+			 */
+			if (ras->gpu_reset_flags & AMDGPU_RAS_GPU_RESET_MODE1_RESET) {
+				ras->gpu_reset_flags &= ~AMDGPU_RAS_GPU_RESET_MODE1_RESET;
+				set_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
+
+				psp_fatal_error_recovery_quirk(&adev->psp);
+			}
+		}
+
+		amdgpu_device_gpu_recover(ras->adev, NULL, &reset_context);
+	}
+	atomic_set(&ras->in_recovery, 0);
+	if (hive) {
+		atomic_set(&hive->ras_recovery, 0);
+		amdgpu_put_xgmi_hive(hive);
+	}
+}
+
+/* alloc/realloc bps array */
+static int amdgpu_ras_realloc_eh_data_space(struct amdgpu_device *adev,
+		struct ras_err_handler_data *data, int pages)
+{
+	unsigned int old_space = data->count + data->space_left;
+	unsigned int new_space = old_space + pages;
+	unsigned int align_space = ALIGN(new_space, 512);
+	void *bps = kmalloc_array(align_space, sizeof(*data->bps), GFP_KERNEL);
+
+	if (!bps) {
+		return -ENOMEM;
+	}
+
+	if (data->bps) {
+		memcpy(bps, data->bps,
+				data->count * sizeof(*data->bps));
+		kfree(data->bps);
+	}
+
+	data->bps = bps;
+	data->space_left += align_space - old_space;
+	return 0;
+}
+
+static int amdgpu_ras_mca2pa_by_idx(struct amdgpu_device *adev,
+			struct eeprom_table_record *bps,
+			struct ras_err_data *err_data)
+{
+	struct ta_ras_query_address_input addr_in;
+	uint32_t socket = 0;
+	int ret = 0;
+
+	if (adev->smuio.funcs && adev->smuio.funcs->get_socket_id)
+		socket = adev->smuio.funcs->get_socket_id(adev);
+
+	/* reinit err_data */
+	err_data->err_addr_cnt = 0;
+	err_data->err_addr_len = adev->umc.retire_unit;
+
+	memset(&addr_in, 0, sizeof(addr_in));
+	addr_in.ma.err_addr = bps->address;
+	addr_in.ma.socket_id = socket;
+	addr_in.ma.ch_inst = bps->mem_channel;
+	/* tell RAS TA the node instance is not used */
+	addr_in.ma.node_inst = TA_RAS_INV_NODE;
+
+	if (adev->umc.ras && adev->umc.ras->convert_ras_err_addr)
+		ret = adev->umc.ras->convert_ras_err_addr(adev, err_data,
+				&addr_in, NULL, false);
+
+	return ret;
+}
+
+static int amdgpu_ras_mca2pa(struct amdgpu_device *adev,
+			struct eeprom_table_record *bps,
+			struct ras_err_data *err_data)
+{
+	struct ta_ras_query_address_input addr_in;
+	uint32_t die_id, socket = 0;
+
+	if (adev->smuio.funcs && adev->smuio.funcs->get_socket_id)
+		socket = adev->smuio.funcs->get_socket_id(adev);
+
+	/* although die id is gotten from PA in nps1 mode, the id is
+	 * fitable for any nps mode
+	 */
+	if (adev->umc.ras && adev->umc.ras->get_die_id_from_pa)
+		die_id = adev->umc.ras->get_die_id_from_pa(adev, bps->address,
+					bps->retired_page << AMDGPU_GPU_PAGE_SHIFT);
+	else
+		return -EINVAL;
+
+	/* reinit err_data */
+	err_data->err_addr_cnt = 0;
+	err_data->err_addr_len = adev->umc.retire_unit;
+
+	memset(&addr_in, 0, sizeof(addr_in));
+	addr_in.ma.err_addr = bps->address;
+	addr_in.ma.ch_inst = bps->mem_channel;
+	addr_in.ma.umc_inst = bps->mcumc_id;
+	addr_in.ma.node_inst = die_id;
+	addr_in.ma.socket_id = socket;
+
+	if (adev->umc.ras && adev->umc.ras->convert_ras_err_addr)
+		return adev->umc.ras->convert_ras_err_addr(adev, err_data,
+					&addr_in, NULL, false);
+	else
+		return  -EINVAL;
+}
+
+static int __amdgpu_ras_restore_bad_pages(struct amdgpu_device *adev,
+					struct eeprom_table_record *bps, int count)
+{
+	int j;
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_err_handler_data *data = con->eh_data;
+
+	for (j = 0; j < count; j++) {
+		if (amdgpu_ras_check_bad_page_unlock(con,
+			bps[j].retired_page << AMDGPU_GPU_PAGE_SHIFT)) {
+			data->count++;
+			data->space_left--;
+			continue;
+		}
+
+		if (!data->space_left &&
+		    amdgpu_ras_realloc_eh_data_space(adev, data, 256)) {
+			return -ENOMEM;
+		}
+
+		amdgpu_ras_reserve_page(adev, bps[j].retired_page);
+
+		memcpy(&data->bps[data->count], &(bps[j]),
+				sizeof(struct eeprom_table_record));
+		data->count++;
+		data->space_left--;
+		con->bad_page_num++;
+	}
+
+	return 0;
+}
+
+static int __amdgpu_ras_convert_rec_array_from_rom(struct amdgpu_device *adev,
+				struct eeprom_table_record *bps, struct ras_err_data *err_data,
+				enum amdgpu_memory_partition nps)
+{
+	int i = 0;
+	enum amdgpu_memory_partition save_nps;
+
+	save_nps = (bps[0].retired_page >> UMC_NPS_SHIFT) & UMC_NPS_MASK;
+
+	/*old asics just have pa in eeprom*/
+	if (IP_VERSION_MAJ(amdgpu_ip_version(adev, UMC_HWIP, 0)) < 12) {
+		memcpy(err_data->err_addr, bps,
+			sizeof(struct eeprom_table_record) * adev->umc.retire_unit);
+		goto out;
+	}
+
+	for (i = 0; i < adev->umc.retire_unit; i++)
+		bps[i].retired_page &= ~(UMC_NPS_MASK << UMC_NPS_SHIFT);
+
+	if (save_nps) {
+		if (save_nps == nps) {
+			if (amdgpu_umc_pages_in_a_row(adev, err_data,
+					bps[0].retired_page << AMDGPU_GPU_PAGE_SHIFT))
+				return -EINVAL;
+			for (i = 0; i < adev->umc.retire_unit; i++) {
+				err_data->err_addr[i].address = bps[0].address;
+				err_data->err_addr[i].mem_channel = bps[0].mem_channel;
+				err_data->err_addr[i].bank = bps[0].bank;
+				err_data->err_addr[i].err_type = bps[0].err_type;
+				err_data->err_addr[i].mcumc_id = bps[0].mcumc_id;
+			}
+		} else {
+			if (amdgpu_ras_mca2pa_by_idx(adev, &bps[0], err_data))
+				return -EINVAL;
+		}
+	} else {
+		if (bps[0].address == 0) {
+			/* for specific old eeprom data, mca address is not stored,
+			 * calc it from pa
+			 */
+			if (amdgpu_umc_pa2mca(adev, bps[0].retired_page << AMDGPU_GPU_PAGE_SHIFT,
+				&(bps[0].address), AMDGPU_NPS1_PARTITION_MODE))
+				return -EINVAL;
+		}
+
+		if (amdgpu_ras_mca2pa(adev, &bps[0], err_data)) {
+			if (nps == AMDGPU_NPS1_PARTITION_MODE)
+				memcpy(err_data->err_addr, bps,
+					sizeof(struct eeprom_table_record) * adev->umc.retire_unit);
+			else
+				return -EOPNOTSUPP;
+		}
+	}
+
+out:
+	return __amdgpu_ras_restore_bad_pages(adev, err_data->err_addr, adev->umc.retire_unit);
+}
+
+static int __amdgpu_ras_convert_rec_from_rom(struct amdgpu_device *adev,
+				struct eeprom_table_record *bps, struct ras_err_data *err_data,
+				enum amdgpu_memory_partition nps)
+{
+	int i = 0;
+	enum amdgpu_memory_partition save_nps;
+
+	save_nps = (bps->retired_page >> UMC_NPS_SHIFT) & UMC_NPS_MASK;
+	bps->retired_page &= ~(UMC_NPS_MASK << UMC_NPS_SHIFT);
+
+	if (save_nps == nps) {
+		if (amdgpu_umc_pages_in_a_row(adev, err_data,
+				bps->retired_page << AMDGPU_GPU_PAGE_SHIFT))
+			return -EINVAL;
+		for (i = 0; i < adev->umc.retire_unit; i++) {
+			err_data->err_addr[i].address = bps->address;
+			err_data->err_addr[i].mem_channel = bps->mem_channel;
+			err_data->err_addr[i].bank = bps->bank;
+			err_data->err_addr[i].err_type = bps->err_type;
+			err_data->err_addr[i].mcumc_id = bps->mcumc_id;
+		}
+	} else {
+		if (bps->address) {
+			if (amdgpu_ras_mca2pa_by_idx(adev, bps, err_data))
+				return -EINVAL;
+		} else {
+			/* for specific old eeprom data, mca address is not stored,
+			 * calc it from pa
+			 */
+			if (amdgpu_umc_pa2mca(adev, bps->retired_page << AMDGPU_GPU_PAGE_SHIFT,
+				&(bps->address), AMDGPU_NPS1_PARTITION_MODE))
+				return -EINVAL;
+
+			if (amdgpu_ras_mca2pa(adev, bps, err_data))
+				return -EOPNOTSUPP;
+		}
+	}
+
+	return __amdgpu_ras_restore_bad_pages(adev, err_data->err_addr,
+									adev->umc.retire_unit);
+}
+
+/* it deal with vram only. */
+int amdgpu_ras_add_bad_pages(struct amdgpu_device *adev,
+		struct eeprom_table_record *bps, int pages, bool from_rom)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_err_data err_data;
+	struct amdgpu_ras_eeprom_control *control =
+			&adev->psp.ras_context.ras->eeprom_control;
+	enum amdgpu_memory_partition nps = AMDGPU_NPS1_PARTITION_MODE;
+	int ret = 0;
+	uint32_t i = 0;
+
+	if (!con || !con->eh_data || !bps || pages <= 0)
+		return 0;
+
+	if (from_rom) {
+		err_data.err_addr =
+			kcalloc(adev->umc.retire_unit,
+				sizeof(struct eeprom_table_record), GFP_KERNEL);
+		if (!err_data.err_addr) {
+			dev_warn(adev->dev, "Failed to alloc UMC error address record in mca2pa conversion!\n");
+			return -ENOMEM;
+		}
+
+		if (adev->gmc.gmc_funcs->query_mem_partition_mode)
+			nps = adev->gmc.gmc_funcs->query_mem_partition_mode(adev);
+	}
+
+	mutex_lock(&con->recovery_lock);
+
+	if (from_rom) {
+		/* there is no pa recs in V3, so skip pa recs processing */
+		if (control->tbl_hdr.version < RAS_TABLE_VER_V3) {
+			for (i = 0; i < pages; i++) {
+				if (control->ras_num_recs - i >= adev->umc.retire_unit) {
+					if ((bps[i].address == bps[i + 1].address) &&
+						(bps[i].mem_channel == bps[i + 1].mem_channel)) {
+						/* deal with retire_unit records a time */
+						ret = __amdgpu_ras_convert_rec_array_from_rom(adev,
+										&bps[i], &err_data, nps);
+						if (ret)
+							con->bad_page_num -= adev->umc.retire_unit;
+						i += (adev->umc.retire_unit - 1);
+					} else {
+						break;
+					}
+				} else {
+					break;
+				}
+			}
+		}
+		for (; i < pages; i++) {
+			ret = __amdgpu_ras_convert_rec_from_rom(adev,
+				&bps[i], &err_data, nps);
+			if (ret)
+				con->bad_page_num -= adev->umc.retire_unit;
+		}
+
+		con->eh_data->count_saved = con->eh_data->count;
+	} else {
+		ret = __amdgpu_ras_restore_bad_pages(adev, bps, pages);
+	}
+
+	if (from_rom)
+		kfree(err_data.err_addr);
+	mutex_unlock(&con->recovery_lock);
+
+	return ret;
+}
+
+/*
+ * write error record array to eeprom, the function should be
+ * protected by recovery_lock
+ * new_cnt: new added UE count, excluding reserved bad pages, can be NULL
+ */
+int amdgpu_ras_save_bad_pages(struct amdgpu_device *adev,
+		unsigned long *new_cnt)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_err_handler_data *data;
+	struct amdgpu_ras_eeprom_control *control;
+	int save_count, unit_num, i;
+
+	if (!con || !con->eh_data) {
+		if (new_cnt)
+			*new_cnt = 0;
+
+		return 0;
+	}
+
+	if (!con->eeprom_control.is_eeprom_valid) {
+		dev_warn(adev->dev,
+			"Failed to save EEPROM table data because of EEPROM data corruption!");
+		if (new_cnt)
+			*new_cnt = 0;
+
+		return 0;
+	}
+
+	mutex_lock(&con->recovery_lock);
+	control = &con->eeprom_control;
+	data = con->eh_data;
+	unit_num = data->count / adev->umc.retire_unit - control->ras_num_recs;
+	save_count = con->bad_page_num - control->ras_num_bad_pages;
+	mutex_unlock(&con->recovery_lock);
+
+	if (new_cnt)
+		*new_cnt = unit_num;
+
+	/* only new entries are saved */
+	if (unit_num > 0) {
+		/*old asics only save pa to eeprom like before*/
+		if (IP_VERSION_MAJ(amdgpu_ip_version(adev, UMC_HWIP, 0)) < 12) {
+			if (amdgpu_ras_eeprom_append(control,
+					&data->bps[data->count_saved], unit_num)) {
+				dev_err(adev->dev, "Failed to save EEPROM table data!");
+				return -EIO;
+			}
+		} else {
+			for (i = 0; i < unit_num; i++) {
+				if (amdgpu_ras_eeprom_append(control,
+						&data->bps[data->count_saved +
+						i * adev->umc.retire_unit], 1)) {
+					dev_err(adev->dev, "Failed to save EEPROM table data!");
+					return -EIO;
+				}
+			}
+		}
+
+		dev_info(adev->dev, "Saved %d pages to EEPROM table.\n", save_count);
+		data->count_saved = data->count;
+	}
+
+	return 0;
+}
+
+/*
+ * read error record array in eeprom and reserve enough space for
+ * storing new bad pages
+ */
+static int amdgpu_ras_load_bad_pages(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras_eeprom_control *control =
+		&adev->psp.ras_context.ras->eeprom_control;
+	struct eeprom_table_record *bps;
+	int ret, i = 0;
+
+	/* no bad page record, skip eeprom access */
+	if (control->ras_num_recs == 0 || amdgpu_bad_page_threshold == 0)
+		return 0;
+
+	bps = kcalloc(control->ras_num_recs, sizeof(*bps), GFP_KERNEL);
+	if (!bps)
+		return -ENOMEM;
+
+	ret = amdgpu_ras_eeprom_read(control, bps, control->ras_num_recs);
+	if (ret) {
+		dev_err(adev->dev, "Failed to load EEPROM table records!");
+	} else {
+		if (adev->umc.ras && adev->umc.ras->convert_ras_err_addr) {
+			/*In V3, there is no pa recs, and some cases(when address==0) may be parsed
+			as pa recs, so add verion check to avoid it.
+			*/
+			if (control->tbl_hdr.version < RAS_TABLE_VER_V3) {
+				for (i = 0; i < control->ras_num_recs; i++) {
+					if ((control->ras_num_recs - i) >= adev->umc.retire_unit) {
+						if ((bps[i].address == bps[i + 1].address) &&
+							(bps[i].mem_channel == bps[i + 1].mem_channel)) {
+							control->ras_num_pa_recs += adev->umc.retire_unit;
+							i += (adev->umc.retire_unit - 1);
+						} else {
+							control->ras_num_mca_recs +=
+										(control->ras_num_recs - i);
+							break;
+						}
+					} else {
+						control->ras_num_mca_recs += (control->ras_num_recs - i);
+						break;
+					}
+				}
+			} else {
+				control->ras_num_mca_recs = control->ras_num_recs;
+			}
+		}
+
+		ret = amdgpu_ras_add_bad_pages(adev, bps, control->ras_num_recs, true);
+		if (ret)
+			goto out;
+
+		ret = amdgpu_ras_eeprom_check(control);
+		if (ret)
+			goto out;
+
+		/* HW not usable */
+		if (amdgpu_ras_is_rma(adev))
+			ret = -EHWPOISON;
+	}
+
+out:
+	kfree(bps);
+	return ret;
+}
+
+static int amdgpu_ras_check_bad_page_unlock(struct amdgpu_ras *con,
+				uint64_t addr)
+{
+	struct ras_err_handler_data *data = con->eh_data;
+	struct amdgpu_device *adev = con->adev;
+	int i;
+
+	if ((addr >= adev->gmc.mc_vram_size &&
+	    adev->gmc.mc_vram_size) ||
+	    (addr >= RAS_UMC_INJECT_ADDR_LIMIT))
+		return -EINVAL;
+
+	addr >>= AMDGPU_GPU_PAGE_SHIFT;
+	for (i = 0; i < data->count; i++)
+		if (addr == data->bps[i].retired_page)
+			return 1;
+
+	return 0;
+}
+
+/*
+ * check if an address belongs to bad page
+ *
+ * Note: this check is only for umc block
+ */
+static int amdgpu_ras_check_bad_page(struct amdgpu_device *adev,
+				uint64_t addr)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	int ret = 0;
+
+	if (!con || !con->eh_data)
+		return ret;
+
+	mutex_lock(&con->recovery_lock);
+	ret = amdgpu_ras_check_bad_page_unlock(con, addr);
+	mutex_unlock(&con->recovery_lock);
+	return ret;
+}
+
+static void amdgpu_ras_validate_threshold(struct amdgpu_device *adev,
+					  uint32_t max_count)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+
+	/*
+	 * amdgpu_bad_page_threshold is used to config
+	 * the threshold for the number of bad pages.
+	 * -1:  Threshold is set to default value
+	 *      Driver will issue a warning message when threshold is reached
+	 *      and continue runtime services.
+	 * 0:   Disable bad page retirement
+	 *      Driver will not retire bad pages
+	 *      which is intended for debugging purpose.
+	 * -2:  Threshold is determined by a formula
+	 *      that assumes 1 bad page per 100M of local memory.
+	 *      Driver will continue runtime services when threhold is reached.
+	 * 0 < threshold < max number of bad page records in EEPROM,
+	 *      A user-defined threshold is set
+	 *      Driver will halt runtime services when this custom threshold is reached.
+	 */
+	if (amdgpu_bad_page_threshold == -2) {
+		u64 val = adev->gmc.mc_vram_size;
+
+		do_div(val, RAS_BAD_PAGE_COVER);
+		con->bad_page_cnt_threshold = min(lower_32_bits(val),
+						  max_count);
+	} else if (amdgpu_bad_page_threshold == -1) {
+		con->bad_page_cnt_threshold = ((con->reserved_pages_in_bytes) >> 21) << 4;
+	} else {
+		con->bad_page_cnt_threshold = min_t(int, max_count,
+						    amdgpu_bad_page_threshold);
+	}
+}
+
+int amdgpu_ras_put_poison_req(struct amdgpu_device *adev,
+		enum amdgpu_ras_block block, uint16_t pasid,
+		pasid_notify pasid_fn, void *data, uint32_t reset)
+{
+	int ret = 0;
+	struct ras_poison_msg poison_msg;
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+
+	memset(&poison_msg, 0, sizeof(poison_msg));
+	poison_msg.block = block;
+	poison_msg.pasid = pasid;
+	poison_msg.reset = reset;
+	poison_msg.pasid_fn = pasid_fn;
+	poison_msg.data = data;
+
+	ret = kfifo_put(&con->poison_fifo, poison_msg);
+	if (!ret) {
+		dev_err(adev->dev, "Poison message fifo is full!\n");
+		return -ENOSPC;
+	}
+
+	return 0;
+}
+
+static int amdgpu_ras_get_poison_req(struct amdgpu_device *adev,
+		struct ras_poison_msg *poison_msg)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+
+	return kfifo_get(&con->poison_fifo, poison_msg);
+}
+
+static void amdgpu_ras_ecc_log_init(struct ras_ecc_log_info *ecc_log)
+{
+	mutex_init(&ecc_log->lock);
+
+	INIT_RADIX_TREE(&ecc_log->de_page_tree, GFP_KERNEL);
+	ecc_log->de_queried_count = 0;
+	ecc_log->consumption_q_count = 0;
+}
+
+static void amdgpu_ras_ecc_log_fini(struct ras_ecc_log_info *ecc_log)
+{
+	struct radix_tree_iter iter;
+	void __rcu **slot;
+	struct ras_ecc_err *ecc_err;
+
+	mutex_lock(&ecc_log->lock);
+	radix_tree_for_each_slot(slot, &ecc_log->de_page_tree, &iter, 0) {
+		ecc_err = radix_tree_deref_slot(slot);
+		kfree(ecc_err->err_pages.pfn);
+		kfree(ecc_err);
+		radix_tree_iter_delete(&ecc_log->de_page_tree, &iter, slot);
+	}
+	mutex_unlock(&ecc_log->lock);
+
+	mutex_destroy(&ecc_log->lock);
+	ecc_log->de_queried_count = 0;
+	ecc_log->consumption_q_count = 0;
+}
+
+static bool amdgpu_ras_schedule_retirement_dwork(struct amdgpu_ras *con,
+				uint32_t delayed_ms)
+{
+	int ret;
+
+	mutex_lock(&con->umc_ecc_log.lock);
+	ret = radix_tree_tagged(&con->umc_ecc_log.de_page_tree,
+			UMC_ECC_NEW_DETECTED_TAG);
+	mutex_unlock(&con->umc_ecc_log.lock);
+
+	if (ret)
+		schedule_delayed_work(&con->page_retirement_dwork,
+			msecs_to_jiffies(delayed_ms));
+
+	return ret ? true : false;
+}
+
+static void amdgpu_ras_do_page_retirement(struct work_struct *work)
+{
+	struct amdgpu_ras *con = container_of(work, struct amdgpu_ras,
+					      page_retirement_dwork.work);
+	struct amdgpu_device *adev = con->adev;
+	struct ras_err_data err_data;
+
+	/* If gpu reset is ongoing, delay retiring the bad pages */
+	if (amdgpu_in_reset(adev) || amdgpu_ras_in_recovery(adev)) {
+		amdgpu_ras_schedule_retirement_dwork(con,
+				AMDGPU_RAS_RETIRE_PAGE_INTERVAL * 3);
+		return;
+	}
+
+	amdgpu_ras_error_data_init(&err_data);
+
+	amdgpu_umc_handle_bad_pages(adev, &err_data);
+
+	amdgpu_ras_error_data_fini(&err_data);
+
+	amdgpu_ras_schedule_retirement_dwork(con,
+			AMDGPU_RAS_RETIRE_PAGE_INTERVAL);
+}
+
+static int amdgpu_ras_poison_creation_handler(struct amdgpu_device *adev,
+				uint32_t poison_creation_count)
+{
+	int ret = 0;
+	struct ras_ecc_log_info *ecc_log;
+	struct ras_query_if info;
+	u32 timeout = MAX_UMC_POISON_POLLING_TIME_ASYNC;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+	u64 de_queried_count;
+	u64 consumption_q_count;
+	enum ras_event_type type = RAS_EVENT_TYPE_POISON_CREATION;
+
+	memset(&info, 0, sizeof(info));
+	info.head.block = AMDGPU_RAS_BLOCK__UMC;
+
+	ecc_log = &ras->umc_ecc_log;
+	ecc_log->de_queried_count = 0;
+	ecc_log->consumption_q_count = 0;
+
+	do {
+		ret = amdgpu_ras_query_error_status_with_event(adev, &info, type);
+		if (ret)
+			return ret;
+
+		de_queried_count = ecc_log->de_queried_count;
+		consumption_q_count = ecc_log->consumption_q_count;
+
+		if (de_queried_count && consumption_q_count)
+			break;
+
+		msleep(100);
+	} while (--timeout);
+
+	if (de_queried_count)
+		schedule_delayed_work(&ras->page_retirement_dwork, 0);
+
+	if (amdgpu_ras_is_rma(adev) && atomic_cmpxchg(&ras->rma_in_recovery, 0, 1) == 0)
+		amdgpu_ras_reset_gpu(adev);
+
+	return 0;
+}
+
+static void amdgpu_ras_clear_poison_fifo(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_poison_msg msg;
+	int ret;
+
+	do {
+		ret = kfifo_get(&con->poison_fifo, &msg);
+	} while (ret);
+}
+
+static int amdgpu_ras_poison_consumption_handler(struct amdgpu_device *adev,
+			uint32_t msg_count, uint32_t *gpu_reset)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	uint32_t reset_flags = 0, reset = 0;
+	struct ras_poison_msg msg;
+	int ret, i;
+
+	kgd2kfd_set_sram_ecc_flag(adev->kfd.dev);
+
+	for (i = 0; i < msg_count; i++) {
+		ret = amdgpu_ras_get_poison_req(adev, &msg);
+		if (!ret)
+			continue;
+
+		if (msg.pasid_fn)
+			msg.pasid_fn(adev, msg.pasid, msg.data);
+
+		reset_flags |= msg.reset;
+	}
+
+	/*
+	 * Try to ensure poison creation handler is completed first
+	 * to set rma if bad page exceed threshold.
+	 */
+	flush_delayed_work(&con->page_retirement_dwork);
+
+	/* for RMA, amdgpu_ras_poison_creation_handler will trigger gpu reset */
+	if (reset_flags && !amdgpu_ras_is_rma(adev)) {
+		if (reset_flags & AMDGPU_RAS_GPU_RESET_MODE1_RESET)
+			reset = AMDGPU_RAS_GPU_RESET_MODE1_RESET;
+		else if (reset_flags & AMDGPU_RAS_GPU_RESET_MODE2_RESET)
+			reset = AMDGPU_RAS_GPU_RESET_MODE2_RESET;
+		else
+			reset = reset_flags;
+
+		con->gpu_reset_flags |= reset;
+		amdgpu_ras_reset_gpu(adev);
+
+		*gpu_reset = reset;
+
+		/* Wait for gpu recovery to complete */
+		flush_work(&con->recovery_work);
+	}
+
+	return 0;
+}
+
+static int amdgpu_ras_page_retirement_thread(void *param)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)param;
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	uint32_t poison_creation_count, msg_count;
+	uint32_t gpu_reset;
+	int ret;
+
+	while (!kthread_should_stop()) {
+
+		wait_event_interruptible(con->page_retirement_wq,
+				kthread_should_stop() ||
+				atomic_read(&con->page_retirement_req_cnt));
+
+		if (kthread_should_stop())
+			break;
+
+		mutex_lock(&con->poison_lock);
+		gpu_reset = 0;
+
+		do {
+			poison_creation_count = atomic_read(&con->poison_creation_count);
+			ret = amdgpu_ras_poison_creation_handler(adev, poison_creation_count);
+			if (ret == -EIO)
+				break;
+
+			if (poison_creation_count) {
+				atomic_sub(poison_creation_count, &con->poison_creation_count);
+				atomic_sub(poison_creation_count, &con->page_retirement_req_cnt);
+			}
+		} while (atomic_read(&con->poison_creation_count) &&
+			!atomic_read(&con->poison_consumption_count));
+
+		if (ret != -EIO) {
+			msg_count = kfifo_len(&con->poison_fifo);
+			if (msg_count) {
+				ret = amdgpu_ras_poison_consumption_handler(adev,
+						msg_count, &gpu_reset);
+				if ((ret != -EIO) &&
+				    (gpu_reset != AMDGPU_RAS_GPU_RESET_MODE1_RESET))
+					atomic_sub(msg_count, &con->page_retirement_req_cnt);
+			}
+		}
+
+		if ((ret == -EIO) || (gpu_reset == AMDGPU_RAS_GPU_RESET_MODE1_RESET)) {
+			/* gpu mode-1 reset is ongoing or just completed ras mode-1 reset */
+			/* Clear poison creation request */
+			atomic_set(&con->poison_creation_count, 0);
+			atomic_set(&con->poison_consumption_count, 0);
+
+			/* Clear poison fifo */
+			amdgpu_ras_clear_poison_fifo(adev);
+
+			/* Clear all poison requests */
+			atomic_set(&con->page_retirement_req_cnt, 0);
+
+			if (ret == -EIO) {
+				/* Wait for mode-1 reset to complete */
+				down_read(&adev->reset_domain->sem);
+				up_read(&adev->reset_domain->sem);
+			}
+
+			/* Wake up work to save bad pages to eeprom */
+			schedule_delayed_work(&con->page_retirement_dwork, 0);
+		} else if (gpu_reset) {
+			/* gpu just completed mode-2 reset or other reset */
+			/* Clear poison consumption messages cached in fifo */
+			msg_count = kfifo_len(&con->poison_fifo);
+			if (msg_count) {
+				amdgpu_ras_clear_poison_fifo(adev);
+				atomic_sub(msg_count, &con->page_retirement_req_cnt);
+			}
+
+			atomic_set(&con->poison_consumption_count, 0);
+
+			/* Wake up work to save bad pages to eeprom */
+			schedule_delayed_work(&con->page_retirement_dwork, 0);
+		}
+		mutex_unlock(&con->poison_lock);
+	}
+
+	return 0;
+}
+
+int amdgpu_ras_init_badpage_info(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct amdgpu_ras_eeprom_control *control;
+	int ret;
+
+	if (!con || amdgpu_sriov_vf(adev))
+		return 0;
+
+	control = &con->eeprom_control;
+	ret = amdgpu_ras_eeprom_init(control);
+	control->is_eeprom_valid = !ret;
+
+	if (!adev->umc.ras || !adev->umc.ras->convert_ras_err_addr)
+		control->ras_num_pa_recs = control->ras_num_recs;
+
+	if (adev->umc.ras &&
+	    adev->umc.ras->get_retire_flip_bits)
+		adev->umc.ras->get_retire_flip_bits(adev);
+
+	if (control->ras_num_recs && control->is_eeprom_valid) {
+		ret = amdgpu_ras_load_bad_pages(adev);
+		if (ret) {
+			control->is_eeprom_valid = false;
+			return 0;
+		}
+
+		amdgpu_dpm_send_hbm_bad_pages_num(
+			adev, control->ras_num_bad_pages);
+
+		if (con->update_channel_flag == true) {
+			amdgpu_dpm_send_hbm_bad_channel_flag(
+				adev, control->bad_channel_bitmap);
+			con->update_channel_flag = false;
+		}
+
+		/* The format action is only applied to new ASICs */
+		if (IP_VERSION_MAJ(amdgpu_ip_version(adev, UMC_HWIP, 0)) >= 12 &&
+		    control->tbl_hdr.version < RAS_TABLE_VER_V3)
+			if (!amdgpu_ras_eeprom_reset_table(control))
+				if (amdgpu_ras_save_bad_pages(adev, NULL))
+					dev_warn(adev->dev, "Failed to format RAS EEPROM data in V3 version!\n");
+	}
+
+	return 0;
+}
+
+int amdgpu_ras_recovery_init(struct amdgpu_device *adev, bool init_bp_info)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_err_handler_data **data;
+	u32  max_eeprom_records_count = 0;
+	int ret;
+
+	if (!con || amdgpu_sriov_vf(adev))
+		return 0;
+
+	/* Allow access to RAS EEPROM via debugfs, when the ASIC
+	 * supports RAS and debugfs is enabled, but when
+	 * adev->ras_enabled is unset, i.e. when "ras_enable"
+	 * module parameter is set to 0.
+	 */
+	con->adev = adev;
+
+	if (!adev->ras_enabled)
+		return 0;
+
+	data = &con->eh_data;
+	*data = kzalloc(sizeof(**data), GFP_KERNEL);
+	if (!*data) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	mutex_init(&con->recovery_lock);
+	mutex_init(&con->poison_lock);
+	INIT_WORK(&con->recovery_work, amdgpu_ras_do_recovery);
+	atomic_set(&con->in_recovery, 0);
+	atomic_set(&con->rma_in_recovery, 0);
+	con->eeprom_control.bad_channel_bitmap = 0;
+
+	max_eeprom_records_count = amdgpu_ras_eeprom_max_record_count(&con->eeprom_control);
+	amdgpu_ras_validate_threshold(adev, max_eeprom_records_count);
+
+	if (init_bp_info) {
+		ret = amdgpu_ras_init_badpage_info(adev);
+		if (ret)
+			goto free;
+	}
+
+	mutex_init(&con->page_rsv_lock);
+	INIT_KFIFO(con->poison_fifo);
+	mutex_init(&con->page_retirement_lock);
+	init_waitqueue_head(&con->page_retirement_wq);
+	atomic_set(&con->page_retirement_req_cnt, 0);
+	atomic_set(&con->poison_creation_count, 0);
+	atomic_set(&con->poison_consumption_count, 0);
+	con->page_retirement_thread =
+		kthread_run(amdgpu_ras_page_retirement_thread, adev, "umc_page_retirement");
+	if (IS_ERR(con->page_retirement_thread)) {
+		con->page_retirement_thread = NULL;
+		dev_warn(adev->dev, "Failed to create umc_page_retirement thread!!!\n");
+	}
+
+	INIT_DELAYED_WORK(&con->page_retirement_dwork, amdgpu_ras_do_page_retirement);
+	amdgpu_ras_ecc_log_init(&con->umc_ecc_log);
+#ifdef CONFIG_X86_MCE_AMD
+	if ((adev->asic_type == CHIP_ALDEBARAN) &&
+	    (adev->gmc.xgmi.connected_to_cpu))
+		amdgpu_register_bad_pages_mca_notifier(adev);
+#endif
+	return 0;
+
+free:
+	kfree((*data)->bps);
+	kfree(*data);
+	con->eh_data = NULL;
+out:
+	dev_warn(adev->dev, "Failed to initialize ras recovery! (%d)\n", ret);
+
+	/*
+	 * Except error threshold exceeding case, other failure cases in this
+	 * function would not fail amdgpu driver init.
+	 */
+	if (!amdgpu_ras_is_rma(adev))
+		ret = 0;
+	else
+		ret = -EINVAL;
+
+	return ret;
+}
+
+static int amdgpu_ras_recovery_fini(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_err_handler_data *data = con->eh_data;
+	int max_flush_timeout = MAX_FLUSH_RETIRE_DWORK_TIMES;
+	bool ret;
+
+	/* recovery_init failed to init it, fini is useless */
+	if (!data)
+		return 0;
+
+	/* Save all cached bad pages to eeprom */
+	do {
+		flush_delayed_work(&con->page_retirement_dwork);
+		ret = amdgpu_ras_schedule_retirement_dwork(con, 0);
+	} while (ret && max_flush_timeout--);
+
+	if (con->page_retirement_thread)
+		kthread_stop(con->page_retirement_thread);
+
+	atomic_set(&con->page_retirement_req_cnt, 0);
+	atomic_set(&con->poison_creation_count, 0);
+
+	mutex_destroy(&con->page_rsv_lock);
+
+	cancel_work_sync(&con->recovery_work);
+
+	cancel_delayed_work_sync(&con->page_retirement_dwork);
+
+	amdgpu_ras_ecc_log_fini(&con->umc_ecc_log);
+
+	mutex_lock(&con->recovery_lock);
+	con->eh_data = NULL;
+	kfree(data->bps);
+	kfree(data);
+	mutex_unlock(&con->recovery_lock);
+
+	amdgpu_ras_critical_region_init(adev);
+
+	return 0;
+}
+/* recovery end */
+
+static bool amdgpu_ras_asic_supported(struct amdgpu_device *adev)
+{
+	if (amdgpu_sriov_vf(adev)) {
+		switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) {
+		case IP_VERSION(13, 0, 2):
+		case IP_VERSION(13, 0, 6):
+		case IP_VERSION(13, 0, 12):
+		case IP_VERSION(13, 0, 14):
+			return true;
+		default:
+			return false;
+		}
+	}
+
+	if (adev->asic_type == CHIP_IP_DISCOVERY) {
+		switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) {
+		case IP_VERSION(13, 0, 0):
+		case IP_VERSION(13, 0, 6):
+		case IP_VERSION(13, 0, 10):
+		case IP_VERSION(13, 0, 12):
+		case IP_VERSION(13, 0, 14):
+		case IP_VERSION(14, 0, 3):
+			return true;
+		default:
+			return false;
+		}
+	}
+
+	return adev->asic_type == CHIP_VEGA10 ||
+		adev->asic_type == CHIP_VEGA20 ||
+		adev->asic_type == CHIP_ARCTURUS ||
+		adev->asic_type == CHIP_ALDEBARAN ||
+		adev->asic_type == CHIP_SIENNA_CICHLID;
+}
+
+/*
+ * this is workaround for vega20 workstation sku,
+ * force enable gfx ras, ignore vbios gfx ras flag
+ * due to GC EDC can not write
+ */
+static void amdgpu_ras_get_quirks(struct amdgpu_device *adev)
+{
+	struct atom_context *ctx = adev->mode_info.atom_context;
+
+	if (!ctx)
+		return;
+
+	if (strnstr(ctx->vbios_pn, "D16406",
+		    sizeof(ctx->vbios_pn)) ||
+		strnstr(ctx->vbios_pn, "D36002",
+			sizeof(ctx->vbios_pn)))
+		adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__GFX);
+}
+
+/* Query ras capablity via atomfirmware interface */
+static void amdgpu_ras_query_ras_capablity_from_vbios(struct amdgpu_device *adev)
+{
+	/* mem_ecc cap */
+	if (amdgpu_atomfirmware_mem_ecc_supported(adev)) {
+		dev_info(adev->dev, "MEM ECC is active.\n");
+		adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__UMC |
+					 1 << AMDGPU_RAS_BLOCK__DF);
+	} else {
+		dev_info(adev->dev, "MEM ECC is not presented.\n");
+	}
+
+	/* sram_ecc cap */
+	if (amdgpu_atomfirmware_sram_ecc_supported(adev)) {
+		dev_info(adev->dev, "SRAM ECC is active.\n");
+		if (!amdgpu_sriov_vf(adev))
+			adev->ras_hw_enabled |= ~(1 << AMDGPU_RAS_BLOCK__UMC |
+						  1 << AMDGPU_RAS_BLOCK__DF);
+		else
+			adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__PCIE_BIF |
+						 1 << AMDGPU_RAS_BLOCK__SDMA |
+						 1 << AMDGPU_RAS_BLOCK__GFX);
+
+		/*
+		 * VCN/JPEG RAS can be supported on both bare metal and
+		 * SRIOV environment
+		 */
+		if (amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(2, 6, 0) ||
+		    amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 0) ||
+		    amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 3) ||
+		    amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(5, 0, 1))
+			adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__VCN |
+						 1 << AMDGPU_RAS_BLOCK__JPEG);
+		else
+			adev->ras_hw_enabled &= ~(1 << AMDGPU_RAS_BLOCK__VCN |
+						  1 << AMDGPU_RAS_BLOCK__JPEG);
+
+		/*
+		 * XGMI RAS is not supported if xgmi num physical nodes
+		 * is zero
+		 */
+		if (!adev->gmc.xgmi.num_physical_nodes)
+			adev->ras_hw_enabled &= ~(1 << AMDGPU_RAS_BLOCK__XGMI_WAFL);
+	} else {
+		dev_info(adev->dev, "SRAM ECC is not presented.\n");
+	}
+}
+
+/* Query poison mode from umc/df IP callbacks */
+static void amdgpu_ras_query_poison_mode(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	bool df_poison, umc_poison;
+
+	/* poison setting is useless on SRIOV guest */
+	if (amdgpu_sriov_vf(adev) || !con)
+		return;
+
+	/* Init poison supported flag, the default value is false */
+	if (adev->gmc.xgmi.connected_to_cpu ||
+	    adev->gmc.is_app_apu) {
+		/* enabled by default when GPU is connected to CPU */
+		con->poison_supported = true;
+	} else if (adev->df.funcs &&
+	    adev->df.funcs->query_ras_poison_mode &&
+	    adev->umc.ras &&
+	    adev->umc.ras->query_ras_poison_mode) {
+		df_poison =
+			adev->df.funcs->query_ras_poison_mode(adev);
+		umc_poison =
+			adev->umc.ras->query_ras_poison_mode(adev);
+
+		/* Only poison is set in both DF and UMC, we can support it */
+		if (df_poison && umc_poison)
+			con->poison_supported = true;
+		else if (df_poison != umc_poison)
+			dev_warn(adev->dev,
+				"Poison setting is inconsistent in DF/UMC(%d:%d)!\n",
+				df_poison, umc_poison);
+	}
+}
+
+/*
+ * check hardware's ras ability which will be saved in hw_supported.
+ * if hardware does not support ras, we can skip some ras initializtion and
+ * forbid some ras operations from IP.
+ * if software itself, say boot parameter, limit the ras ability. We still
+ * need allow IP do some limited operations, like disable. In such case,
+ * we have to initialize ras as normal. but need check if operation is
+ * allowed or not in each function.
+ */
+static void amdgpu_ras_check_supported(struct amdgpu_device *adev)
+{
+	adev->ras_hw_enabled = adev->ras_enabled = 0;
+
+	if (!amdgpu_ras_asic_supported(adev))
+		return;
+
+	if (amdgpu_sriov_vf(adev)) {
+		if (amdgpu_virt_get_ras_capability(adev))
+			goto init_ras_enabled_flag;
+	}
+
+	/* query ras capability from psp */
+	if (amdgpu_psp_get_ras_capability(&adev->psp))
+		goto init_ras_enabled_flag;
+
+	/* query ras capablity from bios */
+	if (!adev->gmc.xgmi.connected_to_cpu && !adev->gmc.is_app_apu) {
+		amdgpu_ras_query_ras_capablity_from_vbios(adev);
+	} else {
+		/* driver only manages a few IP blocks RAS feature
+		 * when GPU is connected cpu through XGMI */
+		adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__GFX |
+					   1 << AMDGPU_RAS_BLOCK__SDMA |
+					   1 << AMDGPU_RAS_BLOCK__MMHUB);
+	}
+
+	/* apply asic specific settings (vega20 only for now) */
+	amdgpu_ras_get_quirks(adev);
+
+	/* query poison mode from umc/df ip callback */
+	amdgpu_ras_query_poison_mode(adev);
+
+init_ras_enabled_flag:
+	/* hw_supported needs to be aligned with RAS block mask. */
+	adev->ras_hw_enabled &= AMDGPU_RAS_BLOCK_MASK;
+
+	adev->ras_enabled = amdgpu_ras_enable == 0 ? 0 :
+		adev->ras_hw_enabled & amdgpu_ras_mask;
+
+	/* aca is disabled by default except for psp v13_0_6/v13_0_12/v13_0_14 */
+	if (!amdgpu_sriov_vf(adev)) {
+		adev->aca.is_enabled =
+			(amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 6) ||
+			amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 12) ||
+			amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 14));
+	}
+
+	/* bad page feature is not applicable to specific app platform */
+	if (adev->gmc.is_app_apu &&
+	    amdgpu_ip_version(adev, UMC_HWIP, 0) == IP_VERSION(12, 0, 0))
+		amdgpu_bad_page_threshold = 0;
+}
+
+static void amdgpu_ras_counte_dw(struct work_struct *work)
+{
+	struct amdgpu_ras *con = container_of(work, struct amdgpu_ras,
+					      ras_counte_delay_work.work);
+	struct amdgpu_device *adev = con->adev;
+	struct drm_device *dev = adev_to_drm(adev);
+	unsigned long ce_count, ue_count;
+	int res;
+
+	res = pm_runtime_get_sync(dev->dev);
+	if (res < 0)
+		goto Out;
+
+	/* Cache new values.
+	 */
+	if (amdgpu_ras_query_error_count(adev, &ce_count, &ue_count, NULL) == 0) {
+		atomic_set(&con->ras_ce_count, ce_count);
+		atomic_set(&con->ras_ue_count, ue_count);
+	}
+
+	pm_runtime_mark_last_busy(dev->dev);
+Out:
+	pm_runtime_put_autosuspend(dev->dev);
+}
+
+static int amdgpu_get_ras_schema(struct amdgpu_device *adev)
+{
+	return  amdgpu_ras_is_poison_mode_supported(adev) ? AMDGPU_RAS_ERROR__POISON : 0 |
+			AMDGPU_RAS_ERROR__SINGLE_CORRECTABLE |
+			AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE |
+			AMDGPU_RAS_ERROR__PARITY;
+}
+
+static void ras_event_mgr_init(struct ras_event_manager *mgr)
+{
+	struct ras_event_state *event_state;
+	int i;
+
+	memset(mgr, 0, sizeof(*mgr));
+	atomic64_set(&mgr->seqno, 0);
+
+	for (i = 0; i < ARRAY_SIZE(mgr->event_state); i++) {
+		event_state = &mgr->event_state[i];
+		event_state->last_seqno = RAS_EVENT_INVALID_ID;
+		atomic64_set(&event_state->count, 0);
+	}
+}
+
+static void amdgpu_ras_event_mgr_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+	struct amdgpu_hive_info *hive;
+
+	if (!ras)
+		return;
+
+	hive = amdgpu_get_xgmi_hive(adev);
+	ras->event_mgr = hive ? &hive->event_mgr : &ras->__event_mgr;
+
+	/* init event manager with node 0 on xgmi system */
+	if (!amdgpu_reset_in_recovery(adev)) {
+		if (!hive || adev->gmc.xgmi.node_id == 0)
+			ras_event_mgr_init(ras->event_mgr);
+	}
+
+	if (hive)
+		amdgpu_put_xgmi_hive(hive);
+}
+
+static void amdgpu_ras_init_reserved_vram_size(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+
+	if (!con || (adev->flags & AMD_IS_APU))
+		return;
+
+	switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) {
+	case IP_VERSION(13, 0, 2):
+	case IP_VERSION(13, 0, 6):
+	case IP_VERSION(13, 0, 12):
+		con->reserved_pages_in_bytes = AMDGPU_RAS_RESERVED_VRAM_SIZE_DEFAULT;
+		break;
+	case IP_VERSION(13, 0, 14):
+		con->reserved_pages_in_bytes = (AMDGPU_RAS_RESERVED_VRAM_SIZE_DEFAULT << 1);
+		break;
+	default:
+		break;
+	}
+}
+
+int amdgpu_ras_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	int r;
+
+	if (con)
+		return 0;
+
+	con = kzalloc(sizeof(*con) +
+			sizeof(struct ras_manager) * AMDGPU_RAS_BLOCK_COUNT +
+			sizeof(struct ras_manager) * AMDGPU_RAS_MCA_BLOCK_COUNT,
+			GFP_KERNEL);
+	if (!con)
+		return -ENOMEM;
+
+	con->adev = adev;
+	INIT_DELAYED_WORK(&con->ras_counte_delay_work, amdgpu_ras_counte_dw);
+	atomic_set(&con->ras_ce_count, 0);
+	atomic_set(&con->ras_ue_count, 0);
+
+	con->objs = (struct ras_manager *)(con + 1);
+
+	amdgpu_ras_set_context(adev, con);
+
+	amdgpu_ras_check_supported(adev);
+
+	if (!adev->ras_enabled || adev->asic_type == CHIP_VEGA10) {
+		/* set gfx block ras context feature for VEGA20 Gaming
+		 * send ras disable cmd to ras ta during ras late init.
+		 */
+		if (!adev->ras_enabled && adev->asic_type == CHIP_VEGA20) {
+			con->features |= BIT(AMDGPU_RAS_BLOCK__GFX);
+
+			return 0;
+		}
+
+		r = 0;
+		goto release_con;
+	}
+
+	con->update_channel_flag = false;
+	con->features = 0;
+	con->schema = 0;
+	INIT_LIST_HEAD(&con->head);
+	/* Might need get this flag from vbios. */
+	con->flags = RAS_DEFAULT_FLAGS;
+
+	/* initialize nbio ras function ahead of any other
+	 * ras functions so hardware fatal error interrupt
+	 * can be enabled as early as possible */
+	switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) {
+	case IP_VERSION(7, 4, 0):
+	case IP_VERSION(7, 4, 1):
+	case IP_VERSION(7, 4, 4):
+		if (!adev->gmc.xgmi.connected_to_cpu)
+			adev->nbio.ras = &nbio_v7_4_ras;
+		break;
+	case IP_VERSION(4, 3, 0):
+		if (adev->ras_hw_enabled & (1 << AMDGPU_RAS_BLOCK__DF))
+			/* unlike other generation of nbio ras,
+			 * nbio v4_3 only support fatal error interrupt
+			 * to inform software that DF is freezed due to
+			 * system fatal error event. driver should not
+			 * enable nbio ras in such case. Instead,
+			 * check DF RAS */
+			adev->nbio.ras = &nbio_v4_3_ras;
+		break;
+	case IP_VERSION(6, 3, 1):
+		if (adev->ras_hw_enabled & (1 << AMDGPU_RAS_BLOCK__DF))
+			/* unlike other generation of nbio ras,
+			 * nbif v6_3_1 only support fatal error interrupt
+			 * to inform software that DF is freezed due to
+			 * system fatal error event. driver should not
+			 * enable nbio ras in such case. Instead,
+			 * check DF RAS
+			 */
+			adev->nbio.ras = &nbif_v6_3_1_ras;
+		break;
+	case IP_VERSION(7, 9, 0):
+	case IP_VERSION(7, 9, 1):
+		if (!adev->gmc.is_app_apu)
+			adev->nbio.ras = &nbio_v7_9_ras;
+		break;
+	default:
+		/* nbio ras is not available */
+		break;
+	}
+
+	/* nbio ras block needs to be enabled ahead of other ras blocks
+	 * to handle fatal error */
+	r = amdgpu_nbio_ras_sw_init(adev);
+	if (r)
+		return r;
+
+	if (adev->nbio.ras &&
+	    adev->nbio.ras->init_ras_controller_interrupt) {
+		r = adev->nbio.ras->init_ras_controller_interrupt(adev);
+		if (r)
+			goto release_con;
+	}
+
+	if (adev->nbio.ras &&
+	    adev->nbio.ras->init_ras_err_event_athub_interrupt) {
+		r = adev->nbio.ras->init_ras_err_event_athub_interrupt(adev);
+		if (r)
+			goto release_con;
+	}
+
+	/* Packed socket_id to ras feature mask bits[31:29] */
+	if (adev->smuio.funcs &&
+	    adev->smuio.funcs->get_socket_id)
+		con->features |= ((adev->smuio.funcs->get_socket_id(adev)) <<
+					AMDGPU_RAS_FEATURES_SOCKETID_SHIFT);
+
+	/* Get RAS schema for particular SOC */
+	con->schema = amdgpu_get_ras_schema(adev);
+
+	amdgpu_ras_init_reserved_vram_size(adev);
+
+	if (amdgpu_ras_fs_init(adev)) {
+		r = -EINVAL;
+		goto release_con;
+	}
+
+	if (amdgpu_ras_aca_is_supported(adev)) {
+		if (amdgpu_aca_is_enabled(adev))
+			r = amdgpu_aca_init(adev);
+		else
+			r = amdgpu_mca_init(adev);
+		if (r)
+			goto release_con;
+	}
+
+	con->init_task_pid = task_pid_nr(current);
+	get_task_comm(con->init_task_comm, current);
+
+	mutex_init(&con->critical_region_lock);
+	INIT_LIST_HEAD(&con->critical_region_head);
+
+	dev_info(adev->dev, "RAS INFO: ras initialized successfully, "
+		 "hardware ability[%x] ras_mask[%x]\n",
+		 adev->ras_hw_enabled, adev->ras_enabled);
+
+	return 0;
+release_con:
+	amdgpu_ras_set_context(adev, NULL);
+	kfree(con);
+
+	return r;
+}
+
+int amdgpu_persistent_edc_harvesting_supported(struct amdgpu_device *adev)
+{
+	if (adev->gmc.xgmi.connected_to_cpu ||
+	    adev->gmc.is_app_apu)
+		return 1;
+	return 0;
+}
+
+static int amdgpu_persistent_edc_harvesting(struct amdgpu_device *adev,
+					struct ras_common_if *ras_block)
+{
+	struct ras_query_if info = {
+		.head = *ras_block,
+	};
+
+	if (!amdgpu_persistent_edc_harvesting_supported(adev))
+		return 0;
+
+	if (amdgpu_ras_query_error_status(adev, &info) != 0)
+		DRM_WARN("RAS init harvest failure");
+
+	if (amdgpu_ras_reset_error_status(adev, ras_block->block) != 0)
+		DRM_WARN("RAS init harvest reset failure");
+
+	return 0;
+}
+
+bool amdgpu_ras_is_poison_mode_supported(struct amdgpu_device *adev)
+{
+       struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+
+       if (!con)
+               return false;
+
+       return con->poison_supported;
+}
+
+/* helper function to handle common stuff in ip late init phase */
+int amdgpu_ras_block_late_init(struct amdgpu_device *adev,
+			 struct ras_common_if *ras_block)
+{
+	struct amdgpu_ras_block_object *ras_obj = NULL;
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_query_if *query_info;
+	unsigned long ue_count, ce_count;
+	int r;
+
+	/* disable RAS feature per IP block if it is not supported */
+	if (!amdgpu_ras_is_supported(adev, ras_block->block)) {
+		amdgpu_ras_feature_enable_on_boot(adev, ras_block, 0);
+		return 0;
+	}
+
+	r = amdgpu_ras_feature_enable_on_boot(adev, ras_block, 1);
+	if (r) {
+		if (adev->in_suspend || amdgpu_reset_in_recovery(adev)) {
+			/* in resume phase, if fail to enable ras,
+			 * clean up all ras fs nodes, and disable ras */
+			goto cleanup;
+		} else
+			return r;
+	}
+
+	/* check for errors on warm reset edc persisant supported ASIC */
+	amdgpu_persistent_edc_harvesting(adev, ras_block);
+
+	/* in resume phase, no need to create ras fs node */
+	if (adev->in_suspend || amdgpu_reset_in_recovery(adev))
+		return 0;
+
+	ras_obj = container_of(ras_block, struct amdgpu_ras_block_object, ras_comm);
+	if (ras_obj->ras_cb || (ras_obj->hw_ops &&
+	    (ras_obj->hw_ops->query_poison_status ||
+	    ras_obj->hw_ops->handle_poison_consumption))) {
+		r = amdgpu_ras_interrupt_add_handler(adev, ras_block);
+		if (r)
+			goto cleanup;
+	}
+
+	if (ras_obj->hw_ops &&
+	    (ras_obj->hw_ops->query_ras_error_count ||
+	     ras_obj->hw_ops->query_ras_error_status)) {
+		r = amdgpu_ras_sysfs_create(adev, ras_block);
+		if (r)
+			goto interrupt;
+
+		/* Those are the cached values at init.
+		 */
+		query_info = kzalloc(sizeof(*query_info), GFP_KERNEL);
+		if (!query_info)
+			return -ENOMEM;
+		memcpy(&query_info->head, ras_block, sizeof(struct ras_common_if));
+
+		if (amdgpu_ras_query_error_count(adev, &ce_count, &ue_count, query_info) == 0) {
+			atomic_set(&con->ras_ce_count, ce_count);
+			atomic_set(&con->ras_ue_count, ue_count);
+		}
+
+		kfree(query_info);
+	}
+
+	return 0;
+
+interrupt:
+	if (ras_obj->ras_cb)
+		amdgpu_ras_interrupt_remove_handler(adev, ras_block);
+cleanup:
+	amdgpu_ras_feature_enable(adev, ras_block, 0);
+	return r;
+}
+
+static int amdgpu_ras_block_late_init_default(struct amdgpu_device *adev,
+			 struct ras_common_if *ras_block)
+{
+	return amdgpu_ras_block_late_init(adev, ras_block);
+}
+
+/* helper function to remove ras fs node and interrupt handler */
+void amdgpu_ras_block_late_fini(struct amdgpu_device *adev,
+			  struct ras_common_if *ras_block)
+{
+	struct amdgpu_ras_block_object *ras_obj;
+	if (!ras_block)
+		return;
+
+	amdgpu_ras_sysfs_remove(adev, ras_block);
+
+	ras_obj = container_of(ras_block, struct amdgpu_ras_block_object, ras_comm);
+	if (ras_obj->ras_cb)
+		amdgpu_ras_interrupt_remove_handler(adev, ras_block);
+}
+
+static void amdgpu_ras_block_late_fini_default(struct amdgpu_device *adev,
+			  struct ras_common_if *ras_block)
+{
+	return amdgpu_ras_block_late_fini(adev, ras_block);
+}
+
+/* do some init work after IP late init as dependence.
+ * and it runs in resume/gpu reset/booting up cases.
+ */
+void amdgpu_ras_resume(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_manager *obj, *tmp;
+
+	if (!adev->ras_enabled || !con) {
+		/* clean ras context for VEGA20 Gaming after send ras disable cmd */
+		amdgpu_release_ras_context(adev);
+
+		return;
+	}
+
+	if (con->flags & AMDGPU_RAS_FLAG_INIT_BY_VBIOS) {
+		/* Set up all other IPs which are not implemented. There is a
+		 * tricky thing that IP's actual ras error type should be
+		 * MULTI_UNCORRECTABLE, but as driver does not handle it, so
+		 * ERROR_NONE make sense anyway.
+		 */
+		amdgpu_ras_enable_all_features(adev, 1);
+
+		/* We enable ras on all hw_supported block, but as boot
+		 * parameter might disable some of them and one or more IP has
+		 * not implemented yet. So we disable them on behalf.
+		 */
+		list_for_each_entry_safe(obj, tmp, &con->head, node) {
+			if (!amdgpu_ras_is_supported(adev, obj->head.block)) {
+				amdgpu_ras_feature_enable(adev, &obj->head, 0);
+				/* there should be no any reference. */
+				WARN_ON(alive_obj(obj));
+			}
+		}
+	}
+}
+
+void amdgpu_ras_suspend(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+
+	if (!adev->ras_enabled || !con)
+		return;
+
+	amdgpu_ras_disable_all_features(adev, 0);
+	/* Make sure all ras objects are disabled. */
+	if (AMDGPU_RAS_GET_FEATURES(con->features))
+		amdgpu_ras_disable_all_features(adev, 1);
+}
+
+int amdgpu_ras_late_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras_block_list *node, *tmp;
+	struct amdgpu_ras_block_object *obj;
+	int r;
+
+	amdgpu_ras_event_mgr_init(adev);
+
+	if (amdgpu_ras_aca_is_supported(adev)) {
+		if (amdgpu_reset_in_recovery(adev)) {
+			if (amdgpu_aca_is_enabled(adev))
+				r = amdgpu_aca_reset(adev);
+			else
+				r = amdgpu_mca_reset(adev);
+			if (r)
+				return r;
+		}
+
+		if (!amdgpu_sriov_vf(adev)) {
+			if (amdgpu_aca_is_enabled(adev))
+				amdgpu_ras_set_aca_debug_mode(adev, false);
+			else
+				amdgpu_ras_set_mca_debug_mode(adev, false);
+		}
+	}
+
+	/* Guest side doesn't need init ras feature */
+	if (amdgpu_sriov_vf(adev) && !amdgpu_sriov_ras_telemetry_en(adev))
+		return 0;
+
+	list_for_each_entry_safe(node, tmp, &adev->ras_list, node) {
+		obj = node->ras_obj;
+		if (!obj) {
+			dev_warn(adev->dev, "Warning: abnormal ras list node.\n");
+			continue;
+		}
+
+		if (!amdgpu_ras_is_supported(adev, obj->ras_comm.block))
+			continue;
+
+		if (obj->ras_late_init) {
+			r = obj->ras_late_init(adev, &obj->ras_comm);
+			if (r) {
+				dev_err(adev->dev, "%s failed to execute ras_late_init! ret:%d\n",
+					obj->ras_comm.name, r);
+				return r;
+			}
+		} else
+			amdgpu_ras_block_late_init_default(adev, &obj->ras_comm);
+	}
+
+	return 0;
+}
+
+/* do some fini work before IP fini as dependence */
+int amdgpu_ras_pre_fini(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+
+	if (!adev->ras_enabled || !con)
+		return 0;
+
+
+	/* Need disable ras on all IPs here before ip [hw/sw]fini */
+	if (AMDGPU_RAS_GET_FEATURES(con->features))
+		amdgpu_ras_disable_all_features(adev, 0);
+	amdgpu_ras_recovery_fini(adev);
+	return 0;
+}
+
+int amdgpu_ras_fini(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras_block_list *ras_node, *tmp;
+	struct amdgpu_ras_block_object *obj = NULL;
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+
+	if (!adev->ras_enabled || !con)
+		return 0;
+
+	amdgpu_ras_critical_region_fini(adev);
+	mutex_destroy(&con->critical_region_lock);
+
+	list_for_each_entry_safe(ras_node, tmp, &adev->ras_list, node) {
+		if (ras_node->ras_obj) {
+			obj = ras_node->ras_obj;
+			if (amdgpu_ras_is_supported(adev, obj->ras_comm.block) &&
+			    obj->ras_fini)
+				obj->ras_fini(adev, &obj->ras_comm);
+			else
+				amdgpu_ras_block_late_fini_default(adev, &obj->ras_comm);
+		}
+
+		/* Clear ras blocks from ras_list and free ras block list node */
+		list_del(&ras_node->node);
+		kfree(ras_node);
+	}
+
+	amdgpu_ras_fs_fini(adev);
+	amdgpu_ras_interrupt_remove_all(adev);
+
+	if (amdgpu_ras_aca_is_supported(adev)) {
+		if (amdgpu_aca_is_enabled(adev))
+			amdgpu_aca_fini(adev);
+		else
+			amdgpu_mca_fini(adev);
+	}
+
+	WARN(AMDGPU_RAS_GET_FEATURES(con->features), "Feature mask is not cleared");
+
+	if (AMDGPU_RAS_GET_FEATURES(con->features))
+		amdgpu_ras_disable_all_features(adev, 0);
+
+	cancel_delayed_work_sync(&con->ras_counte_delay_work);
+
+	amdgpu_ras_set_context(adev, NULL);
+	kfree(con);
+
+	return 0;
+}
+
+bool amdgpu_ras_get_fed_status(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *ras;
+
+	ras = amdgpu_ras_get_context(adev);
+	if (!ras)
+		return false;
+
+	return test_bit(AMDGPU_RAS_BLOCK__LAST, &ras->ras_err_state);
+}
+
+void amdgpu_ras_set_fed(struct amdgpu_device *adev, bool status)
+{
+	struct amdgpu_ras *ras;
+
+	ras = amdgpu_ras_get_context(adev);
+	if (ras) {
+		if (status)
+			set_bit(AMDGPU_RAS_BLOCK__LAST, &ras->ras_err_state);
+		else
+			clear_bit(AMDGPU_RAS_BLOCK__LAST, &ras->ras_err_state);
+	}
+}
+
+void amdgpu_ras_clear_err_state(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *ras;
+
+	ras = amdgpu_ras_get_context(adev);
+	if (ras) {
+		ras->ras_err_state = 0;
+		ras->gpu_reset_flags = 0;
+	}
+}
+
+void amdgpu_ras_set_err_poison(struct amdgpu_device *adev,
+			       enum amdgpu_ras_block block)
+{
+	struct amdgpu_ras *ras;
+
+	ras = amdgpu_ras_get_context(adev);
+	if (ras)
+		set_bit(block, &ras->ras_err_state);
+}
+
+bool amdgpu_ras_is_err_state(struct amdgpu_device *adev, int block)
+{
+	struct amdgpu_ras *ras;
+
+	ras = amdgpu_ras_get_context(adev);
+	if (ras) {
+		if (block == AMDGPU_RAS_BLOCK__ANY)
+			return (ras->ras_err_state != 0);
+		else
+			return test_bit(block, &ras->ras_err_state) ||
+			       test_bit(AMDGPU_RAS_BLOCK__LAST,
+					&ras->ras_err_state);
+	}
+
+	return false;
+}
+
+static struct ras_event_manager *__get_ras_event_mgr(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *ras;
+
+	ras = amdgpu_ras_get_context(adev);
+	if (!ras)
+		return NULL;
+
+	return ras->event_mgr;
+}
+
+int amdgpu_ras_mark_ras_event_caller(struct amdgpu_device *adev, enum ras_event_type type,
+				     const void *caller)
+{
+	struct ras_event_manager *event_mgr;
+	struct ras_event_state *event_state;
+	int ret = 0;
+
+	if (type >= RAS_EVENT_TYPE_COUNT) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	event_mgr = __get_ras_event_mgr(adev);
+	if (!event_mgr) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	event_state = &event_mgr->event_state[type];
+	event_state->last_seqno = atomic64_inc_return(&event_mgr->seqno);
+	atomic64_inc(&event_state->count);
+
+out:
+	if (ret && caller)
+		dev_warn(adev->dev, "failed mark ras event (%d) in %ps, ret:%d\n",
+			 (int)type, caller, ret);
+
+	return ret;
+}
+
+u64 amdgpu_ras_acquire_event_id(struct amdgpu_device *adev, enum ras_event_type type)
+{
+	struct ras_event_manager *event_mgr;
+	u64 id;
+
+	if (type >= RAS_EVENT_TYPE_COUNT)
+		return RAS_EVENT_INVALID_ID;
+
+	switch (type) {
+	case RAS_EVENT_TYPE_FATAL:
+	case RAS_EVENT_TYPE_POISON_CREATION:
+	case RAS_EVENT_TYPE_POISON_CONSUMPTION:
+		event_mgr = __get_ras_event_mgr(adev);
+		if (!event_mgr)
+			return RAS_EVENT_INVALID_ID;
+
+		id = event_mgr->event_state[type].last_seqno;
+		break;
+	case RAS_EVENT_TYPE_INVALID:
+	default:
+		id = RAS_EVENT_INVALID_ID;
+		break;
+	}
+
+	return id;
+}
+
+void amdgpu_ras_global_ras_isr(struct amdgpu_device *adev)
+{
+	if (atomic_cmpxchg(&amdgpu_ras_in_intr, 0, 1) == 0) {
+		struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+		enum ras_event_type type = RAS_EVENT_TYPE_FATAL;
+		u64 event_id;
+
+		if (amdgpu_ras_mark_ras_event(adev, type)) {
+			dev_err(adev->dev,
+				"uncorrectable hardware error (ERREVENT_ATHUB_INTERRUPT) detected!\n");
+			return;
+		}
+
+		event_id = amdgpu_ras_acquire_event_id(adev, type);
+
+		RAS_EVENT_LOG(adev, event_id, "uncorrectable hardware error"
+			      "(ERREVENT_ATHUB_INTERRUPT) detected!\n");
+
+		amdgpu_ras_set_fed(adev, true);
+		ras->gpu_reset_flags |= AMDGPU_RAS_GPU_RESET_MODE1_RESET;
+		amdgpu_ras_reset_gpu(adev);
+	}
+}
+
+bool amdgpu_ras_need_emergency_restart(struct amdgpu_device *adev)
+{
+	if (adev->asic_type == CHIP_VEGA20 &&
+	    adev->pm.fw_version <= 0x283400) {
+		return !(amdgpu_asic_reset_method(adev) == AMD_RESET_METHOD_BACO) &&
+				amdgpu_ras_intr_triggered();
+	}
+
+	return false;
+}
+
+void amdgpu_release_ras_context(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+
+	if (!con)
+		return;
+
+	if (!adev->ras_enabled && con->features & BIT(AMDGPU_RAS_BLOCK__GFX)) {
+		con->features &= ~BIT(AMDGPU_RAS_BLOCK__GFX);
+		amdgpu_ras_set_context(adev, NULL);
+		kfree(con);
+	}
+}
+
+#ifdef CONFIG_X86_MCE_AMD
+static struct amdgpu_device *find_adev(uint32_t node_id)
+{
+	int i;
+	struct amdgpu_device *adev = NULL;
+
+	for (i = 0; i < mce_adev_list.num_gpu; i++) {
+		adev = mce_adev_list.devs[i];
+
+		if (adev && adev->gmc.xgmi.connected_to_cpu &&
+		    adev->gmc.xgmi.physical_node_id == node_id)
+			break;
+		adev = NULL;
+	}
+
+	return adev;
+}
+
+#define GET_MCA_IPID_GPUID(m)	(((m) >> 44) & 0xF)
+#define GET_UMC_INST(m)		(((m) >> 21) & 0x7)
+#define GET_CHAN_INDEX(m)	((((m) >> 12) & 0x3) | (((m) >> 18) & 0x4))
+#define GPU_ID_OFFSET		8
+
+static int amdgpu_bad_page_notifier(struct notifier_block *nb,
+				    unsigned long val, void *data)
+{
+	struct mce *m = (struct mce *)data;
+	struct amdgpu_device *adev = NULL;
+	uint32_t gpu_id = 0;
+	uint32_t umc_inst = 0, ch_inst = 0;
+
+	/*
+	 * If the error was generated in UMC_V2, which belongs to GPU UMCs,
+	 * and error occurred in DramECC (Extended error code = 0) then only
+	 * process the error, else bail out.
+	 */
+	if (!m || !((smca_get_bank_type(m->extcpu, m->bank) == SMCA_UMC_V2) &&
+		    (XEC(m->status, 0x3f) == 0x0)))
+		return NOTIFY_DONE;
+
+	/*
+	 * If it is correctable error, return.
+	 */
+	if (mce_is_correctable(m))
+		return NOTIFY_OK;
+
+	/*
+	 * GPU Id is offset by GPU_ID_OFFSET in MCA_IPID_UMC register.
+	 */
+	gpu_id = GET_MCA_IPID_GPUID(m->ipid) - GPU_ID_OFFSET;
+
+	adev = find_adev(gpu_id);
+	if (!adev) {
+		DRM_WARN("%s: Unable to find adev for gpu_id: %d\n", __func__,
+								gpu_id);
+		return NOTIFY_DONE;
+	}
+
+	/*
+	 * If it is uncorrectable error, then find out UMC instance and
+	 * channel index.
+	 */
+	umc_inst = GET_UMC_INST(m->ipid);
+	ch_inst = GET_CHAN_INDEX(m->ipid);
+
+	dev_info(adev->dev, "Uncorrectable error detected in UMC inst: %d, chan_idx: %d",
+			     umc_inst, ch_inst);
+
+	if (!amdgpu_umc_page_retirement_mca(adev, m->addr, ch_inst, umc_inst))
+		return NOTIFY_OK;
+	else
+		return NOTIFY_DONE;
+}
+
+static struct notifier_block amdgpu_bad_page_nb = {
+	.notifier_call  = amdgpu_bad_page_notifier,
+	.priority       = MCE_PRIO_UC,
+};
+
+static void amdgpu_register_bad_pages_mca_notifier(struct amdgpu_device *adev)
+{
+	/*
+	 * Add the adev to the mce_adev_list.
+	 * During mode2 reset, amdgpu device is temporarily
+	 * removed from the mgpu_info list which can cause
+	 * page retirement to fail.
+	 * Use this list instead of mgpu_info to find the amdgpu
+	 * device on which the UMC error was reported.
+	 */
+	mce_adev_list.devs[mce_adev_list.num_gpu++] = adev;
+
+	/*
+	 * Register the x86 notifier only once
+	 * with MCE subsystem.
+	 */
+	if (notifier_registered == false) {
+		mce_register_decode_chain(&amdgpu_bad_page_nb);
+		notifier_registered = true;
+	}
+}
+#endif
+
+struct amdgpu_ras *amdgpu_ras_get_context(struct amdgpu_device *adev)
+{
+	if (!adev)
+		return NULL;
+
+	return adev->psp.ras_context.ras;
+}
+
+int amdgpu_ras_set_context(struct amdgpu_device *adev, struct amdgpu_ras *ras_con)
+{
+	if (!adev)
+		return -EINVAL;
+
+	adev->psp.ras_context.ras = ras_con;
+	return 0;
+}
+
+/* check if ras is supported on block, say, sdma, gfx */
+int amdgpu_ras_is_supported(struct amdgpu_device *adev,
+		unsigned int block)
+{
+	int ret = 0;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+
+	if (block >= AMDGPU_RAS_BLOCK_COUNT)
+		return 0;
+
+	ret = ras && (adev->ras_enabled & (1 << block));
+
+	/* For the special asic with mem ecc enabled but sram ecc
+	 * not enabled, even if the ras block is not supported on
+	 * .ras_enabled, if the asic supports poison mode and the
+	 * ras block has ras configuration, it can be considered
+	 * that the ras block supports ras function.
+	 */
+	if (!ret &&
+	    (block == AMDGPU_RAS_BLOCK__GFX ||
+	     block == AMDGPU_RAS_BLOCK__SDMA ||
+	     block == AMDGPU_RAS_BLOCK__VCN ||
+	     block == AMDGPU_RAS_BLOCK__JPEG) &&
+		(amdgpu_ras_mask & (1 << block)) &&
+	    amdgpu_ras_is_poison_mode_supported(adev) &&
+	    amdgpu_ras_get_ras_block(adev, block, 0))
+		ret = 1;
+
+	return ret;
+}
+
+int amdgpu_ras_reset_gpu(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+
+	/* mode1 is the only selection for RMA status */
+	if (amdgpu_ras_is_rma(adev)) {
+		ras->gpu_reset_flags = 0;
+		ras->gpu_reset_flags |= AMDGPU_RAS_GPU_RESET_MODE1_RESET;
+	}
+
+	if (atomic_cmpxchg(&ras->in_recovery, 0, 1) == 0) {
+		struct amdgpu_hive_info *hive = amdgpu_get_xgmi_hive(adev);
+		int hive_ras_recovery = 0;
+
+		if (hive) {
+			hive_ras_recovery = atomic_read(&hive->ras_recovery);
+			amdgpu_put_xgmi_hive(hive);
+		}
+		/* In the case of multiple GPUs, after a GPU has started
+		 * resetting all GPUs on hive, other GPUs do not need to
+		 * trigger GPU reset again.
+		 */
+		if (!hive_ras_recovery)
+			amdgpu_reset_domain_schedule(ras->adev->reset_domain, &ras->recovery_work);
+		else
+			atomic_set(&ras->in_recovery, 0);
+	} else {
+		flush_work(&ras->recovery_work);
+		amdgpu_reset_domain_schedule(ras->adev->reset_domain, &ras->recovery_work);
+	}
+
+	return 0;
+}
+
+int amdgpu_ras_set_mca_debug_mode(struct amdgpu_device *adev, bool enable)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	int ret = 0;
+
+	if (con) {
+		ret = amdgpu_mca_smu_set_debug_mode(adev, enable);
+		if (!ret)
+			con->is_aca_debug_mode = enable;
+	}
+
+	return ret;
+}
+
+int amdgpu_ras_set_aca_debug_mode(struct amdgpu_device *adev, bool enable)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	int ret = 0;
+
+	if (con) {
+		if (amdgpu_aca_is_enabled(adev))
+			ret = amdgpu_aca_smu_set_debug_mode(adev, enable);
+		else
+			ret = amdgpu_mca_smu_set_debug_mode(adev, enable);
+		if (!ret)
+			con->is_aca_debug_mode = enable;
+	}
+
+	return ret;
+}
+
+bool amdgpu_ras_get_aca_debug_mode(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	const struct aca_smu_funcs *smu_funcs = adev->aca.smu_funcs;
+	const struct amdgpu_mca_smu_funcs *mca_funcs = adev->mca.mca_funcs;
+
+	if (!con)
+		return false;
+
+	if ((amdgpu_aca_is_enabled(adev) && smu_funcs && smu_funcs->set_debug_mode) ||
+	    (!amdgpu_aca_is_enabled(adev) && mca_funcs && mca_funcs->mca_set_debug_mode))
+		return con->is_aca_debug_mode;
+	else
+		return true;
+}
+
+bool amdgpu_ras_get_error_query_mode(struct amdgpu_device *adev,
+				     unsigned int *error_query_mode)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	const struct amdgpu_mca_smu_funcs *mca_funcs = adev->mca.mca_funcs;
+	const struct aca_smu_funcs *smu_funcs = adev->aca.smu_funcs;
+
+	if (!con) {
+		*error_query_mode = AMDGPU_RAS_INVALID_ERROR_QUERY;
+		return false;
+	}
+
+	if (amdgpu_sriov_vf(adev)) {
+		*error_query_mode = AMDGPU_RAS_VIRT_ERROR_COUNT_QUERY;
+	} else if ((smu_funcs && smu_funcs->set_debug_mode) || (mca_funcs && mca_funcs->mca_set_debug_mode)) {
+		*error_query_mode =
+			(con->is_aca_debug_mode) ? AMDGPU_RAS_DIRECT_ERROR_QUERY : AMDGPU_RAS_FIRMWARE_ERROR_QUERY;
+	} else {
+		*error_query_mode = AMDGPU_RAS_DIRECT_ERROR_QUERY;
+	}
+
+	return true;
+}
+
+/* Register each ip ras block into amdgpu ras */
+int amdgpu_ras_register_ras_block(struct amdgpu_device *adev,
+		struct amdgpu_ras_block_object *ras_block_obj)
+{
+	struct amdgpu_ras_block_list *ras_node;
+	if (!adev || !ras_block_obj)
+		return -EINVAL;
+
+	ras_node = kzalloc(sizeof(*ras_node), GFP_KERNEL);
+	if (!ras_node)
+		return -ENOMEM;
+
+	INIT_LIST_HEAD(&ras_node->node);
+	ras_node->ras_obj = ras_block_obj;
+	list_add_tail(&ras_node->node, &adev->ras_list);
+
+	return 0;
+}
+
+void amdgpu_ras_get_error_type_name(uint32_t err_type, char *err_type_name)
+{
+	if (!err_type_name)
+		return;
+
+	switch (err_type) {
+	case AMDGPU_RAS_ERROR__SINGLE_CORRECTABLE:
+		sprintf(err_type_name, "correctable");
+		break;
+	case AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE:
+		sprintf(err_type_name, "uncorrectable");
+		break;
+	default:
+		sprintf(err_type_name, "unknown");
+		break;
+	}
+}
+
+bool amdgpu_ras_inst_get_memory_id_field(struct amdgpu_device *adev,
+					 const struct amdgpu_ras_err_status_reg_entry *reg_entry,
+					 uint32_t instance,
+					 uint32_t *memory_id)
+{
+	uint32_t err_status_lo_data, err_status_lo_offset;
+
+	if (!reg_entry)
+		return false;
+
+	err_status_lo_offset =
+		AMDGPU_RAS_REG_ENTRY_OFFSET(reg_entry->hwip, instance,
+					    reg_entry->seg_lo, reg_entry->reg_lo);
+	err_status_lo_data = RREG32(err_status_lo_offset);
+
+	if ((reg_entry->flags & AMDGPU_RAS_ERR_STATUS_VALID) &&
+	    !REG_GET_FIELD(err_status_lo_data, ERR_STATUS_LO, ERR_STATUS_VALID_FLAG))
+		return false;
+
+	*memory_id = REG_GET_FIELD(err_status_lo_data, ERR_STATUS_LO, MEMORY_ID);
+
+	return true;
+}
+
+bool amdgpu_ras_inst_get_err_cnt_field(struct amdgpu_device *adev,
+				       const struct amdgpu_ras_err_status_reg_entry *reg_entry,
+				       uint32_t instance,
+				       unsigned long *err_cnt)
+{
+	uint32_t err_status_hi_data, err_status_hi_offset;
+
+	if (!reg_entry)
+		return false;
+
+	err_status_hi_offset =
+		AMDGPU_RAS_REG_ENTRY_OFFSET(reg_entry->hwip, instance,
+					    reg_entry->seg_hi, reg_entry->reg_hi);
+	err_status_hi_data = RREG32(err_status_hi_offset);
+
+	if ((reg_entry->flags & AMDGPU_RAS_ERR_INFO_VALID) &&
+	    !REG_GET_FIELD(err_status_hi_data, ERR_STATUS_HI, ERR_INFO_VALID_FLAG))
+		/* keep the check here in case we need to refer to the result later */
+		dev_dbg(adev->dev, "Invalid err_info field\n");
+
+	/* read err count */
+	*err_cnt = REG_GET_FIELD(err_status_hi_data, ERR_STATUS, ERR_CNT);
+
+	return true;
+}
+
+void amdgpu_ras_inst_query_ras_error_count(struct amdgpu_device *adev,
+					   const struct amdgpu_ras_err_status_reg_entry *reg_list,
+					   uint32_t reg_list_size,
+					   const struct amdgpu_ras_memory_id_entry *mem_list,
+					   uint32_t mem_list_size,
+					   uint32_t instance,
+					   uint32_t err_type,
+					   unsigned long *err_count)
+{
+	uint32_t memory_id;
+	unsigned long err_cnt;
+	char err_type_name[16];
+	uint32_t i, j;
+
+	for (i = 0; i < reg_list_size; i++) {
+		/* query memory_id from err_status_lo */
+		if (!amdgpu_ras_inst_get_memory_id_field(adev, &reg_list[i],
+							 instance, &memory_id))
+			continue;
+
+		/* query err_cnt from err_status_hi */
+		if (!amdgpu_ras_inst_get_err_cnt_field(adev, &reg_list[i],
+						       instance, &err_cnt) ||
+		    !err_cnt)
+			continue;
+
+		*err_count += err_cnt;
+
+		/* log the errors */
+		amdgpu_ras_get_error_type_name(err_type, err_type_name);
+		if (!mem_list) {
+			/* memory_list is not supported */
+			dev_info(adev->dev,
+				 "%ld %s hardware errors detected in %s, instance: %d, memory_id: %d\n",
+				 err_cnt, err_type_name,
+				 reg_list[i].block_name,
+				 instance, memory_id);
+		} else {
+			for (j = 0; j < mem_list_size; j++) {
+				if (memory_id == mem_list[j].memory_id) {
+					dev_info(adev->dev,
+						 "%ld %s hardware errors detected in %s, instance: %d, memory block: %s\n",
+						 err_cnt, err_type_name,
+						 reg_list[i].block_name,
+						 instance, mem_list[j].name);
+					break;
+				}
+			}
+		}
+	}
+}
+
+void amdgpu_ras_inst_reset_ras_error_count(struct amdgpu_device *adev,
+					   const struct amdgpu_ras_err_status_reg_entry *reg_list,
+					   uint32_t reg_list_size,
+					   uint32_t instance)
+{
+	uint32_t err_status_lo_offset, err_status_hi_offset;
+	uint32_t i;
+
+	for (i = 0; i < reg_list_size; i++) {
+		err_status_lo_offset =
+			AMDGPU_RAS_REG_ENTRY_OFFSET(reg_list[i].hwip, instance,
+						    reg_list[i].seg_lo, reg_list[i].reg_lo);
+		err_status_hi_offset =
+			AMDGPU_RAS_REG_ENTRY_OFFSET(reg_list[i].hwip, instance,
+						    reg_list[i].seg_hi, reg_list[i].reg_hi);
+		WREG32(err_status_lo_offset, 0);
+		WREG32(err_status_hi_offset, 0);
+	}
+}
+
+int amdgpu_ras_error_data_init(struct ras_err_data *err_data)
+{
+	memset(err_data, 0, sizeof(*err_data));
+
+	INIT_LIST_HEAD(&err_data->err_node_list);
+
+	return 0;
+}
+
+static void amdgpu_ras_error_node_release(struct ras_err_node *err_node)
+{
+	if (!err_node)
+		return;
+
+	list_del(&err_node->node);
+	kvfree(err_node);
+}
+
+void amdgpu_ras_error_data_fini(struct ras_err_data *err_data)
+{
+	struct ras_err_node *err_node, *tmp;
+
+	list_for_each_entry_safe(err_node, tmp, &err_data->err_node_list, node)
+		amdgpu_ras_error_node_release(err_node);
+}
+
+static struct ras_err_node *amdgpu_ras_error_find_node_by_id(struct ras_err_data *err_data,
+							     struct amdgpu_smuio_mcm_config_info *mcm_info)
+{
+	struct ras_err_node *err_node;
+	struct amdgpu_smuio_mcm_config_info *ref_id;
+
+	if (!err_data || !mcm_info)
+		return NULL;
+
+	for_each_ras_error(err_node, err_data) {
+		ref_id = &err_node->err_info.mcm_info;
+
+		if (mcm_info->socket_id == ref_id->socket_id &&
+		    mcm_info->die_id == ref_id->die_id)
+			return err_node;
+	}
+
+	return NULL;
+}
+
+static struct ras_err_node *amdgpu_ras_error_node_new(void)
+{
+	struct ras_err_node *err_node;
+
+	err_node = kvzalloc(sizeof(*err_node), GFP_KERNEL);
+	if (!err_node)
+		return NULL;
+
+	INIT_LIST_HEAD(&err_node->node);
+
+	return err_node;
+}
+
+static int ras_err_info_cmp(void *priv, const struct list_head *a, const struct list_head *b)
+{
+	struct ras_err_node *nodea = container_of(a, struct ras_err_node, node);
+	struct ras_err_node *nodeb = container_of(b, struct ras_err_node, node);
+	struct amdgpu_smuio_mcm_config_info *infoa = &nodea->err_info.mcm_info;
+	struct amdgpu_smuio_mcm_config_info *infob = &nodeb->err_info.mcm_info;
+
+	if (unlikely(infoa->socket_id != infob->socket_id))
+		return infoa->socket_id - infob->socket_id;
+	else
+		return infoa->die_id - infob->die_id;
+
+	return 0;
+}
+
+static struct ras_err_info *amdgpu_ras_error_get_info(struct ras_err_data *err_data,
+				struct amdgpu_smuio_mcm_config_info *mcm_info)
+{
+	struct ras_err_node *err_node;
+
+	err_node = amdgpu_ras_error_find_node_by_id(err_data, mcm_info);
+	if (err_node)
+		return &err_node->err_info;
+
+	err_node = amdgpu_ras_error_node_new();
+	if (!err_node)
+		return NULL;
+
+	memcpy(&err_node->err_info.mcm_info, mcm_info, sizeof(*mcm_info));
+
+	err_data->err_list_count++;
+	list_add_tail(&err_node->node, &err_data->err_node_list);
+	list_sort(NULL, &err_data->err_node_list, ras_err_info_cmp);
+
+	return &err_node->err_info;
+}
+
+int amdgpu_ras_error_statistic_ue_count(struct ras_err_data *err_data,
+					struct amdgpu_smuio_mcm_config_info *mcm_info,
+					u64 count)
+{
+	struct ras_err_info *err_info;
+
+	if (!err_data || !mcm_info)
+		return -EINVAL;
+
+	if (!count)
+		return 0;
+
+	err_info = amdgpu_ras_error_get_info(err_data, mcm_info);
+	if (!err_info)
+		return -EINVAL;
+
+	err_info->ue_count += count;
+	err_data->ue_count += count;
+
+	return 0;
+}
+
+int amdgpu_ras_error_statistic_ce_count(struct ras_err_data *err_data,
+					struct amdgpu_smuio_mcm_config_info *mcm_info,
+					u64 count)
+{
+	struct ras_err_info *err_info;
+
+	if (!err_data || !mcm_info)
+		return -EINVAL;
+
+	if (!count)
+		return 0;
+
+	err_info = amdgpu_ras_error_get_info(err_data, mcm_info);
+	if (!err_info)
+		return -EINVAL;
+
+	err_info->ce_count += count;
+	err_data->ce_count += count;
+
+	return 0;
+}
+
+int amdgpu_ras_error_statistic_de_count(struct ras_err_data *err_data,
+					struct amdgpu_smuio_mcm_config_info *mcm_info,
+					u64 count)
+{
+	struct ras_err_info *err_info;
+
+	if (!err_data || !mcm_info)
+		return -EINVAL;
+
+	if (!count)
+		return 0;
+
+	err_info = amdgpu_ras_error_get_info(err_data, mcm_info);
+	if (!err_info)
+		return -EINVAL;
+
+	err_info->de_count += count;
+	err_data->de_count += count;
+
+	return 0;
+}
+
+#define mmMP0_SMN_C2PMSG_92	0x1609C
+#define mmMP0_SMN_C2PMSG_126	0x160BE
+static void amdgpu_ras_boot_time_error_reporting(struct amdgpu_device *adev,
+						 u32 instance)
+{
+	u32 socket_id, aid_id, hbm_id;
+	u32 fw_status;
+	u32 boot_error;
+	u64 reg_addr;
+
+	/* The pattern for smn addressing in other SOC could be different from
+	 * the one for aqua_vanjaram. We should revisit the code if the pattern
+	 * is changed. In such case, replace the aqua_vanjaram implementation
+	 * with more common helper */
+	reg_addr = (mmMP0_SMN_C2PMSG_92 << 2) +
+		   aqua_vanjaram_encode_ext_smn_addressing(instance);
+	fw_status = amdgpu_device_indirect_rreg_ext(adev, reg_addr);
+
+	reg_addr = (mmMP0_SMN_C2PMSG_126 << 2) +
+		   aqua_vanjaram_encode_ext_smn_addressing(instance);
+	boot_error = amdgpu_device_indirect_rreg_ext(adev, reg_addr);
+
+	socket_id = AMDGPU_RAS_GPU_ERR_SOCKET_ID(boot_error);
+	aid_id = AMDGPU_RAS_GPU_ERR_AID_ID(boot_error);
+	hbm_id = ((1 == AMDGPU_RAS_GPU_ERR_HBM_ID(boot_error)) ? 0 : 1);
+
+	if (AMDGPU_RAS_GPU_ERR_MEM_TRAINING(boot_error))
+		dev_info(adev->dev,
+			 "socket: %d, aid: %d, hbm: %d, fw_status: 0x%x, memory training failed\n",
+			 socket_id, aid_id, hbm_id, fw_status);
+
+	if (AMDGPU_RAS_GPU_ERR_FW_LOAD(boot_error))
+		dev_info(adev->dev,
+			 "socket: %d, aid: %d, fw_status: 0x%x, firmware load failed at boot time\n",
+			 socket_id, aid_id, fw_status);
+
+	if (AMDGPU_RAS_GPU_ERR_WAFL_LINK_TRAINING(boot_error))
+		dev_info(adev->dev,
+			 "socket: %d, aid: %d, fw_status: 0x%x, wafl link training failed\n",
+			 socket_id, aid_id, fw_status);
+
+	if (AMDGPU_RAS_GPU_ERR_XGMI_LINK_TRAINING(boot_error))
+		dev_info(adev->dev,
+			 "socket: %d, aid: %d, fw_status: 0x%x, xgmi link training failed\n",
+			 socket_id, aid_id, fw_status);
+
+	if (AMDGPU_RAS_GPU_ERR_USR_CP_LINK_TRAINING(boot_error))
+		dev_info(adev->dev,
+			 "socket: %d, aid: %d, fw_status: 0x%x, usr cp link training failed\n",
+			 socket_id, aid_id, fw_status);
+
+	if (AMDGPU_RAS_GPU_ERR_USR_DP_LINK_TRAINING(boot_error))
+		dev_info(adev->dev,
+			 "socket: %d, aid: %d, fw_status: 0x%x, usr dp link training failed\n",
+			 socket_id, aid_id, fw_status);
+
+	if (AMDGPU_RAS_GPU_ERR_HBM_MEM_TEST(boot_error))
+		dev_info(adev->dev,
+			 "socket: %d, aid: %d, hbm: %d, fw_status: 0x%x, hbm memory test failed\n",
+			 socket_id, aid_id, hbm_id, fw_status);
+
+	if (AMDGPU_RAS_GPU_ERR_HBM_BIST_TEST(boot_error))
+		dev_info(adev->dev,
+			 "socket: %d, aid: %d, hbm: %d, fw_status: 0x%x, hbm bist test failed\n",
+			 socket_id, aid_id, hbm_id, fw_status);
+
+	if (AMDGPU_RAS_GPU_ERR_DATA_ABORT(boot_error))
+		dev_info(adev->dev,
+			 "socket: %d, aid: %d, fw_status: 0x%x, data abort exception\n",
+			 socket_id, aid_id, fw_status);
+
+	if (AMDGPU_RAS_GPU_ERR_GENERIC(boot_error))
+		dev_info(adev->dev,
+			 "socket: %d, aid: %d, fw_status: 0x%x, Boot Controller Generic Error\n",
+			 socket_id, aid_id, fw_status);
+}
+
+static bool amdgpu_ras_boot_error_detected(struct amdgpu_device *adev,
+					   u32 instance)
+{
+	u64 reg_addr;
+	u32 reg_data;
+	int retry_loop;
+
+	reg_addr = (mmMP0_SMN_C2PMSG_92 << 2) +
+		   aqua_vanjaram_encode_ext_smn_addressing(instance);
+
+	for (retry_loop = 0; retry_loop < AMDGPU_RAS_BOOT_STATUS_POLLING_LIMIT; retry_loop++) {
+		reg_data = amdgpu_device_indirect_rreg_ext(adev, reg_addr);
+		if ((reg_data & AMDGPU_RAS_BOOT_STATUS_MASK) == AMDGPU_RAS_BOOT_STEADY_STATUS)
+			return false;
+		else
+			msleep(1);
+	}
+
+	return true;
+}
+
+void amdgpu_ras_query_boot_status(struct amdgpu_device *adev, u32 num_instances)
+{
+	u32 i;
+
+	for (i = 0; i < num_instances; i++) {
+		if (amdgpu_ras_boot_error_detected(adev, i))
+			amdgpu_ras_boot_time_error_reporting(adev, i);
+	}
+}
+
+int amdgpu_ras_reserve_page(struct amdgpu_device *adev, uint64_t pfn)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct amdgpu_vram_mgr *mgr = &adev->mman.vram_mgr;
+	uint64_t start = pfn << AMDGPU_GPU_PAGE_SHIFT;
+	int ret = 0;
+
+	if (amdgpu_ras_check_critical_address(adev, start))
+		return 0;
+
+	mutex_lock(&con->page_rsv_lock);
+	ret = amdgpu_vram_mgr_query_page_status(mgr, start);
+	if (ret == -ENOENT)
+		ret = amdgpu_vram_mgr_reserve_range(mgr, start, AMDGPU_GPU_PAGE_SIZE);
+	mutex_unlock(&con->page_rsv_lock);
+
+	return ret;
+}
+
+void amdgpu_ras_event_log_print(struct amdgpu_device *adev, u64 event_id,
+				const char *fmt, ...)
+{
+	struct va_format vaf;
+	va_list args;
+
+	va_start(args, fmt);
+	vaf.fmt = fmt;
+	vaf.va = &args;
+
+	if (RAS_EVENT_ID_IS_VALID(event_id))
+		dev_printk(KERN_INFO, adev->dev, "{%llu}%pV", event_id, &vaf);
+	else
+		dev_printk(KERN_INFO, adev->dev, "%pV", &vaf);
+
+	va_end(args);
+}
+
+bool amdgpu_ras_is_rma(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+
+	if (!con)
+		return false;
+
+	return con->is_rma;
+}
+
+int amdgpu_ras_add_critical_region(struct amdgpu_device *adev,
+			struct amdgpu_bo *bo)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct amdgpu_vram_mgr_resource *vres;
+	struct ras_critical_region *region;
+	struct drm_buddy_block *block;
+	int ret = 0;
+
+	if (!bo || !bo->tbo.resource)
+		return -EINVAL;
+
+	vres = to_amdgpu_vram_mgr_resource(bo->tbo.resource);
+
+	mutex_lock(&con->critical_region_lock);
+
+	/* Check if the bo had been recorded */
+	list_for_each_entry(region, &con->critical_region_head, node)
+		if (region->bo == bo)
+			goto out;
+
+	/* Record new critical amdgpu bo */
+	list_for_each_entry(block, &vres->blocks, link) {
+		region = kzalloc(sizeof(*region), GFP_KERNEL);
+		if (!region) {
+			ret = -ENOMEM;
+			goto out;
+		}
+		region->bo = bo;
+		region->start = amdgpu_vram_mgr_block_start(block);
+		region->size = amdgpu_vram_mgr_block_size(block);
+		list_add_tail(&region->node, &con->critical_region_head);
+	}
+
+out:
+	mutex_unlock(&con->critical_region_lock);
+
+	return ret;
+}
+
+static void amdgpu_ras_critical_region_init(struct amdgpu_device *adev)
+{
+	amdgpu_ras_add_critical_region(adev, adev->mman.fw_reserved_memory);
+}
+
+static void amdgpu_ras_critical_region_fini(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_critical_region *region, *tmp;
+
+	mutex_lock(&con->critical_region_lock);
+	list_for_each_entry_safe(region, tmp, &con->critical_region_head, node) {
+		list_del(&region->node);
+		kfree(region);
+	}
+	mutex_unlock(&con->critical_region_lock);
+}
+
+bool amdgpu_ras_check_critical_address(struct amdgpu_device *adev, uint64_t addr)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_critical_region *region;
+	bool ret = false;
+
+	mutex_lock(&con->critical_region_lock);
+	list_for_each_entry(region, &con->critical_region_head, node) {
+		if ((region->start <= addr) &&
+		    (addr < (region->start + region->size))) {
+			ret = true;
+			break;
+		}
+	}
+	mutex_unlock(&con->critical_region_lock);
+
+	return ret;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
new file mode 100644
index 000000000000..6cf0dfd38be8
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
@@ -0,0 +1,1011 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ *
+ */
+#ifndef _AMDGPU_RAS_H
+#define _AMDGPU_RAS_H
+
+#include <linux/debugfs.h>
+#include <linux/list.h>
+#include <linux/kfifo.h>
+#include <linux/radix-tree.h>
+#include "ta_ras_if.h"
+#include "amdgpu_ras_eeprom.h"
+#include "amdgpu_smuio.h"
+#include "amdgpu_aca.h"
+
+struct amdgpu_iv_entry;
+
+#define AMDGPU_RAS_GPU_ERR_MEM_TRAINING(x)		AMDGPU_GET_REG_FIELD(x, 0, 0)
+#define AMDGPU_RAS_GPU_ERR_FW_LOAD(x)			AMDGPU_GET_REG_FIELD(x, 1, 1)
+#define AMDGPU_RAS_GPU_ERR_WAFL_LINK_TRAINING(x)	AMDGPU_GET_REG_FIELD(x, 2, 2)
+#define AMDGPU_RAS_GPU_ERR_XGMI_LINK_TRAINING(x)	AMDGPU_GET_REG_FIELD(x, 3, 3)
+#define AMDGPU_RAS_GPU_ERR_USR_CP_LINK_TRAINING(x)	AMDGPU_GET_REG_FIELD(x, 4, 4)
+#define AMDGPU_RAS_GPU_ERR_USR_DP_LINK_TRAINING(x)	AMDGPU_GET_REG_FIELD(x, 5, 5)
+#define AMDGPU_RAS_GPU_ERR_HBM_MEM_TEST(x)		AMDGPU_GET_REG_FIELD(x, 6, 6)
+#define AMDGPU_RAS_GPU_ERR_HBM_BIST_TEST(x)		AMDGPU_GET_REG_FIELD(x, 7, 7)
+#define AMDGPU_RAS_GPU_ERR_SOCKET_ID(x)			AMDGPU_GET_REG_FIELD(x, 10, 8)
+#define AMDGPU_RAS_GPU_ERR_AID_ID(x)			AMDGPU_GET_REG_FIELD(x, 12, 11)
+#define AMDGPU_RAS_GPU_ERR_HBM_ID(x)			AMDGPU_GET_REG_FIELD(x, 14, 13)
+#define AMDGPU_RAS_GPU_ERR_DATA_ABORT(x)		AMDGPU_GET_REG_FIELD(x, 29, 29)
+#define AMDGPU_RAS_GPU_ERR_GENERIC(x)			AMDGPU_GET_REG_FIELD(x, 30, 30)
+
+#define AMDGPU_RAS_BOOT_STATUS_POLLING_LIMIT	100
+#define AMDGPU_RAS_BOOT_STEADY_STATUS		0xBA
+#define AMDGPU_RAS_BOOT_STATUS_MASK		0xFF
+
+#define AMDGPU_RAS_FLAG_INIT_BY_VBIOS		(0x1 << 0)
+/* position of instance value in sub_block_index of
+ * ta_ras_trigger_error_input, the sub block uses lower 12 bits
+ */
+#define AMDGPU_RAS_INST_MASK 0xfffff000
+#define AMDGPU_RAS_INST_SHIFT 0xc
+
+#define AMDGPU_RAS_FEATURES_SOCKETID_SHIFT 29
+#define AMDGPU_RAS_FEATURES_SOCKETID_MASK 0xe0000000
+
+/* Reserve 8 physical dram row for possible retirement.
+ * In worst cases, it will lose 8 * 2MB memory in vram domain */
+#define AMDGPU_RAS_RESERVED_VRAM_SIZE_DEFAULT	(16ULL << 20)
+/* The high three bits indicates socketid */
+#define AMDGPU_RAS_GET_FEATURES(val)  ((val) & ~AMDGPU_RAS_FEATURES_SOCKETID_MASK)
+
+#define RAS_EVENT_INVALID_ID		(BIT_ULL(63))
+#define RAS_EVENT_ID_IS_VALID(x)	(!((x) & BIT_ULL(63)))
+
+#define RAS_EVENT_LOG(adev, id, fmt, ...)	\
+	amdgpu_ras_event_log_print((adev), (id), (fmt), ##__VA_ARGS__)
+
+#define amdgpu_ras_mark_ras_event(adev, type)	\
+	(amdgpu_ras_mark_ras_event_caller((adev), (type), __builtin_return_address(0)))
+
+enum amdgpu_ras_block {
+	AMDGPU_RAS_BLOCK__UMC = 0,
+	AMDGPU_RAS_BLOCK__SDMA,
+	AMDGPU_RAS_BLOCK__GFX,
+	AMDGPU_RAS_BLOCK__MMHUB,
+	AMDGPU_RAS_BLOCK__ATHUB,
+	AMDGPU_RAS_BLOCK__PCIE_BIF,
+	AMDGPU_RAS_BLOCK__HDP,
+	AMDGPU_RAS_BLOCK__XGMI_WAFL,
+	AMDGPU_RAS_BLOCK__DF,
+	AMDGPU_RAS_BLOCK__SMN,
+	AMDGPU_RAS_BLOCK__SEM,
+	AMDGPU_RAS_BLOCK__MP0,
+	AMDGPU_RAS_BLOCK__MP1,
+	AMDGPU_RAS_BLOCK__FUSE,
+	AMDGPU_RAS_BLOCK__MCA,
+	AMDGPU_RAS_BLOCK__VCN,
+	AMDGPU_RAS_BLOCK__JPEG,
+	AMDGPU_RAS_BLOCK__IH,
+	AMDGPU_RAS_BLOCK__MPIO,
+	AMDGPU_RAS_BLOCK__MMSCH,
+
+	AMDGPU_RAS_BLOCK__LAST,
+	AMDGPU_RAS_BLOCK__ANY = -1
+};
+
+enum amdgpu_ras_mca_block {
+	AMDGPU_RAS_MCA_BLOCK__MP0 = 0,
+	AMDGPU_RAS_MCA_BLOCK__MP1,
+	AMDGPU_RAS_MCA_BLOCK__MPIO,
+	AMDGPU_RAS_MCA_BLOCK__IOHC,
+
+	AMDGPU_RAS_MCA_BLOCK__LAST
+};
+
+#define AMDGPU_RAS_BLOCK_COUNT	AMDGPU_RAS_BLOCK__LAST
+#define AMDGPU_RAS_MCA_BLOCK_COUNT	AMDGPU_RAS_MCA_BLOCK__LAST
+#define AMDGPU_RAS_BLOCK_MASK	((1ULL << AMDGPU_RAS_BLOCK_COUNT) - 1)
+
+enum amdgpu_ras_gfx_subblock {
+	/* CPC */
+	AMDGPU_RAS_BLOCK__GFX_CPC_INDEX_START = 0,
+	AMDGPU_RAS_BLOCK__GFX_CPC_SCRATCH =
+		AMDGPU_RAS_BLOCK__GFX_CPC_INDEX_START,
+	AMDGPU_RAS_BLOCK__GFX_CPC_UCODE,
+	AMDGPU_RAS_BLOCK__GFX_DC_STATE_ME1,
+	AMDGPU_RAS_BLOCK__GFX_DC_CSINVOC_ME1,
+	AMDGPU_RAS_BLOCK__GFX_DC_RESTORE_ME1,
+	AMDGPU_RAS_BLOCK__GFX_DC_STATE_ME2,
+	AMDGPU_RAS_BLOCK__GFX_DC_CSINVOC_ME2,
+	AMDGPU_RAS_BLOCK__GFX_DC_RESTORE_ME2,
+	AMDGPU_RAS_BLOCK__GFX_CPC_INDEX_END =
+		AMDGPU_RAS_BLOCK__GFX_DC_RESTORE_ME2,
+	/* CPF */
+	AMDGPU_RAS_BLOCK__GFX_CPF_INDEX_START,
+	AMDGPU_RAS_BLOCK__GFX_CPF_ROQ_ME2 =
+		AMDGPU_RAS_BLOCK__GFX_CPF_INDEX_START,
+	AMDGPU_RAS_BLOCK__GFX_CPF_ROQ_ME1,
+	AMDGPU_RAS_BLOCK__GFX_CPF_TAG,
+	AMDGPU_RAS_BLOCK__GFX_CPF_INDEX_END = AMDGPU_RAS_BLOCK__GFX_CPF_TAG,
+	/* CPG */
+	AMDGPU_RAS_BLOCK__GFX_CPG_INDEX_START,
+	AMDGPU_RAS_BLOCK__GFX_CPG_DMA_ROQ =
+		AMDGPU_RAS_BLOCK__GFX_CPG_INDEX_START,
+	AMDGPU_RAS_BLOCK__GFX_CPG_DMA_TAG,
+	AMDGPU_RAS_BLOCK__GFX_CPG_TAG,
+	AMDGPU_RAS_BLOCK__GFX_CPG_INDEX_END = AMDGPU_RAS_BLOCK__GFX_CPG_TAG,
+	/* GDS */
+	AMDGPU_RAS_BLOCK__GFX_GDS_INDEX_START,
+	AMDGPU_RAS_BLOCK__GFX_GDS_MEM = AMDGPU_RAS_BLOCK__GFX_GDS_INDEX_START,
+	AMDGPU_RAS_BLOCK__GFX_GDS_INPUT_QUEUE,
+	AMDGPU_RAS_BLOCK__GFX_GDS_OA_PHY_CMD_RAM_MEM,
+	AMDGPU_RAS_BLOCK__GFX_GDS_OA_PHY_DATA_RAM_MEM,
+	AMDGPU_RAS_BLOCK__GFX_GDS_OA_PIPE_MEM,
+	AMDGPU_RAS_BLOCK__GFX_GDS_INDEX_END =
+		AMDGPU_RAS_BLOCK__GFX_GDS_OA_PIPE_MEM,
+	/* SPI */
+	AMDGPU_RAS_BLOCK__GFX_SPI_SR_MEM,
+	/* SQ */
+	AMDGPU_RAS_BLOCK__GFX_SQ_INDEX_START,
+	AMDGPU_RAS_BLOCK__GFX_SQ_SGPR = AMDGPU_RAS_BLOCK__GFX_SQ_INDEX_START,
+	AMDGPU_RAS_BLOCK__GFX_SQ_LDS_D,
+	AMDGPU_RAS_BLOCK__GFX_SQ_LDS_I,
+	AMDGPU_RAS_BLOCK__GFX_SQ_VGPR,
+	AMDGPU_RAS_BLOCK__GFX_SQ_INDEX_END = AMDGPU_RAS_BLOCK__GFX_SQ_VGPR,
+	/* SQC (3 ranges) */
+	AMDGPU_RAS_BLOCK__GFX_SQC_INDEX_START,
+	/* SQC range 0 */
+	AMDGPU_RAS_BLOCK__GFX_SQC_INDEX0_START =
+		AMDGPU_RAS_BLOCK__GFX_SQC_INDEX_START,
+	AMDGPU_RAS_BLOCK__GFX_SQC_INST_UTCL1_LFIFO =
+		AMDGPU_RAS_BLOCK__GFX_SQC_INDEX0_START,
+	AMDGPU_RAS_BLOCK__GFX_SQC_DATA_CU0_WRITE_DATA_BUF,
+	AMDGPU_RAS_BLOCK__GFX_SQC_DATA_CU0_UTCL1_LFIFO,
+	AMDGPU_RAS_BLOCK__GFX_SQC_DATA_CU1_WRITE_DATA_BUF,
+	AMDGPU_RAS_BLOCK__GFX_SQC_DATA_CU1_UTCL1_LFIFO,
+	AMDGPU_RAS_BLOCK__GFX_SQC_DATA_CU2_WRITE_DATA_BUF,
+	AMDGPU_RAS_BLOCK__GFX_SQC_DATA_CU2_UTCL1_LFIFO,
+	AMDGPU_RAS_BLOCK__GFX_SQC_INDEX0_END =
+		AMDGPU_RAS_BLOCK__GFX_SQC_DATA_CU2_UTCL1_LFIFO,
+	/* SQC range 1 */
+	AMDGPU_RAS_BLOCK__GFX_SQC_INDEX1_START,
+	AMDGPU_RAS_BLOCK__GFX_SQC_INST_BANKA_TAG_RAM =
+		AMDGPU_RAS_BLOCK__GFX_SQC_INDEX1_START,
+	AMDGPU_RAS_BLOCK__GFX_SQC_INST_BANKA_UTCL1_MISS_FIFO,
+	AMDGPU_RAS_BLOCK__GFX_SQC_INST_BANKA_MISS_FIFO,
+	AMDGPU_RAS_BLOCK__GFX_SQC_INST_BANKA_BANK_RAM,
+	AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKA_TAG_RAM,
+	AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKA_HIT_FIFO,
+	AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKA_MISS_FIFO,
+	AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKA_DIRTY_BIT_RAM,
+	AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKA_BANK_RAM,
+	AMDGPU_RAS_BLOCK__GFX_SQC_INDEX1_END =
+		AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKA_BANK_RAM,
+	/* SQC range 2 */
+	AMDGPU_RAS_BLOCK__GFX_SQC_INDEX2_START,
+	AMDGPU_RAS_BLOCK__GFX_SQC_INST_BANKB_TAG_RAM =
+		AMDGPU_RAS_BLOCK__GFX_SQC_INDEX2_START,
+	AMDGPU_RAS_BLOCK__GFX_SQC_INST_BANKB_UTCL1_MISS_FIFO,
+	AMDGPU_RAS_BLOCK__GFX_SQC_INST_BANKB_MISS_FIFO,
+	AMDGPU_RAS_BLOCK__GFX_SQC_INST_BANKB_BANK_RAM,
+	AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKB_TAG_RAM,
+	AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKB_HIT_FIFO,
+	AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKB_MISS_FIFO,
+	AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKB_DIRTY_BIT_RAM,
+	AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKB_BANK_RAM,
+	AMDGPU_RAS_BLOCK__GFX_SQC_INDEX2_END =
+		AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKB_BANK_RAM,
+	AMDGPU_RAS_BLOCK__GFX_SQC_INDEX_END =
+		AMDGPU_RAS_BLOCK__GFX_SQC_INDEX2_END,
+	/* TA */
+	AMDGPU_RAS_BLOCK__GFX_TA_INDEX_START,
+	AMDGPU_RAS_BLOCK__GFX_TA_FS_DFIFO =
+		AMDGPU_RAS_BLOCK__GFX_TA_INDEX_START,
+	AMDGPU_RAS_BLOCK__GFX_TA_FS_AFIFO,
+	AMDGPU_RAS_BLOCK__GFX_TA_FL_LFIFO,
+	AMDGPU_RAS_BLOCK__GFX_TA_FX_LFIFO,
+	AMDGPU_RAS_BLOCK__GFX_TA_FS_CFIFO,
+	AMDGPU_RAS_BLOCK__GFX_TA_INDEX_END = AMDGPU_RAS_BLOCK__GFX_TA_FS_CFIFO,
+	/* TCA */
+	AMDGPU_RAS_BLOCK__GFX_TCA_INDEX_START,
+	AMDGPU_RAS_BLOCK__GFX_TCA_HOLE_FIFO =
+		AMDGPU_RAS_BLOCK__GFX_TCA_INDEX_START,
+	AMDGPU_RAS_BLOCK__GFX_TCA_REQ_FIFO,
+	AMDGPU_RAS_BLOCK__GFX_TCA_INDEX_END =
+		AMDGPU_RAS_BLOCK__GFX_TCA_REQ_FIFO,
+	/* TCC (5 sub-ranges) */
+	AMDGPU_RAS_BLOCK__GFX_TCC_INDEX_START,
+	/* TCC range 0 */
+	AMDGPU_RAS_BLOCK__GFX_TCC_INDEX0_START =
+		AMDGPU_RAS_BLOCK__GFX_TCC_INDEX_START,
+	AMDGPU_RAS_BLOCK__GFX_TCC_CACHE_DATA =
+		AMDGPU_RAS_BLOCK__GFX_TCC_INDEX0_START,
+	AMDGPU_RAS_BLOCK__GFX_TCC_CACHE_DATA_BANK_0_1,
+	AMDGPU_RAS_BLOCK__GFX_TCC_CACHE_DATA_BANK_1_0,
+	AMDGPU_RAS_BLOCK__GFX_TCC_CACHE_DATA_BANK_1_1,
+	AMDGPU_RAS_BLOCK__GFX_TCC_CACHE_DIRTY_BANK_0,
+	AMDGPU_RAS_BLOCK__GFX_TCC_CACHE_DIRTY_BANK_1,
+	AMDGPU_RAS_BLOCK__GFX_TCC_HIGH_RATE_TAG,
+	AMDGPU_RAS_BLOCK__GFX_TCC_LOW_RATE_TAG,
+	AMDGPU_RAS_BLOCK__GFX_TCC_INDEX0_END =
+		AMDGPU_RAS_BLOCK__GFX_TCC_LOW_RATE_TAG,
+	/* TCC range 1 */
+	AMDGPU_RAS_BLOCK__GFX_TCC_INDEX1_START,
+	AMDGPU_RAS_BLOCK__GFX_TCC_IN_USE_DEC =
+		AMDGPU_RAS_BLOCK__GFX_TCC_INDEX1_START,
+	AMDGPU_RAS_BLOCK__GFX_TCC_IN_USE_TRANSFER,
+	AMDGPU_RAS_BLOCK__GFX_TCC_INDEX1_END =
+		AMDGPU_RAS_BLOCK__GFX_TCC_IN_USE_TRANSFER,
+	/* TCC range 2 */
+	AMDGPU_RAS_BLOCK__GFX_TCC_INDEX2_START,
+	AMDGPU_RAS_BLOCK__GFX_TCC_RETURN_DATA =
+		AMDGPU_RAS_BLOCK__GFX_TCC_INDEX2_START,
+	AMDGPU_RAS_BLOCK__GFX_TCC_RETURN_CONTROL,
+	AMDGPU_RAS_BLOCK__GFX_TCC_UC_ATOMIC_FIFO,
+	AMDGPU_RAS_BLOCK__GFX_TCC_WRITE_RETURN,
+	AMDGPU_RAS_BLOCK__GFX_TCC_WRITE_CACHE_READ,
+	AMDGPU_RAS_BLOCK__GFX_TCC_SRC_FIFO,
+	AMDGPU_RAS_BLOCK__GFX_TCC_SRC_FIFO_NEXT_RAM,
+	AMDGPU_RAS_BLOCK__GFX_TCC_CACHE_TAG_PROBE_FIFO,
+	AMDGPU_RAS_BLOCK__GFX_TCC_INDEX2_END =
+		AMDGPU_RAS_BLOCK__GFX_TCC_CACHE_TAG_PROBE_FIFO,
+	/* TCC range 3 */
+	AMDGPU_RAS_BLOCK__GFX_TCC_INDEX3_START,
+	AMDGPU_RAS_BLOCK__GFX_TCC_LATENCY_FIFO =
+		AMDGPU_RAS_BLOCK__GFX_TCC_INDEX3_START,
+	AMDGPU_RAS_BLOCK__GFX_TCC_LATENCY_FIFO_NEXT_RAM,
+	AMDGPU_RAS_BLOCK__GFX_TCC_INDEX3_END =
+		AMDGPU_RAS_BLOCK__GFX_TCC_LATENCY_FIFO_NEXT_RAM,
+	/* TCC range 4 */
+	AMDGPU_RAS_BLOCK__GFX_TCC_INDEX4_START,
+	AMDGPU_RAS_BLOCK__GFX_TCC_WRRET_TAG_WRITE_RETURN =
+		AMDGPU_RAS_BLOCK__GFX_TCC_INDEX4_START,
+	AMDGPU_RAS_BLOCK__GFX_TCC_ATOMIC_RETURN_BUFFER,
+	AMDGPU_RAS_BLOCK__GFX_TCC_INDEX4_END =
+		AMDGPU_RAS_BLOCK__GFX_TCC_ATOMIC_RETURN_BUFFER,
+	AMDGPU_RAS_BLOCK__GFX_TCC_INDEX_END =
+		AMDGPU_RAS_BLOCK__GFX_TCC_INDEX4_END,
+	/* TCI */
+	AMDGPU_RAS_BLOCK__GFX_TCI_WRITE_RAM,
+	/* TCP */
+	AMDGPU_RAS_BLOCK__GFX_TCP_INDEX_START,
+	AMDGPU_RAS_BLOCK__GFX_TCP_CACHE_RAM =
+		AMDGPU_RAS_BLOCK__GFX_TCP_INDEX_START,
+	AMDGPU_RAS_BLOCK__GFX_TCP_LFIFO_RAM,
+	AMDGPU_RAS_BLOCK__GFX_TCP_CMD_FIFO,
+	AMDGPU_RAS_BLOCK__GFX_TCP_VM_FIFO,
+	AMDGPU_RAS_BLOCK__GFX_TCP_DB_RAM,
+	AMDGPU_RAS_BLOCK__GFX_TCP_UTCL1_LFIFO0,
+	AMDGPU_RAS_BLOCK__GFX_TCP_UTCL1_LFIFO1,
+	AMDGPU_RAS_BLOCK__GFX_TCP_INDEX_END =
+		AMDGPU_RAS_BLOCK__GFX_TCP_UTCL1_LFIFO1,
+	/* TD */
+	AMDGPU_RAS_BLOCK__GFX_TD_INDEX_START,
+	AMDGPU_RAS_BLOCK__GFX_TD_SS_FIFO_LO =
+		AMDGPU_RAS_BLOCK__GFX_TD_INDEX_START,
+	AMDGPU_RAS_BLOCK__GFX_TD_SS_FIFO_HI,
+	AMDGPU_RAS_BLOCK__GFX_TD_CS_FIFO,
+	AMDGPU_RAS_BLOCK__GFX_TD_INDEX_END = AMDGPU_RAS_BLOCK__GFX_TD_CS_FIFO,
+	/* EA (3 sub-ranges) */
+	AMDGPU_RAS_BLOCK__GFX_EA_INDEX_START,
+	/* EA range 0 */
+	AMDGPU_RAS_BLOCK__GFX_EA_INDEX0_START =
+		AMDGPU_RAS_BLOCK__GFX_EA_INDEX_START,
+	AMDGPU_RAS_BLOCK__GFX_EA_DRAMRD_CMDMEM =
+		AMDGPU_RAS_BLOCK__GFX_EA_INDEX0_START,
+	AMDGPU_RAS_BLOCK__GFX_EA_DRAMWR_CMDMEM,
+	AMDGPU_RAS_BLOCK__GFX_EA_DRAMWR_DATAMEM,
+	AMDGPU_RAS_BLOCK__GFX_EA_RRET_TAGMEM,
+	AMDGPU_RAS_BLOCK__GFX_EA_WRET_TAGMEM,
+	AMDGPU_RAS_BLOCK__GFX_EA_GMIRD_CMDMEM,
+	AMDGPU_RAS_BLOCK__GFX_EA_GMIWR_CMDMEM,
+	AMDGPU_RAS_BLOCK__GFX_EA_GMIWR_DATAMEM,
+	AMDGPU_RAS_BLOCK__GFX_EA_INDEX0_END =
+		AMDGPU_RAS_BLOCK__GFX_EA_GMIWR_DATAMEM,
+	/* EA range 1 */
+	AMDGPU_RAS_BLOCK__GFX_EA_INDEX1_START,
+	AMDGPU_RAS_BLOCK__GFX_EA_DRAMRD_PAGEMEM =
+		AMDGPU_RAS_BLOCK__GFX_EA_INDEX1_START,
+	AMDGPU_RAS_BLOCK__GFX_EA_DRAMWR_PAGEMEM,
+	AMDGPU_RAS_BLOCK__GFX_EA_IORD_CMDMEM,
+	AMDGPU_RAS_BLOCK__GFX_EA_IOWR_CMDMEM,
+	AMDGPU_RAS_BLOCK__GFX_EA_IOWR_DATAMEM,
+	AMDGPU_RAS_BLOCK__GFX_EA_GMIRD_PAGEMEM,
+	AMDGPU_RAS_BLOCK__GFX_EA_GMIWR_PAGEMEM,
+	AMDGPU_RAS_BLOCK__GFX_EA_INDEX1_END =
+		AMDGPU_RAS_BLOCK__GFX_EA_GMIWR_PAGEMEM,
+	/* EA range 2 */
+	AMDGPU_RAS_BLOCK__GFX_EA_INDEX2_START,
+	AMDGPU_RAS_BLOCK__GFX_EA_MAM_D0MEM =
+		AMDGPU_RAS_BLOCK__GFX_EA_INDEX2_START,
+	AMDGPU_RAS_BLOCK__GFX_EA_MAM_D1MEM,
+	AMDGPU_RAS_BLOCK__GFX_EA_MAM_D2MEM,
+	AMDGPU_RAS_BLOCK__GFX_EA_MAM_D3MEM,
+	AMDGPU_RAS_BLOCK__GFX_EA_INDEX2_END =
+		AMDGPU_RAS_BLOCK__GFX_EA_MAM_D3MEM,
+	AMDGPU_RAS_BLOCK__GFX_EA_INDEX_END =
+		AMDGPU_RAS_BLOCK__GFX_EA_INDEX2_END,
+	/* UTC VM L2 bank */
+	AMDGPU_RAS_BLOCK__UTC_VML2_BANK_CACHE,
+	/* UTC VM walker */
+	AMDGPU_RAS_BLOCK__UTC_VML2_WALKER,
+	/* UTC ATC L2 2MB cache */
+	AMDGPU_RAS_BLOCK__UTC_ATCL2_CACHE_2M_BANK,
+	/* UTC ATC L2 4KB cache */
+	AMDGPU_RAS_BLOCK__UTC_ATCL2_CACHE_4K_BANK,
+	AMDGPU_RAS_BLOCK__GFX_MAX
+};
+
+enum amdgpu_ras_error_type {
+	AMDGPU_RAS_ERROR__NONE							= 0,
+	AMDGPU_RAS_ERROR__PARITY						= 1,
+	AMDGPU_RAS_ERROR__SINGLE_CORRECTABLE					= 2,
+	AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE					= 4,
+	AMDGPU_RAS_ERROR__POISON						= 8,
+};
+
+enum amdgpu_ras_ret {
+	AMDGPU_RAS_SUCCESS = 0,
+	AMDGPU_RAS_FAIL,
+	AMDGPU_RAS_UE,
+	AMDGPU_RAS_CE,
+	AMDGPU_RAS_PT,
+};
+
+enum amdgpu_ras_error_query_mode {
+	AMDGPU_RAS_INVALID_ERROR_QUERY		= 0,
+	AMDGPU_RAS_DIRECT_ERROR_QUERY		= 1,
+	AMDGPU_RAS_FIRMWARE_ERROR_QUERY		= 2,
+	AMDGPU_RAS_VIRT_ERROR_COUNT_QUERY	= 3,
+};
+
+/* ras error status reisger fields */
+#define ERR_STATUS_LO__ERR_STATUS_VALID_FLAG__SHIFT	0x0
+#define ERR_STATUS_LO__ERR_STATUS_VALID_FLAG_MASK	0x00000001L
+#define ERR_STATUS_LO__MEMORY_ID__SHIFT			0x18
+#define ERR_STATUS_LO__MEMORY_ID_MASK			0xFF000000L
+#define ERR_STATUS_HI__ERR_INFO_VALID_FLAG__SHIFT	0x2
+#define ERR_STATUS_HI__ERR_INFO_VALID_FLAG_MASK		0x00000004L
+#define ERR_STATUS__ERR_CNT__SHIFT			0x17
+#define ERR_STATUS__ERR_CNT_MASK			0x03800000L
+
+#define AMDGPU_RAS_REG_ENTRY(ip, inst, reg_lo, reg_hi) \
+	ip##_HWIP, inst, reg_lo##_BASE_IDX, reg_lo, reg_hi##_BASE_IDX, reg_hi
+
+#define AMDGPU_RAS_REG_ENTRY_OFFSET(hwip, ip_inst, segment, reg) \
+	(adev->reg_offset[hwip][ip_inst][segment] + (reg))
+
+#define AMDGPU_RAS_ERR_INFO_VALID	(1 << 0)
+#define AMDGPU_RAS_ERR_STATUS_VALID	(1 << 1)
+#define AMDGPU_RAS_ERR_ADDRESS_VALID	(1 << 2)
+
+#define AMDGPU_RAS_GPU_RESET_MODE2_RESET  (0x1 << 0)
+#define AMDGPU_RAS_GPU_RESET_MODE1_RESET  (0x1 << 1)
+
+struct amdgpu_ras_err_status_reg_entry {
+	uint32_t hwip;
+	uint32_t ip_inst;
+	uint32_t seg_lo;
+	uint32_t reg_lo;
+	uint32_t seg_hi;
+	uint32_t reg_hi;
+	uint32_t reg_inst;
+	uint32_t flags;
+	const char *block_name;
+};
+
+struct amdgpu_ras_memory_id_entry {
+	uint32_t memory_id;
+	const char *name;
+};
+
+struct ras_common_if {
+	enum amdgpu_ras_block block;
+	enum amdgpu_ras_error_type type;
+	uint32_t sub_block_index;
+	char name[32];
+};
+
+#define MAX_UMC_CHANNEL_NUM 32
+
+struct ecc_info_per_ch {
+	uint16_t ce_count_lo_chip;
+	uint16_t ce_count_hi_chip;
+	uint64_t mca_umc_status;
+	uint64_t mca_umc_addr;
+	uint64_t mca_ceumc_addr;
+};
+
+struct umc_ecc_info {
+	struct ecc_info_per_ch ecc[MAX_UMC_CHANNEL_NUM];
+
+	/* Determine smu ecctable whether support
+	 * record correctable error address
+	 */
+	int record_ce_addr_supported;
+};
+
+enum ras_event_type {
+	RAS_EVENT_TYPE_INVALID = 0,
+	RAS_EVENT_TYPE_FATAL,
+	RAS_EVENT_TYPE_POISON_CREATION,
+	RAS_EVENT_TYPE_POISON_CONSUMPTION,
+	RAS_EVENT_TYPE_COUNT,
+};
+
+struct ras_event_state {
+	u64 last_seqno;
+	atomic64_t count;
+};
+
+struct ras_event_manager {
+	atomic64_t seqno;
+	struct ras_event_state event_state[RAS_EVENT_TYPE_COUNT];
+};
+
+struct ras_event_id {
+	enum ras_event_type type;
+	u64 event_id;
+};
+
+struct ras_query_context {
+	struct ras_event_id evid;
+};
+
+typedef int (*pasid_notify)(struct amdgpu_device *adev,
+		uint16_t pasid, void *data);
+
+struct ras_poison_msg {
+	enum amdgpu_ras_block block;
+	uint16_t pasid;
+	uint32_t reset;
+	pasid_notify pasid_fn;
+	void *data;
+};
+
+struct ras_err_pages {
+	uint32_t count;
+	uint64_t *pfn;
+};
+
+struct ras_ecc_err {
+	uint64_t status;
+	uint64_t ipid;
+	uint64_t addr;
+	uint64_t pa_pfn;
+	/* save global channel index across all UMC instances */
+	uint32_t channel_idx;
+	struct ras_err_pages err_pages;
+};
+
+struct ras_ecc_log_info {
+	struct mutex lock;
+	struct radix_tree_root de_page_tree;
+	uint64_t de_queried_count;
+	uint64_t consumption_q_count;
+};
+
+struct ras_critical_region {
+	struct list_head node;
+	struct amdgpu_bo *bo;
+	uint64_t start;
+	uint64_t size;
+};
+
+struct amdgpu_ras {
+	/* ras infrastructure */
+	/* for ras itself. */
+	uint32_t features;
+	uint32_t schema;
+	struct list_head head;
+	/* sysfs */
+	struct device_attribute features_attr;
+	struct device_attribute version_attr;
+	struct device_attribute schema_attr;
+	struct device_attribute event_state_attr;
+	struct bin_attribute badpages_attr;
+	struct dentry *de_ras_eeprom_table;
+	/* block array */
+	struct ras_manager *objs;
+
+	/* gpu recovery */
+	struct work_struct recovery_work;
+	atomic_t in_recovery;
+	atomic_t rma_in_recovery;
+	struct amdgpu_device *adev;
+	/* error handler data */
+	struct ras_err_handler_data *eh_data;
+	struct mutex recovery_lock;
+
+	uint32_t flags;
+	bool reboot;
+	struct amdgpu_ras_eeprom_control eeprom_control;
+
+	bool error_query_ready;
+
+	/* bad page count threshold */
+	uint32_t bad_page_cnt_threshold;
+
+	/* disable ras error count harvest in recovery */
+	bool disable_ras_err_cnt_harvest;
+
+	/* is poison mode supported */
+	bool poison_supported;
+
+	/* RAS count errors delayed work */
+	struct delayed_work ras_counte_delay_work;
+	atomic_t ras_ue_count;
+	atomic_t ras_ce_count;
+
+	/* record umc error info queried from smu */
+	struct umc_ecc_info umc_ecc;
+
+	/* Indicates smu whether need update bad channel info */
+	bool update_channel_flag;
+	/* Record status of smu mca debug mode */
+	bool is_aca_debug_mode;
+	bool is_rma;
+
+	/* Record special requirements of gpu reset caller */
+	uint32_t  gpu_reset_flags;
+
+	struct task_struct *page_retirement_thread;
+	wait_queue_head_t page_retirement_wq;
+	struct mutex page_retirement_lock;
+	atomic_t page_retirement_req_cnt;
+	atomic_t poison_creation_count;
+	atomic_t poison_consumption_count;
+	struct mutex page_rsv_lock;
+	DECLARE_KFIFO(poison_fifo, struct ras_poison_msg, 128);
+	struct ras_ecc_log_info  umc_ecc_log;
+	struct delayed_work page_retirement_dwork;
+
+	/* ras errors detected */
+	unsigned long ras_err_state;
+
+	/* RAS event manager */
+	struct ras_event_manager __event_mgr;
+	struct ras_event_manager *event_mgr;
+
+	uint64_t reserved_pages_in_bytes;
+
+	pid_t init_task_pid;
+	char init_task_comm[TASK_COMM_LEN];
+
+	int bad_page_num;
+
+	struct list_head critical_region_head;
+	struct mutex critical_region_lock;
+
+	/* Protect poison injection */
+	struct mutex poison_lock;
+};
+
+struct ras_fs_data {
+	char sysfs_name[48];
+	char debugfs_name[32];
+};
+
+struct ras_err_info {
+	struct amdgpu_smuio_mcm_config_info mcm_info;
+	u64 ce_count;
+	u64 ue_count;
+	u64 de_count;
+};
+
+struct ras_err_node {
+	struct list_head node;
+	struct ras_err_info err_info;
+};
+
+struct ras_err_data {
+	unsigned long ue_count;
+	unsigned long ce_count;
+	unsigned long de_count;
+	unsigned long err_addr_cnt;
+	struct eeprom_table_record *err_addr;
+	unsigned long err_addr_len;
+	u32 err_list_count;
+	struct list_head err_node_list;
+};
+
+#define for_each_ras_error(err_node, err_data) \
+	list_for_each_entry(err_node, &(err_data)->err_node_list, node)
+
+struct ras_err_handler_data {
+	/* point to bad page records array */
+	struct eeprom_table_record *bps;
+	/* the count of entries */
+	int count;
+	int count_saved;
+	/* the space can place new entries */
+	int space_left;
+};
+
+typedef int (*ras_ih_cb)(struct amdgpu_device *adev,
+		void *err_data,
+		struct amdgpu_iv_entry *entry);
+
+struct ras_ih_data {
+	/* interrupt bottom half */
+	struct work_struct ih_work;
+	int inuse;
+	/* IP callback */
+	ras_ih_cb cb;
+	/* full of entries */
+	unsigned char *ring;
+	unsigned int ring_size;
+	unsigned int element_size;
+	unsigned int aligned_element_size;
+	unsigned int rptr;
+	unsigned int wptr;
+};
+
+struct ras_manager {
+	struct ras_common_if head;
+	/* reference count */
+	int use;
+	/* ras block link */
+	struct list_head node;
+	/* the device */
+	struct amdgpu_device *adev;
+	/* sysfs */
+	struct device_attribute sysfs_attr;
+	int attr_inuse;
+
+	/* fs node name */
+	struct ras_fs_data fs_data;
+
+	/* IH data */
+	struct ras_ih_data ih_data;
+
+	struct ras_err_data err_data;
+
+	struct aca_handle aca_handle;
+};
+
+struct ras_badpage {
+	unsigned int bp;
+	unsigned int size;
+	unsigned int flags;
+};
+
+/* interfaces for IP */
+struct ras_fs_if {
+	struct ras_common_if head;
+	const char* sysfs_name;
+	char debugfs_name[32];
+};
+
+struct ras_query_if {
+	struct ras_common_if head;
+	unsigned long ue_count;
+	unsigned long ce_count;
+	unsigned long de_count;
+};
+
+struct ras_inject_if {
+	struct ras_common_if head;
+	uint64_t address;
+	uint64_t value;
+	uint32_t instance_mask;
+};
+
+struct ras_cure_if {
+	struct ras_common_if head;
+	uint64_t address;
+};
+
+struct ras_ih_if {
+	struct ras_common_if head;
+	ras_ih_cb cb;
+};
+
+struct ras_dispatch_if {
+	struct ras_common_if head;
+	struct amdgpu_iv_entry *entry;
+};
+
+struct ras_debug_if {
+	union {
+		struct ras_common_if head;
+		struct ras_inject_if inject;
+	};
+	int op;
+};
+
+struct amdgpu_ras_block_object {
+	struct ras_common_if  ras_comm;
+
+	int (*ras_block_match)(struct amdgpu_ras_block_object *block_obj,
+				enum amdgpu_ras_block block, uint32_t sub_block_index);
+	int (*ras_late_init)(struct amdgpu_device *adev, struct ras_common_if *ras_block);
+	void (*ras_fini)(struct amdgpu_device *adev, struct ras_common_if *ras_block);
+	ras_ih_cb ras_cb;
+	const struct amdgpu_ras_block_hw_ops *hw_ops;
+};
+
+struct amdgpu_ras_block_hw_ops {
+	int  (*ras_error_inject)(struct amdgpu_device *adev,
+			void *inject_if, uint32_t instance_mask);
+	void (*query_ras_error_count)(struct amdgpu_device *adev, void *ras_error_status);
+	void (*query_ras_error_status)(struct amdgpu_device *adev);
+	void (*query_ras_error_address)(struct amdgpu_device *adev, void *ras_error_status);
+	void (*reset_ras_error_count)(struct amdgpu_device *adev);
+	void (*reset_ras_error_status)(struct amdgpu_device *adev);
+	bool (*query_poison_status)(struct amdgpu_device *adev);
+	bool (*handle_poison_consumption)(struct amdgpu_device *adev);
+};
+
+/* work flow
+ * vbios
+ * 1: ras feature enable (enabled by default)
+ * psp
+ * 2: ras framework init (in ip_init)
+ * IP
+ * 3: IH add
+ * 4: debugfs/sysfs create
+ * 5: query/inject
+ * 6: debugfs/sysfs remove
+ * 7: IH remove
+ * 8: feature disable
+ */
+
+int amdgpu_ras_init_badpage_info(struct amdgpu_device *adev);
+int amdgpu_ras_recovery_init(struct amdgpu_device *adev, bool init_bp_info);
+
+void amdgpu_ras_resume(struct amdgpu_device *adev);
+void amdgpu_ras_suspend(struct amdgpu_device *adev);
+
+int amdgpu_ras_query_error_count(struct amdgpu_device *adev,
+				 unsigned long *ce_count,
+				 unsigned long *ue_count,
+				 struct ras_query_if *query_info);
+
+/* error handling functions */
+int amdgpu_ras_add_bad_pages(struct amdgpu_device *adev,
+		struct eeprom_table_record *bps, int pages, bool from_rom);
+
+int amdgpu_ras_save_bad_pages(struct amdgpu_device *adev,
+		unsigned long *new_cnt);
+
+static inline enum ta_ras_block
+amdgpu_ras_block_to_ta(enum amdgpu_ras_block block) {
+	switch (block) {
+	case AMDGPU_RAS_BLOCK__UMC:
+		return TA_RAS_BLOCK__UMC;
+	case AMDGPU_RAS_BLOCK__SDMA:
+		return TA_RAS_BLOCK__SDMA;
+	case AMDGPU_RAS_BLOCK__GFX:
+		return TA_RAS_BLOCK__GFX;
+	case AMDGPU_RAS_BLOCK__MMHUB:
+		return TA_RAS_BLOCK__MMHUB;
+	case AMDGPU_RAS_BLOCK__ATHUB:
+		return TA_RAS_BLOCK__ATHUB;
+	case AMDGPU_RAS_BLOCK__PCIE_BIF:
+		return TA_RAS_BLOCK__PCIE_BIF;
+	case AMDGPU_RAS_BLOCK__HDP:
+		return TA_RAS_BLOCK__HDP;
+	case AMDGPU_RAS_BLOCK__XGMI_WAFL:
+		return TA_RAS_BLOCK__XGMI_WAFL;
+	case AMDGPU_RAS_BLOCK__DF:
+		return TA_RAS_BLOCK__DF;
+	case AMDGPU_RAS_BLOCK__SMN:
+		return TA_RAS_BLOCK__SMN;
+	case AMDGPU_RAS_BLOCK__SEM:
+		return TA_RAS_BLOCK__SEM;
+	case AMDGPU_RAS_BLOCK__MP0:
+		return TA_RAS_BLOCK__MP0;
+	case AMDGPU_RAS_BLOCK__MP1:
+		return TA_RAS_BLOCK__MP1;
+	case AMDGPU_RAS_BLOCK__FUSE:
+		return TA_RAS_BLOCK__FUSE;
+	case AMDGPU_RAS_BLOCK__MCA:
+		return TA_RAS_BLOCK__MCA;
+	case AMDGPU_RAS_BLOCK__VCN:
+		return TA_RAS_BLOCK__VCN;
+	case AMDGPU_RAS_BLOCK__JPEG:
+		return TA_RAS_BLOCK__JPEG;
+	case AMDGPU_RAS_BLOCK__IH:
+		return TA_RAS_BLOCK__IH;
+	case AMDGPU_RAS_BLOCK__MPIO:
+		return TA_RAS_BLOCK__MPIO;
+	case AMDGPU_RAS_BLOCK__MMSCH:
+		return TA_RAS_BLOCK__MMSCH;
+	default:
+		WARN_ONCE(1, "RAS ERROR: unexpected block id %d\n", block);
+		return TA_RAS_BLOCK__UMC;
+	}
+}
+
+static inline enum ta_ras_error_type
+amdgpu_ras_error_to_ta(enum amdgpu_ras_error_type error) {
+	switch (error) {
+	case AMDGPU_RAS_ERROR__NONE:
+		return TA_RAS_ERROR__NONE;
+	case AMDGPU_RAS_ERROR__PARITY:
+		return TA_RAS_ERROR__PARITY;
+	case AMDGPU_RAS_ERROR__SINGLE_CORRECTABLE:
+		return TA_RAS_ERROR__SINGLE_CORRECTABLE;
+	case AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE:
+		return TA_RAS_ERROR__MULTI_UNCORRECTABLE;
+	case AMDGPU_RAS_ERROR__POISON:
+		return TA_RAS_ERROR__POISON;
+	default:
+		WARN_ONCE(1, "RAS ERROR: unexpected error type %d\n", error);
+		return TA_RAS_ERROR__NONE;
+	}
+}
+
+/* called in ip_init and ip_fini */
+int amdgpu_ras_init(struct amdgpu_device *adev);
+int amdgpu_ras_late_init(struct amdgpu_device *adev);
+int amdgpu_ras_fini(struct amdgpu_device *adev);
+int amdgpu_ras_pre_fini(struct amdgpu_device *adev);
+
+int amdgpu_ras_block_late_init(struct amdgpu_device *adev,
+			struct ras_common_if *ras_block);
+
+void amdgpu_ras_block_late_fini(struct amdgpu_device *adev,
+			  struct ras_common_if *ras_block);
+
+int amdgpu_ras_feature_enable(struct amdgpu_device *adev,
+		struct ras_common_if *head, bool enable);
+
+int amdgpu_ras_feature_enable_on_boot(struct amdgpu_device *adev,
+		struct ras_common_if *head, bool enable);
+
+int amdgpu_ras_sysfs_create(struct amdgpu_device *adev,
+		struct ras_common_if *head);
+
+int amdgpu_ras_sysfs_remove(struct amdgpu_device *adev,
+		struct ras_common_if *head);
+
+void amdgpu_ras_debugfs_create_all(struct amdgpu_device *adev);
+
+int amdgpu_ras_query_error_status(struct amdgpu_device *adev,
+		struct ras_query_if *info);
+
+int amdgpu_ras_reset_error_count(struct amdgpu_device *adev,
+		enum amdgpu_ras_block block);
+int amdgpu_ras_reset_error_status(struct amdgpu_device *adev,
+		enum amdgpu_ras_block block);
+
+int amdgpu_ras_error_inject(struct amdgpu_device *adev,
+		struct ras_inject_if *info);
+
+int amdgpu_ras_interrupt_add_handler(struct amdgpu_device *adev,
+		struct ras_common_if *head);
+
+int amdgpu_ras_interrupt_remove_handler(struct amdgpu_device *adev,
+		struct ras_common_if *head);
+
+int amdgpu_ras_interrupt_dispatch(struct amdgpu_device *adev,
+		struct ras_dispatch_if *info);
+
+struct ras_manager *amdgpu_ras_find_obj(struct amdgpu_device *adev,
+		struct ras_common_if *head);
+
+extern atomic_t amdgpu_ras_in_intr;
+
+static inline bool amdgpu_ras_intr_triggered(void)
+{
+	return !!atomic_read(&amdgpu_ras_in_intr);
+}
+
+static inline void amdgpu_ras_intr_cleared(void)
+{
+	atomic_set(&amdgpu_ras_in_intr, 0);
+}
+
+void amdgpu_ras_global_ras_isr(struct amdgpu_device *adev);
+
+void amdgpu_ras_set_error_query_ready(struct amdgpu_device *adev, bool ready);
+
+bool amdgpu_ras_need_emergency_restart(struct amdgpu_device *adev);
+
+void amdgpu_release_ras_context(struct amdgpu_device *adev);
+
+int amdgpu_persistent_edc_harvesting_supported(struct amdgpu_device *adev);
+
+const char *get_ras_block_str(struct ras_common_if *ras_block);
+
+bool amdgpu_ras_is_poison_mode_supported(struct amdgpu_device *adev);
+
+int amdgpu_ras_is_supported(struct amdgpu_device *adev, unsigned int block);
+
+int amdgpu_ras_reset_gpu(struct amdgpu_device *adev);
+
+struct amdgpu_ras* amdgpu_ras_get_context(struct amdgpu_device *adev);
+
+int amdgpu_ras_set_context(struct amdgpu_device *adev, struct amdgpu_ras *ras_con);
+
+int amdgpu_ras_set_mca_debug_mode(struct amdgpu_device *adev, bool enable);
+int amdgpu_ras_set_aca_debug_mode(struct amdgpu_device *adev, bool enable);
+bool amdgpu_ras_get_aca_debug_mode(struct amdgpu_device *adev);
+bool amdgpu_ras_get_error_query_mode(struct amdgpu_device *adev,
+				     unsigned int *mode);
+
+int amdgpu_ras_register_ras_block(struct amdgpu_device *adev,
+				struct amdgpu_ras_block_object *ras_block_obj);
+void amdgpu_ras_interrupt_fatal_error_handler(struct amdgpu_device *adev);
+void amdgpu_ras_get_error_type_name(uint32_t err_type, char *err_type_name);
+bool amdgpu_ras_inst_get_memory_id_field(struct amdgpu_device *adev,
+					 const struct amdgpu_ras_err_status_reg_entry *reg_entry,
+					 uint32_t instance,
+					 uint32_t *memory_id);
+bool amdgpu_ras_inst_get_err_cnt_field(struct amdgpu_device *adev,
+				       const struct amdgpu_ras_err_status_reg_entry *reg_entry,
+				       uint32_t instance,
+				       unsigned long *err_cnt);
+void amdgpu_ras_inst_query_ras_error_count(struct amdgpu_device *adev,
+					   const struct amdgpu_ras_err_status_reg_entry *reg_list,
+					   uint32_t reg_list_size,
+					   const struct amdgpu_ras_memory_id_entry *mem_list,
+					   uint32_t mem_list_size,
+					   uint32_t instance,
+					   uint32_t err_type,
+					   unsigned long *err_count);
+void amdgpu_ras_inst_reset_ras_error_count(struct amdgpu_device *adev,
+					   const struct amdgpu_ras_err_status_reg_entry *reg_list,
+					   uint32_t reg_list_size,
+					   uint32_t instance);
+
+int amdgpu_ras_error_data_init(struct ras_err_data *err_data);
+void amdgpu_ras_error_data_fini(struct ras_err_data *err_data);
+int amdgpu_ras_error_statistic_ce_count(struct ras_err_data *err_data,
+					struct amdgpu_smuio_mcm_config_info *mcm_info,
+					u64 count);
+int amdgpu_ras_error_statistic_ue_count(struct ras_err_data *err_data,
+					struct amdgpu_smuio_mcm_config_info *mcm_info,
+					u64 count);
+int amdgpu_ras_error_statistic_de_count(struct ras_err_data *err_data,
+					struct amdgpu_smuio_mcm_config_info *mcm_info,
+					u64 count);
+void amdgpu_ras_query_boot_status(struct amdgpu_device *adev, u32 num_instances);
+int amdgpu_ras_bind_aca(struct amdgpu_device *adev, enum amdgpu_ras_block blk,
+			       const struct aca_info *aca_info, void *data);
+int amdgpu_ras_unbind_aca(struct amdgpu_device *adev, enum amdgpu_ras_block blk);
+
+ssize_t amdgpu_ras_aca_sysfs_read(struct device *dev, struct device_attribute *attr,
+				  struct aca_handle *handle, char *buf, void *data);
+
+void amdgpu_ras_set_fed(struct amdgpu_device *adev, bool status);
+bool amdgpu_ras_get_fed_status(struct amdgpu_device *adev);
+void amdgpu_ras_set_err_poison(struct amdgpu_device *adev,
+			       enum amdgpu_ras_block block);
+void amdgpu_ras_clear_err_state(struct amdgpu_device *adev);
+bool amdgpu_ras_is_err_state(struct amdgpu_device *adev, int block);
+
+u64 amdgpu_ras_acquire_event_id(struct amdgpu_device *adev, enum ras_event_type type);
+int amdgpu_ras_mark_ras_event_caller(struct amdgpu_device *adev, enum ras_event_type type,
+				     const void *caller);
+
+int amdgpu_ras_reserve_page(struct amdgpu_device *adev, uint64_t pfn);
+
+int amdgpu_ras_add_critical_region(struct amdgpu_device *adev, struct amdgpu_bo *bo);
+bool amdgpu_ras_check_critical_address(struct amdgpu_device *adev, uint64_t addr);
+
+int amdgpu_ras_put_poison_req(struct amdgpu_device *adev,
+		enum amdgpu_ras_block block, uint16_t pasid,
+		pasid_notify pasid_fn, void *data, uint32_t reset);
+
+bool amdgpu_ras_in_recovery(struct amdgpu_device *adev);
+
+__printf(3, 4)
+void amdgpu_ras_event_log_print(struct amdgpu_device *adev, u64 event_id,
+				const char *fmt, ...);
+
+bool amdgpu_ras_is_rma(struct amdgpu_device *adev);
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c
new file mode 100644
index 000000000000..3eb3fb55ccb0
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c
@@ -0,0 +1,1564 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu_ras_eeprom.h"
+#include "amdgpu.h"
+#include "amdgpu_ras.h"
+#include <linux/bits.h>
+#include "atom.h"
+#include "amdgpu_eeprom.h"
+#include "amdgpu_atomfirmware.h"
+#include <linux/debugfs.h>
+#include <linux/uaccess.h>
+
+#include "amdgpu_reset.h"
+
+/* These are memory addresses as would be seen by one or more EEPROM
+ * chips strung on the I2C bus, usually by manipulating pins 1-3 of a
+ * set of EEPROM devices. They form a continuous memory space.
+ *
+ * The I2C device address includes the device type identifier, 1010b,
+ * which is a reserved value and indicates that this is an I2C EEPROM
+ * device. It also includes the top 3 bits of the 19 bit EEPROM memory
+ * address, namely bits 18, 17, and 16. This makes up the 7 bit
+ * address sent on the I2C bus with bit 0 being the direction bit,
+ * which is not represented here, and sent by the hardware directly.
+ *
+ * For instance,
+ *   50h = 1010000b => device type identifier 1010b, bits 18:16 = 000b, address 0.
+ *   54h = 1010100b => --"--, bits 18:16 = 100b, address 40000h.
+ *   56h = 1010110b => --"--, bits 18:16 = 110b, address 60000h.
+ * Depending on the size of the I2C EEPROM device(s), bits 18:16 may
+ * address memory in a device or a device on the I2C bus, depending on
+ * the status of pins 1-3. See top of amdgpu_eeprom.c.
+ *
+ * The RAS table lives either at address 0 or address 40000h of EEPROM.
+ */
+#define EEPROM_I2C_MADDR_0      0x0
+#define EEPROM_I2C_MADDR_4      0x40000
+
+/*
+ * The 2 macros below represent the actual size in bytes that
+ * those entities occupy in the EEPROM memory.
+ * RAS_TABLE_RECORD_SIZE is different than sizeof(eeprom_table_record) which
+ * uses uint64 to store 6b fields such as retired_page.
+ */
+#define RAS_TABLE_HEADER_SIZE   20
+#define RAS_TABLE_RECORD_SIZE   24
+
+/* Table hdr is 'AMDR' */
+#define RAS_TABLE_HDR_VAL       0x414d4452
+
+/* Bad GPU tag ‘BADG’ */
+#define RAS_TABLE_HDR_BAD       0x42414447
+
+/*
+ * EEPROM Table structure v1
+ * ---------------------------------
+ * |                               |
+ * |     EEPROM TABLE HEADER       |
+ * |      ( size 20 Bytes )        |
+ * |                               |
+ * ---------------------------------
+ * |                               |
+ * |    BAD PAGE RECORD AREA       |
+ * |                               |
+ * ---------------------------------
+ */
+
+/* Assume 2-Mbit size EEPROM and take up the whole space. */
+#define RAS_TBL_SIZE_BYTES      (256 * 1024)
+#define RAS_TABLE_START         0
+#define RAS_HDR_START           RAS_TABLE_START
+#define RAS_RECORD_START        (RAS_HDR_START + RAS_TABLE_HEADER_SIZE)
+#define RAS_MAX_RECORD_COUNT    ((RAS_TBL_SIZE_BYTES - RAS_TABLE_HEADER_SIZE) \
+				 / RAS_TABLE_RECORD_SIZE)
+
+/*
+ * EEPROM Table structrue v2.1
+ * ---------------------------------
+ * |                               |
+ * |     EEPROM TABLE HEADER       |
+ * |      ( size 20 Bytes )        |
+ * |                               |
+ * ---------------------------------
+ * |                               |
+ * |     EEPROM TABLE RAS INFO     |
+ * | (available info size 4 Bytes) |
+ * |  ( reserved size 252 Bytes )  |
+ * |                               |
+ * ---------------------------------
+ * |                               |
+ * |     BAD PAGE RECORD AREA      |
+ * |                               |
+ * ---------------------------------
+ */
+
+/* EEPROM Table V2_1 */
+#define RAS_TABLE_V2_1_INFO_SIZE       256
+#define RAS_TABLE_V2_1_INFO_START      RAS_TABLE_HEADER_SIZE
+#define RAS_RECORD_START_V2_1          (RAS_HDR_START + RAS_TABLE_HEADER_SIZE + \
+					RAS_TABLE_V2_1_INFO_SIZE)
+#define RAS_MAX_RECORD_COUNT_V2_1      ((RAS_TBL_SIZE_BYTES - RAS_TABLE_HEADER_SIZE - \
+					RAS_TABLE_V2_1_INFO_SIZE) \
+					/ RAS_TABLE_RECORD_SIZE)
+
+/* Given a zero-based index of an EEPROM RAS record, yields the EEPROM
+ * offset off of RAS_TABLE_START.  That is, this is something you can
+ * add to control->i2c_address, and then tell I2C layer to read
+ * from/write to there. _N is the so called absolute index,
+ * because it starts right after the table header.
+ */
+#define RAS_INDEX_TO_OFFSET(_C, _N) ((_C)->ras_record_offset + \
+				     (_N) * RAS_TABLE_RECORD_SIZE)
+
+#define RAS_OFFSET_TO_INDEX(_C, _O) (((_O) - \
+				      (_C)->ras_record_offset) / RAS_TABLE_RECORD_SIZE)
+
+/* Given a 0-based relative record index, 0, 1, 2, ..., etc., off
+ * of "fri", return the absolute record index off of the end of
+ * the table header.
+ */
+#define RAS_RI_TO_AI(_C, _I) (((_I) + (_C)->ras_fri) % \
+			      (_C)->ras_max_record_count)
+
+#define RAS_NUM_RECS(_tbl_hdr)  (((_tbl_hdr)->tbl_size - \
+				  RAS_TABLE_HEADER_SIZE) / RAS_TABLE_RECORD_SIZE)
+
+#define RAS_NUM_RECS_V2_1(_tbl_hdr)  (((_tbl_hdr)->tbl_size - \
+				       RAS_TABLE_HEADER_SIZE - \
+				       RAS_TABLE_V2_1_INFO_SIZE) / RAS_TABLE_RECORD_SIZE)
+
+#define to_amdgpu_device(x) ((container_of(x, struct amdgpu_ras, eeprom_control))->adev)
+
+static bool __is_ras_eeprom_supported(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) {
+	case IP_VERSION(11, 0, 2): /* VEGA20 and ARCTURUS */
+	case IP_VERSION(11, 0, 7): /* Sienna cichlid */
+	case IP_VERSION(13, 0, 0):
+	case IP_VERSION(13, 0, 2): /* Aldebaran */
+	case IP_VERSION(13, 0, 10):
+		return true;
+	case IP_VERSION(13, 0, 6):
+	case IP_VERSION(13, 0, 12):
+	case IP_VERSION(13, 0, 14):
+		return (adev->gmc.is_app_apu) ? false : true;
+	default:
+		return false;
+	}
+}
+
+static bool __get_eeprom_i2c_addr(struct amdgpu_device *adev,
+				  struct amdgpu_ras_eeprom_control *control)
+{
+	struct atom_context *atom_ctx = adev->mode_info.atom_context;
+	u8 i2c_addr;
+
+	if (!control)
+		return false;
+
+	if (adev->bios && amdgpu_atomfirmware_ras_rom_addr(adev, &i2c_addr)) {
+		/* The address given by VBIOS is an 8-bit, wire-format
+		 * address, i.e. the most significant byte.
+		 *
+		 * Normalize it to a 19-bit EEPROM address. Remove the
+		 * device type identifier and make it a 7-bit address;
+		 * then make it a 19-bit EEPROM address. See top of
+		 * amdgpu_eeprom.c.
+		 */
+		i2c_addr = (i2c_addr & 0x0F) >> 1;
+		control->i2c_address = ((u32) i2c_addr) << 16;
+
+		return true;
+	}
+
+	switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) {
+	case IP_VERSION(11, 0, 2):
+		/* VEGA20 and ARCTURUS */
+		if (adev->asic_type == CHIP_VEGA20)
+			control->i2c_address = EEPROM_I2C_MADDR_0;
+		else if (strnstr(atom_ctx->vbios_pn,
+				 "D342",
+				 sizeof(atom_ctx->vbios_pn)))
+			control->i2c_address = EEPROM_I2C_MADDR_0;
+		else
+			control->i2c_address = EEPROM_I2C_MADDR_4;
+		return true;
+	case IP_VERSION(11, 0, 7):
+		control->i2c_address = EEPROM_I2C_MADDR_0;
+		return true;
+	case IP_VERSION(13, 0, 2):
+		if (strnstr(atom_ctx->vbios_pn, "D673",
+			    sizeof(atom_ctx->vbios_pn)))
+			control->i2c_address = EEPROM_I2C_MADDR_4;
+		else
+			control->i2c_address = EEPROM_I2C_MADDR_0;
+		return true;
+	case IP_VERSION(13, 0, 0):
+		if (strnstr(atom_ctx->vbios_pn, "D707",
+			    sizeof(atom_ctx->vbios_pn)))
+			control->i2c_address = EEPROM_I2C_MADDR_0;
+		else
+			control->i2c_address = EEPROM_I2C_MADDR_4;
+		return true;
+	case IP_VERSION(13, 0, 6):
+	case IP_VERSION(13, 0, 10):
+	case IP_VERSION(13, 0, 12):
+	case IP_VERSION(13, 0, 14):
+		control->i2c_address = EEPROM_I2C_MADDR_4;
+		return true;
+	default:
+		return false;
+	}
+}
+
+static void
+__encode_table_header_to_buf(struct amdgpu_ras_eeprom_table_header *hdr,
+			     unsigned char *buf)
+{
+	u32 *pp = (uint32_t *)buf;
+
+	pp[0] = cpu_to_le32(hdr->header);
+	pp[1] = cpu_to_le32(hdr->version);
+	pp[2] = cpu_to_le32(hdr->first_rec_offset);
+	pp[3] = cpu_to_le32(hdr->tbl_size);
+	pp[4] = cpu_to_le32(hdr->checksum);
+}
+
+static void
+__decode_table_header_from_buf(struct amdgpu_ras_eeprom_table_header *hdr,
+			       unsigned char *buf)
+{
+	u32 *pp = (uint32_t *)buf;
+
+	hdr->header	      = le32_to_cpu(pp[0]);
+	hdr->version	      = le32_to_cpu(pp[1]);
+	hdr->first_rec_offset = le32_to_cpu(pp[2]);
+	hdr->tbl_size	      = le32_to_cpu(pp[3]);
+	hdr->checksum	      = le32_to_cpu(pp[4]);
+}
+
+static int __write_table_header(struct amdgpu_ras_eeprom_control *control)
+{
+	u8 buf[RAS_TABLE_HEADER_SIZE];
+	struct amdgpu_device *adev = to_amdgpu_device(control);
+	int res;
+
+	memset(buf, 0, sizeof(buf));
+	__encode_table_header_to_buf(&control->tbl_hdr, buf);
+
+	/* i2c may be unstable in gpu reset */
+	down_read(&adev->reset_domain->sem);
+	res = amdgpu_eeprom_write(adev->pm.ras_eeprom_i2c_bus,
+				  control->i2c_address +
+				  control->ras_header_offset,
+				  buf, RAS_TABLE_HEADER_SIZE);
+	up_read(&adev->reset_domain->sem);
+
+	if (res < 0) {
+		dev_err(adev->dev, "Failed to write EEPROM table header:%d",
+			res);
+	} else if (res < RAS_TABLE_HEADER_SIZE) {
+		dev_err(adev->dev, "Short write:%d out of %d\n", res,
+			RAS_TABLE_HEADER_SIZE);
+		res = -EIO;
+	} else {
+		res = 0;
+	}
+
+	return res;
+}
+
+static void
+__encode_table_ras_info_to_buf(struct amdgpu_ras_eeprom_table_ras_info *rai,
+			       unsigned char *buf)
+{
+	u32 *pp = (uint32_t *)buf;
+	u32 tmp;
+
+	tmp = ((uint32_t)(rai->rma_status) & 0xFF) |
+	      (((uint32_t)(rai->health_percent) << 8) & 0xFF00) |
+	      (((uint32_t)(rai->ecc_page_threshold) << 16) & 0xFFFF0000);
+	pp[0] = cpu_to_le32(tmp);
+}
+
+static void
+__decode_table_ras_info_from_buf(struct amdgpu_ras_eeprom_table_ras_info *rai,
+				 unsigned char *buf)
+{
+	u32 *pp = (uint32_t *)buf;
+	u32 tmp;
+
+	tmp = le32_to_cpu(pp[0]);
+	rai->rma_status = tmp & 0xFF;
+	rai->health_percent = (tmp >> 8) & 0xFF;
+	rai->ecc_page_threshold = (tmp >> 16) & 0xFFFF;
+}
+
+static int __write_table_ras_info(struct amdgpu_ras_eeprom_control *control)
+{
+	struct amdgpu_device *adev = to_amdgpu_device(control);
+	u8 *buf;
+	int res;
+
+	buf = kzalloc(RAS_TABLE_V2_1_INFO_SIZE, GFP_KERNEL);
+	if (!buf) {
+		dev_err(adev->dev,
+			"Failed to alloc buf to write table ras info\n");
+		return -ENOMEM;
+	}
+
+	__encode_table_ras_info_to_buf(&control->tbl_rai, buf);
+
+	/* i2c may be unstable in gpu reset */
+	down_read(&adev->reset_domain->sem);
+	res = amdgpu_eeprom_write(adev->pm.ras_eeprom_i2c_bus,
+				  control->i2c_address +
+				  control->ras_info_offset,
+				  buf, RAS_TABLE_V2_1_INFO_SIZE);
+	up_read(&adev->reset_domain->sem);
+
+	if (res < 0) {
+		dev_err(adev->dev, "Failed to write EEPROM table ras info:%d",
+			res);
+	} else if (res < RAS_TABLE_V2_1_INFO_SIZE) {
+		dev_err(adev->dev, "Short write:%d out of %d\n", res,
+			RAS_TABLE_V2_1_INFO_SIZE);
+		res = -EIO;
+	} else {
+		res = 0;
+	}
+
+	kfree(buf);
+
+	return res;
+}
+
+static u8 __calc_hdr_byte_sum(const struct amdgpu_ras_eeprom_control *control)
+{
+	int ii;
+	u8  *pp, csum;
+	size_t sz;
+
+	/* Header checksum, skip checksum field in the calculation */
+	sz = sizeof(control->tbl_hdr) - sizeof(control->tbl_hdr.checksum);
+	pp = (u8 *) &control->tbl_hdr;
+	csum = 0;
+	for (ii = 0; ii < sz; ii++, pp++)
+		csum += *pp;
+
+	return csum;
+}
+
+static u8 __calc_ras_info_byte_sum(const struct amdgpu_ras_eeprom_control *control)
+{
+	int ii;
+	u8  *pp, csum;
+	size_t sz;
+
+	sz = sizeof(control->tbl_rai);
+	pp = (u8 *) &control->tbl_rai;
+	csum = 0;
+	for (ii = 0; ii < sz; ii++, pp++)
+		csum += *pp;
+
+	return csum;
+}
+
+static int amdgpu_ras_eeprom_correct_header_tag(
+	struct amdgpu_ras_eeprom_control *control,
+	uint32_t header)
+{
+	struct amdgpu_ras_eeprom_table_header *hdr = &control->tbl_hdr;
+	u8 *hh;
+	int res;
+	u8 csum;
+
+	csum = -hdr->checksum;
+
+	hh = (void *) &hdr->header;
+	csum -= (hh[0] + hh[1] + hh[2] + hh[3]);
+	hh = (void *) &header;
+	csum += hh[0] + hh[1] + hh[2] + hh[3];
+	csum = -csum;
+	mutex_lock(&control->ras_tbl_mutex);
+	hdr->header = header;
+	hdr->checksum = csum;
+	res = __write_table_header(control);
+	mutex_unlock(&control->ras_tbl_mutex);
+
+	return res;
+}
+
+static void amdgpu_ras_set_eeprom_table_version(struct amdgpu_ras_eeprom_control *control)
+{
+	struct amdgpu_device *adev = to_amdgpu_device(control);
+	struct amdgpu_ras_eeprom_table_header *hdr = &control->tbl_hdr;
+
+	switch (amdgpu_ip_version(adev, UMC_HWIP, 0)) {
+	case IP_VERSION(8, 10, 0):
+		hdr->version = RAS_TABLE_VER_V2_1;
+		return;
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 5, 0):
+		hdr->version = RAS_TABLE_VER_V3;
+		return;
+	default:
+		hdr->version = RAS_TABLE_VER_V1;
+		return;
+	}
+}
+
+/**
+ * amdgpu_ras_eeprom_reset_table -- Reset the RAS EEPROM table
+ * @control: pointer to control structure
+ *
+ * Reset the contents of the header of the RAS EEPROM table.
+ * Return 0 on success, -errno on error.
+ */
+int amdgpu_ras_eeprom_reset_table(struct amdgpu_ras_eeprom_control *control)
+{
+	struct amdgpu_device *adev = to_amdgpu_device(control);
+	struct amdgpu_ras_eeprom_table_header *hdr = &control->tbl_hdr;
+	struct amdgpu_ras_eeprom_table_ras_info *rai = &control->tbl_rai;
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	u8 csum;
+	int res;
+
+	mutex_lock(&control->ras_tbl_mutex);
+
+	hdr->header = RAS_TABLE_HDR_VAL;
+	amdgpu_ras_set_eeprom_table_version(control);
+
+	if (hdr->version >= RAS_TABLE_VER_V2_1) {
+		hdr->first_rec_offset = RAS_RECORD_START_V2_1;
+		hdr->tbl_size = RAS_TABLE_HEADER_SIZE +
+				RAS_TABLE_V2_1_INFO_SIZE;
+		rai->rma_status = GPU_HEALTH_USABLE;
+		/**
+		 * GPU health represented as a percentage.
+		 * 0 means worst health, 100 means fully health.
+		 */
+		rai->health_percent = 100;
+		/* ecc_page_threshold = 0 means disable bad page retirement */
+		rai->ecc_page_threshold = con->bad_page_cnt_threshold;
+	} else {
+		hdr->first_rec_offset = RAS_RECORD_START;
+		hdr->tbl_size = RAS_TABLE_HEADER_SIZE;
+	}
+
+	csum = __calc_hdr_byte_sum(control);
+	if (hdr->version >= RAS_TABLE_VER_V2_1)
+		csum += __calc_ras_info_byte_sum(control);
+	csum = -csum;
+	hdr->checksum = csum;
+	res = __write_table_header(control);
+	if (!res && hdr->version > RAS_TABLE_VER_V1)
+		res = __write_table_ras_info(control);
+
+	control->ras_num_recs = 0;
+	control->ras_num_bad_pages = 0;
+	control->ras_num_mca_recs = 0;
+	control->ras_num_pa_recs = 0;
+	control->ras_fri = 0;
+
+	amdgpu_dpm_send_hbm_bad_pages_num(adev, control->ras_num_bad_pages);
+
+	control->bad_channel_bitmap = 0;
+	amdgpu_dpm_send_hbm_bad_channel_flag(adev, control->bad_channel_bitmap);
+	con->update_channel_flag = false;
+
+	amdgpu_ras_debugfs_set_ret_size(control);
+
+	mutex_unlock(&control->ras_tbl_mutex);
+
+	return res;
+}
+
+static void
+__encode_table_record_to_buf(struct amdgpu_ras_eeprom_control *control,
+			     struct eeprom_table_record *record,
+			     unsigned char *buf)
+{
+	__le64 tmp = 0;
+	int i = 0;
+
+	/* Next are all record fields according to EEPROM page spec in LE foramt */
+	buf[i++] = record->err_type;
+
+	buf[i++] = record->bank;
+
+	tmp = cpu_to_le64(record->ts);
+	memcpy(buf + i, &tmp, 8);
+	i += 8;
+
+	tmp = cpu_to_le64((record->offset & 0xffffffffffff));
+	memcpy(buf + i, &tmp, 6);
+	i += 6;
+
+	buf[i++] = record->mem_channel;
+	buf[i++] = record->mcumc_id;
+
+	tmp = cpu_to_le64((record->retired_page & 0xffffffffffff));
+	memcpy(buf + i, &tmp, 6);
+}
+
+static void
+__decode_table_record_from_buf(struct amdgpu_ras_eeprom_control *control,
+			       struct eeprom_table_record *record,
+			       unsigned char *buf)
+{
+	__le64 tmp = 0;
+	int i =  0;
+
+	/* Next are all record fields according to EEPROM page spec in LE foramt */
+	record->err_type = buf[i++];
+
+	record->bank = buf[i++];
+
+	memcpy(&tmp, buf + i, 8);
+	record->ts = le64_to_cpu(tmp);
+	i += 8;
+
+	memcpy(&tmp, buf + i, 6);
+	record->offset = (le64_to_cpu(tmp) & 0xffffffffffff);
+	i += 6;
+
+	record->mem_channel = buf[i++];
+	record->mcumc_id = buf[i++];
+
+	memcpy(&tmp, buf + i,  6);
+	record->retired_page = (le64_to_cpu(tmp) & 0xffffffffffff);
+}
+
+bool amdgpu_ras_eeprom_check_err_threshold(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+
+	if (!__is_ras_eeprom_supported(adev) ||
+	    !amdgpu_bad_page_threshold)
+		return false;
+
+	/* skip check eeprom table for VEGA20 Gaming */
+	if (!con)
+		return false;
+	else
+		if (!(con->features & BIT(AMDGPU_RAS_BLOCK__UMC)))
+			return false;
+
+	if (con->eeprom_control.tbl_hdr.header == RAS_TABLE_HDR_BAD) {
+		if (con->eeprom_control.ras_num_bad_pages > con->bad_page_cnt_threshold)
+			dev_warn(adev->dev, "RAS records:%d exceed threshold:%d",
+				 con->eeprom_control.ras_num_bad_pages, con->bad_page_cnt_threshold);
+		if ((amdgpu_bad_page_threshold == -1) ||
+		    (amdgpu_bad_page_threshold == -2)) {
+			dev_warn(adev->dev,
+				 "Please consult AMD Service Action Guide (SAG) for appropriate service procedures.\n");
+			return false;
+		} else {
+			dev_warn(adev->dev,
+				 "Please consider adjusting the customized threshold.\n");
+			return true;
+		}
+	}
+
+	return false;
+}
+
+/**
+ * __amdgpu_ras_eeprom_write -- write indexed from buffer to EEPROM
+ * @control: pointer to control structure
+ * @buf: pointer to buffer containing data to write
+ * @fri: start writing at this index
+ * @num: number of records to write
+ *
+ * The caller must hold the table mutex in @control.
+ * Return 0 on success, -errno otherwise.
+ */
+static int __amdgpu_ras_eeprom_write(struct amdgpu_ras_eeprom_control *control,
+				     u8 *buf, const u32 fri, const u32 num)
+{
+	struct amdgpu_device *adev = to_amdgpu_device(control);
+	u32 buf_size;
+	int res;
+
+	/* i2c may be unstable in gpu reset */
+	down_read(&adev->reset_domain->sem);
+	buf_size = num * RAS_TABLE_RECORD_SIZE;
+	res = amdgpu_eeprom_write(adev->pm.ras_eeprom_i2c_bus,
+				  control->i2c_address +
+				  RAS_INDEX_TO_OFFSET(control, fri),
+				  buf, buf_size);
+	up_read(&adev->reset_domain->sem);
+	if (res < 0) {
+		dev_err(adev->dev, "Writing %d EEPROM table records error:%d",
+			num, res);
+	} else if (res < buf_size) {
+		/* Short write, return error.
+		 */
+		dev_err(adev->dev, "Wrote %d records out of %d",
+			res / RAS_TABLE_RECORD_SIZE, num);
+		res = -EIO;
+	} else {
+		res = 0;
+	}
+
+	return res;
+}
+
+static int
+amdgpu_ras_eeprom_append_table(struct amdgpu_ras_eeprom_control *control,
+			       struct eeprom_table_record *record,
+			       const u32 num)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(to_amdgpu_device(control));
+	struct amdgpu_device *adev = to_amdgpu_device(control);
+	u32 a, b, i;
+	u8 *buf, *pp;
+	int res;
+
+	buf = kcalloc(num, RAS_TABLE_RECORD_SIZE, GFP_KERNEL);
+	if (!buf)
+		return -ENOMEM;
+
+	/* Encode all of them in one go.
+	 */
+	pp = buf;
+	for (i = 0; i < num; i++, pp += RAS_TABLE_RECORD_SIZE) {
+		__encode_table_record_to_buf(control, &record[i], pp);
+
+		/* update bad channel bitmap */
+		if ((record[i].mem_channel < BITS_PER_TYPE(control->bad_channel_bitmap)) &&
+		    !(control->bad_channel_bitmap & (1 << record[i].mem_channel))) {
+			control->bad_channel_bitmap |= 1 << record[i].mem_channel;
+			con->update_channel_flag = true;
+		}
+	}
+
+	/* a, first record index to write into.
+	 * b, last record index to write into.
+	 * a = first index to read (fri) + number of records in the table,
+	 * b = a + @num - 1.
+	 * Let N = control->ras_max_num_record_count, then we have,
+	 * case 0: 0 <= a <= b < N,
+	 *   just append @num records starting at a;
+	 * case 1: 0 <= a < N <= b,
+	 *   append (N - a) records starting at a, and
+	 *   append the remainder,  b % N + 1, starting at 0.
+	 * case 2: 0 <= fri < N <= a <= b, then modulo N we get two subcases,
+	 * case 2a: 0 <= a <= b < N
+	 *   append num records starting at a; and fix fri if b overwrote it,
+	 *   and since a <= b, if b overwrote it then a must've also,
+	 *   and if b didn't overwrite it, then a didn't also.
+	 * case 2b: 0 <= b < a < N
+	 *   write num records starting at a, which wraps around 0=N
+	 *   and overwrite fri unconditionally. Now from case 2a,
+	 *   this means that b eclipsed fri to overwrite it and wrap
+	 *   around 0 again, i.e. b = 2N+r pre modulo N, so we unconditionally
+	 *   set fri = b + 1 (mod N).
+	 * Now, since fri is updated in every case, except the trivial case 0,
+	 * the number of records present in the table after writing, is,
+	 * num_recs - 1 = b - fri (mod N), and we take the positive value,
+	 * by adding an arbitrary multiple of N before taking the modulo N
+	 * as shown below.
+	 */
+	a = control->ras_fri + control->ras_num_recs;
+	b = a + num  - 1;
+	if (b < control->ras_max_record_count) {
+		res = __amdgpu_ras_eeprom_write(control, buf, a, num);
+	} else if (a < control->ras_max_record_count) {
+		u32 g0, g1;
+
+		g0 = control->ras_max_record_count - a;
+		g1 = b % control->ras_max_record_count + 1;
+		res = __amdgpu_ras_eeprom_write(control, buf, a, g0);
+		if (res)
+			goto Out;
+		res = __amdgpu_ras_eeprom_write(control,
+						buf + g0 * RAS_TABLE_RECORD_SIZE,
+						0, g1);
+		if (res)
+			goto Out;
+		if (g1 > control->ras_fri)
+			control->ras_fri = g1 % control->ras_max_record_count;
+	} else {
+		a %= control->ras_max_record_count;
+		b %= control->ras_max_record_count;
+
+		if (a <= b) {
+			/* Note that, b - a + 1 = num. */
+			res = __amdgpu_ras_eeprom_write(control, buf, a, num);
+			if (res)
+				goto Out;
+			if (b >= control->ras_fri)
+				control->ras_fri = (b + 1) % control->ras_max_record_count;
+		} else {
+			u32 g0, g1;
+
+			/* b < a, which means, we write from
+			 * a to the end of the table, and from
+			 * the start of the table to b.
+			 */
+			g0 = control->ras_max_record_count - a;
+			g1 = b + 1;
+			res = __amdgpu_ras_eeprom_write(control, buf, a, g0);
+			if (res)
+				goto Out;
+			res = __amdgpu_ras_eeprom_write(control,
+							buf + g0 * RAS_TABLE_RECORD_SIZE,
+							0, g1);
+			if (res)
+				goto Out;
+			control->ras_fri = g1 % control->ras_max_record_count;
+		}
+	}
+	control->ras_num_recs = 1 + (control->ras_max_record_count + b
+				     - control->ras_fri)
+		% control->ras_max_record_count;
+
+	/*old asics only save pa to eeprom like before*/
+	if (IP_VERSION_MAJ(amdgpu_ip_version(adev, UMC_HWIP, 0)) < 12)
+		control->ras_num_pa_recs += num;
+	else
+		control->ras_num_mca_recs += num;
+
+	control->ras_num_bad_pages = con->bad_page_num;
+Out:
+	kfree(buf);
+	return res;
+}
+
+static int
+amdgpu_ras_eeprom_update_header(struct amdgpu_ras_eeprom_control *control)
+{
+	struct amdgpu_device *adev = to_amdgpu_device(control);
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+	u8 *buf, *pp, csum;
+	u32 buf_size;
+	int res;
+
+	/* Modify the header if it exceeds.
+	 */
+	if (amdgpu_bad_page_threshold != 0 &&
+	    control->ras_num_bad_pages > ras->bad_page_cnt_threshold) {
+		dev_warn(adev->dev,
+			"Saved bad pages %d reaches threshold value %d\n",
+			control->ras_num_bad_pages, ras->bad_page_cnt_threshold);
+
+		if (adev->cper.enabled && amdgpu_cper_generate_bp_threshold_record(adev))
+			dev_warn(adev->dev, "fail to generate bad page threshold cper records\n");
+
+		if ((amdgpu_bad_page_threshold != -1) &&
+		    (amdgpu_bad_page_threshold != -2)) {
+			control->tbl_hdr.header = RAS_TABLE_HDR_BAD;
+			if (control->tbl_hdr.version >= RAS_TABLE_VER_V2_1) {
+				control->tbl_rai.rma_status = GPU_RETIRED__ECC_REACH_THRESHOLD;
+				control->tbl_rai.health_percent = 0;
+			}
+			ras->is_rma = true;
+		}
+
+		/* ignore the -ENOTSUPP return value */
+		amdgpu_dpm_send_rma_reason(adev);
+	}
+
+	if (control->tbl_hdr.version >= RAS_TABLE_VER_V2_1)
+		control->tbl_hdr.tbl_size = RAS_TABLE_HEADER_SIZE +
+					    RAS_TABLE_V2_1_INFO_SIZE +
+					    control->ras_num_recs * RAS_TABLE_RECORD_SIZE;
+	else
+		control->tbl_hdr.tbl_size = RAS_TABLE_HEADER_SIZE +
+					    control->ras_num_recs * RAS_TABLE_RECORD_SIZE;
+	control->tbl_hdr.checksum = 0;
+
+	buf_size = control->ras_num_recs * RAS_TABLE_RECORD_SIZE;
+	buf = kcalloc(control->ras_num_recs, RAS_TABLE_RECORD_SIZE, GFP_KERNEL);
+	if (!buf) {
+		dev_err(adev->dev,
+			"allocating memory for table of size %d bytes failed\n",
+			control->tbl_hdr.tbl_size);
+		res = -ENOMEM;
+		goto Out;
+	}
+
+	down_read(&adev->reset_domain->sem);
+	res = amdgpu_eeprom_read(adev->pm.ras_eeprom_i2c_bus,
+				 control->i2c_address +
+				 control->ras_record_offset,
+				 buf, buf_size);
+	up_read(&adev->reset_domain->sem);
+	if (res < 0) {
+		dev_err(adev->dev, "EEPROM failed reading records:%d\n", res);
+		goto Out;
+	} else if (res < buf_size) {
+		dev_err(adev->dev, "EEPROM read %d out of %d bytes\n", res,
+			buf_size);
+		res = -EIO;
+		goto Out;
+	}
+
+	/**
+	 * bad page records have been stored in eeprom,
+	 * now calculate gpu health percent
+	 */
+	if (amdgpu_bad_page_threshold != 0 &&
+	    control->tbl_hdr.version >= RAS_TABLE_VER_V2_1 &&
+	    control->ras_num_bad_pages <= ras->bad_page_cnt_threshold)
+		control->tbl_rai.health_percent = ((ras->bad_page_cnt_threshold -
+						   control->ras_num_bad_pages) * 100) /
+						   ras->bad_page_cnt_threshold;
+
+	/* Recalc the checksum.
+	 */
+	csum = 0;
+	for (pp = buf; pp < buf + buf_size; pp++)
+		csum += *pp;
+
+	csum += __calc_hdr_byte_sum(control);
+	if (control->tbl_hdr.version >= RAS_TABLE_VER_V2_1)
+		csum += __calc_ras_info_byte_sum(control);
+	/* avoid sign extension when assigning to "checksum" */
+	csum = -csum;
+	control->tbl_hdr.checksum = csum;
+	res = __write_table_header(control);
+	if (!res && control->tbl_hdr.version > RAS_TABLE_VER_V1)
+		res = __write_table_ras_info(control);
+Out:
+	kfree(buf);
+	return res;
+}
+
+/**
+ * amdgpu_ras_eeprom_append -- append records to the EEPROM RAS table
+ * @control: pointer to control structure
+ * @record: array of records to append
+ * @num: number of records in @record array
+ *
+ * Append @num records to the table, calculate the checksum and write
+ * the table back to EEPROM. The maximum number of records that
+ * can be appended is between 1 and control->ras_max_record_count,
+ * regardless of how many records are already stored in the table.
+ *
+ * Return 0 on success or if EEPROM is not supported, -errno on error.
+ */
+int amdgpu_ras_eeprom_append(struct amdgpu_ras_eeprom_control *control,
+			     struct eeprom_table_record *record,
+			     const u32 num)
+{
+	struct amdgpu_device *adev = to_amdgpu_device(control);
+	int res, i;
+	uint64_t nps = AMDGPU_NPS1_PARTITION_MODE;
+
+	if (!__is_ras_eeprom_supported(adev))
+		return 0;
+
+	if (num == 0) {
+		dev_err(adev->dev, "will not append 0 records\n");
+		return -EINVAL;
+	} else if (num > control->ras_max_record_count) {
+		dev_err(adev->dev,
+			"cannot append %d records than the size of table %d\n",
+			num, control->ras_max_record_count);
+		return -EINVAL;
+	}
+
+	if (adev->gmc.gmc_funcs->query_mem_partition_mode)
+		nps = adev->gmc.gmc_funcs->query_mem_partition_mode(adev);
+
+	/* set the new channel index flag */
+	for (i = 0; i < num; i++)
+		record[i].retired_page |= (nps << UMC_NPS_SHIFT);
+
+	mutex_lock(&control->ras_tbl_mutex);
+
+	res = amdgpu_ras_eeprom_append_table(control, record, num);
+	if (!res)
+		res = amdgpu_ras_eeprom_update_header(control);
+	if (!res)
+		amdgpu_ras_debugfs_set_ret_size(control);
+
+	mutex_unlock(&control->ras_tbl_mutex);
+
+	/* clear channel index flag, the flag is only saved on eeprom */
+	for (i = 0; i < num; i++)
+		record[i].retired_page &= ~(nps << UMC_NPS_SHIFT);
+
+	return res;
+}
+
+/**
+ * __amdgpu_ras_eeprom_read -- read indexed from EEPROM into buffer
+ * @control: pointer to control structure
+ * @buf: pointer to buffer to read into
+ * @fri: first record index, start reading at this index, absolute index
+ * @num: number of records to read
+ *
+ * The caller must hold the table mutex in @control.
+ * Return 0 on success, -errno otherwise.
+ */
+static int __amdgpu_ras_eeprom_read(struct amdgpu_ras_eeprom_control *control,
+				    u8 *buf, const u32 fri, const u32 num)
+{
+	struct amdgpu_device *adev = to_amdgpu_device(control);
+	u32 buf_size;
+	int res;
+
+	/* i2c may be unstable in gpu reset */
+	down_read(&adev->reset_domain->sem);
+	buf_size = num * RAS_TABLE_RECORD_SIZE;
+	res = amdgpu_eeprom_read(adev->pm.ras_eeprom_i2c_bus,
+				 control->i2c_address +
+				 RAS_INDEX_TO_OFFSET(control, fri),
+				 buf, buf_size);
+	up_read(&adev->reset_domain->sem);
+	if (res < 0) {
+		dev_err(adev->dev, "Reading %d EEPROM table records error:%d",
+			num, res);
+	} else if (res < buf_size) {
+		/* Short read, return error.
+		 */
+		dev_err(adev->dev, "Read %d records out of %d",
+			res / RAS_TABLE_RECORD_SIZE, num);
+		res = -EIO;
+	} else {
+		res = 0;
+	}
+
+	return res;
+}
+
+/**
+ * amdgpu_ras_eeprom_read -- read EEPROM
+ * @control: pointer to control structure
+ * @record: array of records to read into
+ * @num: number of records in @record
+ *
+ * Reads num records from the RAS table in EEPROM and
+ * writes the data into @record array.
+ *
+ * Returns 0 on success, -errno on error.
+ */
+int amdgpu_ras_eeprom_read(struct amdgpu_ras_eeprom_control *control,
+			   struct eeprom_table_record *record,
+			   const u32 num)
+{
+	struct amdgpu_device *adev = to_amdgpu_device(control);
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	int i, res;
+	u8 *buf, *pp;
+	u32 g0, g1;
+
+	if (!__is_ras_eeprom_supported(adev))
+		return 0;
+
+	if (num == 0) {
+		dev_err(adev->dev, "will not read 0 records\n");
+		return -EINVAL;
+	} else if (num > control->ras_num_recs) {
+		dev_err(adev->dev, "too many records to read:%d available:%d\n",
+			num, control->ras_num_recs);
+		return -EINVAL;
+	}
+
+	buf = kcalloc(num, RAS_TABLE_RECORD_SIZE, GFP_KERNEL);
+	if (!buf)
+		return -ENOMEM;
+
+	/* Determine how many records to read, from the first record
+	 * index, fri, to the end of the table, and from the beginning
+	 * of the table, such that the total number of records is
+	 * @num, and we handle wrap around when fri > 0 and
+	 * fri + num > RAS_MAX_RECORD_COUNT.
+	 *
+	 * First we compute the index of the last element
+	 * which would be fetched from each region,
+	 * g0 is in [fri, fri + num - 1], and
+	 * g1 is in [0, RAS_MAX_RECORD_COUNT - 1].
+	 * Then, if g0 < RAS_MAX_RECORD_COUNT, the index of
+	 * the last element to fetch, we set g0 to _the number_
+	 * of elements to fetch, @num, since we know that the last
+	 * indexed to be fetched does not exceed the table.
+	 *
+	 * If, however, g0 >= RAS_MAX_RECORD_COUNT, then
+	 * we set g0 to the number of elements to read
+	 * until the end of the table, and g1 to the number of
+	 * elements to read from the beginning of the table.
+	 */
+	g0 = control->ras_fri + num - 1;
+	g1 = g0 % control->ras_max_record_count;
+	if (g0 < control->ras_max_record_count) {
+		g0 = num;
+		g1 = 0;
+	} else {
+		g0 = control->ras_max_record_count - control->ras_fri;
+		g1 += 1;
+	}
+
+	mutex_lock(&control->ras_tbl_mutex);
+	res = __amdgpu_ras_eeprom_read(control, buf, control->ras_fri, g0);
+	if (res)
+		goto Out;
+	if (g1) {
+		res = __amdgpu_ras_eeprom_read(control,
+					       buf + g0 * RAS_TABLE_RECORD_SIZE,
+					       0, g1);
+		if (res)
+			goto Out;
+	}
+
+	res = 0;
+
+	/* Read up everything? Then transform.
+	 */
+	pp = buf;
+	for (i = 0; i < num; i++, pp += RAS_TABLE_RECORD_SIZE) {
+		__decode_table_record_from_buf(control, &record[i], pp);
+
+		/* update bad channel bitmap */
+		if ((record[i].mem_channel < BITS_PER_TYPE(control->bad_channel_bitmap)) &&
+		    !(control->bad_channel_bitmap & (1 << record[i].mem_channel))) {
+			control->bad_channel_bitmap |= 1 << record[i].mem_channel;
+			con->update_channel_flag = true;
+		}
+	}
+Out:
+	kfree(buf);
+	mutex_unlock(&control->ras_tbl_mutex);
+
+	return res;
+}
+
+uint32_t amdgpu_ras_eeprom_max_record_count(struct amdgpu_ras_eeprom_control *control)
+{
+	/* get available eeprom table version first before eeprom table init */
+	amdgpu_ras_set_eeprom_table_version(control);
+
+	if (control->tbl_hdr.version >= RAS_TABLE_VER_V2_1)
+		return RAS_MAX_RECORD_COUNT_V2_1;
+	else
+		return RAS_MAX_RECORD_COUNT;
+}
+
+static ssize_t
+amdgpu_ras_debugfs_eeprom_size_read(struct file *f, char __user *buf,
+				    size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)file_inode(f)->i_private;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+	struct amdgpu_ras_eeprom_control *control = ras ? &ras->eeprom_control : NULL;
+	u8 data[50];
+	int res;
+
+	if (!size)
+		return size;
+
+	if (!ras || !control) {
+		res = snprintf(data, sizeof(data), "Not supported\n");
+	} else {
+		res = snprintf(data, sizeof(data), "%d bytes or %d records\n",
+			       RAS_TBL_SIZE_BYTES, control->ras_max_record_count);
+	}
+
+	if (*pos >= res)
+		return 0;
+
+	res -= *pos;
+	res = min_t(size_t, res, size);
+
+	if (copy_to_user(buf, &data[*pos], res))
+		return -EFAULT;
+
+	*pos += res;
+
+	return res;
+}
+
+const struct file_operations amdgpu_ras_debugfs_eeprom_size_ops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_ras_debugfs_eeprom_size_read,
+	.write = NULL,
+	.llseek = default_llseek,
+};
+
+static const char *tbl_hdr_str = " Signature    Version  FirstOffs       Size   Checksum\n";
+static const char *tbl_hdr_fmt = "0x%08X 0x%08X 0x%08X 0x%08X 0x%08X\n";
+#define tbl_hdr_fmt_size (5 * (2+8) + 4 + 1)
+static const char *rec_hdr_str = "Index  Offset ErrType Bank/CU          TimeStamp      Offs/Addr MemChl MCUMCID    RetiredPage\n";
+static const char *rec_hdr_fmt = "%5d 0x%05X %7s    0x%02X 0x%016llX 0x%012llX   0x%02X    0x%02X 0x%012llX\n";
+#define rec_hdr_fmt_size (5 + 1 + 7 + 1 + 7 + 1 + 7 + 1 + 18 + 1 + 14 + 1 + 6 + 1 + 7 + 1 + 14 + 1)
+
+static const char *record_err_type_str[AMDGPU_RAS_EEPROM_ERR_COUNT] = {
+	"ignore",
+	"re",
+	"ue",
+};
+
+static loff_t amdgpu_ras_debugfs_table_size(struct amdgpu_ras_eeprom_control *control)
+{
+	return strlen(tbl_hdr_str) + tbl_hdr_fmt_size +
+		strlen(rec_hdr_str) + rec_hdr_fmt_size * control->ras_num_recs;
+}
+
+void amdgpu_ras_debugfs_set_ret_size(struct amdgpu_ras_eeprom_control *control)
+{
+	struct amdgpu_ras *ras = container_of(control, struct amdgpu_ras,
+					      eeprom_control);
+	struct dentry *de = ras->de_ras_eeprom_table;
+
+	if (de)
+		d_inode(de)->i_size = amdgpu_ras_debugfs_table_size(control);
+}
+
+static ssize_t amdgpu_ras_debugfs_table_read(struct file *f, char __user *buf,
+					     size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)file_inode(f)->i_private;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+	struct amdgpu_ras_eeprom_control *control = &ras->eeprom_control;
+	const size_t orig_size = size;
+	int res = -EFAULT;
+	size_t data_len;
+
+	mutex_lock(&control->ras_tbl_mutex);
+
+	/* We want *pos - data_len > 0, which means there's
+	 * bytes to be printed from data.
+	 */
+	data_len = strlen(tbl_hdr_str);
+	if (*pos < data_len) {
+		data_len -= *pos;
+		data_len = min_t(size_t, data_len, size);
+		if (copy_to_user(buf, &tbl_hdr_str[*pos], data_len))
+			goto Out;
+		buf += data_len;
+		size -= data_len;
+		*pos += data_len;
+	}
+
+	data_len = strlen(tbl_hdr_str) + tbl_hdr_fmt_size;
+	if (*pos < data_len && size > 0) {
+		u8 data[tbl_hdr_fmt_size + 1];
+		loff_t lpos;
+
+		snprintf(data, sizeof(data), tbl_hdr_fmt,
+			 control->tbl_hdr.header,
+			 control->tbl_hdr.version,
+			 control->tbl_hdr.first_rec_offset,
+			 control->tbl_hdr.tbl_size,
+			 control->tbl_hdr.checksum);
+
+		data_len -= *pos;
+		data_len = min_t(size_t, data_len, size);
+		lpos = *pos - strlen(tbl_hdr_str);
+		if (copy_to_user(buf, &data[lpos], data_len))
+			goto Out;
+		buf += data_len;
+		size -= data_len;
+		*pos += data_len;
+	}
+
+	data_len = strlen(tbl_hdr_str) + tbl_hdr_fmt_size + strlen(rec_hdr_str);
+	if (*pos < data_len && size > 0) {
+		loff_t lpos;
+
+		data_len -= *pos;
+		data_len = min_t(size_t, data_len, size);
+		lpos = *pos - strlen(tbl_hdr_str) - tbl_hdr_fmt_size;
+		if (copy_to_user(buf, &rec_hdr_str[lpos], data_len))
+			goto Out;
+		buf += data_len;
+		size -= data_len;
+		*pos += data_len;
+	}
+
+	data_len = amdgpu_ras_debugfs_table_size(control);
+	if (*pos < data_len && size > 0) {
+		u8 dare[RAS_TABLE_RECORD_SIZE];
+		u8 data[rec_hdr_fmt_size + 1];
+		struct eeprom_table_record record;
+		int s, r;
+
+		/* Find the starting record index
+		 */
+		s = *pos - strlen(tbl_hdr_str) - tbl_hdr_fmt_size -
+			strlen(rec_hdr_str);
+		s = s / rec_hdr_fmt_size;
+		r = *pos - strlen(tbl_hdr_str) - tbl_hdr_fmt_size -
+			strlen(rec_hdr_str);
+		r = r % rec_hdr_fmt_size;
+
+		for ( ; size > 0 && s < control->ras_num_recs; s++) {
+			u32 ai = RAS_RI_TO_AI(control, s);
+			/* Read a single record
+			 */
+			res = __amdgpu_ras_eeprom_read(control, dare, ai, 1);
+			if (res)
+				goto Out;
+			__decode_table_record_from_buf(control, &record, dare);
+			snprintf(data, sizeof(data), rec_hdr_fmt,
+				 s,
+				 RAS_INDEX_TO_OFFSET(control, ai),
+				 record_err_type_str[record.err_type],
+				 record.bank,
+				 record.ts,
+				 record.offset,
+				 record.mem_channel,
+				 record.mcumc_id,
+				 record.retired_page);
+
+			data_len = min_t(size_t, rec_hdr_fmt_size - r, size);
+			if (copy_to_user(buf, &data[r], data_len)) {
+				res = -EFAULT;
+				goto Out;
+			}
+			buf += data_len;
+			size -= data_len;
+			*pos += data_len;
+			r = 0;
+		}
+	}
+	res = 0;
+Out:
+	mutex_unlock(&control->ras_tbl_mutex);
+	return res < 0 ? res : orig_size - size;
+}
+
+static ssize_t
+amdgpu_ras_debugfs_eeprom_table_read(struct file *f, char __user *buf,
+				     size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)file_inode(f)->i_private;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+	struct amdgpu_ras_eeprom_control *control = ras ? &ras->eeprom_control : NULL;
+	u8 data[81];
+	int res;
+
+	if (!size)
+		return size;
+
+	if (!ras || !control) {
+		res = snprintf(data, sizeof(data), "Not supported\n");
+		if (*pos >= res)
+			return 0;
+
+		res -= *pos;
+		res = min_t(size_t, res, size);
+
+		if (copy_to_user(buf, &data[*pos], res))
+			return -EFAULT;
+
+		*pos += res;
+
+		return res;
+	} else {
+		return amdgpu_ras_debugfs_table_read(f, buf, size, pos);
+	}
+}
+
+const struct file_operations amdgpu_ras_debugfs_eeprom_table_ops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_ras_debugfs_eeprom_table_read,
+	.write = NULL,
+	.llseek = default_llseek,
+};
+
+/**
+ * __verify_ras_table_checksum -- verify the RAS EEPROM table checksum
+ * @control: pointer to control structure
+ *
+ * Check the checksum of the stored in EEPROM RAS table.
+ *
+ * Return 0 if the checksum is correct,
+ * positive if it is not correct, and
+ * -errno on I/O error.
+ */
+static int __verify_ras_table_checksum(struct amdgpu_ras_eeprom_control *control)
+{
+	struct amdgpu_device *adev = to_amdgpu_device(control);
+	int buf_size, res;
+	u8  csum, *buf, *pp;
+
+	if (control->tbl_hdr.version >= RAS_TABLE_VER_V2_1)
+		buf_size = RAS_TABLE_HEADER_SIZE +
+			   RAS_TABLE_V2_1_INFO_SIZE +
+			   control->ras_num_recs * RAS_TABLE_RECORD_SIZE;
+	else
+		buf_size = RAS_TABLE_HEADER_SIZE +
+			   control->ras_num_recs * RAS_TABLE_RECORD_SIZE;
+
+	buf = kzalloc(buf_size, GFP_KERNEL);
+	if (!buf) {
+		dev_err(adev->dev,
+			"Out of memory checking RAS table checksum.\n");
+		return -ENOMEM;
+	}
+
+	res = amdgpu_eeprom_read(adev->pm.ras_eeprom_i2c_bus,
+				 control->i2c_address +
+				 control->ras_header_offset,
+				 buf, buf_size);
+	if (res < buf_size) {
+		dev_err(adev->dev, "Partial read for checksum, res:%d\n", res);
+		/* On partial reads, return -EIO.
+		 */
+		if (res >= 0)
+			res = -EIO;
+		goto Out;
+	}
+
+	csum = 0;
+	for (pp = buf; pp < buf + buf_size; pp++)
+		csum += *pp;
+Out:
+	kfree(buf);
+	return res < 0 ? res : csum;
+}
+
+static int __read_table_ras_info(struct amdgpu_ras_eeprom_control *control)
+{
+	struct amdgpu_ras_eeprom_table_ras_info *rai = &control->tbl_rai;
+	struct amdgpu_device *adev = to_amdgpu_device(control);
+	unsigned char *buf;
+	int res;
+
+	buf = kzalloc(RAS_TABLE_V2_1_INFO_SIZE, GFP_KERNEL);
+	if (!buf) {
+		dev_err(adev->dev,
+			"Failed to alloc buf to read EEPROM table ras info\n");
+		return -ENOMEM;
+	}
+
+	/**
+	 * EEPROM table V2_1 supports ras info,
+	 * read EEPROM table ras info
+	 */
+	res = amdgpu_eeprom_read(adev->pm.ras_eeprom_i2c_bus,
+				 control->i2c_address + control->ras_info_offset,
+				 buf, RAS_TABLE_V2_1_INFO_SIZE);
+	if (res < RAS_TABLE_V2_1_INFO_SIZE) {
+		dev_err(adev->dev,
+			"Failed to read EEPROM table ras info, res:%d", res);
+		res = res >= 0 ? -EIO : res;
+		goto Out;
+	}
+
+	__decode_table_ras_info_from_buf(rai, buf);
+
+Out:
+	kfree(buf);
+	return res == RAS_TABLE_V2_1_INFO_SIZE ? 0 : res;
+}
+
+int amdgpu_ras_eeprom_init(struct amdgpu_ras_eeprom_control *control)
+{
+	struct amdgpu_device *adev = to_amdgpu_device(control);
+	unsigned char buf[RAS_TABLE_HEADER_SIZE] = { 0 };
+	struct amdgpu_ras_eeprom_table_header *hdr = &control->tbl_hdr;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+	int res;
+
+	ras->is_rma = false;
+
+	if (!__is_ras_eeprom_supported(adev))
+		return 0;
+
+	/* Verify i2c adapter is initialized */
+	if (!adev->pm.ras_eeprom_i2c_bus || !adev->pm.ras_eeprom_i2c_bus->algo)
+		return -ENOENT;
+
+	if (!__get_eeprom_i2c_addr(adev, control))
+		return -EINVAL;
+
+	control->ras_header_offset = RAS_HDR_START;
+	control->ras_info_offset = RAS_TABLE_V2_1_INFO_START;
+	mutex_init(&control->ras_tbl_mutex);
+
+	/* Read the table header from EEPROM address */
+	res = amdgpu_eeprom_read(adev->pm.ras_eeprom_i2c_bus,
+				 control->i2c_address + control->ras_header_offset,
+				 buf, RAS_TABLE_HEADER_SIZE);
+	if (res < RAS_TABLE_HEADER_SIZE) {
+		dev_err(adev->dev, "Failed to read EEPROM table header, res:%d",
+			res);
+		return res >= 0 ? -EIO : res;
+	}
+
+	__decode_table_header_from_buf(hdr, buf);
+
+	if (hdr->header != RAS_TABLE_HDR_VAL &&
+	    hdr->header != RAS_TABLE_HDR_BAD) {
+		dev_info(adev->dev, "Creating a new EEPROM table");
+		return amdgpu_ras_eeprom_reset_table(control);
+	}
+
+	switch (hdr->version) {
+	case RAS_TABLE_VER_V2_1:
+	case RAS_TABLE_VER_V3:
+		control->ras_num_recs = RAS_NUM_RECS_V2_1(hdr);
+		control->ras_record_offset = RAS_RECORD_START_V2_1;
+		control->ras_max_record_count = RAS_MAX_RECORD_COUNT_V2_1;
+		break;
+	case RAS_TABLE_VER_V1:
+		control->ras_num_recs = RAS_NUM_RECS(hdr);
+		control->ras_record_offset = RAS_RECORD_START;
+		control->ras_max_record_count = RAS_MAX_RECORD_COUNT;
+		break;
+	default:
+		dev_err(adev->dev,
+			"RAS header invalid, unsupported version: %u",
+			hdr->version);
+		return -EINVAL;
+	}
+
+	if (control->ras_num_recs > control->ras_max_record_count) {
+		dev_err(adev->dev,
+			"RAS header invalid, records in header: %u max allowed :%u",
+			control->ras_num_recs, control->ras_max_record_count);
+		return -EINVAL;
+	}
+
+	control->ras_fri = RAS_OFFSET_TO_INDEX(control, hdr->first_rec_offset);
+	control->ras_num_mca_recs = 0;
+	control->ras_num_pa_recs = 0;
+	return 0;
+}
+
+int amdgpu_ras_eeprom_check(struct amdgpu_ras_eeprom_control *control)
+{
+	struct amdgpu_device *adev = to_amdgpu_device(control);
+	struct amdgpu_ras_eeprom_table_header *hdr = &control->tbl_hdr;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+	int res = 0;
+
+	if (!__is_ras_eeprom_supported(adev))
+		return 0;
+
+	/* Verify i2c adapter is initialized */
+	if (!adev->pm.ras_eeprom_i2c_bus || !adev->pm.ras_eeprom_i2c_bus->algo)
+		return -ENOENT;
+
+	if (!__get_eeprom_i2c_addr(adev, control))
+		return -EINVAL;
+
+	control->ras_num_bad_pages = ras->bad_page_num;
+
+	if (hdr->header == RAS_TABLE_HDR_VAL) {
+		dev_dbg(adev->dev,
+			"Found existing EEPROM table with %d records",
+			control->ras_num_bad_pages);
+
+		if (hdr->version >= RAS_TABLE_VER_V2_1) {
+			res = __read_table_ras_info(control);
+			if (res)
+				return res;
+		}
+
+		res = __verify_ras_table_checksum(control);
+		if (res)
+			dev_err(adev->dev,
+				"RAS table incorrect checksum or error:%d\n",
+				res);
+
+		/* Warn if we are at 90% of the threshold or above
+		 */
+		if (10 * control->ras_num_bad_pages >= 9 * ras->bad_page_cnt_threshold)
+			dev_warn(adev->dev, "RAS records:%u exceeds 90%% of threshold:%d",
+					control->ras_num_bad_pages,
+					ras->bad_page_cnt_threshold);
+	} else if (hdr->header == RAS_TABLE_HDR_BAD &&
+		   amdgpu_bad_page_threshold != 0) {
+		if (hdr->version >= RAS_TABLE_VER_V2_1) {
+			res = __read_table_ras_info(control);
+			if (res)
+				return res;
+		}
+
+		res = __verify_ras_table_checksum(control);
+		if (res) {
+			dev_err(adev->dev,
+				"RAS Table incorrect checksum or error:%d\n",
+				res);
+			return -EINVAL;
+		}
+		if (ras->bad_page_cnt_threshold >= control->ras_num_bad_pages) {
+			/* This means that, the threshold was increased since
+			 * the last time the system was booted, and now,
+			 * ras->bad_page_cnt_threshold - control->num_recs > 0,
+			 * so that at least one more record can be saved,
+			 * before the page count threshold is reached.
+			 */
+			dev_info(adev->dev,
+				 "records:%d threshold:%d, resetting "
+				 "RAS table header signature",
+				 control->ras_num_bad_pages,
+				 ras->bad_page_cnt_threshold);
+			res = amdgpu_ras_eeprom_correct_header_tag(control,
+								   RAS_TABLE_HDR_VAL);
+		} else {
+			dev_warn(adev->dev,
+				"RAS records:%d exceed threshold:%d\n",
+				control->ras_num_bad_pages, ras->bad_page_cnt_threshold);
+			if ((amdgpu_bad_page_threshold == -1) ||
+			    (amdgpu_bad_page_threshold == -2)) {
+				res = 0;
+				dev_warn(adev->dev,
+					 "Please consult AMD Service Action Guide (SAG) for appropriate service procedures\n");
+			} else {
+				ras->is_rma = true;
+				dev_warn(adev->dev,
+					 "User defined threshold is set, runtime service will be halt when threshold is reached\n");
+			}
+		}
+	}
+
+	return res < 0 ? res : 0;
+}
+
+void amdgpu_ras_eeprom_check_and_recover(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+	struct amdgpu_ras_eeprom_control *control;
+	int res;
+
+	if (!__is_ras_eeprom_supported(adev) || !ras)
+		return;
+	control = &ras->eeprom_control;
+	if (!control->is_eeprom_valid)
+		return;
+	res = __verify_ras_table_checksum(control);
+	if (res) {
+		dev_warn(adev->dev,
+			"RAS table incorrect checksum or error:%d, try to recover\n",
+			res);
+		if (!amdgpu_ras_eeprom_reset_table(control))
+			if (!amdgpu_ras_save_bad_pages(adev, NULL))
+				if (!__verify_ras_table_checksum(control)) {
+					dev_info(adev->dev, "RAS table recovery succeed\n");
+					return;
+				}
+		dev_err(adev->dev, "RAS table recovery failed\n");
+		control->is_eeprom_valid = false;
+	}
+	return;
+}
+\ No newline at end of file
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h
new file mode 100644
index 000000000000..ebfca4cb5688
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h
@@ -0,0 +1,169 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef _AMDGPU_RAS_EEPROM_H
+#define _AMDGPU_RAS_EEPROM_H
+
+#include <linux/i2c.h>
+
+#define RAS_TABLE_VER_V1           0x00010000
+#define RAS_TABLE_VER_V2_1         0x00021000
+#define RAS_TABLE_VER_V3           0x00030000
+
+struct amdgpu_device;
+
+enum amdgpu_ras_gpu_health_status {
+	GPU_HEALTH_USABLE = 0,
+	GPU_RETIRED__ECC_REACH_THRESHOLD = 2,
+};
+
+enum amdgpu_ras_eeprom_err_type {
+	AMDGPU_RAS_EEPROM_ERR_NA,
+	AMDGPU_RAS_EEPROM_ERR_RECOVERABLE,
+	AMDGPU_RAS_EEPROM_ERR_NON_RECOVERABLE,
+	AMDGPU_RAS_EEPROM_ERR_COUNT,
+};
+
+struct amdgpu_ras_eeprom_table_header {
+	uint32_t header;
+	uint32_t version;
+	uint32_t first_rec_offset;
+	uint32_t tbl_size;
+	uint32_t checksum;
+} __packed;
+
+struct amdgpu_ras_eeprom_table_ras_info {
+	u8  rma_status;
+	u8  health_percent;
+	u16 ecc_page_threshold;
+	u32 padding[64 - 1];
+} __packed;
+
+struct amdgpu_ras_eeprom_control {
+	struct amdgpu_ras_eeprom_table_header tbl_hdr;
+
+	struct amdgpu_ras_eeprom_table_ras_info tbl_rai;
+
+	/* Base I2C EEPPROM 19-bit memory address,
+	 * where the table is located. For more information,
+	 * see top of amdgpu_eeprom.c.
+	 */
+	u32 i2c_address;
+
+	/* The byte offset off of @i2c_address
+	 * where the table header is found,
+	 * and where the records start--always
+	 * right after the header.
+	 */
+	u32 ras_header_offset;
+	u32 ras_info_offset;
+	u32 ras_record_offset;
+
+	/* Number of records in the table.
+	 */
+	u32 ras_num_recs;
+
+	/* the bad page number is ras_num_recs or
+	 * ras_num_recs * umc.retire_unit
+	 */
+	u32 ras_num_bad_pages;
+
+	/* Number of records store mca address */
+	u32 ras_num_mca_recs;
+
+	/* Number of records store physical address */
+	u32 ras_num_pa_recs;
+
+	/* First record index to read, 0-based.
+	 * Range is [0, num_recs-1]. This is
+	 * an absolute index, starting right after
+	 * the table header.
+	 */
+	u32 ras_fri;
+
+	/* Maximum possible number of records
+	 * we could store, i.e. the maximum capacity
+	 * of the table.
+	 */
+	u32 ras_max_record_count;
+
+	/* Protect table access via this mutex.
+	 */
+	struct mutex ras_tbl_mutex;
+
+	/* Record channel info which occurred bad pages
+	 */
+	u32 bad_channel_bitmap;
+
+	bool is_eeprom_valid;
+};
+
+/*
+ * Represents single table record. Packed to be easily serialized into byte
+ * stream.
+ */
+struct eeprom_table_record {
+
+	union {
+		uint64_t address;
+		uint64_t offset;
+	};
+
+	uint64_t retired_page;
+	uint64_t ts;
+
+	enum amdgpu_ras_eeprom_err_type err_type;
+
+	union {
+		unsigned char bank;
+		unsigned char cu;
+	};
+
+	unsigned char mem_channel;
+	unsigned char mcumc_id;
+} __packed;
+
+int amdgpu_ras_eeprom_init(struct amdgpu_ras_eeprom_control *control);
+
+int amdgpu_ras_eeprom_reset_table(struct amdgpu_ras_eeprom_control *control);
+
+bool amdgpu_ras_eeprom_check_err_threshold(struct amdgpu_device *adev);
+
+int amdgpu_ras_eeprom_read(struct amdgpu_ras_eeprom_control *control,
+			   struct eeprom_table_record *records, const u32 num);
+
+int amdgpu_ras_eeprom_append(struct amdgpu_ras_eeprom_control *control,
+			     struct eeprom_table_record *records, const u32 num);
+
+uint32_t amdgpu_ras_eeprom_max_record_count(struct amdgpu_ras_eeprom_control *control);
+
+void amdgpu_ras_debugfs_set_ret_size(struct amdgpu_ras_eeprom_control *control);
+
+int amdgpu_ras_eeprom_check(struct amdgpu_ras_eeprom_control *control);
+
+void amdgpu_ras_eeprom_check_and_recover(struct amdgpu_device *adev);
+
+extern const struct file_operations amdgpu_ras_debugfs_eeprom_size_ops;
+extern const struct file_operations amdgpu_ras_debugfs_eeprom_table_ops;
+
+#endif // _AMDGPU_RAS_EEPROM_H
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_res_cursor.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_res_cursor.h
new file mode 100644
index 000000000000..be2e56ce1355
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_res_cursor.h
@@ -0,0 +1,194 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Christian König
+ */
+
+#ifndef __AMDGPU_RES_CURSOR_H__
+#define __AMDGPU_RES_CURSOR_H__
+
+#include <drm/drm_mm.h>
+#include <drm/ttm/ttm_resource.h>
+#include <drm/ttm/ttm_range_manager.h>
+
+#include "amdgpu_vram_mgr.h"
+
+/* state back for walking over vram_mgr and gtt_mgr allocations */
+struct amdgpu_res_cursor {
+	uint64_t		start;
+	uint64_t		size;
+	uint64_t		remaining;
+	void			*node;
+	uint32_t		mem_type;
+};
+
+/**
+ * amdgpu_res_first - initialize a amdgpu_res_cursor
+ *
+ * @res: TTM resource object to walk
+ * @start: Start of the range
+ * @size: Size of the range
+ * @cur: cursor object to initialize
+ *
+ * Start walking over the range of allocations between @start and @size.
+ */
+static inline void amdgpu_res_first(struct ttm_resource *res,
+				    uint64_t start, uint64_t size,
+				    struct amdgpu_res_cursor *cur)
+{
+	struct drm_buddy_block *block;
+	struct list_head *head, *next;
+	struct drm_mm_node *node;
+
+	if (!res)
+		goto fallback;
+
+	BUG_ON(start + size > res->size);
+
+	cur->mem_type = res->mem_type;
+
+	switch (cur->mem_type) {
+	case TTM_PL_VRAM:
+		head = &to_amdgpu_vram_mgr_resource(res)->blocks;
+
+		block = list_first_entry_or_null(head,
+						 struct drm_buddy_block,
+						 link);
+		if (!block)
+			goto fallback;
+
+		while (start >= amdgpu_vram_mgr_block_size(block)) {
+			start -= amdgpu_vram_mgr_block_size(block);
+
+			next = block->link.next;
+			if (next != head)
+				block = list_entry(next, struct drm_buddy_block, link);
+		}
+
+		cur->start = amdgpu_vram_mgr_block_start(block) + start;
+		cur->size = min(amdgpu_vram_mgr_block_size(block) - start, size);
+		cur->remaining = size;
+		cur->node = block;
+		break;
+	case TTM_PL_TT:
+	case AMDGPU_PL_DOORBELL:
+	case AMDGPU_PL_MMIO_REMAP:
+		node = to_ttm_range_mgr_node(res)->mm_nodes;
+		while (start >= node->size << PAGE_SHIFT)
+			start -= node++->size << PAGE_SHIFT;
+
+		cur->start = (node->start << PAGE_SHIFT) + start;
+		cur->size = min((node->size << PAGE_SHIFT) - start, size);
+		cur->remaining = size;
+		cur->node = node;
+		break;
+	default:
+		goto fallback;
+	}
+
+	return;
+
+fallback:
+	cur->start = start;
+	cur->size = size;
+	cur->remaining = size;
+	cur->node = NULL;
+	WARN_ON(res && start + size > res->size);
+}
+
+/**
+ * amdgpu_res_next - advance the cursor
+ *
+ * @cur: the cursor to advance
+ * @size: number of bytes to move forward
+ *
+ * Move the cursor @size bytes forwrad, walking to the next node if necessary.
+ */
+static inline void amdgpu_res_next(struct amdgpu_res_cursor *cur, uint64_t size)
+{
+	struct drm_buddy_block *block;
+	struct drm_mm_node *node;
+	struct list_head *next;
+
+	BUG_ON(size > cur->remaining);
+
+	cur->remaining -= size;
+	if (!cur->remaining)
+		return;
+
+	cur->size -= size;
+	if (cur->size) {
+		cur->start += size;
+		return;
+	}
+
+	switch (cur->mem_type) {
+	case TTM_PL_VRAM:
+		block = cur->node;
+
+		next = block->link.next;
+		block = list_entry(next, struct drm_buddy_block, link);
+
+		cur->node = block;
+		cur->start = amdgpu_vram_mgr_block_start(block);
+		cur->size = min(amdgpu_vram_mgr_block_size(block), cur->remaining);
+		break;
+	case TTM_PL_TT:
+	case AMDGPU_PL_DOORBELL:
+	case AMDGPU_PL_MMIO_REMAP:
+		node = cur->node;
+
+		cur->node = ++node;
+		cur->start = node->start << PAGE_SHIFT;
+		cur->size = min(node->size << PAGE_SHIFT, cur->remaining);
+		break;
+	default:
+		return;
+	}
+}
+
+/**
+ * amdgpu_res_cleared - check if blocks are cleared
+ *
+ * @cur: the cursor to extract the block
+ *
+ * Check if the @cur block is cleared
+ */
+static inline bool amdgpu_res_cleared(struct amdgpu_res_cursor *cur)
+{
+	struct drm_buddy_block *block;
+
+	switch (cur->mem_type) {
+	case TTM_PL_VRAM:
+		block = cur->node;
+
+		if (!amdgpu_vram_mgr_is_cleared(block))
+			return false;
+		break;
+	default:
+		return false;
+	}
+
+	return true;
+}
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c
new file mode 100644
index 000000000000..28c4ad62f50e
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c
@@ -0,0 +1,354 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu_reset.h"
+#include "aldebaran.h"
+#include "sienna_cichlid.h"
+#include "smu_v13_0_10.h"
+
+static int amdgpu_reset_xgmi_reset_on_init_suspend(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = adev->num_ip_blocks - 1; i >= 0; i--) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		if (!adev->ip_blocks[i].status.hw)
+			continue;
+		/* displays are handled in phase1 */
+		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_DCE)
+			continue;
+
+		/* XXX handle errors */
+		amdgpu_ip_block_suspend(&adev->ip_blocks[i]);
+		adev->ip_blocks[i].status.hw = false;
+	}
+
+	/* VCN FW shared region is in frambuffer, there are some flags
+	 * initialized in that region during sw_init. Make sure the region is
+	 * backed up.
+	 */
+	amdgpu_vcn_save_vcpu_bo(adev);
+
+	return 0;
+}
+
+static int amdgpu_reset_xgmi_reset_on_init_prep_hwctxt(
+	struct amdgpu_reset_control *reset_ctl,
+	struct amdgpu_reset_context *reset_context)
+{
+	struct list_head *reset_device_list = reset_context->reset_device_list;
+	struct amdgpu_device *tmp_adev;
+	int r;
+
+	list_for_each_entry(tmp_adev, reset_device_list, reset_list) {
+		amdgpu_unregister_gpu_instance(tmp_adev);
+		r = amdgpu_reset_xgmi_reset_on_init_suspend(tmp_adev);
+		if (r) {
+			dev_err(tmp_adev->dev,
+				"xgmi reset on init: prepare for reset failed");
+			return r;
+		}
+	}
+
+	return r;
+}
+
+static int amdgpu_reset_xgmi_reset_on_init_restore_hwctxt(
+	struct amdgpu_reset_control *reset_ctl,
+	struct amdgpu_reset_context *reset_context)
+{
+	struct list_head *reset_device_list = reset_context->reset_device_list;
+	struct amdgpu_device *tmp_adev = NULL;
+	int r;
+
+	r = amdgpu_device_reinit_after_reset(reset_context);
+	if (r)
+		return r;
+	list_for_each_entry(tmp_adev, reset_device_list, reset_list) {
+		if (!tmp_adev->kfd.init_complete) {
+			kgd2kfd_init_zone_device(tmp_adev);
+			amdgpu_amdkfd_device_init(tmp_adev);
+			amdgpu_amdkfd_drm_client_create(tmp_adev);
+		}
+	}
+
+	return r;
+}
+
+static int amdgpu_reset_xgmi_reset_on_init_perform_reset(
+	struct amdgpu_reset_control *reset_ctl,
+	struct amdgpu_reset_context *reset_context)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle;
+	struct list_head *reset_device_list = reset_context->reset_device_list;
+	struct amdgpu_device *tmp_adev = NULL;
+	int r;
+
+	dev_dbg(adev->dev, "xgmi roi - hw reset\n");
+
+	list_for_each_entry(tmp_adev, reset_device_list, reset_list) {
+		mutex_lock(&tmp_adev->reset_cntl->reset_lock);
+		tmp_adev->reset_cntl->active_reset =
+			amdgpu_asic_reset_method(adev);
+	}
+	r = 0;
+	/* Mode1 reset needs to be triggered on all devices together */
+	list_for_each_entry(tmp_adev, reset_device_list, reset_list) {
+		/* For XGMI run all resets in parallel to speed up the process */
+		if (!queue_work(system_unbound_wq, &tmp_adev->xgmi_reset_work))
+			r = -EALREADY;
+		if (r) {
+			dev_err(tmp_adev->dev,
+				"xgmi reset on init: reset failed with error, %d",
+				r);
+			break;
+		}
+	}
+
+	/* For XGMI wait for all resets to complete before proceed */
+	if (!r) {
+		list_for_each_entry(tmp_adev, reset_device_list, reset_list) {
+			flush_work(&tmp_adev->xgmi_reset_work);
+			r = tmp_adev->asic_reset_res;
+			if (r)
+				break;
+		}
+	}
+
+	list_for_each_entry(tmp_adev, reset_device_list, reset_list) {
+		mutex_unlock(&tmp_adev->reset_cntl->reset_lock);
+		tmp_adev->reset_cntl->active_reset = AMD_RESET_METHOD_NONE;
+	}
+
+	return r;
+}
+
+int amdgpu_reset_do_xgmi_reset_on_init(
+	struct amdgpu_reset_context *reset_context)
+{
+	struct list_head *reset_device_list = reset_context->reset_device_list;
+	struct amdgpu_device *adev;
+	int r;
+
+	if (!reset_device_list || list_empty(reset_device_list) ||
+	    list_is_singular(reset_device_list))
+		return -EINVAL;
+
+	adev = list_first_entry(reset_device_list, struct amdgpu_device,
+				reset_list);
+	r = amdgpu_reset_prepare_hwcontext(adev, reset_context);
+	if (r)
+		return r;
+
+	r = amdgpu_reset_perform_reset(adev, reset_context);
+
+	return r;
+}
+
+struct amdgpu_reset_handler xgmi_reset_on_init_handler = {
+	.reset_method = AMD_RESET_METHOD_ON_INIT,
+	.prepare_env = NULL,
+	.prepare_hwcontext = amdgpu_reset_xgmi_reset_on_init_prep_hwctxt,
+	.perform_reset = amdgpu_reset_xgmi_reset_on_init_perform_reset,
+	.restore_hwcontext = amdgpu_reset_xgmi_reset_on_init_restore_hwctxt,
+	.restore_env = NULL,
+	.do_reset = NULL,
+};
+
+int amdgpu_reset_init(struct amdgpu_device *adev)
+{
+	int ret = 0;
+
+	switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) {
+	case IP_VERSION(13, 0, 2):
+	case IP_VERSION(13, 0, 6):
+	case IP_VERSION(13, 0, 12):
+	case IP_VERSION(13, 0, 14):
+		ret = aldebaran_reset_init(adev);
+		break;
+	case IP_VERSION(11, 0, 7):
+		ret = sienna_cichlid_reset_init(adev);
+		break;
+	case IP_VERSION(13, 0, 10):
+		ret = smu_v13_0_10_reset_init(adev);
+		break;
+	default:
+		break;
+	}
+
+	return ret;
+}
+
+int amdgpu_reset_fini(struct amdgpu_device *adev)
+{
+	int ret = 0;
+
+	switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) {
+	case IP_VERSION(13, 0, 2):
+	case IP_VERSION(13, 0, 6):
+	case IP_VERSION(13, 0, 12):
+	case IP_VERSION(13, 0, 14):
+		ret = aldebaran_reset_fini(adev);
+		break;
+	case IP_VERSION(11, 0, 7):
+		ret = sienna_cichlid_reset_fini(adev);
+		break;
+	case IP_VERSION(13, 0, 10):
+		ret = smu_v13_0_10_reset_fini(adev);
+		break;
+	default:
+		break;
+	}
+
+	return ret;
+}
+
+int amdgpu_reset_prepare_hwcontext(struct amdgpu_device *adev,
+				   struct amdgpu_reset_context *reset_context)
+{
+	struct amdgpu_reset_handler *reset_handler = NULL;
+
+	if (adev->reset_cntl && adev->reset_cntl->get_reset_handler)
+		reset_handler = adev->reset_cntl->get_reset_handler(
+			adev->reset_cntl, reset_context);
+	if (!reset_handler)
+		return -EOPNOTSUPP;
+
+	return reset_handler->prepare_hwcontext(adev->reset_cntl,
+						reset_context);
+}
+
+int amdgpu_reset_perform_reset(struct amdgpu_device *adev,
+			       struct amdgpu_reset_context *reset_context)
+{
+	int ret;
+	struct amdgpu_reset_handler *reset_handler = NULL;
+
+	if (adev->reset_cntl)
+		reset_handler = adev->reset_cntl->get_reset_handler(
+			adev->reset_cntl, reset_context);
+	if (!reset_handler)
+		return -EOPNOTSUPP;
+
+	ret = reset_handler->perform_reset(adev->reset_cntl, reset_context);
+	if (ret)
+		return ret;
+
+	return reset_handler->restore_hwcontext(adev->reset_cntl,
+						reset_context);
+}
+
+
+void amdgpu_reset_destroy_reset_domain(struct kref *ref)
+{
+	struct amdgpu_reset_domain *reset_domain = container_of(ref,
+								struct amdgpu_reset_domain,
+								refcount);
+	if (reset_domain->wq)
+		destroy_workqueue(reset_domain->wq);
+
+	kvfree(reset_domain);
+}
+
+struct amdgpu_reset_domain *amdgpu_reset_create_reset_domain(enum amdgpu_reset_domain_type type,
+							     char *wq_name)
+{
+	struct amdgpu_reset_domain *reset_domain;
+
+	reset_domain = kvzalloc(sizeof(struct amdgpu_reset_domain), GFP_KERNEL);
+	if (!reset_domain) {
+		DRM_ERROR("Failed to allocate amdgpu_reset_domain!");
+		return NULL;
+	}
+
+	reset_domain->type = type;
+	kref_init(&reset_domain->refcount);
+
+	reset_domain->wq = create_singlethread_workqueue(wq_name);
+	if (!reset_domain->wq) {
+		DRM_ERROR("Failed to allocate wq for amdgpu_reset_domain!");
+		amdgpu_reset_put_reset_domain(reset_domain);
+		return NULL;
+
+	}
+
+	atomic_set(&reset_domain->in_gpu_reset, 0);
+	atomic_set(&reset_domain->reset_res, 0);
+	init_rwsem(&reset_domain->sem);
+
+	return reset_domain;
+}
+
+void amdgpu_device_lock_reset_domain(struct amdgpu_reset_domain *reset_domain)
+{
+	atomic_set(&reset_domain->in_gpu_reset, 1);
+	down_write(&reset_domain->sem);
+}
+
+
+void amdgpu_device_unlock_reset_domain(struct amdgpu_reset_domain *reset_domain)
+{
+	atomic_set(&reset_domain->in_gpu_reset, 0);
+	up_write(&reset_domain->sem);
+}
+
+void amdgpu_reset_get_desc(struct amdgpu_reset_context *rst_ctxt, char *buf,
+			   size_t len)
+{
+	if (!buf || !len)
+		return;
+
+	switch (rst_ctxt->src) {
+	case AMDGPU_RESET_SRC_JOB:
+		if (rst_ctxt->job) {
+			snprintf(buf, len, "job hang on ring:%s",
+				 rst_ctxt->job->base.sched->name);
+		} else {
+			strscpy(buf, "job hang", len);
+		}
+		break;
+	case AMDGPU_RESET_SRC_RAS:
+		strscpy(buf, "RAS error", len);
+		break;
+	case AMDGPU_RESET_SRC_MES:
+		strscpy(buf, "MES hang", len);
+		break;
+	case AMDGPU_RESET_SRC_HWS:
+		strscpy(buf, "HWS hang", len);
+		break;
+	case AMDGPU_RESET_SRC_USER:
+		strscpy(buf, "user trigger", len);
+		break;
+	case AMDGPU_RESET_SRC_USERQ:
+		strscpy(buf, "user queue trigger", len);
+		break;
+	default:
+		strscpy(buf, "unknown", len);
+	}
+}
+
+bool amdgpu_reset_in_recovery(struct amdgpu_device *adev)
+{
+	return (adev->init_lvl->level == AMDGPU_INIT_LEVEL_RESET_RECOVERY);
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h
new file mode 100644
index 000000000000..07b4d37f1db6
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h
@@ -0,0 +1,176 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_RESET_H__
+#define __AMDGPU_RESET_H__
+
+#include "amdgpu.h"
+
+#define AMDGPU_RESET_MAX_HANDLERS 5
+
+enum AMDGPU_RESET_FLAGS {
+
+	AMDGPU_NEED_FULL_RESET = 0,
+	AMDGPU_SKIP_HW_RESET = 1,
+	AMDGPU_SKIP_COREDUMP = 2,
+	AMDGPU_HOST_FLR = 3,
+};
+
+enum AMDGPU_RESET_SRCS {
+	AMDGPU_RESET_SRC_UNKNOWN,
+	AMDGPU_RESET_SRC_JOB,
+	AMDGPU_RESET_SRC_RAS,
+	AMDGPU_RESET_SRC_MES,
+	AMDGPU_RESET_SRC_HWS,
+	AMDGPU_RESET_SRC_USER,
+	AMDGPU_RESET_SRC_USERQ,
+};
+
+struct amdgpu_reset_context {
+	enum amd_reset_method method;
+	struct amdgpu_device *reset_req_dev;
+	struct amdgpu_job *job;
+	struct amdgpu_hive_info *hive;
+	struct list_head *reset_device_list;
+	unsigned long flags;
+	enum AMDGPU_RESET_SRCS src;
+};
+
+struct amdgpu_reset_handler {
+	enum amd_reset_method reset_method;
+	int (*prepare_env)(struct amdgpu_reset_control *reset_ctl,
+			   struct amdgpu_reset_context *context);
+	int (*prepare_hwcontext)(struct amdgpu_reset_control *reset_ctl,
+				 struct amdgpu_reset_context *context);
+	int (*perform_reset)(struct amdgpu_reset_control *reset_ctl,
+			     struct amdgpu_reset_context *context);
+	int (*restore_hwcontext)(struct amdgpu_reset_control *reset_ctl,
+				 struct amdgpu_reset_context *context);
+	int (*restore_env)(struct amdgpu_reset_control *reset_ctl,
+			   struct amdgpu_reset_context *context);
+
+	int (*do_reset)(struct amdgpu_device *adev);
+};
+
+struct amdgpu_reset_control {
+	void *handle;
+	struct work_struct reset_work;
+	struct mutex reset_lock;
+	struct amdgpu_reset_handler *(
+		*reset_handlers)[AMDGPU_RESET_MAX_HANDLERS];
+	atomic_t in_reset;
+	enum amd_reset_method active_reset;
+	struct amdgpu_reset_handler *(*get_reset_handler)(
+		struct amdgpu_reset_control *reset_ctl,
+		struct amdgpu_reset_context *context);
+	void (*async_reset)(struct work_struct *work);
+};
+
+
+enum amdgpu_reset_domain_type {
+	SINGLE_DEVICE,
+	XGMI_HIVE
+};
+
+struct amdgpu_reset_domain {
+	struct kref refcount;
+	struct workqueue_struct *wq;
+	enum amdgpu_reset_domain_type type;
+	struct rw_semaphore sem;
+	atomic_t in_gpu_reset;
+	atomic_t reset_res;
+};
+
+int amdgpu_reset_init(struct amdgpu_device *adev);
+int amdgpu_reset_fini(struct amdgpu_device *adev);
+
+int amdgpu_reset_prepare_hwcontext(struct amdgpu_device *adev,
+				   struct amdgpu_reset_context *reset_context);
+
+int amdgpu_reset_perform_reset(struct amdgpu_device *adev,
+			       struct amdgpu_reset_context *reset_context);
+
+int amdgpu_reset_prepare_env(struct amdgpu_device *adev,
+			     struct amdgpu_reset_context *reset_context);
+int amdgpu_reset_restore_env(struct amdgpu_device *adev,
+			     struct amdgpu_reset_context *reset_context);
+
+struct amdgpu_reset_domain *amdgpu_reset_create_reset_domain(enum amdgpu_reset_domain_type type,
+							     char *wq_name);
+
+void amdgpu_reset_destroy_reset_domain(struct kref *ref);
+
+static inline bool amdgpu_reset_get_reset_domain(struct amdgpu_reset_domain *domain)
+{
+	return kref_get_unless_zero(&domain->refcount) != 0;
+}
+
+static inline void amdgpu_reset_put_reset_domain(struct amdgpu_reset_domain *domain)
+{
+	if (domain)
+		kref_put(&domain->refcount, amdgpu_reset_destroy_reset_domain);
+}
+
+static inline bool amdgpu_reset_domain_schedule(struct amdgpu_reset_domain *domain,
+						struct work_struct *work)
+{
+	return queue_work(domain->wq, work);
+}
+
+static inline bool amdgpu_reset_pending(struct amdgpu_reset_domain *domain)
+{
+	lockdep_assert_held(&domain->sem);
+	return rwsem_is_contended(&domain->sem);
+}
+
+void amdgpu_device_lock_reset_domain(struct amdgpu_reset_domain *reset_domain);
+
+void amdgpu_device_unlock_reset_domain(struct amdgpu_reset_domain *reset_domain);
+
+void amdgpu_reset_get_desc(struct amdgpu_reset_context *rst_ctxt, char *buf,
+			   size_t len);
+
+#define for_each_handler(i, handler, reset_ctl)                  \
+	for (i = 0; (i < AMDGPU_RESET_MAX_HANDLERS) &&           \
+		    (handler = (*reset_ctl->reset_handlers)[i]); \
+	     ++i)
+
+extern struct amdgpu_reset_handler xgmi_reset_on_init_handler;
+int amdgpu_reset_do_xgmi_reset_on_init(
+	struct amdgpu_reset_context *reset_context);
+
+bool amdgpu_reset_in_recovery(struct amdgpu_device *adev);
+
+static inline void amdgpu_reset_set_dpc_status(struct amdgpu_device *adev,
+					       bool status)
+{
+	adev->pcie_reset_ctx.occurs_dpc = status;
+	adev->no_hw_access = status;
+}
+
+static inline bool amdgpu_reset_in_dpc(struct amdgpu_device *adev)
+{
+	return adev->pcie_reset_ctx.occurs_dpc;
+}
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
index 5ce65280b396..8f6ce948c684 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
@@ -28,8 +28,9 @@
  */
 #include <linux/seq_file.h>
 #include <linux/slab.h>
+#include <linux/uaccess.h>
 #include <linux/debugfs.h>
-#include <drm/drmP.h>
+
 #include <drm/amdgpu_drm.h>
 #include "amdgpu.h"
 #include "atom.h"
@@ -47,21 +48,37 @@
  * wptr.  The GPU then starts fetching commands and executes
  * them until the pointers are equal again.
  */
-static int amdgpu_debugfs_ring_init(struct amdgpu_device *adev,
-				    struct amdgpu_ring *ring);
-static void amdgpu_debugfs_ring_fini(struct amdgpu_ring *ring);
+
+/**
+ * amdgpu_ring_max_ibs - Return max IBs that fit in a single submission.
+ *
+ * @type: ring type for which to return the limit.
+ */
+unsigned int amdgpu_ring_max_ibs(enum amdgpu_ring_type type)
+{
+	switch (type) {
+	case AMDGPU_RING_TYPE_GFX:
+		/* Need to keep at least 192 on GFX7+ for old radv. */
+		return 192;
+	case AMDGPU_RING_TYPE_COMPUTE:
+		return 125;
+	case AMDGPU_RING_TYPE_VCN_JPEG:
+		return 16;
+	default:
+		return 49;
+	}
+}
 
 /**
  * amdgpu_ring_alloc - allocate space on the ring buffer
  *
- * @adev: amdgpu_device pointer
  * @ring: amdgpu_ring structure holding ring information
  * @ndw: number of dwords to allocate in the ring buffer
  *
  * Allocate @ndw dwords in the ring buffer (all asics).
  * Returns 0 on success, error on failure.
  */
-int amdgpu_ring_alloc(struct amdgpu_ring *ring, unsigned ndw)
+int amdgpu_ring_alloc(struct amdgpu_ring *ring, unsigned int ndw)
 {
 	/* Align requested size with padding so unlock_commit can
 	 * pad safely */
@@ -82,6 +99,29 @@ int amdgpu_ring_alloc(struct amdgpu_ring *ring, unsigned ndw)
 	return 0;
 }
 
+/**
+ * amdgpu_ring_alloc_reemit - allocate space on the ring buffer for reemit
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @ndw: number of dwords to allocate in the ring buffer
+ *
+ * Allocate @ndw dwords in the ring buffer (all asics).
+ * doesn't check the max_dw limit as we may be reemitting
+ * several submissions.
+ */
+static void amdgpu_ring_alloc_reemit(struct amdgpu_ring *ring, unsigned int ndw)
+{
+	/* Align requested size with padding so unlock_commit can
+	 * pad safely */
+	ndw = (ndw + ring->funcs->align_mask) & ~ring->funcs->align_mask;
+
+	ring->count_dw = ndw;
+	ring->wptr_old = ring->wptr;
+
+	if (ring->funcs->begin_use)
+		ring->funcs->begin_use(ring);
+}
+
 /** amdgpu_ring_insert_nop - insert NOP packets
  *
  * @ring: amdgpu_ring structure holding ring information
@@ -91,13 +131,26 @@ int amdgpu_ring_alloc(struct amdgpu_ring *ring, unsigned ndw)
  */
 void amdgpu_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
 {
-	int i;
+	uint32_t occupied, chunk1, chunk2;
+
+	occupied = ring->wptr & ring->buf_mask;
+	chunk1 = ring->buf_mask + 1 - occupied;
+	chunk1 = (chunk1 >= count) ? count : chunk1;
+	chunk2 = count - chunk1;
+
+	if (chunk1)
+		memset32(&ring->ring[occupied], ring->funcs->nop, chunk1);
+
+	if (chunk2)
+		memset32(ring->ring, ring->funcs->nop, chunk2);
 
-	for (i = 0; i < count; i++)
-		amdgpu_ring_write(ring, ring->funcs->nop);
+	ring->wptr += count;
+	ring->wptr &= ring->ptr_mask;
+	ring->count_dw -= count;
 }
 
-/** amdgpu_ring_generic_pad_ib - pad IB with NOP packets
+/**
+ * amdgpu_ring_generic_pad_ib - pad IB with NOP packets
  *
  * @ring: amdgpu_ring structure holding ring information
  * @ib: IB to add NOP packets to
@@ -114,7 +167,6 @@ void amdgpu_ring_generic_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib)
  * amdgpu_ring_commit - tell the GPU to execute the new
  * commands on the ring buffer
  *
- * @adev: amdgpu_device pointer
  * @ring: amdgpu_ring structure holding ring information
  *
  * Update the wptr (write pointer) to tell the GPU to
@@ -124,19 +176,22 @@ void amdgpu_ring_commit(struct amdgpu_ring *ring)
 {
 	uint32_t count;
 
+	if (ring->count_dw < 0)
+		DRM_ERROR("amdgpu: writing more dwords to the ring than expected!\n");
+
 	/* We pad to match fetch size */
 	count = ring->funcs->align_mask + 1 -
 		(ring->wptr & ring->funcs->align_mask);
-	count %= ring->funcs->align_mask + 1;
-	ring->funcs->insert_nop(ring, count);
+	count &= ring->funcs->align_mask;
+
+	if (count != 0)
+		ring->funcs->insert_nop(ring, count);
 
 	mb();
 	amdgpu_ring_set_wptr(ring);
 
 	if (ring->funcs->end_use)
 		ring->funcs->end_use(ring);
-
-	amdgpu_ring_lru_touch(ring->adev, ring);
 }
 
 /**
@@ -154,23 +209,36 @@ void amdgpu_ring_undo(struct amdgpu_ring *ring)
 		ring->funcs->end_use(ring);
 }
 
+#define amdgpu_ring_get_gpu_addr(ring, offset)				\
+	 (ring->adev->wb.gpu_addr + offset * 4)
+
+#define amdgpu_ring_get_cpu_addr(ring, offset)				\
+	 (&ring->adev->wb.wb[offset])
+
 /**
  * amdgpu_ring_init - init driver ring struct.
  *
  * @adev: amdgpu_device pointer
  * @ring: amdgpu_ring structure holding ring information
- * @max_ndw: maximum number of dw for ring alloc
- * @nop: nop packet for this ring
+ * @max_dw: maximum number of dw for ring alloc
+ * @irq_src: interrupt source to use for this ring
+ * @irq_type: interrupt type to use for this ring
+ * @hw_prio: ring priority (NORMAL/HIGH)
+ * @sched_score: optional score atomic shared with other schedulers
  *
  * Initialize the driver information for the selected ring (all asics).
  * Returns 0 on success, error on failure.
  */
 int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
-		     unsigned max_dw, struct amdgpu_irq_src *irq_src,
-		     unsigned irq_type)
+		     unsigned int max_dw, struct amdgpu_irq_src *irq_src,
+		     unsigned int irq_type, unsigned int hw_prio,
+		     atomic_t *sched_score)
 {
 	int r;
 	int sched_hw_submission = amdgpu_sched_hw_submission;
+	u32 *num_sched;
+	u32 hw_ip;
+	unsigned int max_ibs_dw;
 
 	/* Set the hw submission limit higher for KIQ because
 	 * it's used for a number of gfx/compute tasks by both
@@ -180,78 +248,144 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
 	 */
 	if (ring->funcs->type == AMDGPU_RING_TYPE_KIQ)
 		sched_hw_submission = max(sched_hw_submission, 256);
+	if (ring->funcs->type == AMDGPU_RING_TYPE_MES)
+		sched_hw_submission = 8;
+	else if (ring == &adev->sdma.instance[0].page)
+		sched_hw_submission = 256;
 
 	if (ring->adev == NULL) {
 		if (adev->num_rings >= AMDGPU_MAX_RINGS)
 			return -EINVAL;
 
 		ring->adev = adev;
+		ring->num_hw_submission = sched_hw_submission;
+		ring->sched_score = sched_score;
+		ring->vmid_wait = dma_fence_get_stub();
+
 		ring->idx = adev->num_rings++;
 		adev->rings[ring->idx] = ring;
-		r = amdgpu_fence_driver_init_ring(ring, sched_hw_submission);
+
+		r = amdgpu_fence_driver_init_ring(ring);
 		if (r)
 			return r;
 	}
 
-	r = amdgpu_wb_get(adev, &ring->rptr_offs);
+	r = amdgpu_device_wb_get(adev, &ring->rptr_offs);
 	if (r) {
 		dev_err(adev->dev, "(%d) ring rptr_offs wb alloc failed\n", r);
 		return r;
 	}
 
-	r = amdgpu_wb_get(adev, &ring->wptr_offs);
+	r = amdgpu_device_wb_get(adev, &ring->wptr_offs);
 	if (r) {
 		dev_err(adev->dev, "(%d) ring wptr_offs wb alloc failed\n", r);
 		return r;
 	}
 
-	r = amdgpu_wb_get(adev, &ring->fence_offs);
+	r = amdgpu_device_wb_get(adev, &ring->fence_offs);
 	if (r) {
 		dev_err(adev->dev, "(%d) ring fence_offs wb alloc failed\n", r);
 		return r;
 	}
 
-	r = amdgpu_wb_get(adev, &ring->cond_exe_offs);
+	r = amdgpu_device_wb_get(adev, &ring->trail_fence_offs);
 	if (r) {
-		dev_err(adev->dev, "(%d) ring cond_exec_polling wb alloc failed\n", r);
+		dev_err(adev->dev, "(%d) ring trail_fence_offs wb alloc failed\n", r);
 		return r;
 	}
-	ring->cond_exe_gpu_addr = adev->wb.gpu_addr + (ring->cond_exe_offs * 4);
-	ring->cond_exe_cpu_addr = &adev->wb.wb[ring->cond_exe_offs];
-	/* always set cond_exec_polling to CONTINUE */
-	*ring->cond_exe_cpu_addr = 1;
 
-	r = amdgpu_fence_driver_start_ring(ring, irq_src, irq_type);
+	r = amdgpu_device_wb_get(adev, &ring->cond_exe_offs);
 	if (r) {
-		dev_err(adev->dev, "failed initializing fences (%d).\n", r);
+		dev_err(adev->dev, "(%d) ring cond_exec_polling wb alloc failed\n", r);
 		return r;
 	}
 
-	ring->ring_size = roundup_pow_of_two(max_dw * 4 * sched_hw_submission);
+	ring->fence_gpu_addr =
+		amdgpu_ring_get_gpu_addr(ring, ring->fence_offs);
+	ring->fence_cpu_addr =
+		amdgpu_ring_get_cpu_addr(ring, ring->fence_offs);
+
+	ring->rptr_gpu_addr =
+		amdgpu_ring_get_gpu_addr(ring, ring->rptr_offs);
+	ring->rptr_cpu_addr =
+		amdgpu_ring_get_cpu_addr(ring, ring->rptr_offs);
+
+	ring->wptr_gpu_addr =
+		amdgpu_ring_get_gpu_addr(ring, ring->wptr_offs);
+	ring->wptr_cpu_addr =
+		amdgpu_ring_get_cpu_addr(ring, ring->wptr_offs);
+
+	ring->trail_fence_gpu_addr =
+		amdgpu_ring_get_gpu_addr(ring, ring->trail_fence_offs);
+	ring->trail_fence_cpu_addr =
+		amdgpu_ring_get_cpu_addr(ring, ring->trail_fence_offs);
+
+	ring->cond_exe_gpu_addr =
+		amdgpu_ring_get_gpu_addr(ring, ring->cond_exe_offs);
+	ring->cond_exe_cpu_addr =
+		amdgpu_ring_get_cpu_addr(ring, ring->cond_exe_offs);
+
+	/* always set cond_exec_polling to CONTINUE */
+	*ring->cond_exe_cpu_addr = 1;
+
+	if (ring->funcs->type != AMDGPU_RING_TYPE_CPER) {
+		r = amdgpu_fence_driver_start_ring(ring, irq_src, irq_type);
+		if (r) {
+			dev_err(adev->dev, "failed initializing fences (%d).\n", r);
+			return r;
+		}
+
+		max_ibs_dw = ring->funcs->emit_frame_size +
+			     amdgpu_ring_max_ibs(ring->funcs->type) * ring->funcs->emit_ib_size;
+		max_ibs_dw = (max_ibs_dw + ring->funcs->align_mask) & ~ring->funcs->align_mask;
+
+		if (WARN_ON(max_ibs_dw > max_dw))
+			max_dw = max_ibs_dw;
+
+		ring->ring_size = roundup_pow_of_two(max_dw * 4 * sched_hw_submission);
+	} else {
+		ring->ring_size = roundup_pow_of_two(max_dw * 4);
+		ring->count_dw = (ring->ring_size - 4) >> 2;
+		/* ring buffer is empty now */
+		ring->wptr = *ring->rptr_cpu_addr = 0;
+	}
 
 	ring->buf_mask = (ring->ring_size / 4) - 1;
 	ring->ptr_mask = ring->funcs->support_64bit_ptrs ?
 		0xffffffffffffffff : ring->buf_mask;
+	/*  Initialize cached_rptr to 0 */
+	ring->cached_rptr = 0;
+
+	if (!ring->ring_backup) {
+		ring->ring_backup = kvzalloc(ring->ring_size, GFP_KERNEL);
+		if (!ring->ring_backup)
+			return -ENOMEM;
+	}
+
 	/* Allocate ring buffer */
 	if (ring->ring_obj == NULL) {
-		r = amdgpu_bo_create_kernel(adev, ring->ring_size, PAGE_SIZE,
+		r = amdgpu_bo_create_kernel(adev, ring->ring_size + ring->funcs->extra_bytes,
+					    PAGE_SIZE,
 					    AMDGPU_GEM_DOMAIN_GTT,
 					    &ring->ring_obj,
 					    &ring->gpu_addr,
 					    (void **)&ring->ring);
 		if (r) {
 			dev_err(adev->dev, "(%d) ring create failed\n", r);
+			kvfree(ring->ring_backup);
 			return r;
 		}
 		amdgpu_ring_clear_ring(ring);
 	}
 
 	ring->max_dw = max_dw;
-	INIT_LIST_HEAD(&ring->lru_list);
-	amdgpu_ring_lru_touch(adev, ring);
+	ring->hw_prio = hw_prio;
 
-	if (amdgpu_debugfs_ring_init(adev, ring)) {
-		DRM_ERROR("Failed to register debugfs file for rings !\n");
+	if (!ring->no_scheduler && ring->funcs->type < AMDGPU_HW_IP_NUM) {
+		hw_ip = ring->funcs->type;
+		num_sched = &adev->gpu_sched[hw_ip][hw_prio].num_scheds;
+		adev->gpu_sched[hw_ip][hw_prio].sched[(*num_sched)++] =
+			&ring->sched;
 	}
 
 	return 0;
@@ -260,110 +394,95 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
 /**
  * amdgpu_ring_fini - tear down the driver ring struct.
  *
- * @adev: amdgpu_device pointer
  * @ring: amdgpu_ring structure holding ring information
  *
  * Tear down the driver information for the selected ring (all asics).
  */
 void amdgpu_ring_fini(struct amdgpu_ring *ring)
 {
-	ring->ready = false;
 
 	/* Not to finish a ring which is not initialized */
 	if (!(ring->adev) || !(ring->adev->rings[ring->idx]))
 		return;
 
-	amdgpu_wb_free(ring->adev, ring->rptr_offs);
-	amdgpu_wb_free(ring->adev, ring->wptr_offs);
+	ring->sched.ready = false;
 
-	amdgpu_wb_free(ring->adev, ring->cond_exe_offs);
-	amdgpu_wb_free(ring->adev, ring->fence_offs);
+	amdgpu_device_wb_free(ring->adev, ring->rptr_offs);
+	amdgpu_device_wb_free(ring->adev, ring->wptr_offs);
+
+	amdgpu_device_wb_free(ring->adev, ring->cond_exe_offs);
+	amdgpu_device_wb_free(ring->adev, ring->fence_offs);
 
 	amdgpu_bo_free_kernel(&ring->ring_obj,
 			      &ring->gpu_addr,
 			      (void **)&ring->ring);
+	kvfree(ring->ring_backup);
+	ring->ring_backup = NULL;
 
-	amdgpu_debugfs_ring_fini(ring);
-
-	ring->adev->rings[ring->idx] = NULL;
-}
-
-static void amdgpu_ring_lru_touch_locked(struct amdgpu_device *adev,
-					 struct amdgpu_ring *ring)
-{
-	/* list_move_tail handles the case where ring isn't part of the list */
-	list_move_tail(&ring->lru_list, &adev->ring_lru_list);
+	dma_fence_put(ring->vmid_wait);
+	ring->vmid_wait = NULL;
+	ring->me = 0;
 }
 
-static bool amdgpu_ring_is_blacklisted(struct amdgpu_ring *ring,
-				       int *blacklist, int num_blacklist)
+/**
+ * amdgpu_ring_emit_reg_write_reg_wait_helper - ring helper
+ *
+ * @ring: ring to write to
+ * @reg0: register to write
+ * @reg1: register to wait on
+ * @ref: reference value to write/wait on
+ * @mask: mask to wait on
+ *
+ * Helper for rings that don't support write and wait in a
+ * single oneshot packet.
+ */
+void amdgpu_ring_emit_reg_write_reg_wait_helper(struct amdgpu_ring *ring,
+						uint32_t reg0, uint32_t reg1,
+						uint32_t ref, uint32_t mask)
 {
-	int i;
-
-	for (i = 0; i < num_blacklist; i++) {
-		if (ring->idx == blacklist[i])
-			return true;
-	}
-
-	return false;
+	amdgpu_ring_emit_wreg(ring, reg0, ref);
+	amdgpu_ring_emit_reg_wait(ring, reg1, mask, mask);
 }
 
 /**
- * amdgpu_ring_lru_get - get the least recently used ring for a HW IP block
+ * amdgpu_ring_soft_recovery - try to soft recover a ring lockup
  *
- * @adev: amdgpu_device pointer
- * @type: amdgpu_ring_type enum
- * @blacklist: blacklisted ring ids array
- * @num_blacklist: number of entries in @blacklist
- * @ring: output ring
+ * @ring: ring to try the recovery on
+ * @vmid: VMID we try to get going again
+ * @fence: timedout fence
  *
- * Retrieve the amdgpu_ring structure for the least recently used ring of
- * a specific IP block (all asics).
- * Returns 0 on success, error on failure.
+ * Tries to get a ring proceeding again when it is stuck.
  */
-int amdgpu_ring_lru_get(struct amdgpu_device *adev, int type, int *blacklist,
-			int num_blacklist, struct amdgpu_ring **ring)
+bool amdgpu_ring_soft_recovery(struct amdgpu_ring *ring, unsigned int vmid,
+			       struct dma_fence *fence)
 {
-	struct amdgpu_ring *entry;
+	unsigned long flags;
+	ktime_t deadline;
+	bool ret;
 
-	/* List is sorted in LRU order, find first entry corresponding
-	 * to the desired HW IP */
-	*ring = NULL;
-	spin_lock(&adev->ring_lru_list_lock);
-	list_for_each_entry(entry, &adev->ring_lru_list, lru_list) {
-		if (entry->funcs->type != type)
-			continue;
+	if (unlikely(ring->adev->debug_disable_soft_recovery))
+		return false;
 
-		if (amdgpu_ring_is_blacklisted(entry, blacklist, num_blacklist))
-			continue;
+	deadline = ktime_add_us(ktime_get(), 10000);
 
-		*ring = entry;
-		amdgpu_ring_lru_touch_locked(adev, *ring);
-		break;
-	}
-	spin_unlock(&adev->ring_lru_list_lock);
+	if (amdgpu_sriov_vf(ring->adev) || !ring->funcs->soft_recovery || !fence)
+		return false;
 
-	if (!*ring) {
-		DRM_ERROR("Ring LRU contains no entries for ring type:%d\n", type);
-		return -EINVAL;
-	}
+	spin_lock_irqsave(fence->lock, flags);
+	if (!dma_fence_is_signaled_locked(fence))
+		dma_fence_set_error(fence, -ENODATA);
+	spin_unlock_irqrestore(fence->lock, flags);
 
-	return 0;
-}
+	while (!dma_fence_is_signaled(fence) &&
+	       ktime_to_ns(ktime_sub(deadline, ktime_get())) > 0)
+		ring->funcs->soft_recovery(ring, vmid);
 
-/**
- * amdgpu_ring_lru_touch - mark a ring as recently being used
- *
- * @adev: amdgpu_device pointer
- * @ring: ring to touch
- *
- * Move @ring to the tail of the lru list
- */
-void amdgpu_ring_lru_touch(struct amdgpu_device *adev, struct amdgpu_ring *ring)
-{
-	spin_lock(&adev->ring_lru_list_lock);
-	amdgpu_ring_lru_touch_locked(adev, ring);
-	spin_unlock(&adev->ring_lru_list_lock);
+	ret = dma_fence_is_signaled(fence);
+	/* increment the counter only if soft reset worked */
+	if (ret)
+		atomic_inc(&ring->adev->gpu_reset_counter);
+
+	return ret;
 }
 
 /*
@@ -382,8 +501,10 @@ static ssize_t amdgpu_debugfs_ring_read(struct file *f, char __user *buf,
 					size_t size, loff_t *pos)
 {
 	struct amdgpu_ring *ring = file_inode(f)->i_private;
-	int r, i;
 	uint32_t value, result, early[3];
+	uint64_t p;
+	loff_t i;
+	int r;
 
 	if (*pos & 3 || size & 3)
 		return -EINVAL;
@@ -391,13 +512,18 @@ static ssize_t amdgpu_debugfs_ring_read(struct file *f, char __user *buf,
 	result = 0;
 
 	if (*pos < 12) {
-		early[0] = amdgpu_ring_get_rptr(ring);
+		if (ring->funcs->type == AMDGPU_RING_TYPE_CPER)
+			mutex_lock(&ring->adev->cper.ring_lock);
+
+		early[0] = amdgpu_ring_get_rptr(ring) & ring->buf_mask;
 		early[1] = amdgpu_ring_get_wptr(ring) & ring->buf_mask;
 		early[2] = ring->wptr & ring->buf_mask;
 		for (i = *pos / 4; i < 3 && size; i++) {
 			r = put_user(early[i], (uint32_t *)buf);
-			if (r)
-				return r;
+			if (r) {
+				result = r;
+				goto out;
+			}
 			buf += 4;
 			result += 4;
 			size -= 4;
@@ -405,56 +531,328 @@ static ssize_t amdgpu_debugfs_ring_read(struct file *f, char __user *buf,
 		}
 	}
 
-	while (size) {
-		if (*pos >= (ring->ring_size + 12))
-			return result;
+	if (ring->funcs->type != AMDGPU_RING_TYPE_CPER) {
+		while (size) {
+			if (*pos >= (ring->ring_size + 12))
+				return result;
 
-		value = ring->ring[(*pos - 12)/4];
-		r = put_user(value, (uint32_t*)buf);
-		if (r)
-			return r;
-		buf += 4;
-		result += 4;
-		size -= 4;
-		*pos += 4;
+			value = ring->ring[(*pos - 12)/4];
+			r = put_user(value, (uint32_t *)buf);
+			if (r)
+				return r;
+			buf += 4;
+			result += 4;
+			size -= 4;
+			*pos += 4;
+		}
+	} else {
+		p = early[0];
+		if (early[0] <= early[1])
+			size = (early[1] - early[0]);
+		else
+			size = ring->ring_size - (early[0] - early[1]);
+
+		while (size) {
+			if (p == early[1])
+				goto out;
+
+			value = ring->ring[p];
+			r = put_user(value, (uint32_t *)buf);
+			if (r) {
+				result = r;
+				goto out;
+			}
+
+			buf += 4;
+			result += 4;
+			size--;
+			p++;
+			p &= ring->ptr_mask;
+		}
 	}
 
+out:
+	if (ring->funcs->type == AMDGPU_RING_TYPE_CPER)
+		mutex_unlock(&ring->adev->cper.ring_lock);
+
 	return result;
 }
 
+static ssize_t amdgpu_debugfs_virt_ring_read(struct file *f, char __user *buf,
+	size_t size, loff_t *pos)
+{
+	struct amdgpu_ring *ring = file_inode(f)->i_private;
+
+	if (*pos & 3 || size & 3)
+		return -EINVAL;
+
+	if (ring->funcs->type == AMDGPU_RING_TYPE_CPER)
+		amdgpu_virt_req_ras_cper_dump(ring->adev, false);
+
+	return amdgpu_debugfs_ring_read(f, buf, size, pos);
+}
+
 static const struct file_operations amdgpu_debugfs_ring_fops = {
 	.owner = THIS_MODULE,
 	.read = amdgpu_debugfs_ring_read,
 	.llseek = default_llseek
 };
 
+static const struct file_operations amdgpu_debugfs_virt_ring_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_virt_ring_read,
+	.llseek = default_llseek
+};
+
+static ssize_t amdgpu_debugfs_mqd_read(struct file *f, char __user *buf,
+				       size_t size, loff_t *pos)
+{
+	struct amdgpu_ring *ring = file_inode(f)->i_private;
+	ssize_t bytes = min_t(ssize_t, ring->mqd_size - *pos, size);
+	void *from = ((u8 *)ring->mqd_ptr) + *pos;
+
+	if (*pos > ring->mqd_size)
+		return 0;
+
+	if (copy_to_user(buf, from, bytes))
+		return -EFAULT;
+
+	*pos += bytes;
+	return bytes;
+}
+
+static const struct file_operations amdgpu_debugfs_mqd_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_mqd_read,
+	.llseek = default_llseek
+};
+
+static int amdgpu_debugfs_ring_error(void *data, u64 val)
+{
+	struct amdgpu_ring *ring = data;
+
+	amdgpu_fence_driver_set_error(ring, val);
+	return 0;
+}
+
+DEFINE_DEBUGFS_ATTRIBUTE_SIGNED(amdgpu_debugfs_error_fops, NULL,
+				amdgpu_debugfs_ring_error, "%lld\n");
+
 #endif
 
-static int amdgpu_debugfs_ring_init(struct amdgpu_device *adev,
-				    struct amdgpu_ring *ring)
+void amdgpu_debugfs_ring_init(struct amdgpu_device *adev,
+			      struct amdgpu_ring *ring)
 {
 #if defined(CONFIG_DEBUG_FS)
-	struct drm_minor *minor = adev->ddev->primary;
-	struct dentry *ent, *root = minor->debugfs_root;
+	struct drm_minor *minor = adev_to_drm(adev)->primary;
+	struct dentry *root = minor->debugfs_root;
 	char name[32];
 
 	sprintf(name, "amdgpu_ring_%s", ring->name);
+	if (amdgpu_sriov_vf(adev))
+		debugfs_create_file_size(name, S_IFREG | 0444, root, ring,
+					 &amdgpu_debugfs_virt_ring_fops,
+					 ring->ring_size + 12);
+	else
+		debugfs_create_file_size(name, S_IFREG | 0444, root, ring,
+					 &amdgpu_debugfs_ring_fops,
+					 ring->ring_size + 12);
+
+	if (ring->mqd_obj) {
+		sprintf(name, "amdgpu_mqd_%s", ring->name);
+		debugfs_create_file_size(name, S_IFREG | 0444, root, ring,
+					 &amdgpu_debugfs_mqd_fops,
+					 ring->mqd_size);
+	}
 
-	ent = debugfs_create_file(name,
-				  S_IFREG | S_IRUGO, root,
-				  ring, &amdgpu_debugfs_ring_fops);
-	if (!ent)
-		return -ENOMEM;
+	sprintf(name, "amdgpu_error_%s", ring->name);
+	debugfs_create_file(name, 0200, root, ring,
+			    &amdgpu_debugfs_error_fops);
 
-	i_size_write(ent->d_inode, ring->ring_size + 12);
-	ring->ent = ent;
 #endif
+}
+
+/**
+ * amdgpu_ring_test_helper - tests ring and set sched readiness status
+ *
+ * @ring: ring to try the recovery on
+ *
+ * Tests ring and set sched readiness status
+ *
+ * Returns 0 on success, error on failure.
+ */
+int amdgpu_ring_test_helper(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int r;
+
+	r = amdgpu_ring_test_ring(ring);
+	if (r)
+		DRM_DEV_ERROR(adev->dev, "ring %s test failed (%d)\n",
+			      ring->name, r);
+	else
+		DRM_DEV_DEBUG(adev->dev, "ring test on %s succeeded\n",
+			      ring->name);
+
+	ring->sched.ready = !r;
+
+	return r;
+}
+
+static void amdgpu_ring_to_mqd_prop(struct amdgpu_ring *ring,
+				    struct amdgpu_mqd_prop *prop)
+{
+	struct amdgpu_device *adev = ring->adev;
+	bool is_high_prio_compute = ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE &&
+				    amdgpu_gfx_is_high_priority_compute_queue(adev, ring);
+	bool is_high_prio_gfx = ring->funcs->type == AMDGPU_RING_TYPE_GFX &&
+				amdgpu_gfx_is_high_priority_graphics_queue(adev, ring);
+
+	memset(prop, 0, sizeof(*prop));
+
+	prop->mqd_gpu_addr = ring->mqd_gpu_addr;
+	prop->hqd_base_gpu_addr = ring->gpu_addr;
+	prop->rptr_gpu_addr = ring->rptr_gpu_addr;
+	prop->wptr_gpu_addr = ring->wptr_gpu_addr;
+	prop->queue_size = ring->ring_size;
+	prop->eop_gpu_addr = ring->eop_gpu_addr;
+	prop->use_doorbell = ring->use_doorbell;
+	prop->doorbell_index = ring->doorbell_index;
+	prop->kernel_queue = true;
+
+	/* map_queues packet doesn't need activate the queue,
+	 * so only kiq need set this field.
+	 */
+	prop->hqd_active = ring->funcs->type == AMDGPU_RING_TYPE_KIQ;
+
+	prop->allow_tunneling = is_high_prio_compute;
+	if (is_high_prio_compute || is_high_prio_gfx) {
+		prop->hqd_pipe_priority = AMDGPU_GFX_PIPE_PRIO_HIGH;
+		prop->hqd_queue_priority = AMDGPU_GFX_QUEUE_PRIORITY_MAXIMUM;
+	}
+}
+
+int amdgpu_ring_init_mqd(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_mqd *mqd_mgr;
+	struct amdgpu_mqd_prop prop;
+
+	amdgpu_ring_to_mqd_prop(ring, &prop);
+
+	ring->wptr = 0;
+
+	if (ring->funcs->type == AMDGPU_RING_TYPE_KIQ)
+		mqd_mgr = &adev->mqds[AMDGPU_HW_IP_COMPUTE];
+	else
+		mqd_mgr = &adev->mqds[ring->funcs->type];
+
+	return mqd_mgr->init_mqd(adev, ring->mqd_ptr, &prop);
+}
+
+void amdgpu_ring_ib_begin(struct amdgpu_ring *ring)
+{
+	if (ring->is_sw_ring)
+		amdgpu_sw_ring_ib_begin(ring);
+}
+
+void amdgpu_ring_ib_end(struct amdgpu_ring *ring)
+{
+	if (ring->is_sw_ring)
+		amdgpu_sw_ring_ib_end(ring);
+}
+
+void amdgpu_ring_ib_on_emit_cntl(struct amdgpu_ring *ring)
+{
+	if (ring->is_sw_ring)
+		amdgpu_sw_ring_ib_mark_offset(ring, AMDGPU_MUX_OFFSET_TYPE_CONTROL);
+}
+
+void amdgpu_ring_ib_on_emit_ce(struct amdgpu_ring *ring)
+{
+	if (ring->is_sw_ring)
+		amdgpu_sw_ring_ib_mark_offset(ring, AMDGPU_MUX_OFFSET_TYPE_CE);
+}
+
+void amdgpu_ring_ib_on_emit_de(struct amdgpu_ring *ring)
+{
+	if (ring->is_sw_ring)
+		amdgpu_sw_ring_ib_mark_offset(ring, AMDGPU_MUX_OFFSET_TYPE_DE);
+}
+
+bool amdgpu_ring_sched_ready(struct amdgpu_ring *ring)
+{
+	if (!ring)
+		return false;
+
+	if (ring->no_scheduler || !drm_sched_wqueue_ready(&ring->sched))
+		return false;
+
+	return true;
+}
+
+void amdgpu_ring_reset_helper_begin(struct amdgpu_ring *ring,
+				    struct amdgpu_fence *guilty_fence)
+{
+	/* Stop the scheduler to prevent anybody else from touching the ring buffer. */
+	drm_sched_wqueue_stop(&ring->sched);
+	/* back up the non-guilty commands */
+	amdgpu_ring_backup_unprocessed_commands(ring, guilty_fence);
+}
+
+int amdgpu_ring_reset_helper_end(struct amdgpu_ring *ring,
+				 struct amdgpu_fence *guilty_fence)
+{
+	unsigned int i;
+	int r;
+
+	/* verify that the ring is functional */
+	r = amdgpu_ring_test_ring(ring);
+	if (r)
+		return r;
+
+	/* signal the fence of the bad job */
+	if (guilty_fence)
+		amdgpu_fence_driver_guilty_force_completion(guilty_fence);
+	/* Re-emit the non-guilty commands */
+	if (ring->ring_backup_entries_to_copy) {
+		amdgpu_ring_alloc_reemit(ring, ring->ring_backup_entries_to_copy);
+		for (i = 0; i < ring->ring_backup_entries_to_copy; i++)
+			amdgpu_ring_write(ring, ring->ring_backup[i]);
+		amdgpu_ring_commit(ring);
+	}
+	/* Start the scheduler again */
+	drm_sched_wqueue_start(&ring->sched);
 	return 0;
 }
 
-static void amdgpu_debugfs_ring_fini(struct amdgpu_ring *ring)
+bool amdgpu_ring_is_reset_type_supported(struct amdgpu_ring *ring,
+					 u32 reset_type)
 {
-#if defined(CONFIG_DEBUG_FS)
-	debugfs_remove(ring->ent);
-#endif
+	switch (ring->funcs->type) {
+	case AMDGPU_RING_TYPE_GFX:
+		if (ring->adev->gfx.gfx_supported_reset & reset_type)
+			return true;
+		break;
+	case AMDGPU_RING_TYPE_COMPUTE:
+		if (ring->adev->gfx.compute_supported_reset & reset_type)
+			return true;
+		break;
+	case AMDGPU_RING_TYPE_SDMA:
+		if (ring->adev->sdma.supported_reset & reset_type)
+			return true;
+		break;
+	case AMDGPU_RING_TYPE_VCN_DEC:
+	case AMDGPU_RING_TYPE_VCN_ENC:
+		if (ring->adev->vcn.supported_reset & reset_type)
+			return true;
+		break;
+	case AMDGPU_RING_TYPE_VCN_JPEG:
+		if (ring->adev->jpeg.supported_reset & reset_type)
+			return true;
+		break;
+	default:
+		break;
+	}
+	return false;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
index 322d25299a00..b6b649179776 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
@@ -24,48 +24,101 @@
 #ifndef __AMDGPU_RING_H__
 #define __AMDGPU_RING_H__
 
-#include "gpu_scheduler.h"
+#include <drm/amdgpu_drm.h>
+#include <drm/gpu_scheduler.h>
+#include <drm/drm_print.h>
+#include <drm/drm_suballoc.h>
+
+struct amdgpu_device;
+struct amdgpu_ring;
+struct amdgpu_ib;
+struct amdgpu_cs_parser;
+struct amdgpu_job;
+struct amdgpu_vm;
 
 /* max number of rings */
-#define AMDGPU_MAX_RINGS		18
-#define AMDGPU_MAX_GFX_RINGS		1
+#define AMDGPU_MAX_RINGS		149
+#define AMDGPU_MAX_HWIP_RINGS		64
+#define AMDGPU_MAX_GFX_RINGS		2
+#define AMDGPU_MAX_SW_GFX_RINGS         2
 #define AMDGPU_MAX_COMPUTE_RINGS	8
 #define AMDGPU_MAX_VCE_RINGS		3
 #define AMDGPU_MAX_UVD_ENC_RINGS	2
+#define AMDGPU_MAX_VPE_RINGS		2
+
+enum amdgpu_ring_priority_level {
+	AMDGPU_RING_PRIO_0,
+	AMDGPU_RING_PRIO_1,
+	AMDGPU_RING_PRIO_DEFAULT = 1,
+	AMDGPU_RING_PRIO_2,
+	AMDGPU_RING_PRIO_MAX
+};
 
 /* some special values for the owner field */
-#define AMDGPU_FENCE_OWNER_UNDEFINED	((void*)0ul)
-#define AMDGPU_FENCE_OWNER_VM		((void*)1ul)
+#define AMDGPU_FENCE_OWNER_UNDEFINED	((void *)0ul)
+#define AMDGPU_FENCE_OWNER_VM		((void *)1ul)
+#define AMDGPU_FENCE_OWNER_KFD		((void *)2ul)
 
 #define AMDGPU_FENCE_FLAG_64BIT         (1 << 0)
 #define AMDGPU_FENCE_FLAG_INT           (1 << 1)
+#define AMDGPU_FENCE_FLAG_TC_WB_ONLY    (1 << 2)
+#define AMDGPU_FENCE_FLAG_EXEC          (1 << 3)
+
+#define to_amdgpu_ring(s) container_of((s), struct amdgpu_ring, sched)
+
+#define AMDGPU_IB_POOL_SIZE	(1024 * 1024)
 
 enum amdgpu_ring_type {
-	AMDGPU_RING_TYPE_GFX,
-	AMDGPU_RING_TYPE_COMPUTE,
-	AMDGPU_RING_TYPE_SDMA,
-	AMDGPU_RING_TYPE_UVD,
-	AMDGPU_RING_TYPE_VCE,
+	AMDGPU_RING_TYPE_GFX		= AMDGPU_HW_IP_GFX,
+	AMDGPU_RING_TYPE_COMPUTE	= AMDGPU_HW_IP_COMPUTE,
+	AMDGPU_RING_TYPE_SDMA		= AMDGPU_HW_IP_DMA,
+	AMDGPU_RING_TYPE_UVD		= AMDGPU_HW_IP_UVD,
+	AMDGPU_RING_TYPE_VCE		= AMDGPU_HW_IP_VCE,
+	AMDGPU_RING_TYPE_UVD_ENC	= AMDGPU_HW_IP_UVD_ENC,
+	AMDGPU_RING_TYPE_VCN_DEC	= AMDGPU_HW_IP_VCN_DEC,
+	AMDGPU_RING_TYPE_VCN_ENC	= AMDGPU_HW_IP_VCN_ENC,
+	AMDGPU_RING_TYPE_VCN_JPEG	= AMDGPU_HW_IP_VCN_JPEG,
+	AMDGPU_RING_TYPE_VPE		= AMDGPU_HW_IP_VPE,
 	AMDGPU_RING_TYPE_KIQ,
-	AMDGPU_RING_TYPE_UVD_ENC,
-	AMDGPU_RING_TYPE_VCN_DEC,
-	AMDGPU_RING_TYPE_VCN_ENC
+	AMDGPU_RING_TYPE_MES,
+	AMDGPU_RING_TYPE_UMSCH_MM,
+	AMDGPU_RING_TYPE_CPER,
 };
 
-struct amdgpu_device;
-struct amdgpu_ring;
-struct amdgpu_ib;
-struct amdgpu_cs_parser;
+enum amdgpu_ib_pool_type {
+	/* Normal submissions to the top of the pipeline. */
+	AMDGPU_IB_POOL_DELAYED,
+	/* Immediate submissions to the bottom of the pipeline. */
+	AMDGPU_IB_POOL_IMMEDIATE,
+	/* Direct submission to the ring buffer during init and reset. */
+	AMDGPU_IB_POOL_DIRECT,
+
+	AMDGPU_IB_POOL_MAX
+};
+
+struct amdgpu_ib {
+	struct drm_suballoc		*sa_bo;
+	uint32_t			length_dw;
+	uint64_t			gpu_addr;
+	uint32_t			*ptr;
+	uint32_t			flags;
+};
+
+struct amdgpu_sched {
+	u32				num_scheds;
+	struct drm_gpu_scheduler	*sched[AMDGPU_MAX_HWIP_RINGS];
+};
 
 /*
  * Fences.
  */
 struct amdgpu_fence_driver {
 	uint64_t			gpu_addr;
-	volatile uint32_t		*cpu_addr;
+	uint32_t			*cpu_addr;
 	/* sync_seq is protected by ring emission lock */
 	uint32_t			sync_seq;
 	atomic_t			last_seq;
+	u64				signalled_wptr;
 	bool				initialized;
 	struct amdgpu_irq_src		*irq_src;
 	unsigned			irq_type;
@@ -75,55 +128,127 @@ struct amdgpu_fence_driver {
 	struct dma_fence		**fences;
 };
 
-int amdgpu_fence_driver_init(struct amdgpu_device *adev);
-void amdgpu_fence_driver_fini(struct amdgpu_device *adev);
-void amdgpu_fence_driver_force_completion(struct amdgpu_device *adev);
-void amdgpu_fence_driver_force_completion_ring(struct amdgpu_ring *ring);
+/*
+ * Fences mark an event in the GPUs pipeline and are used
+ * for GPU/CPU synchronization.  When the fence is written,
+ * it is expected that all buffers associated with that fence
+ * are no longer in use by the associated ring on the GPU and
+ * that the relevant GPU caches have been flushed.
+ */
+
+struct amdgpu_fence {
+	struct dma_fence base;
+
+	/* RB, DMA, etc. */
+	struct amdgpu_ring		*ring;
+	ktime_t				start_timestamp;
+
+	/* wptr for the fence for resets */
+	u64				wptr;
+	/* fence context for resets */
+	u64				context;
+	uint32_t			seq;
+};
 
-int amdgpu_fence_driver_init_ring(struct amdgpu_ring *ring,
-				  unsigned num_hw_submission);
+extern const struct drm_sched_backend_ops amdgpu_sched_ops;
+
+void amdgpu_fence_driver_clear_job_fences(struct amdgpu_ring *ring);
+void amdgpu_fence_driver_set_error(struct amdgpu_ring *ring, int error);
+void amdgpu_fence_driver_force_completion(struct amdgpu_ring *ring);
+void amdgpu_fence_driver_guilty_force_completion(struct amdgpu_fence *fence);
+void amdgpu_fence_save_wptr(struct dma_fence *fence);
+
+int amdgpu_fence_driver_init_ring(struct amdgpu_ring *ring);
 int amdgpu_fence_driver_start_ring(struct amdgpu_ring *ring,
 				   struct amdgpu_irq_src *irq_src,
 				   unsigned irq_type);
-void amdgpu_fence_driver_suspend(struct amdgpu_device *adev);
-void amdgpu_fence_driver_resume(struct amdgpu_device *adev);
-int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **fence);
-void amdgpu_fence_process(struct amdgpu_ring *ring);
+void amdgpu_fence_driver_hw_init(struct amdgpu_device *adev);
+void amdgpu_fence_driver_hw_fini(struct amdgpu_device *adev);
+int amdgpu_fence_driver_sw_init(struct amdgpu_device *adev);
+void amdgpu_fence_driver_sw_fini(struct amdgpu_device *adev);
+int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **f,
+		      struct amdgpu_fence *af, unsigned int flags);
+int amdgpu_fence_emit_polling(struct amdgpu_ring *ring, uint32_t *s,
+			      uint32_t timeout);
+bool amdgpu_fence_process(struct amdgpu_ring *ring);
 int amdgpu_fence_wait_empty(struct amdgpu_ring *ring);
+signed long amdgpu_fence_wait_polling(struct amdgpu_ring *ring,
+				      uint32_t wait_seq,
+				      signed long timeout);
 unsigned amdgpu_fence_count_emitted(struct amdgpu_ring *ring);
 
+void amdgpu_fence_driver_isr_toggle(struct amdgpu_device *adev, bool stop);
+
+u64 amdgpu_fence_last_unsignaled_time_us(struct amdgpu_ring *ring);
+void amdgpu_fence_update_start_timestamp(struct amdgpu_ring *ring, uint32_t seq,
+					 ktime_t timestamp);
+
 /*
  * Rings.
  */
 
 /* provided by hw blocks that expose a ring buffer for commands */
 struct amdgpu_ring_funcs {
+	/**
+	 * @type:
+	 *
+	 * GFX, Compute, SDMA, UVD, VCE, VCN, VPE, KIQ, MES, UMSCH, and CPER
+	 * use ring buffers. The type field just identifies which component the
+	 * ring buffer is associated with.
+	 */
 	enum amdgpu_ring_type	type;
 	uint32_t		align_mask;
+
+	/**
+	 * @nop:
+	 *
+	 * Every block in the amdgpu has no-op instructions (e.g., GFX 10
+	 * uses PACKET3(PACKET3_NOP, 0x3FFF), VCN 5 uses VCN_ENC_CMD_NO_OP,
+	 * etc). This field receives the specific no-op for the component
+	 * that initializes the ring.
+	 */
 	u32			nop;
 	bool			support_64bit_ptrs;
-	unsigned		vmhub;
+	bool			no_user_fence;
+	bool			secure_submission_supported;
+
+	/**
+	 * @extra_bytes:
+	 *
+	 * Optional extra space in bytes that is added to the ring size
+	 * when allocating the BO that holds the contents of the ring.
+	 * This space isn't used for command submission to the ring,
+	 * but is just there to satisfy some hardware requirements or
+	 * implement workarounds. It's up to the implementation of each
+	 * specific ring to initialize this space.
+	 */
+	unsigned		extra_bytes;
 
 	/* ring read/write ptr handling */
 	u64 (*get_rptr)(struct amdgpu_ring *ring);
 	u64 (*get_wptr)(struct amdgpu_ring *ring);
 	void (*set_wptr)(struct amdgpu_ring *ring);
 	/* validating and patching of IBs */
-	int (*parse_cs)(struct amdgpu_cs_parser *p, uint32_t ib_idx);
+	int (*parse_cs)(struct amdgpu_cs_parser *p,
+			struct amdgpu_job *job,
+			struct amdgpu_ib *ib);
+	int (*patch_cs_in_place)(struct amdgpu_cs_parser *p,
+				 struct amdgpu_job *job,
+				 struct amdgpu_ib *ib);
 	/* constants to calculate how many DW are needed for an emit */
 	unsigned emit_frame_size;
 	unsigned emit_ib_size;
 	/* command emit functions */
 	void (*emit_ib)(struct amdgpu_ring *ring,
+			struct amdgpu_job *job,
 			struct amdgpu_ib *ib,
-			unsigned vm_id, bool ctx_switch);
+			uint32_t flags);
 	void (*emit_fence)(struct amdgpu_ring *ring, uint64_t addr,
 			   uint64_t seq, unsigned flags);
 	void (*emit_pipeline_sync)(struct amdgpu_ring *ring);
-	void (*emit_vm_flush)(struct amdgpu_ring *ring, unsigned vm_id,
+	void (*emit_vm_flush)(struct amdgpu_ring *ring, unsigned vmid,
 			      uint64_t pd_addr);
 	void (*emit_hdp_flush)(struct amdgpu_ring *ring);
-	void (*emit_hdp_invalidate)(struct amdgpu_ring *ring);
 	void (*emit_gds_switch)(struct amdgpu_ring *ring, uint32_t vmid,
 				uint32_t gds_base, uint32_t gds_size,
 				uint32_t gws_base, uint32_t gws_size,
@@ -137,85 +262,230 @@ struct amdgpu_ring_funcs {
 	void (*insert_end)(struct amdgpu_ring *ring);
 	/* pad the indirect buffer to the necessary number of dw */
 	void (*pad_ib)(struct amdgpu_ring *ring, struct amdgpu_ib *ib);
-	unsigned (*init_cond_exec)(struct amdgpu_ring *ring);
-	void (*patch_cond_exec)(struct amdgpu_ring *ring, unsigned offset);
+	unsigned (*init_cond_exec)(struct amdgpu_ring *ring, uint64_t addr);
 	/* note usage for clock and power gating */
 	void (*begin_use)(struct amdgpu_ring *ring);
 	void (*end_use)(struct amdgpu_ring *ring);
 	void (*emit_switch_buffer) (struct amdgpu_ring *ring);
 	void (*emit_cntxcntl) (struct amdgpu_ring *ring, uint32_t flags);
-	void (*emit_rreg)(struct amdgpu_ring *ring, uint32_t reg);
+	void (*emit_gfx_shadow)(struct amdgpu_ring *ring, u64 shadow_va, u64 csa_va,
+				u64 gds_va, bool init_shadow, int vmid);
+	void (*emit_rreg)(struct amdgpu_ring *ring, uint32_t reg,
+			  uint32_t reg_val_offs);
 	void (*emit_wreg)(struct amdgpu_ring *ring, uint32_t reg, uint32_t val);
-	void (*emit_tmz)(struct amdgpu_ring *ring, bool start);
+	void (*emit_reg_wait)(struct amdgpu_ring *ring, uint32_t reg,
+			      uint32_t val, uint32_t mask);
+	void (*emit_reg_write_reg_wait)(struct amdgpu_ring *ring,
+					uint32_t reg0, uint32_t reg1,
+					uint32_t ref, uint32_t mask);
+	void (*emit_frame_cntl)(struct amdgpu_ring *ring, bool start,
+				bool secure);
+	/* Try to soft recover the ring to make the fence signal */
+	void (*soft_recovery)(struct amdgpu_ring *ring, unsigned vmid);
+	int (*preempt_ib)(struct amdgpu_ring *ring);
+	void (*emit_mem_sync)(struct amdgpu_ring *ring);
+	void (*emit_wave_limit)(struct amdgpu_ring *ring, bool enable);
+	void (*patch_cntl)(struct amdgpu_ring *ring, unsigned offset);
+	void (*patch_ce)(struct amdgpu_ring *ring, unsigned offset);
+	void (*patch_de)(struct amdgpu_ring *ring, unsigned offset);
+	int (*reset)(struct amdgpu_ring *ring, unsigned int vmid,
+		     struct amdgpu_fence *timedout_fence);
+	void (*emit_cleaner_shader)(struct amdgpu_ring *ring);
 };
 
+/**
+ * amdgpu_ring - Holds ring information
+ */
 struct amdgpu_ring {
 	struct amdgpu_device		*adev;
 	const struct amdgpu_ring_funcs	*funcs;
 	struct amdgpu_fence_driver	fence_drv;
-	struct amd_gpu_scheduler	sched;
-	struct list_head		lru_list;
+	struct drm_gpu_scheduler	sched;
 
 	struct amdgpu_bo	*ring_obj;
-	volatile uint32_t	*ring;
+	uint32_t		*ring;
+	/* backups for resets */
+	uint32_t		*ring_backup;
+	unsigned int		ring_backup_entries_to_copy;
 	unsigned		rptr_offs;
+	u64			rptr_gpu_addr;
+	u32			*rptr_cpu_addr;
+
+	/**
+	 * @wptr:
+	 *
+	 * This is part of the Ring buffer implementation and represents the
+	 * write pointer. The wptr determines where the host has written.
+	 */
 	u64			wptr;
+
+	/**
+	 * @wptr_old:
+	 *
+	 * Before update wptr with the new value, usually the old value is
+	 * stored in the wptr_old.
+	 */
 	u64			wptr_old;
 	unsigned		ring_size;
+
+	/**
+	 * @max_dw:
+	 *
+	 * Maximum number of DWords for ring allocation. This information is
+	 * provided at the ring initialization time, and each IP block can
+	 * specify a specific value. Check places that invoke
+	 * amdgpu_ring_init() to see the maximum size per block.
+	 */
 	unsigned		max_dw;
+
+	/**
+	 * @count_dw:
+	 *
+	 * This value starts with the maximum amount of DWords supported by the
+	 * ring. This value is updated based on the ring manipulation.
+	 */
 	int			count_dw;
 	uint64_t		gpu_addr;
+
+	/**
+	 * @ptr_mask:
+	 *
+	 * Some IPs provide support for 64-bit pointers and others for 32-bit
+	 * only; this behavior is component-specific and defined by the field
+	 * support_64bit_ptr. If the IP block supports 64-bits, the mask
+	 * 0xffffffffffffffff is set; otherwise, this value assumes buf_mask.
+	 * Notice that this field is used to keep wptr under a valid range.
+	 */
 	uint64_t		ptr_mask;
+
+	/**
+	 * @buf_mask:
+	 *
+	 * Buffer mask is a value used to keep wptr count under its
+	 * thresholding. Buffer mask initialized during the ring buffer
+	 * initialization time, and it is defined as (ring_size / 4) -1.
+	 */
 	uint32_t		buf_mask;
-	bool			ready;
 	u32			idx;
+	u32			xcc_id;
+	u32			xcp_id;
 	u32			me;
 	u32			pipe;
 	u32			queue;
 	struct amdgpu_bo	*mqd_obj;
 	uint64_t                mqd_gpu_addr;
 	void                    *mqd_ptr;
+	unsigned                mqd_size;
 	uint64_t                eop_gpu_addr;
 	u32			doorbell_index;
 	bool			use_doorbell;
+	bool			use_pollmem;
 	unsigned		wptr_offs;
+	u64			wptr_gpu_addr;
+
+	/**
+	 * @wptr_cpu_addr:
+	 *
+	 * This is the CPU address pointer in the writeback slot. This is used
+	 * to commit changes to the GPU.
+	 */
+	u32			*wptr_cpu_addr;
 	unsigned		fence_offs;
+	u64			fence_gpu_addr;
+	u32			*fence_cpu_addr;
 	uint64_t		current_ctx;
 	char			name[16];
+	u32                     trail_seq;
+	unsigned		trail_fence_offs;
+	u64			trail_fence_gpu_addr;
+	u32			*trail_fence_cpu_addr;
 	unsigned		cond_exe_offs;
 	u64			cond_exe_gpu_addr;
-	volatile u32		*cond_exe_cpu_addr;
+	u32			*cond_exe_cpu_addr;
+	unsigned int		set_q_mode_offs;
+	u32			*set_q_mode_ptr;
+	u64			set_q_mode_token;
+	unsigned		vm_hub;
 	unsigned		vm_inv_eng;
+	struct dma_fence	*vmid_wait;
 	bool			has_compute_vm_bug;
-#if defined(CONFIG_DEBUG_FS)
-	struct dentry *ent;
-#endif
+	bool			no_scheduler;
+	bool			no_user_submission;
+	int			hw_prio;
+	unsigned 		num_hw_submission;
+	atomic_t		*sched_score;
+
+	bool            is_sw_ring;
+	unsigned int    entry_index;
+	/* store the cached rptr to restore after reset */
+	uint64_t cached_rptr;
 };
 
+#define amdgpu_ring_parse_cs(r, p, job, ib) ((r)->funcs->parse_cs((p), (job), (ib)))
+#define amdgpu_ring_patch_cs_in_place(r, p, job, ib) ((r)->funcs->patch_cs_in_place((p), (job), (ib)))
+#define amdgpu_ring_test_ring(r) (r)->funcs->test_ring((r))
+#define amdgpu_ring_test_ib(r, t) ((r)->funcs->test_ib ? (r)->funcs->test_ib((r), (t)) : 0)
+#define amdgpu_ring_get_rptr(r) (r)->funcs->get_rptr((r))
+#define amdgpu_ring_get_wptr(r) (r)->funcs->get_wptr((r))
+#define amdgpu_ring_set_wptr(r) (r)->funcs->set_wptr((r))
+#define amdgpu_ring_emit_ib(r, job, ib, flags) ((r)->funcs->emit_ib((r), (job), (ib), (flags)))
+#define amdgpu_ring_emit_pipeline_sync(r) (r)->funcs->emit_pipeline_sync((r))
+#define amdgpu_ring_emit_vm_flush(r, vmid, addr) (r)->funcs->emit_vm_flush((r), (vmid), (addr))
+#define amdgpu_ring_emit_fence(r, addr, seq, flags) (r)->funcs->emit_fence((r), (addr), (seq), (flags))
+#define amdgpu_ring_emit_gds_switch(r, v, db, ds, wb, ws, ab, as) (r)->funcs->emit_gds_switch((r), (v), (db), (ds), (wb), (ws), (ab), (as))
+#define amdgpu_ring_emit_hdp_flush(r) (r)->funcs->emit_hdp_flush((r))
+#define amdgpu_ring_emit_switch_buffer(r) (r)->funcs->emit_switch_buffer((r))
+#define amdgpu_ring_emit_cntxcntl(r, d) (r)->funcs->emit_cntxcntl((r), (d))
+#define amdgpu_ring_emit_gfx_shadow(r, s, c, g, i, v) ((r)->funcs->emit_gfx_shadow((r), (s), (c), (g), (i), (v)))
+#define amdgpu_ring_emit_rreg(r, d, o) (r)->funcs->emit_rreg((r), (d), (o))
+#define amdgpu_ring_emit_wreg(r, d, v) (r)->funcs->emit_wreg((r), (d), (v))
+#define amdgpu_ring_emit_reg_wait(r, d, v, m) (r)->funcs->emit_reg_wait((r), (d), (v), (m))
+#define amdgpu_ring_emit_reg_write_reg_wait(r, d0, d1, v, m) (r)->funcs->emit_reg_write_reg_wait((r), (d0), (d1), (v), (m))
+#define amdgpu_ring_emit_frame_cntl(r, b, s) (r)->funcs->emit_frame_cntl((r), (b), (s))
+#define amdgpu_ring_pad_ib(r, ib) ((r)->funcs->pad_ib((r), (ib)))
+#define amdgpu_ring_init_cond_exec(r, a) (r)->funcs->init_cond_exec((r), (a))
+#define amdgpu_ring_preempt_ib(r) (r)->funcs->preempt_ib(r)
+#define amdgpu_ring_patch_cntl(r, o) ((r)->funcs->patch_cntl((r), (o)))
+#define amdgpu_ring_patch_ce(r, o) ((r)->funcs->patch_ce((r), (o)))
+#define amdgpu_ring_patch_de(r, o) ((r)->funcs->patch_de((r), (o)))
+#define amdgpu_ring_reset(r, v, f) (r)->funcs->reset((r), (v), (f))
+
+unsigned int amdgpu_ring_max_ibs(enum amdgpu_ring_type type);
 int amdgpu_ring_alloc(struct amdgpu_ring *ring, unsigned ndw);
+void amdgpu_ring_ib_begin(struct amdgpu_ring *ring);
+void amdgpu_ring_ib_end(struct amdgpu_ring *ring);
+void amdgpu_ring_ib_on_emit_cntl(struct amdgpu_ring *ring);
+void amdgpu_ring_ib_on_emit_ce(struct amdgpu_ring *ring);
+void amdgpu_ring_ib_on_emit_de(struct amdgpu_ring *ring);
+
 void amdgpu_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count);
 void amdgpu_ring_generic_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib);
 void amdgpu_ring_commit(struct amdgpu_ring *ring);
 void amdgpu_ring_undo(struct amdgpu_ring *ring);
 int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
-		     unsigned ring_size, struct amdgpu_irq_src *irq_src,
-		     unsigned irq_type);
+		     unsigned int max_dw, struct amdgpu_irq_src *irq_src,
+		     unsigned int irq_type, unsigned int hw_prio,
+		     atomic_t *sched_score);
 void amdgpu_ring_fini(struct amdgpu_ring *ring);
-int amdgpu_ring_lru_get(struct amdgpu_device *adev, int type, int *blacklist,
-			int num_blacklist, struct amdgpu_ring **ring);
-void amdgpu_ring_lru_touch(struct amdgpu_device *adev, struct amdgpu_ring *ring);
-static inline void amdgpu_ring_clear_ring(struct amdgpu_ring *ring)
+void amdgpu_ring_emit_reg_write_reg_wait_helper(struct amdgpu_ring *ring,
+						uint32_t reg0, uint32_t val0,
+						uint32_t reg1, uint32_t val1);
+bool amdgpu_ring_soft_recovery(struct amdgpu_ring *ring, unsigned int vmid,
+			       struct dma_fence *fence);
+
+static inline void amdgpu_ring_set_preempt_cond_exec(struct amdgpu_ring *ring,
+							bool cond_exec)
 {
-	int i = 0;
-	while (i <= ring->buf_mask)
-		ring->ring[i++] = ring->funcs->nop;
+	*ring->cond_exe_cpu_addr = cond_exec;
+}
 
+static inline void amdgpu_ring_clear_ring(struct amdgpu_ring *ring)
+{
+	memset32(ring->ring, ring->funcs->nop, ring->buf_mask + 1);
 }
 
 static inline void amdgpu_ring_write(struct amdgpu_ring *ring, uint32_t v)
 {
-	if (ring->count_dw <= 0)
-		DRM_ERROR("amdgpu: writing more dwords to the ring than expected!\n");
 	ring->ring[ring->wptr++ & ring->buf_mask] = v;
 	ring->wptr &= ring->ptr_mask;
 	ring->count_dw--;
@@ -225,26 +495,20 @@ static inline void amdgpu_ring_write_multiple(struct amdgpu_ring *ring,
 					      void *src, int count_dw)
 {
 	unsigned occupied, chunk1, chunk2;
-	void *dst;
-
-	if (unlikely(ring->count_dw < count_dw))
-		DRM_ERROR("amdgpu: writing more dwords to the ring than expected!\n");
 
 	occupied = ring->wptr & ring->buf_mask;
-	dst = (void *)&ring->ring[occupied];
 	chunk1 = ring->buf_mask + 1 - occupied;
-	chunk1 = (chunk1 >= count_dw) ? count_dw: chunk1;
+	chunk1 = (chunk1 >= count_dw) ? count_dw : chunk1;
 	chunk2 = count_dw - chunk1;
 	chunk1 <<= 2;
 	chunk2 <<= 2;
 
 	if (chunk1)
-		memcpy(dst, src, chunk1);
+		memcpy(&ring->ring[occupied], src, chunk1);
 
 	if (chunk2) {
 		src += chunk1;
-		dst = (void *)ring->ring;
-		memcpy(dst, src, chunk2);
+		memcpy(ring->ring, src, chunk2);
 	}
 
 	ring->wptr += count_dw;
@@ -252,4 +516,66 @@ static inline void amdgpu_ring_write_multiple(struct amdgpu_ring *ring,
 	ring->count_dw -= count_dw;
 }
 
+/**
+ * amdgpu_ring_patch_cond_exec - patch dw count of conditional execute
+ * @ring: amdgpu_ring structure
+ * @offset: offset returned by amdgpu_ring_init_cond_exec
+ *
+ * Calculate the dw count and patch it into a cond_exec command.
+ */
+static inline void amdgpu_ring_patch_cond_exec(struct amdgpu_ring *ring,
+					       unsigned int offset)
+{
+	unsigned cur;
+
+	if (!ring->funcs->init_cond_exec)
+		return;
+
+	WARN_ON(offset > ring->buf_mask);
+	WARN_ON(ring->ring[offset] != 0);
+
+	cur = (ring->wptr - 1) & ring->buf_mask;
+	if (cur < offset)
+		cur += ring->ring_size >> 2;
+	ring->ring[offset] = cur - offset;
+}
+
+int amdgpu_ring_test_helper(struct amdgpu_ring *ring);
+
+void amdgpu_debugfs_ring_init(struct amdgpu_device *adev,
+			      struct amdgpu_ring *ring);
+
+int amdgpu_ring_init_mqd(struct amdgpu_ring *ring);
+
+static inline u32 amdgpu_ib_get_value(struct amdgpu_ib *ib, int idx)
+{
+	return ib->ptr[idx];
+}
+
+static inline void amdgpu_ib_set_value(struct amdgpu_ib *ib, int idx,
+				       uint32_t value)
+{
+	ib->ptr[idx] = value;
+}
+
+int amdgpu_ib_get(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+		  unsigned size,
+		  enum amdgpu_ib_pool_type pool,
+		  struct amdgpu_ib *ib);
+void amdgpu_ib_free(struct amdgpu_ib *ib, struct dma_fence *f);
+int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
+		       struct amdgpu_ib *ibs, struct amdgpu_job *job,
+		       struct dma_fence **f);
+int amdgpu_ib_pool_init(struct amdgpu_device *adev);
+void amdgpu_ib_pool_fini(struct amdgpu_device *adev);
+int amdgpu_ib_ring_tests(struct amdgpu_device *adev);
+bool amdgpu_ring_sched_ready(struct amdgpu_ring *ring);
+void amdgpu_ring_backup_unprocessed_commands(struct amdgpu_ring *ring,
+					     struct amdgpu_fence *guilty_fence);
+void amdgpu_ring_reset_helper_begin(struct amdgpu_ring *ring,
+				    struct amdgpu_fence *guilty_fence);
+int amdgpu_ring_reset_helper_end(struct amdgpu_ring *ring,
+				 struct amdgpu_fence *guilty_fence);
+bool amdgpu_ring_is_reset_type_supported(struct amdgpu_ring *ring,
+					 u32 reset_type);
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring_mux.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring_mux.c
new file mode 100644
index 000000000000..7e7d6c3865bc
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring_mux.c
@@ -0,0 +1,576 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/slab.h>
+#include <drm/drm_print.h>
+
+#include "amdgpu_ring_mux.h"
+#include "amdgpu_ring.h"
+#include "amdgpu.h"
+
+#define AMDGPU_MUX_RESUBMIT_JIFFIES_TIMEOUT (HZ / 2)
+#define AMDGPU_MAX_LAST_UNSIGNALED_THRESHOLD_US 10000
+
+static const struct ring_info {
+	unsigned int hw_pio;
+	const char *ring_name;
+} sw_ring_info[] = {
+	{ AMDGPU_RING_PRIO_DEFAULT, "gfx_low"},
+	{ AMDGPU_RING_PRIO_2, "gfx_high"},
+};
+
+static struct kmem_cache *amdgpu_mux_chunk_slab;
+
+static inline struct amdgpu_mux_entry *amdgpu_ring_mux_sw_entry(struct amdgpu_ring_mux *mux,
+								struct amdgpu_ring *ring)
+{
+	return ring->entry_index < mux->ring_entry_size ?
+			&mux->ring_entry[ring->entry_index] : NULL;
+}
+
+/* copy packages on sw ring range[begin, end) */
+static void amdgpu_ring_mux_copy_pkt_from_sw_ring(struct amdgpu_ring_mux *mux,
+						  struct amdgpu_ring *ring,
+						  u64 s_start, u64 s_end)
+{
+	u64 start, end;
+	struct amdgpu_ring *real_ring = mux->real_ring;
+
+	start = s_start & ring->buf_mask;
+	end = s_end & ring->buf_mask;
+
+	if (start == end) {
+		DRM_ERROR("no more data copied from sw ring\n");
+		return;
+	}
+	if (start > end) {
+		amdgpu_ring_alloc(real_ring, (ring->ring_size >> 2) + end - start);
+		amdgpu_ring_write_multiple(real_ring, (void *)&ring->ring[start],
+					   (ring->ring_size >> 2) - start);
+		amdgpu_ring_write_multiple(real_ring, (void *)&ring->ring[0], end);
+	} else {
+		amdgpu_ring_alloc(real_ring, end - start);
+		amdgpu_ring_write_multiple(real_ring, (void *)&ring->ring[start], end - start);
+	}
+}
+
+static void amdgpu_mux_resubmit_chunks(struct amdgpu_ring_mux *mux)
+{
+	struct amdgpu_mux_entry *e = NULL;
+	struct amdgpu_mux_chunk *chunk;
+	uint32_t seq, last_seq;
+	int i;
+
+	/*find low priority entries:*/
+	if (!mux->s_resubmit)
+		return;
+
+	for (i = 0; i < mux->num_ring_entries; i++) {
+		if (mux->ring_entry[i].ring->hw_prio <= AMDGPU_RING_PRIO_DEFAULT) {
+			e = &mux->ring_entry[i];
+			break;
+		}
+	}
+
+	if (!e) {
+		DRM_ERROR("%s no low priority ring found\n", __func__);
+		return;
+	}
+
+	last_seq = atomic_read(&e->ring->fence_drv.last_seq);
+	seq = mux->seqno_to_resubmit;
+	if (last_seq < seq) {
+		/*resubmit all the fences between (last_seq, seq]*/
+		list_for_each_entry(chunk, &e->list, entry) {
+			if (chunk->sync_seq > last_seq && chunk->sync_seq <= seq) {
+				amdgpu_fence_update_start_timestamp(e->ring,
+								    chunk->sync_seq,
+								    ktime_get());
+				if (chunk->sync_seq ==
+					le32_to_cpu(*(e->ring->fence_drv.cpu_addr + 2))) {
+					if (chunk->cntl_offset <= e->ring->buf_mask)
+						amdgpu_ring_patch_cntl(e->ring,
+								       chunk->cntl_offset);
+					if (chunk->ce_offset <= e->ring->buf_mask)
+						amdgpu_ring_patch_ce(e->ring, chunk->ce_offset);
+					if (chunk->de_offset <= e->ring->buf_mask)
+						amdgpu_ring_patch_de(e->ring, chunk->de_offset);
+				}
+				amdgpu_ring_mux_copy_pkt_from_sw_ring(mux, e->ring,
+								      chunk->start,
+								      chunk->end);
+				mux->wptr_resubmit = chunk->end;
+				amdgpu_ring_commit(mux->real_ring);
+			}
+		}
+	}
+
+	timer_delete(&mux->resubmit_timer);
+	mux->s_resubmit = false;
+}
+
+static void amdgpu_ring_mux_schedule_resubmit(struct amdgpu_ring_mux *mux)
+{
+	mod_timer(&mux->resubmit_timer, jiffies + AMDGPU_MUX_RESUBMIT_JIFFIES_TIMEOUT);
+}
+
+static void amdgpu_mux_resubmit_fallback(struct timer_list *t)
+{
+	struct amdgpu_ring_mux *mux = timer_container_of(mux, t,
+							 resubmit_timer);
+
+	if (!spin_trylock(&mux->lock)) {
+		amdgpu_ring_mux_schedule_resubmit(mux);
+		DRM_ERROR("reschedule resubmit\n");
+		return;
+	}
+	amdgpu_mux_resubmit_chunks(mux);
+	spin_unlock(&mux->lock);
+}
+
+int amdgpu_ring_mux_init(struct amdgpu_ring_mux *mux, struct amdgpu_ring *ring,
+			 unsigned int entry_size)
+{
+	mux->real_ring = ring;
+	mux->num_ring_entries = 0;
+
+	mux->ring_entry = kcalloc(entry_size, sizeof(struct amdgpu_mux_entry), GFP_KERNEL);
+	if (!mux->ring_entry)
+		return -ENOMEM;
+
+	mux->ring_entry_size = entry_size;
+	mux->s_resubmit = false;
+
+	amdgpu_mux_chunk_slab = KMEM_CACHE(amdgpu_mux_chunk, SLAB_HWCACHE_ALIGN);
+	if (!amdgpu_mux_chunk_slab) {
+		DRM_ERROR("create amdgpu_mux_chunk cache failed\n");
+		return -ENOMEM;
+	}
+
+	spin_lock_init(&mux->lock);
+	timer_setup(&mux->resubmit_timer, amdgpu_mux_resubmit_fallback, 0);
+
+	return 0;
+}
+
+void amdgpu_ring_mux_fini(struct amdgpu_ring_mux *mux)
+{
+	struct amdgpu_mux_entry *e;
+	struct amdgpu_mux_chunk *chunk, *chunk2;
+	int i;
+
+	for (i = 0; i < mux->num_ring_entries; i++) {
+		e = &mux->ring_entry[i];
+		list_for_each_entry_safe(chunk, chunk2, &e->list, entry) {
+			list_del(&chunk->entry);
+			kmem_cache_free(amdgpu_mux_chunk_slab, chunk);
+		}
+	}
+	kmem_cache_destroy(amdgpu_mux_chunk_slab);
+	kfree(mux->ring_entry);
+	mux->ring_entry = NULL;
+	mux->num_ring_entries = 0;
+	mux->ring_entry_size = 0;
+}
+
+int amdgpu_ring_mux_add_sw_ring(struct amdgpu_ring_mux *mux, struct amdgpu_ring *ring)
+{
+	struct amdgpu_mux_entry *e;
+
+	if (mux->num_ring_entries >= mux->ring_entry_size) {
+		DRM_ERROR("add sw ring exceeding max entry size\n");
+		return -ENOENT;
+	}
+
+	e = &mux->ring_entry[mux->num_ring_entries];
+	ring->entry_index = mux->num_ring_entries;
+	e->ring = ring;
+
+	INIT_LIST_HEAD(&e->list);
+	mux->num_ring_entries += 1;
+	return 0;
+}
+
+void amdgpu_ring_mux_set_wptr(struct amdgpu_ring_mux *mux, struct amdgpu_ring *ring, u64 wptr)
+{
+	struct amdgpu_mux_entry *e;
+
+	spin_lock(&mux->lock);
+
+	if (ring->hw_prio <= AMDGPU_RING_PRIO_DEFAULT)
+		amdgpu_mux_resubmit_chunks(mux);
+
+	e = amdgpu_ring_mux_sw_entry(mux, ring);
+	if (!e) {
+		DRM_ERROR("cannot find entry for sw ring\n");
+		spin_unlock(&mux->lock);
+		return;
+	}
+
+	/* We could skip this set wptr as preemption in process. */
+	if (ring->hw_prio <= AMDGPU_RING_PRIO_DEFAULT && mux->pending_trailing_fence_signaled) {
+		spin_unlock(&mux->lock);
+		return;
+	}
+
+	e->sw_cptr = e->sw_wptr;
+	/* Update cptr if the package already copied in resubmit functions */
+	if (ring->hw_prio <= AMDGPU_RING_PRIO_DEFAULT && e->sw_cptr < mux->wptr_resubmit)
+		e->sw_cptr = mux->wptr_resubmit;
+	e->sw_wptr = wptr;
+	e->start_ptr_in_hw_ring = mux->real_ring->wptr;
+
+	/* Skip copying for the packages already resubmitted.*/
+	if (ring->hw_prio > AMDGPU_RING_PRIO_DEFAULT || mux->wptr_resubmit < wptr) {
+		amdgpu_ring_mux_copy_pkt_from_sw_ring(mux, ring, e->sw_cptr, wptr);
+		e->end_ptr_in_hw_ring = mux->real_ring->wptr;
+		amdgpu_ring_commit(mux->real_ring);
+	} else {
+		e->end_ptr_in_hw_ring = mux->real_ring->wptr;
+	}
+	spin_unlock(&mux->lock);
+}
+
+u64 amdgpu_ring_mux_get_wptr(struct amdgpu_ring_mux *mux, struct amdgpu_ring *ring)
+{
+	struct amdgpu_mux_entry *e;
+
+	e = amdgpu_ring_mux_sw_entry(mux, ring);
+	if (!e) {
+		DRM_ERROR("cannot find entry for sw ring\n");
+		return 0;
+	}
+
+	return e->sw_wptr;
+}
+
+/**
+ * amdgpu_ring_mux_get_rptr - get the readptr of the software ring
+ * @mux: the multiplexer the software rings attach to
+ * @ring: the software ring of which we calculate the readptr
+ *
+ * The return value of the readptr is not precise while the other rings could
+ * write data onto the real ring buffer.After overwriting on the real ring, we
+ * can not decide if our packages have been excuted or not read yet. However,
+ * this function is only called by the tools such as umr to collect the latest
+ * packages for the hang analysis. We assume the hang happens near our latest
+ * submit. Thus we could use the following logic to give the clue:
+ * If the readptr is between start and end, then we return the copy pointer
+ * plus the distance from start to readptr. If the readptr is before start, we
+ * return the copy pointer. Lastly, if the readptr is past end, we return the
+ * write pointer.
+ */
+u64 amdgpu_ring_mux_get_rptr(struct amdgpu_ring_mux *mux, struct amdgpu_ring *ring)
+{
+	struct amdgpu_mux_entry *e;
+	u64 readp, offset, start, end;
+
+	e = amdgpu_ring_mux_sw_entry(mux, ring);
+	if (!e) {
+		DRM_ERROR("no sw entry found!\n");
+		return 0;
+	}
+
+	readp = amdgpu_ring_get_rptr(mux->real_ring);
+
+	start = e->start_ptr_in_hw_ring & mux->real_ring->buf_mask;
+	end = e->end_ptr_in_hw_ring & mux->real_ring->buf_mask;
+	if (start > end) {
+		if (readp <= end)
+			readp += mux->real_ring->ring_size >> 2;
+		end += mux->real_ring->ring_size >> 2;
+	}
+
+	if (start <= readp && readp <= end) {
+		offset = readp - start;
+		e->sw_rptr = (e->sw_cptr + offset) & ring->buf_mask;
+	} else if (readp < start) {
+		e->sw_rptr = e->sw_cptr;
+	} else {
+		/* end < readptr */
+		e->sw_rptr = e->sw_wptr;
+	}
+
+	return e->sw_rptr;
+}
+
+u64 amdgpu_sw_ring_get_rptr_gfx(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ring_mux *mux = &adev->gfx.muxer;
+
+	WARN_ON(!ring->is_sw_ring);
+	return amdgpu_ring_mux_get_rptr(mux, ring);
+}
+
+u64 amdgpu_sw_ring_get_wptr_gfx(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ring_mux *mux = &adev->gfx.muxer;
+
+	WARN_ON(!ring->is_sw_ring);
+	return amdgpu_ring_mux_get_wptr(mux, ring);
+}
+
+void amdgpu_sw_ring_set_wptr_gfx(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ring_mux *mux = &adev->gfx.muxer;
+
+	WARN_ON(!ring->is_sw_ring);
+	amdgpu_ring_mux_set_wptr(mux, ring, ring->wptr);
+}
+
+/* Override insert_nop to prevent emitting nops to the software rings */
+void amdgpu_sw_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	WARN_ON(!ring->is_sw_ring);
+}
+
+const char *amdgpu_sw_ring_name(int idx)
+{
+	return idx < ARRAY_SIZE(sw_ring_info) ?
+		sw_ring_info[idx].ring_name : NULL;
+}
+
+unsigned int amdgpu_sw_ring_priority(int idx)
+{
+	return idx < ARRAY_SIZE(sw_ring_info) ?
+		sw_ring_info[idx].hw_pio : AMDGPU_RING_PRIO_DEFAULT;
+}
+
+/*Scan on low prio rings to have unsignaled fence and high ring has no fence.*/
+static int amdgpu_mcbp_scan(struct amdgpu_ring_mux *mux)
+{
+	struct amdgpu_ring *ring;
+	int i, need_preempt;
+
+	need_preempt = 0;
+	for (i = 0; i < mux->num_ring_entries; i++) {
+		ring = mux->ring_entry[i].ring;
+		if (ring->hw_prio > AMDGPU_RING_PRIO_DEFAULT &&
+		    amdgpu_fence_count_emitted(ring) > 0)
+			return 0;
+		if (ring->hw_prio <= AMDGPU_RING_PRIO_DEFAULT &&
+		    amdgpu_fence_last_unsignaled_time_us(ring) >
+		    AMDGPU_MAX_LAST_UNSIGNALED_THRESHOLD_US)
+			need_preempt = 1;
+	}
+	return need_preempt && !mux->s_resubmit;
+}
+
+/* Trigger Mid-Command Buffer Preemption (MCBP) and find if we need to resubmit. */
+static int amdgpu_mcbp_trigger_preempt(struct amdgpu_ring_mux *mux)
+{
+	int r;
+
+	spin_lock(&mux->lock);
+	mux->pending_trailing_fence_signaled = true;
+	r = amdgpu_ring_preempt_ib(mux->real_ring);
+	spin_unlock(&mux->lock);
+	return r;
+}
+
+void amdgpu_sw_ring_ib_begin(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ring_mux *mux = &adev->gfx.muxer;
+
+	WARN_ON(!ring->is_sw_ring);
+	if (adev->gfx.mcbp && ring->hw_prio > AMDGPU_RING_PRIO_DEFAULT) {
+		if (amdgpu_mcbp_scan(mux) > 0)
+			amdgpu_mcbp_trigger_preempt(mux);
+		return;
+	}
+
+	amdgpu_ring_mux_start_ib(mux, ring);
+}
+
+void amdgpu_sw_ring_ib_end(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ring_mux *mux = &adev->gfx.muxer;
+
+	WARN_ON(!ring->is_sw_ring);
+	if (adev->gfx.mcbp && ring->hw_prio > AMDGPU_RING_PRIO_DEFAULT)
+		return;
+	amdgpu_ring_mux_end_ib(mux, ring);
+}
+
+void amdgpu_sw_ring_ib_mark_offset(struct amdgpu_ring *ring, enum amdgpu_ring_mux_offset_type type)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ring_mux *mux = &adev->gfx.muxer;
+	unsigned offset;
+
+	if (ring->hw_prio > AMDGPU_RING_PRIO_DEFAULT)
+		return;
+
+	offset = ring->wptr & ring->buf_mask;
+
+	amdgpu_ring_mux_ib_mark_offset(mux, ring, offset, type);
+}
+
+void amdgpu_ring_mux_start_ib(struct amdgpu_ring_mux *mux, struct amdgpu_ring *ring)
+{
+	struct amdgpu_mux_entry *e;
+	struct amdgpu_mux_chunk *chunk;
+
+	spin_lock(&mux->lock);
+	amdgpu_mux_resubmit_chunks(mux);
+	spin_unlock(&mux->lock);
+
+	e = amdgpu_ring_mux_sw_entry(mux, ring);
+	if (!e) {
+		DRM_ERROR("cannot find entry!\n");
+		return;
+	}
+
+	chunk = kmem_cache_alloc(amdgpu_mux_chunk_slab, GFP_KERNEL);
+	if (!chunk) {
+		DRM_ERROR("alloc amdgpu_mux_chunk_slab failed\n");
+		return;
+	}
+
+	chunk->start = ring->wptr;
+	/* the initialized value used to check if they are set by the ib submission*/
+	chunk->cntl_offset = ring->buf_mask + 1;
+	chunk->de_offset = ring->buf_mask + 1;
+	chunk->ce_offset = ring->buf_mask + 1;
+	list_add_tail(&chunk->entry, &e->list);
+}
+
+static void scan_and_remove_signaled_chunk(struct amdgpu_ring_mux *mux, struct amdgpu_ring *ring)
+{
+	uint32_t last_seq = 0;
+	struct amdgpu_mux_entry *e;
+	struct amdgpu_mux_chunk *chunk, *tmp;
+
+	e = amdgpu_ring_mux_sw_entry(mux, ring);
+	if (!e) {
+		DRM_ERROR("cannot find entry!\n");
+		return;
+	}
+
+	last_seq = atomic_read(&ring->fence_drv.last_seq);
+
+	list_for_each_entry_safe(chunk, tmp, &e->list, entry) {
+		if (chunk->sync_seq <= last_seq) {
+			list_del(&chunk->entry);
+			kmem_cache_free(amdgpu_mux_chunk_slab, chunk);
+		}
+	}
+}
+
+void amdgpu_ring_mux_ib_mark_offset(struct amdgpu_ring_mux *mux,
+				    struct amdgpu_ring *ring, u64 offset,
+				    enum amdgpu_ring_mux_offset_type type)
+{
+	struct amdgpu_mux_entry *e;
+	struct amdgpu_mux_chunk *chunk;
+
+	e = amdgpu_ring_mux_sw_entry(mux, ring);
+	if (!e) {
+		DRM_ERROR("cannot find entry!\n");
+		return;
+	}
+
+	chunk = list_last_entry(&e->list, struct amdgpu_mux_chunk, entry);
+	if (!chunk) {
+		DRM_ERROR("cannot find chunk!\n");
+		return;
+	}
+
+	switch (type) {
+	case AMDGPU_MUX_OFFSET_TYPE_CONTROL:
+		chunk->cntl_offset = offset;
+		break;
+	case AMDGPU_MUX_OFFSET_TYPE_DE:
+		chunk->de_offset = offset;
+		break;
+	case AMDGPU_MUX_OFFSET_TYPE_CE:
+		chunk->ce_offset = offset;
+		break;
+	default:
+		DRM_ERROR("invalid type (%d)\n", type);
+		break;
+	}
+}
+
+void amdgpu_ring_mux_end_ib(struct amdgpu_ring_mux *mux, struct amdgpu_ring *ring)
+{
+	struct amdgpu_mux_entry *e;
+	struct amdgpu_mux_chunk *chunk;
+
+	e = amdgpu_ring_mux_sw_entry(mux, ring);
+	if (!e) {
+		DRM_ERROR("cannot find entry!\n");
+		return;
+	}
+
+	chunk = list_last_entry(&e->list, struct amdgpu_mux_chunk, entry);
+	if (!chunk) {
+		DRM_ERROR("cannot find chunk!\n");
+		return;
+	}
+
+	chunk->end = ring->wptr;
+	chunk->sync_seq = READ_ONCE(ring->fence_drv.sync_seq);
+
+	scan_and_remove_signaled_chunk(mux, ring);
+}
+
+bool amdgpu_mcbp_handle_trailing_fence_irq(struct amdgpu_ring_mux *mux)
+{
+	struct amdgpu_mux_entry *e;
+	struct amdgpu_ring *ring = NULL;
+	int i;
+
+	if (!mux->pending_trailing_fence_signaled)
+		return false;
+
+	if (mux->real_ring->trail_seq != le32_to_cpu(*mux->real_ring->trail_fence_cpu_addr))
+		return false;
+
+	for (i = 0; i < mux->num_ring_entries; i++) {
+		e = &mux->ring_entry[i];
+		if (e->ring->hw_prio <= AMDGPU_RING_PRIO_DEFAULT) {
+			ring = e->ring;
+			break;
+		}
+	}
+
+	if (!ring) {
+		DRM_ERROR("cannot find low priority ring\n");
+		return false;
+	}
+
+	amdgpu_fence_process(ring);
+	if (amdgpu_fence_count_emitted(ring) > 0) {
+		mux->s_resubmit = true;
+		mux->seqno_to_resubmit = ring->fence_drv.sync_seq;
+		amdgpu_ring_mux_schedule_resubmit(mux);
+	}
+
+	mux->pending_trailing_fence_signaled = false;
+	return true;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring_mux.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring_mux.h
new file mode 100644
index 000000000000..d3186b570b82
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring_mux.h
@@ -0,0 +1,127 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_RING_MUX__
+#define __AMDGPU_RING_MUX__
+
+#include <linux/timer.h>
+#include <linux/spinlock.h>
+#include "amdgpu_ring.h"
+
+struct amdgpu_ring;
+
+/**
+ * struct amdgpu_mux_entry - the entry recording software rings copying information.
+ * @ring: the pointer to the software ring.
+ * @start_ptr_in_hw_ring: last start location copied to in the hardware ring.
+ * @end_ptr_in_hw_ring: last end location copied to in the hardware ring.
+ * @sw_cptr: the position of the copy pointer in the sw ring.
+ * @sw_rptr: the read pointer in software ring.
+ * @sw_wptr: the write pointer in software ring.
+ * @list: list head for amdgpu_mux_chunk
+ */
+struct amdgpu_mux_entry {
+	struct amdgpu_ring      *ring;
+	u64                     start_ptr_in_hw_ring;
+	u64                     end_ptr_in_hw_ring;
+	u64                     sw_cptr;
+	u64                     sw_rptr;
+	u64                     sw_wptr;
+	struct list_head        list;
+};
+
+enum amdgpu_ring_mux_offset_type {
+	AMDGPU_MUX_OFFSET_TYPE_CONTROL,
+	AMDGPU_MUX_OFFSET_TYPE_DE,
+	AMDGPU_MUX_OFFSET_TYPE_CE,
+};
+
+enum ib_complete_status {
+	/* IB not started/reset value, default value. */
+	IB_COMPLETION_STATUS_DEFAULT = 0,
+	/* IB preempted, started but not completed. */
+	IB_COMPLETION_STATUS_PREEMPTED = 1,
+	/* IB completed. */
+	IB_COMPLETION_STATUS_COMPLETED = 2,
+};
+
+struct amdgpu_ring_mux {
+	struct amdgpu_ring      *real_ring;
+
+	struct amdgpu_mux_entry *ring_entry;
+	unsigned int            num_ring_entries;
+	unsigned int            ring_entry_size;
+	/*the lock for copy data from different software rings*/
+	spinlock_t              lock;
+	bool                    s_resubmit;
+	uint32_t                seqno_to_resubmit;
+	u64                     wptr_resubmit;
+	struct timer_list       resubmit_timer;
+
+	bool                    pending_trailing_fence_signaled;
+};
+
+/**
+ * struct amdgpu_mux_chunk - save the location of indirect buffer's package on softare rings.
+ * @entry: the list entry.
+ * @sync_seq: the fence seqno related with the saved IB.
+ * @start:- start location on the software ring.
+ * @end:- end location on the software ring.
+ * @control_offset:- the PRE_RESUME bit position used for resubmission.
+ * @de_offset:- the anchor in write_data for de meta of resubmission.
+ * @ce_offset:- the anchor in write_data for ce meta of resubmission.
+ */
+struct amdgpu_mux_chunk {
+	struct list_head        entry;
+	uint32_t                sync_seq;
+	u64                     start;
+	u64                     end;
+	u64                     cntl_offset;
+	u64                     de_offset;
+	u64                     ce_offset;
+};
+
+int amdgpu_ring_mux_init(struct amdgpu_ring_mux *mux, struct amdgpu_ring *ring,
+			 unsigned int entry_size);
+void amdgpu_ring_mux_fini(struct amdgpu_ring_mux *mux);
+int amdgpu_ring_mux_add_sw_ring(struct amdgpu_ring_mux *mux, struct amdgpu_ring *ring);
+void amdgpu_ring_mux_set_wptr(struct amdgpu_ring_mux *mux, struct amdgpu_ring *ring, u64 wptr);
+u64 amdgpu_ring_mux_get_wptr(struct amdgpu_ring_mux *mux, struct amdgpu_ring *ring);
+u64 amdgpu_ring_mux_get_rptr(struct amdgpu_ring_mux *mux, struct amdgpu_ring *ring);
+void amdgpu_ring_mux_start_ib(struct amdgpu_ring_mux *mux, struct amdgpu_ring *ring);
+void amdgpu_ring_mux_end_ib(struct amdgpu_ring_mux *mux, struct amdgpu_ring *ring);
+void amdgpu_ring_mux_ib_mark_offset(struct amdgpu_ring_mux *mux, struct amdgpu_ring *ring,
+				    u64 offset, enum amdgpu_ring_mux_offset_type type);
+bool amdgpu_mcbp_handle_trailing_fence_irq(struct amdgpu_ring_mux *mux);
+
+u64 amdgpu_sw_ring_get_rptr_gfx(struct amdgpu_ring *ring);
+u64 amdgpu_sw_ring_get_wptr_gfx(struct amdgpu_ring *ring);
+void amdgpu_sw_ring_set_wptr_gfx(struct amdgpu_ring *ring);
+void amdgpu_sw_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count);
+void amdgpu_sw_ring_ib_begin(struct amdgpu_ring *ring);
+void amdgpu_sw_ring_ib_end(struct amdgpu_ring *ring);
+void amdgpu_sw_ring_ib_mark_offset(struct amdgpu_ring *ring, enum amdgpu_ring_mux_offset_type type);
+const char *amdgpu_sw_ring_name(int idx);
+unsigned int amdgpu_sw_ring_priority(int idx);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.c
new file mode 100644
index 000000000000..5aa830a02d80
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.c
@@ -0,0 +1,550 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/firmware.h>
+#include "amdgpu.h"
+#include "amdgpu_gfx.h"
+#include "amdgpu_rlc.h"
+
+/**
+ * amdgpu_gfx_rlc_enter_safe_mode - Set RLC into safe mode
+ *
+ * @adev: amdgpu_device pointer
+ * @xcc_id: xcc accelerated compute core id
+ *
+ * Set RLC enter into safe mode if RLC is enabled and haven't in safe mode.
+ */
+void amdgpu_gfx_rlc_enter_safe_mode(struct amdgpu_device *adev, int xcc_id)
+{
+	if (adev->gfx.rlc.in_safe_mode[xcc_id])
+		return;
+
+	/* if RLC is not enabled, do nothing */
+	if (!adev->gfx.rlc.funcs->is_rlc_enabled(adev))
+		return;
+
+	if (adev->cg_flags &
+	    (AMD_CG_SUPPORT_GFX_CGCG | AMD_CG_SUPPORT_GFX_MGCG |
+	     AMD_CG_SUPPORT_GFX_3D_CGCG)) {
+		adev->gfx.rlc.funcs->set_safe_mode(adev, xcc_id);
+		adev->gfx.rlc.in_safe_mode[xcc_id] = true;
+	}
+}
+
+/**
+ * amdgpu_gfx_rlc_exit_safe_mode - Set RLC out of safe mode
+ *
+ * @adev: amdgpu_device pointer
+ * @xcc_id: xcc accelerated compute core id
+ *
+ * Set RLC exit safe mode if RLC is enabled and have entered into safe mode.
+ */
+void amdgpu_gfx_rlc_exit_safe_mode(struct amdgpu_device *adev, int xcc_id)
+{
+	if (!(adev->gfx.rlc.in_safe_mode[xcc_id]))
+		return;
+
+	/* if RLC is not enabled, do nothing */
+	if (!adev->gfx.rlc.funcs->is_rlc_enabled(adev))
+		return;
+
+	if (adev->cg_flags &
+	    (AMD_CG_SUPPORT_GFX_CGCG | AMD_CG_SUPPORT_GFX_MGCG |
+	     AMD_CG_SUPPORT_GFX_3D_CGCG)) {
+		adev->gfx.rlc.funcs->unset_safe_mode(adev, xcc_id);
+		adev->gfx.rlc.in_safe_mode[xcc_id] = false;
+	}
+}
+
+/**
+ * amdgpu_gfx_rlc_init_sr - Init save restore block
+ *
+ * @adev: amdgpu_device pointer
+ * @dws: the size of save restore block
+ *
+ * Allocate and setup value to save restore block of rlc.
+ * Returns 0 on succeess or negative error code if allocate failed.
+ */
+int amdgpu_gfx_rlc_init_sr(struct amdgpu_device *adev, u32 dws)
+{
+	const u32 *src_ptr;
+	u32 *dst_ptr;
+	u32 i;
+	int r;
+
+	/* allocate save restore block */
+	r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_VRAM |
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.rlc.save_restore_obj,
+				      &adev->gfx.rlc.save_restore_gpu_addr,
+				      (void **)&adev->gfx.rlc.sr_ptr);
+	if (r) {
+		dev_warn(adev->dev, "(%d) create RLC sr bo failed\n", r);
+		amdgpu_gfx_rlc_fini(adev);
+		return r;
+	}
+
+	/* write the sr buffer */
+	src_ptr = adev->gfx.rlc.reg_list;
+	dst_ptr = adev->gfx.rlc.sr_ptr;
+	for (i = 0; i < adev->gfx.rlc.reg_list_size; i++)
+		dst_ptr[i] = cpu_to_le32(src_ptr[i]);
+	amdgpu_bo_kunmap(adev->gfx.rlc.save_restore_obj);
+	amdgpu_bo_unreserve(adev->gfx.rlc.save_restore_obj);
+
+	return 0;
+}
+
+/**
+ * amdgpu_gfx_rlc_init_csb - Init clear state block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate and setup value to clear state block of rlc.
+ * Returns 0 on succeess or negative error code if allocate failed.
+ */
+int amdgpu_gfx_rlc_init_csb(struct amdgpu_device *adev)
+{
+	u32 dws;
+	int r;
+
+	/* allocate clear state block */
+	adev->gfx.rlc.clear_state_size = dws = adev->gfx.rlc.funcs->get_csb_size(adev);
+	r = amdgpu_bo_create_kernel(adev, dws * 4, PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_VRAM |
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.rlc.clear_state_obj,
+				      &adev->gfx.rlc.clear_state_gpu_addr,
+				      (void **)&adev->gfx.rlc.cs_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create rlc csb bo\n", r);
+		amdgpu_gfx_rlc_fini(adev);
+		return r;
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_gfx_rlc_init_cpt - Init cp table
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate and setup value to cp table of rlc.
+ * Returns 0 on succeess or negative error code if allocate failed.
+ */
+int amdgpu_gfx_rlc_init_cpt(struct amdgpu_device *adev)
+{
+	int r;
+
+	r = amdgpu_bo_create_reserved(adev, adev->gfx.rlc.cp_table_size,
+				      PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM |
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.rlc.cp_table_obj,
+				      &adev->gfx.rlc.cp_table_gpu_addr,
+				      (void **)&adev->gfx.rlc.cp_table_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create cp table bo\n", r);
+		amdgpu_gfx_rlc_fini(adev);
+		return r;
+	}
+
+	/* set up the cp table */
+	amdgpu_gfx_rlc_setup_cp_table(adev);
+	amdgpu_bo_kunmap(adev->gfx.rlc.cp_table_obj);
+	amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj);
+
+	return 0;
+}
+
+/**
+ * amdgpu_gfx_rlc_setup_cp_table - setup cp the buffer of cp table
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Write cp firmware data into cp table.
+ */
+void amdgpu_gfx_rlc_setup_cp_table(struct amdgpu_device *adev)
+{
+	const __le32 *fw_data;
+	u32 *dst_ptr;
+	int me, i, max_me;
+	u32 bo_offset = 0;
+	u32 table_offset, table_size;
+
+	max_me = adev->gfx.rlc.funcs->get_cp_table_num(adev);
+
+	/* write the cp table buffer */
+	dst_ptr = adev->gfx.rlc.cp_table_ptr;
+	for (me = 0; me < max_me; me++) {
+		if (me == 0) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.ce_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else if (me == 1) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.pfp_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else if (me == 2) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.me_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else if (me == 3) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.mec_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else  if (me == 4) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.mec2_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.mec2_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		}
+
+		for (i = 0; i < table_size; i++) {
+			dst_ptr[bo_offset + i] =
+				cpu_to_le32(le32_to_cpu(fw_data[table_offset + i]));
+		}
+
+		bo_offset += table_size;
+	}
+}
+
+/**
+ * amdgpu_gfx_rlc_fini - Free BO which used for RLC
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Free three BO which is used for rlc_save_restore_block, rlc_clear_state_block
+ * and rlc_jump_table_block.
+ */
+void amdgpu_gfx_rlc_fini(struct amdgpu_device *adev)
+{
+	/* save restore block */
+	if (adev->gfx.rlc.save_restore_obj) {
+		amdgpu_bo_free_kernel(&adev->gfx.rlc.save_restore_obj,
+				      &adev->gfx.rlc.save_restore_gpu_addr,
+				      (void **)&adev->gfx.rlc.sr_ptr);
+	}
+
+	/* clear state block */
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj,
+			      &adev->gfx.rlc.clear_state_gpu_addr,
+			      (void **)&adev->gfx.rlc.cs_ptr);
+
+	/* jump table block */
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj,
+			      &adev->gfx.rlc.cp_table_gpu_addr,
+			      (void **)&adev->gfx.rlc.cp_table_ptr);
+}
+
+static int amdgpu_gfx_rlc_init_microcode_v2_0(struct amdgpu_device *adev)
+{
+	const struct common_firmware_header *common_hdr;
+	const struct rlc_firmware_header_v2_0 *rlc_hdr;
+	struct amdgpu_firmware_info *info;
+	unsigned int *tmp;
+	unsigned int i;
+
+	rlc_hdr = (const struct rlc_firmware_header_v2_0 *)adev->gfx.rlc_fw->data;
+
+	adev->gfx.rlc_fw_version = le32_to_cpu(rlc_hdr->header.ucode_version);
+	adev->gfx.rlc_feature_version = le32_to_cpu(rlc_hdr->ucode_feature_version);
+	adev->gfx.rlc.save_and_restore_offset =
+		le32_to_cpu(rlc_hdr->save_and_restore_offset);
+	adev->gfx.rlc.clear_state_descriptor_offset =
+		le32_to_cpu(rlc_hdr->clear_state_descriptor_offset);
+	adev->gfx.rlc.avail_scratch_ram_locations =
+		le32_to_cpu(rlc_hdr->avail_scratch_ram_locations);
+	adev->gfx.rlc.reg_restore_list_size =
+		le32_to_cpu(rlc_hdr->reg_restore_list_size);
+	adev->gfx.rlc.reg_list_format_start =
+		le32_to_cpu(rlc_hdr->reg_list_format_start);
+	adev->gfx.rlc.reg_list_format_separate_start =
+		le32_to_cpu(rlc_hdr->reg_list_format_separate_start);
+	adev->gfx.rlc.starting_offsets_start =
+		le32_to_cpu(rlc_hdr->starting_offsets_start);
+	adev->gfx.rlc.reg_list_format_size_bytes =
+		le32_to_cpu(rlc_hdr->reg_list_format_size_bytes);
+	adev->gfx.rlc.reg_list_size_bytes =
+		le32_to_cpu(rlc_hdr->reg_list_size_bytes);
+	adev->gfx.rlc.register_list_format =
+		kmalloc(adev->gfx.rlc.reg_list_format_size_bytes +
+			adev->gfx.rlc.reg_list_size_bytes, GFP_KERNEL);
+	if (!adev->gfx.rlc.register_list_format) {
+		dev_err(adev->dev, "failed to allocate memory for rlc register_list_format\n");
+		return -ENOMEM;
+	}
+
+	tmp = (unsigned int *)((uintptr_t)rlc_hdr +
+			le32_to_cpu(rlc_hdr->reg_list_format_array_offset_bytes));
+	for (i = 0 ; i < (rlc_hdr->reg_list_format_size_bytes >> 2); i++)
+		adev->gfx.rlc.register_list_format[i] = le32_to_cpu(tmp[i]);
+
+	adev->gfx.rlc.register_restore = adev->gfx.rlc.register_list_format + i;
+
+	tmp = (unsigned int *)((uintptr_t)rlc_hdr +
+			le32_to_cpu(rlc_hdr->reg_list_array_offset_bytes));
+	for (i = 0 ; i < (rlc_hdr->reg_list_size_bytes >> 2); i++)
+		adev->gfx.rlc.register_restore[i] = le32_to_cpu(tmp[i]);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_RLC_G];
+		info->ucode_id = AMDGPU_UCODE_ID_RLC_G;
+		info->fw = adev->gfx.rlc_fw;
+		if (info->fw) {
+			common_hdr = (const struct common_firmware_header *)info->fw->data;
+			adev->firmware.fw_size +=
+				ALIGN(le32_to_cpu(common_hdr->ucode_size_bytes), PAGE_SIZE);
+		}
+	}
+
+	return 0;
+}
+
+static void amdgpu_gfx_rlc_init_microcode_v2_1(struct amdgpu_device *adev)
+{
+	const struct rlc_firmware_header_v2_1 *rlc_hdr;
+	struct amdgpu_firmware_info *info;
+
+	rlc_hdr = (const struct rlc_firmware_header_v2_1 *)adev->gfx.rlc_fw->data;
+	adev->gfx.rlc_srlc_fw_version = le32_to_cpu(rlc_hdr->save_restore_list_cntl_ucode_ver);
+	adev->gfx.rlc_srlc_feature_version = le32_to_cpu(rlc_hdr->save_restore_list_cntl_feature_ver);
+	adev->gfx.rlc.save_restore_list_cntl_size_bytes = le32_to_cpu(rlc_hdr->save_restore_list_cntl_size_bytes);
+	adev->gfx.rlc.save_restore_list_cntl = (u8 *)rlc_hdr + le32_to_cpu(rlc_hdr->save_restore_list_cntl_offset_bytes);
+	adev->gfx.rlc_srlg_fw_version = le32_to_cpu(rlc_hdr->save_restore_list_gpm_ucode_ver);
+	adev->gfx.rlc_srlg_feature_version = le32_to_cpu(rlc_hdr->save_restore_list_gpm_feature_ver);
+	adev->gfx.rlc.save_restore_list_gpm_size_bytes = le32_to_cpu(rlc_hdr->save_restore_list_gpm_size_bytes);
+	adev->gfx.rlc.save_restore_list_gpm = (u8 *)rlc_hdr + le32_to_cpu(rlc_hdr->save_restore_list_gpm_offset_bytes);
+	adev->gfx.rlc_srls_fw_version = le32_to_cpu(rlc_hdr->save_restore_list_srm_ucode_ver);
+	adev->gfx.rlc_srls_feature_version = le32_to_cpu(rlc_hdr->save_restore_list_srm_feature_ver);
+	adev->gfx.rlc.save_restore_list_srm_size_bytes = le32_to_cpu(rlc_hdr->save_restore_list_srm_size_bytes);
+	adev->gfx.rlc.save_restore_list_srm = (u8 *)rlc_hdr + le32_to_cpu(rlc_hdr->save_restore_list_srm_offset_bytes);
+	adev->gfx.rlc.reg_list_format_direct_reg_list_length =
+		le32_to_cpu(rlc_hdr->reg_list_format_direct_reg_list_length);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		if (adev->gfx.rlc.save_restore_list_cntl_size_bytes) {
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL];
+			info->ucode_id = AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL;
+			info->fw = adev->gfx.rlc_fw;
+			adev->firmware.fw_size +=
+				ALIGN(adev->gfx.rlc.save_restore_list_cntl_size_bytes, PAGE_SIZE);
+		}
+
+		if (adev->gfx.rlc.save_restore_list_gpm_size_bytes) {
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM];
+			info->ucode_id = AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM;
+			info->fw = adev->gfx.rlc_fw;
+			adev->firmware.fw_size +=
+				ALIGN(adev->gfx.rlc.save_restore_list_gpm_size_bytes, PAGE_SIZE);
+		}
+
+		if (adev->gfx.rlc.save_restore_list_srm_size_bytes) {
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM];
+			info->ucode_id = AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM;
+			info->fw = adev->gfx.rlc_fw;
+			adev->firmware.fw_size +=
+				ALIGN(adev->gfx.rlc.save_restore_list_srm_size_bytes, PAGE_SIZE);
+		}
+	}
+}
+
+static void amdgpu_gfx_rlc_init_microcode_v2_2(struct amdgpu_device *adev)
+{
+	const struct rlc_firmware_header_v2_2 *rlc_hdr;
+	struct amdgpu_firmware_info *info;
+
+	rlc_hdr = (const struct rlc_firmware_header_v2_2 *)adev->gfx.rlc_fw->data;
+	adev->gfx.rlc.rlc_iram_ucode_size_bytes = le32_to_cpu(rlc_hdr->rlc_iram_ucode_size_bytes);
+	adev->gfx.rlc.rlc_iram_ucode = (u8 *)rlc_hdr + le32_to_cpu(rlc_hdr->rlc_iram_ucode_offset_bytes);
+	adev->gfx.rlc.rlc_dram_ucode_size_bytes = le32_to_cpu(rlc_hdr->rlc_dram_ucode_size_bytes);
+	adev->gfx.rlc.rlc_dram_ucode = (u8 *)rlc_hdr + le32_to_cpu(rlc_hdr->rlc_dram_ucode_offset_bytes);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		if (adev->gfx.rlc.rlc_iram_ucode_size_bytes) {
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_RLC_IRAM];
+			info->ucode_id = AMDGPU_UCODE_ID_RLC_IRAM;
+			info->fw = adev->gfx.rlc_fw;
+			adev->firmware.fw_size +=
+				ALIGN(adev->gfx.rlc.rlc_iram_ucode_size_bytes, PAGE_SIZE);
+		}
+
+		if (adev->gfx.rlc.rlc_dram_ucode_size_bytes) {
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_RLC_DRAM];
+			info->ucode_id = AMDGPU_UCODE_ID_RLC_DRAM;
+			info->fw = adev->gfx.rlc_fw;
+			adev->firmware.fw_size +=
+				ALIGN(adev->gfx.rlc.rlc_dram_ucode_size_bytes, PAGE_SIZE);
+		}
+	}
+}
+
+static void amdgpu_gfx_rlc_init_microcode_v2_3(struct amdgpu_device *adev)
+{
+	const struct rlc_firmware_header_v2_3 *rlc_hdr;
+	struct amdgpu_firmware_info *info;
+
+	rlc_hdr = (const struct rlc_firmware_header_v2_3 *)adev->gfx.rlc_fw->data;
+	adev->gfx.rlcp_ucode_version = le32_to_cpu(rlc_hdr->rlcp_ucode_version);
+	adev->gfx.rlcp_ucode_feature_version = le32_to_cpu(rlc_hdr->rlcp_ucode_feature_version);
+	adev->gfx.rlc.rlcp_ucode_size_bytes = le32_to_cpu(rlc_hdr->rlcp_ucode_size_bytes);
+	adev->gfx.rlc.rlcp_ucode = (u8 *)rlc_hdr + le32_to_cpu(rlc_hdr->rlcp_ucode_offset_bytes);
+
+	adev->gfx.rlcv_ucode_version = le32_to_cpu(rlc_hdr->rlcv_ucode_version);
+	adev->gfx.rlcv_ucode_feature_version = le32_to_cpu(rlc_hdr->rlcv_ucode_feature_version);
+	adev->gfx.rlc.rlcv_ucode_size_bytes = le32_to_cpu(rlc_hdr->rlcv_ucode_size_bytes);
+	adev->gfx.rlc.rlcv_ucode = (u8 *)rlc_hdr + le32_to_cpu(rlc_hdr->rlcv_ucode_offset_bytes);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		if (adev->gfx.rlc.rlcp_ucode_size_bytes) {
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_RLC_P];
+			info->ucode_id = AMDGPU_UCODE_ID_RLC_P;
+			info->fw = adev->gfx.rlc_fw;
+			adev->firmware.fw_size +=
+				ALIGN(adev->gfx.rlc.rlcp_ucode_size_bytes, PAGE_SIZE);
+		}
+
+		if (adev->gfx.rlc.rlcv_ucode_size_bytes) {
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_RLC_V];
+			info->ucode_id = AMDGPU_UCODE_ID_RLC_V;
+			info->fw = adev->gfx.rlc_fw;
+			adev->firmware.fw_size +=
+				ALIGN(adev->gfx.rlc.rlcv_ucode_size_bytes, PAGE_SIZE);
+		}
+	}
+}
+
+static void amdgpu_gfx_rlc_init_microcode_v2_4(struct amdgpu_device *adev)
+{
+	const struct rlc_firmware_header_v2_4 *rlc_hdr;
+	struct amdgpu_firmware_info *info;
+
+	rlc_hdr = (const struct rlc_firmware_header_v2_4 *)adev->gfx.rlc_fw->data;
+	adev->gfx.rlc.global_tap_delays_ucode_size_bytes = le32_to_cpu(rlc_hdr->global_tap_delays_ucode_size_bytes);
+	adev->gfx.rlc.global_tap_delays_ucode = (u8 *)rlc_hdr + le32_to_cpu(rlc_hdr->global_tap_delays_ucode_offset_bytes);
+	adev->gfx.rlc.se0_tap_delays_ucode_size_bytes = le32_to_cpu(rlc_hdr->se0_tap_delays_ucode_size_bytes);
+	adev->gfx.rlc.se0_tap_delays_ucode = (u8 *)rlc_hdr + le32_to_cpu(rlc_hdr->se0_tap_delays_ucode_offset_bytes);
+	adev->gfx.rlc.se1_tap_delays_ucode_size_bytes = le32_to_cpu(rlc_hdr->se1_tap_delays_ucode_size_bytes);
+	adev->gfx.rlc.se1_tap_delays_ucode = (u8 *)rlc_hdr + le32_to_cpu(rlc_hdr->se1_tap_delays_ucode_offset_bytes);
+	adev->gfx.rlc.se2_tap_delays_ucode_size_bytes = le32_to_cpu(rlc_hdr->se2_tap_delays_ucode_size_bytes);
+	adev->gfx.rlc.se2_tap_delays_ucode = (u8 *)rlc_hdr + le32_to_cpu(rlc_hdr->se2_tap_delays_ucode_offset_bytes);
+	adev->gfx.rlc.se3_tap_delays_ucode_size_bytes = le32_to_cpu(rlc_hdr->se3_tap_delays_ucode_size_bytes);
+	adev->gfx.rlc.se3_tap_delays_ucode = (u8 *)rlc_hdr + le32_to_cpu(rlc_hdr->se3_tap_delays_ucode_offset_bytes);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		if (adev->gfx.rlc.global_tap_delays_ucode_size_bytes) {
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_GLOBAL_TAP_DELAYS];
+			info->ucode_id = AMDGPU_UCODE_ID_GLOBAL_TAP_DELAYS;
+			info->fw = adev->gfx.rlc_fw;
+			adev->firmware.fw_size +=
+				ALIGN(adev->gfx.rlc.global_tap_delays_ucode_size_bytes, PAGE_SIZE);
+		}
+
+		if (adev->gfx.rlc.se0_tap_delays_ucode_size_bytes) {
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_SE0_TAP_DELAYS];
+			info->ucode_id = AMDGPU_UCODE_ID_SE0_TAP_DELAYS;
+			info->fw = adev->gfx.rlc_fw;
+			adev->firmware.fw_size +=
+				ALIGN(adev->gfx.rlc.se0_tap_delays_ucode_size_bytes, PAGE_SIZE);
+		}
+
+		if (adev->gfx.rlc.se1_tap_delays_ucode_size_bytes) {
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_SE1_TAP_DELAYS];
+			info->ucode_id = AMDGPU_UCODE_ID_SE1_TAP_DELAYS;
+			info->fw = adev->gfx.rlc_fw;
+			adev->firmware.fw_size +=
+				ALIGN(adev->gfx.rlc.se1_tap_delays_ucode_size_bytes, PAGE_SIZE);
+		}
+
+		if (adev->gfx.rlc.se2_tap_delays_ucode_size_bytes) {
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_SE2_TAP_DELAYS];
+			info->ucode_id = AMDGPU_UCODE_ID_SE2_TAP_DELAYS;
+			info->fw = adev->gfx.rlc_fw;
+			adev->firmware.fw_size +=
+				ALIGN(adev->gfx.rlc.se2_tap_delays_ucode_size_bytes, PAGE_SIZE);
+		}
+
+		if (adev->gfx.rlc.se3_tap_delays_ucode_size_bytes) {
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_SE3_TAP_DELAYS];
+			info->ucode_id = AMDGPU_UCODE_ID_SE3_TAP_DELAYS;
+			info->fw = adev->gfx.rlc_fw;
+			adev->firmware.fw_size +=
+				ALIGN(adev->gfx.rlc.se3_tap_delays_ucode_size_bytes, PAGE_SIZE);
+		}
+	}
+}
+
+int amdgpu_gfx_rlc_init_microcode(struct amdgpu_device *adev,
+				  uint16_t version_major,
+				  uint16_t version_minor)
+{
+	int err;
+
+	if (version_major < 2) {
+		/* only support rlc_hdr v2.x and onwards */
+		dev_err(adev->dev, "unsupported rlc fw hdr\n");
+		return -EINVAL;
+	}
+
+	/* is_rlc_v2_1 is still used in APU code path */
+	if (version_major == 2 && version_minor == 1)
+		adev->gfx.rlc.is_rlc_v2_1 = true;
+
+	err = amdgpu_gfx_rlc_init_microcode_v2_0(adev);
+	if (err) {
+		dev_err(adev->dev, "fail to init rlc v2_0 microcode\n");
+		return err;
+	}
+
+	if (version_minor >= 1)
+		amdgpu_gfx_rlc_init_microcode_v2_1(adev);
+	if (version_minor >= 2)
+		amdgpu_gfx_rlc_init_microcode_v2_2(adev);
+	if (version_minor == 3)
+		amdgpu_gfx_rlc_init_microcode_v2_3(adev);
+	if (version_minor == 4)
+		amdgpu_gfx_rlc_init_microcode_v2_4(adev);
+
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h
new file mode 100644
index 000000000000..2ce310b31942
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h
@@ -0,0 +1,363 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_RLC_H__
+#define __AMDGPU_RLC_H__
+
+#include "clearstate_defs.h"
+
+#define AMDGPU_MAX_RLC_INSTANCES	8
+
+/* firmware ID used in rlc toc */
+typedef enum _FIRMWARE_ID_ {
+	FIRMWARE_ID_INVALID					= 0,
+	FIRMWARE_ID_RLC_G_UCODE					= 1,
+	FIRMWARE_ID_RLC_TOC					= 2,
+	FIRMWARE_ID_RLCG_SCRATCH                                = 3,
+	FIRMWARE_ID_RLC_SRM_ARAM                                = 4,
+	FIRMWARE_ID_RLC_SRM_INDEX_ADDR                          = 5,
+	FIRMWARE_ID_RLC_SRM_INDEX_DATA                          = 6,
+	FIRMWARE_ID_RLC_P_UCODE                                 = 7,
+	FIRMWARE_ID_RLC_V_UCODE                                 = 8,
+	FIRMWARE_ID_RLX6_UCODE                                  = 9,
+	FIRMWARE_ID_RLX6_DRAM_BOOT                              = 10,
+	FIRMWARE_ID_GLOBAL_TAP_DELAYS                           = 11,
+	FIRMWARE_ID_SE0_TAP_DELAYS                              = 12,
+	FIRMWARE_ID_SE1_TAP_DELAYS                              = 13,
+	FIRMWARE_ID_GLOBAL_SE0_SE1_SKEW_DELAYS                  = 14,
+	FIRMWARE_ID_SDMA0_UCODE                                 = 15,
+	FIRMWARE_ID_SDMA0_JT                                    = 16,
+	FIRMWARE_ID_SDMA1_UCODE                                 = 17,
+	FIRMWARE_ID_SDMA1_JT                                    = 18,
+	FIRMWARE_ID_CP_CE                                       = 19,
+	FIRMWARE_ID_CP_PFP                                      = 20,
+	FIRMWARE_ID_CP_ME                                       = 21,
+	FIRMWARE_ID_CP_MEC                                      = 22,
+	FIRMWARE_ID_CP_MES                                      = 23,
+	FIRMWARE_ID_MES_STACK                                   = 24,
+	FIRMWARE_ID_RLC_SRM_DRAM_SR                             = 25,
+	FIRMWARE_ID_RLCG_SCRATCH_SR                             = 26,
+	FIRMWARE_ID_RLCP_SCRATCH_SR                             = 27,
+	FIRMWARE_ID_RLCV_SCRATCH_SR                             = 28,
+	FIRMWARE_ID_RLX6_DRAM_SR                                = 29,
+	FIRMWARE_ID_SDMA0_PG_CONTEXT                            = 30,
+	FIRMWARE_ID_SDMA1_PG_CONTEXT                            = 31,
+	FIRMWARE_ID_GLOBAL_MUX_SELECT_RAM                       = 32,
+	FIRMWARE_ID_SE0_MUX_SELECT_RAM                          = 33,
+	FIRMWARE_ID_SE1_MUX_SELECT_RAM                          = 34,
+	FIRMWARE_ID_ACCUM_CTRL_RAM                              = 35,
+	FIRMWARE_ID_RLCP_CAM                                    = 36,
+	FIRMWARE_ID_RLC_SPP_CAM_EXT                             = 37,
+	FIRMWARE_ID_MAX                                         = 38,
+} FIRMWARE_ID;
+
+typedef enum _SOC21_FIRMWARE_ID_ {
+    SOC21_FIRMWARE_ID_INVALID                     = 0,
+    SOC21_FIRMWARE_ID_RLC_G_UCODE                 = 1,
+    SOC21_FIRMWARE_ID_RLC_TOC                     = 2,
+    SOC21_FIRMWARE_ID_RLCG_SCRATCH                = 3,
+    SOC21_FIRMWARE_ID_RLC_SRM_ARAM                = 4,
+    SOC21_FIRMWARE_ID_RLC_P_UCODE                 = 5,
+    SOC21_FIRMWARE_ID_RLC_V_UCODE                 = 6,
+    SOC21_FIRMWARE_ID_RLX6_UCODE                  = 7,
+    SOC21_FIRMWARE_ID_RLX6_UCODE_CORE1            = 8,
+    SOC21_FIRMWARE_ID_RLX6_DRAM_BOOT              = 9,
+    SOC21_FIRMWARE_ID_RLX6_DRAM_BOOT_CORE1        = 10,
+    SOC21_FIRMWARE_ID_SDMA_UCODE_TH0              = 11,
+    SOC21_FIRMWARE_ID_SDMA_UCODE_TH1              = 12,
+    SOC21_FIRMWARE_ID_CP_PFP                      = 13,
+    SOC21_FIRMWARE_ID_CP_ME                       = 14,
+    SOC21_FIRMWARE_ID_CP_MEC                      = 15,
+    SOC21_FIRMWARE_ID_RS64_MES_P0                 = 16,
+    SOC21_FIRMWARE_ID_RS64_MES_P1                 = 17,
+    SOC21_FIRMWARE_ID_RS64_PFP                    = 18,
+    SOC21_FIRMWARE_ID_RS64_ME                     = 19,
+    SOC21_FIRMWARE_ID_RS64_MEC                    = 20,
+    SOC21_FIRMWARE_ID_RS64_MES_P0_STACK           = 21,
+    SOC21_FIRMWARE_ID_RS64_MES_P1_STACK           = 22,
+    SOC21_FIRMWARE_ID_RS64_PFP_P0_STACK           = 23,
+    SOC21_FIRMWARE_ID_RS64_PFP_P1_STACK           = 24,
+    SOC21_FIRMWARE_ID_RS64_ME_P0_STACK            = 25,
+    SOC21_FIRMWARE_ID_RS64_ME_P1_STACK            = 26,
+    SOC21_FIRMWARE_ID_RS64_MEC_P0_STACK           = 27,
+    SOC21_FIRMWARE_ID_RS64_MEC_P1_STACK           = 28,
+    SOC21_FIRMWARE_ID_RS64_MEC_P2_STACK           = 29,
+    SOC21_FIRMWARE_ID_RS64_MEC_P3_STACK           = 30,
+    SOC21_FIRMWARE_ID_RLC_SRM_DRAM_SR             = 31,
+    SOC21_FIRMWARE_ID_RLCG_SCRATCH_SR             = 32,
+    SOC21_FIRMWARE_ID_RLCP_SCRATCH_SR             = 33,
+    SOC21_FIRMWARE_ID_RLCV_SCRATCH_SR             = 34,
+    SOC21_FIRMWARE_ID_RLX6_DRAM_SR                = 35,
+    SOC21_FIRMWARE_ID_RLX6_DRAM_SR_CORE1          = 36,
+    SOC21_FIRMWARE_ID_MAX                         = 37
+} SOC21_FIRMWARE_ID;
+
+typedef enum _SOC24_FIRMWARE_ID_ {
+    SOC24_FIRMWARE_ID_INVALID                     = 0,
+    SOC24_FIRMWARE_ID_RLC_G_UCODE                 = 1,
+    SOC24_FIRMWARE_ID_RLC_TOC                     = 2,
+    SOC24_FIRMWARE_ID_RLCG_SCRATCH                = 3,
+    SOC24_FIRMWARE_ID_RLC_SRM_ARAM                = 4,
+    SOC24_FIRMWARE_ID_RLC_P_UCODE                 = 5,
+    SOC24_FIRMWARE_ID_RLC_V_UCODE                 = 6,
+    SOC24_FIRMWARE_ID_RLX6_UCODE                  = 7,
+    SOC24_FIRMWARE_ID_RLX6_UCODE_CORE1            = 8,
+    SOC24_FIRMWARE_ID_RLX6_DRAM_BOOT              = 9,
+    SOC24_FIRMWARE_ID_RLX6_DRAM_BOOT_CORE1        = 10,
+    SOC24_FIRMWARE_ID_SDMA_UCODE_TH0              = 11,
+    SOC24_FIRMWARE_ID_SDMA_UCODE_TH1              = 12,
+    SOC24_FIRMWARE_ID_CP_PFP                      = 13,
+    SOC24_FIRMWARE_ID_CP_ME                       = 14,
+    SOC24_FIRMWARE_ID_CP_MEC                      = 15,
+    SOC24_FIRMWARE_ID_RS64_MES_P0                 = 16,
+    SOC24_FIRMWARE_ID_RS64_MES_P1                 = 17,
+    SOC24_FIRMWARE_ID_RS64_PFP                    = 18,
+    SOC24_FIRMWARE_ID_RS64_ME                     = 19,
+    SOC24_FIRMWARE_ID_RS64_MEC                    = 20,
+    SOC24_FIRMWARE_ID_RS64_MES_P0_STACK           = 21,
+    SOC24_FIRMWARE_ID_RS64_MES_P1_STACK           = 22,
+    SOC24_FIRMWARE_ID_RS64_PFP_P0_STACK           = 23,
+    SOC24_FIRMWARE_ID_RS64_PFP_P1_STACK           = 24,
+    SOC24_FIRMWARE_ID_RS64_ME_P0_STACK            = 25,
+    SOC24_FIRMWARE_ID_RS64_ME_P1_STACK            = 26,
+    SOC24_FIRMWARE_ID_RS64_MEC_P0_STACK           = 27,
+    SOC24_FIRMWARE_ID_RS64_MEC_P1_STACK           = 28,
+    SOC24_FIRMWARE_ID_RS64_MEC_P2_STACK           = 29,
+    SOC24_FIRMWARE_ID_RS64_MEC_P3_STACK           = 30,
+    SOC24_FIRMWARE_ID_RLC_SRM_DRAM_SR             = 31,
+    SOC24_FIRMWARE_ID_RLCG_SCRATCH_SR             = 32,
+    SOC24_FIRMWARE_ID_RLCP_SCRATCH_SR             = 33,
+    SOC24_FIRMWARE_ID_RLCV_SCRATCH_SR             = 34,
+    SOC24_FIRMWARE_ID_RLX6_DRAM_SR                = 35,
+    SOC24_FIRMWARE_ID_RLX6_DRAM_SR_CORE1          = 36,
+    SOC24_FIRMWARE_ID_RLCDEBUGLOG                 = 37,
+    SOC24_FIRMWARE_ID_SRIOV_DEBUG                 = 38,
+    SOC24_FIRMWARE_ID_SRIOV_CSA_RLC               = 39,
+    SOC24_FIRMWARE_ID_SRIOV_CSA_SDMA              = 40,
+    SOC24_FIRMWARE_ID_SRIOV_CSA_CP                = 41,
+    SOC24_FIRMWARE_ID_UMF_ZONE_PAD                = 42,
+    SOC24_FIRMWARE_ID_MAX                         = 43
+} SOC24_FIRMWARE_ID;
+
+typedef struct _RLC_TABLE_OF_CONTENT {
+	union {
+		unsigned int	DW0;
+		struct {
+			unsigned int	offset		: 25;
+			unsigned int	id		: 7;
+		};
+	};
+
+	union {
+		unsigned int	DW1;
+		struct {
+			unsigned int	load_at_boot		: 1;
+			unsigned int	load_at_vddgfx		: 1;
+			unsigned int	load_at_reset		: 1;
+			unsigned int	memory_destination	: 2;
+			unsigned int	vfflr_image_code	: 4;
+			unsigned int	load_mode_direct	: 1;
+			unsigned int	save_for_vddgfx		: 1;
+			unsigned int	save_for_vfflr		: 1;
+			unsigned int	reserved		: 1;
+			unsigned int	signed_source		: 1;
+			unsigned int	size			: 18;
+		};
+	};
+
+	union {
+		unsigned int	DW2;
+		struct {
+			unsigned int	indirect_addr_reg	: 16;
+			unsigned int	index			: 16;
+		};
+	};
+
+	union {
+		unsigned int	DW3;
+		struct {
+			unsigned int	indirect_data_reg	: 16;
+			unsigned int	indirect_start_offset	: 16;
+		};
+	};
+} RLC_TABLE_OF_CONTENT;
+
+typedef struct _RLC_TABLE_OF_CONTENT_V2 {
+	union {
+		unsigned int    DW0;
+		struct {
+			uint32_t offset         : 25;
+			uint32_t id             : 7;
+		};
+	};
+
+	union {
+		unsigned int    DW1;
+		struct {
+			uint32_t reserved0              : 1;
+			uint32_t reserved1              : 1;
+			uint32_t reserved2              : 1;
+			uint32_t memory_destination     : 2;
+			uint32_t vfflr_image_code       : 4;
+			uint32_t reserved9              : 1;
+			uint32_t reserved10             : 1;
+			uint32_t reserved11             : 1;
+			uint32_t size_x16               : 1;
+			uint32_t reserved13             : 1;
+			uint32_t size                   : 18;
+		};
+	};
+} RLC_TABLE_OF_CONTENT_V2;
+
+#define RLC_TOC_MAX_SIZE		64
+
+struct amdgpu_rlc_funcs {
+	bool (*is_rlc_enabled)(struct amdgpu_device *adev);
+	void (*set_safe_mode)(struct amdgpu_device *adev, int xcc_id);
+	void (*unset_safe_mode)(struct amdgpu_device *adev, int xcc_id);
+	int  (*init)(struct amdgpu_device *adev);
+	u32  (*get_csb_size)(struct amdgpu_device *adev);
+
+	/**
+	 * @get_csb_buffer: Get the clear state to be put into the hardware.
+	 *
+	 * The parameter adev is used to get the CS data and other gfx info,
+	 * and buffer is the RLC CS pointer
+	 *
+	 * Sometimes, the user space puts a request to clear the state in the
+	 * command buffer; this function provides the clear state that gets put
+	 * into the hardware. Note that the driver programs Clear State
+	 * Indirect Buffer (CSB) explicitly when it sets up the kernel rings,
+	 * and it also provides a pointer to it which is used by the firmware
+	 * to load the clear state in some cases.
+	 */
+	void (*get_csb_buffer)(struct amdgpu_device *adev, u32 *buffer);
+	int  (*get_cp_table_num)(struct amdgpu_device *adev);
+	int  (*resume)(struct amdgpu_device *adev);
+	void (*stop)(struct amdgpu_device *adev);
+	void (*reset)(struct amdgpu_device *adev);
+	void (*start)(struct amdgpu_device *adev);
+	void (*update_spm_vmid)(struct amdgpu_device *adev, struct amdgpu_ring *ring, unsigned vmid);
+	bool (*is_rlcg_access_range)(struct amdgpu_device *adev, uint32_t reg);
+};
+
+struct amdgpu_rlcg_reg_access_ctrl {
+	uint32_t scratch_reg0;
+	uint32_t scratch_reg1;
+	uint32_t scratch_reg2;
+	uint32_t scratch_reg3;
+	uint32_t grbm_cntl;
+	uint32_t grbm_idx;
+	uint32_t spare_int;
+};
+
+struct amdgpu_rlc {
+	/* for power gating */
+	struct amdgpu_bo        *save_restore_obj;
+	uint64_t                save_restore_gpu_addr;
+	uint32_t		*sr_ptr;
+	const u32               *reg_list;
+	u32                     reg_list_size;
+	/* for clear state */
+	struct amdgpu_bo        *clear_state_obj;
+	uint64_t                clear_state_gpu_addr;
+	uint32_t		*cs_ptr;
+	const struct cs_section_def   *cs_data;
+	u32                     clear_state_size;
+	/* for cp tables */
+	struct amdgpu_bo        *cp_table_obj;
+	uint64_t                cp_table_gpu_addr;
+	uint32_t		*cp_table_ptr;
+	u32                     cp_table_size;
+
+	/* safe mode for updating CG/PG state */
+	bool in_safe_mode[AMDGPU_MAX_RLC_INSTANCES];
+	const struct amdgpu_rlc_funcs *funcs;
+
+	/* for firmware data */
+	u32 save_and_restore_offset;
+	u32 clear_state_descriptor_offset;
+	u32 avail_scratch_ram_locations;
+	u32 reg_restore_list_size;
+	u32 reg_list_format_start;
+	u32 reg_list_format_separate_start;
+	u32 starting_offsets_start;
+	u32 reg_list_format_size_bytes;
+	u32 reg_list_size_bytes;
+	u32 reg_list_format_direct_reg_list_length;
+	u32 save_restore_list_cntl_size_bytes;
+	u32 save_restore_list_gpm_size_bytes;
+	u32 save_restore_list_srm_size_bytes;
+	u32 rlc_iram_ucode_size_bytes;
+	u32 rlc_dram_ucode_size_bytes;
+	u32 rlcp_ucode_size_bytes;
+	u32 rlcv_ucode_size_bytes;
+	u32 global_tap_delays_ucode_size_bytes;
+	u32 se0_tap_delays_ucode_size_bytes;
+	u32 se1_tap_delays_ucode_size_bytes;
+	u32 se2_tap_delays_ucode_size_bytes;
+	u32 se3_tap_delays_ucode_size_bytes;
+
+	u32 *register_list_format;
+	u32 *register_restore;
+	u8 *save_restore_list_cntl;
+	u8 *save_restore_list_gpm;
+	u8 *save_restore_list_srm;
+	u8 *rlc_iram_ucode;
+	u8 *rlc_dram_ucode;
+	u8 *rlcp_ucode;
+	u8 *rlcv_ucode;
+	u8 *global_tap_delays_ucode;
+	u8 *se0_tap_delays_ucode;
+	u8 *se1_tap_delays_ucode;
+	u8 *se2_tap_delays_ucode;
+	u8 *se3_tap_delays_ucode;
+
+	bool is_rlc_v2_1;
+
+	/* for rlc autoload */
+	struct amdgpu_bo	*rlc_autoload_bo;
+	u64			rlc_autoload_gpu_addr;
+	void			*rlc_autoload_ptr;
+
+	/* rlc toc buffer */
+	struct amdgpu_bo	*rlc_toc_bo;
+	uint64_t		rlc_toc_gpu_addr;
+	void			*rlc_toc_buf;
+
+	bool rlcg_reg_access_supported;
+	/* registers for rlcg indirect reg access */
+	struct amdgpu_rlcg_reg_access_ctrl reg_access_ctrl[AMDGPU_MAX_RLC_INSTANCES];
+};
+
+void amdgpu_gfx_rlc_enter_safe_mode(struct amdgpu_device *adev, int xcc_id);
+void amdgpu_gfx_rlc_exit_safe_mode(struct amdgpu_device *adev, int xcc_id);
+int amdgpu_gfx_rlc_init_sr(struct amdgpu_device *adev, u32 dws);
+int amdgpu_gfx_rlc_init_csb(struct amdgpu_device *adev);
+int amdgpu_gfx_rlc_init_cpt(struct amdgpu_device *adev);
+void amdgpu_gfx_rlc_setup_cp_table(struct amdgpu_device *adev);
+void amdgpu_gfx_rlc_fini(struct amdgpu_device *adev);
+int amdgpu_gfx_rlc_init_microcode(struct amdgpu_device *adev,
+				  uint16_t version_major,
+				  uint16_t version_minor);
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c
index 3144400435b7..39070b2a4c04 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c
@@ -41,370 +41,65 @@
  * If we are asked to block we wait on all the oldest fence of all
  * rings. We just wait for any of those fence to complete.
  */
-#include <drm/drmP.h>
-#include "amdgpu.h"
 
-static void amdgpu_sa_bo_remove_locked(struct amdgpu_sa_bo *sa_bo);
-static void amdgpu_sa_bo_try_free(struct amdgpu_sa_manager *sa_manager);
+#include "amdgpu.h"
 
 int amdgpu_sa_bo_manager_init(struct amdgpu_device *adev,
 			      struct amdgpu_sa_manager *sa_manager,
-			      unsigned size, u32 align, u32 domain)
+			      unsigned int size, u32 suballoc_align, u32 domain)
 {
-	int i, r;
-
-	init_waitqueue_head(&sa_manager->wq);
-	sa_manager->bo = NULL;
-	sa_manager->size = size;
-	sa_manager->domain = domain;
-	sa_manager->align = align;
-	sa_manager->hole = &sa_manager->olist;
-	INIT_LIST_HEAD(&sa_manager->olist);
-	for (i = 0; i < AMDGPU_SA_NUM_FENCE_LISTS; ++i)
-		INIT_LIST_HEAD(&sa_manager->flist[i]);
+	int r;
 
-	r = amdgpu_bo_create(adev, size, align, true, domain,
-			     0, NULL, NULL, 0, &sa_manager->bo);
+	r = amdgpu_bo_create_kernel(adev, size, AMDGPU_GPU_PAGE_SIZE, domain,
+				    &sa_manager->bo, &sa_manager->gpu_addr,
+				    &sa_manager->cpu_ptr);
 	if (r) {
 		dev_err(adev->dev, "(%d) failed to allocate bo for manager\n", r);
 		return r;
 	}
 
+	memset(sa_manager->cpu_ptr, 0, size);
+	drm_suballoc_manager_init(&sa_manager->base, size, suballoc_align);
 	return r;
 }
 
 void amdgpu_sa_bo_manager_fini(struct amdgpu_device *adev,
 			       struct amdgpu_sa_manager *sa_manager)
 {
-	struct amdgpu_sa_bo *sa_bo, *tmp;
-
-	if (!list_empty(&sa_manager->olist)) {
-		sa_manager->hole = &sa_manager->olist,
-		amdgpu_sa_bo_try_free(sa_manager);
-		if (!list_empty(&sa_manager->olist)) {
-			dev_err(adev->dev, "sa_manager is not empty, clearing anyway\n");
-		}
-	}
-	list_for_each_entry_safe(sa_bo, tmp, &sa_manager->olist, olist) {
-		amdgpu_sa_bo_remove_locked(sa_bo);
-	}
-	amdgpu_bo_unref(&sa_manager->bo);
-	sa_manager->size = 0;
-}
-
-int amdgpu_sa_bo_manager_start(struct amdgpu_device *adev,
-			       struct amdgpu_sa_manager *sa_manager)
-{
-	int r;
-
-	if (sa_manager->bo == NULL) {
-		dev_err(adev->dev, "no bo for sa manager\n");
-		return -EINVAL;
-	}
-
-	/* map the buffer */
-	r = amdgpu_bo_reserve(sa_manager->bo, false);
-	if (r) {
-		dev_err(adev->dev, "(%d) failed to reserve manager bo\n", r);
-		return r;
-	}
-	r = amdgpu_bo_pin(sa_manager->bo, sa_manager->domain, &sa_manager->gpu_addr);
-	if (r) {
-		amdgpu_bo_unreserve(sa_manager->bo);
-		dev_err(adev->dev, "(%d) failed to pin manager bo\n", r);
-		return r;
-	}
-	r = amdgpu_bo_kmap(sa_manager->bo, &sa_manager->cpu_ptr);
-	memset(sa_manager->cpu_ptr, 0, sa_manager->size);
-	amdgpu_bo_unreserve(sa_manager->bo);
-	return r;
-}
-
-int amdgpu_sa_bo_manager_suspend(struct amdgpu_device *adev,
-				 struct amdgpu_sa_manager *sa_manager)
-{
-	int r;
-
 	if (sa_manager->bo == NULL) {
 		dev_err(adev->dev, "no bo for sa manager\n");
-		return -EINVAL;
-	}
-
-	r = amdgpu_bo_reserve(sa_manager->bo, true);
-	if (!r) {
-		amdgpu_bo_kunmap(sa_manager->bo);
-		amdgpu_bo_unpin(sa_manager->bo);
-		amdgpu_bo_unreserve(sa_manager->bo);
-	}
-	return r;
-}
-
-static void amdgpu_sa_bo_remove_locked(struct amdgpu_sa_bo *sa_bo)
-{
-	struct amdgpu_sa_manager *sa_manager = sa_bo->manager;
-	if (sa_manager->hole == &sa_bo->olist) {
-		sa_manager->hole = sa_bo->olist.prev;
-	}
-	list_del_init(&sa_bo->olist);
-	list_del_init(&sa_bo->flist);
-	dma_fence_put(sa_bo->fence);
-	kfree(sa_bo);
-}
-
-static void amdgpu_sa_bo_try_free(struct amdgpu_sa_manager *sa_manager)
-{
-	struct amdgpu_sa_bo *sa_bo, *tmp;
-
-	if (sa_manager->hole->next == &sa_manager->olist)
 		return;
-
-	sa_bo = list_entry(sa_manager->hole->next, struct amdgpu_sa_bo, olist);
-	list_for_each_entry_safe_from(sa_bo, tmp, &sa_manager->olist, olist) {
-		if (sa_bo->fence == NULL ||
-		    !dma_fence_is_signaled(sa_bo->fence)) {
-			return;
-		}
-		amdgpu_sa_bo_remove_locked(sa_bo);
-	}
-}
-
-static inline unsigned amdgpu_sa_bo_hole_soffset(struct amdgpu_sa_manager *sa_manager)
-{
-	struct list_head *hole = sa_manager->hole;
-
-	if (hole != &sa_manager->olist) {
-		return list_entry(hole, struct amdgpu_sa_bo, olist)->eoffset;
-	}
-	return 0;
-}
-
-static inline unsigned amdgpu_sa_bo_hole_eoffset(struct amdgpu_sa_manager *sa_manager)
-{
-	struct list_head *hole = sa_manager->hole;
-
-	if (hole->next != &sa_manager->olist) {
-		return list_entry(hole->next, struct amdgpu_sa_bo, olist)->soffset;
-	}
-	return sa_manager->size;
-}
-
-static bool amdgpu_sa_bo_try_alloc(struct amdgpu_sa_manager *sa_manager,
-				   struct amdgpu_sa_bo *sa_bo,
-				   unsigned size, unsigned align)
-{
-	unsigned soffset, eoffset, wasted;
-
-	soffset = amdgpu_sa_bo_hole_soffset(sa_manager);
-	eoffset = amdgpu_sa_bo_hole_eoffset(sa_manager);
-	wasted = (align - (soffset % align)) % align;
-
-	if ((eoffset - soffset) >= (size + wasted)) {
-		soffset += wasted;
-
-		sa_bo->manager = sa_manager;
-		sa_bo->soffset = soffset;
-		sa_bo->eoffset = soffset + size;
-		list_add(&sa_bo->olist, sa_manager->hole);
-		INIT_LIST_HEAD(&sa_bo->flist);
-		sa_manager->hole = &sa_bo->olist;
-		return true;
-	}
-	return false;
-}
-
-/**
- * amdgpu_sa_event - Check if we can stop waiting
- *
- * @sa_manager: pointer to the sa_manager
- * @size: number of bytes we want to allocate
- * @align: alignment we need to match
- *
- * Check if either there is a fence we can wait for or
- * enough free memory to satisfy the allocation directly
- */
-static bool amdgpu_sa_event(struct amdgpu_sa_manager *sa_manager,
-			    unsigned size, unsigned align)
-{
-	unsigned soffset, eoffset, wasted;
-	int i;
-
-	for (i = 0; i < AMDGPU_SA_NUM_FENCE_LISTS; ++i)
-		if (!list_empty(&sa_manager->flist[i]))
-			return true;
-
-	soffset = amdgpu_sa_bo_hole_soffset(sa_manager);
-	eoffset = amdgpu_sa_bo_hole_eoffset(sa_manager);
-	wasted = (align - (soffset % align)) % align;
-
-	if ((eoffset - soffset) >= (size + wasted)) {
-		return true;
-	}
-
-	return false;
-}
-
-static bool amdgpu_sa_bo_next_hole(struct amdgpu_sa_manager *sa_manager,
-				   struct dma_fence **fences,
-				   unsigned *tries)
-{
-	struct amdgpu_sa_bo *best_bo = NULL;
-	unsigned i, soffset, best, tmp;
-
-	/* if hole points to the end of the buffer */
-	if (sa_manager->hole->next == &sa_manager->olist) {
-		/* try again with its beginning */
-		sa_manager->hole = &sa_manager->olist;
-		return true;
-	}
-
-	soffset = amdgpu_sa_bo_hole_soffset(sa_manager);
-	/* to handle wrap around we add sa_manager->size */
-	best = sa_manager->size * 2;
-	/* go over all fence list and try to find the closest sa_bo
-	 * of the current last
-	 */
-	for (i = 0; i < AMDGPU_SA_NUM_FENCE_LISTS; ++i) {
-		struct amdgpu_sa_bo *sa_bo;
-
-		if (list_empty(&sa_manager->flist[i]))
-			continue;
-
-		sa_bo = list_first_entry(&sa_manager->flist[i],
-					 struct amdgpu_sa_bo, flist);
-
-		if (!dma_fence_is_signaled(sa_bo->fence)) {
-			fences[i] = sa_bo->fence;
-			continue;
-		}
-
-		/* limit the number of tries each ring gets */
-		if (tries[i] > 2) {
-			continue;
-		}
-
-		tmp = sa_bo->soffset;
-		if (tmp < soffset) {
-			/* wrap around, pretend it's after */
-			tmp += sa_manager->size;
-		}
-		tmp -= soffset;
-		if (tmp < best) {
-			/* this sa bo is the closest one */
-			best = tmp;
-			best_bo = sa_bo;
-		}
 	}
 
-	if (best_bo) {
-		uint32_t idx = best_bo->fence->context;
-
-		idx %= AMDGPU_SA_NUM_FENCE_LISTS;
-		++tries[idx];
-		sa_manager->hole = best_bo->olist.prev;
+	drm_suballoc_manager_fini(&sa_manager->base);
 
-		/* we knew that this one is signaled,
-		   so it's save to remote it */
-		amdgpu_sa_bo_remove_locked(best_bo);
-		return true;
-	}
-	return false;
+	amdgpu_bo_free_kernel(&sa_manager->bo, &sa_manager->gpu_addr, &sa_manager->cpu_ptr);
 }
 
 int amdgpu_sa_bo_new(struct amdgpu_sa_manager *sa_manager,
-		     struct amdgpu_sa_bo **sa_bo,
-		     unsigned size, unsigned align)
+		     struct drm_suballoc **sa_bo,
+		     unsigned int size)
 {
-	struct dma_fence *fences[AMDGPU_SA_NUM_FENCE_LISTS];
-	unsigned tries[AMDGPU_SA_NUM_FENCE_LISTS];
-	unsigned count;
-	int i, r;
-	signed long t;
-
-	if (WARN_ON_ONCE(align > sa_manager->align))
-		return -EINVAL;
-
-	if (WARN_ON_ONCE(size > sa_manager->size))
-		return -EINVAL;
-
-	*sa_bo = kmalloc(sizeof(struct amdgpu_sa_bo), GFP_KERNEL);
-	if (!(*sa_bo))
-		return -ENOMEM;
-	(*sa_bo)->manager = sa_manager;
-	(*sa_bo)->fence = NULL;
-	INIT_LIST_HEAD(&(*sa_bo)->olist);
-	INIT_LIST_HEAD(&(*sa_bo)->flist);
-
-	spin_lock(&sa_manager->wq.lock);
-	do {
-		for (i = 0; i < AMDGPU_SA_NUM_FENCE_LISTS; ++i) {
-			fences[i] = NULL;
-			tries[i] = 0;
-		}
-
-		do {
-			amdgpu_sa_bo_try_free(sa_manager);
-
-			if (amdgpu_sa_bo_try_alloc(sa_manager, *sa_bo,
-						   size, align)) {
-				spin_unlock(&sa_manager->wq.lock);
-				return 0;
-			}
+	struct drm_suballoc *sa = drm_suballoc_new(&sa_manager->base, size,
+						   GFP_KERNEL, false, 0);
 
-			/* see if we can skip over some allocations */
-		} while (amdgpu_sa_bo_next_hole(sa_manager, fences, tries));
+	if (IS_ERR(sa)) {
+		*sa_bo = NULL;
 
-		for (i = 0, count = 0; i < AMDGPU_SA_NUM_FENCE_LISTS; ++i)
-			if (fences[i])
-				fences[count++] = dma_fence_get(fences[i]);
-
-		if (count) {
-			spin_unlock(&sa_manager->wq.lock);
-			t = dma_fence_wait_any_timeout(fences, count, false,
-						       MAX_SCHEDULE_TIMEOUT,
-						       NULL);
-			for (i = 0; i < count; ++i)
-				dma_fence_put(fences[i]);
-
-			r = (t > 0) ? 0 : t;
-			spin_lock(&sa_manager->wq.lock);
-		} else {
-			/* if we have nothing to wait for block */
-			r = wait_event_interruptible_locked(
-				sa_manager->wq,
-				amdgpu_sa_event(sa_manager, size, align)
-			);
-		}
-
-	} while (!r);
+		return PTR_ERR(sa);
+	}
 
-	spin_unlock(&sa_manager->wq.lock);
-	kfree(*sa_bo);
-	*sa_bo = NULL;
-	return r;
+	*sa_bo = sa;
+	return 0;
 }
 
-void amdgpu_sa_bo_free(struct amdgpu_device *adev, struct amdgpu_sa_bo **sa_bo,
-		       struct dma_fence *fence)
+void amdgpu_sa_bo_free(struct drm_suballoc **sa_bo, struct dma_fence *fence)
 {
-	struct amdgpu_sa_manager *sa_manager;
-
 	if (sa_bo == NULL || *sa_bo == NULL) {
 		return;
 	}
 
-	sa_manager = (*sa_bo)->manager;
-	spin_lock(&sa_manager->wq.lock);
-	if (fence && !dma_fence_is_signaled(fence)) {
-		uint32_t idx;
-
-		(*sa_bo)->fence = dma_fence_get(fence);
-		idx = fence->context % AMDGPU_SA_NUM_FENCE_LISTS;
-		list_add_tail(&(*sa_bo)->flist, &sa_manager->flist[idx]);
-	} else {
-		amdgpu_sa_bo_remove_locked(*sa_bo);
-	}
-	wake_up_all_locked(&sa_manager->wq);
-	spin_unlock(&sa_manager->wq.lock);
+	drm_suballoc_free(*sa_bo, fence);
 	*sa_bo = NULL;
 }
 
@@ -413,26 +108,8 @@ void amdgpu_sa_bo_free(struct amdgpu_device *adev, struct amdgpu_sa_bo **sa_bo,
 void amdgpu_sa_bo_dump_debug_info(struct amdgpu_sa_manager *sa_manager,
 				  struct seq_file *m)
 {
-	struct amdgpu_sa_bo *i;
+	struct drm_printer p = drm_seq_file_printer(m);
 
-	spin_lock(&sa_manager->wq.lock);
-	list_for_each_entry(i, &sa_manager->olist, olist) {
-		uint64_t soffset = i->soffset + sa_manager->gpu_addr;
-		uint64_t eoffset = i->eoffset + sa_manager->gpu_addr;
-		if (&i->olist == sa_manager->hole) {
-			seq_printf(m, ">");
-		} else {
-			seq_printf(m, " ");
-		}
-		seq_printf(m, "[0x%010llx 0x%010llx] size %8lld",
-			   soffset, eoffset, eoffset - soffset);
-
-		if (i->fence)
-			seq_printf(m, " protected by 0x%08x on context %llu",
-				   i->fence->seqno, i->fence->context);
-
-		seq_printf(m, "\n");
-	}
-	spin_unlock(&sa_manager->wq.lock);
+	drm_suballoc_dump_debug_info(&sa_manager->base, &p, sa_manager->gpu_addr);
 }
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sched.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sched.c
new file mode 100644
index 000000000000..341beec59537
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sched.c
@@ -0,0 +1,131 @@
+/*
+ * Copyright 2017 Valve Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Andres Rodriguez <andresx7@gmail.com>
+ */
+
+#include <linux/file.h>
+#include <linux/pid.h>
+
+#include <drm/amdgpu_drm.h>
+
+#include "amdgpu.h"
+#include "amdgpu_sched.h"
+#include "amdgpu_vm.h"
+
+static int amdgpu_sched_process_priority_override(struct amdgpu_device *adev,
+						  int fd,
+						  int32_t priority)
+{
+	CLASS(fd, f)(fd);
+	struct amdgpu_fpriv *fpriv;
+	struct amdgpu_ctx_mgr *mgr;
+	struct amdgpu_ctx *ctx;
+	uint32_t id;
+	int r;
+
+	if (fd_empty(f))
+		return -EINVAL;
+
+	r = amdgpu_file_to_fpriv(fd_file(f), &fpriv);
+	if (r)
+		return r;
+
+	mgr = &fpriv->ctx_mgr;
+	mutex_lock(&mgr->lock);
+	idr_for_each_entry(&mgr->ctx_handles, ctx, id)
+		amdgpu_ctx_priority_override(ctx, priority);
+	mutex_unlock(&mgr->lock);
+
+	return 0;
+}
+
+static int amdgpu_sched_context_priority_override(struct amdgpu_device *adev,
+						  int fd,
+						  unsigned ctx_id,
+						  int32_t priority)
+{
+	CLASS(fd, f)(fd);
+	struct amdgpu_fpriv *fpriv;
+	struct amdgpu_ctx *ctx;
+	int r;
+
+	if (fd_empty(f))
+		return -EINVAL;
+
+	r = amdgpu_file_to_fpriv(fd_file(f), &fpriv);
+	if (r)
+		return r;
+
+	ctx = amdgpu_ctx_get(fpriv, ctx_id);
+
+	if (!ctx)
+		return -EINVAL;
+
+	amdgpu_ctx_priority_override(ctx, priority);
+	amdgpu_ctx_put(ctx);
+	return 0;
+}
+
+int amdgpu_sched_ioctl(struct drm_device *dev, void *data,
+		       struct drm_file *filp)
+{
+	union drm_amdgpu_sched *args = data;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	int r;
+
+	/* First check the op, then the op's argument.
+	 */
+	switch (args->in.op) {
+	case AMDGPU_SCHED_OP_PROCESS_PRIORITY_OVERRIDE:
+	case AMDGPU_SCHED_OP_CONTEXT_PRIORITY_OVERRIDE:
+		break;
+	default:
+		DRM_ERROR("Invalid sched op specified: %d\n", args->in.op);
+		return -EINVAL;
+	}
+
+	if (!amdgpu_ctx_priority_is_valid(args->in.priority)) {
+		WARN(1, "Invalid context priority %d\n", args->in.priority);
+		return -EINVAL;
+	}
+
+	switch (args->in.op) {
+	case AMDGPU_SCHED_OP_PROCESS_PRIORITY_OVERRIDE:
+		r = amdgpu_sched_process_priority_override(adev,
+							   args->in.fd,
+							   args->in.priority);
+		break;
+	case AMDGPU_SCHED_OP_CONTEXT_PRIORITY_OVERRIDE:
+		r = amdgpu_sched_context_priority_override(adev,
+							   args->in.fd,
+							   args->in.ctx_id,
+							   args->in.priority);
+		break;
+	default:
+		/* Impossible.
+		 */
+		r = -EINVAL;
+		break;
+	}
+
+	return r;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sched.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_sched.h
new file mode 100644
index 000000000000..67e5b2472f6a
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sched.h
@@ -0,0 +1,38 @@
+/*
+ * Copyright 2017 Valve Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Andres Rodriguez <andresx7@gmail.com>
+ */
+
+#ifndef __AMDGPU_SCHED_H__
+#define __AMDGPU_SCHED_H__
+
+enum drm_sched_priority;
+
+struct drm_device;
+struct drm_file;
+
+int amdgpu_to_sched_priority(int amdgpu_priority,
+			     enum drm_sched_priority *prio);
+int amdgpu_sched_ioctl(struct drm_device *dev, void *data,
+		       struct drm_file *filp);
+
+#endif // __AMDGPU_SCHED_H__
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c
new file mode 100644
index 000000000000..8b8a04138711
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c
@@ -0,0 +1,611 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include "amdgpu.h"
+#include "amdgpu_sdma.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_reset.h"
+#include "gc/gc_10_1_0_offset.h"
+#include "gc/gc_10_3_0_sh_mask.h"
+
+#define AMDGPU_CSA_SDMA_SIZE 64
+/* SDMA CSA reside in the 3rd page of CSA */
+#define AMDGPU_CSA_SDMA_OFFSET (4096 * 2)
+
+/*
+ * GPU SDMA IP block helpers function.
+ */
+
+struct amdgpu_sdma_instance *amdgpu_sdma_get_instance_from_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		if (ring == &adev->sdma.instance[i].ring ||
+		    ring == &adev->sdma.instance[i].page)
+			return &adev->sdma.instance[i];
+
+	return NULL;
+}
+
+int amdgpu_sdma_get_index_from_ring(struct amdgpu_ring *ring, uint32_t *index)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (ring == &adev->sdma.instance[i].ring ||
+			ring == &adev->sdma.instance[i].page) {
+			*index = i;
+			return 0;
+		}
+	}
+
+	return -EINVAL;
+}
+
+uint64_t amdgpu_sdma_get_csa_mc_addr(struct amdgpu_ring *ring,
+				     unsigned int vmid)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint64_t csa_mc_addr;
+	uint32_t index = 0;
+	int r;
+
+	/* don't enable OS preemption on SDMA under SRIOV */
+	if (amdgpu_sriov_vf(adev) || vmid == 0 || !adev->gfx.mcbp)
+		return 0;
+
+	r = amdgpu_sdma_get_index_from_ring(ring, &index);
+
+	if (r || index > 31)
+		csa_mc_addr = 0;
+	else
+		csa_mc_addr = amdgpu_csa_vaddr(adev) +
+			AMDGPU_CSA_SDMA_OFFSET +
+			index * AMDGPU_CSA_SDMA_SIZE;
+
+	return csa_mc_addr;
+}
+
+int amdgpu_sdma_ras_late_init(struct amdgpu_device *adev,
+			      struct ras_common_if *ras_block)
+{
+	int r, i;
+
+	r = amdgpu_ras_block_late_init(adev, ras_block);
+	if (r)
+		return r;
+
+	if (amdgpu_ras_is_supported(adev, ras_block->block)) {
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			r = amdgpu_irq_get(adev, &adev->sdma.ecc_irq,
+				AMDGPU_SDMA_IRQ_INSTANCE0 + i);
+			if (r)
+				goto late_fini;
+		}
+	}
+
+	return 0;
+
+late_fini:
+	amdgpu_ras_block_late_fini(adev, ras_block);
+	return r;
+}
+
+int amdgpu_sdma_process_ras_data_cb(struct amdgpu_device *adev,
+		void *err_data,
+		struct amdgpu_iv_entry *entry)
+{
+	kgd2kfd_set_sram_ecc_flag(adev->kfd.dev);
+
+	if (amdgpu_sriov_vf(adev))
+		return AMDGPU_RAS_SUCCESS;
+
+	amdgpu_ras_reset_gpu(adev);
+
+	return AMDGPU_RAS_SUCCESS;
+}
+
+int amdgpu_sdma_process_ecc_irq(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	struct ras_common_if *ras_if = adev->sdma.ras_if;
+	struct ras_dispatch_if ih_data = {
+		.entry = entry,
+	};
+
+	if (!ras_if)
+		return 0;
+
+	ih_data.head = *ras_if;
+
+	amdgpu_ras_interrupt_dispatch(adev, &ih_data);
+	return 0;
+}
+
+static int amdgpu_sdma_init_inst_ctx(struct amdgpu_sdma_instance *sdma_inst)
+{
+	uint16_t version_major;
+	const struct common_firmware_header *header = NULL;
+	const struct sdma_firmware_header_v1_0 *hdr;
+	const struct sdma_firmware_header_v2_0 *hdr_v2;
+	const struct sdma_firmware_header_v3_0 *hdr_v3;
+
+	header = (const struct common_firmware_header *)
+		sdma_inst->fw->data;
+	version_major = le16_to_cpu(header->header_version_major);
+
+	switch (version_major) {
+	case 1:
+		hdr = (const struct sdma_firmware_header_v1_0 *)sdma_inst->fw->data;
+		sdma_inst->fw_version = le32_to_cpu(hdr->header.ucode_version);
+		sdma_inst->feature_version = le32_to_cpu(hdr->ucode_feature_version);
+		break;
+	case 2:
+		hdr_v2 = (const struct sdma_firmware_header_v2_0 *)sdma_inst->fw->data;
+		sdma_inst->fw_version = le32_to_cpu(hdr_v2->header.ucode_version);
+		sdma_inst->feature_version = le32_to_cpu(hdr_v2->ucode_feature_version);
+		break;
+	case 3:
+		hdr_v3 = (const struct sdma_firmware_header_v3_0 *)sdma_inst->fw->data;
+		sdma_inst->fw_version = le32_to_cpu(hdr_v3->header.ucode_version);
+		sdma_inst->feature_version = le32_to_cpu(hdr_v3->ucode_feature_version);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	if (sdma_inst->feature_version >= 20)
+		sdma_inst->burst_nop = true;
+
+	return 0;
+}
+
+void amdgpu_sdma_destroy_inst_ctx(struct amdgpu_device *adev,
+				  bool duplicate)
+{
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		amdgpu_ucode_release(&adev->sdma.instance[i].fw);
+		if (duplicate)
+			break;
+	}
+
+	memset((void *)adev->sdma.instance, 0,
+	       sizeof(struct amdgpu_sdma_instance) * AMDGPU_MAX_SDMA_INSTANCES);
+}
+
+int amdgpu_sdma_init_microcode(struct amdgpu_device *adev,
+			       u32 instance, bool duplicate)
+{
+	struct amdgpu_firmware_info *info = NULL;
+	const struct common_firmware_header *header = NULL;
+	int err, i;
+	const struct sdma_firmware_header_v2_0 *sdma_hdr;
+	const struct sdma_firmware_header_v3_0 *sdma_hv3;
+	uint16_t version_major;
+	char ucode_prefix[30];
+
+	amdgpu_ucode_ip_version_decode(adev, SDMA0_HWIP, ucode_prefix, sizeof(ucode_prefix));
+	if (instance == 0)
+		err = amdgpu_ucode_request(adev, &adev->sdma.instance[instance].fw,
+					   AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s.bin", ucode_prefix);
+	else
+		err = amdgpu_ucode_request(adev, &adev->sdma.instance[instance].fw,
+					   AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s%d.bin", ucode_prefix, instance);
+	if (err)
+		goto out;
+
+	header = (const struct common_firmware_header *)
+		adev->sdma.instance[instance].fw->data;
+	version_major = le16_to_cpu(header->header_version_major);
+
+	if ((duplicate && instance) || (!duplicate && version_major > 1)) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	err = amdgpu_sdma_init_inst_ctx(&adev->sdma.instance[instance]);
+	if (err)
+		goto out;
+
+	if (duplicate) {
+		for (i = 1; i < adev->sdma.num_instances; i++)
+			memcpy((void *)&adev->sdma.instance[i],
+			       (void *)&adev->sdma.instance[0],
+			       sizeof(struct amdgpu_sdma_instance));
+	}
+
+	DRM_DEBUG("psp_load == '%s'\n",
+		  adev->firmware.load_type == AMDGPU_FW_LOAD_PSP ? "true" : "false");
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		switch (version_major) {
+		case 1:
+			for (i = 0; i < adev->sdma.num_instances; i++) {
+				if (!duplicate && (instance != i))
+					continue;
+				else {
+					/* Use a single copy per SDMA firmware type. PSP uses the same instance for all
+					 * groups of SDMAs */
+					if ((amdgpu_ip_version(adev, SDMA0_HWIP, 0) ==
+						IP_VERSION(4, 4, 2) ||
+					     amdgpu_ip_version(adev, SDMA0_HWIP, 0) ==
+						IP_VERSION(4, 4, 4) ||
+					     amdgpu_ip_version(adev, SDMA0_HWIP, 0) ==
+						IP_VERSION(4, 4, 5)) &&
+					    adev->firmware.load_type ==
+						AMDGPU_FW_LOAD_PSP &&
+					    adev->sdma.num_inst_per_aid == i) {
+						break;
+					}
+					info = &adev->firmware.ucode[AMDGPU_UCODE_ID_SDMA0 + i];
+					info->ucode_id = AMDGPU_UCODE_ID_SDMA0 + i;
+					info->fw = adev->sdma.instance[i].fw;
+					adev->firmware.fw_size +=
+						ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+				}
+			}
+			break;
+		case 2:
+			sdma_hdr = (const struct sdma_firmware_header_v2_0 *)
+				adev->sdma.instance[0].fw->data;
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_SDMA_UCODE_TH0];
+			info->ucode_id = AMDGPU_UCODE_ID_SDMA_UCODE_TH0;
+			info->fw = adev->sdma.instance[0].fw;
+			adev->firmware.fw_size +=
+				ALIGN(le32_to_cpu(sdma_hdr->ctx_ucode_size_bytes), PAGE_SIZE);
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_SDMA_UCODE_TH1];
+			info->ucode_id = AMDGPU_UCODE_ID_SDMA_UCODE_TH1;
+			info->fw = adev->sdma.instance[0].fw;
+			adev->firmware.fw_size +=
+				ALIGN(le32_to_cpu(sdma_hdr->ctl_ucode_size_bytes), PAGE_SIZE);
+			break;
+		case 3:
+			sdma_hv3 = (const struct sdma_firmware_header_v3_0 *)
+				adev->sdma.instance[0].fw->data;
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_SDMA_RS64];
+			info->ucode_id = AMDGPU_UCODE_ID_SDMA_RS64;
+			info->fw = adev->sdma.instance[0].fw;
+			adev->firmware.fw_size +=
+				ALIGN(le32_to_cpu(sdma_hv3->ucode_size_bytes), PAGE_SIZE);
+			break;
+		default:
+			err = -EINVAL;
+		}
+	}
+
+out:
+	if (err)
+		amdgpu_sdma_destroy_inst_ctx(adev, duplicate);
+	return err;
+}
+
+int amdgpu_sdma_ras_sw_init(struct amdgpu_device *adev)
+{
+	int err = 0;
+	struct amdgpu_sdma_ras *ras = NULL;
+
+	/* adev->sdma.ras is NULL, which means sdma does not
+	 * support ras function, then do nothing here.
+	 */
+	if (!adev->sdma.ras)
+		return 0;
+
+	ras = adev->sdma.ras;
+
+	err = amdgpu_ras_register_ras_block(adev, &ras->ras_block);
+	if (err) {
+		dev_err(adev->dev, "Failed to register sdma ras block!\n");
+		return err;
+	}
+
+	strcpy(ras->ras_block.ras_comm.name, "sdma");
+	ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__SDMA;
+	ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
+	adev->sdma.ras_if = &ras->ras_block.ras_comm;
+
+	/* If not define special ras_late_init function, use default ras_late_init */
+	if (!ras->ras_block.ras_late_init)
+		ras->ras_block.ras_late_init = amdgpu_sdma_ras_late_init;
+
+	/* If not defined special ras_cb function, use default ras_cb */
+	if (!ras->ras_block.ras_cb)
+		ras->ras_block.ras_cb = amdgpu_sdma_process_ras_data_cb;
+
+	return 0;
+}
+
+/*
+ * debugfs for to enable/disable sdma job submission to specific core.
+ */
+#if defined(CONFIG_DEBUG_FS)
+static int amdgpu_debugfs_sdma_sched_mask_set(void *data, u64 val)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)data;
+	u64 i, num_ring;
+	u64 mask = 0;
+	struct amdgpu_ring *ring, *page = NULL;
+
+	if (!adev)
+		return -ENODEV;
+
+	/* Determine the number of rings per SDMA instance
+	 * (1 for sdma gfx ring, 2 if page queue exists)
+	 */
+	if (adev->sdma.has_page_queue)
+		num_ring = 2;
+	else
+		num_ring = 1;
+
+	/* Calculate the maximum possible mask value
+	 * based on the number of SDMA instances and rings
+	*/
+	mask = BIT_ULL(adev->sdma.num_instances * num_ring) - 1;
+
+	if ((val & mask) == 0)
+		return -EINVAL;
+
+	for (i = 0; i < adev->sdma.num_instances; ++i) {
+		ring = &adev->sdma.instance[i].ring;
+		if (adev->sdma.has_page_queue)
+			page = &adev->sdma.instance[i].page;
+		if (val & BIT_ULL(i * num_ring))
+			ring->sched.ready = true;
+		else
+			ring->sched.ready = false;
+
+		if (page) {
+			if (val & BIT_ULL(i * num_ring + 1))
+				page->sched.ready = true;
+			else
+				page->sched.ready = false;
+		}
+	}
+	/* publish sched.ready flag update effective immediately across smp */
+	smp_rmb();
+	return 0;
+}
+
+static int amdgpu_debugfs_sdma_sched_mask_get(void *data, u64 *val)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)data;
+	u64 i, num_ring;
+	u64 mask = 0;
+	struct amdgpu_ring *ring, *page = NULL;
+
+	if (!adev)
+		return -ENODEV;
+
+	/* Determine the number of rings per SDMA instance
+	 * (1 for sdma gfx ring, 2 if page queue exists)
+	 */
+	if (adev->sdma.has_page_queue)
+		num_ring = 2;
+	else
+		num_ring = 1;
+
+	for (i = 0; i < adev->sdma.num_instances; ++i) {
+		ring = &adev->sdma.instance[i].ring;
+		if (adev->sdma.has_page_queue)
+			page = &adev->sdma.instance[i].page;
+
+		if (ring->sched.ready)
+			mask |= BIT_ULL(i * num_ring);
+		else
+			mask &= ~BIT_ULL(i * num_ring);
+
+		if (page) {
+			if (page->sched.ready)
+				mask |= BIT_ULL(i * num_ring + 1);
+			else
+				mask &= ~BIT_ULL(i * num_ring + 1);
+		}
+	}
+
+	*val = mask;
+	return 0;
+}
+
+DEFINE_DEBUGFS_ATTRIBUTE(amdgpu_debugfs_sdma_sched_mask_fops,
+			 amdgpu_debugfs_sdma_sched_mask_get,
+			 amdgpu_debugfs_sdma_sched_mask_set, "%llx\n");
+
+#endif
+
+void amdgpu_debugfs_sdma_sched_mask_init(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	struct drm_minor *minor = adev_to_drm(adev)->primary;
+	struct dentry *root = minor->debugfs_root;
+	char name[32];
+
+	if (!(adev->sdma.num_instances > 1))
+		return;
+	sprintf(name, "amdgpu_sdma_sched_mask");
+	debugfs_create_file(name, 0600, root, adev,
+			    &amdgpu_debugfs_sdma_sched_mask_fops);
+#endif
+}
+
+static ssize_t amdgpu_get_sdma_reset_mask(struct device *dev,
+						struct device_attribute *attr,
+						char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	if (!adev)
+		return -ENODEV;
+
+	return amdgpu_show_reset_mask(buf, adev->sdma.supported_reset);
+}
+
+static DEVICE_ATTR(sdma_reset_mask, 0444,
+		   amdgpu_get_sdma_reset_mask, NULL);
+
+int amdgpu_sdma_sysfs_reset_mask_init(struct amdgpu_device *adev)
+{
+	int r = 0;
+
+	if (!amdgpu_gpu_recovery)
+		return r;
+
+	if (adev->sdma.num_instances) {
+		r = device_create_file(adev->dev, &dev_attr_sdma_reset_mask);
+		if (r)
+			return r;
+	}
+
+	return r;
+}
+
+void amdgpu_sdma_sysfs_reset_mask_fini(struct amdgpu_device *adev)
+{
+	if (!amdgpu_gpu_recovery)
+		return;
+
+	if (adev->dev->kobj.sd) {
+		if (adev->sdma.num_instances)
+			device_remove_file(adev->dev, &dev_attr_sdma_reset_mask);
+	}
+}
+
+struct amdgpu_ring *amdgpu_sdma_get_shared_ring(struct amdgpu_device *adev, struct amdgpu_ring *ring)
+{
+	if (adev->sdma.has_page_queue &&
+	    (ring->me < adev->sdma.num_instances) &&
+	    (ring == &adev->sdma.instance[ring->me].ring))
+		return &adev->sdma.instance[ring->me].page;
+	else
+		return NULL;
+}
+
+/**
+* amdgpu_sdma_is_shared_inv_eng - Check if a ring is an SDMA ring that shares a VM invalidation engine
+* @adev: Pointer to the AMDGPU device structure
+* @ring: Pointer to the ring structure to check
+*
+* This function checks if the given ring is an SDMA ring that shares a VM invalidation engine.
+* It returns true if the ring is such an SDMA ring, false otherwise.
+*/
+bool amdgpu_sdma_is_shared_inv_eng(struct amdgpu_device *adev, struct amdgpu_ring *ring)
+{
+	int i = ring->me;
+
+	if (!adev->sdma.has_page_queue || i >= adev->sdma.num_instances)
+		return false;
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0))
+		return (ring == &adev->sdma.instance[i].page);
+	else
+		return false;
+}
+
+static int amdgpu_sdma_soft_reset(struct amdgpu_device *adev, u32 instance_id)
+{
+	struct amdgpu_sdma_instance *sdma_instance = &adev->sdma.instance[instance_id];
+
+	if (sdma_instance->funcs->soft_reset_kernel_queue)
+		return sdma_instance->funcs->soft_reset_kernel_queue(adev, instance_id);
+
+	return -EOPNOTSUPP;
+}
+
+/**
+ * amdgpu_sdma_reset_engine - Reset a specific SDMA engine
+ * @adev: Pointer to the AMDGPU device
+ * @instance_id: Logical ID of the SDMA engine instance to reset
+ * @caller_handles_kernel_queues: Skip kernel queue processing. Caller
+ * will handle it.
+ *
+ * Returns: 0 on success, or a negative error code on failure.
+ */
+int amdgpu_sdma_reset_engine(struct amdgpu_device *adev, uint32_t instance_id,
+			     bool caller_handles_kernel_queues)
+{
+	int ret = 0;
+	struct amdgpu_sdma_instance *sdma_instance = &adev->sdma.instance[instance_id];
+	struct amdgpu_ring *gfx_ring = &sdma_instance->ring;
+	struct amdgpu_ring *page_ring = &sdma_instance->page;
+
+	mutex_lock(&sdma_instance->engine_reset_mutex);
+
+	if (!caller_handles_kernel_queues) {
+		/* Stop the scheduler's work queue for the GFX and page rings if they are running.
+		 * This ensures that no new tasks are submitted to the queues while
+		 * the reset is in progress.
+		 */
+		drm_sched_wqueue_stop(&gfx_ring->sched);
+
+		if (adev->sdma.has_page_queue)
+			drm_sched_wqueue_stop(&page_ring->sched);
+	}
+
+	if (sdma_instance->funcs->stop_kernel_queue) {
+		sdma_instance->funcs->stop_kernel_queue(gfx_ring);
+		if (adev->sdma.has_page_queue)
+			sdma_instance->funcs->stop_kernel_queue(page_ring);
+	}
+
+	/* Perform the SDMA reset for the specified instance */
+	ret = amdgpu_sdma_soft_reset(adev, instance_id);
+	if (ret) {
+		dev_err(adev->dev, "Failed to reset SDMA logical instance %u\n", instance_id);
+		goto exit;
+	}
+
+	if (sdma_instance->funcs->start_kernel_queue) {
+		sdma_instance->funcs->start_kernel_queue(gfx_ring);
+		if (adev->sdma.has_page_queue)
+			sdma_instance->funcs->start_kernel_queue(page_ring);
+	}
+
+exit:
+	if (!caller_handles_kernel_queues) {
+		/* Restart the scheduler's work queue for the GFX and page rings
+		 * if they were stopped by this function. This allows new tasks
+		 * to be submitted to the queues after the reset is complete.
+		 */
+		if (!ret) {
+			amdgpu_fence_driver_force_completion(gfx_ring);
+			drm_sched_wqueue_start(&gfx_ring->sched);
+			if (adev->sdma.has_page_queue) {
+				amdgpu_fence_driver_force_completion(page_ring);
+				drm_sched_wqueue_start(&page_ring->sched);
+			}
+		}
+	}
+	mutex_unlock(&sdma_instance->engine_reset_mutex);
+
+	return ret;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h
new file mode 100644
index 000000000000..34311f32be4c
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h
@@ -0,0 +1,204 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_SDMA_H__
+#define __AMDGPU_SDMA_H__
+#include "amdgpu_ras.h"
+
+/* max number of IP instances */
+#define AMDGPU_MAX_SDMA_INSTANCES		16
+
+enum amdgpu_sdma_irq {
+	AMDGPU_SDMA_IRQ_INSTANCE0  = 0,
+	AMDGPU_SDMA_IRQ_INSTANCE1,
+	AMDGPU_SDMA_IRQ_INSTANCE2,
+	AMDGPU_SDMA_IRQ_INSTANCE3,
+	AMDGPU_SDMA_IRQ_INSTANCE4,
+	AMDGPU_SDMA_IRQ_INSTANCE5,
+	AMDGPU_SDMA_IRQ_INSTANCE6,
+	AMDGPU_SDMA_IRQ_INSTANCE7,
+	AMDGPU_SDMA_IRQ_INSTANCE8,
+	AMDGPU_SDMA_IRQ_INSTANCE9,
+	AMDGPU_SDMA_IRQ_INSTANCE10,
+	AMDGPU_SDMA_IRQ_INSTANCE11,
+	AMDGPU_SDMA_IRQ_INSTANCE12,
+	AMDGPU_SDMA_IRQ_INSTANCE13,
+	AMDGPU_SDMA_IRQ_INSTANCE14,
+	AMDGPU_SDMA_IRQ_INSTANCE15,
+	AMDGPU_SDMA_IRQ_LAST
+};
+
+#define NUM_SDMA(x) hweight32(x)
+
+struct amdgpu_sdma_funcs {
+	int (*stop_kernel_queue)(struct amdgpu_ring *ring);
+	int (*start_kernel_queue)(struct amdgpu_ring *ring);
+	int (*soft_reset_kernel_queue)(struct amdgpu_device *adev, u32 instance_id);
+};
+
+struct amdgpu_sdma_instance {
+	/* SDMA firmware */
+	const struct firmware	*fw;
+	uint32_t		fw_version;
+	uint32_t		feature_version;
+
+	struct amdgpu_ring	ring;
+	struct amdgpu_ring	page;
+	bool			burst_nop;
+	uint32_t		aid_id;
+
+	struct amdgpu_bo	*sdma_fw_obj;
+	uint64_t		sdma_fw_gpu_addr;
+	uint32_t		*sdma_fw_ptr;
+	struct mutex		engine_reset_mutex;
+	/* track guilty state of GFX and PAGE queues */
+	bool			gfx_guilty;
+	bool			page_guilty;
+	const struct amdgpu_sdma_funcs   *funcs;
+};
+
+enum amdgpu_sdma_ras_memory_id {
+	AMDGPU_SDMA_MBANK_DATA_BUF0 = 1,
+	AMDGPU_SDMA_MBANK_DATA_BUF1 = 2,
+	AMDGPU_SDMA_MBANK_DATA_BUF2 = 3,
+	AMDGPU_SDMA_MBANK_DATA_BUF3 = 4,
+	AMDGPU_SDMA_MBANK_DATA_BUF4 = 5,
+	AMDGPU_SDMA_MBANK_DATA_BUF5 = 6,
+	AMDGPU_SDMA_MBANK_DATA_BUF6 = 7,
+	AMDGPU_SDMA_MBANK_DATA_BUF7 = 8,
+	AMDGPU_SDMA_MBANK_DATA_BUF8 = 9,
+	AMDGPU_SDMA_MBANK_DATA_BUF9 = 10,
+	AMDGPU_SDMA_MBANK_DATA_BUF10 = 11,
+	AMDGPU_SDMA_MBANK_DATA_BUF11 = 12,
+	AMDGPU_SDMA_MBANK_DATA_BUF12 = 13,
+	AMDGPU_SDMA_MBANK_DATA_BUF13 = 14,
+	AMDGPU_SDMA_MBANK_DATA_BUF14 = 15,
+	AMDGPU_SDMA_MBANK_DATA_BUF15 = 16,
+	AMDGPU_SDMA_UCODE_BUF = 17,
+	AMDGPU_SDMA_RB_CMD_BUF = 18,
+	AMDGPU_SDMA_IB_CMD_BUF = 19,
+	AMDGPU_SDMA_UTCL1_RD_FIFO = 20,
+	AMDGPU_SDMA_UTCL1_RDBST_FIFO = 21,
+	AMDGPU_SDMA_UTCL1_WR_FIFO = 22,
+	AMDGPU_SDMA_DATA_LUT_FIFO = 23,
+	AMDGPU_SDMA_SPLIT_DAT_BUF = 24,
+	AMDGPU_SDMA_MEMORY_BLOCK_LAST,
+};
+
+struct amdgpu_sdma_ras {
+	struct amdgpu_ras_block_object ras_block;
+};
+
+struct amdgpu_sdma {
+	struct amdgpu_sdma_instance instance[AMDGPU_MAX_SDMA_INSTANCES];
+	struct amdgpu_irq_src	trap_irq;
+	struct amdgpu_irq_src	illegal_inst_irq;
+	struct amdgpu_irq_src	fence_irq;
+	struct amdgpu_irq_src	ecc_irq;
+	struct amdgpu_irq_src	vm_hole_irq;
+	struct amdgpu_irq_src	doorbell_invalid_irq;
+	struct amdgpu_irq_src	pool_timeout_irq;
+	struct amdgpu_irq_src	srbm_write_irq;
+	struct amdgpu_irq_src	ctxt_empty_irq;
+
+	int			num_instances;
+	uint32_t 		sdma_mask;
+	int			num_inst_per_aid;
+	uint32_t                    srbm_soft_reset;
+	bool			has_page_queue;
+	struct ras_common_if	*ras_if;
+	struct amdgpu_sdma_ras	*ras;
+	uint32_t		*ip_dump;
+	uint32_t 		supported_reset;
+	struct list_head	reset_callback_list;
+	bool			no_user_submission;
+	bool			disable_uq;
+};
+
+/*
+ * Provided by hw blocks that can move/clear data.  e.g., gfx or sdma
+ * But currently, we use sdma to move data.
+ */
+struct amdgpu_buffer_funcs {
+	/* maximum bytes in a single operation */
+	uint32_t	copy_max_bytes;
+
+	/* number of dw to reserve per operation */
+	unsigned	copy_num_dw;
+
+	/* used for buffer migration */
+	void (*emit_copy_buffer)(struct amdgpu_ib *ib,
+				 /* src addr in bytes */
+				 uint64_t src_offset,
+				 /* dst addr in bytes */
+				 uint64_t dst_offset,
+				 /* number of byte to transfer */
+				 uint32_t byte_count,
+				 uint32_t copy_flags);
+
+	/* maximum bytes in a single operation */
+	uint32_t	fill_max_bytes;
+
+	/* number of dw to reserve per operation */
+	unsigned	fill_num_dw;
+
+	/* used for buffer clearing */
+	void (*emit_fill_buffer)(struct amdgpu_ib *ib,
+				 /* value to write to memory */
+				 uint32_t src_data,
+				 /* dst addr in bytes */
+				 uint64_t dst_offset,
+				 /* number of byte to fill */
+				 uint32_t byte_count);
+};
+
+int amdgpu_sdma_reset_engine(struct amdgpu_device *adev, uint32_t instance_id,
+			     bool caller_handles_kernel_queues);
+
+#define amdgpu_emit_copy_buffer(adev, ib, s, d, b, t) (adev)->mman.buffer_funcs->emit_copy_buffer((ib),  (s), (d), (b), (t))
+#define amdgpu_emit_fill_buffer(adev, ib, s, d, b) (adev)->mman.buffer_funcs->emit_fill_buffer((ib), (s), (d), (b))
+
+struct amdgpu_sdma_instance *
+amdgpu_sdma_get_instance_from_ring(struct amdgpu_ring *ring);
+int amdgpu_sdma_get_index_from_ring(struct amdgpu_ring *ring, uint32_t *index);
+uint64_t amdgpu_sdma_get_csa_mc_addr(struct amdgpu_ring *ring, unsigned vmid);
+int amdgpu_sdma_ras_late_init(struct amdgpu_device *adev,
+			      struct ras_common_if *ras_block);
+int amdgpu_sdma_process_ras_data_cb(struct amdgpu_device *adev,
+		void *err_data,
+		struct amdgpu_iv_entry *entry);
+int amdgpu_sdma_process_ecc_irq(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry);
+int amdgpu_sdma_init_microcode(struct amdgpu_device *adev, u32 instance,
+			       bool duplicate);
+void amdgpu_sdma_destroy_inst_ctx(struct amdgpu_device *adev,
+        bool duplicate);
+int amdgpu_sdma_ras_sw_init(struct amdgpu_device *adev);
+void amdgpu_debugfs_sdma_sched_mask_init(struct amdgpu_device *adev);
+int amdgpu_sdma_sysfs_reset_mask_init(struct amdgpu_device *adev);
+void amdgpu_sdma_sysfs_reset_mask_fini(struct amdgpu_device *adev);
+bool amdgpu_sdma_is_shared_inv_eng(struct amdgpu_device *adev, struct amdgpu_ring *ring);
+struct amdgpu_ring *amdgpu_sdma_get_shared_ring(struct amdgpu_device *adev,
+	struct amdgpu_ring *ring);
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c
new file mode 100644
index 000000000000..41ebe690eeff
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c
@@ -0,0 +1,187 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ *
+ */
+#include <linux/debugfs.h>
+#include <linux/pm_runtime.h>
+
+#include "amdgpu.h"
+#include "amdgpu_securedisplay.h"
+
+/**
+ * DOC: AMDGPU SECUREDISPLAY debugfs test interface
+ *
+ * how to use?
+ * echo opcode <value> > <debugfs_dir>/dri/xxx/securedisplay_test
+ * eg. echo 1 > <debugfs_dir>/dri/xxx/securedisplay_test
+ * eg. echo 2 phy_id > <debugfs_dir>/dri/xxx/securedisplay_test
+ *
+ * opcode:
+ * 1：Query whether TA is responding used only for validation pupose
+ * 2: Send region of Interest and CRC value to I2C. (uint32)phy_id is
+ * send to determine which DIO scratch register should be used to get
+ * ROI and receive i2c_buf as the output.
+ *
+ * You can refer more detail from header file ta_securedisplay_if.h
+ *
+ */
+
+void psp_securedisplay_parse_resp_status(struct psp_context *psp,
+	enum ta_securedisplay_status status)
+{
+	switch (status) {
+	case TA_SECUREDISPLAY_STATUS__SUCCESS:
+		break;
+	case TA_SECUREDISPLAY_STATUS__GENERIC_FAILURE:
+		dev_err(psp->adev->dev, "Secure display: Generic Failure.");
+		break;
+	case TA_SECUREDISPLAY_STATUS__INVALID_PARAMETER:
+		dev_err(psp->adev->dev, "Secure display: Invalid Parameter.");
+		break;
+	case TA_SECUREDISPLAY_STATUS__NULL_POINTER:
+		dev_err(psp->adev->dev, "Secure display: Null Pointer.");
+		break;
+	case TA_SECUREDISPLAY_STATUS__I2C_WRITE_ERROR:
+		dev_err(psp->adev->dev, "Secure display: Failed to write to I2C.");
+		break;
+	case TA_SECUREDISPLAY_STATUS__READ_DIO_SCRATCH_ERROR:
+		dev_err(psp->adev->dev, "Secure display: Failed to Read DIO Scratch Register.");
+		break;
+	case TA_SECUREDISPLAY_STATUS__READ_CRC_ERROR:
+		dev_err(psp->adev->dev, "Secure display: Failed to Read CRC");
+		break;
+	case TA_SECUREDISPLAY_STATUS__I2C_INIT_ERROR:
+		dev_err(psp->adev->dev, "Secure display: Failed to initialize I2C.");
+		break;
+	default:
+		dev_err(psp->adev->dev, "Secure display: Failed to parse status: %d\n", status);
+	}
+}
+
+void psp_prep_securedisplay_cmd_buf(struct psp_context *psp, struct ta_securedisplay_cmd **cmd,
+	enum ta_securedisplay_command command_id)
+{
+	*cmd = (struct ta_securedisplay_cmd *)psp->securedisplay_context.context.mem_context.shared_buf;
+	memset(*cmd, 0, sizeof(struct ta_securedisplay_cmd));
+	(*cmd)->status = TA_SECUREDISPLAY_STATUS__GENERIC_FAILURE;
+	(*cmd)->cmd_id = command_id;
+}
+
+#if defined(CONFIG_DEBUG_FS)
+
+static ssize_t amdgpu_securedisplay_debugfs_write(struct file *f, const char __user *buf,
+		size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)file_inode(f)->i_private;
+	struct psp_context *psp = &adev->psp;
+	struct ta_securedisplay_cmd *securedisplay_cmd;
+	struct drm_device *dev = adev_to_drm(adev);
+	uint32_t phy_id;
+	uint32_t op;
+	char str[64];
+	int ret;
+
+	if (*pos || size > sizeof(str) - 1)
+		return -EINVAL;
+
+	memset(str,  0, sizeof(str));
+	ret = copy_from_user(str, buf, size);
+	if (ret)
+		return -EFAULT;
+
+	ret = pm_runtime_get_sync(dev->dev);
+	if (ret < 0) {
+		pm_runtime_put_autosuspend(dev->dev);
+		return ret;
+	}
+
+	if (size < 3)
+		sscanf(str, "%u ", &op);
+	else
+		sscanf(str, "%u %u", &op, &phy_id);
+
+	switch (op) {
+	case 1:
+		mutex_lock(&psp->securedisplay_context.mutex);
+		psp_prep_securedisplay_cmd_buf(psp, &securedisplay_cmd,
+			TA_SECUREDISPLAY_COMMAND__QUERY_TA);
+		ret = psp_securedisplay_invoke(psp, TA_SECUREDISPLAY_COMMAND__QUERY_TA);
+		if (!ret) {
+			if (securedisplay_cmd->status == TA_SECUREDISPLAY_STATUS__SUCCESS)
+				dev_info(adev->dev, "SECUREDISPLAY: query securedisplay TA ret is 0x%X\n",
+					securedisplay_cmd->securedisplay_out_message.query_ta.query_cmd_ret);
+			else
+				psp_securedisplay_parse_resp_status(psp, securedisplay_cmd->status);
+		}
+		mutex_unlock(&psp->securedisplay_context.mutex);
+		break;
+	case 2:
+		if (size < 3 || phy_id >= TA_SECUREDISPLAY_MAX_PHY) {
+			dev_err(adev->dev, "Invalid input: %s\n", str);
+			return -EINVAL;
+		}
+		mutex_lock(&psp->securedisplay_context.mutex);
+		psp_prep_securedisplay_cmd_buf(psp, &securedisplay_cmd,
+			TA_SECUREDISPLAY_COMMAND__SEND_ROI_CRC);
+		securedisplay_cmd->securedisplay_in_message.send_roi_crc.phy_id = phy_id;
+		ret = psp_securedisplay_invoke(psp, TA_SECUREDISPLAY_COMMAND__SEND_ROI_CRC);
+		if (!ret) {
+			if (securedisplay_cmd->status == TA_SECUREDISPLAY_STATUS__SUCCESS) {
+				dev_info(adev->dev, "SECUREDISPLAY: I2C buffer out put is: %*ph\n",
+					 TA_SECUREDISPLAY_I2C_BUFFER_SIZE,
+					 securedisplay_cmd->securedisplay_out_message.send_roi_crc.i2c_buf);
+			} else {
+				psp_securedisplay_parse_resp_status(psp, securedisplay_cmd->status);
+			}
+		}
+		mutex_unlock(&psp->securedisplay_context.mutex);
+		break;
+	default:
+		dev_err(adev->dev, "Invalid input: %s\n", str);
+	}
+
+	pm_runtime_mark_last_busy(dev->dev);
+	pm_runtime_put_autosuspend(dev->dev);
+
+	return size;
+}
+
+static const struct file_operations amdgpu_securedisplay_debugfs_ops = {
+	.owner = THIS_MODULE,
+	.read = NULL,
+	.write = amdgpu_securedisplay_debugfs_write,
+	.llseek = default_llseek
+};
+
+#endif
+
+void amdgpu_securedisplay_debugfs_init(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+
+	if (!adev->psp.securedisplay_context.context.initialized)
+		return;
+
+	debugfs_create_file("securedisplay_test", S_IWUSR, adev_to_drm(adev)->primary->debugfs_root,
+				adev, &amdgpu_securedisplay_debugfs_ops);
+#endif
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.h
new file mode 100644
index 000000000000..456ad68ed4b2
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.h
@@ -0,0 +1,36 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ *
+ */
+#ifndef _AMDGPU_SECUREDISPLAY_H
+#define _AMDGPU_SECUREDISPLAY_H
+
+#include "amdgpu.h"
+#include "ta_secureDisplay_if.h"
+
+void amdgpu_securedisplay_debugfs_init(struct amdgpu_device *adev);
+void psp_securedisplay_parse_resp_status(struct psp_context *psp,
+		enum ta_securedisplay_status status);
+void psp_prep_securedisplay_cmd_buf(struct psp_context *psp, struct ta_securedisplay_cmd **cmd,
+		enum ta_securedisplay_command command_id);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.c
new file mode 100644
index 000000000000..a0b479d5fff1
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.c
@@ -0,0 +1,262 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "amdgpu_seq64.h"
+
+#include <drm/drm_exec.h>
+
+/**
+ * DOC: amdgpu_seq64
+ *
+ * amdgpu_seq64 allocates a 64bit memory on each request in sequence order.
+ * seq64 driver is required for user queue fence memory allocation, TLB
+ * counters and VM updates. It has maximum count of 32768 64 bit slots.
+ */
+
+/**
+ * amdgpu_seq64_get_va_base - Get the seq64 va base address
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Returns:
+ * va base address on success
+ */
+static inline u64 amdgpu_seq64_get_va_base(struct amdgpu_device *adev)
+{
+	u64 addr = AMDGPU_VA_RESERVED_SEQ64_START(adev);
+
+	addr = amdgpu_gmc_sign_extend(addr);
+
+	return addr;
+}
+
+/**
+ * amdgpu_seq64_map - Map the seq64 memory to VM
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: vm pointer
+ * @bo_va: bo_va pointer
+ *
+ * Map the seq64 memory to the given VM.
+ *
+ * Returns:
+ * 0 on success or a negative error code on failure
+ */
+int amdgpu_seq64_map(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+		     struct amdgpu_bo_va **bo_va)
+{
+	struct amdgpu_bo *bo;
+	struct drm_exec exec;
+	u64 seq64_addr;
+	int r;
+
+	bo = adev->seq64.sbo;
+	if (!bo)
+		return -EINVAL;
+
+	drm_exec_init(&exec, DRM_EXEC_INTERRUPTIBLE_WAIT, 0);
+	drm_exec_until_all_locked(&exec) {
+		r = amdgpu_vm_lock_pd(vm, &exec, 0);
+		if (likely(!r))
+			r = drm_exec_lock_obj(&exec, &bo->tbo.base);
+		drm_exec_retry_on_contention(&exec);
+		if (unlikely(r))
+			goto error;
+	}
+
+	*bo_va = amdgpu_vm_bo_add(adev, vm, bo);
+	if (!*bo_va) {
+		r = -ENOMEM;
+		goto error;
+	}
+
+	seq64_addr = amdgpu_seq64_get_va_base(adev) & AMDGPU_GMC_HOLE_MASK;
+
+	r = amdgpu_vm_bo_map(adev, *bo_va, seq64_addr, 0,
+			     AMDGPU_VA_RESERVED_SEQ64_SIZE,
+			     AMDGPU_VM_PAGE_READABLE | AMDGPU_VM_MTYPE_UC);
+	if (r) {
+		DRM_ERROR("failed to do bo_map on userq sem, err=%d\n", r);
+		amdgpu_vm_bo_del(adev, *bo_va);
+		goto error;
+	}
+
+	r = amdgpu_vm_bo_update(adev, *bo_va, false);
+	if (r) {
+		DRM_ERROR("failed to do vm_bo_update on userq sem\n");
+		amdgpu_vm_bo_del(adev, *bo_va);
+		goto error;
+	}
+
+error:
+	drm_exec_fini(&exec);
+	return r;
+}
+
+/**
+ * amdgpu_seq64_unmap - Unmap the seq64 memory
+ *
+ * @adev: amdgpu_device pointer
+ * @fpriv: DRM file private
+ *
+ * Unmap the seq64 memory from the given VM.
+ */
+void amdgpu_seq64_unmap(struct amdgpu_device *adev, struct amdgpu_fpriv *fpriv)
+{
+	struct amdgpu_vm *vm;
+	struct amdgpu_bo *bo;
+	struct drm_exec exec;
+	int r;
+
+	if (!fpriv->seq64_va)
+		return;
+
+	bo = adev->seq64.sbo;
+	if (!bo)
+		return;
+
+	vm = &fpriv->vm;
+
+	drm_exec_init(&exec, 0, 0);
+	drm_exec_until_all_locked(&exec) {
+		r = amdgpu_vm_lock_pd(vm, &exec, 0);
+		if (likely(!r))
+			r = drm_exec_lock_obj(&exec, &bo->tbo.base);
+		drm_exec_retry_on_contention(&exec);
+		if (unlikely(r))
+			goto error;
+	}
+
+	amdgpu_vm_bo_del(adev, fpriv->seq64_va);
+
+	fpriv->seq64_va = NULL;
+
+error:
+	drm_exec_fini(&exec);
+}
+
+/**
+ * amdgpu_seq64_alloc - Allocate a 64 bit memory
+ *
+ * @adev: amdgpu_device pointer
+ * @va: VA to access the seq in process address space
+ * @gpu_addr: GPU address to access the seq
+ * @cpu_addr: CPU address to access the seq
+ *
+ * Alloc a 64 bit memory from seq64 pool.
+ *
+ * Returns:
+ * 0 on success or a negative error code on failure
+ */
+int amdgpu_seq64_alloc(struct amdgpu_device *adev, u64 *va,
+		       u64 *gpu_addr, u64 **cpu_addr)
+{
+	unsigned long bit_pos;
+
+	bit_pos = find_first_zero_bit(adev->seq64.used, adev->seq64.num_sem);
+	if (bit_pos >= adev->seq64.num_sem)
+		return -ENOSPC;
+
+	__set_bit(bit_pos, adev->seq64.used);
+
+	*va = bit_pos * sizeof(u64) + amdgpu_seq64_get_va_base(adev);
+
+	if (gpu_addr)
+		*gpu_addr = bit_pos * sizeof(u64) + adev->seq64.gpu_addr;
+
+	*cpu_addr = bit_pos + adev->seq64.cpu_base_addr;
+
+	return 0;
+}
+
+/**
+ * amdgpu_seq64_free - Free the given 64 bit memory
+ *
+ * @adev: amdgpu_device pointer
+ * @va: gpu start address to be freed
+ *
+ * Free the given 64 bit memory from seq64 pool.
+ */
+void amdgpu_seq64_free(struct amdgpu_device *adev, u64 va)
+{
+	unsigned long bit_pos;
+
+	bit_pos = (va - amdgpu_seq64_get_va_base(adev)) / sizeof(u64);
+	if (bit_pos < adev->seq64.num_sem)
+		__clear_bit(bit_pos, adev->seq64.used);
+}
+
+/**
+ * amdgpu_seq64_fini - Cleanup seq64 driver
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Free the memory space allocated for seq64.
+ *
+ */
+void amdgpu_seq64_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->seq64.sbo,
+			      NULL,
+			      (void **)&adev->seq64.cpu_base_addr);
+}
+
+/**
+ * amdgpu_seq64_init - Initialize seq64 driver
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate the required memory space for seq64.
+ *
+ * Returns:
+ * 0 on success or a negative error code on failure
+ */
+int amdgpu_seq64_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->seq64.sbo)
+		return 0;
+
+	/*
+	 * AMDGPU_MAX_SEQ64_SLOTS * sizeof(u64) * 8 = AMDGPU_MAX_SEQ64_SLOTS
+	 * 64bit slots
+	 */
+	r = amdgpu_bo_create_kernel(adev, AMDGPU_VA_RESERVED_SEQ64_SIZE,
+				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
+				    &adev->seq64.sbo, &adev->seq64.gpu_addr,
+				    (void **)&adev->seq64.cpu_base_addr);
+	if (r) {
+		dev_warn(adev->dev, "(%d) create seq64 failed\n", r);
+		return r;
+	}
+
+	memset(adev->seq64.cpu_base_addr, 0, AMDGPU_VA_RESERVED_SEQ64_SIZE);
+
+	adev->seq64.num_sem = AMDGPU_MAX_SEQ64_SLOTS;
+	memset(&adev->seq64.used, 0, sizeof(adev->seq64.used));
+
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.h
new file mode 100644
index 000000000000..26a249aaaee1
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.h
@@ -0,0 +1,49 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_SEQ64_H__
+#define __AMDGPU_SEQ64_H__
+
+#include "amdgpu_vm.h"
+
+#define AMDGPU_MAX_SEQ64_SLOTS         (AMDGPU_VA_RESERVED_SEQ64_SIZE / sizeof(u64))
+
+struct amdgpu_seq64 {
+	struct amdgpu_bo *sbo;
+	u32 num_sem;
+	u64 gpu_addr;
+	u64 *cpu_base_addr;
+	DECLARE_BITMAP(used, AMDGPU_MAX_SEQ64_SLOTS);
+};
+
+void amdgpu_seq64_fini(struct amdgpu_device *adev);
+int amdgpu_seq64_init(struct amdgpu_device *adev);
+int amdgpu_seq64_alloc(struct amdgpu_device *adev, u64 *va, u64 *gpu_addr, u64 **cpu_addr);
+void amdgpu_seq64_free(struct amdgpu_device *adev, u64 gpu_addr);
+int amdgpu_seq64_map(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+		     struct amdgpu_bo_va **bo_va);
+void amdgpu_seq64_unmap(struct amdgpu_device *adev, struct amdgpu_fpriv *fpriv);
+
+#endif
+
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_smuio.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_smuio.h
new file mode 100644
index 000000000000..ec9d12f85f39
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_smuio.h
@@ -0,0 +1,54 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __AMDGPU_SMUIO_H__
+#define __AMDGPU_SMUIO_H__
+
+enum amdgpu_pkg_type {
+	AMDGPU_PKG_TYPE_APU = 2,
+	AMDGPU_PKG_TYPE_CEM = 3,
+	AMDGPU_PKG_TYPE_OAM = 4,
+	AMDGPU_PKG_TYPE_UNKNOWN,
+};
+
+struct amdgpu_smuio_mcm_config_info {
+	int socket_id;
+	int die_id;
+};
+
+struct amdgpu_smuio_funcs {
+	u32 (*get_rom_index_offset)(struct amdgpu_device *adev);
+	u32 (*get_rom_data_offset)(struct amdgpu_device *adev);
+	void (*update_rom_clock_gating)(struct amdgpu_device *adev, bool enable);
+	void (*get_clock_gating_state)(struct amdgpu_device *adev, u64 *flags);
+	u32 (*get_die_id)(struct amdgpu_device *adev);
+	u32 (*get_socket_id)(struct amdgpu_device *adev);
+	enum amdgpu_pkg_type (*get_pkg_type)(struct amdgpu_device *adev);
+	bool (*is_host_gpu_xgmi_supported)(struct amdgpu_device *adev);
+	u64 (*get_gpu_clock_counter)(struct amdgpu_device *adev);
+};
+
+struct amdgpu_smuio {
+	const struct amdgpu_smuio_funcs		*funcs;
+};
+
+#endif /* __AMDGPU_SMUIO_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_socbb.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_socbb.h
new file mode 100644
index 000000000000..f4176cb01790
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_socbb.h
@@ -0,0 +1,82 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __AMDGPU_SOCBB_H__
+#define __AMDGPU_SOCBB_H__
+
+struct gpu_info_voltage_scaling_v1_0 {
+	uint32_t state;
+	uint32_t dscclk_mhz;
+	uint32_t dcfclk_mhz;
+	uint32_t socclk_mhz;
+	uint32_t dram_speed_mts;
+	uint32_t fabricclk_mhz;
+	uint32_t dispclk_mhz;
+	uint32_t phyclk_mhz;
+	uint32_t dppclk_mhz;
+};
+
+struct gpu_info_soc_bounding_box_v1_0 {
+	uint32_t sr_exit_time_us;
+	uint32_t sr_enter_plus_exit_time_us;
+	uint32_t urgent_latency_us;
+	uint32_t urgent_latency_pixel_data_only_us;
+	uint32_t urgent_latency_pixel_mixed_with_vm_data_us;
+	uint32_t urgent_latency_vm_data_only_us;
+	uint32_t writeback_latency_us;
+	uint32_t ideal_dram_bw_after_urgent_percent;
+	uint32_t pct_ideal_dram_sdp_bw_after_urgent_pixel_only; // PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelDataOnly
+	uint32_t pct_ideal_dram_sdp_bw_after_urgent_pixel_and_vm;
+	uint32_t pct_ideal_dram_sdp_bw_after_urgent_vm_only;
+	uint32_t max_avg_sdp_bw_use_normal_percent;
+	uint32_t max_avg_dram_bw_use_normal_percent;
+	uint32_t max_request_size_bytes;
+	uint32_t downspread_percent;
+	uint32_t dram_page_open_time_ns;
+	uint32_t dram_rw_turnaround_time_ns;
+	uint32_t dram_return_buffer_per_channel_bytes;
+	uint32_t dram_channel_width_bytes;
+	uint32_t fabric_datapath_to_dcn_data_return_bytes;
+	uint32_t dcn_downspread_percent;
+	uint32_t dispclk_dppclk_vco_speed_mhz;
+	uint32_t dfs_vco_period_ps;
+	uint32_t urgent_out_of_order_return_per_channel_pixel_only_bytes;
+	uint32_t urgent_out_of_order_return_per_channel_pixel_and_vm_bytes;
+	uint32_t urgent_out_of_order_return_per_channel_vm_only_bytes;
+	uint32_t round_trip_ping_latency_dcfclk_cycles;
+	uint32_t urgent_out_of_order_return_per_channel_bytes;
+	uint32_t channel_interleave_bytes;
+	uint32_t num_banks;
+	uint32_t num_chans;
+	uint32_t vmm_page_size_bytes;
+	uint32_t dram_clock_change_latency_us;
+	uint32_t writeback_dram_clock_change_latency_us;
+	uint32_t return_bus_width_bytes;
+	uint32_t voltage_override;
+	uint32_t xfc_bus_transport_time_us;
+	uint32_t xfc_xbuf_latency_tolerance_us;
+	uint32_t use_urgent_burst_bw;
+	uint32_t num_states;
+	struct gpu_info_voltage_scaling_v1_0 clock_limits[8];
+};
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c
index c586f44312f9..d6ae9974c952 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c
@@ -1,3 +1,4 @@
+// SPDX-License-Identifier: MIT
 /*
  * Copyright 2014 Advanced Micro Devices, Inc.
  * All Rights Reserved.
@@ -28,9 +29,11 @@
  *    Christian König <christian.koenig@amd.com>
  */
 
-#include <drm/drmP.h>
+#include <linux/dma-fence-chain.h>
+
 #include "amdgpu.h"
 #include "amdgpu_trace.h"
+#include "amdgpu_amdkfd.h"
 
 struct amdgpu_sync_entry {
 	struct hlist_node	node;
@@ -49,7 +52,6 @@ static struct kmem_cache *amdgpu_sync_slab;
 void amdgpu_sync_create(struct amdgpu_sync *sync)
 {
 	hash_init(sync->fences);
-	sync->last_vm_update = NULL;
 }
 
 /**
@@ -63,7 +65,7 @@ void amdgpu_sync_create(struct amdgpu_sync *sync)
 static bool amdgpu_sync_same_dev(struct amdgpu_device *adev,
 				 struct dma_fence *f)
 {
-	struct amd_sched_fence *s_fence = to_amd_sched_fence(f);
+	struct drm_sched_fence *s_fence = to_drm_sched_fence(f);
 
 	if (s_fence) {
 		struct amdgpu_ring *ring;
@@ -78,17 +80,26 @@ static bool amdgpu_sync_same_dev(struct amdgpu_device *adev,
 /**
  * amdgpu_sync_get_owner - extract the owner of a fence
  *
- * @fence: fence get the owner from
+ * @f: fence get the owner from
  *
  * Extract who originally created the fence.
  */
 static void *amdgpu_sync_get_owner(struct dma_fence *f)
 {
-	struct amd_sched_fence *s_fence = to_amd_sched_fence(f);
+	struct drm_sched_fence *s_fence;
+	struct amdgpu_amdkfd_fence *kfd_fence;
+
+	if (!f)
+		return AMDGPU_FENCE_OWNER_UNDEFINED;
 
+	s_fence = to_drm_sched_fence(f);
 	if (s_fence)
 		return s_fence->owner;
 
+	kfd_fence = to_amdgpu_amdkfd_fence(f);
+	if (kfd_fence)
+		return AMDGPU_FENCE_OWNER_KFD;
+
 	return AMDGPU_FENCE_OWNER_UNDEFINED;
 }
 
@@ -124,11 +135,16 @@ static bool amdgpu_sync_add_later(struct amdgpu_sync *sync, struct dma_fence *f)
 	struct amdgpu_sync_entry *e;
 
 	hash_for_each_possible(sync->fences, e, node, f->context) {
-		if (unlikely(e->fence->context != f->context))
-			continue;
+		if (dma_fence_is_signaled(e->fence)) {
+			dma_fence_put(e->fence);
+			e->fence = dma_fence_get(f);
+			return true;
+		}
 
-		amdgpu_sync_keep_later(&e->fence, f);
-		return true;
+		if (likely(e->fence->context == f->context)) {
+			amdgpu_sync_keep_later(&e->fence, f);
+			return true;
+		}
 	}
 	return false;
 }
@@ -137,25 +153,23 @@ static bool amdgpu_sync_add_later(struct amdgpu_sync *sync, struct dma_fence *f)
  * amdgpu_sync_fence - remember to sync to this fence
  *
  * @sync: sync object to add fence to
- * @fence: fence to sync to
+ * @f: fence to sync to
+ * @flags: memory allocation flags to use when allocating sync entry
  *
+ * Add the fence to the sync object.
  */
-int amdgpu_sync_fence(struct amdgpu_device *adev, struct amdgpu_sync *sync,
-		      struct dma_fence *f)
+int amdgpu_sync_fence(struct amdgpu_sync *sync, struct dma_fence *f,
+		      gfp_t flags)
 {
 	struct amdgpu_sync_entry *e;
 
 	if (!f)
 		return 0;
 
-	if (amdgpu_sync_same_dev(adev, f) &&
-	    amdgpu_sync_get_owner(f) == AMDGPU_FENCE_OWNER_VM)
-		amdgpu_sync_keep_later(&sync->last_vm_update, f);
-
 	if (amdgpu_sync_add_later(sync, f))
 		return 0;
 
-	e = kmem_cache_alloc(amdgpu_sync_slab, GFP_KERNEL);
+	e = kmem_cache_alloc(amdgpu_sync_slab, flags);
 	if (!e)
 		return -ENOMEM;
 
@@ -164,66 +178,132 @@ int amdgpu_sync_fence(struct amdgpu_device *adev, struct amdgpu_sync *sync,
 	return 0;
 }
 
+/* Determine based on the owner and mode if we should sync to a fence or not */
+static bool amdgpu_sync_test_fence(struct amdgpu_device *adev,
+				   enum amdgpu_sync_mode mode,
+				   void *owner, struct dma_fence *f)
+{
+	void *fence_owner = amdgpu_sync_get_owner(f);
+
+	/* Always sync to moves, no matter what */
+	if (fence_owner == AMDGPU_FENCE_OWNER_UNDEFINED)
+		return true;
+
+	/* We only want to trigger KFD eviction fences on
+	 * evict or move jobs. Skip KFD fences otherwise.
+	 */
+	if (fence_owner == AMDGPU_FENCE_OWNER_KFD &&
+	    owner != AMDGPU_FENCE_OWNER_UNDEFINED)
+		return false;
+
+	/* Never sync to VM updates either. */
+	if (fence_owner == AMDGPU_FENCE_OWNER_VM &&
+	    owner != AMDGPU_FENCE_OWNER_UNDEFINED &&
+	    owner != AMDGPU_FENCE_OWNER_KFD)
+		return false;
+
+	/* Ignore fences depending on the sync mode */
+	switch (mode) {
+	case AMDGPU_SYNC_ALWAYS:
+		return true;
+
+	case AMDGPU_SYNC_NE_OWNER:
+		if (amdgpu_sync_same_dev(adev, f) &&
+		    fence_owner == owner)
+			return false;
+		break;
+
+	case AMDGPU_SYNC_EQ_OWNER:
+		if (amdgpu_sync_same_dev(adev, f) &&
+		    fence_owner != owner)
+			return false;
+		break;
+
+	case AMDGPU_SYNC_EXPLICIT:
+		return false;
+	}
+
+	WARN(debug_evictions && fence_owner == AMDGPU_FENCE_OWNER_KFD,
+	     "Adding eviction fence to sync obj");
+	return true;
+}
+
 /**
  * amdgpu_sync_resv - sync to a reservation object
  *
+ * @adev: amdgpu device
  * @sync: sync object to add fences from reservation object to
  * @resv: reservation object with embedded fence
- * @shared: true if we should only sync to the exclusive fence
+ * @mode: how owner affects which fences we sync to
+ * @owner: owner of the planned job submission
  *
  * Sync to the fence
  */
-int amdgpu_sync_resv(struct amdgpu_device *adev,
-		     struct amdgpu_sync *sync,
-		     struct reservation_object *resv,
+int amdgpu_sync_resv(struct amdgpu_device *adev, struct amdgpu_sync *sync,
+		     struct dma_resv *resv, enum amdgpu_sync_mode mode,
 		     void *owner)
 {
-	struct reservation_object_list *flist;
+	struct dma_resv_iter cursor;
 	struct dma_fence *f;
-	void *fence_owner;
-	unsigned i;
-	int r = 0;
+	int r;
 
 	if (resv == NULL)
 		return -EINVAL;
+	/* Implicitly sync only to KERNEL, WRITE and READ */
+	dma_resv_for_each_fence(&cursor, resv, DMA_RESV_USAGE_READ, f) {
+		dma_fence_chain_for_each(f, f) {
+			struct dma_fence *tmp = dma_fence_chain_contained(f);
+
+			if (amdgpu_sync_test_fence(adev, mode, owner, tmp)) {
+				r = amdgpu_sync_fence(sync, f, GFP_KERNEL);
+				dma_fence_put(f);
+				if (r)
+					return r;
+				break;
+			}
+		}
+	}
+	return 0;
+}
 
-	/* always sync to the exclusive fence */
-	f = reservation_object_get_excl(resv);
-	r = amdgpu_sync_fence(adev, sync, f);
+/**
+ * amdgpu_sync_kfd - sync to KFD fences
+ *
+ * @sync: sync object to add KFD fences to
+ * @resv: reservation object with KFD fences
+ *
+ * Extract all KFD fences and add them to the sync object.
+ */
+int amdgpu_sync_kfd(struct amdgpu_sync *sync, struct dma_resv *resv)
+{
+	struct dma_resv_iter cursor;
+	struct dma_fence *f;
+	int r = 0;
 
-	flist = reservation_object_get_list(resv);
-	if (!flist || r)
-		return r;
+	dma_resv_iter_begin(&cursor, resv, DMA_RESV_USAGE_BOOKKEEP);
+	dma_resv_for_each_fence_unlocked(&cursor, f) {
+		void *fence_owner = amdgpu_sync_get_owner(f);
 
-	for (i = 0; i < flist->shared_count; ++i) {
-		f = rcu_dereference_protected(flist->shared[i],
-					      reservation_object_held(resv));
-		if (amdgpu_sync_same_dev(adev, f)) {
-			/* VM updates are only interesting
-			 * for other VM updates and moves.
-			 */
-			fence_owner = amdgpu_sync_get_owner(f);
-			if ((owner != AMDGPU_FENCE_OWNER_UNDEFINED) &&
-			    (fence_owner != AMDGPU_FENCE_OWNER_UNDEFINED) &&
-			    ((owner == AMDGPU_FENCE_OWNER_VM) !=
-			     (fence_owner == AMDGPU_FENCE_OWNER_VM)))
-				continue;
-
-			/* Ignore fence from the same owner as
-			 * long as it isn't undefined.
-			 */
-			if (owner != AMDGPU_FENCE_OWNER_UNDEFINED &&
-			    fence_owner == owner)
-				continue;
-		}
+		if (fence_owner != AMDGPU_FENCE_OWNER_KFD)
+			continue;
 
-		r = amdgpu_sync_fence(adev, sync, f);
+		r = amdgpu_sync_fence(sync, f, GFP_KERNEL);
 		if (r)
 			break;
 	}
+	dma_resv_iter_end(&cursor);
+
 	return r;
 }
 
+/* Free the entry back to the slab */
+static void amdgpu_sync_entry_free(struct amdgpu_sync_entry *e)
+{
+	hash_del(&e->node);
+	dma_fence_put(e->fence);
+	kmem_cache_free(amdgpu_sync_slab, e);
+}
+
 /**
  * amdgpu_sync_peek_fence - get the next fence not signaled yet
  *
@@ -242,12 +322,10 @@ struct dma_fence *amdgpu_sync_peek_fence(struct amdgpu_sync *sync,
 
 	hash_for_each_safe(sync->fences, i, tmp, e, node) {
 		struct dma_fence *f = e->fence;
-		struct amd_sched_fence *s_fence = to_amd_sched_fence(f);
+		struct drm_sched_fence *s_fence = to_drm_sched_fence(f);
 
 		if (dma_fence_is_signaled(f)) {
-			hash_del(&e->node);
-			dma_fence_put(f);
-			kmem_cache_free(amdgpu_sync_slab, e);
+			amdgpu_sync_entry_free(e);
 			continue;
 		}
 		if (ring && s_fence) {
@@ -297,6 +375,86 @@ struct dma_fence *amdgpu_sync_get_fence(struct amdgpu_sync *sync)
 	return NULL;
 }
 
+/**
+ * amdgpu_sync_clone - clone a sync object
+ *
+ * @source: sync object to clone
+ * @clone: pointer to destination sync object
+ *
+ * Adds references to all unsignaled fences in @source to @clone. Also
+ * removes signaled fences from @source while at it.
+ */
+int amdgpu_sync_clone(struct amdgpu_sync *source, struct amdgpu_sync *clone)
+{
+	struct amdgpu_sync_entry *e;
+	struct hlist_node *tmp;
+	struct dma_fence *f;
+	int i, r;
+
+	hash_for_each_safe(source->fences, i, tmp, e, node) {
+		f = e->fence;
+		if (!dma_fence_is_signaled(f)) {
+			r = amdgpu_sync_fence(clone, f, GFP_KERNEL);
+			if (r)
+				return r;
+		} else {
+			amdgpu_sync_entry_free(e);
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_sync_move - move all fences from src to dst
+ *
+ * @src: source of the fences, empty after function
+ * @dst: destination for the fences
+ *
+ * Moves all fences from source to destination. All fences in destination are
+ * freed and source is empty after the function call.
+ */
+void amdgpu_sync_move(struct amdgpu_sync *src, struct amdgpu_sync *dst)
+{
+	unsigned int i;
+
+	amdgpu_sync_free(dst);
+
+	for (i = 0; i < HASH_SIZE(src->fences); ++i)
+		hlist_move_list(&src->fences[i], &dst->fences[i]);
+}
+
+/**
+ * amdgpu_sync_push_to_job - push fences into job
+ * @sync: sync object to get the fences from
+ * @job: job to push the fences into
+ *
+ * Add all unsignaled fences from sync to job.
+ */
+int amdgpu_sync_push_to_job(struct amdgpu_sync *sync, struct amdgpu_job *job)
+{
+	struct amdgpu_sync_entry *e;
+	struct hlist_node *tmp;
+	struct dma_fence *f;
+	int i, r;
+
+	hash_for_each_safe(sync->fences, i, tmp, e, node) {
+		f = e->fence;
+		if (dma_fence_is_signaled(f)) {
+			amdgpu_sync_entry_free(e);
+			continue;
+		}
+
+		dma_fence_get(f);
+		r = drm_sched_job_add_dependency(&job->base, f);
+		if (r) {
+			dma_fence_put(f);
+			return r;
+		}
+	}
+	return 0;
+}
+
 int amdgpu_sync_wait(struct amdgpu_sync *sync, bool intr)
 {
 	struct amdgpu_sync_entry *e;
@@ -308,9 +466,7 @@ int amdgpu_sync_wait(struct amdgpu_sync *sync, bool intr)
 		if (r)
 			return r;
 
-		hash_del(&e->node);
-		dma_fence_put(e->fence);
-		kmem_cache_free(amdgpu_sync_slab, e);
+		amdgpu_sync_entry_free(e);
 	}
 
 	return 0;
@@ -327,15 +483,10 @@ void amdgpu_sync_free(struct amdgpu_sync *sync)
 {
 	struct amdgpu_sync_entry *e;
 	struct hlist_node *tmp;
-	unsigned i;
-
-	hash_for_each_safe(sync->fences, i, tmp, e, node) {
-		hash_del(&e->node);
-		dma_fence_put(e->fence);
-		kmem_cache_free(amdgpu_sync_slab, e);
-	}
+	unsigned int i;
 
-	dma_fence_put(sync->last_vm_update);
+	hash_for_each_safe(sync->fences, i, tmp, e, node)
+		amdgpu_sync_entry_free(e);
 }
 
 /**
@@ -345,9 +496,7 @@ void amdgpu_sync_free(struct amdgpu_sync *sync)
  */
 int amdgpu_sync_init(void)
 {
-	amdgpu_sync_slab = kmem_cache_create(
-		"amdgpu_sync", sizeof(struct amdgpu_sync_entry), 0,
-		SLAB_HWCACHE_ALIGN, NULL);
+	amdgpu_sync_slab = KMEM_CACHE(amdgpu_sync_entry, SLAB_HWCACHE_ALIGN);
 	if (!amdgpu_sync_slab)
 		return -ENOMEM;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.h
index dc7687993317..51eb4382c91e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.h
@@ -27,28 +27,38 @@
 #include <linux/hashtable.h>
 
 struct dma_fence;
-struct reservation_object;
+struct dma_resv;
 struct amdgpu_device;
 struct amdgpu_ring;
+struct amdgpu_job;
+
+enum amdgpu_sync_mode {
+	AMDGPU_SYNC_ALWAYS,
+	AMDGPU_SYNC_NE_OWNER,
+	AMDGPU_SYNC_EQ_OWNER,
+	AMDGPU_SYNC_EXPLICIT
+};
 
 /*
  * Container for fences used to sync command submissions.
  */
 struct amdgpu_sync {
 	DECLARE_HASHTABLE(fences, 4);
-	struct dma_fence	*last_vm_update;
 };
 
 void amdgpu_sync_create(struct amdgpu_sync *sync);
-int amdgpu_sync_fence(struct amdgpu_device *adev, struct amdgpu_sync *sync,
-		      struct dma_fence *f);
-int amdgpu_sync_resv(struct amdgpu_device *adev,
-		     struct amdgpu_sync *sync,
-		     struct reservation_object *resv,
+int amdgpu_sync_fence(struct amdgpu_sync *sync, struct dma_fence *f,
+		      gfp_t flags);
+int amdgpu_sync_resv(struct amdgpu_device *adev, struct amdgpu_sync *sync,
+		     struct dma_resv *resv, enum amdgpu_sync_mode mode,
 		     void *owner);
+int amdgpu_sync_kfd(struct amdgpu_sync *sync, struct dma_resv *resv);
 struct dma_fence *amdgpu_sync_peek_fence(struct amdgpu_sync *sync,
 				     struct amdgpu_ring *ring);
 struct dma_fence *amdgpu_sync_get_fence(struct amdgpu_sync *sync);
+int amdgpu_sync_clone(struct amdgpu_sync *source, struct amdgpu_sync *clone);
+void amdgpu_sync_move(struct amdgpu_sync *src, struct amdgpu_sync *dst);
+int amdgpu_sync_push_to_job(struct amdgpu_sync *sync, struct amdgpu_job *job);
 int amdgpu_sync_wait(struct amdgpu_sync *sync, bool intr);
 void amdgpu_sync_free(struct amdgpu_sync *sync);
 int amdgpu_sync_init(void);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c
deleted file mode 100644
index ed8c3739015b..000000000000
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c
+++ /dev/null
@@ -1,239 +0,0 @@
-/*
- * Copyright 2009 VMware, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- * Authors: Michel Dänzer
- */
-#include <drm/drmP.h>
-#include <drm/amdgpu_drm.h>
-#include "amdgpu.h"
-#include "amdgpu_uvd.h"
-#include "amdgpu_vce.h"
-
-/* Test BO GTT->VRAM and VRAM->GTT GPU copies across the whole GTT aperture */
-static void amdgpu_do_test_moves(struct amdgpu_device *adev)
-{
-	struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
-	struct amdgpu_bo *vram_obj = NULL;
-	struct amdgpu_bo **gtt_obj = NULL;
-	uint64_t gart_addr, vram_addr;
-	unsigned n, size;
-	int i, r;
-
-	size = 1024 * 1024;
-
-	/* Number of tests =
-	 * (Total GTT - IB pool - writeback page - ring buffers) / test size
-	 */
-	n = adev->mc.gart_size - AMDGPU_IB_POOL_SIZE*64*1024;
-	for (i = 0; i < AMDGPU_MAX_RINGS; ++i)
-		if (adev->rings[i])
-			n -= adev->rings[i]->ring_size;
-	if (adev->wb.wb_obj)
-		n -= AMDGPU_GPU_PAGE_SIZE;
-	if (adev->irq.ih.ring_obj)
-		n -= adev->irq.ih.ring_size;
-	n /= size;
-
-	gtt_obj = kzalloc(n * sizeof(*gtt_obj), GFP_KERNEL);
-	if (!gtt_obj) {
-		DRM_ERROR("Failed to allocate %d pointers\n", n);
-		r = 1;
-		goto out_cleanup;
-	}
-
-	r = amdgpu_bo_create(adev, size, PAGE_SIZE, true,
-			     AMDGPU_GEM_DOMAIN_VRAM, 0,
-			     NULL, NULL, 0, &vram_obj);
-	if (r) {
-		DRM_ERROR("Failed to create VRAM object\n");
-		goto out_cleanup;
-	}
-	r = amdgpu_bo_reserve(vram_obj, false);
-	if (unlikely(r != 0))
-		goto out_unref;
-	r = amdgpu_bo_pin(vram_obj, AMDGPU_GEM_DOMAIN_VRAM, &vram_addr);
-	if (r) {
-		DRM_ERROR("Failed to pin VRAM object\n");
-		goto out_unres;
-	}
-	for (i = 0; i < n; i++) {
-		void *gtt_map, *vram_map;
-		void **gart_start, **gart_end;
-		void **vram_start, **vram_end;
-		struct dma_fence *fence = NULL;
-
-		r = amdgpu_bo_create(adev, size, PAGE_SIZE, true,
-				     AMDGPU_GEM_DOMAIN_GTT, 0, NULL,
-				     NULL, 0, gtt_obj + i);
-		if (r) {
-			DRM_ERROR("Failed to create GTT object %d\n", i);
-			goto out_lclean;
-		}
-
-		r = amdgpu_bo_reserve(gtt_obj[i], false);
-		if (unlikely(r != 0))
-			goto out_lclean_unref;
-		r = amdgpu_bo_pin(gtt_obj[i], AMDGPU_GEM_DOMAIN_GTT, &gart_addr);
-		if (r) {
-			DRM_ERROR("Failed to pin GTT object %d\n", i);
-			goto out_lclean_unres;
-		}
-
-		r = amdgpu_bo_kmap(gtt_obj[i], &gtt_map);
-		if (r) {
-			DRM_ERROR("Failed to map GTT object %d\n", i);
-			goto out_lclean_unpin;
-		}
-
-		for (gart_start = gtt_map, gart_end = gtt_map + size;
-		     gart_start < gart_end;
-		     gart_start++)
-			*gart_start = gart_start;
-
-		amdgpu_bo_kunmap(gtt_obj[i]);
-
-		r = amdgpu_copy_buffer(ring, gart_addr, vram_addr,
-				       size, NULL, &fence, false, false);
-
-		if (r) {
-			DRM_ERROR("Failed GTT->VRAM copy %d\n", i);
-			goto out_lclean_unpin;
-		}
-
-		r = dma_fence_wait(fence, false);
-		if (r) {
-			DRM_ERROR("Failed to wait for GTT->VRAM fence %d\n", i);
-			goto out_lclean_unpin;
-		}
-
-		dma_fence_put(fence);
-
-		r = amdgpu_bo_kmap(vram_obj, &vram_map);
-		if (r) {
-			DRM_ERROR("Failed to map VRAM object after copy %d\n", i);
-			goto out_lclean_unpin;
-		}
-
-		for (gart_start = gtt_map, gart_end = gtt_map + size,
-		     vram_start = vram_map, vram_end = vram_map + size;
-		     vram_start < vram_end;
-		     gart_start++, vram_start++) {
-			if (*vram_start != gart_start) {
-				DRM_ERROR("Incorrect GTT->VRAM copy %d: Got 0x%p, "
-					  "expected 0x%p (GTT/VRAM offset "
-					  "0x%16llx/0x%16llx)\n",
-					  i, *vram_start, gart_start,
-					  (unsigned long long)
-					  (gart_addr - adev->mc.gart_start +
-					   (void*)gart_start - gtt_map),
-					  (unsigned long long)
-					  (vram_addr - adev->mc.vram_start +
-					   (void*)gart_start - gtt_map));
-				amdgpu_bo_kunmap(vram_obj);
-				goto out_lclean_unpin;
-			}
-			*vram_start = vram_start;
-		}
-
-		amdgpu_bo_kunmap(vram_obj);
-
-		r = amdgpu_copy_buffer(ring, vram_addr, gart_addr,
-				       size, NULL, &fence, false, false);
-
-		if (r) {
-			DRM_ERROR("Failed VRAM->GTT copy %d\n", i);
-			goto out_lclean_unpin;
-		}
-
-		r = dma_fence_wait(fence, false);
-		if (r) {
-			DRM_ERROR("Failed to wait for VRAM->GTT fence %d\n", i);
-			goto out_lclean_unpin;
-		}
-
-		dma_fence_put(fence);
-
-		r = amdgpu_bo_kmap(gtt_obj[i], &gtt_map);
-		if (r) {
-			DRM_ERROR("Failed to map GTT object after copy %d\n", i);
-			goto out_lclean_unpin;
-		}
-
-		for (gart_start = gtt_map, gart_end = gtt_map + size,
-		     vram_start = vram_map, vram_end = vram_map + size;
-		     gart_start < gart_end;
-		     gart_start++, vram_start++) {
-			if (*gart_start != vram_start) {
-				DRM_ERROR("Incorrect VRAM->GTT copy %d: Got 0x%p, "
-					  "expected 0x%p (VRAM/GTT offset "
-					  "0x%16llx/0x%16llx)\n",
-					  i, *gart_start, vram_start,
-					  (unsigned long long)
-					  (vram_addr - adev->mc.vram_start +
-					   (void*)vram_start - vram_map),
-					  (unsigned long long)
-					  (gart_addr - adev->mc.gart_start +
-					   (void*)vram_start - vram_map));
-				amdgpu_bo_kunmap(gtt_obj[i]);
-				goto out_lclean_unpin;
-			}
-		}
-
-		amdgpu_bo_kunmap(gtt_obj[i]);
-
-		DRM_INFO("Tested GTT->VRAM and VRAM->GTT copy for GTT offset 0x%llx\n",
-			 gart_addr - adev->mc.gart_start);
-		continue;
-
-out_lclean_unpin:
-		amdgpu_bo_unpin(gtt_obj[i]);
-out_lclean_unres:
-		amdgpu_bo_unreserve(gtt_obj[i]);
-out_lclean_unref:
-		amdgpu_bo_unref(&gtt_obj[i]);
-out_lclean:
-		for (--i; i >= 0; --i) {
-			amdgpu_bo_unpin(gtt_obj[i]);
-			amdgpu_bo_unreserve(gtt_obj[i]);
-			amdgpu_bo_unref(&gtt_obj[i]);
-		}
-		if (fence)
-			dma_fence_put(fence);
-		break;
-	}
-
-	amdgpu_bo_unpin(vram_obj);
-out_unres:
-	amdgpu_bo_unreserve(vram_obj);
-out_unref:
-	amdgpu_bo_unref(&vram_obj);
-out_cleanup:
-	kfree(gtt_obj);
-	if (r) {
-		pr_warn("Error while testing BO move\n");
-	}
-}
-
-void amdgpu_test_moves(struct amdgpu_device *adev)
-{
-	if (adev->mman.buffer_funcs)
-		amdgpu_do_test_moves(adev);
-}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
index 1c88bd5e29ad..d13e64a69e25 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
@@ -1,12 +1,33 @@
-#if !defined(_AMDGPU_TRACE_H) || defined(TRACE_HEADER_MULTI_READ)
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#if !defined(_AMDGPU_TRACE_H_) || defined(TRACE_HEADER_MULTI_READ)
 #define _AMDGPU_TRACE_H_
 
 #include <linux/stringify.h>
 #include <linux/types.h>
 #include <linux/tracepoint.h>
 
-#include <drm/drmP.h>
-
 #undef TRACE_SYSTEM
 #define TRACE_SYSTEM amdgpu
 #define TRACE_INCLUDE_FILE amdgpu_trace
@@ -14,63 +35,7 @@
 #define AMDGPU_JOB_GET_TIMELINE_NAME(job) \
 	 job->base.s_fence->finished.ops->get_timeline_name(&job->base.s_fence->finished)
 
-TRACE_EVENT(amdgpu_ttm_tt_populate,
-	    TP_PROTO(struct amdgpu_device *adev, uint64_t dma_address, uint64_t phys_address),
-	    TP_ARGS(adev, dma_address, phys_address),
-	    TP_STRUCT__entry(
-				__field(uint16_t, domain)
-				__field(uint8_t, bus)
-				__field(uint8_t, slot)
-				__field(uint8_t, func)
-				__field(uint64_t, dma)
-				__field(uint64_t, phys)
-			    ),
-	    TP_fast_assign(
-			   __entry->domain = pci_domain_nr(adev->pdev->bus);
-			   __entry->bus = adev->pdev->bus->number;
-			   __entry->slot = PCI_SLOT(adev->pdev->devfn);
-			   __entry->func = PCI_FUNC(adev->pdev->devfn);
-			   __entry->dma = dma_address;
-			   __entry->phys = phys_address;
-			   ),
-	    TP_printk("%04x:%02x:%02x.%x: 0x%llx => 0x%llx",
-		      (unsigned)__entry->domain,
-		      (unsigned)__entry->bus,
-		      (unsigned)__entry->slot,
-		      (unsigned)__entry->func,
-		      (unsigned long long)__entry->dma,
-		      (unsigned long long)__entry->phys)
-);
-
-TRACE_EVENT(amdgpu_ttm_tt_unpopulate,
-	    TP_PROTO(struct amdgpu_device *adev, uint64_t dma_address, uint64_t phys_address),
-	    TP_ARGS(adev, dma_address, phys_address),
-	    TP_STRUCT__entry(
-				__field(uint16_t, domain)
-				__field(uint8_t, bus)
-				__field(uint8_t, slot)
-				__field(uint8_t, func)
-				__field(uint64_t, dma)
-				__field(uint64_t, phys)
-			    ),
-	    TP_fast_assign(
-			   __entry->domain = pci_domain_nr(adev->pdev->bus);
-			   __entry->bus = adev->pdev->bus->number;
-			   __entry->slot = PCI_SLOT(adev->pdev->devfn);
-			   __entry->func = PCI_FUNC(adev->pdev->devfn);
-			   __entry->dma = dma_address;
-			   __entry->phys = phys_address;
-			   ),
-	    TP_printk("%04x:%02x:%02x.%x: 0x%llx => 0x%llx",
-		      (unsigned)__entry->domain,
-		      (unsigned)__entry->bus,
-		      (unsigned)__entry->slot,
-		      (unsigned)__entry->func,
-		      (unsigned long long)__entry->dma,
-		      (unsigned long long)__entry->phys)
-);
-
-TRACE_EVENT(amdgpu_mm_rreg,
+TRACE_EVENT(amdgpu_device_rreg,
 	    TP_PROTO(unsigned did, uint32_t reg, uint32_t value),
 	    TP_ARGS(did, reg, value),
 	    TP_STRUCT__entry(
@@ -89,7 +54,7 @@ TRACE_EVENT(amdgpu_mm_rreg,
 		      (unsigned long)__entry->value)
 );
 
-TRACE_EVENT(amdgpu_mm_wreg,
+TRACE_EVENT(amdgpu_device_wreg,
 	    TP_PROTO(unsigned did, uint32_t reg, uint32_t value),
 	    TP_ARGS(did, reg, value),
 	    TP_STRUCT__entry(
@@ -109,37 +74,40 @@ TRACE_EVENT(amdgpu_mm_wreg,
 );
 
 TRACE_EVENT(amdgpu_iv,
-	    TP_PROTO(struct amdgpu_iv_entry *iv),
-	    TP_ARGS(iv),
+	    TP_PROTO(unsigned ih, struct amdgpu_iv_entry *iv),
+	    TP_ARGS(ih, iv),
 	    TP_STRUCT__entry(
+			     __field(unsigned, ih)
 			     __field(unsigned, client_id)
 			     __field(unsigned, src_id)
 			     __field(unsigned, ring_id)
-			     __field(unsigned, vm_id)
-			     __field(unsigned, vm_id_src)
+			     __field(unsigned, vmid)
+			     __field(unsigned, vmid_src)
 			     __field(uint64_t, timestamp)
 			     __field(unsigned, timestamp_src)
-			     __field(unsigned, pas_id)
+			     __field(unsigned, pasid)
 			     __array(unsigned, src_data, 4)
 			    ),
 	    TP_fast_assign(
+			   __entry->ih = ih;
 			   __entry->client_id = iv->client_id;
 			   __entry->src_id = iv->src_id;
 			   __entry->ring_id = iv->ring_id;
-			   __entry->vm_id = iv->vm_id;
-			   __entry->vm_id_src = iv->vm_id_src;
+			   __entry->vmid = iv->vmid;
+			   __entry->vmid_src = iv->vmid_src;
 			   __entry->timestamp = iv->timestamp;
 			   __entry->timestamp_src = iv->timestamp_src;
-			   __entry->pas_id = iv->pas_id;
+			   __entry->pasid = iv->pasid;
 			   __entry->src_data[0] = iv->src_data[0];
 			   __entry->src_data[1] = iv->src_data[1];
 			   __entry->src_data[2] = iv->src_data[2];
 			   __entry->src_data[3] = iv->src_data[3];
 			   ),
-	    TP_printk("client_id:%u src_id:%u ring:%u vm_id:%u timestamp: %llu pas_id:%u src_data: %08x %08x %08x %08x\n",
-		      __entry->client_id, __entry->src_id,
-		      __entry->ring_id, __entry->vm_id,
-		      __entry->timestamp, __entry->pas_id,
+	    TP_printk("ih:%u client_id:%u src_id:%u ring:%u vmid:%u "
+		      "timestamp: %llu pasid:%u src_data: %08x %08x %08x %08x",
+		      __entry->ih, __entry->client_id, __entry->src_id,
+		      __entry->ring_id, __entry->vmid,
+		      __entry->timestamp, __entry->pasid,
 		      __entry->src_data[0], __entry->src_data[1],
 		      __entry->src_data[2], __entry->src_data[3])
 );
@@ -159,8 +127,8 @@ TRACE_EVENT(amdgpu_bo_create,
 
 	    TP_fast_assign(
 			   __entry->bo = bo;
-			   __entry->pages = bo->tbo.num_pages;
-			   __entry->type = bo->tbo.mem.mem_type;
+			   __entry->pages = PFN_UP(bo->tbo.resource->size);
+			   __entry->type = bo->tbo.resource->mem_type;
 			   __entry->prefer = bo->preferred_domains;
 			   __entry->allow = bo->allowed_domains;
 			   __entry->visible = bo->flags;
@@ -172,8 +140,10 @@ TRACE_EVENT(amdgpu_bo_create,
 );
 
 TRACE_EVENT(amdgpu_cs,
-	    TP_PROTO(struct amdgpu_cs_parser *p, int i),
-	    TP_ARGS(p, i),
+	    TP_PROTO(struct amdgpu_cs_parser *p,
+		     struct amdgpu_job *job,
+		     struct amdgpu_ib *ib),
+	    TP_ARGS(p, job, ib),
 	    TP_STRUCT__entry(
 			     __field(struct amdgpu_bo_list *, bo_list)
 			     __field(u32, ring)
@@ -183,10 +153,10 @@ TRACE_EVENT(amdgpu_cs,
 
 	    TP_fast_assign(
 			   __entry->bo_list = p->bo_list;
-			   __entry->ring = p->job->ring->idx;
-			   __entry->dw = p->job->ibs[i].length_dw;
+			   __entry->ring = to_amdgpu_ring(job->base.entity->rq->sched)->idx;
+			   __entry->dw = ib->length_dw;
 			   __entry->fences = amdgpu_fence_count_emitted(
-				p->job->ring);
+				to_amdgpu_ring(job->base.entity->rq->sched));
 			   ),
 	    TP_printk("bo_list=%p, ring=%u, dw=%u, fences=%u",
 		      __entry->bo_list, __entry->ring, __entry->dw,
@@ -197,51 +167,47 @@ TRACE_EVENT(amdgpu_cs_ioctl,
 	    TP_PROTO(struct amdgpu_job *job),
 	    TP_ARGS(job),
 	    TP_STRUCT__entry(
-			     __field(uint64_t, sched_job_id)
 			     __string(timeline, AMDGPU_JOB_GET_TIMELINE_NAME(job))
-			     __field(unsigned int, context)
-			     __field(unsigned int, seqno)
+			     __field(u64, context)
+			     __field(u64, seqno)
 			     __field(struct dma_fence *, fence)
-			     __field(char *, ring_name)
+			     __string(ring, to_amdgpu_ring(job->base.sched)->name)
 			     __field(u32, num_ibs)
 			     ),
 
 	    TP_fast_assign(
-			   __entry->sched_job_id = job->base.id;
-			   __assign_str(timeline, AMDGPU_JOB_GET_TIMELINE_NAME(job))
+			   __assign_str(timeline);
 			   __entry->context = job->base.s_fence->finished.context;
 			   __entry->seqno = job->base.s_fence->finished.seqno;
-			   __entry->ring_name = job->ring->name;
+			   __assign_str(ring);
 			   __entry->num_ibs = job->num_ibs;
 			   ),
-	    TP_printk("sched_job=%llu, timeline=%s, context=%u, seqno=%u, ring_name=%s, num_ibs=%u",
-		      __entry->sched_job_id, __get_str(timeline), __entry->context,
-		      __entry->seqno, __entry->ring_name, __entry->num_ibs)
+	    TP_printk("timeline=%s, fence=%llu:%llu, ring_name=%s, num_ibs=%u",
+		      __get_str(timeline), __entry->context,
+		      __entry->seqno, __get_str(ring), __entry->num_ibs)
 );
 
 TRACE_EVENT(amdgpu_sched_run_job,
 	    TP_PROTO(struct amdgpu_job *job),
 	    TP_ARGS(job),
 	    TP_STRUCT__entry(
-			     __field(uint64_t, sched_job_id)
 			     __string(timeline, AMDGPU_JOB_GET_TIMELINE_NAME(job))
-			     __field(unsigned int, context)
-			     __field(unsigned int, seqno)
-			     __field(char *, ring_name)
+			     __field(u64, context)
+			     __field(u64, seqno)
+			     __string(ring, to_amdgpu_ring(job->base.sched)->name)
 			     __field(u32, num_ibs)
 			     ),
 
 	    TP_fast_assign(
-			   __entry->sched_job_id = job->base.id;
-			   __assign_str(timeline, AMDGPU_JOB_GET_TIMELINE_NAME(job))
+			   __assign_str(timeline);
 			   __entry->context = job->base.s_fence->finished.context;
 			   __entry->seqno = job->base.s_fence->finished.seqno;
-			   __entry->ring_name = job->ring->name;
+			   __assign_str(ring);
 			   __entry->num_ibs = job->num_ibs;
 			   ),
-	    TP_printk("sched_job=%llu, timeline=%s, context=%u, seqno=%u, ring_name=%s, num_ibs=%u",
-		      __entry->sched_job_id, __get_str(timeline), __entry->context,
-		      __entry->seqno, __entry->ring_name, __entry->num_ibs)
+	    TP_printk("timeline=%s, fence=%llu:%llu, ring_name=%s, num_ibs=%u",
+		      __get_str(timeline), __entry->context,
+		      __entry->seqno, __get_str(ring), __entry->num_ibs)
 );
 
 
@@ -250,24 +216,25 @@ TRACE_EVENT(amdgpu_vm_grab_id,
 		     struct amdgpu_job *job),
 	    TP_ARGS(vm, ring, job),
 	    TP_STRUCT__entry(
-			     __field(struct amdgpu_vm *, vm)
+			     __field(u32, pasid)
+			     __string(ring, ring->name)
 			     __field(u32, ring)
-			     __field(u32, vm_id)
+			     __field(u32, vmid)
 			     __field(u32, vm_hub)
 			     __field(u64, pd_addr)
 			     __field(u32, needs_flush)
 			     ),
 
 	    TP_fast_assign(
-			   __entry->vm = vm;
-			   __entry->ring = ring->idx;
-			   __entry->vm_id = job->vm_id;
-			   __entry->vm_hub = ring->funcs->vmhub,
+			   __entry->pasid = vm->pasid;
+			   __assign_str(ring);
+			   __entry->vmid = job->vmid;
+			   __entry->vm_hub = ring->vm_hub,
 			   __entry->pd_addr = job->vm_pd_addr;
 			   __entry->needs_flush = job->vm_needs_flush;
 			   ),
-	    TP_printk("vm=%p, ring=%u, id=%u, hub=%u, pd_addr=%010Lx needs_flush=%u",
-		      __entry->vm, __entry->ring, __entry->vm_id,
+	    TP_printk("pasid=%d, ring=%s, id=%u, hub=%u, pd_addr=%010Lx needs_flush=%u",
+		      __entry->pasid, __get_str(ring), __entry->vmid,
 		      __entry->vm_hub, __entry->pd_addr, __entry->needs_flush)
 );
 
@@ -308,7 +275,7 @@ TRACE_EVENT(amdgpu_vm_bo_unmap,
 			     ),
 
 	    TP_fast_assign(
-			   __entry->bo = bo_va->base.bo;
+			   __entry->bo = bo_va ? bo_va->base.bo : NULL;
 			   __entry->start = mapping->start;
 			   __entry->last = mapping->last;
 			   __entry->offset = mapping->offset;
@@ -347,16 +314,65 @@ DEFINE_EVENT(amdgpu_vm_mapping, amdgpu_vm_bo_mapping,
 	    TP_ARGS(mapping)
 );
 
+DEFINE_EVENT(amdgpu_vm_mapping, amdgpu_vm_bo_cs,
+	    TP_PROTO(struct amdgpu_bo_va_mapping *mapping),
+	    TP_ARGS(mapping)
+);
+
+TRACE_EVENT(amdgpu_vm_update_ptes,
+	    TP_PROTO(struct amdgpu_vm_update_params *p,
+		     uint64_t start, uint64_t end,
+		     unsigned int nptes, uint64_t dst,
+		     uint64_t incr, uint64_t flags,
+		     pid_t pid, uint64_t vm_ctx),
+	TP_ARGS(p, start, end, nptes, dst, incr, flags, pid, vm_ctx),
+	TP_STRUCT__entry(
+			 __field(u64, start)
+			 __field(u64, end)
+			 __field(u64, flags)
+			 __field(unsigned int, nptes)
+			 __field(u64, incr)
+			 __field(pid_t, pid)
+			 __field(u64, vm_ctx)
+			 __dynamic_array(u64, dst, nptes)
+	),
+
+	TP_fast_assign(
+			unsigned int i;
+
+			__entry->start = start;
+			__entry->end = end;
+			__entry->flags = flags;
+			__entry->incr = incr;
+			__entry->nptes = nptes;
+			__entry->pid = pid;
+			__entry->vm_ctx = vm_ctx;
+			for (i = 0; i < nptes; ++i) {
+				u64 addr = p->pages_addr ? amdgpu_vm_map_gart(
+					p->pages_addr, dst) : dst;
+
+				((u64 *)__get_dynamic_array(dst))[i] = addr;
+				dst += incr;
+			}
+	),
+	TP_printk("pid:%u vm_ctx:0x%llx start:0x%010llx end:0x%010llx,"
+		  " flags:0x%llx, incr:%llu, dst:\n%s", __entry->pid,
+		  __entry->vm_ctx, __entry->start, __entry->end,
+		  __entry->flags, __entry->incr,  __print_array(
+		  __get_dynamic_array(dst), __entry->nptes, 8))
+);
+
 TRACE_EVENT(amdgpu_vm_set_ptes,
 	    TP_PROTO(uint64_t pe, uint64_t addr, unsigned count,
-		     uint32_t incr, uint64_t flags),
-	    TP_ARGS(pe, addr, count, incr, flags),
+		     uint32_t incr, uint64_t flags, bool immediate),
+	    TP_ARGS(pe, addr, count, incr, flags, immediate),
 	    TP_STRUCT__entry(
 			     __field(u64, pe)
 			     __field(u64, addr)
 			     __field(u32, count)
 			     __field(u32, incr)
 			     __field(u64, flags)
+			     __field(bool, immediate)
 			     ),
 
 	    TP_fast_assign(
@@ -365,50 +381,108 @@ TRACE_EVENT(amdgpu_vm_set_ptes,
 			   __entry->count = count;
 			   __entry->incr = incr;
 			   __entry->flags = flags;
+			   __entry->immediate = immediate;
 			   ),
-	    TP_printk("pe=%010Lx, addr=%010Lx, incr=%u, flags=%llx, count=%u",
-		      __entry->pe, __entry->addr, __entry->incr,
-		      __entry->flags, __entry->count)
+	    TP_printk("pe=%010Lx, addr=%010Lx, incr=%u, flags=%llx, count=%u, "
+		      "immediate=%d", __entry->pe, __entry->addr, __entry->incr,
+		      __entry->flags, __entry->count, __entry->immediate)
 );
 
 TRACE_EVENT(amdgpu_vm_copy_ptes,
-	    TP_PROTO(uint64_t pe, uint64_t src, unsigned count),
-	    TP_ARGS(pe, src, count),
+	    TP_PROTO(uint64_t pe, uint64_t src, unsigned count, bool immediate),
+	    TP_ARGS(pe, src, count, immediate),
 	    TP_STRUCT__entry(
 			     __field(u64, pe)
 			     __field(u64, src)
 			     __field(u32, count)
+			     __field(bool, immediate)
 			     ),
 
 	    TP_fast_assign(
 			   __entry->pe = pe;
 			   __entry->src = src;
 			   __entry->count = count;
+			   __entry->immediate = immediate;
 			   ),
-	    TP_printk("pe=%010Lx, src=%010Lx, count=%u",
-		      __entry->pe, __entry->src, __entry->count)
+	    TP_printk("pe=%010Lx, src=%010Lx, count=%u, immediate=%d",
+		      __entry->pe, __entry->src, __entry->count,
+		      __entry->immediate)
 );
 
 TRACE_EVENT(amdgpu_vm_flush,
-	    TP_PROTO(struct amdgpu_ring *ring, unsigned vm_id,
+	    TP_PROTO(struct amdgpu_ring *ring, unsigned vmid,
 		     uint64_t pd_addr),
-	    TP_ARGS(ring, vm_id, pd_addr),
+	    TP_ARGS(ring, vmid, pd_addr),
 	    TP_STRUCT__entry(
-			     __field(u32, ring)
-			     __field(u32, vm_id)
+			     __string(ring, ring->name)
+			     __field(u32, vmid)
 			     __field(u32, vm_hub)
 			     __field(u64, pd_addr)
 			     ),
 
 	    TP_fast_assign(
-			   __entry->ring = ring->idx;
-			   __entry->vm_id = vm_id;
-			   __entry->vm_hub = ring->funcs->vmhub;
+			   __assign_str(ring);
+			   __entry->vmid = vmid;
+			   __entry->vm_hub = ring->vm_hub;
 			   __entry->pd_addr = pd_addr;
 			   ),
-	    TP_printk("ring=%u, id=%u, hub=%u, pd_addr=%010Lx",
-		      __entry->ring, __entry->vm_id,
-		      __entry->vm_hub,__entry->pd_addr)
+	    TP_printk("ring=%s, id=%u, hub=%u, pd_addr=%010Lx",
+		      __get_str(ring), __entry->vmid,
+		      __entry->vm_hub, __entry->pd_addr)
+);
+
+DECLARE_EVENT_CLASS(amdgpu_pasid,
+	    TP_PROTO(unsigned pasid),
+	    TP_ARGS(pasid),
+	    TP_STRUCT__entry(
+			     __field(unsigned, pasid)
+			     ),
+	    TP_fast_assign(
+			   __entry->pasid = pasid;
+			   ),
+	    TP_printk("pasid=%u", __entry->pasid)
+);
+
+DEFINE_EVENT(amdgpu_pasid, amdgpu_pasid_allocated,
+	    TP_PROTO(unsigned pasid),
+	    TP_ARGS(pasid)
+);
+
+DEFINE_EVENT(amdgpu_pasid, amdgpu_pasid_freed,
+	    TP_PROTO(unsigned pasid),
+	    TP_ARGS(pasid)
+);
+
+TRACE_EVENT(amdgpu_isolation,
+	    TP_PROTO(void *prev, void *next),
+	    TP_ARGS(prev, next),
+	    TP_STRUCT__entry(
+			     __field(void *, prev)
+			     __field(void *, next)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->prev = prev;
+			   __entry->next = next;
+			   ),
+	    TP_printk("prev=%p, next=%p",
+		      __entry->prev,
+		      __entry->next)
+);
+
+TRACE_EVENT(amdgpu_cleaner_shader,
+	    TP_PROTO(struct amdgpu_ring *ring, struct dma_fence *fence),
+	    TP_ARGS(ring, fence),
+	    TP_STRUCT__entry(
+			     __string(ring, ring->name)
+			     __field(u64, seqno)
+			     ),
+
+	    TP_fast_assign(
+			   __assign_str(ring);
+			   __entry->seqno = fence->seqno;
+			   ),
+	    TP_printk("ring=%s, seqno=%Lu", __get_str(ring), __entry->seqno)
 );
 
 TRACE_EVENT(amdgpu_bo_list_set,
@@ -447,8 +521,8 @@ TRACE_EVENT(amdgpu_cs_bo_status,
 			__entry->total_bo, __entry->total_size)
 );
 
-TRACE_EVENT(amdgpu_ttm_bo_move,
-	    TP_PROTO(struct amdgpu_bo* bo, uint32_t new_placement, uint32_t old_placement),
+TRACE_EVENT(amdgpu_bo_move,
+	    TP_PROTO(struct amdgpu_bo *bo, uint32_t new_placement, uint32_t old_placement),
 	    TP_ARGS(bo, new_placement, old_placement),
 	    TP_STRUCT__entry(
 			__field(struct amdgpu_bo *, bo)
@@ -468,10 +542,46 @@ TRACE_EVENT(amdgpu_ttm_bo_move,
 			__entry->new_placement, __entry->bo_size)
 );
 
+TRACE_EVENT(amdgpu_ib_pipe_sync,
+	    TP_PROTO(struct amdgpu_job *sched_job, struct dma_fence *fence),
+	    TP_ARGS(sched_job, fence),
+	    TP_STRUCT__entry(
+			     __string(ring, sched_job->base.sched->name)
+			     __field(struct dma_fence *, fence)
+			     __field(u64, ctx)
+			     __field(u64, seqno)
+			     ),
+
+	    TP_fast_assign(
+			   __assign_str(ring);
+			   __entry->fence = fence;
+			   __entry->ctx = fence->context;
+			   __entry->seqno = fence->seqno;
+			   ),
+	    TP_printk("job ring=%s need pipe sync to fence=%llu:%llu",
+		      __get_str(ring), __entry->ctx, __entry->seqno)
+);
+
+TRACE_EVENT(amdgpu_reset_reg_dumps,
+	    TP_PROTO(uint32_t address, uint32_t value),
+	    TP_ARGS(address, value),
+	    TP_STRUCT__entry(
+			     __field(uint32_t, address)
+			     __field(uint32_t, value)
+			     ),
+	    TP_fast_assign(
+			   __entry->address = address;
+			   __entry->value = value;
+			   ),
+	    TP_printk("amdgpu register dump 0x%x: 0x%x",
+		      __entry->address,
+		      __entry->value)
+);
+
 #undef AMDGPU_JOB_GET_TIMELINE_NAME
 #endif
 
 /* This part must be outside protection */
 #undef TRACE_INCLUDE_PATH
-#define TRACE_INCLUDE_PATH .
+#define TRACE_INCLUDE_PATH ../../drivers/gpu/drm/amd/amdgpu
 #include <trace/define_trace.h>
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace_points.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace_points.c
index 385b7e1d72f9..b96d885f6e33 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace_points.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace_points.c
@@ -1,8 +1,29 @@
+// SPDX-License-Identifier: MIT
 /* Copyright Red Hat Inc 2010.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
  * Author : Dave Airlie <airlied@redhat.com>
  */
-#include <drm/drmP.h>
+
 #include <drm/amdgpu_drm.h>
+#include "amdgpu_cs.h"
 #include "amdgpu.h"
 
 #define CREATE_TRACE_POINTS
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
index 7ef6c28a34d9..aa9ee5dffa45 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
@@ -29,167 +29,64 @@
  *    Thomas Hellstrom <thomas-at-tungstengraphics-dot-com>
  *    Dave Airlie
  */
-#include <drm/ttm/ttm_bo_api.h>
-#include <drm/ttm/ttm_bo_driver.h>
-#include <drm/ttm/ttm_placement.h>
-#include <drm/ttm/ttm_module.h>
-#include <drm/ttm/ttm_page_alloc.h>
-#include <drm/drmP.h>
-#include <drm/amdgpu_drm.h>
+
+#include <linux/dma-mapping.h>
+#include <linux/iommu.h>
+#include <linux/pagemap.h>
+#include <linux/sched/task.h>
+#include <linux/sched/mm.h>
 #include <linux/seq_file.h>
 #include <linux/slab.h>
-#include <linux/swiotlb.h>
 #include <linux/swap.h>
-#include <linux/pagemap.h>
-#include <linux/debugfs.h>
-#include "amdgpu.h"
-#include "amdgpu_trace.h"
-#include "bif/bif_4_1_d.h"
-
-#define DRM_FILE_PAGE_OFFSET (0x100000000ULL >> PAGE_SHIFT)
-
-static int amdgpu_map_buffer(struct ttm_buffer_object *bo,
-			     struct ttm_mem_reg *mem, unsigned num_pages,
-			     uint64_t offset, unsigned window,
-			     struct amdgpu_ring *ring,
-			     uint64_t *addr);
-
-static int amdgpu_ttm_debugfs_init(struct amdgpu_device *adev);
-static void amdgpu_ttm_debugfs_fini(struct amdgpu_device *adev);
-
-/*
- * Global memory.
- */
-static int amdgpu_ttm_mem_global_init(struct drm_global_reference *ref)
-{
-	return ttm_mem_global_init(ref->object);
-}
+#include <linux/dma-buf.h>
+#include <linux/sizes.h>
+#include <linux/module.h>
 
-static void amdgpu_ttm_mem_global_release(struct drm_global_reference *ref)
-{
-	ttm_mem_global_release(ref->object);
-}
-
-static int amdgpu_ttm_global_init(struct amdgpu_device *adev)
-{
-	struct drm_global_reference *global_ref;
-	struct amdgpu_ring *ring;
-	struct amd_sched_rq *rq;
-	int r;
-
-	adev->mman.mem_global_referenced = false;
-	global_ref = &adev->mman.mem_global_ref;
-	global_ref->global_type = DRM_GLOBAL_TTM_MEM;
-	global_ref->size = sizeof(struct ttm_mem_global);
-	global_ref->init = &amdgpu_ttm_mem_global_init;
-	global_ref->release = &amdgpu_ttm_mem_global_release;
-	r = drm_global_item_ref(global_ref);
-	if (r) {
-		DRM_ERROR("Failed setting up TTM memory accounting "
-			  "subsystem.\n");
-		goto error_mem;
-	}
-
-	adev->mman.bo_global_ref.mem_glob =
-		adev->mman.mem_global_ref.object;
-	global_ref = &adev->mman.bo_global_ref.ref;
-	global_ref->global_type = DRM_GLOBAL_TTM_BO;
-	global_ref->size = sizeof(struct ttm_bo_global);
-	global_ref->init = &ttm_bo_global_init;
-	global_ref->release = &ttm_bo_global_release;
-	r = drm_global_item_ref(global_ref);
-	if (r) {
-		DRM_ERROR("Failed setting up TTM BO subsystem.\n");
-		goto error_bo;
-	}
-
-	mutex_init(&adev->mman.gtt_window_lock);
-
-	ring = adev->mman.buffer_funcs_ring;
-	rq = &ring->sched.sched_rq[AMD_SCHED_PRIORITY_KERNEL];
-	r = amd_sched_entity_init(&ring->sched, &adev->mman.entity,
-				  rq, amdgpu_sched_jobs);
-	if (r) {
-		DRM_ERROR("Failed setting up TTM BO move run queue.\n");
-		goto error_entity;
-	}
+#include <drm/drm_drv.h>
+#include <drm/ttm/ttm_bo.h>
+#include <drm/ttm/ttm_placement.h>
+#include <drm/ttm/ttm_range_manager.h>
+#include <drm/ttm/ttm_tt.h>
 
-	adev->mman.mem_global_referenced = true;
+#include <drm/amdgpu_drm.h>
 
-	return 0;
+#include "amdgpu.h"
+#include "amdgpu_object.h"
+#include "amdgpu_trace.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_sdma.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_hmm.h"
+#include "amdgpu_atomfirmware.h"
+#include "amdgpu_res_cursor.h"
+#include "bif/bif_4_1_d.h"
 
-error_entity:
-	drm_global_item_unref(&adev->mman.bo_global_ref.ref);
-error_bo:
-	drm_global_item_unref(&adev->mman.mem_global_ref);
-error_mem:
-	return r;
-}
+MODULE_IMPORT_NS("DMA_BUF");
 
-static void amdgpu_ttm_global_fini(struct amdgpu_device *adev)
-{
-	if (adev->mman.mem_global_referenced) {
-		amd_sched_entity_fini(adev->mman.entity.sched,
-				      &adev->mman.entity);
-		mutex_destroy(&adev->mman.gtt_window_lock);
-		drm_global_item_unref(&adev->mman.bo_global_ref.ref);
-		drm_global_item_unref(&adev->mman.mem_global_ref);
-		adev->mman.mem_global_referenced = false;
-	}
-}
+#define AMDGPU_TTM_VRAM_MAX_DW_READ	((size_t)128)
 
-static int amdgpu_invalidate_caches(struct ttm_bo_device *bdev, uint32_t flags)
-{
-	return 0;
-}
+static int amdgpu_ttm_backend_bind(struct ttm_device *bdev,
+				   struct ttm_tt *ttm,
+				   struct ttm_resource *bo_mem);
+static void amdgpu_ttm_backend_unbind(struct ttm_device *bdev,
+				      struct ttm_tt *ttm);
 
-static int amdgpu_init_mem_type(struct ttm_bo_device *bdev, uint32_t type,
-				struct ttm_mem_type_manager *man)
+static int amdgpu_ttm_init_on_chip(struct amdgpu_device *adev,
+				    unsigned int type,
+				    uint64_t size_in_page)
 {
-	struct amdgpu_device *adev;
-
-	adev = amdgpu_ttm_adev(bdev);
-
-	switch (type) {
-	case TTM_PL_SYSTEM:
-		/* System memory */
-		man->flags = TTM_MEMTYPE_FLAG_MAPPABLE;
-		man->available_caching = TTM_PL_MASK_CACHING;
-		man->default_caching = TTM_PL_FLAG_CACHED;
-		break;
-	case TTM_PL_TT:
-		man->func = &amdgpu_gtt_mgr_func;
-		man->gpu_offset = adev->mc.gart_start;
-		man->available_caching = TTM_PL_MASK_CACHING;
-		man->default_caching = TTM_PL_FLAG_CACHED;
-		man->flags = TTM_MEMTYPE_FLAG_MAPPABLE | TTM_MEMTYPE_FLAG_CMA;
-		break;
-	case TTM_PL_VRAM:
-		/* "On-card" video ram */
-		man->func = &amdgpu_vram_mgr_func;
-		man->gpu_offset = adev->mc.vram_start;
-		man->flags = TTM_MEMTYPE_FLAG_FIXED |
-			     TTM_MEMTYPE_FLAG_MAPPABLE;
-		man->available_caching = TTM_PL_FLAG_UNCACHED | TTM_PL_FLAG_WC;
-		man->default_caching = TTM_PL_FLAG_WC;
-		break;
-	case AMDGPU_PL_GDS:
-	case AMDGPU_PL_GWS:
-	case AMDGPU_PL_OA:
-		/* On-chip GDS memory*/
-		man->func = &ttm_bo_manager_func;
-		man->gpu_offset = 0;
-		man->flags = TTM_MEMTYPE_FLAG_FIXED | TTM_MEMTYPE_FLAG_CMA;
-		man->available_caching = TTM_PL_FLAG_UNCACHED;
-		man->default_caching = TTM_PL_FLAG_UNCACHED;
-		break;
-	default:
-		DRM_ERROR("Unsupported memory type %u\n", (unsigned)type);
-		return -EINVAL;
-	}
-	return 0;
+	return ttm_range_man_init(&adev->mman.bdev, type,
+				  false, size_in_page);
 }
 
+/**
+ * amdgpu_evict_flags - Compute placement flags
+ *
+ * @bo: The buffer object to evict
+ * @placement: Possible destination(s) for evicted BO
+ *
+ * Fill in placement data when ttm_bo_evict() is called
+ */
 static void amdgpu_evict_flags(struct ttm_buffer_object *bo,
 				struct ttm_placement *placement)
 {
@@ -198,334 +95,462 @@ static void amdgpu_evict_flags(struct ttm_buffer_object *bo,
 	static const struct ttm_place placements = {
 		.fpfn = 0,
 		.lpfn = 0,
-		.flags = TTM_PL_MASK_CACHING | TTM_PL_FLAG_SYSTEM
+		.mem_type = TTM_PL_SYSTEM,
+		.flags = 0
 	};
 
-	if (!amdgpu_ttm_bo_is_amdgpu_bo(bo)) {
+	/* Don't handle scatter gather BOs */
+	if (bo->type == ttm_bo_type_sg) {
+		placement->num_placement = 0;
+		return;
+	}
+
+	/* Object isn't an AMDGPU object so ignore */
+	if (!amdgpu_bo_is_amdgpu_bo(bo)) {
 		placement->placement = &placements;
-		placement->busy_placement = &placements;
 		placement->num_placement = 1;
-		placement->num_busy_placement = 1;
 		return;
 	}
-	abo = container_of(bo, struct amdgpu_bo, tbo);
-	switch (bo->mem.mem_type) {
+
+	abo = ttm_to_amdgpu_bo(bo);
+	if (abo->flags & AMDGPU_GEM_CREATE_DISCARDABLE) {
+		placement->num_placement = 0;
+		return;
+	}
+
+	switch (bo->resource->mem_type) {
+	case AMDGPU_PL_GDS:
+	case AMDGPU_PL_GWS:
+	case AMDGPU_PL_OA:
+	case AMDGPU_PL_DOORBELL:
+	case AMDGPU_PL_MMIO_REMAP:
+		placement->num_placement = 0;
+		return;
+
 	case TTM_PL_VRAM:
-		if (adev->mman.buffer_funcs &&
-		    adev->mman.buffer_funcs_ring &&
-		    adev->mman.buffer_funcs_ring->ready == false) {
-			amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_CPU);
-		} else if (adev->mc.visible_vram_size < adev->mc.real_vram_size &&
-			   !(abo->flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED)) {
-			unsigned fpfn = adev->mc.visible_vram_size >> PAGE_SHIFT;
-			struct drm_mm_node *node = bo->mem.mm_node;
-			unsigned long pages_left;
-
-			for (pages_left = bo->mem.num_pages;
-			     pages_left;
-			     pages_left -= node->size, node++) {
-				if (node->start < fpfn)
-					break;
-			}
+		if (!adev->mman.buffer_funcs_enabled) {
+			/* Move to system memory */
+			amdgpu_bo_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_CPU);
 
-			if (!pages_left)
-				goto gtt;
+		} else if (!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
+			   !(abo->flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED) &&
+			   amdgpu_res_cpu_visible(adev, bo->resource)) {
 
 			/* Try evicting to the CPU inaccessible part of VRAM
 			 * first, but only set GTT as busy placement, so this
 			 * BO will be evicted to GTT rather than causing other
 			 * BOs to be evicted from VRAM
 			 */
-			amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_VRAM |
-							 AMDGPU_GEM_DOMAIN_GTT);
-			abo->placements[0].fpfn = fpfn;
+			amdgpu_bo_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_VRAM |
+							AMDGPU_GEM_DOMAIN_GTT |
+							AMDGPU_GEM_DOMAIN_CPU);
+			abo->placements[0].fpfn = adev->gmc.visible_vram_size >> PAGE_SHIFT;
 			abo->placements[0].lpfn = 0;
-			abo->placement.busy_placement = &abo->placements[1];
-			abo->placement.num_busy_placement = 1;
+			abo->placements[0].flags |= TTM_PL_FLAG_DESIRED;
 		} else {
-gtt:
-			amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_GTT);
+			/* Move to GTT memory */
+			amdgpu_bo_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_GTT |
+							AMDGPU_GEM_DOMAIN_CPU);
 		}
 		break;
 	case TTM_PL_TT:
+	case AMDGPU_PL_PREEMPT:
 	default:
-		amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_CPU);
+		amdgpu_bo_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_CPU);
+		break;
 	}
 	*placement = abo->placement;
 }
 
-static int amdgpu_verify_access(struct ttm_buffer_object *bo, struct file *filp)
+/**
+ * amdgpu_ttm_map_buffer - Map memory into the GART windows
+ * @bo: buffer object to map
+ * @mem: memory object to map
+ * @mm_cur: range to map
+ * @window: which GART window to use
+ * @ring: DMA ring to use for the copy
+ * @tmz: if we should setup a TMZ enabled mapping
+ * @size: in number of bytes to map, out number of bytes mapped
+ * @addr: resulting address inside the MC address space
+ *
+ * Setup one of the GART windows to access a specific piece of memory or return
+ * the physical address for local memory.
+ */
+static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo,
+				 struct ttm_resource *mem,
+				 struct amdgpu_res_cursor *mm_cur,
+				 unsigned int window, struct amdgpu_ring *ring,
+				 bool tmz, uint64_t *size, uint64_t *addr)
 {
-	struct amdgpu_bo *abo = container_of(bo, struct amdgpu_bo, tbo);
+	struct amdgpu_device *adev = ring->adev;
+	unsigned int offset, num_pages, num_dw, num_bytes;
+	uint64_t src_addr, dst_addr;
+	struct amdgpu_job *job;
+	void *cpu_addr;
+	uint64_t flags;
+	unsigned int i;
+	int r;
 
-	if (amdgpu_ttm_tt_get_usermm(bo->ttm))
-		return -EPERM;
-	return drm_vma_node_verify_access(&abo->gem_base.vma_node,
-					  filp->private_data);
-}
+	BUG_ON(adev->mman.buffer_funcs->copy_max_bytes <
+	       AMDGPU_GTT_MAX_TRANSFER_SIZE * 8);
 
-static void amdgpu_move_null(struct ttm_buffer_object *bo,
-			     struct ttm_mem_reg *new_mem)
-{
-	struct ttm_mem_reg *old_mem = &bo->mem;
+	if (WARN_ON(mem->mem_type == AMDGPU_PL_PREEMPT))
+		return -EINVAL;
 
-	BUG_ON(old_mem->mm_node != NULL);
-	*old_mem = *new_mem;
-	new_mem->mm_node = NULL;
-}
+	/* Map only what can't be accessed directly */
+	if (!tmz && mem->start != AMDGPU_BO_INVALID_OFFSET) {
+		*addr = amdgpu_ttm_domain_start(adev, mem->mem_type) +
+			mm_cur->start;
+		return 0;
+	}
 
-static uint64_t amdgpu_mm_node_addr(struct ttm_buffer_object *bo,
-				    struct drm_mm_node *mm_node,
-				    struct ttm_mem_reg *mem)
-{
-	uint64_t addr = 0;
 
-	if (mem->mem_type != TTM_PL_TT ||
-	    amdgpu_gtt_mgr_is_allocated(mem)) {
-		addr = mm_node->start << PAGE_SHIFT;
-		addr += bo->bdev->man[mem->mem_type].gpu_offset;
+	/*
+	 * If start begins at an offset inside the page, then adjust the size
+	 * and addr accordingly
+	 */
+	offset = mm_cur->start & ~PAGE_MASK;
+
+	num_pages = PFN_UP(*size + offset);
+	num_pages = min_t(uint32_t, num_pages, AMDGPU_GTT_MAX_TRANSFER_SIZE);
+
+	*size = min(*size, (uint64_t)num_pages * PAGE_SIZE - offset);
+
+	*addr = adev->gmc.gart_start;
+	*addr += (u64)window * AMDGPU_GTT_MAX_TRANSFER_SIZE *
+		AMDGPU_GPU_PAGE_SIZE;
+	*addr += offset;
+
+	num_dw = ALIGN(adev->mman.buffer_funcs->copy_num_dw, 8);
+	num_bytes = num_pages * 8 * AMDGPU_GPU_PAGES_IN_CPU_PAGE;
+
+	r = amdgpu_job_alloc_with_ib(adev, &adev->mman.high_pr,
+				     AMDGPU_FENCE_OWNER_UNDEFINED,
+				     num_dw * 4 + num_bytes,
+				     AMDGPU_IB_POOL_DELAYED, &job,
+				     AMDGPU_KERNEL_JOB_ID_TTM_MAP_BUFFER);
+	if (r)
+		return r;
+
+	src_addr = num_dw * 4;
+	src_addr += job->ibs[0].gpu_addr;
+
+	dst_addr = amdgpu_bo_gpu_offset(adev->gart.bo);
+	dst_addr += window * AMDGPU_GTT_MAX_TRANSFER_SIZE * 8;
+	amdgpu_emit_copy_buffer(adev, &job->ibs[0], src_addr,
+				dst_addr, num_bytes, 0);
+
+	amdgpu_ring_pad_ib(ring, &job->ibs[0]);
+	WARN_ON(job->ibs[0].length_dw > num_dw);
+
+	flags = amdgpu_ttm_tt_pte_flags(adev, bo->ttm, mem);
+	if (tmz)
+		flags |= AMDGPU_PTE_TMZ;
+
+	cpu_addr = &job->ibs[0].ptr[num_dw];
+
+	if (mem->mem_type == TTM_PL_TT) {
+		dma_addr_t *dma_addr;
+
+		dma_addr = &bo->ttm->dma_address[mm_cur->start >> PAGE_SHIFT];
+		amdgpu_gart_map(adev, 0, num_pages, dma_addr, flags, cpu_addr);
+	} else {
+		dma_addr_t dma_address;
+
+		dma_address = mm_cur->start;
+		dma_address += adev->vm_manager.vram_base_offset;
+
+		for (i = 0; i < num_pages; ++i) {
+			amdgpu_gart_map(adev, i << PAGE_SHIFT, 1, &dma_address,
+					flags, cpu_addr);
+			dma_address += PAGE_SIZE;
+		}
 	}
-	return addr;
+
+	dma_fence_put(amdgpu_job_submit(job));
+	return 0;
 }
 
-static int amdgpu_move_blit(struct ttm_buffer_object *bo,
-			    bool evict, bool no_wait_gpu,
-			    struct ttm_mem_reg *new_mem,
-			    struct ttm_mem_reg *old_mem)
+/**
+ * amdgpu_ttm_copy_mem_to_mem - Helper function for copy
+ * @adev: amdgpu device
+ * @src: buffer/address where to read from
+ * @dst: buffer/address where to write to
+ * @size: number of bytes to copy
+ * @tmz: if a secure copy should be used
+ * @resv: resv object to sync to
+ * @f: Returns the last fence if multiple jobs are submitted.
+ *
+ * The function copies @size bytes from {src->mem + src->offset} to
+ * {dst->mem + dst->offset}. src->bo and dst->bo could be same BO for a
+ * move and different for a BO to BO copy.
+ *
+ */
+int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
+			       const struct amdgpu_copy_mem *src,
+			       const struct amdgpu_copy_mem *dst,
+			       uint64_t size, bool tmz,
+			       struct dma_resv *resv,
+			       struct dma_fence **f)
 {
-	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev);
 	struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
-
-	struct drm_mm_node *old_mm, *new_mm;
-	uint64_t old_start, old_size, new_start, new_size;
-	unsigned long num_pages;
+	struct amdgpu_res_cursor src_mm, dst_mm;
 	struct dma_fence *fence = NULL;
-	int r;
-
-	BUILD_BUG_ON((PAGE_SIZE % AMDGPU_GPU_PAGE_SIZE) != 0);
+	int r = 0;
+	uint32_t copy_flags = 0;
+	struct amdgpu_bo *abo_src, *abo_dst;
 
-	if (!ring->ready) {
-		DRM_ERROR("Trying to move memory with ring turned off.\n");
+	if (!adev->mman.buffer_funcs_enabled) {
+		dev_err(adev->dev,
+			"Trying to move memory with ring turned off.\n");
 		return -EINVAL;
 	}
 
-	old_mm = old_mem->mm_node;
-	old_size = old_mm->size;
-	old_start = amdgpu_mm_node_addr(bo, old_mm, old_mem);
+	amdgpu_res_first(src->mem, src->offset, size, &src_mm);
+	amdgpu_res_first(dst->mem, dst->offset, size, &dst_mm);
 
-	new_mm = new_mem->mm_node;
-	new_size = new_mm->size;
-	new_start = amdgpu_mm_node_addr(bo, new_mm, new_mem);
-
-	num_pages = new_mem->num_pages;
 	mutex_lock(&adev->mman.gtt_window_lock);
-	while (num_pages) {
-		unsigned long cur_pages = min(min(old_size, new_size),
-					      (u64)AMDGPU_GTT_MAX_TRANSFER_SIZE);
-		uint64_t from = old_start, to = new_start;
+	while (src_mm.remaining) {
+		uint64_t from, to, cur_size, tiling_flags;
+		uint32_t num_type, data_format, max_com, write_compress_disable;
 		struct dma_fence *next;
 
-		if (old_mem->mem_type == TTM_PL_TT &&
-		    !amdgpu_gtt_mgr_is_allocated(old_mem)) {
-			r = amdgpu_map_buffer(bo, old_mem, cur_pages,
-					      old_start, 0, ring, &from);
-			if (r)
-				goto error;
-		}
+		/* Never copy more than 256MiB at once to avoid a timeout */
+		cur_size = min3(src_mm.size, dst_mm.size, 256ULL << 20);
 
-		if (new_mem->mem_type == TTM_PL_TT &&
-		    !amdgpu_gtt_mgr_is_allocated(new_mem)) {
-			r = amdgpu_map_buffer(bo, new_mem, cur_pages,
-					      new_start, 1, ring, &to);
-			if (r)
-				goto error;
+		/* Map src to window 0 and dst to window 1. */
+		r = amdgpu_ttm_map_buffer(src->bo, src->mem, &src_mm,
+					  0, ring, tmz, &cur_size, &from);
+		if (r)
+			goto error;
+
+		r = amdgpu_ttm_map_buffer(dst->bo, dst->mem, &dst_mm,
+					  1, ring, tmz, &cur_size, &to);
+		if (r)
+			goto error;
+
+		abo_src = ttm_to_amdgpu_bo(src->bo);
+		abo_dst = ttm_to_amdgpu_bo(dst->bo);
+		if (tmz)
+			copy_flags |= AMDGPU_COPY_FLAGS_TMZ;
+		if ((abo_src->flags & AMDGPU_GEM_CREATE_GFX12_DCC) &&
+		    (abo_src->tbo.resource->mem_type == TTM_PL_VRAM))
+			copy_flags |= AMDGPU_COPY_FLAGS_READ_DECOMPRESSED;
+		if ((abo_dst->flags & AMDGPU_GEM_CREATE_GFX12_DCC) &&
+		    (dst->mem->mem_type == TTM_PL_VRAM)) {
+			copy_flags |= AMDGPU_COPY_FLAGS_WRITE_COMPRESSED;
+			amdgpu_bo_get_tiling_flags(abo_dst, &tiling_flags);
+			max_com = AMDGPU_TILING_GET(tiling_flags, GFX12_DCC_MAX_COMPRESSED_BLOCK);
+			num_type = AMDGPU_TILING_GET(tiling_flags, GFX12_DCC_NUMBER_TYPE);
+			data_format = AMDGPU_TILING_GET(tiling_flags, GFX12_DCC_DATA_FORMAT);
+			write_compress_disable =
+				AMDGPU_TILING_GET(tiling_flags, GFX12_DCC_WRITE_COMPRESS_DISABLE);
+			copy_flags |= (AMDGPU_COPY_FLAGS_SET(MAX_COMPRESSED, max_com) |
+				       AMDGPU_COPY_FLAGS_SET(NUMBER_TYPE, num_type) |
+				       AMDGPU_COPY_FLAGS_SET(DATA_FORMAT, data_format) |
+				       AMDGPU_COPY_FLAGS_SET(WRITE_COMPRESS_DISABLE,
+							     write_compress_disable));
 		}
 
-		r = amdgpu_copy_buffer(ring, from, to,
-				       cur_pages * PAGE_SIZE,
-				       bo->resv, &next, false, true);
+		r = amdgpu_copy_buffer(ring, from, to, cur_size, resv,
+				       &next, false, true, copy_flags);
 		if (r)
 			goto error;
 
 		dma_fence_put(fence);
 		fence = next;
 
-		num_pages -= cur_pages;
-		if (!num_pages)
-			break;
+		amdgpu_res_next(&src_mm, cur_size);
+		amdgpu_res_next(&dst_mm, cur_size);
+	}
+error:
+	mutex_unlock(&adev->mman.gtt_window_lock);
+	if (f)
+		*f = dma_fence_get(fence);
+	dma_fence_put(fence);
+	return r;
+}
 
-		old_size -= cur_pages;
-		if (!old_size) {
-			old_start = amdgpu_mm_node_addr(bo, ++old_mm, old_mem);
-			old_size = old_mm->size;
-		} else {
-			old_start += cur_pages * PAGE_SIZE;
-		}
+/*
+ * amdgpu_move_blit - Copy an entire buffer to another buffer
+ *
+ * This is a helper called by amdgpu_bo_move() and amdgpu_move_vram_ram() to
+ * help move buffers to and from VRAM.
+ */
+static int amdgpu_move_blit(struct ttm_buffer_object *bo,
+			    bool evict,
+			    struct ttm_resource *new_mem,
+			    struct ttm_resource *old_mem)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev);
+	struct amdgpu_bo *abo = ttm_to_amdgpu_bo(bo);
+	struct amdgpu_copy_mem src, dst;
+	struct dma_fence *fence = NULL;
+	int r;
 
-		new_size -= cur_pages;
-		if (!new_size) {
-			new_start = amdgpu_mm_node_addr(bo, ++new_mm, new_mem);
-			new_size = new_mm->size;
-		} else {
-			new_start += cur_pages * PAGE_SIZE;
+	src.bo = bo;
+	dst.bo = bo;
+	src.mem = old_mem;
+	dst.mem = new_mem;
+	src.offset = 0;
+	dst.offset = 0;
+
+	r = amdgpu_ttm_copy_mem_to_mem(adev, &src, &dst,
+				       new_mem->size,
+				       amdgpu_bo_encrypted(abo),
+				       bo->base.resv, &fence);
+	if (r)
+		goto error;
+
+	/* clear the space being freed */
+	if (old_mem->mem_type == TTM_PL_VRAM &&
+	    (abo->flags & AMDGPU_GEM_CREATE_VRAM_WIPE_ON_RELEASE)) {
+		struct dma_fence *wipe_fence = NULL;
+
+		r = amdgpu_fill_buffer(abo, 0, NULL, &wipe_fence,
+				       false, AMDGPU_KERNEL_JOB_ID_MOVE_BLIT);
+		if (r) {
+			goto error;
+		} else if (wipe_fence) {
+			amdgpu_vram_mgr_set_cleared(bo->resource);
+			dma_fence_put(fence);
+			fence = wipe_fence;
 		}
 	}
-	mutex_unlock(&adev->mman.gtt_window_lock);
 
-	r = ttm_bo_pipeline_move(bo, fence, evict, new_mem);
+	/* Always block for VM page tables before committing the new location */
+	if (bo->type == ttm_bo_type_kernel)
+		r = ttm_bo_move_accel_cleanup(bo, fence, true, false, new_mem);
+	else
+		r = ttm_bo_move_accel_cleanup(bo, fence, evict, true, new_mem);
 	dma_fence_put(fence);
 	return r;
 
 error:
-	mutex_unlock(&adev->mman.gtt_window_lock);
-
 	if (fence)
 		dma_fence_wait(fence, false);
 	dma_fence_put(fence);
 	return r;
 }
 
-static int amdgpu_move_vram_ram(struct ttm_buffer_object *bo,
-				bool evict, bool interruptible,
-				bool no_wait_gpu,
-				struct ttm_mem_reg *new_mem)
+/**
+ * amdgpu_res_cpu_visible - Check that resource can be accessed by CPU
+ * @adev: amdgpu device
+ * @res: the resource to check
+ *
+ * Returns: true if the full resource is CPU visible, false otherwise.
+ */
+bool amdgpu_res_cpu_visible(struct amdgpu_device *adev,
+			    struct ttm_resource *res)
 {
-	struct amdgpu_device *adev;
-	struct ttm_mem_reg *old_mem = &bo->mem;
-	struct ttm_mem_reg tmp_mem;
-	struct ttm_place placements;
-	struct ttm_placement placement;
-	int r;
+	struct amdgpu_res_cursor cursor;
 
-	adev = amdgpu_ttm_adev(bo->bdev);
-	tmp_mem = *new_mem;
-	tmp_mem.mm_node = NULL;
-	placement.num_placement = 1;
-	placement.placement = &placements;
-	placement.num_busy_placement = 1;
-	placement.busy_placement = &placements;
-	placements.fpfn = 0;
-	placements.lpfn = 0;
-	placements.flags = TTM_PL_MASK_CACHING | TTM_PL_FLAG_TT;
-	r = ttm_bo_mem_space(bo, &placement, &tmp_mem,
-			     interruptible, no_wait_gpu);
-	if (unlikely(r)) {
-		return r;
-	}
+	if (!res)
+		return false;
 
-	r = ttm_tt_set_placement_caching(bo->ttm, tmp_mem.placement);
-	if (unlikely(r)) {
-		goto out_cleanup;
-	}
+	if (res->mem_type == TTM_PL_SYSTEM || res->mem_type == TTM_PL_TT ||
+	    res->mem_type == AMDGPU_PL_PREEMPT || res->mem_type == AMDGPU_PL_DOORBELL ||
+	    res->mem_type == AMDGPU_PL_MMIO_REMAP)
+		return true;
 
-	r = ttm_tt_bind(bo->ttm, &tmp_mem);
-	if (unlikely(r)) {
-		goto out_cleanup;
-	}
-	r = amdgpu_move_blit(bo, true, no_wait_gpu, &tmp_mem, old_mem);
-	if (unlikely(r)) {
-		goto out_cleanup;
+	if (res->mem_type != TTM_PL_VRAM)
+		return false;
+
+	amdgpu_res_first(res, 0, res->size, &cursor);
+	while (cursor.remaining) {
+		if ((cursor.start + cursor.size) > adev->gmc.visible_vram_size)
+			return false;
+		amdgpu_res_next(&cursor, cursor.size);
 	}
-	r = ttm_bo_move_ttm(bo, interruptible, no_wait_gpu, new_mem);
-out_cleanup:
-	ttm_bo_mem_put(bo, &tmp_mem);
-	return r;
+
+	return true;
 }
 
-static int amdgpu_move_ram_vram(struct ttm_buffer_object *bo,
-				bool evict, bool interruptible,
-				bool no_wait_gpu,
-				struct ttm_mem_reg *new_mem)
+/*
+ * amdgpu_res_copyable - Check that memory can be accessed by ttm_bo_move_memcpy
+ *
+ * Called by amdgpu_bo_move()
+ */
+static bool amdgpu_res_copyable(struct amdgpu_device *adev,
+				struct ttm_resource *mem)
 {
-	struct amdgpu_device *adev;
-	struct ttm_mem_reg *old_mem = &bo->mem;
-	struct ttm_mem_reg tmp_mem;
-	struct ttm_placement placement;
-	struct ttm_place placements;
-	int r;
+	if (!amdgpu_res_cpu_visible(adev, mem))
+		return false;
 
-	adev = amdgpu_ttm_adev(bo->bdev);
-	tmp_mem = *new_mem;
-	tmp_mem.mm_node = NULL;
-	placement.num_placement = 1;
-	placement.placement = &placements;
-	placement.num_busy_placement = 1;
-	placement.busy_placement = &placements;
-	placements.fpfn = 0;
-	placements.lpfn = 0;
-	placements.flags = TTM_PL_MASK_CACHING | TTM_PL_FLAG_TT;
-	r = ttm_bo_mem_space(bo, &placement, &tmp_mem,
-			     interruptible, no_wait_gpu);
-	if (unlikely(r)) {
-		return r;
-	}
-	r = ttm_bo_move_ttm(bo, interruptible, no_wait_gpu, &tmp_mem);
-	if (unlikely(r)) {
-		goto out_cleanup;
-	}
-	r = amdgpu_move_blit(bo, true, no_wait_gpu, new_mem, old_mem);
-	if (unlikely(r)) {
-		goto out_cleanup;
-	}
-out_cleanup:
-	ttm_bo_mem_put(bo, &tmp_mem);
-	return r;
+	/* ttm_resource_ioremap only supports contiguous memory */
+	if (mem->mem_type == TTM_PL_VRAM &&
+	    !(mem->placement & TTM_PL_FLAG_CONTIGUOUS))
+		return false;
+
+	return true;
 }
 
-static int amdgpu_bo_move(struct ttm_buffer_object *bo,
-			bool evict, bool interruptible,
-			bool no_wait_gpu,
-			struct ttm_mem_reg *new_mem)
+/*
+ * amdgpu_bo_move - Move a buffer object to a new memory location
+ *
+ * Called by ttm_bo_handle_move_mem()
+ */
+static int amdgpu_bo_move(struct ttm_buffer_object *bo, bool evict,
+			  struct ttm_operation_ctx *ctx,
+			  struct ttm_resource *new_mem,
+			  struct ttm_place *hop)
 {
 	struct amdgpu_device *adev;
 	struct amdgpu_bo *abo;
-	struct ttm_mem_reg *old_mem = &bo->mem;
+	struct ttm_resource *old_mem = bo->resource;
 	int r;
 
-	/* Can't move a pinned BO */
-	abo = container_of(bo, struct amdgpu_bo, tbo);
-	if (WARN_ON_ONCE(abo->pin_count > 0))
-		return -EINVAL;
+	if (new_mem->mem_type == TTM_PL_TT ||
+	    new_mem->mem_type == AMDGPU_PL_PREEMPT) {
+		r = amdgpu_ttm_backend_bind(bo->bdev, bo->ttm, new_mem);
+		if (r)
+			return r;
+	}
 
+	abo = ttm_to_amdgpu_bo(bo);
 	adev = amdgpu_ttm_adev(bo->bdev);
 
-	if (old_mem->mem_type == TTM_PL_SYSTEM && bo->ttm == NULL) {
-		amdgpu_move_null(bo, new_mem);
+	if (!old_mem || (old_mem->mem_type == TTM_PL_SYSTEM &&
+			 bo->ttm == NULL)) {
+		amdgpu_bo_move_notify(bo, evict, new_mem);
+		ttm_bo_move_null(bo, new_mem);
 		return 0;
 	}
-	if ((old_mem->mem_type == TTM_PL_TT &&
-	     new_mem->mem_type == TTM_PL_SYSTEM) ||
-	    (old_mem->mem_type == TTM_PL_SYSTEM &&
-	     new_mem->mem_type == TTM_PL_TT)) {
-		/* bind is enough */
-		amdgpu_move_null(bo, new_mem);
+	if (old_mem->mem_type == TTM_PL_SYSTEM &&
+	    (new_mem->mem_type == TTM_PL_TT ||
+	     new_mem->mem_type == AMDGPU_PL_PREEMPT)) {
+		amdgpu_bo_move_notify(bo, evict, new_mem);
+		ttm_bo_move_null(bo, new_mem);
 		return 0;
 	}
-	if (adev->mman.buffer_funcs == NULL ||
-	    adev->mman.buffer_funcs_ring == NULL ||
-	    !adev->mman.buffer_funcs_ring->ready) {
-		/* use memcpy */
-		goto memcpy;
-	}
-
-	if (old_mem->mem_type == TTM_PL_VRAM &&
+	if ((old_mem->mem_type == TTM_PL_TT ||
+	     old_mem->mem_type == AMDGPU_PL_PREEMPT) &&
 	    new_mem->mem_type == TTM_PL_SYSTEM) {
-		r = amdgpu_move_vram_ram(bo, evict, interruptible,
-					no_wait_gpu, new_mem);
-	} else if (old_mem->mem_type == TTM_PL_SYSTEM &&
-		   new_mem->mem_type == TTM_PL_VRAM) {
-		r = amdgpu_move_ram_vram(bo, evict, interruptible,
-					    no_wait_gpu, new_mem);
-	} else {
-		r = amdgpu_move_blit(bo, evict, no_wait_gpu, new_mem, old_mem);
+		r = ttm_bo_wait_ctx(bo, ctx);
+		if (r)
+			return r;
+
+		amdgpu_ttm_backend_unbind(bo->bdev, bo->ttm);
+		amdgpu_bo_move_notify(bo, evict, new_mem);
+		ttm_resource_free(bo, &bo->resource);
+		ttm_bo_assign_mem(bo, new_mem);
+		return 0;
 	}
 
-	if (r) {
-memcpy:
-		r = ttm_bo_move_memcpy(bo, interruptible, no_wait_gpu, new_mem);
-		if (r) {
-			return r;
-		}
+	if (old_mem->mem_type == AMDGPU_PL_GDS ||
+	    old_mem->mem_type == AMDGPU_PL_GWS ||
+	    old_mem->mem_type == AMDGPU_PL_OA ||
+	    old_mem->mem_type == AMDGPU_PL_DOORBELL ||
+	    old_mem->mem_type == AMDGPU_PL_MMIO_REMAP ||
+	    new_mem->mem_type == AMDGPU_PL_GDS ||
+	    new_mem->mem_type == AMDGPU_PL_GWS ||
+	    new_mem->mem_type == AMDGPU_PL_OA ||
+	    new_mem->mem_type == AMDGPU_PL_DOORBELL ||
+	    new_mem->mem_type == AMDGPU_PL_MMIO_REMAP) {
+		/* Nothing to save here */
+		amdgpu_bo_move_notify(bo, evict, new_mem);
+		ttm_bo_move_null(bo, new_mem);
+		return 0;
 	}
 
 	if (bo->type == ttm_bo_type_device &&
@@ -537,36 +562,83 @@ memcpy:
 		abo->flags &= ~AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
 	}
 
-	/* update statistics */
-	atomic64_add((u64)bo->num_pages << PAGE_SHIFT, &adev->num_bytes_moved);
+	if (adev->mman.buffer_funcs_enabled &&
+	    ((old_mem->mem_type == TTM_PL_SYSTEM &&
+	      new_mem->mem_type == TTM_PL_VRAM) ||
+	     (old_mem->mem_type == TTM_PL_VRAM &&
+	      new_mem->mem_type == TTM_PL_SYSTEM))) {
+		hop->fpfn = 0;
+		hop->lpfn = 0;
+		hop->mem_type = TTM_PL_TT;
+		hop->flags = TTM_PL_FLAG_TEMPORARY;
+		return -EMULTIHOP;
+	}
+
+	amdgpu_bo_move_notify(bo, evict, new_mem);
+	if (adev->mman.buffer_funcs_enabled)
+		r = amdgpu_move_blit(bo, evict, new_mem, old_mem);
+	else
+		r = -ENODEV;
+
+	if (r) {
+		/* Check that all memory is CPU accessible */
+		if (!amdgpu_res_copyable(adev, old_mem) ||
+		    !amdgpu_res_copyable(adev, new_mem)) {
+			pr_err("Move buffer fallback to memcpy unavailable\n");
+			return r;
+		}
+
+		r = ttm_bo_move_memcpy(bo, ctx, new_mem);
+		if (r)
+			return r;
+	}
+
+	/* update statistics after the move */
+	if (evict)
+		atomic64_inc(&adev->num_evictions);
+	atomic64_add(bo->base.size, &adev->num_bytes_moved);
 	return 0;
 }
 
-static int amdgpu_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_mem_reg *mem)
+/*
+ * amdgpu_ttm_io_mem_reserve - Reserve a block of memory during a fault
+ *
+ * Called by ttm_mem_io_reserve() ultimately via ttm_bo_vm_fault()
+ */
+static int amdgpu_ttm_io_mem_reserve(struct ttm_device *bdev,
+				     struct ttm_resource *mem)
 {
-	struct ttm_mem_type_manager *man = &bdev->man[mem->mem_type];
 	struct amdgpu_device *adev = amdgpu_ttm_adev(bdev);
 
-	mem->bus.addr = NULL;
-	mem->bus.offset = 0;
-	mem->bus.size = mem->num_pages << PAGE_SHIFT;
-	mem->bus.base = 0;
-	mem->bus.is_iomem = false;
-	if (!(man->flags & TTM_MEMTYPE_FLAG_MAPPABLE))
-		return -EINVAL;
 	switch (mem->mem_type) {
 	case TTM_PL_SYSTEM:
 		/* system memory */
 		return 0;
 	case TTM_PL_TT:
+	case AMDGPU_PL_PREEMPT:
 		break;
 	case TTM_PL_VRAM:
 		mem->bus.offset = mem->start << PAGE_SHIFT;
-		/* check if it's visible */
-		if ((mem->bus.offset + mem->bus.size) > adev->mc.visible_vram_size)
-			return -EINVAL;
-		mem->bus.base = adev->mc.aper_base;
+
+		if (adev->mman.aper_base_kaddr &&
+		    mem->placement & TTM_PL_FLAG_CONTIGUOUS)
+			mem->bus.addr = (u8 *)adev->mman.aper_base_kaddr +
+					mem->bus.offset;
+
+		mem->bus.offset += adev->gmc.aper_base;
+		mem->bus.is_iomem = true;
+		break;
+	case AMDGPU_PL_DOORBELL:
+		mem->bus.offset = mem->start << PAGE_SHIFT;
+		mem->bus.offset += adev->doorbell.base;
+		mem->bus.is_iomem = true;
+		mem->bus.caching = ttm_uncached;
+		break;
+	case AMDGPU_PL_MMIO_REMAP:
+		mem->bus.offset = mem->start << PAGE_SHIFT;
+		mem->bus.offset += adev->rmmio_remap.bus_addr;
 		mem->bus.is_iomem = true;
+		mem->bus.caching = ttm_uncached;
 		break;
 	default:
 		return -EINVAL;
@@ -574,538 +646,704 @@ static int amdgpu_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_mem_
 	return 0;
 }
 
-static void amdgpu_ttm_io_mem_free(struct ttm_bo_device *bdev, struct ttm_mem_reg *mem)
+static unsigned long amdgpu_ttm_io_mem_pfn(struct ttm_buffer_object *bo,
+					   unsigned long page_offset)
 {
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev);
+	struct amdgpu_res_cursor cursor;
+
+	amdgpu_res_first(bo->resource, (u64)page_offset << PAGE_SHIFT, 0,
+			 &cursor);
+
+	if (bo->resource->mem_type == AMDGPU_PL_DOORBELL)
+		return ((uint64_t)(adev->doorbell.base + cursor.start)) >> PAGE_SHIFT;
+	else if (bo->resource->mem_type == AMDGPU_PL_MMIO_REMAP)
+		return ((uint64_t)(adev->rmmio_remap.bus_addr + cursor.start)) >> PAGE_SHIFT;
+
+	return (adev->gmc.aper_base + cursor.start) >> PAGE_SHIFT;
 }
 
-static unsigned long amdgpu_ttm_io_mem_pfn(struct ttm_buffer_object *bo,
-					   unsigned long page_offset)
+/**
+ * amdgpu_ttm_domain_start - Returns GPU start address
+ * @adev: amdgpu device object
+ * @type: type of the memory
+ *
+ * Returns:
+ * GPU start address of a memory domain
+ */
+
+uint64_t amdgpu_ttm_domain_start(struct amdgpu_device *adev, uint32_t type)
 {
-	struct drm_mm_node *mm = bo->mem.mm_node;
-	uint64_t size = mm->size;
-	uint64_t offset = page_offset;
+	switch (type) {
+	case TTM_PL_TT:
+		return adev->gmc.gart_start;
+	case TTM_PL_VRAM:
+		return adev->gmc.vram_start;
+	}
 
-	page_offset = do_div(offset, size);
-	mm += offset;
-	return (bo->mem.bus.base >> PAGE_SHIFT) + mm->start + page_offset;
+	return 0;
 }
 
 /*
  * TTM backend functions.
  */
-struct amdgpu_ttm_gup_task_list {
-	struct list_head	list;
-	struct task_struct	*task;
-};
-
 struct amdgpu_ttm_tt {
-	struct ttm_dma_tt	ttm;
-	struct amdgpu_device	*adev;
+	struct ttm_tt	ttm;
+	struct drm_gem_object	*gobj;
 	u64			offset;
 	uint64_t		userptr;
-	struct mm_struct	*usermm;
+	struct task_struct	*usertask;
 	uint32_t		userflags;
-	spinlock_t              guptasklock;
-	struct list_head        guptasks;
-	atomic_t		mmu_invalidations;
-	struct list_head        list;
+	bool			bound;
+	int32_t			pool_id;
 };
 
-int amdgpu_ttm_tt_get_user_pages(struct ttm_tt *ttm, struct page **pages)
-{
-	struct amdgpu_ttm_tt *gtt = (void *)ttm;
-	unsigned int flags = 0;
-	unsigned pinned = 0;
-	int r;
+#define ttm_to_amdgpu_ttm_tt(ptr)	container_of(ptr, struct amdgpu_ttm_tt, ttm)
 
-	if (!(gtt->userflags & AMDGPU_GEM_USERPTR_READONLY))
-		flags |= FOLL_WRITE;
+#ifdef CONFIG_DRM_AMDGPU_USERPTR
+/*
+ * amdgpu_ttm_tt_get_user_pages - get device accessible pages that back user
+ * memory and start HMM tracking CPU page table update
+ *
+ * Calling function must call amdgpu_ttm_tt_userptr_range_done() once and only
+ * once afterwards to stop HMM tracking
+ */
+int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo,
+				 struct hmm_range **range)
+{
+	struct ttm_tt *ttm = bo->tbo.ttm;
+	struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(ttm);
+	unsigned long start = gtt->userptr;
+	struct vm_area_struct *vma;
+	struct mm_struct *mm;
+	bool readonly;
+	int r = 0;
 
-	if (gtt->userflags & AMDGPU_GEM_USERPTR_ANONONLY) {
-		/* check that we only use anonymous memory
-		   to prevent problems with writeback */
-		unsigned long end = gtt->userptr + ttm->num_pages * PAGE_SIZE;
-		struct vm_area_struct *vma;
+	/* Make sure get_user_pages_done() can cleanup gracefully */
+	*range = NULL;
 
-		vma = find_vma(gtt->usermm, gtt->userptr);
-		if (!vma || vma->vm_file || vma->vm_end < end)
-			return -EPERM;
+	mm = bo->notifier.mm;
+	if (unlikely(!mm)) {
+		DRM_DEBUG_DRIVER("BO is not registered?\n");
+		return -EFAULT;
 	}
 
-	do {
-		unsigned num_pages = ttm->num_pages - pinned;
-		uint64_t userptr = gtt->userptr + pinned * PAGE_SIZE;
-		struct page **p = pages + pinned;
-		struct amdgpu_ttm_gup_task_list guptask;
-
-		guptask.task = current;
-		spin_lock(&gtt->guptasklock);
-		list_add(&guptask.list, &gtt->guptasks);
-		spin_unlock(&gtt->guptasklock);
-
-		r = get_user_pages(userptr, num_pages, flags, p, NULL);
-
-		spin_lock(&gtt->guptasklock);
-		list_del(&guptask.list);
-		spin_unlock(&gtt->guptasklock);
-
-		if (r < 0)
-			goto release_pages;
+	if (!mmget_not_zero(mm)) /* Happens during process shutdown */
+		return -ESRCH;
 
-		pinned += r;
+	mmap_read_lock(mm);
+	vma = vma_lookup(mm, start);
+	if (unlikely(!vma)) {
+		r = -EFAULT;
+		goto out_unlock;
+	}
+	if (unlikely((gtt->userflags & AMDGPU_GEM_USERPTR_ANONONLY) &&
+		vma->vm_file)) {
+		r = -EPERM;
+		goto out_unlock;
+	}
 
-	} while (pinned < ttm->num_pages);
+	readonly = amdgpu_ttm_tt_is_readonly(ttm);
+	r = amdgpu_hmm_range_get_pages(&bo->notifier, start, ttm->num_pages,
+				       readonly, NULL, range);
+out_unlock:
+	mmap_read_unlock(mm);
+	if (r)
+		pr_debug("failed %d to get user pages 0x%lx\n", r, start);
 
-	return 0;
+	mmput(mm);
 
-release_pages:
-	release_pages(pages, pinned, 0);
 	return r;
 }
 
-static void amdgpu_trace_dma_map(struct ttm_tt *ttm)
+/* amdgpu_ttm_tt_discard_user_pages - Discard range and pfn array allocations
+ */
+void amdgpu_ttm_tt_discard_user_pages(struct ttm_tt *ttm,
+				      struct hmm_range *range)
 {
-	struct amdgpu_device *adev = amdgpu_ttm_adev(ttm->bdev);
 	struct amdgpu_ttm_tt *gtt = (void *)ttm;
-	unsigned i;
-
-	if (unlikely(trace_amdgpu_ttm_tt_populate_enabled())) {
-		for (i = 0; i < ttm->num_pages; i++) {
-			trace_amdgpu_ttm_tt_populate(
-				adev,
-				gtt->ttm.dma_address[i],
-				page_to_phys(ttm->pages[i]));
-		}
-	}
+
+	if (gtt && gtt->userptr && range)
+		amdgpu_hmm_range_get_pages_done(range);
 }
 
-static void amdgpu_trace_dma_unmap(struct ttm_tt *ttm)
+/*
+ * amdgpu_ttm_tt_get_user_pages_done - stop HMM track the CPU page table change
+ * Check if the pages backing this ttm range have been invalidated
+ *
+ * Returns: true if pages are still valid
+ */
+bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm,
+				       struct hmm_range *range)
 {
-	struct amdgpu_device *adev = amdgpu_ttm_adev(ttm->bdev);
-	struct amdgpu_ttm_tt *gtt = (void *)ttm;
-	unsigned i;
-
-	if (unlikely(trace_amdgpu_ttm_tt_unpopulate_enabled())) {
-		for (i = 0; i < ttm->num_pages; i++) {
-			trace_amdgpu_ttm_tt_unpopulate(
-				adev,
-				gtt->ttm.dma_address[i],
-				page_to_phys(ttm->pages[i]));
-		}
-	}
+	struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(ttm);
+
+	if (!gtt || !gtt->userptr || !range)
+		return false;
+
+	DRM_DEBUG_DRIVER("user_pages_done 0x%llx pages 0x%x\n",
+		gtt->userptr, ttm->num_pages);
+
+	WARN_ONCE(!range->hmm_pfns, "No user pages to check\n");
+
+	return !amdgpu_hmm_range_get_pages_done(range);
 }
+#endif
 
-/* prepare the sg table with the user pages */
-static int amdgpu_ttm_tt_pin_userptr(struct ttm_tt *ttm)
+/*
+ * amdgpu_ttm_tt_set_user_pages - Copy pages in, putting old pages as necessary.
+ *
+ * Called by amdgpu_cs_list_validate(). This creates the page list
+ * that backs user memory and will ultimately be mapped into the device
+ * address space.
+ */
+void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct hmm_range *range)
 {
-	struct amdgpu_device *adev = amdgpu_ttm_adev(ttm->bdev);
-	struct amdgpu_ttm_tt *gtt = (void *)ttm;
-	unsigned nents;
-	int r;
+	unsigned long i;
+
+	for (i = 0; i < ttm->num_pages; ++i)
+		ttm->pages[i] = range ? hmm_pfn_to_page(range->hmm_pfns[i]) : NULL;
+}
 
+/*
+ * amdgpu_ttm_tt_pin_userptr - prepare the sg table with the user pages
+ *
+ * Called by amdgpu_ttm_backend_bind()
+ **/
+static int amdgpu_ttm_tt_pin_userptr(struct ttm_device *bdev,
+				     struct ttm_tt *ttm)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bdev);
+	struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(ttm);
 	int write = !(gtt->userflags & AMDGPU_GEM_USERPTR_READONLY);
 	enum dma_data_direction direction = write ?
 		DMA_BIDIRECTIONAL : DMA_TO_DEVICE;
+	int r;
 
+	/* Allocate an SG array and squash pages into it */
 	r = sg_alloc_table_from_pages(ttm->sg, ttm->pages, ttm->num_pages, 0,
-				      ttm->num_pages << PAGE_SHIFT,
+				      (u64)ttm->num_pages << PAGE_SHIFT,
 				      GFP_KERNEL);
 	if (r)
 		goto release_sg;
 
-	r = -ENOMEM;
-	nents = dma_map_sg(adev->dev, ttm->sg->sgl, ttm->sg->nents, direction);
-	if (nents != ttm->sg->nents)
-		goto release_sg;
-
-	drm_prime_sg_to_page_addr_arrays(ttm->sg, ttm->pages,
-					 gtt->ttm.dma_address, ttm->num_pages);
+	/* Map SG to device */
+	r = dma_map_sgtable(adev->dev, ttm->sg, direction, 0);
+	if (r)
+		goto release_sg_table;
 
-	amdgpu_trace_dma_map(ttm);
+	/* convert SG to linear array of pages and dma addresses */
+	drm_prime_sg_to_dma_addr_array(ttm->sg, gtt->ttm.dma_address,
+				       ttm->num_pages);
 
 	return 0;
 
+release_sg_table:
+	sg_free_table(ttm->sg);
 release_sg:
 	kfree(ttm->sg);
+	ttm->sg = NULL;
 	return r;
 }
 
-static void amdgpu_ttm_tt_unpin_userptr(struct ttm_tt *ttm)
+/*
+ * amdgpu_ttm_tt_unpin_userptr - Unpin and unmap userptr pages
+ */
+static void amdgpu_ttm_tt_unpin_userptr(struct ttm_device *bdev,
+					struct ttm_tt *ttm)
 {
-	struct amdgpu_device *adev = amdgpu_ttm_adev(ttm->bdev);
-	struct amdgpu_ttm_tt *gtt = (void *)ttm;
-	struct sg_page_iter sg_iter;
-
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bdev);
+	struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(ttm);
 	int write = !(gtt->userflags & AMDGPU_GEM_USERPTR_READONLY);
 	enum dma_data_direction direction = write ?
 		DMA_BIDIRECTIONAL : DMA_TO_DEVICE;
 
 	/* double check that we don't free the table twice */
-	if (!ttm->sg->sgl)
+	if (!ttm->sg || !ttm->sg->sgl)
 		return;
 
-	/* free the sg table and pages again */
-	dma_unmap_sg(adev->dev, ttm->sg->sgl, ttm->sg->nents, direction);
-
-	for_each_sg_page(ttm->sg->sgl, &sg_iter, ttm->sg->nents, 0) {
-		struct page *page = sg_page_iter_page(&sg_iter);
-		if (!(gtt->userflags & AMDGPU_GEM_USERPTR_READONLY))
-			set_page_dirty(page);
+	/* unmap the pages mapped to the device */
+	dma_unmap_sgtable(adev->dev, ttm->sg, direction, 0);
+	sg_free_table(ttm->sg);
+}
 
-		mark_page_accessed(page);
-		put_page(page);
+/*
+ * total_pages is constructed as MQD0+CtrlStack0 + MQD1+CtrlStack1 + ...
+ * MQDn+CtrlStackn where n is the number of XCCs per partition.
+ * pages_per_xcc is the size of one MQD+CtrlStack. The first page is MQD
+ * and uses memory type default, UC. The rest of pages_per_xcc are
+ * Ctrl stack and modify their memory type to NC.
+ */
+static void amdgpu_ttm_gart_bind_gfx9_mqd(struct amdgpu_device *adev,
+				struct ttm_tt *ttm, uint64_t flags)
+{
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	uint64_t total_pages = ttm->num_pages;
+	int num_xcc = max(1U, adev->gfx.num_xcc_per_xcp);
+	uint64_t page_idx, pages_per_xcc;
+	int i;
+	uint64_t ctrl_flags = AMDGPU_PTE_MTYPE_VG10(flags, AMDGPU_MTYPE_NC);
+
+	pages_per_xcc = total_pages;
+	do_div(pages_per_xcc, num_xcc);
+
+	for (i = 0, page_idx = 0; i < num_xcc; i++, page_idx += pages_per_xcc) {
+		/* MQD page: use default flags */
+		amdgpu_gart_bind(adev,
+				gtt->offset + (page_idx << PAGE_SHIFT),
+				1, &gtt->ttm.dma_address[page_idx], flags);
+		/*
+		 * Ctrl pages - modify the memory type to NC (ctrl_flags) from
+		 * the second page of the BO onward.
+		 */
+		amdgpu_gart_bind(adev,
+				gtt->offset + ((page_idx + 1) << PAGE_SHIFT),
+				pages_per_xcc - 1,
+				&gtt->ttm.dma_address[page_idx + 1],
+				ctrl_flags);
 	}
+}
 
-	amdgpu_trace_dma_unmap(ttm);
+static void amdgpu_ttm_gart_bind(struct amdgpu_device *adev,
+				 struct ttm_buffer_object *tbo,
+				 uint64_t flags)
+{
+	struct amdgpu_bo *abo = ttm_to_amdgpu_bo(tbo);
+	struct ttm_tt *ttm = tbo->ttm;
+	struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(ttm);
 
-	sg_free_table(ttm->sg);
+	if (amdgpu_bo_encrypted(abo))
+		flags |= AMDGPU_PTE_TMZ;
+
+	if (abo->flags & AMDGPU_GEM_CREATE_CP_MQD_GFX9) {
+		amdgpu_ttm_gart_bind_gfx9_mqd(adev, ttm, flags);
+	} else {
+		amdgpu_gart_bind(adev, gtt->offset, ttm->num_pages,
+				 gtt->ttm.dma_address, flags);
+	}
+	gtt->bound = true;
 }
 
-static int amdgpu_ttm_backend_bind(struct ttm_tt *ttm,
-				   struct ttm_mem_reg *bo_mem)
+/*
+ * amdgpu_ttm_backend_bind - Bind GTT memory
+ *
+ * Called by ttm_tt_bind() on behalf of ttm_bo_handle_move_mem().
+ * This handles binding GTT memory to the device address space.
+ */
+static int amdgpu_ttm_backend_bind(struct ttm_device *bdev,
+				   struct ttm_tt *ttm,
+				   struct ttm_resource *bo_mem)
 {
-	struct amdgpu_ttm_tt *gtt = (void*)ttm;
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bdev);
+	struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(ttm);
 	uint64_t flags;
-	int r = 0;
+	int r;
+
+	if (!bo_mem)
+		return -EINVAL;
+
+	if (gtt->bound)
+		return 0;
 
 	if (gtt->userptr) {
-		r = amdgpu_ttm_tt_pin_userptr(ttm);
+		r = amdgpu_ttm_tt_pin_userptr(bdev, ttm);
 		if (r) {
-			DRM_ERROR("failed to pin userptr\n");
+			dev_err(adev->dev, "failed to pin userptr\n");
 			return r;
 		}
+	} else if (ttm->page_flags & TTM_TT_FLAG_EXTERNAL) {
+		if (!ttm->sg) {
+			struct dma_buf_attachment *attach;
+			struct sg_table *sgt;
+
+			attach = gtt->gobj->import_attach;
+			sgt = dma_buf_map_attachment(attach, DMA_BIDIRECTIONAL);
+			if (IS_ERR(sgt))
+				return PTR_ERR(sgt);
+
+			ttm->sg = sgt;
+		}
+
+		drm_prime_sg_to_dma_addr_array(ttm->sg, gtt->ttm.dma_address,
+					       ttm->num_pages);
 	}
+
 	if (!ttm->num_pages) {
-		WARN(1, "nothing to bind %lu pages for mreg %p back %p!\n",
+		WARN(1, "nothing to bind %u pages for mreg %p back %p!\n",
 		     ttm->num_pages, bo_mem, ttm);
 	}
 
-	if (bo_mem->mem_type == AMDGPU_PL_GDS ||
-	    bo_mem->mem_type == AMDGPU_PL_GWS ||
-	    bo_mem->mem_type == AMDGPU_PL_OA)
-		return -EINVAL;
-
-	if (!amdgpu_gtt_mgr_is_allocated(bo_mem))
+	if (bo_mem->mem_type != TTM_PL_TT ||
+	    !amdgpu_gtt_mgr_has_gart_addr(bo_mem)) {
+		gtt->offset = AMDGPU_BO_INVALID_OFFSET;
 		return 0;
-
-	spin_lock(&gtt->adev->gtt_list_lock);
-	flags = amdgpu_ttm_tt_pte_flags(gtt->adev, ttm, bo_mem);
-	gtt->offset = (u64)bo_mem->start << PAGE_SHIFT;
-	r = amdgpu_gart_bind(gtt->adev, gtt->offset, ttm->num_pages,
-		ttm->pages, gtt->ttm.dma_address, flags);
-
-	if (r) {
-		DRM_ERROR("failed to bind %lu pages at 0x%08llX\n",
-			  ttm->num_pages, gtt->offset);
-		goto error_gart_bind;
 	}
 
-	list_add_tail(&gtt->list, &gtt->adev->gtt_list);
-error_gart_bind:
-	spin_unlock(&gtt->adev->gtt_list_lock);
-	return r;
-}
+	/* compute PTE flags relevant to this BO memory */
+	flags = amdgpu_ttm_tt_pte_flags(adev, ttm, bo_mem);
 
-bool amdgpu_ttm_is_bound(struct ttm_tt *ttm)
-{
-	struct amdgpu_ttm_tt *gtt = (void *)ttm;
-
-	return gtt && !list_empty(&gtt->list);
+	/* bind pages into GART page tables */
+	gtt->offset = (u64)bo_mem->start << PAGE_SHIFT;
+	amdgpu_gart_bind(adev, gtt->offset, ttm->num_pages,
+			 gtt->ttm.dma_address, flags);
+	gtt->bound = true;
+	return 0;
 }
 
-int amdgpu_ttm_bind(struct ttm_buffer_object *bo, struct ttm_mem_reg *bo_mem)
+/*
+ * amdgpu_ttm_alloc_gart - Make sure buffer object is accessible either
+ * through AGP or GART aperture.
+ *
+ * If bo is accessible through AGP aperture, then use AGP aperture
+ * to access bo; otherwise allocate logical space in GART aperture
+ * and map bo to GART aperture.
+ */
+int amdgpu_ttm_alloc_gart(struct ttm_buffer_object *bo)
 {
 	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev);
-	struct ttm_tt *ttm = bo->ttm;
-	struct ttm_mem_reg tmp;
-
+	struct ttm_operation_ctx ctx = { false, false };
+	struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(bo->ttm);
 	struct ttm_placement placement;
 	struct ttm_place placements;
+	struct ttm_resource *tmp;
+	uint64_t addr, flags;
 	int r;
 
-	if (!ttm || amdgpu_ttm_is_bound(ttm))
+	if (bo->resource->start != AMDGPU_BO_INVALID_OFFSET)
 		return 0;
 
-	tmp = bo->mem;
-	tmp.mm_node = NULL;
+	addr = amdgpu_gmc_agp_addr(bo);
+	if (addr != AMDGPU_BO_INVALID_OFFSET)
+		return 0;
+
+	/* allocate GART space */
 	placement.num_placement = 1;
 	placement.placement = &placements;
-	placement.num_busy_placement = 1;
-	placement.busy_placement = &placements;
 	placements.fpfn = 0;
-	placements.lpfn = adev->mc.gart_size >> PAGE_SHIFT;
-	placements.flags = TTM_PL_MASK_CACHING | TTM_PL_FLAG_TT;
+	placements.lpfn = adev->gmc.gart_size >> PAGE_SHIFT;
+	placements.mem_type = TTM_PL_TT;
+	placements.flags = bo->resource->placement;
 
-	r = ttm_bo_mem_space(bo, &placement, &tmp, true, false);
+	r = ttm_bo_mem_space(bo, &placement, &tmp, &ctx);
 	if (unlikely(r))
 		return r;
 
-	r = ttm_bo_move_ttm(bo, true, false, &tmp);
-	if (unlikely(r))
-		ttm_bo_mem_put(bo, &tmp);
-	else
-		bo->offset = (bo->mem.start << PAGE_SHIFT) +
-			bo->bdev->man[bo->mem.mem_type].gpu_offset;
+	/* compute PTE flags for this buffer object */
+	flags = amdgpu_ttm_tt_pte_flags(adev, bo->ttm, tmp);
 
-	return r;
+	/* Bind pages */
+	gtt->offset = (u64)tmp->start << PAGE_SHIFT;
+	amdgpu_ttm_gart_bind(adev, bo, flags);
+	amdgpu_gart_invalidate_tlb(adev);
+	ttm_resource_free(bo, &bo->resource);
+	ttm_bo_assign_mem(bo, tmp);
+
+	return 0;
 }
 
-int amdgpu_ttm_recover_gart(struct amdgpu_device *adev)
+/*
+ * amdgpu_ttm_recover_gart - Rebind GTT pages
+ *
+ * Called by amdgpu_gtt_mgr_recover() from amdgpu_device_reset() to
+ * rebind GTT pages during a GPU reset.
+ */
+void amdgpu_ttm_recover_gart(struct ttm_buffer_object *tbo)
 {
-	struct amdgpu_ttm_tt *gtt, *tmp;
-	struct ttm_mem_reg bo_mem;
+	struct amdgpu_device *adev = amdgpu_ttm_adev(tbo->bdev);
 	uint64_t flags;
-	int r;
 
-	bo_mem.mem_type = TTM_PL_TT;
-	spin_lock(&adev->gtt_list_lock);
-	list_for_each_entry_safe(gtt, tmp, &adev->gtt_list, list) {
-		flags = amdgpu_ttm_tt_pte_flags(gtt->adev, &gtt->ttm.ttm, &bo_mem);
-		r = amdgpu_gart_bind(adev, gtt->offset, gtt->ttm.ttm.num_pages,
-				     gtt->ttm.ttm.pages, gtt->ttm.dma_address,
-				     flags);
-		if (r) {
-			spin_unlock(&adev->gtt_list_lock);
-			DRM_ERROR("failed to bind %lu pages at 0x%08llX\n",
-				  gtt->ttm.ttm.num_pages, gtt->offset);
-			return r;
-		}
-	}
-	spin_unlock(&adev->gtt_list_lock);
-	return 0;
+	if (!tbo->ttm)
+		return;
+
+	flags = amdgpu_ttm_tt_pte_flags(adev, tbo->ttm, tbo->resource);
+	amdgpu_ttm_gart_bind(adev, tbo, flags);
 }
 
-static int amdgpu_ttm_backend_unbind(struct ttm_tt *ttm)
+/*
+ * amdgpu_ttm_backend_unbind - Unbind GTT mapped pages
+ *
+ * Called by ttm_tt_unbind() on behalf of ttm_bo_move_ttm() and
+ * ttm_tt_destroy().
+ */
+static void amdgpu_ttm_backend_unbind(struct ttm_device *bdev,
+				      struct ttm_tt *ttm)
 {
-	struct amdgpu_ttm_tt *gtt = (void *)ttm;
-	int r;
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bdev);
+	struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(ttm);
 
-	if (gtt->userptr)
-		amdgpu_ttm_tt_unpin_userptr(ttm);
+	/* if the pages have userptr pinning then clear that first */
+	if (gtt->userptr) {
+		amdgpu_ttm_tt_unpin_userptr(bdev, ttm);
+	} else if (ttm->sg && drm_gem_is_imported(gtt->gobj)) {
+		struct dma_buf_attachment *attach;
 
-	if (!amdgpu_ttm_is_bound(ttm))
-		return 0;
+		attach = gtt->gobj->import_attach;
+		dma_buf_unmap_attachment(attach, ttm->sg, DMA_BIDIRECTIONAL);
+		ttm->sg = NULL;
+	}
+
+	if (!gtt->bound)
+		return;
+
+	if (gtt->offset == AMDGPU_BO_INVALID_OFFSET)
+		return;
 
 	/* unbind shouldn't be done for GDS/GWS/OA in ttm_bo_clean_mm */
-	spin_lock(&gtt->adev->gtt_list_lock);
-	r = amdgpu_gart_unbind(gtt->adev, gtt->offset, ttm->num_pages);
-	if (r) {
-		DRM_ERROR("failed to unbind %lu pages at 0x%08llX\n",
-			  gtt->ttm.ttm.num_pages, gtt->offset);
-		goto error_unbind;
-	}
-	list_del_init(&gtt->list);
-error_unbind:
-	spin_unlock(&gtt->adev->gtt_list_lock);
-	return r;
+	amdgpu_gart_unbind(adev, gtt->offset, ttm->num_pages);
+	gtt->bound = false;
 }
 
-static void amdgpu_ttm_backend_destroy(struct ttm_tt *ttm)
+static void amdgpu_ttm_backend_destroy(struct ttm_device *bdev,
+				       struct ttm_tt *ttm)
 {
-	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(ttm);
+
+	if (gtt->usertask)
+		put_task_struct(gtt->usertask);
 
-	ttm_dma_tt_fini(&gtt->ttm);
+	ttm_tt_fini(&gtt->ttm);
 	kfree(gtt);
 }
 
-static struct ttm_backend_func amdgpu_backend_func = {
-	.bind = &amdgpu_ttm_backend_bind,
-	.unbind = &amdgpu_ttm_backend_unbind,
-	.destroy = &amdgpu_ttm_backend_destroy,
-};
-
-static struct ttm_tt *amdgpu_ttm_tt_create(struct ttm_bo_device *bdev,
-				    unsigned long size, uint32_t page_flags,
-				    struct page *dummy_read_page)
+/**
+ * amdgpu_ttm_tt_create - Create a ttm_tt object for a given BO
+ *
+ * @bo: The buffer object to create a GTT ttm_tt object around
+ * @page_flags: Page flags to be added to the ttm_tt object
+ *
+ * Called by ttm_tt_create().
+ */
+static struct ttm_tt *amdgpu_ttm_tt_create(struct ttm_buffer_object *bo,
+					   uint32_t page_flags)
 {
-	struct amdgpu_device *adev;
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev);
+	struct amdgpu_bo *abo = ttm_to_amdgpu_bo(bo);
 	struct amdgpu_ttm_tt *gtt;
-
-	adev = amdgpu_ttm_adev(bdev);
+	enum ttm_caching caching;
 
 	gtt = kzalloc(sizeof(struct amdgpu_ttm_tt), GFP_KERNEL);
-	if (gtt == NULL) {
+	if (!gtt)
 		return NULL;
-	}
-	gtt->ttm.ttm.func = &amdgpu_backend_func;
-	gtt->adev = adev;
-	if (ttm_dma_tt_init(&gtt->ttm, bdev, size, page_flags, dummy_read_page)) {
+
+	gtt->gobj = &bo->base;
+	if (adev->gmc.mem_partitions && abo->xcp_id >= 0)
+		gtt->pool_id = KFD_XCP_MEM_ID(adev, abo->xcp_id);
+	else
+		gtt->pool_id = abo->xcp_id;
+
+	if (abo->flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC)
+		caching = ttm_write_combined;
+	else
+		caching = ttm_cached;
+
+	/* allocate space for the uninitialized page entries */
+	if (ttm_sg_tt_init(&gtt->ttm, bo, page_flags, caching)) {
 		kfree(gtt);
 		return NULL;
 	}
-	INIT_LIST_HEAD(&gtt->list);
-	return &gtt->ttm.ttm;
+	return &gtt->ttm;
 }
 
-static int amdgpu_ttm_tt_populate(struct ttm_tt *ttm)
+/*
+ * amdgpu_ttm_tt_populate - Map GTT pages visible to the device
+ *
+ * Map the pages of a ttm_tt object to an address space visible
+ * to the underlying device.
+ */
+static int amdgpu_ttm_tt_populate(struct ttm_device *bdev,
+				  struct ttm_tt *ttm,
+				  struct ttm_operation_ctx *ctx)
 {
-	struct amdgpu_device *adev = amdgpu_ttm_adev(ttm->bdev);
-	struct amdgpu_ttm_tt *gtt = (void *)ttm;
-	unsigned i;
-	int r;
-	bool slave = !!(ttm->page_flags & TTM_PAGE_FLAG_SG);
-
-	if (ttm->state != tt_unpopulated)
-		return 0;
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bdev);
+	struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(ttm);
+	struct ttm_pool *pool;
+	pgoff_t i;
+	int ret;
 
-	if (gtt && gtt->userptr) {
+	/* user pages are bound by amdgpu_ttm_tt_pin_userptr() */
+	if (gtt->userptr) {
 		ttm->sg = kzalloc(sizeof(struct sg_table), GFP_KERNEL);
 		if (!ttm->sg)
 			return -ENOMEM;
-
-		ttm->page_flags |= TTM_PAGE_FLAG_SG;
-		ttm->state = tt_unbound;
 		return 0;
 	}
 
-	if (slave && ttm->sg) {
-		drm_prime_sg_to_page_addr_arrays(ttm->sg, ttm->pages,
-						 gtt->ttm.dma_address, ttm->num_pages);
-		ttm->state = tt_unbound;
-		r = 0;
-		goto trace_mappings;
-	}
-
-#ifdef CONFIG_SWIOTLB
-	if (swiotlb_nr_tbl()) {
-		r = ttm_dma_populate(&gtt->ttm, adev->dev);
-		goto trace_mappings;
-	}
-#endif
+	if (ttm->page_flags & TTM_TT_FLAG_EXTERNAL)
+		return 0;
 
-	r = ttm_pool_populate(ttm);
-	if (r) {
-		return r;
-	}
+	if (adev->mman.ttm_pools && gtt->pool_id >= 0)
+		pool = &adev->mman.ttm_pools[gtt->pool_id];
+	else
+		pool = &adev->mman.bdev.pool;
+	ret = ttm_pool_alloc(pool, ttm, ctx);
+	if (ret)
+		return ret;
 
-	for (i = 0; i < ttm->num_pages; i++) {
-		gtt->ttm.dma_address[i] = pci_map_page(adev->pdev, ttm->pages[i],
-						       0, PAGE_SIZE,
-						       PCI_DMA_BIDIRECTIONAL);
-		if (pci_dma_mapping_error(adev->pdev, gtt->ttm.dma_address[i])) {
-			while (i--) {
-				pci_unmap_page(adev->pdev, gtt->ttm.dma_address[i],
-					       PAGE_SIZE, PCI_DMA_BIDIRECTIONAL);
-				gtt->ttm.dma_address[i] = 0;
-			}
-			ttm_pool_unpopulate(ttm);
-			return -EFAULT;
-		}
-	}
+	for (i = 0; i < ttm->num_pages; ++i)
+		ttm->pages[i]->mapping = bdev->dev_mapping;
 
-	r = 0;
-trace_mappings:
-	if (likely(!r))
-		amdgpu_trace_dma_map(ttm);
-	return r;
+	return 0;
 }
 
-static void amdgpu_ttm_tt_unpopulate(struct ttm_tt *ttm)
+/*
+ * amdgpu_ttm_tt_unpopulate - unmap GTT pages and unpopulate page arrays
+ *
+ * Unmaps pages of a ttm_tt object from the device address space and
+ * unpopulates the page array backing it.
+ */
+static void amdgpu_ttm_tt_unpopulate(struct ttm_device *bdev,
+				     struct ttm_tt *ttm)
 {
+	struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(ttm);
 	struct amdgpu_device *adev;
-	struct amdgpu_ttm_tt *gtt = (void *)ttm;
-	unsigned i;
-	bool slave = !!(ttm->page_flags & TTM_PAGE_FLAG_SG);
+	struct ttm_pool *pool;
+	pgoff_t i;
+
+	amdgpu_ttm_backend_unbind(bdev, ttm);
 
-	if (gtt && gtt->userptr) {
+	if (gtt->userptr) {
+		amdgpu_ttm_tt_set_user_pages(ttm, NULL);
 		kfree(ttm->sg);
-		ttm->page_flags &= ~TTM_PAGE_FLAG_SG;
+		ttm->sg = NULL;
 		return;
 	}
 
-	if (slave)
+	if (ttm->page_flags & TTM_TT_FLAG_EXTERNAL)
 		return;
 
-	adev = amdgpu_ttm_adev(ttm->bdev);
-
-	amdgpu_trace_dma_unmap(ttm);
+	for (i = 0; i < ttm->num_pages; ++i)
+		ttm->pages[i]->mapping = NULL;
 
-#ifdef CONFIG_SWIOTLB
-	if (swiotlb_nr_tbl()) {
-		ttm_dma_unpopulate(&gtt->ttm, adev->dev);
-		return;
-	}
-#endif
+	adev = amdgpu_ttm_adev(bdev);
 
-	for (i = 0; i < ttm->num_pages; i++) {
-		if (gtt->ttm.dma_address[i]) {
-			pci_unmap_page(adev->pdev, gtt->ttm.dma_address[i],
-				       PAGE_SIZE, PCI_DMA_BIDIRECTIONAL);
-		}
-	}
+	if (adev->mman.ttm_pools && gtt->pool_id >= 0)
+		pool = &adev->mman.ttm_pools[gtt->pool_id];
+	else
+		pool = &adev->mman.bdev.pool;
 
-	ttm_pool_unpopulate(ttm);
+	return ttm_pool_free(pool, ttm);
 }
 
-int amdgpu_ttm_tt_set_userptr(struct ttm_tt *ttm, uint64_t addr,
-			      uint32_t flags)
+/**
+ * amdgpu_ttm_tt_get_userptr - Return the userptr GTT ttm_tt for the current
+ * task
+ *
+ * @tbo: The ttm_buffer_object that contains the userptr
+ * @user_addr:  The returned value
+ */
+int amdgpu_ttm_tt_get_userptr(const struct ttm_buffer_object *tbo,
+			      uint64_t *user_addr)
 {
-	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	struct amdgpu_ttm_tt *gtt;
 
-	if (gtt == NULL)
+	if (!tbo->ttm)
 		return -EINVAL;
 
+	gtt = (void *)tbo->ttm;
+	*user_addr = gtt->userptr;
+	return 0;
+}
+
+/**
+ * amdgpu_ttm_tt_set_userptr - Initialize userptr GTT ttm_tt for the current
+ * task
+ *
+ * @bo: The ttm_buffer_object to bind this userptr to
+ * @addr:  The address in the current tasks VM space to use
+ * @flags: Requirements of userptr object.
+ *
+ * Called by amdgpu_gem_userptr_ioctl() and kfd_ioctl_alloc_memory_of_gpu() to
+ * bind userptr pages to current task and by kfd_ioctl_acquire_vm() to
+ * initialize GPU VM for a KFD process.
+ */
+int amdgpu_ttm_tt_set_userptr(struct ttm_buffer_object *bo,
+			      uint64_t addr, uint32_t flags)
+{
+	struct amdgpu_ttm_tt *gtt;
+
+	if (!bo->ttm) {
+		/* TODO: We want a separate TTM object type for userptrs */
+		bo->ttm = amdgpu_ttm_tt_create(bo, 0);
+		if (bo->ttm == NULL)
+			return -ENOMEM;
+	}
+
+	/* Set TTM_TT_FLAG_EXTERNAL before populate but after create. */
+	bo->ttm->page_flags |= TTM_TT_FLAG_EXTERNAL;
+
+	gtt = ttm_to_amdgpu_ttm_tt(bo->ttm);
 	gtt->userptr = addr;
-	gtt->usermm = current->mm;
 	gtt->userflags = flags;
-	spin_lock_init(&gtt->guptasklock);
-	INIT_LIST_HEAD(&gtt->guptasks);
-	atomic_set(&gtt->mmu_invalidations, 0);
+
+	if (gtt->usertask)
+		put_task_struct(gtt->usertask);
+	gtt->usertask = current->group_leader;
+	get_task_struct(gtt->usertask);
 
 	return 0;
 }
 
+/*
+ * amdgpu_ttm_tt_get_usermm - Return memory manager for ttm_tt object
+ */
 struct mm_struct *amdgpu_ttm_tt_get_usermm(struct ttm_tt *ttm)
 {
-	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(ttm);
 
 	if (gtt == NULL)
 		return NULL;
 
-	return gtt->usermm;
+	if (gtt->usertask == NULL)
+		return NULL;
+
+	return gtt->usertask->mm;
 }
 
+/*
+ * amdgpu_ttm_tt_affect_userptr - Determine if a ttm_tt object lays inside an
+ * address range for the current task.
+ *
+ */
 bool amdgpu_ttm_tt_affect_userptr(struct ttm_tt *ttm, unsigned long start,
-				  unsigned long end)
+				  unsigned long end, unsigned long *userptr)
 {
-	struct amdgpu_ttm_tt *gtt = (void *)ttm;
-	struct amdgpu_ttm_gup_task_list *entry;
+	struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(ttm);
 	unsigned long size;
 
 	if (gtt == NULL || !gtt->userptr)
 		return false;
 
-	size = (unsigned long)gtt->ttm.ttm.num_pages * PAGE_SIZE;
+	/* Return false if no part of the ttm_tt object lies within
+	 * the range
+	 */
+	size = (unsigned long)gtt->ttm.num_pages * PAGE_SIZE;
 	if (gtt->userptr > end || gtt->userptr + size <= start)
 		return false;
 
-	spin_lock(&gtt->guptasklock);
-	list_for_each_entry(entry, &gtt->guptasks, list) {
-		if (entry->task == current) {
-			spin_unlock(&gtt->guptasklock);
-			return false;
-		}
-	}
-	spin_unlock(&gtt->guptasklock);
-
-	atomic_inc(&gtt->mmu_invalidations);
-
+	if (userptr)
+		*userptr = gtt->userptr;
 	return true;
 }
 
-bool amdgpu_ttm_tt_userptr_invalidated(struct ttm_tt *ttm,
-				       int *last_invalidated)
+/*
+ * amdgpu_ttm_tt_is_userptr - Have the pages backing by userptr?
+ */
+bool amdgpu_ttm_tt_is_userptr(struct ttm_tt *ttm)
 {
-	struct amdgpu_ttm_tt *gtt = (void *)ttm;
-	int prev_invalidated = *last_invalidated;
+	struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(ttm);
 
-	*last_invalidated = atomic_read(&gtt->mmu_invalidations);
-	return prev_invalidated != *last_invalidated;
+	if (gtt == NULL || !gtt->userptr)
+		return false;
+
+	return true;
 }
 
+/*
+ * amdgpu_ttm_tt_is_readonly - Is the ttm_tt object read only?
+ */
 bool amdgpu_ttm_tt_is_readonly(struct ttm_tt *ttm)
 {
-	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(ttm);
 
 	if (gtt == NULL)
 		return false;
@@ -1113,21 +1351,52 @@ bool amdgpu_ttm_tt_is_readonly(struct ttm_tt *ttm)
 	return !!(gtt->userflags & AMDGPU_GEM_USERPTR_READONLY);
 }
 
-uint64_t amdgpu_ttm_tt_pte_flags(struct amdgpu_device *adev, struct ttm_tt *ttm,
-				 struct ttm_mem_reg *mem)
+/**
+ * amdgpu_ttm_tt_pde_flags - Compute PDE flags for ttm_tt object
+ *
+ * @ttm: The ttm_tt object to compute the flags for
+ * @mem: The memory registry backing this ttm_tt object
+ *
+ * Figure out the flags to use for a VM PDE (Page Directory Entry).
+ */
+uint64_t amdgpu_ttm_tt_pde_flags(struct ttm_tt *ttm, struct ttm_resource *mem)
 {
 	uint64_t flags = 0;
 
 	if (mem && mem->mem_type != TTM_PL_SYSTEM)
 		flags |= AMDGPU_PTE_VALID;
 
-	if (mem && mem->mem_type == TTM_PL_TT) {
+	if (mem && (mem->mem_type == TTM_PL_TT ||
+		    mem->mem_type == AMDGPU_PL_DOORBELL ||
+		    mem->mem_type == AMDGPU_PL_PREEMPT ||
+		    mem->mem_type == AMDGPU_PL_MMIO_REMAP)) {
 		flags |= AMDGPU_PTE_SYSTEM;
 
-		if (ttm->caching_state == tt_cached)
+		if (ttm->caching == ttm_cached)
 			flags |= AMDGPU_PTE_SNOOPED;
 	}
 
+	if (mem && mem->mem_type == TTM_PL_VRAM &&
+			mem->bus.caching == ttm_cached)
+		flags |= AMDGPU_PTE_SNOOPED;
+
+	return flags;
+}
+
+/**
+ * amdgpu_ttm_tt_pte_flags - Compute PTE flags for ttm_tt object
+ *
+ * @adev: amdgpu_device pointer
+ * @ttm: The ttm_tt object to compute the flags for
+ * @mem: The memory registry backing this ttm_tt object
+ *
+ * Figure out the flags to use for a VM PTE (Page Table Entry).
+ */
+uint64_t amdgpu_ttm_tt_pte_flags(struct amdgpu_device *adev, struct ttm_tt *ttm,
+				 struct ttm_resource *mem)
+{
+	uint64_t flags = amdgpu_ttm_tt_pde_flags(ttm, mem);
+
 	flags |= adev->gart.gart_pte_flags;
 	flags |= AMDGPU_PTE_READABLE;
 
@@ -1137,575 +1406,1197 @@ uint64_t amdgpu_ttm_tt_pte_flags(struct amdgpu_device *adev, struct ttm_tt *ttm,
 	return flags;
 }
 
+/*
+ * amdgpu_ttm_bo_eviction_valuable - Check to see if we can evict a buffer
+ * object.
+ *
+ * Return true if eviction is sensible. Called by ttm_mem_evict_first() on
+ * behalf of ttm_bo_mem_force_space() which tries to evict buffer objects until
+ * it can find space for a new object and by ttm_bo_force_list_clean() which is
+ * used to clean out a memory space.
+ */
 static bool amdgpu_ttm_bo_eviction_valuable(struct ttm_buffer_object *bo,
 					    const struct ttm_place *place)
 {
-	unsigned long num_pages = bo->mem.num_pages;
-	struct drm_mm_node *node = bo->mem.mm_node;
+	struct dma_resv_iter resv_cursor;
+	struct dma_fence *f;
 
-	if (bo->mem.start != AMDGPU_BO_INVALID_OFFSET)
+	if (!amdgpu_bo_is_amdgpu_bo(bo))
 		return ttm_bo_eviction_valuable(bo, place);
 
-	switch (bo->mem.mem_type) {
-	case TTM_PL_TT:
+	/* Swapout? */
+	if (bo->resource->mem_type == TTM_PL_SYSTEM)
 		return true;
 
-	case TTM_PL_VRAM:
-		/* Check each drm MM node individually */
-		while (num_pages) {
-			if (place->fpfn < (node->start + node->size) &&
-			    !(place->lpfn && place->lpfn <= node->start))
-				return true;
-
-			num_pages -= node->size;
-			++node;
-		}
-		break;
+	if (bo->type == ttm_bo_type_kernel &&
+	    !amdgpu_vm_evictable(ttm_to_amdgpu_bo(bo)))
+		return false;
 
-	default:
-		break;
+	/* If bo is a KFD BO, check if the bo belongs to the current process.
+	 * If true, then return false as any KFD process needs all its BOs to
+	 * be resident to run successfully
+	 */
+	dma_resv_for_each_fence(&resv_cursor, bo->base.resv,
+				DMA_RESV_USAGE_BOOKKEEP, f) {
+		if (amdkfd_fence_check_mm(f, current->mm) &&
+		    !(place->flags & TTM_PL_FLAG_CONTIGUOUS))
+			return false;
 	}
 
+	/* Preemptible BOs don't own system resources managed by the
+	 * driver (pages, VRAM, GART space). They point to resources
+	 * owned by someone else (e.g. pageable memory in user mode
+	 * or a DMABuf). They are used in a preemptible context so we
+	 * can guarantee no deadlocks and good QoS in case of MMU
+	 * notifiers or DMABuf move notifiers from the resource owner.
+	 */
+	if (bo->resource->mem_type == AMDGPU_PL_PREEMPT)
+		return false;
+
+	if (bo->resource->mem_type == TTM_PL_TT &&
+	    amdgpu_bo_encrypted(ttm_to_amdgpu_bo(bo)))
+		return false;
+
 	return ttm_bo_eviction_valuable(bo, place);
 }
 
-static int amdgpu_ttm_access_memory(struct ttm_buffer_object *bo,
-				    unsigned long offset,
-				    void *buf, int len, int write)
+static void amdgpu_ttm_vram_mm_access(struct amdgpu_device *adev, loff_t pos,
+				      void *buf, size_t size, bool write)
 {
-	struct amdgpu_bo *abo = container_of(bo, struct amdgpu_bo, tbo);
-	struct amdgpu_device *adev = amdgpu_ttm_adev(abo->tbo.bdev);
-	struct drm_mm_node *nodes = abo->tbo.mem.mm_node;
-	uint32_t value = 0;
-	int ret = 0;
-	uint64_t pos;
-	unsigned long flags;
-
-	if (bo->mem.mem_type != TTM_PL_VRAM)
-		return -EIO;
-
-	while (offset >= (nodes->size << PAGE_SHIFT)) {
-		offset -= nodes->size << PAGE_SHIFT;
-		++nodes;
-	}
-	pos = (nodes->start << PAGE_SHIFT) + offset;
-
-	while (len && pos < adev->mc.mc_vram_size) {
-		uint64_t aligned_pos = pos & ~(uint64_t)3;
-		uint32_t bytes = 4 - (pos & 3);
-		uint32_t shift = (pos & 3) * 8;
+	while (size) {
+		uint64_t aligned_pos = ALIGN_DOWN(pos, 4);
+		uint64_t bytes = 4 - (pos & 0x3);
+		uint32_t shift = (pos & 0x3) * 8;
 		uint32_t mask = 0xffffffff << shift;
+		uint32_t value = 0;
 
-		if (len < bytes) {
-			mask &= 0xffffffff >> (bytes - len) * 8;
-			bytes = len;
+		if (size < bytes) {
+			mask &= 0xffffffff >> (bytes - size) * 8;
+			bytes = size;
 		}
 
-		spin_lock_irqsave(&adev->mmio_idx_lock, flags);
-		WREG32(mmMM_INDEX, ((uint32_t)aligned_pos) | 0x80000000);
-		WREG32(mmMM_INDEX_HI, aligned_pos >> 31);
-		if (!write || mask != 0xffffffff)
-			value = RREG32(mmMM_DATA);
-		if (write) {
-			value &= ~mask;
-			value |= (*(uint32_t *)buf << shift) & mask;
-			WREG32(mmMM_DATA, value);
-		}
-		spin_unlock_irqrestore(&adev->mmio_idx_lock, flags);
-		if (!write) {
-			value = (value & mask) >> shift;
-			memcpy(buf, &value, bytes);
+		if (mask != 0xffffffff) {
+			amdgpu_device_mm_access(adev, aligned_pos, &value, 4, false);
+			if (write) {
+				value &= ~mask;
+				value |= (*(uint32_t *)buf << shift) & mask;
+				amdgpu_device_mm_access(adev, aligned_pos, &value, 4, true);
+			} else {
+				value = (value & mask) >> shift;
+				memcpy(buf, &value, bytes);
+			}
+		} else {
+			amdgpu_device_mm_access(adev, aligned_pos, buf, 4, write);
 		}
 
-		ret += bytes;
-		buf = (uint8_t *)buf + bytes;
 		pos += bytes;
-		len -= bytes;
-		if (pos >= (nodes->start + nodes->size) << PAGE_SHIFT) {
-			++nodes;
-			pos = (nodes->start << PAGE_SHIFT);
+		buf += bytes;
+		size -= bytes;
+	}
+}
+
+static int amdgpu_ttm_access_memory_sdma(struct ttm_buffer_object *bo,
+					unsigned long offset, void *buf,
+					int len, int write)
+{
+	struct amdgpu_bo *abo = ttm_to_amdgpu_bo(bo);
+	struct amdgpu_device *adev = amdgpu_ttm_adev(abo->tbo.bdev);
+	struct amdgpu_res_cursor src_mm;
+	struct amdgpu_job *job;
+	struct dma_fence *fence;
+	uint64_t src_addr, dst_addr;
+	unsigned int num_dw;
+	int r, idx;
+
+	if (len != PAGE_SIZE)
+		return -EINVAL;
+
+	if (!adev->mman.sdma_access_ptr)
+		return -EACCES;
+
+	if (!drm_dev_enter(adev_to_drm(adev), &idx))
+		return -ENODEV;
+
+	if (write)
+		memcpy(adev->mman.sdma_access_ptr, buf, len);
+
+	num_dw = ALIGN(adev->mman.buffer_funcs->copy_num_dw, 8);
+	r = amdgpu_job_alloc_with_ib(adev, &adev->mman.high_pr,
+				     AMDGPU_FENCE_OWNER_UNDEFINED,
+				     num_dw * 4, AMDGPU_IB_POOL_DELAYED,
+				     &job,
+				     AMDGPU_KERNEL_JOB_ID_TTM_ACCESS_MEMORY_SDMA);
+	if (r)
+		goto out;
+
+	amdgpu_res_first(abo->tbo.resource, offset, len, &src_mm);
+	src_addr = amdgpu_ttm_domain_start(adev, bo->resource->mem_type) +
+		src_mm.start;
+	dst_addr = amdgpu_bo_gpu_offset(adev->mman.sdma_access_bo);
+	if (write)
+		swap(src_addr, dst_addr);
+
+	amdgpu_emit_copy_buffer(adev, &job->ibs[0], src_addr, dst_addr,
+				PAGE_SIZE, 0);
+
+	amdgpu_ring_pad_ib(adev->mman.buffer_funcs_ring, &job->ibs[0]);
+	WARN_ON(job->ibs[0].length_dw > num_dw);
+
+	fence = amdgpu_job_submit(job);
+
+	if (!dma_fence_wait_timeout(fence, false, adev->sdma_timeout))
+		r = -ETIMEDOUT;
+	dma_fence_put(fence);
+
+	if (!(r || write))
+		memcpy(buf, adev->mman.sdma_access_ptr, len);
+out:
+	drm_dev_exit(idx);
+	return r;
+}
+
+/**
+ * amdgpu_ttm_access_memory - Read or Write memory that backs a buffer object.
+ *
+ * @bo:  The buffer object to read/write
+ * @offset:  Offset into buffer object
+ * @buf:  Secondary buffer to write/read from
+ * @len: Length in bytes of access
+ * @write:  true if writing
+ *
+ * This is used to access VRAM that backs a buffer object via MMIO
+ * access for debugging purposes.
+ */
+static int amdgpu_ttm_access_memory(struct ttm_buffer_object *bo,
+				    unsigned long offset, void *buf, int len,
+				    int write)
+{
+	struct amdgpu_bo *abo = ttm_to_amdgpu_bo(bo);
+	struct amdgpu_device *adev = amdgpu_ttm_adev(abo->tbo.bdev);
+	struct amdgpu_res_cursor cursor;
+	int ret = 0;
+
+	if (bo->resource->mem_type != TTM_PL_VRAM)
+		return -EIO;
+
+	if (amdgpu_device_has_timeouts_enabled(adev) &&
+			!amdgpu_ttm_access_memory_sdma(bo, offset, buf, len, write))
+		return len;
+
+	amdgpu_res_first(bo->resource, offset, len, &cursor);
+	while (cursor.remaining) {
+		size_t count, size = cursor.size;
+		loff_t pos = cursor.start;
+
+		count = amdgpu_device_aper_access(adev, pos, buf, size, write);
+		size -= count;
+		if (size) {
+			/* using MM to access rest vram and handle un-aligned address */
+			pos += count;
+			buf += count;
+			amdgpu_ttm_vram_mm_access(adev, pos, buf, size, write);
 		}
+
+		ret += cursor.size;
+		buf += cursor.size;
+		amdgpu_res_next(&cursor, cursor.size);
 	}
 
 	return ret;
 }
 
-static struct ttm_bo_driver amdgpu_bo_driver = {
+static void
+amdgpu_bo_delete_mem_notify(struct ttm_buffer_object *bo)
+{
+	amdgpu_bo_move_notify(bo, false, NULL);
+}
+
+static struct ttm_device_funcs amdgpu_bo_driver = {
 	.ttm_tt_create = &amdgpu_ttm_tt_create,
 	.ttm_tt_populate = &amdgpu_ttm_tt_populate,
 	.ttm_tt_unpopulate = &amdgpu_ttm_tt_unpopulate,
-	.invalidate_caches = &amdgpu_invalidate_caches,
-	.init_mem_type = &amdgpu_init_mem_type,
+	.ttm_tt_destroy = &amdgpu_ttm_backend_destroy,
 	.eviction_valuable = amdgpu_ttm_bo_eviction_valuable,
 	.evict_flags = &amdgpu_evict_flags,
 	.move = &amdgpu_bo_move,
-	.verify_access = &amdgpu_verify_access,
-	.move_notify = &amdgpu_bo_move_notify,
-	.fault_reserve_notify = &amdgpu_bo_fault_reserve_notify,
+	.delete_mem_notify = &amdgpu_bo_delete_mem_notify,
+	.release_notify = &amdgpu_bo_release_notify,
 	.io_mem_reserve = &amdgpu_ttm_io_mem_reserve,
-	.io_mem_free = &amdgpu_ttm_io_mem_free,
 	.io_mem_pfn = amdgpu_ttm_io_mem_pfn,
-	.access_memory = &amdgpu_ttm_access_memory
+	.access_memory = &amdgpu_ttm_access_memory,
 };
 
+/*
+ * Firmware Reservation functions
+ */
+/**
+ * amdgpu_ttm_fw_reserve_vram_fini - free fw reserved vram
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * free fw reserved vram if it has been reserved.
+ */
+static void amdgpu_ttm_fw_reserve_vram_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->mman.fw_vram_usage_reserved_bo,
+		NULL, &adev->mman.fw_vram_usage_va);
+}
+
+/*
+ * Driver Reservation functions
+ */
+/**
+ * amdgpu_ttm_drv_reserve_vram_fini - free drv reserved vram
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * free drv reserved vram if it has been reserved.
+ */
+static void amdgpu_ttm_drv_reserve_vram_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->mman.drv_vram_usage_reserved_bo,
+						  NULL,
+						  &adev->mman.drv_vram_usage_va);
+}
+
+/**
+ * amdgpu_ttm_fw_reserve_vram_init - create bo vram reservation from fw
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * create bo vram reservation from fw.
+ */
+static int amdgpu_ttm_fw_reserve_vram_init(struct amdgpu_device *adev)
+{
+	uint64_t vram_size = adev->gmc.visible_vram_size;
+
+	adev->mman.fw_vram_usage_va = NULL;
+	adev->mman.fw_vram_usage_reserved_bo = NULL;
+
+	if (adev->mman.fw_vram_usage_size == 0 ||
+	    adev->mman.fw_vram_usage_size > vram_size)
+		return 0;
+
+	return amdgpu_bo_create_kernel_at(adev,
+					  adev->mman.fw_vram_usage_start_offset,
+					  adev->mman.fw_vram_usage_size,
+					  &adev->mman.fw_vram_usage_reserved_bo,
+					  &adev->mman.fw_vram_usage_va);
+}
+
+/**
+ * amdgpu_ttm_drv_reserve_vram_init - create bo vram reservation from driver
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * create bo vram reservation from drv.
+ */
+static int amdgpu_ttm_drv_reserve_vram_init(struct amdgpu_device *adev)
+{
+	u64 vram_size = adev->gmc.visible_vram_size;
+
+	adev->mman.drv_vram_usage_va = NULL;
+	adev->mman.drv_vram_usage_reserved_bo = NULL;
+
+	if (adev->mman.drv_vram_usage_size == 0 ||
+	    adev->mman.drv_vram_usage_size > vram_size)
+		return 0;
+
+	return amdgpu_bo_create_kernel_at(adev,
+					  adev->mman.drv_vram_usage_start_offset,
+					  adev->mman.drv_vram_usage_size,
+					  &adev->mman.drv_vram_usage_reserved_bo,
+					  &adev->mman.drv_vram_usage_va);
+}
+
+/*
+ * Memoy training reservation functions
+ */
+
+/**
+ * amdgpu_ttm_training_reserve_vram_fini - free memory training reserved vram
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * free memory training reserved vram if it has been reserved.
+ */
+static int amdgpu_ttm_training_reserve_vram_fini(struct amdgpu_device *adev)
+{
+	struct psp_memory_training_context *ctx = &adev->psp.mem_train_ctx;
+
+	ctx->init = PSP_MEM_TRAIN_NOT_SUPPORT;
+	amdgpu_bo_free_kernel(&ctx->c2p_bo, NULL, NULL);
+	ctx->c2p_bo = NULL;
+
+	return 0;
+}
+
+static void amdgpu_ttm_training_data_block_init(struct amdgpu_device *adev,
+						uint32_t reserve_size)
+{
+	struct psp_memory_training_context *ctx = &adev->psp.mem_train_ctx;
+
+	memset(ctx, 0, sizeof(*ctx));
+
+	ctx->c2p_train_data_offset =
+		ALIGN((adev->gmc.mc_vram_size - reserve_size - SZ_1M), SZ_1M);
+	ctx->p2c_train_data_offset =
+		(adev->gmc.mc_vram_size - GDDR6_MEM_TRAINING_OFFSET);
+	ctx->train_data_size =
+		GDDR6_MEM_TRAINING_DATA_SIZE_IN_BYTES;
+
+	DRM_DEBUG("train_data_size:%llx,p2c_train_data_offset:%llx,c2p_train_data_offset:%llx.\n",
+			ctx->train_data_size,
+			ctx->p2c_train_data_offset,
+			ctx->c2p_train_data_offset);
+}
+
+/*
+ * reserve TMR memory at the top of VRAM which holds
+ * IP Discovery data and is protected by PSP.
+ */
+static int amdgpu_ttm_reserve_tmr(struct amdgpu_device *adev)
+{
+	struct psp_memory_training_context *ctx = &adev->psp.mem_train_ctx;
+	bool mem_train_support = false;
+	uint32_t reserve_size = 0;
+	int ret;
+
+	if (adev->bios && !amdgpu_sriov_vf(adev)) {
+		if (amdgpu_atomfirmware_mem_training_supported(adev))
+			mem_train_support = true;
+		else
+			DRM_DEBUG("memory training does not support!\n");
+	}
+
+	/*
+	 * Query reserved tmr size through atom firmwareinfo for Sienna_Cichlid and onwards for all
+	 * the use cases (IP discovery/G6 memory training/profiling/diagnostic data.etc)
+	 *
+	 * Otherwise, fallback to legacy approach to check and reserve tmr block for ip
+	 * discovery data and G6 memory training data respectively
+	 */
+	if (adev->bios)
+		reserve_size =
+			amdgpu_atomfirmware_get_fw_reserved_fb_size(adev);
+
+	if (!adev->bios &&
+	    (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
+	     amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) ||
+	     amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0)))
+		reserve_size = max(reserve_size, (uint32_t)280 << 20);
+	else if (!reserve_size)
+		reserve_size = DISCOVERY_TMR_OFFSET;
+
+	if (mem_train_support) {
+		/* reserve vram for mem train according to TMR location */
+		amdgpu_ttm_training_data_block_init(adev, reserve_size);
+		ret = amdgpu_bo_create_kernel_at(adev,
+						 ctx->c2p_train_data_offset,
+						 ctx->train_data_size,
+						 &ctx->c2p_bo,
+						 NULL);
+		if (ret) {
+			dev_err(adev->dev, "alloc c2p_bo failed(%d)!\n", ret);
+			amdgpu_ttm_training_reserve_vram_fini(adev);
+			return ret;
+		}
+		ctx->init = PSP_MEM_TRAIN_RESERVE_SUCCESS;
+	}
+
+	if (!adev->gmc.is_app_apu) {
+		ret = amdgpu_bo_create_kernel_at(
+			adev, adev->gmc.real_vram_size - reserve_size,
+			reserve_size, &adev->mman.fw_reserved_memory, NULL);
+		if (ret) {
+			dev_err(adev->dev, "alloc tmr failed(%d)!\n", ret);
+			amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory,
+					      NULL, NULL);
+			return ret;
+		}
+	} else {
+		DRM_DEBUG_DRIVER("backdoor fw loading path for PSP TMR, no reservation needed\n");
+	}
+
+	return 0;
+}
+
+static int amdgpu_ttm_pools_init(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (!adev->gmc.is_app_apu || !adev->gmc.num_mem_partitions)
+		return 0;
+
+	adev->mman.ttm_pools = kcalloc(adev->gmc.num_mem_partitions,
+				       sizeof(*adev->mman.ttm_pools),
+				       GFP_KERNEL);
+	if (!adev->mman.ttm_pools)
+		return -ENOMEM;
+
+	for (i = 0; i < adev->gmc.num_mem_partitions; i++) {
+		ttm_pool_init(&adev->mman.ttm_pools[i], adev->dev,
+			      adev->gmc.mem_partitions[i].numa.node,
+			      false, false);
+	}
+	return 0;
+}
+
+static void amdgpu_ttm_pools_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (!adev->gmc.is_app_apu || !adev->mman.ttm_pools)
+		return;
+
+	for (i = 0; i < adev->gmc.num_mem_partitions; i++)
+		ttm_pool_fini(&adev->mman.ttm_pools[i]);
+
+	kfree(adev->mman.ttm_pools);
+	adev->mman.ttm_pools = NULL;
+}
+
+/**
+ * amdgpu_ttm_mmio_remap_bo_init - Allocate the singleton 4K MMIO_REMAP BO
+ * @adev: amdgpu device
+ *
+ * Allocates a one-page (4K) GEM BO in AMDGPU_GEM_DOMAIN_MMIO_REMAP when the
+ * hardware exposes a remap base (adev->rmmio_remap.bus_addr) and the host
+ * PAGE_SIZE is <= AMDGPU_GPU_PAGE_SIZE (4K). The BO is created as a regular
+ * GEM object (amdgpu_bo_create).
+ *
+ * Return:
+ *  * 0 on success or intentional skip (feature not present/unsupported)
+ *  * negative errno on allocation failure
+ */
+static int amdgpu_ttm_mmio_remap_bo_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_bo_param bp;
+	int r;
+
+	/* Skip if HW doesn't expose remap, or if PAGE_SIZE > AMDGPU_GPU_PAGE_SIZE (4K). */
+	if (!adev->rmmio_remap.bus_addr || PAGE_SIZE > AMDGPU_GPU_PAGE_SIZE)
+		return 0;
+
+	memset(&bp, 0, sizeof(bp));
+
+	/* Create exactly one GEM BO in the MMIO_REMAP domain. */
+	bp.type        = ttm_bo_type_device;          /* userspace-mappable GEM */
+	bp.size        = AMDGPU_GPU_PAGE_SIZE;        /* 4K */
+	bp.byte_align  = AMDGPU_GPU_PAGE_SIZE;
+	bp.domain      = AMDGPU_GEM_DOMAIN_MMIO_REMAP;
+	bp.flags       = 0;
+	bp.resv        = NULL;
+	bp.bo_ptr_size = sizeof(struct amdgpu_bo);
+
+	r = amdgpu_bo_create(adev, &bp, &adev->rmmio_remap.bo);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+/**
+ * amdgpu_ttm_mmio_remap_bo_fini - Free the singleton MMIO_REMAP BO
+ * @adev: amdgpu device
+ *
+ * Frees the kernel-owned MMIO_REMAP BO if it was allocated by
+ * amdgpu_ttm_mmio_remap_bo_init().
+ */
+static void amdgpu_ttm_mmio_remap_bo_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_unref(&adev->rmmio_remap.bo);
+	adev->rmmio_remap.bo = NULL;
+}
+
+/*
+ * amdgpu_ttm_init - Init the memory management (ttm) as well as various
+ * gtt/vram related fields.
+ *
+ * This initializes all of the memory space pools that the TTM layer
+ * will need such as the GTT space (system memory mapped to the device),
+ * VRAM (on-board memory), and on-chip memories (GDS, GWS, OA) which
+ * can be mapped per VMID.
+ */
 int amdgpu_ttm_init(struct amdgpu_device *adev)
 {
 	uint64_t gtt_size;
 	int r;
-	u64 vis_vram_limit;
 
-	r = amdgpu_ttm_global_init(adev);
+	mutex_init(&adev->mman.gtt_window_lock);
+
+	dma_set_max_seg_size(adev->dev, UINT_MAX);
+	/* No others user of address space so set it to 0 */
+	r = ttm_device_init(&adev->mman.bdev, &amdgpu_bo_driver, adev->dev,
+			       adev_to_drm(adev)->anon_inode->i_mapping,
+			       adev_to_drm(adev)->vma_offset_manager,
+			       adev->need_swiotlb,
+			       dma_addressing_limited(adev->dev));
 	if (r) {
+		dev_err(adev->dev,
+			"failed initializing buffer object driver(%d).\n", r);
 		return r;
 	}
-	/* No others user of address space so set it to 0 */
-	r = ttm_bo_device_init(&adev->mman.bdev,
-			       adev->mman.bo_global_ref.ref.object,
-			       &amdgpu_bo_driver,
-			       adev->ddev->anon_inode->i_mapping,
-			       DRM_FILE_PAGE_OFFSET,
-			       adev->need_dma32);
+
+	r = amdgpu_ttm_pools_init(adev);
 	if (r) {
-		DRM_ERROR("failed initializing buffer object driver(%d).\n", r);
+		dev_err(adev->dev, "failed to init ttm pools(%d).\n", r);
 		return r;
 	}
 	adev->mman.initialized = true;
-	r = ttm_bo_init_mm(&adev->mman.bdev, TTM_PL_VRAM,
-				adev->mc.real_vram_size >> PAGE_SHIFT);
+
+	if (!adev->gmc.is_app_apu) {
+		/* Initialize VRAM pool with all of VRAM divided into pages */
+		r = amdgpu_vram_mgr_init(adev);
+		if (r) {
+			dev_err(adev->dev, "Failed initializing VRAM heap.\n");
+			return r;
+		}
+	}
+
+	/* Change the size here instead of the init above so only lpfn is affected */
+	amdgpu_ttm_set_buffer_funcs_status(adev, false);
+#ifdef CONFIG_64BIT
+#ifdef CONFIG_X86
+	if (adev->gmc.xgmi.connected_to_cpu)
+		adev->mman.aper_base_kaddr = ioremap_cache(adev->gmc.aper_base,
+				adev->gmc.visible_vram_size);
+
+	else if (adev->gmc.is_app_apu)
+		DRM_DEBUG_DRIVER(
+			"No need to ioremap when real vram size is 0\n");
+	else
+#endif
+		adev->mman.aper_base_kaddr = ioremap_wc(adev->gmc.aper_base,
+				adev->gmc.visible_vram_size);
+#endif
+
+	/*
+	 *The reserved vram for firmware must be pinned to the specified
+	 *place on the VRAM, so reserve it early.
+	 */
+	r = amdgpu_ttm_fw_reserve_vram_init(adev);
+	if (r)
+		return r;
+
+	/*
+	 *The reserved vram for driver must be pinned to the specified
+	 *place on the VRAM, so reserve it early.
+	 */
+	r = amdgpu_ttm_drv_reserve_vram_init(adev);
+	if (r)
+		return r;
+
+	/*
+	 * only NAVI10 and onwards ASIC support for IP discovery.
+	 * If IP discovery enabled, a block of memory should be
+	 * reserved for IP discovey.
+	 */
+	if (adev->mman.discovery_bin) {
+		r = amdgpu_ttm_reserve_tmr(adev);
+		if (r)
+			return r;
+	}
+
+	/* allocate memory as required for VGA
+	 * This is used for VGA emulation and pre-OS scanout buffers to
+	 * avoid display artifacts while transitioning between pre-OS
+	 * and driver.
+	 */
+	if (!adev->gmc.is_app_apu) {
+		r = amdgpu_bo_create_kernel_at(adev, 0,
+					       adev->mman.stolen_vga_size,
+					       &adev->mman.stolen_vga_memory,
+					       NULL);
+		if (r)
+			return r;
+
+		r = amdgpu_bo_create_kernel_at(adev, adev->mman.stolen_vga_size,
+					       adev->mman.stolen_extended_size,
+					       &adev->mman.stolen_extended_memory,
+					       NULL);
+
+		if (r)
+			return r;
+
+		r = amdgpu_bo_create_kernel_at(adev,
+					       adev->mman.stolen_reserved_offset,
+					       adev->mman.stolen_reserved_size,
+					       &adev->mman.stolen_reserved_memory,
+					       NULL);
+		if (r)
+			return r;
+	} else {
+		DRM_DEBUG_DRIVER("Skipped stolen memory reservation\n");
+	}
+
+	dev_info(adev->dev, "amdgpu: %uM of VRAM memory ready\n",
+		 (unsigned int)(adev->gmc.real_vram_size / (1024 * 1024)));
+
+	/* Compute GTT size, either based on TTM limit
+	 * or whatever the user passed on module init.
+	 */
+	gtt_size = ttm_tt_pages_limit() << PAGE_SHIFT;
+	if (amdgpu_gtt_size != -1) {
+		uint64_t configured_size = (uint64_t)amdgpu_gtt_size << 20;
+
+		drm_warn(&adev->ddev,
+			"Configuring gttsize via module parameter is deprecated, please use ttm.pages_limit\n");
+		if (gtt_size != configured_size)
+			drm_warn(&adev->ddev,
+				"GTT size has been set as %llu but TTM size has been set as %llu, this is unusual\n",
+				configured_size, gtt_size);
+
+		gtt_size = configured_size;
+	}
+
+	/* Initialize GTT memory pool */
+	r = amdgpu_gtt_mgr_init(adev, gtt_size);
 	if (r) {
-		DRM_ERROR("Failed initializing VRAM heap.\n");
+		dev_err(adev->dev, "Failed initializing GTT heap.\n");
 		return r;
 	}
+	dev_info(adev->dev, "amdgpu: %uM of GTT memory ready.\n",
+		 (unsigned int)(gtt_size / (1024 * 1024)));
 
-	/* Reduce size of CPU-visible VRAM if requested */
-	vis_vram_limit = (u64)amdgpu_vis_vram_limit * 1024 * 1024;
-	if (amdgpu_vis_vram_limit > 0 &&
-	    vis_vram_limit <= adev->mc.visible_vram_size)
-		adev->mc.visible_vram_size = vis_vram_limit;
+	if (adev->flags & AMD_IS_APU) {
+		if (adev->gmc.real_vram_size < gtt_size)
+			adev->apu_prefer_gtt = true;
+	}
 
-	/* Change the size here instead of the init above so only lpfn is affected */
-	amdgpu_ttm_set_active_vram_size(adev, adev->mc.visible_vram_size);
+	/* Initialize doorbell pool on PCI BAR */
+	r = amdgpu_ttm_init_on_chip(adev, AMDGPU_PL_DOORBELL, adev->doorbell.size / PAGE_SIZE);
+	if (r) {
+		dev_err(adev->dev, "Failed initializing doorbell heap.\n");
+		return r;
+	}
 
-	r = amdgpu_bo_create_kernel(adev, adev->mc.stolen_size, PAGE_SIZE,
-				    AMDGPU_GEM_DOMAIN_VRAM,
-				    &adev->stolen_vga_memory,
-				    NULL, NULL);
-	if (r)
+	/* Create a boorbell page for kernel usages */
+	r = amdgpu_doorbell_create_kernel_doorbells(adev);
+	if (r) {
+		dev_err(adev->dev, "Failed to initialize kernel doorbells.\n");
 		return r;
-	DRM_INFO("amdgpu: %uM of VRAM memory ready\n",
-		 (unsigned) (adev->mc.real_vram_size / (1024 * 1024)));
+	}
 
-	if (amdgpu_gtt_size == -1)
-		gtt_size = max((AMDGPU_DEFAULT_GTT_SIZE_MB << 20),
-			       adev->mc.mc_vram_size);
-	else
-		gtt_size = (uint64_t)amdgpu_gtt_size << 20;
-	r = ttm_bo_init_mm(&adev->mman.bdev, TTM_PL_TT, gtt_size >> PAGE_SHIFT);
+	/* Initialize MMIO-remap pool (single page 4K) */
+	r = amdgpu_ttm_init_on_chip(adev, AMDGPU_PL_MMIO_REMAP, 1);
 	if (r) {
-		DRM_ERROR("Failed initializing GTT heap.\n");
+		dev_err(adev->dev, "Failed initializing MMIO-remap heap.\n");
 		return r;
 	}
-	DRM_INFO("amdgpu: %uM of GTT memory ready.\n",
-		 (unsigned)(gtt_size / (1024 * 1024)));
-
-	adev->gds.mem.total_size = adev->gds.mem.total_size << AMDGPU_GDS_SHIFT;
-	adev->gds.mem.gfx_partition_size = adev->gds.mem.gfx_partition_size << AMDGPU_GDS_SHIFT;
-	adev->gds.mem.cs_partition_size = adev->gds.mem.cs_partition_size << AMDGPU_GDS_SHIFT;
-	adev->gds.gws.total_size = adev->gds.gws.total_size << AMDGPU_GWS_SHIFT;
-	adev->gds.gws.gfx_partition_size = adev->gds.gws.gfx_partition_size << AMDGPU_GWS_SHIFT;
-	adev->gds.gws.cs_partition_size = adev->gds.gws.cs_partition_size << AMDGPU_GWS_SHIFT;
-	adev->gds.oa.total_size = adev->gds.oa.total_size << AMDGPU_OA_SHIFT;
-	adev->gds.oa.gfx_partition_size = adev->gds.oa.gfx_partition_size << AMDGPU_OA_SHIFT;
-	adev->gds.oa.cs_partition_size = adev->gds.oa.cs_partition_size << AMDGPU_OA_SHIFT;
-	/* GDS Memory */
-	if (adev->gds.mem.total_size) {
-		r = ttm_bo_init_mm(&adev->mman.bdev, AMDGPU_PL_GDS,
-				   adev->gds.mem.total_size >> PAGE_SHIFT);
-		if (r) {
-			DRM_ERROR("Failed initializing GDS heap.\n");
-			return r;
-		}
+
+	/* Allocate the singleton MMIO_REMAP BO (4K) if supported */
+	r = amdgpu_ttm_mmio_remap_bo_init(adev);
+	if (r)
+		return r;
+
+	/* Initialize preemptible memory pool */
+	r = amdgpu_preempt_mgr_init(adev);
+	if (r) {
+		dev_err(adev->dev, "Failed initializing PREEMPT heap.\n");
+		return r;
 	}
 
-	/* GWS */
-	if (adev->gds.gws.total_size) {
-		r = ttm_bo_init_mm(&adev->mman.bdev, AMDGPU_PL_GWS,
-				   adev->gds.gws.total_size >> PAGE_SHIFT);
-		if (r) {
-			DRM_ERROR("Failed initializing gws heap.\n");
-			return r;
-		}
+	/* Initialize various on-chip memory pools */
+	r = amdgpu_ttm_init_on_chip(adev, AMDGPU_PL_GDS, adev->gds.gds_size);
+	if (r) {
+		dev_err(adev->dev, "Failed initializing GDS heap.\n");
+		return r;
 	}
 
-	/* OA */
-	if (adev->gds.oa.total_size) {
-		r = ttm_bo_init_mm(&adev->mman.bdev, AMDGPU_PL_OA,
-				   adev->gds.oa.total_size >> PAGE_SHIFT);
-		if (r) {
-			DRM_ERROR("Failed initializing oa heap.\n");
-			return r;
-		}
+	r = amdgpu_ttm_init_on_chip(adev, AMDGPU_PL_GWS, adev->gds.gws_size);
+	if (r) {
+		dev_err(adev->dev, "Failed initializing gws heap.\n");
+		return r;
 	}
 
-	r = amdgpu_ttm_debugfs_init(adev);
+	r = amdgpu_ttm_init_on_chip(adev, AMDGPU_PL_OA, adev->gds.oa_size);
 	if (r) {
-		DRM_ERROR("Failed to init debugfs\n");
+		dev_err(adev->dev, "Failed initializing oa heap.\n");
 		return r;
 	}
+	if (amdgpu_bo_create_kernel(adev, PAGE_SIZE, PAGE_SIZE,
+				AMDGPU_GEM_DOMAIN_GTT,
+				&adev->mman.sdma_access_bo, NULL,
+				&adev->mman.sdma_access_ptr))
+		DRM_WARN("Debug VRAM access will use slowpath MM access\n");
+
 	return 0;
 }
 
+/*
+ * amdgpu_ttm_fini - De-initialize the TTM memory pools
+ */
 void amdgpu_ttm_fini(struct amdgpu_device *adev)
 {
-	int r;
+	int idx;
 
 	if (!adev->mman.initialized)
 		return;
-	amdgpu_ttm_debugfs_fini(adev);
-	if (adev->stolen_vga_memory) {
-		r = amdgpu_bo_reserve(adev->stolen_vga_memory, true);
-		if (r == 0) {
-			amdgpu_bo_unpin(adev->stolen_vga_memory);
-			amdgpu_bo_unreserve(adev->stolen_vga_memory);
-		}
-		amdgpu_bo_unref(&adev->stolen_vga_memory);
-	}
-	ttm_bo_clean_mm(&adev->mman.bdev, TTM_PL_VRAM);
-	ttm_bo_clean_mm(&adev->mman.bdev, TTM_PL_TT);
-	if (adev->gds.mem.total_size)
-		ttm_bo_clean_mm(&adev->mman.bdev, AMDGPU_PL_GDS);
-	if (adev->gds.gws.total_size)
-		ttm_bo_clean_mm(&adev->mman.bdev, AMDGPU_PL_GWS);
-	if (adev->gds.oa.total_size)
-		ttm_bo_clean_mm(&adev->mman.bdev, AMDGPU_PL_OA);
-	ttm_bo_device_release(&adev->mman.bdev);
-	amdgpu_gart_fini(adev);
-	amdgpu_ttm_global_fini(adev);
+
+	amdgpu_ttm_pools_fini(adev);
+
+	amdgpu_ttm_training_reserve_vram_fini(adev);
+	/* return the stolen vga memory back to VRAM */
+	if (!adev->gmc.is_app_apu) {
+		amdgpu_bo_free_kernel(&adev->mman.stolen_vga_memory, NULL, NULL);
+		amdgpu_bo_free_kernel(&adev->mman.stolen_extended_memory, NULL, NULL);
+		/* return the FW reserved memory back to VRAM */
+		amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory, NULL,
+				      NULL);
+		amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory_extend, NULL,
+				      NULL);
+		if (adev->mman.stolen_reserved_size)
+			amdgpu_bo_free_kernel(&adev->mman.stolen_reserved_memory,
+					      NULL, NULL);
+	}
+	amdgpu_bo_free_kernel(&adev->mman.sdma_access_bo, NULL,
+					&adev->mman.sdma_access_ptr);
+
+	amdgpu_ttm_mmio_remap_bo_fini(adev);
+	amdgpu_ttm_fw_reserve_vram_fini(adev);
+	amdgpu_ttm_drv_reserve_vram_fini(adev);
+
+	if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+
+		if (adev->mman.aper_base_kaddr)
+			iounmap(adev->mman.aper_base_kaddr);
+		adev->mman.aper_base_kaddr = NULL;
+
+		drm_dev_exit(idx);
+	}
+
+	if (!adev->gmc.is_app_apu)
+		amdgpu_vram_mgr_fini(adev);
+	amdgpu_gtt_mgr_fini(adev);
+	amdgpu_preempt_mgr_fini(adev);
+	amdgpu_doorbell_fini(adev);
+
+	ttm_range_man_fini(&adev->mman.bdev, AMDGPU_PL_GDS);
+	ttm_range_man_fini(&adev->mman.bdev, AMDGPU_PL_GWS);
+	ttm_range_man_fini(&adev->mman.bdev, AMDGPU_PL_OA);
+	ttm_range_man_fini(&adev->mman.bdev, AMDGPU_PL_DOORBELL);
+	ttm_range_man_fini(&adev->mman.bdev, AMDGPU_PL_MMIO_REMAP);
+	ttm_device_fini(&adev->mman.bdev);
 	adev->mman.initialized = false;
-	DRM_INFO("amdgpu: ttm finalized\n");
+	dev_info(adev->dev, "amdgpu: ttm finalized\n");
 }
 
-/* this should only be called at bootup or when userspace
- * isn't running */
-void amdgpu_ttm_set_active_vram_size(struct amdgpu_device *adev, u64 size)
+/**
+ * amdgpu_ttm_set_buffer_funcs_status - enable/disable use of buffer functions
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: true when we can use buffer functions.
+ *
+ * Enable/disable use of buffer functions during suspend/resume. This should
+ * only be called at bootup or when userspace isn't running.
+ */
+void amdgpu_ttm_set_buffer_funcs_status(struct amdgpu_device *adev, bool enable)
 {
-	struct ttm_mem_type_manager *man;
+	struct ttm_resource_manager *man = ttm_manager_type(&adev->mman.bdev, TTM_PL_VRAM);
+	uint64_t size;
+	int r;
 
-	if (!adev->mman.initialized)
+	if (!adev->mman.initialized || amdgpu_in_reset(adev) ||
+	    adev->mman.buffer_funcs_enabled == enable || adev->gmc.is_app_apu)
 		return;
 
-	man = &adev->mman.bdev.man[TTM_PL_VRAM];
+	if (enable) {
+		struct amdgpu_ring *ring;
+		struct drm_gpu_scheduler *sched;
+
+		ring = adev->mman.buffer_funcs_ring;
+		sched = &ring->sched;
+		r = drm_sched_entity_init(&adev->mman.high_pr,
+					  DRM_SCHED_PRIORITY_KERNEL, &sched,
+					  1, NULL);
+		if (r) {
+			dev_err(adev->dev,
+				"Failed setting up TTM BO move entity (%d)\n",
+				r);
+			return;
+		}
+
+		r = drm_sched_entity_init(&adev->mman.low_pr,
+					  DRM_SCHED_PRIORITY_NORMAL, &sched,
+					  1, NULL);
+		if (r) {
+			dev_err(adev->dev,
+				"Failed setting up TTM BO move entity (%d)\n",
+				r);
+			goto error_free_entity;
+		}
+	} else {
+		drm_sched_entity_destroy(&adev->mman.high_pr);
+		drm_sched_entity_destroy(&adev->mman.low_pr);
+		dma_fence_put(man->move);
+		man->move = NULL;
+	}
+
 	/* this just adjusts TTM size idea, which sets lpfn to the correct value */
-	man->size = size >> PAGE_SHIFT;
+	if (enable)
+		size = adev->gmc.real_vram_size;
+	else
+		size = adev->gmc.visible_vram_size;
+	man->size = size;
+	adev->mman.buffer_funcs_enabled = enable;
+
+	return;
+
+error_free_entity:
+	drm_sched_entity_destroy(&adev->mman.high_pr);
 }
 
-int amdgpu_mmap(struct file *filp, struct vm_area_struct *vma)
+static int amdgpu_ttm_prepare_job(struct amdgpu_device *adev,
+				  bool direct_submit,
+				  unsigned int num_dw,
+				  struct dma_resv *resv,
+				  bool vm_needs_flush,
+				  struct amdgpu_job **job,
+				  bool delayed, u64 k_job_id)
 {
-	struct drm_file *file_priv;
-	struct amdgpu_device *adev;
-
-	if (unlikely(vma->vm_pgoff < DRM_FILE_PAGE_OFFSET))
-		return -EINVAL;
+	enum amdgpu_ib_pool_type pool = direct_submit ?
+		AMDGPU_IB_POOL_DIRECT :
+		AMDGPU_IB_POOL_DELAYED;
+	int r;
+	struct drm_sched_entity *entity = delayed ? &adev->mman.low_pr :
+						    &adev->mman.high_pr;
+	r = amdgpu_job_alloc_with_ib(adev, entity,
+				     AMDGPU_FENCE_OWNER_UNDEFINED,
+				     num_dw * 4, pool, job, k_job_id);
+	if (r)
+		return r;
 
-	file_priv = filp->private_data;
-	adev = file_priv->minor->dev->dev_private;
-	if (adev == NULL)
-		return -EINVAL;
+	if (vm_needs_flush) {
+		(*job)->vm_pd_addr = amdgpu_gmc_pd_addr(adev->gmc.pdb0_bo ?
+							adev->gmc.pdb0_bo :
+							adev->gart.bo);
+		(*job)->vm_needs_flush = true;
+	}
+	if (!resv)
+		return 0;
 
-	return ttm_bo_mmap(filp, vma, &adev->mman.bdev);
+	return drm_sched_job_add_resv_dependencies(&(*job)->base, resv,
+						   DMA_RESV_USAGE_BOOKKEEP);
 }
 
-static int amdgpu_map_buffer(struct ttm_buffer_object *bo,
-			     struct ttm_mem_reg *mem, unsigned num_pages,
-			     uint64_t offset, unsigned window,
-			     struct amdgpu_ring *ring,
-			     uint64_t *addr)
+int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
+		       uint64_t dst_offset, uint32_t byte_count,
+		       struct dma_resv *resv,
+		       struct dma_fence **fence, bool direct_submit,
+		       bool vm_needs_flush, uint32_t copy_flags)
 {
-	struct amdgpu_ttm_tt *gtt = (void *)bo->ttm;
 	struct amdgpu_device *adev = ring->adev;
-	struct ttm_tt *ttm = bo->ttm;
+	unsigned int num_loops, num_dw;
 	struct amdgpu_job *job;
-	unsigned num_dw, num_bytes;
-	dma_addr_t *dma_address;
-	struct dma_fence *fence;
-	uint64_t src_addr, dst_addr;
-	uint64_t flags;
+	uint32_t max_bytes;
+	unsigned int i;
 	int r;
 
-	BUG_ON(adev->mman.buffer_funcs->copy_max_bytes <
-	       AMDGPU_GTT_MAX_TRANSFER_SIZE * 8);
-
-	*addr = adev->mc.gart_start;
-	*addr += (u64)window * AMDGPU_GTT_MAX_TRANSFER_SIZE *
-		AMDGPU_GPU_PAGE_SIZE;
-
-	num_dw = adev->mman.buffer_funcs->copy_num_dw;
-	while (num_dw & 0x7)
-		num_dw++;
-
-	num_bytes = num_pages * 8;
+	if (!direct_submit && !ring->sched.ready) {
+		dev_err(adev->dev,
+			"Trying to move memory with ring turned off.\n");
+		return -EINVAL;
+	}
 
-	r = amdgpu_job_alloc_with_ib(adev, num_dw * 4 + num_bytes, &job);
+	max_bytes = adev->mman.buffer_funcs->copy_max_bytes;
+	num_loops = DIV_ROUND_UP(byte_count, max_bytes);
+	num_dw = ALIGN(num_loops * adev->mman.buffer_funcs->copy_num_dw, 8);
+	r = amdgpu_ttm_prepare_job(adev, direct_submit, num_dw,
+				   resv, vm_needs_flush, &job, false,
+				   AMDGPU_KERNEL_JOB_ID_TTM_COPY_BUFFER);
 	if (r)
 		return r;
 
-	src_addr = num_dw * 4;
-	src_addr += job->ibs[0].gpu_addr;
+	for (i = 0; i < num_loops; i++) {
+		uint32_t cur_size_in_bytes = min(byte_count, max_bytes);
 
-	dst_addr = adev->gart.table_addr;
-	dst_addr += window * AMDGPU_GTT_MAX_TRANSFER_SIZE * 8;
-	amdgpu_emit_copy_buffer(adev, &job->ibs[0], src_addr,
-				dst_addr, num_bytes);
+		amdgpu_emit_copy_buffer(adev, &job->ibs[0], src_offset,
+					dst_offset, cur_size_in_bytes, copy_flags);
+		src_offset += cur_size_in_bytes;
+		dst_offset += cur_size_in_bytes;
+		byte_count -= cur_size_in_bytes;
+	}
 
 	amdgpu_ring_pad_ib(ring, &job->ibs[0]);
 	WARN_ON(job->ibs[0].length_dw > num_dw);
-
-	dma_address = &gtt->ttm.dma_address[offset >> PAGE_SHIFT];
-	flags = amdgpu_ttm_tt_pte_flags(adev, ttm, mem);
-	r = amdgpu_gart_map(adev, 0, num_pages, dma_address, flags,
-			    &job->ibs[0].ptr[num_dw]);
-	if (r)
-		goto error_free;
-
-	r = amdgpu_job_submit(job, ring, &adev->mman.entity,
-			      AMDGPU_FENCE_OWNER_UNDEFINED, &fence);
+	if (direct_submit)
+		r = amdgpu_job_submit_direct(job, ring, fence);
+	else
+		*fence = amdgpu_job_submit(job);
 	if (r)
 		goto error_free;
 
-	dma_fence_put(fence);
-
 	return r;
 
 error_free:
 	amdgpu_job_free(job);
+	dev_err(adev->dev, "Error scheduling IBs (%d)\n", r);
 	return r;
 }
 
-int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
-		       uint64_t dst_offset, uint32_t byte_count,
-		       struct reservation_object *resv,
-		       struct dma_fence **fence, bool direct_submit,
-		       bool vm_needs_flush)
+static int amdgpu_ttm_fill_mem(struct amdgpu_ring *ring, uint32_t src_data,
+			       uint64_t dst_addr, uint32_t byte_count,
+			       struct dma_resv *resv,
+			       struct dma_fence **fence,
+			       bool vm_needs_flush, bool delayed,
+			       u64 k_job_id)
 {
 	struct amdgpu_device *adev = ring->adev;
+	unsigned int num_loops, num_dw;
 	struct amdgpu_job *job;
-
 	uint32_t max_bytes;
-	unsigned num_loops, num_dw;
-	unsigned i;
+	unsigned int i;
 	int r;
 
-	max_bytes = adev->mman.buffer_funcs->copy_max_bytes;
-	num_loops = DIV_ROUND_UP(byte_count, max_bytes);
-	num_dw = num_loops * adev->mman.buffer_funcs->copy_num_dw;
-
-	/* for IB padding */
-	while (num_dw & 0x7)
-		num_dw++;
-
-	r = amdgpu_job_alloc_with_ib(adev, num_dw * 4, &job);
+	max_bytes = adev->mman.buffer_funcs->fill_max_bytes;
+	num_loops = DIV_ROUND_UP_ULL(byte_count, max_bytes);
+	num_dw = ALIGN(num_loops * adev->mman.buffer_funcs->fill_num_dw, 8);
+	r = amdgpu_ttm_prepare_job(adev, false, num_dw, resv, vm_needs_flush,
+				   &job, delayed, k_job_id);
 	if (r)
 		return r;
 
-	job->vm_needs_flush = vm_needs_flush;
-	if (resv) {
-		r = amdgpu_sync_resv(adev, &job->sync, resv,
-				     AMDGPU_FENCE_OWNER_UNDEFINED);
-		if (r) {
-			DRM_ERROR("sync failed (%d).\n", r);
-			goto error_free;
-		}
-	}
-
 	for (i = 0; i < num_loops; i++) {
-		uint32_t cur_size_in_bytes = min(byte_count, max_bytes);
+		uint32_t cur_size = min(byte_count, max_bytes);
 
-		amdgpu_emit_copy_buffer(adev, &job->ibs[0], src_offset,
-					dst_offset, cur_size_in_bytes);
+		amdgpu_emit_fill_buffer(adev, &job->ibs[0], src_data, dst_addr,
+					cur_size);
 
-		src_offset += cur_size_in_bytes;
-		dst_offset += cur_size_in_bytes;
-		byte_count -= cur_size_in_bytes;
+		dst_addr += cur_size;
+		byte_count -= cur_size;
 	}
 
 	amdgpu_ring_pad_ib(ring, &job->ibs[0]);
 	WARN_ON(job->ibs[0].length_dw > num_dw);
-	if (direct_submit) {
-		r = amdgpu_ib_schedule(ring, job->num_ibs, job->ibs,
-				       NULL, fence);
-		job->fence = dma_fence_get(*fence);
-		if (r)
-			DRM_ERROR("Error scheduling IBs (%d)\n", r);
-		amdgpu_job_free(job);
-	} else {
-		r = amdgpu_job_submit(job, ring, &adev->mman.entity,
-				      AMDGPU_FENCE_OWNER_UNDEFINED, fence);
-		if (r)
-			goto error_free;
-	}
-
-	return r;
-
-error_free:
-	amdgpu_job_free(job);
-	return r;
+	*fence = amdgpu_job_submit(job);
+	return 0;
 }
 
-int amdgpu_fill_buffer(struct amdgpu_bo *bo,
-		       uint64_t src_data,
-		       struct reservation_object *resv,
-		       struct dma_fence **fence)
+/**
+ * amdgpu_ttm_clear_buffer - clear memory buffers
+ * @bo: amdgpu buffer object
+ * @resv: reservation object
+ * @fence: dma_fence associated with the operation
+ *
+ * Clear the memory buffer resource.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_ttm_clear_buffer(struct amdgpu_bo *bo,
+			    struct dma_resv *resv,
+			    struct dma_fence **fence)
 {
 	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
-	/* max_bytes applies to SDMA_OP_PTEPDE as well as SDMA_OP_CONST_FILL*/
-	uint32_t max_bytes = adev->mman.buffer_funcs->fill_max_bytes;
 	struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
+	struct amdgpu_res_cursor cursor;
+	u64 addr;
+	int r = 0;
 
-	struct drm_mm_node *mm_node;
-	unsigned long num_pages;
-	unsigned int num_loops, num_dw;
-
-	struct amdgpu_job *job;
-	int r;
+	if (!adev->mman.buffer_funcs_enabled)
+		return -EINVAL;
 
-	if (!ring->ready) {
-		DRM_ERROR("Trying to clear memory with ring turned off.\n");
+	if (!fence)
 		return -EINVAL;
-	}
 
-	if (bo->tbo.mem.mem_type == TTM_PL_TT) {
-		r = amdgpu_ttm_bind(&bo->tbo, &bo->tbo.mem);
+	*fence = dma_fence_get_stub();
+
+	amdgpu_res_first(bo->tbo.resource, 0, amdgpu_bo_size(bo), &cursor);
+
+	mutex_lock(&adev->mman.gtt_window_lock);
+	while (cursor.remaining) {
+		struct dma_fence *next = NULL;
+		u64 size;
+
+		if (amdgpu_res_cleared(&cursor)) {
+			amdgpu_res_next(&cursor, cursor.size);
+			continue;
+		}
+
+		/* Never clear more than 256MiB at once to avoid timeouts */
+		size = min(cursor.size, 256ULL << 20);
+
+		r = amdgpu_ttm_map_buffer(&bo->tbo, bo->tbo.resource, &cursor,
+					  1, ring, false, &size, &addr);
 		if (r)
-			return r;
-	}
+			goto err;
 
-	num_pages = bo->tbo.num_pages;
-	mm_node = bo->tbo.mem.mm_node;
-	num_loops = 0;
-	while (num_pages) {
-		uint32_t byte_count = mm_node->size << PAGE_SHIFT;
+		r = amdgpu_ttm_fill_mem(ring, 0, addr, size, resv,
+					&next, true, true,
+					AMDGPU_KERNEL_JOB_ID_TTM_CLEAR_BUFFER);
+		if (r)
+			goto err;
 
-		num_loops += DIV_ROUND_UP(byte_count, max_bytes);
-		num_pages -= mm_node->size;
-		++mm_node;
-	}
+		dma_fence_put(*fence);
+		*fence = next;
 
-	/* 10 double words for each SDMA_OP_PTEPDE cmd */
-	num_dw = num_loops * 10;
+		amdgpu_res_next(&cursor, size);
+	}
+err:
+	mutex_unlock(&adev->mman.gtt_window_lock);
 
-	/* for IB padding */
-	num_dw += 64;
+	return r;
+}
 
-	r = amdgpu_job_alloc_with_ib(adev, num_dw * 4, &job);
-	if (r)
-		return r;
+int amdgpu_fill_buffer(struct amdgpu_bo *bo,
+			uint32_t src_data,
+			struct dma_resv *resv,
+			struct dma_fence **f,
+			bool delayed,
+			u64 k_job_id)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
+	struct dma_fence *fence = NULL;
+	struct amdgpu_res_cursor dst;
+	int r;
 
-	if (resv) {
-		r = amdgpu_sync_resv(adev, &job->sync, resv,
-				     AMDGPU_FENCE_OWNER_UNDEFINED);
-		if (r) {
-			DRM_ERROR("sync failed (%d).\n", r);
-			goto error_free;
-		}
+	if (!adev->mman.buffer_funcs_enabled) {
+		dev_err(adev->dev,
+			"Trying to clear memory with ring turned off.\n");
+		return -EINVAL;
 	}
 
-	num_pages = bo->tbo.num_pages;
-	mm_node = bo->tbo.mem.mm_node;
+	amdgpu_res_first(bo->tbo.resource, 0, amdgpu_bo_size(bo), &dst);
 
-	while (num_pages) {
-		uint32_t byte_count = mm_node->size << PAGE_SHIFT;
-		uint64_t dst_addr;
+	mutex_lock(&adev->mman.gtt_window_lock);
+	while (dst.remaining) {
+		struct dma_fence *next;
+		uint64_t cur_size, to;
 
-		WARN_ONCE(byte_count & 0x7, "size should be a multiple of 8");
+		/* Never fill more than 256MiB at once to avoid timeouts */
+		cur_size = min(dst.size, 256ULL << 20);
 
-		dst_addr = amdgpu_mm_node_addr(&bo->tbo, mm_node, &bo->tbo.mem);
-		while (byte_count) {
-			uint32_t cur_size_in_bytes = min(byte_count, max_bytes);
+		r = amdgpu_ttm_map_buffer(&bo->tbo, bo->tbo.resource, &dst,
+					  1, ring, false, &cur_size, &to);
+		if (r)
+			goto error;
 
-			amdgpu_vm_set_pte_pde(adev, &job->ibs[0],
-					dst_addr, 0,
-					cur_size_in_bytes >> 3, 0,
-					src_data);
+		r = amdgpu_ttm_fill_mem(ring, src_data, to, cur_size, resv,
+					&next, true, delayed, k_job_id);
+		if (r)
+			goto error;
 
-			dst_addr += cur_size_in_bytes;
-			byte_count -= cur_size_in_bytes;
-		}
+		dma_fence_put(fence);
+		fence = next;
 
-		num_pages -= mm_node->size;
-		++mm_node;
+		amdgpu_res_next(&dst, cur_size);
 	}
+error:
+	mutex_unlock(&adev->mman.gtt_window_lock);
+	if (f)
+		*f = dma_fence_get(fence);
+	dma_fence_put(fence);
+	return r;
+}
 
-	amdgpu_ring_pad_ib(ring, &job->ibs[0]);
-	WARN_ON(job->ibs[0].length_dw > num_dw);
-	r = amdgpu_job_submit(job, ring, &adev->mman.entity,
-			      AMDGPU_FENCE_OWNER_UNDEFINED, fence);
-	if (r)
-		goto error_free;
+/**
+ * amdgpu_ttm_evict_resources - evict memory buffers
+ * @adev: amdgpu device object
+ * @mem_type: evicted BO's memory type
+ *
+ * Evicts all @mem_type buffers on the lru list of the memory type.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_ttm_evict_resources(struct amdgpu_device *adev, int mem_type)
+{
+	struct ttm_resource_manager *man;
 
-	return 0;
+	switch (mem_type) {
+	case TTM_PL_VRAM:
+	case TTM_PL_TT:
+	case AMDGPU_PL_GWS:
+	case AMDGPU_PL_GDS:
+	case AMDGPU_PL_OA:
+		man = ttm_manager_type(&adev->mman.bdev, mem_type);
+		break;
+	default:
+		dev_err(adev->dev, "Trying to evict invalid memory type\n");
+		return -EINVAL;
+	}
 
-error_free:
-	amdgpu_job_free(job);
-	return r;
+	return ttm_resource_manager_evict_all(&adev->mman.bdev, man);
 }
 
 #if defined(CONFIG_DEBUG_FS)
 
-static int amdgpu_mm_dump_table(struct seq_file *m, void *data)
+static int amdgpu_ttm_page_pool_show(struct seq_file *m, void *unused)
 {
-	struct drm_info_node *node = (struct drm_info_node *)m->private;
-	unsigned ttm_pl = *(int *)node->info_ent->data;
-	struct drm_device *dev = node->minor->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-	struct ttm_mem_type_manager *man = &adev->mman.bdev.man[ttm_pl];
-	struct drm_printer p = drm_seq_file_printer(m);
+	struct amdgpu_device *adev = m->private;
 
-	man->func->debug(man, &p);
-	return 0;
+	return ttm_pool_debugfs(&adev->mman.bdev.pool, m);
 }
 
-static int ttm_pl_vram = TTM_PL_VRAM;
-static int ttm_pl_tt = TTM_PL_TT;
-
-static const struct drm_info_list amdgpu_ttm_debugfs_list[] = {
-	{"amdgpu_vram_mm", amdgpu_mm_dump_table, 0, &ttm_pl_vram},
-	{"amdgpu_gtt_mm", amdgpu_mm_dump_table, 0, &ttm_pl_tt},
-	{"ttm_page_pool", ttm_page_alloc_debugfs, 0, NULL},
-#ifdef CONFIG_SWIOTLB
-	{"ttm_dma_page_pool", ttm_dma_page_alloc_debugfs, 0, NULL}
-#endif
-};
+DEFINE_SHOW_ATTRIBUTE(amdgpu_ttm_page_pool);
 
+/*
+ * amdgpu_ttm_vram_read - Linear read access to VRAM
+ *
+ * Accesses VRAM via MMIO for debugging purposes.
+ */
 static ssize_t amdgpu_ttm_vram_read(struct file *f, char __user *buf,
 				    size_t size, loff_t *pos)
 {
 	struct amdgpu_device *adev = file_inode(f)->i_private;
 	ssize_t result = 0;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	if (*pos >= adev->gmc.mc_vram_size)
+		return -ENXIO;
+
+	size = min(size, (size_t)(adev->gmc.mc_vram_size - *pos));
+	while (size) {
+		size_t bytes = min(size, AMDGPU_TTM_VRAM_MAX_DW_READ * 4);
+		uint32_t value[AMDGPU_TTM_VRAM_MAX_DW_READ];
+
+		amdgpu_device_vram_access(adev, *pos, value, bytes, false);
+		if (copy_to_user(buf, value, bytes))
+			return -EFAULT;
+
+		result += bytes;
+		buf += bytes;
+		*pos += bytes;
+		size -= bytes;
+	}
+
+	return result;
+}
+
+/*
+ * amdgpu_ttm_vram_write - Linear write access to VRAM
+ *
+ * Accesses VRAM via MMIO for debugging purposes.
+ */
+static ssize_t amdgpu_ttm_vram_write(struct file *f, const char __user *buf,
+				    size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
 	int r;
 
 	if (size & 0x3 || *pos & 0x3)
 		return -EINVAL;
 
-	if (*pos >= adev->mc.mc_vram_size)
+	if (*pos >= adev->gmc.mc_vram_size)
 		return -ENXIO;
 
 	while (size) {
-		unsigned long flags;
 		uint32_t value;
 
-		if (*pos >= adev->mc.mc_vram_size)
+		if (*pos >= adev->gmc.mc_vram_size)
 			return result;
 
-		spin_lock_irqsave(&adev->mmio_idx_lock, flags);
-		WREG32(mmMM_INDEX, ((uint32_t)*pos) | 0x80000000);
-		WREG32(mmMM_INDEX_HI, *pos >> 31);
-		value = RREG32(mmMM_DATA);
-		spin_unlock_irqrestore(&adev->mmio_idx_lock, flags);
-
-		r = put_user(value, (uint32_t *)buf);
+		r = get_user(value, (uint32_t *)buf);
 		if (r)
 			return r;
 
+		amdgpu_device_mm_access(adev, *pos, &value, 4, true);
+
 		result += 4;
 		buf += 4;
 		*pos += 4;
@@ -1718,109 +2609,153 @@ static ssize_t amdgpu_ttm_vram_read(struct file *f, char __user *buf,
 static const struct file_operations amdgpu_ttm_vram_fops = {
 	.owner = THIS_MODULE,
 	.read = amdgpu_ttm_vram_read,
-	.llseek = default_llseek
+	.write = amdgpu_ttm_vram_write,
+	.llseek = default_llseek,
 };
 
-#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS
-
-static ssize_t amdgpu_ttm_gtt_read(struct file *f, char __user *buf,
-				   size_t size, loff_t *pos)
+/*
+ * amdgpu_iomem_read - Virtual read access to GPU mapped memory
+ *
+ * This function is used to read memory that has been mapped to the
+ * GPU and the known addresses are not physical addresses but instead
+ * bus addresses (e.g., what you'd put in an IB or ring buffer).
+ */
+static ssize_t amdgpu_iomem_read(struct file *f, char __user *buf,
+				 size_t size, loff_t *pos)
 {
 	struct amdgpu_device *adev = file_inode(f)->i_private;
+	struct iommu_domain *dom;
 	ssize_t result = 0;
 	int r;
 
+	/* retrieve the IOMMU domain if any for this device */
+	dom = iommu_get_domain_for_dev(adev->dev);
+
 	while (size) {
-		loff_t p = *pos / PAGE_SIZE;
-		unsigned off = *pos & ~PAGE_MASK;
-		size_t cur_size = min_t(size_t, size, PAGE_SIZE - off);
-		struct page *page;
+		phys_addr_t addr = *pos & PAGE_MASK;
+		loff_t off = *pos & ~PAGE_MASK;
+		size_t bytes = PAGE_SIZE - off;
+		unsigned long pfn;
+		struct page *p;
 		void *ptr;
 
-		if (p >= adev->gart.num_cpu_pages)
-			return result;
+		bytes = min(bytes, size);
+
+		/* Translate the bus address to a physical address.  If
+		 * the domain is NULL it means there is no IOMMU active
+		 * and the address translation is the identity
+		 */
+		addr = dom ? iommu_iova_to_phys(dom, addr) : addr;
 
-		page = adev->gart.pages[p];
-		if (page) {
-			ptr = kmap(page);
-			ptr += off;
+		pfn = addr >> PAGE_SHIFT;
+		if (!pfn_valid(pfn))
+			return -EPERM;
 
-			r = copy_to_user(buf, ptr, cur_size);
-			kunmap(adev->gart.pages[p]);
-		} else
-			r = clear_user(buf, cur_size);
+		p = pfn_to_page(pfn);
+		if (p->mapping != adev->mman.bdev.dev_mapping)
+			return -EPERM;
 
+		ptr = kmap_local_page(p);
+		r = copy_to_user(buf, ptr + off, bytes);
+		kunmap_local(ptr);
 		if (r)
 			return -EFAULT;
 
-		result += cur_size;
-		buf += cur_size;
-		*pos += cur_size;
-		size -= cur_size;
+		size -= bytes;
+		*pos += bytes;
+		result += bytes;
 	}
 
 	return result;
 }
 
-static const struct file_operations amdgpu_ttm_gtt_fops = {
-	.owner = THIS_MODULE,
-	.read = amdgpu_ttm_gtt_read,
-	.llseek = default_llseek
-};
+/*
+ * amdgpu_iomem_write - Virtual write access to GPU mapped memory
+ *
+ * This function is used to write memory that has been mapped to the
+ * GPU and the known addresses are not physical addresses but instead
+ * bus addresses (e.g., what you'd put in an IB or ring buffer).
+ */
+static ssize_t amdgpu_iomem_write(struct file *f, const char __user *buf,
+				 size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	struct iommu_domain *dom;
+	ssize_t result = 0;
+	int r;
 
-#endif
+	dom = iommu_get_domain_for_dev(adev->dev);
 
-#endif
+	while (size) {
+		phys_addr_t addr = *pos & PAGE_MASK;
+		loff_t off = *pos & ~PAGE_MASK;
+		size_t bytes = PAGE_SIZE - off;
+		unsigned long pfn;
+		struct page *p;
+		void *ptr;
 
-static int amdgpu_ttm_debugfs_init(struct amdgpu_device *adev)
-{
-#if defined(CONFIG_DEBUG_FS)
-	unsigned count;
-
-	struct drm_minor *minor = adev->ddev->primary;
-	struct dentry *ent, *root = minor->debugfs_root;
-
-	ent = debugfs_create_file("amdgpu_vram", S_IFREG | S_IRUGO, root,
-				  adev, &amdgpu_ttm_vram_fops);
-	if (IS_ERR(ent))
-		return PTR_ERR(ent);
-	i_size_write(ent->d_inode, adev->mc.mc_vram_size);
-	adev->mman.vram = ent;
-
-#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS
-	ent = debugfs_create_file("amdgpu_gtt", S_IFREG | S_IRUGO, root,
-				  adev, &amdgpu_ttm_gtt_fops);
-	if (IS_ERR(ent))
-		return PTR_ERR(ent);
-	i_size_write(ent->d_inode, adev->mc.gart_size);
-	adev->mman.gtt = ent;
+		bytes = min(bytes, size);
 
-#endif
-	count = ARRAY_SIZE(amdgpu_ttm_debugfs_list);
+		addr = dom ? iommu_iova_to_phys(dom, addr) : addr;
 
-#ifdef CONFIG_SWIOTLB
-	if (!swiotlb_nr_tbl())
-		--count;
-#endif
+		pfn = addr >> PAGE_SHIFT;
+		if (!pfn_valid(pfn))
+			return -EPERM;
 
-	return amdgpu_debugfs_add_files(adev, amdgpu_ttm_debugfs_list, count);
-#else
+		p = pfn_to_page(pfn);
+		if (p->mapping != adev->mman.bdev.dev_mapping)
+			return -EPERM;
 
-	return 0;
-#endif
-}
+		ptr = kmap_local_page(p);
+		r = copy_from_user(ptr + off, buf, bytes);
+		kunmap_local(ptr);
+		if (r)
+			return -EFAULT;
 
-static void amdgpu_ttm_debugfs_fini(struct amdgpu_device *adev)
-{
-#if defined(CONFIG_DEBUG_FS)
+		size -= bytes;
+		*pos += bytes;
+		result += bytes;
+	}
 
-	debugfs_remove(adev->mman.vram);
-	adev->mman.vram = NULL;
+	return result;
+}
+
+static const struct file_operations amdgpu_ttm_iomem_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_iomem_read,
+	.write = amdgpu_iomem_write,
+	.llseek = default_llseek
+};
 
-#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS
-	debugfs_remove(adev->mman.gtt);
-	adev->mman.gtt = NULL;
 #endif
 
+void amdgpu_ttm_debugfs_init(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	struct drm_minor *minor = adev_to_drm(adev)->primary;
+	struct dentry *root = minor->debugfs_root;
+
+	debugfs_create_file_size("amdgpu_vram", 0444, root, adev,
+				 &amdgpu_ttm_vram_fops, adev->gmc.mc_vram_size);
+	debugfs_create_file("amdgpu_iomem", 0444, root, adev,
+			    &amdgpu_ttm_iomem_fops);
+	debugfs_create_file("ttm_page_pool", 0444, root, adev,
+			    &amdgpu_ttm_page_pool_fops);
+	ttm_resource_manager_create_debugfs(ttm_manager_type(&adev->mman.bdev,
+							     TTM_PL_VRAM),
+					    root, "amdgpu_vram_mm");
+	ttm_resource_manager_create_debugfs(ttm_manager_type(&adev->mman.bdev,
+							     TTM_PL_TT),
+					    root, "amdgpu_gtt_mm");
+	ttm_resource_manager_create_debugfs(ttm_manager_type(&adev->mman.bdev,
+							     AMDGPU_PL_GDS),
+					    root, "amdgpu_gds_mm");
+	ttm_resource_manager_create_debugfs(ttm_manager_type(&adev->mman.bdev,
+							     AMDGPU_PL_GWS),
+					    root, "amdgpu_gws_mm");
+	ttm_resource_manager_create_debugfs(ttm_manager_type(&adev->mman.bdev,
+							     AMDGPU_PL_OA),
+					    root, "amdgpu_oa_mm");
+
 #endif
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
index 43093bffa2cf..0be2728aa872 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
@@ -24,62 +24,214 @@
 #ifndef __AMDGPU_TTM_H__
 #define __AMDGPU_TTM_H__
 
-#include "gpu_scheduler.h"
+#include <linux/dma-direction.h>
+#include <drm/gpu_scheduler.h>
+#include <drm/ttm/ttm_placement.h>
+#include "amdgpu_vram_mgr.h"
 
 #define AMDGPU_PL_GDS		(TTM_PL_PRIV + 0)
 #define AMDGPU_PL_GWS		(TTM_PL_PRIV + 1)
 #define AMDGPU_PL_OA		(TTM_PL_PRIV + 2)
-
-#define AMDGPU_PL_FLAG_GDS		(TTM_PL_FLAG_PRIV << 0)
-#define AMDGPU_PL_FLAG_GWS		(TTM_PL_FLAG_PRIV << 1)
-#define AMDGPU_PL_FLAG_OA		(TTM_PL_FLAG_PRIV << 2)
+#define AMDGPU_PL_PREEMPT	(TTM_PL_PRIV + 3)
+#define AMDGPU_PL_DOORBELL	(TTM_PL_PRIV + 4)
+#define AMDGPU_PL_MMIO_REMAP	(TTM_PL_PRIV + 5)
+#define __AMDGPU_PL_NUM	(TTM_PL_PRIV + 6)
 
 #define AMDGPU_GTT_MAX_TRANSFER_SIZE	512
 #define AMDGPU_GTT_NUM_TRANSFER_WINDOWS	2
 
+extern const struct attribute_group amdgpu_vram_mgr_attr_group;
+extern const struct attribute_group amdgpu_gtt_mgr_attr_group;
+
+struct hmm_range;
+
+struct amdgpu_gtt_mgr {
+	struct ttm_resource_manager manager;
+	struct drm_mm mm;
+	spinlock_t lock;
+};
+
 struct amdgpu_mman {
-	struct ttm_bo_global_ref        bo_global_ref;
-	struct drm_global_reference	mem_global_ref;
-	struct ttm_bo_device		bdev;
-	bool				mem_global_referenced;
+	struct ttm_device		bdev;
+	struct ttm_pool			*ttm_pools;
 	bool				initialized;
-
-#if defined(CONFIG_DEBUG_FS)
-	struct dentry			*vram;
-	struct dentry			*gtt;
-#endif
+	void __iomem			*aper_base_kaddr;
 
 	/* buffer handling */
 	const struct amdgpu_buffer_funcs	*buffer_funcs;
 	struct amdgpu_ring			*buffer_funcs_ring;
+	bool					buffer_funcs_enabled;
 
 	struct mutex				gtt_window_lock;
-	/* Scheduler entity for buffer moves */
-	struct amd_sched_entity			entity;
+	/* High priority scheduler entity for buffer moves */
+	struct drm_sched_entity			high_pr;
+	/* Low priority scheduler entity for VRAM clearing */
+	struct drm_sched_entity			low_pr;
+
+	struct amdgpu_vram_mgr vram_mgr;
+	struct amdgpu_gtt_mgr gtt_mgr;
+	struct ttm_resource_manager preempt_mgr;
+
+	uint64_t		stolen_vga_size;
+	struct amdgpu_bo	*stolen_vga_memory;
+	uint64_t		stolen_extended_size;
+	struct amdgpu_bo	*stolen_extended_memory;
+	bool			keep_stolen_vga_memory;
+
+	struct amdgpu_bo	*stolen_reserved_memory;
+	uint64_t		stolen_reserved_offset;
+	uint64_t		stolen_reserved_size;
+
+	/* discovery */
+	uint8_t				*discovery_bin;
+	uint32_t			discovery_tmr_size;
+	/* fw reserved memory */
+	struct amdgpu_bo		*fw_reserved_memory;
+	struct amdgpu_bo		*fw_reserved_memory_extend;
+
+	/* firmware VRAM reservation */
+	u64		fw_vram_usage_start_offset;
+	u64		fw_vram_usage_size;
+	struct amdgpu_bo	*fw_vram_usage_reserved_bo;
+	void		*fw_vram_usage_va;
+
+	/* driver VRAM reservation */
+	u64		drv_vram_usage_start_offset;
+	u64		drv_vram_usage_size;
+	struct amdgpu_bo	*drv_vram_usage_reserved_bo;
+	void		*drv_vram_usage_va;
+
+	/* PAGE_SIZE'd BO for process memory r/w over SDMA. */
+	struct amdgpu_bo	*sdma_access_bo;
+	void			*sdma_access_ptr;
+};
+
+struct amdgpu_copy_mem {
+	struct ttm_buffer_object	*bo;
+	struct ttm_resource		*mem;
+	unsigned long			offset;
 };
 
-extern const struct ttm_mem_type_manager_func amdgpu_gtt_mgr_func;
-extern const struct ttm_mem_type_manager_func amdgpu_vram_mgr_func;
+#define AMDGPU_COPY_FLAGS_TMZ		(1 << 0)
+#define AMDGPU_COPY_FLAGS_READ_DECOMPRESSED	(1 << 1)
+#define AMDGPU_COPY_FLAGS_WRITE_COMPRESSED	(1 << 2)
+#define AMDGPU_COPY_FLAGS_MAX_COMPRESSED_SHIFT		3
+#define AMDGPU_COPY_FLAGS_MAX_COMPRESSED_MASK		0x03
+#define AMDGPU_COPY_FLAGS_NUMBER_TYPE_SHIFT		5
+#define AMDGPU_COPY_FLAGS_NUMBER_TYPE_MASK		0x07
+#define AMDGPU_COPY_FLAGS_DATA_FORMAT_SHIFT		8
+#define AMDGPU_COPY_FLAGS_DATA_FORMAT_MASK		0x3f
+#define AMDGPU_COPY_FLAGS_WRITE_COMPRESS_DISABLE_SHIFT	14
+#define AMDGPU_COPY_FLAGS_WRITE_COMPRESS_DISABLE_MASK	0x1
+
+#define AMDGPU_COPY_FLAGS_SET(field, value) \
+	(((__u32)(value) & AMDGPU_COPY_FLAGS_##field##_MASK) << AMDGPU_COPY_FLAGS_##field##_SHIFT)
+#define AMDGPU_COPY_FLAGS_GET(value, field) \
+	(((__u32)(value) >> AMDGPU_COPY_FLAGS_##field##_SHIFT) & AMDGPU_COPY_FLAGS_##field##_MASK)
+
+int amdgpu_gtt_mgr_init(struct amdgpu_device *adev, uint64_t gtt_size);
+void amdgpu_gtt_mgr_fini(struct amdgpu_device *adev);
+int amdgpu_preempt_mgr_init(struct amdgpu_device *adev);
+void amdgpu_preempt_mgr_fini(struct amdgpu_device *adev);
+int amdgpu_vram_mgr_init(struct amdgpu_device *adev);
+void amdgpu_vram_mgr_fini(struct amdgpu_device *adev);
+
+bool amdgpu_gtt_mgr_has_gart_addr(struct ttm_resource *mem);
+void amdgpu_gtt_mgr_recover(struct amdgpu_gtt_mgr *mgr);
 
-bool amdgpu_gtt_mgr_is_allocated(struct ttm_mem_reg *mem);
-uint64_t amdgpu_gtt_mgr_usage(struct ttm_mem_type_manager *man);
+uint64_t amdgpu_preempt_mgr_usage(struct ttm_resource_manager *man);
 
-uint64_t amdgpu_vram_mgr_usage(struct ttm_mem_type_manager *man);
-uint64_t amdgpu_vram_mgr_vis_usage(struct ttm_mem_type_manager *man);
+u64 amdgpu_vram_mgr_bo_visible_size(struct amdgpu_bo *bo);
+int amdgpu_vram_mgr_alloc_sgt(struct amdgpu_device *adev,
+			      struct ttm_resource *mem,
+			      u64 offset, u64 size,
+			      struct device *dev,
+			      enum dma_data_direction dir,
+			      struct sg_table **sgt);
+void amdgpu_vram_mgr_free_sgt(struct device *dev,
+			      enum dma_data_direction dir,
+			      struct sg_table *sgt);
+uint64_t amdgpu_vram_mgr_vis_usage(struct amdgpu_vram_mgr *mgr);
+int amdgpu_vram_mgr_reserve_range(struct amdgpu_vram_mgr *mgr,
+				  uint64_t start, uint64_t size);
+int amdgpu_vram_mgr_query_page_status(struct amdgpu_vram_mgr *mgr,
+				      uint64_t start);
+void amdgpu_vram_mgr_clear_reset_blocks(struct amdgpu_device *adev);
 
+bool amdgpu_res_cpu_visible(struct amdgpu_device *adev,
+			    struct ttm_resource *res);
+
+int amdgpu_ttm_init(struct amdgpu_device *adev);
+void amdgpu_ttm_fini(struct amdgpu_device *adev);
+void amdgpu_ttm_set_buffer_funcs_status(struct amdgpu_device *adev,
+					bool enable);
 int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
 		       uint64_t dst_offset, uint32_t byte_count,
-		       struct reservation_object *resv,
+		       struct dma_resv *resv,
 		       struct dma_fence **fence, bool direct_submit,
-		       bool vm_needs_flush);
+		       bool vm_needs_flush, uint32_t copy_flags);
+int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
+			       const struct amdgpu_copy_mem *src,
+			       const struct amdgpu_copy_mem *dst,
+			       uint64_t size, bool tmz,
+			       struct dma_resv *resv,
+			       struct dma_fence **f);
+int amdgpu_ttm_clear_buffer(struct amdgpu_bo *bo,
+			    struct dma_resv *resv,
+			    struct dma_fence **fence);
 int amdgpu_fill_buffer(struct amdgpu_bo *bo,
-			uint64_t src_data,
-			struct reservation_object *resv,
-			struct dma_fence **fence);
-
-int amdgpu_mmap(struct file *filp, struct vm_area_struct *vma);
-bool amdgpu_ttm_is_bound(struct ttm_tt *ttm);
-int amdgpu_ttm_bind(struct ttm_buffer_object *bo, struct ttm_mem_reg *bo_mem);
-int amdgpu_ttm_recover_gart(struct amdgpu_device *adev);
+			uint32_t src_data,
+			struct dma_resv *resv,
+			struct dma_fence **fence,
+			bool delayed,
+			u64 k_job_id);
+
+int amdgpu_ttm_alloc_gart(struct ttm_buffer_object *bo);
+void amdgpu_ttm_recover_gart(struct ttm_buffer_object *tbo);
+uint64_t amdgpu_ttm_domain_start(struct amdgpu_device *adev, uint32_t type);
+
+#if IS_ENABLED(CONFIG_DRM_AMDGPU_USERPTR)
+int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo,
+				 struct hmm_range **range);
+void amdgpu_ttm_tt_discard_user_pages(struct ttm_tt *ttm,
+				      struct hmm_range *range);
+bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm,
+				       struct hmm_range *range);
+#else
+static inline int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo,
+					       struct hmm_range **range)
+{
+	return -EPERM;
+}
+static inline void amdgpu_ttm_tt_discard_user_pages(struct ttm_tt *ttm,
+						    struct hmm_range *range)
+{
+}
+static inline bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm,
+						     struct hmm_range *range)
+{
+	return false;
+}
+#endif
+
+void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct hmm_range *range);
+int amdgpu_ttm_tt_get_userptr(const struct ttm_buffer_object *tbo,
+			      uint64_t *user_addr);
+int amdgpu_ttm_tt_set_userptr(struct ttm_buffer_object *bo,
+			      uint64_t addr, uint32_t flags);
+bool amdgpu_ttm_tt_has_userptr(struct ttm_tt *ttm);
+struct mm_struct *amdgpu_ttm_tt_get_usermm(struct ttm_tt *ttm);
+bool amdgpu_ttm_tt_affect_userptr(struct ttm_tt *ttm, unsigned long start,
+				  unsigned long end, unsigned long *userptr);
+bool amdgpu_ttm_tt_userptr_invalidated(struct ttm_tt *ttm,
+				       int *last_invalidated);
+bool amdgpu_ttm_tt_is_userptr(struct ttm_tt *ttm);
+bool amdgpu_ttm_tt_is_readonly(struct ttm_tt *ttm);
+uint64_t amdgpu_ttm_tt_pde_flags(struct ttm_tt *ttm, struct ttm_resource *mem);
+uint64_t amdgpu_ttm_tt_pte_flags(struct amdgpu_device *adev, struct ttm_tt *ttm,
+				 struct ttm_resource *mem);
+int amdgpu_ttm_evict_resources(struct amdgpu_device *adev, int mem_type);
+
+void amdgpu_ttm_debugfs_init(struct amdgpu_device *adev);
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c
index 36c763310df5..e96f24e9ad57 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c
@@ -24,10 +24,17 @@
 #include <linux/firmware.h>
 #include <linux/slab.h>
 #include <linux/module.h>
-#include <drm/drmP.h>
+
 #include "amdgpu.h"
 #include "amdgpu_ucode.h"
 
+#define AMDGPU_UCODE_NAME_MAX		(128)
+
+static const struct kicker_device kicker_device_list[] = {
+	{0x744B, 0x00},
+	{0x7551, 0xC8}
+};
+
 static void amdgpu_ucode_print_common_hdr(const struct common_firmware_header *hdr)
 {
 	DRM_DEBUG("size_bytes: %u\n", le32_to_cpu(hdr->size_bytes));
@@ -68,15 +75,32 @@ void amdgpu_ucode_print_smc_hdr(const struct common_firmware_header *hdr)
 {
 	uint16_t version_major = le16_to_cpu(hdr->header_version_major);
 	uint16_t version_minor = le16_to_cpu(hdr->header_version_minor);
+	const struct smc_firmware_header_v1_0 *v1_0_hdr;
+	const struct smc_firmware_header_v2_0 *v2_0_hdr;
+	const struct smc_firmware_header_v2_1 *v2_1_hdr;
 
 	DRM_DEBUG("SMC\n");
 	amdgpu_ucode_print_common_hdr(hdr);
 
 	if (version_major == 1) {
-		const struct smc_firmware_header_v1_0 *smc_hdr =
-			container_of(hdr, struct smc_firmware_header_v1_0, header);
+		v1_0_hdr = container_of(hdr, struct smc_firmware_header_v1_0, header);
+		DRM_DEBUG("ucode_start_addr: %u\n", le32_to_cpu(v1_0_hdr->ucode_start_addr));
+	} else if (version_major == 2) {
+		switch (version_minor) {
+		case 0:
+			v2_0_hdr = container_of(hdr, struct smc_firmware_header_v2_0, v1_0.header);
+			DRM_DEBUG("ppt_offset_bytes: %u\n", le32_to_cpu(v2_0_hdr->ppt_offset_bytes));
+			DRM_DEBUG("ppt_size_bytes: %u\n", le32_to_cpu(v2_0_hdr->ppt_size_bytes));
+			break;
+		case 1:
+			v2_1_hdr = container_of(hdr, struct smc_firmware_header_v2_1, v1_0.header);
+			DRM_DEBUG("pptable_count: %u\n", le32_to_cpu(v2_1_hdr->pptable_count));
+			DRM_DEBUG("pptable_entry_offset: %u\n", le32_to_cpu(v2_1_hdr->pptable_entry_offset));
+			break;
+		default:
+			break;
+		}
 
-		DRM_DEBUG("ucode_start_addr: %u\n", le32_to_cpu(smc_hdr->ucode_start_addr));
 	} else {
 		DRM_ERROR("Unknown SMC ucode version: %u.%u\n", version_major, version_minor);
 	}
@@ -98,6 +122,12 @@ void amdgpu_ucode_print_gfx_hdr(const struct common_firmware_header *hdr)
 			  le32_to_cpu(gfx_hdr->ucode_feature_version));
 		DRM_DEBUG("jt_offset: %u\n", le32_to_cpu(gfx_hdr->jt_offset));
 		DRM_DEBUG("jt_size: %u\n", le32_to_cpu(gfx_hdr->jt_size));
+	} else if (version_major == 2) {
+		const struct gfx_firmware_header_v2_0 *gfx_hdr =
+			container_of(hdr, struct gfx_firmware_header_v2_0, header);
+
+		DRM_DEBUG("ucode_feature_version: %u\n",
+			  le32_to_cpu(gfx_hdr->ucode_feature_version));
 	} else {
 		DRM_ERROR("Unknown GFX ucode version: %u.%u\n", version_major, version_minor);
 	}
@@ -128,41 +158,139 @@ void amdgpu_ucode_print_rlc_hdr(const struct common_firmware_header *hdr)
 	} else if (version_major == 2) {
 		const struct rlc_firmware_header_v2_0 *rlc_hdr =
 			container_of(hdr, struct rlc_firmware_header_v2_0, header);
-
-		DRM_DEBUG("ucode_feature_version: %u\n",
-			  le32_to_cpu(rlc_hdr->ucode_feature_version));
-		DRM_DEBUG("jt_offset: %u\n", le32_to_cpu(rlc_hdr->jt_offset));
-		DRM_DEBUG("jt_size: %u\n", le32_to_cpu(rlc_hdr->jt_size));
-		DRM_DEBUG("save_and_restore_offset: %u\n",
-			  le32_to_cpu(rlc_hdr->save_and_restore_offset));
-		DRM_DEBUG("clear_state_descriptor_offset: %u\n",
-			  le32_to_cpu(rlc_hdr->clear_state_descriptor_offset));
-		DRM_DEBUG("avail_scratch_ram_locations: %u\n",
-			  le32_to_cpu(rlc_hdr->avail_scratch_ram_locations));
-		DRM_DEBUG("reg_restore_list_size: %u\n",
-			  le32_to_cpu(rlc_hdr->reg_restore_list_size));
-		DRM_DEBUG("reg_list_format_start: %u\n",
-			  le32_to_cpu(rlc_hdr->reg_list_format_start));
-		DRM_DEBUG("reg_list_format_separate_start: %u\n",
-			  le32_to_cpu(rlc_hdr->reg_list_format_separate_start));
-		DRM_DEBUG("starting_offsets_start: %u\n",
-			  le32_to_cpu(rlc_hdr->starting_offsets_start));
-		DRM_DEBUG("reg_list_format_size_bytes: %u\n",
-			  le32_to_cpu(rlc_hdr->reg_list_format_size_bytes));
-		DRM_DEBUG("reg_list_format_array_offset_bytes: %u\n",
-			  le32_to_cpu(rlc_hdr->reg_list_format_array_offset_bytes));
-		DRM_DEBUG("reg_list_size_bytes: %u\n",
-			  le32_to_cpu(rlc_hdr->reg_list_size_bytes));
-		DRM_DEBUG("reg_list_array_offset_bytes: %u\n",
-			  le32_to_cpu(rlc_hdr->reg_list_array_offset_bytes));
-		DRM_DEBUG("reg_list_format_separate_size_bytes: %u\n",
-			  le32_to_cpu(rlc_hdr->reg_list_format_separate_size_bytes));
-		DRM_DEBUG("reg_list_format_separate_array_offset_bytes: %u\n",
-			  le32_to_cpu(rlc_hdr->reg_list_format_separate_array_offset_bytes));
-		DRM_DEBUG("reg_list_separate_size_bytes: %u\n",
-			  le32_to_cpu(rlc_hdr->reg_list_separate_size_bytes));
-		DRM_DEBUG("reg_list_separate_size_bytes: %u\n",
-			  le32_to_cpu(rlc_hdr->reg_list_separate_size_bytes));
+		const struct rlc_firmware_header_v2_1 *rlc_hdr_v2_1 =
+			container_of(rlc_hdr, struct rlc_firmware_header_v2_1, v2_0);
+		const struct rlc_firmware_header_v2_2 *rlc_hdr_v2_2 =
+			container_of(rlc_hdr_v2_1, struct rlc_firmware_header_v2_2, v2_1);
+		const struct rlc_firmware_header_v2_3 *rlc_hdr_v2_3 =
+			container_of(rlc_hdr_v2_2, struct rlc_firmware_header_v2_3, v2_2);
+		const struct rlc_firmware_header_v2_4 *rlc_hdr_v2_4 =
+			container_of(rlc_hdr_v2_3, struct rlc_firmware_header_v2_4, v2_3);
+
+		switch (version_minor) {
+		case 0:
+			/* rlc_hdr v2_0 */
+			DRM_DEBUG("ucode_feature_version: %u\n",
+				  le32_to_cpu(rlc_hdr->ucode_feature_version));
+			DRM_DEBUG("jt_offset: %u\n", le32_to_cpu(rlc_hdr->jt_offset));
+			DRM_DEBUG("jt_size: %u\n", le32_to_cpu(rlc_hdr->jt_size));
+			DRM_DEBUG("save_and_restore_offset: %u\n",
+				  le32_to_cpu(rlc_hdr->save_and_restore_offset));
+			DRM_DEBUG("clear_state_descriptor_offset: %u\n",
+				  le32_to_cpu(rlc_hdr->clear_state_descriptor_offset));
+			DRM_DEBUG("avail_scratch_ram_locations: %u\n",
+				  le32_to_cpu(rlc_hdr->avail_scratch_ram_locations));
+			DRM_DEBUG("reg_restore_list_size: %u\n",
+				  le32_to_cpu(rlc_hdr->reg_restore_list_size));
+			DRM_DEBUG("reg_list_format_start: %u\n",
+				  le32_to_cpu(rlc_hdr->reg_list_format_start));
+			DRM_DEBUG("reg_list_format_separate_start: %u\n",
+				  le32_to_cpu(rlc_hdr->reg_list_format_separate_start));
+			DRM_DEBUG("starting_offsets_start: %u\n",
+				  le32_to_cpu(rlc_hdr->starting_offsets_start));
+			DRM_DEBUG("reg_list_format_size_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr->reg_list_format_size_bytes));
+			DRM_DEBUG("reg_list_format_array_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr->reg_list_format_array_offset_bytes));
+			DRM_DEBUG("reg_list_size_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr->reg_list_size_bytes));
+			DRM_DEBUG("reg_list_array_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr->reg_list_array_offset_bytes));
+			DRM_DEBUG("reg_list_format_separate_size_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr->reg_list_format_separate_size_bytes));
+			DRM_DEBUG("reg_list_format_separate_array_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr->reg_list_format_separate_array_offset_bytes));
+			DRM_DEBUG("reg_list_separate_size_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr->reg_list_separate_size_bytes));
+			DRM_DEBUG("reg_list_separate_array_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr->reg_list_separate_array_offset_bytes));
+			break;
+		case 1:
+			/* rlc_hdr v2_1 */
+			DRM_DEBUG("reg_list_format_direct_reg_list_length: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_1->reg_list_format_direct_reg_list_length));
+			DRM_DEBUG("save_restore_list_cntl_ucode_ver: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_cntl_ucode_ver));
+			DRM_DEBUG("save_restore_list_cntl_feature_ver: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_cntl_feature_ver));
+			DRM_DEBUG("save_restore_list_cntl_size_bytes %u\n",
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_cntl_size_bytes));
+			DRM_DEBUG("save_restore_list_cntl_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_cntl_offset_bytes));
+			DRM_DEBUG("save_restore_list_gpm_ucode_ver: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_gpm_ucode_ver));
+			DRM_DEBUG("save_restore_list_gpm_feature_ver: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_gpm_feature_ver));
+			DRM_DEBUG("save_restore_list_gpm_size_bytes %u\n",
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_gpm_size_bytes));
+			DRM_DEBUG("save_restore_list_gpm_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_gpm_offset_bytes));
+			DRM_DEBUG("save_restore_list_srm_ucode_ver: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_srm_ucode_ver));
+			DRM_DEBUG("save_restore_list_srm_feature_ver: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_srm_feature_ver));
+			DRM_DEBUG("save_restore_list_srm_size_bytes %u\n",
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_srm_size_bytes));
+			DRM_DEBUG("save_restore_list_srm_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_srm_offset_bytes));
+			break;
+		case 2:
+			/* rlc_hdr v2_2 */
+			DRM_DEBUG("rlc_iram_ucode_size_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_2->rlc_iram_ucode_size_bytes));
+			DRM_DEBUG("rlc_iram_ucode_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_2->rlc_iram_ucode_offset_bytes));
+			DRM_DEBUG("rlc_dram_ucode_size_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_2->rlc_dram_ucode_size_bytes));
+			DRM_DEBUG("rlc_dram_ucode_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_2->rlc_dram_ucode_offset_bytes));
+			break;
+		case 3:
+			/* rlc_hdr v2_3 */
+			DRM_DEBUG("rlcp_ucode_version: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_3->rlcp_ucode_version));
+			DRM_DEBUG("rlcp_ucode_feature_version: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_3->rlcp_ucode_feature_version));
+			DRM_DEBUG("rlcp_ucode_size_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_3->rlcp_ucode_size_bytes));
+			DRM_DEBUG("rlcp_ucode_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_3->rlcp_ucode_offset_bytes));
+			DRM_DEBUG("rlcv_ucode_version: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_3->rlcv_ucode_version));
+			DRM_DEBUG("rlcv_ucode_feature_version: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_3->rlcv_ucode_feature_version));
+			DRM_DEBUG("rlcv_ucode_size_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_3->rlcv_ucode_size_bytes));
+			DRM_DEBUG("rlcv_ucode_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_3->rlcv_ucode_offset_bytes));
+			break;
+		case 4:
+			/* rlc_hdr v2_4 */
+			DRM_DEBUG("global_tap_delays_ucode_size_bytes :%u\n",
+				  le32_to_cpu(rlc_hdr_v2_4->global_tap_delays_ucode_size_bytes));
+			DRM_DEBUG("global_tap_delays_ucode_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_4->global_tap_delays_ucode_offset_bytes));
+			DRM_DEBUG("se0_tap_delays_ucode_size_bytes :%u\n",
+				  le32_to_cpu(rlc_hdr_v2_4->se0_tap_delays_ucode_size_bytes));
+			DRM_DEBUG("se0_tap_delays_ucode_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_4->se0_tap_delays_ucode_offset_bytes));
+			DRM_DEBUG("se1_tap_delays_ucode_size_bytes :%u\n",
+				  le32_to_cpu(rlc_hdr_v2_4->se1_tap_delays_ucode_size_bytes));
+			DRM_DEBUG("se1_tap_delays_ucode_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_4->se1_tap_delays_ucode_offset_bytes));
+			DRM_DEBUG("se2_tap_delays_ucode_size_bytes :%u\n",
+				  le32_to_cpu(rlc_hdr_v2_4->se2_tap_delays_ucode_size_bytes));
+			DRM_DEBUG("se2_tap_delays_ucode_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_4->se2_tap_delays_ucode_offset_bytes));
+			DRM_DEBUG("se3_tap_delays_ucode_size_bytes :%u\n",
+				  le32_to_cpu(rlc_hdr_v2_4->se3_tap_delays_ucode_size_bytes));
+			DRM_DEBUG("se3_tap_delays_ucode_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_4->se3_tap_delays_ucode_offset_bytes));
+			break;
+		default:
+			DRM_ERROR("Unknown RLC v2 ucode: v2.%u\n", version_minor);
+			break;
+		}
 	} else {
 		DRM_ERROR("Unknown RLC ucode version: %u.%u\n", version_major, version_minor);
 	}
@@ -191,12 +319,176 @@ void amdgpu_ucode_print_sdma_hdr(const struct common_firmware_header *hdr)
 				container_of(sdma_hdr, struct sdma_firmware_header_v1_1, v1_0);
 			DRM_DEBUG("digest_size: %u\n", le32_to_cpu(sdma_v1_1_hdr->digest_size));
 		}
+	} else if (version_major == 2) {
+		const struct sdma_firmware_header_v2_0 *sdma_hdr =
+			container_of(hdr, struct sdma_firmware_header_v2_0, header);
+
+		DRM_DEBUG("ucode_feature_version: %u\n",
+			  le32_to_cpu(sdma_hdr->ucode_feature_version));
+		DRM_DEBUG("ctx_jt_offset: %u\n", le32_to_cpu(sdma_hdr->ctx_jt_offset));
+		DRM_DEBUG("ctx_jt_size: %u\n", le32_to_cpu(sdma_hdr->ctx_jt_size));
+		DRM_DEBUG("ctl_ucode_offset: %u\n", le32_to_cpu(sdma_hdr->ctl_ucode_offset));
+		DRM_DEBUG("ctl_jt_offset: %u\n", le32_to_cpu(sdma_hdr->ctl_jt_offset));
+		DRM_DEBUG("ctl_jt_size: %u\n", le32_to_cpu(sdma_hdr->ctl_jt_size));
+	} else if (version_major == 3) {
+		const struct sdma_firmware_header_v3_0 *sdma_hdr =
+			container_of(hdr, struct sdma_firmware_header_v3_0, header);
+
+		DRM_DEBUG("ucode_reversion: %u\n",
+			  le32_to_cpu(sdma_hdr->ucode_feature_version));
 	} else {
 		DRM_ERROR("Unknown SDMA ucode version: %u.%u\n",
 			  version_major, version_minor);
 	}
 }
 
+void amdgpu_ucode_print_psp_hdr(const struct common_firmware_header *hdr)
+{
+	uint16_t version_major = le16_to_cpu(hdr->header_version_major);
+	uint16_t version_minor = le16_to_cpu(hdr->header_version_minor);
+	uint32_t fw_index;
+	const struct psp_fw_bin_desc *desc;
+
+	DRM_DEBUG("PSP\n");
+	amdgpu_ucode_print_common_hdr(hdr);
+
+	if (version_major == 1) {
+		const struct psp_firmware_header_v1_0 *psp_hdr =
+			container_of(hdr, struct psp_firmware_header_v1_0, header);
+
+		DRM_DEBUG("ucode_feature_version: %u\n",
+			  le32_to_cpu(psp_hdr->sos.fw_version));
+		DRM_DEBUG("sos_offset_bytes: %u\n",
+			  le32_to_cpu(psp_hdr->sos.offset_bytes));
+		DRM_DEBUG("sos_size_bytes: %u\n",
+			  le32_to_cpu(psp_hdr->sos.size_bytes));
+		if (version_minor == 1) {
+			const struct psp_firmware_header_v1_1 *psp_hdr_v1_1 =
+				container_of(psp_hdr, struct psp_firmware_header_v1_1, v1_0);
+			DRM_DEBUG("toc_header_version: %u\n",
+				  le32_to_cpu(psp_hdr_v1_1->toc.fw_version));
+			DRM_DEBUG("toc_offset_bytes: %u\n",
+				  le32_to_cpu(psp_hdr_v1_1->toc.offset_bytes));
+			DRM_DEBUG("toc_size_bytes: %u\n",
+				  le32_to_cpu(psp_hdr_v1_1->toc.size_bytes));
+			DRM_DEBUG("kdb_header_version: %u\n",
+				  le32_to_cpu(psp_hdr_v1_1->kdb.fw_version));
+			DRM_DEBUG("kdb_offset_bytes: %u\n",
+				  le32_to_cpu(psp_hdr_v1_1->kdb.offset_bytes));
+			DRM_DEBUG("kdb_size_bytes: %u\n",
+				  le32_to_cpu(psp_hdr_v1_1->kdb.size_bytes));
+		}
+		if (version_minor == 2) {
+			const struct psp_firmware_header_v1_2 *psp_hdr_v1_2 =
+				container_of(psp_hdr, struct psp_firmware_header_v1_2, v1_0);
+			DRM_DEBUG("kdb_header_version: %u\n",
+				  le32_to_cpu(psp_hdr_v1_2->kdb.fw_version));
+			DRM_DEBUG("kdb_offset_bytes: %u\n",
+				  le32_to_cpu(psp_hdr_v1_2->kdb.offset_bytes));
+			DRM_DEBUG("kdb_size_bytes: %u\n",
+				  le32_to_cpu(psp_hdr_v1_2->kdb.size_bytes));
+		}
+		if (version_minor == 3) {
+			const struct psp_firmware_header_v1_1 *psp_hdr_v1_1 =
+				container_of(psp_hdr, struct psp_firmware_header_v1_1, v1_0);
+			const struct psp_firmware_header_v1_3 *psp_hdr_v1_3 =
+				container_of(psp_hdr_v1_1, struct psp_firmware_header_v1_3, v1_1);
+			DRM_DEBUG("toc_header_version: %u\n",
+				  le32_to_cpu(psp_hdr_v1_3->v1_1.toc.fw_version));
+			DRM_DEBUG("toc_offset_bytes: %u\n",
+				  le32_to_cpu(psp_hdr_v1_3->v1_1.toc.offset_bytes));
+			DRM_DEBUG("toc_size_bytes: %u\n",
+				  le32_to_cpu(psp_hdr_v1_3->v1_1.toc.size_bytes));
+			DRM_DEBUG("kdb_header_version: %u\n",
+				  le32_to_cpu(psp_hdr_v1_3->v1_1.kdb.fw_version));
+			DRM_DEBUG("kdb_offset_bytes: %u\n",
+				  le32_to_cpu(psp_hdr_v1_3->v1_1.kdb.offset_bytes));
+			DRM_DEBUG("kdb_size_bytes: %u\n",
+				  le32_to_cpu(psp_hdr_v1_3->v1_1.kdb.size_bytes));
+			DRM_DEBUG("spl_header_version: %u\n",
+				  le32_to_cpu(psp_hdr_v1_3->spl.fw_version));
+			DRM_DEBUG("spl_offset_bytes: %u\n",
+				  le32_to_cpu(psp_hdr_v1_3->spl.offset_bytes));
+			DRM_DEBUG("spl_size_bytes: %u\n",
+				  le32_to_cpu(psp_hdr_v1_3->spl.size_bytes));
+		}
+	} else if (version_major == 2) {
+		const struct psp_firmware_header_v2_0 *psp_hdr_v2_0 =
+			 container_of(hdr, struct psp_firmware_header_v2_0, header);
+		for (fw_index = 0; fw_index < le32_to_cpu(psp_hdr_v2_0->psp_fw_bin_count); fw_index++) {
+			desc = &(psp_hdr_v2_0->psp_fw_bin[fw_index]);
+			switch (desc->fw_type) {
+			case PSP_FW_TYPE_PSP_SOS:
+				DRM_DEBUG("psp_sos_version: %u\n",
+					  le32_to_cpu(desc->fw_version));
+				DRM_DEBUG("psp_sos_size_bytes: %u\n",
+					  le32_to_cpu(desc->size_bytes));
+				break;
+			case PSP_FW_TYPE_PSP_SYS_DRV:
+				DRM_DEBUG("psp_sys_drv_version: %u\n",
+					  le32_to_cpu(desc->fw_version));
+				DRM_DEBUG("psp_sys_drv_size_bytes: %u\n",
+					  le32_to_cpu(desc->size_bytes));
+				break;
+			case PSP_FW_TYPE_PSP_KDB:
+				DRM_DEBUG("psp_kdb_version: %u\n",
+					  le32_to_cpu(desc->fw_version));
+				DRM_DEBUG("psp_kdb_size_bytes: %u\n",
+					  le32_to_cpu(desc->size_bytes));
+				break;
+			case PSP_FW_TYPE_PSP_TOC:
+				DRM_DEBUG("psp_toc_version: %u\n",
+					  le32_to_cpu(desc->fw_version));
+				DRM_DEBUG("psp_toc_size_bytes: %u\n",
+					  le32_to_cpu(desc->size_bytes));
+				break;
+			case PSP_FW_TYPE_PSP_SPL:
+				DRM_DEBUG("psp_spl_version: %u\n",
+					  le32_to_cpu(desc->fw_version));
+				DRM_DEBUG("psp_spl_size_bytes: %u\n",
+					  le32_to_cpu(desc->size_bytes));
+				break;
+			case PSP_FW_TYPE_PSP_RL:
+				DRM_DEBUG("psp_rl_version: %u\n",
+					  le32_to_cpu(desc->fw_version));
+				DRM_DEBUG("psp_rl_size_bytes: %u\n",
+					  le32_to_cpu(desc->size_bytes));
+				break;
+			case PSP_FW_TYPE_PSP_SOC_DRV:
+				DRM_DEBUG("psp_soc_drv_version: %u\n",
+					  le32_to_cpu(desc->fw_version));
+				DRM_DEBUG("psp_soc_drv_size_bytes: %u\n",
+					  le32_to_cpu(desc->size_bytes));
+				break;
+			case PSP_FW_TYPE_PSP_INTF_DRV:
+				DRM_DEBUG("psp_intf_drv_version: %u\n",
+					  le32_to_cpu(desc->fw_version));
+				DRM_DEBUG("psp_intf_drv_size_bytes: %u\n",
+					  le32_to_cpu(desc->size_bytes));
+				break;
+			case PSP_FW_TYPE_PSP_DBG_DRV:
+				DRM_DEBUG("psp_dbg_drv_version: %u\n",
+					  le32_to_cpu(desc->fw_version));
+				DRM_DEBUG("psp_dbg_drv_size_bytes: %u\n",
+					  le32_to_cpu(desc->size_bytes));
+				break;
+			case PSP_FW_TYPE_PSP_RAS_DRV:
+				DRM_DEBUG("psp_ras_drv_version: %u\n",
+					  le32_to_cpu(desc->fw_version));
+				DRM_DEBUG("psp_ras_drv_size_bytes: %u\n",
+					  le32_to_cpu(desc->size_bytes));
+				break;
+			default:
+				DRM_DEBUG("Unsupported PSP fw type: %d\n", desc->fw_type);
+				break;
+			}
+		}
+	} else {
+		DRM_ERROR("Unknown PSP ucode version: %u.%u\n",
+			  version_major, version_minor);
+	}
+}
+
 void amdgpu_ucode_print_gpu_info_hdr(const struct common_firmware_header *hdr)
 {
 	uint16_t version_major = le16_to_cpu(hdr->header_version_major);
@@ -218,7 +510,7 @@ void amdgpu_ucode_print_gpu_info_hdr(const struct common_firmware_header *hdr)
 	}
 }
 
-int amdgpu_ucode_validate(const struct firmware *fw)
+static int amdgpu_ucode_validate(const struct firmware *fw)
 {
 	const struct common_firmware_header *hdr =
 		(const struct common_firmware_header *)fw->data;
@@ -234,8 +526,8 @@ bool amdgpu_ucode_hdr_version(union amdgpu_firmware_header *hdr,
 {
 	if ((hdr->common.header_version_major == hdr_major) &&
 		(hdr->common.header_version_minor == hdr_minor))
-		return false;
-	return true;
+		return true;
+	return false;
 }
 
 enum amdgpu_firmware_load_type
@@ -247,6 +539,7 @@ amdgpu_ucode_get_load_type(struct amdgpu_device *adev, int load_type)
 	case CHIP_PITCAIRN:
 	case CHIP_VERDE:
 	case CHIP_OLAND:
+	case CHIP_HAINAN:
 		return AMDGPU_FW_LOAD_DIRECT;
 #endif
 #ifdef CONFIG_DRM_AMDGPU_CIK
@@ -265,25 +558,268 @@ amdgpu_ucode_get_load_type(struct amdgpu_device *adev, int load_type)
 	case CHIP_POLARIS10:
 	case CHIP_POLARIS11:
 	case CHIP_POLARIS12:
-		if (!load_type)
-			return AMDGPU_FW_LOAD_DIRECT;
-		else
-			return AMDGPU_FW_LOAD_SMU;
-	case CHIP_VEGA10:
-		if (!load_type)
+	case CHIP_VEGAM:
+		return AMDGPU_FW_LOAD_SMU;
+	case CHIP_CYAN_SKILLFISH:
+		if (!(load_type &&
+		      adev->apu_flags & AMD_APU_IS_CYAN_SKILLFISH2))
 			return AMDGPU_FW_LOAD_DIRECT;
 		else
 			return AMDGPU_FW_LOAD_PSP;
-	case CHIP_RAVEN:
-		if (load_type != 2)
+	default:
+		if (!load_type)
 			return AMDGPU_FW_LOAD_DIRECT;
+		else if (load_type == 3)
+			return AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO;
 		else
 			return AMDGPU_FW_LOAD_PSP;
+	}
+}
+
+const char *amdgpu_ucode_name(enum AMDGPU_UCODE_ID ucode_id)
+{
+	switch (ucode_id) {
+	case AMDGPU_UCODE_ID_SDMA0:
+		return "SDMA0";
+	case AMDGPU_UCODE_ID_SDMA1:
+		return "SDMA1";
+	case AMDGPU_UCODE_ID_SDMA2:
+		return "SDMA2";
+	case AMDGPU_UCODE_ID_SDMA3:
+		return "SDMA3";
+	case AMDGPU_UCODE_ID_SDMA4:
+		return "SDMA4";
+	case AMDGPU_UCODE_ID_SDMA5:
+		return "SDMA5";
+	case AMDGPU_UCODE_ID_SDMA6:
+		return "SDMA6";
+	case AMDGPU_UCODE_ID_SDMA7:
+		return "SDMA7";
+	case AMDGPU_UCODE_ID_SDMA_UCODE_TH0:
+		return "SDMA_CTX";
+	case AMDGPU_UCODE_ID_SDMA_UCODE_TH1:
+		return "SDMA_CTL";
+	case AMDGPU_UCODE_ID_CP_CE:
+		return "CP_CE";
+	case AMDGPU_UCODE_ID_CP_PFP:
+		return "CP_PFP";
+	case AMDGPU_UCODE_ID_CP_ME:
+		return "CP_ME";
+	case AMDGPU_UCODE_ID_CP_MEC1:
+		return "CP_MEC1";
+	case AMDGPU_UCODE_ID_CP_MEC1_JT:
+		return "CP_MEC1_JT";
+	case AMDGPU_UCODE_ID_CP_MEC2:
+		return "CP_MEC2";
+	case AMDGPU_UCODE_ID_CP_MEC2_JT:
+		return "CP_MEC2_JT";
+	case AMDGPU_UCODE_ID_CP_MES:
+		return "CP_MES";
+	case AMDGPU_UCODE_ID_CP_MES_DATA:
+		return "CP_MES_DATA";
+	case AMDGPU_UCODE_ID_CP_MES1:
+		return "CP_MES_KIQ";
+	case AMDGPU_UCODE_ID_CP_MES1_DATA:
+		return "CP_MES_KIQ_DATA";
+	case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL:
+		return "RLC_RESTORE_LIST_CNTL";
+	case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM:
+		return "RLC_RESTORE_LIST_GPM_MEM";
+	case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM:
+		return "RLC_RESTORE_LIST_SRM_MEM";
+	case AMDGPU_UCODE_ID_RLC_IRAM:
+		return "RLC_IRAM";
+	case AMDGPU_UCODE_ID_RLC_DRAM:
+		return "RLC_DRAM";
+	case AMDGPU_UCODE_ID_RLC_G:
+		return "RLC_G";
+	case AMDGPU_UCODE_ID_RLC_P:
+		return "RLC_P";
+	case AMDGPU_UCODE_ID_RLC_V:
+		return "RLC_V";
+	case AMDGPU_UCODE_ID_GLOBAL_TAP_DELAYS:
+		return "GLOBAL_TAP_DELAYS";
+	case AMDGPU_UCODE_ID_SE0_TAP_DELAYS:
+		return "SE0_TAP_DELAYS";
+	case AMDGPU_UCODE_ID_SE1_TAP_DELAYS:
+		return "SE1_TAP_DELAYS";
+	case AMDGPU_UCODE_ID_SE2_TAP_DELAYS:
+		return "SE2_TAP_DELAYS";
+	case AMDGPU_UCODE_ID_SE3_TAP_DELAYS:
+		return "SE3_TAP_DELAYS";
+	case AMDGPU_UCODE_ID_IMU_I:
+		return "IMU_I";
+	case AMDGPU_UCODE_ID_IMU_D:
+		return "IMU_D";
+	case AMDGPU_UCODE_ID_STORAGE:
+		return "STORAGE";
+	case AMDGPU_UCODE_ID_SMC:
+		return "SMC";
+	case AMDGPU_UCODE_ID_PPTABLE:
+		return "PPTABLE";
+	case AMDGPU_UCODE_ID_P2S_TABLE:
+		return "P2STABLE";
+	case AMDGPU_UCODE_ID_UVD:
+		return "UVD";
+	case AMDGPU_UCODE_ID_UVD1:
+		return "UVD1";
+	case AMDGPU_UCODE_ID_VCE:
+		return "VCE";
+	case AMDGPU_UCODE_ID_VCN:
+		return "VCN";
+	case AMDGPU_UCODE_ID_VCN1:
+		return "VCN1";
+	case AMDGPU_UCODE_ID_DMCU_ERAM:
+		return "DMCU_ERAM";
+	case AMDGPU_UCODE_ID_DMCU_INTV:
+		return "DMCU_INTV";
+	case AMDGPU_UCODE_ID_VCN0_RAM:
+		return "VCN0_RAM";
+	case AMDGPU_UCODE_ID_VCN1_RAM:
+		return "VCN1_RAM";
+	case AMDGPU_UCODE_ID_DMCUB:
+		return "DMCUB";
+	case AMDGPU_UCODE_ID_CAP:
+		return "CAP";
+	case AMDGPU_UCODE_ID_VPE_CTX:
+		return "VPE_CTX";
+	case AMDGPU_UCODE_ID_VPE_CTL:
+		return "VPE_CTL";
+	case AMDGPU_UCODE_ID_VPE:
+		return "VPE";
+	case AMDGPU_UCODE_ID_UMSCH_MM_UCODE:
+		return "UMSCH_MM_UCODE";
+	case AMDGPU_UCODE_ID_UMSCH_MM_DATA:
+		return "UMSCH_MM_DATA";
+	case AMDGPU_UCODE_ID_UMSCH_MM_CMD_BUFFER:
+		return "UMSCH_MM_CMD_BUFFER";
+	case AMDGPU_UCODE_ID_JPEG_RAM:
+		return "JPEG";
+	case AMDGPU_UCODE_ID_SDMA_RS64:
+		return "RS64_SDMA";
+	case AMDGPU_UCODE_ID_CP_RS64_PFP:
+		return "RS64_PFP";
+	case AMDGPU_UCODE_ID_CP_RS64_ME:
+		return "RS64_ME";
+	case AMDGPU_UCODE_ID_CP_RS64_MEC:
+		return "RS64_MEC";
+	case AMDGPU_UCODE_ID_CP_RS64_PFP_P0_STACK:
+		return "RS64_PFP_P0_STACK";
+	case AMDGPU_UCODE_ID_CP_RS64_PFP_P1_STACK:
+		return "RS64_PFP_P1_STACK";
+	case AMDGPU_UCODE_ID_CP_RS64_ME_P0_STACK:
+		return "RS64_ME_P0_STACK";
+	case AMDGPU_UCODE_ID_CP_RS64_ME_P1_STACK:
+		return "RS64_ME_P1_STACK";
+	case AMDGPU_UCODE_ID_CP_RS64_MEC_P0_STACK:
+		return "RS64_MEC_P0_STACK";
+	case AMDGPU_UCODE_ID_CP_RS64_MEC_P1_STACK:
+		return "RS64_MEC_P1_STACK";
+	case AMDGPU_UCODE_ID_CP_RS64_MEC_P2_STACK:
+		return "RS64_MEC_P2_STACK";
+	case AMDGPU_UCODE_ID_CP_RS64_MEC_P3_STACK:
+		return "RS64_MEC_P3_STACK";
+	case AMDGPU_UCODE_ID_ISP:
+		return "ISP";
 	default:
-		DRM_ERROR("Unknow firmware load type\n");
+		return "UNKNOWN UCODE";
 	}
+}
+
+static inline int amdgpu_ucode_is_valid(uint32_t fw_version)
+{
+	if (!fw_version)
+		return -EINVAL;
+
+	return 0;
+}
+
+#define FW_VERSION_ATTR(name, mode, field)				\
+static ssize_t show_##name(struct device *dev,				\
+			   struct device_attribute *attr, char *buf)	\
+{									\
+	struct drm_device *ddev = dev_get_drvdata(dev);			\
+	struct amdgpu_device *adev = drm_to_adev(ddev);			\
+									\
+	if (!buf)							\
+		return amdgpu_ucode_is_valid(adev->field);		\
+									\
+	return sysfs_emit(buf, "0x%08x\n", adev->field);		\
+}									\
+static DEVICE_ATTR(name, mode, show_##name, NULL)
+
+FW_VERSION_ATTR(vce_fw_version, 0444, vce.fw_version);
+FW_VERSION_ATTR(uvd_fw_version, 0444, uvd.fw_version);
+FW_VERSION_ATTR(mc_fw_version, 0444, gmc.fw_version);
+FW_VERSION_ATTR(me_fw_version, 0444, gfx.me_fw_version);
+FW_VERSION_ATTR(pfp_fw_version, 0444, gfx.pfp_fw_version);
+FW_VERSION_ATTR(ce_fw_version, 0444, gfx.ce_fw_version);
+FW_VERSION_ATTR(rlc_fw_version, 0444, gfx.rlc_fw_version);
+FW_VERSION_ATTR(rlc_srlc_fw_version, 0444, gfx.rlc_srlc_fw_version);
+FW_VERSION_ATTR(rlc_srlg_fw_version, 0444, gfx.rlc_srlg_fw_version);
+FW_VERSION_ATTR(rlc_srls_fw_version, 0444, gfx.rlc_srls_fw_version);
+FW_VERSION_ATTR(mec_fw_version, 0444, gfx.mec_fw_version);
+FW_VERSION_ATTR(mec2_fw_version, 0444, gfx.mec2_fw_version);
+FW_VERSION_ATTR(imu_fw_version, 0444, gfx.imu_fw_version);
+FW_VERSION_ATTR(sos_fw_version, 0444, psp.sos.fw_version);
+FW_VERSION_ATTR(asd_fw_version, 0444, psp.asd_context.bin_desc.fw_version);
+FW_VERSION_ATTR(ta_ras_fw_version, 0444, psp.ras_context.context.bin_desc.fw_version);
+FW_VERSION_ATTR(ta_xgmi_fw_version, 0444, psp.xgmi_context.context.bin_desc.fw_version);
+FW_VERSION_ATTR(smc_fw_version, 0444, pm.fw_version);
+FW_VERSION_ATTR(sdma_fw_version, 0444, sdma.instance[0].fw_version);
+FW_VERSION_ATTR(sdma2_fw_version, 0444, sdma.instance[1].fw_version);
+FW_VERSION_ATTR(vcn_fw_version, 0444, vcn.fw_version);
+FW_VERSION_ATTR(dmcu_fw_version, 0444, dm.dmcu_fw_version);
+FW_VERSION_ATTR(dmcub_fw_version, 0444, dm.dmcub_fw_version);
+FW_VERSION_ATTR(mes_fw_version, 0444, mes.sched_version & AMDGPU_MES_VERSION_MASK);
+FW_VERSION_ATTR(mes_kiq_fw_version, 0444, mes.kiq_version & AMDGPU_MES_VERSION_MASK);
+FW_VERSION_ATTR(pldm_fw_version, 0444, firmware.pldm_version);
+
+static struct attribute *fw_attrs[] = {
+	&dev_attr_vce_fw_version.attr, &dev_attr_uvd_fw_version.attr,
+	&dev_attr_mc_fw_version.attr, &dev_attr_me_fw_version.attr,
+	&dev_attr_pfp_fw_version.attr, &dev_attr_ce_fw_version.attr,
+	&dev_attr_rlc_fw_version.attr, &dev_attr_rlc_srlc_fw_version.attr,
+	&dev_attr_rlc_srlg_fw_version.attr, &dev_attr_rlc_srls_fw_version.attr,
+	&dev_attr_mec_fw_version.attr, &dev_attr_mec2_fw_version.attr,
+	&dev_attr_sos_fw_version.attr, &dev_attr_asd_fw_version.attr,
+	&dev_attr_ta_ras_fw_version.attr, &dev_attr_ta_xgmi_fw_version.attr,
+	&dev_attr_smc_fw_version.attr, &dev_attr_sdma_fw_version.attr,
+	&dev_attr_sdma2_fw_version.attr, &dev_attr_vcn_fw_version.attr,
+	&dev_attr_dmcu_fw_version.attr, &dev_attr_dmcub_fw_version.attr,
+	&dev_attr_imu_fw_version.attr, &dev_attr_mes_fw_version.attr,
+	&dev_attr_mes_kiq_fw_version.attr, &dev_attr_pldm_fw_version.attr,
+	NULL
+};
+
+#define to_dev_attr(x) container_of(x, struct device_attribute, attr)
+
+static umode_t amdgpu_ucode_sys_visible(struct kobject *kobj,
+					struct attribute *attr, int idx)
+{
+	struct device_attribute *dev_attr = to_dev_attr(attr);
+	struct device *dev = kobj_to_dev(kobj);
+
+	if (dev_attr->show(dev, dev_attr, NULL) == -EINVAL)
+		return 0;
+
+	return attr->mode;
+}
 
-	return AMDGPU_FW_LOAD_DIRECT;
+static const struct attribute_group fw_attr_group = {
+	.name = "fw_version",
+	.attrs = fw_attrs,
+	.is_visible = amdgpu_ucode_sys_visible
+};
+
+int amdgpu_ucode_sysfs_init(struct amdgpu_device *adev)
+{
+	return sysfs_create_group(&adev->dev->kobj, &fw_attr_group);
+}
+
+void amdgpu_ucode_sysfs_fini(struct amdgpu_device *adev)
+{
+	sysfs_remove_group(&adev->dev->kobj, &fw_attr_group);
 }
 
 static int amdgpu_ucode_init_single_fw(struct amdgpu_device *adev,
@@ -292,8 +828,18 @@ static int amdgpu_ucode_init_single_fw(struct amdgpu_device *adev,
 {
 	const struct common_firmware_header *header = NULL;
 	const struct gfx_firmware_header_v1_0 *cp_hdr = NULL;
-
-	if (NULL == ucode->fw)
+	const struct gfx_firmware_header_v2_0 *cpv2_hdr = NULL;
+	const struct dmcu_firmware_header_v1_0 *dmcu_hdr = NULL;
+	const struct dmcub_firmware_header_v1_0 *dmcub_hdr = NULL;
+	const struct mes_firmware_header_v1_0 *mes_hdr = NULL;
+	const struct sdma_firmware_header_v2_0 *sdma_hdr = NULL;
+	const struct sdma_firmware_header_v3_0 *sdmav3_hdr = NULL;
+	const struct imu_firmware_header_v1_0 *imu_hdr = NULL;
+	const struct vpe_firmware_header_v1_0 *vpe_hdr = NULL;
+	const struct umsch_mm_firmware_header_v1_0 *umsch_mm_hdr = NULL;
+	u8 *ucode_addr;
+
+	if (!ucode->fw)
 		return 0;
 
 	ucode->mc_addr = mc_addr;
@@ -303,37 +849,241 @@ static int amdgpu_ucode_init_single_fw(struct amdgpu_device *adev,
 		return 0;
 
 	header = (const struct common_firmware_header *)ucode->fw->data;
-
 	cp_hdr = (const struct gfx_firmware_header_v1_0 *)ucode->fw->data;
-
-	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP ||
-	    (ucode->ucode_id != AMDGPU_UCODE_ID_CP_MEC1 &&
-	     ucode->ucode_id != AMDGPU_UCODE_ID_CP_MEC2 &&
-	     ucode->ucode_id != AMDGPU_UCODE_ID_CP_MEC1_JT &&
-	     ucode->ucode_id != AMDGPU_UCODE_ID_CP_MEC2_JT)) {
+	cpv2_hdr = (const struct gfx_firmware_header_v2_0 *)ucode->fw->data;
+	dmcu_hdr = (const struct dmcu_firmware_header_v1_0 *)ucode->fw->data;
+	dmcub_hdr = (const struct dmcub_firmware_header_v1_0 *)ucode->fw->data;
+	mes_hdr = (const struct mes_firmware_header_v1_0 *)ucode->fw->data;
+	sdma_hdr = (const struct sdma_firmware_header_v2_0 *)ucode->fw->data;
+	sdmav3_hdr = (const struct sdma_firmware_header_v3_0 *)ucode->fw->data;
+	imu_hdr = (const struct imu_firmware_header_v1_0 *)ucode->fw->data;
+	vpe_hdr = (const struct vpe_firmware_header_v1_0 *)ucode->fw->data;
+	umsch_mm_hdr = (const struct umsch_mm_firmware_header_v1_0 *)ucode->fw->data;
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		switch (ucode->ucode_id) {
+		case AMDGPU_UCODE_ID_SDMA_UCODE_TH0:
+			ucode->ucode_size = le32_to_cpu(sdma_hdr->ctx_ucode_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(sdma_hdr->header.ucode_array_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_SDMA_UCODE_TH1:
+			ucode->ucode_size = le32_to_cpu(sdma_hdr->ctl_ucode_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(sdma_hdr->ctl_ucode_offset);
+			break;
+		case AMDGPU_UCODE_ID_SDMA_RS64:
+			ucode->ucode_size = le32_to_cpu(sdmav3_hdr->ucode_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(sdmav3_hdr->header.ucode_array_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_CP_MEC1:
+		case AMDGPU_UCODE_ID_CP_MEC2:
+			ucode->ucode_size = le32_to_cpu(header->ucode_size_bytes) -
+				le32_to_cpu(cp_hdr->jt_size) * 4;
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(header->ucode_array_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_CP_MEC1_JT:
+		case AMDGPU_UCODE_ID_CP_MEC2_JT:
+			ucode->ucode_size = le32_to_cpu(cp_hdr->jt_size) * 4;
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(header->ucode_array_offset_bytes) +
+				le32_to_cpu(cp_hdr->jt_offset) * 4;
+			break;
+		case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL:
+			ucode->ucode_size = adev->gfx.rlc.save_restore_list_cntl_size_bytes;
+			ucode_addr = adev->gfx.rlc.save_restore_list_cntl;
+			break;
+		case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM:
+			ucode->ucode_size = adev->gfx.rlc.save_restore_list_gpm_size_bytes;
+			ucode_addr = adev->gfx.rlc.save_restore_list_gpm;
+			break;
+		case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM:
+			ucode->ucode_size = adev->gfx.rlc.save_restore_list_srm_size_bytes;
+			ucode_addr = adev->gfx.rlc.save_restore_list_srm;
+			break;
+		case AMDGPU_UCODE_ID_RLC_IRAM:
+			ucode->ucode_size = adev->gfx.rlc.rlc_iram_ucode_size_bytes;
+			ucode_addr = adev->gfx.rlc.rlc_iram_ucode;
+			break;
+		case AMDGPU_UCODE_ID_RLC_DRAM:
+			ucode->ucode_size = adev->gfx.rlc.rlc_dram_ucode_size_bytes;
+			ucode_addr = adev->gfx.rlc.rlc_dram_ucode;
+			break;
+		case AMDGPU_UCODE_ID_RLC_P:
+			ucode->ucode_size = adev->gfx.rlc.rlcp_ucode_size_bytes;
+			ucode_addr = adev->gfx.rlc.rlcp_ucode;
+			break;
+		case AMDGPU_UCODE_ID_RLC_V:
+			ucode->ucode_size = adev->gfx.rlc.rlcv_ucode_size_bytes;
+			ucode_addr = adev->gfx.rlc.rlcv_ucode;
+			break;
+		case AMDGPU_UCODE_ID_GLOBAL_TAP_DELAYS:
+			ucode->ucode_size = adev->gfx.rlc.global_tap_delays_ucode_size_bytes;
+			ucode_addr = adev->gfx.rlc.global_tap_delays_ucode;
+			break;
+		case AMDGPU_UCODE_ID_SE0_TAP_DELAYS:
+			ucode->ucode_size = adev->gfx.rlc.se0_tap_delays_ucode_size_bytes;
+			ucode_addr = adev->gfx.rlc.se0_tap_delays_ucode;
+			break;
+		case AMDGPU_UCODE_ID_SE1_TAP_DELAYS:
+			ucode->ucode_size = adev->gfx.rlc.se1_tap_delays_ucode_size_bytes;
+			ucode_addr = adev->gfx.rlc.se1_tap_delays_ucode;
+			break;
+		case AMDGPU_UCODE_ID_SE2_TAP_DELAYS:
+			ucode->ucode_size = adev->gfx.rlc.se2_tap_delays_ucode_size_bytes;
+			ucode_addr = adev->gfx.rlc.se2_tap_delays_ucode;
+			break;
+		case AMDGPU_UCODE_ID_SE3_TAP_DELAYS:
+			ucode->ucode_size = adev->gfx.rlc.se3_tap_delays_ucode_size_bytes;
+			ucode_addr = adev->gfx.rlc.se3_tap_delays_ucode;
+			break;
+		case AMDGPU_UCODE_ID_CP_MES:
+			ucode->ucode_size = le32_to_cpu(mes_hdr->mes_ucode_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(mes_hdr->mes_ucode_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_CP_MES_DATA:
+			ucode->ucode_size = le32_to_cpu(mes_hdr->mes_ucode_data_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(mes_hdr->mes_ucode_data_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_CP_MES1:
+			ucode->ucode_size = le32_to_cpu(mes_hdr->mes_ucode_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(mes_hdr->mes_ucode_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_CP_MES1_DATA:
+			ucode->ucode_size = le32_to_cpu(mes_hdr->mes_ucode_data_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(mes_hdr->mes_ucode_data_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_DMCU_ERAM:
+			ucode->ucode_size = le32_to_cpu(header->ucode_size_bytes) -
+				le32_to_cpu(dmcu_hdr->intv_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(header->ucode_array_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_DMCU_INTV:
+			ucode->ucode_size = le32_to_cpu(dmcu_hdr->intv_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(header->ucode_array_offset_bytes) +
+				le32_to_cpu(dmcu_hdr->intv_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_DMCUB:
+			ucode->ucode_size = le32_to_cpu(dmcub_hdr->inst_const_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(header->ucode_array_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_PPTABLE:
+			ucode->ucode_size = ucode->fw->size;
+			ucode_addr = (u8 *)ucode->fw->data;
+			break;
+		case AMDGPU_UCODE_ID_P2S_TABLE:
+			ucode->ucode_size = ucode->fw->size;
+			ucode_addr = (u8 *)ucode->fw->data;
+			break;
+		case AMDGPU_UCODE_ID_IMU_I:
+			ucode->ucode_size = le32_to_cpu(imu_hdr->imu_iram_ucode_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(imu_hdr->header.ucode_array_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_IMU_D:
+			ucode->ucode_size = le32_to_cpu(imu_hdr->imu_dram_ucode_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(imu_hdr->header.ucode_array_offset_bytes) +
+				le32_to_cpu(imu_hdr->imu_iram_ucode_size_bytes);
+			break;
+		case AMDGPU_UCODE_ID_CP_RS64_PFP:
+			ucode->ucode_size = le32_to_cpu(cpv2_hdr->ucode_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(header->ucode_array_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_CP_RS64_PFP_P0_STACK:
+			ucode->ucode_size = le32_to_cpu(cpv2_hdr->data_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(cpv2_hdr->data_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_CP_RS64_PFP_P1_STACK:
+			ucode->ucode_size = le32_to_cpu(cpv2_hdr->data_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(cpv2_hdr->data_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_CP_RS64_ME:
+			ucode->ucode_size = le32_to_cpu(cpv2_hdr->ucode_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(header->ucode_array_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_CP_RS64_ME_P0_STACK:
+			ucode->ucode_size = le32_to_cpu(cpv2_hdr->data_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(cpv2_hdr->data_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_CP_RS64_ME_P1_STACK:
+			ucode->ucode_size = le32_to_cpu(cpv2_hdr->data_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(cpv2_hdr->data_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_CP_RS64_MEC:
+			ucode->ucode_size = le32_to_cpu(cpv2_hdr->ucode_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(header->ucode_array_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_CP_RS64_MEC_P0_STACK:
+			ucode->ucode_size = le32_to_cpu(cpv2_hdr->data_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(cpv2_hdr->data_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_CP_RS64_MEC_P1_STACK:
+			ucode->ucode_size = le32_to_cpu(cpv2_hdr->data_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(cpv2_hdr->data_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_CP_RS64_MEC_P2_STACK:
+			ucode->ucode_size = le32_to_cpu(cpv2_hdr->data_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(cpv2_hdr->data_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_CP_RS64_MEC_P3_STACK:
+			ucode->ucode_size = le32_to_cpu(cpv2_hdr->data_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(cpv2_hdr->data_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_VPE_CTX:
+			ucode->ucode_size = le32_to_cpu(vpe_hdr->ctx_ucode_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(vpe_hdr->header.ucode_array_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_VPE_CTL:
+			ucode->ucode_size = le32_to_cpu(vpe_hdr->ctl_ucode_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(vpe_hdr->ctl_ucode_offset);
+			break;
+		case AMDGPU_UCODE_ID_UMSCH_MM_UCODE:
+			ucode->ucode_size = le32_to_cpu(umsch_mm_hdr->umsch_mm_ucode_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(umsch_mm_hdr->header.ucode_array_offset_bytes);
+			break;
+		case AMDGPU_UCODE_ID_UMSCH_MM_DATA:
+			ucode->ucode_size = le32_to_cpu(umsch_mm_hdr->umsch_mm_ucode_data_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(umsch_mm_hdr->umsch_mm_ucode_data_offset_bytes);
+			break;
+		default:
+			ucode->ucode_size = le32_to_cpu(header->ucode_size_bytes);
+			ucode_addr = (u8 *)ucode->fw->data +
+				le32_to_cpu(header->ucode_array_offset_bytes);
+			break;
+		}
+	} else {
 		ucode->ucode_size = le32_to_cpu(header->ucode_size_bytes);
-
-		memcpy(ucode->kaddr, (void *)((uint8_t *)ucode->fw->data +
-					      le32_to_cpu(header->ucode_array_offset_bytes)),
-		       ucode->ucode_size);
-	} else if (ucode->ucode_id == AMDGPU_UCODE_ID_CP_MEC1 ||
-		   ucode->ucode_id == AMDGPU_UCODE_ID_CP_MEC2) {
-		ucode->ucode_size = le32_to_cpu(header->ucode_size_bytes) -
-			le32_to_cpu(cp_hdr->jt_size) * 4;
-
-		memcpy(ucode->kaddr, (void *)((uint8_t *)ucode->fw->data +
-					      le32_to_cpu(header->ucode_array_offset_bytes)),
-		       ucode->ucode_size);
-	} else if (ucode->ucode_id == AMDGPU_UCODE_ID_CP_MEC1_JT ||
-		   ucode->ucode_id == AMDGPU_UCODE_ID_CP_MEC2_JT) {
-		ucode->ucode_size = le32_to_cpu(cp_hdr->jt_size) * 4;
-
-		memcpy(ucode->kaddr, (void *)((uint8_t *)ucode->fw->data +
-					      le32_to_cpu(header->ucode_array_offset_bytes) +
-					      le32_to_cpu(cp_hdr->jt_offset) * 4),
-		       ucode->ucode_size);
+		ucode_addr = (u8 *)ucode->fw->data +
+			le32_to_cpu(header->ucode_array_offset_bytes);
 	}
 
+	memcpy(ucode->kaddr, ucode_addr, ucode->ucode_size);
+
 	return 0;
 }
 
@@ -342,10 +1092,10 @@ static int amdgpu_ucode_patch_jt(struct amdgpu_firmware_info *ucode,
 {
 	const struct gfx_firmware_header_v1_0 *header = NULL;
 	const struct common_firmware_header *comm_hdr = NULL;
-	uint8_t* src_addr = NULL;
-	uint8_t* dst_addr = NULL;
+	uint8_t *src_addr = NULL;
+	uint8_t *dst_addr = NULL;
 
-	if (NULL == ucode->fw)
+	if (!ucode->fw)
 		return 0;
 
 	comm_hdr = (const struct common_firmware_header *)ucode->fw->data;
@@ -361,53 +1111,42 @@ static int amdgpu_ucode_patch_jt(struct amdgpu_firmware_info *ucode,
 	return 0;
 }
 
+int amdgpu_ucode_create_bo(struct amdgpu_device *adev)
+{
+	if ((adev->firmware.load_type != AMDGPU_FW_LOAD_DIRECT) &&
+	    (adev->firmware.load_type != AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO)) {
+		amdgpu_bo_create_kernel(adev, adev->firmware.fw_size, PAGE_SIZE,
+			(amdgpu_sriov_vf(adev) || adev->debug_use_vram_fw_buf) ?
+			AMDGPU_GEM_DOMAIN_VRAM : AMDGPU_GEM_DOMAIN_GTT,
+			&adev->firmware.fw_buf,
+			&adev->firmware.fw_buf_mc,
+			&adev->firmware.fw_buf_ptr);
+		if (!adev->firmware.fw_buf) {
+			dev_err(adev->dev, "failed to create kernel buffer for firmware.fw_buf\n");
+			return -ENOMEM;
+		} else if (amdgpu_sriov_vf(adev)) {
+			memset(adev->firmware.fw_buf_ptr, 0, adev->firmware.fw_size);
+		}
+	}
+	return 0;
+}
+
+void amdgpu_ucode_free_bo(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->firmware.fw_buf,
+		&adev->firmware.fw_buf_mc,
+		&adev->firmware.fw_buf_ptr);
+}
+
 int amdgpu_ucode_init_bo(struct amdgpu_device *adev)
 {
-	struct amdgpu_bo **bo = &adev->firmware.fw_buf;
-	uint64_t fw_mc_addr;
-	void *fw_buf_ptr = NULL;
 	uint64_t fw_offset = 0;
-	int i, err;
+	int i;
 	struct amdgpu_firmware_info *ucode = NULL;
-	const struct common_firmware_header *header = NULL;
 
-	if (!adev->firmware.fw_size) {
-		dev_warn(adev->dev, "No ip firmware need to load\n");
+ /* for baremetal, the ucode is allocated in gtt, so don't need to fill the bo when reset/suspend */
+	if (!amdgpu_sriov_vf(adev) && (amdgpu_in_reset(adev) || adev->in_suspend))
 		return 0;
-	}
-
-	err = amdgpu_bo_create(adev, adev->firmware.fw_size, PAGE_SIZE, true,
-				amdgpu_sriov_vf(adev) ? AMDGPU_GEM_DOMAIN_VRAM : AMDGPU_GEM_DOMAIN_GTT,
-				AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
-				NULL, NULL, 0, bo);
-	if (err) {
-		dev_err(adev->dev, "(%d) Firmware buffer allocate failed\n", err);
-		goto failed;
-	}
-
-	err = amdgpu_bo_reserve(*bo, false);
-	if (err) {
-		dev_err(adev->dev, "(%d) Firmware buffer reserve failed\n", err);
-		goto failed_reserve;
-	}
-
-	err = amdgpu_bo_pin(*bo, amdgpu_sriov_vf(adev) ? AMDGPU_GEM_DOMAIN_VRAM : AMDGPU_GEM_DOMAIN_GTT,
-				&fw_mc_addr);
-	if (err) {
-		dev_err(adev->dev, "(%d) Firmware buffer pin failed\n", err);
-		goto failed_pin;
-	}
-
-	err = amdgpu_bo_kmap(*bo, &fw_buf_ptr);
-	if (err) {
-		dev_err(adev->dev, "(%d) Firmware buffer kmap failed\n", err);
-		goto failed_kmap;
-	}
-
-	amdgpu_bo_unreserve(*bo);
-
-	memset(fw_buf_ptr, 0, adev->firmware.fw_size);
-
 	/*
 	 * if SMU loaded firmware, it needn't add SMC, UVD, and VCE
 	 * ucode info here
@@ -421,55 +1160,357 @@ int amdgpu_ucode_init_bo(struct amdgpu_device *adev)
 		adev->firmware.max_ucodes = AMDGPU_UCODE_ID_MAXIMUM;
 	}
 
+	if (amdgpu_virt_xgmi_migrate_enabled(adev) && adev->firmware.fw_buf)
+		adev->firmware.fw_buf_mc = amdgpu_bo_fb_aper_addr(adev->firmware.fw_buf);
+
 	for (i = 0; i < adev->firmware.max_ucodes; i++) {
 		ucode = &adev->firmware.ucode[i];
 		if (ucode->fw) {
-			header = (const struct common_firmware_header *)ucode->fw->data;
-			amdgpu_ucode_init_single_fw(adev, ucode, fw_mc_addr + fw_offset,
-						    (void *)((uint8_t *)fw_buf_ptr + fw_offset));
+			amdgpu_ucode_init_single_fw(adev, ucode, adev->firmware.fw_buf_mc + fw_offset,
+						    adev->firmware.fw_buf_ptr + fw_offset);
 			if (i == AMDGPU_UCODE_ID_CP_MEC1 &&
 			    adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
 				const struct gfx_firmware_header_v1_0 *cp_hdr;
+
 				cp_hdr = (const struct gfx_firmware_header_v1_0 *)ucode->fw->data;
-				amdgpu_ucode_patch_jt(ucode, fw_mc_addr + fw_offset,
-						    fw_buf_ptr + fw_offset);
+				amdgpu_ucode_patch_jt(ucode,  adev->firmware.fw_buf_mc + fw_offset,
+						    adev->firmware.fw_buf_ptr + fw_offset);
 				fw_offset += ALIGN(le32_to_cpu(cp_hdr->jt_size) << 2, PAGE_SIZE);
 			}
 			fw_offset += ALIGN(ucode->ucode_size, PAGE_SIZE);
 		}
 	}
 	return 0;
+}
 
-failed_kmap:
-	amdgpu_bo_unpin(*bo);
-failed_pin:
-	amdgpu_bo_unreserve(*bo);
-failed_reserve:
-	amdgpu_bo_unref(bo);
-failed:
-	if (err)
-		adev->firmware.load_type = AMDGPU_FW_LOAD_DIRECT;
-
-	return err;
+static const char *amdgpu_ucode_legacy_naming(struct amdgpu_device *adev, int block_type)
+{
+	if (block_type == MP0_HWIP) {
+		switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) {
+		case IP_VERSION(9, 0, 0):
+			switch (adev->asic_type) {
+			case CHIP_VEGA10:
+				return "vega10";
+			case CHIP_VEGA12:
+				return "vega12";
+			default:
+				return NULL;
+			}
+		case IP_VERSION(10, 0, 0):
+		case IP_VERSION(10, 0, 1):
+			if (adev->asic_type == CHIP_RAVEN) {
+				if (adev->apu_flags & AMD_APU_IS_RAVEN2)
+					return "raven2";
+				else if (adev->apu_flags & AMD_APU_IS_PICASSO)
+					return "picasso";
+				return "raven";
+			}
+			break;
+		case IP_VERSION(11, 0, 0):
+			return "navi10";
+		case IP_VERSION(11, 0, 2):
+			return "vega20";
+		case IP_VERSION(11, 0, 3):
+			return "renoir";
+		case IP_VERSION(11, 0, 4):
+			return "arcturus";
+		case IP_VERSION(11, 0, 5):
+			return "navi14";
+		case IP_VERSION(11, 0, 7):
+			return "sienna_cichlid";
+		case IP_VERSION(11, 0, 9):
+			return "navi12";
+		case IP_VERSION(11, 0, 11):
+			return "navy_flounder";
+		case IP_VERSION(11, 0, 12):
+			return "dimgrey_cavefish";
+		case IP_VERSION(11, 0, 13):
+			return "beige_goby";
+		case IP_VERSION(11, 5, 0):
+		case IP_VERSION(11, 5, 2):
+			return "vangogh";
+		case IP_VERSION(12, 0, 1):
+			return "green_sardine";
+		case IP_VERSION(13, 0, 2):
+			return "aldebaran";
+		case IP_VERSION(13, 0, 1):
+		case IP_VERSION(13, 0, 3):
+			return "yellow_carp";
+		}
+	} else if (block_type == MP1_HWIP) {
+		switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) {
+		case IP_VERSION(9, 0, 0):
+		case IP_VERSION(10, 0, 0):
+		case IP_VERSION(10, 0, 1):
+		case IP_VERSION(11, 0, 2):
+			if (adev->asic_type == CHIP_ARCTURUS)
+				return "arcturus_smc";
+			return NULL;
+		case IP_VERSION(11, 0, 0):
+			return "navi10_smc";
+		case IP_VERSION(11, 0, 5):
+			return "navi14_smc";
+		case IP_VERSION(11, 0, 9):
+			return "navi12_smc";
+		case IP_VERSION(11, 0, 7):
+			return "sienna_cichlid_smc";
+		case IP_VERSION(11, 0, 11):
+			return "navy_flounder_smc";
+		case IP_VERSION(11, 0, 12):
+			return "dimgrey_cavefish_smc";
+		case IP_VERSION(11, 0, 13):
+			return "beige_goby_smc";
+		case IP_VERSION(13, 0, 2):
+			return "aldebaran_smc";
+		}
+	} else if (block_type == SDMA0_HWIP) {
+		switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) {
+		case IP_VERSION(4, 0, 0):
+			return "vega10_sdma";
+		case IP_VERSION(4, 0, 1):
+			return "vega12_sdma";
+		case IP_VERSION(4, 1, 0):
+		case IP_VERSION(4, 1, 1):
+			if (adev->apu_flags & AMD_APU_IS_RAVEN2)
+				return "raven2_sdma";
+			else if (adev->apu_flags & AMD_APU_IS_PICASSO)
+				return "picasso_sdma";
+			return "raven_sdma";
+		case IP_VERSION(4, 1, 2):
+			if (adev->apu_flags & AMD_APU_IS_RENOIR)
+				return "renoir_sdma";
+			return "green_sardine_sdma";
+		case IP_VERSION(4, 2, 0):
+			return "vega20_sdma";
+		case IP_VERSION(4, 2, 2):
+			return "arcturus_sdma";
+		case IP_VERSION(4, 4, 0):
+			return "aldebaran_sdma";
+		case IP_VERSION(5, 0, 0):
+			return "navi10_sdma";
+		case IP_VERSION(5, 0, 1):
+			return "cyan_skillfish2_sdma";
+		case IP_VERSION(5, 0, 2):
+			return "navi14_sdma";
+		case IP_VERSION(5, 0, 5):
+			return "navi12_sdma";
+		case IP_VERSION(5, 2, 0):
+			return "sienna_cichlid_sdma";
+		case IP_VERSION(5, 2, 2):
+			return "navy_flounder_sdma";
+		case IP_VERSION(5, 2, 4):
+			return "dimgrey_cavefish_sdma";
+		case IP_VERSION(5, 2, 5):
+			return "beige_goby_sdma";
+		case IP_VERSION(5, 2, 3):
+			return "yellow_carp_sdma";
+		case IP_VERSION(5, 2, 1):
+			return "vangogh_sdma";
+		}
+	} else if (block_type == UVD_HWIP) {
+		switch (amdgpu_ip_version(adev, UVD_HWIP, 0)) {
+		case IP_VERSION(1, 0, 0):
+		case IP_VERSION(1, 0, 1):
+			if (adev->apu_flags & AMD_APU_IS_RAVEN2)
+				return "raven2_vcn";
+			else if (adev->apu_flags & AMD_APU_IS_PICASSO)
+				return "picasso_vcn";
+			return "raven_vcn";
+		case IP_VERSION(2, 5, 0):
+			return "arcturus_vcn";
+		case IP_VERSION(2, 2, 0):
+			if (adev->apu_flags & AMD_APU_IS_RENOIR)
+				return "renoir_vcn";
+			return "green_sardine_vcn";
+		case IP_VERSION(2, 6, 0):
+			return "aldebaran_vcn";
+		case IP_VERSION(2, 0, 0):
+			return "navi10_vcn";
+		case IP_VERSION(2, 0, 2):
+			if (adev->asic_type == CHIP_NAVI12)
+				return "navi12_vcn";
+			return "navi14_vcn";
+		case IP_VERSION(3, 0, 0):
+		case IP_VERSION(3, 0, 64):
+		case IP_VERSION(3, 0, 192):
+			if (amdgpu_ip_version(adev, GC_HWIP, 0) ==
+			    IP_VERSION(10, 3, 0))
+				return "sienna_cichlid_vcn";
+			return "navy_flounder_vcn";
+		case IP_VERSION(3, 0, 2):
+			return "vangogh_vcn";
+		case IP_VERSION(3, 0, 16):
+			return "dimgrey_cavefish_vcn";
+		case IP_VERSION(3, 0, 33):
+			return "beige_goby_vcn";
+		case IP_VERSION(3, 1, 1):
+			return "yellow_carp_vcn";
+		}
+	} else if (block_type == GC_HWIP) {
+		switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+		case IP_VERSION(9, 0, 1):
+			return "vega10";
+		case IP_VERSION(9, 2, 1):
+			return "vega12";
+		case IP_VERSION(9, 4, 0):
+			return "vega20";
+		case IP_VERSION(9, 2, 2):
+		case IP_VERSION(9, 1, 0):
+			if (adev->apu_flags & AMD_APU_IS_RAVEN2)
+				return "raven2";
+			else if (adev->apu_flags & AMD_APU_IS_PICASSO)
+				return "picasso";
+			return "raven";
+		case IP_VERSION(9, 4, 1):
+			return "arcturus";
+		case IP_VERSION(9, 3, 0):
+			if (adev->apu_flags & AMD_APU_IS_RENOIR)
+				return "renoir";
+			return "green_sardine";
+		case IP_VERSION(9, 4, 2):
+			return "aldebaran";
+		case IP_VERSION(10, 1, 10):
+			return "navi10";
+		case IP_VERSION(10, 1, 1):
+			return "navi14";
+		case IP_VERSION(10, 1, 2):
+			return "navi12";
+		case IP_VERSION(10, 3, 0):
+			return "sienna_cichlid";
+		case IP_VERSION(10, 3, 2):
+			return "navy_flounder";
+		case IP_VERSION(10, 3, 1):
+			return "vangogh";
+		case IP_VERSION(10, 3, 4):
+			return "dimgrey_cavefish";
+		case IP_VERSION(10, 3, 5):
+			return "beige_goby";
+		case IP_VERSION(10, 3, 3):
+			return "yellow_carp";
+		case IP_VERSION(10, 1, 3):
+		case IP_VERSION(10, 1, 4):
+			return "cyan_skillfish2";
+		}
+	}
+	return NULL;
 }
 
-int amdgpu_ucode_fini_bo(struct amdgpu_device *adev)
+bool amdgpu_is_kicker_fw(struct amdgpu_device *adev)
 {
 	int i;
-	struct amdgpu_firmware_info *ucode = NULL;
 
-	if (!adev->firmware.fw_size)
-		return 0;
+	for (i = 0; i < ARRAY_SIZE(kicker_device_list); i++) {
+		if (adev->pdev->device == kicker_device_list[i].device &&
+		    adev->pdev->revision == kicker_device_list[i].revision)
+			return true;
+	}
 
-	for (i = 0; i < adev->firmware.max_ucodes; i++) {
-		ucode = &adev->firmware.ucode[i];
-		if (ucode->fw) {
-			ucode->mc_addr = 0;
-			ucode->kaddr = NULL;
-		}
+	return false;
+}
+
+void amdgpu_ucode_ip_version_decode(struct amdgpu_device *adev, int block_type, char *ucode_prefix, int len)
+{
+	int maj, min, rev;
+	char *ip_name;
+	const char *legacy;
+	uint32_t version = amdgpu_ip_version(adev, block_type, 0);
+
+	legacy = amdgpu_ucode_legacy_naming(adev, block_type);
+	if (legacy) {
+		snprintf(ucode_prefix, len, "%s", legacy);
+		return;
 	}
-	amdgpu_bo_unref(&adev->firmware.fw_buf);
-	adev->firmware.fw_buf = NULL;
 
-	return 0;
+	switch (block_type) {
+	case GC_HWIP:
+		ip_name = "gc";
+		break;
+	case SDMA0_HWIP:
+		ip_name = "sdma";
+		break;
+	case MP0_HWIP:
+		ip_name = "psp";
+		break;
+	case MP1_HWIP:
+		ip_name = "smu";
+		break;
+	case UVD_HWIP:
+		ip_name = "vcn";
+		break;
+	case VPE_HWIP:
+		ip_name = "vpe";
+		break;
+	case ISP_HWIP:
+		ip_name = "isp";
+		break;
+	default:
+		BUG();
+	}
+
+	maj = IP_VERSION_MAJ(version);
+	min = IP_VERSION_MIN(version);
+	rev = IP_VERSION_REV(version);
+
+	snprintf(ucode_prefix, len, "%s_%d_%d_%d", ip_name, maj, min, rev);
+}
+
+/*
+ * amdgpu_ucode_request - Fetch and validate amdgpu microcode
+ *
+ * @adev: amdgpu device
+ * @fw: pointer to load firmware to
+ * @required: whether the firmware is required
+ * @fmt: firmware name format string
+ * @...: variable arguments
+ *
+ * This is a helper that will use request_firmware and amdgpu_ucode_validate
+ * to load and run basic validation on firmware. If the load fails, remap
+ * the error code to -ENODEV, so that early_init functions will fail to load.
+ */
+int amdgpu_ucode_request(struct amdgpu_device *adev, const struct firmware **fw,
+			 enum amdgpu_ucode_required required, const char *fmt, ...)
+{
+	char fname[AMDGPU_UCODE_NAME_MAX];
+	va_list ap;
+	int r;
+
+	va_start(ap, fmt);
+	r = vsnprintf(fname, sizeof(fname), fmt, ap);
+	va_end(ap);
+	if (r == sizeof(fname)) {
+		dev_warn(adev->dev, "amdgpu firmware name buffer overflow\n");
+		return -EOVERFLOW;
+	}
+
+	if (required == AMDGPU_UCODE_REQUIRED)
+		r = request_firmware(fw, fname, adev->dev);
+	else {
+		r = firmware_request_nowarn(fw, fname, adev->dev);
+		if (r)
+			drm_info(&adev->ddev, "Optional firmware \"%s\" was not found\n", fname);
+	}
+	if (r)
+		return -ENODEV;
+
+	r = amdgpu_ucode_validate(*fw);
+	if (r)
+		/*
+		 * The amdgpu_ucode_request() should be paired with amdgpu_ucode_release()
+		 * regardless of success/failure, and the amdgpu_ucode_release() takes care of
+		 * firmware release and need to avoid redundant release FW operation here.
+		 */
+		dev_dbg(adev->dev, "\"%s\" failed to validate\n", fname);
+
+	return r;
+}
+
+/*
+ * amdgpu_ucode_release - Release firmware microcode
+ *
+ * @fw: pointer to firmware to release
+ */
+void amdgpu_ucode_release(const struct firmware **fw)
+{
+	release_firmware(*fw);
+	*fw = NULL;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h
index 30b5500dc152..6349aad6da35 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h
@@ -23,6 +23,10 @@
 #ifndef __AMDGPU_UCODE_H__
 #define __AMDGPU_UCODE_H__
 
+#include "amdgpu_socbb.h"
+
+#define RS64_FW_UC_START_ADDR_LO 0x3000
+
 struct common_firmware_header {
 	uint32_t size_bytes; /* size of the entire header+image(s) in bytes */
 	uint32_t header_size_bytes; /* size of just the header in bytes */
@@ -49,12 +53,128 @@ struct smc_firmware_header_v1_0 {
 	uint32_t ucode_start_addr;
 };
 
+/* version_major=2, version_minor=0 */
+struct smc_firmware_header_v2_0 {
+	struct smc_firmware_header_v1_0 v1_0;
+	uint32_t ppt_offset_bytes; /* soft pptable offset */
+	uint32_t ppt_size_bytes; /* soft pptable size */
+};
+
+struct smc_soft_pptable_entry {
+        uint32_t id;
+        uint32_t ppt_offset_bytes;
+        uint32_t ppt_size_bytes;
+};
+
+/* version_major=2, version_minor=1 */
+struct smc_firmware_header_v2_1 {
+        struct smc_firmware_header_v1_0 v1_0;
+        uint32_t pptable_count;
+        uint32_t pptable_entry_offset;
+};
+
+struct psp_fw_legacy_bin_desc {
+	uint32_t fw_version;
+	uint32_t offset_bytes;
+	uint32_t size_bytes;
+};
+
 /* version_major=1, version_minor=0 */
 struct psp_firmware_header_v1_0 {
 	struct common_firmware_header header;
-	uint32_t ucode_feature_version;
-	uint32_t sos_offset_bytes;
-	uint32_t sos_size_bytes;
+	struct psp_fw_legacy_bin_desc sos;
+};
+
+/* version_major=1, version_minor=1 */
+struct psp_firmware_header_v1_1 {
+	struct psp_firmware_header_v1_0 v1_0;
+	struct psp_fw_legacy_bin_desc toc;
+	struct psp_fw_legacy_bin_desc kdb;
+};
+
+/* version_major=1, version_minor=2 */
+struct psp_firmware_header_v1_2 {
+	struct psp_firmware_header_v1_0 v1_0;
+	struct psp_fw_legacy_bin_desc res;
+	struct psp_fw_legacy_bin_desc kdb;
+};
+
+/* version_major=1, version_minor=3 */
+struct psp_firmware_header_v1_3 {
+	struct psp_firmware_header_v1_1 v1_1;
+	struct psp_fw_legacy_bin_desc spl;
+	struct psp_fw_legacy_bin_desc rl;
+	struct psp_fw_legacy_bin_desc sys_drv_aux;
+	struct psp_fw_legacy_bin_desc sos_aux;
+};
+
+struct psp_fw_bin_desc {
+	uint32_t fw_type;
+	uint32_t fw_version;
+	uint32_t offset_bytes;
+	uint32_t size_bytes;
+};
+
+enum psp_fw_type {
+	PSP_FW_TYPE_UNKOWN,
+	PSP_FW_TYPE_PSP_SOS,
+	PSP_FW_TYPE_PSP_SYS_DRV,
+	PSP_FW_TYPE_PSP_KDB,
+	PSP_FW_TYPE_PSP_TOC,
+	PSP_FW_TYPE_PSP_SPL,
+	PSP_FW_TYPE_PSP_RL,
+	PSP_FW_TYPE_PSP_SOC_DRV,
+	PSP_FW_TYPE_PSP_INTF_DRV,
+	PSP_FW_TYPE_PSP_DBG_DRV,
+	PSP_FW_TYPE_PSP_RAS_DRV,
+	PSP_FW_TYPE_PSP_IPKEYMGR_DRV,
+	PSP_FW_TYPE_PSP_SPDM_DRV,
+	PSP_FW_TYPE_MAX_INDEX,
+};
+
+/* version_major=2, version_minor=0 */
+struct psp_firmware_header_v2_0 {
+	struct common_firmware_header header;
+	uint32_t psp_fw_bin_count;
+	struct psp_fw_bin_desc psp_fw_bin[];
+};
+
+/* version_major=2, version_minor=1 */
+struct psp_firmware_header_v2_1 {
+	struct common_firmware_header header;
+	uint32_t psp_fw_bin_count;
+	uint32_t psp_aux_fw_bin_index;
+	struct psp_fw_bin_desc psp_fw_bin[];
+};
+
+/* version_major=1, version_minor=0 */
+struct ta_firmware_header_v1_0 {
+	struct common_firmware_header header;
+	struct psp_fw_legacy_bin_desc xgmi;
+	struct psp_fw_legacy_bin_desc ras;
+	struct psp_fw_legacy_bin_desc hdcp;
+	struct psp_fw_legacy_bin_desc dtm;
+	struct psp_fw_legacy_bin_desc securedisplay;
+};
+
+enum ta_fw_type {
+	TA_FW_TYPE_UNKOWN,
+	TA_FW_TYPE_PSP_ASD,
+	TA_FW_TYPE_PSP_XGMI,
+	TA_FW_TYPE_PSP_RAS,
+	TA_FW_TYPE_PSP_HDCP,
+	TA_FW_TYPE_PSP_DTM,
+	TA_FW_TYPE_PSP_RAP,
+	TA_FW_TYPE_PSP_SECUREDISPLAY,
+	TA_FW_TYPE_PSP_XGMI_AUX,
+	TA_FW_TYPE_MAX_INDEX,
+};
+
+/* version_major=2, version_minor=0 */
+struct ta_firmware_header_v2_0 {
+	struct common_firmware_header header;
+	uint32_t ta_fw_bin_count;
+	struct psp_fw_bin_desc ta_fw_bin[];
 };
 
 /* version_major=1, version_minor=0 */
@@ -65,6 +185,33 @@ struct gfx_firmware_header_v1_0 {
 	uint32_t jt_size;  /* size of jt */
 };
 
+/* version_major=2, version_minor=0 */
+struct gfx_firmware_header_v2_0 {
+	struct common_firmware_header header;
+	uint32_t ucode_feature_version;
+	uint32_t ucode_size_bytes;
+	uint32_t ucode_offset_bytes;
+	uint32_t data_size_bytes;
+	uint32_t data_offset_bytes;
+	uint32_t ucode_start_addr_lo;
+	uint32_t ucode_start_addr_hi;
+};
+
+/* version_major=1, version_minor=0 */
+struct mes_firmware_header_v1_0 {
+	struct common_firmware_header header;
+	uint32_t mes_ucode_version;
+	uint32_t mes_ucode_size_bytes;
+	uint32_t mes_ucode_offset_bytes;
+	uint32_t mes_ucode_data_version;
+	uint32_t mes_ucode_data_size_bytes;
+	uint32_t mes_ucode_data_offset_bytes;
+	uint32_t mes_uc_start_addr_lo;
+	uint32_t mes_uc_start_addr_hi;
+	uint32_t mes_data_start_addr_lo;
+	uint32_t mes_data_start_addr_hi;
+};
+
 /* version_major=1, version_minor=0 */
 struct rlc_firmware_header_v1_0 {
 	struct common_firmware_header header;
@@ -98,6 +245,61 @@ struct rlc_firmware_header_v2_0 {
 	uint32_t reg_list_separate_array_offset_bytes; /* payload offset from the start of the header */
 };
 
+/* version_major=2, version_minor=1 */
+struct rlc_firmware_header_v2_1 {
+	struct rlc_firmware_header_v2_0 v2_0;
+	uint32_t reg_list_format_direct_reg_list_length; /* length of direct reg list format array */
+	uint32_t save_restore_list_cntl_ucode_ver;
+	uint32_t save_restore_list_cntl_feature_ver;
+	uint32_t save_restore_list_cntl_size_bytes;
+	uint32_t save_restore_list_cntl_offset_bytes;
+	uint32_t save_restore_list_gpm_ucode_ver;
+	uint32_t save_restore_list_gpm_feature_ver;
+	uint32_t save_restore_list_gpm_size_bytes;
+	uint32_t save_restore_list_gpm_offset_bytes;
+	uint32_t save_restore_list_srm_ucode_ver;
+	uint32_t save_restore_list_srm_feature_ver;
+	uint32_t save_restore_list_srm_size_bytes;
+	uint32_t save_restore_list_srm_offset_bytes;
+};
+
+/* version_major=2, version_minor=2 */
+struct rlc_firmware_header_v2_2 {
+	struct rlc_firmware_header_v2_1 v2_1;
+	uint32_t rlc_iram_ucode_size_bytes;
+	uint32_t rlc_iram_ucode_offset_bytes;
+	uint32_t rlc_dram_ucode_size_bytes;
+	uint32_t rlc_dram_ucode_offset_bytes;
+};
+
+/* version_major=2, version_minor=3 */
+struct rlc_firmware_header_v2_3 {
+    struct rlc_firmware_header_v2_2 v2_2;
+    uint32_t rlcp_ucode_version;
+    uint32_t rlcp_ucode_feature_version;
+    uint32_t rlcp_ucode_size_bytes;
+    uint32_t rlcp_ucode_offset_bytes;
+    uint32_t rlcv_ucode_version;
+    uint32_t rlcv_ucode_feature_version;
+    uint32_t rlcv_ucode_size_bytes;
+    uint32_t rlcv_ucode_offset_bytes;
+};
+
+/* version_major=2, version_minor=4 */
+struct rlc_firmware_header_v2_4 {
+    struct rlc_firmware_header_v2_3 v2_3;
+    uint32_t global_tap_delays_ucode_size_bytes;
+    uint32_t global_tap_delays_ucode_offset_bytes;
+    uint32_t se0_tap_delays_ucode_size_bytes;
+    uint32_t se0_tap_delays_ucode_offset_bytes;
+    uint32_t se1_tap_delays_ucode_size_bytes;
+    uint32_t se1_tap_delays_ucode_offset_bytes;
+    uint32_t se2_tap_delays_ucode_size_bytes;
+    uint32_t se2_tap_delays_ucode_offset_bytes;
+    uint32_t se3_tap_delays_ucode_size_bytes;
+    uint32_t se3_tap_delays_ucode_offset_bytes;
+};
+
 /* version_major=1, version_minor=0 */
 struct sdma_firmware_header_v1_0 {
 	struct common_firmware_header header;
@@ -113,6 +315,57 @@ struct sdma_firmware_header_v1_1 {
 	uint32_t digest_size;
 };
 
+/* version_major=2, version_minor=0 */
+struct sdma_firmware_header_v2_0 {
+	struct common_firmware_header header;
+	uint32_t ucode_feature_version;
+	uint32_t ctx_ucode_size_bytes; /* context thread ucode size */
+	uint32_t ctx_jt_offset; /* context thread jt location */
+	uint32_t ctx_jt_size; /* context thread size of jt */
+	uint32_t ctl_ucode_offset;
+	uint32_t ctl_ucode_size_bytes; /* control thread ucode size */
+	uint32_t ctl_jt_offset; /* control thread jt location */
+	uint32_t ctl_jt_size; /* control thread size of jt */
+};
+
+/* version_major=1, version_minor=0 */
+struct vpe_firmware_header_v1_0 {
+	struct common_firmware_header header;
+	uint32_t ucode_feature_version;
+	uint32_t ctx_ucode_size_bytes; /* context thread ucode size */
+	uint32_t ctx_jt_offset; /* context thread jt location */
+	uint32_t ctx_jt_size; /* context thread size of jt */
+	uint32_t ctl_ucode_offset;
+	uint32_t ctl_ucode_size_bytes; /* control thread ucode size */
+	uint32_t ctl_jt_offset; /* control thread jt location */
+	uint32_t ctl_jt_size; /* control thread size of jt */
+};
+
+/* version_major=1, version_minor=0 */
+struct umsch_mm_firmware_header_v1_0 {
+	struct common_firmware_header header;
+	uint32_t umsch_mm_ucode_version;
+	uint32_t umsch_mm_ucode_size_bytes;
+	uint32_t umsch_mm_ucode_offset_bytes;
+	uint32_t umsch_mm_ucode_data_version;
+	uint32_t umsch_mm_ucode_data_size_bytes;
+	uint32_t umsch_mm_ucode_data_offset_bytes;
+	uint32_t umsch_mm_irq_start_addr_lo;
+	uint32_t umsch_mm_irq_start_addr_hi;
+	uint32_t umsch_mm_uc_start_addr_lo;
+	uint32_t umsch_mm_uc_start_addr_hi;
+	uint32_t umsch_mm_data_start_addr_lo;
+	uint32_t umsch_mm_data_start_addr_hi;
+};
+
+/* version_major=3, version_minor=0 */
+struct sdma_firmware_header_v3_0 {
+    struct common_firmware_header header;
+    uint32_t ucode_feature_version;
+    uint32_t ucode_offset_bytes;
+    uint32_t ucode_size_bytes;
+};
+
 /* gpu info payload */
 struct gpu_info_firmware_v1_0 {
 	uint32_t gc_num_se;
@@ -132,6 +385,19 @@ struct gpu_info_firmware_v1_0 {
 	uint32_t gc_lds_size;
 };
 
+struct gpu_info_firmware_v1_1 {
+	struct gpu_info_firmware_v1_0 v1_0;
+	uint32_t num_sc_per_sh;
+	uint32_t num_packer_per_sc;
+};
+
+/* gpu info payload
+ * version_major=1, version_minor=1 */
+struct gpu_info_firmware_v1_2 {
+	struct gpu_info_firmware_v1_1 v1_1;
+	struct gpu_info_soc_bounding_box_v1_0 soc_bounding_box;
+};
+
 /* version_major=1, version_minor=0 */
 struct gpu_info_firmware_header_v1_0 {
 	struct common_firmware_header header;
@@ -139,39 +405,138 @@ struct gpu_info_firmware_header_v1_0 {
 	uint16_t version_minor; /* version */
 };
 
+/* version_major=1, version_minor=0 */
+struct dmcu_firmware_header_v1_0 {
+	struct common_firmware_header header;
+	uint32_t intv_offset_bytes; /* interrupt vectors offset from end of header, in bytes */
+	uint32_t intv_size_bytes;  /* size of interrupt vectors, in bytes */
+};
+
+/* version_major=1, version_minor=0 */
+struct dmcub_firmware_header_v1_0 {
+	struct common_firmware_header header;
+	uint32_t inst_const_bytes; /* size of instruction region, in bytes */
+	uint32_t bss_data_bytes; /* size of bss/data region, in bytes */
+};
+
+/* version_major=1, version_minor=0 */
+struct imu_firmware_header_v1_0 {
+    struct common_firmware_header header;
+    uint32_t imu_iram_ucode_size_bytes;
+    uint32_t imu_iram_ucode_offset_bytes;
+    uint32_t imu_dram_ucode_size_bytes;
+    uint32_t imu_dram_ucode_offset_bytes;
+};
+
 /* header is fixed size */
 union amdgpu_firmware_header {
 	struct common_firmware_header common;
 	struct mc_firmware_header_v1_0 mc;
 	struct smc_firmware_header_v1_0 smc;
+	struct smc_firmware_header_v2_0 smc_v2_0;
 	struct psp_firmware_header_v1_0 psp;
+	struct psp_firmware_header_v1_1 psp_v1_1;
+	struct psp_firmware_header_v1_3 psp_v1_3;
+	struct psp_firmware_header_v2_0 psp_v2_0;
+	struct psp_firmware_header_v2_0 psp_v2_1;
+	struct ta_firmware_header_v1_0 ta;
+	struct ta_firmware_header_v2_0 ta_v2_0;
 	struct gfx_firmware_header_v1_0 gfx;
+	struct gfx_firmware_header_v2_0 gfx_v2_0;
 	struct rlc_firmware_header_v1_0 rlc;
 	struct rlc_firmware_header_v2_0 rlc_v2_0;
+	struct rlc_firmware_header_v2_1 rlc_v2_1;
+	struct rlc_firmware_header_v2_2 rlc_v2_2;
+	struct rlc_firmware_header_v2_3 rlc_v2_3;
+	struct rlc_firmware_header_v2_4 rlc_v2_4;
 	struct sdma_firmware_header_v1_0 sdma;
 	struct sdma_firmware_header_v1_1 sdma_v1_1;
+	struct sdma_firmware_header_v2_0 sdma_v2_0;
+	struct sdma_firmware_header_v3_0 sdma_v3_0;
 	struct gpu_info_firmware_header_v1_0 gpu_info;
+	struct dmcu_firmware_header_v1_0 dmcu;
+	struct dmcub_firmware_header_v1_0 dmcub;
+	struct imu_firmware_header_v1_0 imu;
 	uint8_t raw[0x100];
 };
 
+#define UCODE_MAX_PSP_PACKAGING (((sizeof(union amdgpu_firmware_header) - sizeof(struct common_firmware_header) - 4) / sizeof(struct psp_fw_bin_desc)) * 2)
+
 /*
  * fw loading support
  */
 enum AMDGPU_UCODE_ID {
-	AMDGPU_UCODE_ID_SDMA0 = 0,
+	AMDGPU_UCODE_ID_CAP = 0,
+	AMDGPU_UCODE_ID_SDMA0,
 	AMDGPU_UCODE_ID_SDMA1,
+	AMDGPU_UCODE_ID_SDMA2,
+	AMDGPU_UCODE_ID_SDMA3,
+	AMDGPU_UCODE_ID_SDMA4,
+	AMDGPU_UCODE_ID_SDMA5,
+	AMDGPU_UCODE_ID_SDMA6,
+	AMDGPU_UCODE_ID_SDMA7,
+	AMDGPU_UCODE_ID_SDMA_UCODE_TH0,
+	AMDGPU_UCODE_ID_SDMA_UCODE_TH1,
+	AMDGPU_UCODE_ID_SDMA_RS64,
 	AMDGPU_UCODE_ID_CP_CE,
 	AMDGPU_UCODE_ID_CP_PFP,
 	AMDGPU_UCODE_ID_CP_ME,
+	AMDGPU_UCODE_ID_CP_RS64_PFP,
+	AMDGPU_UCODE_ID_CP_RS64_ME,
+	AMDGPU_UCODE_ID_CP_RS64_MEC,
+	AMDGPU_UCODE_ID_CP_RS64_PFP_P0_STACK,
+	AMDGPU_UCODE_ID_CP_RS64_PFP_P1_STACK,
+	AMDGPU_UCODE_ID_CP_RS64_ME_P0_STACK,
+	AMDGPU_UCODE_ID_CP_RS64_ME_P1_STACK,
+	AMDGPU_UCODE_ID_CP_RS64_MEC_P0_STACK,
+	AMDGPU_UCODE_ID_CP_RS64_MEC_P1_STACK,
+	AMDGPU_UCODE_ID_CP_RS64_MEC_P2_STACK,
+	AMDGPU_UCODE_ID_CP_RS64_MEC_P3_STACK,
 	AMDGPU_UCODE_ID_CP_MEC1,
 	AMDGPU_UCODE_ID_CP_MEC1_JT,
 	AMDGPU_UCODE_ID_CP_MEC2,
 	AMDGPU_UCODE_ID_CP_MEC2_JT,
+	AMDGPU_UCODE_ID_CP_MES,
+	AMDGPU_UCODE_ID_CP_MES_DATA,
+	AMDGPU_UCODE_ID_CP_MES1,
+	AMDGPU_UCODE_ID_CP_MES1_DATA,
+	AMDGPU_UCODE_ID_IMU_I,
+	AMDGPU_UCODE_ID_IMU_D,
+	AMDGPU_UCODE_ID_GLOBAL_TAP_DELAYS,
+	AMDGPU_UCODE_ID_SE0_TAP_DELAYS,
+	AMDGPU_UCODE_ID_SE1_TAP_DELAYS,
+	AMDGPU_UCODE_ID_SE2_TAP_DELAYS,
+	AMDGPU_UCODE_ID_SE3_TAP_DELAYS,
+	AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL,
+	AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM,
+	AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM,
+	AMDGPU_UCODE_ID_RLC_IRAM,
+	AMDGPU_UCODE_ID_RLC_DRAM,
+	AMDGPU_UCODE_ID_RLC_P,
+	AMDGPU_UCODE_ID_RLC_V,
 	AMDGPU_UCODE_ID_RLC_G,
 	AMDGPU_UCODE_ID_STORAGE,
 	AMDGPU_UCODE_ID_SMC,
+	AMDGPU_UCODE_ID_PPTABLE,
 	AMDGPU_UCODE_ID_UVD,
+	AMDGPU_UCODE_ID_UVD1,
 	AMDGPU_UCODE_ID_VCE,
+	AMDGPU_UCODE_ID_VCN,
+	AMDGPU_UCODE_ID_VCN1,
+	AMDGPU_UCODE_ID_DMCU_ERAM,
+	AMDGPU_UCODE_ID_DMCU_INTV,
+	AMDGPU_UCODE_ID_VCN0_RAM,
+	AMDGPU_UCODE_ID_VCN1_RAM,
+	AMDGPU_UCODE_ID_DMCUB,
+	AMDGPU_UCODE_ID_VPE_CTX,
+	AMDGPU_UCODE_ID_VPE_CTL,
+	AMDGPU_UCODE_ID_VPE,
+	AMDGPU_UCODE_ID_UMSCH_MM_UCODE,
+	AMDGPU_UCODE_ID_UMSCH_MM_DATA,
+	AMDGPU_UCODE_ID_UMSCH_MM_CMD_BUFFER,
+	AMDGPU_UCODE_ID_P2S_TABLE,
+	AMDGPU_UCODE_ID_JPEG_RAM,
+	AMDGPU_UCODE_ID_ISP,
 	AMDGPU_UCODE_ID_MAXIMUM,
 };
 
@@ -182,6 +547,18 @@ enum AMDGPU_UCODE_STATUS {
 	AMDGPU_UCODE_STATUS_LOADED,
 };
 
+enum amdgpu_firmware_load_type {
+	AMDGPU_FW_LOAD_DIRECT = 0,
+	AMDGPU_FW_LOAD_PSP,
+	AMDGPU_FW_LOAD_SMU,
+	AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO,
+};
+
+enum amdgpu_ucode_required {
+	AMDGPU_UCODE_OPTIONAL,
+	AMDGPU_UCODE_REQUIRED,
+};
+
 /* conform to smu_ucode_xfer_cz.h */
 #define AMDGPU_SDMA0_UCODE_LOADED	0x00000001
 #define AMDGPU_SDMA1_UCODE_LOADED	0x00000002
@@ -204,21 +581,62 @@ struct amdgpu_firmware_info {
 	void *kaddr;
 	/* ucode_size_bytes */
 	uint32_t ucode_size;
+	/* starting tmr mc address */
+	uint32_t tmr_mc_addr_lo;
+	uint32_t tmr_mc_addr_hi;
+};
+
+struct amdgpu_firmware {
+	struct amdgpu_firmware_info ucode[AMDGPU_UCODE_ID_MAXIMUM];
+	enum amdgpu_firmware_load_type load_type;
+	struct amdgpu_bo *fw_buf;
+	unsigned int fw_size;
+	unsigned int max_ucodes;
+	/* firmwares are loaded by psp instead of smu from vega10 */
+	const struct amdgpu_psp_funcs *funcs;
+	struct amdgpu_bo *rbuf;
+	struct mutex mutex;
+
+	/* gpu info firmware data pointer */
+	const struct firmware *gpu_info_fw;
+
+	void *fw_buf_ptr;
+	uint64_t fw_buf_mc;
+	uint32_t pldm_version;
+};
+
+struct kicker_device{
+	unsigned short device;
+	u8 revision;
 };
 
 void amdgpu_ucode_print_mc_hdr(const struct common_firmware_header *hdr);
 void amdgpu_ucode_print_smc_hdr(const struct common_firmware_header *hdr);
+void amdgpu_ucode_print_imu_hdr(const struct common_firmware_header *hdr);
 void amdgpu_ucode_print_gfx_hdr(const struct common_firmware_header *hdr);
 void amdgpu_ucode_print_rlc_hdr(const struct common_firmware_header *hdr);
 void amdgpu_ucode_print_sdma_hdr(const struct common_firmware_header *hdr);
+void amdgpu_ucode_print_psp_hdr(const struct common_firmware_header *hdr);
 void amdgpu_ucode_print_gpu_info_hdr(const struct common_firmware_header *hdr);
-int amdgpu_ucode_validate(const struct firmware *fw);
+__printf(4, 5)
+int amdgpu_ucode_request(struct amdgpu_device *adev, const struct firmware **fw,
+			 enum amdgpu_ucode_required required, const char *fmt, ...);
+void amdgpu_ucode_release(const struct firmware **fw);
 bool amdgpu_ucode_hdr_version(union amdgpu_firmware_header *hdr,
 				uint16_t hdr_major, uint16_t hdr_minor);
+
 int amdgpu_ucode_init_bo(struct amdgpu_device *adev);
-int amdgpu_ucode_fini_bo(struct amdgpu_device *adev);
+int amdgpu_ucode_create_bo(struct amdgpu_device *adev);
+int amdgpu_ucode_sysfs_init(struct amdgpu_device *adev);
+void amdgpu_ucode_free_bo(struct amdgpu_device *adev);
+void amdgpu_ucode_sysfs_fini(struct amdgpu_device *adev);
 
 enum amdgpu_firmware_load_type
 amdgpu_ucode_get_load_type(struct amdgpu_device *adev, int load_type);
 
+const char *amdgpu_ucode_name(enum AMDGPU_UCODE_ID ucode_id);
+
+void amdgpu_ucode_ip_version_decode(struct amdgpu_device *adev, int block_type, char *ucode_prefix, int len);
+bool amdgpu_is_kicker_fw(struct amdgpu_device *adev);
+
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
new file mode 100644
index 000000000000..2e039fb778ea
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
@@ -0,0 +1,588 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/sort.h>
+#include "amdgpu.h"
+#include "umc_v6_7.h"
+#define MAX_UMC_POISON_POLLING_TIME_SYNC   20  //ms
+
+#define MAX_UMC_HASH_STRING_SIZE  256
+
+static int amdgpu_umc_convert_error_address(struct amdgpu_device *adev,
+				    struct ras_err_data *err_data, uint64_t err_addr,
+				    uint32_t ch_inst, uint32_t umc_inst)
+{
+	switch (amdgpu_ip_version(adev, UMC_HWIP, 0)) {
+	case IP_VERSION(6, 7, 0):
+		umc_v6_7_convert_error_address(adev,
+				err_data, err_addr, ch_inst, umc_inst);
+		break;
+	default:
+		dev_warn(adev->dev,
+			 "UMC address to Physical address translation is not supported\n");
+		return AMDGPU_RAS_FAIL;
+	}
+
+	return AMDGPU_RAS_SUCCESS;
+}
+
+int amdgpu_umc_page_retirement_mca(struct amdgpu_device *adev,
+			uint64_t err_addr, uint32_t ch_inst, uint32_t umc_inst)
+{
+	struct ras_err_data err_data;
+	int ret;
+
+	ret = amdgpu_ras_error_data_init(&err_data);
+	if (ret)
+		return ret;
+
+	err_data.err_addr =
+		kcalloc(adev->umc.max_ras_err_cnt_per_query,
+			sizeof(struct eeprom_table_record), GFP_KERNEL);
+	if (!err_data.err_addr) {
+		dev_warn(adev->dev,
+			"Failed to alloc memory for umc error record in MCA notifier!\n");
+		ret = AMDGPU_RAS_FAIL;
+		goto out_fini_err_data;
+	}
+
+	err_data.err_addr_len = adev->umc.max_ras_err_cnt_per_query;
+
+	/*
+	 * Translate UMC channel address to Physical address
+	 */
+	ret = amdgpu_umc_convert_error_address(adev, &err_data, err_addr,
+					ch_inst, umc_inst);
+	if (ret)
+		goto out_free_err_addr;
+
+	if (amdgpu_bad_page_threshold != 0) {
+		amdgpu_ras_add_bad_pages(adev, err_data.err_addr,
+						err_data.err_addr_cnt, false);
+		amdgpu_ras_save_bad_pages(adev, NULL);
+	}
+
+out_free_err_addr:
+	kfree(err_data.err_addr);
+
+out_fini_err_data:
+	amdgpu_ras_error_data_fini(&err_data);
+
+	return ret;
+}
+
+void amdgpu_umc_handle_bad_pages(struct amdgpu_device *adev,
+			void *ras_error_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	unsigned int error_query_mode;
+	int ret = 0;
+	unsigned long err_count;
+
+	amdgpu_ras_get_error_query_mode(adev, &error_query_mode);
+
+	mutex_lock(&con->page_retirement_lock);
+	ret = amdgpu_dpm_get_ecc_info(adev, (void *)&(con->umc_ecc));
+	if (ret == -EOPNOTSUPP &&
+	    error_query_mode == AMDGPU_RAS_DIRECT_ERROR_QUERY) {
+		if (adev->umc.ras && adev->umc.ras->ras_block.hw_ops &&
+		    adev->umc.ras->ras_block.hw_ops->query_ras_error_count)
+		    adev->umc.ras->ras_block.hw_ops->query_ras_error_count(adev, ras_error_status);
+
+		if (adev->umc.ras && adev->umc.ras->ras_block.hw_ops &&
+		    adev->umc.ras->ras_block.hw_ops->query_ras_error_address &&
+		    adev->umc.max_ras_err_cnt_per_query) {
+			err_data->err_addr =
+				kcalloc(adev->umc.max_ras_err_cnt_per_query,
+					sizeof(struct eeprom_table_record), GFP_KERNEL);
+
+			/* still call query_ras_error_address to clear error status
+			 * even NOMEM error is encountered
+			 */
+			if(!err_data->err_addr)
+				dev_warn(adev->dev, "Failed to alloc memory for "
+						"umc error address record!\n");
+			else
+				err_data->err_addr_len = adev->umc.max_ras_err_cnt_per_query;
+
+			/* umc query_ras_error_address is also responsible for clearing
+			 * error status
+			 */
+			adev->umc.ras->ras_block.hw_ops->query_ras_error_address(adev, ras_error_status);
+		}
+	} else if (error_query_mode == AMDGPU_RAS_FIRMWARE_ERROR_QUERY ||
+	    (!ret && error_query_mode == AMDGPU_RAS_DIRECT_ERROR_QUERY)) {
+		if (adev->umc.ras &&
+		    adev->umc.ras->ecc_info_query_ras_error_count)
+		    adev->umc.ras->ecc_info_query_ras_error_count(adev, ras_error_status);
+
+		if (adev->umc.ras &&
+		    adev->umc.ras->ecc_info_query_ras_error_address &&
+		    adev->umc.max_ras_err_cnt_per_query) {
+			err_data->err_addr =
+				kcalloc(adev->umc.max_ras_err_cnt_per_query,
+					sizeof(struct eeprom_table_record), GFP_KERNEL);
+
+			/* still call query_ras_error_address to clear error status
+			 * even NOMEM error is encountered
+			 */
+			if(!err_data->err_addr)
+				dev_warn(adev->dev, "Failed to alloc memory for "
+						"umc error address record!\n");
+			else
+				err_data->err_addr_len = adev->umc.max_ras_err_cnt_per_query;
+
+			/* umc query_ras_error_address is also responsible for clearing
+			 * error status
+			 */
+			adev->umc.ras->ecc_info_query_ras_error_address(adev, ras_error_status);
+		}
+	}
+
+	/* only uncorrectable error needs gpu reset */
+	if (err_data->ue_count || err_data->de_count) {
+		err_count = err_data->ue_count + err_data->de_count;
+		if ((amdgpu_bad_page_threshold != 0) &&
+			err_data->err_addr_cnt) {
+			amdgpu_ras_add_bad_pages(adev, err_data->err_addr,
+						err_data->err_addr_cnt, false);
+			amdgpu_ras_save_bad_pages(adev, &err_count);
+
+			amdgpu_dpm_send_hbm_bad_pages_num(adev,
+					con->eeprom_control.ras_num_bad_pages);
+
+			if (con->update_channel_flag == true) {
+				amdgpu_dpm_send_hbm_bad_channel_flag(adev, con->eeprom_control.bad_channel_bitmap);
+				con->update_channel_flag = false;
+			}
+		}
+	}
+
+	kfree(err_data->err_addr);
+	err_data->err_addr = NULL;
+
+	mutex_unlock(&con->page_retirement_lock);
+}
+
+static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev,
+		void *ras_error_status,
+		struct amdgpu_iv_entry *entry,
+		uint32_t reset)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+
+	kgd2kfd_set_sram_ecc_flag(adev->kfd.dev);
+	amdgpu_umc_handle_bad_pages(adev, ras_error_status);
+
+	if ((err_data->ue_count || err_data->de_count) &&
+	    (reset || amdgpu_ras_is_rma(adev))) {
+		con->gpu_reset_flags |= reset;
+		amdgpu_ras_reset_gpu(adev);
+	}
+
+	return AMDGPU_RAS_SUCCESS;
+}
+
+int amdgpu_umc_pasid_poison_handler(struct amdgpu_device *adev,
+			enum amdgpu_ras_block block, uint16_t pasid,
+			pasid_notify pasid_fn, void *data, uint32_t reset)
+{
+	int ret = AMDGPU_RAS_SUCCESS;
+
+	if (adev->gmc.xgmi.connected_to_cpu ||
+		adev->gmc.is_app_apu) {
+		if (reset) {
+			/* MCA poison handler is only responsible for GPU reset,
+			 * let MCA notifier do page retirement.
+			 */
+			kgd2kfd_set_sram_ecc_flag(adev->kfd.dev);
+			amdgpu_ras_reset_gpu(adev);
+		}
+		return ret;
+	}
+
+	if (!amdgpu_sriov_vf(adev)) {
+		if (amdgpu_ip_version(adev, UMC_HWIP, 0) < IP_VERSION(12, 0, 0)) {
+			struct ras_err_data err_data;
+			struct ras_common_if head = {
+				.block = AMDGPU_RAS_BLOCK__UMC,
+			};
+			struct ras_manager *obj = amdgpu_ras_find_obj(adev, &head);
+
+			ret = amdgpu_ras_error_data_init(&err_data);
+			if (ret)
+				return ret;
+
+			ret = amdgpu_umc_do_page_retirement(adev, &err_data, NULL, reset);
+
+			if (ret == AMDGPU_RAS_SUCCESS && obj) {
+				obj->err_data.ue_count += err_data.ue_count;
+				obj->err_data.ce_count += err_data.ce_count;
+				obj->err_data.de_count += err_data.de_count;
+			}
+
+			amdgpu_ras_error_data_fini(&err_data);
+		} else {
+			struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+			int ret;
+
+			ret = amdgpu_ras_put_poison_req(adev,
+				block, pasid, pasid_fn, data, reset);
+			if (!ret) {
+				atomic_inc(&con->page_retirement_req_cnt);
+				atomic_inc(&con->poison_consumption_count);
+				wake_up(&con->page_retirement_wq);
+			}
+		}
+	} else {
+		if (adev->virt.ops && adev->virt.ops->ras_poison_handler)
+			adev->virt.ops->ras_poison_handler(adev, block);
+		else
+			dev_warn(adev->dev,
+				"No ras_poison_handler interface in SRIOV!\n");
+	}
+
+	return ret;
+}
+
+int amdgpu_umc_poison_handler(struct amdgpu_device *adev,
+			enum amdgpu_ras_block block, uint32_t reset)
+{
+	return amdgpu_umc_pasid_poison_handler(adev,
+				block, 0, NULL, NULL, reset);
+}
+
+int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev,
+		void *ras_error_status,
+		struct amdgpu_iv_entry *entry)
+{
+	return amdgpu_umc_do_page_retirement(adev, ras_error_status, entry,
+				AMDGPU_RAS_GPU_RESET_MODE1_RESET);
+}
+
+int amdgpu_umc_ras_sw_init(struct amdgpu_device *adev)
+{
+	int err;
+	struct amdgpu_umc_ras *ras;
+
+	if (!adev->umc.ras)
+		return 0;
+
+	ras = adev->umc.ras;
+
+	err = amdgpu_ras_register_ras_block(adev, &ras->ras_block);
+	if (err) {
+		dev_err(adev->dev, "Failed to register umc ras block!\n");
+		return err;
+	}
+
+	strcpy(adev->umc.ras->ras_block.ras_comm.name, "umc");
+	ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__UMC;
+	ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
+	adev->umc.ras_if = &ras->ras_block.ras_comm;
+
+	if (!ras->ras_block.ras_late_init)
+		ras->ras_block.ras_late_init = amdgpu_umc_ras_late_init;
+
+	if (!ras->ras_block.ras_cb)
+		ras->ras_block.ras_cb = amdgpu_umc_process_ras_data_cb;
+
+	return 0;
+}
+
+int amdgpu_umc_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
+{
+	int r;
+
+	r = amdgpu_ras_block_late_init(adev, ras_block);
+	if (r)
+		return r;
+
+	if (amdgpu_sriov_vf(adev))
+		return r;
+
+	if (amdgpu_ras_is_supported(adev, ras_block->block)) {
+		r = amdgpu_irq_get(adev, &adev->gmc.ecc_irq, 0);
+		if (r)
+			goto late_fini;
+	}
+
+	/* ras init of specific umc version */
+	if (adev->umc.ras &&
+	    adev->umc.ras->err_cnt_init)
+		adev->umc.ras->err_cnt_init(adev);
+
+	return 0;
+
+late_fini:
+	amdgpu_ras_block_late_fini(adev, ras_block);
+	return r;
+}
+
+int amdgpu_umc_process_ecc_irq(struct amdgpu_device *adev,
+		struct amdgpu_irq_src *source,
+		struct amdgpu_iv_entry *entry)
+{
+	struct ras_common_if *ras_if = adev->umc.ras_if;
+	struct ras_dispatch_if ih_data = {
+		.entry = entry,
+	};
+
+	if (!ras_if)
+		return 0;
+
+	ih_data.head = *ras_if;
+
+	amdgpu_ras_interrupt_dispatch(adev, &ih_data);
+	return 0;
+}
+
+int amdgpu_umc_fill_error_record(struct ras_err_data *err_data,
+		uint64_t err_addr,
+		uint64_t retired_page,
+		uint32_t channel_index,
+		uint32_t umc_inst)
+{
+	struct eeprom_table_record *err_rec;
+
+	if (!err_data ||
+	    !err_data->err_addr ||
+	    (err_data->err_addr_cnt >= err_data->err_addr_len))
+		return -EINVAL;
+
+	err_rec = &err_data->err_addr[err_data->err_addr_cnt];
+
+	err_rec->address = err_addr;
+	/* page frame address is saved */
+	err_rec->retired_page = retired_page >> AMDGPU_GPU_PAGE_SHIFT;
+	err_rec->ts = (uint64_t)ktime_get_real_seconds();
+	err_rec->err_type = AMDGPU_RAS_EEPROM_ERR_NON_RECOVERABLE;
+	err_rec->cu = 0;
+	err_rec->mem_channel = channel_index;
+	err_rec->mcumc_id = umc_inst;
+
+	err_data->err_addr_cnt++;
+
+	return 0;
+}
+
+static int amdgpu_umc_loop_all_aid(struct amdgpu_device *adev, umc_func func,
+				   void *data)
+{
+	uint32_t umc_node_inst;
+	uint32_t node_inst;
+	uint32_t umc_inst;
+	uint32_t ch_inst;
+	int ret;
+
+	/*
+	 * This loop is done based on the following -
+	 * umc.active mask = mask of active umc instances across all nodes
+	 * umc.umc_inst_num = maximum number of umc instancess per node
+	 * umc.node_inst_num = maximum number of node instances
+	 * Channel instances are not assumed to be harvested.
+	 */
+	dev_dbg(adev->dev, "active umcs :%lx umc_inst per node: %d",
+		adev->umc.active_mask, adev->umc.umc_inst_num);
+	for_each_set_bit(umc_node_inst, &(adev->umc.active_mask),
+			 adev->umc.node_inst_num * adev->umc.umc_inst_num) {
+		node_inst = umc_node_inst / adev->umc.umc_inst_num;
+		umc_inst = umc_node_inst % adev->umc.umc_inst_num;
+		LOOP_UMC_CH_INST(ch_inst) {
+			dev_dbg(adev->dev,
+				"node_inst :%d umc_inst: %d ch_inst: %d",
+				node_inst, umc_inst, ch_inst);
+			ret = func(adev, node_inst, umc_inst, ch_inst, data);
+			if (ret) {
+				dev_err(adev->dev,
+					"Node %d umc %d ch %d func returns %d\n",
+					node_inst, umc_inst, ch_inst, ret);
+				return ret;
+			}
+		}
+	}
+
+	return 0;
+}
+
+int amdgpu_umc_loop_channels(struct amdgpu_device *adev,
+			umc_func func, void *data)
+{
+	uint32_t node_inst       = 0;
+	uint32_t umc_inst        = 0;
+	uint32_t ch_inst         = 0;
+	int ret = 0;
+
+	if (adev->aid_mask)
+		return amdgpu_umc_loop_all_aid(adev, func, data);
+
+	if (adev->umc.node_inst_num) {
+		LOOP_UMC_EACH_NODE_INST_AND_CH(node_inst, umc_inst, ch_inst) {
+			ret = func(adev, node_inst, umc_inst, ch_inst, data);
+			if (ret) {
+				dev_err(adev->dev, "Node %d umc %d ch %d func returns %d\n",
+					node_inst, umc_inst, ch_inst, ret);
+				return ret;
+			}
+		}
+	} else {
+		LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) {
+			ret = func(adev, 0, umc_inst, ch_inst, data);
+			if (ret) {
+				dev_err(adev->dev, "Umc %d ch %d func returns %d\n",
+					umc_inst, ch_inst, ret);
+				return ret;
+			}
+		}
+	}
+
+	return 0;
+}
+
+int amdgpu_umc_update_ecc_status(struct amdgpu_device *adev,
+				uint64_t status, uint64_t ipid, uint64_t addr)
+{
+	if (adev->umc.ras->update_ecc_status)
+		return adev->umc.ras->update_ecc_status(adev,
+					status, ipid, addr);
+	return 0;
+}
+
+int amdgpu_umc_logs_ecc_err(struct amdgpu_device *adev,
+		struct radix_tree_root *ecc_tree, struct ras_ecc_err *ecc_err)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_ecc_log_info *ecc_log;
+	int ret;
+
+	ecc_log = &con->umc_ecc_log;
+
+	mutex_lock(&ecc_log->lock);
+	ret = radix_tree_insert(ecc_tree, ecc_err->pa_pfn, ecc_err);
+	if (!ret)
+		radix_tree_tag_set(ecc_tree,
+			ecc_err->pa_pfn, UMC_ECC_NEW_DETECTED_TAG);
+	mutex_unlock(&ecc_log->lock);
+
+	return ret;
+}
+
+int amdgpu_umc_pages_in_a_row(struct amdgpu_device *adev,
+			struct ras_err_data *err_data, uint64_t pa_addr)
+{
+	struct ta_ras_query_address_output addr_out;
+
+	/* reinit err_data */
+	err_data->err_addr_cnt = 0;
+	err_data->err_addr_len = adev->umc.retire_unit;
+
+	addr_out.pa.pa = pa_addr;
+	if (adev->umc.ras && adev->umc.ras->convert_ras_err_addr)
+		return adev->umc.ras->convert_ras_err_addr(adev, err_data, NULL,
+				&addr_out, false);
+	else
+		return -EINVAL;
+}
+
+int amdgpu_umc_lookup_bad_pages_in_a_row(struct amdgpu_device *adev,
+			uint64_t pa_addr, uint64_t *pfns, int len)
+{
+	int i, ret;
+	struct ras_err_data err_data;
+
+	err_data.err_addr = kcalloc(adev->umc.retire_unit,
+				sizeof(struct eeprom_table_record), GFP_KERNEL);
+	if (!err_data.err_addr) {
+		dev_warn(adev->dev, "Failed to alloc memory in bad page lookup!\n");
+		return 0;
+	}
+
+	ret = amdgpu_umc_pages_in_a_row(adev, &err_data, pa_addr);
+	if (ret)
+		goto out;
+
+	for (i = 0; i < adev->umc.retire_unit; i++) {
+		if (i >= len)
+			goto out;
+
+		pfns[i] = err_data.err_addr[i].retired_page;
+	}
+	ret = i;
+	adev->umc.err_addr_cnt = err_data.err_addr_cnt;
+
+out:
+	kfree(err_data.err_addr);
+	return ret;
+}
+
+int amdgpu_umc_mca_to_addr(struct amdgpu_device *adev,
+			uint64_t err_addr, uint32_t ch, uint32_t umc,
+			uint32_t node, uint32_t socket,
+			struct ta_ras_query_address_output *addr_out, bool dump_addr)
+{
+	struct ta_ras_query_address_input addr_in;
+	int ret;
+
+	memset(&addr_in, 0, sizeof(addr_in));
+	addr_in.ma.err_addr = err_addr;
+	addr_in.ma.ch_inst = ch;
+	addr_in.ma.umc_inst = umc;
+	addr_in.ma.node_inst = node;
+	addr_in.ma.socket_id = socket;
+
+	if (adev->umc.ras && adev->umc.ras->convert_ras_err_addr) {
+		ret = adev->umc.ras->convert_ras_err_addr(adev, NULL, &addr_in,
+				addr_out, dump_addr);
+		if (ret)
+			return ret;
+	} else {
+		return 0;
+	}
+
+	return 0;
+}
+
+int amdgpu_umc_pa2mca(struct amdgpu_device *adev,
+		uint64_t pa, uint64_t *mca, enum amdgpu_memory_partition nps)
+{
+	struct ta_ras_query_address_input addr_in;
+	struct ta_ras_query_address_output addr_out;
+	int ret;
+
+	/* nps: the pa belongs to */
+	addr_in.pa.pa = pa | ((uint64_t)nps << 58);
+	addr_in.addr_type = TA_RAS_PA_TO_MCA;
+	ret = psp_ras_query_address(&adev->psp, &addr_in, &addr_out);
+	if (ret) {
+		dev_warn(adev->dev, "Failed to query RAS MCA address for 0x%llx",
+			pa);
+
+		return ret;
+	}
+
+	*mca = addr_out.ma.err_addr;
+
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h
new file mode 100644
index 000000000000..ec203f9e5ffa
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h
@@ -0,0 +1,194 @@
+/*
+ * Copyright (C) 2019  Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+#ifndef __AMDGPU_UMC_H__
+#define __AMDGPU_UMC_H__
+#include "amdgpu_ras.h"
+#include "amdgpu_mca.h"
+/*
+ * (addr / 256) * 4096, the higher 26 bits in ErrorAddr
+ * is the index of 4KB block
+ */
+#define ADDR_OF_4KB_BLOCK(addr)			(((addr) & ~0xffULL) << 4)
+/*
+ * (addr / 256) * 8192, the higher 26 bits in ErrorAddr
+ * is the index of 8KB block
+ */
+#define ADDR_OF_8KB_BLOCK(addr)			(((addr) & ~0xffULL) << 5)
+/*
+ * (addr / 256) * 32768, the higher 26 bits in ErrorAddr
+ * is the index of 8KB block
+ */
+#define ADDR_OF_32KB_BLOCK(addr)			(((addr) & ~0xffULL) << 7)
+/* channel index is the index of 256B block */
+#define ADDR_OF_256B_BLOCK(channel_index)	((channel_index) << 8)
+/* offset in 256B block */
+#define OFFSET_IN_256B_BLOCK(addr)		((addr) & 0xffULL)
+
+#define LOOP_UMC_INST(umc_inst) for ((umc_inst) = 0; (umc_inst) < adev->umc.umc_inst_num; (umc_inst)++)
+#define LOOP_UMC_CH_INST(ch_inst) for ((ch_inst) = 0; (ch_inst) < adev->umc.channel_inst_num; (ch_inst)++)
+#define LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) LOOP_UMC_INST((umc_inst)) LOOP_UMC_CH_INST((ch_inst))
+
+#define LOOP_UMC_NODE_INST(node_inst) \
+		for_each_set_bit((node_inst), &(adev->umc.active_mask), adev->umc.node_inst_num)
+
+#define LOOP_UMC_EACH_NODE_INST_AND_CH(node_inst, umc_inst, ch_inst) \
+		LOOP_UMC_NODE_INST((node_inst)) LOOP_UMC_INST_AND_CH((umc_inst), (ch_inst))
+
+/* Page retirement tag */
+#define UMC_ECC_NEW_DETECTED_TAG       0x1
+/*
+ * a flag to indicate v2 of channel index stored in eeprom
+ *
+ * v1 (legacy way): store channel index within a umc instance in eeprom
+ *    range in UMC v12: 0 ~ 7
+ * v2: store global channel index in eeprom
+ *    range in UMC v12: 0 ~ 127
+ *
+ * NOTE: it's better to store it in eeprom_table_record.mem_channel,
+ * but there is only 8 bits in mem_channel, and the channel number may
+ * increase in the future, we decide to save it in
+ * eeprom_table_record.retired_page. retired_page is useless in v2,
+ * we depend on eeprom_table_record.address instead of retired_page in v2.
+ * Only 48 bits are saved on eeprom, use bit 47 here.
+ */
+#define UMC_CHANNEL_IDX_V2	BIT_ULL(47)
+
+/*
+ * save nps value to eeprom_table_record.retired_page[47:40],
+ * the channel index flag above will be retired.
+ */
+#define UMC_NPS_SHIFT 40
+#define UMC_NPS_MASK 0xffULL
+
+/* three column bits and one row bit in MCA address flip
+ * in bad page retirement
+ */
+#define RETIRE_FLIP_BITS_NUM 4
+
+struct amdgpu_umc_flip_bits {
+	uint32_t flip_bits_in_pa[RETIRE_FLIP_BITS_NUM];
+	uint32_t flip_row_bit;
+	uint32_t r13_in_pa;
+	uint32_t bit_num;
+};
+
+typedef int (*umc_func)(struct amdgpu_device *adev, uint32_t node_inst,
+			uint32_t umc_inst, uint32_t ch_inst, void *data);
+
+struct amdgpu_umc_ras {
+	struct amdgpu_ras_block_object ras_block;
+	void (*err_cnt_init)(struct amdgpu_device *adev);
+	bool (*query_ras_poison_mode)(struct amdgpu_device *adev);
+	void (*ecc_info_query_ras_error_count)(struct amdgpu_device *adev,
+				      void *ras_error_status);
+	void (*ecc_info_query_ras_error_address)(struct amdgpu_device *adev,
+					void *ras_error_status);
+	bool (*check_ecc_err_status)(struct amdgpu_device *adev,
+			enum amdgpu_mca_error_type type, void *ras_error_status);
+	int (*update_ecc_status)(struct amdgpu_device *adev,
+			uint64_t status, uint64_t ipid, uint64_t addr);
+	int (*convert_ras_err_addr)(struct amdgpu_device *adev,
+			struct ras_err_data *err_data,
+			struct ta_ras_query_address_input *addr_in,
+			struct ta_ras_query_address_output *addr_out,
+			bool dump_addr);
+	uint32_t (*get_die_id_from_pa)(struct amdgpu_device *adev,
+			uint64_t mca_addr, uint64_t retired_page);
+	void (*get_retire_flip_bits)(struct amdgpu_device *adev);
+};
+
+struct amdgpu_umc_funcs {
+	void (*init_registers)(struct amdgpu_device *adev);
+};
+
+struct amdgpu_umc {
+	/* max error count in one ras query call */
+	uint32_t max_ras_err_cnt_per_query;
+	/* number of umc channel instance with memory map register access */
+	uint32_t channel_inst_num;
+	/* number of umc instance with memory map register access */
+	uint32_t umc_inst_num;
+
+	/* Total number of umc node instance including harvest one */
+	uint32_t node_inst_num;
+
+	/* UMC regiser per channel offset */
+	uint32_t channel_offs;
+	/* how many pages are retired in one UE */
+	uint32_t retire_unit;
+	/* channel index table of interleaved memory */
+	const uint32_t *channel_idx_tbl;
+	struct ras_common_if *ras_if;
+
+	const struct amdgpu_umc_funcs *funcs;
+	struct amdgpu_umc_ras *ras;
+
+	/* active mask for umc node instance */
+	unsigned long active_mask;
+
+	struct amdgpu_umc_flip_bits flip_bits;
+
+	unsigned long err_addr_cnt;
+};
+
+int amdgpu_umc_ras_sw_init(struct amdgpu_device *adev);
+int amdgpu_umc_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block);
+int amdgpu_umc_poison_handler(struct amdgpu_device *adev,
+			enum amdgpu_ras_block block, uint32_t reset);
+int amdgpu_umc_pasid_poison_handler(struct amdgpu_device *adev,
+			enum amdgpu_ras_block block, uint16_t pasid,
+			pasid_notify pasid_fn, void *data, uint32_t reset);
+int amdgpu_umc_process_ecc_irq(struct amdgpu_device *adev,
+		struct amdgpu_irq_src *source,
+		struct amdgpu_iv_entry *entry);
+int amdgpu_umc_fill_error_record(struct ras_err_data *err_data,
+		uint64_t err_addr,
+		uint64_t retired_page,
+		uint32_t channel_index,
+		uint32_t umc_inst);
+
+int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev,
+		void *ras_error_status,
+		struct amdgpu_iv_entry *entry);
+int amdgpu_umc_page_retirement_mca(struct amdgpu_device *adev,
+			uint64_t err_addr, uint32_t ch_inst, uint32_t umc_inst);
+
+int amdgpu_umc_loop_channels(struct amdgpu_device *adev,
+			umc_func func, void *data);
+
+int amdgpu_umc_update_ecc_status(struct amdgpu_device *adev,
+				uint64_t status, uint64_t ipid, uint64_t addr);
+int amdgpu_umc_logs_ecc_err(struct amdgpu_device *adev,
+		struct radix_tree_root *ecc_tree, struct ras_ecc_err *ecc_err);
+
+void amdgpu_umc_handle_bad_pages(struct amdgpu_device *adev,
+			void *ras_error_status);
+int amdgpu_umc_pages_in_a_row(struct amdgpu_device *adev,
+			struct ras_err_data *err_data, uint64_t pa_addr);
+int amdgpu_umc_lookup_bad_pages_in_a_row(struct amdgpu_device *adev,
+			uint64_t pa_addr, uint64_t *pfns, int len);
+int amdgpu_umc_mca_to_addr(struct amdgpu_device *adev,
+			uint64_t err_addr, uint32_t ch, uint32_t umc,
+			uint32_t node, uint32_t socket,
+			struct ta_ras_query_address_output *addr_out, bool dump_addr);
+int amdgpu_umc_pa2mca(struct amdgpu_device *adev,
+		uint64_t pa, uint64_t *mca, enum amdgpu_memory_partition nps);
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umr.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_umr.h
new file mode 100644
index 000000000000..5b27fc41ffbf
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umr.h
@@ -0,0 +1,85 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/ioctl.h>
+
+/*
+ * MMIO debugfs IOCTL structure
+ */
+struct amdgpu_debugfs_regs2_iocdata {
+	__u32 use_srbm, use_grbm, pg_lock;
+	struct {
+		__u32 se, sh, instance;
+	} grbm;
+	struct {
+		__u32 me, pipe, queue, vmid;
+	} srbm;
+};
+
+struct amdgpu_debugfs_regs2_iocdata_v2 {
+	__u32 use_srbm, use_grbm, pg_lock;
+	struct {
+		__u32 se, sh, instance;
+	} grbm;
+	struct {
+		__u32 me, pipe, queue, vmid;
+	} srbm;
+	u32 xcc_id;
+};
+
+struct amdgpu_debugfs_gprwave_iocdata {
+	u32 gpr_or_wave, se, sh, cu, wave, simd, xcc_id;
+	struct {
+		u32 thread, vpgr_or_sgpr;
+	} gpr;
+};
+
+/*
+ * MMIO debugfs state data (per file* handle)
+ */
+struct amdgpu_debugfs_regs2_data {
+	struct amdgpu_device *adev;
+	struct mutex lock;
+	struct amdgpu_debugfs_regs2_iocdata_v2 id;
+};
+
+struct amdgpu_debugfs_gprwave_data {
+	struct amdgpu_device *adev;
+	struct mutex lock;
+	struct amdgpu_debugfs_gprwave_iocdata id;
+};
+
+enum AMDGPU_DEBUGFS_REGS2_CMDS {
+	AMDGPU_DEBUGFS_REGS2_CMD_SET_STATE = 0,
+	AMDGPU_DEBUGFS_REGS2_CMD_SET_STATE_V2,
+};
+
+enum AMDGPU_DEBUGFS_GPRWAVE_CMDS {
+	AMDGPU_DEBUGFS_GPRWAVE_CMD_SET_STATE = 0,
+};
+
+//reg2 interface
+#define AMDGPU_DEBUGFS_REGS2_IOC_SET_STATE _IOWR(0x20, AMDGPU_DEBUGFS_REGS2_CMD_SET_STATE, struct amdgpu_debugfs_regs2_iocdata)
+#define AMDGPU_DEBUGFS_REGS2_IOC_SET_STATE_V2 _IOWR(0x20, AMDGPU_DEBUGFS_REGS2_CMD_SET_STATE_V2, struct amdgpu_debugfs_regs2_iocdata_v2)
+
+//gprwave interface
+#define AMDGPU_DEBUGFS_GPRWAVE_IOC_SET_STATE _IOWR(0x20, AMDGPU_DEBUGFS_GPRWAVE_CMD_SET_STATE, struct amdgpu_debugfs_gprwave_iocdata)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c
new file mode 100644
index 000000000000..cd707d70a0bf
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c
@@ -0,0 +1,550 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include <linux/debugfs.h>
+#include <drm/drm_exec.h>
+#include <drm/drm_drv.h>
+
+#include "amdgpu.h"
+#include "amdgpu_umsch_mm.h"
+#include "umsch_mm_v4_0.h"
+
+MODULE_FIRMWARE("amdgpu/umsch_mm_4_0_0.bin");
+
+int amdgpu_umsch_mm_submit_pkt(struct amdgpu_umsch_mm *umsch, void *pkt, int ndws)
+{
+	struct amdgpu_ring *ring = &umsch->ring;
+
+	if (amdgpu_ring_alloc(ring, ndws))
+		return -ENOMEM;
+
+	amdgpu_ring_write_multiple(ring, pkt, ndws);
+	amdgpu_ring_commit(ring);
+
+	return 0;
+}
+
+int amdgpu_umsch_mm_query_fence(struct amdgpu_umsch_mm *umsch)
+{
+	struct amdgpu_ring *ring = &umsch->ring;
+	struct amdgpu_device *adev = ring->adev;
+	int r;
+
+	r = amdgpu_fence_wait_polling(ring, ring->fence_drv.sync_seq, adev->usec_timeout);
+	if (r < 1) {
+		dev_err(adev->dev, "ring umsch timeout, emitted fence %u\n",
+			ring->fence_drv.sync_seq);
+		return -ETIMEDOUT;
+	}
+
+	return 0;
+}
+
+static void umsch_mm_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_umsch_mm *umsch = (struct amdgpu_umsch_mm *)ring;
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell)
+		WDOORBELL32(ring->doorbell_index, ring->wptr << 2);
+	else
+		WREG32(umsch->rb_wptr, ring->wptr << 2);
+}
+
+static u64 umsch_mm_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_umsch_mm *umsch = (struct amdgpu_umsch_mm *)ring;
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32(umsch->rb_rptr);
+}
+
+static u64 umsch_mm_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_umsch_mm *umsch = (struct amdgpu_umsch_mm *)ring;
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32(umsch->rb_wptr);
+}
+
+static const struct amdgpu_ring_funcs umsch_v4_0_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_UMSCH_MM,
+	.align_mask = 0,
+	.nop = 0,
+	.support_64bit_ptrs = false,
+	.get_rptr = umsch_mm_ring_get_rptr,
+	.get_wptr = umsch_mm_ring_get_wptr,
+	.set_wptr = umsch_mm_ring_set_wptr,
+	.insert_nop = amdgpu_ring_insert_nop,
+};
+
+int amdgpu_umsch_mm_ring_init(struct amdgpu_umsch_mm *umsch)
+{
+	struct amdgpu_device *adev = container_of(umsch, struct amdgpu_device, umsch_mm);
+	struct amdgpu_ring *ring = &umsch->ring;
+
+	ring->vm_hub = AMDGPU_MMHUB0(0);
+	ring->use_doorbell = true;
+	ring->no_scheduler = true;
+	ring->doorbell_index = (AMDGPU_NAVI10_DOORBELL64_VCN0_1 << 1) + 6;
+
+	snprintf(ring->name, sizeof(ring->name), "umsch");
+
+	return amdgpu_ring_init(adev, ring, 1024, NULL, 0, AMDGPU_RING_PRIO_DEFAULT, NULL);
+}
+
+int amdgpu_umsch_mm_init_microcode(struct amdgpu_umsch_mm *umsch)
+{
+	const struct umsch_mm_firmware_header_v1_0 *umsch_mm_hdr;
+	struct amdgpu_device *adev = umsch->ring.adev;
+	const char *fw_name = NULL;
+	int r;
+
+	switch (amdgpu_ip_version(adev, VCN_HWIP, 0)) {
+	case IP_VERSION(4, 0, 5):
+	case IP_VERSION(4, 0, 6):
+		fw_name = "4_0_0";
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	r = amdgpu_ucode_request(adev, &adev->umsch_mm.fw, AMDGPU_UCODE_REQUIRED,
+				 "amdgpu/umsch_mm_%s.bin", fw_name);
+	if (r) {
+		release_firmware(adev->umsch_mm.fw);
+		adev->umsch_mm.fw = NULL;
+		return r;
+	}
+
+	umsch_mm_hdr = (const struct umsch_mm_firmware_header_v1_0 *)adev->umsch_mm.fw->data;
+
+	adev->umsch_mm.ucode_size = le32_to_cpu(umsch_mm_hdr->umsch_mm_ucode_size_bytes);
+	adev->umsch_mm.data_size = le32_to_cpu(umsch_mm_hdr->umsch_mm_ucode_data_size_bytes);
+
+	adev->umsch_mm.irq_start_addr =
+		le32_to_cpu(umsch_mm_hdr->umsch_mm_irq_start_addr_lo) |
+		((uint64_t)(le32_to_cpu(umsch_mm_hdr->umsch_mm_irq_start_addr_hi)) << 32);
+	adev->umsch_mm.uc_start_addr =
+		le32_to_cpu(umsch_mm_hdr->umsch_mm_uc_start_addr_lo) |
+		((uint64_t)(le32_to_cpu(umsch_mm_hdr->umsch_mm_uc_start_addr_hi)) << 32);
+	adev->umsch_mm.data_start_addr =
+		le32_to_cpu(umsch_mm_hdr->umsch_mm_data_start_addr_lo) |
+		((uint64_t)(le32_to_cpu(umsch_mm_hdr->umsch_mm_data_start_addr_hi)) << 32);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		struct amdgpu_firmware_info *info;
+
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_UMSCH_MM_UCODE];
+		info->ucode_id = AMDGPU_UCODE_ID_UMSCH_MM_UCODE;
+		info->fw = adev->umsch_mm.fw;
+		adev->firmware.fw_size +=
+			ALIGN(le32_to_cpu(umsch_mm_hdr->umsch_mm_ucode_size_bytes), PAGE_SIZE);
+
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_UMSCH_MM_DATA];
+		info->ucode_id = AMDGPU_UCODE_ID_UMSCH_MM_DATA;
+		info->fw = adev->umsch_mm.fw;
+		adev->firmware.fw_size +=
+			ALIGN(le32_to_cpu(umsch_mm_hdr->umsch_mm_ucode_data_size_bytes), PAGE_SIZE);
+	}
+
+	return 0;
+}
+
+int amdgpu_umsch_mm_allocate_ucode_buffer(struct amdgpu_umsch_mm *umsch)
+{
+	const struct umsch_mm_firmware_header_v1_0 *umsch_mm_hdr;
+	struct amdgpu_device *adev = umsch->ring.adev;
+	const __le32 *fw_data;
+	uint32_t fw_size;
+	int r;
+
+	umsch_mm_hdr = (const struct umsch_mm_firmware_header_v1_0 *)
+		       adev->umsch_mm.fw->data;
+
+	fw_data = (const __le32 *)(adev->umsch_mm.fw->data +
+		  le32_to_cpu(umsch_mm_hdr->umsch_mm_ucode_offset_bytes));
+	fw_size = le32_to_cpu(umsch_mm_hdr->umsch_mm_ucode_size_bytes);
+
+	r = amdgpu_bo_create_reserved(adev, fw_size,
+				      4 * 1024, AMDGPU_GEM_DOMAIN_VRAM,
+				      &adev->umsch_mm.ucode_fw_obj,
+				      &adev->umsch_mm.ucode_fw_gpu_addr,
+				      (void **)&adev->umsch_mm.ucode_fw_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create umsch_mm fw ucode bo\n", r);
+		return r;
+	}
+
+	memcpy(adev->umsch_mm.ucode_fw_ptr, fw_data, fw_size);
+
+	amdgpu_bo_kunmap(adev->umsch_mm.ucode_fw_obj);
+	amdgpu_bo_unreserve(adev->umsch_mm.ucode_fw_obj);
+	return 0;
+}
+
+int amdgpu_umsch_mm_allocate_ucode_data_buffer(struct amdgpu_umsch_mm *umsch)
+{
+	const struct umsch_mm_firmware_header_v1_0 *umsch_mm_hdr;
+	struct amdgpu_device *adev = umsch->ring.adev;
+	const __le32 *fw_data;
+	uint32_t fw_size;
+	int r;
+
+	umsch_mm_hdr = (const struct umsch_mm_firmware_header_v1_0 *)
+		       adev->umsch_mm.fw->data;
+
+	fw_data = (const __le32 *)(adev->umsch_mm.fw->data +
+		  le32_to_cpu(umsch_mm_hdr->umsch_mm_ucode_data_offset_bytes));
+	fw_size = le32_to_cpu(umsch_mm_hdr->umsch_mm_ucode_data_size_bytes);
+
+	r = amdgpu_bo_create_reserved(adev, fw_size,
+				      64 * 1024, AMDGPU_GEM_DOMAIN_VRAM,
+				      &adev->umsch_mm.data_fw_obj,
+				      &adev->umsch_mm.data_fw_gpu_addr,
+				      (void **)&adev->umsch_mm.data_fw_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create umsch_mm fw data bo\n", r);
+		return r;
+	}
+
+	memcpy(adev->umsch_mm.data_fw_ptr, fw_data, fw_size);
+
+	amdgpu_bo_kunmap(adev->umsch_mm.data_fw_obj);
+	amdgpu_bo_unreserve(adev->umsch_mm.data_fw_obj);
+	return 0;
+}
+
+int amdgpu_umsch_mm_psp_execute_cmd_buf(struct amdgpu_umsch_mm *umsch)
+{
+	struct amdgpu_device *adev = umsch->ring.adev;
+	struct amdgpu_firmware_info ucode = {
+		.ucode_id = AMDGPU_UCODE_ID_UMSCH_MM_CMD_BUFFER,
+		.mc_addr = adev->umsch_mm.cmd_buf_gpu_addr,
+		.ucode_size = ((uintptr_t)adev->umsch_mm.cmd_buf_curr_ptr -
+			      (uintptr_t)adev->umsch_mm.cmd_buf_ptr),
+	};
+
+	return psp_execute_ip_fw_load(&adev->psp, &ucode);
+}
+
+static void umsch_mm_agdb_index_init(struct amdgpu_device *adev)
+{
+	uint32_t umsch_mm_agdb_start;
+	int i;
+
+	umsch_mm_agdb_start = adev->doorbell_index.max_assignment + 1;
+	umsch_mm_agdb_start = roundup(umsch_mm_agdb_start, 1024);
+	umsch_mm_agdb_start += (AMDGPU_NAVI10_DOORBELL64_VCN0_1 << 1);
+
+	for (i = 0; i < CONTEXT_PRIORITY_NUM_LEVELS; i++)
+		adev->umsch_mm.agdb_index[i] = umsch_mm_agdb_start + i;
+}
+
+static int umsch_mm_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	adev->umsch_mm.vmid_mask_mm_vpe = 0xf00;
+	adev->umsch_mm.engine_mask = (1 << UMSCH_SWIP_ENGINE_TYPE_VPE);
+	adev->umsch_mm.vpe_hqd_mask = 0xfe;
+
+	r = amdgpu_device_wb_get(adev, &adev->umsch_mm.wb_index);
+	if (r) {
+		dev_err(adev->dev, "failed to alloc wb for umsch: %d\n", r);
+		return r;
+	}
+
+	adev->umsch_mm.sch_ctx_gpu_addr = adev->wb.gpu_addr +
+					  (adev->umsch_mm.wb_index * 4);
+
+	r = amdgpu_bo_create_kernel(adev, PAGE_SIZE, PAGE_SIZE,
+				    AMDGPU_GEM_DOMAIN_GTT,
+				    &adev->umsch_mm.cmd_buf_obj,
+				    &adev->umsch_mm.cmd_buf_gpu_addr,
+				    (void **)&adev->umsch_mm.cmd_buf_ptr);
+	if (r) {
+		dev_err(adev->dev, "failed to allocate cmdbuf bo %d\n", r);
+		amdgpu_device_wb_free(adev, adev->umsch_mm.wb_index);
+		return r;
+	}
+
+	r = amdgpu_bo_create_kernel(adev, AMDGPU_UMSCHFW_LOG_SIZE, PAGE_SIZE,
+				    AMDGPU_GEM_DOMAIN_VRAM |
+				    AMDGPU_GEM_DOMAIN_GTT,
+				    &adev->umsch_mm.dbglog_bo,
+				    &adev->umsch_mm.log_gpu_addr,
+				    &adev->umsch_mm.log_cpu_addr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate umsch debug bo\n", r);
+		return r;
+	}
+
+	mutex_init(&adev->umsch_mm.mutex_hidden);
+
+	umsch_mm_agdb_index_init(adev);
+
+	return 0;
+}
+
+
+static int umsch_mm_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	switch (amdgpu_ip_version(adev, VCN_HWIP, 0)) {
+	case IP_VERSION(4, 0, 5):
+	case IP_VERSION(4, 0, 6):
+		umsch_mm_v4_0_set_funcs(&adev->umsch_mm);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	adev->umsch_mm.ring.funcs = &umsch_v4_0_ring_funcs;
+	umsch_mm_set_regs(&adev->umsch_mm);
+
+	return 0;
+}
+
+static int umsch_mm_late_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_in_reset(adev) || adev->in_s0ix || adev->in_suspend)
+		return 0;
+
+	return 0;
+}
+
+static int umsch_mm_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = umsch_mm_init(adev);
+	if (r)
+		return r;
+
+	amdgpu_umsch_fwlog_init(&adev->umsch_mm);
+	r = umsch_mm_ring_init(&adev->umsch_mm);
+	if (r)
+		return r;
+
+	r = umsch_mm_init_microcode(&adev->umsch_mm);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static int umsch_mm_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	release_firmware(adev->umsch_mm.fw);
+	adev->umsch_mm.fw = NULL;
+
+	amdgpu_ring_fini(&adev->umsch_mm.ring);
+
+	mutex_destroy(&adev->umsch_mm.mutex_hidden);
+
+	amdgpu_bo_free_kernel(&adev->umsch_mm.cmd_buf_obj,
+			      &adev->umsch_mm.cmd_buf_gpu_addr,
+			      (void **)&adev->umsch_mm.cmd_buf_ptr);
+
+	amdgpu_bo_free_kernel(&adev->umsch_mm.dbglog_bo,
+				    &adev->umsch_mm.log_gpu_addr,
+				    (void **)&adev->umsch_mm.log_cpu_addr);
+
+	amdgpu_device_wb_free(adev, adev->umsch_mm.wb_index);
+
+	return 0;
+}
+
+static int umsch_mm_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = umsch_mm_load_microcode(&adev->umsch_mm);
+	if (r)
+		return r;
+
+	umsch_mm_ring_start(&adev->umsch_mm);
+
+	r = umsch_mm_set_hw_resources(&adev->umsch_mm);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static int umsch_mm_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	umsch_mm_ring_stop(&adev->umsch_mm);
+
+	amdgpu_bo_free_kernel(&adev->umsch_mm.data_fw_obj,
+			      &adev->umsch_mm.data_fw_gpu_addr,
+			      (void **)&adev->umsch_mm.data_fw_ptr);
+
+	amdgpu_bo_free_kernel(&adev->umsch_mm.ucode_fw_obj,
+			      &adev->umsch_mm.ucode_fw_gpu_addr,
+			      (void **)&adev->umsch_mm.ucode_fw_ptr);
+	return 0;
+}
+
+static int umsch_mm_suspend(struct amdgpu_ip_block *ip_block)
+{
+	return umsch_mm_hw_fini(ip_block);
+}
+
+static int umsch_mm_resume(struct amdgpu_ip_block *ip_block)
+{
+	return umsch_mm_hw_init(ip_block);
+}
+
+void amdgpu_umsch_fwlog_init(struct amdgpu_umsch_mm *umsch_mm)
+{
+#if defined(CONFIG_DEBUG_FS)
+	void *fw_log_cpu_addr = umsch_mm->log_cpu_addr;
+	volatile struct amdgpu_umsch_fwlog *log_buf = fw_log_cpu_addr;
+
+	log_buf->header_size = sizeof(struct amdgpu_umsch_fwlog);
+	log_buf->buffer_size = AMDGPU_UMSCHFW_LOG_SIZE;
+	log_buf->rptr = log_buf->header_size;
+	log_buf->wptr = log_buf->header_size;
+	log_buf->wrapped = 0;
+#endif
+}
+
+/*
+ * debugfs for mapping umsch firmware log buffer.
+ */
+#if defined(CONFIG_DEBUG_FS)
+static ssize_t amdgpu_debugfs_umsch_fwlog_read(struct file *f, char __user *buf,
+					     size_t size, loff_t *pos)
+{
+	struct amdgpu_umsch_mm *umsch_mm;
+	void *log_buf;
+	volatile struct amdgpu_umsch_fwlog *plog;
+	unsigned int read_pos, write_pos, available, i, read_bytes = 0;
+	unsigned int read_num[2] = {0};
+
+	umsch_mm = file_inode(f)->i_private;
+	if (!umsch_mm)
+		return -ENODEV;
+
+	if (!umsch_mm->log_cpu_addr)
+		return -EFAULT;
+
+	log_buf = umsch_mm->log_cpu_addr;
+
+	plog = (volatile struct amdgpu_umsch_fwlog *)log_buf;
+	read_pos = plog->rptr;
+	write_pos = plog->wptr;
+
+	if (read_pos > AMDGPU_UMSCHFW_LOG_SIZE || write_pos > AMDGPU_UMSCHFW_LOG_SIZE)
+		return -EFAULT;
+
+	if (!size || (read_pos == write_pos))
+		return 0;
+
+	if (write_pos > read_pos) {
+		available = write_pos - read_pos;
+		read_num[0] = min_t(size_t, size, available);
+	} else {
+		read_num[0] = AMDGPU_UMSCHFW_LOG_SIZE - read_pos;
+		available = read_num[0] + write_pos - plog->header_size;
+		if (size > available)
+			read_num[1] = write_pos - plog->header_size;
+		else if (size > read_num[0])
+			read_num[1] = size - read_num[0];
+		else
+			read_num[0] = size;
+	}
+
+	for (i = 0; i < 2; i++) {
+		if (read_num[i]) {
+			if (read_pos == AMDGPU_UMSCHFW_LOG_SIZE)
+				read_pos = plog->header_size;
+			if (read_num[i] == copy_to_user((buf + read_bytes),
+							(log_buf + read_pos), read_num[i]))
+				return -EFAULT;
+
+			read_bytes += read_num[i];
+			read_pos += read_num[i];
+		}
+	}
+
+	plog->rptr = read_pos;
+	*pos += read_bytes;
+	return read_bytes;
+}
+
+static const struct file_operations amdgpu_debugfs_umschfwlog_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_umsch_fwlog_read,
+	.llseek = default_llseek
+};
+#endif
+
+void amdgpu_debugfs_umsch_fwlog_init(struct amdgpu_device *adev,
+			struct amdgpu_umsch_mm *umsch_mm)
+{
+#if defined(CONFIG_DEBUG_FS)
+	struct drm_minor *minor = adev_to_drm(adev)->primary;
+	struct dentry *root = minor->debugfs_root;
+	char name[32];
+
+	sprintf(name, "amdgpu_umsch_fwlog");
+	debugfs_create_file_size(name, S_IFREG | 0444, root, umsch_mm,
+				 &amdgpu_debugfs_umschfwlog_fops,
+				 AMDGPU_UMSCHFW_LOG_SIZE);
+#endif
+}
+
+static const struct amd_ip_funcs umsch_mm_v4_0_ip_funcs = {
+	.name = "umsch_mm_v4_0",
+	.early_init = umsch_mm_early_init,
+	.late_init = umsch_mm_late_init,
+	.sw_init = umsch_mm_sw_init,
+	.sw_fini = umsch_mm_sw_fini,
+	.hw_init = umsch_mm_hw_init,
+	.hw_fini = umsch_mm_hw_fini,
+	.suspend = umsch_mm_suspend,
+	.resume = umsch_mm_resume,
+};
+
+const struct amdgpu_ip_block_version umsch_mm_v4_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_UMSCH_MM,
+	.major = 4,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &umsch_mm_v4_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.h
new file mode 100644
index 000000000000..2c771a753778
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.h
@@ -0,0 +1,246 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_UMSCH_MM_H__
+#define __AMDGPU_UMSCH_MM_H__
+
+enum UMSCH_SWIP_ENGINE_TYPE {
+	UMSCH_SWIP_ENGINE_TYPE_VCN0 = 0,
+	UMSCH_SWIP_ENGINE_TYPE_VCN1 = 1,
+	UMSCH_SWIP_ENGINE_TYPE_VCN = 2,
+	UMSCH_SWIP_ENGINE_TYPE_VPE = 3,
+	UMSCH_SWIP_ENGINE_TYPE_MAX
+};
+
+enum UMSCH_CONTEXT_PRIORITY_LEVEL {
+	CONTEXT_PRIORITY_LEVEL_IDLE = 0,
+	CONTEXT_PRIORITY_LEVEL_NORMAL = 1,
+	CONTEXT_PRIORITY_LEVEL_FOCUS = 2,
+	CONTEXT_PRIORITY_LEVEL_REALTIME = 3,
+	CONTEXT_PRIORITY_NUM_LEVELS
+};
+
+struct umsch_mm_set_resource_input {
+	uint32_t vmid_mask_mm_vcn;
+	uint32_t vmid_mask_mm_vpe;
+	uint32_t collaboration_mask_vpe;
+	uint32_t logging_vmid;
+	uint32_t engine_mask;
+	union {
+		struct {
+			uint32_t disable_reset : 1;
+			uint32_t disable_umsch_mm_log : 1;
+			uint32_t use_rs64mem_for_proc_ctx_csa : 1;
+			uint32_t reserved : 29;
+		};
+		uint32_t uint32_all;
+	};
+};
+
+struct amdgpu_umsch_fwlog {
+	uint32_t rptr;
+	uint32_t wptr;
+	uint32_t buffer_size;
+	uint32_t header_size;
+	uint32_t wrapped;
+};
+
+struct umsch_mm_add_queue_input {
+	uint32_t process_id;
+	uint64_t page_table_base_addr;
+	uint64_t process_va_start;
+	uint64_t process_va_end;
+	uint64_t process_quantum;
+	uint64_t process_csa_addr;
+	uint64_t context_quantum;
+	uint64_t context_csa_addr;
+	uint32_t inprocess_context_priority;
+	enum UMSCH_CONTEXT_PRIORITY_LEVEL context_global_priority_level;
+	uint32_t doorbell_offset_0;
+	uint32_t doorbell_offset_1;
+	enum UMSCH_SWIP_ENGINE_TYPE engine_type;
+	uint32_t affinity;
+	uint64_t mqd_addr;
+	uint64_t h_context;
+	uint64_t h_queue;
+	uint32_t vm_context_cntl;
+
+	uint32_t process_csa_array_index;
+	uint32_t context_csa_array_index;
+
+	struct {
+		uint32_t is_context_suspended : 1;
+		uint32_t collaboration_mode : 1;
+		uint32_t reserved : 30;
+	};
+};
+
+struct umsch_mm_remove_queue_input {
+	uint32_t doorbell_offset_0;
+	uint32_t doorbell_offset_1;
+	uint64_t context_csa_addr;
+	uint32_t context_csa_array_index;
+};
+
+struct MQD_INFO {
+	uint32_t rb_base_hi;
+	uint32_t rb_base_lo;
+	uint32_t rb_size;
+	uint32_t wptr_val;
+	uint32_t rptr_val;
+	uint32_t unmapped;
+	uint32_t vmid;
+};
+
+struct amdgpu_umsch_mm;
+
+struct umsch_mm_funcs {
+	int (*set_hw_resources)(struct amdgpu_umsch_mm *umsch);
+	int (*add_queue)(struct amdgpu_umsch_mm *umsch,
+			 struct umsch_mm_add_queue_input *input);
+	int (*remove_queue)(struct amdgpu_umsch_mm *umsch,
+			    struct umsch_mm_remove_queue_input *input);
+	int (*set_regs)(struct amdgpu_umsch_mm *umsch);
+	int (*init_microcode)(struct amdgpu_umsch_mm *umsch);
+	int (*load_microcode)(struct amdgpu_umsch_mm *umsch);
+	int (*ring_init)(struct amdgpu_umsch_mm *umsch);
+	int (*ring_start)(struct amdgpu_umsch_mm *umsch);
+	int (*ring_stop)(struct amdgpu_umsch_mm *umsch);
+	int (*ring_fini)(struct amdgpu_umsch_mm *umsch);
+};
+
+struct amdgpu_umsch_mm {
+	struct amdgpu_ring		ring;
+
+	uint32_t			rb_wptr;
+	uint32_t			rb_rptr;
+
+	const struct umsch_mm_funcs	*funcs;
+
+	const struct firmware		*fw;
+	uint32_t			fw_version;
+	uint32_t			feature_version;
+
+	struct amdgpu_bo		*ucode_fw_obj;
+	uint64_t			ucode_fw_gpu_addr;
+	uint32_t			*ucode_fw_ptr;
+	uint64_t			irq_start_addr;
+	uint64_t			uc_start_addr;
+	uint32_t			ucode_size;
+
+	struct amdgpu_bo		*data_fw_obj;
+	uint64_t			data_fw_gpu_addr;
+	uint32_t			*data_fw_ptr;
+	uint64_t			data_start_addr;
+	uint32_t			data_size;
+
+	struct amdgpu_bo		*cmd_buf_obj;
+	uint64_t			cmd_buf_gpu_addr;
+	uint32_t			*cmd_buf_ptr;
+	uint32_t			*cmd_buf_curr_ptr;
+
+	uint32_t			wb_index;
+	uint64_t			sch_ctx_gpu_addr;
+	uint32_t			*sch_ctx_cpu_addr;
+
+	uint32_t			vmid_mask_mm_vcn;
+	uint32_t			vmid_mask_mm_vpe;
+	uint32_t			engine_mask;
+	uint32_t			vcn0_hqd_mask;
+	uint32_t			vcn1_hqd_mask;
+	uint32_t			vcn_hqd_mask[2];
+	uint32_t			vpe_hqd_mask;
+	uint32_t			agdb_index[CONTEXT_PRIORITY_NUM_LEVELS];
+
+	struct mutex			mutex_hidden;
+	struct amdgpu_bo		*dbglog_bo;
+	void				*log_cpu_addr;
+	uint64_t			log_gpu_addr;
+	uint32_t			mem_size;
+	uint32_t			log_offset;
+};
+
+int amdgpu_umsch_mm_submit_pkt(struct amdgpu_umsch_mm *umsch, void *pkt, int ndws);
+int amdgpu_umsch_mm_query_fence(struct amdgpu_umsch_mm *umsch);
+
+int amdgpu_umsch_mm_init_microcode(struct amdgpu_umsch_mm *umsch);
+int amdgpu_umsch_mm_allocate_ucode_buffer(struct amdgpu_umsch_mm *umsch);
+int amdgpu_umsch_mm_allocate_ucode_data_buffer(struct amdgpu_umsch_mm *umsch);
+
+int amdgpu_umsch_mm_psp_execute_cmd_buf(struct amdgpu_umsch_mm *umsch);
+
+int amdgpu_umsch_mm_ring_init(struct amdgpu_umsch_mm *umsch);
+
+void amdgpu_debugfs_umsch_fwlog_init(struct amdgpu_device *adev,
+			struct amdgpu_umsch_mm *umsch);
+
+void amdgpu_umsch_fwlog_init(struct amdgpu_umsch_mm *umsch_mm);
+
+#define WREG32_SOC15_UMSCH(reg, value)								\
+	do {											\
+		uint32_t reg_offset = adev->reg_offset[VCN_HWIP][0][reg##_BASE_IDX] + reg;	\
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {				\
+			*adev->umsch_mm.cmd_buf_curr_ptr++ = (reg_offset << 2);			\
+			*adev->umsch_mm.cmd_buf_curr_ptr++ = value;				\
+		} else	{									\
+			WREG32(reg_offset, value);						\
+		}										\
+	} while (0)
+
+#define umsch_mm_set_hw_resources(umsch) \
+	((umsch)->funcs->set_hw_resources ? (umsch)->funcs->set_hw_resources((umsch)) : 0)
+#define umsch_mm_add_queue(umsch, input) \
+	((umsch)->funcs->add_queue ? (umsch)->funcs->add_queue((umsch), (input)) : 0)
+#define umsch_mm_remove_queue(umsch, input) \
+	((umsch)->funcs->remove_queue ? (umsch)->funcs->remove_queue((umsch), (input)) : 0)
+
+#define umsch_mm_set_regs(umsch) \
+	((umsch)->funcs->set_regs ? (umsch)->funcs->set_regs((umsch)) : 0)
+#define umsch_mm_init_microcode(umsch) \
+	((umsch)->funcs->init_microcode ? (umsch)->funcs->init_microcode((umsch)) : 0)
+#define umsch_mm_load_microcode(umsch) \
+	((umsch)->funcs->load_microcode ? (umsch)->funcs->load_microcode((umsch)) : 0)
+
+#define umsch_mm_ring_init(umsch) \
+	((umsch)->funcs->ring_init ? (umsch)->funcs->ring_init((umsch)) : 0)
+#define umsch_mm_ring_start(umsch) \
+	((umsch)->funcs->ring_start ? (umsch)->funcs->ring_start((umsch)) : 0)
+#define umsch_mm_ring_stop(umsch) \
+	((umsch)->funcs->ring_stop ? (umsch)->funcs->ring_stop((umsch)) : 0)
+#define umsch_mm_ring_fini(umsch) \
+	((umsch)->funcs->ring_fini ? (umsch)->funcs->ring_fini((umsch)) : 0)
+
+static inline void amdgpu_umsch_mm_lock(struct amdgpu_umsch_mm *umsch)
+{
+	mutex_lock(&umsch->mutex_hidden);
+}
+
+static inline void amdgpu_umsch_mm_unlock(struct amdgpu_umsch_mm *umsch)
+{
+	mutex_unlock(&umsch->mutex_hidden);
+}
+
+extern const struct amdgpu_ip_block_version umsch_mm_v4_0_ip_block;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c
new file mode 100644
index 000000000000..1add21160d21
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c
@@ -0,0 +1,1090 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <drm/drm_auth.h>
+#include <drm/drm_exec.h>
+#include <linux/pm_runtime.h>
+
+#include "amdgpu.h"
+#include "amdgpu_vm.h"
+#include "amdgpu_userq.h"
+#include "amdgpu_userq_fence.h"
+
+u32 amdgpu_userq_get_supported_ip_mask(struct amdgpu_device *adev)
+{
+	int i;
+	u32 userq_ip_mask = 0;
+
+	for (i = 0; i < AMDGPU_HW_IP_NUM; i++) {
+		if (adev->userq_funcs[i])
+			userq_ip_mask |= (1 << i);
+	}
+
+	return userq_ip_mask;
+}
+
+int amdgpu_userq_input_va_validate(struct amdgpu_vm *vm, u64 addr,
+				   u64 expected_size)
+{
+	struct amdgpu_bo_va_mapping *va_map;
+	u64 user_addr;
+	u64 size;
+	int r = 0;
+
+	user_addr = (addr & AMDGPU_GMC_HOLE_MASK) >> AMDGPU_GPU_PAGE_SHIFT;
+	size = expected_size >> AMDGPU_GPU_PAGE_SHIFT;
+
+	r = amdgpu_bo_reserve(vm->root.bo, false);
+	if (r)
+		return r;
+
+	va_map = amdgpu_vm_bo_lookup_mapping(vm, user_addr);
+	if (!va_map) {
+		r = -EINVAL;
+		goto out_err;
+	}
+	/* Only validate the userq whether resident in the VM mapping range */
+	if (user_addr >= va_map->start  &&
+	    va_map->last - user_addr + 1 >= size) {
+		amdgpu_bo_unreserve(vm->root.bo);
+		return 0;
+	}
+
+	r = -EINVAL;
+out_err:
+	amdgpu_bo_unreserve(vm->root.bo);
+	return r;
+}
+
+static int
+amdgpu_userq_preempt_helper(struct amdgpu_userq_mgr *uq_mgr,
+			  struct amdgpu_usermode_queue *queue)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	const struct amdgpu_userq_funcs *userq_funcs =
+		adev->userq_funcs[queue->queue_type];
+	int r = 0;
+
+	if (queue->state == AMDGPU_USERQ_STATE_MAPPED) {
+		r = userq_funcs->preempt(uq_mgr, queue);
+		if (r) {
+			queue->state = AMDGPU_USERQ_STATE_HUNG;
+		} else {
+			queue->state = AMDGPU_USERQ_STATE_PREEMPTED;
+		}
+	}
+
+	return r;
+}
+
+static int
+amdgpu_userq_restore_helper(struct amdgpu_userq_mgr *uq_mgr,
+			struct amdgpu_usermode_queue *queue)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	const struct amdgpu_userq_funcs *userq_funcs =
+		adev->userq_funcs[queue->queue_type];
+	int r = 0;
+
+	if (queue->state == AMDGPU_USERQ_STATE_PREEMPTED) {
+		r = userq_funcs->restore(uq_mgr, queue);
+		if (r) {
+			queue->state = AMDGPU_USERQ_STATE_HUNG;
+		} else {
+			queue->state = AMDGPU_USERQ_STATE_MAPPED;
+		}
+	}
+
+	return r;
+}
+
+static int
+amdgpu_userq_unmap_helper(struct amdgpu_userq_mgr *uq_mgr,
+			  struct amdgpu_usermode_queue *queue)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	const struct amdgpu_userq_funcs *userq_funcs =
+		adev->userq_funcs[queue->queue_type];
+	int r = 0;
+
+	if ((queue->state == AMDGPU_USERQ_STATE_MAPPED) ||
+		(queue->state == AMDGPU_USERQ_STATE_PREEMPTED)) {
+		r = userq_funcs->unmap(uq_mgr, queue);
+		if (r)
+			queue->state = AMDGPU_USERQ_STATE_HUNG;
+		else
+			queue->state = AMDGPU_USERQ_STATE_UNMAPPED;
+	}
+	return r;
+}
+
+static int
+amdgpu_userq_map_helper(struct amdgpu_userq_mgr *uq_mgr,
+			struct amdgpu_usermode_queue *queue)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	const struct amdgpu_userq_funcs *userq_funcs =
+		adev->userq_funcs[queue->queue_type];
+	int r = 0;
+
+	if (queue->state == AMDGPU_USERQ_STATE_UNMAPPED) {
+		r = userq_funcs->map(uq_mgr, queue);
+		if (r) {
+			queue->state = AMDGPU_USERQ_STATE_HUNG;
+		} else {
+			queue->state = AMDGPU_USERQ_STATE_MAPPED;
+		}
+	}
+	return r;
+}
+
+static void
+amdgpu_userq_wait_for_last_fence(struct amdgpu_userq_mgr *uq_mgr,
+				 struct amdgpu_usermode_queue *queue)
+{
+	struct dma_fence *f = queue->last_fence;
+	int ret;
+
+	if (f && !dma_fence_is_signaled(f)) {
+		ret = dma_fence_wait_timeout(f, true, msecs_to_jiffies(100));
+		if (ret <= 0)
+			drm_file_err(uq_mgr->file, "Timed out waiting for fence=%llu:%llu\n",
+				     f->context, f->seqno);
+	}
+}
+
+static void
+amdgpu_userq_cleanup(struct amdgpu_userq_mgr *uq_mgr,
+		     struct amdgpu_usermode_queue *queue,
+		     int queue_id)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	const struct amdgpu_userq_funcs *uq_funcs = adev->userq_funcs[queue->queue_type];
+
+	uq_funcs->mqd_destroy(uq_mgr, queue);
+	amdgpu_userq_fence_driver_free(queue);
+	idr_remove(&uq_mgr->userq_idr, queue_id);
+	kfree(queue);
+}
+
+static struct amdgpu_usermode_queue *
+amdgpu_userq_find(struct amdgpu_userq_mgr *uq_mgr, int qid)
+{
+	return idr_find(&uq_mgr->userq_idr, qid);
+}
+
+void
+amdgpu_userq_ensure_ev_fence(struct amdgpu_userq_mgr *uq_mgr,
+			     struct amdgpu_eviction_fence_mgr *evf_mgr)
+{
+	struct amdgpu_eviction_fence *ev_fence;
+
+retry:
+	/* Flush any pending resume work to create ev_fence */
+	flush_delayed_work(&uq_mgr->resume_work);
+
+	mutex_lock(&uq_mgr->userq_mutex);
+	spin_lock(&evf_mgr->ev_fence_lock);
+	ev_fence = evf_mgr->ev_fence;
+	spin_unlock(&evf_mgr->ev_fence_lock);
+	if (!ev_fence || dma_fence_is_signaled(&ev_fence->base)) {
+		mutex_unlock(&uq_mgr->userq_mutex);
+		/*
+		 * Looks like there was no pending resume work,
+		 * add one now to create a valid eviction fence
+		 */
+		schedule_delayed_work(&uq_mgr->resume_work, 0);
+		goto retry;
+	}
+}
+
+int amdgpu_userq_create_object(struct amdgpu_userq_mgr *uq_mgr,
+			       struct amdgpu_userq_obj *userq_obj,
+			       int size)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	struct amdgpu_bo_param bp;
+	int r;
+
+	memset(&bp, 0, sizeof(bp));
+	bp.byte_align = PAGE_SIZE;
+	bp.domain = AMDGPU_GEM_DOMAIN_GTT;
+	bp.flags = AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS |
+		   AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
+	bp.type = ttm_bo_type_kernel;
+	bp.size = size;
+	bp.resv = NULL;
+	bp.bo_ptr_size = sizeof(struct amdgpu_bo);
+
+	r = amdgpu_bo_create(adev, &bp, &userq_obj->obj);
+	if (r) {
+		drm_file_err(uq_mgr->file, "Failed to allocate BO for userqueue (%d)", r);
+		return r;
+	}
+
+	r = amdgpu_bo_reserve(userq_obj->obj, true);
+	if (r) {
+		drm_file_err(uq_mgr->file, "Failed to reserve BO to map (%d)", r);
+		goto free_obj;
+	}
+
+	r = amdgpu_ttm_alloc_gart(&(userq_obj->obj)->tbo);
+	if (r) {
+		drm_file_err(uq_mgr->file, "Failed to alloc GART for userqueue object (%d)", r);
+		goto unresv;
+	}
+
+	r = amdgpu_bo_kmap(userq_obj->obj, &userq_obj->cpu_ptr);
+	if (r) {
+		drm_file_err(uq_mgr->file, "Failed to map BO for userqueue (%d)", r);
+		goto unresv;
+	}
+
+	userq_obj->gpu_addr = amdgpu_bo_gpu_offset(userq_obj->obj);
+	amdgpu_bo_unreserve(userq_obj->obj);
+	memset(userq_obj->cpu_ptr, 0, size);
+	return 0;
+
+unresv:
+	amdgpu_bo_unreserve(userq_obj->obj);
+
+free_obj:
+	amdgpu_bo_unref(&userq_obj->obj);
+	return r;
+}
+
+void amdgpu_userq_destroy_object(struct amdgpu_userq_mgr *uq_mgr,
+				 struct amdgpu_userq_obj *userq_obj)
+{
+	amdgpu_bo_kunmap(userq_obj->obj);
+	amdgpu_bo_unref(&userq_obj->obj);
+}
+
+uint64_t
+amdgpu_userq_get_doorbell_index(struct amdgpu_userq_mgr *uq_mgr,
+				struct amdgpu_db_info *db_info,
+				struct drm_file *filp)
+{
+	uint64_t index;
+	struct drm_gem_object *gobj;
+	struct amdgpu_userq_obj *db_obj = db_info->db_obj;
+	int r, db_size;
+
+	gobj = drm_gem_object_lookup(filp, db_info->doorbell_handle);
+	if (gobj == NULL) {
+		drm_file_err(uq_mgr->file, "Can't find GEM object for doorbell\n");
+		return -EINVAL;
+	}
+
+	db_obj->obj = amdgpu_bo_ref(gem_to_amdgpu_bo(gobj));
+	drm_gem_object_put(gobj);
+
+	r = amdgpu_bo_reserve(db_obj->obj, true);
+	if (r) {
+		drm_file_err(uq_mgr->file, "[Usermode queues] Failed to pin doorbell object\n");
+		goto unref_bo;
+	}
+
+	/* Pin the BO before generating the index, unpin in queue destroy */
+	r = amdgpu_bo_pin(db_obj->obj, AMDGPU_GEM_DOMAIN_DOORBELL);
+	if (r) {
+		drm_file_err(uq_mgr->file, "[Usermode queues] Failed to pin doorbell object\n");
+		goto unresv_bo;
+	}
+
+	switch (db_info->queue_type) {
+	case AMDGPU_HW_IP_GFX:
+	case AMDGPU_HW_IP_COMPUTE:
+	case AMDGPU_HW_IP_DMA:
+		db_size = sizeof(u64);
+		break;
+
+	case AMDGPU_HW_IP_VCN_ENC:
+		db_size = sizeof(u32);
+		db_info->doorbell_offset += AMDGPU_NAVI10_DOORBELL64_VCN0_1 << 1;
+		break;
+
+	case AMDGPU_HW_IP_VPE:
+		db_size = sizeof(u32);
+		db_info->doorbell_offset += AMDGPU_NAVI10_DOORBELL64_VPE << 1;
+		break;
+
+	default:
+		drm_file_err(uq_mgr->file, "[Usermode queues] IP %d not support\n",
+			     db_info->queue_type);
+		r = -EINVAL;
+		goto unpin_bo;
+	}
+
+	index = amdgpu_doorbell_index_on_bar(uq_mgr->adev, db_obj->obj,
+					     db_info->doorbell_offset, db_size);
+	drm_dbg_driver(adev_to_drm(uq_mgr->adev),
+		       "[Usermode queues] doorbell index=%lld\n", index);
+	amdgpu_bo_unreserve(db_obj->obj);
+	return index;
+
+unpin_bo:
+	amdgpu_bo_unpin(db_obj->obj);
+unresv_bo:
+	amdgpu_bo_unreserve(db_obj->obj);
+unref_bo:
+	amdgpu_bo_unref(&db_obj->obj);
+	return r;
+}
+
+static int
+amdgpu_userq_destroy(struct drm_file *filp, int queue_id)
+{
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+	struct amdgpu_userq_mgr *uq_mgr = &fpriv->userq_mgr;
+	struct amdgpu_device *adev = uq_mgr->adev;
+	struct amdgpu_usermode_queue *queue;
+	int r = 0;
+
+	cancel_delayed_work_sync(&uq_mgr->resume_work);
+	mutex_lock(&uq_mgr->userq_mutex);
+
+	queue = amdgpu_userq_find(uq_mgr, queue_id);
+	if (!queue) {
+		drm_dbg_driver(adev_to_drm(uq_mgr->adev), "Invalid queue id to destroy\n");
+		mutex_unlock(&uq_mgr->userq_mutex);
+		return -EINVAL;
+	}
+	amdgpu_userq_wait_for_last_fence(uq_mgr, queue);
+	r = amdgpu_bo_reserve(queue->db_obj.obj, true);
+	if (!r) {
+		amdgpu_bo_unpin(queue->db_obj.obj);
+		amdgpu_bo_unreserve(queue->db_obj.obj);
+	}
+	amdgpu_bo_unref(&queue->db_obj.obj);
+
+#if defined(CONFIG_DEBUG_FS)
+	debugfs_remove_recursive(queue->debugfs_queue);
+#endif
+	r = amdgpu_userq_unmap_helper(uq_mgr, queue);
+	/*TODO: It requires a reset for userq hw unmap error*/
+	if (unlikely(r != AMDGPU_USERQ_STATE_UNMAPPED)) {
+		drm_warn(adev_to_drm(uq_mgr->adev), "trying to destroy a HW mapping userq\n");
+		queue->state = AMDGPU_USERQ_STATE_HUNG;
+	}
+	amdgpu_userq_cleanup(uq_mgr, queue, queue_id);
+	mutex_unlock(&uq_mgr->userq_mutex);
+
+	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
+	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+
+	return r;
+}
+
+static int amdgpu_userq_priority_permit(struct drm_file *filp,
+					int priority)
+{
+	if (priority < AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_HIGH)
+		return 0;
+
+	if (capable(CAP_SYS_NICE))
+		return 0;
+
+	if (drm_is_current_master(filp))
+		return 0;
+
+	return -EACCES;
+}
+
+#if defined(CONFIG_DEBUG_FS)
+static int amdgpu_mqd_info_read(struct seq_file *m, void *unused)
+{
+	struct amdgpu_usermode_queue *queue = m->private;
+	struct amdgpu_bo *bo;
+	int r;
+
+	if (!queue || !queue->mqd.obj)
+		return -EINVAL;
+
+	bo = amdgpu_bo_ref(queue->mqd.obj);
+	r = amdgpu_bo_reserve(bo, true);
+	if (r) {
+		amdgpu_bo_unref(&bo);
+		return -EINVAL;
+	}
+
+	seq_printf(m, "queue_type: %d\n", queue->queue_type);
+	seq_printf(m, "mqd_gpu_address: 0x%llx\n", amdgpu_bo_gpu_offset(queue->mqd.obj));
+
+	amdgpu_bo_unreserve(bo);
+	amdgpu_bo_unref(&bo);
+
+	return 0;
+}
+
+static int amdgpu_mqd_info_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, amdgpu_mqd_info_read, inode->i_private);
+}
+
+static const struct file_operations amdgpu_mqd_info_fops = {
+	.owner = THIS_MODULE,
+	.open = amdgpu_mqd_info_open,
+	.read = seq_read,
+	.llseek = seq_lseek,
+	.release = single_release,
+};
+#endif
+
+static int
+amdgpu_userq_create(struct drm_file *filp, union drm_amdgpu_userq *args)
+{
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+	struct amdgpu_userq_mgr *uq_mgr = &fpriv->userq_mgr;
+	struct amdgpu_device *adev = uq_mgr->adev;
+	const struct amdgpu_userq_funcs *uq_funcs;
+	struct amdgpu_usermode_queue *queue;
+	struct amdgpu_db_info db_info;
+	char *queue_name;
+	bool skip_map_queue;
+	uint64_t index;
+	int qid, r = 0;
+	int priority =
+		(args->in.flags & AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_MASK) >>
+		AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_SHIFT;
+
+	r = amdgpu_userq_priority_permit(filp, priority);
+	if (r)
+		return r;
+
+	r = pm_runtime_get_sync(adev_to_drm(adev)->dev);
+	if (r < 0) {
+		drm_file_err(uq_mgr->file, "pm_runtime_get_sync() failed for userqueue create\n");
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	/*
+	 * There could be a situation that we are creating a new queue while
+	 * the other queues under this UQ_mgr are suspended. So if there is any
+	 * resume work pending, wait for it to get done.
+	 *
+	 * This will also make sure we have a valid eviction fence ready to be used.
+	 */
+	mutex_lock(&adev->userq_mutex);
+	amdgpu_userq_ensure_ev_fence(&fpriv->userq_mgr, &fpriv->evf_mgr);
+
+	uq_funcs = adev->userq_funcs[args->in.ip_type];
+	if (!uq_funcs) {
+		drm_file_err(uq_mgr->file, "Usermode queue is not supported for this IP (%u)\n",
+			     args->in.ip_type);
+		r = -EINVAL;
+		goto unlock;
+	}
+
+	queue = kzalloc(sizeof(struct amdgpu_usermode_queue), GFP_KERNEL);
+	if (!queue) {
+		drm_file_err(uq_mgr->file, "Failed to allocate memory for queue\n");
+		r = -ENOMEM;
+		goto unlock;
+	}
+
+	/* Validate the userq virtual address.*/
+	if (amdgpu_userq_input_va_validate(&fpriv->vm, args->in.queue_va, args->in.queue_size) ||
+	    amdgpu_userq_input_va_validate(&fpriv->vm, args->in.rptr_va, AMDGPU_GPU_PAGE_SIZE) ||
+	    amdgpu_userq_input_va_validate(&fpriv->vm, args->in.wptr_va, AMDGPU_GPU_PAGE_SIZE)) {
+		r = -EINVAL;
+		kfree(queue);
+		goto unlock;
+	}
+	queue->doorbell_handle = args->in.doorbell_handle;
+	queue->queue_type = args->in.ip_type;
+	queue->vm = &fpriv->vm;
+	queue->priority = priority;
+
+	db_info.queue_type = queue->queue_type;
+	db_info.doorbell_handle = queue->doorbell_handle;
+	db_info.db_obj = &queue->db_obj;
+	db_info.doorbell_offset = args->in.doorbell_offset;
+
+	/* Convert relative doorbell offset into absolute doorbell index */
+	index = amdgpu_userq_get_doorbell_index(uq_mgr, &db_info, filp);
+	if (index == (uint64_t)-EINVAL) {
+		drm_file_err(uq_mgr->file, "Failed to get doorbell for queue\n");
+		kfree(queue);
+		r = -EINVAL;
+		goto unlock;
+	}
+
+	queue->doorbell_index = index;
+	xa_init_flags(&queue->fence_drv_xa, XA_FLAGS_ALLOC);
+	r = amdgpu_userq_fence_driver_alloc(adev, queue);
+	if (r) {
+		drm_file_err(uq_mgr->file, "Failed to alloc fence driver\n");
+		goto unlock;
+	}
+
+	r = uq_funcs->mqd_create(uq_mgr, &args->in, queue);
+	if (r) {
+		drm_file_err(uq_mgr->file, "Failed to create Queue\n");
+		amdgpu_userq_fence_driver_free(queue);
+		kfree(queue);
+		goto unlock;
+	}
+
+
+	qid = idr_alloc(&uq_mgr->userq_idr, queue, 1, AMDGPU_MAX_USERQ_COUNT, GFP_KERNEL);
+	if (qid < 0) {
+		drm_file_err(uq_mgr->file, "Failed to allocate a queue id\n");
+		amdgpu_userq_fence_driver_free(queue);
+		uq_funcs->mqd_destroy(uq_mgr, queue);
+		kfree(queue);
+		r = -ENOMEM;
+		goto unlock;
+	}
+
+	/* don't map the queue if scheduling is halted */
+	if (adev->userq_halt_for_enforce_isolation &&
+	    ((queue->queue_type == AMDGPU_HW_IP_GFX) ||
+	     (queue->queue_type == AMDGPU_HW_IP_COMPUTE)))
+		skip_map_queue = true;
+	else
+		skip_map_queue = false;
+	if (!skip_map_queue) {
+		r = amdgpu_userq_map_helper(uq_mgr, queue);
+		if (r) {
+			drm_file_err(uq_mgr->file, "Failed to map Queue\n");
+			idr_remove(&uq_mgr->userq_idr, qid);
+			amdgpu_userq_fence_driver_free(queue);
+			uq_funcs->mqd_destroy(uq_mgr, queue);
+			kfree(queue);
+			goto unlock;
+		}
+	}
+
+	queue_name = kasprintf(GFP_KERNEL, "queue-%d", qid);
+	if (!queue_name) {
+		r = -ENOMEM;
+		goto unlock;
+	}
+
+#if defined(CONFIG_DEBUG_FS)
+	/* Queue dentry per client to hold MQD information   */
+	queue->debugfs_queue = debugfs_create_dir(queue_name, filp->debugfs_client);
+	debugfs_create_file("mqd_info", 0444, queue->debugfs_queue, queue, &amdgpu_mqd_info_fops);
+#endif
+	kfree(queue_name);
+
+	args->out.queue_id = qid;
+
+unlock:
+	mutex_unlock(&uq_mgr->userq_mutex);
+	mutex_unlock(&adev->userq_mutex);
+
+	return r;
+}
+
+static int amdgpu_userq_input_args_validate(struct drm_device *dev,
+					union drm_amdgpu_userq *args,
+					struct drm_file *filp)
+{
+	struct amdgpu_device *adev = drm_to_adev(dev);
+
+	switch (args->in.op) {
+	case AMDGPU_USERQ_OP_CREATE:
+		if (args->in.flags & ~(AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_MASK |
+				       AMDGPU_USERQ_CREATE_FLAGS_QUEUE_SECURE))
+			return -EINVAL;
+		/* Usermode queues are only supported for GFX IP as of now */
+		if (args->in.ip_type != AMDGPU_HW_IP_GFX &&
+		    args->in.ip_type != AMDGPU_HW_IP_DMA &&
+		    args->in.ip_type != AMDGPU_HW_IP_COMPUTE) {
+			drm_file_err(filp, "Usermode queue doesn't support IP type %u\n",
+				     args->in.ip_type);
+			return -EINVAL;
+		}
+
+		if ((args->in.flags & AMDGPU_USERQ_CREATE_FLAGS_QUEUE_SECURE) &&
+		    (args->in.ip_type != AMDGPU_HW_IP_GFX) &&
+		    (args->in.ip_type != AMDGPU_HW_IP_COMPUTE) &&
+		    !amdgpu_is_tmz(adev)) {
+			drm_file_err(filp, "Secure only supported on GFX/Compute queues\n");
+			return -EINVAL;
+		}
+
+		if (args->in.queue_va == AMDGPU_BO_INVALID_OFFSET ||
+		    args->in.queue_va == 0 ||
+		    args->in.queue_size == 0) {
+			drm_file_err(filp, "invalidate userq queue va or size\n");
+			return -EINVAL;
+		}
+		if (!args->in.wptr_va || !args->in.rptr_va) {
+			drm_file_err(filp, "invalidate userq queue rptr or wptr\n");
+			return -EINVAL;
+		}
+		break;
+	case AMDGPU_USERQ_OP_FREE:
+		if (args->in.ip_type ||
+		    args->in.doorbell_handle ||
+		    args->in.doorbell_offset ||
+		    args->in.flags ||
+		    args->in.queue_va ||
+		    args->in.queue_size ||
+		    args->in.rptr_va ||
+		    args->in.wptr_va ||
+		    args->in.mqd ||
+		    args->in.mqd_size)
+			return -EINVAL;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+int amdgpu_userq_ioctl(struct drm_device *dev, void *data,
+		       struct drm_file *filp)
+{
+	union drm_amdgpu_userq *args = data;
+	int r;
+
+	if (amdgpu_userq_input_args_validate(dev, args, filp) < 0)
+		return -EINVAL;
+
+	switch (args->in.op) {
+	case AMDGPU_USERQ_OP_CREATE:
+		r = amdgpu_userq_create(filp, args);
+		if (r)
+			drm_file_err(filp, "Failed to create usermode queue\n");
+		break;
+
+	case AMDGPU_USERQ_OP_FREE:
+		r = amdgpu_userq_destroy(filp, args->in.queue_id);
+		if (r)
+			drm_file_err(filp, "Failed to destroy usermode queue\n");
+		break;
+
+	default:
+		drm_dbg_driver(dev, "Invalid user queue op specified: %d\n", args->in.op);
+		return -EINVAL;
+	}
+
+	return r;
+}
+
+static int
+amdgpu_userq_restore_all(struct amdgpu_userq_mgr *uq_mgr)
+{
+	struct amdgpu_usermode_queue *queue;
+	int queue_id;
+	int ret = 0, r;
+
+	/* Resume all the queues for this process */
+	idr_for_each_entry(&uq_mgr->userq_idr, queue, queue_id) {
+		r = amdgpu_userq_restore_helper(uq_mgr, queue);
+		if (r)
+			ret = r;
+	}
+
+	if (ret)
+		drm_file_err(uq_mgr->file, "Failed to map all the queues\n");
+	return ret;
+}
+
+static int amdgpu_userq_validate_vm(void *param, struct amdgpu_bo *bo)
+{
+	struct ttm_operation_ctx ctx = { false, false };
+
+	amdgpu_bo_placement_from_domain(bo, bo->allowed_domains);
+	return ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+}
+
+/* Handle all BOs on the invalidated list, validate them and update the PTs */
+static int
+amdgpu_userq_bo_validate(struct amdgpu_device *adev, struct drm_exec *exec,
+			 struct amdgpu_vm *vm)
+{
+	struct ttm_operation_ctx ctx = { false, false };
+	struct amdgpu_bo_va *bo_va;
+	struct amdgpu_bo *bo;
+	int ret;
+
+	spin_lock(&vm->status_lock);
+	while (!list_empty(&vm->invalidated)) {
+		bo_va = list_first_entry(&vm->invalidated,
+					 struct amdgpu_bo_va,
+					 base.vm_status);
+		spin_unlock(&vm->status_lock);
+
+		bo = bo_va->base.bo;
+		ret = drm_exec_prepare_obj(exec, &bo->tbo.base, 2);
+		if (unlikely(ret))
+			return ret;
+
+		amdgpu_bo_placement_from_domain(bo, bo->allowed_domains);
+		ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+		if (ret)
+			return ret;
+
+		/* This moves the bo_va to the done list */
+		ret = amdgpu_vm_bo_update(adev, bo_va, false);
+		if (ret)
+			return ret;
+
+		spin_lock(&vm->status_lock);
+	}
+	spin_unlock(&vm->status_lock);
+
+	return 0;
+}
+
+/* Make sure the whole VM is ready to be used */
+static int
+amdgpu_userq_vm_validate(struct amdgpu_userq_mgr *uq_mgr)
+{
+	struct amdgpu_fpriv *fpriv = uq_mgr_to_fpriv(uq_mgr);
+	struct amdgpu_device *adev = uq_mgr->adev;
+	struct amdgpu_vm *vm = &fpriv->vm;
+	struct amdgpu_bo_va *bo_va;
+	struct drm_exec exec;
+	int ret;
+
+	drm_exec_init(&exec, DRM_EXEC_IGNORE_DUPLICATES, 0);
+	drm_exec_until_all_locked(&exec) {
+		ret = amdgpu_vm_lock_pd(vm, &exec, 1);
+		drm_exec_retry_on_contention(&exec);
+		if (unlikely(ret))
+			goto unlock_all;
+
+		ret = amdgpu_vm_lock_done_list(vm, &exec, 1);
+		drm_exec_retry_on_contention(&exec);
+		if (unlikely(ret))
+			goto unlock_all;
+
+		/* This validates PDs, PTs and per VM BOs */
+		ret = amdgpu_vm_validate(adev, vm, NULL,
+					 amdgpu_userq_validate_vm,
+					 NULL);
+		if (unlikely(ret))
+			goto unlock_all;
+
+		/* This locks and validates the remaining evicted BOs */
+		ret = amdgpu_userq_bo_validate(adev, &exec, vm);
+		drm_exec_retry_on_contention(&exec);
+		if (unlikely(ret))
+			goto unlock_all;
+	}
+
+	ret = amdgpu_vm_handle_moved(adev, vm, NULL);
+	if (ret)
+		goto unlock_all;
+
+	ret = amdgpu_vm_update_pdes(adev, vm, false);
+	if (ret)
+		goto unlock_all;
+
+	/*
+	 * We need to wait for all VM updates to finish before restarting the
+	 * queues. Using the done list like that is now ok since everything is
+	 * locked in place.
+	 */
+	list_for_each_entry(bo_va, &vm->done, base.vm_status)
+		dma_fence_wait(bo_va->last_pt_update, false);
+	dma_fence_wait(vm->last_update, false);
+
+	ret = amdgpu_eviction_fence_replace_fence(&fpriv->evf_mgr, &exec);
+	if (ret)
+		drm_file_err(uq_mgr->file, "Failed to replace eviction fence\n");
+
+unlock_all:
+	drm_exec_fini(&exec);
+	return ret;
+}
+
+static void amdgpu_userq_restore_worker(struct work_struct *work)
+{
+	struct amdgpu_userq_mgr *uq_mgr = work_to_uq_mgr(work, resume_work.work);
+	struct amdgpu_fpriv *fpriv = uq_mgr_to_fpriv(uq_mgr);
+	int ret;
+
+	flush_delayed_work(&fpriv->evf_mgr.suspend_work);
+
+	mutex_lock(&uq_mgr->userq_mutex);
+
+	ret = amdgpu_userq_vm_validate(uq_mgr);
+	if (ret) {
+		drm_file_err(uq_mgr->file, "Failed to validate BOs to restore\n");
+		goto unlock;
+	}
+
+	ret = amdgpu_userq_restore_all(uq_mgr);
+	if (ret) {
+		drm_file_err(uq_mgr->file, "Failed to restore all queues\n");
+		goto unlock;
+	}
+
+unlock:
+	mutex_unlock(&uq_mgr->userq_mutex);
+}
+
+static int
+amdgpu_userq_evict_all(struct amdgpu_userq_mgr *uq_mgr)
+{
+	struct amdgpu_usermode_queue *queue;
+	int queue_id;
+	int ret = 0, r;
+
+	/* Try to unmap all the queues in this process ctx */
+	idr_for_each_entry(&uq_mgr->userq_idr, queue, queue_id) {
+		r = amdgpu_userq_preempt_helper(uq_mgr, queue);
+		if (r)
+			ret = r;
+	}
+
+	if (ret)
+		drm_file_err(uq_mgr->file, "Couldn't unmap all the queues\n");
+	return ret;
+}
+
+static int
+amdgpu_userq_wait_for_signal(struct amdgpu_userq_mgr *uq_mgr)
+{
+	struct amdgpu_usermode_queue *queue;
+	int queue_id, ret;
+
+	idr_for_each_entry(&uq_mgr->userq_idr, queue, queue_id) {
+		struct dma_fence *f = queue->last_fence;
+
+		if (!f || dma_fence_is_signaled(f))
+			continue;
+		ret = dma_fence_wait_timeout(f, true, msecs_to_jiffies(100));
+		if (ret <= 0) {
+			drm_file_err(uq_mgr->file, "Timed out waiting for fence=%llu:%llu\n",
+				     f->context, f->seqno);
+			return -ETIMEDOUT;
+		}
+	}
+
+	return 0;
+}
+
+void
+amdgpu_userq_evict(struct amdgpu_userq_mgr *uq_mgr,
+		   struct amdgpu_eviction_fence *ev_fence)
+{
+	int ret;
+	struct amdgpu_fpriv *fpriv = uq_mgr_to_fpriv(uq_mgr);
+	struct amdgpu_eviction_fence_mgr *evf_mgr = &fpriv->evf_mgr;
+
+	/* Wait for any pending userqueue fence work to finish */
+	ret = amdgpu_userq_wait_for_signal(uq_mgr);
+	if (ret) {
+		drm_file_err(uq_mgr->file, "Not evicting userqueue, timeout waiting for work\n");
+		return;
+	}
+
+	ret = amdgpu_userq_evict_all(uq_mgr);
+	if (ret) {
+		drm_file_err(uq_mgr->file, "Failed to evict userqueue\n");
+		return;
+	}
+
+	/* Signal current eviction fence */
+	amdgpu_eviction_fence_signal(evf_mgr, ev_fence);
+
+	if (evf_mgr->fd_closing) {
+		cancel_delayed_work_sync(&uq_mgr->resume_work);
+		return;
+	}
+
+	/* Schedule a resume work */
+	schedule_delayed_work(&uq_mgr->resume_work, 0);
+}
+
+int amdgpu_userq_mgr_init(struct amdgpu_userq_mgr *userq_mgr, struct drm_file *file_priv,
+			  struct amdgpu_device *adev)
+{
+	mutex_init(&userq_mgr->userq_mutex);
+	idr_init_base(&userq_mgr->userq_idr, 1);
+	userq_mgr->adev = adev;
+	userq_mgr->file = file_priv;
+
+	mutex_lock(&adev->userq_mutex);
+	list_add(&userq_mgr->list, &adev->userq_mgr_list);
+	mutex_unlock(&adev->userq_mutex);
+
+	INIT_DELAYED_WORK(&userq_mgr->resume_work, amdgpu_userq_restore_worker);
+	return 0;
+}
+
+void amdgpu_userq_mgr_fini(struct amdgpu_userq_mgr *userq_mgr)
+{
+	struct amdgpu_device *adev = userq_mgr->adev;
+	struct amdgpu_usermode_queue *queue;
+	struct amdgpu_userq_mgr *uqm, *tmp;
+	uint32_t queue_id;
+
+	cancel_delayed_work_sync(&userq_mgr->resume_work);
+
+	mutex_lock(&adev->userq_mutex);
+	mutex_lock(&userq_mgr->userq_mutex);
+	idr_for_each_entry(&userq_mgr->userq_idr, queue, queue_id) {
+		amdgpu_userq_wait_for_last_fence(userq_mgr, queue);
+		amdgpu_userq_unmap_helper(userq_mgr, queue);
+		amdgpu_userq_cleanup(userq_mgr, queue, queue_id);
+	}
+
+	list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) {
+		if (uqm == userq_mgr) {
+			list_del(&uqm->list);
+			break;
+		}
+	}
+	idr_destroy(&userq_mgr->userq_idr);
+	mutex_unlock(&userq_mgr->userq_mutex);
+	mutex_unlock(&adev->userq_mutex);
+	mutex_destroy(&userq_mgr->userq_mutex);
+}
+
+int amdgpu_userq_suspend(struct amdgpu_device *adev)
+{
+	u32 ip_mask = amdgpu_userq_get_supported_ip_mask(adev);
+	struct amdgpu_usermode_queue *queue;
+	struct amdgpu_userq_mgr *uqm, *tmp;
+	int queue_id;
+	int ret = 0, r;
+
+	if (!ip_mask)
+		return 0;
+
+	mutex_lock(&adev->userq_mutex);
+	list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) {
+		cancel_delayed_work_sync(&uqm->resume_work);
+		mutex_lock(&uqm->userq_mutex);
+		idr_for_each_entry(&uqm->userq_idr, queue, queue_id) {
+			if (adev->in_s0ix)
+				r = amdgpu_userq_preempt_helper(uqm, queue);
+			else
+				r = amdgpu_userq_unmap_helper(uqm, queue);
+			if (r)
+				ret = r;
+		}
+		mutex_unlock(&uqm->userq_mutex);
+	}
+	mutex_unlock(&adev->userq_mutex);
+	return ret;
+}
+
+int amdgpu_userq_resume(struct amdgpu_device *adev)
+{
+	u32 ip_mask = amdgpu_userq_get_supported_ip_mask(adev);
+	struct amdgpu_usermode_queue *queue;
+	struct amdgpu_userq_mgr *uqm, *tmp;
+	int queue_id;
+	int ret = 0, r;
+
+	if (!ip_mask)
+		return 0;
+
+	mutex_lock(&adev->userq_mutex);
+	list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) {
+		mutex_lock(&uqm->userq_mutex);
+		idr_for_each_entry(&uqm->userq_idr, queue, queue_id) {
+			if (adev->in_s0ix)
+				r = amdgpu_userq_restore_helper(uqm, queue);
+			else
+				r = amdgpu_userq_map_helper(uqm, queue);
+			if (r)
+				ret = r;
+		}
+		mutex_unlock(&uqm->userq_mutex);
+	}
+	mutex_unlock(&adev->userq_mutex);
+	return ret;
+}
+
+int amdgpu_userq_stop_sched_for_enforce_isolation(struct amdgpu_device *adev,
+						  u32 idx)
+{
+	u32 ip_mask = amdgpu_userq_get_supported_ip_mask(adev);
+	struct amdgpu_usermode_queue *queue;
+	struct amdgpu_userq_mgr *uqm, *tmp;
+	int queue_id;
+	int ret = 0, r;
+
+	/* only need to stop gfx/compute */
+	if (!(ip_mask & ((1 << AMDGPU_HW_IP_GFX) | (1 << AMDGPU_HW_IP_COMPUTE))))
+		return 0;
+
+	mutex_lock(&adev->userq_mutex);
+	if (adev->userq_halt_for_enforce_isolation)
+		dev_warn(adev->dev, "userq scheduling already stopped!\n");
+	adev->userq_halt_for_enforce_isolation = true;
+	list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) {
+		cancel_delayed_work_sync(&uqm->resume_work);
+		mutex_lock(&uqm->userq_mutex);
+		idr_for_each_entry(&uqm->userq_idr, queue, queue_id) {
+			if (((queue->queue_type == AMDGPU_HW_IP_GFX) ||
+			     (queue->queue_type == AMDGPU_HW_IP_COMPUTE)) &&
+			    (queue->xcp_id == idx)) {
+				r = amdgpu_userq_preempt_helper(uqm, queue);
+				if (r)
+					ret = r;
+			}
+		}
+		mutex_unlock(&uqm->userq_mutex);
+	}
+	mutex_unlock(&adev->userq_mutex);
+	return ret;
+}
+
+int amdgpu_userq_start_sched_for_enforce_isolation(struct amdgpu_device *adev,
+						   u32 idx)
+{
+	u32 ip_mask = amdgpu_userq_get_supported_ip_mask(adev);
+	struct amdgpu_usermode_queue *queue;
+	struct amdgpu_userq_mgr *uqm, *tmp;
+	int queue_id;
+	int ret = 0, r;
+
+	/* only need to stop gfx/compute */
+	if (!(ip_mask & ((1 << AMDGPU_HW_IP_GFX) | (1 << AMDGPU_HW_IP_COMPUTE))))
+		return 0;
+
+	mutex_lock(&adev->userq_mutex);
+	if (!adev->userq_halt_for_enforce_isolation)
+		dev_warn(adev->dev, "userq scheduling already started!\n");
+	adev->userq_halt_for_enforce_isolation = false;
+	list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) {
+		mutex_lock(&uqm->userq_mutex);
+		idr_for_each_entry(&uqm->userq_idr, queue, queue_id) {
+			if (((queue->queue_type == AMDGPU_HW_IP_GFX) ||
+			     (queue->queue_type == AMDGPU_HW_IP_COMPUTE)) &&
+			    (queue->xcp_id == idx)) {
+				r = amdgpu_userq_restore_helper(uqm, queue);
+				if (r)
+					ret = r;
+			}
+		}
+		mutex_unlock(&uqm->userq_mutex);
+	}
+	mutex_unlock(&adev->userq_mutex);
+	return ret;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h
new file mode 100644
index 000000000000..c027dd916672
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h
@@ -0,0 +1,142 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef AMDGPU_USERQ_H_
+#define AMDGPU_USERQ_H_
+#include "amdgpu_eviction_fence.h"
+
+#define AMDGPU_MAX_USERQ_COUNT 512
+
+#define to_ev_fence(f) container_of(f, struct amdgpu_eviction_fence, base)
+#define uq_mgr_to_fpriv(u) container_of(u, struct amdgpu_fpriv, userq_mgr)
+#define work_to_uq_mgr(w, name) container_of(w, struct amdgpu_userq_mgr, name)
+
+enum amdgpu_userq_state {
+	AMDGPU_USERQ_STATE_UNMAPPED = 0,
+	AMDGPU_USERQ_STATE_MAPPED,
+	AMDGPU_USERQ_STATE_PREEMPTED,
+	AMDGPU_USERQ_STATE_HUNG,
+};
+
+struct amdgpu_mqd_prop;
+
+struct amdgpu_userq_obj {
+	void		 *cpu_ptr;
+	uint64_t	 gpu_addr;
+	struct amdgpu_bo *obj;
+};
+
+struct amdgpu_usermode_queue {
+	int			queue_type;
+	enum amdgpu_userq_state state;
+	uint64_t		doorbell_handle;
+	uint64_t		doorbell_index;
+	uint64_t		flags;
+	struct amdgpu_mqd_prop	*userq_prop;
+	struct amdgpu_userq_mgr *userq_mgr;
+	struct amdgpu_vm	*vm;
+	struct amdgpu_userq_obj mqd;
+	struct amdgpu_userq_obj	db_obj;
+	struct amdgpu_userq_obj fw_obj;
+	struct amdgpu_userq_obj wptr_obj;
+	struct xarray		fence_drv_xa;
+	struct amdgpu_userq_fence_driver *fence_drv;
+	struct dma_fence	*last_fence;
+	u32			xcp_id;
+	int			priority;
+	struct dentry		*debugfs_queue;
+};
+
+struct amdgpu_userq_funcs {
+	int (*mqd_create)(struct amdgpu_userq_mgr *uq_mgr,
+			  struct drm_amdgpu_userq_in *args,
+			  struct amdgpu_usermode_queue *queue);
+	void (*mqd_destroy)(struct amdgpu_userq_mgr *uq_mgr,
+			    struct amdgpu_usermode_queue *uq);
+	int (*unmap)(struct amdgpu_userq_mgr *uq_mgr,
+		     struct amdgpu_usermode_queue *queue);
+	int (*map)(struct amdgpu_userq_mgr *uq_mgr,
+		   struct amdgpu_usermode_queue *queue);
+	int (*preempt)(struct amdgpu_userq_mgr *uq_mgr,
+		   struct amdgpu_usermode_queue *queue);
+	int (*restore)(struct amdgpu_userq_mgr *uq_mgr,
+		   struct amdgpu_usermode_queue *queue);
+	int (*detect_and_reset)(struct amdgpu_device *adev,
+		  int queue_type);
+};
+
+/* Usermode queues for gfx */
+struct amdgpu_userq_mgr {
+	struct idr			userq_idr;
+	struct mutex			userq_mutex;
+	struct amdgpu_device		*adev;
+	struct delayed_work		resume_work;
+	struct list_head		list;
+	struct drm_file			*file;
+};
+
+struct amdgpu_db_info {
+	uint64_t doorbell_handle;
+	uint32_t queue_type;
+	uint32_t doorbell_offset;
+	struct amdgpu_userq_obj	*db_obj;
+};
+
+int amdgpu_userq_ioctl(struct drm_device *dev, void *data, struct drm_file *filp);
+
+int amdgpu_userq_mgr_init(struct amdgpu_userq_mgr *userq_mgr, struct drm_file *file_priv,
+			  struct amdgpu_device *adev);
+
+void amdgpu_userq_mgr_fini(struct amdgpu_userq_mgr *userq_mgr);
+
+int amdgpu_userq_create_object(struct amdgpu_userq_mgr *uq_mgr,
+			       struct amdgpu_userq_obj *userq_obj,
+			       int size);
+
+void amdgpu_userq_destroy_object(struct amdgpu_userq_mgr *uq_mgr,
+				 struct amdgpu_userq_obj *userq_obj);
+
+void amdgpu_userq_evict(struct amdgpu_userq_mgr *uq_mgr,
+			struct amdgpu_eviction_fence *ev_fence);
+
+void amdgpu_userq_ensure_ev_fence(struct amdgpu_userq_mgr *userq_mgr,
+				  struct amdgpu_eviction_fence_mgr *evf_mgr);
+
+uint64_t amdgpu_userq_get_doorbell_index(struct amdgpu_userq_mgr *uq_mgr,
+					 struct amdgpu_db_info *db_info,
+					     struct drm_file *filp);
+
+u32 amdgpu_userq_get_supported_ip_mask(struct amdgpu_device *adev);
+
+int amdgpu_userq_suspend(struct amdgpu_device *adev);
+int amdgpu_userq_resume(struct amdgpu_device *adev);
+
+int amdgpu_userq_stop_sched_for_enforce_isolation(struct amdgpu_device *adev,
+						  u32 idx);
+int amdgpu_userq_start_sched_for_enforce_isolation(struct amdgpu_device *adev,
+						   u32 idx);
+
+int amdgpu_userq_input_va_validate(struct amdgpu_vm *vm, u64 addr,
+				   u64 expected_size);
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c
new file mode 100644
index 000000000000..761bad98da3e
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c
@@ -0,0 +1,1009 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/kref.h>
+#include <linux/slab.h>
+#include <linux/dma-fence-unwrap.h>
+
+#include <drm/drm_exec.h>
+#include <drm/drm_syncobj.h>
+
+#include "amdgpu.h"
+#include "amdgpu_userq_fence.h"
+
+static const struct dma_fence_ops amdgpu_userq_fence_ops;
+static struct kmem_cache *amdgpu_userq_fence_slab;
+
+int amdgpu_userq_fence_slab_init(void)
+{
+	amdgpu_userq_fence_slab = kmem_cache_create("amdgpu_userq_fence",
+						    sizeof(struct amdgpu_userq_fence),
+						    0,
+						    SLAB_HWCACHE_ALIGN,
+						    NULL);
+	if (!amdgpu_userq_fence_slab)
+		return -ENOMEM;
+
+	return 0;
+}
+
+void amdgpu_userq_fence_slab_fini(void)
+{
+	rcu_barrier();
+	kmem_cache_destroy(amdgpu_userq_fence_slab);
+}
+
+static inline struct amdgpu_userq_fence *to_amdgpu_userq_fence(struct dma_fence *f)
+{
+	if (!f || f->ops != &amdgpu_userq_fence_ops)
+		return NULL;
+
+	return container_of(f, struct amdgpu_userq_fence, base);
+}
+
+static u64 amdgpu_userq_fence_read(struct amdgpu_userq_fence_driver *fence_drv)
+{
+	return le64_to_cpu(*fence_drv->cpu_addr);
+}
+
+static void
+amdgpu_userq_fence_write(struct amdgpu_userq_fence_driver *fence_drv,
+			 u64 seq)
+{
+	if (fence_drv->cpu_addr)
+		*fence_drv->cpu_addr = cpu_to_le64(seq);
+}
+
+int amdgpu_userq_fence_driver_alloc(struct amdgpu_device *adev,
+				    struct amdgpu_usermode_queue *userq)
+{
+	struct amdgpu_userq_fence_driver *fence_drv;
+	unsigned long flags;
+	int r;
+
+	fence_drv = kzalloc(sizeof(*fence_drv), GFP_KERNEL);
+	if (!fence_drv)
+		return -ENOMEM;
+
+	/* Acquire seq64 memory */
+	r = amdgpu_seq64_alloc(adev, &fence_drv->va, &fence_drv->gpu_addr,
+			       &fence_drv->cpu_addr);
+	if (r)
+		goto free_fence_drv;
+
+	memset(fence_drv->cpu_addr, 0, sizeof(u64));
+
+	kref_init(&fence_drv->refcount);
+	INIT_LIST_HEAD(&fence_drv->fences);
+	spin_lock_init(&fence_drv->fence_list_lock);
+
+	fence_drv->adev = adev;
+	fence_drv->context = dma_fence_context_alloc(1);
+	get_task_comm(fence_drv->timeline_name, current);
+
+	xa_lock_irqsave(&adev->userq_xa, flags);
+	r = xa_err(__xa_store(&adev->userq_xa, userq->doorbell_index,
+			      fence_drv, GFP_KERNEL));
+	xa_unlock_irqrestore(&adev->userq_xa, flags);
+	if (r)
+		goto free_seq64;
+
+	userq->fence_drv = fence_drv;
+
+	return 0;
+
+free_seq64:
+	amdgpu_seq64_free(adev, fence_drv->va);
+free_fence_drv:
+	kfree(fence_drv);
+
+	return r;
+}
+
+static void amdgpu_userq_walk_and_drop_fence_drv(struct xarray *xa)
+{
+	struct amdgpu_userq_fence_driver *fence_drv;
+	unsigned long index;
+
+	if (xa_empty(xa))
+		return;
+
+	xa_lock(xa);
+	xa_for_each(xa, index, fence_drv) {
+		__xa_erase(xa, index);
+		amdgpu_userq_fence_driver_put(fence_drv);
+	}
+
+	xa_unlock(xa);
+}
+
+void
+amdgpu_userq_fence_driver_free(struct amdgpu_usermode_queue *userq)
+{
+	amdgpu_userq_walk_and_drop_fence_drv(&userq->fence_drv_xa);
+	xa_destroy(&userq->fence_drv_xa);
+	/* Drop the fence_drv reference held by user queue */
+	amdgpu_userq_fence_driver_put(userq->fence_drv);
+}
+
+void amdgpu_userq_fence_driver_process(struct amdgpu_userq_fence_driver *fence_drv)
+{
+	struct amdgpu_userq_fence *userq_fence, *tmp;
+	struct dma_fence *fence;
+	u64 rptr;
+	int i;
+
+	if (!fence_drv)
+		return;
+
+	rptr = amdgpu_userq_fence_read(fence_drv);
+
+	spin_lock(&fence_drv->fence_list_lock);
+	list_for_each_entry_safe(userq_fence, tmp, &fence_drv->fences, link) {
+		fence = &userq_fence->base;
+
+		if (rptr < fence->seqno)
+			break;
+
+		dma_fence_signal(fence);
+
+		for (i = 0; i < userq_fence->fence_drv_array_count; i++)
+			amdgpu_userq_fence_driver_put(userq_fence->fence_drv_array[i]);
+
+		list_del(&userq_fence->link);
+		dma_fence_put(fence);
+	}
+	spin_unlock(&fence_drv->fence_list_lock);
+}
+
+void amdgpu_userq_fence_driver_destroy(struct kref *ref)
+{
+	struct amdgpu_userq_fence_driver *fence_drv = container_of(ref,
+					 struct amdgpu_userq_fence_driver,
+					 refcount);
+	struct amdgpu_userq_fence_driver *xa_fence_drv;
+	struct amdgpu_device *adev = fence_drv->adev;
+	struct amdgpu_userq_fence *fence, *tmp;
+	struct xarray *xa = &adev->userq_xa;
+	unsigned long index, flags;
+	struct dma_fence *f;
+
+	spin_lock_irqsave(&fence_drv->fence_list_lock, flags);
+	list_for_each_entry_safe(fence, tmp, &fence_drv->fences, link) {
+		f = &fence->base;
+
+		if (!dma_fence_is_signaled(f)) {
+			dma_fence_set_error(f, -ECANCELED);
+			dma_fence_signal(f);
+		}
+
+		list_del(&fence->link);
+		dma_fence_put(f);
+	}
+	spin_unlock_irqrestore(&fence_drv->fence_list_lock, flags);
+
+	xa_lock_irqsave(xa, flags);
+	xa_for_each(xa, index, xa_fence_drv)
+		if (xa_fence_drv == fence_drv)
+			__xa_erase(xa, index);
+	xa_unlock_irqrestore(xa, flags);
+
+	/* Free seq64 memory */
+	amdgpu_seq64_free(adev, fence_drv->va);
+	kfree(fence_drv);
+}
+
+void amdgpu_userq_fence_driver_get(struct amdgpu_userq_fence_driver *fence_drv)
+{
+	kref_get(&fence_drv->refcount);
+}
+
+void amdgpu_userq_fence_driver_put(struct amdgpu_userq_fence_driver *fence_drv)
+{
+	kref_put(&fence_drv->refcount, amdgpu_userq_fence_driver_destroy);
+}
+
+static int amdgpu_userq_fence_alloc(struct amdgpu_userq_fence **userq_fence)
+{
+	*userq_fence = kmem_cache_alloc(amdgpu_userq_fence_slab, GFP_ATOMIC);
+	return *userq_fence ? 0 : -ENOMEM;
+}
+
+static int amdgpu_userq_fence_create(struct amdgpu_usermode_queue *userq,
+				     struct amdgpu_userq_fence *userq_fence,
+				     u64 seq, struct dma_fence **f)
+{
+	struct amdgpu_userq_fence_driver *fence_drv;
+	struct dma_fence *fence;
+	unsigned long flags;
+
+	fence_drv = userq->fence_drv;
+	if (!fence_drv)
+		return -EINVAL;
+
+	spin_lock_init(&userq_fence->lock);
+	INIT_LIST_HEAD(&userq_fence->link);
+	fence = &userq_fence->base;
+	userq_fence->fence_drv = fence_drv;
+
+	dma_fence_init64(fence, &amdgpu_userq_fence_ops, &userq_fence->lock,
+			 fence_drv->context, seq);
+
+	amdgpu_userq_fence_driver_get(fence_drv);
+	dma_fence_get(fence);
+
+	if (!xa_empty(&userq->fence_drv_xa)) {
+		struct amdgpu_userq_fence_driver *stored_fence_drv;
+		unsigned long index, count = 0;
+		int i = 0;
+
+		xa_lock(&userq->fence_drv_xa);
+		xa_for_each(&userq->fence_drv_xa, index, stored_fence_drv)
+			count++;
+
+		userq_fence->fence_drv_array =
+			kvmalloc_array(count,
+				       sizeof(struct amdgpu_userq_fence_driver *),
+				       GFP_ATOMIC);
+
+		if (userq_fence->fence_drv_array) {
+			xa_for_each(&userq->fence_drv_xa, index, stored_fence_drv) {
+				userq_fence->fence_drv_array[i] = stored_fence_drv;
+				__xa_erase(&userq->fence_drv_xa, index);
+				i++;
+			}
+		}
+
+		userq_fence->fence_drv_array_count = i;
+		xa_unlock(&userq->fence_drv_xa);
+	} else {
+		userq_fence->fence_drv_array = NULL;
+		userq_fence->fence_drv_array_count = 0;
+	}
+
+	/* Check if hardware has already processed the job */
+	spin_lock_irqsave(&fence_drv->fence_list_lock, flags);
+	if (!dma_fence_is_signaled(fence))
+		list_add_tail(&userq_fence->link, &fence_drv->fences);
+	else
+		dma_fence_put(fence);
+
+	spin_unlock_irqrestore(&fence_drv->fence_list_lock, flags);
+
+	*f = fence;
+
+	return 0;
+}
+
+static const char *amdgpu_userq_fence_get_driver_name(struct dma_fence *f)
+{
+	return "amdgpu_userq_fence";
+}
+
+static const char *amdgpu_userq_fence_get_timeline_name(struct dma_fence *f)
+{
+	struct amdgpu_userq_fence *fence = to_amdgpu_userq_fence(f);
+
+	return fence->fence_drv->timeline_name;
+}
+
+static bool amdgpu_userq_fence_signaled(struct dma_fence *f)
+{
+	struct amdgpu_userq_fence *fence = to_amdgpu_userq_fence(f);
+	struct amdgpu_userq_fence_driver *fence_drv = fence->fence_drv;
+	u64 rptr, wptr;
+
+	rptr = amdgpu_userq_fence_read(fence_drv);
+	wptr = fence->base.seqno;
+
+	if (rptr >= wptr)
+		return true;
+
+	return false;
+}
+
+static void amdgpu_userq_fence_free(struct rcu_head *rcu)
+{
+	struct dma_fence *fence = container_of(rcu, struct dma_fence, rcu);
+	struct amdgpu_userq_fence *userq_fence = to_amdgpu_userq_fence(fence);
+	struct amdgpu_userq_fence_driver *fence_drv = userq_fence->fence_drv;
+
+	/* Release the fence driver reference */
+	amdgpu_userq_fence_driver_put(fence_drv);
+
+	kvfree(userq_fence->fence_drv_array);
+	kmem_cache_free(amdgpu_userq_fence_slab, userq_fence);
+}
+
+static void amdgpu_userq_fence_release(struct dma_fence *f)
+{
+	call_rcu(&f->rcu, amdgpu_userq_fence_free);
+}
+
+static const struct dma_fence_ops amdgpu_userq_fence_ops = {
+	.get_driver_name = amdgpu_userq_fence_get_driver_name,
+	.get_timeline_name = amdgpu_userq_fence_get_timeline_name,
+	.signaled = amdgpu_userq_fence_signaled,
+	.release = amdgpu_userq_fence_release,
+};
+
+/**
+ * amdgpu_userq_fence_read_wptr - Read the userq wptr value
+ *
+ * @queue: user mode queue structure pointer
+ * @wptr: write pointer value
+ *
+ * Read the wptr value from userq's MQD. The userq signal IOCTL
+ * creates a dma_fence for the shared buffers that expects the
+ * RPTR value written to seq64 memory >= WPTR.
+ *
+ * Returns wptr value on success, error on failure.
+ */
+static int amdgpu_userq_fence_read_wptr(struct amdgpu_usermode_queue *queue,
+					u64 *wptr)
+{
+	struct amdgpu_bo_va_mapping *mapping;
+	struct amdgpu_bo *bo;
+	u64 addr, *ptr;
+	int r;
+
+	r = amdgpu_bo_reserve(queue->vm->root.bo, false);
+	if (r)
+		return r;
+
+	addr = queue->userq_prop->wptr_gpu_addr;
+	addr &= AMDGPU_GMC_HOLE_MASK;
+
+	mapping = amdgpu_vm_bo_lookup_mapping(queue->vm, addr >> PAGE_SHIFT);
+	if (!mapping) {
+		amdgpu_bo_unreserve(queue->vm->root.bo);
+		DRM_ERROR("Failed to lookup amdgpu_bo_va_mapping\n");
+		return -EINVAL;
+	}
+
+	bo = amdgpu_bo_ref(mapping->bo_va->base.bo);
+	amdgpu_bo_unreserve(queue->vm->root.bo);
+	r = amdgpu_bo_reserve(bo, true);
+	if (r) {
+		DRM_ERROR("Failed to reserve userqueue wptr bo");
+		return r;
+	}
+
+	r = amdgpu_bo_kmap(bo, (void **)&ptr);
+	if (r) {
+		DRM_ERROR("Failed mapping the userqueue wptr bo");
+		goto map_error;
+	}
+
+	*wptr = le64_to_cpu(*ptr);
+
+	amdgpu_bo_kunmap(bo);
+	amdgpu_bo_unreserve(bo);
+	amdgpu_bo_unref(&bo);
+
+	return 0;
+
+map_error:
+	amdgpu_bo_unreserve(bo);
+	amdgpu_bo_unref(&bo);
+
+	return r;
+}
+
+static void amdgpu_userq_fence_cleanup(struct dma_fence *fence)
+{
+	dma_fence_put(fence);
+}
+
+static void
+amdgpu_userq_fence_driver_set_error(struct amdgpu_userq_fence *fence,
+				    int error)
+{
+	struct amdgpu_userq_fence_driver *fence_drv = fence->fence_drv;
+	unsigned long flags;
+	struct dma_fence *f;
+
+	spin_lock_irqsave(&fence_drv->fence_list_lock, flags);
+
+	f = rcu_dereference_protected(&fence->base,
+				      lockdep_is_held(&fence_drv->fence_list_lock));
+	if (f && !dma_fence_is_signaled_locked(f))
+		dma_fence_set_error(f, error);
+	spin_unlock_irqrestore(&fence_drv->fence_list_lock, flags);
+}
+
+void
+amdgpu_userq_fence_driver_force_completion(struct amdgpu_usermode_queue *userq)
+{
+	struct dma_fence *f = userq->last_fence;
+
+	if (f) {
+		struct amdgpu_userq_fence *fence = to_amdgpu_userq_fence(f);
+		struct amdgpu_userq_fence_driver *fence_drv = fence->fence_drv;
+		u64 wptr = fence->base.seqno;
+
+		amdgpu_userq_fence_driver_set_error(fence, -ECANCELED);
+		amdgpu_userq_fence_write(fence_drv, wptr);
+		amdgpu_userq_fence_driver_process(fence_drv);
+
+	}
+}
+
+int amdgpu_userq_signal_ioctl(struct drm_device *dev, void *data,
+			      struct drm_file *filp)
+{
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+	struct amdgpu_userq_mgr *userq_mgr = &fpriv->userq_mgr;
+	struct drm_amdgpu_userq_signal *args = data;
+	struct drm_gem_object **gobj_write = NULL;
+	struct drm_gem_object **gobj_read = NULL;
+	struct amdgpu_usermode_queue *queue;
+	struct amdgpu_userq_fence *userq_fence;
+	struct drm_syncobj **syncobj = NULL;
+	u32 *bo_handles_write, num_write_bo_handles;
+	u32 *syncobj_handles, num_syncobj_handles;
+	u32 *bo_handles_read, num_read_bo_handles;
+	int r, i, entry, rentry, wentry;
+	struct dma_fence *fence;
+	struct drm_exec exec;
+	u64 wptr;
+
+	num_syncobj_handles = args->num_syncobj_handles;
+	syncobj_handles = memdup_user(u64_to_user_ptr(args->syncobj_handles),
+				      size_mul(sizeof(u32), num_syncobj_handles));
+	if (IS_ERR(syncobj_handles))
+		return PTR_ERR(syncobj_handles);
+
+	/* Array of pointers to the looked up syncobjs */
+	syncobj = kmalloc_array(num_syncobj_handles, sizeof(*syncobj), GFP_KERNEL);
+	if (!syncobj) {
+		r = -ENOMEM;
+		goto free_syncobj_handles;
+	}
+
+	for (entry = 0; entry < num_syncobj_handles; entry++) {
+		syncobj[entry] = drm_syncobj_find(filp, syncobj_handles[entry]);
+		if (!syncobj[entry]) {
+			r = -ENOENT;
+			goto free_syncobj;
+		}
+	}
+
+	num_read_bo_handles = args->num_bo_read_handles;
+	bo_handles_read = memdup_user(u64_to_user_ptr(args->bo_read_handles),
+				      sizeof(u32) * num_read_bo_handles);
+	if (IS_ERR(bo_handles_read)) {
+		r = PTR_ERR(bo_handles_read);
+		goto free_syncobj;
+	}
+
+	/* Array of pointers to the GEM read objects */
+	gobj_read = kmalloc_array(num_read_bo_handles, sizeof(*gobj_read), GFP_KERNEL);
+	if (!gobj_read) {
+		r = -ENOMEM;
+		goto free_bo_handles_read;
+	}
+
+	for (rentry = 0; rentry < num_read_bo_handles; rentry++) {
+		gobj_read[rentry] = drm_gem_object_lookup(filp, bo_handles_read[rentry]);
+		if (!gobj_read[rentry]) {
+			r = -ENOENT;
+			goto put_gobj_read;
+		}
+	}
+
+	num_write_bo_handles = args->num_bo_write_handles;
+	bo_handles_write = memdup_user(u64_to_user_ptr(args->bo_write_handles),
+				       sizeof(u32) * num_write_bo_handles);
+	if (IS_ERR(bo_handles_write)) {
+		r = PTR_ERR(bo_handles_write);
+		goto put_gobj_read;
+	}
+
+	/* Array of pointers to the GEM write objects */
+	gobj_write = kmalloc_array(num_write_bo_handles, sizeof(*gobj_write), GFP_KERNEL);
+	if (!gobj_write) {
+		r = -ENOMEM;
+		goto free_bo_handles_write;
+	}
+
+	for (wentry = 0; wentry < num_write_bo_handles; wentry++) {
+		gobj_write[wentry] = drm_gem_object_lookup(filp, bo_handles_write[wentry]);
+		if (!gobj_write[wentry]) {
+			r = -ENOENT;
+			goto put_gobj_write;
+		}
+	}
+
+	/* Retrieve the user queue */
+	queue = idr_find(&userq_mgr->userq_idr, args->queue_id);
+	if (!queue) {
+		r = -ENOENT;
+		goto put_gobj_write;
+	}
+
+	r = amdgpu_userq_fence_read_wptr(queue, &wptr);
+	if (r)
+		goto put_gobj_write;
+
+	r = amdgpu_userq_fence_alloc(&userq_fence);
+	if (r)
+		goto put_gobj_write;
+
+	/* We are here means UQ is active, make sure the eviction fence is valid */
+	amdgpu_userq_ensure_ev_fence(&fpriv->userq_mgr, &fpriv->evf_mgr);
+
+	/* Create a new fence */
+	r = amdgpu_userq_fence_create(queue, userq_fence, wptr, &fence);
+	if (r) {
+		mutex_unlock(&userq_mgr->userq_mutex);
+		kmem_cache_free(amdgpu_userq_fence_slab, userq_fence);
+		goto put_gobj_write;
+	}
+
+	dma_fence_put(queue->last_fence);
+	queue->last_fence = dma_fence_get(fence);
+	mutex_unlock(&userq_mgr->userq_mutex);
+
+	drm_exec_init(&exec, DRM_EXEC_INTERRUPTIBLE_WAIT,
+		      (num_read_bo_handles + num_write_bo_handles));
+
+	/* Lock all BOs with retry handling */
+	drm_exec_until_all_locked(&exec) {
+		r = drm_exec_prepare_array(&exec, gobj_read, num_read_bo_handles, 1);
+		drm_exec_retry_on_contention(&exec);
+		if (r) {
+			amdgpu_userq_fence_cleanup(fence);
+			goto exec_fini;
+		}
+
+		r = drm_exec_prepare_array(&exec, gobj_write, num_write_bo_handles, 1);
+		drm_exec_retry_on_contention(&exec);
+		if (r) {
+			amdgpu_userq_fence_cleanup(fence);
+			goto exec_fini;
+		}
+	}
+
+	for (i = 0; i < num_read_bo_handles; i++) {
+		if (!gobj_read || !gobj_read[i]->resv)
+			continue;
+
+		dma_resv_add_fence(gobj_read[i]->resv, fence,
+				   DMA_RESV_USAGE_READ);
+	}
+
+	for (i = 0; i < num_write_bo_handles; i++) {
+		if (!gobj_write || !gobj_write[i]->resv)
+			continue;
+
+		dma_resv_add_fence(gobj_write[i]->resv, fence,
+				   DMA_RESV_USAGE_WRITE);
+	}
+
+	/* Add the created fence to syncobj/BO's */
+	for (i = 0; i < num_syncobj_handles; i++)
+		drm_syncobj_replace_fence(syncobj[i], fence);
+
+	/* drop the reference acquired in fence creation function */
+	dma_fence_put(fence);
+
+exec_fini:
+	drm_exec_fini(&exec);
+put_gobj_write:
+	while (wentry-- > 0)
+		drm_gem_object_put(gobj_write[wentry]);
+	kfree(gobj_write);
+free_bo_handles_write:
+	kfree(bo_handles_write);
+put_gobj_read:
+	while (rentry-- > 0)
+		drm_gem_object_put(gobj_read[rentry]);
+	kfree(gobj_read);
+free_bo_handles_read:
+	kfree(bo_handles_read);
+free_syncobj:
+	while (entry-- > 0)
+		if (syncobj[entry])
+			drm_syncobj_put(syncobj[entry]);
+	kfree(syncobj);
+free_syncobj_handles:
+	kfree(syncobj_handles);
+
+	return r;
+}
+
+int amdgpu_userq_wait_ioctl(struct drm_device *dev, void *data,
+			    struct drm_file *filp)
+{
+	u32 *syncobj_handles, *timeline_points, *timeline_handles, *bo_handles_read, *bo_handles_write;
+	u32 num_syncobj, num_read_bo_handles, num_write_bo_handles;
+	struct drm_amdgpu_userq_fence_info *fence_info = NULL;
+	struct drm_amdgpu_userq_wait *wait_info = data;
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+	struct amdgpu_userq_mgr *userq_mgr = &fpriv->userq_mgr;
+	struct amdgpu_usermode_queue *waitq;
+	struct drm_gem_object **gobj_write;
+	struct drm_gem_object **gobj_read;
+	struct dma_fence **fences = NULL;
+	u16 num_points, num_fences = 0;
+	int r, i, rentry, wentry, cnt;
+	struct drm_exec exec;
+
+	num_read_bo_handles = wait_info->num_bo_read_handles;
+	bo_handles_read = memdup_user(u64_to_user_ptr(wait_info->bo_read_handles),
+				      size_mul(sizeof(u32), num_read_bo_handles));
+	if (IS_ERR(bo_handles_read))
+		return PTR_ERR(bo_handles_read);
+
+	num_write_bo_handles = wait_info->num_bo_write_handles;
+	bo_handles_write = memdup_user(u64_to_user_ptr(wait_info->bo_write_handles),
+				       size_mul(sizeof(u32), num_write_bo_handles));
+	if (IS_ERR(bo_handles_write)) {
+		r = PTR_ERR(bo_handles_write);
+		goto free_bo_handles_read;
+	}
+
+	num_syncobj = wait_info->num_syncobj_handles;
+	syncobj_handles = memdup_user(u64_to_user_ptr(wait_info->syncobj_handles),
+				      size_mul(sizeof(u32), num_syncobj));
+	if (IS_ERR(syncobj_handles)) {
+		r = PTR_ERR(syncobj_handles);
+		goto free_bo_handles_write;
+	}
+
+	num_points = wait_info->num_syncobj_timeline_handles;
+	timeline_handles = memdup_user(u64_to_user_ptr(wait_info->syncobj_timeline_handles),
+				       sizeof(u32) * num_points);
+	if (IS_ERR(timeline_handles)) {
+		r = PTR_ERR(timeline_handles);
+		goto free_syncobj_handles;
+	}
+
+	timeline_points = memdup_user(u64_to_user_ptr(wait_info->syncobj_timeline_points),
+				      sizeof(u32) * num_points);
+	if (IS_ERR(timeline_points)) {
+		r = PTR_ERR(timeline_points);
+		goto free_timeline_handles;
+	}
+
+	gobj_read = kmalloc_array(num_read_bo_handles, sizeof(*gobj_read), GFP_KERNEL);
+	if (!gobj_read) {
+		r = -ENOMEM;
+		goto free_timeline_points;
+	}
+
+	for (rentry = 0; rentry < num_read_bo_handles; rentry++) {
+		gobj_read[rentry] = drm_gem_object_lookup(filp, bo_handles_read[rentry]);
+		if (!gobj_read[rentry]) {
+			r = -ENOENT;
+			goto put_gobj_read;
+		}
+	}
+
+	gobj_write = kmalloc_array(num_write_bo_handles, sizeof(*gobj_write), GFP_KERNEL);
+	if (!gobj_write) {
+		r = -ENOMEM;
+		goto put_gobj_read;
+	}
+
+	for (wentry = 0; wentry < num_write_bo_handles; wentry++) {
+		gobj_write[wentry] = drm_gem_object_lookup(filp, bo_handles_write[wentry]);
+		if (!gobj_write[wentry]) {
+			r = -ENOENT;
+			goto put_gobj_write;
+		}
+	}
+
+	drm_exec_init(&exec, DRM_EXEC_INTERRUPTIBLE_WAIT,
+		      (num_read_bo_handles + num_write_bo_handles));
+
+	/* Lock all BOs with retry handling */
+	drm_exec_until_all_locked(&exec) {
+		r = drm_exec_prepare_array(&exec, gobj_read, num_read_bo_handles, 1);
+		drm_exec_retry_on_contention(&exec);
+		if (r) {
+			drm_exec_fini(&exec);
+			goto put_gobj_write;
+		}
+
+		r = drm_exec_prepare_array(&exec, gobj_write, num_write_bo_handles, 1);
+		drm_exec_retry_on_contention(&exec);
+		if (r) {
+			drm_exec_fini(&exec);
+			goto put_gobj_write;
+		}
+	}
+
+	if (!wait_info->num_fences) {
+		if (num_points) {
+			struct dma_fence_unwrap iter;
+			struct dma_fence *fence;
+			struct dma_fence *f;
+
+			for (i = 0; i < num_points; i++) {
+				r = drm_syncobj_find_fence(filp, timeline_handles[i],
+							   timeline_points[i],
+							   DRM_SYNCOBJ_WAIT_FLAGS_WAIT_FOR_SUBMIT,
+							   &fence);
+				if (r)
+					goto exec_fini;
+
+				dma_fence_unwrap_for_each(f, &iter, fence)
+					num_fences++;
+
+				dma_fence_put(fence);
+			}
+		}
+
+		/* Count syncobj's fence */
+		for (i = 0; i < num_syncobj; i++) {
+			struct dma_fence *fence;
+
+			r = drm_syncobj_find_fence(filp, syncobj_handles[i],
+						   0,
+						   DRM_SYNCOBJ_WAIT_FLAGS_WAIT_FOR_SUBMIT,
+						   &fence);
+			if (r)
+				goto exec_fini;
+
+			num_fences++;
+			dma_fence_put(fence);
+		}
+
+		/* Count GEM objects fence */
+		for (i = 0; i < num_read_bo_handles; i++) {
+			struct dma_resv_iter resv_cursor;
+			struct dma_fence *fence;
+
+			dma_resv_for_each_fence(&resv_cursor, gobj_read[i]->resv,
+						DMA_RESV_USAGE_READ, fence)
+				num_fences++;
+		}
+
+		for (i = 0; i < num_write_bo_handles; i++) {
+			struct dma_resv_iter resv_cursor;
+			struct dma_fence *fence;
+
+			dma_resv_for_each_fence(&resv_cursor, gobj_write[i]->resv,
+						DMA_RESV_USAGE_WRITE, fence)
+				num_fences++;
+		}
+
+		/*
+		 * Passing num_fences = 0 means that userspace doesn't want to
+		 * retrieve userq_fence_info. If num_fences = 0 we skip filling
+		 * userq_fence_info and return the actual number of fences on
+		 * args->num_fences.
+		 */
+		wait_info->num_fences = num_fences;
+	} else {
+		/* Array of fence info */
+		fence_info = kmalloc_array(wait_info->num_fences, sizeof(*fence_info), GFP_KERNEL);
+		if (!fence_info) {
+			r = -ENOMEM;
+			goto exec_fini;
+		}
+
+		/* Array of fences */
+		fences = kmalloc_array(wait_info->num_fences, sizeof(*fences), GFP_KERNEL);
+		if (!fences) {
+			r = -ENOMEM;
+			goto free_fence_info;
+		}
+
+		/* Retrieve GEM read objects fence */
+		for (i = 0; i < num_read_bo_handles; i++) {
+			struct dma_resv_iter resv_cursor;
+			struct dma_fence *fence;
+
+			dma_resv_for_each_fence(&resv_cursor, gobj_read[i]->resv,
+						DMA_RESV_USAGE_READ, fence) {
+				if (WARN_ON_ONCE(num_fences >= wait_info->num_fences)) {
+					r = -EINVAL;
+					goto free_fences;
+				}
+
+				fences[num_fences++] = fence;
+				dma_fence_get(fence);
+			}
+		}
+
+		/* Retrieve GEM write objects fence */
+		for (i = 0; i < num_write_bo_handles; i++) {
+			struct dma_resv_iter resv_cursor;
+			struct dma_fence *fence;
+
+			dma_resv_for_each_fence(&resv_cursor, gobj_write[i]->resv,
+						DMA_RESV_USAGE_WRITE, fence) {
+				if (WARN_ON_ONCE(num_fences >= wait_info->num_fences)) {
+					r = -EINVAL;
+					goto free_fences;
+				}
+
+				fences[num_fences++] = fence;
+				dma_fence_get(fence);
+			}
+		}
+
+		if (num_points) {
+			struct dma_fence_unwrap iter;
+			struct dma_fence *fence;
+			struct dma_fence *f;
+
+			for (i = 0; i < num_points; i++) {
+				r = drm_syncobj_find_fence(filp, timeline_handles[i],
+							   timeline_points[i],
+							   DRM_SYNCOBJ_WAIT_FLAGS_WAIT_FOR_SUBMIT,
+							   &fence);
+				if (r)
+					goto free_fences;
+
+				dma_fence_unwrap_for_each(f, &iter, fence) {
+					if (WARN_ON_ONCE(num_fences >= wait_info->num_fences)) {
+						r = -EINVAL;
+						goto free_fences;
+					}
+
+					dma_fence_get(f);
+					fences[num_fences++] = f;
+				}
+
+				dma_fence_put(fence);
+			}
+		}
+
+		/* Retrieve syncobj's fence */
+		for (i = 0; i < num_syncobj; i++) {
+			struct dma_fence *fence;
+
+			r = drm_syncobj_find_fence(filp, syncobj_handles[i],
+						   0,
+						   DRM_SYNCOBJ_WAIT_FLAGS_WAIT_FOR_SUBMIT,
+						   &fence);
+			if (r)
+				goto free_fences;
+
+			if (WARN_ON_ONCE(num_fences >= wait_info->num_fences)) {
+				r = -EINVAL;
+				goto free_fences;
+			}
+
+			fences[num_fences++] = fence;
+		}
+
+		/*
+		 * Keep only the latest fences to reduce the number of values
+		 * given back to userspace.
+		 */
+		num_fences = dma_fence_dedup_array(fences, num_fences);
+
+		waitq = idr_find(&userq_mgr->userq_idr, wait_info->waitq_id);
+		if (!waitq) {
+			r = -EINVAL;
+			goto free_fences;
+		}
+
+		for (i = 0, cnt = 0; i < num_fences; i++) {
+			struct amdgpu_userq_fence_driver *fence_drv;
+			struct amdgpu_userq_fence *userq_fence;
+			u32 index;
+
+			userq_fence = to_amdgpu_userq_fence(fences[i]);
+			if (!userq_fence) {
+				/*
+				 * Just waiting on other driver fences should
+				 * be good for now
+				 */
+				r = dma_fence_wait(fences[i], true);
+				if (r) {
+					dma_fence_put(fences[i]);
+					goto free_fences;
+				}
+
+				dma_fence_put(fences[i]);
+				continue;
+			}
+
+			fence_drv = userq_fence->fence_drv;
+			/*
+			 * We need to make sure the user queue release their reference
+			 * to the fence drivers at some point before queue destruction.
+			 * Otherwise, we would gather those references until we don't
+			 * have any more space left and crash.
+			 */
+			r = xa_alloc(&waitq->fence_drv_xa, &index, fence_drv,
+				     xa_limit_32b, GFP_KERNEL);
+			if (r)
+				goto free_fences;
+
+			amdgpu_userq_fence_driver_get(fence_drv);
+
+			/* Store drm syncobj's gpu va address and value */
+			fence_info[cnt].va = fence_drv->va;
+			fence_info[cnt].value = fences[i]->seqno;
+
+			dma_fence_put(fences[i]);
+			/* Increment the actual userq fence count */
+			cnt++;
+		}
+
+		wait_info->num_fences = cnt;
+		/* Copy userq fence info to user space */
+		if (copy_to_user(u64_to_user_ptr(wait_info->out_fences),
+				 fence_info, wait_info->num_fences * sizeof(*fence_info))) {
+			r = -EFAULT;
+			goto free_fences;
+		}
+
+		kfree(fences);
+		kfree(fence_info);
+	}
+
+	drm_exec_fini(&exec);
+	for (i = 0; i < num_read_bo_handles; i++)
+		drm_gem_object_put(gobj_read[i]);
+	kfree(gobj_read);
+
+	for (i = 0; i < num_write_bo_handles; i++)
+		drm_gem_object_put(gobj_write[i]);
+	kfree(gobj_write);
+
+	kfree(timeline_points);
+	kfree(timeline_handles);
+	kfree(syncobj_handles);
+	kfree(bo_handles_write);
+	kfree(bo_handles_read);
+
+	return 0;
+
+free_fences:
+	while (num_fences-- > 0)
+		dma_fence_put(fences[num_fences]);
+	kfree(fences);
+free_fence_info:
+	kfree(fence_info);
+exec_fini:
+	drm_exec_fini(&exec);
+put_gobj_write:
+	while (wentry-- > 0)
+		drm_gem_object_put(gobj_write[wentry]);
+	kfree(gobj_write);
+put_gobj_read:
+	while (rentry-- > 0)
+		drm_gem_object_put(gobj_read[rentry]);
+	kfree(gobj_read);
+free_timeline_points:
+	kfree(timeline_points);
+free_timeline_handles:
+	kfree(timeline_handles);
+free_syncobj_handles:
+	kfree(syncobj_handles);
+free_bo_handles_write:
+	kfree(bo_handles_write);
+free_bo_handles_read:
+	kfree(bo_handles_read);
+
+	return r;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.h
new file mode 100644
index 000000000000..d76add2afc77
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.h
@@ -0,0 +1,77 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_USERQ_FENCE_H__
+#define __AMDGPU_USERQ_FENCE_H__
+
+#include <linux/types.h>
+
+#include "amdgpu_userq.h"
+
+struct amdgpu_userq_fence {
+	struct dma_fence base;
+	/*
+	 * This lock is necessary to synchronize the
+	 * userqueue dma fence operations.
+	 */
+	spinlock_t lock;
+	struct list_head link;
+	unsigned long fence_drv_array_count;
+	struct amdgpu_userq_fence_driver *fence_drv;
+	struct amdgpu_userq_fence_driver **fence_drv_array;
+};
+
+struct amdgpu_userq_fence_driver {
+	struct kref refcount;
+	u64 va;
+	u64 gpu_addr;
+	u64 *cpu_addr;
+	u64 context;
+	/*
+	 * This lock is necesaary to synchronize the access
+	 * to the fences list by the fence driver.
+	 */
+	spinlock_t fence_list_lock;
+	struct list_head fences;
+	struct amdgpu_device *adev;
+	char timeline_name[TASK_COMM_LEN];
+};
+
+int amdgpu_userq_fence_slab_init(void);
+void amdgpu_userq_fence_slab_fini(void);
+
+void amdgpu_userq_fence_driver_get(struct amdgpu_userq_fence_driver *fence_drv);
+void amdgpu_userq_fence_driver_put(struct amdgpu_userq_fence_driver *fence_drv);
+int amdgpu_userq_fence_driver_alloc(struct amdgpu_device *adev,
+				    struct amdgpu_usermode_queue *userq);
+void amdgpu_userq_fence_driver_free(struct amdgpu_usermode_queue *userq);
+void amdgpu_userq_fence_driver_process(struct amdgpu_userq_fence_driver *fence_drv);
+void amdgpu_userq_fence_driver_force_completion(struct amdgpu_usermode_queue *userq);
+void amdgpu_userq_fence_driver_destroy(struct kref *ref);
+int amdgpu_userq_signal_ioctl(struct drm_device *dev, void *data,
+			      struct drm_file *filp);
+int amdgpu_userq_wait_ioctl(struct drm_device *dev, void *data,
+			    struct drm_file *filp);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_utils.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_utils.h
new file mode 100644
index 000000000000..1e40ca3b1584
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_utils.h
@@ -0,0 +1,91 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2025 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef AMDGPU_UTILS_H_
+#define AMDGPU_UTILS_H_
+
+/* ---------- Generic 2‑bit capability attribute encoding ----------
+ * 00 INVALID, 01 RO, 10 WO, 11 RW
+ */
+enum amdgpu_cap_attr {
+	AMDGPU_CAP_ATTR_INVALID = 0,
+	AMDGPU_CAP_ATTR_RO      = 1 << 0,
+	AMDGPU_CAP_ATTR_WO      = 1 << 1,
+	AMDGPU_CAP_ATTR_RW      = (AMDGPU_CAP_ATTR_RO | AMDGPU_CAP_ATTR_WO),
+};
+
+#define AMDGPU_CAP_ATTR_BITS 2
+#define AMDGPU_CAP_ATTR_MAX  ((1U << AMDGPU_CAP_ATTR_BITS) - 1)
+
+/* Internal helper to build helpers for a given enum NAME */
+#define DECLARE_ATTR_CAP_CLASS_HELPERS(NAME)							\
+enum { NAME##_BITMAP_BITS = NAME##_COUNT * AMDGPU_CAP_ATTR_BITS };				\
+struct NAME##_caps {										\
+	DECLARE_BITMAP(bmap, NAME##_BITMAP_BITS);						\
+};												\
+static inline unsigned int NAME##_ATTR_START(enum NAME##_cap_id cap)				\
+{ return (unsigned int)cap * AMDGPU_CAP_ATTR_BITS; }						\
+static inline void NAME##_attr_init(struct NAME##_caps *c)					\
+{ if (c) bitmap_zero(c->bmap, NAME##_BITMAP_BITS); }						\
+static inline int NAME##_attr_set(struct NAME##_caps *c,					\
+				  enum NAME##_cap_id cap, enum amdgpu_cap_attr attr)		\
+{												\
+	if (!c)											\
+		return -EINVAL;									\
+	if (cap >= NAME##_COUNT)								\
+		return -EINVAL;									\
+	if ((unsigned int)attr > AMDGPU_CAP_ATTR_MAX)						\
+		return -EINVAL;									\
+	bitmap_write(c->bmap, (unsigned long)attr,						\
+			NAME##_ATTR_START(cap), AMDGPU_CAP_ATTR_BITS);				\
+	return 0;										\
+}												\
+static inline int NAME##_attr_get(const struct NAME##_caps *c,					\
+				  enum NAME##_cap_id cap, enum amdgpu_cap_attr *out)		\
+{												\
+	unsigned long v;									\
+	if (!c || !out)										\
+		return -EINVAL;									\
+	if (cap >= NAME##_COUNT)								\
+		return -EINVAL;									\
+	v = bitmap_read(c->bmap, NAME##_ATTR_START(cap), AMDGPU_CAP_ATTR_BITS);			\
+	*out = (enum amdgpu_cap_attr)v;								\
+	return 0;										\
+}												\
+static inline bool NAME##_cap_is_ro(const struct NAME##_caps *c, enum NAME##_cap_id id)		\
+{ enum amdgpu_cap_attr a; return !NAME##_attr_get(c, id, &a) && a == AMDGPU_CAP_ATTR_RO; }	\
+static inline bool NAME##_cap_is_wo(const struct NAME##_caps *c, enum NAME##_cap_id id)		\
+{ enum amdgpu_cap_attr a; return !NAME##_attr_get(c, id, &a) && a == AMDGPU_CAP_ATTR_WO; }	\
+static inline bool NAME##_cap_is_rw(const struct NAME##_caps *c, enum NAME##_cap_id id)		\
+{ enum amdgpu_cap_attr a; return !NAME##_attr_get(c, id, &a) && a == AMDGPU_CAP_ATTR_RW; }
+
+/* Element expander for enum creation */
+#define _CAP_ENUM_ELEM(x) x,
+
+/* Public macro: declare enum + helpers from an X‑macro list */
+#define DECLARE_ATTR_CAP_CLASS(NAME, LIST_MACRO)						\
+	enum NAME##_cap_id { LIST_MACRO(_CAP_ENUM_ELEM) NAME##_COUNT };				\
+	DECLARE_ATTR_CAP_CLASS_HELPERS(NAME)
+
+#endif /* AMDGPU_UTILS_H_ */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
index e19928dae8e3..5c38f0d30c87 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
@@ -30,15 +30,19 @@
 
 #include <linux/firmware.h>
 #include <linux/module.h>
-#include <drm/drmP.h>
+
 #include <drm/drm.h>
+#include <drm/drm_drv.h>
 
 #include "amdgpu.h"
 #include "amdgpu_pm.h"
 #include "amdgpu_uvd.h"
+#include "amdgpu_cs.h"
 #include "cikd.h"
 #include "uvd/uvd_4_2_d.h"
 
+#include "amdgpu_ras.h"
+
 /* 1 second timeout */
 #define UVD_IDLE_TIMEOUT	msecs_to_jiffies(1000)
 
@@ -52,12 +56,18 @@
 #define FW_1_66_16	((1 << 24) | (66 << 16) | (16 << 8))
 
 /* Firmware Names */
+#ifdef CONFIG_DRM_AMDGPU_SI
+#define FIRMWARE_TAHITI		"amdgpu/tahiti_uvd.bin"
+#define FIRMWARE_VERDE		"amdgpu/verde_uvd.bin"
+#define FIRMWARE_PITCAIRN	"amdgpu/pitcairn_uvd.bin"
+#define FIRMWARE_OLAND		"amdgpu/oland_uvd.bin"
+#endif
 #ifdef CONFIG_DRM_AMDGPU_CIK
-#define FIRMWARE_BONAIRE	"radeon/bonaire_uvd.bin"
-#define FIRMWARE_KABINI	"radeon/kabini_uvd.bin"
-#define FIRMWARE_KAVERI	"radeon/kaveri_uvd.bin"
-#define FIRMWARE_HAWAII	"radeon/hawaii_uvd.bin"
-#define FIRMWARE_MULLINS	"radeon/mullins_uvd.bin"
+#define FIRMWARE_BONAIRE	"amdgpu/bonaire_uvd.bin"
+#define FIRMWARE_KABINI	"amdgpu/kabini_uvd.bin"
+#define FIRMWARE_KAVERI	"amdgpu/kaveri_uvd.bin"
+#define FIRMWARE_HAWAII	"amdgpu/hawaii_uvd.bin"
+#define FIRMWARE_MULLINS	"amdgpu/mullins_uvd.bin"
 #endif
 #define FIRMWARE_TONGA		"amdgpu/tonga_uvd.bin"
 #define FIRMWARE_CARRIZO	"amdgpu/carrizo_uvd.bin"
@@ -66,34 +76,44 @@
 #define FIRMWARE_POLARIS10	"amdgpu/polaris10_uvd.bin"
 #define FIRMWARE_POLARIS11	"amdgpu/polaris11_uvd.bin"
 #define FIRMWARE_POLARIS12	"amdgpu/polaris12_uvd.bin"
+#define FIRMWARE_VEGAM		"amdgpu/vegam_uvd.bin"
 
 #define FIRMWARE_VEGA10		"amdgpu/vega10_uvd.bin"
+#define FIRMWARE_VEGA12		"amdgpu/vega12_uvd.bin"
+#define FIRMWARE_VEGA20		"amdgpu/vega20_uvd.bin"
 
-#define mmUVD_GPCOM_VCPU_DATA0_VEGA10 (0x03c4 + 0x7e00)
-#define mmUVD_GPCOM_VCPU_DATA1_VEGA10 (0x03c5 + 0x7e00)
-#define mmUVD_GPCOM_VCPU_CMD_VEGA10 (0x03c3 + 0x7e00)
-#define mmUVD_NO_OP_VEGA10 (0x03ff + 0x7e00)
-#define mmUVD_ENGINE_CNTL_VEGA10 (0x03c6 + 0x7e00)
+/* These are common relative offsets for all asics, from uvd_7_0_offset.h,  */
+#define UVD_GPCOM_VCPU_CMD		0x03c3
+#define UVD_GPCOM_VCPU_DATA0	0x03c4
+#define UVD_GPCOM_VCPU_DATA1	0x03c5
+#define UVD_NO_OP				0x03ff
+#define UVD_BASE_SI				0x3800
 
-/**
+/*
  * amdgpu_uvd_cs_ctx - Command submission parser context
  *
  * Used for emulating virtual memory support on UVD 4.2.
  */
 struct amdgpu_uvd_cs_ctx {
 	struct amdgpu_cs_parser *parser;
-	unsigned reg, count;
-	unsigned data0, data1;
-	unsigned idx;
-	unsigned ib_idx;
+	unsigned int reg, count;
+	unsigned int data0, data1;
+	unsigned int idx;
+	struct amdgpu_ib *ib;
 
 	/* does the IB has a msg command */
 	bool has_msg_cmd;
 
 	/* minimum buffer sizes */
-	unsigned *buf_sizes;
+	unsigned int *buf_sizes;
 };
 
+#ifdef CONFIG_DRM_AMDGPU_SI
+MODULE_FIRMWARE(FIRMWARE_TAHITI);
+MODULE_FIRMWARE(FIRMWARE_VERDE);
+MODULE_FIRMWARE(FIRMWARE_PITCAIRN);
+MODULE_FIRMWARE(FIRMWARE_OLAND);
+#endif
 #ifdef CONFIG_DRM_AMDGPU_CIK
 MODULE_FIRMWARE(FIRMWARE_BONAIRE);
 MODULE_FIRMWARE(FIRMWARE_KABINI);
@@ -108,24 +128,84 @@ MODULE_FIRMWARE(FIRMWARE_STONEY);
 MODULE_FIRMWARE(FIRMWARE_POLARIS10);
 MODULE_FIRMWARE(FIRMWARE_POLARIS11);
 MODULE_FIRMWARE(FIRMWARE_POLARIS12);
+MODULE_FIRMWARE(FIRMWARE_VEGAM);
 
 MODULE_FIRMWARE(FIRMWARE_VEGA10);
+MODULE_FIRMWARE(FIRMWARE_VEGA12);
+MODULE_FIRMWARE(FIRMWARE_VEGA20);
 
 static void amdgpu_uvd_idle_work_handler(struct work_struct *work);
+static void amdgpu_uvd_force_into_uvd_segment(struct amdgpu_bo *abo);
+
+static int amdgpu_uvd_create_msg_bo_helper(struct amdgpu_device *adev,
+					   uint32_t size,
+					   struct amdgpu_bo **bo_ptr)
+{
+	struct ttm_operation_ctx ctx = { true, false };
+	struct amdgpu_bo *bo = NULL;
+	void *addr;
+	int r;
+
+	r = amdgpu_bo_create_reserved(adev, size, PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &bo, NULL, &addr);
+	if (r)
+		return r;
+
+	if (adev->uvd.address_64_bit)
+		goto succ;
+
+	amdgpu_bo_kunmap(bo);
+	amdgpu_bo_unpin(bo);
+	amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_VRAM);
+	amdgpu_uvd_force_into_uvd_segment(bo);
+	r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+	if (r)
+		goto err;
+	r = amdgpu_bo_pin(bo, AMDGPU_GEM_DOMAIN_VRAM);
+	if (r)
+		goto err_pin;
+	r = amdgpu_bo_kmap(bo, &addr);
+	if (r)
+		goto err_kmap;
+succ:
+	amdgpu_bo_unreserve(bo);
+	*bo_ptr = bo;
+	return 0;
+err_kmap:
+	amdgpu_bo_unpin(bo);
+err_pin:
+err:
+	amdgpu_bo_unreserve(bo);
+	amdgpu_bo_unref(&bo);
+	return r;
+}
 
 int amdgpu_uvd_sw_init(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *ring;
-	struct amd_sched_rq *rq;
 	unsigned long bo_size;
 	const char *fw_name;
 	const struct common_firmware_header *hdr;
-	unsigned version_major, version_minor, family_id;
-	int i, r;
+	unsigned int family_id;
+	int i, j, r;
 
 	INIT_DELAYED_WORK(&adev->uvd.idle_work, amdgpu_uvd_idle_work_handler);
 
 	switch (adev->asic_type) {
+#ifdef CONFIG_DRM_AMDGPU_SI
+	case CHIP_TAHITI:
+		fw_name = FIRMWARE_TAHITI;
+		break;
+	case CHIP_VERDE:
+		fw_name = FIRMWARE_VERDE;
+		break;
+	case CHIP_PITCAIRN:
+		fw_name = FIRMWARE_PITCAIRN;
+		break;
+	case CHIP_OLAND:
+		fw_name = FIRMWARE_OLAND;
+		break;
+#endif
 #ifdef CONFIG_DRM_AMDGPU_CIK
 	case CHIP_BONAIRE:
 		fw_name = FIRMWARE_BONAIRE;
@@ -161,29 +241,30 @@ int amdgpu_uvd_sw_init(struct amdgpu_device *adev)
 	case CHIP_POLARIS11:
 		fw_name = FIRMWARE_POLARIS11;
 		break;
+	case CHIP_POLARIS12:
+		fw_name = FIRMWARE_POLARIS12;
+		break;
 	case CHIP_VEGA10:
 		fw_name = FIRMWARE_VEGA10;
 		break;
-	case CHIP_POLARIS12:
-		fw_name = FIRMWARE_POLARIS12;
+	case CHIP_VEGA12:
+		fw_name = FIRMWARE_VEGA12;
+		break;
+	case CHIP_VEGAM:
+		fw_name = FIRMWARE_VEGAM;
+		break;
+	case CHIP_VEGA20:
+		fw_name = FIRMWARE_VEGA20;
 		break;
 	default:
 		return -EINVAL;
 	}
 
-	r = request_firmware(&adev->uvd.fw, fw_name, adev->dev);
-	if (r) {
-		dev_err(adev->dev, "amdgpu_uvd: Can't load firmware \"%s\"\n",
-			fw_name);
-		return r;
-	}
-
-	r = amdgpu_ucode_validate(adev->uvd.fw);
+	r = amdgpu_ucode_request(adev, &adev->uvd.fw, AMDGPU_UCODE_REQUIRED, "%s", fw_name);
 	if (r) {
 		dev_err(adev->dev, "amdgpu_uvd: Can't validate firmware \"%s\"\n",
 			fw_name);
-		release_firmware(adev->uvd.fw);
-		adev->uvd.fw = NULL;
+		amdgpu_ucode_release(&adev->uvd.fw);
 		return r;
 	}
 
@@ -192,50 +273,65 @@ int amdgpu_uvd_sw_init(struct amdgpu_device *adev)
 
 	hdr = (const struct common_firmware_header *)adev->uvd.fw->data;
 	family_id = le32_to_cpu(hdr->ucode_version) & 0xff;
-	version_major = (le32_to_cpu(hdr->ucode_version) >> 24) & 0xff;
-	version_minor = (le32_to_cpu(hdr->ucode_version) >> 8) & 0xff;
-	DRM_INFO("Found UVD firmware Version: %hu.%hu Family ID: %hu\n",
-		version_major, version_minor, family_id);
-
-	/*
-	 * Limit the number of UVD handles depending on microcode major
-	 * and minor versions. The firmware version which has 40 UVD
-	 * instances support is 1.80. So all subsequent versions should
-	 * also have the same support.
-	 */
-	if ((version_major > 0x01) ||
-	    ((version_major == 0x01) && (version_minor >= 0x50)))
-		adev->uvd.max_handles = AMDGPU_MAX_UVD_HANDLES;
 
-	adev->uvd.fw_version = ((version_major << 24) | (version_minor << 16) |
-				(family_id << 8));
+	if (adev->asic_type < CHIP_VEGA20) {
+		unsigned int version_major, version_minor;
+
+		version_major = (le32_to_cpu(hdr->ucode_version) >> 24) & 0xff;
+		version_minor = (le32_to_cpu(hdr->ucode_version) >> 8) & 0xff;
+		DRM_INFO("Found UVD firmware Version: %u.%u Family ID: %u\n",
+			version_major, version_minor, family_id);
+
+		/*
+		 * Limit the number of UVD handles depending on microcode major
+		 * and minor versions. The firmware version which has 40 UVD
+		 * instances support is 1.80. So all subsequent versions should
+		 * also have the same support.
+		 */
+		if ((version_major > 0x01) ||
+		    ((version_major == 0x01) && (version_minor >= 0x50)))
+			adev->uvd.max_handles = AMDGPU_MAX_UVD_HANDLES;
+
+		adev->uvd.fw_version = ((version_major << 24) | (version_minor << 16) |
+					(family_id << 8));
+
+		if ((adev->asic_type == CHIP_POLARIS10 ||
+		     adev->asic_type == CHIP_POLARIS11) &&
+		    (adev->uvd.fw_version < FW_1_66_16))
+			DRM_ERROR("POLARIS10/11 UVD firmware version %u.%u is too old.\n",
+				  version_major, version_minor);
+	} else {
+		unsigned int enc_major, enc_minor, dec_minor;
 
-	if ((adev->asic_type == CHIP_POLARIS10 ||
-	     adev->asic_type == CHIP_POLARIS11) &&
-	    (adev->uvd.fw_version < FW_1_66_16))
-		DRM_ERROR("POLARIS10/11 UVD firmware version %hu.%hu is too old.\n",
-			  version_major, version_minor);
+		dec_minor = (le32_to_cpu(hdr->ucode_version) >> 8) & 0xff;
+		enc_minor = (le32_to_cpu(hdr->ucode_version) >> 24) & 0x3f;
+		enc_major = (le32_to_cpu(hdr->ucode_version) >> 30) & 0x3;
+		DRM_INFO("Found UVD firmware ENC: %u.%u DEC: .%u Family ID: %u\n",
+			enc_major, enc_minor, dec_minor, family_id);
+
+		adev->uvd.max_handles = AMDGPU_MAX_UVD_HANDLES;
+
+		adev->uvd.fw_version = le32_to_cpu(hdr->ucode_version);
+	}
 
 	bo_size = AMDGPU_UVD_STACK_SIZE + AMDGPU_UVD_HEAP_SIZE
 		  +  AMDGPU_UVD_SESSION_SIZE * adev->uvd.max_handles;
 	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
 		bo_size += AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8);
 
-	r = amdgpu_bo_create_kernel(adev, bo_size, PAGE_SIZE,
-				    AMDGPU_GEM_DOMAIN_VRAM, &adev->uvd.vcpu_bo,
-				    &adev->uvd.gpu_addr, &adev->uvd.cpu_addr);
-	if (r) {
-		dev_err(adev->dev, "(%d) failed to allocate UVD bo\n", r);
-		return r;
-	}
-
-	ring = &adev->uvd.ring;
-	rq = &ring->sched.sched_rq[AMD_SCHED_PRIORITY_NORMAL];
-	r = amd_sched_entity_init(&ring->sched, &adev->uvd.entity,
-				  rq, amdgpu_sched_jobs);
-	if (r != 0) {
-		DRM_ERROR("Failed setting up UVD run queue.\n");
-		return r;
+	for (j = 0; j < adev->uvd.num_uvd_inst; j++) {
+		if (adev->uvd.harvest_config & (1 << j))
+			continue;
+		r = amdgpu_bo_create_kernel(adev, bo_size, PAGE_SIZE,
+					    AMDGPU_GEM_DOMAIN_VRAM |
+					    AMDGPU_GEM_DOMAIN_GTT,
+					    &adev->uvd.inst[j].vcpu_bo,
+					    &adev->uvd.inst[j].gpu_addr,
+					    &adev->uvd.inst[j].cpu_addr);
+		if (r) {
+			dev_err(adev->dev, "(%d) failed to allocate UVD bo\n", r);
+			return r;
+		}
 	}
 
 	for (i = 0; i < adev->uvd.max_handles; ++i) {
@@ -244,9 +340,13 @@ int amdgpu_uvd_sw_init(struct amdgpu_device *adev)
 	}
 
 	/* from uvd v5.0 HW addressing capacity increased to 64 bits */
-	if (!amdgpu_ip_block_version_cmp(adev, AMD_IP_BLOCK_TYPE_UVD, 5, 0))
+	if (!amdgpu_device_ip_block_version_cmp(adev, AMD_IP_BLOCK_TYPE_UVD, 5, 0))
 		adev->uvd.address_64_bit = true;
 
+	r = amdgpu_uvd_create_msg_bo_helper(adev, 128 << 10, &adev->uvd.ib_bo);
+	if (r)
+		return r;
+
 	switch (adev->asic_type) {
 	case CHIP_TONGA:
 		adev->uvd.use_ctx_buf = adev->uvd.fw_version >= FW_1_65_10;
@@ -269,98 +369,169 @@ int amdgpu_uvd_sw_init(struct amdgpu_device *adev)
 
 int amdgpu_uvd_sw_fini(struct amdgpu_device *adev)
 {
-	kfree(adev->uvd.saved_bo);
+	void *addr = amdgpu_bo_kptr(adev->uvd.ib_bo);
+	int i, j;
+
+	drm_sched_entity_destroy(&adev->uvd.entity);
 
-	amd_sched_entity_fini(&adev->uvd.ring.sched, &adev->uvd.entity);
+	for (j = 0; j < adev->uvd.num_uvd_inst; ++j) {
+		if (adev->uvd.harvest_config & (1 << j))
+			continue;
+		kvfree(adev->uvd.inst[j].saved_bo);
 
-	amdgpu_bo_free_kernel(&adev->uvd.vcpu_bo,
-			      &adev->uvd.gpu_addr,
-			      (void **)&adev->uvd.cpu_addr);
+		amdgpu_bo_free_kernel(&adev->uvd.inst[j].vcpu_bo,
+				      &adev->uvd.inst[j].gpu_addr,
+				      (void **)&adev->uvd.inst[j].cpu_addr);
 
-	amdgpu_ring_fini(&adev->uvd.ring);
+		amdgpu_ring_fini(&adev->uvd.inst[j].ring);
 
-	release_firmware(adev->uvd.fw);
+		for (i = 0; i < AMDGPU_MAX_UVD_ENC_RINGS; ++i)
+			amdgpu_ring_fini(&adev->uvd.inst[j].ring_enc[i]);
+	}
+	amdgpu_bo_free_kernel(&adev->uvd.ib_bo, NULL, &addr);
+	amdgpu_ucode_release(&adev->uvd.fw);
 
 	return 0;
 }
 
-int amdgpu_uvd_suspend(struct amdgpu_device *adev)
+/**
+ * amdgpu_uvd_entity_init - init entity
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: amdgpu_ring pointer to check
+ *
+ * Initialize the entity used for handle management in the kernel driver.
+ */
+int amdgpu_uvd_entity_init(struct amdgpu_device *adev, struct amdgpu_ring *ring)
+{
+	if (ring == &adev->uvd.inst[0].ring) {
+		struct drm_gpu_scheduler *sched = &ring->sched;
+		int r;
+
+		r = drm_sched_entity_init(&adev->uvd.entity, DRM_SCHED_PRIORITY_NORMAL,
+					  &sched, 1, NULL);
+		if (r) {
+			DRM_ERROR("Failed setting up UVD kernel entity.\n");
+			return r;
+		}
+	}
+
+	return 0;
+}
+
+int amdgpu_uvd_prepare_suspend(struct amdgpu_device *adev)
 {
-	unsigned size;
+	unsigned int size;
 	void *ptr;
-	int i;
+	int i, j, idx;
 
-	if (adev->uvd.vcpu_bo == NULL)
-		return 0;
+	cancel_delayed_work_sync(&adev->uvd.idle_work);
 
-	for (i = 0; i < adev->uvd.max_handles; ++i)
-		if (atomic_read(&adev->uvd.handles[i]))
-			break;
+	/* only valid for physical mode */
+	if (adev->asic_type < CHIP_POLARIS10) {
+		for (i = 0; i < adev->uvd.max_handles; ++i)
+			if (atomic_read(&adev->uvd.handles[i]))
+				break;
 
-	if (i == AMDGPU_MAX_UVD_HANDLES)
-		return 0;
+		if (i == adev->uvd.max_handles)
+			return 0;
+	}
 
-	cancel_delayed_work_sync(&adev->uvd.idle_work);
+	for (j = 0; j < adev->uvd.num_uvd_inst; ++j) {
+		if (adev->uvd.harvest_config & (1 << j))
+			continue;
+		if (adev->uvd.inst[j].vcpu_bo == NULL)
+			continue;
+
+		size = amdgpu_bo_size(adev->uvd.inst[j].vcpu_bo);
+		ptr = adev->uvd.inst[j].cpu_addr;
+
+		adev->uvd.inst[j].saved_bo = kvmalloc(size, GFP_KERNEL);
+		if (!adev->uvd.inst[j].saved_bo)
+			return -ENOMEM;
 
-	size = amdgpu_bo_size(adev->uvd.vcpu_bo);
-	ptr = adev->uvd.cpu_addr;
+		if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+			/* re-write 0 since err_event_athub will corrupt VCPU buffer */
+			if (amdgpu_ras_intr_triggered())
+				memset(adev->uvd.inst[j].saved_bo, 0, size);
+			else
+				memcpy_fromio(adev->uvd.inst[j].saved_bo, ptr, size);
+
+			drm_dev_exit(idx);
+		}
+	}
 
-	adev->uvd.saved_bo = kmalloc(size, GFP_KERNEL);
-	if (!adev->uvd.saved_bo)
-		return -ENOMEM;
+	return 0;
+}
 
-	memcpy_fromio(adev->uvd.saved_bo, ptr, size);
+int amdgpu_uvd_suspend(struct amdgpu_device *adev)
+{
+	if (amdgpu_ras_intr_triggered())
+		DRM_WARN("UVD VCPU state may lost due to RAS ERREVENT_ATHUB_INTERRUPT\n");
 
 	return 0;
 }
 
 int amdgpu_uvd_resume(struct amdgpu_device *adev)
 {
-	unsigned size;
+	unsigned int size;
 	void *ptr;
+	int i, idx;
 
-	if (adev->uvd.vcpu_bo == NULL)
-		return -EINVAL;
+	for (i = 0; i < adev->uvd.num_uvd_inst; i++) {
+		if (adev->uvd.harvest_config & (1 << i))
+			continue;
+		if (adev->uvd.inst[i].vcpu_bo == NULL)
+			return -EINVAL;
 
-	size = amdgpu_bo_size(adev->uvd.vcpu_bo);
-	ptr = adev->uvd.cpu_addr;
+		size = amdgpu_bo_size(adev->uvd.inst[i].vcpu_bo);
+		ptr = adev->uvd.inst[i].cpu_addr;
 
-	if (adev->uvd.saved_bo != NULL) {
-		memcpy_toio(ptr, adev->uvd.saved_bo, size);
-		kfree(adev->uvd.saved_bo);
-		adev->uvd.saved_bo = NULL;
-	} else {
-		const struct common_firmware_header *hdr;
-		unsigned offset;
-
-		hdr = (const struct common_firmware_header *)adev->uvd.fw->data;
-		if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
-			offset = le32_to_cpu(hdr->ucode_array_offset_bytes);
-			memcpy_toio(adev->uvd.cpu_addr, adev->uvd.fw->data + offset,
-				    le32_to_cpu(hdr->ucode_size_bytes));
-			size -= le32_to_cpu(hdr->ucode_size_bytes);
-			ptr += le32_to_cpu(hdr->ucode_size_bytes);
+		if (adev->uvd.inst[i].saved_bo != NULL) {
+			if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+				memcpy_toio(ptr, adev->uvd.inst[i].saved_bo, size);
+				drm_dev_exit(idx);
+			}
+			kvfree(adev->uvd.inst[i].saved_bo);
+			adev->uvd.inst[i].saved_bo = NULL;
+		} else {
+			const struct common_firmware_header *hdr;
+			unsigned int offset;
+
+			hdr = (const struct common_firmware_header *)adev->uvd.fw->data;
+			if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
+				offset = le32_to_cpu(hdr->ucode_array_offset_bytes);
+				if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+					memcpy_toio(adev->uvd.inst[i].cpu_addr, adev->uvd.fw->data + offset,
+						    le32_to_cpu(hdr->ucode_size_bytes));
+					drm_dev_exit(idx);
+				}
+				size -= le32_to_cpu(hdr->ucode_size_bytes);
+				ptr += le32_to_cpu(hdr->ucode_size_bytes);
+			}
+			memset_io(ptr, 0, size);
+			/* to restore uvd fence seq */
+			amdgpu_fence_driver_force_completion(&adev->uvd.inst[i].ring);
 		}
-		memset_io(ptr, 0, size);
 	}
-
 	return 0;
 }
 
 void amdgpu_uvd_free_handles(struct amdgpu_device *adev, struct drm_file *filp)
 {
-	struct amdgpu_ring *ring = &adev->uvd.ring;
+	struct amdgpu_ring *ring = &adev->uvd.inst[0].ring;
 	int i, r;
 
 	for (i = 0; i < adev->uvd.max_handles; ++i) {
 		uint32_t handle = atomic_read(&adev->uvd.handles[i]);
+
 		if (handle != 0 && adev->uvd.filp[i] == filp) {
 			struct dma_fence *fence;
 
-			r = amdgpu_uvd_get_destroy_msg(ring, handle,
-						       false, &fence);
+			r = amdgpu_uvd_get_destroy_msg(ring, handle, false,
+						       &fence);
 			if (r) {
-				DRM_ERROR("Error destroying UVD (%d)!\n", r);
+				DRM_ERROR("Error destroying UVD %d!\n", r);
 				continue;
 			}
 
@@ -376,9 +547,12 @@ void amdgpu_uvd_free_handles(struct amdgpu_device *adev, struct drm_file *filp)
 static void amdgpu_uvd_force_into_uvd_segment(struct amdgpu_bo *abo)
 {
 	int i;
+
 	for (i = 0; i < abo->placement.num_placement; ++i) {
 		abo->placements[i].fpfn = 0 >> PAGE_SHIFT;
 		abo->placements[i].lpfn = (256 * 1024 * 1024) >> PAGE_SHIFT;
+		if (abo->placements[i].mem_type == TTM_PL_VRAM)
+			abo->placements[i].flags |= TTM_PL_FLAG_CONTIGUOUS;
 	}
 }
 
@@ -387,8 +561,8 @@ static u64 amdgpu_uvd_get_addr_from_ctx(struct amdgpu_uvd_cs_ctx *ctx)
 	uint32_t lo, hi;
 	uint64_t addr;
 
-	lo = amdgpu_get_ib_value(ctx->parser, ctx->ib_idx, ctx->data0);
-	hi = amdgpu_get_ib_value(ctx->parser, ctx->ib_idx, ctx->data1);
+	lo = amdgpu_ib_get_value(ctx->ib, ctx->data0);
+	hi = amdgpu_ib_get_value(ctx->ib, ctx->data1);
 	addr = ((uint64_t)lo) | (((uint64_t)hi) << 32);
 
 	return addr;
@@ -404,29 +578,31 @@ static u64 amdgpu_uvd_get_addr_from_ctx(struct amdgpu_uvd_cs_ctx *ctx)
  */
 static int amdgpu_uvd_cs_pass1(struct amdgpu_uvd_cs_ctx *ctx)
 {
+	struct ttm_operation_ctx tctx = { false, false };
 	struct amdgpu_bo_va_mapping *mapping;
 	struct amdgpu_bo *bo;
 	uint32_t cmd;
 	uint64_t addr = amdgpu_uvd_get_addr_from_ctx(ctx);
 	int r = 0;
 
-	mapping = amdgpu_cs_find_mapping(ctx->parser, addr, &bo);
-	if (mapping == NULL) {
-		DRM_ERROR("Can't find BO for addr 0x%08Lx\n", addr);
-		return -EINVAL;
+	r = amdgpu_cs_find_mapping(ctx->parser, addr, &bo, &mapping);
+	if (r) {
+		DRM_ERROR("Can't find BO for addr 0x%08llx\n", addr);
+		return r;
 	}
 
 	if (!ctx->parser->adev->uvd.address_64_bit) {
 		/* check if it's a message or feedback command */
-		cmd = amdgpu_get_ib_value(ctx->parser, ctx->ib_idx, ctx->idx) >> 1;
+		cmd = amdgpu_ib_get_value(ctx->ib, ctx->idx) >> 1;
 		if (cmd == 0x0 || cmd == 0x3) {
 			/* yes, force it into VRAM */
 			uint32_t domain = AMDGPU_GEM_DOMAIN_VRAM;
-			amdgpu_ttm_placement_from_domain(bo, domain);
+
+			amdgpu_bo_placement_from_domain(bo, domain);
 		}
 		amdgpu_uvd_force_into_uvd_segment(bo);
 
-		r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false);
+		r = ttm_bo_validate(&bo->tbo, &bo->placement, &tctx);
 	}
 
 	return r;
@@ -435,27 +611,28 @@ static int amdgpu_uvd_cs_pass1(struct amdgpu_uvd_cs_ctx *ctx)
 /**
  * amdgpu_uvd_cs_msg_decode - handle UVD decode message
  *
+ * @adev: amdgpu_device pointer
  * @msg: pointer to message structure
- * @buf_sizes: returned buffer sizes
+ * @buf_sizes: placeholder to put the different buffer lengths
  *
  * Peek into the decode message and calculate the necessary buffer sizes.
  */
 static int amdgpu_uvd_cs_msg_decode(struct amdgpu_device *adev, uint32_t *msg,
-	unsigned buf_sizes[])
+	unsigned int buf_sizes[])
 {
-	unsigned stream_type = msg[4];
-	unsigned width = msg[6];
-	unsigned height = msg[7];
-	unsigned dpb_size = msg[9];
-	unsigned pitch = msg[28];
-	unsigned level = msg[57];
+	unsigned int stream_type = msg[4];
+	unsigned int width = msg[6];
+	unsigned int height = msg[7];
+	unsigned int dpb_size = msg[9];
+	unsigned int pitch = msg[28];
+	unsigned int level = msg[57];
 
-	unsigned width_in_mb = width / 16;
-	unsigned height_in_mb = ALIGN(height / 16, 2);
-	unsigned fs_in_mb = width_in_mb * height_in_mb;
+	unsigned int width_in_mb = width / 16;
+	unsigned int height_in_mb = ALIGN(height / 16, 2);
+	unsigned int fs_in_mb = width_in_mb * height_in_mb;
 
-	unsigned image_size, tmp, min_dpb_size, num_dpb_buffer;
-	unsigned min_ctx_size = ~0;
+	unsigned int image_size, tmp, min_dpb_size, num_dpb_buffer;
+	unsigned int min_ctx_size = ~0;
 
 	image_size = width * height;
 	image_size += image_size / 2;
@@ -463,7 +640,7 @@ static int amdgpu_uvd_cs_msg_decode(struct amdgpu_device *adev, uint32_t *msg,
 
 	switch (stream_type) {
 	case 0: /* H264 */
-		switch(level) {
+		switch (level) {
 		case 30:
 			num_dpb_buffer = 8100 / fs_in_mb;
 			break;
@@ -541,7 +718,7 @@ static int amdgpu_uvd_cs_msg_decode(struct amdgpu_device *adev, uint32_t *msg,
 		break;
 
 	case 7: /* H264 Perf */
-		switch(level) {
+		switch (level) {
 		case 30:
 			num_dpb_buffer = 8100 / fs_in_mb;
 			break;
@@ -574,7 +751,7 @@ static int amdgpu_uvd_cs_msg_decode(struct amdgpu_device *adev, uint32_t *msg,
 		/* reference picture buffer */
 		min_dpb_size = image_size * num_dpb_buffer;
 
-		if (!adev->uvd.use_ctx_buf){
+		if (!adev->uvd.use_ctx_buf) {
 			/* macroblock context buffer */
 			min_dpb_size +=
 				width_in_mb * height_in_mb * num_dpb_buffer * 192;
@@ -621,6 +798,8 @@ static int amdgpu_uvd_cs_msg_decode(struct amdgpu_device *adev, uint32_t *msg,
 	buf_sizes[0x1] = dpb_size;
 	buf_sizes[0x2] = image_size;
 	buf_sizes[0x4] = min_ctx_size;
+	/* store image width to adjust nb memory pstate */
+	adev->uvd.decode_image_width = width;
 	return 0;
 }
 
@@ -635,7 +814,7 @@ static int amdgpu_uvd_cs_msg_decode(struct amdgpu_device *adev, uint32_t *msg,
  * Make sure that we don't open up to many sessions.
  */
 static int amdgpu_uvd_cs_msg(struct amdgpu_uvd_cs_ctx *ctx,
-			     struct amdgpu_bo *bo, unsigned offset)
+			     struct amdgpu_bo *bo, unsigned int offset)
 {
 	struct amdgpu_device *adev = ctx->parser->adev;
 	int32_t *msg, msg_type, handle;
@@ -650,7 +829,7 @@ static int amdgpu_uvd_cs_msg(struct amdgpu_uvd_cs_ctx *ctx,
 
 	r = amdgpu_bo_kmap(bo, &ptr);
 	if (r) {
-		DRM_ERROR("Failed mapping the UVD message (%ld)!\n", r);
+		DRM_ERROR("Failed mapping the UVD) message (%ld)!\n", r);
 		return r;
 	}
 
@@ -660,6 +839,7 @@ static int amdgpu_uvd_cs_msg(struct amdgpu_uvd_cs_ctx *ctx,
 	handle = msg[2];
 
 	if (handle == 0) {
+		amdgpu_bo_kunmap(bo);
 		DRM_ERROR("Invalid UVD handle!\n");
 		return -EINVAL;
 	}
@@ -672,7 +852,8 @@ static int amdgpu_uvd_cs_msg(struct amdgpu_uvd_cs_ctx *ctx,
 		/* try to alloc a new handle */
 		for (i = 0; i < adev->uvd.max_handles; ++i) {
 			if (atomic_read(&adev->uvd.handles[i]) == handle) {
-				DRM_ERROR("Handle 0x%x already in use!\n", handle);
+				DRM_ERROR(")Handle 0x%x already in use!\n",
+					  handle);
 				return -EINVAL;
 			}
 
@@ -715,9 +896,9 @@ static int amdgpu_uvd_cs_msg(struct amdgpu_uvd_cs_ctx *ctx,
 
 	default:
 		DRM_ERROR("Illegal UVD message type (%d)!\n", msg_type);
-		return -EINVAL;
 	}
-	BUG();
+
+	amdgpu_bo_kunmap(bo);
 	return -EINVAL;
 }
 
@@ -737,10 +918,10 @@ static int amdgpu_uvd_cs_pass2(struct amdgpu_uvd_cs_ctx *ctx)
 	uint64_t addr = amdgpu_uvd_get_addr_from_ctx(ctx);
 	int r;
 
-	mapping = amdgpu_cs_find_mapping(ctx->parser, addr, &bo);
-	if (mapping == NULL) {
-		DRM_ERROR("Can't find BO for addr 0x%08Lx\n", addr);
-		return -EINVAL;
+	r = amdgpu_cs_find_mapping(ctx->parser, addr, &bo, &mapping);
+	if (r) {
+		DRM_ERROR("Can't find BO for addr 0x%08llx\n", addr);
+		return r;
 	}
 
 	start = amdgpu_bo_gpu_offset(bo);
@@ -751,16 +932,14 @@ static int amdgpu_uvd_cs_pass2(struct amdgpu_uvd_cs_ctx *ctx)
 	addr -= mapping->start * AMDGPU_GPU_PAGE_SIZE;
 	start += addr;
 
-	amdgpu_set_ib_value(ctx->parser, ctx->ib_idx, ctx->data0,
-			    lower_32_bits(start));
-	amdgpu_set_ib_value(ctx->parser, ctx->ib_idx, ctx->data1,
-			    upper_32_bits(start));
+	amdgpu_ib_set_value(ctx->ib, ctx->data0, lower_32_bits(start));
+	amdgpu_ib_set_value(ctx->ib, ctx->data1, upper_32_bits(start));
 
-	cmd = amdgpu_get_ib_value(ctx->parser, ctx->ib_idx, ctx->idx) >> 1;
+	cmd = amdgpu_ib_get_value(ctx->ib, ctx->idx) >> 1;
 	if (cmd < 0x4) {
 		if ((end - start) < ctx->buf_sizes[cmd]) {
 			DRM_ERROR("buffer (%d) to small (%d / %d)!\n", cmd,
-				  (unsigned)(end - start),
+				  (unsigned int)(end - start),
 				  ctx->buf_sizes[cmd]);
 			return -EINVAL;
 		}
@@ -768,7 +947,7 @@ static int amdgpu_uvd_cs_pass2(struct amdgpu_uvd_cs_ctx *ctx)
 	} else if (cmd == 0x206) {
 		if ((end - start) < ctx->buf_sizes[4]) {
 			DRM_ERROR("buffer (%d) to small (%d / %d)!\n", cmd,
-					  (unsigned)(end - start),
+					  (unsigned int)(end - start),
 					  ctx->buf_sizes[4]);
 			return -EINVAL;
 		}
@@ -779,14 +958,14 @@ static int amdgpu_uvd_cs_pass2(struct amdgpu_uvd_cs_ctx *ctx)
 
 	if (!ctx->parser->adev->uvd.address_64_bit) {
 		if ((start >> 28) != ((end - 1) >> 28)) {
-			DRM_ERROR("reloc %LX-%LX crossing 256MB boundary!\n",
+			DRM_ERROR("reloc %llx-%llx crossing 256MB boundary!\n",
 				  start, end);
 			return -EINVAL;
 		}
 
 		if ((cmd == 0 || cmd == 0x3) &&
-		    (start >> 28) != (ctx->parser->adev->uvd.gpu_addr >> 28)) {
-			DRM_ERROR("msg/fb buffer %LX-%LX out of 256MB segment!\n",
+		    (start >> 28) != (ctx->parser->adev->uvd.inst->gpu_addr >> 28)) {
+			DRM_ERROR("msg/fb buffer %llx-%llx out of 256MB segment!\n",
 				  start, end);
 			return -EINVAL;
 		}
@@ -816,14 +995,13 @@ static int amdgpu_uvd_cs_pass2(struct amdgpu_uvd_cs_ctx *ctx)
 static int amdgpu_uvd_cs_reg(struct amdgpu_uvd_cs_ctx *ctx,
 			     int (*cb)(struct amdgpu_uvd_cs_ctx *ctx))
 {
-	struct amdgpu_ib *ib = &ctx->parser->job->ibs[ctx->ib_idx];
 	int i, r;
 
 	ctx->idx++;
 	for (i = 0; i <= ctx->count; ++i) {
-		unsigned reg = ctx->reg + i;
+		unsigned int reg = ctx->reg + i;
 
-		if (ctx->idx >= ib->length_dw) {
+		if (ctx->idx >= ctx->ib->length_dw) {
 			DRM_ERROR("Register command after end of CS!\n");
 			return -EINVAL;
 		}
@@ -863,12 +1041,12 @@ static int amdgpu_uvd_cs_reg(struct amdgpu_uvd_cs_ctx *ctx,
 static int amdgpu_uvd_cs_packets(struct amdgpu_uvd_cs_ctx *ctx,
 				 int (*cb)(struct amdgpu_uvd_cs_ctx *ctx))
 {
-	struct amdgpu_ib *ib = &ctx->parser->job->ibs[ctx->ib_idx];
 	int r;
 
-	for (ctx->idx = 0 ; ctx->idx < ib->length_dw; ) {
-		uint32_t cmd = amdgpu_get_ib_value(ctx->parser, ctx->ib_idx, ctx->idx);
-		unsigned type = CP_PACKET_GET_TYPE(cmd);
+	for (ctx->idx = 0 ; ctx->idx < ctx->ib->length_dw; ) {
+		uint32_t cmd = amdgpu_ib_get_value(ctx->ib, ctx->idx);
+		unsigned int type = CP_PACKET_GET_TYPE(cmd);
+
 		switch (type) {
 		case PACKET_TYPE0:
 			ctx->reg = CP_PACKET0_GET_REG(cmd);
@@ -892,24 +1070,26 @@ static int amdgpu_uvd_cs_packets(struct amdgpu_uvd_cs_ctx *ctx,
  * amdgpu_uvd_ring_parse_cs - UVD command submission parser
  *
  * @parser: Command submission parser context
+ * @job: the job to parse
+ * @ib: the IB to patch
  *
  * Parse the command stream, patch in addresses as necessary.
  */
-int amdgpu_uvd_ring_parse_cs(struct amdgpu_cs_parser *parser, uint32_t ib_idx)
+int amdgpu_uvd_ring_parse_cs(struct amdgpu_cs_parser *parser,
+			     struct amdgpu_job *job,
+			     struct amdgpu_ib *ib)
 {
 	struct amdgpu_uvd_cs_ctx ctx = {};
-	unsigned buf_sizes[] = {
+	unsigned int buf_sizes[] = {
 		[0x00000000]	=	2048,
 		[0x00000001]	=	0xFFFFFFFF,
 		[0x00000002]	=	0xFFFFFFFF,
 		[0x00000003]	=	2048,
 		[0x00000004]	=	0xFFFFFFFF,
 	};
-	struct amdgpu_ib *ib = &parser->job->ibs[ib_idx];
 	int r;
 
-	parser->job->vm = NULL;
-	ib->gpu_addr = amdgpu_sa_bo_gpu_addr(ib->sa_bo);
+	job->vm = NULL;
 
 	if (ib->length_dw % 16) {
 		DRM_ERROR("UVD IB length (%d) not 16 dwords aligned!\n",
@@ -917,13 +1097,9 @@ int amdgpu_uvd_ring_parse_cs(struct amdgpu_cs_parser *parser, uint32_t ib_idx)
 		return -EINVAL;
 	}
 
-	r = amdgpu_cs_sysvm_access_required(parser);
-	if (r)
-		return r;
-
 	ctx.parser = parser;
 	ctx.buf_sizes = buf_sizes;
-	ctx.ib_idx = ib_idx;
+	ctx.ib = ib;
 
 	/* first round only required on chips without UVD 64 bit address support */
 	if (!parser->adev->uvd.address_64_bit) {
@@ -949,51 +1125,31 @@ int amdgpu_uvd_ring_parse_cs(struct amdgpu_cs_parser *parser, uint32_t ib_idx)
 static int amdgpu_uvd_send_msg(struct amdgpu_ring *ring, struct amdgpu_bo *bo,
 			       bool direct, struct dma_fence **fence)
 {
-	struct ttm_validate_buffer tv;
-	struct ww_acquire_ctx ticket;
-	struct list_head head;
+	struct amdgpu_device *adev = ring->adev;
+	struct dma_fence *f = NULL;
+	uint32_t offset, data[4];
 	struct amdgpu_job *job;
 	struct amdgpu_ib *ib;
-	struct dma_fence *f = NULL;
-	struct amdgpu_device *adev = ring->adev;
 	uint64_t addr;
-	uint32_t data[4];
 	int i, r;
 
-	memset(&tv, 0, sizeof(tv));
-	tv.bo = &bo->tbo;
-
-	INIT_LIST_HEAD(&head);
-	list_add(&tv.head, &head);
-
-	r = ttm_eu_reserve_buffers(&ticket, &head, true, NULL);
+	r = amdgpu_job_alloc_with_ib(ring->adev, &adev->uvd.entity,
+				     AMDGPU_FENCE_OWNER_UNDEFINED,
+				     64, direct ? AMDGPU_IB_POOL_DIRECT :
+				     AMDGPU_IB_POOL_DELAYED, &job,
+				     AMDGPU_KERNEL_JOB_ID_VCN_RING_TEST);
 	if (r)
 		return r;
 
-	if (!ring->adev->uvd.address_64_bit) {
-		amdgpu_ttm_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_VRAM);
-		amdgpu_uvd_force_into_uvd_segment(bo);
-	}
+	if (adev->asic_type >= CHIP_VEGA10)
+		offset = adev->reg_offset[UVD_HWIP][ring->me][1];
+	else
+		offset = UVD_BASE_SI;
 
-	r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false);
-	if (r)
-		goto err;
-
-	r = amdgpu_job_alloc_with_ib(adev, 64, &job);
-	if (r)
-		goto err;
-
-	if (adev->asic_type >= CHIP_VEGA10) {
-		data[0] = PACKET0(mmUVD_GPCOM_VCPU_DATA0_VEGA10, 0);
-		data[1] = PACKET0(mmUVD_GPCOM_VCPU_DATA1_VEGA10, 0);
-		data[2] = PACKET0(mmUVD_GPCOM_VCPU_CMD_VEGA10, 0);
-		data[3] = PACKET0(mmUVD_NO_OP_VEGA10, 0);
-	} else {
-		data[0] = PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0);
-		data[1] = PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0);
-		data[2] = PACKET0(mmUVD_GPCOM_VCPU_CMD, 0);
-		data[3] = PACKET0(mmUVD_NO_OP, 0);
-	}
+	data[0] = PACKET0(offset + UVD_GPCOM_VCPU_DATA0, 0);
+	data[1] = PACKET0(offset + UVD_GPCOM_VCPU_DATA1, 0);
+	data[2] = PACKET0(offset + UVD_GPCOM_VCPU_CMD, 0);
+	data[3] = PACKET0(offset + UVD_NO_OP, 0);
 
 	ib = &job->ibs[0];
 	addr = amdgpu_bo_gpu_offset(bo);
@@ -1010,68 +1166,47 @@ static int amdgpu_uvd_send_msg(struct amdgpu_ring *ring, struct amdgpu_bo *bo,
 	ib->length_dw = 16;
 
 	if (direct) {
-		r = amdgpu_ib_schedule(ring, 1, ib, NULL, &f);
-		job->fence = dma_fence_get(f);
+		r = amdgpu_job_submit_direct(job, ring, &f);
 		if (r)
 			goto err_free;
-
-		amdgpu_job_free(job);
 	} else {
-		r = amdgpu_job_submit(job, ring, &adev->uvd.entity,
-				      AMDGPU_FENCE_OWNER_UNDEFINED, &f);
+		r = drm_sched_job_add_resv_dependencies(&job->base,
+							bo->tbo.base.resv,
+							DMA_RESV_USAGE_KERNEL);
 		if (r)
 			goto err_free;
+
+		f = amdgpu_job_submit(job);
 	}
 
-	ttm_eu_fence_buffer_objects(&ticket, &head, f);
+	amdgpu_bo_reserve(bo, true);
+	amdgpu_bo_fence(bo, f, false);
+	amdgpu_bo_unreserve(bo);
 
 	if (fence)
 		*fence = dma_fence_get(f);
-	amdgpu_bo_unref(&bo);
 	dma_fence_put(f);
 
 	return 0;
 
 err_free:
 	amdgpu_job_free(job);
-
-err:
-	ttm_eu_backoff_reservation(&ticket, &head);
 	return r;
 }
 
 /* multiple fence commands without any stream commands in between can
-   crash the vcpu so just try to emmit a dummy create/destroy msg to
-   avoid this */
+ * crash the vcpu so just try to emmit a dummy create/destroy msg to
+ * avoid this
+ */
 int amdgpu_uvd_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
 			      struct dma_fence **fence)
 {
 	struct amdgpu_device *adev = ring->adev;
-	struct amdgpu_bo *bo;
+	struct amdgpu_bo *bo = adev->uvd.ib_bo;
 	uint32_t *msg;
-	int r, i;
-
-	r = amdgpu_bo_create(adev, 1024, PAGE_SIZE, true,
-			     AMDGPU_GEM_DOMAIN_VRAM,
-			     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
-			     AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
-			     NULL, NULL, 0, &bo);
-	if (r)
-		return r;
-
-	r = amdgpu_bo_reserve(bo, false);
-	if (r) {
-		amdgpu_bo_unref(&bo);
-		return r;
-	}
-
-	r = amdgpu_bo_kmap(bo, (void **)&msg);
-	if (r) {
-		amdgpu_bo_unreserve(bo);
-		amdgpu_bo_unref(&bo);
-		return r;
-	}
+	int i;
 
+	msg = amdgpu_bo_kptr(bo);
 	/* stitch together an UVD create msg */
 	msg[0] = cpu_to_le32(0x00000de4);
 	msg[1] = cpu_to_le32(0x00000000);
@@ -1087,41 +1222,27 @@ int amdgpu_uvd_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
 	for (i = 11; i < 1024; ++i)
 		msg[i] = cpu_to_le32(0x0);
 
-	amdgpu_bo_kunmap(bo);
-	amdgpu_bo_unreserve(bo);
-
 	return amdgpu_uvd_send_msg(ring, bo, true, fence);
+
 }
 
 int amdgpu_uvd_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
 			       bool direct, struct dma_fence **fence)
 {
 	struct amdgpu_device *adev = ring->adev;
-	struct amdgpu_bo *bo;
+	struct amdgpu_bo *bo = NULL;
 	uint32_t *msg;
 	int r, i;
 
-	r = amdgpu_bo_create(adev, 1024, PAGE_SIZE, true,
-			     AMDGPU_GEM_DOMAIN_VRAM,
-			     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
-			     AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
-			     NULL, NULL, 0, &bo);
-	if (r)
-		return r;
-
-	r = amdgpu_bo_reserve(bo, false);
-	if (r) {
-		amdgpu_bo_unref(&bo);
-		return r;
-	}
-
-	r = amdgpu_bo_kmap(bo, (void **)&msg);
-	if (r) {
-		amdgpu_bo_unreserve(bo);
-		amdgpu_bo_unref(&bo);
-		return r;
+	if (direct) {
+		bo = adev->uvd.ib_bo;
+	} else {
+		r = amdgpu_uvd_create_msg_bo_helper(adev, 4096, &bo);
+		if (r)
+			return r;
 	}
 
+	msg = amdgpu_bo_kptr(bo);
 	/* stitch together an UVD destroy msg */
 	msg[0] = cpu_to_le32(0x00000de4);
 	msg[1] = cpu_to_le32(0x00000002);
@@ -1130,20 +1251,27 @@ int amdgpu_uvd_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
 	for (i = 4; i < 1024; ++i)
 		msg[i] = cpu_to_le32(0x0);
 
-	amdgpu_bo_kunmap(bo);
-	amdgpu_bo_unreserve(bo);
+	r = amdgpu_uvd_send_msg(ring, bo, direct, fence);
+
+	if (!direct)
+		amdgpu_bo_free_kernel(&bo, NULL, (void **)&msg);
 
-	return amdgpu_uvd_send_msg(ring, bo, direct, fence);
+	return r;
 }
 
 static void amdgpu_uvd_idle_work_handler(struct work_struct *work)
 {
 	struct amdgpu_device *adev =
 		container_of(work, struct amdgpu_device, uvd.idle_work.work);
-	unsigned fences = amdgpu_fence_count_emitted(&adev->uvd.ring);
-
-	if (amdgpu_sriov_vf(adev))
-		return;
+	unsigned int fences = 0, i, j;
+
+	for (i = 0; i < adev->uvd.num_uvd_inst; ++i) {
+		if (adev->uvd.harvest_config & (1 << i))
+			continue;
+		fences += amdgpu_fence_count_emitted(&adev->uvd.inst[i].ring);
+		for (j = 0; j < adev->uvd.num_enc_rings; ++j)
+			fences += amdgpu_fence_count_emitted(&adev->uvd.inst[i].ring_enc[j]);
+	}
 
 	if (fences == 0) {
 		if (adev->pm.dpm_enabled) {
@@ -1151,10 +1279,10 @@ static void amdgpu_uvd_idle_work_handler(struct work_struct *work)
 		} else {
 			amdgpu_asic_set_uvd_clocks(adev, 0, 0);
 			/* shutdown the UVD block */
-			amdgpu_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
-							    AMD_PG_STATE_GATE);
-			amdgpu_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
-							    AMD_CG_STATE_GATE);
+			amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+							       AMD_PG_STATE_GATE);
+			amdgpu_device_ip_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+							       AMD_CG_STATE_GATE);
 		}
 	} else {
 		schedule_delayed_work(&adev->uvd.idle_work, UVD_IDLE_TIMEOUT);
@@ -1164,33 +1292,36 @@ static void amdgpu_uvd_idle_work_handler(struct work_struct *work)
 void amdgpu_uvd_ring_begin_use(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
-	bool set_clocks = !cancel_delayed_work_sync(&adev->uvd.idle_work);
+	bool set_clocks;
 
 	if (amdgpu_sriov_vf(adev))
 		return;
 
+	set_clocks = !cancel_delayed_work_sync(&adev->uvd.idle_work);
 	if (set_clocks) {
 		if (adev->pm.dpm_enabled) {
 			amdgpu_dpm_enable_uvd(adev, true);
 		} else {
 			amdgpu_asic_set_uvd_clocks(adev, 53300, 40000);
-			amdgpu_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
-							    AMD_CG_STATE_UNGATE);
-			amdgpu_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
-							    AMD_PG_STATE_UNGATE);
+			amdgpu_device_ip_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+							       AMD_CG_STATE_UNGATE);
+			amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+							       AMD_PG_STATE_UNGATE);
 		}
 	}
 }
 
 void amdgpu_uvd_ring_end_use(struct amdgpu_ring *ring)
 {
-	schedule_delayed_work(&ring->adev->uvd.idle_work, UVD_IDLE_TIMEOUT);
+	if (!amdgpu_sriov_vf(ring->adev))
+		schedule_delayed_work(&ring->adev->uvd.idle_work, UVD_IDLE_TIMEOUT);
 }
 
 /**
  * amdgpu_uvd_ring_test_ib - test ib execution
  *
  * @ring: amdgpu_ring pointer
+ * @timeout: timeout value in jiffies, or MAX_SCHEDULE_TIMEOUT
  *
  * Test if we can successfully execute an IB
  */
@@ -1199,28 +1330,26 @@ int amdgpu_uvd_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 	struct dma_fence *fence;
 	long r;
 
-	r = amdgpu_uvd_get_create_msg(ring, 1, NULL);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get create msg (%ld).\n", r);
+	r = amdgpu_uvd_get_create_msg(ring, 1, &fence);
+	if (r)
+		goto error;
+
+	r = dma_fence_wait_timeout(fence, false, timeout);
+	dma_fence_put(fence);
+	if (r == 0)
+		r = -ETIMEDOUT;
+	if (r < 0)
 		goto error;
-	}
 
 	r = amdgpu_uvd_get_destroy_msg(ring, 1, true, &fence);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get destroy ib (%ld).\n", r);
+	if (r)
 		goto error;
-	}
 
 	r = dma_fence_wait_timeout(fence, false, timeout);
-	if (r == 0) {
-		DRM_ERROR("amdgpu: IB test timed out.\n");
+	if (r == 0)
 		r = -ETIMEDOUT;
-	} else if (r < 0) {
-		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
-	} else {
-		DRM_INFO("ib test on ring %d succeeded\n",  ring->idx);
+	else if (r > 0)
 		r = 0;
-	}
 
 	dma_fence_put(fence);
 
@@ -1237,7 +1366,7 @@ error:
  */
 uint32_t amdgpu_uvd_used_handles(struct amdgpu_device *adev)
 {
-	unsigned i;
+	unsigned int i;
 	uint32_t used_handles = 0;
 
 	for (i = 0; i < adev->uvd.max_handles; ++i) {
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h
index 3553b92bf69a..9dfad2f48ef4 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h
@@ -31,30 +31,50 @@
 #define AMDGPU_UVD_SESSION_SIZE		(50*1024)
 #define AMDGPU_UVD_FIRMWARE_OFFSET	256
 
-struct amdgpu_uvd {
+#define AMDGPU_MAX_UVD_INSTANCES			2
+
+#define AMDGPU_UVD_FIRMWARE_SIZE(adev)    \
+	(AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(((const struct common_firmware_header *)(adev)->uvd.fw->data)->ucode_size_bytes) + \
+			       8) - AMDGPU_UVD_FIRMWARE_OFFSET)
+
+struct amdgpu_uvd_inst {
 	struct amdgpu_bo	*vcpu_bo;
 	void			*cpu_addr;
 	uint64_t		gpu_addr;
-	unsigned		fw_version;
 	void			*saved_bo;
-	unsigned		max_handles;
-	atomic_t		handles[AMDGPU_MAX_UVD_HANDLES];
-	struct drm_file		*filp[AMDGPU_MAX_UVD_HANDLES];
-	struct delayed_work	idle_work;
-	const struct firmware	*fw;	/* UVD firmware */
 	struct amdgpu_ring	ring;
 	struct amdgpu_ring	ring_enc[AMDGPU_MAX_UVD_ENC_RINGS];
 	struct amdgpu_irq_src	irq;
-	bool			address_64_bit;
-	bool			use_ctx_buf;
-	struct amd_sched_entity entity;
-	struct amd_sched_entity entity_enc;
 	uint32_t                srbm_soft_reset;
+};
+
+#define AMDGPU_UVD_HARVEST_UVD0 (1 << 0)
+#define AMDGPU_UVD_HARVEST_UVD1 (1 << 1)
+
+struct amdgpu_uvd {
+	const struct firmware	*fw;	/* UVD firmware */
+	unsigned		fw_version;
+	unsigned		max_handles;
 	unsigned		num_enc_rings;
+	uint8_t			num_uvd_inst;
+	bool			address_64_bit;
+	bool			use_ctx_buf;
+	struct amdgpu_uvd_inst	inst[AMDGPU_MAX_UVD_INSTANCES];
+	struct drm_file		*filp[AMDGPU_MAX_UVD_HANDLES];
+	atomic_t		handles[AMDGPU_MAX_UVD_HANDLES];
+	struct drm_sched_entity entity;
+	struct delayed_work	idle_work;
+	unsigned		harvest_config;
+	/* store image width to adjust nb memory state */
+	unsigned		decode_image_width;
+	uint32_t                keyselect;
+	struct amdgpu_bo	*ib_bo;
 };
 
 int amdgpu_uvd_sw_init(struct amdgpu_device *adev);
 int amdgpu_uvd_sw_fini(struct amdgpu_device *adev);
+int amdgpu_uvd_entity_init(struct amdgpu_device *adev, struct amdgpu_ring *ring);
+int amdgpu_uvd_prepare_suspend(struct amdgpu_device *adev);
 int amdgpu_uvd_suspend(struct amdgpu_device *adev);
 int amdgpu_uvd_resume(struct amdgpu_device *adev);
 int amdgpu_uvd_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
@@ -63,7 +83,9 @@ int amdgpu_uvd_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
 			       bool direct, struct dma_fence **fence);
 void amdgpu_uvd_free_handles(struct amdgpu_device *adev,
 			     struct drm_file *filp);
-int amdgpu_uvd_ring_parse_cs(struct amdgpu_cs_parser *parser, uint32_t ib_idx);
+int amdgpu_uvd_ring_parse_cs(struct amdgpu_cs_parser *parser,
+			     struct amdgpu_job *job,
+			     struct amdgpu_ib *ib);
 void amdgpu_uvd_ring_begin_use(struct amdgpu_ring *ring);
 void amdgpu_uvd_ring_end_use(struct amdgpu_ring *ring);
 int amdgpu_uvd_ring_test_ib(struct amdgpu_ring *ring, long timeout);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
index c855366521ab..ce318f5de047 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
@@ -27,12 +27,14 @@
 
 #include <linux/firmware.h>
 #include <linux/module.h>
-#include <drm/drmP.h>
+
 #include <drm/drm.h>
+#include <drm/drm_drv.h>
 
 #include "amdgpu.h"
 #include "amdgpu_pm.h"
 #include "amdgpu_vce.h"
+#include "amdgpu_cs.h"
 #include "cikd.h"
 
 /* 1 second timeout */
@@ -40,21 +42,24 @@
 
 /* Firmware Names */
 #ifdef CONFIG_DRM_AMDGPU_CIK
-#define FIRMWARE_BONAIRE	"radeon/bonaire_vce.bin"
-#define FIRMWARE_KABINI	"radeon/kabini_vce.bin"
-#define FIRMWARE_KAVERI	"radeon/kaveri_vce.bin"
-#define FIRMWARE_HAWAII	"radeon/hawaii_vce.bin"
-#define FIRMWARE_MULLINS	"radeon/mullins_vce.bin"
+#define FIRMWARE_BONAIRE	"amdgpu/bonaire_vce.bin"
+#define FIRMWARE_KABINI	"amdgpu/kabini_vce.bin"
+#define FIRMWARE_KAVERI	"amdgpu/kaveri_vce.bin"
+#define FIRMWARE_HAWAII	"amdgpu/hawaii_vce.bin"
+#define FIRMWARE_MULLINS	"amdgpu/mullins_vce.bin"
 #endif
 #define FIRMWARE_TONGA		"amdgpu/tonga_vce.bin"
 #define FIRMWARE_CARRIZO	"amdgpu/carrizo_vce.bin"
 #define FIRMWARE_FIJI		"amdgpu/fiji_vce.bin"
 #define FIRMWARE_STONEY		"amdgpu/stoney_vce.bin"
 #define FIRMWARE_POLARIS10	"amdgpu/polaris10_vce.bin"
-#define FIRMWARE_POLARIS11         "amdgpu/polaris11_vce.bin"
-#define FIRMWARE_POLARIS12         "amdgpu/polaris12_vce.bin"
+#define FIRMWARE_POLARIS11	"amdgpu/polaris11_vce.bin"
+#define FIRMWARE_POLARIS12	"amdgpu/polaris12_vce.bin"
+#define FIRMWARE_VEGAM		"amdgpu/vegam_vce.bin"
 
 #define FIRMWARE_VEGA10		"amdgpu/vega10_vce.bin"
+#define FIRMWARE_VEGA12		"amdgpu/vega12_vce.bin"
+#define FIRMWARE_VEGA20		"amdgpu/vega20_vce.bin"
 
 #ifdef CONFIG_DRM_AMDGPU_CIK
 MODULE_FIRMWARE(FIRMWARE_BONAIRE);
@@ -70,25 +75,31 @@ MODULE_FIRMWARE(FIRMWARE_STONEY);
 MODULE_FIRMWARE(FIRMWARE_POLARIS10);
 MODULE_FIRMWARE(FIRMWARE_POLARIS11);
 MODULE_FIRMWARE(FIRMWARE_POLARIS12);
+MODULE_FIRMWARE(FIRMWARE_VEGAM);
 
 MODULE_FIRMWARE(FIRMWARE_VEGA10);
+MODULE_FIRMWARE(FIRMWARE_VEGA12);
+MODULE_FIRMWARE(FIRMWARE_VEGA20);
 
 static void amdgpu_vce_idle_work_handler(struct work_struct *work);
+static int amdgpu_vce_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
+				     struct dma_fence **fence);
+static int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
+				      bool direct, struct dma_fence **fence);
 
 /**
- * amdgpu_vce_init - allocate memory, load vce firmware
+ * amdgpu_vce_sw_init - allocate memory, load vce firmware
  *
  * @adev: amdgpu_device pointer
+ * @size: size for the new BO
  *
  * First step to get VCE online, allocate memory and load the firmware
  */
 int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size)
 {
-	struct amdgpu_ring *ring;
-	struct amd_sched_rq *rq;
 	const char *fw_name;
 	const struct common_firmware_header *hdr;
-	unsigned ucode_version, version_major, version_minor, binary_id;
+	unsigned int ucode_version, version_major, version_minor, binary_id;
 	int i, r;
 
 	switch (adev->asic_type) {
@@ -127,30 +138,31 @@ int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size)
 	case CHIP_POLARIS11:
 		fw_name = FIRMWARE_POLARIS11;
 		break;
+	case CHIP_POLARIS12:
+		fw_name = FIRMWARE_POLARIS12;
+		break;
+	case CHIP_VEGAM:
+		fw_name = FIRMWARE_VEGAM;
+		break;
 	case CHIP_VEGA10:
 		fw_name = FIRMWARE_VEGA10;
 		break;
-	case CHIP_POLARIS12:
-		fw_name = FIRMWARE_POLARIS12;
+	case CHIP_VEGA12:
+		fw_name = FIRMWARE_VEGA12;
+		break;
+	case CHIP_VEGA20:
+		fw_name = FIRMWARE_VEGA20;
 		break;
 
 	default:
 		return -EINVAL;
 	}
 
-	r = request_firmware(&adev->vce.fw, fw_name, adev->dev);
-	if (r) {
-		dev_err(adev->dev, "amdgpu_vce: Can't load firmware \"%s\"\n",
-			fw_name);
-		return r;
-	}
-
-	r = amdgpu_ucode_validate(adev->vce.fw);
+	r = amdgpu_ucode_request(adev, &adev->vce.fw, AMDGPU_UCODE_REQUIRED, "%s", fw_name);
 	if (r) {
 		dev_err(adev->dev, "amdgpu_vce: Can't validate firmware \"%s\"\n",
 			fw_name);
-		release_firmware(adev->vce.fw);
-		adev->vce.fw = NULL;
+		amdgpu_ucode_release(&adev->vce.fw);
 		return r;
 	}
 
@@ -160,28 +172,21 @@ int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size)
 	version_major = (ucode_version >> 20) & 0xfff;
 	version_minor = (ucode_version >> 8) & 0xfff;
 	binary_id = ucode_version & 0xff;
-	DRM_INFO("Found VCE firmware Version: %hhd.%hhd Binary ID: %hhd\n",
+	DRM_INFO("Found VCE firmware Version: %d.%d Binary ID: %d\n",
 		version_major, version_minor, binary_id);
 	adev->vce.fw_version = ((version_major << 24) | (version_minor << 16) |
 				(binary_id << 8));
 
 	r = amdgpu_bo_create_kernel(adev, size, PAGE_SIZE,
-				    AMDGPU_GEM_DOMAIN_VRAM, &adev->vce.vcpu_bo,
+				    AMDGPU_GEM_DOMAIN_VRAM |
+				    AMDGPU_GEM_DOMAIN_GTT,
+				    &adev->vce.vcpu_bo,
 				    &adev->vce.gpu_addr, &adev->vce.cpu_addr);
 	if (r) {
 		dev_err(adev->dev, "(%d) failed to allocate VCE bo\n", r);
 		return r;
 	}
 
-	ring = &adev->vce.ring[0];
-	rq = &ring->sched.sched_rq[AMD_SCHED_PRIORITY_NORMAL];
-	r = amd_sched_entity_init(&ring->sched, &adev->vce.entity,
-				  rq, amdgpu_sched_jobs);
-	if (r != 0) {
-		DRM_ERROR("Failed setting up VCE run queue.\n");
-		return r;
-	}
-
 	for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i) {
 		atomic_set(&adev->vce.handles[i], 0);
 		adev->vce.filp[i] = NULL;
@@ -194,7 +199,7 @@ int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size)
 }
 
 /**
- * amdgpu_vce_fini - free memory
+ * amdgpu_vce_sw_fini - free memory
  *
  * @adev: amdgpu_device pointer
  *
@@ -202,22 +207,47 @@ int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size)
  */
 int amdgpu_vce_sw_fini(struct amdgpu_device *adev)
 {
-	unsigned i;
+	unsigned int i;
 
 	if (adev->vce.vcpu_bo == NULL)
 		return 0;
 
-	amd_sched_entity_fini(&adev->vce.ring[0].sched, &adev->vce.entity);
-
-	amdgpu_bo_free_kernel(&adev->vce.vcpu_bo, &adev->vce.gpu_addr,
-		(void **)&adev->vce.cpu_addr);
+	drm_sched_entity_destroy(&adev->vce.entity);
 
 	for (i = 0; i < adev->vce.num_rings; i++)
 		amdgpu_ring_fini(&adev->vce.ring[i]);
 
-	release_firmware(adev->vce.fw);
+	amdgpu_ucode_release(&adev->vce.fw);
 	mutex_destroy(&adev->vce.idle_mutex);
 
+	amdgpu_bo_free_kernel(&adev->vce.vcpu_bo, &adev->vce.gpu_addr,
+		(void **)&adev->vce.cpu_addr);
+
+	return 0;
+}
+
+/**
+ * amdgpu_vce_entity_init - init entity
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: amdgpu_ring pointer to check
+ *
+ * Initialize the entity used for handle management in the kernel driver.
+ */
+int amdgpu_vce_entity_init(struct amdgpu_device *adev, struct amdgpu_ring *ring)
+{
+	if (ring == &adev->vce.ring[0]) {
+		struct drm_gpu_scheduler *sched = &ring->sched;
+		int r;
+
+		r = drm_sched_entity_init(&adev->vce.entity, DRM_SCHED_PRIORITY_NORMAL,
+					  &sched, 1, NULL);
+		if (r != 0) {
+			DRM_ERROR("Failed setting up VCE run queue.\n");
+			return r;
+		}
+	}
+
 	return 0;
 }
 
@@ -231,6 +261,8 @@ int amdgpu_vce_suspend(struct amdgpu_device *adev)
 {
 	int i;
 
+	cancel_delayed_work_sync(&adev->vce.idle_work);
+
 	if (adev->vce.vcpu_bo == NULL)
 		return 0;
 
@@ -241,7 +273,6 @@ int amdgpu_vce_suspend(struct amdgpu_device *adev)
 	if (i == AMDGPU_MAX_VCE_HANDLES)
 		return 0;
 
-	cancel_delayed_work_sync(&adev->vce.idle_work);
 	/* TODO: suspending running encoding sessions isn't supported */
 	return -EINVAL;
 }
@@ -256,8 +287,8 @@ int amdgpu_vce_resume(struct amdgpu_device *adev)
 {
 	void *cpu_addr;
 	const struct common_firmware_header *hdr;
-	unsigned offset;
-	int r;
+	unsigned int offset;
+	int r, idx;
 
 	if (adev->vce.vcpu_bo == NULL)
 		return -EINVAL;
@@ -277,8 +308,12 @@ int amdgpu_vce_resume(struct amdgpu_device *adev)
 
 	hdr = (const struct common_firmware_header *)adev->vce.fw->data;
 	offset = le32_to_cpu(hdr->ucode_array_offset_bytes);
-	memcpy_toio(cpu_addr, adev->vce.fw->data + offset,
-		    adev->vce.fw->size - offset);
+
+	if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+		memcpy_toio(cpu_addr, adev->vce.fw->data + offset,
+			    adev->vce.fw->size - offset);
+		drm_dev_exit(idx);
+	}
 
 	amdgpu_bo_kunmap(adev->vce.vcpu_bo);
 
@@ -298,10 +333,7 @@ static void amdgpu_vce_idle_work_handler(struct work_struct *work)
 {
 	struct amdgpu_device *adev =
 		container_of(work, struct amdgpu_device, vce.idle_work.work);
-	unsigned i, count = 0;
-
-	if (amdgpu_sriov_vf(adev))
-		return;
+	unsigned int i, count = 0;
 
 	for (i = 0; i < adev->vce.num_rings; i++)
 		count += amdgpu_fence_count_emitted(&adev->vce.ring[i]);
@@ -311,10 +343,10 @@ static void amdgpu_vce_idle_work_handler(struct work_struct *work)
 			amdgpu_dpm_enable_vce(adev, false);
 		} else {
 			amdgpu_asic_set_vce_clocks(adev, 0, 0);
-			amdgpu_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
-							    AMD_PG_STATE_GATE);
-			amdgpu_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
-							    AMD_CG_STATE_GATE);
+			amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							       AMD_PG_STATE_GATE);
+			amdgpu_device_ip_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							       AMD_CG_STATE_GATE);
 		}
 	} else {
 		schedule_delayed_work(&adev->vce.idle_work, VCE_IDLE_TIMEOUT);
@@ -343,10 +375,10 @@ void amdgpu_vce_ring_begin_use(struct amdgpu_ring *ring)
 			amdgpu_dpm_enable_vce(adev, true);
 		} else {
 			amdgpu_asic_set_vce_clocks(adev, 53300, 40000);
-			amdgpu_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
-							    AMD_CG_STATE_UNGATE);
-			amdgpu_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
-							    AMD_PG_STATE_UNGATE);
+			amdgpu_device_ip_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							       AMD_CG_STATE_UNGATE);
+			amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							       AMD_PG_STATE_UNGATE);
 
 		}
 	}
@@ -362,7 +394,8 @@ void amdgpu_vce_ring_begin_use(struct amdgpu_ring *ring)
  */
 void amdgpu_vce_ring_end_use(struct amdgpu_ring *ring)
 {
-	schedule_delayed_work(&ring->adev->vce.idle_work, VCE_IDLE_TIMEOUT);
+	if (!amdgpu_sriov_vf(ring->adev))
+		schedule_delayed_work(&ring->adev->vce.idle_work, VCE_IDLE_TIMEOUT);
 }
 
 /**
@@ -377,6 +410,7 @@ void amdgpu_vce_free_handles(struct amdgpu_device *adev, struct drm_file *filp)
 {
 	struct amdgpu_ring *ring = &adev->vce.ring[0];
 	int i, r;
+
 	for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i) {
 		uint32_t handle = atomic_read(&adev->vce.handles[i]);
 
@@ -395,30 +429,41 @@ void amdgpu_vce_free_handles(struct amdgpu_device *adev, struct drm_file *filp)
 /**
  * amdgpu_vce_get_create_msg - generate a VCE create msg
  *
- * @adev: amdgpu_device pointer
  * @ring: ring we should submit the msg to
  * @handle: VCE session handle to use
  * @fence: optional fence to return
  *
  * Open up a stream for HW test
  */
-int amdgpu_vce_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
-			      struct dma_fence **fence)
+static int amdgpu_vce_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
+				     struct dma_fence **fence)
 {
-	const unsigned ib_size_dw = 1024;
+	const unsigned int ib_size_dw = 1024;
 	struct amdgpu_job *job;
 	struct amdgpu_ib *ib;
+	struct amdgpu_ib ib_msg;
 	struct dma_fence *f = NULL;
-	uint64_t dummy;
+	uint64_t addr;
 	int i, r;
 
-	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4, &job);
+	r = amdgpu_job_alloc_with_ib(ring->adev, &ring->adev->vce.entity,
+				     AMDGPU_FENCE_OWNER_UNDEFINED,
+				     ib_size_dw * 4, AMDGPU_IB_POOL_DIRECT,
+				     &job, AMDGPU_KERNEL_JOB_ID_VCN_RING_TEST);
 	if (r)
 		return r;
 
-	ib = &job->ibs[0];
+	memset(&ib_msg, 0, sizeof(ib_msg));
+	/* only one gpu page is needed, alloc +1 page to make addr aligned. */
+	r = amdgpu_ib_get(ring->adev, NULL, AMDGPU_GPU_PAGE_SIZE * 2,
+			  AMDGPU_IB_POOL_DIRECT,
+			  &ib_msg);
+	if (r)
+		goto err;
 
-	dummy = ib->gpu_addr + 1024;
+	ib = &job->ibs[0];
+	/* let addr point to page boundary */
+	addr = AMDGPU_GPU_PAGE_ALIGN(ib_msg.gpu_addr);
 
 	/* stitch together an VCE create msg */
 	ib->length_dw = 0;
@@ -450,19 +495,18 @@ int amdgpu_vce_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
 
 	ib->ptr[ib->length_dw++] = 0x00000014; /* len */
 	ib->ptr[ib->length_dw++] = 0x05000005; /* feedback buffer */
-	ib->ptr[ib->length_dw++] = upper_32_bits(dummy);
-	ib->ptr[ib->length_dw++] = dummy;
+	ib->ptr[ib->length_dw++] = upper_32_bits(addr);
+	ib->ptr[ib->length_dw++] = addr;
 	ib->ptr[ib->length_dw++] = 0x00000001;
 
 	for (i = ib->length_dw; i < ib_size_dw; ++i)
 		ib->ptr[i] = 0x0;
 
-	r = amdgpu_ib_schedule(ring, 1, ib, NULL, &f);
-	job->fence = dma_fence_get(f);
+	r = amdgpu_job_submit_direct(job, ring, &f);
+	amdgpu_ib_free(&ib_msg, f);
 	if (r)
 		goto err;
 
-	amdgpu_job_free(job);
 	if (fence)
 		*fence = dma_fence_get(f);
 	dma_fence_put(f);
@@ -476,23 +520,28 @@ err:
 /**
  * amdgpu_vce_get_destroy_msg - generate a VCE destroy msg
  *
- * @adev: amdgpu_device pointer
  * @ring: ring we should submit the msg to
  * @handle: VCE session handle to use
+ * @direct: direct or delayed pool
  * @fence: optional fence to return
  *
  * Close up a stream for HW test or if userspace failed to do so
  */
-int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
-			       bool direct, struct dma_fence **fence)
+static int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
+				      bool direct, struct dma_fence **fence)
 {
-	const unsigned ib_size_dw = 1024;
+	const unsigned int ib_size_dw = 1024;
 	struct amdgpu_job *job;
 	struct amdgpu_ib *ib;
 	struct dma_fence *f = NULL;
 	int i, r;
 
-	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4, &job);
+	r = amdgpu_job_alloc_with_ib(ring->adev, &ring->adev->vce.entity,
+				     AMDGPU_FENCE_OWNER_UNDEFINED,
+				     ib_size_dw * 4,
+				     direct ? AMDGPU_IB_POOL_DIRECT :
+				     AMDGPU_IB_POOL_DELAYED, &job,
+				     AMDGPU_KERNEL_JOB_ID_VCN_RING_TEST);
 	if (r)
 		return r;
 
@@ -519,19 +568,12 @@ int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
 	for (i = ib->length_dw; i < ib_size_dw; ++i)
 		ib->ptr[i] = 0x0;
 
-	if (direct) {
-		r = amdgpu_ib_schedule(ring, 1, ib, NULL, &f);
-		job->fence = dma_fence_get(f);
-		if (r)
-			goto err;
-
-		amdgpu_job_free(job);
-	} else {
-		r = amdgpu_job_submit(job, ring, &ring->adev->vce.entity,
-				      AMDGPU_FENCE_OWNER_UNDEFINED, &f);
-		if (r)
-			goto err;
-	}
+	if (direct)
+		r = amdgpu_job_submit_direct(job, ring, &f);
+	else
+		f = amdgpu_job_submit(job);
+	if (r)
+		goto err;
 
 	if (fence)
 		*fence = dma_fence_get(f);
@@ -544,39 +586,93 @@ err:
 }
 
 /**
+ * amdgpu_vce_validate_bo - make sure not to cross 4GB boundary
+ *
+ * @p: cs parser
+ * @ib: indirect buffer to use
+ * @lo: address of lower dword
+ * @hi: address of higher dword
+ * @size: minimum size
+ * @index: bs/fb index
+ *
+ * Make sure that no BO cross a 4GB boundary.
+ */
+static int amdgpu_vce_validate_bo(struct amdgpu_cs_parser *p,
+				  struct amdgpu_ib *ib, int lo, int hi,
+				  unsigned int size, int32_t index)
+{
+	int64_t offset = ((uint64_t)size) * ((int64_t)index);
+	struct ttm_operation_ctx ctx = { false, false };
+	struct amdgpu_bo_va_mapping *mapping;
+	unsigned int i, fpfn, lpfn;
+	struct amdgpu_bo *bo;
+	uint64_t addr;
+	int r;
+
+	addr = ((uint64_t)amdgpu_ib_get_value(ib, lo)) |
+	       ((uint64_t)amdgpu_ib_get_value(ib, hi)) << 32;
+	if (index >= 0) {
+		addr += offset;
+		fpfn = PAGE_ALIGN(offset) >> PAGE_SHIFT;
+		lpfn = 0x100000000ULL >> PAGE_SHIFT;
+	} else {
+		fpfn = 0;
+		lpfn = (0x100000000ULL - PAGE_ALIGN(offset)) >> PAGE_SHIFT;
+	}
+
+	r = amdgpu_cs_find_mapping(p, addr, &bo, &mapping);
+	if (r) {
+		DRM_ERROR("Can't find BO for addr 0x%010llx %d %d %d %d\n",
+			  addr, lo, hi, size, index);
+		return r;
+	}
+
+	for (i = 0; i < bo->placement.num_placement; ++i) {
+		bo->placements[i].fpfn = max(bo->placements[i].fpfn, fpfn);
+		bo->placements[i].lpfn = bo->placements[i].lpfn ?
+			min(bo->placements[i].lpfn, lpfn) : lpfn;
+	}
+	return ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+}
+
+
+/**
  * amdgpu_vce_cs_reloc - command submission relocation
  *
  * @p: parser context
+ * @ib: indirect buffer to use
  * @lo: address of lower dword
  * @hi: address of higher dword
  * @size: minimum size
+ * @index: bs/fb index
  *
  * Patch relocation inside command stream with real buffer address
  */
-static int amdgpu_vce_cs_reloc(struct amdgpu_cs_parser *p, uint32_t ib_idx,
-			       int lo, int hi, unsigned size, uint32_t index)
+static int amdgpu_vce_cs_reloc(struct amdgpu_cs_parser *p, struct amdgpu_ib *ib,
+			       int lo, int hi, unsigned int size, uint32_t index)
 {
 	struct amdgpu_bo_va_mapping *mapping;
 	struct amdgpu_bo *bo;
 	uint64_t addr;
+	int r;
 
 	if (index == 0xffffffff)
 		index = 0;
 
-	addr = ((uint64_t)amdgpu_get_ib_value(p, ib_idx, lo)) |
-	       ((uint64_t)amdgpu_get_ib_value(p, ib_idx, hi)) << 32;
+	addr = ((uint64_t)amdgpu_ib_get_value(ib, lo)) |
+	       ((uint64_t)amdgpu_ib_get_value(ib, hi)) << 32;
 	addr += ((uint64_t)size) * ((uint64_t)index);
 
-	mapping = amdgpu_cs_find_mapping(p, addr, &bo);
-	if (mapping == NULL) {
-		DRM_ERROR("Can't find BO for addr 0x%010Lx %d %d %d %d\n",
+	r = amdgpu_cs_find_mapping(p, addr, &bo, &mapping);
+	if (r) {
+		DRM_ERROR("Can't find BO for addr 0x%010llx %d %d %d %d\n",
 			  addr, lo, hi, size, index);
-		return -EINVAL;
+		return r;
 	}
 
 	if ((addr + (uint64_t)size) >
 	    (mapping->last + 1) * AMDGPU_GPU_PAGE_SIZE) {
-		DRM_ERROR("BO to small for addr 0x%010Lx %d %d\n",
+		DRM_ERROR("BO too small for addr 0x%010llx %d %d\n",
 			  addr, lo, hi);
 		return -EINVAL;
 	}
@@ -585,8 +681,8 @@ static int amdgpu_vce_cs_reloc(struct amdgpu_cs_parser *p, uint32_t ib_idx,
 	addr += amdgpu_bo_gpu_offset(bo);
 	addr -= ((uint64_t)size) * ((uint64_t)index);
 
-	amdgpu_set_ib_value(p, ib_idx, lo, lower_32_bits(addr));
-	amdgpu_set_ib_value(p, ib_idx, hi, upper_32_bits(addr));
+	amdgpu_ib_set_value(ib, lo, lower_32_bits(addr));
+	amdgpu_ib_set_value(ib, hi, upper_32_bits(addr));
 
 	return 0;
 }
@@ -599,12 +695,12 @@ static int amdgpu_vce_cs_reloc(struct amdgpu_cs_parser *p, uint32_t ib_idx,
  * @allocated: allocated a new handle?
  *
  * Validates the handle and return the found session index or -EINVAL
- * we we don't have another free session index.
+ * we don't have another free session index.
  */
 static int amdgpu_vce_validate_handle(struct amdgpu_cs_parser *p,
 				      uint32_t handle, uint32_t *allocated)
 {
-	unsigned i;
+	unsigned int i;
 
 	/* validate the handle */
 	for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i) {
@@ -632,33 +728,32 @@ static int amdgpu_vce_validate_handle(struct amdgpu_cs_parser *p,
 }
 
 /**
- * amdgpu_vce_cs_parse - parse and validate the command stream
+ * amdgpu_vce_ring_parse_cs - parse and validate the command stream
  *
  * @p: parser context
- *
+ * @job: the job to parse
+ * @ib: the IB to patch
  */
-int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx)
+int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p,
+			     struct amdgpu_job *job,
+			     struct amdgpu_ib *ib)
 {
-	struct amdgpu_ib *ib = &p->job->ibs[ib_idx];
-	unsigned fb_idx = 0, bs_idx = 0;
+	unsigned int fb_idx = 0, bs_idx = 0;
 	int session_idx = -1;
 	uint32_t destroyed = 0;
 	uint32_t created = 0;
 	uint32_t allocated = 0;
 	uint32_t tmp, handle = 0;
-	uint32_t *size = &tmp;
-	int i, r, idx = 0;
+	uint32_t dummy = 0xffffffff;
+	uint32_t *size = &dummy;
+	unsigned int idx;
+	int i, r = 0;
 
-	p->job->vm = NULL;
-	ib->gpu_addr = amdgpu_sa_bo_gpu_addr(ib->sa_bo);
+	job->vm = NULL;
 
-	r = amdgpu_cs_sysvm_access_required(p);
-	if (r)
-		return r;
-
-	while (idx < ib->length_dw) {
-		uint32_t len = amdgpu_get_ib_value(p, ib_idx, idx);
-		uint32_t cmd = amdgpu_get_ib_value(p, ib_idx, idx + 1);
+	for (idx = 0; idx < ib->length_dw;) {
+		uint32_t len = amdgpu_ib_get_value(ib, idx);
+		uint32_t cmd = amdgpu_ib_get_value(ib, idx + 1);
 
 		if ((len < 8) || (len & 3)) {
 			DRM_ERROR("invalid VCE command length (%d)!\n", len);
@@ -667,8 +762,68 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx)
 		}
 
 		switch (cmd) {
+		case 0x00000002: /* task info */
+			fb_idx = amdgpu_ib_get_value(ib, idx + 6);
+			bs_idx = amdgpu_ib_get_value(ib, idx + 7);
+			break;
+
+		case 0x03000001: /* encode */
+			r = amdgpu_vce_validate_bo(p, ib, idx + 10, idx + 9,
+						   0, 0);
+			if (r)
+				goto out;
+
+			r = amdgpu_vce_validate_bo(p, ib, idx + 12, idx + 11,
+						   0, 0);
+			if (r)
+				goto out;
+			break;
+
+		case 0x05000001: /* context buffer */
+			r = amdgpu_vce_validate_bo(p, ib, idx + 3, idx + 2,
+						   0, 0);
+			if (r)
+				goto out;
+			break;
+
+		case 0x05000004: /* video bitstream buffer */
+			tmp = amdgpu_ib_get_value(ib, idx + 4);
+			r = amdgpu_vce_validate_bo(p, ib, idx + 3, idx + 2,
+						   tmp, bs_idx);
+			if (r)
+				goto out;
+			break;
+
+		case 0x05000005: /* feedback buffer */
+			r = amdgpu_vce_validate_bo(p, ib, idx + 3, idx + 2,
+						   4096, fb_idx);
+			if (r)
+				goto out;
+			break;
+
+		case 0x0500000d: /* MV buffer */
+			r = amdgpu_vce_validate_bo(p, ib, idx + 3, idx + 2,
+						   0, 0);
+			if (r)
+				goto out;
+
+			r = amdgpu_vce_validate_bo(p, ib, idx + 8, idx + 7,
+						   0, 0);
+			if (r)
+				goto out;
+			break;
+		}
+
+		idx += len / 4;
+	}
+
+	for (idx = 0; idx < ib->length_dw;) {
+		uint32_t len = amdgpu_ib_get_value(ib, idx);
+		uint32_t cmd = amdgpu_ib_get_value(ib, idx + 1);
+
+		switch (cmd) {
 		case 0x00000001: /* session */
-			handle = amdgpu_get_ib_value(p, ib_idx, idx + 2);
+			handle = amdgpu_ib_get_value(ib, idx + 2);
 			session_idx = amdgpu_vce_validate_handle(p, handle,
 								 &allocated);
 			if (session_idx < 0) {
@@ -679,8 +834,8 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx)
 			break;
 
 		case 0x00000002: /* task info */
-			fb_idx = amdgpu_get_ib_value(p, ib_idx, idx + 6);
-			bs_idx = amdgpu_get_ib_value(p, ib_idx, idx + 7);
+			fb_idx = amdgpu_ib_get_value(ib, idx + 6);
+			bs_idx = amdgpu_ib_get_value(ib, idx + 7);
 			break;
 
 		case 0x01000001: /* create */
@@ -695,8 +850,8 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx)
 				goto out;
 			}
 
-			*size = amdgpu_get_ib_value(p, ib_idx, idx + 8) *
-				amdgpu_get_ib_value(p, ib_idx, idx + 10) *
+			*size = amdgpu_ib_get_value(ib, idx + 8) *
+				amdgpu_ib_get_value(ib, idx + 10) *
 				8 * 3 / 2;
 			break;
 
@@ -725,12 +880,12 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx)
 			break;
 
 		case 0x03000001: /* encode */
-			r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 10, idx + 9,
+			r = amdgpu_vce_cs_reloc(p, ib, idx + 10, idx + 9,
 						*size, 0);
 			if (r)
 				goto out;
 
-			r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 12, idx + 11,
+			r = amdgpu_vce_cs_reloc(p, ib, idx + 12, idx + 11,
 						*size / 3, 0);
 			if (r)
 				goto out;
@@ -741,27 +896,39 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx)
 			break;
 
 		case 0x05000001: /* context buffer */
-			r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2,
+			r = amdgpu_vce_cs_reloc(p, ib, idx + 3, idx + 2,
 						*size * 2, 0);
 			if (r)
 				goto out;
 			break;
 
 		case 0x05000004: /* video bitstream buffer */
-			tmp = amdgpu_get_ib_value(p, ib_idx, idx + 4);
-			r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2,
+			tmp = amdgpu_ib_get_value(ib, idx + 4);
+			r = amdgpu_vce_cs_reloc(p, ib, idx + 3, idx + 2,
 						tmp, bs_idx);
 			if (r)
 				goto out;
 			break;
 
 		case 0x05000005: /* feedback buffer */
-			r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2,
+			r = amdgpu_vce_cs_reloc(p, ib, idx + 3, idx + 2,
 						4096, fb_idx);
 			if (r)
 				goto out;
 			break;
 
+		case 0x0500000d: /* MV buffer */
+			r = amdgpu_vce_cs_reloc(p, ib, idx + 3,
+						idx + 2, *size, 0);
+			if (r)
+				goto out;
+
+			r = amdgpu_vce_cs_reloc(p, ib, idx + 8,
+						idx + 7, *size / 12, 0);
+			if (r)
+				goto out;
+			break;
+
 		default:
 			DRM_ERROR("invalid VCE command (0x%x)!\n", cmd);
 			r = -EINVAL;
@@ -799,14 +966,16 @@ out:
 }
 
 /**
- * amdgpu_vce_cs_parse_vm - parse the command stream in VM mode
+ * amdgpu_vce_ring_parse_cs_vm - parse the command stream in VM mode
  *
  * @p: parser context
- *
+ * @job: the job to parse
+ * @ib: the IB to patch
  */
-int amdgpu_vce_ring_parse_cs_vm(struct amdgpu_cs_parser *p, uint32_t ib_idx)
+int amdgpu_vce_ring_parse_cs_vm(struct amdgpu_cs_parser *p,
+				struct amdgpu_job *job,
+				struct amdgpu_ib *ib)
 {
-	struct amdgpu_ib *ib = &p->job->ibs[ib_idx];
 	int session_idx = -1;
 	uint32_t destroyed = 0;
 	uint32_t created = 0;
@@ -815,8 +984,8 @@ int amdgpu_vce_ring_parse_cs_vm(struct amdgpu_cs_parser *p, uint32_t ib_idx)
 	int i, r = 0, idx = 0;
 
 	while (idx < ib->length_dw) {
-		uint32_t len = amdgpu_get_ib_value(p, ib_idx, idx);
-		uint32_t cmd = amdgpu_get_ib_value(p, ib_idx, idx + 1);
+		uint32_t len = amdgpu_ib_get_value(ib, idx);
+		uint32_t cmd = amdgpu_ib_get_value(ib, idx + 1);
 
 		if ((len < 8) || (len & 3)) {
 			DRM_ERROR("invalid VCE command length (%d)!\n", len);
@@ -826,7 +995,7 @@ int amdgpu_vce_ring_parse_cs_vm(struct amdgpu_cs_parser *p, uint32_t ib_idx)
 
 		switch (cmd) {
 		case 0x00000001: /* session */
-			handle = amdgpu_get_ib_value(p, ib_idx, idx + 2);
+			handle = amdgpu_ib_get_value(ib, idx + 2);
 			session_idx = amdgpu_vce_validate_handle(p, handle,
 								 &allocated);
 			if (session_idx < 0) {
@@ -875,7 +1044,6 @@ out:
 	if (!r) {
 		/* No error, free all destroyed handle slots */
 		tmp = destroyed;
-		amdgpu_ib_free(p->adev, ib, NULL);
 	} else {
 		/* Error during parsing, free all allocated handle slots */
 		tmp = allocated;
@@ -892,11 +1060,15 @@ out:
  * amdgpu_vce_ring_emit_ib - execute indirect buffer
  *
  * @ring: engine to use
+ * @job: job to retrieve vmid from
  * @ib: the IB to execute
+ * @flags: unused
  *
  */
-void amdgpu_vce_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib,
-			     unsigned vm_id, bool ctx_switch)
+void amdgpu_vce_ring_emit_ib(struct amdgpu_ring *ring,
+				struct amdgpu_job *job,
+				struct amdgpu_ib *ib,
+				uint32_t flags)
 {
 	amdgpu_ring_write(ring, VCE_CMD_IB);
 	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
@@ -908,11 +1080,13 @@ void amdgpu_vce_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib,
  * amdgpu_vce_ring_emit_fence - add a fence command to the ring
  *
  * @ring: engine to use
- * @fence: the fence
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
  *
  */
 void amdgpu_vce_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
-				unsigned flags)
+				unsigned int flags)
 {
 	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
 
@@ -933,8 +1107,8 @@ void amdgpu_vce_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
 int amdgpu_vce_ring_test_ring(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
-	uint32_t rptr = amdgpu_ring_get_rptr(ring);
-	unsigned i;
+	uint32_t rptr;
+	unsigned int i;
 	int r, timeout = adev->usec_timeout;
 
 	/* skip ring test for sriov*/
@@ -942,28 +1116,22 @@ int amdgpu_vce_ring_test_ring(struct amdgpu_ring *ring)
 		return 0;
 
 	r = amdgpu_ring_alloc(ring, 16);
-	if (r) {
-		DRM_ERROR("amdgpu: vce failed to lock ring %d (%d).\n",
-			  ring->idx, r);
+	if (r)
 		return r;
-	}
+
+	rptr = amdgpu_ring_get_rptr(ring);
+
 	amdgpu_ring_write(ring, VCE_CMD_END);
 	amdgpu_ring_commit(ring);
 
 	for (i = 0; i < timeout; i++) {
 		if (amdgpu_ring_get_rptr(ring) != rptr)
 			break;
-		DRM_UDELAY(1);
+		udelay(1);
 	}
 
-	if (i < timeout) {
-		DRM_INFO("ring test on %d succeeded in %d usecs\n",
-			 ring->idx, i);
-	} else {
-		DRM_ERROR("amdgpu: ring %d test failed\n",
-			  ring->idx);
+	if (i >= timeout)
 		r = -ETIMEDOUT;
-	}
 
 	return r;
 }
@@ -972,6 +1140,7 @@ int amdgpu_vce_ring_test_ring(struct amdgpu_ring *ring)
  * amdgpu_vce_ring_test_ib - test if VCE IBs are working
  *
  * @ring: the engine to test on
+ * @timeout: timeout value in jiffies, or MAX_SCHEDULE_TIMEOUT
  *
  */
 int amdgpu_vce_ring_test_ib(struct amdgpu_ring *ring, long timeout)
@@ -984,28 +1153,34 @@ int amdgpu_vce_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 		return 0;
 
 	r = amdgpu_vce_get_create_msg(ring, 1, NULL);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get create msg (%ld).\n", r);
+	if (r)
 		goto error;
-	}
 
 	r = amdgpu_vce_get_destroy_msg(ring, 1, true, &fence);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get destroy ib (%ld).\n", r);
+	if (r)
 		goto error;
-	}
 
 	r = dma_fence_wait_timeout(fence, false, timeout);
-	if (r == 0) {
-		DRM_ERROR("amdgpu: IB test timed out.\n");
+	if (r == 0)
 		r = -ETIMEDOUT;
-	} else if (r < 0) {
-		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
-	} else {
-		DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
+	else if (r > 0)
 		r = 0;
-	}
+
 error:
 	dma_fence_put(fence);
 	return r;
 }
+
+enum amdgpu_ring_priority_level amdgpu_vce_get_ring_prio(int ring)
+{
+	switch (ring) {
+	case 0:
+		return AMDGPU_RING_PRIO_0;
+	case 1:
+		return AMDGPU_RING_PRIO_1;
+	case 2:
+		return AMDGPU_RING_PRIO_2;
+	default:
+		return AMDGPU_RING_PRIO_0;
+	}
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h
index 5ce54cde472d..6e53f872d084 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h
@@ -30,6 +30,8 @@
 #define AMDGPU_VCE_HARVEST_VCE0 (1 << 0)
 #define AMDGPU_VCE_HARVEST_VCE1 (1 << 1)
 
+#define AMDGPU_VCE_FW_53_45	((53 << 24) | (45 << 16))
+
 struct amdgpu_vce {
 	struct amdgpu_bo	*vcpu_bo;
 	uint64_t		gpu_addr;
@@ -46,24 +48,24 @@ struct amdgpu_vce {
 	struct amdgpu_ring	ring[AMDGPU_MAX_VCE_RINGS];
 	struct amdgpu_irq_src	irq;
 	unsigned		harvest_config;
-	struct amd_sched_entity	entity;
+	struct drm_sched_entity	entity;
 	uint32_t                srbm_soft_reset;
 	unsigned		num_rings;
 };
 
 int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size);
 int amdgpu_vce_sw_fini(struct amdgpu_device *adev);
+int amdgpu_vce_entity_init(struct amdgpu_device *adev, struct amdgpu_ring *ring);
 int amdgpu_vce_suspend(struct amdgpu_device *adev);
 int amdgpu_vce_resume(struct amdgpu_device *adev);
-int amdgpu_vce_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
-			      struct dma_fence **fence);
-int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
-			       bool direct, struct dma_fence **fence);
 void amdgpu_vce_free_handles(struct amdgpu_device *adev, struct drm_file *filp);
-int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx);
-int amdgpu_vce_ring_parse_cs_vm(struct amdgpu_cs_parser *p, uint32_t ib_idx);
-void amdgpu_vce_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib,
-			     unsigned vm_id, bool ctx_switch);
+int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, struct amdgpu_job *job,
+			     struct amdgpu_ib *ib);
+int amdgpu_vce_ring_parse_cs_vm(struct amdgpu_cs_parser *p,
+				struct amdgpu_job *job,
+				struct amdgpu_ib *ib);
+void amdgpu_vce_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_job *job,
+				struct amdgpu_ib *ib, uint32_t flags);
 void amdgpu_vce_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
 				unsigned flags);
 int amdgpu_vce_ring_test_ring(struct amdgpu_ring *ring);
@@ -72,5 +74,6 @@ void amdgpu_vce_ring_begin_use(struct amdgpu_ring *ring);
 void amdgpu_vce_ring_end_use(struct amdgpu_ring *ring);
 unsigned amdgpu_vce_ring_get_emit_ib_size(struct amdgpu_ring *ring);
 unsigned amdgpu_vce_ring_get_dma_frame_size(struct amdgpu_ring *ring);
+enum amdgpu_ring_priority_level amdgpu_vce_get_ring_prio(int ring);
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
index 041e0121590c..5e0786ea911b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
@@ -1,5 +1,5 @@
 /*
- * Copyright 2016 Advanced Micro Devices, Inc.
+ * Copyright 2016-2024 Advanced Micro Devices, Inc.
  * All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -26,356 +26,657 @@
 
 #include <linux/firmware.h>
 #include <linux/module.h>
-#include <drm/drmP.h>
-#include <drm/drm.h>
+#include <linux/dmi.h>
+#include <linux/pci.h>
+#include <linux/debugfs.h>
+#include <drm/drm_drv.h>
 
 #include "amdgpu.h"
 #include "amdgpu_pm.h"
 #include "amdgpu_vcn.h"
 #include "soc15d.h"
-#include "soc15_common.h"
-
-#include "vega10/soc15ip.h"
-#include "raven1/VCN/vcn_1_0_offset.h"
-
-/* 1 second timeout */
-#define VCN_IDLE_TIMEOUT	msecs_to_jiffies(1000)
 
 /* Firmware Names */
-#define FIRMWARE_RAVEN		"amdgpu/raven_vcn.bin"
+#define FIRMWARE_RAVEN			"amdgpu/raven_vcn.bin"
+#define FIRMWARE_PICASSO		"amdgpu/picasso_vcn.bin"
+#define FIRMWARE_RAVEN2			"amdgpu/raven2_vcn.bin"
+#define FIRMWARE_ARCTURUS		"amdgpu/arcturus_vcn.bin"
+#define FIRMWARE_RENOIR			"amdgpu/renoir_vcn.bin"
+#define FIRMWARE_GREEN_SARDINE		"amdgpu/green_sardine_vcn.bin"
+#define FIRMWARE_NAVI10			"amdgpu/navi10_vcn.bin"
+#define FIRMWARE_NAVI14			"amdgpu/navi14_vcn.bin"
+#define FIRMWARE_NAVI12			"amdgpu/navi12_vcn.bin"
+#define FIRMWARE_SIENNA_CICHLID		"amdgpu/sienna_cichlid_vcn.bin"
+#define FIRMWARE_NAVY_FLOUNDER		"amdgpu/navy_flounder_vcn.bin"
+#define FIRMWARE_VANGOGH		"amdgpu/vangogh_vcn.bin"
+#define FIRMWARE_DIMGREY_CAVEFISH	"amdgpu/dimgrey_cavefish_vcn.bin"
+#define FIRMWARE_ALDEBARAN		"amdgpu/aldebaran_vcn.bin"
+#define FIRMWARE_BEIGE_GOBY		"amdgpu/beige_goby_vcn.bin"
+#define FIRMWARE_YELLOW_CARP		"amdgpu/yellow_carp_vcn.bin"
+#define FIRMWARE_VCN_3_1_2		"amdgpu/vcn_3_1_2.bin"
+#define FIRMWARE_VCN4_0_0		"amdgpu/vcn_4_0_0.bin"
+#define FIRMWARE_VCN4_0_2		"amdgpu/vcn_4_0_2.bin"
+#define FIRMWARE_VCN4_0_3		"amdgpu/vcn_4_0_3.bin"
+#define FIRMWARE_VCN4_0_4		"amdgpu/vcn_4_0_4.bin"
+#define FIRMWARE_VCN4_0_5		"amdgpu/vcn_4_0_5.bin"
+#define FIRMWARE_VCN4_0_6		"amdgpu/vcn_4_0_6.bin"
+#define FIRMWARE_VCN4_0_6_1		"amdgpu/vcn_4_0_6_1.bin"
+#define FIRMWARE_VCN5_0_0		"amdgpu/vcn_5_0_0.bin"
+#define FIRMWARE_VCN5_0_1		"amdgpu/vcn_5_0_1.bin"
 
 MODULE_FIRMWARE(FIRMWARE_RAVEN);
+MODULE_FIRMWARE(FIRMWARE_PICASSO);
+MODULE_FIRMWARE(FIRMWARE_RAVEN2);
+MODULE_FIRMWARE(FIRMWARE_ARCTURUS);
+MODULE_FIRMWARE(FIRMWARE_RENOIR);
+MODULE_FIRMWARE(FIRMWARE_GREEN_SARDINE);
+MODULE_FIRMWARE(FIRMWARE_ALDEBARAN);
+MODULE_FIRMWARE(FIRMWARE_NAVI10);
+MODULE_FIRMWARE(FIRMWARE_NAVI14);
+MODULE_FIRMWARE(FIRMWARE_NAVI12);
+MODULE_FIRMWARE(FIRMWARE_SIENNA_CICHLID);
+MODULE_FIRMWARE(FIRMWARE_NAVY_FLOUNDER);
+MODULE_FIRMWARE(FIRMWARE_VANGOGH);
+MODULE_FIRMWARE(FIRMWARE_DIMGREY_CAVEFISH);
+MODULE_FIRMWARE(FIRMWARE_BEIGE_GOBY);
+MODULE_FIRMWARE(FIRMWARE_YELLOW_CARP);
+MODULE_FIRMWARE(FIRMWARE_VCN_3_1_2);
+MODULE_FIRMWARE(FIRMWARE_VCN4_0_0);
+MODULE_FIRMWARE(FIRMWARE_VCN4_0_2);
+MODULE_FIRMWARE(FIRMWARE_VCN4_0_3);
+MODULE_FIRMWARE(FIRMWARE_VCN4_0_4);
+MODULE_FIRMWARE(FIRMWARE_VCN4_0_5);
+MODULE_FIRMWARE(FIRMWARE_VCN4_0_6);
+MODULE_FIRMWARE(FIRMWARE_VCN4_0_6_1);
+MODULE_FIRMWARE(FIRMWARE_VCN5_0_0);
+MODULE_FIRMWARE(FIRMWARE_VCN5_0_1);
 
 static void amdgpu_vcn_idle_work_handler(struct work_struct *work);
+static void amdgpu_vcn_reg_dump_fini(struct amdgpu_device *adev);
 
-int amdgpu_vcn_sw_init(struct amdgpu_device *adev)
+int amdgpu_vcn_early_init(struct amdgpu_device *adev, int i)
+{
+	char ucode_prefix[25];
+	int r;
+
+	adev->vcn.inst[i].adev = adev;
+	adev->vcn.inst[i].inst = i;
+	amdgpu_ucode_ip_version_decode(adev, UVD_HWIP, ucode_prefix, sizeof(ucode_prefix));
+
+	if (i != 0 && adev->vcn.per_inst_fw) {
+		r = amdgpu_ucode_request(adev, &adev->vcn.inst[i].fw,
+					 AMDGPU_UCODE_REQUIRED,
+					 "amdgpu/%s_%d.bin", ucode_prefix, i);
+		if (r)
+			amdgpu_ucode_release(&adev->vcn.inst[i].fw);
+	} else {
+		if (!adev->vcn.inst[0].fw) {
+			r = amdgpu_ucode_request(adev, &adev->vcn.inst[0].fw,
+						 AMDGPU_UCODE_REQUIRED,
+						 "amdgpu/%s.bin", ucode_prefix);
+			if (r)
+				amdgpu_ucode_release(&adev->vcn.inst[0].fw);
+		} else {
+			r = 0;
+		}
+		adev->vcn.inst[i].fw = adev->vcn.inst[0].fw;
+	}
+
+	return r;
+}
+
+int amdgpu_vcn_sw_init(struct amdgpu_device *adev, int i)
 {
-	struct amdgpu_ring *ring;
-	struct amd_sched_rq *rq;
 	unsigned long bo_size;
-	const char *fw_name;
 	const struct common_firmware_header *hdr;
-	unsigned version_major, version_minor, family_id;
+	unsigned char fw_check;
+	unsigned int fw_shared_size, log_offset;
 	int r;
 
-	INIT_DELAYED_WORK(&adev->vcn.idle_work, amdgpu_vcn_idle_work_handler);
-
-	switch (adev->asic_type) {
-	case CHIP_RAVEN:
-		fw_name = FIRMWARE_RAVEN;
-		break;
-	default:
-		return -EINVAL;
+	mutex_init(&adev->vcn.inst[i].vcn1_jpeg1_workaround);
+	mutex_init(&adev->vcn.inst[i].vcn_pg_lock);
+	mutex_init(&adev->vcn.inst[i].engine_reset_mutex);
+	atomic_set(&adev->vcn.inst[i].total_submission_cnt, 0);
+	INIT_DELAYED_WORK(&adev->vcn.inst[i].idle_work, amdgpu_vcn_idle_work_handler);
+	atomic_set(&adev->vcn.inst[i].dpg_enc_submission_cnt, 0);
+	if ((adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) &&
+	    (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG))
+		adev->vcn.inst[i].indirect_sram = true;
+
+	/*
+	 * Some Steam Deck's BIOS versions are incompatible with the
+	 * indirect SRAM mode, leading to amdgpu being unable to get
+	 * properly probed (and even potentially crashing the kernel).
+	 * Hence, check for these versions here - notice this is
+	 * restricted to Vangogh (Deck's APU).
+	 */
+	if (amdgpu_ip_version(adev, UVD_HWIP, 0) == IP_VERSION(3, 0, 2)) {
+		const char *bios_ver = dmi_get_system_info(DMI_BIOS_VERSION);
+
+		if (bios_ver && (!strncmp("F7A0113", bios_ver, 7) ||
+				 !strncmp("F7A0114", bios_ver, 7))) {
+			adev->vcn.inst[i].indirect_sram = false;
+			dev_info(adev->dev,
+				 "Steam Deck quirk: indirect SRAM disabled on BIOS %s\n", bios_ver);
+		}
 	}
 
-	r = request_firmware(&adev->vcn.fw, fw_name, adev->dev);
-	if (r) {
-		dev_err(adev->dev, "amdgpu_vcn: Can't load firmware \"%s\"\n",
-			fw_name);
-		return r;
+	/* from vcn4 and above, only unified queue is used */
+	adev->vcn.inst[i].using_unified_queue =
+		amdgpu_ip_version(adev, UVD_HWIP, 0) >= IP_VERSION(4, 0, 0);
+
+	hdr = (const struct common_firmware_header *)adev->vcn.inst[i].fw->data;
+	adev->vcn.inst[i].fw_version = le32_to_cpu(hdr->ucode_version);
+	adev->vcn.fw_version = le32_to_cpu(hdr->ucode_version);
+
+	/* Bit 20-23, it is encode major and non-zero for new naming convention.
+	 * This field is part of version minor and DRM_DISABLED_FLAG in old naming
+	 * convention. Since the l:wq!atest version minor is 0x5B and DRM_DISABLED_FLAG
+	 * is zero in old naming convention, this field is always zero so far.
+	 * These four bits are used to tell which naming convention is present.
+	 */
+	fw_check = (le32_to_cpu(hdr->ucode_version) >> 20) & 0xf;
+	if (fw_check) {
+		unsigned int dec_ver, enc_major, enc_minor, vep, fw_rev;
+
+		fw_rev = le32_to_cpu(hdr->ucode_version) & 0xfff;
+		enc_minor = (le32_to_cpu(hdr->ucode_version) >> 12) & 0xff;
+		enc_major = fw_check;
+		dec_ver = (le32_to_cpu(hdr->ucode_version) >> 24) & 0xf;
+		vep = (le32_to_cpu(hdr->ucode_version) >> 28) & 0xf;
+		dev_info(adev->dev,
+			 "[VCN instance %d] Found VCN firmware Version ENC: %u.%u DEC: %u VEP: %u Revision: %u\n",
+			 i, enc_major, enc_minor, dec_ver, vep, fw_rev);
+	} else {
+		unsigned int version_major, version_minor, family_id;
+
+		family_id = le32_to_cpu(hdr->ucode_version) & 0xff;
+		version_major = (le32_to_cpu(hdr->ucode_version) >> 24) & 0xff;
+		version_minor = (le32_to_cpu(hdr->ucode_version) >> 8) & 0xff;
+		dev_info(adev->dev, "[VCN instance %d] Found VCN firmware Version: %u.%u Family ID: %u\n",
+			 i, version_major, version_minor, family_id);
 	}
 
-	r = amdgpu_ucode_validate(adev->vcn.fw);
-	if (r) {
-		dev_err(adev->dev, "amdgpu_vcn: Can't validate firmware \"%s\"\n",
-			fw_name);
-		release_firmware(adev->vcn.fw);
-		adev->vcn.fw = NULL;
-		return r;
+	bo_size = AMDGPU_VCN_STACK_SIZE + AMDGPU_VCN_CONTEXT_SIZE;
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
+		bo_size += AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8);
+
+	if (amdgpu_ip_version(adev, UVD_HWIP, 0) >= IP_VERSION(5, 0, 0)) {
+		fw_shared_size = AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn5_fw_shared));
+		log_offset = offsetof(struct amdgpu_vcn5_fw_shared, fw_log);
+	} else if (amdgpu_ip_version(adev, UVD_HWIP, 0) >= IP_VERSION(4, 0, 0)) {
+		fw_shared_size = AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn4_fw_shared));
+		log_offset = offsetof(struct amdgpu_vcn4_fw_shared, fw_log);
+	} else {
+		fw_shared_size = AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_fw_shared));
+		log_offset = offsetof(struct amdgpu_fw_shared, fw_log);
 	}
 
-	hdr = (const struct common_firmware_header *)adev->vcn.fw->data;
-	family_id = le32_to_cpu(hdr->ucode_version) & 0xff;
-	version_major = (le32_to_cpu(hdr->ucode_version) >> 24) & 0xff;
-	version_minor = (le32_to_cpu(hdr->ucode_version) >> 8) & 0xff;
-	DRM_INFO("Found VCN firmware Version: %hu.%hu Family ID: %hu\n",
-		version_major, version_minor, family_id);
+	bo_size += fw_shared_size;
 
+	if (amdgpu_vcnfw_log)
+		bo_size += AMDGPU_VCNFW_LOG_SIZE;
 
-	bo_size = AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8)
-		  +  AMDGPU_VCN_STACK_SIZE + AMDGPU_VCN_HEAP_SIZE
-		  +  AMDGPU_VCN_SESSION_SIZE * 40;
 	r = amdgpu_bo_create_kernel(adev, bo_size, PAGE_SIZE,
-				    AMDGPU_GEM_DOMAIN_VRAM, &adev->vcn.vcpu_bo,
-				    &adev->vcn.gpu_addr, &adev->vcn.cpu_addr);
+				    AMDGPU_GEM_DOMAIN_VRAM |
+				    AMDGPU_GEM_DOMAIN_GTT,
+				    &adev->vcn.inst[i].vcpu_bo,
+				    &adev->vcn.inst[i].gpu_addr,
+				    &adev->vcn.inst[i].cpu_addr);
 	if (r) {
 		dev_err(adev->dev, "(%d) failed to allocate vcn bo\n", r);
 		return r;
 	}
 
-	ring = &adev->vcn.ring_dec;
-	rq = &ring->sched.sched_rq[AMD_SCHED_PRIORITY_NORMAL];
-	r = amd_sched_entity_init(&ring->sched, &adev->vcn.entity_dec,
-				  rq, amdgpu_sched_jobs);
-	if (r != 0) {
-		DRM_ERROR("Failed setting up VCN dec run queue.\n");
-		return r;
+	adev->vcn.inst[i].fw_shared.cpu_addr = adev->vcn.inst[i].cpu_addr +
+		bo_size - fw_shared_size;
+	adev->vcn.inst[i].fw_shared.gpu_addr = adev->vcn.inst[i].gpu_addr +
+		bo_size - fw_shared_size;
+
+	adev->vcn.inst[i].fw_shared.mem_size = fw_shared_size;
+
+	if (amdgpu_vcnfw_log) {
+		adev->vcn.inst[i].fw_shared.cpu_addr -= AMDGPU_VCNFW_LOG_SIZE;
+		adev->vcn.inst[i].fw_shared.gpu_addr -= AMDGPU_VCNFW_LOG_SIZE;
+		adev->vcn.inst[i].fw_shared.log_offset = log_offset;
 	}
 
-	ring = &adev->vcn.ring_enc[0];
-	rq = &ring->sched.sched_rq[AMD_SCHED_PRIORITY_NORMAL];
-	r = amd_sched_entity_init(&ring->sched, &adev->vcn.entity_enc,
-				  rq, amdgpu_sched_jobs);
-	if (r != 0) {
-		DRM_ERROR("Failed setting up VCN enc run queue.\n");
-		return r;
+	if (adev->vcn.inst[i].indirect_sram) {
+		r = amdgpu_bo_create_kernel(adev, 64 * 2 * 4, PAGE_SIZE,
+					    AMDGPU_GEM_DOMAIN_VRAM |
+					    AMDGPU_GEM_DOMAIN_GTT,
+					    &adev->vcn.inst[i].dpg_sram_bo,
+					    &adev->vcn.inst[i].dpg_sram_gpu_addr,
+					    &adev->vcn.inst[i].dpg_sram_cpu_addr);
+		if (r) {
+			dev_err(adev->dev, "VCN %d (%d) failed to allocate DPG bo\n", i, r);
+			return r;
+		}
 	}
 
 	return 0;
 }
 
-int amdgpu_vcn_sw_fini(struct amdgpu_device *adev)
+void amdgpu_vcn_sw_fini(struct amdgpu_device *adev, int i)
 {
-	int i;
+	int j;
 
-	kfree(adev->vcn.saved_bo);
+	if (adev->vcn.harvest_config & (1 << i))
+		return;
 
-	amd_sched_entity_fini(&adev->vcn.ring_dec.sched, &adev->vcn.entity_dec);
+	amdgpu_bo_free_kernel(
+		&adev->vcn.inst[i].dpg_sram_bo,
+		&adev->vcn.inst[i].dpg_sram_gpu_addr,
+		(void **)&adev->vcn.inst[i].dpg_sram_cpu_addr);
 
-	amd_sched_entity_fini(&adev->vcn.ring_enc[0].sched, &adev->vcn.entity_enc);
+	kvfree(adev->vcn.inst[i].saved_bo);
 
-	amdgpu_bo_free_kernel(&adev->vcn.vcpu_bo,
-			      &adev->vcn.gpu_addr,
-			      (void **)&adev->vcn.cpu_addr);
+	amdgpu_bo_free_kernel(&adev->vcn.inst[i].vcpu_bo,
+			      &adev->vcn.inst[i].gpu_addr,
+			      (void **)&adev->vcn.inst[i].cpu_addr);
 
-	amdgpu_ring_fini(&adev->vcn.ring_dec);
+	amdgpu_ring_fini(&adev->vcn.inst[i].ring_dec);
 
-	for (i = 0; i < adev->vcn.num_enc_rings; ++i)
-		amdgpu_ring_fini(&adev->vcn.ring_enc[i]);
+	for (j = 0; j < adev->vcn.inst[i].num_enc_rings; ++j)
+		amdgpu_ring_fini(&adev->vcn.inst[i].ring_enc[j]);
+
+	if (adev->vcn.per_inst_fw) {
+		amdgpu_ucode_release(&adev->vcn.inst[i].fw);
+	} else {
+		amdgpu_ucode_release(&adev->vcn.inst[0].fw);
+		adev->vcn.inst[i].fw = NULL;
+	}
 
-	release_firmware(adev->vcn.fw);
+	if (adev->vcn.reg_list)
+		amdgpu_vcn_reg_dump_fini(adev);
 
-	return 0;
+	mutex_destroy(&adev->vcn.inst[i].vcn_pg_lock);
+	mutex_destroy(&adev->vcn.inst[i].vcn1_jpeg1_workaround);
 }
 
-int amdgpu_vcn_suspend(struct amdgpu_device *adev)
+bool amdgpu_vcn_is_disabled_vcn(struct amdgpu_device *adev, enum vcn_ring_type type, uint32_t vcn_instance)
 {
-	unsigned size;
+	bool ret = false;
+	int vcn_config = adev->vcn.inst[vcn_instance].vcn_config;
+
+	if ((type == VCN_ENCODE_RING) && (vcn_config & VCN_BLOCK_ENCODE_DISABLE_MASK))
+		ret = true;
+	else if ((type == VCN_DECODE_RING) && (vcn_config & VCN_BLOCK_DECODE_DISABLE_MASK))
+		ret = true;
+	else if ((type == VCN_UNIFIED_RING) && (vcn_config & VCN_BLOCK_QUEUE_DISABLE_MASK))
+		ret = true;
+
+	return ret;
+}
+
+static int amdgpu_vcn_save_vcpu_bo_inst(struct amdgpu_device *adev, int i)
+{
+	unsigned int size;
 	void *ptr;
+	int idx;
 
-	if (adev->vcn.vcpu_bo == NULL)
+	if (adev->vcn.harvest_config & (1 << i))
+		return 0;
+	if (adev->vcn.inst[i].vcpu_bo == NULL)
 		return 0;
 
-	cancel_delayed_work_sync(&adev->vcn.idle_work);
-
-	size = amdgpu_bo_size(adev->vcn.vcpu_bo);
-	ptr = adev->vcn.cpu_addr;
+	size = amdgpu_bo_size(adev->vcn.inst[i].vcpu_bo);
+	ptr = adev->vcn.inst[i].cpu_addr;
 
-	adev->vcn.saved_bo = kmalloc(size, GFP_KERNEL);
-	if (!adev->vcn.saved_bo)
+	adev->vcn.inst[i].saved_bo = kvmalloc(size, GFP_KERNEL);
+	if (!adev->vcn.inst[i].saved_bo)
 		return -ENOMEM;
 
-	memcpy_fromio(adev->vcn.saved_bo, ptr, size);
+	if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+		memcpy_fromio(adev->vcn.inst[i].saved_bo, ptr, size);
+		drm_dev_exit(idx);
+	}
+
+	return 0;
+}
+
+int amdgpu_vcn_save_vcpu_bo(struct amdgpu_device *adev)
+{
+	int ret, i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		ret = amdgpu_vcn_save_vcpu_bo_inst(adev, i);
+		if (ret)
+			return ret;
+	}
 
 	return 0;
 }
 
-int amdgpu_vcn_resume(struct amdgpu_device *adev)
+int amdgpu_vcn_suspend(struct amdgpu_device *adev, int i)
 {
-	unsigned size;
+	bool in_ras_intr = amdgpu_ras_intr_triggered();
+
+	if (adev->vcn.harvest_config & (1 << i))
+		return 0;
+
+	/* err_event_athub and dpc recovery will corrupt VCPU buffer, so we need to
+	 * restore fw data and clear buffer in amdgpu_vcn_resume() */
+	if (in_ras_intr || adev->pcie_reset_ctx.in_link_reset)
+		return 0;
+
+	return amdgpu_vcn_save_vcpu_bo_inst(adev, i);
+}
+
+int amdgpu_vcn_resume(struct amdgpu_device *adev, int i)
+{
+	unsigned int size;
 	void *ptr;
+	int idx;
 
-	if (adev->vcn.vcpu_bo == NULL)
+	if (adev->vcn.harvest_config & (1 << i))
+		return 0;
+	if (adev->vcn.inst[i].vcpu_bo == NULL)
 		return -EINVAL;
 
-	size = amdgpu_bo_size(adev->vcn.vcpu_bo);
-	ptr = adev->vcn.cpu_addr;
+	size = amdgpu_bo_size(adev->vcn.inst[i].vcpu_bo);
+	ptr = adev->vcn.inst[i].cpu_addr;
 
-	if (adev->vcn.saved_bo != NULL) {
-		memcpy_toio(ptr, adev->vcn.saved_bo, size);
-		kfree(adev->vcn.saved_bo);
-		adev->vcn.saved_bo = NULL;
+	if (adev->vcn.inst[i].saved_bo != NULL) {
+		if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+			memcpy_toio(ptr, adev->vcn.inst[i].saved_bo, size);
+			drm_dev_exit(idx);
+		}
+		kvfree(adev->vcn.inst[i].saved_bo);
+		adev->vcn.inst[i].saved_bo = NULL;
 	} else {
 		const struct common_firmware_header *hdr;
-		unsigned offset;
-
-		hdr = (const struct common_firmware_header *)adev->vcn.fw->data;
-		offset = le32_to_cpu(hdr->ucode_array_offset_bytes);
-		memcpy_toio(adev->vcn.cpu_addr, adev->vcn.fw->data + offset,
-			    le32_to_cpu(hdr->ucode_size_bytes));
-		size -= le32_to_cpu(hdr->ucode_size_bytes);
-		ptr += le32_to_cpu(hdr->ucode_size_bytes);
+		unsigned int offset;
+
+		hdr = (const struct common_firmware_header *)adev->vcn.inst[i].fw->data;
+		if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
+			offset = le32_to_cpu(hdr->ucode_array_offset_bytes);
+			if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+				memcpy_toio(adev->vcn.inst[i].cpu_addr,
+					    adev->vcn.inst[i].fw->data + offset,
+					    le32_to_cpu(hdr->ucode_size_bytes));
+				drm_dev_exit(idx);
+			}
+			size -= le32_to_cpu(hdr->ucode_size_bytes);
+			ptr += le32_to_cpu(hdr->ucode_size_bytes);
+		}
 		memset_io(ptr, 0, size);
 	}
 
 	return 0;
 }
 
-static void amdgpu_vcn_idle_work_handler(struct work_struct *work)
+void amdgpu_vcn_get_profile(struct amdgpu_device *adev)
 {
-	struct amdgpu_device *adev =
-		container_of(work, struct amdgpu_device, vcn.idle_work.work);
-	unsigned fences = amdgpu_fence_count_emitted(&adev->vcn.ring_dec);
+	int r;
 
-	if (fences == 0) {
-		if (adev->pm.dpm_enabled) {
-			/* might be used when with pg/cg
-			amdgpu_dpm_enable_uvd(adev, false);
-			*/
+	mutex_lock(&adev->vcn.workload_profile_mutex);
+
+	if (adev->vcn.workload_profile_active) {
+		mutex_unlock(&adev->vcn.workload_profile_mutex);
+		return;
+	}
+	r = amdgpu_dpm_switch_power_profile(adev, PP_SMC_POWER_PROFILE_VIDEO,
+					    true);
+	if (r)
+		dev_warn(adev->dev,
+			 "(%d) failed to enable video power profile mode\n", r);
+	else
+		adev->vcn.workload_profile_active = true;
+	mutex_unlock(&adev->vcn.workload_profile_mutex);
+}
+
+void amdgpu_vcn_put_profile(struct amdgpu_device *adev)
+{
+	bool pg = true;
+	int r, i;
+
+	mutex_lock(&adev->vcn.workload_profile_mutex);
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		if (adev->vcn.inst[i].cur_state != AMD_PG_STATE_GATE) {
+			pg = false;
+			break;
 		}
+	}
+
+	if (pg) {
+		r = amdgpu_dpm_switch_power_profile(
+			adev, PP_SMC_POWER_PROFILE_VIDEO, false);
+		if (r)
+			dev_warn(
+				adev->dev,
+				"(%d) failed to disable video power profile mode\n",
+				r);
+		else
+			adev->vcn.workload_profile_active = false;
+	}
+
+	mutex_unlock(&adev->vcn.workload_profile_mutex);
+}
+
+static void amdgpu_vcn_idle_work_handler(struct work_struct *work)
+{
+	struct amdgpu_vcn_inst *vcn_inst =
+		container_of(work, struct amdgpu_vcn_inst, idle_work.work);
+	struct amdgpu_device *adev = vcn_inst->adev;
+	unsigned int fences = 0, fence[AMDGPU_MAX_VCN_INSTANCES] = {0};
+	unsigned int i = vcn_inst->inst, j;
+
+	if (adev->vcn.harvest_config & (1 << i))
+		return;
+
+	for (j = 0; j < adev->vcn.inst[i].num_enc_rings; ++j)
+		fence[i] += amdgpu_fence_count_emitted(&vcn_inst->ring_enc[j]);
+
+	/* Only set DPG pause for VCN3 or below, VCN4 and above will be handled by FW */
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG &&
+	    !adev->vcn.inst[i].using_unified_queue) {
+		struct dpg_pause_state new_state;
+
+		if (fence[i] ||
+		    unlikely(atomic_read(&vcn_inst->dpg_enc_submission_cnt)))
+			new_state.fw_based = VCN_DPG_STATE__PAUSE;
+		else
+			new_state.fw_based = VCN_DPG_STATE__UNPAUSE;
+
+		adev->vcn.inst[i].pause_dpg_mode(vcn_inst, &new_state);
+	}
+
+	fence[i] += amdgpu_fence_count_emitted(&vcn_inst->ring_dec);
+	fences += fence[i];
+
+	if (!fences && !atomic_read(&vcn_inst->total_submission_cnt)) {
+		mutex_lock(&vcn_inst->vcn_pg_lock);
+		vcn_inst->set_pg_state(vcn_inst, AMD_PG_STATE_GATE);
+		mutex_unlock(&vcn_inst->vcn_pg_lock);
+		amdgpu_vcn_put_profile(adev);
+
 	} else {
-		schedule_delayed_work(&adev->vcn.idle_work, VCN_IDLE_TIMEOUT);
+		schedule_delayed_work(&vcn_inst->idle_work, VCN_IDLE_TIMEOUT);
 	}
 }
 
 void amdgpu_vcn_ring_begin_use(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
-	bool set_clocks = !cancel_delayed_work_sync(&adev->vcn.idle_work);
+	struct amdgpu_vcn_inst *vcn_inst = &adev->vcn.inst[ring->me];
+
+	atomic_inc(&vcn_inst->total_submission_cnt);
+
+	cancel_delayed_work_sync(&vcn_inst->idle_work);
+
+	mutex_lock(&vcn_inst->vcn_pg_lock);
+	vcn_inst->set_pg_state(vcn_inst, AMD_PG_STATE_UNGATE);
+
+	/* Only set DPG pause for VCN3 or below, VCN4 and above will be handled by FW */
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG &&
+	    !vcn_inst->using_unified_queue) {
+		struct dpg_pause_state new_state;
 
-	if (set_clocks && adev->pm.dpm_enabled) {
-		/* might be used when with pg/cg
-		amdgpu_dpm_enable_uvd(adev, true);
-		*/
+		if (ring->funcs->type == AMDGPU_RING_TYPE_VCN_ENC) {
+			atomic_inc(&vcn_inst->dpg_enc_submission_cnt);
+			new_state.fw_based = VCN_DPG_STATE__PAUSE;
+		} else {
+			unsigned int fences = 0;
+			unsigned int i;
+
+			for (i = 0; i < vcn_inst->num_enc_rings; ++i)
+				fences += amdgpu_fence_count_emitted(&vcn_inst->ring_enc[i]);
+
+			if (fences || atomic_read(&vcn_inst->dpg_enc_submission_cnt))
+				new_state.fw_based = VCN_DPG_STATE__PAUSE;
+			else
+				new_state.fw_based = VCN_DPG_STATE__UNPAUSE;
+		}
+
+		vcn_inst->pause_dpg_mode(vcn_inst, &new_state);
 	}
+	mutex_unlock(&vcn_inst->vcn_pg_lock);
+	amdgpu_vcn_get_profile(adev);
 }
 
 void amdgpu_vcn_ring_end_use(struct amdgpu_ring *ring)
 {
-	schedule_delayed_work(&ring->adev->vcn.idle_work, VCN_IDLE_TIMEOUT);
+	struct amdgpu_device *adev = ring->adev;
+
+	/* Only set DPG pause for VCN3 or below, VCN4 and above will be handled by FW */
+	if (ring->adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG &&
+	    ring->funcs->type == AMDGPU_RING_TYPE_VCN_ENC &&
+	    !adev->vcn.inst[ring->me].using_unified_queue)
+		atomic_dec(&ring->adev->vcn.inst[ring->me].dpg_enc_submission_cnt);
+
+	atomic_dec(&ring->adev->vcn.inst[ring->me].total_submission_cnt);
+
+	schedule_delayed_work(&ring->adev->vcn.inst[ring->me].idle_work,
+			      VCN_IDLE_TIMEOUT);
 }
 
 int amdgpu_vcn_dec_ring_test_ring(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
 	uint32_t tmp = 0;
-	unsigned i;
+	unsigned int i;
 	int r;
 
-	WREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_CONTEXT_ID), 0xCAFEDEAD);
+	/* VCN in SRIOV does not support direct register read/write */
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	WREG32(adev->vcn.inst[ring->me].external.scratch9, 0xCAFEDEAD);
 	r = amdgpu_ring_alloc(ring, 3);
-	if (r) {
-		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n",
-			  ring->idx, r);
+	if (r)
 		return r;
-	}
-	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_CONTEXT_ID), 0));
+	amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[ring->me].internal.scratch9, 0));
 	amdgpu_ring_write(ring, 0xDEADBEEF);
 	amdgpu_ring_commit(ring);
 	for (i = 0; i < adev->usec_timeout; i++) {
-		tmp = RREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_CONTEXT_ID));
+		tmp = RREG32(adev->vcn.inst[ring->me].external.scratch9);
 		if (tmp == 0xDEADBEEF)
 			break;
-		DRM_UDELAY(1);
+		udelay(1);
 	}
 
-	if (i < adev->usec_timeout) {
-		DRM_INFO("ring test on %d succeeded in %d usecs\n",
-			 ring->idx, i);
-	} else {
-		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
-			  ring->idx, tmp);
-		r = -EINVAL;
-	}
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+
 	return r;
 }
 
-static int amdgpu_vcn_dec_send_msg(struct amdgpu_ring *ring, struct amdgpu_bo *bo,
-			       bool direct, struct dma_fence **fence)
+int amdgpu_vcn_dec_sw_ring_test_ring(struct amdgpu_ring *ring)
 {
-	struct ttm_validate_buffer tv;
-	struct ww_acquire_ctx ticket;
-	struct list_head head;
-	struct amdgpu_job *job;
-	struct amdgpu_ib *ib;
-	struct dma_fence *f = NULL;
 	struct amdgpu_device *adev = ring->adev;
-	uint64_t addr;
-	int i, r;
-
-	memset(&tv, 0, sizeof(tv));
-	tv.bo = &bo->tbo;
+	uint32_t rptr;
+	unsigned int i;
+	int r;
 
-	INIT_LIST_HEAD(&head);
-	list_add(&tv.head, &head);
+	if (amdgpu_sriov_vf(adev))
+		return 0;
 
-	r = ttm_eu_reserve_buffers(&ticket, &head, true, NULL);
+	r = amdgpu_ring_alloc(ring, 16);
 	if (r)
 		return r;
 
-	r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false);
-	if (r)
-		goto err;
+	rptr = amdgpu_ring_get_rptr(ring);
 
-	r = amdgpu_job_alloc_with_ib(adev, 64, &job);
+	amdgpu_ring_write(ring, VCN_DEC_SW_CMD_END);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (amdgpu_ring_get_rptr(ring) != rptr)
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+
+	return r;
+}
+
+static int amdgpu_vcn_dec_send_msg(struct amdgpu_ring *ring,
+				   struct amdgpu_ib *ib_msg,
+				   struct dma_fence **fence)
+{
+	u64 addr = AMDGPU_GPU_PAGE_ALIGN(ib_msg->gpu_addr);
+	struct amdgpu_device *adev = ring->adev;
+	struct dma_fence *f = NULL;
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	int i, r;
+
+	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL,
+				     64, AMDGPU_IB_POOL_DIRECT,
+				     &job, AMDGPU_KERNEL_JOB_ID_VCN_RING_TEST);
 	if (r)
 		goto err;
 
 	ib = &job->ibs[0];
-	addr = amdgpu_bo_gpu_offset(bo);
-	ib->ptr[0] = PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0), 0);
+	ib->ptr[0] = PACKET0(adev->vcn.inst[ring->me].internal.data0, 0);
 	ib->ptr[1] = addr;
-	ib->ptr[2] = PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA1), 0);
+	ib->ptr[2] = PACKET0(adev->vcn.inst[ring->me].internal.data1, 0);
 	ib->ptr[3] = addr >> 32;
-	ib->ptr[4] = PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD), 0);
+	ib->ptr[4] = PACKET0(adev->vcn.inst[ring->me].internal.cmd, 0);
 	ib->ptr[5] = 0;
 	for (i = 6; i < 16; i += 2) {
-		ib->ptr[i] = PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_NO_OP), 0);
+		ib->ptr[i] = PACKET0(adev->vcn.inst[ring->me].internal.nop, 0);
 		ib->ptr[i+1] = 0;
 	}
 	ib->length_dw = 16;
 
-	if (direct) {
-		r = amdgpu_ib_schedule(ring, 1, ib, NULL, &f);
-		job->fence = dma_fence_get(f);
-		if (r)
-			goto err_free;
-
-		amdgpu_job_free(job);
-	} else {
-		r = amdgpu_job_submit(job, ring, &adev->vcn.entity_dec,
-				      AMDGPU_FENCE_OWNER_UNDEFINED, &f);
-		if (r)
-			goto err_free;
-	}
+	r = amdgpu_job_submit_direct(job, ring, &f);
+	if (r)
+		goto err_free;
 
-	ttm_eu_fence_buffer_objects(&ticket, &head, f);
+	amdgpu_ib_free(ib_msg, f);
 
 	if (fence)
 		*fence = dma_fence_get(f);
-	amdgpu_bo_unref(&bo);
 	dma_fence_put(f);
 
 	return 0;
 
 err_free:
 	amdgpu_job_free(job);
-
 err:
-	ttm_eu_backoff_reservation(&ticket, &head);
+	amdgpu_ib_free(ib_msg, f);
 	return r;
 }
 
 static int amdgpu_vcn_dec_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
-			      struct dma_fence **fence)
+		struct amdgpu_ib *ib)
 {
 	struct amdgpu_device *adev = ring->adev;
-	struct amdgpu_bo *bo;
 	uint32_t *msg;
 	int r, i;
 
-	r = amdgpu_bo_create(adev, 1024, PAGE_SIZE, true,
-			     AMDGPU_GEM_DOMAIN_VRAM,
-			     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
-			     AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
-			     NULL, NULL, 0, &bo);
+	memset(ib, 0, sizeof(*ib));
+	r = amdgpu_ib_get(adev, NULL, AMDGPU_GPU_PAGE_SIZE * 2,
+			AMDGPU_IB_POOL_DIRECT,
+			ib);
 	if (r)
 		return r;
 
-	r = amdgpu_bo_reserve(bo, false);
-	if (r) {
-		amdgpu_bo_unref(&bo);
-		return r;
-	}
-
-	r = amdgpu_bo_kmap(bo, (void **)&msg);
-	if (r) {
-		amdgpu_bo_unreserve(bo);
-		amdgpu_bo_unref(&bo);
-		return r;
-	}
-
+	msg = (uint32_t *)AMDGPU_GPU_PAGE_ALIGN((unsigned long)ib->ptr);
 	msg[0] = cpu_to_le32(0x00000028);
 	msg[1] = cpu_to_le32(0x00000038);
 	msg[2] = cpu_to_le32(0x00000001);
@@ -393,41 +694,24 @@ static int amdgpu_vcn_dec_get_create_msg(struct amdgpu_ring *ring, uint32_t hand
 	for (i = 14; i < 1024; ++i)
 		msg[i] = cpu_to_le32(0x0);
 
-	amdgpu_bo_kunmap(bo);
-	amdgpu_bo_unreserve(bo);
-
-	return amdgpu_vcn_dec_send_msg(ring, bo, true, fence);
+	return 0;
 }
 
 static int amdgpu_vcn_dec_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
-			       bool direct, struct dma_fence **fence)
+					  struct amdgpu_ib *ib)
 {
 	struct amdgpu_device *adev = ring->adev;
-	struct amdgpu_bo *bo;
 	uint32_t *msg;
 	int r, i;
 
-	r = amdgpu_bo_create(adev, 1024, PAGE_SIZE, true,
-			     AMDGPU_GEM_DOMAIN_VRAM,
-			     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
-			     AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
-			     NULL, NULL, 0, &bo);
+	memset(ib, 0, sizeof(*ib));
+	r = amdgpu_ib_get(adev, NULL, AMDGPU_GPU_PAGE_SIZE * 2,
+			AMDGPU_IB_POOL_DIRECT,
+			ib);
 	if (r)
 		return r;
 
-	r = amdgpu_bo_reserve(bo, false);
-	if (r) {
-		amdgpu_bo_unref(&bo);
-		return r;
-	}
-
-	r = amdgpu_bo_kmap(bo, (void **)&msg);
-	if (r) {
-		amdgpu_bo_unreserve(bo);
-		amdgpu_bo_unref(&bo);
-		return r;
-	}
-
+	msg = (uint32_t *)AMDGPU_GPU_PAGE_ALIGN((unsigned long)ib->ptr);
 	msg[0] = cpu_to_le32(0x00000028);
 	msg[1] = cpu_to_le32(0x00000018);
 	msg[2] = cpu_to_le32(0x00000000);
@@ -437,42 +721,168 @@ static int amdgpu_vcn_dec_get_destroy_msg(struct amdgpu_ring *ring, uint32_t han
 	for (i = 6; i < 1024; ++i)
 		msg[i] = cpu_to_le32(0x0);
 
-	amdgpu_bo_kunmap(bo);
-	amdgpu_bo_unreserve(bo);
-
-	return amdgpu_vcn_dec_send_msg(ring, bo, direct, fence);
+	return 0;
 }
 
 int amdgpu_vcn_dec_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 {
-	struct dma_fence *fence;
+	struct dma_fence *fence = NULL;
+	struct amdgpu_ib ib;
 	long r;
 
-	r = amdgpu_vcn_dec_get_create_msg(ring, 1, NULL);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get create msg (%ld).\n", r);
+	r = amdgpu_vcn_dec_get_create_msg(ring, 1, &ib);
+	if (r)
 		goto error;
-	}
 
-	r = amdgpu_vcn_dec_get_destroy_msg(ring, 1, true, &fence);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get destroy ib (%ld).\n", r);
+	r = amdgpu_vcn_dec_send_msg(ring, &ib, NULL);
+	if (r)
+		goto error;
+	r = amdgpu_vcn_dec_get_destroy_msg(ring, 1, &ib);
+	if (r)
+		goto error;
+
+	r = amdgpu_vcn_dec_send_msg(ring, &ib, &fence);
+	if (r)
 		goto error;
-	}
 
 	r = dma_fence_wait_timeout(fence, false, timeout);
-	if (r == 0) {
-		DRM_ERROR("amdgpu: IB test timed out.\n");
+	if (r == 0)
 		r = -ETIMEDOUT;
-	} else if (r < 0) {
-		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
-	} else {
-		DRM_INFO("ib test on ring %d succeeded\n",  ring->idx);
+	else if (r > 0)
 		r = 0;
-	}
 
 	dma_fence_put(fence);
+error:
+	return r;
+}
+
+static uint32_t *amdgpu_vcn_unified_ring_ib_header(struct amdgpu_ib *ib,
+						uint32_t ib_pack_in_dw, bool enc)
+{
+	uint32_t *ib_checksum;
+
+	ib->ptr[ib->length_dw++] = 0x00000010; /* single queue checksum */
+	ib->ptr[ib->length_dw++] = 0x30000002;
+	ib_checksum = &ib->ptr[ib->length_dw++];
+	ib->ptr[ib->length_dw++] = ib_pack_in_dw;
+
+	ib->ptr[ib->length_dw++] = 0x00000010; /* engine info */
+	ib->ptr[ib->length_dw++] = 0x30000001;
+	ib->ptr[ib->length_dw++] = enc ? 0x2 : 0x3;
+	ib->ptr[ib->length_dw++] = ib_pack_in_dw * sizeof(uint32_t);
+
+	return ib_checksum;
+}
+
+static void amdgpu_vcn_unified_ring_ib_checksum(uint32_t **ib_checksum,
+						uint32_t ib_pack_in_dw)
+{
+	uint32_t i;
+	uint32_t checksum = 0;
+
+	for (i = 0; i < ib_pack_in_dw; i++)
+		checksum += *(*ib_checksum + 2 + i);
+
+	**ib_checksum = checksum;
+}
+
+static int amdgpu_vcn_dec_sw_send_msg(struct amdgpu_ring *ring,
+				      struct amdgpu_ib *ib_msg,
+				      struct dma_fence **fence)
+{
+	struct amdgpu_vcn_decode_buffer *decode_buffer = NULL;
+	unsigned int ib_size_dw = 64;
+	struct amdgpu_device *adev = ring->adev;
+	struct dma_fence *f = NULL;
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	uint64_t addr = AMDGPU_GPU_PAGE_ALIGN(ib_msg->gpu_addr);
+	uint32_t *ib_checksum;
+	uint32_t ib_pack_in_dw;
+	int i, r;
+
+	if (adev->vcn.inst[ring->me].using_unified_queue)
+		ib_size_dw += 8;
+
+	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL,
+				     ib_size_dw * 4, AMDGPU_IB_POOL_DIRECT,
+				     &job, AMDGPU_KERNEL_JOB_ID_VCN_RING_TEST);
+	if (r)
+		goto err;
+
+	ib = &job->ibs[0];
+	ib->length_dw = 0;
+
+	/* single queue headers */
+	if (adev->vcn.inst[ring->me].using_unified_queue) {
+		ib_pack_in_dw = sizeof(struct amdgpu_vcn_decode_buffer) / sizeof(uint32_t)
+						+ 4 + 2; /* engine info + decoding ib in dw */
+		ib_checksum = amdgpu_vcn_unified_ring_ib_header(ib, ib_pack_in_dw, false);
+	}
+
+	ib->ptr[ib->length_dw++] = sizeof(struct amdgpu_vcn_decode_buffer) + 8;
+	ib->ptr[ib->length_dw++] = cpu_to_le32(AMDGPU_VCN_IB_FLAG_DECODE_BUFFER);
+	decode_buffer = (struct amdgpu_vcn_decode_buffer *)&(ib->ptr[ib->length_dw]);
+	ib->length_dw += sizeof(struct amdgpu_vcn_decode_buffer) / 4;
+	memset(decode_buffer, 0, sizeof(struct amdgpu_vcn_decode_buffer));
+
+	decode_buffer->valid_buf_flag |= cpu_to_le32(AMDGPU_VCN_CMD_FLAG_MSG_BUFFER);
+	decode_buffer->msg_buffer_address_hi = cpu_to_le32(addr >> 32);
+	decode_buffer->msg_buffer_address_lo = cpu_to_le32(addr);
+
+	for (i = ib->length_dw; i < ib_size_dw; ++i)
+		ib->ptr[i] = 0x0;
+
+	if (adev->vcn.inst[ring->me].using_unified_queue)
+		amdgpu_vcn_unified_ring_ib_checksum(&ib_checksum, ib_pack_in_dw);
+
+	r = amdgpu_job_submit_direct(job, ring, &f);
+	if (r)
+		goto err_free;
+
+	amdgpu_ib_free(ib_msg, f);
+
+	if (fence)
+		*fence = dma_fence_get(f);
+	dma_fence_put(f);
+
+	return 0;
+
+err_free:
+	amdgpu_job_free(job);
+err:
+	amdgpu_ib_free(ib_msg, f);
+	return r;
+}
+
+int amdgpu_vcn_dec_sw_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct dma_fence *fence = NULL;
+	struct amdgpu_ib ib;
+	long r;
+
+	r = amdgpu_vcn_dec_get_create_msg(ring, 1, &ib);
+	if (r)
+		goto error;
+
+	r = amdgpu_vcn_dec_sw_send_msg(ring, &ib, NULL);
+	if (r)
+		goto error;
+	r = amdgpu_vcn_dec_get_destroy_msg(ring, 1, &ib);
+	if (r)
+		goto error;
 
+	r = amdgpu_vcn_dec_sw_send_msg(ring, &ib, &fence);
+	if (r)
+		goto error;
+
+	r = dma_fence_wait_timeout(fence, false, timeout);
+	if (r == 0)
+		r = -ETIMEDOUT;
+	else if (r > 0)
+		r = 0;
+
+	dma_fence_put(fence);
 error:
 	return r;
 }
@@ -480,61 +890,70 @@ error:
 int amdgpu_vcn_enc_ring_test_ring(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
-	uint32_t rptr = amdgpu_ring_get_rptr(ring);
-	unsigned i;
+	uint32_t rptr;
+	unsigned int i;
 	int r;
 
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
 	r = amdgpu_ring_alloc(ring, 16);
-	if (r) {
-		DRM_ERROR("amdgpu: vcn enc failed to lock ring %d (%d).\n",
-			  ring->idx, r);
+	if (r)
 		return r;
-	}
+
+	rptr = amdgpu_ring_get_rptr(ring);
+
 	amdgpu_ring_write(ring, VCN_ENC_CMD_END);
 	amdgpu_ring_commit(ring);
 
 	for (i = 0; i < adev->usec_timeout; i++) {
 		if (amdgpu_ring_get_rptr(ring) != rptr)
 			break;
-		DRM_UDELAY(1);
+		udelay(1);
 	}
 
-	if (i < adev->usec_timeout) {
-		DRM_INFO("ring test on %d succeeded in %d usecs\n",
-			 ring->idx, i);
-	} else {
-		DRM_ERROR("amdgpu: ring %d test failed\n",
-			  ring->idx);
+	if (i >= adev->usec_timeout)
 		r = -ETIMEDOUT;
-	}
 
 	return r;
 }
 
 static int amdgpu_vcn_enc_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
-			      struct dma_fence **fence)
+					 struct amdgpu_ib *ib_msg,
+					 struct dma_fence **fence)
 {
-	const unsigned ib_size_dw = 16;
+	unsigned int ib_size_dw = 16;
+	struct amdgpu_device *adev = ring->adev;
 	struct amdgpu_job *job;
 	struct amdgpu_ib *ib;
 	struct dma_fence *f = NULL;
-	uint64_t dummy;
+	uint32_t *ib_checksum = NULL;
+	uint64_t addr;
 	int i, r;
 
-	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4, &job);
+	if (adev->vcn.inst[ring->me].using_unified_queue)
+		ib_size_dw += 8;
+
+	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL,
+				     ib_size_dw * 4, AMDGPU_IB_POOL_DIRECT,
+				     &job, AMDGPU_KERNEL_JOB_ID_VCN_RING_TEST);
 	if (r)
 		return r;
 
 	ib = &job->ibs[0];
-	dummy = ib->gpu_addr + 1024;
+	addr = AMDGPU_GPU_PAGE_ALIGN(ib_msg->gpu_addr);
 
 	ib->length_dw = 0;
+
+	if (adev->vcn.inst[ring->me].using_unified_queue)
+		ib_checksum = amdgpu_vcn_unified_ring_ib_header(ib, 0x11, true);
+
 	ib->ptr[ib->length_dw++] = 0x00000018;
 	ib->ptr[ib->length_dw++] = 0x00000001; /* session info */
 	ib->ptr[ib->length_dw++] = handle;
-	ib->ptr[ib->length_dw++] = upper_32_bits(dummy);
-	ib->ptr[ib->length_dw++] = dummy;
-	ib->ptr[ib->length_dw++] = 0x0000000b;
+	ib->ptr[ib->length_dw++] = upper_32_bits(addr);
+	ib->ptr[ib->length_dw++] = addr;
+	ib->ptr[ib->length_dw++] = 0x00000000;
 
 	ib->ptr[ib->length_dw++] = 0x00000014;
 	ib->ptr[ib->length_dw++] = 0x00000002; /* task info */
@@ -548,12 +967,13 @@ static int amdgpu_vcn_enc_get_create_msg(struct amdgpu_ring *ring, uint32_t hand
 	for (i = ib->length_dw; i < ib_size_dw; ++i)
 		ib->ptr[i] = 0x0;
 
-	r = amdgpu_ib_schedule(ring, 1, ib, NULL, &f);
-	job->fence = dma_fence_get(f);
+	if (adev->vcn.inst[ring->me].using_unified_queue)
+		amdgpu_vcn_unified_ring_ib_checksum(&ib_checksum, 0x11);
+
+	r = amdgpu_job_submit_direct(job, ring, &f);
 	if (r)
 		goto err;
 
-	amdgpu_job_free(job);
 	if (fence)
 		*fence = dma_fence_get(f);
 	dma_fence_put(f);
@@ -566,29 +986,41 @@ err:
 }
 
 static int amdgpu_vcn_enc_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
-				struct dma_fence **fence)
+					  struct amdgpu_ib *ib_msg,
+					  struct dma_fence **fence)
 {
-	const unsigned ib_size_dw = 16;
+	unsigned int ib_size_dw = 16;
+	struct amdgpu_device *adev = ring->adev;
 	struct amdgpu_job *job;
 	struct amdgpu_ib *ib;
 	struct dma_fence *f = NULL;
-	uint64_t dummy;
+	uint32_t *ib_checksum = NULL;
+	uint64_t addr;
 	int i, r;
 
-	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4, &job);
+	if (adev->vcn.inst[ring->me].using_unified_queue)
+		ib_size_dw += 8;
+
+	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL,
+				     ib_size_dw * 4, AMDGPU_IB_POOL_DIRECT,
+				     &job, AMDGPU_KERNEL_JOB_ID_VCN_RING_TEST);
 	if (r)
 		return r;
 
 	ib = &job->ibs[0];
-	dummy = ib->gpu_addr + 1024;
+	addr = AMDGPU_GPU_PAGE_ALIGN(ib_msg->gpu_addr);
 
 	ib->length_dw = 0;
+
+	if (adev->vcn.inst[ring->me].using_unified_queue)
+		ib_checksum = amdgpu_vcn_unified_ring_ib_header(ib, 0x11, true);
+
 	ib->ptr[ib->length_dw++] = 0x00000018;
 	ib->ptr[ib->length_dw++] = 0x00000001;
 	ib->ptr[ib->length_dw++] = handle;
-	ib->ptr[ib->length_dw++] = upper_32_bits(dummy);
-	ib->ptr[ib->length_dw++] = dummy;
-	ib->ptr[ib->length_dw++] = 0x0000000b;
+	ib->ptr[ib->length_dw++] = upper_32_bits(addr);
+	ib->ptr[ib->length_dw++] = addr;
+	ib->ptr[ib->length_dw++] = 0x00000000;
 
 	ib->ptr[ib->length_dw++] = 0x00000014;
 	ib->ptr[ib->length_dw++] = 0x00000002;
@@ -602,12 +1034,13 @@ static int amdgpu_vcn_enc_get_destroy_msg(struct amdgpu_ring *ring, uint32_t han
 	for (i = ib->length_dw; i < ib_size_dw; ++i)
 		ib->ptr[i] = 0x0;
 
-	r = amdgpu_ib_schedule(ring, 1, ib, NULL, &f);
-	job->fence = dma_fence_get(f);
+	if (adev->vcn.inst[ring->me].using_unified_queue)
+		amdgpu_vcn_unified_ring_ib_checksum(&ib_checksum, 0x11);
+
+	r = amdgpu_job_submit_direct(job, ring, &f);
 	if (r)
 		goto err;
 
-	amdgpu_job_free(job);
 	if (fence)
 		*fence = dma_fence_get(f);
 	dma_fence_put(f);
@@ -621,32 +1054,584 @@ err:
 
 int amdgpu_vcn_enc_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 {
+	struct amdgpu_device *adev = ring->adev;
 	struct dma_fence *fence = NULL;
+	struct amdgpu_ib ib;
 	long r;
 
-	r = amdgpu_vcn_enc_get_create_msg(ring, 1, NULL);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get create msg (%ld).\n", r);
+	memset(&ib, 0, sizeof(ib));
+	r = amdgpu_ib_get(adev, NULL, (128 << 10) + AMDGPU_GPU_PAGE_SIZE,
+			AMDGPU_IB_POOL_DIRECT,
+			&ib);
+	if (r)
+		return r;
+
+	r = amdgpu_vcn_enc_get_create_msg(ring, 1, &ib, NULL);
+	if (r)
 		goto error;
-	}
 
-	r = amdgpu_vcn_enc_get_destroy_msg(ring, 1, &fence);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get destroy ib (%ld).\n", r);
+	r = amdgpu_vcn_enc_get_destroy_msg(ring, 1, &ib, &fence);
+	if (r)
 		goto error;
-	}
 
 	r = dma_fence_wait_timeout(fence, false, timeout);
-	if (r == 0) {
-		DRM_ERROR("amdgpu: IB test timed out.\n");
+	if (r == 0)
 		r = -ETIMEDOUT;
-	} else if (r < 0) {
-		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
-	} else {
-		DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
+	else if (r > 0)
 		r = 0;
-	}
+
 error:
+	amdgpu_ib_free(&ib, fence);
 	dma_fence_put(fence);
+
 	return r;
 }
+
+int amdgpu_vcn_unified_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct amdgpu_device *adev = ring->adev;
+	long r;
+
+	if ((amdgpu_ip_version(adev, UVD_HWIP, 0) != IP_VERSION(4, 0, 3)) &&
+	    (amdgpu_ip_version(adev, UVD_HWIP, 0) != IP_VERSION(5, 0, 1))) {
+		r = amdgpu_vcn_enc_ring_test_ib(ring, timeout);
+		if (r)
+			goto error;
+	}
+
+	r =  amdgpu_vcn_dec_sw_ring_test_ib(ring, timeout);
+
+error:
+	return r;
+}
+
+enum amdgpu_ring_priority_level amdgpu_vcn_get_enc_ring_prio(int ring)
+{
+	switch (ring) {
+	case 0:
+		return AMDGPU_RING_PRIO_0;
+	case 1:
+		return AMDGPU_RING_PRIO_1;
+	case 2:
+		return AMDGPU_RING_PRIO_2;
+	default:
+		return AMDGPU_RING_PRIO_0;
+	}
+}
+
+void amdgpu_vcn_setup_ucode(struct amdgpu_device *adev, int i)
+{
+	unsigned int idx;
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		const struct common_firmware_header *hdr;
+
+		if (adev->vcn.harvest_config & (1 << i))
+			return;
+
+		if ((amdgpu_ip_version(adev, UVD_HWIP, 0) == IP_VERSION(4, 0, 3) ||
+		     amdgpu_ip_version(adev, UVD_HWIP, 0) == IP_VERSION(5, 0, 1))
+		    && (i > 0))
+			return;
+
+		hdr = (const struct common_firmware_header *)adev->vcn.inst[i].fw->data;
+		/* currently only support 2 FW instances */
+		if (i >= 2) {
+			dev_info(adev->dev, "More then 2 VCN FW instances!\n");
+			return;
+		}
+		idx = AMDGPU_UCODE_ID_VCN + i;
+		adev->firmware.ucode[idx].ucode_id = idx;
+		adev->firmware.ucode[idx].fw = adev->vcn.inst[i].fw;
+		adev->firmware.fw_size +=
+			ALIGN(le32_to_cpu(hdr->ucode_size_bytes), PAGE_SIZE);
+	}
+}
+
+/*
+ * debugfs for mapping vcn firmware log buffer.
+ */
+#if defined(CONFIG_DEBUG_FS)
+static ssize_t amdgpu_debugfs_vcn_fwlog_read(struct file *f, char __user *buf,
+					     size_t size, loff_t *pos)
+{
+	struct amdgpu_vcn_inst *vcn;
+	void *log_buf;
+	struct amdgpu_vcn_fwlog *plog;
+	unsigned int read_pos, write_pos, available, i, read_bytes = 0;
+	unsigned int read_num[2] = {0};
+
+	vcn = file_inode(f)->i_private;
+	if (!vcn)
+		return -ENODEV;
+
+	if (!vcn->fw_shared.cpu_addr || !amdgpu_vcnfw_log)
+		return -EFAULT;
+
+	log_buf = vcn->fw_shared.cpu_addr + vcn->fw_shared.mem_size;
+
+	plog = (struct amdgpu_vcn_fwlog *)log_buf;
+	read_pos = plog->rptr;
+	write_pos = plog->wptr;
+
+	if (read_pos > AMDGPU_VCNFW_LOG_SIZE || write_pos > AMDGPU_VCNFW_LOG_SIZE)
+		return -EFAULT;
+
+	if (!size || (read_pos == write_pos))
+		return 0;
+
+	if (write_pos > read_pos) {
+		available = write_pos - read_pos;
+		read_num[0] = min_t(size_t, size, available);
+	} else {
+		read_num[0] = AMDGPU_VCNFW_LOG_SIZE - read_pos;
+		available = read_num[0] + write_pos - plog->header_size;
+		if (size > available)
+			read_num[1] = write_pos - plog->header_size;
+		else if (size > read_num[0])
+			read_num[1] = size - read_num[0];
+		else
+			read_num[0] = size;
+	}
+
+	for (i = 0; i < 2; i++) {
+		if (read_num[i]) {
+			if (read_pos == AMDGPU_VCNFW_LOG_SIZE)
+				read_pos = plog->header_size;
+			if (read_num[i] == copy_to_user((buf + read_bytes),
+							(log_buf + read_pos), read_num[i]))
+				return -EFAULT;
+
+			read_bytes += read_num[i];
+			read_pos += read_num[i];
+		}
+	}
+
+	plog->rptr = read_pos;
+	*pos += read_bytes;
+	return read_bytes;
+}
+
+static const struct file_operations amdgpu_debugfs_vcnfwlog_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_vcn_fwlog_read,
+	.llseek = default_llseek
+};
+#endif
+
+void amdgpu_debugfs_vcn_fwlog_init(struct amdgpu_device *adev, uint8_t i,
+				   struct amdgpu_vcn_inst *vcn)
+{
+#if defined(CONFIG_DEBUG_FS)
+	struct drm_minor *minor = adev_to_drm(adev)->primary;
+	struct dentry *root = minor->debugfs_root;
+	char name[32];
+
+	sprintf(name, "amdgpu_vcn_%d_fwlog", i);
+	debugfs_create_file_size(name, S_IFREG | 0444, root, vcn,
+				 &amdgpu_debugfs_vcnfwlog_fops,
+				 AMDGPU_VCNFW_LOG_SIZE);
+#endif
+}
+
+void amdgpu_vcn_fwlog_init(struct amdgpu_vcn_inst *vcn)
+{
+#if defined(CONFIG_DEBUG_FS)
+	uint32_t *flag = vcn->fw_shared.cpu_addr;
+	void *fw_log_cpu_addr = vcn->fw_shared.cpu_addr + vcn->fw_shared.mem_size;
+	uint64_t fw_log_gpu_addr = vcn->fw_shared.gpu_addr + vcn->fw_shared.mem_size;
+	struct amdgpu_vcn_fwlog *log_buf = fw_log_cpu_addr;
+	struct amdgpu_fw_shared_fw_logging *fw_log = vcn->fw_shared.cpu_addr
+							 + vcn->fw_shared.log_offset;
+	*flag |= cpu_to_le32(AMDGPU_VCN_FW_LOGGING_FLAG);
+	fw_log->is_enabled = 1;
+	fw_log->addr_lo = cpu_to_le32(fw_log_gpu_addr & 0xFFFFFFFF);
+	fw_log->addr_hi = cpu_to_le32(fw_log_gpu_addr >> 32);
+	fw_log->size = cpu_to_le32(AMDGPU_VCNFW_LOG_SIZE);
+
+	log_buf->header_size = sizeof(struct amdgpu_vcn_fwlog);
+	log_buf->buffer_size = AMDGPU_VCNFW_LOG_SIZE;
+	log_buf->rptr = log_buf->header_size;
+	log_buf->wptr = log_buf->header_size;
+	log_buf->wrapped = 0;
+#endif
+}
+
+int amdgpu_vcn_process_poison_irq(struct amdgpu_device *adev,
+				struct amdgpu_irq_src *source,
+				struct amdgpu_iv_entry *entry)
+{
+	struct ras_common_if *ras_if = adev->vcn.ras_if;
+	struct ras_dispatch_if ih_data = {
+		.entry = entry,
+	};
+
+	if (!ras_if)
+		return 0;
+
+	if (!amdgpu_sriov_vf(adev)) {
+		ih_data.head = *ras_if;
+		amdgpu_ras_interrupt_dispatch(adev, &ih_data);
+	} else {
+		if (adev->virt.ops && adev->virt.ops->ras_poison_handler)
+			adev->virt.ops->ras_poison_handler(adev, ras_if->block);
+		else
+			dev_warn(adev->dev,
+				"No ras_poison_handler interface in SRIOV for VCN!\n");
+	}
+
+	return 0;
+}
+
+int amdgpu_vcn_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
+{
+	int r, i;
+
+	r = amdgpu_ras_block_late_init(adev, ras_block);
+	if (r)
+		return r;
+
+	if (amdgpu_ras_is_supported(adev, ras_block->block)) {
+		for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+			if (adev->vcn.harvest_config & (1 << i) ||
+			    !adev->vcn.inst[i].ras_poison_irq.funcs)
+				continue;
+
+			r = amdgpu_irq_get(adev, &adev->vcn.inst[i].ras_poison_irq, 0);
+			if (r)
+				goto late_fini;
+		}
+	}
+	return 0;
+
+late_fini:
+	amdgpu_ras_block_late_fini(adev, ras_block);
+	return r;
+}
+
+int amdgpu_vcn_ras_sw_init(struct amdgpu_device *adev)
+{
+	int err;
+	struct amdgpu_vcn_ras *ras;
+
+	if (!adev->vcn.ras)
+		return 0;
+
+	ras = adev->vcn.ras;
+	err = amdgpu_ras_register_ras_block(adev, &ras->ras_block);
+	if (err) {
+		dev_err(adev->dev, "Failed to register vcn ras block!\n");
+		return err;
+	}
+
+	strcpy(ras->ras_block.ras_comm.name, "vcn");
+	ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__VCN;
+	ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__POISON;
+	adev->vcn.ras_if = &ras->ras_block.ras_comm;
+
+	if (!ras->ras_block.ras_late_init)
+		ras->ras_block.ras_late_init = amdgpu_vcn_ras_late_init;
+
+	return 0;
+}
+
+int amdgpu_vcn_psp_update_sram(struct amdgpu_device *adev, int inst_idx,
+			       enum AMDGPU_UCODE_ID ucode_id)
+{
+	struct amdgpu_firmware_info ucode = {
+		.ucode_id = (ucode_id ? ucode_id :
+			    (inst_idx ? AMDGPU_UCODE_ID_VCN1_RAM :
+					AMDGPU_UCODE_ID_VCN0_RAM)),
+		.mc_addr = adev->vcn.inst[inst_idx].dpg_sram_gpu_addr,
+		.ucode_size = ((uintptr_t)adev->vcn.inst[inst_idx].dpg_sram_curr_addr -
+			      (uintptr_t)adev->vcn.inst[inst_idx].dpg_sram_cpu_addr),
+	};
+
+	return psp_execute_ip_fw_load(&adev->psp, &ucode);
+}
+
+static ssize_t amdgpu_get_vcn_reset_mask(struct device *dev,
+						struct device_attribute *attr,
+						char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	if (!adev)
+		return -ENODEV;
+
+	return amdgpu_show_reset_mask(buf, adev->vcn.supported_reset);
+}
+
+static DEVICE_ATTR(vcn_reset_mask, 0444,
+		   amdgpu_get_vcn_reset_mask, NULL);
+
+int amdgpu_vcn_sysfs_reset_mask_init(struct amdgpu_device *adev)
+{
+	int r = 0;
+
+	if (adev->vcn.num_vcn_inst) {
+		r = device_create_file(adev->dev, &dev_attr_vcn_reset_mask);
+		if (r)
+			return r;
+	}
+
+	return r;
+}
+
+void amdgpu_vcn_sysfs_reset_mask_fini(struct amdgpu_device *adev)
+{
+	if (adev->dev->kobj.sd) {
+		if (adev->vcn.num_vcn_inst)
+			device_remove_file(adev->dev, &dev_attr_vcn_reset_mask);
+	}
+}
+
+/*
+ * debugfs to enable/disable vcn job submission to specific core or
+ * instance. It is created only if the queue type is unified.
+ */
+#if defined(CONFIG_DEBUG_FS)
+static int amdgpu_debugfs_vcn_sched_mask_set(void *data, u64 val)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)data;
+	u32 i;
+	u64 mask;
+	struct amdgpu_ring *ring;
+
+	if (!adev)
+		return -ENODEV;
+
+	mask = (1ULL << adev->vcn.num_vcn_inst) - 1;
+	if ((val & mask) == 0)
+		return -EINVAL;
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		ring = &adev->vcn.inst[i].ring_enc[0];
+		if (val & (1ULL << i))
+			ring->sched.ready = true;
+		else
+			ring->sched.ready = false;
+	}
+	/* publish sched.ready flag update effective immediately across smp */
+	smp_rmb();
+	return 0;
+}
+
+static int amdgpu_debugfs_vcn_sched_mask_get(void *data, u64 *val)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)data;
+	u32 i;
+	u64 mask = 0;
+	struct amdgpu_ring *ring;
+
+	if (!adev)
+		return -ENODEV;
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		ring = &adev->vcn.inst[i].ring_enc[0];
+		if (ring->sched.ready)
+			mask |= 1ULL << i;
+		}
+	*val = mask;
+	return 0;
+}
+
+DEFINE_DEBUGFS_ATTRIBUTE(amdgpu_debugfs_vcn_sched_mask_fops,
+			 amdgpu_debugfs_vcn_sched_mask_get,
+			 amdgpu_debugfs_vcn_sched_mask_set, "%llx\n");
+#endif
+
+void amdgpu_debugfs_vcn_sched_mask_init(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	struct drm_minor *minor = adev_to_drm(adev)->primary;
+	struct dentry *root = minor->debugfs_root;
+	char name[32];
+
+	if (adev->vcn.num_vcn_inst <= 1 || !adev->vcn.inst[0].using_unified_queue)
+		return;
+	sprintf(name, "amdgpu_vcn_sched_mask");
+	debugfs_create_file(name, 0600, root, adev,
+			    &amdgpu_debugfs_vcn_sched_mask_fops);
+#endif
+}
+
+/**
+ * vcn_set_powergating_state - set VCN block powergating state
+ *
+ * @ip_block: amdgpu_ip_block pointer
+ * @state: power gating state
+ *
+ * Set VCN block powergating state
+ */
+int vcn_set_powergating_state(struct amdgpu_ip_block *ip_block,
+			      enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int ret = 0, i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[i];
+
+		ret |= vinst->set_pg_state(vinst, state);
+	}
+
+	return ret;
+}
+
+/**
+ * amdgpu_vcn_reset_engine - Reset a specific VCN engine
+ * @adev: Pointer to the AMDGPU device
+ * @instance_id: VCN engine instance to reset
+ *
+ * Returns: 0 on success, or a negative error code on failure.
+ */
+static int amdgpu_vcn_reset_engine(struct amdgpu_device *adev,
+				   uint32_t instance_id)
+{
+	struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[instance_id];
+	int r, i;
+
+	mutex_lock(&vinst->engine_reset_mutex);
+	/* Stop the scheduler's work queue for the dec and enc rings if they are running.
+	 * This ensures that no new tasks are submitted to the queues while
+	 * the reset is in progress.
+	 */
+	drm_sched_wqueue_stop(&vinst->ring_dec.sched);
+	for (i = 0; i < vinst->num_enc_rings; i++)
+		drm_sched_wqueue_stop(&vinst->ring_enc[i].sched);
+
+	/* Perform the VCN reset for the specified instance */
+	r = vinst->reset(vinst);
+	if (r)
+		goto unlock;
+	r = amdgpu_ring_test_ring(&vinst->ring_dec);
+	if (r)
+		goto unlock;
+	for (i = 0; i < vinst->num_enc_rings; i++) {
+		r = amdgpu_ring_test_ring(&vinst->ring_enc[i]);
+		if (r)
+			goto unlock;
+	}
+	amdgpu_fence_driver_force_completion(&vinst->ring_dec);
+	for (i = 0; i < vinst->num_enc_rings; i++)
+		amdgpu_fence_driver_force_completion(&vinst->ring_enc[i]);
+
+	/* Restart the scheduler's work queue for the dec and enc rings
+	 * if they were stopped by this function. This allows new tasks
+	 * to be submitted to the queues after the reset is complete.
+	 */
+	drm_sched_wqueue_start(&vinst->ring_dec.sched);
+	for (i = 0; i < vinst->num_enc_rings; i++)
+		drm_sched_wqueue_start(&vinst->ring_enc[i].sched);
+
+unlock:
+	mutex_unlock(&vinst->engine_reset_mutex);
+
+	return r;
+}
+
+/**
+ * amdgpu_vcn_ring_reset - Reset a VCN ring
+ * @ring: ring to reset
+ * @vmid: vmid of guilty job
+ * @timedout_fence: fence of timed out job
+ *
+ * This helper is for VCN blocks without unified queues because
+ * resetting the engine resets all queues in that case.  With
+ * unified queues we have one queue per engine.
+ * Returns: 0 on success, or a negative error code on failure.
+ */
+int amdgpu_vcn_ring_reset(struct amdgpu_ring *ring,
+			  unsigned int vmid,
+			  struct amdgpu_fence *timedout_fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (adev->vcn.inst[ring->me].using_unified_queue)
+		return -EINVAL;
+
+	return amdgpu_vcn_reset_engine(adev, ring->me);
+}
+
+int amdgpu_vcn_reg_dump_init(struct amdgpu_device *adev,
+			     const struct amdgpu_hwip_reg_entry *reg, u32 count)
+{
+	adev->vcn.ip_dump = kcalloc(adev->vcn.num_vcn_inst * count,
+				     sizeof(uint32_t), GFP_KERNEL);
+	if (!adev->vcn.ip_dump)
+		return -ENOMEM;
+	adev->vcn.reg_list = reg;
+	adev->vcn.reg_count = count;
+
+	return 0;
+}
+
+static void amdgpu_vcn_reg_dump_fini(struct amdgpu_device *adev)
+{
+	kfree(adev->vcn.ip_dump);
+	adev->vcn.ip_dump = NULL;
+	adev->vcn.reg_list = NULL;
+	adev->vcn.reg_count = 0;
+}
+
+void amdgpu_vcn_dump_ip_state(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, j;
+	bool is_powered;
+	u32 inst_off;
+
+	if (!adev->vcn.ip_dump)
+		return;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		inst_off = i * adev->vcn.reg_count;
+		/* mmUVD_POWER_STATUS is always readable and is the first in reg_list */
+		adev->vcn.ip_dump[inst_off] =
+			RREG32(SOC15_REG_ENTRY_OFFSET_INST(adev->vcn.reg_list[0], i));
+		is_powered = (adev->vcn.ip_dump[inst_off] &
+			      UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF) !=
+			      UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF;
+
+		if (is_powered)
+			for (j = 1; j < adev->vcn.reg_count; j++)
+				adev->vcn.ip_dump[inst_off + j] =
+				RREG32(SOC15_REG_ENTRY_OFFSET_INST(adev->vcn.reg_list[j], i));
+	}
+}
+
+void amdgpu_vcn_print_ip_state(struct amdgpu_ip_block *ip_block, struct drm_printer *p)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, j;
+	bool is_powered;
+	u32 inst_off;
+
+	if (!adev->vcn.ip_dump)
+		return;
+
+	drm_printf(p, "num_instances:%d\n", adev->vcn.num_vcn_inst);
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		if (adev->vcn.harvest_config & (1 << i)) {
+			drm_printf(p, "\nHarvested Instance:VCN%d Skipping dump\n", i);
+			continue;
+		}
+
+		inst_off = i * adev->vcn.reg_count;
+		is_powered = (adev->vcn.ip_dump[inst_off] &
+			      UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF) !=
+			      UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF;
+
+		if (is_powered) {
+			drm_printf(p, "\nActive Instance:VCN%d\n", i);
+			for (j = 0; j < adev->vcn.reg_count; j++)
+				drm_printf(p, "%-50s \t 0x%08x\n", adev->vcn.reg_list[j].reg_name,
+					   adev->vcn.ip_dump[inst_off + j]);
+		} else {
+			drm_printf(p, "\nInactive Instance:VCN%d\n", i);
+		}
+	}
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h
index d50ba0657854..dc8a17bcc3c8 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h
@@ -1,5 +1,5 @@
 /*
- * Copyright 2016 Advanced Micro Devices, Inc.
+ * Copyright 2016-2024 Advanced Micro Devices, Inc. All rights reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -24,12 +24,21 @@
 #ifndef __AMDGPU_VCN_H__
 #define __AMDGPU_VCN_H__
 
-#define AMDGPU_VCN_STACK_SIZE		(200*1024)
-#define AMDGPU_VCN_HEAP_SIZE		(256*1024)
-#define AMDGPU_VCN_SESSION_SIZE		(50*1024)
+#include "amdgpu_ras.h"
+
+#define AMDGPU_VCN_STACK_SIZE		(128*1024)
+#define AMDGPU_VCN_CONTEXT_SIZE 	(512*1024)
+
 #define AMDGPU_VCN_FIRMWARE_OFFSET	256
 #define AMDGPU_VCN_MAX_ENC_RINGS	3
 
+#define AMDGPU_MAX_VCN_INSTANCES	4
+#define AMDGPU_MAX_VCN_ENC_RINGS  (AMDGPU_VCN_MAX_ENC_RINGS * AMDGPU_MAX_VCN_INSTANCES)
+
+#define AMDGPU_VCN_HARVEST_VCN0 (1 << 0)
+#define AMDGPU_VCN_HARVEST_VCN1 (1 << 1)
+
+#define VCN_DEC_KMD_CMD 		0x80000000
 #define VCN_DEC_CMD_FENCE		0x00000000
 #define VCN_DEC_CMD_TRAP		0x00000001
 #define VCN_DEC_CMD_WRITE_REG		0x00000004
@@ -37,6 +46,17 @@
 #define VCN_DEC_CMD_PACKET_START	0x0000000a
 #define VCN_DEC_CMD_PACKET_END		0x0000000b
 
+#define VCN_DEC_SW_CMD_NO_OP		0x00000000
+#define VCN_DEC_SW_CMD_END		0x00000001
+#define VCN_DEC_SW_CMD_IB		0x00000002
+#define VCN_DEC_SW_CMD_FENCE		0x00000003
+#define VCN_DEC_SW_CMD_TRAP		0x00000004
+#define VCN_DEC_SW_CMD_IB_AUTO		0x00000005
+#define VCN_DEC_SW_CMD_SEMAPHORE	0x00000006
+#define VCN_DEC_SW_CMD_PREEMPT_FENCE	0x00000009
+#define VCN_DEC_SW_CMD_REG_WRITE	0x0000000b
+#define VCN_DEC_SW_CMD_REG_WAIT		0x0000000c
+
 #define VCN_ENC_CMD_NO_OP		0x00000000
 #define VCN_ENC_CMD_END 		0x00000001
 #define VCN_ENC_CMD_IB			0x00000002
@@ -45,33 +65,507 @@
 #define VCN_ENC_CMD_REG_WRITE		0x0000000b
 #define VCN_ENC_CMD_REG_WAIT		0x0000000c
 
-struct amdgpu_vcn {
+#define VCN_AON_SOC_ADDRESS_2_0 	0x1f800
+#define VCN_VID_IP_ADDRESS_2_0		0x0
+#define VCN_AON_IP_ADDRESS_2_0		0x30000
+
+#define mmUVD_RBC_XX_IB_REG_CHECK 					0x026b
+#define mmUVD_RBC_XX_IB_REG_CHECK_BASE_IDX 				1
+#define mmUVD_REG_XX_MASK 						0x026c
+#define mmUVD_REG_XX_MASK_BASE_IDX 					1
+
+/* 1 second timeout */
+#define VCN_IDLE_TIMEOUT	msecs_to_jiffies(1000)
+
+#define RREG32_SOC15_DPG_MODE_1_0(ip, inst_idx, reg, mask, sram_sel) 			\
+	({	WREG32_SOC15(ip, inst_idx, mmUVD_DPG_LMA_MASK, mask); 			\
+		WREG32_SOC15(ip, inst_idx, mmUVD_DPG_LMA_CTL, 				\
+			UVD_DPG_LMA_CTL__MASK_EN_MASK | 				\
+			((adev->reg_offset[ip##_HWIP][inst_idx][reg##_BASE_IDX] + reg) 	\
+			<< UVD_DPG_LMA_CTL__READ_WRITE_ADDR__SHIFT) | 			\
+			(sram_sel << UVD_DPG_LMA_CTL__SRAM_SEL__SHIFT)); 		\
+		RREG32_SOC15(ip, inst_idx, mmUVD_DPG_LMA_DATA); 			\
+	})
+
+#define WREG32_SOC15_DPG_MODE_1_0(ip, inst_idx, reg, value, mask, sram_sel) 		\
+	do { 										\
+		WREG32_SOC15(ip, inst_idx, mmUVD_DPG_LMA_DATA, value); 			\
+		WREG32_SOC15(ip, inst_idx, mmUVD_DPG_LMA_MASK, mask); 			\
+		WREG32_SOC15(ip, inst_idx, mmUVD_DPG_LMA_CTL, 				\
+			UVD_DPG_LMA_CTL__READ_WRITE_MASK | 				\
+			((adev->reg_offset[ip##_HWIP][inst_idx][reg##_BASE_IDX] + reg) 	\
+			<< UVD_DPG_LMA_CTL__READ_WRITE_ADDR__SHIFT) | 			\
+			(sram_sel << UVD_DPG_LMA_CTL__SRAM_SEL__SHIFT)); 		\
+	} while (0)
+
+#define SOC15_DPG_MODE_OFFSET(ip, inst_idx, reg) 						\
+	({											\
+		uint32_t internal_reg_offset, addr;						\
+		bool video_range, video1_range, aon_range, aon1_range;				\
+												\
+		addr = (adev->reg_offset[ip##_HWIP][inst_idx][reg##_BASE_IDX] + reg);		\
+		addr <<= 2; 									\
+		video_range = ((((0xFFFFF & addr) >= (VCN_VID_SOC_ADDRESS_2_0)) && 		\
+				((0xFFFFF & addr) < ((VCN_VID_SOC_ADDRESS_2_0 + 0x2600)))));	\
+		video1_range = ((((0xFFFFF & addr) >= (VCN1_VID_SOC_ADDRESS_3_0)) && 		\
+				((0xFFFFF & addr) < ((VCN1_VID_SOC_ADDRESS_3_0 + 0x2600)))));	\
+		aon_range   = ((((0xFFFFF & addr) >= (VCN_AON_SOC_ADDRESS_2_0)) && 		\
+				((0xFFFFF & addr) < ((VCN_AON_SOC_ADDRESS_2_0 + 0x600)))));	\
+		aon1_range   = ((((0xFFFFF & addr) >= (VCN1_AON_SOC_ADDRESS_3_0)) && 		\
+				((0xFFFFF & addr) < ((VCN1_AON_SOC_ADDRESS_3_0 + 0x600)))));	\
+		if (video_range) 								\
+			internal_reg_offset = ((0xFFFFF & addr) - (VCN_VID_SOC_ADDRESS_2_0) + 	\
+				(VCN_VID_IP_ADDRESS_2_0));					\
+		else if (aon_range)								\
+			internal_reg_offset = ((0xFFFFF & addr) - (VCN_AON_SOC_ADDRESS_2_0) + 	\
+				(VCN_AON_IP_ADDRESS_2_0));					\
+		else if (video1_range) 								\
+			internal_reg_offset = ((0xFFFFF & addr) - (VCN1_VID_SOC_ADDRESS_3_0) + 	\
+				(VCN_VID_IP_ADDRESS_2_0));					\
+		else if (aon1_range)								\
+			internal_reg_offset = ((0xFFFFF & addr) - (VCN1_AON_SOC_ADDRESS_3_0) + 	\
+				(VCN_AON_IP_ADDRESS_2_0));					\
+		else										\
+			internal_reg_offset = (0xFFFFF & addr);					\
+												\
+		internal_reg_offset >>= 2;							\
+	})
+
+#define RREG32_SOC15_DPG_MODE(inst_idx, offset, mask_en) 					\
+	({											\
+		WREG32_SOC15(VCN, inst_idx, mmUVD_DPG_LMA_CTL, 					\
+			(0x0 << UVD_DPG_LMA_CTL__READ_WRITE__SHIFT |				\
+			mask_en << UVD_DPG_LMA_CTL__MASK_EN__SHIFT |				\
+			offset << UVD_DPG_LMA_CTL__READ_WRITE_ADDR__SHIFT));			\
+		RREG32_SOC15(VCN, inst_idx, mmUVD_DPG_LMA_DATA);				\
+	})
+
+#define WREG32_SOC15_DPG_MODE(inst_idx, offset, value, mask_en, indirect)             \
+	do {                                                                          \
+		if (!indirect) {                                                      \
+			WREG32_SOC15(VCN, GET_INST(VCN, inst_idx),                    \
+				     mmUVD_DPG_LMA_DATA, value);                      \
+			WREG32_SOC15(                                                 \
+				VCN, GET_INST(VCN, inst_idx),                         \
+				mmUVD_DPG_LMA_CTL,                                    \
+				(0x1 << UVD_DPG_LMA_CTL__READ_WRITE__SHIFT |          \
+				 mask_en << UVD_DPG_LMA_CTL__MASK_EN__SHIFT |         \
+				 offset << UVD_DPG_LMA_CTL__READ_WRITE_ADDR__SHIFT)); \
+		} else {                                                              \
+			*adev->vcn.inst[inst_idx].dpg_sram_curr_addr++ =              \
+				offset;                                               \
+			*adev->vcn.inst[inst_idx].dpg_sram_curr_addr++ =              \
+				value;                                                \
+		}                                                                     \
+	} while (0)
+
+#define SOC24_DPG_MODE_OFFSET(ip, inst_idx, reg)						\
+	({											\
+		uint32_t internal_reg_offset, addr;						\
+		bool video_range, video1_range, aon_range, aon1_range;				\
+												\
+		addr = (adev->reg_offset[ip##_HWIP][inst_idx][reg##_BASE_IDX] + reg);		\
+		addr <<= 2;									\
+		video_range = ((((0xFFFFF & addr) >= (VCN_VID_SOC_ADDRESS)) &&			\
+				((0xFFFFF & addr) < ((VCN_VID_SOC_ADDRESS + 0x2600)))));	\
+		video1_range = ((((0xFFFFF & addr) >= (VCN1_VID_SOC_ADDRESS)) &&		\
+				((0xFFFFF & addr) < ((VCN1_VID_SOC_ADDRESS + 0x2600)))));	\
+		aon_range   = ((((0xFFFFF & addr) >= (VCN_AON_SOC_ADDRESS)) &&			\
+				((0xFFFFF & addr) < ((VCN_AON_SOC_ADDRESS + 0x600)))));		\
+		aon1_range   = ((((0xFFFFF & addr) >= (VCN1_AON_SOC_ADDRESS)) &&		\
+				((0xFFFFF & addr) < ((VCN1_AON_SOC_ADDRESS + 0x600)))));	\
+		if (video_range)								\
+			internal_reg_offset = ((0xFFFFF & addr) - (VCN_VID_SOC_ADDRESS) +	\
+				(VCN_VID_IP_ADDRESS));						\
+		else if (aon_range)								\
+			internal_reg_offset = ((0xFFFFF & addr) - (VCN_AON_SOC_ADDRESS) +	\
+				(VCN_AON_IP_ADDRESS));						\
+		else if (video1_range)								\
+			internal_reg_offset = ((0xFFFFF & addr) - (VCN1_VID_SOC_ADDRESS) +	\
+				(VCN_VID_IP_ADDRESS));						\
+		else if (aon1_range)								\
+			internal_reg_offset = ((0xFFFFF & addr) - (VCN1_AON_SOC_ADDRESS) +	\
+				(VCN_AON_IP_ADDRESS));						\
+		else										\
+			internal_reg_offset = (0xFFFFF & addr);					\
+												\
+		internal_reg_offset >>= 2;							\
+	})
+
+#define WREG32_SOC24_DPG_MODE(inst_idx, offset, value, mask_en, indirect)		\
+	do {										\
+		if (!indirect) {							\
+			WREG32_SOC15(VCN, GET_INST(VCN, inst_idx),			\
+				     regUVD_DPG_LMA_DATA, value);			\
+			WREG32_SOC15(							\
+				VCN, GET_INST(VCN, inst_idx),				\
+				regUVD_DPG_LMA_CTL,					\
+				(0x1 << UVD_DPG_LMA_CTL__READ_WRITE__SHIFT |		\
+				 mask_en << UVD_DPG_LMA_CTL__MASK_EN__SHIFT |		\
+				 offset << UVD_DPG_LMA_CTL__READ_WRITE_ADDR__SHIFT));	\
+		} else {								\
+			*adev->vcn.inst[inst_idx].dpg_sram_curr_addr++ =		\
+				offset;							\
+			*adev->vcn.inst[inst_idx].dpg_sram_curr_addr++ =		\
+				value;							\
+		}									\
+	} while (0)
+
+#define AMDGPU_FW_SHARED_FLAG_0_UNIFIED_QUEUE (1 << 2)
+#define AMDGPU_FW_SHARED_FLAG_0_DRM_KEY_INJECT (1 << 4)
+#define AMDGPU_VCN_FW_SHARED_FLAG_0_RB	(1 << 6)
+#define AMDGPU_VCN_MULTI_QUEUE_FLAG	(1 << 8)
+#define AMDGPU_VCN_SW_RING_FLAG		(1 << 9)
+#define AMDGPU_VCN_FW_LOGGING_FLAG	(1 << 10)
+#define AMDGPU_VCN_SMU_VERSION_INFO_FLAG (1 << 11)
+#define AMDGPU_VCN_SMU_DPM_INTERFACE_FLAG (1 << 11)
+#define AMDGPU_VCN_VF_RB_SETUP_FLAG (1 << 14)
+#define AMDGPU_VCN_VF_RB_DECOUPLE_FLAG (1 << 15)
+
+#define MAX_NUM_VCN_RB_SETUP 4
+
+#define AMDGPU_VCN_IB_FLAG_DECODE_BUFFER	0x00000001
+#define AMDGPU_VCN_CMD_FLAG_MSG_BUFFER		0x00000001
+
+#define VCN_CODEC_DISABLE_MASK_AV1  (1 << 0)
+#define VCN_CODEC_DISABLE_MASK_VP9  (1 << 1)
+#define VCN_CODEC_DISABLE_MASK_HEVC (1 << 2)
+#define VCN_CODEC_DISABLE_MASK_H264 (1 << 3)
+
+#define AMDGPU_VCN_SMU_DPM_INTERFACE_DGPU (0)
+#define AMDGPU_VCN_SMU_DPM_INTERFACE_APU (1)
+
+#define AMDGPU_DRM_KEY_INJECT_WORKAROUND_VCNFW_ASD_HANDSHAKING 2
+
+struct amdgpu_hwip_reg_entry;
+
+enum amdgpu_vcn_caps {
+	AMDGPU_VCN_RRMT_ENABLED,
+};
+
+#define AMDGPU_VCN_CAPS(caps) BIT(AMDGPU_VCN_##caps)
+
+enum fw_queue_mode {
+	FW_QUEUE_RING_RESET = 1,
+	FW_QUEUE_DPG_HOLD_OFF = 2,
+};
+
+enum engine_status_constants {
+	UVD_PGFSM_STATUS__UVDM_UVDU_PWR_ON = 0x2AAAA0,
+	UVD_PGFSM_STATUS__UVDM_UVDU_PWR_ON_2_0 = 0xAAAA0,
+	UVD_PGFSM_STATUS__UVDM_UVDU_UVDLM_PWR_ON_3_0 = 0x2A2A8AA0,
+	UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON = 0x00000002,
+	UVD_STATUS__UVD_BUSY = 0x00000004,
+	GB_ADDR_CONFIG_DEFAULT = 0x26010011,
+	UVD_STATUS__IDLE = 0x2,
+	UVD_STATUS__BUSY = 0x5,
+	UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF = 0x1,
+	UVD_STATUS__RBC_BUSY = 0x1,
+	UVD_PGFSM_STATUS_UVDJ_PWR_ON = 0,
+};
+
+enum internal_dpg_state {
+	VCN_DPG_STATE__UNPAUSE = 0,
+	VCN_DPG_STATE__PAUSE,
+};
+
+struct dpg_pause_state {
+	enum internal_dpg_state fw_based;
+	enum internal_dpg_state jpeg;
+};
+
+struct amdgpu_vcn_reg{
+	unsigned	data0;
+	unsigned	data1;
+	unsigned	cmd;
+	unsigned	nop;
+	unsigned	context_id;
+	unsigned	ib_vmid;
+	unsigned	ib_bar_low;
+	unsigned	ib_bar_high;
+	unsigned	ib_size;
+	unsigned	gp_scratch8;
+	unsigned	scratch9;
+};
+
+struct amdgpu_vcn_fw_shared {
+	void        *cpu_addr;
+	uint64_t    gpu_addr;
+	uint32_t    mem_size;
+	uint32_t    log_offset;
+};
+
+struct amdgpu_vcn_inst {
+	struct amdgpu_device	*adev;
+	int			inst;
 	struct amdgpu_bo	*vcpu_bo;
 	void			*cpu_addr;
 	uint64_t		gpu_addr;
-	unsigned		fw_version;
 	void			*saved_bo;
-	struct delayed_work	idle_work;
-	const struct firmware	*fw;	/* VCN firmware */
 	struct amdgpu_ring	ring_dec;
 	struct amdgpu_ring	ring_enc[AMDGPU_VCN_MAX_ENC_RINGS];
+	atomic_t		sched_score;
 	struct amdgpu_irq_src	irq;
-	struct amd_sched_entity entity_dec;
-	struct amd_sched_entity entity_enc;
+	struct amdgpu_irq_src	ras_poison_irq;
+	struct amdgpu_vcn_reg	external;
+	struct amdgpu_bo	*dpg_sram_bo;
+	struct dpg_pause_state	pause_state;
+	void			*dpg_sram_cpu_addr;
+	uint64_t		dpg_sram_gpu_addr;
+	uint32_t		*dpg_sram_curr_addr;
+	atomic_t		dpg_enc_submission_cnt;
+	struct amdgpu_vcn_fw_shared fw_shared;
+	uint8_t			aid_id;
+	const struct firmware	*fw; /* VCN firmware */
+	uint8_t			vcn_config;
+	uint32_t		vcn_codec_disable_mask;
+	atomic_t		total_submission_cnt;
+	struct mutex		vcn_pg_lock;
+	enum amd_powergating_state cur_state;
+	struct delayed_work	idle_work;
+	unsigned		fw_version;
 	unsigned		num_enc_rings;
+	bool			indirect_sram;
+	struct amdgpu_vcn_reg	 internal;
+	struct mutex		vcn1_jpeg1_workaround;
+	int (*pause_dpg_mode)(struct amdgpu_vcn_inst *vinst,
+			      struct dpg_pause_state *new_state);
+	int (*set_pg_state)(struct amdgpu_vcn_inst *vinst,
+			    enum amd_powergating_state state);
+	int (*reset)(struct amdgpu_vcn_inst *vinst);
+	bool using_unified_queue;
+	struct mutex		engine_reset_mutex;
+};
+
+struct amdgpu_vcn_ras {
+	struct amdgpu_ras_block_object ras_block;
+};
+
+struct amdgpu_vcn {
+	uint8_t	num_vcn_inst;
+	struct amdgpu_vcn_inst	 inst[AMDGPU_MAX_VCN_INSTANCES];
+
+	unsigned	harvest_config;
+
+	struct ras_common_if    *ras_if;
+	struct amdgpu_vcn_ras   *ras;
+
+	uint16_t inst_mask;
+	uint8_t	num_inst_per_aid;
+
+	/* IP reg dump */
+	uint32_t		*ip_dump;
+
+	uint32_t		supported_reset;
+	uint32_t		caps;
+
+	bool			per_inst_fw;
+	unsigned		fw_version;
+
+	bool			workload_profile_active;
+	struct mutex            workload_profile_mutex;
+	u32 reg_count;
+	const struct amdgpu_hwip_reg_entry *reg_list;
+};
+
+struct amdgpu_fw_shared_rb_ptrs_struct {
+	/* to WA DPG R/W ptr issues.*/
+	uint32_t  rptr;
+	uint32_t  wptr;
+};
+
+struct amdgpu_fw_shared_multi_queue {
+	uint8_t decode_queue_mode;
+	uint8_t encode_generalpurpose_queue_mode;
+	uint8_t encode_lowlatency_queue_mode;
+	uint8_t encode_realtime_queue_mode;
+	uint8_t padding[4];
+};
+
+struct amdgpu_fw_shared_sw_ring {
+	uint8_t is_enabled;
+	uint8_t padding[3];
+};
+
+struct amdgpu_fw_shared_unified_queue_struct {
+	uint8_t is_enabled;
+	uint8_t queue_mode;
+	uint8_t queue_status;
+	uint8_t padding[5];
+};
+
+struct amdgpu_fw_shared_fw_logging {
+	uint8_t is_enabled;
+	uint32_t addr_lo;
+	uint32_t addr_hi;
+	uint32_t size;
+};
+
+struct amdgpu_fw_shared_smu_interface_info {
+	uint8_t smu_interface_type;
+	uint8_t padding[3];
+};
+
+struct amdgpu_fw_shared {
+	uint32_t present_flag_0;
+	uint8_t pad[44];
+	struct amdgpu_fw_shared_rb_ptrs_struct rb;
+	uint8_t pad1[1];
+	struct amdgpu_fw_shared_multi_queue multi_queue;
+	struct amdgpu_fw_shared_sw_ring sw_ring;
+	struct amdgpu_fw_shared_fw_logging fw_log;
+	struct amdgpu_fw_shared_smu_interface_info smu_interface_info;
+};
+
+struct amdgpu_vcn_rb_setup_info {
+	uint32_t  rb_addr_lo;
+	uint32_t  rb_addr_hi;
+	uint32_t  rb_size;
+};
+
+struct amdgpu_fw_shared_rb_setup {
+	uint32_t is_rb_enabled_flags;
+
+	union {
+		struct {
+			uint32_t rb_addr_lo;
+			uint32_t rb_addr_hi;
+			uint32_t  rb_size;
+			uint32_t  rb4_addr_lo;
+			uint32_t  rb4_addr_hi;
+			uint32_t  rb4_size;
+			uint32_t  reserved[6];
+		};
+
+		struct {
+			struct amdgpu_vcn_rb_setup_info rb_info[MAX_NUM_VCN_RB_SETUP];
+		};
+	};
 };
 
-int amdgpu_vcn_sw_init(struct amdgpu_device *adev);
-int amdgpu_vcn_sw_fini(struct amdgpu_device *adev);
-int amdgpu_vcn_suspend(struct amdgpu_device *adev);
-int amdgpu_vcn_resume(struct amdgpu_device *adev);
+struct amdgpu_fw_shared_drm_key_wa {
+	uint8_t  method;
+	uint8_t  reserved[3];
+};
+
+struct amdgpu_fw_shared_queue_decouple {
+	uint8_t  is_enabled;
+	uint8_t  reserved[7];
+};
+
+struct amdgpu_vcn4_fw_shared {
+	uint32_t present_flag_0;
+	uint8_t pad[12];
+	struct amdgpu_fw_shared_unified_queue_struct sq;
+	uint8_t pad1[8];
+	struct amdgpu_fw_shared_fw_logging fw_log;
+	uint8_t pad2[20];
+	struct amdgpu_fw_shared_rb_setup rb_setup;
+	struct amdgpu_fw_shared_smu_interface_info smu_dpm_interface;
+	struct amdgpu_fw_shared_drm_key_wa drm_key_wa;
+	uint8_t pad3[9];
+	struct amdgpu_fw_shared_queue_decouple decouple;
+};
+
+struct amdgpu_vcn_fwlog {
+	uint32_t rptr;
+	uint32_t wptr;
+	uint32_t buffer_size;
+	uint32_t header_size;
+	uint8_t wrapped;
+};
+
+struct amdgpu_vcn_decode_buffer {
+	uint32_t valid_buf_flag;
+	uint32_t msg_buffer_address_hi;
+	uint32_t msg_buffer_address_lo;
+	uint32_t pad[30];
+};
+
+struct amdgpu_vcn_rb_metadata {
+	uint32_t size;
+	uint32_t present_flag_0;
+
+	uint8_t version;
+	uint8_t ring_id;
+	uint8_t pad[26];
+};
+
+struct amdgpu_vcn5_fw_shared {
+	uint32_t present_flag_0;
+	uint8_t pad[12];
+	struct amdgpu_fw_shared_unified_queue_struct sq;
+	uint8_t pad1[8];
+	struct amdgpu_fw_shared_fw_logging fw_log;
+	uint8_t pad2[20];
+	struct amdgpu_fw_shared_rb_setup rb_setup;
+	struct amdgpu_fw_shared_smu_interface_info smu_dpm_interface;
+	struct amdgpu_fw_shared_drm_key_wa drm_key_wa;
+	uint8_t pad3[404];
+};
+
+#define VCN_BLOCK_ENCODE_DISABLE_MASK 0x80
+#define VCN_BLOCK_DECODE_DISABLE_MASK 0x40
+#define VCN_BLOCK_QUEUE_DISABLE_MASK 0xC0
+
+enum vcn_ring_type {
+	VCN_ENCODE_RING,
+	VCN_DECODE_RING,
+	VCN_UNIFIED_RING,
+};
+
+int amdgpu_vcn_early_init(struct amdgpu_device *adev, int i);
+int amdgpu_vcn_sw_init(struct amdgpu_device *adev, int i);
+void amdgpu_vcn_sw_fini(struct amdgpu_device *adev, int i);
+int amdgpu_vcn_suspend(struct amdgpu_device *adev, int i);
+int amdgpu_vcn_resume(struct amdgpu_device *adev, int i);
 void amdgpu_vcn_ring_begin_use(struct amdgpu_ring *ring);
 void amdgpu_vcn_ring_end_use(struct amdgpu_ring *ring);
 
+bool amdgpu_vcn_is_disabled_vcn(struct amdgpu_device *adev,
+				enum vcn_ring_type type, uint32_t vcn_instance);
+
 int amdgpu_vcn_dec_ring_test_ring(struct amdgpu_ring *ring);
 int amdgpu_vcn_dec_ring_test_ib(struct amdgpu_ring *ring, long timeout);
+int amdgpu_vcn_dec_sw_ring_test_ring(struct amdgpu_ring *ring);
+int amdgpu_vcn_dec_sw_ring_test_ib(struct amdgpu_ring *ring, long timeout);
+int amdgpu_vcn_unified_ring_test_ib(struct amdgpu_ring *ring, long timeout);
 
 int amdgpu_vcn_enc_ring_test_ring(struct amdgpu_ring *ring);
 int amdgpu_vcn_enc_ring_test_ib(struct amdgpu_ring *ring, long timeout);
 
+enum amdgpu_ring_priority_level amdgpu_vcn_get_enc_ring_prio(int ring);
+
+void amdgpu_vcn_setup_ucode(struct amdgpu_device *adev, int i);
+
+void amdgpu_vcn_fwlog_init(struct amdgpu_vcn_inst *vcn);
+void amdgpu_debugfs_vcn_fwlog_init(struct amdgpu_device *adev,
+                                   uint8_t i, struct amdgpu_vcn_inst *vcn);
+
+int amdgpu_vcn_process_poison_irq(struct amdgpu_device *adev,
+			struct amdgpu_irq_src *source,
+			struct amdgpu_iv_entry *entry);
+int amdgpu_vcn_ras_late_init(struct amdgpu_device *adev,
+			struct ras_common_if *ras_block);
+int amdgpu_vcn_ras_sw_init(struct amdgpu_device *adev);
+
+int amdgpu_vcn_psp_update_sram(struct amdgpu_device *adev, int inst_idx,
+			       enum AMDGPU_UCODE_ID ucode_id);
+int amdgpu_vcn_save_vcpu_bo(struct amdgpu_device *adev);
+int amdgpu_vcn_sysfs_reset_mask_init(struct amdgpu_device *adev);
+void amdgpu_vcn_sysfs_reset_mask_fini(struct amdgpu_device *adev);
+void amdgpu_debugfs_vcn_sched_mask_init(struct amdgpu_device *adev);
+
+int vcn_set_powergating_state(struct amdgpu_ip_block *ip_block,
+			      enum amd_powergating_state state);
+int amdgpu_vcn_ring_reset(struct amdgpu_ring *ring,
+			  unsigned int vmid,
+			  struct amdgpu_fence *guilty_fence);
+int amdgpu_vcn_reg_dump_init(struct amdgpu_device *adev,
+			     const struct amdgpu_hwip_reg_entry *reg, u32 count);
+void amdgpu_vcn_dump_ip_state(struct amdgpu_ip_block *ip_block);
+void amdgpu_vcn_print_ip_state(struct amdgpu_ip_block *ip_block, struct drm_printer *p);
+void amdgpu_vcn_get_profile(struct amdgpu_device *adev);
+void amdgpu_vcn_put_profile(struct amdgpu_device *adev);
+
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vf_error.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vf_error.c
index 45ac91861965..f9d3d79f68b1 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vf_error.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vf_error.c
@@ -25,30 +25,26 @@
 #include "amdgpu_vf_error.h"
 #include "mxgpu_ai.h"
 
-#define AMDGPU_VF_ERROR_ENTRY_SIZE    16 
-
-/* struct error_entry - amdgpu VF error information. */
-struct amdgpu_vf_error_buffer {
-	int read_count;
-	int write_count;
-	uint16_t code[AMDGPU_VF_ERROR_ENTRY_SIZE];
-	uint16_t flags[AMDGPU_VF_ERROR_ENTRY_SIZE];
-	uint64_t data[AMDGPU_VF_ERROR_ENTRY_SIZE];
-};
-
-struct amdgpu_vf_error_buffer admgpu_vf_errors;
-
-
-void amdgpu_vf_error_put(uint16_t sub_error_code, uint16_t error_flags, uint64_t error_data)
+void amdgpu_vf_error_put(struct amdgpu_device *adev,
+			 uint16_t sub_error_code,
+			 uint16_t error_flags,
+			 uint64_t error_data)
 {
 	int index;
-	uint16_t error_code = AMDGIM_ERROR_CODE(AMDGIM_ERROR_CATEGORY_VF, sub_error_code);
+	uint16_t error_code;
 
-	index = admgpu_vf_errors.write_count % AMDGPU_VF_ERROR_ENTRY_SIZE;
-	admgpu_vf_errors.code [index] = error_code;
-	admgpu_vf_errors.flags [index] = error_flags;
-	admgpu_vf_errors.data [index] = error_data;
-	admgpu_vf_errors.write_count ++;
+	if (!amdgpu_sriov_vf(adev))
+		return;
+
+	error_code = AMDGIM_ERROR_CODE(AMDGIM_ERROR_CATEGORY_VF, sub_error_code);
+
+	mutex_lock(&adev->virt.vf_errors.lock);
+	index = adev->virt.vf_errors.write_count % AMDGPU_VF_ERROR_ENTRY_SIZE;
+	adev->virt.vf_errors.code [index] = error_code;
+	adev->virt.vf_errors.flags [index] = error_flags;
+	adev->virt.vf_errors.data [index] = error_data;
+	adev->virt.vf_errors.write_count ++;
+	mutex_unlock(&adev->virt.vf_errors.lock);
 }
 
 
@@ -58,28 +54,33 @@ void amdgpu_vf_error_trans_all(struct amdgpu_device *adev)
 	u32 data1, data2, data3;
 	int index;
 
-	if ((NULL == adev) || (!amdgpu_sriov_vf(adev)) || (!adev->virt.ops) || (!adev->virt.ops->trans_msg)) {
+	if ((NULL == adev) || (!amdgpu_sriov_vf(adev)) ||
+	    (!adev->virt.ops) || (!adev->virt.ops->trans_msg)) {
 		return;
 	}
 /*
- 	TODO: Enable these code when pv2vf_info is merged
+	TODO: Enable these code when pv2vf_info is merged
 	AMDGPU_FW_VRAM_PF2VF_READ (adev, feature_flags, &pf2vf_flags);
 	if (!(pf2vf_flags & AMDGIM_FEATURE_ERROR_LOG_COLLECT)) {
 		return;
 	}
 */
+
+	mutex_lock(&adev->virt.vf_errors.lock);
 	/* The errors are overlay of array, correct read_count as full. */
-	if (admgpu_vf_errors.write_count - admgpu_vf_errors.read_count > AMDGPU_VF_ERROR_ENTRY_SIZE) {
-		admgpu_vf_errors.read_count = admgpu_vf_errors.write_count - AMDGPU_VF_ERROR_ENTRY_SIZE;
+	if (adev->virt.vf_errors.write_count - adev->virt.vf_errors.read_count > AMDGPU_VF_ERROR_ENTRY_SIZE) {
+		adev->virt.vf_errors.read_count = adev->virt.vf_errors.write_count - AMDGPU_VF_ERROR_ENTRY_SIZE;
 	}
 
-	while (admgpu_vf_errors.read_count < admgpu_vf_errors.write_count) {
-		index =admgpu_vf_errors.read_count % AMDGPU_VF_ERROR_ENTRY_SIZE;
-		data1 = AMDGIM_ERROR_CODE_FLAGS_TO_MAILBOX (admgpu_vf_errors.code[index], admgpu_vf_errors.flags[index]);
-		data2 = admgpu_vf_errors.data[index] & 0xFFFFFFFF;
-		data3 = (admgpu_vf_errors.data[index] >> 32) & 0xFFFFFFFF;
+	while (adev->virt.vf_errors.read_count < adev->virt.vf_errors.write_count) {
+		index =adev->virt.vf_errors.read_count % AMDGPU_VF_ERROR_ENTRY_SIZE;
+		data1 = AMDGIM_ERROR_CODE_FLAGS_TO_MAILBOX(adev->virt.vf_errors.code[index],
+							   adev->virt.vf_errors.flags[index]);
+		data2 = adev->virt.vf_errors.data[index] & 0xFFFFFFFF;
+		data3 = (adev->virt.vf_errors.data[index] >> 32) & 0xFFFFFFFF;
 
 		adev->virt.ops->trans_msg(adev, IDH_LOG_VF_ERROR, data1, data2, data3);
-		admgpu_vf_errors.read_count ++;
+		adev->virt.vf_errors.read_count ++;
 	}
+	mutex_unlock(&adev->virt.vf_errors.lock);
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vf_error.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vf_error.h
index 2a3278ec76ba..6436bd053325 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vf_error.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vf_error.h
@@ -56,7 +56,10 @@ enum AMDGIM_ERROR_CATEGORY {
 	AMDGIM_ERROR_CATEGORY_MAX
 };
 
-void amdgpu_vf_error_put(uint16_t sub_error_code, uint16_t error_flags, uint64_t error_data);
+void amdgpu_vf_error_put(struct amdgpu_device *adev,
+			 uint16_t sub_error_code,
+			 uint16_t error_flags,
+			 uint64_t error_data);
 void amdgpu_vf_error_trans_all (struct amdgpu_device *adev);
 
 #endif /* __VF_ERROR_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
index ab05121b9272..3328ab63376b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
@@ -21,150 +21,61 @@
  *
  */
 
-#include "amdgpu.h"
-#define MAX_KIQ_REG_WAIT	100000
+#include <linux/module.h>
 
-int amdgpu_allocate_static_csa(struct amdgpu_device *adev)
-{
-	int r;
-	void *ptr;
+#ifdef CONFIG_X86
+#include <asm/hypervisor.h>
+#endif
 
-	r = amdgpu_bo_create_kernel(adev, AMDGPU_CSA_SIZE, PAGE_SIZE,
-				AMDGPU_GEM_DOMAIN_VRAM, &adev->virt.csa_obj,
-				&adev->virt.csa_vmid0_addr, &ptr);
-	if (r)
-		return r;
+#include <drm/drm_drv.h>
+#include <xen/xen.h>
 
-	memset(ptr, 0, AMDGPU_CSA_SIZE);
-	return 0;
-}
+#include "amdgpu.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_reset.h"
+#include "amdgpu_dpm.h"
+#include "vi.h"
+#include "soc15.h"
+#include "nv.h"
 
-/*
- * amdgpu_map_static_csa should be called during amdgpu_vm_init
- * it maps virtual address "AMDGPU_VA_RESERVED_SIZE - AMDGPU_CSA_SIZE"
- * to this VM, and each command submission of GFX should use this virtual
- * address within META_DATA init package to support SRIOV gfx preemption.
- */
+#define POPULATE_UCODE_INFO(vf2pf_info, ucode, ver) \
+	do { \
+		vf2pf_info->ucode_info[ucode].id = ucode; \
+		vf2pf_info->ucode_info[ucode].version = ver; \
+	} while (0)
 
-int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm,
-			  struct amdgpu_bo_va **bo_va)
+bool amdgpu_virt_mmio_blocked(struct amdgpu_device *adev)
 {
-	struct ww_acquire_ctx ticket;
-	struct list_head list;
-	struct amdgpu_bo_list_entry pd;
-	struct ttm_validate_buffer csa_tv;
-	int r;
-
-	INIT_LIST_HEAD(&list);
-	INIT_LIST_HEAD(&csa_tv.head);
-	csa_tv.bo = &adev->virt.csa_obj->tbo;
-	csa_tv.shared = true;
-
-	list_add(&csa_tv.head, &list);
-	amdgpu_vm_get_pd_bo(vm, &list, &pd);
-
-	r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL);
-	if (r) {
-		DRM_ERROR("failed to reserve CSA,PD BOs: err=%d\n", r);
-		return r;
-	}
-
-	*bo_va = amdgpu_vm_bo_add(adev, vm, adev->virt.csa_obj);
-	if (!*bo_va) {
-		ttm_eu_backoff_reservation(&ticket, &list);
-		DRM_ERROR("failed to create bo_va for static CSA\n");
-		return -ENOMEM;
-	}
-
-	r = amdgpu_vm_alloc_pts(adev, (*bo_va)->base.vm, AMDGPU_CSA_VADDR,
-				AMDGPU_CSA_SIZE);
-	if (r) {
-		DRM_ERROR("failed to allocate pts for static CSA, err=%d\n", r);
-		amdgpu_vm_bo_rmv(adev, *bo_va);
-		ttm_eu_backoff_reservation(&ticket, &list);
-		return r;
-	}
-
-	r = amdgpu_vm_bo_map(adev, *bo_va, AMDGPU_CSA_VADDR, 0, AMDGPU_CSA_SIZE,
-			     AMDGPU_PTE_READABLE | AMDGPU_PTE_WRITEABLE |
-			     AMDGPU_PTE_EXECUTABLE);
-
-	if (r) {
-		DRM_ERROR("failed to do bo_map on static CSA, err=%d\n", r);
-		amdgpu_vm_bo_rmv(adev, *bo_va);
-		ttm_eu_backoff_reservation(&ticket, &list);
-		return r;
-	}
-
-	ttm_eu_backoff_reservation(&ticket, &list);
-	return 0;
+	/* By now all MMIO pages except mailbox are blocked */
+	/* if blocking is enabled in hypervisor. Choose the */
+	/* SCRATCH_REG0 to test. */
+	return RREG32_NO_KIQ(0xc040) == 0xffffffff;
 }
 
 void amdgpu_virt_init_setting(struct amdgpu_device *adev)
 {
+	struct drm_device *ddev = adev_to_drm(adev);
+
 	/* enable virtual display */
-	adev->mode_info.num_crtc = 1;
-	adev->enable_virtual_display = true;
+	if (adev->asic_type != CHIP_ALDEBARAN &&
+	    adev->asic_type != CHIP_ARCTURUS &&
+	    ((adev->pdev->class >> 8) != PCI_CLASS_ACCELERATOR_PROCESSING)) {
+		if (adev->mode_info.num_crtc == 0)
+			adev->mode_info.num_crtc = 1;
+		adev->enable_virtual_display = true;
+	}
+	ddev->driver_features &= ~DRIVER_ATOMIC;
 	adev->cg_flags = 0;
 	adev->pg_flags = 0;
 
-	mutex_init(&adev->virt.lock_reset);
-}
-
-uint32_t amdgpu_virt_kiq_rreg(struct amdgpu_device *adev, uint32_t reg)
-{
-	signed long r;
-	uint32_t val;
-	struct dma_fence *f;
-	struct amdgpu_kiq *kiq = &adev->gfx.kiq;
-	struct amdgpu_ring *ring = &kiq->ring;
-
-	BUG_ON(!ring->funcs->emit_rreg);
-
-	mutex_lock(&kiq->ring_mutex);
-	amdgpu_ring_alloc(ring, 32);
-	amdgpu_ring_emit_rreg(ring, reg);
-	amdgpu_fence_emit(ring, &f);
-	amdgpu_ring_commit(ring);
-	mutex_unlock(&kiq->ring_mutex);
-
-	r = dma_fence_wait_timeout(f, false, msecs_to_jiffies(MAX_KIQ_REG_WAIT));
-	dma_fence_put(f);
-	if (r < 1) {
-		DRM_ERROR("wait for kiq fence error: %ld.\n", r);
-		return ~0;
-	}
-
-	val = adev->wb.wb[adev->virt.reg_val_offs];
-
-	return val;
-}
-
-void amdgpu_virt_kiq_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v)
-{
-	signed long r;
-	struct dma_fence *f;
-	struct amdgpu_kiq *kiq = &adev->gfx.kiq;
-	struct amdgpu_ring *ring = &kiq->ring;
-
-	BUG_ON(!ring->funcs->emit_wreg);
-
-	mutex_lock(&kiq->ring_mutex);
-	amdgpu_ring_alloc(ring, 32);
-	amdgpu_ring_emit_wreg(ring, reg, v);
-	amdgpu_fence_emit(ring, &f);
-	amdgpu_ring_commit(ring);
-	mutex_unlock(&kiq->ring_mutex);
-
-	r = dma_fence_wait_timeout(f, false, msecs_to_jiffies(MAX_KIQ_REG_WAIT));
-	if (r < 1)
-		DRM_ERROR("wait for kiq fence error: %ld.\n", r);
-	dma_fence_put(f);
+	/* Reduce kcq number to 2 to reduce latency */
+	if (amdgpu_num_kcq == -1)
+		amdgpu_num_kcq = 2;
 }
 
 /**
  * amdgpu_virt_request_full_gpu() - request full gpu access
- * @amdgpu:	amdgpu device.
+ * @adev:	amdgpu device.
  * @init:	is driver init time.
  * When start to init/fini driver, first need to request full gpu access.
  * Return: Zero if request success, otherwise will return error.
@@ -176,8 +87,10 @@ int amdgpu_virt_request_full_gpu(struct amdgpu_device *adev, bool init)
 
 	if (virt->ops && virt->ops->req_full_gpu) {
 		r = virt->ops->req_full_gpu(adev, init);
-		if (r)
+		if (r) {
+			adev->no_hw_access = true;
 			return r;
+		}
 
 		adev->virt.caps &= ~AMDGPU_SRIOV_CAPS_RUNTIME;
 	}
@@ -187,7 +100,7 @@ int amdgpu_virt_request_full_gpu(struct amdgpu_device *adev, bool init)
 
 /**
  * amdgpu_virt_release_full_gpu() - release full gpu access
- * @amdgpu:	amdgpu device.
+ * @adev:	amdgpu device.
  * @init:	is driver init time.
  * When finishing driver init/fini, need to release full gpu access.
  * Return: Zero if release success, otherwise will returen error.
@@ -209,7 +122,7 @@ int amdgpu_virt_release_full_gpu(struct amdgpu_device *adev, bool init)
 
 /**
  * amdgpu_virt_reset_gpu() - reset gpu
- * @amdgpu:	amdgpu device.
+ * @adev:	amdgpu device.
  * Send reset command to GPU hypervisor to reset GPU that VM is using
  * Return: Zero if reset success, otherwise will return error.
  */
@@ -229,9 +142,52 @@ int amdgpu_virt_reset_gpu(struct amdgpu_device *adev)
 	return 0;
 }
 
+void amdgpu_virt_request_init_data(struct amdgpu_device *adev)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+
+	if (virt->ops && virt->ops->req_init_data)
+		virt->ops->req_init_data(adev);
+
+	if (adev->virt.req_init_data_ver > 0)
+		DRM_INFO("host supports REQ_INIT_DATA handshake\n");
+	else
+		DRM_WARN("host doesn't support REQ_INIT_DATA handshake\n");
+}
+
+/**
+ * amdgpu_virt_ready_to_reset() - send ready to reset to host
+ * @adev:	amdgpu device.
+ * Send ready to reset message to GPU hypervisor to signal we have stopped GPU
+ * activity and is ready for host FLR
+ */
+void amdgpu_virt_ready_to_reset(struct amdgpu_device *adev)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+
+	if (virt->ops && virt->ops->reset_gpu)
+		virt->ops->ready_to_reset(adev);
+}
+
+/**
+ * amdgpu_virt_wait_reset() - wait for reset gpu completed
+ * @adev:	amdgpu device.
+ * Wait for GPU reset completed.
+ * Return: Zero if reset success, otherwise will return error.
+ */
+int amdgpu_virt_wait_reset(struct amdgpu_device *adev)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+
+	if (!virt->ops || !virt->ops->wait_reset)
+		return -EINVAL;
+
+	return virt->ops->wait_reset(adev);
+}
+
 /**
  * amdgpu_virt_alloc_mm_table() - alloc memory for mm table
- * @amdgpu:	amdgpu device.
+ * @adev:	amdgpu device.
  * MM table is used by UVD and VCE for its initialization
  * Return: Zero if allocate success.
  */
@@ -243,7 +199,8 @@ int amdgpu_virt_alloc_mm_table(struct amdgpu_device *adev)
 		return 0;
 
 	r = amdgpu_bo_create_kernel(adev, PAGE_SIZE, PAGE_SIZE,
-				    AMDGPU_GEM_DOMAIN_VRAM,
+				    AMDGPU_GEM_DOMAIN_VRAM |
+				    AMDGPU_GEM_DOMAIN_GTT,
 				    &adev->virt.mm_table.bo,
 				    &adev->virt.mm_table.gpu_addr,
 				    (void *)&adev->virt.mm_table.cpu_addr);
@@ -261,7 +218,7 @@ int amdgpu_virt_alloc_mm_table(struct amdgpu_device *adev)
 
 /**
  * amdgpu_virt_free_mm_table() - free mm table memory
- * @amdgpu:	amdgpu device.
+ * @adev:	amdgpu device.
  * Free MM table memory
  */
 void amdgpu_virt_free_mm_table(struct amdgpu_device *adev)
@@ -274,3 +231,1328 @@ void amdgpu_virt_free_mm_table(struct amdgpu_device *adev)
 			      (void *)&adev->virt.mm_table.cpu_addr);
 	adev->virt.mm_table.gpu_addr = 0;
 }
+
+/**
+ * amdgpu_virt_rcvd_ras_interrupt() - receive ras interrupt
+ * @adev:	amdgpu device.
+ * Check whether host sent RAS error message
+ * Return: true if found, otherwise false
+ */
+bool amdgpu_virt_rcvd_ras_interrupt(struct amdgpu_device *adev)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+
+	if (!virt->ops || !virt->ops->rcvd_ras_intr)
+		return false;
+
+	return virt->ops->rcvd_ras_intr(adev);
+}
+
+
+unsigned int amd_sriov_msg_checksum(void *obj,
+				unsigned long obj_size,
+				unsigned int key,
+				unsigned int checksum)
+{
+	unsigned int ret = key;
+	unsigned long i = 0;
+	unsigned char *pos;
+
+	pos = (char *)obj;
+	/* calculate checksum */
+	for (i = 0; i < obj_size; ++i)
+		ret += *(pos + i);
+	/* minus the checksum itself */
+	pos = (char *)&checksum;
+	for (i = 0; i < sizeof(checksum); ++i)
+		ret -= *(pos + i);
+	return ret;
+}
+
+static int amdgpu_virt_init_ras_err_handler_data(struct amdgpu_device *adev)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+	struct amdgpu_virt_ras_err_handler_data **data = &virt->virt_eh_data;
+	/* GPU will be marked bad on host if bp count more then 10,
+	 * so alloc 512 is enough.
+	 */
+	unsigned int align_space = 512;
+	void *bps = NULL;
+	struct amdgpu_bo **bps_bo = NULL;
+
+	*data = kmalloc(sizeof(struct amdgpu_virt_ras_err_handler_data), GFP_KERNEL);
+	if (!*data)
+		goto data_failure;
+
+	bps = kmalloc_array(align_space, sizeof(*(*data)->bps), GFP_KERNEL);
+	if (!bps)
+		goto bps_failure;
+
+	bps_bo = kmalloc_array(align_space, sizeof(*(*data)->bps_bo), GFP_KERNEL);
+	if (!bps_bo)
+		goto bps_bo_failure;
+
+	(*data)->bps = bps;
+	(*data)->bps_bo = bps_bo;
+	(*data)->count = 0;
+	(*data)->last_reserved = 0;
+
+	virt->ras_init_done = true;
+
+	return 0;
+
+bps_bo_failure:
+	kfree(bps);
+bps_failure:
+	kfree(*data);
+data_failure:
+	return -ENOMEM;
+}
+
+static void amdgpu_virt_ras_release_bp(struct amdgpu_device *adev)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+	struct amdgpu_virt_ras_err_handler_data *data = virt->virt_eh_data;
+	struct amdgpu_bo *bo;
+	int i;
+
+	if (!data)
+		return;
+
+	for (i = data->last_reserved - 1; i >= 0; i--) {
+		bo = data->bps_bo[i];
+		if (bo) {
+			amdgpu_bo_free_kernel(&bo, NULL, NULL);
+			data->bps_bo[i] = bo;
+		}
+		data->last_reserved = i;
+	}
+}
+
+void amdgpu_virt_release_ras_err_handler_data(struct amdgpu_device *adev)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+	struct amdgpu_virt_ras_err_handler_data *data = virt->virt_eh_data;
+
+	virt->ras_init_done = false;
+
+	if (!data)
+		return;
+
+	amdgpu_virt_ras_release_bp(adev);
+
+	kfree(data->bps);
+	kfree(data->bps_bo);
+	kfree(data);
+	virt->virt_eh_data = NULL;
+}
+
+static void amdgpu_virt_ras_add_bps(struct amdgpu_device *adev,
+		struct eeprom_table_record *bps, int pages)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+	struct amdgpu_virt_ras_err_handler_data *data = virt->virt_eh_data;
+
+	if (!data)
+		return;
+
+	memcpy(&data->bps[data->count], bps, pages * sizeof(*data->bps));
+	data->count += pages;
+}
+
+static void amdgpu_virt_ras_reserve_bps(struct amdgpu_device *adev)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+	struct amdgpu_virt_ras_err_handler_data *data = virt->virt_eh_data;
+	struct amdgpu_vram_mgr *mgr = &adev->mman.vram_mgr;
+	struct ttm_resource_manager *man = &mgr->manager;
+	struct amdgpu_bo *bo = NULL;
+	uint64_t bp;
+	int i;
+
+	if (!data)
+		return;
+
+	for (i = data->last_reserved; i < data->count; i++) {
+		bp = data->bps[i].retired_page;
+
+		/* There are two cases of reserve error should be ignored:
+		 * 1) a ras bad page has been allocated (used by someone);
+		 * 2) a ras bad page has been reserved (duplicate error injection
+		 *    for one page);
+		 */
+		if  (ttm_resource_manager_used(man)) {
+			amdgpu_vram_mgr_reserve_range(&adev->mman.vram_mgr,
+				bp << AMDGPU_GPU_PAGE_SHIFT,
+				AMDGPU_GPU_PAGE_SIZE);
+			data->bps_bo[i] = NULL;
+		} else {
+			if (amdgpu_bo_create_kernel_at(adev, bp << AMDGPU_GPU_PAGE_SHIFT,
+							AMDGPU_GPU_PAGE_SIZE,
+							&bo, NULL))
+				DRM_DEBUG("RAS WARN: reserve vram for retired page %llx fail\n", bp);
+			data->bps_bo[i] = bo;
+		}
+		data->last_reserved = i + 1;
+		bo = NULL;
+	}
+}
+
+static bool amdgpu_virt_ras_check_bad_page(struct amdgpu_device *adev,
+		uint64_t retired_page)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+	struct amdgpu_virt_ras_err_handler_data *data = virt->virt_eh_data;
+	int i;
+
+	if (!data)
+		return true;
+
+	for (i = 0; i < data->count; i++)
+		if (retired_page == data->bps[i].retired_page)
+			return true;
+
+	return false;
+}
+
+static void amdgpu_virt_add_bad_page(struct amdgpu_device *adev,
+		uint64_t bp_block_offset, uint32_t bp_block_size)
+{
+	struct eeprom_table_record bp;
+	uint64_t retired_page;
+	uint32_t bp_idx, bp_cnt;
+	void *vram_usage_va = NULL;
+
+	if (adev->mman.fw_vram_usage_va)
+		vram_usage_va = adev->mman.fw_vram_usage_va;
+	else
+		vram_usage_va = adev->mman.drv_vram_usage_va;
+
+	memset(&bp, 0, sizeof(bp));
+
+	if (bp_block_size) {
+		bp_cnt = bp_block_size / sizeof(uint64_t);
+		for (bp_idx = 0; bp_idx < bp_cnt; bp_idx++) {
+			retired_page = *(uint64_t *)(vram_usage_va +
+					bp_block_offset + bp_idx * sizeof(uint64_t));
+			bp.retired_page = retired_page;
+
+			if (amdgpu_virt_ras_check_bad_page(adev, retired_page))
+				continue;
+
+			amdgpu_virt_ras_add_bps(adev, &bp, 1);
+
+			amdgpu_virt_ras_reserve_bps(adev);
+		}
+	}
+}
+
+static int amdgpu_virt_read_pf2vf_data(struct amdgpu_device *adev)
+{
+	struct amd_sriov_msg_pf2vf_info_header *pf2vf_info = adev->virt.fw_reserve.p_pf2vf;
+	uint32_t checksum;
+	uint32_t checkval;
+
+	uint32_t i;
+	uint32_t tmp;
+
+	if (adev->virt.fw_reserve.p_pf2vf == NULL)
+		return -EINVAL;
+
+	if (pf2vf_info->size > 1024) {
+		dev_err(adev->dev, "invalid pf2vf message size: 0x%x\n", pf2vf_info->size);
+		return -EINVAL;
+	}
+
+	switch (pf2vf_info->version) {
+	case 1:
+		checksum = ((struct amdgim_pf2vf_info_v1 *)pf2vf_info)->checksum;
+		checkval = amd_sriov_msg_checksum(
+			adev->virt.fw_reserve.p_pf2vf, pf2vf_info->size,
+			adev->virt.fw_reserve.checksum_key, checksum);
+		if (checksum != checkval) {
+			dev_err(adev->dev,
+				"invalid pf2vf message: header checksum=0x%x calculated checksum=0x%x\n",
+				checksum, checkval);
+			return -EINVAL;
+		}
+
+		adev->virt.gim_feature =
+			((struct amdgim_pf2vf_info_v1 *)pf2vf_info)->feature_flags;
+		break;
+	case 2:
+		/* TODO: missing key, need to add it later */
+		checksum = ((struct amd_sriov_msg_pf2vf_info *)pf2vf_info)->checksum;
+		checkval = amd_sriov_msg_checksum(
+			adev->virt.fw_reserve.p_pf2vf, pf2vf_info->size,
+			0, checksum);
+		if (checksum != checkval) {
+			dev_err(adev->dev,
+				"invalid pf2vf message: header checksum=0x%x calculated checksum=0x%x\n",
+				checksum, checkval);
+			return -EINVAL;
+		}
+
+		adev->virt.vf2pf_update_interval_ms =
+			((struct amd_sriov_msg_pf2vf_info *)pf2vf_info)->vf2pf_update_interval_ms;
+		adev->virt.gim_feature =
+			((struct amd_sriov_msg_pf2vf_info *)pf2vf_info)->feature_flags.all;
+		adev->virt.reg_access =
+			((struct amd_sriov_msg_pf2vf_info *)pf2vf_info)->reg_access_flags.all;
+
+		adev->virt.decode_max_dimension_pixels = 0;
+		adev->virt.decode_max_frame_pixels = 0;
+		adev->virt.encode_max_dimension_pixels = 0;
+		adev->virt.encode_max_frame_pixels = 0;
+		adev->virt.is_mm_bw_enabled = false;
+		for (i = 0; i < AMD_SRIOV_MSG_RESERVE_VCN_INST; i++) {
+			tmp = ((struct amd_sriov_msg_pf2vf_info *)pf2vf_info)->mm_bw_management[i].decode_max_dimension_pixels;
+			adev->virt.decode_max_dimension_pixels = max(tmp, adev->virt.decode_max_dimension_pixels);
+
+			tmp = ((struct amd_sriov_msg_pf2vf_info *)pf2vf_info)->mm_bw_management[i].decode_max_frame_pixels;
+			adev->virt.decode_max_frame_pixels = max(tmp, adev->virt.decode_max_frame_pixels);
+
+			tmp = ((struct amd_sriov_msg_pf2vf_info *)pf2vf_info)->mm_bw_management[i].encode_max_dimension_pixels;
+			adev->virt.encode_max_dimension_pixels = max(tmp, adev->virt.encode_max_dimension_pixels);
+
+			tmp = ((struct amd_sriov_msg_pf2vf_info *)pf2vf_info)->mm_bw_management[i].encode_max_frame_pixels;
+			adev->virt.encode_max_frame_pixels = max(tmp, adev->virt.encode_max_frame_pixels);
+		}
+		if ((adev->virt.decode_max_dimension_pixels > 0) || (adev->virt.encode_max_dimension_pixels > 0))
+			adev->virt.is_mm_bw_enabled = true;
+
+		adev->unique_id =
+			((struct amd_sriov_msg_pf2vf_info *)pf2vf_info)->uuid;
+		adev->virt.ras_en_caps.all = ((struct amd_sriov_msg_pf2vf_info *)pf2vf_info)->ras_en_caps.all;
+		adev->virt.ras_telemetry_en_caps.all =
+			((struct amd_sriov_msg_pf2vf_info *)pf2vf_info)->ras_telemetry_en_caps.all;
+		break;
+	default:
+		dev_err(adev->dev, "invalid pf2vf version: 0x%x\n", pf2vf_info->version);
+		return -EINVAL;
+	}
+
+	/* correct too large or too little interval value */
+	if (adev->virt.vf2pf_update_interval_ms < 200 || adev->virt.vf2pf_update_interval_ms > 10000)
+		adev->virt.vf2pf_update_interval_ms = 2000;
+
+	return 0;
+}
+
+static void amdgpu_virt_populate_vf2pf_ucode_info(struct amdgpu_device *adev)
+{
+	struct amd_sriov_msg_vf2pf_info *vf2pf_info;
+	vf2pf_info = (struct amd_sriov_msg_vf2pf_info *) adev->virt.fw_reserve.p_vf2pf;
+
+	if (adev->virt.fw_reserve.p_vf2pf == NULL)
+		return;
+
+	POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_VCE,      adev->vce.fw_version);
+	POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_UVD,      adev->uvd.fw_version);
+	POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_MC,       adev->gmc.fw_version);
+	POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_ME,       adev->gfx.me_fw_version);
+	POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_PFP,      adev->gfx.pfp_fw_version);
+	POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_CE,       adev->gfx.ce_fw_version);
+	POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_RLC,      adev->gfx.rlc_fw_version);
+	POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_RLC_SRLC, adev->gfx.rlc_srlc_fw_version);
+	POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_RLC_SRLG, adev->gfx.rlc_srlg_fw_version);
+	POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_RLC_SRLS, adev->gfx.rlc_srls_fw_version);
+	POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_MEC,      adev->gfx.mec_fw_version);
+	POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_MEC2,     adev->gfx.mec2_fw_version);
+	POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_SOS,      adev->psp.sos.fw_version);
+	POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_ASD,
+			    adev->psp.asd_context.bin_desc.fw_version);
+	POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_TA_RAS,
+			    adev->psp.ras_context.context.bin_desc.fw_version);
+	POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_TA_XGMI,
+			    adev->psp.xgmi_context.context.bin_desc.fw_version);
+	POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_SMC,      adev->pm.fw_version);
+	POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_SDMA,     adev->sdma.instance[0].fw_version);
+	POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_SDMA2,    adev->sdma.instance[1].fw_version);
+	POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_VCN,      adev->vcn.fw_version);
+	POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_DMCU,     adev->dm.dmcu_fw_version);
+}
+
+static int amdgpu_virt_write_vf2pf_data(struct amdgpu_device *adev)
+{
+	struct amd_sriov_msg_vf2pf_info *vf2pf_info;
+
+	vf2pf_info = (struct amd_sriov_msg_vf2pf_info *) adev->virt.fw_reserve.p_vf2pf;
+
+	if (adev->virt.fw_reserve.p_vf2pf == NULL)
+		return -EINVAL;
+
+	memset(vf2pf_info, 0, sizeof(struct amd_sriov_msg_vf2pf_info));
+
+	vf2pf_info->header.size = sizeof(struct amd_sriov_msg_vf2pf_info);
+	vf2pf_info->header.version = AMD_SRIOV_MSG_FW_VRAM_VF2PF_VER;
+
+#ifdef MODULE
+	if (THIS_MODULE->version != NULL)
+		strcpy(vf2pf_info->driver_version, THIS_MODULE->version);
+	else
+#endif
+		strcpy(vf2pf_info->driver_version, "N/A");
+
+	vf2pf_info->pf2vf_version_required = 0; // no requirement, guest understands all
+	vf2pf_info->driver_cert = 0;
+	vf2pf_info->os_info.all = 0;
+
+	vf2pf_info->fb_usage =
+		ttm_resource_manager_usage(&adev->mman.vram_mgr.manager) >> 20;
+	vf2pf_info->fb_vis_usage =
+		amdgpu_vram_mgr_vis_usage(&adev->mman.vram_mgr) >> 20;
+	vf2pf_info->fb_size = adev->gmc.real_vram_size >> 20;
+	vf2pf_info->fb_vis_size = adev->gmc.visible_vram_size >> 20;
+
+	amdgpu_virt_populate_vf2pf_ucode_info(adev);
+
+	/* TODO: read dynamic info */
+	vf2pf_info->gfx_usage = 0;
+	vf2pf_info->compute_usage = 0;
+	vf2pf_info->encode_usage = 0;
+	vf2pf_info->decode_usage = 0;
+
+	vf2pf_info->dummy_page_addr = (uint64_t)adev->dummy_page_addr;
+	if (amdgpu_sriov_is_mes_info_enable(adev)) {
+		vf2pf_info->mes_info_addr =
+			(uint64_t)(adev->mes.resource_1_gpu_addr[0] + AMDGPU_GPU_PAGE_SIZE);
+		vf2pf_info->mes_info_size =
+			adev->mes.resource_1[0]->tbo.base.size - AMDGPU_GPU_PAGE_SIZE;
+	}
+	vf2pf_info->checksum =
+		amd_sriov_msg_checksum(
+		vf2pf_info, sizeof(*vf2pf_info), 0, 0);
+
+	return 0;
+}
+
+static void amdgpu_virt_update_vf2pf_work_item(struct work_struct *work)
+{
+	struct amdgpu_device *adev = container_of(work, struct amdgpu_device, virt.vf2pf_work.work);
+	int ret;
+
+	ret = amdgpu_virt_read_pf2vf_data(adev);
+	if (ret) {
+		adev->virt.vf2pf_update_retry_cnt++;
+
+		if ((amdgpu_virt_rcvd_ras_interrupt(adev) ||
+			adev->virt.vf2pf_update_retry_cnt >= AMDGPU_VF2PF_UPDATE_MAX_RETRY_LIMIT) &&
+			amdgpu_sriov_runtime(adev)) {
+
+			amdgpu_ras_set_fed(adev, true);
+			if (amdgpu_reset_domain_schedule(adev->reset_domain,
+							&adev->kfd.reset_work))
+				return;
+			else
+				dev_err(adev->dev, "Failed to queue work! at %s", __func__);
+		}
+
+		goto out;
+	}
+
+	adev->virt.vf2pf_update_retry_cnt = 0;
+	amdgpu_virt_write_vf2pf_data(adev);
+
+out:
+	schedule_delayed_work(&(adev->virt.vf2pf_work), adev->virt.vf2pf_update_interval_ms);
+}
+
+void amdgpu_virt_fini_data_exchange(struct amdgpu_device *adev)
+{
+	if (adev->virt.vf2pf_update_interval_ms != 0) {
+		DRM_INFO("clean up the vf2pf work item\n");
+		cancel_delayed_work_sync(&adev->virt.vf2pf_work);
+		adev->virt.vf2pf_update_interval_ms = 0;
+	}
+}
+
+void amdgpu_virt_init_data_exchange(struct amdgpu_device *adev)
+{
+	adev->virt.fw_reserve.p_pf2vf = NULL;
+	adev->virt.fw_reserve.p_vf2pf = NULL;
+	adev->virt.vf2pf_update_interval_ms = 0;
+	adev->virt.vf2pf_update_retry_cnt = 0;
+
+	if (adev->mman.fw_vram_usage_va && adev->mman.drv_vram_usage_va) {
+		DRM_WARN("Currently fw_vram and drv_vram should not have values at the same time!");
+	} else if (adev->mman.fw_vram_usage_va || adev->mman.drv_vram_usage_va) {
+		/* go through this logic in ip_init and reset to init workqueue*/
+		amdgpu_virt_exchange_data(adev);
+
+		INIT_DELAYED_WORK(&adev->virt.vf2pf_work, amdgpu_virt_update_vf2pf_work_item);
+		schedule_delayed_work(&(adev->virt.vf2pf_work), msecs_to_jiffies(adev->virt.vf2pf_update_interval_ms));
+	} else if (adev->bios != NULL) {
+		/* got through this logic in early init stage to get necessary flags, e.g. rlcg_acc related*/
+		adev->virt.fw_reserve.p_pf2vf =
+			(struct amd_sriov_msg_pf2vf_info_header *)
+			(adev->bios + (AMD_SRIOV_MSG_PF2VF_OFFSET_KB << 10));
+
+		amdgpu_virt_read_pf2vf_data(adev);
+	}
+}
+
+
+void amdgpu_virt_exchange_data(struct amdgpu_device *adev)
+{
+	uint64_t bp_block_offset = 0;
+	uint32_t bp_block_size = 0;
+	struct amd_sriov_msg_pf2vf_info *pf2vf_v2 = NULL;
+
+	if (adev->mman.fw_vram_usage_va || adev->mman.drv_vram_usage_va) {
+		if (adev->mman.fw_vram_usage_va) {
+			adev->virt.fw_reserve.p_pf2vf =
+				(struct amd_sriov_msg_pf2vf_info_header *)
+				(adev->mman.fw_vram_usage_va + (AMD_SRIOV_MSG_PF2VF_OFFSET_KB << 10));
+			adev->virt.fw_reserve.p_vf2pf =
+				(struct amd_sriov_msg_vf2pf_info_header *)
+				(adev->mman.fw_vram_usage_va + (AMD_SRIOV_MSG_VF2PF_OFFSET_KB << 10));
+			adev->virt.fw_reserve.ras_telemetry =
+				(adev->mman.fw_vram_usage_va + (AMD_SRIOV_MSG_RAS_TELEMETRY_OFFSET_KB << 10));
+		} else if (adev->mman.drv_vram_usage_va) {
+			adev->virt.fw_reserve.p_pf2vf =
+				(struct amd_sriov_msg_pf2vf_info_header *)
+				(adev->mman.drv_vram_usage_va + (AMD_SRIOV_MSG_PF2VF_OFFSET_KB << 10));
+			adev->virt.fw_reserve.p_vf2pf =
+				(struct amd_sriov_msg_vf2pf_info_header *)
+				(adev->mman.drv_vram_usage_va + (AMD_SRIOV_MSG_VF2PF_OFFSET_KB << 10));
+			adev->virt.fw_reserve.ras_telemetry =
+				(adev->mman.drv_vram_usage_va + (AMD_SRIOV_MSG_RAS_TELEMETRY_OFFSET_KB << 10));
+		}
+
+		amdgpu_virt_read_pf2vf_data(adev);
+		amdgpu_virt_write_vf2pf_data(adev);
+
+		/* bad page handling for version 2 */
+		if (adev->virt.fw_reserve.p_pf2vf->version == 2) {
+			pf2vf_v2 = (struct amd_sriov_msg_pf2vf_info *)adev->virt.fw_reserve.p_pf2vf;
+
+			bp_block_offset = ((uint64_t)pf2vf_v2->bp_block_offset_low & 0xFFFFFFFF) |
+				((((uint64_t)pf2vf_v2->bp_block_offset_high) << 32) & 0xFFFFFFFF00000000);
+			bp_block_size = pf2vf_v2->bp_block_size;
+
+			if (bp_block_size && !adev->virt.ras_init_done)
+				amdgpu_virt_init_ras_err_handler_data(adev);
+
+			if (adev->virt.ras_init_done)
+				amdgpu_virt_add_bad_page(adev, bp_block_offset, bp_block_size);
+		}
+	}
+}
+
+static u32 amdgpu_virt_init_detect_asic(struct amdgpu_device *adev)
+{
+	uint32_t reg;
+
+	switch (adev->asic_type) {
+	case CHIP_TONGA:
+	case CHIP_FIJI:
+		reg = RREG32(mmBIF_IOV_FUNC_IDENTIFIER);
+		break;
+	case CHIP_VEGA10:
+	case CHIP_VEGA20:
+	case CHIP_NAVI10:
+	case CHIP_NAVI12:
+	case CHIP_SIENNA_CICHLID:
+	case CHIP_ARCTURUS:
+	case CHIP_ALDEBARAN:
+	case CHIP_IP_DISCOVERY:
+		reg = RREG32(mmRCC_IOV_FUNC_IDENTIFIER);
+		break;
+	default: /* other chip doesn't support SRIOV */
+		reg = 0;
+		break;
+	}
+
+	if (reg & 1)
+		adev->virt.caps |= AMDGPU_SRIOV_CAPS_IS_VF;
+
+	if (reg & 0x80000000)
+		adev->virt.caps |= AMDGPU_SRIOV_CAPS_ENABLE_IOV;
+
+	if (!reg) {
+		/* passthrough mode exclus sriov mod */
+		if (is_virtual_machine() && !xen_initial_domain())
+			adev->virt.caps |= AMDGPU_PASSTHROUGH_MODE;
+	}
+
+	return reg;
+}
+
+static bool amdgpu_virt_init_req_data(struct amdgpu_device *adev, u32 reg)
+{
+	bool is_sriov = false;
+
+	/* we have the ability to check now */
+	if (amdgpu_sriov_vf(adev)) {
+		is_sriov = true;
+
+		switch (adev->asic_type) {
+		case CHIP_TONGA:
+		case CHIP_FIJI:
+			vi_set_virt_ops(adev);
+			break;
+		case CHIP_VEGA10:
+			soc15_set_virt_ops(adev);
+#ifdef CONFIG_X86
+			/* not send GPU_INIT_DATA with MS_HYPERV*/
+			if (!hypervisor_is_type(X86_HYPER_MS_HYPERV))
+#endif
+				/* send a dummy GPU_INIT_DATA request to host on vega10 */
+				amdgpu_virt_request_init_data(adev);
+			break;
+		case CHIP_VEGA20:
+		case CHIP_ARCTURUS:
+		case CHIP_ALDEBARAN:
+			soc15_set_virt_ops(adev);
+			break;
+		case CHIP_NAVI10:
+		case CHIP_NAVI12:
+		case CHIP_SIENNA_CICHLID:
+		case CHIP_IP_DISCOVERY:
+			nv_set_virt_ops(adev);
+			/* try send GPU_INIT_DATA request to host */
+			amdgpu_virt_request_init_data(adev);
+			break;
+		default: /* other chip doesn't support SRIOV */
+			is_sriov = false;
+			DRM_ERROR("Unknown asic type: %d!\n", adev->asic_type);
+			break;
+		}
+	}
+
+	return is_sriov;
+}
+
+static void amdgpu_virt_init_ras(struct amdgpu_device *adev)
+{
+	ratelimit_state_init(&adev->virt.ras.ras_error_cnt_rs, 5 * HZ, 1);
+	ratelimit_state_init(&adev->virt.ras.ras_cper_dump_rs, 5 * HZ, 1);
+	ratelimit_state_init(&adev->virt.ras.ras_chk_criti_rs, 5 * HZ, 1);
+
+	ratelimit_set_flags(&adev->virt.ras.ras_error_cnt_rs,
+			    RATELIMIT_MSG_ON_RELEASE);
+	ratelimit_set_flags(&adev->virt.ras.ras_cper_dump_rs,
+			    RATELIMIT_MSG_ON_RELEASE);
+	ratelimit_set_flags(&adev->virt.ras.ras_chk_criti_rs,
+			    RATELIMIT_MSG_ON_RELEASE);
+
+	mutex_init(&adev->virt.ras.ras_telemetry_mutex);
+
+	adev->virt.ras.cper_rptr = 0;
+}
+
+void amdgpu_virt_init(struct amdgpu_device *adev)
+{
+	bool is_sriov = false;
+	uint32_t reg = amdgpu_virt_init_detect_asic(adev);
+
+	is_sriov = amdgpu_virt_init_req_data(adev, reg);
+
+	if (is_sriov)
+		amdgpu_virt_init_ras(adev);
+}
+
+static bool amdgpu_virt_access_debugfs_is_mmio(struct amdgpu_device *adev)
+{
+	return amdgpu_sriov_is_debug(adev) ? true : false;
+}
+
+static bool amdgpu_virt_access_debugfs_is_kiq(struct amdgpu_device *adev)
+{
+	return amdgpu_sriov_is_normal(adev) ? true : false;
+}
+
+int amdgpu_virt_enable_access_debugfs(struct amdgpu_device *adev)
+{
+	if (!amdgpu_sriov_vf(adev) ||
+	    amdgpu_virt_access_debugfs_is_kiq(adev))
+		return 0;
+
+	if (amdgpu_virt_access_debugfs_is_mmio(adev))
+		adev->virt.caps &= ~AMDGPU_SRIOV_CAPS_RUNTIME;
+	else
+		return -EPERM;
+
+	return 0;
+}
+
+void amdgpu_virt_disable_access_debugfs(struct amdgpu_device *adev)
+{
+	if (amdgpu_sriov_vf(adev))
+		adev->virt.caps |= AMDGPU_SRIOV_CAPS_RUNTIME;
+}
+
+enum amdgpu_sriov_vf_mode amdgpu_virt_get_sriov_vf_mode(struct amdgpu_device *adev)
+{
+	enum amdgpu_sriov_vf_mode mode;
+
+	if (amdgpu_sriov_vf(adev)) {
+		if (amdgpu_sriov_is_pp_one_vf(adev))
+			mode = SRIOV_VF_MODE_ONE_VF;
+		else
+			mode = SRIOV_VF_MODE_MULTI_VF;
+	} else {
+		mode = SRIOV_VF_MODE_BARE_METAL;
+	}
+
+	return mode;
+}
+
+void amdgpu_virt_pre_reset(struct amdgpu_device *adev)
+{
+	/* stop the data exchange thread */
+	amdgpu_virt_fini_data_exchange(adev);
+	amdgpu_dpm_set_mp1_state(adev, PP_MP1_STATE_FLR);
+}
+
+void amdgpu_virt_post_reset(struct amdgpu_device *adev)
+{
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(11, 0, 3)) {
+		/* force set to GFXOFF state after reset,
+		 * to avoid some invalid operation before GC enable
+		 */
+		adev->gfx.is_poweron = false;
+	}
+
+	adev->mes.ring[0].sched.ready = false;
+}
+
+bool amdgpu_virt_fw_load_skip_check(struct amdgpu_device *adev, uint32_t ucode_id)
+{
+	switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) {
+	case IP_VERSION(13, 0, 0):
+		/* no vf autoload, white list */
+		if (ucode_id == AMDGPU_UCODE_ID_VCN1 ||
+		    ucode_id == AMDGPU_UCODE_ID_VCN)
+			return false;
+		else
+			return true;
+	case IP_VERSION(11, 0, 9):
+	case IP_VERSION(11, 0, 7):
+		/* black list for CHIP_NAVI12 and CHIP_SIENNA_CICHLID */
+		if (ucode_id == AMDGPU_UCODE_ID_RLC_G
+		    || ucode_id == AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL
+		    || ucode_id == AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM
+		    || ucode_id == AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM
+		    || ucode_id == AMDGPU_UCODE_ID_SMC)
+			return true;
+		else
+			return false;
+	case IP_VERSION(13, 0, 10):
+		/* white list */
+		if (ucode_id == AMDGPU_UCODE_ID_CAP
+		|| ucode_id == AMDGPU_UCODE_ID_CP_RS64_PFP
+		|| ucode_id == AMDGPU_UCODE_ID_CP_RS64_ME
+		|| ucode_id == AMDGPU_UCODE_ID_CP_RS64_MEC
+		|| ucode_id == AMDGPU_UCODE_ID_CP_RS64_PFP_P0_STACK
+		|| ucode_id == AMDGPU_UCODE_ID_CP_RS64_PFP_P1_STACK
+		|| ucode_id == AMDGPU_UCODE_ID_CP_RS64_ME_P0_STACK
+		|| ucode_id == AMDGPU_UCODE_ID_CP_RS64_ME_P1_STACK
+		|| ucode_id == AMDGPU_UCODE_ID_CP_RS64_MEC_P0_STACK
+		|| ucode_id == AMDGPU_UCODE_ID_CP_RS64_MEC_P1_STACK
+		|| ucode_id == AMDGPU_UCODE_ID_CP_RS64_MEC_P2_STACK
+		|| ucode_id == AMDGPU_UCODE_ID_CP_RS64_MEC_P3_STACK
+		|| ucode_id == AMDGPU_UCODE_ID_CP_MES
+		|| ucode_id == AMDGPU_UCODE_ID_CP_MES_DATA
+		|| ucode_id == AMDGPU_UCODE_ID_CP_MES1
+		|| ucode_id == AMDGPU_UCODE_ID_CP_MES1_DATA
+		|| ucode_id == AMDGPU_UCODE_ID_VCN1
+		|| ucode_id == AMDGPU_UCODE_ID_VCN)
+			return false;
+		else
+			return true;
+	default:
+		/* lagacy black list */
+		if (ucode_id == AMDGPU_UCODE_ID_SDMA0
+		    || ucode_id == AMDGPU_UCODE_ID_SDMA1
+		    || ucode_id == AMDGPU_UCODE_ID_SDMA2
+		    || ucode_id == AMDGPU_UCODE_ID_SDMA3
+		    || ucode_id == AMDGPU_UCODE_ID_SDMA4
+		    || ucode_id == AMDGPU_UCODE_ID_SDMA5
+		    || ucode_id == AMDGPU_UCODE_ID_SDMA6
+		    || ucode_id == AMDGPU_UCODE_ID_SDMA7
+		    || ucode_id == AMDGPU_UCODE_ID_RLC_G
+		    || ucode_id == AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL
+		    || ucode_id == AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM
+		    || ucode_id == AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM
+		    || ucode_id == AMDGPU_UCODE_ID_SMC)
+			return true;
+		else
+			return false;
+	}
+}
+
+void amdgpu_virt_update_sriov_video_codec(struct amdgpu_device *adev,
+			struct amdgpu_video_codec_info *encode, uint32_t encode_array_size,
+			struct amdgpu_video_codec_info *decode, uint32_t decode_array_size)
+{
+	uint32_t i;
+
+	if (!adev->virt.is_mm_bw_enabled)
+		return;
+
+	if (encode) {
+		for (i = 0; i < encode_array_size; i++) {
+			encode[i].max_width = adev->virt.encode_max_dimension_pixels;
+			encode[i].max_pixels_per_frame = adev->virt.encode_max_frame_pixels;
+			if (encode[i].max_width > 0)
+				encode[i].max_height = encode[i].max_pixels_per_frame / encode[i].max_width;
+			else
+				encode[i].max_height = 0;
+		}
+	}
+
+	if (decode) {
+		for (i = 0; i < decode_array_size; i++) {
+			decode[i].max_width = adev->virt.decode_max_dimension_pixels;
+			decode[i].max_pixels_per_frame = adev->virt.decode_max_frame_pixels;
+			if (decode[i].max_width > 0)
+				decode[i].max_height = decode[i].max_pixels_per_frame / decode[i].max_width;
+			else
+				decode[i].max_height = 0;
+		}
+	}
+}
+
+bool amdgpu_virt_get_rlcg_reg_access_flag(struct amdgpu_device *adev,
+						 u32 acc_flags, u32 hwip,
+						 bool write, u32 *rlcg_flag)
+{
+	bool ret = false;
+
+	switch (hwip) {
+	case GC_HWIP:
+		if (amdgpu_sriov_reg_indirect_gc(adev)) {
+			*rlcg_flag =
+				write ? AMDGPU_RLCG_GC_WRITE : AMDGPU_RLCG_GC_READ;
+			ret = true;
+		/* only in new version, AMDGPU_REGS_NO_KIQ and
+		 * AMDGPU_REGS_RLC are enabled simultaneously */
+		} else if ((acc_flags & AMDGPU_REGS_RLC) &&
+				!(acc_flags & AMDGPU_REGS_NO_KIQ) && write) {
+			*rlcg_flag = AMDGPU_RLCG_GC_WRITE_LEGACY;
+			ret = true;
+		}
+		break;
+	case MMHUB_HWIP:
+		if (amdgpu_sriov_reg_indirect_mmhub(adev) &&
+		    (acc_flags & AMDGPU_REGS_RLC) && write) {
+			*rlcg_flag = AMDGPU_RLCG_MMHUB_WRITE;
+			ret = true;
+		}
+		break;
+	default:
+		break;
+	}
+	return ret;
+}
+
+u32 amdgpu_virt_rlcg_reg_rw(struct amdgpu_device *adev, u32 offset, u32 v, u32 flag, u32 xcc_id)
+{
+	struct amdgpu_rlcg_reg_access_ctrl *reg_access_ctrl;
+	uint32_t timeout = 50000;
+	uint32_t i, tmp;
+	uint32_t ret = 0;
+	void *scratch_reg0;
+	void *scratch_reg1;
+	void *scratch_reg2;
+	void *scratch_reg3;
+	void *spare_int;
+	unsigned long flags;
+
+	if (!adev->gfx.rlc.rlcg_reg_access_supported) {
+		dev_err(adev->dev,
+			"indirect registers access through rlcg is not available\n");
+		return 0;
+	}
+
+	if (adev->gfx.xcc_mask && (((1 << xcc_id) & adev->gfx.xcc_mask) == 0)) {
+		dev_err(adev->dev, "invalid xcc\n");
+		return 0;
+	}
+
+	if (amdgpu_device_skip_hw_access(adev))
+		return 0;
+
+	reg_access_ctrl = &adev->gfx.rlc.reg_access_ctrl[xcc_id];
+	scratch_reg0 = (void __iomem *)adev->rmmio + 4 * reg_access_ctrl->scratch_reg0;
+	scratch_reg1 = (void __iomem *)adev->rmmio + 4 * reg_access_ctrl->scratch_reg1;
+	scratch_reg2 = (void __iomem *)adev->rmmio + 4 * reg_access_ctrl->scratch_reg2;
+	scratch_reg3 = (void __iomem *)adev->rmmio + 4 * reg_access_ctrl->scratch_reg3;
+
+	spin_lock_irqsave(&adev->virt.rlcg_reg_lock, flags);
+
+	if (reg_access_ctrl->spare_int)
+		spare_int = (void __iomem *)adev->rmmio + 4 * reg_access_ctrl->spare_int;
+
+	if (offset == reg_access_ctrl->grbm_cntl) {
+		/* if the target reg offset is grbm_cntl, write to scratch_reg2 */
+		writel(v, scratch_reg2);
+		if (flag == AMDGPU_RLCG_GC_WRITE_LEGACY)
+			writel(v, ((void __iomem *)adev->rmmio) + (offset * 4));
+	} else if (offset == reg_access_ctrl->grbm_idx) {
+		/* if the target reg offset is grbm_idx, write to scratch_reg3 */
+		writel(v, scratch_reg3);
+		if (flag == AMDGPU_RLCG_GC_WRITE_LEGACY)
+			writel(v, ((void __iomem *)adev->rmmio) + (offset * 4));
+	} else {
+		/*
+		 * SCRATCH_REG0 	= read/write value
+		 * SCRATCH_REG1[30:28]	= command
+		 * SCRATCH_REG1[19:0]	= address in dword
+		 * SCRATCH_REG1[27:24]	= Error reporting
+		 */
+		writel(v, scratch_reg0);
+		writel((offset | flag), scratch_reg1);
+		if (reg_access_ctrl->spare_int)
+			writel(1, spare_int);
+
+		for (i = 0; i < timeout; i++) {
+			tmp = readl(scratch_reg1);
+			if (!(tmp & AMDGPU_RLCG_SCRATCH1_ADDRESS_MASK))
+				break;
+			udelay(10);
+		}
+
+		tmp = readl(scratch_reg1);
+		if (i >= timeout || (tmp & AMDGPU_RLCG_SCRATCH1_ERROR_MASK) != 0) {
+			if (amdgpu_sriov_rlcg_error_report_enabled(adev)) {
+				if (tmp & AMDGPU_RLCG_VFGATE_DISABLED) {
+					dev_err(adev->dev,
+						"vfgate is disabled, rlcg failed to program reg: 0x%05x\n", offset);
+				} else if (tmp & AMDGPU_RLCG_WRONG_OPERATION_TYPE) {
+					dev_err(adev->dev,
+						"wrong operation type, rlcg failed to program reg: 0x%05x\n", offset);
+				} else if (tmp & AMDGPU_RLCG_REG_NOT_IN_RANGE) {
+					dev_err(adev->dev,
+						"register is not in range, rlcg failed to program reg: 0x%05x\n", offset);
+				} else {
+					dev_err(adev->dev,
+						"unknown error type, rlcg failed to program reg: 0x%05x\n", offset);
+				}
+			} else {
+				dev_err(adev->dev,
+					"timeout: rlcg faled to program reg: 0x%05x\n", offset);
+			}
+		}
+	}
+
+	ret = readl(scratch_reg0);
+
+	spin_unlock_irqrestore(&adev->virt.rlcg_reg_lock, flags);
+
+	return ret;
+}
+
+void amdgpu_sriov_wreg(struct amdgpu_device *adev,
+		       u32 offset, u32 value,
+		       u32 acc_flags, u32 hwip, u32 xcc_id)
+{
+	u32 rlcg_flag;
+
+	if (amdgpu_device_skip_hw_access(adev))
+		return;
+
+	if (!amdgpu_sriov_runtime(adev) &&
+		amdgpu_virt_get_rlcg_reg_access_flag(adev, acc_flags, hwip, true, &rlcg_flag)) {
+		amdgpu_virt_rlcg_reg_rw(adev, offset, value, rlcg_flag, xcc_id);
+		return;
+	}
+
+	if (acc_flags & AMDGPU_REGS_NO_KIQ)
+		WREG32_NO_KIQ(offset, value);
+	else
+		WREG32(offset, value);
+}
+
+u32 amdgpu_sriov_rreg(struct amdgpu_device *adev,
+		      u32 offset, u32 acc_flags, u32 hwip, u32 xcc_id)
+{
+	u32 rlcg_flag;
+
+	if (amdgpu_device_skip_hw_access(adev))
+		return 0;
+
+	if (!amdgpu_sriov_runtime(adev) &&
+		amdgpu_virt_get_rlcg_reg_access_flag(adev, acc_flags, hwip, false, &rlcg_flag))
+		return amdgpu_virt_rlcg_reg_rw(adev, offset, 0, rlcg_flag, xcc_id);
+
+	if (acc_flags & AMDGPU_REGS_NO_KIQ)
+		return RREG32_NO_KIQ(offset);
+	else
+		return RREG32(offset);
+}
+
+bool amdgpu_sriov_xnack_support(struct amdgpu_device *adev)
+{
+	bool xnack_mode = true;
+
+	if (amdgpu_sriov_vf(adev) &&
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2))
+		xnack_mode = false;
+
+	return xnack_mode;
+}
+
+bool amdgpu_virt_get_ras_capability(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+
+	if (!amdgpu_sriov_ras_caps_en(adev))
+		return false;
+
+	if (adev->virt.ras_en_caps.bits.block_umc)
+		adev->ras_hw_enabled |= BIT(AMDGPU_RAS_BLOCK__UMC);
+	if (adev->virt.ras_en_caps.bits.block_sdma)
+		adev->ras_hw_enabled |= BIT(AMDGPU_RAS_BLOCK__SDMA);
+	if (adev->virt.ras_en_caps.bits.block_gfx)
+		adev->ras_hw_enabled |= BIT(AMDGPU_RAS_BLOCK__GFX);
+	if (adev->virt.ras_en_caps.bits.block_mmhub)
+		adev->ras_hw_enabled |= BIT(AMDGPU_RAS_BLOCK__MMHUB);
+	if (adev->virt.ras_en_caps.bits.block_athub)
+		adev->ras_hw_enabled |= BIT(AMDGPU_RAS_BLOCK__ATHUB);
+	if (adev->virt.ras_en_caps.bits.block_pcie_bif)
+		adev->ras_hw_enabled |= BIT(AMDGPU_RAS_BLOCK__PCIE_BIF);
+	if (adev->virt.ras_en_caps.bits.block_hdp)
+		adev->ras_hw_enabled |= BIT(AMDGPU_RAS_BLOCK__HDP);
+	if (adev->virt.ras_en_caps.bits.block_xgmi_wafl)
+		adev->ras_hw_enabled |= BIT(AMDGPU_RAS_BLOCK__XGMI_WAFL);
+	if (adev->virt.ras_en_caps.bits.block_df)
+		adev->ras_hw_enabled |= BIT(AMDGPU_RAS_BLOCK__DF);
+	if (adev->virt.ras_en_caps.bits.block_smn)
+		adev->ras_hw_enabled |= BIT(AMDGPU_RAS_BLOCK__SMN);
+	if (adev->virt.ras_en_caps.bits.block_sem)
+		adev->ras_hw_enabled |= BIT(AMDGPU_RAS_BLOCK__SEM);
+	if (adev->virt.ras_en_caps.bits.block_mp0)
+		adev->ras_hw_enabled |= BIT(AMDGPU_RAS_BLOCK__MP0);
+	if (adev->virt.ras_en_caps.bits.block_mp1)
+		adev->ras_hw_enabled |= BIT(AMDGPU_RAS_BLOCK__MP1);
+	if (adev->virt.ras_en_caps.bits.block_fuse)
+		adev->ras_hw_enabled |= BIT(AMDGPU_RAS_BLOCK__FUSE);
+	if (adev->virt.ras_en_caps.bits.block_mca)
+		adev->ras_hw_enabled |= BIT(AMDGPU_RAS_BLOCK__MCA);
+	if (adev->virt.ras_en_caps.bits.block_vcn)
+		adev->ras_hw_enabled |= BIT(AMDGPU_RAS_BLOCK__VCN);
+	if (adev->virt.ras_en_caps.bits.block_jpeg)
+		adev->ras_hw_enabled |= BIT(AMDGPU_RAS_BLOCK__JPEG);
+	if (adev->virt.ras_en_caps.bits.block_ih)
+		adev->ras_hw_enabled |= BIT(AMDGPU_RAS_BLOCK__IH);
+	if (adev->virt.ras_en_caps.bits.block_mpio)
+		adev->ras_hw_enabled |= BIT(AMDGPU_RAS_BLOCK__MPIO);
+
+	if (adev->virt.ras_en_caps.bits.poison_propogation_mode)
+		con->poison_supported = true; /* Poison is handled by host */
+
+	return true;
+}
+
+static inline enum amd_sriov_ras_telemetry_gpu_block
+amdgpu_ras_block_to_sriov(struct amdgpu_device *adev, enum amdgpu_ras_block block) {
+	switch (block) {
+	case AMDGPU_RAS_BLOCK__UMC:
+		return RAS_TELEMETRY_GPU_BLOCK_UMC;
+	case AMDGPU_RAS_BLOCK__SDMA:
+		return RAS_TELEMETRY_GPU_BLOCK_SDMA;
+	case AMDGPU_RAS_BLOCK__GFX:
+		return RAS_TELEMETRY_GPU_BLOCK_GFX;
+	case AMDGPU_RAS_BLOCK__MMHUB:
+		return RAS_TELEMETRY_GPU_BLOCK_MMHUB;
+	case AMDGPU_RAS_BLOCK__ATHUB:
+		return RAS_TELEMETRY_GPU_BLOCK_ATHUB;
+	case AMDGPU_RAS_BLOCK__PCIE_BIF:
+		return RAS_TELEMETRY_GPU_BLOCK_PCIE_BIF;
+	case AMDGPU_RAS_BLOCK__HDP:
+		return RAS_TELEMETRY_GPU_BLOCK_HDP;
+	case AMDGPU_RAS_BLOCK__XGMI_WAFL:
+		return RAS_TELEMETRY_GPU_BLOCK_XGMI_WAFL;
+	case AMDGPU_RAS_BLOCK__DF:
+		return RAS_TELEMETRY_GPU_BLOCK_DF;
+	case AMDGPU_RAS_BLOCK__SMN:
+		return RAS_TELEMETRY_GPU_BLOCK_SMN;
+	case AMDGPU_RAS_BLOCK__SEM:
+		return RAS_TELEMETRY_GPU_BLOCK_SEM;
+	case AMDGPU_RAS_BLOCK__MP0:
+		return RAS_TELEMETRY_GPU_BLOCK_MP0;
+	case AMDGPU_RAS_BLOCK__MP1:
+		return RAS_TELEMETRY_GPU_BLOCK_MP1;
+	case AMDGPU_RAS_BLOCK__FUSE:
+		return RAS_TELEMETRY_GPU_BLOCK_FUSE;
+	case AMDGPU_RAS_BLOCK__MCA:
+		return RAS_TELEMETRY_GPU_BLOCK_MCA;
+	case AMDGPU_RAS_BLOCK__VCN:
+		return RAS_TELEMETRY_GPU_BLOCK_VCN;
+	case AMDGPU_RAS_BLOCK__JPEG:
+		return RAS_TELEMETRY_GPU_BLOCK_JPEG;
+	case AMDGPU_RAS_BLOCK__IH:
+		return RAS_TELEMETRY_GPU_BLOCK_IH;
+	case AMDGPU_RAS_BLOCK__MPIO:
+		return RAS_TELEMETRY_GPU_BLOCK_MPIO;
+	default:
+		DRM_WARN_ONCE("Unsupported SRIOV RAS telemetry block 0x%x\n",
+			      block);
+		return RAS_TELEMETRY_GPU_BLOCK_COUNT;
+	}
+}
+
+static int amdgpu_virt_cache_host_error_counts(struct amdgpu_device *adev,
+					       struct amdsriov_ras_telemetry *host_telemetry)
+{
+	struct amd_sriov_ras_telemetry_error_count *tmp = NULL;
+	uint32_t checksum, used_size;
+
+	checksum = host_telemetry->header.checksum;
+	used_size = host_telemetry->header.used_size;
+
+	if (used_size > (AMD_SRIOV_RAS_TELEMETRY_SIZE_KB << 10))
+		return 0;
+
+	tmp = kmemdup(&host_telemetry->body.error_count, used_size, GFP_KERNEL);
+	if (!tmp)
+		return -ENOMEM;
+
+	if (checksum != amd_sriov_msg_checksum(tmp, used_size, 0, 0))
+		goto out;
+
+	memcpy(&adev->virt.count_cache, tmp,
+	       min(used_size, sizeof(adev->virt.count_cache)));
+out:
+	kfree(tmp);
+
+	return 0;
+}
+
+static int amdgpu_virt_req_ras_err_count_internal(struct amdgpu_device *adev, bool force_update)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+
+	if (!virt->ops || !virt->ops->req_ras_err_count)
+		return -EOPNOTSUPP;
+
+	/* Host allows 15 ras telemetry requests per 60 seconds. Afterwhich, the Host
+	 * will ignore incoming guest messages. Ratelimit the guest messages to
+	 * prevent guest self DOS.
+	 */
+	if (__ratelimit(&virt->ras.ras_error_cnt_rs) || force_update) {
+		mutex_lock(&virt->ras.ras_telemetry_mutex);
+		if (!virt->ops->req_ras_err_count(adev))
+			amdgpu_virt_cache_host_error_counts(adev,
+				virt->fw_reserve.ras_telemetry);
+		mutex_unlock(&virt->ras.ras_telemetry_mutex);
+	}
+
+	return 0;
+}
+
+/* Bypass ACA interface and query ECC counts directly from host */
+int amdgpu_virt_req_ras_err_count(struct amdgpu_device *adev, enum amdgpu_ras_block block,
+				  struct ras_err_data *err_data)
+{
+	enum amd_sriov_ras_telemetry_gpu_block sriov_block;
+
+	sriov_block = amdgpu_ras_block_to_sriov(adev, block);
+
+	if (sriov_block >= RAS_TELEMETRY_GPU_BLOCK_COUNT ||
+	    !amdgpu_sriov_ras_telemetry_block_en(adev, sriov_block))
+		return -EOPNOTSUPP;
+
+	/* Host Access may be lost during reset, just return last cached data. */
+	if (down_read_trylock(&adev->reset_domain->sem)) {
+		amdgpu_virt_req_ras_err_count_internal(adev, false);
+		up_read(&adev->reset_domain->sem);
+	}
+
+	err_data->ue_count = adev->virt.count_cache.block[sriov_block].ue_count;
+	err_data->ce_count = adev->virt.count_cache.block[sriov_block].ce_count;
+	err_data->de_count = adev->virt.count_cache.block[sriov_block].de_count;
+
+	return 0;
+}
+
+static int
+amdgpu_virt_write_cpers_to_ring(struct amdgpu_device *adev,
+				struct amdsriov_ras_telemetry *host_telemetry,
+				u32 *more)
+{
+	struct amd_sriov_ras_cper_dump *cper_dump = NULL;
+	struct cper_hdr *entry = NULL;
+	struct amdgpu_ring *ring = &adev->cper.ring_buf;
+	uint32_t checksum, used_size, i;
+	int ret = 0;
+
+	checksum = host_telemetry->header.checksum;
+	used_size = host_telemetry->header.used_size;
+
+	if (used_size > (AMD_SRIOV_RAS_TELEMETRY_SIZE_KB << 10))
+		return -EINVAL;
+
+	cper_dump = kmemdup(&host_telemetry->body.cper_dump, used_size, GFP_KERNEL);
+	if (!cper_dump)
+		return -ENOMEM;
+
+	if (checksum != amd_sriov_msg_checksum(cper_dump, used_size, 0, 0)) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	*more = cper_dump->more;
+
+	if (cper_dump->wptr < adev->virt.ras.cper_rptr) {
+		dev_warn(
+			adev->dev,
+			"guest specified rptr that was too high! guest rptr: 0x%llx, host rptr: 0x%llx\n",
+			adev->virt.ras.cper_rptr, cper_dump->wptr);
+
+		adev->virt.ras.cper_rptr = cper_dump->wptr;
+		goto out;
+	}
+
+	entry = (struct cper_hdr *)&cper_dump->buf[0];
+
+	for (i = 0; i < cper_dump->count; i++) {
+		amdgpu_cper_ring_write(ring, entry, entry->record_length);
+		entry = (struct cper_hdr *)((char *)entry +
+					    entry->record_length);
+	}
+
+	if (cper_dump->overflow_count)
+		dev_warn(adev->dev,
+			 "host reported CPER overflow of 0x%llx entries!\n",
+			 cper_dump->overflow_count);
+
+	adev->virt.ras.cper_rptr = cper_dump->wptr;
+out:
+	kfree(cper_dump);
+
+	return ret;
+}
+
+static int amdgpu_virt_req_ras_cper_dump_internal(struct amdgpu_device *adev)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+	int ret = 0;
+	uint32_t more = 0;
+
+	if (!virt->ops || !virt->ops->req_ras_cper_dump)
+		return -EOPNOTSUPP;
+
+	do {
+		if (!virt->ops->req_ras_cper_dump(adev, virt->ras.cper_rptr))
+			ret = amdgpu_virt_write_cpers_to_ring(
+				adev, virt->fw_reserve.ras_telemetry, &more);
+		else
+			ret = 0;
+	} while (more && !ret);
+
+	return ret;
+}
+
+int amdgpu_virt_req_ras_cper_dump(struct amdgpu_device *adev, bool force_update)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+	int ret = 0;
+
+	if (!amdgpu_sriov_ras_cper_en(adev))
+		return -EOPNOTSUPP;
+
+	if ((__ratelimit(&virt->ras.ras_cper_dump_rs) || force_update) &&
+	    down_read_trylock(&adev->reset_domain->sem)) {
+		mutex_lock(&virt->ras.ras_telemetry_mutex);
+		ret = amdgpu_virt_req_ras_cper_dump_internal(adev);
+		mutex_unlock(&virt->ras.ras_telemetry_mutex);
+		up_read(&adev->reset_domain->sem);
+	}
+
+	return ret;
+}
+
+int amdgpu_virt_ras_telemetry_post_reset(struct amdgpu_device *adev)
+{
+	unsigned long ue_count, ce_count;
+
+	if (amdgpu_sriov_ras_telemetry_en(adev)) {
+		amdgpu_virt_req_ras_err_count_internal(adev, true);
+		amdgpu_ras_query_error_count(adev, &ce_count, &ue_count, NULL);
+	}
+
+	return 0;
+}
+
+bool amdgpu_virt_ras_telemetry_block_en(struct amdgpu_device *adev,
+					enum amdgpu_ras_block block)
+{
+	enum amd_sriov_ras_telemetry_gpu_block sriov_block;
+
+	sriov_block = amdgpu_ras_block_to_sriov(adev, block);
+
+	if (sriov_block >= RAS_TELEMETRY_GPU_BLOCK_COUNT ||
+	    !amdgpu_sriov_ras_telemetry_block_en(adev, sriov_block))
+		return false;
+
+	return true;
+}
+
+/*
+ * amdgpu_virt_request_bad_pages() - request bad pages
+ * @adev: amdgpu device.
+ * Send command to GPU hypervisor to write new bad pages into the shared PF2VF region
+ */
+void amdgpu_virt_request_bad_pages(struct amdgpu_device *adev)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+
+	if (virt->ops && virt->ops->req_bad_pages)
+		virt->ops->req_bad_pages(adev);
+}
+
+static int amdgpu_virt_cache_chk_criti_hit(struct amdgpu_device *adev,
+					   struct amdsriov_ras_telemetry *host_telemetry,
+					   bool *hit)
+{
+	struct amd_sriov_ras_chk_criti *tmp = NULL;
+	uint32_t checksum, used_size;
+
+	checksum = host_telemetry->header.checksum;
+	used_size = host_telemetry->header.used_size;
+
+	if (used_size > (AMD_SRIOV_RAS_TELEMETRY_SIZE_KB << 10))
+		return 0;
+
+	tmp = kmemdup(&host_telemetry->body.chk_criti, used_size, GFP_KERNEL);
+	if (!tmp)
+		return -ENOMEM;
+
+	if (checksum != amd_sriov_msg_checksum(tmp, used_size, 0, 0))
+		goto out;
+
+	if (hit)
+		*hit = tmp->hit ? true : false;
+
+out:
+	kfree(tmp);
+
+	return 0;
+}
+
+int amdgpu_virt_check_vf_critical_region(struct amdgpu_device *adev, u64 addr, bool *hit)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+	int r = -EPERM;
+
+	if (!virt->ops || !virt->ops->req_ras_chk_criti)
+		return -EOPNOTSUPP;
+
+	/* Host allows 15 ras telemetry requests per 60 seconds. Afterwhich, the Host
+	 * will ignore incoming guest messages. Ratelimit the guest messages to
+	 * prevent guest self DOS.
+	 */
+	if (__ratelimit(&virt->ras.ras_chk_criti_rs)) {
+		mutex_lock(&virt->ras.ras_telemetry_mutex);
+		if (!virt->ops->req_ras_chk_criti(adev, addr))
+			r = amdgpu_virt_cache_chk_criti_hit(
+				adev, virt->fw_reserve.ras_telemetry, hit);
+		mutex_unlock(&virt->ras.ras_telemetry_mutex);
+	}
+
+	return r;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
index afcfb8bcfb65..d1172c8e58c4 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
@@ -24,11 +24,41 @@
 #ifndef AMDGPU_VIRT_H
 #define AMDGPU_VIRT_H
 
+#include "amdgv_sriovmsg.h"
+
 #define AMDGPU_SRIOV_CAPS_SRIOV_VBIOS  (1 << 0) /* vBIOS is sr-iov ready */
 #define AMDGPU_SRIOV_CAPS_ENABLE_IOV   (1 << 1) /* sr-iov is enabled on this GPU */
 #define AMDGPU_SRIOV_CAPS_IS_VF        (1 << 2) /* this GPU is a virtual function */
 #define AMDGPU_PASSTHROUGH_MODE        (1 << 3) /* thw whole GPU is pass through for VM */
 #define AMDGPU_SRIOV_CAPS_RUNTIME      (1 << 4) /* is out of full access mode */
+#define AMDGPU_VF_MMIO_ACCESS_PROTECT  (1 << 5) /* MMIO write access is not allowed in sriov runtime */
+
+/* flags for indirect register access path supported by rlcg for sriov */
+#define AMDGPU_RLCG_GC_WRITE_LEGACY    (0x8 << 28)
+#define AMDGPU_RLCG_GC_WRITE           (0x0 << 28)
+#define AMDGPU_RLCG_GC_READ            (0x1 << 28)
+#define AMDGPU_RLCG_MMHUB_WRITE        (0x2 << 28)
+
+/* error code for indirect register access path supported by rlcg for sriov */
+#define AMDGPU_RLCG_VFGATE_DISABLED		0x4000000
+#define AMDGPU_RLCG_WRONG_OPERATION_TYPE	0x2000000
+#define AMDGPU_RLCG_REG_NOT_IN_RANGE		0x1000000
+
+#define AMDGPU_RLCG_SCRATCH1_ADDRESS_MASK	0xFFFFF
+#define AMDGPU_RLCG_SCRATCH1_ERROR_MASK	0xF000000
+
+/* all asic after AI use this offset */
+#define mmRCC_IOV_FUNC_IDENTIFIER 0xDE5
+/* tonga/fiji use this offset */
+#define mmBIF_IOV_FUNC_IDENTIFIER 0x1503
+
+#define AMDGPU_VF2PF_UPDATE_MAX_RETRY_LIMIT 2
+
+enum amdgpu_sriov_vf_mode {
+	SRIOV_VF_MODE_BARE_METAL = 0,
+	SRIOV_VF_MODE_ONE_VF,
+	SRIOV_VF_MODE_MULTI_VF,
+};
 
 struct amdgpu_mm_table {
 	struct amdgpu_bo	*bo;
@@ -36,33 +66,257 @@ struct amdgpu_mm_table {
 	uint64_t		gpu_addr;
 };
 
+#define AMDGPU_VF_ERROR_ENTRY_SIZE    16
+
+/* struct error_entry - amdgpu VF error information. */
+struct amdgpu_vf_error_buffer {
+	struct mutex lock;
+	int read_count;
+	int write_count;
+	uint16_t code[AMDGPU_VF_ERROR_ENTRY_SIZE];
+	uint16_t flags[AMDGPU_VF_ERROR_ENTRY_SIZE];
+	uint64_t data[AMDGPU_VF_ERROR_ENTRY_SIZE];
+};
+
+enum idh_request;
+
 /**
  * struct amdgpu_virt_ops - amdgpu device virt operations
  */
 struct amdgpu_virt_ops {
 	int (*req_full_gpu)(struct amdgpu_device *adev, bool init);
 	int (*rel_full_gpu)(struct amdgpu_device *adev, bool init);
+	int (*req_init_data)(struct amdgpu_device *adev);
 	int (*reset_gpu)(struct amdgpu_device *adev);
-	void (*trans_msg)(struct amdgpu_device *adev, u32 req, u32 data1, u32 data2, u32 data3);
+	void (*ready_to_reset)(struct amdgpu_device *adev);
+	int (*wait_reset)(struct amdgpu_device *adev);
+	void (*trans_msg)(struct amdgpu_device *adev, enum idh_request req,
+			  u32 data1, u32 data2, u32 data3);
+	void (*ras_poison_handler)(struct amdgpu_device *adev,
+					enum amdgpu_ras_block block);
+	bool (*rcvd_ras_intr)(struct amdgpu_device *adev);
+	int (*req_ras_err_count)(struct amdgpu_device *adev);
+	int (*req_ras_cper_dump)(struct amdgpu_device *adev, u64 vf_rptr);
+	int (*req_bad_pages)(struct amdgpu_device *adev);
+	int (*req_ras_chk_criti)(struct amdgpu_device *adev, u64 addr);
+};
+
+/*
+ * Firmware Reserve Frame buffer
+ */
+struct amdgpu_virt_fw_reserve {
+	struct amd_sriov_msg_pf2vf_info_header *p_pf2vf;
+	struct amd_sriov_msg_vf2pf_info_header *p_vf2pf;
+	void *ras_telemetry;
+	unsigned int checksum_key;
+};
+
+/*
+ * Legacy GIM header
+ *
+ * Defination between PF and VF
+ * Structures forcibly aligned to 4 to keep the same style as PF.
+ */
+#define AMDGIM_DATAEXCHANGE_OFFSET		(64 * 1024)
+
+#define AMDGIM_GET_STRUCTURE_RESERVED_SIZE(total, u8, u16, u32, u64) \
+		(total - (((u8)+3) / 4 + ((u16)+1) / 2 + (u32) + (u64)*2))
+
+enum AMDGIM_FEATURE_FLAG {
+	/* GIM supports feature of Error log collecting */
+	AMDGIM_FEATURE_ERROR_LOG_COLLECT = 0x1,
+	/* GIM supports feature of loading uCodes */
+	AMDGIM_FEATURE_GIM_LOAD_UCODES   = 0x2,
+	/* VRAM LOST by GIM */
+	AMDGIM_FEATURE_GIM_FLR_VRAMLOST = 0x4,
+	/* MM bandwidth */
+	AMDGIM_FEATURE_GIM_MM_BW_MGR = 0x8,
+	/* PP ONE VF MODE in GIM */
+	AMDGIM_FEATURE_PP_ONE_VF = (1 << 4),
+	/* Indirect Reg Access enabled */
+	AMDGIM_FEATURE_INDIRECT_REG_ACCESS = (1 << 5),
+	/* AV1 Support MODE*/
+	AMDGIM_FEATURE_AV1_SUPPORT = (1 << 6),
+	/* VCN RB decouple */
+	AMDGIM_FEATURE_VCN_RB_DECOUPLE = (1 << 7),
+	/* MES info */
+	AMDGIM_FEATURE_MES_INFO_ENABLE = (1 << 8),
+	AMDGIM_FEATURE_RAS_CAPS = (1 << 9),
+	AMDGIM_FEATURE_RAS_TELEMETRY = (1 << 10),
+	AMDGIM_FEATURE_RAS_CPER = (1 << 11),
+};
+
+enum AMDGIM_REG_ACCESS_FLAG {
+	/* Use PSP to program IH_RB_CNTL */
+	AMDGIM_FEATURE_IH_REG_PSP_EN      = (1 << 0),
+	/* Use RLC to program MMHUB regs */
+	AMDGIM_FEATURE_MMHUB_REG_RLC_EN   = (1 << 1),
+	/* Use RLC to program GC regs */
+	AMDGIM_FEATURE_GC_REG_RLC_EN      = (1 << 2),
+	/* Use PSP to program L1_TLB_CNTL */
+	AMDGIM_FEATURE_L1_TLB_CNTL_PSP_EN = (1 << 3),
+	/* Use RLCG to program SQ_CONFIG1 */
+	AMDGIM_FEATURE_REG_ACCESS_SQ_CONFIG = (1 << 4),
+};
+
+struct amdgim_pf2vf_info_v1 {
+	/* header contains size and version */
+	struct amd_sriov_msg_pf2vf_info_header header;
+	/* max_width * max_height */
+	unsigned int uvd_enc_max_pixels_count;
+	/* 16x16 pixels/sec, codec independent */
+	unsigned int uvd_enc_max_bandwidth;
+	/* max_width * max_height */
+	unsigned int vce_enc_max_pixels_count;
+	/* 16x16 pixels/sec, codec independent */
+	unsigned int vce_enc_max_bandwidth;
+	/* MEC FW position in kb from the start of visible frame buffer */
+	unsigned int mecfw_kboffset;
+	/* The features flags of the GIM driver supports. */
+	unsigned int feature_flags;
+	/* use private key from mailbox 2 to create chueksum */
+	unsigned int checksum;
+} __aligned(4);
+
+struct amdgim_vf2pf_info_v1 {
+	/* header contains size and version */
+	struct amd_sriov_msg_vf2pf_info_header header;
+	/* driver version */
+	char driver_version[64];
+	/* driver certification, 1=WHQL, 0=None */
+	unsigned int driver_cert;
+	/* guest OS type and version: need a define */
+	unsigned int os_info;
+	/* in the unit of 1M */
+	unsigned int fb_usage;
+	/* guest gfx engine usage percentage */
+	unsigned int gfx_usage;
+	/* guest gfx engine health percentage */
+	unsigned int gfx_health;
+	/* guest compute engine usage percentage */
+	unsigned int compute_usage;
+	/* guest compute engine health percentage */
+	unsigned int compute_health;
+	/* guest vce engine usage percentage. 0xffff means N/A. */
+	unsigned int vce_enc_usage;
+	/* guest vce engine health percentage. 0xffff means N/A. */
+	unsigned int vce_enc_health;
+	/* guest uvd engine usage percentage. 0xffff means N/A. */
+	unsigned int uvd_enc_usage;
+	/* guest uvd engine usage percentage. 0xffff means N/A. */
+	unsigned int uvd_enc_health;
+	unsigned int checksum;
+} __aligned(4);
+
+struct amdgim_vf2pf_info_v2 {
+	/* header contains size and version */
+	struct amd_sriov_msg_vf2pf_info_header header;
+	uint32_t checksum;
+	/* driver version */
+	uint8_t driver_version[64];
+	/* driver certification, 1=WHQL, 0=None */
+	uint32_t driver_cert;
+	/* guest OS type and version: need a define */
+	uint32_t os_info;
+	/* in the unit of 1M */
+	uint32_t fb_usage;
+	/* guest gfx engine usage percentage */
+	uint32_t gfx_usage;
+	/* guest gfx engine health percentage */
+	uint32_t gfx_health;
+	/* guest compute engine usage percentage */
+	uint32_t compute_usage;
+	/* guest compute engine health percentage */
+	uint32_t compute_health;
+	/* guest vce engine usage percentage. 0xffff means N/A. */
+	uint32_t vce_enc_usage;
+	/* guest vce engine health percentage. 0xffff means N/A. */
+	uint32_t vce_enc_health;
+	/* guest uvd engine usage percentage. 0xffff means N/A. */
+	uint32_t uvd_enc_usage;
+	/* guest uvd engine usage percentage. 0xffff means N/A. */
+	uint32_t uvd_enc_health;
+	uint32_t reserved[AMDGIM_GET_STRUCTURE_RESERVED_SIZE(256, 64, 0, (12 + sizeof(struct amd_sriov_msg_vf2pf_info_header)/sizeof(uint32_t)), 0)];
+} __aligned(4);
+
+struct amdgpu_virt_ras_err_handler_data {
+	/* point to bad page records array */
+	struct eeprom_table_record *bps;
+	/* point to reserved bo array */
+	struct amdgpu_bo **bps_bo;
+	/* the count of entries */
+	int count;
+	/* last reserved entry's index + 1 */
+	int last_reserved;
 };
 
+struct amdgpu_virt_ras {
+	struct ratelimit_state ras_error_cnt_rs;
+	struct ratelimit_state ras_cper_dump_rs;
+	struct ratelimit_state ras_chk_criti_rs;
+	struct mutex ras_telemetry_mutex;
+	uint64_t cper_rptr;
+};
+
+#define AMDGPU_VIRT_CAPS_LIST(X) X(AMDGPU_VIRT_CAP_POWER_LIMIT)
+
+DECLARE_ATTR_CAP_CLASS(amdgpu_virt, AMDGPU_VIRT_CAPS_LIST);
+
 /* GPU virtualization */
 struct amdgpu_virt {
 	uint32_t			caps;
 	struct amdgpu_bo		*csa_obj;
-	uint64_t			csa_vmid0_addr;
+	void				*csa_cpu_addr;
 	bool chained_ib_support;
 	uint32_t			reg_val_offs;
-	struct mutex                    lock_reset;
 	struct amdgpu_irq_src		ack_irq;
 	struct amdgpu_irq_src		rcv_irq;
+
 	struct work_struct		flr_work;
+	struct work_struct		req_bad_pages_work;
+	struct work_struct		handle_bad_pages_work;
+
 	struct amdgpu_mm_table		mm_table;
 	const struct amdgpu_virt_ops	*ops;
+	struct amdgpu_vf_error_buffer	vf_errors;
+	struct amdgpu_virt_fw_reserve	fw_reserve;
+	struct amdgpu_virt_caps virt_caps;
+	uint32_t gim_feature;
+	uint32_t reg_access_mode;
+	int req_init_data_ver;
+	bool tdr_debug;
+	struct amdgpu_virt_ras_err_handler_data *virt_eh_data;
+	bool ras_init_done;
+	uint32_t reg_access;
+
+	/* vf2pf message */
+	struct delayed_work vf2pf_work;
+	uint32_t vf2pf_update_interval_ms;
+	int vf2pf_update_retry_cnt;
+
+	/* multimedia bandwidth config */
+	bool     is_mm_bw_enabled;
+	uint32_t decode_max_dimension_pixels;
+	uint32_t decode_max_frame_pixels;
+	uint32_t encode_max_dimension_pixels;
+	uint32_t encode_max_frame_pixels;
+
+	/* the ucode id to signal the autoload */
+	uint32_t autoload_ucode_id;
+
+	/* Spinlock to protect access to the RLCG register interface */
+	spinlock_t rlcg_reg_lock;
+
+	union amd_sriov_ras_caps ras_en_caps;
+	union amd_sriov_ras_caps ras_telemetry_en_caps;
+	struct amdgpu_virt_ras ras;
+	struct amd_sriov_ras_telemetry_error_count count_cache;
+
+	/* hibernate and resume with different VF feature for xgmi enabled system */
+	bool is_xgmi_node_migrate_enabled;
 };
 
-#define AMDGPU_CSA_SIZE    (8 * 1024)
-#define AMDGPU_CSA_VADDR   (AMDGPU_VA_RESERVED_SIZE - AMDGPU_CSA_SIZE)
+struct amdgpu_video_codec_info;
 
 #define amdgpu_sriov_enabled(adev) \
 ((adev)->virt.caps & AMDGPU_SRIOV_CAPS_ENABLE_IOV)
@@ -76,30 +330,130 @@ struct amdgpu_virt {
 #define amdgpu_sriov_runtime(adev) \
 ((adev)->virt.caps & AMDGPU_SRIOV_CAPS_RUNTIME)
 
+#define amdgpu_sriov_fullaccess(adev) \
+(amdgpu_sriov_vf((adev)) && !amdgpu_sriov_runtime((adev)))
+
+#define amdgpu_sriov_reg_indirect_en(adev) \
+(amdgpu_sriov_vf((adev)) && \
+	((adev)->virt.gim_feature & (AMDGIM_FEATURE_INDIRECT_REG_ACCESS)))
+
+#define amdgpu_sriov_reg_indirect_ih(adev) \
+(amdgpu_sriov_vf((adev)) && \
+	((adev)->virt.reg_access & (AMDGIM_FEATURE_IH_REG_PSP_EN)))
+
+#define amdgpu_sriov_reg_indirect_mmhub(adev) \
+(amdgpu_sriov_vf((adev)) && \
+	((adev)->virt.reg_access & (AMDGIM_FEATURE_MMHUB_REG_RLC_EN)))
+
+#define amdgpu_sriov_reg_indirect_gc(adev) \
+(amdgpu_sriov_vf((adev)) && \
+	((adev)->virt.reg_access & (AMDGIM_FEATURE_GC_REG_RLC_EN)))
+
+#define amdgpu_sriov_reg_indirect_l1_tlb_cntl(adev) \
+(amdgpu_sriov_vf((adev)) && \
+	((adev)->virt.reg_access & (AMDGIM_FEATURE_L1_TLB_CNTL_PSP_EN)))
+
+#define amdgpu_sriov_rlcg_error_report_enabled(adev) \
+        (amdgpu_sriov_reg_indirect_mmhub(adev) || amdgpu_sriov_reg_indirect_gc(adev))
+
+#define amdgpu_sriov_reg_access_sq_config(adev) \
+(amdgpu_sriov_vf((adev)) && \
+	((adev)->virt.reg_access & (AMDGIM_FEATURE_REG_ACCESS_SQ_CONFIG)))
+
 #define amdgpu_passthrough(adev) \
 ((adev)->virt.caps & AMDGPU_PASSTHROUGH_MODE)
 
+#define amdgpu_sriov_vf_mmio_access_protection(adev) \
+((adev)->virt.caps & AMDGPU_VF_MMIO_ACCESS_PROTECT)
+
+#define amdgpu_sriov_ras_caps_en(adev) \
+((adev)->virt.gim_feature & AMDGIM_FEATURE_RAS_CAPS)
+
+#define amdgpu_sriov_ras_telemetry_en(adev) \
+(((adev)->virt.gim_feature & AMDGIM_FEATURE_RAS_TELEMETRY) && (adev)->virt.fw_reserve.ras_telemetry)
+
+#define amdgpu_sriov_ras_telemetry_block_en(adev, sriov_blk) \
+(amdgpu_sriov_ras_telemetry_en((adev)) && (adev)->virt.ras_telemetry_en_caps.all & BIT(sriov_blk))
+
+#define amdgpu_sriov_ras_cper_en(adev) \
+((adev)->virt.gim_feature & AMDGIM_FEATURE_RAS_CPER)
+
 static inline bool is_virtual_machine(void)
 {
-#ifdef CONFIG_X86
+#if defined(CONFIG_X86)
 	return boot_cpu_has(X86_FEATURE_HYPERVISOR);
+#elif defined(CONFIG_ARM64)
+	return !is_kernel_in_hyp_mode();
 #else
 	return false;
 #endif
 }
 
-struct amdgpu_vm;
-int amdgpu_allocate_static_csa(struct amdgpu_device *adev);
-int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm,
-			  struct amdgpu_bo_va **bo_va);
+#define amdgpu_sriov_is_pp_one_vf(adev) \
+	((adev)->virt.gim_feature & AMDGIM_FEATURE_PP_ONE_VF)
+#define amdgpu_sriov_multi_vf_mode(adev) \
+	(amdgpu_sriov_vf(adev) && !amdgpu_sriov_is_pp_one_vf(adev))
+#define amdgpu_sriov_is_debug(adev) \
+	((!amdgpu_in_reset(adev)) && adev->virt.tdr_debug)
+#define amdgpu_sriov_is_normal(adev) \
+	((!amdgpu_in_reset(adev)) && (!adev->virt.tdr_debug))
+#define amdgpu_sriov_is_av1_support(adev) \
+	((adev)->virt.gim_feature & AMDGIM_FEATURE_AV1_SUPPORT)
+#define amdgpu_sriov_is_vcn_rb_decouple(adev) \
+	((adev)->virt.gim_feature & AMDGIM_FEATURE_VCN_RB_DECOUPLE)
+#define amdgpu_sriov_is_mes_info_enable(adev) \
+	((adev)->virt.gim_feature & AMDGIM_FEATURE_MES_INFO_ENABLE)
+
+#define amdgpu_virt_xgmi_migrate_enabled(adev) \
+	((adev)->virt.is_xgmi_node_migrate_enabled && (adev)->gmc.xgmi.node_segment_size != 0)
+
+bool amdgpu_virt_mmio_blocked(struct amdgpu_device *adev);
 void amdgpu_virt_init_setting(struct amdgpu_device *adev);
-uint32_t amdgpu_virt_kiq_rreg(struct amdgpu_device *adev, uint32_t reg);
-void amdgpu_virt_kiq_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v);
 int amdgpu_virt_request_full_gpu(struct amdgpu_device *adev, bool init);
 int amdgpu_virt_release_full_gpu(struct amdgpu_device *adev, bool init);
 int amdgpu_virt_reset_gpu(struct amdgpu_device *adev);
-int amdgpu_sriov_gpu_reset(struct amdgpu_device *adev, struct amdgpu_job *job);
+void amdgpu_virt_request_init_data(struct amdgpu_device *adev);
+void amdgpu_virt_ready_to_reset(struct amdgpu_device *adev);
+int amdgpu_virt_wait_reset(struct amdgpu_device *adev);
 int amdgpu_virt_alloc_mm_table(struct amdgpu_device *adev);
 void amdgpu_virt_free_mm_table(struct amdgpu_device *adev);
+bool amdgpu_virt_rcvd_ras_interrupt(struct amdgpu_device *adev);
+void amdgpu_virt_release_ras_err_handler_data(struct amdgpu_device *adev);
+void amdgpu_virt_init_data_exchange(struct amdgpu_device *adev);
+void amdgpu_virt_exchange_data(struct amdgpu_device *adev);
+void amdgpu_virt_fini_data_exchange(struct amdgpu_device *adev);
+void amdgpu_virt_init(struct amdgpu_device *adev);
+
+bool amdgpu_virt_can_access_debugfs(struct amdgpu_device *adev);
+int amdgpu_virt_enable_access_debugfs(struct amdgpu_device *adev);
+void amdgpu_virt_disable_access_debugfs(struct amdgpu_device *adev);
+
+enum amdgpu_sriov_vf_mode amdgpu_virt_get_sriov_vf_mode(struct amdgpu_device *adev);
 
+void amdgpu_virt_update_sriov_video_codec(struct amdgpu_device *adev,
+			struct amdgpu_video_codec_info *encode, uint32_t encode_array_size,
+			struct amdgpu_video_codec_info *decode, uint32_t decode_array_size);
+void amdgpu_sriov_wreg(struct amdgpu_device *adev,
+		       u32 offset, u32 value,
+		       u32 acc_flags, u32 hwip, u32 xcc_id);
+u32 amdgpu_sriov_rreg(struct amdgpu_device *adev,
+		      u32 offset, u32 acc_flags, u32 hwip, u32 xcc_id);
+bool amdgpu_virt_fw_load_skip_check(struct amdgpu_device *adev,
+			uint32_t ucode_id);
+void amdgpu_virt_pre_reset(struct amdgpu_device *adev);
+void amdgpu_virt_post_reset(struct amdgpu_device *adev);
+bool amdgpu_sriov_xnack_support(struct amdgpu_device *adev);
+bool amdgpu_virt_get_rlcg_reg_access_flag(struct amdgpu_device *adev,
+					  u32 acc_flags, u32 hwip,
+					  bool write, u32 *rlcg_flag);
+u32 amdgpu_virt_rlcg_reg_rw(struct amdgpu_device *adev, u32 offset, u32 v, u32 flag, u32 xcc_id);
+bool amdgpu_virt_get_ras_capability(struct amdgpu_device *adev);
+int amdgpu_virt_req_ras_err_count(struct amdgpu_device *adev, enum amdgpu_ras_block block,
+				  struct ras_err_data *err_data);
+int amdgpu_virt_req_ras_cper_dump(struct amdgpu_device *adev, bool force_update);
+int amdgpu_virt_ras_telemetry_post_reset(struct amdgpu_device *adev);
+bool amdgpu_virt_ras_telemetry_block_en(struct amdgpu_device *adev,
+					enum amdgpu_ras_block block);
+void amdgpu_virt_request_bad_pages(struct amdgpu_device *adev);
+int amdgpu_virt_check_vf_critical_region(struct amdgpu_device *adev, u64 addr, bool *hit);
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c
new file mode 100644
index 000000000000..79bad9cbe2ab
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c
@@ -0,0 +1,659 @@
+// SPDX-License-Identifier: GPL-2.0+
+
+#include <drm/drm_atomic_helper.h>
+#include <drm/drm_edid.h>
+#include <drm/drm_simple_kms_helper.h>
+#include <drm/drm_gem_framebuffer_helper.h>
+#include <drm/drm_vblank.h>
+
+#include "amdgpu.h"
+#ifdef CONFIG_DRM_AMDGPU_SI
+#include "dce_v6_0.h"
+#endif
+#ifdef CONFIG_DRM_AMDGPU_CIK
+#include "dce_v8_0.h"
+#endif
+#include "dce_v10_0.h"
+#include "ivsrcid/ivsrcid_vislands30.h"
+#include "amdgpu_vkms.h"
+#include "amdgpu_display.h"
+#include "atom.h"
+#include "amdgpu_irq.h"
+
+/**
+ * DOC: amdgpu_vkms
+ *
+ * The amdgpu vkms interface provides a virtual KMS interface for several use
+ * cases: devices without display hardware, platforms where the actual display
+ * hardware is not useful (e.g., servers), SR-IOV virtual functions, device
+ * emulation/simulation, and device bring up prior to display hardware being
+ * usable. We previously emulated a legacy KMS interface, but there was a desire
+ * to move to the atomic KMS interface. The vkms driver did everything we
+ * needed, but we wanted KMS support natively in the driver without buffer
+ * sharing and the ability to support an instance of VKMS per device. We first
+ * looked at splitting vkms into a stub driver and a helper module that other
+ * drivers could use to implement a virtual display, but this strategy ended up
+ * being messy due to driver specific callbacks needed for buffer management.
+ * Ultimately, it proved easier to import the vkms code as it mostly used core
+ * drm helpers anyway.
+ */
+
+static const u32 amdgpu_vkms_formats[] = {
+	DRM_FORMAT_XRGB8888,
+};
+
+static enum hrtimer_restart amdgpu_vkms_vblank_simulate(struct hrtimer *timer)
+{
+	struct amdgpu_crtc *amdgpu_crtc = container_of(timer, struct amdgpu_crtc, vblank_timer);
+	struct drm_crtc *crtc = &amdgpu_crtc->base;
+	struct amdgpu_vkms_output *output = drm_crtc_to_amdgpu_vkms_output(crtc);
+	u64 ret_overrun;
+	bool ret;
+
+	ret_overrun = hrtimer_forward_now(&amdgpu_crtc->vblank_timer,
+					  output->period_ns);
+	if (ret_overrun != 1)
+		DRM_WARN("%s: vblank timer overrun\n", __func__);
+
+	ret = drm_crtc_handle_vblank(crtc);
+	/* Don't queue timer again when vblank is disabled. */
+	if (!ret)
+		return HRTIMER_NORESTART;
+
+	return HRTIMER_RESTART;
+}
+
+static int amdgpu_vkms_enable_vblank(struct drm_crtc *crtc)
+{
+	struct drm_vblank_crtc *vblank = drm_crtc_vblank_crtc(crtc);
+	struct amdgpu_vkms_output *out = drm_crtc_to_amdgpu_vkms_output(crtc);
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	drm_calc_timestamping_constants(crtc, &crtc->mode);
+
+	out->period_ns = ktime_set(0, vblank->framedur_ns);
+	hrtimer_start(&amdgpu_crtc->vblank_timer, out->period_ns, HRTIMER_MODE_REL);
+
+	return 0;
+}
+
+static void amdgpu_vkms_disable_vblank(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	hrtimer_try_to_cancel(&amdgpu_crtc->vblank_timer);
+}
+
+static bool amdgpu_vkms_get_vblank_timestamp(struct drm_crtc *crtc,
+					     int *max_error,
+					     ktime_t *vblank_time,
+					     bool in_vblank_irq)
+{
+	struct amdgpu_vkms_output *output = drm_crtc_to_amdgpu_vkms_output(crtc);
+	struct drm_vblank_crtc *vblank = drm_crtc_vblank_crtc(crtc);
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	if (!READ_ONCE(vblank->enabled)) {
+		*vblank_time = ktime_get();
+		return true;
+	}
+
+	*vblank_time = READ_ONCE(amdgpu_crtc->vblank_timer.node.expires);
+
+	if (WARN_ON(*vblank_time == vblank->time))
+		return true;
+
+	/*
+	 * To prevent races we roll the hrtimer forward before we do any
+	 * interrupt processing - this is how real hw works (the interrupt is
+	 * only generated after all the vblank registers are updated) and what
+	 * the vblank core expects. Therefore we need to always correct the
+	 * timestampe by one frame.
+	 */
+	*vblank_time -= output->period_ns;
+
+	return true;
+}
+
+static const struct drm_crtc_funcs amdgpu_vkms_crtc_funcs = {
+	.set_config             = drm_atomic_helper_set_config,
+	.destroy                = drm_crtc_cleanup,
+	.page_flip              = drm_atomic_helper_page_flip,
+	.reset                  = drm_atomic_helper_crtc_reset,
+	.atomic_duplicate_state = drm_atomic_helper_crtc_duplicate_state,
+	.atomic_destroy_state   = drm_atomic_helper_crtc_destroy_state,
+	.enable_vblank		= amdgpu_vkms_enable_vblank,
+	.disable_vblank		= amdgpu_vkms_disable_vblank,
+	.get_vblank_timestamp	= amdgpu_vkms_get_vblank_timestamp,
+};
+
+static void amdgpu_vkms_crtc_atomic_enable(struct drm_crtc *crtc,
+					   struct drm_atomic_state *state)
+{
+	drm_crtc_vblank_on(crtc);
+}
+
+static void amdgpu_vkms_crtc_atomic_disable(struct drm_crtc *crtc,
+					    struct drm_atomic_state *state)
+{
+	drm_crtc_vblank_off(crtc);
+}
+
+static void amdgpu_vkms_crtc_atomic_flush(struct drm_crtc *crtc,
+					  struct drm_atomic_state *state)
+{
+	unsigned long flags;
+	if (crtc->state->event) {
+		spin_lock_irqsave(&crtc->dev->event_lock, flags);
+
+		if (drm_crtc_vblank_get(crtc) != 0)
+			drm_crtc_send_vblank_event(crtc, crtc->state->event);
+		else
+			drm_crtc_arm_vblank_event(crtc, crtc->state->event);
+
+		spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
+
+		crtc->state->event = NULL;
+	}
+}
+
+static const struct drm_crtc_helper_funcs amdgpu_vkms_crtc_helper_funcs = {
+	.atomic_flush	= amdgpu_vkms_crtc_atomic_flush,
+	.atomic_enable	= amdgpu_vkms_crtc_atomic_enable,
+	.atomic_disable	= amdgpu_vkms_crtc_atomic_disable,
+};
+
+static int amdgpu_vkms_crtc_init(struct drm_device *dev, struct drm_crtc *crtc,
+			  struct drm_plane *primary, struct drm_plane *cursor)
+{
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	int ret;
+
+	ret = drm_crtc_init_with_planes(dev, crtc, primary, cursor,
+					&amdgpu_vkms_crtc_funcs, NULL);
+	if (ret) {
+		DRM_ERROR("Failed to init CRTC\n");
+		return ret;
+	}
+
+	drm_crtc_helper_add(crtc, &amdgpu_vkms_crtc_helper_funcs);
+
+	amdgpu_crtc->crtc_id = drm_crtc_index(crtc);
+	adev->mode_info.crtcs[drm_crtc_index(crtc)] = amdgpu_crtc;
+
+	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
+	amdgpu_crtc->encoder = NULL;
+	amdgpu_crtc->connector = NULL;
+	amdgpu_crtc->vsync_timer_enabled = AMDGPU_IRQ_STATE_DISABLE;
+
+	hrtimer_setup(&amdgpu_crtc->vblank_timer, &amdgpu_vkms_vblank_simulate, CLOCK_MONOTONIC,
+		      HRTIMER_MODE_REL);
+
+	return ret;
+}
+
+static const struct drm_connector_funcs amdgpu_vkms_connector_funcs = {
+	.fill_modes = drm_helper_probe_single_connector_modes,
+	.destroy = drm_connector_cleanup,
+	.reset = drm_atomic_helper_connector_reset,
+	.atomic_duplicate_state = drm_atomic_helper_connector_duplicate_state,
+	.atomic_destroy_state = drm_atomic_helper_connector_destroy_state,
+};
+
+static int amdgpu_vkms_conn_get_modes(struct drm_connector *connector)
+{
+	struct drm_device *dev = connector->dev;
+	struct drm_display_mode *mode = NULL;
+	unsigned i;
+	static const struct mode_size {
+		int w;
+		int h;
+	} common_modes[] = {
+		{ 640,  480},
+		{ 720,  480},
+		{ 800,  600},
+		{ 848,  480},
+		{1024,  768},
+		{1152,  768},
+		{1280,  720},
+		{1280,  800},
+		{1280,  854},
+		{1280,  960},
+		{1280, 1024},
+		{1440,  900},
+		{1400, 1050},
+		{1680, 1050},
+		{1600, 1200},
+		{1920, 1080},
+		{1920, 1200},
+		{2560, 1440},
+		{4096, 3112},
+		{3656, 2664},
+		{3840, 2160},
+		{4096, 2160},
+	};
+
+	for (i = 0; i < ARRAY_SIZE(common_modes); i++) {
+		mode = drm_cvt_mode(dev, common_modes[i].w, common_modes[i].h, 60, false, false, false);
+		if (!mode)
+			continue;
+		drm_mode_probed_add(connector, mode);
+	}
+
+	drm_set_preferred_mode(connector, XRES_DEF, YRES_DEF);
+
+	return ARRAY_SIZE(common_modes);
+}
+
+static const struct drm_connector_helper_funcs amdgpu_vkms_conn_helper_funcs = {
+	.get_modes    = amdgpu_vkms_conn_get_modes,
+};
+
+static const struct drm_plane_funcs amdgpu_vkms_plane_funcs = {
+	.update_plane		= drm_atomic_helper_update_plane,
+	.disable_plane		= drm_atomic_helper_disable_plane,
+	.destroy		= drm_plane_cleanup,
+	.reset			= drm_atomic_helper_plane_reset,
+	.atomic_duplicate_state = drm_atomic_helper_plane_duplicate_state,
+	.atomic_destroy_state	= drm_atomic_helper_plane_destroy_state,
+};
+
+static void amdgpu_vkms_plane_atomic_update(struct drm_plane *plane,
+					    struct drm_atomic_state *old_state)
+{
+	return;
+}
+
+static int amdgpu_vkms_plane_atomic_check(struct drm_plane *plane,
+					  struct drm_atomic_state *state)
+{
+	struct drm_plane_state *new_plane_state = drm_atomic_get_new_plane_state(state,
+										 plane);
+	struct drm_crtc_state *crtc_state;
+	int ret;
+
+	if (!new_plane_state->fb || WARN_ON(!new_plane_state->crtc))
+		return 0;
+
+	crtc_state = drm_atomic_get_crtc_state(state,
+					       new_plane_state->crtc);
+	if (IS_ERR(crtc_state))
+		return PTR_ERR(crtc_state);
+
+	ret = drm_atomic_helper_check_plane_state(new_plane_state, crtc_state,
+						  DRM_PLANE_NO_SCALING,
+						  DRM_PLANE_NO_SCALING,
+						  false, true);
+	if (ret != 0)
+		return ret;
+
+	/* for now primary plane must be visible and full screen */
+	if (!new_plane_state->visible)
+		return -EINVAL;
+
+	return 0;
+}
+
+static int amdgpu_vkms_prepare_fb(struct drm_plane *plane,
+				  struct drm_plane_state *new_state)
+{
+	struct amdgpu_framebuffer *afb;
+	struct drm_gem_object *obj;
+	struct amdgpu_device *adev;
+	struct amdgpu_bo *rbo;
+	uint32_t domain;
+	int r;
+
+	if (!new_state->fb) {
+		DRM_DEBUG_KMS("No FB bound\n");
+		return 0;
+	}
+	afb = to_amdgpu_framebuffer(new_state->fb);
+
+	obj = drm_gem_fb_get_obj(new_state->fb, 0);
+	if (!obj) {
+		DRM_ERROR("Failed to get obj from framebuffer\n");
+		return -EINVAL;
+	}
+
+	rbo = gem_to_amdgpu_bo(obj);
+	adev = amdgpu_ttm_adev(rbo->tbo.bdev);
+
+	r = amdgpu_bo_reserve(rbo, true);
+	if (r) {
+		dev_err(adev->dev, "fail to reserve bo (%d)\n", r);
+		return r;
+	}
+
+	r = dma_resv_reserve_fences(rbo->tbo.base.resv, 1);
+	if (r) {
+		dev_err(adev->dev, "allocating fence slot failed (%d)\n", r);
+		goto error_unlock;
+	}
+
+	if (plane->type != DRM_PLANE_TYPE_CURSOR)
+		domain = amdgpu_display_supported_domains(adev, rbo->flags);
+	else
+		domain = AMDGPU_GEM_DOMAIN_VRAM;
+
+	rbo->flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+	r = amdgpu_bo_pin(rbo, domain);
+	if (unlikely(r != 0)) {
+		if (r != -ERESTARTSYS)
+			DRM_ERROR("Failed to pin framebuffer with error %d\n", r);
+		goto error_unlock;
+	}
+
+	r = amdgpu_ttm_alloc_gart(&rbo->tbo);
+	if (unlikely(r != 0)) {
+		DRM_ERROR("%p bind failed\n", rbo);
+		goto error_unpin;
+	}
+
+	amdgpu_bo_unreserve(rbo);
+
+	afb->address = amdgpu_bo_gpu_offset(rbo);
+
+	amdgpu_bo_ref(rbo);
+
+	return 0;
+
+error_unpin:
+	amdgpu_bo_unpin(rbo);
+
+error_unlock:
+	amdgpu_bo_unreserve(rbo);
+	return r;
+}
+
+static void amdgpu_vkms_cleanup_fb(struct drm_plane *plane,
+				   struct drm_plane_state *old_state)
+{
+	struct amdgpu_bo *rbo;
+	struct drm_gem_object *obj;
+	int r;
+
+	if (!old_state->fb)
+		return;
+
+	obj = drm_gem_fb_get_obj(old_state->fb, 0);
+	if (!obj) {
+		DRM_ERROR("Failed to get obj from framebuffer\n");
+		return;
+	}
+
+	rbo = gem_to_amdgpu_bo(obj);
+	r = amdgpu_bo_reserve(rbo, false);
+	if (unlikely(r)) {
+		DRM_ERROR("failed to reserve rbo before unpin\n");
+		return;
+	}
+
+	amdgpu_bo_unpin(rbo);
+	amdgpu_bo_unreserve(rbo);
+	amdgpu_bo_unref(&rbo);
+}
+
+static const struct drm_plane_helper_funcs amdgpu_vkms_primary_helper_funcs = {
+	.atomic_update		= amdgpu_vkms_plane_atomic_update,
+	.atomic_check		= amdgpu_vkms_plane_atomic_check,
+	.prepare_fb		= amdgpu_vkms_prepare_fb,
+	.cleanup_fb		= amdgpu_vkms_cleanup_fb,
+};
+
+static struct drm_plane *amdgpu_vkms_plane_init(struct drm_device *dev,
+						enum drm_plane_type type,
+						int index)
+{
+	struct drm_plane *plane;
+	int ret;
+
+	plane = kzalloc(sizeof(*plane), GFP_KERNEL);
+	if (!plane)
+		return ERR_PTR(-ENOMEM);
+
+	ret = drm_universal_plane_init(dev, plane, 1 << index,
+				       &amdgpu_vkms_plane_funcs,
+				       amdgpu_vkms_formats,
+				       ARRAY_SIZE(amdgpu_vkms_formats),
+				       NULL, type, NULL);
+	if (ret) {
+		kfree(plane);
+		return ERR_PTR(ret);
+	}
+
+	drm_plane_helper_add(plane, &amdgpu_vkms_primary_helper_funcs);
+
+	return plane;
+}
+
+static int amdgpu_vkms_output_init(struct drm_device *dev, struct
+				   amdgpu_vkms_output *output, int index)
+{
+	struct drm_connector *connector = &output->connector;
+	struct drm_encoder *encoder = &output->encoder;
+	struct drm_crtc *crtc = &output->crtc.base;
+	struct drm_plane *primary, *cursor = NULL;
+	int ret;
+
+	primary = amdgpu_vkms_plane_init(dev, DRM_PLANE_TYPE_PRIMARY, index);
+	if (IS_ERR(primary))
+		return PTR_ERR(primary);
+
+	ret = amdgpu_vkms_crtc_init(dev, crtc, primary, cursor);
+	if (ret)
+		goto err_crtc;
+
+	ret = drm_connector_init(dev, connector, &amdgpu_vkms_connector_funcs,
+				 DRM_MODE_CONNECTOR_VIRTUAL);
+	if (ret) {
+		DRM_ERROR("Failed to init connector\n");
+		goto err_connector;
+	}
+
+	drm_connector_helper_add(connector, &amdgpu_vkms_conn_helper_funcs);
+
+	ret = drm_simple_encoder_init(dev, encoder, DRM_MODE_ENCODER_VIRTUAL);
+	if (ret) {
+		DRM_ERROR("Failed to init encoder\n");
+		goto err_encoder;
+	}
+	encoder->possible_crtcs = 1 << index;
+
+	ret = drm_connector_attach_encoder(connector, encoder);
+	if (ret) {
+		DRM_ERROR("Failed to attach connector to encoder\n");
+		goto err_attach;
+	}
+
+	drm_mode_config_reset(dev);
+
+	return 0;
+
+err_attach:
+	drm_encoder_cleanup(encoder);
+
+err_encoder:
+	drm_connector_cleanup(connector);
+
+err_connector:
+	drm_crtc_cleanup(crtc);
+
+err_crtc:
+	drm_plane_cleanup(primary);
+
+	return ret;
+}
+
+const struct drm_mode_config_funcs amdgpu_vkms_mode_funcs = {
+	.fb_create = amdgpu_display_user_framebuffer_create,
+	.atomic_check = drm_atomic_helper_check,
+	.atomic_commit = drm_atomic_helper_commit,
+};
+
+static int amdgpu_vkms_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	int r, i;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	adev->amdgpu_vkms_output = kcalloc(adev->mode_info.num_crtc,
+		sizeof(struct amdgpu_vkms_output), GFP_KERNEL);
+	if (!adev->amdgpu_vkms_output)
+		return -ENOMEM;
+
+	adev_to_drm(adev)->max_vblank_count = 0;
+
+	adev_to_drm(adev)->mode_config.funcs = &amdgpu_vkms_mode_funcs;
+
+	adev_to_drm(adev)->mode_config.max_width = XRES_MAX;
+	adev_to_drm(adev)->mode_config.max_height = YRES_MAX;
+
+	adev_to_drm(adev)->mode_config.preferred_depth = 24;
+	adev_to_drm(adev)->mode_config.prefer_shadow = 1;
+
+	adev_to_drm(adev)->mode_config.fb_modifiers_not_supported = true;
+
+	r = amdgpu_display_modeset_create_props(adev);
+	if (r)
+		return r;
+
+	/* allocate crtcs, encoders, connectors */
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		r = amdgpu_vkms_output_init(adev_to_drm(adev), &adev->amdgpu_vkms_output[i], i);
+		if (r)
+			return r;
+	}
+
+	r = drm_vblank_init(adev_to_drm(adev), adev->mode_info.num_crtc);
+	if (r)
+		return r;
+
+	drm_kms_helper_poll_init(adev_to_drm(adev));
+
+	adev->mode_info.mode_config_initialized = true;
+	return 0;
+}
+
+static int amdgpu_vkms_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i = 0;
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++)
+		if (adev->mode_info.crtcs[i])
+			hrtimer_cancel(&adev->mode_info.crtcs[i]->vblank_timer);
+
+	drm_kms_helper_poll_fini(adev_to_drm(adev));
+	drm_mode_config_cleanup(adev_to_drm(adev));
+
+	adev->mode_info.mode_config_initialized = false;
+
+	drm_edid_free(adev->mode_info.bios_hardcoded_edid);
+	kfree(adev->amdgpu_vkms_output);
+	return 0;
+}
+
+static int amdgpu_vkms_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	switch (adev->asic_type) {
+#ifdef CONFIG_DRM_AMDGPU_SI
+	case CHIP_TAHITI:
+	case CHIP_PITCAIRN:
+	case CHIP_VERDE:
+	case CHIP_OLAND:
+		dce_v6_0_disable_dce(adev);
+		break;
+#endif
+#ifdef CONFIG_DRM_AMDGPU_CIK
+	case CHIP_BONAIRE:
+	case CHIP_HAWAII:
+	case CHIP_KAVERI:
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+		dce_v8_0_disable_dce(adev);
+		break;
+#endif
+	case CHIP_FIJI:
+	case CHIP_TONGA:
+		dce_v10_0_disable_dce(adev);
+		break;
+	case CHIP_TOPAZ:
+#ifdef CONFIG_DRM_AMDGPU_SI
+	case CHIP_HAINAN:
+#endif
+		/* no DCE */
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int amdgpu_vkms_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	return 0;
+}
+
+static int amdgpu_vkms_suspend(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = drm_mode_config_helper_suspend(adev_to_drm(adev));
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static int amdgpu_vkms_resume(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	r = amdgpu_vkms_hw_init(ip_block);
+	if (r)
+		return r;
+	return drm_mode_config_helper_resume(adev_to_drm(ip_block->adev));
+}
+
+static bool amdgpu_vkms_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	return true;
+}
+
+static int amdgpu_vkms_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int amdgpu_vkms_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static const struct amd_ip_funcs amdgpu_vkms_ip_funcs = {
+	.name = "amdgpu_vkms",
+	.sw_init = amdgpu_vkms_sw_init,
+	.sw_fini = amdgpu_vkms_sw_fini,
+	.hw_init = amdgpu_vkms_hw_init,
+	.hw_fini = amdgpu_vkms_hw_fini,
+	.suspend = amdgpu_vkms_suspend,
+	.resume = amdgpu_vkms_resume,
+	.is_idle = amdgpu_vkms_is_idle,
+	.set_clockgating_state = amdgpu_vkms_set_clockgating_state,
+	.set_powergating_state = amdgpu_vkms_set_powergating_state,
+};
+
+const struct amdgpu_ip_block_version amdgpu_vkms_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_DCE,
+	.major = 1,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &amdgpu_vkms_ip_funcs,
+};
+
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.h
new file mode 100644
index 000000000000..4f8722ff37c2
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.h
@@ -0,0 +1,25 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+
+#ifndef _AMDGPU_VKMS_H_
+#define _AMDGPU_VKMS_H_
+
+#define XRES_DEF  1024
+#define YRES_DEF   768
+
+#define XRES_MAX  16384
+#define YRES_MAX  16384
+
+#define drm_crtc_to_amdgpu_vkms_output(target) \
+	container_of(target, struct amdgpu_vkms_output, crtc.base)
+
+extern const struct amdgpu_ip_block_version amdgpu_vkms_ip_block;
+
+struct amdgpu_vkms_output {
+	struct amdgpu_crtc crtc;
+	struct drm_encoder encoder;
+	struct drm_connector connector;
+	ktime_t period_ns;
+	struct drm_pending_vblank_event *event;
+};
+
+#endif /* _AMDGPU_VKMS_H_ */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index bd20ff018512..c1a801203949 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -25,31 +25,66 @@
  *          Alex Deucher
  *          Jerome Glisse
  */
+
 #include <linux/dma-fence-array.h>
 #include <linux/interval_tree_generic.h>
-#include <drm/drmP.h>
+#include <linux/idr.h>
+#include <linux/dma-buf.h>
+
 #include <drm/amdgpu_drm.h>
+#include <drm/drm_drv.h>
+#include <drm/ttm/ttm_tt.h>
+#include <drm/drm_exec.h>
 #include "amdgpu.h"
+#include "amdgpu_vm.h"
 #include "amdgpu_trace.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_gmc.h"
+#include "amdgpu_xgmi.h"
+#include "amdgpu_dma_buf.h"
+#include "amdgpu_res_cursor.h"
+#include "kfd_svm.h"
 
-/*
- * GPUVM
- * GPUVM is similar to the legacy gart on older asics, however
- * rather than there being a single global gart table
- * for the entire GPU, there are multiple VM page tables active
- * at any given time.  The VM page tables can contain a mix
- * vram pages and system memory pages and system memory pages
+/**
+ * DOC: GPUVM
+ *
+ * GPUVM is the MMU functionality provided on the GPU.
+ * GPUVM is similar to the legacy GART on older asics, however
+ * rather than there being a single global GART table
+ * for the entire GPU, there can be multiple GPUVM page tables active
+ * at any given time.  The GPUVM page tables can contain a mix
+ * VRAM pages and system pages (both memory and MMIO) and system pages
  * can be mapped as snooped (cached system pages) or unsnooped
  * (uncached system pages).
- * Each VM has an ID associated with it and there is a page table
- * associated with each VMID.  When execting a command buffer,
- * the kernel tells the the ring what VMID to use for that command
+ *
+ * Each active GPUVM has an ID associated with it and there is a page table
+ * linked with each VMID.  When executing a command buffer,
+ * the kernel tells the engine what VMID to use for that command
  * buffer.  VMIDs are allocated dynamically as commands are submitted.
  * The userspace drivers maintain their own address space and the kernel
  * sets up their pages tables accordingly when they submit their
  * command buffers and a VMID is assigned.
- * Cayman/Trinity support up to 8 active VMs at any given time;
- * SI supports 16.
+ * The hardware supports up to 16 active GPUVMs at any given time.
+ *
+ * Each GPUVM is represented by a 1-2 or 1-5 level page table, depending
+ * on the ASIC family.  GPUVM supports RWX attributes on each page as well
+ * as other features such as encryption and caching attributes.
+ *
+ * VMID 0 is special.  It is the GPUVM used for the kernel driver.  In
+ * addition to an aperture managed by a page table, VMID 0 also has
+ * several other apertures.  There is an aperture for direct access to VRAM
+ * and there is a legacy AGP aperture which just forwards accesses directly
+ * to the matching system physical addresses (or IOVAs when an IOMMU is
+ * present).  These apertures provide direct access to these memories without
+ * incurring the overhead of a page table.  VMID 0 is used by the kernel
+ * driver for tasks like memory management.
+ *
+ * GPU clients (i.e., engines on the GPU) use GPUVM VMIDs to access memory.
+ * For user applications, each application can have their own unique GPUVM
+ * address space.  The application manages the address space and the kernel
+ * driver manages the GPUVM page tables for each process.  If an GPU client
+ * accesses an invalid page, it will generate a GPU page fault, similar to
+ * accessing an invalid page on a CPU.
  */
 
 #define START(node) ((node)->start)
@@ -61,597 +96,594 @@ INTERVAL_TREE_DEFINE(struct amdgpu_bo_va_mapping, rb, uint64_t, __subtree_last,
 #undef START
 #undef LAST
 
-/* Local structure. Encapsulate some VM table update parameters to reduce
- * the number of function parameters
+/**
+ * struct amdgpu_prt_cb - Helper to disable partial resident texture feature from a fence callback
  */
-struct amdgpu_pte_update_params {
-	/* amdgpu device we do this update for */
+struct amdgpu_prt_cb {
+
+	/**
+	 * @adev: amdgpu device
+	 */
 	struct amdgpu_device *adev;
-	/* optional amdgpu_vm we do this update for */
-	struct amdgpu_vm *vm;
-	/* address where to copy page table entries from */
-	uint64_t src;
-	/* indirect buffer to fill with commands */
-	struct amdgpu_ib *ib;
-	/* Function which actually does the update */
-	void (*func)(struct amdgpu_pte_update_params *params, uint64_t pe,
-		     uint64_t addr, unsigned count, uint32_t incr,
-		     uint64_t flags);
-	/* The next two are used during VM update by CPU
-	 *  DMA addresses to use for mapping
-	 *  Kernel pointer of PD/PT BO that needs to be updated
+
+	/**
+	 * @cb: callback
 	 */
-	dma_addr_t *pages_addr;
-	void *kptr;
+	struct dma_fence_cb cb;
 };
 
-/* Helper to disable partial resident texture feature from a fence callback */
-struct amdgpu_prt_cb {
-	struct amdgpu_device *adev;
+/**
+ * struct amdgpu_vm_tlb_seq_struct - Helper to increment the TLB flush sequence
+ */
+struct amdgpu_vm_tlb_seq_struct {
+	/**
+	 * @vm: pointer to the amdgpu_vm structure to set the fence sequence on
+	 */
+	struct amdgpu_vm *vm;
+
+	/**
+	 * @cb: callback
+	 */
 	struct dma_fence_cb cb;
 };
 
 /**
- * amdgpu_vm_num_entries - return the number of entries in a PD/PT
+ * amdgpu_vm_assert_locked - check if VM is correctly locked
+ * @vm: the VM which schould be tested
  *
- * @adev: amdgpu_device pointer
+ * Asserts that the VM root PD is locked.
+ */
+static void amdgpu_vm_assert_locked(struct amdgpu_vm *vm)
+{
+	dma_resv_assert_held(vm->root.bo->tbo.base.resv);
+}
+
+/**
+ * amdgpu_vm_bo_evicted - vm_bo is evicted
  *
- * Calculate the number of entries in a page directory or page table.
+ * @vm_bo: vm_bo which is evicted
+ *
+ * State for PDs/PTs and per VM BOs which are not at the location they should
+ * be.
  */
-static unsigned amdgpu_vm_num_entries(struct amdgpu_device *adev,
-				      unsigned level)
+static void amdgpu_vm_bo_evicted(struct amdgpu_vm_bo_base *vm_bo)
 {
-	if (level == 0)
-		/* For the root directory */
-		return adev->vm_manager.max_pfn >>
-			(adev->vm_manager.block_size *
-			 adev->vm_manager.num_level);
-	else if (level == adev->vm_manager.num_level)
-		/* For the page tables on the leaves */
-		return AMDGPU_VM_PTE_COUNT(adev);
+	struct amdgpu_vm *vm = vm_bo->vm;
+	struct amdgpu_bo *bo = vm_bo->bo;
+
+	vm_bo->moved = true;
+	amdgpu_vm_assert_locked(vm);
+	spin_lock(&vm_bo->vm->status_lock);
+	if (bo->tbo.type == ttm_bo_type_kernel)
+		list_move(&vm_bo->vm_status, &vm->evicted);
 	else
-		/* Everything in between */
-		return 1 << adev->vm_manager.block_size;
+		list_move_tail(&vm_bo->vm_status, &vm->evicted);
+	spin_unlock(&vm_bo->vm->status_lock);
+}
+/**
+ * amdgpu_vm_bo_moved - vm_bo is moved
+ *
+ * @vm_bo: vm_bo which is moved
+ *
+ * State for per VM BOs which are moved, but that change is not yet reflected
+ * in the page tables.
+ */
+static void amdgpu_vm_bo_moved(struct amdgpu_vm_bo_base *vm_bo)
+{
+	amdgpu_vm_assert_locked(vm_bo->vm);
+	spin_lock(&vm_bo->vm->status_lock);
+	list_move(&vm_bo->vm_status, &vm_bo->vm->moved);
+	spin_unlock(&vm_bo->vm->status_lock);
 }
 
 /**
- * amdgpu_vm_bo_size - returns the size of the BOs in bytes
+ * amdgpu_vm_bo_idle - vm_bo is idle
  *
- * @adev: amdgpu_device pointer
+ * @vm_bo: vm_bo which is now idle
  *
- * Calculate the size of the BO for a page directory or page table in bytes.
+ * State for PDs/PTs and per VM BOs which have gone through the state machine
+ * and are now idle.
  */
-static unsigned amdgpu_vm_bo_size(struct amdgpu_device *adev, unsigned level)
+static void amdgpu_vm_bo_idle(struct amdgpu_vm_bo_base *vm_bo)
 {
-	return AMDGPU_GPU_PAGE_ALIGN(amdgpu_vm_num_entries(adev, level) * 8);
+	amdgpu_vm_assert_locked(vm_bo->vm);
+	spin_lock(&vm_bo->vm->status_lock);
+	list_move(&vm_bo->vm_status, &vm_bo->vm->idle);
+	spin_unlock(&vm_bo->vm->status_lock);
+	vm_bo->moved = false;
 }
 
 /**
- * amdgpu_vm_get_pd_bo - add the VM PD to a validation list
+ * amdgpu_vm_bo_invalidated - vm_bo is invalidated
  *
- * @vm: vm providing the BOs
- * @validated: head of validation list
- * @entry: entry to add
+ * @vm_bo: vm_bo which is now invalidated
  *
- * Add the page directory to the list of BOs to
- * validate for command submission.
+ * State for normal BOs which are invalidated and that change not yet reflected
+ * in the PTs.
  */
-void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm,
-			 struct list_head *validated,
-			 struct amdgpu_bo_list_entry *entry)
+static void amdgpu_vm_bo_invalidated(struct amdgpu_vm_bo_base *vm_bo)
 {
-	entry->robj = vm->root.bo;
-	entry->priority = 0;
-	entry->tv.bo = &entry->robj->tbo;
-	entry->tv.shared = true;
-	entry->user_pages = NULL;
-	list_add(&entry->tv.head, validated);
+	spin_lock(&vm_bo->vm->status_lock);
+	list_move(&vm_bo->vm_status, &vm_bo->vm->invalidated);
+	spin_unlock(&vm_bo->vm->status_lock);
 }
 
 /**
- * amdgpu_vm_validate_layer - validate a single page table level
+ * amdgpu_vm_bo_evicted_user - vm_bo is evicted
  *
- * @parent: parent page table level
- * @validate: callback to do the validation
- * @param: parameter for the validation callback
+ * @vm_bo: vm_bo which is evicted
  *
- * Validate the page table BOs on command submission if neccessary.
+ * State for BOs used by user mode queues which are not at the location they
+ * should be.
  */
-static int amdgpu_vm_validate_level(struct amdgpu_vm_pt *parent,
-				    int (*validate)(void *, struct amdgpu_bo *),
-				    void *param, bool use_cpu_for_update,
-				    struct ttm_bo_global *glob)
+static void amdgpu_vm_bo_evicted_user(struct amdgpu_vm_bo_base *vm_bo)
 {
-	unsigned i;
-	int r;
+	vm_bo->moved = true;
+	spin_lock(&vm_bo->vm->status_lock);
+	list_move(&vm_bo->vm_status, &vm_bo->vm->evicted_user);
+	spin_unlock(&vm_bo->vm->status_lock);
+}
 
-	if (use_cpu_for_update) {
-		r = amdgpu_bo_kmap(parent->bo, NULL);
-		if (r)
-			return r;
+/**
+ * amdgpu_vm_bo_relocated - vm_bo is reloacted
+ *
+ * @vm_bo: vm_bo which is relocated
+ *
+ * State for PDs/PTs which needs to update their parent PD.
+ * For the root PD, just move to idle state.
+ */
+static void amdgpu_vm_bo_relocated(struct amdgpu_vm_bo_base *vm_bo)
+{
+	amdgpu_vm_assert_locked(vm_bo->vm);
+	if (vm_bo->bo->parent) {
+		spin_lock(&vm_bo->vm->status_lock);
+		list_move(&vm_bo->vm_status, &vm_bo->vm->relocated);
+		spin_unlock(&vm_bo->vm->status_lock);
+	} else {
+		amdgpu_vm_bo_idle(vm_bo);
 	}
+}
 
-	if (!parent->entries)
-		return 0;
+/**
+ * amdgpu_vm_bo_done - vm_bo is done
+ *
+ * @vm_bo: vm_bo which is now done
+ *
+ * State for normal BOs which are invalidated and that change has been updated
+ * in the PTs.
+ */
+static void amdgpu_vm_bo_done(struct amdgpu_vm_bo_base *vm_bo)
+{
+	amdgpu_vm_assert_locked(vm_bo->vm);
+	spin_lock(&vm_bo->vm->status_lock);
+	list_move(&vm_bo->vm_status, &vm_bo->vm->done);
+	spin_unlock(&vm_bo->vm->status_lock);
+}
 
-	for (i = 0; i <= parent->last_entry_used; ++i) {
-		struct amdgpu_vm_pt *entry = &parent->entries[i];
+/**
+ * amdgpu_vm_bo_reset_state_machine - reset the vm_bo state machine
+ * @vm: the VM which state machine to reset
+ *
+ * Move all vm_bo object in the VM into a state where they will be updated
+ * again during validation.
+ */
+static void amdgpu_vm_bo_reset_state_machine(struct amdgpu_vm *vm)
+{
+	struct amdgpu_vm_bo_base *vm_bo, *tmp;
 
-		if (!entry->bo)
-			continue;
+	amdgpu_vm_assert_locked(vm);
 
-		r = validate(param, entry->bo);
-		if (r)
-			return r;
+	spin_lock(&vm->status_lock);
+	list_splice_init(&vm->done, &vm->invalidated);
+	list_for_each_entry(vm_bo, &vm->invalidated, vm_status)
+		vm_bo->moved = true;
 
-		spin_lock(&glob->lru_lock);
-		ttm_bo_move_to_lru_tail(&entry->bo->tbo);
-		if (entry->bo->shadow)
-			ttm_bo_move_to_lru_tail(&entry->bo->shadow->tbo);
-		spin_unlock(&glob->lru_lock);
+	list_for_each_entry_safe(vm_bo, tmp, &vm->idle, vm_status) {
+		struct amdgpu_bo *bo = vm_bo->bo;
 
-		/*
-		 * Recurse into the sub directory. This is harmless because we
-		 * have only a maximum of 5 layers.
-		 */
-		r = amdgpu_vm_validate_level(entry, validate, param,
-					     use_cpu_for_update, glob);
-		if (r)
-			return r;
+		vm_bo->moved = true;
+		if (!bo || bo->tbo.type != ttm_bo_type_kernel)
+			list_move(&vm_bo->vm_status, &vm_bo->vm->moved);
+		else if (bo->parent)
+			list_move(&vm_bo->vm_status, &vm_bo->vm->relocated);
 	}
-
-	return r;
+	spin_unlock(&vm->status_lock);
 }
 
 /**
- * amdgpu_vm_validate_pt_bos - validate the page table BOs
- *
- * @adev: amdgpu device pointer
- * @vm: vm providing the BOs
- * @validate: callback to do the validation
- * @param: parameter for the validation callback
+ * amdgpu_vm_update_shared - helper to update shared memory stat
+ * @base: base structure for tracking BO usage in a VM
  *
- * Validate the page table BOs on command submission if neccessary.
+ * Takes the vm status_lock and updates the shared memory stat. If the basic
+ * stat changed (e.g. buffer was moved) amdgpu_vm_update_stats need to be called
+ * as well.
  */
-int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm,
-			      int (*validate)(void *p, struct amdgpu_bo *bo),
-			      void *param)
+static void amdgpu_vm_update_shared(struct amdgpu_vm_bo_base *base)
 {
-	uint64_t num_evictions;
+	struct amdgpu_vm *vm = base->vm;
+	struct amdgpu_bo *bo = base->bo;
+	uint64_t size = amdgpu_bo_size(bo);
+	uint32_t bo_memtype = amdgpu_bo_mem_stats_placement(bo);
+	bool shared;
 
-	/* We only need to validate the page tables
-	 * if they aren't already valid.
-	 */
-	num_evictions = atomic64_read(&adev->num_evictions);
-	if (num_evictions == vm->last_eviction_counter)
-		return 0;
-
-	return amdgpu_vm_validate_level(&vm->root, validate, param,
-					vm->use_cpu_for_update,
-					adev->mman.bdev.glob);
+	dma_resv_assert_held(bo->tbo.base.resv);
+	spin_lock(&vm->status_lock);
+	shared = drm_gem_object_is_shared_for_memory_stats(&bo->tbo.base);
+	if (base->shared != shared) {
+		base->shared = shared;
+		if (shared) {
+			vm->stats[bo_memtype].drm.shared += size;
+			vm->stats[bo_memtype].drm.private -= size;
+		} else {
+			vm->stats[bo_memtype].drm.shared -= size;
+			vm->stats[bo_memtype].drm.private += size;
+		}
+	}
+	spin_unlock(&vm->status_lock);
 }
 
 /**
- * amdgpu_vm_alloc_levels - allocate the PD/PT levels
- *
- * @adev: amdgpu_device pointer
- * @vm: requested vm
- * @saddr: start of the address range
- * @eaddr: end of the address range
+ * amdgpu_vm_bo_update_shared - callback when bo gets shared/unshared
+ * @bo: amdgpu buffer object
  *
- * Make sure the page directories and page tables are allocated
+ * Update the per VM stats for all the vm if needed from private to shared or
+ * vice versa.
  */
-static int amdgpu_vm_alloc_levels(struct amdgpu_device *adev,
-				  struct amdgpu_vm *vm,
-				  struct amdgpu_vm_pt *parent,
-				  uint64_t saddr, uint64_t eaddr,
-				  unsigned level)
+void amdgpu_vm_bo_update_shared(struct amdgpu_bo *bo)
 {
-	unsigned shift = (adev->vm_manager.num_level - level) *
-		adev->vm_manager.block_size;
-	unsigned pt_idx, from, to;
-	int r;
-	u64 flags;
-	uint64_t init_value = 0;
-
-	if (!parent->entries) {
-		unsigned num_entries = amdgpu_vm_num_entries(adev, level);
-
-		parent->entries = kvmalloc_array(num_entries,
-						   sizeof(struct amdgpu_vm_pt),
-						   GFP_KERNEL | __GFP_ZERO);
-		if (!parent->entries)
-			return -ENOMEM;
-		memset(parent->entries, 0 , sizeof(struct amdgpu_vm_pt));
-	}
-
-	from = saddr >> shift;
-	to = eaddr >> shift;
-	if (from >= amdgpu_vm_num_entries(adev, level) ||
-	    to >= amdgpu_vm_num_entries(adev, level))
-		return -EINVAL;
+	struct amdgpu_vm_bo_base *base;
 
-	if (to > parent->last_entry_used)
-		parent->last_entry_used = to;
+	for (base = bo->vm_bo; base; base = base->next)
+		amdgpu_vm_update_shared(base);
+}
 
-	++level;
-	saddr = saddr & ((1 << shift) - 1);
-	eaddr = eaddr & ((1 << shift) - 1);
+/**
+ * amdgpu_vm_update_stats_locked - helper to update normal memory stat
+ * @base: base structure for tracking BO usage in a VM
+ * @res:  the ttm_resource to use for the purpose of accounting, may or may not
+ *        be bo->tbo.resource
+ * @sign: if we should add (+1) or subtract (-1) from the stat
+ *
+ * Caller need to have the vm status_lock held. Useful for when multiple update
+ * need to happen at the same time.
+ */
+static void amdgpu_vm_update_stats_locked(struct amdgpu_vm_bo_base *base,
+			    struct ttm_resource *res, int sign)
+{
+	struct amdgpu_vm *vm = base->vm;
+	struct amdgpu_bo *bo = base->bo;
+	int64_t size = sign * amdgpu_bo_size(bo);
+	uint32_t bo_memtype = amdgpu_bo_mem_stats_placement(bo);
 
-	flags = AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS |
-			AMDGPU_GEM_CREATE_VRAM_CLEARED;
-	if (vm->use_cpu_for_update)
-		flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
+	/* For drm-total- and drm-shared-, BO are accounted by their preferred
+	 * placement, see also amdgpu_bo_mem_stats_placement.
+	 */
+	if (base->shared)
+		vm->stats[bo_memtype].drm.shared += size;
 	else
-		flags |= (AMDGPU_GEM_CREATE_NO_CPU_ACCESS |
-				AMDGPU_GEM_CREATE_SHADOW);
-
-	if (vm->pte_support_ats) {
-		init_value = AMDGPU_PTE_SYSTEM;
-		if (level != adev->vm_manager.num_level - 1)
-			init_value |= AMDGPU_PDE_PTE;
-	}
-
-	/* walk over the address space and allocate the page tables */
-	for (pt_idx = from; pt_idx <= to; ++pt_idx) {
-		struct reservation_object *resv = vm->root.bo->tbo.resv;
-		struct amdgpu_vm_pt *entry = &parent->entries[pt_idx];
-		struct amdgpu_bo *pt;
-
-		if (!entry->bo) {
-			r = amdgpu_bo_create(adev,
-					     amdgpu_vm_bo_size(adev, level),
-					     AMDGPU_GPU_PAGE_SIZE, true,
-					     AMDGPU_GEM_DOMAIN_VRAM,
-					     flags,
-					     NULL, resv, init_value, &pt);
-			if (r)
-				return r;
-
-			if (vm->use_cpu_for_update) {
-				r = amdgpu_bo_kmap(pt, NULL);
-				if (r) {
-					amdgpu_bo_unref(&pt);
-					return r;
-				}
-			}
+		vm->stats[bo_memtype].drm.private += size;
 
-			/* Keep a reference to the root directory to avoid
-			* freeing them up in the wrong order.
-			*/
-			pt->parent = amdgpu_bo_ref(vm->root.bo);
+	if (res && res->mem_type < __AMDGPU_PL_NUM) {
+		uint32_t res_memtype = res->mem_type;
 
-			entry->bo = pt;
-			entry->addr = 0;
-		}
-
-		if (level < adev->vm_manager.num_level) {
-			uint64_t sub_saddr = (pt_idx == from) ? saddr : 0;
-			uint64_t sub_eaddr = (pt_idx == to) ? eaddr :
-				((1 << shift) - 1);
-			r = amdgpu_vm_alloc_levels(adev, vm, entry, sub_saddr,
-						   sub_eaddr, level);
-			if (r)
-				return r;
-		}
+		vm->stats[res_memtype].drm.resident += size;
+		/* BO only count as purgeable if it is resident,
+		 * since otherwise there's nothing to purge.
+		 */
+		if (bo->flags & AMDGPU_GEM_CREATE_DISCARDABLE)
+			vm->stats[res_memtype].drm.purgeable += size;
+		if (!(bo->preferred_domains & amdgpu_mem_type_to_domain(res_memtype)))
+			vm->stats[bo_memtype].evicted += size;
 	}
+}
 
-	return 0;
+/**
+ * amdgpu_vm_update_stats - helper to update normal memory stat
+ * @base: base structure for tracking BO usage in a VM
+ * @res:  the ttm_resource to use for the purpose of accounting, may or may not
+ *        be bo->tbo.resource
+ * @sign: if we should add (+1) or subtract (-1) from the stat
+ *
+ * Updates the basic memory stat when bo is added/deleted/moved.
+ */
+void amdgpu_vm_update_stats(struct amdgpu_vm_bo_base *base,
+			    struct ttm_resource *res, int sign)
+{
+	struct amdgpu_vm *vm = base->vm;
+
+	spin_lock(&vm->status_lock);
+	amdgpu_vm_update_stats_locked(base, res, sign);
+	spin_unlock(&vm->status_lock);
 }
 
 /**
- * amdgpu_vm_alloc_pts - Allocate page tables.
+ * amdgpu_vm_bo_base_init - Adds bo to the list of bos associated with the vm
  *
- * @adev: amdgpu_device pointer
- * @vm: VM to allocate page tables for
- * @saddr: Start address which needs to be allocated
- * @size: Size from start address we need.
+ * @base: base structure for tracking BO usage in a VM
+ * @vm: vm to which bo is to be added
+ * @bo: amdgpu buffer object
+ *
+ * Initialize a bo_va_base structure and add it to the appropriate lists
  *
- * Make sure the page tables are allocated.
  */
-int amdgpu_vm_alloc_pts(struct amdgpu_device *adev,
-			struct amdgpu_vm *vm,
-			uint64_t saddr, uint64_t size)
+void amdgpu_vm_bo_base_init(struct amdgpu_vm_bo_base *base,
+			    struct amdgpu_vm *vm, struct amdgpu_bo *bo)
 {
-	uint64_t last_pfn;
-	uint64_t eaddr;
+	base->vm = vm;
+	base->bo = bo;
+	base->next = NULL;
+	INIT_LIST_HEAD(&base->vm_status);
 
-	/* validate the parameters */
-	if (saddr & AMDGPU_GPU_PAGE_MASK || size & AMDGPU_GPU_PAGE_MASK)
-		return -EINVAL;
+	if (!bo)
+		return;
+	base->next = bo->vm_bo;
+	bo->vm_bo = base;
 
-	eaddr = saddr + size - 1;
-	last_pfn = eaddr / AMDGPU_GPU_PAGE_SIZE;
-	if (last_pfn >= adev->vm_manager.max_pfn) {
-		dev_err(adev->dev, "va above limit (0x%08llX >= 0x%08llX)\n",
-			last_pfn, adev->vm_manager.max_pfn);
-		return -EINVAL;
-	}
+	spin_lock(&vm->status_lock);
+	base->shared = drm_gem_object_is_shared_for_memory_stats(&bo->tbo.base);
+	amdgpu_vm_update_stats_locked(base, bo->tbo.resource, +1);
+	spin_unlock(&vm->status_lock);
 
-	saddr /= AMDGPU_GPU_PAGE_SIZE;
-	eaddr /= AMDGPU_GPU_PAGE_SIZE;
+	if (!amdgpu_vm_is_bo_always_valid(vm, bo))
+		return;
+
+	dma_resv_assert_held(vm->root.bo->tbo.base.resv);
+
+	ttm_bo_set_bulk_move(&bo->tbo, &vm->lru_bulk_move);
+	if (bo->tbo.type == ttm_bo_type_kernel && bo->parent)
+		amdgpu_vm_bo_relocated(base);
+	else
+		amdgpu_vm_bo_idle(base);
 
-	return amdgpu_vm_alloc_levels(adev, vm, &vm->root, saddr, eaddr, 0);
+	if (bo->preferred_domains &
+	    amdgpu_mem_type_to_domain(bo->tbo.resource->mem_type))
+		return;
+
+	/*
+	 * we checked all the prerequisites, but it looks like this per vm bo
+	 * is currently evicted. add the bo to the evicted list to make sure it
+	 * is validated on next vm use to avoid fault.
+	 * */
+	amdgpu_vm_bo_evicted(base);
 }
 
 /**
- * amdgpu_vm_had_gpu_reset - check if reset occured since last use
+ * amdgpu_vm_lock_pd - lock PD in drm_exec
  *
- * @adev: amdgpu_device pointer
- * @id: VMID structure
+ * @vm: vm providing the BOs
+ * @exec: drm execution context
+ * @num_fences: number of extra fences to reserve
  *
- * Check if GPU reset occured since last use of the VMID.
+ * Lock the VM root PD in the DRM execution context.
  */
-static bool amdgpu_vm_had_gpu_reset(struct amdgpu_device *adev,
-				    struct amdgpu_vm_id *id)
+int amdgpu_vm_lock_pd(struct amdgpu_vm *vm, struct drm_exec *exec,
+		      unsigned int num_fences)
 {
-	return id->current_gpu_reset_count !=
-		atomic_read(&adev->gpu_reset_counter);
+	/* We need at least two fences for the VM PD/PT updates */
+	return drm_exec_prepare_obj(exec, &vm->root.bo->tbo.base,
+				    2 + num_fences);
 }
 
-static bool amdgpu_vm_reserved_vmid_ready(struct amdgpu_vm *vm, unsigned vmhub)
+/**
+ * amdgpu_vm_lock_done_list - lock all BOs on the done list
+ * @vm: vm providing the BOs
+ * @exec: drm execution context
+ * @num_fences: number of extra fences to reserve
+ *
+ * Lock the BOs on the done list in the DRM execution context.
+ */
+int amdgpu_vm_lock_done_list(struct amdgpu_vm *vm, struct drm_exec *exec,
+			     unsigned int num_fences)
 {
-	return !!vm->reserved_vmid[vmhub];
-}
+	struct list_head *prev = &vm->done;
+	struct amdgpu_bo_va *bo_va;
+	struct amdgpu_bo *bo;
+	int ret;
 
-/* idr_mgr->lock must be held */
-static int amdgpu_vm_grab_reserved_vmid_locked(struct amdgpu_vm *vm,
-					       struct amdgpu_ring *ring,
-					       struct amdgpu_sync *sync,
-					       struct dma_fence *fence,
-					       struct amdgpu_job *job)
-{
-	struct amdgpu_device *adev = ring->adev;
-	unsigned vmhub = ring->funcs->vmhub;
-	uint64_t fence_context = adev->fence_context + ring->idx;
-	struct amdgpu_vm_id *id = vm->reserved_vmid[vmhub];
-	struct amdgpu_vm_id_manager *id_mgr = &adev->vm_manager.id_mgr[vmhub];
-	struct dma_fence *updates = sync->last_vm_update;
-	int r = 0;
-	struct dma_fence *flushed, *tmp;
-	bool needs_flush = vm->use_cpu_for_update;
-
-	flushed  = id->flushed_updates;
-	if ((amdgpu_vm_had_gpu_reset(adev, id)) ||
-	    (atomic64_read(&id->owner) != vm->client_id) ||
-	    (job->vm_pd_addr != id->pd_gpu_addr) ||
-	    (updates && (!flushed || updates->context != flushed->context ||
-			dma_fence_is_later(updates, flushed))) ||
-	    (!id->last_flush || (id->last_flush->context != fence_context &&
-				 !dma_fence_is_signaled(id->last_flush)))) {
-		needs_flush = true;
-		/* to prevent one context starved by another context */
-		id->pd_gpu_addr = 0;
-		tmp = amdgpu_sync_peek_fence(&id->active, ring);
-		if (tmp) {
-			r = amdgpu_sync_fence(adev, sync, tmp);
-			return r;
+	/* We can only trust prev->next while holding the lock */
+	spin_lock(&vm->status_lock);
+	while (!list_is_head(prev->next, &vm->done)) {
+		bo_va = list_entry(prev->next, typeof(*bo_va), base.vm_status);
+		spin_unlock(&vm->status_lock);
+
+		bo = bo_va->base.bo;
+		if (bo) {
+			ret = drm_exec_prepare_obj(exec, &bo->tbo.base, 1);
+			if (unlikely(ret))
+				return ret;
 		}
+		spin_lock(&vm->status_lock);
+		prev = prev->next;
 	}
+	spin_unlock(&vm->status_lock);
 
-	/* Good we can use this VMID. Remember this submission as
-	* user of the VMID.
-	*/
-	r = amdgpu_sync_fence(ring->adev, &id->active, fence);
-	if (r)
-		goto out;
-
-	if (updates && (!flushed || updates->context != flushed->context ||
-			dma_fence_is_later(updates, flushed))) {
-		dma_fence_put(id->flushed_updates);
-		id->flushed_updates = dma_fence_get(updates);
-	}
-	id->pd_gpu_addr = job->vm_pd_addr;
-	atomic64_set(&id->owner, vm->client_id);
-	job->vm_needs_flush = needs_flush;
-	if (needs_flush) {
-		dma_fence_put(id->last_flush);
-		id->last_flush = NULL;
-	}
-	job->vm_id = id - id_mgr->ids;
-	trace_amdgpu_vm_grab_id(vm, ring, job);
-out:
-	return r;
+	return 0;
 }
 
 /**
- * amdgpu_vm_grab_id - allocate the next free VMID
+ * amdgpu_vm_move_to_lru_tail - move all BOs to the end of LRU
  *
- * @vm: vm to allocate id for
- * @ring: ring we want to submit job to
- * @sync: sync object where we add dependencies
- * @fence: fence protecting ID from reuse
+ * @adev: amdgpu device pointer
+ * @vm: vm providing the BOs
  *
- * Allocate an id for the vm, adding fences to the sync obj as necessary.
+ * Move all BOs to the end of LRU and remember their positions to put them
+ * together.
  */
-int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
-		      struct amdgpu_sync *sync, struct dma_fence *fence,
-		      struct amdgpu_job *job)
+void amdgpu_vm_move_to_lru_tail(struct amdgpu_device *adev,
+				struct amdgpu_vm *vm)
 {
-	struct amdgpu_device *adev = ring->adev;
-	unsigned vmhub = ring->funcs->vmhub;
-	struct amdgpu_vm_id_manager *id_mgr = &adev->vm_manager.id_mgr[vmhub];
-	uint64_t fence_context = adev->fence_context + ring->idx;
-	struct dma_fence *updates = sync->last_vm_update;
-	struct amdgpu_vm_id *id, *idle;
-	struct dma_fence **fences;
-	unsigned i;
-	int r = 0;
-
-	mutex_lock(&id_mgr->lock);
-	if (amdgpu_vm_reserved_vmid_ready(vm, vmhub)) {
-		r = amdgpu_vm_grab_reserved_vmid_locked(vm, ring, sync, fence, job);
-		mutex_unlock(&id_mgr->lock);
-		return r;
-	}
-	fences = kmalloc_array(sizeof(void *), id_mgr->num_ids, GFP_KERNEL);
-	if (!fences) {
-		mutex_unlock(&id_mgr->lock);
-		return -ENOMEM;
-	}
-	/* Check if we have an idle VMID */
-	i = 0;
-	list_for_each_entry(idle, &id_mgr->ids_lru, list) {
-		fences[i] = amdgpu_sync_peek_fence(&idle->active, ring);
-		if (!fences[i])
-			break;
-		++i;
-	}
-
-	/* If we can't find a idle VMID to use, wait till one becomes available */
-	if (&idle->list == &id_mgr->ids_lru) {
-		u64 fence_context = adev->vm_manager.fence_context + ring->idx;
-		unsigned seqno = ++adev->vm_manager.seqno[ring->idx];
-		struct dma_fence_array *array;
-		unsigned j;
-
-		for (j = 0; j < i; ++j)
-			dma_fence_get(fences[j]);
-
-		array = dma_fence_array_create(i, fences, fence_context,
-					   seqno, true);
-		if (!array) {
-			for (j = 0; j < i; ++j)
-				dma_fence_put(fences[j]);
-			kfree(fences);
-			r = -ENOMEM;
-			goto error;
-		}
-
+	spin_lock(&adev->mman.bdev.lru_lock);
+	ttm_lru_bulk_move_tail(&vm->lru_bulk_move);
+	spin_unlock(&adev->mman.bdev.lru_lock);
+}
 
-		r = amdgpu_sync_fence(ring->adev, sync, &array->base);
-		dma_fence_put(&array->base);
-		if (r)
-			goto error;
+/* Create scheduler entities for page table updates */
+static int amdgpu_vm_init_entities(struct amdgpu_device *adev,
+				   struct amdgpu_vm *vm)
+{
+	int r;
 
-		mutex_unlock(&id_mgr->lock);
-		return 0;
+	r = drm_sched_entity_init(&vm->immediate, DRM_SCHED_PRIORITY_NORMAL,
+				  adev->vm_manager.vm_pte_scheds,
+				  adev->vm_manager.vm_pte_num_scheds, NULL);
+	if (r)
+		goto error;
 
-	}
-	kfree(fences);
+	return drm_sched_entity_init(&vm->delayed, DRM_SCHED_PRIORITY_NORMAL,
+				     adev->vm_manager.vm_pte_scheds,
+				     adev->vm_manager.vm_pte_num_scheds, NULL);
 
-	job->vm_needs_flush = vm->use_cpu_for_update;
-	/* Check if we can use a VMID already assigned to this VM */
-	list_for_each_entry_reverse(id, &id_mgr->ids_lru, list) {
-		struct dma_fence *flushed;
-		bool needs_flush = vm->use_cpu_for_update;
+error:
+	drm_sched_entity_destroy(&vm->immediate);
+	return r;
+}
 
-		/* Check all the prerequisites to using this VMID */
-		if (amdgpu_vm_had_gpu_reset(adev, id))
-			continue;
+/* Destroy the entities for page table updates again */
+static void amdgpu_vm_fini_entities(struct amdgpu_vm *vm)
+{
+	drm_sched_entity_destroy(&vm->immediate);
+	drm_sched_entity_destroy(&vm->delayed);
+}
 
-		if (atomic64_read(&id->owner) != vm->client_id)
-			continue;
+/**
+ * amdgpu_vm_generation - return the page table re-generation counter
+ * @adev: the amdgpu_device
+ * @vm: optional VM to check, might be NULL
+ *
+ * Returns a page table re-generation token to allow checking if submissions
+ * are still valid to use this VM. The VM parameter might be NULL in which case
+ * just the VRAM lost counter will be used.
+ */
+uint64_t amdgpu_vm_generation(struct amdgpu_device *adev, struct amdgpu_vm *vm)
+{
+	uint64_t result = (u64)atomic_read(&adev->vram_lost_counter) << 32;
 
-		if (job->vm_pd_addr != id->pd_gpu_addr)
-			continue;
+	if (!vm)
+		return result;
 
-		if (!id->last_flush ||
-		    (id->last_flush->context != fence_context &&
-		     !dma_fence_is_signaled(id->last_flush)))
-			needs_flush = true;
+	result += lower_32_bits(vm->generation);
+	/* Add one if the page tables will be re-generated on next CS */
+	if (drm_sched_entity_error(&vm->delayed))
+		++result;
 
-		flushed  = id->flushed_updates;
-		if (updates && (!flushed || dma_fence_is_later(updates, flushed)))
-			needs_flush = true;
+	return result;
+}
 
-		/* Concurrent flushes are only possible starting with Vega10 */
-		if (adev->asic_type < CHIP_VEGA10 && needs_flush)
-			continue;
+/**
+ * amdgpu_vm_validate - validate evicted BOs tracked in the VM
+ *
+ * @adev: amdgpu device pointer
+ * @vm: vm providing the BOs
+ * @ticket: optional reservation ticket used to reserve the VM
+ * @validate: callback to do the validation
+ * @param: parameter for the validation callback
+ *
+ * Validate the page table BOs and per-VM BOs on command submission if
+ * necessary. If a ticket is given, also try to validate evicted user queue
+ * BOs. They must already be reserved with the given ticket.
+ *
+ * Returns:
+ * Validation result.
+ */
+int amdgpu_vm_validate(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+		       struct ww_acquire_ctx *ticket,
+		       int (*validate)(void *p, struct amdgpu_bo *bo),
+		       void *param)
+{
+	uint64_t new_vm_generation = amdgpu_vm_generation(adev, vm);
+	struct amdgpu_vm_bo_base *bo_base;
+	struct amdgpu_bo *bo;
+	int r;
 
-		/* Good we can use this VMID. Remember this submission as
-		 * user of the VMID.
-		 */
-		r = amdgpu_sync_fence(ring->adev, &id->active, fence);
+	if (vm->generation != new_vm_generation) {
+		vm->generation = new_vm_generation;
+		amdgpu_vm_bo_reset_state_machine(vm);
+		amdgpu_vm_fini_entities(vm);
+		r = amdgpu_vm_init_entities(adev, vm);
 		if (r)
-			goto error;
+			return r;
+	}
 
-		if (updates && (!flushed || dma_fence_is_later(updates, flushed))) {
-			dma_fence_put(id->flushed_updates);
-			id->flushed_updates = dma_fence_get(updates);
-		}
+	spin_lock(&vm->status_lock);
+	while (!list_empty(&vm->evicted)) {
+		bo_base = list_first_entry(&vm->evicted,
+					   struct amdgpu_vm_bo_base,
+					   vm_status);
+		spin_unlock(&vm->status_lock);
 
-		if (needs_flush)
-			goto needs_flush;
-		else
-			goto no_flush_needed;
+		bo = bo_base->bo;
 
-	};
+		r = validate(param, bo);
+		if (r)
+			return r;
 
-	/* Still no ID to use? Then use the idle one found earlier */
-	id = idle;
+		if (bo->tbo.type != ttm_bo_type_kernel) {
+			amdgpu_vm_bo_moved(bo_base);
+		} else {
+			vm->update_funcs->map_table(to_amdgpu_bo_vm(bo));
+			amdgpu_vm_bo_relocated(bo_base);
+		}
+		spin_lock(&vm->status_lock);
+	}
+	while (ticket && !list_empty(&vm->evicted_user)) {
+		bo_base = list_first_entry(&vm->evicted_user,
+					   struct amdgpu_vm_bo_base,
+					   vm_status);
+		spin_unlock(&vm->status_lock);
 
-	/* Remember this submission as user of the VMID */
-	r = amdgpu_sync_fence(ring->adev, &id->active, fence);
-	if (r)
-		goto error;
+		bo = bo_base->bo;
+		dma_resv_assert_held(bo->tbo.base.resv);
 
-	id->pd_gpu_addr = job->vm_pd_addr;
-	dma_fence_put(id->flushed_updates);
-	id->flushed_updates = dma_fence_get(updates);
-	atomic64_set(&id->owner, vm->client_id);
+		r = validate(param, bo);
+		if (r)
+			return r;
 
-needs_flush:
-	job->vm_needs_flush = true;
-	dma_fence_put(id->last_flush);
-	id->last_flush = NULL;
+		amdgpu_vm_bo_invalidated(bo_base);
 
-no_flush_needed:
-	list_move_tail(&id->list, &id_mgr->ids_lru);
+		spin_lock(&vm->status_lock);
+	}
+	spin_unlock(&vm->status_lock);
 
-	job->vm_id = id - id_mgr->ids;
-	trace_amdgpu_vm_grab_id(vm, ring, job);
+	amdgpu_vm_eviction_lock(vm);
+	vm->evicting = false;
+	amdgpu_vm_eviction_unlock(vm);
 
-error:
-	mutex_unlock(&id_mgr->lock);
-	return r;
+	return 0;
 }
 
-static void amdgpu_vm_free_reserved_vmid(struct amdgpu_device *adev,
-					  struct amdgpu_vm *vm,
-					  unsigned vmhub)
+/**
+ * amdgpu_vm_ready - check VM is ready for updates
+ *
+ * @vm: VM to check
+ *
+ * Check if all VM PDs/PTs are ready for updates
+ *
+ * Returns:
+ * True if VM is not evicting and all VM entities are not stopped
+ */
+bool amdgpu_vm_ready(struct amdgpu_vm *vm)
 {
-	struct amdgpu_vm_id_manager *id_mgr = &adev->vm_manager.id_mgr[vmhub];
+	bool ret;
 
-	mutex_lock(&id_mgr->lock);
-	if (vm->reserved_vmid[vmhub]) {
-		list_add(&vm->reserved_vmid[vmhub]->list,
-			&id_mgr->ids_lru);
-		vm->reserved_vmid[vmhub] = NULL;
-		atomic_dec(&id_mgr->reserved_vmid_num);
-	}
-	mutex_unlock(&id_mgr->lock);
-}
+	amdgpu_vm_assert_locked(vm);
 
-static int amdgpu_vm_alloc_reserved_vmid(struct amdgpu_device *adev,
-					 struct amdgpu_vm *vm,
-					 unsigned vmhub)
-{
-	struct amdgpu_vm_id_manager *id_mgr;
-	struct amdgpu_vm_id *idle;
-	int r = 0;
+	amdgpu_vm_eviction_lock(vm);
+	ret = !vm->evicting;
+	amdgpu_vm_eviction_unlock(vm);
 
-	id_mgr = &adev->vm_manager.id_mgr[vmhub];
-	mutex_lock(&id_mgr->lock);
-	if (vm->reserved_vmid[vmhub])
-		goto unlock;
-	if (atomic_inc_return(&id_mgr->reserved_vmid_num) >
-	    AMDGPU_VM_MAX_RESERVED_VMID) {
-		DRM_ERROR("Over limitation of reserved vmid\n");
-		atomic_dec(&id_mgr->reserved_vmid_num);
-		r = -EINVAL;
-		goto unlock;
-	}
-	/* Select the first entry VMID */
-	idle = list_first_entry(&id_mgr->ids_lru, struct amdgpu_vm_id, list);
-	list_del_init(&idle->list);
-	vm->reserved_vmid[vmhub] = idle;
-	mutex_unlock(&id_mgr->lock);
+	spin_lock(&vm->status_lock);
+	ret &= list_empty(&vm->evicted);
+	spin_unlock(&vm->status_lock);
 
-	return 0;
-unlock:
-	mutex_unlock(&id_mgr->lock);
-	return r;
+	spin_lock(&vm->immediate.lock);
+	ret &= !vm->immediate.stopped;
+	spin_unlock(&vm->immediate.lock);
+
+	spin_lock(&vm->delayed.lock);
+	ret &= !vm->delayed.stopped;
+	spin_unlock(&vm->delayed.lock);
+
+	return ret;
 }
 
 /**
@@ -668,7 +700,7 @@ void amdgpu_vm_check_compute_bug(struct amdgpu_device *adev)
 
 	has_compute_vm_bug = false;
 
-	ip_block = amdgpu_get_ip_block(adev, AMD_IP_BLOCK_TYPE_GFX);
+	ip_block = amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_GFX);
 	if (ip_block) {
 		/* Compute has a VM bug for GFX version < 7.
 		   Compute has a VM bug for GFX 8 MEC firmware version < 673.*/
@@ -689,160 +721,175 @@ void amdgpu_vm_check_compute_bug(struct amdgpu_device *adev)
 	}
 }
 
+/**
+ * amdgpu_vm_need_pipeline_sync - Check if pipe sync is needed for job.
+ *
+ * @ring: ring on which the job will be submitted
+ * @job: job to submit
+ *
+ * Returns:
+ * True if sync is needed.
+ */
 bool amdgpu_vm_need_pipeline_sync(struct amdgpu_ring *ring,
 				  struct amdgpu_job *job)
 {
 	struct amdgpu_device *adev = ring->adev;
-	unsigned vmhub = ring->funcs->vmhub;
-	struct amdgpu_vm_id_manager *id_mgr = &adev->vm_manager.id_mgr[vmhub];
-	struct amdgpu_vm_id *id;
-	bool gds_switch_needed;
-	bool vm_flush_needed = job->vm_needs_flush || ring->has_compute_vm_bug;
+	unsigned vmhub = ring->vm_hub;
+	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
 
-	if (job->vm_id == 0)
+	if (job->vmid == 0)
 		return false;
-	id = &id_mgr->ids[job->vm_id];
-	gds_switch_needed = ring->funcs->emit_gds_switch && (
-		id->gds_base != job->gds_base ||
-		id->gds_size != job->gds_size ||
-		id->gws_base != job->gws_base ||
-		id->gws_size != job->gws_size ||
-		id->oa_base != job->oa_base ||
-		id->oa_size != job->oa_size);
-
-	if (amdgpu_vm_had_gpu_reset(adev, id))
+
+	if (job->vm_needs_flush || ring->has_compute_vm_bug)
 		return true;
 
-	return vm_flush_needed || gds_switch_needed;
-}
+	if (ring->funcs->emit_gds_switch && job->gds_switch_needed)
+		return true;
 
-static bool amdgpu_vm_is_large_bar(struct amdgpu_device *adev)
-{
-	return (adev->mc.real_vram_size == adev->mc.visible_vram_size);
+	if (amdgpu_vmid_had_gpu_reset(adev, &id_mgr->ids[job->vmid]))
+		return true;
+
+	return false;
 }
 
 /**
  * amdgpu_vm_flush - hardware flush the vm
  *
  * @ring: ring to use for flush
- * @vm_id: vmid number to use
- * @pd_addr: address of the page directory
+ * @job:  related job
+ * @need_pipe_sync: is pipe sync needed
  *
  * Emit a VM flush when it is necessary.
+ *
+ * Returns:
+ * 0 on success, errno otherwise.
  */
-int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job, bool need_pipe_sync)
+int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job,
+		    bool need_pipe_sync)
 {
 	struct amdgpu_device *adev = ring->adev;
-	unsigned vmhub = ring->funcs->vmhub;
-	struct amdgpu_vm_id_manager *id_mgr = &adev->vm_manager.id_mgr[vmhub];
-	struct amdgpu_vm_id *id = &id_mgr->ids[job->vm_id];
-	bool gds_switch_needed = ring->funcs->emit_gds_switch && (
-		id->gds_base != job->gds_base ||
-		id->gds_size != job->gds_size ||
-		id->gws_base != job->gws_base ||
-		id->gws_size != job->gws_size ||
-		id->oa_base != job->oa_base ||
-		id->oa_size != job->oa_size);
+	struct amdgpu_isolation *isolation = &adev->isolation[ring->xcp_id];
+	unsigned vmhub = ring->vm_hub;
+	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
+	struct amdgpu_vmid *id = &id_mgr->ids[job->vmid];
+	bool spm_update_needed = job->spm_update_needed;
+	bool gds_switch_needed = ring->funcs->emit_gds_switch &&
+		job->gds_switch_needed;
 	bool vm_flush_needed = job->vm_needs_flush;
-	unsigned patch_offset = 0;
+	bool cleaner_shader_needed = false;
+	bool pasid_mapping_needed = false;
+	struct dma_fence *fence = NULL;
+	struct amdgpu_fence *af;
+	unsigned int patch;
 	int r;
 
-	if (amdgpu_vm_had_gpu_reset(adev, id)) {
+	if (amdgpu_vmid_had_gpu_reset(adev, id)) {
 		gds_switch_needed = true;
 		vm_flush_needed = true;
+		pasid_mapping_needed = true;
+		spm_update_needed = true;
 	}
 
-	if (!vm_flush_needed && !gds_switch_needed && !need_pipe_sync)
+	mutex_lock(&id_mgr->lock);
+	if (id->pasid != job->pasid || !id->pasid_mapping ||
+	    !dma_fence_is_signaled(id->pasid_mapping))
+		pasid_mapping_needed = true;
+	mutex_unlock(&id_mgr->lock);
+
+	gds_switch_needed &= !!ring->funcs->emit_gds_switch;
+	vm_flush_needed &= !!ring->funcs->emit_vm_flush  &&
+			job->vm_pd_addr != AMDGPU_BO_INVALID_OFFSET;
+	pasid_mapping_needed &= adev->gmc.gmc_funcs->emit_pasid_mapping &&
+		ring->funcs->emit_wreg;
+
+	cleaner_shader_needed = job->run_cleaner_shader &&
+		adev->gfx.enable_cleaner_shader &&
+		ring->funcs->emit_cleaner_shader && job->base.s_fence &&
+		&job->base.s_fence->scheduled == isolation->spearhead;
+
+	if (!vm_flush_needed && !gds_switch_needed && !need_pipe_sync &&
+	    !cleaner_shader_needed)
 		return 0;
 
+	amdgpu_ring_ib_begin(ring);
 	if (ring->funcs->init_cond_exec)
-		patch_offset = amdgpu_ring_init_cond_exec(ring);
+		patch = amdgpu_ring_init_cond_exec(ring,
+						   ring->cond_exe_gpu_addr);
 
 	if (need_pipe_sync)
 		amdgpu_ring_emit_pipeline_sync(ring);
 
-	if (ring->funcs->emit_vm_flush && vm_flush_needed) {
-		struct dma_fence *fence;
+	if (cleaner_shader_needed)
+		ring->funcs->emit_cleaner_shader(ring);
+
+	if (vm_flush_needed) {
+		trace_amdgpu_vm_flush(ring, job->vmid, job->vm_pd_addr);
+		amdgpu_ring_emit_vm_flush(ring, job->vmid, job->vm_pd_addr);
+	}
+
+	if (pasid_mapping_needed)
+		amdgpu_gmc_emit_pasid_mapping(ring, job->vmid, job->pasid);
 
-		trace_amdgpu_vm_flush(ring, job->vm_id, job->vm_pd_addr);
-		amdgpu_ring_emit_vm_flush(ring, job->vm_id, job->vm_pd_addr);
+	if (spm_update_needed && adev->gfx.rlc.funcs->update_spm_vmid)
+		adev->gfx.rlc.funcs->update_spm_vmid(adev, ring, job->vmid);
 
-		r = amdgpu_fence_emit(ring, &fence);
+	if (ring->funcs->emit_gds_switch &&
+	    gds_switch_needed) {
+		amdgpu_ring_emit_gds_switch(ring, job->vmid, job->gds_base,
+					    job->gds_size, job->gws_base,
+					    job->gws_size, job->oa_base,
+					    job->oa_size);
+	}
+
+	if (vm_flush_needed || pasid_mapping_needed || cleaner_shader_needed) {
+		r = amdgpu_fence_emit(ring, &fence, NULL, 0);
 		if (r)
 			return r;
+		/* this is part of the job's context */
+		af = container_of(fence, struct amdgpu_fence, base);
+		af->context = job->base.s_fence ? job->base.s_fence->finished.context : 0;
+	}
 
+	if (vm_flush_needed) {
 		mutex_lock(&id_mgr->lock);
 		dma_fence_put(id->last_flush);
-		id->last_flush = fence;
-		id->current_gpu_reset_count = atomic_read(&adev->gpu_reset_counter);
+		id->last_flush = dma_fence_get(fence);
+		id->current_gpu_reset_count =
+			atomic_read(&adev->gpu_reset_counter);
 		mutex_unlock(&id_mgr->lock);
 	}
 
-	if (ring->funcs->emit_gds_switch && gds_switch_needed) {
-		id->gds_base = job->gds_base;
-		id->gds_size = job->gds_size;
-		id->gws_base = job->gws_base;
-		id->gws_size = job->gws_size;
-		id->oa_base = job->oa_base;
-		id->oa_size = job->oa_size;
-		amdgpu_ring_emit_gds_switch(ring, job->vm_id, job->gds_base,
-					    job->gds_size, job->gws_base,
-					    job->gws_size, job->oa_base,
-					    job->oa_size);
+	if (pasid_mapping_needed) {
+		mutex_lock(&id_mgr->lock);
+		id->pasid = job->pasid;
+		dma_fence_put(id->pasid_mapping);
+		id->pasid_mapping = dma_fence_get(fence);
+		mutex_unlock(&id_mgr->lock);
 	}
 
-	if (ring->funcs->patch_cond_exec)
-		amdgpu_ring_patch_cond_exec(ring, patch_offset);
+	/*
+	 * Make sure that all other submissions wait for the cleaner shader to
+	 * finish before we push them to the HW.
+	 */
+	if (cleaner_shader_needed) {
+		trace_amdgpu_cleaner_shader(ring, fence);
+		mutex_lock(&adev->enforce_isolation_mutex);
+		dma_fence_put(isolation->spearhead);
+		isolation->spearhead = dma_fence_get(fence);
+		mutex_unlock(&adev->enforce_isolation_mutex);
+	}
+	dma_fence_put(fence);
+
+	amdgpu_ring_patch_cond_exec(ring, patch);
 
 	/* the double SWITCH_BUFFER here *cannot* be skipped by COND_EXEC */
 	if (ring->funcs->emit_switch_buffer) {
 		amdgpu_ring_emit_switch_buffer(ring);
 		amdgpu_ring_emit_switch_buffer(ring);
 	}
-	return 0;
-}
-
-/**
- * amdgpu_vm_reset_id - reset VMID to zero
- *
- * @adev: amdgpu device structure
- * @vm_id: vmid number to use
- *
- * Reset saved GDW, GWS and OA to force switch on next flush.
- */
-void amdgpu_vm_reset_id(struct amdgpu_device *adev, unsigned vmhub,
-			unsigned vmid)
-{
-	struct amdgpu_vm_id_manager *id_mgr = &adev->vm_manager.id_mgr[vmhub];
-	struct amdgpu_vm_id *id = &id_mgr->ids[vmid];
-
-	atomic64_set(&id->owner, 0);
-	id->gds_base = 0;
-	id->gds_size = 0;
-	id->gws_base = 0;
-	id->gws_size = 0;
-	id->oa_base = 0;
-	id->oa_size = 0;
-}
-
-/**
- * amdgpu_vm_reset_all_id - reset VMID to zero
- *
- * @adev: amdgpu device structure
- *
- * Reset VMID to force flush on next use
- */
-void amdgpu_vm_reset_all_ids(struct amdgpu_device *adev)
-{
-	unsigned i, j;
-
-	for (i = 0; i < AMDGPU_MAX_VMHUBS; ++i) {
-		struct amdgpu_vm_id_manager *id_mgr =
-			&adev->vm_manager.id_mgr[i];
 
-		for (j = 1; j < id_mgr->num_ids; ++j)
-			amdgpu_vm_reset_id(adev, i, j);
-	}
+	amdgpu_ring_ib_end(ring);
+	return 0;
 }
 
 /**
@@ -856,73 +903,22 @@ void amdgpu_vm_reset_all_ids(struct amdgpu_device *adev)
  * Returns the found bo_va or NULL if none is found
  *
  * Object has to be reserved!
+ *
+ * Returns:
+ * Found bo_va or NULL.
  */
 struct amdgpu_bo_va *amdgpu_vm_bo_find(struct amdgpu_vm *vm,
 				       struct amdgpu_bo *bo)
 {
-	struct amdgpu_bo_va *bo_va;
+	struct amdgpu_vm_bo_base *base;
 
-	list_for_each_entry(bo_va, &bo->va, base.bo_list) {
-		if (bo_va->base.vm == vm) {
-			return bo_va;
-		}
-	}
-	return NULL;
-}
-
-/**
- * amdgpu_vm_do_set_ptes - helper to call the right asic function
- *
- * @params: see amdgpu_pte_update_params definition
- * @pe: addr of the page entry
- * @addr: dst addr to write into pe
- * @count: number of page entries to update
- * @incr: increase next addr by incr bytes
- * @flags: hw access flags
- *
- * Traces the parameters and calls the right asic functions
- * to setup the page table using the DMA.
- */
-static void amdgpu_vm_do_set_ptes(struct amdgpu_pte_update_params *params,
-				  uint64_t pe, uint64_t addr,
-				  unsigned count, uint32_t incr,
-				  uint64_t flags)
-{
-	trace_amdgpu_vm_set_ptes(pe, addr, count, incr, flags);
-
-	if (count < 3) {
-		amdgpu_vm_write_pte(params->adev, params->ib, pe,
-				    addr | flags, count, incr);
+	for (base = bo->vm_bo; base; base = base->next) {
+		if (base->vm != vm)
+			continue;
 
-	} else {
-		amdgpu_vm_set_pte_pde(params->adev, params->ib, pe, addr,
-				      count, incr, flags);
+		return container_of(base, struct amdgpu_bo_va, base);
 	}
-}
-
-/**
- * amdgpu_vm_do_copy_ptes - copy the PTEs from the GART
- *
- * @params: see amdgpu_pte_update_params definition
- * @pe: addr of the page entry
- * @addr: dst addr to write into pe
- * @count: number of page entries to update
- * @incr: increase next addr by incr bytes
- * @flags: hw access flags
- *
- * Traces the parameters and calls the DMA function to copy the PTEs.
- */
-static void amdgpu_vm_do_copy_ptes(struct amdgpu_pte_update_params *params,
-				   uint64_t pe, uint64_t addr,
-				   unsigned count, uint32_t incr,
-				   uint64_t flags)
-{
-	uint64_t src = (params->src + (addr >> 12) * 8);
-
-
-	trace_amdgpu_vm_copy_ptes(pe, src, count);
-
-	amdgpu_vm_copy_pte(params->adev, params->ib, pe, src, count);
+	return NULL;
 }
 
 /**
@@ -932,9 +928,12 @@ static void amdgpu_vm_do_copy_ptes(struct amdgpu_pte_update_params *params,
  * @addr: the unmapped addr
  *
  * Look up the physical address of the page that the pte resolves
- * to and return the pointer for the page table entry.
+ * to.
+ *
+ * Returns:
+ * The pointer for the page table entry.
  */
-static uint64_t amdgpu_vm_map_gart(const dma_addr_t *pages_addr, uint64_t addr)
+uint64_t amdgpu_vm_map_gart(const dma_addr_t *pages_addr, uint64_t addr)
 {
 	uint64_t result;
 
@@ -950,764 +949,296 @@ static uint64_t amdgpu_vm_map_gart(const dma_addr_t *pages_addr, uint64_t addr)
 }
 
 /**
- * amdgpu_vm_cpu_set_ptes - helper to update page tables via CPU
+ * amdgpu_vm_update_pdes - make sure that all directories are valid
  *
- * @params: see amdgpu_pte_update_params definition
- * @pe: kmap addr of the page entry
- * @addr: dst addr to write into pe
- * @count: number of page entries to update
- * @incr: increase next addr by incr bytes
- * @flags: hw access flags
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
+ * @immediate: submit immediately to the paging queue
  *
- * Write count number of PT/PD entries directly.
+ * Makes sure all directories are up to date.
+ *
+ * Returns:
+ * 0 for success, error for failure.
  */
-static void amdgpu_vm_cpu_set_ptes(struct amdgpu_pte_update_params *params,
-				   uint64_t pe, uint64_t addr,
-				   unsigned count, uint32_t incr,
-				   uint64_t flags)
+int amdgpu_vm_update_pdes(struct amdgpu_device *adev,
+			  struct amdgpu_vm *vm, bool immediate)
 {
-	unsigned int i;
-	uint64_t value;
+	struct amdgpu_vm_update_params params;
+	struct amdgpu_vm_bo_base *entry;
+	bool flush_tlb_needed = false;
+	LIST_HEAD(relocated);
+	int r, idx;
 
-	trace_amdgpu_vm_set_ptes(pe, addr, count, incr, flags);
-
-	for (i = 0; i < count; i++) {
-		value = params->pages_addr ?
-			amdgpu_vm_map_gart(params->pages_addr, addr) :
-			addr;
-		amdgpu_gart_set_pte_pde(params->adev, (void *)(uintptr_t)pe,
-					i, value, flags);
-		addr += incr;
-	}
-}
+	amdgpu_vm_assert_locked(vm);
 
-static int amdgpu_vm_wait_pd(struct amdgpu_device *adev, struct amdgpu_vm *vm,
-			     void *owner)
-{
-	struct amdgpu_sync sync;
-	int r;
-
-	amdgpu_sync_create(&sync);
-	amdgpu_sync_resv(adev, &sync, vm->root.bo->tbo.resv, owner);
-	r = amdgpu_sync_wait(&sync, true);
-	amdgpu_sync_free(&sync);
-
-	return r;
-}
-
-/*
- * amdgpu_vm_update_level - update a single level in the hierarchy
- *
- * @adev: amdgpu_device pointer
- * @vm: requested vm
- * @parent: parent directory
- *
- * Makes sure all entries in @parent are up to date.
- * Returns 0 for success, error for failure.
- */
-static int amdgpu_vm_update_level(struct amdgpu_device *adev,
-				  struct amdgpu_vm *vm,
-				  struct amdgpu_vm_pt *parent,
-				  unsigned level)
-{
-	struct amdgpu_bo *shadow;
-	struct amdgpu_ring *ring = NULL;
-	uint64_t pd_addr, shadow_addr = 0;
-	uint32_t incr = amdgpu_vm_bo_size(adev, level + 1);
-	uint64_t last_pde = ~0, last_pt = ~0, last_shadow = ~0;
-	unsigned count = 0, pt_idx, ndw = 0;
-	struct amdgpu_job *job;
-	struct amdgpu_pte_update_params params;
-	struct dma_fence *fence = NULL;
-
-	int r;
+	spin_lock(&vm->status_lock);
+	list_splice_init(&vm->relocated, &relocated);
+	spin_unlock(&vm->status_lock);
 
-	if (!parent->entries)
+	if (list_empty(&relocated))
 		return 0;
 
+	if (!drm_dev_enter(adev_to_drm(adev), &idx))
+		return -ENODEV;
+
 	memset(&params, 0, sizeof(params));
 	params.adev = adev;
-	shadow = parent->bo->shadow;
-
-	if (vm->use_cpu_for_update) {
-		pd_addr = (unsigned long)amdgpu_bo_kptr(parent->bo);
-		r = amdgpu_vm_wait_pd(adev, vm, AMDGPU_FENCE_OWNER_VM);
-		if (unlikely(r))
-			return r;
-
-		params.func = amdgpu_vm_cpu_set_ptes;
-	} else {
-		ring = container_of(vm->entity.sched, struct amdgpu_ring,
-				    sched);
-
-		/* padding, etc. */
-		ndw = 64;
-
-		/* assume the worst case */
-		ndw += parent->last_entry_used * 6;
-
-		pd_addr = amdgpu_bo_gpu_offset(parent->bo);
-
-		if (shadow) {
-			shadow_addr = amdgpu_bo_gpu_offset(shadow);
-			ndw *= 2;
-		} else {
-			shadow_addr = 0;
-		}
-
-		r = amdgpu_job_alloc_with_ib(adev, ndw * 4, &job);
-		if (r)
-			return r;
-
-		params.ib = &job->ibs[0];
-		params.func = amdgpu_vm_do_set_ptes;
-	}
-
-
-	/* walk over the address space and update the directory */
-	for (pt_idx = 0; pt_idx <= parent->last_entry_used; ++pt_idx) {
-		struct amdgpu_bo *bo = parent->entries[pt_idx].bo;
-		uint64_t pde, pt;
-
-		if (bo == NULL)
-			continue;
-
-		pt = amdgpu_bo_gpu_offset(bo);
-		pt = amdgpu_gart_get_vm_pde(adev, pt);
-		/* Don't update huge pages here */
-		if ((parent->entries[pt_idx].addr & AMDGPU_PDE_PTE) ||
-		    parent->entries[pt_idx].addr == (pt | AMDGPU_PTE_VALID))
-			continue;
-
-		parent->entries[pt_idx].addr = pt | AMDGPU_PTE_VALID;
-
-		pde = pd_addr + pt_idx * 8;
-		if (((last_pde + 8 * count) != pde) ||
-		    ((last_pt + incr * count) != pt) ||
-		    (count == AMDGPU_VM_MAX_UPDATE_SIZE)) {
-
-			if (count) {
-				if (shadow)
-					params.func(&params,
-						    last_shadow,
-						    last_pt, count,
-						    incr,
-						    AMDGPU_PTE_VALID);
-
-				params.func(&params, last_pde,
-					    last_pt, count, incr,
-					    AMDGPU_PTE_VALID);
-			}
-
-			count = 1;
-			last_pde = pde;
-			last_shadow = shadow_addr + pt_idx * 8;
-			last_pt = pt;
-		} else {
-			++count;
-		}
-	}
-
-	if (count) {
-		if (vm->root.bo->shadow)
-			params.func(&params, last_shadow, last_pt,
-				    count, incr, AMDGPU_PTE_VALID);
-
-		params.func(&params, last_pde, last_pt,
-			    count, incr, AMDGPU_PTE_VALID);
-	}
-
-	if (!vm->use_cpu_for_update) {
-		if (params.ib->length_dw == 0) {
-			amdgpu_job_free(job);
-		} else {
-			amdgpu_ring_pad_ib(ring, params.ib);
-			amdgpu_sync_resv(adev, &job->sync, parent->bo->tbo.resv,
-					 AMDGPU_FENCE_OWNER_VM);
-			if (shadow)
-				amdgpu_sync_resv(adev, &job->sync,
-						 shadow->tbo.resv,
-						 AMDGPU_FENCE_OWNER_VM);
-
-			WARN_ON(params.ib->length_dw > ndw);
-			r = amdgpu_job_submit(job, ring, &vm->entity,
-					AMDGPU_FENCE_OWNER_VM, &fence);
-			if (r)
-				goto error_free;
+	params.vm = vm;
+	params.immediate = immediate;
 
-			amdgpu_bo_fence(parent->bo, fence, true);
-			dma_fence_put(vm->last_dir_update);
-			vm->last_dir_update = dma_fence_get(fence);
-			dma_fence_put(fence);
-		}
-	}
-	/*
-	 * Recurse into the subdirectories. This recursion is harmless because
-	 * we only have a maximum of 5 layers.
-	 */
-	for (pt_idx = 0; pt_idx <= parent->last_entry_used; ++pt_idx) {
-		struct amdgpu_vm_pt *entry = &parent->entries[pt_idx];
+	r = vm->update_funcs->prepare(&params, NULL,
+				      AMDGPU_KERNEL_JOB_ID_VM_UPDATE_PDES);
+	if (r)
+		goto error;
 
-		if (!entry->bo)
-			continue;
+	list_for_each_entry(entry, &relocated, vm_status) {
+		/* vm_flush_needed after updating moved PDEs */
+		flush_tlb_needed |= entry->moved;
 
-		r = amdgpu_vm_update_level(adev, vm, entry, level + 1);
+		r = amdgpu_vm_pde_update(&params, entry);
 		if (r)
-			return r;
-	}
-
-	return 0;
-
-error_free:
-	amdgpu_job_free(job);
-	return r;
-}
-
-/*
- * amdgpu_vm_invalidate_level - mark all PD levels as invalid
- *
- * @parent: parent PD
- *
- * Mark all PD level as invalid after an error.
- */
-static void amdgpu_vm_invalidate_level(struct amdgpu_vm_pt *parent)
-{
-	unsigned pt_idx;
-
-	/*
-	 * Recurse into the subdirectories. This recursion is harmless because
-	 * we only have a maximum of 5 layers.
-	 */
-	for (pt_idx = 0; pt_idx <= parent->last_entry_used; ++pt_idx) {
-		struct amdgpu_vm_pt *entry = &parent->entries[pt_idx];
-
-		if (!entry->bo)
-			continue;
-
-		entry->addr = ~0ULL;
-		amdgpu_vm_invalidate_level(entry);
+			goto error;
 	}
-}
 
-/*
- * amdgpu_vm_update_directories - make sure that all directories are valid
- *
- * @adev: amdgpu_device pointer
- * @vm: requested vm
- *
- * Makes sure all directories are up to date.
- * Returns 0 for success, error for failure.
- */
-int amdgpu_vm_update_directories(struct amdgpu_device *adev,
-				 struct amdgpu_vm *vm)
-{
-	int r;
-
-	r = amdgpu_vm_update_level(adev, vm, &vm->root, 0);
+	r = vm->update_funcs->commit(&params, &vm->last_update);
 	if (r)
-		amdgpu_vm_invalidate_level(&vm->root);
+		goto error;
 
-	if (vm->use_cpu_for_update) {
-		/* Flush HDP */
-		mb();
-		amdgpu_gart_flush_gpu_tlb(adev, 0);
+	if (flush_tlb_needed)
+		atomic64_inc(&vm->tlb_seq);
+
+	while (!list_empty(&relocated)) {
+		entry = list_first_entry(&relocated, struct amdgpu_vm_bo_base,
+					 vm_status);
+		amdgpu_vm_bo_idle(entry);
 	}
 
+error:
+	drm_dev_exit(idx);
 	return r;
 }
 
 /**
- * amdgpu_vm_find_entry - find the entry for an address
+ * amdgpu_vm_tlb_seq_cb - make sure to increment tlb sequence
+ * @fence: unused
+ * @cb: the callback structure
  *
- * @p: see amdgpu_pte_update_params definition
- * @addr: virtual address in question
- * @entry: resulting entry or NULL
- * @parent: parent entry
- *
- * Find the vm_pt entry and it's parent for the given address.
+ * Increments the tlb sequence to make sure that future CS execute a VM flush.
  */
-void amdgpu_vm_get_entry(struct amdgpu_pte_update_params *p, uint64_t addr,
-			 struct amdgpu_vm_pt **entry,
-			 struct amdgpu_vm_pt **parent)
+static void amdgpu_vm_tlb_seq_cb(struct dma_fence *fence,
+				 struct dma_fence_cb *cb)
 {
-	unsigned idx, level = p->adev->vm_manager.num_level;
-
-	*parent = NULL;
-	*entry = &p->vm->root;
-	while ((*entry)->entries) {
-		idx = addr >> (p->adev->vm_manager.block_size * level--);
-		idx %= amdgpu_bo_size((*entry)->bo) / 8;
-		*parent = *entry;
-		*entry = &(*entry)->entries[idx];
-	}
+	struct amdgpu_vm_tlb_seq_struct *tlb_cb;
 
-	if (level)
-		*entry = NULL;
+	tlb_cb = container_of(cb, typeof(*tlb_cb), cb);
+	atomic64_inc(&tlb_cb->vm->tlb_seq);
+	kfree(tlb_cb);
 }
 
 /**
- * amdgpu_vm_handle_huge_pages - handle updating the PD with huge pages
+ * amdgpu_vm_tlb_flush - prepare TLB flush
  *
- * @p: see amdgpu_pte_update_params definition
- * @entry: vm_pt entry to check
- * @parent: parent entry
- * @nptes: number of PTEs updated with this operation
- * @dst: destination address where the PTEs should point to
- * @flags: access flags fro the PTEs
+ * @params: parameters for update
+ * @fence: input fence to sync TLB flush with
+ * @tlb_cb: the callback structure
  *
- * Check if we can update the PD with a huge page.
+ * Increments the tlb sequence to make sure that future CS execute a VM flush.
  */
-static void amdgpu_vm_handle_huge_pages(struct amdgpu_pte_update_params *p,
-					struct amdgpu_vm_pt *entry,
-					struct amdgpu_vm_pt *parent,
-					unsigned nptes, uint64_t dst,
-					uint64_t flags)
+static void
+amdgpu_vm_tlb_flush(struct amdgpu_vm_update_params *params,
+		    struct dma_fence **fence,
+		    struct amdgpu_vm_tlb_seq_struct *tlb_cb)
 {
-	bool use_cpu_update = (p->func == amdgpu_vm_cpu_set_ptes);
-	uint64_t pd_addr, pde;
-
-	/* In the case of a mixed PT the PDE must point to it*/
-	if (p->adev->asic_type < CHIP_VEGA10 ||
-	    nptes != AMDGPU_VM_PTE_COUNT(p->adev) ||
-	    p->src ||
-	    !(flags & AMDGPU_PTE_VALID)) {
-
-		dst = amdgpu_bo_gpu_offset(entry->bo);
-		dst = amdgpu_gart_get_vm_pde(p->adev, dst);
-		flags = AMDGPU_PTE_VALID;
-	} else {
-		/* Set the huge page flag to stop scanning at this PDE */
-		flags |= AMDGPU_PDE_PTE;
-	}
+	struct amdgpu_vm *vm = params->vm;
 
-	if (entry->addr == (dst | flags))
+	tlb_cb->vm = vm;
+	if (!fence || !*fence) {
+		amdgpu_vm_tlb_seq_cb(NULL, &tlb_cb->cb);
 		return;
-
-	entry->addr = (dst | flags);
-
-	if (use_cpu_update) {
-		/* In case a huge page is replaced with a system
-		 * memory mapping, p->pages_addr != NULL and
-		 * amdgpu_vm_cpu_set_ptes would try to translate dst
-		 * through amdgpu_vm_map_gart. But dst is already a
-		 * GPU address (of the page table). Disable
-		 * amdgpu_vm_map_gart temporarily.
-		 */
-		dma_addr_t *tmp;
-
-		tmp = p->pages_addr;
-		p->pages_addr = NULL;
-
-		pd_addr = (unsigned long)amdgpu_bo_kptr(parent->bo);
-		pde = pd_addr + (entry - parent->entries) * 8;
-		amdgpu_vm_cpu_set_ptes(p, pde, dst, 1, 0, flags);
-
-		p->pages_addr = tmp;
-	} else {
-		if (parent->bo->shadow) {
-			pd_addr = amdgpu_bo_gpu_offset(parent->bo->shadow);
-			pde = pd_addr + (entry - parent->entries) * 8;
-			amdgpu_vm_do_set_ptes(p, pde, dst, 1, 0, flags);
-		}
-		pd_addr = amdgpu_bo_gpu_offset(parent->bo);
-		pde = pd_addr + (entry - parent->entries) * 8;
-		amdgpu_vm_do_set_ptes(p, pde, dst, 1, 0, flags);
 	}
-}
-
-/**
- * amdgpu_vm_update_ptes - make sure that page tables are valid
- *
- * @params: see amdgpu_pte_update_params definition
- * @vm: requested vm
- * @start: start of GPU address range
- * @end: end of GPU address range
- * @dst: destination address to map to, the next dst inside the function
- * @flags: mapping flags
- *
- * Update the page tables in the range @start - @end.
- * Returns 0 for success, -EINVAL for failure.
- */
-static int amdgpu_vm_update_ptes(struct amdgpu_pte_update_params *params,
-				  uint64_t start, uint64_t end,
-				  uint64_t dst, uint64_t flags)
-{
-	struct amdgpu_device *adev = params->adev;
-	const uint64_t mask = AMDGPU_VM_PTE_COUNT(adev) - 1;
-
-	uint64_t addr, pe_start;
-	struct amdgpu_bo *pt;
-	unsigned nptes;
-	bool use_cpu_update = (params->func == amdgpu_vm_cpu_set_ptes);
-
-	/* walk over the address space and update the page tables */
-	for (addr = start; addr < end; addr += nptes,
-	     dst += nptes * AMDGPU_GPU_PAGE_SIZE) {
-		struct amdgpu_vm_pt *entry, *parent;
-
-		amdgpu_vm_get_entry(params, addr, &entry, &parent);
-		if (!entry)
-			return -ENOENT;
-
-		if ((addr & ~mask) == (end & ~mask))
-			nptes = end - addr;
-		else
-			nptes = AMDGPU_VM_PTE_COUNT(adev) - (addr & mask);
-
-		amdgpu_vm_handle_huge_pages(params, entry, parent,
-					    nptes, dst, flags);
-		/* We don't need to update PTEs for huge pages */
-		if (entry->addr & AMDGPU_PDE_PTE)
-			continue;
-
-		pt = entry->bo;
-		if (use_cpu_update) {
-			pe_start = (unsigned long)amdgpu_bo_kptr(pt);
-		} else {
-			if (pt->shadow) {
-				pe_start = amdgpu_bo_gpu_offset(pt->shadow);
-				pe_start += (addr & mask) * 8;
-				params->func(params, pe_start, dst, nptes,
-					     AMDGPU_GPU_PAGE_SIZE, flags);
-			}
-			pe_start = amdgpu_bo_gpu_offset(pt);
-		}
-
-		pe_start += (addr & mask) * 8;
-		params->func(params, pe_start, dst, nptes,
-			     AMDGPU_GPU_PAGE_SIZE, flags);
-	}
-
-	return 0;
-}
-
-/*
- * amdgpu_vm_frag_ptes - add fragment information to PTEs
- *
- * @params: see amdgpu_pte_update_params definition
- * @vm: requested vm
- * @start: first PTE to handle
- * @end: last PTE to handle
- * @dst: addr those PTEs should point to
- * @flags: hw mapping flags
- * Returns 0 for success, -EINVAL for failure.
- */
-static int amdgpu_vm_frag_ptes(struct amdgpu_pte_update_params	*params,
-				uint64_t start, uint64_t end,
-				uint64_t dst, uint64_t flags)
-{
-	int r;
 
-	/**
-	 * The MC L1 TLB supports variable sized pages, based on a fragment
-	 * field in the PTE. When this field is set to a non-zero value, page
-	 * granularity is increased from 4KB to (1 << (12 + frag)). The PTE
-	 * flags are considered valid for all PTEs within the fragment range
-	 * and corresponding mappings are assumed to be physically contiguous.
-	 *
-	 * The L1 TLB can store a single PTE for the whole fragment,
-	 * significantly increasing the space available for translation
-	 * caching. This leads to large improvements in throughput when the
-	 * TLB is under pressure.
-	 *
-	 * The L2 TLB distributes small and large fragments into two
-	 * asymmetric partitions. The large fragment cache is significantly
-	 * larger. Thus, we try to use large fragments wherever possible.
-	 * Userspace can support this by aligning virtual base address and
-	 * allocation size to the fragment size.
-	 */
-	unsigned pages_per_frag = params->adev->vm_manager.fragment_size;
-	uint64_t frag_flags = AMDGPU_PTE_FRAG(pages_per_frag);
-	uint64_t frag_align = 1 << pages_per_frag;
-
-	uint64_t frag_start = ALIGN(start, frag_align);
-	uint64_t frag_end = end & ~(frag_align - 1);
-
-	/* system pages are non continuously */
-	if (params->src || !(flags & AMDGPU_PTE_VALID) ||
-	    (frag_start >= frag_end))
-		return amdgpu_vm_update_ptes(params, start, end, dst, flags);
-
-	/* handle the 4K area at the beginning */
-	if (start != frag_start) {
-		r = amdgpu_vm_update_ptes(params, start, frag_start,
-					  dst, flags);
-		if (r)
-			return r;
-		dst += (frag_start - start) * AMDGPU_GPU_PAGE_SIZE;
+	if (!dma_fence_add_callback(*fence, &tlb_cb->cb,
+				    amdgpu_vm_tlb_seq_cb)) {
+		dma_fence_put(vm->last_tlb_flush);
+		vm->last_tlb_flush = dma_fence_get(*fence);
+	} else {
+		amdgpu_vm_tlb_seq_cb(NULL, &tlb_cb->cb);
 	}
 
-	/* handle the area in the middle */
-	r = amdgpu_vm_update_ptes(params, frag_start, frag_end, dst,
-				  flags | frag_flags);
-	if (r)
-		return r;
+	/* Prepare a TLB flush fence to be attached to PTs */
+	if (!params->unlocked && vm->is_compute_context) {
+		amdgpu_vm_tlb_fence_create(params->adev, vm, fence);
 
-	/* handle the 4K area at the end */
-	if (frag_end != end) {
-		dst += (frag_end - frag_start) * AMDGPU_GPU_PAGE_SIZE;
-		r = amdgpu_vm_update_ptes(params, frag_end, end, dst, flags);
+		/* Makes sure no PD/PT is freed before the flush */
+		dma_resv_add_fence(vm->root.bo->tbo.base.resv, *fence,
+				   DMA_RESV_USAGE_BOOKKEEP);
 	}
-	return r;
 }
 
 /**
- * amdgpu_vm_bo_update_mapping - update a mapping in the vm page table
- *
- * @adev: amdgpu_device pointer
- * @exclusive: fence we need to sync to
- * @src: address where to copy page table entries from
- * @pages_addr: DMA addresses to use for mapping
- * @vm: requested vm
+ * amdgpu_vm_update_range - update a range in the vm page table
+ *
+ * @adev: amdgpu_device pointer to use for commands
+ * @vm: the VM to update the range
+ * @immediate: immediate submission in a page fault
+ * @unlocked: unlocked invalidation during MM callback
+ * @flush_tlb: trigger tlb invalidation after update completed
+ * @allow_override: change MTYPE for local NUMA nodes
+ * @sync: fences we need to sync to
  * @start: start of mapped range
  * @last: last mapped entry
  * @flags: flags for the entries
- * @addr: addr to set the area to
+ * @offset: offset into nodes and pages_addr
+ * @vram_base: base for vram mappings
+ * @res: ttm_resource to map
+ * @pages_addr: DMA addresses to use for mapping
  * @fence: optional resulting fence
  *
  * Fill in the page table entries between @start and @last.
- * Returns 0 for success, -EINVAL for failure.
+ *
+ * Returns:
+ * 0 for success, negative erro code for failure.
  */
-static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
-				       struct dma_fence *exclusive,
-				       uint64_t src,
-				       dma_addr_t *pages_addr,
-				       struct amdgpu_vm *vm,
-				       uint64_t start, uint64_t last,
-				       uint64_t flags, uint64_t addr,
-				       struct dma_fence **fence)
+int amdgpu_vm_update_range(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			   bool immediate, bool unlocked, bool flush_tlb,
+			   bool allow_override, struct amdgpu_sync *sync,
+			   uint64_t start, uint64_t last, uint64_t flags,
+			   uint64_t offset, uint64_t vram_base,
+			   struct ttm_resource *res, dma_addr_t *pages_addr,
+			   struct dma_fence **fence)
 {
-	struct amdgpu_ring *ring;
-	void *owner = AMDGPU_FENCE_OWNER_VM;
-	unsigned nptes, ncmds, ndw;
-	struct amdgpu_job *job;
-	struct amdgpu_pte_update_params params;
-	struct dma_fence *f = NULL;
-	int r;
-
-	memset(&params, 0, sizeof(params));
-	params.adev = adev;
-	params.vm = vm;
-	params.src = src;
+	struct amdgpu_vm_tlb_seq_struct *tlb_cb;
+	struct amdgpu_vm_update_params params;
+	struct amdgpu_res_cursor cursor;
+	int r, idx;
 
-	/* sync to everything on unmapping */
-	if (!(flags & AMDGPU_PTE_VALID))
-		owner = AMDGPU_FENCE_OWNER_UNDEFINED;
+	if (!drm_dev_enter(adev_to_drm(adev), &idx))
+		return -ENODEV;
 
-	if (vm->use_cpu_for_update) {
-		/* params.src is used as flag to indicate system Memory */
-		if (pages_addr)
-			params.src = ~0;
-
-		/* Wait for PT BOs to be free. PTs share the same resv. object
-		 * as the root PD BO
-		 */
-		r = amdgpu_vm_wait_pd(adev, vm, owner);
-		if (unlikely(r))
-			return r;
-
-		params.func = amdgpu_vm_cpu_set_ptes;
-		params.pages_addr = pages_addr;
-		return amdgpu_vm_frag_ptes(&params, start, last + 1,
-					   addr, flags);
+	tlb_cb = kmalloc(sizeof(*tlb_cb), GFP_KERNEL);
+	if (!tlb_cb) {
+		drm_dev_exit(idx);
+		return -ENOMEM;
 	}
 
-	ring = container_of(vm->entity.sched, struct amdgpu_ring, sched);
-
-	nptes = last - start + 1;
+	/* Vega20+XGMI where PTEs get inadvertently cached in L2 texture cache,
+	 * heavy-weight flush TLB unconditionally.
+	 */
+	flush_tlb |= adev->gmc.xgmi.num_physical_nodes &&
+		     amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 0);
 
 	/*
-	 * reserve space for one command every (1 << BLOCK_SIZE)
-	 *  entries or 2k dwords (whatever is smaller)
+	 * On GFX8 and older any 8 PTE block with a valid bit set enters the TLB
 	 */
-	ncmds = (nptes >> min(adev->vm_manager.block_size, 11u)) + 1;
+	flush_tlb |= amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(9, 0, 0);
 
-	/* padding, etc. */
-	ndw = 64;
-
-	/* one PDE write for each huge page */
-	ndw += ((nptes >> adev->vm_manager.block_size) + 1) * 6;
-
-	if (src) {
-		/* only copy commands needed */
-		ndw += ncmds * 7;
-
-		params.func = amdgpu_vm_do_copy_ptes;
+	memset(&params, 0, sizeof(params));
+	params.adev = adev;
+	params.vm = vm;
+	params.immediate = immediate;
+	params.pages_addr = pages_addr;
+	params.unlocked = unlocked;
+	params.needs_flush = flush_tlb;
+	params.allow_override = allow_override;
+	INIT_LIST_HEAD(&params.tlb_flush_waitlist);
+
+	amdgpu_vm_eviction_lock(vm);
+	if (vm->evicting) {
+		r = -EBUSY;
+		goto error_free;
+	}
 
-	} else if (pages_addr) {
-		/* copy commands needed */
-		ndw += ncmds * 7;
+	if (!unlocked && !dma_fence_is_signaled(vm->last_unlocked)) {
+		struct dma_fence *tmp = dma_fence_get_stub();
 
-		/* and also PTEs */
-		ndw += nptes * 2;
+		amdgpu_bo_fence(vm->root.bo, vm->last_unlocked, true);
+		swap(vm->last_unlocked, tmp);
+		dma_fence_put(tmp);
+	}
 
-		params.func = amdgpu_vm_do_copy_ptes;
+	r = vm->update_funcs->prepare(&params, sync,
+				      AMDGPU_KERNEL_JOB_ID_VM_UPDATE_RANGE);
+	if (r)
+		goto error_free;
 
-	} else {
-		/* set page commands needed */
-		ndw += ncmds * 10;
+	amdgpu_res_first(pages_addr ? NULL : res, offset,
+			 (last - start + 1) * AMDGPU_GPU_PAGE_SIZE, &cursor);
+	while (cursor.remaining) {
+		uint64_t tmp, num_entries, addr;
 
-		/* two extra commands for begin/end of fragment */
-		ndw += 2 * 10;
+		num_entries = cursor.size >> AMDGPU_GPU_PAGE_SHIFT;
+		if (pages_addr) {
+			bool contiguous = true;
 
-		params.func = amdgpu_vm_do_set_ptes;
-	}
+			if (num_entries > AMDGPU_GPU_PAGES_IN_CPU_PAGE) {
+				uint64_t pfn = cursor.start >> PAGE_SHIFT;
+				uint64_t count;
 
-	r = amdgpu_job_alloc_with_ib(adev, ndw * 4, &job);
-	if (r)
-		return r;
+				contiguous = pages_addr[pfn + 1] ==
+					pages_addr[pfn] + PAGE_SIZE;
 
-	params.ib = &job->ibs[0];
+				tmp = num_entries /
+					AMDGPU_GPU_PAGES_IN_CPU_PAGE;
+				for (count = 2; count < tmp; ++count) {
+					uint64_t idx = pfn + count;
 
-	if (!src && pages_addr) {
-		uint64_t *pte;
-		unsigned i;
+					if (contiguous != (pages_addr[idx] ==
+					    pages_addr[idx - 1] + PAGE_SIZE))
+						break;
+				}
+				if (!contiguous)
+					count--;
+				num_entries = count *
+					AMDGPU_GPU_PAGES_IN_CPU_PAGE;
+			}
 
-		/* Put the PTEs at the end of the IB. */
-		i = ndw - nptes * 2;
-		pte= (uint64_t *)&(job->ibs->ptr[i]);
-		params.src = job->ibs->gpu_addr + i * 4;
+			if (!contiguous) {
+				addr = cursor.start;
+				params.pages_addr = pages_addr;
+			} else {
+				addr = pages_addr[cursor.start >> PAGE_SHIFT];
+				params.pages_addr = NULL;
+			}
 
-		for (i = 0; i < nptes; ++i) {
-			pte[i] = amdgpu_vm_map_gart(pages_addr, addr + i *
-						    AMDGPU_GPU_PAGE_SIZE);
-			pte[i] |= flags;
+		} else if (flags & (AMDGPU_PTE_VALID | AMDGPU_PTE_PRT_FLAG(adev))) {
+			addr = vram_base + cursor.start;
+		} else {
+			addr = 0;
 		}
-		addr = 0;
-	}
 
-	r = amdgpu_sync_fence(adev, &job->sync, exclusive);
-	if (r)
-		goto error_free;
-
-	r = amdgpu_sync_resv(adev, &job->sync, vm->root.bo->tbo.resv,
-			     owner);
-	if (r)
-		goto error_free;
+		tmp = start + num_entries;
+		r = amdgpu_vm_ptes_update(&params, start, tmp, addr, flags);
+		if (r)
+			goto error_free;
 
-	r = reservation_object_reserve_shared(vm->root.bo->tbo.resv);
-	if (r)
-		goto error_free;
+		amdgpu_res_next(&cursor, num_entries * AMDGPU_GPU_PAGE_SIZE);
+		start = tmp;
+	}
 
-	r = amdgpu_vm_frag_ptes(&params, start, last + 1, addr, flags);
+	r = vm->update_funcs->commit(&params, fence);
 	if (r)
 		goto error_free;
 
-	amdgpu_ring_pad_ib(ring, params.ib);
-	WARN_ON(params.ib->length_dw > ndw);
-	r = amdgpu_job_submit(job, ring, &vm->entity,
-			      AMDGPU_FENCE_OWNER_VM, &f);
-	if (r)
-		goto error_free;
+	if (params.needs_flush) {
+		amdgpu_vm_tlb_flush(&params, fence, tlb_cb);
+		tlb_cb = NULL;
+	}
 
-	amdgpu_bo_fence(vm->root.bo, f, true);
-	dma_fence_put(*fence);
-	*fence = f;
-	return 0;
+	amdgpu_vm_pt_free_list(adev, &params);
 
 error_free:
-	amdgpu_job_free(job);
-	amdgpu_vm_invalidate_level(&vm->root);
+	kfree(tlb_cb);
+	amdgpu_vm_eviction_unlock(vm);
+	drm_dev_exit(idx);
 	return r;
 }
 
-/**
- * amdgpu_vm_bo_split_mapping - split a mapping into smaller chunks
- *
- * @adev: amdgpu_device pointer
- * @exclusive: fence we need to sync to
- * @pages_addr: DMA addresses to use for mapping
- * @vm: requested vm
- * @mapping: mapped range and flags to use for the update
- * @flags: HW flags for the mapping
- * @nodes: array of drm_mm_nodes with the MC addresses
- * @fence: optional resulting fence
- *
- * Split the mapping into smaller chunks so that each update fits
- * into a SDMA IB.
- * Returns 0 for success, -EINVAL for failure.
- */
-static int amdgpu_vm_bo_split_mapping(struct amdgpu_device *adev,
-				      struct dma_fence *exclusive,
-				      dma_addr_t *pages_addr,
-				      struct amdgpu_vm *vm,
-				      struct amdgpu_bo_va_mapping *mapping,
-				      uint64_t flags,
-				      struct drm_mm_node *nodes,
-				      struct dma_fence **fence)
+void amdgpu_vm_get_memory(struct amdgpu_vm *vm,
+			  struct amdgpu_mem_stats stats[__AMDGPU_PL_NUM])
 {
-	uint64_t pfn, src = 0, start = mapping->start;
-	int r;
-
-	/* normally,bo_va->flags only contians READABLE and WIRTEABLE bit go here
-	 * but in case of something, we filter the flags in first place
-	 */
-	if (!(mapping->flags & AMDGPU_PTE_READABLE))
-		flags &= ~AMDGPU_PTE_READABLE;
-	if (!(mapping->flags & AMDGPU_PTE_WRITEABLE))
-		flags &= ~AMDGPU_PTE_WRITEABLE;
-
-	flags &= ~AMDGPU_PTE_EXECUTABLE;
-	flags |= mapping->flags & AMDGPU_PTE_EXECUTABLE;
-
-	flags &= ~AMDGPU_PTE_MTYPE_MASK;
-	flags |= (mapping->flags & AMDGPU_PTE_MTYPE_MASK);
-
-	if ((mapping->flags & AMDGPU_PTE_PRT) &&
-	    (adev->asic_type >= CHIP_VEGA10)) {
-		flags |= AMDGPU_PTE_PRT;
-		flags &= ~AMDGPU_PTE_VALID;
-	}
-
-	trace_amdgpu_vm_bo_update(mapping);
-
-	pfn = mapping->offset >> PAGE_SHIFT;
-	if (nodes) {
-		while (pfn >= nodes->size) {
-			pfn -= nodes->size;
-			++nodes;
-		}
-	}
-
-	do {
-		uint64_t max_entries;
-		uint64_t addr, last;
-
-		if (nodes) {
-			addr = nodes->start << PAGE_SHIFT;
-			max_entries = (nodes->size - pfn) *
-				(PAGE_SIZE / AMDGPU_GPU_PAGE_SIZE);
-		} else {
-			addr = 0;
-			max_entries = S64_MAX;
-		}
-
-		if (pages_addr) {
-			max_entries = min(max_entries, 16ull * 1024ull);
-			addr = 0;
-		} else if (flags & AMDGPU_PTE_VALID) {
-			addr += adev->vm_manager.vram_base_offset;
-		}
-		addr += pfn << PAGE_SHIFT;
-
-		last = min((uint64_t)mapping->last, start + max_entries - 1);
-		r = amdgpu_vm_bo_update_mapping(adev, exclusive,
-						src, pages_addr, vm,
-						start, last, flags, addr,
-						fence);
-		if (r)
-			return r;
-
-		pfn += last - start + 1;
-		if (nodes && nodes->size == pfn) {
-			pfn = 0;
-			++nodes;
-		}
-		start = last + 1;
-
-	} while (unlikely(start != mapping->last + 1));
-
-	return 0;
+	spin_lock(&vm->status_lock);
+	memcpy(stats, vm->stats, sizeof(*stats) * __AMDGPU_PL_NUM);
+	spin_unlock(&vm->status_lock);
 }
 
 /**
@@ -1718,83 +1249,160 @@ static int amdgpu_vm_bo_split_mapping(struct amdgpu_device *adev,
  * @clear: if true clear the entries
  *
  * Fill in the page table entries for @bo_va.
- * Returns 0 for success, -EINVAL for failure.
+ *
+ * Returns:
+ * 0 for success, -EINVAL for failure.
  */
-int amdgpu_vm_bo_update(struct amdgpu_device *adev,
-			struct amdgpu_bo_va *bo_va,
+int amdgpu_vm_bo_update(struct amdgpu_device *adev, struct amdgpu_bo_va *bo_va,
 			bool clear)
 {
 	struct amdgpu_bo *bo = bo_va->base.bo;
 	struct amdgpu_vm *vm = bo_va->base.vm;
 	struct amdgpu_bo_va_mapping *mapping;
+	struct dma_fence **last_update;
 	dma_addr_t *pages_addr = NULL;
-	struct ttm_mem_reg *mem;
-	struct drm_mm_node *nodes;
-	struct dma_fence *exclusive;
+	struct ttm_resource *mem;
+	struct amdgpu_sync sync;
+	bool flush_tlb = clear;
+	uint64_t vram_base;
 	uint64_t flags;
+	bool uncached;
 	int r;
 
-	if (clear || !bo_va->base.bo) {
+	amdgpu_sync_create(&sync);
+	if (clear) {
+		mem = NULL;
+
+		/* Implicitly sync to command submissions in the same VM before
+		 * unmapping.
+		 */
+		r = amdgpu_sync_resv(adev, &sync, vm->root.bo->tbo.base.resv,
+				     AMDGPU_SYNC_EQ_OWNER, vm);
+		if (r)
+			goto error_free;
+		if (bo) {
+			r = amdgpu_sync_kfd(&sync, bo->tbo.base.resv);
+			if (r)
+				goto error_free;
+		}
+	} else if (!bo) {
 		mem = NULL;
-		nodes = NULL;
-		exclusive = NULL;
+
+		/* PRT map operations don't need to sync to anything. */
+
 	} else {
-		struct ttm_dma_tt *ttm;
-
-		mem = &bo_va->base.bo->tbo.mem;
-		nodes = mem->mm_node;
-		if (mem->mem_type == TTM_PL_TT) {
-			ttm = container_of(bo_va->base.bo->tbo.ttm,
-					   struct ttm_dma_tt, ttm);
-			pages_addr = ttm->dma_address;
+		struct drm_gem_object *obj = &bo->tbo.base;
+
+		if (drm_gem_is_imported(obj) && bo_va->is_xgmi) {
+			struct dma_buf *dma_buf = obj->import_attach->dmabuf;
+			struct drm_gem_object *gobj = dma_buf->priv;
+			struct amdgpu_bo *abo = gem_to_amdgpu_bo(gobj);
+
+			if (abo->tbo.resource &&
+			    abo->tbo.resource->mem_type == TTM_PL_VRAM)
+				bo = gem_to_amdgpu_bo(gobj);
 		}
-		exclusive = reservation_object_get_excl(bo->tbo.resv);
+		mem = bo->tbo.resource;
+		if (mem && (mem->mem_type == TTM_PL_TT ||
+			    mem->mem_type == AMDGPU_PL_PREEMPT))
+			pages_addr = bo->tbo.ttm->dma_address;
+
+		/* Implicitly sync to moving fences before mapping anything */
+		r = amdgpu_sync_resv(adev, &sync, bo->tbo.base.resv,
+				     AMDGPU_SYNC_EXPLICIT, vm);
+		if (r)
+			goto error_free;
 	}
 
-	if (bo)
+	if (bo) {
+		struct amdgpu_device *bo_adev;
+
 		flags = amdgpu_ttm_tt_pte_flags(adev, bo->tbo.ttm, mem);
-	else
+
+		if (amdgpu_bo_encrypted(bo))
+			flags |= AMDGPU_PTE_TMZ;
+
+		bo_adev = amdgpu_ttm_adev(bo->tbo.bdev);
+		vram_base = bo_adev->vm_manager.vram_base_offset;
+		uncached = (bo->flags & AMDGPU_GEM_CREATE_UNCACHED) != 0;
+	} else {
 		flags = 0x0;
+		vram_base = 0;
+		uncached = false;
+	}
 
-	spin_lock(&vm->status_lock);
-	if (!list_empty(&bo_va->base.vm_status))
+	if (clear || amdgpu_vm_is_bo_always_valid(vm, bo))
+		last_update = &vm->last_update;
+	else
+		last_update = &bo_va->last_pt_update;
+
+	if (!clear && bo_va->base.moved) {
+		flush_tlb = true;
 		list_splice_init(&bo_va->valids, &bo_va->invalids);
-	spin_unlock(&vm->status_lock);
+
+	} else if (bo_va->cleared != clear) {
+		list_splice_init(&bo_va->valids, &bo_va->invalids);
+	}
 
 	list_for_each_entry(mapping, &bo_va->invalids, list) {
-		r = amdgpu_vm_bo_split_mapping(adev, exclusive, pages_addr, vm,
-					       mapping, flags, nodes,
-					       &bo_va->last_pt_update);
+		uint64_t update_flags = flags;
+
+		/* normally,bo_va->flags only contians READABLE and WIRTEABLE bit go here
+		 * but in case of something, we filter the flags in first place
+		 */
+		if (!(mapping->flags & AMDGPU_VM_PAGE_READABLE))
+			update_flags &= ~AMDGPU_PTE_READABLE;
+		if (!(mapping->flags & AMDGPU_VM_PAGE_WRITEABLE))
+			update_flags &= ~AMDGPU_PTE_WRITEABLE;
+
+		/* Apply ASIC specific mapping flags */
+		amdgpu_gmc_get_vm_pte(adev, vm, bo, mapping->flags,
+				      &update_flags);
+
+		trace_amdgpu_vm_bo_update(mapping);
+
+		r = amdgpu_vm_update_range(adev, vm, false, false, flush_tlb,
+					   !uncached, &sync, mapping->start,
+					   mapping->last, update_flags,
+					   mapping->offset, vram_base, mem,
+					   pages_addr, last_update);
 		if (r)
-			return r;
+			goto error_free;
 	}
 
-	if (trace_amdgpu_vm_bo_mapping_enabled()) {
-		list_for_each_entry(mapping, &bo_va->valids, list)
-			trace_amdgpu_vm_bo_mapping(mapping);
-
-		list_for_each_entry(mapping, &bo_va->invalids, list)
-			trace_amdgpu_vm_bo_mapping(mapping);
+	/* If the BO is not in its preferred location add it back to
+	 * the evicted list so that it gets validated again on the
+	 * next command submission.
+	 */
+	if (amdgpu_vm_is_bo_always_valid(vm, bo)) {
+		if (bo->tbo.resource &&
+		    !(bo->preferred_domains &
+		      amdgpu_mem_type_to_domain(bo->tbo.resource->mem_type)))
+			amdgpu_vm_bo_evicted(&bo_va->base);
+		else
+			amdgpu_vm_bo_idle(&bo_va->base);
+	} else {
+		amdgpu_vm_bo_done(&bo_va->base);
 	}
 
-	spin_lock(&vm->status_lock);
 	list_splice_init(&bo_va->invalids, &bo_va->valids);
-	list_del_init(&bo_va->base.vm_status);
-	if (clear)
-		list_add(&bo_va->base.vm_status, &vm->cleared);
-	spin_unlock(&vm->status_lock);
+	bo_va->cleared = clear;
+	bo_va->base.moved = false;
 
-	if (vm->use_cpu_for_update) {
-		/* Flush HDP */
-		mb();
-		amdgpu_gart_flush_gpu_tlb(adev, 0);
+	if (trace_amdgpu_vm_bo_mapping_enabled()) {
+		list_for_each_entry(mapping, &bo_va->valids, list)
+			trace_amdgpu_vm_bo_mapping(mapping);
 	}
 
-	return 0;
+error_free:
+	amdgpu_sync_free(&sync);
+	return r;
 }
 
 /**
  * amdgpu_vm_update_prt_state - update the global PRT state
+ *
+ * @adev: amdgpu_device pointer
  */
 static void amdgpu_vm_update_prt_state(struct amdgpu_device *adev)
 {
@@ -1803,16 +1411,18 @@ static void amdgpu_vm_update_prt_state(struct amdgpu_device *adev)
 
 	spin_lock_irqsave(&adev->vm_manager.prt_lock, flags);
 	enable = !!atomic_read(&adev->vm_manager.num_prt_users);
-	adev->gart.gart_funcs->set_prt(adev, enable);
+	adev->gmc.gmc_funcs->set_prt(adev, enable);
 	spin_unlock_irqrestore(&adev->vm_manager.prt_lock, flags);
 }
 
 /**
  * amdgpu_vm_prt_get - add a PRT user
+ *
+ * @adev: amdgpu_device pointer
  */
 static void amdgpu_vm_prt_get(struct amdgpu_device *adev)
 {
-	if (!adev->gart.gart_funcs->set_prt)
+	if (!adev->gmc.gmc_funcs->set_prt)
 		return;
 
 	if (atomic_inc_return(&adev->vm_manager.num_prt_users) == 1)
@@ -1821,6 +1431,8 @@ static void amdgpu_vm_prt_get(struct amdgpu_device *adev)
 
 /**
  * amdgpu_vm_prt_put - drop a PRT user
+ *
+ * @adev: amdgpu_device pointer
  */
 static void amdgpu_vm_prt_put(struct amdgpu_device *adev)
 {
@@ -1830,6 +1442,9 @@ static void amdgpu_vm_prt_put(struct amdgpu_device *adev)
 
 /**
  * amdgpu_vm_prt_cb - callback for updating the PRT status
+ *
+ * @fence: fence for the callback
+ * @_cb: the callback function
  */
 static void amdgpu_vm_prt_cb(struct dma_fence *fence, struct dma_fence_cb *_cb)
 {
@@ -1841,13 +1456,16 @@ static void amdgpu_vm_prt_cb(struct dma_fence *fence, struct dma_fence_cb *_cb)
 
 /**
  * amdgpu_vm_add_prt_cb - add callback for updating the PRT status
+ *
+ * @adev: amdgpu_device pointer
+ * @fence: fence for the callback
  */
 static void amdgpu_vm_add_prt_cb(struct amdgpu_device *adev,
 				 struct dma_fence *fence)
 {
 	struct amdgpu_prt_cb *cb;
 
-	if (!adev->gart.gart_funcs->set_prt)
+	if (!adev->gmc.gmc_funcs->set_prt)
 		return;
 
 	cb = kmalloc(sizeof(struct amdgpu_prt_cb), GFP_KERNEL);
@@ -1880,7 +1498,7 @@ static void amdgpu_vm_free_mapping(struct amdgpu_device *adev,
 				   struct amdgpu_bo_va_mapping *mapping,
 				   struct dma_fence *fence)
 {
-	if (mapping->flags & AMDGPU_PTE_PRT)
+	if (mapping->flags & AMDGPU_VM_PAGE_PRT)
 		amdgpu_vm_add_prt_cb(adev, fence);
 	kfree(mapping);
 }
@@ -1895,32 +1513,15 @@ static void amdgpu_vm_free_mapping(struct amdgpu_device *adev,
  */
 static void amdgpu_vm_prt_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 {
-	struct reservation_object *resv = vm->root.bo->tbo.resv;
-	struct dma_fence *excl, **shared;
-	unsigned i, shared_count;
-	int r;
+	struct dma_resv *resv = vm->root.bo->tbo.base.resv;
+	struct dma_resv_iter cursor;
+	struct dma_fence *fence;
 
-	r = reservation_object_get_fences_rcu(resv, &excl,
-					      &shared_count, &shared);
-	if (r) {
-		/* Not enough memory to grab the fence list, as last resort
-		 * block for all the fences to complete.
-		 */
-		reservation_object_wait_timeout_rcu(resv, true, false,
-						    MAX_SCHEDULE_TIMEOUT);
-		return;
-	}
-
-	/* Add a callback for each fence in the reservation object */
-	amdgpu_vm_prt_get(adev);
-	amdgpu_vm_add_prt_cb(adev, excl);
-
-	for (i = 0; i < shared_count; ++i) {
+	dma_resv_for_each_fence(&cursor, resv, DMA_RESV_USAGE_BOOKKEEP, fence) {
+		/* Add a callback for each fence in the reservation object */
 		amdgpu_vm_prt_get(adev);
-		amdgpu_vm_add_prt_cb(adev, shared[i]);
+		amdgpu_vm_add_prt_cb(adev, fence);
 	}
-
-	kfree(shared);
 }
 
 /**
@@ -1932,9 +1533,11 @@ static void amdgpu_vm_prt_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm)
  * or if an error occurred)
  *
  * Make sure all freed BOs are cleared in the PT.
- * Returns 0 for success.
- *
  * PTs have to be reserved and mutex must be locked!
+ *
+ * Returns:
+ * 0 for success.
+ *
  */
 int amdgpu_vm_clear_freed(struct amdgpu_device *adev,
 			  struct amdgpu_vm *vm,
@@ -1942,24 +1545,32 @@ int amdgpu_vm_clear_freed(struct amdgpu_device *adev,
 {
 	struct amdgpu_bo_va_mapping *mapping;
 	struct dma_fence *f = NULL;
+	struct amdgpu_sync sync;
 	int r;
-	uint64_t init_pte_value = 0;
+
+
+	/*
+	 * Implicitly sync to command submissions in the same VM before
+	 * unmapping.
+	 */
+	amdgpu_sync_create(&sync);
+	r = amdgpu_sync_resv(adev, &sync, vm->root.bo->tbo.base.resv,
+			     AMDGPU_SYNC_EQ_OWNER, vm);
+	if (r)
+		goto error_free;
 
 	while (!list_empty(&vm->freed)) {
 		mapping = list_first_entry(&vm->freed,
 			struct amdgpu_bo_va_mapping, list);
 		list_del(&mapping->list);
 
-		if (vm->pte_support_ats)
-			init_pte_value = AMDGPU_PTE_SYSTEM;
-
-		r = amdgpu_vm_bo_update_mapping(adev, NULL, 0, NULL, vm,
-						mapping->start, mapping->last,
-						init_pte_value, 0, &f);
+		r = amdgpu_vm_update_range(adev, vm, false, false, true, false,
+					   &sync, mapping->start, mapping->last,
+					   0, 0, 0, NULL, NULL, &f);
 		amdgpu_vm_free_mapping(adev, vm, mapping, f);
 		if (r) {
 			dma_fence_put(f);
-			return r;
+			goto error_free;
 		}
 	}
 
@@ -1970,44 +1581,133 @@ int amdgpu_vm_clear_freed(struct amdgpu_device *adev,
 		dma_fence_put(f);
 	}
 
-	return 0;
+error_free:
+	amdgpu_sync_free(&sync);
+	return r;
 
 }
 
 /**
- * amdgpu_vm_clear_moved - clear moved BOs in the PT
+ * amdgpu_vm_handle_moved - handle moved BOs in the PT
  *
  * @adev: amdgpu_device pointer
  * @vm: requested vm
+ * @ticket: optional reservation ticket used to reserve the VM
  *
- * Make sure all moved BOs are cleared in the PT.
- * Returns 0 for success.
+ * Make sure all BOs which are moved are updated in the PTs.
  *
- * PTs have to be reserved and mutex must be locked!
+ * Returns:
+ * 0 for success.
+ *
+ * PTs have to be reserved!
  */
-int amdgpu_vm_clear_moved(struct amdgpu_device *adev, struct amdgpu_vm *vm,
-			    struct amdgpu_sync *sync)
+int amdgpu_vm_handle_moved(struct amdgpu_device *adev,
+			   struct amdgpu_vm *vm,
+			   struct ww_acquire_ctx *ticket)
 {
-	struct amdgpu_bo_va *bo_va = NULL;
-	int r = 0;
+	struct amdgpu_bo_va *bo_va;
+	struct dma_resv *resv;
+	bool clear, unlock;
+	int r;
 
 	spin_lock(&vm->status_lock);
 	while (!list_empty(&vm->moved)) {
-		bo_va = list_first_entry(&vm->moved,
-			struct amdgpu_bo_va, base.vm_status);
+		bo_va = list_first_entry(&vm->moved, struct amdgpu_bo_va,
+					 base.vm_status);
 		spin_unlock(&vm->status_lock);
 
-		r = amdgpu_vm_bo_update(adev, bo_va, true);
+		/* Per VM BOs never need to bo cleared in the page tables */
+		r = amdgpu_vm_bo_update(adev, bo_va, false);
 		if (r)
 			return r;
+		spin_lock(&vm->status_lock);
+	}
+
+	while (!list_empty(&vm->invalidated)) {
+		bo_va = list_first_entry(&vm->invalidated, struct amdgpu_bo_va,
+					 base.vm_status);
+		resv = bo_va->base.bo->tbo.base.resv;
+		spin_unlock(&vm->status_lock);
+
+		/* Try to reserve the BO to avoid clearing its ptes */
+		if (!adev->debug_vm && dma_resv_trylock(resv)) {
+			clear = false;
+			unlock = true;
+		/* The caller is already holding the reservation lock */
+		} else if (ticket && dma_resv_locking_ctx(resv) == ticket) {
+			clear = false;
+			unlock = false;
+		/* Somebody else is using the BO right now */
+		} else {
+			clear = true;
+			unlock = false;
+		}
+
+		r = amdgpu_vm_bo_update(adev, bo_va, clear);
+
+		if (unlock)
+			dma_resv_unlock(resv);
+		if (r)
+			return r;
+
+		/* Remember evicted DMABuf imports in compute VMs for later
+		 * validation
+		 */
+		if (vm->is_compute_context &&
+		    drm_gem_is_imported(&bo_va->base.bo->tbo.base) &&
+		    (!bo_va->base.bo->tbo.resource ||
+		     bo_va->base.bo->tbo.resource->mem_type == TTM_PL_SYSTEM))
+			amdgpu_vm_bo_evicted_user(&bo_va->base);
 
 		spin_lock(&vm->status_lock);
 	}
 	spin_unlock(&vm->status_lock);
 
-	if (bo_va)
-		r = amdgpu_sync_fence(adev, sync, bo_va->last_pt_update);
+	return 0;
+}
+
+/**
+ * amdgpu_vm_flush_compute_tlb - Flush TLB on compute VM
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
+ * @flush_type: flush type
+ * @xcc_mask: mask of XCCs that belong to the compute partition in need of a TLB flush.
+ *
+ * Flush TLB if needed for a compute VM.
+ *
+ * Returns:
+ * 0 for success.
+ */
+int amdgpu_vm_flush_compute_tlb(struct amdgpu_device *adev,
+				struct amdgpu_vm *vm,
+				uint32_t flush_type,
+				uint32_t xcc_mask)
+{
+	uint64_t tlb_seq = amdgpu_vm_tlb_seq(vm);
+	bool all_hub = false;
+	int xcc = 0, r = 0;
+
+	WARN_ON_ONCE(!vm->is_compute_context);
 
+	/*
+	 * It can be that we race and lose here, but that is extremely unlikely
+	 * and the worst thing which could happen is that we flush the changes
+	 * into the TLB once more which is harmless.
+	 */
+	if (atomic64_xchg(&vm->kfd_last_flushed_seq, tlb_seq) == tlb_seq)
+		return 0;
+
+	if (adev->family == AMDGPU_FAMILY_AI ||
+	    adev->family == AMDGPU_FAMILY_RV)
+		all_hub = true;
+
+	for_each_inst(xcc, xcc_mask) {
+		r = amdgpu_gmc_flush_gpu_tlb_pasid(adev, vm->pasid, flush_type,
+						   all_hub, xcc);
+		if (r)
+			break;
+	}
 	return r;
 }
 
@@ -2020,7 +1720,9 @@ int amdgpu_vm_clear_moved(struct amdgpu_device *adev, struct amdgpu_vm *vm,
  *
  * Add @bo into the requested vm.
  * Add @bo to the list of bos associated with the vm
- * Returns newly added bo_va or NULL for failure
+ *
+ * Returns:
+ * Newly added bo_va or NULL for failure
  *
  * Object has to be reserved!
  */
@@ -2034,21 +1736,87 @@ struct amdgpu_bo_va *amdgpu_vm_bo_add(struct amdgpu_device *adev,
 	if (bo_va == NULL) {
 		return NULL;
 	}
-	bo_va->base.vm = vm;
-	bo_va->base.bo = bo;
-	INIT_LIST_HEAD(&bo_va->base.bo_list);
-	INIT_LIST_HEAD(&bo_va->base.vm_status);
+	amdgpu_vm_bo_base_init(&bo_va->base, vm, bo);
 
 	bo_va->ref_count = 1;
+	bo_va->last_pt_update = dma_fence_get_stub();
 	INIT_LIST_HEAD(&bo_va->valids);
 	INIT_LIST_HEAD(&bo_va->invalids);
 
-	if (bo)
-		list_add_tail(&bo_va->base.bo_list, &bo->va);
+	if (!bo)
+		return bo_va;
+
+	dma_resv_assert_held(bo->tbo.base.resv);
+	if (amdgpu_dmabuf_is_xgmi_accessible(adev, bo)) {
+		bo_va->is_xgmi = true;
+		/* Power up XGMI if it can be potentially used */
+		amdgpu_xgmi_set_pstate(adev, AMDGPU_XGMI_PSTATE_MAX_VEGA20);
+	}
 
 	return bo_va;
 }
 
+
+/**
+ * amdgpu_vm_bo_insert_map - insert a new mapping
+ *
+ * @adev: amdgpu_device pointer
+ * @bo_va: bo_va to store the address
+ * @mapping: the mapping to insert
+ *
+ * Insert a new mapping into all structures.
+ */
+static void amdgpu_vm_bo_insert_map(struct amdgpu_device *adev,
+				    struct amdgpu_bo_va *bo_va,
+				    struct amdgpu_bo_va_mapping *mapping)
+{
+	struct amdgpu_vm *vm = bo_va->base.vm;
+	struct amdgpu_bo *bo = bo_va->base.bo;
+
+	mapping->bo_va = bo_va;
+	list_add(&mapping->list, &bo_va->invalids);
+	amdgpu_vm_it_insert(mapping, &vm->va);
+
+	if (mapping->flags & AMDGPU_VM_PAGE_PRT)
+		amdgpu_vm_prt_get(adev);
+
+	if (amdgpu_vm_is_bo_always_valid(vm, bo) && !bo_va->base.moved)
+		amdgpu_vm_bo_moved(&bo_va->base);
+
+	trace_amdgpu_vm_bo_map(bo_va, mapping);
+}
+
+/* Validate operation parameters to prevent potential abuse */
+static int amdgpu_vm_verify_parameters(struct amdgpu_device *adev,
+					  struct amdgpu_bo *bo,
+					  uint64_t saddr,
+					  uint64_t offset,
+					  uint64_t size)
+{
+	uint64_t tmp, lpfn;
+
+	if (saddr & AMDGPU_GPU_PAGE_MASK
+	    || offset & AMDGPU_GPU_PAGE_MASK
+	    || size & AMDGPU_GPU_PAGE_MASK)
+		return -EINVAL;
+
+	if (check_add_overflow(saddr, size, &tmp)
+	    || check_add_overflow(offset, size, &tmp)
+	    || size == 0 /* which also leads to end < begin */)
+		return -EINVAL;
+
+	/* make sure object fit at this offset */
+	if (bo && offset + size > amdgpu_bo_size(bo))
+		return -EINVAL;
+
+	/* Ensure last pfn not exceed max_pfn */
+	lpfn = (saddr + size - 1) >> AMDGPU_GPU_PAGE_SHIFT;
+	if (lpfn >= adev->vm_manager.max_pfn)
+		return -EINVAL;
+
+	return 0;
+}
+
 /**
  * amdgpu_vm_bo_map - map bo inside a vm
  *
@@ -2056,36 +1824,33 @@ struct amdgpu_bo_va *amdgpu_vm_bo_add(struct amdgpu_device *adev,
  * @bo_va: bo_va to store the address
  * @saddr: where to map the BO
  * @offset: requested offset in the BO
+ * @size: BO size in bytes
  * @flags: attributes of pages (read/write/valid/etc.)
  *
  * Add a mapping of the BO at the specefied addr into the VM.
- * Returns 0 for success, error for failure.
+ *
+ * Returns:
+ * 0 for success, error for failure.
  *
  * Object has to be reserved and unreserved outside!
  */
 int amdgpu_vm_bo_map(struct amdgpu_device *adev,
 		     struct amdgpu_bo_va *bo_va,
 		     uint64_t saddr, uint64_t offset,
-		     uint64_t size, uint64_t flags)
+		     uint64_t size, uint32_t flags)
 {
 	struct amdgpu_bo_va_mapping *mapping, *tmp;
 	struct amdgpu_bo *bo = bo_va->base.bo;
 	struct amdgpu_vm *vm = bo_va->base.vm;
 	uint64_t eaddr;
+	int r;
 
-	/* validate the parameters */
-	if (saddr & AMDGPU_GPU_PAGE_MASK || offset & AMDGPU_GPU_PAGE_MASK ||
-	    size == 0 || size & AMDGPU_GPU_PAGE_MASK)
-		return -EINVAL;
-
-	/* make sure object fit at this offset */
-	eaddr = saddr + size - 1;
-	if (saddr >= eaddr ||
-	    (bo && offset + size > amdgpu_bo_size(bo)))
-		return -EINVAL;
+	r = amdgpu_vm_verify_parameters(adev, bo, saddr, offset, size);
+	if (r)
+		return r;
 
 	saddr /= AMDGPU_GPU_PAGE_SIZE;
-	eaddr /= AMDGPU_GPU_PAGE_SIZE;
+	eaddr = saddr + (size - 1) / AMDGPU_GPU_PAGE_SIZE;
 
 	tmp = amdgpu_vm_it_iter_first(&vm->va, saddr, eaddr);
 	if (tmp) {
@@ -2100,17 +1865,12 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev,
 	if (!mapping)
 		return -ENOMEM;
 
-	INIT_LIST_HEAD(&mapping->list);
 	mapping->start = saddr;
 	mapping->last = eaddr;
 	mapping->offset = offset;
 	mapping->flags = flags;
 
-	list_add(&mapping->list, &bo_va->invalids);
-	amdgpu_vm_it_insert(mapping, &vm->va);
-
-	if (flags & AMDGPU_PTE_PRT)
-		amdgpu_vm_prt_get(adev);
+	amdgpu_vm_bo_insert_map(adev, bo_va, mapping);
 
 	return 0;
 }
@@ -2122,35 +1882,30 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev,
  * @bo_va: bo_va to store the address
  * @saddr: where to map the BO
  * @offset: requested offset in the BO
+ * @size: BO size in bytes
  * @flags: attributes of pages (read/write/valid/etc.)
  *
  * Add a mapping of the BO at the specefied addr into the VM. Replace existing
  * mappings as we do so.
- * Returns 0 for success, error for failure.
+ *
+ * Returns:
+ * 0 for success, error for failure.
  *
  * Object has to be reserved and unreserved outside!
  */
 int amdgpu_vm_bo_replace_map(struct amdgpu_device *adev,
 			     struct amdgpu_bo_va *bo_va,
 			     uint64_t saddr, uint64_t offset,
-			     uint64_t size, uint64_t flags)
+			     uint64_t size, uint32_t flags)
 {
 	struct amdgpu_bo_va_mapping *mapping;
 	struct amdgpu_bo *bo = bo_va->base.bo;
-	struct amdgpu_vm *vm = bo_va->base.vm;
 	uint64_t eaddr;
 	int r;
 
-	/* validate the parameters */
-	if (saddr & AMDGPU_GPU_PAGE_MASK || offset & AMDGPU_GPU_PAGE_MASK ||
-	    size == 0 || size & AMDGPU_GPU_PAGE_MASK)
-		return -EINVAL;
-
-	/* make sure object fit at this offset */
-	eaddr = saddr + size - 1;
-	if (saddr >= eaddr ||
-	    (bo && offset + size > amdgpu_bo_size(bo)))
-		return -EINVAL;
+	r = amdgpu_vm_verify_parameters(adev, bo, saddr, offset, size);
+	if (r)
+		return r;
 
 	/* Allocate all the needed memory */
 	mapping = kmalloc(sizeof(*mapping), GFP_KERNEL);
@@ -2164,18 +1919,14 @@ int amdgpu_vm_bo_replace_map(struct amdgpu_device *adev,
 	}
 
 	saddr /= AMDGPU_GPU_PAGE_SIZE;
-	eaddr /= AMDGPU_GPU_PAGE_SIZE;
+	eaddr = saddr + (size - 1) / AMDGPU_GPU_PAGE_SIZE;
 
 	mapping->start = saddr;
 	mapping->last = eaddr;
 	mapping->offset = offset;
 	mapping->flags = flags;
 
-	list_add(&mapping->list, &bo_va->invalids);
-	amdgpu_vm_it_insert(mapping, &vm->va);
-
-	if (flags & AMDGPU_PTE_PRT)
-		amdgpu_vm_prt_get(adev);
+	amdgpu_vm_bo_insert_map(adev, bo_va, mapping);
 
 	return 0;
 }
@@ -2188,7 +1939,9 @@ int amdgpu_vm_bo_replace_map(struct amdgpu_device *adev,
  * @saddr: where to the BO is mapped
  *
  * Remove a mapping of the BO at the specefied addr from the VM.
- * Returns 0 for success, error for failure.
+ *
+ * Returns:
+ * 0 for success, error for failure.
  *
  * Object has to be reserved and unreserved outside!
  */
@@ -2221,6 +1974,7 @@ int amdgpu_vm_bo_unmap(struct amdgpu_device *adev,
 
 	list_del(&mapping->list);
 	amdgpu_vm_it_remove(mapping, &vm->va);
+	mapping->bo_va = NULL;
 	trace_amdgpu_vm_bo_unmap(bo_va, mapping);
 
 	if (valid)
@@ -2241,7 +1995,9 @@ int amdgpu_vm_bo_unmap(struct amdgpu_device *adev,
  * @size: size of the range
  *
  * Remove all mappings in a range, split them as appropriate.
- * Returns 0 for success, error for failure.
+ *
+ * Returns:
+ * 0 for success, error for failure.
  */
 int amdgpu_vm_bo_clear_mappings(struct amdgpu_device *adev,
 				struct amdgpu_vm *vm,
@@ -2250,10 +2006,14 @@ int amdgpu_vm_bo_clear_mappings(struct amdgpu_device *adev,
 	struct amdgpu_bo_va_mapping *before, *after, *tmp, *next;
 	LIST_HEAD(removed);
 	uint64_t eaddr;
+	int r;
+
+	r = amdgpu_vm_verify_parameters(adev, NULL, saddr, 0, size);
+	if (r)
+		return r;
 
-	eaddr = saddr + size - 1;
 	saddr /= AMDGPU_GPU_PAGE_SIZE;
-	eaddr /= AMDGPU_GPU_PAGE_SIZE;
+	eaddr = saddr + (size - 1) / AMDGPU_GPU_PAGE_SIZE;
 
 	/* Allocate all the needed memory */
 	before = kzalloc(sizeof(*before), GFP_KERNEL);
@@ -2277,7 +2037,8 @@ int amdgpu_vm_bo_clear_mappings(struct amdgpu_device *adev,
 			before->last = saddr - 1;
 			before->offset = tmp->offset;
 			before->flags = tmp->flags;
-			list_add(&before->list, &tmp->list);
+			before->bo_va = tmp->bo_va;
+			list_add(&before->list, &tmp->bo_va->invalids);
 		}
 
 		/* Remember mapping split at the end */
@@ -2285,9 +2046,10 @@ int amdgpu_vm_bo_clear_mappings(struct amdgpu_device *adev,
 			after->start = eaddr + 1;
 			after->last = tmp->last;
 			after->offset = tmp->offset;
-			after->offset += after->start - tmp->start;
+			after->offset += (after->start - tmp->start) << PAGE_SHIFT;
 			after->flags = tmp->flags;
-			list_add(&after->list, &tmp->list);
+			after->bo_va = tmp->bo_va;
+			list_add(&after->list, &tmp->bo_va->invalids);
 		}
 
 		list_del(&tmp->list);
@@ -2306,24 +2068,37 @@ int amdgpu_vm_bo_clear_mappings(struct amdgpu_device *adev,
 		if (tmp->last > eaddr)
 		    tmp->last = eaddr;
 
+		tmp->bo_va = NULL;
 		list_add(&tmp->list, &vm->freed);
 		trace_amdgpu_vm_bo_unmap(NULL, tmp);
 	}
 
 	/* Insert partial mapping before the range */
 	if (!list_empty(&before->list)) {
+		struct amdgpu_bo *bo = before->bo_va->base.bo;
+
 		amdgpu_vm_it_insert(before, &vm->va);
-		if (before->flags & AMDGPU_PTE_PRT)
+		if (before->flags & AMDGPU_PTE_PRT_FLAG(adev))
 			amdgpu_vm_prt_get(adev);
+
+		if (amdgpu_vm_is_bo_always_valid(vm, bo) &&
+		    !before->bo_va->base.moved)
+			amdgpu_vm_bo_moved(&before->bo_va->base);
 	} else {
 		kfree(before);
 	}
 
 	/* Insert partial mapping after the range */
 	if (!list_empty(&after->list)) {
+		struct amdgpu_bo *bo = after->bo_va->base.bo;
+
 		amdgpu_vm_it_insert(after, &vm->va);
-		if (after->flags & AMDGPU_PTE_PRT)
+		if (after->flags & AMDGPU_PTE_PRT_FLAG(adev))
 			amdgpu_vm_prt_get(adev);
+
+		if (amdgpu_vm_is_bo_always_valid(vm, bo) &&
+		    !after->bo_va->base.moved)
+			amdgpu_vm_bo_moved(&after->bo_va->base);
 	} else {
 		kfree(after);
 	}
@@ -2332,7 +2107,55 @@ int amdgpu_vm_bo_clear_mappings(struct amdgpu_device *adev,
 }
 
 /**
- * amdgpu_vm_bo_rmv - remove a bo to a specific vm
+ * amdgpu_vm_bo_lookup_mapping - find mapping by address
+ *
+ * @vm: the requested VM
+ * @addr: the address
+ *
+ * Find a mapping by it's address.
+ *
+ * Returns:
+ * The amdgpu_bo_va_mapping matching for addr or NULL
+ *
+ */
+struct amdgpu_bo_va_mapping *amdgpu_vm_bo_lookup_mapping(struct amdgpu_vm *vm,
+							 uint64_t addr)
+{
+	return amdgpu_vm_it_iter_first(&vm->va, addr, addr);
+}
+
+/**
+ * amdgpu_vm_bo_trace_cs - trace all reserved mappings
+ *
+ * @vm: the requested vm
+ * @ticket: CS ticket
+ *
+ * Trace all mappings of BOs reserved during a command submission.
+ */
+void amdgpu_vm_bo_trace_cs(struct amdgpu_vm *vm, struct ww_acquire_ctx *ticket)
+{
+	struct amdgpu_bo_va_mapping *mapping;
+
+	if (!trace_amdgpu_vm_bo_cs_enabled())
+		return;
+
+	for (mapping = amdgpu_vm_it_iter_first(&vm->va, 0, U64_MAX); mapping;
+	     mapping = amdgpu_vm_it_iter_next(mapping, 0, U64_MAX)) {
+		if (mapping->bo_va && mapping->bo_va->base.bo) {
+			struct amdgpu_bo *bo;
+
+			bo = mapping->bo_va->base.bo;
+			if (dma_resv_locking_ctx(bo->tbo.base.resv) !=
+			    ticket)
+				continue;
+		}
+
+		trace_amdgpu_vm_bo_cs(mapping);
+	}
+}
+
+/**
+ * amdgpu_vm_bo_del - remove a bo from a specific vm
  *
  * @adev: amdgpu_device pointer
  * @bo_va: requested bo_va
@@ -2341,13 +2164,31 @@ int amdgpu_vm_bo_clear_mappings(struct amdgpu_device *adev,
  *
  * Object have to be reserved!
  */
-void amdgpu_vm_bo_rmv(struct amdgpu_device *adev,
+void amdgpu_vm_bo_del(struct amdgpu_device *adev,
 		      struct amdgpu_bo_va *bo_va)
 {
 	struct amdgpu_bo_va_mapping *mapping, *next;
+	struct amdgpu_bo *bo = bo_va->base.bo;
 	struct amdgpu_vm *vm = bo_va->base.vm;
+	struct amdgpu_vm_bo_base **base;
+
+	dma_resv_assert_held(vm->root.bo->tbo.base.resv);
+
+	if (bo) {
+		dma_resv_assert_held(bo->tbo.base.resv);
+		if (amdgpu_vm_is_bo_always_valid(vm, bo))
+			ttm_bo_set_bulk_move(&bo->tbo, NULL);
 
-	list_del(&bo_va->base.bo_list);
+		for (base = &bo_va->base.bo->vm_bo; *base;
+		     base = &(*base)->next) {
+			if (*base != &bo_va->base)
+				continue;
+
+			amdgpu_vm_update_stats(*base, bo->tbo.resource, -1);
+			*base = bo_va->base.next;
+			break;
+		}
+	}
 
 	spin_lock(&vm->status_lock);
 	list_del(&bo_va->base.vm_status);
@@ -2356,6 +2197,7 @@ void amdgpu_vm_bo_rmv(struct amdgpu_device *adev,
 	list_for_each_entry_safe(mapping, next, &bo_va->valids, list) {
 		list_del(&mapping->list);
 		amdgpu_vm_it_remove(mapping, &vm->va);
+		mapping->bo_va = NULL;
 		trace_amdgpu_vm_bo_unmap(bo_va, mapping);
 		list_add(&mapping->list, &vm->freed);
 	}
@@ -2367,32 +2209,114 @@ void amdgpu_vm_bo_rmv(struct amdgpu_device *adev,
 	}
 
 	dma_fence_put(bo_va->last_pt_update);
+
+	if (bo && bo_va->is_xgmi)
+		amdgpu_xgmi_set_pstate(adev, AMDGPU_XGMI_PSTATE_MIN);
+
 	kfree(bo_va);
 }
 
 /**
+ * amdgpu_vm_evictable - check if we can evict a VM
+ *
+ * @bo: A page table of the VM.
+ *
+ * Check if it is possible to evict a VM.
+ */
+bool amdgpu_vm_evictable(struct amdgpu_bo *bo)
+{
+	struct amdgpu_vm_bo_base *bo_base = bo->vm_bo;
+
+	/* Page tables of a destroyed VM can go away immediately */
+	if (!bo_base || !bo_base->vm)
+		return true;
+
+	/* Don't evict VM page tables while they are busy */
+	if (!dma_resv_test_signaled(bo->tbo.base.resv, DMA_RESV_USAGE_BOOKKEEP))
+		return false;
+
+	/* Try to block ongoing updates */
+	if (!amdgpu_vm_eviction_trylock(bo_base->vm))
+		return false;
+
+	/* Don't evict VM page tables while they are updated */
+	if (!dma_fence_is_signaled(bo_base->vm->last_unlocked)) {
+		amdgpu_vm_eviction_unlock(bo_base->vm);
+		return false;
+	}
+
+	bo_base->vm->evicting = true;
+	amdgpu_vm_eviction_unlock(bo_base->vm);
+	return true;
+}
+
+/**
  * amdgpu_vm_bo_invalidate - mark the bo as invalid
  *
- * @adev: amdgpu_device pointer
- * @vm: requested vm
  * @bo: amdgpu buffer object
+ * @evicted: is the BO evicted
  *
  * Mark @bo as invalid.
  */
-void amdgpu_vm_bo_invalidate(struct amdgpu_device *adev,
-			     struct amdgpu_bo *bo)
+void amdgpu_vm_bo_invalidate(struct amdgpu_bo *bo, bool evicted)
 {
 	struct amdgpu_vm_bo_base *bo_base;
 
-	list_for_each_entry(bo_base, &bo->va, bo_list) {
-		spin_lock(&bo_base->vm->status_lock);
-		if (list_empty(&bo_base->vm_status))
-			list_add(&bo_base->vm_status,
-				 &bo_base->vm->moved);
-		spin_unlock(&bo_base->vm->status_lock);
+	for (bo_base = bo->vm_bo; bo_base; bo_base = bo_base->next) {
+		struct amdgpu_vm *vm = bo_base->vm;
+
+		if (evicted && amdgpu_vm_is_bo_always_valid(vm, bo)) {
+			amdgpu_vm_bo_evicted(bo_base);
+			continue;
+		}
+
+		if (bo_base->moved)
+			continue;
+		bo_base->moved = true;
+
+		if (bo->tbo.type == ttm_bo_type_kernel)
+			amdgpu_vm_bo_relocated(bo_base);
+		else if (amdgpu_vm_is_bo_always_valid(vm, bo))
+			amdgpu_vm_bo_moved(bo_base);
+		else
+			amdgpu_vm_bo_invalidated(bo_base);
 	}
 }
 
+/**
+ * amdgpu_vm_bo_move - handle BO move
+ *
+ * @bo: amdgpu buffer object
+ * @new_mem: the new placement of the BO move
+ * @evicted: is the BO evicted
+ *
+ * Update the memory stats for the new placement and mark @bo as invalid.
+ */
+void amdgpu_vm_bo_move(struct amdgpu_bo *bo, struct ttm_resource *new_mem,
+		       bool evicted)
+{
+	struct amdgpu_vm_bo_base *bo_base;
+
+	for (bo_base = bo->vm_bo; bo_base; bo_base = bo_base->next) {
+		struct amdgpu_vm *vm = bo_base->vm;
+
+		spin_lock(&vm->status_lock);
+		amdgpu_vm_update_stats_locked(bo_base, bo->tbo.resource, -1);
+		amdgpu_vm_update_stats_locked(bo_base, new_mem, +1);
+		spin_unlock(&vm->status_lock);
+	}
+
+	amdgpu_vm_bo_invalidate(bo, evicted);
+}
+
+/**
+ * amdgpu_vm_get_block_size - calculate VM page table size as power of two
+ *
+ * @vm_size: VM size
+ *
+ * Returns:
+ * VM page table as power of two
+ */
 static uint32_t amdgpu_vm_get_block_size(uint64_t vm_size)
 {
 	/* Total bits covered by PD + PTs */
@@ -2407,45 +2331,216 @@ static uint32_t amdgpu_vm_get_block_size(uint64_t vm_size)
 }
 
 /**
- * amdgpu_vm_set_fragment_size - adjust fragment size in PTE
+ * amdgpu_vm_adjust_size - adjust vm size, block size and fragment size
  *
  * @adev: amdgpu_device pointer
- * @fragment_size_default: the default fragment size if it's set auto
+ * @min_vm_size: the minimum vm size in GB if it's set auto
+ * @fragment_size_default: Default PTE fragment size
+ * @max_level: max VMPT level
+ * @max_bits: max address space size in bits
+ *
  */
-void amdgpu_vm_set_fragment_size(struct amdgpu_device *adev, uint32_t fragment_size_default)
+void amdgpu_vm_adjust_size(struct amdgpu_device *adev, uint32_t min_vm_size,
+			   uint32_t fragment_size_default, unsigned max_level,
+			   unsigned max_bits)
 {
+	unsigned int max_size = 1 << (max_bits - 30);
+	unsigned int vm_size;
+	uint64_t tmp;
+
+	/* adjust vm size first */
+	if (amdgpu_vm_size != -1) {
+		vm_size = amdgpu_vm_size;
+		if (vm_size > max_size) {
+			dev_warn(adev->dev, "VM size (%d) too large, max is %u GB\n",
+				 amdgpu_vm_size, max_size);
+			vm_size = max_size;
+		}
+	} else {
+		struct sysinfo si;
+		unsigned int phys_ram_gb;
+
+		/* Optimal VM size depends on the amount of physical
+		 * RAM available. Underlying requirements and
+		 * assumptions:
+		 *
+		 *  - Need to map system memory and VRAM from all GPUs
+		 *     - VRAM from other GPUs not known here
+		 *     - Assume VRAM <= system memory
+		 *  - On GFX8 and older, VM space can be segmented for
+		 *    different MTYPEs
+		 *  - Need to allow room for fragmentation, guard pages etc.
+		 *
+		 * This adds up to a rough guess of system memory x3.
+		 * Round up to power of two to maximize the available
+		 * VM size with the given page table size.
+		 */
+		si_meminfo(&si);
+		phys_ram_gb = ((uint64_t)si.totalram * si.mem_unit +
+			       (1 << 30) - 1) >> 30;
+		vm_size = roundup_pow_of_two(
+			clamp(phys_ram_gb * 3, min_vm_size, max_size));
+	}
+
+	adev->vm_manager.max_pfn = (uint64_t)vm_size << 18;
+
+	tmp = roundup_pow_of_two(adev->vm_manager.max_pfn);
+	if (amdgpu_vm_block_size != -1)
+		tmp >>= amdgpu_vm_block_size - 9;
+	tmp = DIV_ROUND_UP(fls64(tmp) - 1, 9) - 1;
+	adev->vm_manager.num_level = min_t(unsigned int, max_level, tmp);
+	switch (adev->vm_manager.num_level) {
+	case 3:
+		adev->vm_manager.root_level = AMDGPU_VM_PDB2;
+		break;
+	case 2:
+		adev->vm_manager.root_level = AMDGPU_VM_PDB1;
+		break;
+	case 1:
+		adev->vm_manager.root_level = AMDGPU_VM_PDB0;
+		break;
+	default:
+		dev_err(adev->dev, "VMPT only supports 2~4+1 levels\n");
+	}
+	/* block size depends on vm size and hw setup*/
+	if (amdgpu_vm_block_size != -1)
+		adev->vm_manager.block_size =
+			min((unsigned)amdgpu_vm_block_size, max_bits
+			    - AMDGPU_GPU_PAGE_SHIFT
+			    - 9 * adev->vm_manager.num_level);
+	else if (adev->vm_manager.num_level > 1)
+		adev->vm_manager.block_size = 9;
+	else
+		adev->vm_manager.block_size = amdgpu_vm_get_block_size(tmp);
+
 	if (amdgpu_vm_fragment_size == -1)
 		adev->vm_manager.fragment_size = fragment_size_default;
 	else
 		adev->vm_manager.fragment_size = amdgpu_vm_fragment_size;
+
+	dev_info(
+		adev->dev,
+		"vm size is %u GB, %u levels, block size is %u-bit, fragment size is %u-bit\n",
+		vm_size, adev->vm_manager.num_level + 1,
+		adev->vm_manager.block_size, adev->vm_manager.fragment_size);
 }
 
 /**
- * amdgpu_vm_adjust_size - adjust vm size, block size and fragment size
+ * amdgpu_vm_wait_idle - wait for the VM to become idle
  *
- * @adev: amdgpu_device pointer
- * @vm_size: the default vm size if it's set auto
+ * @vm: VM object to wait for
+ * @timeout: timeout to wait for VM to become idle
  */
-void amdgpu_vm_adjust_size(struct amdgpu_device *adev, uint64_t vm_size, uint32_t fragment_size_default)
+long amdgpu_vm_wait_idle(struct amdgpu_vm *vm, long timeout)
 {
-	/* adjust vm size firstly */
-	if (amdgpu_vm_size == -1)
-		adev->vm_manager.vm_size = vm_size;
-	else
-		adev->vm_manager.vm_size = amdgpu_vm_size;
+	timeout = drm_sched_entity_flush(&vm->immediate, timeout);
+	if (timeout <= 0)
+		return timeout;
 
-	/* block size depends on vm size */
-	if (amdgpu_vm_block_size == -1)
-		adev->vm_manager.block_size =
-			amdgpu_vm_get_block_size(adev->vm_manager.vm_size);
-	else
-		adev->vm_manager.block_size = amdgpu_vm_block_size;
+	return drm_sched_entity_flush(&vm->delayed, timeout);
+}
+
+static void amdgpu_vm_destroy_task_info(struct kref *kref)
+{
+	struct amdgpu_task_info *ti = container_of(kref, struct amdgpu_task_info, refcount);
+
+	kfree(ti);
+}
+
+static inline struct amdgpu_vm *
+amdgpu_vm_get_vm_from_pasid(struct amdgpu_device *adev, u32 pasid)
+{
+	struct amdgpu_vm *vm;
+	unsigned long flags;
+
+	xa_lock_irqsave(&adev->vm_manager.pasids, flags);
+	vm = xa_load(&adev->vm_manager.pasids, pasid);
+	xa_unlock_irqrestore(&adev->vm_manager.pasids, flags);
+
+	return vm;
+}
+
+/**
+ * amdgpu_vm_put_task_info - reference down the vm task_info ptr
+ *
+ * @task_info: task_info struct under discussion.
+ *
+ * frees the vm task_info ptr at the last put
+ */
+void amdgpu_vm_put_task_info(struct amdgpu_task_info *task_info)
+{
+	if (task_info)
+		kref_put(&task_info->refcount, amdgpu_vm_destroy_task_info);
+}
+
+/**
+ * amdgpu_vm_get_task_info_vm - Extracts task info for a vm.
+ *
+ * @vm: VM to get info from
+ *
+ * Returns the reference counted task_info structure, which must be
+ * referenced down with amdgpu_vm_put_task_info.
+ */
+struct amdgpu_task_info *
+amdgpu_vm_get_task_info_vm(struct amdgpu_vm *vm)
+{
+	struct amdgpu_task_info *ti = NULL;
+
+	if (vm) {
+		ti = vm->task_info;
+		kref_get(&vm->task_info->refcount);
+	}
+
+	return ti;
+}
+
+/**
+ * amdgpu_vm_get_task_info_pasid - Extracts task info for a PASID.
+ *
+ * @adev: drm device pointer
+ * @pasid: PASID identifier for VM
+ *
+ * Returns the reference counted task_info structure, which must be
+ * referenced down with amdgpu_vm_put_task_info.
+ */
+struct amdgpu_task_info *
+amdgpu_vm_get_task_info_pasid(struct amdgpu_device *adev, u32 pasid)
+{
+	return amdgpu_vm_get_task_info_vm(
+			amdgpu_vm_get_vm_from_pasid(adev, pasid));
+}
 
-	amdgpu_vm_set_fragment_size(adev, fragment_size_default);
+static int amdgpu_vm_create_task_info(struct amdgpu_vm *vm)
+{
+	vm->task_info = kzalloc(sizeof(struct amdgpu_task_info), GFP_KERNEL);
+	if (!vm->task_info)
+		return -ENOMEM;
 
-	DRM_INFO("vm size is %llu GB, block size is %u-bit, fragment size is %u-bit\n",
-		adev->vm_manager.vm_size, adev->vm_manager.block_size,
-		adev->vm_manager.fragment_size);
+	kref_init(&vm->task_info->refcount);
+	return 0;
+}
+
+/**
+ * amdgpu_vm_set_task_info - Sets VMs task info.
+ *
+ * @vm: vm for which to set the info
+ */
+void amdgpu_vm_set_task_info(struct amdgpu_vm *vm)
+{
+	if (!vm->task_info)
+		return;
+
+	if (vm->task_info->task.pid == current->pid)
+		return;
+
+	vm->task_info->task.pid = current->pid;
+	get_task_comm(vm->task_info->task.comm, current);
+
+	if (current->group_leader->mm != current->mm)
+		return;
+
+	vm->task_info->tgid = current->group_leader->pid;
+	get_task_comm(vm->task_info->process_name, current->group_leader);
 }
 
 /**
@@ -2453,124 +2548,193 @@ void amdgpu_vm_adjust_size(struct amdgpu_device *adev, uint64_t vm_size, uint32_
  *
  * @adev: amdgpu_device pointer
  * @vm: requested vm
- * @vm_context: Indicates if it GFX or Compute context
+ * @xcp_id: GPU partition selection id
+ * @pasid: the pasid the VM is using on this GPU
  *
  * Init @vm fields.
+ *
+ * Returns:
+ * 0 for success, error for failure.
  */
 int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
-		   int vm_context)
+		   int32_t xcp_id, uint32_t pasid)
 {
-	const unsigned align = min(AMDGPU_VM_PTB_ALIGN_SIZE,
-		AMDGPU_VM_PTE_COUNT(adev) * 8);
-	unsigned ring_instance;
-	struct amdgpu_ring *ring;
-	struct amd_sched_rq *rq;
+	struct amdgpu_bo *root_bo;
+	struct amdgpu_bo_vm *root;
 	int r, i;
-	u64 flags;
-	uint64_t init_pde_value = 0;
 
 	vm->va = RB_ROOT_CACHED;
-	vm->client_id = atomic64_inc_return(&adev->vm_manager.client_counter);
 	for (i = 0; i < AMDGPU_MAX_VMHUBS; i++)
 		vm->reserved_vmid[i] = NULL;
-	spin_lock_init(&vm->status_lock);
+	INIT_LIST_HEAD(&vm->evicted);
+	INIT_LIST_HEAD(&vm->evicted_user);
+	INIT_LIST_HEAD(&vm->relocated);
 	INIT_LIST_HEAD(&vm->moved);
-	INIT_LIST_HEAD(&vm->cleared);
+	INIT_LIST_HEAD(&vm->idle);
+	INIT_LIST_HEAD(&vm->invalidated);
+	spin_lock_init(&vm->status_lock);
 	INIT_LIST_HEAD(&vm->freed);
+	INIT_LIST_HEAD(&vm->done);
+	INIT_KFIFO(vm->faults);
 
-	/* create scheduler entity for page table updates */
-
-	ring_instance = atomic_inc_return(&adev->vm_manager.vm_pte_next_ring);
-	ring_instance %= adev->vm_manager.vm_pte_num_rings;
-	ring = adev->vm_manager.vm_pte_rings[ring_instance];
-	rq = &ring->sched.sched_rq[AMD_SCHED_PRIORITY_KERNEL];
-	r = amd_sched_entity_init(&ring->sched, &vm->entity,
-				  rq, amdgpu_sched_jobs);
+	r = amdgpu_vm_init_entities(adev, vm);
 	if (r)
 		return r;
 
-	vm->pte_support_ats = false;
+	ttm_lru_bulk_move_init(&vm->lru_bulk_move);
 
-	if (vm_context == AMDGPU_VM_CONTEXT_COMPUTE) {
-		vm->use_cpu_for_update = !!(adev->vm_manager.vm_update_mode &
-						AMDGPU_VM_USE_CPU_FOR_COMPUTE);
+	vm->is_compute_context = false;
 
-		if (adev->asic_type == CHIP_RAVEN) {
-			vm->pte_support_ats = true;
-			init_pde_value = AMDGPU_PTE_SYSTEM | AMDGPU_PDE_PTE;
-		}
-	} else
-		vm->use_cpu_for_update = !!(adev->vm_manager.vm_update_mode &
-						AMDGPU_VM_USE_CPU_FOR_GFX);
-	DRM_DEBUG_DRIVER("VM update mode is %s\n",
-			 vm->use_cpu_for_update ? "CPU" : "SDMA");
-	WARN_ONCE((vm->use_cpu_for_update & !amdgpu_vm_is_large_bar(adev)),
+	vm->use_cpu_for_update = !!(adev->vm_manager.vm_update_mode &
+				    AMDGPU_VM_USE_CPU_FOR_GFX);
+
+	dev_dbg(adev->dev, "VM update mode is %s\n",
+		vm->use_cpu_for_update ? "CPU" : "SDMA");
+	WARN_ONCE((vm->use_cpu_for_update &&
+		   !amdgpu_gmc_vram_full_visible(&adev->gmc)),
 		  "CPU update of VM recommended only for large BAR system\n");
-	vm->last_dir_update = NULL;
 
-	flags = AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS |
-			AMDGPU_GEM_CREATE_VRAM_CLEARED;
 	if (vm->use_cpu_for_update)
-		flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
+		vm->update_funcs = &amdgpu_vm_cpu_funcs;
 	else
-		flags |= (AMDGPU_GEM_CREATE_NO_CPU_ACCESS |
-				AMDGPU_GEM_CREATE_SHADOW);
+		vm->update_funcs = &amdgpu_vm_sdma_funcs;
+
+	vm->last_update = dma_fence_get_stub();
+	vm->last_unlocked = dma_fence_get_stub();
+	vm->last_tlb_flush = dma_fence_get_stub();
+	vm->generation = amdgpu_vm_generation(adev, NULL);
+
+	mutex_init(&vm->eviction_lock);
+	vm->evicting = false;
+	vm->tlb_fence_context = dma_fence_context_alloc(1);
+
+	r = amdgpu_vm_pt_create(adev, vm, adev->vm_manager.root_level,
+				false, &root, xcp_id);
+	if (r)
+		goto error_free_delayed;
+
+	root_bo = amdgpu_bo_ref(&root->bo);
+	r = amdgpu_bo_reserve(root_bo, true);
+	if (r) {
+		amdgpu_bo_unref(&root_bo);
+		goto error_free_delayed;
+	}
 
-	r = amdgpu_bo_create(adev, amdgpu_vm_bo_size(adev, 0), align, true,
-			     AMDGPU_GEM_DOMAIN_VRAM,
-			     flags,
-			     NULL, NULL, init_pde_value, &vm->root.bo);
+	amdgpu_vm_bo_base_init(&vm->root, vm, root_bo);
+	r = dma_resv_reserve_fences(root_bo->tbo.base.resv, 1);
 	if (r)
-		goto error_free_sched_entity;
+		goto error_free_root;
 
-	r = amdgpu_bo_reserve(vm->root.bo, false);
+	r = amdgpu_vm_pt_clear(adev, vm, root, false);
 	if (r)
 		goto error_free_root;
 
-	vm->last_eviction_counter = atomic64_read(&adev->num_evictions);
+	r = amdgpu_vm_create_task_info(vm);
+	if (r)
+		dev_dbg(adev->dev, "Failed to create task info for VM\n");
 
-	if (vm->use_cpu_for_update) {
-		r = amdgpu_bo_kmap(vm->root.bo, NULL);
-		if (r)
+	/* Store new PASID in XArray (if non-zero) */
+	if (pasid != 0) {
+		r = xa_err(xa_store_irq(&adev->vm_manager.pasids, pasid, vm, GFP_KERNEL));
+		if (r < 0)
 			goto error_free_root;
+
+		vm->pasid = pasid;
 	}
 
 	amdgpu_bo_unreserve(vm->root.bo);
+	amdgpu_bo_unref(&root_bo);
 
 	return 0;
 
 error_free_root:
-	amdgpu_bo_unref(&vm->root.bo->shadow);
-	amdgpu_bo_unref(&vm->root.bo);
-	vm->root.bo = NULL;
+	/* If PASID was partially set, erase it from XArray before failing */
+	if (vm->pasid != 0) {
+		xa_erase_irq(&adev->vm_manager.pasids, vm->pasid);
+		vm->pasid = 0;
+	}
+	amdgpu_vm_pt_free_root(adev, vm);
+	amdgpu_bo_unreserve(vm->root.bo);
+	amdgpu_bo_unref(&root_bo);
 
-error_free_sched_entity:
-	amd_sched_entity_fini(&ring->sched, &vm->entity);
+error_free_delayed:
+	dma_fence_put(vm->last_tlb_flush);
+	dma_fence_put(vm->last_unlocked);
+	ttm_lru_bulk_move_fini(&adev->mman.bdev, &vm->lru_bulk_move);
+	amdgpu_vm_fini_entities(vm);
 
 	return r;
 }
 
 /**
- * amdgpu_vm_free_levels - free PD/PT levels
+ * amdgpu_vm_make_compute - Turn a GFX VM into a compute VM
  *
- * @level: PD/PT starting level to free
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
  *
- * Free the page directory or page table level and all sub levels.
+ * This only works on GFX VMs that don't have any BOs added and no
+ * page tables allocated yet.
+ *
+ * Changes the following VM parameters:
+ * - use_cpu_for_update
+ * - pte_supports_ats
+ *
+ * Reinitializes the page directory to reflect the changed ATS
+ * setting.
+ *
+ * Returns:
+ * 0 for success, -errno for errors.
  */
-static void amdgpu_vm_free_levels(struct amdgpu_vm_pt *level)
+int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 {
-	unsigned i;
+	int r;
 
-	if (level->bo) {
-		amdgpu_bo_unref(&level->bo->shadow);
-		amdgpu_bo_unref(&level->bo);
+	r = amdgpu_bo_reserve(vm->root.bo, true);
+	if (r)
+		return r;
+
+	/* Update VM state */
+	vm->use_cpu_for_update = !!(adev->vm_manager.vm_update_mode &
+				    AMDGPU_VM_USE_CPU_FOR_COMPUTE);
+	dev_dbg(adev->dev, "VM update mode is %s\n",
+		vm->use_cpu_for_update ? "CPU" : "SDMA");
+	WARN_ONCE((vm->use_cpu_for_update &&
+		   !amdgpu_gmc_vram_full_visible(&adev->gmc)),
+		  "CPU update of VM recommended only for large BAR system\n");
+
+	if (vm->use_cpu_for_update) {
+		/* Sync with last SDMA update/clear before switching to CPU */
+		r = amdgpu_bo_sync_wait(vm->root.bo,
+					AMDGPU_FENCE_OWNER_UNDEFINED, true);
+		if (r)
+			goto unreserve_bo;
+
+		vm->update_funcs = &amdgpu_vm_cpu_funcs;
+		r = amdgpu_vm_pt_map_tables(adev, vm);
+		if (r)
+			goto unreserve_bo;
+
+	} else {
+		vm->update_funcs = &amdgpu_vm_sdma_funcs;
 	}
 
-	if (level->entries)
-		for (i = 0; i <= level->last_entry_used; i++)
-			amdgpu_vm_free_levels(&level->entries[i]);
+	dma_fence_put(vm->last_update);
+	vm->last_update = dma_fence_get_stub();
+	vm->is_compute_context = true;
+
+unreserve_bo:
+	amdgpu_bo_unreserve(vm->root.bo);
+	return r;
+}
 
-	kvfree(level->entries);
+static int amdgpu_vm_stats_is_zero(struct amdgpu_vm *vm)
+{
+	for (int i = 0; i < __AMDGPU_PL_NUM; ++i) {
+		if (!(drm_memory_stats_is_zero(&vm->stats[i].drm) &&
+		      vm->stats[i].evicted == 0))
+			return false;
+	}
+	return true;
 }
 
 /**
@@ -2585,34 +2749,74 @@ static void amdgpu_vm_free_levels(struct amdgpu_vm_pt *level)
 void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 {
 	struct amdgpu_bo_va_mapping *mapping, *tmp;
-	bool prt_fini_needed = !!adev->gart.gart_funcs->set_prt;
+	bool prt_fini_needed = !!adev->gmc.gmc_funcs->set_prt;
+	struct amdgpu_bo *root;
+	unsigned long flags;
 	int i;
 
-	amd_sched_entity_fini(vm->entity.sched, &vm->entity);
+	amdgpu_amdkfd_gpuvm_destroy_cb(adev, vm);
+
+	root = amdgpu_bo_ref(vm->root.bo);
+	amdgpu_bo_reserve(root, true);
+	/* Remove PASID mapping before destroying VM */
+	if (vm->pasid != 0) {
+		xa_erase_irq(&adev->vm_manager.pasids, vm->pasid);
+		vm->pasid = 0;
+	}
+	dma_fence_wait(vm->last_unlocked, false);
+	dma_fence_put(vm->last_unlocked);
+	dma_fence_wait(vm->last_tlb_flush, false);
+	/* Make sure that all fence callbacks have completed */
+	spin_lock_irqsave(vm->last_tlb_flush->lock, flags);
+	spin_unlock_irqrestore(vm->last_tlb_flush->lock, flags);
+	dma_fence_put(vm->last_tlb_flush);
+
+	list_for_each_entry_safe(mapping, tmp, &vm->freed, list) {
+		if (mapping->flags & AMDGPU_VM_PAGE_PRT && prt_fini_needed) {
+			amdgpu_vm_prt_fini(adev, vm);
+			prt_fini_needed = false;
+		}
+
+		list_del(&mapping->list);
+		amdgpu_vm_free_mapping(adev, vm, mapping, NULL);
+	}
+
+	amdgpu_vm_pt_free_root(adev, vm);
+	amdgpu_bo_unreserve(root);
+	amdgpu_bo_unref(&root);
+	WARN_ON(vm->root.bo);
+
+	amdgpu_vm_fini_entities(vm);
 
 	if (!RB_EMPTY_ROOT(&vm->va.rb_root)) {
 		dev_err(adev->dev, "still active bo inside vm\n");
 	}
 	rbtree_postorder_for_each_entry_safe(mapping, tmp,
 					     &vm->va.rb_root, rb) {
+		/* Don't remove the mapping here, we don't want to trigger a
+		 * rebalance and the tree is about to be destroyed anyway.
+		 */
 		list_del(&mapping->list);
-		amdgpu_vm_it_remove(mapping, &vm->va);
 		kfree(mapping);
 	}
-	list_for_each_entry_safe(mapping, tmp, &vm->freed, list) {
-		if (mapping->flags & AMDGPU_PTE_PRT && prt_fini_needed) {
-			amdgpu_vm_prt_fini(adev, vm);
-			prt_fini_needed = false;
-		}
 
-		list_del(&mapping->list);
-		amdgpu_vm_free_mapping(adev, vm, mapping, NULL);
+	dma_fence_put(vm->last_update);
+
+	for (i = 0; i < AMDGPU_MAX_VMHUBS; i++) {
+		amdgpu_vmid_free_reserved(adev, vm, i);
 	}
 
-	amdgpu_vm_free_levels(&vm->root);
-	dma_fence_put(vm->last_dir_update);
-	for (i = 0; i < AMDGPU_MAX_VMHUBS; i++)
-		amdgpu_vm_free_reserved_vmid(adev, vm, i);
+	ttm_lru_bulk_move_fini(&adev->mman.bdev, &vm->lru_bulk_move);
+
+	if (!amdgpu_vm_stats_is_zero(vm)) {
+		struct amdgpu_task_info *ti = vm->task_info;
+
+		dev_warn(adev->dev,
+			 "VM memory stats for proc %s(%d) task %s(%d) is non-zero when fini\n",
+			 ti->process_name, ti->task.pid, ti->task.comm, ti->tgid);
+	}
+
+	amdgpu_vm_put_task_info(vm->task_info);
 }
 
 /**
@@ -2624,31 +2828,21 @@ void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm)
  */
 void amdgpu_vm_manager_init(struct amdgpu_device *adev)
 {
-	unsigned i, j;
-
-	for (i = 0; i < AMDGPU_MAX_VMHUBS; ++i) {
-		struct amdgpu_vm_id_manager *id_mgr =
-			&adev->vm_manager.id_mgr[i];
-
-		mutex_init(&id_mgr->lock);
-		INIT_LIST_HEAD(&id_mgr->ids_lru);
-		atomic_set(&id_mgr->reserved_vmid_num, 0);
+	unsigned i;
 
-		/* skip over VMID 0, since it is the system VM */
-		for (j = 1; j < id_mgr->num_ids; ++j) {
-			amdgpu_vm_reset_id(adev, i, j);
-			amdgpu_sync_create(&id_mgr->ids[i].active);
-			list_add_tail(&id_mgr->ids[j].list, &id_mgr->ids_lru);
-		}
-	}
+	/* Concurrent flushes are only possible starting with Vega10 and
+	 * are broken on Navi10 and Navi14.
+	 */
+	adev->vm_manager.concurrent_flush = !(adev->asic_type < CHIP_VEGA10 ||
+					      adev->asic_type == CHIP_NAVI10 ||
+					      adev->asic_type == CHIP_NAVI14);
+	amdgpu_vmid_mgr_init(adev);
 
 	adev->vm_manager.fence_context =
 		dma_fence_context_alloc(AMDGPU_MAX_RINGS);
 	for (i = 0; i < AMDGPU_MAX_RINGS; ++i)
 		adev->vm_manager.seqno[i] = 0;
 
-	atomic_set(&adev->vm_manager.vm_pte_next_ring, 0);
-	atomic64_set(&adev->vm_manager.client_counter, 0);
 	spin_lock_init(&adev->vm_manager.prt_lock);
 	atomic_set(&adev->vm_manager.num_prt_users, 0);
 
@@ -2657,7 +2851,11 @@ void amdgpu_vm_manager_init(struct amdgpu_device *adev)
 	 */
 #ifdef CONFIG_X86_64
 	if (amdgpu_vm_update_mode == -1) {
-		if (amdgpu_vm_is_large_bar(adev))
+		/* For asic with VF MMIO access protection
+		 * avoid using CPU for VM table updates
+		 */
+		if (amdgpu_gmc_vram_full_visible(&adev->gmc) &&
+		    !amdgpu_sriov_vf_mmio_access_protection(adev))
 			adev->vm_manager.vm_update_mode =
 				AMDGPU_VM_USE_CPU_FOR_COMPUTE;
 		else
@@ -2668,6 +2866,7 @@ void amdgpu_vm_manager_init(struct amdgpu_device *adev)
 	adev->vm_manager.vm_update_mode = 0;
 #endif
 
+	xa_init_flags(&adev->vm_manager.pasids, XA_FLAGS_LOCK_IRQ);
 }
 
 /**
@@ -2679,40 +2878,40 @@ void amdgpu_vm_manager_init(struct amdgpu_device *adev)
  */
 void amdgpu_vm_manager_fini(struct amdgpu_device *adev)
 {
-	unsigned i, j;
+	WARN_ON(!xa_empty(&adev->vm_manager.pasids));
+	xa_destroy(&adev->vm_manager.pasids);
 
-	for (i = 0; i < AMDGPU_MAX_VMHUBS; ++i) {
-		struct amdgpu_vm_id_manager *id_mgr =
-			&adev->vm_manager.id_mgr[i];
-
-		mutex_destroy(&id_mgr->lock);
-		for (j = 0; j < AMDGPU_NUM_VM; ++j) {
-			struct amdgpu_vm_id *id = &id_mgr->ids[j];
-
-			amdgpu_sync_free(&id->active);
-			dma_fence_put(id->flushed_updates);
-			dma_fence_put(id->last_flush);
-		}
-	}
+	amdgpu_vmid_mgr_fini(adev);
 }
 
+/**
+ * amdgpu_vm_ioctl - Manages VMID reservation for vm hubs.
+ *
+ * @dev: drm device pointer
+ * @data: drm_amdgpu_vm
+ * @filp: drm file pointer
+ *
+ * Returns:
+ * 0 for success, -errno for errors.
+ */
 int amdgpu_vm_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 {
 	union drm_amdgpu_vm *args = data;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_fpriv *fpriv = filp->driver_priv;
-	int r;
+	struct amdgpu_vm *vm = &fpriv->vm;
+
+	/* No valid flags defined yet */
+	if (args->in.flags)
+		return -EINVAL;
 
 	switch (args->in.op) {
 	case AMDGPU_VM_OP_RESERVE_VMID:
-		/* current, we only have requirement to reserve vmid from gfxhub */
-		r = amdgpu_vm_alloc_reserved_vmid(adev, &fpriv->vm,
-						  AMDGPU_GFXHUB);
-		if (r)
-			return r;
+		/* We only have requirement to reserve vmid from gfxhub */
+		amdgpu_vmid_alloc_reserved(adev, vm, AMDGPU_GFXHUB(0));
 		break;
 	case AMDGPU_VM_OP_UNRESERVE_VMID:
-		amdgpu_vm_free_reserved_vmid(adev, &fpriv->vm, AMDGPU_GFXHUB);
+		amdgpu_vmid_free_reserved(adev, vm, AMDGPU_GFXHUB(0));
 		break;
 	default:
 		return -EINVAL;
@@ -2720,3 +2919,284 @@ int amdgpu_vm_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 
 	return 0;
 }
+
+/**
+ * amdgpu_vm_handle_fault - graceful handling of VM faults.
+ * @adev: amdgpu device pointer
+ * @pasid: PASID of the VM
+ * @ts: Timestamp of the fault
+ * @vmid: VMID, only used for GFX 9.4.3.
+ * @node_id: Node_id received in IH cookie. Only applicable for
+ *           GFX 9.4.3.
+ * @addr: Address of the fault
+ * @write_fault: true is write fault, false is read fault
+ *
+ * Try to gracefully handle a VM fault. Return true if the fault was handled and
+ * shouldn't be reported any more.
+ */
+bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid,
+			    u32 vmid, u32 node_id, uint64_t addr, uint64_t ts,
+			    bool write_fault)
+{
+	bool is_compute_context = false;
+	struct amdgpu_bo *root;
+	unsigned long irqflags;
+	uint64_t value, flags;
+	struct amdgpu_vm *vm;
+	int r;
+
+	xa_lock_irqsave(&adev->vm_manager.pasids, irqflags);
+	vm = xa_load(&adev->vm_manager.pasids, pasid);
+	if (vm) {
+		root = amdgpu_bo_ref(vm->root.bo);
+		is_compute_context = vm->is_compute_context;
+	} else {
+		root = NULL;
+	}
+	xa_unlock_irqrestore(&adev->vm_manager.pasids, irqflags);
+
+	if (!root)
+		return false;
+
+	addr /= AMDGPU_GPU_PAGE_SIZE;
+
+	if (is_compute_context && !svm_range_restore_pages(adev, pasid, vmid,
+	    node_id, addr, ts, write_fault)) {
+		amdgpu_bo_unref(&root);
+		return true;
+	}
+
+	r = amdgpu_bo_reserve(root, true);
+	if (r)
+		goto error_unref;
+
+	/* Double check that the VM still exists */
+	xa_lock_irqsave(&adev->vm_manager.pasids, irqflags);
+	vm = xa_load(&adev->vm_manager.pasids, pasid);
+	if (vm && vm->root.bo != root)
+		vm = NULL;
+	xa_unlock_irqrestore(&adev->vm_manager.pasids, irqflags);
+	if (!vm)
+		goto error_unlock;
+
+	flags = AMDGPU_PTE_VALID | AMDGPU_PTE_SNOOPED |
+		AMDGPU_PTE_SYSTEM;
+
+	if (is_compute_context) {
+		/* Intentionally setting invalid PTE flag
+		 * combination to force a no-retry-fault
+		 */
+		flags = AMDGPU_VM_NORETRY_FLAGS;
+		value = 0;
+	} else if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_NEVER) {
+		/* Redirect the access to the dummy page */
+		value = adev->dummy_page_addr;
+		flags |= AMDGPU_PTE_EXECUTABLE | AMDGPU_PTE_READABLE |
+			AMDGPU_PTE_WRITEABLE;
+
+	} else {
+		/* Let the hw retry silently on the PTE */
+		value = 0;
+	}
+
+	r = dma_resv_reserve_fences(root->tbo.base.resv, 1);
+	if (r) {
+		pr_debug("failed %d to reserve fence slot\n", r);
+		goto error_unlock;
+	}
+
+	r = amdgpu_vm_update_range(adev, vm, true, false, false, false,
+				   NULL, addr, addr, flags, value, 0, NULL, NULL, NULL);
+	if (r)
+		goto error_unlock;
+
+	r = amdgpu_vm_update_pdes(adev, vm, true);
+
+error_unlock:
+	amdgpu_bo_unreserve(root);
+	if (r < 0)
+		dev_err(adev->dev, "Can't handle page fault (%d)\n", r);
+
+error_unref:
+	amdgpu_bo_unref(&root);
+
+	return false;
+}
+
+#if defined(CONFIG_DEBUG_FS)
+/**
+ * amdgpu_debugfs_vm_bo_info  - print BO info for the VM
+ *
+ * @vm: Requested VM for printing BO info
+ * @m: debugfs file
+ *
+ * Print BO information in debugfs file for the VM
+ */
+void amdgpu_debugfs_vm_bo_info(struct amdgpu_vm *vm, struct seq_file *m)
+{
+	struct amdgpu_bo_va *bo_va, *tmp;
+	u64 total_idle = 0;
+	u64 total_evicted = 0;
+	u64 total_relocated = 0;
+	u64 total_moved = 0;
+	u64 total_invalidated = 0;
+	u64 total_done = 0;
+	unsigned int total_idle_objs = 0;
+	unsigned int total_evicted_objs = 0;
+	unsigned int total_relocated_objs = 0;
+	unsigned int total_moved_objs = 0;
+	unsigned int total_invalidated_objs = 0;
+	unsigned int total_done_objs = 0;
+	unsigned int id = 0;
+
+	amdgpu_vm_assert_locked(vm);
+
+	spin_lock(&vm->status_lock);
+	seq_puts(m, "\tIdle BOs:\n");
+	list_for_each_entry_safe(bo_va, tmp, &vm->idle, base.vm_status) {
+		if (!bo_va->base.bo)
+			continue;
+		total_idle += amdgpu_bo_print_info(id++, bo_va->base.bo, m);
+	}
+	total_idle_objs = id;
+	id = 0;
+
+	seq_puts(m, "\tEvicted BOs:\n");
+	list_for_each_entry_safe(bo_va, tmp, &vm->evicted, base.vm_status) {
+		if (!bo_va->base.bo)
+			continue;
+		total_evicted += amdgpu_bo_print_info(id++, bo_va->base.bo, m);
+	}
+	total_evicted_objs = id;
+	id = 0;
+
+	seq_puts(m, "\tRelocated BOs:\n");
+	list_for_each_entry_safe(bo_va, tmp, &vm->relocated, base.vm_status) {
+		if (!bo_va->base.bo)
+			continue;
+		total_relocated += amdgpu_bo_print_info(id++, bo_va->base.bo, m);
+	}
+	total_relocated_objs = id;
+	id = 0;
+
+	seq_puts(m, "\tMoved BOs:\n");
+	list_for_each_entry_safe(bo_va, tmp, &vm->moved, base.vm_status) {
+		if (!bo_va->base.bo)
+			continue;
+		total_moved += amdgpu_bo_print_info(id++, bo_va->base.bo, m);
+	}
+	total_moved_objs = id;
+	id = 0;
+
+	seq_puts(m, "\tInvalidated BOs:\n");
+	list_for_each_entry_safe(bo_va, tmp, &vm->invalidated, base.vm_status) {
+		if (!bo_va->base.bo)
+			continue;
+		total_invalidated += amdgpu_bo_print_info(id++,	bo_va->base.bo, m);
+	}
+	total_invalidated_objs = id;
+	id = 0;
+
+	seq_puts(m, "\tDone BOs:\n");
+	list_for_each_entry_safe(bo_va, tmp, &vm->done, base.vm_status) {
+		if (!bo_va->base.bo)
+			continue;
+		total_done += amdgpu_bo_print_info(id++, bo_va->base.bo, m);
+	}
+	spin_unlock(&vm->status_lock);
+	total_done_objs = id;
+
+	seq_printf(m, "\tTotal idle size:        %12lld\tobjs:\t%d\n", total_idle,
+		   total_idle_objs);
+	seq_printf(m, "\tTotal evicted size:     %12lld\tobjs:\t%d\n", total_evicted,
+		   total_evicted_objs);
+	seq_printf(m, "\tTotal relocated size:   %12lld\tobjs:\t%d\n", total_relocated,
+		   total_relocated_objs);
+	seq_printf(m, "\tTotal moved size:       %12lld\tobjs:\t%d\n", total_moved,
+		   total_moved_objs);
+	seq_printf(m, "\tTotal invalidated size: %12lld\tobjs:\t%d\n", total_invalidated,
+		   total_invalidated_objs);
+	seq_printf(m, "\tTotal done size:        %12lld\tobjs:\t%d\n", total_done,
+		   total_done_objs);
+}
+#endif
+
+/**
+ * amdgpu_vm_update_fault_cache - update cached fault into.
+ * @adev: amdgpu device pointer
+ * @pasid: PASID of the VM
+ * @addr: Address of the fault
+ * @status: GPUVM fault status register
+ * @vmhub: which vmhub got the fault
+ *
+ * Cache the fault info for later use by userspace in debugging.
+ */
+void amdgpu_vm_update_fault_cache(struct amdgpu_device *adev,
+				  unsigned int pasid,
+				  uint64_t addr,
+				  uint32_t status,
+				  unsigned int vmhub)
+{
+	struct amdgpu_vm *vm;
+	unsigned long flags;
+
+	xa_lock_irqsave(&adev->vm_manager.pasids, flags);
+
+	vm = xa_load(&adev->vm_manager.pasids, pasid);
+	/* Don't update the fault cache if status is 0.  In the multiple
+	 * fault case, subsequent faults will return a 0 status which is
+	 * useless for userspace and replaces the useful fault status, so
+	 * only update if status is non-0.
+	 */
+	if (vm && status) {
+		vm->fault_info.addr = addr;
+		vm->fault_info.status = status;
+		/*
+		 * Update the fault information globally for later usage
+		 * when vm could be stale or freed.
+		 */
+		adev->vm_manager.fault_info.addr = addr;
+		adev->vm_manager.fault_info.vmhub = vmhub;
+		adev->vm_manager.fault_info.status = status;
+
+		if (AMDGPU_IS_GFXHUB(vmhub)) {
+			vm->fault_info.vmhub = AMDGPU_VMHUB_TYPE_GFX;
+			vm->fault_info.vmhub |=
+				(vmhub - AMDGPU_GFXHUB_START) << AMDGPU_VMHUB_IDX_SHIFT;
+		} else if (AMDGPU_IS_MMHUB0(vmhub)) {
+			vm->fault_info.vmhub = AMDGPU_VMHUB_TYPE_MM0;
+			vm->fault_info.vmhub |=
+				(vmhub - AMDGPU_MMHUB0_START) << AMDGPU_VMHUB_IDX_SHIFT;
+		} else if (AMDGPU_IS_MMHUB1(vmhub)) {
+			vm->fault_info.vmhub = AMDGPU_VMHUB_TYPE_MM1;
+			vm->fault_info.vmhub |=
+				(vmhub - AMDGPU_MMHUB1_START) << AMDGPU_VMHUB_IDX_SHIFT;
+		} else {
+			WARN_ONCE(1, "Invalid vmhub %u\n", vmhub);
+		}
+	}
+	xa_unlock_irqrestore(&adev->vm_manager.pasids, flags);
+}
+
+/**
+ * amdgpu_vm_is_bo_always_valid - check if the BO is VM always valid
+ *
+ * @vm: VM to test against.
+ * @bo: BO to be tested.
+ *
+ * Returns true if the BO shares the dma_resv object with the root PD and is
+ * always guaranteed to be valid inside the VM.
+ */
+bool amdgpu_vm_is_bo_always_valid(struct amdgpu_vm *vm, struct amdgpu_bo *bo)
+{
+	return bo && bo->tbo.base.resv == vm->root.bo->tbo.base.resv;
+}
+
+void amdgpu_vm_print_task_info(struct amdgpu_device *adev,
+			       struct amdgpu_task_info *task_info)
+{
+	dev_err(adev->dev,
+		" Process %s pid %d thread %s pid %d\n",
+		task_info->process_name, task_info->tgid,
+		task_info->task.comm, task_info->task.pid);
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
index 6716355403ec..cf0ec94e8a07 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
@@ -24,36 +24,43 @@
 #ifndef __AMDGPU_VM_H__
 #define __AMDGPU_VM_H__
 
+#include <linux/idr.h>
+#include <linux/kfifo.h>
 #include <linux/rbtree.h>
+#include <drm/gpu_scheduler.h>
+#include <drm/drm_file.h>
+#include <drm/ttm/ttm_bo.h>
+#include <linux/sched/mm.h>
 
-#include "gpu_scheduler.h"
 #include "amdgpu_sync.h"
 #include "amdgpu_ring.h"
+#include "amdgpu_ids.h"
+#include "amdgpu_ttm.h"
+
+struct drm_exec;
 
 struct amdgpu_bo_va;
 struct amdgpu_job;
 struct amdgpu_bo_list_entry;
+struct amdgpu_bo_vm;
 
 /*
  * GPUVM handling
  */
 
-/* maximum number of VMIDs */
-#define AMDGPU_NUM_VM	16
-
 /* Maximum number of PTEs the hardware can write with one command */
 #define AMDGPU_VM_MAX_UPDATE_SIZE	0x3FFFF
 
 /* number of entries in page table */
 #define AMDGPU_VM_PTE_COUNT(adev) (1 << (adev)->vm_manager.block_size)
 
-/* PTBs (Page Table Blocks) need to be aligned to 32K */
-#define AMDGPU_VM_PTB_ALIGN_SIZE   32768
-
 #define AMDGPU_PTE_VALID	(1ULL << 0)
 #define AMDGPU_PTE_SYSTEM	(1ULL << 1)
 #define AMDGPU_PTE_SNOOPED	(1ULL << 2)
 
+/* RV+ */
+#define AMDGPU_PTE_TMZ		(1ULL << 3)
+
 /* VI only */
 #define AMDGPU_PTE_EXECUTABLE	(1ULL << 4)
 
@@ -68,32 +75,125 @@ struct amdgpu_bo_list_entry;
 /* PDE is handled as PTE for VEGA10 */
 #define AMDGPU_PDE_PTE		(1ULL << 54)
 
-/* VEGA10 only */
-#define AMDGPU_PTE_MTYPE(a)    ((uint64_t)a << 57)
-#define AMDGPU_PTE_MTYPE_MASK	AMDGPU_PTE_MTYPE(3ULL)
-
-/* How to programm VM fault handling */
+#define AMDGPU_PTE_LOG          (1ULL << 55)
+
+/* PTE is handled as PDE for VEGA10 (Translate Further) */
+#define AMDGPU_PTE_TF		(1ULL << 56)
+
+/* MALL noalloc for sienna_cichlid, reserved for older ASICs  */
+#define AMDGPU_PTE_NOALLOC	(1ULL << 58)
+
+/* PDE Block Fragment Size for VEGA10 */
+#define AMDGPU_PDE_BFS(a)	((uint64_t)a << 59)
+
+/* Flag combination to set no-retry with TF disabled */
+#define AMDGPU_VM_NORETRY_FLAGS	(AMDGPU_PTE_EXECUTABLE | AMDGPU_PDE_PTE | \
+				AMDGPU_PTE_TF)
+
+/* Flag combination to set no-retry with TF enabled */
+#define AMDGPU_VM_NORETRY_FLAGS_TF (AMDGPU_PTE_VALID | AMDGPU_PTE_SYSTEM | \
+				   AMDGPU_PTE_PRT)
+/* For GFX9 */
+#define AMDGPU_PTE_MTYPE_VG10_SHIFT(mtype)	((uint64_t)(mtype) << 57)
+#define AMDGPU_PTE_MTYPE_VG10_MASK	AMDGPU_PTE_MTYPE_VG10_SHIFT(3ULL)
+#define AMDGPU_PTE_MTYPE_VG10(flags, mtype)			\
+	(((uint64_t)(flags) & (~AMDGPU_PTE_MTYPE_VG10_MASK)) |	\
+	  AMDGPU_PTE_MTYPE_VG10_SHIFT(mtype))
+
+#define AMDGPU_MTYPE_NC 0
+#define AMDGPU_MTYPE_CC 2
+
+#define AMDGPU_PTE_DEFAULT_ATC  (AMDGPU_PTE_SYSTEM      \
+                                | AMDGPU_PTE_SNOOPED    \
+                                | AMDGPU_PTE_EXECUTABLE \
+                                | AMDGPU_PTE_READABLE   \
+                                | AMDGPU_PTE_WRITEABLE  \
+                                | AMDGPU_PTE_MTYPE_VG10(AMDGPU_MTYPE_CC))
+
+/* gfx10 */
+#define AMDGPU_PTE_MTYPE_NV10_SHIFT(mtype)	((uint64_t)(mtype) << 48)
+#define AMDGPU_PTE_MTYPE_NV10_MASK     AMDGPU_PTE_MTYPE_NV10_SHIFT(7ULL)
+#define AMDGPU_PTE_MTYPE_NV10(flags, mtype)			\
+	(((uint64_t)(flags) & (~AMDGPU_PTE_MTYPE_NV10_MASK)) |	\
+	  AMDGPU_PTE_MTYPE_NV10_SHIFT(mtype))
+
+/* gfx12 */
+#define AMDGPU_PTE_PRT_GFX12		(1ULL << 56)
+#define AMDGPU_PTE_PRT_FLAG(adev)	\
+	((amdgpu_ip_version((adev), GC_HWIP, 0) >= IP_VERSION(12, 0, 0)) ? AMDGPU_PTE_PRT_GFX12 : AMDGPU_PTE_PRT)
+
+#define AMDGPU_PTE_MTYPE_GFX12_SHIFT(mtype)	((uint64_t)(mtype) << 54)
+#define AMDGPU_PTE_MTYPE_GFX12_MASK	AMDGPU_PTE_MTYPE_GFX12_SHIFT(3ULL)
+#define AMDGPU_PTE_MTYPE_GFX12(flags, mtype)				\
+	(((uint64_t)(flags) & (~AMDGPU_PTE_MTYPE_GFX12_MASK)) |	\
+	  AMDGPU_PTE_MTYPE_GFX12_SHIFT(mtype))
+
+#define AMDGPU_PTE_DCC			(1ULL << 58)
+#define AMDGPU_PTE_IS_PTE		(1ULL << 63)
+
+/* PDE Block Fragment Size for gfx v12 */
+#define AMDGPU_PDE_BFS_GFX12(a)		((uint64_t)((a) & 0x1fULL) << 58)
+#define AMDGPU_PDE_BFS_FLAG(adev, a)	\
+	((amdgpu_ip_version((adev), GC_HWIP, 0) >= IP_VERSION(12, 0, 0)) ? AMDGPU_PDE_BFS_GFX12(a) : AMDGPU_PDE_BFS(a))
+/* PDE is handled as PTE for gfx v12 */
+#define AMDGPU_PDE_PTE_GFX12		(1ULL << 63)
+#define AMDGPU_PDE_PTE_FLAG(adev)	\
+	((amdgpu_ip_version((adev), GC_HWIP, 0) >= IP_VERSION(12, 0, 0)) ? AMDGPU_PDE_PTE_GFX12 : AMDGPU_PDE_PTE)
+
+/* How to program VM fault handling */
 #define AMDGPU_VM_FAULT_STOP_NEVER	0
 #define AMDGPU_VM_FAULT_STOP_FIRST	1
 #define AMDGPU_VM_FAULT_STOP_ALWAYS	2
 
-/* max number of VMHUB */
-#define AMDGPU_MAX_VMHUBS			2
-#define AMDGPU_GFXHUB				0
-#define AMDGPU_MMHUB				1
-
-/* hardcode that limit for now */
-#define AMDGPU_VA_RESERVED_SIZE			(8 << 20)
-/* max vmids dedicated for process */
-#define AMDGPU_VM_MAX_RESERVED_VMID	1
+/* How much VRAM be reserved for page tables */
+#define AMDGPU_VM_RESERVED_VRAM		(8ULL << 20)
 
-#define AMDGPU_VM_CONTEXT_GFX 0
-#define AMDGPU_VM_CONTEXT_COMPUTE 1
+/*
+ * max number of VMHUB
+ * layout: max 8 GFXHUB + 4 MMHUB0 + 1 MMHUB1
+ */
+#define AMDGPU_MAX_VMHUBS			13
+#define AMDGPU_GFXHUB_START			0
+#define AMDGPU_MMHUB0_START			8
+#define AMDGPU_MMHUB1_START			12
+#define AMDGPU_GFXHUB(x)			(AMDGPU_GFXHUB_START + (x))
+#define AMDGPU_MMHUB0(x)			(AMDGPU_MMHUB0_START + (x))
+#define AMDGPU_MMHUB1(x)			(AMDGPU_MMHUB1_START + (x))
+
+#define AMDGPU_IS_GFXHUB(x) ((x) >= AMDGPU_GFXHUB_START && (x) < AMDGPU_MMHUB0_START)
+#define AMDGPU_IS_MMHUB0(x) ((x) >= AMDGPU_MMHUB0_START && (x) < AMDGPU_MMHUB1_START)
+#define AMDGPU_IS_MMHUB1(x) ((x) >= AMDGPU_MMHUB1_START && (x) < AMDGPU_MAX_VMHUBS)
+
+/* Reserve space at top/bottom of address space for kernel use */
+#define AMDGPU_VA_RESERVED_CSA_SIZE		(2ULL << 20)
+#define AMDGPU_VA_RESERVED_CSA_START(adev)	(((adev)->vm_manager.max_pfn \
+						  << AMDGPU_GPU_PAGE_SHIFT)  \
+						 - AMDGPU_VA_RESERVED_CSA_SIZE)
+#define AMDGPU_VA_RESERVED_SEQ64_SIZE		(2ULL << 20)
+#define AMDGPU_VA_RESERVED_SEQ64_START(adev)	(AMDGPU_VA_RESERVED_CSA_START(adev) \
+						 - AMDGPU_VA_RESERVED_SEQ64_SIZE)
+#define AMDGPU_VA_RESERVED_TRAP_SIZE		(2ULL << 12)
+#define AMDGPU_VA_RESERVED_TRAP_START(adev)	(AMDGPU_VA_RESERVED_SEQ64_START(adev) \
+						 - AMDGPU_VA_RESERVED_TRAP_SIZE)
+#define AMDGPU_VA_RESERVED_BOTTOM		(1ULL << 16)
+#define AMDGPU_VA_RESERVED_TOP			(AMDGPU_VA_RESERVED_TRAP_SIZE + \
+						 AMDGPU_VA_RESERVED_SEQ64_SIZE + \
+						 AMDGPU_VA_RESERVED_CSA_SIZE)
 
 /* See vm_update_mode */
 #define AMDGPU_VM_USE_CPU_FOR_GFX (1 << 0)
 #define AMDGPU_VM_USE_CPU_FOR_COMPUTE (1 << 1)
 
+/* VMPT level enumerate, and the hiberachy is:
+ * PDB2->PDB1->PDB0->PTB
+ */
+enum amdgpu_vm_level {
+	AMDGPU_VM_PDB2,
+	AMDGPU_VM_PDB1,
+	AMDGPU_VM_PDB0,
+	AMDGPU_VM_PTB
+};
+
 /* base structure for tracking BO usage in a VM */
 struct amdgpu_vm_bo_base {
 	/* constant after initialization */
@@ -101,91 +201,257 @@ struct amdgpu_vm_bo_base {
 	struct amdgpu_bo		*bo;
 
 	/* protected by bo being reserved */
-	struct list_head		bo_list;
+	struct amdgpu_vm_bo_base	*next;
 
-	/* protected by spinlock */
+	/* protected by vm status_lock */
 	struct list_head		vm_status;
+
+	/* if the bo is counted as shared in mem stats
+	 * protected by vm status_lock */
+	bool				shared;
+
+	/* protected by the BO being reserved */
+	bool				moved;
 };
 
-struct amdgpu_vm_pt {
-	struct amdgpu_bo	*bo;
-	uint64_t		addr;
+/* provided by hw blocks that can write ptes, e.g., sdma */
+struct amdgpu_vm_pte_funcs {
+	/* number of dw to reserve per operation */
+	unsigned	copy_pte_num_dw;
+
+	/* copy pte entries from GART */
+	void (*copy_pte)(struct amdgpu_ib *ib,
+			 uint64_t pe, uint64_t src,
+			 unsigned count);
+
+	/* write pte one entry at a time with addr mapping */
+	void (*write_pte)(struct amdgpu_ib *ib, uint64_t pe,
+			  uint64_t value, unsigned count,
+			  uint32_t incr);
+	/* for linear pte/pde updates without addr mapping */
+	void (*set_pte_pde)(struct amdgpu_ib *ib,
+			    uint64_t pe,
+			    uint64_t addr, unsigned count,
+			    uint32_t incr, uint64_t flags);
+};
 
-	/* array of page tables, one for each directory entry */
-	struct amdgpu_vm_pt	*entries;
-	unsigned		last_entry_used;
+struct amdgpu_task_info {
+	struct drm_wedge_task_info task;
+	char		process_name[TASK_COMM_LEN];
+	pid_t		tgid;
+	struct kref	refcount;
+};
+
+/**
+ * struct amdgpu_vm_update_params
+ *
+ * Encapsulate some VM table update parameters to reduce
+ * the number of function parameters
+ *
+ */
+struct amdgpu_vm_update_params {
+
+	/**
+	 * @adev: amdgpu device we do this update for
+	 */
+	struct amdgpu_device *adev;
+
+	/**
+	 * @vm: optional amdgpu_vm we do this update for
+	 */
+	struct amdgpu_vm *vm;
+
+	/**
+	 * @immediate: if changes should be made immediately
+	 */
+	bool immediate;
+
+	/**
+	 * @unlocked: true if the root BO is not locked
+	 */
+	bool unlocked;
+
+	/**
+	 * @pages_addr:
+	 *
+	 * DMA addresses to use for mapping
+	 */
+	dma_addr_t *pages_addr;
+
+	/**
+	 * @job: job to used for hw submission
+	 */
+	struct amdgpu_job *job;
+
+	/**
+	 * @num_dw_left: number of dw left for the IB
+	 */
+	unsigned int num_dw_left;
+
+	/**
+	 * @needs_flush: true whenever we need to invalidate the TLB
+	 */
+	bool needs_flush;
+
+	/**
+	 * @allow_override: true for memory that is not uncached: allows MTYPE
+	 * to be overridden for NUMA local memory.
+	 */
+	bool allow_override;
+
+	/**
+	 * @tlb_flush_waitlist: temporary storage for BOs until tlb_flush
+	 */
+	struct list_head tlb_flush_waitlist;
+};
+
+struct amdgpu_vm_update_funcs {
+	int (*map_table)(struct amdgpu_bo_vm *bo);
+	int (*prepare)(struct amdgpu_vm_update_params *p,
+		       struct amdgpu_sync *sync, u64 k_job_id);
+	int (*update)(struct amdgpu_vm_update_params *p,
+		      struct amdgpu_bo_vm *bo, uint64_t pe, uint64_t addr,
+		      unsigned count, uint32_t incr, uint64_t flags);
+	int (*commit)(struct amdgpu_vm_update_params *p,
+		      struct dma_fence **fence);
+};
+
+struct amdgpu_vm_fault_info {
+	/* fault address */
+	uint64_t	addr;
+	/* fault status register */
+	uint32_t	status;
+	/* which vmhub? gfxhub, mmhub, etc. */
+	unsigned int	vmhub;
+};
+
+struct amdgpu_mem_stats {
+	struct drm_memory_stats drm;
+
+	/* buffers that requested this placement but are currently evicted */
+	uint64_t evicted;
 };
 
 struct amdgpu_vm {
 	/* tree of virtual addresses mapped */
 	struct rb_root_cached	va;
 
-	/* protecting invalidated */
+	/* Lock to prevent eviction while we are updating page tables
+	 * use vm_eviction_lock/unlock(vm)
+	 */
+	struct mutex		eviction_lock;
+	bool			evicting;
+	unsigned int		saved_flags;
+
+	/* Lock to protect vm_bo add/del/move on all lists of vm */
 	spinlock_t		status_lock;
 
-	/* BOs moved, but not yet updated in the PT */
+	/* Memory statistics for this vm, protected by status_lock */
+	struct amdgpu_mem_stats stats[__AMDGPU_PL_NUM];
+
+	/*
+	 * The following lists contain amdgpu_vm_bo_base objects for either
+	 * PDs, PTs or per VM BOs. The state transits are:
+	 *
+	 * evicted -> relocated (PDs, PTs) or moved (per VM BOs) -> idle
+	 */
+
+	/* Per-VM and PT BOs who needs a validation */
+	struct list_head	evicted;
+
+	/* PT BOs which relocated and their parent need an update */
+	struct list_head	relocated;
+
+	/* per VM BOs moved, but not yet updated in the PT */
 	struct list_head	moved;
 
-	/* BOs cleared in the PT because of a move */
-	struct list_head	cleared;
+	/* All BOs of this VM not currently in the state machine */
+	struct list_head	idle;
 
-	/* BO mappings freed, but not yet updated in the PT */
+	/*
+	 * The following lists contain amdgpu_vm_bo_base objects for BOs which
+	 * have their own dma_resv object and not depend on the root PD. Their
+	 * state transits are:
+	 *
+	 * evicted_user or invalidated -> done
+	 */
+
+	/* BOs for user mode queues that need a validation */
+	struct list_head	evicted_user;
+
+	/* regular invalidated BOs, but not yet updated in the PT */
+	struct list_head	invalidated;
+
+	/* BOs which are invalidated, has been updated in the PTs */
+	struct list_head        done;
+
+	/*
+	 * This list contains amdgpu_bo_va_mapping objects which have been freed
+	 * but not updated in the PTs
+	 */
 	struct list_head	freed;
 
 	/* contains the page directory */
-	struct amdgpu_vm_pt     root;
-	struct dma_fence	*last_dir_update;
-	uint64_t		last_eviction_counter;
+	struct amdgpu_vm_bo_base     root;
+	struct dma_fence	*last_update;
 
-	/* protecting freed */
-	spinlock_t		freed_lock;
+	/* Scheduler entities for page table updates */
+	struct drm_sched_entity	immediate;
+	struct drm_sched_entity	delayed;
 
-	/* Scheduler entity for page table updates */
-	struct amd_sched_entity	entity;
+	/* Last finished delayed update */
+	atomic64_t		tlb_seq;
+	struct dma_fence	*last_tlb_flush;
+	atomic64_t		kfd_last_flushed_seq;
+	uint64_t		tlb_fence_context;
 
-	/* client id */
-	u64                     client_id;
-	/* dedicated to vm */
-	struct amdgpu_vm_id	*reserved_vmid[AMDGPU_MAX_VMHUBS];
+	/* How many times we had to re-generate the page tables */
+	uint64_t		generation;
+
+	/* Last unlocked submission to the scheduler entities */
+	struct dma_fence	*last_unlocked;
+
+	unsigned int		pasid;
+	struct amdgpu_vmid	*reserved_vmid[AMDGPU_MAX_VMHUBS];
 
 	/* Flag to indicate if VM tables are updated by CPU or GPU (SDMA) */
-	bool                    use_cpu_for_update;
+	bool					use_cpu_for_update;
 
-	/* Flag to indicate ATS support from PTE for GFX9 */
-	bool			pte_support_ats;
-};
+	/* Functions to use for VM table updates */
+	const struct amdgpu_vm_update_funcs	*update_funcs;
 
-struct amdgpu_vm_id {
-	struct list_head	list;
-	struct amdgpu_sync	active;
-	struct dma_fence		*last_flush;
-	atomic64_t		owner;
+	/* Up to 128 pending retry page faults */
+	DECLARE_KFIFO(faults, u64, 128);
 
-	uint64_t		pd_gpu_addr;
-	/* last flushed PD/PT update */
-	struct dma_fence		*flushed_updates;
+	/* Points to the KFD process VM info */
+	struct amdkfd_process_info *process_info;
 
-	uint32_t                current_gpu_reset_count;
+	/* List node in amdkfd_process_info.vm_list_head */
+	struct list_head	vm_list_node;
 
-	uint32_t		gds_base;
-	uint32_t		gds_size;
-	uint32_t		gws_base;
-	uint32_t		gws_size;
-	uint32_t		oa_base;
-	uint32_t		oa_size;
-};
+	/* Valid while the PD is reserved or fenced */
+	uint64_t		pd_phys_addr;
+
+	/* Some basic info about the task */
+	struct amdgpu_task_info *task_info;
+
+	/* Store positions of group of BOs */
+	struct ttm_lru_bulk_move lru_bulk_move;
+	/* Flag to indicate if VM is used for compute */
+	bool			is_compute_context;
 
-struct amdgpu_vm_id_manager {
-	struct mutex		lock;
-	unsigned		num_ids;
-	struct list_head	ids_lru;
-	struct amdgpu_vm_id	ids[AMDGPU_NUM_VM];
-	atomic_t		reserved_vmid_num;
+	/* Memory partition number, -1 means any partition */
+	int8_t			mem_id;
+
+	/* cached fault info */
+	struct amdgpu_vm_fault_info fault_info;
 };
 
 struct amdgpu_vm_manager {
 	/* Handling of VMIDs */
-	struct amdgpu_vm_id_manager		id_mgr[AMDGPU_MAX_VMHUBS];
+	struct amdgpu_vmid_mgr			id_mgr[AMDGPU_MAX_VMHUBS];
+	unsigned int				first_kfd_vmid;
+	bool					concurrent_flush;
 
 	/* Handling of VM fences */
 	u64					fence_context;
@@ -193,18 +459,16 @@ struct amdgpu_vm_manager {
 
 	uint64_t				max_pfn;
 	uint32_t				num_level;
-	uint64_t				vm_size;
 	uint32_t				block_size;
 	uint32_t				fragment_size;
+	enum amdgpu_vm_level			root_level;
 	/* vram base address for page table entry  */
 	u64					vram_base_offset;
 	/* vm pte handling */
-	const struct amdgpu_vm_pte_funcs        *vm_pte_funcs;
-	struct amdgpu_ring                      *vm_pte_rings[AMDGPU_MAX_RINGS];
-	unsigned				vm_pte_num_rings;
-	atomic_t				vm_pte_next_ring;
-	/* client id counter */
-	atomic64_t				client_counter;
+	const struct amdgpu_vm_pte_funcs	*vm_pte_funcs;
+	struct drm_gpu_scheduler		*vm_pte_scheds[AMDGPU_MAX_RINGS];
+	unsigned				vm_pte_num_scheds;
+	struct amdgpu_ring			*page_fault;
 
 	/* partial resident texture handling */
 	spinlock_t				prt_lock;
@@ -215,41 +479,74 @@ struct amdgpu_vm_manager {
 	 * BIT1[= 0] Compute updated by SDMA [= 1] by CPU
 	 */
 	int					vm_update_mode;
+
+	/* PASID to VM mapping, will be used in interrupt context to
+	 * look up VM of a page fault
+	 */
+	struct xarray				pasids;
+	/* Global registration of recent page fault information */
+	struct amdgpu_vm_fault_info	fault_info;
 };
 
+struct amdgpu_bo_va_mapping;
+
+#define amdgpu_vm_copy_pte(adev, ib, pe, src, count) ((adev)->vm_manager.vm_pte_funcs->copy_pte((ib), (pe), (src), (count)))
+#define amdgpu_vm_write_pte(adev, ib, pe, value, count, incr) ((adev)->vm_manager.vm_pte_funcs->write_pte((ib), (pe), (value), (count), (incr)))
+#define amdgpu_vm_set_pte_pde(adev, ib, pe, addr, count, incr, flags) ((adev)->vm_manager.vm_pte_funcs->set_pte_pde((ib), (pe), (addr), (count), (incr), (flags)))
+
+extern const struct amdgpu_vm_update_funcs amdgpu_vm_cpu_funcs;
+extern const struct amdgpu_vm_update_funcs amdgpu_vm_sdma_funcs;
+
 void amdgpu_vm_manager_init(struct amdgpu_device *adev);
 void amdgpu_vm_manager_fini(struct amdgpu_device *adev);
-int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
-		   int vm_context);
+
+long amdgpu_vm_wait_idle(struct amdgpu_vm *vm, long timeout);
+int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm, int32_t xcp_id, uint32_t pasid);
+int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm);
 void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm);
-void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm,
-			 struct list_head *validated,
-			 struct amdgpu_bo_list_entry *entry);
-int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm,
-			      int (*callback)(void *p, struct amdgpu_bo *bo),
-			      void *param);
-int amdgpu_vm_alloc_pts(struct amdgpu_device *adev,
-			struct amdgpu_vm *vm,
-			uint64_t saddr, uint64_t size);
-int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
-		      struct amdgpu_sync *sync, struct dma_fence *fence,
-		      struct amdgpu_job *job);
+int amdgpu_vm_lock_pd(struct amdgpu_vm *vm, struct drm_exec *exec,
+		      unsigned int num_fences);
+int amdgpu_vm_lock_done_list(struct amdgpu_vm *vm, struct drm_exec *exec,
+			     unsigned int num_fences);
+bool amdgpu_vm_ready(struct amdgpu_vm *vm);
+uint64_t amdgpu_vm_generation(struct amdgpu_device *adev, struct amdgpu_vm *vm);
+int amdgpu_vm_validate(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+		       struct ww_acquire_ctx *ticket,
+		       int (*callback)(void *p, struct amdgpu_bo *bo),
+		       void *param);
 int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job, bool need_pipe_sync);
-void amdgpu_vm_reset_id(struct amdgpu_device *adev, unsigned vmhub,
-			unsigned vmid);
-void amdgpu_vm_reset_all_ids(struct amdgpu_device *adev);
-int amdgpu_vm_update_directories(struct amdgpu_device *adev,
-				 struct amdgpu_vm *vm);
+int amdgpu_vm_update_pdes(struct amdgpu_device *adev,
+			  struct amdgpu_vm *vm, bool immediate);
 int amdgpu_vm_clear_freed(struct amdgpu_device *adev,
 			  struct amdgpu_vm *vm,
 			  struct dma_fence **fence);
-int amdgpu_vm_clear_moved(struct amdgpu_device *adev, struct amdgpu_vm *vm,
-			  struct amdgpu_sync *sync);
+int amdgpu_vm_handle_moved(struct amdgpu_device *adev,
+			   struct amdgpu_vm *vm,
+			   struct ww_acquire_ctx *ticket);
+int amdgpu_vm_flush_compute_tlb(struct amdgpu_device *adev,
+				struct amdgpu_vm *vm,
+				uint32_t flush_type,
+				uint32_t xcc_mask);
+void amdgpu_vm_bo_base_init(struct amdgpu_vm_bo_base *base,
+			    struct amdgpu_vm *vm, struct amdgpu_bo *bo);
+int amdgpu_vm_update_range(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			   bool immediate, bool unlocked, bool flush_tlb,
+			   bool allow_override, struct amdgpu_sync *sync,
+			   uint64_t start, uint64_t last, uint64_t flags,
+			   uint64_t offset, uint64_t vram_base,
+			   struct ttm_resource *res, dma_addr_t *pages_addr,
+			   struct dma_fence **fence);
 int amdgpu_vm_bo_update(struct amdgpu_device *adev,
 			struct amdgpu_bo_va *bo_va,
 			bool clear);
-void amdgpu_vm_bo_invalidate(struct amdgpu_device *adev,
-			     struct amdgpu_bo *bo);
+bool amdgpu_vm_evictable(struct amdgpu_bo *bo);
+void amdgpu_vm_bo_invalidate(struct amdgpu_bo *bo, bool evicted);
+void amdgpu_vm_update_stats(struct amdgpu_vm_bo_base *base,
+			    struct ttm_resource *new_res, int sign);
+void amdgpu_vm_bo_update_shared(struct amdgpu_bo *bo);
+void amdgpu_vm_bo_move(struct amdgpu_bo *bo, struct ttm_resource *new_mem,
+		       bool evicted);
+uint64_t amdgpu_vm_map_gart(const dma_addr_t *pages_addr, uint64_t addr);
 struct amdgpu_bo_va *amdgpu_vm_bo_find(struct amdgpu_vm *vm,
 				       struct amdgpu_bo *bo);
 struct amdgpu_bo_va *amdgpu_vm_bo_add(struct amdgpu_device *adev,
@@ -258,26 +555,141 @@ struct amdgpu_bo_va *amdgpu_vm_bo_add(struct amdgpu_device *adev,
 int amdgpu_vm_bo_map(struct amdgpu_device *adev,
 		     struct amdgpu_bo_va *bo_va,
 		     uint64_t addr, uint64_t offset,
-		     uint64_t size, uint64_t flags);
+		     uint64_t size, uint32_t flags);
 int amdgpu_vm_bo_replace_map(struct amdgpu_device *adev,
 			     struct amdgpu_bo_va *bo_va,
 			     uint64_t addr, uint64_t offset,
-			     uint64_t size, uint64_t flags);
+			     uint64_t size, uint32_t flags);
 int amdgpu_vm_bo_unmap(struct amdgpu_device *adev,
 		       struct amdgpu_bo_va *bo_va,
 		       uint64_t addr);
 int amdgpu_vm_bo_clear_mappings(struct amdgpu_device *adev,
 				struct amdgpu_vm *vm,
 				uint64_t saddr, uint64_t size);
-void amdgpu_vm_bo_rmv(struct amdgpu_device *adev,
+struct amdgpu_bo_va_mapping *amdgpu_vm_bo_lookup_mapping(struct amdgpu_vm *vm,
+							 uint64_t addr);
+void amdgpu_vm_bo_trace_cs(struct amdgpu_vm *vm, struct ww_acquire_ctx *ticket);
+void amdgpu_vm_bo_del(struct amdgpu_device *adev,
 		      struct amdgpu_bo_va *bo_va);
-void amdgpu_vm_set_fragment_size(struct amdgpu_device *adev,
-				uint32_t fragment_size_default);
-void amdgpu_vm_adjust_size(struct amdgpu_device *adev, uint64_t vm_size,
-				uint32_t fragment_size_default);
+void amdgpu_vm_adjust_size(struct amdgpu_device *adev, uint32_t min_vm_size,
+			   uint32_t fragment_size_default, unsigned max_level,
+			   unsigned max_bits);
 int amdgpu_vm_ioctl(struct drm_device *dev, void *data, struct drm_file *filp);
 bool amdgpu_vm_need_pipeline_sync(struct amdgpu_ring *ring,
 				  struct amdgpu_job *job);
 void amdgpu_vm_check_compute_bug(struct amdgpu_device *adev);
 
+struct amdgpu_task_info *
+amdgpu_vm_get_task_info_pasid(struct amdgpu_device *adev, u32 pasid);
+
+struct amdgpu_task_info *
+amdgpu_vm_get_task_info_vm(struct amdgpu_vm *vm);
+
+void amdgpu_vm_put_task_info(struct amdgpu_task_info *task_info);
+
+bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid,
+			    u32 vmid, u32 node_id, uint64_t addr, uint64_t ts,
+			    bool write_fault);
+
+void amdgpu_vm_set_task_info(struct amdgpu_vm *vm);
+
+void amdgpu_vm_move_to_lru_tail(struct amdgpu_device *adev,
+				struct amdgpu_vm *vm);
+void amdgpu_vm_get_memory(struct amdgpu_vm *vm,
+			  struct amdgpu_mem_stats stats[__AMDGPU_PL_NUM]);
+
+int amdgpu_vm_pt_clear(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+		       struct amdgpu_bo_vm *vmbo, bool immediate);
+int amdgpu_vm_pt_create(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			int level, bool immediate, struct amdgpu_bo_vm **vmbo,
+			int32_t xcp_id);
+void amdgpu_vm_pt_free_root(struct amdgpu_device *adev, struct amdgpu_vm *vm);
+
+int amdgpu_vm_pde_update(struct amdgpu_vm_update_params *params,
+			 struct amdgpu_vm_bo_base *entry);
+int amdgpu_vm_ptes_update(struct amdgpu_vm_update_params *params,
+			  uint64_t start, uint64_t end,
+			  uint64_t dst, uint64_t flags);
+void amdgpu_vm_pt_free_work(struct work_struct *work);
+void amdgpu_vm_pt_free_list(struct amdgpu_device *adev,
+			    struct amdgpu_vm_update_params *params);
+
+#if defined(CONFIG_DEBUG_FS)
+void amdgpu_debugfs_vm_bo_info(struct amdgpu_vm *vm, struct seq_file *m);
+#endif
+
+int amdgpu_vm_pt_map_tables(struct amdgpu_device *adev, struct amdgpu_vm *vm);
+
+bool amdgpu_vm_is_bo_always_valid(struct amdgpu_vm *vm, struct amdgpu_bo *bo);
+
+/**
+ * amdgpu_vm_tlb_seq - return tlb flush sequence number
+ * @vm: the amdgpu_vm structure to query
+ *
+ * Returns the tlb flush sequence number which indicates that the VM TLBs needs
+ * to be invalidated whenever the sequence number change.
+ */
+static inline uint64_t amdgpu_vm_tlb_seq(struct amdgpu_vm *vm)
+{
+	unsigned long flags;
+	spinlock_t *lock;
+
+	/*
+	 * Workaround to stop racing between the fence signaling and handling
+	 * the cb. The lock is static after initially setting it up, just make
+	 * sure that the dma_fence structure isn't freed up.
+	 */
+	rcu_read_lock();
+	lock = vm->last_tlb_flush->lock;
+	rcu_read_unlock();
+
+	spin_lock_irqsave(lock, flags);
+	spin_unlock_irqrestore(lock, flags);
+
+	return atomic64_read(&vm->tlb_seq);
+}
+
+/*
+ * vm eviction_lock can be taken in MMU notifiers. Make sure no reclaim-FS
+ * happens while holding this lock anywhere to prevent deadlocks when
+ * an MMU notifier runs in reclaim-FS context.
+ */
+static inline void amdgpu_vm_eviction_lock(struct amdgpu_vm *vm)
+{
+	mutex_lock(&vm->eviction_lock);
+	vm->saved_flags = memalloc_noreclaim_save();
+}
+
+static inline bool amdgpu_vm_eviction_trylock(struct amdgpu_vm *vm)
+{
+	if (mutex_trylock(&vm->eviction_lock)) {
+		vm->saved_flags = memalloc_noreclaim_save();
+		return true;
+	}
+	return false;
+}
+
+static inline void amdgpu_vm_eviction_unlock(struct amdgpu_vm *vm)
+{
+	memalloc_noreclaim_restore(vm->saved_flags);
+	mutex_unlock(&vm->eviction_lock);
+}
+
+void amdgpu_vm_update_fault_cache(struct amdgpu_device *adev,
+				  unsigned int pasid,
+				  uint64_t addr,
+				  uint32_t status,
+				  unsigned int vmhub);
+void amdgpu_vm_tlb_fence_create(struct amdgpu_device *adev,
+				 struct amdgpu_vm *vm,
+				 struct dma_fence **fence);
+
+void amdgpu_vm_print_task_info(struct amdgpu_device *adev,
+			       struct amdgpu_task_info *task_info);
+
+#define amdgpu_vm_bo_va_for_each_valid_mapping(bo_va, mapping) \
+		list_for_each_entry(mapping, &(bo_va)->valids, list)
+#define amdgpu_vm_bo_va_for_each_invalid_mapping(bo_va, mapping) \
+		list_for_each_entry(mapping, &(bo_va)->invalids, list)
+
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_cpu.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_cpu.c
new file mode 100644
index 000000000000..22e2e5b47341
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_cpu.c
@@ -0,0 +1,124 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include "amdgpu_vm.h"
+#include "amdgpu_object.h"
+#include "amdgpu_trace.h"
+
+/**
+ * amdgpu_vm_cpu_map_table - make sure new PDs/PTs are kmapped
+ *
+ * @table: newly allocated or validated PD/PT
+ */
+static int amdgpu_vm_cpu_map_table(struct amdgpu_bo_vm *table)
+{
+	table->bo.flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
+	return amdgpu_bo_kmap(&table->bo, NULL);
+}
+
+/**
+ * amdgpu_vm_cpu_prepare - prepare page table update with the CPU
+ *
+ * @p: see amdgpu_vm_update_params definition
+ * @sync: sync obj with fences to wait on
+ * @k_job_id: the id for tracing/debug purposes
+ *
+ * Returns:
+ * Negativ errno, 0 for success.
+ */
+static int amdgpu_vm_cpu_prepare(struct amdgpu_vm_update_params *p,
+				 struct amdgpu_sync *sync,
+				 u64 k_job_id)
+{
+	if (!sync)
+		return 0;
+
+	return amdgpu_sync_wait(sync, true);
+}
+
+/**
+ * amdgpu_vm_cpu_update - helper to update page tables via CPU
+ *
+ * @p: see amdgpu_vm_update_params definition
+ * @vmbo: PD/PT to update
+ * @pe: byte offset of the PDE/PTE, relative to start of PDB/PTB
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: hw access flags
+ *
+ * Write count number of PT/PD entries directly.
+ */
+static int amdgpu_vm_cpu_update(struct amdgpu_vm_update_params *p,
+				struct amdgpu_bo_vm *vmbo, uint64_t pe,
+				uint64_t addr, unsigned count, uint32_t incr,
+				uint64_t flags)
+{
+	unsigned int i;
+	uint64_t value;
+	long r;
+
+	r = dma_resv_wait_timeout(vmbo->bo.tbo.base.resv, DMA_RESV_USAGE_KERNEL,
+				  true, MAX_SCHEDULE_TIMEOUT);
+	if (r < 0)
+		return r;
+
+	pe += (unsigned long)amdgpu_bo_kptr(&vmbo->bo);
+
+	trace_amdgpu_vm_set_ptes(pe, addr, count, incr, flags, p->immediate);
+
+	for (i = 0; i < count; i++) {
+		value = p->pages_addr ?
+			amdgpu_vm_map_gart(p->pages_addr, addr) :
+			addr;
+		amdgpu_gmc_set_pte_pde(p->adev, (void *)(uintptr_t)pe,
+				       i, value, flags);
+		addr += incr;
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_vm_cpu_commit - commit page table update to the HW
+ *
+ * @p: see amdgpu_vm_update_params definition
+ * @fence: unused
+ *
+ * Make sure that the hardware sees the page table updates.
+ */
+static int amdgpu_vm_cpu_commit(struct amdgpu_vm_update_params *p,
+				struct dma_fence **fence)
+{
+	if (p->needs_flush)
+		atomic64_inc(&p->vm->tlb_seq);
+
+	mb();
+	amdgpu_device_flush_hdp(p->adev, NULL);
+	return 0;
+}
+
+const struct amdgpu_vm_update_funcs amdgpu_vm_cpu_funcs = {
+	.map_table = amdgpu_vm_cpu_map_table,
+	.prepare = amdgpu_vm_cpu_prepare,
+	.update = amdgpu_vm_cpu_update,
+	.commit = amdgpu_vm_cpu_commit
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c
new file mode 100644
index 000000000000..f794fb1cc06e
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c
@@ -0,0 +1,976 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include <drm/drm_drv.h>
+
+#include "amdgpu.h"
+#include "amdgpu_trace.h"
+#include "amdgpu_vm.h"
+#include "amdgpu_job.h"
+
+/*
+ * amdgpu_vm_pt_cursor - state for for_each_amdgpu_vm_pt
+ */
+struct amdgpu_vm_pt_cursor {
+	uint64_t pfn;
+	struct amdgpu_vm_bo_base *parent;
+	struct amdgpu_vm_bo_base *entry;
+	unsigned int level;
+};
+
+/**
+ * amdgpu_vm_pt_level_shift - return the addr shift for each level
+ *
+ * @adev: amdgpu_device pointer
+ * @level: VMPT level
+ *
+ * Returns:
+ * The number of bits the pfn needs to be right shifted for a level.
+ */
+static unsigned int amdgpu_vm_pt_level_shift(struct amdgpu_device *adev,
+					     unsigned int level)
+{
+	switch (level) {
+	case AMDGPU_VM_PDB2:
+	case AMDGPU_VM_PDB1:
+	case AMDGPU_VM_PDB0:
+		return 9 * (AMDGPU_VM_PDB0 - level) +
+			adev->vm_manager.block_size;
+	case AMDGPU_VM_PTB:
+		return 0;
+	default:
+		return ~0;
+	}
+}
+
+/**
+ * amdgpu_vm_pt_num_entries - return the number of entries in a PD/PT
+ *
+ * @adev: amdgpu_device pointer
+ * @level: VMPT level
+ *
+ * Returns:
+ * The number of entries in a page directory or page table.
+ */
+static unsigned int amdgpu_vm_pt_num_entries(struct amdgpu_device *adev,
+					     unsigned int level)
+{
+	unsigned int shift;
+
+	shift = amdgpu_vm_pt_level_shift(adev, adev->vm_manager.root_level);
+	if (level == adev->vm_manager.root_level)
+		/* For the root directory */
+		return round_up(adev->vm_manager.max_pfn, 1ULL << shift)
+			>> shift;
+	else if (level != AMDGPU_VM_PTB)
+		/* Everything in between */
+		return 512;
+
+	/* For the page tables on the leaves */
+	return AMDGPU_VM_PTE_COUNT(adev);
+}
+
+/**
+ * amdgpu_vm_pt_entries_mask - the mask to get the entry number of a PD/PT
+ *
+ * @adev: amdgpu_device pointer
+ * @level: VMPT level
+ *
+ * Returns:
+ * The mask to extract the entry number of a PD/PT from an address.
+ */
+static uint32_t amdgpu_vm_pt_entries_mask(struct amdgpu_device *adev,
+					  unsigned int level)
+{
+	if (level <= adev->vm_manager.root_level)
+		return 0xffffffff;
+	else if (level != AMDGPU_VM_PTB)
+		return 0x1ff;
+	else
+		return AMDGPU_VM_PTE_COUNT(adev) - 1;
+}
+
+/**
+ * amdgpu_vm_pt_size - returns the size of the page table in bytes
+ *
+ * @adev: amdgpu_device pointer
+ * @level: VMPT level
+ *
+ * Returns:
+ * The size of the BO for a page directory or page table in bytes.
+ */
+static unsigned int amdgpu_vm_pt_size(struct amdgpu_device *adev,
+				      unsigned int level)
+{
+	return AMDGPU_GPU_PAGE_ALIGN(amdgpu_vm_pt_num_entries(adev, level) * 8);
+}
+
+/**
+ * amdgpu_vm_pt_parent - get the parent page directory
+ *
+ * @pt: child page table
+ *
+ * Helper to get the parent entry for the child page table. NULL if we are at
+ * the root page directory.
+ */
+static struct amdgpu_vm_bo_base *
+amdgpu_vm_pt_parent(struct amdgpu_vm_bo_base *pt)
+{
+	struct amdgpu_bo *parent = pt->bo->parent;
+
+	if (!parent)
+		return NULL;
+
+	return parent->vm_bo;
+}
+
+/**
+ * amdgpu_vm_pt_start - start PD/PT walk
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: amdgpu_vm structure
+ * @start: start address of the walk
+ * @cursor: state to initialize
+ *
+ * Initialize a amdgpu_vm_pt_cursor to start a walk.
+ */
+static void amdgpu_vm_pt_start(struct amdgpu_device *adev,
+			       struct amdgpu_vm *vm, uint64_t start,
+			       struct amdgpu_vm_pt_cursor *cursor)
+{
+	cursor->pfn = start;
+	cursor->parent = NULL;
+	cursor->entry = &vm->root;
+	cursor->level = adev->vm_manager.root_level;
+}
+
+/**
+ * amdgpu_vm_pt_descendant - go to child node
+ *
+ * @adev: amdgpu_device pointer
+ * @cursor: current state
+ *
+ * Walk to the child node of the current node.
+ * Returns:
+ * True if the walk was possible, false otherwise.
+ */
+static bool amdgpu_vm_pt_descendant(struct amdgpu_device *adev,
+				    struct amdgpu_vm_pt_cursor *cursor)
+{
+	unsigned int mask, shift, idx;
+
+	if ((cursor->level == AMDGPU_VM_PTB) || !cursor->entry ||
+	    !cursor->entry->bo)
+		return false;
+
+	mask = amdgpu_vm_pt_entries_mask(adev, cursor->level);
+	shift = amdgpu_vm_pt_level_shift(adev, cursor->level);
+
+	++cursor->level;
+	idx = (cursor->pfn >> shift) & mask;
+	cursor->parent = cursor->entry;
+	cursor->entry = &to_amdgpu_bo_vm(cursor->entry->bo)->entries[idx];
+	return true;
+}
+
+/**
+ * amdgpu_vm_pt_sibling - go to sibling node
+ *
+ * @adev: amdgpu_device pointer
+ * @cursor: current state
+ *
+ * Walk to the sibling node of the current node.
+ * Returns:
+ * True if the walk was possible, false otherwise.
+ */
+static bool amdgpu_vm_pt_sibling(struct amdgpu_device *adev,
+				 struct amdgpu_vm_pt_cursor *cursor)
+{
+
+	unsigned int shift, num_entries;
+	struct amdgpu_bo_vm *parent;
+
+	/* Root doesn't have a sibling */
+	if (!cursor->parent)
+		return false;
+
+	/* Go to our parents and see if we got a sibling */
+	shift = amdgpu_vm_pt_level_shift(adev, cursor->level - 1);
+	num_entries = amdgpu_vm_pt_num_entries(adev, cursor->level - 1);
+	parent = to_amdgpu_bo_vm(cursor->parent->bo);
+
+	if (cursor->entry == &parent->entries[num_entries - 1])
+		return false;
+
+	cursor->pfn += 1ULL << shift;
+	cursor->pfn &= ~((1ULL << shift) - 1);
+	++cursor->entry;
+	return true;
+}
+
+/**
+ * amdgpu_vm_pt_ancestor - go to parent node
+ *
+ * @cursor: current state
+ *
+ * Walk to the parent node of the current node.
+ * Returns:
+ * True if the walk was possible, false otherwise.
+ */
+static bool amdgpu_vm_pt_ancestor(struct amdgpu_vm_pt_cursor *cursor)
+{
+	if (!cursor->parent)
+		return false;
+
+	--cursor->level;
+	cursor->entry = cursor->parent;
+	cursor->parent = amdgpu_vm_pt_parent(cursor->parent);
+	return true;
+}
+
+/**
+ * amdgpu_vm_pt_next - get next PD/PT in hieratchy
+ *
+ * @adev: amdgpu_device pointer
+ * @cursor: current state
+ *
+ * Walk the PD/PT tree to the next node.
+ */
+static void amdgpu_vm_pt_next(struct amdgpu_device *adev,
+			      struct amdgpu_vm_pt_cursor *cursor)
+{
+	/* First try a newborn child */
+	if (amdgpu_vm_pt_descendant(adev, cursor))
+		return;
+
+	/* If that didn't worked try to find a sibling */
+	while (!amdgpu_vm_pt_sibling(adev, cursor)) {
+		/* No sibling, go to our parents and grandparents */
+		if (!amdgpu_vm_pt_ancestor(cursor)) {
+			cursor->pfn = ~0ll;
+			return;
+		}
+	}
+}
+
+/**
+ * amdgpu_vm_pt_first_dfs - start a deep first search
+ *
+ * @adev: amdgpu_device structure
+ * @vm: amdgpu_vm structure
+ * @start: optional cursor to start with
+ * @cursor: state to initialize
+ *
+ * Starts a deep first traversal of the PD/PT tree.
+ */
+static void amdgpu_vm_pt_first_dfs(struct amdgpu_device *adev,
+				   struct amdgpu_vm *vm,
+				   struct amdgpu_vm_pt_cursor *start,
+				   struct amdgpu_vm_pt_cursor *cursor)
+{
+	if (start)
+		*cursor = *start;
+	else
+		amdgpu_vm_pt_start(adev, vm, 0, cursor);
+
+	while (amdgpu_vm_pt_descendant(adev, cursor))
+		;
+}
+
+/**
+ * amdgpu_vm_pt_continue_dfs - check if the deep first search should continue
+ *
+ * @start: starting point for the search
+ * @entry: current entry
+ *
+ * Returns:
+ * True when the search should continue, false otherwise.
+ */
+static bool amdgpu_vm_pt_continue_dfs(struct amdgpu_vm_pt_cursor *start,
+				      struct amdgpu_vm_bo_base *entry)
+{
+	return entry && (!start || entry != start->entry);
+}
+
+/**
+ * amdgpu_vm_pt_next_dfs - get the next node for a deep first search
+ *
+ * @adev: amdgpu_device structure
+ * @cursor: current state
+ *
+ * Move the cursor to the next node in a deep first search.
+ */
+static void amdgpu_vm_pt_next_dfs(struct amdgpu_device *adev,
+				  struct amdgpu_vm_pt_cursor *cursor)
+{
+	if (!cursor->entry)
+		return;
+
+	if (!cursor->parent)
+		cursor->entry = NULL;
+	else if (amdgpu_vm_pt_sibling(adev, cursor))
+		while (amdgpu_vm_pt_descendant(adev, cursor))
+			;
+	else
+		amdgpu_vm_pt_ancestor(cursor);
+}
+
+/*
+ * for_each_amdgpu_vm_pt_dfs_safe - safe deep first search of all PDs/PTs
+ */
+#define for_each_amdgpu_vm_pt_dfs_safe(adev, vm, start, cursor, entry)		\
+	for (amdgpu_vm_pt_first_dfs((adev), (vm), (start), &(cursor)),		\
+	     (entry) = (cursor).entry, amdgpu_vm_pt_next_dfs((adev), &(cursor));\
+	     amdgpu_vm_pt_continue_dfs((start), (entry));			\
+	     (entry) = (cursor).entry, amdgpu_vm_pt_next_dfs((adev), &(cursor)))
+
+/**
+ * amdgpu_vm_pt_clear - initially clear the PDs/PTs
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: VM to clear BO from
+ * @vmbo: BO to clear
+ * @immediate: use an immediate update
+ *
+ * Root PD needs to be reserved when calling this.
+ *
+ * Returns:
+ * 0 on success, errno otherwise.
+ */
+int amdgpu_vm_pt_clear(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+		       struct amdgpu_bo_vm *vmbo, bool immediate)
+{
+	unsigned int level = adev->vm_manager.root_level;
+	struct ttm_operation_ctx ctx = { true, false };
+	struct amdgpu_vm_update_params params;
+	struct amdgpu_bo *ancestor = &vmbo->bo;
+	unsigned int entries;
+	struct amdgpu_bo *bo = &vmbo->bo;
+	uint64_t addr;
+	int r, idx;
+
+	/* Figure out our place in the hierarchy */
+	if (ancestor->parent) {
+		++level;
+		while (ancestor->parent->parent) {
+			++level;
+			ancestor = ancestor->parent;
+		}
+	}
+
+	entries = amdgpu_bo_size(bo) / 8;
+
+	r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+	if (r)
+		return r;
+
+	if (!drm_dev_enter(adev_to_drm(adev), &idx))
+		return -ENODEV;
+
+	r = vm->update_funcs->map_table(vmbo);
+	if (r)
+		goto exit;
+
+	memset(&params, 0, sizeof(params));
+	params.adev = adev;
+	params.vm = vm;
+	params.immediate = immediate;
+
+	r = vm->update_funcs->prepare(&params, NULL,
+				      AMDGPU_KERNEL_JOB_ID_VM_PT_CLEAR);
+	if (r)
+		goto exit;
+
+	addr = 0;
+
+	uint64_t value = 0, flags = 0;
+	if (adev->asic_type >= CHIP_VEGA10) {
+		if (level != AMDGPU_VM_PTB) {
+			/* Handle leaf PDEs as PTEs */
+			flags |= AMDGPU_PDE_PTE_FLAG(adev);
+			amdgpu_gmc_get_vm_pde(adev, level,
+					      &value, &flags);
+		} else {
+			/* Workaround for fault priority problem on GMC9 */
+			flags = AMDGPU_PTE_EXECUTABLE;
+		}
+	}
+
+	r = vm->update_funcs->update(&params, vmbo, addr, 0, entries,
+				     value, flags);
+	if (r)
+		goto exit;
+
+	r = vm->update_funcs->commit(&params, NULL);
+exit:
+	drm_dev_exit(idx);
+	return r;
+}
+
+/**
+ * amdgpu_vm_pt_create - create bo for PD/PT
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: requesting vm
+ * @level: the page table level
+ * @immediate: use a immediate update
+ * @vmbo: pointer to the buffer object pointer
+ * @xcp_id: GPU partition id
+ */
+int amdgpu_vm_pt_create(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			int level, bool immediate, struct amdgpu_bo_vm **vmbo,
+			int32_t xcp_id)
+{
+	struct amdgpu_bo_param bp;
+	unsigned int num_entries;
+
+	memset(&bp, 0, sizeof(bp));
+
+	bp.size = amdgpu_vm_pt_size(adev, level);
+	bp.byte_align = AMDGPU_GPU_PAGE_SIZE;
+
+	if (!adev->gmc.is_app_apu)
+		bp.domain = AMDGPU_GEM_DOMAIN_VRAM;
+	else
+		bp.domain = AMDGPU_GEM_DOMAIN_GTT;
+
+	bp.domain = amdgpu_bo_get_preferred_domain(adev, bp.domain);
+	bp.flags = AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS |
+		AMDGPU_GEM_CREATE_CPU_GTT_USWC;
+
+	if (level < AMDGPU_VM_PTB)
+		num_entries = amdgpu_vm_pt_num_entries(adev, level);
+	else
+		num_entries = 0;
+
+	bp.bo_ptr_size = struct_size((*vmbo), entries, num_entries);
+
+	if (vm->use_cpu_for_update)
+		bp.flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
+
+	bp.type = ttm_bo_type_kernel;
+	bp.no_wait_gpu = immediate;
+	bp.xcp_id_plus1 = xcp_id + 1;
+
+	if (vm->root.bo)
+		bp.resv = vm->root.bo->tbo.base.resv;
+
+	return amdgpu_bo_create_vm(adev, &bp, vmbo);
+}
+
+/**
+ * amdgpu_vm_pt_alloc - Allocate a specific page table
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: VM to allocate page tables for
+ * @cursor: Which page table to allocate
+ * @immediate: use an immediate update
+ *
+ * Make sure a specific page table or directory is allocated.
+ *
+ * Returns:
+ * 1 if page table needed to be allocated, 0 if page table was already
+ * allocated, negative errno if an error occurred.
+ */
+static int amdgpu_vm_pt_alloc(struct amdgpu_device *adev,
+			      struct amdgpu_vm *vm,
+			      struct amdgpu_vm_pt_cursor *cursor,
+			      bool immediate)
+{
+	struct amdgpu_vm_bo_base *entry = cursor->entry;
+	struct amdgpu_bo *pt_bo;
+	struct amdgpu_bo_vm *pt;
+	int r;
+
+	if (entry->bo)
+		return 0;
+
+	amdgpu_vm_eviction_unlock(vm);
+	r = amdgpu_vm_pt_create(adev, vm, cursor->level, immediate, &pt,
+				vm->root.bo->xcp_id);
+	amdgpu_vm_eviction_lock(vm);
+	if (r)
+		return r;
+
+	/* Keep a reference to the root directory to avoid
+	 * freeing them up in the wrong order.
+	 */
+	pt_bo = &pt->bo;
+	pt_bo->parent = amdgpu_bo_ref(cursor->parent->bo);
+	amdgpu_vm_bo_base_init(entry, vm, pt_bo);
+	r = amdgpu_vm_pt_clear(adev, vm, pt, immediate);
+	if (r)
+		goto error_free_pt;
+
+	return 0;
+
+error_free_pt:
+	amdgpu_bo_unref(&pt_bo);
+	return r;
+}
+
+/**
+ * amdgpu_vm_pt_free - free one PD/PT
+ *
+ * @entry: PDE to free
+ */
+static void amdgpu_vm_pt_free(struct amdgpu_vm_bo_base *entry)
+{
+	if (!entry->bo)
+		return;
+
+	amdgpu_vm_update_stats(entry, entry->bo->tbo.resource, -1);
+	entry->bo->vm_bo = NULL;
+	ttm_bo_set_bulk_move(&entry->bo->tbo, NULL);
+
+	spin_lock(&entry->vm->status_lock);
+	list_del(&entry->vm_status);
+	spin_unlock(&entry->vm->status_lock);
+	amdgpu_bo_unref(&entry->bo);
+}
+
+/**
+ * amdgpu_vm_pt_free_list - free PD/PT levels
+ *
+ * @adev: amdgpu device structure
+ * @params: see amdgpu_vm_update_params definition
+ *
+ * Free the page directory objects saved in the flush list
+ */
+void amdgpu_vm_pt_free_list(struct amdgpu_device *adev,
+			    struct amdgpu_vm_update_params *params)
+{
+	struct amdgpu_vm_bo_base *entry, *next;
+	bool unlocked = params->unlocked;
+
+	if (list_empty(&params->tlb_flush_waitlist))
+		return;
+
+	/*
+	 * unlocked unmap clear page table leaves, warning to free the page entry.
+	 */
+	WARN_ON(unlocked);
+
+	list_for_each_entry_safe(entry, next, &params->tlb_flush_waitlist, vm_status)
+		amdgpu_vm_pt_free(entry);
+}
+
+/**
+ * amdgpu_vm_pt_add_list - add PD/PT level to the flush list
+ *
+ * @params: parameters for the update
+ * @cursor: first PT entry to start DF search from, non NULL
+ *
+ * This list will be freed after TLB flush.
+ */
+static void amdgpu_vm_pt_add_list(struct amdgpu_vm_update_params *params,
+				  struct amdgpu_vm_pt_cursor *cursor)
+{
+	struct amdgpu_vm_pt_cursor seek;
+	struct amdgpu_vm_bo_base *entry;
+
+	spin_lock(&params->vm->status_lock);
+	for_each_amdgpu_vm_pt_dfs_safe(params->adev, params->vm, cursor, seek, entry) {
+		if (entry && entry->bo)
+			list_move(&entry->vm_status, &params->tlb_flush_waitlist);
+	}
+
+	/* enter start node now */
+	list_move(&cursor->entry->vm_status, &params->tlb_flush_waitlist);
+	spin_unlock(&params->vm->status_lock);
+}
+
+/**
+ * amdgpu_vm_pt_free_root - free root PD
+ * @adev: amdgpu device structure
+ * @vm: amdgpu vm structure
+ *
+ * Free the root page directory and everything below it.
+ */
+void amdgpu_vm_pt_free_root(struct amdgpu_device *adev, struct amdgpu_vm *vm)
+{
+	struct amdgpu_vm_pt_cursor cursor;
+	struct amdgpu_vm_bo_base *entry;
+
+	for_each_amdgpu_vm_pt_dfs_safe(adev, vm, NULL, cursor, entry) {
+		if (entry)
+			amdgpu_vm_pt_free(entry);
+	}
+}
+
+/**
+ * amdgpu_vm_pde_update - update a single level in the hierarchy
+ *
+ * @params: parameters for the update
+ * @entry: entry to update
+ *
+ * Makes sure the requested entry in parent is up to date.
+ */
+int amdgpu_vm_pde_update(struct amdgpu_vm_update_params *params,
+			 struct amdgpu_vm_bo_base *entry)
+{
+	struct amdgpu_vm_bo_base *parent = amdgpu_vm_pt_parent(entry);
+	struct amdgpu_bo *bo, *pbo;
+	struct amdgpu_vm *vm = params->vm;
+	uint64_t pde, pt, flags;
+	unsigned int level;
+
+	if (WARN_ON(!parent))
+		return -EINVAL;
+
+	bo = parent->bo;
+	for (level = 0, pbo = bo->parent; pbo; ++level)
+		pbo = pbo->parent;
+
+	level += params->adev->vm_manager.root_level;
+	amdgpu_gmc_get_pde_for_bo(entry->bo, level, &pt, &flags);
+	pde = (entry - to_amdgpu_bo_vm(parent->bo)->entries) * 8;
+	return vm->update_funcs->update(params, to_amdgpu_bo_vm(bo), pde, pt,
+					1, 0, flags);
+}
+
+/**
+ * amdgpu_vm_pte_update_noretry_flags - Update PTE no-retry flags
+ *
+ * @adev: amdgpu_device pointer
+ * @flags: pointer to PTE flags
+ *
+ * Update PTE no-retry flags when TF is enabled.
+ */
+static void amdgpu_vm_pte_update_noretry_flags(struct amdgpu_device *adev,
+						uint64_t *flags)
+{
+	/*
+	 * Update no-retry flags with the corresponding TF
+	 * no-retry combination.
+	 */
+	if ((*flags & AMDGPU_VM_NORETRY_FLAGS) == AMDGPU_VM_NORETRY_FLAGS) {
+		*flags &= ~AMDGPU_VM_NORETRY_FLAGS;
+		*flags |= adev->gmc.noretry_flags;
+	}
+}
+
+/*
+ * amdgpu_vm_pte_update_flags - figure out flags for PTE updates
+ *
+ * Make sure to set the right flags for the PTEs at the desired level.
+ */
+static void amdgpu_vm_pte_update_flags(struct amdgpu_vm_update_params *params,
+				       struct amdgpu_bo_vm *pt,
+				       unsigned int level,
+				       uint64_t pe, uint64_t addr,
+				       unsigned int count, uint32_t incr,
+				       uint64_t flags)
+{
+	struct amdgpu_device *adev = params->adev;
+
+	if (level != AMDGPU_VM_PTB) {
+		flags |= AMDGPU_PDE_PTE_FLAG(params->adev);
+		amdgpu_gmc_get_vm_pde(adev, level, &addr, &flags);
+
+	} else if (adev->asic_type >= CHIP_VEGA10 &&
+		   !(flags & AMDGPU_PTE_VALID) &&
+		   !(flags & AMDGPU_PTE_PRT_FLAG(params->adev))) {
+
+		/* Workaround for fault priority problem on GMC9 */
+		flags |= AMDGPU_PTE_EXECUTABLE;
+	}
+
+	/*
+	 * Update no-retry flags to use the no-retry flag combination
+	 * with TF enabled. The AMDGPU_VM_NORETRY_FLAGS flag combination
+	 * does not work when TF is enabled. So, replace them with
+	 * AMDGPU_VM_NORETRY_FLAGS_TF flag combination which works for
+	 * all cases.
+	 */
+	if (level == AMDGPU_VM_PTB)
+		amdgpu_vm_pte_update_noretry_flags(adev, &flags);
+
+	/* APUs mapping system memory may need different MTYPEs on different
+	 * NUMA nodes. Only do this for contiguous ranges that can be assumed
+	 * to be on the same NUMA node.
+	 */
+	if ((flags & AMDGPU_PTE_SYSTEM) && (adev->flags & AMD_IS_APU) &&
+	    adev->gmc.gmc_funcs->override_vm_pte_flags &&
+	    num_possible_nodes() > 1 && !params->pages_addr && params->allow_override)
+		amdgpu_gmc_override_vm_pte_flags(adev, params->vm, addr, &flags);
+
+	params->vm->update_funcs->update(params, pt, pe, addr, count, incr,
+					 flags);
+}
+
+/**
+ * amdgpu_vm_pte_fragment - get fragment for PTEs
+ *
+ * @params: see amdgpu_vm_update_params definition
+ * @start: first PTE to handle
+ * @end: last PTE to handle
+ * @flags: hw mapping flags
+ * @frag: resulting fragment size
+ * @frag_end: end of this fragment
+ *
+ * Returns the first possible fragment for the start and end address.
+ */
+static void amdgpu_vm_pte_fragment(struct amdgpu_vm_update_params *params,
+				   uint64_t start, uint64_t end, uint64_t flags,
+				   unsigned int *frag, uint64_t *frag_end)
+{
+	/**
+	 * The MC L1 TLB supports variable sized pages, based on a fragment
+	 * field in the PTE. When this field is set to a non-zero value, page
+	 * granularity is increased from 4KB to (1 << (12 + frag)). The PTE
+	 * flags are considered valid for all PTEs within the fragment range
+	 * and corresponding mappings are assumed to be physically contiguous.
+	 *
+	 * The L1 TLB can store a single PTE for the whole fragment,
+	 * significantly increasing the space available for translation
+	 * caching. This leads to large improvements in throughput when the
+	 * TLB is under pressure.
+	 *
+	 * The L2 TLB distributes small and large fragments into two
+	 * asymmetric partitions. The large fragment cache is significantly
+	 * larger. Thus, we try to use large fragments wherever possible.
+	 * Userspace can support this by aligning virtual base address and
+	 * allocation size to the fragment size.
+	 *
+	 * Starting with Vega10 the fragment size only controls the L1. The L2
+	 * is now directly feed with small/huge/giant pages from the walker.
+	 */
+	unsigned int max_frag;
+
+	if (params->adev->asic_type < CHIP_VEGA10)
+		max_frag = params->adev->vm_manager.fragment_size;
+	else
+		max_frag = 31;
+
+	/* system pages are non continuously */
+	if (params->pages_addr) {
+		*frag = 0;
+		*frag_end = end;
+		return;
+	}
+
+	/* This intentionally wraps around if no bit is set */
+	*frag = min_t(unsigned int, ffs(start) - 1, fls64(end - start) - 1);
+	if (*frag >= max_frag) {
+		*frag = max_frag;
+		*frag_end = end & ~((1ULL << max_frag) - 1);
+	} else {
+		*frag_end = start + (1 << *frag);
+	}
+}
+
+/**
+ * amdgpu_vm_ptes_update - make sure that page tables are valid
+ *
+ * @params: see amdgpu_vm_update_params definition
+ * @start: start of GPU address range
+ * @end: end of GPU address range
+ * @dst: destination address to map to, the next dst inside the function
+ * @flags: mapping flags
+ *
+ * Update the page tables in the range @start - @end.
+ *
+ * Returns:
+ * 0 for success, -EINVAL for failure.
+ */
+int amdgpu_vm_ptes_update(struct amdgpu_vm_update_params *params,
+			  uint64_t start, uint64_t end,
+			  uint64_t dst, uint64_t flags)
+{
+	struct amdgpu_device *adev = params->adev;
+	struct amdgpu_vm_pt_cursor cursor;
+	uint64_t frag_start = start, frag_end;
+	unsigned int frag;
+	int r;
+
+	/* figure out the initial fragment */
+	amdgpu_vm_pte_fragment(params, frag_start, end, flags, &frag,
+			       &frag_end);
+
+	/* walk over the address space and update the PTs */
+	amdgpu_vm_pt_start(adev, params->vm, start, &cursor);
+	while (cursor.pfn < end) {
+		unsigned int shift, parent_shift, mask;
+		uint64_t incr, entry_end, pe_start;
+		struct amdgpu_bo *pt;
+
+		if (!params->unlocked) {
+			/* make sure that the page tables covering the
+			 * address range are actually allocated
+			 */
+			r = amdgpu_vm_pt_alloc(params->adev, params->vm,
+					       &cursor, params->immediate);
+			if (r)
+				return r;
+		}
+
+		shift = amdgpu_vm_pt_level_shift(adev, cursor.level);
+		parent_shift = amdgpu_vm_pt_level_shift(adev, cursor.level - 1);
+		if (params->unlocked) {
+			/* Unlocked updates are only allowed on the leaves */
+			if (amdgpu_vm_pt_descendant(adev, &cursor))
+				continue;
+		} else if (adev->asic_type < CHIP_VEGA10 &&
+			   (flags & AMDGPU_PTE_VALID)) {
+			/* No huge page support before GMC v9 */
+			if (cursor.level != AMDGPU_VM_PTB) {
+				if (!amdgpu_vm_pt_descendant(adev, &cursor))
+					return -ENOENT;
+				continue;
+			}
+		} else if (frag < shift) {
+			/* We can't use this level when the fragment size is
+			 * smaller than the address shift. Go to the next
+			 * child entry and try again.
+			 */
+			if (amdgpu_vm_pt_descendant(adev, &cursor))
+				continue;
+		} else if (frag >= parent_shift) {
+			/* If the fragment size is even larger than the parent
+			 * shift we should go up one level and check it again.
+			 */
+			if (!amdgpu_vm_pt_ancestor(&cursor))
+				return -EINVAL;
+			continue;
+		}
+
+		pt = cursor.entry->bo;
+		if (!pt) {
+			/* We need all PDs and PTs for mapping something, */
+			if (flags & AMDGPU_PTE_VALID)
+				return -ENOENT;
+
+			/* but unmapping something can happen at a higher
+			 * level.
+			 */
+			if (!amdgpu_vm_pt_ancestor(&cursor))
+				return -EINVAL;
+
+			pt = cursor.entry->bo;
+			shift = parent_shift;
+			frag_end = max(frag_end, ALIGN(frag_start + 1,
+				   1ULL << shift));
+		}
+
+		/* Looks good so far, calculate parameters for the update */
+		incr = (uint64_t)AMDGPU_GPU_PAGE_SIZE << shift;
+		mask = amdgpu_vm_pt_entries_mask(adev, cursor.level);
+		pe_start = ((cursor.pfn >> shift) & mask) * 8;
+
+		if (cursor.level < AMDGPU_VM_PTB && params->unlocked)
+			/*
+			 * MMU notifier callback unlocked unmap huge page, leave is PDE entry,
+			 * only clear one entry. Next entry search again for PDE or PTE leave.
+			 */
+			entry_end = 1ULL << shift;
+		else
+			entry_end = ((uint64_t)mask + 1) << shift;
+		entry_end += cursor.pfn & ~(entry_end - 1);
+		entry_end = min(entry_end, end);
+
+		do {
+			struct amdgpu_vm *vm = params->vm;
+			uint64_t upd_end = min(entry_end, frag_end);
+			unsigned int nptes = (upd_end - frag_start) >> shift;
+			uint64_t upd_flags = flags | AMDGPU_PTE_FRAG(frag);
+
+			/* This can happen when we set higher level PDs to
+			 * silent to stop fault floods.
+			 */
+			nptes = max(nptes, 1u);
+
+			trace_amdgpu_vm_update_ptes(params, frag_start, upd_end,
+						    min(nptes, 32u), dst, incr,
+						    upd_flags,
+						    vm->task_info ? vm->task_info->tgid : 0,
+						    vm->immediate.fence_context);
+			amdgpu_vm_pte_update_flags(params, to_amdgpu_bo_vm(pt),
+						   cursor.level, pe_start, dst,
+						   nptes, incr, upd_flags);
+
+			pe_start += nptes * 8;
+			dst += nptes * incr;
+
+			frag_start = upd_end;
+			if (frag_start >= frag_end) {
+				/* figure out the next fragment */
+				amdgpu_vm_pte_fragment(params, frag_start, end,
+						       flags, &frag, &frag_end);
+				if (frag < shift)
+					break;
+			}
+		} while (frag_start < entry_end);
+
+		if (amdgpu_vm_pt_descendant(adev, &cursor)) {
+			/* Free all child entries.
+			 * Update the tables with the flags and addresses and free up subsequent
+			 * tables in the case of huge pages or freed up areas.
+			 * This is the maximum you can free, because all other page tables are not
+			 * completely covered by the range and so potentially still in use.
+			 */
+			while (cursor.pfn < frag_start) {
+				/* Make sure previous mapping is freed */
+				if (cursor.entry->bo) {
+					params->needs_flush = true;
+					amdgpu_vm_pt_add_list(params, &cursor);
+				}
+				amdgpu_vm_pt_next(adev, &cursor);
+			}
+
+		} else if (frag >= shift) {
+			/* or just move on to the next on the same level. */
+			amdgpu_vm_pt_next(adev, &cursor);
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_vm_pt_map_tables - have bo of root PD cpu accessible
+ * @adev: amdgpu device structure
+ * @vm: amdgpu vm structure
+ *
+ * make root page directory and everything below it cpu accessible.
+ */
+int amdgpu_vm_pt_map_tables(struct amdgpu_device *adev, struct amdgpu_vm *vm)
+{
+	struct amdgpu_vm_pt_cursor cursor;
+	struct amdgpu_vm_bo_base *entry;
+
+	for_each_amdgpu_vm_pt_dfs_safe(adev, vm, NULL, cursor, entry) {
+
+		struct amdgpu_bo_vm *bo;
+		int r;
+
+		if (entry->bo) {
+			bo = to_amdgpu_bo_vm(entry->bo);
+			r = vm->update_funcs->map_table(bo);
+			if (r)
+				return r;
+		}
+	}
+
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c
new file mode 100644
index 000000000000..36805dcfa159
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c
@@ -0,0 +1,296 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include "amdgpu_vm.h"
+#include "amdgpu_job.h"
+#include "amdgpu_object.h"
+#include "amdgpu_trace.h"
+
+#define AMDGPU_VM_SDMA_MIN_NUM_DW	256u
+#define AMDGPU_VM_SDMA_MAX_NUM_DW	(16u * 1024u)
+
+/**
+ * amdgpu_vm_sdma_map_table - make sure new PDs/PTs are GTT mapped
+ *
+ * @table: newly allocated or validated PD/PT
+ */
+static int amdgpu_vm_sdma_map_table(struct amdgpu_bo_vm *table)
+{
+	return amdgpu_ttm_alloc_gart(&table->bo.tbo);
+}
+
+/* Allocate a new job for @count PTE updates */
+static int amdgpu_vm_sdma_alloc_job(struct amdgpu_vm_update_params *p,
+				    unsigned int count, u64 k_job_id)
+{
+	enum amdgpu_ib_pool_type pool = p->immediate ? AMDGPU_IB_POOL_IMMEDIATE
+		: AMDGPU_IB_POOL_DELAYED;
+	struct drm_sched_entity *entity = p->immediate ? &p->vm->immediate
+		: &p->vm->delayed;
+	unsigned int ndw;
+	int r;
+
+	/* estimate how many dw we need */
+	ndw = AMDGPU_VM_SDMA_MIN_NUM_DW;
+	if (p->pages_addr)
+		ndw += count * 2;
+	ndw = min(ndw, AMDGPU_VM_SDMA_MAX_NUM_DW);
+
+	r = amdgpu_job_alloc_with_ib(p->adev, entity, AMDGPU_FENCE_OWNER_VM,
+				     ndw * 4, pool, &p->job, k_job_id);
+	if (r)
+		return r;
+
+	p->num_dw_left = ndw;
+	return 0;
+}
+
+/**
+ * amdgpu_vm_sdma_prepare - prepare SDMA command submission
+ *
+ * @p: see amdgpu_vm_update_params definition
+ * @sync: amdgpu_sync object with fences to wait for
+ * @k_job_id: identifier of the job, for tracing purpose
+ *
+ * Returns:
+ * Negativ errno, 0 for success.
+ */
+static int amdgpu_vm_sdma_prepare(struct amdgpu_vm_update_params *p,
+				  struct amdgpu_sync *sync, u64 k_job_id)
+{
+	int r;
+
+	r = amdgpu_vm_sdma_alloc_job(p, 0, k_job_id);
+	if (r)
+		return r;
+
+	if (!sync)
+		return 0;
+
+	r = amdgpu_sync_push_to_job(sync, p->job);
+	if (r) {
+		p->num_dw_left = 0;
+		amdgpu_job_free(p->job);
+	}
+	return r;
+}
+
+/**
+ * amdgpu_vm_sdma_commit - commit SDMA command submission
+ *
+ * @p: see amdgpu_vm_update_params definition
+ * @fence: resulting fence
+ *
+ * Returns:
+ * Negativ errno, 0 for success.
+ */
+static int amdgpu_vm_sdma_commit(struct amdgpu_vm_update_params *p,
+				 struct dma_fence **fence)
+{
+	struct amdgpu_ib *ib = p->job->ibs;
+	struct amdgpu_ring *ring;
+	struct dma_fence *f;
+
+	ring = container_of(p->vm->delayed.rq->sched, struct amdgpu_ring,
+			    sched);
+
+	WARN_ON(ib->length_dw == 0);
+	amdgpu_ring_pad_ib(ring, ib);
+
+	if (p->needs_flush)
+		atomic64_inc(&p->vm->tlb_seq);
+
+	WARN_ON(ib->length_dw > p->num_dw_left);
+	f = amdgpu_job_submit(p->job);
+
+	if (p->unlocked) {
+		struct dma_fence *tmp = dma_fence_get(f);
+
+		swap(p->vm->last_unlocked, tmp);
+		dma_fence_put(tmp);
+	} else {
+		dma_resv_add_fence(p->vm->root.bo->tbo.base.resv, f,
+				   DMA_RESV_USAGE_BOOKKEEP);
+	}
+
+	if (fence && !p->immediate) {
+		/*
+		 * Most hw generations now have a separate queue for page table
+		 * updates, but when the queue is shared with userspace we need
+		 * the extra CPU round trip to correctly flush the TLB.
+		 */
+		set_bit(DRM_SCHED_FENCE_DONT_PIPELINE, &f->flags);
+		swap(*fence, f);
+	}
+	dma_fence_put(f);
+	return 0;
+}
+
+/**
+ * amdgpu_vm_sdma_copy_ptes - copy the PTEs from mapping
+ *
+ * @p: see amdgpu_vm_update_params definition
+ * @bo: PD/PT to update
+ * @pe: addr of the page entry
+ * @count: number of page entries to copy
+ *
+ * Traces the parameters and calls the DMA function to copy the PTEs.
+ */
+static void amdgpu_vm_sdma_copy_ptes(struct amdgpu_vm_update_params *p,
+				     struct amdgpu_bo *bo, uint64_t pe,
+				     unsigned count)
+{
+	struct amdgpu_ib *ib = p->job->ibs;
+	uint64_t src = ib->gpu_addr;
+
+	src += p->num_dw_left * 4;
+
+	pe += amdgpu_bo_gpu_offset_no_check(bo);
+	trace_amdgpu_vm_copy_ptes(pe, src, count, p->immediate);
+
+	amdgpu_vm_copy_pte(p->adev, ib, pe, src, count);
+}
+
+/**
+ * amdgpu_vm_sdma_set_ptes - helper to call the right asic function
+ *
+ * @p: see amdgpu_vm_update_params definition
+ * @bo: PD/PT to update
+ * @pe: byte offset of the PDE/PTE, relative to start of PDB/PTB
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: hw access flags
+ *
+ * Traces the parameters and calls the right asic functions
+ * to setup the page table using the DMA.
+ */
+static void amdgpu_vm_sdma_set_ptes(struct amdgpu_vm_update_params *p,
+				    struct amdgpu_bo *bo, uint64_t pe,
+				    uint64_t addr, unsigned count,
+				    uint32_t incr, uint64_t flags)
+{
+	struct amdgpu_ib *ib = p->job->ibs;
+
+	pe += amdgpu_bo_gpu_offset_no_check(bo);
+	trace_amdgpu_vm_set_ptes(pe, addr, count, incr, flags, p->immediate);
+	if (count < 3) {
+		amdgpu_vm_write_pte(p->adev, ib, pe, addr | flags,
+				    count, incr);
+	} else {
+		amdgpu_vm_set_pte_pde(p->adev, ib, pe, addr,
+				      count, incr, flags);
+	}
+}
+
+/**
+ * amdgpu_vm_sdma_update - execute VM update
+ *
+ * @p: see amdgpu_vm_update_params definition
+ * @vmbo: PD/PT to update
+ * @pe: byte offset of the PDE/PTE, relative to start of PDB/PTB
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: hw access flags
+ *
+ * Reserve space in the IB, setup mapping buffer on demand and write commands to
+ * the IB.
+ */
+static int amdgpu_vm_sdma_update(struct amdgpu_vm_update_params *p,
+				 struct amdgpu_bo_vm *vmbo, uint64_t pe,
+				 uint64_t addr, unsigned count, uint32_t incr,
+				 uint64_t flags)
+{
+	struct amdgpu_bo *bo = &vmbo->bo;
+	struct dma_resv_iter cursor;
+	unsigned int i, ndw, nptes;
+	struct dma_fence *fence;
+	uint64_t *pte;
+	int r;
+
+	/* Wait for PD/PT moves to be completed */
+	dma_resv_iter_begin(&cursor, bo->tbo.base.resv, DMA_RESV_USAGE_KERNEL);
+	dma_resv_for_each_fence_unlocked(&cursor, fence) {
+		dma_fence_get(fence);
+		r = drm_sched_job_add_dependency(&p->job->base, fence);
+		if (r) {
+			dma_fence_put(fence);
+			dma_resv_iter_end(&cursor);
+			return r;
+		}
+	}
+	dma_resv_iter_end(&cursor);
+
+	do {
+		ndw = p->num_dw_left;
+		ndw -= p->job->ibs->length_dw;
+
+		if (ndw < 32) {
+			r = amdgpu_vm_sdma_commit(p, NULL);
+			if (r)
+				return r;
+
+			r = amdgpu_vm_sdma_alloc_job(p, count,
+						     AMDGPU_KERNEL_JOB_ID_VM_UPDATE);
+			if (r)
+				return r;
+		}
+
+		if (!p->pages_addr) {
+			/* set page commands needed */
+			amdgpu_vm_sdma_set_ptes(p, bo, pe, addr, count,
+						incr, flags);
+			return 0;
+		}
+
+		/* copy commands needed */
+		ndw -= p->adev->vm_manager.vm_pte_funcs->copy_pte_num_dw;
+
+		/* for padding */
+		ndw -= 7;
+
+		nptes = min(count, ndw / 2);
+
+		/* Put the PTEs at the end of the IB. */
+		p->num_dw_left -= nptes * 2;
+		pte = (uint64_t *)&(p->job->ibs->ptr[p->num_dw_left]);
+		for (i = 0; i < nptes; ++i, addr += incr) {
+			pte[i] = amdgpu_vm_map_gart(p->pages_addr, addr);
+			pte[i] |= flags;
+		}
+
+		amdgpu_vm_sdma_copy_ptes(p, bo, pe, nptes);
+
+		pe += nptes * 8;
+		count -= nptes;
+	} while (count);
+
+	return 0;
+}
+
+const struct amdgpu_vm_update_funcs amdgpu_vm_sdma_funcs = {
+	.map_table = amdgpu_vm_sdma_map_table,
+	.prepare = amdgpu_vm_sdma_prepare,
+	.update = amdgpu_vm_sdma_update,
+	.commit = amdgpu_vm_sdma_commit
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_tlb_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_tlb_fence.c
new file mode 100644
index 000000000000..5d26797356a3
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_tlb_fence.c
@@ -0,0 +1,111 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include <linux/dma-fence.h>
+#include <linux/workqueue.h>
+
+#include "amdgpu.h"
+#include "amdgpu_vm.h"
+#include "amdgpu_gmc.h"
+
+struct amdgpu_tlb_fence {
+	struct dma_fence	base;
+	struct amdgpu_device	*adev;
+	struct dma_fence	*dependency;
+	struct work_struct	work;
+	spinlock_t		lock;
+	uint16_t		pasid;
+
+};
+
+static const char *amdgpu_tlb_fence_get_driver_name(struct dma_fence *fence)
+{
+	return "amdgpu tlb fence";
+}
+
+static const char *amdgpu_tlb_fence_get_timeline_name(struct dma_fence *f)
+{
+	return "amdgpu tlb timeline";
+}
+
+static void amdgpu_tlb_fence_work(struct work_struct *work)
+{
+	struct amdgpu_tlb_fence *f = container_of(work, typeof(*f), work);
+	int r;
+
+	if (f->dependency) {
+		dma_fence_wait(f->dependency, false);
+		dma_fence_put(f->dependency);
+		f->dependency = NULL;
+	}
+
+	r = amdgpu_gmc_flush_gpu_tlb_pasid(f->adev, f->pasid, 2, true, 0);
+	if (r) {
+		dev_err(f->adev->dev, "TLB flush failed for PASID %d.\n",
+			f->pasid);
+		dma_fence_set_error(&f->base, r);
+	}
+
+	dma_fence_signal(&f->base);
+	dma_fence_put(&f->base);
+}
+
+static const struct dma_fence_ops amdgpu_tlb_fence_ops = {
+	.get_driver_name = amdgpu_tlb_fence_get_driver_name,
+	.get_timeline_name = amdgpu_tlb_fence_get_timeline_name
+};
+
+void amdgpu_vm_tlb_fence_create(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+				struct dma_fence **fence)
+{
+	struct amdgpu_tlb_fence *f;
+
+	f = kmalloc(sizeof(*f), GFP_KERNEL);
+	if (!f) {
+		/*
+		 * We can't fail since the PDEs and PTEs are already updated, so
+		 * just block for the dependency and execute the TLB flush
+		 */
+		if (*fence)
+			dma_fence_wait(*fence, false);
+
+		amdgpu_gmc_flush_gpu_tlb_pasid(adev, vm->pasid, 2, true, 0);
+		*fence = dma_fence_get_stub();
+		return;
+	}
+
+	f->adev = adev;
+	f->dependency = *fence;
+	f->pasid = vm->pasid;
+	INIT_WORK(&f->work, amdgpu_tlb_fence_work);
+	spin_lock_init(&f->lock);
+
+	dma_fence_init64(&f->base, &amdgpu_tlb_fence_ops, &f->lock,
+			 vm->tlb_fence_context, atomic64_read(&vm->tlb_seq));
+
+	/* TODO: We probably need a separate wq here */
+	dma_fence_get(&f->base);
+	schedule_work(&f->work);
+
+	*fence = &f->base;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c
new file mode 100644
index 000000000000..474bfe36c0c2
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c
@@ -0,0 +1,992 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include <linux/firmware.h>
+#include <drm/drm_drv.h>
+
+#include "amdgpu.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_vpe.h"
+#include "amdgpu_smu.h"
+#include "soc15_common.h"
+#include "vpe_v6_1.h"
+
+#define AMDGPU_CSA_VPE_SIZE 	64
+/* VPE CSA resides in the 4th page of CSA */
+#define AMDGPU_CSA_VPE_OFFSET 	(4096 * 3)
+
+/* 1 second timeout */
+#define VPE_IDLE_TIMEOUT	msecs_to_jiffies(1000)
+
+#define VPE_MAX_DPM_LEVEL			4
+#define FIXED1_8_BITS_PER_FRACTIONAL_PART	8
+#define GET_PRATIO_INTEGER_PART(x)		((x) >> FIXED1_8_BITS_PER_FRACTIONAL_PART)
+
+static void vpe_set_ring_funcs(struct amdgpu_device *adev);
+
+static inline uint16_t div16_u16_rem(uint16_t dividend, uint16_t divisor, uint16_t *remainder)
+{
+	*remainder = dividend % divisor;
+	return dividend / divisor;
+}
+
+static inline uint16_t complete_integer_division_u16(
+	uint16_t dividend,
+	uint16_t divisor,
+	uint16_t *remainder)
+{
+	return div16_u16_rem(dividend, divisor, (uint16_t *)remainder);
+}
+
+static uint16_t vpe_u1_8_from_fraction(uint16_t numerator, uint16_t denominator)
+{
+	u16 arg1_value = numerator;
+	u16 arg2_value = denominator;
+
+	uint16_t remainder;
+
+	/* determine integer part */
+	uint16_t res_value = complete_integer_division_u16(
+		arg1_value, arg2_value, &remainder);
+
+	if (res_value > 127 /* CHAR_MAX */)
+		return 0;
+
+	/* determine fractional part */
+	{
+		unsigned int i = FIXED1_8_BITS_PER_FRACTIONAL_PART;
+
+		do {
+			remainder <<= 1;
+
+			res_value <<= 1;
+
+			if (remainder >= arg2_value) {
+				res_value |= 1;
+				remainder -= arg2_value;
+			}
+		} while (--i != 0);
+	}
+
+	/* round up LSB */
+	{
+		uint16_t summand = (remainder << 1) >= arg2_value;
+
+		if ((res_value + summand) > 32767 /* SHRT_MAX */)
+			return 0;
+
+		res_value += summand;
+	}
+
+	return res_value;
+}
+
+static uint16_t vpe_internal_get_pratio(uint16_t from_frequency, uint16_t to_frequency)
+{
+	uint16_t pratio = vpe_u1_8_from_fraction(from_frequency, to_frequency);
+
+	if (GET_PRATIO_INTEGER_PART(pratio) > 1)
+		pratio = 0;
+
+	return pratio;
+}
+
+/*
+ * VPE has 4 DPM levels from level 0 (lowerest) to 3 (highest),
+ * VPE FW will dynamically decide which level should be used according to current loading.
+ *
+ * Get VPE and SOC clocks from PM, and select the appropriate four clock values,
+ * calculate the ratios of adjusting from one clock to another.
+ * The VPE FW can then request the appropriate frequency from the PMFW.
+ */
+int amdgpu_vpe_configure_dpm(struct amdgpu_vpe *vpe)
+{
+	struct amdgpu_device *adev = vpe->ring.adev;
+	uint32_t dpm_ctl;
+
+	if (adev->pm.dpm_enabled) {
+		struct dpm_clocks clock_table = { 0 };
+		struct dpm_clock *VPEClks;
+		struct dpm_clock *SOCClks;
+		uint32_t idx;
+		uint32_t vpeclk_enalbled_num = 0;
+		uint32_t pratio_vmax_vnorm = 0, pratio_vnorm_vmid = 0, pratio_vmid_vmin = 0;
+		uint16_t pratio_vmin_freq = 0, pratio_vmid_freq = 0, pratio_vnorm_freq = 0, pratio_vmax_freq = 0;
+
+		dpm_ctl = RREG32(vpe_get_reg_offset(vpe, 0, vpe->regs.dpm_enable));
+		dpm_ctl |= 1; /* DPM enablement */
+		WREG32(vpe_get_reg_offset(vpe, 0, vpe->regs.dpm_enable), dpm_ctl);
+
+		/* Get VPECLK and SOCCLK */
+		if (amdgpu_dpm_get_dpm_clock_table(adev, &clock_table)) {
+			dev_dbg(adev->dev, "%s: get clock failed!\n", __func__);
+			goto disable_dpm;
+		}
+
+		SOCClks = clock_table.SocClocks;
+		VPEClks = clock_table.VPEClocks;
+
+		/* Comfirm enabled vpe clk num
+		 * Enabled VPE clocks are ordered from low to high in VPEClks
+		 * The highest valid clock index+1 is the number of VPEClks
+		 */
+		for (idx = PP_SMU_NUM_VPECLK_DPM_LEVELS; idx && !vpeclk_enalbled_num; idx--)
+			if (VPEClks[idx-1].Freq)
+				vpeclk_enalbled_num = idx;
+
+		/* vpe dpm only cares 4 levels. */
+		for (idx = 0; idx < VPE_MAX_DPM_LEVEL; idx++) {
+			uint32_t soc_dpm_level;
+			uint32_t min_freq;
+
+			if (idx == 0)
+				soc_dpm_level = 0;
+			else
+				soc_dpm_level = (idx * 2) + 1;
+
+			/* clamp the max level */
+			if (soc_dpm_level > vpeclk_enalbled_num - 1)
+				soc_dpm_level = vpeclk_enalbled_num - 1;
+
+			min_freq = (SOCClks[soc_dpm_level].Freq < VPEClks[soc_dpm_level].Freq) ?
+				   SOCClks[soc_dpm_level].Freq : VPEClks[soc_dpm_level].Freq;
+
+			switch (idx) {
+			case 0:
+				pratio_vmin_freq = min_freq;
+				break;
+			case 1:
+				pratio_vmid_freq = min_freq;
+				break;
+			case 2:
+				pratio_vnorm_freq = min_freq;
+				break;
+			case 3:
+				pratio_vmax_freq = min_freq;
+				break;
+			default:
+				break;
+			}
+		}
+
+		if (pratio_vmin_freq && pratio_vmid_freq && pratio_vnorm_freq && pratio_vmax_freq) {
+			uint32_t pratio_ctl;
+
+			pratio_vmax_vnorm = (uint32_t)vpe_internal_get_pratio(pratio_vmax_freq, pratio_vnorm_freq);
+			pratio_vnorm_vmid = (uint32_t)vpe_internal_get_pratio(pratio_vnorm_freq, pratio_vmid_freq);
+			pratio_vmid_vmin = (uint32_t)vpe_internal_get_pratio(pratio_vmid_freq, pratio_vmin_freq);
+
+			pratio_ctl = pratio_vmax_vnorm | (pratio_vnorm_vmid << 9) | (pratio_vmid_vmin << 18);
+			WREG32(vpe_get_reg_offset(vpe, 0, vpe->regs.dpm_pratio), pratio_ctl);		/* PRatio */
+			WREG32(vpe_get_reg_offset(vpe, 0, vpe->regs.dpm_request_interval), 24000);	/* 1ms, unit=1/24MHz */
+			WREG32(vpe_get_reg_offset(vpe, 0, vpe->regs.dpm_decision_threshold), 1200000);	/* 50ms */
+			WREG32(vpe_get_reg_offset(vpe, 0, vpe->regs.dpm_busy_clamp_threshold), 1200000);/* 50ms */
+			WREG32(vpe_get_reg_offset(vpe, 0, vpe->regs.dpm_idle_clamp_threshold), 1200000);/* 50ms */
+			dev_dbg(adev->dev, "%s: configure vpe dpm pratio done!\n", __func__);
+		} else {
+			dev_dbg(adev->dev, "%s: invalid pratio parameters!\n", __func__);
+			goto disable_dpm;
+		}
+	}
+	return 0;
+
+disable_dpm:
+	dpm_ctl = RREG32(vpe_get_reg_offset(vpe, 0, vpe->regs.dpm_enable));
+	dpm_ctl &= 0xfffffffe; /* Disable DPM */
+	WREG32(vpe_get_reg_offset(vpe, 0, vpe->regs.dpm_enable), dpm_ctl);
+	dev_dbg(adev->dev, "%s: disable vpe dpm\n", __func__);
+	return -EINVAL;
+}
+
+int amdgpu_vpe_psp_update_sram(struct amdgpu_device *adev)
+{
+	struct amdgpu_firmware_info ucode = {
+		.ucode_id = AMDGPU_UCODE_ID_VPE,
+		.mc_addr = adev->vpe.cmdbuf_gpu_addr,
+		.ucode_size = 8,
+	};
+
+	return psp_execute_ip_fw_load(&adev->psp, &ucode);
+}
+
+int amdgpu_vpe_init_microcode(struct amdgpu_vpe *vpe)
+{
+	struct amdgpu_device *adev = vpe->ring.adev;
+	const struct vpe_firmware_header_v1_0 *vpe_hdr;
+	char fw_prefix[32];
+	int ret;
+
+	amdgpu_ucode_ip_version_decode(adev, VPE_HWIP, fw_prefix, sizeof(fw_prefix));
+	ret = amdgpu_ucode_request(adev, &adev->vpe.fw, AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s.bin", fw_prefix);
+	if (ret)
+		goto out;
+
+	vpe_hdr = (const struct vpe_firmware_header_v1_0 *)adev->vpe.fw->data;
+	adev->vpe.fw_version = le32_to_cpu(vpe_hdr->header.ucode_version);
+	adev->vpe.feature_version = le32_to_cpu(vpe_hdr->ucode_feature_version);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		struct amdgpu_firmware_info *info;
+
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_VPE_CTX];
+		info->ucode_id = AMDGPU_UCODE_ID_VPE_CTX;
+		info->fw = adev->vpe.fw;
+		adev->firmware.fw_size +=
+			ALIGN(le32_to_cpu(vpe_hdr->ctx_ucode_size_bytes), PAGE_SIZE);
+
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_VPE_CTL];
+		info->ucode_id = AMDGPU_UCODE_ID_VPE_CTL;
+		info->fw = adev->vpe.fw;
+		adev->firmware.fw_size +=
+			ALIGN(le32_to_cpu(vpe_hdr->ctl_ucode_size_bytes), PAGE_SIZE);
+	}
+
+	return 0;
+out:
+	dev_err(adev->dev, "fail to initialize vpe microcode\n");
+	release_firmware(adev->vpe.fw);
+	adev->vpe.fw = NULL;
+	return ret;
+}
+
+int amdgpu_vpe_ring_init(struct amdgpu_vpe *vpe)
+{
+	struct amdgpu_device *adev = container_of(vpe, struct amdgpu_device, vpe);
+	struct amdgpu_ring *ring = &vpe->ring;
+	int ret;
+
+	ring->ring_obj = NULL;
+	ring->use_doorbell = true;
+	ring->vm_hub = AMDGPU_MMHUB0(0);
+	ring->doorbell_index = (adev->doorbell_index.vpe_ring << 1);
+	snprintf(ring->name, 4, "vpe");
+
+	ret = amdgpu_ring_init(adev, ring, 1024, &vpe->trap_irq, 0,
+			     AMDGPU_RING_PRIO_DEFAULT, NULL);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+int amdgpu_vpe_ring_fini(struct amdgpu_vpe *vpe)
+{
+	amdgpu_ring_fini(&vpe->ring);
+
+	return 0;
+}
+
+static int vpe_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_vpe *vpe = &adev->vpe;
+
+	switch (amdgpu_ip_version(adev, VPE_HWIP, 0)) {
+	case IP_VERSION(6, 1, 0):
+	case IP_VERSION(6, 1, 3):
+		vpe_v6_1_set_funcs(vpe);
+		break;
+	case IP_VERSION(6, 1, 1):
+		vpe_v6_1_set_funcs(vpe);
+		vpe->collaborate_mode = true;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	vpe_set_ring_funcs(adev);
+	vpe_set_regs(vpe);
+
+	dev_info(adev->dev, "VPE: collaborate mode %s", vpe->collaborate_mode ? "true" : "false");
+
+	return 0;
+}
+
+static void vpe_idle_work_handler(struct work_struct *work)
+{
+	struct amdgpu_device *adev =
+		container_of(work, struct amdgpu_device, vpe.idle_work.work);
+	unsigned int fences = 0;
+
+	fences += amdgpu_fence_count_emitted(&adev->vpe.ring);
+
+	if (fences == 0)
+		amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VPE, AMD_PG_STATE_GATE);
+	else
+		schedule_delayed_work(&adev->vpe.idle_work, VPE_IDLE_TIMEOUT);
+}
+
+static int vpe_common_init(struct amdgpu_vpe *vpe)
+{
+	struct amdgpu_device *adev = container_of(vpe, struct amdgpu_device, vpe);
+	int r;
+
+	r = amdgpu_bo_create_kernel(adev, PAGE_SIZE, PAGE_SIZE,
+				    AMDGPU_GEM_DOMAIN_GTT,
+				    &adev->vpe.cmdbuf_obj,
+				    &adev->vpe.cmdbuf_gpu_addr,
+				    (void **)&adev->vpe.cmdbuf_cpu_addr);
+	if (r) {
+		dev_err(adev->dev, "VPE: failed to allocate cmdbuf bo %d\n", r);
+		return r;
+	}
+
+	vpe->context_started = false;
+	INIT_DELAYED_WORK(&adev->vpe.idle_work, vpe_idle_work_handler);
+
+	return 0;
+}
+
+static int vpe_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_vpe *vpe = &adev->vpe;
+	int ret;
+
+	ret = vpe_common_init(vpe);
+	if (ret)
+		goto out;
+
+	ret = vpe_irq_init(vpe);
+	if (ret)
+		goto out;
+
+	ret = vpe_ring_init(vpe);
+	if (ret)
+		goto out;
+
+	ret = vpe_init_microcode(vpe);
+	if (ret)
+		goto out;
+
+	adev->vpe.supported_reset =
+		 amdgpu_get_soft_full_reset_mask(&adev->vpe.ring);
+	if (!amdgpu_sriov_vf(adev))
+		adev->vpe.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+	ret = amdgpu_vpe_sysfs_reset_mask_init(adev);
+	if (ret)
+		goto out;
+out:
+	return ret;
+}
+
+static int vpe_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_vpe *vpe = &adev->vpe;
+
+	release_firmware(vpe->fw);
+	vpe->fw = NULL;
+
+	amdgpu_vpe_sysfs_reset_mask_fini(adev);
+	vpe_ring_fini(vpe);
+
+	amdgpu_bo_free_kernel(&adev->vpe.cmdbuf_obj,
+			      &adev->vpe.cmdbuf_gpu_addr,
+			      (void **)&adev->vpe.cmdbuf_cpu_addr);
+
+	return 0;
+}
+
+static int vpe_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_vpe *vpe = &adev->vpe;
+	int ret;
+
+	/* Power on VPE */
+	ret = amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VPE,
+						     AMD_PG_STATE_UNGATE);
+	if (ret)
+		return ret;
+
+	ret = vpe_load_microcode(vpe);
+	if (ret)
+		return ret;
+
+	ret = vpe_ring_start(vpe);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static int vpe_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_vpe *vpe = &adev->vpe;
+
+	cancel_delayed_work_sync(&adev->vpe.idle_work);
+
+	vpe_ring_stop(vpe);
+
+	/* Power off VPE */
+	amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VPE, AMD_PG_STATE_GATE);
+
+	return 0;
+}
+
+static int vpe_suspend(struct amdgpu_ip_block *ip_block)
+{
+	return vpe_hw_fini(ip_block);
+}
+
+static int vpe_resume(struct amdgpu_ip_block *ip_block)
+{
+	return vpe_hw_init(ip_block);
+}
+
+static void vpe_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	int i;
+
+	for (i = 0; i < count; i++)
+		if (i == 0)
+			amdgpu_ring_write(ring, ring->funcs->nop |
+				VPE_CMD_NOP_HEADER_COUNT(count - 1));
+		else
+			amdgpu_ring_write(ring, ring->funcs->nop);
+}
+
+static uint64_t vpe_get_csa_mc_addr(struct amdgpu_ring *ring, uint32_t vmid)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t index = 0;
+	uint64_t csa_mc_addr;
+
+	if (amdgpu_sriov_vf(adev) || vmid == 0 || !adev->gfx.mcbp)
+		return 0;
+
+	csa_mc_addr = amdgpu_csa_vaddr(adev) + AMDGPU_CSA_VPE_OFFSET +
+		      index * AMDGPU_CSA_VPE_SIZE;
+
+	return csa_mc_addr;
+}
+
+static void vpe_ring_emit_pred_exec(struct amdgpu_ring *ring,
+				    uint32_t device_select,
+				    uint32_t exec_count)
+{
+	if (!ring->adev->vpe.collaborate_mode)
+		return;
+
+	amdgpu_ring_write(ring, VPE_CMD_HEADER(VPE_CMD_OPCODE_PRED_EXE, 0) |
+				(device_select << 16));
+	amdgpu_ring_write(ring, exec_count & 0x1fff);
+}
+
+static void vpe_ring_emit_ib(struct amdgpu_ring *ring,
+			     struct amdgpu_job *job,
+			     struct amdgpu_ib *ib,
+			     uint32_t flags)
+{
+	uint32_t vmid = AMDGPU_JOB_GET_VMID(job);
+	uint64_t csa_mc_addr = vpe_get_csa_mc_addr(ring, vmid);
+
+	amdgpu_ring_write(ring, VPE_CMD_HEADER(VPE_CMD_OPCODE_INDIRECT, 0) |
+				VPE_CMD_INDIRECT_HEADER_VMID(vmid & 0xf));
+
+	/* base must be 32 byte aligned */
+	amdgpu_ring_write(ring, ib->gpu_addr & 0xffffffe0);
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, ib->length_dw);
+	amdgpu_ring_write(ring, lower_32_bits(csa_mc_addr));
+	amdgpu_ring_write(ring, upper_32_bits(csa_mc_addr));
+}
+
+static void vpe_ring_emit_fence(struct amdgpu_ring *ring, uint64_t addr,
+				uint64_t seq, unsigned int flags)
+{
+	int i = 0;
+
+	do {
+		/* write the fence */
+		amdgpu_ring_write(ring, VPE_CMD_HEADER(VPE_CMD_OPCODE_FENCE, 0));
+		/* zero in first two bits */
+		WARN_ON_ONCE(addr & 0x3);
+		amdgpu_ring_write(ring, lower_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(addr));
+		amdgpu_ring_write(ring, i == 0 ? lower_32_bits(seq) : upper_32_bits(seq));
+		addr += 4;
+	} while ((flags & AMDGPU_FENCE_FLAG_64BIT) && (i++ < 1));
+
+	if (flags & AMDGPU_FENCE_FLAG_INT) {
+		/* generate an interrupt */
+		amdgpu_ring_write(ring, VPE_CMD_HEADER(VPE_CMD_OPCODE_TRAP, 0));
+		amdgpu_ring_write(ring, 0);
+	}
+
+}
+
+static void vpe_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	vpe_ring_emit_pred_exec(ring, 0, 6);
+
+	/* wait for idle */
+	amdgpu_ring_write(ring, VPE_CMD_HEADER(VPE_CMD_OPCODE_POLL_REGMEM,
+				VPE_POLL_REGMEM_SUBOP_REGMEM) |
+				VPE_CMD_POLL_REGMEM_HEADER_FUNC(3) | /* equal */
+				VPE_CMD_POLL_REGMEM_HEADER_MEM(1));
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, seq); /* reference */
+	amdgpu_ring_write(ring, 0xffffffff); /* mask */
+	amdgpu_ring_write(ring, VPE_CMD_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+				VPE_CMD_POLL_REGMEM_DW5_INTERVAL(4));
+}
+
+static void vpe_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg, uint32_t val)
+{
+	vpe_ring_emit_pred_exec(ring, 0, 3);
+
+	amdgpu_ring_write(ring, VPE_CMD_HEADER(VPE_CMD_OPCODE_REG_WRITE, 0));
+	amdgpu_ring_write(ring,	reg << 2);
+	amdgpu_ring_write(ring, val);
+}
+
+static void vpe_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+				   uint32_t val, uint32_t mask)
+{
+	vpe_ring_emit_pred_exec(ring, 0, 6);
+
+	amdgpu_ring_write(ring, VPE_CMD_HEADER(VPE_CMD_OPCODE_POLL_REGMEM,
+				VPE_POLL_REGMEM_SUBOP_REGMEM) |
+				VPE_CMD_POLL_REGMEM_HEADER_FUNC(3) | /* equal */
+				VPE_CMD_POLL_REGMEM_HEADER_MEM(0));
+	amdgpu_ring_write(ring, reg << 2);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, val); /* reference */
+	amdgpu_ring_write(ring, mask); /* mask */
+	amdgpu_ring_write(ring, VPE_CMD_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+				VPE_CMD_POLL_REGMEM_DW5_INTERVAL(10));
+}
+
+static void vpe_ring_emit_vm_flush(struct amdgpu_ring *ring, unsigned int vmid,
+				   uint64_t pd_addr)
+{
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+}
+
+static unsigned int vpe_ring_init_cond_exec(struct amdgpu_ring *ring,
+					    uint64_t addr)
+{
+	unsigned int ret;
+
+	amdgpu_ring_write(ring, VPE_CMD_HEADER(VPE_CMD_OPCODE_COND_EXE, 0));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, 1);
+	ret = ring->wptr & ring->buf_mask;
+	amdgpu_ring_write(ring, 0);
+
+	return ret;
+}
+
+static int vpe_ring_preempt_ib(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_vpe *vpe = &adev->vpe;
+	uint32_t preempt_reg = vpe->regs.queue0_preempt;
+	int i, r = 0;
+
+	/* assert preemption condition */
+	amdgpu_ring_set_preempt_cond_exec(ring, false);
+
+	/* emit the trailing fence */
+	ring->trail_seq += 1;
+	amdgpu_ring_alloc(ring, 10);
+	vpe_ring_emit_fence(ring, ring->trail_fence_gpu_addr, ring->trail_seq, 0);
+	amdgpu_ring_commit(ring);
+
+	/* assert IB preemption */
+	WREG32(vpe_get_reg_offset(vpe, ring->me, preempt_reg), 1);
+
+	/* poll the trailing fence */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (ring->trail_seq ==
+		    le32_to_cpu(*(ring->trail_fence_cpu_addr)))
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout) {
+		r = -EINVAL;
+		dev_err(adev->dev, "ring %d failed to be preempted\n", ring->idx);
+	}
+
+	/* deassert IB preemption */
+	WREG32(vpe_get_reg_offset(vpe, ring->me, preempt_reg), 0);
+
+	/* deassert the preemption condition */
+	amdgpu_ring_set_preempt_cond_exec(ring, true);
+
+	return r;
+}
+
+static int vpe_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+				     enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int vpe_set_powergating_state(struct amdgpu_ip_block *ip_block,
+				     enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_vpe *vpe = &adev->vpe;
+
+	if (!adev->pm.dpm_enabled)
+		dev_err(adev->dev, "Without PM, cannot support powergating\n");
+
+	dev_dbg(adev->dev, "%s: %s!\n", __func__, (state == AMD_PG_STATE_GATE) ? "GATE":"UNGATE");
+
+	if (state == AMD_PG_STATE_GATE) {
+		amdgpu_dpm_enable_vpe(adev, false);
+		vpe->context_started = false;
+	} else {
+		amdgpu_dpm_enable_vpe(adev, true);
+	}
+
+	return 0;
+}
+
+static uint64_t vpe_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_vpe *vpe = &adev->vpe;
+	uint64_t rptr;
+
+	if (ring->use_doorbell) {
+		rptr = atomic64_read((atomic64_t *)ring->rptr_cpu_addr);
+		dev_dbg(adev->dev, "rptr/doorbell before shift == 0x%016llx\n", rptr);
+	} else {
+		rptr = RREG32(vpe_get_reg_offset(vpe, ring->me, vpe->regs.queue0_rb_rptr_hi));
+		rptr = rptr << 32;
+		rptr |= RREG32(vpe_get_reg_offset(vpe, ring->me, vpe->regs.queue0_rb_rptr_lo));
+		dev_dbg(adev->dev, "rptr before shift [%i] == 0x%016llx\n", ring->me, rptr);
+	}
+
+	return (rptr >> 2);
+}
+
+static uint64_t vpe_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_vpe *vpe = &adev->vpe;
+	uint64_t wptr;
+
+	if (ring->use_doorbell) {
+		wptr = atomic64_read((atomic64_t *)ring->wptr_cpu_addr);
+		dev_dbg(adev->dev, "wptr/doorbell before shift == 0x%016llx\n", wptr);
+	} else {
+		wptr = RREG32(vpe_get_reg_offset(vpe, ring->me, vpe->regs.queue0_rb_wptr_hi));
+		wptr = wptr << 32;
+		wptr |= RREG32(vpe_get_reg_offset(vpe, ring->me, vpe->regs.queue0_rb_wptr_lo));
+		dev_dbg(adev->dev, "wptr before shift [%i] == 0x%016llx\n", ring->me, wptr);
+	}
+
+	return (wptr >> 2);
+}
+
+static void vpe_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_vpe *vpe = &adev->vpe;
+
+	if (ring->use_doorbell) {
+		dev_dbg(adev->dev, "Using doorbell, \
+			wptr_offs == 0x%08x, \
+			lower_32_bits(ring->wptr) << 2 == 0x%08x, \
+			upper_32_bits(ring->wptr) << 2 == 0x%08x\n",
+			ring->wptr_offs,
+			lower_32_bits(ring->wptr << 2),
+			upper_32_bits(ring->wptr << 2));
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr, ring->wptr << 2);
+		WDOORBELL64(ring->doorbell_index, ring->wptr << 2);
+		if (vpe->collaborate_mode)
+			WDOORBELL64(ring->doorbell_index + 4, ring->wptr << 2);
+	} else {
+		int i;
+
+		for (i = 0; i < vpe->num_instances; i++) {
+			dev_dbg(adev->dev, "Not using doorbell, \
+				regVPEC_QUEUE0_RB_WPTR == 0x%08x, \
+				regVPEC_QUEUE0_RB_WPTR_HI == 0x%08x\n",
+				lower_32_bits(ring->wptr << 2),
+				upper_32_bits(ring->wptr << 2));
+			WREG32(vpe_get_reg_offset(vpe, i, vpe->regs.queue0_rb_wptr_lo),
+			       lower_32_bits(ring->wptr << 2));
+			WREG32(vpe_get_reg_offset(vpe, i, vpe->regs.queue0_rb_wptr_hi),
+			       upper_32_bits(ring->wptr << 2));
+		}
+	}
+}
+
+static int vpe_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	const uint32_t test_pattern = 0xdeadbeef;
+	uint32_t index, i;
+	uint64_t wb_addr;
+	int ret;
+
+	ret = amdgpu_device_wb_get(adev, &index);
+	if (ret) {
+		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", ret);
+		return ret;
+	}
+
+	adev->wb.wb[index] = 0;
+	wb_addr = adev->wb.gpu_addr + (index * 4);
+
+	ret = amdgpu_ring_alloc(ring, 4);
+	if (ret) {
+		dev_err(adev->dev, "amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, ret);
+		goto out;
+	}
+
+	amdgpu_ring_write(ring, VPE_CMD_HEADER(VPE_CMD_OPCODE_FENCE, 0));
+	amdgpu_ring_write(ring, lower_32_bits(wb_addr));
+	amdgpu_ring_write(ring, upper_32_bits(wb_addr));
+	amdgpu_ring_write(ring, test_pattern);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (le32_to_cpu(adev->wb.wb[index]) == test_pattern)
+			goto out;
+		udelay(1);
+	}
+
+	ret = -ETIMEDOUT;
+out:
+	amdgpu_device_wb_free(adev, index);
+
+	return ret;
+}
+
+static int vpe_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct amdgpu_device *adev = ring->adev;
+	const uint32_t test_pattern = 0xdeadbeef;
+	struct amdgpu_ib ib = {};
+	struct dma_fence *f = NULL;
+	uint32_t index;
+	uint64_t wb_addr;
+	int ret;
+
+	ret = amdgpu_device_wb_get(adev, &index);
+	if (ret) {
+		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", ret);
+		return ret;
+	}
+
+	adev->wb.wb[index] = 0;
+	wb_addr = adev->wb.gpu_addr + (index * 4);
+
+	ret = amdgpu_ib_get(adev, NULL, 256, AMDGPU_IB_POOL_DIRECT, &ib);
+	if (ret)
+		goto err0;
+
+	ib.ptr[0] = VPE_CMD_HEADER(VPE_CMD_OPCODE_FENCE, 0);
+	ib.ptr[1] = lower_32_bits(wb_addr);
+	ib.ptr[2] = upper_32_bits(wb_addr);
+	ib.ptr[3] = test_pattern;
+	ib.ptr[4] = VPE_CMD_HEADER(VPE_CMD_OPCODE_NOP, 0);
+	ib.ptr[5] = VPE_CMD_HEADER(VPE_CMD_OPCODE_NOP, 0);
+	ib.ptr[6] = VPE_CMD_HEADER(VPE_CMD_OPCODE_NOP, 0);
+	ib.ptr[7] = VPE_CMD_HEADER(VPE_CMD_OPCODE_NOP, 0);
+	ib.length_dw = 8;
+
+	ret = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
+	if (ret)
+		goto err1;
+
+	ret = dma_fence_wait_timeout(f, false, timeout);
+	if (ret <= 0) {
+		ret = ret ? : -ETIMEDOUT;
+		goto err1;
+	}
+
+	ret = (le32_to_cpu(adev->wb.wb[index]) == test_pattern) ? 0 : -EINVAL;
+
+err1:
+	amdgpu_ib_free(&ib, NULL);
+	dma_fence_put(f);
+err0:
+	amdgpu_device_wb_free(adev, index);
+
+	return ret;
+}
+
+static void vpe_ring_begin_use(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_vpe *vpe = &adev->vpe;
+
+	cancel_delayed_work_sync(&adev->vpe.idle_work);
+
+	/* Power on VPE and notify VPE of new context  */
+	if (!vpe->context_started) {
+		uint32_t context_notify;
+
+		/* Power on VPE */
+		amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VPE, AMD_PG_STATE_UNGATE);
+
+		/* Indicates that a job from a new context has been submitted. */
+		context_notify = RREG32(vpe_get_reg_offset(vpe, 0, vpe->regs.context_indicator));
+		if ((context_notify & 0x1) == 0)
+			context_notify |= 0x1;
+		else
+			context_notify &= ~(0x1);
+		WREG32(vpe_get_reg_offset(vpe, 0, vpe->regs.context_indicator), context_notify);
+		vpe->context_started = true;
+	}
+}
+
+static void vpe_ring_end_use(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	schedule_delayed_work(&adev->vpe.idle_work, VPE_IDLE_TIMEOUT);
+}
+
+static int vpe_ring_reset(struct amdgpu_ring *ring,
+			  unsigned int vmid,
+			  struct amdgpu_fence *timedout_fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int r;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+
+	r = amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VPE,
+						   AMD_PG_STATE_GATE);
+	if (r)
+		return r;
+	r = amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VPE,
+						   AMD_PG_STATE_UNGATE);
+	if (r)
+		return r;
+
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static ssize_t amdgpu_get_vpe_reset_mask(struct device *dev,
+						struct device_attribute *attr,
+						char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	if (!adev)
+		return -ENODEV;
+
+	return amdgpu_show_reset_mask(buf, adev->vpe.supported_reset);
+}
+
+static DEVICE_ATTR(vpe_reset_mask, 0444,
+		   amdgpu_get_vpe_reset_mask, NULL);
+
+int amdgpu_vpe_sysfs_reset_mask_init(struct amdgpu_device *adev)
+{
+	int r = 0;
+
+	if (adev->vpe.num_instances) {
+		r = device_create_file(adev->dev, &dev_attr_vpe_reset_mask);
+		if (r)
+			return r;
+	}
+
+	return r;
+}
+
+void amdgpu_vpe_sysfs_reset_mask_fini(struct amdgpu_device *adev)
+{
+	if (adev->dev->kobj.sd) {
+		if (adev->vpe.num_instances)
+			device_remove_file(adev->dev, &dev_attr_vpe_reset_mask);
+	}
+}
+
+static const struct amdgpu_ring_funcs vpe_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_VPE,
+	.align_mask = 0xf,
+	.nop = VPE_CMD_HEADER(VPE_CMD_OPCODE_NOP, 0),
+	.support_64bit_ptrs = true,
+	.get_rptr = vpe_ring_get_rptr,
+	.get_wptr = vpe_ring_get_wptr,
+	.set_wptr = vpe_ring_set_wptr,
+	.emit_frame_size =
+		5 + /* vpe_ring_init_cond_exec */
+		6 + /* vpe_ring_emit_pipeline_sync */
+		10 + 10 + 10 + /* vpe_ring_emit_fence */
+		/* vpe_ring_emit_vm_flush */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 6,
+	.emit_ib_size = 7 + 6,
+	.emit_ib = vpe_ring_emit_ib,
+	.emit_pipeline_sync = vpe_ring_emit_pipeline_sync,
+	.emit_fence = vpe_ring_emit_fence,
+	.emit_vm_flush = vpe_ring_emit_vm_flush,
+	.emit_wreg = vpe_ring_emit_wreg,
+	.emit_reg_wait = vpe_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.insert_nop = vpe_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.test_ring = vpe_ring_test_ring,
+	.test_ib = vpe_ring_test_ib,
+	.init_cond_exec = vpe_ring_init_cond_exec,
+	.preempt_ib = vpe_ring_preempt_ib,
+	.begin_use = vpe_ring_begin_use,
+	.end_use = vpe_ring_end_use,
+	.reset = vpe_ring_reset,
+};
+
+static void vpe_set_ring_funcs(struct amdgpu_device *adev)
+{
+	adev->vpe.ring.funcs = &vpe_ring_funcs;
+}
+
+const struct amd_ip_funcs vpe_ip_funcs = {
+	.name = "vpe_v6_1",
+	.early_init = vpe_early_init,
+	.sw_init = vpe_sw_init,
+	.sw_fini = vpe_sw_fini,
+	.hw_init = vpe_hw_init,
+	.hw_fini = vpe_hw_fini,
+	.suspend = vpe_suspend,
+	.resume = vpe_resume,
+	.set_clockgating_state = vpe_set_clockgating_state,
+	.set_powergating_state = vpe_set_powergating_state,
+};
+
+const struct amdgpu_ip_block_version vpe_v6_1_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_VPE,
+	.major = 6,
+	.minor = 1,
+	.rev = 0,
+	.funcs = &vpe_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.h
new file mode 100644
index 000000000000..695da740a97e
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.h
@@ -0,0 +1,111 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#ifndef __AMDGPU_VPE_H__
+#define __AMDGPU_VPE_H__
+
+#include "amdgpu_ring.h"
+#include "amdgpu_irq.h"
+#include "vpe_6_1_fw_if.h"
+
+#define AMDGPU_MAX_VPE_INSTANCES 2
+
+struct amdgpu_vpe;
+
+struct vpe_funcs {
+	uint32_t (*get_reg_offset)(struct amdgpu_vpe *vpe, uint32_t inst, uint32_t offset);
+	int (*set_regs)(struct amdgpu_vpe *vpe);
+	int (*irq_init)(struct amdgpu_vpe *vpe);
+	int (*init_microcode)(struct amdgpu_vpe *vpe);
+	int (*load_microcode)(struct amdgpu_vpe *vpe);
+	int (*ring_init)(struct amdgpu_vpe *vpe);
+	int (*ring_start)(struct amdgpu_vpe *vpe);
+	int (*ring_stop)(struct amdgpu_vpe *vpe);
+	int (*ring_fini)(struct amdgpu_vpe *vpe);
+};
+
+struct vpe_regs {
+	uint32_t queue0_rb_rptr_lo;
+	uint32_t queue0_rb_rptr_hi;
+	uint32_t queue0_rb_wptr_lo;
+	uint32_t queue0_rb_wptr_hi;
+	uint32_t queue0_preempt;
+
+	uint32_t dpm_enable;
+	uint32_t dpm_pratio;
+	uint32_t dpm_request_interval;
+	uint32_t dpm_decision_threshold;
+	uint32_t dpm_busy_clamp_threshold;
+	uint32_t dpm_idle_clamp_threshold;
+	uint32_t dpm_request_lv;
+	uint32_t context_indicator;
+};
+
+struct amdgpu_vpe {
+	struct amdgpu_ring		ring;
+	struct amdgpu_irq_src		trap_irq;
+
+	const struct vpe_funcs		*funcs;
+	struct vpe_regs			regs;
+
+	const struct firmware		*fw;
+	uint32_t			fw_version;
+	uint32_t			feature_version;
+
+	struct amdgpu_bo		*cmdbuf_obj;
+	uint64_t			cmdbuf_gpu_addr;
+	uint32_t			*cmdbuf_cpu_addr;
+	struct delayed_work		idle_work;
+	bool				context_started;
+
+	uint32_t			num_instances;
+	bool				collaborate_mode;
+	uint32_t			supported_reset;
+};
+
+int amdgpu_vpe_psp_update_sram(struct amdgpu_device *adev);
+int amdgpu_vpe_init_microcode(struct amdgpu_vpe *vpe);
+int amdgpu_vpe_ring_init(struct amdgpu_vpe *vpe);
+int amdgpu_vpe_ring_fini(struct amdgpu_vpe *vpe);
+int amdgpu_vpe_configure_dpm(struct amdgpu_vpe *vpe);
+void amdgpu_vpe_sysfs_reset_mask_fini(struct amdgpu_device *adev);
+int amdgpu_vpe_sysfs_reset_mask_init(struct amdgpu_device *adev);
+
+#define vpe_ring_init(vpe) ((vpe)->funcs->ring_init ? (vpe)->funcs->ring_init((vpe)) : 0)
+#define vpe_ring_start(vpe) ((vpe)->funcs->ring_start ? (vpe)->funcs->ring_start((vpe)) : 0)
+#define vpe_ring_stop(vpe) ((vpe)->funcs->ring_stop ? (vpe)->funcs->ring_stop((vpe)) : 0)
+#define vpe_ring_fini(vpe) ((vpe)->funcs->ring_fini ? (vpe)->funcs->ring_fini((vpe)) : 0)
+
+#define vpe_get_reg_offset(vpe, inst, offset) \
+		((vpe)->funcs->get_reg_offset ? (vpe)->funcs->get_reg_offset((vpe), (inst), (offset)) : 0)
+#define vpe_set_regs(vpe) \
+		((vpe)->funcs->set_regs ? (vpe)->funcs->set_regs((vpe)) : 0)
+#define vpe_irq_init(vpe) \
+		((vpe)->funcs->irq_init ? (vpe)->funcs->irq_init((vpe)) : 0)
+#define vpe_init_microcode(vpe) \
+		((vpe)->funcs->init_microcode ? (vpe)->funcs->init_microcode((vpe)) : 0)
+#define vpe_load_microcode(vpe) \
+		((vpe)->funcs->load_microcode ? (vpe)->funcs->load_microcode((vpe)) : 0)
+
+extern const struct amdgpu_ip_block_version vpe_v6_1_ip_block;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
index 26e900627971..a5adb2ed9b3c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
@@ -22,83 +22,407 @@
  * Authors: Christian König
  */
 
-#include <drm/drmP.h>
+#include <linux/dma-mapping.h>
+#include <drm/ttm/ttm_range_manager.h>
+#include <drm/drm_drv.h>
+
 #include "amdgpu.h"
+#include "amdgpu_vm.h"
+#include "amdgpu_res_cursor.h"
+#include "atom.h"
+
+#define AMDGPU_MAX_SG_SEGMENT_SIZE	(2UL << 30)
 
-struct amdgpu_vram_mgr {
-	struct drm_mm mm;
-	spinlock_t lock;
-	atomic64_t usage;
-	atomic64_t vis_usage;
+struct amdgpu_vram_reservation {
+	u64 start;
+	u64 size;
+	struct list_head allocated;
+	struct list_head blocks;
 };
 
+static inline struct amdgpu_vram_mgr *
+to_vram_mgr(struct ttm_resource_manager *man)
+{
+	return container_of(man, struct amdgpu_vram_mgr, manager);
+}
+
+static inline struct amdgpu_device *
+to_amdgpu_device(struct amdgpu_vram_mgr *mgr)
+{
+	return container_of(mgr, struct amdgpu_device, mman.vram_mgr);
+}
+
+static inline struct drm_buddy_block *
+amdgpu_vram_mgr_first_block(struct list_head *list)
+{
+	return list_first_entry_or_null(list, struct drm_buddy_block, link);
+}
+
+static inline bool amdgpu_is_vram_mgr_blocks_contiguous(struct list_head *head)
+{
+	struct drm_buddy_block *block;
+	u64 start, size;
+
+	block = amdgpu_vram_mgr_first_block(head);
+	if (!block)
+		return false;
+
+	while (head != block->link.next) {
+		start = amdgpu_vram_mgr_block_start(block);
+		size = amdgpu_vram_mgr_block_size(block);
+
+		block = list_entry(block->link.next, struct drm_buddy_block, link);
+		if (start + size != amdgpu_vram_mgr_block_start(block))
+			return false;
+	}
+
+	return true;
+}
+
+static inline u64 amdgpu_vram_mgr_blocks_size(struct list_head *head)
+{
+	struct drm_buddy_block *block;
+	u64 size = 0;
+
+	list_for_each_entry(block, head, link)
+		size += amdgpu_vram_mgr_block_size(block);
+
+	return size;
+}
+
 /**
- * amdgpu_vram_mgr_init - init VRAM manager and DRM MM
+ * DOC: mem_info_vram_total
  *
- * @man: TTM memory type manager
- * @p_size: maximum size of VRAM
- *
- * Allocate and initialize the VRAM manager.
+ * The amdgpu driver provides a sysfs API for reporting current total VRAM
+ * available on the device
+ * The file mem_info_vram_total is used for this and returns the total
+ * amount of VRAM in bytes
  */
-static int amdgpu_vram_mgr_init(struct ttm_mem_type_manager *man,
-				unsigned long p_size)
+static ssize_t amdgpu_mem_info_vram_total_show(struct device *dev,
+		struct device_attribute *attr, char *buf)
 {
-	struct amdgpu_vram_mgr *mgr;
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
 
-	mgr = kzalloc(sizeof(*mgr), GFP_KERNEL);
-	if (!mgr)
-		return -ENOMEM;
+	return sysfs_emit(buf, "%llu\n", adev->gmc.real_vram_size);
+}
 
-	drm_mm_init(&mgr->mm, 0, p_size);
-	spin_lock_init(&mgr->lock);
-	man->priv = mgr;
-	return 0;
+/**
+ * DOC: mem_info_vis_vram_total
+ *
+ * The amdgpu driver provides a sysfs API for reporting current total
+ * visible VRAM available on the device
+ * The file mem_info_vis_vram_total is used for this and returns the total
+ * amount of visible VRAM in bytes
+ */
+static ssize_t amdgpu_mem_info_vis_vram_total_show(struct device *dev,
+		struct device_attribute *attr, char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	return sysfs_emit(buf, "%llu\n", adev->gmc.visible_vram_size);
 }
 
 /**
- * amdgpu_vram_mgr_fini - free and destroy VRAM manager
+ * DOC: mem_info_vram_used
  *
- * @man: TTM memory type manager
+ * The amdgpu driver provides a sysfs API for reporting current total VRAM
+ * available on the device
+ * The file mem_info_vram_used is used for this and returns the total
+ * amount of currently used VRAM in bytes
+ */
+static ssize_t amdgpu_mem_info_vram_used_show(struct device *dev,
+					      struct device_attribute *attr,
+					      char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	struct ttm_resource_manager *man = &adev->mman.vram_mgr.manager;
+
+	return sysfs_emit(buf, "%llu\n", ttm_resource_manager_usage(man));
+}
+
+/**
+ * DOC: mem_info_vis_vram_used
  *
- * Destroy and free the VRAM manager, returns -EBUSY if ranges are still
- * allocated inside it.
+ * The amdgpu driver provides a sysfs API for reporting current total of
+ * used visible VRAM
+ * The file mem_info_vis_vram_used is used for this and returns the total
+ * amount of currently used visible VRAM in bytes
  */
-static int amdgpu_vram_mgr_fini(struct ttm_mem_type_manager *man)
+static ssize_t amdgpu_mem_info_vis_vram_used_show(struct device *dev,
+						  struct device_attribute *attr,
+						  char *buf)
 {
-	struct amdgpu_vram_mgr *mgr = man->priv;
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
 
-	spin_lock(&mgr->lock);
-	if (!drm_mm_clean(&mgr->mm)) {
-		spin_unlock(&mgr->lock);
-		return -EBUSY;
+	return sysfs_emit(buf, "%llu\n",
+			  amdgpu_vram_mgr_vis_usage(&adev->mman.vram_mgr));
+}
+
+/**
+ * DOC: mem_info_vram_vendor
+ *
+ * The amdgpu driver provides a sysfs API for reporting the vendor of the
+ * installed VRAM
+ * The file mem_info_vram_vendor is used for this and returns the name of the
+ * vendor.
+ */
+static ssize_t amdgpu_mem_info_vram_vendor(struct device *dev,
+					   struct device_attribute *attr,
+					   char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	switch (adev->gmc.vram_vendor) {
+	case SAMSUNG:
+		return sysfs_emit(buf, "samsung\n");
+	case INFINEON:
+		return sysfs_emit(buf, "infineon\n");
+	case ELPIDA:
+		return sysfs_emit(buf, "elpida\n");
+	case ETRON:
+		return sysfs_emit(buf, "etron\n");
+	case NANYA:
+		return sysfs_emit(buf, "nanya\n");
+	case HYNIX:
+		return sysfs_emit(buf, "hynix\n");
+	case MOSEL:
+		return sysfs_emit(buf, "mosel\n");
+	case WINBOND:
+		return sysfs_emit(buf, "winbond\n");
+	case ESMT:
+		return sysfs_emit(buf, "esmt\n");
+	case MICRON:
+		return sysfs_emit(buf, "micron\n");
+	default:
+		return sysfs_emit(buf, "unknown\n");
 	}
+}
 
-	drm_mm_takedown(&mgr->mm);
-	spin_unlock(&mgr->lock);
-	kfree(mgr);
-	man->priv = NULL;
-	return 0;
+static DEVICE_ATTR(mem_info_vram_total, S_IRUGO,
+		   amdgpu_mem_info_vram_total_show, NULL);
+static DEVICE_ATTR(mem_info_vis_vram_total, S_IRUGO,
+		   amdgpu_mem_info_vis_vram_total_show,NULL);
+static DEVICE_ATTR(mem_info_vram_used, S_IRUGO,
+		   amdgpu_mem_info_vram_used_show, NULL);
+static DEVICE_ATTR(mem_info_vis_vram_used, S_IRUGO,
+		   amdgpu_mem_info_vis_vram_used_show, NULL);
+static DEVICE_ATTR(mem_info_vram_vendor, S_IRUGO,
+		   amdgpu_mem_info_vram_vendor, NULL);
+
+static struct attribute *amdgpu_vram_mgr_attributes[] = {
+	&dev_attr_mem_info_vram_total.attr,
+	&dev_attr_mem_info_vis_vram_total.attr,
+	&dev_attr_mem_info_vram_used.attr,
+	&dev_attr_mem_info_vis_vram_used.attr,
+	&dev_attr_mem_info_vram_vendor.attr,
+	NULL
+};
+
+static umode_t amdgpu_vram_attrs_is_visible(struct kobject *kobj,
+					    struct attribute *attr, int i)
+{
+	struct device *dev = kobj_to_dev(kobj);
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	if (attr == &dev_attr_mem_info_vram_vendor.attr &&
+	    !adev->gmc.vram_vendor)
+		return 0;
+
+	return attr->mode;
 }
 
+const struct attribute_group amdgpu_vram_mgr_attr_group = {
+	.attrs = amdgpu_vram_mgr_attributes,
+	.is_visible = amdgpu_vram_attrs_is_visible
+};
+
 /**
- * amdgpu_vram_mgr_vis_size - Calculate visible node size
+ * amdgpu_vram_mgr_vis_size - Calculate visible block size
  *
- * @adev: amdgpu device structure
- * @node: MM node structure
+ * @adev: amdgpu_device pointer
+ * @block: DRM BUDDY block structure
  *
- * Calculate how many bytes of the MM node are inside visible VRAM
+ * Calculate how many bytes of the DRM BUDDY block are inside visible VRAM
  */
 static u64 amdgpu_vram_mgr_vis_size(struct amdgpu_device *adev,
-				    struct drm_mm_node *node)
+				    struct drm_buddy_block *block)
+{
+	u64 start = amdgpu_vram_mgr_block_start(block);
+	u64 end = start + amdgpu_vram_mgr_block_size(block);
+
+	if (start >= adev->gmc.visible_vram_size)
+		return 0;
+
+	return (end > adev->gmc.visible_vram_size ?
+		adev->gmc.visible_vram_size : end) - start;
+}
+
+/**
+ * amdgpu_vram_mgr_bo_visible_size - CPU visible BO size
+ *
+ * @bo: &amdgpu_bo buffer object (must be in VRAM)
+ *
+ * Returns:
+ * How much of the given &amdgpu_bo buffer object lies in CPU visible VRAM.
+ */
+u64 amdgpu_vram_mgr_bo_visible_size(struct amdgpu_bo *bo)
 {
-	uint64_t start = node->start << PAGE_SHIFT;
-	uint64_t end = (node->size + node->start) << PAGE_SHIFT;
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	struct ttm_resource *res = bo->tbo.resource;
+	struct amdgpu_vram_mgr_resource *vres = to_amdgpu_vram_mgr_resource(res);
+	struct drm_buddy_block *block;
+	u64 usage = 0;
+
+	if (amdgpu_gmc_vram_full_visible(&adev->gmc))
+		return amdgpu_bo_size(bo);
 
-	if (start >= adev->mc.visible_vram_size)
+	if (res->start >= adev->gmc.visible_vram_size >> PAGE_SHIFT)
 		return 0;
 
-	return (end > adev->mc.visible_vram_size ?
-		adev->mc.visible_vram_size : end) - start;
+	list_for_each_entry(block, &vres->blocks, link)
+		usage += amdgpu_vram_mgr_vis_size(adev, block);
+
+	return usage;
+}
+
+/* Commit the reservation of VRAM pages */
+static void amdgpu_vram_mgr_do_reserve(struct ttm_resource_manager *man)
+{
+	struct amdgpu_vram_mgr *mgr = to_vram_mgr(man);
+	struct amdgpu_device *adev = to_amdgpu_device(mgr);
+	struct drm_buddy *mm = &mgr->mm;
+	struct amdgpu_vram_reservation *rsv, *temp;
+	struct drm_buddy_block *block;
+	uint64_t vis_usage;
+
+	list_for_each_entry_safe(rsv, temp, &mgr->reservations_pending, blocks) {
+		if (drm_buddy_alloc_blocks(mm, rsv->start, rsv->start + rsv->size,
+					   rsv->size, mm->chunk_size, &rsv->allocated,
+					   DRM_BUDDY_RANGE_ALLOCATION))
+			continue;
+
+		block = amdgpu_vram_mgr_first_block(&rsv->allocated);
+		if (!block)
+			continue;
+
+		dev_dbg(adev->dev, "Reservation 0x%llx - %lld, Succeeded\n",
+			rsv->start, rsv->size);
+
+		vis_usage = amdgpu_vram_mgr_vis_size(adev, block);
+		atomic64_add(vis_usage, &mgr->vis_usage);
+		spin_lock(&man->bdev->lru_lock);
+		man->usage += rsv->size;
+		spin_unlock(&man->bdev->lru_lock);
+		list_move(&rsv->blocks, &mgr->reserved_pages);
+	}
+}
+
+/**
+ * amdgpu_vram_mgr_reserve_range - Reserve a range from VRAM
+ *
+ * @mgr: amdgpu_vram_mgr pointer
+ * @start: start address of the range in VRAM
+ * @size: size of the range
+ *
+ * Reserve memory from start address with the specified size in VRAM
+ */
+int amdgpu_vram_mgr_reserve_range(struct amdgpu_vram_mgr *mgr,
+				  uint64_t start, uint64_t size)
+{
+	struct amdgpu_vram_reservation *rsv;
+
+	rsv = kzalloc(sizeof(*rsv), GFP_KERNEL);
+	if (!rsv)
+		return -ENOMEM;
+
+	INIT_LIST_HEAD(&rsv->allocated);
+	INIT_LIST_HEAD(&rsv->blocks);
+
+	rsv->start = start;
+	rsv->size = size;
+
+	mutex_lock(&mgr->lock);
+	list_add_tail(&rsv->blocks, &mgr->reservations_pending);
+	amdgpu_vram_mgr_do_reserve(&mgr->manager);
+	mutex_unlock(&mgr->lock);
+
+	return 0;
+}
+
+/**
+ * amdgpu_vram_mgr_query_page_status - query the reservation status
+ *
+ * @mgr: amdgpu_vram_mgr pointer
+ * @start: start address of a page in VRAM
+ *
+ * Returns:
+ *	-EBUSY: the page is still hold and in pending list
+ *	0: the page has been reserved
+ *	-ENOENT: the input page is not a reservation
+ */
+int amdgpu_vram_mgr_query_page_status(struct amdgpu_vram_mgr *mgr,
+				      uint64_t start)
+{
+	struct amdgpu_vram_reservation *rsv;
+	int ret;
+
+	mutex_lock(&mgr->lock);
+
+	list_for_each_entry(rsv, &mgr->reservations_pending, blocks) {
+		if (rsv->start <= start &&
+		    (start < (rsv->start + rsv->size))) {
+			ret = -EBUSY;
+			goto out;
+		}
+	}
+
+	list_for_each_entry(rsv, &mgr->reserved_pages, blocks) {
+		if (rsv->start <= start &&
+		    (start < (rsv->start + rsv->size))) {
+			ret = 0;
+			goto out;
+		}
+	}
+
+	ret = -ENOENT;
+out:
+	mutex_unlock(&mgr->lock);
+	return ret;
+}
+
+int amdgpu_vram_mgr_query_address_block_info(struct amdgpu_vram_mgr *mgr,
+			uint64_t address, struct amdgpu_vram_block_info *info)
+{
+	struct amdgpu_vram_mgr_resource *vres;
+	struct drm_buddy_block *block;
+	u64 start, size;
+	int ret = -ENOENT;
+
+	mutex_lock(&mgr->lock);
+	list_for_each_entry(vres, &mgr->allocated_vres_list, vres_node) {
+		list_for_each_entry(block, &vres->blocks, link) {
+			start = amdgpu_vram_mgr_block_start(block);
+			size = amdgpu_vram_mgr_block_size(block);
+			if ((start <= address) && (address < (start + size))) {
+				info->start = start;
+				info->size = size;
+				memcpy(&info->task, &vres->task, sizeof(vres->task));
+				ret = 0;
+				goto out;
+			}
+		}
+	}
+
+out:
+	mutex_unlock(&mgr->lock);
+
+	return ret;
 }
 
 /**
@@ -107,166 +431,438 @@ static u64 amdgpu_vram_mgr_vis_size(struct amdgpu_device *adev,
  * @man: TTM memory type manager
  * @tbo: TTM BO we need this range for
  * @place: placement flags and restrictions
- * @mem: the resulting mem object
+ * @res: the resulting mem object
  *
  * Allocate VRAM for the given BO.
  */
-static int amdgpu_vram_mgr_new(struct ttm_mem_type_manager *man,
+static int amdgpu_vram_mgr_new(struct ttm_resource_manager *man,
 			       struct ttm_buffer_object *tbo,
 			       const struct ttm_place *place,
-			       struct ttm_mem_reg *mem)
+			       struct ttm_resource **res)
 {
-	struct amdgpu_device *adev = amdgpu_ttm_adev(man->bdev);
-	struct amdgpu_vram_mgr *mgr = man->priv;
-	struct drm_mm *mm = &mgr->mm;
-	struct drm_mm_node *nodes;
-	enum drm_mm_insert_mode mode;
-	unsigned long lpfn, num_nodes, pages_per_node, pages_left;
-	uint64_t usage = 0, vis_usage = 0;
-	unsigned i;
+	struct amdgpu_vram_mgr *mgr = to_vram_mgr(man);
+	struct amdgpu_device *adev = to_amdgpu_device(mgr);
+	struct amdgpu_bo *bo = ttm_to_amdgpu_bo(tbo);
+	u64 vis_usage = 0, max_bytes, min_block_size;
+	struct amdgpu_vram_mgr_resource *vres;
+	u64 size, remaining_size, lpfn, fpfn;
+	unsigned int adjust_dcc_size = 0;
+	struct drm_buddy *mm = &mgr->mm;
+	struct drm_buddy_block *block;
+	unsigned long pages_per_block;
 	int r;
 
-	lpfn = place->lpfn;
-	if (!lpfn)
+	lpfn = (u64)place->lpfn << PAGE_SHIFT;
+	if (!lpfn || lpfn > man->size)
 		lpfn = man->size;
 
-	if (place->flags & TTM_PL_FLAG_CONTIGUOUS ||
-	    amdgpu_vram_page_split == -1) {
-		pages_per_node = ~0ul;
-		num_nodes = 1;
+	fpfn = (u64)place->fpfn << PAGE_SHIFT;
+
+	max_bytes = adev->gmc.mc_vram_size;
+	if (tbo->type != ttm_bo_type_kernel)
+		max_bytes -= AMDGPU_VM_RESERVED_VRAM;
+
+	if (bo->flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS) {
+		pages_per_block = ~0ul;
 	} else {
-		pages_per_node = max((uint32_t)amdgpu_vram_page_split,
-				     mem->page_alignment);
-		num_nodes = DIV_ROUND_UP(mem->num_pages, pages_per_node);
+#ifdef CONFIG_TRANSPARENT_HUGEPAGE
+		pages_per_block = HPAGE_PMD_NR;
+#else
+		/* default to 2MB */
+		pages_per_block = 2UL << (20UL - PAGE_SHIFT);
+#endif
+		pages_per_block = max_t(u32, pages_per_block,
+					tbo->page_alignment);
 	}
 
-	nodes = kcalloc(num_nodes, sizeof(*nodes), GFP_KERNEL);
-	if (!nodes)
+	vres = kzalloc(sizeof(*vres), GFP_KERNEL);
+	if (!vres)
 		return -ENOMEM;
 
-	mode = DRM_MM_INSERT_BEST;
+	ttm_resource_init(tbo, place, &vres->base);
+
+	/* bail out quickly if there's likely not enough VRAM for this BO */
+	if (ttm_resource_manager_usage(man) > max_bytes) {
+		r = -ENOSPC;
+		goto error_fini;
+	}
+
+	INIT_LIST_HEAD(&vres->blocks);
+
 	if (place->flags & TTM_PL_FLAG_TOPDOWN)
-		mode = DRM_MM_INSERT_HIGH;
+		vres->flags |= DRM_BUDDY_TOPDOWN_ALLOCATION;
 
-	mem->start = 0;
-	pages_left = mem->num_pages;
+	if (bo->flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS)
+		vres->flags |= DRM_BUDDY_CONTIGUOUS_ALLOCATION;
 
-	spin_lock(&mgr->lock);
-	for (i = 0; i < num_nodes; ++i) {
-		unsigned long pages = min(pages_left, pages_per_node);
-		uint32_t alignment = mem->page_alignment;
-		unsigned long start;
+	if (bo->flags & AMDGPU_GEM_CREATE_VRAM_CLEARED)
+		vres->flags |= DRM_BUDDY_CLEAR_ALLOCATION;
+
+	if (fpfn || lpfn != mgr->mm.size)
+		/* Allocate blocks in desired range */
+		vres->flags |= DRM_BUDDY_RANGE_ALLOCATION;
+
+	if (bo->flags & AMDGPU_GEM_CREATE_GFX12_DCC &&
+	    adev->gmc.gmc_funcs->get_dcc_alignment)
+		adjust_dcc_size = amdgpu_gmc_get_dcc_alignment(adev);
+
+	remaining_size = (u64)vres->base.size;
+	if (bo->flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS && adjust_dcc_size) {
+		unsigned int dcc_size;
+
+		dcc_size = roundup_pow_of_two(vres->base.size + adjust_dcc_size);
+		remaining_size = (u64)dcc_size;
+
+		vres->flags |= DRM_BUDDY_TRIM_DISABLE;
+	}
+
+	mutex_lock(&mgr->lock);
+	while (remaining_size) {
+		if (tbo->page_alignment)
+			min_block_size = (u64)tbo->page_alignment << PAGE_SHIFT;
+		else
+			min_block_size = mgr->default_page_size;
+
+		size = remaining_size;
 
-		if (pages == pages_per_node)
-			alignment = pages_per_node;
+		if (bo->flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS && adjust_dcc_size)
+			min_block_size = size;
+		else if ((size >= (u64)pages_per_block << PAGE_SHIFT) &&
+			 !(size & (((u64)pages_per_block << PAGE_SHIFT) - 1)))
+			min_block_size = (u64)pages_per_block << PAGE_SHIFT;
+
+		BUG_ON(min_block_size < mm->chunk_size);
+
+		r = drm_buddy_alloc_blocks(mm, fpfn,
+					   lpfn,
+					   size,
+					   min_block_size,
+					   &vres->blocks,
+					   vres->flags);
+
+		if (unlikely(r == -ENOSPC) && pages_per_block == ~0ul &&
+		    !(place->flags & TTM_PL_FLAG_CONTIGUOUS)) {
+			vres->flags &= ~DRM_BUDDY_CONTIGUOUS_ALLOCATION;
+			pages_per_block = max_t(u32, 2UL << (20UL - PAGE_SHIFT),
+						tbo->page_alignment);
+
+			continue;
+		}
 
-		r = drm_mm_insert_node_in_range(mm, &nodes[i],
-						pages, alignment, 0,
-						place->fpfn, lpfn,
-						mode);
 		if (unlikely(r))
-			goto error;
+			goto error_free_blocks;
+
+		if (size > remaining_size)
+			remaining_size = 0;
+		else
+			remaining_size -= size;
+	}
 
-		usage += nodes[i].size << PAGE_SHIFT;
-		vis_usage += amdgpu_vram_mgr_vis_size(adev, &nodes[i]);
+	vres->task.pid = task_pid_nr(current);
+	get_task_comm(vres->task.comm, current);
+	list_add_tail(&vres->vres_node, &mgr->allocated_vres_list);
+
+	if (bo->flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS && adjust_dcc_size) {
+		struct drm_buddy_block *dcc_block;
+		unsigned long dcc_start;
+		u64 trim_start;
+
+		dcc_block = amdgpu_vram_mgr_first_block(&vres->blocks);
+		/* Adjust the start address for DCC buffers only */
+		dcc_start =
+			roundup((unsigned long)amdgpu_vram_mgr_block_start(dcc_block),
+				adjust_dcc_size);
+		trim_start = (u64)dcc_start;
+		drm_buddy_block_trim(mm, &trim_start,
+				     (u64)vres->base.size,
+				     &vres->blocks);
+	}
+	mutex_unlock(&mgr->lock);
+
+	vres->base.start = 0;
+	size = max_t(u64, amdgpu_vram_mgr_blocks_size(&vres->blocks),
+		     vres->base.size);
+	list_for_each_entry(block, &vres->blocks, link) {
+		unsigned long start;
+
+		start = amdgpu_vram_mgr_block_start(block) +
+			amdgpu_vram_mgr_block_size(block);
+		start >>= PAGE_SHIFT;
 
-		/* Calculate a virtual BO start address to easily check if
-		 * everything is CPU accessible.
-		 */
-		start = nodes[i].start + nodes[i].size;
-		if (start > mem->num_pages)
-			start -= mem->num_pages;
+		if (start > PFN_UP(size))
+			start -= PFN_UP(size);
 		else
 			start = 0;
-		mem->start = max(mem->start, start);
-		pages_left -= pages;
+		vres->base.start = max(vres->base.start, start);
+
+		vis_usage += amdgpu_vram_mgr_vis_size(adev, block);
 	}
-	spin_unlock(&mgr->lock);
 
-	atomic64_add(usage, &mgr->usage);
-	atomic64_add(vis_usage, &mgr->vis_usage);
+	if (amdgpu_is_vram_mgr_blocks_contiguous(&vres->blocks))
+		vres->base.placement |= TTM_PL_FLAG_CONTIGUOUS;
 
-	mem->mm_node = nodes;
+	if (adev->gmc.xgmi.connected_to_cpu)
+		vres->base.bus.caching = ttm_cached;
+	else
+		vres->base.bus.caching = ttm_write_combined;
 
+	atomic64_add(vis_usage, &mgr->vis_usage);
+	*res = &vres->base;
 	return 0;
 
-error:
-	while (i--)
-		drm_mm_remove_node(&nodes[i]);
-	spin_unlock(&mgr->lock);
+error_free_blocks:
+	drm_buddy_free_list(mm, &vres->blocks, 0);
+	mutex_unlock(&mgr->lock);
+error_fini:
+	ttm_resource_fini(man, &vres->base);
+	kfree(vres);
 
-	kfree(nodes);
-	return r == -ENOSPC ? 0 : r;
+	return r;
 }
 
 /**
  * amdgpu_vram_mgr_del - free ranges
  *
  * @man: TTM memory type manager
- * @tbo: TTM BO we need this range for
- * @place: placement flags and restrictions
- * @mem: TTM memory object
+ * @res: TTM memory object
  *
  * Free the allocated VRAM again.
  */
-static void amdgpu_vram_mgr_del(struct ttm_mem_type_manager *man,
-				struct ttm_mem_reg *mem)
+static void amdgpu_vram_mgr_del(struct ttm_resource_manager *man,
+				struct ttm_resource *res)
 {
-	struct amdgpu_device *adev = amdgpu_ttm_adev(man->bdev);
-	struct amdgpu_vram_mgr *mgr = man->priv;
-	struct drm_mm_node *nodes = mem->mm_node;
-	uint64_t usage = 0, vis_usage = 0;
-	unsigned pages = mem->num_pages;
+	struct amdgpu_vram_mgr_resource *vres = to_amdgpu_vram_mgr_resource(res);
+	struct amdgpu_vram_mgr *mgr = to_vram_mgr(man);
+	struct amdgpu_device *adev = to_amdgpu_device(mgr);
+	struct drm_buddy *mm = &mgr->mm;
+	struct drm_buddy_block *block;
+	uint64_t vis_usage = 0;
 
-	if (!mem->mm_node)
-		return;
+	mutex_lock(&mgr->lock);
 
-	spin_lock(&mgr->lock);
-	while (pages) {
-		pages -= nodes->size;
-		drm_mm_remove_node(nodes);
-		usage += nodes->size << PAGE_SHIFT;
-		vis_usage += amdgpu_vram_mgr_vis_size(adev, nodes);
-		++nodes;
-	}
-	spin_unlock(&mgr->lock);
+	list_del(&vres->vres_node);
+	memset(&vres->task, 0, sizeof(vres->task));
+
+	list_for_each_entry(block, &vres->blocks, link)
+		vis_usage += amdgpu_vram_mgr_vis_size(adev, block);
+
+	drm_buddy_free_list(mm, &vres->blocks, vres->flags);
+	amdgpu_vram_mgr_do_reserve(man);
+	mutex_unlock(&mgr->lock);
 
-	atomic64_sub(usage, &mgr->usage);
 	atomic64_sub(vis_usage, &mgr->vis_usage);
 
-	kfree(mem->mm_node);
-	mem->mm_node = NULL;
+	ttm_resource_fini(man, res);
+	kfree(vres);
 }
 
 /**
- * amdgpu_vram_mgr_usage - how many bytes are used in this domain
+ * amdgpu_vram_mgr_alloc_sgt - allocate and fill a sg table
  *
- * @man: TTM memory type manager
+ * @adev: amdgpu device pointer
+ * @res: TTM memory object
+ * @offset: byte offset from the base of VRAM BO
+ * @length: number of bytes to export in sg_table
+ * @dev: the other device
+ * @dir: dma direction
+ * @sgt: resulting sg table
  *
- * Returns how many bytes are used in this domain.
+ * Allocate and fill a sg table from a VRAM allocation.
  */
-uint64_t amdgpu_vram_mgr_usage(struct ttm_mem_type_manager *man)
+int amdgpu_vram_mgr_alloc_sgt(struct amdgpu_device *adev,
+			      struct ttm_resource *res,
+			      u64 offset, u64 length,
+			      struct device *dev,
+			      enum dma_data_direction dir,
+			      struct sg_table **sgt)
 {
-	struct amdgpu_vram_mgr *mgr = man->priv;
+	struct amdgpu_res_cursor cursor;
+	struct scatterlist *sg;
+	int num_entries = 0;
+	int i, r;
+
+	*sgt = kmalloc(sizeof(**sgt), GFP_KERNEL);
+	if (!*sgt)
+		return -ENOMEM;
+
+	/* Determine the number of DRM_BUDDY blocks to export */
+	amdgpu_res_first(res, offset, length, &cursor);
+	while (cursor.remaining) {
+		num_entries++;
+		amdgpu_res_next(&cursor, min(cursor.size, AMDGPU_MAX_SG_SEGMENT_SIZE));
+	}
+
+	r = sg_alloc_table(*sgt, num_entries, GFP_KERNEL);
+	if (r)
+		goto error_free;
+
+	/* Initialize scatterlist nodes of sg_table */
+	for_each_sgtable_sg((*sgt), sg, i)
+		sg->length = 0;
+
+	/*
+	 * Walk down DRM_BUDDY blocks to populate scatterlist nodes
+	 * @note: Use iterator api to get first the DRM_BUDDY block
+	 * and the number of bytes from it. Access the following
+	 * DRM_BUDDY block(s) if more buffer needs to exported
+	 */
+	amdgpu_res_first(res, offset, length, &cursor);
+	for_each_sgtable_sg((*sgt), sg, i) {
+		phys_addr_t phys = cursor.start + adev->gmc.aper_base;
+		unsigned long size = min(cursor.size, AMDGPU_MAX_SG_SEGMENT_SIZE);
+		dma_addr_t addr;
+
+		addr = dma_map_resource(dev, phys, size, dir,
+					DMA_ATTR_SKIP_CPU_SYNC);
+		r = dma_mapping_error(dev, addr);
+		if (r)
+			goto error_unmap;
+
+		sg_set_page(sg, NULL, size, 0);
+		sg_dma_address(sg) = addr;
+		sg_dma_len(sg) = size;
+
+		amdgpu_res_next(&cursor, size);
+	}
+
+	return 0;
+
+error_unmap:
+	for_each_sgtable_sg((*sgt), sg, i) {
+		if (!sg->length)
+			continue;
+
+		dma_unmap_resource(dev, sg->dma_address,
+				   sg->length, dir,
+				   DMA_ATTR_SKIP_CPU_SYNC);
+	}
+	sg_free_table(*sgt);
 
-	return atomic64_read(&mgr->usage);
+error_free:
+	kfree(*sgt);
+	return r;
+}
+
+/**
+ * amdgpu_vram_mgr_free_sgt - allocate and fill a sg table
+ *
+ * @dev: device pointer
+ * @dir: data direction of resource to unmap
+ * @sgt: sg table to free
+ *
+ * Free a previously allocate sg table.
+ */
+void amdgpu_vram_mgr_free_sgt(struct device *dev,
+			      enum dma_data_direction dir,
+			      struct sg_table *sgt)
+{
+	struct scatterlist *sg;
+	int i;
+
+	for_each_sgtable_sg(sgt, sg, i)
+		dma_unmap_resource(dev, sg->dma_address,
+				   sg->length, dir,
+				   DMA_ATTR_SKIP_CPU_SYNC);
+	sg_free_table(sgt);
+	kfree(sgt);
 }
 
 /**
  * amdgpu_vram_mgr_vis_usage - how many bytes are used in the visible part
  *
- * @man: TTM memory type manager
+ * @mgr: amdgpu_vram_mgr pointer
  *
  * Returns how many bytes are used in the visible part of VRAM
  */
-uint64_t amdgpu_vram_mgr_vis_usage(struct ttm_mem_type_manager *man)
+uint64_t amdgpu_vram_mgr_vis_usage(struct amdgpu_vram_mgr *mgr)
 {
-	struct amdgpu_vram_mgr *mgr = man->priv;
-
 	return atomic64_read(&mgr->vis_usage);
 }
 
 /**
+ * amdgpu_vram_mgr_clear_reset_blocks - reset clear blocks
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Reset the cleared drm buddy blocks.
+ */
+void amdgpu_vram_mgr_clear_reset_blocks(struct amdgpu_device *adev)
+{
+	struct amdgpu_vram_mgr *mgr = &adev->mman.vram_mgr;
+	struct drm_buddy *mm = &mgr->mm;
+
+	mutex_lock(&mgr->lock);
+	drm_buddy_reset_clear(mm, false);
+	mutex_unlock(&mgr->lock);
+}
+
+/**
+ * amdgpu_vram_mgr_intersects - test each drm buddy block for intersection
+ *
+ * @man: TTM memory type manager
+ * @res: The resource to test
+ * @place: The place to test against
+ * @size: Size of the new allocation
+ *
+ * Test each drm buddy block for intersection for eviction decision.
+ */
+static bool amdgpu_vram_mgr_intersects(struct ttm_resource_manager *man,
+				       struct ttm_resource *res,
+				       const struct ttm_place *place,
+				       size_t size)
+{
+	struct amdgpu_vram_mgr_resource *mgr = to_amdgpu_vram_mgr_resource(res);
+	struct drm_buddy_block *block;
+
+	/* Check each drm buddy block individually */
+	list_for_each_entry(block, &mgr->blocks, link) {
+		unsigned long fpfn =
+			amdgpu_vram_mgr_block_start(block) >> PAGE_SHIFT;
+		unsigned long lpfn = fpfn +
+			(amdgpu_vram_mgr_block_size(block) >> PAGE_SHIFT);
+
+		if (place->fpfn < lpfn &&
+		    (!place->lpfn || place->lpfn > fpfn))
+			return true;
+	}
+
+	return false;
+}
+
+/**
+ * amdgpu_vram_mgr_compatible - test each drm buddy block for compatibility
+ *
+ * @man: TTM memory type manager
+ * @res: The resource to test
+ * @place: The place to test against
+ * @size: Size of the new allocation
+ *
+ * Test each drm buddy block for placement compatibility.
+ */
+static bool amdgpu_vram_mgr_compatible(struct ttm_resource_manager *man,
+				       struct ttm_resource *res,
+				       const struct ttm_place *place,
+				       size_t size)
+{
+	struct amdgpu_vram_mgr_resource *mgr = to_amdgpu_vram_mgr_resource(res);
+	struct drm_buddy_block *block;
+
+	/* Check each drm buddy block individually */
+	list_for_each_entry(block, &mgr->blocks, link) {
+		unsigned long fpfn =
+			amdgpu_vram_mgr_block_start(block) >> PAGE_SHIFT;
+		unsigned long lpfn = fpfn +
+			(amdgpu_vram_mgr_block_size(block) >> PAGE_SHIFT);
+
+		if (fpfn < place->fpfn ||
+		    (place->lpfn && lpfn > place->lpfn))
+			return false;
+	}
+
+	return true;
+}
+
+/**
  * amdgpu_vram_mgr_debug - dump VRAM table
  *
  * @man: TTM memory type manager
@@ -274,24 +870,105 @@ uint64_t amdgpu_vram_mgr_vis_usage(struct ttm_mem_type_manager *man)
  *
  * Dump the table content using printk.
  */
-static void amdgpu_vram_mgr_debug(struct ttm_mem_type_manager *man,
+static void amdgpu_vram_mgr_debug(struct ttm_resource_manager *man,
 				  struct drm_printer *printer)
 {
-	struct amdgpu_vram_mgr *mgr = man->priv;
+	struct amdgpu_vram_mgr *mgr = to_vram_mgr(man);
+	struct drm_buddy *mm = &mgr->mm;
+	struct amdgpu_vram_reservation *rsv;
+
+	drm_printf(printer, "  vis usage:%llu\n",
+		   amdgpu_vram_mgr_vis_usage(mgr));
 
-	spin_lock(&mgr->lock);
-	drm_mm_print(&mgr->mm, printer);
-	spin_unlock(&mgr->lock);
+	mutex_lock(&mgr->lock);
+	drm_printf(printer, "default_page_size: %lluKiB\n",
+		   mgr->default_page_size >> 10);
 
-	drm_printf(printer, "man size:%llu pages, ram usage:%lluMB, vis usage:%lluMB\n",
-		   man->size, amdgpu_vram_mgr_usage(man) >> 20,
-		   amdgpu_vram_mgr_vis_usage(man) >> 20);
+	drm_buddy_print(mm, printer);
+
+	drm_printf(printer, "reserved:\n");
+	list_for_each_entry(rsv, &mgr->reserved_pages, blocks)
+		drm_printf(printer, "%#018llx-%#018llx: %llu\n",
+			rsv->start, rsv->start + rsv->size, rsv->size);
+	mutex_unlock(&mgr->lock);
 }
 
-const struct ttm_mem_type_manager_func amdgpu_vram_mgr_func = {
-	.init		= amdgpu_vram_mgr_init,
-	.takedown	= amdgpu_vram_mgr_fini,
-	.get_node	= amdgpu_vram_mgr_new,
-	.put_node	= amdgpu_vram_mgr_del,
-	.debug		= amdgpu_vram_mgr_debug
+static const struct ttm_resource_manager_func amdgpu_vram_mgr_func = {
+	.alloc	= amdgpu_vram_mgr_new,
+	.free	= amdgpu_vram_mgr_del,
+	.intersects = amdgpu_vram_mgr_intersects,
+	.compatible = amdgpu_vram_mgr_compatible,
+	.debug	= amdgpu_vram_mgr_debug
 };
+
+/**
+ * amdgpu_vram_mgr_init - init VRAM manager and DRM MM
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate and initialize the VRAM manager.
+ */
+int amdgpu_vram_mgr_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_vram_mgr *mgr = &adev->mman.vram_mgr;
+	struct ttm_resource_manager *man = &mgr->manager;
+	int err;
+
+	man->cg = drmm_cgroup_register_region(adev_to_drm(adev), "vram", adev->gmc.real_vram_size);
+	if (IS_ERR(man->cg))
+		return PTR_ERR(man->cg);
+	ttm_resource_manager_init(man, &adev->mman.bdev,
+				  adev->gmc.real_vram_size);
+
+	mutex_init(&mgr->lock);
+	INIT_LIST_HEAD(&mgr->reservations_pending);
+	INIT_LIST_HEAD(&mgr->reserved_pages);
+	INIT_LIST_HEAD(&mgr->allocated_vres_list);
+	mgr->default_page_size = PAGE_SIZE;
+
+	man->func = &amdgpu_vram_mgr_func;
+	err = drm_buddy_init(&mgr->mm, man->size, PAGE_SIZE);
+	if (err)
+		return err;
+
+	ttm_set_driver_manager(&adev->mman.bdev, TTM_PL_VRAM, &mgr->manager);
+	ttm_resource_manager_set_used(man, true);
+	return 0;
+}
+
+/**
+ * amdgpu_vram_mgr_fini - free and destroy VRAM manager
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Destroy and free the VRAM manager, returns -EBUSY if ranges are still
+ * allocated inside it.
+ */
+void amdgpu_vram_mgr_fini(struct amdgpu_device *adev)
+{
+	struct amdgpu_vram_mgr *mgr = &adev->mman.vram_mgr;
+	struct ttm_resource_manager *man = &mgr->manager;
+	int ret;
+	struct amdgpu_vram_reservation *rsv, *temp;
+
+	ttm_resource_manager_set_used(man, false);
+
+	ret = ttm_resource_manager_evict_all(&adev->mman.bdev, man);
+	if (ret)
+		return;
+
+	mutex_lock(&mgr->lock);
+	list_for_each_entry_safe(rsv, temp, &mgr->reservations_pending, blocks)
+		kfree(rsv);
+
+	list_for_each_entry_safe(rsv, temp, &mgr->reserved_pages, blocks) {
+		drm_buddy_free_list(&mgr->mm, &rsv->allocated, 0);
+		kfree(rsv);
+	}
+	if (!adev->gmc.is_app_apu)
+		drm_buddy_fini(&mgr->mm);
+	mutex_unlock(&mgr->lock);
+
+	ttm_resource_manager_cleanup(man);
+	ttm_set_driver_manager(&adev->mman.bdev, TTM_PL_VRAM, NULL);
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.h
new file mode 100644
index 000000000000..5f5fd9a911c2
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.h
@@ -0,0 +1,92 @@
+/* SPDX-License-Identifier: MIT
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_VRAM_MGR_H__
+#define __AMDGPU_VRAM_MGR_H__
+
+#include <drm/drm_buddy.h>
+
+struct amdgpu_vram_mgr {
+	struct ttm_resource_manager manager;
+	struct drm_buddy mm;
+	/* protects access to buffer objects */
+	struct mutex lock;
+	struct list_head reservations_pending;
+	struct list_head reserved_pages;
+	atomic64_t vis_usage;
+	u64 default_page_size;
+	struct list_head allocated_vres_list;
+};
+
+struct amdgpu_vres_task {
+	pid_t pid;
+	char  comm[TASK_COMM_LEN];
+};
+
+struct amdgpu_vram_block_info {
+	u64 start;
+	u64 size;
+	struct amdgpu_vres_task task;
+};
+
+struct amdgpu_vram_mgr_resource {
+	struct ttm_resource base;
+	struct list_head blocks;
+	unsigned long flags;
+	struct list_head vres_node;
+	struct amdgpu_vres_task task;
+};
+
+static inline u64 amdgpu_vram_mgr_block_start(struct drm_buddy_block *block)
+{
+	return drm_buddy_block_offset(block);
+}
+
+static inline u64 amdgpu_vram_mgr_block_size(struct drm_buddy_block *block)
+{
+	return (u64)PAGE_SIZE << drm_buddy_block_order(block);
+}
+
+static inline bool amdgpu_vram_mgr_is_cleared(struct drm_buddy_block *block)
+{
+	return drm_buddy_block_is_clear(block);
+}
+
+static inline struct amdgpu_vram_mgr_resource *
+to_amdgpu_vram_mgr_resource(struct ttm_resource *res)
+{
+	return container_of(res, struct amdgpu_vram_mgr_resource, base);
+}
+
+static inline void amdgpu_vram_mgr_set_cleared(struct ttm_resource *res)
+{
+	struct amdgpu_vram_mgr_resource *ares = to_amdgpu_vram_mgr_resource(res);
+
+	WARN_ON(ares->flags & DRM_BUDDY_CLEARED);
+	ares->flags |= DRM_BUDDY_CLEARED;
+}
+
+int amdgpu_vram_mgr_query_address_block_info(struct amdgpu_vram_mgr *mgr,
+		uint64_t address, struct amdgpu_vram_block_info *info);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c
new file mode 100644
index 000000000000..1083db8cea2e
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c
@@ -0,0 +1,1107 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "amdgpu_xcp.h"
+#include "amdgpu_drv.h"
+
+#include <drm/drm_drv.h>
+#include "../amdxcp/amdgpu_xcp_drv.h"
+
+static void amdgpu_xcp_sysfs_entries_init(struct amdgpu_xcp_mgr *xcp_mgr);
+static void amdgpu_xcp_sysfs_entries_update(struct amdgpu_xcp_mgr *xcp_mgr);
+
+static int __amdgpu_xcp_run(struct amdgpu_xcp_mgr *xcp_mgr,
+			    struct amdgpu_xcp_ip *xcp_ip, int xcp_state)
+{
+	int (*run_func)(void *handle, uint32_t inst_mask);
+	int ret = 0;
+
+	if (!xcp_ip || !xcp_ip->valid || !xcp_ip->ip_funcs)
+		return 0;
+
+	run_func = NULL;
+
+	switch (xcp_state) {
+	case AMDGPU_XCP_PREPARE_SUSPEND:
+		run_func = xcp_ip->ip_funcs->prepare_suspend;
+		break;
+	case AMDGPU_XCP_SUSPEND:
+		run_func = xcp_ip->ip_funcs->suspend;
+		break;
+	case AMDGPU_XCP_PREPARE_RESUME:
+		run_func = xcp_ip->ip_funcs->prepare_resume;
+		break;
+	case AMDGPU_XCP_RESUME:
+		run_func = xcp_ip->ip_funcs->resume;
+		break;
+	}
+
+	if (run_func)
+		ret = run_func(xcp_mgr->adev, xcp_ip->inst_mask);
+
+	return ret;
+}
+
+static int amdgpu_xcp_run_transition(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id,
+				     int state)
+{
+	struct amdgpu_xcp_ip *xcp_ip;
+	struct amdgpu_xcp *xcp;
+	int i, ret;
+
+	if (xcp_id >= MAX_XCP || !xcp_mgr->xcp[xcp_id].valid)
+		return -EINVAL;
+
+	xcp = &xcp_mgr->xcp[xcp_id];
+	for (i = 0; i < AMDGPU_XCP_MAX_BLOCKS; ++i) {
+		xcp_ip = &xcp->ip[i];
+		ret = __amdgpu_xcp_run(xcp_mgr, xcp_ip, state);
+		if (ret)
+			break;
+	}
+
+	return ret;
+}
+
+int amdgpu_xcp_prepare_suspend(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id)
+{
+	return amdgpu_xcp_run_transition(xcp_mgr, xcp_id,
+					 AMDGPU_XCP_PREPARE_SUSPEND);
+}
+
+int amdgpu_xcp_suspend(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id)
+{
+	return amdgpu_xcp_run_transition(xcp_mgr, xcp_id, AMDGPU_XCP_SUSPEND);
+}
+
+int amdgpu_xcp_prepare_resume(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id)
+{
+	return amdgpu_xcp_run_transition(xcp_mgr, xcp_id,
+					 AMDGPU_XCP_PREPARE_RESUME);
+}
+
+int amdgpu_xcp_resume(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id)
+{
+	return amdgpu_xcp_run_transition(xcp_mgr, xcp_id, AMDGPU_XCP_RESUME);
+}
+
+static void __amdgpu_xcp_add_block(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id,
+				   struct amdgpu_xcp_ip *ip)
+{
+	struct amdgpu_xcp *xcp;
+
+	if (!ip)
+		return;
+
+	xcp = &xcp_mgr->xcp[xcp_id];
+	xcp->ip[ip->ip_id] = *ip;
+	xcp->ip[ip->ip_id].valid = true;
+
+	xcp->valid = true;
+}
+
+static void __amdgpu_xcp_set_unique_id(struct amdgpu_xcp_mgr *xcp_mgr,
+				       int xcp_id)
+{
+	struct amdgpu_xcp *xcp = &xcp_mgr->xcp[xcp_id];
+	struct amdgpu_device *adev = xcp_mgr->adev;
+	uint32_t inst_mask;
+	uint64_t uid;
+	int i;
+
+	if (!amdgpu_xcp_get_inst_details(xcp, AMDGPU_XCP_GFX, &inst_mask) &&
+	    inst_mask) {
+		i = GET_INST(GC, (ffs(inst_mask) - 1));
+		uid = amdgpu_device_get_uid(xcp_mgr->adev->uid_info,
+					    AMDGPU_UID_TYPE_XCD, i);
+		if (uid)
+			xcp->unique_id = uid;
+	}
+}
+
+int amdgpu_xcp_init(struct amdgpu_xcp_mgr *xcp_mgr, int num_xcps, int mode)
+{
+	struct amdgpu_device *adev = xcp_mgr->adev;
+	struct amdgpu_xcp_ip ip;
+	uint8_t mem_id;
+	int i, j, ret;
+
+	if (!num_xcps || num_xcps > MAX_XCP)
+		return -EINVAL;
+
+	xcp_mgr->mode = mode;
+
+	for (i = 0; i < MAX_XCP; ++i)
+		xcp_mgr->xcp[i].valid = false;
+
+	/* This is needed for figuring out memory id of xcp */
+	xcp_mgr->num_xcp_per_mem_partition = num_xcps / xcp_mgr->adev->gmc.num_mem_partitions;
+
+	for (i = 0; i < num_xcps; ++i) {
+		for (j = AMDGPU_XCP_GFXHUB; j < AMDGPU_XCP_MAX_BLOCKS; ++j) {
+			ret = xcp_mgr->funcs->get_ip_details(xcp_mgr, i, j,
+							     &ip);
+			if (ret)
+				continue;
+
+			__amdgpu_xcp_add_block(xcp_mgr, i, &ip);
+		}
+
+		xcp_mgr->xcp[i].id = i;
+
+		if (xcp_mgr->funcs->get_xcp_mem_id) {
+			ret = xcp_mgr->funcs->get_xcp_mem_id(
+				xcp_mgr, &xcp_mgr->xcp[i], &mem_id);
+			if (ret)
+				continue;
+			else
+				xcp_mgr->xcp[i].mem_id = mem_id;
+		}
+		__amdgpu_xcp_set_unique_id(xcp_mgr, i);
+	}
+
+	xcp_mgr->num_xcps = num_xcps;
+	amdgpu_xcp_update_partition_sched_list(adev);
+
+	return 0;
+}
+
+static int __amdgpu_xcp_switch_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr,
+					      int mode)
+{
+	int ret, curr_mode, num_xcps = 0;
+
+	if (!xcp_mgr->funcs || !xcp_mgr->funcs->switch_partition_mode)
+		return 0;
+
+	mutex_lock(&xcp_mgr->xcp_lock);
+
+	curr_mode = xcp_mgr->mode;
+	/* State set to transient mode */
+	xcp_mgr->mode = AMDGPU_XCP_MODE_TRANS;
+
+	ret = xcp_mgr->funcs->switch_partition_mode(xcp_mgr, mode, &num_xcps);
+
+	if (ret) {
+		/* Failed, get whatever mode it's at now */
+		if (xcp_mgr->funcs->query_partition_mode)
+			xcp_mgr->mode = amdgpu_xcp_query_partition_mode(
+				xcp_mgr, AMDGPU_XCP_FL_LOCKED);
+		else
+			xcp_mgr->mode = curr_mode;
+
+		goto out;
+	}
+	amdgpu_xcp_sysfs_entries_update(xcp_mgr);
+out:
+	mutex_unlock(&xcp_mgr->xcp_lock);
+
+	return ret;
+}
+
+int amdgpu_xcp_switch_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr, int mode)
+{
+	if (!xcp_mgr || mode == AMDGPU_XCP_MODE_NONE)
+		return -EINVAL;
+
+	if (xcp_mgr->mode == mode)
+		return 0;
+
+	return __amdgpu_xcp_switch_partition_mode(xcp_mgr, mode);
+}
+
+int amdgpu_xcp_restore_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr)
+{
+	if (!xcp_mgr || xcp_mgr->mode == AMDGPU_XCP_MODE_NONE)
+		return 0;
+
+	return __amdgpu_xcp_switch_partition_mode(xcp_mgr, xcp_mgr->mode);
+}
+
+static bool __amdgpu_xcp_is_cached_mode_valid(struct amdgpu_xcp_mgr *xcp_mgr)
+{
+	if (!xcp_mgr->funcs || !xcp_mgr->funcs->query_partition_mode)
+		return true;
+
+	if (!amdgpu_sriov_vf(xcp_mgr->adev) &&
+	    xcp_mgr->mode == AMDGPU_XCP_MODE_NONE)
+		return true;
+
+	if (xcp_mgr->mode != AMDGPU_XCP_MODE_NONE &&
+	    xcp_mgr->mode != AMDGPU_XCP_MODE_TRANS)
+		return true;
+
+	return false;
+}
+
+int amdgpu_xcp_query_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr, u32 flags)
+{
+	int mode;
+
+	if (__amdgpu_xcp_is_cached_mode_valid(xcp_mgr))
+		return xcp_mgr->mode;
+
+	if (!(flags & AMDGPU_XCP_FL_LOCKED))
+		mutex_lock(&xcp_mgr->xcp_lock);
+	mode = xcp_mgr->funcs->query_partition_mode(xcp_mgr);
+
+	/* First time query for VF, set the mode here */
+	if (amdgpu_sriov_vf(xcp_mgr->adev) &&
+	    xcp_mgr->mode == AMDGPU_XCP_MODE_NONE)
+		xcp_mgr->mode = mode;
+
+	if (xcp_mgr->mode != AMDGPU_XCP_MODE_TRANS && mode != xcp_mgr->mode)
+		dev_WARN(
+			xcp_mgr->adev->dev,
+			"Cached partition mode %d not matching with device mode %d",
+			xcp_mgr->mode, mode);
+
+	if (!(flags & AMDGPU_XCP_FL_LOCKED))
+		mutex_unlock(&xcp_mgr->xcp_lock);
+
+	return mode;
+}
+
+static int amdgpu_xcp_dev_alloc(struct amdgpu_device *adev)
+{
+	struct drm_device *p_ddev;
+	struct drm_device *ddev;
+	int i, ret;
+
+	ddev = adev_to_drm(adev);
+
+	/* xcp #0 shares drm device setting with adev */
+	adev->xcp_mgr->xcp->ddev = ddev;
+
+	for (i = 1; i < MAX_XCP; i++) {
+		ret = amdgpu_xcp_drm_dev_alloc(&p_ddev);
+		if (ret == -ENOSPC) {
+			dev_warn(adev->dev,
+			"Skip xcp node #%d when out of drm node resource.", i);
+			ret = 0;
+			goto out;
+		} else if (ret) {
+			goto out;
+		}
+
+		/* Redirect all IOCTLs to the primary device */
+		adev->xcp_mgr->xcp[i].rdev = p_ddev->render->dev;
+		adev->xcp_mgr->xcp[i].pdev = p_ddev->primary->dev;
+		adev->xcp_mgr->xcp[i].driver = (struct drm_driver *)p_ddev->driver;
+		adev->xcp_mgr->xcp[i].vma_offset_manager = p_ddev->vma_offset_manager;
+		p_ddev->render->dev = ddev;
+		p_ddev->primary->dev = ddev;
+		p_ddev->vma_offset_manager = ddev->vma_offset_manager;
+		p_ddev->driver = &amdgpu_partition_driver;
+		adev->xcp_mgr->xcp[i].ddev = p_ddev;
+
+		dev_set_drvdata(p_ddev->dev, &adev->xcp_mgr->xcp[i]);
+	}
+	ret = 0;
+out:
+	amdgpu_xcp_sysfs_entries_init(adev->xcp_mgr);
+
+	return ret;
+}
+
+int amdgpu_xcp_mgr_init(struct amdgpu_device *adev, int init_mode,
+			int init_num_xcps,
+			struct amdgpu_xcp_mgr_funcs *xcp_funcs)
+{
+	struct amdgpu_xcp_mgr *xcp_mgr;
+	int i;
+
+	if (!xcp_funcs || !xcp_funcs->get_ip_details)
+		return -EINVAL;
+
+	xcp_mgr = kzalloc(sizeof(*xcp_mgr), GFP_KERNEL);
+
+	if (!xcp_mgr)
+		return -ENOMEM;
+
+	xcp_mgr->adev = adev;
+	xcp_mgr->funcs = xcp_funcs;
+	xcp_mgr->mode = init_mode;
+	mutex_init(&xcp_mgr->xcp_lock);
+
+	if (init_mode != AMDGPU_XCP_MODE_NONE)
+		amdgpu_xcp_init(xcp_mgr, init_num_xcps, init_mode);
+
+	adev->xcp_mgr = xcp_mgr;
+	for (i = 0; i < MAX_XCP; ++i)
+		xcp_mgr->xcp[i].xcp_mgr = xcp_mgr;
+
+	return amdgpu_xcp_dev_alloc(adev);
+}
+
+int amdgpu_xcp_get_partition(struct amdgpu_xcp_mgr *xcp_mgr,
+			     enum AMDGPU_XCP_IP_BLOCK ip, int instance)
+{
+	struct amdgpu_xcp *xcp;
+	int i, id_mask = 0;
+
+	if (ip >= AMDGPU_XCP_MAX_BLOCKS)
+		return -EINVAL;
+
+	for (i = 0; i < xcp_mgr->num_xcps; ++i) {
+		xcp = &xcp_mgr->xcp[i];
+		if ((xcp->valid) && (xcp->ip[ip].valid) &&
+		    (xcp->ip[ip].inst_mask & BIT(instance)))
+			id_mask |= BIT(i);
+	}
+
+	if (!id_mask)
+		id_mask = -ENXIO;
+
+	return id_mask;
+}
+
+int amdgpu_xcp_get_inst_details(struct amdgpu_xcp *xcp,
+				enum AMDGPU_XCP_IP_BLOCK ip,
+				uint32_t *inst_mask)
+{
+	if (!xcp->valid || !inst_mask || !(xcp->ip[ip].valid))
+		return -EINVAL;
+
+	*inst_mask = xcp->ip[ip].inst_mask;
+
+	return 0;
+}
+
+int amdgpu_xcp_dev_register(struct amdgpu_device *adev,
+			const struct pci_device_id *ent)
+{
+	int i, ret;
+
+	if (!adev->xcp_mgr)
+		return 0;
+
+	for (i = 1; i < MAX_XCP; i++) {
+		if (!adev->xcp_mgr->xcp[i].ddev)
+			break;
+
+		ret = drm_dev_register(adev->xcp_mgr->xcp[i].ddev, ent->driver_data);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+void amdgpu_xcp_dev_unplug(struct amdgpu_device *adev)
+{
+	struct drm_device *p_ddev;
+	int i;
+
+	if (!adev->xcp_mgr)
+		return;
+
+	for (i = 1; i < MAX_XCP; i++) {
+		if (!adev->xcp_mgr->xcp[i].ddev)
+			break;
+
+		p_ddev = adev->xcp_mgr->xcp[i].ddev;
+		drm_dev_unplug(p_ddev);
+		p_ddev->render->dev = adev->xcp_mgr->xcp[i].rdev;
+		p_ddev->primary->dev = adev->xcp_mgr->xcp[i].pdev;
+		p_ddev->driver =  adev->xcp_mgr->xcp[i].driver;
+		p_ddev->vma_offset_manager = adev->xcp_mgr->xcp[i].vma_offset_manager;
+		amdgpu_xcp_drm_dev_free(p_ddev);
+	}
+}
+
+int amdgpu_xcp_open_device(struct amdgpu_device *adev,
+			   struct amdgpu_fpriv *fpriv,
+			   struct drm_file *file_priv)
+{
+	int i;
+
+	if (!adev->xcp_mgr)
+		return 0;
+
+	fpriv->xcp_id = AMDGPU_XCP_NO_PARTITION;
+	for (i = 0; i < MAX_XCP; ++i) {
+		if (!adev->xcp_mgr->xcp[i].ddev)
+			break;
+
+		if (file_priv->minor == adev->xcp_mgr->xcp[i].ddev->render) {
+			if (adev->xcp_mgr->xcp[i].valid == FALSE) {
+				dev_err(adev->dev, "renderD%d partition %d not valid!",
+						file_priv->minor->index, i);
+				return -ENOENT;
+			}
+			dev_dbg(adev->dev, "renderD%d partition %d opened!",
+					file_priv->minor->index, i);
+			fpriv->xcp_id = i;
+			break;
+		}
+	}
+
+	fpriv->vm.mem_id = fpriv->xcp_id == AMDGPU_XCP_NO_PARTITION ? -1 :
+				adev->xcp_mgr->xcp[fpriv->xcp_id].mem_id;
+	return 0;
+}
+
+void amdgpu_xcp_release_sched(struct amdgpu_device *adev,
+				  struct amdgpu_ctx_entity *entity)
+{
+	struct drm_gpu_scheduler *sched;
+	struct amdgpu_ring *ring;
+
+	if (!adev->xcp_mgr)
+		return;
+
+	sched = entity->entity.rq->sched;
+	if (drm_sched_wqueue_ready(sched)) {
+		ring = to_amdgpu_ring(entity->entity.rq->sched);
+		atomic_dec(&adev->xcp_mgr->xcp[ring->xcp_id].ref_cnt);
+	}
+}
+
+int amdgpu_xcp_select_scheds(struct amdgpu_device *adev,
+			     u32 hw_ip, u32 hw_prio,
+			     struct amdgpu_fpriv *fpriv,
+			     unsigned int *num_scheds,
+			     struct drm_gpu_scheduler ***scheds)
+{
+	u32 sel_xcp_id;
+	int i;
+	struct amdgpu_xcp_mgr *xcp_mgr = adev->xcp_mgr;
+
+	if (fpriv->xcp_id == AMDGPU_XCP_NO_PARTITION) {
+		u32 least_ref_cnt = ~0;
+
+		fpriv->xcp_id = 0;
+		for (i = 0; i < xcp_mgr->num_xcps; i++) {
+			u32 total_ref_cnt;
+
+			total_ref_cnt = atomic_read(&xcp_mgr->xcp[i].ref_cnt);
+			if (total_ref_cnt < least_ref_cnt) {
+				fpriv->xcp_id = i;
+				least_ref_cnt = total_ref_cnt;
+			}
+		}
+	}
+	sel_xcp_id = fpriv->xcp_id;
+
+	if (xcp_mgr->xcp[sel_xcp_id].gpu_sched[hw_ip][hw_prio].num_scheds) {
+		*num_scheds =
+			xcp_mgr->xcp[fpriv->xcp_id].gpu_sched[hw_ip][hw_prio].num_scheds;
+		*scheds =
+			xcp_mgr->xcp[fpriv->xcp_id].gpu_sched[hw_ip][hw_prio].sched;
+		atomic_inc(&adev->xcp_mgr->xcp[sel_xcp_id].ref_cnt);
+		dev_dbg(adev->dev, "Selected partition #%d", sel_xcp_id);
+	} else {
+		dev_err(adev->dev, "Failed to schedule partition #%d.", sel_xcp_id);
+		return -ENOENT;
+	}
+
+	return 0;
+}
+
+static void amdgpu_set_xcp_id(struct amdgpu_device *adev,
+			      uint32_t inst_idx,
+			      struct amdgpu_ring *ring)
+{
+	int xcp_id;
+	enum AMDGPU_XCP_IP_BLOCK ip_blk;
+	uint32_t inst_mask;
+
+	ring->xcp_id = AMDGPU_XCP_NO_PARTITION;
+	if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE)
+		adev->gfx.enforce_isolation[0].xcp_id = ring->xcp_id;
+	if ((adev->xcp_mgr->mode == AMDGPU_XCP_MODE_NONE) ||
+	    (ring->funcs->type == AMDGPU_RING_TYPE_CPER))
+		return;
+
+	inst_mask = 1 << inst_idx;
+
+	switch (ring->funcs->type) {
+	case AMDGPU_HW_IP_GFX:
+	case AMDGPU_RING_TYPE_COMPUTE:
+	case AMDGPU_RING_TYPE_KIQ:
+		ip_blk = AMDGPU_XCP_GFX;
+		break;
+	case AMDGPU_RING_TYPE_SDMA:
+		ip_blk = AMDGPU_XCP_SDMA;
+		break;
+	case AMDGPU_RING_TYPE_VCN_ENC:
+	case AMDGPU_RING_TYPE_VCN_JPEG:
+		ip_blk = AMDGPU_XCP_VCN;
+		break;
+	default:
+		dev_err(adev->dev, "Not support ring type %d!", ring->funcs->type);
+		return;
+	}
+
+	for (xcp_id = 0; xcp_id < adev->xcp_mgr->num_xcps; xcp_id++) {
+		if (adev->xcp_mgr->xcp[xcp_id].ip[ip_blk].inst_mask & inst_mask) {
+			ring->xcp_id = xcp_id;
+			dev_dbg(adev->dev, "ring:%s xcp_id :%u", ring->name,
+				ring->xcp_id);
+			if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE)
+				adev->gfx.enforce_isolation[xcp_id].xcp_id = xcp_id;
+			break;
+		}
+	}
+}
+
+static void amdgpu_xcp_gpu_sched_update(struct amdgpu_device *adev,
+					struct amdgpu_ring *ring,
+					unsigned int sel_xcp_id)
+{
+	unsigned int *num_gpu_sched;
+
+	num_gpu_sched = &adev->xcp_mgr->xcp[sel_xcp_id]
+			.gpu_sched[ring->funcs->type][ring->hw_prio].num_scheds;
+	adev->xcp_mgr->xcp[sel_xcp_id].gpu_sched[ring->funcs->type][ring->hw_prio]
+			.sched[(*num_gpu_sched)++] = &ring->sched;
+	dev_dbg(adev->dev, "%s :[%d] gpu_sched[%d][%d] = %d",
+		ring->name, sel_xcp_id, ring->funcs->type,
+		ring->hw_prio, *num_gpu_sched);
+}
+
+static int amdgpu_xcp_sched_list_update(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	int i;
+
+	for (i = 0; i < MAX_XCP; i++) {
+		atomic_set(&adev->xcp_mgr->xcp[i].ref_cnt, 0);
+		memset(adev->xcp_mgr->xcp[i].gpu_sched, 0, sizeof(adev->xcp_mgr->xcp->gpu_sched));
+	}
+
+	if (adev->xcp_mgr->mode == AMDGPU_XCP_MODE_NONE)
+		return 0;
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
+		ring = adev->rings[i];
+		if (!ring || !ring->sched.ready || ring->no_scheduler)
+			continue;
+
+		amdgpu_xcp_gpu_sched_update(adev, ring, ring->xcp_id);
+
+		/* VCN may be shared by two partitions under CPX MODE in certain
+		 * configs.
+		 */
+		if ((ring->funcs->type == AMDGPU_RING_TYPE_VCN_ENC ||
+		     ring->funcs->type == AMDGPU_RING_TYPE_VCN_JPEG) &&
+		    (adev->xcp_mgr->num_xcps > adev->vcn.num_vcn_inst))
+			amdgpu_xcp_gpu_sched_update(adev, ring, ring->xcp_id + 1);
+	}
+
+	return 0;
+}
+
+int amdgpu_xcp_update_partition_sched_list(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->num_rings; i++) {
+		struct amdgpu_ring *ring = adev->rings[i];
+
+		if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE ||
+			ring->funcs->type == AMDGPU_RING_TYPE_KIQ)
+			amdgpu_set_xcp_id(adev, ring->xcc_id, ring);
+		else
+			amdgpu_set_xcp_id(adev, ring->me, ring);
+	}
+
+	return amdgpu_xcp_sched_list_update(adev);
+}
+
+void amdgpu_xcp_update_supported_modes(struct amdgpu_xcp_mgr *xcp_mgr)
+{
+	struct amdgpu_device *adev = xcp_mgr->adev;
+
+	xcp_mgr->supp_xcp_modes = 0;
+
+	switch (NUM_XCC(adev->gfx.xcc_mask)) {
+	case 8:
+		xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) |
+					  BIT(AMDGPU_DPX_PARTITION_MODE) |
+					  BIT(AMDGPU_QPX_PARTITION_MODE) |
+					  BIT(AMDGPU_CPX_PARTITION_MODE);
+		break;
+	case 6:
+		xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) |
+					  BIT(AMDGPU_TPX_PARTITION_MODE) |
+					  BIT(AMDGPU_CPX_PARTITION_MODE);
+		break;
+	case 4:
+		xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) |
+					  BIT(AMDGPU_DPX_PARTITION_MODE) |
+					  BIT(AMDGPU_CPX_PARTITION_MODE);
+		break;
+	case 2:
+		xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) |
+					  BIT(AMDGPU_CPX_PARTITION_MODE);
+		break;
+	case 1:
+		xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) |
+					  BIT(AMDGPU_CPX_PARTITION_MODE);
+		break;
+
+	default:
+		break;
+	}
+}
+
+int amdgpu_xcp_pre_partition_switch(struct amdgpu_xcp_mgr *xcp_mgr, u32 flags)
+{
+	/* TODO:
+	 * Stop user queues and threads, and make sure GPU is empty of work.
+	 */
+
+	if (flags & AMDGPU_XCP_OPS_KFD)
+		amdgpu_amdkfd_device_fini_sw(xcp_mgr->adev);
+
+	return 0;
+}
+
+int amdgpu_xcp_post_partition_switch(struct amdgpu_xcp_mgr *xcp_mgr, u32 flags)
+{
+	int ret = 0;
+
+	if (flags & AMDGPU_XCP_OPS_KFD) {
+		amdgpu_amdkfd_device_probe(xcp_mgr->adev);
+		amdgpu_amdkfd_device_init(xcp_mgr->adev);
+		/* If KFD init failed, return failure */
+		if (!xcp_mgr->adev->kfd.init_complete)
+			ret = -EIO;
+	}
+
+	return ret;
+}
+
+/*====================== xcp sysfs - configuration ======================*/
+#define XCP_CFG_SYSFS_RES_ATTR_SHOW(_name)                         \
+	static ssize_t amdgpu_xcp_res_sysfs_##_name##_show(        \
+		struct amdgpu_xcp_res_details *xcp_res, char *buf) \
+	{                                                          \
+		return sysfs_emit(buf, "%d\n", xcp_res->_name);    \
+	}
+
+struct amdgpu_xcp_res_sysfs_attribute {
+	struct attribute attr;
+	ssize_t (*show)(struct amdgpu_xcp_res_details *xcp_res, char *buf);
+};
+
+#define XCP_CFG_SYSFS_RES_ATTR(_name)                                        \
+	struct amdgpu_xcp_res_sysfs_attribute xcp_res_sysfs_attr_##_name = { \
+		.attr = { .name = __stringify(_name), .mode = 0400 },        \
+		.show = amdgpu_xcp_res_sysfs_##_name##_show,                 \
+	}
+
+XCP_CFG_SYSFS_RES_ATTR_SHOW(num_inst)
+XCP_CFG_SYSFS_RES_ATTR(num_inst);
+XCP_CFG_SYSFS_RES_ATTR_SHOW(num_shared)
+XCP_CFG_SYSFS_RES_ATTR(num_shared);
+
+#define XCP_CFG_SYSFS_RES_ATTR_PTR(_name) xcp_res_sysfs_attr_##_name.attr
+
+static struct attribute *xcp_cfg_res_sysfs_attrs[] = {
+	&XCP_CFG_SYSFS_RES_ATTR_PTR(num_inst),
+	&XCP_CFG_SYSFS_RES_ATTR_PTR(num_shared), NULL
+};
+
+static const char *xcp_desc[] = {
+	[AMDGPU_SPX_PARTITION_MODE] = "SPX",
+	[AMDGPU_DPX_PARTITION_MODE] = "DPX",
+	[AMDGPU_TPX_PARTITION_MODE] = "TPX",
+	[AMDGPU_QPX_PARTITION_MODE] = "QPX",
+	[AMDGPU_CPX_PARTITION_MODE] = "CPX",
+};
+
+static const char *nps_desc[] = {
+	[UNKNOWN_MEMORY_PARTITION_MODE] = "UNKNOWN",
+	[AMDGPU_NPS1_PARTITION_MODE] = "NPS1",
+	[AMDGPU_NPS2_PARTITION_MODE] = "NPS2",
+	[AMDGPU_NPS3_PARTITION_MODE] = "NPS3",
+	[AMDGPU_NPS4_PARTITION_MODE] = "NPS4",
+	[AMDGPU_NPS6_PARTITION_MODE] = "NPS6",
+	[AMDGPU_NPS8_PARTITION_MODE] = "NPS8",
+};
+
+ATTRIBUTE_GROUPS(xcp_cfg_res_sysfs);
+
+#define to_xcp_attr(x) \
+	container_of(x, struct amdgpu_xcp_res_sysfs_attribute, attr)
+#define to_xcp_res(x) container_of(x, struct amdgpu_xcp_res_details, kobj)
+
+static ssize_t xcp_cfg_res_sysfs_attr_show(struct kobject *kobj,
+					   struct attribute *attr, char *buf)
+{
+	struct amdgpu_xcp_res_sysfs_attribute *attribute;
+	struct amdgpu_xcp_res_details *xcp_res;
+
+	attribute = to_xcp_attr(attr);
+	xcp_res = to_xcp_res(kobj);
+
+	if (!attribute->show)
+		return -EIO;
+
+	return attribute->show(xcp_res, buf);
+}
+
+static const struct sysfs_ops xcp_cfg_res_sysfs_ops = {
+	.show = xcp_cfg_res_sysfs_attr_show,
+};
+
+static const struct kobj_type xcp_cfg_res_sysfs_ktype = {
+	.sysfs_ops = &xcp_cfg_res_sysfs_ops,
+	.default_groups = xcp_cfg_res_sysfs_groups,
+};
+
+const char *xcp_res_names[] = {
+	[AMDGPU_XCP_RES_XCC] = "xcc",
+	[AMDGPU_XCP_RES_DMA] = "dma",
+	[AMDGPU_XCP_RES_DEC] = "dec",
+	[AMDGPU_XCP_RES_JPEG] = "jpeg",
+};
+
+static int amdgpu_xcp_get_res_info(struct amdgpu_xcp_mgr *xcp_mgr,
+				   int mode,
+				   struct amdgpu_xcp_cfg *xcp_cfg)
+{
+	if (xcp_mgr->funcs && xcp_mgr->funcs->get_xcp_res_info)
+		return xcp_mgr->funcs->get_xcp_res_info(xcp_mgr, mode, xcp_cfg);
+
+	return -EOPNOTSUPP;
+}
+
+#define to_xcp_cfg(x) container_of(x, struct amdgpu_xcp_cfg, kobj)
+static ssize_t supported_xcp_configs_show(struct kobject *kobj,
+					  struct kobj_attribute *attr, char *buf)
+{
+	struct amdgpu_xcp_cfg *xcp_cfg = to_xcp_cfg(kobj);
+	struct amdgpu_xcp_mgr *xcp_mgr = xcp_cfg->xcp_mgr;
+	int size = 0, mode;
+	char *sep = "";
+
+	if (!xcp_mgr || !xcp_mgr->supp_xcp_modes)
+		return sysfs_emit(buf, "Not supported\n");
+
+	for_each_inst(mode, xcp_mgr->supp_xcp_modes) {
+		size += sysfs_emit_at(buf, size, "%s%s", sep, xcp_desc[mode]);
+		sep = ", ";
+	}
+
+	size += sysfs_emit_at(buf, size, "\n");
+
+	return size;
+}
+
+static ssize_t supported_nps_configs_show(struct kobject *kobj,
+					  struct kobj_attribute *attr, char *buf)
+{
+	struct amdgpu_xcp_cfg *xcp_cfg = to_xcp_cfg(kobj);
+	int size = 0, mode;
+	char *sep = "";
+
+	if (!xcp_cfg || !xcp_cfg->compatible_nps_modes)
+		return sysfs_emit(buf, "Not supported\n");
+
+	for_each_inst(mode, xcp_cfg->compatible_nps_modes) {
+		size += sysfs_emit_at(buf, size, "%s%s", sep, nps_desc[mode]);
+		sep = ", ";
+	}
+
+	size += sysfs_emit_at(buf, size, "\n");
+
+	return size;
+}
+
+static ssize_t xcp_config_show(struct kobject *kobj,
+			       struct kobj_attribute *attr, char *buf)
+{
+	struct amdgpu_xcp_cfg *xcp_cfg = to_xcp_cfg(kobj);
+
+	return sysfs_emit(buf, "%s\n",
+			  amdgpu_gfx_compute_mode_desc(xcp_cfg->mode));
+}
+
+static ssize_t xcp_config_store(struct kobject *kobj,
+				struct kobj_attribute *attr,
+				const char *buf, size_t size)
+{
+	struct amdgpu_xcp_cfg *xcp_cfg = to_xcp_cfg(kobj);
+	int mode, r;
+
+	if (!strncasecmp("SPX", buf, strlen("SPX")))
+		mode = AMDGPU_SPX_PARTITION_MODE;
+	else if (!strncasecmp("DPX", buf, strlen("DPX")))
+		mode = AMDGPU_DPX_PARTITION_MODE;
+	else if (!strncasecmp("TPX", buf, strlen("TPX")))
+		mode = AMDGPU_TPX_PARTITION_MODE;
+	else if (!strncasecmp("QPX", buf, strlen("QPX")))
+		mode = AMDGPU_QPX_PARTITION_MODE;
+	else if (!strncasecmp("CPX", buf, strlen("CPX")))
+		mode = AMDGPU_CPX_PARTITION_MODE;
+	else
+		return -EINVAL;
+
+	r = amdgpu_xcp_get_res_info(xcp_cfg->xcp_mgr, mode, xcp_cfg);
+
+	if (r)
+		return r;
+
+	xcp_cfg->mode = mode;
+	return size;
+}
+
+static struct kobj_attribute xcp_cfg_sysfs_mode =
+	__ATTR_RW_MODE(xcp_config, 0644);
+
+static void xcp_cfg_sysfs_release(struct kobject *kobj)
+{
+	struct amdgpu_xcp_cfg *xcp_cfg = to_xcp_cfg(kobj);
+
+	kfree(xcp_cfg);
+}
+
+static const struct kobj_type xcp_cfg_sysfs_ktype = {
+	.release = xcp_cfg_sysfs_release,
+	.sysfs_ops = &kobj_sysfs_ops,
+};
+
+static struct kobj_attribute supp_part_sysfs_mode =
+	__ATTR_RO(supported_xcp_configs);
+
+static struct kobj_attribute supp_nps_sysfs_mode =
+	__ATTR_RO(supported_nps_configs);
+
+static const struct attribute *xcp_attrs[] = {
+	&supp_part_sysfs_mode.attr,
+	&xcp_cfg_sysfs_mode.attr,
+	NULL,
+};
+
+static void amdgpu_xcp_cfg_sysfs_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_xcp_res_details *xcp_res;
+	struct amdgpu_xcp_cfg *xcp_cfg;
+	int i, r, j, rid, mode;
+
+	if (!adev->xcp_mgr)
+		return;
+
+	xcp_cfg = kzalloc(sizeof(*xcp_cfg), GFP_KERNEL);
+	if (!xcp_cfg)
+		return;
+	xcp_cfg->xcp_mgr = adev->xcp_mgr;
+
+	r = kobject_init_and_add(&xcp_cfg->kobj, &xcp_cfg_sysfs_ktype,
+				 &adev->dev->kobj, "compute_partition_config");
+	if (r)
+		goto err1;
+
+	r = sysfs_create_files(&xcp_cfg->kobj, xcp_attrs);
+	if (r)
+		goto err1;
+
+	if (adev->gmc.supported_nps_modes != 0) {
+		r = sysfs_create_file(&xcp_cfg->kobj, &supp_nps_sysfs_mode.attr);
+		if (r) {
+			sysfs_remove_files(&xcp_cfg->kobj, xcp_attrs);
+			goto err1;
+		}
+	}
+
+	mode = (xcp_cfg->xcp_mgr->mode ==
+		AMDGPU_UNKNOWN_COMPUTE_PARTITION_MODE) ?
+		       AMDGPU_SPX_PARTITION_MODE :
+		       xcp_cfg->xcp_mgr->mode;
+	r = amdgpu_xcp_get_res_info(xcp_cfg->xcp_mgr, mode, xcp_cfg);
+	if (r) {
+		sysfs_remove_file(&xcp_cfg->kobj, &supp_nps_sysfs_mode.attr);
+		sysfs_remove_files(&xcp_cfg->kobj, xcp_attrs);
+		goto err1;
+	}
+
+	xcp_cfg->mode = mode;
+	for (i = 0; i < xcp_cfg->num_res; i++) {
+		xcp_res = &xcp_cfg->xcp_res[i];
+		rid = xcp_res->id;
+		r = kobject_init_and_add(&xcp_res->kobj,
+					 &xcp_cfg_res_sysfs_ktype,
+					 &xcp_cfg->kobj, "%s",
+					 xcp_res_names[rid]);
+		if (r)
+			goto err;
+	}
+
+	adev->xcp_mgr->xcp_cfg = xcp_cfg;
+	return;
+err:
+	for (j = 0; j < i; j++) {
+		xcp_res = &xcp_cfg->xcp_res[i];
+		kobject_put(&xcp_res->kobj);
+	}
+
+	sysfs_remove_file(&xcp_cfg->kobj, &supp_nps_sysfs_mode.attr);
+	sysfs_remove_files(&xcp_cfg->kobj, xcp_attrs);
+err1:
+	kobject_put(&xcp_cfg->kobj);
+}
+
+static void amdgpu_xcp_cfg_sysfs_fini(struct amdgpu_device *adev)
+{
+	struct amdgpu_xcp_res_details *xcp_res;
+	struct amdgpu_xcp_cfg *xcp_cfg;
+	int i;
+
+	if (!adev->xcp_mgr || !adev->xcp_mgr->xcp_cfg)
+		return;
+
+	xcp_cfg = adev->xcp_mgr->xcp_cfg;
+	for (i = 0; i < xcp_cfg->num_res; i++) {
+		xcp_res = &xcp_cfg->xcp_res[i];
+		kobject_put(&xcp_res->kobj);
+	}
+
+	sysfs_remove_file(&xcp_cfg->kobj, &supp_nps_sysfs_mode.attr);
+	sysfs_remove_files(&xcp_cfg->kobj, xcp_attrs);
+	kobject_put(&xcp_cfg->kobj);
+}
+
+/*====================== xcp sysfs - data entries ======================*/
+
+#define to_xcp(x) container_of(x, struct amdgpu_xcp, kobj)
+
+static ssize_t xcp_metrics_show(struct kobject *kobj,
+				struct kobj_attribute *attr, char *buf)
+{
+	struct amdgpu_xcp *xcp = to_xcp(kobj);
+	struct amdgpu_xcp_mgr *xcp_mgr;
+	ssize_t size;
+
+	xcp_mgr = xcp->xcp_mgr;
+	size = amdgpu_dpm_get_xcp_metrics(xcp_mgr->adev, xcp->id, NULL);
+	if (size <= 0)
+		return size;
+
+	if (size > PAGE_SIZE)
+		return -ENOSPC;
+
+	return amdgpu_dpm_get_xcp_metrics(xcp_mgr->adev, xcp->id, buf);
+}
+
+static umode_t amdgpu_xcp_attrs_is_visible(struct kobject *kobj,
+					   struct attribute *attr, int n)
+{
+	struct amdgpu_xcp *xcp = to_xcp(kobj);
+
+	if (!xcp || !xcp->valid)
+		return 0;
+
+	return attr->mode;
+}
+
+static struct kobj_attribute xcp_sysfs_metrics = __ATTR_RO(xcp_metrics);
+
+static struct attribute *amdgpu_xcp_attrs[] = {
+	&xcp_sysfs_metrics.attr,
+	NULL,
+};
+
+static const struct attribute_group amdgpu_xcp_attrs_group = {
+	.attrs = amdgpu_xcp_attrs,
+	.is_visible = amdgpu_xcp_attrs_is_visible
+};
+
+static const struct kobj_type xcp_sysfs_ktype = {
+	.sysfs_ops = &kobj_sysfs_ops,
+};
+
+static void amdgpu_xcp_sysfs_entries_fini(struct amdgpu_xcp_mgr *xcp_mgr, int n)
+{
+	struct amdgpu_xcp *xcp;
+
+	for (n--; n >= 0; n--) {
+		xcp = &xcp_mgr->xcp[n];
+		if (!xcp->ddev || !xcp->valid)
+			continue;
+		sysfs_remove_group(&xcp->kobj, &amdgpu_xcp_attrs_group);
+		kobject_put(&xcp->kobj);
+	}
+}
+
+static void amdgpu_xcp_sysfs_entries_init(struct amdgpu_xcp_mgr *xcp_mgr)
+{
+	struct amdgpu_xcp *xcp;
+	int i, r;
+
+	for (i = 0; i < MAX_XCP; i++) {
+		/* Redirect all IOCTLs to the primary device */
+		xcp = &xcp_mgr->xcp[i];
+		if (!xcp->ddev)
+			break;
+		r = kobject_init_and_add(&xcp->kobj, &xcp_sysfs_ktype,
+					 &xcp->ddev->dev->kobj, "xcp");
+		if (r)
+			goto out;
+
+		r = sysfs_create_group(&xcp->kobj, &amdgpu_xcp_attrs_group);
+		if (r)
+			goto out;
+	}
+
+	return;
+out:
+	kobject_put(&xcp->kobj);
+}
+
+static void amdgpu_xcp_sysfs_entries_update(struct amdgpu_xcp_mgr *xcp_mgr)
+{
+	struct amdgpu_xcp *xcp;
+	int i;
+
+	for (i = 0; i < MAX_XCP; i++) {
+		/* Redirect all IOCTLs to the primary device */
+		xcp = &xcp_mgr->xcp[i];
+		if (!xcp->ddev)
+			continue;
+		sysfs_update_group(&xcp->kobj, &amdgpu_xcp_attrs_group);
+	}
+
+	return;
+}
+
+void amdgpu_xcp_sysfs_init(struct amdgpu_device *adev)
+{
+	if (!adev->xcp_mgr)
+		return;
+
+	amdgpu_xcp_cfg_sysfs_init(adev);
+
+	return;
+}
+
+void amdgpu_xcp_sysfs_fini(struct amdgpu_device *adev)
+{
+	if (!adev->xcp_mgr)
+		return;
+	amdgpu_xcp_sysfs_entries_fini(adev->xcp_mgr, MAX_XCP);
+	amdgpu_xcp_cfg_sysfs_fini(adev);
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h
new file mode 100644
index 000000000000..1928d9e224fc
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h
@@ -0,0 +1,217 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef AMDGPU_XCP_H
+#define AMDGPU_XCP_H
+
+#include <linux/pci.h>
+#include <linux/xarray.h>
+
+#include "amdgpu_ctx.h"
+
+#define MAX_XCP 8
+
+#define AMDGPU_XCP_MODE_NONE -1
+#define AMDGPU_XCP_MODE_TRANS -2
+
+#define AMDGPU_XCP_FL_NONE 0
+#define AMDGPU_XCP_FL_LOCKED (1 << 0)
+
+#define AMDGPU_XCP_NO_PARTITION (~0)
+
+#define AMDGPU_XCP_OPS_KFD	(1 << 0)
+
+struct amdgpu_fpriv;
+
+enum AMDGPU_XCP_IP_BLOCK {
+	AMDGPU_XCP_GFXHUB,
+	AMDGPU_XCP_GFX,
+	AMDGPU_XCP_SDMA,
+	AMDGPU_XCP_VCN,
+	AMDGPU_XCP_MAX_BLOCKS
+};
+
+enum AMDGPU_XCP_STATE {
+	AMDGPU_XCP_PREPARE_SUSPEND,
+	AMDGPU_XCP_SUSPEND,
+	AMDGPU_XCP_PREPARE_RESUME,
+	AMDGPU_XCP_RESUME,
+};
+
+enum amdgpu_xcp_res_id {
+	AMDGPU_XCP_RES_XCC,
+	AMDGPU_XCP_RES_DMA,
+	AMDGPU_XCP_RES_DEC,
+	AMDGPU_XCP_RES_JPEG,
+	AMDGPU_XCP_RES_MAX,
+};
+
+struct amdgpu_xcp_res_details {
+	enum amdgpu_xcp_res_id id;
+	u8 num_inst;
+	u8 num_shared;
+	struct kobject kobj;
+};
+
+struct amdgpu_xcp_cfg {
+	u8 mode;
+	struct amdgpu_xcp_res_details xcp_res[AMDGPU_XCP_RES_MAX];
+	u8 num_res;
+	struct amdgpu_xcp_mgr *xcp_mgr;
+	struct kobject kobj;
+	u16 compatible_nps_modes;
+};
+
+struct amdgpu_xcp_ip_funcs {
+	int (*prepare_suspend)(void *handle, uint32_t inst_mask);
+	int (*suspend)(void *handle, uint32_t inst_mask);
+	int (*prepare_resume)(void *handle, uint32_t inst_mask);
+	int (*resume)(void *handle, uint32_t inst_mask);
+};
+
+struct amdgpu_xcp_ip {
+	struct amdgpu_xcp_ip_funcs *ip_funcs;
+	uint32_t inst_mask;
+
+	enum AMDGPU_XCP_IP_BLOCK ip_id;
+	bool valid;
+};
+
+struct amdgpu_xcp {
+	struct amdgpu_xcp_ip ip[AMDGPU_XCP_MAX_BLOCKS];
+
+	uint8_t id;
+	uint8_t mem_id;
+	bool valid;
+	atomic_t	ref_cnt;
+	struct drm_device *ddev;
+	struct drm_device *rdev;
+	struct drm_device *pdev;
+	struct drm_driver *driver;
+	struct drm_vma_offset_manager *vma_offset_manager;
+	struct amdgpu_sched	gpu_sched[AMDGPU_HW_IP_NUM][AMDGPU_RING_PRIO_MAX];
+	struct amdgpu_xcp_mgr *xcp_mgr;
+	struct kobject kobj;
+	uint64_t unique_id;
+};
+
+struct amdgpu_xcp_mgr {
+	struct amdgpu_device *adev;
+	struct mutex xcp_lock;
+	struct amdgpu_xcp_mgr_funcs *funcs;
+
+	struct amdgpu_xcp xcp[MAX_XCP];
+	uint8_t num_xcps;
+	int8_t mode;
+
+	 /* Used to determine KFD memory size limits per XCP */
+	unsigned int num_xcp_per_mem_partition;
+	struct amdgpu_xcp_cfg *xcp_cfg;
+	uint32_t supp_xcp_modes;
+	uint32_t avail_xcp_modes;
+};
+
+struct amdgpu_xcp_mgr_funcs {
+	int (*switch_partition_mode)(struct amdgpu_xcp_mgr *xcp_mgr, int mode,
+				     int *num_xcps);
+	int (*query_partition_mode)(struct amdgpu_xcp_mgr *xcp_mgr);
+	int (*get_ip_details)(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id,
+			      enum AMDGPU_XCP_IP_BLOCK ip_id,
+			      struct amdgpu_xcp_ip *ip);
+	int (*get_xcp_mem_id)(struct amdgpu_xcp_mgr *xcp_mgr,
+			      struct amdgpu_xcp *xcp, uint8_t *mem_id);
+	int (*get_xcp_res_info)(struct amdgpu_xcp_mgr *xcp_mgr,
+				int mode,
+				struct amdgpu_xcp_cfg *xcp_cfg);
+	int (*prepare_suspend)(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id);
+	int (*suspend)(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id);
+	int (*prepare_resume)(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id);
+	int (*resume)(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id);
+};
+
+int amdgpu_xcp_prepare_suspend(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id);
+int amdgpu_xcp_suspend(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id);
+int amdgpu_xcp_prepare_resume(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id);
+int amdgpu_xcp_resume(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id);
+
+int amdgpu_xcp_mgr_init(struct amdgpu_device *adev, int init_mode,
+			int init_xcps, struct amdgpu_xcp_mgr_funcs *xcp_funcs);
+int amdgpu_xcp_init(struct amdgpu_xcp_mgr *xcp_mgr, int num_xcps, int mode);
+int amdgpu_xcp_query_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr, u32 flags);
+int amdgpu_xcp_switch_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr, int mode);
+int amdgpu_xcp_restore_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr);
+int amdgpu_xcp_get_partition(struct amdgpu_xcp_mgr *xcp_mgr,
+			     enum AMDGPU_XCP_IP_BLOCK ip, int instance);
+
+int amdgpu_xcp_get_inst_details(struct amdgpu_xcp *xcp,
+				enum AMDGPU_XCP_IP_BLOCK ip,
+				uint32_t *inst_mask);
+
+int amdgpu_xcp_dev_register(struct amdgpu_device *adev,
+				const struct pci_device_id *ent);
+void amdgpu_xcp_dev_unplug(struct amdgpu_device *adev);
+int amdgpu_xcp_open_device(struct amdgpu_device *adev,
+			   struct amdgpu_fpriv *fpriv,
+			   struct drm_file *file_priv);
+void amdgpu_xcp_release_sched(struct amdgpu_device *adev,
+			      struct amdgpu_ctx_entity *entity);
+int amdgpu_xcp_select_scheds(struct amdgpu_device *adev,
+			     u32 hw_ip, u32 hw_prio,
+			     struct amdgpu_fpriv *fpriv,
+			     unsigned int *num_scheds,
+			     struct drm_gpu_scheduler ***scheds);
+void amdgpu_xcp_update_supported_modes(struct amdgpu_xcp_mgr *xcp_mgr);
+int amdgpu_xcp_update_partition_sched_list(struct amdgpu_device *adev);
+int amdgpu_xcp_pre_partition_switch(struct amdgpu_xcp_mgr *xcp_mgr, u32 flags);
+int amdgpu_xcp_post_partition_switch(struct amdgpu_xcp_mgr *xcp_mgr, u32 flags);
+void amdgpu_xcp_sysfs_init(struct amdgpu_device *adev);
+void amdgpu_xcp_sysfs_fini(struct amdgpu_device *adev);
+
+static inline int amdgpu_xcp_get_num_xcp(struct amdgpu_xcp_mgr *xcp_mgr)
+{
+	if (!xcp_mgr)
+		return 1;
+	else
+		return xcp_mgr->num_xcps;
+}
+
+static inline struct amdgpu_xcp *
+amdgpu_get_next_xcp(struct amdgpu_xcp_mgr *xcp_mgr, int *from)
+{
+	if (!xcp_mgr)
+		return NULL;
+
+	while (*from < MAX_XCP) {
+		if (xcp_mgr->xcp[*from].valid)
+			return &xcp_mgr->xcp[*from];
+		++(*from);
+	}
+
+	return NULL;
+}
+
+#define for_each_xcp(xcp_mgr, xcp, i)                            \
+	for (i = 0, xcp = amdgpu_get_next_xcp(xcp_mgr, &i); xcp; \
+	     ++i, xcp = amdgpu_get_next_xcp(xcp_mgr, &i))
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
new file mode 100644
index 000000000000..1ede308a7c67
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
@@ -0,0 +1,1795 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ *
+ */
+#include <linux/list.h>
+#include "amdgpu.h"
+#include "amdgpu_xgmi.h"
+#include "amdgpu_ras.h"
+#include "soc15.h"
+#include "df/df_3_6_offset.h"
+#include "xgmi/xgmi_4_0_0_smn.h"
+#include "xgmi/xgmi_4_0_0_sh_mask.h"
+#include "xgmi/xgmi_6_1_0_sh_mask.h"
+#include "wafl/wafl2_4_0_0_smn.h"
+#include "wafl/wafl2_4_0_0_sh_mask.h"
+
+#include "amdgpu_reset.h"
+
+#define smnPCS_XGMI3X16_PCS_ERROR_STATUS 0x11a0020c
+#define smnPCS_XGMI3X16_PCS_ERROR_NONCORRECTABLE_MASK   0x11a00218
+#define smnPCS_GOPX1_PCS_ERROR_STATUS    0x12200210
+#define smnPCS_GOPX1_PCS_ERROR_NONCORRECTABLE_MASK      0x12200218
+
+#define XGMI_STATE_DISABLE                      0xD1
+#define XGMI_STATE_LS0                          0x81
+#define XGMI_LINK_ACTIVE			1
+#define XGMI_LINK_INACTIVE			0
+
+static DEFINE_MUTEX(xgmi_mutex);
+
+#define AMDGPU_MAX_XGMI_DEVICE_PER_HIVE		4
+
+static LIST_HEAD(xgmi_hive_list);
+
+static const int xgmi_pcs_err_status_reg_vg20[] = {
+	smnXGMI0_PCS_GOPX16_PCS_ERROR_STATUS,
+	smnXGMI0_PCS_GOPX16_PCS_ERROR_STATUS + 0x100000,
+};
+
+static const int wafl_pcs_err_status_reg_vg20[] = {
+	smnPCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS,
+	smnPCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS + 0x100000,
+};
+
+static const int xgmi_pcs_err_status_reg_arct[] = {
+	smnXGMI0_PCS_GOPX16_PCS_ERROR_STATUS,
+	smnXGMI0_PCS_GOPX16_PCS_ERROR_STATUS + 0x100000,
+	smnXGMI0_PCS_GOPX16_PCS_ERROR_STATUS + 0x500000,
+	smnXGMI0_PCS_GOPX16_PCS_ERROR_STATUS + 0x600000,
+	smnXGMI0_PCS_GOPX16_PCS_ERROR_STATUS + 0x700000,
+	smnXGMI0_PCS_GOPX16_PCS_ERROR_STATUS + 0x800000,
+};
+
+/* same as vg20*/
+static const int wafl_pcs_err_status_reg_arct[] = {
+	smnPCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS,
+	smnPCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS + 0x100000,
+};
+
+static const int xgmi3x16_pcs_err_status_reg_aldebaran[] = {
+	smnPCS_XGMI3X16_PCS_ERROR_STATUS,
+	smnPCS_XGMI3X16_PCS_ERROR_STATUS + 0x100000,
+	smnPCS_XGMI3X16_PCS_ERROR_STATUS + 0x200000,
+	smnPCS_XGMI3X16_PCS_ERROR_STATUS + 0x300000,
+	smnPCS_XGMI3X16_PCS_ERROR_STATUS + 0x400000,
+	smnPCS_XGMI3X16_PCS_ERROR_STATUS + 0x500000,
+	smnPCS_XGMI3X16_PCS_ERROR_STATUS + 0x600000,
+	smnPCS_XGMI3X16_PCS_ERROR_STATUS + 0x700000
+};
+
+static const int xgmi3x16_pcs_err_noncorrectable_mask_reg_aldebaran[] = {
+	smnPCS_XGMI3X16_PCS_ERROR_NONCORRECTABLE_MASK,
+	smnPCS_XGMI3X16_PCS_ERROR_NONCORRECTABLE_MASK + 0x100000,
+	smnPCS_XGMI3X16_PCS_ERROR_NONCORRECTABLE_MASK + 0x200000,
+	smnPCS_XGMI3X16_PCS_ERROR_NONCORRECTABLE_MASK + 0x300000,
+	smnPCS_XGMI3X16_PCS_ERROR_NONCORRECTABLE_MASK + 0x400000,
+	smnPCS_XGMI3X16_PCS_ERROR_NONCORRECTABLE_MASK + 0x500000,
+	smnPCS_XGMI3X16_PCS_ERROR_NONCORRECTABLE_MASK + 0x600000,
+	smnPCS_XGMI3X16_PCS_ERROR_NONCORRECTABLE_MASK + 0x700000
+};
+
+static const int walf_pcs_err_status_reg_aldebaran[] = {
+	smnPCS_GOPX1_PCS_ERROR_STATUS,
+	smnPCS_GOPX1_PCS_ERROR_STATUS + 0x100000
+};
+
+static const int walf_pcs_err_noncorrectable_mask_reg_aldebaran[] = {
+	smnPCS_GOPX1_PCS_ERROR_NONCORRECTABLE_MASK,
+	smnPCS_GOPX1_PCS_ERROR_NONCORRECTABLE_MASK + 0x100000
+};
+
+static const int xgmi3x16_pcs_err_status_reg_v6_4[] = {
+	smnPCS_XGMI3X16_PCS_ERROR_STATUS,
+	smnPCS_XGMI3X16_PCS_ERROR_STATUS + 0x100000
+};
+
+static const int xgmi3x16_pcs_err_noncorrectable_mask_reg_v6_4[] = {
+	smnPCS_XGMI3X16_PCS_ERROR_NONCORRECTABLE_MASK,
+	smnPCS_XGMI3X16_PCS_ERROR_NONCORRECTABLE_MASK + 0x100000
+};
+
+static const u64 xgmi_v6_4_0_mca_base_array[] = {
+	0x11a09200,
+	0x11b09200,
+};
+
+static const char *xgmi_v6_4_0_ras_error_code_ext[32] = {
+	[0x00] = "XGMI PCS DataLossErr",
+	[0x01] = "XGMI PCS TrainingErr",
+	[0x02] = "XGMI PCS FlowCtrlAckErr",
+	[0x03] = "XGMI PCS RxFifoUnderflowErr",
+	[0x04] = "XGMI PCS RxFifoOverflowErr",
+	[0x05] = "XGMI PCS CRCErr",
+	[0x06] = "XGMI PCS BERExceededErr",
+	[0x07] = "XGMI PCS TxMetaDataErr",
+	[0x08] = "XGMI PCS ReplayBufParityErr",
+	[0x09] = "XGMI PCS DataParityErr",
+	[0x0a] = "XGMI PCS ReplayFifoOverflowErr",
+	[0x0b] = "XGMI PCS ReplayFifoUnderflowErr",
+	[0x0c] = "XGMI PCS ElasticFifoOverflowErr",
+	[0x0d] = "XGMI PCS DeskewErr",
+	[0x0e] = "XGMI PCS FlowCtrlCRCErr",
+	[0x0f] = "XGMI PCS DataStartupLimitErr",
+	[0x10] = "XGMI PCS FCInitTimeoutErr",
+	[0x11] = "XGMI PCS RecoveryTimeoutErr",
+	[0x12] = "XGMI PCS ReadySerialTimeoutErr",
+	[0x13] = "XGMI PCS ReadySerialAttemptErr",
+	[0x14] = "XGMI PCS RecoveryAttemptErr",
+	[0x15] = "XGMI PCS RecoveryRelockAttemptErr",
+	[0x16] = "XGMI PCS ReplayAttemptErr",
+	[0x17] = "XGMI PCS SyncHdrErr",
+	[0x18] = "XGMI PCS TxReplayTimeoutErr",
+	[0x19] = "XGMI PCS RxReplayTimeoutErr",
+	[0x1a] = "XGMI PCS LinkSubTxTimeoutErr",
+	[0x1b] = "XGMI PCS LinkSubRxTimeoutErr",
+	[0x1c] = "XGMI PCS RxCMDPktErr",
+};
+
+static const struct amdgpu_pcs_ras_field xgmi_pcs_ras_fields[] = {
+	{"XGMI PCS DataLossErr",
+	 SOC15_REG_FIELD(XGMI0_PCS_GOPX16_PCS_ERROR_STATUS, DataLossErr)},
+	{"XGMI PCS TrainingErr",
+	 SOC15_REG_FIELD(XGMI0_PCS_GOPX16_PCS_ERROR_STATUS, TrainingErr)},
+	{"XGMI PCS CRCErr",
+	 SOC15_REG_FIELD(XGMI0_PCS_GOPX16_PCS_ERROR_STATUS, CRCErr)},
+	{"XGMI PCS BERExceededErr",
+	 SOC15_REG_FIELD(XGMI0_PCS_GOPX16_PCS_ERROR_STATUS, BERExceededErr)},
+	{"XGMI PCS TxMetaDataErr",
+	 SOC15_REG_FIELD(XGMI0_PCS_GOPX16_PCS_ERROR_STATUS, TxMetaDataErr)},
+	{"XGMI PCS ReplayBufParityErr",
+	 SOC15_REG_FIELD(XGMI0_PCS_GOPX16_PCS_ERROR_STATUS, ReplayBufParityErr)},
+	{"XGMI PCS DataParityErr",
+	 SOC15_REG_FIELD(XGMI0_PCS_GOPX16_PCS_ERROR_STATUS, DataParityErr)},
+	{"XGMI PCS ReplayFifoOverflowErr",
+	 SOC15_REG_FIELD(XGMI0_PCS_GOPX16_PCS_ERROR_STATUS, ReplayFifoOverflowErr)},
+	{"XGMI PCS ReplayFifoUnderflowErr",
+	 SOC15_REG_FIELD(XGMI0_PCS_GOPX16_PCS_ERROR_STATUS, ReplayFifoUnderflowErr)},
+	{"XGMI PCS ElasticFifoOverflowErr",
+	 SOC15_REG_FIELD(XGMI0_PCS_GOPX16_PCS_ERROR_STATUS, ElasticFifoOverflowErr)},
+	{"XGMI PCS DeskewErr",
+	 SOC15_REG_FIELD(XGMI0_PCS_GOPX16_PCS_ERROR_STATUS, DeskewErr)},
+	{"XGMI PCS DataStartupLimitErr",
+	 SOC15_REG_FIELD(XGMI0_PCS_GOPX16_PCS_ERROR_STATUS, DataStartupLimitErr)},
+	{"XGMI PCS FCInitTimeoutErr",
+	 SOC15_REG_FIELD(XGMI0_PCS_GOPX16_PCS_ERROR_STATUS, FCInitTimeoutErr)},
+	{"XGMI PCS RecoveryTimeoutErr",
+	 SOC15_REG_FIELD(XGMI0_PCS_GOPX16_PCS_ERROR_STATUS, RecoveryTimeoutErr)},
+	{"XGMI PCS ReadySerialTimeoutErr",
+	 SOC15_REG_FIELD(XGMI0_PCS_GOPX16_PCS_ERROR_STATUS, ReadySerialTimeoutErr)},
+	{"XGMI PCS ReadySerialAttemptErr",
+	 SOC15_REG_FIELD(XGMI0_PCS_GOPX16_PCS_ERROR_STATUS, ReadySerialAttemptErr)},
+	{"XGMI PCS RecoveryAttemptErr",
+	 SOC15_REG_FIELD(XGMI0_PCS_GOPX16_PCS_ERROR_STATUS, RecoveryAttemptErr)},
+	{"XGMI PCS RecoveryRelockAttemptErr",
+	 SOC15_REG_FIELD(XGMI0_PCS_GOPX16_PCS_ERROR_STATUS, RecoveryRelockAttemptErr)},
+};
+
+static const struct amdgpu_pcs_ras_field wafl_pcs_ras_fields[] = {
+	{"WAFL PCS DataLossErr",
+	 SOC15_REG_FIELD(PCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS, DataLossErr)},
+	{"WAFL PCS TrainingErr",
+	 SOC15_REG_FIELD(PCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS, TrainingErr)},
+	{"WAFL PCS CRCErr",
+	 SOC15_REG_FIELD(PCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS, CRCErr)},
+	{"WAFL PCS BERExceededErr",
+	 SOC15_REG_FIELD(PCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS, BERExceededErr)},
+	{"WAFL PCS TxMetaDataErr",
+	 SOC15_REG_FIELD(PCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS, TxMetaDataErr)},
+	{"WAFL PCS ReplayBufParityErr",
+	 SOC15_REG_FIELD(PCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS, ReplayBufParityErr)},
+	{"WAFL PCS DataParityErr",
+	 SOC15_REG_FIELD(PCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS, DataParityErr)},
+	{"WAFL PCS ReplayFifoOverflowErr",
+	 SOC15_REG_FIELD(PCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS, ReplayFifoOverflowErr)},
+	{"WAFL PCS ReplayFifoUnderflowErr",
+	 SOC15_REG_FIELD(PCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS, ReplayFifoUnderflowErr)},
+	{"WAFL PCS ElasticFifoOverflowErr",
+	 SOC15_REG_FIELD(PCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS, ElasticFifoOverflowErr)},
+	{"WAFL PCS DeskewErr",
+	 SOC15_REG_FIELD(PCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS, DeskewErr)},
+	{"WAFL PCS DataStartupLimitErr",
+	 SOC15_REG_FIELD(PCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS, DataStartupLimitErr)},
+	{"WAFL PCS FCInitTimeoutErr",
+	 SOC15_REG_FIELD(PCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS, FCInitTimeoutErr)},
+	{"WAFL PCS RecoveryTimeoutErr",
+	 SOC15_REG_FIELD(PCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS, RecoveryTimeoutErr)},
+	{"WAFL PCS ReadySerialTimeoutErr",
+	 SOC15_REG_FIELD(PCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS, ReadySerialTimeoutErr)},
+	{"WAFL PCS ReadySerialAttemptErr",
+	 SOC15_REG_FIELD(PCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS, ReadySerialAttemptErr)},
+	{"WAFL PCS RecoveryAttemptErr",
+	 SOC15_REG_FIELD(PCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS, RecoveryAttemptErr)},
+	{"WAFL PCS RecoveryRelockAttemptErr",
+	 SOC15_REG_FIELD(PCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS, RecoveryRelockAttemptErr)},
+};
+
+static const struct amdgpu_pcs_ras_field xgmi3x16_pcs_ras_fields[] = {
+	{"XGMI3X16 PCS DataLossErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, DataLossErr)},
+	{"XGMI3X16 PCS TrainingErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, TrainingErr)},
+	{"XGMI3X16 PCS FlowCtrlAckErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, FlowCtrlAckErr)},
+	{"XGMI3X16 PCS RxFifoUnderflowErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, RxFifoUnderflowErr)},
+	{"XGMI3X16 PCS RxFifoOverflowErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, RxFifoOverflowErr)},
+	{"XGMI3X16 PCS CRCErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, CRCErr)},
+	{"XGMI3X16 PCS BERExceededErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, BERExceededErr)},
+	{"XGMI3X16 PCS TxVcidDataErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, TxVcidDataErr)},
+	{"XGMI3X16 PCS ReplayBufParityErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, ReplayBufParityErr)},
+	{"XGMI3X16 PCS DataParityErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, DataParityErr)},
+	{"XGMI3X16 PCS ReplayFifoOverflowErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, ReplayFifoOverflowErr)},
+	{"XGMI3X16 PCS ReplayFifoUnderflowErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, ReplayFifoUnderflowErr)},
+	{"XGMI3X16 PCS ElasticFifoOverflowErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, ElasticFifoOverflowErr)},
+	{"XGMI3X16 PCS DeskewErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, DeskewErr)},
+	{"XGMI3X16 PCS FlowCtrlCRCErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, FlowCtrlCRCErr)},
+	{"XGMI3X16 PCS DataStartupLimitErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, DataStartupLimitErr)},
+	{"XGMI3X16 PCS FCInitTimeoutErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, FCInitTimeoutErr)},
+	{"XGMI3X16 PCS RecoveryTimeoutErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, RecoveryTimeoutErr)},
+	{"XGMI3X16 PCS ReadySerialTimeoutErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, ReadySerialTimeoutErr)},
+	{"XGMI3X16 PCS ReadySerialAttemptErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, ReadySerialAttemptErr)},
+	{"XGMI3X16 PCS RecoveryAttemptErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, RecoveryAttemptErr)},
+	{"XGMI3X16 PCS RecoveryRelockAttemptErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, RecoveryRelockAttemptErr)},
+	{"XGMI3X16 PCS ReplayAttemptErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, ReplayAttemptErr)},
+	{"XGMI3X16 PCS SyncHdrErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, SyncHdrErr)},
+	{"XGMI3X16 PCS TxReplayTimeoutErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, TxReplayTimeoutErr)},
+	{"XGMI3X16 PCS RxReplayTimeoutErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, RxReplayTimeoutErr)},
+	{"XGMI3X16 PCS LinkSubTxTimeoutErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, LinkSubTxTimeoutErr)},
+	{"XGMI3X16 PCS LinkSubRxTimeoutErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, LinkSubRxTimeoutErr)},
+	{"XGMI3X16 PCS RxCMDPktErr",
+	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, RxCMDPktErr)},
+};
+
+int amdgpu_xgmi_get_ext_link(struct amdgpu_device *adev, int link_num)
+{
+	int link_map_6_4_x[8] = { 0, 3, 1, 2, 7, 6, 4, 5 };
+
+	switch (amdgpu_ip_version(adev, XGMI_HWIP, 0)) {
+	case IP_VERSION(6, 4, 0):
+	case IP_VERSION(6, 4, 1):
+		if (link_num < ARRAY_SIZE(link_map_6_4_x))
+			return link_map_6_4_x[link_num];
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return -EINVAL;
+}
+
+static u32 xgmi_v6_4_get_link_status(struct amdgpu_device *adev, int global_link_num)
+{
+	const u32 smn_xgmi_6_4_pcs_state_hist1[2] = { 0x11a00070, 0x11b00070 };
+	const u32 smn_xgmi_6_4_1_pcs_state_hist1[2] = { 0x12100070,
+							0x11b00070 };
+	u32 i, n;
+	u64 addr;
+
+	switch (amdgpu_ip_version(adev, XGMI_HWIP, 0)) {
+	case IP_VERSION(6, 4, 0):
+		n = ARRAY_SIZE(smn_xgmi_6_4_pcs_state_hist1);
+		addr = smn_xgmi_6_4_pcs_state_hist1[global_link_num % n];
+		break;
+	case IP_VERSION(6, 4, 1):
+		n = ARRAY_SIZE(smn_xgmi_6_4_1_pcs_state_hist1);
+		addr = smn_xgmi_6_4_1_pcs_state_hist1[global_link_num % n];
+		break;
+	default:
+		return U32_MAX;
+	}
+
+	i = global_link_num / n;
+	addr += adev->asic_funcs->encode_ext_smn_addressing(i);
+
+	return RREG32_PCIE_EXT(addr);
+}
+
+int amdgpu_get_xgmi_link_status(struct amdgpu_device *adev, int global_link_num)
+{
+	u32 xgmi_state_reg_val;
+
+	switch (amdgpu_ip_version(adev, XGMI_HWIP, 0)) {
+	case IP_VERSION(6, 4, 0):
+	case IP_VERSION(6, 4, 1):
+		xgmi_state_reg_val = xgmi_v6_4_get_link_status(adev, global_link_num);
+		break;
+	default:
+		return -EOPNOTSUPP;
+	}
+
+	if ((xgmi_state_reg_val & 0xFF) == XGMI_STATE_DISABLE)
+		return -ENOLINK;
+
+	if ((xgmi_state_reg_val & 0xFF) == XGMI_STATE_LS0)
+		return XGMI_LINK_ACTIVE;
+
+	return XGMI_LINK_INACTIVE;
+}
+
+/**
+ * DOC: AMDGPU XGMI Support
+ *
+ * XGMI is a high speed interconnect that joins multiple GPU cards
+ * into a homogeneous memory space that is organized by a collective
+ * hive ID and individual node IDs, both of which are 64-bit numbers.
+ *
+ * The file xgmi_device_id contains the unique per GPU device ID and
+ * is stored in the /sys/class/drm/card${cardno}/device/ directory.
+ *
+ * Inside the device directory a sub-directory 'xgmi_hive_info' is
+ * created which contains the hive ID and the list of nodes.
+ *
+ * The hive ID is stored in:
+ *   /sys/class/drm/card${cardno}/device/xgmi_hive_info/xgmi_hive_id
+ *
+ * The node information is stored in numbered directories:
+ *   /sys/class/drm/card${cardno}/device/xgmi_hive_info/node${nodeno}/xgmi_device_id
+ *
+ * Each device has their own xgmi_hive_info direction with a mirror
+ * set of node sub-directories.
+ *
+ * The XGMI memory space is built by contiguously adding the power of
+ * two padded VRAM space from each node to each other.
+ *
+ */
+
+static struct attribute amdgpu_xgmi_hive_id = {
+	.name = "xgmi_hive_id",
+	.mode = S_IRUGO
+};
+
+static struct attribute *amdgpu_xgmi_hive_attrs[] = {
+	&amdgpu_xgmi_hive_id,
+	NULL
+};
+ATTRIBUTE_GROUPS(amdgpu_xgmi_hive);
+
+static ssize_t amdgpu_xgmi_show_attrs(struct kobject *kobj,
+	struct attribute *attr, char *buf)
+{
+	struct amdgpu_hive_info *hive = container_of(
+		kobj, struct amdgpu_hive_info, kobj);
+
+	if (attr == &amdgpu_xgmi_hive_id)
+		return snprintf(buf, PAGE_SIZE, "%llu\n", hive->hive_id);
+
+	return 0;
+}
+
+static void amdgpu_xgmi_hive_release(struct kobject *kobj)
+{
+	struct amdgpu_hive_info *hive = container_of(
+		kobj, struct amdgpu_hive_info, kobj);
+
+	amdgpu_reset_put_reset_domain(hive->reset_domain);
+	hive->reset_domain = NULL;
+
+	mutex_destroy(&hive->hive_lock);
+	kfree(hive);
+}
+
+static const struct sysfs_ops amdgpu_xgmi_hive_ops = {
+	.show = amdgpu_xgmi_show_attrs,
+};
+
+static const struct kobj_type amdgpu_xgmi_hive_type = {
+	.release = amdgpu_xgmi_hive_release,
+	.sysfs_ops = &amdgpu_xgmi_hive_ops,
+	.default_groups = amdgpu_xgmi_hive_groups,
+};
+
+static ssize_t amdgpu_xgmi_show_device_id(struct device *dev,
+				     struct device_attribute *attr,
+				     char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	return sysfs_emit(buf, "%llu\n", adev->gmc.xgmi.node_id);
+
+}
+
+static ssize_t amdgpu_xgmi_show_physical_id(struct device *dev,
+				     struct device_attribute *attr,
+				     char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	return sysfs_emit(buf, "%u\n", adev->gmc.xgmi.physical_node_id);
+
+}
+
+static ssize_t amdgpu_xgmi_show_num_hops(struct device *dev,
+					struct device_attribute *attr,
+					char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	struct psp_xgmi_topology_info *top = &adev->psp.xgmi_context.top_info;
+	int i;
+
+	for (i = 0; i < top->num_nodes; i++)
+		sprintf(buf + 3 * i, "%02x ", top->nodes[i].num_hops);
+
+	return sysfs_emit(buf, "%s\n", buf);
+}
+
+static ssize_t amdgpu_xgmi_show_num_links(struct device *dev,
+					struct device_attribute *attr,
+					char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	struct psp_xgmi_topology_info *top = &adev->psp.xgmi_context.top_info;
+	int i;
+
+	for (i = 0; i < top->num_nodes; i++)
+		sprintf(buf + 3 * i, "%02x ", top->nodes[i].num_links);
+
+	return sysfs_emit(buf, "%s\n", buf);
+}
+
+static ssize_t amdgpu_xgmi_show_connected_port_num(struct device *dev,
+					struct device_attribute *attr,
+					char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	struct psp_xgmi_topology_info *top = &adev->psp.xgmi_context.top_info;
+	int i, j, size = 0;
+	int current_node;
+	/*
+	 * get the node id in the sysfs for the current socket and show
+	 * it in the port num info output in the sysfs for easy reading.
+	 * it is NOT the one retrieved from xgmi ta.
+	 */
+	for (i = 0; i < top->num_nodes; i++) {
+		if (top->nodes[i].node_id == adev->gmc.xgmi.node_id) {
+			current_node = i;
+			break;
+		}
+	}
+
+	if (i == top->num_nodes)
+		return -EINVAL;
+
+	for (i = 0; i < top->num_nodes; i++) {
+		for (j = 0; j < top->nodes[i].num_links; j++)
+			/* node id in sysfs starts from 1 rather than 0 so +1 here */
+			size += sysfs_emit_at(buf, size, "%02x:%02x ->  %02x:%02x\n", current_node + 1,
+					      top->nodes[i].port_num[j].src_xgmi_port_num, i + 1,
+					      top->nodes[i].port_num[j].dst_xgmi_port_num);
+	}
+
+	return size;
+}
+
+#define AMDGPU_XGMI_SET_FICAA(o)	((o) | 0x456801)
+static ssize_t amdgpu_xgmi_show_error(struct device *dev,
+				      struct device_attribute *attr,
+				      char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	uint32_t ficaa_pie_ctl_in, ficaa_pie_status_in;
+	uint64_t fica_out;
+	unsigned int error_count = 0;
+
+	ficaa_pie_ctl_in = AMDGPU_XGMI_SET_FICAA(0x200);
+	ficaa_pie_status_in = AMDGPU_XGMI_SET_FICAA(0x208);
+
+	if ((!adev->df.funcs) ||
+	    (!adev->df.funcs->get_fica) ||
+	    (!adev->df.funcs->set_fica))
+		return -EINVAL;
+
+	fica_out = adev->df.funcs->get_fica(adev, ficaa_pie_ctl_in);
+	if (fica_out != 0x1f)
+		pr_err("xGMI error counters not enabled!\n");
+
+	fica_out = adev->df.funcs->get_fica(adev, ficaa_pie_status_in);
+
+	if ((fica_out & 0xffff) == 2)
+		error_count = ((fica_out >> 62) & 0x1) + (fica_out >> 63);
+
+	adev->df.funcs->set_fica(adev, ficaa_pie_status_in, 0, 0);
+
+	return sysfs_emit(buf, "%u\n", error_count);
+}
+
+
+static DEVICE_ATTR(xgmi_device_id, S_IRUGO, amdgpu_xgmi_show_device_id, NULL);
+static DEVICE_ATTR(xgmi_physical_id, 0444, amdgpu_xgmi_show_physical_id, NULL);
+static DEVICE_ATTR(xgmi_error, S_IRUGO, amdgpu_xgmi_show_error, NULL);
+static DEVICE_ATTR(xgmi_num_hops, S_IRUGO, amdgpu_xgmi_show_num_hops, NULL);
+static DEVICE_ATTR(xgmi_num_links, S_IRUGO, amdgpu_xgmi_show_num_links, NULL);
+static DEVICE_ATTR(xgmi_port_num, S_IRUGO, amdgpu_xgmi_show_connected_port_num, NULL);
+
+static int amdgpu_xgmi_sysfs_add_dev_info(struct amdgpu_device *adev,
+					 struct amdgpu_hive_info *hive)
+{
+	int ret = 0;
+	char node[10] = { 0 };
+
+	/* Create xgmi device id file */
+	ret = device_create_file(adev->dev, &dev_attr_xgmi_device_id);
+	if (ret) {
+		dev_err(adev->dev, "XGMI: Failed to create device file xgmi_device_id\n");
+		return ret;
+	}
+
+	ret = device_create_file(adev->dev, &dev_attr_xgmi_physical_id);
+	if (ret) {
+		dev_err(adev->dev, "XGMI: Failed to create device file xgmi_physical_id\n");
+		return ret;
+	}
+
+	/* Create xgmi error file */
+	ret = device_create_file(adev->dev, &dev_attr_xgmi_error);
+	if (ret)
+		pr_err("failed to create xgmi_error\n");
+
+	/* Create xgmi num hops file */
+	ret = device_create_file(adev->dev, &dev_attr_xgmi_num_hops);
+	if (ret)
+		pr_err("failed to create xgmi_num_hops\n");
+
+	/* Create xgmi num links file */
+	ret = device_create_file(adev->dev, &dev_attr_xgmi_num_links);
+	if (ret)
+		pr_err("failed to create xgmi_num_links\n");
+
+	/* Create xgmi port num file if supported */
+	if (adev->psp.xgmi_context.xgmi_ta_caps & EXTEND_PEER_LINK_INFO_CMD_FLAG) {
+		ret = device_create_file(adev->dev, &dev_attr_xgmi_port_num);
+		if (ret)
+			dev_err(adev->dev, "failed to create xgmi_port_num\n");
+	}
+
+	/* Create sysfs link to hive info folder on the first device */
+	if (hive->kobj.parent != (&adev->dev->kobj)) {
+		ret = sysfs_create_link(&adev->dev->kobj, &hive->kobj,
+					"xgmi_hive_info");
+		if (ret) {
+			dev_err(adev->dev, "XGMI: Failed to create link to hive info");
+			goto remove_file;
+		}
+	}
+
+	sprintf(node, "node%d", atomic_read(&hive->number_devices));
+	/* Create sysfs link form the hive folder to yourself */
+	ret = sysfs_create_link(&hive->kobj, &adev->dev->kobj, node);
+	if (ret) {
+		dev_err(adev->dev, "XGMI: Failed to create link from hive info");
+		goto remove_link;
+	}
+
+	goto success;
+
+
+remove_link:
+	sysfs_remove_link(&adev->dev->kobj, adev_to_drm(adev)->unique);
+
+remove_file:
+	device_remove_file(adev->dev, &dev_attr_xgmi_device_id);
+	device_remove_file(adev->dev, &dev_attr_xgmi_physical_id);
+	device_remove_file(adev->dev, &dev_attr_xgmi_error);
+	device_remove_file(adev->dev, &dev_attr_xgmi_num_hops);
+	device_remove_file(adev->dev, &dev_attr_xgmi_num_links);
+	if (adev->psp.xgmi_context.xgmi_ta_caps & EXTEND_PEER_LINK_INFO_CMD_FLAG)
+		device_remove_file(adev->dev, &dev_attr_xgmi_port_num);
+
+success:
+	return ret;
+}
+
+static void amdgpu_xgmi_sysfs_rem_dev_info(struct amdgpu_device *adev,
+					  struct amdgpu_hive_info *hive)
+{
+	char node[10];
+	memset(node, 0, sizeof(node));
+
+	device_remove_file(adev->dev, &dev_attr_xgmi_device_id);
+	device_remove_file(adev->dev, &dev_attr_xgmi_physical_id);
+	device_remove_file(adev->dev, &dev_attr_xgmi_error);
+	device_remove_file(adev->dev, &dev_attr_xgmi_num_hops);
+	device_remove_file(adev->dev, &dev_attr_xgmi_num_links);
+	if (adev->psp.xgmi_context.xgmi_ta_caps & EXTEND_PEER_LINK_INFO_CMD_FLAG)
+		device_remove_file(adev->dev, &dev_attr_xgmi_port_num);
+
+	if (hive->kobj.parent != (&adev->dev->kobj))
+		sysfs_remove_link(&adev->dev->kobj,"xgmi_hive_info");
+
+	sprintf(node, "node%d", atomic_read(&hive->number_devices));
+	sysfs_remove_link(&hive->kobj, node);
+
+}
+
+
+
+struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev)
+{
+	struct amdgpu_hive_info *hive = NULL;
+	int ret;
+
+	if (!adev->gmc.xgmi.hive_id)
+		return NULL;
+
+	if (adev->hive) {
+		kobject_get(&adev->hive->kobj);
+		return adev->hive;
+	}
+
+	mutex_lock(&xgmi_mutex);
+
+	list_for_each_entry(hive, &xgmi_hive_list, node)  {
+		if (hive->hive_id == adev->gmc.xgmi.hive_id)
+			goto pro_end;
+	}
+
+	hive = kzalloc(sizeof(*hive), GFP_KERNEL);
+	if (!hive) {
+		dev_err(adev->dev, "XGMI: allocation failed\n");
+		ret = -ENOMEM;
+		hive = NULL;
+		goto pro_end;
+	}
+
+	/* initialize new hive if not exist */
+	ret = kobject_init_and_add(&hive->kobj,
+			&amdgpu_xgmi_hive_type,
+			&adev->dev->kobj,
+			"%s", "xgmi_hive_info");
+	if (ret) {
+		dev_err(adev->dev, "XGMI: failed initializing kobject for xgmi hive\n");
+		kobject_put(&hive->kobj);
+		hive = NULL;
+		goto pro_end;
+	}
+
+	/**
+	 * Only init hive->reset_domain for none SRIOV configuration. For SRIOV,
+	 * Host driver decide how to reset the GPU either through FLR or chain reset.
+	 * Guest side will get individual notifications from the host for the FLR
+	 * if necessary.
+	 */
+	if (!amdgpu_sriov_vf(adev)) {
+	/**
+	 * Avoid recreating reset domain when hive is reconstructed for the case
+	 * of reset the devices in the XGMI hive during probe for passthrough GPU
+	 * See https://www.spinics.net/lists/amd-gfx/msg58836.html
+	 */
+		if (adev->reset_domain->type != XGMI_HIVE) {
+			hive->reset_domain =
+				amdgpu_reset_create_reset_domain(XGMI_HIVE, "amdgpu-reset-hive");
+			if (!hive->reset_domain) {
+				dev_err(adev->dev, "XGMI: failed initializing reset domain for xgmi hive\n");
+				ret = -ENOMEM;
+				kobject_put(&hive->kobj);
+				hive = NULL;
+				goto pro_end;
+			}
+		} else {
+			amdgpu_reset_get_reset_domain(adev->reset_domain);
+			hive->reset_domain = adev->reset_domain;
+		}
+	}
+
+	hive->hive_id = adev->gmc.xgmi.hive_id;
+	INIT_LIST_HEAD(&hive->device_list);
+	INIT_LIST_HEAD(&hive->node);
+	mutex_init(&hive->hive_lock);
+	atomic_set(&hive->number_devices, 0);
+	task_barrier_init(&hive->tb);
+	hive->pstate = AMDGPU_XGMI_PSTATE_UNKNOWN;
+	hive->hi_req_gpu = NULL;
+	atomic_set(&hive->requested_nps_mode, UNKNOWN_MEMORY_PARTITION_MODE);
+
+	/*
+	 * hive pstate on boot is high in vega20 so we have to go to low
+	 * pstate on after boot.
+	 */
+	hive->hi_req_count = AMDGPU_MAX_XGMI_DEVICE_PER_HIVE;
+	list_add_tail(&hive->node, &xgmi_hive_list);
+
+pro_end:
+	if (hive)
+		kobject_get(&hive->kobj);
+	mutex_unlock(&xgmi_mutex);
+	return hive;
+}
+
+void amdgpu_put_xgmi_hive(struct amdgpu_hive_info *hive)
+{
+	if (hive)
+		kobject_put(&hive->kobj);
+}
+
+int amdgpu_xgmi_set_pstate(struct amdgpu_device *adev, int pstate)
+{
+	int ret = 0;
+	struct amdgpu_hive_info *hive;
+	struct amdgpu_device *request_adev;
+	bool is_hi_req = pstate == AMDGPU_XGMI_PSTATE_MAX_VEGA20;
+	bool init_low;
+
+	hive = amdgpu_get_xgmi_hive(adev);
+	if (!hive)
+		return 0;
+
+	request_adev = hive->hi_req_gpu ? hive->hi_req_gpu : adev;
+	init_low = hive->pstate == AMDGPU_XGMI_PSTATE_UNKNOWN;
+	amdgpu_put_xgmi_hive(hive);
+	/* fw bug so temporarily disable pstate switching */
+	return 0;
+
+	if (!hive || adev->asic_type != CHIP_VEGA20)
+		return 0;
+
+	mutex_lock(&hive->hive_lock);
+
+	if (is_hi_req)
+		hive->hi_req_count++;
+	else
+		hive->hi_req_count--;
+
+	/*
+	 * Vega20 only needs single peer to request pstate high for the hive to
+	 * go high but all peers must request pstate low for the hive to go low
+	 */
+	if (hive->pstate == pstate ||
+			(!is_hi_req && hive->hi_req_count && !init_low))
+		goto out;
+
+	dev_dbg(request_adev->dev, "Set xgmi pstate %d.\n", pstate);
+
+	ret = amdgpu_dpm_set_xgmi_pstate(request_adev, pstate);
+	if (ret) {
+		dev_err(request_adev->dev,
+			"XGMI: Set pstate failure on device %llx, hive %llx, ret %d",
+			request_adev->gmc.xgmi.node_id,
+			request_adev->gmc.xgmi.hive_id, ret);
+		goto out;
+	}
+
+	if (init_low)
+		hive->pstate = hive->hi_req_count ?
+					hive->pstate : AMDGPU_XGMI_PSTATE_MIN;
+	else {
+		hive->pstate = pstate;
+		hive->hi_req_gpu = pstate != AMDGPU_XGMI_PSTATE_MIN ?
+							adev : NULL;
+	}
+out:
+	mutex_unlock(&hive->hive_lock);
+	return ret;
+}
+
+int amdgpu_xgmi_update_topology(struct amdgpu_hive_info *hive, struct amdgpu_device *adev)
+{
+	int ret;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	/* Each psp need to set the latest topology */
+	ret = psp_xgmi_set_topology_info(&adev->psp,
+					 atomic_read(&hive->number_devices),
+					 &adev->psp.xgmi_context.top_info);
+	if (ret)
+		dev_err(adev->dev,
+			"XGMI: Set topology failure on device %llx, hive %llx, ret %d",
+			adev->gmc.xgmi.node_id,
+			adev->gmc.xgmi.hive_id, ret);
+
+	return ret;
+}
+
+
+/*
+ * NOTE psp_xgmi_node_info.num_hops layout is as follows:
+ * num_hops[7:6] = link type (0 = xGMI2, 1 = xGMI3, 2/3 = reserved)
+ * num_hops[5:3] = reserved
+ * num_hops[2:0] = number of hops
+ */
+int amdgpu_xgmi_get_hops_count(struct amdgpu_device *adev,
+			       struct amdgpu_device *peer_adev)
+{
+	struct psp_xgmi_topology_info *top = &adev->psp.xgmi_context.top_info;
+	uint8_t num_hops_mask = 0x7;
+	int i;
+
+	if (!adev->gmc.xgmi.supported)
+		return 0;
+
+	for (i = 0 ; i < top->num_nodes; ++i)
+		if (top->nodes[i].node_id == peer_adev->gmc.xgmi.node_id)
+			return top->nodes[i].num_hops & num_hops_mask;
+
+	dev_err(adev->dev, "Failed to get xgmi hops count for peer %d.\n",
+		peer_adev->gmc.xgmi.physical_node_id);
+
+	return 0;
+}
+
+int amdgpu_xgmi_get_bandwidth(struct amdgpu_device *adev, struct amdgpu_device *peer_adev,
+			      enum amdgpu_xgmi_bw_mode bw_mode, enum amdgpu_xgmi_bw_unit bw_unit,
+			      uint32_t *min_bw, uint32_t *max_bw)
+{
+	bool peer_mode = bw_mode == AMDGPU_XGMI_BW_MODE_PER_PEER;
+	int unit_scale = bw_unit == AMDGPU_XGMI_BW_UNIT_MBYTES ? 1000 : 1;
+	int num_lanes = adev->gmc.xgmi.max_width;
+	int speed = adev->gmc.xgmi.max_speed;
+	int num_links = !peer_mode ? 1 : -1;
+
+	if (!(min_bw && max_bw))
+		return -EINVAL;
+
+	*min_bw = 0;
+	*max_bw = 0;
+
+	if (!adev->gmc.xgmi.supported)
+		return -ENODATA;
+
+	if (peer_mode && !peer_adev)
+		return -EINVAL;
+
+	if (peer_mode) {
+		struct psp_xgmi_topology_info *top = &adev->psp.xgmi_context.top_info;
+		int i;
+
+		for (i = 0 ; i < top->num_nodes; ++i) {
+			if (top->nodes[i].node_id != peer_adev->gmc.xgmi.node_id)
+				continue;
+
+			num_links =  top->nodes[i].num_links;
+			break;
+		}
+	}
+
+	if (num_links == -1) {
+		dev_err(adev->dev, "Failed to get number of xgmi links for peer %d.\n",
+			peer_adev->gmc.xgmi.physical_node_id);
+	} else if (num_links) {
+		int per_link_bw = (speed * num_lanes * unit_scale)/BITS_PER_BYTE;
+
+		*min_bw = per_link_bw;
+		*max_bw = num_links * per_link_bw;
+	}
+
+	return 0;
+}
+
+bool amdgpu_xgmi_get_is_sharing_enabled(struct amdgpu_device *adev,
+					struct amdgpu_device *peer_adev)
+{
+	struct psp_xgmi_topology_info *top = &adev->psp.xgmi_context.top_info;
+	int i;
+
+	/* Sharing should always be enabled for non-SRIOV. */
+	if (!amdgpu_sriov_vf(adev))
+		return true;
+
+	for (i = 0 ; i < top->num_nodes; ++i)
+		if (top->nodes[i].node_id == peer_adev->gmc.xgmi.node_id)
+			return !!top->nodes[i].is_sharing_enabled;
+
+	return false;
+}
+
+/*
+ * Devices that support extended data require the entire hive to initialize with
+ * the shared memory buffer flag set.
+ *
+ * Hive locks and conditions apply - see amdgpu_xgmi_add_device
+ */
+static int amdgpu_xgmi_initialize_hive_get_data_partition(struct amdgpu_hive_info *hive,
+							bool set_extended_data)
+{
+	struct amdgpu_device *tmp_adev;
+	int ret;
+
+	list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) {
+		ret = psp_xgmi_initialize(&tmp_adev->psp, set_extended_data, false);
+		if (ret) {
+			dev_err(tmp_adev->dev,
+				"XGMI: Failed to initialize xgmi session for data partition %i\n",
+				set_extended_data);
+			return ret;
+		}
+
+	}
+
+	return 0;
+}
+
+static void amdgpu_xgmi_fill_topology_info(struct amdgpu_device *adev,
+	struct amdgpu_device *peer_adev)
+{
+	struct psp_xgmi_topology_info *top_info = &adev->psp.xgmi_context.top_info;
+	struct psp_xgmi_topology_info *peer_info = &peer_adev->psp.xgmi_context.top_info;
+
+	for (int i = 0; i < peer_info->num_nodes; i++) {
+		if (peer_info->nodes[i].node_id == adev->gmc.xgmi.node_id) {
+			for (int j = 0; j < top_info->num_nodes; j++) {
+				if (top_info->nodes[j].node_id == peer_adev->gmc.xgmi.node_id) {
+					peer_info->nodes[i].num_hops = top_info->nodes[j].num_hops;
+					peer_info->nodes[i].is_sharing_enabled =
+							top_info->nodes[j].is_sharing_enabled;
+					peer_info->nodes[i].num_links =
+							top_info->nodes[j].num_links;
+					return;
+				}
+			}
+		}
+	}
+}
+
+int amdgpu_xgmi_add_device(struct amdgpu_device *adev)
+{
+	struct psp_xgmi_topology_info *top_info;
+	struct amdgpu_hive_info *hive;
+	struct amdgpu_xgmi	*entry;
+	struct amdgpu_device *tmp_adev = NULL;
+
+	int count = 0, ret = 0;
+
+	if (!adev->gmc.xgmi.supported)
+		return 0;
+
+	if (amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_PSP)) {
+		ret = psp_xgmi_initialize(&adev->psp, false, true);
+		if (ret) {
+			dev_err(adev->dev,
+				"XGMI: Failed to initialize xgmi session\n");
+			return ret;
+		}
+
+		ret = psp_xgmi_get_hive_id(&adev->psp, &adev->gmc.xgmi.hive_id);
+		if (ret) {
+			dev_err(adev->dev,
+				"XGMI: Failed to get hive id\n");
+			return ret;
+		}
+
+		ret = psp_xgmi_get_node_id(&adev->psp, &adev->gmc.xgmi.node_id);
+		if (ret) {
+			dev_err(adev->dev,
+				"XGMI: Failed to get node id\n");
+			return ret;
+		}
+	} else {
+		adev->gmc.xgmi.hive_id = 16;
+		adev->gmc.xgmi.node_id = adev->gmc.xgmi.physical_node_id + 16;
+	}
+
+	hive = amdgpu_get_xgmi_hive(adev);
+	if (!hive) {
+		ret = -EINVAL;
+		dev_err(adev->dev,
+			"XGMI: node 0x%llx, can not match hive 0x%llx in the hive list.\n",
+			adev->gmc.xgmi.node_id, adev->gmc.xgmi.hive_id);
+		goto exit;
+	}
+	mutex_lock(&hive->hive_lock);
+
+	top_info = &adev->psp.xgmi_context.top_info;
+
+	list_add_tail(&adev->gmc.xgmi.head, &hive->device_list);
+	list_for_each_entry(entry, &hive->device_list, head)
+		top_info->nodes[count++].node_id = entry->node_id;
+	top_info->num_nodes = count;
+	atomic_set(&hive->number_devices, count);
+
+	task_barrier_add_task(&hive->tb);
+
+	if (amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_PSP)) {
+		list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) {
+			/* update node list for other device in the hive */
+			if (tmp_adev != adev) {
+				top_info = &tmp_adev->psp.xgmi_context.top_info;
+				top_info->nodes[count - 1].node_id =
+					adev->gmc.xgmi.node_id;
+				top_info->num_nodes = count;
+			}
+			ret = amdgpu_xgmi_update_topology(hive, tmp_adev);
+			if (ret)
+				goto exit_unlock;
+		}
+
+		if (amdgpu_sriov_vf(adev) &&
+			adev->psp.xgmi_context.xgmi_ta_caps & EXTEND_PEER_LINK_INFO_CMD_FLAG) {
+			/* only get topology for VF being init if it can support full duplex */
+			ret = psp_xgmi_get_topology_info(&adev->psp, count,
+						&adev->psp.xgmi_context.top_info, false);
+			if (ret) {
+				dev_err(adev->dev,
+					"XGMI: Get topology failure on device %llx, hive %llx, ret %d",
+					adev->gmc.xgmi.node_id,
+					adev->gmc.xgmi.hive_id, ret);
+				/* To do: continue with some node failed or disable the whole hive*/
+				goto exit_unlock;
+			}
+
+			/* fill the topology info for peers instead of getting from PSP */
+			list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) {
+				amdgpu_xgmi_fill_topology_info(adev, tmp_adev);
+			}
+		} else {
+			/* get latest topology info for each device from psp */
+			list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) {
+				ret = psp_xgmi_get_topology_info(&tmp_adev->psp, count,
+					&tmp_adev->psp.xgmi_context.top_info, false);
+				if (ret) {
+					dev_err(tmp_adev->dev,
+						"XGMI: Get topology failure on device %llx, hive %llx, ret %d",
+						tmp_adev->gmc.xgmi.node_id,
+						tmp_adev->gmc.xgmi.hive_id, ret);
+					/* To do : continue with some node failed or disable the whole hive */
+					goto exit_unlock;
+				}
+			}
+		}
+
+		/* get topology again for hives that support extended data */
+		if (adev->psp.xgmi_context.supports_extended_data) {
+
+			/* initialize the hive to get extended data.  */
+			ret = amdgpu_xgmi_initialize_hive_get_data_partition(hive, true);
+			if (ret)
+				goto exit_unlock;
+
+			/* get the extended data. */
+			list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) {
+				ret = psp_xgmi_get_topology_info(&tmp_adev->psp, count,
+						&tmp_adev->psp.xgmi_context.top_info, true);
+				if (ret) {
+					dev_err(tmp_adev->dev,
+						"XGMI: Get topology for extended data failure on device %llx, hive %llx, ret %d",
+						tmp_adev->gmc.xgmi.node_id,
+						tmp_adev->gmc.xgmi.hive_id, ret);
+					goto exit_unlock;
+				}
+			}
+
+			/* initialize the hive to get non-extended data for the next round. */
+			ret = amdgpu_xgmi_initialize_hive_get_data_partition(hive, false);
+			if (ret)
+				goto exit_unlock;
+
+		}
+	}
+
+	if (!ret)
+		ret = amdgpu_xgmi_sysfs_add_dev_info(adev, hive);
+
+exit_unlock:
+	mutex_unlock(&hive->hive_lock);
+exit:
+	if (!ret) {
+		adev->hive = hive;
+		dev_info(adev->dev, "XGMI: Add node %d, hive 0x%llx.\n",
+			 adev->gmc.xgmi.physical_node_id, adev->gmc.xgmi.hive_id);
+	} else {
+		amdgpu_put_xgmi_hive(hive);
+		dev_err(adev->dev, "XGMI: Failed to add node %d, hive 0x%llx ret: %d\n",
+			adev->gmc.xgmi.physical_node_id, adev->gmc.xgmi.hive_id,
+			ret);
+	}
+
+	return ret;
+}
+
+int amdgpu_xgmi_remove_device(struct amdgpu_device *adev)
+{
+	struct amdgpu_hive_info *hive = adev->hive;
+
+	if (!adev->gmc.xgmi.supported)
+		return -EINVAL;
+
+	if (!hive)
+		return -EINVAL;
+
+	mutex_lock(&hive->hive_lock);
+	task_barrier_rem_task(&hive->tb);
+	amdgpu_xgmi_sysfs_rem_dev_info(adev, hive);
+	if (hive->hi_req_gpu == adev)
+		hive->hi_req_gpu = NULL;
+	list_del(&adev->gmc.xgmi.head);
+	mutex_unlock(&hive->hive_lock);
+
+	amdgpu_put_xgmi_hive(hive);
+	adev->hive = NULL;
+
+	if (atomic_dec_return(&hive->number_devices) == 0) {
+		/* Remove the hive from global hive list */
+		mutex_lock(&xgmi_mutex);
+		list_del(&hive->node);
+		mutex_unlock(&xgmi_mutex);
+
+		amdgpu_put_xgmi_hive(hive);
+	}
+
+	return 0;
+}
+
+static int xgmi_v6_4_0_aca_bank_parser(struct aca_handle *handle, struct aca_bank *bank,
+				       enum aca_smu_type type, void *data)
+{
+	struct amdgpu_device *adev = handle->adev;
+	struct aca_bank_info info;
+	const char *error_str;
+	u64 status, count;
+	int ret, ext_error_code;
+
+	ret = aca_bank_info_decode(bank, &info);
+	if (ret)
+		return ret;
+
+	status = bank->regs[ACA_REG_IDX_STATUS];
+	ext_error_code = ACA_REG__STATUS__ERRORCODEEXT(status);
+
+	error_str = ext_error_code < ARRAY_SIZE(xgmi_v6_4_0_ras_error_code_ext) ?
+		xgmi_v6_4_0_ras_error_code_ext[ext_error_code] : NULL;
+	if (error_str)
+		dev_info(adev->dev, "%s detected\n", error_str);
+
+	count = ACA_REG__MISC0__ERRCNT(bank->regs[ACA_REG_IDX_MISC0]);
+
+	switch (type) {
+	case ACA_SMU_TYPE_UE:
+		if (ext_error_code != 0 && ext_error_code != 9)
+			count = 0ULL;
+
+		bank->aca_err_type = ACA_ERROR_TYPE_UE;
+		ret = aca_error_cache_log_bank_error(handle, &info, ACA_ERROR_TYPE_UE, count);
+		break;
+	case ACA_SMU_TYPE_CE:
+		count = ext_error_code == 6 ? count : 0ULL;
+		bank->aca_err_type = ACA_ERROR_TYPE_CE;
+		ret = aca_error_cache_log_bank_error(handle, &info, bank->aca_err_type, count);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return ret;
+}
+
+static const struct aca_bank_ops xgmi_v6_4_0_aca_bank_ops = {
+	.aca_bank_parser = xgmi_v6_4_0_aca_bank_parser,
+};
+
+static const struct aca_info xgmi_v6_4_0_aca_info = {
+	.hwip = ACA_HWIP_TYPE_PCS_XGMI,
+	.mask = ACA_ERROR_UE_MASK | ACA_ERROR_CE_MASK,
+	.bank_ops = &xgmi_v6_4_0_aca_bank_ops,
+};
+
+static int amdgpu_xgmi_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
+{
+	int r;
+
+	if (!adev->gmc.xgmi.supported ||
+	    adev->gmc.xgmi.num_physical_nodes == 0)
+		return 0;
+
+	amdgpu_ras_reset_error_count(adev, AMDGPU_RAS_BLOCK__XGMI_WAFL);
+
+	r = amdgpu_ras_block_late_init(adev, ras_block);
+	if (r)
+		return r;
+
+	switch (amdgpu_ip_version(adev, XGMI_HWIP, 0)) {
+	case IP_VERSION(6, 4, 0):
+	case IP_VERSION(6, 4, 1):
+		r = amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__XGMI_WAFL,
+					&xgmi_v6_4_0_aca_info, NULL);
+		if (r)
+			goto late_fini;
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+
+late_fini:
+	amdgpu_ras_block_late_fini(adev, ras_block);
+
+	return r;
+}
+
+uint64_t amdgpu_xgmi_get_relative_phy_addr(struct amdgpu_device *adev,
+					   uint64_t addr)
+{
+	struct amdgpu_xgmi *xgmi = &adev->gmc.xgmi;
+	return (addr + xgmi->physical_node_id * xgmi->node_segment_size);
+}
+
+static void pcs_clear_status(struct amdgpu_device *adev, uint32_t pcs_status_reg)
+{
+	WREG32_PCIE(pcs_status_reg, 0xFFFFFFFF);
+	WREG32_PCIE(pcs_status_reg, 0);
+}
+
+static void amdgpu_xgmi_legacy_reset_ras_error_count(struct amdgpu_device *adev)
+{
+	uint32_t i;
+
+	switch (adev->asic_type) {
+	case CHIP_ARCTURUS:
+		for (i = 0; i < ARRAY_SIZE(xgmi_pcs_err_status_reg_arct); i++)
+			pcs_clear_status(adev,
+					 xgmi_pcs_err_status_reg_arct[i]);
+		break;
+	case CHIP_VEGA20:
+		for (i = 0; i < ARRAY_SIZE(xgmi_pcs_err_status_reg_vg20); i++)
+			pcs_clear_status(adev,
+					 xgmi_pcs_err_status_reg_vg20[i]);
+		break;
+	case CHIP_ALDEBARAN:
+		for (i = 0; i < ARRAY_SIZE(xgmi3x16_pcs_err_status_reg_aldebaran); i++)
+			pcs_clear_status(adev,
+					 xgmi3x16_pcs_err_status_reg_aldebaran[i]);
+		for (i = 0; i < ARRAY_SIZE(walf_pcs_err_status_reg_aldebaran); i++)
+			pcs_clear_status(adev,
+					 walf_pcs_err_status_reg_aldebaran[i]);
+		break;
+	default:
+		break;
+	}
+
+	switch (amdgpu_ip_version(adev, XGMI_HWIP, 0)) {
+	case IP_VERSION(6, 4, 0):
+	case IP_VERSION(6, 4, 1):
+		for (i = 0; i < ARRAY_SIZE(xgmi3x16_pcs_err_status_reg_v6_4); i++)
+			pcs_clear_status(adev,
+					xgmi3x16_pcs_err_status_reg_v6_4[i]);
+		break;
+	default:
+		break;
+	}
+}
+
+static void __xgmi_v6_4_0_reset_error_count(struct amdgpu_device *adev, int xgmi_inst, u64 mca_base)
+{
+	WREG64_MCA(xgmi_inst, mca_base, ACA_REG_IDX_STATUS, 0ULL);
+}
+
+static void xgmi_v6_4_0_reset_error_count(struct amdgpu_device *adev, int xgmi_inst)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(xgmi_v6_4_0_mca_base_array); i++)
+		__xgmi_v6_4_0_reset_error_count(adev, xgmi_inst, xgmi_v6_4_0_mca_base_array[i]);
+}
+
+static void xgmi_v6_4_0_reset_ras_error_count(struct amdgpu_device *adev)
+{
+	int i;
+
+	for_each_inst(i, adev->aid_mask)
+		xgmi_v6_4_0_reset_error_count(adev, i);
+}
+
+static void amdgpu_xgmi_reset_ras_error_count(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, XGMI_HWIP, 0)) {
+	case IP_VERSION(6, 4, 0):
+	case IP_VERSION(6, 4, 1):
+		xgmi_v6_4_0_reset_ras_error_count(adev);
+		break;
+	default:
+		amdgpu_xgmi_legacy_reset_ras_error_count(adev);
+		break;
+	}
+}
+
+static int amdgpu_xgmi_query_pcs_error_status(struct amdgpu_device *adev,
+					      uint32_t value,
+						  uint32_t mask_value,
+					      uint32_t *ue_count,
+					      uint32_t *ce_count,
+					      bool is_xgmi_pcs,
+						  bool check_mask)
+{
+	int i;
+	int ue_cnt = 0;
+	const struct amdgpu_pcs_ras_field *pcs_ras_fields = NULL;
+	uint32_t field_array_size = 0;
+
+	if (is_xgmi_pcs) {
+		if (amdgpu_ip_version(adev, XGMI_HWIP, 0) ==
+		    IP_VERSION(6, 1, 0) ||
+		    amdgpu_ip_version(adev, XGMI_HWIP, 0) ==
+		    IP_VERSION(6, 4, 0) ||
+		    amdgpu_ip_version(adev, XGMI_HWIP, 0) ==
+		    IP_VERSION(6, 4, 1)) {
+			pcs_ras_fields = &xgmi3x16_pcs_ras_fields[0];
+			field_array_size = ARRAY_SIZE(xgmi3x16_pcs_ras_fields);
+		} else {
+			pcs_ras_fields = &xgmi_pcs_ras_fields[0];
+			field_array_size = ARRAY_SIZE(xgmi_pcs_ras_fields);
+		}
+	} else {
+		pcs_ras_fields = &wafl_pcs_ras_fields[0];
+		field_array_size = ARRAY_SIZE(wafl_pcs_ras_fields);
+	}
+
+	if (check_mask)
+		value = value & ~mask_value;
+
+	/* query xgmi/walf pcs error status,
+	 * only ue is supported */
+	for (i = 0; value && i < field_array_size; i++) {
+		ue_cnt = (value &
+				pcs_ras_fields[i].pcs_err_mask) >>
+				pcs_ras_fields[i].pcs_err_shift;
+		if (ue_cnt) {
+			dev_info(adev->dev, "%s detected\n",
+				 pcs_ras_fields[i].err_name);
+			*ue_count += ue_cnt;
+		}
+
+		/* reset bit value if the bit is checked */
+		value &= ~(pcs_ras_fields[i].pcs_err_mask);
+	}
+
+	return 0;
+}
+
+static void amdgpu_xgmi_legacy_query_ras_error_count(struct amdgpu_device *adev,
+						     void *ras_error_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+	int i, supported = 1;
+	uint32_t data, mask_data = 0;
+	uint32_t ue_cnt = 0, ce_cnt = 0;
+
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__XGMI_WAFL))
+		return ;
+
+	err_data->ue_count = 0;
+	err_data->ce_count = 0;
+
+	switch (adev->asic_type) {
+	case CHIP_ARCTURUS:
+		/* check xgmi pcs error */
+		for (i = 0; i < ARRAY_SIZE(xgmi_pcs_err_status_reg_arct); i++) {
+			data = RREG32_PCIE(xgmi_pcs_err_status_reg_arct[i]);
+			if (data)
+				amdgpu_xgmi_query_pcs_error_status(adev, data,
+						mask_data, &ue_cnt, &ce_cnt, true, false);
+		}
+		/* check wafl pcs error */
+		for (i = 0; i < ARRAY_SIZE(wafl_pcs_err_status_reg_arct); i++) {
+			data = RREG32_PCIE(wafl_pcs_err_status_reg_arct[i]);
+			if (data)
+				amdgpu_xgmi_query_pcs_error_status(adev, data,
+						mask_data, &ue_cnt, &ce_cnt, false, false);
+		}
+		break;
+	case CHIP_VEGA20:
+		/* check xgmi pcs error */
+		for (i = 0; i < ARRAY_SIZE(xgmi_pcs_err_status_reg_vg20); i++) {
+			data = RREG32_PCIE(xgmi_pcs_err_status_reg_vg20[i]);
+			if (data)
+				amdgpu_xgmi_query_pcs_error_status(adev, data,
+						mask_data, &ue_cnt, &ce_cnt, true, false);
+		}
+		/* check wafl pcs error */
+		for (i = 0; i < ARRAY_SIZE(wafl_pcs_err_status_reg_vg20); i++) {
+			data = RREG32_PCIE(wafl_pcs_err_status_reg_vg20[i]);
+			if (data)
+				amdgpu_xgmi_query_pcs_error_status(adev, data,
+						mask_data, &ue_cnt, &ce_cnt, false, false);
+		}
+		break;
+	case CHIP_ALDEBARAN:
+		/* check xgmi3x16 pcs error */
+		for (i = 0; i < ARRAY_SIZE(xgmi3x16_pcs_err_status_reg_aldebaran); i++) {
+			data = RREG32_PCIE(xgmi3x16_pcs_err_status_reg_aldebaran[i]);
+			mask_data =
+				RREG32_PCIE(xgmi3x16_pcs_err_noncorrectable_mask_reg_aldebaran[i]);
+			if (data)
+				amdgpu_xgmi_query_pcs_error_status(adev, data,
+						mask_data, &ue_cnt, &ce_cnt, true, true);
+		}
+		/* check wafl pcs error */
+		for (i = 0; i < ARRAY_SIZE(walf_pcs_err_status_reg_aldebaran); i++) {
+			data = RREG32_PCIE(walf_pcs_err_status_reg_aldebaran[i]);
+			mask_data =
+				RREG32_PCIE(walf_pcs_err_noncorrectable_mask_reg_aldebaran[i]);
+			if (data)
+				amdgpu_xgmi_query_pcs_error_status(adev, data,
+						mask_data, &ue_cnt, &ce_cnt, false, true);
+		}
+		break;
+	default:
+		supported = 0;
+		break;
+	}
+
+	switch (amdgpu_ip_version(adev, XGMI_HWIP, 0)) {
+	case IP_VERSION(6, 4, 0):
+	case IP_VERSION(6, 4, 1):
+		/* check xgmi3x16 pcs error */
+		for (i = 0; i < ARRAY_SIZE(xgmi3x16_pcs_err_status_reg_v6_4); i++) {
+			data = RREG32_PCIE(xgmi3x16_pcs_err_status_reg_v6_4[i]);
+			mask_data =
+				RREG32_PCIE(xgmi3x16_pcs_err_noncorrectable_mask_reg_v6_4[i]);
+			if (data)
+				amdgpu_xgmi_query_pcs_error_status(adev, data,
+						mask_data, &ue_cnt, &ce_cnt, true, true);
+		}
+		break;
+	default:
+		if (!supported)
+			dev_warn(adev->dev, "XGMI RAS error query not supported");
+		break;
+	}
+
+	amdgpu_ras_reset_error_count(adev, AMDGPU_RAS_BLOCK__XGMI_WAFL);
+
+	err_data->ue_count += ue_cnt;
+	err_data->ce_count += ce_cnt;
+}
+
+static enum aca_error_type xgmi_v6_4_0_pcs_mca_get_error_type(struct amdgpu_device *adev, u64 status)
+{
+	const char *error_str;
+	int ext_error_code;
+
+	ext_error_code = ACA_REG__STATUS__ERRORCODEEXT(status);
+
+	error_str = ext_error_code < ARRAY_SIZE(xgmi_v6_4_0_ras_error_code_ext) ?
+		xgmi_v6_4_0_ras_error_code_ext[ext_error_code] : NULL;
+	if (error_str)
+		dev_info(adev->dev, "%s detected\n", error_str);
+
+	switch (ext_error_code) {
+	case 0:
+		return ACA_ERROR_TYPE_UE;
+	case 6:
+		return ACA_ERROR_TYPE_CE;
+	default:
+		return -EINVAL;
+	}
+
+	return -EINVAL;
+}
+
+static void __xgmi_v6_4_0_query_error_count(struct amdgpu_device *adev, struct amdgpu_smuio_mcm_config_info *mcm_info,
+					    u64 mca_base, struct ras_err_data *err_data)
+{
+	int xgmi_inst = mcm_info->die_id;
+	u64 status = 0;
+
+	status = RREG64_MCA(xgmi_inst, mca_base, ACA_REG_IDX_STATUS);
+	if (!ACA_REG__STATUS__VAL(status))
+		return;
+
+	switch (xgmi_v6_4_0_pcs_mca_get_error_type(adev, status)) {
+	case ACA_ERROR_TYPE_UE:
+		amdgpu_ras_error_statistic_ue_count(err_data, mcm_info, 1ULL);
+		break;
+	case ACA_ERROR_TYPE_CE:
+		amdgpu_ras_error_statistic_ce_count(err_data, mcm_info, 1ULL);
+		break;
+	default:
+		break;
+	}
+
+	WREG64_MCA(xgmi_inst, mca_base, ACA_REG_IDX_STATUS, 0ULL);
+}
+
+static void xgmi_v6_4_0_query_error_count(struct amdgpu_device *adev, int xgmi_inst, struct ras_err_data *err_data)
+{
+	struct amdgpu_smuio_mcm_config_info mcm_info = {
+		.socket_id = adev->smuio.funcs->get_socket_id(adev),
+		.die_id = xgmi_inst,
+	};
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(xgmi_v6_4_0_mca_base_array); i++)
+		__xgmi_v6_4_0_query_error_count(adev, &mcm_info, xgmi_v6_4_0_mca_base_array[i], err_data);
+}
+
+static void xgmi_v6_4_0_query_ras_error_count(struct amdgpu_device *adev, void *ras_error_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+	int i;
+
+	for_each_inst(i, adev->aid_mask)
+		xgmi_v6_4_0_query_error_count(adev, i, err_data);
+}
+
+static void amdgpu_xgmi_query_ras_error_count(struct amdgpu_device *adev,
+					      void *ras_error_status)
+{
+	switch (amdgpu_ip_version(adev, XGMI_HWIP, 0)) {
+	case IP_VERSION(6, 4, 0):
+	case IP_VERSION(6, 4, 1):
+		xgmi_v6_4_0_query_ras_error_count(adev, ras_error_status);
+		break;
+	default:
+		amdgpu_xgmi_legacy_query_ras_error_count(adev, ras_error_status);
+		break;
+	}
+}
+
+/* Trigger XGMI/WAFL error */
+static int amdgpu_ras_error_inject_xgmi(struct amdgpu_device *adev,
+			void *inject_if, uint32_t instance_mask)
+{
+	int ret1, ret2;
+	struct ta_ras_trigger_error_input *block_info =
+				(struct ta_ras_trigger_error_input *)inject_if;
+
+	if (amdgpu_dpm_set_df_cstate(adev, DF_CSTATE_DISALLOW))
+		dev_warn(adev->dev, "Failed to disallow df cstate");
+
+	ret1 = amdgpu_dpm_set_pm_policy(adev, PP_PM_POLICY_XGMI_PLPD, XGMI_PLPD_DISALLOW);
+	if (ret1 && ret1 != -EOPNOTSUPP)
+		dev_warn(adev->dev, "Failed to disallow XGMI power down");
+
+	ret2 = psp_ras_trigger_error(&adev->psp, block_info, instance_mask);
+
+	if (amdgpu_ras_intr_triggered())
+		return ret2;
+
+	ret1 = amdgpu_dpm_set_pm_policy(adev, PP_PM_POLICY_XGMI_PLPD, XGMI_PLPD_DEFAULT);
+	if (ret1 && ret1 != -EOPNOTSUPP)
+		dev_warn(adev->dev, "Failed to allow XGMI power down");
+
+	if (amdgpu_dpm_set_df_cstate(adev, DF_CSTATE_ALLOW))
+		dev_warn(adev->dev, "Failed to allow df cstate");
+
+	return ret2;
+}
+
+struct amdgpu_ras_block_hw_ops  xgmi_ras_hw_ops = {
+	.query_ras_error_count = amdgpu_xgmi_query_ras_error_count,
+	.reset_ras_error_count = amdgpu_xgmi_reset_ras_error_count,
+	.ras_error_inject = amdgpu_ras_error_inject_xgmi,
+};
+
+struct amdgpu_xgmi_ras xgmi_ras = {
+	.ras_block = {
+		.hw_ops = &xgmi_ras_hw_ops,
+		.ras_late_init = amdgpu_xgmi_ras_late_init,
+	},
+};
+
+int amdgpu_xgmi_ras_sw_init(struct amdgpu_device *adev)
+{
+	int err;
+	struct amdgpu_xgmi_ras *ras;
+
+	if (!adev->gmc.xgmi.ras)
+		return 0;
+
+	ras = adev->gmc.xgmi.ras;
+	err = amdgpu_ras_register_ras_block(adev, &ras->ras_block);
+	if (err) {
+		dev_err(adev->dev, "Failed to register xgmi_wafl_pcs ras block!\n");
+		return err;
+	}
+
+	strcpy(ras->ras_block.ras_comm.name, "xgmi_wafl");
+	ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__XGMI_WAFL;
+	ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
+	adev->gmc.xgmi.ras_if = &ras->ras_block.ras_comm;
+
+	return 0;
+}
+
+static void amdgpu_xgmi_reset_on_init_work(struct work_struct *work)
+{
+	struct amdgpu_hive_info *hive =
+		container_of(work, struct amdgpu_hive_info, reset_on_init_work);
+	struct amdgpu_reset_context reset_context;
+	struct amdgpu_device *tmp_adev;
+	struct list_head device_list;
+	int r;
+
+	mutex_lock(&hive->hive_lock);
+
+	INIT_LIST_HEAD(&device_list);
+	list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head)
+		list_add_tail(&tmp_adev->reset_list, &device_list);
+
+	tmp_adev = list_first_entry(&device_list, struct amdgpu_device,
+				    reset_list);
+	amdgpu_device_lock_reset_domain(tmp_adev->reset_domain);
+
+	reset_context.method = AMD_RESET_METHOD_ON_INIT;
+	reset_context.reset_req_dev = tmp_adev;
+	reset_context.hive = hive;
+	reset_context.reset_device_list = &device_list;
+	set_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
+	set_bit(AMDGPU_SKIP_COREDUMP, &reset_context.flags);
+
+	amdgpu_reset_do_xgmi_reset_on_init(&reset_context);
+	mutex_unlock(&hive->hive_lock);
+	amdgpu_device_unlock_reset_domain(tmp_adev->reset_domain);
+
+	list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) {
+		r = amdgpu_ras_init_badpage_info(tmp_adev);
+		if (r && r != -EHWPOISON)
+			dev_err(tmp_adev->dev,
+				"error during bad page data initialization");
+	}
+}
+
+static void amdgpu_xgmi_schedule_reset_on_init(struct amdgpu_hive_info *hive)
+{
+	INIT_WORK(&hive->reset_on_init_work, amdgpu_xgmi_reset_on_init_work);
+	amdgpu_reset_domain_schedule(hive->reset_domain,
+				     &hive->reset_on_init_work);
+}
+
+int amdgpu_xgmi_reset_on_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_hive_info *hive;
+	bool reset_scheduled;
+	int num_devs;
+
+	hive = amdgpu_get_xgmi_hive(adev);
+	if (!hive)
+		return -EINVAL;
+
+	mutex_lock(&hive->hive_lock);
+	num_devs = atomic_read(&hive->number_devices);
+	reset_scheduled = false;
+	if (num_devs == adev->gmc.xgmi.num_physical_nodes) {
+		amdgpu_xgmi_schedule_reset_on_init(hive);
+		reset_scheduled = true;
+	}
+
+	mutex_unlock(&hive->hive_lock);
+	amdgpu_put_xgmi_hive(hive);
+
+	if (reset_scheduled)
+		flush_work(&hive->reset_on_init_work);
+
+	return 0;
+}
+
+int amdgpu_xgmi_request_nps_change(struct amdgpu_device *adev,
+				   struct amdgpu_hive_info *hive,
+				   int req_nps_mode)
+{
+	struct amdgpu_device *tmp_adev;
+	int cur_nps_mode, r;
+
+	/* This is expected to be called only during unload of driver. The
+	 * request needs to be placed only once for all devices in the hive. If
+	 * one of them fail, revert the request for previous successful devices.
+	 * After placing the request, make hive mode as UNKNOWN so that other
+	 * devices don't request anymore.
+	 */
+	mutex_lock(&hive->hive_lock);
+	if (atomic_read(&hive->requested_nps_mode) ==
+	    UNKNOWN_MEMORY_PARTITION_MODE) {
+		dev_dbg(adev->dev, "Unexpected entry for hive NPS change");
+		mutex_unlock(&hive->hive_lock);
+		return 0;
+	}
+	list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) {
+		r = adev->gmc.gmc_funcs->request_mem_partition_mode(
+			tmp_adev, req_nps_mode);
+		if (r)
+			break;
+	}
+	if (r) {
+		/* Request back current mode if one of the requests failed */
+		cur_nps_mode =
+			adev->gmc.gmc_funcs->query_mem_partition_mode(tmp_adev);
+		list_for_each_entry_continue_reverse(
+			tmp_adev, &hive->device_list, gmc.xgmi.head)
+			adev->gmc.gmc_funcs->request_mem_partition_mode(
+				tmp_adev, cur_nps_mode);
+	}
+	/* Set to UNKNOWN so that other devices don't request anymore */
+	atomic_set(&hive->requested_nps_mode, UNKNOWN_MEMORY_PARTITION_MODE);
+	mutex_unlock(&hive->hive_lock);
+
+	return r;
+}
+
+bool amdgpu_xgmi_same_hive(struct amdgpu_device *adev,
+			   struct amdgpu_device *bo_adev)
+{
+	return (amdgpu_use_xgmi_p2p && adev != bo_adev &&
+		adev->gmc.xgmi.hive_id &&
+		adev->gmc.xgmi.hive_id == bo_adev->gmc.xgmi.hive_id);
+}
+
+void amdgpu_xgmi_early_init(struct amdgpu_device *adev)
+{
+	if (!adev->gmc.xgmi.supported)
+		return;
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 4, 0):
+	case IP_VERSION(9, 4, 1):
+	case IP_VERSION(9, 4, 2):
+		/* 25 GT/s */
+		adev->gmc.xgmi.max_speed = 25;
+		adev->gmc.xgmi.max_width = 16;
+		break;
+	case IP_VERSION(9, 4, 3):
+	case IP_VERSION(9, 4, 4):
+	case IP_VERSION(9, 5, 0):
+		/* 32 GT/s */
+		adev->gmc.xgmi.max_speed = 32;
+		adev->gmc.xgmi.max_width = 16;
+		break;
+	default:
+		break;
+	}
+}
+
+void amgpu_xgmi_set_max_speed_width(struct amdgpu_device *adev,
+				    uint16_t max_speed, uint8_t max_width)
+{
+	adev->gmc.xgmi.max_speed = max_speed;
+	adev->gmc.xgmi.max_width = max_width;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h
new file mode 100644
index 000000000000..5f36aff17e79
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h
@@ -0,0 +1,133 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+#ifndef __AMDGPU_XGMI_H__
+#define __AMDGPU_XGMI_H__
+
+#include <drm/task_barrier.h>
+#include "amdgpu_ras.h"
+
+struct amdgpu_hive_info {
+	struct kobject kobj;
+	uint64_t hive_id;
+	struct list_head device_list;
+	struct list_head node;
+	atomic_t number_devices;
+	struct mutex hive_lock;
+	int hi_req_count;
+	struct amdgpu_device *hi_req_gpu;
+	struct task_barrier tb;
+	enum {
+		AMDGPU_XGMI_PSTATE_MIN,
+		AMDGPU_XGMI_PSTATE_MAX_VEGA20,
+		AMDGPU_XGMI_PSTATE_UNKNOWN
+	} pstate;
+
+	struct amdgpu_reset_domain *reset_domain;
+	atomic_t ras_recovery;
+	struct ras_event_manager event_mgr;
+	struct work_struct reset_on_init_work;
+	atomic_t requested_nps_mode;
+};
+
+struct amdgpu_pcs_ras_field {
+	const char *err_name;
+	uint32_t pcs_err_mask;
+	uint32_t pcs_err_shift;
+};
+
+/**
+ * Bandwidth range reporting comes in two modes.
+ *
+ * PER_LINK - range for any xgmi link
+ * PER_PEER - range of max of single xgmi link to max of multiple links based on source peer
+ */
+enum amdgpu_xgmi_bw_mode {
+	AMDGPU_XGMI_BW_MODE_PER_LINK = 0,
+	AMDGPU_XGMI_BW_MODE_PER_PEER
+};
+
+enum amdgpu_xgmi_bw_unit {
+	AMDGPU_XGMI_BW_UNIT_GBYTES = 0,
+	AMDGPU_XGMI_BW_UNIT_MBYTES
+};
+
+struct amdgpu_xgmi_ras {
+	struct amdgpu_ras_block_object ras_block;
+};
+extern struct amdgpu_xgmi_ras xgmi_ras;
+
+struct amdgpu_xgmi {
+	/* from psp */
+	u64 node_id;
+	u64 hive_id;
+	/* fixed per family */
+	u64 node_segment_size;
+	/* physical node (0-3) */
+	unsigned physical_node_id;
+	/* number of nodes (0-4) */
+	unsigned num_physical_nodes;
+	/* gpu list in the same hive */
+	struct list_head head;
+	bool supported;
+	struct ras_common_if *ras_if;
+	bool connected_to_cpu;
+	struct amdgpu_xgmi_ras *ras;
+	uint16_t max_speed;
+	uint8_t max_width;
+};
+
+struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev);
+void amdgpu_put_xgmi_hive(struct amdgpu_hive_info *hive);
+int amdgpu_xgmi_update_topology(struct amdgpu_hive_info *hive, struct amdgpu_device *adev);
+int amdgpu_xgmi_add_device(struct amdgpu_device *adev);
+int amdgpu_xgmi_remove_device(struct amdgpu_device *adev);
+int amdgpu_xgmi_set_pstate(struct amdgpu_device *adev, int pstate);
+int amdgpu_xgmi_get_hops_count(struct amdgpu_device *adev, struct amdgpu_device *peer_adev);
+int amdgpu_xgmi_get_bandwidth(struct amdgpu_device *adev, struct amdgpu_device *peer_adev,
+			      enum amdgpu_xgmi_bw_mode bw_mode, enum amdgpu_xgmi_bw_unit bw_unit,
+			      uint32_t *min_bw, uint32_t *max_bw);
+bool amdgpu_xgmi_get_is_sharing_enabled(struct amdgpu_device *adev,
+					struct amdgpu_device *peer_adev);
+uint64_t amdgpu_xgmi_get_relative_phy_addr(struct amdgpu_device *adev,
+					   uint64_t addr);
+bool amdgpu_xgmi_same_hive(struct amdgpu_device *adev,
+			   struct amdgpu_device *bo_adev);
+int amdgpu_xgmi_ras_sw_init(struct amdgpu_device *adev);
+int amdgpu_xgmi_reset_on_init(struct amdgpu_device *adev);
+
+int amdgpu_xgmi_request_nps_change(struct amdgpu_device *adev,
+				   struct amdgpu_hive_info *hive,
+				   int req_nps_mode);
+int amdgpu_get_xgmi_link_status(struct amdgpu_device *adev,
+				int global_link_num);
+int amdgpu_xgmi_get_ext_link(struct amdgpu_device *adev, int link_num);
+
+void amdgpu_xgmi_early_init(struct amdgpu_device *adev);
+uint32_t amdgpu_xgmi_get_max_bandwidth(struct amdgpu_device *adev);
+
+void amgpu_xgmi_set_max_speed_width(struct amdgpu_device *adev,
+				    uint16_t max_speed, uint8_t max_width);
+
+/* Cleanup macro for use with __free(xgmi_put_hive) */
+DEFINE_FREE(xgmi_put_hive, struct amdgpu_hive_info *, if (_T) amdgpu_put_xgmi_hive(_T))
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h b/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h
new file mode 100644
index 000000000000..3a79ed7d8031
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h
@@ -0,0 +1,456 @@
+/*
+ * Copyright (c) 2018-2021 Advanced Micro Devices, Inc. All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#ifndef AMDGV_SRIOV_MSG__H_
+#define AMDGV_SRIOV_MSG__H_
+
+/* unit in kilobytes */
+#define AMD_SRIOV_MSG_VBIOS_OFFSET	     0
+#define AMD_SRIOV_MSG_VBIOS_SIZE_KB	     64
+#define AMD_SRIOV_MSG_DATAEXCHANGE_OFFSET_KB AMD_SRIOV_MSG_VBIOS_SIZE_KB
+#define AMD_SRIOV_MSG_DATAEXCHANGE_SIZE_KB   4
+#define AMD_SRIOV_MSG_TMR_OFFSET_KB	     2048
+#define AMD_SRIOV_MSG_BAD_PAGE_SIZE_KB	     2
+#define AMD_SRIOV_RAS_TELEMETRY_SIZE_KB	     64
+/*
+ * layout
+ * 0           64KB        65KB        66KB           68KB                   132KB
+ * |   VBIOS   |   PF2VF   |   VF2PF   |   Bad Page   | RAS Telemetry Region | ...
+ * |   64KB    |   1KB     |   1KB     |   2KB        | 64KB                 | ...
+ */
+
+#define AMD_SRIOV_MSG_SIZE_KB                   1
+#define AMD_SRIOV_MSG_PF2VF_OFFSET_KB           AMD_SRIOV_MSG_DATAEXCHANGE_OFFSET_KB
+#define AMD_SRIOV_MSG_VF2PF_OFFSET_KB           (AMD_SRIOV_MSG_PF2VF_OFFSET_KB + AMD_SRIOV_MSG_SIZE_KB)
+#define AMD_SRIOV_MSG_BAD_PAGE_OFFSET_KB        (AMD_SRIOV_MSG_VF2PF_OFFSET_KB + AMD_SRIOV_MSG_SIZE_KB)
+#define AMD_SRIOV_MSG_RAS_TELEMETRY_OFFSET_KB   (AMD_SRIOV_MSG_BAD_PAGE_OFFSET_KB + AMD_SRIOV_MSG_BAD_PAGE_SIZE_KB)
+
+/*
+ * PF2VF history log:
+ * v1 defined in amdgim
+ * v2 current
+ *
+ * VF2PF history log:
+ * v1 defined in amdgim
+ * v2 defined in amdgim
+ * v3 current
+ */
+#define AMD_SRIOV_MSG_FW_VRAM_PF2VF_VER 2
+#define AMD_SRIOV_MSG_FW_VRAM_VF2PF_VER 3
+
+#define AMD_SRIOV_MSG_RESERVE_UCODE 24
+
+#define AMD_SRIOV_MSG_RESERVE_VCN_INST 4
+
+enum amd_sriov_ucode_engine_id {
+	AMD_SRIOV_UCODE_ID_VCE = 0,
+	AMD_SRIOV_UCODE_ID_UVD,
+	AMD_SRIOV_UCODE_ID_MC,
+	AMD_SRIOV_UCODE_ID_ME,
+	AMD_SRIOV_UCODE_ID_PFP,
+	AMD_SRIOV_UCODE_ID_CE,
+	AMD_SRIOV_UCODE_ID_RLC,
+	AMD_SRIOV_UCODE_ID_RLC_SRLC,
+	AMD_SRIOV_UCODE_ID_RLC_SRLG,
+	AMD_SRIOV_UCODE_ID_RLC_SRLS,
+	AMD_SRIOV_UCODE_ID_MEC,
+	AMD_SRIOV_UCODE_ID_MEC2,
+	AMD_SRIOV_UCODE_ID_SOS,
+	AMD_SRIOV_UCODE_ID_ASD,
+	AMD_SRIOV_UCODE_ID_TA_RAS,
+	AMD_SRIOV_UCODE_ID_TA_XGMI,
+	AMD_SRIOV_UCODE_ID_SMC,
+	AMD_SRIOV_UCODE_ID_SDMA,
+	AMD_SRIOV_UCODE_ID_SDMA2,
+	AMD_SRIOV_UCODE_ID_VCN,
+	AMD_SRIOV_UCODE_ID_DMCU,
+	AMD_SRIOV_UCODE_ID__MAX
+};
+
+#pragma pack(push, 1) // PF2VF / VF2PF data areas are byte packed
+
+union amd_sriov_msg_feature_flags {
+	struct {
+		uint32_t error_log_collect	: 1;
+		uint32_t host_load_ucodes	: 1;
+		uint32_t host_flr_vramlost	: 1;
+		uint32_t mm_bw_management	: 1;
+		uint32_t pp_one_vf_mode		: 1;
+		uint32_t reg_indirect_acc	: 1;
+		uint32_t av1_support		: 1;
+		uint32_t vcn_rb_decouple	: 1;
+		uint32_t mes_info_dump_enable	: 1;
+		uint32_t ras_caps		: 1;
+		uint32_t ras_telemetry		: 1;
+		uint32_t ras_cper		: 1;
+		uint32_t reserved		: 20;
+	} flags;
+	uint32_t all;
+};
+
+union amd_sriov_reg_access_flags {
+	struct {
+		uint32_t vf_reg_access_ih		: 1;
+		uint32_t vf_reg_access_mmhub		: 1;
+		uint32_t vf_reg_access_gc		: 1;
+		uint32_t vf_reg_access_l1_tlb_cntl	: 1;
+		uint32_t vf_reg_access_sq_config	: 1;
+		uint32_t reserved			: 27;
+	} flags;
+	uint32_t all;
+};
+
+union amd_sriov_ras_caps {
+	struct {
+		uint64_t block_umc			: 1;
+		uint64_t block_sdma			: 1;
+		uint64_t block_gfx			: 1;
+		uint64_t block_mmhub			: 1;
+		uint64_t block_athub			: 1;
+		uint64_t block_pcie_bif			: 1;
+		uint64_t block_hdp			: 1;
+		uint64_t block_xgmi_wafl		: 1;
+		uint64_t block_df			: 1;
+		uint64_t block_smn			: 1;
+		uint64_t block_sem			: 1;
+		uint64_t block_mp0			: 1;
+		uint64_t block_mp1			: 1;
+		uint64_t block_fuse			: 1;
+		uint64_t block_mca			: 1;
+		uint64_t block_vcn			: 1;
+		uint64_t block_jpeg			: 1;
+		uint64_t block_ih			: 1;
+		uint64_t block_mpio			: 1;
+		uint64_t poison_propogation_mode	: 1;
+		uint64_t reserved			: 44;
+	} bits;
+	uint64_t all;
+};
+
+union amd_sriov_msg_os_info {
+	struct {
+		uint32_t windows  : 1;
+		uint32_t reserved : 31;
+	} info;
+	uint32_t all;
+};
+
+struct amd_sriov_msg_uuid_info {
+	union {
+		struct {
+			uint32_t did	: 16;
+			uint32_t fcn	: 8;
+			uint32_t asic_7 : 8;
+		};
+		uint32_t time_low;
+	};
+
+	struct {
+		uint32_t time_mid  : 16;
+		uint32_t time_high : 12;
+		uint32_t version   : 4;
+	};
+
+	struct {
+		struct {
+			uint8_t clk_seq_hi : 6;
+			uint8_t variant    : 2;
+		};
+		union {
+			uint8_t clk_seq_low;
+			uint8_t asic_6;
+		};
+		uint16_t asic_4;
+	};
+
+	uint32_t asic_0;
+};
+
+struct amd_sriov_msg_pf2vf_info_header {
+	/* the total structure size in byte */
+	uint32_t size;
+	/* version of this structure, written by the HOST */
+	uint32_t version;
+	/* reserved */
+	uint32_t reserved[2];
+};
+
+#define AMD_SRIOV_MSG_PF2VF_INFO_FILLED_SIZE (55)
+struct amd_sriov_msg_pf2vf_info {
+	/* header contains size and version */
+	struct amd_sriov_msg_pf2vf_info_header header;
+	/* use private key from mailbox 2 to create checksum */
+	uint32_t checksum;
+	/* The features flags of the HOST driver supports */
+	union amd_sriov_msg_feature_flags feature_flags;
+	/* (max_width * max_height * fps) / (16 * 16) */
+	uint32_t hevc_enc_max_mb_per_second;
+	/* (max_width * max_height) / (16 * 16) */
+	uint32_t hevc_enc_max_mb_per_frame;
+	/* (max_width * max_height * fps) / (16 * 16) */
+	uint32_t avc_enc_max_mb_per_second;
+	/* (max_width * max_height) / (16 * 16) */
+	uint32_t avc_enc_max_mb_per_frame;
+	/* MEC FW position in BYTE from the start of VF visible frame buffer */
+	uint64_t mecfw_offset;
+	/* MEC FW size in BYTE */
+	uint32_t mecfw_size;
+	/* UVD FW position in BYTE from the start of VF visible frame buffer */
+	uint64_t uvdfw_offset;
+	/* UVD FW size in BYTE */
+	uint32_t uvdfw_size;
+	/* VCE FW position in BYTE from the start of VF visible frame buffer */
+	uint64_t vcefw_offset;
+	/* VCE FW size in BYTE */
+	uint32_t vcefw_size;
+	/* Bad pages block position in BYTE */
+	uint32_t bp_block_offset_low;
+	uint32_t bp_block_offset_high;
+	/* Bad pages block size in BYTE */
+	uint32_t bp_block_size;
+	/* frequency for VF to update the VF2PF area in msec, 0 = manual */
+	uint32_t vf2pf_update_interval_ms;
+	/* identification in ROCm SMI */
+	uint64_t uuid;
+	uint32_t fcn_idx;
+	/* flags to indicate which register access method VF should use */
+	union amd_sriov_reg_access_flags reg_access_flags;
+	/* MM BW management */
+	struct {
+		uint32_t decode_max_dimension_pixels;
+		uint32_t decode_max_frame_pixels;
+		uint32_t encode_max_dimension_pixels;
+		uint32_t encode_max_frame_pixels;
+	} mm_bw_management[AMD_SRIOV_MSG_RESERVE_VCN_INST];
+	/* UUID info */
+	struct amd_sriov_msg_uuid_info uuid_info;
+	/* PCIE atomic ops support flag */
+	uint32_t pcie_atomic_ops_support_flags;
+	/* Portion of GPU memory occupied by VF.  MAX value is 65535, but set to uint32_t to maintain alignment with reserved size */
+	uint32_t gpu_capacity;
+	/* vf bdf on host pci tree for debug only */
+	uint32_t bdf_on_host;
+	uint32_t more_bp;	//Reserved for future use.
+	union amd_sriov_ras_caps ras_en_caps;
+	union amd_sriov_ras_caps ras_telemetry_en_caps;
+
+	/* reserved */
+	uint32_t reserved[256 - AMD_SRIOV_MSG_PF2VF_INFO_FILLED_SIZE];
+} __packed;
+
+struct amd_sriov_msg_vf2pf_info_header {
+	/* the total structure size in byte */
+	uint32_t size;
+	/* version of this structure, written by the guest */
+	uint32_t version;
+	/* reserved */
+	uint32_t reserved[2];
+};
+
+#define AMD_SRIOV_MSG_VF2PF_INFO_FILLED_SIZE (73)
+struct amd_sriov_msg_vf2pf_info {
+	/* header contains size and version */
+	struct amd_sriov_msg_vf2pf_info_header header;
+	uint32_t checksum;
+	/* driver version */
+	uint8_t driver_version[64];
+	/* driver certification, 1=WHQL, 0=None */
+	uint32_t driver_cert;
+	/* guest OS type and version */
+	union amd_sriov_msg_os_info os_info;
+	/* guest fb information in the unit of MB */
+	uint32_t fb_usage;
+	/* guest gfx engine usage percentage */
+	uint32_t gfx_usage;
+	/* guest gfx engine health percentage */
+	uint32_t gfx_health;
+	/* guest compute engine usage percentage */
+	uint32_t compute_usage;
+	/* guest compute engine health percentage */
+	uint32_t compute_health;
+	/* guest avc engine usage percentage. 0xffff means N/A */
+	uint32_t avc_enc_usage;
+	/* guest avc engine health percentage. 0xffff means N/A */
+	uint32_t avc_enc_health;
+	/* guest hevc engine usage percentage. 0xffff means N/A */
+	uint32_t hevc_enc_usage;
+	/* guest hevc engine usage percentage. 0xffff means N/A */
+	uint32_t hevc_enc_health;
+	/* combined encode/decode usage */
+	uint32_t encode_usage;
+	uint32_t decode_usage;
+	/* Version of PF2VF that VF understands */
+	uint32_t pf2vf_version_required;
+	/* additional FB usage */
+	uint32_t fb_vis_usage;
+	uint32_t fb_vis_size;
+	uint32_t fb_size;
+	/* guest ucode data, each one is 1.25 Dword */
+	struct {
+		uint8_t id;
+		uint32_t version;
+	} ucode_info[AMD_SRIOV_MSG_RESERVE_UCODE];
+	uint64_t dummy_page_addr;
+	/* FB allocated for guest MES to record UQ info */
+	uint64_t mes_info_addr;
+	uint32_t mes_info_size;
+	/* reserved */
+	uint32_t reserved[256 - AMD_SRIOV_MSG_VF2PF_INFO_FILLED_SIZE];
+} __packed;
+
+/* mailbox message send from guest to host  */
+enum amd_sriov_mailbox_request_message {
+	MB_REQ_MSG_REQ_GPU_INIT_ACCESS = 1,
+	MB_REQ_MSG_REL_GPU_INIT_ACCESS,
+	MB_REQ_MSG_REQ_GPU_FINI_ACCESS,
+	MB_REQ_MSG_REL_GPU_FINI_ACCESS,
+	MB_REQ_MSG_REQ_GPU_RESET_ACCESS,
+	MB_REQ_MSG_REQ_GPU_INIT_DATA,
+	MB_REQ_MSG_PSP_VF_CMD_RELAY,
+
+	MB_REQ_MSG_LOG_VF_ERROR = 200,
+	MB_REQ_MSG_READY_TO_RESET = 201,
+	MB_REQ_MSG_RAS_POISON = 202,
+	MB_REQ_RAS_ERROR_COUNT = 203,
+	MB_REQ_RAS_CPER_DUMP = 204,
+	MB_REQ_RAS_BAD_PAGES = 205,
+};
+
+/* mailbox message send from host to guest  */
+enum amd_sriov_mailbox_response_message {
+	MB_RES_MSG_CLR_MSG_BUF			= 0,
+	MB_RES_MSG_READY_TO_ACCESS_GPU		= 1,
+	MB_RES_MSG_FLR_NOTIFICATION		= 2,
+	MB_RES_MSG_FLR_NOTIFICATION_COMPLETION  = 3,
+	MB_RES_MSG_SUCCESS			= 4,
+	MB_RES_MSG_FAIL				= 5,
+	MB_RES_MSG_QUERY_ALIVE			= 6,
+	MB_RES_MSG_GPU_INIT_DATA_READY		= 7,
+	MB_RES_MSG_RAS_POISON_READY		= 8,
+	MB_RES_MSG_PF_SOFT_FLR_NOTIFICATION	= 9,
+	MB_RES_MSG_GPU_RMA			= 10,
+	MB_RES_MSG_RAS_ERROR_COUNT_READY	= 11,
+	MB_REQ_RAS_CPER_DUMP_READY		= 14,
+	MB_RES_MSG_RAS_BAD_PAGES_READY		= 15,
+	MB_RES_MSG_RAS_BAD_PAGES_NOTIFICATION	= 16,
+	MB_RES_MSG_UNRECOV_ERR_NOTIFICATION	= 17,
+	MB_RES_MSG_TEXT_MESSAGE			= 255
+};
+
+enum amd_sriov_ras_telemetry_gpu_block {
+	RAS_TELEMETRY_GPU_BLOCK_UMC		= 0,
+	RAS_TELEMETRY_GPU_BLOCK_SDMA		= 1,
+	RAS_TELEMETRY_GPU_BLOCK_GFX		= 2,
+	RAS_TELEMETRY_GPU_BLOCK_MMHUB		= 3,
+	RAS_TELEMETRY_GPU_BLOCK_ATHUB		= 4,
+	RAS_TELEMETRY_GPU_BLOCK_PCIE_BIF	= 5,
+	RAS_TELEMETRY_GPU_BLOCK_HDP		= 6,
+	RAS_TELEMETRY_GPU_BLOCK_XGMI_WAFL	= 7,
+	RAS_TELEMETRY_GPU_BLOCK_DF		= 8,
+	RAS_TELEMETRY_GPU_BLOCK_SMN		= 9,
+	RAS_TELEMETRY_GPU_BLOCK_SEM		= 10,
+	RAS_TELEMETRY_GPU_BLOCK_MP0		= 11,
+	RAS_TELEMETRY_GPU_BLOCK_MP1		= 12,
+	RAS_TELEMETRY_GPU_BLOCK_FUSE		= 13,
+	RAS_TELEMETRY_GPU_BLOCK_MCA		= 14,
+	RAS_TELEMETRY_GPU_BLOCK_VCN		= 15,
+	RAS_TELEMETRY_GPU_BLOCK_JPEG		= 16,
+	RAS_TELEMETRY_GPU_BLOCK_IH		= 17,
+	RAS_TELEMETRY_GPU_BLOCK_MPIO		= 18,
+	RAS_TELEMETRY_GPU_BLOCK_COUNT		= 19,
+};
+
+struct amd_sriov_ras_telemetry_header {
+	uint32_t checksum;
+	uint32_t used_size;
+	uint32_t reserved[2];
+};
+
+struct amd_sriov_ras_telemetry_error_count {
+	struct {
+		uint32_t ce_count;
+		uint32_t ue_count;
+		uint32_t de_count;
+		uint32_t ce_overflow_count;
+		uint32_t ue_overflow_count;
+		uint32_t de_overflow_count;
+		uint32_t reserved[6];
+	} block[RAS_TELEMETRY_GPU_BLOCK_COUNT];
+};
+
+struct amd_sriov_ras_cper_dump {
+	uint32_t more;
+	uint64_t overflow_count;
+	uint64_t count;
+	uint64_t wptr;
+	uint32_t buf[];
+};
+
+struct amd_sriov_ras_chk_criti {
+	uint32_t hit;
+};
+
+struct amdsriov_ras_telemetry {
+	struct amd_sriov_ras_telemetry_header header;
+
+	union {
+		struct amd_sriov_ras_telemetry_error_count error_count;
+		struct amd_sriov_ras_cper_dump cper_dump;
+		struct amd_sriov_ras_chk_criti chk_criti;
+	} body;
+};
+
+/* version data stored in MAILBOX_MSGBUF_RCV_DW1 for future expansion */
+enum amd_sriov_gpu_init_data_version {
+	GPU_INIT_DATA_READY_V1 = 1,
+};
+
+#pragma pack(pop) // Restore previous packing option
+
+/* checksum function between host and guest */
+unsigned int amd_sriov_msg_checksum(void *obj, unsigned long obj_size, unsigned int key,
+				    unsigned int checksum);
+
+/* assertion at compile time */
+#ifdef __linux__
+#define stringification(s)  _stringification(s)
+#define _stringification(s) #s
+
+_Static_assert(
+	sizeof(struct amd_sriov_msg_vf2pf_info) == AMD_SRIOV_MSG_SIZE_KB << 10,
+	"amd_sriov_msg_vf2pf_info must be " stringification(AMD_SRIOV_MSG_SIZE_KB) " KB");
+
+_Static_assert(
+	sizeof(struct amd_sriov_msg_pf2vf_info) == AMD_SRIOV_MSG_SIZE_KB << 10,
+	"amd_sriov_msg_pf2vf_info must be " stringification(AMD_SRIOV_MSG_SIZE_KB) " KB");
+
+_Static_assert(AMD_SRIOV_MSG_RESERVE_UCODE % 4 == 0,
+	       "AMD_SRIOV_MSG_RESERVE_UCODE must be multiple of 4");
+
+_Static_assert(AMD_SRIOV_MSG_RESERVE_UCODE > AMD_SRIOV_UCODE_ID__MAX,
+	       "AMD_SRIOV_MSG_RESERVE_UCODE must be bigger than AMD_SRIOV_UCODE_ID__MAX");
+
+#undef _stringification
+#undef stringification
+#endif
+
+#endif /* AMDGV_SRIOV_MSG__H_ */
diff --git a/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram.c b/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram.c
new file mode 100644
index 000000000000..811124ff88a8
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram.c
@@ -0,0 +1,985 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "soc15.h"
+
+#include "soc15_common.h"
+#include "amdgpu_reg_state.h"
+#include "amdgpu_xcp.h"
+#include "gfx_v9_4_3.h"
+#include "gfxhub_v1_2.h"
+#include "sdma_v4_4_2.h"
+#include "amdgpu_ip.h"
+
+#define XCP_INST_MASK(num_inst, xcp_id)                                        \
+	(num_inst ? GENMASK(num_inst - 1, 0) << (xcp_id * num_inst) : 0)
+
+void aqua_vanjaram_doorbell_index_init(struct amdgpu_device *adev)
+{
+	int i;
+
+	adev->doorbell_index.kiq = AMDGPU_DOORBELL_LAYOUT1_KIQ_START;
+
+	adev->doorbell_index.mec_ring0 = AMDGPU_DOORBELL_LAYOUT1_MEC_RING_START;
+
+	adev->doorbell_index.userqueue_start = AMDGPU_DOORBELL_LAYOUT1_USERQUEUE_START;
+	adev->doorbell_index.userqueue_end = AMDGPU_DOORBELL_LAYOUT1_USERQUEUE_END;
+	adev->doorbell_index.xcc_doorbell_range = AMDGPU_DOORBELL_LAYOUT1_XCC_RANGE;
+
+	adev->doorbell_index.sdma_doorbell_range = 20;
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		adev->doorbell_index.sdma_engine[i] =
+			AMDGPU_DOORBELL_LAYOUT1_sDMA_ENGINE_START +
+			i * (adev->doorbell_index.sdma_doorbell_range >> 1);
+
+	adev->doorbell_index.ih = AMDGPU_DOORBELL_LAYOUT1_IH;
+	adev->doorbell_index.vcn.vcn_ring0_1 = AMDGPU_DOORBELL_LAYOUT1_VCN_START;
+
+	adev->doorbell_index.first_non_cp = AMDGPU_DOORBELL_LAYOUT1_FIRST_NON_CP;
+	adev->doorbell_index.last_non_cp = AMDGPU_DOORBELL_LAYOUT1_LAST_NON_CP;
+
+	adev->doorbell_index.max_assignment = AMDGPU_DOORBELL_LAYOUT1_MAX_ASSIGNMENT << 1;
+}
+
+/* Fixed pattern for smn addressing on different AIDs:
+ *   bit[34]: indicate cross AID access
+ *   bit[33:32]: indicate target AID id
+ * AID id range is 0 ~ 3 as maximum AID number is 4.
+ */
+u64 aqua_vanjaram_encode_ext_smn_addressing(int ext_id)
+{
+	u64 ext_offset;
+
+	/* local routing and bit[34:32] will be zeros */
+	if (ext_id == 0)
+		return 0;
+
+	/* Initiated from host, accessing to all non-zero aids are cross traffic */
+	ext_offset = ((u64)(ext_id & 0x3) << 32) | (1ULL << 34);
+
+	return ext_offset;
+}
+
+static enum amdgpu_gfx_partition
+__aqua_vanjaram_calc_xcp_mode(struct amdgpu_xcp_mgr *xcp_mgr)
+{
+	struct amdgpu_device *adev = xcp_mgr->adev;
+	int num_xcc, num_xcc_per_xcp = 0, mode = 0;
+
+	num_xcc = NUM_XCC(xcp_mgr->adev->gfx.xcc_mask);
+	if (adev->gfx.funcs->get_xccs_per_xcp)
+		num_xcc_per_xcp = adev->gfx.funcs->get_xccs_per_xcp(adev);
+	if ((num_xcc_per_xcp) && (num_xcc % num_xcc_per_xcp == 0))
+		mode = num_xcc / num_xcc_per_xcp;
+
+	if (num_xcc_per_xcp == 1)
+		return AMDGPU_CPX_PARTITION_MODE;
+
+	switch (mode) {
+	case 1:
+		return AMDGPU_SPX_PARTITION_MODE;
+	case 2:
+		return AMDGPU_DPX_PARTITION_MODE;
+	case 3:
+		return AMDGPU_TPX_PARTITION_MODE;
+	case 4:
+		return AMDGPU_QPX_PARTITION_MODE;
+	default:
+		return AMDGPU_UNKNOWN_COMPUTE_PARTITION_MODE;
+	}
+
+	return AMDGPU_UNKNOWN_COMPUTE_PARTITION_MODE;
+}
+
+static int aqua_vanjaram_query_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr)
+{
+	enum amdgpu_gfx_partition derv_mode,
+		mode = AMDGPU_UNKNOWN_COMPUTE_PARTITION_MODE;
+	struct amdgpu_device *adev = xcp_mgr->adev;
+
+	derv_mode = __aqua_vanjaram_calc_xcp_mode(xcp_mgr);
+
+	if (amdgpu_sriov_vf(adev))
+		return derv_mode;
+
+	if (adev->nbio.funcs->get_compute_partition_mode) {
+		mode = adev->nbio.funcs->get_compute_partition_mode(adev);
+		if (mode != derv_mode) {
+			dev_warn(
+				adev->dev,
+				"Mismatch in compute partition mode - reported : %d derived : %d",
+				mode, derv_mode);
+			if (derv_mode == AMDGPU_UNKNOWN_COMPUTE_PARTITION_MODE)
+				amdgpu_device_bus_status_check(adev);
+		}
+	}
+
+	return mode;
+}
+
+static int __aqua_vanjaram_get_xcc_per_xcp(struct amdgpu_xcp_mgr *xcp_mgr, int mode)
+{
+	int num_xcc, num_xcc_per_xcp = 0;
+
+	num_xcc = NUM_XCC(xcp_mgr->adev->gfx.xcc_mask);
+
+	switch (mode) {
+	case AMDGPU_SPX_PARTITION_MODE:
+		num_xcc_per_xcp = num_xcc;
+		break;
+	case AMDGPU_DPX_PARTITION_MODE:
+		num_xcc_per_xcp = num_xcc / 2;
+		break;
+	case AMDGPU_TPX_PARTITION_MODE:
+		num_xcc_per_xcp = num_xcc / 3;
+		break;
+	case AMDGPU_QPX_PARTITION_MODE:
+		num_xcc_per_xcp = num_xcc / 4;
+		break;
+	case AMDGPU_CPX_PARTITION_MODE:
+		num_xcc_per_xcp = 1;
+		break;
+	}
+
+	return num_xcc_per_xcp;
+}
+
+static int __aqua_vanjaram_get_xcp_ip_info(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id,
+				    enum AMDGPU_XCP_IP_BLOCK ip_id,
+				    struct amdgpu_xcp_ip *ip)
+{
+	struct amdgpu_device *adev = xcp_mgr->adev;
+	int num_sdma, num_vcn, num_shared_vcn, num_xcp;
+	int num_xcc_xcp, num_sdma_xcp, num_vcn_xcp;
+
+	num_sdma = adev->sdma.num_instances;
+	num_vcn = adev->vcn.num_vcn_inst;
+	num_shared_vcn = 1;
+
+	num_xcc_xcp = adev->gfx.num_xcc_per_xcp;
+	num_xcp = NUM_XCC(adev->gfx.xcc_mask) / num_xcc_xcp;
+
+	switch (xcp_mgr->mode) {
+	case AMDGPU_SPX_PARTITION_MODE:
+	case AMDGPU_DPX_PARTITION_MODE:
+	case AMDGPU_TPX_PARTITION_MODE:
+	case AMDGPU_QPX_PARTITION_MODE:
+	case AMDGPU_CPX_PARTITION_MODE:
+		num_sdma_xcp = DIV_ROUND_UP(num_sdma, num_xcp);
+		num_vcn_xcp = DIV_ROUND_UP(num_vcn, num_xcp);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	if (num_vcn && num_xcp > num_vcn)
+		num_shared_vcn = num_xcp / num_vcn;
+
+	switch (ip_id) {
+	case AMDGPU_XCP_GFXHUB:
+		ip->inst_mask = XCP_INST_MASK(num_xcc_xcp, xcp_id);
+		ip->ip_funcs = &gfxhub_v1_2_xcp_funcs;
+		break;
+	case AMDGPU_XCP_GFX:
+		ip->inst_mask = XCP_INST_MASK(num_xcc_xcp, xcp_id);
+		ip->ip_funcs = &gfx_v9_4_3_xcp_funcs;
+		break;
+	case AMDGPU_XCP_SDMA:
+		ip->inst_mask = XCP_INST_MASK(num_sdma_xcp, xcp_id);
+		ip->ip_funcs = &sdma_v4_4_2_xcp_funcs;
+		break;
+	case AMDGPU_XCP_VCN:
+		ip->inst_mask =
+			XCP_INST_MASK(num_vcn_xcp, xcp_id / num_shared_vcn);
+		/* TODO : Assign IP funcs */
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	ip->ip_id = ip_id;
+
+	return 0;
+}
+
+static int __aqua_vanjaram_get_px_mode_info(struct amdgpu_xcp_mgr *xcp_mgr,
+					    int px_mode, int *num_xcp,
+					    uint16_t *nps_modes)
+{
+	struct amdgpu_device *adev = xcp_mgr->adev;
+	uint32_t gc_ver = amdgpu_ip_version(adev, GC_HWIP, 0);
+
+	if (!num_xcp || !nps_modes || !(xcp_mgr->supp_xcp_modes & BIT(px_mode)))
+		return -EINVAL;
+
+	switch (px_mode) {
+	case AMDGPU_SPX_PARTITION_MODE:
+		*num_xcp = 1;
+		*nps_modes = BIT(AMDGPU_NPS1_PARTITION_MODE);
+		break;
+	case AMDGPU_DPX_PARTITION_MODE:
+		*num_xcp = 2;
+		*nps_modes = BIT(AMDGPU_NPS1_PARTITION_MODE) |
+			     BIT(AMDGPU_NPS2_PARTITION_MODE);
+		break;
+	case AMDGPU_TPX_PARTITION_MODE:
+		*num_xcp = 3;
+		*nps_modes = BIT(AMDGPU_NPS1_PARTITION_MODE) |
+			     BIT(AMDGPU_NPS4_PARTITION_MODE);
+		break;
+	case AMDGPU_QPX_PARTITION_MODE:
+		*num_xcp = 4;
+		*nps_modes = BIT(AMDGPU_NPS1_PARTITION_MODE) |
+			     BIT(AMDGPU_NPS4_PARTITION_MODE);
+		if (gc_ver == IP_VERSION(9, 5, 0))
+			*nps_modes |= BIT(AMDGPU_NPS2_PARTITION_MODE);
+		break;
+	case AMDGPU_CPX_PARTITION_MODE:
+		*num_xcp = NUM_XCC(adev->gfx.xcc_mask);
+		*nps_modes = BIT(AMDGPU_NPS1_PARTITION_MODE) |
+			     BIT(AMDGPU_NPS4_PARTITION_MODE);
+		if (gc_ver == IP_VERSION(9, 5, 0))
+			*nps_modes |= BIT(AMDGPU_NPS2_PARTITION_MODE);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int aqua_vanjaram_get_xcp_res_info(struct amdgpu_xcp_mgr *xcp_mgr,
+					  int mode,
+					  struct amdgpu_xcp_cfg *xcp_cfg)
+{
+	struct amdgpu_device *adev = xcp_mgr->adev;
+	int max_res[AMDGPU_XCP_RES_MAX] = {};
+	bool res_lt_xcp;
+	int num_xcp, i, r;
+	u16 nps_modes;
+
+	if (!(xcp_mgr->supp_xcp_modes & BIT(mode)))
+		return -EINVAL;
+
+	max_res[AMDGPU_XCP_RES_XCC] = NUM_XCC(adev->gfx.xcc_mask);
+	max_res[AMDGPU_XCP_RES_DMA] = adev->sdma.num_instances;
+	max_res[AMDGPU_XCP_RES_DEC] = adev->vcn.num_vcn_inst;
+	max_res[AMDGPU_XCP_RES_JPEG] = adev->jpeg.num_jpeg_inst;
+
+	r = __aqua_vanjaram_get_px_mode_info(xcp_mgr, mode, &num_xcp, &nps_modes);
+	if (r)
+		return r;
+
+	xcp_cfg->compatible_nps_modes =
+		(adev->gmc.supported_nps_modes & nps_modes);
+	xcp_cfg->num_res = ARRAY_SIZE(max_res);
+
+	for (i = 0; i < xcp_cfg->num_res; i++) {
+		res_lt_xcp = max_res[i] < num_xcp;
+		xcp_cfg->xcp_res[i].id = i;
+		xcp_cfg->xcp_res[i].num_inst =
+			res_lt_xcp ? 1 : max_res[i] / num_xcp;
+		xcp_cfg->xcp_res[i].num_inst =
+			i == AMDGPU_XCP_RES_JPEG ?
+			xcp_cfg->xcp_res[i].num_inst *
+			adev->jpeg.num_jpeg_rings : xcp_cfg->xcp_res[i].num_inst;
+		xcp_cfg->xcp_res[i].num_shared =
+			res_lt_xcp ? num_xcp / max_res[i] : 1;
+	}
+
+	return 0;
+}
+
+static enum amdgpu_gfx_partition
+__aqua_vanjaram_get_auto_mode(struct amdgpu_xcp_mgr *xcp_mgr)
+{
+	struct amdgpu_device *adev = xcp_mgr->adev;
+	int num_xcc;
+
+	num_xcc = NUM_XCC(xcp_mgr->adev->gfx.xcc_mask);
+
+	if (adev->gmc.num_mem_partitions == 1)
+		return AMDGPU_SPX_PARTITION_MODE;
+
+	if (adev->gmc.num_mem_partitions == num_xcc)
+		return AMDGPU_CPX_PARTITION_MODE;
+
+	if (adev->gmc.num_mem_partitions == num_xcc / 2)
+		return (adev->flags & AMD_IS_APU) ? AMDGPU_TPX_PARTITION_MODE :
+						    AMDGPU_CPX_PARTITION_MODE;
+
+	if (adev->gmc.num_mem_partitions == 2 && !(adev->flags & AMD_IS_APU))
+		return AMDGPU_DPX_PARTITION_MODE;
+
+	return AMDGPU_UNKNOWN_COMPUTE_PARTITION_MODE;
+}
+
+static bool __aqua_vanjaram_is_valid_mode(struct amdgpu_xcp_mgr *xcp_mgr,
+					  enum amdgpu_gfx_partition mode)
+{
+	struct amdgpu_device *adev = xcp_mgr->adev;
+	int num_xcc, num_xccs_per_xcp, r;
+	int num_xcp, nps_mode;
+	u16 supp_nps_modes;
+	bool comp_mode;
+
+	nps_mode = adev->gmc.gmc_funcs->query_mem_partition_mode(adev);
+	r = __aqua_vanjaram_get_px_mode_info(xcp_mgr, mode, &num_xcp,
+					       &supp_nps_modes);
+	if (r)
+		return false;
+
+	comp_mode = !!(BIT(nps_mode) & supp_nps_modes);
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	switch (mode) {
+	case AMDGPU_SPX_PARTITION_MODE:
+		return comp_mode && num_xcc > 0;
+	case AMDGPU_DPX_PARTITION_MODE:
+		return comp_mode && (num_xcc % 4) == 0;
+	case AMDGPU_TPX_PARTITION_MODE:
+		return comp_mode && ((num_xcc % 3) == 0);
+	case AMDGPU_QPX_PARTITION_MODE:
+		num_xccs_per_xcp = num_xcc / 4;
+		return comp_mode && (num_xccs_per_xcp >= 2);
+	case AMDGPU_CPX_PARTITION_MODE:
+		return comp_mode && (num_xcc > 1);
+	default:
+		return false;
+	}
+
+	return false;
+}
+
+static void __aqua_vanjaram_update_available_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr)
+{
+	int mode;
+
+	xcp_mgr->avail_xcp_modes = 0;
+
+	for_each_inst(mode, xcp_mgr->supp_xcp_modes) {
+		if (__aqua_vanjaram_is_valid_mode(xcp_mgr, mode))
+			xcp_mgr->avail_xcp_modes |= BIT(mode);
+	}
+}
+
+static int aqua_vanjaram_switch_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr,
+					       int mode, int *num_xcps)
+{
+	int num_xcc_per_xcp, num_xcc, ret;
+	struct amdgpu_device *adev;
+	u32 flags = 0;
+
+	adev = xcp_mgr->adev;
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+
+	if (mode == AMDGPU_AUTO_COMPUTE_PARTITION_MODE) {
+		mode = __aqua_vanjaram_get_auto_mode(xcp_mgr);
+		if (mode == AMDGPU_UNKNOWN_COMPUTE_PARTITION_MODE) {
+			dev_err(adev->dev,
+				"Invalid config, no compatible compute partition mode found, available memory partitions: %d",
+				adev->gmc.num_mem_partitions);
+			return -EINVAL;
+		}
+	} else if (!__aqua_vanjaram_is_valid_mode(xcp_mgr, mode)) {
+		dev_err(adev->dev,
+			"Invalid compute partition mode requested, requested: %s, available memory partitions: %d",
+			amdgpu_gfx_compute_mode_desc(mode), adev->gmc.num_mem_partitions);
+		return -EINVAL;
+	}
+
+	if (adev->kfd.init_complete && !amdgpu_in_reset(adev))
+		flags |= AMDGPU_XCP_OPS_KFD;
+
+	if (flags & AMDGPU_XCP_OPS_KFD) {
+		ret = amdgpu_amdkfd_check_and_lock_kfd(adev);
+		if (ret)
+			goto out;
+	}
+
+	ret = amdgpu_xcp_pre_partition_switch(xcp_mgr, flags);
+	if (ret)
+		goto unlock;
+
+	num_xcc_per_xcp = __aqua_vanjaram_get_xcc_per_xcp(xcp_mgr, mode);
+	if (adev->gfx.funcs->switch_partition_mode)
+		adev->gfx.funcs->switch_partition_mode(xcp_mgr->adev,
+						       num_xcc_per_xcp);
+
+	/* Init info about new xcps */
+	*num_xcps = num_xcc / num_xcc_per_xcp;
+	amdgpu_xcp_init(xcp_mgr, *num_xcps, mode);
+
+	ret = amdgpu_xcp_post_partition_switch(xcp_mgr, flags);
+	if (!ret)
+		__aqua_vanjaram_update_available_partition_mode(xcp_mgr);
+unlock:
+	if (flags & AMDGPU_XCP_OPS_KFD)
+		amdgpu_amdkfd_unlock_kfd(adev);
+out:
+	return ret;
+}
+
+static int __aqua_vanjaram_get_xcp_mem_id(struct amdgpu_device *adev,
+					  int xcc_id, uint8_t *mem_id)
+{
+	/* memory/spatial modes validation check is already done */
+	*mem_id = xcc_id / adev->gfx.num_xcc_per_xcp;
+	*mem_id /= adev->xcp_mgr->num_xcp_per_mem_partition;
+
+	return 0;
+}
+
+static int aqua_vanjaram_get_xcp_mem_id(struct amdgpu_xcp_mgr *xcp_mgr,
+					struct amdgpu_xcp *xcp, uint8_t *mem_id)
+{
+	struct amdgpu_numa_info numa_info;
+	struct amdgpu_device *adev;
+	uint32_t xcc_mask;
+	int r, i, xcc_id;
+
+	adev = xcp_mgr->adev;
+	/* TODO: BIOS is not returning the right info now
+	 * Check on this later
+	 */
+	/*
+	if (adev->gmc.gmc_funcs->query_mem_partition_mode)
+		mode = adev->gmc.gmc_funcs->query_mem_partition_mode(adev);
+	*/
+	if (adev->gmc.num_mem_partitions == 1) {
+		/* Only one range */
+		*mem_id = 0;
+		return 0;
+	}
+
+	r = amdgpu_xcp_get_inst_details(xcp, AMDGPU_XCP_GFX, &xcc_mask);
+	if (r || !xcc_mask)
+		return -EINVAL;
+
+	xcc_id = ffs(xcc_mask) - 1;
+	if (!adev->gmc.is_app_apu)
+		return __aqua_vanjaram_get_xcp_mem_id(adev, xcc_id, mem_id);
+
+	r = amdgpu_acpi_get_mem_info(adev, xcc_id, &numa_info);
+
+	if (r)
+		return r;
+
+	r = -EINVAL;
+	for (i = 0; i < adev->gmc.num_mem_partitions; ++i) {
+		if (adev->gmc.mem_partitions[i].numa.node == numa_info.nid) {
+			*mem_id = i;
+			r = 0;
+			break;
+		}
+	}
+
+	return r;
+}
+
+static int aqua_vanjaram_get_xcp_ip_details(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id,
+				     enum AMDGPU_XCP_IP_BLOCK ip_id,
+				     struct amdgpu_xcp_ip *ip)
+{
+	if (!ip)
+		return -EINVAL;
+
+	return __aqua_vanjaram_get_xcp_ip_info(xcp_mgr, xcp_id, ip_id, ip);
+}
+
+struct amdgpu_xcp_mgr_funcs aqua_vanjaram_xcp_funcs = {
+	.switch_partition_mode = &aqua_vanjaram_switch_partition_mode,
+	.query_partition_mode = &aqua_vanjaram_query_partition_mode,
+	.get_ip_details = &aqua_vanjaram_get_xcp_ip_details,
+	.get_xcp_res_info = &aqua_vanjaram_get_xcp_res_info,
+	.get_xcp_mem_id = &aqua_vanjaram_get_xcp_mem_id,
+};
+
+static int aqua_vanjaram_xcp_mgr_init(struct amdgpu_device *adev)
+{
+	int ret;
+
+	if (amdgpu_sriov_vf(adev))
+		aqua_vanjaram_xcp_funcs.switch_partition_mode = NULL;
+
+	ret = amdgpu_xcp_mgr_init(adev, AMDGPU_UNKNOWN_COMPUTE_PARTITION_MODE, 1,
+				  &aqua_vanjaram_xcp_funcs);
+	if (ret)
+		return ret;
+
+	amdgpu_xcp_update_supported_modes(adev->xcp_mgr);
+	/* TODO: Default memory node affinity init */
+
+	return ret;
+}
+
+int aqua_vanjaram_init_soc_config(struct amdgpu_device *adev)
+{
+	u32 mask, avail_inst, inst_mask = adev->sdma.sdma_mask;
+	int ret, i;
+
+	/* generally 1 AID supports 4 instances */
+	adev->sdma.num_inst_per_aid = 4;
+	adev->sdma.num_instances = NUM_SDMA(adev->sdma.sdma_mask);
+
+	adev->aid_mask = i = 1;
+	inst_mask >>= adev->sdma.num_inst_per_aid;
+
+	for (mask = (1 << adev->sdma.num_inst_per_aid) - 1; inst_mask;
+	     inst_mask >>= adev->sdma.num_inst_per_aid, ++i) {
+		avail_inst = inst_mask & mask;
+		if (avail_inst == mask || avail_inst == 0x3 ||
+		    avail_inst == 0xc)
+			adev->aid_mask |= (1 << i);
+	}
+
+	/* Harvest config is not used for aqua vanjaram. VCN and JPEGs will be
+	 * addressed based on logical instance ids.
+	 */
+	adev->vcn.harvest_config = 0;
+	adev->vcn.num_inst_per_aid = 1;
+	adev->vcn.num_vcn_inst = hweight32(adev->vcn.inst_mask);
+	adev->jpeg.harvest_config = 0;
+	adev->jpeg.num_inst_per_aid = 1;
+	adev->jpeg.num_jpeg_inst = hweight32(adev->jpeg.inst_mask);
+
+	ret = aqua_vanjaram_xcp_mgr_init(adev);
+	if (ret)
+		return ret;
+
+	amdgpu_ip_map_init(adev);
+
+	return 0;
+}
+
+static void aqua_read_smn(struct amdgpu_device *adev,
+			  struct amdgpu_smn_reg_data *regdata,
+			  uint64_t smn_addr)
+{
+	regdata->addr = smn_addr;
+	regdata->value = RREG32_PCIE(smn_addr);
+}
+
+struct aqua_reg_list {
+	uint64_t start_addr;
+	uint32_t num_regs;
+	uint32_t incrx;
+};
+
+#define DW_ADDR_INCR	4
+
+static void aqua_read_smn_ext(struct amdgpu_device *adev,
+			      struct amdgpu_smn_reg_data *regdata,
+			      uint64_t smn_addr, int i)
+{
+	regdata->addr =
+		smn_addr + adev->asic_funcs->encode_ext_smn_addressing(i);
+	regdata->value = RREG32_PCIE_EXT(regdata->addr);
+}
+
+#define smnreg_0x1A340218	0x1A340218
+#define smnreg_0x1A3402E4	0x1A3402E4
+#define smnreg_0x1A340294	0x1A340294
+#define smreg_0x1A380088	0x1A380088
+
+#define NUM_PCIE_SMN_REGS	14
+
+static struct aqua_reg_list pcie_reg_addrs[] = {
+	{ smnreg_0x1A340218, 1, 0 },
+	{ smnreg_0x1A3402E4, 1, 0 },
+	{ smnreg_0x1A340294, 6, DW_ADDR_INCR },
+	{ smreg_0x1A380088, 6, DW_ADDR_INCR },
+};
+
+static ssize_t aqua_vanjaram_read_pcie_state(struct amdgpu_device *adev,
+					     void *buf, size_t max_size)
+{
+	struct amdgpu_reg_state_pcie_v1_0 *pcie_reg_state;
+	uint32_t start_addr, incrx, num_regs, szbuf;
+	struct amdgpu_regs_pcie_v1_0 *pcie_regs;
+	struct amdgpu_smn_reg_data *reg_data;
+	struct pci_dev *us_pdev, *ds_pdev;
+	int aer_cap, r, n;
+
+	if (!buf || !max_size)
+		return -EINVAL;
+
+	pcie_reg_state = (struct amdgpu_reg_state_pcie_v1_0 *)buf;
+
+	szbuf = sizeof(*pcie_reg_state) +
+		amdgpu_reginst_size(1, sizeof(*pcie_regs), NUM_PCIE_SMN_REGS);
+	/* Only one instance of pcie regs */
+	if (max_size < szbuf)
+		return -EOVERFLOW;
+
+	pcie_regs = (struct amdgpu_regs_pcie_v1_0 *)((uint8_t *)buf +
+						     sizeof(*pcie_reg_state));
+	pcie_regs->inst_header.instance = 0;
+	pcie_regs->inst_header.state = AMDGPU_INST_S_OK;
+	pcie_regs->inst_header.num_smn_regs = NUM_PCIE_SMN_REGS;
+
+	reg_data = pcie_regs->smn_reg_values;
+
+	for (r = 0; r < ARRAY_SIZE(pcie_reg_addrs); r++) {
+		start_addr = pcie_reg_addrs[r].start_addr;
+		incrx = pcie_reg_addrs[r].incrx;
+		num_regs = pcie_reg_addrs[r].num_regs;
+		for (n = 0; n < num_regs; n++) {
+			aqua_read_smn(adev, reg_data, start_addr + n * incrx);
+			++reg_data;
+		}
+	}
+
+	ds_pdev = pci_upstream_bridge(adev->pdev);
+	us_pdev = pci_upstream_bridge(ds_pdev);
+
+	pcie_capability_read_word(us_pdev, PCI_EXP_DEVSTA,
+				  &pcie_regs->device_status);
+	pcie_capability_read_word(us_pdev, PCI_EXP_LNKSTA,
+				  &pcie_regs->link_status);
+
+	aer_cap = pci_find_ext_capability(us_pdev, PCI_EXT_CAP_ID_ERR);
+	if (aer_cap) {
+		pci_read_config_dword(us_pdev, aer_cap + PCI_ERR_COR_STATUS,
+				      &pcie_regs->pcie_corr_err_status);
+		pci_read_config_dword(us_pdev, aer_cap + PCI_ERR_UNCOR_STATUS,
+				      &pcie_regs->pcie_uncorr_err_status);
+	}
+
+	pci_read_config_dword(us_pdev, PCI_PRIMARY_BUS,
+			      &pcie_regs->sub_bus_number_latency);
+
+	pcie_reg_state->common_header.structure_size = szbuf;
+	pcie_reg_state->common_header.format_revision = 1;
+	pcie_reg_state->common_header.content_revision = 0;
+	pcie_reg_state->common_header.state_type = AMDGPU_REG_STATE_TYPE_PCIE;
+	pcie_reg_state->common_header.num_instances = 1;
+
+	return pcie_reg_state->common_header.structure_size;
+}
+
+#define smnreg_0x11A00050	0x11A00050
+#define smnreg_0x11A00180	0x11A00180
+#define smnreg_0x11A00070	0x11A00070
+#define smnreg_0x11A00200	0x11A00200
+#define smnreg_0x11A0020C	0x11A0020C
+#define smnreg_0x11A00210	0x11A00210
+#define smnreg_0x11A00108	0x11A00108
+
+#define XGMI_LINK_REG(smnreg, l) ((smnreg) | (l << 20))
+
+#define NUM_XGMI_SMN_REGS 25
+
+static struct aqua_reg_list xgmi_reg_addrs[] = {
+	{ smnreg_0x11A00050, 1, 0 },
+	{ smnreg_0x11A00180, 16, DW_ADDR_INCR },
+	{ smnreg_0x11A00070, 4, DW_ADDR_INCR },
+	{ smnreg_0x11A00200, 1, 0 },
+	{ smnreg_0x11A0020C, 1, 0 },
+	{ smnreg_0x11A00210, 1, 0 },
+	{ smnreg_0x11A00108, 1, 0 },
+};
+
+static ssize_t aqua_vanjaram_read_xgmi_state(struct amdgpu_device *adev,
+					     void *buf, size_t max_size)
+{
+	struct amdgpu_reg_state_xgmi_v1_0 *xgmi_reg_state;
+	uint32_t start_addr, incrx, num_regs, szbuf;
+	struct amdgpu_regs_xgmi_v1_0 *xgmi_regs;
+	struct amdgpu_smn_reg_data *reg_data;
+	const int max_xgmi_instances = 8;
+	int inst = 0, i, j, r, n;
+	const int xgmi_inst = 2;
+	void *p;
+
+	if (!buf || !max_size)
+		return -EINVAL;
+
+	xgmi_reg_state = (struct amdgpu_reg_state_xgmi_v1_0 *)buf;
+
+	szbuf = sizeof(*xgmi_reg_state) +
+		amdgpu_reginst_size(max_xgmi_instances, sizeof(*xgmi_regs),
+				    NUM_XGMI_SMN_REGS);
+	/* Only one instance of pcie regs */
+	if (max_size < szbuf)
+		return -EOVERFLOW;
+
+	p = &xgmi_reg_state->xgmi_state_regs[0];
+	for_each_inst(i, adev->aid_mask) {
+		for (j = 0; j < xgmi_inst; ++j) {
+			xgmi_regs = (struct amdgpu_regs_xgmi_v1_0 *)p;
+			xgmi_regs->inst_header.instance = inst++;
+
+			xgmi_regs->inst_header.state = AMDGPU_INST_S_OK;
+			xgmi_regs->inst_header.num_smn_regs = NUM_XGMI_SMN_REGS;
+
+			reg_data = xgmi_regs->smn_reg_values;
+
+			for (r = 0; r < ARRAY_SIZE(xgmi_reg_addrs); r++) {
+				start_addr = xgmi_reg_addrs[r].start_addr;
+				incrx = xgmi_reg_addrs[r].incrx;
+				num_regs = xgmi_reg_addrs[r].num_regs;
+
+				for (n = 0; n < num_regs; n++) {
+					aqua_read_smn_ext(
+						adev, reg_data,
+						XGMI_LINK_REG(start_addr, j) +
+							n * incrx,
+						i);
+					++reg_data;
+				}
+			}
+			p = reg_data;
+		}
+	}
+
+	xgmi_reg_state->common_header.structure_size = szbuf;
+	xgmi_reg_state->common_header.format_revision = 1;
+	xgmi_reg_state->common_header.content_revision = 0;
+	xgmi_reg_state->common_header.state_type = AMDGPU_REG_STATE_TYPE_XGMI;
+	xgmi_reg_state->common_header.num_instances = max_xgmi_instances;
+
+	return xgmi_reg_state->common_header.structure_size;
+}
+
+#define smnreg_0x11C00070	0x11C00070
+#define smnreg_0x11C00210	0x11C00210
+
+static struct aqua_reg_list wafl_reg_addrs[] = {
+	{ smnreg_0x11C00070, 4, DW_ADDR_INCR },
+	{ smnreg_0x11C00210, 1, 0 },
+};
+
+#define WAFL_LINK_REG(smnreg, l) ((smnreg) | (l << 20))
+
+#define NUM_WAFL_SMN_REGS 5
+
+static ssize_t aqua_vanjaram_read_wafl_state(struct amdgpu_device *adev,
+					     void *buf, size_t max_size)
+{
+	struct amdgpu_reg_state_wafl_v1_0 *wafl_reg_state;
+	uint32_t start_addr, incrx, num_regs, szbuf;
+	struct amdgpu_regs_wafl_v1_0 *wafl_regs;
+	struct amdgpu_smn_reg_data *reg_data;
+	const int max_wafl_instances = 8;
+	int inst = 0, i, j, r, n;
+	const int wafl_inst = 2;
+	void *p;
+
+	if (!buf || !max_size)
+		return -EINVAL;
+
+	wafl_reg_state = (struct amdgpu_reg_state_wafl_v1_0 *)buf;
+
+	szbuf = sizeof(*wafl_reg_state) +
+		amdgpu_reginst_size(max_wafl_instances, sizeof(*wafl_regs),
+				    NUM_WAFL_SMN_REGS);
+
+	if (max_size < szbuf)
+		return -EOVERFLOW;
+
+	p = &wafl_reg_state->wafl_state_regs[0];
+	for_each_inst(i, adev->aid_mask) {
+		for (j = 0; j < wafl_inst; ++j) {
+			wafl_regs = (struct amdgpu_regs_wafl_v1_0 *)p;
+			wafl_regs->inst_header.instance = inst++;
+
+			wafl_regs->inst_header.state = AMDGPU_INST_S_OK;
+			wafl_regs->inst_header.num_smn_regs = NUM_WAFL_SMN_REGS;
+
+			reg_data = wafl_regs->smn_reg_values;
+
+			for (r = 0; r < ARRAY_SIZE(wafl_reg_addrs); r++) {
+				start_addr = wafl_reg_addrs[r].start_addr;
+				incrx = wafl_reg_addrs[r].incrx;
+				num_regs = wafl_reg_addrs[r].num_regs;
+				for (n = 0; n < num_regs; n++) {
+					aqua_read_smn_ext(
+						adev, reg_data,
+						WAFL_LINK_REG(start_addr, j) +
+							n * incrx,
+						i);
+					++reg_data;
+				}
+			}
+			p = reg_data;
+		}
+	}
+
+	wafl_reg_state->common_header.structure_size = szbuf;
+	wafl_reg_state->common_header.format_revision = 1;
+	wafl_reg_state->common_header.content_revision = 0;
+	wafl_reg_state->common_header.state_type = AMDGPU_REG_STATE_TYPE_WAFL;
+	wafl_reg_state->common_header.num_instances = max_wafl_instances;
+
+	return wafl_reg_state->common_header.structure_size;
+}
+
+#define smnreg_0x1B311060 0x1B311060
+#define smnreg_0x1B411060 0x1B411060
+#define smnreg_0x1B511060 0x1B511060
+#define smnreg_0x1B611060 0x1B611060
+
+#define smnreg_0x1C307120 0x1C307120
+#define smnreg_0x1C317120 0x1C317120
+
+#define smnreg_0x1C320830 0x1C320830
+#define smnreg_0x1C380830 0x1C380830
+#define smnreg_0x1C3D0830 0x1C3D0830
+#define smnreg_0x1C420830 0x1C420830
+
+#define smnreg_0x1C320100 0x1C320100
+#define smnreg_0x1C380100 0x1C380100
+#define smnreg_0x1C3D0100 0x1C3D0100
+#define smnreg_0x1C420100 0x1C420100
+
+#define smnreg_0x1B310500 0x1B310500
+#define smnreg_0x1C300400 0x1C300400
+
+#define USR_CAKE_INCR 0x11000
+#define USR_LINK_INCR 0x100000
+#define USR_CP_INCR 0x10000
+
+#define NUM_USR_SMN_REGS	20
+
+struct aqua_reg_list usr_reg_addrs[] = {
+	{ smnreg_0x1B311060, 4, DW_ADDR_INCR },
+	{ smnreg_0x1B411060, 4, DW_ADDR_INCR },
+	{ smnreg_0x1B511060, 4, DW_ADDR_INCR },
+	{ smnreg_0x1B611060, 4, DW_ADDR_INCR },
+	{ smnreg_0x1C307120, 2, DW_ADDR_INCR },
+	{ smnreg_0x1C317120, 2, DW_ADDR_INCR },
+};
+
+#define NUM_USR1_SMN_REGS	46
+struct aqua_reg_list usr1_reg_addrs[] = {
+	{ smnreg_0x1C320830, 6, USR_CAKE_INCR },
+	{ smnreg_0x1C380830, 5, USR_CAKE_INCR },
+	{ smnreg_0x1C3D0830, 5, USR_CAKE_INCR },
+	{ smnreg_0x1C420830, 4, USR_CAKE_INCR },
+	{ smnreg_0x1C320100, 6, USR_CAKE_INCR },
+	{ smnreg_0x1C380100, 5, USR_CAKE_INCR },
+	{ smnreg_0x1C3D0100, 5, USR_CAKE_INCR },
+	{ smnreg_0x1C420100, 4, USR_CAKE_INCR },
+	{ smnreg_0x1B310500, 4, USR_LINK_INCR },
+	{ smnreg_0x1C300400, 2, USR_CP_INCR },
+};
+
+static ssize_t aqua_vanjaram_read_usr_state(struct amdgpu_device *adev,
+					    void *buf, size_t max_size,
+					    int reg_state)
+{
+	uint32_t start_addr, incrx, num_regs, szbuf, num_smn;
+	struct amdgpu_reg_state_usr_v1_0 *usr_reg_state;
+	struct amdgpu_regs_usr_v1_0 *usr_regs;
+	struct amdgpu_smn_reg_data *reg_data;
+	const int max_usr_instances = 4;
+	struct aqua_reg_list *reg_addrs;
+	int inst = 0, i, n, r, arr_size;
+	void *p;
+
+	if (!buf || !max_size)
+		return -EINVAL;
+
+	switch (reg_state) {
+	case AMDGPU_REG_STATE_TYPE_USR:
+		arr_size = ARRAY_SIZE(usr_reg_addrs);
+		reg_addrs = usr_reg_addrs;
+		num_smn = NUM_USR_SMN_REGS;
+		break;
+	case AMDGPU_REG_STATE_TYPE_USR_1:
+		arr_size = ARRAY_SIZE(usr1_reg_addrs);
+		reg_addrs = usr1_reg_addrs;
+		num_smn = NUM_USR1_SMN_REGS;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	usr_reg_state = (struct amdgpu_reg_state_usr_v1_0 *)buf;
+
+	szbuf = sizeof(*usr_reg_state) + amdgpu_reginst_size(max_usr_instances,
+							     sizeof(*usr_regs),
+							     num_smn);
+	if (max_size < szbuf)
+		return -EOVERFLOW;
+
+	p = &usr_reg_state->usr_state_regs[0];
+	for_each_inst(i, adev->aid_mask) {
+		usr_regs = (struct amdgpu_regs_usr_v1_0 *)p;
+		usr_regs->inst_header.instance = inst++;
+		usr_regs->inst_header.state = AMDGPU_INST_S_OK;
+		usr_regs->inst_header.num_smn_regs = num_smn;
+		reg_data = usr_regs->smn_reg_values;
+
+		for (r = 0; r < arr_size; r++) {
+			start_addr = reg_addrs[r].start_addr;
+			incrx = reg_addrs[r].incrx;
+			num_regs = reg_addrs[r].num_regs;
+			for (n = 0; n < num_regs; n++) {
+				aqua_read_smn_ext(adev, reg_data,
+						  start_addr + n * incrx, i);
+				reg_data++;
+			}
+		}
+		p = reg_data;
+	}
+
+	usr_reg_state->common_header.structure_size = szbuf;
+	usr_reg_state->common_header.format_revision = 1;
+	usr_reg_state->common_header.content_revision = 0;
+	usr_reg_state->common_header.state_type = AMDGPU_REG_STATE_TYPE_USR;
+	usr_reg_state->common_header.num_instances = max_usr_instances;
+
+	return usr_reg_state->common_header.structure_size;
+}
+
+ssize_t aqua_vanjaram_get_reg_state(struct amdgpu_device *adev,
+				    enum amdgpu_reg_state reg_state, void *buf,
+				    size_t max_size)
+{
+	ssize_t size;
+
+	switch (reg_state) {
+	case AMDGPU_REG_STATE_TYPE_PCIE:
+		size = aqua_vanjaram_read_pcie_state(adev, buf, max_size);
+		break;
+	case AMDGPU_REG_STATE_TYPE_XGMI:
+		size = aqua_vanjaram_read_xgmi_state(adev, buf, max_size);
+		break;
+	case AMDGPU_REG_STATE_TYPE_WAFL:
+		size = aqua_vanjaram_read_wafl_state(adev, buf, max_size);
+		break;
+	case AMDGPU_REG_STATE_TYPE_USR:
+		size = aqua_vanjaram_read_usr_state(adev, buf, max_size,
+						    AMDGPU_REG_STATE_TYPE_USR);
+		break;
+	case AMDGPU_REG_STATE_TYPE_USR_1:
+		size = aqua_vanjaram_read_usr_state(
+			adev, buf, max_size, AMDGPU_REG_STATE_TYPE_USR_1);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return size;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/arct_reg_init.c b/drivers/gpu/drm/amd/amdgpu/arct_reg_init.c
new file mode 100644
index 000000000000..fda99c958c3b
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/arct_reg_init.c
@@ -0,0 +1,60 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "soc15.h"
+
+#include "soc15_common.h"
+#include "arct_ip_offset.h"
+
+int arct_reg_base_init(struct amdgpu_device *adev)
+{
+	/* HW has more IP blocks,  only initialized the block needed by our driver  */
+	uint32_t i;
+	for (i = 0 ; i < MAX_INSTANCE ; ++i) {
+		adev->reg_offset[GC_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i]));
+		adev->reg_offset[HDP_HWIP][i] = (uint32_t *)(&(HDP_BASE.instance[i]));
+		adev->reg_offset[MMHUB_HWIP][i] = (uint32_t *)(&(MMHUB_BASE.instance[i]));
+		adev->reg_offset[ATHUB_HWIP][i] = (uint32_t *)(&(ATHUB_BASE.instance[i]));
+		adev->reg_offset[NBIO_HWIP][i] = (uint32_t *)(&(NBIF0_BASE.instance[i]));
+		adev->reg_offset[MP0_HWIP][i] = (uint32_t *)(&(MP0_BASE.instance[i]));
+		adev->reg_offset[MP1_HWIP][i] = (uint32_t *)(&(MP1_BASE.instance[i]));
+		adev->reg_offset[UVD_HWIP][i] = (uint32_t *)(&(UVD_BASE.instance[i]));
+		adev->reg_offset[DF_HWIP][i] = (uint32_t *)(&(DF_BASE.instance[i]));
+		adev->reg_offset[OSSSYS_HWIP][i] = (uint32_t *)(&(OSSSYS_BASE.instance[i]));
+		adev->reg_offset[SDMA0_HWIP][i] = (uint32_t *)(&(SDMA0_BASE.instance[i]));
+		adev->reg_offset[SDMA1_HWIP][i] = (uint32_t *)(&(SDMA1_BASE.instance[i]));
+		adev->reg_offset[SDMA2_HWIP][i] = (uint32_t *)(&(SDMA2_BASE.instance[i]));
+		adev->reg_offset[SDMA3_HWIP][i] = (uint32_t *)(&(SDMA3_BASE.instance[i]));
+		adev->reg_offset[SDMA4_HWIP][i] = (uint32_t *)(&(SDMA4_BASE.instance[i]));
+		adev->reg_offset[SDMA5_HWIP][i] = (uint32_t *)(&(SDMA5_BASE.instance[i]));
+		adev->reg_offset[SDMA6_HWIP][i] = (uint32_t *)(&(SDMA6_BASE.instance[i]));
+		adev->reg_offset[SDMA7_HWIP][i] = (uint32_t *)(&(SDMA7_BASE.instance[i]));
+		adev->reg_offset[SMUIO_HWIP][i] = (uint32_t *)(&(SMUIO_BASE.instance[i]));
+		adev->reg_offset[THM_HWIP][i] = (uint32_t *)(&(THM_BASE.instance[i]));
+		adev->reg_offset[UMC_HWIP][i] = (uint32_t *)(&(UMC_BASE.instance[i]));
+		adev->reg_offset[RSMU_HWIP][i] = (uint32_t *)(&(RSMU_BASE.instance[i]));
+	}
+	return 0;
+}
+
+
diff --git a/drivers/gpu/drm/amd/amdgpu/athub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/athub_v1_0.c
new file mode 100644
index 000000000000..42f4e163e251
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/athub_v1_0.c
@@ -0,0 +1,105 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "athub_v1_0.h"
+
+#include "athub/athub_1_0_offset.h"
+#include "athub/athub_1_0_sh_mask.h"
+#include "vega10_enum.h"
+
+#include "soc15_common.h"
+
+static void athub_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						   bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG))
+		data |= ATHUB_MISC_CNTL__CG_ENABLE_MASK;
+	else
+		data &= ~ATHUB_MISC_CNTL__CG_ENABLE_MASK;
+
+	if (def != data)
+		WREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL, data);
+}
+
+static void athub_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+						  bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_LS) &&
+	    (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS))
+		data |= ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK;
+	else
+		data &= ~ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK;
+
+	if(def != data)
+		WREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL, data);
+}
+
+int athub_v1_0_set_clockgating(struct amdgpu_device *adev,
+			       enum amd_clockgating_state state)
+{
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (amdgpu_ip_version(adev, ATHUB_HWIP, 0)) {
+	case IP_VERSION(9, 0, 0):
+	case IP_VERSION(9, 1, 0):
+	case IP_VERSION(9, 2, 0):
+	case IP_VERSION(9, 3, 0):
+	case IP_VERSION(9, 4, 0):
+	case IP_VERSION(1, 5, 0):
+		athub_update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		athub_update_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+void athub_v1_0_get_clockgating(struct amdgpu_device *adev, u64 *flags)
+{
+	int data;
+
+	if (amdgpu_sriov_vf(adev))
+		*flags = 0;
+
+	/* AMD_CG_SUPPORT_ATHUB_MGCG */
+	data = RREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL);
+	if (data & ATHUB_MISC_CNTL__CG_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_ATHUB_MGCG;
+
+	/* AMD_CG_SUPPORT_ATHUB_LS */
+	if (data & ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_ATHUB_LS;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/athub_v1_0.h b/drivers/gpu/drm/amd/amdgpu/athub_v1_0.h
new file mode 100644
index 000000000000..6be0a6704ea7
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/athub_v1_0.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __ATHUB_V1_0_H__
+#define __ATHUB_V1_0_H__
+
+int athub_v1_0_set_clockgating(struct amdgpu_device *adev,
+			       enum amd_clockgating_state state);
+void athub_v1_0_get_clockgating(struct amdgpu_device *adev, u64 *flags);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/athub_v2_0.c b/drivers/gpu/drm/amd/amdgpu/athub_v2_0.c
new file mode 100644
index 000000000000..5a122f50a6e7
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/athub_v2_0.c
@@ -0,0 +1,108 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "athub_v2_0.h"
+
+#include "athub/athub_2_0_0_offset.h"
+#include "athub/athub_2_0_0_sh_mask.h"
+#include "athub/athub_2_0_0_default.h"
+
+#include "soc15_common.h"
+
+static void
+athub_v2_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+					    bool enable)
+{
+	uint32_t def, data;
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG))
+		return;
+
+	def = data = RREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL);
+
+	if (enable)
+		data |= ATHUB_MISC_CNTL__CG_ENABLE_MASK;
+	else
+		data &= ~ATHUB_MISC_CNTL__CG_ENABLE_MASK;
+
+	if (def != data)
+		WREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL, data);
+}
+
+static void
+athub_v2_0_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+					   bool enable)
+{
+	uint32_t def, data;
+
+	if (!((adev->cg_flags & AMD_CG_SUPPORT_MC_LS) &&
+	       (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS)))
+		return;
+
+	def = data = RREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL);
+
+	if (enable)
+		data |= ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK;
+	else
+		data &= ~ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK;
+
+	if (def != data)
+		WREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL, data);
+}
+
+int athub_v2_0_set_clockgating(struct amdgpu_device *adev,
+			       enum amd_clockgating_state state)
+{
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (amdgpu_ip_version(adev, ATHUB_HWIP, 0)) {
+	case IP_VERSION(1, 3, 1):
+	case IP_VERSION(2, 0, 0):
+	case IP_VERSION(2, 0, 2):
+		athub_v2_0_update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		athub_v2_0_update_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+void athub_v2_0_get_clockgating(struct amdgpu_device *adev, u64 *flags)
+{
+	int data;
+
+	/* AMD_CG_SUPPORT_ATHUB_MGCG */
+	data = RREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL);
+	if (data & ATHUB_MISC_CNTL__CG_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_ATHUB_MGCG;
+
+	/* AMD_CG_SUPPORT_ATHUB_LS */
+	if (data & ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_ATHUB_LS;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/athub_v2_0.h b/drivers/gpu/drm/amd/amdgpu/athub_v2_0.h
new file mode 100644
index 000000000000..8b763f6dfd81
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/athub_v2_0.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __ATHUB_V2_0_H__
+#define __ATHUB_V2_0_H__
+
+int athub_v2_0_set_clockgating(struct amdgpu_device *adev,
+			       enum amd_clockgating_state state);
+void athub_v2_0_get_clockgating(struct amdgpu_device *adev, u64 *flags);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/athub_v2_1.c b/drivers/gpu/drm/amd/amdgpu/athub_v2_1.c
new file mode 100644
index 000000000000..e143fcc46148
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/athub_v2_1.c
@@ -0,0 +1,100 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "athub_v2_1.h"
+
+#include "athub/athub_2_1_0_offset.h"
+#include "athub/athub_2_1_0_sh_mask.h"
+
+#include "soc15_common.h"
+
+static void
+athub_v2_1_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+					    bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG))
+		data |= ATHUB_MISC_CNTL__CG_ENABLE_MASK;
+	else
+		data &= ~ATHUB_MISC_CNTL__CG_ENABLE_MASK;
+
+	if (def != data)
+		WREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL, data);
+}
+
+static void
+athub_v2_1_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+					   bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_LS) &&
+	    (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS))
+		data |= ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK;
+	else
+		data &= ~ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK;
+
+	if(def != data)
+		WREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL, data);
+}
+
+int athub_v2_1_set_clockgating(struct amdgpu_device *adev,
+			       enum amd_clockgating_state state)
+{
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (amdgpu_ip_version(adev, ATHUB_HWIP, 0)) {
+	case IP_VERSION(2, 1, 0):
+	case IP_VERSION(2, 1, 1):
+	case IP_VERSION(2, 1, 2):
+	case IP_VERSION(2, 4, 0):
+		athub_v2_1_update_medium_grain_clock_gating(adev, state == AMD_CG_STATE_GATE);
+		athub_v2_1_update_medium_grain_light_sleep(adev, state == AMD_CG_STATE_GATE);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+void athub_v2_1_get_clockgating(struct amdgpu_device *adev, u64 *flags)
+{
+	int data;
+
+	/* AMD_CG_SUPPORT_ATHUB_MGCG */
+	data = RREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL);
+	if (data & ATHUB_MISC_CNTL__CG_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_ATHUB_MGCG;
+
+	/* AMD_CG_SUPPORT_ATHUB_LS */
+	if (data & ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_ATHUB_LS;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/athub_v2_1.h b/drivers/gpu/drm/amd/amdgpu/athub_v2_1.h
new file mode 100644
index 000000000000..b799f14bce03
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/athub_v2_1.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __ATHUB_V2_1_H__
+#define __ATHUB_V2_1_H__
+
+int athub_v2_1_set_clockgating(struct amdgpu_device *adev,
+			       enum amd_clockgating_state state);
+void athub_v2_1_get_clockgating(struct amdgpu_device *adev, u64 *flags);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/athub_v3_0.c b/drivers/gpu/drm/amd/amdgpu/athub_v3_0.c
new file mode 100644
index 000000000000..d1bba9c64e16
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/athub_v3_0.c
@@ -0,0 +1,139 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "athub_v3_0.h"
+#include "athub/athub_3_0_0_offset.h"
+#include "athub/athub_3_0_0_sh_mask.h"
+#include "navi10_enum.h"
+#include "soc15_common.h"
+
+#define regATHUB_MISC_CNTL_V3_0_1			0x00d7
+#define regATHUB_MISC_CNTL_V3_0_1_BASE_IDX		0
+#define regATHUB_MISC_CNTL_V3_3_0			0x00d8
+#define regATHUB_MISC_CNTL_V3_3_0_BASE_IDX		0
+
+
+static uint32_t athub_v3_0_get_cg_cntl(struct amdgpu_device *adev)
+{
+	uint32_t data;
+
+	switch (amdgpu_ip_version(adev, ATHUB_HWIP, 0)) {
+	case IP_VERSION(3, 0, 1):
+		data = RREG32_SOC15(ATHUB, 0, regATHUB_MISC_CNTL_V3_0_1);
+		break;
+	case IP_VERSION(3, 3, 0):
+		data = RREG32_SOC15(ATHUB, 0, regATHUB_MISC_CNTL_V3_3_0);
+		break;
+	default:
+		data = RREG32_SOC15(ATHUB, 0, regATHUB_MISC_CNTL);
+		break;
+	}
+	return data;
+}
+
+static void athub_v3_0_set_cg_cntl(struct amdgpu_device *adev, uint32_t data)
+{
+	switch (amdgpu_ip_version(adev, ATHUB_HWIP, 0)) {
+	case IP_VERSION(3, 0, 1):
+		WREG32_SOC15(ATHUB, 0, regATHUB_MISC_CNTL_V3_0_1, data);
+		break;
+	case IP_VERSION(3, 3, 0):
+		WREG32_SOC15(ATHUB, 0, regATHUB_MISC_CNTL_V3_3_0, data);
+		break;
+	default:
+		WREG32_SOC15(ATHUB, 0, regATHUB_MISC_CNTL, data);
+		break;
+	}
+}
+
+static void
+athub_v3_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+					    bool enable)
+{
+	uint32_t def, data;
+
+	def = data = athub_v3_0_get_cg_cntl(adev);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_ATHUB_MGCG))
+		data |= ATHUB_MISC_CNTL__CG_ENABLE_MASK;
+	else
+		data &= ~ATHUB_MISC_CNTL__CG_ENABLE_MASK;
+
+	if (def != data)
+		athub_v3_0_set_cg_cntl(adev, data);
+}
+
+static void
+athub_v3_0_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+					   bool enable)
+{
+	uint32_t def, data;
+
+	def = data = athub_v3_0_get_cg_cntl(adev);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_ATHUB_LS))
+		data |= ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK;
+	else
+		data &= ~ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK;
+
+	if (def != data)
+		athub_v3_0_set_cg_cntl(adev, data);
+}
+
+int athub_v3_0_set_clockgating(struct amdgpu_device *adev,
+			       enum amd_clockgating_state state)
+{
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (amdgpu_ip_version(adev, ATHUB_HWIP, 0)) {
+	case IP_VERSION(3, 0, 0):
+	case IP_VERSION(3, 0, 1):
+	case IP_VERSION(3, 0, 2):
+	case IP_VERSION(3, 3, 0):
+		athub_v3_0_update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		athub_v3_0_update_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+void athub_v3_0_get_clockgating(struct amdgpu_device *adev, u64 *flags)
+{
+	int data;
+
+	/* AMD_CG_SUPPORT_ATHUB_MGCG */
+	data = athub_v3_0_get_cg_cntl(adev);
+	if (data & ATHUB_MISC_CNTL__CG_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_ATHUB_MGCG;
+
+	/* AMD_CG_SUPPORT_ATHUB_LS */
+	if (data & ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_ATHUB_LS;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/athub_v3_0.h b/drivers/gpu/drm/amd/amdgpu/athub_v3_0.h
new file mode 100644
index 000000000000..e08a7d564365
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/athub_v3_0.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __ATHUB_V3_0_H__
+#define __ATHUB_V3_0_H__
+
+int athub_v3_0_set_clockgating(struct amdgpu_device *adev,
+			       enum amd_clockgating_state state);
+void athub_v3_0_get_clockgating(struct amdgpu_device *adev, u64 *flags);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/athub_v4_1_0.c b/drivers/gpu/drm/amd/amdgpu/athub_v4_1_0.c
new file mode 100644
index 000000000000..8a0773b80864
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/athub_v4_1_0.c
@@ -0,0 +1,122 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "athub_v4_1_0.h"
+#include "athub/athub_4_1_0_offset.h"
+#include "athub/athub_4_1_0_sh_mask.h"
+#include "soc15_common.h"
+
+static uint32_t athub_v4_1_0_get_cg_cntl(struct amdgpu_device *adev)
+{
+	uint32_t data;
+
+	switch (amdgpu_ip_version(adev, ATHUB_HWIP, 0)) {
+	case IP_VERSION(4, 1, 0):
+		data = RREG32_SOC15(ATHUB, 0, regATHUB_MISC_CNTL);
+		break;
+	default:
+		data = 0;
+		break;
+	}
+	return data;
+}
+
+static void athub_v4_1_0_set_cg_cntl(struct amdgpu_device *adev, uint32_t data)
+{
+	switch (amdgpu_ip_version(adev, ATHUB_HWIP, 0)) {
+	case IP_VERSION(4, 1, 0):
+		WREG32_SOC15(ATHUB, 0, regATHUB_MISC_CNTL, data);
+		break;
+	default:
+		break;
+	}
+}
+
+static void
+athub_v4_1_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+					      bool enable)
+{
+	uint32_t def, data;
+
+	def = data = athub_v4_1_0_get_cg_cntl(adev);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_ATHUB_MGCG))
+		data |= ATHUB_MISC_CNTL__CG_ENABLE_MASK;
+	else
+		data &= ~ATHUB_MISC_CNTL__CG_ENABLE_MASK;
+
+	if (def != data)
+		athub_v4_1_0_set_cg_cntl(adev, data);
+}
+
+static void
+athub_v4_1_0_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+					     bool enable)
+{
+	uint32_t def, data;
+
+	def = data = athub_v4_1_0_get_cg_cntl(adev);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_ATHUB_LS))
+		data |= ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK;
+	else
+		data &= ~ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK;
+
+	if (def != data)
+		athub_v4_1_0_set_cg_cntl(adev, data);
+}
+
+int athub_v4_1_0_set_clockgating(struct amdgpu_device *adev,
+				 enum amd_clockgating_state state)
+{
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (amdgpu_ip_version(adev, ATHUB_HWIP, 0)) {
+	case IP_VERSION(4, 1, 0):
+		athub_v4_1_0_update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		athub_v4_1_0_update_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+void athub_v4_1_0_get_clockgating(struct amdgpu_device *adev, u64 *flags)
+{
+	int data;
+
+	/* AMD_CG_SUPPORT_ATHUB_MGCG */
+	data = athub_v4_1_0_get_cg_cntl(adev);
+	if (data & ATHUB_MISC_CNTL__CG_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_ATHUB_MGCG;
+
+	/* AMD_CG_SUPPORT_ATHUB_LS */
+	if (data & ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_ATHUB_LS;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/athub_v4_1_0.h b/drivers/gpu/drm/amd/amdgpu/athub_v4_1_0.h
new file mode 100644
index 000000000000..4d18d0998fa8
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/athub_v4_1_0.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __ATHUB_V4_1_0_H__
+#define __ATHUB_V4_1_0_H__
+
+int athub_v4_1_0_set_clockgating(struct amdgpu_device *adev,
+				 enum amd_clockgating_state state);
+void athub_v4_1_0_get_clockgating(struct amdgpu_device *adev, u64 *flags);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/atom.c b/drivers/gpu/drm/amd/amdgpu/atom.c
index d69aa2e179bb..7a063e44d429 100644
--- a/drivers/gpu/drm/amd/amdgpu/atom.c
+++ b/drivers/gpu/drm/amd/amdgpu/atom.c
@@ -25,10 +25,15 @@
 #include <linux/module.h>
 #include <linux/sched.h>
 #include <linux/slab.h>
-#include <asm/unaligned.h>
+#include <linux/string_helpers.h>
+
+#include <linux/unaligned.h>
+
+#include <drm/drm_util.h>
 
 #define ATOM_DEBUG
 
+#include "atomfirmware.h"
 #include "atom.h"
 #include "atom-names.h"
 #include "atom-bits.h"
@@ -52,9 +57,12 @@
 #define PLL_INDEX	2
 #define PLL_DATA	3
 
+#define ATOM_CMD_TIMEOUT_SEC	20
+
 typedef struct {
 	struct atom_context *ctx;
 	uint32_t *ps, *ws;
+	int ps_size, ws_size;
 	int ps_shift;
 	uint16_t start;
 	unsigned last_jump;
@@ -62,13 +70,13 @@ typedef struct {
 	bool abort;
 } atom_exec_context;
 
-int amdgpu_atom_debug = 0;
-static int amdgpu_atom_execute_table_locked(struct atom_context *ctx, int index, uint32_t * params);
-int amdgpu_atom_execute_table(struct atom_context *ctx, int index, uint32_t * params);
+int amdgpu_atom_debug;
+static int amdgpu_atom_execute_table_locked(struct atom_context *ctx, int index, uint32_t *params, int params_size);
+int amdgpu_atom_execute_table(struct atom_context *ctx, int index, uint32_t *params, int params_size);
 
 static uint32_t atom_arg_mask[8] =
-    { 0xFFFFFFFF, 0xFFFF, 0xFFFF00, 0xFFFF0000, 0xFF, 0xFF00, 0xFF0000,
-0xFF000000 };
+	{ 0xFFFFFFFF, 0xFFFF, 0xFFFF00, 0xFFFF0000, 0xFF, 0xFF00, 0xFF0000,
+	  0xFF000000 };
 static int atom_arg_shift[8] = { 0, 0, 8, 16, 0, 8, 16, 24 };
 
 static int atom_dst_to_src[8][4] = {
@@ -84,7 +92,7 @@ static int atom_dst_to_src[8][4] = {
 };
 static int atom_def_dst[8] = { 0, 0, 1, 2, 0, 1, 2, 3 };
 
-static int debug_depth = 0;
+static int debug_depth;
 #ifdef ATOM_DEBUG
 static void debug_print_spaces(int n)
 {
@@ -110,11 +118,11 @@ static uint32_t atom_iio_execute(struct atom_context *ctx, int base,
 			base++;
 			break;
 		case ATOM_IIO_READ:
-			temp = ctx->card->ioreg_read(ctx->card, CU16(base + 1));
+			temp = ctx->card->reg_read(ctx->card, CU16(base + 1));
 			base += 3;
 			break;
 		case ATOM_IIO_WRITE:
-			ctx->card->ioreg_write(ctx->card, CU16(base + 1), temp);
+			ctx->card->reg_write(ctx->card, CU16(base + 1), temp);
 			base += 3;
 			break;
 		case ATOM_IIO_CLEAR:
@@ -216,7 +224,10 @@ static uint32_t atom_get_src_int(atom_exec_context *ctx, uint8_t attr,
 		(*ptr)++;
 		/* get_unaligned_le32 avoids unaligned accesses from atombios
 		 * tables, noticed on a DEC Alpha. */
-		val = get_unaligned_le32((u32 *)&ctx->ps[idx]);
+		if (idx < ctx->ps_size)
+			val = get_unaligned_le32((u32 *)&ctx->ps[idx]);
+		else
+			pr_info("PS index out of range: %i > %i\n", idx, ctx->ps_size);
 		if (print)
 			DEBUG("PS[0x%02X,0x%04X]", idx, val);
 		break;
@@ -254,7 +265,10 @@ static uint32_t atom_get_src_int(atom_exec_context *ctx, uint8_t attr,
 			val = gctx->reg_block;
 			break;
 		default:
-			val = ctx->ws[idx];
+			if (idx < ctx->ws_size)
+				val = ctx->ws[idx];
+			else
+				pr_info("WS index out of range: %i > %i\n", idx, ctx->ws_size);
 		}
 		break;
 	case ATOM_ARG_ID:
@@ -287,7 +301,7 @@ static uint32_t atom_get_src_int(atom_exec_context *ctx, uint8_t attr,
 			(*ptr) += 4;
 			if (print)
 				DEBUG("IMM 0x%08X\n", val);
-			return val;
+			break;
 		case ATOM_SRC_WORD0:
 		case ATOM_SRC_WORD8:
 		case ATOM_SRC_WORD16:
@@ -295,7 +309,7 @@ static uint32_t atom_get_src_int(atom_exec_context *ctx, uint8_t attr,
 			(*ptr) += 2;
 			if (print)
 				DEBUG("IMM 0x%04X\n", val);
-			return val;
+			break;
 		case ATOM_SRC_BYTE0:
 		case ATOM_SRC_BYTE8:
 		case ATOM_SRC_BYTE16:
@@ -304,9 +318,9 @@ static uint32_t atom_get_src_int(atom_exec_context *ctx, uint8_t attr,
 			(*ptr)++;
 			if (print)
 				DEBUG("IMM 0x%02X\n", val);
-			return val;
+			break;
 		}
-		return 0;
+		return val;
 	case ATOM_ARG_PLL:
 		idx = U8(*ptr);
 		(*ptr)++;
@@ -388,7 +402,6 @@ static void atom_skip_src_int(atom_exec_context *ctx, uint8_t attr, int *ptr)
 			(*ptr)++;
 			return;
 		}
-		return;
 	}
 }
 
@@ -489,6 +502,10 @@ static void atom_put_dst(atom_exec_context *ctx, int arg, uint8_t attr,
 		idx = U8(*ptr);
 		(*ptr)++;
 		DEBUG("PS[0x%02X]", idx);
+		if (idx >= ctx->ps_size) {
+			pr_info("PS index out of range: %i > %i\n", idx, ctx->ps_size);
+			return;
+		}
 		ctx->ps[idx] = cpu_to_le32(val);
 		break;
 	case ATOM_ARG_WS:
@@ -521,6 +538,10 @@ static void atom_put_dst(atom_exec_context *ctx, int arg, uint8_t attr,
 			gctx->reg_block = val;
 			break;
 		default:
+			if (idx >= ctx->ws_size) {
+				pr_info("WS index out of range: %i > %i\n", idx, ctx->ws_size);
+				return;
+			}
 			ctx->ws[idx] = val;
 		}
 		break;
@@ -618,7 +639,7 @@ static void atom_op_calltable(atom_exec_context *ctx, int *ptr, int arg)
 	else
 		SDEBUG("   table: %d\n", idx);
 	if (U16(ctx->ctx->cmd_table + 4 + 2 * idx))
-		r = amdgpu_atom_execute_table_locked(ctx->ctx, idx, ctx->ps + ctx->ps_shift);
+		r = amdgpu_atom_execute_table_locked(ctx->ctx, idx, ctx->ps + ctx->ps_shift, ctx->ps_size - ctx->ps_shift);
 	if (r) {
 		ctx->abort = true;
 	}
@@ -735,15 +756,16 @@ static void atom_op_jump(atom_exec_context *ctx, int *ptr, int arg)
 		break;
 	}
 	if (arg != ATOM_COND_ALWAYS)
-		SDEBUG("   taken: %s\n", execute ? "yes" : "no");
+		SDEBUG("   taken: %s\n", str_yes_no(execute));
 	SDEBUG("   target: 0x%04X\n", target);
 	if (execute) {
 		if (ctx->last_jump == (ctx->start + target)) {
 			cjiffies = jiffies;
 			if (time_after(cjiffies, ctx->last_jump_jiffies)) {
 				cjiffies -= ctx->last_jump_jiffies;
-				if ((jiffies_to_msecs(cjiffies) > 5000)) {
-					DRM_ERROR("atombios stuck in loop for more than 5secs aborting\n");
+				if ((jiffies_to_msecs(cjiffies) > ATOM_CMD_TIMEOUT_SEC*1000)) {
+					DRM_ERROR("atombios stuck in loop for more than %dsecs aborting\n",
+						  ATOM_CMD_TIMEOUT_SEC);
 					ctx->abort = true;
 				}
 			} else {
@@ -1196,7 +1218,7 @@ static struct {
 	atom_op_div32, ATOM_ARG_WS},
 };
 
-static int amdgpu_atom_execute_table_locked(struct atom_context *ctx, int index, uint32_t * params)
+static int amdgpu_atom_execute_table_locked(struct atom_context *ctx, int index, uint32_t *params, int params_size)
 {
 	int base = CU16(ctx->cmd_table + 4 + 2 * index);
 	int len, ws, ps, ptr;
@@ -1218,12 +1240,21 @@ static int amdgpu_atom_execute_table_locked(struct atom_context *ctx, int index,
 	ectx.ps_shift = ps / 4;
 	ectx.start = base;
 	ectx.ps = params;
+	ectx.ps_size = params_size;
 	ectx.abort = false;
 	ectx.last_jump = 0;
-	if (ws)
-		ectx.ws = kzalloc(4 * ws, GFP_KERNEL);
-	else
+	ectx.last_jump_jiffies = 0;
+	if (ws) {
+		ectx.ws = kcalloc(4, ws, GFP_KERNEL);
+		if (!ectx.ws) {
+			ret = -ENOMEM;
+			goto free;
+		}
+		ectx.ws_size = ws;
+	} else {
 		ectx.ws = NULL;
+		ectx.ws_size = 0;
+	}
 
 	debug_depth++;
 	while (1) {
@@ -1257,7 +1288,7 @@ free:
 	return ret;
 }
 
-int amdgpu_atom_execute_table(struct atom_context *ctx, int index, uint32_t * params)
+int amdgpu_atom_execute_table(struct atom_context *ctx, int index, uint32_t *params, int params_size)
 {
 	int r;
 
@@ -1273,7 +1304,7 @@ int amdgpu_atom_execute_table(struct atom_context *ctx, int index, uint32_t * pa
 	/* reset divmul */
 	ctx->divmul[0] = 0;
 	ctx->divmul[1] = 0;
-	r = amdgpu_atom_execute_table_locked(ctx, index, params);
+	r = amdgpu_atom_execute_table_locked(ctx, index, params, params_size);
 	mutex_unlock(&ctx->mutex);
 	return r;
 }
@@ -1294,13 +1325,209 @@ static void atom_index_iio(struct atom_context *ctx, int base)
 	}
 }
 
+static void atom_get_vbios_name(struct atom_context *ctx)
+{
+	unsigned char *p_rom;
+	unsigned char str_num;
+	unsigned short off_to_vbios_str;
+	unsigned char *c_ptr;
+	int name_size;
+	int i;
+
+	const char *na = "--N/A--";
+	char *back;
+
+	p_rom = ctx->bios;
+
+	str_num = *(p_rom + OFFSET_TO_GET_ATOMBIOS_NUMBER_OF_STRINGS);
+	if (str_num != 0) {
+		off_to_vbios_str =
+			*(unsigned short *)(p_rom + OFFSET_TO_GET_ATOMBIOS_STRING_START);
+
+		c_ptr = (unsigned char *)(p_rom + off_to_vbios_str);
+	} else {
+		/* do not know where to find name */
+		memcpy(ctx->name, na, 7);
+		ctx->name[7] = 0;
+		return;
+	}
+
+	/*
+	 * skip the atombios strings, usually 4
+	 * 1st is P/N, 2nd is ASIC, 3rd is PCI type, 4th is Memory type
+	 */
+	for (i = 0; i < str_num; i++) {
+		while (*c_ptr != 0)
+			c_ptr++;
+		c_ptr++;
+	}
+
+	/* skip the following 2 chars: 0x0D 0x0A */
+	c_ptr += 2;
+
+	name_size = strnlen(c_ptr, STRLEN_LONG - 1);
+	memcpy(ctx->name, c_ptr, name_size);
+	back = ctx->name + name_size;
+	while ((*--back) == ' ')
+		;
+	*(back + 1) = '\0';
+}
+
+static void atom_get_vbios_date(struct atom_context *ctx)
+{
+	unsigned char *p_rom;
+	unsigned char *date_in_rom;
+
+	p_rom = ctx->bios;
+
+	date_in_rom = p_rom + OFFSET_TO_VBIOS_DATE;
+
+	ctx->date[0] = '2';
+	ctx->date[1] = '0';
+	ctx->date[2] = date_in_rom[6];
+	ctx->date[3] = date_in_rom[7];
+	ctx->date[4] = '/';
+	ctx->date[5] = date_in_rom[0];
+	ctx->date[6] = date_in_rom[1];
+	ctx->date[7] = '/';
+	ctx->date[8] = date_in_rom[3];
+	ctx->date[9] = date_in_rom[4];
+	ctx->date[10] = ' ';
+	ctx->date[11] = date_in_rom[9];
+	ctx->date[12] = date_in_rom[10];
+	ctx->date[13] = date_in_rom[11];
+	ctx->date[14] = date_in_rom[12];
+	ctx->date[15] = date_in_rom[13];
+	ctx->date[16] = '\0';
+}
+
+static unsigned char *atom_find_str_in_rom(struct atom_context *ctx, char *str, int start,
+					   int end, int maxlen)
+{
+	unsigned long str_off;
+	unsigned char *p_rom;
+	unsigned short str_len;
+
+	str_off = 0;
+	str_len = strnlen(str, maxlen);
+	p_rom = ctx->bios;
+
+	for (; start <= end; ++start) {
+		for (str_off = 0; str_off < str_len; ++str_off) {
+			if (str[str_off] != *(p_rom + start + str_off))
+				break;
+		}
+
+		if (str_off == str_len || str[str_off] == 0)
+			return p_rom + start;
+	}
+	return NULL;
+}
+
+static void atom_get_vbios_pn(struct atom_context *ctx)
+{
+	unsigned char *p_rom;
+	unsigned short off_to_vbios_str;
+	unsigned char *vbios_str;
+	int count;
+
+	off_to_vbios_str = 0;
+	p_rom = ctx->bios;
+
+	if (*(p_rom + OFFSET_TO_GET_ATOMBIOS_NUMBER_OF_STRINGS) != 0) {
+		off_to_vbios_str =
+			*(unsigned short *)(p_rom + OFFSET_TO_GET_ATOMBIOS_STRING_START);
+
+		vbios_str = (unsigned char *)(p_rom + off_to_vbios_str);
+	} else {
+		vbios_str = p_rom + OFFSET_TO_VBIOS_PART_NUMBER;
+	}
+
+	if (*vbios_str == 0) {
+		vbios_str = atom_find_str_in_rom(ctx, BIOS_ATOM_PREFIX, 3, 1024, 64);
+		if (vbios_str == NULL)
+			vbios_str += sizeof(BIOS_ATOM_PREFIX) - 1;
+	}
+	OPTIMIZER_HIDE_VAR(vbios_str);
+	if (vbios_str != NULL && *vbios_str == 0)
+		vbios_str++;
+
+	if (vbios_str != NULL) {
+		count = 0;
+		while ((count < BIOS_STRING_LENGTH) && vbios_str[count] >= ' ' &&
+		       vbios_str[count] <= 'z') {
+			ctx->vbios_pn[count] = vbios_str[count];
+			count++;
+		}
+
+		ctx->vbios_pn[count] = 0;
+	}
+
+	pr_info("ATOM BIOS: %s\n", ctx->vbios_pn);
+}
+
+static void atom_get_vbios_version(struct atom_context *ctx)
+{
+	unsigned short start = 3, end;
+	unsigned char *vbios_ver;
+	unsigned char *p_rom;
+
+	p_rom = ctx->bios;
+	/* Search from strings offset if it's present */
+	start = *(unsigned short *)(p_rom +
+				    OFFSET_TO_GET_ATOMBIOS_STRING_START);
+
+	/* Search till atom rom header start point */
+	end = *(unsigned short *)(p_rom + OFFSET_TO_ATOM_ROM_HEADER_POINTER);
+
+	/* Use hardcoded offsets, if the offsets are not populated */
+	if (end <= start) {
+		start = 3;
+		end = 1024;
+	}
+
+	/* find anchor ATOMBIOSBK-AMD */
+	vbios_ver =
+		atom_find_str_in_rom(ctx, BIOS_VERSION_PREFIX, start, end, 64);
+	if (vbios_ver != NULL) {
+		/* skip ATOMBIOSBK-AMD VER */
+		vbios_ver += 18;
+		memcpy(ctx->vbios_ver_str, vbios_ver, STRLEN_NORMAL);
+	} else {
+		ctx->vbios_ver_str[0] = '\0';
+	}
+}
+
+static void atom_get_vbios_build(struct atom_context *ctx)
+{
+	unsigned char *atom_rom_hdr;
+	unsigned char *str;
+	uint16_t base, len;
+
+	base = CU16(ATOM_ROM_TABLE_PTR);
+	atom_rom_hdr = CSTR(base);
+
+	str = CSTR(CU16(base + ATOM_ROM_CFG_PTR));
+	/* Skip config string */
+	while (str < atom_rom_hdr && *str++)
+		;
+	/* Skip change list string */
+	while (str < atom_rom_hdr && *str++)
+		;
+
+	len = min(atom_rom_hdr - str, STRLEN_NORMAL);
+	if (len)
+		strscpy(ctx->build_num, str, len);
+}
+
 struct atom_context *amdgpu_atom_parse(struct card_info *card, void *bios)
 {
 	int base;
 	struct atom_context *ctx =
 	    kzalloc(sizeof(struct atom_context), GFP_KERNEL);
-	char *str;
-	u16 idx;
+	struct _ATOM_ROM_HEADER *atom_rom_header;
+	struct _ATOM_MASTER_DATA_TABLE *master_table;
+	struct _ATOM_FIRMWARE_INFO *atom_fw_info;
 
 	if (!ctx)
 		return NULL;
@@ -1338,13 +1565,22 @@ struct atom_context *amdgpu_atom_parse(struct card_info *card, void *bios)
 		return NULL;
 	}
 
-	idx = CU16(ATOM_ROM_PART_NUMBER_PTR);
-	if (idx == 0)
-		idx = 0x80;
+	atom_rom_header = (struct _ATOM_ROM_HEADER *)CSTR(base);
+	if (atom_rom_header->usMasterDataTableOffset != 0) {
+		master_table = (struct _ATOM_MASTER_DATA_TABLE *)
+				CSTR(atom_rom_header->usMasterDataTableOffset);
+		if (master_table->ListOfDataTables.FirmwareInfo != 0) {
+			atom_fw_info = (struct _ATOM_FIRMWARE_INFO *)
+					CSTR(master_table->ListOfDataTables.FirmwareInfo);
+			ctx->version = atom_fw_info->ulFirmwareRevision;
+		}
+	}
 
-	str = CSTR(idx);
-	if (*str != '\0')
-		pr_info("ATOM BIOS: %s\n", str);
+	atom_get_vbios_name(ctx);
+	atom_get_vbios_pn(ctx);
+	atom_get_vbios_date(ctx);
+	atom_get_vbios_version(ctx);
+	atom_get_vbios_build(ctx);
 
 	return ctx;
 }
@@ -1364,7 +1600,7 @@ int amdgpu_atom_asic_init(struct atom_context *ctx)
 
 	if (!CU16(ctx->cmd_table + 4 + 2 * ATOM_CMD_INIT))
 		return 1;
-	ret = amdgpu_atom_execute_table(ctx, ATOM_CMD_INIT, ps);
+	ret = amdgpu_atom_execute_table(ctx, ATOM_CMD_INIT, ps, 16);
 	if (ret)
 		return ret;
 
@@ -1380,8 +1616,8 @@ void amdgpu_atom_destroy(struct atom_context *ctx)
 }
 
 bool amdgpu_atom_parse_data_header(struct atom_context *ctx, int index,
-			    uint16_t * size, uint8_t * frev, uint8_t * crev,
-			    uint16_t * data_start)
+			    uint16_t *size, uint8_t *frev, uint8_t *crev,
+			    uint16_t *data_start)
 {
 	int offset = index * 2 + 4;
 	int idx = CU16(ctx->data_table + offset);
@@ -1400,8 +1636,8 @@ bool amdgpu_atom_parse_data_header(struct atom_context *ctx, int index,
 	return true;
 }
 
-bool amdgpu_atom_parse_cmd_header(struct atom_context *ctx, int index, uint8_t * frev,
-			   uint8_t * crev)
+bool amdgpu_atom_parse_cmd_header(struct atom_context *ctx, int index, uint8_t *frev,
+			   uint8_t *crev)
 {
 	int offset = index * 2 + 4;
 	int idx = CU16(ctx->cmd_table + offset);
diff --git a/drivers/gpu/drm/amd/amdgpu/atom.h b/drivers/gpu/drm/amd/amdgpu/atom.h
index ddd8045accf3..825ff28731f5 100644
--- a/drivers/gpu/drm/amd/amdgpu/atom.h
+++ b/drivers/gpu/drm/amd/amdgpu/atom.h
@@ -26,17 +26,18 @@
 #define ATOM_H
 
 #include <linux/types.h>
-#include <drm/drmP.h>
+
+struct drm_device;
 
 #define ATOM_BIOS_MAGIC		0xAA55
 #define ATOM_ATI_MAGIC_PTR	0x30
 #define ATOM_ATI_MAGIC		" 761295520"
 #define ATOM_ROM_TABLE_PTR	0x48
-#define ATOM_ROM_PART_NUMBER_PTR	0x6E
 
 #define ATOM_ROM_MAGIC		"ATOM"
 #define ATOM_ROM_MAGIC_PTR	4
 
+#define ATOM_ROM_CFG_PTR 0xC
 #define ATOM_ROM_MSG_PTR	0x10
 #define ATOM_ROM_CMD_PTR	0x1E
 #define ATOM_ROM_DATA_PTR	0x20
@@ -111,16 +112,21 @@
 #define ATOM_IO_SYSIO		2
 #define ATOM_IO_IIO		0x80
 
+#define STRLEN_NORMAL		32
+#define STRLEN_LONG		64
+#define STRLEN_VERYLONG		254
+
 struct card_info {
 	struct drm_device *dev;
-	void (* reg_write)(struct card_info *, uint32_t, uint32_t);   /*  filled by driver */
-	uint32_t (* reg_read)(struct card_info *, uint32_t);          /*  filled by driver */
-	void (* ioreg_write)(struct card_info *, uint32_t, uint32_t);   /*  filled by driver */
-	uint32_t (* ioreg_read)(struct card_info *, uint32_t);          /*  filled by driver */
-	void (* mc_write)(struct card_info *, uint32_t, uint32_t);   /*  filled by driver */
-	uint32_t (* mc_read)(struct card_info *, uint32_t);          /*  filled by driver */
-	void (* pll_write)(struct card_info *, uint32_t, uint32_t);   /*  filled by driver */
-	uint32_t (* pll_read)(struct card_info *, uint32_t);          /*  filled by driver */
+	void (*reg_write)(struct card_info *info,
+			  u32 reg, uint32_t val);   /*  filled by driver */
+	uint32_t (*reg_read)(struct card_info *info, uint32_t reg);          /*  filled by driver */
+	void (*mc_write)(struct card_info *info,
+			 u32 reg, uint32_t val);   /*  filled by driver */
+	uint32_t (*mc_read)(struct card_info *info, uint32_t reg);          /*  filled by driver */
+	void (*pll_write)(struct card_info *info,
+			  u32 reg, uint32_t val);   /*  filled by driver */
+	uint32_t (*pll_read)(struct card_info *info, uint32_t reg);          /*  filled by driver */
 };
 
 struct atom_context {
@@ -140,14 +146,21 @@ struct atom_context {
 	int io_mode;
 	uint32_t *scratch;
 	int scratch_size_bytes;
+
+	uint8_t name[STRLEN_LONG];
+	uint8_t vbios_pn[STRLEN_LONG];
+	uint32_t version;
+	uint8_t vbios_ver_str[STRLEN_NORMAL];
+	uint8_t date[STRLEN_NORMAL];
+	uint8_t build_num[STRLEN_NORMAL];
 };
 
 extern int amdgpu_atom_debug;
 
-struct atom_context *amdgpu_atom_parse(struct card_info *, void *);
-int amdgpu_atom_execute_table(struct atom_context *, int, uint32_t *);
-int amdgpu_atom_asic_init(struct atom_context *);
-void amdgpu_atom_destroy(struct atom_context *);
+struct atom_context *amdgpu_atom_parse(struct card_info *card, void *bios);
+int amdgpu_atom_execute_table(struct atom_context *ctx, int index, uint32_t *params, int params_size);
+int amdgpu_atom_asic_init(struct atom_context *ctx);
+void amdgpu_atom_destroy(struct atom_context *ctx);
 bool amdgpu_atom_parse_data_header(struct atom_context *ctx, int index, uint16_t *size,
 			    uint8_t *frev, uint8_t *crev, uint16_t *data_start);
 bool amdgpu_atom_parse_cmd_header(struct atom_context *ctx, int index,
diff --git a/drivers/gpu/drm/amd/amdgpu/atombios_crtc.c b/drivers/gpu/drm/amd/amdgpu/atombios_crtc.c
index 8a0818b23ea4..3dfc28840a7d 100644
--- a/drivers/gpu/drm/amd/amdgpu/atombios_crtc.c
+++ b/drivers/gpu/drm/amd/amdgpu/atombios_crtc.c
@@ -23,8 +23,7 @@
  * Authors: Dave Airlie
  *          Alex Deucher
  */
-#include <drm/drmP.h>
-#include <drm/drm_crtc_helper.h>
+
 #include <drm/amdgpu_drm.h>
 #include <drm/drm_fixed.h>
 #include "amdgpu.h"
@@ -41,7 +40,7 @@ void amdgpu_atombios_crtc_overscan_setup(struct drm_crtc *crtc,
 				  struct drm_display_mode *adjusted_mode)
 {
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	SET_CRTC_OVERSCAN_PS_ALLOCATION args;
 	int index = GetIndexIntoMasterTable(COMMAND, SetCRTC_OverScan);
@@ -78,13 +77,13 @@ void amdgpu_atombios_crtc_overscan_setup(struct drm_crtc *crtc,
 		args.usOverscanTop = cpu_to_le16(amdgpu_crtc->v_border);
 		break;
 	}
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 }
 
 void amdgpu_atombios_crtc_scaler_setup(struct drm_crtc *crtc)
 {
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	ENABLE_SCALER_PS_ALLOCATION args;
 	int index = GetIndexIntoMasterTable(COMMAND, EnableScaler);
@@ -107,14 +106,14 @@ void amdgpu_atombios_crtc_scaler_setup(struct drm_crtc *crtc)
 		args.ucEnable = ATOM_SCALER_DISABLE;
 		break;
 	}
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 }
 
 void amdgpu_atombios_crtc_lock(struct drm_crtc *crtc, int lock)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	int index =
 	    GetIndexIntoMasterTable(COMMAND, UpdateCRTC_DoubleBufferRegisters);
 	ENABLE_CRTC_PS_ALLOCATION args;
@@ -124,14 +123,14 @@ void amdgpu_atombios_crtc_lock(struct drm_crtc *crtc, int lock)
 	args.ucCRTC = amdgpu_crtc->crtc_id;
 	args.ucEnable = lock;
 
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 }
 
 void amdgpu_atombios_crtc_enable(struct drm_crtc *crtc, int state)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	int index = GetIndexIntoMasterTable(COMMAND, EnableCRTC);
 	ENABLE_CRTC_PS_ALLOCATION args;
 
@@ -140,14 +139,14 @@ void amdgpu_atombios_crtc_enable(struct drm_crtc *crtc, int state)
 	args.ucCRTC = amdgpu_crtc->crtc_id;
 	args.ucEnable = state;
 
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 }
 
 void amdgpu_atombios_crtc_blank(struct drm_crtc *crtc, int state)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	int index = GetIndexIntoMasterTable(COMMAND, BlankCRTC);
 	BLANK_CRTC_PS_ALLOCATION args;
 
@@ -156,14 +155,14 @@ void amdgpu_atombios_crtc_blank(struct drm_crtc *crtc, int state)
 	args.ucCRTC = amdgpu_crtc->crtc_id;
 	args.ucBlanking = state;
 
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 }
 
 void amdgpu_atombios_crtc_powergate(struct drm_crtc *crtc, int state)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	int index = GetIndexIntoMasterTable(COMMAND, EnableDispPowerGating);
 	ENABLE_DISP_POWER_GATING_PS_ALLOCATION args;
 
@@ -172,7 +171,7 @@ void amdgpu_atombios_crtc_powergate(struct drm_crtc *crtc, int state)
 	args.ucDispPipeId = amdgpu_crtc->crtc_id;
 	args.ucEnable = state;
 
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 }
 
 void amdgpu_atombios_crtc_powergate_init(struct amdgpu_device *adev)
@@ -184,7 +183,7 @@ void amdgpu_atombios_crtc_powergate_init(struct amdgpu_device *adev)
 
 	args.ucEnable = ATOM_INIT;
 
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 }
 
 void amdgpu_atombios_crtc_set_dtd_timing(struct drm_crtc *crtc,
@@ -192,7 +191,7 @@ void amdgpu_atombios_crtc_set_dtd_timing(struct drm_crtc *crtc,
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	SET_CRTC_USING_DTD_TIMING_PARAMETERS args;
 	int index = GetIndexIntoMasterTable(COMMAND, SetCRTC_UsingDTDTiming);
 	u16 misc = 0;
@@ -229,7 +228,7 @@ void amdgpu_atombios_crtc_set_dtd_timing(struct drm_crtc *crtc,
 	args.susModeMiscInfo.usAccess = cpu_to_le16(misc);
 	args.ucCRTC = amdgpu_crtc->crtc_id;
 
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 }
 
 union atom_enable_ss {
@@ -294,7 +293,7 @@ static void amdgpu_atombios_crtc_program_ss(struct amdgpu_device *adev,
 	args.v3.usSpreadSpectrumStep = cpu_to_le16(ss->step);
 	args.v3.ucEnable = enable;
 
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 }
 
 union adjust_pixel_clock {
@@ -307,7 +306,7 @@ static u32 amdgpu_atombios_crtc_adjust_pll(struct drm_crtc *crtc,
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct drm_encoder *encoder = amdgpu_crtc->encoder;
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
@@ -396,7 +395,7 @@ static u32 amdgpu_atombios_crtc_adjust_pll(struct drm_crtc *crtc,
 					ADJUST_DISPLAY_CONFIG_SS_ENABLE;
 
 			amdgpu_atom_execute_table(adev->mode_info.atom_context,
-					   index, (uint32_t *)&args);
+					   index, (uint32_t *)&args, sizeof(args));
 			adjusted_clock = le16_to_cpu(args.v1.usPixelClock) * 10;
 			break;
 		case 3:
@@ -429,7 +428,7 @@ static u32 amdgpu_atombios_crtc_adjust_pll(struct drm_crtc *crtc,
 				args.v3.sInput.ucExtTransmitterID = 0;
 
 			amdgpu_atom_execute_table(adev->mode_info.atom_context,
-					   index, (uint32_t *)&args);
+					   index, (uint32_t *)&args, sizeof(args));
 			adjusted_clock = le32_to_cpu(args.v3.sOutput.ulDispPllFreq) * 10;
 			if (args.v3.sOutput.ucRefDiv) {
 				amdgpu_crtc->pll_flags |= AMDGPU_PLL_USE_FRAC_FB_DIV;
@@ -515,7 +514,7 @@ void amdgpu_atombios_crtc_set_disp_eng_pll(struct amdgpu_device *adev,
 		DRM_ERROR("Unknown table version %d %d\n", frev, crev);
 		return;
 	}
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 }
 
 union set_dce_clock {
@@ -545,7 +544,7 @@ u32 amdgpu_atombios_crtc_set_dce_clock(struct amdgpu_device *adev,
 			args.v2_1.asParam.ulDCEClkFreq = cpu_to_le32(freq); /* 10kHz units */
 			args.v2_1.asParam.ucDCEClkType = clk_type;
 			args.v2_1.asParam.ucDCEClkSrc = clk_src;
-			amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+			amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 			ret_freq = le32_to_cpu(args.v2_1.asParam.ulDCEClkFreq) * 10;
 			break;
 		default:
@@ -588,7 +587,7 @@ void amdgpu_atombios_crtc_program_pll(struct drm_crtc *crtc,
 				      struct amdgpu_atom_ss *ss)
 {
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	u8 frev, crev;
 	int index = GetIndexIntoMasterTable(COMMAND, SetPixelClock);
 	union set_pixel_clock args;
@@ -741,7 +740,7 @@ void amdgpu_atombios_crtc_program_pll(struct drm_crtc *crtc,
 		return;
 	}
 
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 }
 
 int amdgpu_atombios_crtc_prepare_pll(struct drm_crtc *crtc,
@@ -749,7 +748,7 @@ int amdgpu_atombios_crtc_prepare_pll(struct drm_crtc *crtc,
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder =
 		to_amdgpu_encoder(amdgpu_crtc->encoder);
 	int encoder_mode = amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder);
@@ -818,7 +817,7 @@ void amdgpu_atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder =
 		to_amdgpu_encoder(amdgpu_crtc->encoder);
 	u32 pll_clock = mode->clock;
@@ -851,7 +850,7 @@ void amdgpu_atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode
 	pll->reference_div = amdgpu_crtc->pll_reference_div;
 	pll->post_div = amdgpu_crtc->pll_post_div;
 
-	amdgpu_pll_compute(pll, amdgpu_crtc->adjusted_clock, &pll_clock,
+	amdgpu_pll_compute(adev, pll, amdgpu_crtc->adjusted_clock, &pll_clock,
 			    &fb_div, &frac_fb_div, &ref_div, &post_div);
 
 	amdgpu_atombios_crtc_program_ss(adev, ATOM_DISABLE, amdgpu_crtc->pll_id,
diff --git a/drivers/gpu/drm/amd/amdgpu/atombios_dp.c b/drivers/gpu/drm/amd/amdgpu/atombios_dp.c
index f81068ba4cc6..492813ab1b54 100644
--- a/drivers/gpu/drm/amd/amdgpu/atombios_dp.c
+++ b/drivers/gpu/drm/amd/amdgpu/atombios_dp.c
@@ -24,8 +24,10 @@
  *          Alex Deucher
  *          Jerome Glisse
  */
-#include <drm/drmP.h>
+
 #include <drm/amdgpu_drm.h>
+#include <drm/display/drm_dp_helper.h>
+
 #include "amdgpu.h"
 
 #include "atom.h"
@@ -34,7 +36,6 @@
 #include "atombios_dp.h"
 #include "amdgpu_connectors.h"
 #include "amdgpu_atombios.h"
-#include <drm/drm_dp_helper.h>
 
 /* move these to drm_dp_helper.c/h */
 #define DP_LINK_CONFIGURATION_SIZE 9
@@ -60,7 +61,7 @@ static int amdgpu_atombios_dp_process_aux_ch(struct amdgpu_i2c_chan *chan,
 				      u8 delay, u8 *ack)
 {
 	struct drm_device *dev = chan->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	union aux_channel_transaction args;
 	int index = GetIndexIntoMasterTable(COMMAND, ProcessAuxChannelTransaction);
 	unsigned char *base;
@@ -82,7 +83,7 @@ static int amdgpu_atombios_dp_process_aux_ch(struct amdgpu_i2c_chan *chan,
 	args.v2.ucDelay = delay / 10;
 	args.v2.ucHPD_ID = chan->rec.hpd;
 
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 
 	*ack = args.v2.ucReplyStatus;
 
@@ -186,16 +187,12 @@ amdgpu_atombios_dp_aux_transfer(struct drm_dp_aux *aux, struct drm_dp_aux_msg *m
 
 void amdgpu_atombios_dp_aux_init(struct amdgpu_connector *amdgpu_connector)
 {
-	int ret;
-
 	amdgpu_connector->ddc_bus->rec.hpd = amdgpu_connector->hpd.hpd;
-	amdgpu_connector->ddc_bus->aux.dev = amdgpu_connector->base.kdev;
 	amdgpu_connector->ddc_bus->aux.transfer = amdgpu_atombios_dp_aux_transfer;
-	ret = drm_dp_aux_register(&amdgpu_connector->ddc_bus->aux);
-	if (!ret)
-		amdgpu_connector->ddc_bus->has_aux = true;
+	amdgpu_connector->ddc_bus->aux.drm_dev = amdgpu_connector->base.dev;
 
-	WARN(ret, "drm_dp_aux_register_i2c_bus() failed with error %d\n", ret);
+	drm_dp_aux_init(&amdgpu_connector->ddc_bus->aux);
+	amdgpu_connector->ddc_bus->has_aux = true;
 }
 
 /***** general DP utility functions *****/
@@ -304,14 +301,14 @@ static u8 amdgpu_atombios_dp_encoder_service(struct amdgpu_device *adev,
 	args.ucLaneNum = lane_num;
 	args.ucStatus = 0;
 
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 	return args.ucStatus;
 }
 
 u8 amdgpu_atombios_dp_get_sinktype(struct amdgpu_connector *amdgpu_connector)
 {
 	struct drm_device *dev = amdgpu_connector->base.dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 
 	return amdgpu_atombios_dp_encoder_service(adev, ATOM_DP_ACTION_GET_SINK_TYPE, 0,
 					   amdgpu_connector->ddc_bus->rec.i2c_id, 0);
@@ -334,6 +331,22 @@ static void amdgpu_atombios_dp_probe_oui(struct amdgpu_connector *amdgpu_connect
 			      buf[0], buf[1], buf[2]);
 }
 
+static void amdgpu_atombios_dp_ds_ports(struct amdgpu_connector *amdgpu_connector)
+{
+	struct amdgpu_connector_atom_dig *dig_connector = amdgpu_connector->con_priv;
+	int ret;
+
+	if (dig_connector->dpcd[DP_DPCD_REV] > 0x10) {
+		ret = drm_dp_dpcd_read(&amdgpu_connector->ddc_bus->aux,
+				       DP_DOWNSTREAM_PORT_0,
+				       dig_connector->downstream_ports,
+				       DP_MAX_DOWNSTREAM_PORTS);
+		if (ret)
+			memset(dig_connector->downstream_ports, 0,
+			       DP_MAX_DOWNSTREAM_PORTS);
+	}
+}
+
 int amdgpu_atombios_dp_get_dpcd(struct amdgpu_connector *amdgpu_connector)
 {
 	struct amdgpu_connector_atom_dig *dig_connector = amdgpu_connector->con_priv;
@@ -349,7 +362,7 @@ int amdgpu_atombios_dp_get_dpcd(struct amdgpu_connector *amdgpu_connector)
 			      dig_connector->dpcd);
 
 		amdgpu_atombios_dp_probe_oui(amdgpu_connector);
-
+		amdgpu_atombios_dp_ds_ports(amdgpu_connector);
 		return 0;
 	}
 
@@ -361,7 +374,6 @@ int amdgpu_atombios_dp_get_panel_mode(struct drm_encoder *encoder,
 			       struct drm_connector *connector)
 {
 	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
-	struct amdgpu_connector_atom_dig *dig_connector;
 	int panel_mode = DP_PANEL_MODE_EXTERNAL_DP_MODE;
 	u16 dp_bridge = amdgpu_connector_encoder_get_dp_bridge_encoder_id(connector);
 	u8 tmp;
@@ -369,8 +381,6 @@ int amdgpu_atombios_dp_get_panel_mode(struct drm_encoder *encoder,
 	if (!amdgpu_connector->con_priv)
 		return panel_mode;
 
-	dig_connector = amdgpu_connector->con_priv;
-
 	if (dp_bridge != ENCODER_OBJECT_ID_NONE) {
 		/* DP bridge chips */
 		if (drm_dp_dpcd_readb(&amdgpu_connector->ddc_bus->aux,
@@ -420,7 +430,7 @@ void amdgpu_atombios_dp_set_link_config(struct drm_connector *connector,
 }
 
 int amdgpu_atombios_dp_mode_valid_helper(struct drm_connector *connector,
-				  struct drm_display_mode *mode)
+				  const struct drm_display_mode *mode)
 {
 	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 	struct amdgpu_connector_atom_dig *dig_connector;
@@ -448,8 +458,8 @@ bool amdgpu_atombios_dp_needs_link_train(struct amdgpu_connector *amdgpu_connect
 	u8 link_status[DP_LINK_STATUS_SIZE];
 	struct amdgpu_connector_atom_dig *dig = amdgpu_connector->con_priv;
 
-	if (drm_dp_dpcd_read_link_status(&amdgpu_connector->ddc_bus->aux, link_status)
-	    <= 0)
+	if (drm_dp_dpcd_read_link_status(&amdgpu_connector->ddc_bus->aux,
+					 link_status) < 0)
 		return false;
 	if (drm_dp_channel_eq_ok(link_status, dig->dp_lane_count))
 		return false;
@@ -603,10 +613,10 @@ amdgpu_atombios_dp_link_train_cr(struct amdgpu_atombios_dp_link_train_info *dp_i
 	dp_info->tries = 0;
 	voltage = 0xff;
 	while (1) {
-		drm_dp_link_train_clock_recovery_delay(dp_info->dpcd);
+		drm_dp_link_train_clock_recovery_delay(dp_info->aux, dp_info->dpcd);
 
 		if (drm_dp_dpcd_read_link_status(dp_info->aux,
-						 dp_info->link_status) <= 0) {
+						 dp_info->link_status) < 0) {
 			DRM_ERROR("displayport link status failed\n");
 			break;
 		}
@@ -668,10 +678,10 @@ amdgpu_atombios_dp_link_train_ce(struct amdgpu_atombios_dp_link_train_info *dp_i
 	dp_info->tries = 0;
 	channel_eq = false;
 	while (1) {
-		drm_dp_link_train_channel_eq_delay(dp_info->dpcd);
+		drm_dp_link_train_channel_eq_delay(dp_info->aux, dp_info->dpcd);
 
 		if (drm_dp_dpcd_read_link_status(dp_info->aux,
-						 dp_info->link_status) <= 0) {
+						 dp_info->link_status) < 0) {
 			DRM_ERROR("displayport link status failed\n");
 			break;
 		}
@@ -711,9 +721,8 @@ void amdgpu_atombios_dp_link_train(struct drm_encoder *encoder,
 			    struct drm_connector *connector)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig;
 	struct amdgpu_connector *amdgpu_connector;
 	struct amdgpu_connector_atom_dig *dig_connector;
 	struct amdgpu_atombios_dp_link_train_info dp_info;
@@ -721,7 +730,6 @@ void amdgpu_atombios_dp_link_train(struct drm_encoder *encoder,
 
 	if (!amdgpu_encoder->enc_priv)
 		return;
-	dig = amdgpu_encoder->enc_priv;
 
 	amdgpu_connector = to_amdgpu_connector(connector);
 	if (!amdgpu_connector->con_priv)
diff --git a/drivers/gpu/drm/amd/amdgpu/atombios_dp.h b/drivers/gpu/drm/amd/amdgpu/atombios_dp.h
index f59d85eaddf0..3e24acf8133f 100644
--- a/drivers/gpu/drm/amd/amdgpu/atombios_dp.h
+++ b/drivers/gpu/drm/amd/amdgpu/atombios_dp.h
@@ -32,7 +32,7 @@ int amdgpu_atombios_dp_get_panel_mode(struct drm_encoder *encoder,
 void amdgpu_atombios_dp_set_link_config(struct drm_connector *connector,
 				 const struct drm_display_mode *mode);
 int amdgpu_atombios_dp_mode_valid_helper(struct drm_connector *connector,
-				  struct drm_display_mode *mode);
+				  const struct drm_display_mode *mode);
 bool amdgpu_atombios_dp_needs_link_train(struct amdgpu_connector *amdgpu_connector);
 void amdgpu_atombios_dp_set_rx_power_state(struct drm_connector *connector,
 				    u8 power_state);
diff --git a/drivers/gpu/drm/amd/amdgpu/atombios_encoders.c b/drivers/gpu/drm/amd/amdgpu/atombios_encoders.c
index 2af26d2da127..a51f3414b65d 100644
--- a/drivers/gpu/drm/amd/amdgpu/atombios_encoders.c
+++ b/drivers/gpu/drm/amd/amdgpu/atombios_encoders.c
@@ -23,18 +23,23 @@
  * Authors: Dave Airlie
  *          Alex Deucher
  */
-#include <drm/drmP.h>
-#include <drm/drm_crtc_helper.h>
+
+#include <linux/pci.h>
+
+#include <acpi/video.h>
+
+#include <drm/drm_edid.h>
 #include <drm/amdgpu_drm.h>
 #include "amdgpu.h"
 #include "amdgpu_connectors.h"
+#include "amdgpu_display.h"
 #include "atom.h"
 #include "atombios_encoders.h"
 #include "atombios_dp.h"
 #include <linux/backlight.h>
 #include "bif/bif_4_1_d.h"
 
-static u8
+u8
 amdgpu_atombios_encoder_get_backlight_level_from_reg(struct amdgpu_device *adev)
 {
 	u8 backlight_level;
@@ -48,7 +53,7 @@ amdgpu_atombios_encoder_get_backlight_level_from_reg(struct amdgpu_device *adev)
 	return backlight_level;
 }
 
-static void
+void
 amdgpu_atombios_encoder_set_backlight_level_to_reg(struct amdgpu_device *adev,
 					    u8 backlight_level)
 {
@@ -67,7 +72,7 @@ u8
 amdgpu_atombios_encoder_get_backlight_level(struct amdgpu_encoder *amdgpu_encoder)
 {
 	struct drm_device *dev = amdgpu_encoder->base.dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 
 	if (!(adev->mode_info.firmware_flags & ATOM_BIOS_INFO_BL_CONTROLLED_BY_GPU))
 		return 0;
@@ -81,7 +86,7 @@ amdgpu_atombios_encoder_set_backlight_level(struct amdgpu_encoder *amdgpu_encode
 {
 	struct drm_encoder *encoder = &amdgpu_encoder->base;
 	struct drm_device *dev = amdgpu_encoder->base.dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder_atom_dig *dig;
 
 	if (!(adev->mode_info.firmware_flags & ATOM_BIOS_INFO_BL_CONTROLLED_BY_GPU))
@@ -115,8 +120,6 @@ amdgpu_atombios_encoder_set_backlight_level(struct amdgpu_encoder *amdgpu_encode
 	}
 }
 
-#if defined(CONFIG_BACKLIGHT_CLASS_DEVICE) || defined(CONFIG_BACKLIGHT_CLASS_DEVICE_MODULE)
-
 static u8 amdgpu_atombios_encoder_backlight_level(struct backlight_device *bd)
 {
 	u8 level;
@@ -149,7 +152,7 @@ amdgpu_atombios_encoder_get_backlight_brightness(struct backlight_device *bd)
 	struct amdgpu_backlight_privdata *pdata = bl_get_data(bd);
 	struct amdgpu_encoder *amdgpu_encoder = pdata->encoder;
 	struct drm_device *dev = amdgpu_encoder->base.dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 
 	return amdgpu_atombios_encoder_get_backlight_level_from_reg(adev);
 }
@@ -163,12 +166,11 @@ void amdgpu_atombios_encoder_init_backlight(struct amdgpu_encoder *amdgpu_encode
 				     struct drm_connector *drm_connector)
 {
 	struct drm_device *dev = amdgpu_encoder->base.dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct backlight_device *bd;
 	struct backlight_properties props;
 	struct amdgpu_backlight_privdata *pdata;
 	struct amdgpu_encoder_atom_dig *dig;
-	u8 backlight_level;
 	char bl_name[16];
 
 	/* Mac laptops with multiple GPUs use the gmux driver for backlight
@@ -182,7 +184,12 @@ void amdgpu_atombios_encoder_init_backlight(struct amdgpu_encoder *amdgpu_encode
 		return;
 
 	if (!(adev->mode_info.firmware_flags & ATOM_BIOS_INFO_BL_CONTROLLED_BY_GPU))
-		return;
+		goto register_acpi_backlight;
+
+	if (!acpi_video_backlight_use_native()) {
+		drm_info(dev, "Skipping amdgpu atom DIG backlight registration\n");
+		goto register_acpi_backlight;
+	}
 
 	pdata = kmalloc(sizeof(struct amdgpu_backlight_privdata), GFP_KERNEL);
 	if (!pdata) {
@@ -204,13 +211,11 @@ void amdgpu_atombios_encoder_init_backlight(struct amdgpu_encoder *amdgpu_encode
 
 	pdata->encoder = amdgpu_encoder;
 
-	backlight_level = amdgpu_atombios_encoder_get_backlight_level_from_reg(adev);
-
 	dig = amdgpu_encoder->enc_priv;
 	dig->bl_dev = bd;
 
 	bd->props.brightness = amdgpu_atombios_encoder_get_backlight_brightness(bd);
-	bd->props.power = FB_BLANK_UNBLANK;
+	bd->props.power = BACKLIGHT_POWER_ON;
 	backlight_update_status(bd);
 
 	DRM_INFO("amdgpu atom DIG backlight initialized\n");
@@ -220,13 +225,17 @@ void amdgpu_atombios_encoder_init_backlight(struct amdgpu_encoder *amdgpu_encode
 error:
 	kfree(pdata);
 	return;
+
+register_acpi_backlight:
+	/* Try registering an ACPI video backlight device instead. */
+	acpi_video_register_backlight();
 }
 
 void
 amdgpu_atombios_encoder_fini_backlight(struct amdgpu_encoder *amdgpu_encoder)
 {
 	struct drm_device *dev = amdgpu_encoder->base.dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct backlight_device *bd = NULL;
 	struct amdgpu_encoder_atom_dig *dig;
 
@@ -251,18 +260,6 @@ amdgpu_atombios_encoder_fini_backlight(struct amdgpu_encoder *amdgpu_encoder)
 	}
 }
 
-#else /* !CONFIG_BACKLIGHT_CLASS_DEVICE */
-
-void amdgpu_atombios_encoder_init_backlight(struct amdgpu_encoder *encoder)
-{
-}
-
-void amdgpu_atombios_encoder_fini_backlight(struct amdgpu_encoder *encoder)
-{
-}
-
-#endif
-
 bool amdgpu_atombios_encoder_is_digital(struct drm_encoder *encoder)
 {
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
@@ -316,7 +313,7 @@ static void
 amdgpu_atombios_encoder_setup_dac(struct drm_encoder *encoder, int action)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	DAC_ENCODER_CONTROL_PS_ALLOCATION args;
 	int index = 0;
@@ -338,7 +335,7 @@ amdgpu_atombios_encoder_setup_dac(struct drm_encoder *encoder, int action)
 	args.ucDacStandard = ATOM_DAC1_PS2;
 	args.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
 
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 
 }
 
@@ -379,7 +376,7 @@ static void
 amdgpu_atombios_encoder_setup_dvo(struct drm_encoder *encoder, int action)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	union dvo_encoder_control args;
 	int index = GetIndexIntoMasterTable(COMMAND, DVOEncoderControl);
@@ -435,7 +432,7 @@ amdgpu_atombios_encoder_setup_dvo(struct drm_encoder *encoder, int action)
 		break;
 	}
 
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 }
 
 int amdgpu_atombios_encoder_get_encoder_mode(struct drm_encoder *encoder)
@@ -469,7 +466,7 @@ int amdgpu_atombios_encoder_get_encoder_mode(struct drm_encoder *encoder)
 			if (amdgpu_connector->use_digital &&
 			    (amdgpu_connector->audio == AMDGPU_AUDIO_ENABLE))
 				return ATOM_ENCODER_MODE_HDMI;
-			else if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector)) &&
+			else if (connector->display_info.is_hdmi &&
 				 (amdgpu_connector->audio == AMDGPU_AUDIO_AUTO))
 				return ATOM_ENCODER_MODE_HDMI;
 			else if (amdgpu_connector->use_digital)
@@ -488,7 +485,7 @@ int amdgpu_atombios_encoder_get_encoder_mode(struct drm_encoder *encoder)
 		if (amdgpu_audio != 0) {
 			if (amdgpu_connector->audio == AMDGPU_AUDIO_ENABLE)
 				return ATOM_ENCODER_MODE_HDMI;
-			else if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector)) &&
+			else if (connector->display_info.is_hdmi &&
 				 (amdgpu_connector->audio == AMDGPU_AUDIO_AUTO))
 				return ATOM_ENCODER_MODE_HDMI;
 			else
@@ -496,10 +493,8 @@ int amdgpu_atombios_encoder_get_encoder_mode(struct drm_encoder *encoder)
 		} else {
 			return ATOM_ENCODER_MODE_DVI;
 		}
-		break;
 	case DRM_MODE_CONNECTOR_LVDS:
 		return ATOM_ENCODER_MODE_LVDS;
-		break;
 	case DRM_MODE_CONNECTOR_DisplayPort:
 		dig_connector = amdgpu_connector->con_priv;
 		if ((dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) ||
@@ -508,7 +503,7 @@ int amdgpu_atombios_encoder_get_encoder_mode(struct drm_encoder *encoder)
 		} else if (amdgpu_audio != 0) {
 			if (amdgpu_connector->audio == AMDGPU_AUDIO_ENABLE)
 				return ATOM_ENCODER_MODE_HDMI;
-			else if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector)) &&
+			else if (connector->display_info.is_hdmi &&
 				 (amdgpu_connector->audio == AMDGPU_AUDIO_AUTO))
 				return ATOM_ENCODER_MODE_HDMI;
 			else
@@ -516,20 +511,16 @@ int amdgpu_atombios_encoder_get_encoder_mode(struct drm_encoder *encoder)
 		} else {
 			return ATOM_ENCODER_MODE_DVI;
 		}
-		break;
 	case DRM_MODE_CONNECTOR_eDP:
 		return ATOM_ENCODER_MODE_DP;
 	case DRM_MODE_CONNECTOR_DVIA:
 	case DRM_MODE_CONNECTOR_VGA:
 		return ATOM_ENCODER_MODE_CRT;
-		break;
 	case DRM_MODE_CONNECTOR_Composite:
 	case DRM_MODE_CONNECTOR_SVIDEO:
 	case DRM_MODE_CONNECTOR_9PinDIN:
 		/* fix me */
 		return ATOM_ENCODER_MODE_TV;
-		/*return ATOM_ENCODER_MODE_CV;*/
-		break;
 	}
 }
 
@@ -570,7 +561,7 @@ amdgpu_atombios_encoder_setup_dig_encoder(struct drm_encoder *encoder,
 				   int action, int panel_mode)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
@@ -741,7 +732,7 @@ amdgpu_atombios_encoder_setup_dig_encoder(struct drm_encoder *encoder,
 		break;
 	}
 
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 
 }
 
@@ -759,7 +750,7 @@ amdgpu_atombios_encoder_setup_dig_transmitter(struct drm_encoder *encoder, int a
 					      uint8_t lane_num, uint8_t lane_set)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector;
@@ -771,7 +762,6 @@ amdgpu_atombios_encoder_setup_dig_transmitter(struct drm_encoder *encoder, int a
 	int dp_clock = 0;
 	int dp_lane_count = 0;
 	int connector_object_id = 0;
-	int igp_lane_info = 0;
 	int dig_encoder = dig->dig_encoder;
 	int hpd_id = AMDGPU_HPD_NONE;
 
@@ -854,26 +844,6 @@ amdgpu_atombios_encoder_setup_dig_transmitter(struct drm_encoder *encoder, int a
 			else
 				args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_DIG1_ENCODER;
 
-			if ((adev->flags & AMD_IS_APU) &&
-			    (amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_UNIPHY)) {
-				if (is_dp ||
-				    !amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock)) {
-					if (igp_lane_info & 0x1)
-						args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_0_3;
-					else if (igp_lane_info & 0x2)
-						args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_4_7;
-					else if (igp_lane_info & 0x4)
-						args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_8_11;
-					else if (igp_lane_info & 0x8)
-						args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_12_15;
-				} else {
-					if (igp_lane_info & 0x3)
-						args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_0_7;
-					else if (igp_lane_info & 0xc)
-						args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_8_15;
-				}
-			}
-
 			if (dig->linkb)
 				args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LINKB;
 			else
@@ -1166,7 +1136,7 @@ amdgpu_atombios_encoder_setup_dig_transmitter(struct drm_encoder *encoder, int a
 		break;
 	}
 
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 }
 
 bool
@@ -1175,7 +1145,7 @@ amdgpu_atombios_encoder_set_edp_panel_power(struct drm_connector *connector,
 {
 	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 	struct drm_device *dev = amdgpu_connector->base.dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	union dig_transmitter_control args;
 	int index = GetIndexIntoMasterTable(COMMAND, UNIPHYTransmitterControl);
 	uint8_t frev, crev;
@@ -1194,7 +1164,7 @@ amdgpu_atombios_encoder_set_edp_panel_power(struct drm_connector *connector,
 
 	args.v1.ucAction = action;
 
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 
 	/* wait for the panel to power up */
 	if (action == ATOM_TRANSMITTER_ACTION_POWER_ON) {
@@ -1222,7 +1192,7 @@ amdgpu_atombios_encoder_setup_external_encoder(struct drm_encoder *encoder,
 					int action)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder *ext_amdgpu_encoder = to_amdgpu_encoder(ext_encoder);
 	union external_encoder_control args;
@@ -1318,7 +1288,7 @@ amdgpu_atombios_encoder_setup_external_encoder(struct drm_encoder *encoder,
 		DRM_ERROR("Unknown table version: %d, %d\n", frev, crev);
 		return;
 	}
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 }
 
 static void
@@ -1463,7 +1433,7 @@ void
 amdgpu_atombios_encoder_set_crtc_source(struct drm_encoder *encoder)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
 	union crtc_source_param args;
@@ -1663,14 +1633,14 @@ amdgpu_atombios_encoder_set_crtc_source(struct drm_encoder *encoder)
 		return;
 	}
 
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 }
 
 /* This only needs to be called once at startup */
 void
 amdgpu_atombios_encoder_init_dig(struct amdgpu_device *adev)
 {
-	struct drm_device *dev = adev->ddev;
+	struct drm_device *dev = adev_to_drm(adev);
 	struct drm_encoder *encoder;
 
 	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
@@ -1698,7 +1668,7 @@ amdgpu_atombios_encoder_dac_load_detect(struct drm_encoder *encoder,
 				 struct drm_connector *connector)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 
@@ -1736,7 +1706,7 @@ amdgpu_atombios_encoder_dac_load_detect(struct drm_encoder *encoder,
 				args.sDacload.ucMisc = DAC_LOAD_MISC_YPrPb;
 		}
 
-		amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+		amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 
 		return true;
 	} else
@@ -1748,7 +1718,7 @@ amdgpu_atombios_encoder_dac_detect(struct drm_encoder *encoder,
 			    struct drm_connector *connector)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 	uint32_t bios_0_scratch;
@@ -1787,7 +1757,7 @@ amdgpu_atombios_encoder_dig_detect(struct drm_encoder *encoder,
 			    struct drm_connector *connector)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 	struct drm_encoder *ext_encoder = amdgpu_get_external_encoder(encoder);
@@ -1845,7 +1815,7 @@ amdgpu_atombios_encoder_set_bios_scratch_regs(struct drm_connector *connector,
 				       bool connected)
 {
 	struct drm_device *dev = connector->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_connector *amdgpu_connector =
 	    to_amdgpu_connector(connector);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
@@ -1996,7 +1966,7 @@ struct amdgpu_encoder_atom_dig *
 amdgpu_atombios_encoder_get_lcd_info(struct amdgpu_encoder *encoder)
 {
 	struct drm_device *dev = encoder->base.dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_mode_info *mode_info = &adev->mode_info;
 	int index = GetIndexIntoMasterTable(DATA, LVDS_Info);
 	uint16_t data_offset, misc;
@@ -2094,24 +2064,25 @@ amdgpu_atombios_encoder_get_lcd_info(struct amdgpu_encoder *encoder)
 				case LCD_FAKE_EDID_PATCH_RECORD_TYPE:
 					fake_edid_record = (ATOM_FAKE_EDID_PATCH_RECORD *)record;
 					if (fake_edid_record->ucFakeEDIDLength) {
-						struct edid *edid;
-						int edid_size =
-							max((int)EDID_LENGTH, (int)fake_edid_record->ucFakeEDIDLength);
-						edid = kmalloc(edid_size, GFP_KERNEL);
-						if (edid) {
-							memcpy((u8 *)edid, (u8 *)&fake_edid_record->ucFakeEDIDString[0],
-							       fake_edid_record->ucFakeEDIDLength);
-
-							if (drm_edid_is_valid(edid)) {
-								adev->mode_info.bios_hardcoded_edid = edid;
-								adev->mode_info.bios_hardcoded_edid_size = edid_size;
-							} else
-								kfree(edid);
-						}
+						const struct drm_edid *edid;
+						int edid_size;
+
+						if (fake_edid_record->ucFakeEDIDLength == 128)
+							edid_size = fake_edid_record->ucFakeEDIDLength;
+						else
+							edid_size = fake_edid_record->ucFakeEDIDLength * 128;
+						edid = drm_edid_alloc(fake_edid_record->ucFakeEDIDString, edid_size);
+						if (drm_edid_valid(edid))
+							adev->mode_info.bios_hardcoded_edid = edid;
+						else
+							drm_edid_free(edid);
+						record += struct_size(fake_edid_record,
+								      ucFakeEDIDString,
+								      edid_size);
+					} else {
+						/* empty fake edid record must be 3 bytes long */
+						record += sizeof(ATOM_FAKE_EDID_PATCH_RECORD) + 1;
 					}
-					record += fake_edid_record->ucFakeEDIDLength ?
-						fake_edid_record->ucFakeEDIDLength + 2 :
-						sizeof(ATOM_FAKE_EDID_PATCH_RECORD);
 					break;
 				case LCD_PANEL_RESOLUTION_RECORD_TYPE:
 					panel_res_record = (ATOM_PANEL_RESOLUTION_PATCH_RECORD *)record;
diff --git a/drivers/gpu/drm/amd/amdgpu/atombios_encoders.h b/drivers/gpu/drm/amd/amdgpu/atombios_encoders.h
index 2bdec40515ce..f77cbdef679e 100644
--- a/drivers/gpu/drm/amd/amdgpu/atombios_encoders.h
+++ b/drivers/gpu/drm/amd/amdgpu/atombios_encoders.h
@@ -25,6 +25,11 @@
 #define __ATOMBIOS_ENCODER_H__
 
 u8
+amdgpu_atombios_encoder_get_backlight_level_from_reg(struct amdgpu_device *adev);
+void
+amdgpu_atombios_encoder_set_backlight_level_to_reg(struct amdgpu_device *adev,
+						   u8 backlight_level);
+u8
 amdgpu_atombios_encoder_get_backlight_level(struct amdgpu_encoder *amdgpu_encoder);
 void
 amdgpu_atombios_encoder_set_backlight_level(struct amdgpu_encoder *amdgpu_encoder,
diff --git a/drivers/gpu/drm/amd/amdgpu/atombios_i2c.c b/drivers/gpu/drm/amd/amdgpu/atombios_i2c.c
index b374653bd6cf..a6501114322f 100644
--- a/drivers/gpu/drm/amd/amdgpu/atombios_i2c.c
+++ b/drivers/gpu/drm/amd/amdgpu/atombios_i2c.c
@@ -22,7 +22,7 @@
  * Authors: Alex Deucher
  *
  */
-#include <drm/drmP.h>
+
 #include <drm/amdgpu_drm.h>
 #include "amdgpu.h"
 #include "atom.h"
@@ -40,7 +40,7 @@ static int amdgpu_atombios_i2c_process_i2c_ch(struct amdgpu_i2c_chan *chan,
 				       u8 *buf, u8 num)
 {
 	struct drm_device *dev = chan->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	PROCESS_I2C_CHANNEL_TRANSACTION_PS_ALLOCATION args;
 	int index = GetIndexIntoMasterTable(COMMAND, ProcessI2cChannelTransaction);
 	unsigned char *base;
@@ -65,15 +65,17 @@ static int amdgpu_atombios_i2c_process_i2c_ch(struct amdgpu_i2c_chan *chan,
 			args.ucRegIndex = buf[0];
 		if (num)
 			num--;
-		if (num)
-			memcpy(&out, &buf[1], num);
+		if (num) {
+			if (buf) {
+				memcpy(&out, &buf[1], num);
+			} else {
+				DRM_ERROR("hw i2c: missing buf with num > 1\n");
+				r = -EINVAL;
+				goto done;
+			}
+		}
 		args.lpI2CDataOut = cpu_to_le16(out);
 	} else {
-		if (num > ATOM_MAX_HW_I2C_READ) {
-			DRM_ERROR("hw i2c: tried to read too many bytes (%d vs 255)\n", num);
-			r = -EINVAL;
-			goto done;
-		}
 		args.ucRegIndex = 0;
 		args.lpI2CDataOut = 0;
 	}
@@ -84,7 +86,7 @@ static int amdgpu_atombios_i2c_process_i2c_ch(struct amdgpu_i2c_chan *chan,
 	args.ucSlaveAddr = slave_addr << 1;
 	args.ucLineNumber = chan->rec.i2c_id;
 
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 
 	/* error */
 	if (args.ucStatus != HW_ASSISTED_I2C_STATUS_SUCCESS) {
@@ -157,7 +159,7 @@ u32 amdgpu_atombios_i2c_func(struct i2c_adapter *adap)
 	return I2C_FUNC_I2C | I2C_FUNC_SMBUS_EMUL;
 }
 
-void amdgpu_atombios_i2c_channel_trans(struct amdgpu_device* adev, u8 slave_addr, u8 line_number, u8 offset, u8 data)
+void amdgpu_atombios_i2c_channel_trans(struct amdgpu_device *adev, u8 slave_addr, u8 line_number, u8 offset, u8 data)
 {
 	PROCESS_I2C_CHANNEL_TRANSACTION_PS_ALLOCATION args;
 	int index = GetIndexIntoMasterTable(COMMAND, ProcessI2cChannelTransaction);
@@ -170,5 +172,5 @@ void amdgpu_atombios_i2c_channel_trans(struct amdgpu_device* adev, u8 slave_addr
 	args.ucSlaveAddr = slave_addr;
 	args.ucLineNumber = line_number;
 
-	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/ci_dpm.c b/drivers/gpu/drm/amd/amdgpu/ci_dpm.c
deleted file mode 100644
index cb508a211b2f..000000000000
--- a/drivers/gpu/drm/amd/amdgpu/ci_dpm.c
+++ /dev/null
@@ -1,7053 +0,0 @@
-/*
- * Copyright 2013 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-
-#include <linux/firmware.h>
-#include <drm/drmP.h>
-#include "amdgpu.h"
-#include "amdgpu_pm.h"
-#include "amdgpu_ucode.h"
-#include "cikd.h"
-#include "amdgpu_dpm.h"
-#include "ci_dpm.h"
-#include "gfx_v7_0.h"
-#include "atom.h"
-#include "amd_pcie.h"
-#include <linux/seq_file.h>
-
-#include "smu/smu_7_0_1_d.h"
-#include "smu/smu_7_0_1_sh_mask.h"
-
-#include "dce/dce_8_0_d.h"
-#include "dce/dce_8_0_sh_mask.h"
-
-#include "bif/bif_4_1_d.h"
-#include "bif/bif_4_1_sh_mask.h"
-
-#include "gca/gfx_7_2_d.h"
-#include "gca/gfx_7_2_sh_mask.h"
-
-#include "gmc/gmc_7_1_d.h"
-#include "gmc/gmc_7_1_sh_mask.h"
-
-MODULE_FIRMWARE("radeon/bonaire_smc.bin");
-MODULE_FIRMWARE("radeon/bonaire_k_smc.bin");
-MODULE_FIRMWARE("radeon/hawaii_smc.bin");
-MODULE_FIRMWARE("radeon/hawaii_k_smc.bin");
-
-#define MC_CG_ARB_FREQ_F0           0x0a
-#define MC_CG_ARB_FREQ_F1           0x0b
-#define MC_CG_ARB_FREQ_F2           0x0c
-#define MC_CG_ARB_FREQ_F3           0x0d
-
-#define SMC_RAM_END 0x40000
-
-#define VOLTAGE_SCALE               4
-#define VOLTAGE_VID_OFFSET_SCALE1    625
-#define VOLTAGE_VID_OFFSET_SCALE2    100
-
-static const struct ci_pt_defaults defaults_hawaii_xt =
-{
-	1, 0xF, 0xFD, 0x19, 5, 0x14, 0, 0xB0000,
-	{ 0x2E,  0x00,  0x00,  0x88,  0x00,  0x00,  0x72,  0x60,  0x51,  0xA7,  0x79,  0x6B,  0x90,  0xBD,  0x79  },
-	{ 0x217, 0x217, 0x217, 0x242, 0x242, 0x242, 0x269, 0x269, 0x269, 0x2A1, 0x2A1, 0x2A1, 0x2C9, 0x2C9, 0x2C9 }
-};
-
-static const struct ci_pt_defaults defaults_hawaii_pro =
-{
-	1, 0xF, 0xFD, 0x19, 5, 0x14, 0, 0x65062,
-	{ 0x2E,  0x00,  0x00,  0x88,  0x00,  0x00,  0x72,  0x60,  0x51,  0xA7,  0x79,  0x6B,  0x90,  0xBD,  0x79  },
-	{ 0x217, 0x217, 0x217, 0x242, 0x242, 0x242, 0x269, 0x269, 0x269, 0x2A1, 0x2A1, 0x2A1, 0x2C9, 0x2C9, 0x2C9 }
-};
-
-static const struct ci_pt_defaults defaults_bonaire_xt =
-{
-	1, 0xF, 0xFD, 0x19, 5, 45, 0, 0xB0000,
-	{ 0x79,  0x253, 0x25D, 0xAE,  0x72,  0x80,  0x83,  0x86,  0x6F,  0xC8,  0xC9,  0xC9,  0x2F,  0x4D,  0x61  },
-	{ 0x17C, 0x172, 0x180, 0x1BC, 0x1B3, 0x1BD, 0x206, 0x200, 0x203, 0x25D, 0x25A, 0x255, 0x2C3, 0x2C5, 0x2B4 }
-};
-
-#if 0
-static const struct ci_pt_defaults defaults_bonaire_pro =
-{
-	1, 0xF, 0xFD, 0x19, 5, 45, 0, 0x65062,
-	{ 0x8C,  0x23F, 0x244, 0xA6,  0x83,  0x85,  0x86,  0x86,  0x83,  0xDB,  0xDB,  0xDA,  0x67,  0x60,  0x5F  },
-	{ 0x187, 0x193, 0x193, 0x1C7, 0x1D1, 0x1D1, 0x210, 0x219, 0x219, 0x266, 0x26C, 0x26C, 0x2C9, 0x2CB, 0x2CB }
-};
-#endif
-
-static const struct ci_pt_defaults defaults_saturn_xt =
-{
-	1, 0xF, 0xFD, 0x19, 5, 55, 0, 0x70000,
-	{ 0x8C,  0x247, 0x249, 0xA6,  0x80,  0x81,  0x8B,  0x89,  0x86,  0xC9,  0xCA,  0xC9,  0x4D,  0x4D,  0x4D  },
-	{ 0x187, 0x187, 0x187, 0x1C7, 0x1C7, 0x1C7, 0x210, 0x210, 0x210, 0x266, 0x266, 0x266, 0x2C9, 0x2C9, 0x2C9 }
-};
-
-#if 0
-static const struct ci_pt_defaults defaults_saturn_pro =
-{
-	1, 0xF, 0xFD, 0x19, 5, 55, 0, 0x30000,
-	{ 0x96,  0x21D, 0x23B, 0xA1,  0x85,  0x87,  0x83,  0x84,  0x81,  0xE6,  0xE6,  0xE6,  0x71,  0x6A,  0x6A  },
-	{ 0x193, 0x19E, 0x19E, 0x1D2, 0x1DC, 0x1DC, 0x21A, 0x223, 0x223, 0x26E, 0x27E, 0x274, 0x2CF, 0x2D2, 0x2D2 }
-};
-#endif
-
-static const struct ci_pt_config_reg didt_config_ci[] =
-{
-	{ 0x10, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x10, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x10, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x10, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x11, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x11, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x11, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x11, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x12, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x12, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x12, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x12, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x2, 0x00003fff, 0, 0x4, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x2, 0x03ff0000, 16, 0x80, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x2, 0x78000000, 27, 0x3, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x1, 0x0000ffff, 0, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x1, 0xffff0000, 16, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x0, 0x00000001, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x30, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x30, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x30, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x30, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x31, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x31, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x31, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x31, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x32, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x32, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x32, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x32, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x22, 0x00003fff, 0, 0x4, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x22, 0x03ff0000, 16, 0x80, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x22, 0x78000000, 27, 0x3, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x21, 0x0000ffff, 0, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x21, 0xffff0000, 16, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x20, 0x00000001, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x50, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x50, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x50, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x50, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x51, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x51, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x51, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x51, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x52, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x52, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x52, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x52, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x42, 0x00003fff, 0, 0x4, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x42, 0x03ff0000, 16, 0x80, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x42, 0x78000000, 27, 0x3, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x41, 0x0000ffff, 0, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x41, 0xffff0000, 16, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x40, 0x00000001, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x70, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x70, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x70, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x70, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x71, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x71, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x71, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x71, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x72, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x72, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x72, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x72, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x62, 0x00003fff, 0, 0x4, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x62, 0x03ff0000, 16, 0x80, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x62, 0x78000000, 27, 0x3, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x61, 0x0000ffff, 0, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x61, 0xffff0000, 16, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0x60, 0x00000001, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
-	{ 0xFFFFFFFF }
-};
-
-static u8 ci_get_memory_module_index(struct amdgpu_device *adev)
-{
-	return (u8) ((RREG32(mmBIOS_SCRATCH_4) >> 16) & 0xff);
-}
-
-#define MC_CG_ARB_FREQ_F0           0x0a
-#define MC_CG_ARB_FREQ_F1           0x0b
-#define MC_CG_ARB_FREQ_F2           0x0c
-#define MC_CG_ARB_FREQ_F3           0x0d
-
-static int ci_copy_and_switch_arb_sets(struct amdgpu_device *adev,
-				       u32 arb_freq_src, u32 arb_freq_dest)
-{
-	u32 mc_arb_dram_timing;
-	u32 mc_arb_dram_timing2;
-	u32 burst_time;
-	u32 mc_cg_config;
-
-	switch (arb_freq_src) {
-	case MC_CG_ARB_FREQ_F0:
-		mc_arb_dram_timing  = RREG32(mmMC_ARB_DRAM_TIMING);
-		mc_arb_dram_timing2 = RREG32(mmMC_ARB_DRAM_TIMING2);
-		burst_time = (RREG32(mmMC_ARB_BURST_TIME) & MC_ARB_BURST_TIME__STATE0_MASK) >>
-			 MC_ARB_BURST_TIME__STATE0__SHIFT;
-		break;
-	case MC_CG_ARB_FREQ_F1:
-		mc_arb_dram_timing  = RREG32(mmMC_ARB_DRAM_TIMING_1);
-		mc_arb_dram_timing2 = RREG32(mmMC_ARB_DRAM_TIMING2_1);
-		burst_time = (RREG32(mmMC_ARB_BURST_TIME) & MC_ARB_BURST_TIME__STATE1_MASK) >>
-			 MC_ARB_BURST_TIME__STATE1__SHIFT;
-		break;
-	default:
-		return -EINVAL;
-	}
-
-	switch (arb_freq_dest) {
-	case MC_CG_ARB_FREQ_F0:
-		WREG32(mmMC_ARB_DRAM_TIMING, mc_arb_dram_timing);
-		WREG32(mmMC_ARB_DRAM_TIMING2, mc_arb_dram_timing2);
-		WREG32_P(mmMC_ARB_BURST_TIME, (burst_time << MC_ARB_BURST_TIME__STATE0__SHIFT),
-			~MC_ARB_BURST_TIME__STATE0_MASK);
-		break;
-	case MC_CG_ARB_FREQ_F1:
-		WREG32(mmMC_ARB_DRAM_TIMING_1, mc_arb_dram_timing);
-		WREG32(mmMC_ARB_DRAM_TIMING2_1, mc_arb_dram_timing2);
-		WREG32_P(mmMC_ARB_BURST_TIME, (burst_time << MC_ARB_BURST_TIME__STATE1__SHIFT),
-			~MC_ARB_BURST_TIME__STATE1_MASK);
-		break;
-	default:
-		return -EINVAL;
-	}
-
-	mc_cg_config = RREG32(mmMC_CG_CONFIG) | 0x0000000F;
-	WREG32(mmMC_CG_CONFIG, mc_cg_config);
-	WREG32_P(mmMC_ARB_CG, (arb_freq_dest) << MC_ARB_CG__CG_ARB_REQ__SHIFT,
-		~MC_ARB_CG__CG_ARB_REQ_MASK);
-
-	return 0;
-}
-
-static u8 ci_get_ddr3_mclk_frequency_ratio(u32 memory_clock)
-{
-	u8 mc_para_index;
-
-	if (memory_clock < 10000)
-		mc_para_index = 0;
-	else if (memory_clock >= 80000)
-		mc_para_index = 0x0f;
-	else
-		mc_para_index = (u8)((memory_clock - 10000) / 5000 + 1);
-	return mc_para_index;
-}
-
-static u8 ci_get_mclk_frequency_ratio(u32 memory_clock, bool strobe_mode)
-{
-	u8 mc_para_index;
-
-	if (strobe_mode) {
-		if (memory_clock < 12500)
-			mc_para_index = 0x00;
-		else if (memory_clock > 47500)
-			mc_para_index = 0x0f;
-		else
-			mc_para_index = (u8)((memory_clock - 10000) / 2500);
-	} else {
-		if (memory_clock < 65000)
-			mc_para_index = 0x00;
-		else if (memory_clock > 135000)
-			mc_para_index = 0x0f;
-		else
-			mc_para_index = (u8)((memory_clock - 60000) / 5000);
-	}
-	return mc_para_index;
-}
-
-static void ci_trim_voltage_table_to_fit_state_table(struct amdgpu_device *adev,
-						     u32 max_voltage_steps,
-						     struct atom_voltage_table *voltage_table)
-{
-	unsigned int i, diff;
-
-	if (voltage_table->count <= max_voltage_steps)
-		return;
-
-	diff = voltage_table->count - max_voltage_steps;
-
-	for (i = 0; i < max_voltage_steps; i++)
-		voltage_table->entries[i] = voltage_table->entries[i + diff];
-
-	voltage_table->count = max_voltage_steps;
-}
-
-static int ci_get_std_voltage_value_sidd(struct amdgpu_device *adev,
-					 struct atom_voltage_table_entry *voltage_table,
-					 u16 *std_voltage_hi_sidd, u16 *std_voltage_lo_sidd);
-static int ci_set_power_limit(struct amdgpu_device *adev, u32 n);
-static int ci_set_overdrive_target_tdp(struct amdgpu_device *adev,
-				       u32 target_tdp);
-static int ci_update_uvd_dpm(struct amdgpu_device *adev, bool gate);
-static void ci_dpm_set_dpm_funcs(struct amdgpu_device *adev);
-static void ci_dpm_set_irq_funcs(struct amdgpu_device *adev);
-
-static PPSMC_Result amdgpu_ci_send_msg_to_smc_with_parameter(struct amdgpu_device *adev,
-							     PPSMC_Msg msg, u32 parameter);
-static void ci_thermal_start_smc_fan_control(struct amdgpu_device *adev);
-static void ci_fan_ctrl_set_default_mode(struct amdgpu_device *adev);
-
-static struct ci_power_info *ci_get_pi(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = adev->pm.dpm.priv;
-
-	return pi;
-}
-
-static struct ci_ps *ci_get_ps(struct amdgpu_ps *rps)
-{
-	struct ci_ps *ps = rps->ps_priv;
-
-	return ps;
-}
-
-static void ci_initialize_powertune_defaults(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	switch (adev->pdev->device) {
-	case 0x6649:
-	case 0x6650:
-	case 0x6651:
-	case 0x6658:
-	case 0x665C:
-	case 0x665D:
-	default:
-		pi->powertune_defaults = &defaults_bonaire_xt;
-		break;
-	case 0x6640:
-	case 0x6641:
-	case 0x6646:
-	case 0x6647:
-		pi->powertune_defaults = &defaults_saturn_xt;
-		break;
-	case 0x67B8:
-	case 0x67B0:
-		pi->powertune_defaults = &defaults_hawaii_xt;
-		break;
-	case 0x67BA:
-	case 0x67B1:
-		pi->powertune_defaults = &defaults_hawaii_pro;
-		break;
-	case 0x67A0:
-	case 0x67A1:
-	case 0x67A2:
-	case 0x67A8:
-	case 0x67A9:
-	case 0x67AA:
-	case 0x67B9:
-	case 0x67BE:
-		pi->powertune_defaults = &defaults_bonaire_xt;
-		break;
-	}
-
-	pi->dte_tj_offset = 0;
-
-	pi->caps_power_containment = true;
-	pi->caps_cac = false;
-	pi->caps_sq_ramping = false;
-	pi->caps_db_ramping = false;
-	pi->caps_td_ramping = false;
-	pi->caps_tcp_ramping = false;
-
-	if (pi->caps_power_containment) {
-		pi->caps_cac = true;
-		if (adev->asic_type == CHIP_HAWAII)
-			pi->enable_bapm_feature = false;
-		else
-			pi->enable_bapm_feature = true;
-		pi->enable_tdc_limit_feature = true;
-		pi->enable_pkg_pwr_tracking_feature = true;
-	}
-}
-
-static u8 ci_convert_to_vid(u16 vddc)
-{
-	return (6200 - (vddc * VOLTAGE_SCALE)) / 25;
-}
-
-static int ci_populate_bapm_vddc_vid_sidd(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u8 *hi_vid = pi->smc_powertune_table.BapmVddCVidHiSidd;
-	u8 *lo_vid = pi->smc_powertune_table.BapmVddCVidLoSidd;
-	u8 *hi2_vid = pi->smc_powertune_table.BapmVddCVidHiSidd2;
-	u32 i;
-
-	if (adev->pm.dpm.dyn_state.cac_leakage_table.entries == NULL)
-		return -EINVAL;
-	if (adev->pm.dpm.dyn_state.cac_leakage_table.count > 8)
-		return -EINVAL;
-	if (adev->pm.dpm.dyn_state.cac_leakage_table.count !=
-	    adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.count)
-		return -EINVAL;
-
-	for (i = 0; i < adev->pm.dpm.dyn_state.cac_leakage_table.count; i++) {
-		if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_EVV) {
-			lo_vid[i] = ci_convert_to_vid(adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc1);
-			hi_vid[i] = ci_convert_to_vid(adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc2);
-			hi2_vid[i] = ci_convert_to_vid(adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc3);
-		} else {
-			lo_vid[i] = ci_convert_to_vid(adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc);
-			hi_vid[i] = ci_convert_to_vid((u16)adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].leakage);
-		}
-	}
-	return 0;
-}
-
-static int ci_populate_vddc_vid(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u8 *vid = pi->smc_powertune_table.VddCVid;
-	u32 i;
-
-	if (pi->vddc_voltage_table.count > 8)
-		return -EINVAL;
-
-	for (i = 0; i < pi->vddc_voltage_table.count; i++)
-		vid[i] = ci_convert_to_vid(pi->vddc_voltage_table.entries[i].value);
-
-	return 0;
-}
-
-static int ci_populate_svi_load_line(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	const struct ci_pt_defaults *pt_defaults = pi->powertune_defaults;
-
-	pi->smc_powertune_table.SviLoadLineEn = pt_defaults->svi_load_line_en;
-	pi->smc_powertune_table.SviLoadLineVddC = pt_defaults->svi_load_line_vddc;
-	pi->smc_powertune_table.SviLoadLineTrimVddC = 3;
-	pi->smc_powertune_table.SviLoadLineOffsetVddC = 0;
-
-	return 0;
-}
-
-static int ci_populate_tdc_limit(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	const struct ci_pt_defaults *pt_defaults = pi->powertune_defaults;
-	u16 tdc_limit;
-
-	tdc_limit = adev->pm.dpm.dyn_state.cac_tdp_table->tdc * 256;
-	pi->smc_powertune_table.TDC_VDDC_PkgLimit = cpu_to_be16(tdc_limit);
-	pi->smc_powertune_table.TDC_VDDC_ThrottleReleaseLimitPerc =
-		pt_defaults->tdc_vddc_throttle_release_limit_perc;
-	pi->smc_powertune_table.TDC_MAWt = pt_defaults->tdc_mawt;
-
-	return 0;
-}
-
-static int ci_populate_dw8(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	const struct ci_pt_defaults *pt_defaults = pi->powertune_defaults;
-	int ret;
-
-	ret = amdgpu_ci_read_smc_sram_dword(adev,
-				     SMU7_FIRMWARE_HEADER_LOCATION +
-				     offsetof(SMU7_Firmware_Header, PmFuseTable) +
-				     offsetof(SMU7_Discrete_PmFuses, TdcWaterfallCtl),
-				     (u32 *)&pi->smc_powertune_table.TdcWaterfallCtl,
-				     pi->sram_end);
-	if (ret)
-		return -EINVAL;
-	else
-		pi->smc_powertune_table.TdcWaterfallCtl = pt_defaults->tdc_waterfall_ctl;
-
-	return 0;
-}
-
-static int ci_populate_fuzzy_fan(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	if ((adev->pm.dpm.fan.fan_output_sensitivity & (1 << 15)) ||
-	    (adev->pm.dpm.fan.fan_output_sensitivity == 0))
-		adev->pm.dpm.fan.fan_output_sensitivity =
-			adev->pm.dpm.fan.default_fan_output_sensitivity;
-
-	pi->smc_powertune_table.FuzzyFan_PwmSetDelta =
-		cpu_to_be16(adev->pm.dpm.fan.fan_output_sensitivity);
-
-	return 0;
-}
-
-static int ci_min_max_v_gnbl_pm_lid_from_bapm_vddc(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u8 *hi_vid = pi->smc_powertune_table.BapmVddCVidHiSidd;
-	u8 *lo_vid = pi->smc_powertune_table.BapmVddCVidLoSidd;
-	int i, min, max;
-
-	min = max = hi_vid[0];
-	for (i = 0; i < 8; i++) {
-		if (0 != hi_vid[i]) {
-			if (min > hi_vid[i])
-				min = hi_vid[i];
-			if (max < hi_vid[i])
-				max = hi_vid[i];
-		}
-
-		if (0 != lo_vid[i]) {
-			if (min > lo_vid[i])
-				min = lo_vid[i];
-			if (max < lo_vid[i])
-				max = lo_vid[i];
-		}
-	}
-
-	if ((min == 0) || (max == 0))
-		return -EINVAL;
-	pi->smc_powertune_table.GnbLPMLMaxVid = (u8)max;
-	pi->smc_powertune_table.GnbLPMLMinVid = (u8)min;
-
-	return 0;
-}
-
-static int ci_populate_bapm_vddc_base_leakage_sidd(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u16 hi_sidd = pi->smc_powertune_table.BapmVddCBaseLeakageHiSidd;
-	u16 lo_sidd = pi->smc_powertune_table.BapmVddCBaseLeakageLoSidd;
-	struct amdgpu_cac_tdp_table *cac_tdp_table =
-		adev->pm.dpm.dyn_state.cac_tdp_table;
-
-	hi_sidd = cac_tdp_table->high_cac_leakage / 100 * 256;
-	lo_sidd = cac_tdp_table->low_cac_leakage / 100 * 256;
-
-	pi->smc_powertune_table.BapmVddCBaseLeakageHiSidd = cpu_to_be16(hi_sidd);
-	pi->smc_powertune_table.BapmVddCBaseLeakageLoSidd = cpu_to_be16(lo_sidd);
-
-	return 0;
-}
-
-static int ci_populate_bapm_parameters_in_dpm_table(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	const struct ci_pt_defaults *pt_defaults = pi->powertune_defaults;
-	SMU7_Discrete_DpmTable  *dpm_table = &pi->smc_state_table;
-	struct amdgpu_cac_tdp_table *cac_tdp_table =
-		adev->pm.dpm.dyn_state.cac_tdp_table;
-	struct amdgpu_ppm_table *ppm = adev->pm.dpm.dyn_state.ppm_table;
-	int i, j, k;
-	const u16 *def1;
-	const u16 *def2;
-
-	dpm_table->DefaultTdp = cac_tdp_table->tdp * 256;
-	dpm_table->TargetTdp = cac_tdp_table->configurable_tdp * 256;
-
-	dpm_table->DTETjOffset = (u8)pi->dte_tj_offset;
-	dpm_table->GpuTjMax =
-		(u8)(pi->thermal_temp_setting.temperature_high / 1000);
-	dpm_table->GpuTjHyst = 8;
-
-	dpm_table->DTEAmbientTempBase = pt_defaults->dte_ambient_temp_base;
-
-	if (ppm) {
-		dpm_table->PPM_PkgPwrLimit = cpu_to_be16((u16)ppm->dgpu_tdp * 256 / 1000);
-		dpm_table->PPM_TemperatureLimit = cpu_to_be16((u16)ppm->tj_max * 256);
-	} else {
-		dpm_table->PPM_PkgPwrLimit = cpu_to_be16(0);
-		dpm_table->PPM_TemperatureLimit = cpu_to_be16(0);
-	}
-
-	dpm_table->BAPM_TEMP_GRADIENT = cpu_to_be32(pt_defaults->bapm_temp_gradient);
-	def1 = pt_defaults->bapmti_r;
-	def2 = pt_defaults->bapmti_rc;
-
-	for (i = 0; i < SMU7_DTE_ITERATIONS; i++) {
-		for (j = 0; j < SMU7_DTE_SOURCES; j++) {
-			for (k = 0; k < SMU7_DTE_SINKS; k++) {
-				dpm_table->BAPMTI_R[i][j][k] = cpu_to_be16(*def1);
-				dpm_table->BAPMTI_RC[i][j][k] = cpu_to_be16(*def2);
-				def1++;
-				def2++;
-			}
-		}
-	}
-
-	return 0;
-}
-
-static int ci_populate_pm_base(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u32 pm_fuse_table_offset;
-	int ret;
-
-	if (pi->caps_power_containment) {
-		ret = amdgpu_ci_read_smc_sram_dword(adev,
-					     SMU7_FIRMWARE_HEADER_LOCATION +
-					     offsetof(SMU7_Firmware_Header, PmFuseTable),
-					     &pm_fuse_table_offset, pi->sram_end);
-		if (ret)
-			return ret;
-		ret = ci_populate_bapm_vddc_vid_sidd(adev);
-		if (ret)
-			return ret;
-		ret = ci_populate_vddc_vid(adev);
-		if (ret)
-			return ret;
-		ret = ci_populate_svi_load_line(adev);
-		if (ret)
-			return ret;
-		ret = ci_populate_tdc_limit(adev);
-		if (ret)
-			return ret;
-		ret = ci_populate_dw8(adev);
-		if (ret)
-			return ret;
-		ret = ci_populate_fuzzy_fan(adev);
-		if (ret)
-			return ret;
-		ret = ci_min_max_v_gnbl_pm_lid_from_bapm_vddc(adev);
-		if (ret)
-			return ret;
-		ret = ci_populate_bapm_vddc_base_leakage_sidd(adev);
-		if (ret)
-			return ret;
-		ret = amdgpu_ci_copy_bytes_to_smc(adev, pm_fuse_table_offset,
-					   (u8 *)&pi->smc_powertune_table,
-					   sizeof(SMU7_Discrete_PmFuses), pi->sram_end);
-		if (ret)
-			return ret;
-	}
-
-	return 0;
-}
-
-static void ci_do_enable_didt(struct amdgpu_device *adev, const bool enable)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u32 data;
-
-	if (pi->caps_sq_ramping) {
-		data = RREG32_DIDT(ixDIDT_SQ_CTRL0);
-		if (enable)
-			data |= DIDT_SQ_CTRL0__DIDT_CTRL_EN_MASK;
-		else
-			data &= ~DIDT_SQ_CTRL0__DIDT_CTRL_EN_MASK;
-		WREG32_DIDT(ixDIDT_SQ_CTRL0, data);
-	}
-
-	if (pi->caps_db_ramping) {
-		data = RREG32_DIDT(ixDIDT_DB_CTRL0);
-		if (enable)
-			data |= DIDT_DB_CTRL0__DIDT_CTRL_EN_MASK;
-		else
-			data &= ~DIDT_DB_CTRL0__DIDT_CTRL_EN_MASK;
-		WREG32_DIDT(ixDIDT_DB_CTRL0, data);
-	}
-
-	if (pi->caps_td_ramping) {
-		data = RREG32_DIDT(ixDIDT_TD_CTRL0);
-		if (enable)
-			data |= DIDT_TD_CTRL0__DIDT_CTRL_EN_MASK;
-		else
-			data &= ~DIDT_TD_CTRL0__DIDT_CTRL_EN_MASK;
-		WREG32_DIDT(ixDIDT_TD_CTRL0, data);
-	}
-
-	if (pi->caps_tcp_ramping) {
-		data = RREG32_DIDT(ixDIDT_TCP_CTRL0);
-		if (enable)
-			data |= DIDT_TCP_CTRL0__DIDT_CTRL_EN_MASK;
-		else
-			data &= ~DIDT_TCP_CTRL0__DIDT_CTRL_EN_MASK;
-		WREG32_DIDT(ixDIDT_TCP_CTRL0, data);
-	}
-}
-
-static int ci_program_pt_config_registers(struct amdgpu_device *adev,
-					  const struct ci_pt_config_reg *cac_config_regs)
-{
-	const struct ci_pt_config_reg *config_regs = cac_config_regs;
-	u32 data;
-	u32 cache = 0;
-
-	if (config_regs == NULL)
-		return -EINVAL;
-
-	while (config_regs->offset != 0xFFFFFFFF) {
-		if (config_regs->type == CISLANDS_CONFIGREG_CACHE) {
-			cache |= ((config_regs->value << config_regs->shift) & config_regs->mask);
-		} else {
-			switch (config_regs->type) {
-			case CISLANDS_CONFIGREG_SMC_IND:
-				data = RREG32_SMC(config_regs->offset);
-				break;
-			case CISLANDS_CONFIGREG_DIDT_IND:
-				data = RREG32_DIDT(config_regs->offset);
-				break;
-			default:
-				data = RREG32(config_regs->offset);
-				break;
-			}
-
-			data &= ~config_regs->mask;
-			data |= ((config_regs->value << config_regs->shift) & config_regs->mask);
-			data |= cache;
-
-			switch (config_regs->type) {
-			case CISLANDS_CONFIGREG_SMC_IND:
-				WREG32_SMC(config_regs->offset, data);
-				break;
-			case CISLANDS_CONFIGREG_DIDT_IND:
-				WREG32_DIDT(config_regs->offset, data);
-				break;
-			default:
-				WREG32(config_regs->offset, data);
-				break;
-			}
-			cache = 0;
-		}
-		config_regs++;
-	}
-	return 0;
-}
-
-static int ci_enable_didt(struct amdgpu_device *adev, bool enable)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	int ret;
-
-	if (pi->caps_sq_ramping || pi->caps_db_ramping ||
-	    pi->caps_td_ramping || pi->caps_tcp_ramping) {
-		adev->gfx.rlc.funcs->enter_safe_mode(adev);
-
-		if (enable) {
-			ret = ci_program_pt_config_registers(adev, didt_config_ci);
-			if (ret) {
-				adev->gfx.rlc.funcs->exit_safe_mode(adev);
-				return ret;
-			}
-		}
-
-		ci_do_enable_didt(adev, enable);
-
-		adev->gfx.rlc.funcs->exit_safe_mode(adev);
-	}
-
-	return 0;
-}
-
-static int ci_enable_power_containment(struct amdgpu_device *adev, bool enable)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	PPSMC_Result smc_result;
-	int ret = 0;
-
-	if (enable) {
-		pi->power_containment_features = 0;
-		if (pi->caps_power_containment) {
-			if (pi->enable_bapm_feature) {
-				smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_EnableDTE);
-				if (smc_result != PPSMC_Result_OK)
-					ret = -EINVAL;
-				else
-					pi->power_containment_features |= POWERCONTAINMENT_FEATURE_BAPM;
-			}
-
-			if (pi->enable_tdc_limit_feature) {
-				smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_TDCLimitEnable);
-				if (smc_result != PPSMC_Result_OK)
-					ret = -EINVAL;
-				else
-					pi->power_containment_features |= POWERCONTAINMENT_FEATURE_TDCLimit;
-			}
-
-			if (pi->enable_pkg_pwr_tracking_feature) {
-				smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_PkgPwrLimitEnable);
-				if (smc_result != PPSMC_Result_OK) {
-					ret = -EINVAL;
-				} else {
-					struct amdgpu_cac_tdp_table *cac_tdp_table =
-						adev->pm.dpm.dyn_state.cac_tdp_table;
-					u32 default_pwr_limit =
-						(u32)(cac_tdp_table->maximum_power_delivery_limit * 256);
-
-					pi->power_containment_features |= POWERCONTAINMENT_FEATURE_PkgPwrLimit;
-
-					ci_set_power_limit(adev, default_pwr_limit);
-				}
-			}
-		}
-	} else {
-		if (pi->caps_power_containment && pi->power_containment_features) {
-			if (pi->power_containment_features & POWERCONTAINMENT_FEATURE_TDCLimit)
-				amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_TDCLimitDisable);
-
-			if (pi->power_containment_features & POWERCONTAINMENT_FEATURE_BAPM)
-				amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_DisableDTE);
-
-			if (pi->power_containment_features & POWERCONTAINMENT_FEATURE_PkgPwrLimit)
-				amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_PkgPwrLimitDisable);
-			pi->power_containment_features = 0;
-		}
-	}
-
-	return ret;
-}
-
-static int ci_enable_smc_cac(struct amdgpu_device *adev, bool enable)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	PPSMC_Result smc_result;
-	int ret = 0;
-
-	if (pi->caps_cac) {
-		if (enable) {
-			smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_EnableCac);
-			if (smc_result != PPSMC_Result_OK) {
-				ret = -EINVAL;
-				pi->cac_enabled = false;
-			} else {
-				pi->cac_enabled = true;
-			}
-		} else if (pi->cac_enabled) {
-			amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_DisableCac);
-			pi->cac_enabled = false;
-		}
-	}
-
-	return ret;
-}
-
-static int ci_enable_thermal_based_sclk_dpm(struct amdgpu_device *adev,
-					    bool enable)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	PPSMC_Result smc_result = PPSMC_Result_OK;
-
-	if (pi->thermal_sclk_dpm_enabled) {
-		if (enable)
-			smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_ENABLE_THERMAL_DPM);
-		else
-			smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_DISABLE_THERMAL_DPM);
-	}
-
-	if (smc_result == PPSMC_Result_OK)
-		return 0;
-	else
-		return -EINVAL;
-}
-
-static int ci_power_control_set_level(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct amdgpu_cac_tdp_table *cac_tdp_table =
-		adev->pm.dpm.dyn_state.cac_tdp_table;
-	s32 adjust_percent;
-	s32 target_tdp;
-	int ret = 0;
-	bool adjust_polarity = false; /* ??? */
-
-	if (pi->caps_power_containment) {
-		adjust_percent = adjust_polarity ?
-			adev->pm.dpm.tdp_adjustment : (-1 * adev->pm.dpm.tdp_adjustment);
-		target_tdp = ((100 + adjust_percent) *
-			      (s32)cac_tdp_table->configurable_tdp) / 100;
-
-		ret = ci_set_overdrive_target_tdp(adev, (u32)target_tdp);
-	}
-
-	return ret;
-}
-
-static void ci_dpm_powergate_uvd(struct amdgpu_device *adev, bool gate)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	pi->uvd_power_gated = gate;
-
-	if (gate) {
-		/* stop the UVD block */
-		amdgpu_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
-							AMD_PG_STATE_GATE);
-		ci_update_uvd_dpm(adev, gate);
-	} else {
-		amdgpu_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
-							AMD_PG_STATE_UNGATE);
-		ci_update_uvd_dpm(adev, gate);
-	}
-}
-
-static bool ci_dpm_vblank_too_short(struct amdgpu_device *adev)
-{
-	u32 vblank_time = amdgpu_dpm_get_vblank_time(adev);
-	u32 switch_limit = adev->mc.vram_type == AMDGPU_VRAM_TYPE_GDDR5 ? 450 : 300;
-
-	/* disable mclk switching if the refresh is >120Hz, even if the
-	 * blanking period would allow it
-	 */
-	if (amdgpu_dpm_get_vrefresh(adev) > 120)
-		return true;
-
-	if (vblank_time < switch_limit)
-		return true;
-	else
-		return false;
-
-}
-
-static void ci_apply_state_adjust_rules(struct amdgpu_device *adev,
-					struct amdgpu_ps *rps)
-{
-	struct ci_ps *ps = ci_get_ps(rps);
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct amdgpu_clock_and_voltage_limits *max_limits;
-	bool disable_mclk_switching;
-	u32 sclk, mclk;
-	int i;
-
-	if (rps->vce_active) {
-		rps->evclk = adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].evclk;
-		rps->ecclk = adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].ecclk;
-	} else {
-		rps->evclk = 0;
-		rps->ecclk = 0;
-	}
-
-	if ((adev->pm.dpm.new_active_crtc_count > 1) ||
-	    ci_dpm_vblank_too_short(adev))
-		disable_mclk_switching = true;
-	else
-		disable_mclk_switching = false;
-
-	if ((rps->class & ATOM_PPLIB_CLASSIFICATION_UI_MASK) == ATOM_PPLIB_CLASSIFICATION_UI_BATTERY)
-		pi->battery_state = true;
-	else
-		pi->battery_state = false;
-
-	if (adev->pm.dpm.ac_power)
-		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
-	else
-		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_dc;
-
-	if (adev->pm.dpm.ac_power == false) {
-		for (i = 0; i < ps->performance_level_count; i++) {
-			if (ps->performance_levels[i].mclk > max_limits->mclk)
-				ps->performance_levels[i].mclk = max_limits->mclk;
-			if (ps->performance_levels[i].sclk > max_limits->sclk)
-				ps->performance_levels[i].sclk = max_limits->sclk;
-		}
-	}
-
-	/* XXX validate the min clocks required for display */
-
-	if (disable_mclk_switching) {
-		mclk  = ps->performance_levels[ps->performance_level_count - 1].mclk;
-		sclk = ps->performance_levels[0].sclk;
-	} else {
-		mclk = ps->performance_levels[0].mclk;
-		sclk = ps->performance_levels[0].sclk;
-	}
-
-	if (adev->pm.pm_display_cfg.min_core_set_clock > sclk)
-		sclk = adev->pm.pm_display_cfg.min_core_set_clock;
-
-	if (adev->pm.pm_display_cfg.min_mem_set_clock > mclk)
-		mclk = adev->pm.pm_display_cfg.min_mem_set_clock;
-
-	if (rps->vce_active) {
-		if (sclk < adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].sclk)
-			sclk = adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].sclk;
-		if (mclk < adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].mclk)
-			mclk = adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].mclk;
-	}
-
-	ps->performance_levels[0].sclk = sclk;
-	ps->performance_levels[0].mclk = mclk;
-
-	if (ps->performance_levels[1].sclk < ps->performance_levels[0].sclk)
-		ps->performance_levels[1].sclk = ps->performance_levels[0].sclk;
-
-	if (disable_mclk_switching) {
-		if (ps->performance_levels[0].mclk < ps->performance_levels[1].mclk)
-			ps->performance_levels[0].mclk = ps->performance_levels[1].mclk;
-	} else {
-		if (ps->performance_levels[1].mclk < ps->performance_levels[0].mclk)
-			ps->performance_levels[1].mclk = ps->performance_levels[0].mclk;
-	}
-}
-
-static int ci_thermal_set_temperature_range(struct amdgpu_device *adev,
-					    int min_temp, int max_temp)
-{
-	int low_temp = 0 * 1000;
-	int high_temp = 255 * 1000;
-	u32 tmp;
-
-	if (low_temp < min_temp)
-		low_temp = min_temp;
-	if (high_temp > max_temp)
-		high_temp = max_temp;
-	if (high_temp < low_temp) {
-		DRM_ERROR("invalid thermal range: %d - %d\n", low_temp, high_temp);
-		return -EINVAL;
-	}
-
-	tmp = RREG32_SMC(ixCG_THERMAL_INT);
-	tmp &= ~(CG_THERMAL_INT__DIG_THERM_INTH_MASK | CG_THERMAL_INT__DIG_THERM_INTL_MASK);
-	tmp |= ((high_temp / 1000) << CG_THERMAL_INT__DIG_THERM_INTH__SHIFT) |
-		((low_temp / 1000)) << CG_THERMAL_INT__DIG_THERM_INTL__SHIFT;
-	WREG32_SMC(ixCG_THERMAL_INT, tmp);
-
-#if 0
-	/* XXX: need to figure out how to handle this properly */
-	tmp = RREG32_SMC(ixCG_THERMAL_CTRL);
-	tmp &= DIG_THERM_DPM_MASK;
-	tmp |= DIG_THERM_DPM(high_temp / 1000);
-	WREG32_SMC(ixCG_THERMAL_CTRL, tmp);
-#endif
-
-	adev->pm.dpm.thermal.min_temp = low_temp;
-	adev->pm.dpm.thermal.max_temp = high_temp;
-	return 0;
-}
-
-static int ci_thermal_enable_alert(struct amdgpu_device *adev,
-				   bool enable)
-{
-	u32 thermal_int = RREG32_SMC(ixCG_THERMAL_INT);
-	PPSMC_Result result;
-
-	if (enable) {
-		thermal_int &= ~(CG_THERMAL_INT_CTRL__THERM_INTH_MASK_MASK |
-				 CG_THERMAL_INT_CTRL__THERM_INTL_MASK_MASK);
-		WREG32_SMC(ixCG_THERMAL_INT, thermal_int);
-		result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_Thermal_Cntl_Enable);
-		if (result != PPSMC_Result_OK) {
-			DRM_DEBUG_KMS("Could not enable thermal interrupts.\n");
-			return -EINVAL;
-		}
-	} else {
-		thermal_int |= CG_THERMAL_INT_CTRL__THERM_INTH_MASK_MASK |
-			CG_THERMAL_INT_CTRL__THERM_INTL_MASK_MASK;
-		WREG32_SMC(ixCG_THERMAL_INT, thermal_int);
-		result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_Thermal_Cntl_Disable);
-		if (result != PPSMC_Result_OK) {
-			DRM_DEBUG_KMS("Could not disable thermal interrupts.\n");
-			return -EINVAL;
-		}
-	}
-
-	return 0;
-}
-
-static void ci_fan_ctrl_set_static_mode(struct amdgpu_device *adev, u32 mode)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u32 tmp;
-
-	if (pi->fan_ctrl_is_in_default_mode) {
-		tmp = (RREG32_SMC(ixCG_FDO_CTRL2) & CG_FDO_CTRL2__FDO_PWM_MODE_MASK)
-			>> CG_FDO_CTRL2__FDO_PWM_MODE__SHIFT;
-		pi->fan_ctrl_default_mode = tmp;
-		tmp = (RREG32_SMC(ixCG_FDO_CTRL2) & CG_FDO_CTRL2__TMIN_MASK)
-			>> CG_FDO_CTRL2__TMIN__SHIFT;
-		pi->t_min = tmp;
-		pi->fan_ctrl_is_in_default_mode = false;
-	}
-
-	tmp = RREG32_SMC(ixCG_FDO_CTRL2) & ~CG_FDO_CTRL2__TMIN_MASK;
-	tmp |= 0 << CG_FDO_CTRL2__TMIN__SHIFT;
-	WREG32_SMC(ixCG_FDO_CTRL2, tmp);
-
-	tmp = RREG32_SMC(ixCG_FDO_CTRL2) & ~CG_FDO_CTRL2__FDO_PWM_MODE_MASK;
-	tmp |= mode << CG_FDO_CTRL2__FDO_PWM_MODE__SHIFT;
-	WREG32_SMC(ixCG_FDO_CTRL2, tmp);
-}
-
-static int ci_thermal_setup_fan_table(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	SMU7_Discrete_FanTable fan_table = { FDO_MODE_HARDWARE };
-	u32 duty100;
-	u32 t_diff1, t_diff2, pwm_diff1, pwm_diff2;
-	u16 fdo_min, slope1, slope2;
-	u32 reference_clock, tmp;
-	int ret;
-	u64 tmp64;
-
-	if (!pi->fan_table_start) {
-		adev->pm.dpm.fan.ucode_fan_control = false;
-		return 0;
-	}
-
-	duty100 = (RREG32_SMC(ixCG_FDO_CTRL1) & CG_FDO_CTRL1__FMAX_DUTY100_MASK)
-		>> CG_FDO_CTRL1__FMAX_DUTY100__SHIFT;
-
-	if (duty100 == 0) {
-		adev->pm.dpm.fan.ucode_fan_control = false;
-		return 0;
-	}
-
-	tmp64 = (u64)adev->pm.dpm.fan.pwm_min * duty100;
-	do_div(tmp64, 10000);
-	fdo_min = (u16)tmp64;
-
-	t_diff1 = adev->pm.dpm.fan.t_med - adev->pm.dpm.fan.t_min;
-	t_diff2 = adev->pm.dpm.fan.t_high - adev->pm.dpm.fan.t_med;
-
-	pwm_diff1 = adev->pm.dpm.fan.pwm_med - adev->pm.dpm.fan.pwm_min;
-	pwm_diff2 = adev->pm.dpm.fan.pwm_high - adev->pm.dpm.fan.pwm_med;
-
-	slope1 = (u16)((50 + ((16 * duty100 * pwm_diff1) / t_diff1)) / 100);
-	slope2 = (u16)((50 + ((16 * duty100 * pwm_diff2) / t_diff2)) / 100);
-
-	fan_table.TempMin = cpu_to_be16((50 + adev->pm.dpm.fan.t_min) / 100);
-	fan_table.TempMed = cpu_to_be16((50 + adev->pm.dpm.fan.t_med) / 100);
-	fan_table.TempMax = cpu_to_be16((50 + adev->pm.dpm.fan.t_max) / 100);
-
-	fan_table.Slope1 = cpu_to_be16(slope1);
-	fan_table.Slope2 = cpu_to_be16(slope2);
-
-	fan_table.FdoMin = cpu_to_be16(fdo_min);
-
-	fan_table.HystDown = cpu_to_be16(adev->pm.dpm.fan.t_hyst);
-
-	fan_table.HystUp = cpu_to_be16(1);
-
-	fan_table.HystSlope = cpu_to_be16(1);
-
-	fan_table.TempRespLim = cpu_to_be16(5);
-
-	reference_clock = amdgpu_asic_get_xclk(adev);
-
-	fan_table.RefreshPeriod = cpu_to_be32((adev->pm.dpm.fan.cycle_delay *
-					       reference_clock) / 1600);
-
-	fan_table.FdoMax = cpu_to_be16((u16)duty100);
-
-	tmp = (RREG32_SMC(ixCG_MULT_THERMAL_CTRL) & CG_MULT_THERMAL_CTRL__TEMP_SEL_MASK)
-		>> CG_MULT_THERMAL_CTRL__TEMP_SEL__SHIFT;
-	fan_table.TempSrc = (uint8_t)tmp;
-
-	ret = amdgpu_ci_copy_bytes_to_smc(adev,
-					  pi->fan_table_start,
-					  (u8 *)(&fan_table),
-					  sizeof(fan_table),
-					  pi->sram_end);
-
-	if (ret) {
-		DRM_ERROR("Failed to load fan table to the SMC.");
-		adev->pm.dpm.fan.ucode_fan_control = false;
-	}
-
-	return 0;
-}
-
-static int ci_fan_ctrl_start_smc_fan_control(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	PPSMC_Result ret;
-
-	if (pi->caps_od_fuzzy_fan_control_support) {
-		ret = amdgpu_ci_send_msg_to_smc_with_parameter(adev,
-							       PPSMC_StartFanControl,
-							       FAN_CONTROL_FUZZY);
-		if (ret != PPSMC_Result_OK)
-			return -EINVAL;
-		ret = amdgpu_ci_send_msg_to_smc_with_parameter(adev,
-							       PPSMC_MSG_SetFanPwmMax,
-							       adev->pm.dpm.fan.default_max_fan_pwm);
-		if (ret != PPSMC_Result_OK)
-			return -EINVAL;
-	} else {
-		ret = amdgpu_ci_send_msg_to_smc_with_parameter(adev,
-							       PPSMC_StartFanControl,
-							       FAN_CONTROL_TABLE);
-		if (ret != PPSMC_Result_OK)
-			return -EINVAL;
-	}
-
-	pi->fan_is_controlled_by_smc = true;
-	return 0;
-}
-
-
-static int ci_fan_ctrl_stop_smc_fan_control(struct amdgpu_device *adev)
-{
-	PPSMC_Result ret;
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	ret = amdgpu_ci_send_msg_to_smc(adev, PPSMC_StopFanControl);
-	if (ret == PPSMC_Result_OK) {
-		pi->fan_is_controlled_by_smc = false;
-		return 0;
-	} else {
-		return -EINVAL;
-	}
-}
-
-static int ci_dpm_get_fan_speed_percent(struct amdgpu_device *adev,
-					u32 *speed)
-{
-	u32 duty, duty100;
-	u64 tmp64;
-
-	if (adev->pm.no_fan)
-		return -ENOENT;
-
-	duty100 = (RREG32_SMC(ixCG_FDO_CTRL1) & CG_FDO_CTRL1__FMAX_DUTY100_MASK)
-		>> CG_FDO_CTRL1__FMAX_DUTY100__SHIFT;
-	duty = (RREG32_SMC(ixCG_THERMAL_STATUS) & CG_THERMAL_STATUS__FDO_PWM_DUTY_MASK)
-		>> CG_THERMAL_STATUS__FDO_PWM_DUTY__SHIFT;
-
-	if (duty100 == 0)
-		return -EINVAL;
-
-	tmp64 = (u64)duty * 100;
-	do_div(tmp64, duty100);
-	*speed = (u32)tmp64;
-
-	if (*speed > 100)
-		*speed = 100;
-
-	return 0;
-}
-
-static int ci_dpm_set_fan_speed_percent(struct amdgpu_device *adev,
-					u32 speed)
-{
-	u32 tmp;
-	u32 duty, duty100;
-	u64 tmp64;
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	if (adev->pm.no_fan)
-		return -ENOENT;
-
-	if (pi->fan_is_controlled_by_smc)
-		return -EINVAL;
-
-	if (speed > 100)
-		return -EINVAL;
-
-	duty100 = (RREG32_SMC(ixCG_FDO_CTRL1) & CG_FDO_CTRL1__FMAX_DUTY100_MASK)
-		>> CG_FDO_CTRL1__FMAX_DUTY100__SHIFT;
-
-	if (duty100 == 0)
-		return -EINVAL;
-
-	tmp64 = (u64)speed * duty100;
-	do_div(tmp64, 100);
-	duty = (u32)tmp64;
-
-	tmp = RREG32_SMC(ixCG_FDO_CTRL0) & ~CG_FDO_CTRL0__FDO_STATIC_DUTY_MASK;
-	tmp |= duty << CG_FDO_CTRL0__FDO_STATIC_DUTY__SHIFT;
-	WREG32_SMC(ixCG_FDO_CTRL0, tmp);
-
-	return 0;
-}
-
-static void ci_dpm_set_fan_control_mode(struct amdgpu_device *adev, u32 mode)
-{
-	switch (mode) {
-	case AMD_FAN_CTRL_NONE:
-		if (adev->pm.dpm.fan.ucode_fan_control)
-			ci_fan_ctrl_stop_smc_fan_control(adev);
-		ci_dpm_set_fan_speed_percent(adev, 100);
-		break;
-	case AMD_FAN_CTRL_MANUAL:
-		if (adev->pm.dpm.fan.ucode_fan_control)
-			ci_fan_ctrl_stop_smc_fan_control(adev);
-		break;
-	case AMD_FAN_CTRL_AUTO:
-		if (adev->pm.dpm.fan.ucode_fan_control)
-			ci_thermal_start_smc_fan_control(adev);
-		break;
-	default:
-		break;
-	}
-}
-
-static u32 ci_dpm_get_fan_control_mode(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	if (pi->fan_is_controlled_by_smc)
-		return AMD_FAN_CTRL_AUTO;
-	else
-		return AMD_FAN_CTRL_MANUAL;
-}
-
-#if 0
-static int ci_fan_ctrl_get_fan_speed_rpm(struct amdgpu_device *adev,
-					 u32 *speed)
-{
-	u32 tach_period;
-	u32 xclk = amdgpu_asic_get_xclk(adev);
-
-	if (adev->pm.no_fan)
-		return -ENOENT;
-
-	if (adev->pm.fan_pulses_per_revolution == 0)
-		return -ENOENT;
-
-	tach_period = (RREG32_SMC(ixCG_TACH_STATUS) & CG_TACH_STATUS__TACH_PERIOD_MASK)
-		>> CG_TACH_STATUS__TACH_PERIOD__SHIFT;
-	if (tach_period == 0)
-		return -ENOENT;
-
-	*speed = 60 * xclk * 10000 / tach_period;
-
-	return 0;
-}
-
-static int ci_fan_ctrl_set_fan_speed_rpm(struct amdgpu_device *adev,
-					 u32 speed)
-{
-	u32 tach_period, tmp;
-	u32 xclk = amdgpu_asic_get_xclk(adev);
-
-	if (adev->pm.no_fan)
-		return -ENOENT;
-
-	if (adev->pm.fan_pulses_per_revolution == 0)
-		return -ENOENT;
-
-	if ((speed < adev->pm.fan_min_rpm) ||
-	    (speed > adev->pm.fan_max_rpm))
-		return -EINVAL;
-
-	if (adev->pm.dpm.fan.ucode_fan_control)
-		ci_fan_ctrl_stop_smc_fan_control(adev);
-
-	tach_period = 60 * xclk * 10000 / (8 * speed);
-	tmp = RREG32_SMC(ixCG_TACH_CTRL) & ~CG_TACH_CTRL__TARGET_PERIOD_MASK;
-	tmp |= tach_period << CG_TACH_CTRL__TARGET_PERIOD__SHIFT;
-	WREG32_SMC(CG_TACH_CTRL, tmp);
-
-	ci_fan_ctrl_set_static_mode(adev, FDO_PWM_MODE_STATIC_RPM);
-
-	return 0;
-}
-#endif
-
-static void ci_fan_ctrl_set_default_mode(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u32 tmp;
-
-	if (!pi->fan_ctrl_is_in_default_mode) {
-		tmp = RREG32_SMC(ixCG_FDO_CTRL2) & ~CG_FDO_CTRL2__FDO_PWM_MODE_MASK;
-		tmp |= pi->fan_ctrl_default_mode << CG_FDO_CTRL2__FDO_PWM_MODE__SHIFT;
-		WREG32_SMC(ixCG_FDO_CTRL2, tmp);
-
-		tmp = RREG32_SMC(ixCG_FDO_CTRL2) & ~CG_FDO_CTRL2__TMIN_MASK;
-		tmp |= pi->t_min << CG_FDO_CTRL2__TMIN__SHIFT;
-		WREG32_SMC(ixCG_FDO_CTRL2, tmp);
-		pi->fan_ctrl_is_in_default_mode = true;
-	}
-}
-
-static void ci_thermal_start_smc_fan_control(struct amdgpu_device *adev)
-{
-	if (adev->pm.dpm.fan.ucode_fan_control) {
-		ci_fan_ctrl_start_smc_fan_control(adev);
-		ci_fan_ctrl_set_static_mode(adev, FDO_PWM_MODE_STATIC);
-	}
-}
-
-static void ci_thermal_initialize(struct amdgpu_device *adev)
-{
-	u32 tmp;
-
-	if (adev->pm.fan_pulses_per_revolution) {
-		tmp = RREG32_SMC(ixCG_TACH_CTRL) & ~CG_TACH_CTRL__EDGE_PER_REV_MASK;
-		tmp |= (adev->pm.fan_pulses_per_revolution - 1)
-			<< CG_TACH_CTRL__EDGE_PER_REV__SHIFT;
-		WREG32_SMC(ixCG_TACH_CTRL, tmp);
-	}
-
-	tmp = RREG32_SMC(ixCG_FDO_CTRL2) & ~CG_FDO_CTRL2__TACH_PWM_RESP_RATE_MASK;
-	tmp |= 0x28 << CG_FDO_CTRL2__TACH_PWM_RESP_RATE__SHIFT;
-	WREG32_SMC(ixCG_FDO_CTRL2, tmp);
-}
-
-static int ci_thermal_start_thermal_controller(struct amdgpu_device *adev)
-{
-	int ret;
-
-	ci_thermal_initialize(adev);
-	ret = ci_thermal_set_temperature_range(adev, CISLANDS_TEMP_RANGE_MIN, CISLANDS_TEMP_RANGE_MAX);
-	if (ret)
-		return ret;
-	ret = ci_thermal_enable_alert(adev, true);
-	if (ret)
-		return ret;
-	if (adev->pm.dpm.fan.ucode_fan_control) {
-		ret = ci_thermal_setup_fan_table(adev);
-		if (ret)
-			return ret;
-		ci_thermal_start_smc_fan_control(adev);
-	}
-
-	return 0;
-}
-
-static void ci_thermal_stop_thermal_controller(struct amdgpu_device *adev)
-{
-	if (!adev->pm.no_fan)
-		ci_fan_ctrl_set_default_mode(adev);
-}
-
-static int ci_read_smc_soft_register(struct amdgpu_device *adev,
-				     u16 reg_offset, u32 *value)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	return amdgpu_ci_read_smc_sram_dword(adev,
-				      pi->soft_regs_start + reg_offset,
-				      value, pi->sram_end);
-}
-
-static int ci_write_smc_soft_register(struct amdgpu_device *adev,
-				      u16 reg_offset, u32 value)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	return amdgpu_ci_write_smc_sram_dword(adev,
-				       pi->soft_regs_start + reg_offset,
-				       value, pi->sram_end);
-}
-
-static void ci_init_fps_limits(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	SMU7_Discrete_DpmTable *table = &pi->smc_state_table;
-
-	if (pi->caps_fps) {
-		u16 tmp;
-
-		tmp = 45;
-		table->FpsHighT = cpu_to_be16(tmp);
-
-		tmp = 30;
-		table->FpsLowT = cpu_to_be16(tmp);
-	}
-}
-
-static int ci_update_sclk_t(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	int ret = 0;
-	u32 low_sclk_interrupt_t = 0;
-
-	if (pi->caps_sclk_throttle_low_notification) {
-		low_sclk_interrupt_t = cpu_to_be32(pi->low_sclk_interrupt_t);
-
-		ret = amdgpu_ci_copy_bytes_to_smc(adev,
-					   pi->dpm_table_start +
-					   offsetof(SMU7_Discrete_DpmTable, LowSclkInterruptT),
-					   (u8 *)&low_sclk_interrupt_t,
-					   sizeof(u32), pi->sram_end);
-
-	}
-
-	return ret;
-}
-
-static void ci_get_leakage_voltages(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u16 leakage_id, virtual_voltage_id;
-	u16 vddc, vddci;
-	int i;
-
-	pi->vddc_leakage.count = 0;
-	pi->vddci_leakage.count = 0;
-
-	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_EVV) {
-		for (i = 0; i < CISLANDS_MAX_LEAKAGE_COUNT; i++) {
-			virtual_voltage_id = ATOM_VIRTUAL_VOLTAGE_ID0 + i;
-			if (amdgpu_atombios_get_voltage_evv(adev, virtual_voltage_id, &vddc) != 0)
-				continue;
-			if (vddc != 0 && vddc != virtual_voltage_id) {
-				pi->vddc_leakage.actual_voltage[pi->vddc_leakage.count] = vddc;
-				pi->vddc_leakage.leakage_id[pi->vddc_leakage.count] = virtual_voltage_id;
-				pi->vddc_leakage.count++;
-			}
-		}
-	} else if (amdgpu_atombios_get_leakage_id_from_vbios(adev, &leakage_id) == 0) {
-		for (i = 0; i < CISLANDS_MAX_LEAKAGE_COUNT; i++) {
-			virtual_voltage_id = ATOM_VIRTUAL_VOLTAGE_ID0 + i;
-			if (amdgpu_atombios_get_leakage_vddc_based_on_leakage_params(adev, &vddc, &vddci,
-										     virtual_voltage_id,
-										     leakage_id) == 0) {
-				if (vddc != 0 && vddc != virtual_voltage_id) {
-					pi->vddc_leakage.actual_voltage[pi->vddc_leakage.count] = vddc;
-					pi->vddc_leakage.leakage_id[pi->vddc_leakage.count] = virtual_voltage_id;
-					pi->vddc_leakage.count++;
-				}
-				if (vddci != 0 && vddci != virtual_voltage_id) {
-					pi->vddci_leakage.actual_voltage[pi->vddci_leakage.count] = vddci;
-					pi->vddci_leakage.leakage_id[pi->vddci_leakage.count] = virtual_voltage_id;
-					pi->vddci_leakage.count++;
-				}
-			}
-		}
-	}
-}
-
-static void ci_set_dpm_event_sources(struct amdgpu_device *adev, u32 sources)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	bool want_thermal_protection;
-	enum amdgpu_dpm_event_src dpm_event_src;
-	u32 tmp;
-
-	switch (sources) {
-	case 0:
-	default:
-		want_thermal_protection = false;
-		break;
-	case (1 << AMDGPU_DPM_AUTO_THROTTLE_SRC_THERMAL):
-		want_thermal_protection = true;
-		dpm_event_src = AMDGPU_DPM_EVENT_SRC_DIGITAL;
-		break;
-	case (1 << AMDGPU_DPM_AUTO_THROTTLE_SRC_EXTERNAL):
-		want_thermal_protection = true;
-		dpm_event_src = AMDGPU_DPM_EVENT_SRC_EXTERNAL;
-		break;
-	case ((1 << AMDGPU_DPM_AUTO_THROTTLE_SRC_EXTERNAL) |
-	      (1 << AMDGPU_DPM_AUTO_THROTTLE_SRC_THERMAL)):
-		want_thermal_protection = true;
-		dpm_event_src = AMDGPU_DPM_EVENT_SRC_DIGIAL_OR_EXTERNAL;
-		break;
-	}
-
-	if (want_thermal_protection) {
-#if 0
-		/* XXX: need to figure out how to handle this properly */
-		tmp = RREG32_SMC(ixCG_THERMAL_CTRL);
-		tmp &= DPM_EVENT_SRC_MASK;
-		tmp |= DPM_EVENT_SRC(dpm_event_src);
-		WREG32_SMC(ixCG_THERMAL_CTRL, tmp);
-#endif
-
-		tmp = RREG32_SMC(ixGENERAL_PWRMGT);
-		if (pi->thermal_protection)
-			tmp &= ~GENERAL_PWRMGT__THERMAL_PROTECTION_DIS_MASK;
-		else
-			tmp |= GENERAL_PWRMGT__THERMAL_PROTECTION_DIS_MASK;
-		WREG32_SMC(ixGENERAL_PWRMGT, tmp);
-	} else {
-		tmp = RREG32_SMC(ixGENERAL_PWRMGT);
-		tmp |= GENERAL_PWRMGT__THERMAL_PROTECTION_DIS_MASK;
-		WREG32_SMC(ixGENERAL_PWRMGT, tmp);
-	}
-}
-
-static void ci_enable_auto_throttle_source(struct amdgpu_device *adev,
-					   enum amdgpu_dpm_auto_throttle_src source,
-					   bool enable)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	if (enable) {
-		if (!(pi->active_auto_throttle_sources & (1 << source))) {
-			pi->active_auto_throttle_sources |= 1 << source;
-			ci_set_dpm_event_sources(adev, pi->active_auto_throttle_sources);
-		}
-	} else {
-		if (pi->active_auto_throttle_sources & (1 << source)) {
-			pi->active_auto_throttle_sources &= ~(1 << source);
-			ci_set_dpm_event_sources(adev, pi->active_auto_throttle_sources);
-		}
-	}
-}
-
-static void ci_enable_vr_hot_gpio_interrupt(struct amdgpu_device *adev)
-{
-	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_REGULATOR_HOT)
-		amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_EnableVRHotGPIOInterrupt);
-}
-
-static int ci_unfreeze_sclk_mclk_dpm(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	PPSMC_Result smc_result;
-
-	if (!pi->need_update_smu7_dpm_table)
-		return 0;
-
-	if ((!pi->sclk_dpm_key_disabled) &&
-	    (pi->need_update_smu7_dpm_table & (DPMTABLE_OD_UPDATE_SCLK | DPMTABLE_UPDATE_SCLK))) {
-		smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_SCLKDPM_UnfreezeLevel);
-		if (smc_result != PPSMC_Result_OK)
-			return -EINVAL;
-	}
-
-	if ((!pi->mclk_dpm_key_disabled) &&
-	    (pi->need_update_smu7_dpm_table & DPMTABLE_OD_UPDATE_MCLK)) {
-		smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_MCLKDPM_UnfreezeLevel);
-		if (smc_result != PPSMC_Result_OK)
-			return -EINVAL;
-	}
-
-	pi->need_update_smu7_dpm_table = 0;
-	return 0;
-}
-
-static int ci_enable_sclk_mclk_dpm(struct amdgpu_device *adev, bool enable)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	PPSMC_Result smc_result;
-
-	if (enable) {
-		if (!pi->sclk_dpm_key_disabled) {
-			smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_DPM_Enable);
-			if (smc_result != PPSMC_Result_OK)
-				return -EINVAL;
-		}
-
-		if (!pi->mclk_dpm_key_disabled) {
-			smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_MCLKDPM_Enable);
-			if (smc_result != PPSMC_Result_OK)
-				return -EINVAL;
-
-			WREG32_P(mmMC_SEQ_CNTL_3, MC_SEQ_CNTL_3__CAC_EN_MASK,
-					~MC_SEQ_CNTL_3__CAC_EN_MASK);
-
-			WREG32_SMC(ixLCAC_MC0_CNTL, 0x05);
-			WREG32_SMC(ixLCAC_MC1_CNTL, 0x05);
-			WREG32_SMC(ixLCAC_CPL_CNTL, 0x100005);
-
-			udelay(10);
-
-			WREG32_SMC(ixLCAC_MC0_CNTL, 0x400005);
-			WREG32_SMC(ixLCAC_MC1_CNTL, 0x400005);
-			WREG32_SMC(ixLCAC_CPL_CNTL, 0x500005);
-		}
-	} else {
-		if (!pi->sclk_dpm_key_disabled) {
-			smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_DPM_Disable);
-			if (smc_result != PPSMC_Result_OK)
-				return -EINVAL;
-		}
-
-		if (!pi->mclk_dpm_key_disabled) {
-			smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_MCLKDPM_Disable);
-			if (smc_result != PPSMC_Result_OK)
-				return -EINVAL;
-		}
-	}
-
-	return 0;
-}
-
-static int ci_start_dpm(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	PPSMC_Result smc_result;
-	int ret;
-	u32 tmp;
-
-	tmp = RREG32_SMC(ixGENERAL_PWRMGT);
-	tmp |= GENERAL_PWRMGT__GLOBAL_PWRMGT_EN_MASK;
-	WREG32_SMC(ixGENERAL_PWRMGT, tmp);
-
-	tmp = RREG32_SMC(ixSCLK_PWRMGT_CNTL);
-	tmp |= SCLK_PWRMGT_CNTL__DYNAMIC_PM_EN_MASK;
-	WREG32_SMC(ixSCLK_PWRMGT_CNTL, tmp);
-
-	ci_write_smc_soft_register(adev, offsetof(SMU7_SoftRegisters, VoltageChangeTimeout), 0x1000);
-
-	WREG32_P(mmBIF_LNCNT_RESET, 0, ~BIF_LNCNT_RESET__RESET_LNCNT_EN_MASK);
-
-	smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_Voltage_Cntl_Enable);
-	if (smc_result != PPSMC_Result_OK)
-		return -EINVAL;
-
-	ret = ci_enable_sclk_mclk_dpm(adev, true);
-	if (ret)
-		return ret;
-
-	if (!pi->pcie_dpm_key_disabled) {
-		smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_PCIeDPM_Enable);
-		if (smc_result != PPSMC_Result_OK)
-			return -EINVAL;
-	}
-
-	return 0;
-}
-
-static int ci_freeze_sclk_mclk_dpm(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	PPSMC_Result smc_result;
-
-	if (!pi->need_update_smu7_dpm_table)
-		return 0;
-
-	if ((!pi->sclk_dpm_key_disabled) &&
-	    (pi->need_update_smu7_dpm_table & (DPMTABLE_OD_UPDATE_SCLK | DPMTABLE_UPDATE_SCLK))) {
-		smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_SCLKDPM_FreezeLevel);
-		if (smc_result != PPSMC_Result_OK)
-			return -EINVAL;
-	}
-
-	if ((!pi->mclk_dpm_key_disabled) &&
-	    (pi->need_update_smu7_dpm_table & DPMTABLE_OD_UPDATE_MCLK)) {
-		smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_MCLKDPM_FreezeLevel);
-		if (smc_result != PPSMC_Result_OK)
-			return -EINVAL;
-	}
-
-	return 0;
-}
-
-static int ci_stop_dpm(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	PPSMC_Result smc_result;
-	int ret;
-	u32 tmp;
-
-	tmp = RREG32_SMC(ixGENERAL_PWRMGT);
-	tmp &= ~GENERAL_PWRMGT__GLOBAL_PWRMGT_EN_MASK;
-	WREG32_SMC(ixGENERAL_PWRMGT, tmp);
-
-	tmp = RREG32_SMC(ixSCLK_PWRMGT_CNTL);
-	tmp &= ~SCLK_PWRMGT_CNTL__DYNAMIC_PM_EN_MASK;
-	WREG32_SMC(ixSCLK_PWRMGT_CNTL, tmp);
-
-	if (!pi->pcie_dpm_key_disabled) {
-		smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_PCIeDPM_Disable);
-		if (smc_result != PPSMC_Result_OK)
-			return -EINVAL;
-	}
-
-	ret = ci_enable_sclk_mclk_dpm(adev, false);
-	if (ret)
-		return ret;
-
-	smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_Voltage_Cntl_Disable);
-	if (smc_result != PPSMC_Result_OK)
-		return -EINVAL;
-
-	return 0;
-}
-
-static void ci_enable_sclk_control(struct amdgpu_device *adev, bool enable)
-{
-	u32 tmp = RREG32_SMC(ixSCLK_PWRMGT_CNTL);
-
-	if (enable)
-		tmp &= ~SCLK_PWRMGT_CNTL__SCLK_PWRMGT_OFF_MASK;
-	else
-		tmp |= SCLK_PWRMGT_CNTL__SCLK_PWRMGT_OFF_MASK;
-	WREG32_SMC(ixSCLK_PWRMGT_CNTL, tmp);
-}
-
-#if 0
-static int ci_notify_hw_of_power_source(struct amdgpu_device *adev,
-					bool ac_power)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct amdgpu_cac_tdp_table *cac_tdp_table =
-		adev->pm.dpm.dyn_state.cac_tdp_table;
-	u32 power_limit;
-
-	if (ac_power)
-		power_limit = (u32)(cac_tdp_table->maximum_power_delivery_limit * 256);
-	else
-		power_limit = (u32)(cac_tdp_table->battery_power_limit * 256);
-
-	ci_set_power_limit(adev, power_limit);
-
-	if (pi->caps_automatic_dc_transition) {
-		if (ac_power)
-			amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_RunningOnAC);
-		else
-			amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_Remove_DC_Clamp);
-	}
-
-	return 0;
-}
-#endif
-
-static PPSMC_Result amdgpu_ci_send_msg_to_smc_with_parameter(struct amdgpu_device *adev,
-						      PPSMC_Msg msg, u32 parameter)
-{
-	WREG32(mmSMC_MSG_ARG_0, parameter);
-	return amdgpu_ci_send_msg_to_smc(adev, msg);
-}
-
-static PPSMC_Result amdgpu_ci_send_msg_to_smc_return_parameter(struct amdgpu_device *adev,
-							PPSMC_Msg msg, u32 *parameter)
-{
-	PPSMC_Result smc_result;
-
-	smc_result = amdgpu_ci_send_msg_to_smc(adev, msg);
-
-	if ((smc_result == PPSMC_Result_OK) && parameter)
-		*parameter = RREG32(mmSMC_MSG_ARG_0);
-
-	return smc_result;
-}
-
-static int ci_dpm_force_state_sclk(struct amdgpu_device *adev, u32 n)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	if (!pi->sclk_dpm_key_disabled) {
-		PPSMC_Result smc_result =
-			amdgpu_ci_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_SCLKDPM_SetEnabledMask, 1 << n);
-		if (smc_result != PPSMC_Result_OK)
-			return -EINVAL;
-	}
-
-	return 0;
-}
-
-static int ci_dpm_force_state_mclk(struct amdgpu_device *adev, u32 n)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	if (!pi->mclk_dpm_key_disabled) {
-		PPSMC_Result smc_result =
-			amdgpu_ci_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_MCLKDPM_SetEnabledMask, 1 << n);
-		if (smc_result != PPSMC_Result_OK)
-			return -EINVAL;
-	}
-
-	return 0;
-}
-
-static int ci_dpm_force_state_pcie(struct amdgpu_device *adev, u32 n)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	if (!pi->pcie_dpm_key_disabled) {
-		PPSMC_Result smc_result =
-			amdgpu_ci_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_PCIeDPM_ForceLevel, n);
-		if (smc_result != PPSMC_Result_OK)
-			return -EINVAL;
-	}
-
-	return 0;
-}
-
-static int ci_set_power_limit(struct amdgpu_device *adev, u32 n)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	if (pi->power_containment_features & POWERCONTAINMENT_FEATURE_PkgPwrLimit) {
-		PPSMC_Result smc_result =
-			amdgpu_ci_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_PkgPwrSetLimit, n);
-		if (smc_result != PPSMC_Result_OK)
-			return -EINVAL;
-	}
-
-	return 0;
-}
-
-static int ci_set_overdrive_target_tdp(struct amdgpu_device *adev,
-				       u32 target_tdp)
-{
-	PPSMC_Result smc_result =
-		amdgpu_ci_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_OverDriveSetTargetTdp, target_tdp);
-	if (smc_result != PPSMC_Result_OK)
-		return -EINVAL;
-	return 0;
-}
-
-#if 0
-static int ci_set_boot_state(struct amdgpu_device *adev)
-{
-	return ci_enable_sclk_mclk_dpm(adev, false);
-}
-#endif
-
-static u32 ci_get_average_sclk_freq(struct amdgpu_device *adev)
-{
-	u32 sclk_freq;
-	PPSMC_Result smc_result =
-		amdgpu_ci_send_msg_to_smc_return_parameter(adev,
-						    PPSMC_MSG_API_GetSclkFrequency,
-						    &sclk_freq);
-	if (smc_result != PPSMC_Result_OK)
-		sclk_freq = 0;
-
-	return sclk_freq;
-}
-
-static u32 ci_get_average_mclk_freq(struct amdgpu_device *adev)
-{
-	u32 mclk_freq;
-	PPSMC_Result smc_result =
-		amdgpu_ci_send_msg_to_smc_return_parameter(adev,
-						    PPSMC_MSG_API_GetMclkFrequency,
-						    &mclk_freq);
-	if (smc_result != PPSMC_Result_OK)
-		mclk_freq = 0;
-
-	return mclk_freq;
-}
-
-static void ci_dpm_start_smc(struct amdgpu_device *adev)
-{
-	int i;
-
-	amdgpu_ci_program_jump_on_start(adev);
-	amdgpu_ci_start_smc_clock(adev);
-	amdgpu_ci_start_smc(adev);
-	for (i = 0; i < adev->usec_timeout; i++) {
-		if (RREG32_SMC(ixFIRMWARE_FLAGS) & FIRMWARE_FLAGS__INTERRUPTS_ENABLED_MASK)
-			break;
-	}
-}
-
-static void ci_dpm_stop_smc(struct amdgpu_device *adev)
-{
-	amdgpu_ci_reset_smc(adev);
-	amdgpu_ci_stop_smc_clock(adev);
-}
-
-static int ci_process_firmware_header(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u32 tmp;
-	int ret;
-
-	ret = amdgpu_ci_read_smc_sram_dword(adev,
-				     SMU7_FIRMWARE_HEADER_LOCATION +
-				     offsetof(SMU7_Firmware_Header, DpmTable),
-				     &tmp, pi->sram_end);
-	if (ret)
-		return ret;
-
-	pi->dpm_table_start = tmp;
-
-	ret = amdgpu_ci_read_smc_sram_dword(adev,
-				     SMU7_FIRMWARE_HEADER_LOCATION +
-				     offsetof(SMU7_Firmware_Header, SoftRegisters),
-				     &tmp, pi->sram_end);
-	if (ret)
-		return ret;
-
-	pi->soft_regs_start = tmp;
-
-	ret = amdgpu_ci_read_smc_sram_dword(adev,
-				     SMU7_FIRMWARE_HEADER_LOCATION +
-				     offsetof(SMU7_Firmware_Header, mcRegisterTable),
-				     &tmp, pi->sram_end);
-	if (ret)
-		return ret;
-
-	pi->mc_reg_table_start = tmp;
-
-	ret = amdgpu_ci_read_smc_sram_dword(adev,
-				     SMU7_FIRMWARE_HEADER_LOCATION +
-				     offsetof(SMU7_Firmware_Header, FanTable),
-				     &tmp, pi->sram_end);
-	if (ret)
-		return ret;
-
-	pi->fan_table_start = tmp;
-
-	ret = amdgpu_ci_read_smc_sram_dword(adev,
-				     SMU7_FIRMWARE_HEADER_LOCATION +
-				     offsetof(SMU7_Firmware_Header, mcArbDramTimingTable),
-				     &tmp, pi->sram_end);
-	if (ret)
-		return ret;
-
-	pi->arb_table_start = tmp;
-
-	return 0;
-}
-
-static void ci_read_clock_registers(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	pi->clock_registers.cg_spll_func_cntl =
-		RREG32_SMC(ixCG_SPLL_FUNC_CNTL);
-	pi->clock_registers.cg_spll_func_cntl_2 =
-		RREG32_SMC(ixCG_SPLL_FUNC_CNTL_2);
-	pi->clock_registers.cg_spll_func_cntl_3 =
-		RREG32_SMC(ixCG_SPLL_FUNC_CNTL_3);
-	pi->clock_registers.cg_spll_func_cntl_4 =
-		RREG32_SMC(ixCG_SPLL_FUNC_CNTL_4);
-	pi->clock_registers.cg_spll_spread_spectrum =
-		RREG32_SMC(ixCG_SPLL_SPREAD_SPECTRUM);
-	pi->clock_registers.cg_spll_spread_spectrum_2 =
-		RREG32_SMC(ixCG_SPLL_SPREAD_SPECTRUM_2);
-	pi->clock_registers.dll_cntl = RREG32(mmDLL_CNTL);
-	pi->clock_registers.mclk_pwrmgt_cntl = RREG32(mmMCLK_PWRMGT_CNTL);
-	pi->clock_registers.mpll_ad_func_cntl = RREG32(mmMPLL_AD_FUNC_CNTL);
-	pi->clock_registers.mpll_dq_func_cntl = RREG32(mmMPLL_DQ_FUNC_CNTL);
-	pi->clock_registers.mpll_func_cntl = RREG32(mmMPLL_FUNC_CNTL);
-	pi->clock_registers.mpll_func_cntl_1 = RREG32(mmMPLL_FUNC_CNTL_1);
-	pi->clock_registers.mpll_func_cntl_2 = RREG32(mmMPLL_FUNC_CNTL_2);
-	pi->clock_registers.mpll_ss1 = RREG32(mmMPLL_SS1);
-	pi->clock_registers.mpll_ss2 = RREG32(mmMPLL_SS2);
-}
-
-static void ci_init_sclk_t(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	pi->low_sclk_interrupt_t = 0;
-}
-
-static void ci_enable_thermal_protection(struct amdgpu_device *adev,
-					 bool enable)
-{
-	u32 tmp = RREG32_SMC(ixGENERAL_PWRMGT);
-
-	if (enable)
-		tmp &= ~GENERAL_PWRMGT__THERMAL_PROTECTION_DIS_MASK;
-	else
-		tmp |= GENERAL_PWRMGT__THERMAL_PROTECTION_DIS_MASK;
-	WREG32_SMC(ixGENERAL_PWRMGT, tmp);
-}
-
-static void ci_enable_acpi_power_management(struct amdgpu_device *adev)
-{
-	u32 tmp = RREG32_SMC(ixGENERAL_PWRMGT);
-
-	tmp |= GENERAL_PWRMGT__STATIC_PM_EN_MASK;
-
-	WREG32_SMC(ixGENERAL_PWRMGT, tmp);
-}
-
-#if 0
-static int ci_enter_ulp_state(struct amdgpu_device *adev)
-{
-
-	WREG32(mmSMC_MESSAGE_0, PPSMC_MSG_SwitchToMinimumPower);
-
-	udelay(25000);
-
-	return 0;
-}
-
-static int ci_exit_ulp_state(struct amdgpu_device *adev)
-{
-	int i;
-
-	WREG32(mmSMC_MESSAGE_0, PPSMC_MSG_ResumeFromMinimumPower);
-
-	udelay(7000);
-
-	for (i = 0; i < adev->usec_timeout; i++) {
-		if (RREG32(mmSMC_RESP_0) == 1)
-			break;
-		udelay(1000);
-	}
-
-	return 0;
-}
-#endif
-
-static int ci_notify_smc_display_change(struct amdgpu_device *adev,
-					bool has_display)
-{
-	PPSMC_Msg msg = has_display ? PPSMC_MSG_HasDisplay : PPSMC_MSG_NoDisplay;
-
-	return (amdgpu_ci_send_msg_to_smc(adev, msg) == PPSMC_Result_OK) ?  0 : -EINVAL;
-}
-
-static int ci_enable_ds_master_switch(struct amdgpu_device *adev,
-				      bool enable)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	if (enable) {
-		if (pi->caps_sclk_ds) {
-			if (amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_MASTER_DeepSleep_ON) != PPSMC_Result_OK)
-				return -EINVAL;
-		} else {
-			if (amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_MASTER_DeepSleep_OFF) != PPSMC_Result_OK)
-				return -EINVAL;
-		}
-	} else {
-		if (pi->caps_sclk_ds) {
-			if (amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_MASTER_DeepSleep_OFF) != PPSMC_Result_OK)
-				return -EINVAL;
-		}
-	}
-
-	return 0;
-}
-
-static void ci_program_display_gap(struct amdgpu_device *adev)
-{
-	u32 tmp = RREG32_SMC(ixCG_DISPLAY_GAP_CNTL);
-	u32 pre_vbi_time_in_us;
-	u32 frame_time_in_us;
-	u32 ref_clock = adev->clock.spll.reference_freq;
-	u32 refresh_rate = amdgpu_dpm_get_vrefresh(adev);
-	u32 vblank_time = amdgpu_dpm_get_vblank_time(adev);
-
-	tmp &= ~CG_DISPLAY_GAP_CNTL__DISP_GAP_MASK;
-	if (adev->pm.dpm.new_active_crtc_count > 0)
-		tmp |= (AMDGPU_PM_DISPLAY_GAP_VBLANK_OR_WM << CG_DISPLAY_GAP_CNTL__DISP_GAP__SHIFT);
-	else
-		tmp |= (AMDGPU_PM_DISPLAY_GAP_IGNORE << CG_DISPLAY_GAP_CNTL__DISP_GAP__SHIFT);
-	WREG32_SMC(ixCG_DISPLAY_GAP_CNTL, tmp);
-
-	if (refresh_rate == 0)
-		refresh_rate = 60;
-	if (vblank_time == 0xffffffff)
-		vblank_time = 500;
-	frame_time_in_us = 1000000 / refresh_rate;
-	pre_vbi_time_in_us =
-		frame_time_in_us - 200 - vblank_time;
-	tmp = pre_vbi_time_in_us * (ref_clock / 100);
-
-	WREG32_SMC(ixCG_DISPLAY_GAP_CNTL2, tmp);
-	ci_write_smc_soft_register(adev, offsetof(SMU7_SoftRegisters, PreVBlankGap), 0x64);
-	ci_write_smc_soft_register(adev, offsetof(SMU7_SoftRegisters, VBlankTimeout), (frame_time_in_us - pre_vbi_time_in_us));
-
-
-	ci_notify_smc_display_change(adev, (adev->pm.dpm.new_active_crtc_count == 1));
-
-}
-
-static void ci_enable_spread_spectrum(struct amdgpu_device *adev, bool enable)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u32 tmp;
-
-	if (enable) {
-		if (pi->caps_sclk_ss_support) {
-			tmp = RREG32_SMC(ixGENERAL_PWRMGT);
-			tmp |= GENERAL_PWRMGT__DYN_SPREAD_SPECTRUM_EN_MASK;
-			WREG32_SMC(ixGENERAL_PWRMGT, tmp);
-		}
-	} else {
-		tmp = RREG32_SMC(ixCG_SPLL_SPREAD_SPECTRUM);
-		tmp &= ~CG_SPLL_SPREAD_SPECTRUM__SSEN_MASK;
-		WREG32_SMC(ixCG_SPLL_SPREAD_SPECTRUM, tmp);
-
-		tmp = RREG32_SMC(ixGENERAL_PWRMGT);
-		tmp &= ~GENERAL_PWRMGT__DYN_SPREAD_SPECTRUM_EN_MASK;
-		WREG32_SMC(ixGENERAL_PWRMGT, tmp);
-	}
-}
-
-static void ci_program_sstp(struct amdgpu_device *adev)
-{
-	WREG32_SMC(ixCG_STATIC_SCREEN_PARAMETER,
-	((CISLANDS_SSTU_DFLT << CG_STATIC_SCREEN_PARAMETER__STATIC_SCREEN_THRESHOLD_UNIT__SHIFT) |
-	 (CISLANDS_SST_DFLT << CG_STATIC_SCREEN_PARAMETER__STATIC_SCREEN_THRESHOLD__SHIFT)));
-}
-
-static void ci_enable_display_gap(struct amdgpu_device *adev)
-{
-	u32 tmp = RREG32_SMC(ixCG_DISPLAY_GAP_CNTL);
-
-	tmp &= ~(CG_DISPLAY_GAP_CNTL__DISP_GAP_MASK |
-			CG_DISPLAY_GAP_CNTL__DISP_GAP_MCHG_MASK);
-	tmp |= ((AMDGPU_PM_DISPLAY_GAP_IGNORE << CG_DISPLAY_GAP_CNTL__DISP_GAP__SHIFT) |
-		(AMDGPU_PM_DISPLAY_GAP_VBLANK << CG_DISPLAY_GAP_CNTL__DISP_GAP_MCHG__SHIFT));
-
-	WREG32_SMC(ixCG_DISPLAY_GAP_CNTL, tmp);
-}
-
-static void ci_program_vc(struct amdgpu_device *adev)
-{
-	u32 tmp;
-
-	tmp = RREG32_SMC(ixSCLK_PWRMGT_CNTL);
-	tmp &= ~(SCLK_PWRMGT_CNTL__RESET_SCLK_CNT_MASK | SCLK_PWRMGT_CNTL__RESET_BUSY_CNT_MASK);
-	WREG32_SMC(ixSCLK_PWRMGT_CNTL, tmp);
-
-	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_0, CISLANDS_VRC_DFLT0);
-	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_1, CISLANDS_VRC_DFLT1);
-	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_2, CISLANDS_VRC_DFLT2);
-	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_3, CISLANDS_VRC_DFLT3);
-	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_4, CISLANDS_VRC_DFLT4);
-	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_5, CISLANDS_VRC_DFLT5);
-	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_6, CISLANDS_VRC_DFLT6);
-	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_7, CISLANDS_VRC_DFLT7);
-}
-
-static void ci_clear_vc(struct amdgpu_device *adev)
-{
-	u32 tmp;
-
-	tmp = RREG32_SMC(ixSCLK_PWRMGT_CNTL);
-	tmp |= (SCLK_PWRMGT_CNTL__RESET_SCLK_CNT_MASK | SCLK_PWRMGT_CNTL__RESET_BUSY_CNT_MASK);
-	WREG32_SMC(ixSCLK_PWRMGT_CNTL, tmp);
-
-	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_0, 0);
-	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_1, 0);
-	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_2, 0);
-	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_3, 0);
-	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_4, 0);
-	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_5, 0);
-	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_6, 0);
-	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_7, 0);
-}
-
-static int ci_upload_firmware(struct amdgpu_device *adev)
-{
-	int i, ret;
-
-	if (amdgpu_ci_is_smc_running(adev)) {
-		DRM_INFO("smc is running, no need to load smc firmware\n");
-		return 0;
-	}
-
-	for (i = 0; i < adev->usec_timeout; i++) {
-		if (RREG32_SMC(ixRCU_UC_EVENTS) & RCU_UC_EVENTS__boot_seq_done_MASK)
-			break;
-	}
-	WREG32_SMC(ixSMC_SYSCON_MISC_CNTL, 1);
-
-	amdgpu_ci_stop_smc_clock(adev);
-	amdgpu_ci_reset_smc(adev);
-
-	ret = amdgpu_ci_load_smc_ucode(adev, SMC_RAM_END);
-
-	return ret;
-
-}
-
-static int ci_get_svi2_voltage_table(struct amdgpu_device *adev,
-				     struct amdgpu_clock_voltage_dependency_table *voltage_dependency_table,
-				     struct atom_voltage_table *voltage_table)
-{
-	u32 i;
-
-	if (voltage_dependency_table == NULL)
-		return -EINVAL;
-
-	voltage_table->mask_low = 0;
-	voltage_table->phase_delay = 0;
-
-	voltage_table->count = voltage_dependency_table->count;
-	for (i = 0; i < voltage_table->count; i++) {
-		voltage_table->entries[i].value = voltage_dependency_table->entries[i].v;
-		voltage_table->entries[i].smio_low = 0;
-	}
-
-	return 0;
-}
-
-static int ci_construct_voltage_tables(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	int ret;
-
-	if (pi->voltage_control == CISLANDS_VOLTAGE_CONTROL_BY_GPIO) {
-		ret = amdgpu_atombios_get_voltage_table(adev, VOLTAGE_TYPE_VDDC,
-							VOLTAGE_OBJ_GPIO_LUT,
-							&pi->vddc_voltage_table);
-		if (ret)
-			return ret;
-	} else if (pi->voltage_control == CISLANDS_VOLTAGE_CONTROL_BY_SVID2) {
-		ret = ci_get_svi2_voltage_table(adev,
-						&adev->pm.dpm.dyn_state.vddc_dependency_on_mclk,
-						&pi->vddc_voltage_table);
-		if (ret)
-			return ret;
-	}
-
-	if (pi->vddc_voltage_table.count > SMU7_MAX_LEVELS_VDDC)
-		ci_trim_voltage_table_to_fit_state_table(adev, SMU7_MAX_LEVELS_VDDC,
-							 &pi->vddc_voltage_table);
-
-	if (pi->vddci_control == CISLANDS_VOLTAGE_CONTROL_BY_GPIO) {
-		ret = amdgpu_atombios_get_voltage_table(adev, VOLTAGE_TYPE_VDDCI,
-							VOLTAGE_OBJ_GPIO_LUT,
-							&pi->vddci_voltage_table);
-		if (ret)
-			return ret;
-	} else if (pi->vddci_control == CISLANDS_VOLTAGE_CONTROL_BY_SVID2) {
-		ret = ci_get_svi2_voltage_table(adev,
-						&adev->pm.dpm.dyn_state.vddci_dependency_on_mclk,
-						&pi->vddci_voltage_table);
-		if (ret)
-			return ret;
-	}
-
-	if (pi->vddci_voltage_table.count > SMU7_MAX_LEVELS_VDDCI)
-		ci_trim_voltage_table_to_fit_state_table(adev, SMU7_MAX_LEVELS_VDDCI,
-							 &pi->vddci_voltage_table);
-
-	if (pi->mvdd_control == CISLANDS_VOLTAGE_CONTROL_BY_GPIO) {
-		ret = amdgpu_atombios_get_voltage_table(adev, VOLTAGE_TYPE_MVDDC,
-							VOLTAGE_OBJ_GPIO_LUT,
-							&pi->mvdd_voltage_table);
-		if (ret)
-			return ret;
-	} else if (pi->mvdd_control == CISLANDS_VOLTAGE_CONTROL_BY_SVID2) {
-		ret = ci_get_svi2_voltage_table(adev,
-						&adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk,
-						&pi->mvdd_voltage_table);
-		if (ret)
-			return ret;
-	}
-
-	if (pi->mvdd_voltage_table.count > SMU7_MAX_LEVELS_MVDD)
-		ci_trim_voltage_table_to_fit_state_table(adev, SMU7_MAX_LEVELS_MVDD,
-							 &pi->mvdd_voltage_table);
-
-	return 0;
-}
-
-static void ci_populate_smc_voltage_table(struct amdgpu_device *adev,
-					  struct atom_voltage_table_entry *voltage_table,
-					  SMU7_Discrete_VoltageLevel *smc_voltage_table)
-{
-	int ret;
-
-	ret = ci_get_std_voltage_value_sidd(adev, voltage_table,
-					    &smc_voltage_table->StdVoltageHiSidd,
-					    &smc_voltage_table->StdVoltageLoSidd);
-
-	if (ret) {
-		smc_voltage_table->StdVoltageHiSidd = voltage_table->value * VOLTAGE_SCALE;
-		smc_voltage_table->StdVoltageLoSidd = voltage_table->value * VOLTAGE_SCALE;
-	}
-
-	smc_voltage_table->Voltage = cpu_to_be16(voltage_table->value * VOLTAGE_SCALE);
-	smc_voltage_table->StdVoltageHiSidd =
-		cpu_to_be16(smc_voltage_table->StdVoltageHiSidd);
-	smc_voltage_table->StdVoltageLoSidd =
-		cpu_to_be16(smc_voltage_table->StdVoltageLoSidd);
-}
-
-static int ci_populate_smc_vddc_table(struct amdgpu_device *adev,
-				      SMU7_Discrete_DpmTable *table)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	unsigned int count;
-
-	table->VddcLevelCount = pi->vddc_voltage_table.count;
-	for (count = 0; count < table->VddcLevelCount; count++) {
-		ci_populate_smc_voltage_table(adev,
-					      &pi->vddc_voltage_table.entries[count],
-					      &table->VddcLevel[count]);
-
-		if (pi->voltage_control == CISLANDS_VOLTAGE_CONTROL_BY_GPIO)
-			table->VddcLevel[count].Smio |=
-				pi->vddc_voltage_table.entries[count].smio_low;
-		else
-			table->VddcLevel[count].Smio = 0;
-	}
-	table->VddcLevelCount = cpu_to_be32(table->VddcLevelCount);
-
-	return 0;
-}
-
-static int ci_populate_smc_vddci_table(struct amdgpu_device *adev,
-				       SMU7_Discrete_DpmTable *table)
-{
-	unsigned int count;
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	table->VddciLevelCount = pi->vddci_voltage_table.count;
-	for (count = 0; count < table->VddciLevelCount; count++) {
-		ci_populate_smc_voltage_table(adev,
-					      &pi->vddci_voltage_table.entries[count],
-					      &table->VddciLevel[count]);
-
-		if (pi->vddci_control == CISLANDS_VOLTAGE_CONTROL_BY_GPIO)
-			table->VddciLevel[count].Smio |=
-				pi->vddci_voltage_table.entries[count].smio_low;
-		else
-			table->VddciLevel[count].Smio = 0;
-	}
-	table->VddciLevelCount = cpu_to_be32(table->VddciLevelCount);
-
-	return 0;
-}
-
-static int ci_populate_smc_mvdd_table(struct amdgpu_device *adev,
-				      SMU7_Discrete_DpmTable *table)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	unsigned int count;
-
-	table->MvddLevelCount = pi->mvdd_voltage_table.count;
-	for (count = 0; count < table->MvddLevelCount; count++) {
-		ci_populate_smc_voltage_table(adev,
-					      &pi->mvdd_voltage_table.entries[count],
-					      &table->MvddLevel[count]);
-
-		if (pi->mvdd_control == CISLANDS_VOLTAGE_CONTROL_BY_GPIO)
-			table->MvddLevel[count].Smio |=
-				pi->mvdd_voltage_table.entries[count].smio_low;
-		else
-			table->MvddLevel[count].Smio = 0;
-	}
-	table->MvddLevelCount = cpu_to_be32(table->MvddLevelCount);
-
-	return 0;
-}
-
-static int ci_populate_smc_voltage_tables(struct amdgpu_device *adev,
-					  SMU7_Discrete_DpmTable *table)
-{
-	int ret;
-
-	ret = ci_populate_smc_vddc_table(adev, table);
-	if (ret)
-		return ret;
-
-	ret = ci_populate_smc_vddci_table(adev, table);
-	if (ret)
-		return ret;
-
-	ret = ci_populate_smc_mvdd_table(adev, table);
-	if (ret)
-		return ret;
-
-	return 0;
-}
-
-static int ci_populate_mvdd_value(struct amdgpu_device *adev, u32 mclk,
-				  SMU7_Discrete_VoltageLevel *voltage)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u32 i = 0;
-
-	if (pi->mvdd_control != CISLANDS_VOLTAGE_CONTROL_NONE) {
-		for (i = 0; i < adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk.count; i++) {
-			if (mclk <= adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk.entries[i].clk) {
-				voltage->Voltage = pi->mvdd_voltage_table.entries[i].value;
-				break;
-			}
-		}
-
-		if (i >= adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk.count)
-			return -EINVAL;
-	}
-
-	return -EINVAL;
-}
-
-static int ci_get_std_voltage_value_sidd(struct amdgpu_device *adev,
-					 struct atom_voltage_table_entry *voltage_table,
-					 u16 *std_voltage_hi_sidd, u16 *std_voltage_lo_sidd)
-{
-	u16 v_index, idx;
-	bool voltage_found = false;
-	*std_voltage_hi_sidd = voltage_table->value * VOLTAGE_SCALE;
-	*std_voltage_lo_sidd = voltage_table->value * VOLTAGE_SCALE;
-
-	if (adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries == NULL)
-		return -EINVAL;
-
-	if (adev->pm.dpm.dyn_state.cac_leakage_table.entries) {
-		for (v_index = 0; (u32)v_index < adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.count; v_index++) {
-			if (voltage_table->value ==
-			    adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[v_index].v) {
-				voltage_found = true;
-				if ((u32)v_index < adev->pm.dpm.dyn_state.cac_leakage_table.count)
-					idx = v_index;
-				else
-					idx = adev->pm.dpm.dyn_state.cac_leakage_table.count - 1;
-				*std_voltage_lo_sidd =
-					adev->pm.dpm.dyn_state.cac_leakage_table.entries[idx].vddc * VOLTAGE_SCALE;
-				*std_voltage_hi_sidd =
-					adev->pm.dpm.dyn_state.cac_leakage_table.entries[idx].leakage * VOLTAGE_SCALE;
-				break;
-			}
-		}
-
-		if (!voltage_found) {
-			for (v_index = 0; (u32)v_index < adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.count; v_index++) {
-				if (voltage_table->value <=
-				    adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[v_index].v) {
-					voltage_found = true;
-					if ((u32)v_index < adev->pm.dpm.dyn_state.cac_leakage_table.count)
-						idx = v_index;
-					else
-						idx = adev->pm.dpm.dyn_state.cac_leakage_table.count - 1;
-					*std_voltage_lo_sidd =
-						adev->pm.dpm.dyn_state.cac_leakage_table.entries[idx].vddc * VOLTAGE_SCALE;
-					*std_voltage_hi_sidd =
-						adev->pm.dpm.dyn_state.cac_leakage_table.entries[idx].leakage * VOLTAGE_SCALE;
-					break;
-				}
-			}
-		}
-	}
-
-	return 0;
-}
-
-static void ci_populate_phase_value_based_on_sclk(struct amdgpu_device *adev,
-						  const struct amdgpu_phase_shedding_limits_table *limits,
-						  u32 sclk,
-						  u32 *phase_shedding)
-{
-	unsigned int i;
-
-	*phase_shedding = 1;
-
-	for (i = 0; i < limits->count; i++) {
-		if (sclk < limits->entries[i].sclk) {
-			*phase_shedding = i;
-			break;
-		}
-	}
-}
-
-static void ci_populate_phase_value_based_on_mclk(struct amdgpu_device *adev,
-						  const struct amdgpu_phase_shedding_limits_table *limits,
-						  u32 mclk,
-						  u32 *phase_shedding)
-{
-	unsigned int i;
-
-	*phase_shedding = 1;
-
-	for (i = 0; i < limits->count; i++) {
-		if (mclk < limits->entries[i].mclk) {
-			*phase_shedding = i;
-			break;
-		}
-	}
-}
-
-static int ci_init_arb_table_index(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u32 tmp;
-	int ret;
-
-	ret = amdgpu_ci_read_smc_sram_dword(adev, pi->arb_table_start,
-				     &tmp, pi->sram_end);
-	if (ret)
-		return ret;
-
-	tmp &= 0x00FFFFFF;
-	tmp |= MC_CG_ARB_FREQ_F1 << 24;
-
-	return amdgpu_ci_write_smc_sram_dword(adev, pi->arb_table_start,
-				       tmp, pi->sram_end);
-}
-
-static int ci_get_dependency_volt_by_clk(struct amdgpu_device *adev,
-					 struct amdgpu_clock_voltage_dependency_table *allowed_clock_voltage_table,
-					 u32 clock, u32 *voltage)
-{
-	u32 i = 0;
-
-	if (allowed_clock_voltage_table->count == 0)
-		return -EINVAL;
-
-	for (i = 0; i < allowed_clock_voltage_table->count; i++) {
-		if (allowed_clock_voltage_table->entries[i].clk >= clock) {
-			*voltage = allowed_clock_voltage_table->entries[i].v;
-			return 0;
-		}
-	}
-
-	*voltage = allowed_clock_voltage_table->entries[i-1].v;
-
-	return 0;
-}
-
-static u8 ci_get_sleep_divider_id_from_clock(u32 sclk, u32 min_sclk_in_sr)
-{
-	u32 i;
-	u32 tmp;
-	u32 min = max(min_sclk_in_sr, (u32)CISLAND_MINIMUM_ENGINE_CLOCK);
-
-	if (sclk < min)
-		return 0;
-
-	for (i = CISLAND_MAX_DEEPSLEEP_DIVIDER_ID;  ; i--) {
-		tmp = sclk >> i;
-		if (tmp >= min || i == 0)
-			break;
-	}
-
-	return (u8)i;
-}
-
-static int ci_initial_switch_from_arb_f0_to_f1(struct amdgpu_device *adev)
-{
-	return ci_copy_and_switch_arb_sets(adev, MC_CG_ARB_FREQ_F0, MC_CG_ARB_FREQ_F1);
-}
-
-static int ci_reset_to_default(struct amdgpu_device *adev)
-{
-	return (amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_ResetToDefaults) == PPSMC_Result_OK) ?
-		0 : -EINVAL;
-}
-
-static int ci_force_switch_to_arb_f0(struct amdgpu_device *adev)
-{
-	u32 tmp;
-
-	tmp = (RREG32_SMC(ixSMC_SCRATCH9) & 0x0000ff00) >> 8;
-
-	if (tmp == MC_CG_ARB_FREQ_F0)
-		return 0;
-
-	return ci_copy_and_switch_arb_sets(adev, tmp, MC_CG_ARB_FREQ_F0);
-}
-
-static void ci_register_patching_mc_arb(struct amdgpu_device *adev,
-					const u32 engine_clock,
-					const u32 memory_clock,
-					u32 *dram_timimg2)
-{
-	bool patch;
-	u32 tmp, tmp2;
-
-	tmp = RREG32(mmMC_SEQ_MISC0);
-	patch = ((tmp & 0x0000f00) == 0x300) ? true : false;
-
-	if (patch &&
-	    ((adev->pdev->device == 0x67B0) ||
-	     (adev->pdev->device == 0x67B1))) {
-		if ((memory_clock > 100000) && (memory_clock <= 125000)) {
-			tmp2 = (((0x31 * engine_clock) / 125000) - 1) & 0xff;
-			*dram_timimg2 &= ~0x00ff0000;
-			*dram_timimg2 |= tmp2 << 16;
-		} else if ((memory_clock > 125000) && (memory_clock <= 137500)) {
-			tmp2 = (((0x36 * engine_clock) / 137500) - 1) & 0xff;
-			*dram_timimg2 &= ~0x00ff0000;
-			*dram_timimg2 |= tmp2 << 16;
-		}
-	}
-}
-
-static int ci_populate_memory_timing_parameters(struct amdgpu_device *adev,
-						u32 sclk,
-						u32 mclk,
-						SMU7_Discrete_MCArbDramTimingTableEntry *arb_regs)
-{
-	u32 dram_timing;
-	u32 dram_timing2;
-	u32 burst_time;
-
-	amdgpu_atombios_set_engine_dram_timings(adev, sclk, mclk);
-
-	dram_timing  = RREG32(mmMC_ARB_DRAM_TIMING);
-	dram_timing2 = RREG32(mmMC_ARB_DRAM_TIMING2);
-	burst_time = RREG32(mmMC_ARB_BURST_TIME) & MC_ARB_BURST_TIME__STATE0_MASK;
-
-	ci_register_patching_mc_arb(adev, sclk, mclk, &dram_timing2);
-
-	arb_regs->McArbDramTiming  = cpu_to_be32(dram_timing);
-	arb_regs->McArbDramTiming2 = cpu_to_be32(dram_timing2);
-	arb_regs->McArbBurstTime = (u8)burst_time;
-
-	return 0;
-}
-
-static int ci_do_program_memory_timing_parameters(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	SMU7_Discrete_MCArbDramTimingTable arb_regs;
-	u32 i, j;
-	int ret =  0;
-
-	memset(&arb_regs, 0, sizeof(SMU7_Discrete_MCArbDramTimingTable));
-
-	for (i = 0; i < pi->dpm_table.sclk_table.count; i++) {
-		for (j = 0; j < pi->dpm_table.mclk_table.count; j++) {
-			ret = ci_populate_memory_timing_parameters(adev,
-								   pi->dpm_table.sclk_table.dpm_levels[i].value,
-								   pi->dpm_table.mclk_table.dpm_levels[j].value,
-								   &arb_regs.entries[i][j]);
-			if (ret)
-				break;
-		}
-	}
-
-	if (ret == 0)
-		ret = amdgpu_ci_copy_bytes_to_smc(adev,
-					   pi->arb_table_start,
-					   (u8 *)&arb_regs,
-					   sizeof(SMU7_Discrete_MCArbDramTimingTable),
-					   pi->sram_end);
-
-	return ret;
-}
-
-static int ci_program_memory_timing_parameters(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	if (pi->need_update_smu7_dpm_table == 0)
-		return 0;
-
-	return ci_do_program_memory_timing_parameters(adev);
-}
-
-static void ci_populate_smc_initial_state(struct amdgpu_device *adev,
-					  struct amdgpu_ps *amdgpu_boot_state)
-{
-	struct ci_ps *boot_state = ci_get_ps(amdgpu_boot_state);
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u32 level = 0;
-
-	for (level = 0; level < adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.count; level++) {
-		if (adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[level].clk >=
-		    boot_state->performance_levels[0].sclk) {
-			pi->smc_state_table.GraphicsBootLevel = level;
-			break;
-		}
-	}
-
-	for (level = 0; level < adev->pm.dpm.dyn_state.vddc_dependency_on_mclk.count; level++) {
-		if (adev->pm.dpm.dyn_state.vddc_dependency_on_mclk.entries[level].clk >=
-		    boot_state->performance_levels[0].mclk) {
-			pi->smc_state_table.MemoryBootLevel = level;
-			break;
-		}
-	}
-}
-
-static u32 ci_get_dpm_level_enable_mask_value(struct ci_single_dpm_table *dpm_table)
-{
-	u32 i;
-	u32 mask_value = 0;
-
-	for (i = dpm_table->count; i > 0; i--) {
-		mask_value = mask_value << 1;
-		if (dpm_table->dpm_levels[i-1].enabled)
-			mask_value |= 0x1;
-		else
-			mask_value &= 0xFFFFFFFE;
-	}
-
-	return mask_value;
-}
-
-static void ci_populate_smc_link_level(struct amdgpu_device *adev,
-				       SMU7_Discrete_DpmTable *table)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct ci_dpm_table *dpm_table = &pi->dpm_table;
-	u32 i;
-
-	for (i = 0; i < dpm_table->pcie_speed_table.count; i++) {
-		table->LinkLevel[i].PcieGenSpeed =
-			(u8)dpm_table->pcie_speed_table.dpm_levels[i].value;
-		table->LinkLevel[i].PcieLaneCount =
-			amdgpu_encode_pci_lane_width(dpm_table->pcie_speed_table.dpm_levels[i].param1);
-		table->LinkLevel[i].EnabledForActivity = 1;
-		table->LinkLevel[i].DownT = cpu_to_be32(5);
-		table->LinkLevel[i].UpT = cpu_to_be32(30);
-	}
-
-	pi->smc_state_table.LinkLevelCount = (u8)dpm_table->pcie_speed_table.count;
-	pi->dpm_level_enable_mask.pcie_dpm_enable_mask =
-		ci_get_dpm_level_enable_mask_value(&dpm_table->pcie_speed_table);
-}
-
-static int ci_populate_smc_uvd_level(struct amdgpu_device *adev,
-				     SMU7_Discrete_DpmTable *table)
-{
-	u32 count;
-	struct atom_clock_dividers dividers;
-	int ret = -EINVAL;
-
-	table->UvdLevelCount =
-		adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.count;
-
-	for (count = 0; count < table->UvdLevelCount; count++) {
-		table->UvdLevel[count].VclkFrequency =
-			adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries[count].vclk;
-		table->UvdLevel[count].DclkFrequency =
-			adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries[count].dclk;
-		table->UvdLevel[count].MinVddc =
-			adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries[count].v * VOLTAGE_SCALE;
-		table->UvdLevel[count].MinVddcPhases = 1;
-
-		ret = amdgpu_atombios_get_clock_dividers(adev,
-							 COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
-							 table->UvdLevel[count].VclkFrequency, false, &dividers);
-		if (ret)
-			return ret;
-
-		table->UvdLevel[count].VclkDivider = (u8)dividers.post_divider;
-
-		ret = amdgpu_atombios_get_clock_dividers(adev,
-							 COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
-							 table->UvdLevel[count].DclkFrequency, false, &dividers);
-		if (ret)
-			return ret;
-
-		table->UvdLevel[count].DclkDivider = (u8)dividers.post_divider;
-
-		table->UvdLevel[count].VclkFrequency = cpu_to_be32(table->UvdLevel[count].VclkFrequency);
-		table->UvdLevel[count].DclkFrequency = cpu_to_be32(table->UvdLevel[count].DclkFrequency);
-		table->UvdLevel[count].MinVddc = cpu_to_be16(table->UvdLevel[count].MinVddc);
-	}
-
-	return ret;
-}
-
-static int ci_populate_smc_vce_level(struct amdgpu_device *adev,
-				     SMU7_Discrete_DpmTable *table)
-{
-	u32 count;
-	struct atom_clock_dividers dividers;
-	int ret = -EINVAL;
-
-	table->VceLevelCount =
-		adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.count;
-
-	for (count = 0; count < table->VceLevelCount; count++) {
-		table->VceLevel[count].Frequency =
-			adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries[count].evclk;
-		table->VceLevel[count].MinVoltage =
-			(u16)adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries[count].v * VOLTAGE_SCALE;
-		table->VceLevel[count].MinPhases = 1;
-
-		ret = amdgpu_atombios_get_clock_dividers(adev,
-							 COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
-							 table->VceLevel[count].Frequency, false, &dividers);
-		if (ret)
-			return ret;
-
-		table->VceLevel[count].Divider = (u8)dividers.post_divider;
-
-		table->VceLevel[count].Frequency = cpu_to_be32(table->VceLevel[count].Frequency);
-		table->VceLevel[count].MinVoltage = cpu_to_be16(table->VceLevel[count].MinVoltage);
-	}
-
-	return ret;
-
-}
-
-static int ci_populate_smc_acp_level(struct amdgpu_device *adev,
-				     SMU7_Discrete_DpmTable *table)
-{
-	u32 count;
-	struct atom_clock_dividers dividers;
-	int ret = -EINVAL;
-
-	table->AcpLevelCount = (u8)
-		(adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.count);
-
-	for (count = 0; count < table->AcpLevelCount; count++) {
-		table->AcpLevel[count].Frequency =
-			adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.entries[count].clk;
-		table->AcpLevel[count].MinVoltage =
-			adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.entries[count].v;
-		table->AcpLevel[count].MinPhases = 1;
-
-		ret = amdgpu_atombios_get_clock_dividers(adev,
-							 COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
-							 table->AcpLevel[count].Frequency, false, &dividers);
-		if (ret)
-			return ret;
-
-		table->AcpLevel[count].Divider = (u8)dividers.post_divider;
-
-		table->AcpLevel[count].Frequency = cpu_to_be32(table->AcpLevel[count].Frequency);
-		table->AcpLevel[count].MinVoltage = cpu_to_be16(table->AcpLevel[count].MinVoltage);
-	}
-
-	return ret;
-}
-
-static int ci_populate_smc_samu_level(struct amdgpu_device *adev,
-				      SMU7_Discrete_DpmTable *table)
-{
-	u32 count;
-	struct atom_clock_dividers dividers;
-	int ret = -EINVAL;
-
-	table->SamuLevelCount =
-		adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.count;
-
-	for (count = 0; count < table->SamuLevelCount; count++) {
-		table->SamuLevel[count].Frequency =
-			adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.entries[count].clk;
-		table->SamuLevel[count].MinVoltage =
-			adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.entries[count].v * VOLTAGE_SCALE;
-		table->SamuLevel[count].MinPhases = 1;
-
-		ret = amdgpu_atombios_get_clock_dividers(adev,
-							 COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
-							 table->SamuLevel[count].Frequency, false, &dividers);
-		if (ret)
-			return ret;
-
-		table->SamuLevel[count].Divider = (u8)dividers.post_divider;
-
-		table->SamuLevel[count].Frequency = cpu_to_be32(table->SamuLevel[count].Frequency);
-		table->SamuLevel[count].MinVoltage = cpu_to_be16(table->SamuLevel[count].MinVoltage);
-	}
-
-	return ret;
-}
-
-static int ci_calculate_mclk_params(struct amdgpu_device *adev,
-				    u32 memory_clock,
-				    SMU7_Discrete_MemoryLevel *mclk,
-				    bool strobe_mode,
-				    bool dll_state_on)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u32  dll_cntl = pi->clock_registers.dll_cntl;
-	u32  mclk_pwrmgt_cntl = pi->clock_registers.mclk_pwrmgt_cntl;
-	u32  mpll_ad_func_cntl = pi->clock_registers.mpll_ad_func_cntl;
-	u32  mpll_dq_func_cntl = pi->clock_registers.mpll_dq_func_cntl;
-	u32  mpll_func_cntl = pi->clock_registers.mpll_func_cntl;
-	u32  mpll_func_cntl_1 = pi->clock_registers.mpll_func_cntl_1;
-	u32  mpll_func_cntl_2 = pi->clock_registers.mpll_func_cntl_2;
-	u32  mpll_ss1 = pi->clock_registers.mpll_ss1;
-	u32  mpll_ss2 = pi->clock_registers.mpll_ss2;
-	struct atom_mpll_param mpll_param;
-	int ret;
-
-	ret = amdgpu_atombios_get_memory_pll_dividers(adev, memory_clock, strobe_mode, &mpll_param);
-	if (ret)
-		return ret;
-
-	mpll_func_cntl &= ~MPLL_FUNC_CNTL__BWCTRL_MASK;
-	mpll_func_cntl |= (mpll_param.bwcntl << MPLL_FUNC_CNTL__BWCTRL__SHIFT);
-
-	mpll_func_cntl_1 &= ~(MPLL_FUNC_CNTL_1__CLKF_MASK | MPLL_FUNC_CNTL_1__CLKFRAC_MASK |
-			MPLL_FUNC_CNTL_1__VCO_MODE_MASK);
-	mpll_func_cntl_1 |= (mpll_param.clkf) << MPLL_FUNC_CNTL_1__CLKF__SHIFT |
-		(mpll_param.clkfrac << MPLL_FUNC_CNTL_1__CLKFRAC__SHIFT) |
-		(mpll_param.vco_mode << MPLL_FUNC_CNTL_1__VCO_MODE__SHIFT);
-
-	mpll_ad_func_cntl &= ~MPLL_AD_FUNC_CNTL__YCLK_POST_DIV_MASK;
-	mpll_ad_func_cntl |= (mpll_param.post_div << MPLL_AD_FUNC_CNTL__YCLK_POST_DIV__SHIFT);
-
-	if (adev->mc.vram_type == AMDGPU_VRAM_TYPE_GDDR5) {
-		mpll_dq_func_cntl &= ~(MPLL_DQ_FUNC_CNTL__YCLK_SEL_MASK |
-				MPLL_AD_FUNC_CNTL__YCLK_POST_DIV_MASK);
-		mpll_dq_func_cntl |= (mpll_param.yclk_sel << MPLL_DQ_FUNC_CNTL__YCLK_SEL__SHIFT) |
-				(mpll_param.post_div << MPLL_AD_FUNC_CNTL__YCLK_POST_DIV__SHIFT);
-	}
-
-	if (pi->caps_mclk_ss_support) {
-		struct amdgpu_atom_ss ss;
-		u32 freq_nom;
-		u32 tmp;
-		u32 reference_clock = adev->clock.mpll.reference_freq;
-
-		if (mpll_param.qdr == 1)
-			freq_nom = memory_clock * 4 * (1 << mpll_param.post_div);
-		else
-			freq_nom = memory_clock * 2 * (1 << mpll_param.post_div);
-
-		tmp = (freq_nom / reference_clock);
-		tmp = tmp * tmp;
-		if (amdgpu_atombios_get_asic_ss_info(adev, &ss,
-						     ASIC_INTERNAL_MEMORY_SS, freq_nom)) {
-			u32 clks = reference_clock * 5 / ss.rate;
-			u32 clkv = (u32)((((131 * ss.percentage * ss.rate) / 100) * tmp) / freq_nom);
-
-			mpll_ss1 &= ~MPLL_SS1__CLKV_MASK;
-			mpll_ss1 |= (clkv << MPLL_SS1__CLKV__SHIFT);
-
-			mpll_ss2 &= ~MPLL_SS2__CLKS_MASK;
-			mpll_ss2 |= (clks << MPLL_SS2__CLKS__SHIFT);
-		}
-	}
-
-	mclk_pwrmgt_cntl &= ~MCLK_PWRMGT_CNTL__DLL_SPEED_MASK;
-	mclk_pwrmgt_cntl |= (mpll_param.dll_speed << MCLK_PWRMGT_CNTL__DLL_SPEED__SHIFT);
-
-	if (dll_state_on)
-		mclk_pwrmgt_cntl |= MCLK_PWRMGT_CNTL__MRDCK0_PDNB_MASK |
-			MCLK_PWRMGT_CNTL__MRDCK1_PDNB_MASK;
-	else
-		mclk_pwrmgt_cntl &= ~(MCLK_PWRMGT_CNTL__MRDCK0_PDNB_MASK |
-			MCLK_PWRMGT_CNTL__MRDCK1_PDNB_MASK);
-
-	mclk->MclkFrequency = memory_clock;
-	mclk->MpllFuncCntl = mpll_func_cntl;
-	mclk->MpllFuncCntl_1 = mpll_func_cntl_1;
-	mclk->MpllFuncCntl_2 = mpll_func_cntl_2;
-	mclk->MpllAdFuncCntl = mpll_ad_func_cntl;
-	mclk->MpllDqFuncCntl = mpll_dq_func_cntl;
-	mclk->MclkPwrmgtCntl = mclk_pwrmgt_cntl;
-	mclk->DllCntl = dll_cntl;
-	mclk->MpllSs1 = mpll_ss1;
-	mclk->MpllSs2 = mpll_ss2;
-
-	return 0;
-}
-
-static int ci_populate_single_memory_level(struct amdgpu_device *adev,
-					   u32 memory_clock,
-					   SMU7_Discrete_MemoryLevel *memory_level)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	int ret;
-	bool dll_state_on;
-
-	if (adev->pm.dpm.dyn_state.vddc_dependency_on_mclk.entries) {
-		ret = ci_get_dependency_volt_by_clk(adev,
-						    &adev->pm.dpm.dyn_state.vddc_dependency_on_mclk,
-						    memory_clock, &memory_level->MinVddc);
-		if (ret)
-			return ret;
-	}
-
-	if (adev->pm.dpm.dyn_state.vddci_dependency_on_mclk.entries) {
-		ret = ci_get_dependency_volt_by_clk(adev,
-						    &adev->pm.dpm.dyn_state.vddci_dependency_on_mclk,
-						    memory_clock, &memory_level->MinVddci);
-		if (ret)
-			return ret;
-	}
-
-	if (adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk.entries) {
-		ret = ci_get_dependency_volt_by_clk(adev,
-						    &adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk,
-						    memory_clock, &memory_level->MinMvdd);
-		if (ret)
-			return ret;
-	}
-
-	memory_level->MinVddcPhases = 1;
-
-	if (pi->vddc_phase_shed_control)
-		ci_populate_phase_value_based_on_mclk(adev,
-						      &adev->pm.dpm.dyn_state.phase_shedding_limits_table,
-						      memory_clock,
-						      &memory_level->MinVddcPhases);
-
-	memory_level->EnabledForActivity = 1;
-	memory_level->EnabledForThrottle = 1;
-	memory_level->UpH = 0;
-	memory_level->DownH = 100;
-	memory_level->VoltageDownH = 0;
-	memory_level->ActivityLevel = (u16)pi->mclk_activity_target;
-
-	memory_level->StutterEnable = false;
-	memory_level->StrobeEnable = false;
-	memory_level->EdcReadEnable = false;
-	memory_level->EdcWriteEnable = false;
-	memory_level->RttEnable = false;
-
-	memory_level->DisplayWatermark = PPSMC_DISPLAY_WATERMARK_LOW;
-
-	if (pi->mclk_stutter_mode_threshold &&
-	    (memory_clock <= pi->mclk_stutter_mode_threshold) &&
-	    (!pi->uvd_enabled) &&
-	    (RREG32(mmDPG_PIPE_STUTTER_CONTROL) & DPG_PIPE_STUTTER_CONTROL__STUTTER_ENABLE_MASK) &&
-	    (adev->pm.dpm.new_active_crtc_count <= 2))
-		memory_level->StutterEnable = true;
-
-	if (pi->mclk_strobe_mode_threshold &&
-	    (memory_clock <= pi->mclk_strobe_mode_threshold))
-		memory_level->StrobeEnable = 1;
-
-	if (adev->mc.vram_type == AMDGPU_VRAM_TYPE_GDDR5) {
-		memory_level->StrobeRatio =
-			ci_get_mclk_frequency_ratio(memory_clock, memory_level->StrobeEnable);
-		if (pi->mclk_edc_enable_threshold &&
-		    (memory_clock > pi->mclk_edc_enable_threshold))
-			memory_level->EdcReadEnable = true;
-
-		if (pi->mclk_edc_wr_enable_threshold &&
-		    (memory_clock > pi->mclk_edc_wr_enable_threshold))
-			memory_level->EdcWriteEnable = true;
-
-		if (memory_level->StrobeEnable) {
-			if (ci_get_mclk_frequency_ratio(memory_clock, true) >=
-			    ((RREG32(mmMC_SEQ_MISC7) >> 16) & 0xf))
-				dll_state_on = ((RREG32(mmMC_SEQ_MISC5) >> 1) & 0x1) ? true : false;
-			else
-				dll_state_on = ((RREG32(mmMC_SEQ_MISC6) >> 1) & 0x1) ? true : false;
-		} else {
-			dll_state_on = pi->dll_default_on;
-		}
-	} else {
-		memory_level->StrobeRatio = ci_get_ddr3_mclk_frequency_ratio(memory_clock);
-		dll_state_on = ((RREG32(mmMC_SEQ_MISC5) >> 1) & 0x1) ? true : false;
-	}
-
-	ret = ci_calculate_mclk_params(adev, memory_clock, memory_level, memory_level->StrobeEnable, dll_state_on);
-	if (ret)
-		return ret;
-
-	memory_level->MinVddc = cpu_to_be32(memory_level->MinVddc * VOLTAGE_SCALE);
-	memory_level->MinVddcPhases = cpu_to_be32(memory_level->MinVddcPhases);
-	memory_level->MinVddci = cpu_to_be32(memory_level->MinVddci * VOLTAGE_SCALE);
-	memory_level->MinMvdd = cpu_to_be32(memory_level->MinMvdd * VOLTAGE_SCALE);
-
-	memory_level->MclkFrequency = cpu_to_be32(memory_level->MclkFrequency);
-	memory_level->ActivityLevel = cpu_to_be16(memory_level->ActivityLevel);
-	memory_level->MpllFuncCntl = cpu_to_be32(memory_level->MpllFuncCntl);
-	memory_level->MpllFuncCntl_1 = cpu_to_be32(memory_level->MpllFuncCntl_1);
-	memory_level->MpllFuncCntl_2 = cpu_to_be32(memory_level->MpllFuncCntl_2);
-	memory_level->MpllAdFuncCntl = cpu_to_be32(memory_level->MpllAdFuncCntl);
-	memory_level->MpllDqFuncCntl = cpu_to_be32(memory_level->MpllDqFuncCntl);
-	memory_level->MclkPwrmgtCntl = cpu_to_be32(memory_level->MclkPwrmgtCntl);
-	memory_level->DllCntl = cpu_to_be32(memory_level->DllCntl);
-	memory_level->MpllSs1 = cpu_to_be32(memory_level->MpllSs1);
-	memory_level->MpllSs2 = cpu_to_be32(memory_level->MpllSs2);
-
-	return 0;
-}
-
-static int ci_populate_smc_acpi_level(struct amdgpu_device *adev,
-				      SMU7_Discrete_DpmTable *table)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct atom_clock_dividers dividers;
-	SMU7_Discrete_VoltageLevel voltage_level;
-	u32 spll_func_cntl = pi->clock_registers.cg_spll_func_cntl;
-	u32 spll_func_cntl_2 = pi->clock_registers.cg_spll_func_cntl_2;
-	u32 dll_cntl = pi->clock_registers.dll_cntl;
-	u32 mclk_pwrmgt_cntl = pi->clock_registers.mclk_pwrmgt_cntl;
-	int ret;
-
-	table->ACPILevel.Flags &= ~PPSMC_SWSTATE_FLAG_DC;
-
-	if (pi->acpi_vddc)
-		table->ACPILevel.MinVddc = cpu_to_be32(pi->acpi_vddc * VOLTAGE_SCALE);
-	else
-		table->ACPILevel.MinVddc = cpu_to_be32(pi->min_vddc_in_pp_table * VOLTAGE_SCALE);
-
-	table->ACPILevel.MinVddcPhases = pi->vddc_phase_shed_control ? 0 : 1;
-
-	table->ACPILevel.SclkFrequency = adev->clock.spll.reference_freq;
-
-	ret = amdgpu_atombios_get_clock_dividers(adev,
-						 COMPUTE_GPUCLK_INPUT_FLAG_SCLK,
-						 table->ACPILevel.SclkFrequency, false, &dividers);
-	if (ret)
-		return ret;
-
-	table->ACPILevel.SclkDid = (u8)dividers.post_divider;
-	table->ACPILevel.DisplayWatermark = PPSMC_DISPLAY_WATERMARK_LOW;
-	table->ACPILevel.DeepSleepDivId = 0;
-
-	spll_func_cntl &= ~CG_SPLL_FUNC_CNTL__SPLL_PWRON_MASK;
-	spll_func_cntl |= CG_SPLL_FUNC_CNTL__SPLL_RESET_MASK;
-
-	spll_func_cntl_2 &= ~CG_SPLL_FUNC_CNTL_2__SCLK_MUX_SEL_MASK;
-	spll_func_cntl_2 |= (4 << CG_SPLL_FUNC_CNTL_2__SCLK_MUX_SEL__SHIFT);
-
-	table->ACPILevel.CgSpllFuncCntl = spll_func_cntl;
-	table->ACPILevel.CgSpllFuncCntl2 = spll_func_cntl_2;
-	table->ACPILevel.CgSpllFuncCntl3 = pi->clock_registers.cg_spll_func_cntl_3;
-	table->ACPILevel.CgSpllFuncCntl4 = pi->clock_registers.cg_spll_func_cntl_4;
-	table->ACPILevel.SpllSpreadSpectrum = pi->clock_registers.cg_spll_spread_spectrum;
-	table->ACPILevel.SpllSpreadSpectrum2 = pi->clock_registers.cg_spll_spread_spectrum_2;
-	table->ACPILevel.CcPwrDynRm = 0;
-	table->ACPILevel.CcPwrDynRm1 = 0;
-
-	table->ACPILevel.Flags = cpu_to_be32(table->ACPILevel.Flags);
-	table->ACPILevel.MinVddcPhases = cpu_to_be32(table->ACPILevel.MinVddcPhases);
-	table->ACPILevel.SclkFrequency = cpu_to_be32(table->ACPILevel.SclkFrequency);
-	table->ACPILevel.CgSpllFuncCntl = cpu_to_be32(table->ACPILevel.CgSpllFuncCntl);
-	table->ACPILevel.CgSpllFuncCntl2 = cpu_to_be32(table->ACPILevel.CgSpllFuncCntl2);
-	table->ACPILevel.CgSpllFuncCntl3 = cpu_to_be32(table->ACPILevel.CgSpllFuncCntl3);
-	table->ACPILevel.CgSpllFuncCntl4 = cpu_to_be32(table->ACPILevel.CgSpllFuncCntl4);
-	table->ACPILevel.SpllSpreadSpectrum = cpu_to_be32(table->ACPILevel.SpllSpreadSpectrum);
-	table->ACPILevel.SpllSpreadSpectrum2 = cpu_to_be32(table->ACPILevel.SpllSpreadSpectrum2);
-	table->ACPILevel.CcPwrDynRm = cpu_to_be32(table->ACPILevel.CcPwrDynRm);
-	table->ACPILevel.CcPwrDynRm1 = cpu_to_be32(table->ACPILevel.CcPwrDynRm1);
-
-	table->MemoryACPILevel.MinVddc = table->ACPILevel.MinVddc;
-	table->MemoryACPILevel.MinVddcPhases = table->ACPILevel.MinVddcPhases;
-
-	if (pi->vddci_control != CISLANDS_VOLTAGE_CONTROL_NONE) {
-		if (pi->acpi_vddci)
-			table->MemoryACPILevel.MinVddci =
-				cpu_to_be32(pi->acpi_vddci * VOLTAGE_SCALE);
-		else
-			table->MemoryACPILevel.MinVddci =
-				cpu_to_be32(pi->min_vddci_in_pp_table * VOLTAGE_SCALE);
-	}
-
-	if (ci_populate_mvdd_value(adev, 0, &voltage_level))
-		table->MemoryACPILevel.MinMvdd = 0;
-	else
-		table->MemoryACPILevel.MinMvdd =
-			cpu_to_be32(voltage_level.Voltage * VOLTAGE_SCALE);
-
-	mclk_pwrmgt_cntl |= MCLK_PWRMGT_CNTL__MRDCK0_RESET_MASK |
-		MCLK_PWRMGT_CNTL__MRDCK1_RESET_MASK;
-	mclk_pwrmgt_cntl &= ~(MCLK_PWRMGT_CNTL__MRDCK0_PDNB_MASK |
-			MCLK_PWRMGT_CNTL__MRDCK1_PDNB_MASK);
-
-	dll_cntl &= ~(DLL_CNTL__MRDCK0_BYPASS_MASK | DLL_CNTL__MRDCK1_BYPASS_MASK);
-
-	table->MemoryACPILevel.DllCntl = cpu_to_be32(dll_cntl);
-	table->MemoryACPILevel.MclkPwrmgtCntl = cpu_to_be32(mclk_pwrmgt_cntl);
-	table->MemoryACPILevel.MpllAdFuncCntl =
-		cpu_to_be32(pi->clock_registers.mpll_ad_func_cntl);
-	table->MemoryACPILevel.MpllDqFuncCntl =
-		cpu_to_be32(pi->clock_registers.mpll_dq_func_cntl);
-	table->MemoryACPILevel.MpllFuncCntl =
-		cpu_to_be32(pi->clock_registers.mpll_func_cntl);
-	table->MemoryACPILevel.MpllFuncCntl_1 =
-		cpu_to_be32(pi->clock_registers.mpll_func_cntl_1);
-	table->MemoryACPILevel.MpllFuncCntl_2 =
-		cpu_to_be32(pi->clock_registers.mpll_func_cntl_2);
-	table->MemoryACPILevel.MpllSs1 = cpu_to_be32(pi->clock_registers.mpll_ss1);
-	table->MemoryACPILevel.MpllSs2 = cpu_to_be32(pi->clock_registers.mpll_ss2);
-
-	table->MemoryACPILevel.EnabledForThrottle = 0;
-	table->MemoryACPILevel.EnabledForActivity = 0;
-	table->MemoryACPILevel.UpH = 0;
-	table->MemoryACPILevel.DownH = 100;
-	table->MemoryACPILevel.VoltageDownH = 0;
-	table->MemoryACPILevel.ActivityLevel =
-		cpu_to_be16((u16)pi->mclk_activity_target);
-
-	table->MemoryACPILevel.StutterEnable = false;
-	table->MemoryACPILevel.StrobeEnable = false;
-	table->MemoryACPILevel.EdcReadEnable = false;
-	table->MemoryACPILevel.EdcWriteEnable = false;
-	table->MemoryACPILevel.RttEnable = false;
-
-	return 0;
-}
-
-
-static int ci_enable_ulv(struct amdgpu_device *adev, bool enable)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct ci_ulv_parm *ulv = &pi->ulv;
-
-	if (ulv->supported) {
-		if (enable)
-			return (amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_EnableULV) == PPSMC_Result_OK) ?
-				0 : -EINVAL;
-		else
-			return (amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_DisableULV) == PPSMC_Result_OK) ?
-				0 : -EINVAL;
-	}
-
-	return 0;
-}
-
-static int ci_populate_ulv_level(struct amdgpu_device *adev,
-				 SMU7_Discrete_Ulv *state)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u16 ulv_voltage = adev->pm.dpm.backbias_response_time;
-
-	state->CcPwrDynRm = 0;
-	state->CcPwrDynRm1 = 0;
-
-	if (ulv_voltage == 0) {
-		pi->ulv.supported = false;
-		return 0;
-	}
-
-	if (pi->voltage_control != CISLANDS_VOLTAGE_CONTROL_BY_SVID2) {
-		if (ulv_voltage > adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[0].v)
-			state->VddcOffset = 0;
-		else
-			state->VddcOffset =
-				adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[0].v - ulv_voltage;
-	} else {
-		if (ulv_voltage > adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[0].v)
-			state->VddcOffsetVid = 0;
-		else
-			state->VddcOffsetVid = (u8)
-				((adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[0].v - ulv_voltage) *
-				 VOLTAGE_VID_OFFSET_SCALE2 / VOLTAGE_VID_OFFSET_SCALE1);
-	}
-	state->VddcPhase = pi->vddc_phase_shed_control ? 0 : 1;
-
-	state->CcPwrDynRm = cpu_to_be32(state->CcPwrDynRm);
-	state->CcPwrDynRm1 = cpu_to_be32(state->CcPwrDynRm1);
-	state->VddcOffset = cpu_to_be16(state->VddcOffset);
-
-	return 0;
-}
-
-static int ci_calculate_sclk_params(struct amdgpu_device *adev,
-				    u32 engine_clock,
-				    SMU7_Discrete_GraphicsLevel *sclk)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct atom_clock_dividers dividers;
-	u32 spll_func_cntl_3 = pi->clock_registers.cg_spll_func_cntl_3;
-	u32 spll_func_cntl_4 = pi->clock_registers.cg_spll_func_cntl_4;
-	u32 cg_spll_spread_spectrum = pi->clock_registers.cg_spll_spread_spectrum;
-	u32 cg_spll_spread_spectrum_2 = pi->clock_registers.cg_spll_spread_spectrum_2;
-	u32 reference_clock = adev->clock.spll.reference_freq;
-	u32 reference_divider;
-	u32 fbdiv;
-	int ret;
-
-	ret = amdgpu_atombios_get_clock_dividers(adev,
-						 COMPUTE_GPUCLK_INPUT_FLAG_SCLK,
-						 engine_clock, false, &dividers);
-	if (ret)
-		return ret;
-
-	reference_divider = 1 + dividers.ref_div;
-	fbdiv = dividers.fb_div & 0x3FFFFFF;
-
-	spll_func_cntl_3 &= ~CG_SPLL_FUNC_CNTL_3__SPLL_FB_DIV_MASK;
-	spll_func_cntl_3 |= (fbdiv << CG_SPLL_FUNC_CNTL_3__SPLL_FB_DIV__SHIFT);
-	spll_func_cntl_3 |= CG_SPLL_FUNC_CNTL_3__SPLL_DITHEN_MASK;
-
-	if (pi->caps_sclk_ss_support) {
-		struct amdgpu_atom_ss ss;
-		u32 vco_freq = engine_clock * dividers.post_div;
-
-		if (amdgpu_atombios_get_asic_ss_info(adev, &ss,
-						     ASIC_INTERNAL_ENGINE_SS, vco_freq)) {
-			u32 clk_s = reference_clock * 5 / (reference_divider * ss.rate);
-			u32 clk_v = 4 * ss.percentage * fbdiv / (clk_s * 10000);
-
-			cg_spll_spread_spectrum &= ~(CG_SPLL_SPREAD_SPECTRUM__CLKS_MASK | CG_SPLL_SPREAD_SPECTRUM__SSEN_MASK);
-			cg_spll_spread_spectrum |= (clk_s << CG_SPLL_SPREAD_SPECTRUM__CLKS__SHIFT);
-			cg_spll_spread_spectrum |= (1 << CG_SPLL_SPREAD_SPECTRUM__SSEN__SHIFT);
-
-			cg_spll_spread_spectrum_2 &= ~CG_SPLL_SPREAD_SPECTRUM_2__CLKV_MASK;
-			cg_spll_spread_spectrum_2 |= (clk_v << CG_SPLL_SPREAD_SPECTRUM_2__CLKV__SHIFT);
-		}
-	}
-
-	sclk->SclkFrequency = engine_clock;
-	sclk->CgSpllFuncCntl3 = spll_func_cntl_3;
-	sclk->CgSpllFuncCntl4 = spll_func_cntl_4;
-	sclk->SpllSpreadSpectrum = cg_spll_spread_spectrum;
-	sclk->SpllSpreadSpectrum2  = cg_spll_spread_spectrum_2;
-	sclk->SclkDid = (u8)dividers.post_divider;
-
-	return 0;
-}
-
-static int ci_populate_single_graphic_level(struct amdgpu_device *adev,
-					    u32 engine_clock,
-					    u16 sclk_activity_level_t,
-					    SMU7_Discrete_GraphicsLevel *graphic_level)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	int ret;
-
-	ret = ci_calculate_sclk_params(adev, engine_clock, graphic_level);
-	if (ret)
-		return ret;
-
-	ret = ci_get_dependency_volt_by_clk(adev,
-					    &adev->pm.dpm.dyn_state.vddc_dependency_on_sclk,
-					    engine_clock, &graphic_level->MinVddc);
-	if (ret)
-		return ret;
-
-	graphic_level->SclkFrequency = engine_clock;
-
-	graphic_level->Flags =  0;
-	graphic_level->MinVddcPhases = 1;
-
-	if (pi->vddc_phase_shed_control)
-		ci_populate_phase_value_based_on_sclk(adev,
-						      &adev->pm.dpm.dyn_state.phase_shedding_limits_table,
-						      engine_clock,
-						      &graphic_level->MinVddcPhases);
-
-	graphic_level->ActivityLevel = sclk_activity_level_t;
-
-	graphic_level->CcPwrDynRm = 0;
-	graphic_level->CcPwrDynRm1 = 0;
-	graphic_level->EnabledForThrottle = 1;
-	graphic_level->UpH = 0;
-	graphic_level->DownH = 0;
-	graphic_level->VoltageDownH = 0;
-	graphic_level->PowerThrottle = 0;
-
-	if (pi->caps_sclk_ds)
-		graphic_level->DeepSleepDivId = ci_get_sleep_divider_id_from_clock(engine_clock,
-										   CISLAND_MINIMUM_ENGINE_CLOCK);
-
-	graphic_level->DisplayWatermark = PPSMC_DISPLAY_WATERMARK_LOW;
-
-	graphic_level->Flags = cpu_to_be32(graphic_level->Flags);
-	graphic_level->MinVddc = cpu_to_be32(graphic_level->MinVddc * VOLTAGE_SCALE);
-	graphic_level->MinVddcPhases = cpu_to_be32(graphic_level->MinVddcPhases);
-	graphic_level->SclkFrequency = cpu_to_be32(graphic_level->SclkFrequency);
-	graphic_level->ActivityLevel = cpu_to_be16(graphic_level->ActivityLevel);
-	graphic_level->CgSpllFuncCntl3 = cpu_to_be32(graphic_level->CgSpllFuncCntl3);
-	graphic_level->CgSpllFuncCntl4 = cpu_to_be32(graphic_level->CgSpllFuncCntl4);
-	graphic_level->SpllSpreadSpectrum = cpu_to_be32(graphic_level->SpllSpreadSpectrum);
-	graphic_level->SpllSpreadSpectrum2 = cpu_to_be32(graphic_level->SpllSpreadSpectrum2);
-	graphic_level->CcPwrDynRm = cpu_to_be32(graphic_level->CcPwrDynRm);
-	graphic_level->CcPwrDynRm1 = cpu_to_be32(graphic_level->CcPwrDynRm1);
-
-	return 0;
-}
-
-static int ci_populate_all_graphic_levels(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct ci_dpm_table *dpm_table = &pi->dpm_table;
-	u32 level_array_address = pi->dpm_table_start +
-		offsetof(SMU7_Discrete_DpmTable, GraphicsLevel);
-	u32 level_array_size = sizeof(SMU7_Discrete_GraphicsLevel) *
-		SMU7_MAX_LEVELS_GRAPHICS;
-	SMU7_Discrete_GraphicsLevel *levels = pi->smc_state_table.GraphicsLevel;
-	u32 i, ret;
-
-	memset(levels, 0, level_array_size);
-
-	for (i = 0; i < dpm_table->sclk_table.count; i++) {
-		ret = ci_populate_single_graphic_level(adev,
-						       dpm_table->sclk_table.dpm_levels[i].value,
-						       (u16)pi->activity_target[i],
-						       &pi->smc_state_table.GraphicsLevel[i]);
-		if (ret)
-			return ret;
-		if (i > 1)
-			pi->smc_state_table.GraphicsLevel[i].DeepSleepDivId = 0;
-		if (i == (dpm_table->sclk_table.count - 1))
-			pi->smc_state_table.GraphicsLevel[i].DisplayWatermark =
-				PPSMC_DISPLAY_WATERMARK_HIGH;
-	}
-	pi->smc_state_table.GraphicsLevel[0].EnabledForActivity = 1;
-
-	pi->smc_state_table.GraphicsDpmLevelCount = (u8)dpm_table->sclk_table.count;
-	pi->dpm_level_enable_mask.sclk_dpm_enable_mask =
-		ci_get_dpm_level_enable_mask_value(&dpm_table->sclk_table);
-
-	ret = amdgpu_ci_copy_bytes_to_smc(adev, level_array_address,
-				   (u8 *)levels, level_array_size,
-				   pi->sram_end);
-	if (ret)
-		return ret;
-
-	return 0;
-}
-
-static int ci_populate_ulv_state(struct amdgpu_device *adev,
-				 SMU7_Discrete_Ulv *ulv_level)
-{
-	return ci_populate_ulv_level(adev, ulv_level);
-}
-
-static int ci_populate_all_memory_levels(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct ci_dpm_table *dpm_table = &pi->dpm_table;
-	u32 level_array_address = pi->dpm_table_start +
-		offsetof(SMU7_Discrete_DpmTable, MemoryLevel);
-	u32 level_array_size = sizeof(SMU7_Discrete_MemoryLevel) *
-		SMU7_MAX_LEVELS_MEMORY;
-	SMU7_Discrete_MemoryLevel *levels = pi->smc_state_table.MemoryLevel;
-	u32 i, ret;
-
-	memset(levels, 0, level_array_size);
-
-	for (i = 0; i < dpm_table->mclk_table.count; i++) {
-		if (dpm_table->mclk_table.dpm_levels[i].value == 0)
-			return -EINVAL;
-		ret = ci_populate_single_memory_level(adev,
-						      dpm_table->mclk_table.dpm_levels[i].value,
-						      &pi->smc_state_table.MemoryLevel[i]);
-		if (ret)
-			return ret;
-	}
-
-	if ((dpm_table->mclk_table.count >= 2) &&
-	    ((adev->pdev->device == 0x67B0) || (adev->pdev->device == 0x67B1))) {
-		pi->smc_state_table.MemoryLevel[1].MinVddc =
-			pi->smc_state_table.MemoryLevel[0].MinVddc;
-		pi->smc_state_table.MemoryLevel[1].MinVddcPhases =
-			pi->smc_state_table.MemoryLevel[0].MinVddcPhases;
-	}
-
-	pi->smc_state_table.MemoryLevel[0].ActivityLevel = cpu_to_be16(0x1F);
-
-	pi->smc_state_table.MemoryDpmLevelCount = (u8)dpm_table->mclk_table.count;
-	pi->dpm_level_enable_mask.mclk_dpm_enable_mask =
-		ci_get_dpm_level_enable_mask_value(&dpm_table->mclk_table);
-
-	pi->smc_state_table.MemoryLevel[dpm_table->mclk_table.count - 1].DisplayWatermark =
-		PPSMC_DISPLAY_WATERMARK_HIGH;
-
-	ret = amdgpu_ci_copy_bytes_to_smc(adev, level_array_address,
-				   (u8 *)levels, level_array_size,
-				   pi->sram_end);
-	if (ret)
-		return ret;
-
-	return 0;
-}
-
-static void ci_reset_single_dpm_table(struct amdgpu_device *adev,
-				      struct ci_single_dpm_table* dpm_table,
-				      u32 count)
-{
-	u32 i;
-
-	dpm_table->count = count;
-	for (i = 0; i < MAX_REGULAR_DPM_NUMBER; i++)
-		dpm_table->dpm_levels[i].enabled = false;
-}
-
-static void ci_setup_pcie_table_entry(struct ci_single_dpm_table* dpm_table,
-				      u32 index, u32 pcie_gen, u32 pcie_lanes)
-{
-	dpm_table->dpm_levels[index].value = pcie_gen;
-	dpm_table->dpm_levels[index].param1 = pcie_lanes;
-	dpm_table->dpm_levels[index].enabled = true;
-}
-
-static int ci_setup_default_pcie_tables(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	if (!pi->use_pcie_performance_levels && !pi->use_pcie_powersaving_levels)
-		return -EINVAL;
-
-	if (pi->use_pcie_performance_levels && !pi->use_pcie_powersaving_levels) {
-		pi->pcie_gen_powersaving = pi->pcie_gen_performance;
-		pi->pcie_lane_powersaving = pi->pcie_lane_performance;
-	} else if (!pi->use_pcie_performance_levels && pi->use_pcie_powersaving_levels) {
-		pi->pcie_gen_performance = pi->pcie_gen_powersaving;
-		pi->pcie_lane_performance = pi->pcie_lane_powersaving;
-	}
-
-	ci_reset_single_dpm_table(adev,
-				  &pi->dpm_table.pcie_speed_table,
-				  SMU7_MAX_LEVELS_LINK);
-
-	if (adev->asic_type == CHIP_BONAIRE)
-		ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 0,
-					  pi->pcie_gen_powersaving.min,
-					  pi->pcie_lane_powersaving.max);
-	else
-		ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 0,
-					  pi->pcie_gen_powersaving.min,
-					  pi->pcie_lane_powersaving.min);
-	ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 1,
-				  pi->pcie_gen_performance.min,
-				  pi->pcie_lane_performance.min);
-	ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 2,
-				  pi->pcie_gen_powersaving.min,
-				  pi->pcie_lane_powersaving.max);
-	ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 3,
-				  pi->pcie_gen_performance.min,
-				  pi->pcie_lane_performance.max);
-	ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 4,
-				  pi->pcie_gen_powersaving.max,
-				  pi->pcie_lane_powersaving.max);
-	ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 5,
-				  pi->pcie_gen_performance.max,
-				  pi->pcie_lane_performance.max);
-
-	pi->dpm_table.pcie_speed_table.count = 6;
-
-	return 0;
-}
-
-static int ci_setup_default_dpm_tables(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct amdgpu_clock_voltage_dependency_table *allowed_sclk_vddc_table =
-		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
-	struct amdgpu_clock_voltage_dependency_table *allowed_mclk_table =
-		&adev->pm.dpm.dyn_state.vddc_dependency_on_mclk;
-	struct amdgpu_cac_leakage_table *std_voltage_table =
-		&adev->pm.dpm.dyn_state.cac_leakage_table;
-	u32 i;
-
-	if (allowed_sclk_vddc_table == NULL)
-		return -EINVAL;
-	if (allowed_sclk_vddc_table->count < 1)
-		return -EINVAL;
-	if (allowed_mclk_table == NULL)
-		return -EINVAL;
-	if (allowed_mclk_table->count < 1)
-		return -EINVAL;
-
-	memset(&pi->dpm_table, 0, sizeof(struct ci_dpm_table));
-
-	ci_reset_single_dpm_table(adev,
-				  &pi->dpm_table.sclk_table,
-				  SMU7_MAX_LEVELS_GRAPHICS);
-	ci_reset_single_dpm_table(adev,
-				  &pi->dpm_table.mclk_table,
-				  SMU7_MAX_LEVELS_MEMORY);
-	ci_reset_single_dpm_table(adev,
-				  &pi->dpm_table.vddc_table,
-				  SMU7_MAX_LEVELS_VDDC);
-	ci_reset_single_dpm_table(adev,
-				  &pi->dpm_table.vddci_table,
-				  SMU7_MAX_LEVELS_VDDCI);
-	ci_reset_single_dpm_table(adev,
-				  &pi->dpm_table.mvdd_table,
-				  SMU7_MAX_LEVELS_MVDD);
-
-	pi->dpm_table.sclk_table.count = 0;
-	for (i = 0; i < allowed_sclk_vddc_table->count; i++) {
-		if ((i == 0) ||
-		    (pi->dpm_table.sclk_table.dpm_levels[pi->dpm_table.sclk_table.count-1].value !=
-		     allowed_sclk_vddc_table->entries[i].clk)) {
-			pi->dpm_table.sclk_table.dpm_levels[pi->dpm_table.sclk_table.count].value =
-				allowed_sclk_vddc_table->entries[i].clk;
-			pi->dpm_table.sclk_table.dpm_levels[pi->dpm_table.sclk_table.count].enabled =
-				(i == 0) ? true : false;
-			pi->dpm_table.sclk_table.count++;
-		}
-	}
-
-	pi->dpm_table.mclk_table.count = 0;
-	for (i = 0; i < allowed_mclk_table->count; i++) {
-		if ((i == 0) ||
-		    (pi->dpm_table.mclk_table.dpm_levels[pi->dpm_table.mclk_table.count-1].value !=
-		     allowed_mclk_table->entries[i].clk)) {
-			pi->dpm_table.mclk_table.dpm_levels[pi->dpm_table.mclk_table.count].value =
-				allowed_mclk_table->entries[i].clk;
-			pi->dpm_table.mclk_table.dpm_levels[pi->dpm_table.mclk_table.count].enabled =
-				(i == 0) ? true : false;
-			pi->dpm_table.mclk_table.count++;
-		}
-	}
-
-	for (i = 0; i < allowed_sclk_vddc_table->count; i++) {
-		pi->dpm_table.vddc_table.dpm_levels[i].value =
-			allowed_sclk_vddc_table->entries[i].v;
-		pi->dpm_table.vddc_table.dpm_levels[i].param1 =
-			std_voltage_table->entries[i].leakage;
-		pi->dpm_table.vddc_table.dpm_levels[i].enabled = true;
-	}
-	pi->dpm_table.vddc_table.count = allowed_sclk_vddc_table->count;
-
-	allowed_mclk_table = &adev->pm.dpm.dyn_state.vddci_dependency_on_mclk;
-	if (allowed_mclk_table) {
-		for (i = 0; i < allowed_mclk_table->count; i++) {
-			pi->dpm_table.vddci_table.dpm_levels[i].value =
-				allowed_mclk_table->entries[i].v;
-			pi->dpm_table.vddci_table.dpm_levels[i].enabled = true;
-		}
-		pi->dpm_table.vddci_table.count = allowed_mclk_table->count;
-	}
-
-	allowed_mclk_table = &adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk;
-	if (allowed_mclk_table) {
-		for (i = 0; i < allowed_mclk_table->count; i++) {
-			pi->dpm_table.mvdd_table.dpm_levels[i].value =
-				allowed_mclk_table->entries[i].v;
-			pi->dpm_table.mvdd_table.dpm_levels[i].enabled = true;
-		}
-		pi->dpm_table.mvdd_table.count = allowed_mclk_table->count;
-	}
-
-	ci_setup_default_pcie_tables(adev);
-
-	/* save a copy of the default DPM table */
-	memcpy(&(pi->golden_dpm_table), &(pi->dpm_table),
-			sizeof(struct ci_dpm_table));
-
-	return 0;
-}
-
-static int ci_find_boot_level(struct ci_single_dpm_table *table,
-			      u32 value, u32 *boot_level)
-{
-	u32 i;
-	int ret = -EINVAL;
-
-	for(i = 0; i < table->count; i++) {
-		if (value == table->dpm_levels[i].value) {
-			*boot_level = i;
-			ret = 0;
-		}
-	}
-
-	return ret;
-}
-
-static void ci_save_default_power_profile(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct SMU7_Discrete_GraphicsLevel *levels =
-				pi->smc_state_table.GraphicsLevel;
-	uint32_t min_level = 0;
-
-	pi->default_gfx_power_profile.activity_threshold =
-			be16_to_cpu(levels[0].ActivityLevel);
-	pi->default_gfx_power_profile.up_hyst = levels[0].UpH;
-	pi->default_gfx_power_profile.down_hyst = levels[0].DownH;
-	pi->default_gfx_power_profile.type = AMD_PP_GFX_PROFILE;
-
-	pi->default_compute_power_profile = pi->default_gfx_power_profile;
-	pi->default_compute_power_profile.type = AMD_PP_COMPUTE_PROFILE;
-
-	/* Optimize compute power profile: Use only highest
-	 * 2 power levels (if more than 2 are available), Hysteresis:
-	 * 0ms up, 5ms down
-	 */
-	if (pi->smc_state_table.GraphicsDpmLevelCount > 2)
-		min_level = pi->smc_state_table.GraphicsDpmLevelCount - 2;
-	else if (pi->smc_state_table.GraphicsDpmLevelCount == 2)
-		min_level = 1;
-	pi->default_compute_power_profile.min_sclk =
-			be32_to_cpu(levels[min_level].SclkFrequency);
-
-	pi->default_compute_power_profile.up_hyst = 0;
-	pi->default_compute_power_profile.down_hyst = 5;
-
-	pi->gfx_power_profile = pi->default_gfx_power_profile;
-	pi->compute_power_profile = pi->default_compute_power_profile;
-}
-
-static int ci_init_smc_table(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct ci_ulv_parm *ulv = &pi->ulv;
-	struct amdgpu_ps *amdgpu_boot_state = adev->pm.dpm.boot_ps;
-	SMU7_Discrete_DpmTable *table = &pi->smc_state_table;
-	int ret;
-
-	ret = ci_setup_default_dpm_tables(adev);
-	if (ret)
-		return ret;
-
-	if (pi->voltage_control != CISLANDS_VOLTAGE_CONTROL_NONE)
-		ci_populate_smc_voltage_tables(adev, table);
-
-	ci_init_fps_limits(adev);
-
-	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_HARDWAREDC)
-		table->SystemFlags |= PPSMC_SYSTEMFLAG_GPIO_DC;
-
-	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_STEPVDDC)
-		table->SystemFlags |= PPSMC_SYSTEMFLAG_STEPVDDC;
-
-	if (adev->mc.vram_type == AMDGPU_VRAM_TYPE_GDDR5)
-		table->SystemFlags |= PPSMC_SYSTEMFLAG_GDDR5;
-
-	if (ulv->supported) {
-		ret = ci_populate_ulv_state(adev, &pi->smc_state_table.Ulv);
-		if (ret)
-			return ret;
-		WREG32_SMC(ixCG_ULV_PARAMETER, ulv->cg_ulv_parameter);
-	}
-
-	ret = ci_populate_all_graphic_levels(adev);
-	if (ret)
-		return ret;
-
-	ret = ci_populate_all_memory_levels(adev);
-	if (ret)
-		return ret;
-
-	ci_populate_smc_link_level(adev, table);
-
-	ret = ci_populate_smc_acpi_level(adev, table);
-	if (ret)
-		return ret;
-
-	ret = ci_populate_smc_vce_level(adev, table);
-	if (ret)
-		return ret;
-
-	ret = ci_populate_smc_acp_level(adev, table);
-	if (ret)
-		return ret;
-
-	ret = ci_populate_smc_samu_level(adev, table);
-	if (ret)
-		return ret;
-
-	ret = ci_do_program_memory_timing_parameters(adev);
-	if (ret)
-		return ret;
-
-	ret = ci_populate_smc_uvd_level(adev, table);
-	if (ret)
-		return ret;
-
-	table->UvdBootLevel  = 0;
-	table->VceBootLevel  = 0;
-	table->AcpBootLevel  = 0;
-	table->SamuBootLevel  = 0;
-	table->GraphicsBootLevel  = 0;
-	table->MemoryBootLevel  = 0;
-
-	ret = ci_find_boot_level(&pi->dpm_table.sclk_table,
-				 pi->vbios_boot_state.sclk_bootup_value,
-				 (u32 *)&pi->smc_state_table.GraphicsBootLevel);
-
-	ret = ci_find_boot_level(&pi->dpm_table.mclk_table,
-				 pi->vbios_boot_state.mclk_bootup_value,
-				 (u32 *)&pi->smc_state_table.MemoryBootLevel);
-
-	table->BootVddc = pi->vbios_boot_state.vddc_bootup_value;
-	table->BootVddci = pi->vbios_boot_state.vddci_bootup_value;
-	table->BootMVdd = pi->vbios_boot_state.mvdd_bootup_value;
-
-	ci_populate_smc_initial_state(adev, amdgpu_boot_state);
-
-	ret = ci_populate_bapm_parameters_in_dpm_table(adev);
-	if (ret)
-		return ret;
-
-	table->UVDInterval = 1;
-	table->VCEInterval = 1;
-	table->ACPInterval = 1;
-	table->SAMUInterval = 1;
-	table->GraphicsVoltageChangeEnable = 1;
-	table->GraphicsThermThrottleEnable = 1;
-	table->GraphicsInterval = 1;
-	table->VoltageInterval = 1;
-	table->ThermalInterval = 1;
-	table->TemperatureLimitHigh = (u16)((pi->thermal_temp_setting.temperature_high *
-					     CISLANDS_Q88_FORMAT_CONVERSION_UNIT) / 1000);
-	table->TemperatureLimitLow = (u16)((pi->thermal_temp_setting.temperature_low *
-					    CISLANDS_Q88_FORMAT_CONVERSION_UNIT) / 1000);
-	table->MemoryVoltageChangeEnable = 1;
-	table->MemoryInterval = 1;
-	table->VoltageResponseTime = 0;
-	table->VddcVddciDelta = 4000;
-	table->PhaseResponseTime = 0;
-	table->MemoryThermThrottleEnable = 1;
-	table->PCIeBootLinkLevel = pi->dpm_table.pcie_speed_table.count - 1;
-	table->PCIeGenInterval = 1;
-	if (pi->voltage_control == CISLANDS_VOLTAGE_CONTROL_BY_SVID2)
-		table->SVI2Enable  = 1;
-	else
-		table->SVI2Enable  = 0;
-
-	table->ThermGpio = 17;
-	table->SclkStepSize = 0x4000;
-
-	table->SystemFlags = cpu_to_be32(table->SystemFlags);
-	table->SmioMaskVddcVid = cpu_to_be32(table->SmioMaskVddcVid);
-	table->SmioMaskVddcPhase = cpu_to_be32(table->SmioMaskVddcPhase);
-	table->SmioMaskVddciVid = cpu_to_be32(table->SmioMaskVddciVid);
-	table->SmioMaskMvddVid = cpu_to_be32(table->SmioMaskMvddVid);
-	table->SclkStepSize = cpu_to_be32(table->SclkStepSize);
-	table->TemperatureLimitHigh = cpu_to_be16(table->TemperatureLimitHigh);
-	table->TemperatureLimitLow = cpu_to_be16(table->TemperatureLimitLow);
-	table->VddcVddciDelta = cpu_to_be16(table->VddcVddciDelta);
-	table->VoltageResponseTime = cpu_to_be16(table->VoltageResponseTime);
-	table->PhaseResponseTime = cpu_to_be16(table->PhaseResponseTime);
-	table->BootVddc = cpu_to_be16(table->BootVddc * VOLTAGE_SCALE);
-	table->BootVddci = cpu_to_be16(table->BootVddci * VOLTAGE_SCALE);
-	table->BootMVdd = cpu_to_be16(table->BootMVdd * VOLTAGE_SCALE);
-
-	ret = amdgpu_ci_copy_bytes_to_smc(adev,
-				   pi->dpm_table_start +
-				   offsetof(SMU7_Discrete_DpmTable, SystemFlags),
-				   (u8 *)&table->SystemFlags,
-				   sizeof(SMU7_Discrete_DpmTable) - 3 * sizeof(SMU7_PIDController),
-				   pi->sram_end);
-	if (ret)
-		return ret;
-
-	ci_save_default_power_profile(adev);
-
-	return 0;
-}
-
-static void ci_trim_single_dpm_states(struct amdgpu_device *adev,
-				      struct ci_single_dpm_table *dpm_table,
-				      u32 low_limit, u32 high_limit)
-{
-	u32 i;
-
-	for (i = 0; i < dpm_table->count; i++) {
-		if ((dpm_table->dpm_levels[i].value < low_limit) ||
-		    (dpm_table->dpm_levels[i].value > high_limit))
-			dpm_table->dpm_levels[i].enabled = false;
-		else
-			dpm_table->dpm_levels[i].enabled = true;
-	}
-}
-
-static void ci_trim_pcie_dpm_states(struct amdgpu_device *adev,
-				    u32 speed_low, u32 lanes_low,
-				    u32 speed_high, u32 lanes_high)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct ci_single_dpm_table *pcie_table = &pi->dpm_table.pcie_speed_table;
-	u32 i, j;
-
-	for (i = 0; i < pcie_table->count; i++) {
-		if ((pcie_table->dpm_levels[i].value < speed_low) ||
-		    (pcie_table->dpm_levels[i].param1 < lanes_low) ||
-		    (pcie_table->dpm_levels[i].value > speed_high) ||
-		    (pcie_table->dpm_levels[i].param1 > lanes_high))
-			pcie_table->dpm_levels[i].enabled = false;
-		else
-			pcie_table->dpm_levels[i].enabled = true;
-	}
-
-	for (i = 0; i < pcie_table->count; i++) {
-		if (pcie_table->dpm_levels[i].enabled) {
-			for (j = i + 1; j < pcie_table->count; j++) {
-				if (pcie_table->dpm_levels[j].enabled) {
-					if ((pcie_table->dpm_levels[i].value == pcie_table->dpm_levels[j].value) &&
-					    (pcie_table->dpm_levels[i].param1 == pcie_table->dpm_levels[j].param1))
-						pcie_table->dpm_levels[j].enabled = false;
-				}
-			}
-		}
-	}
-}
-
-static int ci_trim_dpm_states(struct amdgpu_device *adev,
-			      struct amdgpu_ps *amdgpu_state)
-{
-	struct ci_ps *state = ci_get_ps(amdgpu_state);
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u32 high_limit_count;
-
-	if (state->performance_level_count < 1)
-		return -EINVAL;
-
-	if (state->performance_level_count == 1)
-		high_limit_count = 0;
-	else
-		high_limit_count = 1;
-
-	ci_trim_single_dpm_states(adev,
-				  &pi->dpm_table.sclk_table,
-				  state->performance_levels[0].sclk,
-				  state->performance_levels[high_limit_count].sclk);
-
-	ci_trim_single_dpm_states(adev,
-				  &pi->dpm_table.mclk_table,
-				  state->performance_levels[0].mclk,
-				  state->performance_levels[high_limit_count].mclk);
-
-	ci_trim_pcie_dpm_states(adev,
-				state->performance_levels[0].pcie_gen,
-				state->performance_levels[0].pcie_lane,
-				state->performance_levels[high_limit_count].pcie_gen,
-				state->performance_levels[high_limit_count].pcie_lane);
-
-	return 0;
-}
-
-static int ci_apply_disp_minimum_voltage_request(struct amdgpu_device *adev)
-{
-	struct amdgpu_clock_voltage_dependency_table *disp_voltage_table =
-		&adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk;
-	struct amdgpu_clock_voltage_dependency_table *vddc_table =
-		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
-	u32 requested_voltage = 0;
-	u32 i;
-
-	if (disp_voltage_table == NULL)
-		return -EINVAL;
-	if (!disp_voltage_table->count)
-		return -EINVAL;
-
-	for (i = 0; i < disp_voltage_table->count; i++) {
-		if (adev->clock.current_dispclk == disp_voltage_table->entries[i].clk)
-			requested_voltage = disp_voltage_table->entries[i].v;
-	}
-
-	for (i = 0; i < vddc_table->count; i++) {
-		if (requested_voltage <= vddc_table->entries[i].v) {
-			requested_voltage = vddc_table->entries[i].v;
-			return (amdgpu_ci_send_msg_to_smc_with_parameter(adev,
-								  PPSMC_MSG_VddC_Request,
-								  requested_voltage * VOLTAGE_SCALE) == PPSMC_Result_OK) ?
-				0 : -EINVAL;
-		}
-	}
-
-	return -EINVAL;
-}
-
-static int ci_upload_dpm_level_enable_mask(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	PPSMC_Result result;
-
-	ci_apply_disp_minimum_voltage_request(adev);
-
-	if (!pi->sclk_dpm_key_disabled) {
-		if (pi->dpm_level_enable_mask.sclk_dpm_enable_mask) {
-			result = amdgpu_ci_send_msg_to_smc_with_parameter(adev,
-								   PPSMC_MSG_SCLKDPM_SetEnabledMask,
-								   pi->dpm_level_enable_mask.sclk_dpm_enable_mask);
-			if (result != PPSMC_Result_OK)
-				return -EINVAL;
-		}
-	}
-
-	if (!pi->mclk_dpm_key_disabled) {
-		if (pi->dpm_level_enable_mask.mclk_dpm_enable_mask) {
-			result = amdgpu_ci_send_msg_to_smc_with_parameter(adev,
-								   PPSMC_MSG_MCLKDPM_SetEnabledMask,
-								   pi->dpm_level_enable_mask.mclk_dpm_enable_mask);
-			if (result != PPSMC_Result_OK)
-				return -EINVAL;
-		}
-	}
-
-#if 0
-	if (!pi->pcie_dpm_key_disabled) {
-		if (pi->dpm_level_enable_mask.pcie_dpm_enable_mask) {
-			result = amdgpu_ci_send_msg_to_smc_with_parameter(adev,
-								   PPSMC_MSG_PCIeDPM_SetEnabledMask,
-								   pi->dpm_level_enable_mask.pcie_dpm_enable_mask);
-			if (result != PPSMC_Result_OK)
-				return -EINVAL;
-		}
-	}
-#endif
-
-	return 0;
-}
-
-static void ci_find_dpm_states_clocks_in_dpm_table(struct amdgpu_device *adev,
-						   struct amdgpu_ps *amdgpu_state)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct ci_ps *state = ci_get_ps(amdgpu_state);
-	struct ci_single_dpm_table *sclk_table = &pi->dpm_table.sclk_table;
-	u32 sclk = state->performance_levels[state->performance_level_count-1].sclk;
-	struct ci_single_dpm_table *mclk_table = &pi->dpm_table.mclk_table;
-	u32 mclk = state->performance_levels[state->performance_level_count-1].mclk;
-	u32 i;
-
-	pi->need_update_smu7_dpm_table = 0;
-
-	for (i = 0; i < sclk_table->count; i++) {
-		if (sclk == sclk_table->dpm_levels[i].value)
-			break;
-	}
-
-	if (i >= sclk_table->count) {
-		pi->need_update_smu7_dpm_table |= DPMTABLE_OD_UPDATE_SCLK;
-	} else {
-		/* XXX check display min clock requirements */
-		if (CISLAND_MINIMUM_ENGINE_CLOCK != CISLAND_MINIMUM_ENGINE_CLOCK)
-			pi->need_update_smu7_dpm_table |= DPMTABLE_UPDATE_SCLK;
-	}
-
-	for (i = 0; i < mclk_table->count; i++) {
-		if (mclk == mclk_table->dpm_levels[i].value)
-			break;
-	}
-
-	if (i >= mclk_table->count)
-		pi->need_update_smu7_dpm_table |= DPMTABLE_OD_UPDATE_MCLK;
-
-	if (adev->pm.dpm.current_active_crtc_count !=
-	    adev->pm.dpm.new_active_crtc_count)
-		pi->need_update_smu7_dpm_table |= DPMTABLE_UPDATE_MCLK;
-}
-
-static int ci_populate_and_upload_sclk_mclk_dpm_levels(struct amdgpu_device *adev,
-						       struct amdgpu_ps *amdgpu_state)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct ci_ps *state = ci_get_ps(amdgpu_state);
-	u32 sclk = state->performance_levels[state->performance_level_count-1].sclk;
-	u32 mclk = state->performance_levels[state->performance_level_count-1].mclk;
-	struct ci_dpm_table *dpm_table = &pi->dpm_table;
-	int ret;
-
-	if (!pi->need_update_smu7_dpm_table)
-		return 0;
-
-	if (pi->need_update_smu7_dpm_table & DPMTABLE_OD_UPDATE_SCLK)
-		dpm_table->sclk_table.dpm_levels[dpm_table->sclk_table.count-1].value = sclk;
-
-	if (pi->need_update_smu7_dpm_table & DPMTABLE_OD_UPDATE_MCLK)
-		dpm_table->mclk_table.dpm_levels[dpm_table->mclk_table.count-1].value = mclk;
-
-	if (pi->need_update_smu7_dpm_table & (DPMTABLE_OD_UPDATE_SCLK | DPMTABLE_UPDATE_SCLK)) {
-		ret = ci_populate_all_graphic_levels(adev);
-		if (ret)
-			return ret;
-	}
-
-	if (pi->need_update_smu7_dpm_table & (DPMTABLE_OD_UPDATE_MCLK | DPMTABLE_UPDATE_MCLK)) {
-		ret = ci_populate_all_memory_levels(adev);
-		if (ret)
-			return ret;
-	}
-
-	return 0;
-}
-
-static int ci_enable_uvd_dpm(struct amdgpu_device *adev, bool enable)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	const struct amdgpu_clock_and_voltage_limits *max_limits;
-	int i;
-
-	if (adev->pm.dpm.ac_power)
-		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
-	else
-		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_dc;
-
-	if (enable) {
-		pi->dpm_level_enable_mask.uvd_dpm_enable_mask = 0;
-
-		for (i = adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.count - 1; i >= 0; i--) {
-			if (adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries[i].v <= max_limits->vddc) {
-				pi->dpm_level_enable_mask.uvd_dpm_enable_mask |= 1 << i;
-
-				if (!pi->caps_uvd_dpm)
-					break;
-			}
-		}
-
-		amdgpu_ci_send_msg_to_smc_with_parameter(adev,
-						  PPSMC_MSG_UVDDPM_SetEnabledMask,
-						  pi->dpm_level_enable_mask.uvd_dpm_enable_mask);
-
-		if (pi->last_mclk_dpm_enable_mask & 0x1) {
-			pi->uvd_enabled = true;
-			pi->dpm_level_enable_mask.mclk_dpm_enable_mask &= 0xFFFFFFFE;
-			amdgpu_ci_send_msg_to_smc_with_parameter(adev,
-							  PPSMC_MSG_MCLKDPM_SetEnabledMask,
-							  pi->dpm_level_enable_mask.mclk_dpm_enable_mask);
-		}
-	} else {
-		if (pi->uvd_enabled) {
-			pi->uvd_enabled = false;
-			pi->dpm_level_enable_mask.mclk_dpm_enable_mask |= 1;
-			amdgpu_ci_send_msg_to_smc_with_parameter(adev,
-							  PPSMC_MSG_MCLKDPM_SetEnabledMask,
-							  pi->dpm_level_enable_mask.mclk_dpm_enable_mask);
-		}
-	}
-
-	return (amdgpu_ci_send_msg_to_smc(adev, enable ?
-				   PPSMC_MSG_UVDDPM_Enable : PPSMC_MSG_UVDDPM_Disable) == PPSMC_Result_OK) ?
-		0 : -EINVAL;
-}
-
-static int ci_enable_vce_dpm(struct amdgpu_device *adev, bool enable)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	const struct amdgpu_clock_and_voltage_limits *max_limits;
-	int i;
-
-	if (adev->pm.dpm.ac_power)
-		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
-	else
-		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_dc;
-
-	if (enable) {
-		pi->dpm_level_enable_mask.vce_dpm_enable_mask = 0;
-		for (i = adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.count - 1; i >= 0; i--) {
-			if (adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries[i].v <= max_limits->vddc) {
-				pi->dpm_level_enable_mask.vce_dpm_enable_mask |= 1 << i;
-
-				if (!pi->caps_vce_dpm)
-					break;
-			}
-		}
-
-		amdgpu_ci_send_msg_to_smc_with_parameter(adev,
-						  PPSMC_MSG_VCEDPM_SetEnabledMask,
-						  pi->dpm_level_enable_mask.vce_dpm_enable_mask);
-	}
-
-	return (amdgpu_ci_send_msg_to_smc(adev, enable ?
-				   PPSMC_MSG_VCEDPM_Enable : PPSMC_MSG_VCEDPM_Disable) == PPSMC_Result_OK) ?
-		0 : -EINVAL;
-}
-
-#if 0
-static int ci_enable_samu_dpm(struct amdgpu_device *adev, bool enable)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	const struct amdgpu_clock_and_voltage_limits *max_limits;
-	int i;
-
-	if (adev->pm.dpm.ac_power)
-		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
-	else
-		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_dc;
-
-	if (enable) {
-		pi->dpm_level_enable_mask.samu_dpm_enable_mask = 0;
-		for (i = adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.count - 1; i >= 0; i--) {
-			if (adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.entries[i].v <= max_limits->vddc) {
-				pi->dpm_level_enable_mask.samu_dpm_enable_mask |= 1 << i;
-
-				if (!pi->caps_samu_dpm)
-					break;
-			}
-		}
-
-		amdgpu_ci_send_msg_to_smc_with_parameter(adev,
-						  PPSMC_MSG_SAMUDPM_SetEnabledMask,
-						  pi->dpm_level_enable_mask.samu_dpm_enable_mask);
-	}
-	return (amdgpu_ci_send_msg_to_smc(adev, enable ?
-				   PPSMC_MSG_SAMUDPM_Enable : PPSMC_MSG_SAMUDPM_Disable) == PPSMC_Result_OK) ?
-		0 : -EINVAL;
-}
-
-static int ci_enable_acp_dpm(struct amdgpu_device *adev, bool enable)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	const struct amdgpu_clock_and_voltage_limits *max_limits;
-	int i;
-
-	if (adev->pm.dpm.ac_power)
-		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
-	else
-		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_dc;
-
-	if (enable) {
-		pi->dpm_level_enable_mask.acp_dpm_enable_mask = 0;
-		for (i = adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.count - 1; i >= 0; i--) {
-			if (adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.entries[i].v <= max_limits->vddc) {
-				pi->dpm_level_enable_mask.acp_dpm_enable_mask |= 1 << i;
-
-				if (!pi->caps_acp_dpm)
-					break;
-			}
-		}
-
-		amdgpu_ci_send_msg_to_smc_with_parameter(adev,
-						  PPSMC_MSG_ACPDPM_SetEnabledMask,
-						  pi->dpm_level_enable_mask.acp_dpm_enable_mask);
-	}
-
-	return (amdgpu_ci_send_msg_to_smc(adev, enable ?
-				   PPSMC_MSG_ACPDPM_Enable : PPSMC_MSG_ACPDPM_Disable) == PPSMC_Result_OK) ?
-		0 : -EINVAL;
-}
-#endif
-
-static int ci_update_uvd_dpm(struct amdgpu_device *adev, bool gate)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u32 tmp;
-	int ret = 0;
-
-	if (!gate) {
-		/* turn the clocks on when decoding */
-		if (pi->caps_uvd_dpm ||
-		    (adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.count <= 0))
-			pi->smc_state_table.UvdBootLevel = 0;
-		else
-			pi->smc_state_table.UvdBootLevel =
-				adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.count - 1;
-
-		tmp = RREG32_SMC(ixDPM_TABLE_475);
-		tmp &= ~DPM_TABLE_475__UvdBootLevel_MASK;
-		tmp |= (pi->smc_state_table.UvdBootLevel << DPM_TABLE_475__UvdBootLevel__SHIFT);
-		WREG32_SMC(ixDPM_TABLE_475, tmp);
-		ret = ci_enable_uvd_dpm(adev, true);
-	} else {
-		ret = ci_enable_uvd_dpm(adev, false);
-		if (ret)
-			return ret;
-	}
-
-	return ret;
-}
-
-static u8 ci_get_vce_boot_level(struct amdgpu_device *adev)
-{
-	u8 i;
-	u32 min_evclk = 30000; /* ??? */
-	struct amdgpu_vce_clock_voltage_dependency_table *table =
-		&adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table;
-
-	for (i = 0; i < table->count; i++) {
-		if (table->entries[i].evclk >= min_evclk)
-			return i;
-	}
-
-	return table->count - 1;
-}
-
-static int ci_update_vce_dpm(struct amdgpu_device *adev,
-			     struct amdgpu_ps *amdgpu_new_state,
-			     struct amdgpu_ps *amdgpu_current_state)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	int ret = 0;
-	u32 tmp;
-
-	if (amdgpu_current_state->evclk != amdgpu_new_state->evclk) {
-		if (amdgpu_new_state->evclk) {
-			pi->smc_state_table.VceBootLevel = ci_get_vce_boot_level(adev);
-			tmp = RREG32_SMC(ixDPM_TABLE_475);
-			tmp &= ~DPM_TABLE_475__VceBootLevel_MASK;
-			tmp |= (pi->smc_state_table.VceBootLevel << DPM_TABLE_475__VceBootLevel__SHIFT);
-			WREG32_SMC(ixDPM_TABLE_475, tmp);
-
-			ret = ci_enable_vce_dpm(adev, true);
-		} else {
-			ret = ci_enable_vce_dpm(adev, false);
-			if (ret)
-				return ret;
-		}
-	}
-	return ret;
-}
-
-#if 0
-static int ci_update_samu_dpm(struct amdgpu_device *adev, bool gate)
-{
-	return ci_enable_samu_dpm(adev, gate);
-}
-
-static int ci_update_acp_dpm(struct amdgpu_device *adev, bool gate)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u32 tmp;
-
-	if (!gate) {
-		pi->smc_state_table.AcpBootLevel = 0;
-
-		tmp = RREG32_SMC(ixDPM_TABLE_475);
-		tmp &= ~AcpBootLevel_MASK;
-		tmp |= AcpBootLevel(pi->smc_state_table.AcpBootLevel);
-		WREG32_SMC(ixDPM_TABLE_475, tmp);
-	}
-
-	return ci_enable_acp_dpm(adev, !gate);
-}
-#endif
-
-static int ci_generate_dpm_level_enable_mask(struct amdgpu_device *adev,
-					     struct amdgpu_ps *amdgpu_state)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	int ret;
-
-	ret = ci_trim_dpm_states(adev, amdgpu_state);
-	if (ret)
-		return ret;
-
-	pi->dpm_level_enable_mask.sclk_dpm_enable_mask =
-		ci_get_dpm_level_enable_mask_value(&pi->dpm_table.sclk_table);
-	pi->dpm_level_enable_mask.mclk_dpm_enable_mask =
-		ci_get_dpm_level_enable_mask_value(&pi->dpm_table.mclk_table);
-	pi->last_mclk_dpm_enable_mask =
-		pi->dpm_level_enable_mask.mclk_dpm_enable_mask;
-	if (pi->uvd_enabled) {
-		if (pi->dpm_level_enable_mask.mclk_dpm_enable_mask & 1)
-			pi->dpm_level_enable_mask.mclk_dpm_enable_mask &= 0xFFFFFFFE;
-	}
-	pi->dpm_level_enable_mask.pcie_dpm_enable_mask =
-		ci_get_dpm_level_enable_mask_value(&pi->dpm_table.pcie_speed_table);
-
-	return 0;
-}
-
-static u32 ci_get_lowest_enabled_level(struct amdgpu_device *adev,
-				       u32 level_mask)
-{
-	u32 level = 0;
-
-	while ((level_mask & (1 << level)) == 0)
-		level++;
-
-	return level;
-}
-
-
-static int ci_dpm_force_performance_level(struct amdgpu_device *adev,
-					  enum amd_dpm_forced_level level)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u32 tmp, levels, i;
-	int ret;
-
-	if (level == AMD_DPM_FORCED_LEVEL_HIGH) {
-		if ((!pi->pcie_dpm_key_disabled) &&
-		    pi->dpm_level_enable_mask.pcie_dpm_enable_mask) {
-			levels = 0;
-			tmp = pi->dpm_level_enable_mask.pcie_dpm_enable_mask;
-			while (tmp >>= 1)
-				levels++;
-			if (levels) {
-				ret = ci_dpm_force_state_pcie(adev, level);
-				if (ret)
-					return ret;
-				for (i = 0; i < adev->usec_timeout; i++) {
-					tmp = (RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX_1) &
-					TARGET_AND_CURRENT_PROFILE_INDEX_1__CURR_PCIE_INDEX_MASK) >>
-					TARGET_AND_CURRENT_PROFILE_INDEX_1__CURR_PCIE_INDEX__SHIFT;
-					if (tmp == levels)
-						break;
-					udelay(1);
-				}
-			}
-		}
-		if ((!pi->sclk_dpm_key_disabled) &&
-		    pi->dpm_level_enable_mask.sclk_dpm_enable_mask) {
-			levels = 0;
-			tmp = pi->dpm_level_enable_mask.sclk_dpm_enable_mask;
-			while (tmp >>= 1)
-				levels++;
-			if (levels) {
-				ret = ci_dpm_force_state_sclk(adev, levels);
-				if (ret)
-					return ret;
-				for (i = 0; i < adev->usec_timeout; i++) {
-					tmp = (RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX) &
-					TARGET_AND_CURRENT_PROFILE_INDEX__CURR_SCLK_INDEX_MASK) >>
-					TARGET_AND_CURRENT_PROFILE_INDEX__CURR_SCLK_INDEX__SHIFT;
-					if (tmp == levels)
-						break;
-					udelay(1);
-				}
-			}
-		}
-		if ((!pi->mclk_dpm_key_disabled) &&
-		    pi->dpm_level_enable_mask.mclk_dpm_enable_mask) {
-			levels = 0;
-			tmp = pi->dpm_level_enable_mask.mclk_dpm_enable_mask;
-			while (tmp >>= 1)
-				levels++;
-			if (levels) {
-				ret = ci_dpm_force_state_mclk(adev, levels);
-				if (ret)
-					return ret;
-				for (i = 0; i < adev->usec_timeout; i++) {
-					tmp = (RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX) &
-					TARGET_AND_CURRENT_PROFILE_INDEX__CURR_MCLK_INDEX_MASK) >>
-					TARGET_AND_CURRENT_PROFILE_INDEX__CURR_MCLK_INDEX__SHIFT;
-					if (tmp == levels)
-						break;
-					udelay(1);
-				}
-			}
-		}
-	} else if (level == AMD_DPM_FORCED_LEVEL_LOW) {
-		if ((!pi->sclk_dpm_key_disabled) &&
-		    pi->dpm_level_enable_mask.sclk_dpm_enable_mask) {
-			levels = ci_get_lowest_enabled_level(adev,
-							     pi->dpm_level_enable_mask.sclk_dpm_enable_mask);
-			ret = ci_dpm_force_state_sclk(adev, levels);
-			if (ret)
-				return ret;
-			for (i = 0; i < adev->usec_timeout; i++) {
-				tmp = (RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX) &
-				TARGET_AND_CURRENT_PROFILE_INDEX__CURR_SCLK_INDEX_MASK) >>
-				TARGET_AND_CURRENT_PROFILE_INDEX__CURR_SCLK_INDEX__SHIFT;
-				if (tmp == levels)
-					break;
-				udelay(1);
-			}
-		}
-		if ((!pi->mclk_dpm_key_disabled) &&
-		    pi->dpm_level_enable_mask.mclk_dpm_enable_mask) {
-			levels = ci_get_lowest_enabled_level(adev,
-							     pi->dpm_level_enable_mask.mclk_dpm_enable_mask);
-			ret = ci_dpm_force_state_mclk(adev, levels);
-			if (ret)
-				return ret;
-			for (i = 0; i < adev->usec_timeout; i++) {
-				tmp = (RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX) &
-				TARGET_AND_CURRENT_PROFILE_INDEX__CURR_MCLK_INDEX_MASK) >>
-				TARGET_AND_CURRENT_PROFILE_INDEX__CURR_MCLK_INDEX__SHIFT;
-				if (tmp == levels)
-					break;
-				udelay(1);
-			}
-		}
-		if ((!pi->pcie_dpm_key_disabled) &&
-		    pi->dpm_level_enable_mask.pcie_dpm_enable_mask) {
-			levels = ci_get_lowest_enabled_level(adev,
-							     pi->dpm_level_enable_mask.pcie_dpm_enable_mask);
-			ret = ci_dpm_force_state_pcie(adev, levels);
-			if (ret)
-				return ret;
-			for (i = 0; i < adev->usec_timeout; i++) {
-				tmp = (RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX_1) &
-				TARGET_AND_CURRENT_PROFILE_INDEX_1__CURR_PCIE_INDEX_MASK) >>
-				TARGET_AND_CURRENT_PROFILE_INDEX_1__CURR_PCIE_INDEX__SHIFT;
-				if (tmp == levels)
-					break;
-				udelay(1);
-			}
-		}
-	} else if (level == AMD_DPM_FORCED_LEVEL_AUTO) {
-		if (!pi->pcie_dpm_key_disabled) {
-			PPSMC_Result smc_result;
-
-			smc_result = amdgpu_ci_send_msg_to_smc(adev,
-							       PPSMC_MSG_PCIeDPM_UnForceLevel);
-			if (smc_result != PPSMC_Result_OK)
-				return -EINVAL;
-		}
-		ret = ci_upload_dpm_level_enable_mask(adev);
-		if (ret)
-			return ret;
-	}
-
-	adev->pm.dpm.forced_level = level;
-
-	return 0;
-}
-
-static int ci_set_mc_special_registers(struct amdgpu_device *adev,
-				       struct ci_mc_reg_table *table)
-{
-	u8 i, j, k;
-	u32 temp_reg;
-
-	for (i = 0, j = table->last; i < table->last; i++) {
-		if (j >= SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE)
-			return -EINVAL;
-		switch(table->mc_reg_address[i].s1) {
-		case mmMC_SEQ_MISC1:
-			temp_reg = RREG32(mmMC_PMG_CMD_EMRS);
-			table->mc_reg_address[j].s1 = mmMC_PMG_CMD_EMRS;
-			table->mc_reg_address[j].s0 = mmMC_SEQ_PMG_CMD_EMRS_LP;
-			for (k = 0; k < table->num_entries; k++) {
-				table->mc_reg_table_entry[k].mc_data[j] =
-					((temp_reg & 0xffff0000)) | ((table->mc_reg_table_entry[k].mc_data[i] & 0xffff0000) >> 16);
-			}
-			j++;
-			if (j >= SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE)
-				return -EINVAL;
-
-			temp_reg = RREG32(mmMC_PMG_CMD_MRS);
-			table->mc_reg_address[j].s1 = mmMC_PMG_CMD_MRS;
-			table->mc_reg_address[j].s0 = mmMC_SEQ_PMG_CMD_MRS_LP;
-			for (k = 0; k < table->num_entries; k++) {
-				table->mc_reg_table_entry[k].mc_data[j] =
-					(temp_reg & 0xffff0000) | (table->mc_reg_table_entry[k].mc_data[i] & 0x0000ffff);
-				if (adev->mc.vram_type != AMDGPU_VRAM_TYPE_GDDR5)
-					table->mc_reg_table_entry[k].mc_data[j] |= 0x100;
-			}
-			j++;
-			if (j > SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE)
-				return -EINVAL;
-
-			if (adev->mc.vram_type != AMDGPU_VRAM_TYPE_GDDR5) {
-				table->mc_reg_address[j].s1 = mmMC_PMG_AUTO_CMD;
-				table->mc_reg_address[j].s0 = mmMC_PMG_AUTO_CMD;
-				for (k = 0; k < table->num_entries; k++) {
-					table->mc_reg_table_entry[k].mc_data[j] =
-						(table->mc_reg_table_entry[k].mc_data[i] & 0xffff0000) >> 16;
-				}
-				j++;
-				if (j > SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE)
-					return -EINVAL;
-			}
-			break;
-		case mmMC_SEQ_RESERVE_M:
-			temp_reg = RREG32(mmMC_PMG_CMD_MRS1);
-			table->mc_reg_address[j].s1 = mmMC_PMG_CMD_MRS1;
-			table->mc_reg_address[j].s0 = mmMC_SEQ_PMG_CMD_MRS1_LP;
-			for (k = 0; k < table->num_entries; k++) {
-				table->mc_reg_table_entry[k].mc_data[j] =
-					(temp_reg & 0xffff0000) | (table->mc_reg_table_entry[k].mc_data[i] & 0x0000ffff);
-			}
-			j++;
-			if (j > SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE)
-				return -EINVAL;
-			break;
-		default:
-			break;
-		}
-
-	}
-
-	table->last = j;
-
-	return 0;
-}
-
-static bool ci_check_s0_mc_reg_index(u16 in_reg, u16 *out_reg)
-{
-	bool result = true;
-
-	switch(in_reg) {
-	case mmMC_SEQ_RAS_TIMING:
-		*out_reg = mmMC_SEQ_RAS_TIMING_LP;
-		break;
-	case mmMC_SEQ_DLL_STBY:
-		*out_reg = mmMC_SEQ_DLL_STBY_LP;
-		break;
-	case mmMC_SEQ_G5PDX_CMD0:
-		*out_reg = mmMC_SEQ_G5PDX_CMD0_LP;
-		break;
-	case mmMC_SEQ_G5PDX_CMD1:
-		*out_reg = mmMC_SEQ_G5PDX_CMD1_LP;
-		break;
-	case mmMC_SEQ_G5PDX_CTRL:
-		*out_reg = mmMC_SEQ_G5PDX_CTRL_LP;
-		break;
-	case mmMC_SEQ_CAS_TIMING:
-		*out_reg = mmMC_SEQ_CAS_TIMING_LP;
-	    break;
-	case mmMC_SEQ_MISC_TIMING:
-		*out_reg = mmMC_SEQ_MISC_TIMING_LP;
-		break;
-	case mmMC_SEQ_MISC_TIMING2:
-		*out_reg = mmMC_SEQ_MISC_TIMING2_LP;
-		break;
-	case mmMC_SEQ_PMG_DVS_CMD:
-		*out_reg = mmMC_SEQ_PMG_DVS_CMD_LP;
-		break;
-	case mmMC_SEQ_PMG_DVS_CTL:
-		*out_reg = mmMC_SEQ_PMG_DVS_CTL_LP;
-		break;
-	case mmMC_SEQ_RD_CTL_D0:
-		*out_reg = mmMC_SEQ_RD_CTL_D0_LP;
-		break;
-	case mmMC_SEQ_RD_CTL_D1:
-		*out_reg = mmMC_SEQ_RD_CTL_D1_LP;
-		break;
-	case mmMC_SEQ_WR_CTL_D0:
-		*out_reg = mmMC_SEQ_WR_CTL_D0_LP;
-		break;
-	case mmMC_SEQ_WR_CTL_D1:
-		*out_reg = mmMC_SEQ_WR_CTL_D1_LP;
-		break;
-	case mmMC_PMG_CMD_EMRS:
-		*out_reg = mmMC_SEQ_PMG_CMD_EMRS_LP;
-		break;
-	case mmMC_PMG_CMD_MRS:
-		*out_reg = mmMC_SEQ_PMG_CMD_MRS_LP;
-		break;
-	case mmMC_PMG_CMD_MRS1:
-		*out_reg = mmMC_SEQ_PMG_CMD_MRS1_LP;
-		break;
-	case mmMC_SEQ_PMG_TIMING:
-		*out_reg = mmMC_SEQ_PMG_TIMING_LP;
-		break;
-	case mmMC_PMG_CMD_MRS2:
-		*out_reg = mmMC_SEQ_PMG_CMD_MRS2_LP;
-		break;
-	case mmMC_SEQ_WR_CTL_2:
-		*out_reg = mmMC_SEQ_WR_CTL_2_LP;
-		break;
-	default:
-		result = false;
-		break;
-	}
-
-	return result;
-}
-
-static void ci_set_valid_flag(struct ci_mc_reg_table *table)
-{
-	u8 i, j;
-
-	for (i = 0; i < table->last; i++) {
-		for (j = 1; j < table->num_entries; j++) {
-			if (table->mc_reg_table_entry[j-1].mc_data[i] !=
-			    table->mc_reg_table_entry[j].mc_data[i]) {
-				table->valid_flag |= 1 << i;
-				break;
-			}
-		}
-	}
-}
-
-static void ci_set_s0_mc_reg_index(struct ci_mc_reg_table *table)
-{
-	u32 i;
-	u16 address;
-
-	for (i = 0; i < table->last; i++) {
-		table->mc_reg_address[i].s0 =
-			ci_check_s0_mc_reg_index(table->mc_reg_address[i].s1, &address) ?
-			address : table->mc_reg_address[i].s1;
-	}
-}
-
-static int ci_copy_vbios_mc_reg_table(const struct atom_mc_reg_table *table,
-				      struct ci_mc_reg_table *ci_table)
-{
-	u8 i, j;
-
-	if (table->last > SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE)
-		return -EINVAL;
-	if (table->num_entries > MAX_AC_TIMING_ENTRIES)
-		return -EINVAL;
-
-	for (i = 0; i < table->last; i++)
-		ci_table->mc_reg_address[i].s1 = table->mc_reg_address[i].s1;
-
-	ci_table->last = table->last;
-
-	for (i = 0; i < table->num_entries; i++) {
-		ci_table->mc_reg_table_entry[i].mclk_max =
-			table->mc_reg_table_entry[i].mclk_max;
-		for (j = 0; j < table->last; j++)
-			ci_table->mc_reg_table_entry[i].mc_data[j] =
-				table->mc_reg_table_entry[i].mc_data[j];
-	}
-	ci_table->num_entries = table->num_entries;
-
-	return 0;
-}
-
-static int ci_register_patching_mc_seq(struct amdgpu_device *adev,
-				       struct ci_mc_reg_table *table)
-{
-	u8 i, k;
-	u32 tmp;
-	bool patch;
-
-	tmp = RREG32(mmMC_SEQ_MISC0);
-	patch = ((tmp & 0x0000f00) == 0x300) ? true : false;
-
-	if (patch &&
-	    ((adev->pdev->device == 0x67B0) ||
-	     (adev->pdev->device == 0x67B1))) {
-		for (i = 0; i < table->last; i++) {
-			if (table->last >= SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE)
-				return -EINVAL;
-			switch (table->mc_reg_address[i].s1) {
-			case mmMC_SEQ_MISC1:
-				for (k = 0; k < table->num_entries; k++) {
-					if ((table->mc_reg_table_entry[k].mclk_max == 125000) ||
-					    (table->mc_reg_table_entry[k].mclk_max == 137500))
-						table->mc_reg_table_entry[k].mc_data[i] =
-							(table->mc_reg_table_entry[k].mc_data[i] & 0xFFFFFFF8) |
-							0x00000007;
-				}
-				break;
-			case mmMC_SEQ_WR_CTL_D0:
-				for (k = 0; k < table->num_entries; k++) {
-					if ((table->mc_reg_table_entry[k].mclk_max == 125000) ||
-					    (table->mc_reg_table_entry[k].mclk_max == 137500))
-						table->mc_reg_table_entry[k].mc_data[i] =
-							(table->mc_reg_table_entry[k].mc_data[i] & 0xFFFF0F00) |
-							0x0000D0DD;
-				}
-				break;
-			case mmMC_SEQ_WR_CTL_D1:
-				for (k = 0; k < table->num_entries; k++) {
-					if ((table->mc_reg_table_entry[k].mclk_max == 125000) ||
-					    (table->mc_reg_table_entry[k].mclk_max == 137500))
-						table->mc_reg_table_entry[k].mc_data[i] =
-							(table->mc_reg_table_entry[k].mc_data[i] & 0xFFFF0F00) |
-							0x0000D0DD;
-				}
-				break;
-			case mmMC_SEQ_WR_CTL_2:
-				for (k = 0; k < table->num_entries; k++) {
-					if ((table->mc_reg_table_entry[k].mclk_max == 125000) ||
-					    (table->mc_reg_table_entry[k].mclk_max == 137500))
-						table->mc_reg_table_entry[k].mc_data[i] = 0;
-				}
-				break;
-			case mmMC_SEQ_CAS_TIMING:
-				for (k = 0; k < table->num_entries; k++) {
-					if (table->mc_reg_table_entry[k].mclk_max == 125000)
-						table->mc_reg_table_entry[k].mc_data[i] =
-							(table->mc_reg_table_entry[k].mc_data[i] & 0xFFE0FE0F) |
-							0x000C0140;
-					else if (table->mc_reg_table_entry[k].mclk_max == 137500)
-						table->mc_reg_table_entry[k].mc_data[i] =
-							(table->mc_reg_table_entry[k].mc_data[i] & 0xFFE0FE0F) |
-							0x000C0150;
-				}
-				break;
-			case mmMC_SEQ_MISC_TIMING:
-				for (k = 0; k < table->num_entries; k++) {
-					if (table->mc_reg_table_entry[k].mclk_max == 125000)
-						table->mc_reg_table_entry[k].mc_data[i] =
-							(table->mc_reg_table_entry[k].mc_data[i] & 0xFFFFFFE0) |
-							0x00000030;
-					else if (table->mc_reg_table_entry[k].mclk_max == 137500)
-						table->mc_reg_table_entry[k].mc_data[i] =
-							(table->mc_reg_table_entry[k].mc_data[i] & 0xFFFFFFE0) |
-							0x00000035;
-				}
-				break;
-			default:
-				break;
-			}
-		}
-
-		WREG32(mmMC_SEQ_IO_DEBUG_INDEX, 3);
-		tmp = RREG32(mmMC_SEQ_IO_DEBUG_DATA);
-		tmp = (tmp & 0xFFF8FFFF) | (1 << 16);
-		WREG32(mmMC_SEQ_IO_DEBUG_INDEX, 3);
-		WREG32(mmMC_SEQ_IO_DEBUG_DATA, tmp);
-	}
-
-	return 0;
-}
-
-static int ci_initialize_mc_reg_table(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct atom_mc_reg_table *table;
-	struct ci_mc_reg_table *ci_table = &pi->mc_reg_table;
-	u8 module_index = ci_get_memory_module_index(adev);
-	int ret;
-
-	table = kzalloc(sizeof(struct atom_mc_reg_table), GFP_KERNEL);
-	if (!table)
-		return -ENOMEM;
-
-	WREG32(mmMC_SEQ_RAS_TIMING_LP, RREG32(mmMC_SEQ_RAS_TIMING));
-	WREG32(mmMC_SEQ_CAS_TIMING_LP, RREG32(mmMC_SEQ_CAS_TIMING));
-	WREG32(mmMC_SEQ_DLL_STBY_LP, RREG32(mmMC_SEQ_DLL_STBY));
-	WREG32(mmMC_SEQ_G5PDX_CMD0_LP, RREG32(mmMC_SEQ_G5PDX_CMD0));
-	WREG32(mmMC_SEQ_G5PDX_CMD1_LP, RREG32(mmMC_SEQ_G5PDX_CMD1));
-	WREG32(mmMC_SEQ_G5PDX_CTRL_LP, RREG32(mmMC_SEQ_G5PDX_CTRL));
-	WREG32(mmMC_SEQ_PMG_DVS_CMD_LP, RREG32(mmMC_SEQ_PMG_DVS_CMD));
-	WREG32(mmMC_SEQ_PMG_DVS_CTL_LP, RREG32(mmMC_SEQ_PMG_DVS_CTL));
-	WREG32(mmMC_SEQ_MISC_TIMING_LP, RREG32(mmMC_SEQ_MISC_TIMING));
-	WREG32(mmMC_SEQ_MISC_TIMING2_LP, RREG32(mmMC_SEQ_MISC_TIMING2));
-	WREG32(mmMC_SEQ_PMG_CMD_EMRS_LP, RREG32(mmMC_PMG_CMD_EMRS));
-	WREG32(mmMC_SEQ_PMG_CMD_MRS_LP, RREG32(mmMC_PMG_CMD_MRS));
-	WREG32(mmMC_SEQ_PMG_CMD_MRS1_LP, RREG32(mmMC_PMG_CMD_MRS1));
-	WREG32(mmMC_SEQ_WR_CTL_D0_LP, RREG32(mmMC_SEQ_WR_CTL_D0));
-	WREG32(mmMC_SEQ_WR_CTL_D1_LP, RREG32(mmMC_SEQ_WR_CTL_D1));
-	WREG32(mmMC_SEQ_RD_CTL_D0_LP, RREG32(mmMC_SEQ_RD_CTL_D0));
-	WREG32(mmMC_SEQ_RD_CTL_D1_LP, RREG32(mmMC_SEQ_RD_CTL_D1));
-	WREG32(mmMC_SEQ_PMG_TIMING_LP, RREG32(mmMC_SEQ_PMG_TIMING));
-	WREG32(mmMC_SEQ_PMG_CMD_MRS2_LP, RREG32(mmMC_PMG_CMD_MRS2));
-	WREG32(mmMC_SEQ_WR_CTL_2_LP, RREG32(mmMC_SEQ_WR_CTL_2));
-
-	ret = amdgpu_atombios_init_mc_reg_table(adev, module_index, table);
-	if (ret)
-		goto init_mc_done;
-
-	ret = ci_copy_vbios_mc_reg_table(table, ci_table);
-	if (ret)
-		goto init_mc_done;
-
-	ci_set_s0_mc_reg_index(ci_table);
-
-	ret = ci_register_patching_mc_seq(adev, ci_table);
-	if (ret)
-		goto init_mc_done;
-
-	ret = ci_set_mc_special_registers(adev, ci_table);
-	if (ret)
-		goto init_mc_done;
-
-	ci_set_valid_flag(ci_table);
-
-init_mc_done:
-	kfree(table);
-
-	return ret;
-}
-
-static int ci_populate_mc_reg_addresses(struct amdgpu_device *adev,
-					SMU7_Discrete_MCRegisters *mc_reg_table)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u32 i, j;
-
-	for (i = 0, j = 0; j < pi->mc_reg_table.last; j++) {
-		if (pi->mc_reg_table.valid_flag & (1 << j)) {
-			if (i >= SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE)
-				return -EINVAL;
-			mc_reg_table->address[i].s0 = cpu_to_be16(pi->mc_reg_table.mc_reg_address[j].s0);
-			mc_reg_table->address[i].s1 = cpu_to_be16(pi->mc_reg_table.mc_reg_address[j].s1);
-			i++;
-		}
-	}
-
-	mc_reg_table->last = (u8)i;
-
-	return 0;
-}
-
-static void ci_convert_mc_registers(const struct ci_mc_reg_entry *entry,
-				    SMU7_Discrete_MCRegisterSet *data,
-				    u32 num_entries, u32 valid_flag)
-{
-	u32 i, j;
-
-	for (i = 0, j = 0; j < num_entries; j++) {
-		if (valid_flag & (1 << j)) {
-			data->value[i] = cpu_to_be32(entry->mc_data[j]);
-			i++;
-		}
-	}
-}
-
-static void ci_convert_mc_reg_table_entry_to_smc(struct amdgpu_device *adev,
-						 const u32 memory_clock,
-						 SMU7_Discrete_MCRegisterSet *mc_reg_table_data)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u32 i = 0;
-
-	for(i = 0; i < pi->mc_reg_table.num_entries; i++) {
-		if (memory_clock <= pi->mc_reg_table.mc_reg_table_entry[i].mclk_max)
-			break;
-	}
-
-	if ((i == pi->mc_reg_table.num_entries) && (i > 0))
-		--i;
-
-	ci_convert_mc_registers(&pi->mc_reg_table.mc_reg_table_entry[i],
-				mc_reg_table_data, pi->mc_reg_table.last,
-				pi->mc_reg_table.valid_flag);
-}
-
-static void ci_convert_mc_reg_table_to_smc(struct amdgpu_device *adev,
-					   SMU7_Discrete_MCRegisters *mc_reg_table)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	u32 i;
-
-	for (i = 0; i < pi->dpm_table.mclk_table.count; i++)
-		ci_convert_mc_reg_table_entry_to_smc(adev,
-						     pi->dpm_table.mclk_table.dpm_levels[i].value,
-						     &mc_reg_table->data[i]);
-}
-
-static int ci_populate_initial_mc_reg_table(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	int ret;
-
-	memset(&pi->smc_mc_reg_table, 0, sizeof(SMU7_Discrete_MCRegisters));
-
-	ret = ci_populate_mc_reg_addresses(adev, &pi->smc_mc_reg_table);
-	if (ret)
-		return ret;
-	ci_convert_mc_reg_table_to_smc(adev, &pi->smc_mc_reg_table);
-
-	return amdgpu_ci_copy_bytes_to_smc(adev,
-				    pi->mc_reg_table_start,
-				    (u8 *)&pi->smc_mc_reg_table,
-				    sizeof(SMU7_Discrete_MCRegisters),
-				    pi->sram_end);
-}
-
-static int ci_update_and_upload_mc_reg_table(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	if (!(pi->need_update_smu7_dpm_table & DPMTABLE_OD_UPDATE_MCLK))
-		return 0;
-
-	memset(&pi->smc_mc_reg_table, 0, sizeof(SMU7_Discrete_MCRegisters));
-
-	ci_convert_mc_reg_table_to_smc(adev, &pi->smc_mc_reg_table);
-
-	return amdgpu_ci_copy_bytes_to_smc(adev,
-				    pi->mc_reg_table_start +
-				    offsetof(SMU7_Discrete_MCRegisters, data[0]),
-				    (u8 *)&pi->smc_mc_reg_table.data[0],
-				    sizeof(SMU7_Discrete_MCRegisterSet) *
-				    pi->dpm_table.mclk_table.count,
-				    pi->sram_end);
-}
-
-static void ci_enable_voltage_control(struct amdgpu_device *adev)
-{
-	u32 tmp = RREG32_SMC(ixGENERAL_PWRMGT);
-
-	tmp |= GENERAL_PWRMGT__VOLT_PWRMGT_EN_MASK;
-	WREG32_SMC(ixGENERAL_PWRMGT, tmp);
-}
-
-static enum amdgpu_pcie_gen ci_get_maximum_link_speed(struct amdgpu_device *adev,
-						      struct amdgpu_ps *amdgpu_state)
-{
-	struct ci_ps *state = ci_get_ps(amdgpu_state);
-	int i;
-	u16 pcie_speed, max_speed = 0;
-
-	for (i = 0; i < state->performance_level_count; i++) {
-		pcie_speed = state->performance_levels[i].pcie_gen;
-		if (max_speed < pcie_speed)
-			max_speed = pcie_speed;
-	}
-
-	return max_speed;
-}
-
-static u16 ci_get_current_pcie_speed(struct amdgpu_device *adev)
-{
-	u32 speed_cntl = 0;
-
-	speed_cntl = RREG32_PCIE(ixPCIE_LC_SPEED_CNTL) &
-		PCIE_LC_SPEED_CNTL__LC_CURRENT_DATA_RATE_MASK;
-	speed_cntl >>= PCIE_LC_SPEED_CNTL__LC_CURRENT_DATA_RATE__SHIFT;
-
-	return (u16)speed_cntl;
-}
-
-static int ci_get_current_pcie_lane_number(struct amdgpu_device *adev)
-{
-	u32 link_width = 0;
-
-	link_width = RREG32_PCIE(ixPCIE_LC_LINK_WIDTH_CNTL) &
-		PCIE_LC_LINK_WIDTH_CNTL__LC_LINK_WIDTH_RD_MASK;
-	link_width >>= PCIE_LC_LINK_WIDTH_CNTL__LC_LINK_WIDTH_RD__SHIFT;
-
-	switch (link_width) {
-	case 1:
-		return 1;
-	case 2:
-		return 2;
-	case 3:
-		return 4;
-	case 4:
-		return 8;
-	case 0:
-	case 6:
-	default:
-		return 16;
-	}
-}
-
-static void ci_request_link_speed_change_before_state_change(struct amdgpu_device *adev,
-							     struct amdgpu_ps *amdgpu_new_state,
-							     struct amdgpu_ps *amdgpu_current_state)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	enum amdgpu_pcie_gen target_link_speed =
-		ci_get_maximum_link_speed(adev, amdgpu_new_state);
-	enum amdgpu_pcie_gen current_link_speed;
-
-	if (pi->force_pcie_gen == AMDGPU_PCIE_GEN_INVALID)
-		current_link_speed = ci_get_maximum_link_speed(adev, amdgpu_current_state);
-	else
-		current_link_speed = pi->force_pcie_gen;
-
-	pi->force_pcie_gen = AMDGPU_PCIE_GEN_INVALID;
-	pi->pspp_notify_required = false;
-	if (target_link_speed > current_link_speed) {
-		switch (target_link_speed) {
-#ifdef CONFIG_ACPI
-		case AMDGPU_PCIE_GEN3:
-			if (amdgpu_acpi_pcie_performance_request(adev, PCIE_PERF_REQ_PECI_GEN3, false) == 0)
-				break;
-			pi->force_pcie_gen = AMDGPU_PCIE_GEN2;
-			if (current_link_speed == AMDGPU_PCIE_GEN2)
-				break;
-		case AMDGPU_PCIE_GEN2:
-			if (amdgpu_acpi_pcie_performance_request(adev, PCIE_PERF_REQ_PECI_GEN2, false) == 0)
-				break;
-#endif
-		default:
-			pi->force_pcie_gen = ci_get_current_pcie_speed(adev);
-			break;
-		}
-	} else {
-		if (target_link_speed < current_link_speed)
-			pi->pspp_notify_required = true;
-	}
-}
-
-static void ci_notify_link_speed_change_after_state_change(struct amdgpu_device *adev,
-							   struct amdgpu_ps *amdgpu_new_state,
-							   struct amdgpu_ps *amdgpu_current_state)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	enum amdgpu_pcie_gen target_link_speed =
-		ci_get_maximum_link_speed(adev, amdgpu_new_state);
-	u8 request;
-
-	if (pi->pspp_notify_required) {
-		if (target_link_speed == AMDGPU_PCIE_GEN3)
-			request = PCIE_PERF_REQ_PECI_GEN3;
-		else if (target_link_speed == AMDGPU_PCIE_GEN2)
-			request = PCIE_PERF_REQ_PECI_GEN2;
-		else
-			request = PCIE_PERF_REQ_PECI_GEN1;
-
-		if ((request == PCIE_PERF_REQ_PECI_GEN1) &&
-		    (ci_get_current_pcie_speed(adev) > 0))
-			return;
-
-#ifdef CONFIG_ACPI
-		amdgpu_acpi_pcie_performance_request(adev, request, false);
-#endif
-	}
-}
-
-static int ci_set_private_data_variables_based_on_pptable(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct amdgpu_clock_voltage_dependency_table *allowed_sclk_vddc_table =
-		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
-	struct amdgpu_clock_voltage_dependency_table *allowed_mclk_vddc_table =
-		&adev->pm.dpm.dyn_state.vddc_dependency_on_mclk;
-	struct amdgpu_clock_voltage_dependency_table *allowed_mclk_vddci_table =
-		&adev->pm.dpm.dyn_state.vddci_dependency_on_mclk;
-
-	if (allowed_sclk_vddc_table == NULL)
-		return -EINVAL;
-	if (allowed_sclk_vddc_table->count < 1)
-		return -EINVAL;
-	if (allowed_mclk_vddc_table == NULL)
-		return -EINVAL;
-	if (allowed_mclk_vddc_table->count < 1)
-		return -EINVAL;
-	if (allowed_mclk_vddci_table == NULL)
-		return -EINVAL;
-	if (allowed_mclk_vddci_table->count < 1)
-		return -EINVAL;
-
-	pi->min_vddc_in_pp_table = allowed_sclk_vddc_table->entries[0].v;
-	pi->max_vddc_in_pp_table =
-		allowed_sclk_vddc_table->entries[allowed_sclk_vddc_table->count - 1].v;
-
-	pi->min_vddci_in_pp_table = allowed_mclk_vddci_table->entries[0].v;
-	pi->max_vddci_in_pp_table =
-		allowed_mclk_vddci_table->entries[allowed_mclk_vddci_table->count - 1].v;
-
-	adev->pm.dpm.dyn_state.max_clock_voltage_on_ac.sclk =
-		allowed_sclk_vddc_table->entries[allowed_sclk_vddc_table->count - 1].clk;
-	adev->pm.dpm.dyn_state.max_clock_voltage_on_ac.mclk =
-		allowed_mclk_vddc_table->entries[allowed_sclk_vddc_table->count - 1].clk;
-	adev->pm.dpm.dyn_state.max_clock_voltage_on_ac.vddc =
-		allowed_sclk_vddc_table->entries[allowed_sclk_vddc_table->count - 1].v;
-	adev->pm.dpm.dyn_state.max_clock_voltage_on_ac.vddci =
-		allowed_mclk_vddci_table->entries[allowed_mclk_vddci_table->count - 1].v;
-
-	return 0;
-}
-
-static void ci_patch_with_vddc_leakage(struct amdgpu_device *adev, u16 *vddc)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct ci_leakage_voltage *leakage_table = &pi->vddc_leakage;
-	u32 leakage_index;
-
-	for (leakage_index = 0; leakage_index < leakage_table->count; leakage_index++) {
-		if (leakage_table->leakage_id[leakage_index] == *vddc) {
-			*vddc = leakage_table->actual_voltage[leakage_index];
-			break;
-		}
-	}
-}
-
-static void ci_patch_with_vddci_leakage(struct amdgpu_device *adev, u16 *vddci)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct ci_leakage_voltage *leakage_table = &pi->vddci_leakage;
-	u32 leakage_index;
-
-	for (leakage_index = 0; leakage_index < leakage_table->count; leakage_index++) {
-		if (leakage_table->leakage_id[leakage_index] == *vddci) {
-			*vddci = leakage_table->actual_voltage[leakage_index];
-			break;
-		}
-	}
-}
-
-static void ci_patch_clock_voltage_dependency_table_with_vddc_leakage(struct amdgpu_device *adev,
-								      struct amdgpu_clock_voltage_dependency_table *table)
-{
-	u32 i;
-
-	if (table) {
-		for (i = 0; i < table->count; i++)
-			ci_patch_with_vddc_leakage(adev, &table->entries[i].v);
-	}
-}
-
-static void ci_patch_clock_voltage_dependency_table_with_vddci_leakage(struct amdgpu_device *adev,
-								       struct amdgpu_clock_voltage_dependency_table *table)
-{
-	u32 i;
-
-	if (table) {
-		for (i = 0; i < table->count; i++)
-			ci_patch_with_vddci_leakage(adev, &table->entries[i].v);
-	}
-}
-
-static void ci_patch_vce_clock_voltage_dependency_table_with_vddc_leakage(struct amdgpu_device *adev,
-									  struct amdgpu_vce_clock_voltage_dependency_table *table)
-{
-	u32 i;
-
-	if (table) {
-		for (i = 0; i < table->count; i++)
-			ci_patch_with_vddc_leakage(adev, &table->entries[i].v);
-	}
-}
-
-static void ci_patch_uvd_clock_voltage_dependency_table_with_vddc_leakage(struct amdgpu_device *adev,
-									  struct amdgpu_uvd_clock_voltage_dependency_table *table)
-{
-	u32 i;
-
-	if (table) {
-		for (i = 0; i < table->count; i++)
-			ci_patch_with_vddc_leakage(adev, &table->entries[i].v);
-	}
-}
-
-static void ci_patch_vddc_phase_shed_limit_table_with_vddc_leakage(struct amdgpu_device *adev,
-								   struct amdgpu_phase_shedding_limits_table *table)
-{
-	u32 i;
-
-	if (table) {
-		for (i = 0; i < table->count; i++)
-			ci_patch_with_vddc_leakage(adev, &table->entries[i].voltage);
-	}
-}
-
-static void ci_patch_clock_voltage_limits_with_vddc_leakage(struct amdgpu_device *adev,
-							    struct amdgpu_clock_and_voltage_limits *table)
-{
-	if (table) {
-		ci_patch_with_vddc_leakage(adev, (u16 *)&table->vddc);
-		ci_patch_with_vddci_leakage(adev, (u16 *)&table->vddci);
-	}
-}
-
-static void ci_patch_cac_leakage_table_with_vddc_leakage(struct amdgpu_device *adev,
-							 struct amdgpu_cac_leakage_table *table)
-{
-	u32 i;
-
-	if (table) {
-		for (i = 0; i < table->count; i++)
-			ci_patch_with_vddc_leakage(adev, &table->entries[i].vddc);
-	}
-}
-
-static void ci_patch_dependency_tables_with_leakage(struct amdgpu_device *adev)
-{
-
-	ci_patch_clock_voltage_dependency_table_with_vddc_leakage(adev,
-								  &adev->pm.dpm.dyn_state.vddc_dependency_on_sclk);
-	ci_patch_clock_voltage_dependency_table_with_vddc_leakage(adev,
-								  &adev->pm.dpm.dyn_state.vddc_dependency_on_mclk);
-	ci_patch_clock_voltage_dependency_table_with_vddc_leakage(adev,
-								  &adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk);
-	ci_patch_clock_voltage_dependency_table_with_vddci_leakage(adev,
-								   &adev->pm.dpm.dyn_state.vddci_dependency_on_mclk);
-	ci_patch_vce_clock_voltage_dependency_table_with_vddc_leakage(adev,
-								      &adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table);
-	ci_patch_uvd_clock_voltage_dependency_table_with_vddc_leakage(adev,
-								      &adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table);
-	ci_patch_clock_voltage_dependency_table_with_vddc_leakage(adev,
-								  &adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table);
-	ci_patch_clock_voltage_dependency_table_with_vddc_leakage(adev,
-								  &adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table);
-	ci_patch_vddc_phase_shed_limit_table_with_vddc_leakage(adev,
-							       &adev->pm.dpm.dyn_state.phase_shedding_limits_table);
-	ci_patch_clock_voltage_limits_with_vddc_leakage(adev,
-							&adev->pm.dpm.dyn_state.max_clock_voltage_on_ac);
-	ci_patch_clock_voltage_limits_with_vddc_leakage(adev,
-							&adev->pm.dpm.dyn_state.max_clock_voltage_on_dc);
-	ci_patch_cac_leakage_table_with_vddc_leakage(adev,
-						     &adev->pm.dpm.dyn_state.cac_leakage_table);
-
-}
-
-static void ci_update_current_ps(struct amdgpu_device *adev,
-				 struct amdgpu_ps *rps)
-{
-	struct ci_ps *new_ps = ci_get_ps(rps);
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	pi->current_rps = *rps;
-	pi->current_ps = *new_ps;
-	pi->current_rps.ps_priv = &pi->current_ps;
-	adev->pm.dpm.current_ps = &pi->current_rps;
-}
-
-static void ci_update_requested_ps(struct amdgpu_device *adev,
-				   struct amdgpu_ps *rps)
-{
-	struct ci_ps *new_ps = ci_get_ps(rps);
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	pi->requested_rps = *rps;
-	pi->requested_ps = *new_ps;
-	pi->requested_rps.ps_priv = &pi->requested_ps;
-	adev->pm.dpm.requested_ps = &pi->requested_rps;
-}
-
-static int ci_dpm_pre_set_power_state(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct amdgpu_ps requested_ps = *adev->pm.dpm.requested_ps;
-	struct amdgpu_ps *new_ps = &requested_ps;
-
-	ci_update_requested_ps(adev, new_ps);
-
-	ci_apply_state_adjust_rules(adev, &pi->requested_rps);
-
-	return 0;
-}
-
-static void ci_dpm_post_set_power_state(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct amdgpu_ps *new_ps = &pi->requested_rps;
-
-	ci_update_current_ps(adev, new_ps);
-}
-
-
-static void ci_dpm_setup_asic(struct amdgpu_device *adev)
-{
-	ci_read_clock_registers(adev);
-	ci_enable_acpi_power_management(adev);
-	ci_init_sclk_t(adev);
-}
-
-static int ci_dpm_enable(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct amdgpu_ps *boot_ps = adev->pm.dpm.boot_ps;
-	int ret;
-
-	if (pi->voltage_control != CISLANDS_VOLTAGE_CONTROL_NONE) {
-		ci_enable_voltage_control(adev);
-		ret = ci_construct_voltage_tables(adev);
-		if (ret) {
-			DRM_ERROR("ci_construct_voltage_tables failed\n");
-			return ret;
-		}
-	}
-	if (pi->caps_dynamic_ac_timing) {
-		ret = ci_initialize_mc_reg_table(adev);
-		if (ret)
-			pi->caps_dynamic_ac_timing = false;
-	}
-	if (pi->dynamic_ss)
-		ci_enable_spread_spectrum(adev, true);
-	if (pi->thermal_protection)
-		ci_enable_thermal_protection(adev, true);
-	ci_program_sstp(adev);
-	ci_enable_display_gap(adev);
-	ci_program_vc(adev);
-	ret = ci_upload_firmware(adev);
-	if (ret) {
-		DRM_ERROR("ci_upload_firmware failed\n");
-		return ret;
-	}
-	ret = ci_process_firmware_header(adev);
-	if (ret) {
-		DRM_ERROR("ci_process_firmware_header failed\n");
-		return ret;
-	}
-	ret = ci_initial_switch_from_arb_f0_to_f1(adev);
-	if (ret) {
-		DRM_ERROR("ci_initial_switch_from_arb_f0_to_f1 failed\n");
-		return ret;
-	}
-	ret = ci_init_smc_table(adev);
-	if (ret) {
-		DRM_ERROR("ci_init_smc_table failed\n");
-		return ret;
-	}
-	ret = ci_init_arb_table_index(adev);
-	if (ret) {
-		DRM_ERROR("ci_init_arb_table_index failed\n");
-		return ret;
-	}
-	if (pi->caps_dynamic_ac_timing) {
-		ret = ci_populate_initial_mc_reg_table(adev);
-		if (ret) {
-			DRM_ERROR("ci_populate_initial_mc_reg_table failed\n");
-			return ret;
-		}
-	}
-	ret = ci_populate_pm_base(adev);
-	if (ret) {
-		DRM_ERROR("ci_populate_pm_base failed\n");
-		return ret;
-	}
-	ci_dpm_start_smc(adev);
-	ci_enable_vr_hot_gpio_interrupt(adev);
-	ret = ci_notify_smc_display_change(adev, false);
-	if (ret) {
-		DRM_ERROR("ci_notify_smc_display_change failed\n");
-		return ret;
-	}
-	ci_enable_sclk_control(adev, true);
-	ret = ci_enable_ulv(adev, true);
-	if (ret) {
-		DRM_ERROR("ci_enable_ulv failed\n");
-		return ret;
-	}
-	ret = ci_enable_ds_master_switch(adev, true);
-	if (ret) {
-		DRM_ERROR("ci_enable_ds_master_switch failed\n");
-		return ret;
-	}
-	ret = ci_start_dpm(adev);
-	if (ret) {
-		DRM_ERROR("ci_start_dpm failed\n");
-		return ret;
-	}
-	ret = ci_enable_didt(adev, true);
-	if (ret) {
-		DRM_ERROR("ci_enable_didt failed\n");
-		return ret;
-	}
-	ret = ci_enable_smc_cac(adev, true);
-	if (ret) {
-		DRM_ERROR("ci_enable_smc_cac failed\n");
-		return ret;
-	}
-	ret = ci_enable_power_containment(adev, true);
-	if (ret) {
-		DRM_ERROR("ci_enable_power_containment failed\n");
-		return ret;
-	}
-
-	ret = ci_power_control_set_level(adev);
-	if (ret) {
-		DRM_ERROR("ci_power_control_set_level failed\n");
-		return ret;
-	}
-
-	ci_enable_auto_throttle_source(adev, AMDGPU_DPM_AUTO_THROTTLE_SRC_THERMAL, true);
-
-	ret = ci_enable_thermal_based_sclk_dpm(adev, true);
-	if (ret) {
-		DRM_ERROR("ci_enable_thermal_based_sclk_dpm failed\n");
-		return ret;
-	}
-
-	ci_thermal_start_thermal_controller(adev);
-
-	ci_update_current_ps(adev, boot_ps);
-
-	return 0;
-}
-
-static void ci_dpm_disable(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct amdgpu_ps *boot_ps = adev->pm.dpm.boot_ps;
-
-	amdgpu_irq_put(adev, &adev->pm.dpm.thermal.irq,
-		       AMDGPU_THERMAL_IRQ_LOW_TO_HIGH);
-	amdgpu_irq_put(adev, &adev->pm.dpm.thermal.irq,
-		       AMDGPU_THERMAL_IRQ_HIGH_TO_LOW);
-
-	ci_dpm_powergate_uvd(adev, true);
-
-	if (!amdgpu_ci_is_smc_running(adev))
-		return;
-
-	ci_thermal_stop_thermal_controller(adev);
-
-	if (pi->thermal_protection)
-		ci_enable_thermal_protection(adev, false);
-	ci_enable_power_containment(adev, false);
-	ci_enable_smc_cac(adev, false);
-	ci_enable_didt(adev, false);
-	ci_enable_spread_spectrum(adev, false);
-	ci_enable_auto_throttle_source(adev, AMDGPU_DPM_AUTO_THROTTLE_SRC_THERMAL, false);
-	ci_stop_dpm(adev);
-	ci_enable_ds_master_switch(adev, false);
-	ci_enable_ulv(adev, false);
-	ci_clear_vc(adev);
-	ci_reset_to_default(adev);
-	ci_dpm_stop_smc(adev);
-	ci_force_switch_to_arb_f0(adev);
-	ci_enable_thermal_based_sclk_dpm(adev, false);
-
-	ci_update_current_ps(adev, boot_ps);
-}
-
-static int ci_dpm_set_power_state(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct amdgpu_ps *new_ps = &pi->requested_rps;
-	struct amdgpu_ps *old_ps = &pi->current_rps;
-	int ret;
-
-	ci_find_dpm_states_clocks_in_dpm_table(adev, new_ps);
-	if (pi->pcie_performance_request)
-		ci_request_link_speed_change_before_state_change(adev, new_ps, old_ps);
-	ret = ci_freeze_sclk_mclk_dpm(adev);
-	if (ret) {
-		DRM_ERROR("ci_freeze_sclk_mclk_dpm failed\n");
-		return ret;
-	}
-	ret = ci_populate_and_upload_sclk_mclk_dpm_levels(adev, new_ps);
-	if (ret) {
-		DRM_ERROR("ci_populate_and_upload_sclk_mclk_dpm_levels failed\n");
-		return ret;
-	}
-	ret = ci_generate_dpm_level_enable_mask(adev, new_ps);
-	if (ret) {
-		DRM_ERROR("ci_generate_dpm_level_enable_mask failed\n");
-		return ret;
-	}
-
-	ret = ci_update_vce_dpm(adev, new_ps, old_ps);
-	if (ret) {
-		DRM_ERROR("ci_update_vce_dpm failed\n");
-		return ret;
-	}
-
-	ret = ci_update_sclk_t(adev);
-	if (ret) {
-		DRM_ERROR("ci_update_sclk_t failed\n");
-		return ret;
-	}
-	if (pi->caps_dynamic_ac_timing) {
-		ret = ci_update_and_upload_mc_reg_table(adev);
-		if (ret) {
-			DRM_ERROR("ci_update_and_upload_mc_reg_table failed\n");
-			return ret;
-		}
-	}
-	ret = ci_program_memory_timing_parameters(adev);
-	if (ret) {
-		DRM_ERROR("ci_program_memory_timing_parameters failed\n");
-		return ret;
-	}
-	ret = ci_unfreeze_sclk_mclk_dpm(adev);
-	if (ret) {
-		DRM_ERROR("ci_unfreeze_sclk_mclk_dpm failed\n");
-		return ret;
-	}
-	ret = ci_upload_dpm_level_enable_mask(adev);
-	if (ret) {
-		DRM_ERROR("ci_upload_dpm_level_enable_mask failed\n");
-		return ret;
-	}
-	if (pi->pcie_performance_request)
-		ci_notify_link_speed_change_after_state_change(adev, new_ps, old_ps);
-
-	return 0;
-}
-
-#if 0
-static void ci_dpm_reset_asic(struct amdgpu_device *adev)
-{
-	ci_set_boot_state(adev);
-}
-#endif
-
-static void ci_dpm_display_configuration_changed(struct amdgpu_device *adev)
-{
-	ci_program_display_gap(adev);
-}
-
-union power_info {
-	struct _ATOM_POWERPLAY_INFO info;
-	struct _ATOM_POWERPLAY_INFO_V2 info_2;
-	struct _ATOM_POWERPLAY_INFO_V3 info_3;
-	struct _ATOM_PPLIB_POWERPLAYTABLE pplib;
-	struct _ATOM_PPLIB_POWERPLAYTABLE2 pplib2;
-	struct _ATOM_PPLIB_POWERPLAYTABLE3 pplib3;
-};
-
-union pplib_clock_info {
-	struct _ATOM_PPLIB_R600_CLOCK_INFO r600;
-	struct _ATOM_PPLIB_RS780_CLOCK_INFO rs780;
-	struct _ATOM_PPLIB_EVERGREEN_CLOCK_INFO evergreen;
-	struct _ATOM_PPLIB_SUMO_CLOCK_INFO sumo;
-	struct _ATOM_PPLIB_SI_CLOCK_INFO si;
-	struct _ATOM_PPLIB_CI_CLOCK_INFO ci;
-};
-
-union pplib_power_state {
-	struct _ATOM_PPLIB_STATE v1;
-	struct _ATOM_PPLIB_STATE_V2 v2;
-};
-
-static void ci_parse_pplib_non_clock_info(struct amdgpu_device *adev,
-					  struct amdgpu_ps *rps,
-					  struct _ATOM_PPLIB_NONCLOCK_INFO *non_clock_info,
-					  u8 table_rev)
-{
-	rps->caps = le32_to_cpu(non_clock_info->ulCapsAndSettings);
-	rps->class = le16_to_cpu(non_clock_info->usClassification);
-	rps->class2 = le16_to_cpu(non_clock_info->usClassification2);
-
-	if (ATOM_PPLIB_NONCLOCKINFO_VER1 < table_rev) {
-		rps->vclk = le32_to_cpu(non_clock_info->ulVCLK);
-		rps->dclk = le32_to_cpu(non_clock_info->ulDCLK);
-	} else {
-		rps->vclk = 0;
-		rps->dclk = 0;
-	}
-
-	if (rps->class & ATOM_PPLIB_CLASSIFICATION_BOOT)
-		adev->pm.dpm.boot_ps = rps;
-	if (rps->class & ATOM_PPLIB_CLASSIFICATION_UVDSTATE)
-		adev->pm.dpm.uvd_ps = rps;
-}
-
-static void ci_parse_pplib_clock_info(struct amdgpu_device *adev,
-				      struct amdgpu_ps *rps, int index,
-				      union pplib_clock_info *clock_info)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct ci_ps *ps = ci_get_ps(rps);
-	struct ci_pl *pl = &ps->performance_levels[index];
-
-	ps->performance_level_count = index + 1;
-
-	pl->sclk = le16_to_cpu(clock_info->ci.usEngineClockLow);
-	pl->sclk |= clock_info->ci.ucEngineClockHigh << 16;
-	pl->mclk = le16_to_cpu(clock_info->ci.usMemoryClockLow);
-	pl->mclk |= clock_info->ci.ucMemoryClockHigh << 16;
-
-	pl->pcie_gen = amdgpu_get_pcie_gen_support(adev,
-						   pi->sys_pcie_mask,
-						   pi->vbios_boot_state.pcie_gen_bootup_value,
-						   clock_info->ci.ucPCIEGen);
-	pl->pcie_lane = amdgpu_get_pcie_lane_support(adev,
-						     pi->vbios_boot_state.pcie_lane_bootup_value,
-						     le16_to_cpu(clock_info->ci.usPCIELane));
-
-	if (rps->class & ATOM_PPLIB_CLASSIFICATION_ACPI) {
-		pi->acpi_pcie_gen = pl->pcie_gen;
-	}
-
-	if (rps->class2 & ATOM_PPLIB_CLASSIFICATION2_ULV) {
-		pi->ulv.supported = true;
-		pi->ulv.pl = *pl;
-		pi->ulv.cg_ulv_parameter = CISLANDS_CGULVPARAMETER_DFLT;
-	}
-
-	/* patch up boot state */
-	if (rps->class & ATOM_PPLIB_CLASSIFICATION_BOOT) {
-		pl->mclk = pi->vbios_boot_state.mclk_bootup_value;
-		pl->sclk = pi->vbios_boot_state.sclk_bootup_value;
-		pl->pcie_gen = pi->vbios_boot_state.pcie_gen_bootup_value;
-		pl->pcie_lane = pi->vbios_boot_state.pcie_lane_bootup_value;
-	}
-
-	switch (rps->class & ATOM_PPLIB_CLASSIFICATION_UI_MASK) {
-	case ATOM_PPLIB_CLASSIFICATION_UI_BATTERY:
-		pi->use_pcie_powersaving_levels = true;
-		if (pi->pcie_gen_powersaving.max < pl->pcie_gen)
-			pi->pcie_gen_powersaving.max = pl->pcie_gen;
-		if (pi->pcie_gen_powersaving.min > pl->pcie_gen)
-			pi->pcie_gen_powersaving.min = pl->pcie_gen;
-		if (pi->pcie_lane_powersaving.max < pl->pcie_lane)
-			pi->pcie_lane_powersaving.max = pl->pcie_lane;
-		if (pi->pcie_lane_powersaving.min > pl->pcie_lane)
-			pi->pcie_lane_powersaving.min = pl->pcie_lane;
-		break;
-	case ATOM_PPLIB_CLASSIFICATION_UI_PERFORMANCE:
-		pi->use_pcie_performance_levels = true;
-		if (pi->pcie_gen_performance.max < pl->pcie_gen)
-			pi->pcie_gen_performance.max = pl->pcie_gen;
-		if (pi->pcie_gen_performance.min > pl->pcie_gen)
-			pi->pcie_gen_performance.min = pl->pcie_gen;
-		if (pi->pcie_lane_performance.max < pl->pcie_lane)
-			pi->pcie_lane_performance.max = pl->pcie_lane;
-		if (pi->pcie_lane_performance.min > pl->pcie_lane)
-			pi->pcie_lane_performance.min = pl->pcie_lane;
-		break;
-	default:
-		break;
-	}
-}
-
-static int ci_parse_power_table(struct amdgpu_device *adev)
-{
-	struct amdgpu_mode_info *mode_info = &adev->mode_info;
-	struct _ATOM_PPLIB_NONCLOCK_INFO *non_clock_info;
-	union pplib_power_state *power_state;
-	int i, j, k, non_clock_array_index, clock_array_index;
-	union pplib_clock_info *clock_info;
-	struct _StateArray *state_array;
-	struct _ClockInfoArray *clock_info_array;
-	struct _NonClockInfoArray *non_clock_info_array;
-	union power_info *power_info;
-	int index = GetIndexIntoMasterTable(DATA, PowerPlayInfo);
-	u16 data_offset;
-	u8 frev, crev;
-	u8 *power_state_offset;
-	struct ci_ps *ps;
-
-	if (!amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
-				   &frev, &crev, &data_offset))
-		return -EINVAL;
-	power_info = (union power_info *)(mode_info->atom_context->bios + data_offset);
-
-	amdgpu_add_thermal_controller(adev);
-
-	state_array = (struct _StateArray *)
-		(mode_info->atom_context->bios + data_offset +
-		 le16_to_cpu(power_info->pplib.usStateArrayOffset));
-	clock_info_array = (struct _ClockInfoArray *)
-		(mode_info->atom_context->bios + data_offset +
-		 le16_to_cpu(power_info->pplib.usClockInfoArrayOffset));
-	non_clock_info_array = (struct _NonClockInfoArray *)
-		(mode_info->atom_context->bios + data_offset +
-		 le16_to_cpu(power_info->pplib.usNonClockInfoArrayOffset));
-
-	adev->pm.dpm.ps = kzalloc(sizeof(struct amdgpu_ps) *
-				  state_array->ucNumEntries, GFP_KERNEL);
-	if (!adev->pm.dpm.ps)
-		return -ENOMEM;
-	power_state_offset = (u8 *)state_array->states;
-	for (i = 0; i < state_array->ucNumEntries; i++) {
-		u8 *idx;
-		power_state = (union pplib_power_state *)power_state_offset;
-		non_clock_array_index = power_state->v2.nonClockInfoIndex;
-		non_clock_info = (struct _ATOM_PPLIB_NONCLOCK_INFO *)
-			&non_clock_info_array->nonClockInfo[non_clock_array_index];
-		ps = kzalloc(sizeof(struct ci_ps), GFP_KERNEL);
-		if (ps == NULL) {
-			kfree(adev->pm.dpm.ps);
-			return -ENOMEM;
-		}
-		adev->pm.dpm.ps[i].ps_priv = ps;
-		ci_parse_pplib_non_clock_info(adev, &adev->pm.dpm.ps[i],
-					      non_clock_info,
-					      non_clock_info_array->ucEntrySize);
-		k = 0;
-		idx = (u8 *)&power_state->v2.clockInfoIndex[0];
-		for (j = 0; j < power_state->v2.ucNumDPMLevels; j++) {
-			clock_array_index = idx[j];
-			if (clock_array_index >= clock_info_array->ucNumEntries)
-				continue;
-			if (k >= CISLANDS_MAX_HARDWARE_POWERLEVELS)
-				break;
-			clock_info = (union pplib_clock_info *)
-				((u8 *)&clock_info_array->clockInfo[0] +
-				 (clock_array_index * clock_info_array->ucEntrySize));
-			ci_parse_pplib_clock_info(adev,
-						  &adev->pm.dpm.ps[i], k,
-						  clock_info);
-			k++;
-		}
-		power_state_offset += 2 + power_state->v2.ucNumDPMLevels;
-	}
-	adev->pm.dpm.num_ps = state_array->ucNumEntries;
-
-	/* fill in the vce power states */
-	for (i = 0; i < adev->pm.dpm.num_of_vce_states; i++) {
-		u32 sclk, mclk;
-		clock_array_index = adev->pm.dpm.vce_states[i].clk_idx;
-		clock_info = (union pplib_clock_info *)
-			&clock_info_array->clockInfo[clock_array_index * clock_info_array->ucEntrySize];
-		sclk = le16_to_cpu(clock_info->ci.usEngineClockLow);
-		sclk |= clock_info->ci.ucEngineClockHigh << 16;
-		mclk = le16_to_cpu(clock_info->ci.usMemoryClockLow);
-		mclk |= clock_info->ci.ucMemoryClockHigh << 16;
-		adev->pm.dpm.vce_states[i].sclk = sclk;
-		adev->pm.dpm.vce_states[i].mclk = mclk;
-	}
-
-	return 0;
-}
-
-static int ci_get_vbios_boot_values(struct amdgpu_device *adev,
-				    struct ci_vbios_boot_state *boot_state)
-{
-	struct amdgpu_mode_info *mode_info = &adev->mode_info;
-	int index = GetIndexIntoMasterTable(DATA, FirmwareInfo);
-	ATOM_FIRMWARE_INFO_V2_2 *firmware_info;
-	u8 frev, crev;
-	u16 data_offset;
-
-	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
-				   &frev, &crev, &data_offset)) {
-		firmware_info =
-			(ATOM_FIRMWARE_INFO_V2_2 *)(mode_info->atom_context->bios +
-						    data_offset);
-		boot_state->mvdd_bootup_value = le16_to_cpu(firmware_info->usBootUpMVDDCVoltage);
-		boot_state->vddc_bootup_value = le16_to_cpu(firmware_info->usBootUpVDDCVoltage);
-		boot_state->vddci_bootup_value = le16_to_cpu(firmware_info->usBootUpVDDCIVoltage);
-		boot_state->pcie_gen_bootup_value = ci_get_current_pcie_speed(adev);
-		boot_state->pcie_lane_bootup_value = ci_get_current_pcie_lane_number(adev);
-		boot_state->sclk_bootup_value = le32_to_cpu(firmware_info->ulDefaultEngineClock);
-		boot_state->mclk_bootup_value = le32_to_cpu(firmware_info->ulDefaultMemoryClock);
-
-		return 0;
-	}
-	return -EINVAL;
-}
-
-static void ci_dpm_fini(struct amdgpu_device *adev)
-{
-	int i;
-
-	for (i = 0; i < adev->pm.dpm.num_ps; i++) {
-		kfree(adev->pm.dpm.ps[i].ps_priv);
-	}
-	kfree(adev->pm.dpm.ps);
-	kfree(adev->pm.dpm.priv);
-	kfree(adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries);
-	amdgpu_free_extended_power_table(adev);
-}
-
-/**
- * ci_dpm_init_microcode - load ucode images from disk
- *
- * @adev: amdgpu_device pointer
- *
- * Use the firmware interface to load the ucode images into
- * the driver (not loaded into hw).
- * Returns 0 on success, error on failure.
- */
-static int ci_dpm_init_microcode(struct amdgpu_device *adev)
-{
-	const char *chip_name;
-	char fw_name[30];
-	int err;
-
-	DRM_DEBUG("\n");
-
-	switch (adev->asic_type) {
-	case CHIP_BONAIRE:
-		if ((adev->pdev->revision == 0x80) ||
-		    (adev->pdev->revision == 0x81) ||
-		    (adev->pdev->device == 0x665f))
-			chip_name = "bonaire_k";
-		else
-			chip_name = "bonaire";
-		break;
-	case CHIP_HAWAII:
-		if (adev->pdev->revision == 0x80)
-			chip_name = "hawaii_k";
-		else
-			chip_name = "hawaii";
-		break;
-	case CHIP_KAVERI:
-	case CHIP_KABINI:
-	case CHIP_MULLINS:
-	default: BUG();
-	}
-
-	snprintf(fw_name, sizeof(fw_name), "radeon/%s_smc.bin", chip_name);
-	err = request_firmware(&adev->pm.fw, fw_name, adev->dev);
-	if (err)
-		goto out;
-	err = amdgpu_ucode_validate(adev->pm.fw);
-
-out:
-	if (err) {
-		pr_err("cik_smc: Failed to load firmware \"%s\"\n", fw_name);
-		release_firmware(adev->pm.fw);
-		adev->pm.fw = NULL;
-	}
-	return err;
-}
-
-static int ci_dpm_init(struct amdgpu_device *adev)
-{
-	int index = GetIndexIntoMasterTable(DATA, ASIC_InternalSS_Info);
-	SMU7_Discrete_DpmTable *dpm_table;
-	struct amdgpu_gpio_rec gpio;
-	u16 data_offset, size;
-	u8 frev, crev;
-	struct ci_power_info *pi;
-	int ret;
-
-	pi = kzalloc(sizeof(struct ci_power_info), GFP_KERNEL);
-	if (pi == NULL)
-		return -ENOMEM;
-	adev->pm.dpm.priv = pi;
-
-	pi->sys_pcie_mask =
-		(adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_MASK) >>
-		CAIL_PCIE_LINK_SPEED_SUPPORT_SHIFT;
-
-	pi->force_pcie_gen = AMDGPU_PCIE_GEN_INVALID;
-
-	pi->pcie_gen_performance.max = AMDGPU_PCIE_GEN1;
-	pi->pcie_gen_performance.min = AMDGPU_PCIE_GEN3;
-	pi->pcie_gen_powersaving.max = AMDGPU_PCIE_GEN1;
-	pi->pcie_gen_powersaving.min = AMDGPU_PCIE_GEN3;
-
-	pi->pcie_lane_performance.max = 0;
-	pi->pcie_lane_performance.min = 16;
-	pi->pcie_lane_powersaving.max = 0;
-	pi->pcie_lane_powersaving.min = 16;
-
-	ret = ci_get_vbios_boot_values(adev, &pi->vbios_boot_state);
-	if (ret) {
-		ci_dpm_fini(adev);
-		return ret;
-	}
-
-	ret = amdgpu_get_platform_caps(adev);
-	if (ret) {
-		ci_dpm_fini(adev);
-		return ret;
-	}
-
-	ret = amdgpu_parse_extended_power_table(adev);
-	if (ret) {
-		ci_dpm_fini(adev);
-		return ret;
-	}
-
-	ret = ci_parse_power_table(adev);
-	if (ret) {
-		ci_dpm_fini(adev);
-		return ret;
-	}
-
-	pi->dll_default_on = false;
-	pi->sram_end = SMC_RAM_END;
-
-	pi->activity_target[0] = CISLAND_TARGETACTIVITY_DFLT;
-	pi->activity_target[1] = CISLAND_TARGETACTIVITY_DFLT;
-	pi->activity_target[2] = CISLAND_TARGETACTIVITY_DFLT;
-	pi->activity_target[3] = CISLAND_TARGETACTIVITY_DFLT;
-	pi->activity_target[4] = CISLAND_TARGETACTIVITY_DFLT;
-	pi->activity_target[5] = CISLAND_TARGETACTIVITY_DFLT;
-	pi->activity_target[6] = CISLAND_TARGETACTIVITY_DFLT;
-	pi->activity_target[7] = CISLAND_TARGETACTIVITY_DFLT;
-
-	pi->mclk_activity_target = CISLAND_MCLK_TARGETACTIVITY_DFLT;
-
-	pi->sclk_dpm_key_disabled = 0;
-	pi->mclk_dpm_key_disabled = 0;
-	pi->pcie_dpm_key_disabled = 0;
-	pi->thermal_sclk_dpm_enabled = 0;
-
-	if (amdgpu_pp_feature_mask & SCLK_DEEP_SLEEP_MASK)
-		pi->caps_sclk_ds = true;
-	else
-		pi->caps_sclk_ds = false;
-
-	pi->mclk_strobe_mode_threshold = 40000;
-	pi->mclk_stutter_mode_threshold = 40000;
-	pi->mclk_edc_enable_threshold = 40000;
-	pi->mclk_edc_wr_enable_threshold = 40000;
-
-	ci_initialize_powertune_defaults(adev);
-
-	pi->caps_fps = false;
-
-	pi->caps_sclk_throttle_low_notification = false;
-
-	pi->caps_uvd_dpm = true;
-	pi->caps_vce_dpm = true;
-
-	ci_get_leakage_voltages(adev);
-	ci_patch_dependency_tables_with_leakage(adev);
-	ci_set_private_data_variables_based_on_pptable(adev);
-
-	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries =
-		kzalloc(4 * sizeof(struct amdgpu_clock_voltage_dependency_entry), GFP_KERNEL);
-	if (!adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries) {
-		ci_dpm_fini(adev);
-		return -ENOMEM;
-	}
-	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.count = 4;
-	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[0].clk = 0;
-	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[0].v = 0;
-	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[1].clk = 36000;
-	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[1].v = 720;
-	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[2].clk = 54000;
-	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[2].v = 810;
-	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[3].clk = 72000;
-	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[3].v = 900;
-
-	adev->pm.dpm.dyn_state.mclk_sclk_ratio = 4;
-	adev->pm.dpm.dyn_state.sclk_mclk_delta = 15000;
-	adev->pm.dpm.dyn_state.vddc_vddci_delta = 200;
-
-	adev->pm.dpm.dyn_state.valid_sclk_values.count = 0;
-	adev->pm.dpm.dyn_state.valid_sclk_values.values = NULL;
-	adev->pm.dpm.dyn_state.valid_mclk_values.count = 0;
-	adev->pm.dpm.dyn_state.valid_mclk_values.values = NULL;
-
-	if (adev->asic_type == CHIP_HAWAII) {
-		pi->thermal_temp_setting.temperature_low = 94500;
-		pi->thermal_temp_setting.temperature_high = 95000;
-		pi->thermal_temp_setting.temperature_shutdown = 104000;
-	} else {
-		pi->thermal_temp_setting.temperature_low = 99500;
-		pi->thermal_temp_setting.temperature_high = 100000;
-		pi->thermal_temp_setting.temperature_shutdown = 104000;
-	}
-
-	pi->uvd_enabled = false;
-
-	dpm_table = &pi->smc_state_table;
-
-	gpio = amdgpu_atombios_lookup_gpio(adev, VDDC_VRHOT_GPIO_PINID);
-	if (gpio.valid) {
-		dpm_table->VRHotGpio = gpio.shift;
-		adev->pm.dpm.platform_caps |= ATOM_PP_PLATFORM_CAP_REGULATOR_HOT;
-	} else {
-		dpm_table->VRHotGpio = CISLANDS_UNUSED_GPIO_PIN;
-		adev->pm.dpm.platform_caps &= ~ATOM_PP_PLATFORM_CAP_REGULATOR_HOT;
-	}
-
-	gpio = amdgpu_atombios_lookup_gpio(adev, PP_AC_DC_SWITCH_GPIO_PINID);
-	if (gpio.valid) {
-		dpm_table->AcDcGpio = gpio.shift;
-		adev->pm.dpm.platform_caps |= ATOM_PP_PLATFORM_CAP_HARDWAREDC;
-	} else {
-		dpm_table->AcDcGpio = CISLANDS_UNUSED_GPIO_PIN;
-		adev->pm.dpm.platform_caps &= ~ATOM_PP_PLATFORM_CAP_HARDWAREDC;
-	}
-
-	gpio = amdgpu_atombios_lookup_gpio(adev, VDDC_PCC_GPIO_PINID);
-	if (gpio.valid) {
-		u32 tmp = RREG32_SMC(ixCNB_PWRMGT_CNTL);
-
-		switch (gpio.shift) {
-		case 0:
-			tmp &= ~CNB_PWRMGT_CNTL__GNB_SLOW_MODE_MASK;
-			tmp |= 1 << CNB_PWRMGT_CNTL__GNB_SLOW_MODE__SHIFT;
-			break;
-		case 1:
-			tmp &= ~CNB_PWRMGT_CNTL__GNB_SLOW_MODE_MASK;
-			tmp |= 2 << CNB_PWRMGT_CNTL__GNB_SLOW_MODE__SHIFT;
-			break;
-		case 2:
-			tmp |= CNB_PWRMGT_CNTL__GNB_SLOW_MASK;
-			break;
-		case 3:
-			tmp |= CNB_PWRMGT_CNTL__FORCE_NB_PS1_MASK;
-			break;
-		case 4:
-			tmp |= CNB_PWRMGT_CNTL__DPM_ENABLED_MASK;
-			break;
-		default:
-			DRM_INFO("Invalid PCC GPIO: %u!\n", gpio.shift);
-			break;
-		}
-		WREG32_SMC(ixCNB_PWRMGT_CNTL, tmp);
-	}
-
-	pi->voltage_control = CISLANDS_VOLTAGE_CONTROL_NONE;
-	pi->vddci_control = CISLANDS_VOLTAGE_CONTROL_NONE;
-	pi->mvdd_control = CISLANDS_VOLTAGE_CONTROL_NONE;
-	if (amdgpu_atombios_is_voltage_gpio(adev, VOLTAGE_TYPE_VDDC, VOLTAGE_OBJ_GPIO_LUT))
-		pi->voltage_control = CISLANDS_VOLTAGE_CONTROL_BY_GPIO;
-	else if (amdgpu_atombios_is_voltage_gpio(adev, VOLTAGE_TYPE_VDDC, VOLTAGE_OBJ_SVID2))
-		pi->voltage_control = CISLANDS_VOLTAGE_CONTROL_BY_SVID2;
-
-	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_VDDCI_CONTROL) {
-		if (amdgpu_atombios_is_voltage_gpio(adev, VOLTAGE_TYPE_VDDCI, VOLTAGE_OBJ_GPIO_LUT))
-			pi->vddci_control = CISLANDS_VOLTAGE_CONTROL_BY_GPIO;
-		else if (amdgpu_atombios_is_voltage_gpio(adev, VOLTAGE_TYPE_VDDCI, VOLTAGE_OBJ_SVID2))
-			pi->vddci_control = CISLANDS_VOLTAGE_CONTROL_BY_SVID2;
-		else
-			adev->pm.dpm.platform_caps &= ~ATOM_PP_PLATFORM_CAP_VDDCI_CONTROL;
-	}
-
-	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_MVDDCONTROL) {
-		if (amdgpu_atombios_is_voltage_gpio(adev, VOLTAGE_TYPE_MVDDC, VOLTAGE_OBJ_GPIO_LUT))
-			pi->mvdd_control = CISLANDS_VOLTAGE_CONTROL_BY_GPIO;
-		else if (amdgpu_atombios_is_voltage_gpio(adev, VOLTAGE_TYPE_MVDDC, VOLTAGE_OBJ_SVID2))
-			pi->mvdd_control = CISLANDS_VOLTAGE_CONTROL_BY_SVID2;
-		else
-			adev->pm.dpm.platform_caps &= ~ATOM_PP_PLATFORM_CAP_MVDDCONTROL;
-	}
-
-	pi->vddc_phase_shed_control = true;
-
-#if defined(CONFIG_ACPI)
-	pi->pcie_performance_request =
-		amdgpu_acpi_is_pcie_performance_request_supported(adev);
-#else
-	pi->pcie_performance_request = false;
-#endif
-
-	if (amdgpu_atom_parse_data_header(adev->mode_info.atom_context, index, &size,
-				   &frev, &crev, &data_offset)) {
-		pi->caps_sclk_ss_support = true;
-		pi->caps_mclk_ss_support = true;
-		pi->dynamic_ss = true;
-	} else {
-		pi->caps_sclk_ss_support = false;
-		pi->caps_mclk_ss_support = false;
-		pi->dynamic_ss = true;
-	}
-
-	if (adev->pm.int_thermal_type != THERMAL_TYPE_NONE)
-		pi->thermal_protection = true;
-	else
-		pi->thermal_protection = false;
-
-	pi->caps_dynamic_ac_timing = true;
-
-	pi->uvd_power_gated = true;
-
-	/* make sure dc limits are valid */
-	if ((adev->pm.dpm.dyn_state.max_clock_voltage_on_dc.sclk == 0) ||
-	    (adev->pm.dpm.dyn_state.max_clock_voltage_on_dc.mclk == 0))
-		adev->pm.dpm.dyn_state.max_clock_voltage_on_dc =
-			adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
-
-	pi->fan_ctrl_is_in_default_mode = true;
-
-	return 0;
-}
-
-static void
-ci_dpm_debugfs_print_current_performance_level(struct amdgpu_device *adev,
-					       struct seq_file *m)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct amdgpu_ps *rps = &pi->current_rps;
-	u32 sclk = ci_get_average_sclk_freq(adev);
-	u32 mclk = ci_get_average_mclk_freq(adev);
-	u32 activity_percent = 50;
-	int ret;
-
-	ret = ci_read_smc_soft_register(adev, offsetof(SMU7_SoftRegisters, AverageGraphicsA),
-					&activity_percent);
-
-	if (ret == 0) {
-		activity_percent += 0x80;
-		activity_percent >>= 8;
-		activity_percent = activity_percent > 100 ? 100 : activity_percent;
-	}
-
-	seq_printf(m, "uvd %sabled\n", pi->uvd_power_gated ? "dis" : "en");
-	seq_printf(m, "vce %sabled\n", rps->vce_active ? "en" : "dis");
-	seq_printf(m, "power level avg    sclk: %u mclk: %u\n",
-		   sclk, mclk);
-	seq_printf(m, "GPU load: %u %%\n", activity_percent);
-}
-
-static void ci_dpm_print_power_state(struct amdgpu_device *adev,
-				     struct amdgpu_ps *rps)
-{
-	struct ci_ps *ps = ci_get_ps(rps);
-	struct ci_pl *pl;
-	int i;
-
-	amdgpu_dpm_print_class_info(rps->class, rps->class2);
-	amdgpu_dpm_print_cap_info(rps->caps);
-	printk("\tuvd    vclk: %d dclk: %d\n", rps->vclk, rps->dclk);
-	for (i = 0; i < ps->performance_level_count; i++) {
-		pl = &ps->performance_levels[i];
-		printk("\t\tpower level %d    sclk: %u mclk: %u pcie gen: %u pcie lanes: %u\n",
-		       i, pl->sclk, pl->mclk, pl->pcie_gen + 1, pl->pcie_lane);
-	}
-	amdgpu_dpm_print_ps_status(adev, rps);
-}
-
-static inline bool ci_are_power_levels_equal(const struct ci_pl *ci_cpl1,
-						const struct ci_pl *ci_cpl2)
-{
-	return ((ci_cpl1->mclk == ci_cpl2->mclk) &&
-		  (ci_cpl1->sclk == ci_cpl2->sclk) &&
-		  (ci_cpl1->pcie_gen == ci_cpl2->pcie_gen) &&
-		  (ci_cpl1->pcie_lane == ci_cpl2->pcie_lane));
-}
-
-static int ci_check_state_equal(struct amdgpu_device *adev,
-				struct amdgpu_ps *cps,
-				struct amdgpu_ps *rps,
-				bool *equal)
-{
-	struct ci_ps *ci_cps;
-	struct ci_ps *ci_rps;
-	int i;
-
-	if (adev == NULL || cps == NULL || rps == NULL || equal == NULL)
-		return -EINVAL;
-
-	ci_cps = ci_get_ps(cps);
-	ci_rps = ci_get_ps(rps);
-
-	if (ci_cps == NULL) {
-		*equal = false;
-		return 0;
-	}
-
-	if (ci_cps->performance_level_count != ci_rps->performance_level_count) {
-
-		*equal = false;
-		return 0;
-	}
-
-	for (i = 0; i < ci_cps->performance_level_count; i++) {
-		if (!ci_are_power_levels_equal(&(ci_cps->performance_levels[i]),
-					&(ci_rps->performance_levels[i]))) {
-			*equal = false;
-			return 0;
-		}
-	}
-
-	/* If all performance levels are the same try to use the UVD clocks to break the tie.*/
-	*equal = ((cps->vclk == rps->vclk) && (cps->dclk == rps->dclk));
-	*equal &= ((cps->evclk == rps->evclk) && (cps->ecclk == rps->ecclk));
-
-	return 0;
-}
-
-static u32 ci_dpm_get_sclk(struct amdgpu_device *adev, bool low)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct ci_ps *requested_state = ci_get_ps(&pi->requested_rps);
-
-	if (low)
-		return requested_state->performance_levels[0].sclk;
-	else
-		return requested_state->performance_levels[requested_state->performance_level_count - 1].sclk;
-}
-
-static u32 ci_dpm_get_mclk(struct amdgpu_device *adev, bool low)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct ci_ps *requested_state = ci_get_ps(&pi->requested_rps);
-
-	if (low)
-		return requested_state->performance_levels[0].mclk;
-	else
-		return requested_state->performance_levels[requested_state->performance_level_count - 1].mclk;
-}
-
-/* get temperature in millidegrees */
-static int ci_dpm_get_temp(struct amdgpu_device *adev)
-{
-	u32 temp;
-	int actual_temp = 0;
-
-	temp = (RREG32_SMC(ixCG_MULT_THERMAL_STATUS) & CG_MULT_THERMAL_STATUS__CTF_TEMP_MASK) >>
-		CG_MULT_THERMAL_STATUS__CTF_TEMP__SHIFT;
-
-	if (temp & 0x200)
-		actual_temp = 255;
-	else
-		actual_temp = temp & 0x1ff;
-
-	actual_temp = actual_temp * 1000;
-
-	return actual_temp;
-}
-
-static int ci_set_temperature_range(struct amdgpu_device *adev)
-{
-	int ret;
-
-	ret = ci_thermal_enable_alert(adev, false);
-	if (ret)
-		return ret;
-	ret = ci_thermal_set_temperature_range(adev, CISLANDS_TEMP_RANGE_MIN,
-					       CISLANDS_TEMP_RANGE_MAX);
-	if (ret)
-		return ret;
-	ret = ci_thermal_enable_alert(adev, true);
-	if (ret)
-		return ret;
-	return ret;
-}
-
-static int ci_dpm_early_init(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	ci_dpm_set_dpm_funcs(adev);
-	ci_dpm_set_irq_funcs(adev);
-
-	return 0;
-}
-
-static int ci_dpm_late_init(void *handle)
-{
-	int ret;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (!amdgpu_dpm)
-		return 0;
-
-	/* init the sysfs and debugfs files late */
-	ret = amdgpu_pm_sysfs_init(adev);
-	if (ret)
-		return ret;
-
-	ret = ci_set_temperature_range(adev);
-	if (ret)
-		return ret;
-
-	return 0;
-}
-
-static int ci_dpm_sw_init(void *handle)
-{
-	int ret;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	ret = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 230,
-				&adev->pm.dpm.thermal.irq);
-	if (ret)
-		return ret;
-
-	ret = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 231,
-				&adev->pm.dpm.thermal.irq);
-	if (ret)
-		return ret;
-
-	/* default to balanced state */
-	adev->pm.dpm.state = POWER_STATE_TYPE_BALANCED;
-	adev->pm.dpm.user_state = POWER_STATE_TYPE_BALANCED;
-	adev->pm.dpm.forced_level = AMD_DPM_FORCED_LEVEL_AUTO;
-	adev->pm.default_sclk = adev->clock.default_sclk;
-	adev->pm.default_mclk = adev->clock.default_mclk;
-	adev->pm.current_sclk = adev->clock.default_sclk;
-	adev->pm.current_mclk = adev->clock.default_mclk;
-	adev->pm.int_thermal_type = THERMAL_TYPE_NONE;
-
-	ret = ci_dpm_init_microcode(adev);
-	if (ret)
-		return ret;
-
-	if (amdgpu_dpm == 0)
-		return 0;
-
-	INIT_WORK(&adev->pm.dpm.thermal.work, amdgpu_dpm_thermal_work_handler);
-	mutex_lock(&adev->pm.mutex);
-	ret = ci_dpm_init(adev);
-	if (ret)
-		goto dpm_failed;
-	adev->pm.dpm.current_ps = adev->pm.dpm.requested_ps = adev->pm.dpm.boot_ps;
-	if (amdgpu_dpm == 1)
-		amdgpu_pm_print_power_states(adev);
-	mutex_unlock(&adev->pm.mutex);
-	DRM_INFO("amdgpu: dpm initialized\n");
-
-	return 0;
-
-dpm_failed:
-	ci_dpm_fini(adev);
-	mutex_unlock(&adev->pm.mutex);
-	DRM_ERROR("amdgpu: dpm initialization failed\n");
-	return ret;
-}
-
-static int ci_dpm_sw_fini(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	flush_work(&adev->pm.dpm.thermal.work);
-
-	mutex_lock(&adev->pm.mutex);
-	amdgpu_pm_sysfs_fini(adev);
-	ci_dpm_fini(adev);
-	mutex_unlock(&adev->pm.mutex);
-
-	release_firmware(adev->pm.fw);
-	adev->pm.fw = NULL;
-
-	return 0;
-}
-
-static int ci_dpm_hw_init(void *handle)
-{
-	int ret;
-
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (!amdgpu_dpm) {
-		ret = ci_upload_firmware(adev);
-		if (ret) {
-			DRM_ERROR("ci_upload_firmware failed\n");
-			return ret;
-		}
-		ci_dpm_start_smc(adev);
-		return 0;
-	}
-
-	mutex_lock(&adev->pm.mutex);
-	ci_dpm_setup_asic(adev);
-	ret = ci_dpm_enable(adev);
-	if (ret)
-		adev->pm.dpm_enabled = false;
-	else
-		adev->pm.dpm_enabled = true;
-	mutex_unlock(&adev->pm.mutex);
-
-	return ret;
-}
-
-static int ci_dpm_hw_fini(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (adev->pm.dpm_enabled) {
-		mutex_lock(&adev->pm.mutex);
-		ci_dpm_disable(adev);
-		mutex_unlock(&adev->pm.mutex);
-	} else {
-		ci_dpm_stop_smc(adev);
-	}
-
-	return 0;
-}
-
-static int ci_dpm_suspend(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (adev->pm.dpm_enabled) {
-		mutex_lock(&adev->pm.mutex);
-		amdgpu_irq_put(adev, &adev->pm.dpm.thermal.irq,
-			       AMDGPU_THERMAL_IRQ_LOW_TO_HIGH);
-		amdgpu_irq_put(adev, &adev->pm.dpm.thermal.irq,
-			       AMDGPU_THERMAL_IRQ_HIGH_TO_LOW);
-		adev->pm.dpm.last_user_state = adev->pm.dpm.user_state;
-		adev->pm.dpm.last_state = adev->pm.dpm.state;
-		adev->pm.dpm.user_state = POWER_STATE_TYPE_INTERNAL_BOOT;
-		adev->pm.dpm.state = POWER_STATE_TYPE_INTERNAL_BOOT;
-		mutex_unlock(&adev->pm.mutex);
-		amdgpu_pm_compute_clocks(adev);
-
-	}
-
-	return 0;
-}
-
-static int ci_dpm_resume(void *handle)
-{
-	int ret;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (adev->pm.dpm_enabled) {
-		/* asic init will reset to the boot state */
-		mutex_lock(&adev->pm.mutex);
-		ci_dpm_setup_asic(adev);
-		ret = ci_dpm_enable(adev);
-		if (ret)
-			adev->pm.dpm_enabled = false;
-		else
-			adev->pm.dpm_enabled = true;
-		adev->pm.dpm.user_state = adev->pm.dpm.last_user_state;
-		adev->pm.dpm.state = adev->pm.dpm.last_state;
-		mutex_unlock(&adev->pm.mutex);
-		if (adev->pm.dpm_enabled)
-			amdgpu_pm_compute_clocks(adev);
-	}
-	return 0;
-}
-
-static bool ci_dpm_is_idle(void *handle)
-{
-	/* XXX */
-	return true;
-}
-
-static int ci_dpm_wait_for_idle(void *handle)
-{
-	/* XXX */
-	return 0;
-}
-
-static int ci_dpm_soft_reset(void *handle)
-{
-	return 0;
-}
-
-static int ci_dpm_set_interrupt_state(struct amdgpu_device *adev,
-				      struct amdgpu_irq_src *source,
-				      unsigned type,
-				      enum amdgpu_interrupt_state state)
-{
-	u32 cg_thermal_int;
-
-	switch (type) {
-	case AMDGPU_THERMAL_IRQ_LOW_TO_HIGH:
-		switch (state) {
-		case AMDGPU_IRQ_STATE_DISABLE:
-			cg_thermal_int = RREG32_SMC(ixCG_THERMAL_INT);
-			cg_thermal_int |= CG_THERMAL_INT_CTRL__THERM_INTH_MASK_MASK;
-			WREG32_SMC(ixCG_THERMAL_INT, cg_thermal_int);
-			break;
-		case AMDGPU_IRQ_STATE_ENABLE:
-			cg_thermal_int = RREG32_SMC(ixCG_THERMAL_INT);
-			cg_thermal_int &= ~CG_THERMAL_INT_CTRL__THERM_INTH_MASK_MASK;
-			WREG32_SMC(ixCG_THERMAL_INT, cg_thermal_int);
-			break;
-		default:
-			break;
-		}
-		break;
-
-	case AMDGPU_THERMAL_IRQ_HIGH_TO_LOW:
-		switch (state) {
-		case AMDGPU_IRQ_STATE_DISABLE:
-			cg_thermal_int = RREG32_SMC(ixCG_THERMAL_INT);
-			cg_thermal_int |= CG_THERMAL_INT_CTRL__THERM_INTL_MASK_MASK;
-			WREG32_SMC(ixCG_THERMAL_INT, cg_thermal_int);
-			break;
-		case AMDGPU_IRQ_STATE_ENABLE:
-			cg_thermal_int = RREG32_SMC(ixCG_THERMAL_INT);
-			cg_thermal_int &= ~CG_THERMAL_INT_CTRL__THERM_INTL_MASK_MASK;
-			WREG32_SMC(ixCG_THERMAL_INT, cg_thermal_int);
-			break;
-		default:
-			break;
-		}
-		break;
-
-	default:
-		break;
-	}
-	return 0;
-}
-
-static int ci_dpm_process_interrupt(struct amdgpu_device *adev,
-				    struct amdgpu_irq_src *source,
-				    struct amdgpu_iv_entry *entry)
-{
-	bool queue_thermal = false;
-
-	if (entry == NULL)
-		return -EINVAL;
-
-	switch (entry->src_id) {
-	case 230: /* thermal low to high */
-		DRM_DEBUG("IH: thermal low to high\n");
-		adev->pm.dpm.thermal.high_to_low = false;
-		queue_thermal = true;
-		break;
-	case 231: /* thermal high to low */
-		DRM_DEBUG("IH: thermal high to low\n");
-		adev->pm.dpm.thermal.high_to_low = true;
-		queue_thermal = true;
-		break;
-	default:
-		break;
-	}
-
-	if (queue_thermal)
-		schedule_work(&adev->pm.dpm.thermal.work);
-
-	return 0;
-}
-
-static int ci_dpm_set_clockgating_state(void *handle,
-					  enum amd_clockgating_state state)
-{
-	return 0;
-}
-
-static int ci_dpm_set_powergating_state(void *handle,
-					  enum amd_powergating_state state)
-{
-	return 0;
-}
-
-static int ci_dpm_print_clock_levels(struct amdgpu_device *adev,
-		enum pp_clock_type type, char *buf)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct ci_single_dpm_table *sclk_table = &pi->dpm_table.sclk_table;
-	struct ci_single_dpm_table *mclk_table = &pi->dpm_table.mclk_table;
-	struct ci_single_dpm_table *pcie_table = &pi->dpm_table.pcie_speed_table;
-
-	int i, now, size = 0;
-	uint32_t clock, pcie_speed;
-
-	switch (type) {
-	case PP_SCLK:
-		amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_API_GetSclkFrequency);
-		clock = RREG32(mmSMC_MSG_ARG_0);
-
-		for (i = 0; i < sclk_table->count; i++) {
-			if (clock > sclk_table->dpm_levels[i].value)
-				continue;
-			break;
-		}
-		now = i;
-
-		for (i = 0; i < sclk_table->count; i++)
-			size += sprintf(buf + size, "%d: %uMhz %s\n",
-					i, sclk_table->dpm_levels[i].value / 100,
-					(i == now) ? "*" : "");
-		break;
-	case PP_MCLK:
-		amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_API_GetMclkFrequency);
-		clock = RREG32(mmSMC_MSG_ARG_0);
-
-		for (i = 0; i < mclk_table->count; i++) {
-			if (clock > mclk_table->dpm_levels[i].value)
-				continue;
-			break;
-		}
-		now = i;
-
-		for (i = 0; i < mclk_table->count; i++)
-			size += sprintf(buf + size, "%d: %uMhz %s\n",
-					i, mclk_table->dpm_levels[i].value / 100,
-					(i == now) ? "*" : "");
-		break;
-	case PP_PCIE:
-		pcie_speed = ci_get_current_pcie_speed(adev);
-		for (i = 0; i < pcie_table->count; i++) {
-			if (pcie_speed != pcie_table->dpm_levels[i].value)
-				continue;
-			break;
-		}
-		now = i;
-
-		for (i = 0; i < pcie_table->count; i++)
-			size += sprintf(buf + size, "%d: %s %s\n", i,
-					(pcie_table->dpm_levels[i].value == 0) ? "2.5GB, x1" :
-					(pcie_table->dpm_levels[i].value == 1) ? "5.0GB, x16" :
-					(pcie_table->dpm_levels[i].value == 2) ? "8.0GB, x16" : "",
-					(i == now) ? "*" : "");
-		break;
-	default:
-		break;
-	}
-
-	return size;
-}
-
-static int ci_dpm_force_clock_level(struct amdgpu_device *adev,
-		enum pp_clock_type type, uint32_t mask)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	if (adev->pm.dpm.forced_level & (AMD_DPM_FORCED_LEVEL_AUTO |
-				AMD_DPM_FORCED_LEVEL_LOW |
-				AMD_DPM_FORCED_LEVEL_HIGH))
-		return -EINVAL;
-
-	switch (type) {
-	case PP_SCLK:
-		if (!pi->sclk_dpm_key_disabled)
-			amdgpu_ci_send_msg_to_smc_with_parameter(adev,
-					PPSMC_MSG_SCLKDPM_SetEnabledMask,
-					pi->dpm_level_enable_mask.sclk_dpm_enable_mask & mask);
-		break;
-
-	case PP_MCLK:
-		if (!pi->mclk_dpm_key_disabled)
-			amdgpu_ci_send_msg_to_smc_with_parameter(adev,
-					PPSMC_MSG_MCLKDPM_SetEnabledMask,
-					pi->dpm_level_enable_mask.mclk_dpm_enable_mask & mask);
-		break;
-
-	case PP_PCIE:
-	{
-		uint32_t tmp = mask & pi->dpm_level_enable_mask.pcie_dpm_enable_mask;
-		uint32_t level = 0;
-
-		while (tmp >>= 1)
-			level++;
-
-		if (!pi->pcie_dpm_key_disabled)
-			amdgpu_ci_send_msg_to_smc_with_parameter(adev,
-					PPSMC_MSG_PCIeDPM_ForceLevel,
-					level);
-		break;
-	}
-	default:
-		break;
-	}
-
-	return 0;
-}
-
-static int ci_dpm_get_sclk_od(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct ci_single_dpm_table *sclk_table = &(pi->dpm_table.sclk_table);
-	struct ci_single_dpm_table *golden_sclk_table =
-			&(pi->golden_dpm_table.sclk_table);
-	int value;
-
-	value = (sclk_table->dpm_levels[sclk_table->count - 1].value -
-			golden_sclk_table->dpm_levels[golden_sclk_table->count - 1].value) *
-			100 /
-			golden_sclk_table->dpm_levels[golden_sclk_table->count - 1].value;
-
-	return value;
-}
-
-static int ci_dpm_set_sclk_od(struct amdgpu_device *adev, uint32_t value)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct ci_ps *ps = ci_get_ps(adev->pm.dpm.requested_ps);
-	struct ci_single_dpm_table *golden_sclk_table =
-			&(pi->golden_dpm_table.sclk_table);
-
-	if (value > 20)
-		value = 20;
-
-	ps->performance_levels[ps->performance_level_count - 1].sclk =
-			golden_sclk_table->dpm_levels[golden_sclk_table->count - 1].value *
-			value / 100 +
-			golden_sclk_table->dpm_levels[golden_sclk_table->count - 1].value;
-
-	return 0;
-}
-
-static int ci_dpm_get_mclk_od(struct amdgpu_device *adev)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct ci_single_dpm_table *mclk_table = &(pi->dpm_table.mclk_table);
-	struct ci_single_dpm_table *golden_mclk_table =
-			&(pi->golden_dpm_table.mclk_table);
-	int value;
-
-	value = (mclk_table->dpm_levels[mclk_table->count - 1].value -
-			golden_mclk_table->dpm_levels[golden_mclk_table->count - 1].value) *
-			100 /
-			golden_mclk_table->dpm_levels[golden_mclk_table->count - 1].value;
-
-	return value;
-}
-
-static int ci_dpm_set_mclk_od(struct amdgpu_device *adev, uint32_t value)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct ci_ps *ps = ci_get_ps(adev->pm.dpm.requested_ps);
-	struct ci_single_dpm_table *golden_mclk_table =
-			&(pi->golden_dpm_table.mclk_table);
-
-	if (value > 20)
-		value = 20;
-
-	ps->performance_levels[ps->performance_level_count - 1].mclk =
-			golden_mclk_table->dpm_levels[golden_mclk_table->count - 1].value *
-			value / 100 +
-			golden_mclk_table->dpm_levels[golden_mclk_table->count - 1].value;
-
-	return 0;
-}
-
-static int ci_dpm_get_power_profile_state(struct amdgpu_device *adev,
-		struct amd_pp_profile *query)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	if (!pi || !query)
-		return -EINVAL;
-
-	if (query->type == AMD_PP_GFX_PROFILE)
-		memcpy(query, &pi->gfx_power_profile,
-				sizeof(struct amd_pp_profile));
-	else if (query->type == AMD_PP_COMPUTE_PROFILE)
-		memcpy(query, &pi->compute_power_profile,
-				sizeof(struct amd_pp_profile));
-	else
-		return -EINVAL;
-
-	return 0;
-}
-
-static int ci_populate_requested_graphic_levels(struct amdgpu_device *adev,
-		struct amd_pp_profile *request)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct ci_dpm_table *dpm_table = &(pi->dpm_table);
-	struct SMU7_Discrete_GraphicsLevel *levels =
-			pi->smc_state_table.GraphicsLevel;
-	uint32_t array = pi->dpm_table_start +
-			offsetof(SMU7_Discrete_DpmTable, GraphicsLevel);
-	uint32_t array_size = sizeof(struct SMU7_Discrete_GraphicsLevel) *
-			SMU7_MAX_LEVELS_GRAPHICS;
-	uint32_t i;
-
-	for (i = 0; i < dpm_table->sclk_table.count; i++) {
-		levels[i].ActivityLevel =
-				cpu_to_be16(request->activity_threshold);
-		levels[i].EnabledForActivity = 1;
-		levels[i].UpH = request->up_hyst;
-		levels[i].DownH = request->down_hyst;
-	}
-
-	return amdgpu_ci_copy_bytes_to_smc(adev, array, (uint8_t *)levels,
-				array_size, pi->sram_end);
-}
-
-static void ci_find_min_clock_masks(struct amdgpu_device *adev,
-		uint32_t *sclk_mask, uint32_t *mclk_mask,
-		uint32_t min_sclk, uint32_t min_mclk)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct ci_dpm_table *dpm_table = &(pi->dpm_table);
-	uint32_t i;
-
-	for (i = 0; i < dpm_table->sclk_table.count; i++) {
-		if (dpm_table->sclk_table.dpm_levels[i].enabled &&
-			dpm_table->sclk_table.dpm_levels[i].value >= min_sclk)
-			*sclk_mask |= 1 << i;
-	}
-
-	for (i = 0; i < dpm_table->mclk_table.count; i++) {
-		if (dpm_table->mclk_table.dpm_levels[i].enabled &&
-			dpm_table->mclk_table.dpm_levels[i].value >= min_mclk)
-			*mclk_mask |= 1 << i;
-	}
-}
-
-static int ci_set_power_profile_state(struct amdgpu_device *adev,
-		struct amd_pp_profile *request)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	int tmp_result, result = 0;
-	uint32_t sclk_mask = 0, mclk_mask = 0;
-
-	tmp_result = ci_freeze_sclk_mclk_dpm(adev);
-	if (tmp_result) {
-		DRM_ERROR("Failed to freeze SCLK MCLK DPM!");
-		result = tmp_result;
-	}
-
-	tmp_result = ci_populate_requested_graphic_levels(adev,
-			request);
-	if (tmp_result) {
-		DRM_ERROR("Failed to populate requested graphic levels!");
-		result = tmp_result;
-	}
-
-	tmp_result = ci_unfreeze_sclk_mclk_dpm(adev);
-	if (tmp_result) {
-		DRM_ERROR("Failed to unfreeze SCLK MCLK DPM!");
-		result = tmp_result;
-	}
-
-	ci_find_min_clock_masks(adev, &sclk_mask, &mclk_mask,
-			request->min_sclk, request->min_mclk);
-
-	if (sclk_mask) {
-		if (!pi->sclk_dpm_key_disabled)
-			amdgpu_ci_send_msg_to_smc_with_parameter(
-				adev,
-				PPSMC_MSG_SCLKDPM_SetEnabledMask,
-				pi->dpm_level_enable_mask.
-				sclk_dpm_enable_mask &
-				sclk_mask);
-	}
-
-	if (mclk_mask) {
-		if (!pi->mclk_dpm_key_disabled)
-			amdgpu_ci_send_msg_to_smc_with_parameter(
-				adev,
-				PPSMC_MSG_MCLKDPM_SetEnabledMask,
-				pi->dpm_level_enable_mask.
-				mclk_dpm_enable_mask &
-				mclk_mask);
-	}
-
-
-	return result;
-}
-
-static int ci_dpm_set_power_profile_state(struct amdgpu_device *adev,
-		struct amd_pp_profile *request)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	int ret = -1;
-
-	if (!pi || !request)
-		return -EINVAL;
-
-	if (adev->pm.dpm.forced_level !=
-			AMD_DPM_FORCED_LEVEL_AUTO)
-		return -EINVAL;
-
-	if (request->min_sclk ||
-		request->min_mclk ||
-		request->activity_threshold ||
-		request->up_hyst ||
-		request->down_hyst) {
-		if (request->type == AMD_PP_GFX_PROFILE)
-			memcpy(&pi->gfx_power_profile, request,
-					sizeof(struct amd_pp_profile));
-		else if (request->type == AMD_PP_COMPUTE_PROFILE)
-			memcpy(&pi->compute_power_profile, request,
-					sizeof(struct amd_pp_profile));
-		else
-			return -EINVAL;
-
-		if (request->type == pi->current_power_profile)
-			ret = ci_set_power_profile_state(
-					adev,
-					request);
-	} else {
-		/* set power profile if it exists */
-		switch (request->type) {
-		case AMD_PP_GFX_PROFILE:
-			ret = ci_set_power_profile_state(
-				adev,
-				&pi->gfx_power_profile);
-			break;
-		case AMD_PP_COMPUTE_PROFILE:
-			ret = ci_set_power_profile_state(
-				adev,
-				&pi->compute_power_profile);
-			break;
-		default:
-			return -EINVAL;
-		}
-	}
-
-	if (!ret)
-		pi->current_power_profile = request->type;
-
-	return 0;
-}
-
-static int ci_dpm_reset_power_profile_state(struct amdgpu_device *adev,
-		struct amd_pp_profile *request)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-
-	if (!pi || !request)
-		return -EINVAL;
-
-	if (request->type == AMD_PP_GFX_PROFILE) {
-		pi->gfx_power_profile = pi->default_gfx_power_profile;
-		return ci_dpm_set_power_profile_state(adev,
-					  &pi->gfx_power_profile);
-	} else if (request->type == AMD_PP_COMPUTE_PROFILE) {
-		pi->compute_power_profile =
-			pi->default_compute_power_profile;
-		return ci_dpm_set_power_profile_state(adev,
-					  &pi->compute_power_profile);
-	} else
-		return -EINVAL;
-}
-
-static int ci_dpm_switch_power_profile(struct amdgpu_device *adev,
-		enum amd_pp_profile_type type)
-{
-	struct ci_power_info *pi = ci_get_pi(adev);
-	struct amd_pp_profile request = {0};
-
-	if (!pi)
-		return -EINVAL;
-
-	if (pi->current_power_profile != type) {
-		request.type = type;
-		return ci_dpm_set_power_profile_state(adev, &request);
-	}
-
-	return 0;
-}
-
-static int ci_dpm_read_sensor(struct amdgpu_device *adev, int idx,
-			      void *value, int *size)
-{
-	u32 activity_percent = 50;
-	int ret;
-
-	/* size must be at least 4 bytes for all sensors */
-	if (*size < 4)
-		return -EINVAL;
-
-	switch (idx) {
-	case AMDGPU_PP_SENSOR_GFX_SCLK:
-		*((uint32_t *)value) = ci_get_average_sclk_freq(adev);
-		*size = 4;
-		return 0;
-	case AMDGPU_PP_SENSOR_GFX_MCLK:
-		*((uint32_t *)value) = ci_get_average_mclk_freq(adev);
-		*size = 4;
-		return 0;
-	case AMDGPU_PP_SENSOR_GPU_TEMP:
-		*((uint32_t *)value) = ci_dpm_get_temp(adev);
-		*size = 4;
-		return 0;
-	case AMDGPU_PP_SENSOR_GPU_LOAD:
-		ret = ci_read_smc_soft_register(adev,
-						offsetof(SMU7_SoftRegisters,
-							 AverageGraphicsA),
-						&activity_percent);
-		if (ret == 0) {
-			activity_percent += 0x80;
-			activity_percent >>= 8;
-			activity_percent =
-				activity_percent > 100 ? 100 : activity_percent;
-		}
-		*((uint32_t *)value) = activity_percent;
-		*size = 4;
-		return 0;
-	default:
-		return -EINVAL;
-	}
-}
-
-const struct amd_ip_funcs ci_dpm_ip_funcs = {
-	.name = "ci_dpm",
-	.early_init = ci_dpm_early_init,
-	.late_init = ci_dpm_late_init,
-	.sw_init = ci_dpm_sw_init,
-	.sw_fini = ci_dpm_sw_fini,
-	.hw_init = ci_dpm_hw_init,
-	.hw_fini = ci_dpm_hw_fini,
-	.suspend = ci_dpm_suspend,
-	.resume = ci_dpm_resume,
-	.is_idle = ci_dpm_is_idle,
-	.wait_for_idle = ci_dpm_wait_for_idle,
-	.soft_reset = ci_dpm_soft_reset,
-	.set_clockgating_state = ci_dpm_set_clockgating_state,
-	.set_powergating_state = ci_dpm_set_powergating_state,
-};
-
-static const struct amdgpu_dpm_funcs ci_dpm_funcs = {
-	.get_temperature = &ci_dpm_get_temp,
-	.pre_set_power_state = &ci_dpm_pre_set_power_state,
-	.set_power_state = &ci_dpm_set_power_state,
-	.post_set_power_state = &ci_dpm_post_set_power_state,
-	.display_configuration_changed = &ci_dpm_display_configuration_changed,
-	.get_sclk = &ci_dpm_get_sclk,
-	.get_mclk = &ci_dpm_get_mclk,
-	.print_power_state = &ci_dpm_print_power_state,
-	.debugfs_print_current_performance_level = &ci_dpm_debugfs_print_current_performance_level,
-	.force_performance_level = &ci_dpm_force_performance_level,
-	.vblank_too_short = &ci_dpm_vblank_too_short,
-	.powergate_uvd = &ci_dpm_powergate_uvd,
-	.set_fan_control_mode = &ci_dpm_set_fan_control_mode,
-	.get_fan_control_mode = &ci_dpm_get_fan_control_mode,
-	.set_fan_speed_percent = &ci_dpm_set_fan_speed_percent,
-	.get_fan_speed_percent = &ci_dpm_get_fan_speed_percent,
-	.print_clock_levels = ci_dpm_print_clock_levels,
-	.force_clock_level = ci_dpm_force_clock_level,
-	.get_sclk_od = ci_dpm_get_sclk_od,
-	.set_sclk_od = ci_dpm_set_sclk_od,
-	.get_mclk_od = ci_dpm_get_mclk_od,
-	.set_mclk_od = ci_dpm_set_mclk_od,
-	.check_state_equal = ci_check_state_equal,
-	.get_vce_clock_state = amdgpu_get_vce_clock_state,
-	.get_power_profile_state = ci_dpm_get_power_profile_state,
-	.set_power_profile_state = ci_dpm_set_power_profile_state,
-	.reset_power_profile_state = ci_dpm_reset_power_profile_state,
-	.switch_power_profile = ci_dpm_switch_power_profile,
-	.read_sensor = ci_dpm_read_sensor,
-};
-
-static void ci_dpm_set_dpm_funcs(struct amdgpu_device *adev)
-{
-	if (adev->pm.funcs == NULL)
-		adev->pm.funcs = &ci_dpm_funcs;
-}
-
-static const struct amdgpu_irq_src_funcs ci_dpm_irq_funcs = {
-	.set = ci_dpm_set_interrupt_state,
-	.process = ci_dpm_process_interrupt,
-};
-
-static void ci_dpm_set_irq_funcs(struct amdgpu_device *adev)
-{
-	adev->pm.dpm.thermal.irq.num_types = AMDGPU_THERMAL_IRQ_LAST;
-	adev->pm.dpm.thermal.irq.funcs = &ci_dpm_irq_funcs;
-}
diff --git a/drivers/gpu/drm/amd/amdgpu/ci_dpm.h b/drivers/gpu/drm/amd/amdgpu/ci_dpm.h
deleted file mode 100644
index 84cbc9c45f4d..000000000000
--- a/drivers/gpu/drm/amd/amdgpu/ci_dpm.h
+++ /dev/null
@@ -1,356 +0,0 @@
-/*
- * Copyright 2013 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-#ifndef __CI_DPM_H__
-#define __CI_DPM_H__
-
-#include "amdgpu_atombios.h"
-#include "ppsmc.h"
-
-#define SMU__NUM_SCLK_DPM_STATE  8
-#define SMU__NUM_MCLK_DPM_LEVELS 6
-#define SMU__NUM_LCLK_DPM_LEVELS 8
-#define SMU__NUM_PCIE_DPM_LEVELS 8
-#include "smu7_discrete.h"
-
-#define CISLANDS_MAX_HARDWARE_POWERLEVELS 2
-
-#define CISLANDS_UNUSED_GPIO_PIN 0x7F
-
-struct ci_pl {
-	u32 mclk;
-	u32 sclk;
-	enum amdgpu_pcie_gen pcie_gen;
-	u16 pcie_lane;
-};
-
-struct ci_ps {
-	u16 performance_level_count;
-	bool dc_compatible;
-	u32 sclk_t;
-	struct ci_pl performance_levels[CISLANDS_MAX_HARDWARE_POWERLEVELS];
-};
-
-struct ci_dpm_level {
-	bool enabled;
-	u32 value;
-	u32 param1;
-};
-
-#define CISLAND_MAX_DEEPSLEEP_DIVIDER_ID 5
-#define MAX_REGULAR_DPM_NUMBER 8
-#define CISLAND_MINIMUM_ENGINE_CLOCK 800
-
-struct ci_single_dpm_table {
-	u32 count;
-	struct ci_dpm_level dpm_levels[MAX_REGULAR_DPM_NUMBER];
-};
-
-struct ci_dpm_table {
-	struct ci_single_dpm_table sclk_table;
-	struct ci_single_dpm_table mclk_table;
-	struct ci_single_dpm_table pcie_speed_table;
-	struct ci_single_dpm_table vddc_table;
-	struct ci_single_dpm_table vddci_table;
-	struct ci_single_dpm_table mvdd_table;
-};
-
-struct ci_mc_reg_entry {
-	u32 mclk_max;
-	u32 mc_data[SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE];
-};
-
-struct ci_mc_reg_table {
-	u8 last;
-	u8 num_entries;
-	u16 valid_flag;
-	struct ci_mc_reg_entry mc_reg_table_entry[MAX_AC_TIMING_ENTRIES];
-	SMU7_Discrete_MCRegisterAddress mc_reg_address[SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE];
-};
-
-struct ci_ulv_parm
-{
-	bool supported;
-	u32 cg_ulv_parameter;
-	u32 volt_change_delay;
-	struct ci_pl pl;
-};
-
-#define CISLANDS_MAX_LEAKAGE_COUNT  8
-
-struct ci_leakage_voltage {
-	u16 count;
-	u16 leakage_id[CISLANDS_MAX_LEAKAGE_COUNT];
-	u16 actual_voltage[CISLANDS_MAX_LEAKAGE_COUNT];
-};
-
-struct ci_dpm_level_enable_mask {
-	u32 uvd_dpm_enable_mask;
-	u32 vce_dpm_enable_mask;
-	u32 acp_dpm_enable_mask;
-	u32 samu_dpm_enable_mask;
-	u32 sclk_dpm_enable_mask;
-	u32 mclk_dpm_enable_mask;
-	u32 pcie_dpm_enable_mask;
-};
-
-struct ci_vbios_boot_state
-{
-	u16 mvdd_bootup_value;
-	u16 vddc_bootup_value;
-	u16 vddci_bootup_value;
-	u32 sclk_bootup_value;
-	u32 mclk_bootup_value;
-	u16 pcie_gen_bootup_value;
-	u16 pcie_lane_bootup_value;
-};
-
-struct ci_clock_registers {
-	u32 cg_spll_func_cntl;
-	u32 cg_spll_func_cntl_2;
-	u32 cg_spll_func_cntl_3;
-	u32 cg_spll_func_cntl_4;
-	u32 cg_spll_spread_spectrum;
-	u32 cg_spll_spread_spectrum_2;
-	u32 dll_cntl;
-	u32 mclk_pwrmgt_cntl;
-	u32 mpll_ad_func_cntl;
-	u32 mpll_dq_func_cntl;
-	u32 mpll_func_cntl;
-	u32 mpll_func_cntl_1;
-	u32 mpll_func_cntl_2;
-	u32 mpll_ss1;
-	u32 mpll_ss2;
-};
-
-struct ci_thermal_temperature_setting {
-	s32 temperature_low;
-	s32 temperature_high;
-	s32 temperature_shutdown;
-};
-
-struct ci_pcie_perf_range {
-	u16 max;
-	u16 min;
-};
-
-enum ci_pt_config_reg_type {
-	CISLANDS_CONFIGREG_MMR = 0,
-	CISLANDS_CONFIGREG_SMC_IND,
-	CISLANDS_CONFIGREG_DIDT_IND,
-	CISLANDS_CONFIGREG_CACHE,
-	CISLANDS_CONFIGREG_MAX
-};
-
-#define POWERCONTAINMENT_FEATURE_BAPM            0x00000001
-#define POWERCONTAINMENT_FEATURE_TDCLimit        0x00000002
-#define POWERCONTAINMENT_FEATURE_PkgPwrLimit     0x00000004
-
-struct ci_pt_config_reg {
-	u32 offset;
-	u32 mask;
-	u32 shift;
-	u32 value;
-	enum ci_pt_config_reg_type type;
-};
-
-struct ci_pt_defaults {
-	u8 svi_load_line_en;
-	u8 svi_load_line_vddc;
-	u8 tdc_vddc_throttle_release_limit_perc;
-	u8 tdc_mawt;
-	u8 tdc_waterfall_ctl;
-	u8 dte_ambient_temp_base;
-	u32 display_cac;
-	u32 bapm_temp_gradient;
-	u16 bapmti_r[SMU7_DTE_ITERATIONS * SMU7_DTE_SOURCES * SMU7_DTE_SINKS];
-	u16 bapmti_rc[SMU7_DTE_ITERATIONS * SMU7_DTE_SOURCES * SMU7_DTE_SINKS];
-};
-
-#define DPMTABLE_OD_UPDATE_SCLK     0x00000001
-#define DPMTABLE_OD_UPDATE_MCLK     0x00000002
-#define DPMTABLE_UPDATE_SCLK        0x00000004
-#define DPMTABLE_UPDATE_MCLK        0x00000008
-
-struct ci_power_info {
-	struct ci_dpm_table dpm_table;
-	struct ci_dpm_table golden_dpm_table;
-	u32 voltage_control;
-	u32 mvdd_control;
-	u32 vddci_control;
-	u32 active_auto_throttle_sources;
-	struct ci_clock_registers clock_registers;
-	u16 acpi_vddc;
-	u16 acpi_vddci;
-	enum amdgpu_pcie_gen force_pcie_gen;
-	enum amdgpu_pcie_gen acpi_pcie_gen;
-	struct ci_leakage_voltage vddc_leakage;
-	struct ci_leakage_voltage vddci_leakage;
-	u16 max_vddc_in_pp_table;
-	u16 min_vddc_in_pp_table;
-	u16 max_vddci_in_pp_table;
-	u16 min_vddci_in_pp_table;
-	u32 mclk_strobe_mode_threshold;
-	u32 mclk_stutter_mode_threshold;
-	u32 mclk_edc_enable_threshold;
-	u32 mclk_edc_wr_enable_threshold;
-	struct ci_vbios_boot_state vbios_boot_state;
-	/* smc offsets */
-	u32 sram_end;
-	u32 dpm_table_start;
-	u32 soft_regs_start;
-	u32 mc_reg_table_start;
-	u32 fan_table_start;
-	u32 arb_table_start;
-	/* smc tables */
-	SMU7_Discrete_DpmTable smc_state_table;
-	SMU7_Discrete_MCRegisters smc_mc_reg_table;
-	SMU7_Discrete_PmFuses smc_powertune_table;
-	/* other stuff */
-	struct ci_mc_reg_table mc_reg_table;
-	struct atom_voltage_table vddc_voltage_table;
-	struct atom_voltage_table vddci_voltage_table;
-	struct atom_voltage_table mvdd_voltage_table;
-	struct ci_ulv_parm ulv;
-	u32 power_containment_features;
-	const struct ci_pt_defaults *powertune_defaults;
-	u32 dte_tj_offset;
-	bool vddc_phase_shed_control;
-	struct ci_thermal_temperature_setting thermal_temp_setting;
-	struct ci_dpm_level_enable_mask dpm_level_enable_mask;
-	u32 need_update_smu7_dpm_table;
-	u32 sclk_dpm_key_disabled;
-	u32 mclk_dpm_key_disabled;
-	u32 pcie_dpm_key_disabled;
-	u32 thermal_sclk_dpm_enabled;
-	struct ci_pcie_perf_range pcie_gen_performance;
-	struct ci_pcie_perf_range pcie_lane_performance;
-	struct ci_pcie_perf_range pcie_gen_powersaving;
-	struct ci_pcie_perf_range pcie_lane_powersaving;
-	u32 activity_target[SMU7_MAX_LEVELS_GRAPHICS];
-	u32 mclk_activity_target;
-	u32 low_sclk_interrupt_t;
-	u32 last_mclk_dpm_enable_mask;
-	u32 sys_pcie_mask;
-	/* caps */
-	bool caps_power_containment;
-	bool caps_cac;
-	bool caps_sq_ramping;
-	bool caps_db_ramping;
-	bool caps_td_ramping;
-	bool caps_tcp_ramping;
-	bool caps_fps;
-	bool caps_sclk_ds;
-	bool caps_sclk_ss_support;
-	bool caps_mclk_ss_support;
-	bool caps_uvd_dpm;
-	bool caps_vce_dpm;
-	bool caps_samu_dpm;
-	bool caps_acp_dpm;
-	bool caps_automatic_dc_transition;
-	bool caps_sclk_throttle_low_notification;
-	bool caps_dynamic_ac_timing;
-	bool caps_od_fuzzy_fan_control_support;
-	/* flags */
-	bool thermal_protection;
-	bool pcie_performance_request;
-	bool dynamic_ss;
-	bool dll_default_on;
-	bool cac_enabled;
-	bool uvd_enabled;
-	bool battery_state;
-	bool pspp_notify_required;
-	bool enable_bapm_feature;
-	bool enable_tdc_limit_feature;
-	bool enable_pkg_pwr_tracking_feature;
-	bool use_pcie_performance_levels;
-	bool use_pcie_powersaving_levels;
-	bool uvd_power_gated;
-	/* driver states */
-	struct amdgpu_ps current_rps;
-	struct ci_ps current_ps;
-	struct amdgpu_ps requested_rps;
-	struct ci_ps requested_ps;
-	/* fan control */
-	bool fan_ctrl_is_in_default_mode;
-	bool fan_is_controlled_by_smc;
-	u32 t_min;
-	u32 fan_ctrl_default_mode;
-
-	/* power profile */
-	struct amd_pp_profile gfx_power_profile;
-	struct amd_pp_profile compute_power_profile;
-	struct amd_pp_profile default_gfx_power_profile;
-	struct amd_pp_profile default_compute_power_profile;
-	enum amd_pp_profile_type current_power_profile;
-};
-
-#define CISLANDS_VOLTAGE_CONTROL_NONE                   0x0
-#define CISLANDS_VOLTAGE_CONTROL_BY_GPIO                0x1
-#define CISLANDS_VOLTAGE_CONTROL_BY_SVID2               0x2
-
-#define CISLANDS_Q88_FORMAT_CONVERSION_UNIT             256
-
-#define CISLANDS_VRC_DFLT0                              0x3FFFC000
-#define CISLANDS_VRC_DFLT1                              0x000400
-#define CISLANDS_VRC_DFLT2                              0xC00080
-#define CISLANDS_VRC_DFLT3                              0xC00200
-#define CISLANDS_VRC_DFLT4                              0xC01680
-#define CISLANDS_VRC_DFLT5                              0xC00033
-#define CISLANDS_VRC_DFLT6                              0xC00033
-#define CISLANDS_VRC_DFLT7                              0x3FFFC000
-
-#define CISLANDS_CGULVPARAMETER_DFLT                    0x00040035
-#define CISLAND_TARGETACTIVITY_DFLT                     30
-#define CISLAND_MCLK_TARGETACTIVITY_DFLT                10
-
-#define PCIE_PERF_REQ_REMOVE_REGISTRY   0
-#define PCIE_PERF_REQ_FORCE_LOWPOWER    1
-#define PCIE_PERF_REQ_PECI_GEN1         2
-#define PCIE_PERF_REQ_PECI_GEN2         3
-#define PCIE_PERF_REQ_PECI_GEN3         4
-
-#define CISLANDS_SSTU_DFLT                               0
-#define CISLANDS_SST_DFLT                                0x00C8
-
-/* XXX are these ok? */
-#define CISLANDS_TEMP_RANGE_MIN (90 * 1000)
-#define CISLANDS_TEMP_RANGE_MAX (120 * 1000)
-
-int amdgpu_ci_copy_bytes_to_smc(struct amdgpu_device *adev,
-			 u32 smc_start_address,
-			 const u8 *src, u32 byte_count, u32 limit);
-void amdgpu_ci_start_smc(struct amdgpu_device *adev);
-void amdgpu_ci_reset_smc(struct amdgpu_device *adev);
-int amdgpu_ci_program_jump_on_start(struct amdgpu_device *adev);
-void amdgpu_ci_stop_smc_clock(struct amdgpu_device *adev);
-void amdgpu_ci_start_smc_clock(struct amdgpu_device *adev);
-bool amdgpu_ci_is_smc_running(struct amdgpu_device *adev);
-PPSMC_Result amdgpu_ci_send_msg_to_smc(struct amdgpu_device *adev, PPSMC_Msg msg);
-PPSMC_Result amdgpu_ci_wait_for_smc_inactive(struct amdgpu_device *adev);
-int amdgpu_ci_load_smc_ucode(struct amdgpu_device *adev, u32 limit);
-int amdgpu_ci_read_smc_sram_dword(struct amdgpu_device *adev,
-			   u32 smc_address, u32 *value, u32 limit);
-int amdgpu_ci_write_smc_sram_dword(struct amdgpu_device *adev,
-			    u32 smc_address, u32 value, u32 limit);
-
-#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/ci_smc.c b/drivers/gpu/drm/amd/amdgpu/ci_smc.c
deleted file mode 100644
index b8ba51e045b5..000000000000
--- a/drivers/gpu/drm/amd/amdgpu/ci_smc.c
+++ /dev/null
@@ -1,279 +0,0 @@
-/*
- * Copyright 2011 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- * Authors: Alex Deucher
- */
-
-#include <linux/firmware.h>
-#include <drm/drmP.h>
-#include "amdgpu.h"
-#include "cikd.h"
-#include "ppsmc.h"
-#include "amdgpu_ucode.h"
-#include "ci_dpm.h"
-
-#include "smu/smu_7_0_1_d.h"
-#include "smu/smu_7_0_1_sh_mask.h"
-
-static int ci_set_smc_sram_address(struct amdgpu_device *adev,
-				   u32 smc_address, u32 limit)
-{
-	if (smc_address & 3)
-		return -EINVAL;
-	if ((smc_address + 3) > limit)
-		return -EINVAL;
-
-	WREG32(mmSMC_IND_INDEX_0, smc_address);
-	WREG32_P(mmSMC_IND_ACCESS_CNTL, 0, ~SMC_IND_ACCESS_CNTL__AUTO_INCREMENT_IND_0_MASK);
-
-	return 0;
-}
-
-int amdgpu_ci_copy_bytes_to_smc(struct amdgpu_device *adev,
-			 u32 smc_start_address,
-			 const u8 *src, u32 byte_count, u32 limit)
-{
-	unsigned long flags;
-	u32 data, original_data;
-	u32 addr;
-	u32 extra_shift;
-	int ret = 0;
-
-	if (smc_start_address & 3)
-		return -EINVAL;
-	if ((smc_start_address + byte_count) > limit)
-		return -EINVAL;
-
-	addr = smc_start_address;
-
-	spin_lock_irqsave(&adev->smc_idx_lock, flags);
-	while (byte_count >= 4) {
-		/* SMC address space is BE */
-		data = (src[0] << 24) | (src[1] << 16) | (src[2] << 8) | src[3];
-
-		ret = ci_set_smc_sram_address(adev, addr, limit);
-		if (ret)
-			goto done;
-
-		WREG32(mmSMC_IND_DATA_0, data);
-
-		src += 4;
-		byte_count -= 4;
-		addr += 4;
-	}
-
-	/* RMW for the final bytes */
-	if (byte_count > 0) {
-		data = 0;
-
-		ret = ci_set_smc_sram_address(adev, addr, limit);
-		if (ret)
-			goto done;
-
-		original_data = RREG32(mmSMC_IND_DATA_0);
-
-		extra_shift = 8 * (4 - byte_count);
-
-		while (byte_count > 0) {
-			data = (data << 8) + *src++;
-			byte_count--;
-		}
-
-		data <<= extra_shift;
-
-		data |= (original_data & ~((~0UL) << extra_shift));
-
-		ret = ci_set_smc_sram_address(adev, addr, limit);
-		if (ret)
-			goto done;
-
-		WREG32(mmSMC_IND_DATA_0, data);
-	}
-
-done:
-	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
-
-	return ret;
-}
-
-void amdgpu_ci_start_smc(struct amdgpu_device *adev)
-{
-	u32 tmp = RREG32_SMC(ixSMC_SYSCON_RESET_CNTL);
-
-	tmp &= ~SMC_SYSCON_RESET_CNTL__rst_reg_MASK;
-	WREG32_SMC(ixSMC_SYSCON_RESET_CNTL, tmp);
-}
-
-void amdgpu_ci_reset_smc(struct amdgpu_device *adev)
-{
-	u32 tmp = RREG32_SMC(ixSMC_SYSCON_RESET_CNTL);
-
-	tmp |= SMC_SYSCON_RESET_CNTL__rst_reg_MASK;
-	WREG32_SMC(ixSMC_SYSCON_RESET_CNTL, tmp);
-}
-
-int amdgpu_ci_program_jump_on_start(struct amdgpu_device *adev)
-{
-	static u8 data[] = { 0xE0, 0x00, 0x80, 0x40 };
-
-	return amdgpu_ci_copy_bytes_to_smc(adev, 0x0, data, 4, sizeof(data)+1);
-}
-
-void amdgpu_ci_stop_smc_clock(struct amdgpu_device *adev)
-{
-	u32 tmp = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
-
-	tmp |= SMC_SYSCON_CLOCK_CNTL_0__ck_disable_MASK;
-
-	WREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0, tmp);
-}
-
-void amdgpu_ci_start_smc_clock(struct amdgpu_device *adev)
-{
-	u32 tmp = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
-
-	tmp &= ~SMC_SYSCON_CLOCK_CNTL_0__ck_disable_MASK;
-
-	WREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0, tmp);
-}
-
-bool amdgpu_ci_is_smc_running(struct amdgpu_device *adev)
-{
-	u32 clk = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
-	u32 pc_c = RREG32_SMC(ixSMC_PC_C);
-
-	if (!(clk & SMC_SYSCON_CLOCK_CNTL_0__ck_disable_MASK) && (0x20100 <= pc_c))
-		return true;
-
-	return false;
-}
-
-PPSMC_Result amdgpu_ci_send_msg_to_smc(struct amdgpu_device *adev, PPSMC_Msg msg)
-{
-	u32 tmp;
-	int i;
-
-	if (!amdgpu_ci_is_smc_running(adev))
-		return PPSMC_Result_Failed;
-
-	WREG32(mmSMC_MESSAGE_0, msg);
-
-	for (i = 0; i < adev->usec_timeout; i++) {
-		tmp = RREG32(mmSMC_RESP_0);
-		if (tmp != 0)
-			break;
-		udelay(1);
-	}
-	tmp = RREG32(mmSMC_RESP_0);
-
-	return (PPSMC_Result)tmp;
-}
-
-PPSMC_Result amdgpu_ci_wait_for_smc_inactive(struct amdgpu_device *adev)
-{
-	u32 tmp;
-	int i;
-
-	if (!amdgpu_ci_is_smc_running(adev))
-		return PPSMC_Result_OK;
-
-	for (i = 0; i < adev->usec_timeout; i++) {
-		tmp = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
-		if ((tmp & SMC_SYSCON_CLOCK_CNTL_0__cken_MASK) == 0)
-			break;
-		udelay(1);
-	}
-
-	return PPSMC_Result_OK;
-}
-
-int amdgpu_ci_load_smc_ucode(struct amdgpu_device *adev, u32 limit)
-{
-	const struct smc_firmware_header_v1_0 *hdr;
-	unsigned long flags;
-	u32 ucode_start_address;
-	u32 ucode_size;
-	const u8 *src;
-	u32 data;
-
-	if (!adev->pm.fw)
-		return -EINVAL;
-
-	hdr = (const struct smc_firmware_header_v1_0 *)adev->pm.fw->data;
-	amdgpu_ucode_print_smc_hdr(&hdr->header);
-
-	adev->pm.fw_version = le32_to_cpu(hdr->header.ucode_version);
-	ucode_start_address = le32_to_cpu(hdr->ucode_start_addr);
-	ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes);
-	src = (const u8 *)
-		(adev->pm.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
-
-	if (ucode_size & 3)
-		return -EINVAL;
-
-	spin_lock_irqsave(&adev->smc_idx_lock, flags);
-	WREG32(mmSMC_IND_INDEX_0, ucode_start_address);
-	WREG32_P(mmSMC_IND_ACCESS_CNTL, SMC_IND_ACCESS_CNTL__AUTO_INCREMENT_IND_0_MASK,
-		~SMC_IND_ACCESS_CNTL__AUTO_INCREMENT_IND_0_MASK);
-	while (ucode_size >= 4) {
-		/* SMC address space is BE */
-		data = (src[0] << 24) | (src[1] << 16) | (src[2] << 8) | src[3];
-
-		WREG32(mmSMC_IND_DATA_0, data);
-
-		src += 4;
-		ucode_size -= 4;
-	}
-	WREG32_P(mmSMC_IND_ACCESS_CNTL, 0, ~SMC_IND_ACCESS_CNTL__AUTO_INCREMENT_IND_0_MASK);
-	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
-
-	return 0;
-}
-
-int amdgpu_ci_read_smc_sram_dword(struct amdgpu_device *adev,
-			   u32 smc_address, u32 *value, u32 limit)
-{
-	unsigned long flags;
-	int ret;
-
-	spin_lock_irqsave(&adev->smc_idx_lock, flags);
-	ret = ci_set_smc_sram_address(adev, smc_address, limit);
-	if (ret == 0)
-		*value = RREG32(mmSMC_IND_DATA_0);
-	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
-
-	return ret;
-}
-
-int amdgpu_ci_write_smc_sram_dword(struct amdgpu_device *adev,
-			    u32 smc_address, u32 value, u32 limit)
-{
-	unsigned long flags;
-	int ret;
-
-	spin_lock_irqsave(&adev->smc_idx_lock, flags);
-	ret = ci_set_smc_sram_address(adev, smc_address, limit);
-	if (ret == 0)
-		WREG32(mmSMC_IND_DATA_0, value);
-	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
-
-	return ret;
-}
diff --git a/drivers/gpu/drm/amd/amdgpu/cik.c b/drivers/gpu/drm/amd/amdgpu/cik.c
index 567c4a5cf90c..9cd63b4177bf 100644
--- a/drivers/gpu/drm/amd/amdgpu/cik.c
+++ b/drivers/gpu/drm/amd/amdgpu/cik.c
@@ -24,7 +24,10 @@
 #include <linux/firmware.h>
 #include <linux/slab.h>
 #include <linux/module.h>
-#include <drm/drmP.h>
+#include <linux/pci.h>
+
+#include <drm/amdgpu_drm.h>
+
 #include "amdgpu.h"
 #include "amdgpu_atombios.h"
 #include "amdgpu_ih.h"
@@ -65,9 +68,83 @@
 #include "oss/oss_2_0_d.h"
 #include "oss/oss_2_0_sh_mask.h"
 
+#include "amdgpu_dm.h"
 #include "amdgpu_amdkfd.h"
-#include "amdgpu_powerplay.h"
-#include "dce_virtual.h"
+#include "amdgpu_vkms.h"
+
+static const struct amdgpu_video_codec_info cik_video_codecs_encode_array[] =
+{
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC,
+		.max_width = 2048,
+		.max_height = 1152,
+		.max_pixels_per_frame = 2048 * 1152,
+		.max_level = 0,
+	},
+};
+
+static const struct amdgpu_video_codecs cik_video_codecs_encode =
+{
+	.codec_count = ARRAY_SIZE(cik_video_codecs_encode_array),
+	.codec_array = cik_video_codecs_encode_array,
+};
+
+static const struct amdgpu_video_codec_info cik_video_codecs_decode_array[] =
+{
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2,
+		.max_width = 2048,
+		.max_height = 1152,
+		.max_pixels_per_frame = 2048 * 1152,
+		.max_level = 3,
+	},
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4,
+		.max_width = 2048,
+		.max_height = 1152,
+		.max_pixels_per_frame = 2048 * 1152,
+		.max_level = 5,
+	},
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC,
+		.max_width = 2048,
+		.max_height = 1152,
+		.max_pixels_per_frame = 2048 * 1152,
+		.max_level = 41,
+	},
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1,
+		.max_width = 2048,
+		.max_height = 1152,
+		.max_pixels_per_frame = 2048 * 1152,
+		.max_level = 4,
+	},
+};
+
+static const struct amdgpu_video_codecs cik_video_codecs_decode =
+{
+	.codec_count = ARRAY_SIZE(cik_video_codecs_decode_array),
+	.codec_array = cik_video_codecs_decode_array,
+};
+
+static int cik_query_video_codecs(struct amdgpu_device *adev, bool encode,
+				  const struct amdgpu_video_codecs **codecs)
+{
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+	case CHIP_HAWAII:
+	case CHIP_KAVERI:
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+		if (encode)
+			*codecs = &cik_video_codecs_encode;
+		else
+			*codecs = &cik_video_codecs_decode;
+		return 0;
+	default:
+		return -EINVAL;
+	}
+}
 
 /*
  * Indirect registers accessor
@@ -754,74 +831,74 @@ static void cik_init_golden_registers(struct amdgpu_device *adev)
 
 	switch (adev->asic_type) {
 	case CHIP_BONAIRE:
-		amdgpu_program_register_sequence(adev,
-						 bonaire_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(bonaire_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 bonaire_golden_registers,
-						 (const u32)ARRAY_SIZE(bonaire_golden_registers));
-		amdgpu_program_register_sequence(adev,
-						 bonaire_golden_common_registers,
-						 (const u32)ARRAY_SIZE(bonaire_golden_common_registers));
-		amdgpu_program_register_sequence(adev,
-						 bonaire_golden_spm_registers,
-						 (const u32)ARRAY_SIZE(bonaire_golden_spm_registers));
+		amdgpu_device_program_register_sequence(adev,
+							bonaire_mgcg_cgcg_init,
+							ARRAY_SIZE(bonaire_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							bonaire_golden_registers,
+							ARRAY_SIZE(bonaire_golden_registers));
+		amdgpu_device_program_register_sequence(adev,
+							bonaire_golden_common_registers,
+							ARRAY_SIZE(bonaire_golden_common_registers));
+		amdgpu_device_program_register_sequence(adev,
+							bonaire_golden_spm_registers,
+							ARRAY_SIZE(bonaire_golden_spm_registers));
 		break;
 	case CHIP_KABINI:
-		amdgpu_program_register_sequence(adev,
-						 kalindi_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(kalindi_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 kalindi_golden_registers,
-						 (const u32)ARRAY_SIZE(kalindi_golden_registers));
-		amdgpu_program_register_sequence(adev,
-						 kalindi_golden_common_registers,
-						 (const u32)ARRAY_SIZE(kalindi_golden_common_registers));
-		amdgpu_program_register_sequence(adev,
-						 kalindi_golden_spm_registers,
-						 (const u32)ARRAY_SIZE(kalindi_golden_spm_registers));
+		amdgpu_device_program_register_sequence(adev,
+							kalindi_mgcg_cgcg_init,
+							ARRAY_SIZE(kalindi_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							kalindi_golden_registers,
+							ARRAY_SIZE(kalindi_golden_registers));
+		amdgpu_device_program_register_sequence(adev,
+							kalindi_golden_common_registers,
+							ARRAY_SIZE(kalindi_golden_common_registers));
+		amdgpu_device_program_register_sequence(adev,
+							kalindi_golden_spm_registers,
+							ARRAY_SIZE(kalindi_golden_spm_registers));
 		break;
 	case CHIP_MULLINS:
-		amdgpu_program_register_sequence(adev,
-						 kalindi_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(kalindi_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 godavari_golden_registers,
-						 (const u32)ARRAY_SIZE(godavari_golden_registers));
-		amdgpu_program_register_sequence(adev,
-						 kalindi_golden_common_registers,
-						 (const u32)ARRAY_SIZE(kalindi_golden_common_registers));
-		amdgpu_program_register_sequence(adev,
-						 kalindi_golden_spm_registers,
-						 (const u32)ARRAY_SIZE(kalindi_golden_spm_registers));
+		amdgpu_device_program_register_sequence(adev,
+							kalindi_mgcg_cgcg_init,
+							ARRAY_SIZE(kalindi_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							godavari_golden_registers,
+							ARRAY_SIZE(godavari_golden_registers));
+		amdgpu_device_program_register_sequence(adev,
+							kalindi_golden_common_registers,
+							ARRAY_SIZE(kalindi_golden_common_registers));
+		amdgpu_device_program_register_sequence(adev,
+							kalindi_golden_spm_registers,
+							ARRAY_SIZE(kalindi_golden_spm_registers));
 		break;
 	case CHIP_KAVERI:
-		amdgpu_program_register_sequence(adev,
-						 spectre_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(spectre_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 spectre_golden_registers,
-						 (const u32)ARRAY_SIZE(spectre_golden_registers));
-		amdgpu_program_register_sequence(adev,
-						 spectre_golden_common_registers,
-						 (const u32)ARRAY_SIZE(spectre_golden_common_registers));
-		amdgpu_program_register_sequence(adev,
-						 spectre_golden_spm_registers,
-						 (const u32)ARRAY_SIZE(spectre_golden_spm_registers));
+		amdgpu_device_program_register_sequence(adev,
+							spectre_mgcg_cgcg_init,
+							ARRAY_SIZE(spectre_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							spectre_golden_registers,
+							ARRAY_SIZE(spectre_golden_registers));
+		amdgpu_device_program_register_sequence(adev,
+							spectre_golden_common_registers,
+							ARRAY_SIZE(spectre_golden_common_registers));
+		amdgpu_device_program_register_sequence(adev,
+							spectre_golden_spm_registers,
+							ARRAY_SIZE(spectre_golden_spm_registers));
 		break;
 	case CHIP_HAWAII:
-		amdgpu_program_register_sequence(adev,
-						 hawaii_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(hawaii_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 hawaii_golden_registers,
-						 (const u32)ARRAY_SIZE(hawaii_golden_registers));
-		amdgpu_program_register_sequence(adev,
-						 hawaii_golden_common_registers,
-						 (const u32)ARRAY_SIZE(hawaii_golden_common_registers));
-		amdgpu_program_register_sequence(adev,
-						 hawaii_golden_spm_registers,
-						 (const u32)ARRAY_SIZE(hawaii_golden_spm_registers));
+		amdgpu_device_program_register_sequence(adev,
+							hawaii_mgcg_cgcg_init,
+							ARRAY_SIZE(hawaii_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							hawaii_golden_registers,
+							ARRAY_SIZE(hawaii_golden_registers));
+		amdgpu_device_program_register_sequence(adev,
+							hawaii_golden_common_registers,
+							ARRAY_SIZE(hawaii_golden_common_registers));
+		amdgpu_device_program_register_sequence(adev,
+							hawaii_golden_spm_registers,
+							ARRAY_SIZE(hawaii_golden_spm_registers));
 		break;
 	default:
 		break;
@@ -965,6 +1042,25 @@ static bool cik_read_bios_from_rom(struct amdgpu_device *adev,
 
 static const struct amdgpu_allowed_register_entry cik_allowed_read_registers[] = {
 	{mmGRBM_STATUS},
+	{mmGRBM_STATUS2},
+	{mmGRBM_STATUS_SE0},
+	{mmGRBM_STATUS_SE1},
+	{mmGRBM_STATUS_SE2},
+	{mmGRBM_STATUS_SE3},
+	{mmSRBM_STATUS},
+	{mmSRBM_STATUS2},
+	{mmSDMA0_STATUS_REG + SDMA0_REGISTER_OFFSET},
+	{mmSDMA0_STATUS_REG + SDMA1_REGISTER_OFFSET},
+	{mmCP_STAT},
+	{mmCP_STALLED_STAT1},
+	{mmCP_STALLED_STAT2},
+	{mmCP_STALLED_STAT3},
+	{mmCP_CPF_BUSY_STAT},
+	{mmCP_CPF_STALLED_STAT1},
+	{mmCP_CPF_STATUS},
+	{mmCP_CPC_BUSY_STAT},
+	{mmCP_CPC_STALLED_STAT1},
+	{mmCP_CPC_STATUS},
 	{mmGB_ADDR_CONFIG},
 	{mmMC_ARB_RAMCFG},
 	{mmGB_TILE_MODE0},
@@ -1022,22 +1118,101 @@ static const struct amdgpu_allowed_register_entry cik_allowed_read_registers[] =
 	{mmPA_SC_RASTER_CONFIG_1, true},
 };
 
-static uint32_t cik_read_indexed_register(struct amdgpu_device *adev,
-					  u32 se_num, u32 sh_num,
-					  u32 reg_offset)
+
+static uint32_t cik_get_register_value(struct amdgpu_device *adev,
+				       bool indexed, u32 se_num,
+				       u32 sh_num, u32 reg_offset)
 {
-	uint32_t val;
+	if (indexed) {
+		uint32_t val;
+		unsigned se_idx = (se_num == 0xffffffff) ? 0 : se_num;
+		unsigned sh_idx = (sh_num == 0xffffffff) ? 0 : sh_num;
+
+		switch (reg_offset) {
+		case mmCC_RB_BACKEND_DISABLE:
+			return adev->gfx.config.rb_config[se_idx][sh_idx].rb_backend_disable;
+		case mmGC_USER_RB_BACKEND_DISABLE:
+			return adev->gfx.config.rb_config[se_idx][sh_idx].user_rb_backend_disable;
+		case mmPA_SC_RASTER_CONFIG:
+			return adev->gfx.config.rb_config[se_idx][sh_idx].raster_config;
+		case mmPA_SC_RASTER_CONFIG_1:
+			return adev->gfx.config.rb_config[se_idx][sh_idx].raster_config_1;
+		}
 
-	mutex_lock(&adev->grbm_idx_mutex);
-	if (se_num != 0xffffffff || sh_num != 0xffffffff)
-		amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff);
+		mutex_lock(&adev->grbm_idx_mutex);
+		if (se_num != 0xffffffff || sh_num != 0xffffffff)
+			amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff, 0);
 
-	val = RREG32(reg_offset);
+		val = RREG32(reg_offset);
 
-	if (se_num != 0xffffffff || sh_num != 0xffffffff)
-		amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
-	mutex_unlock(&adev->grbm_idx_mutex);
-	return val;
+		if (se_num != 0xffffffff || sh_num != 0xffffffff)
+			amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
+		mutex_unlock(&adev->grbm_idx_mutex);
+		return val;
+	} else {
+		unsigned idx;
+
+		switch (reg_offset) {
+		case mmGB_ADDR_CONFIG:
+			return adev->gfx.config.gb_addr_config;
+		case mmMC_ARB_RAMCFG:
+			return adev->gfx.config.mc_arb_ramcfg;
+		case mmGB_TILE_MODE0:
+		case mmGB_TILE_MODE1:
+		case mmGB_TILE_MODE2:
+		case mmGB_TILE_MODE3:
+		case mmGB_TILE_MODE4:
+		case mmGB_TILE_MODE5:
+		case mmGB_TILE_MODE6:
+		case mmGB_TILE_MODE7:
+		case mmGB_TILE_MODE8:
+		case mmGB_TILE_MODE9:
+		case mmGB_TILE_MODE10:
+		case mmGB_TILE_MODE11:
+		case mmGB_TILE_MODE12:
+		case mmGB_TILE_MODE13:
+		case mmGB_TILE_MODE14:
+		case mmGB_TILE_MODE15:
+		case mmGB_TILE_MODE16:
+		case mmGB_TILE_MODE17:
+		case mmGB_TILE_MODE18:
+		case mmGB_TILE_MODE19:
+		case mmGB_TILE_MODE20:
+		case mmGB_TILE_MODE21:
+		case mmGB_TILE_MODE22:
+		case mmGB_TILE_MODE23:
+		case mmGB_TILE_MODE24:
+		case mmGB_TILE_MODE25:
+		case mmGB_TILE_MODE26:
+		case mmGB_TILE_MODE27:
+		case mmGB_TILE_MODE28:
+		case mmGB_TILE_MODE29:
+		case mmGB_TILE_MODE30:
+		case mmGB_TILE_MODE31:
+			idx = (reg_offset - mmGB_TILE_MODE0);
+			return adev->gfx.config.tile_mode_array[idx];
+		case mmGB_MACROTILE_MODE0:
+		case mmGB_MACROTILE_MODE1:
+		case mmGB_MACROTILE_MODE2:
+		case mmGB_MACROTILE_MODE3:
+		case mmGB_MACROTILE_MODE4:
+		case mmGB_MACROTILE_MODE5:
+		case mmGB_MACROTILE_MODE6:
+		case mmGB_MACROTILE_MODE7:
+		case mmGB_MACROTILE_MODE8:
+		case mmGB_MACROTILE_MODE9:
+		case mmGB_MACROTILE_MODE10:
+		case mmGB_MACROTILE_MODE11:
+		case mmGB_MACROTILE_MODE12:
+		case mmGB_MACROTILE_MODE13:
+		case mmGB_MACROTILE_MODE14:
+		case mmGB_MACROTILE_MODE15:
+			idx = (reg_offset - mmGB_MACROTILE_MODE0);
+			return adev->gfx.config.macrotile_mode_array[idx];
+		default:
+			return RREG32(reg_offset);
+		}
+	}
 }
 
 static int cik_read_register(struct amdgpu_device *adev, u32 se_num,
@@ -1047,13 +1222,13 @@ static int cik_read_register(struct amdgpu_device *adev, u32 se_num,
 
 	*value = 0;
 	for (i = 0; i < ARRAY_SIZE(cik_allowed_read_registers); i++) {
+		bool indexed = cik_allowed_read_registers[i].grbm_indexed;
+
 		if (reg_offset != cik_allowed_read_registers[i].reg_offset)
 			continue;
 
-		*value = cik_allowed_read_registers[i].grbm_indexed ?
-			 cik_read_indexed_register(adev, se_num,
-						   sh_num, reg_offset) :
-			 RREG32(reg_offset);
+		*value = cik_get_register_value(adev, indexed, se_num, sh_num,
+						reg_offset);
 		return 0;
 	}
 	return -EINVAL;
@@ -1152,13 +1327,22 @@ static void kv_restore_regs_for_reset(struct amdgpu_device *adev,
 	WREG32(mmGMCON_RENG_EXECUTE, save->gmcon_reng_execute);
 }
 
-static int cik_gpu_pci_config_reset(struct amdgpu_device *adev)
+/**
+ * cik_asic_pci_config_reset - soft reset GPU
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Use PCI Config method to reset the GPU.
+ *
+ * Returns 0 for success.
+ */
+static int cik_asic_pci_config_reset(struct amdgpu_device *adev)
 {
 	struct kv_reset_save_regs kv_save = { 0 };
 	u32 i;
 	int r = -EINVAL;
 
-	dev_info(adev->dev, "GPU pci config reset\n");
+	amdgpu_atombios_scratch_regs_engine_hung(adev, true);
 
 	if (adev->flags & AMD_IS_APU)
 		kv_save_regs_for_reset(adev, &kv_save);
@@ -1166,7 +1350,7 @@ static int cik_gpu_pci_config_reset(struct amdgpu_device *adev)
 	/* disable BM */
 	pci_clear_master(adev->pdev);
 	/* reset */
-	amdgpu_pci_config_reset(adev);
+	amdgpu_device_pci_config_reset(adev);
 
 	udelay(100);
 
@@ -1186,9 +1370,51 @@ static int cik_gpu_pci_config_reset(struct amdgpu_device *adev)
 	if (adev->flags & AMD_IS_APU)
 		kv_restore_regs_for_reset(adev, &kv_save);
 
+	amdgpu_atombios_scratch_regs_engine_hung(adev, false);
+
 	return r;
 }
 
+static int cik_asic_supports_baco(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+	case CHIP_HAWAII:
+		return amdgpu_dpm_is_baco_supported(adev);
+	default:
+		return 0;
+	}
+}
+
+static enum amd_reset_method
+cik_asic_reset_method(struct amdgpu_device *adev)
+{
+	bool baco_reset;
+
+	if (amdgpu_reset_method == AMD_RESET_METHOD_LEGACY ||
+	    amdgpu_reset_method == AMD_RESET_METHOD_BACO)
+		return amdgpu_reset_method;
+
+	if (amdgpu_reset_method != -1)
+		dev_warn(adev->dev, "Specified reset:%d isn't supported, using AUTO instead.\n",
+				  amdgpu_reset_method);
+
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+	case CHIP_HAWAII:
+		baco_reset = cik_asic_supports_baco(adev);
+		break;
+	default:
+		baco_reset = false;
+		break;
+	}
+
+	if (baco_reset)
+		return AMD_RESET_METHOD_BACO;
+	else
+		return AMD_RESET_METHOD_LEGACY;
+}
+
 /**
  * cik_asic_reset - soft reset GPU
  *
@@ -1202,11 +1428,17 @@ static int cik_asic_reset(struct amdgpu_device *adev)
 {
 	int r;
 
-	amdgpu_atombios_scratch_regs_engine_hung(adev, true);
-
-	r = cik_gpu_pci_config_reset(adev);
+	/* APUs don't have full asic reset */
+	if (adev->flags & AMD_IS_APU)
+		return 0;
 
-	amdgpu_atombios_scratch_regs_engine_hung(adev, false);
+	if (cik_asic_reset_method(adev) == AMD_RESET_METHOD_BACO) {
+		dev_info(adev->dev, "BACO reset\n");
+		r = amdgpu_dpm_baco_reset(adev);
+	} else {
+		dev_info(adev->dev, "PCI CONFIG reset\n");
+		r = cik_asic_pci_config_reset(adev);
+	}
 
 	return r;
 }
@@ -1298,7 +1530,6 @@ static int cik_set_vce_clocks(struct amdgpu_device *adev, u32 evclk, u32 ecclk)
 static void cik_pcie_gen3_enable(struct amdgpu_device *adev)
 {
 	struct pci_dev *root = adev->pdev->bus->self;
-	int bridge_pos, gpu_pos;
 	u32 speed_cntl, current_data_rate;
 	int i;
 	u16 tmp16;
@@ -1333,12 +1564,7 @@ static void cik_pcie_gen3_enable(struct amdgpu_device *adev)
 		DRM_INFO("enabling PCIE gen 2 link speeds, disable with amdgpu.pcie_gen2=0\n");
 	}
 
-	bridge_pos = pci_pcie_cap(root);
-	if (!bridge_pos)
-		return;
-
-	gpu_pos = pci_pcie_cap(adev->pdev);
-	if (!gpu_pos)
+	if (!pci_is_pcie(root) || !pci_is_pcie(adev->pdev))
 		return;
 
 	if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3) {
@@ -1348,14 +1574,8 @@ static void cik_pcie_gen3_enable(struct amdgpu_device *adev)
 			u16 bridge_cfg2, gpu_cfg2;
 			u32 max_lw, current_lw, tmp;
 
-			pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL, &bridge_cfg);
-			pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, &gpu_cfg);
-
-			tmp16 = bridge_cfg | PCI_EXP_LNKCTL_HAWD;
-			pci_write_config_word(root, bridge_pos + PCI_EXP_LNKCTL, tmp16);
-
-			tmp16 = gpu_cfg | PCI_EXP_LNKCTL_HAWD;
-			pci_write_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, tmp16);
+			pcie_capability_set_word(root, PCI_EXP_LNKCTL, PCI_EXP_LNKCTL_HAWD);
+			pcie_capability_set_word(adev->pdev, PCI_EXP_LNKCTL, PCI_EXP_LNKCTL_HAWD);
 
 			tmp = RREG32_PCIE(ixPCIE_LC_STATUS1);
 			max_lw = (tmp & PCIE_LC_STATUS1__LC_DETECTED_LINK_WIDTH_MASK) >>
@@ -1379,15 +1599,23 @@ static void cik_pcie_gen3_enable(struct amdgpu_device *adev)
 
 			for (i = 0; i < 10; i++) {
 				/* check status */
-				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_DEVSTA, &tmp16);
+				pcie_capability_read_word(adev->pdev,
+							  PCI_EXP_DEVSTA,
+							  &tmp16);
 				if (tmp16 & PCI_EXP_DEVSTA_TRPND)
 					break;
 
-				pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL, &bridge_cfg);
-				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, &gpu_cfg);
+				pcie_capability_read_word(root, PCI_EXP_LNKCTL,
+							  &bridge_cfg);
+				pcie_capability_read_word(adev->pdev,
+							  PCI_EXP_LNKCTL,
+							  &gpu_cfg);
 
-				pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL2, &bridge_cfg2);
-				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, &gpu_cfg2);
+				pcie_capability_read_word(root, PCI_EXP_LNKCTL2,
+							  &bridge_cfg2);
+				pcie_capability_read_word(adev->pdev,
+							  PCI_EXP_LNKCTL2,
+							  &gpu_cfg2);
 
 				tmp = RREG32_PCIE(ixPCIE_LC_CNTL4);
 				tmp |= PCIE_LC_CNTL4__LC_SET_QUIESCE_MASK;
@@ -1397,29 +1625,31 @@ static void cik_pcie_gen3_enable(struct amdgpu_device *adev)
 				tmp |= PCIE_LC_CNTL4__LC_REDO_EQ_MASK;
 				WREG32_PCIE(ixPCIE_LC_CNTL4, tmp);
 
-				mdelay(100);
+				msleep(100);
 
 				/* linkctl */
-				pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL, &tmp16);
-				tmp16 &= ~PCI_EXP_LNKCTL_HAWD;
-				tmp16 |= (bridge_cfg & PCI_EXP_LNKCTL_HAWD);
-				pci_write_config_word(root, bridge_pos + PCI_EXP_LNKCTL, tmp16);
-
-				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, &tmp16);
-				tmp16 &= ~PCI_EXP_LNKCTL_HAWD;
-				tmp16 |= (gpu_cfg & PCI_EXP_LNKCTL_HAWD);
-				pci_write_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, tmp16);
+				pcie_capability_clear_and_set_word(root, PCI_EXP_LNKCTL,
+								   PCI_EXP_LNKCTL_HAWD,
+								   bridge_cfg &
+								   PCI_EXP_LNKCTL_HAWD);
+				pcie_capability_clear_and_set_word(adev->pdev, PCI_EXP_LNKCTL,
+								   PCI_EXP_LNKCTL_HAWD,
+								   gpu_cfg &
+								   PCI_EXP_LNKCTL_HAWD);
 
 				/* linkctl2 */
-				pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL2, &tmp16);
-				tmp16 &= ~((1 << 4) | (7 << 9));
-				tmp16 |= (bridge_cfg2 & ((1 << 4) | (7 << 9)));
-				pci_write_config_word(root, bridge_pos + PCI_EXP_LNKCTL2, tmp16);
-
-				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, &tmp16);
-				tmp16 &= ~((1 << 4) | (7 << 9));
-				tmp16 |= (gpu_cfg2 & ((1 << 4) | (7 << 9)));
-				pci_write_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, tmp16);
+				pcie_capability_clear_and_set_word(root, PCI_EXP_LNKCTL2,
+								   PCI_EXP_LNKCTL2_ENTER_COMP |
+								   PCI_EXP_LNKCTL2_TX_MARGIN,
+								   bridge_cfg2 &
+								   (PCI_EXP_LNKCTL2_ENTER_COMP |
+								    PCI_EXP_LNKCTL2_TX_MARGIN));
+				pcie_capability_clear_and_set_word(adev->pdev, PCI_EXP_LNKCTL2,
+								   PCI_EXP_LNKCTL2_ENTER_COMP |
+								   PCI_EXP_LNKCTL2_TX_MARGIN,
+								   gpu_cfg2 &
+								   (PCI_EXP_LNKCTL2_ENTER_COMP |
+								    PCI_EXP_LNKCTL2_TX_MARGIN));
 
 				tmp = RREG32_PCIE(ixPCIE_LC_CNTL4);
 				tmp &= ~PCIE_LC_CNTL4__LC_SET_QUIESCE_MASK;
@@ -1434,15 +1664,15 @@ static void cik_pcie_gen3_enable(struct amdgpu_device *adev)
 	speed_cntl &= ~PCIE_LC_SPEED_CNTL__LC_FORCE_DIS_SW_SPEED_CHANGE_MASK;
 	WREG32_PCIE(ixPCIE_LC_SPEED_CNTL, speed_cntl);
 
-	pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, &tmp16);
-	tmp16 &= ~0xf;
+	tmp16 = 0;
 	if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3)
-		tmp16 |= 3; /* gen3 */
+		tmp16 |= PCI_EXP_LNKCTL2_TLS_8_0GT; /* gen3 */
 	else if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2)
-		tmp16 |= 2; /* gen2 */
+		tmp16 |= PCI_EXP_LNKCTL2_TLS_5_0GT; /* gen2 */
 	else
-		tmp16 |= 1; /* gen1 */
-	pci_write_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, tmp16);
+		tmp16 |= PCI_EXP_LNKCTL2_TLS_2_5GT; /* gen1 */
+	pcie_capability_clear_and_set_word(adev->pdev, PCI_EXP_LNKCTL2,
+					   PCI_EXP_LNKCTL2_TLS, tmp16);
 
 	speed_cntl = RREG32_PCIE(ixPCIE_LC_SPEED_CNTL);
 	speed_cntl |= PCIE_LC_SPEED_CNTL__LC_INITIATE_LINK_SPEED_CHANGE_MASK;
@@ -1462,16 +1692,12 @@ static void cik_program_aspm(struct amdgpu_device *adev)
 	bool disable_l0s = false, disable_l1 = false, disable_plloff_in_l1 = false;
 	bool disable_clkreq = false;
 
-	if (amdgpu_aspm == 0)
+	if (!amdgpu_device_should_use_aspm(adev))
 		return;
 
 	if (pci_is_root_bus(adev->pdev->bus))
 		return;
 
-	/* XXX double check APUs */
-	if (adev->flags & AMD_IS_APU)
-		return;
-
 	orig = data = RREG32_PCIE(ixPCIE_LC_N_FTS_CNTL);
 	data &= ~PCIE_LC_N_FTS_CNTL__LC_XMIT_N_FTS_MASK;
 	data |= (0x24 << PCIE_LC_N_FTS_CNTL__LC_XMIT_N_FTS__SHIFT) |
@@ -1629,10 +1855,110 @@ static uint32_t cik_get_rev_id(struct amdgpu_device *adev)
 		>> CC_DRM_ID_STRAPS__ATI_REV_ID__SHIFT;
 }
 
-static void cik_detect_hw_virtualization(struct amdgpu_device *adev)
+static void cik_flush_hdp(struct amdgpu_device *adev, struct amdgpu_ring *ring)
+{
+	if (!ring || !ring->funcs->emit_wreg) {
+		WREG32(mmHDP_MEM_COHERENCY_FLUSH_CNTL, 1);
+		RREG32(mmHDP_MEM_COHERENCY_FLUSH_CNTL);
+	} else {
+		amdgpu_ring_emit_wreg(ring, mmHDP_MEM_COHERENCY_FLUSH_CNTL, 1);
+	}
+}
+
+static void cik_invalidate_hdp(struct amdgpu_device *adev,
+			       struct amdgpu_ring *ring)
+{
+	if (!ring || !ring->funcs->emit_wreg) {
+		WREG32(mmHDP_DEBUG0, 1);
+		RREG32(mmHDP_DEBUG0);
+	} else {
+		amdgpu_ring_emit_wreg(ring, mmHDP_DEBUG0, 1);
+	}
+}
+
+static bool cik_need_full_reset(struct amdgpu_device *adev)
+{
+	/* change this when we support soft reset */
+	return true;
+}
+
+static void cik_get_pcie_usage(struct amdgpu_device *adev, uint64_t *count0,
+			       uint64_t *count1)
+{
+	uint32_t perfctr = 0;
+	uint64_t cnt0_of, cnt1_of;
+	int tmp;
+
+	/* This reports 0 on APUs, so return to avoid writing/reading registers
+	 * that may or may not be different from their GPU counterparts
+	 */
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	/* Set the 2 events that we wish to watch, defined above */
+	/* Reg 40 is # received msgs, Reg 104 is # of posted requests sent */
+	perfctr = REG_SET_FIELD(perfctr, PCIE_PERF_CNTL_TXCLK, EVENT0_SEL, 40);
+	perfctr = REG_SET_FIELD(perfctr, PCIE_PERF_CNTL_TXCLK, EVENT1_SEL, 104);
+
+	/* Write to enable desired perf counters */
+	WREG32_PCIE(ixPCIE_PERF_CNTL_TXCLK, perfctr);
+	/* Zero out and enable the perf counters
+	 * Write 0x5:
+	 * Bit 0 = Start all counters(1)
+	 * Bit 2 = Global counter reset enable(1)
+	 */
+	WREG32_PCIE(ixPCIE_PERF_COUNT_CNTL, 0x00000005);
+
+	msleep(1000);
+
+	/* Load the shadow and disable the perf counters
+	 * Write 0x2:
+	 * Bit 0 = Stop counters(0)
+	 * Bit 1 = Load the shadow counters(1)
+	 */
+	WREG32_PCIE(ixPCIE_PERF_COUNT_CNTL, 0x00000002);
+
+	/* Read register values to get any >32bit overflow */
+	tmp = RREG32_PCIE(ixPCIE_PERF_CNTL_TXCLK);
+	cnt0_of = REG_GET_FIELD(tmp, PCIE_PERF_CNTL_TXCLK, COUNTER0_UPPER);
+	cnt1_of = REG_GET_FIELD(tmp, PCIE_PERF_CNTL_TXCLK, COUNTER1_UPPER);
+
+	/* Get the values and add the overflow */
+	*count0 = RREG32_PCIE(ixPCIE_PERF_COUNT0_TXCLK) | (cnt0_of << 32);
+	*count1 = RREG32_PCIE(ixPCIE_PERF_COUNT1_TXCLK) | (cnt1_of << 32);
+}
+
+static bool cik_need_reset_on_init(struct amdgpu_device *adev)
+{
+	u32 clock_cntl, pc;
+
+	if (adev->flags & AMD_IS_APU)
+		return false;
+
+	/* check if the SMC is already running */
+	clock_cntl = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+	pc = RREG32_SMC(ixSMC_PC_C);
+	if ((0 == REG_GET_FIELD(clock_cntl, SMC_SYSCON_CLOCK_CNTL_0, ck_disable)) &&
+	    (0x20100 <= pc))
+		return true;
+
+	return false;
+}
+
+static uint64_t cik_get_pcie_replay_count(struct amdgpu_device *adev)
+{
+	uint64_t nak_r, nak_g;
+
+	/* Get the number of NAKs received and generated */
+	nak_r = RREG32_PCIE(ixPCIE_RX_NUM_NAK);
+	nak_g = RREG32_PCIE(ixPCIE_RX_NUM_NAK_GENERATED);
+
+	/* Add the total number of NAKs, i.e the number of replays */
+	return (nak_r + nak_g);
+}
+
+static void cik_pre_asic_init(struct amdgpu_device *adev)
 {
-	if (is_virtual_machine()) /* passthrough mode */
-		adev->virt.caps |= AMDGPU_PASSTHROUGH_MODE;
 }
 
 static const struct amdgpu_asic_funcs cik_asic_funcs =
@@ -1641,16 +1967,27 @@ static const struct amdgpu_asic_funcs cik_asic_funcs =
 	.read_bios_from_rom = &cik_read_bios_from_rom,
 	.read_register = &cik_read_register,
 	.reset = &cik_asic_reset,
+	.reset_method = &cik_asic_reset_method,
 	.set_vga_state = &cik_vga_set_state,
 	.get_xclk = &cik_get_xclk,
 	.set_uvd_clocks = &cik_set_uvd_clocks,
 	.set_vce_clocks = &cik_set_vce_clocks,
 	.get_config_memsize = &cik_get_config_memsize,
+	.flush_hdp = &cik_flush_hdp,
+	.invalidate_hdp = &cik_invalidate_hdp,
+	.need_full_reset = &cik_need_full_reset,
+	.init_doorbell_index = &legacy_doorbell_index_init,
+	.get_pcie_usage = &cik_get_pcie_usage,
+	.need_reset_on_init = &cik_need_reset_on_init,
+	.get_pcie_replay_count = &cik_get_pcie_replay_count,
+	.supports_baco = &cik_asic_supports_baco,
+	.pre_asic_init = &cik_pre_asic_init,
+	.query_video_codecs = &cik_query_video_codecs,
 };
 
-static int cik_common_early_init(void *handle)
+static int cik_common_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	adev->smc_rreg = &cik_smc_rreg;
 	adev->smc_wreg = &cik_smc_wreg;
@@ -1784,26 +2121,12 @@ static int cik_common_early_init(void *handle)
 		return -EINVAL;
 	}
 
-	adev->firmware.load_type = amdgpu_ucode_get_load_type(adev, amdgpu_fw_load_type);
-
-	amdgpu_get_pcie_info(adev);
-
 	return 0;
 }
 
-static int cik_common_sw_init(void *handle)
+static int cik_common_hw_init(struct amdgpu_ip_block *ip_block)
 {
-	return 0;
-}
-
-static int cik_common_sw_fini(void *handle)
-{
-	return 0;
-}
-
-static int cik_common_hw_init(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	/* move the golden regs per IP block */
 	cik_init_golden_registers(adev);
@@ -1815,48 +2138,36 @@ static int cik_common_hw_init(void *handle)
 	return 0;
 }
 
-static int cik_common_hw_fini(void *handle)
+static int cik_common_hw_fini(struct amdgpu_ip_block *ip_block)
 {
 	return 0;
 }
 
-static int cik_common_suspend(void *handle)
+static int cik_common_resume(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return cik_common_hw_fini(adev);
+	return cik_common_hw_init(ip_block);
 }
 
-static int cik_common_resume(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return cik_common_hw_init(adev);
-}
-
-static bool cik_common_is_idle(void *handle)
+static bool cik_common_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	return true;
 }
 
-static int cik_common_wait_for_idle(void *handle)
-{
-	return 0;
-}
 
-static int cik_common_soft_reset(void *handle)
+
+static int cik_common_soft_reset(struct amdgpu_ip_block *ip_block)
 {
 	/* XXX hard reset?? */
 	return 0;
 }
 
-static int cik_common_set_clockgating_state(void *handle,
+static int cik_common_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					    enum amd_clockgating_state state)
 {
 	return 0;
 }
 
-static int cik_common_set_powergating_state(void *handle,
+static int cik_common_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					    enum amd_powergating_state state)
 {
 	return 0;
@@ -1865,15 +2176,10 @@ static int cik_common_set_powergating_state(void *handle,
 static const struct amd_ip_funcs cik_common_ip_funcs = {
 	.name = "cik_common",
 	.early_init = cik_common_early_init,
-	.late_init = NULL,
-	.sw_init = cik_common_sw_init,
-	.sw_fini = cik_common_sw_fini,
 	.hw_init = cik_common_hw_init,
 	.hw_fini = cik_common_hw_fini,
-	.suspend = cik_common_suspend,
 	.resume = cik_common_resume,
 	.is_idle = cik_common_is_idle,
-	.wait_for_idle = cik_common_wait_for_idle,
 	.soft_reset = cik_common_soft_reset,
 	.set_clockgating_state = cik_common_set_clockgating_state,
 	.set_powergating_state = cik_common_set_powergating_state,
@@ -1890,65 +2196,80 @@ static const struct amdgpu_ip_block_version cik_common_ip_block =
 
 int cik_set_ip_blocks(struct amdgpu_device *adev)
 {
-	cik_detect_hw_virtualization(adev);
-
 	switch (adev->asic_type) {
 	case CHIP_BONAIRE:
-		amdgpu_ip_block_add(adev, &cik_common_ip_block);
-		amdgpu_ip_block_add(adev, &gmc_v7_0_ip_block);
-		amdgpu_ip_block_add(adev, &cik_ih_ip_block);
-		amdgpu_ip_block_add(adev, &amdgpu_pp_ip_block);
+		amdgpu_device_ip_block_add(adev, &cik_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v7_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &cik_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v7_2_ip_block);
+		amdgpu_device_ip_block_add(adev, &cik_sdma_ip_block);
+		amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
 		if (adev->enable_virtual_display)
-			amdgpu_ip_block_add(adev, &dce_virtual_ip_block);
+			amdgpu_device_ip_block_add(adev, &amdgpu_vkms_ip_block);
+#if defined(CONFIG_DRM_AMD_DC)
+		else if (amdgpu_device_has_dc_support(adev))
+			amdgpu_device_ip_block_add(adev, &dm_ip_block);
+#endif
 		else
-			amdgpu_ip_block_add(adev, &dce_v8_2_ip_block);
-		amdgpu_ip_block_add(adev, &gfx_v7_2_ip_block);
-		amdgpu_ip_block_add(adev, &cik_sdma_ip_block);
-		amdgpu_ip_block_add(adev, &uvd_v4_2_ip_block);
-		amdgpu_ip_block_add(adev, &vce_v2_0_ip_block);
+			amdgpu_device_ip_block_add(adev, &dce_v8_2_ip_block);
+		amdgpu_device_ip_block_add(adev, &uvd_v4_2_ip_block);
+		amdgpu_device_ip_block_add(adev, &vce_v2_0_ip_block);
 		break;
 	case CHIP_HAWAII:
-		amdgpu_ip_block_add(adev, &cik_common_ip_block);
-		amdgpu_ip_block_add(adev, &gmc_v7_0_ip_block);
-		amdgpu_ip_block_add(adev, &cik_ih_ip_block);
-		amdgpu_ip_block_add(adev, &amdgpu_pp_ip_block);
+		amdgpu_device_ip_block_add(adev, &cik_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v7_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &cik_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v7_3_ip_block);
+		amdgpu_device_ip_block_add(adev, &cik_sdma_ip_block);
+		amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
 		if (adev->enable_virtual_display)
-			amdgpu_ip_block_add(adev, &dce_virtual_ip_block);
+			amdgpu_device_ip_block_add(adev, &amdgpu_vkms_ip_block);
+#if defined(CONFIG_DRM_AMD_DC)
+		else if (amdgpu_device_has_dc_support(adev))
+			amdgpu_device_ip_block_add(adev, &dm_ip_block);
+#endif
 		else
-			amdgpu_ip_block_add(adev, &dce_v8_5_ip_block);
-		amdgpu_ip_block_add(adev, &gfx_v7_3_ip_block);
-		amdgpu_ip_block_add(adev, &cik_sdma_ip_block);
-		amdgpu_ip_block_add(adev, &uvd_v4_2_ip_block);
-		amdgpu_ip_block_add(adev, &vce_v2_0_ip_block);
+			amdgpu_device_ip_block_add(adev, &dce_v8_5_ip_block);
+		amdgpu_device_ip_block_add(adev, &uvd_v4_2_ip_block);
+		amdgpu_device_ip_block_add(adev, &vce_v2_0_ip_block);
 		break;
 	case CHIP_KAVERI:
-		amdgpu_ip_block_add(adev, &cik_common_ip_block);
-		amdgpu_ip_block_add(adev, &gmc_v7_0_ip_block);
-		amdgpu_ip_block_add(adev, &cik_ih_ip_block);
-		amdgpu_ip_block_add(adev, &amdgpu_pp_ip_block);
+		amdgpu_device_ip_block_add(adev, &cik_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v7_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &cik_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v7_1_ip_block);
+		amdgpu_device_ip_block_add(adev, &cik_sdma_ip_block);
+		amdgpu_device_ip_block_add(adev, &kv_smu_ip_block);
 		if (adev->enable_virtual_display)
-			amdgpu_ip_block_add(adev, &dce_virtual_ip_block);
+			amdgpu_device_ip_block_add(adev, &amdgpu_vkms_ip_block);
+#if defined(CONFIG_DRM_AMD_DC)
+		else if (amdgpu_device_has_dc_support(adev))
+			amdgpu_device_ip_block_add(adev, &dm_ip_block);
+#endif
 		else
-			amdgpu_ip_block_add(adev, &dce_v8_1_ip_block);
-		amdgpu_ip_block_add(adev, &gfx_v7_1_ip_block);
-		amdgpu_ip_block_add(adev, &cik_sdma_ip_block);
-		amdgpu_ip_block_add(adev, &uvd_v4_2_ip_block);
-		amdgpu_ip_block_add(adev, &vce_v2_0_ip_block);
+			amdgpu_device_ip_block_add(adev, &dce_v8_1_ip_block);
+
+		amdgpu_device_ip_block_add(adev, &uvd_v4_2_ip_block);
+		amdgpu_device_ip_block_add(adev, &vce_v2_0_ip_block);
 		break;
 	case CHIP_KABINI:
 	case CHIP_MULLINS:
-		amdgpu_ip_block_add(adev, &cik_common_ip_block);
-		amdgpu_ip_block_add(adev, &gmc_v7_0_ip_block);
-		amdgpu_ip_block_add(adev, &cik_ih_ip_block);
-		amdgpu_ip_block_add(adev, &amdgpu_pp_ip_block);
+		amdgpu_device_ip_block_add(adev, &cik_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v7_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &cik_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v7_2_ip_block);
+		amdgpu_device_ip_block_add(adev, &cik_sdma_ip_block);
+		amdgpu_device_ip_block_add(adev, &kv_smu_ip_block);
 		if (adev->enable_virtual_display)
-			amdgpu_ip_block_add(adev, &dce_virtual_ip_block);
+			amdgpu_device_ip_block_add(adev, &amdgpu_vkms_ip_block);
+#if defined(CONFIG_DRM_AMD_DC)
+		else if (amdgpu_device_has_dc_support(adev))
+			amdgpu_device_ip_block_add(adev, &dm_ip_block);
+#endif
 		else
-			amdgpu_ip_block_add(adev, &dce_v8_3_ip_block);
-		amdgpu_ip_block_add(adev, &gfx_v7_2_ip_block);
-		amdgpu_ip_block_add(adev, &cik_sdma_ip_block);
-		amdgpu_ip_block_add(adev, &uvd_v4_2_ip_block);
-		amdgpu_ip_block_add(adev, &vce_v2_0_ip_block);
+			amdgpu_device_ip_block_add(adev, &dce_v8_3_ip_block);
+		amdgpu_device_ip_block_add(adev, &uvd_v4_2_ip_block);
+		amdgpu_device_ip_block_add(adev, &vce_v2_0_ip_block);
 		break;
 	default:
 		/* FIXME: not supported yet */
diff --git a/drivers/gpu/drm/amd/amdgpu/cik.h b/drivers/gpu/drm/amd/amdgpu/cik.h
index c4989f51ecef..f91ab4c246b7 100644
--- a/drivers/gpu/drm/amd/amdgpu/cik.h
+++ b/drivers/gpu/drm/amd/amdgpu/cik.h
@@ -24,8 +24,12 @@
 #ifndef __CIK_H__
 #define __CIK_H__
 
+#define CIK_FLUSH_GPU_TLB_NUM_WREG	3
+
 void cik_srbm_select(struct amdgpu_device *adev,
 		     u32 me, u32 pipe, u32 queue, u32 vmid);
 int cik_set_ip_blocks(struct amdgpu_device *adev);
 
+void legacy_doorbell_index_init(struct amdgpu_device *adev);
+
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/cik_ih.c b/drivers/gpu/drm/amd/amdgpu/cik_ih.c
index b8918432c572..41f4705bdbbd 100644
--- a/drivers/gpu/drm/amd/amdgpu/cik_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/cik_ih.c
@@ -20,7 +20,9 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  *
  */
-#include <drm/drmP.h>
+
+#include <linux/pci.h>
+
 #include "amdgpu.h"
 #include "amdgpu_ih.h"
 #include "cikd.h"
@@ -103,15 +105,15 @@ static void cik_ih_disable_interrupts(struct amdgpu_device *adev)
  */
 static int cik_ih_irq_init(struct amdgpu_device *adev)
 {
+	struct amdgpu_ih_ring *ih = &adev->irq.ih;
 	int rb_bufsz;
 	u32 interrupt_cntl, ih_cntl, ih_rb_cntl;
-	u64 wptr_off;
 
 	/* disable irqs */
 	cik_ih_disable_interrupts(adev);
 
 	/* setup interrupt control */
-	WREG32(mmINTERRUPT_CNTL2, adev->dummy_page.addr >> 8);
+	WREG32(mmINTERRUPT_CNTL2, adev->dummy_page_addr >> 8);
 	interrupt_cntl = RREG32(mmINTERRUPT_CNTL);
 	/* INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
 	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
@@ -131,9 +133,8 @@ static int cik_ih_irq_init(struct amdgpu_device *adev)
 	ih_rb_cntl |= IH_RB_CNTL__WPTR_WRITEBACK_ENABLE_MASK;
 
 	/* set the writeback address whether it's enabled or not */
-	wptr_off = adev->wb.gpu_addr + (adev->irq.ih.wptr_offs * 4);
-	WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(wptr_off));
-	WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(wptr_off) & 0xFF);
+	WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(ih->wptr_addr));
+	WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(ih->wptr_addr) & 0xFF);
 
 	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
 
@@ -176,6 +177,7 @@ static void cik_ih_irq_disable(struct amdgpu_device *adev)
  * cik_ih_get_wptr - get the IH ring buffer wptr
  *
  * @adev: amdgpu_device pointer
+ * @ih: IH ring buffer to fetch wptr
  *
  * Get the IH ring buffer wptr from either the register
  * or the writeback memory buffer (CIK).  Also check for
@@ -183,11 +185,12 @@ static void cik_ih_irq_disable(struct amdgpu_device *adev)
  * Used by cik_irq_process().
  * Returns the value of the wptr.
  */
-static u32 cik_ih_get_wptr(struct amdgpu_device *adev)
+static u32 cik_ih_get_wptr(struct amdgpu_device *adev,
+			   struct amdgpu_ih_ring *ih)
 {
 	u32 wptr, tmp;
 
-	wptr = le32_to_cpu(adev->wb.wb[adev->irq.ih.wptr_offs]);
+	wptr = le32_to_cpu(*ih->wptr_cpu);
 
 	if (wptr & IH_RB_WPTR__RB_OVERFLOW_MASK) {
 		wptr &= ~IH_RB_WPTR__RB_OVERFLOW_MASK;
@@ -196,13 +199,19 @@ static u32 cik_ih_get_wptr(struct amdgpu_device *adev)
 		 * this should allow us to catchup.
 		 */
 		dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
-			wptr, adev->irq.ih.rptr, (wptr + 16) & adev->irq.ih.ptr_mask);
-		adev->irq.ih.rptr = (wptr + 16) & adev->irq.ih.ptr_mask;
+			 wptr, ih->rptr, (wptr + 16) & ih->ptr_mask);
+		ih->rptr = (wptr + 16) & ih->ptr_mask;
 		tmp = RREG32(mmIH_RB_CNTL);
 		tmp |= IH_RB_CNTL__WPTR_OVERFLOW_CLEAR_MASK;
 		WREG32(mmIH_RB_CNTL, tmp);
+
+		/* Unset the CLEAR_OVERFLOW bit immediately so new overflows
+		 * can be detected.
+		 */
+		tmp &= ~IH_RB_CNTL__WPTR_OVERFLOW_CLEAR_MASK;
+		WREG32(mmIH_RB_CNTL, tmp);
 	}
-	return (wptr & adev->irq.ih.ptr_mask);
+	return (wptr & ih->ptr_mask);
 }
 
 /*        CIK IV Ring
@@ -237,43 +246,46 @@ static u32 cik_ih_get_wptr(struct amdgpu_device *adev)
  * position and also advance the position.
  */
 static void cik_ih_decode_iv(struct amdgpu_device *adev,
+			     struct amdgpu_ih_ring *ih,
 			     struct amdgpu_iv_entry *entry)
 {
 	/* wptr/rptr are in bytes! */
-	u32 ring_index = adev->irq.ih.rptr >> 2;
+	u32 ring_index = ih->rptr >> 2;
 	uint32_t dw[4];
 
-	dw[0] = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
-	dw[1] = le32_to_cpu(adev->irq.ih.ring[ring_index + 1]);
-	dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
-	dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
+	dw[0] = le32_to_cpu(ih->ring[ring_index + 0]);
+	dw[1] = le32_to_cpu(ih->ring[ring_index + 1]);
+	dw[2] = le32_to_cpu(ih->ring[ring_index + 2]);
+	dw[3] = le32_to_cpu(ih->ring[ring_index + 3]);
 
-	entry->client_id = AMDGPU_IH_CLIENTID_LEGACY;
+	entry->client_id = AMDGPU_IRQ_CLIENTID_LEGACY;
 	entry->src_id = dw[0] & 0xff;
 	entry->src_data[0] = dw[1] & 0xfffffff;
 	entry->ring_id = dw[2] & 0xff;
-	entry->vm_id = (dw[2] >> 8) & 0xff;
-	entry->pas_id = (dw[2] >> 16) & 0xffff;
+	entry->vmid = (dw[2] >> 8) & 0xff;
+	entry->pasid = (dw[2] >> 16) & 0xffff;
 
 	/* wptr/rptr are in bytes! */
-	adev->irq.ih.rptr += 16;
+	ih->rptr += 16;
 }
 
 /**
  * cik_ih_set_rptr - set the IH ring buffer rptr
  *
  * @adev: amdgpu_device pointer
+ * @ih: IH ring buffer to set wptr
  *
  * Set the IH ring buffer rptr.
  */
-static void cik_ih_set_rptr(struct amdgpu_device *adev)
+static void cik_ih_set_rptr(struct amdgpu_device *adev,
+			    struct amdgpu_ih_ring *ih)
 {
-	WREG32(mmIH_RB_RPTR, adev->irq.ih.rptr);
+	WREG32(mmIH_RB_RPTR, ih->rptr);
 }
 
-static int cik_ih_early_init(void *handle)
+static int cik_ih_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int ret;
 
 	ret = amdgpu_irq_add_domain(adev);
@@ -285,12 +297,12 @@ static int cik_ih_early_init(void *handle)
 	return 0;
 }
 
-static int cik_ih_sw_init(void *handle)
+static int cik_ih_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	r = amdgpu_ih_ring_init(adev, 64 * 1024, false);
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih, 64 * 1024, false);
 	if (r)
 		return r;
 
@@ -299,55 +311,43 @@ static int cik_ih_sw_init(void *handle)
 	return r;
 }
 
-static int cik_ih_sw_fini(void *handle)
+static int cik_ih_sw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	amdgpu_irq_fini(adev);
-	amdgpu_ih_ring_fini(adev);
+	amdgpu_irq_fini_sw(adev);
 	amdgpu_irq_remove_domain(adev);
 
 	return 0;
 }
 
-static int cik_ih_hw_init(void *handle)
+static int cik_ih_hw_init(struct amdgpu_ip_block *ip_block)
 {
-	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	r = cik_ih_irq_init(adev);
-	if (r)
-		return r;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	return 0;
+	return cik_ih_irq_init(adev);
 }
 
-static int cik_ih_hw_fini(void *handle)
+static int cik_ih_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	cik_ih_irq_disable(adev);
+	cik_ih_irq_disable(ip_block->adev);
 
 	return 0;
 }
 
-static int cik_ih_suspend(void *handle)
+static int cik_ih_suspend(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return cik_ih_hw_fini(adev);
+	return cik_ih_hw_fini(ip_block);
 }
 
-static int cik_ih_resume(void *handle)
+static int cik_ih_resume(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return cik_ih_hw_init(adev);
+	return cik_ih_hw_init(ip_block);
 }
 
-static bool cik_ih_is_idle(void *handle)
+static bool cik_ih_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 tmp = RREG32(mmSRBM_STATUS);
 
 	if (tmp & SRBM_STATUS__IH_BUSY_MASK)
@@ -356,11 +356,11 @@ static bool cik_ih_is_idle(void *handle)
 	return true;
 }
 
-static int cik_ih_wait_for_idle(void *handle)
+static int cik_ih_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
 	unsigned i;
 	u32 tmp;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
 		/* read MC_STATUS */
@@ -372,9 +372,9 @@ static int cik_ih_wait_for_idle(void *handle)
 	return -ETIMEDOUT;
 }
 
-static int cik_ih_soft_reset(void *handle)
+static int cik_ih_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	u32 srbm_soft_reset = 0;
 	u32 tmp = RREG32(mmSRBM_STATUS);
@@ -402,13 +402,13 @@ static int cik_ih_soft_reset(void *handle)
 	return 0;
 }
 
-static int cik_ih_set_clockgating_state(void *handle,
+static int cik_ih_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
 	return 0;
 }
 
-static int cik_ih_set_powergating_state(void *handle,
+static int cik_ih_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	return 0;
@@ -417,7 +417,6 @@ static int cik_ih_set_powergating_state(void *handle,
 static const struct amd_ip_funcs cik_ih_ip_funcs = {
 	.name = "cik_ih",
 	.early_init = cik_ih_early_init,
-	.late_init = NULL,
 	.sw_init = cik_ih_sw_init,
 	.sw_fini = cik_ih_sw_fini,
 	.hw_init = cik_ih_hw_init,
@@ -439,12 +438,10 @@ static const struct amdgpu_ih_funcs cik_ih_funcs = {
 
 static void cik_ih_set_interrupt_funcs(struct amdgpu_device *adev)
 {
-	if (adev->irq.ih_funcs == NULL)
-		adev->irq.ih_funcs = &cik_ih_funcs;
+	adev->irq.ih_funcs = &cik_ih_funcs;
 }
 
-const struct amdgpu_ip_block_version cik_ih_ip_block =
-{
+const struct amdgpu_ip_block_version cik_ih_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_IH,
 	.major = 2,
 	.minor = 0,
diff --git a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
index f508f4d01e4a..9e8715b4739d 100644
--- a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
+++ b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
@@ -21,8 +21,10 @@
  *
  * Authors: Alex Deucher
  */
+
 #include <linux/firmware.h>
-#include <drm/drmP.h>
+#include <linux/module.h>
+
 #include "amdgpu.h"
 #include "amdgpu_ucode.h"
 #include "amdgpu_trace.h"
@@ -52,29 +54,27 @@ static void cik_sdma_set_ring_funcs(struct amdgpu_device *adev);
 static void cik_sdma_set_irq_funcs(struct amdgpu_device *adev);
 static void cik_sdma_set_buffer_funcs(struct amdgpu_device *adev);
 static void cik_sdma_set_vm_pte_funcs(struct amdgpu_device *adev);
-static int cik_sdma_soft_reset(void *handle);
-
-MODULE_FIRMWARE("radeon/bonaire_sdma.bin");
-MODULE_FIRMWARE("radeon/bonaire_sdma1.bin");
-MODULE_FIRMWARE("radeon/hawaii_sdma.bin");
-MODULE_FIRMWARE("radeon/hawaii_sdma1.bin");
-MODULE_FIRMWARE("radeon/kaveri_sdma.bin");
-MODULE_FIRMWARE("radeon/kaveri_sdma1.bin");
-MODULE_FIRMWARE("radeon/kabini_sdma.bin");
-MODULE_FIRMWARE("radeon/kabini_sdma1.bin");
-MODULE_FIRMWARE("radeon/mullins_sdma.bin");
-MODULE_FIRMWARE("radeon/mullins_sdma1.bin");
+static int cik_sdma_soft_reset(struct amdgpu_ip_block *ip_block);
 
 u32 amdgpu_cik_gpu_check_soft_reset(struct amdgpu_device *adev);
 
+MODULE_FIRMWARE("amdgpu/bonaire_sdma.bin");
+MODULE_FIRMWARE("amdgpu/bonaire_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/hawaii_sdma.bin");
+MODULE_FIRMWARE("amdgpu/hawaii_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/kaveri_sdma.bin");
+MODULE_FIRMWARE("amdgpu/kaveri_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/kabini_sdma.bin");
+MODULE_FIRMWARE("amdgpu/kabini_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/mullins_sdma.bin");
+MODULE_FIRMWARE("amdgpu/mullins_sdma1.bin");
 
 static void cik_sdma_free_microcode(struct amdgpu_device *adev)
 {
 	int i;
-	for (i = 0; i < adev->sdma.num_instances; i++) {
-			release_firmware(adev->sdma.instance[i].fw);
-			adev->sdma.instance[i].fw = NULL;
-	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		amdgpu_ucode_release(&adev->sdma.instance[i].fw);
 }
 
 /*
@@ -106,7 +106,6 @@ static void cik_sdma_free_microcode(struct amdgpu_device *adev)
 static int cik_sdma_init_microcode(struct amdgpu_device *adev)
 {
 	const char *chip_name;
-	char fw_name[30];
 	int err = 0, i;
 
 	DRM_DEBUG("\n");
@@ -132,21 +131,22 @@ static int cik_sdma_init_microcode(struct amdgpu_device *adev)
 
 	for (i = 0; i < adev->sdma.num_instances; i++) {
 		if (i == 0)
-			snprintf(fw_name, sizeof(fw_name), "radeon/%s_sdma.bin", chip_name);
+			err = amdgpu_ucode_request(adev, &adev->sdma.instance[i].fw,
+						   AMDGPU_UCODE_REQUIRED,
+						   "amdgpu/%s_sdma.bin", chip_name);
 		else
-			snprintf(fw_name, sizeof(fw_name), "radeon/%s_sdma1.bin", chip_name);
-		err = request_firmware(&adev->sdma.instance[i].fw, fw_name, adev->dev);
+			err = amdgpu_ucode_request(adev, &adev->sdma.instance[i].fw,
+						   AMDGPU_UCODE_REQUIRED,
+						   "amdgpu/%s_sdma1.bin", chip_name);
 		if (err)
 			goto out;
-		err = amdgpu_ucode_validate(adev->sdma.instance[i].fw);
 	}
 out:
 	if (err) {
-		pr_err("cik_sdma: Failed to load firmware \"%s\"\n", fw_name);
-		for (i = 0; i < adev->sdma.num_instances; i++) {
-			release_firmware(adev->sdma.instance[i].fw);
-			adev->sdma.instance[i].fw = NULL;
-		}
+		pr_err("cik_sdma: Failed to load firmware \"%s_sdma%s.bin\"\n",
+		       chip_name, i == 0 ? "" : "1");
+		for (i = 0; i < adev->sdma.num_instances; i++)
+			amdgpu_ucode_release(&adev->sdma.instance[i].fw);
 	}
 	return err;
 }
@@ -162,7 +162,7 @@ static uint64_t cik_sdma_ring_get_rptr(struct amdgpu_ring *ring)
 {
 	u32 rptr;
 
-	rptr = ring->adev->wb.wb[ring->rptr_offs];
+	rptr = *ring->rptr_cpu_addr;
 
 	return (rptr & 0x3fffc) >> 2;
 }
@@ -177,9 +177,8 @@ static uint64_t cik_sdma_ring_get_rptr(struct amdgpu_ring *ring)
 static uint64_t cik_sdma_ring_get_wptr(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
-	u32 me = (ring == &adev->sdma.instance[0].ring) ? 0 : 1;
 
-	return (RREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) & 0x3fffc) >> 2;
+	return (RREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) & 0x3fffc) >> 2;
 }
 
 /**
@@ -192,15 +191,14 @@ static uint64_t cik_sdma_ring_get_wptr(struct amdgpu_ring *ring)
 static void cik_sdma_ring_set_wptr(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
-	u32 me = (ring == &adev->sdma.instance[0].ring) ? 0 : 1;
 
-	WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[me],
-		       	(lower_32_bits(ring->wptr) << 2) & 0x3fffc);
+	WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me],
+	       (ring->wptr << 2) & 0x3fffc);
 }
 
 static void cik_sdma_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
 {
-	struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring);
+	struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring);
 	int i;
 
 	for (i = 0; i < count; i++)
@@ -215,18 +213,22 @@ static void cik_sdma_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
  * cik_sdma_ring_emit_ib - Schedule an IB on the DMA engine
  *
  * @ring: amdgpu ring pointer
+ * @job: job to retrive vmid from
  * @ib: IB object to schedule
+ * @flags: unused
  *
  * Schedule an IB in the DMA ring (CIK).
  */
 static void cik_sdma_ring_emit_ib(struct amdgpu_ring *ring,
+				  struct amdgpu_job *job,
 				  struct amdgpu_ib *ib,
-				  unsigned vm_id, bool ctx_switch)
+				  uint32_t flags)
 {
-	u32 extra_bits = vm_id & 0xf;
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+	u32 extra_bits = vmid & 0xf;
 
 	/* IB packet must end on a 8 DW boundary */
-	cik_sdma_ring_insert_nop(ring, (12 - (lower_32_bits(ring->wptr) & 7)) % 8);
+	cik_sdma_ring_insert_nop(ring, (4 - lower_32_bits(ring->wptr)) & 7);
 
 	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_INDIRECT_BUFFER, 0, extra_bits));
 	amdgpu_ring_write(ring, ib->gpu_addr & 0xffffffe0); /* base must be 32 byte aligned */
@@ -248,7 +250,7 @@ static void cik_sdma_ring_emit_hdp_flush(struct amdgpu_ring *ring)
 			  SDMA_POLL_REG_MEM_EXTRA_FUNC(3)); /* == */
 	u32 ref_and_mask;
 
-	if (ring == &ring->adev->sdma.instance[0].ring)
+	if (ring->me == 0)
 		ref_and_mask = GPU_HDP_FLUSH_DONE__SDMA0_MASK;
 	else
 		ref_and_mask = GPU_HDP_FLUSH_DONE__SDMA1_MASK;
@@ -261,18 +263,13 @@ static void cik_sdma_ring_emit_hdp_flush(struct amdgpu_ring *ring)
 	amdgpu_ring_write(ring, (0xfff << 16) | 10); /* retry count, poll interval */
 }
 
-static void cik_sdma_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
-{
-	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_SRBM_WRITE, 0, 0xf000));
-	amdgpu_ring_write(ring, mmHDP_DEBUG0);
-	amdgpu_ring_write(ring, 1);
-}
-
 /**
  * cik_sdma_ring_emit_fence - emit a fence on the DMA ring
  *
  * @ring: amdgpu ring pointer
- * @fence: amdgpu fence object
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
  *
  * Add a DMA fence packet to the ring to write
  * the fence seq number and DMA trap packet to generate
@@ -310,23 +307,15 @@ static void cik_sdma_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq
  */
 static void cik_sdma_gfx_stop(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *sdma0 = &adev->sdma.instance[0].ring;
-	struct amdgpu_ring *sdma1 = &adev->sdma.instance[1].ring;
 	u32 rb_cntl;
 	int i;
 
-	if ((adev->mman.buffer_funcs_ring == sdma0) ||
-	    (adev->mman.buffer_funcs_ring == sdma1))
-		amdgpu_ttm_set_active_vram_size(adev, adev->mc.visible_vram_size);
-
 	for (i = 0; i < adev->sdma.num_instances; i++) {
 		rb_cntl = RREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i]);
 		rb_cntl &= ~SDMA0_GFX_RB_CNTL__RB_ENABLE_MASK;
 		WREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i], rb_cntl);
 		WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], 0);
 	}
-	sdma0->ready = false;
-	sdma1->ready = false;
 }
 
 /**
@@ -439,12 +428,10 @@ static int cik_sdma_gfx_resume(struct amdgpu_device *adev)
 	struct amdgpu_ring *ring;
 	u32 rb_cntl, ib_cntl;
 	u32 rb_bufsz;
-	u32 wb_offset;
 	int i, j, r;
 
 	for (i = 0; i < adev->sdma.num_instances; i++) {
 		ring = &adev->sdma.instance[i].ring;
-		wb_offset = (ring->rptr_offs * 4);
 
 		mutex_lock(&adev->srbm_mutex);
 		for (j = 0; j < 16; j++) {
@@ -480,9 +467,9 @@ static int cik_sdma_gfx_resume(struct amdgpu_device *adev)
 
 		/* set the wb address whether it's enabled or not */
 		WREG32(mmSDMA0_GFX_RB_RPTR_ADDR_HI + sdma_offsets[i],
-		       upper_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFF);
+		       upper_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFF);
 		WREG32(mmSDMA0_GFX_RB_RPTR_ADDR_LO + sdma_offsets[i],
-		       ((adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFC));
+		       ((ring->rptr_gpu_addr) & 0xFFFFFFFC));
 
 		rb_cntl |= SDMA0_GFX_RB_CNTL__RPTR_WRITEBACK_ENABLE_MASK;
 
@@ -490,7 +477,7 @@ static int cik_sdma_gfx_resume(struct amdgpu_device *adev)
 		WREG32(mmSDMA0_GFX_RB_BASE_HI + sdma_offsets[i], ring->gpu_addr >> 40);
 
 		ring->wptr = 0;
-		WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[i], lower_32_bits(ring->wptr) << 2);
+		WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[i], ring->wptr << 2);
 
 		/* enable DMA RB */
 		WREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i],
@@ -502,22 +489,15 @@ static int cik_sdma_gfx_resume(struct amdgpu_device *adev)
 #endif
 		/* enable DMA IBs */
 		WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl);
-
-		ring->ready = true;
 	}
 
 	cik_sdma_enable(adev, true);
 
 	for (i = 0; i < adev->sdma.num_instances; i++) {
 		ring = &adev->sdma.instance[i].ring;
-		r = amdgpu_ring_test_ring(ring);
-		if (r) {
-			ring->ready = false;
+		r = amdgpu_ring_test_helper(ring);
+		if (r)
 			return r;
-		}
-
-		if (adev->mman.buffer_funcs_ring == ring)
-			amdgpu_ttm_set_active_vram_size(adev, adev->mc.real_vram_size);
 	}
 
 	return 0;
@@ -626,22 +606,18 @@ static int cik_sdma_ring_test_ring(struct amdgpu_ring *ring)
 	u32 tmp;
 	u64 gpu_addr;
 
-	r = amdgpu_wb_get(adev, &index);
-	if (r) {
-		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r)
 		return r;
-	}
 
 	gpu_addr = adev->wb.gpu_addr + (index * 4);
 	tmp = 0xCAFEDEAD;
 	adev->wb.wb[index] = cpu_to_le32(tmp);
 
 	r = amdgpu_ring_alloc(ring, 5);
-	if (r) {
-		DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r);
-		amdgpu_wb_free(adev, index);
-		return r;
-	}
+	if (r)
+		goto error_free_wb;
+
 	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_WRITE, SDMA_WRITE_SUB_OPCODE_LINEAR, 0));
 	amdgpu_ring_write(ring, lower_32_bits(gpu_addr));
 	amdgpu_ring_write(ring, upper_32_bits(gpu_addr));
@@ -653,18 +629,14 @@ static int cik_sdma_ring_test_ring(struct amdgpu_ring *ring)
 		tmp = le32_to_cpu(adev->wb.wb[index]);
 		if (tmp == 0xDEADBEEF)
 			break;
-		DRM_UDELAY(1);
+		udelay(1);
 	}
 
-	if (i < adev->usec_timeout) {
-		DRM_INFO("ring test on %d succeeded in %d usecs\n", ring->idx, i);
-	} else {
-		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
-			  ring->idx, tmp);
-		r = -EINVAL;
-	}
-	amdgpu_wb_free(adev, index);
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
 
+error_free_wb:
+	amdgpu_device_wb_free(adev, index);
 	return r;
 }
 
@@ -672,6 +644,7 @@ static int cik_sdma_ring_test_ring(struct amdgpu_ring *ring)
  * cik_sdma_ring_test_ib - test an IB on the DMA engine
  *
  * @ring: amdgpu_ring structure holding ring information
+ * @timeout: timeout value in jiffies, or MAX_SCHEDULE_TIMEOUT
  *
  * Test a simple IB in the DMA ring (CIK).
  * Returns 0 on success, error on failure.
@@ -686,21 +659,18 @@ static int cik_sdma_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 	u64 gpu_addr;
 	long r;
 
-	r = amdgpu_wb_get(adev, &index);
-	if (r) {
-		dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r)
 		return r;
-	}
 
 	gpu_addr = adev->wb.gpu_addr + (index * 4);
 	tmp = 0xCAFEDEAD;
 	adev->wb.wb[index] = cpu_to_le32(tmp);
 	memset(&ib, 0, sizeof(ib));
-	r = amdgpu_ib_get(adev, NULL, 256, &ib);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
+	r = amdgpu_ib_get(adev, NULL, 256,
+					AMDGPU_IB_POOL_DIRECT, &ib);
+	if (r)
 		goto err0;
-	}
 
 	ib.ptr[0] = SDMA_PACKET(SDMA_OPCODE_WRITE,
 				SDMA_WRITE_SUB_OPCODE_LINEAR, 0);
@@ -715,32 +685,27 @@ static int cik_sdma_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 
 	r = dma_fence_wait_timeout(f, false, timeout);
 	if (r == 0) {
-		DRM_ERROR("amdgpu: IB test timed out\n");
 		r = -ETIMEDOUT;
 		goto err1;
 	} else if (r < 0) {
-		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
 		goto err1;
 	}
 	tmp = le32_to_cpu(adev->wb.wb[index]);
-	if (tmp == 0xDEADBEEF) {
-		DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
+	if (tmp == 0xDEADBEEF)
 		r = 0;
-	} else {
-		DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp);
+	else
 		r = -EINVAL;
-	}
 
 err1:
-	amdgpu_ib_free(adev, &ib, NULL);
+	amdgpu_ib_free(&ib, NULL);
 	dma_fence_put(f);
 err0:
-	amdgpu_wb_free(adev, index);
+	amdgpu_device_wb_free(adev, index);
 	return r;
 }
 
 /**
- * cik_sdma_vm_copy_pages - update PTEs by copying them from the GART
+ * cik_sdma_vm_copy_pte - update PTEs by copying them from the GART
  *
  * @ib: indirect buffer to fill with commands
  * @pe: addr of the page entry
@@ -766,7 +731,7 @@ static void cik_sdma_vm_copy_pte(struct amdgpu_ib *ib,
 }
 
 /**
- * cik_sdma_vm_write_pages - update PTEs by writing them manually
+ * cik_sdma_vm_write_pte - update PTEs by writing them manually
  *
  * @ib: indirect buffer to fill with commands
  * @pe: addr of the page entry
@@ -795,7 +760,7 @@ static void cik_sdma_vm_write_pte(struct amdgpu_ib *ib, uint64_t pe,
 }
 
 /**
- * cik_sdma_vm_set_pages - update the page tables using sDMA
+ * cik_sdma_vm_set_pte_pde - update the page tables using sDMA
  *
  * @ib: indirect buffer to fill with commands
  * @pe: addr of the page entry
@@ -824,18 +789,19 @@ static void cik_sdma_vm_set_pte_pde(struct amdgpu_ib *ib, uint64_t pe,
 }
 
 /**
- * cik_sdma_vm_pad_ib - pad the IB to the required number of dw
+ * cik_sdma_ring_pad_ib - pad the IB to the required number of dw
  *
+ * @ring: amdgpu_ring structure holding ring information
  * @ib: indirect buffer to fill with padding
  *
  */
 static void cik_sdma_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib)
 {
-	struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring);
+	struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring);
 	u32 pad_count;
 	int i;
 
-	pad_count = (8 - (ib->length_dw & 0x7)) % 8;
+	pad_count = (-ib->length_dw) & 7;
 	for (i = 0; i < pad_count; i++)
 		if (sdma && sdma->burst_nop && (i == 0))
 			ib->ptr[ib->length_dw++] =
@@ -866,7 +832,7 @@ static void cik_sdma_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
 	amdgpu_ring_write(ring, addr & 0xfffffffc);
 	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xffffffff);
 	amdgpu_ring_write(ring, seq); /* reference */
-	amdgpu_ring_write(ring, 0xfffffff); /* mask */
+	amdgpu_ring_write(ring, 0xffffffff); /* mask */
 	amdgpu_ring_write(ring, (0xfff << 16) | 4); /* retry count, poll interval */
 }
 
@@ -874,29 +840,19 @@ static void cik_sdma_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
  * cik_sdma_ring_emit_vm_flush - cik vm flush using sDMA
  *
  * @ring: amdgpu_ring pointer
- * @vm: amdgpu_vm pointer
+ * @vmid: vmid number to use
+ * @pd_addr: address
  *
  * Update the page table base and flush the VM TLB
  * using sDMA (CIK).
  */
 static void cik_sdma_ring_emit_vm_flush(struct amdgpu_ring *ring,
-					unsigned vm_id, uint64_t pd_addr)
+					unsigned vmid, uint64_t pd_addr)
 {
 	u32 extra_bits = (SDMA_POLL_REG_MEM_EXTRA_OP(0) |
 			  SDMA_POLL_REG_MEM_EXTRA_FUNC(0)); /* always */
 
-	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_SRBM_WRITE, 0, 0xf000));
-	if (vm_id < 8) {
-		amdgpu_ring_write(ring, (mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + vm_id));
-	} else {
-		amdgpu_ring_write(ring, (mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vm_id - 8));
-	}
-	amdgpu_ring_write(ring, pd_addr >> 12);
-
-	/* flush TLB */
-	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_SRBM_WRITE, 0, 0xf000));
-	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST);
-	amdgpu_ring_write(ring, 1 << vm_id);
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
 
 	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_POLL_REG_MEM, 0, extra_bits));
 	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST << 2);
@@ -906,6 +862,14 @@ static void cik_sdma_ring_emit_vm_flush(struct amdgpu_ring *ring,
 	amdgpu_ring_write(ring, (0xfff << 16) | 10); /* retry count, poll interval */
 }
 
+static void cik_sdma_ring_emit_wreg(struct amdgpu_ring *ring,
+				    uint32_t reg, uint32_t val)
+{
+	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_SRBM_WRITE, 0, 0xf000));
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, val);
+}
+
 static void cik_enable_sdma_mgcg(struct amdgpu_device *adev,
 				 bool enable)
 {
@@ -955,12 +919,17 @@ static void cik_enable_sdma_mgls(struct amdgpu_device *adev,
 	}
 }
 
-static int cik_sdma_early_init(void *handle)
+static int cik_sdma_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
 
 	adev->sdma.num_instances = SDMA_MAX_INSTANCE;
 
+	r = cik_sdma_init_microcode(adev);
+	if (r)
+		return r;
+
 	cik_sdma_set_ring_funcs(adev);
 	cik_sdma_set_irq_funcs(adev);
 	cik_sdma_set_buffer_funcs(adev);
@@ -969,32 +938,26 @@ static int cik_sdma_early_init(void *handle)
 	return 0;
 }
 
-static int cik_sdma_sw_init(void *handle)
+static int cik_sdma_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_ring *ring;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int r, i;
 
-	r = cik_sdma_init_microcode(adev);
-	if (r) {
-		DRM_ERROR("Failed to load sdma firmware!\n");
-		return r;
-	}
-
 	/* SDMA trap event */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 224,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 224,
 			      &adev->sdma.trap_irq);
 	if (r)
 		return r;
 
 	/* SDMA Privileged inst */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 241,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 241,
 			      &adev->sdma.illegal_inst_irq);
 	if (r)
 		return r;
 
 	/* SDMA Privileged inst */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 247,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 247,
 			      &adev->sdma.illegal_inst_irq);
 	if (r)
 		return r;
@@ -1005,9 +968,9 @@ static int cik_sdma_sw_init(void *handle)
 		sprintf(ring->name, "sdma%d", i);
 		r = amdgpu_ring_init(adev, ring, 1024,
 				     &adev->sdma.trap_irq,
-				     (i == 0) ?
-				     AMDGPU_SDMA_IRQ_TRAP0 :
-				     AMDGPU_SDMA_IRQ_TRAP1);
+				     (i == 0) ? AMDGPU_SDMA_IRQ_INSTANCE0 :
+				     AMDGPU_SDMA_IRQ_INSTANCE1,
+				     AMDGPU_RING_PRIO_DEFAULT, NULL);
 		if (r)
 			return r;
 	}
@@ -1015,9 +978,9 @@ static int cik_sdma_sw_init(void *handle)
 	return r;
 }
 
-static int cik_sdma_sw_fini(void *handle)
+static int cik_sdma_sw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int i;
 
 	for (i = 0; i < adev->sdma.num_instances; i++)
@@ -1027,21 +990,16 @@ static int cik_sdma_sw_fini(void *handle)
 	return 0;
 }
 
-static int cik_sdma_hw_init(void *handle)
+static int cik_sdma_hw_init(struct amdgpu_ip_block *ip_block)
 {
-	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	r = cik_sdma_start(adev);
-	if (r)
-		return r;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	return r;
+	return cik_sdma_start(adev);
 }
 
-static int cik_sdma_hw_fini(void *handle)
+static int cik_sdma_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	cik_ctx_switch_enable(adev, false);
 	cik_sdma_enable(adev, false);
@@ -1049,25 +1007,21 @@ static int cik_sdma_hw_fini(void *handle)
 	return 0;
 }
 
-static int cik_sdma_suspend(void *handle)
+static int cik_sdma_suspend(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return cik_sdma_hw_fini(adev);
+	return cik_sdma_hw_fini(ip_block);
 }
 
-static int cik_sdma_resume(void *handle)
+static int cik_sdma_resume(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	cik_sdma_soft_reset(handle);
+	cik_sdma_soft_reset(ip_block);
 
-	return cik_sdma_hw_init(adev);
+	return cik_sdma_hw_init(ip_block);
 }
 
-static bool cik_sdma_is_idle(void *handle)
+static bool cik_sdma_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 tmp = RREG32(mmSRBM_STATUS2);
 
 	if (tmp & (SRBM_STATUS2__SDMA_BUSY_MASK |
@@ -1077,43 +1031,36 @@ static bool cik_sdma_is_idle(void *handle)
 	return true;
 }
 
-static int cik_sdma_wait_for_idle(void *handle)
+static int cik_sdma_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
 	unsigned i;
-	u32 tmp;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		tmp = RREG32(mmSRBM_STATUS2) & (SRBM_STATUS2__SDMA_BUSY_MASK |
-				SRBM_STATUS2__SDMA1_BUSY_MASK);
-
-		if (!tmp)
+		if (cik_sdma_is_idle(ip_block))
 			return 0;
 		udelay(1);
 	}
 	return -ETIMEDOUT;
 }
 
-static int cik_sdma_soft_reset(void *handle)
+static int cik_sdma_soft_reset(struct amdgpu_ip_block *ip_block)
 {
 	u32 srbm_soft_reset = 0;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	u32 tmp = RREG32(mmSRBM_STATUS2);
+	struct amdgpu_device *adev = ip_block->adev;
+	u32 tmp;
 
-	if (tmp & SRBM_STATUS2__SDMA_BUSY_MASK) {
-		/* sdma0 */
-		tmp = RREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET);
-		tmp |= SDMA0_F32_CNTL__HALT_MASK;
-		WREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET, tmp);
-		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA_MASK;
-	}
-	if (tmp & SRBM_STATUS2__SDMA1_BUSY_MASK) {
-		/* sdma1 */
-		tmp = RREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET);
-		tmp |= SDMA0_F32_CNTL__HALT_MASK;
-		WREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET, tmp);
-		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA1_MASK;
-	}
+	/* sdma0 */
+	tmp = RREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET);
+	tmp |= SDMA0_F32_CNTL__HALT_MASK;
+	WREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET, tmp);
+	srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA_MASK;
+
+	/* sdma1 */
+	tmp = RREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET);
+	tmp |= SDMA0_F32_CNTL__HALT_MASK;
+	WREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET, tmp);
+	srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA1_MASK;
 
 	if (srbm_soft_reset) {
 		tmp = RREG32(mmSRBM_SOFT_RESET);
@@ -1143,7 +1090,7 @@ static int cik_sdma_set_trap_irq_state(struct amdgpu_device *adev,
 	u32 sdma_cntl;
 
 	switch (type) {
-	case AMDGPU_SDMA_IRQ_TRAP0:
+	case AMDGPU_SDMA_IRQ_INSTANCE0:
 		switch (state) {
 		case AMDGPU_IRQ_STATE_DISABLE:
 			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET);
@@ -1159,7 +1106,7 @@ static int cik_sdma_set_trap_irq_state(struct amdgpu_device *adev,
 			break;
 		}
 		break;
-	case AMDGPU_SDMA_IRQ_TRAP1:
+	case AMDGPU_SDMA_IRQ_INSTANCE1:
 		switch (state) {
 		case AMDGPU_IRQ_STATE_DISABLE:
 			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET);
@@ -1226,16 +1173,19 @@ static int cik_sdma_process_illegal_inst_irq(struct amdgpu_device *adev,
 					     struct amdgpu_irq_src *source,
 					     struct amdgpu_iv_entry *entry)
 {
+	u8 instance_id;
+
 	DRM_ERROR("Illegal instruction in SDMA command stream\n");
-	schedule_work(&adev->reset_work);
+	instance_id = (entry->ring_id & 0x3) >> 0;
+	drm_sched_fault(&adev->sdma.instance[instance_id].ring.sched);
 	return 0;
 }
 
-static int cik_sdma_set_clockgating_state(void *handle,
+static int cik_sdma_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
 	bool gate = false;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (state == AMD_CG_STATE_GATE)
 		gate = true;
@@ -1246,7 +1196,7 @@ static int cik_sdma_set_clockgating_state(void *handle,
 	return 0;
 }
 
-static int cik_sdma_set_powergating_state(void *handle,
+static int cik_sdma_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	return 0;
@@ -1255,7 +1205,6 @@ static int cik_sdma_set_powergating_state(void *handle,
 static const struct amd_ip_funcs cik_sdma_ip_funcs = {
 	.name = "cik_sdma",
 	.early_init = cik_sdma_early_init,
-	.late_init = NULL,
 	.sw_init = cik_sdma_sw_init,
 	.sw_fini = cik_sdma_sw_fini,
 	.hw_init = cik_sdma_hw_init,
@@ -1279,9 +1228,9 @@ static const struct amdgpu_ring_funcs cik_sdma_ring_funcs = {
 	.set_wptr = cik_sdma_ring_set_wptr,
 	.emit_frame_size =
 		6 + /* cik_sdma_ring_emit_hdp_flush */
-		3 + /* cik_sdma_ring_emit_hdp_invalidate */
+		3 + /* hdp invalidate */
 		6 + /* cik_sdma_ring_emit_pipeline_sync */
-		12 + /* cik_sdma_ring_emit_vm_flush */
+		CIK_FLUSH_GPU_TLB_NUM_WREG * 3 + 6 + /* cik_sdma_ring_emit_vm_flush */
 		9 + 9 + 9, /* cik_sdma_ring_emit_fence x3 for user fence, vm fence */
 	.emit_ib_size = 7 + 4, /* cik_sdma_ring_emit_ib */
 	.emit_ib = cik_sdma_ring_emit_ib,
@@ -1289,19 +1238,21 @@ static const struct amdgpu_ring_funcs cik_sdma_ring_funcs = {
 	.emit_pipeline_sync = cik_sdma_ring_emit_pipeline_sync,
 	.emit_vm_flush = cik_sdma_ring_emit_vm_flush,
 	.emit_hdp_flush = cik_sdma_ring_emit_hdp_flush,
-	.emit_hdp_invalidate = cik_sdma_ring_emit_hdp_invalidate,
 	.test_ring = cik_sdma_ring_test_ring,
 	.test_ib = cik_sdma_ring_test_ib,
 	.insert_nop = cik_sdma_ring_insert_nop,
 	.pad_ib = cik_sdma_ring_pad_ib,
+	.emit_wreg = cik_sdma_ring_emit_wreg,
 };
 
 static void cik_sdma_set_ring_funcs(struct amdgpu_device *adev)
 {
 	int i;
 
-	for (i = 0; i < adev->sdma.num_instances; i++)
+	for (i = 0; i < adev->sdma.num_instances; i++) {
 		adev->sdma.instance[i].ring.funcs = &cik_sdma_ring_funcs;
+		adev->sdma.instance[i].ring.me = i;
+	}
 }
 
 static const struct amdgpu_irq_src_funcs cik_sdma_trap_irq_funcs = {
@@ -1323,10 +1274,11 @@ static void cik_sdma_set_irq_funcs(struct amdgpu_device *adev)
 /**
  * cik_sdma_emit_copy_buffer - copy buffer using the sDMA engine
  *
- * @ring: amdgpu_ring structure holding ring information
+ * @ib: indirect buffer to copy to
  * @src_offset: src GPU address
  * @dst_offset: dst GPU address
  * @byte_count: number of bytes to xfer
+ * @copy_flags: unused
  *
  * Copy GPU buffers using the DMA engine (CIK).
  * Used by the amdgpu ttm implementation to move pages if
@@ -1335,7 +1287,8 @@ static void cik_sdma_set_irq_funcs(struct amdgpu_device *adev)
 static void cik_sdma_emit_copy_buffer(struct amdgpu_ib *ib,
 				      uint64_t src_offset,
 				      uint64_t dst_offset,
-				      uint32_t byte_count)
+				      uint32_t byte_count,
+				      uint32_t copy_flags)
 {
 	ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_COPY, SDMA_COPY_SUB_OPCODE_LINEAR, 0);
 	ib->ptr[ib->length_dw++] = byte_count;
@@ -1349,7 +1302,7 @@ static void cik_sdma_emit_copy_buffer(struct amdgpu_ib *ib,
 /**
  * cik_sdma_emit_fill_buffer - fill buffer using the sDMA engine
  *
- * @ring: amdgpu_ring structure holding ring information
+ * @ib: indirect buffer to fill
  * @src_data: value to write to buffer
  * @dst_offset: dst GPU address
  * @byte_count: number of bytes to xfer
@@ -1380,14 +1333,14 @@ static const struct amdgpu_buffer_funcs cik_sdma_buffer_funcs = {
 
 static void cik_sdma_set_buffer_funcs(struct amdgpu_device *adev)
 {
-	if (adev->mman.buffer_funcs == NULL) {
-		adev->mman.buffer_funcs = &cik_sdma_buffer_funcs;
-		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
-	}
+	adev->mman.buffer_funcs = &cik_sdma_buffer_funcs;
+	adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
 }
 
 static const struct amdgpu_vm_pte_funcs cik_sdma_vm_pte_funcs = {
+	.copy_pte_num_dw = 7,
 	.copy_pte = cik_sdma_vm_copy_pte,
+
 	.write_pte = cik_sdma_vm_write_pte,
 	.set_pte_pde = cik_sdma_vm_set_pte_pde,
 };
@@ -1396,14 +1349,12 @@ static void cik_sdma_set_vm_pte_funcs(struct amdgpu_device *adev)
 {
 	unsigned i;
 
-	if (adev->vm_manager.vm_pte_funcs == NULL) {
-		adev->vm_manager.vm_pte_funcs = &cik_sdma_vm_pte_funcs;
-		for (i = 0; i < adev->sdma.num_instances; i++)
-			adev->vm_manager.vm_pte_rings[i] =
-				&adev->sdma.instance[i].ring;
-
-		adev->vm_manager.vm_pte_num_rings = adev->sdma.num_instances;
+	adev->vm_manager.vm_pte_funcs = &cik_sdma_vm_pte_funcs;
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		adev->vm_manager.vm_pte_scheds[i] =
+			&adev->sdma.instance[i].ring.sched;
 	}
+	adev->vm_manager.vm_pte_num_scheds = adev->sdma.num_instances;
 }
 
 const struct amdgpu_ip_block_version cik_sdma_ip_block =
diff --git a/drivers/gpu/drm/amd/amdgpu/cikd.h b/drivers/gpu/drm/amd/amdgpu/cikd.h
index 6a9e38a3d2a0..8aca4f2734f2 100644
--- a/drivers/gpu/drm/amd/amdgpu/cikd.h
+++ b/drivers/gpu/drm/amd/amdgpu/cikd.h
@@ -51,10 +51,14 @@
 #define HPD4_REGISTER_OFFSET                 (0x1813 - 0x1807)
 #define HPD5_REGISTER_OFFSET                 (0x1816 - 0x1807)
 
-#define BONAIRE_GB_ADDR_CONFIG_GOLDEN        0x12010001
-#define HAWAII_GB_ADDR_CONFIG_GOLDEN         0x12011003
-
-#define AMDGPU_NUM_OF_VMIDS	8
+/* audio endpt instance offsets */
+#define AUD0_REGISTER_OFFSET                 (0x1780 - 0x1780)
+#define AUD1_REGISTER_OFFSET                 (0x1786 - 0x1780)
+#define AUD2_REGISTER_OFFSET                 (0x178c - 0x1780)
+#define AUD3_REGISTER_OFFSET                 (0x1792 - 0x1780)
+#define AUD4_REGISTER_OFFSET                 (0x1798 - 0x1780)
+#define AUD5_REGISTER_OFFSET                 (0x179d - 0x1780)
+#define AUD6_REGISTER_OFFSET                 (0x17a4 - 0x1780)
 
 #define		PIPEID(x)					((x) << 0)
 #define		MEID(x)						((x) << 2)
@@ -366,6 +370,7 @@
 		 * 1 - Stream
 		 * 2 - Bypass
 		 */
+#define		EOP_EXEC				(1 << 28) /* For Trailing Fence */
 #define		DATA_SEL(x)                             ((x) << 29)
 		/* 0 - discard
 		 * 1 - send low 32bit data
@@ -450,7 +455,7 @@
 #              define PACKET3_DMA_DATA_CMD_SAIC    (1 << 28)
 #              define PACKET3_DMA_DATA_CMD_DAIC    (1 << 29)
 #              define PACKET3_DMA_DATA_CMD_RAW_WAIT  (1 << 30)
-#define	PACKET3_AQUIRE_MEM				0x58
+#define	PACKET3_ACQUIRE_MEM				0x58
 #define	PACKET3_REWIND					0x59
 #define	PACKET3_LOAD_UCONFIG_REG			0x5E
 #define	PACKET3_LOAD_SH_REG				0x5F
@@ -562,7 +567,7 @@
 #define	PRIVATE_BASE(x)	((x) << 0) /* scratch */
 #define	SHARED_BASE(x)	((x) << 16) /* LDS */
 
-#define KFD_CIK_SDMA_QUEUE_OFFSET	0x200
+#define KFD_CIK_SDMA_QUEUE_OFFSET (mmSDMA0_RLC1_RB_CNTL - mmSDMA0_RLC0_RB_CNTL)
 
 /* valid for both DEFAULT_MTYPE and APE1_MTYPE */
 enum {
diff --git a/drivers/gpu/drm/amd/amdgpu/clearstate_gfx10.h b/drivers/gpu/drm/amd/amdgpu/clearstate_gfx10.h
new file mode 100644
index 000000000000..27a2ff0a07d2
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/clearstate_gfx10.h
@@ -0,0 +1,975 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+static const unsigned int gfx10_SECT_CONTEXT_def_1[] = {
+	0x00000000, // DB_RENDER_CONTROL
+	0x00000000, // DB_COUNT_CONTROL
+	0x00000000, // DB_DEPTH_VIEW
+	0x00000000, // DB_RENDER_OVERRIDE
+	0x00000000, // DB_RENDER_OVERRIDE2
+	0x00000000, // DB_HTILE_DATA_BASE
+	0x00000000, // HOLE
+	0x00000000, // DB_DEPTH_SIZE_XY
+	0x00000000, // DB_DEPTH_BOUNDS_MIN
+	0x00000000, // DB_DEPTH_BOUNDS_MAX
+	0x00000000, // DB_STENCIL_CLEAR
+	0x00000000, // DB_DEPTH_CLEAR
+	0x00000000, // PA_SC_SCREEN_SCISSOR_TL
+	0x40004000, // PA_SC_SCREEN_SCISSOR_BR
+	0x00000000, // DB_DFSM_CONTROL
+	0x00000000, // DB_DEPTH_INFO
+	0x00000000, // DB_Z_INFO
+	0x00000000, // DB_STENCIL_INFO
+	0x00000000, // DB_Z_READ_BASE
+	0x00000000, // DB_STENCIL_READ_BASE
+	0x00000000, // DB_Z_WRITE_BASE
+	0x00000000, // DB_STENCIL_WRITE_BASE
+	0x00000000, // DB_DEPTH_SIZE
+	0x00000000, // DB_DEPTH_SLICE
+	0x00000000, // DB_Z_INFO2
+	0x00000000, // DB_STENCIL_INFO2
+	0x00000000, // DB_Z_READ_BASE_HI
+	0x00000000, // DB_STENCIL_READ_BASE_HI
+	0x00000000, // DB_Z_WRITE_BASE_HI
+	0x00000000, // DB_STENCIL_WRITE_BASE_HI
+	0x00000000, // DB_HTILE_DATA_BASE_HI
+	0x00150055, // DB_RMI_L2_CACHE_CONTROL
+	0x00000000, // TA_BC_BASE_ADDR
+	0x00000000, // TA_BC_BASE_ADDR_HI
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0x00000000, // COHER_DEST_BASE_HI_0
+	0x00000000, // COHER_DEST_BASE_HI_1
+	0x00000000, // COHER_DEST_BASE_HI_2
+	0x00000000, // COHER_DEST_BASE_HI_3
+	0x00000000, // COHER_DEST_BASE_2
+	0x00000000, // COHER_DEST_BASE_3
+	0x00000000, // PA_SC_WINDOW_OFFSET
+	0x80000000, // PA_SC_WINDOW_SCISSOR_TL
+	0x40004000, // PA_SC_WINDOW_SCISSOR_BR
+	0x0000ffff, // PA_SC_CLIPRECT_RULE
+	0x00000000, // PA_SC_CLIPRECT_0_TL
+	0x40004000, // PA_SC_CLIPRECT_0_BR
+	0x00000000, // PA_SC_CLIPRECT_1_TL
+	0x40004000, // PA_SC_CLIPRECT_1_BR
+	0x00000000, // PA_SC_CLIPRECT_2_TL
+	0x40004000, // PA_SC_CLIPRECT_2_BR
+	0x00000000, // PA_SC_CLIPRECT_3_TL
+	0x40004000, // PA_SC_CLIPRECT_3_BR
+	0xaa99aaaa, // PA_SC_EDGERULE
+	0x00000000, // PA_SU_HARDWARE_SCREEN_OFFSET
+	0xffffffff, // CB_TARGET_MASK
+	0xffffffff, // CB_SHADER_MASK
+	0x80000000, // PA_SC_GENERIC_SCISSOR_TL
+	0x40004000, // PA_SC_GENERIC_SCISSOR_BR
+	0x00000000, // COHER_DEST_BASE_0
+	0x00000000, // COHER_DEST_BASE_1
+	0x80000000, // PA_SC_VPORT_SCISSOR_0_TL
+	0x40004000, // PA_SC_VPORT_SCISSOR_0_BR
+	0x80000000, // PA_SC_VPORT_SCISSOR_1_TL
+	0x40004000, // PA_SC_VPORT_SCISSOR_1_BR
+	0x80000000, // PA_SC_VPORT_SCISSOR_2_TL
+	0x40004000, // PA_SC_VPORT_SCISSOR_2_BR
+	0x80000000, // PA_SC_VPORT_SCISSOR_3_TL
+	0x40004000, // PA_SC_VPORT_SCISSOR_3_BR
+	0x80000000, // PA_SC_VPORT_SCISSOR_4_TL
+	0x40004000, // PA_SC_VPORT_SCISSOR_4_BR
+	0x80000000, // PA_SC_VPORT_SCISSOR_5_TL
+	0x40004000, // PA_SC_VPORT_SCISSOR_5_BR
+	0x80000000, // PA_SC_VPORT_SCISSOR_6_TL
+	0x40004000, // PA_SC_VPORT_SCISSOR_6_BR
+	0x80000000, // PA_SC_VPORT_SCISSOR_7_TL
+	0x40004000, // PA_SC_VPORT_SCISSOR_7_BR
+	0x80000000, // PA_SC_VPORT_SCISSOR_8_TL
+	0x40004000, // PA_SC_VPORT_SCISSOR_8_BR
+	0x80000000, // PA_SC_VPORT_SCISSOR_9_TL
+	0x40004000, // PA_SC_VPORT_SCISSOR_9_BR
+	0x80000000, // PA_SC_VPORT_SCISSOR_10_TL
+	0x40004000, // PA_SC_VPORT_SCISSOR_10_BR
+	0x80000000, // PA_SC_VPORT_SCISSOR_11_TL
+	0x40004000, // PA_SC_VPORT_SCISSOR_11_BR
+	0x80000000, // PA_SC_VPORT_SCISSOR_12_TL
+	0x40004000, // PA_SC_VPORT_SCISSOR_12_BR
+	0x80000000, // PA_SC_VPORT_SCISSOR_13_TL
+	0x40004000, // PA_SC_VPORT_SCISSOR_13_BR
+	0x80000000, // PA_SC_VPORT_SCISSOR_14_TL
+	0x40004000, // PA_SC_VPORT_SCISSOR_14_BR
+	0x80000000, // PA_SC_VPORT_SCISSOR_15_TL
+	0x40004000, // PA_SC_VPORT_SCISSOR_15_BR
+	0x00000000, // PA_SC_VPORT_ZMIN_0
+	0x3f800000, // PA_SC_VPORT_ZMAX_0
+	0x00000000, // PA_SC_VPORT_ZMIN_1
+	0x3f800000, // PA_SC_VPORT_ZMAX_1
+	0x00000000, // PA_SC_VPORT_ZMIN_2
+	0x3f800000, // PA_SC_VPORT_ZMAX_2
+	0x00000000, // PA_SC_VPORT_ZMIN_3
+	0x3f800000, // PA_SC_VPORT_ZMAX_3
+	0x00000000, // PA_SC_VPORT_ZMIN_4
+	0x3f800000, // PA_SC_VPORT_ZMAX_4
+	0x00000000, // PA_SC_VPORT_ZMIN_5
+	0x3f800000, // PA_SC_VPORT_ZMAX_5
+	0x00000000, // PA_SC_VPORT_ZMIN_6
+	0x3f800000, // PA_SC_VPORT_ZMAX_6
+	0x00000000, // PA_SC_VPORT_ZMIN_7
+	0x3f800000, // PA_SC_VPORT_ZMAX_7
+	0x00000000, // PA_SC_VPORT_ZMIN_8
+	0x3f800000, // PA_SC_VPORT_ZMAX_8
+	0x00000000, // PA_SC_VPORT_ZMIN_9
+	0x3f800000, // PA_SC_VPORT_ZMAX_9
+	0x00000000, // PA_SC_VPORT_ZMIN_10
+	0x3f800000, // PA_SC_VPORT_ZMAX_10
+	0x00000000, // PA_SC_VPORT_ZMIN_11
+	0x3f800000, // PA_SC_VPORT_ZMAX_11
+	0x00000000, // PA_SC_VPORT_ZMIN_12
+	0x3f800000, // PA_SC_VPORT_ZMAX_12
+	0x00000000, // PA_SC_VPORT_ZMIN_13
+	0x3f800000, // PA_SC_VPORT_ZMAX_13
+	0x00000000, // PA_SC_VPORT_ZMIN_14
+	0x3f800000, // PA_SC_VPORT_ZMAX_14
+	0x00000000, // PA_SC_VPORT_ZMIN_15
+	0x3f800000, // PA_SC_VPORT_ZMAX_15
+	0x00000000, // PA_SC_RASTER_CONFIG
+	0x00000000, // PA_SC_RASTER_CONFIG_1
+	0x00000000, // PA_SC_SCREEN_EXTENT_CONTROL
+};
+static const unsigned int gfx10_SECT_CONTEXT_def_2[] = {
+	0x00000000, // CP_PERFMON_CNTX_CNTL
+	0x00000000, // CP_RINGID
+	0x00000000, // CP_VMID
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0x00000000, // PA_SC_RIGHT_VERT_GRID
+	0x00000000, // PA_SC_LEFT_VERT_GRID
+	0x00000000, // PA_SC_HORIZ_GRID
+	0x00000000, // HOLE
+	0x00000000, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0xffffffff, // VGT_MAX_VTX_INDX
+	0x00000000, // VGT_MIN_VTX_INDX
+	0x00000000, // VGT_INDX_OFFSET
+	0x00000000, // VGT_MULTI_PRIM_IB_RESET_INDX
+	0x00550055, // CB_RMI_GL2_CACHE_CONTROL
+	0x00000000, // CB_BLEND_RED
+	0x00000000, // CB_BLEND_GREEN
+	0x00000000, // CB_BLEND_BLUE
+	0x00000000, // CB_BLEND_ALPHA
+	0x00000000, // CB_DCC_CONTROL
+	0x00000000, // CB_COVERAGE_OUT_CONTROL
+	0x00000000, // DB_STENCIL_CONTROL
+	0x01000000, // DB_STENCILREFMASK
+	0x01000000, // DB_STENCILREFMASK_BF
+	0, // HOLE
+	0x00000000, // PA_CL_VPORT_XSCALE
+	0x00000000, // PA_CL_VPORT_XOFFSET
+	0x00000000, // PA_CL_VPORT_YSCALE
+	0x00000000, // PA_CL_VPORT_YOFFSET
+	0x00000000, // PA_CL_VPORT_ZSCALE
+	0x00000000, // PA_CL_VPORT_ZOFFSET
+	0x00000000, // PA_CL_VPORT_XSCALE_1
+	0x00000000, // PA_CL_VPORT_XOFFSET_1
+	0x00000000, // PA_CL_VPORT_YSCALE_1
+	0x00000000, // PA_CL_VPORT_YOFFSET_1
+	0x00000000, // PA_CL_VPORT_ZSCALE_1
+	0x00000000, // PA_CL_VPORT_ZOFFSET_1
+	0x00000000, // PA_CL_VPORT_XSCALE_2
+	0x00000000, // PA_CL_VPORT_XOFFSET_2
+	0x00000000, // PA_CL_VPORT_YSCALE_2
+	0x00000000, // PA_CL_VPORT_YOFFSET_2
+	0x00000000, // PA_CL_VPORT_ZSCALE_2
+	0x00000000, // PA_CL_VPORT_ZOFFSET_2
+	0x00000000, // PA_CL_VPORT_XSCALE_3
+	0x00000000, // PA_CL_VPORT_XOFFSET_3
+	0x00000000, // PA_CL_VPORT_YSCALE_3
+	0x00000000, // PA_CL_VPORT_YOFFSET_3
+	0x00000000, // PA_CL_VPORT_ZSCALE_3
+	0x00000000, // PA_CL_VPORT_ZOFFSET_3
+	0x00000000, // PA_CL_VPORT_XSCALE_4
+	0x00000000, // PA_CL_VPORT_XOFFSET_4
+	0x00000000, // PA_CL_VPORT_YSCALE_4
+	0x00000000, // PA_CL_VPORT_YOFFSET_4
+	0x00000000, // PA_CL_VPORT_ZSCALE_4
+	0x00000000, // PA_CL_VPORT_ZOFFSET_4
+	0x00000000, // PA_CL_VPORT_XSCALE_5
+	0x00000000, // PA_CL_VPORT_XOFFSET_5
+	0x00000000, // PA_CL_VPORT_YSCALE_5
+	0x00000000, // PA_CL_VPORT_YOFFSET_5
+	0x00000000, // PA_CL_VPORT_ZSCALE_5
+	0x00000000, // PA_CL_VPORT_ZOFFSET_5
+	0x00000000, // PA_CL_VPORT_XSCALE_6
+	0x00000000, // PA_CL_VPORT_XOFFSET_6
+	0x00000000, // PA_CL_VPORT_YSCALE_6
+	0x00000000, // PA_CL_VPORT_YOFFSET_6
+	0x00000000, // PA_CL_VPORT_ZSCALE_6
+	0x00000000, // PA_CL_VPORT_ZOFFSET_6
+	0x00000000, // PA_CL_VPORT_XSCALE_7
+	0x00000000, // PA_CL_VPORT_XOFFSET_7
+	0x00000000, // PA_CL_VPORT_YSCALE_7
+	0x00000000, // PA_CL_VPORT_YOFFSET_7
+	0x00000000, // PA_CL_VPORT_ZSCALE_7
+	0x00000000, // PA_CL_VPORT_ZOFFSET_7
+	0x00000000, // PA_CL_VPORT_XSCALE_8
+	0x00000000, // PA_CL_VPORT_XOFFSET_8
+	0x00000000, // PA_CL_VPORT_YSCALE_8
+	0x00000000, // PA_CL_VPORT_YOFFSET_8
+	0x00000000, // PA_CL_VPORT_ZSCALE_8
+	0x00000000, // PA_CL_VPORT_ZOFFSET_8
+	0x00000000, // PA_CL_VPORT_XSCALE_9
+	0x00000000, // PA_CL_VPORT_XOFFSET_9
+	0x00000000, // PA_CL_VPORT_YSCALE_9
+	0x00000000, // PA_CL_VPORT_YOFFSET_9
+	0x00000000, // PA_CL_VPORT_ZSCALE_9
+	0x00000000, // PA_CL_VPORT_ZOFFSET_9
+	0x00000000, // PA_CL_VPORT_XSCALE_10
+	0x00000000, // PA_CL_VPORT_XOFFSET_10
+	0x00000000, // PA_CL_VPORT_YSCALE_10
+	0x00000000, // PA_CL_VPORT_YOFFSET_10
+	0x00000000, // PA_CL_VPORT_ZSCALE_10
+	0x00000000, // PA_CL_VPORT_ZOFFSET_10
+	0x00000000, // PA_CL_VPORT_XSCALE_11
+	0x00000000, // PA_CL_VPORT_XOFFSET_11
+	0x00000000, // PA_CL_VPORT_YSCALE_11
+	0x00000000, // PA_CL_VPORT_YOFFSET_11
+	0x00000000, // PA_CL_VPORT_ZSCALE_11
+	0x00000000, // PA_CL_VPORT_ZOFFSET_11
+	0x00000000, // PA_CL_VPORT_XSCALE_12
+	0x00000000, // PA_CL_VPORT_XOFFSET_12
+	0x00000000, // PA_CL_VPORT_YSCALE_12
+	0x00000000, // PA_CL_VPORT_YOFFSET_12
+	0x00000000, // PA_CL_VPORT_ZSCALE_12
+	0x00000000, // PA_CL_VPORT_ZOFFSET_12
+	0x00000000, // PA_CL_VPORT_XSCALE_13
+	0x00000000, // PA_CL_VPORT_XOFFSET_13
+	0x00000000, // PA_CL_VPORT_YSCALE_13
+	0x00000000, // PA_CL_VPORT_YOFFSET_13
+	0x00000000, // PA_CL_VPORT_ZSCALE_13
+	0x00000000, // PA_CL_VPORT_ZOFFSET_13
+	0x00000000, // PA_CL_VPORT_XSCALE_14
+	0x00000000, // PA_CL_VPORT_XOFFSET_14
+	0x00000000, // PA_CL_VPORT_YSCALE_14
+	0x00000000, // PA_CL_VPORT_YOFFSET_14
+	0x00000000, // PA_CL_VPORT_ZSCALE_14
+	0x00000000, // PA_CL_VPORT_ZOFFSET_14
+	0x00000000, // PA_CL_VPORT_XSCALE_15
+	0x00000000, // PA_CL_VPORT_XOFFSET_15
+	0x00000000, // PA_CL_VPORT_YSCALE_15
+	0x00000000, // PA_CL_VPORT_YOFFSET_15
+	0x00000000, // PA_CL_VPORT_ZSCALE_15
+	0x00000000, // PA_CL_VPORT_ZOFFSET_15
+	0x00000000, // PA_CL_UCP_0_X
+	0x00000000, // PA_CL_UCP_0_Y
+	0x00000000, // PA_CL_UCP_0_Z
+	0x00000000, // PA_CL_UCP_0_W
+	0x00000000, // PA_CL_UCP_1_X
+	0x00000000, // PA_CL_UCP_1_Y
+	0x00000000, // PA_CL_UCP_1_Z
+	0x00000000, // PA_CL_UCP_1_W
+	0x00000000, // PA_CL_UCP_2_X
+	0x00000000, // PA_CL_UCP_2_Y
+	0x00000000, // PA_CL_UCP_2_Z
+	0x00000000, // PA_CL_UCP_2_W
+	0x00000000, // PA_CL_UCP_3_X
+	0x00000000, // PA_CL_UCP_3_Y
+	0x00000000, // PA_CL_UCP_3_Z
+	0x00000000, // PA_CL_UCP_3_W
+	0x00000000, // PA_CL_UCP_4_X
+	0x00000000, // PA_CL_UCP_4_Y
+	0x00000000, // PA_CL_UCP_4_Z
+	0x00000000, // PA_CL_UCP_4_W
+	0x00000000, // PA_CL_UCP_5_X
+	0x00000000, // PA_CL_UCP_5_Y
+	0x00000000, // PA_CL_UCP_5_Z
+	0x00000000, // PA_CL_UCP_5_W
+	0x00000000, // PA_CL_PROG_NEAR_CLIP_Z
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0x00000000, // SPI_PS_INPUT_CNTL_0
+	0x00000000, // SPI_PS_INPUT_CNTL_1
+	0x00000000, // SPI_PS_INPUT_CNTL_2
+	0x00000000, // SPI_PS_INPUT_CNTL_3
+	0x00000000, // SPI_PS_INPUT_CNTL_4
+	0x00000000, // SPI_PS_INPUT_CNTL_5
+	0x00000000, // SPI_PS_INPUT_CNTL_6
+	0x00000000, // SPI_PS_INPUT_CNTL_7
+	0x00000000, // SPI_PS_INPUT_CNTL_8
+	0x00000000, // SPI_PS_INPUT_CNTL_9
+	0x00000000, // SPI_PS_INPUT_CNTL_10
+	0x00000000, // SPI_PS_INPUT_CNTL_11
+	0x00000000, // SPI_PS_INPUT_CNTL_12
+	0x00000000, // SPI_PS_INPUT_CNTL_13
+	0x00000000, // SPI_PS_INPUT_CNTL_14
+	0x00000000, // SPI_PS_INPUT_CNTL_15
+	0x00000000, // SPI_PS_INPUT_CNTL_16
+	0x00000000, // SPI_PS_INPUT_CNTL_17
+	0x00000000, // SPI_PS_INPUT_CNTL_18
+	0x00000000, // SPI_PS_INPUT_CNTL_19
+	0x00000000, // SPI_PS_INPUT_CNTL_20
+	0x00000000, // SPI_PS_INPUT_CNTL_21
+	0x00000000, // SPI_PS_INPUT_CNTL_22
+	0x00000000, // SPI_PS_INPUT_CNTL_23
+	0x00000000, // SPI_PS_INPUT_CNTL_24
+	0x00000000, // SPI_PS_INPUT_CNTL_25
+	0x00000000, // SPI_PS_INPUT_CNTL_26
+	0x00000000, // SPI_PS_INPUT_CNTL_27
+	0x00000000, // SPI_PS_INPUT_CNTL_28
+	0x00000000, // SPI_PS_INPUT_CNTL_29
+	0x00000000, // SPI_PS_INPUT_CNTL_30
+	0x00000000, // SPI_PS_INPUT_CNTL_31
+	0x00000000, // SPI_VS_OUT_CONFIG
+	0, // HOLE
+	0x00000000, // SPI_PS_INPUT_ENA
+	0x00000000, // SPI_PS_INPUT_ADDR
+	0x00000000, // SPI_INTERP_CONTROL_0
+	0x00000002, // SPI_PS_IN_CONTROL
+	0, // HOLE
+	0x00000000, // SPI_BARYC_CNTL
+	0, // HOLE
+	0x00000000, // SPI_TMPRING_SIZE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0x00000000, // SPI_SHADER_IDX_FORMAT
+	0x00000000, // SPI_SHADER_POS_FORMAT
+	0x00000000, // SPI_SHADER_Z_FORMAT
+	0x00000000, // SPI_SHADER_COL_FORMAT
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0x00000000, // SX_PS_DOWNCONVERT
+	0x00000000, // SX_BLEND_OPT_EPSILON
+	0x00000000, // SX_BLEND_OPT_CONTROL
+	0x00000000, // SX_MRT0_BLEND_OPT
+	0x00000000, // SX_MRT1_BLEND_OPT
+	0x00000000, // SX_MRT2_BLEND_OPT
+	0x00000000, // SX_MRT3_BLEND_OPT
+	0x00000000, // SX_MRT4_BLEND_OPT
+	0x00000000, // SX_MRT5_BLEND_OPT
+	0x00000000, // SX_MRT6_BLEND_OPT
+	0x00000000, // SX_MRT7_BLEND_OPT
+	0x00000000, // CB_BLEND0_CONTROL
+	0x00000000, // CB_BLEND1_CONTROL
+	0x00000000, // CB_BLEND2_CONTROL
+	0x00000000, // CB_BLEND3_CONTROL
+	0x00000000, // CB_BLEND4_CONTROL
+	0x00000000, // CB_BLEND5_CONTROL
+	0x00000000, // CB_BLEND6_CONTROL
+	0x00000000, // CB_BLEND7_CONTROL
+};
+static const unsigned int gfx10_SECT_CONTEXT_def_3[] = {
+	0x00000000, // PA_CL_POINT_X_RAD
+	0x00000000, // PA_CL_POINT_Y_RAD
+	0x00000000, // PA_CL_POINT_SIZE
+	0x00000000, // PA_CL_POINT_CULL_RAD
+};
+static const unsigned int gfx10_SECT_CONTEXT_def_4[] = {
+	0x00000000, // VGT_GS_MAX_PRIMS_PER_SUBGROUP
+	0x00000000, // DB_DEPTH_CONTROL
+	0x00000000, // DB_EQAA
+	0x00000000, // CB_COLOR_CONTROL
+	0x00000000, // DB_SHADER_CONTROL
+	0x00090000, // PA_CL_CLIP_CNTL
+	0x00000004, // PA_SU_SC_MODE_CNTL
+	0x00000000, // PA_CL_VTE_CNTL
+	0x00000000, // PA_CL_VS_OUT_CNTL
+	0x00000000, // PA_CL_NANINF_CNTL
+	0x00000000, // PA_SU_LINE_STIPPLE_CNTL
+	0x00000000, // PA_SU_LINE_STIPPLE_SCALE
+	0x00000000, // PA_SU_PRIM_FILTER_CNTL
+	0x00000000, // PA_SU_SMALL_PRIM_FILTER_CNTL
+	0x00000000, // PA_CL_OBJPRIM_ID_CNTL
+	0x00000000, // PA_CL_NGG_CNTL
+	0x00000000, // PA_SU_OVER_RASTERIZATION_CNTL
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0x00000000, // PA_SU_POINT_SIZE
+	0x00000000, // PA_SU_POINT_MINMAX
+	0x00000000, // PA_SU_LINE_CNTL
+	0x00000000, // PA_SC_LINE_STIPPLE
+	0x00000000, // VGT_OUTPUT_PATH_CNTL
+	0x00000000, // VGT_HOS_CNTL
+	0x00000000, // VGT_HOS_MAX_TESS_LEVEL
+	0x00000000, // VGT_HOS_MIN_TESS_LEVEL
+	0x00000000, // VGT_HOS_REUSE_DEPTH
+	0x00000000, // VGT_GROUP_PRIM_TYPE
+	0x00000000, // VGT_GROUP_FIRST_DECR
+	0x00000000, // VGT_GROUP_DECR
+	0x00000000, // VGT_GROUP_VECT_0_CNTL
+	0x00000000, // VGT_GROUP_VECT_1_CNTL
+	0x00000000, // VGT_GROUP_VECT_0_FMT_CNTL
+	0x00000000, // VGT_GROUP_VECT_1_FMT_CNTL
+	0x00000000, // VGT_GS_MODE
+	0x00000000, // VGT_GS_ONCHIP_CNTL
+	0x00000000, // PA_SC_MODE_CNTL_0
+	0x00000000, // PA_SC_MODE_CNTL_1
+	0x00000000, // VGT_ENHANCE
+	0x00000100, // VGT_GS_PER_ES
+	0x00000080, // VGT_ES_PER_GS
+	0x00000002, // VGT_GS_PER_VS
+	0x00000000, // VGT_GSVS_RING_OFFSET_1
+	0x00000000, // VGT_GSVS_RING_OFFSET_2
+	0x00000000, // VGT_GSVS_RING_OFFSET_3
+	0x00000000, // VGT_GS_OUT_PRIM_TYPE
+	0x00000000, // IA_ENHANCE
+};
+static const unsigned int gfx10_SECT_CONTEXT_def_5[] = {
+	0x00000000, // WD_ENHANCE
+	0x00000000, // VGT_PRIMITIVEID_EN
+};
+static const unsigned int gfx10_SECT_CONTEXT_def_6[] = {
+	0x00000000, // VGT_PRIMITIVEID_RESET
+};
+static const unsigned int gfx10_SECT_CONTEXT_def_7[] = {
+	0x00000000, // VGT_MULTI_PRIM_IB_RESET_EN
+	0x00000000, // VGT_DRAW_PAYLOAD_CNTL
+	0x00000000, // HOLE
+	0x00000000, // VGT_INSTANCE_STEP_RATE_0
+	0x00000000, // VGT_INSTANCE_STEP_RATE_1
+	0x000000ff, // IA_MULTI_VGT_PARAM
+	0x00000000, // VGT_ESGS_RING_ITEMSIZE
+	0x00000000, // VGT_GSVS_RING_ITEMSIZE
+	0x00000000, // VGT_REUSE_OFF
+	0x00000000, // VGT_VTX_CNT_EN
+	0x00000000, // DB_HTILE_SURFACE
+	0x00000000, // DB_SRESULTS_COMPARE_STATE0
+	0x00000000, // DB_SRESULTS_COMPARE_STATE1
+	0x00000000, // DB_PRELOAD_CONTROL
+	0, // HOLE
+	0x00000000, // VGT_STRMOUT_BUFFER_SIZE_0
+	0x00000000, // VGT_STRMOUT_VTX_STRIDE_0
+	0, // HOLE
+	0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_0
+	0x00000000, // VGT_STRMOUT_BUFFER_SIZE_1
+	0x00000000, // VGT_STRMOUT_VTX_STRIDE_1
+	0, // HOLE
+	0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_1
+	0x00000000, // VGT_STRMOUT_BUFFER_SIZE_2
+	0x00000000, // VGT_STRMOUT_VTX_STRIDE_2
+	0, // HOLE
+	0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_2
+	0x00000000, // VGT_STRMOUT_BUFFER_SIZE_3
+	0x00000000, // VGT_STRMOUT_VTX_STRIDE_3
+	0, // HOLE
+	0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_3
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_OFFSET
+	0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_BUFFER_FILLED_SIZE
+	0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_VERTEX_STRIDE
+	0, // HOLE
+	0x00000000, // VGT_GS_MAX_VERT_OUT
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0, // HOLE
+	0x00000000, // VGT_TESS_DISTRIBUTION
+	0x00000000, // VGT_SHADER_STAGES_EN
+	0x00000000, // VGT_LS_HS_CONFIG
+	0x00000000, // VGT_GS_VERT_ITEMSIZE
+	0x00000000, // VGT_GS_VERT_ITEMSIZE_1
+	0x00000000, // VGT_GS_VERT_ITEMSIZE_2
+	0x00000000, // VGT_GS_VERT_ITEMSIZE_3
+	0x00000000, // VGT_TF_PARAM
+	0x00000000, // DB_ALPHA_TO_MASK
+	0x00000000, // VGT_DISPATCH_DRAW_INDEX
+	0x00000000, // PA_SU_POLY_OFFSET_DB_FMT_CNTL
+	0x00000000, // PA_SU_POLY_OFFSET_CLAMP
+	0x00000000, // PA_SU_POLY_OFFSET_FRONT_SCALE
+	0x00000000, // PA_SU_POLY_OFFSET_FRONT_OFFSET
+	0x00000000, // PA_SU_POLY_OFFSET_BACK_SCALE
+	0x00000000, // PA_SU_POLY_OFFSET_BACK_OFFSET
+	0x00000000, // VGT_GS_INSTANCE_CNT
+	0x00000000, // VGT_STRMOUT_CONFIG
+	0x00000000, // VGT_STRMOUT_BUFFER_CONFIG
+};
+static const unsigned int gfx10_SECT_CONTEXT_def_8[] = {
+	0x00000000, // PA_SC_CENTROID_PRIORITY_0
+	0x00000000, // PA_SC_CENTROID_PRIORITY_1
+	0x00001000, // PA_SC_LINE_CNTL
+	0x00000000, // PA_SC_AA_CONFIG
+	0x00000005, // PA_SU_VTX_CNTL
+	0x3f800000, // PA_CL_GB_VERT_CLIP_ADJ
+	0x3f800000, // PA_CL_GB_VERT_DISC_ADJ
+	0x3f800000, // PA_CL_GB_HORZ_CLIP_ADJ
+	0x3f800000, // PA_CL_GB_HORZ_DISC_ADJ
+	0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_0
+	0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_1
+	0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_2
+	0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_3
+	0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_0
+	0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_1
+	0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_2
+	0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_3
+	0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_0
+	0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_1
+	0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_2
+	0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_3
+	0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_0
+	0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_1
+	0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_2
+	0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_3
+	0xffffffff, // PA_SC_AA_MASK_X0Y0_X1Y0
+	0xffffffff, // PA_SC_AA_MASK_X0Y1_X1Y1
+	0x00000000, // PA_SC_SHADER_CONTROL
+	0x00000003, // PA_SC_BINNER_CNTL_0
+	0x00000000, // PA_SC_BINNER_CNTL_1
+	0x00100000, // PA_SC_CONSERVATIVE_RASTERIZATION_CNTL
+	0x00000000, // PA_SC_NGG_MODE_CNTL
+	0, // HOLE
+	0x0000001e, // VGT_VERTEX_REUSE_BLOCK_CNTL
+	0x00000020, // VGT_OUT_DEALLOC_CNTL
+	0x00000000, // CB_COLOR0_BASE
+	0x00000000, // CB_COLOR0_PITCH
+	0x00000000, // CB_COLOR0_SLICE
+	0x00000000, // CB_COLOR0_VIEW
+	0x00000000, // CB_COLOR0_INFO
+	0x00000000, // CB_COLOR0_ATTRIB
+	0x00000000, // CB_COLOR0_DCC_CONTROL
+	0x00000000, // CB_COLOR0_CMASK
+	0x00000000, // CB_COLOR0_CMASK_SLICE
+	0x00000000, // CB_COLOR0_FMASK
+	0x00000000, // CB_COLOR0_FMASK_SLICE
+	0x00000000, // CB_COLOR0_CLEAR_WORD0
+	0x00000000, // CB_COLOR0_CLEAR_WORD1
+	0x00000000, // CB_COLOR0_DCC_BASE
+	0, // HOLE
+	0x00000000, // CB_COLOR1_BASE
+	0x00000000, // CB_COLOR1_PITCH
+	0x00000000, // CB_COLOR1_SLICE
+	0x00000000, // CB_COLOR1_VIEW
+	0x00000000, // CB_COLOR1_INFO
+	0x00000000, // CB_COLOR1_ATTRIB
+	0x00000000, // CB_COLOR1_DCC_CONTROL
+	0x00000000, // CB_COLOR1_CMASK
+	0x00000000, // CB_COLOR1_CMASK_SLICE
+	0x00000000, // CB_COLOR1_FMASK
+	0x00000000, // CB_COLOR1_FMASK_SLICE
+	0x00000000, // CB_COLOR1_CLEAR_WORD0
+	0x00000000, // CB_COLOR1_CLEAR_WORD1
+	0x00000000, // CB_COLOR1_DCC_BASE
+	0, // HOLE
+	0x00000000, // CB_COLOR2_BASE
+	0x00000000, // CB_COLOR2_PITCH
+	0x00000000, // CB_COLOR2_SLICE
+	0x00000000, // CB_COLOR2_VIEW
+	0x00000000, // CB_COLOR2_INFO
+	0x00000000, // CB_COLOR2_ATTRIB
+	0x00000000, // CB_COLOR2_DCC_CONTROL
+	0x00000000, // CB_COLOR2_CMASK
+	0x00000000, // CB_COLOR2_CMASK_SLICE
+	0x00000000, // CB_COLOR2_FMASK
+	0x00000000, // CB_COLOR2_FMASK_SLICE
+	0x00000000, // CB_COLOR2_CLEAR_WORD0
+	0x00000000, // CB_COLOR2_CLEAR_WORD1
+	0x00000000, // CB_COLOR2_DCC_BASE
+	0, // HOLE
+	0x00000000, // CB_COLOR3_BASE
+	0x00000000, // CB_COLOR3_PITCH
+	0x00000000, // CB_COLOR3_SLICE
+	0x00000000, // CB_COLOR3_VIEW
+	0x00000000, // CB_COLOR3_INFO
+	0x00000000, // CB_COLOR3_ATTRIB
+	0x00000000, // CB_COLOR3_DCC_CONTROL
+	0x00000000, // CB_COLOR3_CMASK
+	0x00000000, // CB_COLOR3_CMASK_SLICE
+	0x00000000, // CB_COLOR3_FMASK
+	0x00000000, // CB_COLOR3_FMASK_SLICE
+	0x00000000, // CB_COLOR3_CLEAR_WORD0
+	0x00000000, // CB_COLOR3_CLEAR_WORD1
+	0x00000000, // CB_COLOR3_DCC_BASE
+	0, // HOLE
+	0x00000000, // CB_COLOR4_BASE
+	0x00000000, // CB_COLOR4_PITCH
+	0x00000000, // CB_COLOR4_SLICE
+	0x00000000, // CB_COLOR4_VIEW
+	0x00000000, // CB_COLOR4_INFO
+	0x00000000, // CB_COLOR4_ATTRIB
+	0x00000000, // CB_COLOR4_DCC_CONTROL
+	0x00000000, // CB_COLOR4_CMASK
+	0x00000000, // CB_COLOR4_CMASK_SLICE
+	0x00000000, // CB_COLOR4_FMASK
+	0x00000000, // CB_COLOR4_FMASK_SLICE
+	0x00000000, // CB_COLOR4_CLEAR_WORD0
+	0x00000000, // CB_COLOR4_CLEAR_WORD1
+	0x00000000, // CB_COLOR4_DCC_BASE
+	0, // HOLE
+	0x00000000, // CB_COLOR5_BASE
+	0x00000000, // CB_COLOR5_PITCH
+	0x00000000, // CB_COLOR5_SLICE
+	0x00000000, // CB_COLOR5_VIEW
+	0x00000000, // CB_COLOR5_INFO
+	0x00000000, // CB_COLOR5_ATTRIB
+	0x00000000, // CB_COLOR5_DCC_CONTROL
+	0x00000000, // CB_COLOR5_CMASK
+	0x00000000, // CB_COLOR5_CMASK_SLICE
+	0x00000000, // CB_COLOR5_FMASK
+	0x00000000, // CB_COLOR5_FMASK_SLICE
+	0x00000000, // CB_COLOR5_CLEAR_WORD0
+	0x00000000, // CB_COLOR5_CLEAR_WORD1
+	0x00000000, // CB_COLOR5_DCC_BASE
+	0, // HOLE
+	0x00000000, // CB_COLOR6_BASE
+	0x00000000, // CB_COLOR6_PITCH
+	0x00000000, // CB_COLOR6_SLICE
+	0x00000000, // CB_COLOR6_VIEW
+	0x00000000, // CB_COLOR6_INFO
+	0x00000000, // CB_COLOR6_ATTRIB
+	0x00000000, // CB_COLOR6_DCC_CONTROL
+	0x00000000, // CB_COLOR6_CMASK
+	0x00000000, // CB_COLOR6_CMASK_SLICE
+	0x00000000, // CB_COLOR6_FMASK
+	0x00000000, // CB_COLOR6_FMASK_SLICE
+	0x00000000, // CB_COLOR6_CLEAR_WORD0
+	0x00000000, // CB_COLOR6_CLEAR_WORD1
+	0x00000000, // CB_COLOR6_DCC_BASE
+	0, // HOLE
+	0x00000000, // CB_COLOR7_BASE
+	0x00000000, // CB_COLOR7_PITCH
+	0x00000000, // CB_COLOR7_SLICE
+	0x00000000, // CB_COLOR7_VIEW
+	0x00000000, // CB_COLOR7_INFO
+	0x00000000, // CB_COLOR7_ATTRIB
+	0x00000000, // CB_COLOR7_DCC_CONTROL
+	0x00000000, // CB_COLOR7_CMASK
+	0x00000000, // CB_COLOR7_CMASK_SLICE
+	0x00000000, // CB_COLOR7_FMASK
+	0x00000000, // CB_COLOR7_FMASK_SLICE
+	0x00000000, // CB_COLOR7_CLEAR_WORD0
+	0x00000000, // CB_COLOR7_CLEAR_WORD1
+	0x00000000, // CB_COLOR7_DCC_BASE
+	0, // HOLE
+	0x00000000, // CB_COLOR0_BASE_EXT
+	0x00000000, // CB_COLOR1_BASE_EXT
+	0x00000000, // CB_COLOR2_BASE_EXT
+	0x00000000, // CB_COLOR3_BASE_EXT
+	0x00000000, // CB_COLOR4_BASE_EXT
+	0x00000000, // CB_COLOR5_BASE_EXT
+	0x00000000, // CB_COLOR6_BASE_EXT
+	0x00000000, // CB_COLOR7_BASE_EXT
+	0x00000000, // CB_COLOR0_CMASK_BASE_EXT
+	0x00000000, // CB_COLOR1_CMASK_BASE_EXT
+	0x00000000, // CB_COLOR2_CMASK_BASE_EXT
+	0x00000000, // CB_COLOR3_CMASK_BASE_EXT
+	0x00000000, // CB_COLOR4_CMASK_BASE_EXT
+	0x00000000, // CB_COLOR5_CMASK_BASE_EXT
+	0x00000000, // CB_COLOR6_CMASK_BASE_EXT
+	0x00000000, // CB_COLOR7_CMASK_BASE_EXT
+	0x00000000, // CB_COLOR0_FMASK_BASE_EXT
+	0x00000000, // CB_COLOR1_FMASK_BASE_EXT
+	0x00000000, // CB_COLOR2_FMASK_BASE_EXT
+	0x00000000, // CB_COLOR3_FMASK_BASE_EXT
+	0x00000000, // CB_COLOR4_FMASK_BASE_EXT
+	0x00000000, // CB_COLOR5_FMASK_BASE_EXT
+	0x00000000, // CB_COLOR6_FMASK_BASE_EXT
+	0x00000000, // CB_COLOR7_FMASK_BASE_EXT
+	0x00000000, // CB_COLOR0_DCC_BASE_EXT
+	0x00000000, // CB_COLOR1_DCC_BASE_EXT
+	0x00000000, // CB_COLOR2_DCC_BASE_EXT
+	0x00000000, // CB_COLOR3_DCC_BASE_EXT
+	0x00000000, // CB_COLOR4_DCC_BASE_EXT
+	0x00000000, // CB_COLOR5_DCC_BASE_EXT
+	0x00000000, // CB_COLOR6_DCC_BASE_EXT
+	0x00000000, // CB_COLOR7_DCC_BASE_EXT
+	0x00000000, // CB_COLOR0_ATTRIB2
+	0x00000000, // CB_COLOR1_ATTRIB2
+	0x00000000, // CB_COLOR2_ATTRIB2
+	0x00000000, // CB_COLOR3_ATTRIB2
+	0x00000000, // CB_COLOR4_ATTRIB2
+	0x00000000, // CB_COLOR5_ATTRIB2
+	0x00000000, // CB_COLOR6_ATTRIB2
+	0x00000000, // CB_COLOR7_ATTRIB2
+	0x00000000, // CB_COLOR0_ATTRIB3
+	0x00000000, // CB_COLOR1_ATTRIB3
+	0x00000000, // CB_COLOR2_ATTRIB3
+	0x00000000, // CB_COLOR3_ATTRIB3
+	0x00000000, // CB_COLOR4_ATTRIB3
+	0x00000000, // CB_COLOR5_ATTRIB3
+	0x00000000, // CB_COLOR6_ATTRIB3
+	0x00000000, // CB_COLOR7_ATTRIB3
+};
+static const struct cs_extent_def gfx10_SECT_CONTEXT_defs[] = {
+	{gfx10_SECT_CONTEXT_def_1, 0x0000a000, 215 },
+	{gfx10_SECT_CONTEXT_def_2, 0x0000a0d8, 272 },
+	{gfx10_SECT_CONTEXT_def_3, 0x0000a1f5, 4 },
+	{gfx10_SECT_CONTEXT_def_4, 0x0000a1ff, 158 },
+	{gfx10_SECT_CONTEXT_def_5, 0x0000a2a0, 2 },
+	{gfx10_SECT_CONTEXT_def_6, 0x0000a2a3, 1 },
+	{gfx10_SECT_CONTEXT_def_7, 0x0000a2a5, 66 },
+	{gfx10_SECT_CONTEXT_def_8, 0x0000a2f5, 203 },
+	{ 0, 0, 0 }
+};
+static const struct cs_section_def gfx10_cs_data[] = {
+	{ gfx10_SECT_CONTEXT_defs, SECT_CONTEXT },
+	{ 0, SECT_NONE }
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/clearstate_gfx11.h b/drivers/gpu/drm/amd/amdgpu/clearstate_gfx11.h
new file mode 100644
index 000000000000..a8b29d33c464
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/clearstate_gfx11.h
@@ -0,0 +1,997 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __CLEARSTATE_GFX11_H_
+#define __CLEARSTATE_GFX11_H_
+
+static const unsigned int gfx11_SECT_CONTEXT_def_1[] =
+{
+    0x00000000, // DB_RENDER_CONTROL
+    0x00000000, // DB_COUNT_CONTROL
+    0x00000000, // DB_DEPTH_VIEW
+    0x00000000, // DB_RENDER_OVERRIDE
+    0x00000000, // DB_RENDER_OVERRIDE2
+    0x00000000, // DB_HTILE_DATA_BASE
+    0, // HOLE
+    0x00000000, // DB_DEPTH_SIZE_XY
+    0x00000000, // DB_DEPTH_BOUNDS_MIN
+    0x00000000, // DB_DEPTH_BOUNDS_MAX
+    0x00000000, // DB_STENCIL_CLEAR
+    0x00000000, // DB_DEPTH_CLEAR
+    0x00000000, // PA_SC_SCREEN_SCISSOR_TL
+    0x40004000, // PA_SC_SCREEN_SCISSOR_BR
+    0, // HOLE
+    0x00000000, // DB_RESERVED_REG_2
+    0x00000000, // DB_Z_INFO
+    0x00000000, // DB_STENCIL_INFO
+    0x00000000, // DB_Z_READ_BASE
+    0x00000000, // DB_STENCIL_READ_BASE
+    0x00000000, // DB_Z_WRITE_BASE
+    0x00000000, // DB_STENCIL_WRITE_BASE
+    0x00000000, // DB_RESERVED_REG_1
+    0x00000000, // DB_RESERVED_REG_3
+    0x00000000, // DB_SPI_VRS_CENTER_LOCATION
+    0, // HOLE
+    0x00000000, // DB_Z_READ_BASE_HI
+    0x00000000, // DB_STENCIL_READ_BASE_HI
+    0x00000000, // DB_Z_WRITE_BASE_HI
+    0x00000000, // DB_STENCIL_WRITE_BASE_HI
+    0x00000000, // DB_HTILE_DATA_BASE_HI
+    0x00150055, // DB_RMI_L2_CACHE_CONTROL
+    0x00000000, // TA_BC_BASE_ADDR
+    0x00000000, // TA_BC_BASE_ADDR_HI
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // COHER_DEST_BASE_HI_0
+    0x00000000, // COHER_DEST_BASE_HI_1
+    0x00000000, // COHER_DEST_BASE_HI_2
+    0x00000000, // COHER_DEST_BASE_HI_3
+    0x00000000, // COHER_DEST_BASE_2
+    0x00000000, // COHER_DEST_BASE_3
+    0x00000000, // PA_SC_WINDOW_OFFSET
+    0x80000000, // PA_SC_WINDOW_SCISSOR_TL
+    0x40004000, // PA_SC_WINDOW_SCISSOR_BR
+    0x0000ffff, // PA_SC_CLIPRECT_RULE
+    0x00000000, // PA_SC_CLIPRECT_0_TL
+    0x40004000, // PA_SC_CLIPRECT_0_BR
+    0x00000000, // PA_SC_CLIPRECT_1_TL
+    0x40004000, // PA_SC_CLIPRECT_1_BR
+    0x00000000, // PA_SC_CLIPRECT_2_TL
+    0x40004000, // PA_SC_CLIPRECT_2_BR
+    0x00000000, // PA_SC_CLIPRECT_3_TL
+    0x40004000, // PA_SC_CLIPRECT_3_BR
+    0xaa99aaaa, // PA_SC_EDGERULE
+    0x00000000, // PA_SU_HARDWARE_SCREEN_OFFSET
+    0xffffffff, // CB_TARGET_MASK
+    0xffffffff, // CB_SHADER_MASK
+    0x80000000, // PA_SC_GENERIC_SCISSOR_TL
+    0x40004000, // PA_SC_GENERIC_SCISSOR_BR
+    0x00000000, // COHER_DEST_BASE_0
+    0x00000000, // COHER_DEST_BASE_1
+    0x80000000, // PA_SC_VPORT_SCISSOR_0_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_0_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_1_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_1_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_2_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_2_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_3_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_3_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_4_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_4_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_5_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_5_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_6_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_6_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_7_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_7_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_8_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_8_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_9_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_9_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_10_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_10_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_11_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_11_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_12_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_12_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_13_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_13_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_14_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_14_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_15_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_15_BR
+    0x00000000, // PA_SC_VPORT_ZMIN_0
+    0x3f800000, // PA_SC_VPORT_ZMAX_0
+    0x00000000, // PA_SC_VPORT_ZMIN_1
+    0x3f800000, // PA_SC_VPORT_ZMAX_1
+    0x00000000, // PA_SC_VPORT_ZMIN_2
+    0x3f800000, // PA_SC_VPORT_ZMAX_2
+    0x00000000, // PA_SC_VPORT_ZMIN_3
+    0x3f800000, // PA_SC_VPORT_ZMAX_3
+    0x00000000, // PA_SC_VPORT_ZMIN_4
+    0x3f800000, // PA_SC_VPORT_ZMAX_4
+    0x00000000, // PA_SC_VPORT_ZMIN_5
+    0x3f800000, // PA_SC_VPORT_ZMAX_5
+    0x00000000, // PA_SC_VPORT_ZMIN_6
+    0x3f800000, // PA_SC_VPORT_ZMAX_6
+    0x00000000, // PA_SC_VPORT_ZMIN_7
+    0x3f800000, // PA_SC_VPORT_ZMAX_7
+    0x00000000, // PA_SC_VPORT_ZMIN_8
+    0x3f800000, // PA_SC_VPORT_ZMAX_8
+    0x00000000, // PA_SC_VPORT_ZMIN_9
+    0x3f800000, // PA_SC_VPORT_ZMAX_9
+    0x00000000, // PA_SC_VPORT_ZMIN_10
+    0x3f800000, // PA_SC_VPORT_ZMAX_10
+    0x00000000, // PA_SC_VPORT_ZMIN_11
+    0x3f800000, // PA_SC_VPORT_ZMAX_11
+    0x00000000, // PA_SC_VPORT_ZMIN_12
+    0x3f800000, // PA_SC_VPORT_ZMAX_12
+    0x00000000, // PA_SC_VPORT_ZMIN_13
+    0x3f800000, // PA_SC_VPORT_ZMAX_13
+    0x00000000, // PA_SC_VPORT_ZMIN_14
+    0x3f800000, // PA_SC_VPORT_ZMAX_14
+    0x00000000, // PA_SC_VPORT_ZMIN_15
+    0x3f800000, // PA_SC_VPORT_ZMAX_15
+    0x00000000, // PA_SC_RASTER_CONFIG
+    0x00000000, // PA_SC_RASTER_CONFIG_1
+    0x00000000, // PA_SC_SCREEN_EXTENT_CONTROL
+};
+static const unsigned int gfx11_SECT_CONTEXT_def_2[] =
+{
+    0x00000000, // CP_PERFMON_CNTX_CNTL
+    0x00000000, // CP_PIPEID
+    0x00000000, // CP_VMID
+    0x00000000, // CONTEXT_RESERVED_REG0
+    0x00000000, // CONTEXT_RESERVED_REG1
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // PA_SC_FSR_EN
+    0x00000000, // PA_SC_FSR_FBW_RECURSIONS_X
+    0x00000000, // PA_SC_FSR_FBW_RECURSIONS_Y
+    0, // HOLE
+    0x00000000, // PA_SC_VRS_OVERRIDE_CNTL
+    0x00000000, // PA_SC_VRS_RATE_FEEDBACK_BASE
+    0x00000000, // PA_SC_VRS_RATE_FEEDBACK_BASE_EXT
+    0x00000000, // PA_SC_VRS_RATE_FEEDBACK_SIZE_XY
+    0x00000000, // PA_SC_BINNER_OUTPUT_TIMEOUT_CNTL
+    0x00000000, // PA_SC_VRS_RATE_CACHE_CNTL
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // PA_SC_VRS_RATE_BASE
+    0x00000000, // PA_SC_VRS_RATE_BASE_EXT
+    0x00000000, // PA_SC_VRS_RATE_SIZE_XY
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_MULTI_PRIM_IB_RESET_INDX
+    0x00550055, // CB_RMI_GL2_CACHE_CONTROL
+    0x00000000, // CB_BLEND_RED
+    0x00000000, // CB_BLEND_GREEN
+    0x00000000, // CB_BLEND_BLUE
+    0x00000000, // CB_BLEND_ALPHA
+    0x00000000, // CB_FDCC_CONTROL
+    0x00000000, // CB_COVERAGE_OUT_CONTROL
+    0x00000000, // DB_STENCIL_CONTROL
+    0x01000000, // DB_STENCILREFMASK
+    0x01000000, // DB_STENCILREFMASK_BF
+    0, // HOLE
+    0x00000000, // PA_CL_VPORT_XSCALE
+    0x00000000, // PA_CL_VPORT_XOFFSET
+    0x00000000, // PA_CL_VPORT_YSCALE
+    0x00000000, // PA_CL_VPORT_YOFFSET
+    0x00000000, // PA_CL_VPORT_ZSCALE
+    0x00000000, // PA_CL_VPORT_ZOFFSET
+    0x00000000, // PA_CL_VPORT_XSCALE_1
+    0x00000000, // PA_CL_VPORT_XOFFSET_1
+    0x00000000, // PA_CL_VPORT_YSCALE_1
+    0x00000000, // PA_CL_VPORT_YOFFSET_1
+    0x00000000, // PA_CL_VPORT_ZSCALE_1
+    0x00000000, // PA_CL_VPORT_ZOFFSET_1
+    0x00000000, // PA_CL_VPORT_XSCALE_2
+    0x00000000, // PA_CL_VPORT_XOFFSET_2
+    0x00000000, // PA_CL_VPORT_YSCALE_2
+    0x00000000, // PA_CL_VPORT_YOFFSET_2
+    0x00000000, // PA_CL_VPORT_ZSCALE_2
+    0x00000000, // PA_CL_VPORT_ZOFFSET_2
+    0x00000000, // PA_CL_VPORT_XSCALE_3
+    0x00000000, // PA_CL_VPORT_XOFFSET_3
+    0x00000000, // PA_CL_VPORT_YSCALE_3
+    0x00000000, // PA_CL_VPORT_YOFFSET_3
+    0x00000000, // PA_CL_VPORT_ZSCALE_3
+    0x00000000, // PA_CL_VPORT_ZOFFSET_3
+    0x00000000, // PA_CL_VPORT_XSCALE_4
+    0x00000000, // PA_CL_VPORT_XOFFSET_4
+    0x00000000, // PA_CL_VPORT_YSCALE_4
+    0x00000000, // PA_CL_VPORT_YOFFSET_4
+    0x00000000, // PA_CL_VPORT_ZSCALE_4
+    0x00000000, // PA_CL_VPORT_ZOFFSET_4
+    0x00000000, // PA_CL_VPORT_XSCALE_5
+    0x00000000, // PA_CL_VPORT_XOFFSET_5
+    0x00000000, // PA_CL_VPORT_YSCALE_5
+    0x00000000, // PA_CL_VPORT_YOFFSET_5
+    0x00000000, // PA_CL_VPORT_ZSCALE_5
+    0x00000000, // PA_CL_VPORT_ZOFFSET_5
+    0x00000000, // PA_CL_VPORT_XSCALE_6
+    0x00000000, // PA_CL_VPORT_XOFFSET_6
+    0x00000000, // PA_CL_VPORT_YSCALE_6
+    0x00000000, // PA_CL_VPORT_YOFFSET_6
+    0x00000000, // PA_CL_VPORT_ZSCALE_6
+    0x00000000, // PA_CL_VPORT_ZOFFSET_6
+    0x00000000, // PA_CL_VPORT_XSCALE_7
+    0x00000000, // PA_CL_VPORT_XOFFSET_7
+    0x00000000, // PA_CL_VPORT_YSCALE_7
+    0x00000000, // PA_CL_VPORT_YOFFSET_7
+    0x00000000, // PA_CL_VPORT_ZSCALE_7
+    0x00000000, // PA_CL_VPORT_ZOFFSET_7
+    0x00000000, // PA_CL_VPORT_XSCALE_8
+    0x00000000, // PA_CL_VPORT_XOFFSET_8
+    0x00000000, // PA_CL_VPORT_YSCALE_8
+    0x00000000, // PA_CL_VPORT_YOFFSET_8
+    0x00000000, // PA_CL_VPORT_ZSCALE_8
+    0x00000000, // PA_CL_VPORT_ZOFFSET_8
+    0x00000000, // PA_CL_VPORT_XSCALE_9
+    0x00000000, // PA_CL_VPORT_XOFFSET_9
+    0x00000000, // PA_CL_VPORT_YSCALE_9
+    0x00000000, // PA_CL_VPORT_YOFFSET_9
+    0x00000000, // PA_CL_VPORT_ZSCALE_9
+    0x00000000, // PA_CL_VPORT_ZOFFSET_9
+    0x00000000, // PA_CL_VPORT_XSCALE_10
+    0x00000000, // PA_CL_VPORT_XOFFSET_10
+    0x00000000, // PA_CL_VPORT_YSCALE_10
+    0x00000000, // PA_CL_VPORT_YOFFSET_10
+    0x00000000, // PA_CL_VPORT_ZSCALE_10
+    0x00000000, // PA_CL_VPORT_ZOFFSET_10
+    0x00000000, // PA_CL_VPORT_XSCALE_11
+    0x00000000, // PA_CL_VPORT_XOFFSET_11
+    0x00000000, // PA_CL_VPORT_YSCALE_11
+    0x00000000, // PA_CL_VPORT_YOFFSET_11
+    0x00000000, // PA_CL_VPORT_ZSCALE_11
+    0x00000000, // PA_CL_VPORT_ZOFFSET_11
+    0x00000000, // PA_CL_VPORT_XSCALE_12
+    0x00000000, // PA_CL_VPORT_XOFFSET_12
+    0x00000000, // PA_CL_VPORT_YSCALE_12
+    0x00000000, // PA_CL_VPORT_YOFFSET_12
+    0x00000000, // PA_CL_VPORT_ZSCALE_12
+    0x00000000, // PA_CL_VPORT_ZOFFSET_12
+    0x00000000, // PA_CL_VPORT_XSCALE_13
+    0x00000000, // PA_CL_VPORT_XOFFSET_13
+    0x00000000, // PA_CL_VPORT_YSCALE_13
+    0x00000000, // PA_CL_VPORT_YOFFSET_13
+    0x00000000, // PA_CL_VPORT_ZSCALE_13
+    0x00000000, // PA_CL_VPORT_ZOFFSET_13
+    0x00000000, // PA_CL_VPORT_XSCALE_14
+    0x00000000, // PA_CL_VPORT_XOFFSET_14
+    0x00000000, // PA_CL_VPORT_YSCALE_14
+    0x00000000, // PA_CL_VPORT_YOFFSET_14
+    0x00000000, // PA_CL_VPORT_ZSCALE_14
+    0x00000000, // PA_CL_VPORT_ZOFFSET_14
+    0x00000000, // PA_CL_VPORT_XSCALE_15
+    0x00000000, // PA_CL_VPORT_XOFFSET_15
+    0x00000000, // PA_CL_VPORT_YSCALE_15
+    0x00000000, // PA_CL_VPORT_YOFFSET_15
+    0x00000000, // PA_CL_VPORT_ZSCALE_15
+    0x00000000, // PA_CL_VPORT_ZOFFSET_15
+    0x00000000, // PA_CL_UCP_0_X
+    0x00000000, // PA_CL_UCP_0_Y
+    0x00000000, // PA_CL_UCP_0_Z
+    0x00000000, // PA_CL_UCP_0_W
+    0x00000000, // PA_CL_UCP_1_X
+    0x00000000, // PA_CL_UCP_1_Y
+    0x00000000, // PA_CL_UCP_1_Z
+    0x00000000, // PA_CL_UCP_1_W
+    0x00000000, // PA_CL_UCP_2_X
+    0x00000000, // PA_CL_UCP_2_Y
+    0x00000000, // PA_CL_UCP_2_Z
+    0x00000000, // PA_CL_UCP_2_W
+    0x00000000, // PA_CL_UCP_3_X
+    0x00000000, // PA_CL_UCP_3_Y
+    0x00000000, // PA_CL_UCP_3_Z
+    0x00000000, // PA_CL_UCP_3_W
+    0x00000000, // PA_CL_UCP_4_X
+    0x00000000, // PA_CL_UCP_4_Y
+    0x00000000, // PA_CL_UCP_4_Z
+    0x00000000, // PA_CL_UCP_4_W
+    0x00000000, // PA_CL_UCP_5_X
+    0x00000000, // PA_CL_UCP_5_Y
+    0x00000000, // PA_CL_UCP_5_Z
+    0x00000000, // PA_CL_UCP_5_W
+    0x00000000, // PA_CL_PROG_NEAR_CLIP_Z
+    0x00000000, // PA_RATE_CNTL
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // SPI_PS_INPUT_CNTL_0
+    0x00000000, // SPI_PS_INPUT_CNTL_1
+    0x00000000, // SPI_PS_INPUT_CNTL_2
+    0x00000000, // SPI_PS_INPUT_CNTL_3
+    0x00000000, // SPI_PS_INPUT_CNTL_4
+    0x00000000, // SPI_PS_INPUT_CNTL_5
+    0x00000000, // SPI_PS_INPUT_CNTL_6
+    0x00000000, // SPI_PS_INPUT_CNTL_7
+    0x00000000, // SPI_PS_INPUT_CNTL_8
+    0x00000000, // SPI_PS_INPUT_CNTL_9
+    0x00000000, // SPI_PS_INPUT_CNTL_10
+    0x00000000, // SPI_PS_INPUT_CNTL_11
+    0x00000000, // SPI_PS_INPUT_CNTL_12
+    0x00000000, // SPI_PS_INPUT_CNTL_13
+    0x00000000, // SPI_PS_INPUT_CNTL_14
+    0x00000000, // SPI_PS_INPUT_CNTL_15
+    0x00000000, // SPI_PS_INPUT_CNTL_16
+    0x00000000, // SPI_PS_INPUT_CNTL_17
+    0x00000000, // SPI_PS_INPUT_CNTL_18
+    0x00000000, // SPI_PS_INPUT_CNTL_19
+    0x00000000, // SPI_PS_INPUT_CNTL_20
+    0x00000000, // SPI_PS_INPUT_CNTL_21
+    0x00000000, // SPI_PS_INPUT_CNTL_22
+    0x00000000, // SPI_PS_INPUT_CNTL_23
+    0x00000000, // SPI_PS_INPUT_CNTL_24
+    0x00000000, // SPI_PS_INPUT_CNTL_25
+    0x00000000, // SPI_PS_INPUT_CNTL_26
+    0x00000000, // SPI_PS_INPUT_CNTL_27
+    0x00000000, // SPI_PS_INPUT_CNTL_28
+    0x00000000, // SPI_PS_INPUT_CNTL_29
+    0x00000000, // SPI_PS_INPUT_CNTL_30
+    0x00000000, // SPI_PS_INPUT_CNTL_31
+    0x00000000, // SPI_VS_OUT_CONFIG
+    0, // HOLE
+    0x00000000, // SPI_PS_INPUT_ENA
+    0x00000000, // SPI_PS_INPUT_ADDR
+    0x00000000, // SPI_INTERP_CONTROL_0
+    0x00000002, // SPI_PS_IN_CONTROL
+    0x00000000, // SPI_BARYC_SSAA_CNTL
+    0x00000000, // SPI_BARYC_CNTL
+    0, // HOLE
+    0x00000000, // SPI_TMPRING_SIZE
+    0x00000000, // SPI_GFX_SCRATCH_BASE_LO
+    0x00000000, // SPI_GFX_SCRATCH_BASE_HI
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // SPI_SHADER_IDX_FORMAT
+    0x00000000, // SPI_SHADER_POS_FORMAT
+    0x00000000, // SPI_SHADER_Z_FORMAT
+    0x00000000, // SPI_SHADER_COL_FORMAT
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // SX_PS_DOWNCONVERT_CONTROL
+    0x00000000, // SX_PS_DOWNCONVERT
+    0x00000000, // SX_BLEND_OPT_EPSILON
+    0x00000000, // SX_BLEND_OPT_CONTROL
+    0x00000000, // SX_MRT0_BLEND_OPT
+    0x00000000, // SX_MRT1_BLEND_OPT
+    0x00000000, // SX_MRT2_BLEND_OPT
+    0x00000000, // SX_MRT3_BLEND_OPT
+    0x00000000, // SX_MRT4_BLEND_OPT
+    0x00000000, // SX_MRT5_BLEND_OPT
+    0x00000000, // SX_MRT6_BLEND_OPT
+    0x00000000, // SX_MRT7_BLEND_OPT
+    0x00000000, // CB_BLEND0_CONTROL
+    0x00000000, // CB_BLEND1_CONTROL
+    0x00000000, // CB_BLEND2_CONTROL
+    0x00000000, // CB_BLEND3_CONTROL
+    0x00000000, // CB_BLEND4_CONTROL
+    0x00000000, // CB_BLEND5_CONTROL
+    0x00000000, // CB_BLEND6_CONTROL
+    0x00000000, // CB_BLEND7_CONTROL
+};
+static const unsigned int gfx11_SECT_CONTEXT_def_3[] =
+{
+    0x00000000, // PA_CL_POINT_X_RAD
+    0x00000000, // PA_CL_POINT_Y_RAD
+    0x00000000, // PA_CL_POINT_SIZE
+    0x00000000, // PA_CL_POINT_CULL_RAD
+};
+static const unsigned int gfx11_SECT_CONTEXT_def_4[] =
+{
+    0x00000000, // GE_MAX_OUTPUT_PER_SUBGROUP
+    0x00000000, // DB_DEPTH_CONTROL
+    0x00000000, // DB_EQAA
+    0x00000000, // CB_COLOR_CONTROL
+    0x00000000, // DB_SHADER_CONTROL
+    0x00090000, // PA_CL_CLIP_CNTL
+    0x00000004, // PA_SU_SC_MODE_CNTL
+    0x00000000, // PA_CL_VTE_CNTL
+    0x00000000, // PA_CL_VS_OUT_CNTL
+    0x00000000, // PA_CL_NANINF_CNTL
+    0x00000000, // PA_SU_LINE_STIPPLE_CNTL
+    0x00000000, // PA_SU_LINE_STIPPLE_SCALE
+    0x00000000, // PA_SU_PRIM_FILTER_CNTL
+    0x00000000, // PA_SU_SMALL_PRIM_FILTER_CNTL
+    0, // HOLE
+    0x00000000, // PA_CL_NGG_CNTL
+    0x00000000, // PA_SU_OVER_RASTERIZATION_CNTL
+    0x00000000, // PA_STEREO_CNTL
+    0x00000000, // PA_STATE_STEREO_X
+    0x00000000, // PA_CL_VRS_CNTL
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // PA_SU_POINT_SIZE
+    0x00000000, // PA_SU_POINT_MINMAX
+    0x00000000, // PA_SU_LINE_CNTL
+    0x00000000, // PA_SC_LINE_STIPPLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_HOS_MAX_TESS_LEVEL
+    0x00000000, // VGT_HOS_MIN_TESS_LEVEL
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_GS_ONCHIP_CNTL
+    0x00000000, // PA_SC_MODE_CNTL_0
+    0x00000000, // PA_SC_MODE_CNTL_1
+    0x00000000, // VGT_ENHANCE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // IA_ENHANCE
+};
+static const unsigned int gfx11_SECT_CONTEXT_def_5[] =
+{
+    0x00000000, // WD_ENHANCE
+    0x00000000, // VGT_PRIMITIVEID_EN
+};
+static const unsigned int gfx11_SECT_CONTEXT_def_6[] =
+{
+    0x00000000, // VGT_PRIMITIVEID_RESET
+};
+static const unsigned int gfx11_SECT_CONTEXT_def_7[] =
+{
+    0x00000000, // VGT_DRAW_PAYLOAD_CNTL
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_ESGS_RING_ITEMSIZE
+    0, // HOLE
+    0x00000000, // VGT_REUSE_OFF
+    0, // HOLE
+    0x00000000, // DB_HTILE_SURFACE
+    0x00000000, // DB_SRESULTS_COMPARE_STATE0
+    0x00000000, // DB_SRESULTS_COMPARE_STATE1
+    0x00000000, // DB_PRELOAD_CONTROL
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_OFFSET
+    0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_BUFFER_FILLED_SIZE
+    0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_VERTEX_STRIDE
+    0, // HOLE
+    0x00000000, // VGT_GS_MAX_VERT_OUT
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // GE_NGG_SUBGRP_CNTL
+    0x00000000, // VGT_TESS_DISTRIBUTION
+    0x00000000, // VGT_SHADER_STAGES_EN
+    0x00000000, // VGT_LS_HS_CONFIG
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_TF_PARAM
+    0x00000000, // DB_ALPHA_TO_MASK
+    0, // HOLE
+    0x00000000, // PA_SU_POLY_OFFSET_DB_FMT_CNTL
+    0x00000000, // PA_SU_POLY_OFFSET_CLAMP
+    0x00000000, // PA_SU_POLY_OFFSET_FRONT_SCALE
+    0x00000000, // PA_SU_POLY_OFFSET_FRONT_OFFSET
+    0x00000000, // PA_SU_POLY_OFFSET_BACK_SCALE
+    0x00000000, // PA_SU_POLY_OFFSET_BACK_OFFSET
+    0x00000000, // VGT_GS_INSTANCE_CNT
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // PA_SC_CENTROID_PRIORITY_0
+    0x00000000, // PA_SC_CENTROID_PRIORITY_1
+    0x00001000, // PA_SC_LINE_CNTL
+    0x00000000, // PA_SC_AA_CONFIG
+    0x00000005, // PA_SU_VTX_CNTL
+    0x3f800000, // PA_CL_GB_VERT_CLIP_ADJ
+    0x3f800000, // PA_CL_GB_VERT_DISC_ADJ
+    0x3f800000, // PA_CL_GB_HORZ_CLIP_ADJ
+    0x3f800000, // PA_CL_GB_HORZ_DISC_ADJ
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_3
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_3
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_3
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_3
+    0xffffffff, // PA_SC_AA_MASK_X0Y0_X1Y0
+    0xffffffff, // PA_SC_AA_MASK_X0Y1_X1Y1
+    0x00000000, // PA_SC_SHADER_CONTROL
+    0x00000003, // PA_SC_BINNER_CNTL_0
+    0x00000000, // PA_SC_BINNER_CNTL_1
+    0x00100000, // PA_SC_CONSERVATIVE_RASTERIZATION_CNTL
+    0x00000000, // PA_SC_NGG_MODE_CNTL
+    0x00000000, // PA_SC_BINNER_CNTL_2
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR0_BASE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR0_VIEW
+    0x00000000, // CB_COLOR0_INFO
+    0x00000000, // CB_COLOR0_ATTRIB
+    0x00000000, // CB_COLOR0_FDCC_CONTROL
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR0_DCC_BASE
+    0, // HOLE
+    0x00000000, // CB_COLOR1_BASE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR1_VIEW
+    0x00000000, // CB_COLOR1_INFO
+    0x00000000, // CB_COLOR1_ATTRIB
+    0x00000000, // CB_COLOR1_FDCC_CONTROL
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR1_DCC_BASE
+    0, // HOLE
+    0x00000000, // CB_COLOR2_BASE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR2_VIEW
+    0x00000000, // CB_COLOR2_INFO
+    0x00000000, // CB_COLOR2_ATTRIB
+    0x00000000, // CB_COLOR2_FDCC_CONTROL
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR2_DCC_BASE
+    0, // HOLE
+    0x00000000, // CB_COLOR3_BASE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR3_VIEW
+    0x00000000, // CB_COLOR3_INFO
+    0x00000000, // CB_COLOR3_ATTRIB
+    0x00000000, // CB_COLOR3_FDCC_CONTROL
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR3_DCC_BASE
+    0, // HOLE
+    0x00000000, // CB_COLOR4_BASE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR4_VIEW
+    0x00000000, // CB_COLOR4_INFO
+    0x00000000, // CB_COLOR4_ATTRIB
+    0x00000000, // CB_COLOR4_FDCC_CONTROL
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR4_DCC_BASE
+    0, // HOLE
+    0x00000000, // CB_COLOR5_BASE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR5_VIEW
+    0x00000000, // CB_COLOR5_INFO
+    0x00000000, // CB_COLOR5_ATTRIB
+    0x00000000, // CB_COLOR5_FDCC_CONTROL
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR5_DCC_BASE
+    0, // HOLE
+    0x00000000, // CB_COLOR6_BASE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR6_VIEW
+    0x00000000, // CB_COLOR6_INFO
+    0x00000000, // CB_COLOR6_ATTRIB
+    0x00000000, // CB_COLOR6_FDCC_CONTROL
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR6_DCC_BASE
+    0, // HOLE
+    0x00000000, // CB_COLOR7_BASE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR7_VIEW
+    0x00000000, // CB_COLOR7_INFO
+    0x00000000, // CB_COLOR7_ATTRIB
+    0x00000000, // CB_COLOR7_FDCC_CONTROL
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR7_DCC_BASE
+    0, // HOLE
+    0x00000000, // CB_COLOR0_BASE_EXT
+    0x00000000, // CB_COLOR1_BASE_EXT
+    0x00000000, // CB_COLOR2_BASE_EXT
+    0x00000000, // CB_COLOR3_BASE_EXT
+    0x00000000, // CB_COLOR4_BASE_EXT
+    0x00000000, // CB_COLOR5_BASE_EXT
+    0x00000000, // CB_COLOR6_BASE_EXT
+    0x00000000, // CB_COLOR7_BASE_EXT
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR0_DCC_BASE_EXT
+    0x00000000, // CB_COLOR1_DCC_BASE_EXT
+    0x00000000, // CB_COLOR2_DCC_BASE_EXT
+    0x00000000, // CB_COLOR3_DCC_BASE_EXT
+    0x00000000, // CB_COLOR4_DCC_BASE_EXT
+    0x00000000, // CB_COLOR5_DCC_BASE_EXT
+    0x00000000, // CB_COLOR6_DCC_BASE_EXT
+    0x00000000, // CB_COLOR7_DCC_BASE_EXT
+    0x00000000, // CB_COLOR0_ATTRIB2
+    0x00000000, // CB_COLOR1_ATTRIB2
+    0x00000000, // CB_COLOR2_ATTRIB2
+    0x00000000, // CB_COLOR3_ATTRIB2
+    0x00000000, // CB_COLOR4_ATTRIB2
+    0x00000000, // CB_COLOR5_ATTRIB2
+    0x00000000, // CB_COLOR6_ATTRIB2
+    0x00000000, // CB_COLOR7_ATTRIB2
+    0x00000000, // CB_COLOR0_ATTRIB3
+    0x00000000, // CB_COLOR1_ATTRIB3
+    0x00000000, // CB_COLOR2_ATTRIB3
+    0x00000000, // CB_COLOR3_ATTRIB3
+    0x00000000, // CB_COLOR4_ATTRIB3
+    0x00000000, // CB_COLOR5_ATTRIB3
+    0x00000000, // CB_COLOR6_ATTRIB3
+    0x00000000, // CB_COLOR7_ATTRIB3
+};
+static const struct cs_extent_def gfx11_SECT_CONTEXT_defs[] =
+{
+    {gfx11_SECT_CONTEXT_def_1, 0x0000a000, 215 },
+    {gfx11_SECT_CONTEXT_def_2, 0x0000a0d8, 272 },
+    {gfx11_SECT_CONTEXT_def_3, 0x0000a1f5, 4 },
+    {gfx11_SECT_CONTEXT_def_4, 0x0000a1ff, 158 },
+    {gfx11_SECT_CONTEXT_def_5, 0x0000a2a0, 2 },
+    {gfx11_SECT_CONTEXT_def_6, 0x0000a2a3, 1 },
+    {gfx11_SECT_CONTEXT_def_7, 0x0000a2a6, 282 },
+    { 0, 0, 0 }
+};
+static const struct cs_section_def gfx11_cs_data[] = {
+    { gfx11_SECT_CONTEXT_defs, SECT_CONTEXT },
+    { 0, SECT_NONE }
+};
+
+#endif /* __CLEARSTATE_GFX11_H_ */
diff --git a/drivers/gpu/drm/amd/amdgpu/clearstate_gfx12.h b/drivers/gpu/drm/amd/amdgpu/clearstate_gfx12.h
new file mode 100644
index 000000000000..2f6c9d11d5ae
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/clearstate_gfx12.h
@@ -0,0 +1,121 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __CLEARSTATE_GFX12_H_
+#define __CLEARSTATE_GFX12_H_
+
+static const unsigned int gfx12_SECT_CONTEXT_def_1[] = {
+0x00000000, //mmSC_MEM_TEMPORAL
+0x00000000, //mmSC_MEM_SPEC_READ
+0x00000000, //mmPA_SC_VPORT_0_TL
+0x00000000, //mmPA_SC_VPORT_0_BR
+0x00000000, //mmPA_SC_VPORT_1_TL
+0x00000000, //mmPA_SC_VPORT_1_BR
+0x00000000, //mmPA_SC_VPORT_2_TL
+0x00000000, //mmPA_SC_VPORT_2_BR
+0x00000000, //mmPA_SC_VPORT_3_TL
+0x00000000, //mmPA_SC_VPORT_3_BR
+0x00000000, //mmPA_SC_VPORT_4_TL
+0x00000000, //mmPA_SC_VPORT_4_BR
+0x00000000, //mmPA_SC_VPORT_5_TL
+0x00000000, //mmPA_SC_VPORT_5_BR
+0x00000000, //mmPA_SC_VPORT_6_TL
+0x00000000, //mmPA_SC_VPORT_6_BR
+0x00000000, //mmPA_SC_VPORT_7_TL
+0x00000000, //mmPA_SC_VPORT_7_BR
+0x00000000, //mmPA_SC_VPORT_8_TL
+0x00000000, //mmPA_SC_VPORT_8_BR
+0x00000000, //mmPA_SC_VPORT_9_TL
+0x00000000, //mmPA_SC_VPORT_9_BR
+0x00000000, //mmPA_SC_VPORT_10_TL
+0x00000000, //mmPA_SC_VPORT_10_BR
+0x00000000, //mmPA_SC_VPORT_11_TL
+0x00000000, //mmPA_SC_VPORT_11_BR
+0x00000000, //mmPA_SC_VPORT_12_TL
+0x00000000, //mmPA_SC_VPORT_12_BR
+0x00000000, //mmPA_SC_VPORT_13_TL
+0x00000000, //mmPA_SC_VPORT_13_BR
+0x00000000, //mmPA_SC_VPORT_14_TL
+0x00000000, //mmPA_SC_VPORT_14_BR
+0x00000000, //mmPA_SC_VPORT_15_TL
+0x00000000, //mmPA_SC_VPORT_15_BR
+};
+
+static const unsigned int gfx12_SECT_CONTEXT_def_2[] = {
+0x00000000, //mmPA_CL_PROG_NEAR_CLIP_Z
+0x00000000, //mmPA_RATE_CNTL
+};
+
+static const unsigned int gfx12_SECT_CONTEXT_def_3[] = {
+0x00000000, //mmCP_PERFMON_CNTX_CNTL
+};
+
+static const unsigned int gfx12_SECT_CONTEXT_def_4[] = {
+0x00000000, //mmCONTEXT_RESERVED_REG0
+0x00000000, //mmCONTEXT_RESERVED_REG1
+0x00000000, //mmPA_SC_CLIPRECT_0_EXT
+0x00000000, //mmPA_SC_CLIPRECT_1_EXT
+0x00000000, //mmPA_SC_CLIPRECT_2_EXT
+0x00000000, //mmPA_SC_CLIPRECT_3_EXT
+};
+
+static const unsigned int gfx12_SECT_CONTEXT_def_5[] = {
+0x00000000, //mmPA_SC_HIZ_INFO
+0x00000000, //mmPA_SC_HIS_INFO
+0x00000000, //mmPA_SC_HIZ_BASE
+0x00000000, //mmPA_SC_HIZ_BASE_EXT
+0x00000000, //mmPA_SC_HIZ_SIZE_XY
+0x00000000, //mmPA_SC_HIS_BASE
+0x00000000, //mmPA_SC_HIS_BASE_EXT
+0x00000000, //mmPA_SC_HIS_SIZE_XY
+0x00000000, //mmPA_SC_BINNER_OUTPUT_TIMEOUT_CNTL
+0x00000000, //mmPA_SC_BINNER_DYNAMIC_BATCH_LIMIT
+0x00000000, //mmPA_SC_HISZ_CONTROL
+};
+
+static const unsigned int gfx12_SECT_CONTEXT_def_6[] = {
+0x00000000, //mmCB_MEM0_INFO
+0x00000000, //mmCB_MEM1_INFO
+0x00000000, //mmCB_MEM2_INFO
+0x00000000, //mmCB_MEM3_INFO
+0x00000000, //mmCB_MEM4_INFO
+0x00000000, //mmCB_MEM5_INFO
+0x00000000, //mmCB_MEM6_INFO
+0x00000000, //mmCB_MEM7_INFO
+};
+
+static const struct cs_extent_def gfx12_SECT_CONTEXT_defs[] = {
+    {gfx12_SECT_CONTEXT_def_1, 0x0000a03e, 34 },
+    {gfx12_SECT_CONTEXT_def_2, 0x0000a0cc, 2 },
+    {gfx12_SECT_CONTEXT_def_3, 0x0000a0d8, 1 },
+    {gfx12_SECT_CONTEXT_def_4, 0x0000a0db, 6 },
+    {gfx12_SECT_CONTEXT_def_5, 0x0000a2e5, 11 },
+    {gfx12_SECT_CONTEXT_def_6, 0x0000a3c0, 8 },
+    { 0, 0, 0 }
+};
+
+static const struct cs_section_def gfx12_cs_data[] = {
+    { gfx12_SECT_CONTEXT_defs, SECT_CONTEXT },
+    { 0, SECT_NONE }
+};
+
+#endif /* __CLEARSTATE_GFX12_H_ */
diff --git a/drivers/gpu/drm/amd/amdgpu/clearstate_gfx9.h b/drivers/gpu/drm/amd/amdgpu/clearstate_gfx9.h
index 003a131bad47..9c85ca6358c1 100644
--- a/drivers/gpu/drm/amd/amdgpu/clearstate_gfx9.h
+++ b/drivers/gpu/drm/amd/amdgpu/clearstate_gfx9.h
@@ -21,8 +21,7 @@
  *
  */
 
-static const unsigned int gfx9_SECT_CONTEXT_def_1[] =
-{
+static const unsigned int gfx9_SECT_CONTEXT_def_1[] = {
     0x00000000, // DB_RENDER_CONTROL
     0x00000000, // DB_COUNT_CONTROL
     0x00000000, // DB_DEPTH_VIEW
@@ -48,7 +47,7 @@ static const unsigned int gfx9_SECT_CONTEXT_def_1[] =
     0x00000000, // DB_STENCIL_WRITE_BASE
     0x00000000, // DB_STENCIL_WRITE_BASE_HI
     0x00000000, // DB_DFSM_CONTROL
-    0x00000000, // DB_RENDER_FILTER
+    0, // HOLE
     0x00000000, // DB_Z_INFO2
     0x00000000, // DB_STENCIL_INFO2
     0, // HOLE
@@ -236,8 +235,7 @@ static const unsigned int gfx9_SECT_CONTEXT_def_1[] =
     0x00000000, // PA_SC_VPORT_ZMIN_15
     0x3f800000, // PA_SC_VPORT_ZMAX_15
 };
-static const unsigned int gfx9_SECT_CONTEXT_def_2[] =
-{
+static const unsigned int gfx9_SECT_CONTEXT_def_2[] = {
     0x00000000, // PA_SC_SCREEN_EXTENT_CONTROL
     0x00000000, // PA_SC_TILE_STEERING_OVERRIDE
     0x00000000, // CP_PERFMON_CNTX_CNTL
@@ -259,8 +257,8 @@ static const unsigned int gfx9_SECT_CONTEXT_def_2[] =
     0x00000000, // PA_SC_RIGHT_VERT_GRID
     0x00000000, // PA_SC_LEFT_VERT_GRID
     0x00000000, // PA_SC_HORIZ_GRID
-    0x00000000, // PA_SC_FOV_WINDOW_LR
-    0x00000000, // PA_SC_FOV_WINDOW_TB
+    0, // HOLE
+    0, // HOLE
     0, // HOLE
     0, // HOLE
     0, // HOLE
@@ -521,15 +519,13 @@ static const unsigned int gfx9_SECT_CONTEXT_def_2[] =
     0x00000000, // CB_MRT6_EPITCH
     0x00000000, // CB_MRT7_EPITCH
 };
-static const unsigned int gfx9_SECT_CONTEXT_def_3[] =
-{
+static const unsigned int gfx9_SECT_CONTEXT_def_3[] = {
     0x00000000, // PA_CL_POINT_X_RAD
     0x00000000, // PA_CL_POINT_Y_RAD
     0x00000000, // PA_CL_POINT_SIZE
     0x00000000, // PA_CL_POINT_CULL_RAD
 };
-static const unsigned int gfx9_SECT_CONTEXT_def_4[] =
-{
+static const unsigned int gfx9_SECT_CONTEXT_def_4[] = {
     0x00000000, // DB_DEPTH_CONTROL
     0x00000000, // DB_EQAA
     0x00000000, // CB_COLOR_CONTROL
@@ -688,20 +684,17 @@ static const unsigned int gfx9_SECT_CONTEXT_def_4[] =
     0x00000000, // VGT_GS_OUT_PRIM_TYPE
     0x00000000, // IA_ENHANCE
 };
-static const unsigned int gfx9_SECT_CONTEXT_def_5[] =
-{
+static const unsigned int gfx9_SECT_CONTEXT_def_5[] = {
     0x00000000, // WD_ENHANCE
     0x00000000, // VGT_PRIMITIVEID_EN
 };
-static const unsigned int gfx9_SECT_CONTEXT_def_6[] =
-{
+static const unsigned int gfx9_SECT_CONTEXT_def_6[] = {
     0x00000000, // VGT_PRIMITIVEID_RESET
 };
-static const unsigned int gfx9_SECT_CONTEXT_def_7[] =
-{
+static const unsigned int gfx9_SECT_CONTEXT_def_7[] = {
     0x00000000, // VGT_GS_MAX_PRIMS_PER_SUBGROUP
     0x00000000, // VGT_DRAW_PAYLOAD_CNTL
-    0x00000000, // VGT_INDEX_PAYLOAD_CNTL
+    0, // HOLE
     0x00000000, // VGT_INSTANCE_STEP_RATE_0
     0x00000000, // VGT_INSTANCE_STEP_RATE_1
     0, // HOLE
@@ -766,8 +759,7 @@ static const unsigned int gfx9_SECT_CONTEXT_def_7[] =
     0x00000000, // VGT_STRMOUT_CONFIG
     0x00000000, // VGT_STRMOUT_BUFFER_CONFIG
 };
-static const unsigned int gfx9_SECT_CONTEXT_def_8[] =
-{
+static const unsigned int gfx9_SECT_CONTEXT_def_8[] = {
     0x00000000, // PA_SC_CENTROID_PRIORITY_0
     0x00000000, // PA_SC_CENTROID_PRIORITY_1
     0x00001000, // PA_SC_LINE_CNTL
@@ -924,8 +916,7 @@ static const unsigned int gfx9_SECT_CONTEXT_def_8[] =
     0x00000000, // CB_COLOR7_DCC_BASE
     0x00000000, // CB_COLOR7_DCC_BASE_EXT
 };
-static const struct cs_extent_def gfx9_SECT_CONTEXT_defs[] =
-{
+static const struct cs_extent_def gfx9_SECT_CONTEXT_defs[] = {
     {gfx9_SECT_CONTEXT_def_1, 0x0000a000, 212 },
     {gfx9_SECT_CONTEXT_def_2, 0x0000a0d6, 282 },
     {gfx9_SECT_CONTEXT_def_3, 0x0000a1f5, 4 },
diff --git a/drivers/gpu/drm/amd/amdgpu/clearstate_si.h b/drivers/gpu/drm/amd/amdgpu/clearstate_si.h
index 66e39cdb5cb0..5fd96ddd7f0f 100644
--- a/drivers/gpu/drm/amd/amdgpu/clearstate_si.h
+++ b/drivers/gpu/drm/amd/amdgpu/clearstate_si.h
@@ -21,8 +21,7 @@
  *
  */
 
-static const u32 si_SECT_CONTEXT_def_1[] =
-{
+static const u32 si_SECT_CONTEXT_def_1[] = {
     0x00000000, // DB_RENDER_CONTROL
     0x00000000, // DB_COUNT_CONTROL
     0x00000000, // DB_DEPTH_VIEW
@@ -236,8 +235,7 @@ static const u32 si_SECT_CONTEXT_def_1[] =
     0x00000000, // PA_SC_VPORT_ZMIN_15
     0x3f800000, // PA_SC_VPORT_ZMAX_15
 };
-static const u32 si_SECT_CONTEXT_def_2[] =
-{
+static const u32 si_SECT_CONTEXT_def_2[] = {
     0x00000000, // CP_PERFMON_CNTX_CNTL
     0x00000000, // CP_RINGID
     0x00000000, // CP_VMID
@@ -511,8 +509,7 @@ static const u32 si_SECT_CONTEXT_def_2[] =
     0x00000000, // CB_BLEND6_CONTROL
     0x00000000, // CB_BLEND7_CONTROL
 };
-static const u32 si_SECT_CONTEXT_def_3[] =
-{
+static const u32 si_SECT_CONTEXT_def_3[] = {
     0x00000000, // PA_CL_POINT_X_RAD
     0x00000000, // PA_CL_POINT_Y_RAD
     0x00000000, // PA_CL_POINT_SIZE
@@ -520,8 +517,7 @@ static const u32 si_SECT_CONTEXT_def_3[] =
     0x00000000, // VGT_DMA_BASE_HI
     0x00000000, // VGT_DMA_BASE
 };
-static const u32 si_SECT_CONTEXT_def_4[] =
-{
+static const u32 si_SECT_CONTEXT_def_4[] = {
     0x00000000, // DB_DEPTH_CONTROL
     0x00000000, // DB_EQAA
     0x00000000, // CB_COLOR_CONTROL
@@ -680,16 +676,13 @@ static const u32 si_SECT_CONTEXT_def_4[] =
     0x00000000, // VGT_GS_OUT_PRIM_TYPE
     0x00000000, // IA_ENHANCE
 };
-static const u32 si_SECT_CONTEXT_def_5[] =
-{
+static const u32 si_SECT_CONTEXT_def_5[] = {
     0x00000000, // VGT_PRIMITIVEID_EN
 };
-static const u32 si_SECT_CONTEXT_def_6[] =
-{
+static const u32 si_SECT_CONTEXT_def_6[] = {
     0x00000000, // VGT_PRIMITIVEID_RESET
 };
-static const u32 si_SECT_CONTEXT_def_7[] =
-{
+static const u32 si_SECT_CONTEXT_def_7[] = {
     0x00000000, // VGT_MULTI_PRIM_IB_RESET_EN
     0, // HOLE
     0, // HOLE
@@ -924,8 +917,7 @@ static const u32 si_SECT_CONTEXT_def_7[] =
     0x00000000, // CB_COLOR7_CLEAR_WORD0
     0x00000000, // CB_COLOR7_CLEAR_WORD1
 };
-static const struct cs_extent_def si_SECT_CONTEXT_defs[] =
-{
+static const struct cs_extent_def si_SECT_CONTEXT_defs[] = {
     {si_SECT_CONTEXT_def_1, 0x0000a000, 212 },
     {si_SECT_CONTEXT_def_2, 0x0000a0d8, 272 },
     {si_SECT_CONTEXT_def_3, 0x0000a1f5, 6 },
diff --git a/drivers/gpu/drm/amd/amdgpu/cyan_skillfish_reg_init.c b/drivers/gpu/drm/amd/amdgpu/cyan_skillfish_reg_init.c
new file mode 100644
index 000000000000..96616a865aac
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/cyan_skillfish_reg_init.c
@@ -0,0 +1,56 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "nv.h"
+
+#include "soc15_common.h"
+#include "soc15_hw_ip.h"
+#include "cyan_skillfish_ip_offset.h"
+
+int cyan_skillfish_reg_base_init(struct amdgpu_device *adev)
+{
+	/* HW has more IP blocks,  only initialized the blocke needed by driver */
+	uint32_t i;
+
+	adev->gfx.xcc_mask = 1;
+	for (i = 0 ; i < MAX_INSTANCE ; ++i) {
+		adev->reg_offset[GC_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i]));
+		adev->reg_offset[HDP_HWIP][i] = (uint32_t *)(&(HDP_BASE.instance[i]));
+		adev->reg_offset[MMHUB_HWIP][i] = (uint32_t *)(&(MMHUB_BASE.instance[i]));
+		adev->reg_offset[ATHUB_HWIP][i] = (uint32_t *)(&(ATHUB_BASE.instance[i]));
+		adev->reg_offset[NBIO_HWIP][i] = (uint32_t *)(&(NBIO_BASE.instance[i]));
+		adev->reg_offset[MP0_HWIP][i] = (uint32_t *)(&(MP0_BASE.instance[i]));
+		adev->reg_offset[MP1_HWIP][i] = (uint32_t *)(&(MP1_BASE.instance[i]));
+		adev->reg_offset[VCN_HWIP][i] = (uint32_t *)(&(UVD0_BASE.instance[i]));
+		adev->reg_offset[DF_HWIP][i] = (uint32_t *)(&(DF_BASE.instance[i]));
+		adev->reg_offset[DCE_HWIP][i] = (uint32_t *)(&(DMU_BASE.instance[i]));
+		adev->reg_offset[OSSSYS_HWIP][i] = (uint32_t *)(&(OSSSYS_BASE.instance[i]));
+		adev->reg_offset[SDMA0_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i]));
+		adev->reg_offset[SDMA1_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i]));
+		adev->reg_offset[SMUIO_HWIP][i] = (uint32_t *)(&(SMUIO_BASE.instance[i]));
+		adev->reg_offset[THM_HWIP][i] = (uint32_t *)(&(THM_BASE.instance[i]));
+		adev->reg_offset[CLK_HWIP][i] = (uint32_t *)(&(CLK_BASE.instance[i]));
+	}
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/cz_ih.c b/drivers/gpu/drm/amd/amdgpu/cz_ih.c
index 0c1209cdd1cb..2f891fb846d5 100644
--- a/drivers/gpu/drm/amd/amdgpu/cz_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/cz_ih.c
@@ -20,7 +20,9 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  *
  */
-#include <drm/drmP.h>
+
+#include <linux/pci.h>
+
 #include "amdgpu.h"
 #include "amdgpu_ih.h"
 #include "vid.h"
@@ -103,15 +105,15 @@ static void cz_ih_disable_interrupts(struct amdgpu_device *adev)
  */
 static int cz_ih_irq_init(struct amdgpu_device *adev)
 {
-	int rb_bufsz;
+	struct amdgpu_ih_ring *ih = &adev->irq.ih;
 	u32 interrupt_cntl, ih_cntl, ih_rb_cntl;
-	u64 wptr_off;
+	int rb_bufsz;
 
 	/* disable irqs */
 	cz_ih_disable_interrupts(adev);
 
 	/* setup interrupt control */
-	WREG32(mmINTERRUPT_CNTL2, adev->dummy_page.addr >> 8);
+	WREG32(mmINTERRUPT_CNTL2, adev->dummy_page_addr >> 8);
 	interrupt_cntl = RREG32(mmINTERRUPT_CNTL);
 	/* INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
 	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
@@ -133,9 +135,8 @@ static int cz_ih_irq_init(struct amdgpu_device *adev)
 	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_WRITEBACK_ENABLE, 1);
 
 	/* set the writeback address whether it's enabled or not */
-	wptr_off = adev->wb.gpu_addr + (adev->irq.ih.wptr_offs * 4);
-	WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(wptr_off));
-	WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(wptr_off) & 0xFF);
+	WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(ih->wptr_addr));
+	WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(ih->wptr_addr) & 0xFF);
 
 	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
 
@@ -178,6 +179,7 @@ static void cz_ih_irq_disable(struct amdgpu_device *adev)
  * cz_ih_get_wptr - get the IH ring buffer wptr
  *
  * @adev: amdgpu_device pointer
+ * @ih: IH ring buffer to fetch wptr
  *
  * Get the IH ring buffer wptr from either the register
  * or the writeback memory buffer (VI).  Also check for
@@ -185,74 +187,96 @@ static void cz_ih_irq_disable(struct amdgpu_device *adev)
  * Used by cz_irq_process(VI).
  * Returns the value of the wptr.
  */
-static u32 cz_ih_get_wptr(struct amdgpu_device *adev)
+static u32 cz_ih_get_wptr(struct amdgpu_device *adev,
+			  struct amdgpu_ih_ring *ih)
 {
 	u32 wptr, tmp;
 
-	wptr = le32_to_cpu(adev->wb.wb[adev->irq.ih.wptr_offs]);
-
-	if (REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW)) {
-		wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
-		/* When a ring buffer overflow happen start parsing interrupt
-		 * from the last not overwritten vector (wptr + 16). Hopefully
-		 * this should allow us to catchup.
-		 */
-		dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
-			wptr, adev->irq.ih.rptr, (wptr + 16) & adev->irq.ih.ptr_mask);
-		adev->irq.ih.rptr = (wptr + 16) & adev->irq.ih.ptr_mask;
-		tmp = RREG32(mmIH_RB_CNTL);
-		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
-		WREG32(mmIH_RB_CNTL, tmp);
-	}
-	return (wptr & adev->irq.ih.ptr_mask);
+	wptr = le32_to_cpu(*ih->wptr_cpu);
+
+	if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
+		goto out;
+
+	/* Double check that the overflow wasn't already cleared. */
+	wptr = RREG32(mmIH_RB_WPTR);
+
+	if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
+		goto out;
+
+	wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
+
+	/* When a ring buffer overflow happen start parsing interrupt
+	 * from the last not overwritten vector (wptr + 16). Hopefully
+	 * this should allow us to catchup.
+	 */
+	dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
+		wptr, ih->rptr, (wptr + 16) & ih->ptr_mask);
+	ih->rptr = (wptr + 16) & ih->ptr_mask;
+	tmp = RREG32(mmIH_RB_CNTL);
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+	WREG32(mmIH_RB_CNTL, tmp);
+
+	/* Unset the CLEAR_OVERFLOW bit immediately so new overflows
+	 * can be detected.
+	 */
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 0);
+	WREG32(mmIH_RB_CNTL, tmp);
+
+out:
+	return (wptr & ih->ptr_mask);
 }
 
 /**
  * cz_ih_decode_iv - decode an interrupt vector
  *
  * @adev: amdgpu_device pointer
+ * @ih: IH ring buffer to decode
+ * @entry: IV entry to place decoded information into
  *
  * Decodes the interrupt vector at the current rptr
  * position and also advance the position.
  */
 static void cz_ih_decode_iv(struct amdgpu_device *adev,
-				 struct amdgpu_iv_entry *entry)
+			    struct amdgpu_ih_ring *ih,
+			    struct amdgpu_iv_entry *entry)
 {
 	/* wptr/rptr are in bytes! */
-	u32 ring_index = adev->irq.ih.rptr >> 2;
+	u32 ring_index = ih->rptr >> 2;
 	uint32_t dw[4];
 
-	dw[0] = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
-	dw[1] = le32_to_cpu(adev->irq.ih.ring[ring_index + 1]);
-	dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
-	dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
+	dw[0] = le32_to_cpu(ih->ring[ring_index + 0]);
+	dw[1] = le32_to_cpu(ih->ring[ring_index + 1]);
+	dw[2] = le32_to_cpu(ih->ring[ring_index + 2]);
+	dw[3] = le32_to_cpu(ih->ring[ring_index + 3]);
 
-	entry->client_id = AMDGPU_IH_CLIENTID_LEGACY;
+	entry->client_id = AMDGPU_IRQ_CLIENTID_LEGACY;
 	entry->src_id = dw[0] & 0xff;
 	entry->src_data[0] = dw[1] & 0xfffffff;
 	entry->ring_id = dw[2] & 0xff;
-	entry->vm_id = (dw[2] >> 8) & 0xff;
-	entry->pas_id = (dw[2] >> 16) & 0xffff;
+	entry->vmid = (dw[2] >> 8) & 0xff;
+	entry->pasid = (dw[2] >> 16) & 0xffff;
 
 	/* wptr/rptr are in bytes! */
-	adev->irq.ih.rptr += 16;
+	ih->rptr += 16;
 }
 
 /**
  * cz_ih_set_rptr - set the IH ring buffer rptr
  *
  * @adev: amdgpu_device pointer
+ * @ih: IH ring buffer to set rptr
  *
  * Set the IH ring buffer rptr.
  */
-static void cz_ih_set_rptr(struct amdgpu_device *adev)
+static void cz_ih_set_rptr(struct amdgpu_device *adev,
+			   struct amdgpu_ih_ring *ih)
 {
-	WREG32(mmIH_RB_RPTR, adev->irq.ih.rptr);
+	WREG32(mmIH_RB_RPTR, ih->rptr);
 }
 
-static int cz_ih_early_init(void *handle)
+static int cz_ih_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int ret;
 
 	ret = amdgpu_irq_add_domain(adev);
@@ -264,12 +288,12 @@ static int cz_ih_early_init(void *handle)
 	return 0;
 }
 
-static int cz_ih_sw_init(void *handle)
+static int cz_ih_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	r = amdgpu_ih_ring_init(adev, 64 * 1024, false);
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih, 64 * 1024, false);
 	if (r)
 		return r;
 
@@ -278,21 +302,20 @@ static int cz_ih_sw_init(void *handle)
 	return r;
 }
 
-static int cz_ih_sw_fini(void *handle)
+static int cz_ih_sw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	amdgpu_irq_fini(adev);
-	amdgpu_ih_ring_fini(adev);
+	amdgpu_irq_fini_sw(adev);
 	amdgpu_irq_remove_domain(adev);
 
 	return 0;
 }
 
-static int cz_ih_hw_init(void *handle)
+static int cz_ih_hw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	r = cz_ih_irq_init(adev);
 	if (r)
@@ -301,32 +324,26 @@ static int cz_ih_hw_init(void *handle)
 	return 0;
 }
 
-static int cz_ih_hw_fini(void *handle)
+static int cz_ih_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	cz_ih_irq_disable(adev);
+	cz_ih_irq_disable(ip_block->adev);
 
 	return 0;
 }
 
-static int cz_ih_suspend(void *handle)
+static int cz_ih_suspend(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return cz_ih_hw_fini(adev);
+	return cz_ih_hw_fini(ip_block);
 }
 
-static int cz_ih_resume(void *handle)
+static int cz_ih_resume(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return cz_ih_hw_init(adev);
+	return cz_ih_hw_init(ip_block);
 }
 
-static bool cz_ih_is_idle(void *handle)
+static bool cz_ih_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 tmp = RREG32(mmSRBM_STATUS);
 
 	if (REG_GET_FIELD(tmp, SRBM_STATUS, IH_BUSY))
@@ -335,11 +352,11 @@ static bool cz_ih_is_idle(void *handle)
 	return true;
 }
 
-static int cz_ih_wait_for_idle(void *handle)
+static int cz_ih_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
 	unsigned i;
 	u32 tmp;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
 		/* read MC_STATUS */
@@ -351,10 +368,10 @@ static int cz_ih_wait_for_idle(void *handle)
 	return -ETIMEDOUT;
 }
 
-static int cz_ih_soft_reset(void *handle)
+static int cz_ih_soft_reset(struct amdgpu_ip_block *ip_block)
 {
 	u32 srbm_soft_reset = 0;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 tmp = RREG32(mmSRBM_STATUS);
 
 	if (tmp & SRBM_STATUS__IH_BUSY_MASK)
@@ -381,14 +398,14 @@ static int cz_ih_soft_reset(void *handle)
 	return 0;
 }
 
-static int cz_ih_set_clockgating_state(void *handle,
+static int cz_ih_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
 	// TODO
 	return 0;
 }
 
-static int cz_ih_set_powergating_state(void *handle,
+static int cz_ih_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	// TODO
@@ -398,7 +415,6 @@ static int cz_ih_set_powergating_state(void *handle,
 static const struct amd_ip_funcs cz_ih_ip_funcs = {
 	.name = "cz_ih",
 	.early_init = cz_ih_early_init,
-	.late_init = NULL,
 	.sw_init = cz_ih_sw_init,
 	.sw_fini = cz_ih_sw_fini,
 	.hw_init = cz_ih_hw_init,
@@ -420,8 +436,7 @@ static const struct amdgpu_ih_funcs cz_ih_funcs = {
 
 static void cz_ih_set_interrupt_funcs(struct amdgpu_device *adev)
 {
-	if (adev->irq.ih_funcs == NULL)
-		adev->irq.ih_funcs = &cz_ih_funcs;
+	adev->irq.ih_funcs = &cz_ih_funcs;
 }
 
 const struct amdgpu_ip_block_version cz_ih_ip_block =
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
index 4e519dc42916..72ca6538b2e4 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
@@ -20,7 +20,13 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  *
  */
-#include <drm/drmP.h>
+
+#include <drm/drm_edid.h>
+#include <drm/drm_fourcc.h>
+#include <drm/drm_modeset_helper.h>
+#include <drm/drm_modeset_helper_vtables.h>
+#include <drm/drm_vblank.h>
+
 #include "amdgpu.h"
 #include "amdgpu_pm.h"
 #include "amdgpu_i2c.h"
@@ -31,6 +37,7 @@
 #include "atombios_encoders.h"
 #include "amdgpu_pll.h"
 #include "amdgpu_connectors.h"
+#include "amdgpu_display.h"
 #include "dce_v10_0.h"
 
 #include "dce/dce_10_0_d.h"
@@ -41,11 +48,13 @@
 #include "gmc/gmc_8_1_d.h"
 #include "gmc/gmc_8_1_sh_mask.h"
 
+#include "ivsrcid/ivsrcid_vislands30.h"
+
 static void dce_v10_0_set_display_funcs(struct amdgpu_device *adev);
 static void dce_v10_0_set_irq_funcs(struct amdgpu_device *adev);
+static void dce_v10_0_hpd_int_ack(struct amdgpu_device *adev, int hpd);
 
-static const u32 crtc_offsets[] =
-{
+static const u32 crtc_offsets[] = {
 	CRTC0_REGISTER_OFFSET,
 	CRTC1_REGISTER_OFFSET,
 	CRTC2_REGISTER_OFFSET,
@@ -55,8 +64,7 @@ static const u32 crtc_offsets[] =
 	CRTC6_REGISTER_OFFSET
 };
 
-static const u32 hpd_offsets[] =
-{
+static const u32 hpd_offsets[] = {
 	HPD0_REGISTER_OFFSET,
 	HPD1_REGISTER_OFFSET,
 	HPD2_REGISTER_OFFSET,
@@ -113,30 +121,26 @@ static const struct {
 	.hpd = DISP_INTERRUPT_STATUS_CONTINUE5__DC_HPD6_INTERRUPT_MASK
 } };
 
-static const u32 golden_settings_tonga_a11[] =
-{
+static const u32 golden_settings_tonga_a11[] = {
 	mmDCI_CLK_CNTL, 0x00000080, 0x00000000,
 	mmFBC_DEBUG_COMP, 0x000000f0, 0x00000070,
 	mmFBC_MISC, 0x1f311fff, 0x12300000,
 	mmHDMI_CONTROL, 0x31000111, 0x00000011,
 };
 
-static const u32 tonga_mgcg_cgcg_init[] =
-{
+static const u32 tonga_mgcg_cgcg_init[] = {
 	mmXDMA_CLOCK_GATING_CNTL, 0xffffffff, 0x00000100,
 	mmXDMA_MEM_POWER_CNTL, 0x00000101, 0x00000000,
 };
 
-static const u32 golden_settings_fiji_a10[] =
-{
+static const u32 golden_settings_fiji_a10[] = {
 	mmDCI_CLK_CNTL, 0x00000080, 0x00000000,
 	mmFBC_DEBUG_COMP, 0x000000f0, 0x00000070,
 	mmFBC_MISC, 0x1f311fff, 0x12300000,
 	mmHDMI_CONTROL, 0x31000111, 0x00000011,
 };
 
-static const u32 fiji_mgcg_cgcg_init[] =
-{
+static const u32 fiji_mgcg_cgcg_init[] = {
 	mmXDMA_CLOCK_GATING_CNTL, 0xffffffff, 0x00000100,
 	mmXDMA_MEM_POWER_CNTL, 0x00000101, 0x00000000,
 };
@@ -145,20 +149,20 @@ static void dce_v10_0_init_golden_registers(struct amdgpu_device *adev)
 {
 	switch (adev->asic_type) {
 	case CHIP_FIJI:
-		amdgpu_program_register_sequence(adev,
-						 fiji_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(fiji_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_fiji_a10,
-						 (const u32)ARRAY_SIZE(golden_settings_fiji_a10));
+		amdgpu_device_program_register_sequence(adev,
+							fiji_mgcg_cgcg_init,
+							ARRAY_SIZE(fiji_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_fiji_a10,
+							ARRAY_SIZE(golden_settings_fiji_a10));
 		break;
 	case CHIP_TONGA:
-		amdgpu_program_register_sequence(adev,
-						 tonga_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(tonga_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_tonga_a11,
-						 (const u32)ARRAY_SIZE(golden_settings_tonga_a11));
+		amdgpu_device_program_register_sequence(adev,
+							tonga_mgcg_cgcg_init,
+							ARRAY_SIZE(tonga_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_tonga_a11,
+							ARRAY_SIZE(golden_settings_tonga_a11));
 		break;
 	default:
 		break;
@@ -190,66 +194,6 @@ static void dce_v10_0_audio_endpt_wreg(struct amdgpu_device *adev,
 	spin_unlock_irqrestore(&adev->audio_endpt_idx_lock, flags);
 }
 
-static bool dce_v10_0_is_in_vblank(struct amdgpu_device *adev, int crtc)
-{
-	if (RREG32(mmCRTC_STATUS + crtc_offsets[crtc]) &
-			CRTC_V_BLANK_START_END__CRTC_V_BLANK_START_MASK)
-		return true;
-	else
-		return false;
-}
-
-static bool dce_v10_0_is_counter_moving(struct amdgpu_device *adev, int crtc)
-{
-	u32 pos1, pos2;
-
-	pos1 = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
-	pos2 = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
-
-	if (pos1 != pos2)
-		return true;
-	else
-		return false;
-}
-
-/**
- * dce_v10_0_vblank_wait - vblank wait asic callback.
- *
- * @adev: amdgpu_device pointer
- * @crtc: crtc to wait for vblank on
- *
- * Wait for vblank on the requested crtc (evergreen+).
- */
-static void dce_v10_0_vblank_wait(struct amdgpu_device *adev, int crtc)
-{
-	unsigned i = 100;
-
-	if (crtc >= adev->mode_info.num_crtc)
-		return;
-
-	if (!(RREG32(mmCRTC_CONTROL + crtc_offsets[crtc]) & CRTC_CONTROL__CRTC_MASTER_EN_MASK))
-		return;
-
-	/* depending on when we hit vblank, we may be close to active; if so,
-	 * wait for another frame.
-	 */
-	while (dce_v10_0_is_in_vblank(adev, crtc)) {
-		if (i++ == 100) {
-			i = 0;
-			if (!dce_v10_0_is_counter_moving(adev, crtc))
-				break;
-		}
-	}
-
-	while (!dce_v10_0_is_in_vblank(adev, crtc)) {
-		if (i++ == 100) {
-			i = 0;
-			if (!dce_v10_0_is_counter_moving(adev, crtc))
-				break;
-		}
-	}
-}
-
 static u32 dce_v10_0_vblank_get_counter(struct amdgpu_device *adev, int crtc)
 {
 	if (crtc >= adev->mode_info.num_crtc)
@@ -282,6 +226,7 @@ static void dce_v10_0_pageflip_interrupt_fini(struct amdgpu_device *adev)
  * @adev: amdgpu_device pointer
  * @crtc_id: crtc to cleanup pageflip on
  * @crtc_base: new address of the crtc (GPU MC address)
+ * @async: asynchronous flip
  *
  * Triggers the actual pageflip by updating the primary
  * surface base address.
@@ -290,6 +235,7 @@ static void dce_v10_0_page_flip(struct amdgpu_device *adev,
 				int crtc_id, u64 crtc_base, bool async)
 {
 	struct amdgpu_crtc *amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
+	struct drm_framebuffer *fb = amdgpu_crtc->base.primary->fb;
 	u32 tmp;
 
 	/* flip at hsync for async, default is vsync */
@@ -297,6 +243,9 @@ static void dce_v10_0_page_flip(struct amdgpu_device *adev,
 	tmp = REG_SET_FIELD(tmp, GRPH_FLIP_CONTROL,
 			    GRPH_SURFACE_UPDATE_H_RETRACE_EN, async ? 1 : 0);
 	WREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	/* update pitch */
+	WREG32(mmGRPH_PITCH + amdgpu_crtc->crtc_offset,
+	       fb->pitches[0] / fb->format->cpp[0]);
 	/* update the primary scanout address */
 	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
 	       upper_32_bits(crtc_base));
@@ -378,11 +327,13 @@ static void dce_v10_0_hpd_set_polarity(struct amdgpu_device *adev,
  */
 static void dce_v10_0_hpd_init(struct amdgpu_device *adev)
 {
-	struct drm_device *dev = adev->ddev;
+	struct drm_device *dev = adev_to_drm(adev);
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	u32 tmp;
 
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 
 		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
@@ -414,10 +365,12 @@ static void dce_v10_0_hpd_init(struct amdgpu_device *adev)
 				    AMDGPU_HPD_DISCONNECT_INT_DELAY_IN_MS);
 		WREG32(mmDC_HPD_TOGGLE_FILT_CNTL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
 
+		dce_v10_0_hpd_int_ack(adev, amdgpu_connector->hpd.hpd);
 		dce_v10_0_hpd_set_polarity(adev, amdgpu_connector->hpd.hpd);
 		amdgpu_irq_get(adev, &adev->hpd_irq,
 			       amdgpu_connector->hpd.hpd);
 	}
+	drm_connector_list_iter_end(&iter);
 }
 
 /**
@@ -430,11 +383,13 @@ static void dce_v10_0_hpd_init(struct amdgpu_device *adev)
  */
 static void dce_v10_0_hpd_fini(struct amdgpu_device *adev)
 {
-	struct drm_device *dev = adev->ddev;
+	struct drm_device *dev = adev_to_drm(adev);
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	u32 tmp;
 
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 
 		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
@@ -447,6 +402,7 @@ static void dce_v10_0_hpd_fini(struct amdgpu_device *adev)
 		amdgpu_irq_put(adev, &adev->hpd_irq,
 			       amdgpu_connector->hpd.hpd);
 	}
+	drm_connector_list_iter_end(&iter);
 }
 
 static u32 dce_v10_0_hpd_get_gpio_reg(struct amdgpu_device *adev)
@@ -548,7 +504,7 @@ void dce_v10_0_disable_dce(struct amdgpu_device *adev)
 static void dce_v10_0_program_fmt(struct drm_encoder *encoder)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
 	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
@@ -1083,7 +1039,7 @@ static void dce_v10_0_program_watermarks(struct amdgpu_device *adev,
 					    (u32)mode->clock);
 		line_time = (u32) div_u64((u64)mode->crtc_htotal * 1000000,
 					  (u32)mode->clock);
-		line_time = min(line_time, (u32)65535);
+		line_time = min_t(u32, line_time, 65535);
 
 		/* watermark for high clocks */
 		if (adev->pm.dpm_enabled) {
@@ -1113,7 +1069,7 @@ static void dce_v10_0_program_watermarks(struct amdgpu_device *adev,
 		wm_high.num_heads = num_heads;
 
 		/* set for high clocks */
-		latency_watermark_a = min(dce_v10_0_latency_watermark(&wm_high), (u32)65535);
+		latency_watermark_a = min_t(u32, dce_v10_0_latency_watermark(&wm_high), 65535);
 
 		/* possibly force display priority to high */
 		/* should really do this at mode validation time... */
@@ -1152,7 +1108,7 @@ static void dce_v10_0_program_watermarks(struct amdgpu_device *adev,
 		wm_low.num_heads = num_heads;
 
 		/* set for low clocks */
-		latency_watermark_b = min(dce_v10_0_latency_watermark(&wm_low), (u32)65535);
+		latency_watermark_b = min_t(u32, dce_v10_0_latency_watermark(&wm_low), 65535);
 
 		/* possibly force display priority to high */
 		/* should really do this at mode validation time... */
@@ -1185,8 +1141,7 @@ static void dce_v10_0_program_watermarks(struct amdgpu_device *adev,
 
 	/* save values for DPM */
 	amdgpu_crtc->line_time = line_time;
-	amdgpu_crtc->wm_high = latency_watermark_a;
-	amdgpu_crtc->wm_low = latency_watermark_b;
+
 	/* Save number of lines the linebuffer leads before the scanout */
 	amdgpu_crtc->lb_vblank_lead_lines = lb_vblank_lead_lines;
 }
@@ -1205,7 +1160,7 @@ static void dce_v10_0_bandwidth_update(struct amdgpu_device *adev)
 	u32 num_heads = 0, lb_size;
 	int i;
 
-	amdgpu_update_display_priority(adev);
+	amdgpu_display_update_priority(adev);
 
 	for (i = 0; i < adev->mode_info.num_crtc; i++) {
 		if (adev->mode_info.crtcs[i]->base.enabled)
@@ -1253,7 +1208,7 @@ static struct amdgpu_audio_pin *dce_v10_0_audio_get_pin(struct amdgpu_device *ad
 
 static void dce_v10_0_afmt_audio_select_pin(struct drm_encoder *encoder)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(encoder->dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	u32 tmp;
@@ -1269,10 +1224,12 @@ static void dce_v10_0_afmt_audio_select_pin(struct drm_encoder *encoder)
 static void dce_v10_0_audio_write_latency_fields(struct drm_encoder *encoder,
 						struct drm_display_mode *mode)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	u32 tmp;
 	int interlace = 0;
@@ -1280,12 +1237,14 @@ static void dce_v10_0_audio_write_latency_fields(struct drm_encoder *encoder,
 	if (!dig || !dig->afmt || !dig->afmt->pin)
 		return;
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
@@ -1311,10 +1270,12 @@ static void dce_v10_0_audio_write_latency_fields(struct drm_encoder *encoder,
 
 static void dce_v10_0_audio_write_speaker_allocation(struct drm_encoder *encoder)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	u32 tmp;
 	u8 *sadb = NULL;
@@ -1323,19 +1284,21 @@ static void dce_v10_0_audio_write_speaker_allocation(struct drm_encoder *encoder
 	if (!dig || !dig->afmt || !dig->afmt->pin)
 		return;
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
 		return;
 	}
 
-	sad_count = drm_edid_to_speaker_allocation(amdgpu_connector_edid(connector), &sadb);
+	sad_count = drm_edid_to_speaker_allocation(amdgpu_connector->edid, &sadb);
 	if (sad_count < 0) {
 		DRM_ERROR("Couldn't read Speaker Allocation Data Block: %d\n", sad_count);
 		sad_count = 0;
@@ -1363,10 +1326,12 @@ static void dce_v10_0_audio_write_speaker_allocation(struct drm_encoder *encoder
 
 static void dce_v10_0_audio_write_sad_regs(struct drm_encoder *encoder)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	struct cea_sad *sads;
 	int i, sad_count;
@@ -1389,23 +1354,25 @@ static void dce_v10_0_audio_write_sad_regs(struct drm_encoder *encoder)
 	if (!dig || !dig->afmt || !dig->afmt->pin)
 		return;
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
 		return;
 	}
 
-	sad_count = drm_edid_to_sad(amdgpu_connector_edid(connector), &sads);
-	if (sad_count <= 0) {
+	sad_count = drm_edid_to_sad(amdgpu_connector->edid, &sads);
+	if (sad_count < 0)
 		DRM_ERROR("Couldn't read SADs: %d\n", sad_count);
+	if (sad_count <= 0)
 		return;
-	}
 	BUG_ON(!sads);
 
 	for (i = 0; i < ARRAY_SIZE(eld_reg_to_type); i++) {
@@ -1454,8 +1421,7 @@ static void dce_v10_0_audio_enable(struct amdgpu_device *adev,
 			   enable ? AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL__AUDIO_ENABLED_MASK : 0);
 }
 
-static const u32 pin_offsets[] =
-{
+static const u32 pin_offsets[] = {
 	AUD0_REGISTER_OFFSET,
 	AUD1_REGISTER_OFFSET,
 	AUD2_REGISTER_OFFSET,
@@ -1495,17 +1461,12 @@ static int dce_v10_0_audio_init(struct amdgpu_device *adev)
 
 static void dce_v10_0_audio_fini(struct amdgpu_device *adev)
 {
-	int i;
-
 	if (!amdgpu_audio)
 		return;
 
 	if (!adev->mode_info.audio.enabled)
 		return;
 
-	for (i = 0; i < adev->mode_info.audio.num_pins; i++)
-		dce_v10_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
-
 	adev->mode_info.audio.enabled = false;
 }
 
@@ -1515,7 +1476,7 @@ static void dce_v10_0_audio_fini(struct amdgpu_device *adev)
 static void dce_v10_0_afmt_update_ACR(struct drm_encoder *encoder, uint32_t clock)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_afmt_acr acr = amdgpu_afmt_acr(clock);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
@@ -1551,7 +1512,7 @@ static void dce_v10_0_afmt_update_avi_infoframe(struct drm_encoder *encoder,
 					       void *buffer, size_t size)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	uint8_t *frame = buffer + 3;
@@ -1570,7 +1531,7 @@ static void dce_v10_0_afmt_update_avi_infoframe(struct drm_encoder *encoder,
 static void dce_v10_0_audio_set_dto(struct drm_encoder *encoder, u32 clock)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
@@ -1601,7 +1562,7 @@ static void dce_v10_0_afmt_setmode(struct drm_encoder *encoder,
 				  struct drm_display_mode *mode)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
@@ -1739,7 +1700,7 @@ static void dce_v10_0_afmt_setmode(struct drm_encoder *encoder,
 	dce_v10_0_audio_write_sad_regs(encoder);
 	dce_v10_0_audio_write_latency_fields(encoder, mode);
 
-	err = drm_hdmi_avi_infoframe_from_display_mode(&frame, mode, false);
+	err = drm_hdmi_avi_infoframe_from_display_mode(&frame, connector, mode);
 	if (err < 0) {
 		DRM_ERROR("failed to setup AVI infoframe: %zd\n", err);
 		return;
@@ -1781,7 +1742,7 @@ static void dce_v10_0_afmt_setmode(struct drm_encoder *encoder,
 static void dce_v10_0_afmt_enable(struct drm_encoder *encoder, bool enable)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 
@@ -1840,8 +1801,7 @@ static void dce_v10_0_afmt_fini(struct amdgpu_device *adev)
 	}
 }
 
-static const u32 vga_control_regs[6] =
-{
+static const u32 vga_control_regs[6] = {
 	mmD1VGA_CONTROL,
 	mmD2VGA_CONTROL,
 	mmD3VGA_CONTROL,
@@ -1854,7 +1814,7 @@ static void dce_v10_0_vga_enable(struct drm_crtc *crtc, bool enable)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	u32 vga_control;
 
 	vga_control = RREG32(vga_control_regs[amdgpu_crtc->crtc_id]) & ~1;
@@ -1868,7 +1828,7 @@ static void dce_v10_0_grph_enable(struct drm_crtc *crtc, bool enable)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 
 	if (enable)
 		WREG32(mmGRPH_ENABLE + amdgpu_crtc->crtc_offset, 1);
@@ -1882,8 +1842,7 @@ static int dce_v10_0_crtc_do_set_base(struct drm_crtc *crtc,
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-	struct amdgpu_framebuffer *amdgpu_fb;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct drm_framebuffer *target_fb;
 	struct drm_gem_object *obj;
 	struct amdgpu_bo *abo;
@@ -1894,7 +1853,6 @@ static int dce_v10_0_crtc_do_set_base(struct drm_crtc *crtc,
 	u32 tmp, viewport_w, viewport_h;
 	int r;
 	bool bypass_lut = false;
-	struct drm_format_name_buf format_name;
 
 	/* no fb bound */
 	if (!atomic && !crtc->primary->fb) {
@@ -1902,32 +1860,29 @@ static int dce_v10_0_crtc_do_set_base(struct drm_crtc *crtc,
 		return 0;
 	}
 
-	if (atomic) {
-		amdgpu_fb = to_amdgpu_framebuffer(fb);
+	if (atomic)
 		target_fb = fb;
-	} else {
-		amdgpu_fb = to_amdgpu_framebuffer(crtc->primary->fb);
+	else
 		target_fb = crtc->primary->fb;
-	}
 
 	/* If atomic, assume fb object is pinned & idle & fenced and
 	 * just update base pointers
 	 */
-	obj = amdgpu_fb->obj;
+	obj = target_fb->obj[0];
 	abo = gem_to_amdgpu_bo(obj);
 	r = amdgpu_bo_reserve(abo, false);
 	if (unlikely(r != 0))
 		return r;
 
-	if (atomic) {
-		fb_location = amdgpu_bo_gpu_offset(abo);
-	} else {
-		r = amdgpu_bo_pin(abo, AMDGPU_GEM_DOMAIN_VRAM, &fb_location);
+	if (!atomic) {
+		abo->flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+		r = amdgpu_bo_pin(abo, AMDGPU_GEM_DOMAIN_VRAM);
 		if (unlikely(r != 0)) {
 			amdgpu_bo_unreserve(abo);
 			return -EINVAL;
 		}
 	}
+	fb_location = amdgpu_bo_gpu_offset(abo);
 
 	amdgpu_bo_get_tiling_flags(abo, &tiling_flags);
 	amdgpu_bo_unreserve(abo);
@@ -2005,9 +1960,20 @@ static int dce_v10_0_crtc_do_set_base(struct drm_crtc *crtc,
 		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
 		bypass_lut = true;
 		break;
+	case DRM_FORMAT_XBGR8888:
+	case DRM_FORMAT_ABGR8888:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 2);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 0);
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_RED_CROSSBAR, 2);
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_BLUE_CROSSBAR, 2);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN32);
+#endif
+		break;
 	default:
-		DRM_ERROR("Unsupported screen format %s\n",
-		          drm_get_format_name(target_fb->format->format, &format_name));
+		DRM_ERROR("Unsupported screen format %p4cc\n",
+			  &target_fb->format->format);
 		return -EINVAL;
 	}
 
@@ -2103,8 +2069,7 @@ static int dce_v10_0_crtc_do_set_base(struct drm_crtc *crtc,
 	WREG32(mmMASTER_UPDATE_MODE + amdgpu_crtc->crtc_offset, 0);
 
 	if (!atomic && fb && fb != crtc->primary->fb) {
-		amdgpu_fb = to_amdgpu_framebuffer(fb);
-		abo = gem_to_amdgpu_bo(amdgpu_fb->obj);
+		abo = gem_to_amdgpu_bo(fb->obj[0]);
 		r = amdgpu_bo_reserve(abo, true);
 		if (unlikely(r != 0))
 			return r;
@@ -2122,7 +2087,7 @@ static void dce_v10_0_set_interleave(struct drm_crtc *crtc,
 				     struct drm_display_mode *mode)
 {
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	u32 tmp;
 
@@ -2138,7 +2103,7 @@ static void dce_v10_0_crtc_load_lut(struct drm_crtc *crtc)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	u16 *r, *g, *b;
 	int i;
 	u32 tmp;
@@ -2229,22 +2194,18 @@ static int dce_v10_0_pick_dig_encoder(struct drm_encoder *encoder)
 			return 1;
 		else
 			return 0;
-		break;
 	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
 		if (dig->linkb)
 			return 3;
 		else
 			return 2;
-		break;
 	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
 		if (dig->linkb)
 			return 5;
 		else
 			return 4;
-		break;
 	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
 		return 6;
-		break;
 	default:
 		DRM_ERROR("invalid encoder_id: 0x%x\n", amdgpu_encoder->encoder_id);
 		return 0;
@@ -2277,7 +2238,7 @@ static u32 dce_v10_0_pick_pll(struct drm_crtc *crtc)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	u32 pll_in_use;
 	int pll;
 
@@ -2312,7 +2273,7 @@ static u32 dce_v10_0_pick_pll(struct drm_crtc *crtc)
 
 static void dce_v10_0_lock_cursor(struct drm_crtc *crtc, bool lock)
 {
-	struct amdgpu_device *adev = crtc->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	uint32_t cur_lock;
 
@@ -2327,18 +2288,18 @@ static void dce_v10_0_lock_cursor(struct drm_crtc *crtc, bool lock)
 static void dce_v10_0_hide_cursor(struct drm_crtc *crtc)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct amdgpu_device *adev = crtc->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
 	u32 tmp;
 
-	tmp = RREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = RREG32(mmCUR_CONTROL + amdgpu_crtc->crtc_offset);
 	tmp = REG_SET_FIELD(tmp, CUR_CONTROL, CURSOR_EN, 0);
-	WREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	WREG32(mmCUR_CONTROL + amdgpu_crtc->crtc_offset, tmp);
 }
 
 static void dce_v10_0_show_cursor(struct drm_crtc *crtc)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct amdgpu_device *adev = crtc->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
 	u32 tmp;
 
 	WREG32(mmCUR_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
@@ -2346,17 +2307,17 @@ static void dce_v10_0_show_cursor(struct drm_crtc *crtc)
 	WREG32(mmCUR_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
 	       lower_32_bits(amdgpu_crtc->cursor_addr));
 
-	tmp = RREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = RREG32(mmCUR_CONTROL + amdgpu_crtc->crtc_offset);
 	tmp = REG_SET_FIELD(tmp, CUR_CONTROL, CURSOR_EN, 1);
 	tmp = REG_SET_FIELD(tmp, CUR_CONTROL, CURSOR_MODE, 2);
-	WREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	WREG32(mmCUR_CONTROL + amdgpu_crtc->crtc_offset, tmp);
 }
 
 static int dce_v10_0_cursor_move_locked(struct drm_crtc *crtc,
 					int x, int y)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct amdgpu_device *adev = crtc->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
 	int xorigin = 0, yorigin = 0;
 
 	amdgpu_crtc->cursor_x = x;
@@ -2431,17 +2392,19 @@ static int dce_v10_0_crtc_cursor_set2(struct drm_crtc *crtc,
 	aobj = gem_to_amdgpu_bo(obj);
 	ret = amdgpu_bo_reserve(aobj, false);
 	if (ret != 0) {
-		drm_gem_object_put_unlocked(obj);
+		drm_gem_object_put(obj);
 		return ret;
 	}
 
-	ret = amdgpu_bo_pin(aobj, AMDGPU_GEM_DOMAIN_VRAM, &amdgpu_crtc->cursor_addr);
+	aobj->flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+	ret = amdgpu_bo_pin(aobj, AMDGPU_GEM_DOMAIN_VRAM);
 	amdgpu_bo_unreserve(aobj);
 	if (ret) {
 		DRM_ERROR("Failed to pin new cursor BO (%d)\n", ret);
-		drm_gem_object_put_unlocked(obj);
+		drm_gem_object_put(obj);
 		return ret;
 	}
+	amdgpu_crtc->cursor_addr = amdgpu_bo_gpu_offset(aobj);
 
 	dce_v10_0_lock_cursor(crtc, true);
 
@@ -2473,7 +2436,7 @@ unpin:
 			amdgpu_bo_unpin(aobj);
 			amdgpu_bo_unreserve(aobj);
 		}
-		drm_gem_object_put_unlocked(amdgpu_crtc->cursor_bo);
+		drm_gem_object_put(amdgpu_crtc->cursor_bo);
 	}
 
 	amdgpu_crtc->cursor_bo = obj;
@@ -2517,15 +2480,19 @@ static const struct drm_crtc_funcs dce_v10_0_crtc_funcs = {
 	.cursor_set2 = dce_v10_0_crtc_cursor_set2,
 	.cursor_move = dce_v10_0_crtc_cursor_move,
 	.gamma_set = dce_v10_0_crtc_gamma_set,
-	.set_config = amdgpu_crtc_set_config,
+	.set_config = amdgpu_display_crtc_set_config,
 	.destroy = dce_v10_0_crtc_destroy,
-	.page_flip_target = amdgpu_crtc_page_flip_target,
+	.page_flip_target = amdgpu_display_crtc_page_flip_target,
+	.get_vblank_counter = amdgpu_get_vblank_counter_kms,
+	.enable_vblank = amdgpu_enable_vblank_kms,
+	.disable_vblank = amdgpu_disable_vblank_kms,
+	.get_vblank_timestamp = drm_crtc_vblank_helper_get_vblank_timestamp,
 };
 
 static void dce_v10_0_crtc_dpms(struct drm_crtc *crtc, int mode)
 {
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	unsigned type;
 
@@ -2537,7 +2504,8 @@ static void dce_v10_0_crtc_dpms(struct drm_crtc *crtc, int mode)
 		amdgpu_atombios_crtc_blank(crtc, ATOM_DISABLE);
 		dce_v10_0_vga_enable(crtc, false);
 		/* Make sure VBLANK and PFLIP interrupts are still enabled */
-		type = amdgpu_crtc_idx_to_irq_type(adev, amdgpu_crtc->crtc_id);
+		type = amdgpu_display_crtc_idx_to_irq_type(adev,
+						amdgpu_crtc->crtc_id);
 		amdgpu_irq_update(adev, &adev->crtc_irq, type);
 		amdgpu_irq_update(adev, &adev->pageflip_irq, type);
 		drm_crtc_vblank_on(crtc);
@@ -2557,7 +2525,7 @@ static void dce_v10_0_crtc_dpms(struct drm_crtc *crtc, int mode)
 		break;
 	}
 	/* adjust pm to dpms */
-	amdgpu_pm_compute_clocks(adev);
+	amdgpu_dpm_compute_clocks(adev);
 }
 
 static void dce_v10_0_crtc_prepare(struct drm_crtc *crtc)
@@ -2578,18 +2546,16 @@ static void dce_v10_0_crtc_disable(struct drm_crtc *crtc)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_atom_ss ss;
 	int i;
 
 	dce_v10_0_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
 	if (crtc->primary->fb) {
 		int r;
-		struct amdgpu_framebuffer *amdgpu_fb;
 		struct amdgpu_bo *abo;
 
-		amdgpu_fb = to_amdgpu_framebuffer(crtc->primary->fb);
-		abo = gem_to_amdgpu_bo(amdgpu_fb->obj);
+		abo = gem_to_amdgpu_bo(crtc->primary->fb->obj[0]);
 		r = amdgpu_bo_reserve(abo, true);
 		if (unlikely(r))
 			DRM_ERROR("failed to reserve abo before unpin\n");
@@ -2676,7 +2642,7 @@ static bool dce_v10_0_crtc_mode_fixup(struct drm_crtc *crtc,
 		amdgpu_crtc->connector = NULL;
 		return false;
 	}
-	if (!amdgpu_crtc_scaling_mode_fixup(crtc, mode, adjusted_mode))
+	if (!amdgpu_display_crtc_scaling_mode_fixup(crtc, mode, adjusted_mode))
 		return false;
 	if (amdgpu_atombios_crtc_prepare_pll(crtc, adjusted_mode))
 		return false;
@@ -2700,7 +2666,7 @@ static int dce_v10_0_crtc_set_base_atomic(struct drm_crtc *crtc,
 					 struct drm_framebuffer *fb,
 					 int x, int y, enum mode_set_atomic state)
 {
-       return dce_v10_0_crtc_do_set_base(crtc, fb, x, y, 1);
+	return dce_v10_0_crtc_do_set_base(crtc, fb, x, y, 1);
 }
 
 static const struct drm_crtc_helper_funcs dce_v10_0_crtc_helper_funcs = {
@@ -2712,6 +2678,33 @@ static const struct drm_crtc_helper_funcs dce_v10_0_crtc_helper_funcs = {
 	.prepare = dce_v10_0_crtc_prepare,
 	.commit = dce_v10_0_crtc_commit,
 	.disable = dce_v10_0_crtc_disable,
+	.get_scanout_position = amdgpu_crtc_get_scanout_position,
+};
+
+static void dce_v10_0_panic_flush(struct drm_plane *plane)
+{
+	struct drm_framebuffer *fb;
+	struct amdgpu_crtc *amdgpu_crtc;
+	struct amdgpu_device *adev;
+	uint32_t fb_format;
+
+	if (!plane->fb)
+		return;
+
+	fb = plane->fb;
+	amdgpu_crtc = to_amdgpu_crtc(plane->crtc);
+	adev = drm_to_adev(fb->dev);
+
+	/* Disable DC tiling */
+	fb_format = RREG32(mmGRPH_CONTROL + amdgpu_crtc->crtc_offset);
+	fb_format &= ~GRPH_CONTROL__GRPH_ARRAY_MODE_MASK;
+	WREG32(mmGRPH_CONTROL + amdgpu_crtc->crtc_offset, fb_format);
+
+}
+
+static const struct drm_plane_helper_funcs dce_v10_0_drm_primary_plane_helper_funcs = {
+	.get_scanout_buffer = amdgpu_display_get_scanout_buffer,
+	.panic_flush = dce_v10_0_panic_flush,
 };
 
 static int dce_v10_0_crtc_init(struct amdgpu_device *adev, int index)
@@ -2723,7 +2716,7 @@ static int dce_v10_0_crtc_init(struct amdgpu_device *adev, int index)
 	if (amdgpu_crtc == NULL)
 		return -ENOMEM;
 
-	drm_crtc_init(adev->ddev, &amdgpu_crtc->base, &dce_v10_0_crtc_funcs);
+	drm_crtc_init(adev_to_drm(adev), &amdgpu_crtc->base, &dce_v10_0_crtc_funcs);
 
 	drm_mode_crtc_set_gamma_size(&amdgpu_crtc->base, 256);
 	amdgpu_crtc->crtc_id = index;
@@ -2731,8 +2724,8 @@ static int dce_v10_0_crtc_init(struct amdgpu_device *adev, int index)
 
 	amdgpu_crtc->max_cursor_width = 128;
 	amdgpu_crtc->max_cursor_height = 128;
-	adev->ddev->mode_config.cursor_width = amdgpu_crtc->max_cursor_width;
-	adev->ddev->mode_config.cursor_height = amdgpu_crtc->max_cursor_height;
+	adev_to_drm(adev)->mode_config.cursor_width = amdgpu_crtc->max_cursor_width;
+	adev_to_drm(adev)->mode_config.cursor_height = amdgpu_crtc->max_cursor_height;
 
 	switch (amdgpu_crtc->crtc_id) {
 	case 0:
@@ -2761,19 +2754,19 @@ static int dce_v10_0_crtc_init(struct amdgpu_device *adev, int index)
 	amdgpu_crtc->encoder = NULL;
 	amdgpu_crtc->connector = NULL;
 	drm_crtc_helper_add(&amdgpu_crtc->base, &dce_v10_0_crtc_helper_funcs);
+	drm_plane_helper_add(amdgpu_crtc->base.primary, &dce_v10_0_drm_primary_plane_helper_funcs);
 
 	return 0;
 }
 
-static int dce_v10_0_early_init(void *handle)
+static int dce_v10_0_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	adev->audio_endpt_rreg = &dce_v10_0_audio_endpt_rreg;
 	adev->audio_endpt_wreg = &dce_v10_0_audio_endpt_wreg;
 
 	dce_v10_0_set_display_funcs(adev);
-	dce_v10_0_set_irq_funcs(adev);
 
 	adev->mode_info.num_crtc = dce_v10_0_get_num_crtc(adev);
 
@@ -2788,49 +2781,51 @@ static int dce_v10_0_early_init(void *handle)
 		return -EINVAL;
 	}
 
+	dce_v10_0_set_irq_funcs(adev);
+
 	return 0;
 }
 
-static int dce_v10_0_sw_init(void *handle)
+static int dce_v10_0_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r, i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->mode_info.num_crtc; i++) {
-		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i + 1, &adev->crtc_irq);
+		r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, i + 1, &adev->crtc_irq);
 		if (r)
 			return r;
 	}
 
-	for (i = 8; i < 20; i += 2) {
-		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i, &adev->pageflip_irq);
+	for (i = VISLANDS30_IV_SRCID_D1_GRPH_PFLIP; i < 20; i += 2) {
+		r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, i, &adev->pageflip_irq);
 		if (r)
 			return r;
 	}
 
 	/* HPD hotplug */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 42, &adev->hpd_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_HOTPLUG_DETECT_A, &adev->hpd_irq);
 	if (r)
 		return r;
 
-	adev->ddev->mode_config.funcs = &amdgpu_mode_funcs;
+	adev_to_drm(adev)->mode_config.funcs = &amdgpu_mode_funcs;
 
-	adev->ddev->mode_config.async_page_flip = true;
+	adev_to_drm(adev)->mode_config.async_page_flip = true;
 
-	adev->ddev->mode_config.max_width = 16384;
-	adev->ddev->mode_config.max_height = 16384;
+	adev_to_drm(adev)->mode_config.max_width = 16384;
+	adev_to_drm(adev)->mode_config.max_height = 16384;
 
-	adev->ddev->mode_config.preferred_depth = 24;
-	adev->ddev->mode_config.prefer_shadow = 1;
+	adev_to_drm(adev)->mode_config.preferred_depth = 24;
+	adev_to_drm(adev)->mode_config.prefer_shadow = 1;
 
-	adev->ddev->mode_config.fb_base = adev->mc.aper_base;
+	adev_to_drm(adev)->mode_config.fb_modifiers_not_supported = true;
 
-	r = amdgpu_modeset_create_props(adev);
+	r = amdgpu_display_modeset_create_props(adev);
 	if (r)
 		return r;
 
-	adev->ddev->mode_config.max_width = 16384;
-	adev->ddev->mode_config.max_height = 16384;
+	adev_to_drm(adev)->mode_config.max_width = 16384;
+	adev_to_drm(adev)->mode_config.max_height = 16384;
 
 	/* allocate crtcs */
 	for (i = 0; i < adev->mode_info.num_crtc; i++) {
@@ -2840,7 +2835,7 @@ static int dce_v10_0_sw_init(void *handle)
 	}
 
 	if (amdgpu_atombios_get_connector_info_from_object_table(adev))
-		amdgpu_print_display_setup(adev->ddev);
+		amdgpu_display_print_display_setup(adev_to_drm(adev));
 	else
 		return -EINVAL;
 
@@ -2853,34 +2848,45 @@ static int dce_v10_0_sw_init(void *handle)
 	if (r)
 		return r;
 
-	drm_kms_helper_poll_init(adev->ddev);
+	/* Disable vblank IRQs aggressively for power-saving */
+	/* XXX: can this be enabled for DC? */
+	adev_to_drm(adev)->vblank_disable_immediate = true;
+
+	r = drm_vblank_init(adev_to_drm(adev), adev->mode_info.num_crtc);
+	if (r)
+		return r;
+
+	INIT_DELAYED_WORK(&adev->hotplug_work,
+		  amdgpu_display_hotplug_work_func);
+
+	drm_kms_helper_poll_init(adev_to_drm(adev));
 
 	adev->mode_info.mode_config_initialized = true;
 	return 0;
 }
 
-static int dce_v10_0_sw_fini(void *handle)
+static int dce_v10_0_sw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	kfree(adev->mode_info.bios_hardcoded_edid);
+	drm_edid_free(adev->mode_info.bios_hardcoded_edid);
 
-	drm_kms_helper_poll_fini(adev->ddev);
+	drm_kms_helper_poll_fini(adev_to_drm(adev));
 
 	dce_v10_0_audio_fini(adev);
 
 	dce_v10_0_afmt_fini(adev);
 
-	drm_mode_config_cleanup(adev->ddev);
+	drm_mode_config_cleanup(adev_to_drm(adev));
 	adev->mode_info.mode_config_initialized = false;
 
 	return 0;
 }
 
-static int dce_v10_0_hw_init(void *handle)
+static int dce_v10_0_hw_init(struct amdgpu_ip_block *ip_block)
 {
 	int i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	dce_v10_0_init_golden_registers(adev);
 
@@ -2902,10 +2908,10 @@ static int dce_v10_0_hw_init(void *handle)
 	return 0;
 }
 
-static int dce_v10_0_hw_fini(void *handle)
+static int dce_v10_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
 	int i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	dce_v10_0_hpd_fini(adev);
 
@@ -2915,20 +2921,35 @@ static int dce_v10_0_hw_fini(void *handle)
 
 	dce_v10_0_pageflip_interrupt_fini(adev);
 
+	flush_delayed_work(&adev->hotplug_work);
+
 	return 0;
 }
 
-static int dce_v10_0_suspend(void *handle)
+static int dce_v10_0_suspend(struct amdgpu_ip_block *ip_block)
 {
-	return dce_v10_0_hw_fini(handle);
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = amdgpu_display_suspend_helper(adev);
+	if (r)
+		return r;
+
+	adev->mode_info.bl_level =
+		amdgpu_atombios_encoder_get_backlight_level_from_reg(adev);
+
+	return dce_v10_0_hw_fini(ip_block);
 }
 
-static int dce_v10_0_resume(void *handle)
+static int dce_v10_0_resume(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int ret;
 
-	ret = dce_v10_0_hw_init(handle);
+	amdgpu_atombios_encoder_set_backlight_level_to_reg(adev,
+							   adev->mode_info.bl_level);
+
+	ret = dce_v10_0_hw_init(ip_block);
 
 	/* turn on the BL */
 	if (adev->mode_info.bl_encoder) {
@@ -2937,31 +2958,28 @@ static int dce_v10_0_resume(void *handle)
 		amdgpu_display_backlight_set_level(adev, adev->mode_info.bl_encoder,
 						    bl_level);
 	}
+	if (ret)
+		return ret;
 
-	return ret;
+	return amdgpu_display_resume_helper(adev);
 }
 
-static bool dce_v10_0_is_idle(void *handle)
+static bool dce_v10_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	return true;
 }
 
-static int dce_v10_0_wait_for_idle(void *handle)
+static bool dce_v10_0_check_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	return 0;
-}
-
-static bool dce_v10_0_check_soft_reset(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	return dce_v10_0_is_display_hung(adev);
 }
 
-static int dce_v10_0_soft_reset(void *handle)
+static int dce_v10_0_soft_reset(struct amdgpu_ip_block *ip_block)
 {
 	u32 srbm_soft_reset = 0, tmp;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (dce_v10_0_is_display_hung(adev))
 		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_DC_MASK;
@@ -3051,7 +3069,7 @@ static int dce_v10_0_set_hpd_irq_state(struct amdgpu_device *adev,
 	u32 tmp;
 
 	if (hpd >= adev->mode_info.num_hpd) {
-		DRM_DEBUG("invalid hdp %d\n", hpd);
+		DRM_DEBUG("invalid hpd %d\n", hpd);
 		return 0;
 	}
 
@@ -3170,14 +3188,14 @@ static int dce_v10_0_pageflip_irq(struct amdgpu_device *adev,
 	if (amdgpu_crtc == NULL)
 		return 0;
 
-	spin_lock_irqsave(&adev->ddev->event_lock, flags);
+	spin_lock_irqsave(&adev_to_drm(adev)->event_lock, flags);
 	works = amdgpu_crtc->pflip_works;
 	if (amdgpu_crtc->pflip_status != AMDGPU_FLIP_SUBMITTED) {
 		DRM_DEBUG_DRIVER("amdgpu_crtc->pflip_status = %d != "
 						 "AMDGPU_FLIP_SUBMITTED(%d)\n",
 						 amdgpu_crtc->pflip_status,
 						 AMDGPU_FLIP_SUBMITTED);
-		spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
+		spin_unlock_irqrestore(&adev_to_drm(adev)->event_lock, flags);
 		return 0;
 	}
 
@@ -3189,7 +3207,7 @@ static int dce_v10_0_pageflip_irq(struct amdgpu_device *adev,
 	if (works->event)
 		drm_crtc_send_vblank_event(&amdgpu_crtc->base, works->event);
 
-	spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
+	spin_unlock_irqrestore(&adev_to_drm(adev)->event_lock, flags);
 
 	drm_crtc_vblank_put(&amdgpu_crtc->base);
 	schedule_work(&works->unpin_work);
@@ -3203,7 +3221,7 @@ static void dce_v10_0_hpd_int_ack(struct amdgpu_device *adev,
 	u32 tmp;
 
 	if (hpd >= adev->mode_info.num_hpd) {
-		DRM_DEBUG("invalid hdp %d\n", hpd);
+		DRM_DEBUG("invalid hpd %d\n", hpd);
 		return;
 	}
 
@@ -3248,7 +3266,7 @@ static int dce_v10_0_crtc_irq(struct amdgpu_device *adev,
 {
 	unsigned crtc = entry->src_id - 1;
 	uint32_t disp_int = RREG32(interrupt_status_offsets[crtc].reg);
-	unsigned irq_type = amdgpu_crtc_idx_to_irq_type(adev, crtc);
+	unsigned int irq_type = amdgpu_display_crtc_idx_to_irq_type(adev, crtc);
 
 	switch (entry->src_data[0]) {
 	case 0: /* vblank */
@@ -3258,7 +3276,7 @@ static int dce_v10_0_crtc_irq(struct amdgpu_device *adev,
 			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
 
 		if (amdgpu_irq_enabled(adev, source, irq_type)) {
-			drm_handle_vblank(adev->ddev, crtc);
+			drm_handle_vblank(adev_to_drm(adev), crtc);
 		}
 		DRM_DEBUG("IH: D%d vblank\n", crtc + 1);
 
@@ -3298,20 +3316,20 @@ static int dce_v10_0_hpd_irq(struct amdgpu_device *adev,
 
 	if (disp_int & mask) {
 		dce_v10_0_hpd_int_ack(adev, hpd);
-		schedule_work(&adev->hotplug_work);
+		schedule_delayed_work(&adev->hotplug_work, 0);
 		DRM_DEBUG("IH: HPD%d\n", hpd + 1);
 	}
 
 	return 0;
 }
 
-static int dce_v10_0_set_clockgating_state(void *handle,
+static int dce_v10_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
 	return 0;
 }
 
-static int dce_v10_0_set_powergating_state(void *handle,
+static int dce_v10_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	return 0;
@@ -3320,7 +3338,6 @@ static int dce_v10_0_set_powergating_state(void *handle,
 static const struct amd_ip_funcs dce_v10_0_ip_funcs = {
 	.name = "dce_v10_0",
 	.early_init = dce_v10_0_early_init,
-	.late_init = NULL,
 	.sw_init = dce_v10_0_sw_init,
 	.sw_fini = dce_v10_0_sw_fini,
 	.hw_init = dce_v10_0_hw_init,
@@ -3328,7 +3345,6 @@ static const struct amd_ip_funcs dce_v10_0_ip_funcs = {
 	.suspend = dce_v10_0_suspend,
 	.resume = dce_v10_0_resume,
 	.is_idle = dce_v10_0_is_idle,
-	.wait_for_idle = dce_v10_0_wait_for_idle,
 	.check_soft_reset = dce_v10_0_check_soft_reset,
 	.soft_reset = dce_v10_0_soft_reset,
 	.set_clockgating_state = dce_v10_0_set_clockgating_state,
@@ -3358,7 +3374,7 @@ dce_v10_0_encoder_mode_set(struct drm_encoder *encoder,
 
 static void dce_v10_0_encoder_prepare(struct drm_encoder *encoder)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(encoder->dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
 
@@ -3398,7 +3414,7 @@ static void dce_v10_0_encoder_prepare(struct drm_encoder *encoder)
 static void dce_v10_0_encoder_commit(struct drm_encoder *encoder)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 
 	/* need to call this here as we need the crtc set up */
 	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_ON);
@@ -3498,7 +3514,7 @@ static void dce_v10_0_encoder_add(struct amdgpu_device *adev,
 				 uint32_t supported_device,
 				 u16 caps)
 {
-	struct drm_device *dev = adev->ddev;
+	struct drm_device *dev = adev_to_drm(adev);
 	struct drm_encoder *encoder;
 	struct amdgpu_encoder *amdgpu_encoder;
 
@@ -3600,7 +3616,6 @@ static void dce_v10_0_encoder_add(struct amdgpu_device *adev,
 static const struct amdgpu_display_funcs dce_v10_0_display_funcs = {
 	.bandwidth_update = &dce_v10_0_bandwidth_update,
 	.vblank_get_counter = &dce_v10_0_vblank_get_counter,
-	.vblank_wait = &dce_v10_0_vblank_wait,
 	.backlight_set_level = &amdgpu_atombios_encoder_set_backlight_level,
 	.backlight_get_level = &amdgpu_atombios_encoder_get_backlight_level,
 	.hpd_sense = &dce_v10_0_hpd_sense,
@@ -3614,8 +3629,7 @@ static const struct amdgpu_display_funcs dce_v10_0_display_funcs = {
 
 static void dce_v10_0_set_display_funcs(struct amdgpu_device *adev)
 {
-	if (adev->mode_info.funcs == NULL)
-		adev->mode_info.funcs = &dce_v10_0_display_funcs;
+	adev->mode_info.funcs = &dce_v10_0_display_funcs;
 }
 
 static const struct amdgpu_irq_src_funcs dce_v10_0_crtc_irq_funcs = {
@@ -3635,18 +3649,20 @@ static const struct amdgpu_irq_src_funcs dce_v10_0_hpd_irq_funcs = {
 
 static void dce_v10_0_set_irq_funcs(struct amdgpu_device *adev)
 {
-	adev->crtc_irq.num_types = AMDGPU_CRTC_IRQ_LAST;
+	if (adev->mode_info.num_crtc > 0)
+		adev->crtc_irq.num_types = AMDGPU_CRTC_IRQ_VLINE1 + adev->mode_info.num_crtc;
+	else
+		adev->crtc_irq.num_types = 0;
 	adev->crtc_irq.funcs = &dce_v10_0_crtc_irq_funcs;
 
-	adev->pageflip_irq.num_types = AMDGPU_PAGEFLIP_IRQ_LAST;
+	adev->pageflip_irq.num_types = adev->mode_info.num_crtc;
 	adev->pageflip_irq.funcs = &dce_v10_0_pageflip_irq_funcs;
 
-	adev->hpd_irq.num_types = AMDGPU_HPD_LAST;
+	adev->hpd_irq.num_types = adev->mode_info.num_hpd;
 	adev->hpd_irq.funcs = &dce_v10_0_hpd_irq_funcs;
 }
 
-const struct amdgpu_ip_block_version dce_v10_0_ip_block =
-{
+const struct amdgpu_ip_block_version dce_v10_0_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_DCE,
 	.major = 10,
 	.minor = 0,
@@ -3654,8 +3670,7 @@ const struct amdgpu_ip_block_version dce_v10_0_ip_block =
 	.funcs = &dce_v10_0_ip_funcs,
 };
 
-const struct amdgpu_ip_block_version dce_v10_1_ip_block =
-{
+const struct amdgpu_ip_block_version dce_v10_1_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_DCE,
 	.major = 10,
 	.minor = 1,
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
deleted file mode 100644
index 11edc75edaa9..000000000000
--- a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
+++ /dev/null
@@ -1,3788 +0,0 @@
-/*
- * Copyright 2014 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-#include <drm/drmP.h>
-#include "amdgpu.h"
-#include "amdgpu_pm.h"
-#include "amdgpu_i2c.h"
-#include "vid.h"
-#include "atom.h"
-#include "amdgpu_atombios.h"
-#include "atombios_crtc.h"
-#include "atombios_encoders.h"
-#include "amdgpu_pll.h"
-#include "amdgpu_connectors.h"
-#include "dce_v11_0.h"
-
-#include "dce/dce_11_0_d.h"
-#include "dce/dce_11_0_sh_mask.h"
-#include "dce/dce_11_0_enum.h"
-#include "oss/oss_3_0_d.h"
-#include "oss/oss_3_0_sh_mask.h"
-#include "gmc/gmc_8_1_d.h"
-#include "gmc/gmc_8_1_sh_mask.h"
-
-static void dce_v11_0_set_display_funcs(struct amdgpu_device *adev);
-static void dce_v11_0_set_irq_funcs(struct amdgpu_device *adev);
-
-static const u32 crtc_offsets[] =
-{
-	CRTC0_REGISTER_OFFSET,
-	CRTC1_REGISTER_OFFSET,
-	CRTC2_REGISTER_OFFSET,
-	CRTC3_REGISTER_OFFSET,
-	CRTC4_REGISTER_OFFSET,
-	CRTC5_REGISTER_OFFSET,
-	CRTC6_REGISTER_OFFSET
-};
-
-static const u32 hpd_offsets[] =
-{
-	HPD0_REGISTER_OFFSET,
-	HPD1_REGISTER_OFFSET,
-	HPD2_REGISTER_OFFSET,
-	HPD3_REGISTER_OFFSET,
-	HPD4_REGISTER_OFFSET,
-	HPD5_REGISTER_OFFSET
-};
-
-static const uint32_t dig_offsets[] = {
-	DIG0_REGISTER_OFFSET,
-	DIG1_REGISTER_OFFSET,
-	DIG2_REGISTER_OFFSET,
-	DIG3_REGISTER_OFFSET,
-	DIG4_REGISTER_OFFSET,
-	DIG5_REGISTER_OFFSET,
-	DIG6_REGISTER_OFFSET,
-	DIG7_REGISTER_OFFSET,
-	DIG8_REGISTER_OFFSET
-};
-
-static const struct {
-	uint32_t        reg;
-	uint32_t        vblank;
-	uint32_t        vline;
-	uint32_t        hpd;
-
-} interrupt_status_offsets[] = { {
-	.reg = mmDISP_INTERRUPT_STATUS,
-	.vblank = DISP_INTERRUPT_STATUS__LB_D1_VBLANK_INTERRUPT_MASK,
-	.vline = DISP_INTERRUPT_STATUS__LB_D1_VLINE_INTERRUPT_MASK,
-	.hpd = DISP_INTERRUPT_STATUS__DC_HPD1_INTERRUPT_MASK
-}, {
-	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE,
-	.vblank = DISP_INTERRUPT_STATUS_CONTINUE__LB_D2_VBLANK_INTERRUPT_MASK,
-	.vline = DISP_INTERRUPT_STATUS_CONTINUE__LB_D2_VLINE_INTERRUPT_MASK,
-	.hpd = DISP_INTERRUPT_STATUS_CONTINUE__DC_HPD2_INTERRUPT_MASK
-}, {
-	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE2,
-	.vblank = DISP_INTERRUPT_STATUS_CONTINUE2__LB_D3_VBLANK_INTERRUPT_MASK,
-	.vline = DISP_INTERRUPT_STATUS_CONTINUE2__LB_D3_VLINE_INTERRUPT_MASK,
-	.hpd = DISP_INTERRUPT_STATUS_CONTINUE2__DC_HPD3_INTERRUPT_MASK
-}, {
-	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE3,
-	.vblank = DISP_INTERRUPT_STATUS_CONTINUE3__LB_D4_VBLANK_INTERRUPT_MASK,
-	.vline = DISP_INTERRUPT_STATUS_CONTINUE3__LB_D4_VLINE_INTERRUPT_MASK,
-	.hpd = DISP_INTERRUPT_STATUS_CONTINUE3__DC_HPD4_INTERRUPT_MASK
-}, {
-	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE4,
-	.vblank = DISP_INTERRUPT_STATUS_CONTINUE4__LB_D5_VBLANK_INTERRUPT_MASK,
-	.vline = DISP_INTERRUPT_STATUS_CONTINUE4__LB_D5_VLINE_INTERRUPT_MASK,
-	.hpd = DISP_INTERRUPT_STATUS_CONTINUE4__DC_HPD5_INTERRUPT_MASK
-}, {
-	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE5,
-	.vblank = DISP_INTERRUPT_STATUS_CONTINUE5__LB_D6_VBLANK_INTERRUPT_MASK,
-	.vline = DISP_INTERRUPT_STATUS_CONTINUE5__LB_D6_VLINE_INTERRUPT_MASK,
-	.hpd = DISP_INTERRUPT_STATUS_CONTINUE5__DC_HPD6_INTERRUPT_MASK
-} };
-
-static const u32 cz_golden_settings_a11[] =
-{
-	mmCRTC_DOUBLE_BUFFER_CONTROL, 0x00010101, 0x00010000,
-	mmFBC_MISC, 0x1f311fff, 0x14300000,
-};
-
-static const u32 cz_mgcg_cgcg_init[] =
-{
-	mmXDMA_CLOCK_GATING_CNTL, 0xffffffff, 0x00000100,
-	mmXDMA_MEM_POWER_CNTL, 0x00000101, 0x00000000,
-};
-
-static const u32 stoney_golden_settings_a11[] =
-{
-	mmCRTC_DOUBLE_BUFFER_CONTROL, 0x00010101, 0x00010000,
-	mmFBC_MISC, 0x1f311fff, 0x14302000,
-};
-
-static const u32 polaris11_golden_settings_a11[] =
-{
-	mmDCI_CLK_CNTL, 0x00000080, 0x00000000,
-	mmFBC_DEBUG_COMP, 0x000000f0, 0x00000070,
-	mmFBC_DEBUG1, 0xffffffff, 0x00000008,
-	mmFBC_MISC, 0x9f313fff, 0x14302008,
-	mmHDMI_CONTROL, 0x313f031f, 0x00000011,
-};
-
-static const u32 polaris10_golden_settings_a11[] =
-{
-	mmDCI_CLK_CNTL, 0x00000080, 0x00000000,
-	mmFBC_DEBUG_COMP, 0x000000f0, 0x00000070,
-	mmFBC_MISC, 0x9f313fff, 0x14302008,
-	mmHDMI_CONTROL, 0x313f031f, 0x00000011,
-};
-
-static void dce_v11_0_init_golden_registers(struct amdgpu_device *adev)
-{
-	switch (adev->asic_type) {
-	case CHIP_CARRIZO:
-		amdgpu_program_register_sequence(adev,
-						 cz_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(cz_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 cz_golden_settings_a11,
-						 (const u32)ARRAY_SIZE(cz_golden_settings_a11));
-		break;
-	case CHIP_STONEY:
-		amdgpu_program_register_sequence(adev,
-						 stoney_golden_settings_a11,
-						 (const u32)ARRAY_SIZE(stoney_golden_settings_a11));
-		break;
-	case CHIP_POLARIS11:
-	case CHIP_POLARIS12:
-		amdgpu_program_register_sequence(adev,
-						 polaris11_golden_settings_a11,
-						 (const u32)ARRAY_SIZE(polaris11_golden_settings_a11));
-		break;
-	case CHIP_POLARIS10:
-		amdgpu_program_register_sequence(adev,
-						 polaris10_golden_settings_a11,
-						 (const u32)ARRAY_SIZE(polaris10_golden_settings_a11));
-		break;
-	default:
-		break;
-	}
-}
-
-static u32 dce_v11_0_audio_endpt_rreg(struct amdgpu_device *adev,
-				     u32 block_offset, u32 reg)
-{
-	unsigned long flags;
-	u32 r;
-
-	spin_lock_irqsave(&adev->audio_endpt_idx_lock, flags);
-	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_INDEX + block_offset, reg);
-	r = RREG32(mmAZALIA_F0_CODEC_ENDPOINT_DATA + block_offset);
-	spin_unlock_irqrestore(&adev->audio_endpt_idx_lock, flags);
-
-	return r;
-}
-
-static void dce_v11_0_audio_endpt_wreg(struct amdgpu_device *adev,
-				      u32 block_offset, u32 reg, u32 v)
-{
-	unsigned long flags;
-
-	spin_lock_irqsave(&adev->audio_endpt_idx_lock, flags);
-	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_INDEX + block_offset, reg);
-	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_DATA + block_offset, v);
-	spin_unlock_irqrestore(&adev->audio_endpt_idx_lock, flags);
-}
-
-static bool dce_v11_0_is_in_vblank(struct amdgpu_device *adev, int crtc)
-{
-	if (RREG32(mmCRTC_STATUS + crtc_offsets[crtc]) &
-			CRTC_V_BLANK_START_END__CRTC_V_BLANK_START_MASK)
-		return true;
-	else
-		return false;
-}
-
-static bool dce_v11_0_is_counter_moving(struct amdgpu_device *adev, int crtc)
-{
-	u32 pos1, pos2;
-
-	pos1 = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
-	pos2 = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
-
-	if (pos1 != pos2)
-		return true;
-	else
-		return false;
-}
-
-/**
- * dce_v11_0_vblank_wait - vblank wait asic callback.
- *
- * @adev: amdgpu_device pointer
- * @crtc: crtc to wait for vblank on
- *
- * Wait for vblank on the requested crtc (evergreen+).
- */
-static void dce_v11_0_vblank_wait(struct amdgpu_device *adev, int crtc)
-{
-	unsigned i = 100;
-
-	if (crtc < 0 || crtc >= adev->mode_info.num_crtc)
-		return;
-
-	if (!(RREG32(mmCRTC_CONTROL + crtc_offsets[crtc]) & CRTC_CONTROL__CRTC_MASTER_EN_MASK))
-		return;
-
-	/* depending on when we hit vblank, we may be close to active; if so,
-	 * wait for another frame.
-	 */
-	while (dce_v11_0_is_in_vblank(adev, crtc)) {
-		if (i++ == 100) {
-			i = 0;
-			if (!dce_v11_0_is_counter_moving(adev, crtc))
-				break;
-		}
-	}
-
-	while (!dce_v11_0_is_in_vblank(adev, crtc)) {
-		if (i++ == 100) {
-			i = 0;
-			if (!dce_v11_0_is_counter_moving(adev, crtc))
-				break;
-		}
-	}
-}
-
-static u32 dce_v11_0_vblank_get_counter(struct amdgpu_device *adev, int crtc)
-{
-	if (crtc < 0 || crtc >= adev->mode_info.num_crtc)
-		return 0;
-	else
-		return RREG32(mmCRTC_STATUS_FRAME_COUNT + crtc_offsets[crtc]);
-}
-
-static void dce_v11_0_pageflip_interrupt_init(struct amdgpu_device *adev)
-{
-	unsigned i;
-
-	/* Enable pflip interrupts */
-	for (i = 0; i < adev->mode_info.num_crtc; i++)
-		amdgpu_irq_get(adev, &adev->pageflip_irq, i);
-}
-
-static void dce_v11_0_pageflip_interrupt_fini(struct amdgpu_device *adev)
-{
-	unsigned i;
-
-	/* Disable pflip interrupts */
-	for (i = 0; i < adev->mode_info.num_crtc; i++)
-		amdgpu_irq_put(adev, &adev->pageflip_irq, i);
-}
-
-/**
- * dce_v11_0_page_flip - pageflip callback.
- *
- * @adev: amdgpu_device pointer
- * @crtc_id: crtc to cleanup pageflip on
- * @crtc_base: new address of the crtc (GPU MC address)
- *
- * Triggers the actual pageflip by updating the primary
- * surface base address.
- */
-static void dce_v11_0_page_flip(struct amdgpu_device *adev,
-				int crtc_id, u64 crtc_base, bool async)
-{
-	struct amdgpu_crtc *amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
-	u32 tmp;
-
-	/* flip immediate for async, default is vsync */
-	tmp = RREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, GRPH_FLIP_CONTROL,
-			    GRPH_SURFACE_UPDATE_IMMEDIATE_EN, async ? 1 : 0);
-	WREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-	/* update the scanout addresses */
-	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
-	       upper_32_bits(crtc_base));
-	/* writing to the low address triggers the update */
-	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
-	       lower_32_bits(crtc_base));
-	/* post the write */
-	RREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset);
-}
-
-static int dce_v11_0_crtc_get_scanoutpos(struct amdgpu_device *adev, int crtc,
-					u32 *vbl, u32 *position)
-{
-	if ((crtc < 0) || (crtc >= adev->mode_info.num_crtc))
-		return -EINVAL;
-
-	*vbl = RREG32(mmCRTC_V_BLANK_START_END + crtc_offsets[crtc]);
-	*position = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
-
-	return 0;
-}
-
-/**
- * dce_v11_0_hpd_sense - hpd sense callback.
- *
- * @adev: amdgpu_device pointer
- * @hpd: hpd (hotplug detect) pin
- *
- * Checks if a digital monitor is connected (evergreen+).
- * Returns true if connected, false if not connected.
- */
-static bool dce_v11_0_hpd_sense(struct amdgpu_device *adev,
-			       enum amdgpu_hpd_id hpd)
-{
-	bool connected = false;
-
-	if (hpd >= adev->mode_info.num_hpd)
-		return connected;
-
-	if (RREG32(mmDC_HPD_INT_STATUS + hpd_offsets[hpd]) &
-	    DC_HPD_INT_STATUS__DC_HPD_SENSE_MASK)
-		connected = true;
-
-	return connected;
-}
-
-/**
- * dce_v11_0_hpd_set_polarity - hpd set polarity callback.
- *
- * @adev: amdgpu_device pointer
- * @hpd: hpd (hotplug detect) pin
- *
- * Set the polarity of the hpd pin (evergreen+).
- */
-static void dce_v11_0_hpd_set_polarity(struct amdgpu_device *adev,
-				      enum amdgpu_hpd_id hpd)
-{
-	u32 tmp;
-	bool connected = dce_v11_0_hpd_sense(adev, hpd);
-
-	if (hpd >= adev->mode_info.num_hpd)
-		return;
-
-	tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd]);
-	if (connected)
-		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_POLARITY, 0);
-	else
-		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_POLARITY, 1);
-	WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd], tmp);
-}
-
-/**
- * dce_v11_0_hpd_init - hpd setup callback.
- *
- * @adev: amdgpu_device pointer
- *
- * Setup the hpd pins used by the card (evergreen+).
- * Enable the pin, set the polarity, and enable the hpd interrupts.
- */
-static void dce_v11_0_hpd_init(struct amdgpu_device *adev)
-{
-	struct drm_device *dev = adev->ddev;
-	struct drm_connector *connector;
-	u32 tmp;
-
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
-		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
-
-		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
-			continue;
-
-		if (connector->connector_type == DRM_MODE_CONNECTOR_eDP ||
-		    connector->connector_type == DRM_MODE_CONNECTOR_LVDS) {
-			/* don't try to enable hpd on eDP or LVDS avoid breaking the
-			 * aux dp channel on imac and help (but not completely fix)
-			 * https://bugzilla.redhat.com/show_bug.cgi?id=726143
-			 * also avoid interrupt storms during dpms.
-			 */
-			tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd]);
-			tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_EN, 0);
-			WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
-			continue;
-		}
-
-		tmp = RREG32(mmDC_HPD_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd]);
-		tmp = REG_SET_FIELD(tmp, DC_HPD_CONTROL, DC_HPD_EN, 1);
-		WREG32(mmDC_HPD_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
-
-		tmp = RREG32(mmDC_HPD_TOGGLE_FILT_CNTL + hpd_offsets[amdgpu_connector->hpd.hpd]);
-		tmp = REG_SET_FIELD(tmp, DC_HPD_TOGGLE_FILT_CNTL,
-				    DC_HPD_CONNECT_INT_DELAY,
-				    AMDGPU_HPD_CONNECT_INT_DELAY_IN_MS);
-		tmp = REG_SET_FIELD(tmp, DC_HPD_TOGGLE_FILT_CNTL,
-				    DC_HPD_DISCONNECT_INT_DELAY,
-				    AMDGPU_HPD_DISCONNECT_INT_DELAY_IN_MS);
-		WREG32(mmDC_HPD_TOGGLE_FILT_CNTL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
-
-		dce_v11_0_hpd_set_polarity(adev, amdgpu_connector->hpd.hpd);
-		amdgpu_irq_get(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
-	}
-}
-
-/**
- * dce_v11_0_hpd_fini - hpd tear down callback.
- *
- * @adev: amdgpu_device pointer
- *
- * Tear down the hpd pins used by the card (evergreen+).
- * Disable the hpd interrupts.
- */
-static void dce_v11_0_hpd_fini(struct amdgpu_device *adev)
-{
-	struct drm_device *dev = adev->ddev;
-	struct drm_connector *connector;
-	u32 tmp;
-
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
-		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
-
-		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
-			continue;
-
-		tmp = RREG32(mmDC_HPD_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd]);
-		tmp = REG_SET_FIELD(tmp, DC_HPD_CONTROL, DC_HPD_EN, 0);
-		WREG32(mmDC_HPD_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
-
-		amdgpu_irq_put(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
-	}
-}
-
-static u32 dce_v11_0_hpd_get_gpio_reg(struct amdgpu_device *adev)
-{
-	return mmDC_GPIO_HPD_A;
-}
-
-static bool dce_v11_0_is_display_hung(struct amdgpu_device *adev)
-{
-	u32 crtc_hung = 0;
-	u32 crtc_status[6];
-	u32 i, j, tmp;
-
-	for (i = 0; i < adev->mode_info.num_crtc; i++) {
-		tmp = RREG32(mmCRTC_CONTROL + crtc_offsets[i]);
-		if (REG_GET_FIELD(tmp, CRTC_CONTROL, CRTC_MASTER_EN)) {
-			crtc_status[i] = RREG32(mmCRTC_STATUS_HV_COUNT + crtc_offsets[i]);
-			crtc_hung |= (1 << i);
-		}
-	}
-
-	for (j = 0; j < 10; j++) {
-		for (i = 0; i < adev->mode_info.num_crtc; i++) {
-			if (crtc_hung & (1 << i)) {
-				tmp = RREG32(mmCRTC_STATUS_HV_COUNT + crtc_offsets[i]);
-				if (tmp != crtc_status[i])
-					crtc_hung &= ~(1 << i);
-			}
-		}
-		if (crtc_hung == 0)
-			return false;
-		udelay(100);
-	}
-
-	return true;
-}
-
-static void dce_v11_0_set_vga_render_state(struct amdgpu_device *adev,
-					   bool render)
-{
-	u32 tmp;
-
-	/* Lockout access through VGA aperture*/
-	tmp = RREG32(mmVGA_HDP_CONTROL);
-	if (render)
-		tmp = REG_SET_FIELD(tmp, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 0);
-	else
-		tmp = REG_SET_FIELD(tmp, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 1);
-	WREG32(mmVGA_HDP_CONTROL, tmp);
-
-	/* disable VGA render */
-	tmp = RREG32(mmVGA_RENDER_CONTROL);
-	if (render)
-		tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 1);
-	else
-		tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 0);
-	WREG32(mmVGA_RENDER_CONTROL, tmp);
-}
-
-static int dce_v11_0_get_num_crtc (struct amdgpu_device *adev)
-{
-	int num_crtc = 0;
-
-	switch (adev->asic_type) {
-	case CHIP_CARRIZO:
-		num_crtc = 3;
-		break;
-	case CHIP_STONEY:
-		num_crtc = 2;
-		break;
-	case CHIP_POLARIS10:
-		num_crtc = 6;
-		break;
-	case CHIP_POLARIS11:
-	case CHIP_POLARIS12:
-		num_crtc = 5;
-		break;
-	default:
-		num_crtc = 0;
-	}
-	return num_crtc;
-}
-
-void dce_v11_0_disable_dce(struct amdgpu_device *adev)
-{
-	/*Disable VGA render and enabled crtc, if has DCE engine*/
-	if (amdgpu_atombios_has_dce_engine_info(adev)) {
-		u32 tmp;
-		int crtc_enabled, i;
-
-		dce_v11_0_set_vga_render_state(adev, false);
-
-		/*Disable crtc*/
-		for (i = 0; i < dce_v11_0_get_num_crtc(adev); i++) {
-			crtc_enabled = REG_GET_FIELD(RREG32(mmCRTC_CONTROL + crtc_offsets[i]),
-									 CRTC_CONTROL, CRTC_MASTER_EN);
-			if (crtc_enabled) {
-				WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
-				tmp = RREG32(mmCRTC_CONTROL + crtc_offsets[i]);
-				tmp = REG_SET_FIELD(tmp, CRTC_CONTROL, CRTC_MASTER_EN, 0);
-				WREG32(mmCRTC_CONTROL + crtc_offsets[i], tmp);
-				WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
-			}
-		}
-	}
-}
-
-static void dce_v11_0_program_fmt(struct drm_encoder *encoder)
-{
-	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
-	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
-	int bpc = 0;
-	u32 tmp = 0;
-	enum amdgpu_connector_dither dither = AMDGPU_FMT_DITHER_DISABLE;
-
-	if (connector) {
-		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
-		bpc = amdgpu_connector_get_monitor_bpc(connector);
-		dither = amdgpu_connector->dither;
-	}
-
-	/* LVDS/eDP FMT is set up by atom */
-	if (amdgpu_encoder->devices & ATOM_DEVICE_LCD_SUPPORT)
-		return;
-
-	/* not needed for analog */
-	if ((amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1) ||
-	    (amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2))
-		return;
-
-	if (bpc == 0)
-		return;
-
-	switch (bpc) {
-	case 6:
-		if (dither == AMDGPU_FMT_DITHER_ENABLE) {
-			/* XXX sort out optimal dither settings */
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_FRAME_RANDOM_ENABLE, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_HIGHPASS_RANDOM_ENABLE, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_EN, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_DEPTH, 0);
-		} else {
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_DEPTH, 0);
-		}
-		break;
-	case 8:
-		if (dither == AMDGPU_FMT_DITHER_ENABLE) {
-			/* XXX sort out optimal dither settings */
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_FRAME_RANDOM_ENABLE, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_HIGHPASS_RANDOM_ENABLE, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_RGB_RANDOM_ENABLE, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_EN, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_DEPTH, 1);
-		} else {
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_DEPTH, 1);
-		}
-		break;
-	case 10:
-		if (dither == AMDGPU_FMT_DITHER_ENABLE) {
-			/* XXX sort out optimal dither settings */
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_FRAME_RANDOM_ENABLE, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_HIGHPASS_RANDOM_ENABLE, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_RGB_RANDOM_ENABLE, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_EN, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_DEPTH, 2);
-		} else {
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_DEPTH, 2);
-		}
-		break;
-	default:
-		/* not needed */
-		break;
-	}
-
-	WREG32(mmFMT_BIT_DEPTH_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-}
-
-
-/* display watermark setup */
-/**
- * dce_v11_0_line_buffer_adjust - Set up the line buffer
- *
- * @adev: amdgpu_device pointer
- * @amdgpu_crtc: the selected display controller
- * @mode: the current display mode on the selected display
- * controller
- *
- * Setup up the line buffer allocation for
- * the selected display controller (CIK).
- * Returns the line buffer size in pixels.
- */
-static u32 dce_v11_0_line_buffer_adjust(struct amdgpu_device *adev,
-				       struct amdgpu_crtc *amdgpu_crtc,
-				       struct drm_display_mode *mode)
-{
-	u32 tmp, buffer_alloc, i, mem_cfg;
-	u32 pipe_offset = amdgpu_crtc->crtc_id;
-	/*
-	 * Line Buffer Setup
-	 * There are 6 line buffers, one for each display controllers.
-	 * There are 3 partitions per LB. Select the number of partitions
-	 * to enable based on the display width.  For display widths larger
-	 * than 4096, you need use to use 2 display controllers and combine
-	 * them using the stereo blender.
-	 */
-	if (amdgpu_crtc->base.enabled && mode) {
-		if (mode->crtc_hdisplay < 1920) {
-			mem_cfg = 1;
-			buffer_alloc = 2;
-		} else if (mode->crtc_hdisplay < 2560) {
-			mem_cfg = 2;
-			buffer_alloc = 2;
-		} else if (mode->crtc_hdisplay < 4096) {
-			mem_cfg = 0;
-			buffer_alloc = (adev->flags & AMD_IS_APU) ? 2 : 4;
-		} else {
-			DRM_DEBUG_KMS("Mode too big for LB!\n");
-			mem_cfg = 0;
-			buffer_alloc = (adev->flags & AMD_IS_APU) ? 2 : 4;
-		}
-	} else {
-		mem_cfg = 1;
-		buffer_alloc = 0;
-	}
-
-	tmp = RREG32(mmLB_MEMORY_CTRL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, LB_MEMORY_CTRL, LB_MEMORY_CONFIG, mem_cfg);
-	WREG32(mmLB_MEMORY_CTRL + amdgpu_crtc->crtc_offset, tmp);
-
-	tmp = RREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset);
-	tmp = REG_SET_FIELD(tmp, PIPE0_DMIF_BUFFER_CONTROL, DMIF_BUFFERS_ALLOCATED, buffer_alloc);
-	WREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset, tmp);
-
-	for (i = 0; i < adev->usec_timeout; i++) {
-		tmp = RREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset);
-		if (REG_GET_FIELD(tmp, PIPE0_DMIF_BUFFER_CONTROL, DMIF_BUFFERS_ALLOCATION_COMPLETED))
-			break;
-		udelay(1);
-	}
-
-	if (amdgpu_crtc->base.enabled && mode) {
-		switch (mem_cfg) {
-		case 0:
-		default:
-			return 4096 * 2;
-		case 1:
-			return 1920 * 2;
-		case 2:
-			return 2560 * 2;
-		}
-	}
-
-	/* controller not enabled, so no lb used */
-	return 0;
-}
-
-/**
- * cik_get_number_of_dram_channels - get the number of dram channels
- *
- * @adev: amdgpu_device pointer
- *
- * Look up the number of video ram channels (CIK).
- * Used for display watermark bandwidth calculations
- * Returns the number of dram channels
- */
-static u32 cik_get_number_of_dram_channels(struct amdgpu_device *adev)
-{
-	u32 tmp = RREG32(mmMC_SHARED_CHMAP);
-
-	switch (REG_GET_FIELD(tmp, MC_SHARED_CHMAP, NOOFCHAN)) {
-	case 0:
-	default:
-		return 1;
-	case 1:
-		return 2;
-	case 2:
-		return 4;
-	case 3:
-		return 8;
-	case 4:
-		return 3;
-	case 5:
-		return 6;
-	case 6:
-		return 10;
-	case 7:
-		return 12;
-	case 8:
-		return 16;
-	}
-}
-
-struct dce10_wm_params {
-	u32 dram_channels; /* number of dram channels */
-	u32 yclk;          /* bandwidth per dram data pin in kHz */
-	u32 sclk;          /* engine clock in kHz */
-	u32 disp_clk;      /* display clock in kHz */
-	u32 src_width;     /* viewport width */
-	u32 active_time;   /* active display time in ns */
-	u32 blank_time;    /* blank time in ns */
-	bool interlaced;    /* mode is interlaced */
-	fixed20_12 vsc;    /* vertical scale ratio */
-	u32 num_heads;     /* number of active crtcs */
-	u32 bytes_per_pixel; /* bytes per pixel display + overlay */
-	u32 lb_size;       /* line buffer allocated to pipe */
-	u32 vtaps;         /* vertical scaler taps */
-};
-
-/**
- * dce_v11_0_dram_bandwidth - get the dram bandwidth
- *
- * @wm: watermark calculation data
- *
- * Calculate the raw dram bandwidth (CIK).
- * Used for display watermark bandwidth calculations
- * Returns the dram bandwidth in MBytes/s
- */
-static u32 dce_v11_0_dram_bandwidth(struct dce10_wm_params *wm)
-{
-	/* Calculate raw DRAM Bandwidth */
-	fixed20_12 dram_efficiency; /* 0.7 */
-	fixed20_12 yclk, dram_channels, bandwidth;
-	fixed20_12 a;
-
-	a.full = dfixed_const(1000);
-	yclk.full = dfixed_const(wm->yclk);
-	yclk.full = dfixed_div(yclk, a);
-	dram_channels.full = dfixed_const(wm->dram_channels * 4);
-	a.full = dfixed_const(10);
-	dram_efficiency.full = dfixed_const(7);
-	dram_efficiency.full = dfixed_div(dram_efficiency, a);
-	bandwidth.full = dfixed_mul(dram_channels, yclk);
-	bandwidth.full = dfixed_mul(bandwidth, dram_efficiency);
-
-	return dfixed_trunc(bandwidth);
-}
-
-/**
- * dce_v11_0_dram_bandwidth_for_display - get the dram bandwidth for display
- *
- * @wm: watermark calculation data
- *
- * Calculate the dram bandwidth used for display (CIK).
- * Used for display watermark bandwidth calculations
- * Returns the dram bandwidth for display in MBytes/s
- */
-static u32 dce_v11_0_dram_bandwidth_for_display(struct dce10_wm_params *wm)
-{
-	/* Calculate DRAM Bandwidth and the part allocated to display. */
-	fixed20_12 disp_dram_allocation; /* 0.3 to 0.7 */
-	fixed20_12 yclk, dram_channels, bandwidth;
-	fixed20_12 a;
-
-	a.full = dfixed_const(1000);
-	yclk.full = dfixed_const(wm->yclk);
-	yclk.full = dfixed_div(yclk, a);
-	dram_channels.full = dfixed_const(wm->dram_channels * 4);
-	a.full = dfixed_const(10);
-	disp_dram_allocation.full = dfixed_const(3); /* XXX worse case value 0.3 */
-	disp_dram_allocation.full = dfixed_div(disp_dram_allocation, a);
-	bandwidth.full = dfixed_mul(dram_channels, yclk);
-	bandwidth.full = dfixed_mul(bandwidth, disp_dram_allocation);
-
-	return dfixed_trunc(bandwidth);
-}
-
-/**
- * dce_v11_0_data_return_bandwidth - get the data return bandwidth
- *
- * @wm: watermark calculation data
- *
- * Calculate the data return bandwidth used for display (CIK).
- * Used for display watermark bandwidth calculations
- * Returns the data return bandwidth in MBytes/s
- */
-static u32 dce_v11_0_data_return_bandwidth(struct dce10_wm_params *wm)
-{
-	/* Calculate the display Data return Bandwidth */
-	fixed20_12 return_efficiency; /* 0.8 */
-	fixed20_12 sclk, bandwidth;
-	fixed20_12 a;
-
-	a.full = dfixed_const(1000);
-	sclk.full = dfixed_const(wm->sclk);
-	sclk.full = dfixed_div(sclk, a);
-	a.full = dfixed_const(10);
-	return_efficiency.full = dfixed_const(8);
-	return_efficiency.full = dfixed_div(return_efficiency, a);
-	a.full = dfixed_const(32);
-	bandwidth.full = dfixed_mul(a, sclk);
-	bandwidth.full = dfixed_mul(bandwidth, return_efficiency);
-
-	return dfixed_trunc(bandwidth);
-}
-
-/**
- * dce_v11_0_dmif_request_bandwidth - get the dmif bandwidth
- *
- * @wm: watermark calculation data
- *
- * Calculate the dmif bandwidth used for display (CIK).
- * Used for display watermark bandwidth calculations
- * Returns the dmif bandwidth in MBytes/s
- */
-static u32 dce_v11_0_dmif_request_bandwidth(struct dce10_wm_params *wm)
-{
-	/* Calculate the DMIF Request Bandwidth */
-	fixed20_12 disp_clk_request_efficiency; /* 0.8 */
-	fixed20_12 disp_clk, bandwidth;
-	fixed20_12 a, b;
-
-	a.full = dfixed_const(1000);
-	disp_clk.full = dfixed_const(wm->disp_clk);
-	disp_clk.full = dfixed_div(disp_clk, a);
-	a.full = dfixed_const(32);
-	b.full = dfixed_mul(a, disp_clk);
-
-	a.full = dfixed_const(10);
-	disp_clk_request_efficiency.full = dfixed_const(8);
-	disp_clk_request_efficiency.full = dfixed_div(disp_clk_request_efficiency, a);
-
-	bandwidth.full = dfixed_mul(b, disp_clk_request_efficiency);
-
-	return dfixed_trunc(bandwidth);
-}
-
-/**
- * dce_v11_0_available_bandwidth - get the min available bandwidth
- *
- * @wm: watermark calculation data
- *
- * Calculate the min available bandwidth used for display (CIK).
- * Used for display watermark bandwidth calculations
- * Returns the min available bandwidth in MBytes/s
- */
-static u32 dce_v11_0_available_bandwidth(struct dce10_wm_params *wm)
-{
-	/* Calculate the Available bandwidth. Display can use this temporarily but not in average. */
-	u32 dram_bandwidth = dce_v11_0_dram_bandwidth(wm);
-	u32 data_return_bandwidth = dce_v11_0_data_return_bandwidth(wm);
-	u32 dmif_req_bandwidth = dce_v11_0_dmif_request_bandwidth(wm);
-
-	return min(dram_bandwidth, min(data_return_bandwidth, dmif_req_bandwidth));
-}
-
-/**
- * dce_v11_0_average_bandwidth - get the average available bandwidth
- *
- * @wm: watermark calculation data
- *
- * Calculate the average available bandwidth used for display (CIK).
- * Used for display watermark bandwidth calculations
- * Returns the average available bandwidth in MBytes/s
- */
-static u32 dce_v11_0_average_bandwidth(struct dce10_wm_params *wm)
-{
-	/* Calculate the display mode Average Bandwidth
-	 * DisplayMode should contain the source and destination dimensions,
-	 * timing, etc.
-	 */
-	fixed20_12 bpp;
-	fixed20_12 line_time;
-	fixed20_12 src_width;
-	fixed20_12 bandwidth;
-	fixed20_12 a;
-
-	a.full = dfixed_const(1000);
-	line_time.full = dfixed_const(wm->active_time + wm->blank_time);
-	line_time.full = dfixed_div(line_time, a);
-	bpp.full = dfixed_const(wm->bytes_per_pixel);
-	src_width.full = dfixed_const(wm->src_width);
-	bandwidth.full = dfixed_mul(src_width, bpp);
-	bandwidth.full = dfixed_mul(bandwidth, wm->vsc);
-	bandwidth.full = dfixed_div(bandwidth, line_time);
-
-	return dfixed_trunc(bandwidth);
-}
-
-/**
- * dce_v11_0_latency_watermark - get the latency watermark
- *
- * @wm: watermark calculation data
- *
- * Calculate the latency watermark (CIK).
- * Used for display watermark bandwidth calculations
- * Returns the latency watermark in ns
- */
-static u32 dce_v11_0_latency_watermark(struct dce10_wm_params *wm)
-{
-	/* First calculate the latency in ns */
-	u32 mc_latency = 2000; /* 2000 ns. */
-	u32 available_bandwidth = dce_v11_0_available_bandwidth(wm);
-	u32 worst_chunk_return_time = (512 * 8 * 1000) / available_bandwidth;
-	u32 cursor_line_pair_return_time = (128 * 4 * 1000) / available_bandwidth;
-	u32 dc_latency = 40000000 / wm->disp_clk; /* dc pipe latency */
-	u32 other_heads_data_return_time = ((wm->num_heads + 1) * worst_chunk_return_time) +
-		(wm->num_heads * cursor_line_pair_return_time);
-	u32 latency = mc_latency + other_heads_data_return_time + dc_latency;
-	u32 max_src_lines_per_dst_line, lb_fill_bw, line_fill_time;
-	u32 tmp, dmif_size = 12288;
-	fixed20_12 a, b, c;
-
-	if (wm->num_heads == 0)
-		return 0;
-
-	a.full = dfixed_const(2);
-	b.full = dfixed_const(1);
-	if ((wm->vsc.full > a.full) ||
-	    ((wm->vsc.full > b.full) && (wm->vtaps >= 3)) ||
-	    (wm->vtaps >= 5) ||
-	    ((wm->vsc.full >= a.full) && wm->interlaced))
-		max_src_lines_per_dst_line = 4;
-	else
-		max_src_lines_per_dst_line = 2;
-
-	a.full = dfixed_const(available_bandwidth);
-	b.full = dfixed_const(wm->num_heads);
-	a.full = dfixed_div(a, b);
-	tmp = div_u64((u64) dmif_size * (u64) wm->disp_clk, mc_latency + 512);
-	tmp = min(dfixed_trunc(a), tmp);
-
-	lb_fill_bw = min(tmp, wm->disp_clk * wm->bytes_per_pixel / 1000);
-
-	a.full = dfixed_const(max_src_lines_per_dst_line * wm->src_width * wm->bytes_per_pixel);
-	b.full = dfixed_const(1000);
-	c.full = dfixed_const(lb_fill_bw);
-	b.full = dfixed_div(c, b);
-	a.full = dfixed_div(a, b);
-	line_fill_time = dfixed_trunc(a);
-
-	if (line_fill_time < wm->active_time)
-		return latency;
-	else
-		return latency + (line_fill_time - wm->active_time);
-
-}
-
-/**
- * dce_v11_0_average_bandwidth_vs_dram_bandwidth_for_display - check
- * average and available dram bandwidth
- *
- * @wm: watermark calculation data
- *
- * Check if the display average bandwidth fits in the display
- * dram bandwidth (CIK).
- * Used for display watermark bandwidth calculations
- * Returns true if the display fits, false if not.
- */
-static bool dce_v11_0_average_bandwidth_vs_dram_bandwidth_for_display(struct dce10_wm_params *wm)
-{
-	if (dce_v11_0_average_bandwidth(wm) <=
-	    (dce_v11_0_dram_bandwidth_for_display(wm) / wm->num_heads))
-		return true;
-	else
-		return false;
-}
-
-/**
- * dce_v11_0_average_bandwidth_vs_available_bandwidth - check
- * average and available bandwidth
- *
- * @wm: watermark calculation data
- *
- * Check if the display average bandwidth fits in the display
- * available bandwidth (CIK).
- * Used for display watermark bandwidth calculations
- * Returns true if the display fits, false if not.
- */
-static bool dce_v11_0_average_bandwidth_vs_available_bandwidth(struct dce10_wm_params *wm)
-{
-	if (dce_v11_0_average_bandwidth(wm) <=
-	    (dce_v11_0_available_bandwidth(wm) / wm->num_heads))
-		return true;
-	else
-		return false;
-}
-
-/**
- * dce_v11_0_check_latency_hiding - check latency hiding
- *
- * @wm: watermark calculation data
- *
- * Check latency hiding (CIK).
- * Used for display watermark bandwidth calculations
- * Returns true if the display fits, false if not.
- */
-static bool dce_v11_0_check_latency_hiding(struct dce10_wm_params *wm)
-{
-	u32 lb_partitions = wm->lb_size / wm->src_width;
-	u32 line_time = wm->active_time + wm->blank_time;
-	u32 latency_tolerant_lines;
-	u32 latency_hiding;
-	fixed20_12 a;
-
-	a.full = dfixed_const(1);
-	if (wm->vsc.full > a.full)
-		latency_tolerant_lines = 1;
-	else {
-		if (lb_partitions <= (wm->vtaps + 1))
-			latency_tolerant_lines = 1;
-		else
-			latency_tolerant_lines = 2;
-	}
-
-	latency_hiding = (latency_tolerant_lines * line_time + wm->blank_time);
-
-	if (dce_v11_0_latency_watermark(wm) <= latency_hiding)
-		return true;
-	else
-		return false;
-}
-
-/**
- * dce_v11_0_program_watermarks - program display watermarks
- *
- * @adev: amdgpu_device pointer
- * @amdgpu_crtc: the selected display controller
- * @lb_size: line buffer size
- * @num_heads: number of display controllers in use
- *
- * Calculate and program the display watermarks for the
- * selected display controller (CIK).
- */
-static void dce_v11_0_program_watermarks(struct amdgpu_device *adev,
-					struct amdgpu_crtc *amdgpu_crtc,
-					u32 lb_size, u32 num_heads)
-{
-	struct drm_display_mode *mode = &amdgpu_crtc->base.mode;
-	struct dce10_wm_params wm_low, wm_high;
-	u32 active_time;
-	u32 line_time = 0;
-	u32 latency_watermark_a = 0, latency_watermark_b = 0;
-	u32 tmp, wm_mask, lb_vblank_lead_lines = 0;
-
-	if (amdgpu_crtc->base.enabled && num_heads && mode) {
-		active_time = (u32) div_u64((u64)mode->crtc_hdisplay * 1000000,
-					    (u32)mode->clock);
-		line_time = (u32) div_u64((u64)mode->crtc_htotal * 1000000,
-					  (u32)mode->clock);
-		line_time = min(line_time, (u32)65535);
-
-		/* watermark for high clocks */
-		if (adev->pm.dpm_enabled) {
-			wm_high.yclk =
-				amdgpu_dpm_get_mclk(adev, false) * 10;
-			wm_high.sclk =
-				amdgpu_dpm_get_sclk(adev, false) * 10;
-		} else {
-			wm_high.yclk = adev->pm.current_mclk * 10;
-			wm_high.sclk = adev->pm.current_sclk * 10;
-		}
-
-		wm_high.disp_clk = mode->clock;
-		wm_high.src_width = mode->crtc_hdisplay;
-		wm_high.active_time = active_time;
-		wm_high.blank_time = line_time - wm_high.active_time;
-		wm_high.interlaced = false;
-		if (mode->flags & DRM_MODE_FLAG_INTERLACE)
-			wm_high.interlaced = true;
-		wm_high.vsc = amdgpu_crtc->vsc;
-		wm_high.vtaps = 1;
-		if (amdgpu_crtc->rmx_type != RMX_OFF)
-			wm_high.vtaps = 2;
-		wm_high.bytes_per_pixel = 4; /* XXX: get this from fb config */
-		wm_high.lb_size = lb_size;
-		wm_high.dram_channels = cik_get_number_of_dram_channels(adev);
-		wm_high.num_heads = num_heads;
-
-		/* set for high clocks */
-		latency_watermark_a = min(dce_v11_0_latency_watermark(&wm_high), (u32)65535);
-
-		/* possibly force display priority to high */
-		/* should really do this at mode validation time... */
-		if (!dce_v11_0_average_bandwidth_vs_dram_bandwidth_for_display(&wm_high) ||
-		    !dce_v11_0_average_bandwidth_vs_available_bandwidth(&wm_high) ||
-		    !dce_v11_0_check_latency_hiding(&wm_high) ||
-		    (adev->mode_info.disp_priority == 2)) {
-			DRM_DEBUG_KMS("force priority to high\n");
-		}
-
-		/* watermark for low clocks */
-		if (adev->pm.dpm_enabled) {
-			wm_low.yclk =
-				amdgpu_dpm_get_mclk(adev, true) * 10;
-			wm_low.sclk =
-				amdgpu_dpm_get_sclk(adev, true) * 10;
-		} else {
-			wm_low.yclk = adev->pm.current_mclk * 10;
-			wm_low.sclk = adev->pm.current_sclk * 10;
-		}
-
-		wm_low.disp_clk = mode->clock;
-		wm_low.src_width = mode->crtc_hdisplay;
-		wm_low.active_time = active_time;
-		wm_low.blank_time = line_time - wm_low.active_time;
-		wm_low.interlaced = false;
-		if (mode->flags & DRM_MODE_FLAG_INTERLACE)
-			wm_low.interlaced = true;
-		wm_low.vsc = amdgpu_crtc->vsc;
-		wm_low.vtaps = 1;
-		if (amdgpu_crtc->rmx_type != RMX_OFF)
-			wm_low.vtaps = 2;
-		wm_low.bytes_per_pixel = 4; /* XXX: get this from fb config */
-		wm_low.lb_size = lb_size;
-		wm_low.dram_channels = cik_get_number_of_dram_channels(adev);
-		wm_low.num_heads = num_heads;
-
-		/* set for low clocks */
-		latency_watermark_b = min(dce_v11_0_latency_watermark(&wm_low), (u32)65535);
-
-		/* possibly force display priority to high */
-		/* should really do this at mode validation time... */
-		if (!dce_v11_0_average_bandwidth_vs_dram_bandwidth_for_display(&wm_low) ||
-		    !dce_v11_0_average_bandwidth_vs_available_bandwidth(&wm_low) ||
-		    !dce_v11_0_check_latency_hiding(&wm_low) ||
-		    (adev->mode_info.disp_priority == 2)) {
-			DRM_DEBUG_KMS("force priority to high\n");
-		}
-		lb_vblank_lead_lines = DIV_ROUND_UP(lb_size, mode->crtc_hdisplay);
-	}
-
-	/* select wm A */
-	wm_mask = RREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(wm_mask, DPG_WATERMARK_MASK_CONTROL, URGENCY_WATERMARK_MASK, 1);
-	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-	tmp = RREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_LOW_WATERMARK, latency_watermark_a);
-	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_HIGH_WATERMARK, line_time);
-	WREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-	/* select wm B */
-	tmp = REG_SET_FIELD(wm_mask, DPG_WATERMARK_MASK_CONTROL, URGENCY_WATERMARK_MASK, 2);
-	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-	tmp = RREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_LOW_WATERMARK, latency_watermark_b);
-	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_HIGH_WATERMARK, line_time);
-	WREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-	/* restore original selection */
-	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, wm_mask);
-
-	/* save values for DPM */
-	amdgpu_crtc->line_time = line_time;
-	amdgpu_crtc->wm_high = latency_watermark_a;
-	amdgpu_crtc->wm_low = latency_watermark_b;
-	/* Save number of lines the linebuffer leads before the scanout */
-	amdgpu_crtc->lb_vblank_lead_lines = lb_vblank_lead_lines;
-}
-
-/**
- * dce_v11_0_bandwidth_update - program display watermarks
- *
- * @adev: amdgpu_device pointer
- *
- * Calculate and program the display watermarks and line
- * buffer allocation (CIK).
- */
-static void dce_v11_0_bandwidth_update(struct amdgpu_device *adev)
-{
-	struct drm_display_mode *mode = NULL;
-	u32 num_heads = 0, lb_size;
-	int i;
-
-	amdgpu_update_display_priority(adev);
-
-	for (i = 0; i < adev->mode_info.num_crtc; i++) {
-		if (adev->mode_info.crtcs[i]->base.enabled)
-			num_heads++;
-	}
-	for (i = 0; i < adev->mode_info.num_crtc; i++) {
-		mode = &adev->mode_info.crtcs[i]->base.mode;
-		lb_size = dce_v11_0_line_buffer_adjust(adev, adev->mode_info.crtcs[i], mode);
-		dce_v11_0_program_watermarks(adev, adev->mode_info.crtcs[i],
-					    lb_size, num_heads);
-	}
-}
-
-static void dce_v11_0_audio_get_connected_pins(struct amdgpu_device *adev)
-{
-	int i;
-	u32 offset, tmp;
-
-	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
-		offset = adev->mode_info.audio.pin[i].offset;
-		tmp = RREG32_AUDIO_ENDPT(offset,
-					 ixAZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT);
-		if (((tmp &
-		AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT__PORT_CONNECTIVITY_MASK) >>
-		AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT__PORT_CONNECTIVITY__SHIFT) == 1)
-			adev->mode_info.audio.pin[i].connected = false;
-		else
-			adev->mode_info.audio.pin[i].connected = true;
-	}
-}
-
-static struct amdgpu_audio_pin *dce_v11_0_audio_get_pin(struct amdgpu_device *adev)
-{
-	int i;
-
-	dce_v11_0_audio_get_connected_pins(adev);
-
-	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
-		if (adev->mode_info.audio.pin[i].connected)
-			return &adev->mode_info.audio.pin[i];
-	}
-	DRM_ERROR("No connected audio pins found!\n");
-	return NULL;
-}
-
-static void dce_v11_0_afmt_audio_select_pin(struct drm_encoder *encoder)
-{
-	struct amdgpu_device *adev = encoder->dev->dev_private;
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-	u32 tmp;
-
-	if (!dig || !dig->afmt || !dig->afmt->pin)
-		return;
-
-	tmp = RREG32(mmAFMT_AUDIO_SRC_CONTROL + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_SRC_CONTROL, AFMT_AUDIO_SRC_SELECT, dig->afmt->pin->id);
-	WREG32(mmAFMT_AUDIO_SRC_CONTROL + dig->afmt->offset, tmp);
-}
-
-static void dce_v11_0_audio_write_latency_fields(struct drm_encoder *encoder,
-						struct drm_display_mode *mode)
-{
-	struct amdgpu_device *adev = encoder->dev->dev_private;
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-	struct drm_connector *connector;
-	struct amdgpu_connector *amdgpu_connector = NULL;
-	u32 tmp;
-	int interlace = 0;
-
-	if (!dig || !dig->afmt || !dig->afmt->pin)
-		return;
-
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
-		if (connector->encoder == encoder) {
-			amdgpu_connector = to_amdgpu_connector(connector);
-			break;
-		}
-	}
-
-	if (!amdgpu_connector) {
-		DRM_ERROR("Couldn't find encoder's connector\n");
-		return;
-	}
-
-	if (mode->flags & DRM_MODE_FLAG_INTERLACE)
-		interlace = 1;
-	if (connector->latency_present[interlace]) {
-		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
-				    VIDEO_LIPSYNC, connector->video_latency[interlace]);
-		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
-				    AUDIO_LIPSYNC, connector->audio_latency[interlace]);
-	} else {
-		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
-				    VIDEO_LIPSYNC, 0);
-		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
-				    AUDIO_LIPSYNC, 0);
-	}
-	WREG32_AUDIO_ENDPT(dig->afmt->pin->offset,
-			   ixAZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC, tmp);
-}
-
-static void dce_v11_0_audio_write_speaker_allocation(struct drm_encoder *encoder)
-{
-	struct amdgpu_device *adev = encoder->dev->dev_private;
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-	struct drm_connector *connector;
-	struct amdgpu_connector *amdgpu_connector = NULL;
-	u32 tmp;
-	u8 *sadb = NULL;
-	int sad_count;
-
-	if (!dig || !dig->afmt || !dig->afmt->pin)
-		return;
-
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
-		if (connector->encoder == encoder) {
-			amdgpu_connector = to_amdgpu_connector(connector);
-			break;
-		}
-	}
-
-	if (!amdgpu_connector) {
-		DRM_ERROR("Couldn't find encoder's connector\n");
-		return;
-	}
-
-	sad_count = drm_edid_to_speaker_allocation(amdgpu_connector_edid(connector), &sadb);
-	if (sad_count < 0) {
-		DRM_ERROR("Couldn't read Speaker Allocation Data Block: %d\n", sad_count);
-		sad_count = 0;
-	}
-
-	/* program the speaker allocation */
-	tmp = RREG32_AUDIO_ENDPT(dig->afmt->pin->offset,
-				 ixAZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER);
-	tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
-			    DP_CONNECTION, 0);
-	/* set HDMI mode */
-	tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
-			    HDMI_CONNECTION, 1);
-	if (sad_count)
-		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
-				    SPEAKER_ALLOCATION, sadb[0]);
-	else
-		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
-				    SPEAKER_ALLOCATION, 5); /* stereo */
-	WREG32_AUDIO_ENDPT(dig->afmt->pin->offset,
-			   ixAZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER, tmp);
-
-	kfree(sadb);
-}
-
-static void dce_v11_0_audio_write_sad_regs(struct drm_encoder *encoder)
-{
-	struct amdgpu_device *adev = encoder->dev->dev_private;
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-	struct drm_connector *connector;
-	struct amdgpu_connector *amdgpu_connector = NULL;
-	struct cea_sad *sads;
-	int i, sad_count;
-
-	static const u16 eld_reg_to_type[][2] = {
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0, HDMI_AUDIO_CODING_TYPE_PCM },
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR1, HDMI_AUDIO_CODING_TYPE_AC3 },
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR2, HDMI_AUDIO_CODING_TYPE_MPEG1 },
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR3, HDMI_AUDIO_CODING_TYPE_MP3 },
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR4, HDMI_AUDIO_CODING_TYPE_MPEG2 },
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR5, HDMI_AUDIO_CODING_TYPE_AAC_LC },
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR6, HDMI_AUDIO_CODING_TYPE_DTS },
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR7, HDMI_AUDIO_CODING_TYPE_ATRAC },
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR9, HDMI_AUDIO_CODING_TYPE_EAC3 },
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR10, HDMI_AUDIO_CODING_TYPE_DTS_HD },
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR11, HDMI_AUDIO_CODING_TYPE_MLP },
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR13, HDMI_AUDIO_CODING_TYPE_WMA_PRO },
-	};
-
-	if (!dig || !dig->afmt || !dig->afmt->pin)
-		return;
-
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
-		if (connector->encoder == encoder) {
-			amdgpu_connector = to_amdgpu_connector(connector);
-			break;
-		}
-	}
-
-	if (!amdgpu_connector) {
-		DRM_ERROR("Couldn't find encoder's connector\n");
-		return;
-	}
-
-	sad_count = drm_edid_to_sad(amdgpu_connector_edid(connector), &sads);
-	if (sad_count <= 0) {
-		DRM_ERROR("Couldn't read SADs: %d\n", sad_count);
-		return;
-	}
-	BUG_ON(!sads);
-
-	for (i = 0; i < ARRAY_SIZE(eld_reg_to_type); i++) {
-		u32 tmp = 0;
-		u8 stereo_freqs = 0;
-		int max_channels = -1;
-		int j;
-
-		for (j = 0; j < sad_count; j++) {
-			struct cea_sad *sad = &sads[j];
-
-			if (sad->format == eld_reg_to_type[i][1]) {
-				if (sad->channels > max_channels) {
-					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
-							    MAX_CHANNELS, sad->channels);
-					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
-							    DESCRIPTOR_BYTE_2, sad->byte2);
-					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
-							    SUPPORTED_FREQUENCIES, sad->freq);
-					max_channels = sad->channels;
-				}
-
-				if (sad->format == HDMI_AUDIO_CODING_TYPE_PCM)
-					stereo_freqs |= sad->freq;
-				else
-					break;
-			}
-		}
-
-		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
-				    SUPPORTED_FREQUENCIES_STEREO, stereo_freqs);
-		WREG32_AUDIO_ENDPT(dig->afmt->pin->offset, eld_reg_to_type[i][0], tmp);
-	}
-
-	kfree(sads);
-}
-
-static void dce_v11_0_audio_enable(struct amdgpu_device *adev,
-				  struct amdgpu_audio_pin *pin,
-				  bool enable)
-{
-	if (!pin)
-		return;
-
-	WREG32_AUDIO_ENDPT(pin->offset, ixAZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL,
-			   enable ? AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL__AUDIO_ENABLED_MASK : 0);
-}
-
-static const u32 pin_offsets[] =
-{
-	AUD0_REGISTER_OFFSET,
-	AUD1_REGISTER_OFFSET,
-	AUD2_REGISTER_OFFSET,
-	AUD3_REGISTER_OFFSET,
-	AUD4_REGISTER_OFFSET,
-	AUD5_REGISTER_OFFSET,
-	AUD6_REGISTER_OFFSET,
-	AUD7_REGISTER_OFFSET,
-};
-
-static int dce_v11_0_audio_init(struct amdgpu_device *adev)
-{
-	int i;
-
-	if (!amdgpu_audio)
-		return 0;
-
-	adev->mode_info.audio.enabled = true;
-
-	switch (adev->asic_type) {
-	case CHIP_CARRIZO:
-	case CHIP_STONEY:
-		adev->mode_info.audio.num_pins = 7;
-		break;
-	case CHIP_POLARIS10:
-		adev->mode_info.audio.num_pins = 8;
-		break;
-	case CHIP_POLARIS11:
-	case CHIP_POLARIS12:
-		adev->mode_info.audio.num_pins = 6;
-		break;
-	default:
-		return -EINVAL;
-	}
-
-	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
-		adev->mode_info.audio.pin[i].channels = -1;
-		adev->mode_info.audio.pin[i].rate = -1;
-		adev->mode_info.audio.pin[i].bits_per_sample = -1;
-		adev->mode_info.audio.pin[i].status_bits = 0;
-		adev->mode_info.audio.pin[i].category_code = 0;
-		adev->mode_info.audio.pin[i].connected = false;
-		adev->mode_info.audio.pin[i].offset = pin_offsets[i];
-		adev->mode_info.audio.pin[i].id = i;
-		/* disable audio.  it will be set up later */
-		/* XXX remove once we switch to ip funcs */
-		dce_v11_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
-	}
-
-	return 0;
-}
-
-static void dce_v11_0_audio_fini(struct amdgpu_device *adev)
-{
-	int i;
-
-	if (!amdgpu_audio)
-		return;
-
-	if (!adev->mode_info.audio.enabled)
-		return;
-
-	for (i = 0; i < adev->mode_info.audio.num_pins; i++)
-		dce_v11_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
-
-	adev->mode_info.audio.enabled = false;
-}
-
-/*
- * update the N and CTS parameters for a given pixel clock rate
- */
-static void dce_v11_0_afmt_update_ACR(struct drm_encoder *encoder, uint32_t clock)
-{
-	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-	struct amdgpu_afmt_acr acr = amdgpu_afmt_acr(clock);
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-	u32 tmp;
-
-	tmp = RREG32(mmHDMI_ACR_32_0 + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, HDMI_ACR_32_0, HDMI_ACR_CTS_32, acr.cts_32khz);
-	WREG32(mmHDMI_ACR_32_0 + dig->afmt->offset, tmp);
-	tmp = RREG32(mmHDMI_ACR_32_1 + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, HDMI_ACR_32_1, HDMI_ACR_N_32, acr.n_32khz);
-	WREG32(mmHDMI_ACR_32_1 + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmHDMI_ACR_44_0 + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, HDMI_ACR_44_0, HDMI_ACR_CTS_44, acr.cts_44_1khz);
-	WREG32(mmHDMI_ACR_44_0 + dig->afmt->offset, tmp);
-	tmp = RREG32(mmHDMI_ACR_44_1 + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, HDMI_ACR_44_1, HDMI_ACR_N_44, acr.n_44_1khz);
-	WREG32(mmHDMI_ACR_44_1 + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmHDMI_ACR_48_0 + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, HDMI_ACR_48_0, HDMI_ACR_CTS_48, acr.cts_48khz);
-	WREG32(mmHDMI_ACR_48_0 + dig->afmt->offset, tmp);
-	tmp = RREG32(mmHDMI_ACR_48_1 + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, HDMI_ACR_48_1, HDMI_ACR_N_48, acr.n_48khz);
-	WREG32(mmHDMI_ACR_48_1 + dig->afmt->offset, tmp);
-
-}
-
-/*
- * build a HDMI Video Info Frame
- */
-static void dce_v11_0_afmt_update_avi_infoframe(struct drm_encoder *encoder,
-					       void *buffer, size_t size)
-{
-	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-	uint8_t *frame = buffer + 3;
-	uint8_t *header = buffer;
-
-	WREG32(mmAFMT_AVI_INFO0 + dig->afmt->offset,
-		frame[0x0] | (frame[0x1] << 8) | (frame[0x2] << 16) | (frame[0x3] << 24));
-	WREG32(mmAFMT_AVI_INFO1 + dig->afmt->offset,
-		frame[0x4] | (frame[0x5] << 8) | (frame[0x6] << 16) | (frame[0x7] << 24));
-	WREG32(mmAFMT_AVI_INFO2 + dig->afmt->offset,
-		frame[0x8] | (frame[0x9] << 8) | (frame[0xA] << 16) | (frame[0xB] << 24));
-	WREG32(mmAFMT_AVI_INFO3 + dig->afmt->offset,
-		frame[0xC] | (frame[0xD] << 8) | (header[1] << 24));
-}
-
-static void dce_v11_0_audio_set_dto(struct drm_encoder *encoder, u32 clock)
-{
-	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
-	u32 dto_phase = 24 * 1000;
-	u32 dto_modulo = clock;
-	u32 tmp;
-
-	if (!dig || !dig->afmt)
-		return;
-
-	/* XXX two dtos; generally use dto0 for hdmi */
-	/* Express [24MHz / target pixel clock] as an exact rational
-	 * number (coefficient of two integer numbers.  DCCG_AUDIO_DTOx_PHASE
-	 * is the numerator, DCCG_AUDIO_DTOx_MODULE is the denominator
-	 */
-	tmp = RREG32(mmDCCG_AUDIO_DTO_SOURCE);
-	tmp = REG_SET_FIELD(tmp, DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO0_SOURCE_SEL,
-			    amdgpu_crtc->crtc_id);
-	WREG32(mmDCCG_AUDIO_DTO_SOURCE, tmp);
-	WREG32(mmDCCG_AUDIO_DTO0_PHASE, dto_phase);
-	WREG32(mmDCCG_AUDIO_DTO0_MODULE, dto_modulo);
-}
-
-/*
- * update the info frames with the data from the current display mode
- */
-static void dce_v11_0_afmt_setmode(struct drm_encoder *encoder,
-				  struct drm_display_mode *mode)
-{
-	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
-	u8 buffer[HDMI_INFOFRAME_HEADER_SIZE + HDMI_AVI_INFOFRAME_SIZE];
-	struct hdmi_avi_infoframe frame;
-	ssize_t err;
-	u32 tmp;
-	int bpc = 8;
-
-	if (!dig || !dig->afmt)
-		return;
-
-	/* Silent, r600_hdmi_enable will raise WARN for us */
-	if (!dig->afmt->enabled)
-		return;
-
-	/* hdmi deep color mode general control packets setup, if bpc > 8 */
-	if (encoder->crtc) {
-		struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
-		bpc = amdgpu_crtc->bpc;
-	}
-
-	/* disable audio prior to setting up hw */
-	dig->afmt->pin = dce_v11_0_audio_get_pin(adev);
-	dce_v11_0_audio_enable(adev, dig->afmt->pin, false);
-
-	dce_v11_0_audio_set_dto(encoder, mode->clock);
-
-	tmp = RREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_NULL_SEND, 1);
-	WREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset, tmp); /* send null packets when required */
-
-	WREG32(mmAFMT_AUDIO_CRC_CONTROL + dig->afmt->offset, 0x1000);
-
-	tmp = RREG32(mmHDMI_CONTROL + dig->afmt->offset);
-	switch (bpc) {
-	case 0:
-	case 6:
-	case 8:
-	case 16:
-	default:
-		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, 0);
-		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 0);
-		DRM_DEBUG("%s: Disabling hdmi deep color for %d bpc.\n",
-			  connector->name, bpc);
-		break;
-	case 10:
-		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, 1);
-		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 1);
-		DRM_DEBUG("%s: Enabling hdmi deep color 30 for 10 bpc.\n",
-			  connector->name);
-		break;
-	case 12:
-		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, 1);
-		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 2);
-		DRM_DEBUG("%s: Enabling hdmi deep color 36 for 12 bpc.\n",
-			  connector->name);
-		break;
-	}
-	WREG32(mmHDMI_CONTROL + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_NULL_SEND, 1); /* send null packets when required */
-	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_GC_SEND, 1); /* send general control packets */
-	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_GC_CONT, 1); /* send general control packets every frame */
-	WREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset);
-	/* enable audio info frames (frames won't be set until audio is enabled) */
-	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, 1);
-	/* required for audio info values to be updated */
-	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_CONT, 1);
-	WREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmAFMT_INFOFRAME_CONTROL0 + dig->afmt->offset);
-	/* required for audio info values to be updated */
-	tmp = REG_SET_FIELD(tmp, AFMT_INFOFRAME_CONTROL0, AFMT_AUDIO_INFO_UPDATE, 1);
-	WREG32(mmAFMT_INFOFRAME_CONTROL0 + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset);
-	/* anything other than 0 */
-	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL1, HDMI_AUDIO_INFO_LINE, 2);
-	WREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset, tmp);
-
-	WREG32(mmHDMI_GC + dig->afmt->offset, 0); /* unset HDMI_GC_AVMUTE */
-
-	tmp = RREG32(mmHDMI_AUDIO_PACKET_CONTROL + dig->afmt->offset);
-	/* set the default audio delay */
-	tmp = REG_SET_FIELD(tmp, HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_DELAY_EN, 1);
-	/* should be suffient for all audio modes and small enough for all hblanks */
-	tmp = REG_SET_FIELD(tmp, HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_PACKETS_PER_LINE, 3);
-	WREG32(mmHDMI_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset);
-	/* allow 60958 channel status fields to be updated */
-	tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_PACKET_CONTROL, AFMT_60958_CS_UPDATE, 1);
-	WREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmHDMI_ACR_PACKET_CONTROL + dig->afmt->offset);
-	if (bpc > 8)
-		/* clear SW CTS value */
-		tmp = REG_SET_FIELD(tmp, HDMI_ACR_PACKET_CONTROL, HDMI_ACR_SOURCE, 0);
-	else
-		/* select SW CTS value */
-		tmp = REG_SET_FIELD(tmp, HDMI_ACR_PACKET_CONTROL, HDMI_ACR_SOURCE, 1);
-	/* allow hw to sent ACR packets when required */
-	tmp = REG_SET_FIELD(tmp, HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUTO_SEND, 1);
-	WREG32(mmHDMI_ACR_PACKET_CONTROL + dig->afmt->offset, tmp);
-
-	dce_v11_0_afmt_update_ACR(encoder, mode->clock);
-
-	tmp = RREG32(mmAFMT_60958_0 + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, AFMT_60958_0, AFMT_60958_CS_CHANNEL_NUMBER_L, 1);
-	WREG32(mmAFMT_60958_0 + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmAFMT_60958_1 + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, AFMT_60958_1, AFMT_60958_CS_CHANNEL_NUMBER_R, 2);
-	WREG32(mmAFMT_60958_1 + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmAFMT_60958_2 + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_2, 3);
-	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_3, 4);
-	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_4, 5);
-	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_5, 6);
-	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_6, 7);
-	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_7, 8);
-	WREG32(mmAFMT_60958_2 + dig->afmt->offset, tmp);
-
-	dce_v11_0_audio_write_speaker_allocation(encoder);
-
-	WREG32(mmAFMT_AUDIO_PACKET_CONTROL2 + dig->afmt->offset,
-	       (0xff << AFMT_AUDIO_PACKET_CONTROL2__AFMT_AUDIO_CHANNEL_ENABLE__SHIFT));
-
-	dce_v11_0_afmt_audio_select_pin(encoder);
-	dce_v11_0_audio_write_sad_regs(encoder);
-	dce_v11_0_audio_write_latency_fields(encoder, mode);
-
-	err = drm_hdmi_avi_infoframe_from_display_mode(&frame, mode, false);
-	if (err < 0) {
-		DRM_ERROR("failed to setup AVI infoframe: %zd\n", err);
-		return;
-	}
-
-	err = hdmi_avi_infoframe_pack(&frame, buffer, sizeof(buffer));
-	if (err < 0) {
-		DRM_ERROR("failed to pack AVI infoframe: %zd\n", err);
-		return;
-	}
-
-	dce_v11_0_afmt_update_avi_infoframe(encoder, buffer, sizeof(buffer));
-
-	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset);
-	/* enable AVI info frames */
-	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AVI_INFO_SEND, 1);
-	/* required for audio info values to be updated */
-	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AVI_INFO_CONT, 1);
-	WREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL1, HDMI_AVI_INFO_LINE, 2);
-	WREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset);
-	/* send audio packets */
-	tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_PACKET_CONTROL, AFMT_AUDIO_SAMPLE_SEND, 1);
-	WREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
-
-	WREG32(mmAFMT_RAMP_CONTROL0 + dig->afmt->offset, 0x00FFFFFF);
-	WREG32(mmAFMT_RAMP_CONTROL1 + dig->afmt->offset, 0x007FFFFF);
-	WREG32(mmAFMT_RAMP_CONTROL2 + dig->afmt->offset, 0x00000001);
-	WREG32(mmAFMT_RAMP_CONTROL3 + dig->afmt->offset, 0x00000001);
-
-	/* enable audio after to setting up hw */
-	dce_v11_0_audio_enable(adev, dig->afmt->pin, true);
-}
-
-static void dce_v11_0_afmt_enable(struct drm_encoder *encoder, bool enable)
-{
-	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-
-	if (!dig || !dig->afmt)
-		return;
-
-	/* Silent, r600_hdmi_enable will raise WARN for us */
-	if (enable && dig->afmt->enabled)
-		return;
-	if (!enable && !dig->afmt->enabled)
-		return;
-
-	if (!enable && dig->afmt->pin) {
-		dce_v11_0_audio_enable(adev, dig->afmt->pin, false);
-		dig->afmt->pin = NULL;
-	}
-
-	dig->afmt->enabled = enable;
-
-	DRM_DEBUG("%sabling AFMT interface @ 0x%04X for encoder 0x%x\n",
-		  enable ? "En" : "Dis", dig->afmt->offset, amdgpu_encoder->encoder_id);
-}
-
-static int dce_v11_0_afmt_init(struct amdgpu_device *adev)
-{
-	int i;
-
-	for (i = 0; i < adev->mode_info.num_dig; i++)
-		adev->mode_info.afmt[i] = NULL;
-
-	/* DCE11 has audio blocks tied to DIG encoders */
-	for (i = 0; i < adev->mode_info.num_dig; i++) {
-		adev->mode_info.afmt[i] = kzalloc(sizeof(struct amdgpu_afmt), GFP_KERNEL);
-		if (adev->mode_info.afmt[i]) {
-			adev->mode_info.afmt[i]->offset = dig_offsets[i];
-			adev->mode_info.afmt[i]->id = i;
-		} else {
-			int j;
-			for (j = 0; j < i; j++) {
-				kfree(adev->mode_info.afmt[j]);
-				adev->mode_info.afmt[j] = NULL;
-			}
-			return -ENOMEM;
-		}
-	}
-	return 0;
-}
-
-static void dce_v11_0_afmt_fini(struct amdgpu_device *adev)
-{
-	int i;
-
-	for (i = 0; i < adev->mode_info.num_dig; i++) {
-		kfree(adev->mode_info.afmt[i]);
-		adev->mode_info.afmt[i] = NULL;
-	}
-}
-
-static const u32 vga_control_regs[6] =
-{
-	mmD1VGA_CONTROL,
-	mmD2VGA_CONTROL,
-	mmD3VGA_CONTROL,
-	mmD4VGA_CONTROL,
-	mmD5VGA_CONTROL,
-	mmD6VGA_CONTROL,
-};
-
-static void dce_v11_0_vga_enable(struct drm_crtc *crtc, bool enable)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-	u32 vga_control;
-
-	vga_control = RREG32(vga_control_regs[amdgpu_crtc->crtc_id]) & ~1;
-	if (enable)
-		WREG32(vga_control_regs[amdgpu_crtc->crtc_id], vga_control | 1);
-	else
-		WREG32(vga_control_regs[amdgpu_crtc->crtc_id], vga_control);
-}
-
-static void dce_v11_0_grph_enable(struct drm_crtc *crtc, bool enable)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-
-	if (enable)
-		WREG32(mmGRPH_ENABLE + amdgpu_crtc->crtc_offset, 1);
-	else
-		WREG32(mmGRPH_ENABLE + amdgpu_crtc->crtc_offset, 0);
-}
-
-static int dce_v11_0_crtc_do_set_base(struct drm_crtc *crtc,
-				     struct drm_framebuffer *fb,
-				     int x, int y, int atomic)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-	struct amdgpu_framebuffer *amdgpu_fb;
-	struct drm_framebuffer *target_fb;
-	struct drm_gem_object *obj;
-	struct amdgpu_bo *abo;
-	uint64_t fb_location, tiling_flags;
-	uint32_t fb_format, fb_pitch_pixels;
-	u32 fb_swap = REG_SET_FIELD(0, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP, ENDIAN_NONE);
-	u32 pipe_config;
-	u32 tmp, viewport_w, viewport_h;
-	int r;
-	bool bypass_lut = false;
-	struct drm_format_name_buf format_name;
-
-	/* no fb bound */
-	if (!atomic && !crtc->primary->fb) {
-		DRM_DEBUG_KMS("No FB bound\n");
-		return 0;
-	}
-
-	if (atomic) {
-		amdgpu_fb = to_amdgpu_framebuffer(fb);
-		target_fb = fb;
-	} else {
-		amdgpu_fb = to_amdgpu_framebuffer(crtc->primary->fb);
-		target_fb = crtc->primary->fb;
-	}
-
-	/* If atomic, assume fb object is pinned & idle & fenced and
-	 * just update base pointers
-	 */
-	obj = amdgpu_fb->obj;
-	abo = gem_to_amdgpu_bo(obj);
-	r = amdgpu_bo_reserve(abo, false);
-	if (unlikely(r != 0))
-		return r;
-
-	if (atomic) {
-		fb_location = amdgpu_bo_gpu_offset(abo);
-	} else {
-		r = amdgpu_bo_pin(abo, AMDGPU_GEM_DOMAIN_VRAM, &fb_location);
-		if (unlikely(r != 0)) {
-			amdgpu_bo_unreserve(abo);
-			return -EINVAL;
-		}
-	}
-
-	amdgpu_bo_get_tiling_flags(abo, &tiling_flags);
-	amdgpu_bo_unreserve(abo);
-
-	pipe_config = AMDGPU_TILING_GET(tiling_flags, PIPE_CONFIG);
-
-	switch (target_fb->format->format) {
-	case DRM_FORMAT_C8:
-		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 0);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 0);
-		break;
-	case DRM_FORMAT_XRGB4444:
-	case DRM_FORMAT_ARGB4444:
-		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 2);
-#ifdef __BIG_ENDIAN
-		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
-					ENDIAN_8IN16);
-#endif
-		break;
-	case DRM_FORMAT_XRGB1555:
-	case DRM_FORMAT_ARGB1555:
-		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 0);
-#ifdef __BIG_ENDIAN
-		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
-					ENDIAN_8IN16);
-#endif
-		break;
-	case DRM_FORMAT_BGRX5551:
-	case DRM_FORMAT_BGRA5551:
-		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 5);
-#ifdef __BIG_ENDIAN
-		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
-					ENDIAN_8IN16);
-#endif
-		break;
-	case DRM_FORMAT_RGB565:
-		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 1);
-#ifdef __BIG_ENDIAN
-		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
-					ENDIAN_8IN16);
-#endif
-		break;
-	case DRM_FORMAT_XRGB8888:
-	case DRM_FORMAT_ARGB8888:
-		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 2);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 0);
-#ifdef __BIG_ENDIAN
-		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
-					ENDIAN_8IN32);
-#endif
-		break;
-	case DRM_FORMAT_XRGB2101010:
-	case DRM_FORMAT_ARGB2101010:
-		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 2);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 1);
-#ifdef __BIG_ENDIAN
-		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
-					ENDIAN_8IN32);
-#endif
-		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
-		bypass_lut = true;
-		break;
-	case DRM_FORMAT_BGRX1010102:
-	case DRM_FORMAT_BGRA1010102:
-		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 2);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 4);
-#ifdef __BIG_ENDIAN
-		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
-					ENDIAN_8IN32);
-#endif
-		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
-		bypass_lut = true;
-		break;
-	default:
-		DRM_ERROR("Unsupported screen format %s\n",
-		          drm_get_format_name(target_fb->format->format, &format_name));
-		return -EINVAL;
-	}
-
-	if (AMDGPU_TILING_GET(tiling_flags, ARRAY_MODE) == ARRAY_2D_TILED_THIN1) {
-		unsigned bankw, bankh, mtaspect, tile_split, num_banks;
-
-		bankw = AMDGPU_TILING_GET(tiling_flags, BANK_WIDTH);
-		bankh = AMDGPU_TILING_GET(tiling_flags, BANK_HEIGHT);
-		mtaspect = AMDGPU_TILING_GET(tiling_flags, MACRO_TILE_ASPECT);
-		tile_split = AMDGPU_TILING_GET(tiling_flags, TILE_SPLIT);
-		num_banks = AMDGPU_TILING_GET(tiling_flags, NUM_BANKS);
-
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_NUM_BANKS, num_banks);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_ARRAY_MODE,
-					  ARRAY_2D_TILED_THIN1);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_TILE_SPLIT,
-					  tile_split);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_BANK_WIDTH, bankw);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_BANK_HEIGHT, bankh);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_MACRO_TILE_ASPECT,
-					  mtaspect);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_MICRO_TILE_MODE,
-					  ADDR_SURF_MICRO_TILING_DISPLAY);
-	} else if (AMDGPU_TILING_GET(tiling_flags, ARRAY_MODE) == ARRAY_1D_TILED_THIN1) {
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_ARRAY_MODE,
-					  ARRAY_1D_TILED_THIN1);
-	}
-
-	fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_PIPE_CONFIG,
-				  pipe_config);
-
-	dce_v11_0_vga_enable(crtc, false);
-
-	/* Make sure surface address is updated at vertical blank rather than
-	 * horizontal blank
-	 */
-	tmp = RREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, GRPH_FLIP_CONTROL,
-			    GRPH_SURFACE_UPDATE_H_RETRACE_EN, 0);
-	WREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-
-	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
-	       upper_32_bits(fb_location));
-	WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
-	       upper_32_bits(fb_location));
-	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
-	       (u32)fb_location & GRPH_PRIMARY_SURFACE_ADDRESS__GRPH_PRIMARY_SURFACE_ADDRESS_MASK);
-	WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
-	       (u32) fb_location & GRPH_SECONDARY_SURFACE_ADDRESS__GRPH_SECONDARY_SURFACE_ADDRESS_MASK);
-	WREG32(mmGRPH_CONTROL + amdgpu_crtc->crtc_offset, fb_format);
-	WREG32(mmGRPH_SWAP_CNTL + amdgpu_crtc->crtc_offset, fb_swap);
-
-	/*
-	 * The LUT only has 256 slots for indexing by a 8 bpc fb. Bypass the LUT
-	 * for > 8 bpc scanout to avoid truncation of fb indices to 8 msb's, to
-	 * retain the full precision throughout the pipeline.
-	 */
-	tmp = RREG32(mmGRPH_LUT_10BIT_BYPASS + amdgpu_crtc->crtc_offset);
-	if (bypass_lut)
-		tmp = REG_SET_FIELD(tmp, GRPH_LUT_10BIT_BYPASS, GRPH_LUT_10BIT_BYPASS_EN, 1);
-	else
-		tmp = REG_SET_FIELD(tmp, GRPH_LUT_10BIT_BYPASS, GRPH_LUT_10BIT_BYPASS_EN, 0);
-	WREG32(mmGRPH_LUT_10BIT_BYPASS + amdgpu_crtc->crtc_offset, tmp);
-
-	if (bypass_lut)
-		DRM_DEBUG_KMS("Bypassing hardware LUT due to 10 bit fb scanout.\n");
-
-	WREG32(mmGRPH_SURFACE_OFFSET_X + amdgpu_crtc->crtc_offset, 0);
-	WREG32(mmGRPH_SURFACE_OFFSET_Y + amdgpu_crtc->crtc_offset, 0);
-	WREG32(mmGRPH_X_START + amdgpu_crtc->crtc_offset, 0);
-	WREG32(mmGRPH_Y_START + amdgpu_crtc->crtc_offset, 0);
-	WREG32(mmGRPH_X_END + amdgpu_crtc->crtc_offset, target_fb->width);
-	WREG32(mmGRPH_Y_END + amdgpu_crtc->crtc_offset, target_fb->height);
-
-	fb_pitch_pixels = target_fb->pitches[0] / target_fb->format->cpp[0];
-	WREG32(mmGRPH_PITCH + amdgpu_crtc->crtc_offset, fb_pitch_pixels);
-
-	dce_v11_0_grph_enable(crtc, true);
-
-	WREG32(mmLB_DESKTOP_HEIGHT + amdgpu_crtc->crtc_offset,
-	       target_fb->height);
-
-	x &= ~3;
-	y &= ~1;
-	WREG32(mmVIEWPORT_START + amdgpu_crtc->crtc_offset,
-	       (x << 16) | y);
-	viewport_w = crtc->mode.hdisplay;
-	viewport_h = (crtc->mode.vdisplay + 1) & ~1;
-	WREG32(mmVIEWPORT_SIZE + amdgpu_crtc->crtc_offset,
-	       (viewport_w << 16) | viewport_h);
-
-	/* set pageflip to happen anywhere in vblank interval */
-	WREG32(mmCRTC_MASTER_UPDATE_MODE + amdgpu_crtc->crtc_offset, 0);
-
-	if (!atomic && fb && fb != crtc->primary->fb) {
-		amdgpu_fb = to_amdgpu_framebuffer(fb);
-		abo = gem_to_amdgpu_bo(amdgpu_fb->obj);
-		r = amdgpu_bo_reserve(abo, true);
-		if (unlikely(r != 0))
-			return r;
-		amdgpu_bo_unpin(abo);
-		amdgpu_bo_unreserve(abo);
-	}
-
-	/* Bytes per pixel may have changed */
-	dce_v11_0_bandwidth_update(adev);
-
-	return 0;
-}
-
-static void dce_v11_0_set_interleave(struct drm_crtc *crtc,
-				     struct drm_display_mode *mode)
-{
-	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	u32 tmp;
-
-	tmp = RREG32(mmLB_DATA_FORMAT + amdgpu_crtc->crtc_offset);
-	if (mode->flags & DRM_MODE_FLAG_INTERLACE)
-		tmp = REG_SET_FIELD(tmp, LB_DATA_FORMAT, INTERLEAVE_EN, 1);
-	else
-		tmp = REG_SET_FIELD(tmp, LB_DATA_FORMAT, INTERLEAVE_EN, 0);
-	WREG32(mmLB_DATA_FORMAT + amdgpu_crtc->crtc_offset, tmp);
-}
-
-static void dce_v11_0_crtc_load_lut(struct drm_crtc *crtc)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-	u16 *r, *g, *b;
-	int i;
-	u32 tmp;
-
-	DRM_DEBUG_KMS("%d\n", amdgpu_crtc->crtc_id);
-
-	tmp = RREG32(mmINPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, INPUT_CSC_CONTROL, INPUT_CSC_GRPH_MODE, 0);
-	WREG32(mmINPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-
-	tmp = RREG32(mmPRESCALE_GRPH_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, PRESCALE_GRPH_CONTROL, GRPH_PRESCALE_BYPASS, 1);
-	WREG32(mmPRESCALE_GRPH_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-
-	tmp = RREG32(mmINPUT_GAMMA_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, INPUT_GAMMA_CONTROL, GRPH_INPUT_GAMMA_MODE, 0);
-	WREG32(mmINPUT_GAMMA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-
-	WREG32(mmDC_LUT_CONTROL + amdgpu_crtc->crtc_offset, 0);
-
-	WREG32(mmDC_LUT_BLACK_OFFSET_BLUE + amdgpu_crtc->crtc_offset, 0);
-	WREG32(mmDC_LUT_BLACK_OFFSET_GREEN + amdgpu_crtc->crtc_offset, 0);
-	WREG32(mmDC_LUT_BLACK_OFFSET_RED + amdgpu_crtc->crtc_offset, 0);
-
-	WREG32(mmDC_LUT_WHITE_OFFSET_BLUE + amdgpu_crtc->crtc_offset, 0xffff);
-	WREG32(mmDC_LUT_WHITE_OFFSET_GREEN + amdgpu_crtc->crtc_offset, 0xffff);
-	WREG32(mmDC_LUT_WHITE_OFFSET_RED + amdgpu_crtc->crtc_offset, 0xffff);
-
-	WREG32(mmDC_LUT_RW_MODE + amdgpu_crtc->crtc_offset, 0);
-	WREG32(mmDC_LUT_WRITE_EN_MASK + amdgpu_crtc->crtc_offset, 0x00000007);
-
-	WREG32(mmDC_LUT_RW_INDEX + amdgpu_crtc->crtc_offset, 0);
-	r = crtc->gamma_store;
-	g = r + crtc->gamma_size;
-	b = g + crtc->gamma_size;
-	for (i = 0; i < 256; i++) {
-		WREG32(mmDC_LUT_30_COLOR + amdgpu_crtc->crtc_offset,
-		       ((*r++ & 0xffc0) << 14) |
-		       ((*g++ & 0xffc0) << 4) |
-		       (*b++ >> 6));
-	}
-
-	tmp = RREG32(mmDEGAMMA_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, DEGAMMA_CONTROL, GRPH_DEGAMMA_MODE, 0);
-	tmp = REG_SET_FIELD(tmp, DEGAMMA_CONTROL, CURSOR_DEGAMMA_MODE, 0);
-	tmp = REG_SET_FIELD(tmp, DEGAMMA_CONTROL, CURSOR2_DEGAMMA_MODE, 0);
-	WREG32(mmDEGAMMA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-
-	tmp = RREG32(mmGAMUT_REMAP_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, GAMUT_REMAP_CONTROL, GRPH_GAMUT_REMAP_MODE, 0);
-	WREG32(mmGAMUT_REMAP_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-
-	tmp = RREG32(mmREGAMMA_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, REGAMMA_CONTROL, GRPH_REGAMMA_MODE, 0);
-	WREG32(mmREGAMMA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-
-	tmp = RREG32(mmOUTPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, OUTPUT_CSC_CONTROL, OUTPUT_CSC_GRPH_MODE, 0);
-	WREG32(mmOUTPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-
-	/* XXX match this to the depth of the crtc fmt block, move to modeset? */
-	WREG32(mmDENORM_CONTROL + amdgpu_crtc->crtc_offset, 0);
-	/* XXX this only needs to be programmed once per crtc at startup,
-	 * not sure where the best place for it is
-	 */
-	tmp = RREG32(mmALPHA_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, ALPHA_CONTROL, CURSOR_ALPHA_BLND_ENA, 1);
-	WREG32(mmALPHA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-}
-
-static int dce_v11_0_pick_dig_encoder(struct drm_encoder *encoder)
-{
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-
-	switch (amdgpu_encoder->encoder_id) {
-	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
-		if (dig->linkb)
-			return 1;
-		else
-			return 0;
-		break;
-	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
-		if (dig->linkb)
-			return 3;
-		else
-			return 2;
-		break;
-	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
-		if (dig->linkb)
-			return 5;
-		else
-			return 4;
-		break;
-	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
-		return 6;
-		break;
-	default:
-		DRM_ERROR("invalid encoder_id: 0x%x\n", amdgpu_encoder->encoder_id);
-		return 0;
-	}
-}
-
-/**
- * dce_v11_0_pick_pll - Allocate a PPLL for use by the crtc.
- *
- * @crtc: drm crtc
- *
- * Returns the PPLL (Pixel PLL) to be used by the crtc.  For DP monitors
- * a single PPLL can be used for all DP crtcs/encoders.  For non-DP
- * monitors a dedicated PPLL must be used.  If a particular board has
- * an external DP PLL, return ATOM_PPLL_INVALID to skip PLL programming
- * as there is no need to program the PLL itself.  If we are not able to
- * allocate a PLL, return ATOM_PPLL_INVALID to skip PLL programming to
- * avoid messing up an existing monitor.
- *
- * Asic specific PLL information
- *
- * DCE 10.x
- * Tonga
- * - PPLL1, PPLL2 are available for all UNIPHY (both DP and non-DP)
- * CI
- * - PPLL0, PPLL1, PPLL2 are available for all UNIPHY (both DP and non-DP) and DAC
- *
- */
-static u32 dce_v11_0_pick_pll(struct drm_crtc *crtc)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-	u32 pll_in_use;
-	int pll;
-
-	if ((adev->asic_type == CHIP_POLARIS10) ||
-	    (adev->asic_type == CHIP_POLARIS11) ||
-	    (adev->asic_type == CHIP_POLARIS12)) {
-		struct amdgpu_encoder *amdgpu_encoder =
-			to_amdgpu_encoder(amdgpu_crtc->encoder);
-		struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-
-		if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder)))
-			return ATOM_DP_DTO;
-
-		switch (amdgpu_encoder->encoder_id) {
-		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
-			if (dig->linkb)
-				return ATOM_COMBOPHY_PLL1;
-			else
-				return ATOM_COMBOPHY_PLL0;
-			break;
-		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
-			if (dig->linkb)
-				return ATOM_COMBOPHY_PLL3;
-			else
-				return ATOM_COMBOPHY_PLL2;
-			break;
-		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
-			if (dig->linkb)
-				return ATOM_COMBOPHY_PLL5;
-			else
-				return ATOM_COMBOPHY_PLL4;
-			break;
-		default:
-			DRM_ERROR("invalid encoder_id: 0x%x\n", amdgpu_encoder->encoder_id);
-			return ATOM_PPLL_INVALID;
-		}
-	}
-
-	if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder))) {
-		if (adev->clock.dp_extclk)
-			/* skip PPLL programming if using ext clock */
-			return ATOM_PPLL_INVALID;
-		else {
-			/* use the same PPLL for all DP monitors */
-			pll = amdgpu_pll_get_shared_dp_ppll(crtc);
-			if (pll != ATOM_PPLL_INVALID)
-				return pll;
-		}
-	} else {
-		/* use the same PPLL for all monitors with the same clock */
-		pll = amdgpu_pll_get_shared_nondp_ppll(crtc);
-		if (pll != ATOM_PPLL_INVALID)
-			return pll;
-	}
-
-	/* XXX need to determine what plls are available on each DCE11 part */
-	pll_in_use = amdgpu_pll_get_use_mask(crtc);
-	if (adev->asic_type == CHIP_CARRIZO || adev->asic_type == CHIP_STONEY) {
-		if (!(pll_in_use & (1 << ATOM_PPLL1)))
-			return ATOM_PPLL1;
-		if (!(pll_in_use & (1 << ATOM_PPLL0)))
-			return ATOM_PPLL0;
-		DRM_ERROR("unable to allocate a PPLL\n");
-		return ATOM_PPLL_INVALID;
-	} else {
-		if (!(pll_in_use & (1 << ATOM_PPLL2)))
-			return ATOM_PPLL2;
-		if (!(pll_in_use & (1 << ATOM_PPLL1)))
-			return ATOM_PPLL1;
-		if (!(pll_in_use & (1 << ATOM_PPLL0)))
-			return ATOM_PPLL0;
-		DRM_ERROR("unable to allocate a PPLL\n");
-		return ATOM_PPLL_INVALID;
-	}
-	return ATOM_PPLL_INVALID;
-}
-
-static void dce_v11_0_lock_cursor(struct drm_crtc *crtc, bool lock)
-{
-	struct amdgpu_device *adev = crtc->dev->dev_private;
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	uint32_t cur_lock;
-
-	cur_lock = RREG32(mmCUR_UPDATE + amdgpu_crtc->crtc_offset);
-	if (lock)
-		cur_lock = REG_SET_FIELD(cur_lock, CUR_UPDATE, CURSOR_UPDATE_LOCK, 1);
-	else
-		cur_lock = REG_SET_FIELD(cur_lock, CUR_UPDATE, CURSOR_UPDATE_LOCK, 0);
-	WREG32(mmCUR_UPDATE + amdgpu_crtc->crtc_offset, cur_lock);
-}
-
-static void dce_v11_0_hide_cursor(struct drm_crtc *crtc)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct amdgpu_device *adev = crtc->dev->dev_private;
-	u32 tmp;
-
-	tmp = RREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, CUR_CONTROL, CURSOR_EN, 0);
-	WREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-}
-
-static void dce_v11_0_show_cursor(struct drm_crtc *crtc)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct amdgpu_device *adev = crtc->dev->dev_private;
-	u32 tmp;
-
-	WREG32(mmCUR_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
-	       upper_32_bits(amdgpu_crtc->cursor_addr));
-	WREG32(mmCUR_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
-	       lower_32_bits(amdgpu_crtc->cursor_addr));
-
-	tmp = RREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, CUR_CONTROL, CURSOR_EN, 1);
-	tmp = REG_SET_FIELD(tmp, CUR_CONTROL, CURSOR_MODE, 2);
-	WREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-}
-
-static int dce_v11_0_cursor_move_locked(struct drm_crtc *crtc,
-					int x, int y)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct amdgpu_device *adev = crtc->dev->dev_private;
-	int xorigin = 0, yorigin = 0;
-
-	amdgpu_crtc->cursor_x = x;
-	amdgpu_crtc->cursor_y = y;
-
-	/* avivo cursor are offset into the total surface */
-	x += crtc->x;
-	y += crtc->y;
-	DRM_DEBUG("x %d y %d c->x %d c->y %d\n", x, y, crtc->x, crtc->y);
-
-	if (x < 0) {
-		xorigin = min(-x, amdgpu_crtc->max_cursor_width - 1);
-		x = 0;
-	}
-	if (y < 0) {
-		yorigin = min(-y, amdgpu_crtc->max_cursor_height - 1);
-		y = 0;
-	}
-
-	WREG32(mmCUR_POSITION + amdgpu_crtc->crtc_offset, (x << 16) | y);
-	WREG32(mmCUR_HOT_SPOT + amdgpu_crtc->crtc_offset, (xorigin << 16) | yorigin);
-	WREG32(mmCUR_SIZE + amdgpu_crtc->crtc_offset,
-	       ((amdgpu_crtc->cursor_width - 1) << 16) | (amdgpu_crtc->cursor_height - 1));
-
-	return 0;
-}
-
-static int dce_v11_0_crtc_cursor_move(struct drm_crtc *crtc,
-				      int x, int y)
-{
-	int ret;
-
-	dce_v11_0_lock_cursor(crtc, true);
-	ret = dce_v11_0_cursor_move_locked(crtc, x, y);
-	dce_v11_0_lock_cursor(crtc, false);
-
-	return ret;
-}
-
-static int dce_v11_0_crtc_cursor_set2(struct drm_crtc *crtc,
-				      struct drm_file *file_priv,
-				      uint32_t handle,
-				      uint32_t width,
-				      uint32_t height,
-				      int32_t hot_x,
-				      int32_t hot_y)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct drm_gem_object *obj;
-	struct amdgpu_bo *aobj;
-	int ret;
-
-	if (!handle) {
-		/* turn off cursor */
-		dce_v11_0_hide_cursor(crtc);
-		obj = NULL;
-		goto unpin;
-	}
-
-	if ((width > amdgpu_crtc->max_cursor_width) ||
-	    (height > amdgpu_crtc->max_cursor_height)) {
-		DRM_ERROR("bad cursor width or height %d x %d\n", width, height);
-		return -EINVAL;
-	}
-
-	obj = drm_gem_object_lookup(file_priv, handle);
-	if (!obj) {
-		DRM_ERROR("Cannot find cursor object %x for crtc %d\n", handle, amdgpu_crtc->crtc_id);
-		return -ENOENT;
-	}
-
-	aobj = gem_to_amdgpu_bo(obj);
-	ret = amdgpu_bo_reserve(aobj, false);
-	if (ret != 0) {
-		drm_gem_object_put_unlocked(obj);
-		return ret;
-	}
-
-	ret = amdgpu_bo_pin(aobj, AMDGPU_GEM_DOMAIN_VRAM, &amdgpu_crtc->cursor_addr);
-	amdgpu_bo_unreserve(aobj);
-	if (ret) {
-		DRM_ERROR("Failed to pin new cursor BO (%d)\n", ret);
-		drm_gem_object_put_unlocked(obj);
-		return ret;
-	}
-
-	dce_v11_0_lock_cursor(crtc, true);
-
-	if (width != amdgpu_crtc->cursor_width ||
-	    height != amdgpu_crtc->cursor_height ||
-	    hot_x != amdgpu_crtc->cursor_hot_x ||
-	    hot_y != amdgpu_crtc->cursor_hot_y) {
-		int x, y;
-
-		x = amdgpu_crtc->cursor_x + amdgpu_crtc->cursor_hot_x - hot_x;
-		y = amdgpu_crtc->cursor_y + amdgpu_crtc->cursor_hot_y - hot_y;
-
-		dce_v11_0_cursor_move_locked(crtc, x, y);
-
-		amdgpu_crtc->cursor_width = width;
-		amdgpu_crtc->cursor_height = height;
-		amdgpu_crtc->cursor_hot_x = hot_x;
-		amdgpu_crtc->cursor_hot_y = hot_y;
-	}
-
-	dce_v11_0_show_cursor(crtc);
-	dce_v11_0_lock_cursor(crtc, false);
-
-unpin:
-	if (amdgpu_crtc->cursor_bo) {
-		struct amdgpu_bo *aobj = gem_to_amdgpu_bo(amdgpu_crtc->cursor_bo);
-		ret = amdgpu_bo_reserve(aobj, true);
-		if (likely(ret == 0)) {
-			amdgpu_bo_unpin(aobj);
-			amdgpu_bo_unreserve(aobj);
-		}
-		drm_gem_object_put_unlocked(amdgpu_crtc->cursor_bo);
-	}
-
-	amdgpu_crtc->cursor_bo = obj;
-	return 0;
-}
-
-static void dce_v11_0_cursor_reset(struct drm_crtc *crtc)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-
-	if (amdgpu_crtc->cursor_bo) {
-		dce_v11_0_lock_cursor(crtc, true);
-
-		dce_v11_0_cursor_move_locked(crtc, amdgpu_crtc->cursor_x,
-					     amdgpu_crtc->cursor_y);
-
-		dce_v11_0_show_cursor(crtc);
-
-		dce_v11_0_lock_cursor(crtc, false);
-	}
-}
-
-static int dce_v11_0_crtc_gamma_set(struct drm_crtc *crtc, u16 *red, u16 *green,
-				    u16 *blue, uint32_t size,
-				    struct drm_modeset_acquire_ctx *ctx)
-{
-	dce_v11_0_crtc_load_lut(crtc);
-
-	return 0;
-}
-
-static void dce_v11_0_crtc_destroy(struct drm_crtc *crtc)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-
-	drm_crtc_cleanup(crtc);
-	kfree(amdgpu_crtc);
-}
-
-static const struct drm_crtc_funcs dce_v11_0_crtc_funcs = {
-	.cursor_set2 = dce_v11_0_crtc_cursor_set2,
-	.cursor_move = dce_v11_0_crtc_cursor_move,
-	.gamma_set = dce_v11_0_crtc_gamma_set,
-	.set_config = amdgpu_crtc_set_config,
-	.destroy = dce_v11_0_crtc_destroy,
-	.page_flip_target = amdgpu_crtc_page_flip_target,
-};
-
-static void dce_v11_0_crtc_dpms(struct drm_crtc *crtc, int mode)
-{
-	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	unsigned type;
-
-	switch (mode) {
-	case DRM_MODE_DPMS_ON:
-		amdgpu_crtc->enabled = true;
-		amdgpu_atombios_crtc_enable(crtc, ATOM_ENABLE);
-		dce_v11_0_vga_enable(crtc, true);
-		amdgpu_atombios_crtc_blank(crtc, ATOM_DISABLE);
-		dce_v11_0_vga_enable(crtc, false);
-		/* Make sure VBLANK and PFLIP interrupts are still enabled */
-		type = amdgpu_crtc_idx_to_irq_type(adev, amdgpu_crtc->crtc_id);
-		amdgpu_irq_update(adev, &adev->crtc_irq, type);
-		amdgpu_irq_update(adev, &adev->pageflip_irq, type);
-		drm_crtc_vblank_on(crtc);
-		dce_v11_0_crtc_load_lut(crtc);
-		break;
-	case DRM_MODE_DPMS_STANDBY:
-	case DRM_MODE_DPMS_SUSPEND:
-	case DRM_MODE_DPMS_OFF:
-		drm_crtc_vblank_off(crtc);
-		if (amdgpu_crtc->enabled) {
-			dce_v11_0_vga_enable(crtc, true);
-			amdgpu_atombios_crtc_blank(crtc, ATOM_ENABLE);
-			dce_v11_0_vga_enable(crtc, false);
-		}
-		amdgpu_atombios_crtc_enable(crtc, ATOM_DISABLE);
-		amdgpu_crtc->enabled = false;
-		break;
-	}
-	/* adjust pm to dpms */
-	amdgpu_pm_compute_clocks(adev);
-}
-
-static void dce_v11_0_crtc_prepare(struct drm_crtc *crtc)
-{
-	/* disable crtc pair power gating before programming */
-	amdgpu_atombios_crtc_powergate(crtc, ATOM_DISABLE);
-	amdgpu_atombios_crtc_lock(crtc, ATOM_ENABLE);
-	dce_v11_0_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
-}
-
-static void dce_v11_0_crtc_commit(struct drm_crtc *crtc)
-{
-	dce_v11_0_crtc_dpms(crtc, DRM_MODE_DPMS_ON);
-	amdgpu_atombios_crtc_lock(crtc, ATOM_DISABLE);
-}
-
-static void dce_v11_0_crtc_disable(struct drm_crtc *crtc)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-	struct amdgpu_atom_ss ss;
-	int i;
-
-	dce_v11_0_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
-	if (crtc->primary->fb) {
-		int r;
-		struct amdgpu_framebuffer *amdgpu_fb;
-		struct amdgpu_bo *abo;
-
-		amdgpu_fb = to_amdgpu_framebuffer(crtc->primary->fb);
-		abo = gem_to_amdgpu_bo(amdgpu_fb->obj);
-		r = amdgpu_bo_reserve(abo, true);
-		if (unlikely(r))
-			DRM_ERROR("failed to reserve abo before unpin\n");
-		else {
-			amdgpu_bo_unpin(abo);
-			amdgpu_bo_unreserve(abo);
-		}
-	}
-	/* disable the GRPH */
-	dce_v11_0_grph_enable(crtc, false);
-
-	amdgpu_atombios_crtc_powergate(crtc, ATOM_ENABLE);
-
-	for (i = 0; i < adev->mode_info.num_crtc; i++) {
-		if (adev->mode_info.crtcs[i] &&
-		    adev->mode_info.crtcs[i]->enabled &&
-		    i != amdgpu_crtc->crtc_id &&
-		    amdgpu_crtc->pll_id == adev->mode_info.crtcs[i]->pll_id) {
-			/* one other crtc is using this pll don't turn
-			 * off the pll
-			 */
-			goto done;
-		}
-	}
-
-	switch (amdgpu_crtc->pll_id) {
-	case ATOM_PPLL0:
-	case ATOM_PPLL1:
-	case ATOM_PPLL2:
-		/* disable the ppll */
-		amdgpu_atombios_crtc_program_pll(crtc, amdgpu_crtc->crtc_id, amdgpu_crtc->pll_id,
-						 0, 0, ATOM_DISABLE, 0, 0, 0, 0, 0, false, &ss);
-		break;
-	case ATOM_COMBOPHY_PLL0:
-	case ATOM_COMBOPHY_PLL1:
-	case ATOM_COMBOPHY_PLL2:
-	case ATOM_COMBOPHY_PLL3:
-	case ATOM_COMBOPHY_PLL4:
-	case ATOM_COMBOPHY_PLL5:
-		/* disable the ppll */
-		amdgpu_atombios_crtc_program_pll(crtc, ATOM_CRTC_INVALID, amdgpu_crtc->pll_id,
-						 0, 0, ATOM_DISABLE, 0, 0, 0, 0, 0, false, &ss);
-		break;
-	default:
-		break;
-	}
-done:
-	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
-	amdgpu_crtc->adjusted_clock = 0;
-	amdgpu_crtc->encoder = NULL;
-	amdgpu_crtc->connector = NULL;
-}
-
-static int dce_v11_0_crtc_mode_set(struct drm_crtc *crtc,
-				  struct drm_display_mode *mode,
-				  struct drm_display_mode *adjusted_mode,
-				  int x, int y, struct drm_framebuffer *old_fb)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-
-	if (!amdgpu_crtc->adjusted_clock)
-		return -EINVAL;
-
-	if ((adev->asic_type == CHIP_POLARIS10) ||
-	    (adev->asic_type == CHIP_POLARIS11) ||
-	    (adev->asic_type == CHIP_POLARIS12)) {
-		struct amdgpu_encoder *amdgpu_encoder =
-			to_amdgpu_encoder(amdgpu_crtc->encoder);
-		int encoder_mode =
-			amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder);
-
-		/* SetPixelClock calculates the plls and ss values now */
-		amdgpu_atombios_crtc_program_pll(crtc, amdgpu_crtc->crtc_id,
-						 amdgpu_crtc->pll_id,
-						 encoder_mode, amdgpu_encoder->encoder_id,
-						 adjusted_mode->clock, 0, 0, 0, 0,
-						 amdgpu_crtc->bpc, amdgpu_crtc->ss_enabled, &amdgpu_crtc->ss);
-	} else {
-		amdgpu_atombios_crtc_set_pll(crtc, adjusted_mode);
-	}
-	amdgpu_atombios_crtc_set_dtd_timing(crtc, adjusted_mode);
-	dce_v11_0_crtc_do_set_base(crtc, old_fb, x, y, 0);
-	amdgpu_atombios_crtc_overscan_setup(crtc, mode, adjusted_mode);
-	amdgpu_atombios_crtc_scaler_setup(crtc);
-	dce_v11_0_cursor_reset(crtc);
-	/* update the hw version fpr dpm */
-	amdgpu_crtc->hw_mode = *adjusted_mode;
-
-	return 0;
-}
-
-static bool dce_v11_0_crtc_mode_fixup(struct drm_crtc *crtc,
-				     const struct drm_display_mode *mode,
-				     struct drm_display_mode *adjusted_mode)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct drm_device *dev = crtc->dev;
-	struct drm_encoder *encoder;
-
-	/* assign the encoder to the amdgpu crtc to avoid repeated lookups later */
-	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
-		if (encoder->crtc == crtc) {
-			amdgpu_crtc->encoder = encoder;
-			amdgpu_crtc->connector = amdgpu_get_connector_for_encoder(encoder);
-			break;
-		}
-	}
-	if ((amdgpu_crtc->encoder == NULL) || (amdgpu_crtc->connector == NULL)) {
-		amdgpu_crtc->encoder = NULL;
-		amdgpu_crtc->connector = NULL;
-		return false;
-	}
-	if (!amdgpu_crtc_scaling_mode_fixup(crtc, mode, adjusted_mode))
-		return false;
-	if (amdgpu_atombios_crtc_prepare_pll(crtc, adjusted_mode))
-		return false;
-	/* pick pll */
-	amdgpu_crtc->pll_id = dce_v11_0_pick_pll(crtc);
-	/* if we can't get a PPLL for a non-DP encoder, fail */
-	if ((amdgpu_crtc->pll_id == ATOM_PPLL_INVALID) &&
-	    !ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder)))
-		return false;
-
-	return true;
-}
-
-static int dce_v11_0_crtc_set_base(struct drm_crtc *crtc, int x, int y,
-				  struct drm_framebuffer *old_fb)
-{
-	return dce_v11_0_crtc_do_set_base(crtc, old_fb, x, y, 0);
-}
-
-static int dce_v11_0_crtc_set_base_atomic(struct drm_crtc *crtc,
-					 struct drm_framebuffer *fb,
-					 int x, int y, enum mode_set_atomic state)
-{
-       return dce_v11_0_crtc_do_set_base(crtc, fb, x, y, 1);
-}
-
-static const struct drm_crtc_helper_funcs dce_v11_0_crtc_helper_funcs = {
-	.dpms = dce_v11_0_crtc_dpms,
-	.mode_fixup = dce_v11_0_crtc_mode_fixup,
-	.mode_set = dce_v11_0_crtc_mode_set,
-	.mode_set_base = dce_v11_0_crtc_set_base,
-	.mode_set_base_atomic = dce_v11_0_crtc_set_base_atomic,
-	.prepare = dce_v11_0_crtc_prepare,
-	.commit = dce_v11_0_crtc_commit,
-	.disable = dce_v11_0_crtc_disable,
-};
-
-static int dce_v11_0_crtc_init(struct amdgpu_device *adev, int index)
-{
-	struct amdgpu_crtc *amdgpu_crtc;
-
-	amdgpu_crtc = kzalloc(sizeof(struct amdgpu_crtc) +
-			      (AMDGPUFB_CONN_LIMIT * sizeof(struct drm_connector *)), GFP_KERNEL);
-	if (amdgpu_crtc == NULL)
-		return -ENOMEM;
-
-	drm_crtc_init(adev->ddev, &amdgpu_crtc->base, &dce_v11_0_crtc_funcs);
-
-	drm_mode_crtc_set_gamma_size(&amdgpu_crtc->base, 256);
-	amdgpu_crtc->crtc_id = index;
-	adev->mode_info.crtcs[index] = amdgpu_crtc;
-
-	amdgpu_crtc->max_cursor_width = 128;
-	amdgpu_crtc->max_cursor_height = 128;
-	adev->ddev->mode_config.cursor_width = amdgpu_crtc->max_cursor_width;
-	adev->ddev->mode_config.cursor_height = amdgpu_crtc->max_cursor_height;
-
-	switch (amdgpu_crtc->crtc_id) {
-	case 0:
-	default:
-		amdgpu_crtc->crtc_offset = CRTC0_REGISTER_OFFSET;
-		break;
-	case 1:
-		amdgpu_crtc->crtc_offset = CRTC1_REGISTER_OFFSET;
-		break;
-	case 2:
-		amdgpu_crtc->crtc_offset = CRTC2_REGISTER_OFFSET;
-		break;
-	case 3:
-		amdgpu_crtc->crtc_offset = CRTC3_REGISTER_OFFSET;
-		break;
-	case 4:
-		amdgpu_crtc->crtc_offset = CRTC4_REGISTER_OFFSET;
-		break;
-	case 5:
-		amdgpu_crtc->crtc_offset = CRTC5_REGISTER_OFFSET;
-		break;
-	}
-
-	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
-	amdgpu_crtc->adjusted_clock = 0;
-	amdgpu_crtc->encoder = NULL;
-	amdgpu_crtc->connector = NULL;
-	drm_crtc_helper_add(&amdgpu_crtc->base, &dce_v11_0_crtc_helper_funcs);
-
-	return 0;
-}
-
-static int dce_v11_0_early_init(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	adev->audio_endpt_rreg = &dce_v11_0_audio_endpt_rreg;
-	adev->audio_endpt_wreg = &dce_v11_0_audio_endpt_wreg;
-
-	dce_v11_0_set_display_funcs(adev);
-	dce_v11_0_set_irq_funcs(adev);
-
-	adev->mode_info.num_crtc = dce_v11_0_get_num_crtc(adev);
-
-	switch (adev->asic_type) {
-	case CHIP_CARRIZO:
-		adev->mode_info.num_hpd = 6;
-		adev->mode_info.num_dig = 9;
-		break;
-	case CHIP_STONEY:
-		adev->mode_info.num_hpd = 6;
-		adev->mode_info.num_dig = 9;
-		break;
-	case CHIP_POLARIS10:
-		adev->mode_info.num_hpd = 6;
-		adev->mode_info.num_dig = 6;
-		break;
-	case CHIP_POLARIS11:
-	case CHIP_POLARIS12:
-		adev->mode_info.num_hpd = 5;
-		adev->mode_info.num_dig = 5;
-		break;
-	default:
-		/* FIXME: not supported yet */
-		return -EINVAL;
-	}
-
-	return 0;
-}
-
-static int dce_v11_0_sw_init(void *handle)
-{
-	int r, i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	for (i = 0; i < adev->mode_info.num_crtc; i++) {
-		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i + 1, &adev->crtc_irq);
-		if (r)
-			return r;
-	}
-
-	for (i = 8; i < 20; i += 2) {
-		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i, &adev->pageflip_irq);
-		if (r)
-			return r;
-	}
-
-	/* HPD hotplug */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 42, &adev->hpd_irq);
-	if (r)
-		return r;
-
-	adev->ddev->mode_config.funcs = &amdgpu_mode_funcs;
-
-	adev->ddev->mode_config.async_page_flip = true;
-
-	adev->ddev->mode_config.max_width = 16384;
-	adev->ddev->mode_config.max_height = 16384;
-
-	adev->ddev->mode_config.preferred_depth = 24;
-	adev->ddev->mode_config.prefer_shadow = 1;
-
-	adev->ddev->mode_config.fb_base = adev->mc.aper_base;
-
-	r = amdgpu_modeset_create_props(adev);
-	if (r)
-		return r;
-
-	adev->ddev->mode_config.max_width = 16384;
-	adev->ddev->mode_config.max_height = 16384;
-
-
-	/* allocate crtcs */
-	for (i = 0; i < adev->mode_info.num_crtc; i++) {
-		r = dce_v11_0_crtc_init(adev, i);
-		if (r)
-			return r;
-	}
-
-	if (amdgpu_atombios_get_connector_info_from_object_table(adev))
-		amdgpu_print_display_setup(adev->ddev);
-	else
-		return -EINVAL;
-
-	/* setup afmt */
-	r = dce_v11_0_afmt_init(adev);
-	if (r)
-		return r;
-
-	r = dce_v11_0_audio_init(adev);
-	if (r)
-		return r;
-
-	drm_kms_helper_poll_init(adev->ddev);
-
-	adev->mode_info.mode_config_initialized = true;
-	return 0;
-}
-
-static int dce_v11_0_sw_fini(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	kfree(adev->mode_info.bios_hardcoded_edid);
-
-	drm_kms_helper_poll_fini(adev->ddev);
-
-	dce_v11_0_audio_fini(adev);
-
-	dce_v11_0_afmt_fini(adev);
-
-	drm_mode_config_cleanup(adev->ddev);
-	adev->mode_info.mode_config_initialized = false;
-
-	return 0;
-}
-
-static int dce_v11_0_hw_init(void *handle)
-{
-	int i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	dce_v11_0_init_golden_registers(adev);
-
-	/* disable vga render */
-	dce_v11_0_set_vga_render_state(adev, false);
-	/* init dig PHYs, disp eng pll */
-	amdgpu_atombios_crtc_powergate_init(adev);
-	amdgpu_atombios_encoder_init_dig(adev);
-	if ((adev->asic_type == CHIP_POLARIS10) ||
-	    (adev->asic_type == CHIP_POLARIS11) ||
-	    (adev->asic_type == CHIP_POLARIS12)) {
-		amdgpu_atombios_crtc_set_dce_clock(adev, adev->clock.default_dispclk,
-						   DCE_CLOCK_TYPE_DISPCLK, ATOM_GCK_DFS);
-		amdgpu_atombios_crtc_set_dce_clock(adev, 0,
-						   DCE_CLOCK_TYPE_DPREFCLK, ATOM_GCK_DFS);
-	} else {
-		amdgpu_atombios_crtc_set_disp_eng_pll(adev, adev->clock.default_dispclk);
-	}
-
-	/* initialize hpd */
-	dce_v11_0_hpd_init(adev);
-
-	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
-		dce_v11_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
-	}
-
-	dce_v11_0_pageflip_interrupt_init(adev);
-
-	return 0;
-}
-
-static int dce_v11_0_hw_fini(void *handle)
-{
-	int i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	dce_v11_0_hpd_fini(adev);
-
-	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
-		dce_v11_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
-	}
-
-	dce_v11_0_pageflip_interrupt_fini(adev);
-
-	return 0;
-}
-
-static int dce_v11_0_suspend(void *handle)
-{
-	return dce_v11_0_hw_fini(handle);
-}
-
-static int dce_v11_0_resume(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	int ret;
-
-	ret = dce_v11_0_hw_init(handle);
-
-	/* turn on the BL */
-	if (adev->mode_info.bl_encoder) {
-		u8 bl_level = amdgpu_display_backlight_get_level(adev,
-								  adev->mode_info.bl_encoder);
-		amdgpu_display_backlight_set_level(adev, adev->mode_info.bl_encoder,
-						    bl_level);
-	}
-
-	return ret;
-}
-
-static bool dce_v11_0_is_idle(void *handle)
-{
-	return true;
-}
-
-static int dce_v11_0_wait_for_idle(void *handle)
-{
-	return 0;
-}
-
-static int dce_v11_0_soft_reset(void *handle)
-{
-	u32 srbm_soft_reset = 0, tmp;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (dce_v11_0_is_display_hung(adev))
-		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_DC_MASK;
-
-	if (srbm_soft_reset) {
-		tmp = RREG32(mmSRBM_SOFT_RESET);
-		tmp |= srbm_soft_reset;
-		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
-		WREG32(mmSRBM_SOFT_RESET, tmp);
-		tmp = RREG32(mmSRBM_SOFT_RESET);
-
-		udelay(50);
-
-		tmp &= ~srbm_soft_reset;
-		WREG32(mmSRBM_SOFT_RESET, tmp);
-		tmp = RREG32(mmSRBM_SOFT_RESET);
-
-		/* Wait a little for things to settle down */
-		udelay(50);
-	}
-	return 0;
-}
-
-static void dce_v11_0_set_crtc_vblank_interrupt_state(struct amdgpu_device *adev,
-						     int crtc,
-						     enum amdgpu_interrupt_state state)
-{
-	u32 lb_interrupt_mask;
-
-	if (crtc >= adev->mode_info.num_crtc) {
-		DRM_DEBUG("invalid crtc %d\n", crtc);
-		return;
-	}
-
-	switch (state) {
-	case AMDGPU_IRQ_STATE_DISABLE:
-		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
-		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
-						  VBLANK_INTERRUPT_MASK, 0);
-		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
-		break;
-	case AMDGPU_IRQ_STATE_ENABLE:
-		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
-		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
-						  VBLANK_INTERRUPT_MASK, 1);
-		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
-		break;
-	default:
-		break;
-	}
-}
-
-static void dce_v11_0_set_crtc_vline_interrupt_state(struct amdgpu_device *adev,
-						    int crtc,
-						    enum amdgpu_interrupt_state state)
-{
-	u32 lb_interrupt_mask;
-
-	if (crtc >= adev->mode_info.num_crtc) {
-		DRM_DEBUG("invalid crtc %d\n", crtc);
-		return;
-	}
-
-	switch (state) {
-	case AMDGPU_IRQ_STATE_DISABLE:
-		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
-		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
-						  VLINE_INTERRUPT_MASK, 0);
-		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
-		break;
-	case AMDGPU_IRQ_STATE_ENABLE:
-		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
-		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
-						  VLINE_INTERRUPT_MASK, 1);
-		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
-		break;
-	default:
-		break;
-	}
-}
-
-static int dce_v11_0_set_hpd_irq_state(struct amdgpu_device *adev,
-					struct amdgpu_irq_src *source,
-					unsigned hpd,
-					enum amdgpu_interrupt_state state)
-{
-	u32 tmp;
-
-	if (hpd >= adev->mode_info.num_hpd) {
-		DRM_DEBUG("invalid hdp %d\n", hpd);
-		return 0;
-	}
-
-	switch (state) {
-	case AMDGPU_IRQ_STATE_DISABLE:
-		tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd]);
-		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_EN, 0);
-		WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd], tmp);
-		break;
-	case AMDGPU_IRQ_STATE_ENABLE:
-		tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd]);
-		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_EN, 1);
-		WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd], tmp);
-		break;
-	default:
-		break;
-	}
-
-	return 0;
-}
-
-static int dce_v11_0_set_crtc_irq_state(struct amdgpu_device *adev,
-					struct amdgpu_irq_src *source,
-					unsigned type,
-					enum amdgpu_interrupt_state state)
-{
-	switch (type) {
-	case AMDGPU_CRTC_IRQ_VBLANK1:
-		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 0, state);
-		break;
-	case AMDGPU_CRTC_IRQ_VBLANK2:
-		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 1, state);
-		break;
-	case AMDGPU_CRTC_IRQ_VBLANK3:
-		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 2, state);
-		break;
-	case AMDGPU_CRTC_IRQ_VBLANK4:
-		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 3, state);
-		break;
-	case AMDGPU_CRTC_IRQ_VBLANK5:
-		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 4, state);
-		break;
-	case AMDGPU_CRTC_IRQ_VBLANK6:
-		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 5, state);
-		break;
-	case AMDGPU_CRTC_IRQ_VLINE1:
-		dce_v11_0_set_crtc_vline_interrupt_state(adev, 0, state);
-		break;
-	case AMDGPU_CRTC_IRQ_VLINE2:
-		dce_v11_0_set_crtc_vline_interrupt_state(adev, 1, state);
-		break;
-	case AMDGPU_CRTC_IRQ_VLINE3:
-		dce_v11_0_set_crtc_vline_interrupt_state(adev, 2, state);
-		break;
-	case AMDGPU_CRTC_IRQ_VLINE4:
-		dce_v11_0_set_crtc_vline_interrupt_state(adev, 3, state);
-		break;
-	case AMDGPU_CRTC_IRQ_VLINE5:
-		dce_v11_0_set_crtc_vline_interrupt_state(adev, 4, state);
-		break;
-	 case AMDGPU_CRTC_IRQ_VLINE6:
-		dce_v11_0_set_crtc_vline_interrupt_state(adev, 5, state);
-		break;
-	default:
-		break;
-	}
-	return 0;
-}
-
-static int dce_v11_0_set_pageflip_irq_state(struct amdgpu_device *adev,
-					    struct amdgpu_irq_src *src,
-					    unsigned type,
-					    enum amdgpu_interrupt_state state)
-{
-	u32 reg;
-
-	if (type >= adev->mode_info.num_crtc) {
-		DRM_ERROR("invalid pageflip crtc %d\n", type);
-		return -EINVAL;
-	}
-
-	reg = RREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type]);
-	if (state == AMDGPU_IRQ_STATE_DISABLE)
-		WREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type],
-		       reg & ~GRPH_INTERRUPT_CONTROL__GRPH_PFLIP_INT_MASK_MASK);
-	else
-		WREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type],
-		       reg | GRPH_INTERRUPT_CONTROL__GRPH_PFLIP_INT_MASK_MASK);
-
-	return 0;
-}
-
-static int dce_v11_0_pageflip_irq(struct amdgpu_device *adev,
-				  struct amdgpu_irq_src *source,
-				  struct amdgpu_iv_entry *entry)
-{
-	unsigned long flags;
-	unsigned crtc_id;
-	struct amdgpu_crtc *amdgpu_crtc;
-	struct amdgpu_flip_work *works;
-
-	crtc_id = (entry->src_id - 8) >> 1;
-	amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
-
-	if (crtc_id >= adev->mode_info.num_crtc) {
-		DRM_ERROR("invalid pageflip crtc %d\n", crtc_id);
-		return -EINVAL;
-	}
-
-	if (RREG32(mmGRPH_INTERRUPT_STATUS + crtc_offsets[crtc_id]) &
-	    GRPH_INTERRUPT_STATUS__GRPH_PFLIP_INT_OCCURRED_MASK)
-		WREG32(mmGRPH_INTERRUPT_STATUS + crtc_offsets[crtc_id],
-		       GRPH_INTERRUPT_STATUS__GRPH_PFLIP_INT_CLEAR_MASK);
-
-	/* IRQ could occur when in initial stage */
-	if(amdgpu_crtc == NULL)
-		return 0;
-
-	spin_lock_irqsave(&adev->ddev->event_lock, flags);
-	works = amdgpu_crtc->pflip_works;
-	if (amdgpu_crtc->pflip_status != AMDGPU_FLIP_SUBMITTED){
-		DRM_DEBUG_DRIVER("amdgpu_crtc->pflip_status = %d != "
-						 "AMDGPU_FLIP_SUBMITTED(%d)\n",
-						 amdgpu_crtc->pflip_status,
-						 AMDGPU_FLIP_SUBMITTED);
-		spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
-		return 0;
-	}
-
-	/* page flip completed. clean up */
-	amdgpu_crtc->pflip_status = AMDGPU_FLIP_NONE;
-	amdgpu_crtc->pflip_works = NULL;
-
-	/* wakeup usersapce */
-	if(works->event)
-		drm_crtc_send_vblank_event(&amdgpu_crtc->base, works->event);
-
-	spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
-
-	drm_crtc_vblank_put(&amdgpu_crtc->base);
-	schedule_work(&works->unpin_work);
-
-	return 0;
-}
-
-static void dce_v11_0_hpd_int_ack(struct amdgpu_device *adev,
-				  int hpd)
-{
-	u32 tmp;
-
-	if (hpd >= adev->mode_info.num_hpd) {
-		DRM_DEBUG("invalid hdp %d\n", hpd);
-		return;
-	}
-
-	tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd]);
-	tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_ACK, 1);
-	WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd], tmp);
-}
-
-static void dce_v11_0_crtc_vblank_int_ack(struct amdgpu_device *adev,
-					  int crtc)
-{
-	u32 tmp;
-
-	if (crtc < 0 || crtc >= adev->mode_info.num_crtc) {
-		DRM_DEBUG("invalid crtc %d\n", crtc);
-		return;
-	}
-
-	tmp = RREG32(mmLB_VBLANK_STATUS + crtc_offsets[crtc]);
-	tmp = REG_SET_FIELD(tmp, LB_VBLANK_STATUS, VBLANK_ACK, 1);
-	WREG32(mmLB_VBLANK_STATUS + crtc_offsets[crtc], tmp);
-}
-
-static void dce_v11_0_crtc_vline_int_ack(struct amdgpu_device *adev,
-					 int crtc)
-{
-	u32 tmp;
-
-	if (crtc < 0 || crtc >= adev->mode_info.num_crtc) {
-		DRM_DEBUG("invalid crtc %d\n", crtc);
-		return;
-	}
-
-	tmp = RREG32(mmLB_VLINE_STATUS + crtc_offsets[crtc]);
-	tmp = REG_SET_FIELD(tmp, LB_VLINE_STATUS, VLINE_ACK, 1);
-	WREG32(mmLB_VLINE_STATUS + crtc_offsets[crtc], tmp);
-}
-
-static int dce_v11_0_crtc_irq(struct amdgpu_device *adev,
-				struct amdgpu_irq_src *source,
-				struct amdgpu_iv_entry *entry)
-{
-	unsigned crtc = entry->src_id - 1;
-	uint32_t disp_int = RREG32(interrupt_status_offsets[crtc].reg);
-	unsigned irq_type = amdgpu_crtc_idx_to_irq_type(adev, crtc);
-
-	switch (entry->src_data[0]) {
-	case 0: /* vblank */
-		if (disp_int & interrupt_status_offsets[crtc].vblank)
-			dce_v11_0_crtc_vblank_int_ack(adev, crtc);
-		else
-			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
-
-		if (amdgpu_irq_enabled(adev, source, irq_type)) {
-			drm_handle_vblank(adev->ddev, crtc);
-		}
-		DRM_DEBUG("IH: D%d vblank\n", crtc + 1);
-
-		break;
-	case 1: /* vline */
-		if (disp_int & interrupt_status_offsets[crtc].vline)
-			dce_v11_0_crtc_vline_int_ack(adev, crtc);
-		else
-			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
-
-		DRM_DEBUG("IH: D%d vline\n", crtc + 1);
-
-		break;
-	default:
-		DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data[0]);
-		break;
-	}
-
-	return 0;
-}
-
-static int dce_v11_0_hpd_irq(struct amdgpu_device *adev,
-			     struct amdgpu_irq_src *source,
-			     struct amdgpu_iv_entry *entry)
-{
-	uint32_t disp_int, mask;
-	unsigned hpd;
-
-	if (entry->src_data[0] >= adev->mode_info.num_hpd) {
-		DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data[0]);
-		return 0;
-	}
-
-	hpd = entry->src_data[0];
-	disp_int = RREG32(interrupt_status_offsets[hpd].reg);
-	mask = interrupt_status_offsets[hpd].hpd;
-
-	if (disp_int & mask) {
-		dce_v11_0_hpd_int_ack(adev, hpd);
-		schedule_work(&adev->hotplug_work);
-		DRM_DEBUG("IH: HPD%d\n", hpd + 1);
-	}
-
-	return 0;
-}
-
-static int dce_v11_0_set_clockgating_state(void *handle,
-					  enum amd_clockgating_state state)
-{
-	return 0;
-}
-
-static int dce_v11_0_set_powergating_state(void *handle,
-					  enum amd_powergating_state state)
-{
-	return 0;
-}
-
-static const struct amd_ip_funcs dce_v11_0_ip_funcs = {
-	.name = "dce_v11_0",
-	.early_init = dce_v11_0_early_init,
-	.late_init = NULL,
-	.sw_init = dce_v11_0_sw_init,
-	.sw_fini = dce_v11_0_sw_fini,
-	.hw_init = dce_v11_0_hw_init,
-	.hw_fini = dce_v11_0_hw_fini,
-	.suspend = dce_v11_0_suspend,
-	.resume = dce_v11_0_resume,
-	.is_idle = dce_v11_0_is_idle,
-	.wait_for_idle = dce_v11_0_wait_for_idle,
-	.soft_reset = dce_v11_0_soft_reset,
-	.set_clockgating_state = dce_v11_0_set_clockgating_state,
-	.set_powergating_state = dce_v11_0_set_powergating_state,
-};
-
-static void
-dce_v11_0_encoder_mode_set(struct drm_encoder *encoder,
-			  struct drm_display_mode *mode,
-			  struct drm_display_mode *adjusted_mode)
-{
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-
-	amdgpu_encoder->pixel_clock = adjusted_mode->clock;
-
-	/* need to call this here rather than in prepare() since we need some crtc info */
-	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_OFF);
-
-	/* set scaler clears this on some chips */
-	dce_v11_0_set_interleave(encoder->crtc, mode);
-
-	if (amdgpu_atombios_encoder_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_HDMI) {
-		dce_v11_0_afmt_enable(encoder, true);
-		dce_v11_0_afmt_setmode(encoder, adjusted_mode);
-	}
-}
-
-static void dce_v11_0_encoder_prepare(struct drm_encoder *encoder)
-{
-	struct amdgpu_device *adev = encoder->dev->dev_private;
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
-
-	if ((amdgpu_encoder->active_device &
-	     (ATOM_DEVICE_DFP_SUPPORT | ATOM_DEVICE_LCD_SUPPORT)) ||
-	    (amdgpu_encoder_get_dp_bridge_encoder_id(encoder) !=
-	     ENCODER_OBJECT_ID_NONE)) {
-		struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-		if (dig) {
-			dig->dig_encoder = dce_v11_0_pick_dig_encoder(encoder);
-			if (amdgpu_encoder->active_device & ATOM_DEVICE_DFP_SUPPORT)
-				dig->afmt = adev->mode_info.afmt[dig->dig_encoder];
-		}
-	}
-
-	amdgpu_atombios_scratch_regs_lock(adev, true);
-
-	if (connector) {
-		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
-
-		/* select the clock/data port if it uses a router */
-		if (amdgpu_connector->router.cd_valid)
-			amdgpu_i2c_router_select_cd_port(amdgpu_connector);
-
-		/* turn eDP panel on for mode set */
-		if (connector->connector_type == DRM_MODE_CONNECTOR_eDP)
-			amdgpu_atombios_encoder_set_edp_panel_power(connector,
-							     ATOM_TRANSMITTER_ACTION_POWER_ON);
-	}
-
-	/* this is needed for the pll/ss setup to work correctly in some cases */
-	amdgpu_atombios_encoder_set_crtc_source(encoder);
-	/* set up the FMT blocks */
-	dce_v11_0_program_fmt(encoder);
-}
-
-static void dce_v11_0_encoder_commit(struct drm_encoder *encoder)
-{
-	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-
-	/* need to call this here as we need the crtc set up */
-	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_ON);
-	amdgpu_atombios_scratch_regs_lock(adev, false);
-}
-
-static void dce_v11_0_encoder_disable(struct drm_encoder *encoder)
-{
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig;
-
-	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_OFF);
-
-	if (amdgpu_atombios_encoder_is_digital(encoder)) {
-		if (amdgpu_atombios_encoder_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_HDMI)
-			dce_v11_0_afmt_enable(encoder, false);
-		dig = amdgpu_encoder->enc_priv;
-		dig->dig_encoder = -1;
-	}
-	amdgpu_encoder->active_device = 0;
-}
-
-/* these are handled by the primary encoders */
-static void dce_v11_0_ext_prepare(struct drm_encoder *encoder)
-{
-
-}
-
-static void dce_v11_0_ext_commit(struct drm_encoder *encoder)
-{
-
-}
-
-static void
-dce_v11_0_ext_mode_set(struct drm_encoder *encoder,
-		      struct drm_display_mode *mode,
-		      struct drm_display_mode *adjusted_mode)
-{
-
-}
-
-static void dce_v11_0_ext_disable(struct drm_encoder *encoder)
-{
-
-}
-
-static void
-dce_v11_0_ext_dpms(struct drm_encoder *encoder, int mode)
-{
-
-}
-
-static const struct drm_encoder_helper_funcs dce_v11_0_ext_helper_funcs = {
-	.dpms = dce_v11_0_ext_dpms,
-	.prepare = dce_v11_0_ext_prepare,
-	.mode_set = dce_v11_0_ext_mode_set,
-	.commit = dce_v11_0_ext_commit,
-	.disable = dce_v11_0_ext_disable,
-	/* no detect for TMDS/LVDS yet */
-};
-
-static const struct drm_encoder_helper_funcs dce_v11_0_dig_helper_funcs = {
-	.dpms = amdgpu_atombios_encoder_dpms,
-	.mode_fixup = amdgpu_atombios_encoder_mode_fixup,
-	.prepare = dce_v11_0_encoder_prepare,
-	.mode_set = dce_v11_0_encoder_mode_set,
-	.commit = dce_v11_0_encoder_commit,
-	.disable = dce_v11_0_encoder_disable,
-	.detect = amdgpu_atombios_encoder_dig_detect,
-};
-
-static const struct drm_encoder_helper_funcs dce_v11_0_dac_helper_funcs = {
-	.dpms = amdgpu_atombios_encoder_dpms,
-	.mode_fixup = amdgpu_atombios_encoder_mode_fixup,
-	.prepare = dce_v11_0_encoder_prepare,
-	.mode_set = dce_v11_0_encoder_mode_set,
-	.commit = dce_v11_0_encoder_commit,
-	.detect = amdgpu_atombios_encoder_dac_detect,
-};
-
-static void dce_v11_0_encoder_destroy(struct drm_encoder *encoder)
-{
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
-		amdgpu_atombios_encoder_fini_backlight(amdgpu_encoder);
-	kfree(amdgpu_encoder->enc_priv);
-	drm_encoder_cleanup(encoder);
-	kfree(amdgpu_encoder);
-}
-
-static const struct drm_encoder_funcs dce_v11_0_encoder_funcs = {
-	.destroy = dce_v11_0_encoder_destroy,
-};
-
-static void dce_v11_0_encoder_add(struct amdgpu_device *adev,
-				 uint32_t encoder_enum,
-				 uint32_t supported_device,
-				 u16 caps)
-{
-	struct drm_device *dev = adev->ddev;
-	struct drm_encoder *encoder;
-	struct amdgpu_encoder *amdgpu_encoder;
-
-	/* see if we already added it */
-	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
-		amdgpu_encoder = to_amdgpu_encoder(encoder);
-		if (amdgpu_encoder->encoder_enum == encoder_enum) {
-			amdgpu_encoder->devices |= supported_device;
-			return;
-		}
-
-	}
-
-	/* add a new one */
-	amdgpu_encoder = kzalloc(sizeof(struct amdgpu_encoder), GFP_KERNEL);
-	if (!amdgpu_encoder)
-		return;
-
-	encoder = &amdgpu_encoder->base;
-	switch (adev->mode_info.num_crtc) {
-	case 1:
-		encoder->possible_crtcs = 0x1;
-		break;
-	case 2:
-	default:
-		encoder->possible_crtcs = 0x3;
-		break;
-	case 3:
-		encoder->possible_crtcs = 0x7;
-		break;
-	case 4:
-		encoder->possible_crtcs = 0xf;
-		break;
-	case 5:
-		encoder->possible_crtcs = 0x1f;
-		break;
-	case 6:
-		encoder->possible_crtcs = 0x3f;
-		break;
-	}
-
-	amdgpu_encoder->enc_priv = NULL;
-
-	amdgpu_encoder->encoder_enum = encoder_enum;
-	amdgpu_encoder->encoder_id = (encoder_enum & OBJECT_ID_MASK) >> OBJECT_ID_SHIFT;
-	amdgpu_encoder->devices = supported_device;
-	amdgpu_encoder->rmx_type = RMX_OFF;
-	amdgpu_encoder->underscan_type = UNDERSCAN_OFF;
-	amdgpu_encoder->is_ext_encoder = false;
-	amdgpu_encoder->caps = caps;
-
-	switch (amdgpu_encoder->encoder_id) {
-	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
-	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2:
-		drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
-				 DRM_MODE_ENCODER_DAC, NULL);
-		drm_encoder_helper_add(encoder, &dce_v11_0_dac_helper_funcs);
-		break;
-	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
-	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
-	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
-	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
-	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
-		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) {
-			amdgpu_encoder->rmx_type = RMX_FULL;
-			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
-					 DRM_MODE_ENCODER_LVDS, NULL);
-			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_lcd_info(amdgpu_encoder);
-		} else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT)) {
-			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
-					 DRM_MODE_ENCODER_DAC, NULL);
-			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder);
-		} else {
-			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
-					 DRM_MODE_ENCODER_TMDS, NULL);
-			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder);
-		}
-		drm_encoder_helper_add(encoder, &dce_v11_0_dig_helper_funcs);
-		break;
-	case ENCODER_OBJECT_ID_SI170B:
-	case ENCODER_OBJECT_ID_CH7303:
-	case ENCODER_OBJECT_ID_EXTERNAL_SDVOA:
-	case ENCODER_OBJECT_ID_EXTERNAL_SDVOB:
-	case ENCODER_OBJECT_ID_TITFP513:
-	case ENCODER_OBJECT_ID_VT1623:
-	case ENCODER_OBJECT_ID_HDMI_SI1930:
-	case ENCODER_OBJECT_ID_TRAVIS:
-	case ENCODER_OBJECT_ID_NUTMEG:
-		/* these are handled by the primary encoders */
-		amdgpu_encoder->is_ext_encoder = true;
-		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
-			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
-					 DRM_MODE_ENCODER_LVDS, NULL);
-		else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT))
-			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
-					 DRM_MODE_ENCODER_DAC, NULL);
-		else
-			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
-					 DRM_MODE_ENCODER_TMDS, NULL);
-		drm_encoder_helper_add(encoder, &dce_v11_0_ext_helper_funcs);
-		break;
-	}
-}
-
-static const struct amdgpu_display_funcs dce_v11_0_display_funcs = {
-	.bandwidth_update = &dce_v11_0_bandwidth_update,
-	.vblank_get_counter = &dce_v11_0_vblank_get_counter,
-	.vblank_wait = &dce_v11_0_vblank_wait,
-	.backlight_set_level = &amdgpu_atombios_encoder_set_backlight_level,
-	.backlight_get_level = &amdgpu_atombios_encoder_get_backlight_level,
-	.hpd_sense = &dce_v11_0_hpd_sense,
-	.hpd_set_polarity = &dce_v11_0_hpd_set_polarity,
-	.hpd_get_gpio_reg = &dce_v11_0_hpd_get_gpio_reg,
-	.page_flip = &dce_v11_0_page_flip,
-	.page_flip_get_scanoutpos = &dce_v11_0_crtc_get_scanoutpos,
-	.add_encoder = &dce_v11_0_encoder_add,
-	.add_connector = &amdgpu_connector_add,
-};
-
-static void dce_v11_0_set_display_funcs(struct amdgpu_device *adev)
-{
-	if (adev->mode_info.funcs == NULL)
-		adev->mode_info.funcs = &dce_v11_0_display_funcs;
-}
-
-static const struct amdgpu_irq_src_funcs dce_v11_0_crtc_irq_funcs = {
-	.set = dce_v11_0_set_crtc_irq_state,
-	.process = dce_v11_0_crtc_irq,
-};
-
-static const struct amdgpu_irq_src_funcs dce_v11_0_pageflip_irq_funcs = {
-	.set = dce_v11_0_set_pageflip_irq_state,
-	.process = dce_v11_0_pageflip_irq,
-};
-
-static const struct amdgpu_irq_src_funcs dce_v11_0_hpd_irq_funcs = {
-	.set = dce_v11_0_set_hpd_irq_state,
-	.process = dce_v11_0_hpd_irq,
-};
-
-static void dce_v11_0_set_irq_funcs(struct amdgpu_device *adev)
-{
-	adev->crtc_irq.num_types = AMDGPU_CRTC_IRQ_LAST;
-	adev->crtc_irq.funcs = &dce_v11_0_crtc_irq_funcs;
-
-	adev->pageflip_irq.num_types = AMDGPU_PAGEFLIP_IRQ_LAST;
-	adev->pageflip_irq.funcs = &dce_v11_0_pageflip_irq_funcs;
-
-	adev->hpd_irq.num_types = AMDGPU_HPD_LAST;
-	adev->hpd_irq.funcs = &dce_v11_0_hpd_irq_funcs;
-}
-
-const struct amdgpu_ip_block_version dce_v11_0_ip_block =
-{
-	.type = AMD_IP_BLOCK_TYPE_DCE,
-	.major = 11,
-	.minor = 0,
-	.rev = 0,
-	.funcs = &dce_v11_0_ip_funcs,
-};
-
-const struct amdgpu_ip_block_version dce_v11_2_ip_block =
-{
-	.type = AMD_IP_BLOCK_TYPE_DCE,
-	.major = 11,
-	.minor = 2,
-	.rev = 0,
-	.funcs = &dce_v11_0_ip_funcs,
-};
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.h b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.h
deleted file mode 100644
index 0d878ca3acba..000000000000
--- a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.h
+++ /dev/null
@@ -1,32 +0,0 @@
-/*
- * Copyright 2014 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-
-#ifndef __DCE_V11_0_H__
-#define __DCE_V11_0_H__
-
-extern const struct amdgpu_ip_block_version dce_v11_0_ip_block;
-extern const struct amdgpu_ip_block_version dce_v11_2_ip_block;
-
-void dce_v11_0_disable_dce(struct amdgpu_device *adev);
-
-#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c
index a51e35f824a1..acc887a58518 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c
@@ -20,7 +20,15 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  *
  */
-#include <drm/drmP.h>
+
+#include <linux/pci.h>
+
+#include <drm/drm_edid.h>
+#include <drm/drm_fourcc.h>
+#include <drm/drm_modeset_helper.h>
+#include <drm/drm_modeset_helper_vtables.h>
+#include <drm/drm_vblank.h>
+
 #include "amdgpu.h"
 #include "amdgpu_pm.h"
 #include "amdgpu_i2c.h"
@@ -30,19 +38,27 @@
 #include "atombios_encoders.h"
 #include "amdgpu_pll.h"
 #include "amdgpu_connectors.h"
+#include "amdgpu_display.h"
+
+#include "dce_v6_0.h"
+#include "sid.h"
 
 #include "bif/bif_3_0_d.h"
 #include "bif/bif_3_0_sh_mask.h"
+
 #include "oss/oss_1_0_d.h"
 #include "oss/oss_1_0_sh_mask.h"
+
 #include "gca/gfx_6_0_d.h"
 #include "gca/gfx_6_0_sh_mask.h"
+#include "gca/gfx_7_2_enum.h"
+
 #include "gmc/gmc_6_0_d.h"
 #include "gmc/gmc_6_0_sh_mask.h"
+
 #include "dce/dce_6_0_d.h"
 #include "dce/dce_6_0_sh_mask.h"
-#include "gca/gfx_7_2_enum.h"
-#include "dce_v6_0.h"
+
 #include "si_enums.h"
 
 static void dce_v6_0_set_display_funcs(struct amdgpu_device *adev);
@@ -50,31 +66,31 @@ static void dce_v6_0_set_irq_funcs(struct amdgpu_device *adev);
 
 static const u32 crtc_offsets[6] =
 {
-	SI_CRTC0_REGISTER_OFFSET,
-	SI_CRTC1_REGISTER_OFFSET,
-	SI_CRTC2_REGISTER_OFFSET,
-	SI_CRTC3_REGISTER_OFFSET,
-	SI_CRTC4_REGISTER_OFFSET,
-	SI_CRTC5_REGISTER_OFFSET
+	CRTC0_REGISTER_OFFSET,
+	CRTC1_REGISTER_OFFSET,
+	CRTC2_REGISTER_OFFSET,
+	CRTC3_REGISTER_OFFSET,
+	CRTC4_REGISTER_OFFSET,
+	CRTC5_REGISTER_OFFSET
 };
 
 static const u32 hpd_offsets[] =
 {
-	mmDC_HPD1_INT_STATUS - mmDC_HPD1_INT_STATUS,
-	mmDC_HPD2_INT_STATUS - mmDC_HPD1_INT_STATUS,
-	mmDC_HPD3_INT_STATUS - mmDC_HPD1_INT_STATUS,
-	mmDC_HPD4_INT_STATUS - mmDC_HPD1_INT_STATUS,
-	mmDC_HPD5_INT_STATUS - mmDC_HPD1_INT_STATUS,
-	mmDC_HPD6_INT_STATUS - mmDC_HPD1_INT_STATUS,
+	HPD0_REGISTER_OFFSET,
+	HPD1_REGISTER_OFFSET,
+	HPD2_REGISTER_OFFSET,
+	HPD3_REGISTER_OFFSET,
+	HPD4_REGISTER_OFFSET,
+	HPD5_REGISTER_OFFSET
 };
 
 static const uint32_t dig_offsets[] = {
-	SI_CRTC0_REGISTER_OFFSET,
-	SI_CRTC1_REGISTER_OFFSET,
-	SI_CRTC2_REGISTER_OFFSET,
-	SI_CRTC3_REGISTER_OFFSET,
-	SI_CRTC4_REGISTER_OFFSET,
-	SI_CRTC5_REGISTER_OFFSET,
+	CRTC0_REGISTER_OFFSET,
+	CRTC1_REGISTER_OFFSET,
+	CRTC2_REGISTER_OFFSET,
+	CRTC3_REGISTER_OFFSET,
+	CRTC4_REGISTER_OFFSET,
+	CRTC5_REGISTER_OFFSET,
 	(0x13830 - 0x7030) >> 2,
 };
 
@@ -142,64 +158,6 @@ static void dce_v6_0_audio_endpt_wreg(struct amdgpu_device *adev,
 	spin_unlock_irqrestore(&adev->audio_endpt_idx_lock, flags);
 }
 
-static bool dce_v6_0_is_in_vblank(struct amdgpu_device *adev, int crtc)
-{
-	if (RREG32(mmCRTC_STATUS + crtc_offsets[crtc]) & CRTC_STATUS__CRTC_V_BLANK_MASK)
-		return true;
-	else
-		return false;
-}
-
-static bool dce_v6_0_is_counter_moving(struct amdgpu_device *adev, int crtc)
-{
-	u32 pos1, pos2;
-
-	pos1 = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
-	pos2 = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
-
-	if (pos1 != pos2)
-		return true;
-	else
-		return false;
-}
-
-/**
- * dce_v6_0_wait_for_vblank - vblank wait asic callback.
- *
- * @crtc: crtc to wait for vblank on
- *
- * Wait for vblank on the requested crtc (evergreen+).
- */
-static void dce_v6_0_vblank_wait(struct amdgpu_device *adev, int crtc)
-{
-	unsigned i = 100;
-
-	if (crtc >= adev->mode_info.num_crtc)
-		return;
-
-	if (!(RREG32(mmCRTC_CONTROL + crtc_offsets[crtc]) & CRTC_CONTROL__CRTC_MASTER_EN_MASK))
-		return;
-
-	/* depending on when we hit vblank, we may be close to active; if so,
-	 * wait for another frame.
-	 */
-	while (dce_v6_0_is_in_vblank(adev, crtc)) {
-		if (i++ == 100) {
-			i = 0;
-			if (!dce_v6_0_is_counter_moving(adev, crtc))
-				break;
-		}
-	}
-
-	while (!dce_v6_0_is_in_vblank(adev, crtc)) {
-		if (i++ == 100) {
-			i = 0;
-			if (!dce_v6_0_is_counter_moving(adev, crtc))
-				break;
-		}
-	}
-}
-
 static u32 dce_v6_0_vblank_get_counter(struct amdgpu_device *adev, int crtc)
 {
 	if (crtc >= adev->mode_info.num_crtc)
@@ -232,6 +190,7 @@ static void dce_v6_0_pageflip_interrupt_fini(struct amdgpu_device *adev)
  * @adev: amdgpu_device pointer
  * @crtc_id: crtc to cleanup pageflip on
  * @crtc_base: new address of the crtc (GPU MC address)
+ * @async: asynchronous flip
  *
  * Does the actual pageflip (evergreen+).
  * During vblank we take the crtc lock and wait for the update_pending
@@ -243,16 +202,20 @@ static void dce_v6_0_page_flip(struct amdgpu_device *adev,
 			       int crtc_id, u64 crtc_base, bool async)
 {
 	struct amdgpu_crtc *amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
+	struct drm_framebuffer *fb = amdgpu_crtc->base.primary->fb;
 
 	/* flip at hsync for async, default is vsync */
 	WREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset, async ?
 	       GRPH_FLIP_CONTROL__GRPH_SURFACE_UPDATE_H_RETRACE_EN_MASK : 0);
+	/* update pitch */
+	WREG32(mmGRPH_PITCH + amdgpu_crtc->crtc_offset,
+	       fb->pitches[0] / fb->format->cpp[0]);
 	/* update the scanout addresses */
 	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
 	       upper_32_bits(crtc_base));
+	/* writing to the low address triggers the update */
 	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
 	       (u32)crtc_base);
-
 	/* post the write */
 	RREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset);
 }
@@ -262,11 +225,11 @@ static int dce_v6_0_crtc_get_scanoutpos(struct amdgpu_device *adev, int crtc,
 {
 	if ((crtc < 0) || (crtc >= adev->mode_info.num_crtc))
 		return -EINVAL;
+
 	*vbl = RREG32(mmCRTC_V_BLANK_START_END + crtc_offsets[crtc]);
 	*position = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
 
 	return 0;
-
 }
 
 /**
@@ -286,7 +249,8 @@ static bool dce_v6_0_hpd_sense(struct amdgpu_device *adev,
 	if (hpd >= adev->mode_info.num_hpd)
 		return connected;
 
-	if (RREG32(mmDC_HPD1_INT_STATUS + hpd_offsets[hpd]) & DC_HPD1_INT_STATUS__DC_HPD1_SENSE_MASK)
+	if (RREG32(mmDC_HPD1_INT_STATUS + hpd_offsets[hpd]) &
+	    DC_HPD1_INT_STATUS__DC_HPD1_SENSE_MASK)
 		connected = true;
 
 	return connected;
@@ -317,6 +281,21 @@ static void dce_v6_0_hpd_set_polarity(struct amdgpu_device *adev,
 	WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd], tmp);
 }
 
+static void dce_v6_0_hpd_int_ack(struct amdgpu_device *adev,
+				 int hpd)
+{
+	u32 tmp;
+
+	if (hpd >= adev->mode_info.num_hpd) {
+		DRM_DEBUG("invalid hpd %d\n", hpd);
+		return;
+	}
+
+	tmp = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd]);
+	tmp |= DC_HPD1_INT_CONTROL__DC_HPD1_INT_ACK_MASK;
+	WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd], tmp);
+}
+
 /**
  * dce_v6_0_hpd_init - hpd setup callback.
  *
@@ -327,11 +306,13 @@ static void dce_v6_0_hpd_set_polarity(struct amdgpu_device *adev,
  */
 static void dce_v6_0_hpd_init(struct amdgpu_device *adev)
 {
-	struct drm_device *dev = adev->ddev;
+	struct drm_device *dev = adev_to_drm(adev);
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	u32 tmp;
 
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 
 		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
@@ -354,10 +335,11 @@ static void dce_v6_0_hpd_init(struct amdgpu_device *adev)
 			continue;
 		}
 
+		dce_v6_0_hpd_int_ack(adev, amdgpu_connector->hpd.hpd);
 		dce_v6_0_hpd_set_polarity(adev, amdgpu_connector->hpd.hpd);
 		amdgpu_irq_get(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
 	}
-
+	drm_connector_list_iter_end(&iter);
 }
 
 /**
@@ -370,11 +352,13 @@ static void dce_v6_0_hpd_init(struct amdgpu_device *adev)
  */
 static void dce_v6_0_hpd_fini(struct amdgpu_device *adev)
 {
-	struct drm_device *dev = adev->ddev;
+	struct drm_device *dev = adev_to_drm(adev);
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	u32 tmp;
 
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 
 		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
@@ -382,10 +366,11 @@ static void dce_v6_0_hpd_fini(struct amdgpu_device *adev)
 
 		tmp = RREG32(mmDC_HPD1_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd]);
 		tmp &= ~DC_HPD1_CONTROL__DC_HPD1_EN_MASK;
-		WREG32(mmDC_HPD1_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd], 0);
+		WREG32(mmDC_HPD1_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
 
 		amdgpu_irq_put(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
 	}
+	drm_connector_list_iter_end(&iter);
 }
 
 static u32 dce_v6_0_hpd_get_gpio_reg(struct amdgpu_device *adev)
@@ -393,13 +378,41 @@ static u32 dce_v6_0_hpd_get_gpio_reg(struct amdgpu_device *adev)
 	return mmDC_GPIO_HPD_A;
 }
 
+static bool dce_v6_0_is_display_hung(struct amdgpu_device *adev)
+{
+	u32 crtc_hung = 0;
+	u32 crtc_status[6];
+	u32 i, j, tmp;
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		if (RREG32(mmCRTC_CONTROL + crtc_offsets[i]) & CRTC_CONTROL__CRTC_MASTER_EN_MASK) {
+			crtc_status[i] = RREG32(mmCRTC_STATUS_HV_COUNT + crtc_offsets[i]);
+			crtc_hung |= (1 << i);
+		}
+	}
+
+	for (j = 0; j < 10; j++) {
+		for (i = 0; i < adev->mode_info.num_crtc; i++) {
+			if (crtc_hung & (1 << i)) {
+				tmp = RREG32(mmCRTC_STATUS_HV_COUNT + crtc_offsets[i]);
+				if (tmp != crtc_status[i])
+					crtc_hung &= ~(1 << i);
+			}
+		}
+		if (crtc_hung == 0)
+			return false;
+		udelay(100);
+	}
+
+	return true;
+}
+
 static void dce_v6_0_set_vga_render_state(struct amdgpu_device *adev,
 					  bool render)
 {
 	if (!render)
 		WREG32(mmVGA_RENDER_CONTROL,
-			RREG32(mmVGA_RENDER_CONTROL) & VGA_VSTATUS_CNTL);
-
+		       RREG32(mmVGA_RENDER_CONTROL) & ~VGA_RENDER_CONTROL__VGA_VSTATUS_CNTL_MASK);
 }
 
 static int dce_v6_0_get_num_crtc(struct amdgpu_device *adev)
@@ -442,9 +455,8 @@ void dce_v6_0_disable_dce(struct amdgpu_device *adev)
 
 static void dce_v6_0_program_fmt(struct drm_encoder *encoder)
 {
-
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
@@ -498,7 +510,7 @@ static void dce_v6_0_program_fmt(struct drm_encoder *encoder)
 }
 
 /**
- * cik_get_number_of_dram_channels - get the number of dram channels
+ * si_get_number_of_dram_channels - get the number of dram channels
  *
  * @adev: amdgpu_device pointer
  *
@@ -885,7 +897,7 @@ static void dce_v6_0_program_watermarks(struct amdgpu_device *adev,
 					    (u32)mode->clock);
 		line_time = (u32) div_u64((u64)mode->crtc_htotal * 1000000,
 					  (u32)mode->clock);
-		line_time = min(line_time, (u32)65535);
+		line_time = min_t(u32, line_time, 65535);
 		priority_a_cnt = 0;
 		priority_b_cnt = 0;
 
@@ -918,8 +930,8 @@ static void dce_v6_0_program_watermarks(struct amdgpu_device *adev,
 		wm_high.dram_channels = dram_channels;
 		wm_high.num_heads = num_heads;
 
-		if (adev->pm.dpm_enabled) {
 		/* watermark for low clocks */
+		if (adev->pm.dpm_enabled) {
 			wm_low.yclk =
 				amdgpu_dpm_get_mclk(adev, true) * 10;
 			wm_low.sclk =
@@ -946,9 +958,9 @@ static void dce_v6_0_program_watermarks(struct amdgpu_device *adev,
 		wm_low.num_heads = num_heads;
 
 		/* set for high clocks */
-		latency_watermark_a = min(dce_v6_0_latency_watermark(&wm_high), (u32)65535);
+		latency_watermark_a = min_t(u32, dce_v6_0_latency_watermark(&wm_high), 65535);
 		/* set for low clocks */
-		latency_watermark_b = min(dce_v6_0_latency_watermark(&wm_low), (u32)65535);
+		latency_watermark_b = min_t(u32, dce_v6_0_latency_watermark(&wm_low), 65535);
 
 		/* possibly force display priority to high */
 		/* should really do this at mode validation time... */
@@ -999,16 +1011,16 @@ static void dce_v6_0_program_watermarks(struct amdgpu_device *adev,
 	/* select wm A */
 	arb_control3 = RREG32(mmDPG_PIPE_ARBITRATION_CONTROL3 + amdgpu_crtc->crtc_offset);
 	tmp = arb_control3;
-	tmp &= ~LATENCY_WATERMARK_MASK(3);
-	tmp |= LATENCY_WATERMARK_MASK(1);
+	tmp &= ~(3 << DPG_PIPE_ARBITRATION_CONTROL3__URGENCY_WATERMARK_MASK__SHIFT);
+	tmp |= (1 << DPG_PIPE_ARBITRATION_CONTROL3__URGENCY_WATERMARK_MASK__SHIFT);
 	WREG32(mmDPG_PIPE_ARBITRATION_CONTROL3 + amdgpu_crtc->crtc_offset, tmp);
 	WREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset,
 	       ((latency_watermark_a << DPG_PIPE_URGENCY_CONTROL__URGENCY_LOW_WATERMARK__SHIFT)  |
 		(line_time << DPG_PIPE_URGENCY_CONTROL__URGENCY_HIGH_WATERMARK__SHIFT)));
 	/* select wm B */
 	tmp = RREG32(mmDPG_PIPE_ARBITRATION_CONTROL3 + amdgpu_crtc->crtc_offset);
-	tmp &= ~LATENCY_WATERMARK_MASK(3);
-	tmp |= LATENCY_WATERMARK_MASK(2);
+	tmp &= ~(3 << DPG_PIPE_ARBITRATION_CONTROL3__URGENCY_WATERMARK_MASK__SHIFT);
+	tmp |= (2 << DPG_PIPE_ARBITRATION_CONTROL3__URGENCY_WATERMARK_MASK__SHIFT);
 	WREG32(mmDPG_PIPE_ARBITRATION_CONTROL3 + amdgpu_crtc->crtc_offset, tmp);
 	WREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset,
 	       ((latency_watermark_b << DPG_PIPE_URGENCY_CONTROL__URGENCY_LOW_WATERMARK__SHIFT) |
@@ -1022,13 +1034,26 @@ static void dce_v6_0_program_watermarks(struct amdgpu_device *adev,
 
 	/* save values for DPM */
 	amdgpu_crtc->line_time = line_time;
-	amdgpu_crtc->wm_high = latency_watermark_a;
 
 	/* Save number of lines the linebuffer leads before the scanout */
 	amdgpu_crtc->lb_vblank_lead_lines = lb_vblank_lead_lines;
 }
 
 /* watermark setup */
+/**
+ * dce_v6_0_line_buffer_adjust - Set up the line buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @amdgpu_crtc: the selected display controller
+ * @mode: the current display mode on the selected display
+ * controller
+ * @other_mode: the display mode of another display controller
+ *              that may be sharing the line buffer
+ *
+ * Setup up the line buffer allocation for
+ * the selected display controller (CIK).
+ * Returns the line buffer size in pixels.
+ */
 static u32 dce_v6_0_line_buffer_adjust(struct amdgpu_device *adev,
 				   struct amdgpu_crtc *amdgpu_crtc,
 				   struct drm_display_mode *mode,
@@ -1063,7 +1088,7 @@ static u32 dce_v6_0_line_buffer_adjust(struct amdgpu_device *adev,
 	}
 
 	WREG32(mmDC_LB_MEMORY_SPLIT + amdgpu_crtc->crtc_offset,
-	       DC_LB_MEMORY_CONFIG(tmp));
+	       (tmp << DC_LB_MEMORY_SPLIT__DC_LB_MEMORY_CONFIG__SHIFT));
 
 	WREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset,
 	       (buffer_alloc << PIPE0_DMIF_BUFFER_CONTROL__DMIF_BUFFERS_ALLOCATED__SHIFT));
@@ -1090,7 +1115,6 @@ static u32 dce_v6_0_line_buffer_adjust(struct amdgpu_device *adev,
 
 
 /**
- *
  * dce_v6_0_bandwidth_update - program display watermarks
  *
  * @adev: amdgpu_device pointer
@@ -1108,7 +1132,7 @@ static void dce_v6_0_bandwidth_update(struct amdgpu_device *adev)
 	if (!adev->mode_info.mode_config_initialized)
 		return;
 
-	amdgpu_update_display_priority(adev);
+	amdgpu_display_update_priority(adev);
 
 	for (i = 0; i < adev->mode_info.num_crtc; i++) {
 		if (adev->mode_info.crtcs[i]->base.enabled)
@@ -1157,7 +1181,7 @@ static struct amdgpu_audio_pin *dce_v6_0_audio_get_pin(struct amdgpu_device *ade
 
 static void dce_v6_0_audio_select_pin(struct drm_encoder *encoder)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(encoder->dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 
@@ -1172,20 +1196,24 @@ static void dce_v6_0_audio_select_pin(struct drm_encoder *encoder)
 static void dce_v6_0_audio_write_latency_fields(struct drm_encoder *encoder,
 						struct drm_display_mode *mode)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	int interlace = 0;
 	u32 tmp;
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
@@ -1212,28 +1240,32 @@ static void dce_v6_0_audio_write_latency_fields(struct drm_encoder *encoder,
 
 static void dce_v6_0_audio_write_speaker_allocation(struct drm_encoder *encoder)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	u8 *sadb = NULL;
 	int sad_count;
 	u32 tmp;
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
 		return;
 	}
 
-	sad_count = drm_edid_to_speaker_allocation(amdgpu_connector_edid(connector), &sadb);
+	sad_count = drm_edid_to_speaker_allocation(amdgpu_connector->edid, &sadb);
 	if (sad_count < 0) {
 		DRM_ERROR("Couldn't read Speaker Allocation Data Block: %d\n", sad_count);
 		sad_count = 0;
@@ -1269,10 +1301,13 @@ static void dce_v6_0_audio_write_speaker_allocation(struct drm_encoder *encoder)
 
 static void dce_v6_0_audio_write_sad_regs(struct drm_encoder *encoder)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	u32 offset;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	struct cea_sad *sads;
 	int i, sad_count;
@@ -1292,26 +1327,33 @@ static void dce_v6_0_audio_write_sad_regs(struct drm_encoder *encoder)
 		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR13, HDMI_AUDIO_CODING_TYPE_WMA_PRO },
 	};
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	offset = dig->afmt->pin->offset;
+
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
 		return;
 	}
 
-	sad_count = drm_edid_to_sad(amdgpu_connector_edid(connector), &sads);
-	if (sad_count <= 0) {
+	sad_count = drm_edid_to_sad(amdgpu_connector->edid, &sads);
+	if (sad_count < 0)
 		DRM_ERROR("Couldn't read SADs: %d\n", sad_count);
+	if (sad_count <= 0)
 		return;
-	}
 
 	for (i = 0; i < ARRAY_SIZE(eld_reg_to_type); i++) {
-		u32 tmp = 0;
+		u32 value = 0;
 		u8 stereo_freqs = 0;
 		int max_channels = -1;
 		int j;
@@ -1321,12 +1363,12 @@ static void dce_v6_0_audio_write_sad_regs(struct drm_encoder *encoder)
 
 			if (sad->format == eld_reg_to_type[i][1]) {
 				if (sad->channels > max_channels) {
-					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
-							MAX_CHANNELS, sad->channels);
-					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
-							DESCRIPTOR_BYTE_2, sad->byte2);
-					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
-							SUPPORTED_FREQUENCIES, sad->freq);
+					value = (sad->channels <<
+						AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__MAX_CHANNELS__SHIFT) |
+					       (sad->byte2 <<
+						AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__DESCRIPTOR_BYTE_2__SHIFT) |
+					       (sad->freq <<
+						AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__SUPPORTED_FREQUENCIES__SHIFT);
 					max_channels = sad->channels;
 				}
 
@@ -1337,13 +1379,13 @@ static void dce_v6_0_audio_write_sad_regs(struct drm_encoder *encoder)
 			}
 		}
 
-		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
-				SUPPORTED_FREQUENCIES_STEREO, stereo_freqs);
-		WREG32_AUDIO_ENDPT(dig->afmt->pin->offset, eld_reg_to_type[i][0], tmp);
+		value |= (stereo_freqs <<
+			AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__SUPPORTED_FREQUENCIES_STEREO__SHIFT);
+
+		WREG32_AUDIO_ENDPT(offset, eld_reg_to_type[i][0], value);
 	}
 
 	kfree(sads);
-
 }
 
 static void dce_v6_0_audio_enable(struct amdgpu_device *adev,
@@ -1359,13 +1401,13 @@ static void dce_v6_0_audio_enable(struct amdgpu_device *adev,
 
 static const u32 pin_offsets[7] =
 {
-	(0x1780 - 0x1780),
-	(0x1786 - 0x1780),
-	(0x178c - 0x1780),
-	(0x1792 - 0x1780),
-	(0x1798 - 0x1780),
-	(0x179d - 0x1780),
-	(0x17a4 - 0x1780),
+	AUD0_REGISTER_OFFSET,
+	AUD1_REGISTER_OFFSET,
+	AUD2_REGISTER_OFFSET,
+	AUD3_REGISTER_OFFSET,
+	AUD4_REGISTER_OFFSET,
+	AUD5_REGISTER_OFFSET,
+	AUD6_REGISTER_OFFSET,
 };
 
 static int dce_v6_0_audio_init(struct amdgpu_device *adev)
@@ -1398,6 +1440,8 @@ static int dce_v6_0_audio_init(struct amdgpu_device *adev)
 		adev->mode_info.audio.pin[i].connected = false;
 		adev->mode_info.audio.pin[i].offset = pin_offsets[i];
 		adev->mode_info.audio.pin[i].id = i;
+		/* disable audio.  it will be set up later */
+		/* XXX remove once we switch to ip funcs */
 		dce_v6_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
 	}
 
@@ -1406,24 +1450,19 @@ static int dce_v6_0_audio_init(struct amdgpu_device *adev)
 
 static void dce_v6_0_audio_fini(struct amdgpu_device *adev)
 {
-	int i;
-
 	if (!amdgpu_audio)
 		return;
 
 	if (!adev->mode_info.audio.enabled)
 		return;
 
-	for (i = 0; i < adev->mode_info.audio.num_pins; i++)
-		dce_v6_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
-
 	adev->mode_info.audio.enabled = false;
 }
 
 static void dce_v6_0_audio_set_vbi_packet(struct drm_encoder *encoder)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	u32 tmp;
@@ -1439,7 +1478,7 @@ static void dce_v6_0_audio_set_acr(struct drm_encoder *encoder,
 				   uint32_t clock, int bpc)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_afmt_acr acr = amdgpu_afmt_acr(clock);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
@@ -1477,9 +1516,10 @@ static void dce_v6_0_audio_set_avi_infoframe(struct drm_encoder *encoder,
 					       struct drm_display_mode *mode)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
 	struct hdmi_avi_infoframe frame;
 	u8 buffer[HDMI_INFOFRAME_HEADER_SIZE + HDMI_AVI_INFOFRAME_SIZE];
 	uint8_t *payload = buffer + 3;
@@ -1487,7 +1527,7 @@ static void dce_v6_0_audio_set_avi_infoframe(struct drm_encoder *encoder,
 	ssize_t err;
 	u32 tmp;
 
-	err = drm_hdmi_avi_infoframe_from_display_mode(&frame, mode, false);
+	err = drm_hdmi_avi_infoframe_from_display_mode(&frame, connector, mode);
 	if (err < 0) {
 		DRM_ERROR("failed to setup AVI infoframe: %zd\n", err);
 		return;
@@ -1518,7 +1558,7 @@ static void dce_v6_0_audio_set_avi_infoframe(struct drm_encoder *encoder,
 static void dce_v6_0_audio_set_dto(struct drm_encoder *encoder, u32 clock)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
 	int em = amdgpu_atombios_encoder_get_encoder_mode(encoder);
 	u32 tmp;
@@ -1552,7 +1592,7 @@ static void dce_v6_0_audio_set_dto(struct drm_encoder *encoder, u32 clock)
 static void dce_v6_0_audio_set_packet(struct drm_encoder *encoder)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	u32 tmp;
@@ -1596,7 +1636,7 @@ static void dce_v6_0_audio_set_packet(struct drm_encoder *encoder)
 static void dce_v6_0_audio_set_mute(struct drm_encoder *encoder, bool mute)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	u32 tmp;
@@ -1609,7 +1649,7 @@ static void dce_v6_0_audio_set_mute(struct drm_encoder *encoder, bool mute)
 static void dce_v6_0_audio_hdmi_enable(struct drm_encoder *encoder, bool enable)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	u32 tmp;
@@ -1646,7 +1686,7 @@ static void dce_v6_0_audio_hdmi_enable(struct drm_encoder *encoder, bool enable)
 static void dce_v6_0_audio_dp_enable(struct drm_encoder *encoder, bool enable)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	u32 tmp;
@@ -1675,10 +1715,11 @@ static void dce_v6_0_afmt_setmode(struct drm_encoder *encoder,
 				  struct drm_display_mode *mode)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	int em = amdgpu_atombios_encoder_get_encoder_mode(encoder);
 	int bpc = 8;
@@ -1686,12 +1727,14 @@ static void dce_v6_0_afmt_setmode(struct drm_encoder *encoder,
 	if (!dig || !dig->afmt)
 		return;
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
@@ -1741,7 +1784,7 @@ static void dce_v6_0_afmt_setmode(struct drm_encoder *encoder,
 static void dce_v6_0_afmt_enable(struct drm_encoder *encoder, bool enable)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 
@@ -1815,7 +1858,7 @@ static void dce_v6_0_vga_enable(struct drm_crtc *crtc, bool enable)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	u32 vga_control;
 
 	vga_control = RREG32(vga_control_regs[amdgpu_crtc->crtc_id]) & ~1;
@@ -1826,7 +1869,7 @@ static void dce_v6_0_grph_enable(struct drm_crtc *crtc, bool enable)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 
 	WREG32(mmGRPH_ENABLE + amdgpu_crtc->crtc_offset, enable ? 1 : 0);
 }
@@ -1837,18 +1880,16 @@ static int dce_v6_0_crtc_do_set_base(struct drm_crtc *crtc,
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-	struct amdgpu_framebuffer *amdgpu_fb;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct drm_framebuffer *target_fb;
 	struct drm_gem_object *obj;
 	struct amdgpu_bo *abo;
 	uint64_t fb_location, tiling_flags;
 	uint32_t fb_format, fb_pitch_pixels, pipe_config;
-	u32 fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_NONE);
+	u32 fb_swap = (GRPH_ENDIAN_NONE << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
 	u32 viewport_w, viewport_h;
 	int r;
 	bool bypass_lut = false;
-	struct drm_format_name_buf format_name;
 
 	/* no fb bound */
 	if (!atomic && !crtc->primary->fb) {
@@ -1856,103 +1897,110 @@ static int dce_v6_0_crtc_do_set_base(struct drm_crtc *crtc,
 		return 0;
 	}
 
-	if (atomic) {
-		amdgpu_fb = to_amdgpu_framebuffer(fb);
+	if (atomic)
 		target_fb = fb;
-	} else {
-		amdgpu_fb = to_amdgpu_framebuffer(crtc->primary->fb);
+	else
 		target_fb = crtc->primary->fb;
-	}
 
 	/* If atomic, assume fb object is pinned & idle & fenced and
 	 * just update base pointers
 	 */
-	obj = amdgpu_fb->obj;
+	obj = target_fb->obj[0];
 	abo = gem_to_amdgpu_bo(obj);
 	r = amdgpu_bo_reserve(abo, false);
 	if (unlikely(r != 0))
 		return r;
 
-	if (atomic) {
-		fb_location = amdgpu_bo_gpu_offset(abo);
-	} else {
-		r = amdgpu_bo_pin(abo, AMDGPU_GEM_DOMAIN_VRAM, &fb_location);
+	if (!atomic) {
+		abo->flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+		r = amdgpu_bo_pin(abo, AMDGPU_GEM_DOMAIN_VRAM);
 		if (unlikely(r != 0)) {
 			amdgpu_bo_unreserve(abo);
 			return -EINVAL;
 		}
 	}
+	fb_location = amdgpu_bo_gpu_offset(abo);
 
 	amdgpu_bo_get_tiling_flags(abo, &tiling_flags);
 	amdgpu_bo_unreserve(abo);
 
 	switch (target_fb->format->format) {
 	case DRM_FORMAT_C8:
-		fb_format = (GRPH_DEPTH(GRPH_DEPTH_8BPP) |
-			     GRPH_FORMAT(GRPH_FORMAT_INDEXED));
+		fb_format = ((GRPH_DEPTH_8BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_INDEXED << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
 		break;
 	case DRM_FORMAT_XRGB4444:
 	case DRM_FORMAT_ARGB4444:
-		fb_format = (GRPH_DEPTH(GRPH_DEPTH_16BPP) |
-			     GRPH_FORMAT(GRPH_FORMAT_ARGB4444));
+		fb_format = ((GRPH_DEPTH_16BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_ARGB4444 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
 #ifdef __BIG_ENDIAN
-		fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_8IN16);
+		fb_swap = (GRPH_ENDIAN_8IN16 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
 #endif
 		break;
 	case DRM_FORMAT_XRGB1555:
 	case DRM_FORMAT_ARGB1555:
-		fb_format = (GRPH_DEPTH(GRPH_DEPTH_16BPP) |
-			     GRPH_FORMAT(GRPH_FORMAT_ARGB1555));
+		fb_format = ((GRPH_DEPTH_16BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_ARGB1555 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
 #ifdef __BIG_ENDIAN
-		fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_8IN16);
+		fb_swap = (GRPH_ENDIAN_8IN16 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
 #endif
 		break;
 	case DRM_FORMAT_BGRX5551:
 	case DRM_FORMAT_BGRA5551:
-		fb_format = (GRPH_DEPTH(GRPH_DEPTH_16BPP) |
-			     GRPH_FORMAT(GRPH_FORMAT_BGRA5551));
+		fb_format = ((GRPH_DEPTH_16BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_BGRA5551 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
 #ifdef __BIG_ENDIAN
-		fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_8IN16);
+		fb_swap = (GRPH_ENDIAN_8IN16 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
 #endif
 		break;
 	case DRM_FORMAT_RGB565:
-		fb_format = (GRPH_DEPTH(GRPH_DEPTH_16BPP) |
-			     GRPH_FORMAT(GRPH_FORMAT_ARGB565));
+		fb_format = ((GRPH_DEPTH_16BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_ARGB565 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
 #ifdef __BIG_ENDIAN
-		fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_8IN16);
+		fb_swap = (GRPH_ENDIAN_8IN16 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
 #endif
 		break;
 	case DRM_FORMAT_XRGB8888:
 	case DRM_FORMAT_ARGB8888:
-		fb_format = (GRPH_DEPTH(GRPH_DEPTH_32BPP) |
-			     GRPH_FORMAT(GRPH_FORMAT_ARGB8888));
+		fb_format = ((GRPH_DEPTH_32BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_ARGB8888 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
 #ifdef __BIG_ENDIAN
-		fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_8IN32);
+		fb_swap = (GRPH_ENDIAN_8IN32 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
 #endif
 		break;
 	case DRM_FORMAT_XRGB2101010:
 	case DRM_FORMAT_ARGB2101010:
-		fb_format = (GRPH_DEPTH(GRPH_DEPTH_32BPP) |
-			     GRPH_FORMAT(GRPH_FORMAT_ARGB2101010));
+		fb_format = ((GRPH_DEPTH_32BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_ARGB2101010 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
 #ifdef __BIG_ENDIAN
-		fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_8IN32);
+		fb_swap = (GRPH_ENDIAN_8IN32 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
 #endif
 		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
 		bypass_lut = true;
 		break;
 	case DRM_FORMAT_BGRX1010102:
 	case DRM_FORMAT_BGRA1010102:
-		fb_format = (GRPH_DEPTH(GRPH_DEPTH_32BPP) |
-			     GRPH_FORMAT(GRPH_FORMAT_BGRA1010102));
+		fb_format = ((GRPH_DEPTH_32BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_BGRA1010102 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
 #ifdef __BIG_ENDIAN
-		fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_8IN32);
+		fb_swap = (GRPH_ENDIAN_8IN32 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
 #endif
 		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
 		bypass_lut = true;
 		break;
+	case DRM_FORMAT_XBGR8888:
+	case DRM_FORMAT_ABGR8888:
+		fb_format = ((GRPH_DEPTH_32BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_ARGB8888 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
+		fb_swap = ((GRPH_RED_SEL_B << GRPH_SWAP_CNTL__GRPH_RED_CROSSBAR__SHIFT) |
+			   (GRPH_BLUE_SEL_R << GRPH_SWAP_CNTL__GRPH_BLUE_CROSSBAR__SHIFT));
+#ifdef __BIG_ENDIAN
+		fb_swap |= (GRPH_ENDIAN_8IN32 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
+#endif
+		break;
 	default:
-		DRM_ERROR("Unsupported screen format %s\n",
-		          drm_get_format_name(target_fb->format->format, &format_name));
+		DRM_ERROR("Unsupported screen format %p4cc\n",
+			  &target_fb->format->format);
 		return -EINVAL;
 	}
 
@@ -1965,18 +2013,18 @@ static int dce_v6_0_crtc_do_set_base(struct drm_crtc *crtc,
 		tile_split = AMDGPU_TILING_GET(tiling_flags, TILE_SPLIT);
 		num_banks = AMDGPU_TILING_GET(tiling_flags, NUM_BANKS);
 
-		fb_format |= GRPH_NUM_BANKS(num_banks);
-		fb_format |= GRPH_ARRAY_MODE(GRPH_ARRAY_2D_TILED_THIN1);
-		fb_format |= GRPH_TILE_SPLIT(tile_split);
-		fb_format |= GRPH_BANK_WIDTH(bankw);
-		fb_format |= GRPH_BANK_HEIGHT(bankh);
-		fb_format |= GRPH_MACRO_TILE_ASPECT(mtaspect);
+		fb_format |= (num_banks << GRPH_CONTROL__GRPH_NUM_BANKS__SHIFT);
+		fb_format |= (GRPH_ARRAY_2D_TILED_THIN1 << GRPH_CONTROL__GRPH_ARRAY_MODE__SHIFT);
+		fb_format |= (tile_split << GRPH_CONTROL__GRPH_TILE_SPLIT__SHIFT);
+		fb_format |= (bankw << GRPH_CONTROL__GRPH_BANK_WIDTH__SHIFT);
+		fb_format |= (bankh << GRPH_CONTROL__GRPH_BANK_HEIGHT__SHIFT);
+		fb_format |= (mtaspect << GRPH_CONTROL__GRPH_MACRO_TILE_ASPECT__SHIFT);
 	} else if (AMDGPU_TILING_GET(tiling_flags, ARRAY_MODE) == ARRAY_1D_TILED_THIN1) {
-		fb_format |= GRPH_ARRAY_MODE(GRPH_ARRAY_1D_TILED_THIN1);
+		fb_format |= (GRPH_ARRAY_1D_TILED_THIN1 << GRPH_CONTROL__GRPH_ARRAY_MODE__SHIFT);
 	}
 
 	pipe_config = AMDGPU_TILING_GET(tiling_flags, PIPE_CONFIG);
-	fb_format |= GRPH_PIPE_CONFIG(pipe_config);
+	fb_format |= (pipe_config << GRPH_CONTROL__GRPH_PIPE_CONFIG__SHIFT);
 
 	dce_v6_0_vga_enable(crtc, false);
 
@@ -1992,7 +2040,7 @@ static int dce_v6_0_crtc_do_set_base(struct drm_crtc *crtc,
 	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
 	       (u32)fb_location & GRPH_PRIMARY_SURFACE_ADDRESS__GRPH_PRIMARY_SURFACE_ADDRESS_MASK);
 	WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
-	       (u32) fb_location & GRPH_PRIMARY_SURFACE_ADDRESS__GRPH_PRIMARY_SURFACE_ADDRESS_MASK);
+	       (u32) fb_location & GRPH_SECONDARY_SURFACE_ADDRESS__GRPH_SECONDARY_SURFACE_ADDRESS_MASK);
 	WREG32(mmGRPH_CONTROL + amdgpu_crtc->crtc_offset, fb_format);
 	WREG32(mmGRPH_SWAP_CNTL + amdgpu_crtc->crtc_offset, fb_swap);
 
@@ -2036,8 +2084,7 @@ static int dce_v6_0_crtc_do_set_base(struct drm_crtc *crtc,
 	WREG32(mmMASTER_UPDATE_MODE + amdgpu_crtc->crtc_offset, 0);
 
 	if (!atomic && fb && fb != crtc->primary->fb) {
-		amdgpu_fb = to_amdgpu_framebuffer(fb);
-		abo = gem_to_amdgpu_bo(amdgpu_fb->obj);
+		abo = gem_to_amdgpu_bo(fb->obj[0]);
 		r = amdgpu_bo_reserve(abo, true);
 		if (unlikely(r != 0))
 			return r;
@@ -2056,37 +2103,36 @@ static void dce_v6_0_set_interleave(struct drm_crtc *crtc,
 				    struct drm_display_mode *mode)
 {
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 
 	if (mode->flags & DRM_MODE_FLAG_INTERLACE)
 		WREG32(mmDATA_FORMAT + amdgpu_crtc->crtc_offset,
-		       INTERLEAVE_EN);
+			DATA_FORMAT__INTERLEAVE_EN_MASK);
 	else
 		WREG32(mmDATA_FORMAT + amdgpu_crtc->crtc_offset, 0);
 }
 
 static void dce_v6_0_crtc_load_lut(struct drm_crtc *crtc)
 {
-
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	u16 *r, *g, *b;
 	int i;
 
 	DRM_DEBUG_KMS("%d\n", amdgpu_crtc->crtc_id);
 
 	WREG32(mmINPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset,
-	       ((0 << INPUT_CSC_CONTROL__INPUT_CSC_GRPH_MODE__SHIFT) |
-		(0 << INPUT_CSC_CONTROL__INPUT_CSC_OVL_MODE__SHIFT)));
+	       ((INPUT_CSC_BYPASS << INPUT_CSC_CONTROL__INPUT_CSC_GRPH_MODE__SHIFT) |
+		(INPUT_CSC_BYPASS << INPUT_CSC_CONTROL__INPUT_CSC_OVL_MODE__SHIFT)));
 	WREG32(mmPRESCALE_GRPH_CONTROL + amdgpu_crtc->crtc_offset,
 	       PRESCALE_GRPH_CONTROL__GRPH_PRESCALE_BYPASS_MASK);
 	WREG32(mmPRESCALE_OVL_CONTROL + amdgpu_crtc->crtc_offset,
 	       PRESCALE_OVL_CONTROL__OVL_PRESCALE_BYPASS_MASK);
 	WREG32(mmINPUT_GAMMA_CONTROL + amdgpu_crtc->crtc_offset,
-	       ((0 << INPUT_GAMMA_CONTROL__GRPH_INPUT_GAMMA_MODE__SHIFT) |
-		(0 << INPUT_GAMMA_CONTROL__OVL_INPUT_GAMMA_MODE__SHIFT)));
+	       ((INPUT_GAMMA_USE_LUT << INPUT_GAMMA_CONTROL__GRPH_INPUT_GAMMA_MODE__SHIFT) |
+		(INPUT_GAMMA_USE_LUT << INPUT_GAMMA_CONTROL__OVL_INPUT_GAMMA_MODE__SHIFT)));
 
 	WREG32(mmDC_LUT_CONTROL + amdgpu_crtc->crtc_offset, 0);
 
@@ -2113,19 +2159,19 @@ static void dce_v6_0_crtc_load_lut(struct drm_crtc *crtc)
 	}
 
 	WREG32(mmDEGAMMA_CONTROL + amdgpu_crtc->crtc_offset,
-	       ((0 << DEGAMMA_CONTROL__GRPH_DEGAMMA_MODE__SHIFT) |
-		(0 << DEGAMMA_CONTROL__OVL_DEGAMMA_MODE__SHIFT) |
-		ICON_DEGAMMA_MODE(0) |
-		(0 << DEGAMMA_CONTROL__CURSOR_DEGAMMA_MODE__SHIFT)));
+	       ((DEGAMMA_BYPASS << DEGAMMA_CONTROL__GRPH_DEGAMMA_MODE__SHIFT) |
+		(DEGAMMA_BYPASS << DEGAMMA_CONTROL__OVL_DEGAMMA_MODE__SHIFT) |
+		(DEGAMMA_BYPASS << DEGAMMA_CONTROL__ICON_DEGAMMA_MODE__SHIFT) |
+		(DEGAMMA_BYPASS << DEGAMMA_CONTROL__CURSOR_DEGAMMA_MODE__SHIFT)));
 	WREG32(mmGAMUT_REMAP_CONTROL + amdgpu_crtc->crtc_offset,
-	       ((0 << GAMUT_REMAP_CONTROL__GRPH_GAMUT_REMAP_MODE__SHIFT) |
-		(0 << GAMUT_REMAP_CONTROL__OVL_GAMUT_REMAP_MODE__SHIFT)));
+	       ((GAMUT_REMAP_BYPASS << GAMUT_REMAP_CONTROL__GRPH_GAMUT_REMAP_MODE__SHIFT) |
+		(GAMUT_REMAP_BYPASS << GAMUT_REMAP_CONTROL__OVL_GAMUT_REMAP_MODE__SHIFT)));
 	WREG32(mmREGAMMA_CONTROL + amdgpu_crtc->crtc_offset,
-	       ((0 << REGAMMA_CONTROL__GRPH_REGAMMA_MODE__SHIFT) |
-		(0 << REGAMMA_CONTROL__OVL_REGAMMA_MODE__SHIFT)));
+	       ((REGAMMA_BYPASS << REGAMMA_CONTROL__GRPH_REGAMMA_MODE__SHIFT) |
+		(REGAMMA_BYPASS << REGAMMA_CONTROL__OVL_REGAMMA_MODE__SHIFT)));
 	WREG32(mmOUTPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset,
-	       ((0 << OUTPUT_CSC_CONTROL__OUTPUT_CSC_GRPH_MODE__SHIFT) |
-		(0 << OUTPUT_CSC_CONTROL__OUTPUT_CSC_OVL_MODE__SHIFT)));
+	       ((OUTPUT_CSC_BYPASS << OUTPUT_CSC_CONTROL__OUTPUT_CSC_GRPH_MODE__SHIFT) |
+		(OUTPUT_CSC_BYPASS << OUTPUT_CSC_CONTROL__OUTPUT_CSC_OVL_MODE__SHIFT)));
 	/* XXX match this to the depth of the crtc fmt block, move to modeset? */
 	WREG32(0x1a50 + amdgpu_crtc->crtc_offset, 0);
 
@@ -2171,7 +2217,7 @@ static u32 dce_v6_0_pick_pll(struct drm_crtc *crtc)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	u32 pll_in_use;
 	int pll;
 
@@ -2200,7 +2246,7 @@ static u32 dce_v6_0_pick_pll(struct drm_crtc *crtc)
 
 static void dce_v6_0_lock_cursor(struct drm_crtc *crtc, bool lock)
 {
-	struct amdgpu_device *adev = crtc->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	uint32_t cur_lock;
 
@@ -2215,37 +2261,34 @@ static void dce_v6_0_lock_cursor(struct drm_crtc *crtc, bool lock)
 static void dce_v6_0_hide_cursor(struct drm_crtc *crtc)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct amdgpu_device *adev = crtc->dev->dev_private;
-
-	WREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset,
-		   (CURSOR_24_8_PRE_MULT << CUR_CONTROL__CURSOR_MODE__SHIFT) |
-		   (CURSOR_URGENT_1_2 << CUR_CONTROL__CURSOR_URGENT_CONTROL__SHIFT));
-
+	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
 
+	WREG32(mmCUR_CONTROL + amdgpu_crtc->crtc_offset,
+	       (CURSOR_24_8_PRE_MULT << CUR_CONTROL__CURSOR_MODE__SHIFT) |
+	       (CURSOR_URGENT_1_2 << CUR_CONTROL__CURSOR_URGENT_CONTROL__SHIFT));
 }
 
 static void dce_v6_0_show_cursor(struct drm_crtc *crtc)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct amdgpu_device *adev = crtc->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
 
 	WREG32(mmCUR_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
 	       upper_32_bits(amdgpu_crtc->cursor_addr));
 	WREG32(mmCUR_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
 	       lower_32_bits(amdgpu_crtc->cursor_addr));
 
-	WREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset,
-		   CUR_CONTROL__CURSOR_EN_MASK |
-		   (CURSOR_24_8_PRE_MULT << CUR_CONTROL__CURSOR_MODE__SHIFT) |
-		   (CURSOR_URGENT_1_2 << CUR_CONTROL__CURSOR_URGENT_CONTROL__SHIFT));
-
+	WREG32(mmCUR_CONTROL + amdgpu_crtc->crtc_offset,
+	       CUR_CONTROL__CURSOR_EN_MASK |
+	       (CURSOR_24_8_PRE_MULT << CUR_CONTROL__CURSOR_MODE__SHIFT) |
+	       (CURSOR_URGENT_1_2 << CUR_CONTROL__CURSOR_URGENT_CONTROL__SHIFT));
 }
 
 static int dce_v6_0_cursor_move_locked(struct drm_crtc *crtc,
 				       int x, int y)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct amdgpu_device *adev = crtc->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
 	int xorigin = 0, yorigin = 0;
 
 	int w = amdgpu_crtc->cursor_width;
@@ -2322,17 +2365,19 @@ static int dce_v6_0_crtc_cursor_set2(struct drm_crtc *crtc,
 	aobj = gem_to_amdgpu_bo(obj);
 	ret = amdgpu_bo_reserve(aobj, false);
 	if (ret != 0) {
-		drm_gem_object_put_unlocked(obj);
+		drm_gem_object_put(obj);
 		return ret;
 	}
 
-	ret = amdgpu_bo_pin(aobj, AMDGPU_GEM_DOMAIN_VRAM, &amdgpu_crtc->cursor_addr);
+	aobj->flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+	ret = amdgpu_bo_pin(aobj, AMDGPU_GEM_DOMAIN_VRAM);
 	amdgpu_bo_unreserve(aobj);
 	if (ret) {
 		DRM_ERROR("Failed to pin new cursor BO (%d)\n", ret);
-		drm_gem_object_put_unlocked(obj);
+		drm_gem_object_put(obj);
 		return ret;
 	}
+	amdgpu_crtc->cursor_addr = amdgpu_bo_gpu_offset(aobj);
 
 	dce_v6_0_lock_cursor(crtc, true);
 
@@ -2364,7 +2409,7 @@ unpin:
 			amdgpu_bo_unpin(aobj);
 			amdgpu_bo_unreserve(aobj);
 		}
-		drm_gem_object_put_unlocked(amdgpu_crtc->cursor_bo);
+		drm_gem_object_put(amdgpu_crtc->cursor_bo);
 	}
 
 	amdgpu_crtc->cursor_bo = obj;
@@ -2407,15 +2452,19 @@ static const struct drm_crtc_funcs dce_v6_0_crtc_funcs = {
 	.cursor_set2 = dce_v6_0_crtc_cursor_set2,
 	.cursor_move = dce_v6_0_crtc_cursor_move,
 	.gamma_set = dce_v6_0_crtc_gamma_set,
-	.set_config = amdgpu_crtc_set_config,
+	.set_config = amdgpu_display_crtc_set_config,
 	.destroy = dce_v6_0_crtc_destroy,
-	.page_flip_target = amdgpu_crtc_page_flip_target,
+	.page_flip_target = amdgpu_display_crtc_page_flip_target,
+	.get_vblank_counter = amdgpu_get_vblank_counter_kms,
+	.enable_vblank = amdgpu_enable_vblank_kms,
+	.disable_vblank = amdgpu_disable_vblank_kms,
+	.get_vblank_timestamp = drm_crtc_vblank_helper_get_vblank_timestamp,
 };
 
 static void dce_v6_0_crtc_dpms(struct drm_crtc *crtc, int mode)
 {
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	unsigned type;
 
@@ -2425,7 +2474,8 @@ static void dce_v6_0_crtc_dpms(struct drm_crtc *crtc, int mode)
 		amdgpu_atombios_crtc_enable(crtc, ATOM_ENABLE);
 		amdgpu_atombios_crtc_blank(crtc, ATOM_DISABLE);
 		/* Make sure VBLANK and PFLIP interrupts are still enabled */
-		type = amdgpu_crtc_idx_to_irq_type(adev, amdgpu_crtc->crtc_id);
+		type = amdgpu_display_crtc_idx_to_irq_type(adev,
+						amdgpu_crtc->crtc_id);
 		amdgpu_irq_update(adev, &adev->crtc_irq, type);
 		amdgpu_irq_update(adev, &adev->pageflip_irq, type);
 		drm_crtc_vblank_on(crtc);
@@ -2442,7 +2492,7 @@ static void dce_v6_0_crtc_dpms(struct drm_crtc *crtc, int mode)
 		break;
 	}
 	/* adjust pm to dpms */
-	amdgpu_pm_compute_clocks(adev);
+	amdgpu_dpm_compute_clocks(adev);
 }
 
 static void dce_v6_0_crtc_prepare(struct drm_crtc *crtc)
@@ -2464,18 +2514,16 @@ static void dce_v6_0_crtc_disable(struct drm_crtc *crtc)
 
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_atom_ss ss;
 	int i;
 
 	dce_v6_0_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
 	if (crtc->primary->fb) {
 		int r;
-		struct amdgpu_framebuffer *amdgpu_fb;
 		struct amdgpu_bo *abo;
 
-		amdgpu_fb = to_amdgpu_framebuffer(crtc->primary->fb);
-		abo = gem_to_amdgpu_bo(amdgpu_fb->obj);
+		abo = gem_to_amdgpu_bo(crtc->primary->fb->obj[0]);
 		r = amdgpu_bo_reserve(abo, true);
 		if (unlikely(r))
 			DRM_ERROR("failed to reserve abo before unpin\n");
@@ -2544,7 +2592,6 @@ static bool dce_v6_0_crtc_mode_fixup(struct drm_crtc *crtc,
 				     const struct drm_display_mode *mode,
 				     struct drm_display_mode *adjusted_mode)
 {
-
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
 	struct drm_encoder *encoder;
@@ -2562,7 +2609,7 @@ static bool dce_v6_0_crtc_mode_fixup(struct drm_crtc *crtc,
 		amdgpu_crtc->connector = NULL;
 		return false;
 	}
-	if (!amdgpu_crtc_scaling_mode_fixup(crtc, mode, adjusted_mode))
+	if (!amdgpu_display_crtc_scaling_mode_fixup(crtc, mode, adjusted_mode))
 		return false;
 	if (amdgpu_atombios_crtc_prepare_pll(crtc, adjusted_mode))
 		return false;
@@ -2586,7 +2633,7 @@ static int dce_v6_0_crtc_set_base_atomic(struct drm_crtc *crtc,
 					 struct drm_framebuffer *fb,
 					 int x, int y, enum mode_set_atomic state)
 {
-       return dce_v6_0_crtc_do_set_base(crtc, fb, x, y, 1);
+	return dce_v6_0_crtc_do_set_base(crtc, fb, x, y, 1);
 }
 
 static const struct drm_crtc_helper_funcs dce_v6_0_crtc_helper_funcs = {
@@ -2598,6 +2645,33 @@ static const struct drm_crtc_helper_funcs dce_v6_0_crtc_helper_funcs = {
 	.prepare = dce_v6_0_crtc_prepare,
 	.commit = dce_v6_0_crtc_commit,
 	.disable = dce_v6_0_crtc_disable,
+	.get_scanout_position = amdgpu_crtc_get_scanout_position,
+};
+
+static void dce_v6_0_panic_flush(struct drm_plane *plane)
+{
+	struct drm_framebuffer *fb;
+	struct amdgpu_crtc *amdgpu_crtc;
+	struct amdgpu_device *adev;
+	uint32_t fb_format;
+
+	if (!plane->fb)
+		return;
+
+	fb = plane->fb;
+	amdgpu_crtc = to_amdgpu_crtc(plane->crtc);
+	adev = drm_to_adev(fb->dev);
+
+	/* Disable DC tiling */
+	fb_format = RREG32(mmGRPH_CONTROL + amdgpu_crtc->crtc_offset);
+	fb_format &= ~GRPH_CONTROL__GRPH_ARRAY_MODE_MASK;
+	WREG32(mmGRPH_CONTROL + amdgpu_crtc->crtc_offset, fb_format);
+
+}
+
+static const struct drm_plane_helper_funcs dce_v6_0_drm_primary_plane_helper_funcs = {
+	.get_scanout_buffer = amdgpu_display_get_scanout_buffer,
+	.panic_flush = dce_v6_0_panic_flush,
 };
 
 static int dce_v6_0_crtc_init(struct amdgpu_device *adev, int index)
@@ -2609,7 +2683,7 @@ static int dce_v6_0_crtc_init(struct amdgpu_device *adev, int index)
 	if (amdgpu_crtc == NULL)
 		return -ENOMEM;
 
-	drm_crtc_init(adev->ddev, &amdgpu_crtc->base, &dce_v6_0_crtc_funcs);
+	drm_crtc_init(adev_to_drm(adev), &amdgpu_crtc->base, &dce_v6_0_crtc_funcs);
 
 	drm_mode_crtc_set_gamma_size(&amdgpu_crtc->base, 256);
 	amdgpu_crtc->crtc_id = index;
@@ -2617,8 +2691,8 @@ static int dce_v6_0_crtc_init(struct amdgpu_device *adev, int index)
 
 	amdgpu_crtc->max_cursor_width = CURSOR_WIDTH;
 	amdgpu_crtc->max_cursor_height = CURSOR_HEIGHT;
-	adev->ddev->mode_config.cursor_width = amdgpu_crtc->max_cursor_width;
-	adev->ddev->mode_config.cursor_height = amdgpu_crtc->max_cursor_height;
+	adev_to_drm(adev)->mode_config.cursor_width = amdgpu_crtc->max_cursor_width;
+	adev_to_drm(adev)->mode_config.cursor_height = amdgpu_crtc->max_cursor_height;
 
 	amdgpu_crtc->crtc_offset = crtc_offsets[amdgpu_crtc->crtc_id];
 
@@ -2627,19 +2701,19 @@ static int dce_v6_0_crtc_init(struct amdgpu_device *adev, int index)
 	amdgpu_crtc->encoder = NULL;
 	amdgpu_crtc->connector = NULL;
 	drm_crtc_helper_add(&amdgpu_crtc->base, &dce_v6_0_crtc_helper_funcs);
+	drm_plane_helper_add(amdgpu_crtc->base.primary, &dce_v6_0_drm_primary_plane_helper_funcs);
 
 	return 0;
 }
 
-static int dce_v6_0_early_init(void *handle)
+static int dce_v6_0_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	adev->audio_endpt_rreg = &dce_v6_0_audio_endpt_rreg;
 	adev->audio_endpt_wreg = &dce_v6_0_audio_endpt_wreg;
 
 	dce_v6_0_set_display_funcs(adev);
-	dce_v6_0_set_irq_funcs(adev);
 
 	adev->mode_info.num_crtc = dce_v6_0_get_num_crtc(adev);
 
@@ -2658,48 +2732,49 @@ static int dce_v6_0_early_init(void *handle)
 		return -EINVAL;
 	}
 
+	dce_v6_0_set_irq_funcs(adev);
+
 	return 0;
 }
 
-static int dce_v6_0_sw_init(void *handle)
+static int dce_v6_0_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r, i;
-	bool ret;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->mode_info.num_crtc; i++) {
-		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i + 1, &adev->crtc_irq);
+		r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, i + 1, &adev->crtc_irq);
 		if (r)
 			return r;
 	}
 
 	for (i = 8; i < 20; i += 2) {
-		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i, &adev->pageflip_irq);
+		r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, i, &adev->pageflip_irq);
 		if (r)
 			return r;
 	}
 
 	/* HPD hotplug */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 42, &adev->hpd_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 42, &adev->hpd_irq);
 	if (r)
 		return r;
 
 	adev->mode_info.mode_config_initialized = true;
 
-	adev->ddev->mode_config.funcs = &amdgpu_mode_funcs;
-	adev->ddev->mode_config.async_page_flip = true;
-	adev->ddev->mode_config.max_width = 16384;
-	adev->ddev->mode_config.max_height = 16384;
-	adev->ddev->mode_config.preferred_depth = 24;
-	adev->ddev->mode_config.prefer_shadow = 1;
-	adev->ddev->mode_config.fb_base = adev->mc.aper_base;
+	adev_to_drm(adev)->mode_config.funcs = &amdgpu_mode_funcs;
+	adev_to_drm(adev)->mode_config.async_page_flip = true;
+	adev_to_drm(adev)->mode_config.max_width = 16384;
+	adev_to_drm(adev)->mode_config.max_height = 16384;
+	adev_to_drm(adev)->mode_config.preferred_depth = 24;
+	adev_to_drm(adev)->mode_config.prefer_shadow = 1;
+	adev_to_drm(adev)->mode_config.fb_modifiers_not_supported = true;
 
-	r = amdgpu_modeset_create_props(adev);
+	r = amdgpu_display_modeset_create_props(adev);
 	if (r)
 		return r;
 
-	adev->ddev->mode_config.max_width = 16384;
-	adev->ddev->mode_config.max_height = 16384;
+	adev_to_drm(adev)->mode_config.max_width = 16384;
+	adev_to_drm(adev)->mode_config.max_height = 16384;
 
 	/* allocate crtcs */
 	for (i = 0; i < adev->mode_info.num_crtc; i++) {
@@ -2708,9 +2783,8 @@ static int dce_v6_0_sw_init(void *handle)
 			return r;
 	}
 
-	ret = amdgpu_atombios_get_connector_info_from_object_table(adev);
-	if (ret)
-		amdgpu_print_display_setup(adev->ddev);
+	if (amdgpu_atombios_get_connector_info_from_object_table(adev))
+		amdgpu_display_print_display_setup(adev_to_drm(adev));
 	else
 		return -EINVAL;
 
@@ -2723,32 +2797,44 @@ static int dce_v6_0_sw_init(void *handle)
 	if (r)
 		return r;
 
-	drm_kms_helper_poll_init(adev->ddev);
+	/* Disable vblank IRQs aggressively for power-saving */
+	/* XXX: can this be enabled for DC? */
+	adev_to_drm(adev)->vblank_disable_immediate = true;
+
+	r = drm_vblank_init(adev_to_drm(adev), adev->mode_info.num_crtc);
+	if (r)
+		return r;
+
+	/* Pre-DCE11 */
+	INIT_DELAYED_WORK(&adev->hotplug_work,
+		  amdgpu_display_hotplug_work_func);
+
+	drm_kms_helper_poll_init(adev_to_drm(adev));
 
 	return r;
 }
 
-static int dce_v6_0_sw_fini(void *handle)
+static int dce_v6_0_sw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	kfree(adev->mode_info.bios_hardcoded_edid);
+	drm_edid_free(adev->mode_info.bios_hardcoded_edid);
 
-	drm_kms_helper_poll_fini(adev->ddev);
+	drm_kms_helper_poll_fini(adev_to_drm(adev));
 
 	dce_v6_0_audio_fini(adev);
 	dce_v6_0_afmt_fini(adev);
 
-	drm_mode_config_cleanup(adev->ddev);
+	drm_mode_config_cleanup(adev_to_drm(adev));
 	adev->mode_info.mode_config_initialized = false;
 
 	return 0;
 }
 
-static int dce_v6_0_hw_init(void *handle)
+static int dce_v6_0_hw_init(struct amdgpu_ip_block *ip_block)
 {
 	int i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	/* disable vga render */
 	dce_v6_0_set_vga_render_state(adev, false);
@@ -2768,10 +2854,10 @@ static int dce_v6_0_hw_init(void *handle)
 	return 0;
 }
 
-static int dce_v6_0_hw_fini(void *handle)
+static int dce_v6_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
 	int i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	dce_v6_0_hpd_fini(adev);
 
@@ -2781,20 +2867,34 @@ static int dce_v6_0_hw_fini(void *handle)
 
 	dce_v6_0_pageflip_interrupt_fini(adev);
 
+	flush_delayed_work(&adev->hotplug_work);
+
 	return 0;
 }
 
-static int dce_v6_0_suspend(void *handle)
+static int dce_v6_0_suspend(struct amdgpu_ip_block *ip_block)
 {
-	return dce_v6_0_hw_fini(handle);
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = amdgpu_display_suspend_helper(adev);
+	if (r)
+		return r;
+	adev->mode_info.bl_level =
+		amdgpu_atombios_encoder_get_backlight_level_from_reg(adev);
+
+	return dce_v6_0_hw_fini(ip_block);
 }
 
-static int dce_v6_0_resume(void *handle)
+static int dce_v6_0_resume(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int ret;
 
-	ret = dce_v6_0_hw_init(handle);
+	amdgpu_atombios_encoder_set_backlight_level_to_reg(adev,
+							   adev->mode_info.bl_level);
+
+	ret = dce_v6_0_hw_init(ip_block);
 
 	/* turn on the BL */
 	if (adev->mode_info.bl_encoder) {
@@ -2803,23 +2903,41 @@ static int dce_v6_0_resume(void *handle)
 		amdgpu_display_backlight_set_level(adev, adev->mode_info.bl_encoder,
 						    bl_level);
 	}
+	if (ret)
+		return ret;
 
-	return ret;
+	return amdgpu_display_resume_helper(adev);
 }
 
-static bool dce_v6_0_is_idle(void *handle)
+static bool dce_v6_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	return true;
 }
 
-static int dce_v6_0_wait_for_idle(void *handle)
+static int dce_v6_0_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	return 0;
-}
+	u32 srbm_soft_reset = 0, tmp;
+	struct amdgpu_device *adev = ip_block->adev;
 
-static int dce_v6_0_soft_reset(void *handle)
-{
-	DRM_INFO("xxxx: dce_v6_0_soft_reset --- no impl!!\n");
+	if (dce_v6_0_is_display_hung(adev))
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_DC_MASK;
+
+	if (srbm_soft_reset) {
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+	}
 	return 0;
 }
 
@@ -2836,22 +2954,22 @@ static void dce_v6_0_set_crtc_vblank_interrupt_state(struct amdgpu_device *adev,
 
 	switch (crtc) {
 	case 0:
-		reg_block = SI_CRTC0_REGISTER_OFFSET;
+		reg_block = CRTC0_REGISTER_OFFSET;
 		break;
 	case 1:
-		reg_block = SI_CRTC1_REGISTER_OFFSET;
+		reg_block = CRTC1_REGISTER_OFFSET;
 		break;
 	case 2:
-		reg_block = SI_CRTC2_REGISTER_OFFSET;
+		reg_block = CRTC2_REGISTER_OFFSET;
 		break;
 	case 3:
-		reg_block = SI_CRTC3_REGISTER_OFFSET;
+		reg_block = CRTC3_REGISTER_OFFSET;
 		break;
 	case 4:
-		reg_block = SI_CRTC4_REGISTER_OFFSET;
+		reg_block = CRTC4_REGISTER_OFFSET;
 		break;
 	case 5:
-		reg_block = SI_CRTC5_REGISTER_OFFSET;
+		reg_block = CRTC5_REGISTER_OFFSET;
 		break;
 	default:
 		DRM_DEBUG("invalid crtc %d\n", crtc);
@@ -2861,12 +2979,12 @@ static void dce_v6_0_set_crtc_vblank_interrupt_state(struct amdgpu_device *adev,
 	switch (state) {
 	case AMDGPU_IRQ_STATE_DISABLE:
 		interrupt_mask = RREG32(mmINT_MASK + reg_block);
-		interrupt_mask &= ~VBLANK_INT_MASK;
+		interrupt_mask &= ~INT_MASK__VBLANK_INT_MASK;
 		WREG32(mmINT_MASK + reg_block, interrupt_mask);
 		break;
 	case AMDGPU_IRQ_STATE_ENABLE:
 		interrupt_mask = RREG32(mmINT_MASK + reg_block);
-		interrupt_mask |= VBLANK_INT_MASK;
+		interrupt_mask |= INT_MASK__VBLANK_INT_MASK;
 		WREG32(mmINT_MASK + reg_block, interrupt_mask);
 		break;
 	default:
@@ -2881,28 +2999,28 @@ static void dce_v6_0_set_crtc_vline_interrupt_state(struct amdgpu_device *adev,
 
 }
 
-static int dce_v6_0_set_hpd_interrupt_state(struct amdgpu_device *adev,
+static int dce_v6_0_set_hpd_irq_state(struct amdgpu_device *adev,
 					    struct amdgpu_irq_src *src,
-					    unsigned type,
+					    unsigned hpd,
 					    enum amdgpu_interrupt_state state)
 {
 	u32 dc_hpd_int_cntl;
 
-	if (type >= adev->mode_info.num_hpd) {
-		DRM_DEBUG("invalid hdp %d\n", type);
+	if (hpd >= adev->mode_info.num_hpd) {
+		DRM_DEBUG("invalid hpd %d\n", hpd);
 		return 0;
 	}
 
 	switch (state) {
 	case AMDGPU_IRQ_STATE_DISABLE:
-		dc_hpd_int_cntl = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[type]);
-		dc_hpd_int_cntl &= ~DC_HPDx_INT_EN;
-		WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[type], dc_hpd_int_cntl);
+		dc_hpd_int_cntl = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd]);
+		dc_hpd_int_cntl &= ~DC_HPD1_INT_CONTROL__DC_HPD1_INT_EN_MASK;
+		WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd], dc_hpd_int_cntl);
 		break;
 	case AMDGPU_IRQ_STATE_ENABLE:
-		dc_hpd_int_cntl = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[type]);
-		dc_hpd_int_cntl |= DC_HPDx_INT_EN;
-		WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[type], dc_hpd_int_cntl);
+		dc_hpd_int_cntl = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd]);
+		dc_hpd_int_cntl |= DC_HPD1_INT_CONTROL__DC_HPD1_INT_EN_MASK;
+		WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd], dc_hpd_int_cntl);
 		break;
 	default:
 		break;
@@ -2911,7 +3029,7 @@ static int dce_v6_0_set_hpd_interrupt_state(struct amdgpu_device *adev,
 	return 0;
 }
 
-static int dce_v6_0_set_crtc_interrupt_state(struct amdgpu_device *adev,
+static int dce_v6_0_set_crtc_irq_state(struct amdgpu_device *adev,
 					     struct amdgpu_irq_src *src,
 					     unsigned type,
 					     enum amdgpu_interrupt_state state)
@@ -2965,23 +3083,24 @@ static int dce_v6_0_crtc_irq(struct amdgpu_device *adev,
 {
 	unsigned crtc = entry->src_id - 1;
 	uint32_t disp_int = RREG32(interrupt_status_offsets[crtc].reg);
-	unsigned irq_type = amdgpu_crtc_idx_to_irq_type(adev, crtc);
+	unsigned int irq_type = amdgpu_display_crtc_idx_to_irq_type(adev,
+								    crtc);
 
 	switch (entry->src_data[0]) {
 	case 0: /* vblank */
 		if (disp_int & interrupt_status_offsets[crtc].vblank)
-			WREG32(mmVBLANK_STATUS + crtc_offsets[crtc], VBLANK_ACK);
+			WREG32(mmVBLANK_STATUS + crtc_offsets[crtc], VBLANK_STATUS__VBLANK_ACK_MASK);
 		else
 			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
 
 		if (amdgpu_irq_enabled(adev, source, irq_type)) {
-			drm_handle_vblank(adev->ddev, crtc);
+			drm_handle_vblank(adev_to_drm(adev), crtc);
 		}
 		DRM_DEBUG("IH: D%d vblank\n", crtc + 1);
 		break;
 	case 1: /* vline */
 		if (disp_int & interrupt_status_offsets[crtc].vline)
-			WREG32(mmVLINE_STATUS + crtc_offsets[crtc], VLINE_ACK);
+			WREG32(mmVLINE_STATUS + crtc_offsets[crtc], VLINE_STATUS__VLINE_ACK_MASK);
 		else
 			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
 
@@ -2995,7 +3114,7 @@ static int dce_v6_0_crtc_irq(struct amdgpu_device *adev,
 	return 0;
 }
 
-static int dce_v6_0_set_pageflip_interrupt_state(struct amdgpu_device *adev,
+static int dce_v6_0_set_pageflip_irq_state(struct amdgpu_device *adev,
 						 struct amdgpu_irq_src *src,
 						 unsigned type,
 						 enum amdgpu_interrupt_state state)
@@ -3022,7 +3141,7 @@ static int dce_v6_0_pageflip_irq(struct amdgpu_device *adev,
 				 struct amdgpu_irq_src *source,
 				 struct amdgpu_iv_entry *entry)
 {
-		unsigned long flags;
+	unsigned long flags;
 	unsigned crtc_id;
 	struct amdgpu_crtc *amdgpu_crtc;
 	struct amdgpu_flip_work *works;
@@ -3044,14 +3163,14 @@ static int dce_v6_0_pageflip_irq(struct amdgpu_device *adev,
 	if (amdgpu_crtc == NULL)
 		return 0;
 
-	spin_lock_irqsave(&adev->ddev->event_lock, flags);
+	spin_lock_irqsave(&adev_to_drm(adev)->event_lock, flags);
 	works = amdgpu_crtc->pflip_works;
-	if (amdgpu_crtc->pflip_status != AMDGPU_FLIP_SUBMITTED){
+	if (amdgpu_crtc->pflip_status != AMDGPU_FLIP_SUBMITTED) {
 		DRM_DEBUG_DRIVER("amdgpu_crtc->pflip_status = %d != "
 						"AMDGPU_FLIP_SUBMITTED(%d)\n",
 						amdgpu_crtc->pflip_status,
 						AMDGPU_FLIP_SUBMITTED);
-		spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
+		spin_unlock_irqrestore(&adev_to_drm(adev)->event_lock, flags);
 		return 0;
 	}
 
@@ -3063,7 +3182,7 @@ static int dce_v6_0_pageflip_irq(struct amdgpu_device *adev,
 	if (works->event)
 		drm_crtc_send_vblank_event(&amdgpu_crtc->base, works->event);
 
-	spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
+	spin_unlock_irqrestore(&adev_to_drm(adev)->event_lock, flags);
 
 	drm_crtc_vblank_put(&amdgpu_crtc->base);
 	schedule_work(&works->unpin_work);
@@ -3075,7 +3194,7 @@ static int dce_v6_0_hpd_irq(struct amdgpu_device *adev,
 			    struct amdgpu_irq_src *source,
 			    struct amdgpu_iv_entry *entry)
 {
-	uint32_t disp_int, mask, tmp;
+	uint32_t disp_int, mask;
 	unsigned hpd;
 
 	if (entry->src_data[0] >= adev->mode_info.num_hpd) {
@@ -3088,24 +3207,21 @@ static int dce_v6_0_hpd_irq(struct amdgpu_device *adev,
 	mask = interrupt_status_offsets[hpd].hpd;
 
 	if (disp_int & mask) {
-		tmp = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd]);
-		tmp |= DC_HPD1_INT_CONTROL__DC_HPD1_INT_ACK_MASK;
-		WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd], tmp);
-		schedule_work(&adev->hotplug_work);
-		DRM_INFO("IH: HPD%d\n", hpd + 1);
+		dce_v6_0_hpd_int_ack(adev, hpd);
+		schedule_delayed_work(&adev->hotplug_work, 0);
+		DRM_DEBUG("IH: HPD%d\n", hpd + 1);
 	}
 
 	return 0;
-
 }
 
-static int dce_v6_0_set_clockgating_state(void *handle,
+static int dce_v6_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
 	return 0;
 }
 
-static int dce_v6_0_set_powergating_state(void *handle,
+static int dce_v6_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	return 0;
@@ -3114,7 +3230,6 @@ static int dce_v6_0_set_powergating_state(void *handle,
 static const struct amd_ip_funcs dce_v6_0_ip_funcs = {
 	.name = "dce_v6_0",
 	.early_init = dce_v6_0_early_init,
-	.late_init = NULL,
 	.sw_init = dce_v6_0_sw_init,
 	.sw_fini = dce_v6_0_sw_fini,
 	.hw_init = dce_v6_0_hw_init,
@@ -3122,18 +3237,15 @@ static const struct amd_ip_funcs dce_v6_0_ip_funcs = {
 	.suspend = dce_v6_0_suspend,
 	.resume = dce_v6_0_resume,
 	.is_idle = dce_v6_0_is_idle,
-	.wait_for_idle = dce_v6_0_wait_for_idle,
 	.soft_reset = dce_v6_0_soft_reset,
 	.set_clockgating_state = dce_v6_0_set_clockgating_state,
 	.set_powergating_state = dce_v6_0_set_powergating_state,
 };
 
-static void
-dce_v6_0_encoder_mode_set(struct drm_encoder *encoder,
+static void dce_v6_0_encoder_mode_set(struct drm_encoder *encoder,
 			  struct drm_display_mode *mode,
 			  struct drm_display_mode *adjusted_mode)
 {
-
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	int em = amdgpu_atombios_encoder_get_encoder_mode(encoder);
 
@@ -3153,8 +3265,7 @@ dce_v6_0_encoder_mode_set(struct drm_encoder *encoder,
 
 static void dce_v6_0_encoder_prepare(struct drm_encoder *encoder)
 {
-
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(encoder->dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
 
@@ -3193,9 +3304,8 @@ static void dce_v6_0_encoder_prepare(struct drm_encoder *encoder)
 
 static void dce_v6_0_encoder_commit(struct drm_encoder *encoder)
 {
-
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 
 	/* need to call this here as we need the crtc set up */
 	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_ON);
@@ -3204,7 +3314,6 @@ static void dce_v6_0_encoder_commit(struct drm_encoder *encoder)
 
 static void dce_v6_0_encoder_disable(struct drm_encoder *encoder)
 {
-
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig;
 	int em = amdgpu_atombios_encoder_get_encoder_mode(encoder);
@@ -3231,8 +3340,7 @@ static void dce_v6_0_ext_commit(struct drm_encoder *encoder)
 
 }
 
-static void
-dce_v6_0_ext_mode_set(struct drm_encoder *encoder,
+static void dce_v6_0_ext_mode_set(struct drm_encoder *encoder,
 		      struct drm_display_mode *mode,
 		      struct drm_display_mode *adjusted_mode)
 {
@@ -3244,8 +3352,7 @@ static void dce_v6_0_ext_disable(struct drm_encoder *encoder)
 
 }
 
-static void
-dce_v6_0_ext_dpms(struct drm_encoder *encoder, int mode)
+static void dce_v6_0_ext_dpms(struct drm_encoder *encoder, int mode)
 {
 
 }
@@ -3305,7 +3412,7 @@ static void dce_v6_0_encoder_add(struct amdgpu_device *adev,
 				 uint32_t supported_device,
 				 u16 caps)
 {
-	struct drm_device *dev = adev->ddev;
+	struct drm_device *dev = adev_to_drm(adev);
 	struct drm_encoder *encoder;
 	struct amdgpu_encoder *amdgpu_encoder;
 
@@ -3316,7 +3423,6 @@ static void dce_v6_0_encoder_add(struct amdgpu_device *adev,
 			amdgpu_encoder->devices |= supported_device;
 			return;
 		}
-
 	}
 
 	/* add a new one */
@@ -3406,7 +3512,6 @@ static void dce_v6_0_encoder_add(struct amdgpu_device *adev,
 static const struct amdgpu_display_funcs dce_v6_0_display_funcs = {
 	.bandwidth_update = &dce_v6_0_bandwidth_update,
 	.vblank_get_counter = &dce_v6_0_vblank_get_counter,
-	.vblank_wait = &dce_v6_0_vblank_wait,
 	.backlight_set_level = &amdgpu_atombios_encoder_set_backlight_level,
 	.backlight_get_level = &amdgpu_atombios_encoder_get_backlight_level,
 	.hpd_sense = &dce_v6_0_hpd_sense,
@@ -3420,34 +3525,36 @@ static const struct amdgpu_display_funcs dce_v6_0_display_funcs = {
 
 static void dce_v6_0_set_display_funcs(struct amdgpu_device *adev)
 {
-	if (adev->mode_info.funcs == NULL)
-		adev->mode_info.funcs = &dce_v6_0_display_funcs;
+	adev->mode_info.funcs = &dce_v6_0_display_funcs;
 }
 
 static const struct amdgpu_irq_src_funcs dce_v6_0_crtc_irq_funcs = {
-	.set = dce_v6_0_set_crtc_interrupt_state,
+	.set = dce_v6_0_set_crtc_irq_state,
 	.process = dce_v6_0_crtc_irq,
 };
 
 static const struct amdgpu_irq_src_funcs dce_v6_0_pageflip_irq_funcs = {
-	.set = dce_v6_0_set_pageflip_interrupt_state,
+	.set = dce_v6_0_set_pageflip_irq_state,
 	.process = dce_v6_0_pageflip_irq,
 };
 
 static const struct amdgpu_irq_src_funcs dce_v6_0_hpd_irq_funcs = {
-	.set = dce_v6_0_set_hpd_interrupt_state,
+	.set = dce_v6_0_set_hpd_irq_state,
 	.process = dce_v6_0_hpd_irq,
 };
 
 static void dce_v6_0_set_irq_funcs(struct amdgpu_device *adev)
 {
-	adev->crtc_irq.num_types = AMDGPU_CRTC_IRQ_LAST;
+	if (adev->mode_info.num_crtc > 0)
+		adev->crtc_irq.num_types = AMDGPU_CRTC_IRQ_VLINE1 + adev->mode_info.num_crtc;
+	else
+		adev->crtc_irq.num_types = 0;
 	adev->crtc_irq.funcs = &dce_v6_0_crtc_irq_funcs;
 
-	adev->pageflip_irq.num_types = AMDGPU_PAGEFLIP_IRQ_LAST;
+	adev->pageflip_irq.num_types = adev->mode_info.num_crtc;
 	adev->pageflip_irq.funcs = &dce_v6_0_pageflip_irq_funcs;
 
-	adev->hpd_irq.num_types = AMDGPU_HPD_LAST;
+	adev->hpd_irq.num_types = adev->mode_info.num_hpd;
 	adev->hpd_irq.funcs = &dce_v6_0_hpd_irq_funcs;
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
index 9cf14b8b2db9..2ccd6aad8dd6 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
@@ -20,7 +20,13 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  *
  */
-#include <drm/drmP.h>
+
+#include <drm/drm_edid.h>
+#include <drm/drm_fourcc.h>
+#include <drm/drm_modeset_helper.h>
+#include <drm/drm_modeset_helper_vtables.h>
+#include <drm/drm_vblank.h>
+
 #include "amdgpu.h"
 #include "amdgpu_pm.h"
 #include "amdgpu_i2c.h"
@@ -31,6 +37,7 @@
 #include "atombios_encoders.h"
 #include "amdgpu_pll.h"
 #include "amdgpu_connectors.h"
+#include "amdgpu_display.h"
 #include "dce_v8_0.h"
 
 #include "dce/dce_8_0_d.h"
@@ -47,8 +54,7 @@
 static void dce_v8_0_set_display_funcs(struct amdgpu_device *adev);
 static void dce_v8_0_set_irq_funcs(struct amdgpu_device *adev);
 
-static const u32 crtc_offsets[6] =
-{
+static const u32 crtc_offsets[6] = {
 	CRTC0_REGISTER_OFFSET,
 	CRTC1_REGISTER_OFFSET,
 	CRTC2_REGISTER_OFFSET,
@@ -57,8 +63,7 @@ static const u32 crtc_offsets[6] =
 	CRTC5_REGISTER_OFFSET
 };
 
-static const u32 hpd_offsets[] =
-{
+static const u32 hpd_offsets[] = {
 	HPD0_REGISTER_OFFSET,
 	HPD1_REGISTER_OFFSET,
 	HPD2_REGISTER_OFFSET,
@@ -140,66 +145,6 @@ static void dce_v8_0_audio_endpt_wreg(struct amdgpu_device *adev,
 	spin_unlock_irqrestore(&adev->audio_endpt_idx_lock, flags);
 }
 
-static bool dce_v8_0_is_in_vblank(struct amdgpu_device *adev, int crtc)
-{
-	if (RREG32(mmCRTC_STATUS + crtc_offsets[crtc]) &
-			CRTC_V_BLANK_START_END__CRTC_V_BLANK_START_MASK)
-		return true;
-	else
-		return false;
-}
-
-static bool dce_v8_0_is_counter_moving(struct amdgpu_device *adev, int crtc)
-{
-	u32 pos1, pos2;
-
-	pos1 = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
-	pos2 = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
-
-	if (pos1 != pos2)
-		return true;
-	else
-		return false;
-}
-
-/**
- * dce_v8_0_vblank_wait - vblank wait asic callback.
- *
- * @adev: amdgpu_device pointer
- * @crtc: crtc to wait for vblank on
- *
- * Wait for vblank on the requested crtc (evergreen+).
- */
-static void dce_v8_0_vblank_wait(struct amdgpu_device *adev, int crtc)
-{
-	unsigned i = 100;
-
-	if (crtc >= adev->mode_info.num_crtc)
-		return;
-
-	if (!(RREG32(mmCRTC_CONTROL + crtc_offsets[crtc]) & CRTC_CONTROL__CRTC_MASTER_EN_MASK))
-		return;
-
-	/* depending on when we hit vblank, we may be close to active; if so,
-	 * wait for another frame.
-	 */
-	while (dce_v8_0_is_in_vblank(adev, crtc)) {
-		if (i++ == 100) {
-			i = 0;
-			if (!dce_v8_0_is_counter_moving(adev, crtc))
-				break;
-		}
-	}
-
-	while (!dce_v8_0_is_in_vblank(adev, crtc)) {
-		if (i++ == 100) {
-			i = 0;
-			if (!dce_v8_0_is_counter_moving(adev, crtc))
-				break;
-		}
-	}
-}
-
 static u32 dce_v8_0_vblank_get_counter(struct amdgpu_device *adev, int crtc)
 {
 	if (crtc >= adev->mode_info.num_crtc)
@@ -232,6 +177,7 @@ static void dce_v8_0_pageflip_interrupt_fini(struct amdgpu_device *adev)
  * @adev: amdgpu_device pointer
  * @crtc_id: crtc to cleanup pageflip on
  * @crtc_base: new address of the crtc (GPU MC address)
+ * @async: asynchronous flip
  *
  * Triggers the actual pageflip by updating the primary
  * surface base address.
@@ -240,10 +186,14 @@ static void dce_v8_0_page_flip(struct amdgpu_device *adev,
 			       int crtc_id, u64 crtc_base, bool async)
 {
 	struct amdgpu_crtc *amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
+	struct drm_framebuffer *fb = amdgpu_crtc->base.primary->fb;
 
 	/* flip at hsync for async, default is vsync */
 	WREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset, async ?
 	       GRPH_FLIP_CONTROL__GRPH_SURFACE_UPDATE_H_RETRACE_EN_MASK : 0);
+	/* update pitch */
+	WREG32(mmGRPH_PITCH + amdgpu_crtc->crtc_offset,
+	       fb->pitches[0] / fb->format->cpp[0]);
 	/* update the primary scanout addresses */
 	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
 	       upper_32_bits(crtc_base));
@@ -315,6 +265,21 @@ static void dce_v8_0_hpd_set_polarity(struct amdgpu_device *adev,
 	WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd], tmp);
 }
 
+static void dce_v8_0_hpd_int_ack(struct amdgpu_device *adev,
+				 int hpd)
+{
+	u32 tmp;
+
+	if (hpd >= adev->mode_info.num_hpd) {
+		DRM_DEBUG("invalid hpd %d\n", hpd);
+		return;
+	}
+
+	tmp = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd]);
+	tmp |= DC_HPD1_INT_CONTROL__DC_HPD1_INT_ACK_MASK;
+	WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd], tmp);
+}
+
 /**
  * dce_v8_0_hpd_init - hpd setup callback.
  *
@@ -325,11 +290,13 @@ static void dce_v8_0_hpd_set_polarity(struct amdgpu_device *adev,
  */
 static void dce_v8_0_hpd_init(struct amdgpu_device *adev)
 {
-	struct drm_device *dev = adev->ddev;
+	struct drm_device *dev = adev_to_drm(adev);
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	u32 tmp;
 
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 
 		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
@@ -352,9 +319,11 @@ static void dce_v8_0_hpd_init(struct amdgpu_device *adev)
 			continue;
 		}
 
+		dce_v8_0_hpd_int_ack(adev, amdgpu_connector->hpd.hpd);
 		dce_v8_0_hpd_set_polarity(adev, amdgpu_connector->hpd.hpd);
 		amdgpu_irq_get(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
 	}
+	drm_connector_list_iter_end(&iter);
 }
 
 /**
@@ -367,11 +336,13 @@ static void dce_v8_0_hpd_init(struct amdgpu_device *adev)
  */
 static void dce_v8_0_hpd_fini(struct amdgpu_device *adev)
 {
-	struct drm_device *dev = adev->ddev;
+	struct drm_device *dev = adev_to_drm(adev);
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	u32 tmp;
 
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 
 		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
@@ -379,10 +350,11 @@ static void dce_v8_0_hpd_fini(struct amdgpu_device *adev)
 
 		tmp = RREG32(mmDC_HPD1_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd]);
 		tmp &= ~DC_HPD1_CONTROL__DC_HPD1_EN_MASK;
-		WREG32(mmDC_HPD1_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd], 0);
+		WREG32(mmDC_HPD1_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
 
 		amdgpu_irq_put(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
 	}
+	drm_connector_list_iter_end(&iter);
 }
 
 static u32 dce_v8_0_hpd_get_gpio_reg(struct amdgpu_device *adev)
@@ -490,7 +462,7 @@ void dce_v8_0_disable_dce(struct amdgpu_device *adev)
 static void dce_v8_0_program_fmt(struct drm_encoder *encoder)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
 	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
@@ -1020,7 +992,7 @@ static void dce_v8_0_program_watermarks(struct amdgpu_device *adev,
 					    (u32)mode->clock);
 		line_time = (u32) div_u64((u64)mode->crtc_htotal * 1000000,
 					  (u32)mode->clock);
-		line_time = min(line_time, (u32)65535);
+		line_time = min_t(u32, line_time, 65535);
 
 		/* watermark for high clocks */
 		if (adev->pm.dpm_enabled) {
@@ -1050,7 +1022,7 @@ static void dce_v8_0_program_watermarks(struct amdgpu_device *adev,
 		wm_high.num_heads = num_heads;
 
 		/* set for high clocks */
-		latency_watermark_a = min(dce_v8_0_latency_watermark(&wm_high), (u32)65535);
+		latency_watermark_a = min_t(u32, dce_v8_0_latency_watermark(&wm_high), 65535);
 
 		/* possibly force display priority to high */
 		/* should really do this at mode validation time... */
@@ -1089,7 +1061,7 @@ static void dce_v8_0_program_watermarks(struct amdgpu_device *adev,
 		wm_low.num_heads = num_heads;
 
 		/* set for low clocks */
-		latency_watermark_b = min(dce_v8_0_latency_watermark(&wm_low), (u32)65535);
+		latency_watermark_b = min_t(u32, dce_v8_0_latency_watermark(&wm_low), 65535);
 
 		/* possibly force display priority to high */
 		/* should really do this at mode validation time... */
@@ -1124,8 +1096,7 @@ static void dce_v8_0_program_watermarks(struct amdgpu_device *adev,
 
 	/* save values for DPM */
 	amdgpu_crtc->line_time = line_time;
-	amdgpu_crtc->wm_high = latency_watermark_a;
-	amdgpu_crtc->wm_low = latency_watermark_b;
+
 	/* Save number of lines the linebuffer leads before the scanout */
 	amdgpu_crtc->lb_vblank_lead_lines = lb_vblank_lead_lines;
 }
@@ -1144,7 +1115,7 @@ static void dce_v8_0_bandwidth_update(struct amdgpu_device *adev)
 	u32 num_heads = 0, lb_size;
 	int i;
 
-	amdgpu_update_display_priority(adev);
+	amdgpu_display_update_priority(adev);
 
 	for (i = 0; i < adev->mode_info.num_crtc; i++) {
 		if (adev->mode_info.crtcs[i]->base.enabled)
@@ -1192,7 +1163,7 @@ static struct amdgpu_audio_pin *dce_v8_0_audio_get_pin(struct amdgpu_device *ade
 
 static void dce_v8_0_afmt_audio_select_pin(struct drm_encoder *encoder)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(encoder->dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	u32 offset;
@@ -1209,10 +1180,12 @@ static void dce_v8_0_afmt_audio_select_pin(struct drm_encoder *encoder)
 static void dce_v8_0_audio_write_latency_fields(struct drm_encoder *encoder,
 						struct drm_display_mode *mode)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	u32 tmp = 0, offset;
 
@@ -1221,12 +1194,14 @@ static void dce_v8_0_audio_write_latency_fields(struct drm_encoder *encoder,
 
 	offset = dig->afmt->pin->offset;
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
@@ -1266,10 +1241,12 @@ static void dce_v8_0_audio_write_latency_fields(struct drm_encoder *encoder,
 
 static void dce_v8_0_audio_write_speaker_allocation(struct drm_encoder *encoder)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	u32 offset, tmp;
 	u8 *sadb = NULL;
@@ -1280,19 +1257,21 @@ static void dce_v8_0_audio_write_speaker_allocation(struct drm_encoder *encoder)
 
 	offset = dig->afmt->pin->offset;
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
 		return;
 	}
 
-	sad_count = drm_edid_to_speaker_allocation(amdgpu_connector_edid(connector), &sadb);
+	sad_count = drm_edid_to_speaker_allocation(amdgpu_connector->edid, &sadb);
 	if (sad_count < 0) {
 		DRM_ERROR("Couldn't read Speaker Allocation Data Block: %d\n", sad_count);
 		sad_count = 0;
@@ -1315,11 +1294,13 @@ static void dce_v8_0_audio_write_speaker_allocation(struct drm_encoder *encoder)
 
 static void dce_v8_0_audio_write_sad_regs(struct drm_encoder *encoder)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	u32 offset;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	struct cea_sad *sads;
 	int i, sad_count;
@@ -1344,23 +1325,25 @@ static void dce_v8_0_audio_write_sad_regs(struct drm_encoder *encoder)
 
 	offset = dig->afmt->pin->offset;
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
 		return;
 	}
 
-	sad_count = drm_edid_to_sad(amdgpu_connector_edid(connector), &sads);
-	if (sad_count <= 0) {
+	sad_count = drm_edid_to_sad(amdgpu_connector->edid, &sads);
+	if (sad_count < 0)
 		DRM_ERROR("Couldn't read SADs: %d\n", sad_count);
+	if (sad_count <= 0)
 		return;
-	}
 	BUG_ON(!sads);
 
 	for (i = 0; i < ARRAY_SIZE(eld_reg_to_type); i++) {
@@ -1376,9 +1359,9 @@ static void dce_v8_0_audio_write_sad_regs(struct drm_encoder *encoder)
 				if (sad->channels > max_channels) {
 					value = (sad->channels <<
 						 AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__MAX_CHANNELS__SHIFT) |
-					        (sad->byte2 <<
+						(sad->byte2 <<
 						 AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__DESCRIPTOR_BYTE_2__SHIFT) |
-					        (sad->freq <<
+						(sad->freq <<
 						 AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__SUPPORTED_FREQUENCIES__SHIFT);
 					max_channels = sad->channels;
 				}
@@ -1410,15 +1393,14 @@ static void dce_v8_0_audio_enable(struct amdgpu_device *adev,
 		enable ? AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL__AUDIO_ENABLED_MASK : 0);
 }
 
-static const u32 pin_offsets[7] =
-{
-	(0x1780 - 0x1780),
-	(0x1786 - 0x1780),
-	(0x178c - 0x1780),
-	(0x1792 - 0x1780),
-	(0x1798 - 0x1780),
-	(0x179d - 0x1780),
-	(0x17a4 - 0x1780),
+static const u32 pin_offsets[7] = {
+	AUD0_REGISTER_OFFSET,
+	AUD1_REGISTER_OFFSET,
+	AUD2_REGISTER_OFFSET,
+	AUD3_REGISTER_OFFSET,
+	AUD4_REGISTER_OFFSET,
+	AUD5_REGISTER_OFFSET,
+	AUD6_REGISTER_OFFSET,
 };
 
 static int dce_v8_0_audio_init(struct amdgpu_device *adev)
@@ -1460,17 +1442,12 @@ static int dce_v8_0_audio_init(struct amdgpu_device *adev)
 
 static void dce_v8_0_audio_fini(struct amdgpu_device *adev)
 {
-	int i;
-
 	if (!amdgpu_audio)
 		return;
 
 	if (!adev->mode_info.audio.enabled)
 		return;
 
-	for (i = 0; i < adev->mode_info.audio.num_pins; i++)
-		dce_v8_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
-
 	adev->mode_info.audio.enabled = false;
 }
 
@@ -1480,7 +1457,7 @@ static void dce_v8_0_audio_fini(struct amdgpu_device *adev)
 static void dce_v8_0_afmt_update_ACR(struct drm_encoder *encoder, uint32_t clock)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_afmt_acr acr = amdgpu_afmt_acr(clock);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
@@ -1503,7 +1480,7 @@ static void dce_v8_0_afmt_update_avi_infoframe(struct drm_encoder *encoder,
 					       void *buffer, size_t size)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	uint32_t offset = dig->afmt->offset;
@@ -1523,7 +1500,7 @@ static void dce_v8_0_afmt_update_avi_infoframe(struct drm_encoder *encoder,
 static void dce_v8_0_audio_set_dto(struct drm_encoder *encoder, u32 clock)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
@@ -1550,7 +1527,7 @@ static void dce_v8_0_afmt_setmode(struct drm_encoder *encoder,
 				  struct drm_display_mode *mode)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
@@ -1675,7 +1652,7 @@ static void dce_v8_0_afmt_setmode(struct drm_encoder *encoder,
 	dce_v8_0_audio_write_sad_regs(encoder);
 	dce_v8_0_audio_write_latency_fields(encoder, mode);
 
-	err = drm_hdmi_avi_infoframe_from_display_mode(&frame, mode, false);
+	err = drm_hdmi_avi_infoframe_from_display_mode(&frame, connector, mode);
 	if (err < 0) {
 		DRM_ERROR("failed to setup AVI infoframe: %zd\n", err);
 		return;
@@ -1712,7 +1689,7 @@ static void dce_v8_0_afmt_setmode(struct drm_encoder *encoder,
 static void dce_v8_0_afmt_enable(struct drm_encoder *encoder, bool enable)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 
@@ -1771,8 +1748,7 @@ static void dce_v8_0_afmt_fini(struct amdgpu_device *adev)
 	}
 }
 
-static const u32 vga_control_regs[6] =
-{
+static const u32 vga_control_regs[6] = {
 	mmD1VGA_CONTROL,
 	mmD2VGA_CONTROL,
 	mmD3VGA_CONTROL,
@@ -1785,7 +1761,7 @@ static void dce_v8_0_vga_enable(struct drm_crtc *crtc, bool enable)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	u32 vga_control;
 
 	vga_control = RREG32(vga_control_regs[amdgpu_crtc->crtc_id]) & ~1;
@@ -1799,7 +1775,7 @@ static void dce_v8_0_grph_enable(struct drm_crtc *crtc, bool enable)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 
 	if (enable)
 		WREG32(mmGRPH_ENABLE + amdgpu_crtc->crtc_offset, 1);
@@ -1813,8 +1789,7 @@ static int dce_v8_0_crtc_do_set_base(struct drm_crtc *crtc,
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-	struct amdgpu_framebuffer *amdgpu_fb;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct drm_framebuffer *target_fb;
 	struct drm_gem_object *obj;
 	struct amdgpu_bo *abo;
@@ -1825,7 +1800,6 @@ static int dce_v8_0_crtc_do_set_base(struct drm_crtc *crtc,
 	u32 viewport_w, viewport_h;
 	int r;
 	bool bypass_lut = false;
-	struct drm_format_name_buf format_name;
 
 	/* no fb bound */
 	if (!atomic && !crtc->primary->fb) {
@@ -1833,32 +1807,29 @@ static int dce_v8_0_crtc_do_set_base(struct drm_crtc *crtc,
 		return 0;
 	}
 
-	if (atomic) {
-		amdgpu_fb = to_amdgpu_framebuffer(fb);
+	if (atomic)
 		target_fb = fb;
-	} else {
-		amdgpu_fb = to_amdgpu_framebuffer(crtc->primary->fb);
+	else
 		target_fb = crtc->primary->fb;
-	}
 
 	/* If atomic, assume fb object is pinned & idle & fenced and
 	 * just update base pointers
 	 */
-	obj = amdgpu_fb->obj;
+	obj = target_fb->obj[0];
 	abo = gem_to_amdgpu_bo(obj);
 	r = amdgpu_bo_reserve(abo, false);
 	if (unlikely(r != 0))
 		return r;
 
-	if (atomic) {
-		fb_location = amdgpu_bo_gpu_offset(abo);
-	} else {
-		r = amdgpu_bo_pin(abo, AMDGPU_GEM_DOMAIN_VRAM, &fb_location);
+	if (!atomic) {
+		abo->flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+		r = amdgpu_bo_pin(abo, AMDGPU_GEM_DOMAIN_VRAM);
 		if (unlikely(r != 0)) {
 			amdgpu_bo_unreserve(abo);
 			return -EINVAL;
 		}
 	}
+	fb_location = amdgpu_bo_gpu_offset(abo);
 
 	amdgpu_bo_get_tiling_flags(abo, &tiling_flags);
 	amdgpu_bo_unreserve(abo);
@@ -1929,9 +1900,19 @@ static int dce_v8_0_crtc_do_set_base(struct drm_crtc *crtc,
 		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
 		bypass_lut = true;
 		break;
+	case DRM_FORMAT_XBGR8888:
+	case DRM_FORMAT_ABGR8888:
+		fb_format = ((GRPH_DEPTH_32BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+				(GRPH_FORMAT_ARGB8888 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
+		fb_swap = ((GRPH_RED_SEL_B << GRPH_SWAP_CNTL__GRPH_RED_CROSSBAR__SHIFT) |
+			(GRPH_BLUE_SEL_R << GRPH_SWAP_CNTL__GRPH_BLUE_CROSSBAR__SHIFT));
+#ifdef __BIG_ENDIAN
+		fb_swap |= (GRPH_ENDIAN_8IN32 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
+#endif
+		break;
 	default:
-		DRM_ERROR("Unsupported screen format %s\n",
-		          drm_get_format_name(target_fb->format->format, &format_name));
+		DRM_ERROR("Unsupported screen format %p4cc\n",
+			  &target_fb->format->format);
 		return -EINVAL;
 	}
 
@@ -2015,8 +1996,7 @@ static int dce_v8_0_crtc_do_set_base(struct drm_crtc *crtc,
 	WREG32(mmMASTER_UPDATE_MODE + amdgpu_crtc->crtc_offset, 0);
 
 	if (!atomic && fb && fb != crtc->primary->fb) {
-		amdgpu_fb = to_amdgpu_framebuffer(fb);
-		abo = gem_to_amdgpu_bo(amdgpu_fb->obj);
+		abo = gem_to_amdgpu_bo(fb->obj[0]);
 		r = amdgpu_bo_reserve(abo, true);
 		if (unlikely(r != 0))
 			return r;
@@ -2034,7 +2014,7 @@ static void dce_v8_0_set_interleave(struct drm_crtc *crtc,
 				    struct drm_display_mode *mode)
 {
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 
 	if (mode->flags & DRM_MODE_FLAG_INTERLACE)
@@ -2048,7 +2028,7 @@ static void dce_v8_0_crtc_load_lut(struct drm_crtc *crtc)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	u16 *r, *g, *b;
 	int i;
 
@@ -2122,22 +2102,18 @@ static int dce_v8_0_pick_dig_encoder(struct drm_encoder *encoder)
 			return 1;
 		else
 			return 0;
-		break;
 	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
 		if (dig->linkb)
 			return 3;
 		else
 			return 2;
-		break;
 	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
 		if (dig->linkb)
 			return 5;
 		else
 			return 4;
-		break;
 	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
 		return 6;
-		break;
 	default:
 		DRM_ERROR("invalid encoder_id: 0x%x\n", amdgpu_encoder->encoder_id);
 		return 0;
@@ -2170,7 +2146,7 @@ static u32 dce_v8_0_pick_pll(struct drm_crtc *crtc)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	u32 pll_in_use;
 	int pll;
 
@@ -2218,7 +2194,7 @@ static u32 dce_v8_0_pick_pll(struct drm_crtc *crtc)
 
 static void dce_v8_0_lock_cursor(struct drm_crtc *crtc, bool lock)
 {
-	struct amdgpu_device *adev = crtc->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	uint32_t cur_lock;
 
@@ -2233,34 +2209,34 @@ static void dce_v8_0_lock_cursor(struct drm_crtc *crtc, bool lock)
 static void dce_v8_0_hide_cursor(struct drm_crtc *crtc)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct amdgpu_device *adev = crtc->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
 
-	WREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset,
-		   (CURSOR_24_8_PRE_MULT << CUR_CONTROL__CURSOR_MODE__SHIFT) |
-		   (CURSOR_URGENT_1_2 << CUR_CONTROL__CURSOR_URGENT_CONTROL__SHIFT));
+	WREG32(mmCUR_CONTROL + amdgpu_crtc->crtc_offset,
+	       (CURSOR_24_8_PRE_MULT << CUR_CONTROL__CURSOR_MODE__SHIFT) |
+	       (CURSOR_URGENT_1_2 << CUR_CONTROL__CURSOR_URGENT_CONTROL__SHIFT));
 }
 
 static void dce_v8_0_show_cursor(struct drm_crtc *crtc)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct amdgpu_device *adev = crtc->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
 
 	WREG32(mmCUR_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
 	       upper_32_bits(amdgpu_crtc->cursor_addr));
 	WREG32(mmCUR_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
 	       lower_32_bits(amdgpu_crtc->cursor_addr));
 
-	WREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset,
-		   CUR_CONTROL__CURSOR_EN_MASK |
-		   (CURSOR_24_8_PRE_MULT << CUR_CONTROL__CURSOR_MODE__SHIFT) |
-		   (CURSOR_URGENT_1_2 << CUR_CONTROL__CURSOR_URGENT_CONTROL__SHIFT));
+	WREG32(mmCUR_CONTROL + amdgpu_crtc->crtc_offset,
+	       CUR_CONTROL__CURSOR_EN_MASK |
+	       (CURSOR_24_8_PRE_MULT << CUR_CONTROL__CURSOR_MODE__SHIFT) |
+	       (CURSOR_URGENT_1_2 << CUR_CONTROL__CURSOR_URGENT_CONTROL__SHIFT));
 }
 
 static int dce_v8_0_cursor_move_locked(struct drm_crtc *crtc,
 				       int x, int y)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct amdgpu_device *adev = crtc->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
 	int xorigin = 0, yorigin = 0;
 
 	amdgpu_crtc->cursor_x = x;
@@ -2335,17 +2311,19 @@ static int dce_v8_0_crtc_cursor_set2(struct drm_crtc *crtc,
 	aobj = gem_to_amdgpu_bo(obj);
 	ret = amdgpu_bo_reserve(aobj, false);
 	if (ret != 0) {
-		drm_gem_object_put_unlocked(obj);
+		drm_gem_object_put(obj);
 		return ret;
 	}
 
-	ret = amdgpu_bo_pin(aobj, AMDGPU_GEM_DOMAIN_VRAM, &amdgpu_crtc->cursor_addr);
+	aobj->flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+	ret = amdgpu_bo_pin(aobj, AMDGPU_GEM_DOMAIN_VRAM);
 	amdgpu_bo_unreserve(aobj);
 	if (ret) {
 		DRM_ERROR("Failed to pin new cursor BO (%d)\n", ret);
-		drm_gem_object_put_unlocked(obj);
+		drm_gem_object_put(obj);
 		return ret;
 	}
+	amdgpu_crtc->cursor_addr = amdgpu_bo_gpu_offset(aobj);
 
 	dce_v8_0_lock_cursor(crtc, true);
 
@@ -2377,7 +2355,7 @@ unpin:
 			amdgpu_bo_unpin(aobj);
 			amdgpu_bo_unreserve(aobj);
 		}
-		drm_gem_object_put_unlocked(amdgpu_crtc->cursor_bo);
+		drm_gem_object_put(amdgpu_crtc->cursor_bo);
 	}
 
 	amdgpu_crtc->cursor_bo = obj;
@@ -2421,15 +2399,19 @@ static const struct drm_crtc_funcs dce_v8_0_crtc_funcs = {
 	.cursor_set2 = dce_v8_0_crtc_cursor_set2,
 	.cursor_move = dce_v8_0_crtc_cursor_move,
 	.gamma_set = dce_v8_0_crtc_gamma_set,
-	.set_config = amdgpu_crtc_set_config,
+	.set_config = amdgpu_display_crtc_set_config,
 	.destroy = dce_v8_0_crtc_destroy,
-	.page_flip_target = amdgpu_crtc_page_flip_target,
+	.page_flip_target = amdgpu_display_crtc_page_flip_target,
+	.get_vblank_counter = amdgpu_get_vblank_counter_kms,
+	.enable_vblank = amdgpu_enable_vblank_kms,
+	.disable_vblank = amdgpu_disable_vblank_kms,
+	.get_vblank_timestamp = drm_crtc_vblank_helper_get_vblank_timestamp,
 };
 
 static void dce_v8_0_crtc_dpms(struct drm_crtc *crtc, int mode)
 {
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	unsigned type;
 
@@ -2441,7 +2423,8 @@ static void dce_v8_0_crtc_dpms(struct drm_crtc *crtc, int mode)
 		amdgpu_atombios_crtc_blank(crtc, ATOM_DISABLE);
 		dce_v8_0_vga_enable(crtc, false);
 		/* Make sure VBLANK and PFLIP interrupts are still enabled */
-		type = amdgpu_crtc_idx_to_irq_type(adev, amdgpu_crtc->crtc_id);
+		type = amdgpu_display_crtc_idx_to_irq_type(adev,
+						amdgpu_crtc->crtc_id);
 		amdgpu_irq_update(adev, &adev->crtc_irq, type);
 		amdgpu_irq_update(adev, &adev->pageflip_irq, type);
 		drm_crtc_vblank_on(crtc);
@@ -2461,7 +2444,7 @@ static void dce_v8_0_crtc_dpms(struct drm_crtc *crtc, int mode)
 		break;
 	}
 	/* adjust pm to dpms */
-	amdgpu_pm_compute_clocks(adev);
+	amdgpu_dpm_compute_clocks(adev);
 }
 
 static void dce_v8_0_crtc_prepare(struct drm_crtc *crtc)
@@ -2482,18 +2465,16 @@ static void dce_v8_0_crtc_disable(struct drm_crtc *crtc)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_atom_ss ss;
 	int i;
 
 	dce_v8_0_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
 	if (crtc->primary->fb) {
 		int r;
-		struct amdgpu_framebuffer *amdgpu_fb;
 		struct amdgpu_bo *abo;
 
-		amdgpu_fb = to_amdgpu_framebuffer(crtc->primary->fb);
-		abo = gem_to_amdgpu_bo(amdgpu_fb->obj);
+		abo = gem_to_amdgpu_bo(crtc->primary->fb->obj[0]);
 		r = amdgpu_bo_reserve(abo, true);
 		if (unlikely(r))
 			DRM_ERROR("failed to reserve abo before unpin\n");
@@ -2524,7 +2505,7 @@ static void dce_v8_0_crtc_disable(struct drm_crtc *crtc)
 	case ATOM_PPLL2:
 		/* disable the ppll */
 		amdgpu_atombios_crtc_program_pll(crtc, amdgpu_crtc->crtc_id, amdgpu_crtc->pll_id,
-                                                 0, 0, ATOM_DISABLE, 0, 0, 0, 0, 0, false, &ss);
+						 0, 0, ATOM_DISABLE, 0, 0, 0, 0, 0, false, &ss);
 		break;
 	case ATOM_PPLL0:
 		/* disable the ppll */
@@ -2587,7 +2568,7 @@ static bool dce_v8_0_crtc_mode_fixup(struct drm_crtc *crtc,
 		amdgpu_crtc->connector = NULL;
 		return false;
 	}
-	if (!amdgpu_crtc_scaling_mode_fixup(crtc, mode, adjusted_mode))
+	if (!amdgpu_display_crtc_scaling_mode_fixup(crtc, mode, adjusted_mode))
 		return false;
 	if (amdgpu_atombios_crtc_prepare_pll(crtc, adjusted_mode))
 		return false;
@@ -2611,7 +2592,7 @@ static int dce_v8_0_crtc_set_base_atomic(struct drm_crtc *crtc,
 					 struct drm_framebuffer *fb,
 					 int x, int y, enum mode_set_atomic state)
 {
-       return dce_v8_0_crtc_do_set_base(crtc, fb, x, y, 1);
+	return dce_v8_0_crtc_do_set_base(crtc, fb, x, y, 1);
 }
 
 static const struct drm_crtc_helper_funcs dce_v8_0_crtc_helper_funcs = {
@@ -2623,6 +2604,32 @@ static const struct drm_crtc_helper_funcs dce_v8_0_crtc_helper_funcs = {
 	.prepare = dce_v8_0_crtc_prepare,
 	.commit = dce_v8_0_crtc_commit,
 	.disable = dce_v8_0_crtc_disable,
+	.get_scanout_position = amdgpu_crtc_get_scanout_position,
+};
+
+static void dce_v8_0_panic_flush(struct drm_plane *plane)
+{
+	struct drm_framebuffer *fb;
+	struct amdgpu_crtc *amdgpu_crtc;
+	struct amdgpu_device *adev;
+	uint32_t fb_format;
+
+	if (!plane->fb)
+		return;
+
+	fb = plane->fb;
+	amdgpu_crtc = to_amdgpu_crtc(plane->crtc);
+	adev = drm_to_adev(fb->dev);
+
+	/* Disable DC tiling */
+	fb_format = RREG32(mmGRPH_CONTROL + amdgpu_crtc->crtc_offset);
+	fb_format &= ~GRPH_CONTROL__GRPH_ARRAY_MODE_MASK;
+	WREG32(mmGRPH_CONTROL + amdgpu_crtc->crtc_offset, fb_format);
+}
+
+static const struct drm_plane_helper_funcs dce_v8_0_drm_primary_plane_helper_funcs = {
+	.get_scanout_buffer = amdgpu_display_get_scanout_buffer,
+	.panic_flush = dce_v8_0_panic_flush,
 };
 
 static int dce_v8_0_crtc_init(struct amdgpu_device *adev, int index)
@@ -2634,7 +2641,7 @@ static int dce_v8_0_crtc_init(struct amdgpu_device *adev, int index)
 	if (amdgpu_crtc == NULL)
 		return -ENOMEM;
 
-	drm_crtc_init(adev->ddev, &amdgpu_crtc->base, &dce_v8_0_crtc_funcs);
+	drm_crtc_init(adev_to_drm(adev), &amdgpu_crtc->base, &dce_v8_0_crtc_funcs);
 
 	drm_mode_crtc_set_gamma_size(&amdgpu_crtc->base, 256);
 	amdgpu_crtc->crtc_id = index;
@@ -2642,8 +2649,8 @@ static int dce_v8_0_crtc_init(struct amdgpu_device *adev, int index)
 
 	amdgpu_crtc->max_cursor_width = CIK_CURSOR_WIDTH;
 	amdgpu_crtc->max_cursor_height = CIK_CURSOR_HEIGHT;
-	adev->ddev->mode_config.cursor_width = amdgpu_crtc->max_cursor_width;
-	adev->ddev->mode_config.cursor_height = amdgpu_crtc->max_cursor_height;
+	adev_to_drm(adev)->mode_config.cursor_width = amdgpu_crtc->max_cursor_width;
+	adev_to_drm(adev)->mode_config.cursor_height = amdgpu_crtc->max_cursor_height;
 
 	amdgpu_crtc->crtc_offset = crtc_offsets[amdgpu_crtc->crtc_id];
 
@@ -2652,19 +2659,19 @@ static int dce_v8_0_crtc_init(struct amdgpu_device *adev, int index)
 	amdgpu_crtc->encoder = NULL;
 	amdgpu_crtc->connector = NULL;
 	drm_crtc_helper_add(&amdgpu_crtc->base, &dce_v8_0_crtc_helper_funcs);
+	drm_plane_helper_add(amdgpu_crtc->base.primary, &dce_v8_0_drm_primary_plane_helper_funcs);
 
 	return 0;
 }
 
-static int dce_v8_0_early_init(void *handle)
+static int dce_v8_0_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	adev->audio_endpt_rreg = &dce_v8_0_audio_endpt_rreg;
 	adev->audio_endpt_wreg = &dce_v8_0_audio_endpt_wreg;
 
 	dce_v8_0_set_display_funcs(adev);
-	dce_v8_0_set_irq_funcs(adev);
 
 	adev->mode_info.num_crtc = dce_v8_0_get_num_crtc(adev);
 
@@ -2688,49 +2695,55 @@ static int dce_v8_0_early_init(void *handle)
 		return -EINVAL;
 	}
 
+	dce_v8_0_set_irq_funcs(adev);
+
 	return 0;
 }
 
-static int dce_v8_0_sw_init(void *handle)
+static int dce_v8_0_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r, i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->mode_info.num_crtc; i++) {
-		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i + 1, &adev->crtc_irq);
+		r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, i + 1, &adev->crtc_irq);
 		if (r)
 			return r;
 	}
 
 	for (i = 8; i < 20; i += 2) {
-		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i, &adev->pageflip_irq);
+		r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, i, &adev->pageflip_irq);
 		if (r)
 			return r;
 	}
 
 	/* HPD hotplug */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 42, &adev->hpd_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 42, &adev->hpd_irq);
 	if (r)
 		return r;
 
-	adev->ddev->mode_config.funcs = &amdgpu_mode_funcs;
+	adev_to_drm(adev)->mode_config.funcs = &amdgpu_mode_funcs;
 
-	adev->ddev->mode_config.async_page_flip = true;
+	adev_to_drm(adev)->mode_config.async_page_flip = true;
 
-	adev->ddev->mode_config.max_width = 16384;
-	adev->ddev->mode_config.max_height = 16384;
+	adev_to_drm(adev)->mode_config.max_width = 16384;
+	adev_to_drm(adev)->mode_config.max_height = 16384;
 
-	adev->ddev->mode_config.preferred_depth = 24;
-	adev->ddev->mode_config.prefer_shadow = 1;
+	adev_to_drm(adev)->mode_config.preferred_depth = 24;
+	if (adev->asic_type == CHIP_HAWAII)
+		/* disable prefer shadow for now due to hibernation issues */
+		adev_to_drm(adev)->mode_config.prefer_shadow = 0;
+	else
+		adev_to_drm(adev)->mode_config.prefer_shadow = 1;
 
-	adev->ddev->mode_config.fb_base = adev->mc.aper_base;
+	adev_to_drm(adev)->mode_config.fb_modifiers_not_supported = true;
 
-	r = amdgpu_modeset_create_props(adev);
+	r = amdgpu_display_modeset_create_props(adev);
 	if (r)
 		return r;
 
-	adev->ddev->mode_config.max_width = 16384;
-	adev->ddev->mode_config.max_height = 16384;
+	adev_to_drm(adev)->mode_config.max_width = 16384;
+	adev_to_drm(adev)->mode_config.max_height = 16384;
 
 	/* allocate crtcs */
 	for (i = 0; i < adev->mode_info.num_crtc; i++) {
@@ -2740,7 +2753,7 @@ static int dce_v8_0_sw_init(void *handle)
 	}
 
 	if (amdgpu_atombios_get_connector_info_from_object_table(adev))
-		amdgpu_print_display_setup(adev->ddev);
+		amdgpu_display_print_display_setup(adev_to_drm(adev));
 	else
 		return -EINVAL;
 
@@ -2753,34 +2766,46 @@ static int dce_v8_0_sw_init(void *handle)
 	if (r)
 		return r;
 
-	drm_kms_helper_poll_init(adev->ddev);
+	/* Disable vblank IRQs aggressively for power-saving */
+	/* XXX: can this be enabled for DC? */
+	adev_to_drm(adev)->vblank_disable_immediate = true;
+
+	r = drm_vblank_init(adev_to_drm(adev), adev->mode_info.num_crtc);
+	if (r)
+		return r;
+
+	/* Pre-DCE11 */
+	INIT_DELAYED_WORK(&adev->hotplug_work,
+		  amdgpu_display_hotplug_work_func);
+
+	drm_kms_helper_poll_init(adev_to_drm(adev));
 
 	adev->mode_info.mode_config_initialized = true;
 	return 0;
 }
 
-static int dce_v8_0_sw_fini(void *handle)
+static int dce_v8_0_sw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	kfree(adev->mode_info.bios_hardcoded_edid);
+	drm_edid_free(adev->mode_info.bios_hardcoded_edid);
 
-	drm_kms_helper_poll_fini(adev->ddev);
+	drm_kms_helper_poll_fini(adev_to_drm(adev));
 
 	dce_v8_0_audio_fini(adev);
 
 	dce_v8_0_afmt_fini(adev);
 
-	drm_mode_config_cleanup(adev->ddev);
+	drm_mode_config_cleanup(adev_to_drm(adev));
 	adev->mode_info.mode_config_initialized = false;
 
 	return 0;
 }
 
-static int dce_v8_0_hw_init(void *handle)
+static int dce_v8_0_hw_init(struct amdgpu_ip_block *ip_block)
 {
 	int i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	/* disable vga render */
 	dce_v8_0_set_vga_render_state(adev, false);
@@ -2800,10 +2825,10 @@ static int dce_v8_0_hw_init(void *handle)
 	return 0;
 }
 
-static int dce_v8_0_hw_fini(void *handle)
+static int dce_v8_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
 	int i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	dce_v8_0_hpd_fini(adev);
 
@@ -2813,20 +2838,35 @@ static int dce_v8_0_hw_fini(void *handle)
 
 	dce_v8_0_pageflip_interrupt_fini(adev);
 
+	flush_delayed_work(&adev->hotplug_work);
+
 	return 0;
 }
 
-static int dce_v8_0_suspend(void *handle)
+static int dce_v8_0_suspend(struct amdgpu_ip_block *ip_block)
 {
-	return dce_v8_0_hw_fini(handle);
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = amdgpu_display_suspend_helper(adev);
+	if (r)
+		return r;
+
+	adev->mode_info.bl_level =
+		amdgpu_atombios_encoder_get_backlight_level_from_reg(adev);
+
+	return dce_v8_0_hw_fini(ip_block);
 }
 
-static int dce_v8_0_resume(void *handle)
+static int dce_v8_0_resume(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int ret;
 
-	ret = dce_v8_0_hw_init(handle);
+	amdgpu_atombios_encoder_set_backlight_level_to_reg(adev,
+							   adev->mode_info.bl_level);
+
+	ret = dce_v8_0_hw_init(ip_block);
 
 	/* turn on the BL */
 	if (adev->mode_info.bl_encoder) {
@@ -2835,24 +2875,21 @@ static int dce_v8_0_resume(void *handle)
 		amdgpu_display_backlight_set_level(adev, adev->mode_info.bl_encoder,
 						    bl_level);
 	}
+	if (ret)
+		return ret;
 
-	return ret;
+	return amdgpu_display_resume_helper(adev);
 }
 
-static bool dce_v8_0_is_idle(void *handle)
+static bool dce_v8_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	return true;
 }
 
-static int dce_v8_0_wait_for_idle(void *handle)
-{
-	return 0;
-}
-
-static int dce_v8_0_soft_reset(void *handle)
+static int dce_v8_0_soft_reset(struct amdgpu_ip_block *ip_block)
 {
 	u32 srbm_soft_reset = 0, tmp;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (dce_v8_0_is_display_hung(adev))
 		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_DC_MASK;
@@ -2978,7 +3015,7 @@ static void dce_v8_0_set_crtc_vline_interrupt_state(struct amdgpu_device *adev,
 	}
 }
 
-static int dce_v8_0_set_hpd_interrupt_state(struct amdgpu_device *adev,
+static int dce_v8_0_set_hpd_irq_state(struct amdgpu_device *adev,
 					    struct amdgpu_irq_src *src,
 					    unsigned type,
 					    enum amdgpu_interrupt_state state)
@@ -2986,7 +3023,7 @@ static int dce_v8_0_set_hpd_interrupt_state(struct amdgpu_device *adev,
 	u32 dc_hpd_int_cntl;
 
 	if (type >= adev->mode_info.num_hpd) {
-		DRM_DEBUG("invalid hdp %d\n", type);
+		DRM_DEBUG("invalid hpd %d\n", type);
 		return 0;
 	}
 
@@ -3008,7 +3045,7 @@ static int dce_v8_0_set_hpd_interrupt_state(struct amdgpu_device *adev,
 	return 0;
 }
 
-static int dce_v8_0_set_crtc_interrupt_state(struct amdgpu_device *adev,
+static int dce_v8_0_set_crtc_irq_state(struct amdgpu_device *adev,
 					     struct amdgpu_irq_src *src,
 					     unsigned type,
 					     enum amdgpu_interrupt_state state)
@@ -3062,7 +3099,8 @@ static int dce_v8_0_crtc_irq(struct amdgpu_device *adev,
 {
 	unsigned crtc = entry->src_id - 1;
 	uint32_t disp_int = RREG32(interrupt_status_offsets[crtc].reg);
-	unsigned irq_type = amdgpu_crtc_idx_to_irq_type(adev, crtc);
+	unsigned int irq_type = amdgpu_display_crtc_idx_to_irq_type(adev,
+								    crtc);
 
 	switch (entry->src_data[0]) {
 	case 0: /* vblank */
@@ -3072,7 +3110,7 @@ static int dce_v8_0_crtc_irq(struct amdgpu_device *adev,
 			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
 
 		if (amdgpu_irq_enabled(adev, source, irq_type)) {
-			drm_handle_vblank(adev->ddev, crtc);
+			drm_handle_vblank(adev_to_drm(adev), crtc);
 		}
 		DRM_DEBUG("IH: D%d vblank\n", crtc + 1);
 		break;
@@ -3092,7 +3130,7 @@ static int dce_v8_0_crtc_irq(struct amdgpu_device *adev,
 	return 0;
 }
 
-static int dce_v8_0_set_pageflip_interrupt_state(struct amdgpu_device *adev,
+static int dce_v8_0_set_pageflip_irq_state(struct amdgpu_device *adev,
 						 struct amdgpu_irq_src *src,
 						 unsigned type,
 						 enum amdgpu_interrupt_state state)
@@ -3141,14 +3179,14 @@ static int dce_v8_0_pageflip_irq(struct amdgpu_device *adev,
 	if (amdgpu_crtc == NULL)
 		return 0;
 
-	spin_lock_irqsave(&adev->ddev->event_lock, flags);
+	spin_lock_irqsave(&adev_to_drm(adev)->event_lock, flags);
 	works = amdgpu_crtc->pflip_works;
-	if (amdgpu_crtc->pflip_status != AMDGPU_FLIP_SUBMITTED){
+	if (amdgpu_crtc->pflip_status != AMDGPU_FLIP_SUBMITTED) {
 		DRM_DEBUG_DRIVER("amdgpu_crtc->pflip_status = %d != "
 						"AMDGPU_FLIP_SUBMITTED(%d)\n",
 						amdgpu_crtc->pflip_status,
 						AMDGPU_FLIP_SUBMITTED);
-		spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
+		spin_unlock_irqrestore(&adev_to_drm(adev)->event_lock, flags);
 		return 0;
 	}
 
@@ -3160,7 +3198,7 @@ static int dce_v8_0_pageflip_irq(struct amdgpu_device *adev,
 	if (works->event)
 		drm_crtc_send_vblank_event(&amdgpu_crtc->base, works->event);
 
-	spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
+	spin_unlock_irqrestore(&adev_to_drm(adev)->event_lock, flags);
 
 	drm_crtc_vblank_put(&amdgpu_crtc->base);
 	schedule_work(&works->unpin_work);
@@ -3172,7 +3210,7 @@ static int dce_v8_0_hpd_irq(struct amdgpu_device *adev,
 			    struct amdgpu_irq_src *source,
 			    struct amdgpu_iv_entry *entry)
 {
-	uint32_t disp_int, mask, tmp;
+	uint32_t disp_int, mask;
 	unsigned hpd;
 
 	if (entry->src_data[0] >= adev->mode_info.num_hpd) {
@@ -3185,10 +3223,8 @@ static int dce_v8_0_hpd_irq(struct amdgpu_device *adev,
 	mask = interrupt_status_offsets[hpd].hpd;
 
 	if (disp_int & mask) {
-		tmp = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd]);
-		tmp |= DC_HPD1_INT_CONTROL__DC_HPD1_INT_ACK_MASK;
-		WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd], tmp);
-		schedule_work(&adev->hotplug_work);
+		dce_v8_0_hpd_int_ack(adev, hpd);
+		schedule_delayed_work(&adev->hotplug_work, 0);
 		DRM_DEBUG("IH: HPD%d\n", hpd + 1);
 	}
 
@@ -3196,13 +3232,13 @@ static int dce_v8_0_hpd_irq(struct amdgpu_device *adev,
 
 }
 
-static int dce_v8_0_set_clockgating_state(void *handle,
+static int dce_v8_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
 	return 0;
 }
 
-static int dce_v8_0_set_powergating_state(void *handle,
+static int dce_v8_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	return 0;
@@ -3211,7 +3247,6 @@ static int dce_v8_0_set_powergating_state(void *handle,
 static const struct amd_ip_funcs dce_v8_0_ip_funcs = {
 	.name = "dce_v8_0",
 	.early_init = dce_v8_0_early_init,
-	.late_init = NULL,
 	.sw_init = dce_v8_0_sw_init,
 	.sw_fini = dce_v8_0_sw_fini,
 	.hw_init = dce_v8_0_hw_init,
@@ -3219,7 +3254,6 @@ static const struct amd_ip_funcs dce_v8_0_ip_funcs = {
 	.suspend = dce_v8_0_suspend,
 	.resume = dce_v8_0_resume,
 	.is_idle = dce_v8_0_is_idle,
-	.wait_for_idle = dce_v8_0_wait_for_idle,
 	.soft_reset = dce_v8_0_soft_reset,
 	.set_clockgating_state = dce_v8_0_set_clockgating_state,
 	.set_powergating_state = dce_v8_0_set_powergating_state,
@@ -3248,7 +3282,7 @@ dce_v8_0_encoder_mode_set(struct drm_encoder *encoder,
 
 static void dce_v8_0_encoder_prepare(struct drm_encoder *encoder)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(encoder->dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
 
@@ -3288,7 +3322,7 @@ static void dce_v8_0_encoder_prepare(struct drm_encoder *encoder)
 static void dce_v8_0_encoder_commit(struct drm_encoder *encoder)
 {
 	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_device *adev = drm_to_adev(dev);
 
 	/* need to call this here as we need the crtc set up */
 	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_ON);
@@ -3388,7 +3422,7 @@ static void dce_v8_0_encoder_add(struct amdgpu_device *adev,
 				 uint32_t supported_device,
 				 u16 caps)
 {
-	struct drm_device *dev = adev->ddev;
+	struct drm_device *dev = adev_to_drm(adev);
 	struct drm_encoder *encoder;
 	struct amdgpu_encoder *amdgpu_encoder;
 
@@ -3490,7 +3524,6 @@ static void dce_v8_0_encoder_add(struct amdgpu_device *adev,
 static const struct amdgpu_display_funcs dce_v8_0_display_funcs = {
 	.bandwidth_update = &dce_v8_0_bandwidth_update,
 	.vblank_get_counter = &dce_v8_0_vblank_get_counter,
-	.vblank_wait = &dce_v8_0_vblank_wait,
 	.backlight_set_level = &amdgpu_atombios_encoder_set_backlight_level,
 	.backlight_get_level = &amdgpu_atombios_encoder_get_backlight_level,
 	.hpd_sense = &dce_v8_0_hpd_sense,
@@ -3504,39 +3537,40 @@ static const struct amdgpu_display_funcs dce_v8_0_display_funcs = {
 
 static void dce_v8_0_set_display_funcs(struct amdgpu_device *adev)
 {
-	if (adev->mode_info.funcs == NULL)
-		adev->mode_info.funcs = &dce_v8_0_display_funcs;
+	adev->mode_info.funcs = &dce_v8_0_display_funcs;
 }
 
 static const struct amdgpu_irq_src_funcs dce_v8_0_crtc_irq_funcs = {
-	.set = dce_v8_0_set_crtc_interrupt_state,
+	.set = dce_v8_0_set_crtc_irq_state,
 	.process = dce_v8_0_crtc_irq,
 };
 
 static const struct amdgpu_irq_src_funcs dce_v8_0_pageflip_irq_funcs = {
-	.set = dce_v8_0_set_pageflip_interrupt_state,
+	.set = dce_v8_0_set_pageflip_irq_state,
 	.process = dce_v8_0_pageflip_irq,
 };
 
 static const struct amdgpu_irq_src_funcs dce_v8_0_hpd_irq_funcs = {
-	.set = dce_v8_0_set_hpd_interrupt_state,
+	.set = dce_v8_0_set_hpd_irq_state,
 	.process = dce_v8_0_hpd_irq,
 };
 
 static void dce_v8_0_set_irq_funcs(struct amdgpu_device *adev)
 {
-	adev->crtc_irq.num_types = AMDGPU_CRTC_IRQ_LAST;
+	if (adev->mode_info.num_crtc > 0)
+		adev->crtc_irq.num_types = AMDGPU_CRTC_IRQ_VLINE1 + adev->mode_info.num_crtc;
+	else
+		adev->crtc_irq.num_types = 0;
 	adev->crtc_irq.funcs = &dce_v8_0_crtc_irq_funcs;
 
-	adev->pageflip_irq.num_types = AMDGPU_PAGEFLIP_IRQ_LAST;
+	adev->pageflip_irq.num_types = adev->mode_info.num_crtc;
 	adev->pageflip_irq.funcs = &dce_v8_0_pageflip_irq_funcs;
 
-	adev->hpd_irq.num_types = AMDGPU_HPD_LAST;
+	adev->hpd_irq.num_types = adev->mode_info.num_hpd;
 	adev->hpd_irq.funcs = &dce_v8_0_hpd_irq_funcs;
 }
 
-const struct amdgpu_ip_block_version dce_v8_0_ip_block =
-{
+const struct amdgpu_ip_block_version dce_v8_0_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_DCE,
 	.major = 8,
 	.minor = 0,
@@ -3544,8 +3578,7 @@ const struct amdgpu_ip_block_version dce_v8_0_ip_block =
 	.funcs = &dce_v8_0_ip_funcs,
 };
 
-const struct amdgpu_ip_block_version dce_v8_1_ip_block =
-{
+const struct amdgpu_ip_block_version dce_v8_1_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_DCE,
 	.major = 8,
 	.minor = 1,
@@ -3553,8 +3586,7 @@ const struct amdgpu_ip_block_version dce_v8_1_ip_block =
 	.funcs = &dce_v8_0_ip_funcs,
 };
 
-const struct amdgpu_ip_block_version dce_v8_2_ip_block =
-{
+const struct amdgpu_ip_block_version dce_v8_2_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_DCE,
 	.major = 8,
 	.minor = 2,
@@ -3562,8 +3594,7 @@ const struct amdgpu_ip_block_version dce_v8_2_ip_block =
 	.funcs = &dce_v8_0_ip_funcs,
 };
 
-const struct amdgpu_ip_block_version dce_v8_3_ip_block =
-{
+const struct amdgpu_ip_block_version dce_v8_3_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_DCE,
 	.major = 8,
 	.minor = 3,
@@ -3571,8 +3602,7 @@ const struct amdgpu_ip_block_version dce_v8_3_ip_block =
 	.funcs = &dce_v8_0_ip_funcs,
 };
 
-const struct amdgpu_ip_block_version dce_v8_5_ip_block =
-{
+const struct amdgpu_ip_block_version dce_v8_5_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_DCE,
 	.major = 8,
 	.minor = 5,
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_virtual.c b/drivers/gpu/drm/amd/amdgpu/dce_virtual.c
deleted file mode 100644
index b9ee9073cb0d..000000000000
--- a/drivers/gpu/drm/amd/amdgpu/dce_virtual.c
+++ /dev/null
@@ -1,782 +0,0 @@
-/*
- * Copyright 2014 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-#include <drm/drmP.h>
-#include "amdgpu.h"
-#include "amdgpu_pm.h"
-#include "amdgpu_i2c.h"
-#include "atom.h"
-#include "amdgpu_pll.h"
-#include "amdgpu_connectors.h"
-#ifdef CONFIG_DRM_AMDGPU_SI
-#include "dce_v6_0.h"
-#endif
-#ifdef CONFIG_DRM_AMDGPU_CIK
-#include "dce_v8_0.h"
-#endif
-#include "dce_v10_0.h"
-#include "dce_v11_0.h"
-#include "dce_virtual.h"
-
-#define DCE_VIRTUAL_VBLANK_PERIOD 16666666
-
-
-static void dce_virtual_set_display_funcs(struct amdgpu_device *adev);
-static void dce_virtual_set_irq_funcs(struct amdgpu_device *adev);
-static int dce_virtual_connector_encoder_init(struct amdgpu_device *adev,
-					      int index);
-
-/**
- * dce_virtual_vblank_wait - vblank wait asic callback.
- *
- * @adev: amdgpu_device pointer
- * @crtc: crtc to wait for vblank on
- *
- * Wait for vblank on the requested crtc (evergreen+).
- */
-static void dce_virtual_vblank_wait(struct amdgpu_device *adev, int crtc)
-{
-	return;
-}
-
-static u32 dce_virtual_vblank_get_counter(struct amdgpu_device *adev, int crtc)
-{
-	return 0;
-}
-
-static void dce_virtual_page_flip(struct amdgpu_device *adev,
-			      int crtc_id, u64 crtc_base, bool async)
-{
-	return;
-}
-
-static int dce_virtual_crtc_get_scanoutpos(struct amdgpu_device *adev, int crtc,
-					u32 *vbl, u32 *position)
-{
-	*vbl = 0;
-	*position = 0;
-
-	return -EINVAL;
-}
-
-static bool dce_virtual_hpd_sense(struct amdgpu_device *adev,
-			       enum amdgpu_hpd_id hpd)
-{
-	return true;
-}
-
-static void dce_virtual_hpd_set_polarity(struct amdgpu_device *adev,
-				      enum amdgpu_hpd_id hpd)
-{
-	return;
-}
-
-static u32 dce_virtual_hpd_get_gpio_reg(struct amdgpu_device *adev)
-{
-	return 0;
-}
-
-/**
- * dce_virtual_bandwidth_update - program display watermarks
- *
- * @adev: amdgpu_device pointer
- *
- * Calculate and program the display watermarks and line
- * buffer allocation (CIK).
- */
-static void dce_virtual_bandwidth_update(struct amdgpu_device *adev)
-{
-	return;
-}
-
-static int dce_virtual_crtc_gamma_set(struct drm_crtc *crtc, u16 *red,
-				      u16 *green, u16 *blue, uint32_t size,
-				      struct drm_modeset_acquire_ctx *ctx)
-{
-	return 0;
-}
-
-static void dce_virtual_crtc_destroy(struct drm_crtc *crtc)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-
-	drm_crtc_cleanup(crtc);
-	kfree(amdgpu_crtc);
-}
-
-static const struct drm_crtc_funcs dce_virtual_crtc_funcs = {
-	.cursor_set2 = NULL,
-	.cursor_move = NULL,
-	.gamma_set = dce_virtual_crtc_gamma_set,
-	.set_config = amdgpu_crtc_set_config,
-	.destroy = dce_virtual_crtc_destroy,
-	.page_flip_target = amdgpu_crtc_page_flip_target,
-};
-
-static void dce_virtual_crtc_dpms(struct drm_crtc *crtc, int mode)
-{
-	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	unsigned type;
-
-	if (amdgpu_sriov_vf(adev))
-		return;
-
-	switch (mode) {
-	case DRM_MODE_DPMS_ON:
-		amdgpu_crtc->enabled = true;
-		/* Make sure VBLANK interrupts are still enabled */
-		type = amdgpu_crtc_idx_to_irq_type(adev, amdgpu_crtc->crtc_id);
-		amdgpu_irq_update(adev, &adev->crtc_irq, type);
-		drm_crtc_vblank_on(crtc);
-		break;
-	case DRM_MODE_DPMS_STANDBY:
-	case DRM_MODE_DPMS_SUSPEND:
-	case DRM_MODE_DPMS_OFF:
-		drm_crtc_vblank_off(crtc);
-		amdgpu_crtc->enabled = false;
-		break;
-	}
-}
-
-
-static void dce_virtual_crtc_prepare(struct drm_crtc *crtc)
-{
-	dce_virtual_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
-}
-
-static void dce_virtual_crtc_commit(struct drm_crtc *crtc)
-{
-	dce_virtual_crtc_dpms(crtc, DRM_MODE_DPMS_ON);
-}
-
-static void dce_virtual_crtc_disable(struct drm_crtc *crtc)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-
-	dce_virtual_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
-	if (crtc->primary->fb) {
-		int r;
-		struct amdgpu_framebuffer *amdgpu_fb;
-		struct amdgpu_bo *abo;
-
-		amdgpu_fb = to_amdgpu_framebuffer(crtc->primary->fb);
-		abo = gem_to_amdgpu_bo(amdgpu_fb->obj);
-		r = amdgpu_bo_reserve(abo, true);
-		if (unlikely(r))
-			DRM_ERROR("failed to reserve abo before unpin\n");
-		else {
-			amdgpu_bo_unpin(abo);
-			amdgpu_bo_unreserve(abo);
-		}
-	}
-
-	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
-	amdgpu_crtc->encoder = NULL;
-	amdgpu_crtc->connector = NULL;
-}
-
-static int dce_virtual_crtc_mode_set(struct drm_crtc *crtc,
-				  struct drm_display_mode *mode,
-				  struct drm_display_mode *adjusted_mode,
-				  int x, int y, struct drm_framebuffer *old_fb)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-
-	/* update the hw version fpr dpm */
-	amdgpu_crtc->hw_mode = *adjusted_mode;
-
-	return 0;
-}
-
-static bool dce_virtual_crtc_mode_fixup(struct drm_crtc *crtc,
-				     const struct drm_display_mode *mode,
-				     struct drm_display_mode *adjusted_mode)
-{
-	return true;
-}
-
-
-static int dce_virtual_crtc_set_base(struct drm_crtc *crtc, int x, int y,
-				  struct drm_framebuffer *old_fb)
-{
-	return 0;
-}
-
-static int dce_virtual_crtc_set_base_atomic(struct drm_crtc *crtc,
-					 struct drm_framebuffer *fb,
-					 int x, int y, enum mode_set_atomic state)
-{
-	return 0;
-}
-
-static const struct drm_crtc_helper_funcs dce_virtual_crtc_helper_funcs = {
-	.dpms = dce_virtual_crtc_dpms,
-	.mode_fixup = dce_virtual_crtc_mode_fixup,
-	.mode_set = dce_virtual_crtc_mode_set,
-	.mode_set_base = dce_virtual_crtc_set_base,
-	.mode_set_base_atomic = dce_virtual_crtc_set_base_atomic,
-	.prepare = dce_virtual_crtc_prepare,
-	.commit = dce_virtual_crtc_commit,
-	.disable = dce_virtual_crtc_disable,
-};
-
-static int dce_virtual_crtc_init(struct amdgpu_device *adev, int index)
-{
-	struct amdgpu_crtc *amdgpu_crtc;
-
-	amdgpu_crtc = kzalloc(sizeof(struct amdgpu_crtc) +
-			      (AMDGPUFB_CONN_LIMIT * sizeof(struct drm_connector *)), GFP_KERNEL);
-	if (amdgpu_crtc == NULL)
-		return -ENOMEM;
-
-	drm_crtc_init(adev->ddev, &amdgpu_crtc->base, &dce_virtual_crtc_funcs);
-
-	drm_mode_crtc_set_gamma_size(&amdgpu_crtc->base, 256);
-	amdgpu_crtc->crtc_id = index;
-	adev->mode_info.crtcs[index] = amdgpu_crtc;
-
-	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
-	amdgpu_crtc->encoder = NULL;
-	amdgpu_crtc->connector = NULL;
-	amdgpu_crtc->vsync_timer_enabled = AMDGPU_IRQ_STATE_DISABLE;
-	drm_crtc_helper_add(&amdgpu_crtc->base, &dce_virtual_crtc_helper_funcs);
-
-	return 0;
-}
-
-static int dce_virtual_early_init(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	dce_virtual_set_display_funcs(adev);
-	dce_virtual_set_irq_funcs(adev);
-
-	adev->mode_info.num_hpd = 1;
-	adev->mode_info.num_dig = 1;
-	return 0;
-}
-
-static struct drm_encoder *
-dce_virtual_encoder(struct drm_connector *connector)
-{
-	int enc_id = connector->encoder_ids[0];
-	struct drm_encoder *encoder;
-	int i;
-
-	for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) {
-		if (connector->encoder_ids[i] == 0)
-			break;
-
-		encoder = drm_encoder_find(connector->dev, connector->encoder_ids[i]);
-		if (!encoder)
-			continue;
-
-		if (encoder->encoder_type == DRM_MODE_ENCODER_VIRTUAL)
-			return encoder;
-	}
-
-	/* pick the first one */
-	if (enc_id)
-		return drm_encoder_find(connector->dev, enc_id);
-	return NULL;
-}
-
-static int dce_virtual_get_modes(struct drm_connector *connector)
-{
-	struct drm_device *dev = connector->dev;
-	struct drm_display_mode *mode = NULL;
-	unsigned i;
-	static const struct mode_size {
-		int w;
-		int h;
-	} common_modes[17] = {
-		{ 640,  480},
-		{ 720,  480},
-		{ 800,  600},
-		{ 848,  480},
-		{1024,  768},
-		{1152,  768},
-		{1280,  720},
-		{1280,  800},
-		{1280,  854},
-		{1280,  960},
-		{1280, 1024},
-		{1440,  900},
-		{1400, 1050},
-		{1680, 1050},
-		{1600, 1200},
-		{1920, 1080},
-		{1920, 1200}
-	};
-
-	for (i = 0; i < 17; i++) {
-		mode = drm_cvt_mode(dev, common_modes[i].w, common_modes[i].h, 60, false, false, false);
-		drm_mode_probed_add(connector, mode);
-	}
-
-	return 0;
-}
-
-static int dce_virtual_mode_valid(struct drm_connector *connector,
-				  struct drm_display_mode *mode)
-{
-	return MODE_OK;
-}
-
-static int
-dce_virtual_dpms(struct drm_connector *connector, int mode)
-{
-	return 0;
-}
-
-static int
-dce_virtual_set_property(struct drm_connector *connector,
-			 struct drm_property *property,
-			 uint64_t val)
-{
-	return 0;
-}
-
-static void dce_virtual_destroy(struct drm_connector *connector)
-{
-	drm_connector_unregister(connector);
-	drm_connector_cleanup(connector);
-	kfree(connector);
-}
-
-static void dce_virtual_force(struct drm_connector *connector)
-{
-	return;
-}
-
-static const struct drm_connector_helper_funcs dce_virtual_connector_helper_funcs = {
-	.get_modes = dce_virtual_get_modes,
-	.mode_valid = dce_virtual_mode_valid,
-	.best_encoder = dce_virtual_encoder,
-};
-
-static const struct drm_connector_funcs dce_virtual_connector_funcs = {
-	.dpms = dce_virtual_dpms,
-	.fill_modes = drm_helper_probe_single_connector_modes,
-	.set_property = dce_virtual_set_property,
-	.destroy = dce_virtual_destroy,
-	.force = dce_virtual_force,
-};
-
-static int dce_virtual_sw_init(void *handle)
-{
-	int r, i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 229, &adev->crtc_irq);
-	if (r)
-		return r;
-
-	adev->ddev->max_vblank_count = 0;
-
-	adev->ddev->mode_config.funcs = &amdgpu_mode_funcs;
-
-	adev->ddev->mode_config.max_width = 16384;
-	adev->ddev->mode_config.max_height = 16384;
-
-	adev->ddev->mode_config.preferred_depth = 24;
-	adev->ddev->mode_config.prefer_shadow = 1;
-
-	adev->ddev->mode_config.fb_base = adev->mc.aper_base;
-
-	r = amdgpu_modeset_create_props(adev);
-	if (r)
-		return r;
-
-	adev->ddev->mode_config.max_width = 16384;
-	adev->ddev->mode_config.max_height = 16384;
-
-	/* allocate crtcs, encoders, connectors */
-	for (i = 0; i < adev->mode_info.num_crtc; i++) {
-		r = dce_virtual_crtc_init(adev, i);
-		if (r)
-			return r;
-		r = dce_virtual_connector_encoder_init(adev, i);
-		if (r)
-			return r;
-	}
-
-	drm_kms_helper_poll_init(adev->ddev);
-
-	adev->mode_info.mode_config_initialized = true;
-	return 0;
-}
-
-static int dce_virtual_sw_fini(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	kfree(adev->mode_info.bios_hardcoded_edid);
-
-	drm_kms_helper_poll_fini(adev->ddev);
-
-	drm_mode_config_cleanup(adev->ddev);
-	adev->mode_info.mode_config_initialized = false;
-	return 0;
-}
-
-static int dce_virtual_hw_init(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	switch (adev->asic_type) {
-#ifdef CONFIG_DRM_AMDGPU_SI
-	case CHIP_TAHITI:
-	case CHIP_PITCAIRN:
-	case CHIP_VERDE:
-	case CHIP_OLAND:
-		dce_v6_0_disable_dce(adev);
-		break;
-#endif
-#ifdef CONFIG_DRM_AMDGPU_CIK
-	case CHIP_BONAIRE:
-	case CHIP_HAWAII:
-	case CHIP_KAVERI:
-	case CHIP_KABINI:
-	case CHIP_MULLINS:
-		dce_v8_0_disable_dce(adev);
-		break;
-#endif
-	case CHIP_FIJI:
-	case CHIP_TONGA:
-		dce_v10_0_disable_dce(adev);
-		break;
-	case CHIP_CARRIZO:
-	case CHIP_STONEY:
-	case CHIP_POLARIS11:
-	case CHIP_POLARIS10:
-		dce_v11_0_disable_dce(adev);
-		break;
-	case CHIP_TOPAZ:
-#ifdef CONFIG_DRM_AMDGPU_SI
-	case CHIP_HAINAN:
-#endif
-		/* no DCE */
-		break;
-	case CHIP_VEGA10:
-		break;
-	default:
-		DRM_ERROR("Virtual display unsupported ASIC type: 0x%X\n", adev->asic_type);
-	}
-	return 0;
-}
-
-static int dce_virtual_hw_fini(void *handle)
-{
-	return 0;
-}
-
-static int dce_virtual_suspend(void *handle)
-{
-	return dce_virtual_hw_fini(handle);
-}
-
-static int dce_virtual_resume(void *handle)
-{
-	return dce_virtual_hw_init(handle);
-}
-
-static bool dce_virtual_is_idle(void *handle)
-{
-	return true;
-}
-
-static int dce_virtual_wait_for_idle(void *handle)
-{
-	return 0;
-}
-
-static int dce_virtual_soft_reset(void *handle)
-{
-	return 0;
-}
-
-static int dce_virtual_set_clockgating_state(void *handle,
-					  enum amd_clockgating_state state)
-{
-	return 0;
-}
-
-static int dce_virtual_set_powergating_state(void *handle,
-					  enum amd_powergating_state state)
-{
-	return 0;
-}
-
-static const struct amd_ip_funcs dce_virtual_ip_funcs = {
-	.name = "dce_virtual",
-	.early_init = dce_virtual_early_init,
-	.late_init = NULL,
-	.sw_init = dce_virtual_sw_init,
-	.sw_fini = dce_virtual_sw_fini,
-	.hw_init = dce_virtual_hw_init,
-	.hw_fini = dce_virtual_hw_fini,
-	.suspend = dce_virtual_suspend,
-	.resume = dce_virtual_resume,
-	.is_idle = dce_virtual_is_idle,
-	.wait_for_idle = dce_virtual_wait_for_idle,
-	.soft_reset = dce_virtual_soft_reset,
-	.set_clockgating_state = dce_virtual_set_clockgating_state,
-	.set_powergating_state = dce_virtual_set_powergating_state,
-};
-
-/* these are handled by the primary encoders */
-static void dce_virtual_encoder_prepare(struct drm_encoder *encoder)
-{
-	return;
-}
-
-static void dce_virtual_encoder_commit(struct drm_encoder *encoder)
-{
-	return;
-}
-
-static void
-dce_virtual_encoder_mode_set(struct drm_encoder *encoder,
-			     struct drm_display_mode *mode,
-			     struct drm_display_mode *adjusted_mode)
-{
-	return;
-}
-
-static void dce_virtual_encoder_disable(struct drm_encoder *encoder)
-{
-	return;
-}
-
-static void
-dce_virtual_encoder_dpms(struct drm_encoder *encoder, int mode)
-{
-	return;
-}
-
-static bool dce_virtual_encoder_mode_fixup(struct drm_encoder *encoder,
-				    const struct drm_display_mode *mode,
-				    struct drm_display_mode *adjusted_mode)
-{
-	return true;
-}
-
-static const struct drm_encoder_helper_funcs dce_virtual_encoder_helper_funcs = {
-	.dpms = dce_virtual_encoder_dpms,
-	.mode_fixup = dce_virtual_encoder_mode_fixup,
-	.prepare = dce_virtual_encoder_prepare,
-	.mode_set = dce_virtual_encoder_mode_set,
-	.commit = dce_virtual_encoder_commit,
-	.disable = dce_virtual_encoder_disable,
-};
-
-static void dce_virtual_encoder_destroy(struct drm_encoder *encoder)
-{
-	drm_encoder_cleanup(encoder);
-	kfree(encoder);
-}
-
-static const struct drm_encoder_funcs dce_virtual_encoder_funcs = {
-	.destroy = dce_virtual_encoder_destroy,
-};
-
-static int dce_virtual_connector_encoder_init(struct amdgpu_device *adev,
-					      int index)
-{
-	struct drm_encoder *encoder;
-	struct drm_connector *connector;
-
-	/* add a new encoder */
-	encoder = kzalloc(sizeof(struct drm_encoder), GFP_KERNEL);
-	if (!encoder)
-		return -ENOMEM;
-	encoder->possible_crtcs = 1 << index;
-	drm_encoder_init(adev->ddev, encoder, &dce_virtual_encoder_funcs,
-			 DRM_MODE_ENCODER_VIRTUAL, NULL);
-	drm_encoder_helper_add(encoder, &dce_virtual_encoder_helper_funcs);
-
-	connector = kzalloc(sizeof(struct drm_connector), GFP_KERNEL);
-	if (!connector) {
-		kfree(encoder);
-		return -ENOMEM;
-	}
-
-	/* add a new connector */
-	drm_connector_init(adev->ddev, connector, &dce_virtual_connector_funcs,
-			   DRM_MODE_CONNECTOR_VIRTUAL);
-	drm_connector_helper_add(connector, &dce_virtual_connector_helper_funcs);
-	connector->display_info.subpixel_order = SubPixelHorizontalRGB;
-	connector->interlace_allowed = false;
-	connector->doublescan_allowed = false;
-	drm_connector_register(connector);
-
-	/* link them */
-	drm_mode_connector_attach_encoder(connector, encoder);
-
-	return 0;
-}
-
-static const struct amdgpu_display_funcs dce_virtual_display_funcs = {
-	.bandwidth_update = &dce_virtual_bandwidth_update,
-	.vblank_get_counter = &dce_virtual_vblank_get_counter,
-	.vblank_wait = &dce_virtual_vblank_wait,
-	.backlight_set_level = NULL,
-	.backlight_get_level = NULL,
-	.hpd_sense = &dce_virtual_hpd_sense,
-	.hpd_set_polarity = &dce_virtual_hpd_set_polarity,
-	.hpd_get_gpio_reg = &dce_virtual_hpd_get_gpio_reg,
-	.page_flip = &dce_virtual_page_flip,
-	.page_flip_get_scanoutpos = &dce_virtual_crtc_get_scanoutpos,
-	.add_encoder = NULL,
-	.add_connector = NULL,
-};
-
-static void dce_virtual_set_display_funcs(struct amdgpu_device *adev)
-{
-	if (adev->mode_info.funcs == NULL)
-		adev->mode_info.funcs = &dce_virtual_display_funcs;
-}
-
-static int dce_virtual_pageflip(struct amdgpu_device *adev,
-				unsigned crtc_id)
-{
-	unsigned long flags;
-	struct amdgpu_crtc *amdgpu_crtc;
-	struct amdgpu_flip_work *works;
-
-	amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
-
-	if (crtc_id >= adev->mode_info.num_crtc) {
-		DRM_ERROR("invalid pageflip crtc %d\n", crtc_id);
-		return -EINVAL;
-	}
-
-	/* IRQ could occur when in initial stage */
-	if (amdgpu_crtc == NULL)
-		return 0;
-
-	spin_lock_irqsave(&adev->ddev->event_lock, flags);
-	works = amdgpu_crtc->pflip_works;
-	if (amdgpu_crtc->pflip_status != AMDGPU_FLIP_SUBMITTED) {
-		DRM_DEBUG_DRIVER("amdgpu_crtc->pflip_status = %d != "
-			"AMDGPU_FLIP_SUBMITTED(%d)\n",
-			amdgpu_crtc->pflip_status,
-			AMDGPU_FLIP_SUBMITTED);
-		spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
-		return 0;
-	}
-
-	/* page flip completed. clean up */
-	amdgpu_crtc->pflip_status = AMDGPU_FLIP_NONE;
-	amdgpu_crtc->pflip_works = NULL;
-
-	/* wakeup usersapce */
-	if (works->event)
-		drm_crtc_send_vblank_event(&amdgpu_crtc->base, works->event);
-
-	spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
-
-	drm_crtc_vblank_put(&amdgpu_crtc->base);
-	schedule_work(&works->unpin_work);
-
-	return 0;
-}
-
-static enum hrtimer_restart dce_virtual_vblank_timer_handle(struct hrtimer *vblank_timer)
-{
-	struct amdgpu_crtc *amdgpu_crtc = container_of(vblank_timer,
-						       struct amdgpu_crtc, vblank_timer);
-	struct drm_device *ddev = amdgpu_crtc->base.dev;
-	struct amdgpu_device *adev = ddev->dev_private;
-
-	drm_handle_vblank(ddev, amdgpu_crtc->crtc_id);
-	dce_virtual_pageflip(adev, amdgpu_crtc->crtc_id);
-	hrtimer_start(vblank_timer, DCE_VIRTUAL_VBLANK_PERIOD,
-		      HRTIMER_MODE_REL);
-
-	return HRTIMER_NORESTART;
-}
-
-static void dce_virtual_set_crtc_vblank_interrupt_state(struct amdgpu_device *adev,
-							int crtc,
-							enum amdgpu_interrupt_state state)
-{
-	if (crtc >= adev->mode_info.num_crtc) {
-		DRM_DEBUG("invalid crtc %d\n", crtc);
-		return;
-	}
-
-	if (state && !adev->mode_info.crtcs[crtc]->vsync_timer_enabled) {
-		DRM_DEBUG("Enable software vsync timer\n");
-		hrtimer_init(&adev->mode_info.crtcs[crtc]->vblank_timer,
-			     CLOCK_MONOTONIC, HRTIMER_MODE_REL);
-		hrtimer_set_expires(&adev->mode_info.crtcs[crtc]->vblank_timer,
-				    DCE_VIRTUAL_VBLANK_PERIOD);
-		adev->mode_info.crtcs[crtc]->vblank_timer.function =
-			dce_virtual_vblank_timer_handle;
-		hrtimer_start(&adev->mode_info.crtcs[crtc]->vblank_timer,
-			      DCE_VIRTUAL_VBLANK_PERIOD, HRTIMER_MODE_REL);
-	} else if (!state && adev->mode_info.crtcs[crtc]->vsync_timer_enabled) {
-		DRM_DEBUG("Disable software vsync timer\n");
-		hrtimer_cancel(&adev->mode_info.crtcs[crtc]->vblank_timer);
-	}
-
-	adev->mode_info.crtcs[crtc]->vsync_timer_enabled = state;
-	DRM_DEBUG("[FM]set crtc %d vblank interrupt state %d\n", crtc, state);
-}
-
-
-static int dce_virtual_set_crtc_irq_state(struct amdgpu_device *adev,
-					  struct amdgpu_irq_src *source,
-					  unsigned type,
-					  enum amdgpu_interrupt_state state)
-{
-	if (type > AMDGPU_CRTC_IRQ_VBLANK6)
-		return -EINVAL;
-
-	dce_virtual_set_crtc_vblank_interrupt_state(adev, type, state);
-
-	return 0;
-}
-
-static const struct amdgpu_irq_src_funcs dce_virtual_crtc_irq_funcs = {
-	.set = dce_virtual_set_crtc_irq_state,
-	.process = NULL,
-};
-
-static void dce_virtual_set_irq_funcs(struct amdgpu_device *adev)
-{
-	adev->crtc_irq.num_types = AMDGPU_CRTC_IRQ_VBLANK6 + 1;
-	adev->crtc_irq.funcs = &dce_virtual_crtc_irq_funcs;
-}
-
-const struct amdgpu_ip_block_version dce_virtual_ip_block =
-{
-	.type = AMD_IP_BLOCK_TYPE_DCE,
-	.major = 1,
-	.minor = 0,
-	.rev = 0,
-	.funcs = &dce_virtual_ip_funcs,
-};
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_virtual.h b/drivers/gpu/drm/amd/amdgpu/dce_virtual.h
deleted file mode 100644
index ed422012c8c6..000000000000
--- a/drivers/gpu/drm/amd/amdgpu/dce_virtual.h
+++ /dev/null
@@ -1,30 +0,0 @@
-/*
- * Copyright 2014 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-
-#ifndef __DCE_VIRTUAL_H__
-#define __DCE_VIRTUAL_H__
-
-extern const struct amdgpu_ip_block_version dce_virtual_ip_block;
-
-#endif
-
diff --git a/drivers/gpu/drm/amd/amdgpu/df_v1_7.c b/drivers/gpu/drm/amd/amdgpu/df_v1_7.c
new file mode 100644
index 000000000000..cd298556f7a6
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/df_v1_7.c
@@ -0,0 +1,130 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "df_v1_7.h"
+
+#include "df/df_1_7_default.h"
+#include "df/df_1_7_offset.h"
+#include "df/df_1_7_sh_mask.h"
+
+static u32 df_v1_7_channel_number[] = {1, 2, 0, 4, 0, 8, 0, 16, 2};
+
+static void df_v1_7_sw_init(struct amdgpu_device *adev)
+{
+	adev->df.hash_status.hash_64k = false;
+	adev->df.hash_status.hash_2m = false;
+	adev->df.hash_status.hash_1g = false;
+}
+
+static void df_v1_7_sw_fini(struct amdgpu_device *adev)
+{
+}
+
+static void df_v1_7_enable_broadcast_mode(struct amdgpu_device *adev,
+					  bool enable)
+{
+	u32 tmp;
+
+	if (enable) {
+		tmp = RREG32_SOC15(DF, 0, mmFabricConfigAccessControl);
+		tmp &= ~FabricConfigAccessControl__CfgRegInstAccEn_MASK;
+		WREG32_SOC15(DF, 0, mmFabricConfigAccessControl, tmp);
+	} else
+		WREG32_SOC15(DF, 0, mmFabricConfigAccessControl,
+			     mmFabricConfigAccessControl_DEFAULT);
+}
+
+static u32 df_v1_7_get_fb_channel_number(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	tmp = RREG32_SOC15(DF, 0, mmDF_CS_AON0_DramBaseAddress0);
+	tmp &= DF_CS_AON0_DramBaseAddress0__IntLvNumChan_MASK;
+	tmp >>= DF_CS_AON0_DramBaseAddress0__IntLvNumChan__SHIFT;
+
+	return tmp;
+}
+
+static u32 df_v1_7_get_hbm_channel_number(struct amdgpu_device *adev)
+{
+	int fb_channel_number;
+
+	fb_channel_number = adev->df.funcs->get_fb_channel_number(adev);
+	if (fb_channel_number >= ARRAY_SIZE(df_v1_7_channel_number))
+		fb_channel_number = 0;
+
+	return df_v1_7_channel_number[fb_channel_number];
+}
+
+static void df_v1_7_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						     bool enable)
+{
+	u32 tmp;
+
+	/* Put DF on broadcast mode */
+	adev->df.funcs->enable_broadcast_mode(adev, true);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_DF_MGCG)) {
+		tmp = RREG32_SOC15(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater);
+		tmp &= ~DF_PIE_AON0_DfGlobalClkGater__MGCGMode_MASK;
+		tmp |= DF_V1_7_MGCG_ENABLE_15_CYCLE_DELAY;
+		WREG32_SOC15(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater, tmp);
+	} else {
+		tmp = RREG32_SOC15(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater);
+		tmp &= ~DF_PIE_AON0_DfGlobalClkGater__MGCGMode_MASK;
+		tmp |= DF_V1_7_MGCG_DISABLE;
+		WREG32_SOC15(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater, tmp);
+	}
+
+	/* Exit broadcast mode */
+	adev->df.funcs->enable_broadcast_mode(adev, false);
+}
+
+static void df_v1_7_get_clockgating_state(struct amdgpu_device *adev,
+					  u64 *flags)
+{
+	u32 tmp;
+
+	/* AMD_CG_SUPPORT_DF_MGCG */
+	tmp = RREG32_SOC15(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater);
+	if (tmp & DF_V1_7_MGCG_ENABLE_15_CYCLE_DELAY)
+		*flags |= AMD_CG_SUPPORT_DF_MGCG;
+}
+
+static void df_v1_7_enable_ecc_force_par_wr_rmw(struct amdgpu_device *adev,
+						bool enable)
+{
+	WREG32_FIELD15(DF, 0, DF_CS_AON0_CoherentSlaveModeCtrlA0,
+		       ForceParWrRMW, enable);
+}
+
+const struct amdgpu_df_funcs df_v1_7_funcs = {
+	.sw_init = df_v1_7_sw_init,
+	.sw_fini = df_v1_7_sw_fini,
+	.enable_broadcast_mode = df_v1_7_enable_broadcast_mode,
+	.get_fb_channel_number = df_v1_7_get_fb_channel_number,
+	.get_hbm_channel_number = df_v1_7_get_hbm_channel_number,
+	.update_medium_grain_clock_gating = df_v1_7_update_medium_grain_clock_gating,
+	.get_clockgating_state = df_v1_7_get_clockgating_state,
+	.enable_ecc_force_par_wr_rmw = df_v1_7_enable_ecc_force_par_wr_rmw,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/df_v1_7.h b/drivers/gpu/drm/amd/amdgpu/df_v1_7.h
new file mode 100644
index 000000000000..74621104c487
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/df_v1_7.h
@@ -0,0 +1,40 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __DF_V1_7_H__
+#define __DF_V1_7_H__
+
+#include "soc15_common.h"
+enum DF_V1_7_MGCG
+{
+	DF_V1_7_MGCG_DISABLE = 0,
+	DF_V1_7_MGCG_ENABLE_00_CYCLE_DELAY =1,
+	DF_V1_7_MGCG_ENABLE_01_CYCLE_DELAY =2,
+	DF_V1_7_MGCG_ENABLE_15_CYCLE_DELAY =13,
+	DF_V1_7_MGCG_ENABLE_31_CYCLE_DELAY =14,
+	DF_V1_7_MGCG_ENABLE_63_CYCLE_DELAY =15
+};
+
+extern const struct amdgpu_df_funcs df_v1_7_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/df_v3_6.c b/drivers/gpu/drm/amd/amdgpu/df_v3_6.c
new file mode 100644
index 000000000000..621aeca53880
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/df_v3_6.c
@@ -0,0 +1,685 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "df_v3_6.h"
+
+#include "df/df_3_6_default.h"
+#include "df/df_3_6_offset.h"
+#include "df/df_3_6_sh_mask.h"
+
+#define DF_3_6_SMN_REG_INST_DIST        0x8
+#define DF_3_6_INST_CNT                 8
+
+/* Defined in global_features.h as FTI_PERFMON_VISIBLE */
+#define DF_V3_6_MAX_COUNTERS		4
+
+/* get flags from df perfmon config */
+#define DF_V3_6_GET_EVENT(x)		(x & 0xFFUL)
+#define DF_V3_6_GET_INSTANCE(x)		((x >> 8) & 0xFFUL)
+#define DF_V3_6_GET_UNITMASK(x)		((x >> 16) & 0xFFUL)
+#define DF_V3_6_PERFMON_OVERFLOW	0xFFFFFFFFFFFFULL
+
+static u32 df_v3_6_channel_number[] = {1, 2, 0, 4, 0, 8, 0,
+				       16, 32, 0, 0, 0, 2, 4, 8};
+
+static uint64_t df_v3_6_get_fica(struct amdgpu_device *adev,
+				 uint32_t ficaa_val)
+{
+	unsigned long flags, address, data;
+	uint32_t ficadl_val, ficadh_val;
+
+	address = adev->nbio.funcs->get_pcie_index_offset(adev);
+	data = adev->nbio.funcs->get_pcie_data_offset(adev);
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(address, smnDF_PIE_AON_FabricIndirectConfigAccessAddress3);
+	WREG32(data, ficaa_val);
+
+	WREG32(address, smnDF_PIE_AON_FabricIndirectConfigAccessDataLo3);
+	ficadl_val = RREG32(data);
+
+	WREG32(address, smnDF_PIE_AON_FabricIndirectConfigAccessDataHi3);
+	ficadh_val = RREG32(data);
+
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+
+	return (((ficadh_val & 0xFFFFFFFFFFFFFFFF) << 32) | ficadl_val);
+}
+
+static void df_v3_6_set_fica(struct amdgpu_device *adev, uint32_t ficaa_val,
+			     uint32_t ficadl_val, uint32_t ficadh_val)
+{
+	unsigned long flags, address, data;
+
+	address = adev->nbio.funcs->get_pcie_index_offset(adev);
+	data = adev->nbio.funcs->get_pcie_data_offset(adev);
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(address, smnDF_PIE_AON_FabricIndirectConfigAccessAddress3);
+	WREG32(data, ficaa_val);
+
+	WREG32(address, smnDF_PIE_AON_FabricIndirectConfigAccessDataLo3);
+	WREG32(data, ficadl_val);
+
+	WREG32(address, smnDF_PIE_AON_FabricIndirectConfigAccessDataHi3);
+	WREG32(data, ficadh_val);
+
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+}
+
+/*
+ * df_v3_6_perfmon_rreg - read perfmon lo and hi
+ *
+ * required to be atomic.  no mmio method provided so subsequent reads for lo
+ * and hi require to preserve df finite state machine
+ */
+static void df_v3_6_perfmon_rreg(struct amdgpu_device *adev,
+			    uint32_t lo_addr, uint32_t *lo_val,
+			    uint32_t hi_addr, uint32_t *hi_val)
+{
+	unsigned long flags, address, data;
+
+	address = adev->nbio.funcs->get_pcie_index_offset(adev);
+	data = adev->nbio.funcs->get_pcie_data_offset(adev);
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(address, lo_addr);
+	*lo_val = RREG32(data);
+	WREG32(address, hi_addr);
+	*hi_val = RREG32(data);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+}
+
+/*
+ * df_v3_6_perfmon_wreg - write to perfmon lo and hi
+ *
+ * required to be atomic.  no mmio method provided so subsequent reads after
+ * data writes cannot occur to preserve data fabrics finite state machine.
+ */
+static void df_v3_6_perfmon_wreg(struct amdgpu_device *adev, uint32_t lo_addr,
+			    uint32_t lo_val, uint32_t hi_addr, uint32_t hi_val)
+{
+	unsigned long flags, address, data;
+
+	address = adev->nbio.funcs->get_pcie_index_offset(adev);
+	data = adev->nbio.funcs->get_pcie_data_offset(adev);
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(address, lo_addr);
+	WREG32(data, lo_val);
+	WREG32(address, hi_addr);
+	WREG32(data, hi_val);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+}
+
+/* same as perfmon_wreg but return status on write value check */
+static int df_v3_6_perfmon_arm_with_status(struct amdgpu_device *adev,
+					  uint32_t lo_addr, uint32_t lo_val,
+					  uint32_t hi_addr, uint32_t  hi_val)
+{
+	unsigned long flags, address, data;
+	uint32_t lo_val_rb, hi_val_rb;
+
+	address = adev->nbio.funcs->get_pcie_index_offset(adev);
+	data = adev->nbio.funcs->get_pcie_data_offset(adev);
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(address, lo_addr);
+	WREG32(data, lo_val);
+	WREG32(address, hi_addr);
+	WREG32(data, hi_val);
+
+	WREG32(address, lo_addr);
+	lo_val_rb = RREG32(data);
+	WREG32(address, hi_addr);
+	hi_val_rb = RREG32(data);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+
+	if (!(lo_val == lo_val_rb && hi_val == hi_val_rb))
+		return -EBUSY;
+
+	return 0;
+}
+
+
+/*
+ * retry arming counters every 100 usecs within 1 millisecond interval.
+ * if retry fails after time out, return error.
+ */
+#define ARM_RETRY_USEC_TIMEOUT	1000
+#define ARM_RETRY_USEC_INTERVAL	100
+static int df_v3_6_perfmon_arm_with_retry(struct amdgpu_device *adev,
+					  uint32_t lo_addr, uint32_t lo_val,
+					  uint32_t hi_addr, uint32_t  hi_val)
+{
+	int countdown = ARM_RETRY_USEC_TIMEOUT;
+
+	while (countdown) {
+
+		if (!df_v3_6_perfmon_arm_with_status(adev, lo_addr, lo_val,
+						     hi_addr, hi_val))
+			break;
+
+		countdown -= ARM_RETRY_USEC_INTERVAL;
+		udelay(ARM_RETRY_USEC_INTERVAL);
+	}
+
+	return countdown > 0 ? 0 : -ETIME;
+}
+
+/* get the number of df counters available */
+static ssize_t df_v3_6_get_df_cntr_avail(struct device *dev,
+		struct device_attribute *attr,
+		char *buf)
+{
+	struct amdgpu_device *adev;
+	struct drm_device *ddev;
+	int i, count;
+
+	ddev = dev_get_drvdata(dev);
+	adev = drm_to_adev(ddev);
+	count = 0;
+
+	for (i = 0; i < DF_V3_6_MAX_COUNTERS; i++) {
+		if (adev->df_perfmon_config_assign_mask[i] == 0)
+			count++;
+	}
+
+	return sysfs_emit(buf, "%i\n", count);
+}
+
+/* device attr for available perfmon counters */
+static DEVICE_ATTR(df_cntr_avail, S_IRUGO, df_v3_6_get_df_cntr_avail, NULL);
+
+static void df_v3_6_query_hashes(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	adev->df.hash_status.hash_64k = false;
+	adev->df.hash_status.hash_2m = false;
+	adev->df.hash_status.hash_1g = false;
+
+	/* encoding for hash-enabled on Arcturus and Aldebaran */
+	if ((adev->asic_type == CHIP_ARCTURUS &&
+	     adev->df.funcs->get_fb_channel_number(adev) == 0xe) ||
+	     (adev->asic_type == CHIP_ALDEBARAN &&
+	      adev->df.funcs->get_fb_channel_number(adev) == 0x1e)) {
+		tmp = RREG32_SOC15(DF, 0, mmDF_CS_UMC_AON0_DfGlobalCtrl);
+		adev->df.hash_status.hash_64k = REG_GET_FIELD(tmp,
+						DF_CS_UMC_AON0_DfGlobalCtrl,
+						GlbHashIntlvCtl64K);
+		adev->df.hash_status.hash_2m = REG_GET_FIELD(tmp,
+						DF_CS_UMC_AON0_DfGlobalCtrl,
+						GlbHashIntlvCtl2M);
+		adev->df.hash_status.hash_1g = REG_GET_FIELD(tmp,
+						DF_CS_UMC_AON0_DfGlobalCtrl,
+						GlbHashIntlvCtl1G);
+	}
+}
+
+/* init perfmons */
+static void df_v3_6_sw_init(struct amdgpu_device *adev)
+{
+	int i, ret;
+
+	ret = device_create_file(adev->dev, &dev_attr_df_cntr_avail);
+	if (ret)
+		DRM_ERROR("failed to create file for available df counters\n");
+
+	for (i = 0; i < AMDGPU_MAX_DF_PERFMONS; i++)
+		adev->df_perfmon_config_assign_mask[i] = 0;
+
+	df_v3_6_query_hashes(adev);
+}
+
+static void df_v3_6_sw_fini(struct amdgpu_device *adev)
+{
+	if (adev->dev->kobj.sd)
+		device_remove_file(adev->dev, &dev_attr_df_cntr_avail);
+
+}
+
+static void df_v3_6_enable_broadcast_mode(struct amdgpu_device *adev,
+					  bool enable)
+{
+	u32 tmp;
+
+	if (enable) {
+		tmp = RREG32_SOC15(DF, 0, mmFabricConfigAccessControl);
+		tmp &= ~FabricConfigAccessControl__CfgRegInstAccEn_MASK;
+		WREG32_SOC15(DF, 0, mmFabricConfigAccessControl, tmp);
+	} else
+		WREG32_SOC15(DF, 0, mmFabricConfigAccessControl,
+			     mmFabricConfigAccessControl_DEFAULT);
+}
+
+static u32 df_v3_6_get_fb_channel_number(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	if (adev->asic_type == CHIP_ALDEBARAN) {
+		tmp = RREG32_SOC15(DF, 0, mmDF_GCM_AON0_DramMegaBaseAddress0);
+		tmp &=
+		ALDEBARAN_DF_CS_UMC_AON0_DramBaseAddress0__IntLvNumChan_MASK;
+	} else {
+		tmp = RREG32_SOC15(DF, 0, mmDF_CS_UMC_AON0_DramBaseAddress0);
+		tmp &= DF_CS_UMC_AON0_DramBaseAddress0__IntLvNumChan_MASK;
+	}
+	tmp >>= DF_CS_UMC_AON0_DramBaseAddress0__IntLvNumChan__SHIFT;
+
+	return tmp;
+}
+
+static u32 df_v3_6_get_hbm_channel_number(struct amdgpu_device *adev)
+{
+	int fb_channel_number;
+
+	fb_channel_number = adev->df.funcs->get_fb_channel_number(adev);
+	if (fb_channel_number >= ARRAY_SIZE(df_v3_6_channel_number))
+		fb_channel_number = 0;
+
+	return df_v3_6_channel_number[fb_channel_number];
+}
+
+static void df_v3_6_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						     bool enable)
+{
+	u32 tmp;
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_DF_MGCG) {
+		/* Put DF on broadcast mode */
+		adev->df.funcs->enable_broadcast_mode(adev, true);
+
+		if (enable) {
+			tmp = RREG32_SOC15(DF, 0,
+					mmDF_PIE_AON0_DfGlobalClkGater);
+			tmp &= ~DF_PIE_AON0_DfGlobalClkGater__MGCGMode_MASK;
+			tmp |= DF_V3_6_MGCG_ENABLE_15_CYCLE_DELAY;
+			WREG32_SOC15(DF, 0,
+					mmDF_PIE_AON0_DfGlobalClkGater, tmp);
+		} else {
+			tmp = RREG32_SOC15(DF, 0,
+					mmDF_PIE_AON0_DfGlobalClkGater);
+			tmp &= ~DF_PIE_AON0_DfGlobalClkGater__MGCGMode_MASK;
+			tmp |= DF_V3_6_MGCG_DISABLE;
+			WREG32_SOC15(DF, 0,
+					mmDF_PIE_AON0_DfGlobalClkGater, tmp);
+		}
+
+		/* Exit broadcast mode */
+		adev->df.funcs->enable_broadcast_mode(adev, false);
+	}
+}
+
+static void df_v3_6_get_clockgating_state(struct amdgpu_device *adev,
+					  u64 *flags)
+{
+	u32 tmp;
+
+	/* AMD_CG_SUPPORT_DF_MGCG */
+	tmp = RREG32_SOC15(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater);
+	if (tmp & DF_V3_6_MGCG_ENABLE_15_CYCLE_DELAY)
+		*flags |= AMD_CG_SUPPORT_DF_MGCG;
+}
+
+/* get assigned df perfmon ctr as int */
+static bool df_v3_6_pmc_has_counter(struct amdgpu_device *adev,
+				      uint64_t config,
+				      int counter_idx)
+{
+
+	return ((config & 0x0FFFFFFUL) ==
+			adev->df_perfmon_config_assign_mask[counter_idx]);
+
+}
+
+/* get address based on counter assignment */
+static void df_v3_6_pmc_get_addr(struct amdgpu_device *adev,
+				 uint64_t config,
+				 int counter_idx,
+				 int is_ctrl,
+				 uint32_t *lo_base_addr,
+				 uint32_t *hi_base_addr)
+{
+	if (!df_v3_6_pmc_has_counter(adev, config, counter_idx))
+		return;
+
+	switch (counter_idx) {
+
+	case 0:
+		*lo_base_addr = is_ctrl ? smnPerfMonCtlLo4 : smnPerfMonCtrLo4;
+		*hi_base_addr = is_ctrl ? smnPerfMonCtlHi4 : smnPerfMonCtrHi4;
+		break;
+	case 1:
+		*lo_base_addr = is_ctrl ? smnPerfMonCtlLo5 : smnPerfMonCtrLo5;
+		*hi_base_addr = is_ctrl ? smnPerfMonCtlHi5 : smnPerfMonCtrHi5;
+		break;
+	case 2:
+		*lo_base_addr = is_ctrl ? smnPerfMonCtlLo6 : smnPerfMonCtrLo6;
+		*hi_base_addr = is_ctrl ? smnPerfMonCtlHi6 : smnPerfMonCtrHi6;
+		break;
+	case 3:
+		*lo_base_addr = is_ctrl ? smnPerfMonCtlLo7 : smnPerfMonCtrLo7;
+		*hi_base_addr = is_ctrl ? smnPerfMonCtlHi7 : smnPerfMonCtrHi7;
+		break;
+
+	}
+
+}
+
+/* get read counter address */
+static void df_v3_6_pmc_get_read_settings(struct amdgpu_device *adev,
+					  uint64_t config,
+					  int counter_idx,
+					  uint32_t *lo_base_addr,
+					  uint32_t *hi_base_addr)
+{
+	df_v3_6_pmc_get_addr(adev, config, counter_idx, 0, lo_base_addr,
+								hi_base_addr);
+}
+
+/* get control counter settings i.e. address and values to set */
+static int df_v3_6_pmc_get_ctrl_settings(struct amdgpu_device *adev,
+					  uint64_t config,
+					  int counter_idx,
+					  uint32_t *lo_base_addr,
+					  uint32_t *hi_base_addr,
+					  uint32_t *lo_val,
+					  uint32_t *hi_val,
+					  bool is_enable)
+{
+
+	uint32_t eventsel, instance, unitmask;
+	uint32_t instance_10, instance_5432, instance_76;
+
+	df_v3_6_pmc_get_addr(adev, config, counter_idx, 1, lo_base_addr,
+				hi_base_addr);
+
+	if ((*lo_base_addr == 0) || (*hi_base_addr == 0)) {
+		DRM_ERROR("[DF PMC] addressing not retrieved! Lo: %x, Hi: %x",
+				*lo_base_addr, *hi_base_addr);
+		return -ENXIO;
+	}
+
+	eventsel = DF_V3_6_GET_EVENT(config) & 0x3f;
+	unitmask = DF_V3_6_GET_UNITMASK(config) & 0xf;
+	instance = DF_V3_6_GET_INSTANCE(config);
+
+	instance_10 = instance & 0x3;
+	instance_5432 = (instance >> 2) & 0xf;
+	instance_76 = (instance >> 6) & 0x3;
+
+	*lo_val = (unitmask << 8) | (instance_10 << 6) | eventsel;
+	*lo_val = is_enable ? *lo_val | (1 << 22) : *lo_val & ~(1 << 22);
+	*hi_val = (instance_76 << 29) | instance_5432;
+
+	DRM_DEBUG_DRIVER("config=%llx addr=%08x:%08x val=%08x:%08x",
+		config, *lo_base_addr, *hi_base_addr, *lo_val, *hi_val);
+
+	return 0;
+}
+
+/* add df performance counters for read */
+static int df_v3_6_pmc_add_cntr(struct amdgpu_device *adev,
+				   uint64_t config)
+{
+	int i;
+
+	for (i = 0; i < DF_V3_6_MAX_COUNTERS; i++) {
+		if (adev->df_perfmon_config_assign_mask[i] == 0U) {
+			adev->df_perfmon_config_assign_mask[i] =
+							config & 0x0FFFFFFUL;
+			return i;
+		}
+	}
+
+	return -ENOSPC;
+}
+
+#define DEFERRED_ARM_MASK	(1 << 31)
+static int df_v3_6_pmc_set_deferred(struct amdgpu_device *adev,
+				    uint64_t config, int counter_idx,
+				    bool is_deferred)
+{
+
+	if (!df_v3_6_pmc_has_counter(adev, config, counter_idx))
+		return -EINVAL;
+
+	if (is_deferred)
+		adev->df_perfmon_config_assign_mask[counter_idx] |=
+							DEFERRED_ARM_MASK;
+	else
+		adev->df_perfmon_config_assign_mask[counter_idx] &=
+							~DEFERRED_ARM_MASK;
+
+	return 0;
+}
+
+static bool df_v3_6_pmc_is_deferred(struct amdgpu_device *adev,
+				    uint64_t config,
+				    int counter_idx)
+{
+	return	(df_v3_6_pmc_has_counter(adev, config, counter_idx) &&
+			(adev->df_perfmon_config_assign_mask[counter_idx]
+				& DEFERRED_ARM_MASK));
+
+}
+
+/* release performance counter */
+static void df_v3_6_pmc_release_cntr(struct amdgpu_device *adev,
+				     uint64_t config,
+				     int counter_idx)
+{
+	if (df_v3_6_pmc_has_counter(adev, config, counter_idx))
+		adev->df_perfmon_config_assign_mask[counter_idx] = 0ULL;
+}
+
+
+static void df_v3_6_reset_perfmon_cntr(struct amdgpu_device *adev,
+					 uint64_t config,
+					 int counter_idx)
+{
+	uint32_t lo_base_addr = 0, hi_base_addr = 0;
+
+	df_v3_6_pmc_get_read_settings(adev, config, counter_idx, &lo_base_addr,
+				      &hi_base_addr);
+
+	if ((lo_base_addr == 0) || (hi_base_addr == 0))
+		return;
+
+	df_v3_6_perfmon_wreg(adev, lo_base_addr, 0, hi_base_addr, 0);
+}
+
+/* return available counter if is_add == 1 otherwise return error status. */
+static int df_v3_6_pmc_start(struct amdgpu_device *adev, uint64_t config,
+			     int counter_idx, int is_add)
+{
+	uint32_t lo_base_addr, hi_base_addr, lo_val, hi_val;
+	int err = 0, ret = 0;
+
+	switch (adev->asic_type) {
+	case CHIP_VEGA20:
+	case CHIP_ARCTURUS:
+		if (is_add)
+			return df_v3_6_pmc_add_cntr(adev, config);
+
+		ret = df_v3_6_pmc_get_ctrl_settings(adev,
+					config,
+					counter_idx,
+					&lo_base_addr,
+					&hi_base_addr,
+					&lo_val,
+					&hi_val,
+					true);
+
+		if (ret)
+			return ret;
+
+		err = df_v3_6_perfmon_arm_with_retry(adev,
+						     lo_base_addr,
+						     lo_val,
+						     hi_base_addr,
+						     hi_val);
+
+		if (err)
+			ret = df_v3_6_pmc_set_deferred(adev, config,
+							counter_idx, true);
+
+		break;
+	default:
+		break;
+	}
+
+	return ret;
+}
+
+static int df_v3_6_pmc_stop(struct amdgpu_device *adev, uint64_t config,
+			    int counter_idx, int is_remove)
+{
+	uint32_t lo_base_addr, hi_base_addr, lo_val, hi_val;
+	int ret = 0;
+
+	switch (adev->asic_type) {
+	case CHIP_VEGA20:
+	case CHIP_ARCTURUS:
+		ret = df_v3_6_pmc_get_ctrl_settings(adev,
+			config,
+			counter_idx,
+			&lo_base_addr,
+			&hi_base_addr,
+			&lo_val,
+			&hi_val,
+			false);
+
+		if (ret)
+			return ret;
+
+		df_v3_6_perfmon_wreg(adev, lo_base_addr, lo_val,
+							hi_base_addr, hi_val);
+
+		if (is_remove) {
+			df_v3_6_reset_perfmon_cntr(adev, config, counter_idx);
+			df_v3_6_pmc_release_cntr(adev, config, counter_idx);
+		}
+
+		break;
+	default:
+		break;
+	}
+
+	return ret;
+}
+
+static void df_v3_6_pmc_get_count(struct amdgpu_device *adev,
+				  uint64_t config,
+				  int counter_idx,
+				  uint64_t *count)
+{
+	uint32_t lo_base_addr = 0, hi_base_addr = 0, lo_val = 0, hi_val = 0;
+	*count = 0;
+
+	switch (adev->asic_type) {
+	case CHIP_VEGA20:
+	case CHIP_ARCTURUS:
+		df_v3_6_pmc_get_read_settings(adev, config, counter_idx,
+						&lo_base_addr, &hi_base_addr);
+
+		if ((lo_base_addr == 0) || (hi_base_addr == 0))
+			return;
+
+		/* rearm the counter or throw away count value on failure */
+		if (df_v3_6_pmc_is_deferred(adev, config, counter_idx)) {
+			int rearm_err = df_v3_6_perfmon_arm_with_status(adev,
+							lo_base_addr, lo_val,
+							hi_base_addr, hi_val);
+
+			if (rearm_err)
+				return;
+
+			df_v3_6_pmc_set_deferred(adev, config, counter_idx,
+									false);
+		}
+
+		df_v3_6_perfmon_rreg(adev, lo_base_addr, &lo_val,
+				hi_base_addr, &hi_val);
+
+		*count  = ((hi_val | 0ULL) << 32) | (lo_val | 0ULL);
+
+		if (*count >= DF_V3_6_PERFMON_OVERFLOW)
+			*count = 0;
+
+		DRM_DEBUG_DRIVER("config=%llx addr=%08x:%08x val=%08x:%08x",
+			 config, lo_base_addr, hi_base_addr, lo_val, hi_val);
+
+		break;
+	default:
+		break;
+	}
+}
+
+static bool df_v3_6_query_ras_poison_mode(struct amdgpu_device *adev)
+{
+	uint32_t hw_assert_msklo, hw_assert_mskhi;
+	uint32_t v0, v1, v28, v31;
+
+	hw_assert_msklo = RREG32_SOC15(DF, 0,
+				mmDF_CS_UMC_AON0_HardwareAssertMaskLow);
+	hw_assert_mskhi = RREG32_SOC15(DF, 0,
+				mmDF_NCS_PG0_HardwareAssertMaskHigh);
+
+	v0 = REG_GET_FIELD(hw_assert_msklo,
+		DF_CS_UMC_AON0_HardwareAssertMaskLow, HWAssertMsk0);
+	v1 = REG_GET_FIELD(hw_assert_msklo,
+		DF_CS_UMC_AON0_HardwareAssertMaskLow, HWAssertMsk1);
+	v28 = REG_GET_FIELD(hw_assert_mskhi,
+		DF_NCS_PG0_HardwareAssertMaskHigh, HWAssertMsk28);
+	v31 = REG_GET_FIELD(hw_assert_mskhi,
+		DF_NCS_PG0_HardwareAssertMaskHigh, HWAssertMsk31);
+
+	if (v0 && v1 && v28 && v31)
+		return true;
+	else if (!v0 && !v1 && !v28 && !v31)
+		return false;
+	else {
+		dev_warn(adev->dev, "DF poison setting is inconsistent(%d:%d:%d:%d)!\n",
+				v0, v1, v28, v31);
+		return false;
+	}
+}
+
+const struct amdgpu_df_funcs df_v3_6_funcs = {
+	.sw_init = df_v3_6_sw_init,
+	.sw_fini = df_v3_6_sw_fini,
+	.enable_broadcast_mode = df_v3_6_enable_broadcast_mode,
+	.get_fb_channel_number = df_v3_6_get_fb_channel_number,
+	.get_hbm_channel_number = df_v3_6_get_hbm_channel_number,
+	.update_medium_grain_clock_gating =
+			df_v3_6_update_medium_grain_clock_gating,
+	.get_clockgating_state = df_v3_6_get_clockgating_state,
+	.pmc_start = df_v3_6_pmc_start,
+	.pmc_stop = df_v3_6_pmc_stop,
+	.pmc_get_count = df_v3_6_pmc_get_count,
+	.get_fica = df_v3_6_get_fica,
+	.set_fica = df_v3_6_set_fica,
+	.query_ras_poison_mode = df_v3_6_query_ras_poison_mode,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/df_v3_6.h b/drivers/gpu/drm/amd/amdgpu/df_v3_6.h
new file mode 100644
index 000000000000..2505c7ef258a
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/df_v3_6.h
@@ -0,0 +1,41 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __DF_V3_6_H__
+#define __DF_V3_6_H__
+
+#include "soc15_common.h"
+
+enum DF_V3_6_MGCG {
+	DF_V3_6_MGCG_DISABLE = 0,
+	DF_V3_6_MGCG_ENABLE_00_CYCLE_DELAY = 1,
+	DF_V3_6_MGCG_ENABLE_01_CYCLE_DELAY = 2,
+	DF_V3_6_MGCG_ENABLE_15_CYCLE_DELAY = 13,
+	DF_V3_6_MGCG_ENABLE_31_CYCLE_DELAY = 14,
+	DF_V3_6_MGCG_ENABLE_63_CYCLE_DELAY = 15
+};
+
+extern const struct attribute_group *df_v3_6_attr_groups[];
+extern const struct amdgpu_df_funcs df_v3_6_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/df_v4_15.c b/drivers/gpu/drm/amd/amdgpu/df_v4_15.c
new file mode 100644
index 000000000000..2a573e33908b
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/df_v4_15.c
@@ -0,0 +1,45 @@
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "df_v4_15.h"
+
+#include "df/df_4_15_offset.h"
+#include "df/df_4_15_sh_mask.h"
+
+static void df_v4_15_hw_init(struct amdgpu_device *adev)
+{
+	if (adev->have_atomics_support) {
+		uint32_t tmp;
+		uint32_t dis_lcl_proc = (1 <<  1 |
+					1 <<  2 |
+					1 << 13);
+
+		tmp = RREG32_SOC15(DF, 0, regNCSConfigurationRegister1);
+		tmp |= (dis_lcl_proc << NCSConfigurationRegister1__DisIntAtomicsLclProcessing__SHIFT);
+		WREG32_SOC15(DF, 0, regNCSConfigurationRegister1, tmp);
+	}
+}
+
+const struct amdgpu_df_funcs df_v4_15_funcs = {
+	.hw_init = df_v4_15_hw_init
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/df_v4_15.h b/drivers/gpu/drm/amd/amdgpu/df_v4_15.h
new file mode 100644
index 000000000000..dddf2422112a
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/df_v4_15.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __DF_V4_15_H__
+#define __DF_V4_15_H__
+
+extern const struct amdgpu_df_funcs df_v4_15_funcs;
+
+#endif /* __DF_V4_15_H__ */
+
diff --git a/drivers/gpu/drm/amd/amdgpu/df_v4_3.c b/drivers/gpu/drm/amd/amdgpu/df_v4_3.c
new file mode 100644
index 000000000000..e8b9e19ede2e
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/df_v4_3.c
@@ -0,0 +1,61 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "df_v4_3.h"
+
+#include "df/df_4_3_offset.h"
+#include "df/df_4_3_sh_mask.h"
+
+static bool df_v4_3_query_ras_poison_mode(struct amdgpu_device *adev)
+{
+	uint32_t hw_assert_msklo, hw_assert_mskhi;
+	uint32_t v0, v1, v28, v31;
+
+	hw_assert_msklo = RREG32_SOC15(DF, 0,
+				regDF_CS_UMC_AON0_HardwareAssertMaskLow);
+	hw_assert_mskhi = RREG32_SOC15(DF, 0,
+				regDF_NCS_PG0_HardwareAssertMaskHigh);
+
+	v0 = REG_GET_FIELD(hw_assert_msklo,
+		DF_CS_UMC_AON0_HardwareAssertMaskLow, HWAssertMsk0);
+	v1 = REG_GET_FIELD(hw_assert_msklo,
+		DF_CS_UMC_AON0_HardwareAssertMaskLow, HWAssertMsk1);
+	v28 = REG_GET_FIELD(hw_assert_mskhi,
+		DF_NCS_PG0_HardwareAssertMaskHigh, HWAssertMsk28);
+	v31 = REG_GET_FIELD(hw_assert_mskhi,
+		DF_NCS_PG0_HardwareAssertMaskHigh, HWAssertMsk31);
+
+	if (v0 && v1 && v28 && v31)
+		return true;
+	else if (!v0 && !v1 && !v28 && !v31)
+		return false;
+	else {
+		dev_warn(adev->dev, "DF poison setting is inconsistent(%d:%d:%d:%d)!\n",
+				v0, v1, v28, v31);
+		return false;
+	}
+}
+
+const struct amdgpu_df_funcs df_v4_3_funcs = {
+	.query_ras_poison_mode = df_v4_3_query_ras_poison_mode,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/df_v4_3.h b/drivers/gpu/drm/amd/amdgpu/df_v4_3.h
new file mode 100644
index 000000000000..06ef0724edd3
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/df_v4_3.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __DF_V4_3_H__
+#define __DF_V4_3_H__
+
+#include "soc15_common.h"
+
+extern const struct amdgpu_df_funcs df_v4_3_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/df_v4_6_2.c b/drivers/gpu/drm/amd/amdgpu/df_v4_6_2.c
new file mode 100644
index 000000000000..a47960a0babd
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/df_v4_6_2.c
@@ -0,0 +1,34 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "df_v4_6_2.h"
+
+static bool df_v4_6_2_query_ras_poison_mode(struct amdgpu_device *adev)
+{
+	/* return true since related regs are inaccessible */
+	return true;
+}
+
+const struct amdgpu_df_funcs df_v4_6_2_funcs = {
+	.query_ras_poison_mode = df_v4_6_2_query_ras_poison_mode,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/df_v4_6_2.h b/drivers/gpu/drm/amd/amdgpu/df_v4_6_2.h
new file mode 100644
index 000000000000..3bc3e6d216e2
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/df_v4_6_2.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __DF_V4_6_2_H__
+#define __DF_V4_6_2_H__
+
+#include "soc15_common.h"
+
+extern const struct amdgpu_df_funcs df_v4_6_2_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/dimgrey_cavefish_reg_init.c b/drivers/gpu/drm/amd/amdgpu/dimgrey_cavefish_reg_init.c
new file mode 100644
index 000000000000..e9f177e9e3cf
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/dimgrey_cavefish_reg_init.c
@@ -0,0 +1,54 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "nv.h"
+
+#include "soc15_common.h"
+#include "soc15_hw_ip.h"
+#include "dimgrey_cavefish_ip_offset.h"
+
+int dimgrey_cavefish_reg_base_init(struct amdgpu_device *adev)
+{
+	/* HW has more IP blocks,  only initialize the block needed by driver */
+	uint32_t i;
+	for (i = 0 ; i < MAX_INSTANCE ; ++i) {
+		adev->reg_offset[GC_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i]));
+		adev->reg_offset[HDP_HWIP][i] = (uint32_t *)(&(HDP_BASE.instance[i]));
+		adev->reg_offset[MMHUB_HWIP][i] = (uint32_t *)(&(MMHUB_BASE.instance[i]));
+		adev->reg_offset[ATHUB_HWIP][i] = (uint32_t *)(&(ATHUB_BASE.instance[i]));
+		adev->reg_offset[NBIO_HWIP][i] = (uint32_t *)(&(NBIO_BASE.instance[i]));
+		adev->reg_offset[MP0_HWIP][i] = (uint32_t *)(&(MP0_BASE.instance[i]));
+		adev->reg_offset[MP1_HWIP][i] = (uint32_t *)(&(MP1_BASE.instance[i]));
+		adev->reg_offset[VCN_HWIP][i] = (uint32_t *)(&(VCN0_BASE.instance[i]));
+		adev->reg_offset[DF_HWIP][i] = (uint32_t *)(&(DF_BASE.instance[i]));
+		adev->reg_offset[DCE_HWIP][i] = (uint32_t *)(&(DCN_BASE.instance[i]));
+		adev->reg_offset[OSSSYS_HWIP][i] = (uint32_t *)(&(OSSSYS_BASE.instance[i]));
+		adev->reg_offset[SDMA0_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i]));
+		adev->reg_offset[SDMA1_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i]));
+		adev->reg_offset[SDMA2_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i]));
+		adev->reg_offset[SDMA3_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i]));
+		adev->reg_offset[SMUIO_HWIP][i] = (uint32_t *)(&(SMUIO_BASE.instance[i]));
+		adev->reg_offset[THM_HWIP][i] = (uint32_t *)(&(THM_BASE.instance[i]));
+	}
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/emu_soc.c b/drivers/gpu/drm/amd/amdgpu/emu_soc.c
new file mode 100644
index 000000000000..d72c25c1b987
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/emu_soc.c
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "soc15.h"
+
+#include "soc15_common.h"
+#include "soc15_hw_ip.h"
+
+int emu_soc_asic_init(struct amdgpu_device *adev)
+{
+	return 0;
+}
+
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
new file mode 100644
index 000000000000..8841d7213de4
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
@@ -0,0 +1,10247 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/delay.h>
+#include <linux/kernel.h>
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+#include "amdgpu.h"
+#include "amdgpu_gfx.h"
+#include "amdgpu_psp.h"
+#include "nv.h"
+#include "nvd.h"
+
+#include "gc/gc_10_1_0_offset.h"
+#include "gc/gc_10_1_0_sh_mask.h"
+#include "smuio/smuio_11_0_0_offset.h"
+#include "smuio/smuio_11_0_0_sh_mask.h"
+#include "navi10_enum.h"
+#include "ivsrcid/gfx/irqsrcs_gfx_10_1.h"
+
+#include "soc15.h"
+#include "soc15_common.h"
+#include "clearstate_gfx10.h"
+#include "v10_structs.h"
+#include "gfx_v10_0.h"
+#include "gfx_v10_0_cleaner_shader.h"
+#include "nbio_v2_3.h"
+
+/*
+ * Navi10 has two graphic rings to share each graphic pipe.
+ * 1. Primary ring
+ * 2. Async ring
+ */
+#define GFX10_NUM_GFX_RINGS_NV1X	1
+#define GFX10_NUM_GFX_RINGS_Sienna_Cichlid	2
+#define GFX10_MEC_HPD_SIZE	2048
+
+#define F32_CE_PROGRAM_RAM_SIZE		65536
+#define RLCG_UCODE_LOADING_START_ADDRESS	0x00002000L
+
+#define mmCGTT_GS_NGG_CLK_CTRL	0x5087
+#define mmCGTT_GS_NGG_CLK_CTRL_BASE_IDX	1
+#define mmCGTT_SPI_RA0_CLK_CTRL 0x507a
+#define mmCGTT_SPI_RA0_CLK_CTRL_BASE_IDX 1
+#define mmCGTT_SPI_RA1_CLK_CTRL 0x507b
+#define mmCGTT_SPI_RA1_CLK_CTRL_BASE_IDX 1
+
+#define GB_ADDR_CONFIG__NUM_PKRS__SHIFT                                                                       0x8
+#define GB_ADDR_CONFIG__NUM_PKRS_MASK                                                                         0x00000700L
+
+#define mmCGTS_TCC_DISABLE_gc_10_3                 0x5006
+#define mmCGTS_TCC_DISABLE_gc_10_3_BASE_IDX        1
+#define mmCGTS_USER_TCC_DISABLE_gc_10_3            0x5007
+#define mmCGTS_USER_TCC_DISABLE_gc_10_3_BASE_IDX   1
+
+#define mmCP_MEC_CNTL_Sienna_Cichlid                      0x0f55
+#define mmCP_MEC_CNTL_Sienna_Cichlid_BASE_IDX             0
+#define mmRLC_SAFE_MODE_Sienna_Cichlid			0x4ca0
+#define mmRLC_SAFE_MODE_Sienna_Cichlid_BASE_IDX		1
+#define mmRLC_CP_SCHEDULERS_Sienna_Cichlid		0x4ca1
+#define mmRLC_CP_SCHEDULERS_Sienna_Cichlid_BASE_IDX	1
+#define mmSPI_CONFIG_CNTL_Sienna_Cichlid			0x11ec
+#define mmSPI_CONFIG_CNTL_Sienna_Cichlid_BASE_IDX		0
+#define mmVGT_ESGS_RING_SIZE_Sienna_Cichlid		0x0fc1
+#define mmVGT_ESGS_RING_SIZE_Sienna_Cichlid_BASE_IDX	0
+#define mmVGT_GSVS_RING_SIZE_Sienna_Cichlid		0x0fc2
+#define mmVGT_GSVS_RING_SIZE_Sienna_Cichlid_BASE_IDX	0
+#define mmVGT_TF_RING_SIZE_Sienna_Cichlid			0x0fc3
+#define mmVGT_TF_RING_SIZE_Sienna_Cichlid_BASE_IDX	0
+#define mmVGT_HS_OFFCHIP_PARAM_Sienna_Cichlid		0x0fc4
+#define mmVGT_HS_OFFCHIP_PARAM_Sienna_Cichlid_BASE_IDX	0
+#define mmVGT_TF_MEMORY_BASE_Sienna_Cichlid		0x0fc5
+#define mmVGT_TF_MEMORY_BASE_Sienna_Cichlid_BASE_IDX	0
+#define mmVGT_TF_MEMORY_BASE_HI_Sienna_Cichlid		0x0fc6
+#define mmVGT_TF_MEMORY_BASE_HI_Sienna_Cichlid_BASE_IDX	0
+#define GRBM_STATUS2__RLC_BUSY_Sienna_Cichlid__SHIFT	0x1a
+#define GRBM_STATUS2__RLC_BUSY_Sienna_Cichlid_MASK	0x04000000L
+#define CP_RB_DOORBELL_RANGE_LOWER__DOORBELL_RANGE_LOWER_Sienna_Cichlid_MASK	0x00000FFCL
+#define CP_RB_DOORBELL_RANGE_LOWER__DOORBELL_RANGE_LOWER_Sienna_Cichlid__SHIFT	0x2
+#define CP_RB_DOORBELL_RANGE_UPPER__DOORBELL_RANGE_UPPER_Sienna_Cichlid_MASK	0x00000FFCL
+#define mmGCR_GENERAL_CNTL_Sienna_Cichlid			0x1580
+#define mmGCR_GENERAL_CNTL_Sienna_Cichlid_BASE_IDX	0
+
+#define mmGOLDEN_TSC_COUNT_UPPER_Cyan_Skillfish                0x0105
+#define mmGOLDEN_TSC_COUNT_UPPER_Cyan_Skillfish_BASE_IDX       1
+#define mmGOLDEN_TSC_COUNT_LOWER_Cyan_Skillfish                0x0106
+#define mmGOLDEN_TSC_COUNT_LOWER_Cyan_Skillfish_BASE_IDX       1
+
+#define mmGOLDEN_TSC_COUNT_UPPER_Vangogh                0x0025
+#define mmGOLDEN_TSC_COUNT_UPPER_Vangogh_BASE_IDX       1
+#define mmGOLDEN_TSC_COUNT_LOWER_Vangogh                0x0026
+#define mmGOLDEN_TSC_COUNT_LOWER_Vangogh_BASE_IDX       1
+
+#define mmGOLDEN_TSC_COUNT_UPPER_GC_10_3_6                0x002d
+#define mmGOLDEN_TSC_COUNT_UPPER_GC_10_3_6_BASE_IDX       1
+#define mmGOLDEN_TSC_COUNT_LOWER_GC_10_3_6                0x002e
+#define mmGOLDEN_TSC_COUNT_LOWER_GC_10_3_6_BASE_IDX       1
+
+#define mmSPI_CONFIG_CNTL_1_Vangogh		 0x2441
+#define mmSPI_CONFIG_CNTL_1_Vangogh_BASE_IDX	 1
+#define mmVGT_TF_MEMORY_BASE_HI_Vangogh          0x2261
+#define mmVGT_TF_MEMORY_BASE_HI_Vangogh_BASE_IDX 1
+#define mmVGT_HS_OFFCHIP_PARAM_Vangogh           0x224f
+#define mmVGT_HS_OFFCHIP_PARAM_Vangogh_BASE_IDX  1
+#define mmVGT_TF_RING_SIZE_Vangogh               0x224e
+#define mmVGT_TF_RING_SIZE_Vangogh_BASE_IDX      1
+#define mmVGT_GSVS_RING_SIZE_Vangogh             0x2241
+#define mmVGT_GSVS_RING_SIZE_Vangogh_BASE_IDX    1
+#define mmVGT_TF_MEMORY_BASE_Vangogh             0x2250
+#define mmVGT_TF_MEMORY_BASE_Vangogh_BASE_IDX    1
+#define mmVGT_ESGS_RING_SIZE_Vangogh             0x2240
+#define mmVGT_ESGS_RING_SIZE_Vangogh_BASE_IDX    1
+#define mmSPI_CONFIG_CNTL_Vangogh                0x2440
+#define mmSPI_CONFIG_CNTL_Vangogh_BASE_IDX       1
+#define mmGCR_GENERAL_CNTL_Vangogh               0x1580
+#define mmGCR_GENERAL_CNTL_Vangogh_BASE_IDX      0
+#define RLC_PG_DELAY_3__CGCG_ACTIVE_BEFORE_CGPG_MASK_Vangogh   0x0000FFFFL
+
+#define mmCP_HYP_PFP_UCODE_ADDR			0x5814
+#define mmCP_HYP_PFP_UCODE_ADDR_BASE_IDX	1
+#define mmCP_HYP_PFP_UCODE_DATA			0x5815
+#define mmCP_HYP_PFP_UCODE_DATA_BASE_IDX	1
+#define mmCP_HYP_CE_UCODE_ADDR			0x5818
+#define mmCP_HYP_CE_UCODE_ADDR_BASE_IDX		1
+#define mmCP_HYP_CE_UCODE_DATA			0x5819
+#define mmCP_HYP_CE_UCODE_DATA_BASE_IDX		1
+#define mmCP_HYP_ME_UCODE_ADDR			0x5816
+#define mmCP_HYP_ME_UCODE_ADDR_BASE_IDX		1
+#define mmCP_HYP_ME_UCODE_DATA			0x5817
+#define mmCP_HYP_ME_UCODE_DATA_BASE_IDX		1
+
+#define mmCPG_PSP_DEBUG				0x5c10
+#define mmCPG_PSP_DEBUG_BASE_IDX		1
+#define mmCPC_PSP_DEBUG				0x5c11
+#define mmCPC_PSP_DEBUG_BASE_IDX		1
+#define CPC_PSP_DEBUG__GPA_OVERRIDE_MASK	0x00000008L
+#define CPG_PSP_DEBUG__GPA_OVERRIDE_MASK	0x00000008L
+
+//CC_GC_SA_UNIT_DISABLE
+#define mmCC_GC_SA_UNIT_DISABLE                 0x0fe9
+#define mmCC_GC_SA_UNIT_DISABLE_BASE_IDX        0
+#define CC_GC_SA_UNIT_DISABLE__SA_DISABLE__SHIFT	0x8
+#define CC_GC_SA_UNIT_DISABLE__SA_DISABLE_MASK		0x0000FF00L
+//GC_USER_SA_UNIT_DISABLE
+#define mmGC_USER_SA_UNIT_DISABLE               0x0fea
+#define mmGC_USER_SA_UNIT_DISABLE_BASE_IDX      0
+#define GC_USER_SA_UNIT_DISABLE__SA_DISABLE__SHIFT	0x8
+#define GC_USER_SA_UNIT_DISABLE__SA_DISABLE_MASK	0x0000FF00L
+//PA_SC_ENHANCE_3
+#define mmPA_SC_ENHANCE_3                       0x1085
+#define mmPA_SC_ENHANCE_3_BASE_IDX              0
+#define PA_SC_ENHANCE_3__FORCE_PBB_WORKLOAD_MODE_TO_ZERO__SHIFT 0x3
+#define PA_SC_ENHANCE_3__FORCE_PBB_WORKLOAD_MODE_TO_ZERO_MASK   0x00000008L
+
+#define mmCGTT_SPI_CS_CLK_CTRL			0x507c
+#define mmCGTT_SPI_CS_CLK_CTRL_BASE_IDX         1
+
+#define mmGCUTCL2_CGTT_CLK_CTRL_Sienna_Cichlid		0x16f3
+#define mmGCUTCL2_CGTT_CLK_CTRL_Sienna_Cichlid_BASE_IDX	0
+#define mmGCVM_L2_CGTT_CLK_CTRL_Sienna_Cichlid          0x15db
+#define mmGCVM_L2_CGTT_CLK_CTRL_Sienna_Cichlid_BASE_IDX	0
+
+#define mmGC_THROTTLE_CTRL_Sienna_Cichlid              0x2030
+#define mmGC_THROTTLE_CTRL_Sienna_Cichlid_BASE_IDX     0
+
+#define mmRLC_SPARE_INT_0_Sienna_Cichlid               0x4ca5
+#define mmRLC_SPARE_INT_0_Sienna_Cichlid_BASE_IDX      1
+
+MODULE_FIRMWARE("amdgpu/navi10_ce.bin");
+MODULE_FIRMWARE("amdgpu/navi10_pfp.bin");
+MODULE_FIRMWARE("amdgpu/navi10_me.bin");
+MODULE_FIRMWARE("amdgpu/navi10_mec.bin");
+MODULE_FIRMWARE("amdgpu/navi10_mec2.bin");
+MODULE_FIRMWARE("amdgpu/navi10_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/navi14_ce_wks.bin");
+MODULE_FIRMWARE("amdgpu/navi14_pfp_wks.bin");
+MODULE_FIRMWARE("amdgpu/navi14_me_wks.bin");
+MODULE_FIRMWARE("amdgpu/navi14_mec_wks.bin");
+MODULE_FIRMWARE("amdgpu/navi14_mec2_wks.bin");
+MODULE_FIRMWARE("amdgpu/navi14_ce.bin");
+MODULE_FIRMWARE("amdgpu/navi14_pfp.bin");
+MODULE_FIRMWARE("amdgpu/navi14_me.bin");
+MODULE_FIRMWARE("amdgpu/navi14_mec.bin");
+MODULE_FIRMWARE("amdgpu/navi14_mec2.bin");
+MODULE_FIRMWARE("amdgpu/navi14_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/navi12_ce.bin");
+MODULE_FIRMWARE("amdgpu/navi12_pfp.bin");
+MODULE_FIRMWARE("amdgpu/navi12_me.bin");
+MODULE_FIRMWARE("amdgpu/navi12_mec.bin");
+MODULE_FIRMWARE("amdgpu/navi12_mec2.bin");
+MODULE_FIRMWARE("amdgpu/navi12_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/sienna_cichlid_ce.bin");
+MODULE_FIRMWARE("amdgpu/sienna_cichlid_pfp.bin");
+MODULE_FIRMWARE("amdgpu/sienna_cichlid_me.bin");
+MODULE_FIRMWARE("amdgpu/sienna_cichlid_mec.bin");
+MODULE_FIRMWARE("amdgpu/sienna_cichlid_mec2.bin");
+MODULE_FIRMWARE("amdgpu/sienna_cichlid_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/navy_flounder_ce.bin");
+MODULE_FIRMWARE("amdgpu/navy_flounder_pfp.bin");
+MODULE_FIRMWARE("amdgpu/navy_flounder_me.bin");
+MODULE_FIRMWARE("amdgpu/navy_flounder_mec.bin");
+MODULE_FIRMWARE("amdgpu/navy_flounder_mec2.bin");
+MODULE_FIRMWARE("amdgpu/navy_flounder_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/vangogh_ce.bin");
+MODULE_FIRMWARE("amdgpu/vangogh_pfp.bin");
+MODULE_FIRMWARE("amdgpu/vangogh_me.bin");
+MODULE_FIRMWARE("amdgpu/vangogh_mec.bin");
+MODULE_FIRMWARE("amdgpu/vangogh_mec2.bin");
+MODULE_FIRMWARE("amdgpu/vangogh_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_ce.bin");
+MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_pfp.bin");
+MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_me.bin");
+MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_mec.bin");
+MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_mec2.bin");
+MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/beige_goby_ce.bin");
+MODULE_FIRMWARE("amdgpu/beige_goby_pfp.bin");
+MODULE_FIRMWARE("amdgpu/beige_goby_me.bin");
+MODULE_FIRMWARE("amdgpu/beige_goby_mec.bin");
+MODULE_FIRMWARE("amdgpu/beige_goby_mec2.bin");
+MODULE_FIRMWARE("amdgpu/beige_goby_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/yellow_carp_ce.bin");
+MODULE_FIRMWARE("amdgpu/yellow_carp_pfp.bin");
+MODULE_FIRMWARE("amdgpu/yellow_carp_me.bin");
+MODULE_FIRMWARE("amdgpu/yellow_carp_mec.bin");
+MODULE_FIRMWARE("amdgpu/yellow_carp_mec2.bin");
+MODULE_FIRMWARE("amdgpu/yellow_carp_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/cyan_skillfish2_ce.bin");
+MODULE_FIRMWARE("amdgpu/cyan_skillfish2_pfp.bin");
+MODULE_FIRMWARE("amdgpu/cyan_skillfish2_me.bin");
+MODULE_FIRMWARE("amdgpu/cyan_skillfish2_mec.bin");
+MODULE_FIRMWARE("amdgpu/cyan_skillfish2_mec2.bin");
+MODULE_FIRMWARE("amdgpu/cyan_skillfish2_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/gc_10_3_6_ce.bin");
+MODULE_FIRMWARE("amdgpu/gc_10_3_6_pfp.bin");
+MODULE_FIRMWARE("amdgpu/gc_10_3_6_me.bin");
+MODULE_FIRMWARE("amdgpu/gc_10_3_6_mec.bin");
+MODULE_FIRMWARE("amdgpu/gc_10_3_6_mec2.bin");
+MODULE_FIRMWARE("amdgpu/gc_10_3_6_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/gc_10_3_7_ce.bin");
+MODULE_FIRMWARE("amdgpu/gc_10_3_7_pfp.bin");
+MODULE_FIRMWARE("amdgpu/gc_10_3_7_me.bin");
+MODULE_FIRMWARE("amdgpu/gc_10_3_7_mec.bin");
+MODULE_FIRMWARE("amdgpu/gc_10_3_7_mec2.bin");
+MODULE_FIRMWARE("amdgpu/gc_10_3_7_rlc.bin");
+
+static const struct amdgpu_hwip_reg_entry gc_reg_list_10_1[] = {
+	SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS2),
+	SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS3),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_STALLED_STAT1),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_STALLED_STAT2),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CPC_STALLED_STAT1),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CPF_STALLED_STAT1),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_BUSY_STAT),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CPC_BUSY_STAT),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CPF_BUSY_STAT),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CPC_BUSY_STAT2),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CPF_BUSY_STAT2),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CPF_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_ERROR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_HPD_STATUS0),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB_BASE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB0_BASE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB0_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB0_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB1_BASE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB1_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB1_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB2_BASE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB2_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB2_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB1_CMD_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB2_CMD_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB1_CMD_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB2_CMD_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB1_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB1_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB1_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB2_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB2_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB2_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB1_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB1_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB1_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB2_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB2_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB2_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCPF_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCPC_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCPG_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmGDS_PROTECTION_FAULT),
+	SOC15_REG_ENTRY_STR(GC, 0, mmGDS_VM_PROTECTION_FAULT),
+	SOC15_REG_ENTRY_STR(GC, 0, mmIA_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmIA_UTCL1_STATUS_2),
+	SOC15_REG_ENTRY_STR(GC, 0, mmPA_CL_CNTL_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRMI_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSQC_DCACHE_UTCL0_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSQC_ICACHE_UTCL0_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSQG_UTCL0_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmTCP_UTCL0_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmWD_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmGCVM_L2_PROTECTION_FAULT_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmGCVM_L2_PROTECTION_FAULT_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_DEBUG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MES_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_INSTR_PNTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC1_INSTR_PNTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC2_INSTR_PNTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MES_DEBUG_INTERRUPT_INSTR_PNTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MES_INSTR_PNTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_INSTR_PNTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_INSTR_PNTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CPC_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_STAT),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SMU_COMMAND),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SMU_MESSAGE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SMU_ARGUMENT_1),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SMU_ARGUMENT_2),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SMU_ARGUMENT_3),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SMU_ARGUMENT_4),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSMU_RLC_RESPONSE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SAFE_MODE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SMU_SAFE_MODE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_RLCS_GPM_STAT_2),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SPP_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_RLCS_BOOTLOAD_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_INT_STAT),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_GPM_GENERAL_6),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_GPM_DEBUG_INST_A),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_GPM_DEBUG_INST_B),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_GPM_DEBUG_INST_ADDR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_LX6_CORE_PDEBUG_INST),
+	/* cp header registers */
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MES_HEADER_DUMP),
+	/* SE status registers */
+	SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS_SE0),
+	SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS_SE1),
+	SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS_SE2),
+	SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS_SE3)
+};
+
+static const struct amdgpu_hwip_reg_entry gc_cp_reg_list_10[] = {
+	/* compute registers */
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_VMID),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PERSISTENT_STATE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PIPE_PRIORITY),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_QUEUE_PRIORITY),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_QUANTUM),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PQ_BASE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PQ_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PQ_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PQ_WPTR_POLL_ADDR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PQ_WPTR_POLL_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PQ_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_IB_BASE_ADDR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_IB_BASE_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_IB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_IB_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_DEQUEUE_REQUEST),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_EOP_BASE_ADDR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_EOP_BASE_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_EOP_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_EOP_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_EOP_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_EOP_EVENTS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_CTX_SAVE_BASE_ADDR_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_CTX_SAVE_BASE_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_CTX_SAVE_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_CNTL_STACK_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_CNTL_STACK_SIZE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_WG_STATE_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_CTX_SAVE_SIZE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_GDS_RESOURCE_STATE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_ERROR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_EOP_WPTR_MEM),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PQ_WPTR_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PQ_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_SUSPEND_CNTL_STACK_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_SUSPEND_CNTL_STACK_DW_CNT),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_SUSPEND_WG_STATE_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_DEQUEUE_STATUS),
+	/* cp header registers */
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+};
+
+static const struct amdgpu_hwip_reg_entry gc_gfx_queue_reg_list_10[] = {
+	/* gfx queue registers */
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_HQD_ACTIVE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_HQD_QUEUE_PRIORITY),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_HQD_BASE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_HQD_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_HQD_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_HQD_CSMD_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_HQD_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_HQD_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_HQD_DEQUEUE_REQUEST),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_HQD_MAPPED),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_HQD_QUE_MGR_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_HQD_HQ_CONTROL0),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_HQD_HQ_STATUS0),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_HQD_CE_WPTR_POLL_ADDR_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_HQD_CE_WPTR_POLL_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_HQD_CE_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_HQD_CE_CSMD_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_HQD_CE_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_HQD_CE_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_MQD_BASE_ADDR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_MQD_BASE_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB_WPTR_POLL_ADDR_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB_WPTR_POLL_ADDR_HI),
+	/* gfx header registers */
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+};
+
+static const struct soc15_reg_golden golden_settings_gc_10_1[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL_4, 0xffffffff, 0x00400014),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_CPF_CLK_CTRL, 0xfcff8fff, 0xf8000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_CLK_CTRL, 0xcd000000, 0x0d000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SQ_CLK_CTRL, 0x60000ff0, 0x60000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SQG_CLK_CTRL, 0x40000000, 0x40000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_VGT_CLK_CTRL, 0xffff8fff, 0xffff8100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_WD_CLK_CTRL, 0xfeff8fff, 0xfeff8100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCH_PIPE_STEER, 0xffffffff, 0xe4e4e4e4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCH_VC5_ENABLE, 0x00000002, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCP_SD_CNTL, 0x000007ff, 0x000005ff),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG, 0x20000000, 0x20000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG2, 0xffffffff, 0x00000420),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG3, 0x00000200, 0x00000200),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG4, 0x07900000, 0x04900000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DFSM_TILES_IN_FLIGHT, 0x0000ffff, 0x0000003f),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_LAST_OF_BURST_CONFIG, 0xffffffff, 0x03860204),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCR_GENERAL_CNTL, 0x1ff0ffff, 0x00000500),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGE_PRIV_CONTROL, 0x000007ff, 0x000001fe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL1_PIPE_STEER, 0xffffffff, 0xe4e4e4e4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2_PIPE_STEER_0, 0x77777777, 0x10321032),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2_PIPE_STEER_1, 0x77777777, 0x02310231),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2A_ADDR_MATCH_MASK, 0xffffffff, 0xffffffcf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_ADDR_MATCH_MASK, 0xffffffff, 0xffffffcf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CGTT_SCLK_CTRL, 0x10000000, 0x10000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CTRL2, 0xffffffff, 0x1402002f),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CTRL3, 0xffff9fff, 0x00001188),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_TIMEOUT_COUNTER, 0xffffffff, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE, 0x3fffffff, 0x08000009),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_1, 0x00400000, 0x04440000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_2, 0x00000800, 0x00000820),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRMI_SPARE, 0xffffffff, 0xffff3101),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_CONFIG_CNTL_1, 0x001f0000, 0x00070104),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_ALU_CLK_CTRL, 0xffffffff, 0xffffffff),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_ARB_CONFIG, 0x00000100, 0x00000130),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_LDS_CLK_CTRL, 0xffffffff, 0xffffffff),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfff7ffff, 0x01030000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CNTL, 0x60000010, 0x479c0010),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CGTT_CLK_CTRL, 0xfeff0fff, 0x40000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0x00c00000, 0x00c00000)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_10_0_nv10[] = {
+	/* Pending on emulation bring up */
+};
+
+static const struct soc15_reg_golden golden_settings_gc_rlc_spm_10_0_nv10[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xe0000000, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x28),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x9),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1b),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1b),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x20),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xcc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xd0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xd4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x24),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x24),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x11),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x11),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x88),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x8c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xb0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xb4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xb8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xbc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x90),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x94),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x98),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x9c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xa0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xa4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xa8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xac),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x9),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xb),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x38),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x3c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x40),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x44),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x48),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x5),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x4c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x9),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x70),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x9),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x74),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x9),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x78),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xb),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x7c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xd),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x80),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x84),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xb),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x50),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x7),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x54),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x5),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x58),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x9),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x5c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x60),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xb),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x64),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x9),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x68),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x9),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x6c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xb),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x1c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x3),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x20),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x20),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x24),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x24),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x28),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x28),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x2c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1b),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x2c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1b),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x30),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x30),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x34),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x34),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x38),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x38),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x3c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1a),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x3c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1a),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x50),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x50),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x54),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x54),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x58),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x58),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x5c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x5c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1d),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1d),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x48),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1a),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x48),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1a),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x4c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1a),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x4c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1a),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x40),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x40),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x44),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1b),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x44),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1b),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x17),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x17),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x60),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x60),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x64),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x64),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x70),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x16),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x70),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x16),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x74),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x74),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x68),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x68),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x6c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x6c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x78),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xd),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x78),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xd),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x7c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x7c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x88),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x88),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x8c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x8c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x80),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x80),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x84),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x84),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x90),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x90),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x94),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x94),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xa0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xa0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xa4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x3),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xa4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x3),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x98),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x98),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x9c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x9c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xa8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xa8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xac),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xac),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xb8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x13),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xb8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x13),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xbc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xbc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xb0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xb0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xb4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xb4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xd0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xb),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xd0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xb),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xd4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xd4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xcc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xb),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xcc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xb),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xe8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xe8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xec),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xec),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xf0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xf0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xf4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xf4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xf8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xf8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xfc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xfc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x104),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x104),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xe0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x13),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xe0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x13),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x118),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x13),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x118),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x13),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x11c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x13),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x11c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x13),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x120),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x13),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x120),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x13),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x124),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x13),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x124),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x13),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xdc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1b),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xdc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1b),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x110),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x110),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x114),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x114),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x108),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x108),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x10c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x10c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xd8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x17),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xd8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x17),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x128),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x128),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x12c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x12c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x138),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x138),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x13c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x13c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x130),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x130),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x134),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x134),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x140),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x140),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x144),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x7),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x144),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x7),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x150),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x150),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x154),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x7),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x154),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x7),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x148),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x148),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x14c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x7),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x14c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x7),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x158),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x5),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x158),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x5),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x15c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x2),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x15c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x2),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x168),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x168),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x16c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x16c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x160),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x160),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x164),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x2),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x164),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x2),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x170),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x11),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x170),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x11),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x174),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xd),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x174),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xd),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x180),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x180),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x184),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x184),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x178),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x11),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x178),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x11),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x17c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xd),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x17c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xd),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x188),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x188),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x18c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x18c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x198),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x9),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x198),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x9),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x19c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x3),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x19c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x3),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x190),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x190),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x194),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x194),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x30),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x11),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x34),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1d),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1d),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1f),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1f),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x2c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xb),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLOBALS_SAMPLE_SKEW, 0x000000FF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLOBALS_MUXSEL_SKEW, 0x000000FF, 0x20),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLE_SKEW, 0x000000FF, 0x5),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLE_SKEW, 0x000000FF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_MUXSEL_SKEW, 0x000000FF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_MUXSEL_SKEW, 0x000000FF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_DESER_START_SKEW, 0x000000FF, 0x33),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xFFFFFFFF, 0xe0000000)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_10_1_1[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL_4, 0xffffffff, 0x003c0014),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_GS_NGG_CLK_CTRL, 0xffff8fff, 0xffff8100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_IA_CLK_CTRL, 0xffff0fff, 0xffff0100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_CLK_CTRL, 0xcd000000, 0x0d000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SQ_CLK_CTRL, 0xf8ff0fff, 0x60000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SQG_CLK_CTRL, 0x40000ff0, 0x40000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_VGT_CLK_CTRL, 0xffff8fff, 0xffff8100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_WD_CLK_CTRL, 0xffff8fff, 0xffff8100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCH_PIPE_STEER, 0xffffffff, 0xe4e4e4e4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCH_VC5_ENABLE, 0x00000002, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCP_SD_CNTL, 0x800007ff, 0x000005ff),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG, 0xffffffff, 0x20000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG2, 0xffffffff, 0x00000420),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG3, 0x00000200, 0x00000200),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG4, 0xffffffff, 0x04900000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DFSM_TILES_IN_FLIGHT, 0x0000ffff, 0x0000003f),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_LAST_OF_BURST_CONFIG, 0xffffffff, 0x03860204),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCR_GENERAL_CNTL, 0x1ff0ffff, 0x00000500),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGE_PRIV_CONTROL, 0x000007ff, 0x000001fe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL1_PIPE_STEER, 0xffffffff, 0xe4e4e4e4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2A_ADDR_MATCH_MASK, 0xffffffff, 0xffffffe7),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_ADDR_MATCH_MASK, 0xffffffff, 0xffffffe7),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CGTT_SCLK_CTRL, 0xffff0fff, 0x10000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CTRL2, 0xffffffff, 0x1402002f),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CTRL3, 0xffffbfff, 0x00000188),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_TIMEOUT_COUNTER, 0xffffffff, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE, 0x3fffffff, 0x08000009),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_1, 0x00400000, 0x04440000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_2, 0x00000800, 0x00000820),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRMI_SPARE, 0xffffffff, 0xffff3101),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_CONFIG_CNTL_1, 0x001f0000, 0x00070105),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_ALU_CLK_CTRL, 0xffffffff, 0xffffffff),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_ARB_CONFIG, 0x00000133, 0x00000130),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_LDS_CLK_CTRL, 0xffffffff, 0xffffffff),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfff7ffff, 0x01030000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CNTL, 0x60000010, 0x479c0010),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0x00c00000, 0x00c00000),
+};
+
+static const struct soc15_reg_golden golden_settings_gc_10_1_2[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL_4, 0x003e001f, 0x003c0014),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_GS_NGG_CLK_CTRL, 0xffff8fff, 0xffff8100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_IA_CLK_CTRL, 0xffff0fff, 0xffff0100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_CLK_CTRL, 0xff7f0fff, 0x0d000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SQ_CLK_CTRL, 0xffffcfff, 0x60000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SQG_CLK_CTRL, 0xffff0fff, 0x40000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_VGT_CLK_CTRL, 0xffff8fff, 0xffff8100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_WD_CLK_CTRL, 0xffff8fff, 0xffff8100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCH_PIPE_STEER, 0xffffffff, 0xe4e4e4e4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCH_VC5_ENABLE, 0x00000003, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCP_SD_CNTL, 0x800007ff, 0x000005ff),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG, 0xffffffff, 0x20000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG2, 0xffffffff, 0x00000420),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG3, 0xffffffff, 0x00000200),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG4, 0xffffffff, 0x04900000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DFSM_TILES_IN_FLIGHT, 0x0000ffff, 0x0000003f),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_LAST_OF_BURST_CONFIG, 0xffffffff, 0x03860204),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG, 0x0c1800ff, 0x00000044),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCR_GENERAL_CNTL, 0x1ff0ffff, 0x00000500),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGE_PRIV_CONTROL, 0x00007fff, 0x000001fe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL1_PIPE_STEER, 0xffffffff, 0xe4e4e4e4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2_PIPE_STEER_0, 0x77777777, 0x10321032),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2_PIPE_STEER_1, 0x77777777, 0x02310231),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2A_ADDR_MATCH_MASK, 0xffffffff, 0xffffffcf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_ADDR_MATCH_MASK, 0xffffffff, 0xffffffcf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CGTT_SCLK_CTRL, 0xffff0fff, 0x10000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CTRL2, 0xffffffff, 0x1402002f),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CTRL3, 0xffffbfff, 0x00000188),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_EVENT_CNTL_0, 0xffffffff, 0x842a4c02),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_TIMEOUT_COUNTER, 0xffffffff, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE, 0x3fffffff, 0x08000009),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_1, 0xffffffff, 0x04440000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_2, 0x00000820, 0x00000820),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRMI_SPARE, 0xffffffff, 0xffff3101),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_CONFIG_CNTL_1, 0x001f0000, 0x00070104),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_ALU_CLK_CTRL, 0xffffffff, 0xffffffff),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_ARB_CONFIG, 0x00000133, 0x00000130),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_LDS_CLK_CTRL, 0xffffffff, 0xffffffff),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfff7ffff, 0x01030000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CNTL, 0xffdf80ff, 0x479c0010),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0xffffffff, 0x00c00000)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_10_1_nv14[] = {
+	/* Pending on emulation bring up */
+};
+
+static const struct soc15_reg_golden golden_settings_gc_rlc_spm_10_1_nv14[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xE0000000L, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x28),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x20),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x80),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x84),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x88),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x8c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x24),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x24),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x20),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x60),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x64),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x2),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x68),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x2),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x6c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x2),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x70),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x2),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x74),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x2),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x78),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x7c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x2),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x38),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x3c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x40),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x3),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x44),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x9),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x48),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x7),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x4c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x3),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x50),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x3),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x54),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x3),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x58),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x7),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x5c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x9),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x1c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x2),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x20),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x24),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x16),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x28),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x2c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1a),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x30),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x16),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x34),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x38),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1a),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x3c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x50),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1e),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x54),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1e),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x58),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1e),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x5c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1e),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x20),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x48),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1a),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x4c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1a),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x40),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1e),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x44),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1e),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1a),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x60),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x16),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x64),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x70),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x16),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x74),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x16),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x68),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x16),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x6c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x78),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x7c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x88),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x8c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x80),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x84),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x90),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x94),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xa0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xa4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x98),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x9c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xa8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xac),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xb8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xbc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xb0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xb4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x16),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xd0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x16),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xd4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x16),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xcc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xd8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xdc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xe8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xec),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xe0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xe4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x104),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x108),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x10c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x110),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x114),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x118),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x11c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x16),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xf8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x17),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x130),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x134),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x138),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x13c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xf4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x20),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x128),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x12c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x120),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x124),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xf0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1a),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x140),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x144),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x150),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x154),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x148),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x14c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x158),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x15c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x168),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x16c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x160),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x164),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x170),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x174),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x180),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x184),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x178),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x17c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x188),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x16),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x18c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x198),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x19c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x190),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x16),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x194),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x1a0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x1a4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x1b0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x1b4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x1a8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x1ac),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x1b8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x1bc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x1c8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x1cc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x1c0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x1c4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x30),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x34),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x20),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x20),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x2c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLOBALS_SAMPLE_SKEW, 0x000000FF, 0x26),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLOBALS_MUXSEL_SKEW, 0x000000FF, 0x28),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLE_SKEW, 0x000000FF, 0xf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLE_SKEW, 0x000000FF, 0x15),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_MUXSEL_SKEW, 0x000000FF, 0x1f),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_MUXSEL_SKEW, 0x000000FF, 0x25),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_DESER_START_SKEW, 0x000000FF, 0x3b),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xFFFFFFFF, 0xe0000000)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_10_1_2_nv12[] = {
+	/* Pending on emulation bring up */
+};
+
+static const struct soc15_reg_golden golden_settings_gc_rlc_spm_10_1_2_nv12[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xe0000000L, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x28),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x5),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1b),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1b),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x20),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xd),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xcc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xd),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xd0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xd),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xd4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xd),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x24),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xd),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x24),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xd),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x11),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x11),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x88),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x8c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x2),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xb0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xb4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xb8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xbc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x7),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x90),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x2),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x94),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x98),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x5),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x9c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xa0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xa4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xa8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xac),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x5),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xd),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x9),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x38),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x7),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x3c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x7),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x40),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xd),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x44),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x48),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x4c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x7),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x70),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x74),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x7),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x78),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x7c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x80),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xb),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x84),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x50),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x54),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x3),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x58),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x5c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x7),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x60),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x64),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x68),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x7),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x6c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x1c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x3),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x11),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x20),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x20),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x24),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x24),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x28),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x28),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x2c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1a),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x2c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1a),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x30),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1d),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x30),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1d),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x34),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1a),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x34),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1a),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x38),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x16),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x38),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x16),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x3c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x3c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x18),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x50),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x50),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x54),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x54),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x58),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x58),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x5c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x5c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1a),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1a),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x48),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x48),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x4c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x4c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x40),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1a),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x40),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1a),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x44),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1a),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x44),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1a),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1b),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1b),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x60),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x15),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x60),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x15),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x64),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x13),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x64),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x13),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x70),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x16),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x70),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x16),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x74),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x74),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x68),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x15),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x68),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x15),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x6c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x13),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x6c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x13),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x78),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x78),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x7c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xb),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x7c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xb),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x88),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x88),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x8c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x8c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x80),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x80),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x84),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xb),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x84),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xb),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x90),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x9),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x90),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x9),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x94),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x94),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xa0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xa0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xa4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xd),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xa4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xd),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x98),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x9),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x98),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x9),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x9c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x9c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xa8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xa8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xac),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xac),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xb8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xb8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xbc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xbc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xb0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xb0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xb4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xb4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x11),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x11),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xb),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xb),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xd0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xd0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xd4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xd4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x11),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xc8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x11),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xcc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xb),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xcc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xb),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xe8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xe8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xec),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x16),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xec),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x16),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xf0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x15),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xf0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x15),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xf4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xf4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xf8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xf8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xfc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x17),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xfc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x17),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x13),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x13),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x104),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x11),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x104),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x11),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xe0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xe0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x118),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x15),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x118),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x15),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x11c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x15),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x11c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x15),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x120),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x15),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x120),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x15),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x124),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x15),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x124),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x15),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xdc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xdc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x110),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x15),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x110),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x15),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x114),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x114),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x14),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x108),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x108),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x10c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x10c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x19),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xd8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1b),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0xd8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1b),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x128),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x128),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x12c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x12c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x138),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x138),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x13c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x13c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x130),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x130),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x12),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x134),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x134),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x140),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x140),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x144),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x7),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x144),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x7),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x150),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x150),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x154),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xd),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x154),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xd),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x148),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x148),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x14c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x7),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x14c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x7),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x158),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x5),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x158),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x5),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x15c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x15c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x168),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x168),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xa),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x16c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x9),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x16c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x9),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x160),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x5),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x160),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x5),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x164),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x164),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x170),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x170),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x174),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x174),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x180),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x180),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x184),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x184),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x178),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x178),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x17c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x17c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x188),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x188),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x18c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x5),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x18c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x5),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x198),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x198),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xc),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x19c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xb),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x19c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xb),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x190),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x190),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x194),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x194),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x30),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xd),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x34),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x11),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1d),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1d),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1f),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0x1f),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_ADDR, 0xFFFFFFFF, 0x2c),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLB_SAMPLEDELAY_IND_DATA, 0xFFFFFFFF, 0xb),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLOBALS_SAMPLE_SKEW, 0x000000FF, 0x1f),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_GLOBALS_MUXSEL_SKEW, 0x000000FF, 0x22),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLE_SKEW, 0x000000FF, 0x1),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_SAMPLE_SKEW, 0x000000FF, 0x6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_MUXSEL_SKEW, 0x000000FF, 0x10),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x10000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_SE_MUXSEL_SKEW, 0x000000FF, 0x15),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffff, 0x0),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_DESER_START_SKEW, 0x000000FF, 0x35),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xFFFFFFFF, 0xe0000000)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_10_3[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_CS_CLK_CTRL, 0x78000000, 0x78000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_PS_CLK_CTRL, 0xff7f0fff, 0x78000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_RA0_CLK_CTRL, 0xff7f0fff, 0x30000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_RA1_CLK_CTRL, 0xff7f0fff, 0x7e000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPF_GCR_CNTL, 0x0007ffff, 0x0000c000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG3, 0xffffffff, 0x00000280),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG4, 0xffffffff, 0x00800000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_EXCEPTION_CONTROL, 0x7fff0f1f, 0x00b80000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCEA_SDP_TAG_RESERVE0, 0xffffffff, 0x10100100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCEA_SDP_TAG_RESERVE1, 0xffffffff, 0x17000088),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCR_GENERAL_CNTL_Sienna_Cichlid, 0x1ff1ffff, 0x00000500),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCUTCL2_CGTT_CLK_CTRL_Sienna_Cichlid, 0xff000000, 0xff008080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCVM_L2_CGTT_CLK_CTRL_Sienna_Cichlid, 0xff000000, 0xff008080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGE_PC_CNTL, 0x003fffff, 0x00280400),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2A_ADDR_MATCH_MASK, 0xffffffff, 0xffffffcf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_ADDR_MATCH_MASK, 0xffffffff, 0xffffffcf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CM_CTRL1, 0xff8fff0f, 0x580f1008),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CTRL3, 0xf7ffffff, 0x10f80988),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmLDS_CONFIG,  0x00000020, 0x00000020),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_CL_ENHANCE, 0xf17fffff, 0x01200007),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_TIMEOUT_COUNTER, 0xffffffff, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_2, 0xffffffbf, 0x00000820),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_CONFIG_CNTL_1, 0xffffffff, 0x00070104),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_CONFIG, 0xe07df47f, 0x00180070),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER0_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER1_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER10_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER11_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER12_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER13_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER14_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER15_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER2_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER3_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER4_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER5_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER6_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER7_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER8_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER9_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSX_DEBUG_1, 0x00010000, 0x00010020),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfff7ffff, 0x01030000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0xffbfffff, 0x00a00000)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_10_3_sienna_cichlid[] = {
+	/* Pending on emulation bring up */
+};
+
+static const struct soc15_reg_golden golden_settings_gc_10_3_2[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_CS_CLK_CTRL, 0xff7f0fff, 0x78000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_PS_CLK_CTRL, 0xff7f0fff, 0x78000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_RA0_CLK_CTRL, 0xff7f0fff, 0x30000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_RA1_CLK_CTRL, 0xff7f0fff, 0x7e000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPF_GCR_CNTL, 0x0007ffff, 0x0000c000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG3, 0xffffffff, 0x00000280),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG4, 0xffffffff, 0x00800000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_EXCEPTION_CONTROL, 0x7fff0f1f, 0x00b80000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCR_GENERAL_CNTL_Sienna_Cichlid, 0x1ff1ffff, 0x00000500),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCUTCL2_CGTT_CLK_CTRL_Sienna_Cichlid, 0xffffffff, 0xff008080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCVM_L2_CGTT_CLK_CTRL_Sienna_Cichlid, 0xffff8fff, 0xff008080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGE_PC_CNTL, 0x003fffff, 0x00280400),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2A_ADDR_MATCH_MASK, 0xffffffff, 0xffffffcf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_ADDR_MATCH_MASK, 0xffffffff, 0xffffffcf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CM_CTRL1, 0xff8fff0f, 0x580f1008),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CTRL3, 0xf7ffffff, 0x00f80988),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_CL_ENHANCE, 0xf17fffff, 0x01200007),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_TIMEOUT_COUNTER, 0xffffffff, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_2, 0xffffffbf, 0x00000820),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_CONFIG_CNTL_1, 0xffffffff, 0x00070104),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_START_PHASE, 0x000000ff, 0x00000004),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_CONFIG, 0xe07df47f, 0x00180070),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER0_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER1_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER10_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER11_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER12_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER13_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER14_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER15_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER2_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER3_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER4_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER5_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER6_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER7_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER8_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER9_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfff7ffff, 0x01030000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0xffbfffff, 0x00a00000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_GS_MAX_WAVE_ID, 0x00000fff, 0x000003ff),
+
+	/* This is not in GDB yet. Don't remove it. It fixes a GPU hang on Navy Flounder. */
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmLDS_CONFIG,  0x00000020, 0x00000020),
+};
+
+static const struct soc15_reg_golden golden_settings_gc_10_3_vangogh[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_RA0_CLK_CTRL, 0xff7f0fff, 0x30000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_RA1_CLK_CTRL, 0xff7f0fff, 0x7e000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCH_PIPE_STEER, 0x000000ff, 0x000000e4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG3, 0xffffffff, 0x00000200),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG4, 0xffffffff, 0x00800000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_EXCEPTION_CONTROL, 0x7fff0f1f, 0x00b80000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG, 0x0c1807ff, 0x00000142),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCR_GENERAL_CNTL_Vangogh, 0x1ff1ffff, 0x00000500),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL1_PIPE_STEER, 0x000000ff, 0x000000e4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2_PIPE_STEER_0, 0x77777777, 0x32103210),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2_PIPE_STEER_1, 0x77777777, 0x32103210),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2A_ADDR_MATCH_MASK, 0xffffffff, 0xfffffff3),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_ADDR_MATCH_MASK, 0xffffffff, 0xfffffff3),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CM_CTRL1, 0xff8fff0f, 0x580f1008),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CTRL3, 0xf7ffffff, 0x00f80988),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_CL_ENHANCE, 0xf17fffff, 0x01200007),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_TIMEOUT_COUNTER, 0xffffffff, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_2, 0xffffffbf, 0x00000020),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_CONFIG_CNTL_1_Vangogh, 0xffffffff, 0x00070103),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQG_CONFIG, 0x000017ff, 0x00001000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSX_DEBUG_1, 0x00010000, 0x00010020),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfff7ffff, 0x01030000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0xffffffff, 0x00400000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_GS_MAX_WAVE_ID, 0x00000fff, 0x000000ff),
+
+	/* This is not in GDB yet. Don't remove it. It fixes a GPU hang on VanGogh. */
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmLDS_CONFIG,  0x00000020, 0x00000020),
+};
+
+static const struct soc15_reg_golden golden_settings_gc_10_3_3[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_CS_CLK_CTRL, 0xff7f0fff, 0x78000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCH_PIPE_STEER, 0x000000ff, 0x000000e4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPF_GCR_CNTL, 0x0007ffff, 0x0000c200),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG3, 0xffffffff, 0x00000280),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG4, 0xffffffff, 0x00800000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG, 0x0c1807ff, 0x00000242),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCR_GENERAL_CNTL_Vangogh, 0x1ff1ffff, 0x00000500),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL1_PIPE_STEER, 0x000000ff, 0x000000e4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2_PIPE_STEER_0, 0x77777777, 0x32103210),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2_PIPE_STEER_1, 0x77777777, 0x32103210),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2A_ADDR_MATCH_MASK, 0xffffffff, 0xfffffff3),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_ADDR_MATCH_MASK, 0xffffffff, 0xfffffff3),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CM_CTRL1, 0xff8fff0f, 0x580f1008),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CTRL3, 0xf7ffffff, 0x00f80988),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmLDS_CONFIG, 0x000001ff, 0x00000020),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_CL_ENHANCE, 0xf17fffff, 0x01200007),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_TIMEOUT_COUNTER, 0xffffffff, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_2, 0xffffffbf, 0x00000820),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfff7ffff, 0x01030000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0xffffffff, 0x00100000)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_10_3_4[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_CS_CLK_CTRL, 0x78000000, 0x78000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_RA0_CLK_CTRL, 0x30000000, 0x30000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_RA1_CLK_CTRL, 0x7e000000, 0x7e000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPF_GCR_CNTL, 0x0007ffff, 0x0000c000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG3, 0x00000280, 0x00000280),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG4, 0x07800000, 0x00800000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCR_GENERAL_CNTL_Sienna_Cichlid, 0x00001d00, 0x00000500),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGE_PC_CNTL, 0x003c0000, 0x00280400),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2A_ADDR_MATCH_MASK, 0xffffffff, 0xffffffcf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_ADDR_MATCH_MASK, 0xffffffff, 0xffffffcf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CM_CTRL1, 0x40000000, 0x580f1008),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CTRL3, 0x00040000, 0x00f80988),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_CL_ENHANCE, 0x01000000, 0x01200007),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_TIMEOUT_COUNTER, 0xffffffff, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_2, 0x00000800, 0x00000820),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_CONFIG, 0x0000001f, 0x00180070),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER0_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER1_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER10_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER11_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER12_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER13_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER14_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER15_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER2_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER3_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER4_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER5_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER6_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER7_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER8_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER9_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSX_DEBUG_1, 0x00010000, 0x00010020),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0x01030000, 0x01030000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0x03a00000, 0x00a00000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmLDS_CONFIG,  0x00000020, 0x00000020)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_10_3_5[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_CS_CLK_CTRL, 0x78000000, 0x78000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_RA0_CLK_CTRL, 0xb0000ff0, 0x30000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_RA1_CLK_CTRL, 0xff000000, 0x7e000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPF_GCR_CNTL, 0x0007ffff, 0x0000c000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG3, 0xffffffff, 0x00000280),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG4, 0xffffffff, 0x00800000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCR_GENERAL_CNTL_Sienna_Cichlid, 0x1ff1ffff, 0x00000500),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2A_ADDR_MATCH_MASK, 0xffffffff, 0xffffffcf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_ADDR_MATCH_MASK, 0xffffffff, 0xffffffcf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CM_CTRL1, 0xff8fff0f, 0x580f1008),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CTRL3, 0xf7ffffff, 0x00f80988),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmLDS_CONFIG, 0x000001ff, 0x00000020),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_CL_ENHANCE, 0xf17fffff, 0x01200007),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_CONFIG, 0xe07df47f, 0x00180070),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER0_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER1_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER10_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER11_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER12_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER13_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER14_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER15_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER2_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER3_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER4_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER5_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER6_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER7_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER8_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER9_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfff7ffff, 0x01030000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0xffbfffff, 0x00a00000)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_10_0_cyan_skillfish[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGE_FAST_CLKS, 0x3fffffff, 0x0000493e),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_CPF_CLK_CTRL, 0xfcff8fff, 0xf8000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_CLK_CTRL, 0xff7f0fff, 0x3c000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL_3, 0xa0000000, 0xa0000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL_4, 0x00008000, 0x003c8014),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCH_DRAM_BURST_CTRL, 0x00000010, 0x00000017),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCH_PIPE_STEER, 0xffffffff, 0xd8d8d8d8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCH_VC5_ENABLE, 0x00000003, 0x00000003),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCP_SD_CNTL, 0x800007ff, 0x000005ff),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG, 0xffffffff, 0x20000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG3, 0xffffffff, 0x00000200),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG4, 0xffffffff, 0x04800000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_LAST_OF_BURST_CONFIG, 0xffffffff, 0x03860210),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG, 0x0c1800ff, 0x00000044),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCR_GENERAL_CNTL, 0x00009d00, 0x00008500),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCMC_VM_CACHEABLE_DRAM_ADDRESS_END, 0xffffffff, 0x000fffff),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL1_DRAM_BURST_CTRL, 0x00000010, 0x00000017),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL1_PIPE_STEER, 0xfcfcfcfc, 0xd8d8d8d8),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2_PIPE_STEER_0, 0x77707770, 0x21302130),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2_PIPE_STEER_1, 0x77707770, 0x21302130),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2A_ADDR_MATCH_MASK, 0xffffffff, 0xffffffcf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_ADDR_MATCH_MASK, 0xffffffff, 0xffffffcf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CGTT_SCLK_CTRL, 0x10000000, 0x10000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CTRL2, 0xfc02002f, 0x9402002f),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CTRL3, 0x00002188, 0x00000188),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE, 0x08000009, 0x08000009),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_EVENT_CNTL_0, 0xcc3fcc03, 0x842a4c02),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_LINE_STIPPLE_STATE, 0x0000000f, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRMI_SPARE, 0xffff3109, 0xffff3101),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_ARB_CONFIG, 0x00000100, 0x00000130),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_LDS_CLK_CTRL, 0xffffffff, 0xffffffff),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0x00030008, 0x01030000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0x00800000, 0x00800000)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_10_3_6[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_CS_CLK_CTRL, 0xff7f0fff, 0x78000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCH_PIPE_STEER, 0x000000ff, 0x00000044),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPF_GCR_CNTL, 0x0007ffff, 0x0000c200),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG3, 0xffffffff, 0x00000280),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG4, 0xffffffff, 0x00800000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG, 0x0c1807ff, 0x00000042),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCR_GENERAL_CNTL_Vangogh, 0x1ff1ffff, 0x00000500),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL1_PIPE_STEER, 0x000000ff, 0x00000044),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2_PIPE_STEER_0, 0x77777777, 0x32103210),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2_PIPE_STEER_1, 0x77777777, 0x32103210),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2A_ADDR_MATCH_MASK, 0xffffffff, 0xfffffff3),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_ADDR_MATCH_MASK, 0xffffffff, 0xfffffff3),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CM_CTRL1, 0xff8fff0f, 0x580f1008),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CTRL3, 0xf7ffffff, 0x00f80988),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmLDS_CONFIG, 0x000001ff, 0x00000020),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_CL_ENHANCE, 0xf17fffff, 0x01200007),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_TIMEOUT_COUNTER, 0xffffffff, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_2, 0xffffffbf, 0x00000820),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQG_CONFIG, 0x000017ff, 0x00001000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSX_DEBUG_1, 0xffffff7f, 0x00010020),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfff7ffff, 0x01030000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0xffffffff, 0x00100000)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_10_3_7[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_CS_CLK_CTRL, 0xff7f0fff, 0x78000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCH_PIPE_STEER, 0x000000ff, 0x000000e4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPF_GCR_CNTL, 0x0007ffff, 0x0000c200),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG3, 0xffffffff, 0x00000280),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG4, 0xffffffff, 0x00800000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG, 0x0c1807ff, 0x00000041),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCR_GENERAL_CNTL_Vangogh, 0x1ff1ffff, 0x00000500),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL1_PIPE_STEER, 0x000000ff, 0x000000e4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2_PIPE_STEER_0, 0x77777777, 0x32103210),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2_PIPE_STEER_1, 0x77777777, 0x32103210),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2A_ADDR_MATCH_MASK, 0xffffffff, 0xffffffff),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_ADDR_MATCH_MASK, 0xffffffff, 0xffffffff),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CM_CTRL1, 0xff8fff0f, 0x580f1008),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CTRL3, 0xf7ffffff, 0x00f80988),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmLDS_CONFIG, 0x000001ff, 0x00000020),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_CL_ENHANCE, 0xf000003f, 0x01200007),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_TIMEOUT_COUNTER, 0xffffffff, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_2, 0xffffffbf, 0x00000820),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQG_CONFIG, 0x000017ff, 0x00001000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSX_DEBUG_1, 0xffffff7f, 0x00010020),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfff7ffff, 0x01030000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0xffffffff, 0x00100000)
+};
+
+#define DEFAULT_SH_MEM_CONFIG \
+	((SH_MEM_ADDRESS_MODE_64 << SH_MEM_CONFIG__ADDRESS_MODE__SHIFT) | \
+	 (SH_MEM_ALIGNMENT_MODE_UNALIGNED << SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) | \
+	 (SH_MEM_RETRY_MODE_ALL << SH_MEM_CONFIG__RETRY_MODE__SHIFT) | \
+	 (3 << SH_MEM_CONFIG__INITIAL_INST_PREFETCH__SHIFT))
+
+/* TODO: pending on golden setting value of gb address config */
+#define CYAN_SKILLFISH_GB_ADDR_CONFIG_GOLDEN 0x00100044
+
+static void gfx_v10_0_set_ring_funcs(struct amdgpu_device *adev);
+static void gfx_v10_0_set_irq_funcs(struct amdgpu_device *adev);
+static void gfx_v10_0_set_gds_init(struct amdgpu_device *adev);
+static void gfx_v10_0_set_rlc_funcs(struct amdgpu_device *adev);
+static void gfx_v10_0_set_mqd_funcs(struct amdgpu_device *adev);
+static int gfx_v10_0_get_cu_info(struct amdgpu_device *adev,
+				 struct amdgpu_cu_info *cu_info);
+static uint64_t gfx_v10_0_get_gpu_clock_counter(struct amdgpu_device *adev);
+static void gfx_v10_0_select_se_sh(struct amdgpu_device *adev, u32 se_num,
+				   u32 sh_num, u32 instance, int xcc_id);
+static u32 gfx_v10_0_get_wgp_active_bitmap_per_sh(struct amdgpu_device *adev);
+
+static int gfx_v10_0_rlc_backdoor_autoload_buffer_init(struct amdgpu_device *adev);
+static void gfx_v10_0_rlc_backdoor_autoload_buffer_fini(struct amdgpu_device *adev);
+static int gfx_v10_0_rlc_backdoor_autoload_enable(struct amdgpu_device *adev);
+static int gfx_v10_0_wait_for_rlc_autoload_complete(struct amdgpu_device *adev);
+static void gfx_v10_0_ring_emit_ce_meta(struct amdgpu_ring *ring, bool resume);
+static void gfx_v10_0_ring_emit_de_meta(struct amdgpu_ring *ring, bool resume);
+static void gfx_v10_0_ring_emit_frame_cntl(struct amdgpu_ring *ring, bool start, bool secure);
+static u32 gfx_v10_3_get_disabled_sa(struct amdgpu_device *adev);
+static void gfx_v10_3_program_pbb_mode(struct amdgpu_device *adev);
+static void gfx_v10_3_set_power_brake_sequence(struct amdgpu_device *adev);
+static void gfx_v10_0_ring_invalidate_tlbs(struct amdgpu_ring *ring,
+					   uint16_t pasid, uint32_t flush_type,
+					   bool all_hub, uint8_t dst_sel);
+static void gfx_v10_0_update_spm_vmid_internal(struct amdgpu_device *adev,
+					       unsigned int vmid);
+
+static int gfx_v10_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_powergating_state state);
+static void gfx10_kiq_set_resources(struct amdgpu_ring *kiq_ring, uint64_t queue_mask)
+{
+	struct amdgpu_device *adev = kiq_ring->adev;
+	u64 shader_mc_addr;
+
+	/* Cleaner shader MC address */
+	shader_mc_addr = adev->gfx.cleaner_shader_gpu_addr >> 8;
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_SET_RESOURCES, 6));
+	amdgpu_ring_write(kiq_ring, PACKET3_SET_RESOURCES_VMID_MASK(0) |
+			  PACKET3_SET_RESOURCES_QUEUE_TYPE(0));	/* vmid_mask:0 queue_type:0 (KIQ) */
+	amdgpu_ring_write(kiq_ring, lower_32_bits(queue_mask));	/* queue mask lo */
+	amdgpu_ring_write(kiq_ring, upper_32_bits(queue_mask));	/* queue mask hi */
+	amdgpu_ring_write(kiq_ring, lower_32_bits(shader_mc_addr)); /* cleaner shader addr lo */
+	amdgpu_ring_write(kiq_ring, upper_32_bits(shader_mc_addr)); /* cleaner shader addr hi */
+	amdgpu_ring_write(kiq_ring, 0);	/* oac mask */
+	amdgpu_ring_write(kiq_ring, 0);	/* gds heap base:0, gds heap size:0 */
+}
+
+static void gfx10_kiq_map_queues(struct amdgpu_ring *kiq_ring,
+				 struct amdgpu_ring *ring)
+{
+	uint64_t mqd_addr = amdgpu_bo_gpu_offset(ring->mqd_obj);
+	uint64_t wptr_addr = ring->wptr_gpu_addr;
+	uint32_t eng_sel = 0;
+
+	switch (ring->funcs->type) {
+	case AMDGPU_RING_TYPE_COMPUTE:
+		eng_sel = 0;
+		break;
+	case AMDGPU_RING_TYPE_GFX:
+		eng_sel = 4;
+		break;
+	case AMDGPU_RING_TYPE_MES:
+		eng_sel = 5;
+		break;
+	default:
+		WARN_ON(1);
+	}
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_MAP_QUEUES, 5));
+	/* Q_sel:0, vmid:0, vidmem: 1, engine:0, num_Q:1*/
+	amdgpu_ring_write(kiq_ring, /* Q_sel: 0, vmid: 0, engine: 0, num_Q: 1 */
+			  PACKET3_MAP_QUEUES_QUEUE_SEL(0) | /* Queue_Sel */
+			  PACKET3_MAP_QUEUES_VMID(0) | /* VMID */
+			  PACKET3_MAP_QUEUES_QUEUE(ring->queue) |
+			  PACKET3_MAP_QUEUES_PIPE(ring->pipe) |
+			  PACKET3_MAP_QUEUES_ME((ring->me == 1 ? 0 : 1)) |
+			  PACKET3_MAP_QUEUES_QUEUE_TYPE(0) | /*queue_type: normal compute queue */
+			  PACKET3_MAP_QUEUES_ALLOC_FORMAT(0) | /* alloc format: all_on_one_pipe */
+			  PACKET3_MAP_QUEUES_ENGINE_SEL(eng_sel) |
+			  PACKET3_MAP_QUEUES_NUM_QUEUES(1)); /* num_queues: must be 1 */
+	amdgpu_ring_write(kiq_ring, PACKET3_MAP_QUEUES_DOORBELL_OFFSET(ring->doorbell_index));
+	amdgpu_ring_write(kiq_ring, lower_32_bits(mqd_addr));
+	amdgpu_ring_write(kiq_ring, upper_32_bits(mqd_addr));
+	amdgpu_ring_write(kiq_ring, lower_32_bits(wptr_addr));
+	amdgpu_ring_write(kiq_ring, upper_32_bits(wptr_addr));
+}
+
+static void gfx10_kiq_unmap_queues(struct amdgpu_ring *kiq_ring,
+				   struct amdgpu_ring *ring,
+				   enum amdgpu_unmap_queues_action action,
+				   u64 gpu_addr, u64 seq)
+{
+	uint32_t eng_sel = ring->funcs->type == AMDGPU_RING_TYPE_GFX ? 4 : 0;
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_UNMAP_QUEUES, 4));
+	amdgpu_ring_write(kiq_ring, /* Q_sel: 0, vmid: 0, engine: 0, num_Q: 1 */
+			  PACKET3_UNMAP_QUEUES_ACTION(action) |
+			  PACKET3_UNMAP_QUEUES_QUEUE_SEL(0) |
+			  PACKET3_UNMAP_QUEUES_ENGINE_SEL(eng_sel) |
+			  PACKET3_UNMAP_QUEUES_NUM_QUEUES(1));
+	amdgpu_ring_write(kiq_ring,
+		  PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET0(ring->doorbell_index));
+
+	if (action == PREEMPT_QUEUES_NO_UNMAP) {
+		amdgpu_ring_write(kiq_ring, lower_32_bits(gpu_addr));
+		amdgpu_ring_write(kiq_ring, upper_32_bits(gpu_addr));
+		amdgpu_ring_write(kiq_ring, seq);
+	} else {
+		amdgpu_ring_write(kiq_ring, 0);
+		amdgpu_ring_write(kiq_ring, 0);
+		amdgpu_ring_write(kiq_ring, 0);
+	}
+}
+
+static void gfx10_kiq_query_status(struct amdgpu_ring *kiq_ring,
+				   struct amdgpu_ring *ring,
+				   u64 addr,
+				   u64 seq)
+{
+	uint32_t eng_sel = ring->funcs->type == AMDGPU_RING_TYPE_GFX ? 4 : 0;
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_QUERY_STATUS, 5));
+	amdgpu_ring_write(kiq_ring,
+			  PACKET3_QUERY_STATUS_CONTEXT_ID(0) |
+			  PACKET3_QUERY_STATUS_INTERRUPT_SEL(0) |
+			  PACKET3_QUERY_STATUS_COMMAND(2));
+	amdgpu_ring_write(kiq_ring, /* Q_sel: 0, vmid: 0, engine: 0, num_Q: 1 */
+			  PACKET3_QUERY_STATUS_DOORBELL_OFFSET(ring->doorbell_index) |
+			  PACKET3_QUERY_STATUS_ENG_SEL(eng_sel));
+	amdgpu_ring_write(kiq_ring, lower_32_bits(addr));
+	amdgpu_ring_write(kiq_ring, upper_32_bits(addr));
+	amdgpu_ring_write(kiq_ring, lower_32_bits(seq));
+	amdgpu_ring_write(kiq_ring, upper_32_bits(seq));
+}
+
+static void gfx10_kiq_invalidate_tlbs(struct amdgpu_ring *kiq_ring,
+				uint16_t pasid, uint32_t flush_type,
+				bool all_hub)
+{
+	gfx_v10_0_ring_invalidate_tlbs(kiq_ring, pasid, flush_type, all_hub, 1);
+}
+
+static void gfx_v10_0_kiq_reset_hw_queue(struct amdgpu_ring *kiq_ring, uint32_t queue_type,
+					uint32_t me_id, uint32_t pipe_id, uint32_t queue_id,
+					uint32_t xcc_id, uint32_t vmid)
+{
+	struct amdgpu_device *adev = kiq_ring->adev;
+	unsigned i;
+	uint32_t tmp;
+
+	/* enter save mode */
+	amdgpu_gfx_rlc_enter_safe_mode(adev, xcc_id);
+	mutex_lock(&adev->srbm_mutex);
+	nv_grbm_select(adev, me_id, pipe_id, queue_id, 0);
+
+	if (queue_type == AMDGPU_RING_TYPE_COMPUTE) {
+		WREG32_SOC15(GC, 0, mmCP_HQD_DEQUEUE_REQUEST, 0x2);
+		WREG32_SOC15(GC, 0, mmSPI_COMPUTE_QUEUE_RESET, 0x1);
+		/* wait till dequeue take effects */
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (!(RREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE) & 1))
+				break;
+			udelay(1);
+		}
+		if (i >= adev->usec_timeout)
+			dev_err(adev->dev, "fail to wait on hqd deactive\n");
+	} else if (queue_type == AMDGPU_RING_TYPE_GFX) {
+		WREG32_SOC15(GC, 0, mmGRBM_GFX_INDEX,
+			     (uint32_t)(0x1 << GRBM_GFX_INDEX__SE_BROADCAST_WRITES__SHIFT));
+		tmp = REG_SET_FIELD(0, CP_VMID_RESET, RESET_REQUEST, 1 << vmid);
+		if (pipe_id == 0)
+			tmp = REG_SET_FIELD(tmp, CP_VMID_RESET, PIPE0_QUEUES, 1 << queue_id);
+		else
+			tmp = REG_SET_FIELD(tmp, CP_VMID_RESET, PIPE1_QUEUES, 1 << queue_id);
+		WREG32_SOC15(GC, 0, mmCP_VMID_RESET, tmp);
+
+		/* wait till dequeue take effects */
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (!(RREG32_SOC15(GC, 0, mmCP_GFX_HQD_ACTIVE) & 1))
+				break;
+			udelay(1);
+		}
+		if (i >= adev->usec_timeout)
+			dev_err(adev->dev, "failed to wait on gfx hqd deactivate\n");
+	} else {
+		dev_err(adev->dev, "reset queue_type(%d) not supported\n", queue_type);
+	}
+
+	nv_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+	/* exit safe mode */
+	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
+}
+
+static const struct kiq_pm4_funcs gfx_v10_0_kiq_pm4_funcs = {
+	.kiq_set_resources = gfx10_kiq_set_resources,
+	.kiq_map_queues = gfx10_kiq_map_queues,
+	.kiq_unmap_queues = gfx10_kiq_unmap_queues,
+	.kiq_query_status = gfx10_kiq_query_status,
+	.kiq_invalidate_tlbs = gfx10_kiq_invalidate_tlbs,
+	.kiq_reset_hw_queue = gfx_v10_0_kiq_reset_hw_queue,
+	.set_resources_size = 8,
+	.map_queues_size = 7,
+	.unmap_queues_size = 6,
+	.query_status_size = 7,
+	.invalidate_tlbs_size = 2,
+};
+
+static void gfx_v10_0_set_kiq_pm4_funcs(struct amdgpu_device *adev)
+{
+	adev->gfx.kiq[0].pmf = &gfx_v10_0_kiq_pm4_funcs;
+}
+
+static void gfx_v10_0_init_spm_golden_registers(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 1, 10):
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_rlc_spm_10_0_nv10,
+						(const u32)ARRAY_SIZE(golden_settings_gc_rlc_spm_10_0_nv10));
+		break;
+	case IP_VERSION(10, 1, 1):
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_rlc_spm_10_1_nv14,
+						(const u32)ARRAY_SIZE(golden_settings_gc_rlc_spm_10_1_nv14));
+		break;
+	case IP_VERSION(10, 1, 2):
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_rlc_spm_10_1_2_nv12,
+						(const u32)ARRAY_SIZE(golden_settings_gc_rlc_spm_10_1_2_nv12));
+		break;
+	default:
+		break;
+	}
+}
+
+static void gfx_v10_0_init_golden_registers(struct amdgpu_device *adev)
+{
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 1, 10):
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_10_1,
+						(const u32)ARRAY_SIZE(golden_settings_gc_10_1));
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_10_0_nv10,
+						(const u32)ARRAY_SIZE(golden_settings_gc_10_0_nv10));
+		break;
+	case IP_VERSION(10, 1, 1):
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_10_1_1,
+						(const u32)ARRAY_SIZE(golden_settings_gc_10_1_1));
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_10_1_nv14,
+						(const u32)ARRAY_SIZE(golden_settings_gc_10_1_nv14));
+		break;
+	case IP_VERSION(10, 1, 2):
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_10_1_2,
+						(const u32)ARRAY_SIZE(golden_settings_gc_10_1_2));
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_10_1_2_nv12,
+						(const u32)ARRAY_SIZE(golden_settings_gc_10_1_2_nv12));
+		break;
+	case IP_VERSION(10, 3, 0):
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_10_3,
+						(const u32)ARRAY_SIZE(golden_settings_gc_10_3));
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_10_3_sienna_cichlid,
+						(const u32)ARRAY_SIZE(golden_settings_gc_10_3_sienna_cichlid));
+		break;
+	case IP_VERSION(10, 3, 2):
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_10_3_2,
+						(const u32)ARRAY_SIZE(golden_settings_gc_10_3_2));
+		break;
+	case IP_VERSION(10, 3, 1):
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_10_3_vangogh,
+						(const u32)ARRAY_SIZE(golden_settings_gc_10_3_vangogh));
+		break;
+	case IP_VERSION(10, 3, 3):
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_10_3_3,
+						(const u32)ARRAY_SIZE(golden_settings_gc_10_3_3));
+		break;
+	case IP_VERSION(10, 3, 4):
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_10_3_4,
+						(const u32)ARRAY_SIZE(golden_settings_gc_10_3_4));
+		break;
+	case IP_VERSION(10, 3, 5):
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_10_3_5,
+						(const u32)ARRAY_SIZE(golden_settings_gc_10_3_5));
+		break;
+	case IP_VERSION(10, 1, 3):
+	case IP_VERSION(10, 1, 4):
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_10_0_cyan_skillfish,
+						(const u32)ARRAY_SIZE(golden_settings_gc_10_0_cyan_skillfish));
+		break;
+	case IP_VERSION(10, 3, 6):
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_10_3_6,
+						(const u32)ARRAY_SIZE(golden_settings_gc_10_3_6));
+		break;
+	case IP_VERSION(10, 3, 7):
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_10_3_7,
+						(const u32)ARRAY_SIZE(golden_settings_gc_10_3_7));
+		break;
+	default:
+		break;
+	}
+	gfx_v10_0_init_spm_golden_registers(adev);
+}
+
+static void gfx_v10_0_write_data_to_reg(struct amdgpu_ring *ring, int eng_sel,
+				       bool wc, uint32_t reg, uint32_t val)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, WRITE_DATA_ENGINE_SEL(eng_sel) |
+			  WRITE_DATA_DST_SEL(0) | (wc ? WR_CONFIRM : 0));
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, val);
+}
+
+static void gfx_v10_0_wait_reg_mem(struct amdgpu_ring *ring, int eng_sel,
+				  int mem_space, int opt, uint32_t addr0,
+				  uint32_t addr1, uint32_t ref, uint32_t mask,
+				  uint32_t inv)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+	amdgpu_ring_write(ring,
+			  /* memory (1) or register (0) */
+			  (WAIT_REG_MEM_MEM_SPACE(mem_space) |
+			   WAIT_REG_MEM_OPERATION(opt) | /* wait */
+			   WAIT_REG_MEM_FUNCTION(3) |  /* equal */
+			   WAIT_REG_MEM_ENGINE(eng_sel)));
+
+	if (mem_space)
+		BUG_ON(addr0 & 0x3); /* Dword align */
+	amdgpu_ring_write(ring, addr0);
+	amdgpu_ring_write(ring, addr1);
+	amdgpu_ring_write(ring, ref);
+	amdgpu_ring_write(ring, mask);
+	amdgpu_ring_write(ring, inv); /* poll interval */
+}
+
+static int gfx_v10_0_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t scratch = SOC15_REG_OFFSET(GC, 0, mmSCRATCH_REG0);
+	uint32_t tmp = 0;
+	unsigned int i;
+	int r;
+
+	WREG32(scratch, 0xCAFEDEAD);
+	r = amdgpu_ring_alloc(ring, 3);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n",
+			  ring->idx, r);
+		return r;
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_UCONFIG_REG, 1));
+	amdgpu_ring_write(ring, scratch -
+			  PACKET3_SET_UCONFIG_REG_START);
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(scratch);
+		if (tmp == 0xDEADBEEF)
+			break;
+		if (amdgpu_emu_mode == 1)
+			msleep(1);
+		else
+			udelay(1);
+	}
+
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+
+	return r;
+}
+
+static int gfx_v10_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct dma_fence *f = NULL;
+	unsigned int index;
+	uint64_t gpu_addr;
+	uint32_t *cpu_ptr;
+	long r;
+
+	memset(&ib, 0, sizeof(ib));
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r)
+		return r;
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	adev->wb.wb[index] = cpu_to_le32(0xCAFEDEAD);
+	cpu_ptr = &adev->wb.wb[index];
+
+	r = amdgpu_ib_get(adev, NULL, 20, AMDGPU_IB_POOL_DIRECT, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
+		goto err1;
+	}
+
+	ib.ptr[0] = PACKET3(PACKET3_WRITE_DATA, 3);
+	ib.ptr[1] = WRITE_DATA_DST_SEL(5) | WR_CONFIRM;
+	ib.ptr[2] = lower_32_bits(gpu_addr);
+	ib.ptr[3] = upper_32_bits(gpu_addr);
+	ib.ptr[4] = 0xDEADBEEF;
+	ib.length_dw = 5;
+
+	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
+	if (r)
+		goto err2;
+
+	r = dma_fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		r = -ETIMEDOUT;
+		goto err2;
+	} else if (r < 0) {
+		goto err2;
+	}
+
+	if (le32_to_cpu(*cpu_ptr) == 0xDEADBEEF)
+		r = 0;
+	else
+		r = -EINVAL;
+err2:
+	amdgpu_ib_free(&ib, NULL);
+	dma_fence_put(f);
+err1:
+	amdgpu_device_wb_free(adev, index);
+	return r;
+}
+
+static void gfx_v10_0_free_microcode(struct amdgpu_device *adev)
+{
+	amdgpu_ucode_release(&adev->gfx.pfp_fw);
+	amdgpu_ucode_release(&adev->gfx.me_fw);
+	amdgpu_ucode_release(&adev->gfx.ce_fw);
+	amdgpu_ucode_release(&adev->gfx.rlc_fw);
+	amdgpu_ucode_release(&adev->gfx.mec_fw);
+	amdgpu_ucode_release(&adev->gfx.mec2_fw);
+
+	kfree(adev->gfx.rlc.register_list_format);
+}
+
+static void gfx_v10_0_check_fw_write_wait(struct amdgpu_device *adev)
+{
+	adev->gfx.cp_fw_write_wait = false;
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 1, 10):
+	case IP_VERSION(10, 1, 2):
+	case IP_VERSION(10, 1, 1):
+	case IP_VERSION(10, 1, 3):
+	case IP_VERSION(10, 1, 4):
+		if ((adev->gfx.me_fw_version >= 0x00000046) &&
+		    (adev->gfx.me_feature_version >= 27) &&
+		    (adev->gfx.pfp_fw_version >= 0x00000068) &&
+		    (adev->gfx.pfp_feature_version >= 27) &&
+		    (adev->gfx.mec_fw_version >= 0x0000005b) &&
+		    (adev->gfx.mec_feature_version >= 27))
+			adev->gfx.cp_fw_write_wait = true;
+		break;
+	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+	case IP_VERSION(10, 3, 6):
+	case IP_VERSION(10, 3, 3):
+	case IP_VERSION(10, 3, 7):
+		adev->gfx.cp_fw_write_wait = true;
+		break;
+	default:
+		break;
+	}
+
+	if (!adev->gfx.cp_fw_write_wait)
+		DRM_WARN_ONCE("CP firmware version too old, please update!");
+}
+
+static bool gfx_v10_0_navi10_gfxoff_should_enable(struct amdgpu_device *adev)
+{
+	bool ret = false;
+
+	switch (adev->pdev->revision) {
+	case 0xc2:
+	case 0xc3:
+		ret = true;
+		break;
+	default:
+		ret = false;
+		break;
+	}
+
+	return ret;
+}
+
+static void gfx_v10_0_check_gfxoff_flag(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 1, 10):
+		if (!gfx_v10_0_navi10_gfxoff_should_enable(adev))
+			adev->pm.pp_feature &= ~PP_GFXOFF_MASK;
+		break;
+	default:
+		break;
+	}
+}
+
+static int gfx_v10_0_init_microcode(struct amdgpu_device *adev)
+{
+	char fw_name[53];
+	char ucode_prefix[30];
+	const char *wks = "";
+	int err;
+	const struct rlc_firmware_header_v2_0 *rlc_hdr;
+	uint16_t version_major;
+	uint16_t version_minor;
+
+	DRM_DEBUG("\n");
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 1, 1) &&
+	    (!(adev->pdev->device == 0x7340 && adev->pdev->revision != 0x00)))
+		wks = "_wks";
+	amdgpu_ucode_ip_version_decode(adev, GC_HWIP, ucode_prefix, sizeof(ucode_prefix));
+
+	err = amdgpu_ucode_request(adev, &adev->gfx.pfp_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_pfp%s.bin", ucode_prefix, wks);
+	if (err)
+		goto out;
+	amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_PFP);
+
+	err = amdgpu_ucode_request(adev, &adev->gfx.me_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_me%s.bin", ucode_prefix, wks);
+	if (err)
+		goto out;
+	amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_ME);
+
+	err = amdgpu_ucode_request(adev, &adev->gfx.ce_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_ce%s.bin", ucode_prefix, wks);
+	if (err)
+		goto out;
+	amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_CE);
+
+	if (!amdgpu_sriov_vf(adev)) {
+		snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_rlc.bin", ucode_prefix);
+		err = request_firmware(&adev->gfx.rlc_fw, fw_name, adev->dev);
+		if (err)
+			goto out;
+
+		/* don't validate this firmware. There are apparently firmwares
+		 * in the wild with incorrect size in the header
+		 */
+		rlc_hdr = (const struct rlc_firmware_header_v2_0 *)adev->gfx.rlc_fw->data;
+		version_major = le16_to_cpu(rlc_hdr->header.header_version_major);
+		version_minor = le16_to_cpu(rlc_hdr->header.header_version_minor);
+		err = amdgpu_gfx_rlc_init_microcode(adev, version_major, version_minor);
+		if (err)
+			goto out;
+	}
+
+	err = amdgpu_ucode_request(adev, &adev->gfx.mec_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_mec%s.bin", ucode_prefix, wks);
+	if (err)
+		goto out;
+	amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_MEC1);
+	amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_MEC1_JT);
+
+	err = amdgpu_ucode_request(adev, &adev->gfx.mec2_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_mec2%s.bin", ucode_prefix, wks);
+	if (!err) {
+		amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_MEC2);
+		amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_MEC2_JT);
+	} else {
+		err = 0;
+		adev->gfx.mec2_fw = NULL;
+	}
+
+	gfx_v10_0_check_fw_write_wait(adev);
+out:
+	if (err) {
+		amdgpu_ucode_release(&adev->gfx.pfp_fw);
+		amdgpu_ucode_release(&adev->gfx.me_fw);
+		amdgpu_ucode_release(&adev->gfx.ce_fw);
+		amdgpu_ucode_release(&adev->gfx.rlc_fw);
+		amdgpu_ucode_release(&adev->gfx.mec_fw);
+		amdgpu_ucode_release(&adev->gfx.mec2_fw);
+	}
+
+	gfx_v10_0_check_gfxoff_flag(adev);
+
+	return err;
+}
+
+static u32 gfx_v10_0_get_csb_size(struct amdgpu_device *adev)
+{
+	u32 count = 0;
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+
+	/* begin clear state */
+	count += 2;
+	/* context control state */
+	count += 3;
+
+	for (sect = gfx10_cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT)
+				count += 2 + ext->reg_count;
+			else
+				return 0;
+		}
+	}
+
+	/* set PA_SC_TILE_STEERING_OVERRIDE */
+	count += 3;
+	/* end clear state */
+	count += 2;
+	/* clear state */
+	count += 2;
+
+	return count;
+}
+
+static void gfx_v10_0_get_csb_buffer(struct amdgpu_device *adev, u32 *buffer)
+{
+	u32 count = 0;
+	int ctx_reg_offset;
+
+	if (adev->gfx.rlc.cs_data == NULL)
+		return;
+	if (buffer == NULL)
+		return;
+
+	count = amdgpu_gfx_csb_preamble_start(buffer);
+	count = amdgpu_gfx_csb_data_parser(adev, buffer, count);
+
+	ctx_reg_offset = SOC15_REG_OFFSET(GC, 0, mmPA_SC_TILE_STEERING_OVERRIDE) - PACKET3_SET_CONTEXT_REG_START;
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, 1));
+	buffer[count++] = cpu_to_le32(ctx_reg_offset);
+	buffer[count++] = cpu_to_le32(adev->gfx.config.pa_sc_tile_steering_override);
+
+	amdgpu_gfx_csb_preamble_end(buffer, count);
+}
+
+static void gfx_v10_0_rlc_fini(struct amdgpu_device *adev)
+{
+	/* clear state block */
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj,
+			&adev->gfx.rlc.clear_state_gpu_addr,
+			(void **)&adev->gfx.rlc.cs_ptr);
+
+	/* jump table block */
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj,
+			&adev->gfx.rlc.cp_table_gpu_addr,
+			(void **)&adev->gfx.rlc.cp_table_ptr);
+}
+
+static void gfx_v10_0_init_rlcg_reg_access_ctrl(struct amdgpu_device *adev)
+{
+	struct amdgpu_rlcg_reg_access_ctrl *reg_access_ctrl;
+
+	reg_access_ctrl = &adev->gfx.rlc.reg_access_ctrl[0];
+	reg_access_ctrl->scratch_reg0 = SOC15_REG_OFFSET(GC, 0, mmSCRATCH_REG0);
+	reg_access_ctrl->scratch_reg1 = SOC15_REG_OFFSET(GC, 0, mmSCRATCH_REG1);
+	reg_access_ctrl->scratch_reg2 = SOC15_REG_OFFSET(GC, 0, mmSCRATCH_REG2);
+	reg_access_ctrl->scratch_reg3 = SOC15_REG_OFFSET(GC, 0, mmSCRATCH_REG3);
+	reg_access_ctrl->grbm_cntl = SOC15_REG_OFFSET(GC, 0, mmGRBM_GFX_CNTL);
+	reg_access_ctrl->grbm_idx = SOC15_REG_OFFSET(GC, 0, mmGRBM_GFX_INDEX);
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 3, 0):
+		reg_access_ctrl->spare_int =
+			SOC15_REG_OFFSET(GC, 0, mmRLC_SPARE_INT_0_Sienna_Cichlid);
+		break;
+	default:
+		reg_access_ctrl->spare_int =
+			SOC15_REG_OFFSET(GC, 0, mmRLC_SPARE_INT);
+		break;
+	}
+	adev->gfx.rlc.rlcg_reg_access_supported = true;
+}
+
+static int gfx_v10_0_rlc_init(struct amdgpu_device *adev)
+{
+	const struct cs_section_def *cs_data;
+	int r;
+
+	adev->gfx.rlc.cs_data = gfx10_cs_data;
+
+	cs_data = adev->gfx.rlc.cs_data;
+
+	if (cs_data) {
+		/* init clear state block */
+		r = amdgpu_gfx_rlc_init_csb(adev);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+static void gfx_v10_0_mec_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->gfx.mec.hpd_eop_obj, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gfx.mec.mec_fw_obj, NULL, NULL);
+}
+
+static void gfx_v10_0_me_init(struct amdgpu_device *adev)
+{
+	bitmap_zero(adev->gfx.me.queue_bitmap, AMDGPU_MAX_GFX_QUEUES);
+
+	amdgpu_gfx_graphics_queue_acquire(adev);
+}
+
+static int gfx_v10_0_mec_init(struct amdgpu_device *adev)
+{
+	int r;
+	u32 *hpd;
+	const __le32 *fw_data = NULL;
+	unsigned int fw_size;
+	u32 *fw = NULL;
+	size_t mec_hpd_size;
+
+	const struct gfx_firmware_header_v1_0 *mec_hdr = NULL;
+
+	bitmap_zero(adev->gfx.mec_bitmap[0].queue_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
+
+	/* take ownership of the relevant compute queues */
+	amdgpu_gfx_compute_queue_acquire(adev);
+	mec_hpd_size = adev->gfx.num_compute_rings * GFX10_MEC_HPD_SIZE;
+
+	if (mec_hpd_size) {
+		r = amdgpu_bo_create_reserved(adev, mec_hpd_size, PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_GTT,
+					      &adev->gfx.mec.hpd_eop_obj,
+					      &adev->gfx.mec.hpd_eop_gpu_addr,
+					      (void **)&hpd);
+		if (r) {
+			dev_warn(adev->dev, "(%d) create HDP EOP bo failed\n", r);
+			gfx_v10_0_mec_fini(adev);
+			return r;
+		}
+
+		memset(hpd, 0, mec_hpd_size);
+
+		amdgpu_bo_kunmap(adev->gfx.mec.hpd_eop_obj);
+		amdgpu_bo_unreserve(adev->gfx.mec.hpd_eop_obj);
+	}
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+		mec_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+
+		fw_data = (const __le32 *) (adev->gfx.mec_fw->data +
+			 le32_to_cpu(mec_hdr->header.ucode_array_offset_bytes));
+		fw_size = le32_to_cpu(mec_hdr->header.ucode_size_bytes);
+
+		r = amdgpu_bo_create_reserved(adev, mec_hdr->header.ucode_size_bytes,
+					      PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
+					      &adev->gfx.mec.mec_fw_obj,
+					      &adev->gfx.mec.mec_fw_gpu_addr,
+					      (void **)&fw);
+		if (r) {
+			dev_err(adev->dev, "(%d) failed to create mec fw bo\n", r);
+			gfx_v10_0_mec_fini(adev);
+			return r;
+		}
+
+		memcpy(fw, fw_data, fw_size);
+
+		amdgpu_bo_kunmap(adev->gfx.mec.mec_fw_obj);
+		amdgpu_bo_unreserve(adev->gfx.mec.mec_fw_obj);
+	}
+
+	return 0;
+}
+
+static uint32_t wave_read_ind(struct amdgpu_device *adev, uint32_t wave, uint32_t address)
+{
+	WREG32_SOC15(GC, 0, mmSQ_IND_INDEX,
+		(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
+		(address << SQ_IND_INDEX__INDEX__SHIFT));
+	return RREG32_SOC15(GC, 0, mmSQ_IND_DATA);
+}
+
+static void wave_read_regs(struct amdgpu_device *adev, uint32_t wave,
+			   uint32_t thread, uint32_t regno,
+			   uint32_t num, uint32_t *out)
+{
+	WREG32_SOC15(GC, 0, mmSQ_IND_INDEX,
+		(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
+		(regno << SQ_IND_INDEX__INDEX__SHIFT) |
+		(thread << SQ_IND_INDEX__WORKITEM_ID__SHIFT) |
+		(SQ_IND_INDEX__AUTO_INCR_MASK));
+	while (num--)
+		*(out++) = RREG32_SOC15(GC, 0, mmSQ_IND_DATA);
+}
+
+static void gfx_v10_0_read_wave_data(struct amdgpu_device *adev, uint32_t xcc_id, uint32_t simd, uint32_t wave, uint32_t *dst, int *no_fields)
+{
+	/* in gfx10 the SIMD_ID is specified as part of the INSTANCE
+	 * field when performing a select_se_sh so it should be
+	 * zero here
+	 */
+	WARN_ON(simd != 0);
+
+	/* type 2 wave data */
+	dst[(*no_fields)++] = 2;
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_STATUS);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_PC_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_PC_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_EXEC_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_EXEC_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_HW_ID1);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_HW_ID2);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_INST_DW0);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_GPR_ALLOC);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_LDS_ALLOC);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_TRAPSTS);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_IB_STS);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_IB_STS2);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_IB_DBG1);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_M0);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_MODE);
+}
+
+static void gfx_v10_0_read_wave_sgprs(struct amdgpu_device *adev, uint32_t xcc_id, uint32_t simd,
+				     uint32_t wave, uint32_t start,
+				     uint32_t size, uint32_t *dst)
+{
+	WARN_ON(simd != 0);
+
+	wave_read_regs(
+		adev, wave, 0, start + SQIND_WAVE_SGPRS_OFFSET, size,
+		dst);
+}
+
+static void gfx_v10_0_read_wave_vgprs(struct amdgpu_device *adev, uint32_t xcc_id, uint32_t simd,
+				      uint32_t wave, uint32_t thread,
+				      uint32_t start, uint32_t size,
+				      uint32_t *dst)
+{
+	wave_read_regs(
+		adev, wave, thread,
+		start + SQIND_WAVE_VGPRS_OFFSET, size, dst);
+}
+
+static void gfx_v10_0_select_me_pipe_q(struct amdgpu_device *adev,
+				       u32 me, u32 pipe, u32 q, u32 vm, u32 xcc_id)
+{
+	nv_grbm_select(adev, me, pipe, q, vm);
+}
+
+static void gfx_v10_0_update_perfmon_mgcg(struct amdgpu_device *adev,
+					  bool enable)
+{
+	uint32_t data, def;
+
+	data = def = RREG32_SOC15(GC, 0, mmRLC_PERFMON_CLK_CNTL);
+
+	if (enable)
+		data |= RLC_PERFMON_CLK_CNTL__PERFMON_CLOCK_STATE_MASK;
+	else
+		data &= ~RLC_PERFMON_CLK_CNTL__PERFMON_CLOCK_STATE_MASK;
+
+	if (data != def)
+		WREG32_SOC15(GC, 0, mmRLC_PERFMON_CLK_CNTL, data);
+}
+
+static const struct amdgpu_gfx_funcs gfx_v10_0_gfx_funcs = {
+	.get_gpu_clock_counter = &gfx_v10_0_get_gpu_clock_counter,
+	.select_se_sh = &gfx_v10_0_select_se_sh,
+	.read_wave_data = &gfx_v10_0_read_wave_data,
+	.read_wave_sgprs = &gfx_v10_0_read_wave_sgprs,
+	.read_wave_vgprs = &gfx_v10_0_read_wave_vgprs,
+	.select_me_pipe_q = &gfx_v10_0_select_me_pipe_q,
+	.init_spm_golden = &gfx_v10_0_init_spm_golden_registers,
+	.update_perfmon_mgcg = &gfx_v10_0_update_perfmon_mgcg,
+};
+
+static void gfx_v10_0_gpu_early_init(struct amdgpu_device *adev)
+{
+	u32 gb_addr_config;
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 1, 10):
+	case IP_VERSION(10, 1, 1):
+	case IP_VERSION(10, 1, 2):
+		adev->gfx.config.max_hw_contexts = 8;
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x4C0;
+		gb_addr_config = RREG32_SOC15(GC, 0, mmGB_ADDR_CONFIG);
+		break;
+	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+	case IP_VERSION(10, 3, 6):
+	case IP_VERSION(10, 3, 3):
+	case IP_VERSION(10, 3, 7):
+		adev->gfx.config.max_hw_contexts = 8;
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x4C0;
+		gb_addr_config = RREG32_SOC15(GC, 0, mmGB_ADDR_CONFIG);
+		adev->gfx.config.gb_addr_config_fields.num_pkrs =
+			1 << REG_GET_FIELD(gb_addr_config, GB_ADDR_CONFIG, NUM_PKRS);
+		break;
+	case IP_VERSION(10, 1, 3):
+	case IP_VERSION(10, 1, 4):
+		adev->gfx.config.max_hw_contexts = 8;
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x4C0;
+		gb_addr_config = CYAN_SKILLFISH_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	default:
+		BUG();
+		break;
+	}
+
+	adev->gfx.config.gb_addr_config = gb_addr_config;
+
+	adev->gfx.config.gb_addr_config_fields.num_pipes = 1 <<
+			REG_GET_FIELD(adev->gfx.config.gb_addr_config,
+				      GB_ADDR_CONFIG, NUM_PIPES);
+
+	adev->gfx.config.max_tile_pipes =
+		adev->gfx.config.gb_addr_config_fields.num_pipes;
+
+	adev->gfx.config.gb_addr_config_fields.max_compress_frags = 1 <<
+			REG_GET_FIELD(adev->gfx.config.gb_addr_config,
+				      GB_ADDR_CONFIG, MAX_COMPRESSED_FRAGS);
+	adev->gfx.config.gb_addr_config_fields.num_rb_per_se = 1 <<
+			REG_GET_FIELD(adev->gfx.config.gb_addr_config,
+				      GB_ADDR_CONFIG, NUM_RB_PER_SE);
+	adev->gfx.config.gb_addr_config_fields.num_se = 1 <<
+			REG_GET_FIELD(adev->gfx.config.gb_addr_config,
+				      GB_ADDR_CONFIG, NUM_SHADER_ENGINES);
+	adev->gfx.config.gb_addr_config_fields.pipe_interleave_size = 1 << (8 +
+			REG_GET_FIELD(adev->gfx.config.gb_addr_config,
+				      GB_ADDR_CONFIG, PIPE_INTERLEAVE_SIZE));
+}
+
+static int gfx_v10_0_gfx_ring_init(struct amdgpu_device *adev, int ring_id,
+				   int me, int pipe, int queue)
+{
+	struct amdgpu_ring *ring;
+	unsigned int irq_type;
+	unsigned int hw_prio;
+
+	ring = &adev->gfx.gfx_ring[ring_id];
+
+	ring->me = me;
+	ring->pipe = pipe;
+	ring->queue = queue;
+
+	ring->ring_obj = NULL;
+	ring->use_doorbell = true;
+
+	if (!ring_id)
+		ring->doorbell_index = adev->doorbell_index.gfx_ring0 << 1;
+	else
+		ring->doorbell_index = adev->doorbell_index.gfx_ring1 << 1;
+	ring->vm_hub = AMDGPU_GFXHUB(0);
+	sprintf(ring->name, "gfx_%d.%d.%d", ring->me, ring->pipe, ring->queue);
+
+	irq_type = AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP + ring->pipe;
+	hw_prio = amdgpu_gfx_is_high_priority_graphics_queue(adev, ring) ?
+			AMDGPU_GFX_PIPE_PRIO_HIGH : AMDGPU_GFX_PIPE_PRIO_NORMAL;
+	return amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq, irq_type,
+				hw_prio, NULL);
+}
+
+static int gfx_v10_0_compute_ring_init(struct amdgpu_device *adev, int ring_id,
+				       int mec, int pipe, int queue)
+{
+	unsigned int irq_type;
+	struct amdgpu_ring *ring;
+	unsigned int hw_prio;
+
+	ring = &adev->gfx.compute_ring[ring_id];
+
+	/* mec0 is me1 */
+	ring->me = mec + 1;
+	ring->pipe = pipe;
+	ring->queue = queue;
+
+	ring->ring_obj = NULL;
+	ring->use_doorbell = true;
+	ring->doorbell_index = (adev->doorbell_index.mec_ring0 + ring_id) << 1;
+	ring->eop_gpu_addr = adev->gfx.mec.hpd_eop_gpu_addr
+				+ (ring_id * GFX10_MEC_HPD_SIZE);
+	ring->vm_hub = AMDGPU_GFXHUB(0);
+	sprintf(ring->name, "comp_%d.%d.%d", ring->me, ring->pipe, ring->queue);
+
+	irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP
+		+ ((ring->me - 1) * adev->gfx.mec.num_pipe_per_mec)
+		+ ring->pipe;
+	hw_prio = amdgpu_gfx_is_high_priority_compute_queue(adev, ring) ?
+			AMDGPU_RING_PRIO_2 : AMDGPU_RING_PRIO_DEFAULT;
+	/* type-2 packets are deprecated on MEC, use type-3 instead */
+	return amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq, irq_type,
+			     hw_prio, NULL);
+}
+
+static void gfx_v10_0_alloc_ip_dump(struct amdgpu_device *adev)
+{
+	uint32_t reg_count = ARRAY_SIZE(gc_reg_list_10_1);
+	uint32_t *ptr;
+	uint32_t inst;
+
+	ptr = kcalloc(reg_count, sizeof(uint32_t), GFP_KERNEL);
+	if (!ptr) {
+		DRM_ERROR("Failed to allocate memory for GFX IP Dump\n");
+		adev->gfx.ip_dump_core = NULL;
+	} else {
+		adev->gfx.ip_dump_core = ptr;
+	}
+
+	/* Allocate memory for compute queue registers for all the instances */
+	reg_count = ARRAY_SIZE(gc_cp_reg_list_10);
+	inst = adev->gfx.mec.num_mec * adev->gfx.mec.num_pipe_per_mec *
+		adev->gfx.mec.num_queue_per_pipe;
+
+	ptr = kcalloc(reg_count * inst, sizeof(uint32_t), GFP_KERNEL);
+	if (!ptr) {
+		DRM_ERROR("Failed to allocate memory for Compute Queues IP Dump\n");
+		adev->gfx.ip_dump_compute_queues = NULL;
+	} else {
+		adev->gfx.ip_dump_compute_queues = ptr;
+	}
+
+	/* Allocate memory for gfx queue registers for all the instances */
+	reg_count = ARRAY_SIZE(gc_gfx_queue_reg_list_10);
+	inst = adev->gfx.me.num_me * adev->gfx.me.num_pipe_per_me *
+		adev->gfx.me.num_queue_per_pipe;
+
+	ptr = kcalloc(reg_count * inst, sizeof(uint32_t), GFP_KERNEL);
+	if (!ptr) {
+		DRM_ERROR("Failed to allocate memory for GFX Queues IP Dump\n");
+		adev->gfx.ip_dump_gfx_queues = NULL;
+	} else {
+		adev->gfx.ip_dump_gfx_queues = ptr;
+	}
+}
+
+static int gfx_v10_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	int i, j, k, r, ring_id = 0;
+	int xcc_id = 0;
+	struct amdgpu_device *adev = ip_block->adev;
+	int num_queue_per_pipe = 1; /* we only enable 1 KGQ per pipe */
+
+	INIT_DELAYED_WORK(&adev->gfx.idle_work, amdgpu_gfx_profile_idle_work_handler);
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 1, 10):
+	case IP_VERSION(10, 1, 1):
+	case IP_VERSION(10, 1, 2):
+	case IP_VERSION(10, 1, 3):
+	case IP_VERSION(10, 1, 4):
+		adev->gfx.me.num_me = 1;
+		adev->gfx.me.num_pipe_per_me = 1;
+		adev->gfx.me.num_queue_per_pipe = 8;
+		adev->gfx.mec.num_mec = 2;
+		adev->gfx.mec.num_pipe_per_mec = 4;
+		adev->gfx.mec.num_queue_per_pipe = 8;
+		break;
+	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+	case IP_VERSION(10, 3, 6):
+	case IP_VERSION(10, 3, 3):
+	case IP_VERSION(10, 3, 7):
+		adev->gfx.me.num_me = 1;
+		adev->gfx.me.num_pipe_per_me = 2;
+		adev->gfx.me.num_queue_per_pipe = 2;
+		adev->gfx.mec.num_mec = 2;
+		adev->gfx.mec.num_pipe_per_mec = 4;
+		adev->gfx.mec.num_queue_per_pipe = 4;
+		break;
+	default:
+		adev->gfx.me.num_me = 1;
+		adev->gfx.me.num_pipe_per_me = 1;
+		adev->gfx.me.num_queue_per_pipe = 1;
+		adev->gfx.mec.num_mec = 1;
+		adev->gfx.mec.num_pipe_per_mec = 4;
+		adev->gfx.mec.num_queue_per_pipe = 8;
+		break;
+	}
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 1, 10):
+	case IP_VERSION(10, 1, 1):
+	case IP_VERSION(10, 1, 2):
+		adev->gfx.cleaner_shader_ptr = gfx_10_1_10_cleaner_shader_hex;
+		adev->gfx.cleaner_shader_size = sizeof(gfx_10_1_10_cleaner_shader_hex);
+		if (adev->gfx.me_fw_version >= 101 &&
+		    adev->gfx.pfp_fw_version  >= 158 &&
+		    adev->gfx.mec_fw_version >= 151) {
+			adev->gfx.enable_cleaner_shader = true;
+			r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);
+			if (r) {
+				adev->gfx.enable_cleaner_shader = false;
+				dev_err(adev->dev, "Failed to initialize cleaner shader\n");
+			}
+		}
+		break;
+	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 3):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+		adev->gfx.cleaner_shader_ptr = gfx_10_3_0_cleaner_shader_hex;
+		adev->gfx.cleaner_shader_size = sizeof(gfx_10_3_0_cleaner_shader_hex);
+		if (adev->gfx.me_fw_version >= 64 &&
+		    adev->gfx.pfp_fw_version >= 100 &&
+		    adev->gfx.mec_fw_version >= 122) {
+			adev->gfx.enable_cleaner_shader = true;
+			r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);
+			if (r) {
+				adev->gfx.enable_cleaner_shader = false;
+				dev_err(adev->dev, "Failed to initialize cleaner shader\n");
+			}
+		}
+		break;
+	case IP_VERSION(10, 3, 6):
+		adev->gfx.cleaner_shader_ptr = gfx_10_3_0_cleaner_shader_hex;
+		adev->gfx.cleaner_shader_size = sizeof(gfx_10_3_0_cleaner_shader_hex);
+		if (adev->gfx.me_fw_version >= 14 &&
+		    adev->gfx.pfp_fw_version >= 17 &&
+		    adev->gfx.mec_fw_version >= 24) {
+			adev->gfx.enable_cleaner_shader = true;
+			r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);
+			if (r) {
+				adev->gfx.enable_cleaner_shader = false;
+				dev_err(adev->dev, "Failed to initialize cleaner shader\n");
+			}
+		}
+		break;
+	case IP_VERSION(10, 3, 7):
+		adev->gfx.cleaner_shader_ptr = gfx_10_3_0_cleaner_shader_hex;
+		adev->gfx.cleaner_shader_size = sizeof(gfx_10_3_0_cleaner_shader_hex);
+		if (adev->gfx.me_fw_version >= 4 &&
+		    adev->gfx.pfp_fw_version >= 9 &&
+		    adev->gfx.mec_fw_version >= 12) {
+			adev->gfx.enable_cleaner_shader = true;
+			r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);
+			if (r) {
+				adev->gfx.enable_cleaner_shader = false;
+				dev_err(adev->dev, "Failed to initialize cleaner shader\n");
+			}
+		}
+		break;
+	default:
+		adev->gfx.enable_cleaner_shader = false;
+		break;
+	}
+
+	/* KIQ event */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_GRBM_CP,
+			      GFX_10_1__SRCID__CP_IB2_INTERRUPT_PKT,
+			      &adev->gfx.kiq[0].irq);
+	if (r)
+		return r;
+
+	/* EOP Event */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_GRBM_CP,
+			      GFX_10_1__SRCID__CP_EOP_INTERRUPT,
+			      &adev->gfx.eop_irq);
+	if (r)
+		return r;
+
+	/* Bad opcode Event */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_GRBM_CP,
+			      GFX_10_1__SRCID__CP_BAD_OPCODE_ERROR,
+			      &adev->gfx.bad_op_irq);
+	if (r)
+		return r;
+
+	/* Privileged reg */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_GRBM_CP, GFX_10_1__SRCID__CP_PRIV_REG_FAULT,
+			      &adev->gfx.priv_reg_irq);
+	if (r)
+		return r;
+
+	/* Privileged inst */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_GRBM_CP, GFX_10_1__SRCID__CP_PRIV_INSTR_FAULT,
+			      &adev->gfx.priv_inst_irq);
+	if (r)
+		return r;
+
+	adev->gfx.gfx_current_status = AMDGPU_GFX_NORMAL_MODE;
+
+	gfx_v10_0_me_init(adev);
+
+	if (adev->gfx.rlc.funcs) {
+		if (adev->gfx.rlc.funcs->init) {
+			r = adev->gfx.rlc.funcs->init(adev);
+			if (r) {
+				dev_err(adev->dev, "Failed to init rlc BOs!\n");
+				return r;
+			}
+		}
+	}
+
+	r = gfx_v10_0_mec_init(adev);
+	if (r) {
+		DRM_ERROR("Failed to init MEC BOs!\n");
+		return r;
+	}
+
+	/* set up the gfx ring */
+	for (i = 0; i < adev->gfx.me.num_me; i++) {
+		for (j = 0; j < num_queue_per_pipe; j++) {
+			for (k = 0; k < adev->gfx.me.num_pipe_per_me; k++) {
+				if (!amdgpu_gfx_is_me_queue_enabled(adev, i, k, j))
+					continue;
+
+				r = gfx_v10_0_gfx_ring_init(adev, ring_id,
+							    i, k, j);
+				if (r)
+					return r;
+				ring_id++;
+			}
+		}
+	}
+
+	ring_id = 0;
+	/* set up the compute queues - allocate horizontally across pipes */
+	for (i = 0; i < adev->gfx.mec.num_mec; ++i) {
+		for (j = 0; j < adev->gfx.mec.num_queue_per_pipe; j++) {
+			for (k = 0; k < adev->gfx.mec.num_pipe_per_mec; k++) {
+				if (!amdgpu_gfx_is_mec_queue_enabled(adev, 0, i,
+								     k, j))
+					continue;
+
+				r = gfx_v10_0_compute_ring_init(adev, ring_id,
+								i, k, j);
+				if (r)
+					return r;
+
+				ring_id++;
+			}
+		}
+	}
+
+	adev->gfx.gfx_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.gfx_ring[0]);
+	adev->gfx.compute_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.compute_ring[0]);
+	if (!amdgpu_sriov_vf(adev)) {
+		adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+		adev->gfx.gfx_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+	}
+
+	r = amdgpu_gfx_kiq_init(adev, GFX10_MEC_HPD_SIZE, 0);
+	if (r) {
+		DRM_ERROR("Failed to init KIQ BOs!\n");
+		return r;
+	}
+
+	r = amdgpu_gfx_kiq_init_ring(adev, xcc_id);
+	if (r)
+		return r;
+
+	r = amdgpu_gfx_mqd_sw_init(adev, sizeof(struct v10_compute_mqd), 0);
+	if (r)
+		return r;
+
+	/* allocate visible FB for rlc auto-loading fw */
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO) {
+		r = gfx_v10_0_rlc_backdoor_autoload_buffer_init(adev);
+		if (r)
+			return r;
+	}
+
+	adev->gfx.ce_ram_size = F32_CE_PROGRAM_RAM_SIZE;
+
+	gfx_v10_0_gpu_early_init(adev);
+
+	gfx_v10_0_alloc_ip_dump(adev);
+
+	r = amdgpu_gfx_sysfs_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static void gfx_v10_0_pfp_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->gfx.pfp.pfp_fw_obj,
+			      &adev->gfx.pfp.pfp_fw_gpu_addr,
+			      (void **)&adev->gfx.pfp.pfp_fw_ptr);
+}
+
+static void gfx_v10_0_ce_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->gfx.ce.ce_fw_obj,
+			      &adev->gfx.ce.ce_fw_gpu_addr,
+			      (void **)&adev->gfx.ce.ce_fw_ptr);
+}
+
+static void gfx_v10_0_me_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->gfx.me.me_fw_obj,
+			      &adev->gfx.me.me_fw_gpu_addr,
+			      (void **)&adev->gfx.me.me_fw_ptr);
+}
+
+static int gfx_v10_0_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	int i;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+		amdgpu_ring_fini(&adev->gfx.gfx_ring[i]);
+	for (i = 0; i < adev->gfx.num_compute_rings; i++)
+		amdgpu_ring_fini(&adev->gfx.compute_ring[i]);
+
+	amdgpu_gfx_mqd_sw_fini(adev, 0);
+
+	amdgpu_gfx_kiq_free_ring(&adev->gfx.kiq[0].ring);
+	amdgpu_gfx_kiq_fini(adev, 0);
+
+	amdgpu_gfx_cleaner_shader_sw_fini(adev);
+
+	gfx_v10_0_pfp_fini(adev);
+	gfx_v10_0_ce_fini(adev);
+	gfx_v10_0_me_fini(adev);
+	gfx_v10_0_rlc_fini(adev);
+	gfx_v10_0_mec_fini(adev);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO)
+		gfx_v10_0_rlc_backdoor_autoload_buffer_fini(adev);
+
+	gfx_v10_0_free_microcode(adev);
+	amdgpu_gfx_sysfs_fini(adev);
+
+	kfree(adev->gfx.ip_dump_core);
+	kfree(adev->gfx.ip_dump_compute_queues);
+	kfree(adev->gfx.ip_dump_gfx_queues);
+
+	return 0;
+}
+
+static void gfx_v10_0_select_se_sh(struct amdgpu_device *adev, u32 se_num,
+				   u32 sh_num, u32 instance, int xcc_id)
+{
+	u32 data;
+
+	if (instance == 0xffffffff)
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX,
+				     INSTANCE_BROADCAST_WRITES, 1);
+	else
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_INDEX,
+				     instance);
+
+	if (se_num == 0xffffffff)
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_BROADCAST_WRITES,
+				     1);
+	else
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_INDEX, se_num);
+
+	if (sh_num == 0xffffffff)
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SA_BROADCAST_WRITES,
+				     1);
+	else
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SA_INDEX, sh_num);
+
+	WREG32_SOC15(GC, 0, mmGRBM_GFX_INDEX, data);
+}
+
+static u32 gfx_v10_0_get_rb_active_bitmap(struct amdgpu_device *adev)
+{
+	u32 data, mask;
+
+	data = RREG32_SOC15(GC, 0, mmCC_RB_BACKEND_DISABLE);
+	data |= RREG32_SOC15(GC, 0, mmGC_USER_RB_BACKEND_DISABLE);
+
+	data &= CC_RB_BACKEND_DISABLE__BACKEND_DISABLE_MASK;
+	data >>= GC_USER_RB_BACKEND_DISABLE__BACKEND_DISABLE__SHIFT;
+
+	mask = amdgpu_gfx_create_bitmask(adev->gfx.config.max_backends_per_se /
+					 adev->gfx.config.max_sh_per_se);
+
+	return (~data) & mask;
+}
+
+static void gfx_v10_0_setup_rb(struct amdgpu_device *adev)
+{
+	int i, j;
+	u32 data;
+	u32 active_rbs = 0;
+	u32 bitmap;
+	u32 rb_bitmap_width_per_sh = adev->gfx.config.max_backends_per_se /
+					adev->gfx.config.max_sh_per_se;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			bitmap = i * adev->gfx.config.max_sh_per_se + j;
+			if (((amdgpu_ip_version(adev, GC_HWIP, 0) ==
+			      IP_VERSION(10, 3, 0)) ||
+			     (amdgpu_ip_version(adev, GC_HWIP, 0) ==
+			      IP_VERSION(10, 3, 3)) ||
+			     (amdgpu_ip_version(adev, GC_HWIP, 0) ==
+			      IP_VERSION(10, 3, 6))) &&
+			    ((gfx_v10_3_get_disabled_sa(adev) >> bitmap) & 1))
+				continue;
+			gfx_v10_0_select_se_sh(adev, i, j, 0xffffffff, 0);
+			data = gfx_v10_0_get_rb_active_bitmap(adev);
+			active_rbs |= data << ((i * adev->gfx.config.max_sh_per_se + j) *
+					       rb_bitmap_width_per_sh);
+		}
+	}
+	gfx_v10_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	adev->gfx.config.backend_enable_mask = active_rbs;
+	adev->gfx.config.num_rbs = hweight32(active_rbs);
+}
+
+static u32 gfx_v10_0_init_pa_sc_tile_steering_override(struct amdgpu_device *adev)
+{
+	uint32_t num_sc;
+	uint32_t enabled_rb_per_sh;
+	uint32_t active_rb_bitmap;
+	uint32_t num_rb_per_sc;
+	uint32_t num_packer_per_sc;
+	uint32_t pa_sc_tile_steering_override;
+
+	/* for ASICs that integrates GFX v10.3
+	 * pa_sc_tile_steering_override should be set to 0
+	 */
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(10, 3, 0))
+		return 0;
+
+	/* init num_sc */
+	num_sc = adev->gfx.config.max_shader_engines * adev->gfx.config.max_sh_per_se *
+			adev->gfx.config.num_sc_per_sh;
+	/* init num_rb_per_sc */
+	active_rb_bitmap = gfx_v10_0_get_rb_active_bitmap(adev);
+	enabled_rb_per_sh = hweight32(active_rb_bitmap);
+	num_rb_per_sc = enabled_rb_per_sh / adev->gfx.config.num_sc_per_sh;
+	/* init num_packer_per_sc */
+	num_packer_per_sc = adev->gfx.config.num_packer_per_sc;
+
+	pa_sc_tile_steering_override = 0;
+	pa_sc_tile_steering_override |=
+		(order_base_2(num_sc) << PA_SC_TILE_STEERING_OVERRIDE__NUM_SC__SHIFT) &
+		PA_SC_TILE_STEERING_OVERRIDE__NUM_SC_MASK;
+	pa_sc_tile_steering_override |=
+		(order_base_2(num_rb_per_sc) << PA_SC_TILE_STEERING_OVERRIDE__NUM_RB_PER_SC__SHIFT) &
+		PA_SC_TILE_STEERING_OVERRIDE__NUM_RB_PER_SC_MASK;
+	pa_sc_tile_steering_override |=
+		(order_base_2(num_packer_per_sc) << PA_SC_TILE_STEERING_OVERRIDE__NUM_PACKER_PER_SC__SHIFT) &
+		PA_SC_TILE_STEERING_OVERRIDE__NUM_PACKER_PER_SC_MASK;
+
+	return pa_sc_tile_steering_override;
+}
+
+#define DEFAULT_SH_MEM_BASES	(0x6000)
+
+static void gfx_v10_0_debug_trap_config_init(struct amdgpu_device *adev,
+				uint32_t first_vmid,
+				uint32_t last_vmid)
+{
+	uint32_t data;
+	uint32_t trap_config_vmid_mask = 0;
+	int i;
+
+	/* Calculate trap config vmid mask */
+	for (i = first_vmid; i < last_vmid; i++)
+		trap_config_vmid_mask |= (1 << i);
+
+	data = REG_SET_FIELD(0, SPI_GDBG_TRAP_CONFIG,
+			VMID_SEL, trap_config_vmid_mask);
+	data = REG_SET_FIELD(data, SPI_GDBG_TRAP_CONFIG,
+			TRAP_EN, 1);
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_TRAP_CONFIG), data);
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_TRAP_MASK), 0);
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_TRAP_DATA0), 0);
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_TRAP_DATA1), 0);
+}
+
+static void gfx_v10_0_init_compute_vmid(struct amdgpu_device *adev)
+{
+	int i;
+	uint32_t sh_mem_bases;
+
+	/*
+	 * Configure apertures:
+	 * LDS:         0x60000000'00000000 - 0x60000001'00000000 (4GB)
+	 * Scratch:     0x60000001'00000000 - 0x60000002'00000000 (4GB)
+	 * GPUVM:       0x60010000'00000000 - 0x60020000'00000000 (1TB)
+	 */
+	sh_mem_bases = DEFAULT_SH_MEM_BASES | (DEFAULT_SH_MEM_BASES << 16);
+
+	mutex_lock(&adev->srbm_mutex);
+	for (i = adev->vm_manager.first_kfd_vmid; i < AMDGPU_NUM_VMID; i++) {
+		nv_grbm_select(adev, 0, 0, 0, i);
+		/* CP and shaders */
+		WREG32_SOC15(GC, 0, mmSH_MEM_CONFIG, DEFAULT_SH_MEM_CONFIG);
+		WREG32_SOC15(GC, 0, mmSH_MEM_BASES, sh_mem_bases);
+	}
+	nv_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	/*
+	 * Initialize all compute VMIDs to have no GDS, GWS, or OA
+	 * access. These should be enabled by FW for target VMIDs.
+	 */
+	for (i = adev->vm_manager.first_kfd_vmid; i < AMDGPU_NUM_VMID; i++) {
+		WREG32_SOC15_OFFSET(GC, 0, mmGDS_VMID0_BASE, 2 * i, 0);
+		WREG32_SOC15_OFFSET(GC, 0, mmGDS_VMID0_SIZE, 2 * i, 0);
+		WREG32_SOC15_OFFSET(GC, 0, mmGDS_GWS_VMID0, i, 0);
+		WREG32_SOC15_OFFSET(GC, 0, mmGDS_OA_VMID0, i, 0);
+	}
+
+	gfx_v10_0_debug_trap_config_init(adev, adev->vm_manager.first_kfd_vmid,
+					AMDGPU_NUM_VMID);
+}
+
+static void gfx_v10_0_init_gds_vmid(struct amdgpu_device *adev)
+{
+	int vmid;
+
+	/*
+	 * Initialize all compute and user-gfx VMIDs to have no GDS, GWS, or OA
+	 * access. Compute VMIDs should be enabled by FW for target VMIDs,
+	 * the driver can enable them for graphics. VMID0 should maintain
+	 * access so that HWS firmware can save/restore entries.
+	 */
+	for (vmid = 1; vmid < AMDGPU_NUM_VMID; vmid++) {
+		WREG32_SOC15_OFFSET(GC, 0, mmGDS_VMID0_BASE, 2 * vmid, 0);
+		WREG32_SOC15_OFFSET(GC, 0, mmGDS_VMID0_SIZE, 2 * vmid, 0);
+		WREG32_SOC15_OFFSET(GC, 0, mmGDS_GWS_VMID0, vmid, 0);
+		WREG32_SOC15_OFFSET(GC, 0, mmGDS_OA_VMID0, vmid, 0);
+	}
+}
+
+
+static void gfx_v10_0_tcp_harvest(struct amdgpu_device *adev)
+{
+	int i, j, k;
+	int max_wgp_per_sh = adev->gfx.config.max_cu_per_sh >> 1;
+	u32 tmp, wgp_active_bitmap = 0;
+	u32 gcrd_targets_disable_tcp = 0;
+	u32 utcl_invreq_disable = 0;
+	/*
+	 * GCRD_TARGETS_DISABLE field contains
+	 * for Navi10/Navi12: GL1C=[18:15], SQC=[14:10], TCP=[9:0]
+	 * for Navi14: GL1C=[21:18], SQC=[17:12], TCP=[11:0]
+	 */
+	u32 gcrd_targets_disable_mask = amdgpu_gfx_create_bitmask(
+		2 * max_wgp_per_sh + /* TCP */
+		max_wgp_per_sh + /* SQC */
+		4); /* GL1C */
+	/*
+	 * UTCL1_UTCL0_INVREQ_DISABLE field contains
+	 * for Navi10Navi12: SQG=[24], RMI=[23:20], SQC=[19:10], TCP=[9:0]
+	 * for Navi14: SQG=[28], RMI=[27:24], SQC=[23:12], TCP=[11:0]
+	 */
+	u32 utcl_invreq_disable_mask = amdgpu_gfx_create_bitmask(
+		2 * max_wgp_per_sh + /* TCP */
+		2 * max_wgp_per_sh + /* SQC */
+		4 + /* RMI */
+		1); /* SQG */
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			gfx_v10_0_select_se_sh(adev, i, j, 0xffffffff, 0);
+			wgp_active_bitmap = gfx_v10_0_get_wgp_active_bitmap_per_sh(adev);
+			/*
+			 * Set corresponding TCP bits for the inactive WGPs in
+			 * GCRD_SA_TARGETS_DISABLE
+			 */
+			gcrd_targets_disable_tcp = 0;
+			/* Set TCP & SQC bits in UTCL1_UTCL0_INVREQ_DISABLE */
+			utcl_invreq_disable = 0;
+
+			for (k = 0; k < max_wgp_per_sh; k++) {
+				if (!(wgp_active_bitmap & (1 << k))) {
+					gcrd_targets_disable_tcp |= 3 << (2 * k);
+					gcrd_targets_disable_tcp |= 1 << (k + (max_wgp_per_sh * 2));
+					utcl_invreq_disable |= (3 << (2 * k)) |
+						(3 << (2 * (max_wgp_per_sh + k)));
+				}
+			}
+
+			tmp = RREG32_SOC15(GC, 0, mmUTCL1_UTCL0_INVREQ_DISABLE);
+			/* only override TCP & SQC bits */
+			tmp &= (0xffffffffU << (4 * max_wgp_per_sh));
+			tmp |= (utcl_invreq_disable & utcl_invreq_disable_mask);
+			WREG32_SOC15(GC, 0, mmUTCL1_UTCL0_INVREQ_DISABLE, tmp);
+
+			tmp = RREG32_SOC15(GC, 0, mmGCRD_SA_TARGETS_DISABLE);
+			/* only override TCP & SQC bits */
+			tmp &= (0xffffffffU << (3 * max_wgp_per_sh));
+			tmp |= (gcrd_targets_disable_tcp & gcrd_targets_disable_mask);
+			WREG32_SOC15(GC, 0, mmGCRD_SA_TARGETS_DISABLE, tmp);
+		}
+	}
+
+	gfx_v10_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
+	mutex_unlock(&adev->grbm_idx_mutex);
+}
+
+static void gfx_v10_0_get_tcc_info(struct amdgpu_device *adev)
+{
+	/* TCCs are global (not instanced). */
+	uint32_t tcc_disable;
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(10, 3, 0)) {
+		tcc_disable = RREG32_SOC15(GC, 0, mmCGTS_TCC_DISABLE_gc_10_3) |
+			      RREG32_SOC15(GC, 0, mmCGTS_USER_TCC_DISABLE_gc_10_3);
+	} else {
+		tcc_disable = RREG32_SOC15(GC, 0, mmCGTS_TCC_DISABLE) |
+			      RREG32_SOC15(GC, 0, mmCGTS_USER_TCC_DISABLE);
+	}
+
+	adev->gfx.config.tcc_disabled_mask =
+		REG_GET_FIELD(tcc_disable, CGTS_TCC_DISABLE, TCC_DISABLE) |
+		(REG_GET_FIELD(tcc_disable, CGTS_TCC_DISABLE, HI_TCC_DISABLE) << 16);
+}
+
+static void gfx_v10_0_constants_init(struct amdgpu_device *adev)
+{
+	u32 tmp;
+	int i;
+
+	if (!amdgpu_sriov_vf(adev))
+		WREG32_FIELD15(GC, 0, GRBM_CNTL, READ_TIMEOUT, 0xff);
+
+	gfx_v10_0_setup_rb(adev);
+	gfx_v10_0_get_cu_info(adev, &adev->gfx.cu_info);
+	gfx_v10_0_get_tcc_info(adev);
+	adev->gfx.config.pa_sc_tile_steering_override =
+		gfx_v10_0_init_pa_sc_tile_steering_override(adev);
+
+	/* XXX SH_MEM regs */
+	/* where to put LDS, scratch, GPUVM in FSA64 space */
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < adev->vm_manager.id_mgr[AMDGPU_GFXHUB(0)].num_ids; i++) {
+		nv_grbm_select(adev, 0, 0, 0, i);
+		/* CP and shaders */
+		WREG32_SOC15(GC, 0, mmSH_MEM_CONFIG, DEFAULT_SH_MEM_CONFIG);
+		if (i != 0) {
+			tmp = REG_SET_FIELD(0, SH_MEM_BASES, PRIVATE_BASE,
+				(adev->gmc.private_aperture_start >> 48));
+			tmp = REG_SET_FIELD(tmp, SH_MEM_BASES, SHARED_BASE,
+				(adev->gmc.shared_aperture_start >> 48));
+			WREG32_SOC15(GC, 0, mmSH_MEM_BASES, tmp);
+		}
+	}
+	nv_grbm_select(adev, 0, 0, 0, 0);
+
+	mutex_unlock(&adev->srbm_mutex);
+
+	gfx_v10_0_init_compute_vmid(adev);
+	gfx_v10_0_init_gds_vmid(adev);
+
+}
+
+static u32 gfx_v10_0_get_cpg_int_cntl(struct amdgpu_device *adev,
+				      int me, int pipe)
+{
+	if (me != 0)
+		return 0;
+
+	switch (pipe) {
+	case 0:
+		return SOC15_REG_OFFSET(GC, 0, mmCP_INT_CNTL_RING0);
+	case 1:
+		return SOC15_REG_OFFSET(GC, 0, mmCP_INT_CNTL_RING1);
+	default:
+		return 0;
+	}
+}
+
+static u32 gfx_v10_0_get_cpc_int_cntl(struct amdgpu_device *adev,
+				      int me, int pipe)
+{
+	/*
+	 * amdgpu controls only the first MEC. That's why this function only
+	 * handles the setting of interrupts for this specific MEC. All other
+	 * pipes' interrupts are set by amdkfd.
+	 */
+	if (me != 1)
+		return 0;
+
+	switch (pipe) {
+	case 0:
+		return SOC15_REG_OFFSET(GC, 0, mmCP_ME1_PIPE0_INT_CNTL);
+	case 1:
+		return SOC15_REG_OFFSET(GC, 0, mmCP_ME1_PIPE1_INT_CNTL);
+	case 2:
+		return SOC15_REG_OFFSET(GC, 0, mmCP_ME1_PIPE2_INT_CNTL);
+	case 3:
+		return SOC15_REG_OFFSET(GC, 0, mmCP_ME1_PIPE3_INT_CNTL);
+	default:
+		return 0;
+	}
+}
+
+static void gfx_v10_0_enable_gui_idle_interrupt(struct amdgpu_device *adev,
+					       bool enable)
+{
+	u32 tmp, cp_int_cntl_reg;
+	int i, j;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	for (i = 0; i < adev->gfx.me.num_me; i++) {
+		for (j = 0; j < adev->gfx.me.num_pipe_per_me; j++) {
+			cp_int_cntl_reg = gfx_v10_0_get_cpg_int_cntl(adev, i, j);
+
+			if (cp_int_cntl_reg) {
+				tmp = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+				tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_BUSY_INT_ENABLE,
+						    enable ? 1 : 0);
+				tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_EMPTY_INT_ENABLE,
+						    enable ? 1 : 0);
+				tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CMP_BUSY_INT_ENABLE,
+						    enable ? 1 : 0);
+				tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, GFX_IDLE_INT_ENABLE,
+						    enable ? 1 : 0);
+				WREG32_SOC15_IP(GC, cp_int_cntl_reg, tmp);
+			}
+		}
+	}
+}
+
+static int gfx_v10_0_init_csb(struct amdgpu_device *adev)
+{
+	adev->gfx.rlc.funcs->get_csb_buffer(adev, adev->gfx.rlc.cs_ptr);
+
+	/* csib */
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 1, 2)) {
+		WREG32_SOC15_RLC(GC, 0, mmRLC_CSIB_ADDR_HI,
+				adev->gfx.rlc.clear_state_gpu_addr >> 32);
+		WREG32_SOC15_RLC(GC, 0, mmRLC_CSIB_ADDR_LO,
+				adev->gfx.rlc.clear_state_gpu_addr & 0xfffffffc);
+		WREG32_SOC15_RLC(GC, 0, mmRLC_CSIB_LENGTH, adev->gfx.rlc.clear_state_size);
+	} else {
+		WREG32_SOC15(GC, 0, mmRLC_CSIB_ADDR_HI,
+				adev->gfx.rlc.clear_state_gpu_addr >> 32);
+		WREG32_SOC15(GC, 0, mmRLC_CSIB_ADDR_LO,
+				adev->gfx.rlc.clear_state_gpu_addr & 0xfffffffc);
+		WREG32_SOC15(GC, 0, mmRLC_CSIB_LENGTH, adev->gfx.rlc.clear_state_size);
+	}
+	return 0;
+}
+
+static void gfx_v10_0_rlc_stop(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SOC15(GC, 0, mmRLC_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, RLC_CNTL, RLC_ENABLE_F32, 0);
+	WREG32_SOC15(GC, 0, mmRLC_CNTL, tmp);
+}
+
+static void gfx_v10_0_rlc_reset(struct amdgpu_device *adev)
+{
+	WREG32_FIELD15(GC, 0, GRBM_SOFT_RESET, SOFT_RESET_RLC, 1);
+	udelay(50);
+	WREG32_FIELD15(GC, 0, GRBM_SOFT_RESET, SOFT_RESET_RLC, 0);
+	udelay(50);
+}
+
+static void gfx_v10_0_rlc_smu_handshake_cntl(struct amdgpu_device *adev,
+					     bool enable)
+{
+	uint32_t rlc_pg_cntl;
+
+	rlc_pg_cntl = RREG32_SOC15(GC, 0, mmRLC_PG_CNTL);
+
+	if (!enable) {
+		/* RLC_PG_CNTL[23] = 0 (default)
+		 * RLC will wait for handshake acks with SMU
+		 * GFXOFF will be enabled
+		 * RLC_PG_CNTL[23] = 1
+		 * RLC will not issue any message to SMU
+		 * hence no handshake between SMU & RLC
+		 * GFXOFF will be disabled
+		 */
+		rlc_pg_cntl |= 0x800000;
+	} else
+		rlc_pg_cntl &= ~0x800000;
+	WREG32_SOC15(GC, 0, mmRLC_PG_CNTL, rlc_pg_cntl);
+}
+
+static void gfx_v10_0_rlc_start(struct amdgpu_device *adev)
+{
+	/*
+	 * TODO: enable rlc & smu handshake until smu
+	 * and gfxoff feature works as expected
+	 */
+	if (!(amdgpu_pp_feature_mask & PP_GFXOFF_MASK))
+		gfx_v10_0_rlc_smu_handshake_cntl(adev, false);
+
+	WREG32_FIELD15(GC, 0, RLC_CNTL, RLC_ENABLE_F32, 1);
+	udelay(50);
+}
+
+static void gfx_v10_0_rlc_enable_srm(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* enable Save Restore Machine */
+	tmp = RREG32_SOC15(GC, 0, mmRLC_SRM_CNTL);
+	tmp |= RLC_SRM_CNTL__AUTO_INCR_ADDR_MASK;
+	tmp |= RLC_SRM_CNTL__SRM_ENABLE_MASK;
+	WREG32_SOC15(GC, 0, mmRLC_SRM_CNTL, tmp);
+}
+
+static int gfx_v10_0_rlc_load_microcode(struct amdgpu_device *adev)
+{
+	const struct rlc_firmware_header_v2_0 *hdr;
+	const __le32 *fw_data;
+	unsigned int i, fw_size;
+
+	if (!adev->gfx.rlc_fw)
+		return -EINVAL;
+
+	hdr = (const struct rlc_firmware_header_v2_0 *)adev->gfx.rlc_fw->data;
+	amdgpu_ucode_print_rlc_hdr(&hdr->header);
+
+	fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+			   le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+
+	WREG32_SOC15(GC, 0, mmRLC_GPM_UCODE_ADDR,
+		     RLCG_UCODE_LOADING_START_ADDRESS);
+
+	for (i = 0; i < fw_size; i++)
+		WREG32_SOC15(GC, 0, mmRLC_GPM_UCODE_DATA,
+			     le32_to_cpup(fw_data++));
+
+	WREG32_SOC15(GC, 0, mmRLC_GPM_UCODE_ADDR, adev->gfx.rlc_fw_version);
+
+	return 0;
+}
+
+static int gfx_v10_0_rlc_resume(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP &&
+		adev->psp.autoload_supported) {
+
+		r = gfx_v10_0_wait_for_rlc_autoload_complete(adev);
+		if (r)
+			return r;
+
+		gfx_v10_0_init_csb(adev);
+
+		gfx_v10_0_update_spm_vmid_internal(adev, 0xf);
+
+		if (!amdgpu_sriov_vf(adev)) /* enable RLC SRM */
+			gfx_v10_0_rlc_enable_srm(adev);
+	} else {
+		if (amdgpu_sriov_vf(adev)) {
+			gfx_v10_0_init_csb(adev);
+			return 0;
+		}
+
+		adev->gfx.rlc.funcs->stop(adev);
+
+		/* disable CG */
+		WREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL, 0);
+
+		/* disable PG */
+		WREG32_SOC15(GC, 0, mmRLC_PG_CNTL, 0);
+
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+			/* legacy rlc firmware loading */
+			r = gfx_v10_0_rlc_load_microcode(adev);
+			if (r)
+				return r;
+		} else if (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO) {
+			/* rlc backdoor autoload firmware */
+			r = gfx_v10_0_rlc_backdoor_autoload_enable(adev);
+			if (r)
+				return r;
+		}
+
+		gfx_v10_0_init_csb(adev);
+
+		gfx_v10_0_update_spm_vmid_internal(adev, 0xf);
+
+		adev->gfx.rlc.funcs->start(adev);
+
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO) {
+			r = gfx_v10_0_wait_for_rlc_autoload_complete(adev);
+			if (r)
+				return r;
+		}
+	}
+
+	return 0;
+}
+
+static struct {
+	FIRMWARE_ID	id;
+	unsigned int	offset;
+	unsigned int	size;
+} rlc_autoload_info[FIRMWARE_ID_MAX];
+
+static int gfx_v10_0_parse_rlc_toc(struct amdgpu_device *adev)
+{
+	int ret;
+	RLC_TABLE_OF_CONTENT *rlc_toc;
+
+	ret = amdgpu_bo_create_reserved(adev, adev->psp.toc.size_bytes, PAGE_SIZE,
+					AMDGPU_GEM_DOMAIN_GTT,
+					&adev->gfx.rlc.rlc_toc_bo,
+					&adev->gfx.rlc.rlc_toc_gpu_addr,
+					(void **)&adev->gfx.rlc.rlc_toc_buf);
+	if (ret) {
+		dev_err(adev->dev, "(%d) failed to create rlc toc bo\n", ret);
+		return ret;
+	}
+
+	/* Copy toc from psp sos fw to rlc toc buffer */
+	memcpy(adev->gfx.rlc.rlc_toc_buf, adev->psp.toc.start_addr, adev->psp.toc.size_bytes);
+
+	rlc_toc = (RLC_TABLE_OF_CONTENT *)adev->gfx.rlc.rlc_toc_buf;
+	while (rlc_toc && (rlc_toc->id > FIRMWARE_ID_INVALID) &&
+		(rlc_toc->id < FIRMWARE_ID_MAX)) {
+		if ((rlc_toc->id >= FIRMWARE_ID_CP_CE) &&
+		    (rlc_toc->id <= FIRMWARE_ID_CP_MES)) {
+			/* Offset needs 4KB alignment */
+			rlc_toc->offset = ALIGN(rlc_toc->offset * 4, PAGE_SIZE);
+		}
+
+		rlc_autoload_info[rlc_toc->id].id = rlc_toc->id;
+		rlc_autoload_info[rlc_toc->id].offset = rlc_toc->offset * 4;
+		rlc_autoload_info[rlc_toc->id].size = rlc_toc->size * 4;
+
+		rlc_toc++;
+	}
+
+	return 0;
+}
+
+static uint32_t gfx_v10_0_calc_toc_total_size(struct amdgpu_device *adev)
+{
+	uint32_t total_size = 0;
+	FIRMWARE_ID id;
+	int ret;
+
+	ret = gfx_v10_0_parse_rlc_toc(adev);
+	if (ret) {
+		dev_err(adev->dev, "failed to parse rlc toc\n");
+		return 0;
+	}
+
+	for (id = FIRMWARE_ID_RLC_G_UCODE; id < FIRMWARE_ID_MAX; id++)
+		total_size += rlc_autoload_info[id].size;
+
+	/* In case the offset in rlc toc ucode is aligned */
+	if (total_size < rlc_autoload_info[FIRMWARE_ID_MAX-1].offset)
+		total_size = rlc_autoload_info[FIRMWARE_ID_MAX-1].offset +
+				rlc_autoload_info[FIRMWARE_ID_MAX-1].size;
+
+	return total_size;
+}
+
+static int gfx_v10_0_rlc_backdoor_autoload_buffer_init(struct amdgpu_device *adev)
+{
+	int r;
+	uint32_t total_size;
+
+	total_size = gfx_v10_0_calc_toc_total_size(adev);
+
+	r = amdgpu_bo_create_reserved(adev, total_size, PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.rlc.rlc_autoload_bo,
+				      &adev->gfx.rlc.rlc_autoload_gpu_addr,
+				      (void **)&adev->gfx.rlc.rlc_autoload_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create fw autoload bo\n", r);
+		return r;
+	}
+
+	return 0;
+}
+
+static void gfx_v10_0_rlc_backdoor_autoload_buffer_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.rlc_toc_bo,
+			      &adev->gfx.rlc.rlc_toc_gpu_addr,
+			      (void **)&adev->gfx.rlc.rlc_toc_buf);
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.rlc_autoload_bo,
+			      &adev->gfx.rlc.rlc_autoload_gpu_addr,
+			      (void **)&adev->gfx.rlc.rlc_autoload_ptr);
+}
+
+static void gfx_v10_0_rlc_backdoor_autoload_copy_ucode(struct amdgpu_device *adev,
+						       FIRMWARE_ID id,
+						       const void *fw_data,
+						       uint32_t fw_size)
+{
+	uint32_t toc_offset;
+	uint32_t toc_fw_size;
+	char *ptr = adev->gfx.rlc.rlc_autoload_ptr;
+
+	if (id <= FIRMWARE_ID_INVALID || id >= FIRMWARE_ID_MAX)
+		return;
+
+	toc_offset = rlc_autoload_info[id].offset;
+	toc_fw_size = rlc_autoload_info[id].size;
+
+	if (fw_size == 0)
+		fw_size = toc_fw_size;
+
+	if (fw_size > toc_fw_size)
+		fw_size = toc_fw_size;
+
+	memcpy(ptr + toc_offset, fw_data, fw_size);
+
+	if (fw_size < toc_fw_size)
+		memset(ptr + toc_offset + fw_size, 0, toc_fw_size - fw_size);
+}
+
+static void gfx_v10_0_rlc_backdoor_autoload_copy_toc_ucode(struct amdgpu_device *adev)
+{
+	void *data;
+	uint32_t size;
+
+	data = adev->gfx.rlc.rlc_toc_buf;
+	size = rlc_autoload_info[FIRMWARE_ID_RLC_TOC].size;
+
+	gfx_v10_0_rlc_backdoor_autoload_copy_ucode(adev,
+						   FIRMWARE_ID_RLC_TOC,
+						   data, size);
+}
+
+static void gfx_v10_0_rlc_backdoor_autoload_copy_gfx_ucode(struct amdgpu_device *adev)
+{
+	const __le32 *fw_data;
+	uint32_t fw_size;
+	const struct gfx_firmware_header_v1_0 *cp_hdr;
+	const struct rlc_firmware_header_v2_0 *rlc_hdr;
+
+	/* pfp ucode */
+	cp_hdr = (const struct gfx_firmware_header_v1_0 *)
+		adev->gfx.pfp_fw->data;
+	fw_data = (const __le32 *)(adev->gfx.pfp_fw->data +
+		le32_to_cpu(cp_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(cp_hdr->header.ucode_size_bytes);
+	gfx_v10_0_rlc_backdoor_autoload_copy_ucode(adev,
+						   FIRMWARE_ID_CP_PFP,
+						   fw_data, fw_size);
+
+	/* ce ucode */
+	cp_hdr = (const struct gfx_firmware_header_v1_0 *)
+		adev->gfx.ce_fw->data;
+	fw_data = (const __le32 *)(adev->gfx.ce_fw->data +
+		le32_to_cpu(cp_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(cp_hdr->header.ucode_size_bytes);
+	gfx_v10_0_rlc_backdoor_autoload_copy_ucode(adev,
+						   FIRMWARE_ID_CP_CE,
+						   fw_data, fw_size);
+
+	/* me ucode */
+	cp_hdr = (const struct gfx_firmware_header_v1_0 *)
+		adev->gfx.me_fw->data;
+	fw_data = (const __le32 *)(adev->gfx.me_fw->data +
+		le32_to_cpu(cp_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(cp_hdr->header.ucode_size_bytes);
+	gfx_v10_0_rlc_backdoor_autoload_copy_ucode(adev,
+						   FIRMWARE_ID_CP_ME,
+						   fw_data, fw_size);
+
+	/* rlc ucode */
+	rlc_hdr = (const struct rlc_firmware_header_v2_0 *)
+		adev->gfx.rlc_fw->data;
+	fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+		le32_to_cpu(rlc_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(rlc_hdr->header.ucode_size_bytes);
+	gfx_v10_0_rlc_backdoor_autoload_copy_ucode(adev,
+						   FIRMWARE_ID_RLC_G_UCODE,
+						   fw_data, fw_size);
+
+	/* mec1 ucode */
+	cp_hdr = (const struct gfx_firmware_header_v1_0 *)
+		adev->gfx.mec_fw->data;
+	fw_data = (const __le32 *) (adev->gfx.mec_fw->data +
+		le32_to_cpu(cp_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(cp_hdr->header.ucode_size_bytes) -
+		cp_hdr->jt_size * 4;
+	gfx_v10_0_rlc_backdoor_autoload_copy_ucode(adev,
+						   FIRMWARE_ID_CP_MEC,
+						   fw_data, fw_size);
+	/* mec2 ucode is not necessary if mec2 ucode is same as mec1 */
+}
+
+/* Temporarily put sdma part here */
+static void gfx_v10_0_rlc_backdoor_autoload_copy_sdma_ucode(struct amdgpu_device *adev)
+{
+	const __le32 *fw_data;
+	uint32_t fw_size;
+	const struct sdma_firmware_header_v1_0 *sdma_hdr;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		sdma_hdr = (const struct sdma_firmware_header_v1_0 *)
+			adev->sdma.instance[i].fw->data;
+		fw_data = (const __le32 *) (adev->sdma.instance[i].fw->data +
+			le32_to_cpu(sdma_hdr->header.ucode_array_offset_bytes));
+		fw_size = le32_to_cpu(sdma_hdr->header.ucode_size_bytes);
+
+		if (i == 0) {
+			gfx_v10_0_rlc_backdoor_autoload_copy_ucode(adev,
+				FIRMWARE_ID_SDMA0_UCODE, fw_data, fw_size);
+			gfx_v10_0_rlc_backdoor_autoload_copy_ucode(adev,
+				FIRMWARE_ID_SDMA0_JT,
+				(uint32_t *)fw_data +
+				sdma_hdr->jt_offset,
+				sdma_hdr->jt_size * 4);
+		} else if (i == 1) {
+			gfx_v10_0_rlc_backdoor_autoload_copy_ucode(adev,
+				FIRMWARE_ID_SDMA1_UCODE, fw_data, fw_size);
+			gfx_v10_0_rlc_backdoor_autoload_copy_ucode(adev,
+				FIRMWARE_ID_SDMA1_JT,
+				(uint32_t *)fw_data +
+				sdma_hdr->jt_offset,
+				sdma_hdr->jt_size * 4);
+		}
+	}
+}
+
+static int gfx_v10_0_rlc_backdoor_autoload_enable(struct amdgpu_device *adev)
+{
+	uint32_t rlc_g_offset, rlc_g_size, tmp;
+	uint64_t gpu_addr;
+
+	gfx_v10_0_rlc_backdoor_autoload_copy_toc_ucode(adev);
+	gfx_v10_0_rlc_backdoor_autoload_copy_sdma_ucode(adev);
+	gfx_v10_0_rlc_backdoor_autoload_copy_gfx_ucode(adev);
+
+	rlc_g_offset = rlc_autoload_info[FIRMWARE_ID_RLC_G_UCODE].offset;
+	rlc_g_size = rlc_autoload_info[FIRMWARE_ID_RLC_G_UCODE].size;
+	gpu_addr = adev->gfx.rlc.rlc_autoload_gpu_addr + rlc_g_offset;
+
+	WREG32_SOC15(GC, 0, mmRLC_HYP_BOOTLOAD_ADDR_HI, upper_32_bits(gpu_addr));
+	WREG32_SOC15(GC, 0, mmRLC_HYP_BOOTLOAD_ADDR_LO, lower_32_bits(gpu_addr));
+	WREG32_SOC15(GC, 0, mmRLC_HYP_BOOTLOAD_SIZE, rlc_g_size);
+
+	tmp = RREG32_SOC15(GC, 0, mmRLC_HYP_RESET_VECTOR);
+	if (!(tmp & (RLC_HYP_RESET_VECTOR__COLD_BOOT_EXIT_MASK |
+		   RLC_HYP_RESET_VECTOR__VDDGFX_EXIT_MASK))) {
+		DRM_ERROR("Neither COLD_BOOT_EXIT nor VDDGFX_EXIT is set\n");
+		return -EINVAL;
+	}
+
+	tmp = RREG32_SOC15(GC, 0, mmRLC_CNTL);
+	if (tmp & RLC_CNTL__RLC_ENABLE_F32_MASK) {
+		DRM_ERROR("RLC ROM should halt itself\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int gfx_v10_0_rlc_backdoor_autoload_config_me_cache(struct amdgpu_device *adev)
+{
+	uint32_t usec_timeout = 50000;  /* wait for 50ms */
+	uint32_t tmp;
+	int i;
+	uint64_t addr;
+
+	/* Trigger an invalidation of the L1 instruction caches */
+	tmp = RREG32_SOC15(GC, 0, mmCP_ME_IC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_OP_CNTL, INVALIDATE_CACHE, 1);
+	WREG32_SOC15(GC, 0, mmCP_ME_IC_OP_CNTL, tmp);
+
+	/* Wait for invalidation complete */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, mmCP_ME_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_ME_IC_OP_CNTL,
+			INVALIDATE_CACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	/* Program me ucode address into intruction cache address register */
+	addr = adev->gfx.rlc.rlc_autoload_gpu_addr +
+		rlc_autoload_info[FIRMWARE_ID_CP_ME].offset;
+	WREG32_SOC15(GC, 0, mmCP_ME_IC_BASE_LO,
+			lower_32_bits(addr) & 0xFFFFF000);
+	WREG32_SOC15(GC, 0, mmCP_ME_IC_BASE_HI,
+			upper_32_bits(addr));
+
+	return 0;
+}
+
+static int gfx_v10_0_rlc_backdoor_autoload_config_ce_cache(struct amdgpu_device *adev)
+{
+	uint32_t usec_timeout = 50000;  /* wait for 50ms */
+	uint32_t tmp;
+	int i;
+	uint64_t addr;
+
+	/* Trigger an invalidation of the L1 instruction caches */
+	tmp = RREG32_SOC15(GC, 0, mmCP_CE_IC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_CE_IC_OP_CNTL, INVALIDATE_CACHE, 1);
+	WREG32_SOC15(GC, 0, mmCP_CE_IC_OP_CNTL, tmp);
+
+	/* Wait for invalidation complete */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, mmCP_CE_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_CE_IC_OP_CNTL,
+			INVALIDATE_CACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	/* Program ce ucode address into intruction cache address register */
+	addr = adev->gfx.rlc.rlc_autoload_gpu_addr +
+		rlc_autoload_info[FIRMWARE_ID_CP_CE].offset;
+	WREG32_SOC15(GC, 0, mmCP_CE_IC_BASE_LO,
+			lower_32_bits(addr) & 0xFFFFF000);
+	WREG32_SOC15(GC, 0, mmCP_CE_IC_BASE_HI,
+			upper_32_bits(addr));
+
+	return 0;
+}
+
+static int gfx_v10_0_rlc_backdoor_autoload_config_pfp_cache(struct amdgpu_device *adev)
+{
+	uint32_t usec_timeout = 50000;  /* wait for 50ms */
+	uint32_t tmp;
+	int i;
+	uint64_t addr;
+
+	/* Trigger an invalidation of the L1 instruction caches */
+	tmp = RREG32_SOC15(GC, 0, mmCP_PFP_IC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_OP_CNTL, INVALIDATE_CACHE, 1);
+	WREG32_SOC15(GC, 0, mmCP_PFP_IC_OP_CNTL, tmp);
+
+	/* Wait for invalidation complete */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, mmCP_PFP_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_PFP_IC_OP_CNTL,
+			INVALIDATE_CACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	/* Program pfp ucode address into intruction cache address register */
+	addr = adev->gfx.rlc.rlc_autoload_gpu_addr +
+		rlc_autoload_info[FIRMWARE_ID_CP_PFP].offset;
+	WREG32_SOC15(GC, 0, mmCP_PFP_IC_BASE_LO,
+			lower_32_bits(addr) & 0xFFFFF000);
+	WREG32_SOC15(GC, 0, mmCP_PFP_IC_BASE_HI,
+			upper_32_bits(addr));
+
+	return 0;
+}
+
+static int gfx_v10_0_rlc_backdoor_autoload_config_mec_cache(struct amdgpu_device *adev)
+{
+	uint32_t usec_timeout = 50000;  /* wait for 50ms */
+	uint32_t tmp;
+	int i;
+	uint64_t addr;
+
+	/* Trigger an invalidation of the L1 instruction caches */
+	tmp = RREG32_SOC15(GC, 0, mmCP_CPC_IC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_OP_CNTL, INVALIDATE_CACHE, 1);
+	WREG32_SOC15(GC, 0, mmCP_CPC_IC_OP_CNTL, tmp);
+
+	/* Wait for invalidation complete */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, mmCP_CPC_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_CPC_IC_OP_CNTL,
+			INVALIDATE_CACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	/* Program mec1 ucode address into intruction cache address register */
+	addr = adev->gfx.rlc.rlc_autoload_gpu_addr +
+		rlc_autoload_info[FIRMWARE_ID_CP_MEC].offset;
+	WREG32_SOC15(GC, 0, mmCP_CPC_IC_BASE_LO,
+			lower_32_bits(addr) & 0xFFFFF000);
+	WREG32_SOC15(GC, 0, mmCP_CPC_IC_BASE_HI,
+			upper_32_bits(addr));
+
+	return 0;
+}
+
+static int gfx_v10_0_wait_for_rlc_autoload_complete(struct amdgpu_device *adev)
+{
+	uint32_t cp_status;
+	uint32_t bootload_status;
+	int i, r;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		cp_status = RREG32_SOC15(GC, 0, mmCP_STAT);
+		bootload_status = RREG32_SOC15(GC, 0, mmRLC_RLCS_BOOTLOAD_STATUS);
+		if ((cp_status == 0) &&
+		    (REG_GET_FIELD(bootload_status,
+			RLC_RLCS_BOOTLOAD_STATUS, BOOTLOAD_COMPLETE) == 1)) {
+			break;
+		}
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout) {
+		dev_err(adev->dev, "rlc autoload: gc ucode autoload timeout\n");
+		return -ETIMEDOUT;
+	}
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO) {
+		r = gfx_v10_0_rlc_backdoor_autoload_config_me_cache(adev);
+		if (r)
+			return r;
+
+		r = gfx_v10_0_rlc_backdoor_autoload_config_ce_cache(adev);
+		if (r)
+			return r;
+
+		r = gfx_v10_0_rlc_backdoor_autoload_config_pfp_cache(adev);
+		if (r)
+			return r;
+
+		r = gfx_v10_0_rlc_backdoor_autoload_config_mec_cache(adev);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+static int gfx_v10_0_cp_gfx_enable(struct amdgpu_device *adev, bool enable)
+{
+	int i;
+	u32 tmp = RREG32_SOC15(GC, 0, mmCP_ME_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_HALT, enable ? 0 : 1);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_HALT, enable ? 0 : 1);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, CE_HALT, enable ? 0 : 1);
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 1, 2))
+		WREG32_SOC15_RLC(GC, 0, mmCP_ME_CNTL, tmp);
+	else
+		WREG32_SOC15(GC, 0, mmCP_ME_CNTL, tmp);
+
+	if (amdgpu_in_reset(adev) && !enable)
+		return 0;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (RREG32_SOC15(GC, 0, mmCP_STAT) == 0)
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout)
+		DRM_ERROR("failed to %s cp gfx\n", enable ? "unhalt" : "halt");
+
+	return 0;
+}
+
+static int gfx_v10_0_cp_gfx_load_pfp_microcode(struct amdgpu_device *adev)
+{
+	int r;
+	const struct gfx_firmware_header_v1_0 *pfp_hdr;
+	const __le32 *fw_data;
+	unsigned int i, fw_size;
+	uint32_t tmp;
+	uint32_t usec_timeout = 50000;  /* wait for 50ms */
+
+	pfp_hdr = (const struct gfx_firmware_header_v1_0 *)
+		adev->gfx.pfp_fw->data;
+
+	amdgpu_ucode_print_gfx_hdr(&pfp_hdr->header);
+
+	fw_data = (const __le32 *)(adev->gfx.pfp_fw->data +
+		le32_to_cpu(pfp_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(pfp_hdr->header.ucode_size_bytes);
+
+	r = amdgpu_bo_create_reserved(adev, pfp_hdr->header.ucode_size_bytes,
+				      PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.pfp.pfp_fw_obj,
+				      &adev->gfx.pfp.pfp_fw_gpu_addr,
+				      (void **)&adev->gfx.pfp.pfp_fw_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create pfp fw bo\n", r);
+		gfx_v10_0_pfp_fini(adev);
+		return r;
+	}
+
+	memcpy(adev->gfx.pfp.pfp_fw_ptr, fw_data, fw_size);
+
+	amdgpu_bo_kunmap(adev->gfx.pfp.pfp_fw_obj);
+	amdgpu_bo_unreserve(adev->gfx.pfp.pfp_fw_obj);
+
+	/* Trigger an invalidation of the L1 instruction caches */
+	tmp = RREG32_SOC15(GC, 0, mmCP_PFP_IC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_OP_CNTL, INVALIDATE_CACHE, 1);
+	WREG32_SOC15(GC, 0, mmCP_PFP_IC_OP_CNTL, tmp);
+
+	/* Wait for invalidation complete */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, mmCP_PFP_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_PFP_IC_OP_CNTL,
+			INVALIDATE_CACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	if (amdgpu_emu_mode == 1)
+		amdgpu_device_flush_hdp(adev, NULL);
+
+	tmp = RREG32_SOC15(GC, 0, mmCP_PFP_IC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_BASE_CNTL, CACHE_POLICY, 0);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_BASE_CNTL, EXE_DISABLE, 0);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_BASE_CNTL, ADDRESS_CLAMP, 1);
+	WREG32_SOC15(GC, 0, mmCP_PFP_IC_BASE_CNTL, tmp);
+	WREG32_SOC15(GC, 0, mmCP_PFP_IC_BASE_LO,
+		adev->gfx.pfp.pfp_fw_gpu_addr & 0xFFFFF000);
+	WREG32_SOC15(GC, 0, mmCP_PFP_IC_BASE_HI,
+		upper_32_bits(adev->gfx.pfp.pfp_fw_gpu_addr));
+
+	WREG32_SOC15(GC, 0, mmCP_HYP_PFP_UCODE_ADDR, 0);
+
+	for (i = 0; i < pfp_hdr->jt_size; i++)
+		WREG32_SOC15(GC, 0, mmCP_HYP_PFP_UCODE_DATA,
+			     le32_to_cpup(fw_data + pfp_hdr->jt_offset + i));
+
+	WREG32_SOC15(GC, 0, mmCP_HYP_PFP_UCODE_ADDR, adev->gfx.pfp_fw_version);
+
+	return 0;
+}
+
+static int gfx_v10_0_cp_gfx_load_ce_microcode(struct amdgpu_device *adev)
+{
+	int r;
+	const struct gfx_firmware_header_v1_0 *ce_hdr;
+	const __le32 *fw_data;
+	unsigned int i, fw_size;
+	uint32_t tmp;
+	uint32_t usec_timeout = 50000;  /* wait for 50ms */
+
+	ce_hdr = (const struct gfx_firmware_header_v1_0 *)
+		adev->gfx.ce_fw->data;
+
+	amdgpu_ucode_print_gfx_hdr(&ce_hdr->header);
+
+	fw_data = (const __le32 *)(adev->gfx.ce_fw->data +
+		le32_to_cpu(ce_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(ce_hdr->header.ucode_size_bytes);
+
+	r = amdgpu_bo_create_reserved(adev, ce_hdr->header.ucode_size_bytes,
+				      PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.ce.ce_fw_obj,
+				      &adev->gfx.ce.ce_fw_gpu_addr,
+				      (void **)&adev->gfx.ce.ce_fw_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create ce fw bo\n", r);
+		gfx_v10_0_ce_fini(adev);
+		return r;
+	}
+
+	memcpy(adev->gfx.ce.ce_fw_ptr, fw_data, fw_size);
+
+	amdgpu_bo_kunmap(adev->gfx.ce.ce_fw_obj);
+	amdgpu_bo_unreserve(adev->gfx.ce.ce_fw_obj);
+
+	/* Trigger an invalidation of the L1 instruction caches */
+	tmp = RREG32_SOC15(GC, 0, mmCP_CE_IC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_CE_IC_OP_CNTL, INVALIDATE_CACHE, 1);
+	WREG32_SOC15(GC, 0, mmCP_CE_IC_OP_CNTL, tmp);
+
+	/* Wait for invalidation complete */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, mmCP_CE_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_CE_IC_OP_CNTL,
+			INVALIDATE_CACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	if (amdgpu_emu_mode == 1)
+		amdgpu_device_flush_hdp(adev, NULL);
+
+	tmp = RREG32_SOC15(GC, 0, mmCP_CE_IC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_CE_IC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_CE_IC_BASE_CNTL, CACHE_POLICY, 0);
+	tmp = REG_SET_FIELD(tmp, CP_CE_IC_BASE_CNTL, EXE_DISABLE, 0);
+	tmp = REG_SET_FIELD(tmp, CP_CE_IC_BASE_CNTL, ADDRESS_CLAMP, 1);
+	WREG32_SOC15(GC, 0, mmCP_CE_IC_BASE_LO,
+		adev->gfx.ce.ce_fw_gpu_addr & 0xFFFFF000);
+	WREG32_SOC15(GC, 0, mmCP_CE_IC_BASE_HI,
+		upper_32_bits(adev->gfx.ce.ce_fw_gpu_addr));
+
+	WREG32_SOC15(GC, 0, mmCP_HYP_CE_UCODE_ADDR, 0);
+
+	for (i = 0; i < ce_hdr->jt_size; i++)
+		WREG32_SOC15(GC, 0, mmCP_HYP_CE_UCODE_DATA,
+			     le32_to_cpup(fw_data + ce_hdr->jt_offset + i));
+
+	WREG32_SOC15(GC, 0, mmCP_HYP_CE_UCODE_ADDR, adev->gfx.ce_fw_version);
+
+	return 0;
+}
+
+static int gfx_v10_0_cp_gfx_load_me_microcode(struct amdgpu_device *adev)
+{
+	int r;
+	const struct gfx_firmware_header_v1_0 *me_hdr;
+	const __le32 *fw_data;
+	unsigned int i, fw_size;
+	uint32_t tmp;
+	uint32_t usec_timeout = 50000;  /* wait for 50ms */
+
+	me_hdr = (const struct gfx_firmware_header_v1_0 *)
+		adev->gfx.me_fw->data;
+
+	amdgpu_ucode_print_gfx_hdr(&me_hdr->header);
+
+	fw_data = (const __le32 *)(adev->gfx.me_fw->data +
+		le32_to_cpu(me_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(me_hdr->header.ucode_size_bytes);
+
+	r = amdgpu_bo_create_reserved(adev, me_hdr->header.ucode_size_bytes,
+				      PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.me.me_fw_obj,
+				      &adev->gfx.me.me_fw_gpu_addr,
+				      (void **)&adev->gfx.me.me_fw_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create me fw bo\n", r);
+		gfx_v10_0_me_fini(adev);
+		return r;
+	}
+
+	memcpy(adev->gfx.me.me_fw_ptr, fw_data, fw_size);
+
+	amdgpu_bo_kunmap(adev->gfx.me.me_fw_obj);
+	amdgpu_bo_unreserve(adev->gfx.me.me_fw_obj);
+
+	/* Trigger an invalidation of the L1 instruction caches */
+	tmp = RREG32_SOC15(GC, 0, mmCP_ME_IC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_OP_CNTL, INVALIDATE_CACHE, 1);
+	WREG32_SOC15(GC, 0, mmCP_ME_IC_OP_CNTL, tmp);
+
+	/* Wait for invalidation complete */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, mmCP_ME_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_ME_IC_OP_CNTL,
+			INVALIDATE_CACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	if (amdgpu_emu_mode == 1)
+		amdgpu_device_flush_hdp(adev, NULL);
+
+	tmp = RREG32_SOC15(GC, 0, mmCP_ME_IC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_BASE_CNTL, CACHE_POLICY, 0);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_BASE_CNTL, EXE_DISABLE, 0);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_BASE_CNTL, ADDRESS_CLAMP, 1);
+	WREG32_SOC15(GC, 0, mmCP_ME_IC_BASE_LO,
+		adev->gfx.me.me_fw_gpu_addr & 0xFFFFF000);
+	WREG32_SOC15(GC, 0, mmCP_ME_IC_BASE_HI,
+		upper_32_bits(adev->gfx.me.me_fw_gpu_addr));
+
+	WREG32_SOC15(GC, 0, mmCP_HYP_ME_UCODE_ADDR, 0);
+
+	for (i = 0; i < me_hdr->jt_size; i++)
+		WREG32_SOC15(GC, 0, mmCP_HYP_ME_UCODE_DATA,
+			     le32_to_cpup(fw_data + me_hdr->jt_offset + i));
+
+	WREG32_SOC15(GC, 0, mmCP_HYP_ME_UCODE_ADDR, adev->gfx.me_fw_version);
+
+	return 0;
+}
+
+static int gfx_v10_0_cp_gfx_load_microcode(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (!adev->gfx.me_fw || !adev->gfx.pfp_fw || !adev->gfx.ce_fw)
+		return -EINVAL;
+
+	gfx_v10_0_cp_gfx_enable(adev, false);
+
+	r = gfx_v10_0_cp_gfx_load_pfp_microcode(adev);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to load pfp fw\n", r);
+		return r;
+	}
+
+	r = gfx_v10_0_cp_gfx_load_ce_microcode(adev);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to load ce fw\n", r);
+		return r;
+	}
+
+	r = gfx_v10_0_cp_gfx_load_me_microcode(adev);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to load me fw\n", r);
+		return r;
+	}
+
+	return 0;
+}
+
+static int gfx_v10_0_cp_gfx_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+	int r, i;
+	int ctx_reg_offset;
+
+	/* init the CP */
+	WREG32_SOC15(GC, 0, mmCP_MAX_CONTEXT,
+		     adev->gfx.config.max_hw_contexts - 1);
+	WREG32_SOC15(GC, 0, mmCP_DEVICE_ID, 1);
+
+	gfx_v10_0_cp_gfx_enable(adev, true);
+
+	ring = &adev->gfx.gfx_ring[0];
+	r = amdgpu_ring_alloc(ring, gfx_v10_0_get_csb_size(adev) + 4);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring (%d).\n", r);
+		return r;
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	amdgpu_ring_write(ring, PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	amdgpu_ring_write(ring, 0x80000000);
+	amdgpu_ring_write(ring, 0x80000000);
+
+	for (sect = gfx10_cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT) {
+				amdgpu_ring_write(ring,
+						  PACKET3(PACKET3_SET_CONTEXT_REG,
+							  ext->reg_count));
+				amdgpu_ring_write(ring, ext->reg_index -
+						  PACKET3_SET_CONTEXT_REG_START);
+				for (i = 0; i < ext->reg_count; i++)
+					amdgpu_ring_write(ring, ext->extent[i]);
+			}
+		}
+	}
+
+	ctx_reg_offset =
+		SOC15_REG_OFFSET(GC, 0, mmPA_SC_TILE_STEERING_OVERRIDE) - PACKET3_SET_CONTEXT_REG_START;
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONTEXT_REG, 1));
+	amdgpu_ring_write(ring, ctx_reg_offset);
+	amdgpu_ring_write(ring, adev->gfx.config.pa_sc_tile_steering_override);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	amdgpu_ring_write(ring, PACKET3_PREAMBLE_END_CLEAR_STATE);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CLEAR_STATE, 0));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_BASE, 2));
+	amdgpu_ring_write(ring, PACKET3_BASE_INDEX(CE_PARTITION_BASE));
+	amdgpu_ring_write(ring, 0x8000);
+	amdgpu_ring_write(ring, 0x8000);
+
+	amdgpu_ring_commit(ring);
+
+	/* submit cs packet to copy state 0 to next available state */
+	if (adev->gfx.num_gfx_rings > 1) {
+		/* maximum supported gfx ring is 2 */
+		ring = &adev->gfx.gfx_ring[1];
+		r = amdgpu_ring_alloc(ring, 2);
+		if (r) {
+			DRM_ERROR("amdgpu: cp failed to lock ring (%d).\n", r);
+			return r;
+		}
+
+		amdgpu_ring_write(ring, PACKET3(PACKET3_CLEAR_STATE, 0));
+		amdgpu_ring_write(ring, 0);
+
+		amdgpu_ring_commit(ring);
+	}
+	return 0;
+}
+
+static void gfx_v10_0_cp_gfx_switch_pipe(struct amdgpu_device *adev,
+					 CP_PIPE_ID pipe)
+{
+	u32 tmp;
+
+	tmp = RREG32_SOC15(GC, 0, mmGRBM_GFX_CNTL);
+	tmp = REG_SET_FIELD(tmp, GRBM_GFX_CNTL, PIPEID, pipe);
+
+	WREG32_SOC15(GC, 0, mmGRBM_GFX_CNTL, tmp);
+}
+
+static void gfx_v10_0_cp_gfx_set_doorbell(struct amdgpu_device *adev,
+					  struct amdgpu_ring *ring)
+{
+	u32 tmp;
+
+	if (!amdgpu_async_gfx_ring) {
+		tmp = RREG32_SOC15(GC, 0, mmCP_RB_DOORBELL_CONTROL);
+		if (ring->use_doorbell) {
+			tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+						DOORBELL_OFFSET, ring->doorbell_index);
+			tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+						DOORBELL_EN, 1);
+		} else {
+			tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+						DOORBELL_EN, 0);
+		}
+		WREG32_SOC15(GC, 0, mmCP_RB_DOORBELL_CONTROL, tmp);
+	}
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+	case IP_VERSION(10, 3, 6):
+	case IP_VERSION(10, 3, 3):
+	case IP_VERSION(10, 3, 7):
+		tmp = REG_SET_FIELD(0, CP_RB_DOORBELL_RANGE_LOWER,
+				    DOORBELL_RANGE_LOWER_Sienna_Cichlid, ring->doorbell_index);
+		WREG32_SOC15(GC, 0, mmCP_RB_DOORBELL_RANGE_LOWER, tmp);
+
+		WREG32_SOC15(GC, 0, mmCP_RB_DOORBELL_RANGE_UPPER,
+			     CP_RB_DOORBELL_RANGE_UPPER__DOORBELL_RANGE_UPPER_Sienna_Cichlid_MASK);
+		break;
+	default:
+		tmp = REG_SET_FIELD(0, CP_RB_DOORBELL_RANGE_LOWER,
+				    DOORBELL_RANGE_LOWER, ring->doorbell_index);
+		WREG32_SOC15(GC, 0, mmCP_RB_DOORBELL_RANGE_LOWER, tmp);
+
+		WREG32_SOC15(GC, 0, mmCP_RB_DOORBELL_RANGE_UPPER,
+			     CP_RB_DOORBELL_RANGE_UPPER__DOORBELL_RANGE_UPPER_MASK);
+		break;
+	}
+}
+
+static int gfx_v10_0_cp_gfx_resume(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	u32 tmp;
+	u32 rb_bufsz;
+	u64 rb_addr, rptr_addr, wptr_gpu_addr;
+
+	/* Set the write pointer delay */
+	WREG32_SOC15(GC, 0, mmCP_RB_WPTR_DELAY, 0);
+
+	/* set the RB to use vmid 0 */
+	WREG32_SOC15(GC, 0, mmCP_RB_VMID, 0);
+
+	/* Init gfx ring 0 for pipe 0 */
+	mutex_lock(&adev->srbm_mutex);
+	gfx_v10_0_cp_gfx_switch_pipe(adev, PIPE_ID0);
+
+	/* Set ring buffer size */
+	ring = &adev->gfx.gfx_ring[0];
+	rb_bufsz = order_base_2(ring->ring_size / 8);
+	tmp = REG_SET_FIELD(0, CP_RB0_CNTL, RB_BUFSZ, rb_bufsz);
+	tmp = REG_SET_FIELD(tmp, CP_RB0_CNTL, RB_BLKSZ, rb_bufsz - 2);
+#ifdef __BIG_ENDIAN
+	tmp = REG_SET_FIELD(tmp, CP_RB0_CNTL, BUF_SWAP, 1);
+#endif
+	WREG32_SOC15(GC, 0, mmCP_RB0_CNTL, tmp);
+
+	/* Initialize the ring buffer's write pointers */
+	ring->wptr = 0;
+	WREG32_SOC15(GC, 0, mmCP_RB0_WPTR, lower_32_bits(ring->wptr));
+	WREG32_SOC15(GC, 0, mmCP_RB0_WPTR_HI, upper_32_bits(ring->wptr));
+
+	/* set the wb address whether it's enabled or not */
+	rptr_addr = ring->rptr_gpu_addr;
+	WREG32_SOC15(GC, 0, mmCP_RB0_RPTR_ADDR, lower_32_bits(rptr_addr));
+	WREG32_SOC15(GC, 0, mmCP_RB0_RPTR_ADDR_HI, upper_32_bits(rptr_addr) &
+		     CP_RB_RPTR_ADDR_HI__RB_RPTR_ADDR_HI_MASK);
+
+	wptr_gpu_addr = ring->wptr_gpu_addr;
+	WREG32_SOC15(GC, 0, mmCP_RB_WPTR_POLL_ADDR_LO,
+		     lower_32_bits(wptr_gpu_addr));
+	WREG32_SOC15(GC, 0, mmCP_RB_WPTR_POLL_ADDR_HI,
+		     upper_32_bits(wptr_gpu_addr));
+
+	mdelay(1);
+	WREG32_SOC15(GC, 0, mmCP_RB0_CNTL, tmp);
+
+	rb_addr = ring->gpu_addr >> 8;
+	WREG32_SOC15(GC, 0, mmCP_RB0_BASE, rb_addr);
+	WREG32_SOC15(GC, 0, mmCP_RB0_BASE_HI, upper_32_bits(rb_addr));
+
+	WREG32_SOC15(GC, 0, mmCP_RB_ACTIVE, 1);
+
+	gfx_v10_0_cp_gfx_set_doorbell(adev, ring);
+	mutex_unlock(&adev->srbm_mutex);
+
+	/* Init gfx ring 1 for pipe 1 */
+	if (adev->gfx.num_gfx_rings > 1) {
+		mutex_lock(&adev->srbm_mutex);
+		gfx_v10_0_cp_gfx_switch_pipe(adev, PIPE_ID1);
+		/* maximum supported gfx ring is 2 */
+		ring = &adev->gfx.gfx_ring[1];
+		rb_bufsz = order_base_2(ring->ring_size / 8);
+		tmp = REG_SET_FIELD(0, CP_RB1_CNTL, RB_BUFSZ, rb_bufsz);
+		tmp = REG_SET_FIELD(tmp, CP_RB1_CNTL, RB_BLKSZ, rb_bufsz - 2);
+		WREG32_SOC15(GC, 0, mmCP_RB1_CNTL, tmp);
+		/* Initialize the ring buffer's write pointers */
+		ring->wptr = 0;
+		WREG32_SOC15(GC, 0, mmCP_RB1_WPTR, lower_32_bits(ring->wptr));
+		WREG32_SOC15(GC, 0, mmCP_RB1_WPTR_HI, upper_32_bits(ring->wptr));
+		/* Set the wb address whether it's enabled or not */
+		rptr_addr = ring->rptr_gpu_addr;
+		WREG32_SOC15(GC, 0, mmCP_RB1_RPTR_ADDR, lower_32_bits(rptr_addr));
+		WREG32_SOC15(GC, 0, mmCP_RB1_RPTR_ADDR_HI, upper_32_bits(rptr_addr) &
+			     CP_RB1_RPTR_ADDR_HI__RB_RPTR_ADDR_HI_MASK);
+		wptr_gpu_addr = ring->wptr_gpu_addr;
+		WREG32_SOC15(GC, 0, mmCP_RB_WPTR_POLL_ADDR_LO,
+			     lower_32_bits(wptr_gpu_addr));
+		WREG32_SOC15(GC, 0, mmCP_RB_WPTR_POLL_ADDR_HI,
+			     upper_32_bits(wptr_gpu_addr));
+
+		mdelay(1);
+		WREG32_SOC15(GC, 0, mmCP_RB1_CNTL, tmp);
+
+		rb_addr = ring->gpu_addr >> 8;
+		WREG32_SOC15(GC, 0, mmCP_RB1_BASE, rb_addr);
+		WREG32_SOC15(GC, 0, mmCP_RB1_BASE_HI, upper_32_bits(rb_addr));
+		WREG32_SOC15(GC, 0, mmCP_RB1_ACTIVE, 1);
+
+		gfx_v10_0_cp_gfx_set_doorbell(adev, ring);
+		mutex_unlock(&adev->srbm_mutex);
+	}
+	/* Switch to pipe 0 */
+	mutex_lock(&adev->srbm_mutex);
+	gfx_v10_0_cp_gfx_switch_pipe(adev, PIPE_ID0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	/* start the ring */
+	gfx_v10_0_cp_gfx_start(adev);
+
+	return 0;
+}
+
+static void gfx_v10_0_cp_compute_enable(struct amdgpu_device *adev, bool enable)
+{
+	if (enable) {
+		switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+		case IP_VERSION(10, 3, 0):
+		case IP_VERSION(10, 3, 2):
+		case IP_VERSION(10, 3, 1):
+		case IP_VERSION(10, 3, 4):
+		case IP_VERSION(10, 3, 5):
+		case IP_VERSION(10, 3, 6):
+		case IP_VERSION(10, 3, 3):
+		case IP_VERSION(10, 3, 7):
+			WREG32_SOC15(GC, 0, mmCP_MEC_CNTL_Sienna_Cichlid, 0);
+			break;
+		default:
+			WREG32_SOC15(GC, 0, mmCP_MEC_CNTL, 0);
+			break;
+		}
+	} else {
+		switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+		case IP_VERSION(10, 3, 0):
+		case IP_VERSION(10, 3, 2):
+		case IP_VERSION(10, 3, 1):
+		case IP_VERSION(10, 3, 4):
+		case IP_VERSION(10, 3, 5):
+		case IP_VERSION(10, 3, 6):
+		case IP_VERSION(10, 3, 3):
+		case IP_VERSION(10, 3, 7):
+			WREG32_SOC15(GC, 0, mmCP_MEC_CNTL_Sienna_Cichlid,
+				     (CP_MEC_CNTL__MEC_ME1_HALT_MASK |
+				      CP_MEC_CNTL__MEC_ME2_HALT_MASK));
+			break;
+		default:
+			WREG32_SOC15(GC, 0, mmCP_MEC_CNTL,
+				     (CP_MEC_CNTL__MEC_ME1_HALT_MASK |
+				      CP_MEC_CNTL__MEC_ME2_HALT_MASK));
+			break;
+		}
+		adev->gfx.kiq[0].ring.sched.ready = false;
+	}
+	udelay(50);
+}
+
+static int gfx_v10_0_cp_compute_load_microcode(struct amdgpu_device *adev)
+{
+	const struct gfx_firmware_header_v1_0 *mec_hdr;
+	const __le32 *fw_data;
+	unsigned int i;
+	u32 tmp;
+	u32 usec_timeout = 50000; /* Wait for 50 ms */
+
+	if (!adev->gfx.mec_fw)
+		return -EINVAL;
+
+	gfx_v10_0_cp_compute_enable(adev, false);
+
+	mec_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+	amdgpu_ucode_print_gfx_hdr(&mec_hdr->header);
+
+	fw_data = (const __le32 *)
+		(adev->gfx.mec_fw->data +
+		 le32_to_cpu(mec_hdr->header.ucode_array_offset_bytes));
+
+	/* Trigger an invalidation of the L1 instruction caches */
+	tmp = RREG32_SOC15(GC, 0, mmCP_CPC_IC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_OP_CNTL, INVALIDATE_CACHE, 1);
+	WREG32_SOC15(GC, 0, mmCP_CPC_IC_OP_CNTL, tmp);
+
+	/* Wait for invalidation complete */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, mmCP_CPC_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_CPC_IC_OP_CNTL,
+				       INVALIDATE_CACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	if (amdgpu_emu_mode == 1)
+		amdgpu_device_flush_hdp(adev, NULL);
+
+	tmp = RREG32_SOC15(GC, 0, mmCP_CPC_IC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, CACHE_POLICY, 0);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, EXE_DISABLE, 0);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, ADDRESS_CLAMP, 1);
+	WREG32_SOC15(GC, 0, mmCP_CPC_IC_BASE_CNTL, tmp);
+
+	WREG32_SOC15(GC, 0, mmCP_CPC_IC_BASE_LO, adev->gfx.mec.mec_fw_gpu_addr &
+		     0xFFFFF000);
+	WREG32_SOC15(GC, 0, mmCP_CPC_IC_BASE_HI,
+		     upper_32_bits(adev->gfx.mec.mec_fw_gpu_addr));
+
+	/* MEC1 */
+	WREG32_SOC15(GC, 0, mmCP_MEC_ME1_UCODE_ADDR, 0);
+
+	for (i = 0; i < mec_hdr->jt_size; i++)
+		WREG32_SOC15(GC, 0, mmCP_MEC_ME1_UCODE_DATA,
+			     le32_to_cpup(fw_data + mec_hdr->jt_offset + i));
+
+	WREG32_SOC15(GC, 0, mmCP_MEC_ME1_UCODE_ADDR, adev->gfx.mec_fw_version);
+
+	/*
+	 * TODO: Loading MEC2 firmware is only necessary if MEC2 should run
+	 * different microcode than MEC1.
+	 */
+
+	return 0;
+}
+
+static void gfx_v10_0_kiq_setting(struct amdgpu_ring *ring)
+{
+	uint32_t tmp;
+	struct amdgpu_device *adev = ring->adev;
+
+	/* tell RLC which is KIQ queue */
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+	case IP_VERSION(10, 3, 6):
+	case IP_VERSION(10, 3, 3):
+	case IP_VERSION(10, 3, 7):
+		tmp = RREG32_SOC15(GC, 0, mmRLC_CP_SCHEDULERS_Sienna_Cichlid);
+		tmp &= 0xffffff00;
+		tmp |= (ring->me << 5) | (ring->pipe << 3) | (ring->queue);
+		WREG32_SOC15(GC, 0, mmRLC_CP_SCHEDULERS_Sienna_Cichlid, tmp | 0x80);
+		break;
+	default:
+		tmp = RREG32_SOC15(GC, 0, mmRLC_CP_SCHEDULERS);
+		tmp &= 0xffffff00;
+		tmp |= (ring->me << 5) | (ring->pipe << 3) | (ring->queue);
+		WREG32_SOC15(GC, 0, mmRLC_CP_SCHEDULERS, tmp | 0x80);
+		break;
+	}
+}
+
+static void gfx_v10_0_gfx_mqd_set_priority(struct amdgpu_device *adev,
+					   struct v10_gfx_mqd *mqd,
+					   struct amdgpu_mqd_prop *prop)
+{
+	bool priority = 0;
+	u32 tmp;
+
+	/* set up default queue priority level
+	 * 0x0 = low priority, 0x1 = high priority
+	 */
+	if (prop->hqd_pipe_priority == AMDGPU_GFX_PIPE_PRIO_HIGH)
+		priority = 1;
+
+	tmp = RREG32_SOC15(GC, 0, mmCP_GFX_HQD_QUEUE_PRIORITY);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_QUEUE_PRIORITY, PRIORITY_LEVEL, priority);
+	mqd->cp_gfx_hqd_queue_priority = tmp;
+}
+
+static int gfx_v10_0_gfx_mqd_init(struct amdgpu_device *adev, void *m,
+				  struct amdgpu_mqd_prop *prop)
+{
+	struct v10_gfx_mqd *mqd = m;
+	uint64_t hqd_gpu_addr, wb_gpu_addr;
+	uint32_t tmp;
+	uint32_t rb_bufsz;
+
+	/* set up gfx hqd wptr */
+	mqd->cp_gfx_hqd_wptr = 0;
+	mqd->cp_gfx_hqd_wptr_hi = 0;
+
+	/* set the pointer to the MQD */
+	mqd->cp_mqd_base_addr = prop->mqd_gpu_addr & 0xfffffffc;
+	mqd->cp_mqd_base_addr_hi = upper_32_bits(prop->mqd_gpu_addr);
+
+	/* set up mqd control */
+	tmp = RREG32_SOC15(GC, 0, mmCP_GFX_MQD_CONTROL);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_MQD_CONTROL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_MQD_CONTROL, PRIV_STATE, 1);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_MQD_CONTROL, CACHE_POLICY, 0);
+	mqd->cp_gfx_mqd_control = tmp;
+
+	/* set up gfx_hqd_vimd with 0x0 to indicate the ring buffer's vmid */
+	tmp = RREG32_SOC15(GC, 0, mmCP_GFX_HQD_VMID);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_VMID, VMID, 0);
+	mqd->cp_gfx_hqd_vmid = 0;
+
+	/* set up gfx queue priority */
+	gfx_v10_0_gfx_mqd_set_priority(adev, mqd, prop);
+
+	/* set up time quantum */
+	tmp = RREG32_SOC15(GC, 0, mmCP_GFX_HQD_QUANTUM);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_QUANTUM, QUANTUM_EN, 1);
+	mqd->cp_gfx_hqd_quantum = tmp;
+
+	/* set up gfx hqd base. this is similar as CP_RB_BASE */
+	hqd_gpu_addr = prop->hqd_base_gpu_addr >> 8;
+	mqd->cp_gfx_hqd_base = hqd_gpu_addr;
+	mqd->cp_gfx_hqd_base_hi = upper_32_bits(hqd_gpu_addr);
+
+	/* set up hqd_rptr_addr/_hi, similar as CP_RB_RPTR */
+	wb_gpu_addr = prop->rptr_gpu_addr;
+	mqd->cp_gfx_hqd_rptr_addr = wb_gpu_addr & 0xfffffffc;
+	mqd->cp_gfx_hqd_rptr_addr_hi =
+		upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	/* set up rb_wptr_poll addr */
+	wb_gpu_addr = prop->wptr_gpu_addr;
+	mqd->cp_rb_wptr_poll_addr_lo = wb_gpu_addr & 0xfffffffc;
+	mqd->cp_rb_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	/* set up the gfx_hqd_control, similar as CP_RB0_CNTL */
+	rb_bufsz = order_base_2(prop->queue_size / 4) - 1;
+	tmp = RREG32_SOC15(GC, 0, mmCP_GFX_HQD_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, RB_BUFSZ, rb_bufsz);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, RB_BLKSZ, rb_bufsz - 2);
+#ifdef __BIG_ENDIAN
+	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, BUF_SWAP, 1);
+#endif
+	mqd->cp_gfx_hqd_cntl = tmp;
+
+	/* set up cp_doorbell_control */
+	tmp = RREG32_SOC15(GC, 0, mmCP_RB_DOORBELL_CONTROL);
+	if (prop->use_doorbell) {
+		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+				    DOORBELL_OFFSET, prop->doorbell_index);
+		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+				    DOORBELL_EN, 1);
+	} else
+		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+				    DOORBELL_EN, 0);
+	mqd->cp_rb_doorbell_control = tmp;
+
+	/* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */
+	mqd->cp_gfx_hqd_rptr = RREG32_SOC15(GC, 0, mmCP_GFX_HQD_RPTR);
+
+	/* active the queue */
+	mqd->cp_gfx_hqd_active = 1;
+
+	return 0;
+}
+
+static int gfx_v10_0_kgq_init_queue(struct amdgpu_ring *ring, bool reset)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v10_gfx_mqd *mqd = ring->mqd_ptr;
+	int mqd_idx = ring - &adev->gfx.gfx_ring[0];
+
+	if (!reset && !amdgpu_in_reset(adev) && !adev->in_suspend) {
+		memset((void *)mqd, 0, sizeof(*mqd));
+		mutex_lock(&adev->srbm_mutex);
+		nv_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		amdgpu_ring_init_mqd(ring);
+
+		/*
+		 * if there are 2 gfx rings, set the lower doorbell
+		 * range of the first ring, otherwise the range of
+		 * the second ring will override the first ring
+		 */
+		if (ring->doorbell_index == adev->doorbell_index.gfx_ring0 << 1)
+			gfx_v10_0_cp_gfx_set_doorbell(adev, ring);
+
+		nv_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+		if (adev->gfx.me.mqd_backup[mqd_idx])
+			memcpy_fromio(adev->gfx.me.mqd_backup[mqd_idx], mqd, sizeof(*mqd));
+	} else {
+		mutex_lock(&adev->srbm_mutex);
+		nv_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		if (ring->doorbell_index == adev->doorbell_index.gfx_ring0 << 1)
+			gfx_v10_0_cp_gfx_set_doorbell(adev, ring);
+
+		nv_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+		/* restore mqd with the backup copy */
+		if (adev->gfx.me.mqd_backup[mqd_idx])
+			memcpy_toio(mqd, adev->gfx.me.mqd_backup[mqd_idx], sizeof(*mqd));
+		/* reset the ring */
+		ring->wptr = 0;
+		*ring->wptr_cpu_addr = 0;
+		amdgpu_ring_clear_ring(ring);
+	}
+
+	return 0;
+}
+
+static int gfx_v10_0_cp_async_gfx_ring_resume(struct amdgpu_device *adev)
+{
+	int r, i;
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+		r = gfx_v10_0_kgq_init_queue(&adev->gfx.gfx_ring[i], false);
+		if (r)
+			return r;
+	}
+
+	r = amdgpu_gfx_enable_kgq(adev, 0);
+	if (r)
+		return r;
+
+	return gfx_v10_0_cp_gfx_start(adev);
+}
+
+static int gfx_v10_0_compute_mqd_init(struct amdgpu_device *adev, void *m,
+				      struct amdgpu_mqd_prop *prop)
+{
+	struct v10_compute_mqd *mqd = m;
+	uint64_t hqd_gpu_addr, wb_gpu_addr, eop_base_addr;
+	uint32_t tmp;
+
+	mqd->header = 0xC0310800;
+	mqd->compute_pipelinestat_enable = 0x00000001;
+	mqd->compute_static_thread_mgmt_se0 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se1 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se2 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se3 = 0xffffffff;
+	mqd->compute_misc_reserved = 0x00000003;
+
+	eop_base_addr = prop->eop_gpu_addr >> 8;
+	mqd->cp_hqd_eop_base_addr_lo = eop_base_addr;
+	mqd->cp_hqd_eop_base_addr_hi = upper_32_bits(eop_base_addr);
+
+	/* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */
+	tmp = RREG32_SOC15(GC, 0, mmCP_HQD_EOP_CONTROL);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_EOP_CONTROL, EOP_SIZE,
+			(order_base_2(GFX10_MEC_HPD_SIZE / 4) - 1));
+
+	mqd->cp_hqd_eop_control = tmp;
+
+	/* enable doorbell? */
+	tmp = RREG32_SOC15(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL);
+
+	if (prop->use_doorbell) {
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_OFFSET, prop->doorbell_index);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_EN, 1);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_SOURCE, 0);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_HIT, 0);
+	} else {
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_EN, 0);
+	}
+
+	mqd->cp_hqd_pq_doorbell_control = tmp;
+
+	/* disable the queue if it's active */
+	mqd->cp_hqd_dequeue_request = 0;
+	mqd->cp_hqd_pq_rptr = 0;
+	mqd->cp_hqd_pq_wptr_lo = 0;
+	mqd->cp_hqd_pq_wptr_hi = 0;
+
+	/* set the pointer to the MQD */
+	mqd->cp_mqd_base_addr_lo = prop->mqd_gpu_addr & 0xfffffffc;
+	mqd->cp_mqd_base_addr_hi = upper_32_bits(prop->mqd_gpu_addr);
+
+	/* set MQD vmid to 0 */
+	tmp = RREG32_SOC15(GC, 0, mmCP_MQD_CONTROL);
+	tmp = REG_SET_FIELD(tmp, CP_MQD_CONTROL, VMID, 0);
+	mqd->cp_mqd_control = tmp;
+
+	/* set the pointer to the HQD, this is similar CP_RB0_BASE/_HI */
+	hqd_gpu_addr = prop->hqd_base_gpu_addr >> 8;
+	mqd->cp_hqd_pq_base_lo = hqd_gpu_addr;
+	mqd->cp_hqd_pq_base_hi = upper_32_bits(hqd_gpu_addr);
+
+	/* set up the HQD, this is similar to CP_RB0_CNTL */
+	tmp = RREG32_SOC15(GC, 0, mmCP_HQD_PQ_CONTROL);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, QUEUE_SIZE,
+			    (order_base_2(prop->queue_size / 4) - 1));
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, RPTR_BLOCK_SIZE,
+			    (order_base_2(AMDGPU_GPU_PAGE_SIZE / 4) - 1));
+#ifdef __BIG_ENDIAN
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, ENDIAN_SWAP, 1);
+#endif
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, UNORD_DISPATCH, 1);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, TUNNEL_DISPATCH,
+			    prop->allow_tunneling);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, PRIV_STATE, 1);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, KMD_QUEUE, 1);
+	mqd->cp_hqd_pq_control = tmp;
+
+	/* set the wb address whether it's enabled or not */
+	wb_gpu_addr = prop->rptr_gpu_addr;
+	mqd->cp_hqd_pq_rptr_report_addr_lo = wb_gpu_addr & 0xfffffffc;
+	mqd->cp_hqd_pq_rptr_report_addr_hi =
+		upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	/* only used if CP_PQ_WPTR_POLL_CNTL.CP_PQ_WPTR_POLL_CNTL__EN_MASK=1 */
+	wb_gpu_addr = prop->wptr_gpu_addr;
+	mqd->cp_hqd_pq_wptr_poll_addr_lo = wb_gpu_addr & 0xfffffffc;
+	mqd->cp_hqd_pq_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	/* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */
+	mqd->cp_hqd_pq_rptr = RREG32_SOC15(GC, 0, mmCP_HQD_PQ_RPTR);
+
+	/* set the vmid for the queue */
+	mqd->cp_hqd_vmid = 0;
+
+	tmp = RREG32_SOC15(GC, 0, mmCP_HQD_PERSISTENT_STATE);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PERSISTENT_STATE, PRELOAD_SIZE, 0x53);
+	mqd->cp_hqd_persistent_state = tmp;
+
+	/* set MIN_IB_AVAIL_SIZE */
+	tmp = RREG32_SOC15(GC, 0, mmCP_HQD_IB_CONTROL);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_IB_CONTROL, MIN_IB_AVAIL_SIZE, 3);
+	mqd->cp_hqd_ib_control = tmp;
+
+	/* set static priority for a compute queue/ring */
+	mqd->cp_hqd_pipe_priority = prop->hqd_pipe_priority;
+	mqd->cp_hqd_queue_priority = prop->hqd_queue_priority;
+
+	mqd->cp_hqd_active = prop->hqd_active;
+
+	return 0;
+}
+
+static int gfx_v10_0_kiq_init_register(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v10_compute_mqd *mqd = ring->mqd_ptr;
+	int j;
+
+	/* inactivate the queue */
+	if (amdgpu_sriov_vf(adev))
+		WREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE, 0);
+
+	/* disable wptr polling */
+	WREG32_FIELD15(GC, 0, CP_PQ_WPTR_POLL_CNTL, EN, 0);
+
+	/* disable the queue if it's active */
+	if (RREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE) & 1) {
+		WREG32_SOC15(GC, 0, mmCP_HQD_DEQUEUE_REQUEST, 1);
+		for (j = 0; j < adev->usec_timeout; j++) {
+			if (!(RREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE) & 1))
+				break;
+			udelay(1);
+		}
+		WREG32_SOC15(GC, 0, mmCP_HQD_DEQUEUE_REQUEST,
+		       mqd->cp_hqd_dequeue_request);
+		WREG32_SOC15(GC, 0, mmCP_HQD_PQ_RPTR,
+		       mqd->cp_hqd_pq_rptr);
+		WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_LO,
+		       mqd->cp_hqd_pq_wptr_lo);
+		WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_HI,
+		       mqd->cp_hqd_pq_wptr_hi);
+	}
+
+	/* disable doorbells */
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL, 0);
+
+	/* write the EOP addr */
+	WREG32_SOC15(GC, 0, mmCP_HQD_EOP_BASE_ADDR,
+	       mqd->cp_hqd_eop_base_addr_lo);
+	WREG32_SOC15(GC, 0, mmCP_HQD_EOP_BASE_ADDR_HI,
+	       mqd->cp_hqd_eop_base_addr_hi);
+
+	/* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */
+	WREG32_SOC15(GC, 0, mmCP_HQD_EOP_CONTROL,
+	       mqd->cp_hqd_eop_control);
+
+	/* set the pointer to the MQD */
+	WREG32_SOC15(GC, 0, mmCP_MQD_BASE_ADDR,
+	       mqd->cp_mqd_base_addr_lo);
+	WREG32_SOC15(GC, 0, mmCP_MQD_BASE_ADDR_HI,
+	       mqd->cp_mqd_base_addr_hi);
+
+	/* set MQD vmid to 0 */
+	WREG32_SOC15(GC, 0, mmCP_MQD_CONTROL,
+	       mqd->cp_mqd_control);
+
+	/* set the pointer to the HQD, this is similar CP_RB0_BASE/_HI */
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_BASE,
+	       mqd->cp_hqd_pq_base_lo);
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_BASE_HI,
+	       mqd->cp_hqd_pq_base_hi);
+
+	/* set up the HQD, this is similar to CP_RB0_CNTL */
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_CONTROL,
+	       mqd->cp_hqd_pq_control);
+
+	/* set the wb address whether it's enabled or not */
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_RPTR_REPORT_ADDR,
+		mqd->cp_hqd_pq_rptr_report_addr_lo);
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_RPTR_REPORT_ADDR_HI,
+		mqd->cp_hqd_pq_rptr_report_addr_hi);
+
+	/* only used if CP_PQ_WPTR_POLL_CNTL.CP_PQ_WPTR_POLL_CNTL__EN_MASK=1 */
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_POLL_ADDR,
+	       mqd->cp_hqd_pq_wptr_poll_addr_lo);
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_POLL_ADDR_HI,
+	       mqd->cp_hqd_pq_wptr_poll_addr_hi);
+
+	/* enable the doorbell if requested */
+	if (ring->use_doorbell) {
+		WREG32_SOC15(GC, 0, mmCP_MEC_DOORBELL_RANGE_LOWER,
+			(adev->doorbell_index.kiq * 2) << 2);
+		WREG32_SOC15(GC, 0, mmCP_MEC_DOORBELL_RANGE_UPPER,
+			(adev->doorbell_index.userqueue_end * 2) << 2);
+	}
+
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL,
+	       mqd->cp_hqd_pq_doorbell_control);
+
+	/* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_LO,
+	       mqd->cp_hqd_pq_wptr_lo);
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_HI,
+	       mqd->cp_hqd_pq_wptr_hi);
+
+	/* set the vmid for the queue */
+	WREG32_SOC15(GC, 0, mmCP_HQD_VMID, mqd->cp_hqd_vmid);
+
+	WREG32_SOC15(GC, 0, mmCP_HQD_PERSISTENT_STATE,
+	       mqd->cp_hqd_persistent_state);
+
+	/* activate the queue */
+	WREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE,
+	       mqd->cp_hqd_active);
+
+	if (ring->use_doorbell)
+		WREG32_FIELD15(GC, 0, CP_PQ_STATUS, DOORBELL_ENABLE, 1);
+
+	return 0;
+}
+
+static int gfx_v10_0_kiq_init_queue(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v10_compute_mqd *mqd = ring->mqd_ptr;
+
+	gfx_v10_0_kiq_setting(ring);
+
+	if (amdgpu_in_reset(adev)) { /* for GPU_RESET case */
+		/* reset MQD to a clean status */
+		if (adev->gfx.kiq[0].mqd_backup)
+			memcpy_toio(mqd, adev->gfx.kiq[0].mqd_backup, sizeof(*mqd));
+
+		/* reset ring buffer */
+		ring->wptr = 0;
+		amdgpu_ring_clear_ring(ring);
+
+		mutex_lock(&adev->srbm_mutex);
+		nv_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		gfx_v10_0_kiq_init_register(ring);
+		nv_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+	} else {
+		memset((void *)mqd, 0, sizeof(*mqd));
+		if (amdgpu_sriov_vf(adev) && adev->in_suspend)
+			amdgpu_ring_clear_ring(ring);
+		mutex_lock(&adev->srbm_mutex);
+		nv_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		amdgpu_ring_init_mqd(ring);
+		gfx_v10_0_kiq_init_register(ring);
+		nv_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+
+		if (adev->gfx.kiq[0].mqd_backup)
+			memcpy_fromio(adev->gfx.kiq[0].mqd_backup, mqd, sizeof(*mqd));
+	}
+
+	return 0;
+}
+
+static int gfx_v10_0_kcq_init_queue(struct amdgpu_ring *ring, bool restore)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v10_compute_mqd *mqd = ring->mqd_ptr;
+	int mqd_idx = ring - &adev->gfx.compute_ring[0];
+
+	if (!restore && !amdgpu_in_reset(adev) && !adev->in_suspend) {
+		memset((void *)mqd, 0, sizeof(*mqd));
+		mutex_lock(&adev->srbm_mutex);
+		nv_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		amdgpu_ring_init_mqd(ring);
+		nv_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+
+		if (adev->gfx.mec.mqd_backup[mqd_idx])
+			memcpy_fromio(adev->gfx.mec.mqd_backup[mqd_idx], mqd, sizeof(*mqd));
+	} else {
+		/* restore MQD to a clean status */
+		if (adev->gfx.mec.mqd_backup[mqd_idx])
+			memcpy_toio(mqd, adev->gfx.mec.mqd_backup[mqd_idx], sizeof(*mqd));
+		/* reset ring buffer */
+		ring->wptr = 0;
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr, 0);
+		amdgpu_ring_clear_ring(ring);
+	}
+
+	return 0;
+}
+
+static int gfx_v10_0_kiq_resume(struct amdgpu_device *adev)
+{
+	gfx_v10_0_kiq_init_queue(&adev->gfx.kiq[0].ring);
+	return 0;
+}
+
+static int gfx_v10_0_kcq_resume(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	gfx_v10_0_cp_compute_enable(adev, true);
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		r = gfx_v10_0_kcq_init_queue(&adev->gfx.compute_ring[i],
+					     false);
+		if (r)
+			return r;
+	}
+
+	return amdgpu_gfx_enable_kcq(adev, 0);
+}
+
+static int gfx_v10_0_cp_resume(struct amdgpu_device *adev)
+{
+	int r, i;
+	struct amdgpu_ring *ring;
+
+	if (!(adev->flags & AMD_IS_APU))
+		gfx_v10_0_enable_gui_idle_interrupt(adev, false);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+		/* legacy firmware loading */
+		r = gfx_v10_0_cp_gfx_load_microcode(adev);
+		if (r)
+			return r;
+
+		r = gfx_v10_0_cp_compute_load_microcode(adev);
+		if (r)
+			return r;
+	}
+
+	r = gfx_v10_0_kiq_resume(adev);
+	if (r)
+		return r;
+
+	r = gfx_v10_0_kcq_resume(adev);
+	if (r)
+		return r;
+
+	if (!amdgpu_async_gfx_ring) {
+		r = gfx_v10_0_cp_gfx_resume(adev);
+		if (r)
+			return r;
+	} else {
+		r = gfx_v10_0_cp_async_gfx_ring_resume(adev);
+		if (r)
+			return r;
+	}
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+		ring = &adev->gfx.gfx_ring[i];
+		r = amdgpu_ring_test_helper(ring);
+		if (r)
+			return r;
+	}
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		ring = &adev->gfx.compute_ring[i];
+		r = amdgpu_ring_test_helper(ring);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+static void gfx_v10_0_cp_enable(struct amdgpu_device *adev, bool enable)
+{
+	gfx_v10_0_cp_gfx_enable(adev, enable);
+	gfx_v10_0_cp_compute_enable(adev, enable);
+}
+
+static bool gfx_v10_0_check_grbm_cam_remapping(struct amdgpu_device *adev)
+{
+	uint32_t data, pattern = 0xDEADBEEF;
+
+	/*
+	 * check if mmVGT_ESGS_RING_SIZE_UMD
+	 * has been remapped to mmVGT_ESGS_RING_SIZE
+	 */
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+		data = RREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE_Sienna_Cichlid);
+		WREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE_Sienna_Cichlid, 0);
+		WREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE_UMD, pattern);
+
+		if (RREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE_Sienna_Cichlid) == pattern) {
+			WREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE_UMD, data);
+			return true;
+		}
+		WREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE_Sienna_Cichlid, data);
+		break;
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 3):
+	case IP_VERSION(10, 3, 6):
+	case IP_VERSION(10, 3, 7):
+		return true;
+	default:
+		data = RREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE);
+		WREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE, 0);
+		WREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE_UMD, pattern);
+
+		if (RREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE) == pattern) {
+			WREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE_UMD, data);
+			return true;
+		}
+		WREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE, data);
+		break;
+	}
+
+	return false;
+}
+
+static void gfx_v10_0_setup_grbm_cam_remapping(struct amdgpu_device *adev)
+{
+	uint32_t data;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/*
+	 * Initialize cam_index to 0
+	 * index will auto-inc after each data writing
+	 */
+	WREG32_SOC15(GC, 0, mmGRBM_CAM_INDEX, 0);
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+	case IP_VERSION(10, 3, 6):
+	case IP_VERSION(10, 3, 3):
+	case IP_VERSION(10, 3, 7):
+		/* mmVGT_TF_RING_SIZE_UMD -> mmVGT_TF_RING_SIZE */
+		data = (SOC15_REG_OFFSET(GC, 0, mmVGT_TF_RING_SIZE_UMD) <<
+			GRBM_CAM_DATA__CAM_ADDR__SHIFT) |
+		       (SOC15_REG_OFFSET(GC, 0, mmVGT_TF_RING_SIZE_Sienna_Cichlid) <<
+			GRBM_CAM_DATA__CAM_REMAPADDR__SHIFT);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA_UPPER, 0);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA, data);
+
+		/* mmVGT_TF_MEMORY_BASE_UMD -> mmVGT_TF_MEMORY_BASE */
+		data = (SOC15_REG_OFFSET(GC, 0, mmVGT_TF_MEMORY_BASE_UMD) <<
+			GRBM_CAM_DATA__CAM_ADDR__SHIFT) |
+		       (SOC15_REG_OFFSET(GC, 0, mmVGT_TF_MEMORY_BASE_Sienna_Cichlid) <<
+			GRBM_CAM_DATA__CAM_REMAPADDR__SHIFT);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA_UPPER, 0);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA, data);
+
+		/* mmVGT_TF_MEMORY_BASE_HI_UMD -> mmVGT_TF_MEMORY_BASE_HI */
+		data = (SOC15_REG_OFFSET(GC, 0, mmVGT_TF_MEMORY_BASE_HI_UMD) <<
+			GRBM_CAM_DATA__CAM_ADDR__SHIFT) |
+		       (SOC15_REG_OFFSET(GC, 0, mmVGT_TF_MEMORY_BASE_HI_Sienna_Cichlid) <<
+			GRBM_CAM_DATA__CAM_REMAPADDR__SHIFT);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA_UPPER, 0);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA, data);
+
+		/* mmVGT_HS_OFFCHIP_PARAM_UMD -> mmVGT_HS_OFFCHIP_PARAM */
+		data = (SOC15_REG_OFFSET(GC, 0, mmVGT_HS_OFFCHIP_PARAM_UMD) <<
+			GRBM_CAM_DATA__CAM_ADDR__SHIFT) |
+		       (SOC15_REG_OFFSET(GC, 0, mmVGT_HS_OFFCHIP_PARAM_Sienna_Cichlid) <<
+			GRBM_CAM_DATA__CAM_REMAPADDR__SHIFT);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA_UPPER, 0);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA, data);
+
+		/* mmVGT_ESGS_RING_SIZE_UMD -> mmVGT_ESGS_RING_SIZE */
+		data = (SOC15_REG_OFFSET(GC, 0, mmVGT_ESGS_RING_SIZE_UMD) <<
+			GRBM_CAM_DATA__CAM_ADDR__SHIFT) |
+		       (SOC15_REG_OFFSET(GC, 0, mmVGT_ESGS_RING_SIZE_Sienna_Cichlid) <<
+			GRBM_CAM_DATA__CAM_REMAPADDR__SHIFT);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA_UPPER, 0);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA, data);
+
+		/* mmVGT_GSVS_RING_SIZE_UMD -> mmVGT_GSVS_RING_SIZE */
+		data = (SOC15_REG_OFFSET(GC, 0, mmVGT_GSVS_RING_SIZE_UMD) <<
+			GRBM_CAM_DATA__CAM_ADDR__SHIFT) |
+		       (SOC15_REG_OFFSET(GC, 0, mmVGT_GSVS_RING_SIZE_Sienna_Cichlid) <<
+			GRBM_CAM_DATA__CAM_REMAPADDR__SHIFT);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA_UPPER, 0);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA, data);
+
+		/* mmSPI_CONFIG_CNTL_REMAP -> mmSPI_CONFIG_CNTL */
+		data = (SOC15_REG_OFFSET(GC, 0, mmSPI_CONFIG_CNTL_REMAP) <<
+			GRBM_CAM_DATA__CAM_ADDR__SHIFT) |
+		       (SOC15_REG_OFFSET(GC, 0, mmSPI_CONFIG_CNTL_Sienna_Cichlid) <<
+			GRBM_CAM_DATA__CAM_REMAPADDR__SHIFT);
+		break;
+	default:
+		/* mmVGT_TF_RING_SIZE_UMD -> mmVGT_TF_RING_SIZE */
+		data = (SOC15_REG_OFFSET(GC, 0, mmVGT_TF_RING_SIZE_UMD) <<
+			GRBM_CAM_DATA__CAM_ADDR__SHIFT) |
+		       (SOC15_REG_OFFSET(GC, 0, mmVGT_TF_RING_SIZE) <<
+			GRBM_CAM_DATA__CAM_REMAPADDR__SHIFT);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA_UPPER, 0);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA, data);
+
+		/* mmVGT_TF_MEMORY_BASE_UMD -> mmVGT_TF_MEMORY_BASE */
+		data = (SOC15_REG_OFFSET(GC, 0, mmVGT_TF_MEMORY_BASE_UMD) <<
+			GRBM_CAM_DATA__CAM_ADDR__SHIFT) |
+		       (SOC15_REG_OFFSET(GC, 0, mmVGT_TF_MEMORY_BASE) <<
+			GRBM_CAM_DATA__CAM_REMAPADDR__SHIFT);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA_UPPER, 0);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA, data);
+
+		/* mmVGT_TF_MEMORY_BASE_HI_UMD -> mmVGT_TF_MEMORY_BASE_HI */
+		data = (SOC15_REG_OFFSET(GC, 0, mmVGT_TF_MEMORY_BASE_HI_UMD) <<
+			GRBM_CAM_DATA__CAM_ADDR__SHIFT) |
+		       (SOC15_REG_OFFSET(GC, 0, mmVGT_TF_MEMORY_BASE_HI) <<
+			GRBM_CAM_DATA__CAM_REMAPADDR__SHIFT);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA_UPPER, 0);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA, data);
+
+		/* mmVGT_HS_OFFCHIP_PARAM_UMD -> mmVGT_HS_OFFCHIP_PARAM */
+		data = (SOC15_REG_OFFSET(GC, 0, mmVGT_HS_OFFCHIP_PARAM_UMD) <<
+			GRBM_CAM_DATA__CAM_ADDR__SHIFT) |
+		       (SOC15_REG_OFFSET(GC, 0, mmVGT_HS_OFFCHIP_PARAM) <<
+			GRBM_CAM_DATA__CAM_REMAPADDR__SHIFT);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA_UPPER, 0);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA, data);
+
+		/* mmVGT_ESGS_RING_SIZE_UMD -> mmVGT_ESGS_RING_SIZE */
+		data = (SOC15_REG_OFFSET(GC, 0, mmVGT_ESGS_RING_SIZE_UMD) <<
+			GRBM_CAM_DATA__CAM_ADDR__SHIFT) |
+		       (SOC15_REG_OFFSET(GC, 0, mmVGT_ESGS_RING_SIZE) <<
+			GRBM_CAM_DATA__CAM_REMAPADDR__SHIFT);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA_UPPER, 0);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA, data);
+
+		/* mmVGT_GSVS_RING_SIZE_UMD -> mmVGT_GSVS_RING_SIZE */
+		data = (SOC15_REG_OFFSET(GC, 0, mmVGT_GSVS_RING_SIZE_UMD) <<
+			GRBM_CAM_DATA__CAM_ADDR__SHIFT) |
+		       (SOC15_REG_OFFSET(GC, 0, mmVGT_GSVS_RING_SIZE) <<
+			GRBM_CAM_DATA__CAM_REMAPADDR__SHIFT);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA_UPPER, 0);
+		WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA, data);
+
+		/* mmSPI_CONFIG_CNTL_REMAP -> mmSPI_CONFIG_CNTL */
+		data = (SOC15_REG_OFFSET(GC, 0, mmSPI_CONFIG_CNTL_REMAP) <<
+			GRBM_CAM_DATA__CAM_ADDR__SHIFT) |
+		       (SOC15_REG_OFFSET(GC, 0, mmSPI_CONFIG_CNTL) <<
+			GRBM_CAM_DATA__CAM_REMAPADDR__SHIFT);
+		break;
+	}
+
+	WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA_UPPER, 0);
+	WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA, data);
+}
+
+static void gfx_v10_0_disable_gpa_mode(struct amdgpu_device *adev)
+{
+	uint32_t data;
+
+	data = RREG32_SOC15(GC, 0, mmCPC_PSP_DEBUG);
+	data |= CPC_PSP_DEBUG__GPA_OVERRIDE_MASK;
+	WREG32_SOC15(GC, 0, mmCPC_PSP_DEBUG, data);
+
+	data = RREG32_SOC15(GC, 0, mmCPG_PSP_DEBUG);
+	data |= CPG_PSP_DEBUG__GPA_OVERRIDE_MASK;
+	WREG32_SOC15(GC, 0, mmCPG_PSP_DEBUG, data);
+}
+
+static int gfx_v10_0_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (!amdgpu_emu_mode)
+		gfx_v10_0_init_golden_registers(adev);
+
+	amdgpu_gfx_cleaner_shader_init(adev, adev->gfx.cleaner_shader_size,
+				       adev->gfx.cleaner_shader_ptr);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+		/**
+		 * For gfx 10, rlc firmware loading relies on smu firmware is
+		 * loaded firstly, so in direct type, it has to load smc ucode
+		 * here before rlc.
+		 */
+		r = amdgpu_pm_load_smu_firmware(adev, NULL);
+		if (r)
+			return r;
+		gfx_v10_0_disable_gpa_mode(adev);
+	}
+
+	/* if GRBM CAM not remapped, set up the remapping */
+	if (!gfx_v10_0_check_grbm_cam_remapping(adev))
+		gfx_v10_0_setup_grbm_cam_remapping(adev);
+
+	gfx_v10_0_constants_init(adev);
+
+	r = gfx_v10_0_rlc_resume(adev);
+	if (r)
+		return r;
+
+	/*
+	 * init golden registers and rlc resume may override some registers,
+	 * reconfig them here
+	 */
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 1, 10) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 1, 1) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 1, 2))
+		gfx_v10_0_tcp_harvest(adev);
+
+	r = gfx_v10_0_cp_resume(adev);
+	if (r)
+		return r;
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 3, 0))
+		gfx_v10_3_program_pbb_mode(adev);
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(10, 3, 0) && !amdgpu_sriov_vf(adev))
+		gfx_v10_3_set_power_brake_sequence(adev);
+
+	return r;
+}
+
+static int gfx_v10_0_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	cancel_delayed_work_sync(&adev->gfx.idle_work);
+
+	amdgpu_irq_put(adev, &adev->gfx.priv_reg_irq, 0);
+	amdgpu_irq_put(adev, &adev->gfx.priv_inst_irq, 0);
+	amdgpu_irq_put(adev, &adev->gfx.bad_op_irq, 0);
+
+	/* WA added for Vangogh asic fixing the SMU suspend failure
+	 * It needs to set power gating again during gfxoff control
+	 * otherwise the gfxoff disallowing will be failed to set.
+	 */
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 3, 1))
+		gfx_v10_0_set_powergating_state(ip_block, AMD_PG_STATE_UNGATE);
+
+	if (!adev->no_hw_access) {
+		if (amdgpu_async_gfx_ring) {
+			if (amdgpu_gfx_disable_kgq(adev, 0))
+				DRM_ERROR("KGQ disable failed\n");
+		}
+
+		if (amdgpu_gfx_disable_kcq(adev, 0))
+			DRM_ERROR("KCQ disable failed\n");
+	}
+
+	if (amdgpu_sriov_vf(adev)) {
+		gfx_v10_0_cp_gfx_enable(adev, false);
+		/* Remove the steps of clearing KIQ position.
+		 * It causes GFX hang when another Win guest is rendering.
+		 */
+		return 0;
+	}
+	gfx_v10_0_cp_enable(adev, false);
+	gfx_v10_0_enable_gui_idle_interrupt(adev, false);
+
+	return 0;
+}
+
+static int gfx_v10_0_suspend(struct amdgpu_ip_block *ip_block)
+{
+	return gfx_v10_0_hw_fini(ip_block);
+}
+
+static int gfx_v10_0_resume(struct amdgpu_ip_block *ip_block)
+{
+	return gfx_v10_0_hw_init(ip_block);
+}
+
+static bool gfx_v10_0_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (REG_GET_FIELD(RREG32_SOC15(GC, 0, mmGRBM_STATUS),
+				GRBM_STATUS, GUI_ACTIVE))
+		return false;
+	else
+		return true;
+}
+
+static int gfx_v10_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	unsigned int i;
+	u32 tmp;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		/* read MC_STATUS */
+		tmp = RREG32_SOC15(GC, 0, mmGRBM_STATUS) &
+			GRBM_STATUS__GUI_ACTIVE_MASK;
+
+		if (!REG_GET_FIELD(tmp, GRBM_STATUS, GUI_ACTIVE))
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static int gfx_v10_0_soft_reset(struct amdgpu_ip_block *ip_block)
+{
+	u32 grbm_soft_reset = 0;
+	u32 tmp;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	/* GRBM_STATUS */
+	tmp = RREG32_SOC15(GC, 0, mmGRBM_STATUS);
+	if (tmp & (GRBM_STATUS__PA_BUSY_MASK | GRBM_STATUS__SC_BUSY_MASK |
+		   GRBM_STATUS__BCI_BUSY_MASK | GRBM_STATUS__SX_BUSY_MASK |
+		   GRBM_STATUS__TA_BUSY_MASK | GRBM_STATUS__DB_BUSY_MASK |
+		   GRBM_STATUS__CB_BUSY_MASK | GRBM_STATUS__GDS_BUSY_MASK |
+		   GRBM_STATUS__SPI_BUSY_MASK | GRBM_STATUS__GE_BUSY_NO_DMA_MASK)) {
+		grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset,
+						GRBM_SOFT_RESET, SOFT_RESET_CP,
+						1);
+		grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset,
+						GRBM_SOFT_RESET, SOFT_RESET_GFX,
+						1);
+	}
+
+	if (tmp & (GRBM_STATUS__CP_BUSY_MASK | GRBM_STATUS__CP_COHERENCY_BUSY_MASK)) {
+		grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset,
+						GRBM_SOFT_RESET, SOFT_RESET_CP,
+						1);
+	}
+
+	/* GRBM_STATUS2 */
+	tmp = RREG32_SOC15(GC, 0, mmGRBM_STATUS2);
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+	case IP_VERSION(10, 3, 6):
+	case IP_VERSION(10, 3, 3):
+		if (REG_GET_FIELD(tmp, GRBM_STATUS2, RLC_BUSY_Sienna_Cichlid))
+			grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset,
+							GRBM_SOFT_RESET,
+							SOFT_RESET_RLC,
+							1);
+		break;
+	default:
+		if (REG_GET_FIELD(tmp, GRBM_STATUS2, RLC_BUSY))
+			grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset,
+							GRBM_SOFT_RESET,
+							SOFT_RESET_RLC,
+							1);
+		break;
+	}
+
+	if (grbm_soft_reset) {
+		/* stop the rlc */
+		gfx_v10_0_rlc_stop(adev);
+
+		/* Disable GFX parsing/prefetching */
+		gfx_v10_0_cp_gfx_enable(adev, false);
+
+		/* Disable MEC parsing/prefetching */
+		gfx_v10_0_cp_compute_enable(adev, false);
+
+		tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
+		tmp |= grbm_soft_reset;
+		dev_info(adev->dev, "GRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET, tmp);
+		tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~grbm_soft_reset;
+		WREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET, tmp);
+		tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+	}
+	return 0;
+}
+
+static uint64_t gfx_v10_0_get_gpu_clock_counter(struct amdgpu_device *adev)
+{
+	uint64_t clock, clock_lo, clock_hi, hi_check;
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 1, 3):
+	case IP_VERSION(10, 1, 4):
+		preempt_disable();
+		clock_hi = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_UPPER_Cyan_Skillfish);
+		clock_lo = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_LOWER_Cyan_Skillfish);
+		hi_check = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_UPPER_Cyan_Skillfish);
+		/* The SMUIO TSC clock frequency is 100MHz, which sets 32-bit carry over
+		 * roughly every 42 seconds.
+		 */
+		if (hi_check != clock_hi) {
+			clock_lo = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_LOWER_Cyan_Skillfish);
+			clock_hi = hi_check;
+		}
+		preempt_enable();
+		clock = clock_lo | (clock_hi << 32ULL);
+		break;
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 3):
+	case IP_VERSION(10, 3, 7):
+		preempt_disable();
+		clock_hi = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_UPPER_Vangogh);
+		clock_lo = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_LOWER_Vangogh);
+		hi_check = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_UPPER_Vangogh);
+		/* The SMUIO TSC clock frequency is 100MHz, which sets 32-bit carry over
+		 * roughly every 42 seconds.
+		 */
+		if (hi_check != clock_hi) {
+			clock_lo = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_LOWER_Vangogh);
+			clock_hi = hi_check;
+		}
+		preempt_enable();
+		clock = clock_lo | (clock_hi << 32ULL);
+		break;
+	case IP_VERSION(10, 3, 6):
+		preempt_disable();
+		clock_hi = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_UPPER_GC_10_3_6);
+		clock_lo = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_LOWER_GC_10_3_6);
+		hi_check = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_UPPER_GC_10_3_6);
+		/* The SMUIO TSC clock frequency is 100MHz, which sets 32-bit carry over
+		 * roughly every 42 seconds.
+		 */
+		if (hi_check != clock_hi) {
+			clock_lo = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_LOWER_GC_10_3_6);
+			clock_hi = hi_check;
+		}
+		preempt_enable();
+		clock = clock_lo | (clock_hi << 32ULL);
+		break;
+	default:
+		preempt_disable();
+		clock_hi = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_UPPER);
+		clock_lo = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_LOWER);
+		hi_check = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_UPPER);
+		/* The SMUIO TSC clock frequency is 100MHz, which sets 32-bit carry over
+		 * roughly every 42 seconds.
+		 */
+		if (hi_check != clock_hi) {
+			clock_lo = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_LOWER);
+			clock_hi = hi_check;
+		}
+		preempt_enable();
+		clock = clock_lo | (clock_hi << 32ULL);
+		break;
+	}
+	return clock;
+}
+
+static void gfx_v10_0_ring_emit_gds_switch(struct amdgpu_ring *ring,
+					   uint32_t vmid,
+					   uint32_t gds_base, uint32_t gds_size,
+					   uint32_t gws_base, uint32_t gws_size,
+					   uint32_t oa_base, uint32_t oa_size)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	/* GDS Base */
+	gfx_v10_0_write_data_to_reg(ring, 0, false,
+				    SOC15_REG_OFFSET(GC, 0, mmGDS_VMID0_BASE) + 2 * vmid,
+				    gds_base);
+
+	/* GDS Size */
+	gfx_v10_0_write_data_to_reg(ring, 0, false,
+				    SOC15_REG_OFFSET(GC, 0, mmGDS_VMID0_SIZE) + 2 * vmid,
+				    gds_size);
+
+	/* GWS */
+	gfx_v10_0_write_data_to_reg(ring, 0, false,
+				    SOC15_REG_OFFSET(GC, 0, mmGDS_GWS_VMID0) + vmid,
+				    gws_size << GDS_GWS_VMID0__SIZE__SHIFT | gws_base);
+
+	/* OA */
+	gfx_v10_0_write_data_to_reg(ring, 0, false,
+				    SOC15_REG_OFFSET(GC, 0, mmGDS_OA_VMID0) + vmid,
+				    (1 << (oa_size + oa_base)) - (1 << oa_base));
+}
+
+static int gfx_v10_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	adev->gfx.funcs = &gfx_v10_0_gfx_funcs;
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 1, 10):
+	case IP_VERSION(10, 1, 1):
+	case IP_VERSION(10, 1, 2):
+	case IP_VERSION(10, 1, 3):
+	case IP_VERSION(10, 1, 4):
+		adev->gfx.num_gfx_rings = GFX10_NUM_GFX_RINGS_NV1X;
+		break;
+	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+	case IP_VERSION(10, 3, 6):
+	case IP_VERSION(10, 3, 3):
+	case IP_VERSION(10, 3, 7):
+		adev->gfx.num_gfx_rings = GFX10_NUM_GFX_RINGS_Sienna_Cichlid;
+		break;
+	default:
+		break;
+	}
+
+	adev->gfx.num_compute_rings = min(amdgpu_gfx_get_num_kcq(adev),
+					  AMDGPU_MAX_COMPUTE_RINGS);
+
+	gfx_v10_0_set_kiq_pm4_funcs(adev);
+	gfx_v10_0_set_ring_funcs(adev);
+	gfx_v10_0_set_irq_funcs(adev);
+	gfx_v10_0_set_gds_init(adev);
+	gfx_v10_0_set_rlc_funcs(adev);
+	gfx_v10_0_set_mqd_funcs(adev);
+
+	/* init rlcg reg access ctrl */
+	gfx_v10_0_init_rlcg_reg_access_ctrl(adev);
+
+	return gfx_v10_0_init_microcode(adev);
+}
+
+static int gfx_v10_0_late_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = amdgpu_irq_get(adev, &adev->gfx.priv_reg_irq, 0);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_get(adev, &adev->gfx.priv_inst_irq, 0);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_get(adev, &adev->gfx.bad_op_irq, 0);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static bool gfx_v10_0_is_rlc_enabled(struct amdgpu_device *adev)
+{
+	uint32_t rlc_cntl;
+
+	/* if RLC is not enabled, do nothing */
+	rlc_cntl = RREG32_SOC15(GC, 0, mmRLC_CNTL);
+	return (REG_GET_FIELD(rlc_cntl, RLC_CNTL, RLC_ENABLE_F32)) ? true : false;
+}
+
+static void gfx_v10_0_set_safe_mode(struct amdgpu_device *adev, int xcc_id)
+{
+	uint32_t data;
+	unsigned int i;
+
+	data = RLC_SAFE_MODE__CMD_MASK;
+	data |= (1 << RLC_SAFE_MODE__MESSAGE__SHIFT);
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+	case IP_VERSION(10, 3, 6):
+	case IP_VERSION(10, 3, 3):
+	case IP_VERSION(10, 3, 7):
+		WREG32_SOC15(GC, 0, mmRLC_SAFE_MODE_Sienna_Cichlid, data);
+
+		/* wait for RLC_SAFE_MODE */
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (!REG_GET_FIELD(RREG32_SOC15(GC, 0, mmRLC_SAFE_MODE_Sienna_Cichlid),
+					   RLC_SAFE_MODE, CMD))
+				break;
+			udelay(1);
+		}
+		break;
+	default:
+		WREG32_SOC15(GC, 0, mmRLC_SAFE_MODE, data);
+
+		/* wait for RLC_SAFE_MODE */
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (!REG_GET_FIELD(RREG32_SOC15(GC, 0, mmRLC_SAFE_MODE),
+					   RLC_SAFE_MODE, CMD))
+				break;
+			udelay(1);
+		}
+		break;
+	}
+}
+
+static void gfx_v10_0_unset_safe_mode(struct amdgpu_device *adev, int xcc_id)
+{
+	uint32_t data;
+
+	data = RLC_SAFE_MODE__CMD_MASK;
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+	case IP_VERSION(10, 3, 6):
+	case IP_VERSION(10, 3, 3):
+	case IP_VERSION(10, 3, 7):
+		WREG32_SOC15(GC, 0, mmRLC_SAFE_MODE_Sienna_Cichlid, data);
+		break;
+	default:
+		WREG32_SOC15(GC, 0, mmRLC_SAFE_MODE, data);
+		break;
+	}
+}
+
+static void gfx_v10_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						      bool enable)
+{
+	uint32_t data, def;
+
+	if (!(adev->cg_flags & (AMD_CG_SUPPORT_GFX_MGCG | AMD_CG_SUPPORT_GFX_MGLS)))
+		return;
+
+	/* It is disabled by HW by default */
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGCG)) {
+		/* 0 - Disable some blocks' MGCG */
+		WREG32_SOC15(GC, 0, mmGRBM_GFX_INDEX, 0xe0000000);
+		WREG32_SOC15(GC, 0, mmCGTT_WD_CLK_CTRL, 0xff000000);
+		WREG32_SOC15(GC, 0, mmCGTT_VGT_CLK_CTRL, 0xff000000);
+		WREG32_SOC15(GC, 0, mmCGTT_IA_CLK_CTRL, 0xff000000);
+
+		/* 1 - RLC_CGTT_MGCG_OVERRIDE */
+		def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE);
+		data &= ~(RLC_CGTT_MGCG_OVERRIDE__GRBM_CGTT_SCLK_OVERRIDE_MASK |
+			  RLC_CGTT_MGCG_OVERRIDE__RLC_CGTT_SCLK_OVERRIDE_MASK |
+			  RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGCG_OVERRIDE_MASK |
+			  RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGLS_OVERRIDE_MASK |
+			  RLC_CGTT_MGCG_OVERRIDE__GFXIP_FGCG_OVERRIDE_MASK |
+			  RLC_CGTT_MGCG_OVERRIDE__ENABLE_CGTS_LEGACY_MASK);
+
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE, data);
+
+		/* MGLS is a global flag to control all MGLS in GFX */
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGLS) {
+			/* 2 - RLC memory Light sleep */
+			if (adev->cg_flags & AMD_CG_SUPPORT_GFX_RLC_LS) {
+				def = data = RREG32_SOC15(GC, 0, mmRLC_MEM_SLP_CNTL);
+				data |= RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK;
+				if (def != data)
+					WREG32_SOC15(GC, 0, mmRLC_MEM_SLP_CNTL, data);
+			}
+			/* 3 - CP memory Light sleep */
+			if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CP_LS) {
+				def = data = RREG32_SOC15(GC, 0, mmCP_MEM_SLP_CNTL);
+				data |= CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK;
+				if (def != data)
+					WREG32_SOC15(GC, 0, mmCP_MEM_SLP_CNTL, data);
+			}
+		}
+	} else if (!enable || !(adev->cg_flags & AMD_CG_SUPPORT_GFX_MGCG)) {
+		/* 1 - MGCG_OVERRIDE */
+		def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE);
+		data |= (RLC_CGTT_MGCG_OVERRIDE__RLC_CGTT_SCLK_OVERRIDE_MASK |
+			 RLC_CGTT_MGCG_OVERRIDE__GRBM_CGTT_SCLK_OVERRIDE_MASK |
+			 RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGCG_OVERRIDE_MASK |
+			 RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGLS_OVERRIDE_MASK |
+			 RLC_CGTT_MGCG_OVERRIDE__GFXIP_FGCG_OVERRIDE_MASK |
+			 RLC_CGTT_MGCG_OVERRIDE__ENABLE_CGTS_LEGACY_MASK);
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE, data);
+
+		/* 2 - disable MGLS in CP */
+		data = RREG32_SOC15(GC, 0, mmCP_MEM_SLP_CNTL);
+		if (data & CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK) {
+			data &= ~CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK;
+			WREG32_SOC15(GC, 0, mmCP_MEM_SLP_CNTL, data);
+		}
+
+		/* 3 - disable MGLS in RLC */
+		data = RREG32_SOC15(GC, 0, mmRLC_MEM_SLP_CNTL);
+		if (data & RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK) {
+			data &= ~RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK;
+			WREG32_SOC15(GC, 0, mmRLC_MEM_SLP_CNTL, data);
+		}
+
+	}
+}
+
+static void gfx_v10_0_update_3d_clock_gating(struct amdgpu_device *adev,
+					   bool enable)
+{
+	uint32_t data, def;
+
+	if (!(adev->cg_flags & (AMD_CG_SUPPORT_GFX_3D_CGCG | AMD_CG_SUPPORT_GFX_3D_CGLS)))
+		return;
+
+	/* Enable 3D CGCG/CGLS */
+	if (enable) {
+		/* write cmd to clear cgcg/cgls ov */
+		def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE);
+
+		/* unset CGCG override */
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGCG)
+			data &= ~RLC_CGTT_MGCG_OVERRIDE__GFXIP_GFX3D_CG_OVERRIDE_MASK;
+
+		/* update CGCG and CGLS override bits */
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE, data);
+
+		/* enable 3Dcgcg FSM(0x0000363f) */
+		def = RREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL_3D);
+		data = 0;
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGCG)
+			data = (0x36 << RLC_CGCG_CGLS_CTRL_3D__CGCG_GFX_IDLE_THRESHOLD__SHIFT) |
+				RLC_CGCG_CGLS_CTRL_3D__CGCG_EN_MASK;
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGLS)
+			data |= (0x000F << RLC_CGCG_CGLS_CTRL_3D__CGLS_REP_COMPANSAT_DELAY__SHIFT) |
+				RLC_CGCG_CGLS_CTRL_3D__CGLS_EN_MASK;
+
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL_3D, data);
+
+		/* set IDLE_POLL_COUNT(0x00900100) */
+		def = RREG32_SOC15(GC, 0, mmCP_RB_WPTR_POLL_CNTL);
+		data = (0x0100 << CP_RB_WPTR_POLL_CNTL__POLL_FREQUENCY__SHIFT) |
+			(0x0090 << CP_RB_WPTR_POLL_CNTL__IDLE_POLL_COUNT__SHIFT);
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmCP_RB_WPTR_POLL_CNTL, data);
+	} else {
+		/* Disable CGCG/CGLS */
+		def = data = RREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL_3D);
+
+		/* disable cgcg, cgls should be disabled */
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGCG)
+			data &= ~RLC_CGCG_CGLS_CTRL_3D__CGCG_EN_MASK;
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGLS)
+			data &= ~RLC_CGCG_CGLS_CTRL_3D__CGLS_EN_MASK;
+
+		/* disable cgcg and cgls in FSM */
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL_3D, data);
+	}
+}
+
+static void gfx_v10_0_update_coarse_grain_clock_gating(struct amdgpu_device *adev,
+						      bool enable)
+{
+	uint32_t def, data;
+
+	if (!(adev->cg_flags & (AMD_CG_SUPPORT_GFX_CGCG | AMD_CG_SUPPORT_GFX_CGLS)))
+		return;
+
+	if (enable) {
+		def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE);
+
+		/* unset CGCG override */
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGCG)
+			data &= ~RLC_CGTT_MGCG_OVERRIDE__GFXIP_CGCG_OVERRIDE_MASK;
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGLS)
+			data &= ~RLC_CGTT_MGCG_OVERRIDE__GFXIP_CGLS_OVERRIDE_MASK;
+
+		/* update CGCG and CGLS override bits */
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE, data);
+
+		/* enable cgcg FSM(0x0000363F) */
+		def = RREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL);
+		data = 0;
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGCG)
+			data = (0x36 << RLC_CGCG_CGLS_CTRL__CGCG_GFX_IDLE_THRESHOLD__SHIFT) |
+				RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK;
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGLS)
+			data |= (0x000F << RLC_CGCG_CGLS_CTRL__CGLS_REP_COMPANSAT_DELAY__SHIFT) |
+				RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK;
+
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL, data);
+
+		/* set IDLE_POLL_COUNT(0x00900100) */
+		def = RREG32_SOC15(GC, 0, mmCP_RB_WPTR_POLL_CNTL);
+		data = (0x0100 << CP_RB_WPTR_POLL_CNTL__POLL_FREQUENCY__SHIFT) |
+			(0x0090 << CP_RB_WPTR_POLL_CNTL__IDLE_POLL_COUNT__SHIFT);
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmCP_RB_WPTR_POLL_CNTL, data);
+	} else {
+		def = data = RREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL);
+
+		/* reset CGCG/CGLS bits */
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGCG)
+			data &= ~RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK;
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGLS)
+			data &= ~RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK;
+
+		/* disable cgcg and cgls in FSM */
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL, data);
+	}
+}
+
+static void gfx_v10_0_update_fine_grain_clock_gating(struct amdgpu_device *adev,
+						      bool enable)
+{
+	uint32_t def, data;
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_GFX_FGCG))
+		return;
+
+	if (enable) {
+		def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE);
+		/* unset FGCG override */
+		data &= ~RLC_CGTT_MGCG_OVERRIDE__GFXIP_FGCG_OVERRIDE_MASK;
+		/* update FGCG override bits */
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE, data);
+
+		def = data = RREG32_SOC15(GC, 0, mmRLC_CLK_CNTL);
+		/* unset RLC SRAM CLK GATER override */
+		data &= ~RLC_CLK_CNTL__RLC_SRAM_CLK_GATER_OVERRIDE_MASK;
+		/* update RLC SRAM CLK GATER override bits */
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CLK_CNTL, data);
+	} else {
+		def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE);
+		/* reset FGCG bits */
+		data |= RLC_CGTT_MGCG_OVERRIDE__GFXIP_FGCG_OVERRIDE_MASK;
+		/* disable FGCG*/
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE, data);
+
+		def = data = RREG32_SOC15(GC, 0, mmRLC_CLK_CNTL);
+		/* reset RLC SRAM CLK GATER bits */
+		data |= RLC_CLK_CNTL__RLC_SRAM_CLK_GATER_OVERRIDE_MASK;
+		/* disable RLC SRAM CLK*/
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CLK_CNTL, data);
+	}
+}
+
+static void gfx_v10_0_apply_medium_grain_clock_gating_workaround(struct amdgpu_device *adev)
+{
+	uint32_t reg_data = 0;
+	uint32_t reg_idx = 0;
+	uint32_t i;
+
+	const uint32_t tcp_ctrl_regs[] = {
+		mmCGTS_SA0_WGP00_CU0_TCP_CTRL_REG,
+		mmCGTS_SA0_WGP00_CU1_TCP_CTRL_REG,
+		mmCGTS_SA0_WGP01_CU0_TCP_CTRL_REG,
+		mmCGTS_SA0_WGP01_CU1_TCP_CTRL_REG,
+		mmCGTS_SA0_WGP02_CU0_TCP_CTRL_REG,
+		mmCGTS_SA0_WGP02_CU1_TCP_CTRL_REG,
+		mmCGTS_SA0_WGP10_CU0_TCP_CTRL_REG,
+		mmCGTS_SA0_WGP10_CU1_TCP_CTRL_REG,
+		mmCGTS_SA0_WGP11_CU0_TCP_CTRL_REG,
+		mmCGTS_SA0_WGP11_CU1_TCP_CTRL_REG,
+		mmCGTS_SA0_WGP12_CU0_TCP_CTRL_REG,
+		mmCGTS_SA0_WGP12_CU1_TCP_CTRL_REG,
+		mmCGTS_SA1_WGP00_CU0_TCP_CTRL_REG,
+		mmCGTS_SA1_WGP00_CU1_TCP_CTRL_REG,
+		mmCGTS_SA1_WGP01_CU0_TCP_CTRL_REG,
+		mmCGTS_SA1_WGP01_CU1_TCP_CTRL_REG,
+		mmCGTS_SA1_WGP02_CU0_TCP_CTRL_REG,
+		mmCGTS_SA1_WGP02_CU1_TCP_CTRL_REG,
+		mmCGTS_SA1_WGP10_CU0_TCP_CTRL_REG,
+		mmCGTS_SA1_WGP10_CU1_TCP_CTRL_REG,
+		mmCGTS_SA1_WGP11_CU0_TCP_CTRL_REG,
+		mmCGTS_SA1_WGP11_CU1_TCP_CTRL_REG,
+		mmCGTS_SA1_WGP12_CU0_TCP_CTRL_REG,
+		mmCGTS_SA1_WGP12_CU1_TCP_CTRL_REG
+	};
+
+	const uint32_t tcp_ctrl_regs_nv12[] = {
+		mmCGTS_SA0_WGP00_CU0_TCP_CTRL_REG,
+		mmCGTS_SA0_WGP00_CU1_TCP_CTRL_REG,
+		mmCGTS_SA0_WGP01_CU0_TCP_CTRL_REG,
+		mmCGTS_SA0_WGP01_CU1_TCP_CTRL_REG,
+		mmCGTS_SA0_WGP02_CU0_TCP_CTRL_REG,
+		mmCGTS_SA0_WGP02_CU1_TCP_CTRL_REG,
+		mmCGTS_SA0_WGP10_CU0_TCP_CTRL_REG,
+		mmCGTS_SA0_WGP10_CU1_TCP_CTRL_REG,
+		mmCGTS_SA0_WGP11_CU0_TCP_CTRL_REG,
+		mmCGTS_SA0_WGP11_CU1_TCP_CTRL_REG,
+		mmCGTS_SA1_WGP00_CU0_TCP_CTRL_REG,
+		mmCGTS_SA1_WGP00_CU1_TCP_CTRL_REG,
+		mmCGTS_SA1_WGP01_CU0_TCP_CTRL_REG,
+		mmCGTS_SA1_WGP01_CU1_TCP_CTRL_REG,
+		mmCGTS_SA1_WGP02_CU0_TCP_CTRL_REG,
+		mmCGTS_SA1_WGP02_CU1_TCP_CTRL_REG,
+		mmCGTS_SA1_WGP10_CU0_TCP_CTRL_REG,
+		mmCGTS_SA1_WGP10_CU1_TCP_CTRL_REG,
+		mmCGTS_SA1_WGP11_CU0_TCP_CTRL_REG,
+		mmCGTS_SA1_WGP11_CU1_TCP_CTRL_REG,
+	};
+
+	const uint32_t sm_ctlr_regs[] = {
+		mmCGTS_SA0_QUAD0_SM_CTRL_REG,
+		mmCGTS_SA0_QUAD1_SM_CTRL_REG,
+		mmCGTS_SA1_QUAD0_SM_CTRL_REG,
+		mmCGTS_SA1_QUAD1_SM_CTRL_REG
+	};
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 1, 2)) {
+		for (i = 0; i < ARRAY_SIZE(tcp_ctrl_regs_nv12); i++) {
+			reg_idx = adev->reg_offset[GC_HWIP][0][mmCGTS_SA0_WGP00_CU0_TCP_CTRL_REG_BASE_IDX] +
+				  tcp_ctrl_regs_nv12[i];
+			reg_data = RREG32(reg_idx);
+			reg_data |= CGTS_SA0_WGP00_CU0_TCP_CTRL_REG__TCPI_LS_OVERRIDE_MASK;
+			WREG32(reg_idx, reg_data);
+		}
+	} else {
+		for (i = 0; i < ARRAY_SIZE(tcp_ctrl_regs); i++) {
+			reg_idx = adev->reg_offset[GC_HWIP][0][mmCGTS_SA0_WGP00_CU0_TCP_CTRL_REG_BASE_IDX] +
+				  tcp_ctrl_regs[i];
+			reg_data = RREG32(reg_idx);
+			reg_data |= CGTS_SA0_WGP00_CU0_TCP_CTRL_REG__TCPI_LS_OVERRIDE_MASK;
+			WREG32(reg_idx, reg_data);
+		}
+	}
+
+	for (i = 0; i < ARRAY_SIZE(sm_ctlr_regs); i++) {
+		reg_idx = adev->reg_offset[GC_HWIP][0][mmCGTS_SA0_QUAD0_SM_CTRL_REG_BASE_IDX] +
+			  sm_ctlr_regs[i];
+		reg_data = RREG32(reg_idx);
+		reg_data &= ~CGTS_SA0_QUAD0_SM_CTRL_REG__SM_MODE_MASK;
+		reg_data |= 2 << CGTS_SA0_QUAD0_SM_CTRL_REG__SM_MODE__SHIFT;
+		WREG32(reg_idx, reg_data);
+	}
+}
+
+static int gfx_v10_0_update_gfx_clock_gating(struct amdgpu_device *adev,
+					    bool enable)
+{
+	amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
+
+	if (enable) {
+		/* enable FGCG firstly*/
+		gfx_v10_0_update_fine_grain_clock_gating(adev, enable);
+		/* CGCG/CGLS should be enabled after MGCG/MGLS
+		 * ===  MGCG + MGLS ===
+		 */
+		gfx_v10_0_update_medium_grain_clock_gating(adev, enable);
+		/* ===  CGCG /CGLS for GFX 3D Only === */
+		gfx_v10_0_update_3d_clock_gating(adev, enable);
+		/* ===  CGCG + CGLS === */
+		gfx_v10_0_update_coarse_grain_clock_gating(adev, enable);
+
+		if ((amdgpu_ip_version(adev, GC_HWIP, 0) ==
+		     IP_VERSION(10, 1, 10)) ||
+		    (amdgpu_ip_version(adev, GC_HWIP, 0) ==
+		     IP_VERSION(10, 1, 1)) ||
+		    (amdgpu_ip_version(adev, GC_HWIP, 0) ==
+		     IP_VERSION(10, 1, 2)))
+			gfx_v10_0_apply_medium_grain_clock_gating_workaround(adev);
+	} else {
+		/* CGCG/CGLS should be disabled before MGCG/MGLS
+		 * ===  CGCG + CGLS ===
+		 */
+		gfx_v10_0_update_coarse_grain_clock_gating(adev, enable);
+		/* ===  CGCG /CGLS for GFX 3D Only === */
+		gfx_v10_0_update_3d_clock_gating(adev, enable);
+		/* ===  MGCG + MGLS === */
+		gfx_v10_0_update_medium_grain_clock_gating(adev, enable);
+		/* disable fgcg at last*/
+		gfx_v10_0_update_fine_grain_clock_gating(adev, enable);
+	}
+
+	if (adev->cg_flags &
+	    (AMD_CG_SUPPORT_GFX_MGCG |
+	     AMD_CG_SUPPORT_GFX_CGLS |
+	     AMD_CG_SUPPORT_GFX_CGCG |
+	     AMD_CG_SUPPORT_GFX_3D_CGCG |
+	     AMD_CG_SUPPORT_GFX_3D_CGLS))
+		gfx_v10_0_enable_gui_idle_interrupt(adev, enable);
+
+	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
+
+	return 0;
+}
+
+static void gfx_v10_0_update_spm_vmid_internal(struct amdgpu_device *adev,
+					       unsigned int vmid)
+{
+	u32 reg, pre_data, data;
+
+	reg = SOC15_REG_OFFSET(GC, 0, mmRLC_SPM_MC_CNTL);
+	/* not for *_SOC15 */
+	if (amdgpu_sriov_is_pp_one_vf(adev) && !amdgpu_sriov_runtime(adev))
+		pre_data = RREG32_NO_KIQ(reg);
+	else
+		pre_data = RREG32(reg);
+
+	data = pre_data & (~RLC_SPM_MC_CNTL__RLC_SPM_VMID_MASK);
+	data |= (vmid & RLC_SPM_MC_CNTL__RLC_SPM_VMID_MASK) << RLC_SPM_MC_CNTL__RLC_SPM_VMID__SHIFT;
+
+	if (pre_data != data) {
+		if (amdgpu_sriov_is_pp_one_vf(adev) && !amdgpu_sriov_runtime(adev)) {
+			WREG32_SOC15_NO_KIQ(GC, 0, mmRLC_SPM_MC_CNTL, data);
+		} else
+			WREG32_SOC15(GC, 0, mmRLC_SPM_MC_CNTL, data);
+	}
+}
+
+static void gfx_v10_0_update_spm_vmid(struct amdgpu_device *adev, struct amdgpu_ring *ring, unsigned int vmid)
+{
+	amdgpu_gfx_off_ctrl(adev, false);
+
+	gfx_v10_0_update_spm_vmid_internal(adev, vmid);
+
+	amdgpu_gfx_off_ctrl(adev, true);
+}
+
+static bool gfx_v10_0_check_rlcg_range(struct amdgpu_device *adev,
+					uint32_t offset,
+					struct soc15_reg_rlcg *entries, int arr_size)
+{
+	int i;
+	uint32_t reg;
+
+	if (!entries)
+		return false;
+
+	for (i = 0; i < arr_size; i++) {
+		const struct soc15_reg_rlcg *entry;
+
+		entry = &entries[i];
+		reg = adev->reg_offset[entry->hwip][entry->instance][entry->segment] + entry->reg;
+		if (offset == reg)
+			return true;
+	}
+
+	return false;
+}
+
+static bool gfx_v10_0_is_rlcg_access_range(struct amdgpu_device *adev, u32 offset)
+{
+	return gfx_v10_0_check_rlcg_range(adev, offset, NULL, 0);
+}
+
+static void gfx_v10_cntl_power_gating(struct amdgpu_device *adev, bool enable)
+{
+	u32 data = RREG32_SOC15(GC, 0, mmRLC_PG_CNTL);
+
+	if (enable && (adev->pg_flags & AMD_PG_SUPPORT_GFX_PG))
+		data |= RLC_PG_CNTL__GFX_POWER_GATING_ENABLE_MASK;
+	else
+		data &= ~RLC_PG_CNTL__GFX_POWER_GATING_ENABLE_MASK;
+
+	WREG32_SOC15(GC, 0, mmRLC_PG_CNTL, data);
+
+	/*
+	 * CGPG enablement required and the register to program the hysteresis value
+	 * RLC_PG_DELAY_3.CGCG_ACTIVE_BEFORE_CGPG to the desired CGPG hysteresis value
+	 * in refclk count. Note that RLC FW is modified to take 16 bits from
+	 * RLC_PG_DELAY_3[15:0] as the hysteresis instead of just 8 bits.
+	 *
+	 * The recommendation from RLC team is setting RLC_PG_DELAY_3 to 200us as part)
+	 * of CGPG enablement starting point.
+	 * Power/performance team will optimize it and might give a new value later.
+	 */
+	if (enable && (adev->pg_flags & AMD_PG_SUPPORT_GFX_PG)) {
+		switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+		case IP_VERSION(10, 3, 1):
+		case IP_VERSION(10, 3, 3):
+		case IP_VERSION(10, 3, 6):
+		case IP_VERSION(10, 3, 7):
+			data = 0x4E20 & RLC_PG_DELAY_3__CGCG_ACTIVE_BEFORE_CGPG_MASK_Vangogh;
+			WREG32_SOC15(GC, 0, mmRLC_PG_DELAY_3, data);
+			break;
+		default:
+			break;
+		}
+	}
+}
+
+static void gfx_v10_cntl_pg(struct amdgpu_device *adev, bool enable)
+{
+	amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
+
+	gfx_v10_cntl_power_gating(adev, enable);
+
+	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
+}
+
+static const struct amdgpu_rlc_funcs gfx_v10_0_rlc_funcs = {
+	.is_rlc_enabled = gfx_v10_0_is_rlc_enabled,
+	.set_safe_mode = gfx_v10_0_set_safe_mode,
+	.unset_safe_mode = gfx_v10_0_unset_safe_mode,
+	.init = gfx_v10_0_rlc_init,
+	.get_csb_size = gfx_v10_0_get_csb_size,
+	.get_csb_buffer = gfx_v10_0_get_csb_buffer,
+	.resume = gfx_v10_0_rlc_resume,
+	.stop = gfx_v10_0_rlc_stop,
+	.reset = gfx_v10_0_rlc_reset,
+	.start = gfx_v10_0_rlc_start,
+	.update_spm_vmid = gfx_v10_0_update_spm_vmid,
+};
+
+static const struct amdgpu_rlc_funcs gfx_v10_0_rlc_funcs_sriov = {
+	.is_rlc_enabled = gfx_v10_0_is_rlc_enabled,
+	.set_safe_mode = gfx_v10_0_set_safe_mode,
+	.unset_safe_mode = gfx_v10_0_unset_safe_mode,
+	.init = gfx_v10_0_rlc_init,
+	.get_csb_size = gfx_v10_0_get_csb_size,
+	.get_csb_buffer = gfx_v10_0_get_csb_buffer,
+	.resume = gfx_v10_0_rlc_resume,
+	.stop = gfx_v10_0_rlc_stop,
+	.reset = gfx_v10_0_rlc_reset,
+	.start = gfx_v10_0_rlc_start,
+	.update_spm_vmid = gfx_v10_0_update_spm_vmid,
+	.is_rlcg_access_range = gfx_v10_0_is_rlcg_access_range,
+};
+
+static int gfx_v10_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = (state == AMD_PG_STATE_GATE);
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 1, 10):
+	case IP_VERSION(10, 1, 1):
+	case IP_VERSION(10, 1, 2):
+	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+		amdgpu_gfx_off_ctrl(adev, enable);
+		break;
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 3):
+	case IP_VERSION(10, 3, 6):
+	case IP_VERSION(10, 3, 7):
+		if (!enable)
+			amdgpu_gfx_off_ctrl(adev, false);
+
+		gfx_v10_cntl_pg(adev, enable);
+
+		if (enable)
+			amdgpu_gfx_off_ctrl(adev, true);
+
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int gfx_v10_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 1, 10):
+	case IP_VERSION(10, 1, 1):
+	case IP_VERSION(10, 1, 2):
+	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+	case IP_VERSION(10, 3, 6):
+	case IP_VERSION(10, 3, 3):
+	case IP_VERSION(10, 3, 7):
+		gfx_v10_0_update_gfx_clock_gating(adev,
+						 state == AMD_CG_STATE_GATE);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static void gfx_v10_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int data;
+
+	/* AMD_CG_SUPPORT_GFX_FGCG */
+	data = RREG32_KIQ(SOC15_REG_OFFSET(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE));
+	if (!(data & RLC_CGTT_MGCG_OVERRIDE__GFXIP_FGCG_OVERRIDE_MASK))
+		*flags |= AMD_CG_SUPPORT_GFX_FGCG;
+
+	/* AMD_CG_SUPPORT_GFX_MGCG */
+	data = RREG32_KIQ(SOC15_REG_OFFSET(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE));
+	if (!(data & RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGCG_OVERRIDE_MASK))
+		*flags |= AMD_CG_SUPPORT_GFX_MGCG;
+
+	/* AMD_CG_SUPPORT_GFX_CGCG */
+	data = RREG32_KIQ(SOC15_REG_OFFSET(GC, 0, mmRLC_CGCG_CGLS_CTRL));
+	if (data & RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_CGCG;
+
+	/* AMD_CG_SUPPORT_GFX_CGLS */
+	if (data & RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_CGLS;
+
+	/* AMD_CG_SUPPORT_GFX_RLC_LS */
+	data = RREG32_KIQ(SOC15_REG_OFFSET(GC, 0, mmRLC_MEM_SLP_CNTL));
+	if (data & RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_RLC_LS | AMD_CG_SUPPORT_GFX_MGLS;
+
+	/* AMD_CG_SUPPORT_GFX_CP_LS */
+	data = RREG32_KIQ(SOC15_REG_OFFSET(GC, 0, mmCP_MEM_SLP_CNTL));
+	if (data & CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_CP_LS | AMD_CG_SUPPORT_GFX_MGLS;
+
+	/* AMD_CG_SUPPORT_GFX_3D_CGCG */
+	data = RREG32_KIQ(SOC15_REG_OFFSET(GC, 0, mmRLC_CGCG_CGLS_CTRL_3D));
+	if (data & RLC_CGCG_CGLS_CTRL_3D__CGCG_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_3D_CGCG;
+
+	/* AMD_CG_SUPPORT_GFX_3D_CGLS */
+	if (data & RLC_CGCG_CGLS_CTRL_3D__CGLS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_3D_CGLS;
+}
+
+static u64 gfx_v10_0_ring_get_rptr_gfx(struct amdgpu_ring *ring)
+{
+	/* gfx10 is 32bit rptr*/
+	return *(uint32_t *)ring->rptr_cpu_addr;
+}
+
+static u64 gfx_v10_0_ring_get_wptr_gfx(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u64 wptr;
+
+	/* XXX check if swapping is necessary on BE */
+	if (ring->use_doorbell) {
+		wptr = atomic64_read((atomic64_t *)ring->wptr_cpu_addr);
+	} else {
+		wptr = RREG32_SOC15(GC, 0, mmCP_RB0_WPTR);
+		wptr += (u64)RREG32_SOC15(GC, 0, mmCP_RB0_WPTR_HI) << 32;
+	}
+
+	return wptr;
+}
+
+static void gfx_v10_0_ring_set_wptr_gfx(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr,
+			     ring->wptr);
+		WDOORBELL64(ring->doorbell_index, ring->wptr);
+	} else {
+		WREG32_SOC15(GC, 0, mmCP_RB0_WPTR,
+			     lower_32_bits(ring->wptr));
+		WREG32_SOC15(GC, 0, mmCP_RB0_WPTR_HI,
+			     upper_32_bits(ring->wptr));
+	}
+}
+
+static u64 gfx_v10_0_ring_get_rptr_compute(struct amdgpu_ring *ring)
+{
+	/* gfx10 hardware is 32bit rptr */
+	return *(uint32_t *)ring->rptr_cpu_addr;
+}
+
+static u64 gfx_v10_0_ring_get_wptr_compute(struct amdgpu_ring *ring)
+{
+	u64 wptr;
+
+	/* XXX check if swapping is necessary on BE */
+	if (ring->use_doorbell)
+		wptr = atomic64_read((atomic64_t *)ring->wptr_cpu_addr);
+	else
+		BUG();
+	return wptr;
+}
+
+static void gfx_v10_0_ring_set_wptr_compute(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr,
+			     ring->wptr);
+		WDOORBELL64(ring->doorbell_index, ring->wptr);
+	} else {
+		BUG(); /* only DOORBELL method supported on gfx10 now */
+	}
+}
+
+static void gfx_v10_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 ref_and_mask, reg_mem_engine;
+	const struct nbio_hdp_flush_reg *nbio_hf_reg = adev->nbio.hdp_flush_reg;
+
+	if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE) {
+		switch (ring->me) {
+		case 1:
+			ref_and_mask = nbio_hf_reg->ref_and_mask_cp2 << ring->pipe;
+			break;
+		case 2:
+			ref_and_mask = nbio_hf_reg->ref_and_mask_cp6 << ring->pipe;
+			break;
+		default:
+			return;
+		}
+		reg_mem_engine = 0;
+	} else {
+		ref_and_mask = nbio_hf_reg->ref_and_mask_cp0 << ring->pipe;
+		reg_mem_engine = 1; /* pfp */
+	}
+
+	gfx_v10_0_wait_reg_mem(ring, reg_mem_engine, 0, 1,
+			       adev->nbio.funcs->get_hdp_flush_req_offset(adev),
+			       adev->nbio.funcs->get_hdp_flush_done_offset(adev),
+			       ref_and_mask, ref_and_mask, 0x20);
+}
+
+static void gfx_v10_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
+				       struct amdgpu_job *job,
+				       struct amdgpu_ib *ib,
+				       uint32_t flags)
+{
+	unsigned int vmid = AMDGPU_JOB_GET_VMID(job);
+	u32 header, control = 0;
+
+	if (ib->flags & AMDGPU_IB_FLAG_CE)
+		header = PACKET3(PACKET3_INDIRECT_BUFFER_CNST, 2);
+	else
+		header = PACKET3(PACKET3_INDIRECT_BUFFER, 2);
+
+	control |= ib->length_dw | (vmid << 24);
+
+	if (ring->adev->gfx.mcbp && (ib->flags & AMDGPU_IB_FLAG_PREEMPT)) {
+		control |= INDIRECT_BUFFER_PRE_ENB(1);
+
+		if (flags & AMDGPU_IB_PREEMPTED)
+			control |= INDIRECT_BUFFER_PRE_RESUME(1);
+
+		if (!(ib->flags & AMDGPU_IB_FLAG_CE) && vmid)
+			gfx_v10_0_ring_emit_de_meta(ring,
+				    (!amdgpu_sriov_vf(ring->adev) && flags & AMDGPU_IB_PREEMPTED) ? true : false);
+	}
+
+	amdgpu_ring_write(ring, header);
+	BUG_ON(ib->gpu_addr & 0x3); /* Dword align */
+	amdgpu_ring_write(ring,
+#ifdef __BIG_ENDIAN
+		(2 << 0) |
+#endif
+		lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, control);
+}
+
+static void gfx_v10_0_ring_emit_ib_compute(struct amdgpu_ring *ring,
+					   struct amdgpu_job *job,
+					   struct amdgpu_ib *ib,
+					   uint32_t flags)
+{
+	unsigned int vmid = AMDGPU_JOB_GET_VMID(job);
+	u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vmid << 24);
+
+	/* Currently, there is a high possibility to get wave ID mismatch
+	 * between ME and GDS, leading to a hw deadlock, because ME generates
+	 * different wave IDs than the GDS expects. This situation happens
+	 * randomly when at least 5 compute pipes use GDS ordered append.
+	 * The wave IDs generated by ME are also wrong after suspend/resume.
+	 * Those are probably bugs somewhere else in the kernel driver.
+	 *
+	 * Writing GDS_COMPUTE_MAX_WAVE_ID resets wave ID counters in ME and
+	 * GDS to 0 for this ring (me/pipe).
+	 */
+	if (ib->flags & AMDGPU_IB_FLAG_RESET_GDS_MAX_WAVE_ID) {
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1));
+		amdgpu_ring_write(ring, mmGDS_COMPUTE_MAX_WAVE_ID);
+		amdgpu_ring_write(ring, ring->adev->gds.gds_compute_max_wave_id);
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2));
+	BUG_ON(ib->gpu_addr & 0x3); /* Dword align */
+	amdgpu_ring_write(ring,
+#ifdef __BIG_ENDIAN
+				(2 << 0) |
+#endif
+				lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, control);
+}
+
+static void gfx_v10_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
+				     u64 seq, unsigned int flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	bool int_sel = flags & AMDGPU_FENCE_FLAG_INT;
+
+	/* RELEASE_MEM - flush caches, send int */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_RELEASE_MEM, 6));
+	amdgpu_ring_write(ring, (PACKET3_RELEASE_MEM_GCR_SEQ |
+				 PACKET3_RELEASE_MEM_GCR_GL2_WB |
+				 PACKET3_RELEASE_MEM_GCR_GLM_INV | /* must be set with GLM_WB */
+				 PACKET3_RELEASE_MEM_GCR_GLM_WB |
+				 PACKET3_RELEASE_MEM_CACHE_POLICY(3) |
+				 PACKET3_RELEASE_MEM_EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) |
+				 PACKET3_RELEASE_MEM_EVENT_INDEX(5)));
+	amdgpu_ring_write(ring, (PACKET3_RELEASE_MEM_DATA_SEL(write64bit ? 2 : 1) |
+				 PACKET3_RELEASE_MEM_INT_SEL(int_sel ? 2 : 0)));
+
+	/*
+	 * the address should be Qword aligned if 64bit write, Dword
+	 * aligned if only send 32bit data low (discard data high)
+	 */
+	if (write64bit)
+		BUG_ON(addr & 0x7);
+	else
+		BUG_ON(addr & 0x3);
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+	amdgpu_ring_write(ring, upper_32_bits(seq));
+	amdgpu_ring_write(ring, 0);
+}
+
+static void gfx_v10_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	gfx_v10_0_wait_reg_mem(ring, usepfp, 1, 0, lower_32_bits(addr),
+			       upper_32_bits(addr), seq, 0xffffffff, 4);
+}
+
+static void gfx_v10_0_ring_invalidate_tlbs(struct amdgpu_ring *ring,
+				   uint16_t pasid, uint32_t flush_type,
+				   bool all_hub, uint8_t dst_sel)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_INVALIDATE_TLBS, 0));
+	amdgpu_ring_write(ring,
+			  PACKET3_INVALIDATE_TLBS_DST_SEL(dst_sel) |
+			  PACKET3_INVALIDATE_TLBS_ALL_HUB(all_hub) |
+			  PACKET3_INVALIDATE_TLBS_PASID(pasid) |
+			  PACKET3_INVALIDATE_TLBS_FLUSH_TYPE(flush_type));
+}
+
+static void gfx_v10_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					 unsigned int vmid, uint64_t pd_addr)
+{
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* compute doesn't have PFP */
+	if (ring->funcs->type == AMDGPU_RING_TYPE_GFX) {
+		/* sync PFP to ME, otherwise we might get invalid PFP reads */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_PFP_SYNC_ME, 0));
+		amdgpu_ring_write(ring, 0x0);
+	}
+}
+
+static void gfx_v10_0_ring_emit_fence_kiq(struct amdgpu_ring *ring, u64 addr,
+					  u64 seq, unsigned int flags)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	/* we only allocate 32bit for each seq wb address */
+	BUG_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	/* write fence seq to the "addr" */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				 WRITE_DATA_DST_SEL(5) | WR_CONFIRM));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+
+	if (flags & AMDGPU_FENCE_FLAG_INT) {
+		/* set register to trigger INT */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+		amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+					 WRITE_DATA_DST_SEL(0) | WR_CONFIRM));
+		amdgpu_ring_write(ring, SOC15_REG_OFFSET(GC, 0, mmCPC_INT_STATUS));
+		amdgpu_ring_write(ring, 0);
+		amdgpu_ring_write(ring, 0x20000000); /* src_id is 178 */
+	}
+}
+
+static void gfx_v10_0_ring_emit_sb(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+	amdgpu_ring_write(ring, 0);
+}
+
+static void gfx_v10_0_ring_emit_cntxcntl(struct amdgpu_ring *ring,
+					 uint32_t flags)
+{
+	uint32_t dw2 = 0;
+
+	if (ring->adev->gfx.mcbp)
+		gfx_v10_0_ring_emit_ce_meta(ring,
+				    (!amdgpu_sriov_vf(ring->adev) && flags & AMDGPU_IB_PREEMPTED) ? true : false);
+
+	dw2 |= 0x80000000; /* set load_enable otherwise this package is just NOPs */
+	if (flags & AMDGPU_HAVE_CTX_SWITCH) {
+		/* set load_global_config & load_global_uconfig */
+		dw2 |= 0x8001;
+		/* set load_cs_sh_regs */
+		dw2 |= 0x01000000;
+		/* set load_per_context_state & load_gfx_sh_regs for GFX */
+		dw2 |= 0x10002;
+
+		/* set load_ce_ram if preamble presented */
+		if (AMDGPU_PREAMBLE_IB_PRESENT & flags)
+			dw2 |= 0x10000000;
+	} else {
+		/* still load_ce_ram if this is the first time preamble presented
+		 * although there is no context switch happens.
+		 */
+		if (AMDGPU_PREAMBLE_IB_PRESENT_FIRST & flags)
+			dw2 |= 0x10000000;
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	amdgpu_ring_write(ring, dw2);
+	amdgpu_ring_write(ring, 0);
+}
+
+static unsigned int gfx_v10_0_ring_emit_init_cond_exec(struct amdgpu_ring *ring,
+						       uint64_t addr)
+{
+	unsigned int ret;
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_COND_EXEC, 3));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	/* discard following DWs if *cond_exec_gpu_addr==0 */
+	amdgpu_ring_write(ring, 0);
+	ret = ring->wptr & ring->buf_mask;
+	/* patch dummy value later */
+	amdgpu_ring_write(ring, 0);
+
+	return ret;
+}
+
+static int gfx_v10_0_ring_preempt_ib(struct amdgpu_ring *ring)
+{
+	int i, r = 0;
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[0];
+	struct amdgpu_ring *kiq_ring = &kiq->ring;
+	unsigned long flags;
+
+	if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
+		return -EINVAL;
+
+	spin_lock_irqsave(&kiq->ring_lock, flags);
+
+	if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->unmap_queues_size)) {
+		spin_unlock_irqrestore(&kiq->ring_lock, flags);
+		return -ENOMEM;
+	}
+
+	/* assert preemption condition */
+	amdgpu_ring_set_preempt_cond_exec(ring, false);
+
+	/* assert IB preemption, emit the trailing fence */
+	kiq->pmf->kiq_unmap_queues(kiq_ring, ring, PREEMPT_QUEUES_NO_UNMAP,
+				   ring->trail_fence_gpu_addr,
+				   ++ring->trail_seq);
+	amdgpu_ring_commit(kiq_ring);
+
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+
+	/* poll the trailing fence */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (ring->trail_seq ==
+		    le32_to_cpu(*(ring->trail_fence_cpu_addr)))
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout) {
+		r = -EINVAL;
+		DRM_ERROR("ring %d failed to preempt ib\n", ring->idx);
+	}
+
+	/* deassert preemption condition */
+	amdgpu_ring_set_preempt_cond_exec(ring, true);
+	return r;
+}
+
+static void gfx_v10_0_ring_emit_ce_meta(struct amdgpu_ring *ring, bool resume)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v10_ce_ib_state ce_payload = {0};
+	uint64_t offset, ce_payload_gpu_addr;
+	void *ce_payload_cpu_addr;
+	int cnt;
+
+	cnt = (sizeof(ce_payload) >> 2) + 4 - 2;
+
+	offset = offsetof(struct v10_gfx_meta_data, ce_payload);
+	ce_payload_gpu_addr = amdgpu_csa_vaddr(ring->adev) + offset;
+	ce_payload_cpu_addr = adev->virt.csa_cpu_addr + offset;
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, cnt));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(2) |
+				 WRITE_DATA_DST_SEL(8) |
+				 WR_CONFIRM) |
+				 WRITE_DATA_CACHE_POLICY(0));
+	amdgpu_ring_write(ring, lower_32_bits(ce_payload_gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ce_payload_gpu_addr));
+
+	if (resume)
+		amdgpu_ring_write_multiple(ring, ce_payload_cpu_addr,
+					   sizeof(ce_payload) >> 2);
+	else
+		amdgpu_ring_write_multiple(ring, (void *)&ce_payload,
+					   sizeof(ce_payload) >> 2);
+}
+
+static void gfx_v10_0_ring_emit_de_meta(struct amdgpu_ring *ring, bool resume)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v10_de_ib_state de_payload = {0};
+	uint64_t offset, gds_addr, de_payload_gpu_addr;
+	void *de_payload_cpu_addr;
+	int cnt;
+
+	offset = offsetof(struct v10_gfx_meta_data, de_payload);
+	de_payload_gpu_addr = amdgpu_csa_vaddr(ring->adev) + offset;
+	de_payload_cpu_addr = adev->virt.csa_cpu_addr + offset;
+
+	gds_addr = ALIGN(amdgpu_csa_vaddr(ring->adev) +
+			 AMDGPU_CSA_SIZE - adev->gds.gds_size,
+			 PAGE_SIZE);
+
+	de_payload.gds_backup_addrlo = lower_32_bits(gds_addr);
+	de_payload.gds_backup_addrhi = upper_32_bits(gds_addr);
+
+	cnt = (sizeof(de_payload) >> 2) + 4 - 2;
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, cnt));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(1) |
+				 WRITE_DATA_DST_SEL(8) |
+				 WR_CONFIRM) |
+				 WRITE_DATA_CACHE_POLICY(0));
+	amdgpu_ring_write(ring, lower_32_bits(de_payload_gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(de_payload_gpu_addr));
+
+	if (resume)
+		amdgpu_ring_write_multiple(ring, de_payload_cpu_addr,
+					   sizeof(de_payload) >> 2);
+	else
+		amdgpu_ring_write_multiple(ring, (void *)&de_payload,
+					   sizeof(de_payload) >> 2);
+}
+
+static void gfx_v10_0_ring_emit_frame_cntl(struct amdgpu_ring *ring, bool start,
+				    bool secure)
+{
+	uint32_t v = secure ? FRAME_TMZ : 0;
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_FRAME_CONTROL, 0));
+	amdgpu_ring_write(ring, v | FRAME_CMD(start ? 0 : 1));
+}
+
+static void gfx_v10_0_ring_emit_rreg(struct amdgpu_ring *ring, uint32_t reg,
+				     uint32_t reg_val_offs)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_COPY_DATA, 4));
+	amdgpu_ring_write(ring, 0 |	/* src: register*/
+				(5 << 8) |	/* dst: memory */
+				(1 << 20));	/* write confirm */
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, lower_32_bits(adev->wb.gpu_addr +
+				reg_val_offs * 4));
+	amdgpu_ring_write(ring, upper_32_bits(adev->wb.gpu_addr +
+				reg_val_offs * 4));
+}
+
+static void gfx_v10_0_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg,
+				   uint32_t val)
+{
+	uint32_t cmd = 0;
+
+	switch (ring->funcs->type) {
+	case AMDGPU_RING_TYPE_GFX:
+		cmd = WRITE_DATA_ENGINE_SEL(1) | WR_CONFIRM;
+		break;
+	case AMDGPU_RING_TYPE_KIQ:
+		cmd = (1 << 16); /* no inc addr */
+		break;
+	default:
+		cmd = WR_CONFIRM;
+		break;
+	}
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, cmd);
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, val);
+}
+
+static void gfx_v10_0_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+					uint32_t val, uint32_t mask)
+{
+	gfx_v10_0_wait_reg_mem(ring, 0, 0, 0, reg, 0, val, mask, 0x20);
+}
+
+static void gfx_v10_0_ring_emit_reg_write_reg_wait(struct amdgpu_ring *ring,
+						   uint32_t reg0, uint32_t reg1,
+						   uint32_t ref, uint32_t mask)
+{
+	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
+	struct amdgpu_device *adev = ring->adev;
+	bool fw_version_ok = false;
+
+	fw_version_ok = adev->gfx.cp_fw_write_wait;
+
+	if (fw_version_ok)
+		gfx_v10_0_wait_reg_mem(ring, usepfp, 0, 1, reg0, reg1,
+				       ref, mask, 0x20);
+	else
+		amdgpu_ring_emit_reg_write_reg_wait_helper(ring, reg0, reg1,
+							   ref, mask);
+}
+
+static void
+gfx_v10_0_set_gfx_eop_interrupt_state(struct amdgpu_device *adev,
+				      uint32_t me, uint32_t pipe,
+				      enum amdgpu_interrupt_state state)
+{
+	uint32_t cp_int_cntl, cp_int_cntl_reg;
+
+	if (!me) {
+		switch (pipe) {
+		case 0:
+			cp_int_cntl_reg = SOC15_REG_OFFSET(GC, 0, mmCP_INT_CNTL_RING0);
+			break;
+		case 1:
+			cp_int_cntl_reg = SOC15_REG_OFFSET(GC, 0, mmCP_INT_CNTL_RING1);
+			break;
+		default:
+			DRM_DEBUG("invalid pipe %d\n", pipe);
+			return;
+		}
+	} else {
+		DRM_DEBUG("invalid me %d\n", me);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+		cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+					    TIME_STAMP_INT_ENABLE, 0);
+		WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+		cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+					    TIME_STAMP_INT_ENABLE, 1);
+		WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+		break;
+	default:
+		break;
+	}
+}
+
+static void gfx_v10_0_set_compute_eop_interrupt_state(struct amdgpu_device *adev,
+						     int me, int pipe,
+						     enum amdgpu_interrupt_state state)
+{
+	u32 mec_int_cntl, mec_int_cntl_reg;
+
+	/*
+	 * amdgpu controls only the first MEC. That's why this function only
+	 * handles the setting of interrupts for this specific MEC. All other
+	 * pipes' interrupts are set by amdkfd.
+	 */
+
+	if (me == 1) {
+		switch (pipe) {
+		case 0:
+			mec_int_cntl_reg = SOC15_REG_OFFSET(GC, 0, mmCP_ME1_PIPE0_INT_CNTL);
+			break;
+		case 1:
+			mec_int_cntl_reg = SOC15_REG_OFFSET(GC, 0, mmCP_ME1_PIPE1_INT_CNTL);
+			break;
+		case 2:
+			mec_int_cntl_reg = SOC15_REG_OFFSET(GC, 0, mmCP_ME1_PIPE2_INT_CNTL);
+			break;
+		case 3:
+			mec_int_cntl_reg = SOC15_REG_OFFSET(GC, 0, mmCP_ME1_PIPE3_INT_CNTL);
+			break;
+		default:
+			DRM_DEBUG("invalid pipe %d\n", pipe);
+			return;
+		}
+	} else {
+		DRM_DEBUG("invalid me %d\n", me);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		mec_int_cntl = RREG32_SOC15_IP(GC, mec_int_cntl_reg);
+		mec_int_cntl = REG_SET_FIELD(mec_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+					     TIME_STAMP_INT_ENABLE, 0);
+		WREG32_SOC15_IP(GC, mec_int_cntl_reg, mec_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		mec_int_cntl = RREG32_SOC15_IP(GC, mec_int_cntl_reg);
+		mec_int_cntl = REG_SET_FIELD(mec_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+					     TIME_STAMP_INT_ENABLE, 1);
+		WREG32_SOC15_IP(GC, mec_int_cntl_reg, mec_int_cntl);
+		break;
+	default:
+		break;
+	}
+}
+
+static int gfx_v10_0_set_eop_interrupt_state(struct amdgpu_device *adev,
+					    struct amdgpu_irq_src *src,
+					    unsigned int type,
+					    enum amdgpu_interrupt_state state)
+{
+	switch (type) {
+	case AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP:
+		gfx_v10_0_set_gfx_eop_interrupt_state(adev, 0, 0, state);
+		break;
+	case AMDGPU_CP_IRQ_GFX_ME0_PIPE1_EOP:
+		gfx_v10_0_set_gfx_eop_interrupt_state(adev, 0, 1, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP:
+		gfx_v10_0_set_compute_eop_interrupt_state(adev, 1, 0, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE1_EOP:
+		gfx_v10_0_set_compute_eop_interrupt_state(adev, 1, 1, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE2_EOP:
+		gfx_v10_0_set_compute_eop_interrupt_state(adev, 1, 2, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE3_EOP:
+		gfx_v10_0_set_compute_eop_interrupt_state(adev, 1, 3, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE0_EOP:
+		gfx_v10_0_set_compute_eop_interrupt_state(adev, 2, 0, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE1_EOP:
+		gfx_v10_0_set_compute_eop_interrupt_state(adev, 2, 1, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE2_EOP:
+		gfx_v10_0_set_compute_eop_interrupt_state(adev, 2, 2, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE3_EOP:
+		gfx_v10_0_set_compute_eop_interrupt_state(adev, 2, 3, state);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int gfx_v10_0_eop_irq(struct amdgpu_device *adev,
+			     struct amdgpu_irq_src *source,
+			     struct amdgpu_iv_entry *entry)
+{
+	int i;
+	u8 me_id, pipe_id, queue_id;
+	struct amdgpu_ring *ring;
+
+	DRM_DEBUG("IH: CP EOP\n");
+
+	me_id = (entry->ring_id & 0x0c) >> 2;
+	pipe_id = (entry->ring_id & 0x03) >> 0;
+	queue_id = (entry->ring_id & 0x70) >> 4;
+
+	switch (me_id) {
+	case 0:
+		if (pipe_id == 0)
+			amdgpu_fence_process(&adev->gfx.gfx_ring[0]);
+		else
+			amdgpu_fence_process(&adev->gfx.gfx_ring[1]);
+		break;
+	case 1:
+	case 2:
+		for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+			ring = &adev->gfx.compute_ring[i];
+			/* Per-queue interrupt is supported for MEC starting from VI.
+			 * The interrupt can only be enabled/disabled per pipe instead
+			 * of per queue.
+			 */
+			if ((ring->me == me_id) &&
+			    (ring->pipe == pipe_id) &&
+			    (ring->queue == queue_id))
+				amdgpu_fence_process(ring);
+		}
+		break;
+	}
+
+	return 0;
+}
+
+static int gfx_v10_0_set_priv_reg_fault_state(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      unsigned int type,
+					      enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl_reg, cp_int_cntl;
+	int i, j;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+	case AMDGPU_IRQ_STATE_ENABLE:
+		for (i = 0; i < adev->gfx.me.num_me; i++) {
+			for (j = 0; j < adev->gfx.me.num_pipe_per_me; j++) {
+				cp_int_cntl_reg = gfx_v10_0_get_cpg_int_cntl(adev, i, j);
+
+				if (cp_int_cntl_reg) {
+					cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+					cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+								    PRIV_REG_INT_ENABLE,
+								    state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+					WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+				}
+			}
+		}
+		for (i = 0; i < adev->gfx.mec.num_mec; i++) {
+			for (j = 0; j < adev->gfx.mec.num_pipe_per_mec; j++) {
+				/* MECs start at 1 */
+				cp_int_cntl_reg = gfx_v10_0_get_cpc_int_cntl(adev, i + 1, j);
+
+				if (cp_int_cntl_reg) {
+					cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+					cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+								    PRIV_REG_INT_ENABLE,
+								    state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+					WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+				}
+			}
+		}
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gfx_v10_0_set_bad_op_fault_state(struct amdgpu_device *adev,
+					    struct amdgpu_irq_src *source,
+					    unsigned type,
+					    enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl_reg, cp_int_cntl;
+	int i, j;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+	case AMDGPU_IRQ_STATE_ENABLE:
+		for (i = 0; i < adev->gfx.me.num_me; i++) {
+			for (j = 0; j < adev->gfx.me.num_pipe_per_me; j++) {
+				cp_int_cntl_reg = gfx_v10_0_get_cpg_int_cntl(adev, i, j);
+
+				if (cp_int_cntl_reg) {
+					cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+					cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+								    OPCODE_ERROR_INT_ENABLE,
+								    state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+					WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+				}
+			}
+		}
+		for (i = 0; i < adev->gfx.mec.num_mec; i++) {
+			for (j = 0; j < adev->gfx.mec.num_pipe_per_mec; j++) {
+				/* MECs start at 1 */
+				cp_int_cntl_reg = gfx_v10_0_get_cpc_int_cntl(adev, i + 1, j);
+
+				if (cp_int_cntl_reg) {
+					cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+					cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+								    OPCODE_ERROR_INT_ENABLE,
+								    state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+					WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+				}
+			}
+		}
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int gfx_v10_0_set_priv_inst_fault_state(struct amdgpu_device *adev,
+					       struct amdgpu_irq_src *source,
+					       unsigned int type,
+					       enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl_reg, cp_int_cntl;
+	int i, j;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+	case AMDGPU_IRQ_STATE_ENABLE:
+		for (i = 0; i < adev->gfx.me.num_me; i++) {
+			for (j = 0; j < adev->gfx.me.num_pipe_per_me; j++) {
+				cp_int_cntl_reg = gfx_v10_0_get_cpg_int_cntl(adev, i, j);
+
+				if (cp_int_cntl_reg) {
+					cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+					cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+								    PRIV_INSTR_INT_ENABLE,
+								    state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+					WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+				}
+			}
+		}
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static void gfx_v10_0_handle_priv_fault(struct amdgpu_device *adev,
+					struct amdgpu_iv_entry *entry)
+{
+	u8 me_id, pipe_id, queue_id;
+	struct amdgpu_ring *ring;
+	int i;
+
+	me_id = (entry->ring_id & 0x0c) >> 2;
+	pipe_id = (entry->ring_id & 0x03) >> 0;
+	queue_id = (entry->ring_id & 0x70) >> 4;
+
+	switch (me_id) {
+	case 0:
+		for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+			ring = &adev->gfx.gfx_ring[i];
+			if (ring->me == me_id && ring->pipe == pipe_id &&
+			    ring->queue == queue_id)
+				drm_sched_fault(&ring->sched);
+		}
+		break;
+	case 1:
+	case 2:
+		for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+			ring = &adev->gfx.compute_ring[i];
+			if (ring->me == me_id && ring->pipe == pipe_id &&
+			    ring->queue == queue_id)
+				drm_sched_fault(&ring->sched);
+		}
+		break;
+	default:
+		BUG();
+	}
+}
+
+static int gfx_v10_0_priv_reg_irq(struct amdgpu_device *adev,
+				  struct amdgpu_irq_src *source,
+				  struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal register access in command stream\n");
+	gfx_v10_0_handle_priv_fault(adev, entry);
+	return 0;
+}
+
+static int gfx_v10_0_bad_op_irq(struct amdgpu_device *adev,
+				struct amdgpu_irq_src *source,
+				struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal opcode in command stream \n");
+	gfx_v10_0_handle_priv_fault(adev, entry);
+	return 0;
+}
+
+static int gfx_v10_0_priv_inst_irq(struct amdgpu_device *adev,
+				   struct amdgpu_irq_src *source,
+				   struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal instruction in command stream\n");
+	gfx_v10_0_handle_priv_fault(adev, entry);
+	return 0;
+}
+
+static int gfx_v10_0_kiq_set_interrupt_state(struct amdgpu_device *adev,
+					     struct amdgpu_irq_src *src,
+					     unsigned int type,
+					     enum amdgpu_interrupt_state state)
+{
+	uint32_t tmp, target;
+	struct amdgpu_ring *ring = &(adev->gfx.kiq[0].ring);
+
+	if (ring->me == 1)
+		target = SOC15_REG_OFFSET(GC, 0, mmCP_ME1_PIPE0_INT_CNTL);
+	else
+		target = SOC15_REG_OFFSET(GC, 0, mmCP_ME2_PIPE0_INT_CNTL);
+	target += ring->pipe;
+
+	switch (type) {
+	case AMDGPU_CP_KIQ_IRQ_DRIVER0:
+		if (state == AMDGPU_IRQ_STATE_DISABLE) {
+			tmp = RREG32_SOC15(GC, 0, mmCPC_INT_CNTL);
+			tmp = REG_SET_FIELD(tmp, CPC_INT_CNTL,
+					    GENERIC2_INT_ENABLE, 0);
+			WREG32_SOC15(GC, 0, mmCPC_INT_CNTL, tmp);
+
+			tmp = RREG32_SOC15_IP(GC, target);
+			tmp = REG_SET_FIELD(tmp, CP_ME2_PIPE0_INT_CNTL,
+					    GENERIC2_INT_ENABLE, 0);
+			WREG32_SOC15_IP(GC, target, tmp);
+		} else {
+			tmp = RREG32_SOC15(GC, 0, mmCPC_INT_CNTL);
+			tmp = REG_SET_FIELD(tmp, CPC_INT_CNTL,
+					    GENERIC2_INT_ENABLE, 1);
+			WREG32_SOC15(GC, 0, mmCPC_INT_CNTL, tmp);
+
+			tmp = RREG32_SOC15_IP(GC, target);
+			tmp = REG_SET_FIELD(tmp, CP_ME2_PIPE0_INT_CNTL,
+					    GENERIC2_INT_ENABLE, 1);
+			WREG32_SOC15_IP(GC, target, tmp);
+		}
+		break;
+	default:
+		BUG(); /* kiq only support GENERIC2_INT now */
+		break;
+	}
+	return 0;
+}
+
+static int gfx_v10_0_kiq_irq(struct amdgpu_device *adev,
+			     struct amdgpu_irq_src *source,
+			     struct amdgpu_iv_entry *entry)
+{
+	u8 me_id, pipe_id, queue_id;
+	struct amdgpu_ring *ring = &(adev->gfx.kiq[0].ring);
+
+	me_id = (entry->ring_id & 0x0c) >> 2;
+	pipe_id = (entry->ring_id & 0x03) >> 0;
+	queue_id = (entry->ring_id & 0x70) >> 4;
+	DRM_DEBUG("IH: CPC GENERIC2_INT, me:%d, pipe:%d, queue:%d\n",
+		   me_id, pipe_id, queue_id);
+
+	amdgpu_fence_process(ring);
+	return 0;
+}
+
+static void gfx_v10_0_emit_mem_sync(struct amdgpu_ring *ring)
+{
+	const unsigned int gcr_cntl =
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GL2_INV(1) |
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GL2_WB(1) |
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GLM_INV(1) |
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GLM_WB(1) |
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GL1_INV(1) |
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GLV_INV(1) |
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GLK_INV(1) |
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GLI_INV(1);
+
+	/* ACQUIRE_MEM - make one or more surfaces valid for use by the subsequent operations */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_ACQUIRE_MEM, 6));
+	amdgpu_ring_write(ring, 0); /* CP_COHER_CNTL */
+	amdgpu_ring_write(ring, 0xffffffff);  /* CP_COHER_SIZE */
+	amdgpu_ring_write(ring, 0xffffff);  /* CP_COHER_SIZE_HI */
+	amdgpu_ring_write(ring, 0); /* CP_COHER_BASE */
+	amdgpu_ring_write(ring, 0);  /* CP_COHER_BASE_HI */
+	amdgpu_ring_write(ring, 0x0000000A); /* POLL_INTERVAL */
+	amdgpu_ring_write(ring, gcr_cntl); /* GCR_CNTL */
+}
+
+static void gfx_v10_ring_insert_nop(struct amdgpu_ring *ring, uint32_t num_nop)
+{
+	/* Header itself is a NOP packet */
+	if (num_nop == 1) {
+		amdgpu_ring_write(ring, ring->funcs->nop);
+		return;
+	}
+
+	/* Max HW optimization till 0x3ffe, followed by remaining one NOP at a time*/
+	amdgpu_ring_write(ring, PACKET3(PACKET3_NOP, min(num_nop - 2, 0x3ffe)));
+
+	/* Header is at index 0, followed by num_nops - 1 NOP packet's */
+	amdgpu_ring_insert_nop(ring, num_nop - 1);
+}
+
+static int gfx_v10_0_reset_kgq(struct amdgpu_ring *ring,
+			       unsigned int vmid,
+			       struct amdgpu_fence *timedout_fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[0];
+	struct amdgpu_ring *kiq_ring = &kiq->ring;
+	unsigned long flags;
+	u32 tmp;
+	u64 addr;
+	int r;
+
+	if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
+		return -EINVAL;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+
+	spin_lock_irqsave(&kiq->ring_lock, flags);
+
+	if (amdgpu_ring_alloc(kiq_ring, 5 + 7 + 7)) {
+		spin_unlock_irqrestore(&kiq->ring_lock, flags);
+		return -ENOMEM;
+	}
+
+	addr = amdgpu_bo_gpu_offset(ring->mqd_obj) +
+		offsetof(struct v10_gfx_mqd, cp_gfx_hqd_active);
+	tmp = REG_SET_FIELD(0, CP_VMID_RESET, RESET_REQUEST, 1 << vmid);
+	if (ring->pipe == 0)
+		tmp = REG_SET_FIELD(tmp, CP_VMID_RESET, PIPE0_QUEUES, 1 << ring->queue);
+	else
+		tmp = REG_SET_FIELD(tmp, CP_VMID_RESET, PIPE1_QUEUES, 1 << ring->queue);
+
+	gfx_v10_0_ring_emit_wreg(kiq_ring,
+				 SOC15_REG_OFFSET(GC, 0, mmCP_VMID_RESET), tmp);
+	gfx_v10_0_wait_reg_mem(kiq_ring, 0, 1, 0,
+			       lower_32_bits(addr), upper_32_bits(addr),
+			       0, 1, 0x20);
+	gfx_v10_0_ring_emit_reg_wait(kiq_ring,
+				     SOC15_REG_OFFSET(GC, 0, mmCP_VMID_RESET), 0, 0xffffffff);
+	amdgpu_ring_commit(kiq_ring);
+	r = amdgpu_ring_test_ring(kiq_ring);
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+	if (r)
+		return r;
+
+	r = gfx_v10_0_kgq_init_queue(ring, true);
+	if (r) {
+		DRM_ERROR("fail to init kgq\n");
+		return r;
+	}
+
+	spin_lock_irqsave(&kiq->ring_lock, flags);
+
+	if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->map_queues_size)) {
+		spin_unlock_irqrestore(&kiq->ring_lock, flags);
+		return -ENOMEM;
+	}
+	kiq->pmf->kiq_map_queues(kiq_ring, ring);
+	amdgpu_ring_commit(kiq_ring);
+	r = amdgpu_ring_test_ring(kiq_ring);
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+	if (r)
+		return r;
+
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static int gfx_v10_0_reset_kcq(struct amdgpu_ring *ring,
+			       unsigned int vmid,
+			       struct amdgpu_fence *timedout_fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[0];
+	struct amdgpu_ring *kiq_ring = &kiq->ring;
+	unsigned long flags;
+	int i, r;
+
+	if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
+		return -EINVAL;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+
+	spin_lock_irqsave(&kiq->ring_lock, flags);
+
+	if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->unmap_queues_size)) {
+		spin_unlock_irqrestore(&kiq->ring_lock, flags);
+		return -ENOMEM;
+	}
+
+	kiq->pmf->kiq_unmap_queues(kiq_ring, ring, RESET_QUEUES,
+				   0, 0);
+	amdgpu_ring_commit(kiq_ring);
+	r = amdgpu_ring_test_ring(kiq_ring);
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+	if (r)
+		return r;
+
+	/* make sure dequeue is complete*/
+	amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
+	mutex_lock(&adev->srbm_mutex);
+	nv_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (!(RREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE) & 1))
+			break;
+		udelay(1);
+	}
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+	nv_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
+	if (r) {
+		dev_err(adev->dev, "fail to wait on hqd deactivate\n");
+		return r;
+	}
+
+	r = gfx_v10_0_kcq_init_queue(ring, true);
+	if (r) {
+		dev_err(adev->dev, "fail to init kcq\n");
+		return r;
+	}
+
+	spin_lock_irqsave(&kiq->ring_lock, flags);
+	if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->map_queues_size)) {
+		spin_unlock_irqrestore(&kiq->ring_lock, flags);
+		return -ENOMEM;
+	}
+	kiq->pmf->kiq_map_queues(kiq_ring, ring);
+	amdgpu_ring_commit(kiq_ring);
+	r = amdgpu_ring_test_ring(kiq_ring);
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+	if (r)
+		return r;
+
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static void gfx_v10_ip_print(struct amdgpu_ip_block *ip_block, struct drm_printer *p)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	uint32_t i, j, k, reg, index = 0;
+	uint32_t reg_count = ARRAY_SIZE(gc_reg_list_10_1);
+
+	if (!adev->gfx.ip_dump_core)
+		return;
+
+	for (i = 0; i < reg_count; i++)
+		drm_printf(p, "%-50s \t 0x%08x\n",
+			   gc_reg_list_10_1[i].reg_name,
+			   adev->gfx.ip_dump_core[i]);
+
+	/* print compute queue registers for all instances */
+	if (!adev->gfx.ip_dump_compute_queues)
+		return;
+
+	reg_count = ARRAY_SIZE(gc_cp_reg_list_10);
+	drm_printf(p, "\nnum_mec: %d num_pipe: %d num_queue: %d\n",
+		   adev->gfx.mec.num_mec,
+		   adev->gfx.mec.num_pipe_per_mec,
+		   adev->gfx.mec.num_queue_per_pipe);
+
+	for (i = 0; i < adev->gfx.mec.num_mec; i++) {
+		for (j = 0; j < adev->gfx.mec.num_pipe_per_mec; j++) {
+			for (k = 0; k < adev->gfx.mec.num_queue_per_pipe; k++) {
+				drm_printf(p, "\nmec %d, pipe %d, queue %d\n", i, j, k);
+				for (reg = 0; reg < reg_count; reg++) {
+					if (i && gc_cp_reg_list_10[reg].reg_offset == mmCP_MEC_ME1_HEADER_DUMP)
+						drm_printf(p, "%-50s \t 0x%08x\n",
+							   "mmCP_MEC_ME2_HEADER_DUMP",
+							   adev->gfx.ip_dump_compute_queues[index + reg]);
+					else
+						drm_printf(p, "%-50s \t 0x%08x\n",
+							   gc_cp_reg_list_10[reg].reg_name,
+							   adev->gfx.ip_dump_compute_queues[index + reg]);
+				}
+				index += reg_count;
+			}
+		}
+	}
+
+	/* print gfx queue registers for all instances */
+	if (!adev->gfx.ip_dump_gfx_queues)
+		return;
+
+	index = 0;
+	reg_count = ARRAY_SIZE(gc_gfx_queue_reg_list_10);
+	drm_printf(p, "\nnum_me: %d num_pipe: %d num_queue: %d\n",
+		   adev->gfx.me.num_me,
+		   adev->gfx.me.num_pipe_per_me,
+		   adev->gfx.me.num_queue_per_pipe);
+
+	for (i = 0; i < adev->gfx.me.num_me; i++) {
+		for (j = 0; j < adev->gfx.me.num_pipe_per_me; j++) {
+			for (k = 0; k < adev->gfx.me.num_queue_per_pipe; k++) {
+				drm_printf(p, "\nme %d, pipe %d, queue %d\n", i, j, k);
+				for (reg = 0; reg < reg_count; reg++) {
+					drm_printf(p, "%-50s \t 0x%08x\n",
+						   gc_gfx_queue_reg_list_10[reg].reg_name,
+						   adev->gfx.ip_dump_gfx_queues[index + reg]);
+				}
+				index += reg_count;
+			}
+		}
+	}
+}
+
+static void gfx_v10_ip_dump(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	uint32_t i, j, k, reg, index = 0;
+	uint32_t reg_count = ARRAY_SIZE(gc_reg_list_10_1);
+
+	if (!adev->gfx.ip_dump_core)
+		return;
+
+	amdgpu_gfx_off_ctrl(adev, false);
+	for (i = 0; i < reg_count; i++)
+		adev->gfx.ip_dump_core[i] = RREG32(SOC15_REG_ENTRY_OFFSET(gc_reg_list_10_1[i]));
+	amdgpu_gfx_off_ctrl(adev, true);
+
+	/* dump compute queue registers for all instances */
+	if (!adev->gfx.ip_dump_compute_queues)
+		return;
+
+	reg_count = ARRAY_SIZE(gc_cp_reg_list_10);
+	amdgpu_gfx_off_ctrl(adev, false);
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < adev->gfx.mec.num_mec; i++) {
+		for (j = 0; j < adev->gfx.mec.num_pipe_per_mec; j++) {
+			for (k = 0; k < adev->gfx.mec.num_queue_per_pipe; k++) {
+				/* ME0 is for GFX so start from 1 for CP */
+				nv_grbm_select(adev, adev->gfx.me.num_me + i, j, k, 0);
+
+				for (reg = 0; reg < reg_count; reg++) {
+					if (i && gc_cp_reg_list_10[reg].reg_offset == mmCP_MEC_ME1_HEADER_DUMP)
+						adev->gfx.ip_dump_compute_queues[index + reg] =
+							RREG32(SOC15_REG_OFFSET(GC, 0, mmCP_MEC_ME2_HEADER_DUMP));
+					else
+						adev->gfx.ip_dump_compute_queues[index + reg] =
+							RREG32(SOC15_REG_ENTRY_OFFSET(
+								       gc_cp_reg_list_10[reg]));
+				}
+				index += reg_count;
+			}
+		}
+	}
+	nv_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+	amdgpu_gfx_off_ctrl(adev, true);
+
+	/* dump gfx queue registers for all instances */
+	if (!adev->gfx.ip_dump_gfx_queues)
+		return;
+
+	index = 0;
+	reg_count = ARRAY_SIZE(gc_gfx_queue_reg_list_10);
+	amdgpu_gfx_off_ctrl(adev, false);
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < adev->gfx.me.num_me; i++) {
+		for (j = 0; j < adev->gfx.me.num_pipe_per_me; j++) {
+			for (k = 0; k < adev->gfx.me.num_queue_per_pipe; k++) {
+				nv_grbm_select(adev, i, j, k, 0);
+
+				for (reg = 0; reg < reg_count; reg++) {
+					adev->gfx.ip_dump_gfx_queues[index + reg] =
+						RREG32(SOC15_REG_ENTRY_OFFSET(
+							gc_gfx_queue_reg_list_10[reg]));
+				}
+				index += reg_count;
+			}
+		}
+	}
+	nv_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+	amdgpu_gfx_off_ctrl(adev, true);
+}
+
+static void gfx_v10_0_ring_emit_cleaner_shader(struct amdgpu_ring *ring)
+{
+	/* Emit the cleaner shader */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_RUN_CLEANER_SHADER, 0));
+	amdgpu_ring_write(ring, 0);  /* RESERVED field, programmed to zero */
+}
+
+static void gfx_v10_0_ring_begin_use(struct amdgpu_ring *ring)
+{
+	amdgpu_gfx_profile_ring_begin_use(ring);
+
+	amdgpu_gfx_enforce_isolation_ring_begin_use(ring);
+}
+
+static void gfx_v10_0_ring_end_use(struct amdgpu_ring *ring)
+{
+	amdgpu_gfx_profile_ring_end_use(ring);
+
+	amdgpu_gfx_enforce_isolation_ring_end_use(ring);
+}
+
+static const struct amd_ip_funcs gfx_v10_0_ip_funcs = {
+	.name = "gfx_v10_0",
+	.early_init = gfx_v10_0_early_init,
+	.late_init = gfx_v10_0_late_init,
+	.sw_init = gfx_v10_0_sw_init,
+	.sw_fini = gfx_v10_0_sw_fini,
+	.hw_init = gfx_v10_0_hw_init,
+	.hw_fini = gfx_v10_0_hw_fini,
+	.suspend = gfx_v10_0_suspend,
+	.resume = gfx_v10_0_resume,
+	.is_idle = gfx_v10_0_is_idle,
+	.wait_for_idle = gfx_v10_0_wait_for_idle,
+	.soft_reset = gfx_v10_0_soft_reset,
+	.set_clockgating_state = gfx_v10_0_set_clockgating_state,
+	.set_powergating_state = gfx_v10_0_set_powergating_state,
+	.get_clockgating_state = gfx_v10_0_get_clockgating_state,
+	.dump_ip_state = gfx_v10_ip_dump,
+	.print_ip_state = gfx_v10_ip_print,
+};
+
+static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_gfx = {
+	.type = AMDGPU_RING_TYPE_GFX,
+	.align_mask = 0xff,
+	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
+	.support_64bit_ptrs = true,
+	.secure_submission_supported = true,
+	.get_rptr = gfx_v10_0_ring_get_rptr_gfx,
+	.get_wptr = gfx_v10_0_ring_get_wptr_gfx,
+	.set_wptr = gfx_v10_0_ring_set_wptr_gfx,
+	.emit_frame_size = /* totally 242 maximum if 16 IBs */
+		5 + /* COND_EXEC */
+		7 + /* PIPELINE_SYNC */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 5 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 +
+		4 + /* VM_FLUSH */
+		8 + /* FENCE for VM_FLUSH */
+		20 + /* GDS switch */
+		4 + /* double SWITCH_BUFFER,
+		     * the first COND_EXEC jump to the place
+		     * just prior to this double SWITCH_BUFFER
+		     */
+		5 + /* COND_EXEC */
+		7 + /* HDP_flush */
+		4 + /* VGT_flush */
+		14 + /*	CE_META */
+		31 + /*	DE_META */
+		3 + /* CNTX_CTRL */
+		5 + /* HDP_INVL */
+		8 + 8 + /* FENCE x2 */
+		2 + /* SWITCH_BUFFER */
+		8 + /* gfx_v10_0_emit_mem_sync */
+		2, /* gfx_v10_0_ring_emit_cleaner_shader */
+	.emit_ib_size =	4, /* gfx_v10_0_ring_emit_ib_gfx */
+	.emit_ib = gfx_v10_0_ring_emit_ib_gfx,
+	.emit_fence = gfx_v10_0_ring_emit_fence,
+	.emit_pipeline_sync = gfx_v10_0_ring_emit_pipeline_sync,
+	.emit_vm_flush = gfx_v10_0_ring_emit_vm_flush,
+	.emit_gds_switch = gfx_v10_0_ring_emit_gds_switch,
+	.emit_hdp_flush = gfx_v10_0_ring_emit_hdp_flush,
+	.test_ring = gfx_v10_0_ring_test_ring,
+	.test_ib = gfx_v10_0_ring_test_ib,
+	.insert_nop = gfx_v10_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_switch_buffer = gfx_v10_0_ring_emit_sb,
+	.emit_cntxcntl = gfx_v10_0_ring_emit_cntxcntl,
+	.init_cond_exec = gfx_v10_0_ring_emit_init_cond_exec,
+	.preempt_ib = gfx_v10_0_ring_preempt_ib,
+	.emit_frame_cntl = gfx_v10_0_ring_emit_frame_cntl,
+	.emit_wreg = gfx_v10_0_ring_emit_wreg,
+	.emit_reg_wait = gfx_v10_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = gfx_v10_0_ring_emit_reg_write_reg_wait,
+	.emit_mem_sync = gfx_v10_0_emit_mem_sync,
+	.reset = gfx_v10_0_reset_kgq,
+	.emit_cleaner_shader = gfx_v10_0_ring_emit_cleaner_shader,
+	.begin_use = gfx_v10_0_ring_begin_use,
+	.end_use = gfx_v10_0_ring_end_use,
+};
+
+static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_compute = {
+	.type = AMDGPU_RING_TYPE_COMPUTE,
+	.align_mask = 0xff,
+	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
+	.support_64bit_ptrs = true,
+	.get_rptr = gfx_v10_0_ring_get_rptr_compute,
+	.get_wptr = gfx_v10_0_ring_get_wptr_compute,
+	.set_wptr = gfx_v10_0_ring_set_wptr_compute,
+	.emit_frame_size =
+		20 + /* gfx_v10_0_ring_emit_gds_switch */
+		7 + /* gfx_v10_0_ring_emit_hdp_flush */
+		5 + /* hdp invalidate */
+		7 + /* gfx_v10_0_ring_emit_pipeline_sync */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 5 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 +
+		2 + /* gfx_v10_0_ring_emit_vm_flush */
+		8 + 8 + 8 + /* gfx_v10_0_ring_emit_fence x3 for user fence, vm fence */
+		8 + /* gfx_v10_0_emit_mem_sync */
+		2, /* gfx_v10_0_ring_emit_cleaner_shader */
+	.emit_ib_size =	7, /* gfx_v10_0_ring_emit_ib_compute */
+	.emit_ib = gfx_v10_0_ring_emit_ib_compute,
+	.emit_fence = gfx_v10_0_ring_emit_fence,
+	.emit_pipeline_sync = gfx_v10_0_ring_emit_pipeline_sync,
+	.emit_vm_flush = gfx_v10_0_ring_emit_vm_flush,
+	.emit_gds_switch = gfx_v10_0_ring_emit_gds_switch,
+	.emit_hdp_flush = gfx_v10_0_ring_emit_hdp_flush,
+	.test_ring = gfx_v10_0_ring_test_ring,
+	.test_ib = gfx_v10_0_ring_test_ib,
+	.insert_nop = gfx_v10_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_wreg = gfx_v10_0_ring_emit_wreg,
+	.emit_reg_wait = gfx_v10_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = gfx_v10_0_ring_emit_reg_write_reg_wait,
+	.emit_mem_sync = gfx_v10_0_emit_mem_sync,
+	.reset = gfx_v10_0_reset_kcq,
+	.emit_cleaner_shader = gfx_v10_0_ring_emit_cleaner_shader,
+	.begin_use = gfx_v10_0_ring_begin_use,
+	.end_use = gfx_v10_0_ring_end_use,
+};
+
+static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_kiq = {
+	.type = AMDGPU_RING_TYPE_KIQ,
+	.align_mask = 0xff,
+	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
+	.support_64bit_ptrs = true,
+	.get_rptr = gfx_v10_0_ring_get_rptr_compute,
+	.get_wptr = gfx_v10_0_ring_get_wptr_compute,
+	.set_wptr = gfx_v10_0_ring_set_wptr_compute,
+	.emit_frame_size =
+		20 + /* gfx_v10_0_ring_emit_gds_switch */
+		7 + /* gfx_v10_0_ring_emit_hdp_flush */
+		5 + /*hdp invalidate */
+		7 + /* gfx_v10_0_ring_emit_pipeline_sync */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 5 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 +
+		8 + 8 + 8, /* gfx_v10_0_ring_emit_fence_kiq x3 for user fence, vm fence */
+	.emit_ib_size =	7, /* gfx_v10_0_ring_emit_ib_compute */
+	.emit_ib = gfx_v10_0_ring_emit_ib_compute,
+	.emit_fence = gfx_v10_0_ring_emit_fence_kiq,
+	.test_ring = gfx_v10_0_ring_test_ring,
+	.test_ib = gfx_v10_0_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_rreg = gfx_v10_0_ring_emit_rreg,
+	.emit_wreg = gfx_v10_0_ring_emit_wreg,
+	.emit_reg_wait = gfx_v10_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = gfx_v10_0_ring_emit_reg_write_reg_wait,
+};
+
+static void gfx_v10_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	adev->gfx.kiq[0].ring.funcs = &gfx_v10_0_ring_funcs_kiq;
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+		adev->gfx.gfx_ring[i].funcs = &gfx_v10_0_ring_funcs_gfx;
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++)
+		adev->gfx.compute_ring[i].funcs = &gfx_v10_0_ring_funcs_compute;
+}
+
+static const struct amdgpu_irq_src_funcs gfx_v10_0_eop_irq_funcs = {
+	.set = gfx_v10_0_set_eop_interrupt_state,
+	.process = gfx_v10_0_eop_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v10_0_priv_reg_irq_funcs = {
+	.set = gfx_v10_0_set_priv_reg_fault_state,
+	.process = gfx_v10_0_priv_reg_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v10_0_bad_op_irq_funcs = {
+	.set = gfx_v10_0_set_bad_op_fault_state,
+	.process = gfx_v10_0_bad_op_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v10_0_priv_inst_irq_funcs = {
+	.set = gfx_v10_0_set_priv_inst_fault_state,
+	.process = gfx_v10_0_priv_inst_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v10_0_kiq_irq_funcs = {
+	.set = gfx_v10_0_kiq_set_interrupt_state,
+	.process = gfx_v10_0_kiq_irq,
+};
+
+static void gfx_v10_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->gfx.eop_irq.num_types = AMDGPU_CP_IRQ_LAST;
+	adev->gfx.eop_irq.funcs = &gfx_v10_0_eop_irq_funcs;
+
+	adev->gfx.kiq[0].irq.num_types = AMDGPU_CP_KIQ_IRQ_LAST;
+	adev->gfx.kiq[0].irq.funcs = &gfx_v10_0_kiq_irq_funcs;
+
+	adev->gfx.priv_reg_irq.num_types = 1;
+	adev->gfx.priv_reg_irq.funcs = &gfx_v10_0_priv_reg_irq_funcs;
+
+	adev->gfx.bad_op_irq.num_types = 1;
+	adev->gfx.bad_op_irq.funcs = &gfx_v10_0_bad_op_irq_funcs;
+
+	adev->gfx.priv_inst_irq.num_types = 1;
+	adev->gfx.priv_inst_irq.funcs = &gfx_v10_0_priv_inst_irq_funcs;
+}
+
+static void gfx_v10_0_set_rlc_funcs(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 1, 10):
+	case IP_VERSION(10, 1, 1):
+	case IP_VERSION(10, 1, 3):
+	case IP_VERSION(10, 1, 4):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+	case IP_VERSION(10, 3, 6):
+	case IP_VERSION(10, 3, 3):
+	case IP_VERSION(10, 3, 7):
+		adev->gfx.rlc.funcs = &gfx_v10_0_rlc_funcs;
+		break;
+	case IP_VERSION(10, 1, 2):
+	case IP_VERSION(10, 3, 0):
+		adev->gfx.rlc.funcs = &gfx_v10_0_rlc_funcs_sriov;
+		break;
+	default:
+		break;
+	}
+}
+
+static void gfx_v10_0_set_gds_init(struct amdgpu_device *adev)
+{
+	unsigned int total_cu = adev->gfx.config.max_cu_per_sh *
+			    adev->gfx.config.max_sh_per_se *
+			    adev->gfx.config.max_shader_engines;
+
+	adev->gds.gds_size = 0x10000;
+	adev->gds.gds_compute_max_wave_id = total_cu * 32 - 1;
+	adev->gds.gws_size = 64;
+	adev->gds.oa_size = 16;
+}
+
+static void gfx_v10_0_set_mqd_funcs(struct amdgpu_device *adev)
+{
+	/* set gfx eng mqd */
+	adev->mqds[AMDGPU_HW_IP_GFX].mqd_size =
+		sizeof(struct v10_gfx_mqd);
+	adev->mqds[AMDGPU_HW_IP_GFX].init_mqd =
+		gfx_v10_0_gfx_mqd_init;
+	/* set compute eng mqd */
+	adev->mqds[AMDGPU_HW_IP_COMPUTE].mqd_size =
+		sizeof(struct v10_compute_mqd);
+	adev->mqds[AMDGPU_HW_IP_COMPUTE].init_mqd =
+		gfx_v10_0_compute_mqd_init;
+}
+
+static void gfx_v10_0_set_user_wgp_inactive_bitmap_per_sh(struct amdgpu_device *adev,
+							  u32 bitmap)
+{
+	u32 data;
+
+	if (!bitmap)
+		return;
+
+	data = bitmap << GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_WGPS__SHIFT;
+	data &= GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_WGPS_MASK;
+
+	WREG32_SOC15(GC, 0, mmGC_USER_SHADER_ARRAY_CONFIG, data);
+}
+
+static u32 gfx_v10_0_get_wgp_active_bitmap_per_sh(struct amdgpu_device *adev)
+{
+	u32 disabled_mask =
+		~amdgpu_gfx_create_bitmask(adev->gfx.config.max_cu_per_sh >> 1);
+	u32 efuse_setting = 0;
+	u32 vbios_setting = 0;
+
+	efuse_setting = RREG32_SOC15(GC, 0, mmCC_GC_SHADER_ARRAY_CONFIG);
+	efuse_setting &= CC_GC_SHADER_ARRAY_CONFIG__INACTIVE_WGPS_MASK;
+	efuse_setting >>= CC_GC_SHADER_ARRAY_CONFIG__INACTIVE_WGPS__SHIFT;
+
+	vbios_setting = RREG32_SOC15(GC, 0, mmGC_USER_SHADER_ARRAY_CONFIG);
+	vbios_setting &= GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_WGPS_MASK;
+	vbios_setting >>= GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_WGPS__SHIFT;
+
+	disabled_mask |= efuse_setting | vbios_setting;
+
+	return (~disabled_mask);
+}
+
+static u32 gfx_v10_0_get_cu_active_bitmap_per_sh(struct amdgpu_device *adev)
+{
+	u32 wgp_idx, wgp_active_bitmap;
+	u32 cu_bitmap_per_wgp, cu_active_bitmap;
+
+	wgp_active_bitmap = gfx_v10_0_get_wgp_active_bitmap_per_sh(adev);
+	cu_active_bitmap = 0;
+
+	for (wgp_idx = 0; wgp_idx < 16; wgp_idx++) {
+		/* if there is one WGP enabled, it means 2 CUs will be enabled */
+		cu_bitmap_per_wgp = 3 << (2 * wgp_idx);
+		if (wgp_active_bitmap & (1 << wgp_idx))
+			cu_active_bitmap |= cu_bitmap_per_wgp;
+	}
+
+	return cu_active_bitmap;
+}
+
+static int gfx_v10_0_get_cu_info(struct amdgpu_device *adev,
+				 struct amdgpu_cu_info *cu_info)
+{
+	int i, j, k, counter, active_cu_number = 0;
+	u32 mask, bitmap, ao_bitmap, ao_cu_mask = 0;
+	unsigned int disable_masks[4 * 2];
+
+	if (!adev || !cu_info)
+		return -EINVAL;
+
+	amdgpu_gfx_parse_disable_cu(disable_masks, 4, 2);
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			bitmap = i * adev->gfx.config.max_sh_per_se + j;
+			if (((amdgpu_ip_version(adev, GC_HWIP, 0) ==
+			      IP_VERSION(10, 3, 0)) ||
+			     (amdgpu_ip_version(adev, GC_HWIP, 0) ==
+			      IP_VERSION(10, 3, 3)) ||
+			     (amdgpu_ip_version(adev, GC_HWIP, 0) ==
+			      IP_VERSION(10, 3, 6)) ||
+			     (amdgpu_ip_version(adev, GC_HWIP, 0) ==
+			      IP_VERSION(10, 3, 7))) &&
+			    ((gfx_v10_3_get_disabled_sa(adev) >> bitmap) & 1))
+				continue;
+			mask = 1;
+			ao_bitmap = 0;
+			counter = 0;
+			gfx_v10_0_select_se_sh(adev, i, j, 0xffffffff, 0);
+			if (i < 4 && j < 2)
+				gfx_v10_0_set_user_wgp_inactive_bitmap_per_sh(
+					adev, disable_masks[i * 2 + j]);
+			bitmap = gfx_v10_0_get_cu_active_bitmap_per_sh(adev);
+			cu_info->bitmap[0][i][j] = bitmap;
+
+			for (k = 0; k < adev->gfx.config.max_cu_per_sh; k++) {
+				if (bitmap & mask) {
+					if (counter < adev->gfx.config.max_cu_per_sh)
+						ao_bitmap |= mask;
+					counter++;
+				}
+				mask <<= 1;
+			}
+			active_cu_number += counter;
+			if (i < 2 && j < 2)
+				ao_cu_mask |= (ao_bitmap << (i * 16 + j * 8));
+			cu_info->ao_cu_bitmap[i][j] = ao_bitmap;
+		}
+	}
+	gfx_v10_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	cu_info->number = active_cu_number;
+	cu_info->ao_cu_mask = ao_cu_mask;
+	cu_info->simd_per_cu = NUM_SIMD_PER_CU;
+
+	return 0;
+}
+
+static u32 gfx_v10_3_get_disabled_sa(struct amdgpu_device *adev)
+{
+	uint32_t efuse_setting, vbios_setting, disabled_sa, max_sa_mask;
+
+	efuse_setting = RREG32_SOC15(GC, 0, mmCC_GC_SA_UNIT_DISABLE);
+	efuse_setting &= CC_GC_SA_UNIT_DISABLE__SA_DISABLE_MASK;
+	efuse_setting >>= CC_GC_SA_UNIT_DISABLE__SA_DISABLE__SHIFT;
+
+	vbios_setting = RREG32_SOC15(GC, 0, mmGC_USER_SA_UNIT_DISABLE);
+	vbios_setting &= GC_USER_SA_UNIT_DISABLE__SA_DISABLE_MASK;
+	vbios_setting >>= GC_USER_SA_UNIT_DISABLE__SA_DISABLE__SHIFT;
+
+	max_sa_mask = amdgpu_gfx_create_bitmask(adev->gfx.config.max_sh_per_se *
+						adev->gfx.config.max_shader_engines);
+	disabled_sa = efuse_setting | vbios_setting;
+	disabled_sa &= max_sa_mask;
+
+	return disabled_sa;
+}
+
+static void gfx_v10_3_program_pbb_mode(struct amdgpu_device *adev)
+{
+	uint32_t max_sa_per_se, max_sa_per_se_mask, max_shader_engines;
+	uint32_t disabled_sa_mask, se_index, disabled_sa_per_se;
+
+	disabled_sa_mask = gfx_v10_3_get_disabled_sa(adev);
+
+	max_sa_per_se = adev->gfx.config.max_sh_per_se;
+	max_sa_per_se_mask = (1 << max_sa_per_se) - 1;
+	max_shader_engines = adev->gfx.config.max_shader_engines;
+
+	for (se_index = 0; max_shader_engines > se_index; se_index++) {
+		disabled_sa_per_se = disabled_sa_mask >> (se_index * max_sa_per_se);
+		disabled_sa_per_se &= max_sa_per_se_mask;
+		if (disabled_sa_per_se == max_sa_per_se_mask) {
+			WREG32_FIELD15(GC, 0, PA_SC_ENHANCE_3, FORCE_PBB_WORKLOAD_MODE_TO_ZERO, 1);
+			break;
+		}
+	}
+}
+
+static void gfx_v10_3_set_power_brake_sequence(struct amdgpu_device *adev)
+{
+	WREG32_SOC15(GC, 0, mmGRBM_GFX_INDEX,
+		     (0x1 << GRBM_GFX_INDEX__SA_BROADCAST_WRITES__SHIFT) |
+		     (0x1 << GRBM_GFX_INDEX__INSTANCE_BROADCAST_WRITES__SHIFT) |
+		     (0x1 << GRBM_GFX_INDEX__SE_BROADCAST_WRITES__SHIFT));
+
+	WREG32_SOC15(GC, 0, mmGC_CAC_IND_INDEX, ixPWRBRK_STALL_PATTERN_CTRL);
+	WREG32_SOC15(GC, 0, mmGC_CAC_IND_DATA,
+		     (0x1 << PWRBRK_STALL_PATTERN_CTRL__PWRBRK_STEP_INTERVAL__SHIFT) |
+		     (0x12 << PWRBRK_STALL_PATTERN_CTRL__PWRBRK_BEGIN_STEP__SHIFT) |
+		     (0x13 << PWRBRK_STALL_PATTERN_CTRL__PWRBRK_END_STEP__SHIFT) |
+		     (0xf << PWRBRK_STALL_PATTERN_CTRL__PWRBRK_THROTTLE_PATTERN_BIT_NUMS__SHIFT));
+
+	WREG32_SOC15(GC, 0, mmGC_THROTTLE_CTRL_Sienna_Cichlid,
+		     (0x1 << GC_THROTTLE_CTRL__PWRBRK_STALL_EN__SHIFT) |
+		     (0x1 << GC_THROTTLE_CTRL__PATTERN_MODE__SHIFT) |
+		     (0x5 << GC_THROTTLE_CTRL__RELEASE_STEP_INTERVAL__SHIFT));
+
+	WREG32_SOC15(GC, 0, mmDIDT_IND_INDEX, ixDIDT_SQ_THROTTLE_CTRL);
+
+	WREG32_SOC15(GC, 0, mmDIDT_IND_DATA,
+		     (0x1 << DIDT_SQ_THROTTLE_CTRL__PWRBRK_STALL_EN__SHIFT));
+}
+
+const struct amdgpu_ip_block_version gfx_v10_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_GFX,
+	.major = 10,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &gfx_v10_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.h b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.h
new file mode 100644
index 000000000000..b442e50324d0
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2019 dvanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFX_V10_0_H__
+#define __GFX_V10_0_H__
+
+extern const struct amdgpu_ip_block_version gfx_v10_0_ip_block;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0_cleaner_shader.h b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0_cleaner_shader.h
new file mode 100644
index 000000000000..f67569ccf9f6
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0_cleaner_shader.h
@@ -0,0 +1,91 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2025 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/* Define the cleaner shader gfx_10_1_10 */
+static const u32 gfx_10_1_10_cleaner_shader_hex[] = {
+	0xb0804004, 0xbf8a0000,
+	0xbf068100, 0xbf840023,
+	0xbe8203b8, 0xbefc0380,
+	0x7e008480, 0x7e028480,
+	0x7e048480, 0x7e068480,
+	0x7e088480, 0x7e0a8480,
+	0x7e0c8480, 0x7e0e8480,
+	0xbefc0302, 0x80828802,
+	0xbf84fff5, 0xbe8203ff,
+	0x80000000, 0x87020102,
+	0xbf840012, 0xbefe03c1,
+	0xbeff03c1, 0xd7650001,
+	0x0001007f, 0xd7660001,
+	0x0002027e, 0x16020288,
+	0xbe8203bf, 0xbefc03c1,
+	0xd9382000, 0x00020201,
+	0xd9386040, 0x00040401,
+	0xd70f6a01, 0x000202ff,
+	0x00000400, 0x80828102,
+	0xbf84fff7, 0xbefc03ff,
+	0x00000068, 0xbe803000,
+	0xbe813000, 0xbe823000,
+	0xbe833000, 0x80fc847c,
+	0xbf84fffa, 0xbeea0480,
+	0xbeec0480, 0xbeee0480,
+	0xbef00480, 0xbef20480,
+	0xbef40480, 0xbef60480,
+	0xbef80480, 0xbefa0480,
+	0xbf810000, 0xbf9f0000,
+	0xbf9f0000, 0xbf9f0000,
+	0xbf9f0000, 0xbf9f0000,
+};
+
+/* Define the cleaner shader gfx_10_3_0 */
+static const u32 gfx_10_3_0_cleaner_shader_hex[] = {
+	0xb0804004, 0xbf8a0000,
+	0xbe8203b8, 0xbefc0380,
+	0x7e008480, 0x7e028480,
+	0x7e048480, 0x7e068480,
+	0x7e088480, 0x7e0a8480,
+	0x7e0c8480, 0x7e0e8480,
+	0xbefc0302, 0x80828802,
+	0xbf84fff5, 0xbe8203ff,
+	0x80000000, 0x87020002,
+	0xbf840012, 0xbefe03c1,
+	0xbeff03c1, 0xd7650001,
+	0x0001007f, 0xd7660001,
+	0x0002027e, 0x16020288,
+	0xbe8203bf, 0xbefc03c1,
+	0xd9382000, 0x00020201,
+	0xd9386040, 0x00040401,
+	0xd70f6a01, 0x000202ff,
+	0x00000400, 0x80828102,
+	0xbf84fff7, 0xbefc03ff,
+	0x00000068, 0xbe803080,
+	0xbe813080, 0xbe823080,
+	0xbe833080, 0x80fc847c,
+	0xbf84fffa, 0xbeea0480,
+	0xbeec0480, 0xbeee0480,
+	0xbef00480, 0xbef20480,
+	0xbef40480, 0xbef60480,
+	0xbef80480, 0xbefa0480,
+	0xbf810000, 0xbf9f0000,
+	0xbf9f0000, 0xbf9f0000,
+	0xbf9f0000, 0xbf9f0000,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_1_10_cleaner_shader.asm b/drivers/gpu/drm/amd/amdgpu/gfx_v10_1_10_cleaner_shader.asm
new file mode 100644
index 000000000000..54f7ed9e2801
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_1_10_cleaner_shader.asm
@@ -0,0 +1,125 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2025 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+// This shader is to clean LDS, SGPRs and VGPRs. It is  first 64 Dwords or 256 bytes of 256 Dwords cleaner shader.
+
+// GFX10.1 : Clear SGPRs, VGPRs and LDS
+//   Launch 32 waves per CU (16 per SIMD) as a workgroup (threadgroup) to fill every wave slot
+//   Waves are "wave32" and have 64 VGPRs each, which uses all 1024 VGPRs per SIMD
+//   Waves are launched in "CU" mode, and the workgroup shares 64KB of LDS (half of the WGP's LDS)
+//      It takes 2 workgroups to use all of LDS: one on each CU of the WGP
+//   Each wave clears SGPRs 0 - 107
+//   Each wave clears VGPRs 0 - 63
+//   The first wave of the workgroup clears its 64KB of LDS
+//   The shader starts with "S_BARRIER" to ensure SPI has launched all waves of the workgroup
+//       before any wave in the workgroup could end.  Without this, it is possible not all SGPRs get cleared.
+
+
+shader main
+  asic(GFX10.1)
+  type(CS)
+  wave_size(32)
+// Note: original source code from SQ team
+//
+// Create 32 waves in a threadgroup (CS waves)
+// Each allocates 64 VGPRs
+// The workgroup allocates all of LDS (64kbytes)
+//
+// Takes about 2500 clocks to run.
+//   (theorhetical fastest = 1024clks vgpr + 640lds = 1660 clks)
+//
+  S_BARRIER
+  s_cmp_eq_u32 s0, 1                                // Bit0 is set, sgpr0 is set then clear VGPRS and LDS as FW set COMPUTE_USER_DATA_0
+  s_cbranch_scc0  label_0023                        // Clean VGPRs and LDS if sgpr0 of wave is set, scc = (s0 == 1)
+
+  s_mov_b32     s2, 0x00000038  // Loop 64/8=8 times  (loop unrolled for performance)
+  s_mov_b32     m0, 0
+  //
+  // CLEAR VGPRs
+  //
+label_0005:
+  v_movreld_b32     v0, 0
+  v_movreld_b32     v1, 0
+  v_movreld_b32     v2, 0
+  v_movreld_b32     v3, 0
+  v_movreld_b32     v4, 0
+  v_movreld_b32     v5, 0
+  v_movreld_b32     v6, 0
+  v_movreld_b32     v7, 0
+  s_mov_b32         m0, s2
+  s_sub_u32     s2, s2, 8
+  s_cbranch_scc0  label_0005
+  //
+  s_mov_b32     s2, 0x80000000                       // Bit31 is first_wave
+  s_and_b32     s2, s2, s1                           // sgpr0 has tg_size (first_wave) term as in ucode only COMPUTE_PGM_RSRC2.tg_size_en is set
+  s_cbranch_scc0  label_0023                         // Clean LDS if its first wave of ThreadGroup/WorkGroup
+  // CLEAR LDS
+  //
+  s_mov_b32 exec_lo, 0xffffffff
+  s_mov_b32 exec_hi, 0xffffffff
+  v_mbcnt_lo_u32_b32  v1, exec_hi, 0          // Set V1 to thread-ID (0..63)
+  v_mbcnt_hi_u32_b32  v1, exec_lo, v1        // Set V1 to thread-ID (0..63)
+  v_mul_u32_u24  v1, 0x00000008, v1          // * 8, so each thread is a double-dword address (8byte)
+  s_mov_b32     s2, 0x00000003f                    // 64 loop iterations
+  s_mov_b32     m0, 0xffffffff
+  // Clear all of LDS space
+  // Each FirstWave of WorkGroup clears 64kbyte block
+
+label_001F:
+  ds_write2_b64  v1, v[2:3], v[2:3] offset1:32
+  ds_write2_b64  v1, v[4:5], v[4:5] offset0:64 offset1:96
+  v_add_co_u32     v1, vcc, 0x00000400, v1
+  s_sub_u32     s2, s2, 1
+  s_cbranch_scc0  label_001F
+
+  //
+  // CLEAR SGPRs
+  //
+label_0023:
+  s_mov_b32     m0, 0x00000068  // Loop 108/4=27 times  (loop unrolled for performance)
+label_sgpr_loop:
+  s_movreld_b32     s0, s0
+  s_movreld_b32     s1, s0
+  s_movreld_b32     s2, s0
+  s_movreld_b32     s3, s0
+  s_sub_u32         m0, m0, 4
+  s_cbranch_scc0  label_sgpr_loop
+
+  //clear vcc
+  s_mov_b64 vcc, 0          //clear vcc
+  //s_setreg_imm32_b32 hw_reg_shader_flat_scratch_lo, 0   //clear  flat scratch lo SGPR
+  //s_setreg_imm32_b32 hw_reg_shader_flat_scratch_hi, 0    //clear  flat scratch hi SGPR
+  s_mov_b64 ttmp0, 0        //Clear ttmp0 and ttmp1
+  s_mov_b64 ttmp2, 0        //Clear ttmp2 and ttmp3
+  s_mov_b64 ttmp4, 0        //Clear ttmp4 and ttmp5
+  s_mov_b64 ttmp6, 0        //Clear ttmp6 and ttmp7
+  s_mov_b64 ttmp8, 0        //Clear ttmp8 and ttmp9
+  s_mov_b64 ttmp10, 0       //Clear ttmp10 and ttmp11
+  s_mov_b64 ttmp12, 0       //Clear ttmp12 and ttmp13
+  s_mov_b64 ttmp14, 0       //Clear ttmp14 and ttmp15
+
+ s_endpgm
+
+end
+
+
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_3_0_cleaner_shader.asm b/drivers/gpu/drm/amd/amdgpu/gfx_v10_3_0_cleaner_shader.asm
new file mode 100644
index 000000000000..0e1c246166c0
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_3_0_cleaner_shader.asm
@@ -0,0 +1,124 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2025 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+// This shader is to clean LDS, SGPRs and VGPRs. It is  first 64 Dwords or 256 bytes of 192 Dwords cleaner shader.
+//To turn this shader program on for complitaion change this to main and lower shader main to main_1
+
+// GFX10.3 : Clear SGPRs, VGPRs and LDS
+//   Launch 32 waves per CU (16 per SIMD) as a workgroup (threadgroup) to fill every wave slot
+//   Waves are "wave32" and have 64 VGPRs each, which uses all 1024 VGPRs per SIMD
+//   Waves are launched in "CU" mode, and the workgroup shares 64KB of LDS (half of the WGP's LDS)
+//      It takes 2 workgroups to use all of LDS: one on each CU of the WGP
+//   Each wave clears SGPRs 0 - 107
+//   Each wave clears VGPRs 0 - 63
+//   The first wave of the workgroup clears its 64KB of LDS
+//   The shader starts with "S_BARRIER" to ensure SPI has launched all waves of the workgroup
+//       before any wave in the workgroup could end.  Without this, it is possible not all SGPRs get cleared.
+
+
+shader main
+  asic(GFX10)
+  type(CS)
+  wave_size(32)
+// Note: original source code from SQ team
+
+//
+// Create 32 waves in a threadgroup (CS waves)
+// Each allocates 64 VGPRs
+// The workgroup allocates all of LDS (64kbytes)
+//
+// Takes about 2500 clocks to run.
+//   (theorhetical fastest = 1024clks vgpr + 640lds = 1660 clks)
+//
+  S_BARRIER
+  s_mov_b32     s2, 0x00000038  // Loop 64/8=8 times  (loop unrolled for performance)
+  s_mov_b32     m0, 0
+  //
+  // CLEAR VGPRs
+  //
+label_0005:
+  v_movreld_b32     v0, 0
+  v_movreld_b32     v1, 0
+  v_movreld_b32     v2, 0
+  v_movreld_b32     v3, 0
+  v_movreld_b32     v4, 0
+  v_movreld_b32     v5, 0
+  v_movreld_b32     v6, 0
+  v_movreld_b32     v7, 0
+  s_mov_b32         m0, s2
+  s_sub_u32     s2, s2, 8
+  s_cbranch_scc0  label_0005
+  //
+  s_mov_b32     s2, 0x80000000                     // Bit31 is first_wave
+  s_and_b32     s2, s2, s0                                  // sgpr0 has tg_size (first_wave) term as in ucode only COMPUTE_PGM_RSRC2.tg_size_en is set
+  s_cbranch_scc0  label_0023                         // Clean LDS if its first wave of ThreadGroup/WorkGroup
+  // CLEAR LDS
+  //
+  s_mov_b32 exec_lo, 0xffffffff
+  s_mov_b32 exec_hi, 0xffffffff
+  v_mbcnt_lo_u32_b32  v1, exec_hi, 0          // Set V1 to thread-ID (0..63)
+  v_mbcnt_hi_u32_b32  v1, exec_lo, v1        // Set V1 to thread-ID (0..63)
+  v_mul_u32_u24  v1, 0x00000008, v1          // * 8, so each thread is a double-dword address (8byte)
+  s_mov_b32     s2, 0x00000003f                    // 64 loop iterations
+  s_mov_b32     m0, 0xffffffff
+  // Clear all of LDS space
+  // Each FirstWave of WorkGroup clears 64kbyte block
+
+label_001F:
+  ds_write2_b64  v1, v[2:3], v[2:3] offset1:32
+  ds_write2_b64  v1, v[4:5], v[4:5] offset0:64 offset1:96
+  v_add_co_u32     v1, vcc, 0x00000400, v1
+  s_sub_u32     s2, s2, 1
+  s_cbranch_scc0  label_001F
+
+  //
+  // CLEAR SGPRs
+  //
+label_0023:
+  s_mov_b32     m0, 0x00000068  // Loop 108/4=27 times  (loop unrolled for performance)
+label_sgpr_loop:
+  s_movreld_b32     s0, 0
+  s_movreld_b32     s1, 0
+  s_movreld_b32     s2, 0
+  s_movreld_b32     s3, 0
+  s_sub_u32         m0, m0, 4
+  s_cbranch_scc0  label_sgpr_loop
+
+  //clear vcc
+  s_mov_b32 flat_scratch_lo, 0   //clear  flat scratch lo SGPR
+  s_mov_b32 flat_scratch_hi, 0   //clear  flat scratch hi SGPR
+  s_mov_b64 vcc, 0          //clear vcc
+  s_mov_b64 ttmp0, 0        //Clear ttmp0 and ttmp1
+  s_mov_b64 ttmp2, 0        //Clear ttmp2 and ttmp3
+  s_mov_b64 ttmp4, 0        //Clear ttmp4 and ttmp5
+  s_mov_b64 ttmp6, 0        //Clear ttmp6 and ttmp7
+  s_mov_b64 ttmp8, 0        //Clear ttmp8 and ttmp9
+  s_mov_b64 ttmp10, 0       //Clear ttmp10 and ttmp11
+  s_mov_b64 ttmp12, 0       //Clear ttmp12 and ttmp13
+  s_mov_b64 ttmp14, 0       //Clear ttmp14 and ttmp15
+
+ s_endpgm
+
+end
+
+
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
new file mode 100644
index 000000000000..66c47c466532
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
@@ -0,0 +1,7537 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/delay.h>
+#include <linux/kernel.h>
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+#include "amdgpu.h"
+#include "amdgpu_gfx.h"
+#include "amdgpu_psp.h"
+#include "amdgpu_smu.h"
+#include "imu_v11_0.h"
+#include "soc21.h"
+#include "nvd.h"
+
+#include "gc/gc_11_0_0_offset.h"
+#include "gc/gc_11_0_0_sh_mask.h"
+#include "smuio/smuio_13_0_6_offset.h"
+#include "smuio/smuio_13_0_6_sh_mask.h"
+#include "navi10_enum.h"
+#include "ivsrcid/gfx/irqsrcs_gfx_11_0_0.h"
+
+#include "soc15.h"
+#include "clearstate_gfx11.h"
+#include "v11_structs.h"
+#include "gfx_v11_0.h"
+#include "gfx_v11_0_cleaner_shader.h"
+#include "gfx_v11_0_3.h"
+#include "nbio_v4_3.h"
+#include "mes_v11_0.h"
+#include "mes_userqueue.h"
+#include "amdgpu_userq_fence.h"
+
+#define GFX11_NUM_GFX_RINGS		1
+#define GFX11_MEC_HPD_SIZE	2048
+
+#define RLCG_UCODE_LOADING_START_ADDRESS	0x00002000L
+#define RLC_PG_DELAY_3_DEFAULT_GC_11_0_1	0x1388
+
+#define regCGTT_WD_CLK_CTRL		0x5086
+#define regCGTT_WD_CLK_CTRL_BASE_IDX	1
+#define regRLC_RLCS_BOOTLOAD_STATUS_gc_11_0_1	0x4e7e
+#define regRLC_RLCS_BOOTLOAD_STATUS_gc_11_0_1_BASE_IDX	1
+#define regPC_CONFIG_CNTL_1		0x194d
+#define regPC_CONFIG_CNTL_1_BASE_IDX	1
+
+#define regCP_GFX_MQD_CONTROL_DEFAULT                                             0x00000100
+#define regCP_GFX_HQD_VMID_DEFAULT                                                0x00000000
+#define regCP_GFX_HQD_QUEUE_PRIORITY_DEFAULT                                      0x00000000
+#define regCP_GFX_HQD_QUANTUM_DEFAULT                                             0x00000a01
+#define regCP_GFX_HQD_CNTL_DEFAULT                                                0x00a00000
+#define regCP_RB_DOORBELL_CONTROL_DEFAULT                                         0x00000000
+#define regCP_GFX_HQD_RPTR_DEFAULT                                                0x00000000
+
+#define regCP_HQD_EOP_CONTROL_DEFAULT                                             0x00000006
+#define regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT                                     0x00000000
+#define regCP_MQD_CONTROL_DEFAULT                                                 0x00000100
+#define regCP_HQD_PQ_CONTROL_DEFAULT                                              0x00308509
+#define regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT                                     0x00000000
+#define regCP_HQD_PQ_RPTR_DEFAULT                                                 0x00000000
+#define regCP_HQD_PERSISTENT_STATE_DEFAULT                                        0x0be05501
+#define regCP_HQD_IB_CONTROL_DEFAULT                                              0x00300000
+
+MODULE_FIRMWARE("amdgpu/gc_11_0_0_pfp.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_0_me.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_0_mec.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_0_rlc.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_0_rlc_kicker.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_0_rlc_1.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_0_toc.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_1_pfp.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_1_me.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_1_mec.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_1_rlc.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_2_pfp.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_2_me.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_2_mec.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_2_rlc.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_3_pfp.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_3_me.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_3_mec.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_3_rlc.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_4_pfp.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_4_me.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_4_mec.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_4_rlc.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_0_pfp.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_0_me.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_0_mec.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_0_rlc.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_1_pfp.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_1_me.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_1_mec.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_1_rlc.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_2_pfp.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_2_me.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_2_mec.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_2_rlc.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_3_pfp.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_3_me.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_3_mec.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_3_rlc.bin");
+
+static const struct amdgpu_hwip_reg_entry gc_reg_list_11_0[] = {
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS2),
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS3),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_STALLED_STAT1),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_STALLED_STAT2),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_STALLED_STAT3),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPC_STALLED_STAT1),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPF_STALLED_STAT1),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_BUSY_STAT),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPC_BUSY_STAT),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPF_BUSY_STAT),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPC_BUSY_STAT2),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPF_BUSY_STAT2),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPF_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_ERROR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HPD_STATUS0),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_RB_BASE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_RB_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_RB0_BASE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_RB0_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_RB0_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_RB1_BASE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_RB1_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_RB1_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_CMD_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB2_CMD_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB2_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB2_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB2_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, regCPF_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regCPC_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regCPG_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regGDS_PROTECTION_FAULT),
+	SOC15_REG_ENTRY_STR(GC, 0, regGDS_VM_PROTECTION_FAULT),
+	SOC15_REG_ENTRY_STR(GC, 0, regIA_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regIA_UTCL1_STATUS_2),
+	SOC15_REG_ENTRY_STR(GC, 0, regPA_CL_CNTL_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regRLC_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regRMI_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regSQC_CACHES),
+	SOC15_REG_ENTRY_STR(GC, 0, regSQG_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regWD_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regGCVM_L2_PROTECTION_FAULT_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_DEBUG),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC1_INSTR_PNTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_DEBUG_INTERRUPT_INSTR_PNTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_INSTR_PNTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_INSTR_PNTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_INSTR_PNTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPC_STATUS),
+	/* cp header registers */
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	/* SE status registers */
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS_SE0),
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS_SE1),
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS_SE2),
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS_SE3),
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS_SE4),
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS_SE5)
+};
+
+static const struct amdgpu_hwip_reg_entry gc_cp_reg_list_11[] = {
+	/* compute registers */
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_VMID),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PERSISTENT_STATE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PIPE_PRIORITY),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_QUEUE_PRIORITY),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_QUANTUM),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_BASE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_WPTR_POLL_ADDR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_WPTR_POLL_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_IB_BASE_ADDR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_IB_BASE_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_IB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_IB_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_DEQUEUE_REQUEST),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_BASE_ADDR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_BASE_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_EVENTS),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CTX_SAVE_BASE_ADDR_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CTX_SAVE_BASE_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CTX_SAVE_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CNTL_STACK_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CNTL_STACK_SIZE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_WG_STATE_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CTX_SAVE_SIZE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_GDS_RESOURCE_STATE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_ERROR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_WPTR_MEM),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_WPTR_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_SUSPEND_CNTL_STACK_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_SUSPEND_CNTL_STACK_DW_CNT),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_SUSPEND_WG_STATE_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_DEQUEUE_STATUS),
+	/* cp header registers */
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+};
+
+static const struct amdgpu_hwip_reg_entry gc_gfx_queue_reg_list_11[] = {
+	/* gfx queue registers */
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_ACTIVE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_VMID),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_QUEUE_PRIORITY),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_QUANTUM),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_BASE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_CSMD_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_DEQUEUE_REQUEST),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_MAPPED),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_QUE_MGR_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_HQ_CONTROL0),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_HQ_STATUS0),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_MQD_BASE_ADDR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_MQD_BASE_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_RB_WPTR_POLL_ADDR_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_RB_WPTR_POLL_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_CMD_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_BUFSZ),
+	/* cp header registers */
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+};
+
+static const struct soc15_reg_golden golden_settings_gc_11_0[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CNTL, 0x20000000, 0x20000000)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_11_0_1[] =
+{
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regCGTT_GS_NGG_CLK_CTRL, 0x9fff8fff, 0x00000010),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regCGTT_WD_CLK_CTRL, 0xffff8fff, 0x00000010),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regCPF_GCR_CNTL, 0x0007ffff, 0x0000c200),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regGL2C_CTRL3, 0xffff001b, 0x00f01988),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regPA_CL_ENHANCE, 0xf0ffffff, 0x00880007),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regPA_SC_ENHANCE_3, 0xfffffffd, 0x00000008),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regPA_SC_VRS_SURFACE_CNTL_1, 0xfff891ff, 0x55480100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regTA_CNTL_AUX, 0xf7f7ffff, 0x01030000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CNTL2, 0xfcffffff, 0x0000000a)
+};
+
+#define DEFAULT_SH_MEM_CONFIG \
+	((SH_MEM_ADDRESS_MODE_64 << SH_MEM_CONFIG__ADDRESS_MODE__SHIFT) | \
+	 (SH_MEM_ALIGNMENT_MODE_UNALIGNED << SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) | \
+	 (3 << SH_MEM_CONFIG__INITIAL_INST_PREFETCH__SHIFT))
+
+static void gfx_v11_0_disable_gpa_mode(struct amdgpu_device *adev);
+static void gfx_v11_0_set_ring_funcs(struct amdgpu_device *adev);
+static void gfx_v11_0_set_irq_funcs(struct amdgpu_device *adev);
+static void gfx_v11_0_set_gds_init(struct amdgpu_device *adev);
+static void gfx_v11_0_set_rlc_funcs(struct amdgpu_device *adev);
+static void gfx_v11_0_set_mqd_funcs(struct amdgpu_device *adev);
+static void gfx_v11_0_set_imu_funcs(struct amdgpu_device *adev);
+static int gfx_v11_0_get_cu_info(struct amdgpu_device *adev,
+                                 struct amdgpu_cu_info *cu_info);
+static uint64_t gfx_v11_0_get_gpu_clock_counter(struct amdgpu_device *adev);
+static void gfx_v11_0_select_se_sh(struct amdgpu_device *adev, u32 se_num,
+				   u32 sh_num, u32 instance, int xcc_id);
+static u32 gfx_v11_0_get_wgp_active_bitmap_per_sh(struct amdgpu_device *adev);
+
+static void gfx_v11_0_ring_emit_de_meta(struct amdgpu_ring *ring, bool resume);
+static void gfx_v11_0_ring_emit_frame_cntl(struct amdgpu_ring *ring, bool start, bool secure);
+static void gfx_v11_0_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg,
+				     uint32_t val);
+static int gfx_v11_0_wait_for_rlc_autoload_complete(struct amdgpu_device *adev);
+static void gfx_v11_0_ring_invalidate_tlbs(struct amdgpu_ring *ring,
+					   uint16_t pasid, uint32_t flush_type,
+					   bool all_hub, uint8_t dst_sel);
+static void gfx_v11_0_set_safe_mode(struct amdgpu_device *adev, int xcc_id);
+static void gfx_v11_0_unset_safe_mode(struct amdgpu_device *adev, int xcc_id);
+static void gfx_v11_0_update_perf_clk(struct amdgpu_device *adev,
+				      bool enable);
+
+static void gfx11_kiq_set_resources(struct amdgpu_ring *kiq_ring, uint64_t queue_mask)
+{
+	struct amdgpu_device *adev = kiq_ring->adev;
+	u64 shader_mc_addr;
+
+	/* Cleaner shader MC address */
+	shader_mc_addr = adev->gfx.cleaner_shader_gpu_addr >> 8;
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_SET_RESOURCES, 6));
+	amdgpu_ring_write(kiq_ring, PACKET3_SET_RESOURCES_VMID_MASK(0) |
+			  PACKET3_SET_RESOURCES_UNMAP_LATENTY(0xa) | /* unmap_latency: 0xa (~ 1s) */
+			  PACKET3_SET_RESOURCES_QUEUE_TYPE(0));	/* vmid_mask:0 queue_type:0 (KIQ) */
+	amdgpu_ring_write(kiq_ring, lower_32_bits(queue_mask));	/* queue mask lo */
+	amdgpu_ring_write(kiq_ring, upper_32_bits(queue_mask));	/* queue mask hi */
+	amdgpu_ring_write(kiq_ring, lower_32_bits(shader_mc_addr)); /* cleaner shader addr lo */
+	amdgpu_ring_write(kiq_ring, upper_32_bits(shader_mc_addr)); /* cleaner shader addr hi */
+	amdgpu_ring_write(kiq_ring, 0);	/* oac mask */
+	amdgpu_ring_write(kiq_ring, 0);	/* gds heap base:0, gds heap size:0 */
+}
+
+static void gfx11_kiq_map_queues(struct amdgpu_ring *kiq_ring,
+				 struct amdgpu_ring *ring)
+{
+	uint64_t mqd_addr = amdgpu_bo_gpu_offset(ring->mqd_obj);
+	uint64_t wptr_addr = ring->wptr_gpu_addr;
+	uint32_t me = 0, eng_sel = 0;
+
+	switch (ring->funcs->type) {
+	case AMDGPU_RING_TYPE_COMPUTE:
+		me = 1;
+		eng_sel = 0;
+		break;
+	case AMDGPU_RING_TYPE_GFX:
+		me = 0;
+		eng_sel = 4;
+		break;
+	case AMDGPU_RING_TYPE_MES:
+		me = 2;
+		eng_sel = 5;
+		break;
+	default:
+		WARN_ON(1);
+	}
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_MAP_QUEUES, 5));
+	/* Q_sel:0, vmid:0, vidmem: 1, engine:0, num_Q:1*/
+	amdgpu_ring_write(kiq_ring, /* Q_sel: 0, vmid: 0, engine: 0, num_Q: 1 */
+			  PACKET3_MAP_QUEUES_QUEUE_SEL(0) | /* Queue_Sel */
+			  PACKET3_MAP_QUEUES_VMID(0) | /* VMID */
+			  PACKET3_MAP_QUEUES_QUEUE(ring->queue) |
+			  PACKET3_MAP_QUEUES_PIPE(ring->pipe) |
+			  PACKET3_MAP_QUEUES_ME((me)) |
+			  PACKET3_MAP_QUEUES_QUEUE_TYPE(0) | /*queue_type: normal compute queue */
+			  PACKET3_MAP_QUEUES_ALLOC_FORMAT(0) | /* alloc format: all_on_one_pipe */
+			  PACKET3_MAP_QUEUES_ENGINE_SEL(eng_sel) |
+			  PACKET3_MAP_QUEUES_NUM_QUEUES(1)); /* num_queues: must be 1 */
+	amdgpu_ring_write(kiq_ring, PACKET3_MAP_QUEUES_DOORBELL_OFFSET(ring->doorbell_index));
+	amdgpu_ring_write(kiq_ring, lower_32_bits(mqd_addr));
+	amdgpu_ring_write(kiq_ring, upper_32_bits(mqd_addr));
+	amdgpu_ring_write(kiq_ring, lower_32_bits(wptr_addr));
+	amdgpu_ring_write(kiq_ring, upper_32_bits(wptr_addr));
+}
+
+static void gfx11_kiq_unmap_queues(struct amdgpu_ring *kiq_ring,
+				   struct amdgpu_ring *ring,
+				   enum amdgpu_unmap_queues_action action,
+				   u64 gpu_addr, u64 seq)
+{
+	struct amdgpu_device *adev = kiq_ring->adev;
+	uint32_t eng_sel = ring->funcs->type == AMDGPU_RING_TYPE_GFX ? 4 : 0;
+
+	if (adev->enable_mes && !adev->gfx.kiq[0].ring.sched.ready) {
+		amdgpu_mes_unmap_legacy_queue(adev, ring, action, gpu_addr, seq);
+		return;
+	}
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_UNMAP_QUEUES, 4));
+	amdgpu_ring_write(kiq_ring, /* Q_sel: 0, vmid: 0, engine: 0, num_Q: 1 */
+			  PACKET3_UNMAP_QUEUES_ACTION(action) |
+			  PACKET3_UNMAP_QUEUES_QUEUE_SEL(0) |
+			  PACKET3_UNMAP_QUEUES_ENGINE_SEL(eng_sel) |
+			  PACKET3_UNMAP_QUEUES_NUM_QUEUES(1));
+	amdgpu_ring_write(kiq_ring,
+		  PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET0(ring->doorbell_index));
+
+	if (action == PREEMPT_QUEUES_NO_UNMAP) {
+		amdgpu_ring_write(kiq_ring, lower_32_bits(gpu_addr));
+		amdgpu_ring_write(kiq_ring, upper_32_bits(gpu_addr));
+		amdgpu_ring_write(kiq_ring, seq);
+	} else {
+		amdgpu_ring_write(kiq_ring, 0);
+		amdgpu_ring_write(kiq_ring, 0);
+		amdgpu_ring_write(kiq_ring, 0);
+	}
+}
+
+static void gfx11_kiq_query_status(struct amdgpu_ring *kiq_ring,
+				   struct amdgpu_ring *ring,
+				   u64 addr,
+				   u64 seq)
+{
+	uint32_t eng_sel = ring->funcs->type == AMDGPU_RING_TYPE_GFX ? 4 : 0;
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_QUERY_STATUS, 5));
+	amdgpu_ring_write(kiq_ring,
+			  PACKET3_QUERY_STATUS_CONTEXT_ID(0) |
+			  PACKET3_QUERY_STATUS_INTERRUPT_SEL(0) |
+			  PACKET3_QUERY_STATUS_COMMAND(2));
+	amdgpu_ring_write(kiq_ring, /* Q_sel: 0, vmid: 0, engine: 0, num_Q: 1 */
+			  PACKET3_QUERY_STATUS_DOORBELL_OFFSET(ring->doorbell_index) |
+			  PACKET3_QUERY_STATUS_ENG_SEL(eng_sel));
+	amdgpu_ring_write(kiq_ring, lower_32_bits(addr));
+	amdgpu_ring_write(kiq_ring, upper_32_bits(addr));
+	amdgpu_ring_write(kiq_ring, lower_32_bits(seq));
+	amdgpu_ring_write(kiq_ring, upper_32_bits(seq));
+}
+
+static void gfx11_kiq_invalidate_tlbs(struct amdgpu_ring *kiq_ring,
+				uint16_t pasid, uint32_t flush_type,
+				bool all_hub)
+{
+	gfx_v11_0_ring_invalidate_tlbs(kiq_ring, pasid, flush_type, all_hub, 1);
+}
+
+static const struct kiq_pm4_funcs gfx_v11_0_kiq_pm4_funcs = {
+	.kiq_set_resources = gfx11_kiq_set_resources,
+	.kiq_map_queues = gfx11_kiq_map_queues,
+	.kiq_unmap_queues = gfx11_kiq_unmap_queues,
+	.kiq_query_status = gfx11_kiq_query_status,
+	.kiq_invalidate_tlbs = gfx11_kiq_invalidate_tlbs,
+	.set_resources_size = 8,
+	.map_queues_size = 7,
+	.unmap_queues_size = 6,
+	.query_status_size = 7,
+	.invalidate_tlbs_size = 2,
+};
+
+static void gfx_v11_0_set_kiq_pm4_funcs(struct amdgpu_device *adev)
+{
+	adev->gfx.kiq[0].pmf = &gfx_v11_0_kiq_pm4_funcs;
+}
+
+static void gfx_v11_0_init_golden_registers(struct amdgpu_device *adev)
+{
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(11, 0, 1):
+	case IP_VERSION(11, 0, 4):
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_11_0_1,
+						(const u32)ARRAY_SIZE(golden_settings_gc_11_0_1));
+		break;
+	default:
+		break;
+	}
+	soc15_program_register_sequence(adev,
+					golden_settings_gc_11_0,
+					(const u32)ARRAY_SIZE(golden_settings_gc_11_0));
+
+}
+
+static void gfx_v11_0_write_data_to_reg(struct amdgpu_ring *ring, int eng_sel,
+				       bool wc, uint32_t reg, uint32_t val)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, WRITE_DATA_ENGINE_SEL(eng_sel) |
+			  WRITE_DATA_DST_SEL(0) | (wc ? WR_CONFIRM : 0));
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, val);
+}
+
+static void gfx_v11_0_wait_reg_mem(struct amdgpu_ring *ring, int eng_sel,
+				  int mem_space, int opt, uint32_t addr0,
+				  uint32_t addr1, uint32_t ref, uint32_t mask,
+				  uint32_t inv)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+	amdgpu_ring_write(ring,
+			  /* memory (1) or register (0) */
+			  (WAIT_REG_MEM_MEM_SPACE(mem_space) |
+			   WAIT_REG_MEM_OPERATION(opt) | /* wait */
+			   WAIT_REG_MEM_FUNCTION(3) |  /* equal */
+			   WAIT_REG_MEM_ENGINE(eng_sel)));
+
+	if (mem_space)
+		BUG_ON(addr0 & 0x3); /* Dword align */
+	amdgpu_ring_write(ring, addr0);
+	amdgpu_ring_write(ring, addr1);
+	amdgpu_ring_write(ring, ref);
+	amdgpu_ring_write(ring, mask);
+	amdgpu_ring_write(ring, inv); /* poll interval */
+}
+
+static void gfx_v11_ring_insert_nop(struct amdgpu_ring *ring, uint32_t num_nop)
+{
+	/* Header itself is a NOP packet */
+	if (num_nop == 1) {
+		amdgpu_ring_write(ring, ring->funcs->nop);
+		return;
+	}
+
+	/* Max HW optimization till 0x3ffe, followed by remaining one NOP at a time*/
+	amdgpu_ring_write(ring, PACKET3(PACKET3_NOP, min(num_nop - 2, 0x3ffe)));
+
+	/* Header is at index 0, followed by num_nops - 1 NOP packet's */
+	amdgpu_ring_insert_nop(ring, num_nop - 1);
+}
+
+static int gfx_v11_0_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t scratch = SOC15_REG_OFFSET(GC, 0, regSCRATCH_REG0);
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	WREG32(scratch, 0xCAFEDEAD);
+	r = amdgpu_ring_alloc(ring, 5);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n",
+			  ring->idx, r);
+		return r;
+	}
+
+	if (ring->funcs->type == AMDGPU_RING_TYPE_KIQ) {
+		gfx_v11_0_ring_emit_wreg(ring, scratch, 0xDEADBEEF);
+	} else {
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SET_UCONFIG_REG, 1));
+		amdgpu_ring_write(ring, scratch -
+				  PACKET3_SET_UCONFIG_REG_START);
+		amdgpu_ring_write(ring, 0xDEADBEEF);
+	}
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(scratch);
+		if (tmp == 0xDEADBEEF)
+			break;
+		if (amdgpu_emu_mode == 1)
+			msleep(1);
+		else
+			udelay(1);
+	}
+
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+	return r;
+}
+
+static int gfx_v11_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct dma_fence *f = NULL;
+	unsigned index;
+	uint64_t gpu_addr;
+	uint32_t *cpu_ptr;
+	long r;
+
+	/* MES KIQ fw hasn't indirect buffer support for now */
+	if (adev->enable_mes_kiq &&
+	    ring->funcs->type == AMDGPU_RING_TYPE_KIQ)
+		return 0;
+
+	memset(&ib, 0, sizeof(ib));
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r)
+		return r;
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	adev->wb.wb[index] = cpu_to_le32(0xCAFEDEAD);
+	cpu_ptr = &adev->wb.wb[index];
+
+	r = amdgpu_ib_get(adev, NULL, 20, AMDGPU_IB_POOL_DIRECT, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
+		goto err1;
+	}
+
+	ib.ptr[0] = PACKET3(PACKET3_WRITE_DATA, 3);
+	ib.ptr[1] = WRITE_DATA_DST_SEL(5) | WR_CONFIRM;
+	ib.ptr[2] = lower_32_bits(gpu_addr);
+	ib.ptr[3] = upper_32_bits(gpu_addr);
+	ib.ptr[4] = 0xDEADBEEF;
+	ib.length_dw = 5;
+
+	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
+	if (r)
+		goto err2;
+
+	r = dma_fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		r = -ETIMEDOUT;
+		goto err2;
+	} else if (r < 0) {
+		goto err2;
+	}
+
+	if (le32_to_cpu(*cpu_ptr) == 0xDEADBEEF)
+		r = 0;
+	else
+		r = -EINVAL;
+err2:
+	amdgpu_ib_free(&ib, NULL);
+	dma_fence_put(f);
+err1:
+	amdgpu_device_wb_free(adev, index);
+	return r;
+}
+
+static void gfx_v11_0_free_microcode(struct amdgpu_device *adev)
+{
+	amdgpu_ucode_release(&adev->gfx.pfp_fw);
+	amdgpu_ucode_release(&adev->gfx.me_fw);
+	amdgpu_ucode_release(&adev->gfx.rlc_fw);
+	amdgpu_ucode_release(&adev->gfx.mec_fw);
+
+	kfree(adev->gfx.rlc.register_list_format);
+}
+
+static int gfx_v11_0_init_toc_microcode(struct amdgpu_device *adev, const char *ucode_prefix)
+{
+	const struct psp_firmware_header_v1_0 *toc_hdr;
+	int err = 0;
+
+	err = amdgpu_ucode_request(adev, &adev->psp.toc_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_toc.bin", ucode_prefix);
+	if (err)
+		goto out;
+
+	toc_hdr = (const struct psp_firmware_header_v1_0 *)adev->psp.toc_fw->data;
+	adev->psp.toc.fw_version = le32_to_cpu(toc_hdr->header.ucode_version);
+	adev->psp.toc.feature_version = le32_to_cpu(toc_hdr->sos.fw_version);
+	adev->psp.toc.size_bytes = le32_to_cpu(toc_hdr->header.ucode_size_bytes);
+	adev->psp.toc.start_addr = (uint8_t *)toc_hdr +
+				le32_to_cpu(toc_hdr->header.ucode_array_offset_bytes);
+	return 0;
+out:
+	amdgpu_ucode_release(&adev->psp.toc_fw);
+	return err;
+}
+
+static void gfx_v11_0_check_fw_cp_gfx_shadow(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 2):
+	case IP_VERSION(11, 0, 3):
+		if ((adev->gfx.me_fw_version >= 1505) &&
+		    (adev->gfx.pfp_fw_version >= 1600) &&
+		    (adev->gfx.mec_fw_version >= 512)) {
+			if (amdgpu_sriov_vf(adev))
+				adev->gfx.cp_gfx_shadow = true;
+			else
+				adev->gfx.cp_gfx_shadow = false;
+		}
+		break;
+	default:
+		adev->gfx.cp_gfx_shadow = false;
+		break;
+	}
+}
+
+static int gfx_v11_0_init_microcode(struct amdgpu_device *adev)
+{
+	char ucode_prefix[25];
+	int err;
+	const struct rlc_firmware_header_v2_0 *rlc_hdr;
+	uint16_t version_major;
+	uint16_t version_minor;
+
+	DRM_DEBUG("\n");
+
+	amdgpu_ucode_ip_version_decode(adev, GC_HWIP, ucode_prefix, sizeof(ucode_prefix));
+	err = amdgpu_ucode_request(adev, &adev->gfx.pfp_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_pfp.bin", ucode_prefix);
+	if (err)
+		goto out;
+	/* check pfp fw hdr version to decide if enable rs64 for gfx11.*/
+	adev->gfx.rs64_enable = amdgpu_ucode_hdr_version(
+				(union amdgpu_firmware_header *)
+				adev->gfx.pfp_fw->data, 2, 0);
+	if (adev->gfx.rs64_enable) {
+		dev_info(adev->dev, "CP RS64 enable\n");
+		amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_RS64_PFP);
+		amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_RS64_PFP_P0_STACK);
+		amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_RS64_PFP_P1_STACK);
+	} else {
+		amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_PFP);
+	}
+
+	err = amdgpu_ucode_request(adev, &adev->gfx.me_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_me.bin", ucode_prefix);
+	if (err)
+		goto out;
+	if (adev->gfx.rs64_enable) {
+		amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_RS64_ME);
+		amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_RS64_ME_P0_STACK);
+		amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_RS64_ME_P1_STACK);
+	} else {
+		amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_ME);
+	}
+
+	if (!amdgpu_sriov_vf(adev)) {
+		if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(11, 0, 0) &&
+		    adev->pdev->revision == 0xCE)
+			err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw,
+						   AMDGPU_UCODE_REQUIRED,
+						   "amdgpu/gc_11_0_0_rlc_1.bin");
+		else if (amdgpu_is_kicker_fw(adev))
+			err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw,
+						   AMDGPU_UCODE_REQUIRED,
+						   "amdgpu/%s_rlc_kicker.bin", ucode_prefix);
+		else
+			err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw,
+						   AMDGPU_UCODE_REQUIRED,
+						   "amdgpu/%s_rlc.bin", ucode_prefix);
+		if (err)
+			goto out;
+		rlc_hdr = (const struct rlc_firmware_header_v2_0 *)adev->gfx.rlc_fw->data;
+		version_major = le16_to_cpu(rlc_hdr->header.header_version_major);
+		version_minor = le16_to_cpu(rlc_hdr->header.header_version_minor);
+		err = amdgpu_gfx_rlc_init_microcode(adev, version_major, version_minor);
+		if (err)
+			goto out;
+	}
+
+	err = amdgpu_ucode_request(adev, &adev->gfx.mec_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_mec.bin", ucode_prefix);
+	if (err)
+		goto out;
+	if (adev->gfx.rs64_enable) {
+		amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_RS64_MEC);
+		amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_RS64_MEC_P0_STACK);
+		amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_RS64_MEC_P1_STACK);
+		amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_RS64_MEC_P2_STACK);
+		amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_RS64_MEC_P3_STACK);
+	} else {
+		amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_MEC1);
+		amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_MEC1_JT);
+	}
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO)
+		err = gfx_v11_0_init_toc_microcode(adev, ucode_prefix);
+
+	/* only one MEC for gfx 11.0.0. */
+	adev->gfx.mec2_fw = NULL;
+
+	gfx_v11_0_check_fw_cp_gfx_shadow(adev);
+
+	if (adev->gfx.imu.funcs && adev->gfx.imu.funcs->init_microcode) {
+		err = adev->gfx.imu.funcs->init_microcode(adev);
+		if (err)
+			DRM_ERROR("Failed to init imu firmware!\n");
+		return err;
+	}
+
+out:
+	if (err) {
+		amdgpu_ucode_release(&adev->gfx.pfp_fw);
+		amdgpu_ucode_release(&adev->gfx.me_fw);
+		amdgpu_ucode_release(&adev->gfx.rlc_fw);
+		amdgpu_ucode_release(&adev->gfx.mec_fw);
+	}
+
+	return err;
+}
+
+static u32 gfx_v11_0_get_csb_size(struct amdgpu_device *adev)
+{
+	u32 count = 0;
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+
+	/* begin clear state */
+	count += 2;
+	/* context control state */
+	count += 3;
+
+	for (sect = gfx11_cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT)
+				count += 2 + ext->reg_count;
+			else
+				return 0;
+		}
+	}
+
+	/* set PA_SC_TILE_STEERING_OVERRIDE */
+	count += 3;
+	/* end clear state */
+	count += 2;
+	/* clear state */
+	count += 2;
+
+	return count;
+}
+
+static void gfx_v11_0_get_csb_buffer(struct amdgpu_device *adev, u32 *buffer)
+{
+	u32 count = 0;
+	int ctx_reg_offset;
+
+	if (adev->gfx.rlc.cs_data == NULL)
+		return;
+	if (buffer == NULL)
+		return;
+
+	count = amdgpu_gfx_csb_preamble_start(buffer);
+	count = amdgpu_gfx_csb_data_parser(adev, buffer, count);
+
+	ctx_reg_offset = SOC15_REG_OFFSET(GC, 0, regPA_SC_TILE_STEERING_OVERRIDE) - PACKET3_SET_CONTEXT_REG_START;
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, 1));
+	buffer[count++] = cpu_to_le32(ctx_reg_offset);
+	buffer[count++] = cpu_to_le32(adev->gfx.config.pa_sc_tile_steering_override);
+
+	amdgpu_gfx_csb_preamble_end(buffer, count);
+}
+
+static void gfx_v11_0_rlc_fini(struct amdgpu_device *adev)
+{
+	/* clear state block */
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj,
+			&adev->gfx.rlc.clear_state_gpu_addr,
+			(void **)&adev->gfx.rlc.cs_ptr);
+
+	/* jump table block */
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj,
+			&adev->gfx.rlc.cp_table_gpu_addr,
+			(void **)&adev->gfx.rlc.cp_table_ptr);
+}
+
+static void gfx_v11_0_init_rlcg_reg_access_ctrl(struct amdgpu_device *adev)
+{
+	struct amdgpu_rlcg_reg_access_ctrl *reg_access_ctrl;
+
+	reg_access_ctrl = &adev->gfx.rlc.reg_access_ctrl[0];
+	reg_access_ctrl->scratch_reg0 = SOC15_REG_OFFSET(GC, 0, regSCRATCH_REG0);
+	reg_access_ctrl->scratch_reg1 = SOC15_REG_OFFSET(GC, 0, regSCRATCH_REG1);
+	reg_access_ctrl->scratch_reg2 = SOC15_REG_OFFSET(GC, 0, regSCRATCH_REG2);
+	reg_access_ctrl->scratch_reg3 = SOC15_REG_OFFSET(GC, 0, regSCRATCH_REG3);
+	reg_access_ctrl->grbm_cntl = SOC15_REG_OFFSET(GC, 0, regGRBM_GFX_CNTL);
+	reg_access_ctrl->grbm_idx = SOC15_REG_OFFSET(GC, 0, regGRBM_GFX_INDEX);
+	reg_access_ctrl->spare_int = SOC15_REG_OFFSET(GC, 0, regRLC_SPARE_INT_0);
+	adev->gfx.rlc.rlcg_reg_access_supported = true;
+}
+
+static int gfx_v11_0_rlc_init(struct amdgpu_device *adev)
+{
+	const struct cs_section_def *cs_data;
+	int r;
+
+	adev->gfx.rlc.cs_data = gfx11_cs_data;
+
+	cs_data = adev->gfx.rlc.cs_data;
+
+	if (cs_data) {
+		/* init clear state block */
+		r = amdgpu_gfx_rlc_init_csb(adev);
+		if (r)
+			return r;
+	}
+
+	/* init spm vmid with 0xf */
+	if (adev->gfx.rlc.funcs->update_spm_vmid)
+		adev->gfx.rlc.funcs->update_spm_vmid(adev, NULL, 0xf);
+
+	return 0;
+}
+
+static void gfx_v11_0_mec_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->gfx.mec.hpd_eop_obj, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gfx.mec.mec_fw_obj, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gfx.mec.mec_fw_data_obj, NULL, NULL);
+}
+
+static void gfx_v11_0_me_init(struct amdgpu_device *adev)
+{
+	bitmap_zero(adev->gfx.me.queue_bitmap, AMDGPU_MAX_GFX_QUEUES);
+
+	amdgpu_gfx_graphics_queue_acquire(adev);
+}
+
+static int gfx_v11_0_mec_init(struct amdgpu_device *adev)
+{
+	int r;
+	u32 *hpd;
+	size_t mec_hpd_size;
+
+	bitmap_zero(adev->gfx.mec_bitmap[0].queue_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
+
+	/* take ownership of the relevant compute queues */
+	amdgpu_gfx_compute_queue_acquire(adev);
+	mec_hpd_size = adev->gfx.num_compute_rings * GFX11_MEC_HPD_SIZE;
+
+	if (mec_hpd_size) {
+		r = amdgpu_bo_create_reserved(adev, mec_hpd_size, PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_GTT,
+					      &adev->gfx.mec.hpd_eop_obj,
+					      &adev->gfx.mec.hpd_eop_gpu_addr,
+					      (void **)&hpd);
+		if (r) {
+			dev_warn(adev->dev, "(%d) create HDP EOP bo failed\n", r);
+			gfx_v11_0_mec_fini(adev);
+			return r;
+		}
+
+		memset(hpd, 0, mec_hpd_size);
+
+		amdgpu_bo_kunmap(adev->gfx.mec.hpd_eop_obj);
+		amdgpu_bo_unreserve(adev->gfx.mec.hpd_eop_obj);
+	}
+
+	return 0;
+}
+
+static uint32_t wave_read_ind(struct amdgpu_device *adev, uint32_t wave, uint32_t address)
+{
+	WREG32_SOC15(GC, 0, regSQ_IND_INDEX,
+		(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
+		(address << SQ_IND_INDEX__INDEX__SHIFT));
+	return RREG32_SOC15(GC, 0, regSQ_IND_DATA);
+}
+
+static void wave_read_regs(struct amdgpu_device *adev, uint32_t wave,
+			   uint32_t thread, uint32_t regno,
+			   uint32_t num, uint32_t *out)
+{
+	WREG32_SOC15(GC, 0, regSQ_IND_INDEX,
+		(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
+		(regno << SQ_IND_INDEX__INDEX__SHIFT) |
+		(thread << SQ_IND_INDEX__WORKITEM_ID__SHIFT) |
+		(SQ_IND_INDEX__AUTO_INCR_MASK));
+	while (num--)
+		*(out++) = RREG32_SOC15(GC, 0, regSQ_IND_DATA);
+}
+
+static void gfx_v11_0_read_wave_data(struct amdgpu_device *adev, uint32_t xcc_id, uint32_t simd, uint32_t wave, uint32_t *dst, int *no_fields)
+{
+	/* in gfx11 the SIMD_ID is specified as part of the INSTANCE
+	 * field when performing a select_se_sh so it should be
+	 * zero here */
+	WARN_ON(simd != 0);
+
+	/* type 3 wave data */
+	dst[(*no_fields)++] = 3;
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_STATUS);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_PC_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_PC_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_EXEC_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_EXEC_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_HW_ID1);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_HW_ID2);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_GPR_ALLOC);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_LDS_ALLOC);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_TRAPSTS);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_IB_STS);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_IB_STS2);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_IB_DBG1);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_M0);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_MODE);
+}
+
+static void gfx_v11_0_read_wave_sgprs(struct amdgpu_device *adev, uint32_t xcc_id, uint32_t simd,
+				     uint32_t wave, uint32_t start,
+				     uint32_t size, uint32_t *dst)
+{
+	WARN_ON(simd != 0);
+
+	wave_read_regs(
+		adev, wave, 0, start + SQIND_WAVE_SGPRS_OFFSET, size,
+		dst);
+}
+
+static void gfx_v11_0_read_wave_vgprs(struct amdgpu_device *adev, uint32_t xcc_id, uint32_t simd,
+				      uint32_t wave, uint32_t thread,
+				      uint32_t start, uint32_t size,
+				      uint32_t *dst)
+{
+	wave_read_regs(
+		adev, wave, thread,
+		start + SQIND_WAVE_VGPRS_OFFSET, size, dst);
+}
+
+static void gfx_v11_0_select_me_pipe_q(struct amdgpu_device *adev,
+					u32 me, u32 pipe, u32 q, u32 vm, u32 xcc_id)
+{
+	soc21_grbm_select(adev, me, pipe, q, vm);
+}
+
+/* all sizes are in bytes */
+#define MQD_SHADOW_BASE_SIZE      73728
+#define MQD_SHADOW_BASE_ALIGNMENT 256
+#define MQD_FWWORKAREA_SIZE       484
+#define MQD_FWWORKAREA_ALIGNMENT  256
+
+static void gfx_v11_0_get_gfx_shadow_info_nocheck(struct amdgpu_device *adev,
+					 struct amdgpu_gfx_shadow_info *shadow_info)
+{
+	shadow_info->shadow_size = MQD_SHADOW_BASE_SIZE;
+	shadow_info->shadow_alignment = MQD_SHADOW_BASE_ALIGNMENT;
+	shadow_info->csa_size = MQD_FWWORKAREA_SIZE;
+	shadow_info->csa_alignment = MQD_FWWORKAREA_ALIGNMENT;
+}
+
+static int gfx_v11_0_get_gfx_shadow_info(struct amdgpu_device *adev,
+					 struct amdgpu_gfx_shadow_info *shadow_info,
+					 bool skip_check)
+{
+	if (adev->gfx.cp_gfx_shadow || skip_check) {
+		gfx_v11_0_get_gfx_shadow_info_nocheck(adev, shadow_info);
+		return 0;
+	} else {
+		memset(shadow_info, 0, sizeof(struct amdgpu_gfx_shadow_info));
+		return -ENOTSUPP;
+	}
+}
+
+static const struct amdgpu_gfx_funcs gfx_v11_0_gfx_funcs = {
+	.get_gpu_clock_counter = &gfx_v11_0_get_gpu_clock_counter,
+	.select_se_sh = &gfx_v11_0_select_se_sh,
+	.read_wave_data = &gfx_v11_0_read_wave_data,
+	.read_wave_sgprs = &gfx_v11_0_read_wave_sgprs,
+	.read_wave_vgprs = &gfx_v11_0_read_wave_vgprs,
+	.select_me_pipe_q = &gfx_v11_0_select_me_pipe_q,
+	.update_perfmon_mgcg = &gfx_v11_0_update_perf_clk,
+	.get_gfx_shadow_info = &gfx_v11_0_get_gfx_shadow_info,
+};
+
+static int gfx_v11_0_gpu_early_init(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 2):
+		adev->gfx.config.max_hw_contexts = 8;
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x4C0;
+		break;
+	case IP_VERSION(11, 0, 3):
+		adev->gfx.ras = &gfx_v11_0_3_ras;
+		adev->gfx.config.max_hw_contexts = 8;
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x4C0;
+		break;
+	case IP_VERSION(11, 0, 1):
+	case IP_VERSION(11, 0, 4):
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 1):
+	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
+		adev->gfx.config.max_hw_contexts = 8;
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x80;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x300;
+		break;
+	default:
+		BUG();
+		break;
+	}
+
+	return 0;
+}
+
+static int gfx_v11_0_gfx_ring_init(struct amdgpu_device *adev, int ring_id,
+				   int me, int pipe, int queue)
+{
+	struct amdgpu_ring *ring;
+	unsigned int irq_type;
+	unsigned int hw_prio;
+
+	ring = &adev->gfx.gfx_ring[ring_id];
+
+	ring->me = me;
+	ring->pipe = pipe;
+	ring->queue = queue;
+
+	ring->ring_obj = NULL;
+	ring->use_doorbell = true;
+	if (adev->gfx.disable_kq) {
+		ring->no_scheduler = true;
+		ring->no_user_submission = true;
+	}
+
+	if (!ring_id)
+		ring->doorbell_index = adev->doorbell_index.gfx_ring0 << 1;
+	else
+		ring->doorbell_index = adev->doorbell_index.gfx_ring1 << 1;
+	ring->vm_hub = AMDGPU_GFXHUB(0);
+	sprintf(ring->name, "gfx_%d.%d.%d", ring->me, ring->pipe, ring->queue);
+
+	irq_type = AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP + ring->pipe;
+	hw_prio = amdgpu_gfx_is_high_priority_graphics_queue(adev, ring) ?
+		AMDGPU_GFX_PIPE_PRIO_HIGH : AMDGPU_GFX_PIPE_PRIO_NORMAL;
+	return amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq, irq_type,
+				hw_prio, NULL);
+}
+
+static int gfx_v11_0_compute_ring_init(struct amdgpu_device *adev, int ring_id,
+				       int mec, int pipe, int queue)
+{
+	int r;
+	unsigned irq_type;
+	struct amdgpu_ring *ring;
+	unsigned int hw_prio;
+
+	ring = &adev->gfx.compute_ring[ring_id];
+
+	/* mec0 is me1 */
+	ring->me = mec + 1;
+	ring->pipe = pipe;
+	ring->queue = queue;
+
+	ring->ring_obj = NULL;
+	ring->use_doorbell = true;
+	ring->doorbell_index = (adev->doorbell_index.mec_ring0 + ring_id) << 1;
+	ring->eop_gpu_addr = adev->gfx.mec.hpd_eop_gpu_addr
+				+ (ring_id * GFX11_MEC_HPD_SIZE);
+	ring->vm_hub = AMDGPU_GFXHUB(0);
+	sprintf(ring->name, "comp_%d.%d.%d", ring->me, ring->pipe, ring->queue);
+
+	irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP
+		+ ((ring->me - 1) * adev->gfx.mec.num_pipe_per_mec)
+		+ ring->pipe;
+	hw_prio = amdgpu_gfx_is_high_priority_compute_queue(adev, ring) ?
+			AMDGPU_GFX_PIPE_PRIO_HIGH : AMDGPU_GFX_PIPE_PRIO_NORMAL;
+	/* type-2 packets are deprecated on MEC, use type-3 instead */
+	r = amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq, irq_type,
+			     hw_prio, NULL);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static struct {
+	SOC21_FIRMWARE_ID	id;
+	unsigned int		offset;
+	unsigned int		size;
+} rlc_autoload_info[SOC21_FIRMWARE_ID_MAX];
+
+static void gfx_v11_0_parse_rlc_toc(struct amdgpu_device *adev, void *rlc_toc)
+{
+	RLC_TABLE_OF_CONTENT *ucode = rlc_toc;
+
+	while (ucode && (ucode->id > SOC21_FIRMWARE_ID_INVALID) &&
+			(ucode->id < SOC21_FIRMWARE_ID_MAX)) {
+		rlc_autoload_info[ucode->id].id = ucode->id;
+		rlc_autoload_info[ucode->id].offset = ucode->offset * 4;
+		rlc_autoload_info[ucode->id].size = ucode->size * 4;
+
+		ucode++;
+	}
+}
+
+static uint32_t gfx_v11_0_calc_toc_total_size(struct amdgpu_device *adev)
+{
+	uint32_t total_size = 0;
+	SOC21_FIRMWARE_ID id;
+
+	gfx_v11_0_parse_rlc_toc(adev, adev->psp.toc.start_addr);
+
+	for (id = SOC21_FIRMWARE_ID_RLC_G_UCODE; id < SOC21_FIRMWARE_ID_MAX; id++)
+		total_size += rlc_autoload_info[id].size;
+
+	/* In case the offset in rlc toc ucode is aligned */
+	if (total_size < rlc_autoload_info[SOC21_FIRMWARE_ID_MAX-1].offset)
+		total_size = rlc_autoload_info[SOC21_FIRMWARE_ID_MAX-1].offset +
+			rlc_autoload_info[SOC21_FIRMWARE_ID_MAX-1].size;
+
+	return total_size;
+}
+
+static int gfx_v11_0_rlc_autoload_buffer_init(struct amdgpu_device *adev)
+{
+	int r;
+	uint32_t total_size;
+
+	total_size = gfx_v11_0_calc_toc_total_size(adev);
+
+	r = amdgpu_bo_create_reserved(adev, total_size, 64 * 1024,
+				      AMDGPU_GEM_DOMAIN_VRAM |
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.rlc.rlc_autoload_bo,
+				      &adev->gfx.rlc.rlc_autoload_gpu_addr,
+				      (void **)&adev->gfx.rlc.rlc_autoload_ptr);
+
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create fw autoload bo\n", r);
+		return r;
+	}
+
+	return 0;
+}
+
+static void gfx_v11_0_rlc_backdoor_autoload_copy_ucode(struct amdgpu_device *adev,
+					      SOC21_FIRMWARE_ID id,
+			    		      const void *fw_data,
+					      uint32_t fw_size,
+					      uint32_t *fw_autoload_mask)
+{
+	uint32_t toc_offset;
+	uint32_t toc_fw_size;
+	char *ptr = adev->gfx.rlc.rlc_autoload_ptr;
+
+	if (id <= SOC21_FIRMWARE_ID_INVALID || id >= SOC21_FIRMWARE_ID_MAX)
+		return;
+
+	toc_offset = rlc_autoload_info[id].offset;
+	toc_fw_size = rlc_autoload_info[id].size;
+
+	if (fw_size == 0)
+		fw_size = toc_fw_size;
+
+	if (fw_size > toc_fw_size)
+		fw_size = toc_fw_size;
+
+	memcpy(ptr + toc_offset, fw_data, fw_size);
+
+	if (fw_size < toc_fw_size)
+		memset(ptr + toc_offset + fw_size, 0, toc_fw_size - fw_size);
+
+	if ((id != SOC21_FIRMWARE_ID_RS64_PFP) && (id != SOC21_FIRMWARE_ID_RS64_ME))
+		*(uint64_t *)fw_autoload_mask |= 1ULL << id;
+}
+
+static void gfx_v11_0_rlc_backdoor_autoload_copy_toc_ucode(struct amdgpu_device *adev,
+							uint32_t *fw_autoload_mask)
+{
+	void *data;
+	uint32_t size;
+	uint64_t *toc_ptr;
+
+	*(uint64_t *)fw_autoload_mask |= 0x1;
+
+	DRM_DEBUG("rlc autoload enabled fw: 0x%llx\n", *(uint64_t *)fw_autoload_mask);
+
+	data = adev->psp.toc.start_addr;
+	size = rlc_autoload_info[SOC21_FIRMWARE_ID_RLC_TOC].size;
+
+	toc_ptr = (uint64_t *)data + size / 8 - 1;
+	*toc_ptr = *(uint64_t *)fw_autoload_mask;
+
+	gfx_v11_0_rlc_backdoor_autoload_copy_ucode(adev, SOC21_FIRMWARE_ID_RLC_TOC,
+					data, size, fw_autoload_mask);
+}
+
+static void gfx_v11_0_rlc_backdoor_autoload_copy_gfx_ucode(struct amdgpu_device *adev,
+							uint32_t *fw_autoload_mask)
+{
+	const __le32 *fw_data;
+	uint32_t fw_size;
+	const struct gfx_firmware_header_v1_0 *cp_hdr;
+	const struct gfx_firmware_header_v2_0 *cpv2_hdr;
+	const struct rlc_firmware_header_v2_0 *rlc_hdr;
+	const struct rlc_firmware_header_v2_2 *rlcv22_hdr;
+	uint16_t version_major, version_minor;
+
+	if (adev->gfx.rs64_enable) {
+		/* pfp ucode */
+		cpv2_hdr = (const struct gfx_firmware_header_v2_0 *)
+			adev->gfx.pfp_fw->data;
+		/* instruction */
+		fw_data = (const __le32 *)(adev->gfx.pfp_fw->data +
+			le32_to_cpu(cpv2_hdr->ucode_offset_bytes));
+		fw_size = le32_to_cpu(cpv2_hdr->ucode_size_bytes);
+		gfx_v11_0_rlc_backdoor_autoload_copy_ucode(adev, SOC21_FIRMWARE_ID_RS64_PFP,
+						fw_data, fw_size, fw_autoload_mask);
+		/* data */
+		fw_data = (const __le32 *)(adev->gfx.pfp_fw->data +
+			le32_to_cpu(cpv2_hdr->data_offset_bytes));
+		fw_size = le32_to_cpu(cpv2_hdr->data_size_bytes);
+		gfx_v11_0_rlc_backdoor_autoload_copy_ucode(adev, SOC21_FIRMWARE_ID_RS64_PFP_P0_STACK,
+						fw_data, fw_size, fw_autoload_mask);
+		gfx_v11_0_rlc_backdoor_autoload_copy_ucode(adev, SOC21_FIRMWARE_ID_RS64_PFP_P1_STACK,
+						fw_data, fw_size, fw_autoload_mask);
+		/* me ucode */
+		cpv2_hdr = (const struct gfx_firmware_header_v2_0 *)
+			adev->gfx.me_fw->data;
+		/* instruction */
+		fw_data = (const __le32 *)(adev->gfx.me_fw->data +
+			le32_to_cpu(cpv2_hdr->ucode_offset_bytes));
+		fw_size = le32_to_cpu(cpv2_hdr->ucode_size_bytes);
+		gfx_v11_0_rlc_backdoor_autoload_copy_ucode(adev, SOC21_FIRMWARE_ID_RS64_ME,
+						fw_data, fw_size, fw_autoload_mask);
+		/* data */
+		fw_data = (const __le32 *)(adev->gfx.me_fw->data +
+			le32_to_cpu(cpv2_hdr->data_offset_bytes));
+		fw_size = le32_to_cpu(cpv2_hdr->data_size_bytes);
+		gfx_v11_0_rlc_backdoor_autoload_copy_ucode(adev, SOC21_FIRMWARE_ID_RS64_ME_P0_STACK,
+						fw_data, fw_size, fw_autoload_mask);
+		gfx_v11_0_rlc_backdoor_autoload_copy_ucode(adev, SOC21_FIRMWARE_ID_RS64_ME_P1_STACK,
+						fw_data, fw_size, fw_autoload_mask);
+		/* mec ucode */
+		cpv2_hdr = (const struct gfx_firmware_header_v2_0 *)
+			adev->gfx.mec_fw->data;
+		/* instruction */
+		fw_data = (const __le32 *) (adev->gfx.mec_fw->data +
+			le32_to_cpu(cpv2_hdr->ucode_offset_bytes));
+		fw_size = le32_to_cpu(cpv2_hdr->ucode_size_bytes);
+		gfx_v11_0_rlc_backdoor_autoload_copy_ucode(adev, SOC21_FIRMWARE_ID_RS64_MEC,
+						fw_data, fw_size, fw_autoload_mask);
+		/* data */
+		fw_data = (const __le32 *) (adev->gfx.mec_fw->data +
+			le32_to_cpu(cpv2_hdr->data_offset_bytes));
+		fw_size = le32_to_cpu(cpv2_hdr->data_size_bytes);
+		gfx_v11_0_rlc_backdoor_autoload_copy_ucode(adev, SOC21_FIRMWARE_ID_RS64_MEC_P0_STACK,
+						fw_data, fw_size, fw_autoload_mask);
+		gfx_v11_0_rlc_backdoor_autoload_copy_ucode(adev, SOC21_FIRMWARE_ID_RS64_MEC_P1_STACK,
+						fw_data, fw_size, fw_autoload_mask);
+		gfx_v11_0_rlc_backdoor_autoload_copy_ucode(adev, SOC21_FIRMWARE_ID_RS64_MEC_P2_STACK,
+						fw_data, fw_size, fw_autoload_mask);
+		gfx_v11_0_rlc_backdoor_autoload_copy_ucode(adev, SOC21_FIRMWARE_ID_RS64_MEC_P3_STACK,
+						fw_data, fw_size, fw_autoload_mask);
+	} else {
+		/* pfp ucode */
+		cp_hdr = (const struct gfx_firmware_header_v1_0 *)
+			adev->gfx.pfp_fw->data;
+		fw_data = (const __le32 *)(adev->gfx.pfp_fw->data +
+				le32_to_cpu(cp_hdr->header.ucode_array_offset_bytes));
+		fw_size = le32_to_cpu(cp_hdr->header.ucode_size_bytes);
+		gfx_v11_0_rlc_backdoor_autoload_copy_ucode(adev, SOC21_FIRMWARE_ID_CP_PFP,
+						fw_data, fw_size, fw_autoload_mask);
+
+		/* me ucode */
+		cp_hdr = (const struct gfx_firmware_header_v1_0 *)
+			adev->gfx.me_fw->data;
+		fw_data = (const __le32 *)(adev->gfx.me_fw->data +
+				le32_to_cpu(cp_hdr->header.ucode_array_offset_bytes));
+		fw_size = le32_to_cpu(cp_hdr->header.ucode_size_bytes);
+		gfx_v11_0_rlc_backdoor_autoload_copy_ucode(adev, SOC21_FIRMWARE_ID_CP_ME,
+						fw_data, fw_size, fw_autoload_mask);
+
+		/* mec ucode */
+		cp_hdr = (const struct gfx_firmware_header_v1_0 *)
+			adev->gfx.mec_fw->data;
+		fw_data = (const __le32 *) (adev->gfx.mec_fw->data +
+				le32_to_cpu(cp_hdr->header.ucode_array_offset_bytes));
+		fw_size = le32_to_cpu(cp_hdr->header.ucode_size_bytes) -
+			cp_hdr->jt_size * 4;
+		gfx_v11_0_rlc_backdoor_autoload_copy_ucode(adev, SOC21_FIRMWARE_ID_CP_MEC,
+						fw_data, fw_size, fw_autoload_mask);
+	}
+
+	/* rlc ucode */
+	rlc_hdr = (const struct rlc_firmware_header_v2_0 *)
+		adev->gfx.rlc_fw->data;
+	fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+			le32_to_cpu(rlc_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(rlc_hdr->header.ucode_size_bytes);
+	gfx_v11_0_rlc_backdoor_autoload_copy_ucode(adev, SOC21_FIRMWARE_ID_RLC_G_UCODE,
+					fw_data, fw_size, fw_autoload_mask);
+
+	version_major = le16_to_cpu(rlc_hdr->header.header_version_major);
+	version_minor = le16_to_cpu(rlc_hdr->header.header_version_minor);
+	if (version_major == 2) {
+		if (version_minor >= 2) {
+			rlcv22_hdr = (const struct rlc_firmware_header_v2_2 *)adev->gfx.rlc_fw->data;
+
+			fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+					le32_to_cpu(rlcv22_hdr->rlc_iram_ucode_offset_bytes));
+			fw_size = le32_to_cpu(rlcv22_hdr->rlc_iram_ucode_size_bytes);
+			gfx_v11_0_rlc_backdoor_autoload_copy_ucode(adev, SOC21_FIRMWARE_ID_RLX6_UCODE,
+					fw_data, fw_size, fw_autoload_mask);
+
+			fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+					le32_to_cpu(rlcv22_hdr->rlc_dram_ucode_offset_bytes));
+			fw_size = le32_to_cpu(rlcv22_hdr->rlc_dram_ucode_size_bytes);
+			gfx_v11_0_rlc_backdoor_autoload_copy_ucode(adev, SOC21_FIRMWARE_ID_RLX6_DRAM_BOOT,
+					fw_data, fw_size, fw_autoload_mask);
+		}
+	}
+}
+
+static void gfx_v11_0_rlc_backdoor_autoload_copy_sdma_ucode(struct amdgpu_device *adev,
+							uint32_t *fw_autoload_mask)
+{
+	const __le32 *fw_data;
+	uint32_t fw_size;
+	const struct sdma_firmware_header_v2_0 *sdma_hdr;
+
+	sdma_hdr = (const struct sdma_firmware_header_v2_0 *)
+		adev->sdma.instance[0].fw->data;
+	fw_data = (const __le32 *) (adev->sdma.instance[0].fw->data +
+			le32_to_cpu(sdma_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(sdma_hdr->ctx_ucode_size_bytes);
+
+	gfx_v11_0_rlc_backdoor_autoload_copy_ucode(adev,
+			SOC21_FIRMWARE_ID_SDMA_UCODE_TH0, fw_data, fw_size, fw_autoload_mask);
+
+	fw_data = (const __le32 *) (adev->sdma.instance[0].fw->data +
+			le32_to_cpu(sdma_hdr->ctl_ucode_offset));
+	fw_size = le32_to_cpu(sdma_hdr->ctl_ucode_size_bytes);
+
+	gfx_v11_0_rlc_backdoor_autoload_copy_ucode(adev,
+			SOC21_FIRMWARE_ID_SDMA_UCODE_TH1, fw_data, fw_size, fw_autoload_mask);
+}
+
+static void gfx_v11_0_rlc_backdoor_autoload_copy_mes_ucode(struct amdgpu_device *adev,
+							uint32_t *fw_autoload_mask)
+{
+	const __le32 *fw_data;
+	unsigned fw_size;
+	const struct mes_firmware_header_v1_0 *mes_hdr;
+	int pipe, ucode_id, data_id;
+
+	for (pipe = 0; pipe < 2; pipe++) {
+		if (pipe==0) {
+			ucode_id = SOC21_FIRMWARE_ID_RS64_MES_P0;
+			data_id  = SOC21_FIRMWARE_ID_RS64_MES_P0_STACK;
+		} else {
+			ucode_id = SOC21_FIRMWARE_ID_RS64_MES_P1;
+			data_id  = SOC21_FIRMWARE_ID_RS64_MES_P1_STACK;
+		}
+
+		mes_hdr = (const struct mes_firmware_header_v1_0 *)
+			adev->mes.fw[pipe]->data;
+
+		fw_data = (const __le32 *)(adev->mes.fw[pipe]->data +
+				le32_to_cpu(mes_hdr->mes_ucode_offset_bytes));
+		fw_size = le32_to_cpu(mes_hdr->mes_ucode_size_bytes);
+
+		gfx_v11_0_rlc_backdoor_autoload_copy_ucode(adev,
+				ucode_id, fw_data, fw_size, fw_autoload_mask);
+
+		fw_data = (const __le32 *)(adev->mes.fw[pipe]->data +
+				le32_to_cpu(mes_hdr->mes_ucode_data_offset_bytes));
+		fw_size = le32_to_cpu(mes_hdr->mes_ucode_data_size_bytes);
+
+		gfx_v11_0_rlc_backdoor_autoload_copy_ucode(adev,
+				data_id, fw_data, fw_size, fw_autoload_mask);
+	}
+}
+
+static int gfx_v11_0_rlc_backdoor_autoload_enable(struct amdgpu_device *adev)
+{
+	uint32_t rlc_g_offset, rlc_g_size;
+	uint64_t gpu_addr;
+	uint32_t autoload_fw_id[2];
+
+	memset(autoload_fw_id, 0, sizeof(uint32_t) * 2);
+
+	/* RLC autoload sequence 2: copy ucode */
+	gfx_v11_0_rlc_backdoor_autoload_copy_sdma_ucode(adev, autoload_fw_id);
+	gfx_v11_0_rlc_backdoor_autoload_copy_gfx_ucode(adev, autoload_fw_id);
+	gfx_v11_0_rlc_backdoor_autoload_copy_mes_ucode(adev, autoload_fw_id);
+	gfx_v11_0_rlc_backdoor_autoload_copy_toc_ucode(adev, autoload_fw_id);
+
+	rlc_g_offset = rlc_autoload_info[SOC21_FIRMWARE_ID_RLC_G_UCODE].offset;
+	rlc_g_size = rlc_autoload_info[SOC21_FIRMWARE_ID_RLC_G_UCODE].size;
+	gpu_addr = adev->gfx.rlc.rlc_autoload_gpu_addr + rlc_g_offset;
+
+	WREG32_SOC15(GC, 0, regGFX_IMU_RLC_BOOTLOADER_ADDR_HI, upper_32_bits(gpu_addr));
+	WREG32_SOC15(GC, 0, regGFX_IMU_RLC_BOOTLOADER_ADDR_LO, lower_32_bits(gpu_addr));
+
+	WREG32_SOC15(GC, 0, regGFX_IMU_RLC_BOOTLOADER_SIZE, rlc_g_size);
+
+	/* RLC autoload sequence 3: load IMU fw */
+	if (adev->gfx.imu.funcs->load_microcode)
+		adev->gfx.imu.funcs->load_microcode(adev);
+	/* RLC autoload sequence 4 init IMU fw */
+	if (adev->gfx.imu.funcs->setup_imu)
+		adev->gfx.imu.funcs->setup_imu(adev);
+	if (adev->gfx.imu.funcs->start_imu)
+		adev->gfx.imu.funcs->start_imu(adev);
+
+	/* RLC autoload sequence 5 disable gpa mode */
+	gfx_v11_0_disable_gpa_mode(adev);
+
+	return 0;
+}
+
+static void gfx_v11_0_alloc_ip_dump(struct amdgpu_device *adev)
+{
+	uint32_t reg_count = ARRAY_SIZE(gc_reg_list_11_0);
+	uint32_t *ptr;
+	uint32_t inst;
+
+	ptr = kcalloc(reg_count, sizeof(uint32_t), GFP_KERNEL);
+	if (!ptr) {
+		DRM_ERROR("Failed to allocate memory for GFX IP Dump\n");
+		adev->gfx.ip_dump_core = NULL;
+	} else {
+		adev->gfx.ip_dump_core = ptr;
+	}
+
+	/* Allocate memory for compute queue registers for all the instances */
+	reg_count = ARRAY_SIZE(gc_cp_reg_list_11);
+	inst = adev->gfx.mec.num_mec * adev->gfx.mec.num_pipe_per_mec *
+		adev->gfx.mec.num_queue_per_pipe;
+
+	ptr = kcalloc(reg_count * inst, sizeof(uint32_t), GFP_KERNEL);
+	if (!ptr) {
+		DRM_ERROR("Failed to allocate memory for Compute Queues IP Dump\n");
+		adev->gfx.ip_dump_compute_queues = NULL;
+	} else {
+		adev->gfx.ip_dump_compute_queues = ptr;
+	}
+
+	/* Allocate memory for gfx queue registers for all the instances */
+	reg_count = ARRAY_SIZE(gc_gfx_queue_reg_list_11);
+	inst = adev->gfx.me.num_me * adev->gfx.me.num_pipe_per_me *
+		adev->gfx.me.num_queue_per_pipe;
+
+	ptr = kcalloc(reg_count * inst, sizeof(uint32_t), GFP_KERNEL);
+	if (!ptr) {
+		DRM_ERROR("Failed to allocate memory for GFX Queues IP Dump\n");
+		adev->gfx.ip_dump_gfx_queues = NULL;
+	} else {
+		adev->gfx.ip_dump_gfx_queues = ptr;
+	}
+}
+
+static int gfx_v11_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	int i, j, k, r, ring_id;
+	int xcc_id = 0;
+	struct amdgpu_device *adev = ip_block->adev;
+	int num_queue_per_pipe = 1; /* we only enable 1 KGQ per pipe */
+
+	INIT_DELAYED_WORK(&adev->gfx.idle_work, amdgpu_gfx_profile_idle_work_handler);
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 1):
+	case IP_VERSION(11, 0, 2):
+	case IP_VERSION(11, 0, 3):
+	case IP_VERSION(11, 0, 4):
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 1):
+	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
+		adev->gfx.me.num_me = 1;
+		adev->gfx.me.num_pipe_per_me = 1;
+		adev->gfx.me.num_queue_per_pipe = 2;
+		adev->gfx.mec.num_mec = 1;
+		adev->gfx.mec.num_pipe_per_mec = 4;
+		adev->gfx.mec.num_queue_per_pipe = 4;
+		break;
+	default:
+		adev->gfx.me.num_me = 1;
+		adev->gfx.me.num_pipe_per_me = 1;
+		adev->gfx.me.num_queue_per_pipe = 1;
+		adev->gfx.mec.num_mec = 1;
+		adev->gfx.mec.num_pipe_per_mec = 4;
+		adev->gfx.mec.num_queue_per_pipe = 8;
+		break;
+	}
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 2):
+	case IP_VERSION(11, 0, 3):
+		if (!adev->gfx.disable_uq &&
+		    adev->gfx.me_fw_version  >= 2420 &&
+		    adev->gfx.pfp_fw_version >= 2580 &&
+		    adev->gfx.mec_fw_version >= 2650 &&
+		    adev->mes.fw_version[0] >= 120) {
+			adev->userq_funcs[AMDGPU_HW_IP_GFX] = &userq_mes_funcs;
+			adev->userq_funcs[AMDGPU_HW_IP_COMPUTE] = &userq_mes_funcs;
+		}
+		break;
+	case IP_VERSION(11, 0, 1):
+	case IP_VERSION(11, 0, 4):
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 1):
+	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
+		/* add firmware version checks here */
+		if (0 && !adev->gfx.disable_uq) {
+			adev->userq_funcs[AMDGPU_HW_IP_GFX] = &userq_mes_funcs;
+			adev->userq_funcs[AMDGPU_HW_IP_COMPUTE] = &userq_mes_funcs;
+		}
+		break;
+	default:
+		break;
+	}
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 2):
+	case IP_VERSION(11, 0, 3):
+		adev->gfx.cleaner_shader_ptr = gfx_11_0_3_cleaner_shader_hex;
+		adev->gfx.cleaner_shader_size = sizeof(gfx_11_0_3_cleaner_shader_hex);
+		if (adev->gfx.me_fw_version  >= 2280 &&
+		    adev->gfx.pfp_fw_version >= 2370 &&
+		    adev->gfx.mec_fw_version >= 2450  &&
+		    adev->mes.fw_version[0] >= 99) {
+			adev->gfx.enable_cleaner_shader = true;
+			r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);
+			if (r) {
+				adev->gfx.enable_cleaner_shader = false;
+				dev_err(adev->dev, "Failed to initialize cleaner shader\n");
+			}
+		}
+		break;
+	case IP_VERSION(11, 0, 1):
+	case IP_VERSION(11, 0, 4):
+		adev->gfx.cleaner_shader_ptr = gfx_11_0_3_cleaner_shader_hex;
+		adev->gfx.cleaner_shader_size = sizeof(gfx_11_0_3_cleaner_shader_hex);
+		if (adev->gfx.pfp_fw_version >= 102 &&
+		    adev->gfx.mec_fw_version >= 66 &&
+		    adev->mes.fw_version[0] >= 128) {
+			adev->gfx.enable_cleaner_shader = true;
+			r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);
+			if (r) {
+				adev->gfx.enable_cleaner_shader = false;
+				dev_err(adev->dev, "Failed to initialize cleaner shader\n");
+			}
+		}
+		break;
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 1):
+		adev->gfx.cleaner_shader_ptr = gfx_11_0_3_cleaner_shader_hex;
+		adev->gfx.cleaner_shader_size = sizeof(gfx_11_0_3_cleaner_shader_hex);
+		if (adev->gfx.mec_fw_version >= 26 &&
+		    adev->mes.fw_version[0] >= 114) {
+			adev->gfx.enable_cleaner_shader = true;
+			r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);
+			if (r) {
+				adev->gfx.enable_cleaner_shader = false;
+				dev_err(adev->dev, "Failed to initialize cleaner shader\n");
+			}
+		}
+		break;
+	case IP_VERSION(11, 5, 2):
+		adev->gfx.cleaner_shader_ptr = gfx_11_0_3_cleaner_shader_hex;
+		adev->gfx.cleaner_shader_size = sizeof(gfx_11_0_3_cleaner_shader_hex);
+		if (adev->gfx.me_fw_version  >= 12 &&
+		    adev->gfx.pfp_fw_version >= 15 &&
+		    adev->gfx.mec_fw_version >= 15) {
+			adev->gfx.enable_cleaner_shader = true;
+			r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);
+			if (r) {
+				adev->gfx.enable_cleaner_shader = false;
+				dev_err(adev->dev, "Failed to initialize cleaner shader\n");
+			}
+		}
+		break;
+	case IP_VERSION(11, 5, 3):
+		adev->gfx.cleaner_shader_ptr = gfx_11_0_3_cleaner_shader_hex;
+		adev->gfx.cleaner_shader_size = sizeof(gfx_11_0_3_cleaner_shader_hex);
+		if (adev->gfx.me_fw_version  >= 7 &&
+		    adev->gfx.pfp_fw_version >= 8 &&
+		    adev->gfx.mec_fw_version >= 8) {
+			adev->gfx.enable_cleaner_shader = true;
+			r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);
+			if (r) {
+				adev->gfx.enable_cleaner_shader = false;
+				dev_err(adev->dev, "Failed to initialize cleaner shader\n");
+			}
+		}
+		break;
+	default:
+		adev->gfx.enable_cleaner_shader = false;
+		break;
+	}
+
+	/* Enable CG flag in one VF mode for enabling RLC safe mode enter/exit */
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(11, 0, 3) &&
+	    amdgpu_sriov_is_pp_one_vf(adev))
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_CGCG;
+
+	/* EOP Event */
+	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GRBM_CP,
+			      GFX_11_0_0__SRCID__CP_EOP_INTERRUPT,
+			      &adev->gfx.eop_irq);
+	if (r)
+		return r;
+
+	/* Bad opcode Event */
+	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GRBM_CP,
+			      GFX_11_0_0__SRCID__CP_BAD_OPCODE_ERROR,
+			      &adev->gfx.bad_op_irq);
+	if (r)
+		return r;
+
+	/* Privileged reg */
+	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GRBM_CP,
+			      GFX_11_0_0__SRCID__CP_PRIV_REG_FAULT,
+			      &adev->gfx.priv_reg_irq);
+	if (r)
+		return r;
+
+	/* Privileged inst */
+	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GRBM_CP,
+			      GFX_11_0_0__SRCID__CP_PRIV_INSTR_FAULT,
+			      &adev->gfx.priv_inst_irq);
+	if (r)
+		return r;
+
+	/* FED error */
+	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GFX,
+				  GFX_11_0_0__SRCID__RLC_GC_FED_INTERRUPT,
+				  &adev->gfx.rlc_gc_fed_irq);
+	if (r)
+		return r;
+
+	adev->gfx.gfx_current_status = AMDGPU_GFX_NORMAL_MODE;
+
+	gfx_v11_0_me_init(adev);
+
+	r = gfx_v11_0_rlc_init(adev);
+	if (r) {
+		DRM_ERROR("Failed to init rlc BOs!\n");
+		return r;
+	}
+
+	r = gfx_v11_0_mec_init(adev);
+	if (r) {
+		DRM_ERROR("Failed to init MEC BOs!\n");
+		return r;
+	}
+
+	if (adev->gfx.num_gfx_rings) {
+		ring_id = 0;
+		/* set up the gfx ring */
+		for (i = 0; i < adev->gfx.me.num_me; i++) {
+			for (j = 0; j < num_queue_per_pipe; j++) {
+				for (k = 0; k < adev->gfx.me.num_pipe_per_me; k++) {
+					if (!amdgpu_gfx_is_me_queue_enabled(adev, i, k, j))
+						continue;
+
+					r = gfx_v11_0_gfx_ring_init(adev, ring_id,
+								    i, k, j);
+					if (r)
+						return r;
+					ring_id++;
+				}
+			}
+		}
+	}
+
+	if (adev->gfx.num_compute_rings) {
+		ring_id = 0;
+		/* set up the compute queues - allocate horizontally across pipes */
+		for (i = 0; i < adev->gfx.mec.num_mec; ++i) {
+			for (j = 0; j < adev->gfx.mec.num_queue_per_pipe; j++) {
+				for (k = 0; k < adev->gfx.mec.num_pipe_per_mec; k++) {
+					if (!amdgpu_gfx_is_mec_queue_enabled(adev, 0, i,
+									     k, j))
+						continue;
+
+					r = gfx_v11_0_compute_ring_init(adev, ring_id,
+									i, k, j);
+					if (r)
+						return r;
+
+					ring_id++;
+				}
+			}
+		}
+	}
+
+	adev->gfx.gfx_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.gfx_ring[0]);
+	adev->gfx.compute_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.compute_ring[0]);
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 2):
+	case IP_VERSION(11, 0, 3):
+		if ((adev->gfx.me_fw_version >= 2280) &&
+		    (adev->gfx.mec_fw_version >= 2410) &&
+		    !amdgpu_sriov_vf(adev)) {
+			adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+			adev->gfx.gfx_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+		}
+		break;
+	default:
+		if (!amdgpu_sriov_vf(adev)) {
+			adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+			adev->gfx.gfx_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+		}
+		break;
+	}
+
+	if (!adev->enable_mes_kiq) {
+		r = amdgpu_gfx_kiq_init(adev, GFX11_MEC_HPD_SIZE, 0);
+		if (r) {
+			DRM_ERROR("Failed to init KIQ BOs!\n");
+			return r;
+		}
+
+		r = amdgpu_gfx_kiq_init_ring(adev, xcc_id);
+		if (r)
+			return r;
+	}
+
+	r = amdgpu_gfx_mqd_sw_init(adev, sizeof(struct v11_compute_mqd), 0);
+	if (r)
+		return r;
+
+	/* allocate visible FB for rlc auto-loading fw */
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO) {
+		r = gfx_v11_0_rlc_autoload_buffer_init(adev);
+		if (r)
+			return r;
+	}
+
+	r = gfx_v11_0_gpu_early_init(adev);
+	if (r)
+		return r;
+
+	if (amdgpu_gfx_ras_sw_init(adev)) {
+		dev_err(adev->dev, "Failed to initialize gfx ras block!\n");
+		return -EINVAL;
+	}
+
+	gfx_v11_0_alloc_ip_dump(adev);
+
+	r = amdgpu_gfx_sysfs_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static void gfx_v11_0_pfp_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->gfx.pfp.pfp_fw_obj,
+			      &adev->gfx.pfp.pfp_fw_gpu_addr,
+			      (void **)&adev->gfx.pfp.pfp_fw_ptr);
+
+	amdgpu_bo_free_kernel(&adev->gfx.pfp.pfp_fw_data_obj,
+			      &adev->gfx.pfp.pfp_fw_data_gpu_addr,
+			      (void **)&adev->gfx.pfp.pfp_fw_data_ptr);
+}
+
+static void gfx_v11_0_me_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->gfx.me.me_fw_obj,
+			      &adev->gfx.me.me_fw_gpu_addr,
+			      (void **)&adev->gfx.me.me_fw_ptr);
+
+	amdgpu_bo_free_kernel(&adev->gfx.me.me_fw_data_obj,
+			       &adev->gfx.me.me_fw_data_gpu_addr,
+			       (void **)&adev->gfx.me.me_fw_data_ptr);
+}
+
+static void gfx_v11_0_rlc_autoload_buffer_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.rlc_autoload_bo,
+			&adev->gfx.rlc.rlc_autoload_gpu_addr,
+			(void **)&adev->gfx.rlc.rlc_autoload_ptr);
+}
+
+static int gfx_v11_0_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	int i;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+		amdgpu_ring_fini(&adev->gfx.gfx_ring[i]);
+	for (i = 0; i < adev->gfx.num_compute_rings; i++)
+		amdgpu_ring_fini(&adev->gfx.compute_ring[i]);
+
+	amdgpu_gfx_mqd_sw_fini(adev, 0);
+
+	if (!adev->enable_mes_kiq) {
+		amdgpu_gfx_kiq_free_ring(&adev->gfx.kiq[0].ring);
+		amdgpu_gfx_kiq_fini(adev, 0);
+	}
+
+	amdgpu_gfx_cleaner_shader_sw_fini(adev);
+
+	gfx_v11_0_pfp_fini(adev);
+	gfx_v11_0_me_fini(adev);
+	gfx_v11_0_rlc_fini(adev);
+	gfx_v11_0_mec_fini(adev);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO)
+		gfx_v11_0_rlc_autoload_buffer_fini(adev);
+
+	gfx_v11_0_free_microcode(adev);
+
+	amdgpu_gfx_sysfs_fini(adev);
+
+	kfree(adev->gfx.ip_dump_core);
+	kfree(adev->gfx.ip_dump_compute_queues);
+	kfree(adev->gfx.ip_dump_gfx_queues);
+
+	return 0;
+}
+
+static void gfx_v11_0_select_se_sh(struct amdgpu_device *adev, u32 se_num,
+				   u32 sh_num, u32 instance, int xcc_id)
+{
+	u32 data;
+
+	if (instance == 0xffffffff)
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX,
+				     INSTANCE_BROADCAST_WRITES, 1);
+	else
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_INDEX,
+				     instance);
+
+	if (se_num == 0xffffffff)
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_BROADCAST_WRITES,
+				     1);
+	else
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_INDEX, se_num);
+
+	if (sh_num == 0xffffffff)
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SA_BROADCAST_WRITES,
+				     1);
+	else
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SA_INDEX, sh_num);
+
+	WREG32_SOC15(GC, 0, regGRBM_GFX_INDEX, data);
+}
+
+static u32 gfx_v11_0_get_sa_active_bitmap(struct amdgpu_device *adev)
+{
+	u32 gc_disabled_sa_mask, gc_user_disabled_sa_mask, sa_mask;
+
+	gc_disabled_sa_mask = RREG32_SOC15(GC, 0, regCC_GC_SA_UNIT_DISABLE);
+	gc_disabled_sa_mask = REG_GET_FIELD(gc_disabled_sa_mask,
+					   CC_GC_SA_UNIT_DISABLE,
+					   SA_DISABLE);
+	gc_user_disabled_sa_mask = RREG32_SOC15(GC, 0, regGC_USER_SA_UNIT_DISABLE);
+	gc_user_disabled_sa_mask = REG_GET_FIELD(gc_user_disabled_sa_mask,
+						 GC_USER_SA_UNIT_DISABLE,
+						 SA_DISABLE);
+	sa_mask = amdgpu_gfx_create_bitmask(adev->gfx.config.max_sh_per_se *
+					    adev->gfx.config.max_shader_engines);
+
+	return sa_mask & (~(gc_disabled_sa_mask | gc_user_disabled_sa_mask));
+}
+
+static u32 gfx_v11_0_get_rb_active_bitmap(struct amdgpu_device *adev)
+{
+	u32 gc_disabled_rb_mask, gc_user_disabled_rb_mask;
+	u32 rb_mask;
+
+	gc_disabled_rb_mask = RREG32_SOC15(GC, 0, regCC_RB_BACKEND_DISABLE);
+	gc_disabled_rb_mask = REG_GET_FIELD(gc_disabled_rb_mask,
+					    CC_RB_BACKEND_DISABLE,
+					    BACKEND_DISABLE);
+	gc_user_disabled_rb_mask = RREG32_SOC15(GC, 0, regGC_USER_RB_BACKEND_DISABLE);
+	gc_user_disabled_rb_mask = REG_GET_FIELD(gc_user_disabled_rb_mask,
+						 GC_USER_RB_BACKEND_DISABLE,
+						 BACKEND_DISABLE);
+	rb_mask = amdgpu_gfx_create_bitmask(adev->gfx.config.max_backends_per_se *
+					    adev->gfx.config.max_shader_engines);
+
+	return rb_mask & (~(gc_disabled_rb_mask | gc_user_disabled_rb_mask));
+}
+
+static void gfx_v11_0_setup_rb(struct amdgpu_device *adev)
+{
+	u32 rb_bitmap_per_sa;
+	u32 rb_bitmap_width_per_sa;
+	u32 max_sa;
+	u32 active_sa_bitmap;
+	u32 global_active_rb_bitmap;
+	u32 active_rb_bitmap = 0;
+	u32 i;
+
+	/* query sa bitmap from SA_UNIT_DISABLE registers */
+	active_sa_bitmap = gfx_v11_0_get_sa_active_bitmap(adev);
+	/* query rb bitmap from RB_BACKEND_DISABLE registers */
+	global_active_rb_bitmap = gfx_v11_0_get_rb_active_bitmap(adev);
+
+	/* generate active rb bitmap according to active sa bitmap */
+	max_sa = adev->gfx.config.max_shader_engines *
+		 adev->gfx.config.max_sh_per_se;
+	rb_bitmap_width_per_sa = adev->gfx.config.max_backends_per_se /
+				 adev->gfx.config.max_sh_per_se;
+	rb_bitmap_per_sa = amdgpu_gfx_create_bitmask(rb_bitmap_width_per_sa);
+
+	for (i = 0; i < max_sa; i++) {
+		if (active_sa_bitmap & (1 << i))
+			active_rb_bitmap |= (rb_bitmap_per_sa << (i * rb_bitmap_width_per_sa));
+	}
+
+	active_rb_bitmap &= global_active_rb_bitmap;
+	adev->gfx.config.backend_enable_mask = active_rb_bitmap;
+	adev->gfx.config.num_rbs = hweight32(active_rb_bitmap);
+}
+
+#define DEFAULT_SH_MEM_BASES	(0x6000)
+#define LDS_APP_BASE           0x1
+#define SCRATCH_APP_BASE       0x2
+
+static void gfx_v11_0_init_compute_vmid(struct amdgpu_device *adev)
+{
+	int i;
+	uint32_t sh_mem_bases;
+	uint32_t data;
+
+	/*
+	 * Configure apertures:
+	 * LDS:         0x60000000'00000000 - 0x60000001'00000000 (4GB)
+	 * Scratch:     0x60000001'00000000 - 0x60000002'00000000 (4GB)
+	 * GPUVM:       0x60010000'00000000 - 0x60020000'00000000 (1TB)
+	 */
+	sh_mem_bases = (LDS_APP_BASE << SH_MEM_BASES__SHARED_BASE__SHIFT) |
+			SCRATCH_APP_BASE;
+
+	mutex_lock(&adev->srbm_mutex);
+	for (i = adev->vm_manager.first_kfd_vmid; i < AMDGPU_NUM_VMID; i++) {
+		soc21_grbm_select(adev, 0, 0, 0, i);
+		/* CP and shaders */
+		WREG32_SOC15(GC, 0, regSH_MEM_CONFIG, DEFAULT_SH_MEM_CONFIG);
+		WREG32_SOC15(GC, 0, regSH_MEM_BASES, sh_mem_bases);
+
+		/* Enable trap for each kfd vmid. */
+		data = RREG32_SOC15(GC, 0, regSPI_GDBG_PER_VMID_CNTL);
+		data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, TRAP_EN, 1);
+		WREG32_SOC15(GC, 0, regSPI_GDBG_PER_VMID_CNTL, data);
+	}
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	/*
+	 * Initialize all compute VMIDs to have no GDS, GWS, or OA
+	 * access. These should be enabled by FW for target VMIDs.
+	 */
+	for (i = adev->vm_manager.first_kfd_vmid; i < AMDGPU_NUM_VMID; i++) {
+		WREG32_SOC15_OFFSET(GC, 0, regGDS_VMID0_BASE, 2 * i, 0);
+		WREG32_SOC15_OFFSET(GC, 0, regGDS_VMID0_SIZE, 2 * i, 0);
+		WREG32_SOC15_OFFSET(GC, 0, regGDS_GWS_VMID0, i, 0);
+		WREG32_SOC15_OFFSET(GC, 0, regGDS_OA_VMID0, i, 0);
+	}
+}
+
+static void gfx_v11_0_init_gds_vmid(struct amdgpu_device *adev)
+{
+	int vmid;
+
+	/*
+	 * Initialize all compute and user-gfx VMIDs to have no GDS, GWS, or OA
+	 * access. Compute VMIDs should be enabled by FW for target VMIDs,
+	 * the driver can enable them for graphics. VMID0 should maintain
+	 * access so that HWS firmware can save/restore entries.
+	 */
+	for (vmid = 1; vmid < 16; vmid++) {
+		WREG32_SOC15_OFFSET(GC, 0, regGDS_VMID0_BASE, 2 * vmid, 0);
+		WREG32_SOC15_OFFSET(GC, 0, regGDS_VMID0_SIZE, 2 * vmid, 0);
+		WREG32_SOC15_OFFSET(GC, 0, regGDS_GWS_VMID0, vmid, 0);
+		WREG32_SOC15_OFFSET(GC, 0, regGDS_OA_VMID0, vmid, 0);
+	}
+}
+
+static void gfx_v11_0_tcp_harvest(struct amdgpu_device *adev)
+{
+	/* TODO: harvest feature to be added later. */
+}
+
+static void gfx_v11_0_get_tcc_info(struct amdgpu_device *adev)
+{
+	/* TCCs are global (not instanced). */
+	uint32_t tcc_disable = RREG32_SOC15(GC, 0, regCGTS_TCC_DISABLE) |
+			       RREG32_SOC15(GC, 0, regCGTS_USER_TCC_DISABLE);
+
+	adev->gfx.config.tcc_disabled_mask =
+		REG_GET_FIELD(tcc_disable, CGTS_TCC_DISABLE, TCC_DISABLE) |
+		(REG_GET_FIELD(tcc_disable, CGTS_TCC_DISABLE, HI_TCC_DISABLE) << 16);
+}
+
+static void gfx_v11_0_constants_init(struct amdgpu_device *adev)
+{
+	u32 tmp;
+	int i;
+
+	if (!amdgpu_sriov_vf(adev))
+		WREG32_FIELD15_PREREG(GC, 0, GRBM_CNTL, READ_TIMEOUT, 0xff);
+
+	gfx_v11_0_setup_rb(adev);
+	gfx_v11_0_get_cu_info(adev, &adev->gfx.cu_info);
+	gfx_v11_0_get_tcc_info(adev);
+	adev->gfx.config.pa_sc_tile_steering_override = 0;
+
+	/* Set whether texture coordinate truncation is conformant. */
+	tmp = RREG32_SOC15(GC, 0, regTA_CNTL2);
+	adev->gfx.config.ta_cntl2_truncate_coord_mode =
+		REG_GET_FIELD(tmp, TA_CNTL2, TRUNCATE_COORD_MODE);
+
+	/* XXX SH_MEM regs */
+	/* where to put LDS, scratch, GPUVM in FSA64 space */
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < adev->vm_manager.id_mgr[AMDGPU_GFXHUB(0)].num_ids; i++) {
+		soc21_grbm_select(adev, 0, 0, 0, i);
+		/* CP and shaders */
+		WREG32_SOC15(GC, 0, regSH_MEM_CONFIG, DEFAULT_SH_MEM_CONFIG);
+		if (i != 0) {
+			tmp = REG_SET_FIELD(0, SH_MEM_BASES, PRIVATE_BASE,
+				(adev->gmc.private_aperture_start >> 48));
+			tmp = REG_SET_FIELD(tmp, SH_MEM_BASES, SHARED_BASE,
+				(adev->gmc.shared_aperture_start >> 48));
+			WREG32_SOC15(GC, 0, regSH_MEM_BASES, tmp);
+		}
+	}
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+
+	mutex_unlock(&adev->srbm_mutex);
+
+	gfx_v11_0_init_compute_vmid(adev);
+	gfx_v11_0_init_gds_vmid(adev);
+}
+
+static u32 gfx_v11_0_get_cpg_int_cntl(struct amdgpu_device *adev,
+				      int me, int pipe)
+{
+	if (me != 0)
+		return 0;
+
+	switch (pipe) {
+	case 0:
+		return SOC15_REG_OFFSET(GC, 0, regCP_INT_CNTL_RING0);
+	case 1:
+		return SOC15_REG_OFFSET(GC, 0, regCP_INT_CNTL_RING1);
+	default:
+		return 0;
+	}
+}
+
+static u32 gfx_v11_0_get_cpc_int_cntl(struct amdgpu_device *adev,
+				      int me, int pipe)
+{
+	/*
+	 * amdgpu controls only the first MEC. That's why this function only
+	 * handles the setting of interrupts for this specific MEC. All other
+	 * pipes' interrupts are set by amdkfd.
+	 */
+	if (me != 1)
+		return 0;
+
+	switch (pipe) {
+	case 0:
+		return SOC15_REG_OFFSET(GC, 0, regCP_ME1_PIPE0_INT_CNTL);
+	case 1:
+		return SOC15_REG_OFFSET(GC, 0, regCP_ME1_PIPE1_INT_CNTL);
+	case 2:
+		return SOC15_REG_OFFSET(GC, 0, regCP_ME1_PIPE2_INT_CNTL);
+	case 3:
+		return SOC15_REG_OFFSET(GC, 0, regCP_ME1_PIPE3_INT_CNTL);
+	default:
+		return 0;
+	}
+}
+
+static void gfx_v11_0_enable_gui_idle_interrupt(struct amdgpu_device *adev,
+					       bool enable)
+{
+	u32 tmp, cp_int_cntl_reg;
+	int i, j;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	for (i = 0; i < adev->gfx.me.num_me; i++) {
+		for (j = 0; j < adev->gfx.me.num_pipe_per_me; j++) {
+			cp_int_cntl_reg = gfx_v11_0_get_cpg_int_cntl(adev, i, j);
+
+			if (cp_int_cntl_reg) {
+				tmp = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+				tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_BUSY_INT_ENABLE,
+						    enable ? 1 : 0);
+				tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_EMPTY_INT_ENABLE,
+						    enable ? 1 : 0);
+				tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CMP_BUSY_INT_ENABLE,
+						    enable ? 1 : 0);
+				tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, GFX_IDLE_INT_ENABLE,
+						    enable ? 1 : 0);
+				WREG32_SOC15_IP(GC, cp_int_cntl_reg, tmp);
+			}
+		}
+	}
+}
+
+static int gfx_v11_0_init_csb(struct amdgpu_device *adev)
+{
+	adev->gfx.rlc.funcs->get_csb_buffer(adev, adev->gfx.rlc.cs_ptr);
+
+	WREG32_SOC15(GC, 0, regRLC_CSIB_ADDR_HI,
+			adev->gfx.rlc.clear_state_gpu_addr >> 32);
+	WREG32_SOC15(GC, 0, regRLC_CSIB_ADDR_LO,
+			adev->gfx.rlc.clear_state_gpu_addr & 0xfffffffc);
+	WREG32_SOC15(GC, 0, regRLC_CSIB_LENGTH, adev->gfx.rlc.clear_state_size);
+
+	return 0;
+}
+
+static void gfx_v11_0_rlc_stop(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SOC15(GC, 0, regRLC_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, RLC_CNTL, RLC_ENABLE_F32, 0);
+	WREG32_SOC15(GC, 0, regRLC_CNTL, tmp);
+}
+
+static void gfx_v11_0_rlc_reset(struct amdgpu_device *adev)
+{
+	WREG32_FIELD15_PREREG(GC, 0, GRBM_SOFT_RESET, SOFT_RESET_RLC, 1);
+	udelay(50);
+	WREG32_FIELD15_PREREG(GC, 0, GRBM_SOFT_RESET, SOFT_RESET_RLC, 0);
+	udelay(50);
+}
+
+static void gfx_v11_0_rlc_smu_handshake_cntl(struct amdgpu_device *adev,
+					     bool enable)
+{
+	uint32_t rlc_pg_cntl;
+
+	rlc_pg_cntl = RREG32_SOC15(GC, 0, regRLC_PG_CNTL);
+
+	if (!enable) {
+		/* RLC_PG_CNTL[23] = 0 (default)
+		 * RLC will wait for handshake acks with SMU
+		 * GFXOFF will be enabled
+		 * RLC_PG_CNTL[23] = 1
+		 * RLC will not issue any message to SMU
+		 * hence no handshake between SMU & RLC
+		 * GFXOFF will be disabled
+		 */
+		rlc_pg_cntl |= RLC_PG_CNTL__SMU_HANDSHAKE_DISABLE_MASK;
+	} else
+		rlc_pg_cntl &= ~RLC_PG_CNTL__SMU_HANDSHAKE_DISABLE_MASK;
+	WREG32_SOC15(GC, 0, regRLC_PG_CNTL, rlc_pg_cntl);
+}
+
+static void gfx_v11_0_rlc_start(struct amdgpu_device *adev)
+{
+	/* TODO: enable rlc & smu handshake until smu
+	 * and gfxoff feature works as expected */
+	if (!(amdgpu_pp_feature_mask & PP_GFXOFF_MASK))
+		gfx_v11_0_rlc_smu_handshake_cntl(adev, false);
+
+	WREG32_FIELD15_PREREG(GC, 0, RLC_CNTL, RLC_ENABLE_F32, 1);
+	udelay(50);
+}
+
+static void gfx_v11_0_rlc_enable_srm(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* enable Save Restore Machine */
+	tmp = RREG32(SOC15_REG_OFFSET(GC, 0, regRLC_SRM_CNTL));
+	tmp |= RLC_SRM_CNTL__AUTO_INCR_ADDR_MASK;
+	tmp |= RLC_SRM_CNTL__SRM_ENABLE_MASK;
+	WREG32(SOC15_REG_OFFSET(GC, 0, regRLC_SRM_CNTL), tmp);
+}
+
+static void gfx_v11_0_load_rlcg_microcode(struct amdgpu_device *adev)
+{
+	const struct rlc_firmware_header_v2_0 *hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+
+	hdr = (const struct rlc_firmware_header_v2_0 *)adev->gfx.rlc_fw->data;
+	fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+			   le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+
+	WREG32_SOC15(GC, 0, regRLC_GPM_UCODE_ADDR,
+		     RLCG_UCODE_LOADING_START_ADDRESS);
+
+	for (i = 0; i < fw_size; i++)
+		WREG32_SOC15(GC, 0, regRLC_GPM_UCODE_DATA,
+			     le32_to_cpup(fw_data++));
+
+	WREG32_SOC15(GC, 0, regRLC_GPM_UCODE_ADDR, adev->gfx.rlc_fw_version);
+}
+
+static void gfx_v11_0_load_rlc_iram_dram_microcode(struct amdgpu_device *adev)
+{
+	const struct rlc_firmware_header_v2_2 *hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+	u32 tmp;
+
+	hdr = (const struct rlc_firmware_header_v2_2 *)adev->gfx.rlc_fw->data;
+
+	fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+			le32_to_cpu(hdr->rlc_iram_ucode_offset_bytes));
+	fw_size = le32_to_cpu(hdr->rlc_iram_ucode_size_bytes) / 4;
+
+	WREG32_SOC15(GC, 0, regRLC_LX6_IRAM_ADDR, 0);
+
+	for (i = 0; i < fw_size; i++) {
+		if ((amdgpu_emu_mode == 1) && (i % 100 == 99))
+			msleep(1);
+		WREG32_SOC15(GC, 0, regRLC_LX6_IRAM_DATA,
+				le32_to_cpup(fw_data++));
+	}
+
+	WREG32_SOC15(GC, 0, regRLC_LX6_IRAM_ADDR, adev->gfx.rlc_fw_version);
+
+	fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+			le32_to_cpu(hdr->rlc_dram_ucode_offset_bytes));
+	fw_size = le32_to_cpu(hdr->rlc_dram_ucode_size_bytes) / 4;
+
+	WREG32_SOC15(GC, 0, regRLC_LX6_DRAM_ADDR, 0);
+	for (i = 0; i < fw_size; i++) {
+		if ((amdgpu_emu_mode == 1) && (i % 100 == 99))
+			msleep(1);
+		WREG32_SOC15(GC, 0, regRLC_LX6_DRAM_DATA,
+				le32_to_cpup(fw_data++));
+	}
+
+	WREG32_SOC15(GC, 0, regRLC_LX6_IRAM_ADDR, adev->gfx.rlc_fw_version);
+
+	tmp = RREG32_SOC15(GC, 0, regRLC_LX6_CNTL);
+	tmp = REG_SET_FIELD(tmp, RLC_LX6_CNTL, PDEBUG_ENABLE, 1);
+	tmp = REG_SET_FIELD(tmp, RLC_LX6_CNTL, BRESET, 0);
+	WREG32_SOC15(GC, 0, regRLC_LX6_CNTL, tmp);
+}
+
+static void gfx_v11_0_load_rlcp_rlcv_microcode(struct amdgpu_device *adev)
+{
+	const struct rlc_firmware_header_v2_3 *hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+	u32 tmp;
+
+	hdr = (const struct rlc_firmware_header_v2_3 *)adev->gfx.rlc_fw->data;
+
+	fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+			le32_to_cpu(hdr->rlcp_ucode_offset_bytes));
+	fw_size = le32_to_cpu(hdr->rlcp_ucode_size_bytes) / 4;
+
+	WREG32_SOC15(GC, 0, regRLC_PACE_UCODE_ADDR, 0);
+
+	for (i = 0; i < fw_size; i++) {
+		if ((amdgpu_emu_mode == 1) && (i % 100 == 99))
+			msleep(1);
+		WREG32_SOC15(GC, 0, regRLC_PACE_UCODE_DATA,
+				le32_to_cpup(fw_data++));
+	}
+
+	WREG32_SOC15(GC, 0, regRLC_PACE_UCODE_ADDR, adev->gfx.rlc_fw_version);
+
+	tmp = RREG32_SOC15(GC, 0, regRLC_GPM_THREAD_ENABLE);
+	tmp = REG_SET_FIELD(tmp, RLC_GPM_THREAD_ENABLE, THREAD1_ENABLE, 1);
+	WREG32_SOC15(GC, 0, regRLC_GPM_THREAD_ENABLE, tmp);
+
+	fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+			le32_to_cpu(hdr->rlcv_ucode_offset_bytes));
+	fw_size = le32_to_cpu(hdr->rlcv_ucode_size_bytes) / 4;
+
+	WREG32_SOC15(GC, 0, regRLC_GPU_IOV_UCODE_ADDR, 0);
+
+	for (i = 0; i < fw_size; i++) {
+		if ((amdgpu_emu_mode == 1) && (i % 100 == 99))
+			msleep(1);
+		WREG32_SOC15(GC, 0, regRLC_GPU_IOV_UCODE_DATA,
+				le32_to_cpup(fw_data++));
+	}
+
+	WREG32_SOC15(GC, 0, regRLC_GPU_IOV_UCODE_ADDR, adev->gfx.rlc_fw_version);
+
+	tmp = RREG32_SOC15(GC, 0, regRLC_GPU_IOV_F32_CNTL);
+	tmp = REG_SET_FIELD(tmp, RLC_GPU_IOV_F32_CNTL, ENABLE, 1);
+	WREG32_SOC15(GC, 0, regRLC_GPU_IOV_F32_CNTL, tmp);
+}
+
+static int gfx_v11_0_rlc_load_microcode(struct amdgpu_device *adev)
+{
+	const struct rlc_firmware_header_v2_0 *hdr;
+	uint16_t version_major;
+	uint16_t version_minor;
+
+	if (!adev->gfx.rlc_fw)
+		return -EINVAL;
+
+	hdr = (const struct rlc_firmware_header_v2_0 *)adev->gfx.rlc_fw->data;
+	amdgpu_ucode_print_rlc_hdr(&hdr->header);
+
+	version_major = le16_to_cpu(hdr->header.header_version_major);
+	version_minor = le16_to_cpu(hdr->header.header_version_minor);
+
+	if (version_major == 2) {
+		gfx_v11_0_load_rlcg_microcode(adev);
+		if (amdgpu_dpm == 1) {
+			if (version_minor >= 2)
+				gfx_v11_0_load_rlc_iram_dram_microcode(adev);
+			if (version_minor == 3)
+				gfx_v11_0_load_rlcp_rlcv_microcode(adev);
+		}
+		
+		return 0;
+	}
+
+	return -EINVAL;
+}
+
+static int gfx_v11_0_rlc_resume(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		gfx_v11_0_init_csb(adev);
+
+		if (!amdgpu_sriov_vf(adev)) /* enable RLC SRM */
+			gfx_v11_0_rlc_enable_srm(adev);
+	} else {
+		if (amdgpu_sriov_vf(adev)) {
+			gfx_v11_0_init_csb(adev);
+			return 0;
+		}
+
+		adev->gfx.rlc.funcs->stop(adev);
+
+		/* disable CG */
+		WREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL, 0);
+
+		/* disable PG */
+		WREG32_SOC15(GC, 0, regRLC_PG_CNTL, 0);
+
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+			/* legacy rlc firmware loading */
+			r = gfx_v11_0_rlc_load_microcode(adev);
+			if (r)
+				return r;
+		}
+
+		gfx_v11_0_init_csb(adev);
+
+		adev->gfx.rlc.funcs->start(adev);
+	}
+	return 0;
+}
+
+static int gfx_v11_0_config_me_cache(struct amdgpu_device *adev, uint64_t addr)
+{
+	uint32_t usec_timeout = 50000;  /* wait for 50ms */
+	uint32_t tmp;
+	int i;
+
+	/* Trigger an invalidation of the L1 instruction caches */
+	tmp = RREG32_SOC15(GC, 0, regCP_ME_IC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_OP_CNTL, INVALIDATE_CACHE, 1);
+	WREG32_SOC15(GC, 0, regCP_ME_IC_OP_CNTL, tmp);
+
+	/* Wait for invalidation complete */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_ME_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_ME_IC_OP_CNTL,
+					INVALIDATE_CACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	if (amdgpu_emu_mode == 1)
+		amdgpu_device_flush_hdp(adev, NULL);
+
+	tmp = RREG32_SOC15(GC, 0, regCP_ME_IC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_BASE_CNTL, CACHE_POLICY, 0);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_BASE_CNTL, EXE_DISABLE, 0);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_BASE_CNTL, ADDRESS_CLAMP, 1);
+	WREG32_SOC15(GC, 0, regCP_ME_IC_BASE_CNTL, tmp);
+
+	/* Program me ucode address into intruction cache address register */
+	WREG32_SOC15(GC, 0, regCP_ME_IC_BASE_LO,
+			lower_32_bits(addr) & 0xFFFFF000);
+	WREG32_SOC15(GC, 0, regCP_ME_IC_BASE_HI,
+			upper_32_bits(addr));
+
+	return 0;
+}
+
+static int gfx_v11_0_config_pfp_cache(struct amdgpu_device *adev, uint64_t addr)
+{
+	uint32_t usec_timeout = 50000;  /* wait for 50ms */
+	uint32_t tmp;
+	int i;
+
+	/* Trigger an invalidation of the L1 instruction caches */
+	tmp = RREG32_SOC15(GC, 0, regCP_PFP_IC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_OP_CNTL, INVALIDATE_CACHE, 1);
+	WREG32_SOC15(GC, 0, regCP_PFP_IC_OP_CNTL, tmp);
+
+	/* Wait for invalidation complete */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_PFP_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_PFP_IC_OP_CNTL,
+					INVALIDATE_CACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	if (amdgpu_emu_mode == 1)
+		amdgpu_device_flush_hdp(adev, NULL);
+
+	tmp = RREG32_SOC15(GC, 0, regCP_PFP_IC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_BASE_CNTL, CACHE_POLICY, 0);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_BASE_CNTL, EXE_DISABLE, 0);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_BASE_CNTL, ADDRESS_CLAMP, 1);
+	WREG32_SOC15(GC, 0, regCP_PFP_IC_BASE_CNTL, tmp);
+
+	/* Program pfp ucode address into intruction cache address register */
+	WREG32_SOC15(GC, 0, regCP_PFP_IC_BASE_LO,
+			lower_32_bits(addr) & 0xFFFFF000);
+	WREG32_SOC15(GC, 0, regCP_PFP_IC_BASE_HI,
+			upper_32_bits(addr));
+
+	return 0;
+}
+
+static int gfx_v11_0_config_mec_cache(struct amdgpu_device *adev, uint64_t addr)
+{
+	uint32_t usec_timeout = 50000;  /* wait for 50ms */
+	uint32_t tmp;
+	int i;
+
+	/* Trigger an invalidation of the L1 instruction caches */
+	tmp = RREG32_SOC15(GC, 0, regCP_CPC_IC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_OP_CNTL, INVALIDATE_CACHE, 1);
+
+	WREG32_SOC15(GC, 0, regCP_CPC_IC_OP_CNTL, tmp);
+
+	/* Wait for invalidation complete */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_CPC_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_CPC_IC_OP_CNTL,
+					INVALIDATE_CACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	if (amdgpu_emu_mode == 1)
+		amdgpu_device_flush_hdp(adev, NULL);
+
+	tmp = RREG32_SOC15(GC, 0, regCP_CPC_IC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, CACHE_POLICY, 0);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, EXE_DISABLE, 0);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, ADDRESS_CLAMP, 1);
+	WREG32_SOC15(GC, 0, regCP_CPC_IC_BASE_CNTL, tmp);
+
+	/* Program mec1 ucode address into intruction cache address register */
+	WREG32_SOC15(GC, 0, regCP_CPC_IC_BASE_LO,
+			lower_32_bits(addr) & 0xFFFFF000);
+	WREG32_SOC15(GC, 0, regCP_CPC_IC_BASE_HI,
+			upper_32_bits(addr));
+
+	return 0;
+}
+
+static int gfx_v11_0_config_pfp_cache_rs64(struct amdgpu_device *adev, uint64_t addr, uint64_t addr2)
+{
+	uint32_t usec_timeout = 50000;  /* wait for 50ms */
+	uint32_t tmp;
+	unsigned i, pipe_id;
+	const struct gfx_firmware_header_v2_0 *pfp_hdr;
+
+	pfp_hdr = (const struct gfx_firmware_header_v2_0 *)
+		adev->gfx.pfp_fw->data;
+
+	WREG32_SOC15(GC, 0, regCP_PFP_IC_BASE_LO,
+		lower_32_bits(addr));
+	WREG32_SOC15(GC, 0, regCP_PFP_IC_BASE_HI,
+		upper_32_bits(addr));
+
+	tmp = RREG32_SOC15(GC, 0, regCP_PFP_IC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_BASE_CNTL, CACHE_POLICY, 0);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_BASE_CNTL, EXE_DISABLE, 0);
+	WREG32_SOC15(GC, 0, regCP_PFP_IC_BASE_CNTL, tmp);
+
+	/*
+	 * Programming any of the CP_PFP_IC_BASE registers
+	 * forces invalidation of the ME L1 I$. Wait for the
+	 * invalidation complete
+	 */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_PFP_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_PFP_IC_OP_CNTL,
+			INVALIDATE_CACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	/* Prime the L1 instruction caches */
+	tmp = RREG32_SOC15(GC, 0, regCP_PFP_IC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_OP_CNTL, PRIME_ICACHE, 1);
+	WREG32_SOC15(GC, 0, regCP_PFP_IC_OP_CNTL, tmp);
+	/* Waiting for cache primed*/
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_PFP_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_PFP_IC_OP_CNTL,
+			ICACHE_PRIMED))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to prime instruction cache\n");
+		return -EINVAL;
+	}
+
+	mutex_lock(&adev->srbm_mutex);
+	for (pipe_id = 0; pipe_id < adev->gfx.me.num_pipe_per_me; pipe_id++) {
+		soc21_grbm_select(adev, 0, pipe_id, 0, 0);
+		WREG32_SOC15(GC, 0, regCP_PFP_PRGRM_CNTR_START,
+			(pfp_hdr->ucode_start_addr_hi << 30) |
+			(pfp_hdr->ucode_start_addr_lo >> 2));
+		WREG32_SOC15(GC, 0, regCP_PFP_PRGRM_CNTR_START_HI,
+			pfp_hdr->ucode_start_addr_hi >> 2);
+
+		/*
+		 * Program CP_ME_CNTL to reset given PIPE to take
+		 * effect of CP_PFP_PRGRM_CNTR_START.
+		 */
+		tmp = RREG32_SOC15(GC, 0, regCP_ME_CNTL);
+		if (pipe_id == 0)
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					PFP_PIPE0_RESET, 1);
+		else
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					PFP_PIPE1_RESET, 1);
+		WREG32_SOC15(GC, 0, regCP_ME_CNTL, tmp);
+
+		/* Clear pfp pipe0 reset bit. */
+		if (pipe_id == 0)
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					PFP_PIPE0_RESET, 0);
+		else
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					PFP_PIPE1_RESET, 0);
+		WREG32_SOC15(GC, 0, regCP_ME_CNTL, tmp);
+
+		WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE0_LO,
+			lower_32_bits(addr2));
+		WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE0_HI,
+			upper_32_bits(addr2));
+	}
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	tmp = RREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_RS64_DC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_RS64_DC_BASE_CNTL, CACHE_POLICY, 0);
+	WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE_CNTL, tmp);
+
+	/* Invalidate the data caches */
+	tmp = RREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_RS64_DC_OP_CNTL, INVALIDATE_DCACHE, 1);
+	WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_OP_CNTL, tmp);
+
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_GFX_RS64_DC_OP_CNTL,
+			INVALIDATE_DCACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate RS64 data cache\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int gfx_v11_0_config_me_cache_rs64(struct amdgpu_device *adev, uint64_t addr, uint64_t addr2)
+{
+	uint32_t usec_timeout = 50000;  /* wait for 50ms */
+	uint32_t tmp;
+	unsigned i, pipe_id;
+	const struct gfx_firmware_header_v2_0 *me_hdr;
+
+	me_hdr = (const struct gfx_firmware_header_v2_0 *)
+		adev->gfx.me_fw->data;
+
+	WREG32_SOC15(GC, 0, regCP_ME_IC_BASE_LO,
+		lower_32_bits(addr));
+	WREG32_SOC15(GC, 0, regCP_ME_IC_BASE_HI,
+		upper_32_bits(addr));
+
+	tmp = RREG32_SOC15(GC, 0, regCP_ME_IC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_BASE_CNTL, CACHE_POLICY, 0);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_BASE_CNTL, EXE_DISABLE, 0);
+	WREG32_SOC15(GC, 0, regCP_ME_IC_BASE_CNTL, tmp);
+
+	/*
+	 * Programming any of the CP_ME_IC_BASE registers
+	 * forces invalidation of the ME L1 I$. Wait for the
+	 * invalidation complete
+	 */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_ME_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_ME_IC_OP_CNTL,
+			INVALIDATE_CACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	/* Prime the instruction caches */
+	tmp = RREG32_SOC15(GC, 0, regCP_ME_IC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_OP_CNTL, PRIME_ICACHE, 1);
+	WREG32_SOC15(GC, 0, regCP_ME_IC_OP_CNTL, tmp);
+
+	/* Waiting for instruction cache primed*/
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_ME_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_ME_IC_OP_CNTL,
+			ICACHE_PRIMED))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to prime instruction cache\n");
+		return -EINVAL;
+	}
+
+	mutex_lock(&adev->srbm_mutex);
+	for (pipe_id = 0; pipe_id < adev->gfx.me.num_pipe_per_me; pipe_id++) {
+		soc21_grbm_select(adev, 0, pipe_id, 0, 0);
+		WREG32_SOC15(GC, 0, regCP_ME_PRGRM_CNTR_START,
+			(me_hdr->ucode_start_addr_hi << 30) |
+			(me_hdr->ucode_start_addr_lo >> 2) );
+		WREG32_SOC15(GC, 0, regCP_ME_PRGRM_CNTR_START_HI,
+			me_hdr->ucode_start_addr_hi>>2);
+
+		/*
+		 * Program CP_ME_CNTL to reset given PIPE to take
+		 * effect of CP_PFP_PRGRM_CNTR_START.
+		 */
+		tmp = RREG32_SOC15(GC, 0, regCP_ME_CNTL);
+		if (pipe_id == 0)
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					ME_PIPE0_RESET, 1);
+		else
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					ME_PIPE1_RESET, 1);
+		WREG32_SOC15(GC, 0, regCP_ME_CNTL, tmp);
+
+		/* Clear pfp pipe0 reset bit. */
+		if (pipe_id == 0)
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					ME_PIPE0_RESET, 0);
+		else
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					ME_PIPE1_RESET, 0);
+		WREG32_SOC15(GC, 0, regCP_ME_CNTL, tmp);
+
+		WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE1_LO,
+			lower_32_bits(addr2));
+		WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE1_HI,
+			upper_32_bits(addr2));
+	}
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	tmp = RREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_RS64_DC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_RS64_DC_BASE_CNTL, CACHE_POLICY, 0);
+	WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE_CNTL, tmp);
+
+	/* Invalidate the data caches */
+	tmp = RREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_RS64_DC_OP_CNTL, INVALIDATE_DCACHE, 1);
+	WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_OP_CNTL, tmp);
+
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_GFX_RS64_DC_OP_CNTL,
+			INVALIDATE_DCACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate RS64 data cache\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int gfx_v11_0_config_mec_cache_rs64(struct amdgpu_device *adev, uint64_t addr, uint64_t addr2)
+{
+	uint32_t usec_timeout = 50000;  /* wait for 50ms */
+	uint32_t tmp;
+	unsigned i;
+	const struct gfx_firmware_header_v2_0 *mec_hdr;
+
+	mec_hdr = (const struct gfx_firmware_header_v2_0 *)
+		adev->gfx.mec_fw->data;
+
+	tmp = RREG32_SOC15(GC, 0, regCP_CPC_IC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, EXE_DISABLE, 0);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, CACHE_POLICY, 0);
+	WREG32_SOC15(GC, 0, regCP_CPC_IC_BASE_CNTL, tmp);
+
+	tmp = RREG32_SOC15(GC, 0, regCP_MEC_DC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_DC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_DC_BASE_CNTL, CACHE_POLICY, 0);
+	WREG32_SOC15(GC, 0, regCP_MEC_DC_BASE_CNTL, tmp);
+
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < adev->gfx.mec.num_pipe_per_mec; i++) {
+		soc21_grbm_select(adev, 1, i, 0, 0);
+
+		WREG32_SOC15(GC, 0, regCP_MEC_MDBASE_LO, addr2);
+		WREG32_SOC15(GC, 0, regCP_MEC_MDBASE_HI,
+		     upper_32_bits(addr2));
+
+		WREG32_SOC15(GC, 0, regCP_MEC_RS64_PRGRM_CNTR_START,
+					mec_hdr->ucode_start_addr_lo >> 2 |
+					mec_hdr->ucode_start_addr_hi << 30);
+		WREG32_SOC15(GC, 0, regCP_MEC_RS64_PRGRM_CNTR_START_HI,
+					mec_hdr->ucode_start_addr_hi >> 2);
+
+		WREG32_SOC15(GC, 0, regCP_CPC_IC_BASE_LO, addr);
+		WREG32_SOC15(GC, 0, regCP_CPC_IC_BASE_HI,
+		     upper_32_bits(addr));
+	}
+	mutex_unlock(&adev->srbm_mutex);
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+
+	/* Trigger an invalidation of the L1 instruction caches */
+	tmp = RREG32_SOC15(GC, 0, regCP_MEC_DC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_DC_OP_CNTL, INVALIDATE_DCACHE, 1);
+	WREG32_SOC15(GC, 0, regCP_MEC_DC_OP_CNTL, tmp);
+
+	/* Wait for invalidation complete */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_MEC_DC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_MEC_DC_OP_CNTL,
+				       INVALIDATE_DCACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	/* Trigger an invalidation of the L1 instruction caches */
+	tmp = RREG32_SOC15(GC, 0, regCP_CPC_IC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_OP_CNTL, INVALIDATE_CACHE, 1);
+	WREG32_SOC15(GC, 0, regCP_CPC_IC_OP_CNTL, tmp);
+
+	/* Wait for invalidation complete */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_CPC_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_CPC_IC_OP_CNTL,
+				       INVALIDATE_CACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static void gfx_v11_0_config_gfx_rs64(struct amdgpu_device *adev)
+{
+	const struct gfx_firmware_header_v2_0 *pfp_hdr;
+	const struct gfx_firmware_header_v2_0 *me_hdr;
+	const struct gfx_firmware_header_v2_0 *mec_hdr;
+	uint32_t pipe_id, tmp;
+
+	mec_hdr = (const struct gfx_firmware_header_v2_0 *)
+		adev->gfx.mec_fw->data;
+	me_hdr = (const struct gfx_firmware_header_v2_0 *)
+		adev->gfx.me_fw->data;
+	pfp_hdr = (const struct gfx_firmware_header_v2_0 *)
+		adev->gfx.pfp_fw->data;
+
+	/* config pfp program start addr */
+	for (pipe_id = 0; pipe_id < 2; pipe_id++) {
+		soc21_grbm_select(adev, 0, pipe_id, 0, 0);
+		WREG32_SOC15(GC, 0, regCP_PFP_PRGRM_CNTR_START,
+			(pfp_hdr->ucode_start_addr_hi << 30) |
+			(pfp_hdr->ucode_start_addr_lo >> 2));
+		WREG32_SOC15(GC, 0, regCP_PFP_PRGRM_CNTR_START_HI,
+			pfp_hdr->ucode_start_addr_hi >> 2);
+	}
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+
+	/* reset pfp pipe */
+	tmp = RREG32_SOC15(GC, 0, regCP_ME_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_PIPE0_RESET, 1);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_PIPE1_RESET, 1);
+	WREG32_SOC15(GC, 0, regCP_ME_CNTL, tmp);
+
+	/* clear pfp pipe reset */
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_PIPE0_RESET, 0);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_PIPE1_RESET, 0);
+	WREG32_SOC15(GC, 0, regCP_ME_CNTL, tmp);
+
+	/* config me program start addr */
+	for (pipe_id = 0; pipe_id < 2; pipe_id++) {
+		soc21_grbm_select(adev, 0, pipe_id, 0, 0);
+		WREG32_SOC15(GC, 0, regCP_ME_PRGRM_CNTR_START,
+			(me_hdr->ucode_start_addr_hi << 30) |
+			(me_hdr->ucode_start_addr_lo >> 2) );
+		WREG32_SOC15(GC, 0, regCP_ME_PRGRM_CNTR_START_HI,
+			me_hdr->ucode_start_addr_hi>>2);
+	}
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+
+	/* reset me pipe */
+	tmp = RREG32_SOC15(GC, 0, regCP_ME_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_PIPE0_RESET, 1);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_PIPE1_RESET, 1);
+	WREG32_SOC15(GC, 0, regCP_ME_CNTL, tmp);
+
+	/* clear me pipe reset */
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_PIPE0_RESET, 0);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_PIPE1_RESET, 0);
+	WREG32_SOC15(GC, 0, regCP_ME_CNTL, tmp);
+
+	/* config mec program start addr */
+	for (pipe_id = 0; pipe_id < 4; pipe_id++) {
+		soc21_grbm_select(adev, 1, pipe_id, 0, 0);
+		WREG32_SOC15(GC, 0, regCP_MEC_RS64_PRGRM_CNTR_START,
+					mec_hdr->ucode_start_addr_lo >> 2 |
+					mec_hdr->ucode_start_addr_hi << 30);
+		WREG32_SOC15(GC, 0, regCP_MEC_RS64_PRGRM_CNTR_START_HI,
+					mec_hdr->ucode_start_addr_hi >> 2);
+	}
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+
+	/* reset mec pipe */
+	tmp = RREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_RS64_CNTL, MEC_PIPE0_RESET, 1);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_RS64_CNTL, MEC_PIPE1_RESET, 1);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_RS64_CNTL, MEC_PIPE2_RESET, 1);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_RS64_CNTL, MEC_PIPE3_RESET, 1);
+	WREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL, tmp);
+
+	/* clear mec pipe reset */
+	tmp = REG_SET_FIELD(tmp, CP_MEC_RS64_CNTL, MEC_PIPE0_RESET, 0);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_RS64_CNTL, MEC_PIPE1_RESET, 0);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_RS64_CNTL, MEC_PIPE2_RESET, 0);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_RS64_CNTL, MEC_PIPE3_RESET, 0);
+	WREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL, tmp);
+}
+
+static int gfx_v11_0_wait_for_rlc_autoload_complete(struct amdgpu_device *adev)
+{
+	uint32_t cp_status;
+	uint32_t bootload_status;
+	int i, r;
+	uint64_t addr, addr2;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		cp_status = RREG32_SOC15(GC, 0, regCP_STAT);
+
+		if (amdgpu_ip_version(adev, GC_HWIP, 0) ==
+			    IP_VERSION(11, 0, 1) ||
+		    amdgpu_ip_version(adev, GC_HWIP, 0) ==
+			    IP_VERSION(11, 0, 4) ||
+		    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(11, 5, 0) ||
+		    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(11, 5, 1) ||
+		    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(11, 5, 2) ||
+		    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(11, 5, 3))
+			bootload_status = RREG32_SOC15(GC, 0,
+					regRLC_RLCS_BOOTLOAD_STATUS_gc_11_0_1);
+		else
+			bootload_status = RREG32_SOC15(GC, 0, regRLC_RLCS_BOOTLOAD_STATUS);
+
+		if ((cp_status == 0) &&
+		    (REG_GET_FIELD(bootload_status,
+			RLC_RLCS_BOOTLOAD_STATUS, BOOTLOAD_COMPLETE) == 1)) {
+			break;
+		}
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout) {
+		dev_err(adev->dev, "rlc autoload: gc ucode autoload timeout\n");
+		return -ETIMEDOUT;
+	}
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO) {
+		if (adev->gfx.rs64_enable) {
+			addr = adev->gfx.rlc.rlc_autoload_gpu_addr +
+				rlc_autoload_info[SOC21_FIRMWARE_ID_RS64_ME].offset;
+			addr2 = adev->gfx.rlc.rlc_autoload_gpu_addr +
+				rlc_autoload_info[SOC21_FIRMWARE_ID_RS64_ME_P0_STACK].offset;
+			r = gfx_v11_0_config_me_cache_rs64(adev, addr, addr2);
+			if (r)
+				return r;
+			addr = adev->gfx.rlc.rlc_autoload_gpu_addr +
+				rlc_autoload_info[SOC21_FIRMWARE_ID_RS64_PFP].offset;
+			addr2 = adev->gfx.rlc.rlc_autoload_gpu_addr +
+				rlc_autoload_info[SOC21_FIRMWARE_ID_RS64_PFP_P0_STACK].offset;
+			r = gfx_v11_0_config_pfp_cache_rs64(adev, addr, addr2);
+			if (r)
+				return r;
+			addr = adev->gfx.rlc.rlc_autoload_gpu_addr +
+				rlc_autoload_info[SOC21_FIRMWARE_ID_RS64_MEC].offset;
+			addr2 = adev->gfx.rlc.rlc_autoload_gpu_addr +
+				rlc_autoload_info[SOC21_FIRMWARE_ID_RS64_MEC_P0_STACK].offset;
+			r = gfx_v11_0_config_mec_cache_rs64(adev, addr, addr2);
+			if (r)
+				return r;
+		} else {
+			addr = adev->gfx.rlc.rlc_autoload_gpu_addr +
+				rlc_autoload_info[SOC21_FIRMWARE_ID_CP_ME].offset;
+			r = gfx_v11_0_config_me_cache(adev, addr);
+			if (r)
+				return r;
+			addr = adev->gfx.rlc.rlc_autoload_gpu_addr +
+				rlc_autoload_info[SOC21_FIRMWARE_ID_CP_PFP].offset;
+			r = gfx_v11_0_config_pfp_cache(adev, addr);
+			if (r)
+				return r;
+			addr = adev->gfx.rlc.rlc_autoload_gpu_addr +
+				rlc_autoload_info[SOC21_FIRMWARE_ID_CP_MEC].offset;
+			r = gfx_v11_0_config_mec_cache(adev, addr);
+			if (r)
+				return r;
+		}
+	}
+
+	return 0;
+}
+
+static int gfx_v11_0_cp_gfx_enable(struct amdgpu_device *adev, bool enable)
+{
+	int i;
+	u32 tmp = RREG32_SOC15(GC, 0, regCP_ME_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_HALT, enable ? 0 : 1);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_HALT, enable ? 0 : 1);
+	WREG32_SOC15(GC, 0, regCP_ME_CNTL, tmp);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (RREG32_SOC15(GC, 0, regCP_STAT) == 0)
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout)
+		DRM_ERROR("failed to %s cp gfx\n", enable ? "unhalt" : "halt");
+
+	return 0;
+}
+
+static int gfx_v11_0_cp_gfx_load_pfp_microcode(struct amdgpu_device *adev)
+{
+	int r;
+	const struct gfx_firmware_header_v1_0 *pfp_hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+
+	pfp_hdr = (const struct gfx_firmware_header_v1_0 *)
+		adev->gfx.pfp_fw->data;
+
+	amdgpu_ucode_print_gfx_hdr(&pfp_hdr->header);
+
+	fw_data = (const __le32 *)(adev->gfx.pfp_fw->data +
+		le32_to_cpu(pfp_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(pfp_hdr->header.ucode_size_bytes);
+
+	r = amdgpu_bo_create_reserved(adev, pfp_hdr->header.ucode_size_bytes,
+				      PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.pfp.pfp_fw_obj,
+				      &adev->gfx.pfp.pfp_fw_gpu_addr,
+				      (void **)&adev->gfx.pfp.pfp_fw_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create pfp fw bo\n", r);
+		gfx_v11_0_pfp_fini(adev);
+		return r;
+	}
+
+	memcpy(adev->gfx.pfp.pfp_fw_ptr, fw_data, fw_size);
+
+	amdgpu_bo_kunmap(adev->gfx.pfp.pfp_fw_obj);
+	amdgpu_bo_unreserve(adev->gfx.pfp.pfp_fw_obj);
+
+	gfx_v11_0_config_pfp_cache(adev, adev->gfx.pfp.pfp_fw_gpu_addr);
+
+	WREG32_SOC15(GC, 0, regCP_HYP_PFP_UCODE_ADDR, 0);
+
+	for (i = 0; i < pfp_hdr->jt_size; i++)
+		WREG32_SOC15(GC, 0, regCP_HYP_PFP_UCODE_DATA,
+			     le32_to_cpup(fw_data + pfp_hdr->jt_offset + i));
+
+	WREG32_SOC15(GC, 0, regCP_HYP_PFP_UCODE_ADDR, adev->gfx.pfp_fw_version);
+
+	return 0;
+}
+
+static int gfx_v11_0_cp_gfx_load_pfp_microcode_rs64(struct amdgpu_device *adev)
+{
+	int r;
+	const struct gfx_firmware_header_v2_0 *pfp_hdr;
+	const __le32 *fw_ucode, *fw_data;
+	unsigned i, pipe_id, fw_ucode_size, fw_data_size;
+	uint32_t tmp;
+	uint32_t usec_timeout = 50000;  /* wait for 50ms */
+
+	pfp_hdr = (const struct gfx_firmware_header_v2_0 *)
+		adev->gfx.pfp_fw->data;
+
+	amdgpu_ucode_print_gfx_hdr(&pfp_hdr->header);
+
+	/* instruction */
+	fw_ucode = (const __le32 *)(adev->gfx.pfp_fw->data +
+		le32_to_cpu(pfp_hdr->ucode_offset_bytes));
+	fw_ucode_size = le32_to_cpu(pfp_hdr->ucode_size_bytes);
+	/* data */
+	fw_data = (const __le32 *)(adev->gfx.pfp_fw->data +
+		le32_to_cpu(pfp_hdr->data_offset_bytes));
+	fw_data_size = le32_to_cpu(pfp_hdr->data_size_bytes);
+
+	/* 64kb align */
+	r = amdgpu_bo_create_reserved(adev, fw_ucode_size,
+				      64 * 1024,
+				      AMDGPU_GEM_DOMAIN_VRAM |
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.pfp.pfp_fw_obj,
+				      &adev->gfx.pfp.pfp_fw_gpu_addr,
+				      (void **)&adev->gfx.pfp.pfp_fw_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create pfp ucode fw bo\n", r);
+		gfx_v11_0_pfp_fini(adev);
+		return r;
+	}
+
+	r = amdgpu_bo_create_reserved(adev, fw_data_size,
+				      64 * 1024,
+				      AMDGPU_GEM_DOMAIN_VRAM |
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.pfp.pfp_fw_data_obj,
+				      &adev->gfx.pfp.pfp_fw_data_gpu_addr,
+				      (void **)&adev->gfx.pfp.pfp_fw_data_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create pfp data fw bo\n", r);
+		gfx_v11_0_pfp_fini(adev);
+		return r;
+	}
+
+	memcpy(adev->gfx.pfp.pfp_fw_ptr, fw_ucode, fw_ucode_size);
+	memcpy(adev->gfx.pfp.pfp_fw_data_ptr, fw_data, fw_data_size);
+
+	amdgpu_bo_kunmap(adev->gfx.pfp.pfp_fw_obj);
+	amdgpu_bo_kunmap(adev->gfx.pfp.pfp_fw_data_obj);
+	amdgpu_bo_unreserve(adev->gfx.pfp.pfp_fw_obj);
+	amdgpu_bo_unreserve(adev->gfx.pfp.pfp_fw_data_obj);
+
+	if (amdgpu_emu_mode == 1)
+		amdgpu_device_flush_hdp(adev, NULL);
+
+	WREG32_SOC15(GC, 0, regCP_PFP_IC_BASE_LO,
+		lower_32_bits(adev->gfx.pfp.pfp_fw_gpu_addr));
+	WREG32_SOC15(GC, 0, regCP_PFP_IC_BASE_HI,
+		upper_32_bits(adev->gfx.pfp.pfp_fw_gpu_addr));
+
+	tmp = RREG32_SOC15(GC, 0, regCP_PFP_IC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_BASE_CNTL, CACHE_POLICY, 0);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_BASE_CNTL, EXE_DISABLE, 0);
+	WREG32_SOC15(GC, 0, regCP_PFP_IC_BASE_CNTL, tmp);
+
+	/*
+	 * Programming any of the CP_PFP_IC_BASE registers
+	 * forces invalidation of the ME L1 I$. Wait for the
+	 * invalidation complete
+	 */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_PFP_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_PFP_IC_OP_CNTL,
+			INVALIDATE_CACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	/* Prime the L1 instruction caches */
+	tmp = RREG32_SOC15(GC, 0, regCP_PFP_IC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_OP_CNTL, PRIME_ICACHE, 1);
+	WREG32_SOC15(GC, 0, regCP_PFP_IC_OP_CNTL, tmp);
+	/* Waiting for cache primed*/
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_PFP_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_PFP_IC_OP_CNTL,
+			ICACHE_PRIMED))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to prime instruction cache\n");
+		return -EINVAL;
+	}
+
+	mutex_lock(&adev->srbm_mutex);
+	for (pipe_id = 0; pipe_id < adev->gfx.me.num_pipe_per_me; pipe_id++) {
+		soc21_grbm_select(adev, 0, pipe_id, 0, 0);
+		WREG32_SOC15(GC, 0, regCP_PFP_PRGRM_CNTR_START,
+			(pfp_hdr->ucode_start_addr_hi << 30) |
+			(pfp_hdr->ucode_start_addr_lo >> 2) );
+		WREG32_SOC15(GC, 0, regCP_PFP_PRGRM_CNTR_START_HI,
+			pfp_hdr->ucode_start_addr_hi>>2);
+
+		/*
+		 * Program CP_ME_CNTL to reset given PIPE to take
+		 * effect of CP_PFP_PRGRM_CNTR_START.
+		 */
+		tmp = RREG32_SOC15(GC, 0, regCP_ME_CNTL);
+		if (pipe_id == 0)
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					PFP_PIPE0_RESET, 1);
+		else
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					PFP_PIPE1_RESET, 1);
+		WREG32_SOC15(GC, 0, regCP_ME_CNTL, tmp);
+
+		/* Clear pfp pipe0 reset bit. */
+		if (pipe_id == 0)
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					PFP_PIPE0_RESET, 0);
+		else
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					PFP_PIPE1_RESET, 0);
+		WREG32_SOC15(GC, 0, regCP_ME_CNTL, tmp);
+
+		WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE0_LO,
+			lower_32_bits(adev->gfx.pfp.pfp_fw_data_gpu_addr));
+		WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE0_HI,
+			upper_32_bits(adev->gfx.pfp.pfp_fw_data_gpu_addr));
+	}
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	tmp = RREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_RS64_DC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_RS64_DC_BASE_CNTL, CACHE_POLICY, 0);
+	WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE_CNTL, tmp);
+
+	/* Invalidate the data caches */
+	tmp = RREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_RS64_DC_OP_CNTL, INVALIDATE_DCACHE, 1);
+	WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_OP_CNTL, tmp);
+
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_GFX_RS64_DC_OP_CNTL,
+			INVALIDATE_DCACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate RS64 data cache\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int gfx_v11_0_cp_gfx_load_me_microcode(struct amdgpu_device *adev)
+{
+	int r;
+	const struct gfx_firmware_header_v1_0 *me_hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+
+	me_hdr = (const struct gfx_firmware_header_v1_0 *)
+		adev->gfx.me_fw->data;
+
+	amdgpu_ucode_print_gfx_hdr(&me_hdr->header);
+
+	fw_data = (const __le32 *)(adev->gfx.me_fw->data +
+		le32_to_cpu(me_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(me_hdr->header.ucode_size_bytes);
+
+	r = amdgpu_bo_create_reserved(adev, me_hdr->header.ucode_size_bytes,
+				      PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.me.me_fw_obj,
+				      &adev->gfx.me.me_fw_gpu_addr,
+				      (void **)&adev->gfx.me.me_fw_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create me fw bo\n", r);
+		gfx_v11_0_me_fini(adev);
+		return r;
+	}
+
+	memcpy(adev->gfx.me.me_fw_ptr, fw_data, fw_size);
+
+	amdgpu_bo_kunmap(adev->gfx.me.me_fw_obj);
+	amdgpu_bo_unreserve(adev->gfx.me.me_fw_obj);
+
+	gfx_v11_0_config_me_cache(adev, adev->gfx.me.me_fw_gpu_addr);
+
+	WREG32_SOC15(GC, 0, regCP_HYP_ME_UCODE_ADDR, 0);
+
+	for (i = 0; i < me_hdr->jt_size; i++)
+		WREG32_SOC15(GC, 0, regCP_HYP_ME_UCODE_DATA,
+			     le32_to_cpup(fw_data + me_hdr->jt_offset + i));
+
+	WREG32_SOC15(GC, 0, regCP_HYP_ME_UCODE_ADDR, adev->gfx.me_fw_version);
+
+	return 0;
+}
+
+static int gfx_v11_0_cp_gfx_load_me_microcode_rs64(struct amdgpu_device *adev)
+{
+	int r;
+	const struct gfx_firmware_header_v2_0 *me_hdr;
+	const __le32 *fw_ucode, *fw_data;
+	unsigned i, pipe_id, fw_ucode_size, fw_data_size;
+	uint32_t tmp;
+	uint32_t usec_timeout = 50000;  /* wait for 50ms */
+
+	me_hdr = (const struct gfx_firmware_header_v2_0 *)
+		adev->gfx.me_fw->data;
+
+	amdgpu_ucode_print_gfx_hdr(&me_hdr->header);
+
+	/* instruction */
+	fw_ucode = (const __le32 *)(adev->gfx.me_fw->data +
+		le32_to_cpu(me_hdr->ucode_offset_bytes));
+	fw_ucode_size = le32_to_cpu(me_hdr->ucode_size_bytes);
+	/* data */
+	fw_data = (const __le32 *)(adev->gfx.me_fw->data +
+		le32_to_cpu(me_hdr->data_offset_bytes));
+	fw_data_size = le32_to_cpu(me_hdr->data_size_bytes);
+
+	/* 64kb align*/
+	r = amdgpu_bo_create_reserved(adev, fw_ucode_size,
+				      64 * 1024,
+				      AMDGPU_GEM_DOMAIN_VRAM |
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.me.me_fw_obj,
+				      &adev->gfx.me.me_fw_gpu_addr,
+				      (void **)&adev->gfx.me.me_fw_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create me ucode bo\n", r);
+		gfx_v11_0_me_fini(adev);
+		return r;
+	}
+
+	r = amdgpu_bo_create_reserved(adev, fw_data_size,
+				      64 * 1024,
+				      AMDGPU_GEM_DOMAIN_VRAM |
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.me.me_fw_data_obj,
+				      &adev->gfx.me.me_fw_data_gpu_addr,
+				      (void **)&adev->gfx.me.me_fw_data_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create me data bo\n", r);
+		gfx_v11_0_pfp_fini(adev);
+		return r;
+	}
+
+	memcpy(adev->gfx.me.me_fw_ptr, fw_ucode, fw_ucode_size);
+	memcpy(adev->gfx.me.me_fw_data_ptr, fw_data, fw_data_size);
+
+	amdgpu_bo_kunmap(adev->gfx.me.me_fw_obj);
+	amdgpu_bo_kunmap(adev->gfx.me.me_fw_data_obj);
+	amdgpu_bo_unreserve(adev->gfx.me.me_fw_obj);
+	amdgpu_bo_unreserve(adev->gfx.me.me_fw_data_obj);
+
+	if (amdgpu_emu_mode == 1)
+		amdgpu_device_flush_hdp(adev, NULL);
+
+	WREG32_SOC15(GC, 0, regCP_ME_IC_BASE_LO,
+		lower_32_bits(adev->gfx.me.me_fw_gpu_addr));
+	WREG32_SOC15(GC, 0, regCP_ME_IC_BASE_HI,
+		upper_32_bits(adev->gfx.me.me_fw_gpu_addr));
+
+	tmp = RREG32_SOC15(GC, 0, regCP_ME_IC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_BASE_CNTL, CACHE_POLICY, 0);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_BASE_CNTL, EXE_DISABLE, 0);
+	WREG32_SOC15(GC, 0, regCP_ME_IC_BASE_CNTL, tmp);
+
+	/*
+	 * Programming any of the CP_ME_IC_BASE registers
+	 * forces invalidation of the ME L1 I$. Wait for the
+	 * invalidation complete
+	 */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_ME_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_ME_IC_OP_CNTL,
+			INVALIDATE_CACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	/* Prime the instruction caches */
+	tmp = RREG32_SOC15(GC, 0, regCP_ME_IC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_OP_CNTL, PRIME_ICACHE, 1);
+	WREG32_SOC15(GC, 0, regCP_ME_IC_OP_CNTL, tmp);
+
+	/* Waiting for instruction cache primed*/
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_ME_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_ME_IC_OP_CNTL,
+			ICACHE_PRIMED))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to prime instruction cache\n");
+		return -EINVAL;
+	}
+
+	mutex_lock(&adev->srbm_mutex);
+	for (pipe_id = 0; pipe_id < adev->gfx.me.num_pipe_per_me; pipe_id++) {
+		soc21_grbm_select(adev, 0, pipe_id, 0, 0);
+		WREG32_SOC15(GC, 0, regCP_ME_PRGRM_CNTR_START,
+			(me_hdr->ucode_start_addr_hi << 30) |
+			(me_hdr->ucode_start_addr_lo >> 2) );
+		WREG32_SOC15(GC, 0, regCP_ME_PRGRM_CNTR_START_HI,
+			me_hdr->ucode_start_addr_hi>>2);
+
+		/*
+		 * Program CP_ME_CNTL to reset given PIPE to take
+		 * effect of CP_PFP_PRGRM_CNTR_START.
+		 */
+		tmp = RREG32_SOC15(GC, 0, regCP_ME_CNTL);
+		if (pipe_id == 0)
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					ME_PIPE0_RESET, 1);
+		else
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					ME_PIPE1_RESET, 1);
+		WREG32_SOC15(GC, 0, regCP_ME_CNTL, tmp);
+
+		/* Clear pfp pipe0 reset bit. */
+		if (pipe_id == 0)
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					ME_PIPE0_RESET, 0);
+		else
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					ME_PIPE1_RESET, 0);
+		WREG32_SOC15(GC, 0, regCP_ME_CNTL, tmp);
+
+		WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE1_LO,
+			lower_32_bits(adev->gfx.me.me_fw_data_gpu_addr));
+		WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE1_HI,
+			upper_32_bits(adev->gfx.me.me_fw_data_gpu_addr));
+	}
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	tmp = RREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_RS64_DC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_RS64_DC_BASE_CNTL, CACHE_POLICY, 0);
+	WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE_CNTL, tmp);
+
+	/* Invalidate the data caches */
+	tmp = RREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_RS64_DC_OP_CNTL, INVALIDATE_DCACHE, 1);
+	WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_OP_CNTL, tmp);
+
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_GFX_RS64_DC_OP_CNTL,
+			INVALIDATE_DCACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate RS64 data cache\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int gfx_v11_0_cp_gfx_load_microcode(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (!adev->gfx.me_fw || !adev->gfx.pfp_fw)
+		return -EINVAL;
+
+	gfx_v11_0_cp_gfx_enable(adev, false);
+
+	if (adev->gfx.rs64_enable)
+		r = gfx_v11_0_cp_gfx_load_pfp_microcode_rs64(adev);
+	else
+		r = gfx_v11_0_cp_gfx_load_pfp_microcode(adev);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to load pfp fw\n", r);
+		return r;
+	}
+
+	if (adev->gfx.rs64_enable)
+		r = gfx_v11_0_cp_gfx_load_me_microcode_rs64(adev);
+	else
+		r = gfx_v11_0_cp_gfx_load_me_microcode(adev);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to load me fw\n", r);
+		return r;
+	}
+
+	return 0;
+}
+
+static int gfx_v11_0_cp_gfx_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+	int r, i;
+	int ctx_reg_offset;
+
+	/* init the CP */
+	WREG32_SOC15(GC, 0, regCP_MAX_CONTEXT,
+		     adev->gfx.config.max_hw_contexts - 1);
+	WREG32_SOC15(GC, 0, regCP_DEVICE_ID, 1);
+
+	if (!amdgpu_async_gfx_ring)
+		gfx_v11_0_cp_gfx_enable(adev, true);
+
+	ring = &adev->gfx.gfx_ring[0];
+	r = amdgpu_ring_alloc(ring, gfx_v11_0_get_csb_size(adev));
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring (%d).\n", r);
+		return r;
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	amdgpu_ring_write(ring, PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	amdgpu_ring_write(ring, 0x80000000);
+	amdgpu_ring_write(ring, 0x80000000);
+
+	for (sect = gfx11_cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT) {
+				amdgpu_ring_write(ring,
+						  PACKET3(PACKET3_SET_CONTEXT_REG,
+							  ext->reg_count));
+				amdgpu_ring_write(ring, ext->reg_index -
+						  PACKET3_SET_CONTEXT_REG_START);
+				for (i = 0; i < ext->reg_count; i++)
+					amdgpu_ring_write(ring, ext->extent[i]);
+			}
+		}
+	}
+
+	ctx_reg_offset =
+		SOC15_REG_OFFSET(GC, 0, regPA_SC_TILE_STEERING_OVERRIDE) - PACKET3_SET_CONTEXT_REG_START;
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONTEXT_REG, 1));
+	amdgpu_ring_write(ring, ctx_reg_offset);
+	amdgpu_ring_write(ring, adev->gfx.config.pa_sc_tile_steering_override);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	amdgpu_ring_write(ring, PACKET3_PREAMBLE_END_CLEAR_STATE);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CLEAR_STATE, 0));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_commit(ring);
+
+	/* submit cs packet to copy state 0 to next available state */
+	if (adev->gfx.num_gfx_rings > 1) {
+		/* maximum supported gfx ring is 2 */
+		ring = &adev->gfx.gfx_ring[1];
+		r = amdgpu_ring_alloc(ring, 2);
+		if (r) {
+			DRM_ERROR("amdgpu: cp failed to lock ring (%d).\n", r);
+			return r;
+		}
+
+		amdgpu_ring_write(ring, PACKET3(PACKET3_CLEAR_STATE, 0));
+		amdgpu_ring_write(ring, 0);
+
+		amdgpu_ring_commit(ring);
+	}
+	return 0;
+}
+
+static void gfx_v11_0_cp_gfx_switch_pipe(struct amdgpu_device *adev,
+					 CP_PIPE_ID pipe)
+{
+	u32 tmp;
+
+	tmp = RREG32_SOC15(GC, 0, regGRBM_GFX_CNTL);
+	tmp = REG_SET_FIELD(tmp, GRBM_GFX_CNTL, PIPEID, pipe);
+
+	WREG32_SOC15(GC, 0, regGRBM_GFX_CNTL, tmp);
+}
+
+static void gfx_v11_0_cp_gfx_set_doorbell(struct amdgpu_device *adev,
+					  struct amdgpu_ring *ring)
+{
+	u32 tmp;
+
+	tmp = RREG32_SOC15(GC, 0, regCP_RB_DOORBELL_CONTROL);
+	if (ring->use_doorbell) {
+		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+				    DOORBELL_OFFSET, ring->doorbell_index);
+		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+				    DOORBELL_EN, 1);
+	} else {
+		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+				    DOORBELL_EN, 0);
+	}
+	WREG32_SOC15(GC, 0, regCP_RB_DOORBELL_CONTROL, tmp);
+
+	tmp = REG_SET_FIELD(0, CP_RB_DOORBELL_RANGE_LOWER,
+			    DOORBELL_RANGE_LOWER, ring->doorbell_index);
+	WREG32_SOC15(GC, 0, regCP_RB_DOORBELL_RANGE_LOWER, tmp);
+
+	WREG32_SOC15(GC, 0, regCP_RB_DOORBELL_RANGE_UPPER,
+		     CP_RB_DOORBELL_RANGE_UPPER__DOORBELL_RANGE_UPPER_MASK);
+}
+
+static int gfx_v11_0_cp_gfx_resume(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	u32 tmp;
+	u32 rb_bufsz;
+	u64 rb_addr, rptr_addr, wptr_gpu_addr;
+
+	/* Set the write pointer delay */
+	WREG32_SOC15(GC, 0, regCP_RB_WPTR_DELAY, 0);
+
+	/* set the RB to use vmid 0 */
+	WREG32_SOC15(GC, 0, regCP_RB_VMID, 0);
+
+	/* Init gfx ring 0 for pipe 0 */
+	mutex_lock(&adev->srbm_mutex);
+	gfx_v11_0_cp_gfx_switch_pipe(adev, PIPE_ID0);
+
+	/* Set ring buffer size */
+	ring = &adev->gfx.gfx_ring[0];
+	rb_bufsz = order_base_2(ring->ring_size / 8);
+	tmp = REG_SET_FIELD(0, CP_RB0_CNTL, RB_BUFSZ, rb_bufsz);
+	tmp = REG_SET_FIELD(tmp, CP_RB0_CNTL, RB_BLKSZ, rb_bufsz - 2);
+	WREG32_SOC15(GC, 0, regCP_RB0_CNTL, tmp);
+
+	/* Initialize the ring buffer's write pointers */
+	ring->wptr = 0;
+	WREG32_SOC15(GC, 0, regCP_RB0_WPTR, lower_32_bits(ring->wptr));
+	WREG32_SOC15(GC, 0, regCP_RB0_WPTR_HI, upper_32_bits(ring->wptr));
+
+	/* set the wb address whether it's enabled or not */
+	rptr_addr = ring->rptr_gpu_addr;
+	WREG32_SOC15(GC, 0, regCP_RB0_RPTR_ADDR, lower_32_bits(rptr_addr));
+	WREG32_SOC15(GC, 0, regCP_RB0_RPTR_ADDR_HI, upper_32_bits(rptr_addr) &
+		     CP_RB_RPTR_ADDR_HI__RB_RPTR_ADDR_HI_MASK);
+
+	wptr_gpu_addr = ring->wptr_gpu_addr;
+	WREG32_SOC15(GC, 0, regCP_RB_WPTR_POLL_ADDR_LO,
+		     lower_32_bits(wptr_gpu_addr));
+	WREG32_SOC15(GC, 0, regCP_RB_WPTR_POLL_ADDR_HI,
+		     upper_32_bits(wptr_gpu_addr));
+
+	mdelay(1);
+	WREG32_SOC15(GC, 0, regCP_RB0_CNTL, tmp);
+
+	rb_addr = ring->gpu_addr >> 8;
+	WREG32_SOC15(GC, 0, regCP_RB0_BASE, rb_addr);
+	WREG32_SOC15(GC, 0, regCP_RB0_BASE_HI, upper_32_bits(rb_addr));
+
+	WREG32_SOC15(GC, 0, regCP_RB_ACTIVE, 1);
+
+	gfx_v11_0_cp_gfx_set_doorbell(adev, ring);
+	mutex_unlock(&adev->srbm_mutex);
+
+	/* Init gfx ring 1 for pipe 1 */
+	if (adev->gfx.num_gfx_rings > 1) {
+		mutex_lock(&adev->srbm_mutex);
+		gfx_v11_0_cp_gfx_switch_pipe(adev, PIPE_ID1);
+		/* maximum supported gfx ring is 2 */
+		ring = &adev->gfx.gfx_ring[1];
+		rb_bufsz = order_base_2(ring->ring_size / 8);
+		tmp = REG_SET_FIELD(0, CP_RB1_CNTL, RB_BUFSZ, rb_bufsz);
+		tmp = REG_SET_FIELD(tmp, CP_RB1_CNTL, RB_BLKSZ, rb_bufsz - 2);
+		WREG32_SOC15(GC, 0, regCP_RB1_CNTL, tmp);
+		/* Initialize the ring buffer's write pointers */
+		ring->wptr = 0;
+		WREG32_SOC15(GC, 0, regCP_RB1_WPTR, lower_32_bits(ring->wptr));
+		WREG32_SOC15(GC, 0, regCP_RB1_WPTR_HI, upper_32_bits(ring->wptr));
+		/* Set the wb address whether it's enabled or not */
+		rptr_addr = ring->rptr_gpu_addr;
+		WREG32_SOC15(GC, 0, regCP_RB1_RPTR_ADDR, lower_32_bits(rptr_addr));
+		WREG32_SOC15(GC, 0, regCP_RB1_RPTR_ADDR_HI, upper_32_bits(rptr_addr) &
+			     CP_RB1_RPTR_ADDR_HI__RB_RPTR_ADDR_HI_MASK);
+		wptr_gpu_addr = ring->wptr_gpu_addr;
+		WREG32_SOC15(GC, 0, regCP_RB_WPTR_POLL_ADDR_LO,
+			     lower_32_bits(wptr_gpu_addr));
+		WREG32_SOC15(GC, 0, regCP_RB_WPTR_POLL_ADDR_HI,
+			     upper_32_bits(wptr_gpu_addr));
+
+		mdelay(1);
+		WREG32_SOC15(GC, 0, regCP_RB1_CNTL, tmp);
+
+		rb_addr = ring->gpu_addr >> 8;
+		WREG32_SOC15(GC, 0, regCP_RB1_BASE, rb_addr);
+		WREG32_SOC15(GC, 0, regCP_RB1_BASE_HI, upper_32_bits(rb_addr));
+		WREG32_SOC15(GC, 0, regCP_RB1_ACTIVE, 1);
+
+		gfx_v11_0_cp_gfx_set_doorbell(adev, ring);
+		mutex_unlock(&adev->srbm_mutex);
+	}
+	/* Switch to pipe 0 */
+	mutex_lock(&adev->srbm_mutex);
+	gfx_v11_0_cp_gfx_switch_pipe(adev, PIPE_ID0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	/* start the ring */
+	gfx_v11_0_cp_gfx_start(adev);
+
+	return 0;
+}
+
+static void gfx_v11_0_cp_compute_enable(struct amdgpu_device *adev, bool enable)
+{
+	u32 data;
+
+	if (adev->gfx.rs64_enable) {
+		data = RREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL);
+		data = REG_SET_FIELD(data, CP_MEC_RS64_CNTL, MEC_INVALIDATE_ICACHE,
+							 enable ? 0 : 1);
+		data = REG_SET_FIELD(data, CP_MEC_RS64_CNTL, MEC_PIPE0_RESET,
+							 enable ? 0 : 1);
+		data = REG_SET_FIELD(data, CP_MEC_RS64_CNTL, MEC_PIPE1_RESET,
+							 enable ? 0 : 1);
+		data = REG_SET_FIELD(data, CP_MEC_RS64_CNTL, MEC_PIPE2_RESET,
+							 enable ? 0 : 1);
+		data = REG_SET_FIELD(data, CP_MEC_RS64_CNTL, MEC_PIPE3_RESET,
+							 enable ? 0 : 1);
+		data = REG_SET_FIELD(data, CP_MEC_RS64_CNTL, MEC_PIPE0_ACTIVE,
+							 enable ? 1 : 0);
+		data = REG_SET_FIELD(data, CP_MEC_RS64_CNTL, MEC_PIPE1_ACTIVE,
+				                         enable ? 1 : 0);
+		data = REG_SET_FIELD(data, CP_MEC_RS64_CNTL, MEC_PIPE2_ACTIVE,
+							 enable ? 1 : 0);
+		data = REG_SET_FIELD(data, CP_MEC_RS64_CNTL, MEC_PIPE3_ACTIVE,
+							 enable ? 1 : 0);
+		data = REG_SET_FIELD(data, CP_MEC_RS64_CNTL, MEC_HALT,
+							 enable ? 0 : 1);
+		WREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL, data);
+	} else {
+		data = RREG32_SOC15(GC, 0, regCP_MEC_CNTL);
+
+		if (enable) {
+			data = REG_SET_FIELD(data, CP_MEC_CNTL, MEC_ME1_HALT, 0);
+			if (!adev->enable_mes_kiq)
+				data = REG_SET_FIELD(data, CP_MEC_CNTL,
+						     MEC_ME2_HALT, 0);
+		} else {
+			data = REG_SET_FIELD(data, CP_MEC_CNTL, MEC_ME1_HALT, 1);
+			data = REG_SET_FIELD(data, CP_MEC_CNTL, MEC_ME2_HALT, 1);
+		}
+		WREG32_SOC15(GC, 0, regCP_MEC_CNTL, data);
+	}
+
+	udelay(50);
+}
+
+static int gfx_v11_0_cp_compute_load_microcode(struct amdgpu_device *adev)
+{
+	const struct gfx_firmware_header_v1_0 *mec_hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+	u32 *fw = NULL;
+	int r;
+
+	if (!adev->gfx.mec_fw)
+		return -EINVAL;
+
+	gfx_v11_0_cp_compute_enable(adev, false);
+
+	mec_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+	amdgpu_ucode_print_gfx_hdr(&mec_hdr->header);
+
+	fw_data = (const __le32 *)
+		(adev->gfx.mec_fw->data +
+		 le32_to_cpu(mec_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(mec_hdr->header.ucode_size_bytes);
+
+	r = amdgpu_bo_create_reserved(adev, mec_hdr->header.ucode_size_bytes,
+					  PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
+					  &adev->gfx.mec.mec_fw_obj,
+					  &adev->gfx.mec.mec_fw_gpu_addr,
+					  (void **)&fw);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create mec fw bo\n", r);
+		gfx_v11_0_mec_fini(adev);
+		return r;
+	}
+
+	memcpy(fw, fw_data, fw_size);
+	
+	amdgpu_bo_kunmap(adev->gfx.mec.mec_fw_obj);
+	amdgpu_bo_unreserve(adev->gfx.mec.mec_fw_obj);
+
+	gfx_v11_0_config_mec_cache(adev, adev->gfx.mec.mec_fw_gpu_addr);
+
+	/* MEC1 */
+	WREG32_SOC15(GC, 0, regCP_MEC_ME1_UCODE_ADDR, 0);
+
+	for (i = 0; i < mec_hdr->jt_size; i++)
+		WREG32_SOC15(GC, 0, regCP_MEC_ME1_UCODE_DATA,
+			     le32_to_cpup(fw_data + mec_hdr->jt_offset + i));
+
+	WREG32_SOC15(GC, 0, regCP_MEC_ME1_UCODE_ADDR, adev->gfx.mec_fw_version);
+
+	return 0;
+}
+
+static int gfx_v11_0_cp_compute_load_microcode_rs64(struct amdgpu_device *adev)
+{
+	const struct gfx_firmware_header_v2_0 *mec_hdr;
+	const __le32 *fw_ucode, *fw_data;
+	u32 tmp, fw_ucode_size, fw_data_size;
+	u32 i, usec_timeout = 50000; /* Wait for 50 ms */
+	u32 *fw_ucode_ptr, *fw_data_ptr;
+	int r;
+
+	if (!adev->gfx.mec_fw)
+		return -EINVAL;
+
+	gfx_v11_0_cp_compute_enable(adev, false);
+
+	mec_hdr = (const struct gfx_firmware_header_v2_0 *)adev->gfx.mec_fw->data;
+	amdgpu_ucode_print_gfx_hdr(&mec_hdr->header);
+
+	fw_ucode = (const __le32 *) (adev->gfx.mec_fw->data +
+				le32_to_cpu(mec_hdr->ucode_offset_bytes));
+	fw_ucode_size = le32_to_cpu(mec_hdr->ucode_size_bytes);
+
+	fw_data = (const __le32 *) (adev->gfx.mec_fw->data +
+				le32_to_cpu(mec_hdr->data_offset_bytes));
+	fw_data_size = le32_to_cpu(mec_hdr->data_size_bytes);
+
+	r = amdgpu_bo_create_reserved(adev, fw_ucode_size,
+				      64 * 1024,
+				      AMDGPU_GEM_DOMAIN_VRAM |
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.mec.mec_fw_obj,
+				      &adev->gfx.mec.mec_fw_gpu_addr,
+				      (void **)&fw_ucode_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create mec fw ucode bo\n", r);
+		gfx_v11_0_mec_fini(adev);
+		return r;
+	}
+
+	r = amdgpu_bo_create_reserved(adev, fw_data_size,
+				      64 * 1024,
+				      AMDGPU_GEM_DOMAIN_VRAM |
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.mec.mec_fw_data_obj,
+				      &adev->gfx.mec.mec_fw_data_gpu_addr,
+				      (void **)&fw_data_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create mec fw ucode bo\n", r);
+		gfx_v11_0_mec_fini(adev);
+		return r;
+	}
+
+	memcpy(fw_ucode_ptr, fw_ucode, fw_ucode_size);
+	memcpy(fw_data_ptr, fw_data, fw_data_size);
+
+	amdgpu_bo_kunmap(adev->gfx.mec.mec_fw_obj);
+	amdgpu_bo_kunmap(adev->gfx.mec.mec_fw_data_obj);
+	amdgpu_bo_unreserve(adev->gfx.mec.mec_fw_obj);
+	amdgpu_bo_unreserve(adev->gfx.mec.mec_fw_data_obj);
+
+	tmp = RREG32_SOC15(GC, 0, regCP_CPC_IC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, EXE_DISABLE, 0);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, CACHE_POLICY, 0);
+	WREG32_SOC15(GC, 0, regCP_CPC_IC_BASE_CNTL, tmp);
+
+	tmp = RREG32_SOC15(GC, 0, regCP_MEC_DC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_DC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_DC_BASE_CNTL, CACHE_POLICY, 0);
+	WREG32_SOC15(GC, 0, regCP_MEC_DC_BASE_CNTL, tmp);
+
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < adev->gfx.mec.num_pipe_per_mec; i++) {
+		soc21_grbm_select(adev, 1, i, 0, 0);
+
+		WREG32_SOC15(GC, 0, regCP_MEC_MDBASE_LO, adev->gfx.mec.mec_fw_data_gpu_addr);
+		WREG32_SOC15(GC, 0, regCP_MEC_MDBASE_HI,
+		     upper_32_bits(adev->gfx.mec.mec_fw_data_gpu_addr));
+
+		WREG32_SOC15(GC, 0, regCP_MEC_RS64_PRGRM_CNTR_START,
+					mec_hdr->ucode_start_addr_lo >> 2 |
+					mec_hdr->ucode_start_addr_hi << 30);
+		WREG32_SOC15(GC, 0, regCP_MEC_RS64_PRGRM_CNTR_START_HI,
+					mec_hdr->ucode_start_addr_hi >> 2);
+
+		WREG32_SOC15(GC, 0, regCP_CPC_IC_BASE_LO, adev->gfx.mec.mec_fw_gpu_addr);
+		WREG32_SOC15(GC, 0, regCP_CPC_IC_BASE_HI,
+		     upper_32_bits(adev->gfx.mec.mec_fw_gpu_addr));
+	}
+	mutex_unlock(&adev->srbm_mutex);
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+
+	/* Trigger an invalidation of the L1 instruction caches */
+	tmp = RREG32_SOC15(GC, 0, regCP_MEC_DC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_DC_OP_CNTL, INVALIDATE_DCACHE, 1);
+	WREG32_SOC15(GC, 0, regCP_MEC_DC_OP_CNTL, tmp);
+
+	/* Wait for invalidation complete */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_MEC_DC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_MEC_DC_OP_CNTL,
+				       INVALIDATE_DCACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	/* Trigger an invalidation of the L1 instruction caches */
+	tmp = RREG32_SOC15(GC, 0, regCP_CPC_IC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_OP_CNTL, INVALIDATE_CACHE, 1);
+	WREG32_SOC15(GC, 0, regCP_CPC_IC_OP_CNTL, tmp);
+
+	/* Wait for invalidation complete */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_CPC_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_CPC_IC_OP_CNTL,
+				       INVALIDATE_CACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static void gfx_v11_0_kiq_setting(struct amdgpu_ring *ring)
+{
+	uint32_t tmp;
+	struct amdgpu_device *adev = ring->adev;
+
+	/* tell RLC which is KIQ queue */
+	tmp = RREG32_SOC15(GC, 0, regRLC_CP_SCHEDULERS);
+	tmp &= 0xffffff00;
+	tmp |= (ring->me << 5) | (ring->pipe << 3) | (ring->queue);
+	WREG32_SOC15(GC, 0, regRLC_CP_SCHEDULERS, tmp | 0x80);
+}
+
+static void gfx_v11_0_cp_set_doorbell_range(struct amdgpu_device *adev)
+{
+	/* set graphics engine doorbell range */
+	WREG32_SOC15(GC, 0, regCP_RB_DOORBELL_RANGE_LOWER,
+		     (adev->doorbell_index.gfx_ring0 * 2) << 2);
+	WREG32_SOC15(GC, 0, regCP_RB_DOORBELL_RANGE_UPPER,
+		     (adev->doorbell_index.gfx_userqueue_end * 2) << 2);
+
+	/* set compute engine doorbell range */
+	WREG32_SOC15(GC, 0, regCP_MEC_DOORBELL_RANGE_LOWER,
+		     (adev->doorbell_index.kiq * 2) << 2);
+	WREG32_SOC15(GC, 0, regCP_MEC_DOORBELL_RANGE_UPPER,
+		     (adev->doorbell_index.userqueue_end * 2) << 2);
+}
+
+static void gfx_v11_0_gfx_mqd_set_priority(struct amdgpu_device *adev,
+					   struct v11_gfx_mqd *mqd,
+					   struct amdgpu_mqd_prop *prop)
+{
+	bool priority = 0;
+	u32 tmp;
+
+	/* set up default queue priority level
+	 * 0x0 = low priority, 0x1 = high priority
+	 */
+	if (prop->hqd_pipe_priority == AMDGPU_GFX_PIPE_PRIO_HIGH)
+		priority = 1;
+
+	tmp = regCP_GFX_HQD_QUEUE_PRIORITY_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_QUEUE_PRIORITY, PRIORITY_LEVEL, priority);
+	mqd->cp_gfx_hqd_queue_priority = tmp;
+}
+
+static int gfx_v11_0_gfx_mqd_init(struct amdgpu_device *adev, void *m,
+				  struct amdgpu_mqd_prop *prop)
+{
+	struct v11_gfx_mqd *mqd = m;
+	uint64_t hqd_gpu_addr, wb_gpu_addr;
+	uint32_t tmp;
+	uint32_t rb_bufsz;
+
+	/* set up gfx hqd wptr */
+	mqd->cp_gfx_hqd_wptr = 0;
+	mqd->cp_gfx_hqd_wptr_hi = 0;
+
+	/* set the pointer to the MQD */
+	mqd->cp_mqd_base_addr = prop->mqd_gpu_addr & 0xfffffffc;
+	mqd->cp_mqd_base_addr_hi = upper_32_bits(prop->mqd_gpu_addr);
+
+	/* set up mqd control */
+	tmp = regCP_GFX_MQD_CONTROL_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_GFX_MQD_CONTROL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_MQD_CONTROL, PRIV_STATE, 1);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_MQD_CONTROL, CACHE_POLICY, 0);
+	mqd->cp_gfx_mqd_control = tmp;
+
+	/* set up gfx_hqd_vimd with 0x0 to indicate the ring buffer's vmid */
+	tmp = regCP_GFX_HQD_VMID_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_VMID, VMID, 0);
+	mqd->cp_gfx_hqd_vmid = 0;
+
+	/* set up gfx queue priority */
+	gfx_v11_0_gfx_mqd_set_priority(adev, mqd, prop);
+
+	/* set up time quantum */
+	tmp = regCP_GFX_HQD_QUANTUM_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_QUANTUM, QUANTUM_EN, 1);
+	mqd->cp_gfx_hqd_quantum = tmp;
+
+	/* set up gfx hqd base. this is similar as CP_RB_BASE */
+	hqd_gpu_addr = prop->hqd_base_gpu_addr >> 8;
+	mqd->cp_gfx_hqd_base = hqd_gpu_addr;
+	mqd->cp_gfx_hqd_base_hi = upper_32_bits(hqd_gpu_addr);
+
+	/* set up hqd_rptr_addr/_hi, similar as CP_RB_RPTR */
+	wb_gpu_addr = prop->rptr_gpu_addr;
+	mqd->cp_gfx_hqd_rptr_addr = wb_gpu_addr & 0xfffffffc;
+	mqd->cp_gfx_hqd_rptr_addr_hi =
+		upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	/* set up rb_wptr_poll addr */
+	wb_gpu_addr = prop->wptr_gpu_addr;
+	mqd->cp_rb_wptr_poll_addr_lo = wb_gpu_addr & 0xfffffffc;
+	mqd->cp_rb_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	/* set up the gfx_hqd_control, similar as CP_RB0_CNTL */
+	rb_bufsz = order_base_2(prop->queue_size / 4) - 1;
+	tmp = regCP_GFX_HQD_CNTL_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, RB_BUFSZ, rb_bufsz);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, RB_BLKSZ, rb_bufsz - 2);
+#ifdef __BIG_ENDIAN
+	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, BUF_SWAP, 1);
+#endif
+	if (prop->tmz_queue)
+		tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, TMZ_MATCH, 1);
+	if (!prop->kernel_queue)
+		tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, RB_NON_PRIV, 1);
+	mqd->cp_gfx_hqd_cntl = tmp;
+
+	/* set up cp_doorbell_control */
+	tmp = regCP_RB_DOORBELL_CONTROL_DEFAULT;
+	if (prop->use_doorbell) {
+		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+				    DOORBELL_OFFSET, prop->doorbell_index);
+		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+				    DOORBELL_EN, 1);
+	} else
+		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+				    DOORBELL_EN, 0);
+	mqd->cp_rb_doorbell_control = tmp;
+
+	/* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */
+	mqd->cp_gfx_hqd_rptr = regCP_GFX_HQD_RPTR_DEFAULT;
+
+	/* active the queue */
+	mqd->cp_gfx_hqd_active = 1;
+
+	/* set gfx UQ items */
+	mqd->shadow_base_lo = lower_32_bits(prop->shadow_addr);
+	mqd->shadow_base_hi = upper_32_bits(prop->shadow_addr);
+	mqd->gds_bkup_base_lo = lower_32_bits(prop->gds_bkup_addr);
+	mqd->gds_bkup_base_hi = upper_32_bits(prop->gds_bkup_addr);
+	mqd->fw_work_area_base_lo = lower_32_bits(prop->csa_addr);
+	mqd->fw_work_area_base_hi = upper_32_bits(prop->csa_addr);
+	mqd->fence_address_lo = lower_32_bits(prop->fence_address);
+	mqd->fence_address_hi = upper_32_bits(prop->fence_address);
+
+	return 0;
+}
+
+static int gfx_v11_0_kgq_init_queue(struct amdgpu_ring *ring, bool reset)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v11_gfx_mqd *mqd = ring->mqd_ptr;
+	int mqd_idx = ring - &adev->gfx.gfx_ring[0];
+
+	if (!reset && !amdgpu_in_reset(adev) && !adev->in_suspend) {
+		memset((void *)mqd, 0, sizeof(*mqd));
+		mutex_lock(&adev->srbm_mutex);
+		soc21_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		amdgpu_ring_init_mqd(ring);
+		soc21_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+		if (adev->gfx.me.mqd_backup[mqd_idx])
+			memcpy_fromio(adev->gfx.me.mqd_backup[mqd_idx], mqd, sizeof(*mqd));
+	} else {
+		/* restore mqd with the backup copy */
+		if (adev->gfx.me.mqd_backup[mqd_idx])
+			memcpy_toio(mqd, adev->gfx.me.mqd_backup[mqd_idx], sizeof(*mqd));
+		/* reset the ring */
+		ring->wptr = 0;
+		*ring->wptr_cpu_addr = 0;
+		amdgpu_ring_clear_ring(ring);
+	}
+
+	return 0;
+}
+
+static int gfx_v11_0_cp_async_gfx_ring_resume(struct amdgpu_device *adev)
+{
+	int r, i;
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+		r = gfx_v11_0_kgq_init_queue(&adev->gfx.gfx_ring[i], false);
+		if (r)
+			return r;
+	}
+
+	r = amdgpu_gfx_enable_kgq(adev, 0);
+	if (r)
+		return r;
+
+	return gfx_v11_0_cp_gfx_start(adev);
+}
+
+static int gfx_v11_0_compute_mqd_init(struct amdgpu_device *adev, void *m,
+				      struct amdgpu_mqd_prop *prop)
+{
+	struct v11_compute_mqd *mqd = m;
+	uint64_t hqd_gpu_addr, wb_gpu_addr, eop_base_addr;
+	uint32_t tmp;
+
+	mqd->header = 0xC0310800;
+	mqd->compute_pipelinestat_enable = 0x00000001;
+	mqd->compute_static_thread_mgmt_se0 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se1 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se2 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se3 = 0xffffffff;
+	mqd->compute_misc_reserved = 0x00000007;
+
+	eop_base_addr = prop->eop_gpu_addr >> 8;
+	mqd->cp_hqd_eop_base_addr_lo = eop_base_addr;
+	mqd->cp_hqd_eop_base_addr_hi = upper_32_bits(eop_base_addr);
+
+	/* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */
+	tmp = regCP_HQD_EOP_CONTROL_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_HQD_EOP_CONTROL, EOP_SIZE,
+			(order_base_2(GFX11_MEC_HPD_SIZE / 4) - 1));
+
+	mqd->cp_hqd_eop_control = tmp;
+
+	/* enable doorbell? */
+	tmp = regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT;
+
+	if (prop->use_doorbell) {
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_OFFSET, prop->doorbell_index);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_EN, 1);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_SOURCE, 0);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_HIT, 0);
+	} else {
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_EN, 0);
+	}
+
+	mqd->cp_hqd_pq_doorbell_control = tmp;
+
+	/* disable the queue if it's active */
+	mqd->cp_hqd_dequeue_request = 0;
+	mqd->cp_hqd_pq_rptr = 0;
+	mqd->cp_hqd_pq_wptr_lo = 0;
+	mqd->cp_hqd_pq_wptr_hi = 0;
+
+	/* set the pointer to the MQD */
+	mqd->cp_mqd_base_addr_lo = prop->mqd_gpu_addr & 0xfffffffc;
+	mqd->cp_mqd_base_addr_hi = upper_32_bits(prop->mqd_gpu_addr);
+
+	/* set MQD vmid to 0 */
+	tmp = regCP_MQD_CONTROL_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_MQD_CONTROL, VMID, 0);
+	mqd->cp_mqd_control = tmp;
+
+	/* set the pointer to the HQD, this is similar CP_RB0_BASE/_HI */
+	hqd_gpu_addr = prop->hqd_base_gpu_addr >> 8;
+	mqd->cp_hqd_pq_base_lo = hqd_gpu_addr;
+	mqd->cp_hqd_pq_base_hi = upper_32_bits(hqd_gpu_addr);
+
+	/* set up the HQD, this is similar to CP_RB0_CNTL */
+	tmp = regCP_HQD_PQ_CONTROL_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, QUEUE_SIZE,
+			    (order_base_2(prop->queue_size / 4) - 1));
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, RPTR_BLOCK_SIZE,
+			    (order_base_2(AMDGPU_GPU_PAGE_SIZE / 4) - 1));
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, UNORD_DISPATCH, 1);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, TUNNEL_DISPATCH,
+			    prop->allow_tunneling);
+	if (prop->kernel_queue) {
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, PRIV_STATE, 1);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, KMD_QUEUE, 1);
+	}
+	if (prop->tmz_queue)
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, TMZ, 1);
+	mqd->cp_hqd_pq_control = tmp;
+
+	/* set the wb address whether it's enabled or not */
+	wb_gpu_addr = prop->rptr_gpu_addr;
+	mqd->cp_hqd_pq_rptr_report_addr_lo = wb_gpu_addr & 0xfffffffc;
+	mqd->cp_hqd_pq_rptr_report_addr_hi =
+		upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	/* only used if CP_PQ_WPTR_POLL_CNTL.CP_PQ_WPTR_POLL_CNTL__EN_MASK=1 */
+	wb_gpu_addr = prop->wptr_gpu_addr;
+	mqd->cp_hqd_pq_wptr_poll_addr_lo = wb_gpu_addr & 0xfffffffc;
+	mqd->cp_hqd_pq_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	tmp = 0;
+	/* enable the doorbell if requested */
+	if (prop->use_doorbell) {
+		tmp = regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT;
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				DOORBELL_OFFSET, prop->doorbell_index);
+
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_EN, 1);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_SOURCE, 0);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_HIT, 0);
+	}
+
+	mqd->cp_hqd_pq_doorbell_control = tmp;
+
+	/* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */
+	mqd->cp_hqd_pq_rptr = regCP_HQD_PQ_RPTR_DEFAULT;
+
+	/* set the vmid for the queue */
+	mqd->cp_hqd_vmid = 0;
+
+	tmp = regCP_HQD_PERSISTENT_STATE_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PERSISTENT_STATE, PRELOAD_SIZE, 0x55);
+	mqd->cp_hqd_persistent_state = tmp;
+
+	/* set MIN_IB_AVAIL_SIZE */
+	tmp = regCP_HQD_IB_CONTROL_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_HQD_IB_CONTROL, MIN_IB_AVAIL_SIZE, 3);
+	mqd->cp_hqd_ib_control = tmp;
+
+	/* set static priority for a compute queue/ring */
+	mqd->cp_hqd_pipe_priority = prop->hqd_pipe_priority;
+	mqd->cp_hqd_queue_priority = prop->hqd_queue_priority;
+
+	mqd->cp_hqd_active = prop->hqd_active;
+
+	/* set UQ fenceaddress */
+	mqd->fence_address_lo = lower_32_bits(prop->fence_address);
+	mqd->fence_address_hi = upper_32_bits(prop->fence_address);
+
+	return 0;
+}
+
+static int gfx_v11_0_kiq_init_register(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v11_compute_mqd *mqd = ring->mqd_ptr;
+	int j;
+
+	/* inactivate the queue */
+	if (amdgpu_sriov_vf(adev))
+		WREG32_SOC15(GC, 0, regCP_HQD_ACTIVE, 0);
+
+	/* disable wptr polling */
+	WREG32_FIELD15_PREREG(GC, 0, CP_PQ_WPTR_POLL_CNTL, EN, 0);
+
+	/* write the EOP addr */
+	WREG32_SOC15(GC, 0, regCP_HQD_EOP_BASE_ADDR,
+	       mqd->cp_hqd_eop_base_addr_lo);
+	WREG32_SOC15(GC, 0, regCP_HQD_EOP_BASE_ADDR_HI,
+	       mqd->cp_hqd_eop_base_addr_hi);
+
+	/* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */
+	WREG32_SOC15(GC, 0, regCP_HQD_EOP_CONTROL,
+	       mqd->cp_hqd_eop_control);
+
+	/* enable doorbell? */
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL,
+	       mqd->cp_hqd_pq_doorbell_control);
+
+	/* disable the queue if it's active */
+	if (RREG32_SOC15(GC, 0, regCP_HQD_ACTIVE) & 1) {
+		WREG32_SOC15(GC, 0, regCP_HQD_DEQUEUE_REQUEST, 1);
+		for (j = 0; j < adev->usec_timeout; j++) {
+			if (!(RREG32_SOC15(GC, 0, regCP_HQD_ACTIVE) & 1))
+				break;
+			udelay(1);
+		}
+		WREG32_SOC15(GC, 0, regCP_HQD_DEQUEUE_REQUEST,
+		       mqd->cp_hqd_dequeue_request);
+		WREG32_SOC15(GC, 0, regCP_HQD_PQ_RPTR,
+		       mqd->cp_hqd_pq_rptr);
+		WREG32_SOC15(GC, 0, regCP_HQD_PQ_WPTR_LO,
+		       mqd->cp_hqd_pq_wptr_lo);
+		WREG32_SOC15(GC, 0, regCP_HQD_PQ_WPTR_HI,
+		       mqd->cp_hqd_pq_wptr_hi);
+	}
+
+	/* set the pointer to the MQD */
+	WREG32_SOC15(GC, 0, regCP_MQD_BASE_ADDR,
+	       mqd->cp_mqd_base_addr_lo);
+	WREG32_SOC15(GC, 0, regCP_MQD_BASE_ADDR_HI,
+	       mqd->cp_mqd_base_addr_hi);
+
+	/* set MQD vmid to 0 */
+	WREG32_SOC15(GC, 0, regCP_MQD_CONTROL,
+	       mqd->cp_mqd_control);
+
+	/* set the pointer to the HQD, this is similar CP_RB0_BASE/_HI */
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_BASE,
+	       mqd->cp_hqd_pq_base_lo);
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_BASE_HI,
+	       mqd->cp_hqd_pq_base_hi);
+
+	/* set up the HQD, this is similar to CP_RB0_CNTL */
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_CONTROL,
+	       mqd->cp_hqd_pq_control);
+
+	/* set the wb address whether it's enabled or not */
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_RPTR_REPORT_ADDR,
+		mqd->cp_hqd_pq_rptr_report_addr_lo);
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_RPTR_REPORT_ADDR_HI,
+		mqd->cp_hqd_pq_rptr_report_addr_hi);
+
+	/* only used if CP_PQ_WPTR_POLL_CNTL.CP_PQ_WPTR_POLL_CNTL__EN_MASK=1 */
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_WPTR_POLL_ADDR,
+	       mqd->cp_hqd_pq_wptr_poll_addr_lo);
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_WPTR_POLL_ADDR_HI,
+	       mqd->cp_hqd_pq_wptr_poll_addr_hi);
+
+	/* enable the doorbell if requested */
+	if (ring->use_doorbell) {
+		WREG32_SOC15(GC, 0, regCP_MEC_DOORBELL_RANGE_LOWER,
+			(adev->doorbell_index.kiq * 2) << 2);
+		WREG32_SOC15(GC, 0, regCP_MEC_DOORBELL_RANGE_UPPER,
+			(adev->doorbell_index.userqueue_end * 2) << 2);
+	}
+
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL,
+	       mqd->cp_hqd_pq_doorbell_control);
+
+	/* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_WPTR_LO,
+	       mqd->cp_hqd_pq_wptr_lo);
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_WPTR_HI,
+	       mqd->cp_hqd_pq_wptr_hi);
+
+	/* set the vmid for the queue */
+	WREG32_SOC15(GC, 0, regCP_HQD_VMID, mqd->cp_hqd_vmid);
+
+	WREG32_SOC15(GC, 0, regCP_HQD_PERSISTENT_STATE,
+	       mqd->cp_hqd_persistent_state);
+
+	/* activate the queue */
+	WREG32_SOC15(GC, 0, regCP_HQD_ACTIVE,
+	       mqd->cp_hqd_active);
+
+	if (ring->use_doorbell)
+		WREG32_FIELD15_PREREG(GC, 0, CP_PQ_STATUS, DOORBELL_ENABLE, 1);
+
+	return 0;
+}
+
+static int gfx_v11_0_kiq_init_queue(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v11_compute_mqd *mqd = ring->mqd_ptr;
+
+	gfx_v11_0_kiq_setting(ring);
+
+	if (amdgpu_in_reset(adev)) { /* for GPU_RESET case */
+		/* reset MQD to a clean status */
+		if (adev->gfx.kiq[0].mqd_backup)
+			memcpy_toio(mqd, adev->gfx.kiq[0].mqd_backup, sizeof(*mqd));
+
+		/* reset ring buffer */
+		ring->wptr = 0;
+		amdgpu_ring_clear_ring(ring);
+
+		mutex_lock(&adev->srbm_mutex);
+		soc21_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		gfx_v11_0_kiq_init_register(ring);
+		soc21_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+	} else {
+		memset((void *)mqd, 0, sizeof(*mqd));
+		if (amdgpu_sriov_vf(adev) && adev->in_suspend)
+			amdgpu_ring_clear_ring(ring);
+		mutex_lock(&adev->srbm_mutex);
+		soc21_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		amdgpu_ring_init_mqd(ring);
+		gfx_v11_0_kiq_init_register(ring);
+		soc21_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+
+		if (adev->gfx.kiq[0].mqd_backup)
+			memcpy_fromio(adev->gfx.kiq[0].mqd_backup, mqd, sizeof(*mqd));
+	}
+
+	return 0;
+}
+
+static int gfx_v11_0_kcq_init_queue(struct amdgpu_ring *ring, bool reset)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v11_compute_mqd *mqd = ring->mqd_ptr;
+	int mqd_idx = ring - &adev->gfx.compute_ring[0];
+
+	if (!reset && !amdgpu_in_reset(adev) && !adev->in_suspend) {
+		memset((void *)mqd, 0, sizeof(*mqd));
+		mutex_lock(&adev->srbm_mutex);
+		soc21_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		amdgpu_ring_init_mqd(ring);
+		soc21_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+
+		if (adev->gfx.mec.mqd_backup[mqd_idx])
+			memcpy_fromio(adev->gfx.mec.mqd_backup[mqd_idx], mqd, sizeof(*mqd));
+	} else {
+		/* restore MQD to a clean status */
+		if (adev->gfx.mec.mqd_backup[mqd_idx])
+			memcpy_toio(mqd, adev->gfx.mec.mqd_backup[mqd_idx], sizeof(*mqd));
+		/* reset ring buffer */
+		ring->wptr = 0;
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr, 0);
+		amdgpu_ring_clear_ring(ring);
+	}
+
+	return 0;
+}
+
+static int gfx_v11_0_kiq_resume(struct amdgpu_device *adev)
+{
+	gfx_v11_0_kiq_init_queue(&adev->gfx.kiq[0].ring);
+	return 0;
+}
+
+static int gfx_v11_0_kcq_resume(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	if (!amdgpu_async_gfx_ring)
+		gfx_v11_0_cp_compute_enable(adev, true);
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		r = gfx_v11_0_kcq_init_queue(&adev->gfx.compute_ring[i], false);
+		if (r)
+			return r;
+	}
+
+	return amdgpu_gfx_enable_kcq(adev, 0);
+}
+
+static int gfx_v11_0_cp_resume(struct amdgpu_device *adev)
+{
+	int r, i;
+	struct amdgpu_ring *ring;
+
+	if (!(adev->flags & AMD_IS_APU))
+		gfx_v11_0_enable_gui_idle_interrupt(adev, false);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+		/* legacy firmware loading */
+		r = gfx_v11_0_cp_gfx_load_microcode(adev);
+		if (r)
+			return r;
+
+		if (adev->gfx.rs64_enable)
+			r = gfx_v11_0_cp_compute_load_microcode_rs64(adev);
+		else
+			r = gfx_v11_0_cp_compute_load_microcode(adev);
+		if (r)
+			return r;
+	}
+
+	gfx_v11_0_cp_set_doorbell_range(adev);
+
+	if (amdgpu_async_gfx_ring) {
+		gfx_v11_0_cp_compute_enable(adev, true);
+		gfx_v11_0_cp_gfx_enable(adev, true);
+	}
+
+	if (adev->enable_mes_kiq && adev->mes.kiq_hw_init)
+		r = amdgpu_mes_kiq_hw_init(adev);
+	else
+		r = gfx_v11_0_kiq_resume(adev);
+	if (r)
+		return r;
+
+	r = gfx_v11_0_kcq_resume(adev);
+	if (r)
+		return r;
+
+	if (!amdgpu_async_gfx_ring) {
+		r = gfx_v11_0_cp_gfx_resume(adev);
+		if (r)
+			return r;
+	} else {
+		r = gfx_v11_0_cp_async_gfx_ring_resume(adev);
+		if (r)
+			return r;
+	}
+
+	if (adev->gfx.disable_kq) {
+		for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+			ring = &adev->gfx.gfx_ring[i];
+			/* we don't want to set ring->ready */
+			r = amdgpu_ring_test_ring(ring);
+			if (r)
+				return r;
+		}
+		if (amdgpu_async_gfx_ring)
+			amdgpu_gfx_disable_kgq(adev, 0);
+	} else {
+		for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+			ring = &adev->gfx.gfx_ring[i];
+			r = amdgpu_ring_test_helper(ring);
+			if (r)
+				return r;
+		}
+	}
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		ring = &adev->gfx.compute_ring[i];
+		r = amdgpu_ring_test_helper(ring);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+static void gfx_v11_0_cp_enable(struct amdgpu_device *adev, bool enable)
+{
+	gfx_v11_0_cp_gfx_enable(adev, enable);
+	gfx_v11_0_cp_compute_enable(adev, enable);
+}
+
+static int gfx_v11_0_gfxhub_enable(struct amdgpu_device *adev)
+{
+	int r;
+	bool value;
+
+	r = adev->gfxhub.funcs->gart_enable(adev);
+	if (r)
+		return r;
+
+	amdgpu_device_flush_hdp(adev, NULL);
+
+	value = amdgpu_vm_fault_stop != AMDGPU_VM_FAULT_STOP_ALWAYS;
+
+	adev->gfxhub.funcs->set_fault_enable_default(adev, value);
+	/* TODO investigate why this and the hdp flush above is needed,
+	 * are we missing a flush somewhere else? */
+	adev->gmc.gmc_funcs->flush_gpu_tlb(adev, 0, AMDGPU_GFXHUB(0), 0);
+
+	return 0;
+}
+
+static void gfx_v11_0_select_cp_fw_arch(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	/* select RS64 */
+	if (adev->gfx.rs64_enable) {
+		tmp = RREG32_SOC15(GC, 0, regCP_GFX_CNTL);
+		tmp = REG_SET_FIELD(tmp, CP_GFX_CNTL, ENGINE_SEL, 1);
+		WREG32_SOC15(GC, 0, regCP_GFX_CNTL, tmp);
+
+		tmp = RREG32_SOC15(GC, 0, regCP_MEC_ISA_CNTL);
+		tmp = REG_SET_FIELD(tmp, CP_MEC_ISA_CNTL, ISA_MODE, 1);
+		WREG32_SOC15(GC, 0, regCP_MEC_ISA_CNTL, tmp);
+	}
+
+	if (amdgpu_emu_mode == 1)
+		msleep(100);
+}
+
+static int get_gb_addr_config(struct amdgpu_device * adev)
+{
+	u32 gb_addr_config;
+
+	gb_addr_config = RREG32_SOC15(GC, 0, regGB_ADDR_CONFIG);
+	if (gb_addr_config == 0)
+		return -EINVAL;
+
+	adev->gfx.config.gb_addr_config_fields.num_pkrs =
+		1 << REG_GET_FIELD(gb_addr_config, GB_ADDR_CONFIG, NUM_PKRS);
+
+	adev->gfx.config.gb_addr_config = gb_addr_config;
+
+	adev->gfx.config.gb_addr_config_fields.num_pipes = 1 <<
+			REG_GET_FIELD(adev->gfx.config.gb_addr_config,
+				      GB_ADDR_CONFIG, NUM_PIPES);
+
+	adev->gfx.config.max_tile_pipes =
+		adev->gfx.config.gb_addr_config_fields.num_pipes;
+
+	adev->gfx.config.gb_addr_config_fields.max_compress_frags = 1 <<
+			REG_GET_FIELD(adev->gfx.config.gb_addr_config,
+				      GB_ADDR_CONFIG, MAX_COMPRESSED_FRAGS);
+	adev->gfx.config.gb_addr_config_fields.num_rb_per_se = 1 <<
+			REG_GET_FIELD(adev->gfx.config.gb_addr_config,
+				      GB_ADDR_CONFIG, NUM_RB_PER_SE);
+	adev->gfx.config.gb_addr_config_fields.num_se = 1 <<
+			REG_GET_FIELD(adev->gfx.config.gb_addr_config,
+				      GB_ADDR_CONFIG, NUM_SHADER_ENGINES);
+	adev->gfx.config.gb_addr_config_fields.pipe_interleave_size = 1 << (8 +
+			REG_GET_FIELD(adev->gfx.config.gb_addr_config,
+				      GB_ADDR_CONFIG, PIPE_INTERLEAVE_SIZE));
+
+	return 0;
+}
+
+static void gfx_v11_0_disable_gpa_mode(struct amdgpu_device *adev)
+{
+	uint32_t data;
+
+	data = RREG32_SOC15(GC, 0, regCPC_PSP_DEBUG);
+	data |= CPC_PSP_DEBUG__GPA_OVERRIDE_MASK;
+	WREG32_SOC15(GC, 0, regCPC_PSP_DEBUG, data);
+
+	data = RREG32_SOC15(GC, 0, regCPG_PSP_DEBUG);
+	data |= CPG_PSP_DEBUG__GPA_OVERRIDE_MASK;
+	WREG32_SOC15(GC, 0, regCPG_PSP_DEBUG, data);
+}
+
+static int gfx_v11_0_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	amdgpu_gfx_cleaner_shader_init(adev, adev->gfx.cleaner_shader_size,
+				       adev->gfx.cleaner_shader_ptr);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO) {
+		if (adev->gfx.imu.funcs) {
+			/* RLC autoload sequence 1: Program rlc ram */
+			if (adev->gfx.imu.funcs->program_rlc_ram)
+				adev->gfx.imu.funcs->program_rlc_ram(adev);
+			/* rlc autoload firmware */
+			r = gfx_v11_0_rlc_backdoor_autoload_enable(adev);
+			if (r)
+				return r;
+		}
+	} else {
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+			if (adev->gfx.imu.funcs && (amdgpu_dpm > 0)) {
+				if (adev->gfx.imu.funcs->load_microcode)
+					adev->gfx.imu.funcs->load_microcode(adev);
+				if (adev->gfx.imu.funcs->setup_imu)
+					adev->gfx.imu.funcs->setup_imu(adev);
+				if (adev->gfx.imu.funcs->start_imu)
+					adev->gfx.imu.funcs->start_imu(adev);
+			}
+
+			/* disable gpa mode in backdoor loading */
+			gfx_v11_0_disable_gpa_mode(adev);
+		}
+	}
+
+	if ((adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO) ||
+	    (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP)) {
+		r = gfx_v11_0_wait_for_rlc_autoload_complete(adev);
+		if (r) {
+			dev_err(adev->dev, "(%d) failed to wait rlc autoload complete\n", r);
+			return r;
+		}
+	}
+
+	adev->gfx.is_poweron = true;
+
+	if(get_gb_addr_config(adev))
+		DRM_WARN("Invalid gb_addr_config !\n");
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP &&
+	    adev->gfx.rs64_enable)
+		gfx_v11_0_config_gfx_rs64(adev);
+
+	r = gfx_v11_0_gfxhub_enable(adev);
+	if (r)
+		return r;
+
+	if (!amdgpu_emu_mode)
+		gfx_v11_0_init_golden_registers(adev);
+
+	if ((adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) ||
+	    (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO && amdgpu_dpm == 1)) {
+		/**
+		 * For gfx 11, rlc firmware loading relies on smu firmware is
+		 * loaded firstly, so in direct type, it has to load smc ucode
+		 * here before rlc.
+		 */
+		r = amdgpu_pm_load_smu_firmware(adev, NULL);
+		if (r)
+			return r;
+	}
+
+	gfx_v11_0_constants_init(adev);
+
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
+		gfx_v11_0_select_cp_fw_arch(adev);
+
+	if (adev->nbio.funcs->gc_doorbell_init)
+		adev->nbio.funcs->gc_doorbell_init(adev);
+
+	r = gfx_v11_0_rlc_resume(adev);
+	if (r)
+		return r;
+
+	/*
+	 * init golden registers and rlc resume may override some registers,
+	 * reconfig them here
+	 */
+	gfx_v11_0_tcp_harvest(adev);
+
+	r = gfx_v11_0_cp_resume(adev);
+	if (r)
+		return r;
+
+	/* get IMU version from HW if it's not set */
+	if (!adev->gfx.imu_fw_version)
+		adev->gfx.imu_fw_version = RREG32_SOC15(GC, 0, regGFX_IMU_SCRATCH_0);
+
+	return r;
+}
+
+static int gfx_v11_0_set_userq_eop_interrupts(struct amdgpu_device *adev,
+					      bool enable)
+{
+	unsigned int irq_type;
+	int m, p, r;
+
+	if (adev->userq_funcs[AMDGPU_HW_IP_GFX]) {
+		for (m = 0; m < adev->gfx.me.num_me; m++) {
+			for (p = 0; p < adev->gfx.me.num_pipe_per_me; p++) {
+				irq_type = AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP + p;
+				if (enable)
+					r = amdgpu_irq_get(adev, &adev->gfx.eop_irq,
+							   irq_type);
+				else
+					r = amdgpu_irq_put(adev, &adev->gfx.eop_irq,
+							   irq_type);
+				if (r)
+					return r;
+			}
+		}
+	}
+
+	if (adev->userq_funcs[AMDGPU_HW_IP_COMPUTE]) {
+		for (m = 0; m < adev->gfx.mec.num_mec; ++m) {
+			for (p = 0; p < adev->gfx.mec.num_pipe_per_mec; p++) {
+				irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP
+					+ (m * adev->gfx.mec.num_pipe_per_mec)
+					+ p;
+				if (enable)
+					r = amdgpu_irq_get(adev, &adev->gfx.eop_irq,
+							   irq_type);
+				else
+					r = amdgpu_irq_put(adev, &adev->gfx.eop_irq,
+							   irq_type);
+				if (r)
+					return r;
+			}
+		}
+	}
+
+	return 0;
+}
+
+static int gfx_v11_0_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	cancel_delayed_work_sync(&adev->gfx.idle_work);
+
+	amdgpu_irq_put(adev, &adev->gfx.priv_reg_irq, 0);
+	amdgpu_irq_put(adev, &adev->gfx.priv_inst_irq, 0);
+	amdgpu_irq_put(adev, &adev->gfx.bad_op_irq, 0);
+	gfx_v11_0_set_userq_eop_interrupts(adev, false);
+
+	if (!adev->no_hw_access) {
+		if (amdgpu_async_gfx_ring &&
+		    !adev->gfx.disable_kq) {
+			if (amdgpu_gfx_disable_kgq(adev, 0))
+				DRM_ERROR("KGQ disable failed\n");
+		}
+
+		if (amdgpu_gfx_disable_kcq(adev, 0))
+			DRM_ERROR("KCQ disable failed\n");
+
+		amdgpu_mes_kiq_hw_fini(adev);
+	}
+
+	if (amdgpu_sriov_vf(adev))
+		/* Remove the steps disabling CPG and clearing KIQ position,
+		 * so that CP could perform IDLE-SAVE during switch. Those
+		 * steps are necessary to avoid a DMAR error in gfx9 but it is
+		 * not reproduced on gfx11.
+		 */
+		return 0;
+
+	gfx_v11_0_cp_enable(adev, false);
+	gfx_v11_0_enable_gui_idle_interrupt(adev, false);
+
+	adev->gfxhub.funcs->gart_disable(adev);
+
+	adev->gfx.is_poweron = false;
+
+	return 0;
+}
+
+static int gfx_v11_0_suspend(struct amdgpu_ip_block *ip_block)
+{
+	return gfx_v11_0_hw_fini(ip_block);
+}
+
+static int gfx_v11_0_resume(struct amdgpu_ip_block *ip_block)
+{
+	return gfx_v11_0_hw_init(ip_block);
+}
+
+static bool gfx_v11_0_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (REG_GET_FIELD(RREG32_SOC15(GC, 0, regGRBM_STATUS),
+				GRBM_STATUS, GUI_ACTIVE))
+		return false;
+	else
+		return true;
+}
+
+static int gfx_v11_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		/* read MC_STATUS */
+		tmp = RREG32_SOC15(GC, 0, regGRBM_STATUS) &
+			GRBM_STATUS__GUI_ACTIVE_MASK;
+
+		if (!REG_GET_FIELD(tmp, GRBM_STATUS, GUI_ACTIVE))
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+int gfx_v11_0_request_gfx_index_mutex(struct amdgpu_device *adev,
+				      bool req)
+{
+	u32 i, tmp, val;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		/* Request with MeId=2, PipeId=0 */
+		tmp = REG_SET_FIELD(0, CP_GFX_INDEX_MUTEX, REQUEST, req);
+		tmp = REG_SET_FIELD(tmp, CP_GFX_INDEX_MUTEX, CLIENTID, 4);
+		WREG32_SOC15(GC, 0, regCP_GFX_INDEX_MUTEX, tmp);
+
+		val = RREG32_SOC15(GC, 0, regCP_GFX_INDEX_MUTEX);
+		if (req) {
+			if (val == tmp)
+				break;
+		} else {
+			tmp = REG_SET_FIELD(tmp, CP_GFX_INDEX_MUTEX,
+					    REQUEST, 1);
+
+			/* unlocked or locked by firmware */
+			if (val != tmp)
+				break;
+		}
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout)
+		return -EINVAL;
+
+	return 0;
+}
+
+static int gfx_v11_0_soft_reset(struct amdgpu_ip_block *ip_block)
+{
+	u32 grbm_soft_reset = 0;
+	u32 tmp;
+	int r, i, j, k;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
+
+	tmp = RREG32_SOC15(GC, 0, regCP_INT_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL, CMP_BUSY_INT_ENABLE, 0);
+	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL, CNTX_BUSY_INT_ENABLE, 0);
+	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL, CNTX_EMPTY_INT_ENABLE, 0);
+	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL, GFX_IDLE_INT_ENABLE, 0);
+	WREG32_SOC15(GC, 0, regCP_INT_CNTL, tmp);
+
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < adev->gfx.mec.num_mec; ++i) {
+		for (j = 0; j < adev->gfx.mec.num_queue_per_pipe; j++) {
+			for (k = 0; k < adev->gfx.mec.num_pipe_per_mec; k++) {
+				soc21_grbm_select(adev, i, k, j, 0);
+
+				WREG32_SOC15(GC, 0, regCP_HQD_DEQUEUE_REQUEST, 0x2);
+				WREG32_SOC15(GC, 0, regSPI_COMPUTE_QUEUE_RESET, 0x1);
+			}
+		}
+	}
+	for (i = 0; i < adev->gfx.me.num_me; ++i) {
+		for (j = 0; j < adev->gfx.me.num_queue_per_pipe; j++) {
+			for (k = 0; k < adev->gfx.me.num_pipe_per_me; k++) {
+				soc21_grbm_select(adev, i, k, j, 0);
+
+				WREG32_SOC15(GC, 0, regCP_GFX_HQD_DEQUEUE_REQUEST, 0x1);
+			}
+		}
+	}
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	/* Try to acquire the gfx mutex before access to CP_VMID_RESET */
+	mutex_lock(&adev->gfx.reset_sem_mutex);
+	r = gfx_v11_0_request_gfx_index_mutex(adev, true);
+	if (r) {
+		mutex_unlock(&adev->gfx.reset_sem_mutex);
+		DRM_ERROR("Failed to acquire the gfx mutex during soft reset\n");
+		return r;
+	}
+
+	WREG32_SOC15(GC, 0, regCP_VMID_RESET, 0xfffffffe);
+
+	// Read CP_VMID_RESET register three times.
+	// to get sufficient time for GFX_HQD_ACTIVE reach 0
+	RREG32_SOC15(GC, 0, regCP_VMID_RESET);
+	RREG32_SOC15(GC, 0, regCP_VMID_RESET);
+	RREG32_SOC15(GC, 0, regCP_VMID_RESET);
+
+	/* release the gfx mutex */
+	r = gfx_v11_0_request_gfx_index_mutex(adev, false);
+	mutex_unlock(&adev->gfx.reset_sem_mutex);
+	if (r) {
+		DRM_ERROR("Failed to release the gfx mutex during soft reset\n");
+		return r;
+	}
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (!RREG32_SOC15(GC, 0, regCP_HQD_ACTIVE) &&
+		    !RREG32_SOC15(GC, 0, regCP_GFX_HQD_ACTIVE))
+			break;
+		udelay(1);
+	}
+	if (i >= adev->usec_timeout) {
+		printk("Failed to wait all pipes clean\n");
+		return -EINVAL;
+	}
+
+	/**********  trigger soft reset  ***********/
+	grbm_soft_reset = RREG32_SOC15(GC, 0, regGRBM_SOFT_RESET);
+	grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET,
+					SOFT_RESET_CP, 1);
+	grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET,
+					SOFT_RESET_GFX, 1);
+	grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET,
+					SOFT_RESET_CPF, 1);
+	grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET,
+					SOFT_RESET_CPC, 1);
+	grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET,
+					SOFT_RESET_CPG, 1);
+	WREG32_SOC15(GC, 0, regGRBM_SOFT_RESET, grbm_soft_reset);
+	/**********  exit soft reset  ***********/
+	grbm_soft_reset = RREG32_SOC15(GC, 0, regGRBM_SOFT_RESET);
+	grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET,
+					SOFT_RESET_CP, 0);
+	grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET,
+					SOFT_RESET_GFX, 0);
+	grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET,
+					SOFT_RESET_CPF, 0);
+	grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET,
+					SOFT_RESET_CPC, 0);
+	grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET,
+					SOFT_RESET_CPG, 0);
+	WREG32_SOC15(GC, 0, regGRBM_SOFT_RESET, grbm_soft_reset);
+
+	tmp = RREG32_SOC15(GC, 0, regCP_SOFT_RESET_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_SOFT_RESET_CNTL, CMP_HQD_REG_RESET, 0x1);
+	WREG32_SOC15(GC, 0, regCP_SOFT_RESET_CNTL, tmp);
+
+	WREG32_SOC15(GC, 0, regCP_ME_CNTL, 0x0);
+	WREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL, 0x0);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (!RREG32_SOC15(GC, 0, regCP_VMID_RESET))
+			break;
+		udelay(1);
+	}
+	if (i >= adev->usec_timeout) {
+		printk("Failed to wait CP_VMID_RESET to 0\n");
+		return -EINVAL;
+	}
+
+	tmp = RREG32_SOC15(GC, 0, regCP_INT_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL, CMP_BUSY_INT_ENABLE, 1);
+	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL, CNTX_BUSY_INT_ENABLE, 1);
+	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL, CNTX_EMPTY_INT_ENABLE, 1);
+	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL, GFX_IDLE_INT_ENABLE, 1);
+	WREG32_SOC15(GC, 0, regCP_INT_CNTL, tmp);
+
+	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
+
+	return gfx_v11_0_cp_resume(adev);
+}
+
+static bool gfx_v11_0_check_soft_reset(struct amdgpu_ip_block *ip_block)
+{
+	int i, r;
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	long tmo = msecs_to_jiffies(1000);
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+		ring = &adev->gfx.gfx_ring[i];
+		r = amdgpu_ring_test_ib(ring, tmo);
+		if (r)
+			return true;
+	}
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		ring = &adev->gfx.compute_ring[i];
+		r = amdgpu_ring_test_ib(ring, tmo);
+		if (r)
+			return true;
+	}
+
+	return false;
+}
+
+static int gfx_v11_0_post_soft_reset(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	/**
+	 * GFX soft reset will impact MES, need resume MES when do GFX soft reset
+	 */
+	return amdgpu_mes_resume(adev);
+}
+
+static uint64_t gfx_v11_0_get_gpu_clock_counter(struct amdgpu_device *adev)
+{
+	uint64_t clock;
+	uint64_t clock_counter_lo, clock_counter_hi_pre, clock_counter_hi_after;
+
+	if (amdgpu_sriov_vf(adev)) {
+		amdgpu_gfx_off_ctrl(adev, false);
+		mutex_lock(&adev->gfx.gpu_clock_mutex);
+		clock_counter_hi_pre = (uint64_t)RREG32_SOC15(GC, 0, regCP_MES_MTIME_HI);
+		clock_counter_lo = (uint64_t)RREG32_SOC15(GC, 0, regCP_MES_MTIME_LO);
+		clock_counter_hi_after = (uint64_t)RREG32_SOC15(GC, 0, regCP_MES_MTIME_HI);
+		if (clock_counter_hi_pre != clock_counter_hi_after)
+			clock_counter_lo = (uint64_t)RREG32_SOC15(GC, 0, regCP_MES_MTIME_LO);
+		mutex_unlock(&adev->gfx.gpu_clock_mutex);
+		amdgpu_gfx_off_ctrl(adev, true);
+	} else {
+		preempt_disable();
+		clock_counter_hi_pre = (uint64_t)RREG32_SOC15(SMUIO, 0, regGOLDEN_TSC_COUNT_UPPER);
+		clock_counter_lo = (uint64_t)RREG32_SOC15(SMUIO, 0, regGOLDEN_TSC_COUNT_LOWER);
+		clock_counter_hi_after = (uint64_t)RREG32_SOC15(SMUIO, 0, regGOLDEN_TSC_COUNT_UPPER);
+		if (clock_counter_hi_pre != clock_counter_hi_after)
+			clock_counter_lo = (uint64_t)RREG32_SOC15(SMUIO, 0, regGOLDEN_TSC_COUNT_LOWER);
+		preempt_enable();
+	}
+	clock = clock_counter_lo | (clock_counter_hi_after << 32ULL);
+
+	return clock;
+}
+
+static void gfx_v11_0_ring_emit_gds_switch(struct amdgpu_ring *ring,
+					   uint32_t vmid,
+					   uint32_t gds_base, uint32_t gds_size,
+					   uint32_t gws_base, uint32_t gws_size,
+					   uint32_t oa_base, uint32_t oa_size)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	/* GDS Base */
+	gfx_v11_0_write_data_to_reg(ring, 0, false,
+				    SOC15_REG_OFFSET(GC, 0, regGDS_VMID0_BASE) + 2 * vmid,
+				    gds_base);
+
+	/* GDS Size */
+	gfx_v11_0_write_data_to_reg(ring, 0, false,
+				    SOC15_REG_OFFSET(GC, 0, regGDS_VMID0_SIZE) + 2 * vmid,
+				    gds_size);
+
+	/* GWS */
+	gfx_v11_0_write_data_to_reg(ring, 0, false,
+				    SOC15_REG_OFFSET(GC, 0, regGDS_GWS_VMID0) + vmid,
+				    gws_size << GDS_GWS_VMID0__SIZE__SHIFT | gws_base);
+
+	/* OA */
+	gfx_v11_0_write_data_to_reg(ring, 0, false,
+				    SOC15_REG_OFFSET(GC, 0, regGDS_OA_VMID0) + vmid,
+				    (1 << (oa_size + oa_base)) - (1 << oa_base));
+}
+
+static int gfx_v11_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	switch (amdgpu_user_queue) {
+	case -1:
+	case 0:
+	default:
+		adev->gfx.disable_kq = false;
+		adev->gfx.disable_uq = true;
+		break;
+	case 1:
+		adev->gfx.disable_kq = false;
+		adev->gfx.disable_uq = false;
+		break;
+	case 2:
+		adev->gfx.disable_kq = true;
+		adev->gfx.disable_uq = false;
+		break;
+	}
+
+	adev->gfx.funcs = &gfx_v11_0_gfx_funcs;
+
+	if (adev->gfx.disable_kq) {
+		/* We need one GFX ring temporarily to set up
+		 * the clear state.
+		 */
+		adev->gfx.num_gfx_rings = 1;
+		adev->gfx.num_compute_rings = 0;
+	} else {
+		adev->gfx.num_gfx_rings = GFX11_NUM_GFX_RINGS;
+		adev->gfx.num_compute_rings = min(amdgpu_gfx_get_num_kcq(adev),
+						  AMDGPU_MAX_COMPUTE_RINGS);
+	}
+
+	gfx_v11_0_set_kiq_pm4_funcs(adev);
+	gfx_v11_0_set_ring_funcs(adev);
+	gfx_v11_0_set_irq_funcs(adev);
+	gfx_v11_0_set_gds_init(adev);
+	gfx_v11_0_set_rlc_funcs(adev);
+	gfx_v11_0_set_mqd_funcs(adev);
+	gfx_v11_0_set_imu_funcs(adev);
+
+	gfx_v11_0_init_rlcg_reg_access_ctrl(adev);
+
+	return gfx_v11_0_init_microcode(adev);
+}
+
+static int gfx_v11_0_late_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = amdgpu_irq_get(adev, &adev->gfx.priv_reg_irq, 0);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_get(adev, &adev->gfx.priv_inst_irq, 0);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_get(adev, &adev->gfx.bad_op_irq, 0);
+	if (r)
+		return r;
+
+	r = gfx_v11_0_set_userq_eop_interrupts(adev, true);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static bool gfx_v11_0_is_rlc_enabled(struct amdgpu_device *adev)
+{
+	uint32_t rlc_cntl;
+
+	/* if RLC is not enabled, do nothing */
+	rlc_cntl = RREG32_SOC15(GC, 0, regRLC_CNTL);
+	return (REG_GET_FIELD(rlc_cntl, RLC_CNTL, RLC_ENABLE_F32)) ? true : false;
+}
+
+static void gfx_v11_0_set_safe_mode(struct amdgpu_device *adev, int xcc_id)
+{
+	uint32_t data;
+	unsigned i;
+
+	data = RLC_SAFE_MODE__CMD_MASK;
+	data |= (1 << RLC_SAFE_MODE__MESSAGE__SHIFT);
+
+	WREG32_SOC15(GC, 0, regRLC_SAFE_MODE, data);
+
+	/* wait for RLC_SAFE_MODE */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (!REG_GET_FIELD(RREG32_SOC15(GC, 0, regRLC_SAFE_MODE),
+				   RLC_SAFE_MODE, CMD))
+			break;
+		udelay(1);
+	}
+}
+
+static void gfx_v11_0_unset_safe_mode(struct amdgpu_device *adev, int xcc_id)
+{
+	WREG32_SOC15(GC, 0, regRLC_SAFE_MODE, RLC_SAFE_MODE__CMD_MASK);
+}
+
+static void gfx_v11_0_update_perf_clk(struct amdgpu_device *adev,
+				      bool enable)
+{
+	uint32_t def, data;
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_GFX_PERF_CLK))
+		return;
+
+	def = data = RREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE);
+
+	if (enable)
+		data &= ~RLC_CGTT_MGCG_OVERRIDE__PERFMON_CLOCK_STATE_MASK;
+	else
+		data |= RLC_CGTT_MGCG_OVERRIDE__PERFMON_CLOCK_STATE_MASK;
+
+	if (def != data)
+		WREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE, data);
+}
+
+static void gfx_v11_0_update_sram_fgcg(struct amdgpu_device *adev,
+				       bool enable)
+{
+	uint32_t def, data;
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_GFX_FGCG))
+		return;
+
+	def = data = RREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE);
+
+	if (enable)
+		data &= ~RLC_CGTT_MGCG_OVERRIDE__GFXIP_FGCG_OVERRIDE_MASK;
+	else
+		data |= RLC_CGTT_MGCG_OVERRIDE__GFXIP_FGCG_OVERRIDE_MASK;
+
+	if (def != data)
+		WREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE, data);
+}
+
+static void gfx_v11_0_update_repeater_fgcg(struct amdgpu_device *adev,
+					   bool enable)
+{
+	uint32_t def, data;
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_REPEATER_FGCG))
+		return;
+
+	def = data = RREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE);
+
+	if (enable)
+		data &= ~RLC_CGTT_MGCG_OVERRIDE__GFXIP_REPEATER_FGCG_OVERRIDE_MASK;
+	else
+		data |= RLC_CGTT_MGCG_OVERRIDE__GFXIP_REPEATER_FGCG_OVERRIDE_MASK;
+
+	if (def != data)
+		WREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE, data);
+}
+
+static void gfx_v11_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						       bool enable)
+{
+	uint32_t data, def;
+
+	if (!(adev->cg_flags & (AMD_CG_SUPPORT_GFX_MGCG | AMD_CG_SUPPORT_GFX_MGLS)))
+		return;
+
+	/* It is disabled by HW by default */
+	if (enable) {
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGCG) {
+			/* 1 - RLC_CGTT_MGCG_OVERRIDE */
+			def = data = RREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE);
+
+			data &= ~(RLC_CGTT_MGCG_OVERRIDE__GRBM_CGTT_SCLK_OVERRIDE_MASK |
+				  RLC_CGTT_MGCG_OVERRIDE__RLC_CGTT_SCLK_OVERRIDE_MASK |
+				  RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGCG_OVERRIDE_MASK);
+
+			if (def != data)
+				WREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE, data);
+		}
+	} else {
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGCG) {
+			def = data = RREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE);
+
+			data |= (RLC_CGTT_MGCG_OVERRIDE__RLC_CGTT_SCLK_OVERRIDE_MASK |
+				 RLC_CGTT_MGCG_OVERRIDE__GRBM_CGTT_SCLK_OVERRIDE_MASK |
+				 RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGCG_OVERRIDE_MASK);
+
+			if (def != data)
+				WREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE, data);
+		}
+	}
+}
+
+static void gfx_v11_0_update_coarse_grain_clock_gating(struct amdgpu_device *adev,
+						       bool enable)
+{
+	uint32_t def, data;
+
+	if (!(adev->cg_flags &
+	      (AMD_CG_SUPPORT_GFX_CGCG |
+	      AMD_CG_SUPPORT_GFX_CGLS |
+	      AMD_CG_SUPPORT_GFX_3D_CGCG |
+	      AMD_CG_SUPPORT_GFX_3D_CGLS)))
+		return;
+
+	if (enable) {
+		def = data = RREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE);
+
+		/* unset CGCG override */
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGCG)
+			data &= ~RLC_CGTT_MGCG_OVERRIDE__GFXIP_CGCG_OVERRIDE_MASK;
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGLS)
+			data &= ~RLC_CGTT_MGCG_OVERRIDE__GFXIP_CGLS_OVERRIDE_MASK;
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGCG ||
+		    adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGLS)
+			data &= ~RLC_CGTT_MGCG_OVERRIDE__GFXIP_GFX3D_CG_OVERRIDE_MASK;
+
+		/* update CGCG override bits */
+		if (def != data)
+			WREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE, data);
+
+		/* enable cgcg FSM(0x0000363F) */
+		def = data = RREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL);
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGCG) {
+			data &= ~RLC_CGCG_CGLS_CTRL__CGCG_GFX_IDLE_THRESHOLD_MASK;
+			data |= (0x36 << RLC_CGCG_CGLS_CTRL__CGCG_GFX_IDLE_THRESHOLD__SHIFT) |
+				 RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK;
+		}
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGLS) {
+			data &= ~RLC_CGCG_CGLS_CTRL__CGLS_REP_COMPANSAT_DELAY_MASK;
+			data |= (0x000F << RLC_CGCG_CGLS_CTRL__CGLS_REP_COMPANSAT_DELAY__SHIFT) |
+				 RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK;
+		}
+
+		if (def != data)
+			WREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL, data);
+
+		/* Program RLC_CGCG_CGLS_CTRL_3D */
+		def = data = RREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL_3D);
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGCG) {
+			data &= ~RLC_CGCG_CGLS_CTRL_3D__CGCG_GFX_IDLE_THRESHOLD_MASK;
+			data |= (0x36 << RLC_CGCG_CGLS_CTRL_3D__CGCG_GFX_IDLE_THRESHOLD__SHIFT) |
+				 RLC_CGCG_CGLS_CTRL_3D__CGCG_EN_MASK;
+		}
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGLS) {
+			data &= ~RLC_CGCG_CGLS_CTRL_3D__CGLS_REP_COMPANSAT_DELAY_MASK;
+			data |= (0xf << RLC_CGCG_CGLS_CTRL_3D__CGLS_REP_COMPANSAT_DELAY__SHIFT) |
+				 RLC_CGCG_CGLS_CTRL_3D__CGLS_EN_MASK;
+		}
+
+		if (def != data)
+			WREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL_3D, data);
+
+		/* set IDLE_POLL_COUNT(0x00900100) */
+		def = data = RREG32_SOC15(GC, 0, regCP_RB_WPTR_POLL_CNTL);
+
+		data &= ~(CP_RB_WPTR_POLL_CNTL__POLL_FREQUENCY_MASK | CP_RB_WPTR_POLL_CNTL__IDLE_POLL_COUNT_MASK);
+		data |= (0x0100 << CP_RB_WPTR_POLL_CNTL__POLL_FREQUENCY__SHIFT) |
+			(0x0090 << CP_RB_WPTR_POLL_CNTL__IDLE_POLL_COUNT__SHIFT);
+
+		if (def != data)
+			WREG32_SOC15(GC, 0, regCP_RB_WPTR_POLL_CNTL, data);
+
+		data = RREG32_SOC15(GC, 0, regCP_INT_CNTL);
+		data = REG_SET_FIELD(data, CP_INT_CNTL, CNTX_BUSY_INT_ENABLE, 1);
+		data = REG_SET_FIELD(data, CP_INT_CNTL, CNTX_EMPTY_INT_ENABLE, 1);
+		data = REG_SET_FIELD(data, CP_INT_CNTL, CMP_BUSY_INT_ENABLE, 1);
+		data = REG_SET_FIELD(data, CP_INT_CNTL, GFX_IDLE_INT_ENABLE, 1);
+		WREG32_SOC15(GC, 0, regCP_INT_CNTL, data);
+
+		data = RREG32_SOC15(GC, 0, regSDMA0_RLC_CGCG_CTRL);
+		data = REG_SET_FIELD(data, SDMA0_RLC_CGCG_CTRL, CGCG_INT_ENABLE, 1);
+		WREG32_SOC15(GC, 0, regSDMA0_RLC_CGCG_CTRL, data);
+
+		/* Some ASICs only have one SDMA instance, not need to configure SDMA1 */
+		if (adev->sdma.num_instances > 1) {
+			data = RREG32_SOC15(GC, 0, regSDMA1_RLC_CGCG_CTRL);
+			data = REG_SET_FIELD(data, SDMA1_RLC_CGCG_CTRL, CGCG_INT_ENABLE, 1);
+			WREG32_SOC15(GC, 0, regSDMA1_RLC_CGCG_CTRL, data);
+		}
+	} else {
+		/* Program RLC_CGCG_CGLS_CTRL */
+		def = data = RREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL);
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGCG)
+			data &= ~RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK;
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGLS)
+			data &= ~RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK;
+
+		if (def != data)
+			WREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL, data);
+
+		/* Program RLC_CGCG_CGLS_CTRL_3D */
+		def = data = RREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL_3D);
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGCG)
+			data &= ~RLC_CGCG_CGLS_CTRL_3D__CGCG_EN_MASK;
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGLS)
+			data &= ~RLC_CGCG_CGLS_CTRL_3D__CGLS_EN_MASK;
+
+		if (def != data)
+			WREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL_3D, data);
+
+		data = RREG32_SOC15(GC, 0, regSDMA0_RLC_CGCG_CTRL);
+		data &= ~SDMA0_RLC_CGCG_CTRL__CGCG_INT_ENABLE_MASK;
+		WREG32_SOC15(GC, 0, regSDMA0_RLC_CGCG_CTRL, data);
+
+		/* Some ASICs only have one SDMA instance, not need to configure SDMA1 */
+		if (adev->sdma.num_instances > 1) {
+			data = RREG32_SOC15(GC, 0, regSDMA1_RLC_CGCG_CTRL);
+			data &= ~SDMA1_RLC_CGCG_CTRL__CGCG_INT_ENABLE_MASK;
+			WREG32_SOC15(GC, 0, regSDMA1_RLC_CGCG_CTRL, data);
+		}
+	}
+}
+
+static int gfx_v11_0_update_gfx_clock_gating(struct amdgpu_device *adev,
+					    bool enable)
+{
+	amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
+
+	gfx_v11_0_update_coarse_grain_clock_gating(adev, enable);
+
+	gfx_v11_0_update_medium_grain_clock_gating(adev, enable);
+
+	gfx_v11_0_update_repeater_fgcg(adev, enable);
+
+	gfx_v11_0_update_sram_fgcg(adev, enable);
+
+	gfx_v11_0_update_perf_clk(adev, enable);
+
+	if (adev->cg_flags &
+	    (AMD_CG_SUPPORT_GFX_MGCG |
+	     AMD_CG_SUPPORT_GFX_CGLS |
+	     AMD_CG_SUPPORT_GFX_CGCG |
+	     AMD_CG_SUPPORT_GFX_3D_CGCG |
+	     AMD_CG_SUPPORT_GFX_3D_CGLS))
+	        gfx_v11_0_enable_gui_idle_interrupt(adev, enable);
+
+	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
+
+	return 0;
+}
+
+static void gfx_v11_0_update_spm_vmid(struct amdgpu_device *adev, struct amdgpu_ring *ring, unsigned vmid)
+{
+	u32 reg, pre_data, data;
+
+	amdgpu_gfx_off_ctrl(adev, false);
+	reg = SOC15_REG_OFFSET(GC, 0, regRLC_SPM_MC_CNTL);
+	if (amdgpu_sriov_is_pp_one_vf(adev) && !amdgpu_sriov_runtime(adev))
+		pre_data = RREG32_NO_KIQ(reg);
+	else
+		pre_data = RREG32(reg);
+
+	data = pre_data & (~RLC_SPM_MC_CNTL__RLC_SPM_VMID_MASK);
+	data |= (vmid & RLC_SPM_MC_CNTL__RLC_SPM_VMID_MASK) << RLC_SPM_MC_CNTL__RLC_SPM_VMID__SHIFT;
+
+	if (pre_data != data) {
+		if (amdgpu_sriov_is_pp_one_vf(adev) && !amdgpu_sriov_runtime(adev)) {
+			WREG32_SOC15_NO_KIQ(GC, 0, regRLC_SPM_MC_CNTL, data);
+		} else
+			WREG32_SOC15(GC, 0, regRLC_SPM_MC_CNTL, data);
+	}
+	amdgpu_gfx_off_ctrl(adev, true);
+
+	if (ring
+		&& amdgpu_sriov_is_pp_one_vf(adev)
+		&& (pre_data != data)
+		&& ((ring->funcs->type == AMDGPU_RING_TYPE_GFX)
+			|| (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE))) {
+		amdgpu_ring_emit_wreg(ring, reg, data);
+	}
+}
+
+static const struct amdgpu_rlc_funcs gfx_v11_0_rlc_funcs = {
+	.is_rlc_enabled = gfx_v11_0_is_rlc_enabled,
+	.set_safe_mode = gfx_v11_0_set_safe_mode,
+	.unset_safe_mode = gfx_v11_0_unset_safe_mode,
+	.init = gfx_v11_0_rlc_init,
+	.get_csb_size = gfx_v11_0_get_csb_size,
+	.get_csb_buffer = gfx_v11_0_get_csb_buffer,
+	.resume = gfx_v11_0_rlc_resume,
+	.stop = gfx_v11_0_rlc_stop,
+	.reset = gfx_v11_0_rlc_reset,
+	.start = gfx_v11_0_rlc_start,
+	.update_spm_vmid = gfx_v11_0_update_spm_vmid,
+};
+
+static void gfx_v11_cntl_power_gating(struct amdgpu_device *adev, bool enable)
+{
+	u32 data = RREG32_SOC15(GC, 0, regRLC_PG_CNTL);
+
+	if (enable && (adev->pg_flags & AMD_PG_SUPPORT_GFX_PG))
+		data |= RLC_PG_CNTL__GFX_POWER_GATING_ENABLE_MASK;
+	else
+		data &= ~RLC_PG_CNTL__GFX_POWER_GATING_ENABLE_MASK;
+
+	WREG32_SOC15(GC, 0, regRLC_PG_CNTL, data);
+
+	// Program RLC_PG_DELAY3 for CGPG hysteresis
+	if (enable && (adev->pg_flags & AMD_PG_SUPPORT_GFX_PG)) {
+		switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+		case IP_VERSION(11, 0, 1):
+		case IP_VERSION(11, 0, 4):
+		case IP_VERSION(11, 5, 0):
+		case IP_VERSION(11, 5, 1):
+		case IP_VERSION(11, 5, 2):
+		case IP_VERSION(11, 5, 3):
+			WREG32_SOC15(GC, 0, regRLC_PG_DELAY_3, RLC_PG_DELAY_3_DEFAULT_GC_11_0_1);
+			break;
+		default:
+			break;
+		}
+	}
+}
+
+static void gfx_v11_cntl_pg(struct amdgpu_device *adev, bool enable)
+{
+	amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
+
+	gfx_v11_cntl_power_gating(adev, enable);
+
+	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
+}
+
+static int gfx_v11_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					   enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = (state == AMD_PG_STATE_GATE);
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 2):
+	case IP_VERSION(11, 0, 3):
+		amdgpu_gfx_off_ctrl(adev, enable);
+		break;
+	case IP_VERSION(11, 0, 1):
+	case IP_VERSION(11, 0, 4):
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 1):
+	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
+		if (!enable)
+			amdgpu_gfx_off_ctrl(adev, false);
+
+		gfx_v11_cntl_pg(adev, enable);
+
+		if (enable)
+			amdgpu_gfx_off_ctrl(adev, true);
+
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gfx_v11_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_sriov_vf(adev))
+	        return 0;
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 1):
+	case IP_VERSION(11, 0, 2):
+	case IP_VERSION(11, 0, 3):
+	case IP_VERSION(11, 0, 4):
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 1):
+	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
+	        gfx_v11_0_update_gfx_clock_gating(adev,
+	                        state ==  AMD_CG_STATE_GATE);
+	        break;
+	default:
+	        break;
+	}
+
+	return 0;
+}
+
+static void gfx_v11_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int data;
+
+	/* AMD_CG_SUPPORT_GFX_MGCG */
+	data = RREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE);
+	if (!(data & RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGCG_OVERRIDE_MASK))
+		*flags |= AMD_CG_SUPPORT_GFX_MGCG;
+
+	/* AMD_CG_SUPPORT_REPEATER_FGCG */
+	if (!(data & RLC_CGTT_MGCG_OVERRIDE__GFXIP_REPEATER_FGCG_OVERRIDE_MASK))
+		*flags |= AMD_CG_SUPPORT_REPEATER_FGCG;
+
+	/* AMD_CG_SUPPORT_GFX_FGCG */
+	if (!(data & RLC_CGTT_MGCG_OVERRIDE__GFXIP_FGCG_OVERRIDE_MASK))
+		*flags |= AMD_CG_SUPPORT_GFX_FGCG;
+
+	/* AMD_CG_SUPPORT_GFX_PERF_CLK */
+	if (!(data & RLC_CGTT_MGCG_OVERRIDE__PERFMON_CLOCK_STATE_MASK))
+		*flags |= AMD_CG_SUPPORT_GFX_PERF_CLK;
+
+	/* AMD_CG_SUPPORT_GFX_CGCG */
+	data = RREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL);
+	if (data & RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_CGCG;
+
+	/* AMD_CG_SUPPORT_GFX_CGLS */
+	if (data & RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_CGLS;
+
+	/* AMD_CG_SUPPORT_GFX_3D_CGCG */
+	data = RREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL_3D);
+	if (data & RLC_CGCG_CGLS_CTRL_3D__CGCG_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_3D_CGCG;
+
+	/* AMD_CG_SUPPORT_GFX_3D_CGLS */
+	if (data & RLC_CGCG_CGLS_CTRL_3D__CGLS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_3D_CGLS;
+}
+
+static u64 gfx_v11_0_ring_get_rptr_gfx(struct amdgpu_ring *ring)
+{
+	/* gfx11 is 32bit rptr*/
+	return *(uint32_t *)ring->rptr_cpu_addr;
+}
+
+static u64 gfx_v11_0_ring_get_wptr_gfx(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u64 wptr;
+
+	/* XXX check if swapping is necessary on BE */
+	if (ring->use_doorbell) {
+		wptr = atomic64_read((atomic64_t *)ring->wptr_cpu_addr);
+	} else {
+		wptr = RREG32_SOC15(GC, 0, regCP_RB0_WPTR);
+		wptr += (u64)RREG32_SOC15(GC, 0, regCP_RB0_WPTR_HI) << 32;
+	}
+
+	return wptr;
+}
+
+static void gfx_v11_0_ring_set_wptr_gfx(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr,
+			     ring->wptr);
+		WDOORBELL64(ring->doorbell_index, ring->wptr);
+	} else {
+		WREG32_SOC15(GC, 0, regCP_RB0_WPTR,
+			     lower_32_bits(ring->wptr));
+		WREG32_SOC15(GC, 0, regCP_RB0_WPTR_HI,
+			     upper_32_bits(ring->wptr));
+	}
+}
+
+static u64 gfx_v11_0_ring_get_rptr_compute(struct amdgpu_ring *ring)
+{
+	/* gfx11 hardware is 32bit rptr */
+	return *(uint32_t *)ring->rptr_cpu_addr;
+}
+
+static u64 gfx_v11_0_ring_get_wptr_compute(struct amdgpu_ring *ring)
+{
+	u64 wptr;
+
+	/* XXX check if swapping is necessary on BE */
+	if (ring->use_doorbell)
+		wptr = atomic64_read((atomic64_t *)ring->wptr_cpu_addr);
+	else
+		BUG();
+	return wptr;
+}
+
+static void gfx_v11_0_ring_set_wptr_compute(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	/* XXX check if swapping is necessary on BE */
+	if (ring->use_doorbell) {
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr,
+			     ring->wptr);
+		WDOORBELL64(ring->doorbell_index, ring->wptr);
+	} else {
+		BUG(); /* only DOORBELL method supported on gfx11 now */
+	}
+}
+
+static void gfx_v11_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 ref_and_mask, reg_mem_engine;
+	const struct nbio_hdp_flush_reg *nbio_hf_reg = adev->nbio.hdp_flush_reg;
+
+	if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE) {
+		switch (ring->me) {
+		case 1:
+			ref_and_mask = nbio_hf_reg->ref_and_mask_cp2 << ring->pipe;
+			break;
+		case 2:
+			ref_and_mask = nbio_hf_reg->ref_and_mask_cp6 << ring->pipe;
+			break;
+		default:
+			return;
+		}
+		reg_mem_engine = 0;
+	} else {
+		ref_and_mask = nbio_hf_reg->ref_and_mask_cp0 << ring->pipe;
+		reg_mem_engine = 1; /* pfp */
+	}
+
+	gfx_v11_0_wait_reg_mem(ring, reg_mem_engine, 0, 1,
+			       adev->nbio.funcs->get_hdp_flush_req_offset(adev),
+			       adev->nbio.funcs->get_hdp_flush_done_offset(adev),
+			       ref_and_mask, ref_and_mask, 0x20);
+}
+
+static void gfx_v11_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
+				       struct amdgpu_job *job,
+				       struct amdgpu_ib *ib,
+				       uint32_t flags)
+{
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+	u32 header, control = 0;
+
+	BUG_ON(ib->flags & AMDGPU_IB_FLAG_CE);
+
+	header = PACKET3(PACKET3_INDIRECT_BUFFER, 2);
+
+	control |= ib->length_dw | (vmid << 24);
+
+	if (ring->adev->gfx.mcbp && (ib->flags & AMDGPU_IB_FLAG_PREEMPT)) {
+		control |= INDIRECT_BUFFER_PRE_ENB(1);
+
+		if (flags & AMDGPU_IB_PREEMPTED)
+			control |= INDIRECT_BUFFER_PRE_RESUME(1);
+
+		if (vmid)
+			gfx_v11_0_ring_emit_de_meta(ring,
+				    (!amdgpu_sriov_vf(ring->adev) && flags & AMDGPU_IB_PREEMPTED) ? true : false);
+	}
+
+	amdgpu_ring_write(ring, header);
+	BUG_ON(ib->gpu_addr & 0x3); /* Dword align */
+	amdgpu_ring_write(ring,
+#ifdef __BIG_ENDIAN
+		(2 << 0) |
+#endif
+		lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, control);
+}
+
+static void gfx_v11_0_ring_emit_ib_compute(struct amdgpu_ring *ring,
+					   struct amdgpu_job *job,
+					   struct amdgpu_ib *ib,
+					   uint32_t flags)
+{
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+	u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vmid << 24);
+
+	/* Currently, there is a high possibility to get wave ID mismatch
+	 * between ME and GDS, leading to a hw deadlock, because ME generates
+	 * different wave IDs than the GDS expects. This situation happens
+	 * randomly when at least 5 compute pipes use GDS ordered append.
+	 * The wave IDs generated by ME are also wrong after suspend/resume.
+	 * Those are probably bugs somewhere else in the kernel driver.
+	 *
+	 * Writing GDS_COMPUTE_MAX_WAVE_ID resets wave ID counters in ME and
+	 * GDS to 0 for this ring (me/pipe).
+	 */
+	if (ib->flags & AMDGPU_IB_FLAG_RESET_GDS_MAX_WAVE_ID) {
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1));
+		amdgpu_ring_write(ring, regGDS_COMPUTE_MAX_WAVE_ID);
+		amdgpu_ring_write(ring, ring->adev->gds.gds_compute_max_wave_id);
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2));
+	BUG_ON(ib->gpu_addr & 0x3); /* Dword align */
+	amdgpu_ring_write(ring,
+#ifdef __BIG_ENDIAN
+				(2 << 0) |
+#endif
+				lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, control);
+}
+
+static void gfx_v11_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
+				     u64 seq, unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	bool int_sel = flags & AMDGPU_FENCE_FLAG_INT;
+
+	/* RELEASE_MEM - flush caches, send int */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_RELEASE_MEM, 6));
+	amdgpu_ring_write(ring, (PACKET3_RELEASE_MEM_GCR_SEQ |
+				 PACKET3_RELEASE_MEM_GCR_GL2_WB |
+				 PACKET3_RELEASE_MEM_GCR_GLM_INV | /* must be set with GLM_WB */
+				 PACKET3_RELEASE_MEM_GCR_GLM_WB |
+				 PACKET3_RELEASE_MEM_CACHE_POLICY(3) |
+				 PACKET3_RELEASE_MEM_EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) |
+				 PACKET3_RELEASE_MEM_EVENT_INDEX(5)));
+	amdgpu_ring_write(ring, (PACKET3_RELEASE_MEM_DATA_SEL(write64bit ? 2 : 1) |
+				 PACKET3_RELEASE_MEM_INT_SEL(int_sel ? 2 : 0)));
+
+	/*
+	 * the address should be Qword aligned if 64bit write, Dword
+	 * aligned if only send 32bit data low (discard data high)
+	 */
+	if (write64bit)
+		BUG_ON(addr & 0x7);
+	else
+		BUG_ON(addr & 0x3);
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+	amdgpu_ring_write(ring, upper_32_bits(seq));
+	amdgpu_ring_write(ring, 0);
+}
+
+static void gfx_v11_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	gfx_v11_0_wait_reg_mem(ring, usepfp, 1, 0, lower_32_bits(addr),
+			       upper_32_bits(addr), seq, 0xffffffff, 4);
+}
+
+static void gfx_v11_0_ring_invalidate_tlbs(struct amdgpu_ring *ring,
+				   uint16_t pasid, uint32_t flush_type,
+				   bool all_hub, uint8_t dst_sel)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_INVALIDATE_TLBS, 0));
+	amdgpu_ring_write(ring,
+			  PACKET3_INVALIDATE_TLBS_DST_SEL(dst_sel) |
+			  PACKET3_INVALIDATE_TLBS_ALL_HUB(all_hub) |
+			  PACKET3_INVALIDATE_TLBS_PASID(pasid) |
+			  PACKET3_INVALIDATE_TLBS_FLUSH_TYPE(flush_type));
+}
+
+static void gfx_v11_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					 unsigned vmid, uint64_t pd_addr)
+{
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* compute doesn't have PFP */
+	if (ring->funcs->type == AMDGPU_RING_TYPE_GFX) {
+		/* sync PFP to ME, otherwise we might get invalid PFP reads */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_PFP_SYNC_ME, 0));
+		amdgpu_ring_write(ring, 0x0);
+	}
+
+	/* Make sure that we can't skip the SET_Q_MODE packets when the VM
+	 * changed in any way.
+	 */
+	ring->set_q_mode_offs = 0;
+	ring->set_q_mode_ptr = NULL;
+}
+
+static void gfx_v11_0_ring_emit_fence_kiq(struct amdgpu_ring *ring, u64 addr,
+					  u64 seq, unsigned int flags)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	/* we only allocate 32bit for each seq wb address */
+	BUG_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	/* write fence seq to the "addr" */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				 WRITE_DATA_DST_SEL(5) | WR_CONFIRM));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+
+	if (flags & AMDGPU_FENCE_FLAG_INT) {
+		/* set register to trigger INT */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+		amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+					 WRITE_DATA_DST_SEL(0) | WR_CONFIRM));
+		amdgpu_ring_write(ring, SOC15_REG_OFFSET(GC, 0, regCPC_INT_STATUS));
+		amdgpu_ring_write(ring, 0);
+		amdgpu_ring_write(ring, 0x20000000); /* src_id is 178 */
+	}
+}
+
+static void gfx_v11_0_ring_emit_cntxcntl(struct amdgpu_ring *ring,
+					 uint32_t flags)
+{
+	uint32_t dw2 = 0;
+
+	dw2 |= 0x80000000; /* set load_enable otherwise this package is just NOPs */
+	if (flags & AMDGPU_HAVE_CTX_SWITCH) {
+		/* set load_global_config & load_global_uconfig */
+		dw2 |= 0x8001;
+		/* set load_cs_sh_regs */
+		dw2 |= 0x01000000;
+		/* set load_per_context_state & load_gfx_sh_regs for GFX */
+		dw2 |= 0x10002;
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	amdgpu_ring_write(ring, dw2);
+	amdgpu_ring_write(ring, 0);
+}
+
+static unsigned gfx_v11_0_ring_emit_init_cond_exec(struct amdgpu_ring *ring,
+						   uint64_t addr)
+{
+	unsigned ret;
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_COND_EXEC, 3));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	/* discard following DWs if *cond_exec_gpu_addr==0 */
+	amdgpu_ring_write(ring, 0);
+	ret = ring->wptr & ring->buf_mask;
+	/* patch dummy value later */
+	amdgpu_ring_write(ring, 0);
+
+	return ret;
+}
+
+static void gfx_v11_0_ring_emit_gfx_shadow(struct amdgpu_ring *ring,
+					   u64 shadow_va, u64 csa_va,
+					   u64 gds_va, bool init_shadow,
+					   int vmid)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned int offs, end;
+
+	if (!adev->gfx.cp_gfx_shadow || !ring->ring_obj)
+		return;
+
+	/*
+	 * The logic here isn't easy to understand because we need to keep state
+	 * accross multiple executions of the function as well as between the
+	 * CPU and GPU. The general idea is that the newly written GPU command
+	 * has a condition on the previous one and only executed if really
+	 * necessary.
+	 */
+
+	/*
+	 * The dw in the NOP controls if the next SET_Q_MODE packet should be
+	 * executed or not. Reserve 64bits just to be on the save side.
+	 */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_NOP, 1));
+	offs = ring->wptr & ring->buf_mask;
+
+	/*
+	 * We start with skipping the prefix SET_Q_MODE and always executing
+	 * the postfix SET_Q_MODE packet. This is changed below with a
+	 * WRITE_DATA command when the postfix executed.
+	 */
+	amdgpu_ring_write(ring, shadow_va ? 1 : 0);
+	amdgpu_ring_write(ring, 0);
+
+	if (ring->set_q_mode_offs) {
+		uint64_t addr;
+
+		addr = amdgpu_bo_gpu_offset(ring->ring_obj);
+		addr += ring->set_q_mode_offs << 2;
+		end = gfx_v11_0_ring_emit_init_cond_exec(ring, addr);
+	}
+
+	/*
+	 * When the postfix SET_Q_MODE packet executes we need to make sure that the
+	 * next prefix SET_Q_MODE packet executes as well.
+	 */
+	if (!shadow_va) {
+		uint64_t addr;
+
+		addr = amdgpu_bo_gpu_offset(ring->ring_obj);
+		addr += offs << 2;
+		amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+		amdgpu_ring_write(ring, WRITE_DATA_DST_SEL(5) | WR_CONFIRM);
+		amdgpu_ring_write(ring, lower_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(addr));
+		amdgpu_ring_write(ring, 0x1);
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_Q_PREEMPTION_MODE, 7));
+	amdgpu_ring_write(ring, lower_32_bits(shadow_va));
+	amdgpu_ring_write(ring, upper_32_bits(shadow_va));
+	amdgpu_ring_write(ring, lower_32_bits(gds_va));
+	amdgpu_ring_write(ring, upper_32_bits(gds_va));
+	amdgpu_ring_write(ring, lower_32_bits(csa_va));
+	amdgpu_ring_write(ring, upper_32_bits(csa_va));
+	amdgpu_ring_write(ring, shadow_va ?
+			  PACKET3_SET_Q_PREEMPTION_MODE_IB_VMID(vmid) : 0);
+	amdgpu_ring_write(ring, init_shadow ?
+			  PACKET3_SET_Q_PREEMPTION_MODE_INIT_SHADOW_MEM : 0);
+
+	if (ring->set_q_mode_offs)
+		amdgpu_ring_patch_cond_exec(ring, end);
+
+	if (shadow_va) {
+		uint64_t token = shadow_va ^ csa_va ^ gds_va ^ vmid;
+
+		/*
+		 * If the tokens match try to skip the last postfix SET_Q_MODE
+		 * packet to avoid saving/restoring the state all the time.
+		 */
+		if (ring->set_q_mode_ptr && ring->set_q_mode_token == token)
+			*ring->set_q_mode_ptr = 0;
+
+		ring->set_q_mode_token = token;
+	} else {
+		ring->set_q_mode_ptr = &ring->ring[ring->set_q_mode_offs];
+	}
+
+	ring->set_q_mode_offs = offs;
+}
+
+static int gfx_v11_0_ring_preempt_ib(struct amdgpu_ring *ring)
+{
+	int i, r = 0;
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[0];
+	struct amdgpu_ring *kiq_ring = &kiq->ring;
+	unsigned long flags;
+
+	if (adev->enable_mes)
+		return -EINVAL;
+
+	if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
+		return -EINVAL;
+
+	spin_lock_irqsave(&kiq->ring_lock, flags);
+
+	if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->unmap_queues_size)) {
+		spin_unlock_irqrestore(&kiq->ring_lock, flags);
+		return -ENOMEM;
+	}
+
+	/* assert preemption condition */
+	amdgpu_ring_set_preempt_cond_exec(ring, false);
+
+	/* assert IB preemption, emit the trailing fence */
+	kiq->pmf->kiq_unmap_queues(kiq_ring, ring, PREEMPT_QUEUES_NO_UNMAP,
+				   ring->trail_fence_gpu_addr,
+				   ++ring->trail_seq);
+	amdgpu_ring_commit(kiq_ring);
+
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+
+	/* poll the trailing fence */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (ring->trail_seq ==
+		    le32_to_cpu(*(ring->trail_fence_cpu_addr)))
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout) {
+		r = -EINVAL;
+		DRM_ERROR("ring %d failed to preempt ib\n", ring->idx);
+	}
+
+	/* deassert preemption condition */
+	amdgpu_ring_set_preempt_cond_exec(ring, true);
+	return r;
+}
+
+static void gfx_v11_0_ring_emit_de_meta(struct amdgpu_ring *ring, bool resume)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v10_de_ib_state de_payload = {0};
+	uint64_t offset, gds_addr, de_payload_gpu_addr;
+	void *de_payload_cpu_addr;
+	int cnt;
+
+	offset = offsetof(struct v10_gfx_meta_data, de_payload);
+	de_payload_gpu_addr = amdgpu_csa_vaddr(ring->adev) + offset;
+	de_payload_cpu_addr = adev->virt.csa_cpu_addr + offset;
+
+	gds_addr = ALIGN(amdgpu_csa_vaddr(ring->adev) +
+			 AMDGPU_CSA_SIZE - adev->gds.gds_size,
+			 PAGE_SIZE);
+
+	de_payload.gds_backup_addrlo = lower_32_bits(gds_addr);
+	de_payload.gds_backup_addrhi = upper_32_bits(gds_addr);
+
+	cnt = (sizeof(de_payload) >> 2) + 4 - 2;
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, cnt));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(1) |
+				 WRITE_DATA_DST_SEL(8) |
+				 WR_CONFIRM) |
+				 WRITE_DATA_CACHE_POLICY(0));
+	amdgpu_ring_write(ring, lower_32_bits(de_payload_gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(de_payload_gpu_addr));
+
+	if (resume)
+		amdgpu_ring_write_multiple(ring, de_payload_cpu_addr,
+					   sizeof(de_payload) >> 2);
+	else
+		amdgpu_ring_write_multiple(ring, (void *)&de_payload,
+					   sizeof(de_payload) >> 2);
+}
+
+static void gfx_v11_0_ring_emit_frame_cntl(struct amdgpu_ring *ring, bool start,
+				    bool secure)
+{
+	uint32_t v = secure ? FRAME_TMZ : 0;
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_FRAME_CONTROL, 0));
+	amdgpu_ring_write(ring, v | FRAME_CMD(start ? 0 : 1));
+}
+
+static void gfx_v11_0_ring_emit_rreg(struct amdgpu_ring *ring, uint32_t reg,
+				     uint32_t reg_val_offs)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_COPY_DATA, 4));
+	amdgpu_ring_write(ring, 0 |	/* src: register*/
+				(5 << 8) |	/* dst: memory */
+				(1 << 20));	/* write confirm */
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, lower_32_bits(adev->wb.gpu_addr +
+				reg_val_offs * 4));
+	amdgpu_ring_write(ring, upper_32_bits(adev->wb.gpu_addr +
+				reg_val_offs * 4));
+}
+
+static void gfx_v11_0_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg,
+				   uint32_t val)
+{
+	uint32_t cmd = 0;
+
+	switch (ring->funcs->type) {
+	case AMDGPU_RING_TYPE_GFX:
+		cmd = WRITE_DATA_ENGINE_SEL(1) | WR_CONFIRM;
+		break;
+	case AMDGPU_RING_TYPE_KIQ:
+		cmd = (1 << 16); /* no inc addr */
+		break;
+	default:
+		cmd = WR_CONFIRM;
+		break;
+	}
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, cmd);
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, val);
+}
+
+static void gfx_v11_0_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+					uint32_t val, uint32_t mask)
+{
+	gfx_v11_0_wait_reg_mem(ring, 0, 0, 0, reg, 0, val, mask, 0x20);
+}
+
+static void gfx_v11_0_ring_emit_reg_write_reg_wait(struct amdgpu_ring *ring,
+						   uint32_t reg0, uint32_t reg1,
+						   uint32_t ref, uint32_t mask)
+{
+	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
+
+	gfx_v11_0_wait_reg_mem(ring, usepfp, 0, 1, reg0, reg1,
+			       ref, mask, 0x20);
+}
+
+static void
+gfx_v11_0_set_gfx_eop_interrupt_state(struct amdgpu_device *adev,
+				      uint32_t me, uint32_t pipe,
+				      enum amdgpu_interrupt_state state)
+{
+	uint32_t cp_int_cntl, cp_int_cntl_reg;
+
+	if (!me) {
+		switch (pipe) {
+		case 0:
+			cp_int_cntl_reg = SOC15_REG_OFFSET(GC, 0, regCP_INT_CNTL_RING0);
+			break;
+		case 1:
+			cp_int_cntl_reg = SOC15_REG_OFFSET(GC, 0, regCP_INT_CNTL_RING1);
+			break;
+		default:
+			DRM_DEBUG("invalid pipe %d\n", pipe);
+			return;
+		}
+	} else {
+		DRM_DEBUG("invalid me %d\n", me);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+		cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+					    TIME_STAMP_INT_ENABLE, 0);
+		cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+					    GENERIC0_INT_ENABLE, 0);
+		WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+		cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+					    TIME_STAMP_INT_ENABLE, 1);
+		cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+					    GENERIC0_INT_ENABLE, 1);
+		WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+		break;
+	default:
+		break;
+	}
+}
+
+static void gfx_v11_0_set_compute_eop_interrupt_state(struct amdgpu_device *adev,
+						     int me, int pipe,
+						     enum amdgpu_interrupt_state state)
+{
+	u32 mec_int_cntl, mec_int_cntl_reg;
+
+	/*
+	 * amdgpu controls only the first MEC. That's why this function only
+	 * handles the setting of interrupts for this specific MEC. All other
+	 * pipes' interrupts are set by amdkfd.
+	 */
+
+	if (me == 1) {
+		switch (pipe) {
+		case 0:
+			mec_int_cntl_reg = SOC15_REG_OFFSET(GC, 0, regCP_ME1_PIPE0_INT_CNTL);
+			break;
+		case 1:
+			mec_int_cntl_reg = SOC15_REG_OFFSET(GC, 0, regCP_ME1_PIPE1_INT_CNTL);
+			break;
+		case 2:
+			mec_int_cntl_reg = SOC15_REG_OFFSET(GC, 0, regCP_ME1_PIPE2_INT_CNTL);
+			break;
+		case 3:
+			mec_int_cntl_reg = SOC15_REG_OFFSET(GC, 0, regCP_ME1_PIPE3_INT_CNTL);
+			break;
+		default:
+			DRM_DEBUG("invalid pipe %d\n", pipe);
+			return;
+		}
+	} else {
+		DRM_DEBUG("invalid me %d\n", me);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		mec_int_cntl = RREG32_SOC15_IP(GC, mec_int_cntl_reg);
+		mec_int_cntl = REG_SET_FIELD(mec_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+					     TIME_STAMP_INT_ENABLE, 0);
+		mec_int_cntl = REG_SET_FIELD(mec_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+					     GENERIC0_INT_ENABLE, 0);
+		WREG32_SOC15_IP(GC, mec_int_cntl_reg, mec_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		mec_int_cntl = RREG32_SOC15_IP(GC, mec_int_cntl_reg);
+		mec_int_cntl = REG_SET_FIELD(mec_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+					     TIME_STAMP_INT_ENABLE, 1);
+		mec_int_cntl = REG_SET_FIELD(mec_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+					     GENERIC0_INT_ENABLE, 1);
+		WREG32_SOC15_IP(GC, mec_int_cntl_reg, mec_int_cntl);
+		break;
+	default:
+		break;
+	}
+}
+
+static int gfx_v11_0_set_eop_interrupt_state(struct amdgpu_device *adev,
+					    struct amdgpu_irq_src *src,
+					    unsigned type,
+					    enum amdgpu_interrupt_state state)
+{
+	switch (type) {
+	case AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP:
+		gfx_v11_0_set_gfx_eop_interrupt_state(adev, 0, 0, state);
+		break;
+	case AMDGPU_CP_IRQ_GFX_ME0_PIPE1_EOP:
+		gfx_v11_0_set_gfx_eop_interrupt_state(adev, 0, 1, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP:
+		gfx_v11_0_set_compute_eop_interrupt_state(adev, 1, 0, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE1_EOP:
+		gfx_v11_0_set_compute_eop_interrupt_state(adev, 1, 1, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE2_EOP:
+		gfx_v11_0_set_compute_eop_interrupt_state(adev, 1, 2, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE3_EOP:
+		gfx_v11_0_set_compute_eop_interrupt_state(adev, 1, 3, state);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int gfx_v11_0_eop_irq(struct amdgpu_device *adev,
+			     struct amdgpu_irq_src *source,
+			     struct amdgpu_iv_entry *entry)
+{
+	u32 doorbell_offset = entry->src_data[0];
+	u8 me_id, pipe_id, queue_id;
+	struct amdgpu_ring *ring;
+	int i;
+
+	DRM_DEBUG("IH: CP EOP\n");
+
+	if (adev->enable_mes && doorbell_offset) {
+		struct amdgpu_userq_fence_driver *fence_drv = NULL;
+		struct xarray *xa = &adev->userq_xa;
+		unsigned long flags;
+
+		xa_lock_irqsave(xa, flags);
+		fence_drv = xa_load(xa, doorbell_offset);
+		if (fence_drv)
+			amdgpu_userq_fence_driver_process(fence_drv);
+		xa_unlock_irqrestore(xa, flags);
+	} else {
+		me_id = (entry->ring_id & 0x0c) >> 2;
+		pipe_id = (entry->ring_id & 0x03) >> 0;
+		queue_id = (entry->ring_id & 0x70) >> 4;
+
+		switch (me_id) {
+		case 0:
+			if (pipe_id == 0)
+				amdgpu_fence_process(&adev->gfx.gfx_ring[0]);
+			else
+				amdgpu_fence_process(&adev->gfx.gfx_ring[1]);
+			break;
+		case 1:
+		case 2:
+			for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+				ring = &adev->gfx.compute_ring[i];
+				/* Per-queue interrupt is supported for MEC starting from VI.
+				 * The interrupt can only be enabled/disabled per pipe instead
+				 * of per queue.
+				 */
+				if ((ring->me == me_id) &&
+				    (ring->pipe == pipe_id) &&
+				    (ring->queue == queue_id))
+					amdgpu_fence_process(ring);
+			}
+			break;
+		}
+	}
+
+	return 0;
+}
+
+static int gfx_v11_0_set_priv_reg_fault_state(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      unsigned int type,
+					      enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl_reg, cp_int_cntl;
+	int i, j;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+	case AMDGPU_IRQ_STATE_ENABLE:
+		for (i = 0; i < adev->gfx.me.num_me; i++) {
+			for (j = 0; j < adev->gfx.me.num_pipe_per_me; j++) {
+				cp_int_cntl_reg = gfx_v11_0_get_cpg_int_cntl(adev, i, j);
+
+				if (cp_int_cntl_reg) {
+					cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+					cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+								    PRIV_REG_INT_ENABLE,
+								    state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+					WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+				}
+			}
+		}
+		for (i = 0; i < adev->gfx.mec.num_mec; i++) {
+			for (j = 0; j < adev->gfx.mec.num_pipe_per_mec; j++) {
+				/* MECs start at 1 */
+				cp_int_cntl_reg = gfx_v11_0_get_cpc_int_cntl(adev, i + 1, j);
+
+				if (cp_int_cntl_reg) {
+					cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+					cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+								    PRIV_REG_INT_ENABLE,
+								    state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+					WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+				}
+			}
+		}
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gfx_v11_0_set_bad_op_fault_state(struct amdgpu_device *adev,
+					    struct amdgpu_irq_src *source,
+					    unsigned type,
+					    enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl_reg, cp_int_cntl;
+	int i, j;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+	case AMDGPU_IRQ_STATE_ENABLE:
+		for (i = 0; i < adev->gfx.me.num_me; i++) {
+			for (j = 0; j < adev->gfx.me.num_pipe_per_me; j++) {
+				cp_int_cntl_reg = gfx_v11_0_get_cpg_int_cntl(adev, i, j);
+
+				if (cp_int_cntl_reg) {
+					cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+					cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+								    OPCODE_ERROR_INT_ENABLE,
+								    state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+					WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+				}
+			}
+		}
+		for (i = 0; i < adev->gfx.mec.num_mec; i++) {
+			for (j = 0; j < adev->gfx.mec.num_pipe_per_mec; j++) {
+				/* MECs start at 1 */
+				cp_int_cntl_reg = gfx_v11_0_get_cpc_int_cntl(adev, i + 1, j);
+
+				if (cp_int_cntl_reg) {
+					cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+					cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+								    OPCODE_ERROR_INT_ENABLE,
+								    state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+					WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+				}
+			}
+		}
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int gfx_v11_0_set_priv_inst_fault_state(struct amdgpu_device *adev,
+					       struct amdgpu_irq_src *source,
+					       unsigned int type,
+					       enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl_reg, cp_int_cntl;
+	int i, j;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+	case AMDGPU_IRQ_STATE_ENABLE:
+		for (i = 0; i < adev->gfx.me.num_me; i++) {
+			for (j = 0; j < adev->gfx.me.num_pipe_per_me; j++) {
+				cp_int_cntl_reg = gfx_v11_0_get_cpg_int_cntl(adev, i, j);
+
+				if (cp_int_cntl_reg) {
+					cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+					cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+								    PRIV_INSTR_INT_ENABLE,
+								    state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+					WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+				}
+			}
+		}
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static void gfx_v11_0_handle_priv_fault(struct amdgpu_device *adev,
+					struct amdgpu_iv_entry *entry)
+{
+	u8 me_id, pipe_id, queue_id;
+	struct amdgpu_ring *ring;
+	int i;
+
+	me_id = (entry->ring_id & 0x0c) >> 2;
+	pipe_id = (entry->ring_id & 0x03) >> 0;
+	queue_id = (entry->ring_id & 0x70) >> 4;
+
+	if (!adev->gfx.disable_kq) {
+		switch (me_id) {
+		case 0:
+			for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+				ring = &adev->gfx.gfx_ring[i];
+				if (ring->me == me_id && ring->pipe == pipe_id &&
+				    ring->queue == queue_id)
+					drm_sched_fault(&ring->sched);
+			}
+			break;
+		case 1:
+		case 2:
+			for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+				ring = &adev->gfx.compute_ring[i];
+				if (ring->me == me_id && ring->pipe == pipe_id &&
+				    ring->queue == queue_id)
+					drm_sched_fault(&ring->sched);
+			}
+			break;
+		default:
+			BUG();
+			break;
+		}
+	}
+}
+
+static int gfx_v11_0_priv_reg_irq(struct amdgpu_device *adev,
+				  struct amdgpu_irq_src *source,
+				  struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal register access in command stream\n");
+	gfx_v11_0_handle_priv_fault(adev, entry);
+	return 0;
+}
+
+static int gfx_v11_0_bad_op_irq(struct amdgpu_device *adev,
+				struct amdgpu_irq_src *source,
+				struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal opcode in command stream \n");
+	gfx_v11_0_handle_priv_fault(adev, entry);
+	return 0;
+}
+
+static int gfx_v11_0_priv_inst_irq(struct amdgpu_device *adev,
+				   struct amdgpu_irq_src *source,
+				   struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal instruction in command stream\n");
+	gfx_v11_0_handle_priv_fault(adev, entry);
+	return 0;
+}
+
+static int gfx_v11_0_rlc_gc_fed_irq(struct amdgpu_device *adev,
+				  struct amdgpu_irq_src *source,
+				  struct amdgpu_iv_entry *entry)
+{
+	if (adev->gfx.ras && adev->gfx.ras->rlc_gc_fed_irq)
+		return adev->gfx.ras->rlc_gc_fed_irq(adev, source, entry);
+
+	return 0;
+}
+
+#if 0
+static int gfx_v11_0_kiq_set_interrupt_state(struct amdgpu_device *adev,
+					     struct amdgpu_irq_src *src,
+					     unsigned int type,
+					     enum amdgpu_interrupt_state state)
+{
+	uint32_t tmp, target;
+	struct amdgpu_ring *ring = &(adev->gfx.kiq[0].ring);
+
+	target = SOC15_REG_OFFSET(GC, 0, regCP_ME1_PIPE0_INT_CNTL);
+	target += ring->pipe;
+
+	switch (type) {
+	case AMDGPU_CP_KIQ_IRQ_DRIVER0:
+		if (state == AMDGPU_IRQ_STATE_DISABLE) {
+			tmp = RREG32_SOC15(GC, 0, regCPC_INT_CNTL);
+			tmp = REG_SET_FIELD(tmp, CPC_INT_CNTL,
+					    GENERIC2_INT_ENABLE, 0);
+			WREG32_SOC15(GC, 0, regCPC_INT_CNTL, tmp);
+
+			tmp = RREG32_SOC15_IP(GC, target);
+			tmp = REG_SET_FIELD(tmp, CP_ME1_PIPE0_INT_CNTL,
+					    GENERIC2_INT_ENABLE, 0);
+			WREG32_SOC15_IP(GC, target, tmp);
+		} else {
+			tmp = RREG32_SOC15(GC, 0, regCPC_INT_CNTL);
+			tmp = REG_SET_FIELD(tmp, CPC_INT_CNTL,
+					    GENERIC2_INT_ENABLE, 1);
+			WREG32_SOC15(GC, 0, regCPC_INT_CNTL, tmp);
+
+			tmp = RREG32_SOC15_IP(GC, target);
+			tmp = REG_SET_FIELD(tmp, CP_ME1_PIPE0_INT_CNTL,
+					    GENERIC2_INT_ENABLE, 1);
+			WREG32_SOC15_IP(GC, target, tmp);
+		}
+		break;
+	default:
+		BUG(); /* kiq only support GENERIC2_INT now */
+		break;
+	}
+	return 0;
+}
+#endif
+
+static void gfx_v11_0_emit_mem_sync(struct amdgpu_ring *ring)
+{
+	const unsigned int gcr_cntl =
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GL2_INV(1) |
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GL2_WB(1) |
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GLM_INV(1) |
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GLM_WB(1) |
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GL1_INV(1) |
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GLV_INV(1) |
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GLK_INV(1) |
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GLI_INV(1);
+
+	/* ACQUIRE_MEM - make one or more surfaces valid for use by the subsequent operations */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_ACQUIRE_MEM, 6));
+	amdgpu_ring_write(ring, 0); /* CP_COHER_CNTL */
+	amdgpu_ring_write(ring, 0xffffffff);  /* CP_COHER_SIZE */
+	amdgpu_ring_write(ring, 0xffffff);  /* CP_COHER_SIZE_HI */
+	amdgpu_ring_write(ring, 0); /* CP_COHER_BASE */
+	amdgpu_ring_write(ring, 0);  /* CP_COHER_BASE_HI */
+	amdgpu_ring_write(ring, 0x0000000A); /* POLL_INTERVAL */
+	amdgpu_ring_write(ring, gcr_cntl); /* GCR_CNTL */
+}
+
+static bool gfx_v11_pipe_reset_support(struct amdgpu_device *adev)
+{
+	/* Disable the pipe reset until the CPFW fully support it.*/
+	dev_warn_once(adev->dev, "The CPFW hasn't support pipe reset yet.\n");
+	return false;
+}
+
+
+static int gfx_v11_reset_gfx_pipe(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t reset_pipe = 0, clean_pipe = 0;
+	int r;
+
+	if (!gfx_v11_pipe_reset_support(adev))
+		return -EOPNOTSUPP;
+
+	gfx_v11_0_set_safe_mode(adev, 0);
+	mutex_lock(&adev->srbm_mutex);
+	soc21_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+
+	switch (ring->pipe) {
+	case 0:
+		reset_pipe = REG_SET_FIELD(reset_pipe, CP_ME_CNTL,
+					   PFP_PIPE0_RESET, 1);
+		reset_pipe = REG_SET_FIELD(reset_pipe, CP_ME_CNTL,
+					   ME_PIPE0_RESET, 1);
+		clean_pipe = REG_SET_FIELD(clean_pipe, CP_ME_CNTL,
+					   PFP_PIPE0_RESET, 0);
+		clean_pipe = REG_SET_FIELD(clean_pipe, CP_ME_CNTL,
+					   ME_PIPE0_RESET, 0);
+		break;
+	case 1:
+		reset_pipe = REG_SET_FIELD(reset_pipe, CP_ME_CNTL,
+					   PFP_PIPE1_RESET, 1);
+		reset_pipe = REG_SET_FIELD(reset_pipe, CP_ME_CNTL,
+					   ME_PIPE1_RESET, 1);
+		clean_pipe = REG_SET_FIELD(clean_pipe, CP_ME_CNTL,
+					   PFP_PIPE1_RESET, 0);
+		clean_pipe = REG_SET_FIELD(clean_pipe, CP_ME_CNTL,
+					   ME_PIPE1_RESET, 0);
+		break;
+	default:
+		break;
+	}
+
+	WREG32_SOC15(GC, 0, regCP_ME_CNTL, reset_pipe);
+	WREG32_SOC15(GC, 0, regCP_ME_CNTL, clean_pipe);
+
+	r = (RREG32(SOC15_REG_OFFSET(GC, 0, regCP_GFX_RS64_INSTR_PNTR1)) << 2) -
+						RS64_FW_UC_START_ADDR_LO;
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+	gfx_v11_0_unset_safe_mode(adev, 0);
+
+	dev_info(adev->dev, "The ring %s pipe reset to the ME firmware start PC: %s\n", ring->name,
+			r == 0 ? "successfully" : "failed");
+	/* FIXME: Sometimes driver can't cache the ME firmware start PC correctly,
+	 * so the pipe reset status relies on the later gfx ring test result.
+	 */
+	return 0;
+}
+
+static int gfx_v11_0_reset_kgq(struct amdgpu_ring *ring,
+			       unsigned int vmid,
+			       struct amdgpu_fence *timedout_fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int r;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+
+	r = amdgpu_mes_reset_legacy_queue(ring->adev, ring, vmid, false);
+	if (r) {
+
+		dev_warn(adev->dev, "reset via MES failed and try pipe reset %d\n", r);
+		r = gfx_v11_reset_gfx_pipe(ring);
+		if (r)
+			return r;
+	}
+
+	r = gfx_v11_0_kgq_init_queue(ring, true);
+	if (r) {
+		dev_err(adev->dev, "failed to init kgq\n");
+		return r;
+	}
+
+	r = amdgpu_mes_map_legacy_queue(adev, ring);
+	if (r) {
+		dev_err(adev->dev, "failed to remap kgq\n");
+		return r;
+	}
+
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static int gfx_v11_0_reset_compute_pipe(struct amdgpu_ring *ring)
+{
+
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t reset_pipe = 0, clean_pipe = 0;
+	int r;
+
+	if (!gfx_v11_pipe_reset_support(adev))
+		return -EOPNOTSUPP;
+
+	gfx_v11_0_set_safe_mode(adev, 0);
+	mutex_lock(&adev->srbm_mutex);
+	soc21_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+
+	reset_pipe = RREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL);
+	clean_pipe = reset_pipe;
+
+	if (adev->gfx.rs64_enable) {
+
+		switch (ring->pipe) {
+		case 0:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE0_RESET, 1);
+			clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE0_RESET, 0);
+			break;
+		case 1:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE1_RESET, 1);
+			clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE1_RESET, 0);
+			break;
+		case 2:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE2_RESET, 1);
+			clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE2_RESET, 0);
+			break;
+		case 3:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE3_RESET, 1);
+			clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE3_RESET, 0);
+			break;
+		default:
+			break;
+		}
+		WREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL, reset_pipe);
+		WREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL, clean_pipe);
+		r = (RREG32_SOC15(GC, 0, regCP_MEC_RS64_INSTR_PNTR) << 2) -
+					RS64_FW_UC_START_ADDR_LO;
+	} else {
+		if (ring->me == 1) {
+			switch (ring->pipe) {
+			case 0:
+				reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE0_RESET, 1);
+				clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE0_RESET, 0);
+				break;
+			case 1:
+				reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE1_RESET, 1);
+				clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE1_RESET, 0);
+				break;
+			case 2:
+				reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE2_RESET, 1);
+				clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE2_RESET, 0);
+				break;
+			case 3:
+				reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE3_RESET, 1);
+				clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE3_RESET, 0);
+				break;
+			default:
+				break;
+			}
+			/* mec1 fw pc: CP_MEC1_INSTR_PNTR */
+		} else {
+			switch (ring->pipe) {
+			case 0:
+				reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+							   MEC_ME2_PIPE0_RESET, 1);
+				clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_CNTL,
+							   MEC_ME2_PIPE0_RESET, 0);
+				break;
+			case 1:
+				reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+							   MEC_ME2_PIPE1_RESET, 1);
+				clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_CNTL,
+							   MEC_ME2_PIPE1_RESET, 0);
+				break;
+			case 2:
+				reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+							   MEC_ME2_PIPE2_RESET, 1);
+				clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_CNTL,
+							   MEC_ME2_PIPE2_RESET, 0);
+				break;
+			case 3:
+				reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+							   MEC_ME2_PIPE3_RESET, 1);
+				clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_CNTL,
+							   MEC_ME2_PIPE3_RESET, 0);
+				break;
+			default:
+				break;
+			}
+			/* mec2 fw pc: CP:CP_MEC2_INSTR_PNTR */
+		}
+		WREG32_SOC15(GC, 0, regCP_MEC_CNTL, reset_pipe);
+		WREG32_SOC15(GC, 0, regCP_MEC_CNTL, clean_pipe);
+		r = RREG32(SOC15_REG_OFFSET(GC, 0, regCP_MEC1_INSTR_PNTR));
+	}
+
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+	gfx_v11_0_unset_safe_mode(adev, 0);
+
+	dev_info(adev->dev, "The ring %s pipe resets to MEC FW start PC: %s\n", ring->name,
+			r == 0 ? "successfully" : "failed");
+	/*FIXME:Sometimes driver can't cache the MEC firmware start PC correctly, so the pipe
+	 * reset status relies on the compute ring test result.
+	 */
+	return 0;
+}
+
+static int gfx_v11_0_reset_kcq(struct amdgpu_ring *ring,
+			       unsigned int vmid,
+			       struct amdgpu_fence *timedout_fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int r = 0;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+
+	r = amdgpu_mes_reset_legacy_queue(ring->adev, ring, vmid, true);
+	if (r) {
+		dev_warn(adev->dev, "fail(%d) to reset kcq and try pipe reset\n", r);
+		r = gfx_v11_0_reset_compute_pipe(ring);
+		if (r)
+			return r;
+	}
+
+	r = gfx_v11_0_kcq_init_queue(ring, true);
+	if (r) {
+		dev_err(adev->dev, "fail to init kcq\n");
+		return r;
+	}
+	r = amdgpu_mes_map_legacy_queue(adev, ring);
+	if (r) {
+		dev_err(adev->dev, "failed to remap kcq\n");
+		return r;
+	}
+
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static void gfx_v11_ip_print(struct amdgpu_ip_block *ip_block, struct drm_printer *p)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	uint32_t i, j, k, reg, index = 0;
+	uint32_t reg_count = ARRAY_SIZE(gc_reg_list_11_0);
+
+	if (!adev->gfx.ip_dump_core)
+		return;
+
+	for (i = 0; i < reg_count; i++)
+		drm_printf(p, "%-50s \t 0x%08x\n",
+			   gc_reg_list_11_0[i].reg_name,
+			   adev->gfx.ip_dump_core[i]);
+
+	/* print compute queue registers for all instances */
+	if (!adev->gfx.ip_dump_compute_queues)
+		return;
+
+	reg_count = ARRAY_SIZE(gc_cp_reg_list_11);
+	drm_printf(p, "\nnum_mec: %d num_pipe: %d num_queue: %d\n",
+		   adev->gfx.mec.num_mec,
+		   adev->gfx.mec.num_pipe_per_mec,
+		   adev->gfx.mec.num_queue_per_pipe);
+
+	for (i = 0; i < adev->gfx.mec.num_mec; i++) {
+		for (j = 0; j < adev->gfx.mec.num_pipe_per_mec; j++) {
+			for (k = 0; k < adev->gfx.mec.num_queue_per_pipe; k++) {
+				drm_printf(p, "\nmec %d, pipe %d, queue %d\n", i, j, k);
+				for (reg = 0; reg < reg_count; reg++) {
+					if (i && gc_cp_reg_list_11[reg].reg_offset == regCP_MEC_ME1_HEADER_DUMP)
+						drm_printf(p, "%-50s \t 0x%08x\n",
+							   "regCP_MEC_ME2_HEADER_DUMP",
+							   adev->gfx.ip_dump_compute_queues[index + reg]);
+					else
+						drm_printf(p, "%-50s \t 0x%08x\n",
+							   gc_cp_reg_list_11[reg].reg_name,
+							   adev->gfx.ip_dump_compute_queues[index + reg]);
+				}
+				index += reg_count;
+			}
+		}
+	}
+
+	/* print gfx queue registers for all instances */
+	if (!adev->gfx.ip_dump_gfx_queues)
+		return;
+
+	index = 0;
+	reg_count = ARRAY_SIZE(gc_gfx_queue_reg_list_11);
+	drm_printf(p, "\nnum_me: %d num_pipe: %d num_queue: %d\n",
+		   adev->gfx.me.num_me,
+		   adev->gfx.me.num_pipe_per_me,
+		   adev->gfx.me.num_queue_per_pipe);
+
+	for (i = 0; i < adev->gfx.me.num_me; i++) {
+		for (j = 0; j < adev->gfx.me.num_pipe_per_me; j++) {
+			for (k = 0; k < adev->gfx.me.num_queue_per_pipe; k++) {
+				drm_printf(p, "\nme %d, pipe %d, queue %d\n", i, j, k);
+				for (reg = 0; reg < reg_count; reg++) {
+					drm_printf(p, "%-50s \t 0x%08x\n",
+						   gc_gfx_queue_reg_list_11[reg].reg_name,
+						   adev->gfx.ip_dump_gfx_queues[index + reg]);
+				}
+				index += reg_count;
+			}
+		}
+	}
+}
+
+static void gfx_v11_ip_dump(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	uint32_t i, j, k, reg, index = 0;
+	uint32_t reg_count = ARRAY_SIZE(gc_reg_list_11_0);
+
+	if (!adev->gfx.ip_dump_core)
+		return;
+
+	amdgpu_gfx_off_ctrl(adev, false);
+	for (i = 0; i < reg_count; i++)
+		adev->gfx.ip_dump_core[i] = RREG32(SOC15_REG_ENTRY_OFFSET(gc_reg_list_11_0[i]));
+	amdgpu_gfx_off_ctrl(adev, true);
+
+	/* dump compute queue registers for all instances */
+	if (!adev->gfx.ip_dump_compute_queues)
+		return;
+
+	reg_count = ARRAY_SIZE(gc_cp_reg_list_11);
+	amdgpu_gfx_off_ctrl(adev, false);
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < adev->gfx.mec.num_mec; i++) {
+		for (j = 0; j < adev->gfx.mec.num_pipe_per_mec; j++) {
+			for (k = 0; k < adev->gfx.mec.num_queue_per_pipe; k++) {
+				/* ME0 is for GFX so start from 1 for CP */
+				soc21_grbm_select(adev, adev->gfx.me.num_me + i, j, k, 0);
+				for (reg = 0; reg < reg_count; reg++) {
+					if (i &&
+					    gc_cp_reg_list_11[reg].reg_offset ==
+						    regCP_MEC_ME1_HEADER_DUMP)
+						adev->gfx.ip_dump_compute_queues[index + reg] =
+							RREG32(SOC15_REG_OFFSET(GC, 0,
+							       regCP_MEC_ME2_HEADER_DUMP));
+					else
+						adev->gfx.ip_dump_compute_queues[index + reg] =
+							RREG32(SOC15_REG_ENTRY_OFFSET(
+								       gc_cp_reg_list_11[reg]));
+				}
+				index += reg_count;
+			}
+		}
+	}
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+	amdgpu_gfx_off_ctrl(adev, true);
+
+	/* dump gfx queue registers for all instances */
+	if (!adev->gfx.ip_dump_gfx_queues)
+		return;
+
+	index = 0;
+	reg_count = ARRAY_SIZE(gc_gfx_queue_reg_list_11);
+	amdgpu_gfx_off_ctrl(adev, false);
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < adev->gfx.me.num_me; i++) {
+		for (j = 0; j < adev->gfx.me.num_pipe_per_me; j++) {
+			for (k = 0; k < adev->gfx.me.num_queue_per_pipe; k++) {
+				soc21_grbm_select(adev, i, j, k, 0);
+
+				for (reg = 0; reg < reg_count; reg++) {
+					adev->gfx.ip_dump_gfx_queues[index + reg] =
+						RREG32(SOC15_REG_ENTRY_OFFSET(
+							gc_gfx_queue_reg_list_11[reg]));
+				}
+				index += reg_count;
+			}
+		}
+	}
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+	amdgpu_gfx_off_ctrl(adev, true);
+}
+
+static void gfx_v11_0_ring_emit_cleaner_shader(struct amdgpu_ring *ring)
+{
+	/* Emit the cleaner shader */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_RUN_CLEANER_SHADER, 0));
+	amdgpu_ring_write(ring, 0);  /* RESERVED field, programmed to zero */
+}
+
+static void gfx_v11_0_ring_begin_use(struct amdgpu_ring *ring)
+{
+	amdgpu_gfx_profile_ring_begin_use(ring);
+
+	amdgpu_gfx_enforce_isolation_ring_begin_use(ring);
+}
+
+static void gfx_v11_0_ring_end_use(struct amdgpu_ring *ring)
+{
+	amdgpu_gfx_profile_ring_end_use(ring);
+
+	amdgpu_gfx_enforce_isolation_ring_end_use(ring);
+}
+
+static const struct amd_ip_funcs gfx_v11_0_ip_funcs = {
+	.name = "gfx_v11_0",
+	.early_init = gfx_v11_0_early_init,
+	.late_init = gfx_v11_0_late_init,
+	.sw_init = gfx_v11_0_sw_init,
+	.sw_fini = gfx_v11_0_sw_fini,
+	.hw_init = gfx_v11_0_hw_init,
+	.hw_fini = gfx_v11_0_hw_fini,
+	.suspend = gfx_v11_0_suspend,
+	.resume = gfx_v11_0_resume,
+	.is_idle = gfx_v11_0_is_idle,
+	.wait_for_idle = gfx_v11_0_wait_for_idle,
+	.soft_reset = gfx_v11_0_soft_reset,
+	.check_soft_reset = gfx_v11_0_check_soft_reset,
+	.post_soft_reset = gfx_v11_0_post_soft_reset,
+	.set_clockgating_state = gfx_v11_0_set_clockgating_state,
+	.set_powergating_state = gfx_v11_0_set_powergating_state,
+	.get_clockgating_state = gfx_v11_0_get_clockgating_state,
+	.dump_ip_state = gfx_v11_ip_dump,
+	.print_ip_state = gfx_v11_ip_print,
+};
+
+static const struct amdgpu_ring_funcs gfx_v11_0_ring_funcs_gfx = {
+	.type = AMDGPU_RING_TYPE_GFX,
+	.align_mask = 0xff,
+	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
+	.support_64bit_ptrs = true,
+	.secure_submission_supported = true,
+	.get_rptr = gfx_v11_0_ring_get_rptr_gfx,
+	.get_wptr = gfx_v11_0_ring_get_wptr_gfx,
+	.set_wptr = gfx_v11_0_ring_set_wptr_gfx,
+	.emit_frame_size = /* totally 247 maximum if 16 IBs */
+		5 + /* update_spm_vmid */
+		5 + /* COND_EXEC */
+		22 + /* SET_Q_PREEMPTION_MODE */
+		7 + /* PIPELINE_SYNC */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 5 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 +
+		4 + /* VM_FLUSH */
+		8 + /* FENCE for VM_FLUSH */
+		20 + /* GDS switch */
+		5 + /* COND_EXEC */
+		7 + /* HDP_flush */
+		4 + /* VGT_flush */
+		31 + /*	DE_META */
+		3 + /* CNTX_CTRL */
+		5 + /* HDP_INVL */
+		22 + /* SET_Q_PREEMPTION_MODE */
+		8 + 8 + /* FENCE x2 */
+		8 + /* gfx_v11_0_emit_mem_sync */
+		2, /* gfx_v11_0_ring_emit_cleaner_shader */
+	.emit_ib_size =	4, /* gfx_v11_0_ring_emit_ib_gfx */
+	.emit_ib = gfx_v11_0_ring_emit_ib_gfx,
+	.emit_fence = gfx_v11_0_ring_emit_fence,
+	.emit_pipeline_sync = gfx_v11_0_ring_emit_pipeline_sync,
+	.emit_vm_flush = gfx_v11_0_ring_emit_vm_flush,
+	.emit_gds_switch = gfx_v11_0_ring_emit_gds_switch,
+	.emit_hdp_flush = gfx_v11_0_ring_emit_hdp_flush,
+	.test_ring = gfx_v11_0_ring_test_ring,
+	.test_ib = gfx_v11_0_ring_test_ib,
+	.insert_nop = gfx_v11_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_cntxcntl = gfx_v11_0_ring_emit_cntxcntl,
+	.emit_gfx_shadow = gfx_v11_0_ring_emit_gfx_shadow,
+	.init_cond_exec = gfx_v11_0_ring_emit_init_cond_exec,
+	.preempt_ib = gfx_v11_0_ring_preempt_ib,
+	.emit_frame_cntl = gfx_v11_0_ring_emit_frame_cntl,
+	.emit_wreg = gfx_v11_0_ring_emit_wreg,
+	.emit_reg_wait = gfx_v11_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = gfx_v11_0_ring_emit_reg_write_reg_wait,
+	.emit_mem_sync = gfx_v11_0_emit_mem_sync,
+	.reset = gfx_v11_0_reset_kgq,
+	.emit_cleaner_shader = gfx_v11_0_ring_emit_cleaner_shader,
+	.begin_use = gfx_v11_0_ring_begin_use,
+	.end_use = gfx_v11_0_ring_end_use,
+};
+
+static const struct amdgpu_ring_funcs gfx_v11_0_ring_funcs_compute = {
+	.type = AMDGPU_RING_TYPE_COMPUTE,
+	.align_mask = 0xff,
+	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
+	.support_64bit_ptrs = true,
+	.get_rptr = gfx_v11_0_ring_get_rptr_compute,
+	.get_wptr = gfx_v11_0_ring_get_wptr_compute,
+	.set_wptr = gfx_v11_0_ring_set_wptr_compute,
+	.emit_frame_size =
+		5 + /* update_spm_vmid */
+		20 + /* gfx_v11_0_ring_emit_gds_switch */
+		7 + /* gfx_v11_0_ring_emit_hdp_flush */
+		5 + /* hdp invalidate */
+		7 + /* gfx_v11_0_ring_emit_pipeline_sync */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 5 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 +
+		2 + /* gfx_v11_0_ring_emit_vm_flush */
+		8 + 8 + 8 + /* gfx_v11_0_ring_emit_fence x3 for user fence, vm fence */
+		8 + /* gfx_v11_0_emit_mem_sync */
+		2, /* gfx_v11_0_ring_emit_cleaner_shader */
+	.emit_ib_size =	7, /* gfx_v11_0_ring_emit_ib_compute */
+	.emit_ib = gfx_v11_0_ring_emit_ib_compute,
+	.emit_fence = gfx_v11_0_ring_emit_fence,
+	.emit_pipeline_sync = gfx_v11_0_ring_emit_pipeline_sync,
+	.emit_vm_flush = gfx_v11_0_ring_emit_vm_flush,
+	.emit_gds_switch = gfx_v11_0_ring_emit_gds_switch,
+	.emit_hdp_flush = gfx_v11_0_ring_emit_hdp_flush,
+	.test_ring = gfx_v11_0_ring_test_ring,
+	.test_ib = gfx_v11_0_ring_test_ib,
+	.insert_nop = gfx_v11_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_wreg = gfx_v11_0_ring_emit_wreg,
+	.emit_reg_wait = gfx_v11_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = gfx_v11_0_ring_emit_reg_write_reg_wait,
+	.emit_mem_sync = gfx_v11_0_emit_mem_sync,
+	.reset = gfx_v11_0_reset_kcq,
+	.emit_cleaner_shader = gfx_v11_0_ring_emit_cleaner_shader,
+	.begin_use = gfx_v11_0_ring_begin_use,
+	.end_use = gfx_v11_0_ring_end_use,
+};
+
+static const struct amdgpu_ring_funcs gfx_v11_0_ring_funcs_kiq = {
+	.type = AMDGPU_RING_TYPE_KIQ,
+	.align_mask = 0xff,
+	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
+	.support_64bit_ptrs = true,
+	.get_rptr = gfx_v11_0_ring_get_rptr_compute,
+	.get_wptr = gfx_v11_0_ring_get_wptr_compute,
+	.set_wptr = gfx_v11_0_ring_set_wptr_compute,
+	.emit_frame_size =
+		20 + /* gfx_v11_0_ring_emit_gds_switch */
+		7 + /* gfx_v11_0_ring_emit_hdp_flush */
+		5 + /*hdp invalidate */
+		7 + /* gfx_v11_0_ring_emit_pipeline_sync */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 5 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 +
+		8 + 8 + 8, /* gfx_v11_0_ring_emit_fence_kiq x3 for user fence, vm fence */
+	.emit_ib_size =	7, /* gfx_v11_0_ring_emit_ib_compute */
+	.emit_ib = gfx_v11_0_ring_emit_ib_compute,
+	.emit_fence = gfx_v11_0_ring_emit_fence_kiq,
+	.test_ring = gfx_v11_0_ring_test_ring,
+	.test_ib = gfx_v11_0_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_rreg = gfx_v11_0_ring_emit_rreg,
+	.emit_wreg = gfx_v11_0_ring_emit_wreg,
+	.emit_reg_wait = gfx_v11_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = gfx_v11_0_ring_emit_reg_write_reg_wait,
+};
+
+static void gfx_v11_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	adev->gfx.kiq[0].ring.funcs = &gfx_v11_0_ring_funcs_kiq;
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+		adev->gfx.gfx_ring[i].funcs = &gfx_v11_0_ring_funcs_gfx;
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++)
+		adev->gfx.compute_ring[i].funcs = &gfx_v11_0_ring_funcs_compute;
+}
+
+static const struct amdgpu_irq_src_funcs gfx_v11_0_eop_irq_funcs = {
+	.set = gfx_v11_0_set_eop_interrupt_state,
+	.process = gfx_v11_0_eop_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v11_0_priv_reg_irq_funcs = {
+	.set = gfx_v11_0_set_priv_reg_fault_state,
+	.process = gfx_v11_0_priv_reg_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v11_0_bad_op_irq_funcs = {
+	.set = gfx_v11_0_set_bad_op_fault_state,
+	.process = gfx_v11_0_bad_op_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v11_0_priv_inst_irq_funcs = {
+	.set = gfx_v11_0_set_priv_inst_fault_state,
+	.process = gfx_v11_0_priv_inst_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v11_0_rlc_gc_fed_irq_funcs = {
+	.process = gfx_v11_0_rlc_gc_fed_irq,
+};
+
+static void gfx_v11_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->gfx.eop_irq.num_types = AMDGPU_CP_IRQ_LAST;
+	adev->gfx.eop_irq.funcs = &gfx_v11_0_eop_irq_funcs;
+
+	adev->gfx.priv_reg_irq.num_types = 1;
+	adev->gfx.priv_reg_irq.funcs = &gfx_v11_0_priv_reg_irq_funcs;
+
+	adev->gfx.bad_op_irq.num_types = 1;
+	adev->gfx.bad_op_irq.funcs = &gfx_v11_0_bad_op_irq_funcs;
+
+	adev->gfx.priv_inst_irq.num_types = 1;
+	adev->gfx.priv_inst_irq.funcs = &gfx_v11_0_priv_inst_irq_funcs;
+
+	adev->gfx.rlc_gc_fed_irq.num_types = 1; /* 0x80 FED error */
+	adev->gfx.rlc_gc_fed_irq.funcs = &gfx_v11_0_rlc_gc_fed_irq_funcs;
+
+}
+
+static void gfx_v11_0_set_imu_funcs(struct amdgpu_device *adev)
+{
+	if (adev->flags & AMD_IS_APU)
+		adev->gfx.imu.mode = MISSION_MODE;
+	else
+		adev->gfx.imu.mode = DEBUG_MODE;
+
+	adev->gfx.imu.funcs = &gfx_v11_0_imu_funcs;
+}
+
+static void gfx_v11_0_set_rlc_funcs(struct amdgpu_device *adev)
+{
+	adev->gfx.rlc.funcs = &gfx_v11_0_rlc_funcs;
+}
+
+static void gfx_v11_0_set_gds_init(struct amdgpu_device *adev)
+{
+	unsigned total_cu = adev->gfx.config.max_cu_per_sh *
+			    adev->gfx.config.max_sh_per_se *
+			    adev->gfx.config.max_shader_engines;
+
+	adev->gds.gds_size = 0x1000;
+	adev->gds.gds_compute_max_wave_id = total_cu * 32 - 1;
+	adev->gds.gws_size = 64;
+	adev->gds.oa_size = 16;
+}
+
+static void gfx_v11_0_set_mqd_funcs(struct amdgpu_device *adev)
+{
+	/* set gfx eng mqd */
+	adev->mqds[AMDGPU_HW_IP_GFX].mqd_size =
+		sizeof(struct v11_gfx_mqd);
+	adev->mqds[AMDGPU_HW_IP_GFX].init_mqd =
+		gfx_v11_0_gfx_mqd_init;
+	/* set compute eng mqd */
+	adev->mqds[AMDGPU_HW_IP_COMPUTE].mqd_size =
+		sizeof(struct v11_compute_mqd);
+	adev->mqds[AMDGPU_HW_IP_COMPUTE].init_mqd =
+		gfx_v11_0_compute_mqd_init;
+}
+
+static void gfx_v11_0_set_user_wgp_inactive_bitmap_per_sh(struct amdgpu_device *adev,
+							  u32 bitmap)
+{
+	u32 data;
+
+	if (!bitmap)
+		return;
+
+	data = bitmap << GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_WGPS__SHIFT;
+	data &= GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_WGPS_MASK;
+
+	WREG32_SOC15(GC, 0, regGC_USER_SHADER_ARRAY_CONFIG, data);
+}
+
+static u32 gfx_v11_0_get_wgp_active_bitmap_per_sh(struct amdgpu_device *adev)
+{
+	u32 data, wgp_bitmask;
+	data = RREG32_SOC15(GC, 0, regCC_GC_SHADER_ARRAY_CONFIG);
+	data |= RREG32_SOC15(GC, 0, regGC_USER_SHADER_ARRAY_CONFIG);
+
+	data &= CC_GC_SHADER_ARRAY_CONFIG__INACTIVE_WGPS_MASK;
+	data >>= CC_GC_SHADER_ARRAY_CONFIG__INACTIVE_WGPS__SHIFT;
+
+	wgp_bitmask =
+		amdgpu_gfx_create_bitmask(adev->gfx.config.max_cu_per_sh >> 1);
+
+	return (~data) & wgp_bitmask;
+}
+
+static u32 gfx_v11_0_get_cu_active_bitmap_per_sh(struct amdgpu_device *adev)
+{
+	u32 wgp_idx, wgp_active_bitmap;
+	u32 cu_bitmap_per_wgp, cu_active_bitmap;
+
+	wgp_active_bitmap = gfx_v11_0_get_wgp_active_bitmap_per_sh(adev);
+	cu_active_bitmap = 0;
+
+	for (wgp_idx = 0; wgp_idx < 16; wgp_idx++) {
+		/* if there is one WGP enabled, it means 2 CUs will be enabled */
+		cu_bitmap_per_wgp = 3 << (2 * wgp_idx);
+		if (wgp_active_bitmap & (1 << wgp_idx))
+			cu_active_bitmap |= cu_bitmap_per_wgp;
+	}
+
+	return cu_active_bitmap;
+}
+
+static int gfx_v11_0_get_cu_info(struct amdgpu_device *adev,
+				 struct amdgpu_cu_info *cu_info)
+{
+	int i, j, k, counter, active_cu_number = 0;
+	u32 mask, bitmap;
+	unsigned disable_masks[8 * 2];
+
+	if (!adev || !cu_info)
+		return -EINVAL;
+
+	amdgpu_gfx_parse_disable_cu(disable_masks, 8, 2);
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			bitmap = i * adev->gfx.config.max_sh_per_se + j;
+			if (!((gfx_v11_0_get_sa_active_bitmap(adev) >> bitmap) & 1))
+				continue;
+			mask = 1;
+			counter = 0;
+			gfx_v11_0_select_se_sh(adev, i, j, 0xffffffff, 0);
+			if (i < 8 && j < 2)
+				gfx_v11_0_set_user_wgp_inactive_bitmap_per_sh(
+					adev, disable_masks[i * 2 + j]);
+			bitmap = gfx_v11_0_get_cu_active_bitmap_per_sh(adev);
+
+			/**
+			 * GFX11 could support more than 4 SEs, while the bitmap
+			 * in cu_info struct is 4x4 and ioctl interface struct
+			 * drm_amdgpu_info_device should keep stable.
+			 * So we use last two columns of bitmap to store cu mask for
+			 * SEs 4 to 7, the layout of the bitmap is as below:
+			 *    SE0: {SH0,SH1} --> {bitmap[0][0], bitmap[0][1]}
+			 *    SE1: {SH0,SH1} --> {bitmap[1][0], bitmap[1][1]}
+			 *    SE2: {SH0,SH1} --> {bitmap[2][0], bitmap[2][1]}
+			 *    SE3: {SH0,SH1} --> {bitmap[3][0], bitmap[3][1]}
+			 *    SE4: {SH0,SH1} --> {bitmap[0][2], bitmap[0][3]}
+			 *    SE5: {SH0,SH1} --> {bitmap[1][2], bitmap[1][3]}
+			 *    SE6: {SH0,SH1} --> {bitmap[2][2], bitmap[2][3]}
+			 *    SE7: {SH0,SH1} --> {bitmap[3][2], bitmap[3][3]}
+			 */
+			cu_info->bitmap[0][i % 4][j + (i / 4) * 2] = bitmap;
+
+			for (k = 0; k < adev->gfx.config.max_cu_per_sh; k++) {
+				if (bitmap & mask)
+					counter++;
+
+				mask <<= 1;
+			}
+			active_cu_number += counter;
+		}
+	}
+	gfx_v11_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	cu_info->number = active_cu_number;
+	cu_info->simd_per_cu = NUM_SIMD_PER_CU;
+
+	return 0;
+}
+
+const struct amdgpu_ip_block_version gfx_v11_0_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_GFX,
+	.major = 11,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &gfx_v11_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.h b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.h
new file mode 100644
index 000000000000..157a5c812259
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright 2021 dvanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFX_V11_0_H__
+#define __GFX_V11_0_H__
+
+extern const struct amdgpu_ip_block_version gfx_v11_0_ip_block;
+
+int gfx_v11_0_request_gfx_index_mutex(struct amdgpu_device *adev,
+				      bool req);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3.c
new file mode 100644
index 000000000000..999bb3cc88b7
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3.c
@@ -0,0 +1,110 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "soc21.h"
+#include "gc/gc_11_0_3_offset.h"
+#include "gc/gc_11_0_3_sh_mask.h"
+#include "ivsrcid/gfx/irqsrcs_gfx_11_0_0.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "gfx_v11_0.h"
+
+
+static int gfx_v11_0_3_rlc_gc_fed_irq(struct amdgpu_device *adev,
+				  struct amdgpu_irq_src *source,
+				  struct amdgpu_iv_entry *entry)
+{
+	uint32_t rlc_status0 = 0, rlc_status1 = 0;
+	struct ras_common_if *ras_if = NULL;
+	struct ras_dispatch_if ih_data = {
+		.entry = entry,
+	};
+
+	rlc_status0 = RREG32(SOC15_REG_OFFSET(GC, 0, regRLC_RLCS_FED_STATUS_0));
+	rlc_status1 = RREG32(SOC15_REG_OFFSET(GC, 0, regRLC_RLCS_FED_STATUS_1));
+
+	if (!rlc_status0 && !rlc_status1) {
+		dev_warn(adev->dev, "RLC_GC_FED irq is generated, but rlc_status0 and rlc_status1 are empty!\n");
+		return 0;
+	}
+
+	/* Use RLC_RLCS_FED_STATUS_0/1 to distinguish FED error block. */
+	if (REG_GET_FIELD(rlc_status0, RLC_RLCS_FED_STATUS_0, SDMA0_FED_ERR) ||
+	    REG_GET_FIELD(rlc_status0, RLC_RLCS_FED_STATUS_0, SDMA1_FED_ERR))
+		ras_if = adev->sdma.ras_if;
+	else
+		ras_if = adev->gfx.ras_if;
+
+	if (!ras_if) {
+		dev_err(adev->dev, "Gfx or sdma ras block not initialized, rlc_status0:0x%x.\n",
+				rlc_status0);
+		return -EINVAL;
+	}
+
+	dev_warn(adev->dev, "RLC %s FED IRQ\n", ras_if->name);
+
+	if (!amdgpu_sriov_vf(adev)) {
+		ih_data.head = *ras_if;
+		amdgpu_ras_interrupt_dispatch(adev, &ih_data);
+	} else {
+		if (adev->virt.ops && adev->virt.ops->ras_poison_handler)
+			adev->virt.ops->ras_poison_handler(adev, ras_if->block);
+		else
+			dev_warn(adev->dev,
+				"No ras_poison_handler interface in SRIOV for %s!\n", ras_if->name);
+	}
+
+	return 0;
+}
+
+static int gfx_v11_0_3_poison_consumption_handler(struct amdgpu_device *adev,
+					struct amdgpu_iv_entry *entry)
+{
+	/* Workaround: when vmid and pasid are both zero, trigger gpu reset in KGD. */
+	if (entry && (entry->client_id == SOC21_IH_CLIENTID_GFX) &&
+	    (entry->src_id == GFX_11_0_0__SRCID__RLC_GC_FED_INTERRUPT) &&
+	     !entry->vmid && !entry->pasid) {
+		struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+		uint32_t rlc_status0 = 0;
+
+		rlc_status0 = RREG32_SOC15(GC, 0, regRLC_RLCS_FED_STATUS_0);
+
+		if (REG_GET_FIELD(rlc_status0, RLC_RLCS_FED_STATUS_0, SDMA0_FED_ERR) ||
+		    REG_GET_FIELD(rlc_status0, RLC_RLCS_FED_STATUS_0, SDMA1_FED_ERR)) {
+			struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+
+			ras->gpu_reset_flags |= AMDGPU_RAS_GPU_RESET_MODE2_RESET;
+		}
+
+		if (con && !amdgpu_ras_is_rma(adev))
+			amdgpu_ras_reset_gpu(adev);
+	}
+
+	return 0;
+}
+
+struct amdgpu_gfx_ras gfx_v11_0_3_ras = {
+	.rlc_gc_fed_irq = gfx_v11_0_3_rlc_gc_fed_irq,
+	.poison_consumption_handler = gfx_v11_0_3_poison_consumption_handler,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3.h b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3.h
new file mode 100644
index 000000000000..672c7920b3d0
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFX_V11_0_3_H__
+#define __GFX_V11_0_3_H__
+
+extern struct amdgpu_gfx_ras gfx_v11_0_3_ras;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3_cleaner_shader.asm b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3_cleaner_shader.asm
new file mode 100644
index 000000000000..9b90b66368c7
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3_cleaner_shader.asm
@@ -0,0 +1,118 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+// This shader is to clean LDS, SGPRs and VGPRs. It is  first 64 Dwords or 256 bytes of 192 Dwords cleaner shader.
+//To turn this shader program on for complitaion change this to main and lower shader main to main_1
+
+// Navi3 : Clear SGPRs, VGPRs and LDS
+//   Launch 32 waves per CU (16 per SIMD) as a workgroup (threadgroup) to fill every wave slot
+//   Waves are "wave32" and have 64 VGPRs each, which uses all 1024 VGPRs per SIMD
+//   Waves are launched in "CU" mode, and the workgroup shares 64KB of LDS (half of the WGP's LDS)
+//      It takes 2 workgroups to use all of LDS: one on each CU of the WGP
+//   Each wave clears SGPRs 0 - 107
+//   Each wave clears VGPRs 0 - 63
+//   The first wave of the workgroup clears its 64KB of LDS
+//   The shader starts with "S_BARRIER" to ensure SPI has launched all waves of the workgroup
+//       before any wave in the workgroup could end.  Without this, it is possible not all SGPRs get cleared.
+
+shader main
+  asic(GFX11)
+  type(CS)
+  wave_size(32)
+// Note: original source code from SQ team
+
+// Takes about 2500 clocks to run.
+//   (theorhetical fastest = 1024clks vgpr + 640lds = 1660 clks)
+//
+  S_BARRIER
+
+  //
+  // CLEAR VGPRs
+  //
+  s_mov_b32     m0, 0x00000058  // Loop 96/8=12 times  (loop unrolled for performance)
+
+label_0005:
+  v_movreld_b32     v0, 0
+  v_movreld_b32     v1, 0
+  v_movreld_b32     v2, 0
+  v_movreld_b32     v3, 0
+  v_movreld_b32     v4, 0
+  v_movreld_b32     v5, 0
+  v_movreld_b32     v6, 0
+  v_movreld_b32     v7, 0
+  s_sub_u32     m0, m0, 8
+  s_cbranch_scc0  label_0005
+  //
+  //
+
+  s_mov_b32     s2, 0x80000000                      // Bit31 is first_wave
+  s_and_b32     s2, s2, s0                          // sgpr0 has tg_size (first_wave) term as in ucode only COMPUTE_PGM_RSRC2.tg_size_en is set
+  s_cbranch_scc0  label_0023                        // Clean LDS if its first wave of ThreadGroup/WorkGroup
+  // CLEAR LDS
+  //
+  s_mov_b32 exec_lo, 0xffffffff
+  s_mov_b32 exec_hi, 0xffffffff
+  v_mbcnt_lo_u32_b32  v1, exec_hi, 0          // Set V1 to thread-ID (0..63)
+  v_mbcnt_hi_u32_b32  v1, exec_lo, v1        // Set V1 to thread-ID (0..63)
+  v_mul_u32_u24  v1, 0x00000008, v1          // * 8, so each thread is a double-dword address (8byte)
+  s_mov_b32     s2, 0x00000003f                    // 64 loop iterations
+  s_mov_b32     m0, 0xffffffff
+  // Clear all of LDS space
+  // Each FirstWave of WorkGroup clears 64kbyte block
+
+label_001F:
+  ds_write2_b64  v1, v[2:3], v[2:3] offset1:32
+  ds_write2_b64  v1, v[4:5], v[4:5] offset0:64 offset1:96
+  v_add_co_u32     v1, vcc, 0x00000400, v1
+  s_sub_u32     s2, s2, 1
+  s_cbranch_scc0  label_001F
+  //
+  // CLEAR SGPRs
+  //
+label_0023:
+  s_mov_b32     m0, 0x00000068  // Loop 108/4=27 times  (loop unrolled for performance)
+label_sgpr_loop:
+  s_movreld_b32     s0, 0
+  s_movreld_b32     s1, 0
+  s_movreld_b32     s2, 0
+  s_movreld_b32     s3, 0
+  s_sub_u32         m0, m0, 4
+  s_cbranch_scc0  label_sgpr_loop
+
+  //clear vcc
+  s_mov_b64 vcc, 0          //clear vcc
+  s_mov_b32 flat_scratch_lo, 0   //clear  flat scratch lo SGPR
+  s_mov_b32 flat_scratch_hi, 0   //clear  flat scratch hi SGPR
+  s_mov_b64 ttmp0, 0        //Clear ttmp0 and ttmp1
+  s_mov_b64 ttmp2, 0        //Clear ttmp2 and ttmp3
+  s_mov_b64 ttmp4, 0        //Clear ttmp4 and ttmp5
+  s_mov_b64 ttmp6, 0        //Clear ttmp6 and ttmp7
+  s_mov_b64 ttmp8, 0        //Clear ttmp8 and ttmp9
+  s_mov_b64 ttmp10, 0       //Clear ttmp10 and ttmp11
+  s_mov_b64 ttmp12, 0       //Clear ttmp12 and ttmp13
+  s_mov_b64 ttmp14, 0       //Clear ttmp14 and ttmp15
+
+ s_endpgm
+
+end
+
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_cleaner_shader.h b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_cleaner_shader.h
new file mode 100644
index 000000000000..3218cc04f543
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_cleaner_shader.h
@@ -0,0 +1,56 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/* Define the cleaner shader gfx_11_0_3 */
+static const u32 gfx_11_0_3_cleaner_shader_hex[] = {
+	0xb0804006, 0xbe8200ff,
+	0x00000058, 0xbefd0080,
+	0x7e008480, 0x7e028480,
+	0x7e048480, 0x7e068480,
+	0x7e088480, 0x7e0a8480,
+	0x7e0c8480, 0x7e0e8480,
+	0xbefd0002, 0x80828802,
+	0xbfa1fff5, 0xbe8200ff,
+	0x80000000, 0x8b020002,
+	0xbfa10012, 0xbefe00c1,
+	0xbeff00c1, 0xd71f0001,
+	0x0001007f, 0xd7200001,
+	0x0002027e, 0x16020288,
+	0xbe8200bf, 0xbefd00c1,
+	0xd9382000, 0x00020201,
+	0xd9386040, 0x00040401,
+	0xd7006a01, 0x000202ff,
+	0x00000400, 0x80828102,
+	0xbfa1fff7, 0xbefd00ff,
+	0x00000068, 0xbe804280,
+	0xbe814280, 0xbe824280,
+	0xbe834280, 0x80fd847d,
+	0xbfa1fffa, 0xbeea0180,
+	0xbeec0180, 0xbeee0180,
+	0xbef00180, 0xbef20180,
+	0xbef40180, 0xbef60180,
+	0xbef80180, 0xbefa0180,
+	0xbfb00000, 0xbf9f0000,
+	0xbf9f0000, 0xbf9f0000,
+	0xbf9f0000, 0xbf9f0000,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c
new file mode 100644
index 000000000000..710ec9c34e43
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c
@@ -0,0 +1,5793 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/delay.h>
+#include <linux/kernel.h>
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+#include "amdgpu.h"
+#include "amdgpu_gfx.h"
+#include "amdgpu_psp.h"
+#include "amdgpu_smu.h"
+#include "imu_v12_0.h"
+#include "soc24.h"
+#include "nvd.h"
+
+#include "gc/gc_12_0_0_offset.h"
+#include "gc/gc_12_0_0_sh_mask.h"
+#include "soc24_enum.h"
+#include "ivsrcid/gfx/irqsrcs_gfx_12_0_0.h"
+
+#include "soc15.h"
+#include "clearstate_gfx12.h"
+#include "v12_structs.h"
+#include "gfx_v12_0.h"
+#include "nbif_v6_3_1.h"
+#include "mes_v12_0.h"
+#include "mes_userqueue.h"
+#include "amdgpu_userq_fence.h"
+
+#define GFX12_NUM_GFX_RINGS	1
+#define GFX12_MEC_HPD_SIZE	2048
+
+#define RLCG_UCODE_LOADING_START_ADDRESS	0x00002000L
+
+#define regCP_GFX_MQD_CONTROL_DEFAULT                                             0x00000100
+#define regCP_GFX_HQD_VMID_DEFAULT                                                0x00000000
+#define regCP_GFX_HQD_QUEUE_PRIORITY_DEFAULT                                      0x00000000
+#define regCP_GFX_HQD_QUANTUM_DEFAULT                                             0x00000a01
+#define regCP_GFX_HQD_CNTL_DEFAULT                                                0x00f00000
+#define regCP_RB_DOORBELL_CONTROL_DEFAULT                                         0x00000000
+#define regCP_GFX_HQD_RPTR_DEFAULT                                                0x00000000
+
+#define regCP_HQD_EOP_CONTROL_DEFAULT                                             0x00000006
+#define regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT                                     0x00000000
+#define regCP_MQD_CONTROL_DEFAULT                                                 0x00000100
+#define regCP_HQD_PQ_CONTROL_DEFAULT                                              0x00308509
+#define regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT                                     0x00000000
+#define regCP_HQD_PQ_RPTR_DEFAULT                                                 0x00000000
+#define regCP_HQD_PERSISTENT_STATE_DEFAULT                                        0x0be05501
+#define regCP_HQD_IB_CONTROL_DEFAULT                                              0x00300000
+
+
+MODULE_FIRMWARE("amdgpu/gc_12_0_0_pfp.bin");
+MODULE_FIRMWARE("amdgpu/gc_12_0_0_me.bin");
+MODULE_FIRMWARE("amdgpu/gc_12_0_0_mec.bin");
+MODULE_FIRMWARE("amdgpu/gc_12_0_0_rlc.bin");
+MODULE_FIRMWARE("amdgpu/gc_12_0_0_toc.bin");
+MODULE_FIRMWARE("amdgpu/gc_12_0_1_pfp.bin");
+MODULE_FIRMWARE("amdgpu/gc_12_0_1_me.bin");
+MODULE_FIRMWARE("amdgpu/gc_12_0_1_mec.bin");
+MODULE_FIRMWARE("amdgpu/gc_12_0_1_rlc.bin");
+MODULE_FIRMWARE("amdgpu/gc_12_0_1_rlc_kicker.bin");
+MODULE_FIRMWARE("amdgpu/gc_12_0_1_toc.bin");
+
+static const struct amdgpu_hwip_reg_entry gc_reg_list_12_0[] = {
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS2),
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS3),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_STALLED_STAT1),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_STALLED_STAT2),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_STALLED_STAT3),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPC_STALLED_STAT1),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPF_STALLED_STAT1),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_BUSY_STAT),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPC_BUSY_STAT),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPF_BUSY_STAT),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPC_BUSY_STAT2),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPF_BUSY_STAT2),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPF_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_ERROR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HPD_STATUS0),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_RB_BASE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_RB_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_RB0_BASE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_RB0_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_RB0_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_CMD_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB2_CMD_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB2_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB2_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB2_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, regCPF_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regCPC_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regCPG_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regIA_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regIA_UTCL1_STATUS_2),
+	SOC15_REG_ENTRY_STR(GC, 0, regPA_CL_CNTL_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regRMI_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regSQC_CACHES),
+	SOC15_REG_ENTRY_STR(GC, 0, regSQG_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regWD_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regGCVM_L2_PROTECTION_FAULT_STATUS_LO32),
+	SOC15_REG_ENTRY_STR(GC, 0, regGCVM_L2_PROTECTION_FAULT_STATUS_HI32),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_DEBUG),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_INSTR_PNTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_INSTR_PNTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_INSTR_PNTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPC_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_RS64_INSTR_PNTR0),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_RS64_INSTR_PNTR1),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_RS64_INSTR_PNTR),
+	/* cp header registers */
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	/* SE status registers */
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS_SE0),
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS_SE1),
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS_SE2),
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS_SE3)
+};
+
+static const struct amdgpu_hwip_reg_entry gc_cp_reg_list_12[] = {
+	/* compute registers */
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_VMID),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PERSISTENT_STATE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PIPE_PRIORITY),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_QUEUE_PRIORITY),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_QUANTUM),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_BASE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_WPTR_POLL_ADDR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_WPTR_POLL_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_IB_BASE_ADDR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_IB_BASE_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_IB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_IB_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_DEQUEUE_REQUEST),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_BASE_ADDR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_BASE_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_EVENTS),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CTX_SAVE_BASE_ADDR_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CTX_SAVE_BASE_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CTX_SAVE_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CNTL_STACK_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CNTL_STACK_SIZE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_WG_STATE_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CTX_SAVE_SIZE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_GDS_RESOURCE_STATE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_ERROR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_WPTR_MEM),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_WPTR_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_SUSPEND_CNTL_STACK_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_SUSPEND_CNTL_STACK_DW_CNT),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_SUSPEND_WG_STATE_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_DEQUEUE_STATUS),
+	/* cp header registers */
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+};
+
+static const struct amdgpu_hwip_reg_entry gc_gfx_queue_reg_list_12[] = {
+	/* gfx queue registers */
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_ACTIVE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_VMID),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_QUEUE_PRIORITY),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_QUANTUM),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_BASE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_CSMD_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_DEQUEUE_REQUEST),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_MAPPED),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_QUE_MGR_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_HQ_CONTROL0),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_HQD_HQ_STATUS0),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_MQD_BASE_ADDR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_MQD_BASE_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_RB_WPTR_POLL_ADDR_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_RB_WPTR_POLL_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_CMD_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_BUFSZ),
+	/* cp header registers */
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+};
+
+static const struct soc15_reg_golden golden_settings_gc_12_0_rev0[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regDB_MEM_CONFIG, 0x0000000f, 0x0000000f),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regCB_HW_CONTROL_1, 0x03000000, 0x03000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regGL2C_CTRL5, 0x00000070, 0x00000020)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_12_0[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regDB_MEM_CONFIG, 0x00008000, 0x00008000),
+};
+
+#define DEFAULT_SH_MEM_CONFIG \
+	((SH_MEM_ADDRESS_MODE_64 << SH_MEM_CONFIG__ADDRESS_MODE__SHIFT) | \
+	 (SH_MEM_ALIGNMENT_MODE_UNALIGNED << SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) | \
+	 (3 << SH_MEM_CONFIG__INITIAL_INST_PREFETCH__SHIFT))
+
+static void gfx_v12_0_disable_gpa_mode(struct amdgpu_device *adev);
+static void gfx_v12_0_set_ring_funcs(struct amdgpu_device *adev);
+static void gfx_v12_0_set_irq_funcs(struct amdgpu_device *adev);
+static void gfx_v12_0_set_rlc_funcs(struct amdgpu_device *adev);
+static void gfx_v12_0_set_mqd_funcs(struct amdgpu_device *adev);
+static void gfx_v12_0_set_imu_funcs(struct amdgpu_device *adev);
+static int gfx_v12_0_get_cu_info(struct amdgpu_device *adev,
+				 struct amdgpu_cu_info *cu_info);
+static uint64_t gfx_v12_0_get_gpu_clock_counter(struct amdgpu_device *adev);
+static void gfx_v12_0_select_se_sh(struct amdgpu_device *adev, u32 se_num,
+				   u32 sh_num, u32 instance, int xcc_id);
+static u32 gfx_v12_0_get_wgp_active_bitmap_per_sh(struct amdgpu_device *adev);
+
+static void gfx_v12_0_ring_emit_frame_cntl(struct amdgpu_ring *ring, bool start, bool secure);
+static void gfx_v12_0_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg,
+				     uint32_t val);
+static int gfx_v12_0_wait_for_rlc_autoload_complete(struct amdgpu_device *adev);
+static void gfx_v12_0_ring_invalidate_tlbs(struct amdgpu_ring *ring,
+					   uint16_t pasid, uint32_t flush_type,
+					   bool all_hub, uint8_t dst_sel);
+static void gfx_v12_0_set_safe_mode(struct amdgpu_device *adev, int xcc_id);
+static void gfx_v12_0_unset_safe_mode(struct amdgpu_device *adev, int xcc_id);
+static void gfx_v12_0_update_perf_clk(struct amdgpu_device *adev,
+				      bool enable);
+
+static void gfx_v12_0_kiq_set_resources(struct amdgpu_ring *kiq_ring,
+					uint64_t queue_mask)
+{
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_SET_RESOURCES, 6));
+	amdgpu_ring_write(kiq_ring, PACKET3_SET_RESOURCES_VMID_MASK(0) |
+			  PACKET3_SET_RESOURCES_QUEUE_TYPE(0));	/* vmid_mask:0 queue_type:0 (KIQ) */
+	amdgpu_ring_write(kiq_ring, lower_32_bits(queue_mask));	/* queue mask lo */
+	amdgpu_ring_write(kiq_ring, upper_32_bits(queue_mask));	/* queue mask hi */
+	amdgpu_ring_write(kiq_ring, 0);	/* gws mask lo */
+	amdgpu_ring_write(kiq_ring, 0);	/* gws mask hi */
+	amdgpu_ring_write(kiq_ring, 0);	/* oac mask */
+	amdgpu_ring_write(kiq_ring, 0);
+}
+
+static void gfx_v12_0_kiq_map_queues(struct amdgpu_ring *kiq_ring,
+				     struct amdgpu_ring *ring)
+{
+	uint64_t mqd_addr = amdgpu_bo_gpu_offset(ring->mqd_obj);
+	uint64_t wptr_addr = ring->wptr_gpu_addr;
+	uint32_t me = 0, eng_sel = 0;
+
+	switch (ring->funcs->type) {
+	case AMDGPU_RING_TYPE_COMPUTE:
+		me = 1;
+		eng_sel = 0;
+		break;
+	case AMDGPU_RING_TYPE_GFX:
+		me = 0;
+		eng_sel = 4;
+		break;
+	case AMDGPU_RING_TYPE_MES:
+		me = 2;
+		eng_sel = 5;
+		break;
+	default:
+		WARN_ON(1);
+	}
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_MAP_QUEUES, 5));
+	/* Q_sel:0, vmid:0, vidmem: 1, engine:0, num_Q:1*/
+	amdgpu_ring_write(kiq_ring, /* Q_sel: 0, vmid: 0, engine: 0, num_Q: 1 */
+			  PACKET3_MAP_QUEUES_QUEUE_SEL(0) | /* Queue_Sel */
+			  PACKET3_MAP_QUEUES_VMID(0) | /* VMID */
+			  PACKET3_MAP_QUEUES_QUEUE(ring->queue) |
+			  PACKET3_MAP_QUEUES_PIPE(ring->pipe) |
+			  PACKET3_MAP_QUEUES_ME((me)) |
+			  PACKET3_MAP_QUEUES_QUEUE_TYPE(0) | /*queue_type: normal compute queue */
+			  PACKET3_MAP_QUEUES_ALLOC_FORMAT(0) | /* alloc format: all_on_one_pipe */
+			  PACKET3_MAP_QUEUES_ENGINE_SEL(eng_sel) |
+			  PACKET3_MAP_QUEUES_NUM_QUEUES(1)); /* num_queues: must be 1 */
+	amdgpu_ring_write(kiq_ring, PACKET3_MAP_QUEUES_DOORBELL_OFFSET(ring->doorbell_index));
+	amdgpu_ring_write(kiq_ring, lower_32_bits(mqd_addr));
+	amdgpu_ring_write(kiq_ring, upper_32_bits(mqd_addr));
+	amdgpu_ring_write(kiq_ring, lower_32_bits(wptr_addr));
+	amdgpu_ring_write(kiq_ring, upper_32_bits(wptr_addr));
+}
+
+static void gfx_v12_0_kiq_unmap_queues(struct amdgpu_ring *kiq_ring,
+				       struct amdgpu_ring *ring,
+				       enum amdgpu_unmap_queues_action action,
+				       u64 gpu_addr, u64 seq)
+{
+	struct amdgpu_device *adev = kiq_ring->adev;
+	uint32_t eng_sel = ring->funcs->type == AMDGPU_RING_TYPE_GFX ? 4 : 0;
+
+	if (adev->enable_mes && !adev->gfx.kiq[0].ring.sched.ready) {
+		amdgpu_mes_unmap_legacy_queue(adev, ring, action, gpu_addr, seq);
+		return;
+	}
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_UNMAP_QUEUES, 4));
+	amdgpu_ring_write(kiq_ring, /* Q_sel: 0, vmid: 0, engine: 0, num_Q: 1 */
+			  PACKET3_UNMAP_QUEUES_ACTION(action) |
+			  PACKET3_UNMAP_QUEUES_QUEUE_SEL(0) |
+			  PACKET3_UNMAP_QUEUES_ENGINE_SEL(eng_sel) |
+			  PACKET3_UNMAP_QUEUES_NUM_QUEUES(1));
+	amdgpu_ring_write(kiq_ring,
+		  PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET0(ring->doorbell_index));
+
+	if (action == PREEMPT_QUEUES_NO_UNMAP) {
+		amdgpu_ring_write(kiq_ring, lower_32_bits(gpu_addr));
+		amdgpu_ring_write(kiq_ring, upper_32_bits(gpu_addr));
+		amdgpu_ring_write(kiq_ring, seq);
+	} else {
+		amdgpu_ring_write(kiq_ring, 0);
+		amdgpu_ring_write(kiq_ring, 0);
+		amdgpu_ring_write(kiq_ring, 0);
+	}
+}
+
+static void gfx_v12_0_kiq_query_status(struct amdgpu_ring *kiq_ring,
+				       struct amdgpu_ring *ring,
+				       u64 addr, u64 seq)
+{
+	uint32_t eng_sel = ring->funcs->type == AMDGPU_RING_TYPE_GFX ? 4 : 0;
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_QUERY_STATUS, 5));
+	amdgpu_ring_write(kiq_ring,
+			  PACKET3_QUERY_STATUS_CONTEXT_ID(0) |
+			  PACKET3_QUERY_STATUS_INTERRUPT_SEL(0) |
+			  PACKET3_QUERY_STATUS_COMMAND(2));
+	amdgpu_ring_write(kiq_ring, /* Q_sel: 0, vmid: 0, engine: 0, num_Q: 1 */
+			  PACKET3_QUERY_STATUS_DOORBELL_OFFSET(ring->doorbell_index) |
+			  PACKET3_QUERY_STATUS_ENG_SEL(eng_sel));
+	amdgpu_ring_write(kiq_ring, lower_32_bits(addr));
+	amdgpu_ring_write(kiq_ring, upper_32_bits(addr));
+	amdgpu_ring_write(kiq_ring, lower_32_bits(seq));
+	amdgpu_ring_write(kiq_ring, upper_32_bits(seq));
+}
+
+static void gfx_v12_0_kiq_invalidate_tlbs(struct amdgpu_ring *kiq_ring,
+					  uint16_t pasid,
+					  uint32_t flush_type,
+					  bool all_hub)
+{
+	gfx_v12_0_ring_invalidate_tlbs(kiq_ring, pasid, flush_type, all_hub, 1);
+}
+
+static const struct kiq_pm4_funcs gfx_v12_0_kiq_pm4_funcs = {
+	.kiq_set_resources = gfx_v12_0_kiq_set_resources,
+	.kiq_map_queues = gfx_v12_0_kiq_map_queues,
+	.kiq_unmap_queues = gfx_v12_0_kiq_unmap_queues,
+	.kiq_query_status = gfx_v12_0_kiq_query_status,
+	.kiq_invalidate_tlbs = gfx_v12_0_kiq_invalidate_tlbs,
+	.set_resources_size = 8,
+	.map_queues_size = 7,
+	.unmap_queues_size = 6,
+	.query_status_size = 7,
+	.invalidate_tlbs_size = 2,
+};
+
+static void gfx_v12_0_set_kiq_pm4_funcs(struct amdgpu_device *adev)
+{
+	adev->gfx.kiq[0].pmf = &gfx_v12_0_kiq_pm4_funcs;
+}
+
+static void gfx_v12_0_wait_reg_mem(struct amdgpu_ring *ring, int eng_sel,
+				   int mem_space, int opt, uint32_t addr0,
+				   uint32_t addr1, uint32_t ref,
+				   uint32_t mask, uint32_t inv)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+	amdgpu_ring_write(ring,
+			  /* memory (1) or register (0) */
+			  (WAIT_REG_MEM_MEM_SPACE(mem_space) |
+			   WAIT_REG_MEM_OPERATION(opt) | /* wait */
+			   WAIT_REG_MEM_FUNCTION(3) |  /* equal */
+			   WAIT_REG_MEM_ENGINE(eng_sel)));
+
+	if (mem_space)
+		BUG_ON(addr0 & 0x3); /* Dword align */
+	amdgpu_ring_write(ring, addr0);
+	amdgpu_ring_write(ring, addr1);
+	amdgpu_ring_write(ring, ref);
+	amdgpu_ring_write(ring, mask);
+	amdgpu_ring_write(ring, inv); /* poll interval */
+}
+
+static int gfx_v12_0_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t scratch = SOC15_REG_OFFSET(GC, 0, regSCRATCH_REG0);
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	WREG32(scratch, 0xCAFEDEAD);
+	r = amdgpu_ring_alloc(ring, 5);
+	if (r) {
+		dev_err(adev->dev,
+			"amdgpu: cp failed to lock ring %d (%d).\n",
+			ring->idx, r);
+		return r;
+	}
+
+	if (ring->funcs->type == AMDGPU_RING_TYPE_KIQ) {
+		gfx_v12_0_ring_emit_wreg(ring, scratch, 0xDEADBEEF);
+	} else {
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SET_UCONFIG_REG, 1));
+		amdgpu_ring_write(ring, scratch -
+				  PACKET3_SET_UCONFIG_REG_START);
+		amdgpu_ring_write(ring, 0xDEADBEEF);
+	}
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(scratch);
+		if (tmp == 0xDEADBEEF)
+			break;
+		if (amdgpu_emu_mode == 1)
+			msleep(1);
+		else
+			udelay(1);
+	}
+
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+	return r;
+}
+
+static int gfx_v12_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct dma_fence *f = NULL;
+	unsigned index;
+	uint64_t gpu_addr;
+	uint32_t *cpu_ptr;
+	long r;
+
+	/* MES KIQ fw hasn't indirect buffer support for now */
+	if (adev->enable_mes_kiq &&
+	    ring->funcs->type == AMDGPU_RING_TYPE_KIQ)
+		return 0;
+
+	memset(&ib, 0, sizeof(ib));
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r)
+		return r;
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	adev->wb.wb[index] = cpu_to_le32(0xCAFEDEAD);
+	cpu_ptr = &adev->wb.wb[index];
+
+	r = amdgpu_ib_get(adev, NULL, 16, AMDGPU_IB_POOL_DIRECT, &ib);
+	if (r) {
+		dev_err(adev->dev, "amdgpu: failed to get ib (%ld).\n", r);
+		goto err1;
+	}
+
+	ib.ptr[0] = PACKET3(PACKET3_WRITE_DATA, 3);
+	ib.ptr[1] = WRITE_DATA_DST_SEL(5) | WR_CONFIRM;
+	ib.ptr[2] = lower_32_bits(gpu_addr);
+	ib.ptr[3] = upper_32_bits(gpu_addr);
+	ib.ptr[4] = 0xDEADBEEF;
+	ib.length_dw = 5;
+
+	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
+	if (r)
+		goto err2;
+
+	r = dma_fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		r = -ETIMEDOUT;
+		goto err2;
+	} else if (r < 0) {
+		goto err2;
+	}
+
+	if (le32_to_cpu(*cpu_ptr) == 0xDEADBEEF)
+		r = 0;
+	else
+		r = -EINVAL;
+err2:
+	amdgpu_ib_free(&ib, NULL);
+	dma_fence_put(f);
+err1:
+	amdgpu_device_wb_free(adev, index);
+	return r;
+}
+
+static void gfx_v12_0_free_microcode(struct amdgpu_device *adev)
+{
+	amdgpu_ucode_release(&adev->gfx.pfp_fw);
+	amdgpu_ucode_release(&adev->gfx.me_fw);
+	amdgpu_ucode_release(&adev->gfx.rlc_fw);
+	amdgpu_ucode_release(&adev->gfx.mec_fw);
+
+	kfree(adev->gfx.rlc.register_list_format);
+}
+
+static int gfx_v12_0_init_toc_microcode(struct amdgpu_device *adev, const char *ucode_prefix)
+{
+	const struct psp_firmware_header_v1_0 *toc_hdr;
+	int err = 0;
+
+	err = amdgpu_ucode_request(adev, &adev->psp.toc_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_toc.bin", ucode_prefix);
+	if (err)
+		goto out;
+
+	toc_hdr = (const struct psp_firmware_header_v1_0 *)adev->psp.toc_fw->data;
+	adev->psp.toc.fw_version = le32_to_cpu(toc_hdr->header.ucode_version);
+	adev->psp.toc.feature_version = le32_to_cpu(toc_hdr->sos.fw_version);
+	adev->psp.toc.size_bytes = le32_to_cpu(toc_hdr->header.ucode_size_bytes);
+	adev->psp.toc.start_addr = (uint8_t *)toc_hdr +
+			le32_to_cpu(toc_hdr->header.ucode_array_offset_bytes);
+	return 0;
+out:
+	amdgpu_ucode_release(&adev->psp.toc_fw);
+	return err;
+}
+
+static int gfx_v12_0_init_microcode(struct amdgpu_device *adev)
+{
+	char ucode_prefix[30];
+	int err;
+	const struct rlc_firmware_header_v2_0 *rlc_hdr;
+	uint16_t version_major;
+	uint16_t version_minor;
+
+	DRM_DEBUG("\n");
+
+	amdgpu_ucode_ip_version_decode(adev, GC_HWIP, ucode_prefix, sizeof(ucode_prefix));
+
+	err = amdgpu_ucode_request(adev, &adev->gfx.pfp_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_pfp.bin", ucode_prefix);
+	if (err)
+		goto out;
+	amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_RS64_PFP);
+	amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_RS64_PFP_P0_STACK);
+
+	err = amdgpu_ucode_request(adev, &adev->gfx.me_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_me.bin", ucode_prefix);
+	if (err)
+		goto out;
+	amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_RS64_ME);
+	amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_RS64_ME_P0_STACK);
+
+	if (!amdgpu_sriov_vf(adev)) {
+		if (amdgpu_is_kicker_fw(adev))
+			err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw,
+						   AMDGPU_UCODE_REQUIRED,
+						   "amdgpu/%s_rlc_kicker.bin", ucode_prefix);
+		else
+			err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw,
+						   AMDGPU_UCODE_REQUIRED,
+						   "amdgpu/%s_rlc.bin", ucode_prefix);
+		if (err)
+			goto out;
+		rlc_hdr = (const struct rlc_firmware_header_v2_0 *)adev->gfx.rlc_fw->data;
+		version_major = le16_to_cpu(rlc_hdr->header.header_version_major);
+		version_minor = le16_to_cpu(rlc_hdr->header.header_version_minor);
+		err = amdgpu_gfx_rlc_init_microcode(adev, version_major, version_minor);
+		if (err)
+			goto out;
+	}
+
+	err = amdgpu_ucode_request(adev, &adev->gfx.mec_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_mec.bin", ucode_prefix);
+	if (err)
+		goto out;
+	amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_RS64_MEC);
+	amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_RS64_MEC_P0_STACK);
+	amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_RS64_MEC_P1_STACK);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO)
+		err = gfx_v12_0_init_toc_microcode(adev, ucode_prefix);
+
+	/* only one MEC for gfx 12 */
+	adev->gfx.mec2_fw = NULL;
+
+	if (adev->gfx.imu.funcs) {
+		if (adev->gfx.imu.funcs->init_microcode) {
+			err = adev->gfx.imu.funcs->init_microcode(adev);
+			if (err)
+				dev_err(adev->dev, "Failed to load imu firmware!\n");
+		}
+	}
+
+out:
+	if (err) {
+		amdgpu_ucode_release(&adev->gfx.pfp_fw);
+		amdgpu_ucode_release(&adev->gfx.me_fw);
+		amdgpu_ucode_release(&adev->gfx.rlc_fw);
+		amdgpu_ucode_release(&adev->gfx.mec_fw);
+	}
+
+	return err;
+}
+
+static u32 gfx_v12_0_get_csb_size(struct amdgpu_device *adev)
+{
+	u32 count = 0;
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+
+	count += 1;
+
+	for (sect = gfx12_cs_data; sect->section != NULL; ++sect) {
+		if (sect->id == SECT_CONTEXT) {
+			for (ext = sect->section; ext->extent != NULL; ++ext)
+				count += 2 + ext->reg_count;
+		} else
+			return 0;
+	}
+
+	return count;
+}
+
+static void gfx_v12_0_get_csb_buffer(struct amdgpu_device *adev, u32 *buffer)
+{
+	u32 count = 0, clustercount = 0, i;
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+
+	if (adev->gfx.rlc.cs_data == NULL)
+		return;
+	if (buffer == NULL)
+		return;
+
+	count += 1;
+
+	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
+		if (sect->id == SECT_CONTEXT) {
+			for (ext = sect->section; ext->extent != NULL; ++ext) {
+				clustercount++;
+				buffer[count++] = ext->reg_count;
+				buffer[count++] = ext->reg_index;
+
+				for (i = 0; i < ext->reg_count; i++)
+					buffer[count++] = cpu_to_le32(ext->extent[i]);
+			}
+		} else
+			return;
+	}
+
+	buffer[0] = clustercount;
+}
+
+static void gfx_v12_0_rlc_fini(struct amdgpu_device *adev)
+{
+	/* clear state block */
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj,
+			&adev->gfx.rlc.clear_state_gpu_addr,
+			(void **)&adev->gfx.rlc.cs_ptr);
+
+	/* jump table block */
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj,
+			&adev->gfx.rlc.cp_table_gpu_addr,
+			(void **)&adev->gfx.rlc.cp_table_ptr);
+}
+
+static void gfx_v12_0_init_rlcg_reg_access_ctrl(struct amdgpu_device *adev)
+{
+	struct amdgpu_rlcg_reg_access_ctrl *reg_access_ctrl;
+
+	reg_access_ctrl = &adev->gfx.rlc.reg_access_ctrl[0];
+	reg_access_ctrl->scratch_reg0 = SOC15_REG_OFFSET(GC, 0, regSCRATCH_REG0);
+	reg_access_ctrl->scratch_reg1 = SOC15_REG_OFFSET(GC, 0, regSCRATCH_REG1);
+	reg_access_ctrl->scratch_reg2 = SOC15_REG_OFFSET(GC, 0, regSCRATCH_REG2);
+	reg_access_ctrl->scratch_reg3 = SOC15_REG_OFFSET(GC, 0, regSCRATCH_REG3);
+	reg_access_ctrl->grbm_cntl = SOC15_REG_OFFSET(GC, 0, regGRBM_GFX_CNTL);
+	reg_access_ctrl->grbm_idx = SOC15_REG_OFFSET(GC, 0, regGRBM_GFX_INDEX);
+	reg_access_ctrl->spare_int = SOC15_REG_OFFSET(GC, 0, regRLC_SPARE_INT_0);
+	adev->gfx.rlc.rlcg_reg_access_supported = true;
+}
+
+static int gfx_v12_0_rlc_init(struct amdgpu_device *adev)
+{
+	const struct cs_section_def *cs_data;
+	int r;
+
+	adev->gfx.rlc.cs_data = gfx12_cs_data;
+
+	cs_data = adev->gfx.rlc.cs_data;
+
+	if (cs_data) {
+		/* init clear state block */
+		r = amdgpu_gfx_rlc_init_csb(adev);
+		if (r)
+			return r;
+	}
+
+	/* init spm vmid with 0xf */
+	if (adev->gfx.rlc.funcs->update_spm_vmid)
+		adev->gfx.rlc.funcs->update_spm_vmid(adev, NULL, 0xf);
+
+	return 0;
+}
+
+static void gfx_v12_0_mec_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->gfx.mec.hpd_eop_obj, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gfx.mec.mec_fw_obj, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gfx.mec.mec_fw_data_obj, NULL, NULL);
+}
+
+static void gfx_v12_0_me_init(struct amdgpu_device *adev)
+{
+	bitmap_zero(adev->gfx.me.queue_bitmap, AMDGPU_MAX_GFX_QUEUES);
+
+	amdgpu_gfx_graphics_queue_acquire(adev);
+}
+
+static int gfx_v12_0_mec_init(struct amdgpu_device *adev)
+{
+	int r;
+	u32 *hpd;
+	size_t mec_hpd_size;
+
+	bitmap_zero(adev->gfx.mec_bitmap[0].queue_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
+
+	/* take ownership of the relevant compute queues */
+	amdgpu_gfx_compute_queue_acquire(adev);
+	mec_hpd_size = adev->gfx.num_compute_rings * GFX12_MEC_HPD_SIZE;
+
+	if (mec_hpd_size) {
+		r = amdgpu_bo_create_reserved(adev, mec_hpd_size, PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_GTT,
+					      &adev->gfx.mec.hpd_eop_obj,
+					      &adev->gfx.mec.hpd_eop_gpu_addr,
+					      (void **)&hpd);
+		if (r) {
+			dev_warn(adev->dev, "(%d) create HDP EOP bo failed\n", r);
+			gfx_v12_0_mec_fini(adev);
+			return r;
+		}
+
+		memset(hpd, 0, mec_hpd_size);
+
+		amdgpu_bo_kunmap(adev->gfx.mec.hpd_eop_obj);
+		amdgpu_bo_unreserve(adev->gfx.mec.hpd_eop_obj);
+	}
+
+	return 0;
+}
+
+static uint32_t wave_read_ind(struct amdgpu_device *adev, uint32_t wave, uint32_t address)
+{
+	WREG32_SOC15(GC, 0, regSQ_IND_INDEX,
+		(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
+		(address << SQ_IND_INDEX__INDEX__SHIFT));
+	return RREG32_SOC15(GC, 0, regSQ_IND_DATA);
+}
+
+static void wave_read_regs(struct amdgpu_device *adev, uint32_t wave,
+			   uint32_t thread, uint32_t regno,
+			   uint32_t num, uint32_t *out)
+{
+	WREG32_SOC15(GC, 0, regSQ_IND_INDEX,
+		(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
+		(regno << SQ_IND_INDEX__INDEX__SHIFT) |
+		(thread << SQ_IND_INDEX__WORKITEM_ID__SHIFT) |
+		(SQ_IND_INDEX__AUTO_INCR_MASK));
+	while (num--)
+		*(out++) = RREG32_SOC15(GC, 0, regSQ_IND_DATA);
+}
+
+static void gfx_v12_0_read_wave_data(struct amdgpu_device *adev,
+				     uint32_t xcc_id,
+				     uint32_t simd, uint32_t wave,
+				     uint32_t *dst, int *no_fields)
+{
+	/* in gfx12 the SIMD_ID is specified as part of the INSTANCE
+	 * field when performing a select_se_sh so it should be
+	 * zero here */
+	WARN_ON(simd != 0);
+
+	/* type 4 wave data */
+	dst[(*no_fields)++] = 4;
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_STATUS);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_PC_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_PC_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_EXEC_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_EXEC_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_HW_ID1);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_HW_ID2);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_GPR_ALLOC);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_LDS_ALLOC);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_IB_STS);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_IB_STS2);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_IB_DBG1);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_M0);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_MODE);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_STATE_PRIV);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_EXCP_FLAG_PRIV);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_EXCP_FLAG_USER);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_TRAP_CTRL);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_ACTIVE);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_VALID_AND_IDLE);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_DVGPR_ALLOC_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_DVGPR_ALLOC_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, wave, ixSQ_WAVE_SCHED_MODE);
+}
+
+static void gfx_v12_0_read_wave_sgprs(struct amdgpu_device *adev,
+				      uint32_t xcc_id, uint32_t simd,
+				      uint32_t wave, uint32_t start,
+				      uint32_t size, uint32_t *dst)
+{
+	WARN_ON(simd != 0);
+
+	wave_read_regs(
+		adev, wave, 0, start + SQIND_WAVE_SGPRS_OFFSET, size,
+		dst);
+}
+
+static void gfx_v12_0_read_wave_vgprs(struct amdgpu_device *adev,
+				      uint32_t xcc_id, uint32_t simd,
+				      uint32_t wave, uint32_t thread,
+				      uint32_t start, uint32_t size,
+				      uint32_t *dst)
+{
+	wave_read_regs(
+		adev, wave, thread,
+		start + SQIND_WAVE_VGPRS_OFFSET, size, dst);
+}
+
+static void gfx_v12_0_select_me_pipe_q(struct amdgpu_device *adev,
+				       u32 me, u32 pipe, u32 q, u32 vm, u32 xcc_id)
+{
+	soc24_grbm_select(adev, me, pipe, q, vm);
+}
+
+/* all sizes are in bytes */
+#define MQD_SHADOW_BASE_SIZE      73728
+#define MQD_SHADOW_BASE_ALIGNMENT 256
+#define MQD_FWWORKAREA_SIZE       484
+#define MQD_FWWORKAREA_ALIGNMENT  256
+
+static void gfx_v12_0_get_gfx_shadow_info_nocheck(struct amdgpu_device *adev,
+						  struct amdgpu_gfx_shadow_info *shadow_info)
+{
+	shadow_info->shadow_size = MQD_SHADOW_BASE_SIZE;
+	shadow_info->shadow_alignment = MQD_SHADOW_BASE_ALIGNMENT;
+	shadow_info->csa_size = MQD_FWWORKAREA_SIZE;
+	shadow_info->csa_alignment = MQD_FWWORKAREA_ALIGNMENT;
+}
+
+static int gfx_v12_0_get_gfx_shadow_info(struct amdgpu_device *adev,
+					 struct amdgpu_gfx_shadow_info *shadow_info,
+					 bool skip_check)
+{
+	if (adev->gfx.cp_gfx_shadow || skip_check) {
+		gfx_v12_0_get_gfx_shadow_info_nocheck(adev, shadow_info);
+		return 0;
+	}
+
+	memset(shadow_info, 0, sizeof(struct amdgpu_gfx_shadow_info));
+	return -EINVAL;
+}
+
+static const struct amdgpu_gfx_funcs gfx_v12_0_gfx_funcs = {
+	.get_gpu_clock_counter = &gfx_v12_0_get_gpu_clock_counter,
+	.select_se_sh = &gfx_v12_0_select_se_sh,
+	.read_wave_data = &gfx_v12_0_read_wave_data,
+	.read_wave_sgprs = &gfx_v12_0_read_wave_sgprs,
+	.read_wave_vgprs = &gfx_v12_0_read_wave_vgprs,
+	.select_me_pipe_q = &gfx_v12_0_select_me_pipe_q,
+	.update_perfmon_mgcg = &gfx_v12_0_update_perf_clk,
+	.get_gfx_shadow_info = &gfx_v12_0_get_gfx_shadow_info,
+};
+
+static int gfx_v12_0_gpu_early_init(struct amdgpu_device *adev)
+{
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 0, 1):
+		adev->gfx.config.max_hw_contexts = 8;
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x4C0;
+		break;
+	default:
+		BUG();
+		break;
+	}
+
+	return 0;
+}
+
+static int gfx_v12_0_gfx_ring_init(struct amdgpu_device *adev, int ring_id,
+				   int me, int pipe, int queue)
+{
+	int r;
+	struct amdgpu_ring *ring;
+	unsigned int irq_type;
+
+	ring = &adev->gfx.gfx_ring[ring_id];
+
+	ring->me = me;
+	ring->pipe = pipe;
+	ring->queue = queue;
+
+	ring->ring_obj = NULL;
+	ring->use_doorbell = true;
+
+	if (!ring_id)
+		ring->doorbell_index = adev->doorbell_index.gfx_ring0 << 1;
+	else
+		ring->doorbell_index = adev->doorbell_index.gfx_ring1 << 1;
+	ring->vm_hub = AMDGPU_GFXHUB(0);
+	sprintf(ring->name, "gfx_%d.%d.%d", ring->me, ring->pipe, ring->queue);
+
+	irq_type = AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP + ring->pipe;
+	r = amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq, irq_type,
+			     AMDGPU_RING_PRIO_DEFAULT, NULL);
+	if (r)
+		return r;
+	return 0;
+}
+
+static int gfx_v12_0_compute_ring_init(struct amdgpu_device *adev, int ring_id,
+				       int mec, int pipe, int queue)
+{
+	int r;
+	unsigned irq_type;
+	struct amdgpu_ring *ring;
+	unsigned int hw_prio;
+
+	ring = &adev->gfx.compute_ring[ring_id];
+
+	/* mec0 is me1 */
+	ring->me = mec + 1;
+	ring->pipe = pipe;
+	ring->queue = queue;
+
+	ring->ring_obj = NULL;
+	ring->use_doorbell = true;
+	ring->doorbell_index = (adev->doorbell_index.mec_ring0 + ring_id) << 1;
+	ring->eop_gpu_addr = adev->gfx.mec.hpd_eop_gpu_addr
+				+ (ring_id * GFX12_MEC_HPD_SIZE);
+	ring->vm_hub = AMDGPU_GFXHUB(0);
+	sprintf(ring->name, "comp_%d.%d.%d", ring->me, ring->pipe, ring->queue);
+
+	irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP
+		+ ((ring->me - 1) * adev->gfx.mec.num_pipe_per_mec)
+		+ ring->pipe;
+	hw_prio = amdgpu_gfx_is_high_priority_compute_queue(adev, ring) ?
+			AMDGPU_GFX_PIPE_PRIO_HIGH : AMDGPU_GFX_PIPE_PRIO_NORMAL;
+	/* type-2 packets are deprecated on MEC, use type-3 instead */
+	r = amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq, irq_type,
+			     hw_prio, NULL);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static struct {
+	SOC24_FIRMWARE_ID	id;
+	unsigned int		offset;
+	unsigned int		size;
+	unsigned int		size_x16;
+} rlc_autoload_info[SOC24_FIRMWARE_ID_MAX];
+
+#define RLC_TOC_OFFSET_DWUNIT   8
+#define RLC_SIZE_MULTIPLE       1024
+#define RLC_TOC_UMF_SIZE_inM	23ULL
+#define RLC_TOC_FORMAT_API	165ULL
+
+static void gfx_v12_0_parse_rlc_toc(struct amdgpu_device *adev, void *rlc_toc)
+{
+	RLC_TABLE_OF_CONTENT_V2 *ucode = rlc_toc;
+
+	while (ucode && (ucode->id > SOC24_FIRMWARE_ID_INVALID)) {
+		rlc_autoload_info[ucode->id].id = ucode->id;
+		rlc_autoload_info[ucode->id].offset =
+			ucode->offset * RLC_TOC_OFFSET_DWUNIT * 4;
+		rlc_autoload_info[ucode->id].size =
+			ucode->size_x16 ? ucode->size * RLC_SIZE_MULTIPLE * 4 :
+					  ucode->size * 4;
+		ucode++;
+	}
+}
+
+static uint32_t gfx_v12_0_calc_toc_total_size(struct amdgpu_device *adev)
+{
+	uint32_t total_size = 0;
+	SOC24_FIRMWARE_ID id;
+
+	gfx_v12_0_parse_rlc_toc(adev, adev->psp.toc.start_addr);
+
+	for (id = SOC24_FIRMWARE_ID_RLC_G_UCODE; id < SOC24_FIRMWARE_ID_MAX; id++)
+		total_size += rlc_autoload_info[id].size;
+
+	/* In case the offset in rlc toc ucode is aligned */
+	if (total_size < rlc_autoload_info[SOC24_FIRMWARE_ID_MAX-1].offset)
+		total_size = rlc_autoload_info[SOC24_FIRMWARE_ID_MAX-1].offset +
+			rlc_autoload_info[SOC24_FIRMWARE_ID_MAX-1].size;
+	if (total_size < (RLC_TOC_UMF_SIZE_inM << 20))
+		total_size = RLC_TOC_UMF_SIZE_inM << 20;
+
+	return total_size;
+}
+
+static int gfx_v12_0_rlc_autoload_buffer_init(struct amdgpu_device *adev)
+{
+	int r;
+	uint32_t total_size;
+
+	total_size = gfx_v12_0_calc_toc_total_size(adev);
+
+	r = amdgpu_bo_create_reserved(adev, total_size, 64 * 1024,
+				      AMDGPU_GEM_DOMAIN_VRAM,
+				      &adev->gfx.rlc.rlc_autoload_bo,
+				      &adev->gfx.rlc.rlc_autoload_gpu_addr,
+				      (void **)&adev->gfx.rlc.rlc_autoload_ptr);
+
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create fw autoload bo\n", r);
+		return r;
+	}
+
+	return 0;
+}
+
+static void gfx_v12_0_rlc_backdoor_autoload_copy_ucode(struct amdgpu_device *adev,
+						       SOC24_FIRMWARE_ID id,
+						       const void *fw_data,
+						       uint32_t fw_size)
+{
+	uint32_t toc_offset;
+	uint32_t toc_fw_size;
+	char *ptr = adev->gfx.rlc.rlc_autoload_ptr;
+
+	if (id <= SOC24_FIRMWARE_ID_INVALID || id >= SOC24_FIRMWARE_ID_MAX)
+		return;
+
+	toc_offset = rlc_autoload_info[id].offset;
+	toc_fw_size = rlc_autoload_info[id].size;
+
+	if (fw_size == 0)
+		fw_size = toc_fw_size;
+
+	if (fw_size > toc_fw_size)
+		fw_size = toc_fw_size;
+
+	memcpy(ptr + toc_offset, fw_data, fw_size);
+
+	if (fw_size < toc_fw_size)
+		memset(ptr + toc_offset + fw_size, 0, toc_fw_size - fw_size);
+}
+
+static void
+gfx_v12_0_rlc_backdoor_autoload_copy_toc_ucode(struct amdgpu_device *adev)
+{
+	void *data;
+	uint32_t size;
+	uint32_t *toc_ptr;
+
+	data = adev->psp.toc.start_addr;
+	size = rlc_autoload_info[SOC24_FIRMWARE_ID_RLC_TOC].size;
+
+	toc_ptr = (uint32_t *)data + size / 4 - 2;
+	*toc_ptr = (RLC_TOC_FORMAT_API << 24) | 0x1;
+
+	gfx_v12_0_rlc_backdoor_autoload_copy_ucode(adev, SOC24_FIRMWARE_ID_RLC_TOC,
+						   data, size);
+}
+
+static void
+gfx_v12_0_rlc_backdoor_autoload_copy_gfx_ucode(struct amdgpu_device *adev)
+{
+	const __le32 *fw_data;
+	uint32_t fw_size;
+	const struct gfx_firmware_header_v2_0 *cpv2_hdr;
+	const struct rlc_firmware_header_v2_0 *rlc_hdr;
+	const struct rlc_firmware_header_v2_1 *rlcv21_hdr;
+	const struct rlc_firmware_header_v2_2 *rlcv22_hdr;
+	uint16_t version_major, version_minor;
+
+	/* pfp ucode */
+	cpv2_hdr = (const struct gfx_firmware_header_v2_0 *)
+		adev->gfx.pfp_fw->data;
+	/* instruction */
+	fw_data = (const __le32 *)(adev->gfx.pfp_fw->data +
+		le32_to_cpu(cpv2_hdr->ucode_offset_bytes));
+	fw_size = le32_to_cpu(cpv2_hdr->ucode_size_bytes);
+	gfx_v12_0_rlc_backdoor_autoload_copy_ucode(adev, SOC24_FIRMWARE_ID_RS64_PFP,
+						   fw_data, fw_size);
+	/* data */
+	fw_data = (const __le32 *)(adev->gfx.pfp_fw->data +
+		le32_to_cpu(cpv2_hdr->data_offset_bytes));
+	fw_size = le32_to_cpu(cpv2_hdr->data_size_bytes);
+	gfx_v12_0_rlc_backdoor_autoload_copy_ucode(adev, SOC24_FIRMWARE_ID_RS64_PFP_P0_STACK,
+						   fw_data, fw_size);
+	gfx_v12_0_rlc_backdoor_autoload_copy_ucode(adev, SOC24_FIRMWARE_ID_RS64_PFP_P1_STACK,
+						   fw_data, fw_size);
+	/* me ucode */
+	cpv2_hdr = (const struct gfx_firmware_header_v2_0 *)
+		adev->gfx.me_fw->data;
+	/* instruction */
+	fw_data = (const __le32 *)(adev->gfx.me_fw->data +
+		le32_to_cpu(cpv2_hdr->ucode_offset_bytes));
+	fw_size = le32_to_cpu(cpv2_hdr->ucode_size_bytes);
+	gfx_v12_0_rlc_backdoor_autoload_copy_ucode(adev, SOC24_FIRMWARE_ID_RS64_ME,
+						   fw_data, fw_size);
+	/* data */
+	fw_data = (const __le32 *)(adev->gfx.me_fw->data +
+		le32_to_cpu(cpv2_hdr->data_offset_bytes));
+	fw_size = le32_to_cpu(cpv2_hdr->data_size_bytes);
+	gfx_v12_0_rlc_backdoor_autoload_copy_ucode(adev, SOC24_FIRMWARE_ID_RS64_ME_P0_STACK,
+						   fw_data, fw_size);
+	gfx_v12_0_rlc_backdoor_autoload_copy_ucode(adev, SOC24_FIRMWARE_ID_RS64_ME_P1_STACK,
+						   fw_data, fw_size);
+	/* mec ucode */
+	cpv2_hdr = (const struct gfx_firmware_header_v2_0 *)
+		adev->gfx.mec_fw->data;
+	/* instruction */
+	fw_data = (const __le32 *) (adev->gfx.mec_fw->data +
+		le32_to_cpu(cpv2_hdr->ucode_offset_bytes));
+	fw_size = le32_to_cpu(cpv2_hdr->ucode_size_bytes);
+	gfx_v12_0_rlc_backdoor_autoload_copy_ucode(adev, SOC24_FIRMWARE_ID_RS64_MEC,
+						   fw_data, fw_size);
+	/* data */
+	fw_data = (const __le32 *) (adev->gfx.mec_fw->data +
+		le32_to_cpu(cpv2_hdr->data_offset_bytes));
+	fw_size = le32_to_cpu(cpv2_hdr->data_size_bytes);
+	gfx_v12_0_rlc_backdoor_autoload_copy_ucode(adev, SOC24_FIRMWARE_ID_RS64_MEC_P0_STACK,
+						   fw_data, fw_size);
+	gfx_v12_0_rlc_backdoor_autoload_copy_ucode(adev, SOC24_FIRMWARE_ID_RS64_MEC_P1_STACK,
+						   fw_data, fw_size);
+	gfx_v12_0_rlc_backdoor_autoload_copy_ucode(adev, SOC24_FIRMWARE_ID_RS64_MEC_P2_STACK,
+						   fw_data, fw_size);
+	gfx_v12_0_rlc_backdoor_autoload_copy_ucode(adev, SOC24_FIRMWARE_ID_RS64_MEC_P3_STACK,
+						   fw_data, fw_size);
+
+	/* rlc ucode */
+	rlc_hdr = (const struct rlc_firmware_header_v2_0 *)
+		adev->gfx.rlc_fw->data;
+	fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+			le32_to_cpu(rlc_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(rlc_hdr->header.ucode_size_bytes);
+	gfx_v12_0_rlc_backdoor_autoload_copy_ucode(adev, SOC24_FIRMWARE_ID_RLC_G_UCODE,
+						   fw_data, fw_size);
+
+	version_major = le16_to_cpu(rlc_hdr->header.header_version_major);
+	version_minor = le16_to_cpu(rlc_hdr->header.header_version_minor);
+	if (version_major == 2) {
+		if (version_minor >= 1) {
+			rlcv21_hdr = (const struct rlc_firmware_header_v2_1 *)adev->gfx.rlc_fw->data;
+
+			fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+					le32_to_cpu(rlcv21_hdr->save_restore_list_gpm_offset_bytes));
+			fw_size = le32_to_cpu(rlcv21_hdr->save_restore_list_gpm_size_bytes);
+			gfx_v12_0_rlc_backdoor_autoload_copy_ucode(adev, SOC24_FIRMWARE_ID_RLCG_SCRATCH,
+						   fw_data, fw_size);
+
+			fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+					le32_to_cpu(rlcv21_hdr->save_restore_list_srm_offset_bytes));
+			fw_size = le32_to_cpu(rlcv21_hdr->save_restore_list_srm_size_bytes);
+			gfx_v12_0_rlc_backdoor_autoload_copy_ucode(adev, SOC24_FIRMWARE_ID_RLC_SRM_ARAM,
+						   fw_data, fw_size);
+		}
+		if (version_minor >= 2) {
+			rlcv22_hdr = (const struct rlc_firmware_header_v2_2 *)adev->gfx.rlc_fw->data;
+
+			fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+					le32_to_cpu(rlcv22_hdr->rlc_iram_ucode_offset_bytes));
+			fw_size = le32_to_cpu(rlcv22_hdr->rlc_iram_ucode_size_bytes);
+			gfx_v12_0_rlc_backdoor_autoload_copy_ucode(adev, SOC24_FIRMWARE_ID_RLX6_UCODE,
+						   fw_data, fw_size);
+
+			fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+					le32_to_cpu(rlcv22_hdr->rlc_dram_ucode_offset_bytes));
+			fw_size = le32_to_cpu(rlcv22_hdr->rlc_dram_ucode_size_bytes);
+			gfx_v12_0_rlc_backdoor_autoload_copy_ucode(adev, SOC24_FIRMWARE_ID_RLX6_DRAM_BOOT,
+						   fw_data, fw_size);
+		}
+	}
+}
+
+static void
+gfx_v12_0_rlc_backdoor_autoload_copy_sdma_ucode(struct amdgpu_device *adev)
+{
+	const __le32 *fw_data;
+	uint32_t fw_size;
+	const struct sdma_firmware_header_v3_0 *sdma_hdr;
+
+	sdma_hdr = (const struct sdma_firmware_header_v3_0 *)
+		adev->sdma.instance[0].fw->data;
+	fw_data = (const __le32 *) (adev->sdma.instance[0].fw->data +
+			le32_to_cpu(sdma_hdr->ucode_offset_bytes));
+	fw_size = le32_to_cpu(sdma_hdr->ucode_size_bytes);
+
+	gfx_v12_0_rlc_backdoor_autoload_copy_ucode(adev, SOC24_FIRMWARE_ID_SDMA_UCODE_TH0,
+						   fw_data, fw_size);
+}
+
+static void
+gfx_v12_0_rlc_backdoor_autoload_copy_mes_ucode(struct amdgpu_device *adev)
+{
+	const __le32 *fw_data;
+	unsigned fw_size;
+	const struct mes_firmware_header_v1_0 *mes_hdr;
+	int pipe, ucode_id, data_id;
+
+	for (pipe = 0; pipe < 2; pipe++) {
+		if (pipe == 0) {
+			ucode_id = SOC24_FIRMWARE_ID_RS64_MES_P0;
+			data_id  = SOC24_FIRMWARE_ID_RS64_MES_P0_STACK;
+		} else {
+			ucode_id = SOC24_FIRMWARE_ID_RS64_MES_P1;
+			data_id  = SOC24_FIRMWARE_ID_RS64_MES_P1_STACK;
+		}
+
+		mes_hdr = (const struct mes_firmware_header_v1_0 *)
+			adev->mes.fw[pipe]->data;
+
+		fw_data = (const __le32 *)(adev->mes.fw[pipe]->data +
+				le32_to_cpu(mes_hdr->mes_ucode_offset_bytes));
+		fw_size = le32_to_cpu(mes_hdr->mes_ucode_size_bytes);
+
+		gfx_v12_0_rlc_backdoor_autoload_copy_ucode(adev, ucode_id, fw_data, fw_size);
+
+		fw_data = (const __le32 *)(adev->mes.fw[pipe]->data +
+				le32_to_cpu(mes_hdr->mes_ucode_data_offset_bytes));
+		fw_size = le32_to_cpu(mes_hdr->mes_ucode_data_size_bytes);
+
+		gfx_v12_0_rlc_backdoor_autoload_copy_ucode(adev, data_id, fw_data, fw_size);
+	}
+}
+
+static int gfx_v12_0_rlc_backdoor_autoload_enable(struct amdgpu_device *adev)
+{
+	uint32_t rlc_g_offset, rlc_g_size;
+	uint64_t gpu_addr;
+	uint32_t data;
+
+	/* RLC autoload sequence 2: copy ucode */
+	gfx_v12_0_rlc_backdoor_autoload_copy_sdma_ucode(adev);
+	gfx_v12_0_rlc_backdoor_autoload_copy_gfx_ucode(adev);
+	gfx_v12_0_rlc_backdoor_autoload_copy_mes_ucode(adev);
+	gfx_v12_0_rlc_backdoor_autoload_copy_toc_ucode(adev);
+
+	rlc_g_offset = rlc_autoload_info[SOC24_FIRMWARE_ID_RLC_G_UCODE].offset;
+	rlc_g_size = rlc_autoload_info[SOC24_FIRMWARE_ID_RLC_G_UCODE].size;
+	gpu_addr = adev->gfx.rlc.rlc_autoload_gpu_addr + rlc_g_offset - adev->gmc.vram_start;
+
+	WREG32_SOC15(GC, 0, regGFX_IMU_RLC_BOOTLOADER_ADDR_HI, upper_32_bits(gpu_addr));
+	WREG32_SOC15(GC, 0, regGFX_IMU_RLC_BOOTLOADER_ADDR_LO, lower_32_bits(gpu_addr));
+
+	WREG32_SOC15(GC, 0, regGFX_IMU_RLC_BOOTLOADER_SIZE, rlc_g_size);
+
+	if (adev->gfx.imu.funcs && (amdgpu_dpm > 0)) {
+		/* RLC autoload sequence 3: load IMU fw */
+		if (adev->gfx.imu.funcs->load_microcode)
+			adev->gfx.imu.funcs->load_microcode(adev);
+		/* RLC autoload sequence 4 init IMU fw */
+		if (adev->gfx.imu.funcs->setup_imu)
+			adev->gfx.imu.funcs->setup_imu(adev);
+		if (adev->gfx.imu.funcs->start_imu)
+			adev->gfx.imu.funcs->start_imu(adev);
+
+		/* RLC autoload sequence 5 disable gpa mode */
+		gfx_v12_0_disable_gpa_mode(adev);
+	} else {
+		/* unhalt rlc to start autoload without imu */
+		data = RREG32_SOC15(GC, 0, regRLC_GPM_THREAD_ENABLE);
+		data = REG_SET_FIELD(data, RLC_GPM_THREAD_ENABLE, THREAD0_ENABLE, 1);
+		data = REG_SET_FIELD(data, RLC_GPM_THREAD_ENABLE, THREAD1_ENABLE, 1);
+		WREG32_SOC15(GC, 0, regRLC_GPM_THREAD_ENABLE, data);
+		WREG32_SOC15(GC, 0, regRLC_CNTL, RLC_CNTL__RLC_ENABLE_F32_MASK);
+	}
+
+	return 0;
+}
+
+static void gfx_v12_0_alloc_ip_dump(struct amdgpu_device *adev)
+{
+	uint32_t reg_count = ARRAY_SIZE(gc_reg_list_12_0);
+	uint32_t *ptr;
+	uint32_t inst;
+
+	ptr = kcalloc(reg_count, sizeof(uint32_t), GFP_KERNEL);
+	if (!ptr) {
+		DRM_ERROR("Failed to allocate memory for GFX IP Dump\n");
+		adev->gfx.ip_dump_core = NULL;
+	} else {
+		adev->gfx.ip_dump_core = ptr;
+	}
+
+	/* Allocate memory for compute queue registers for all the instances */
+	reg_count = ARRAY_SIZE(gc_cp_reg_list_12);
+	inst = adev->gfx.mec.num_mec * adev->gfx.mec.num_pipe_per_mec *
+		adev->gfx.mec.num_queue_per_pipe;
+
+	ptr = kcalloc(reg_count * inst, sizeof(uint32_t), GFP_KERNEL);
+	if (!ptr) {
+		DRM_ERROR("Failed to allocate memory for Compute Queues IP Dump\n");
+		adev->gfx.ip_dump_compute_queues = NULL;
+	} else {
+		adev->gfx.ip_dump_compute_queues = ptr;
+	}
+
+	/* Allocate memory for gfx queue registers for all the instances */
+	reg_count = ARRAY_SIZE(gc_gfx_queue_reg_list_12);
+	inst = adev->gfx.me.num_me * adev->gfx.me.num_pipe_per_me *
+		adev->gfx.me.num_queue_per_pipe;
+
+	ptr = kcalloc(reg_count * inst, sizeof(uint32_t), GFP_KERNEL);
+	if (!ptr) {
+		DRM_ERROR("Failed to allocate memory for GFX Queues IP Dump\n");
+		adev->gfx.ip_dump_gfx_queues = NULL;
+	} else {
+		adev->gfx.ip_dump_gfx_queues = ptr;
+	}
+}
+
+static int gfx_v12_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	int i, j, k, r, ring_id = 0;
+	unsigned num_compute_rings;
+	int xcc_id = 0;
+	struct amdgpu_device *adev = ip_block->adev;
+	int num_queue_per_pipe = 1; /* we only enable 1 KGQ per pipe */
+
+	INIT_DELAYED_WORK(&adev->gfx.idle_work, amdgpu_gfx_profile_idle_work_handler);
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 0, 1):
+		adev->gfx.me.num_me = 1;
+		adev->gfx.me.num_pipe_per_me = 1;
+		adev->gfx.me.num_queue_per_pipe = 8;
+		adev->gfx.mec.num_mec = 1;
+		adev->gfx.mec.num_pipe_per_mec = 2;
+		adev->gfx.mec.num_queue_per_pipe = 4;
+		break;
+	default:
+		adev->gfx.me.num_me = 1;
+		adev->gfx.me.num_pipe_per_me = 1;
+		adev->gfx.me.num_queue_per_pipe = 1;
+		adev->gfx.mec.num_mec = 1;
+		adev->gfx.mec.num_pipe_per_mec = 4;
+		adev->gfx.mec.num_queue_per_pipe = 8;
+		break;
+	}
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 0, 1):
+		if (!adev->gfx.disable_uq &&
+		    adev->gfx.me_fw_version  >= 2780 &&
+		    adev->gfx.pfp_fw_version >= 2840 &&
+		    adev->gfx.mec_fw_version >= 3050 &&
+		    adev->mes.fw_version[0] >= 123) {
+			adev->userq_funcs[AMDGPU_HW_IP_GFX] = &userq_mes_funcs;
+			adev->userq_funcs[AMDGPU_HW_IP_COMPUTE] = &userq_mes_funcs;
+		}
+		break;
+	default:
+		break;
+	}
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 0, 1):
+		if (adev->gfx.me_fw_version  >= 2480 &&
+		    adev->gfx.pfp_fw_version >= 2530 &&
+		    adev->gfx.mec_fw_version >= 2680 &&
+		    adev->mes.fw_version[0] >= 100)
+			adev->gfx.enable_cleaner_shader = true;
+		break;
+	default:
+		adev->gfx.enable_cleaner_shader = false;
+		break;
+	}
+
+	if (adev->gfx.num_compute_rings) {
+		/* recalculate compute rings to use based on hardware configuration */
+		num_compute_rings = (adev->gfx.mec.num_pipe_per_mec *
+				     adev->gfx.mec.num_queue_per_pipe) / 2;
+		adev->gfx.num_compute_rings = min(adev->gfx.num_compute_rings,
+						  num_compute_rings);
+	}
+
+	/* EOP Event */
+	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GRBM_CP,
+			      GFX_12_0_0__SRCID__CP_EOP_INTERRUPT,
+			      &adev->gfx.eop_irq);
+	if (r)
+		return r;
+
+	/* Bad opcode Event */
+	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GRBM_CP,
+			      GFX_12_0_0__SRCID__CP_BAD_OPCODE_ERROR,
+			      &adev->gfx.bad_op_irq);
+	if (r)
+		return r;
+
+	/* Privileged reg */
+	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GRBM_CP,
+			      GFX_12_0_0__SRCID__CP_PRIV_REG_FAULT,
+			      &adev->gfx.priv_reg_irq);
+	if (r)
+		return r;
+
+	/* Privileged inst */
+	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GRBM_CP,
+			      GFX_12_0_0__SRCID__CP_PRIV_INSTR_FAULT,
+			      &adev->gfx.priv_inst_irq);
+	if (r)
+		return r;
+
+	adev->gfx.gfx_current_status = AMDGPU_GFX_NORMAL_MODE;
+
+	gfx_v12_0_me_init(adev);
+
+	r = gfx_v12_0_rlc_init(adev);
+	if (r) {
+		dev_err(adev->dev, "Failed to init rlc BOs!\n");
+		return r;
+	}
+
+	r = gfx_v12_0_mec_init(adev);
+	if (r) {
+		dev_err(adev->dev, "Failed to init MEC BOs!\n");
+		return r;
+	}
+
+	if (adev->gfx.num_gfx_rings) {
+		/* set up the gfx ring */
+		for (i = 0; i < adev->gfx.me.num_me; i++) {
+			for (j = 0; j < num_queue_per_pipe; j++) {
+				for (k = 0; k < adev->gfx.me.num_pipe_per_me; k++) {
+					if (!amdgpu_gfx_is_me_queue_enabled(adev, i, k, j))
+						continue;
+
+					r = gfx_v12_0_gfx_ring_init(adev, ring_id,
+								    i, k, j);
+					if (r)
+						return r;
+					ring_id++;
+				}
+			}
+		}
+	}
+
+	if (adev->gfx.num_compute_rings) {
+		ring_id = 0;
+		/* set up the compute queues - allocate horizontally across pipes */
+		for (i = 0; i < adev->gfx.mec.num_mec; ++i) {
+			for (j = 0; j < adev->gfx.mec.num_queue_per_pipe; j++) {
+				for (k = 0; k < adev->gfx.mec.num_pipe_per_mec; k++) {
+					if (!amdgpu_gfx_is_mec_queue_enabled(adev,
+									     0, i, k, j))
+						continue;
+
+					r = gfx_v12_0_compute_ring_init(adev, ring_id,
+									i, k, j);
+					if (r)
+						return r;
+
+					ring_id++;
+				}
+			}
+		}
+	}
+
+	adev->gfx.gfx_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.gfx_ring[0]);
+	adev->gfx.compute_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.compute_ring[0]);
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 0, 1):
+		if ((adev->gfx.me_fw_version >= 2660) &&
+		    (adev->gfx.mec_fw_version >= 2920) &&
+		    !amdgpu_sriov_vf(adev)) {
+			adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+			adev->gfx.gfx_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+		}
+		break;
+	default:
+		break;
+	}
+
+	if (!adev->enable_mes_kiq) {
+		r = amdgpu_gfx_kiq_init(adev, GFX12_MEC_HPD_SIZE, 0);
+		if (r) {
+			dev_err(adev->dev, "Failed to init KIQ BOs!\n");
+			return r;
+		}
+
+		r = amdgpu_gfx_kiq_init_ring(adev, xcc_id);
+		if (r)
+			return r;
+	}
+
+	r = amdgpu_gfx_mqd_sw_init(adev, sizeof(struct v12_compute_mqd), 0);
+	if (r)
+		return r;
+
+	/* allocate visible FB for rlc auto-loading fw */
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO) {
+		r = gfx_v12_0_rlc_autoload_buffer_init(adev);
+		if (r)
+			return r;
+	}
+
+	r = gfx_v12_0_gpu_early_init(adev);
+	if (r)
+		return r;
+
+	gfx_v12_0_alloc_ip_dump(adev);
+
+	r = amdgpu_gfx_sysfs_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static void gfx_v12_0_pfp_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->gfx.pfp.pfp_fw_obj,
+			      &adev->gfx.pfp.pfp_fw_gpu_addr,
+			      (void **)&adev->gfx.pfp.pfp_fw_ptr);
+
+	amdgpu_bo_free_kernel(&adev->gfx.pfp.pfp_fw_data_obj,
+			      &adev->gfx.pfp.pfp_fw_data_gpu_addr,
+			      (void **)&adev->gfx.pfp.pfp_fw_data_ptr);
+}
+
+static void gfx_v12_0_me_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->gfx.me.me_fw_obj,
+			      &adev->gfx.me.me_fw_gpu_addr,
+			      (void **)&adev->gfx.me.me_fw_ptr);
+
+	amdgpu_bo_free_kernel(&adev->gfx.me.me_fw_data_obj,
+			       &adev->gfx.me.me_fw_data_gpu_addr,
+			       (void **)&adev->gfx.me.me_fw_data_ptr);
+}
+
+static void gfx_v12_0_rlc_autoload_buffer_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.rlc_autoload_bo,
+			&adev->gfx.rlc.rlc_autoload_gpu_addr,
+			(void **)&adev->gfx.rlc.rlc_autoload_ptr);
+}
+
+static int gfx_v12_0_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	int i;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+		amdgpu_ring_fini(&adev->gfx.gfx_ring[i]);
+	for (i = 0; i < adev->gfx.num_compute_rings; i++)
+		amdgpu_ring_fini(&adev->gfx.compute_ring[i]);
+
+	amdgpu_gfx_mqd_sw_fini(adev, 0);
+
+	if (!adev->enable_mes_kiq) {
+		amdgpu_gfx_kiq_free_ring(&adev->gfx.kiq[0].ring);
+		amdgpu_gfx_kiq_fini(adev, 0);
+	}
+
+	gfx_v12_0_pfp_fini(adev);
+	gfx_v12_0_me_fini(adev);
+	gfx_v12_0_rlc_fini(adev);
+	gfx_v12_0_mec_fini(adev);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO)
+		gfx_v12_0_rlc_autoload_buffer_fini(adev);
+
+	gfx_v12_0_free_microcode(adev);
+
+	amdgpu_gfx_sysfs_fini(adev);
+
+	kfree(adev->gfx.ip_dump_core);
+	kfree(adev->gfx.ip_dump_compute_queues);
+	kfree(adev->gfx.ip_dump_gfx_queues);
+
+	return 0;
+}
+
+static void gfx_v12_0_select_se_sh(struct amdgpu_device *adev, u32 se_num,
+				   u32 sh_num, u32 instance, int xcc_id)
+{
+	u32 data;
+
+	if (instance == 0xffffffff)
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX,
+				     INSTANCE_BROADCAST_WRITES, 1);
+	else
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_INDEX,
+				     instance);
+
+	if (se_num == 0xffffffff)
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_BROADCAST_WRITES,
+				     1);
+	else
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_INDEX, se_num);
+
+	if (sh_num == 0xffffffff)
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SA_BROADCAST_WRITES,
+				     1);
+	else
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SA_INDEX, sh_num);
+
+	WREG32_SOC15(GC, 0, regGRBM_GFX_INDEX, data);
+}
+
+static u32 gfx_v12_0_get_sa_active_bitmap(struct amdgpu_device *adev)
+{
+	u32 gc_disabled_sa_mask, gc_user_disabled_sa_mask, sa_mask;
+
+	gc_disabled_sa_mask = RREG32_SOC15(GC, 0, regGRBM_CC_GC_SA_UNIT_DISABLE);
+	gc_disabled_sa_mask = REG_GET_FIELD(gc_disabled_sa_mask,
+					    GRBM_CC_GC_SA_UNIT_DISABLE,
+					    SA_DISABLE);
+	gc_user_disabled_sa_mask = RREG32_SOC15(GC, 0, regGRBM_GC_USER_SA_UNIT_DISABLE);
+	gc_user_disabled_sa_mask = REG_GET_FIELD(gc_user_disabled_sa_mask,
+						 GRBM_GC_USER_SA_UNIT_DISABLE,
+						 SA_DISABLE);
+	sa_mask = amdgpu_gfx_create_bitmask(adev->gfx.config.max_sh_per_se *
+					    adev->gfx.config.max_shader_engines);
+
+	return sa_mask & (~(gc_disabled_sa_mask | gc_user_disabled_sa_mask));
+}
+
+static u32 gfx_v12_0_get_rb_active_bitmap(struct amdgpu_device *adev)
+{
+	u32 gc_disabled_rb_mask, gc_user_disabled_rb_mask;
+	u32 rb_mask;
+
+	gc_disabled_rb_mask = RREG32_SOC15(GC, 0, regCC_RB_BACKEND_DISABLE);
+	gc_disabled_rb_mask = REG_GET_FIELD(gc_disabled_rb_mask,
+					    CC_RB_BACKEND_DISABLE,
+					    BACKEND_DISABLE);
+	gc_user_disabled_rb_mask = RREG32_SOC15(GC, 0, regGC_USER_RB_BACKEND_DISABLE);
+	gc_user_disabled_rb_mask = REG_GET_FIELD(gc_user_disabled_rb_mask,
+						 GC_USER_RB_BACKEND_DISABLE,
+						 BACKEND_DISABLE);
+	rb_mask = amdgpu_gfx_create_bitmask(adev->gfx.config.max_backends_per_se *
+					    adev->gfx.config.max_shader_engines);
+
+	return rb_mask & (~(gc_disabled_rb_mask | gc_user_disabled_rb_mask));
+}
+
+static void gfx_v12_0_setup_rb(struct amdgpu_device *adev)
+{
+	u32 rb_bitmap_per_sa;
+	u32 rb_bitmap_width_per_sa;
+	u32 max_sa;
+	u32 active_sa_bitmap;
+	u32 global_active_rb_bitmap;
+	u32 active_rb_bitmap = 0;
+	u32 i;
+
+	/* query sa bitmap from SA_UNIT_DISABLE registers */
+	active_sa_bitmap = gfx_v12_0_get_sa_active_bitmap(adev);
+	/* query rb bitmap from RB_BACKEND_DISABLE registers */
+	global_active_rb_bitmap = gfx_v12_0_get_rb_active_bitmap(adev);
+
+	/* generate active rb bitmap according to active sa bitmap */
+	max_sa = adev->gfx.config.max_shader_engines *
+		 adev->gfx.config.max_sh_per_se;
+	rb_bitmap_width_per_sa = adev->gfx.config.max_backends_per_se /
+				 adev->gfx.config.max_sh_per_se;
+	rb_bitmap_per_sa = amdgpu_gfx_create_bitmask(rb_bitmap_width_per_sa);
+
+	for (i = 0; i < max_sa; i++) {
+		if (active_sa_bitmap & (1 << i))
+			active_rb_bitmap |= (rb_bitmap_per_sa << (i * rb_bitmap_width_per_sa));
+	}
+
+	active_rb_bitmap &= global_active_rb_bitmap;
+	adev->gfx.config.backend_enable_mask = active_rb_bitmap;
+	adev->gfx.config.num_rbs = hweight32(active_rb_bitmap);
+}
+
+#define LDS_APP_BASE           0x1
+#define SCRATCH_APP_BASE       0x2
+
+static void gfx_v12_0_init_compute_vmid(struct amdgpu_device *adev)
+{
+	int i;
+	uint32_t sh_mem_bases;
+	uint32_t data;
+
+	/*
+	 * Configure apertures:
+	 * LDS:         0x60000000'00000000 - 0x60000001'00000000 (4GB)
+	 * Scratch:     0x60000001'00000000 - 0x60000002'00000000 (4GB)
+	 * GPUVM:       0x60010000'00000000 - 0x60020000'00000000 (1TB)
+	 */
+	sh_mem_bases = (LDS_APP_BASE << SH_MEM_BASES__SHARED_BASE__SHIFT) |
+			SCRATCH_APP_BASE;
+
+	mutex_lock(&adev->srbm_mutex);
+	for (i = adev->vm_manager.first_kfd_vmid; i < AMDGPU_NUM_VMID; i++) {
+		soc24_grbm_select(adev, 0, 0, 0, i);
+		/* CP and shaders */
+		WREG32_SOC15(GC, 0, regSH_MEM_CONFIG, DEFAULT_SH_MEM_CONFIG);
+		WREG32_SOC15(GC, 0, regSH_MEM_BASES, sh_mem_bases);
+
+		/* Enable trap for each kfd vmid. */
+		data = RREG32_SOC15(GC, 0, regSPI_GDBG_PER_VMID_CNTL);
+		data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, TRAP_EN, 1);
+		WREG32_SOC15(GC, 0, regSPI_GDBG_PER_VMID_CNTL, data);
+	}
+	soc24_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static void gfx_v12_0_tcp_harvest(struct amdgpu_device *adev)
+{
+	/* TODO: harvest feature to be added later. */
+}
+
+static void gfx_v12_0_get_tcc_info(struct amdgpu_device *adev)
+{
+}
+
+static void gfx_v12_0_constants_init(struct amdgpu_device *adev)
+{
+	u32 tmp;
+	int i;
+
+	if (!amdgpu_sriov_vf(adev))
+		WREG32_FIELD15_PREREG(GC, 0, GRBM_CNTL, READ_TIMEOUT, 0xff);
+
+	gfx_v12_0_setup_rb(adev);
+	gfx_v12_0_get_cu_info(adev, &adev->gfx.cu_info);
+	gfx_v12_0_get_tcc_info(adev);
+	adev->gfx.config.pa_sc_tile_steering_override = 0;
+
+	/* XXX SH_MEM regs */
+	/* where to put LDS, scratch, GPUVM in FSA64 space */
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < adev->vm_manager.id_mgr[AMDGPU_GFXHUB(0)].num_ids; i++) {
+		soc24_grbm_select(adev, 0, 0, 0, i);
+		/* CP and shaders */
+		WREG32_SOC15(GC, 0, regSH_MEM_CONFIG, DEFAULT_SH_MEM_CONFIG);
+		if (i != 0) {
+			tmp = REG_SET_FIELD(0, SH_MEM_BASES, PRIVATE_BASE,
+				(adev->gmc.private_aperture_start >> 48));
+			tmp = REG_SET_FIELD(tmp, SH_MEM_BASES, SHARED_BASE,
+				(adev->gmc.shared_aperture_start >> 48));
+			WREG32_SOC15(GC, 0, regSH_MEM_BASES, tmp);
+		}
+	}
+	soc24_grbm_select(adev, 0, 0, 0, 0);
+
+	mutex_unlock(&adev->srbm_mutex);
+
+	gfx_v12_0_init_compute_vmid(adev);
+}
+
+static u32 gfx_v12_0_get_cpg_int_cntl(struct amdgpu_device *adev,
+				      int me, int pipe)
+{
+	if (me != 0)
+		return 0;
+
+	switch (pipe) {
+	case 0:
+		return SOC15_REG_OFFSET(GC, 0, regCP_INT_CNTL_RING0);
+	default:
+		return 0;
+	}
+}
+
+static u32 gfx_v12_0_get_cpc_int_cntl(struct amdgpu_device *adev,
+				      int me, int pipe)
+{
+	/*
+	 * amdgpu controls only the first MEC. That's why this function only
+	 * handles the setting of interrupts for this specific MEC. All other
+	 * pipes' interrupts are set by amdkfd.
+	 */
+	if (me != 1)
+		return 0;
+
+	switch (pipe) {
+	case 0:
+		return SOC15_REG_OFFSET(GC, 0, regCP_ME1_PIPE0_INT_CNTL);
+	case 1:
+		return SOC15_REG_OFFSET(GC, 0, regCP_ME1_PIPE1_INT_CNTL);
+	default:
+		return 0;
+	}
+}
+
+static void gfx_v12_0_enable_gui_idle_interrupt(struct amdgpu_device *adev,
+					       bool enable)
+{
+	u32 tmp, cp_int_cntl_reg;
+	int i, j;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	for (i = 0; i < adev->gfx.me.num_me; i++) {
+		for (j = 0; j < adev->gfx.me.num_pipe_per_me; j++) {
+			cp_int_cntl_reg = gfx_v12_0_get_cpg_int_cntl(adev, i, j);
+
+			if (cp_int_cntl_reg) {
+				tmp = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+				tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_BUSY_INT_ENABLE,
+						    enable ? 1 : 0);
+				tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_EMPTY_INT_ENABLE,
+						    enable ? 1 : 0);
+				tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CMP_BUSY_INT_ENABLE,
+						    enable ? 1 : 0);
+				tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, GFX_IDLE_INT_ENABLE,
+						    enable ? 1 : 0);
+				WREG32_SOC15_IP(GC, cp_int_cntl_reg, tmp);
+			}
+		}
+	}
+}
+
+static int gfx_v12_0_init_csb(struct amdgpu_device *adev)
+{
+	adev->gfx.rlc.funcs->get_csb_buffer(adev, adev->gfx.rlc.cs_ptr);
+
+	WREG32_SOC15(GC, 0, regRLC_CSIB_ADDR_HI,
+			adev->gfx.rlc.clear_state_gpu_addr >> 32);
+	WREG32_SOC15(GC, 0, regRLC_CSIB_ADDR_LO,
+			adev->gfx.rlc.clear_state_gpu_addr & 0xfffffffc);
+	WREG32_SOC15(GC, 0, regRLC_CSIB_LENGTH, adev->gfx.rlc.clear_state_size);
+
+	return 0;
+}
+
+static void gfx_v12_0_rlc_stop(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SOC15(GC, 0, regRLC_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, RLC_CNTL, RLC_ENABLE_F32, 0);
+	WREG32_SOC15(GC, 0, regRLC_CNTL, tmp);
+}
+
+static void gfx_v12_0_rlc_reset(struct amdgpu_device *adev)
+{
+	WREG32_FIELD15_PREREG(GC, 0, GRBM_SOFT_RESET, SOFT_RESET_RLC, 1);
+	udelay(50);
+	WREG32_FIELD15_PREREG(GC, 0, GRBM_SOFT_RESET, SOFT_RESET_RLC, 0);
+	udelay(50);
+}
+
+static void gfx_v12_0_rlc_smu_handshake_cntl(struct amdgpu_device *adev,
+					     bool enable)
+{
+	uint32_t rlc_pg_cntl;
+
+	rlc_pg_cntl = RREG32_SOC15(GC, 0, regRLC_PG_CNTL);
+
+	if (!enable) {
+		/* RLC_PG_CNTL[23] = 0 (default)
+		 * RLC will wait for handshake acks with SMU
+		 * GFXOFF will be enabled
+		 * RLC_PG_CNTL[23] = 1
+		 * RLC will not issue any message to SMU
+		 * hence no handshake between SMU & RLC
+		 * GFXOFF will be disabled
+		 */
+		rlc_pg_cntl |= RLC_PG_CNTL__SMU_HANDSHAKE_DISABLE_MASK;
+	} else
+		rlc_pg_cntl &= ~RLC_PG_CNTL__SMU_HANDSHAKE_DISABLE_MASK;
+	WREG32_SOC15(GC, 0, regRLC_PG_CNTL, rlc_pg_cntl);
+}
+
+static void gfx_v12_0_rlc_start(struct amdgpu_device *adev)
+{
+	/* TODO: enable rlc & smu handshake until smu
+	 * and gfxoff feature works as expected */
+	if (!(amdgpu_pp_feature_mask & PP_GFXOFF_MASK))
+		gfx_v12_0_rlc_smu_handshake_cntl(adev, false);
+
+	WREG32_FIELD15_PREREG(GC, 0, RLC_CNTL, RLC_ENABLE_F32, 1);
+	udelay(50);
+}
+
+static void gfx_v12_0_rlc_enable_srm(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* enable Save Restore Machine */
+	tmp = RREG32(SOC15_REG_OFFSET(GC, 0, regRLC_SRM_CNTL));
+	tmp |= RLC_SRM_CNTL__AUTO_INCR_ADDR_MASK;
+	tmp |= RLC_SRM_CNTL__SRM_ENABLE_MASK;
+	WREG32(SOC15_REG_OFFSET(GC, 0, regRLC_SRM_CNTL), tmp);
+}
+
+static void gfx_v12_0_load_rlcg_microcode(struct amdgpu_device *adev)
+{
+	const struct rlc_firmware_header_v2_0 *hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+
+	hdr = (const struct rlc_firmware_header_v2_0 *)adev->gfx.rlc_fw->data;
+	fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+			   le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+
+	WREG32_SOC15(GC, 0, regRLC_GPM_UCODE_ADDR,
+		     RLCG_UCODE_LOADING_START_ADDRESS);
+
+	for (i = 0; i < fw_size; i++)
+		WREG32_SOC15(GC, 0, regRLC_GPM_UCODE_DATA,
+			     le32_to_cpup(fw_data++));
+
+	WREG32_SOC15(GC, 0, regRLC_GPM_UCODE_ADDR, adev->gfx.rlc_fw_version);
+}
+
+static void gfx_v12_0_load_rlc_iram_dram_microcode(struct amdgpu_device *adev)
+{
+	const struct rlc_firmware_header_v2_2 *hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+	u32 tmp;
+
+	hdr = (const struct rlc_firmware_header_v2_2 *)adev->gfx.rlc_fw->data;
+
+	fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+			le32_to_cpu(hdr->rlc_iram_ucode_offset_bytes));
+	fw_size = le32_to_cpu(hdr->rlc_iram_ucode_size_bytes) / 4;
+
+	WREG32_SOC15(GC, 0, regRLC_LX6_IRAM_ADDR, 0);
+
+	for (i = 0; i < fw_size; i++) {
+		if ((amdgpu_emu_mode == 1) && (i % 100 == 99))
+			msleep(1);
+		WREG32_SOC15(GC, 0, regRLC_LX6_IRAM_DATA,
+				le32_to_cpup(fw_data++));
+	}
+
+	WREG32_SOC15(GC, 0, regRLC_LX6_IRAM_ADDR, adev->gfx.rlc_fw_version);
+
+	fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+			le32_to_cpu(hdr->rlc_dram_ucode_offset_bytes));
+	fw_size = le32_to_cpu(hdr->rlc_dram_ucode_size_bytes) / 4;
+
+	WREG32_SOC15(GC, 0, regRLC_LX6_DRAM_ADDR, 0);
+	for (i = 0; i < fw_size; i++) {
+		if ((amdgpu_emu_mode == 1) && (i % 100 == 99))
+			msleep(1);
+		WREG32_SOC15(GC, 0, regRLC_LX6_DRAM_DATA,
+				le32_to_cpup(fw_data++));
+	}
+
+	WREG32_SOC15(GC, 0, regRLC_LX6_IRAM_ADDR, adev->gfx.rlc_fw_version);
+
+	tmp = RREG32_SOC15(GC, 0, regRLC_LX6_CNTL);
+	tmp = REG_SET_FIELD(tmp, RLC_LX6_CNTL, PDEBUG_ENABLE, 1);
+	tmp = REG_SET_FIELD(tmp, RLC_LX6_CNTL, BRESET, 0);
+	WREG32_SOC15(GC, 0, regRLC_LX6_CNTL, tmp);
+}
+
+static int gfx_v12_0_rlc_load_microcode(struct amdgpu_device *adev)
+{
+	const struct rlc_firmware_header_v2_0 *hdr;
+	uint16_t version_major;
+	uint16_t version_minor;
+
+	if (!adev->gfx.rlc_fw)
+		return -EINVAL;
+
+	hdr = (const struct rlc_firmware_header_v2_0 *)adev->gfx.rlc_fw->data;
+	amdgpu_ucode_print_rlc_hdr(&hdr->header);
+
+	version_major = le16_to_cpu(hdr->header.header_version_major);
+	version_minor = le16_to_cpu(hdr->header.header_version_minor);
+
+	if (version_major == 2) {
+		gfx_v12_0_load_rlcg_microcode(adev);
+		if (amdgpu_dpm == 1) {
+			if (version_minor >= 2)
+				gfx_v12_0_load_rlc_iram_dram_microcode(adev);
+		}
+
+		return 0;
+	}
+
+	return -EINVAL;
+}
+
+static int gfx_v12_0_rlc_resume(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		gfx_v12_0_init_csb(adev);
+
+		if (!amdgpu_sriov_vf(adev)) /* enable RLC SRM */
+			gfx_v12_0_rlc_enable_srm(adev);
+	} else {
+		if (amdgpu_sriov_vf(adev)) {
+			gfx_v12_0_init_csb(adev);
+			return 0;
+		}
+
+		adev->gfx.rlc.funcs->stop(adev);
+
+		/* disable CG */
+		WREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL, 0);
+
+		/* disable PG */
+		WREG32_SOC15(GC, 0, regRLC_PG_CNTL, 0);
+
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+			/* legacy rlc firmware loading */
+			r = gfx_v12_0_rlc_load_microcode(adev);
+			if (r)
+				return r;
+		}
+
+		gfx_v12_0_init_csb(adev);
+
+		adev->gfx.rlc.funcs->start(adev);
+	}
+
+	return 0;
+}
+
+static void gfx_v12_0_config_gfx_rs64(struct amdgpu_device *adev)
+{
+	const struct gfx_firmware_header_v2_0 *pfp_hdr;
+	const struct gfx_firmware_header_v2_0 *me_hdr;
+	const struct gfx_firmware_header_v2_0 *mec_hdr;
+	uint32_t pipe_id, tmp;
+
+	mec_hdr = (const struct gfx_firmware_header_v2_0 *)
+		adev->gfx.mec_fw->data;
+	me_hdr = (const struct gfx_firmware_header_v2_0 *)
+		adev->gfx.me_fw->data;
+	pfp_hdr = (const struct gfx_firmware_header_v2_0 *)
+		adev->gfx.pfp_fw->data;
+
+	/* config pfp program start addr */
+	for (pipe_id = 0; pipe_id < 2; pipe_id++) {
+		soc24_grbm_select(adev, 0, pipe_id, 0, 0);
+		WREG32_SOC15(GC, 0, regCP_PFP_PRGRM_CNTR_START,
+			(pfp_hdr->ucode_start_addr_hi << 30) |
+			(pfp_hdr->ucode_start_addr_lo >> 2));
+		WREG32_SOC15(GC, 0, regCP_PFP_PRGRM_CNTR_START_HI,
+			pfp_hdr->ucode_start_addr_hi >> 2);
+	}
+	soc24_grbm_select(adev, 0, 0, 0, 0);
+
+	/* reset pfp pipe */
+	tmp = RREG32_SOC15(GC, 0, regCP_ME_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_PIPE0_RESET, 1);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_PIPE1_RESET, 1);
+	WREG32_SOC15(GC, 0, regCP_ME_CNTL, tmp);
+
+	/* clear pfp pipe reset */
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_PIPE0_RESET, 0);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_PIPE1_RESET, 0);
+	WREG32_SOC15(GC, 0, regCP_ME_CNTL, tmp);
+
+	/* config me program start addr */
+	for (pipe_id = 0; pipe_id < 2; pipe_id++) {
+		soc24_grbm_select(adev, 0, pipe_id, 0, 0);
+		WREG32_SOC15(GC, 0, regCP_ME_PRGRM_CNTR_START,
+			(me_hdr->ucode_start_addr_hi << 30) |
+			(me_hdr->ucode_start_addr_lo >> 2));
+		WREG32_SOC15(GC, 0, regCP_ME_PRGRM_CNTR_START_HI,
+			me_hdr->ucode_start_addr_hi>>2);
+	}
+	soc24_grbm_select(adev, 0, 0, 0, 0);
+
+	/* reset me pipe */
+	tmp = RREG32_SOC15(GC, 0, regCP_ME_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_PIPE0_RESET, 1);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_PIPE1_RESET, 1);
+	WREG32_SOC15(GC, 0, regCP_ME_CNTL, tmp);
+
+	/* clear me pipe reset */
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_PIPE0_RESET, 0);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_PIPE1_RESET, 0);
+	WREG32_SOC15(GC, 0, regCP_ME_CNTL, tmp);
+
+	/* config mec program start addr */
+	for (pipe_id = 0; pipe_id < 4; pipe_id++) {
+		soc24_grbm_select(adev, 1, pipe_id, 0, 0);
+		WREG32_SOC15(GC, 0, regCP_MEC_RS64_PRGRM_CNTR_START,
+					mec_hdr->ucode_start_addr_lo >> 2 |
+					mec_hdr->ucode_start_addr_hi << 30);
+		WREG32_SOC15(GC, 0, regCP_MEC_RS64_PRGRM_CNTR_START_HI,
+					mec_hdr->ucode_start_addr_hi >> 2);
+	}
+	soc24_grbm_select(adev, 0, 0, 0, 0);
+
+	/* reset mec pipe */
+	tmp = RREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_RS64_CNTL, MEC_PIPE0_RESET, 1);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_RS64_CNTL, MEC_PIPE1_RESET, 1);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_RS64_CNTL, MEC_PIPE2_RESET, 1);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_RS64_CNTL, MEC_PIPE3_RESET, 1);
+	WREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL, tmp);
+
+	/* clear mec pipe reset */
+	tmp = REG_SET_FIELD(tmp, CP_MEC_RS64_CNTL, MEC_PIPE0_RESET, 0);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_RS64_CNTL, MEC_PIPE1_RESET, 0);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_RS64_CNTL, MEC_PIPE2_RESET, 0);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_RS64_CNTL, MEC_PIPE3_RESET, 0);
+	WREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL, tmp);
+}
+
+static void gfx_v12_0_set_pfp_ucode_start_addr(struct amdgpu_device *adev)
+{
+	const struct gfx_firmware_header_v2_0 *cp_hdr;
+	unsigned pipe_id, tmp;
+
+	cp_hdr = (const struct gfx_firmware_header_v2_0 *)
+		adev->gfx.pfp_fw->data;
+	mutex_lock(&adev->srbm_mutex);
+	for (pipe_id = 0; pipe_id < adev->gfx.me.num_pipe_per_me; pipe_id++) {
+		soc24_grbm_select(adev, 0, pipe_id, 0, 0);
+		WREG32_SOC15(GC, 0, regCP_PFP_PRGRM_CNTR_START,
+			     (cp_hdr->ucode_start_addr_hi << 30) |
+			     (cp_hdr->ucode_start_addr_lo >> 2));
+		WREG32_SOC15(GC, 0, regCP_PFP_PRGRM_CNTR_START_HI,
+			     cp_hdr->ucode_start_addr_hi>>2);
+
+		/*
+		 * Program CP_ME_CNTL to reset given PIPE to take
+		 * effect of CP_PFP_PRGRM_CNTR_START.
+		 */
+		tmp = RREG32_SOC15(GC, 0, regCP_ME_CNTL);
+		if (pipe_id == 0)
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					PFP_PIPE0_RESET, 1);
+		else
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					PFP_PIPE1_RESET, 1);
+		WREG32_SOC15(GC, 0, regCP_ME_CNTL, tmp);
+
+		/* Clear pfp pipe0 reset bit. */
+		if (pipe_id == 0)
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					PFP_PIPE0_RESET, 0);
+		else
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					PFP_PIPE1_RESET, 0);
+		WREG32_SOC15(GC, 0, regCP_ME_CNTL, tmp);
+	}
+	soc24_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static void gfx_v12_0_set_me_ucode_start_addr(struct amdgpu_device *adev)
+{
+	const struct gfx_firmware_header_v2_0 *cp_hdr;
+	unsigned pipe_id, tmp;
+
+	cp_hdr = (const struct gfx_firmware_header_v2_0 *)
+		adev->gfx.me_fw->data;
+	mutex_lock(&adev->srbm_mutex);
+	for (pipe_id = 0; pipe_id < adev->gfx.me.num_pipe_per_me; pipe_id++) {
+		soc24_grbm_select(adev, 0, pipe_id, 0, 0);
+		WREG32_SOC15(GC, 0, regCP_ME_PRGRM_CNTR_START,
+			     (cp_hdr->ucode_start_addr_hi << 30) |
+			     (cp_hdr->ucode_start_addr_lo >> 2) );
+		WREG32_SOC15(GC, 0, regCP_ME_PRGRM_CNTR_START_HI,
+			     cp_hdr->ucode_start_addr_hi>>2);
+
+		/*
+		 * Program CP_ME_CNTL to reset given PIPE to take
+		 * effect of CP_ME_PRGRM_CNTR_START.
+		 */
+		tmp = RREG32_SOC15(GC, 0, regCP_ME_CNTL);
+		if (pipe_id == 0)
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					ME_PIPE0_RESET, 1);
+		else
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					ME_PIPE1_RESET, 1);
+		WREG32_SOC15(GC, 0, regCP_ME_CNTL, tmp);
+
+		/* Clear pfp pipe0 reset bit. */
+		if (pipe_id == 0)
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					ME_PIPE0_RESET, 0);
+		else
+			tmp = REG_SET_FIELD(tmp, CP_ME_CNTL,
+					ME_PIPE1_RESET, 0);
+		WREG32_SOC15(GC, 0, regCP_ME_CNTL, tmp);
+	}
+	soc24_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static void gfx_v12_0_set_mec_ucode_start_addr(struct amdgpu_device *adev)
+{
+	const struct gfx_firmware_header_v2_0 *cp_hdr;
+	unsigned pipe_id;
+
+	cp_hdr = (const struct gfx_firmware_header_v2_0 *)
+		adev->gfx.mec_fw->data;
+	mutex_lock(&adev->srbm_mutex);
+	for (pipe_id = 0; pipe_id < adev->gfx.mec.num_pipe_per_mec; pipe_id++) {
+		soc24_grbm_select(adev, 1, pipe_id, 0, 0);
+		WREG32_SOC15(GC, 0, regCP_MEC_RS64_PRGRM_CNTR_START,
+			     cp_hdr->ucode_start_addr_lo >> 2 |
+			     cp_hdr->ucode_start_addr_hi << 30);
+		WREG32_SOC15(GC, 0, regCP_MEC_RS64_PRGRM_CNTR_START_HI,
+			     cp_hdr->ucode_start_addr_hi >> 2);
+	}
+	soc24_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static int gfx_v12_0_wait_for_rlc_autoload_complete(struct amdgpu_device *adev)
+{
+	uint32_t cp_status;
+	uint32_t bootload_status;
+	int i;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		cp_status = RREG32_SOC15(GC, 0, regCP_STAT);
+		bootload_status = RREG32_SOC15(GC, 0, regRLC_RLCS_BOOTLOAD_STATUS);
+
+		if ((cp_status == 0) &&
+		    (REG_GET_FIELD(bootload_status,
+			RLC_RLCS_BOOTLOAD_STATUS, BOOTLOAD_COMPLETE) == 1)) {
+			break;
+		}
+		udelay(1);
+		if (amdgpu_emu_mode)
+			msleep(10);
+	}
+
+	if (i >= adev->usec_timeout) {
+		dev_err(adev->dev, "rlc autoload: gc ucode autoload timeout\n");
+		return -ETIMEDOUT;
+	}
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO) {
+		gfx_v12_0_set_pfp_ucode_start_addr(adev);
+		gfx_v12_0_set_me_ucode_start_addr(adev);
+		gfx_v12_0_set_mec_ucode_start_addr(adev);
+	}
+
+	return 0;
+}
+
+static int gfx_v12_0_cp_gfx_enable(struct amdgpu_device *adev, bool enable)
+{
+	int i;
+	u32 tmp = RREG32_SOC15(GC, 0, regCP_ME_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_HALT, enable ? 0 : 1);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_HALT, enable ? 0 : 1);
+	WREG32_SOC15(GC, 0, regCP_ME_CNTL, tmp);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (RREG32_SOC15(GC, 0, regCP_STAT) == 0)
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout)
+		DRM_ERROR("failed to %s cp gfx\n", enable ? "unhalt" : "halt");
+
+	return 0;
+}
+
+static int gfx_v12_0_cp_gfx_load_pfp_microcode_rs64(struct amdgpu_device *adev)
+{
+	int r;
+	const struct gfx_firmware_header_v2_0 *pfp_hdr;
+	const __le32 *fw_ucode, *fw_data;
+	unsigned i, pipe_id, fw_ucode_size, fw_data_size;
+	uint32_t tmp;
+	uint32_t usec_timeout = 50000;  /* wait for 50ms */
+
+	pfp_hdr = (const struct gfx_firmware_header_v2_0 *)
+		adev->gfx.pfp_fw->data;
+
+	amdgpu_ucode_print_gfx_hdr(&pfp_hdr->header);
+
+	/* instruction */
+	fw_ucode = (const __le32 *)(adev->gfx.pfp_fw->data +
+		le32_to_cpu(pfp_hdr->ucode_offset_bytes));
+	fw_ucode_size = le32_to_cpu(pfp_hdr->ucode_size_bytes);
+	/* data */
+	fw_data = (const __le32 *)(adev->gfx.pfp_fw->data +
+		le32_to_cpu(pfp_hdr->data_offset_bytes));
+	fw_data_size = le32_to_cpu(pfp_hdr->data_size_bytes);
+
+	/* 64kb align */
+	r = amdgpu_bo_create_reserved(adev, fw_ucode_size,
+				      64 * 1024, AMDGPU_GEM_DOMAIN_VRAM,
+				      &adev->gfx.pfp.pfp_fw_obj,
+				      &adev->gfx.pfp.pfp_fw_gpu_addr,
+				      (void **)&adev->gfx.pfp.pfp_fw_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create pfp ucode fw bo\n", r);
+		gfx_v12_0_pfp_fini(adev);
+		return r;
+	}
+
+	r = amdgpu_bo_create_reserved(adev, fw_data_size,
+				      64 * 1024, AMDGPU_GEM_DOMAIN_VRAM,
+				      &adev->gfx.pfp.pfp_fw_data_obj,
+				      &adev->gfx.pfp.pfp_fw_data_gpu_addr,
+				      (void **)&adev->gfx.pfp.pfp_fw_data_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create pfp data fw bo\n", r);
+		gfx_v12_0_pfp_fini(adev);
+		return r;
+	}
+
+	memcpy(adev->gfx.pfp.pfp_fw_ptr, fw_ucode, fw_ucode_size);
+	memcpy(adev->gfx.pfp.pfp_fw_data_ptr, fw_data, fw_data_size);
+
+	amdgpu_bo_kunmap(adev->gfx.pfp.pfp_fw_obj);
+	amdgpu_bo_kunmap(adev->gfx.pfp.pfp_fw_data_obj);
+	amdgpu_bo_unreserve(adev->gfx.pfp.pfp_fw_obj);
+	amdgpu_bo_unreserve(adev->gfx.pfp.pfp_fw_data_obj);
+
+	if (amdgpu_emu_mode == 1)
+		amdgpu_device_flush_hdp(adev, NULL);
+
+	WREG32_SOC15(GC, 0, regCP_PFP_IC_BASE_LO,
+		lower_32_bits(adev->gfx.pfp.pfp_fw_gpu_addr));
+	WREG32_SOC15(GC, 0, regCP_PFP_IC_BASE_HI,
+		upper_32_bits(adev->gfx.pfp.pfp_fw_gpu_addr));
+
+	tmp = RREG32_SOC15(GC, 0, regCP_PFP_IC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_BASE_CNTL, CACHE_POLICY, 0);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_BASE_CNTL, EXE_DISABLE, 0);
+	WREG32_SOC15(GC, 0, regCP_PFP_IC_BASE_CNTL, tmp);
+
+	/*
+	 * Programming any of the CP_PFP_IC_BASE registers
+	 * forces invalidation of the ME L1 I$. Wait for the
+	 * invalidation complete
+	 */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_PFP_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_PFP_IC_OP_CNTL,
+			INVALIDATE_CACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	/* Prime the L1 instruction caches */
+	tmp = RREG32_SOC15(GC, 0, regCP_PFP_IC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_OP_CNTL, PRIME_ICACHE, 1);
+	WREG32_SOC15(GC, 0, regCP_PFP_IC_OP_CNTL, tmp);
+	/* Waiting for cache primed*/
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_PFP_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_PFP_IC_OP_CNTL,
+			ICACHE_PRIMED))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to prime instruction cache\n");
+		return -EINVAL;
+	}
+
+	mutex_lock(&adev->srbm_mutex);
+	for (pipe_id = 0; pipe_id < adev->gfx.me.num_pipe_per_me; pipe_id++) {
+		soc24_grbm_select(adev, 0, pipe_id, 0, 0);
+
+		WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE0_LO,
+			lower_32_bits(adev->gfx.pfp.pfp_fw_data_gpu_addr));
+		WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE0_HI,
+			upper_32_bits(adev->gfx.pfp.pfp_fw_data_gpu_addr));
+	}
+	soc24_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	tmp = RREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_RS64_DC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_RS64_DC_BASE_CNTL, CACHE_POLICY, 0);
+	WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE_CNTL, tmp);
+
+	/* Invalidate the data caches */
+	tmp = RREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_RS64_DC_OP_CNTL, INVALIDATE_DCACHE, 1);
+	WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_OP_CNTL, tmp);
+
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_GFX_RS64_DC_OP_CNTL,
+			INVALIDATE_DCACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate RS64 data cache\n");
+		return -EINVAL;
+	}
+
+	gfx_v12_0_set_pfp_ucode_start_addr(adev);
+
+	return 0;
+}
+
+static int gfx_v12_0_cp_gfx_load_me_microcode_rs64(struct amdgpu_device *adev)
+{
+	int r;
+	const struct gfx_firmware_header_v2_0 *me_hdr;
+	const __le32 *fw_ucode, *fw_data;
+	unsigned i, pipe_id, fw_ucode_size, fw_data_size;
+	uint32_t tmp;
+	uint32_t usec_timeout = 50000;  /* wait for 50ms */
+
+	me_hdr = (const struct gfx_firmware_header_v2_0 *)
+		adev->gfx.me_fw->data;
+
+	amdgpu_ucode_print_gfx_hdr(&me_hdr->header);
+
+	/* instruction */
+	fw_ucode = (const __le32 *)(adev->gfx.me_fw->data +
+		le32_to_cpu(me_hdr->ucode_offset_bytes));
+	fw_ucode_size = le32_to_cpu(me_hdr->ucode_size_bytes);
+	/* data */
+	fw_data = (const __le32 *)(adev->gfx.me_fw->data +
+		le32_to_cpu(me_hdr->data_offset_bytes));
+	fw_data_size = le32_to_cpu(me_hdr->data_size_bytes);
+
+	/* 64kb align*/
+	r = amdgpu_bo_create_reserved(adev, fw_ucode_size,
+				      64 * 1024, AMDGPU_GEM_DOMAIN_VRAM,
+				      &adev->gfx.me.me_fw_obj,
+				      &adev->gfx.me.me_fw_gpu_addr,
+				      (void **)&adev->gfx.me.me_fw_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create me ucode bo\n", r);
+		gfx_v12_0_me_fini(adev);
+		return r;
+	}
+
+	r = amdgpu_bo_create_reserved(adev, fw_data_size,
+				      64 * 1024, AMDGPU_GEM_DOMAIN_VRAM,
+				      &adev->gfx.me.me_fw_data_obj,
+				      &adev->gfx.me.me_fw_data_gpu_addr,
+				      (void **)&adev->gfx.me.me_fw_data_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create me data bo\n", r);
+		gfx_v12_0_me_fini(adev);
+		return r;
+	}
+
+	memcpy(adev->gfx.me.me_fw_ptr, fw_ucode, fw_ucode_size);
+	memcpy(adev->gfx.me.me_fw_data_ptr, fw_data, fw_data_size);
+
+	amdgpu_bo_kunmap(adev->gfx.me.me_fw_obj);
+	amdgpu_bo_kunmap(adev->gfx.me.me_fw_data_obj);
+	amdgpu_bo_unreserve(adev->gfx.me.me_fw_obj);
+	amdgpu_bo_unreserve(adev->gfx.me.me_fw_data_obj);
+
+	if (amdgpu_emu_mode == 1)
+		amdgpu_device_flush_hdp(adev, NULL);
+
+	WREG32_SOC15(GC, 0, regCP_ME_IC_BASE_LO,
+		lower_32_bits(adev->gfx.me.me_fw_gpu_addr));
+	WREG32_SOC15(GC, 0, regCP_ME_IC_BASE_HI,
+		upper_32_bits(adev->gfx.me.me_fw_gpu_addr));
+
+	tmp = RREG32_SOC15(GC, 0, regCP_ME_IC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_BASE_CNTL, CACHE_POLICY, 0);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_BASE_CNTL, EXE_DISABLE, 0);
+	WREG32_SOC15(GC, 0, regCP_ME_IC_BASE_CNTL, tmp);
+
+	/*
+	 * Programming any of the CP_ME_IC_BASE registers
+	 * forces invalidation of the ME L1 I$. Wait for the
+	 * invalidation complete
+	 */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_ME_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_ME_IC_OP_CNTL,
+			INVALIDATE_CACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	/* Prime the instruction caches */
+	tmp = RREG32_SOC15(GC, 0, regCP_ME_IC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_ME_IC_OP_CNTL, PRIME_ICACHE, 1);
+	WREG32_SOC15(GC, 0, regCP_ME_IC_OP_CNTL, tmp);
+
+	/* Waiting for instruction cache primed*/
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_ME_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_ME_IC_OP_CNTL,
+			ICACHE_PRIMED))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to prime instruction cache\n");
+		return -EINVAL;
+	}
+
+	mutex_lock(&adev->srbm_mutex);
+	for (pipe_id = 0; pipe_id < adev->gfx.me.num_pipe_per_me; pipe_id++) {
+		soc24_grbm_select(adev, 0, pipe_id, 0, 0);
+
+		WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE1_LO,
+			lower_32_bits(adev->gfx.me.me_fw_data_gpu_addr));
+		WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE1_HI,
+			upper_32_bits(adev->gfx.me.me_fw_data_gpu_addr));
+	}
+	soc24_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	tmp = RREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_RS64_DC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_RS64_DC_BASE_CNTL, CACHE_POLICY, 0);
+	WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_BASE_CNTL, tmp);
+
+	/* Invalidate the data caches */
+	tmp = RREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_RS64_DC_OP_CNTL, INVALIDATE_DCACHE, 1);
+	WREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_OP_CNTL, tmp);
+
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_GFX_RS64_DC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_GFX_RS64_DC_OP_CNTL,
+			INVALIDATE_DCACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate RS64 data cache\n");
+		return -EINVAL;
+	}
+
+	gfx_v12_0_set_me_ucode_start_addr(adev);
+
+	return 0;
+}
+
+static int gfx_v12_0_cp_gfx_load_microcode(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (!adev->gfx.me_fw || !adev->gfx.pfp_fw)
+		return -EINVAL;
+
+	gfx_v12_0_cp_gfx_enable(adev, false);
+
+	r = gfx_v12_0_cp_gfx_load_pfp_microcode_rs64(adev);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to load pfp fw\n", r);
+		return r;
+	}
+
+	r = gfx_v12_0_cp_gfx_load_me_microcode_rs64(adev);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to load me fw\n", r);
+		return r;
+	}
+
+	return 0;
+}
+
+static int gfx_v12_0_cp_gfx_start(struct amdgpu_device *adev)
+{
+	/* init the CP */
+	WREG32_SOC15(GC, 0, regCP_MAX_CONTEXT,
+		     adev->gfx.config.max_hw_contexts - 1);
+	WREG32_SOC15(GC, 0, regCP_DEVICE_ID, 1);
+
+	if (!amdgpu_async_gfx_ring)
+		gfx_v12_0_cp_gfx_enable(adev, true);
+
+	return 0;
+}
+
+static void gfx_v12_0_cp_gfx_switch_pipe(struct amdgpu_device *adev,
+					 CP_PIPE_ID pipe)
+{
+	u32 tmp;
+
+	tmp = RREG32_SOC15(GC, 0, regGRBM_GFX_CNTL);
+	tmp = REG_SET_FIELD(tmp, GRBM_GFX_CNTL, PIPEID, pipe);
+
+	WREG32_SOC15(GC, 0, regGRBM_GFX_CNTL, tmp);
+}
+
+static void gfx_v12_0_cp_gfx_set_doorbell(struct amdgpu_device *adev,
+					  struct amdgpu_ring *ring)
+{
+	u32 tmp;
+
+	tmp = RREG32_SOC15(GC, 0, regCP_RB_DOORBELL_CONTROL);
+	if (ring->use_doorbell) {
+		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+				    DOORBELL_OFFSET, ring->doorbell_index);
+		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+				    DOORBELL_EN, 1);
+	} else {
+		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+				    DOORBELL_EN, 0);
+	}
+	WREG32_SOC15(GC, 0, regCP_RB_DOORBELL_CONTROL, tmp);
+
+	tmp = REG_SET_FIELD(0, CP_RB_DOORBELL_RANGE_LOWER,
+			    DOORBELL_RANGE_LOWER, ring->doorbell_index);
+	WREG32_SOC15(GC, 0, regCP_RB_DOORBELL_RANGE_LOWER, tmp);
+
+	WREG32_SOC15(GC, 0, regCP_RB_DOORBELL_RANGE_UPPER,
+		     CP_RB_DOORBELL_RANGE_UPPER__DOORBELL_RANGE_UPPER_MASK);
+}
+
+static int gfx_v12_0_cp_gfx_resume(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	u32 tmp;
+	u32 rb_bufsz;
+	u64 rb_addr, rptr_addr, wptr_gpu_addr;
+
+	/* Set the write pointer delay */
+	WREG32_SOC15(GC, 0, regCP_RB_WPTR_DELAY, 0);
+
+	/* set the RB to use vmid 0 */
+	WREG32_SOC15(GC, 0, regCP_RB_VMID, 0);
+
+	/* Init gfx ring 0 for pipe 0 */
+	mutex_lock(&adev->srbm_mutex);
+	gfx_v12_0_cp_gfx_switch_pipe(adev, PIPE_ID0);
+
+	/* Set ring buffer size */
+	ring = &adev->gfx.gfx_ring[0];
+	rb_bufsz = order_base_2(ring->ring_size / 8);
+	tmp = REG_SET_FIELD(0, CP_RB0_CNTL, RB_BUFSZ, rb_bufsz);
+	tmp = REG_SET_FIELD(tmp, CP_RB0_CNTL, RB_BLKSZ, rb_bufsz - 2);
+	WREG32_SOC15(GC, 0, regCP_RB0_CNTL, tmp);
+
+	/* Initialize the ring buffer's write pointers */
+	ring->wptr = 0;
+	WREG32_SOC15(GC, 0, regCP_RB0_WPTR, lower_32_bits(ring->wptr));
+	WREG32_SOC15(GC, 0, regCP_RB0_WPTR_HI, upper_32_bits(ring->wptr));
+
+	/* set the wb address whether it's enabled or not */
+	rptr_addr = ring->rptr_gpu_addr;
+	WREG32_SOC15(GC, 0, regCP_RB0_RPTR_ADDR, lower_32_bits(rptr_addr));
+	WREG32_SOC15(GC, 0, regCP_RB0_RPTR_ADDR_HI, upper_32_bits(rptr_addr) &
+		     CP_RB_RPTR_ADDR_HI__RB_RPTR_ADDR_HI_MASK);
+
+	wptr_gpu_addr = ring->wptr_gpu_addr;
+	WREG32_SOC15(GC, 0, regCP_RB_WPTR_POLL_ADDR_LO,
+		     lower_32_bits(wptr_gpu_addr));
+	WREG32_SOC15(GC, 0, regCP_RB_WPTR_POLL_ADDR_HI,
+		     upper_32_bits(wptr_gpu_addr));
+
+	mdelay(1);
+	WREG32_SOC15(GC, 0, regCP_RB0_CNTL, tmp);
+
+	rb_addr = ring->gpu_addr >> 8;
+	WREG32_SOC15(GC, 0, regCP_RB0_BASE, rb_addr);
+	WREG32_SOC15(GC, 0, regCP_RB0_BASE_HI, upper_32_bits(rb_addr));
+
+	WREG32_SOC15(GC, 0, regCP_RB_ACTIVE, 1);
+
+	gfx_v12_0_cp_gfx_set_doorbell(adev, ring);
+	mutex_unlock(&adev->srbm_mutex);
+
+	/* Switch to pipe 0 */
+	mutex_lock(&adev->srbm_mutex);
+	gfx_v12_0_cp_gfx_switch_pipe(adev, PIPE_ID0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	/* start the ring */
+	gfx_v12_0_cp_gfx_start(adev);
+	return 0;
+}
+
+static void gfx_v12_0_cp_compute_enable(struct amdgpu_device *adev, bool enable)
+{
+	u32 data;
+
+	data = RREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL);
+	data = REG_SET_FIELD(data, CP_MEC_RS64_CNTL, MEC_INVALIDATE_ICACHE,
+						 enable ? 0 : 1);
+	data = REG_SET_FIELD(data, CP_MEC_RS64_CNTL, MEC_PIPE0_RESET,
+						 enable ? 0 : 1);
+	data = REG_SET_FIELD(data, CP_MEC_RS64_CNTL, MEC_PIPE1_RESET,
+						 enable ? 0 : 1);
+	data = REG_SET_FIELD(data, CP_MEC_RS64_CNTL, MEC_PIPE2_RESET,
+						 enable ? 0 : 1);
+	data = REG_SET_FIELD(data, CP_MEC_RS64_CNTL, MEC_PIPE3_RESET,
+						 enable ? 0 : 1);
+	data = REG_SET_FIELD(data, CP_MEC_RS64_CNTL, MEC_PIPE0_ACTIVE,
+						 enable ? 1 : 0);
+	data = REG_SET_FIELD(data, CP_MEC_RS64_CNTL, MEC_PIPE1_ACTIVE,
+			                         enable ? 1 : 0);
+	data = REG_SET_FIELD(data, CP_MEC_RS64_CNTL, MEC_PIPE2_ACTIVE,
+						 enable ? 1 : 0);
+	data = REG_SET_FIELD(data, CP_MEC_RS64_CNTL, MEC_PIPE3_ACTIVE,
+						 enable ? 1 : 0);
+	data = REG_SET_FIELD(data, CP_MEC_RS64_CNTL, MEC_HALT,
+						 enable ? 0 : 1);
+	WREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL, data);
+
+	adev->gfx.kiq[0].ring.sched.ready = enable;
+
+	udelay(50);
+}
+
+static int gfx_v12_0_cp_compute_load_microcode_rs64(struct amdgpu_device *adev)
+{
+	const struct gfx_firmware_header_v2_0 *mec_hdr;
+	const __le32 *fw_ucode, *fw_data;
+	u32 tmp, fw_ucode_size, fw_data_size;
+	u32 i, usec_timeout = 50000; /* Wait for 50 ms */
+	u32 *fw_ucode_ptr, *fw_data_ptr;
+	int r;
+
+	if (!adev->gfx.mec_fw)
+		return -EINVAL;
+
+	gfx_v12_0_cp_compute_enable(adev, false);
+
+	mec_hdr = (const struct gfx_firmware_header_v2_0 *)adev->gfx.mec_fw->data;
+	amdgpu_ucode_print_gfx_hdr(&mec_hdr->header);
+
+	fw_ucode = (const __le32 *) (adev->gfx.mec_fw->data +
+				le32_to_cpu(mec_hdr->ucode_offset_bytes));
+	fw_ucode_size = le32_to_cpu(mec_hdr->ucode_size_bytes);
+
+	fw_data = (const __le32 *) (adev->gfx.mec_fw->data +
+				le32_to_cpu(mec_hdr->data_offset_bytes));
+	fw_data_size = le32_to_cpu(mec_hdr->data_size_bytes);
+
+	r = amdgpu_bo_create_reserved(adev, fw_ucode_size,
+				      64 * 1024, AMDGPU_GEM_DOMAIN_VRAM,
+				      &adev->gfx.mec.mec_fw_obj,
+				      &adev->gfx.mec.mec_fw_gpu_addr,
+				      (void **)&fw_ucode_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create mec fw ucode bo\n", r);
+		gfx_v12_0_mec_fini(adev);
+		return r;
+	}
+
+	r = amdgpu_bo_create_reserved(adev,
+				      ALIGN(fw_data_size, 64 * 1024) *
+				      adev->gfx.mec.num_pipe_per_mec,
+				      64 * 1024, AMDGPU_GEM_DOMAIN_VRAM,
+				      &adev->gfx.mec.mec_fw_data_obj,
+				      &adev->gfx.mec.mec_fw_data_gpu_addr,
+				      (void **)&fw_data_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create mec fw ucode bo\n", r);
+		gfx_v12_0_mec_fini(adev);
+		return r;
+	}
+
+	memcpy(fw_ucode_ptr, fw_ucode, fw_ucode_size);
+	for (i = 0; i < adev->gfx.mec.num_pipe_per_mec; i++) {
+		memcpy(fw_data_ptr + i * ALIGN(fw_data_size, 64 * 1024) / 4, fw_data, fw_data_size);
+	}
+
+	amdgpu_bo_kunmap(adev->gfx.mec.mec_fw_obj);
+	amdgpu_bo_kunmap(adev->gfx.mec.mec_fw_data_obj);
+	amdgpu_bo_unreserve(adev->gfx.mec.mec_fw_obj);
+	amdgpu_bo_unreserve(adev->gfx.mec.mec_fw_data_obj);
+
+	tmp = RREG32_SOC15(GC, 0, regCP_CPC_IC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, EXE_DISABLE, 0);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, CACHE_POLICY, 0);
+	WREG32_SOC15(GC, 0, regCP_CPC_IC_BASE_CNTL, tmp);
+
+	tmp = RREG32_SOC15(GC, 0, regCP_MEC_DC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_DC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_DC_BASE_CNTL, CACHE_POLICY, 0);
+	WREG32_SOC15(GC, 0, regCP_MEC_DC_BASE_CNTL, tmp);
+
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < adev->gfx.mec.num_pipe_per_mec; i++) {
+		soc24_grbm_select(adev, 1, i, 0, 0);
+
+		WREG32_SOC15(GC, 0, regCP_MEC_MDBASE_LO,
+			     lower_32_bits(adev->gfx.mec.mec_fw_data_gpu_addr +
+					   i * ALIGN(fw_data_size, 64 * 1024)));
+		WREG32_SOC15(GC, 0, regCP_MEC_MDBASE_HI,
+			     upper_32_bits(adev->gfx.mec.mec_fw_data_gpu_addr +
+					   i * ALIGN(fw_data_size, 64 * 1024)));
+
+		WREG32_SOC15(GC, 0, regCP_CPC_IC_BASE_LO,
+			     lower_32_bits(adev->gfx.mec.mec_fw_gpu_addr));
+		WREG32_SOC15(GC, 0, regCP_CPC_IC_BASE_HI,
+			     upper_32_bits(adev->gfx.mec.mec_fw_gpu_addr));
+	}
+	mutex_unlock(&adev->srbm_mutex);
+	soc24_grbm_select(adev, 0, 0, 0, 0);
+
+	/* Trigger an invalidation of the L1 instruction caches */
+	tmp = RREG32_SOC15(GC, 0, regCP_MEC_DC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_DC_OP_CNTL, INVALIDATE_DCACHE, 1);
+	WREG32_SOC15(GC, 0, regCP_MEC_DC_OP_CNTL, tmp);
+
+	/* Wait for invalidation complete */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_MEC_DC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_MEC_DC_OP_CNTL,
+				       INVALIDATE_DCACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	/* Trigger an invalidation of the L1 instruction caches */
+	tmp = RREG32_SOC15(GC, 0, regCP_CPC_IC_OP_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_OP_CNTL, INVALIDATE_CACHE, 1);
+	WREG32_SOC15(GC, 0, regCP_CPC_IC_OP_CNTL, tmp);
+
+	/* Wait for invalidation complete */
+	for (i = 0; i < usec_timeout; i++) {
+		tmp = RREG32_SOC15(GC, 0, regCP_CPC_IC_OP_CNTL);
+		if (1 == REG_GET_FIELD(tmp, CP_CPC_IC_OP_CNTL,
+				       INVALIDATE_CACHE_COMPLETE))
+			break;
+		udelay(1);
+	}
+
+	if (i >= usec_timeout) {
+		dev_err(adev->dev, "failed to invalidate instruction cache\n");
+		return -EINVAL;
+	}
+
+	gfx_v12_0_set_mec_ucode_start_addr(adev);
+
+	return 0;
+}
+
+static void gfx_v12_0_kiq_setting(struct amdgpu_ring *ring)
+{
+	uint32_t tmp;
+	struct amdgpu_device *adev = ring->adev;
+
+	/* tell RLC which is KIQ queue */
+	tmp = RREG32_SOC15(GC, 0, regRLC_CP_SCHEDULERS);
+	tmp &= 0xffffff00;
+	tmp |= (ring->me << 5) | (ring->pipe << 3) | (ring->queue);
+	WREG32_SOC15(GC, 0, regRLC_CP_SCHEDULERS, tmp | 0x80);
+}
+
+static void gfx_v12_0_cp_set_doorbell_range(struct amdgpu_device *adev)
+{
+	/* set graphics engine doorbell range */
+	WREG32_SOC15(GC, 0, regCP_RB_DOORBELL_RANGE_LOWER,
+		     (adev->doorbell_index.gfx_ring0 * 2) << 2);
+	WREG32_SOC15(GC, 0, regCP_RB_DOORBELL_RANGE_UPPER,
+		     (adev->doorbell_index.gfx_userqueue_end * 2) << 2);
+
+	/* set compute engine doorbell range */
+	WREG32_SOC15(GC, 0, regCP_MEC_DOORBELL_RANGE_LOWER,
+		     (adev->doorbell_index.kiq * 2) << 2);
+	WREG32_SOC15(GC, 0, regCP_MEC_DOORBELL_RANGE_UPPER,
+		     (adev->doorbell_index.userqueue_end * 2) << 2);
+}
+
+static int gfx_v12_0_gfx_mqd_init(struct amdgpu_device *adev, void *m,
+				  struct amdgpu_mqd_prop *prop)
+{
+	struct v12_gfx_mqd *mqd = m;
+	uint64_t hqd_gpu_addr, wb_gpu_addr;
+	uint32_t tmp;
+	uint32_t rb_bufsz;
+
+	/* set up gfx hqd wptr */
+	mqd->cp_gfx_hqd_wptr = 0;
+	mqd->cp_gfx_hqd_wptr_hi = 0;
+
+	/* set the pointer to the MQD */
+	mqd->cp_mqd_base_addr = prop->mqd_gpu_addr & 0xfffffffc;
+	mqd->cp_mqd_base_addr_hi = upper_32_bits(prop->mqd_gpu_addr);
+
+	/* set up mqd control */
+	tmp = regCP_GFX_MQD_CONTROL_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_GFX_MQD_CONTROL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_MQD_CONTROL, PRIV_STATE, 1);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_MQD_CONTROL, CACHE_POLICY, 0);
+	mqd->cp_gfx_mqd_control = tmp;
+
+	/* set up gfx_hqd_vimd with 0x0 to indicate the ring buffer's vmid */
+	tmp = regCP_GFX_HQD_VMID_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_VMID, VMID, 0);
+	mqd->cp_gfx_hqd_vmid = 0;
+
+	/* set up default queue priority level
+	 * 0x0 = low priority, 0x1 = high priority */
+	tmp = regCP_GFX_HQD_QUEUE_PRIORITY_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_QUEUE_PRIORITY, PRIORITY_LEVEL, 0);
+	mqd->cp_gfx_hqd_queue_priority = tmp;
+
+	/* set up time quantum */
+	tmp = regCP_GFX_HQD_QUANTUM_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_QUANTUM, QUANTUM_EN, 1);
+	mqd->cp_gfx_hqd_quantum = tmp;
+
+	/* set up gfx hqd base. this is similar as CP_RB_BASE */
+	hqd_gpu_addr = prop->hqd_base_gpu_addr >> 8;
+	mqd->cp_gfx_hqd_base = hqd_gpu_addr;
+	mqd->cp_gfx_hqd_base_hi = upper_32_bits(hqd_gpu_addr);
+
+	/* set up hqd_rptr_addr/_hi, similar as CP_RB_RPTR */
+	wb_gpu_addr = prop->rptr_gpu_addr;
+	mqd->cp_gfx_hqd_rptr_addr = wb_gpu_addr & 0xfffffffc;
+	mqd->cp_gfx_hqd_rptr_addr_hi =
+		upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	/* set up rb_wptr_poll addr */
+	wb_gpu_addr = prop->wptr_gpu_addr;
+	mqd->cp_rb_wptr_poll_addr_lo = wb_gpu_addr & 0xfffffffc;
+	mqd->cp_rb_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	/* set up the gfx_hqd_control, similar as CP_RB0_CNTL */
+	rb_bufsz = order_base_2(prop->queue_size / 4) - 1;
+	tmp = regCP_GFX_HQD_CNTL_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, RB_BUFSZ, rb_bufsz);
+	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, RB_BLKSZ, rb_bufsz - 2);
+#ifdef __BIG_ENDIAN
+	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, BUF_SWAP, 1);
+#endif
+	if (prop->tmz_queue)
+		tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, TMZ_MATCH, 1);
+	if (!prop->kernel_queue)
+		tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, RB_NON_PRIV, 1);
+	mqd->cp_gfx_hqd_cntl = tmp;
+
+	/* set up cp_doorbell_control */
+	tmp = regCP_RB_DOORBELL_CONTROL_DEFAULT;
+	if (prop->use_doorbell) {
+		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+				    DOORBELL_OFFSET, prop->doorbell_index);
+		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+				    DOORBELL_EN, 1);
+	} else
+		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+				    DOORBELL_EN, 0);
+	mqd->cp_rb_doorbell_control = tmp;
+
+	/* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */
+	mqd->cp_gfx_hqd_rptr = regCP_GFX_HQD_RPTR_DEFAULT;
+
+	/* active the queue */
+	mqd->cp_gfx_hqd_active = 1;
+
+	/* set gfx UQ items */
+	mqd->shadow_base_lo = lower_32_bits(prop->shadow_addr);
+	mqd->shadow_base_hi = upper_32_bits(prop->shadow_addr);
+	mqd->fw_work_area_base_lo = lower_32_bits(prop->csa_addr);
+	mqd->fw_work_area_base_hi = upper_32_bits(prop->csa_addr);
+	mqd->fence_address_lo = lower_32_bits(prop->fence_address);
+	mqd->fence_address_hi = upper_32_bits(prop->fence_address);
+
+	return 0;
+}
+
+static int gfx_v12_0_kgq_init_queue(struct amdgpu_ring *ring, bool reset)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v12_gfx_mqd *mqd = ring->mqd_ptr;
+	int mqd_idx = ring - &adev->gfx.gfx_ring[0];
+
+	if (!reset && !amdgpu_in_reset(adev) && !adev->in_suspend) {
+		memset((void *)mqd, 0, sizeof(*mqd));
+		mutex_lock(&adev->srbm_mutex);
+		soc24_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		amdgpu_ring_init_mqd(ring);
+		soc24_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+		if (adev->gfx.me.mqd_backup[mqd_idx])
+			memcpy_fromio(adev->gfx.me.mqd_backup[mqd_idx], mqd, sizeof(*mqd));
+	} else {
+		/* restore mqd with the backup copy */
+		if (adev->gfx.me.mqd_backup[mqd_idx])
+			memcpy_toio(mqd, adev->gfx.me.mqd_backup[mqd_idx], sizeof(*mqd));
+		/* reset the ring */
+		ring->wptr = 0;
+		*ring->wptr_cpu_addr = 0;
+		amdgpu_ring_clear_ring(ring);
+	}
+
+	return 0;
+}
+
+static int gfx_v12_0_cp_async_gfx_ring_resume(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+		r = gfx_v12_0_kgq_init_queue(&adev->gfx.gfx_ring[i], false);
+		if (r)
+			return r;
+	}
+
+	r = amdgpu_gfx_enable_kgq(adev, 0);
+	if (r)
+		return r;
+
+	return gfx_v12_0_cp_gfx_start(adev);
+}
+
+static int gfx_v12_0_compute_mqd_init(struct amdgpu_device *adev, void *m,
+				      struct amdgpu_mqd_prop *prop)
+{
+	struct v12_compute_mqd *mqd = m;
+	uint64_t hqd_gpu_addr, wb_gpu_addr, eop_base_addr;
+	uint32_t tmp;
+
+	mqd->header = 0xC0310800;
+	mqd->compute_pipelinestat_enable = 0x00000001;
+	mqd->compute_static_thread_mgmt_se0 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se1 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se2 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se3 = 0xffffffff;
+	mqd->compute_misc_reserved = 0x00000007;
+
+	eop_base_addr = prop->eop_gpu_addr >> 8;
+	mqd->cp_hqd_eop_base_addr_lo = eop_base_addr;
+	mqd->cp_hqd_eop_base_addr_hi = upper_32_bits(eop_base_addr);
+
+	/* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */
+	tmp = regCP_HQD_EOP_CONTROL_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_HQD_EOP_CONTROL, EOP_SIZE,
+			(order_base_2(GFX12_MEC_HPD_SIZE / 4) - 1));
+
+	mqd->cp_hqd_eop_control = tmp;
+
+	/* enable doorbell? */
+	tmp = regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT;
+
+	if (prop->use_doorbell) {
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_OFFSET, prop->doorbell_index);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_EN, 1);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_SOURCE, 0);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_HIT, 0);
+	} else {
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_EN, 0);
+	}
+
+	mqd->cp_hqd_pq_doorbell_control = tmp;
+
+	/* disable the queue if it's active */
+	mqd->cp_hqd_dequeue_request = 0;
+	mqd->cp_hqd_pq_rptr = 0;
+	mqd->cp_hqd_pq_wptr_lo = 0;
+	mqd->cp_hqd_pq_wptr_hi = 0;
+
+	/* set the pointer to the MQD */
+	mqd->cp_mqd_base_addr_lo = prop->mqd_gpu_addr & 0xfffffffc;
+	mqd->cp_mqd_base_addr_hi = upper_32_bits(prop->mqd_gpu_addr);
+
+	/* set MQD vmid to 0 */
+	tmp = regCP_MQD_CONTROL_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_MQD_CONTROL, VMID, 0);
+	mqd->cp_mqd_control = tmp;
+
+	/* set the pointer to the HQD, this is similar CP_RB0_BASE/_HI */
+	hqd_gpu_addr = prop->hqd_base_gpu_addr >> 8;
+	mqd->cp_hqd_pq_base_lo = hqd_gpu_addr;
+	mqd->cp_hqd_pq_base_hi = upper_32_bits(hqd_gpu_addr);
+
+	/* set up the HQD, this is similar to CP_RB0_CNTL */
+	tmp = regCP_HQD_PQ_CONTROL_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, QUEUE_SIZE,
+			    (order_base_2(prop->queue_size / 4) - 1));
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, RPTR_BLOCK_SIZE,
+			    (order_base_2(AMDGPU_GPU_PAGE_SIZE / 4) - 1));
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, UNORD_DISPATCH, 1);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, TUNNEL_DISPATCH, 0);
+	if (prop->kernel_queue) {
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, PRIV_STATE, 1);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, KMD_QUEUE, 1);
+	}
+	if (prop->tmz_queue)
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, TMZ, 1);
+	mqd->cp_hqd_pq_control = tmp;
+
+	/* set the wb address whether it's enabled or not */
+	wb_gpu_addr = prop->rptr_gpu_addr;
+	mqd->cp_hqd_pq_rptr_report_addr_lo = wb_gpu_addr & 0xfffffffc;
+	mqd->cp_hqd_pq_rptr_report_addr_hi =
+		upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	/* only used if CP_PQ_WPTR_POLL_CNTL.CP_PQ_WPTR_POLL_CNTL__EN_MASK=1 */
+	wb_gpu_addr = prop->wptr_gpu_addr;
+	mqd->cp_hqd_pq_wptr_poll_addr_lo = wb_gpu_addr & 0xfffffffc;
+	mqd->cp_hqd_pq_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	tmp = 0;
+	/* enable the doorbell if requested */
+	if (prop->use_doorbell) {
+		tmp = regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT;
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				DOORBELL_OFFSET, prop->doorbell_index);
+
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_EN, 1);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_SOURCE, 0);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_HIT, 0);
+	}
+
+	mqd->cp_hqd_pq_doorbell_control = tmp;
+
+	/* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */
+	mqd->cp_hqd_pq_rptr = regCP_HQD_PQ_RPTR_DEFAULT;
+
+	/* set the vmid for the queue */
+	mqd->cp_hqd_vmid = 0;
+
+	tmp = regCP_HQD_PERSISTENT_STATE_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PERSISTENT_STATE, PRELOAD_SIZE, 0x55);
+	mqd->cp_hqd_persistent_state = tmp;
+
+	/* set MIN_IB_AVAIL_SIZE */
+	tmp = regCP_HQD_IB_CONTROL_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_HQD_IB_CONTROL, MIN_IB_AVAIL_SIZE, 3);
+	mqd->cp_hqd_ib_control = tmp;
+
+	/* set static priority for a compute queue/ring */
+	mqd->cp_hqd_pipe_priority = prop->hqd_pipe_priority;
+	mqd->cp_hqd_queue_priority = prop->hqd_queue_priority;
+
+	mqd->cp_hqd_active = prop->hqd_active;
+
+	/* set UQ fenceaddress */
+	mqd->fence_address_lo = lower_32_bits(prop->fence_address);
+	mqd->fence_address_hi = upper_32_bits(prop->fence_address);
+
+	return 0;
+}
+
+static int gfx_v12_0_kiq_init_register(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v12_compute_mqd *mqd = ring->mqd_ptr;
+	int j;
+
+	/* inactivate the queue */
+	if (amdgpu_sriov_vf(adev))
+		WREG32_SOC15(GC, 0, regCP_HQD_ACTIVE, 0);
+
+	/* disable wptr polling */
+	WREG32_FIELD15_PREREG(GC, 0, CP_PQ_WPTR_POLL_CNTL, EN, 0);
+
+	/* write the EOP addr */
+	WREG32_SOC15(GC, 0, regCP_HQD_EOP_BASE_ADDR,
+	       mqd->cp_hqd_eop_base_addr_lo);
+	WREG32_SOC15(GC, 0, regCP_HQD_EOP_BASE_ADDR_HI,
+	       mqd->cp_hqd_eop_base_addr_hi);
+
+	/* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */
+	WREG32_SOC15(GC, 0, regCP_HQD_EOP_CONTROL,
+	       mqd->cp_hqd_eop_control);
+
+	/* enable doorbell? */
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL,
+	       mqd->cp_hqd_pq_doorbell_control);
+
+	/* disable the queue if it's active */
+	if (RREG32_SOC15(GC, 0, regCP_HQD_ACTIVE) & 1) {
+		WREG32_SOC15(GC, 0, regCP_HQD_DEQUEUE_REQUEST, 1);
+		for (j = 0; j < adev->usec_timeout; j++) {
+			if (!(RREG32_SOC15(GC, 0, regCP_HQD_ACTIVE) & 1))
+				break;
+			udelay(1);
+		}
+		WREG32_SOC15(GC, 0, regCP_HQD_DEQUEUE_REQUEST,
+		       mqd->cp_hqd_dequeue_request);
+		WREG32_SOC15(GC, 0, regCP_HQD_PQ_RPTR,
+		       mqd->cp_hqd_pq_rptr);
+		WREG32_SOC15(GC, 0, regCP_HQD_PQ_WPTR_LO,
+		       mqd->cp_hqd_pq_wptr_lo);
+		WREG32_SOC15(GC, 0, regCP_HQD_PQ_WPTR_HI,
+		       mqd->cp_hqd_pq_wptr_hi);
+	}
+
+	/* set the pointer to the MQD */
+	WREG32_SOC15(GC, 0, regCP_MQD_BASE_ADDR,
+	       mqd->cp_mqd_base_addr_lo);
+	WREG32_SOC15(GC, 0, regCP_MQD_BASE_ADDR_HI,
+	       mqd->cp_mqd_base_addr_hi);
+
+	/* set MQD vmid to 0 */
+	WREG32_SOC15(GC, 0, regCP_MQD_CONTROL,
+	       mqd->cp_mqd_control);
+
+	/* set the pointer to the HQD, this is similar CP_RB0_BASE/_HI */
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_BASE,
+	       mqd->cp_hqd_pq_base_lo);
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_BASE_HI,
+	       mqd->cp_hqd_pq_base_hi);
+
+	/* set up the HQD, this is similar to CP_RB0_CNTL */
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_CONTROL,
+	       mqd->cp_hqd_pq_control);
+
+	/* set the wb address whether it's enabled or not */
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_RPTR_REPORT_ADDR,
+		mqd->cp_hqd_pq_rptr_report_addr_lo);
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_RPTR_REPORT_ADDR_HI,
+		mqd->cp_hqd_pq_rptr_report_addr_hi);
+
+	/* only used if CP_PQ_WPTR_POLL_CNTL.CP_PQ_WPTR_POLL_CNTL__EN_MASK=1 */
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_WPTR_POLL_ADDR,
+	       mqd->cp_hqd_pq_wptr_poll_addr_lo);
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_WPTR_POLL_ADDR_HI,
+	       mqd->cp_hqd_pq_wptr_poll_addr_hi);
+
+	/* enable the doorbell if requested */
+	if (ring->use_doorbell) {
+		WREG32_SOC15(GC, 0, regCP_MEC_DOORBELL_RANGE_LOWER,
+			(adev->doorbell_index.kiq * 2) << 2);
+		WREG32_SOC15(GC, 0, regCP_MEC_DOORBELL_RANGE_UPPER,
+			(adev->doorbell_index.userqueue_end * 2) << 2);
+	}
+
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL,
+	       mqd->cp_hqd_pq_doorbell_control);
+
+	/* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_WPTR_LO,
+	       mqd->cp_hqd_pq_wptr_lo);
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_WPTR_HI,
+	       mqd->cp_hqd_pq_wptr_hi);
+
+	/* set the vmid for the queue */
+	WREG32_SOC15(GC, 0, regCP_HQD_VMID, mqd->cp_hqd_vmid);
+
+	WREG32_SOC15(GC, 0, regCP_HQD_PERSISTENT_STATE,
+	       mqd->cp_hqd_persistent_state);
+
+	/* activate the queue */
+	WREG32_SOC15(GC, 0, regCP_HQD_ACTIVE,
+	       mqd->cp_hqd_active);
+
+	if (ring->use_doorbell)
+		WREG32_FIELD15_PREREG(GC, 0, CP_PQ_STATUS, DOORBELL_ENABLE, 1);
+
+	return 0;
+}
+
+static int gfx_v12_0_kiq_init_queue(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v12_compute_mqd *mqd = ring->mqd_ptr;
+	int mqd_idx = AMDGPU_MAX_COMPUTE_RINGS;
+
+	gfx_v12_0_kiq_setting(ring);
+
+	if (amdgpu_in_reset(adev)) { /* for GPU_RESET case */
+		/* reset MQD to a clean status */
+		if (adev->gfx.mec.mqd_backup[mqd_idx])
+			memcpy(mqd, adev->gfx.mec.mqd_backup[mqd_idx], sizeof(*mqd));
+
+		/* reset ring buffer */
+		ring->wptr = 0;
+		amdgpu_ring_clear_ring(ring);
+
+		mutex_lock(&adev->srbm_mutex);
+		soc24_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		gfx_v12_0_kiq_init_register(ring);
+		soc24_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+	} else {
+		memset((void *)mqd, 0, sizeof(*mqd));
+		if (amdgpu_sriov_vf(adev) && adev->in_suspend)
+			amdgpu_ring_clear_ring(ring);
+		mutex_lock(&adev->srbm_mutex);
+		soc24_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		amdgpu_ring_init_mqd(ring);
+		gfx_v12_0_kiq_init_register(ring);
+		soc24_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+
+		if (adev->gfx.mec.mqd_backup[mqd_idx])
+			memcpy(adev->gfx.mec.mqd_backup[mqd_idx], mqd, sizeof(*mqd));
+	}
+
+	return 0;
+}
+
+static int gfx_v12_0_kcq_init_queue(struct amdgpu_ring *ring, bool reset)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v12_compute_mqd *mqd = ring->mqd_ptr;
+	int mqd_idx = ring - &adev->gfx.compute_ring[0];
+
+	if (!reset && !amdgpu_in_reset(adev) && !adev->in_suspend) {
+		memset((void *)mqd, 0, sizeof(*mqd));
+		mutex_lock(&adev->srbm_mutex);
+		soc24_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		amdgpu_ring_init_mqd(ring);
+		soc24_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+
+		if (adev->gfx.mec.mqd_backup[mqd_idx])
+			memcpy_fromio(adev->gfx.mec.mqd_backup[mqd_idx], mqd, sizeof(*mqd));
+	} else {
+		/* restore MQD to a clean status */
+		if (adev->gfx.mec.mqd_backup[mqd_idx])
+			memcpy_toio(mqd, adev->gfx.mec.mqd_backup[mqd_idx], sizeof(*mqd));
+		/* reset ring buffer */
+		ring->wptr = 0;
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr, 0);
+		amdgpu_ring_clear_ring(ring);
+	}
+
+	return 0;
+}
+
+static int gfx_v12_0_kiq_resume(struct amdgpu_device *adev)
+{
+	gfx_v12_0_kiq_init_queue(&adev->gfx.kiq[0].ring);
+	adev->gfx.kiq[0].ring.sched.ready = true;
+	return 0;
+}
+
+static int gfx_v12_0_kcq_resume(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	if (!amdgpu_async_gfx_ring)
+		gfx_v12_0_cp_compute_enable(adev, true);
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		r = gfx_v12_0_kcq_init_queue(&adev->gfx.compute_ring[i], false);
+		if (r)
+			return r;
+	}
+
+	return amdgpu_gfx_enable_kcq(adev, 0);
+}
+
+static int gfx_v12_0_cp_resume(struct amdgpu_device *adev)
+{
+	int r, i;
+	struct amdgpu_ring *ring;
+
+	if (!(adev->flags & AMD_IS_APU))
+		gfx_v12_0_enable_gui_idle_interrupt(adev, false);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+		/* legacy firmware loading */
+		r = gfx_v12_0_cp_gfx_load_microcode(adev);
+		if (r)
+			return r;
+
+		r = gfx_v12_0_cp_compute_load_microcode_rs64(adev);
+		if (r)
+			return r;
+	}
+
+	gfx_v12_0_cp_set_doorbell_range(adev);
+
+	if (amdgpu_async_gfx_ring) {
+		gfx_v12_0_cp_compute_enable(adev, true);
+		gfx_v12_0_cp_gfx_enable(adev, true);
+	}
+
+	if (adev->enable_mes_kiq && adev->mes.kiq_hw_init)
+		r = amdgpu_mes_kiq_hw_init(adev);
+	else
+		r = gfx_v12_0_kiq_resume(adev);
+	if (r)
+		return r;
+
+	r = gfx_v12_0_kcq_resume(adev);
+	if (r)
+		return r;
+
+	if (!amdgpu_async_gfx_ring) {
+		r = gfx_v12_0_cp_gfx_resume(adev);
+		if (r)
+			return r;
+	} else {
+		r = gfx_v12_0_cp_async_gfx_ring_resume(adev);
+		if (r)
+			return r;
+	}
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+		ring = &adev->gfx.gfx_ring[i];
+		r = amdgpu_ring_test_helper(ring);
+		if (r)
+			return r;
+	}
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		ring = &adev->gfx.compute_ring[i];
+		r = amdgpu_ring_test_helper(ring);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+static void gfx_v12_0_cp_enable(struct amdgpu_device *adev, bool enable)
+{
+	gfx_v12_0_cp_gfx_enable(adev, enable);
+	gfx_v12_0_cp_compute_enable(adev, enable);
+}
+
+static int gfx_v12_0_gfxhub_enable(struct amdgpu_device *adev)
+{
+	int r;
+	bool value;
+
+	r = adev->gfxhub.funcs->gart_enable(adev);
+	if (r)
+		return r;
+
+	amdgpu_device_flush_hdp(adev, NULL);
+
+	value = amdgpu_vm_fault_stop != AMDGPU_VM_FAULT_STOP_ALWAYS;
+
+	adev->gfxhub.funcs->set_fault_enable_default(adev, value);
+	/* TODO investigate why this and the hdp flush above is needed,
+	 * are we missing a flush somewhere else? */
+	adev->gmc.gmc_funcs->flush_gpu_tlb(adev, 0, AMDGPU_GFXHUB(0), 0);
+
+	return 0;
+}
+
+static int get_gb_addr_config(struct amdgpu_device *adev)
+{
+	u32 gb_addr_config;
+
+	gb_addr_config = RREG32_SOC15(GC, 0, regGB_ADDR_CONFIG);
+	if (gb_addr_config == 0)
+		return -EINVAL;
+
+	adev->gfx.config.gb_addr_config_fields.num_pkrs =
+		1 << REG_GET_FIELD(gb_addr_config, GB_ADDR_CONFIG, NUM_PKRS);
+
+	adev->gfx.config.gb_addr_config = gb_addr_config;
+
+	adev->gfx.config.gb_addr_config_fields.num_pipes = 1 <<
+			REG_GET_FIELD(adev->gfx.config.gb_addr_config,
+				      GB_ADDR_CONFIG, NUM_PIPES);
+
+	adev->gfx.config.max_tile_pipes =
+		adev->gfx.config.gb_addr_config_fields.num_pipes;
+
+	adev->gfx.config.gb_addr_config_fields.max_compress_frags = 1 <<
+			REG_GET_FIELD(adev->gfx.config.gb_addr_config,
+				      GB_ADDR_CONFIG, MAX_COMPRESSED_FRAGS);
+	adev->gfx.config.gb_addr_config_fields.num_rb_per_se = 1 <<
+			REG_GET_FIELD(adev->gfx.config.gb_addr_config,
+				      GB_ADDR_CONFIG, NUM_RB_PER_SE);
+	adev->gfx.config.gb_addr_config_fields.num_se = 1 <<
+			REG_GET_FIELD(adev->gfx.config.gb_addr_config,
+				      GB_ADDR_CONFIG, NUM_SHADER_ENGINES);
+	adev->gfx.config.gb_addr_config_fields.pipe_interleave_size = 1 << (8 +
+			REG_GET_FIELD(adev->gfx.config.gb_addr_config,
+				      GB_ADDR_CONFIG, PIPE_INTERLEAVE_SIZE));
+
+	return 0;
+}
+
+static void gfx_v12_0_disable_gpa_mode(struct amdgpu_device *adev)
+{
+	uint32_t data;
+
+	data = RREG32_SOC15(GC, 0, regCPC_PSP_DEBUG);
+	data |= CPC_PSP_DEBUG__GPA_OVERRIDE_MASK;
+	WREG32_SOC15(GC, 0, regCPC_PSP_DEBUG, data);
+
+	data = RREG32_SOC15(GC, 0, regCPG_PSP_DEBUG);
+	data |= CPG_PSP_DEBUG__GPA_OVERRIDE_MASK;
+	WREG32_SOC15(GC, 0, regCPG_PSP_DEBUG, data);
+}
+
+static void gfx_v12_0_init_golden_registers(struct amdgpu_device *adev)
+{
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 0, 1):
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_12_0,
+						(const u32)ARRAY_SIZE(golden_settings_gc_12_0));
+
+		if (adev->rev_id == 0)
+			soc15_program_register_sequence(adev,
+					golden_settings_gc_12_0_rev0,
+					(const u32)ARRAY_SIZE(golden_settings_gc_12_0_rev0));
+		break;
+	default:
+		break;
+	}
+}
+
+static int gfx_v12_0_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO) {
+		if (adev->gfx.imu.funcs && (amdgpu_dpm > 0)) {
+			/* RLC autoload sequence 1: Program rlc ram */
+			if (adev->gfx.imu.funcs->program_rlc_ram)
+				adev->gfx.imu.funcs->program_rlc_ram(adev);
+		}
+		/* rlc autoload firmware */
+		r = gfx_v12_0_rlc_backdoor_autoload_enable(adev);
+		if (r)
+			return r;
+	} else {
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+			if (adev->gfx.imu.funcs && (amdgpu_dpm > 0)) {
+				if (adev->gfx.imu.funcs->load_microcode)
+					adev->gfx.imu.funcs->load_microcode(adev);
+				if (adev->gfx.imu.funcs->setup_imu)
+					adev->gfx.imu.funcs->setup_imu(adev);
+				if (adev->gfx.imu.funcs->start_imu)
+					adev->gfx.imu.funcs->start_imu(adev);
+			}
+
+			/* disable gpa mode in backdoor loading */
+			gfx_v12_0_disable_gpa_mode(adev);
+		}
+	}
+
+	if ((adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO) ||
+	    (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP)) {
+		r = gfx_v12_0_wait_for_rlc_autoload_complete(adev);
+		if (r) {
+			dev_err(adev->dev, "(%d) failed to wait rlc autoload complete\n", r);
+			return r;
+		}
+	}
+
+	if (!amdgpu_emu_mode)
+		gfx_v12_0_init_golden_registers(adev);
+
+	adev->gfx.is_poweron = true;
+
+	if (get_gb_addr_config(adev))
+		DRM_WARN("Invalid gb_addr_config !\n");
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP)
+		gfx_v12_0_config_gfx_rs64(adev);
+
+	r = gfx_v12_0_gfxhub_enable(adev);
+	if (r)
+		return r;
+
+	if ((adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT ||
+	     adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO) &&
+	     (amdgpu_dpm == 1)) {
+		/**
+		 * For gfx 12, rlc firmware loading relies on smu firmware is
+		 * loaded firstly, so in direct type, it has to load smc ucode
+		 * here before rlc.
+		 */
+		r = amdgpu_pm_load_smu_firmware(adev, NULL);
+		if (r)
+			return r;
+	}
+
+	gfx_v12_0_constants_init(adev);
+
+	if (adev->nbio.funcs->gc_doorbell_init)
+		adev->nbio.funcs->gc_doorbell_init(adev);
+
+	r = gfx_v12_0_rlc_resume(adev);
+	if (r)
+		return r;
+
+	/*
+	 * init golden registers and rlc resume may override some registers,
+	 * reconfig them here
+	 */
+	gfx_v12_0_tcp_harvest(adev);
+
+	r = gfx_v12_0_cp_resume(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int gfx_v12_0_set_userq_eop_interrupts(struct amdgpu_device *adev,
+					      bool enable)
+{
+	unsigned int irq_type;
+	int m, p, r;
+
+	if (adev->userq_funcs[AMDGPU_HW_IP_GFX]) {
+		for (m = 0; m < adev->gfx.me.num_me; m++) {
+			for (p = 0; p < adev->gfx.me.num_pipe_per_me; p++) {
+				irq_type = AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP + p;
+				if (enable)
+					r = amdgpu_irq_get(adev, &adev->gfx.eop_irq,
+							   irq_type);
+				else
+					r = amdgpu_irq_put(adev, &adev->gfx.eop_irq,
+							   irq_type);
+				if (r)
+					return r;
+			}
+		}
+	}
+
+	if (adev->userq_funcs[AMDGPU_HW_IP_COMPUTE]) {
+		for (m = 0; m < adev->gfx.mec.num_mec; ++m) {
+			for (p = 0; p < adev->gfx.mec.num_pipe_per_mec; p++) {
+				irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP
+					+ (m * adev->gfx.mec.num_pipe_per_mec)
+					+ p;
+				if (enable)
+					r = amdgpu_irq_get(adev, &adev->gfx.eop_irq,
+							   irq_type);
+				else
+					r = amdgpu_irq_put(adev, &adev->gfx.eop_irq,
+							   irq_type);
+				if (r)
+					return r;
+			}
+		}
+	}
+
+	return 0;
+}
+
+static int gfx_v12_0_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	uint32_t tmp;
+
+	cancel_delayed_work_sync(&adev->gfx.idle_work);
+
+	amdgpu_irq_put(adev, &adev->gfx.priv_reg_irq, 0);
+	amdgpu_irq_put(adev, &adev->gfx.priv_inst_irq, 0);
+	amdgpu_irq_put(adev, &adev->gfx.bad_op_irq, 0);
+	gfx_v12_0_set_userq_eop_interrupts(adev, false);
+
+	if (!adev->no_hw_access) {
+		if (amdgpu_async_gfx_ring) {
+			if (amdgpu_gfx_disable_kgq(adev, 0))
+				DRM_ERROR("KGQ disable failed\n");
+		}
+
+		if (amdgpu_gfx_disable_kcq(adev, 0))
+			DRM_ERROR("KCQ disable failed\n");
+
+		amdgpu_mes_kiq_hw_fini(adev);
+	}
+
+	if (amdgpu_sriov_vf(adev)) {
+		gfx_v12_0_cp_gfx_enable(adev, false);
+		/* Program KIQ position of RLC_CP_SCHEDULERS during destroy */
+		tmp = RREG32_SOC15(GC, 0, regRLC_CP_SCHEDULERS);
+		tmp &= 0xffffff00;
+		WREG32_SOC15(GC, 0, regRLC_CP_SCHEDULERS, tmp);
+
+		return 0;
+	}
+	gfx_v12_0_cp_enable(adev, false);
+	gfx_v12_0_enable_gui_idle_interrupt(adev, false);
+
+	adev->gfxhub.funcs->gart_disable(adev);
+
+	adev->gfx.is_poweron = false;
+
+	return 0;
+}
+
+static int gfx_v12_0_suspend(struct amdgpu_ip_block *ip_block)
+{
+	return gfx_v12_0_hw_fini(ip_block);
+}
+
+static int gfx_v12_0_resume(struct amdgpu_ip_block *ip_block)
+{
+	return gfx_v12_0_hw_init(ip_block);
+}
+
+static bool gfx_v12_0_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (REG_GET_FIELD(RREG32_SOC15(GC, 0, regGRBM_STATUS),
+				GRBM_STATUS, GUI_ACTIVE))
+		return false;
+	else
+		return true;
+}
+
+static int gfx_v12_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		/* read MC_STATUS */
+		tmp = RREG32_SOC15(GC, 0, regGRBM_STATUS) &
+			GRBM_STATUS__GUI_ACTIVE_MASK;
+
+		if (!REG_GET_FIELD(tmp, GRBM_STATUS, GUI_ACTIVE))
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static uint64_t gfx_v12_0_get_gpu_clock_counter(struct amdgpu_device *adev)
+{
+	uint64_t clock = 0;
+
+	if (adev->smuio.funcs &&
+	    adev->smuio.funcs->get_gpu_clock_counter)
+		clock = adev->smuio.funcs->get_gpu_clock_counter(adev);
+	else
+		dev_warn(adev->dev, "query gpu clock counter is not supported\n");
+
+	return clock;
+}
+
+static int gfx_v12_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	switch (amdgpu_user_queue) {
+	case -1:
+	case 0:
+	default:
+		adev->gfx.disable_kq = false;
+		adev->gfx.disable_uq = true;
+		break;
+	case 1:
+		adev->gfx.disable_kq = false;
+		adev->gfx.disable_uq = false;
+		break;
+	case 2:
+		adev->gfx.disable_kq = true;
+		adev->gfx.disable_uq = false;
+		break;
+	}
+
+	adev->gfx.funcs = &gfx_v12_0_gfx_funcs;
+
+	if (adev->gfx.disable_kq) {
+		adev->gfx.num_gfx_rings = 0;
+		adev->gfx.num_compute_rings = 0;
+	} else {
+		adev->gfx.num_gfx_rings = GFX12_NUM_GFX_RINGS;
+		adev->gfx.num_compute_rings = min(amdgpu_gfx_get_num_kcq(adev),
+						  AMDGPU_MAX_COMPUTE_RINGS);
+	}
+
+	gfx_v12_0_set_kiq_pm4_funcs(adev);
+	gfx_v12_0_set_ring_funcs(adev);
+	gfx_v12_0_set_irq_funcs(adev);
+	gfx_v12_0_set_rlc_funcs(adev);
+	gfx_v12_0_set_mqd_funcs(adev);
+	gfx_v12_0_set_imu_funcs(adev);
+
+	gfx_v12_0_init_rlcg_reg_access_ctrl(adev);
+
+	return gfx_v12_0_init_microcode(adev);
+}
+
+static int gfx_v12_0_late_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = amdgpu_irq_get(adev, &adev->gfx.priv_reg_irq, 0);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_get(adev, &adev->gfx.priv_inst_irq, 0);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_get(adev, &adev->gfx.bad_op_irq, 0);
+	if (r)
+		return r;
+
+	r = gfx_v12_0_set_userq_eop_interrupts(adev, true);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static bool gfx_v12_0_is_rlc_enabled(struct amdgpu_device *adev)
+{
+	uint32_t rlc_cntl;
+
+	/* if RLC is not enabled, do nothing */
+	rlc_cntl = RREG32_SOC15(GC, 0, regRLC_CNTL);
+	return (REG_GET_FIELD(rlc_cntl, RLC_CNTL, RLC_ENABLE_F32)) ? true : false;
+}
+
+static void gfx_v12_0_set_safe_mode(struct amdgpu_device *adev,
+				    int xcc_id)
+{
+	uint32_t data;
+	unsigned i;
+
+	data = RLC_SAFE_MODE__CMD_MASK;
+	data |= (1 << RLC_SAFE_MODE__MESSAGE__SHIFT);
+
+	WREG32_SOC15(GC, 0, regRLC_SAFE_MODE, data);
+
+	/* wait for RLC_SAFE_MODE */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (!REG_GET_FIELD(RREG32_SOC15(GC, 0, regRLC_SAFE_MODE),
+				   RLC_SAFE_MODE, CMD))
+			break;
+		udelay(1);
+	}
+}
+
+static void gfx_v12_0_unset_safe_mode(struct amdgpu_device *adev,
+				      int xcc_id)
+{
+	WREG32_SOC15(GC, 0, regRLC_SAFE_MODE, RLC_SAFE_MODE__CMD_MASK);
+}
+
+static void gfx_v12_0_update_perf_clk(struct amdgpu_device *adev,
+				      bool enable)
+{
+	uint32_t def, data;
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_GFX_PERF_CLK))
+		return;
+
+	def = data = RREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE);
+
+	if (enable)
+		data &= ~RLC_CGTT_MGCG_OVERRIDE__PERFMON_CLOCK_STATE_MASK;
+	else
+		data |= RLC_CGTT_MGCG_OVERRIDE__PERFMON_CLOCK_STATE_MASK;
+
+	if (def != data)
+		WREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE, data);
+}
+
+static void gfx_v12_0_update_spm_vmid(struct amdgpu_device *adev,
+				      struct amdgpu_ring *ring,
+				      unsigned vmid)
+{
+	u32 reg, data;
+
+	reg = SOC15_REG_OFFSET(GC, 0, regRLC_SPM_MC_CNTL);
+	if (amdgpu_sriov_is_pp_one_vf(adev))
+		data = RREG32_NO_KIQ(reg);
+	else
+		data = RREG32(reg);
+
+	data &= ~RLC_SPM_MC_CNTL__RLC_SPM_VMID_MASK;
+	data |= (vmid & RLC_SPM_MC_CNTL__RLC_SPM_VMID_MASK) << RLC_SPM_MC_CNTL__RLC_SPM_VMID__SHIFT;
+
+	if (amdgpu_sriov_is_pp_one_vf(adev))
+		WREG32_SOC15_NO_KIQ(GC, 0, regRLC_SPM_MC_CNTL, data);
+	else
+		WREG32_SOC15(GC, 0, regRLC_SPM_MC_CNTL, data);
+
+	if (ring
+	    && amdgpu_sriov_is_pp_one_vf(adev)
+	    && ((ring->funcs->type == AMDGPU_RING_TYPE_GFX)
+		|| (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE))) {
+		uint32_t reg = SOC15_REG_OFFSET(GC, 0, regRLC_SPM_MC_CNTL);
+		amdgpu_ring_emit_wreg(ring, reg, data);
+	}
+}
+
+static const struct amdgpu_rlc_funcs gfx_v12_0_rlc_funcs = {
+	.is_rlc_enabled = gfx_v12_0_is_rlc_enabled,
+	.set_safe_mode = gfx_v12_0_set_safe_mode,
+	.unset_safe_mode = gfx_v12_0_unset_safe_mode,
+	.init = gfx_v12_0_rlc_init,
+	.get_csb_size = gfx_v12_0_get_csb_size,
+	.get_csb_buffer = gfx_v12_0_get_csb_buffer,
+	.resume = gfx_v12_0_rlc_resume,
+	.stop = gfx_v12_0_rlc_stop,
+	.reset = gfx_v12_0_rlc_reset,
+	.start = gfx_v12_0_rlc_start,
+	.update_spm_vmid = gfx_v12_0_update_spm_vmid,
+};
+
+#if 0
+static void gfx_v12_cntl_power_gating(struct amdgpu_device *adev, bool enable)
+{
+	/* TODO */
+}
+
+static void gfx_v12_cntl_pg(struct amdgpu_device *adev, bool enable)
+{
+	/* TODO */
+}
+#endif
+
+static int gfx_v12_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					   enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = (state == AMD_PG_STATE_GATE);
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 0, 1):
+		amdgpu_gfx_off_ctrl(adev, enable);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static void gfx_v12_0_update_coarse_grain_clock_gating(struct amdgpu_device *adev,
+						       bool enable)
+{
+	uint32_t def, data;
+
+	if (!(adev->cg_flags &
+	      (AMD_CG_SUPPORT_GFX_CGCG |
+	      AMD_CG_SUPPORT_GFX_CGLS |
+	      AMD_CG_SUPPORT_GFX_3D_CGCG |
+	      AMD_CG_SUPPORT_GFX_3D_CGLS)))
+		return;
+
+	if (enable) {
+		def = data = RREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE);
+
+		/* unset CGCG override */
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGCG)
+			data &= ~RLC_CGTT_MGCG_OVERRIDE__GFXIP_CGCG_OVERRIDE_MASK;
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGLS)
+			data &= ~RLC_CGTT_MGCG_OVERRIDE__GFXIP_CGLS_OVERRIDE_MASK;
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGCG ||
+		    adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGLS)
+			data &= ~RLC_CGTT_MGCG_OVERRIDE__GFXIP_GFX3D_CG_OVERRIDE_MASK;
+
+		/* update CGCG override bits */
+		if (def != data)
+			WREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE, data);
+
+		/* enable cgcg FSM(0x0000363F) */
+		def = data = RREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL);
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGCG) {
+			data &= ~RLC_CGCG_CGLS_CTRL__CGCG_GFX_IDLE_THRESHOLD_MASK;
+			data |= (0x36 << RLC_CGCG_CGLS_CTRL__CGCG_GFX_IDLE_THRESHOLD__SHIFT) |
+				 RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK;
+		}
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGLS) {
+			data &= ~RLC_CGCG_CGLS_CTRL__CGLS_REP_COMPANSAT_DELAY_MASK;
+			data |= (0x000F << RLC_CGCG_CGLS_CTRL__CGLS_REP_COMPANSAT_DELAY__SHIFT) |
+				 RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK;
+		}
+
+		if (def != data)
+			WREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL, data);
+
+		/* Program RLC_CGCG_CGLS_CTRL_3D */
+		def = data = RREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL_3D);
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGCG) {
+			data &= ~RLC_CGCG_CGLS_CTRL_3D__CGCG_GFX_IDLE_THRESHOLD_MASK;
+			data |= (0x36 << RLC_CGCG_CGLS_CTRL_3D__CGCG_GFX_IDLE_THRESHOLD__SHIFT) |
+				 RLC_CGCG_CGLS_CTRL_3D__CGCG_EN_MASK;
+		}
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGLS) {
+			data &= ~RLC_CGCG_CGLS_CTRL_3D__CGLS_REP_COMPANSAT_DELAY_MASK;
+			data |= (0xf << RLC_CGCG_CGLS_CTRL_3D__CGLS_REP_COMPANSAT_DELAY__SHIFT) |
+				 RLC_CGCG_CGLS_CTRL_3D__CGLS_EN_MASK;
+		}
+
+		if (def != data)
+			WREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL_3D, data);
+
+		/* set IDLE_POLL_COUNT(0x00900100) */
+		def = data = RREG32_SOC15(GC, 0, regCP_RB_WPTR_POLL_CNTL);
+
+		data &= ~(CP_RB_WPTR_POLL_CNTL__POLL_FREQUENCY_MASK | CP_RB_WPTR_POLL_CNTL__IDLE_POLL_COUNT_MASK);
+		data |= (0x0100 << CP_RB_WPTR_POLL_CNTL__POLL_FREQUENCY__SHIFT) |
+			(0x0090 << CP_RB_WPTR_POLL_CNTL__IDLE_POLL_COUNT__SHIFT);
+
+		if (def != data)
+			WREG32_SOC15(GC, 0, regCP_RB_WPTR_POLL_CNTL, data);
+
+		data = RREG32_SOC15(GC, 0, regCP_INT_CNTL);
+		data = REG_SET_FIELD(data, CP_INT_CNTL, CNTX_BUSY_INT_ENABLE, 1);
+		data = REG_SET_FIELD(data, CP_INT_CNTL, CNTX_EMPTY_INT_ENABLE, 1);
+		data = REG_SET_FIELD(data, CP_INT_CNTL, CMP_BUSY_INT_ENABLE, 1);
+		data = REG_SET_FIELD(data, CP_INT_CNTL, GFX_IDLE_INT_ENABLE, 1);
+		WREG32_SOC15(GC, 0, regCP_INT_CNTL, data);
+
+		data = RREG32_SOC15(GC, 0, regSDMA0_RLC_CGCG_CTRL);
+		data = REG_SET_FIELD(data, SDMA0_RLC_CGCG_CTRL, CGCG_INT_ENABLE, 1);
+		WREG32_SOC15(GC, 0, regSDMA0_RLC_CGCG_CTRL, data);
+
+		/* Some ASICs only have one SDMA instance, not need to configure SDMA1 */
+		if (adev->sdma.num_instances > 1) {
+			data = RREG32_SOC15(GC, 0, regSDMA1_RLC_CGCG_CTRL);
+			data = REG_SET_FIELD(data, SDMA1_RLC_CGCG_CTRL, CGCG_INT_ENABLE, 1);
+			WREG32_SOC15(GC, 0, regSDMA1_RLC_CGCG_CTRL, data);
+		}
+	} else {
+		/* Program RLC_CGCG_CGLS_CTRL */
+		def = data = RREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL);
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGCG)
+			data &= ~RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK;
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGLS)
+			data &= ~RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK;
+
+		if (def != data)
+			WREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL, data);
+
+		/* Program RLC_CGCG_CGLS_CTRL_3D */
+		def = data = RREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL_3D);
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGCG)
+			data &= ~RLC_CGCG_CGLS_CTRL_3D__CGCG_EN_MASK;
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGLS)
+			data &= ~RLC_CGCG_CGLS_CTRL_3D__CGLS_EN_MASK;
+
+		if (def != data)
+			WREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL_3D, data);
+	}
+}
+
+static void gfx_v12_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						       bool enable)
+{
+	uint32_t data, def;
+	if (!(adev->cg_flags & (AMD_CG_SUPPORT_GFX_MGCG | AMD_CG_SUPPORT_GFX_MGLS)))
+		return;
+
+	/* It is disabled by HW by default */
+	if (enable) {
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGCG) {
+			/* 1 - RLC_CGTT_MGCG_OVERRIDE */
+			def = data = RREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE);
+
+			data &= ~(RLC_CGTT_MGCG_OVERRIDE__GRBM_CGTT_SCLK_OVERRIDE_MASK |
+				  RLC_CGTT_MGCG_OVERRIDE__RLC_CGTT_SCLK_OVERRIDE_MASK |
+				  RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGCG_OVERRIDE_MASK);
+
+			if (def != data)
+				WREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE, data);
+		}
+	} else {
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGCG) {
+			def = data = RREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE);
+
+			data |= (RLC_CGTT_MGCG_OVERRIDE__RLC_CGTT_SCLK_OVERRIDE_MASK |
+				 RLC_CGTT_MGCG_OVERRIDE__GRBM_CGTT_SCLK_OVERRIDE_MASK |
+				 RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGCG_OVERRIDE_MASK);
+
+			if (def != data)
+				WREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE, data);
+		}
+	}
+}
+
+static void gfx_v12_0_update_repeater_fgcg(struct amdgpu_device *adev,
+					   bool enable)
+{
+	uint32_t def, data;
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_REPEATER_FGCG))
+		return;
+
+	def = data = RREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE);
+
+	if (enable)
+		data &= ~(RLC_CGTT_MGCG_OVERRIDE__GFXIP_REPEATER_FGCG_OVERRIDE_MASK |
+				  RLC_CGTT_MGCG_OVERRIDE__RLC_REPEATER_FGCG_OVERRIDE_MASK);
+	else
+		data |= RLC_CGTT_MGCG_OVERRIDE__GFXIP_REPEATER_FGCG_OVERRIDE_MASK |
+				RLC_CGTT_MGCG_OVERRIDE__RLC_REPEATER_FGCG_OVERRIDE_MASK;
+
+	if (def != data)
+		WREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE, data);
+}
+
+static void gfx_v12_0_update_sram_fgcg(struct amdgpu_device *adev,
+				       bool enable)
+{
+	uint32_t def, data;
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_GFX_FGCG))
+		return;
+
+	def = data = RREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE);
+
+	if (enable)
+		data &= ~RLC_CGTT_MGCG_OVERRIDE__GFXIP_FGCG_OVERRIDE_MASK;
+	else
+		data |= RLC_CGTT_MGCG_OVERRIDE__GFXIP_FGCG_OVERRIDE_MASK;
+
+	if (def != data)
+		WREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE, data);
+}
+
+static int gfx_v12_0_update_gfx_clock_gating(struct amdgpu_device *adev,
+					    bool enable)
+{
+	amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
+
+	gfx_v12_0_update_coarse_grain_clock_gating(adev, enable);
+
+	gfx_v12_0_update_medium_grain_clock_gating(adev, enable);
+
+	gfx_v12_0_update_repeater_fgcg(adev, enable);
+
+	gfx_v12_0_update_sram_fgcg(adev, enable);
+
+	gfx_v12_0_update_perf_clk(adev, enable);
+
+	if (adev->cg_flags &
+	    (AMD_CG_SUPPORT_GFX_MGCG |
+	     AMD_CG_SUPPORT_GFX_CGLS |
+	     AMD_CG_SUPPORT_GFX_CGCG |
+	     AMD_CG_SUPPORT_GFX_3D_CGCG |
+	     AMD_CG_SUPPORT_GFX_3D_CGLS))
+		gfx_v12_0_enable_gui_idle_interrupt(adev, enable);
+
+	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
+
+	return 0;
+}
+
+static int gfx_v12_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					   enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 0, 1):
+		gfx_v12_0_update_gfx_clock_gating(adev,
+						  state == AMD_CG_STATE_GATE);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static void gfx_v12_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int data;
+
+	/* AMD_CG_SUPPORT_GFX_MGCG */
+	data = RREG32_SOC15(GC, 0, regRLC_CGTT_MGCG_OVERRIDE);
+	if (!(data & RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGCG_OVERRIDE_MASK))
+		*flags |= AMD_CG_SUPPORT_GFX_MGCG;
+
+	/* AMD_CG_SUPPORT_REPEATER_FGCG */
+	if (!(data & RLC_CGTT_MGCG_OVERRIDE__GFXIP_REPEATER_FGCG_OVERRIDE_MASK))
+		*flags |= AMD_CG_SUPPORT_REPEATER_FGCG;
+
+	/* AMD_CG_SUPPORT_GFX_FGCG */
+	if (!(data & RLC_CGTT_MGCG_OVERRIDE__GFXIP_FGCG_OVERRIDE_MASK))
+		*flags |= AMD_CG_SUPPORT_GFX_FGCG;
+
+	/* AMD_CG_SUPPORT_GFX_PERF_CLK */
+	if (!(data & RLC_CGTT_MGCG_OVERRIDE__PERFMON_CLOCK_STATE_MASK))
+		*flags |= AMD_CG_SUPPORT_GFX_PERF_CLK;
+
+	/* AMD_CG_SUPPORT_GFX_CGCG */
+	data = RREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL);
+	if (data & RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_CGCG;
+
+	/* AMD_CG_SUPPORT_GFX_CGLS */
+	if (data & RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_CGLS;
+
+	/* AMD_CG_SUPPORT_GFX_3D_CGCG */
+	data = RREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL_3D);
+	if (data & RLC_CGCG_CGLS_CTRL_3D__CGCG_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_3D_CGCG;
+
+	/* AMD_CG_SUPPORT_GFX_3D_CGLS */
+	if (data & RLC_CGCG_CGLS_CTRL_3D__CGLS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_3D_CGLS;
+}
+
+static u64 gfx_v12_0_ring_get_rptr_gfx(struct amdgpu_ring *ring)
+{
+	/* gfx12 is 32bit rptr*/
+	return *(uint32_t *)ring->rptr_cpu_addr;
+}
+
+static u64 gfx_v12_0_ring_get_wptr_gfx(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u64 wptr;
+
+	/* XXX check if swapping is necessary on BE */
+	if (ring->use_doorbell) {
+		wptr = atomic64_read((atomic64_t *)ring->wptr_cpu_addr);
+	} else {
+		wptr = RREG32_SOC15(GC, 0, regCP_RB0_WPTR);
+		wptr += (u64)RREG32_SOC15(GC, 0, regCP_RB0_WPTR_HI) << 32;
+	}
+
+	return wptr;
+}
+
+static void gfx_v12_0_ring_set_wptr_gfx(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr,
+			     ring->wptr);
+		WDOORBELL64(ring->doorbell_index, ring->wptr);
+	} else {
+		WREG32_SOC15(GC, 0, regCP_RB0_WPTR,
+			     lower_32_bits(ring->wptr));
+		WREG32_SOC15(GC, 0, regCP_RB0_WPTR_HI,
+			     upper_32_bits(ring->wptr));
+	}
+}
+
+static u64 gfx_v12_0_ring_get_rptr_compute(struct amdgpu_ring *ring)
+{
+	/* gfx12 hardware is 32bit rptr */
+	return *(uint32_t *)ring->rptr_cpu_addr;
+}
+
+static u64 gfx_v12_0_ring_get_wptr_compute(struct amdgpu_ring *ring)
+{
+	u64 wptr;
+
+	/* XXX check if swapping is necessary on BE */
+	if (ring->use_doorbell)
+		wptr = atomic64_read((atomic64_t *)ring->wptr_cpu_addr);
+	else
+		BUG();
+	return wptr;
+}
+
+static void gfx_v12_0_ring_set_wptr_compute(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	/* XXX check if swapping is necessary on BE */
+	if (ring->use_doorbell) {
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr,
+			     ring->wptr);
+		WDOORBELL64(ring->doorbell_index, ring->wptr);
+	} else {
+		BUG(); /* only DOORBELL method supported on gfx12 now */
+	}
+}
+
+static void gfx_v12_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 ref_and_mask, reg_mem_engine;
+	const struct nbio_hdp_flush_reg *nbio_hf_reg = adev->nbio.hdp_flush_reg;
+
+	if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE) {
+		switch (ring->me) {
+		case 1:
+			ref_and_mask = nbio_hf_reg->ref_and_mask_cp2 << ring->pipe;
+			break;
+		case 2:
+			ref_and_mask = nbio_hf_reg->ref_and_mask_cp6 << ring->pipe;
+			break;
+		default:
+			return;
+		}
+		reg_mem_engine = 0;
+	} else {
+		ref_and_mask = nbio_hf_reg->ref_and_mask_cp0;
+		reg_mem_engine = 1; /* pfp */
+	}
+
+	gfx_v12_0_wait_reg_mem(ring, reg_mem_engine, 0, 1,
+			       adev->nbio.funcs->get_hdp_flush_req_offset(adev),
+			       adev->nbio.funcs->get_hdp_flush_done_offset(adev),
+			       ref_and_mask, ref_and_mask, 0x20);
+}
+
+static void gfx_v12_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
+				       struct amdgpu_job *job,
+				       struct amdgpu_ib *ib,
+				       uint32_t flags)
+{
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+	u32 header, control = 0;
+
+	BUG_ON(ib->flags & AMDGPU_IB_FLAG_CE);
+
+	header = PACKET3(PACKET3_INDIRECT_BUFFER, 2);
+
+	control |= ib->length_dw | (vmid << 24);
+
+	amdgpu_ring_write(ring, header);
+	BUG_ON(ib->gpu_addr & 0x3); /* Dword align */
+	amdgpu_ring_write(ring,
+#ifdef __BIG_ENDIAN
+		(2 << 0) |
+#endif
+		lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, control);
+}
+
+static void gfx_v12_0_ring_emit_ib_compute(struct amdgpu_ring *ring,
+					   struct amdgpu_job *job,
+					   struct amdgpu_ib *ib,
+					   uint32_t flags)
+{
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+	u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vmid << 24);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2));
+	BUG_ON(ib->gpu_addr & 0x3); /* Dword align */
+	amdgpu_ring_write(ring,
+#ifdef __BIG_ENDIAN
+				(2 << 0) |
+#endif
+				lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, control);
+}
+
+static void gfx_v12_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
+				     u64 seq, unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	bool int_sel = flags & AMDGPU_FENCE_FLAG_INT;
+
+	/* RELEASE_MEM - flush caches, send int */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_RELEASE_MEM, 6));
+	amdgpu_ring_write(ring, (PACKET3_RELEASE_MEM_GCR_SEQ |
+				 PACKET3_RELEASE_MEM_GCR_GL2_WB |
+				 PACKET3_RELEASE_MEM_CACHE_POLICY(3) |
+				 PACKET3_RELEASE_MEM_EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) |
+				 PACKET3_RELEASE_MEM_EVENT_INDEX(5)));
+	amdgpu_ring_write(ring, (PACKET3_RELEASE_MEM_DATA_SEL(write64bit ? 2 : 1) |
+				 PACKET3_RELEASE_MEM_INT_SEL(int_sel ? 2 : 0)));
+
+	/*
+	 * the address should be Qword aligned if 64bit write, Dword
+	 * aligned if only send 32bit data low (discard data high)
+	 */
+	if (write64bit)
+		BUG_ON(addr & 0x7);
+	else
+		BUG_ON(addr & 0x3);
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+	amdgpu_ring_write(ring, upper_32_bits(seq));
+	amdgpu_ring_write(ring, 0);
+}
+
+static void gfx_v12_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	gfx_v12_0_wait_reg_mem(ring, usepfp, 1, 0, lower_32_bits(addr),
+			       upper_32_bits(addr), seq, 0xffffffff, 4);
+}
+
+static void gfx_v12_0_ring_invalidate_tlbs(struct amdgpu_ring *ring,
+				   uint16_t pasid, uint32_t flush_type,
+				   bool all_hub, uint8_t dst_sel)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_INVALIDATE_TLBS, 0));
+	amdgpu_ring_write(ring,
+			  PACKET3_INVALIDATE_TLBS_DST_SEL(dst_sel) |
+			  PACKET3_INVALIDATE_TLBS_ALL_HUB(all_hub) |
+			  PACKET3_INVALIDATE_TLBS_PASID(pasid) |
+			  PACKET3_INVALIDATE_TLBS_FLUSH_TYPE(flush_type));
+}
+
+static void gfx_v12_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					 unsigned vmid, uint64_t pd_addr)
+{
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* compute doesn't have PFP */
+	if (ring->funcs->type == AMDGPU_RING_TYPE_GFX) {
+		/* sync PFP to ME, otherwise we might get invalid PFP reads */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_PFP_SYNC_ME, 0));
+		amdgpu_ring_write(ring, 0x0);
+	}
+}
+
+static void gfx_v12_0_ring_emit_fence_kiq(struct amdgpu_ring *ring, u64 addr,
+					  u64 seq, unsigned int flags)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	/* we only allocate 32bit for each seq wb address */
+	BUG_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	/* write fence seq to the "addr" */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				 WRITE_DATA_DST_SEL(5) | WR_CONFIRM));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+
+	if (flags & AMDGPU_FENCE_FLAG_INT) {
+		/* set register to trigger INT */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+		amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+					 WRITE_DATA_DST_SEL(0) | WR_CONFIRM));
+		amdgpu_ring_write(ring, SOC15_REG_OFFSET(GC, 0, regCPC_INT_STATUS));
+		amdgpu_ring_write(ring, 0);
+		amdgpu_ring_write(ring, 0x20000000); /* src_id is 178 */
+	}
+}
+
+static void gfx_v12_0_ring_emit_cntxcntl(struct amdgpu_ring *ring,
+					 uint32_t flags)
+{
+	uint32_t dw2 = 0;
+
+	dw2 |= 0x80000000; /* set load_enable otherwise this package is just NOPs */
+	if (flags & AMDGPU_HAVE_CTX_SWITCH) {
+		/* set load_global_config & load_global_uconfig */
+		dw2 |= 0x8001;
+		/* set load_cs_sh_regs */
+		dw2 |= 0x01000000;
+		/* set load_per_context_state & load_gfx_sh_regs for GFX */
+		dw2 |= 0x10002;
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	amdgpu_ring_write(ring, dw2);
+	amdgpu_ring_write(ring, 0);
+}
+
+static unsigned gfx_v12_0_ring_emit_init_cond_exec(struct amdgpu_ring *ring,
+						   uint64_t addr)
+{
+	unsigned ret;
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_COND_EXEC, 3));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	/* discard following DWs if *cond_exec_gpu_addr==0 */
+	amdgpu_ring_write(ring, 0);
+	ret = ring->wptr & ring->buf_mask;
+	/* patch dummy value later */
+	amdgpu_ring_write(ring, 0);
+
+	return ret;
+}
+
+static int gfx_v12_0_ring_preempt_ib(struct amdgpu_ring *ring)
+{
+	int i, r = 0;
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[0];
+	struct amdgpu_ring *kiq_ring = &kiq->ring;
+	unsigned long flags;
+
+	if (adev->enable_mes)
+		return -EINVAL;
+
+	if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
+		return -EINVAL;
+
+	spin_lock_irqsave(&kiq->ring_lock, flags);
+
+	if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->unmap_queues_size)) {
+		spin_unlock_irqrestore(&kiq->ring_lock, flags);
+		return -ENOMEM;
+	}
+
+	/* assert preemption condition */
+	amdgpu_ring_set_preempt_cond_exec(ring, false);
+
+	/* assert IB preemption, emit the trailing fence */
+	kiq->pmf->kiq_unmap_queues(kiq_ring, ring, PREEMPT_QUEUES_NO_UNMAP,
+				   ring->trail_fence_gpu_addr,
+				   ++ring->trail_seq);
+	amdgpu_ring_commit(kiq_ring);
+
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+
+	/* poll the trailing fence */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (ring->trail_seq ==
+		    le32_to_cpu(*(ring->trail_fence_cpu_addr)))
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout) {
+		r = -EINVAL;
+		DRM_ERROR("ring %d failed to preempt ib\n", ring->idx);
+	}
+
+	/* deassert preemption condition */
+	amdgpu_ring_set_preempt_cond_exec(ring, true);
+	return r;
+}
+
+static void gfx_v12_0_ring_emit_frame_cntl(struct amdgpu_ring *ring,
+					   bool start,
+					   bool secure)
+{
+	uint32_t v = secure ? FRAME_TMZ : 0;
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_FRAME_CONTROL, 0));
+	amdgpu_ring_write(ring, v | FRAME_CMD(start ? 0 : 1));
+}
+
+static void gfx_v12_0_ring_emit_rreg(struct amdgpu_ring *ring, uint32_t reg,
+				     uint32_t reg_val_offs)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_COPY_DATA, 4));
+	amdgpu_ring_write(ring, 0 |	/* src: register*/
+				(5 << 8) |	/* dst: memory */
+				(1 << 20));	/* write confirm */
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, lower_32_bits(adev->wb.gpu_addr +
+				reg_val_offs * 4));
+	amdgpu_ring_write(ring, upper_32_bits(adev->wb.gpu_addr +
+				reg_val_offs * 4));
+}
+
+static void gfx_v12_0_ring_emit_wreg(struct amdgpu_ring *ring,
+				     uint32_t reg,
+				     uint32_t val)
+{
+	uint32_t cmd = 0;
+
+	switch (ring->funcs->type) {
+	case AMDGPU_RING_TYPE_GFX:
+		cmd = WRITE_DATA_ENGINE_SEL(1) | WR_CONFIRM;
+		break;
+	case AMDGPU_RING_TYPE_KIQ:
+		cmd = (1 << 16); /* no inc addr */
+		break;
+	default:
+		cmd = WR_CONFIRM;
+		break;
+	}
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, cmd);
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, val);
+}
+
+static void gfx_v12_0_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+					uint32_t val, uint32_t mask)
+{
+	gfx_v12_0_wait_reg_mem(ring, 0, 0, 0, reg, 0, val, mask, 0x20);
+}
+
+static void gfx_v12_0_ring_emit_reg_write_reg_wait(struct amdgpu_ring *ring,
+						   uint32_t reg0, uint32_t reg1,
+						   uint32_t ref, uint32_t mask)
+{
+	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
+
+	gfx_v12_0_wait_reg_mem(ring, usepfp, 0, 1, reg0, reg1,
+			       ref, mask, 0x20);
+}
+
+static void
+gfx_v12_0_set_gfx_eop_interrupt_state(struct amdgpu_device *adev,
+				      uint32_t me, uint32_t pipe,
+				      enum amdgpu_interrupt_state state)
+{
+	uint32_t cp_int_cntl, cp_int_cntl_reg;
+
+	if (!me) {
+		switch (pipe) {
+		case 0:
+			cp_int_cntl_reg = SOC15_REG_OFFSET(GC, 0, regCP_INT_CNTL_RING0);
+			break;
+		default:
+			DRM_DEBUG("invalid pipe %d\n", pipe);
+			return;
+		}
+	} else {
+		DRM_DEBUG("invalid me %d\n", me);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+		cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+					    TIME_STAMP_INT_ENABLE, 0);
+		cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+					    GENERIC0_INT_ENABLE, 0);
+		WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+		cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+					    TIME_STAMP_INT_ENABLE, 1);
+		cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+					    GENERIC0_INT_ENABLE, 1);
+		WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+		break;
+	default:
+		break;
+	}
+}
+
+static void gfx_v12_0_set_compute_eop_interrupt_state(struct amdgpu_device *adev,
+						     int me, int pipe,
+						     enum amdgpu_interrupt_state state)
+{
+	u32 mec_int_cntl, mec_int_cntl_reg;
+
+	/*
+	 * amdgpu controls only the first MEC. That's why this function only
+	 * handles the setting of interrupts for this specific MEC. All other
+	 * pipes' interrupts are set by amdkfd.
+	 */
+
+	if (me == 1) {
+		switch (pipe) {
+		case 0:
+			mec_int_cntl_reg = SOC15_REG_OFFSET(GC, 0, regCP_ME1_PIPE0_INT_CNTL);
+			break;
+		case 1:
+			mec_int_cntl_reg = SOC15_REG_OFFSET(GC, 0, regCP_ME1_PIPE1_INT_CNTL);
+			break;
+		default:
+			DRM_DEBUG("invalid pipe %d\n", pipe);
+			return;
+		}
+	} else {
+		DRM_DEBUG("invalid me %d\n", me);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		mec_int_cntl = RREG32_SOC15_IP(GC, mec_int_cntl_reg);
+		mec_int_cntl = REG_SET_FIELD(mec_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+					     TIME_STAMP_INT_ENABLE, 0);
+		mec_int_cntl = REG_SET_FIELD(mec_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+					     GENERIC0_INT_ENABLE, 0);
+		WREG32_SOC15_IP(GC, mec_int_cntl_reg, mec_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		mec_int_cntl = RREG32_SOC15_IP(GC, mec_int_cntl_reg);
+		mec_int_cntl = REG_SET_FIELD(mec_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+					     TIME_STAMP_INT_ENABLE, 1);
+		mec_int_cntl = REG_SET_FIELD(mec_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+					     GENERIC0_INT_ENABLE, 1);
+		WREG32_SOC15_IP(GC, mec_int_cntl_reg, mec_int_cntl);
+		break;
+	default:
+		break;
+	}
+}
+
+static int gfx_v12_0_set_eop_interrupt_state(struct amdgpu_device *adev,
+					    struct amdgpu_irq_src *src,
+					    unsigned type,
+					    enum amdgpu_interrupt_state state)
+{
+	switch (type) {
+	case AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP:
+		gfx_v12_0_set_gfx_eop_interrupt_state(adev, 0, 0, state);
+		break;
+	case AMDGPU_CP_IRQ_GFX_ME0_PIPE1_EOP:
+		gfx_v12_0_set_gfx_eop_interrupt_state(adev, 0, 1, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP:
+		gfx_v12_0_set_compute_eop_interrupt_state(adev, 1, 0, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE1_EOP:
+		gfx_v12_0_set_compute_eop_interrupt_state(adev, 1, 1, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE2_EOP:
+		gfx_v12_0_set_compute_eop_interrupt_state(adev, 1, 2, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE3_EOP:
+		gfx_v12_0_set_compute_eop_interrupt_state(adev, 1, 3, state);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int gfx_v12_0_eop_irq(struct amdgpu_device *adev,
+			     struct amdgpu_irq_src *source,
+			     struct amdgpu_iv_entry *entry)
+{
+	u32 doorbell_offset = entry->src_data[0];
+	u8 me_id, pipe_id, queue_id;
+	struct amdgpu_ring *ring;
+	int i;
+
+	DRM_DEBUG("IH: CP EOP\n");
+
+	if (adev->enable_mes && doorbell_offset) {
+		struct amdgpu_userq_fence_driver *fence_drv = NULL;
+		struct xarray *xa = &adev->userq_xa;
+		unsigned long flags;
+
+		xa_lock_irqsave(xa, flags);
+		fence_drv = xa_load(xa, doorbell_offset);
+		if (fence_drv)
+			amdgpu_userq_fence_driver_process(fence_drv);
+		xa_unlock_irqrestore(xa, flags);
+	} else {
+		me_id = (entry->ring_id & 0x0c) >> 2;
+		pipe_id = (entry->ring_id & 0x03) >> 0;
+		queue_id = (entry->ring_id & 0x70) >> 4;
+
+		switch (me_id) {
+		case 0:
+			if (pipe_id == 0)
+				amdgpu_fence_process(&adev->gfx.gfx_ring[0]);
+			else
+				amdgpu_fence_process(&adev->gfx.gfx_ring[1]);
+			break;
+		case 1:
+		case 2:
+			for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+				ring = &adev->gfx.compute_ring[i];
+				/* Per-queue interrupt is supported for MEC starting from VI.
+				 * The interrupt can only be enabled/disabled per pipe instead
+				 * of per queue.
+				 */
+				if ((ring->me == me_id) &&
+				    (ring->pipe == pipe_id) &&
+				    (ring->queue == queue_id))
+					amdgpu_fence_process(ring);
+			}
+			break;
+		}
+	}
+
+	return 0;
+}
+
+static int gfx_v12_0_set_priv_reg_fault_state(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      unsigned int type,
+					      enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl_reg, cp_int_cntl;
+	int i, j;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+	case AMDGPU_IRQ_STATE_ENABLE:
+		for (i = 0; i < adev->gfx.me.num_me; i++) {
+			for (j = 0; j < adev->gfx.me.num_pipe_per_me; j++) {
+				cp_int_cntl_reg = gfx_v12_0_get_cpg_int_cntl(adev, i, j);
+
+				if (cp_int_cntl_reg) {
+					cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+					cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+								    PRIV_REG_INT_ENABLE,
+								    state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+					WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+				}
+			}
+		}
+		for (i = 0; i < adev->gfx.mec.num_mec; i++) {
+			for (j = 0; j < adev->gfx.mec.num_pipe_per_mec; j++) {
+				/* MECs start at 1 */
+				cp_int_cntl_reg = gfx_v12_0_get_cpc_int_cntl(adev, i + 1, j);
+
+				if (cp_int_cntl_reg) {
+					cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+					cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+								    PRIV_REG_INT_ENABLE,
+								    state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+					WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+				}
+			}
+		}
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gfx_v12_0_set_bad_op_fault_state(struct amdgpu_device *adev,
+					    struct amdgpu_irq_src *source,
+					    unsigned type,
+					    enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl_reg, cp_int_cntl;
+	int i, j;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+	case AMDGPU_IRQ_STATE_ENABLE:
+		for (i = 0; i < adev->gfx.me.num_me; i++) {
+			for (j = 0; j < adev->gfx.me.num_pipe_per_me; j++) {
+				cp_int_cntl_reg = gfx_v12_0_get_cpg_int_cntl(adev, i, j);
+
+				if (cp_int_cntl_reg) {
+					cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+					cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+								    OPCODE_ERROR_INT_ENABLE,
+								    state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+					WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+				}
+			}
+		}
+		for (i = 0; i < adev->gfx.mec.num_mec; i++) {
+			for (j = 0; j < adev->gfx.mec.num_pipe_per_mec; j++) {
+				/* MECs start at 1 */
+				cp_int_cntl_reg = gfx_v12_0_get_cpc_int_cntl(adev, i + 1, j);
+
+				if (cp_int_cntl_reg) {
+					cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+					cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+								    OPCODE_ERROR_INT_ENABLE,
+								    state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+					WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+				}
+			}
+		}
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int gfx_v12_0_set_priv_inst_fault_state(struct amdgpu_device *adev,
+					       struct amdgpu_irq_src *source,
+					       unsigned int type,
+					       enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl_reg, cp_int_cntl;
+	int i, j;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+	case AMDGPU_IRQ_STATE_ENABLE:
+		for (i = 0; i < adev->gfx.me.num_me; i++) {
+			for (j = 0; j < adev->gfx.me.num_pipe_per_me; j++) {
+				cp_int_cntl_reg = gfx_v12_0_get_cpg_int_cntl(adev, i, j);
+
+				if (cp_int_cntl_reg) {
+					cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+					cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+								    PRIV_INSTR_INT_ENABLE,
+								    state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+					WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+				}
+			}
+		}
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static void gfx_v12_0_handle_priv_fault(struct amdgpu_device *adev,
+					struct amdgpu_iv_entry *entry)
+{
+	u8 me_id, pipe_id, queue_id;
+	struct amdgpu_ring *ring;
+	int i;
+
+	me_id = (entry->ring_id & 0x0c) >> 2;
+	pipe_id = (entry->ring_id & 0x03) >> 0;
+	queue_id = (entry->ring_id & 0x70) >> 4;
+
+	if (!adev->gfx.disable_kq) {
+		switch (me_id) {
+		case 0:
+			for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+				ring = &adev->gfx.gfx_ring[i];
+				if (ring->me == me_id && ring->pipe == pipe_id &&
+				    ring->queue == queue_id)
+					drm_sched_fault(&ring->sched);
+			}
+			break;
+		case 1:
+		case 2:
+			for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+				ring = &adev->gfx.compute_ring[i];
+				if (ring->me == me_id && ring->pipe == pipe_id &&
+				    ring->queue == queue_id)
+					drm_sched_fault(&ring->sched);
+			}
+			break;
+		default:
+			BUG();
+			break;
+		}
+	}
+}
+
+static int gfx_v12_0_priv_reg_irq(struct amdgpu_device *adev,
+				  struct amdgpu_irq_src *source,
+				  struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal register access in command stream\n");
+	gfx_v12_0_handle_priv_fault(adev, entry);
+	return 0;
+}
+
+static int gfx_v12_0_bad_op_irq(struct amdgpu_device *adev,
+				struct amdgpu_irq_src *source,
+				struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal opcode in command stream \n");
+	gfx_v12_0_handle_priv_fault(adev, entry);
+	return 0;
+}
+
+static int gfx_v12_0_priv_inst_irq(struct amdgpu_device *adev,
+				   struct amdgpu_irq_src *source,
+				   struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal instruction in command stream\n");
+	gfx_v12_0_handle_priv_fault(adev, entry);
+	return 0;
+}
+
+static void gfx_v12_0_emit_mem_sync(struct amdgpu_ring *ring)
+{
+	const unsigned int gcr_cntl =
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GL2_INV(1) |
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GL2_WB(1) |
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GLM_INV(1) |
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GLM_WB(1) |
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GL1_INV(1) |
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GLV_INV(1) |
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GLK_INV(1) |
+			PACKET3_ACQUIRE_MEM_GCR_CNTL_GLI_INV(1);
+
+	/* ACQUIRE_MEM - make one or more surfaces valid for use by the subsequent operations */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_ACQUIRE_MEM, 6));
+	amdgpu_ring_write(ring, 0); /* CP_COHER_CNTL */
+	amdgpu_ring_write(ring, 0xffffffff);  /* CP_COHER_SIZE */
+	amdgpu_ring_write(ring, 0xffffff);  /* CP_COHER_SIZE_HI */
+	amdgpu_ring_write(ring, 0); /* CP_COHER_BASE */
+	amdgpu_ring_write(ring, 0);  /* CP_COHER_BASE_HI */
+	amdgpu_ring_write(ring, 0x0000000A); /* POLL_INTERVAL */
+	amdgpu_ring_write(ring, gcr_cntl); /* GCR_CNTL */
+}
+
+static void gfx_v12_ring_insert_nop(struct amdgpu_ring *ring, uint32_t num_nop)
+{
+	/* Header itself is a NOP packet */
+	if (num_nop == 1) {
+		amdgpu_ring_write(ring, ring->funcs->nop);
+		return;
+	}
+
+	/* Max HW optimization till 0x3ffe, followed by remaining one NOP at a time*/
+	amdgpu_ring_write(ring, PACKET3(PACKET3_NOP, min(num_nop - 2, 0x3ffe)));
+
+	/* Header is at index 0, followed by num_nops - 1 NOP packet's */
+	amdgpu_ring_insert_nop(ring, num_nop - 1);
+}
+
+static void gfx_v12_0_ring_emit_cleaner_shader(struct amdgpu_ring *ring)
+{
+	/* Emit the cleaner shader */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_RUN_CLEANER_SHADER, 0));
+	amdgpu_ring_write(ring, 0);  /* RESERVED field, programmed to zero */
+}
+
+static void gfx_v12_ip_print(struct amdgpu_ip_block *ip_block, struct drm_printer *p)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	uint32_t i, j, k, reg, index = 0;
+	uint32_t reg_count = ARRAY_SIZE(gc_reg_list_12_0);
+
+	if (!adev->gfx.ip_dump_core)
+		return;
+
+	for (i = 0; i < reg_count; i++)
+		drm_printf(p, "%-50s \t 0x%08x\n",
+			   gc_reg_list_12_0[i].reg_name,
+			   adev->gfx.ip_dump_core[i]);
+
+	/* print compute queue registers for all instances */
+	if (!adev->gfx.ip_dump_compute_queues)
+		return;
+
+	reg_count = ARRAY_SIZE(gc_cp_reg_list_12);
+	drm_printf(p, "\nnum_mec: %d num_pipe: %d num_queue: %d\n",
+		   adev->gfx.mec.num_mec,
+		   adev->gfx.mec.num_pipe_per_mec,
+		   adev->gfx.mec.num_queue_per_pipe);
+
+	for (i = 0; i < adev->gfx.mec.num_mec; i++) {
+		for (j = 0; j < adev->gfx.mec.num_pipe_per_mec; j++) {
+			for (k = 0; k < adev->gfx.mec.num_queue_per_pipe; k++) {
+				drm_printf(p, "\nmec %d, pipe %d, queue %d\n", i, j, k);
+				for (reg = 0; reg < reg_count; reg++) {
+					drm_printf(p, "%-50s \t 0x%08x\n",
+						   gc_cp_reg_list_12[reg].reg_name,
+						   adev->gfx.ip_dump_compute_queues[index + reg]);
+				}
+				index += reg_count;
+			}
+		}
+	}
+
+	/* print gfx queue registers for all instances */
+	if (!adev->gfx.ip_dump_gfx_queues)
+		return;
+
+	index = 0;
+	reg_count = ARRAY_SIZE(gc_gfx_queue_reg_list_12);
+	drm_printf(p, "\nnum_me: %d num_pipe: %d num_queue: %d\n",
+		   adev->gfx.me.num_me,
+		   adev->gfx.me.num_pipe_per_me,
+		   adev->gfx.me.num_queue_per_pipe);
+
+	for (i = 0; i < adev->gfx.me.num_me; i++) {
+		for (j = 0; j < adev->gfx.me.num_pipe_per_me; j++) {
+			for (k = 0; k < adev->gfx.me.num_queue_per_pipe; k++) {
+				drm_printf(p, "\nme %d, pipe %d, queue %d\n", i, j, k);
+				for (reg = 0; reg < reg_count; reg++) {
+					drm_printf(p, "%-50s \t 0x%08x\n",
+						   gc_gfx_queue_reg_list_12[reg].reg_name,
+						   adev->gfx.ip_dump_gfx_queues[index + reg]);
+				}
+				index += reg_count;
+			}
+		}
+	}
+}
+
+static void gfx_v12_ip_dump(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	uint32_t i, j, k, reg, index = 0;
+	uint32_t reg_count = ARRAY_SIZE(gc_reg_list_12_0);
+
+	if (!adev->gfx.ip_dump_core)
+		return;
+
+	amdgpu_gfx_off_ctrl(adev, false);
+	for (i = 0; i < reg_count; i++)
+		adev->gfx.ip_dump_core[i] = RREG32(SOC15_REG_ENTRY_OFFSET(gc_reg_list_12_0[i]));
+	amdgpu_gfx_off_ctrl(adev, true);
+
+	/* dump compute queue registers for all instances */
+	if (!adev->gfx.ip_dump_compute_queues)
+		return;
+
+	reg_count = ARRAY_SIZE(gc_cp_reg_list_12);
+	amdgpu_gfx_off_ctrl(adev, false);
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < adev->gfx.mec.num_mec; i++) {
+		for (j = 0; j < adev->gfx.mec.num_pipe_per_mec; j++) {
+			for (k = 0; k < adev->gfx.mec.num_queue_per_pipe; k++) {
+				/* ME0 is for GFX so start from 1 for CP */
+				soc24_grbm_select(adev, adev->gfx.me.num_me + i, j, k, 0);
+				for (reg = 0; reg < reg_count; reg++) {
+					adev->gfx.ip_dump_compute_queues[index + reg] =
+						RREG32(SOC15_REG_ENTRY_OFFSET(
+							gc_cp_reg_list_12[reg]));
+				}
+				index += reg_count;
+			}
+		}
+	}
+	soc24_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+	amdgpu_gfx_off_ctrl(adev, true);
+
+	/* dump gfx queue registers for all instances */
+	if (!adev->gfx.ip_dump_gfx_queues)
+		return;
+
+	index = 0;
+	reg_count = ARRAY_SIZE(gc_gfx_queue_reg_list_12);
+	amdgpu_gfx_off_ctrl(adev, false);
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < adev->gfx.me.num_me; i++) {
+		for (j = 0; j < adev->gfx.me.num_pipe_per_me; j++) {
+			for (k = 0; k < adev->gfx.me.num_queue_per_pipe; k++) {
+				soc24_grbm_select(adev, i, j, k, 0);
+
+				for (reg = 0; reg < reg_count; reg++) {
+					adev->gfx.ip_dump_gfx_queues[index + reg] =
+						RREG32(SOC15_REG_ENTRY_OFFSET(
+							gc_gfx_queue_reg_list_12[reg]));
+				}
+				index += reg_count;
+			}
+		}
+	}
+	soc24_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+	amdgpu_gfx_off_ctrl(adev, true);
+}
+
+static bool gfx_v12_pipe_reset_support(struct amdgpu_device *adev)
+{
+	/* Disable the pipe reset until the CPFW fully support it.*/
+	dev_warn_once(adev->dev, "The CPFW hasn't support pipe reset yet.\n");
+	return false;
+}
+
+static int gfx_v12_reset_gfx_pipe(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t reset_pipe = 0, clean_pipe = 0;
+	int r;
+
+	if (!gfx_v12_pipe_reset_support(adev))
+		return -EOPNOTSUPP;
+
+	gfx_v12_0_set_safe_mode(adev, 0);
+	mutex_lock(&adev->srbm_mutex);
+	soc24_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+
+	switch (ring->pipe) {
+	case 0:
+		reset_pipe = REG_SET_FIELD(reset_pipe, CP_ME_CNTL,
+					   PFP_PIPE0_RESET, 1);
+		reset_pipe = REG_SET_FIELD(reset_pipe, CP_ME_CNTL,
+					   ME_PIPE0_RESET, 1);
+		clean_pipe = REG_SET_FIELD(clean_pipe, CP_ME_CNTL,
+					   PFP_PIPE0_RESET, 0);
+		clean_pipe = REG_SET_FIELD(clean_pipe, CP_ME_CNTL,
+					   ME_PIPE0_RESET, 0);
+		break;
+	case 1:
+		reset_pipe = REG_SET_FIELD(reset_pipe, CP_ME_CNTL,
+					   PFP_PIPE1_RESET, 1);
+		reset_pipe = REG_SET_FIELD(reset_pipe, CP_ME_CNTL,
+					   ME_PIPE1_RESET, 1);
+		clean_pipe = REG_SET_FIELD(clean_pipe, CP_ME_CNTL,
+					   PFP_PIPE1_RESET, 0);
+		clean_pipe = REG_SET_FIELD(clean_pipe, CP_ME_CNTL,
+					   ME_PIPE1_RESET, 0);
+		break;
+	default:
+		break;
+	}
+
+	WREG32_SOC15(GC, 0, regCP_ME_CNTL, reset_pipe);
+	WREG32_SOC15(GC, 0, regCP_ME_CNTL, clean_pipe);
+
+	r = (RREG32(SOC15_REG_OFFSET(GC, 0, regCP_GFX_RS64_INSTR_PNTR1)) << 2) -
+					RS64_FW_UC_START_ADDR_LO;
+	soc24_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+	gfx_v12_0_unset_safe_mode(adev, 0);
+
+	dev_info(adev->dev, "The ring %s pipe reset: %s\n", ring->name,
+			r == 0 ? "successfully" : "failed");
+	/* Sometimes the ME start pc counter can't cache correctly, so the
+	 * PC check only as a reference and pipe reset result rely on the
+	 * later ring test.
+	 */
+	return 0;
+}
+
+static int gfx_v12_0_reset_kgq(struct amdgpu_ring *ring,
+			       unsigned int vmid,
+			       struct amdgpu_fence *timedout_fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int r;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+
+	r = amdgpu_mes_reset_legacy_queue(ring->adev, ring, vmid, false);
+	if (r) {
+		dev_warn(adev->dev, "reset via MES failed and try pipe reset %d\n", r);
+		r = gfx_v12_reset_gfx_pipe(ring);
+		if (r)
+			return r;
+	}
+
+	r = gfx_v12_0_kgq_init_queue(ring, true);
+	if (r) {
+		dev_err(adev->dev, "failed to init kgq\n");
+		return r;
+	}
+
+	r = amdgpu_mes_map_legacy_queue(adev, ring);
+	if (r) {
+		dev_err(adev->dev, "failed to remap kgq\n");
+		return r;
+	}
+
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static int gfx_v12_0_reset_compute_pipe(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t reset_pipe = 0, clean_pipe = 0;
+	int r = 0;
+
+	if (!gfx_v12_pipe_reset_support(adev))
+		return -EOPNOTSUPP;
+
+	gfx_v12_0_set_safe_mode(adev, 0);
+	mutex_lock(&adev->srbm_mutex);
+	soc24_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+
+	reset_pipe = RREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL);
+	clean_pipe = reset_pipe;
+
+	if (adev->gfx.rs64_enable) {
+		switch (ring->pipe) {
+		case 0:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE0_RESET, 1);
+			clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE0_RESET, 0);
+			break;
+		case 1:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE1_RESET, 1);
+			clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE1_RESET, 0);
+			break;
+		case 2:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE2_RESET, 1);
+			clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE2_RESET, 0);
+			break;
+		case 3:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE3_RESET, 1);
+			clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE3_RESET, 0);
+			break;
+		default:
+			break;
+		}
+		WREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL, reset_pipe);
+		WREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL, clean_pipe);
+		r = (RREG32_SOC15(GC, 0, regCP_MEC_RS64_INSTR_PNTR) << 2) -
+				RS64_FW_UC_START_ADDR_LO;
+	} else {
+		switch (ring->pipe) {
+		case 0:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE0_RESET, 1);
+			clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE0_RESET, 0);
+			break;
+		case 1:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE1_RESET, 1);
+			clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE1_RESET, 0);
+			break;
+		default:
+		break;
+		}
+		WREG32_SOC15(GC, 0, regCP_MEC_CNTL, reset_pipe);
+		WREG32_SOC15(GC, 0, regCP_MEC_CNTL, clean_pipe);
+		/* Doesn't find the F32 MEC instruction pointer register, and suppose
+		 * the driver won't run into the F32 mode.
+		 */
+	}
+
+	soc24_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+	gfx_v12_0_unset_safe_mode(adev, 0);
+
+	dev_info(adev->dev, "The ring %s pipe resets: %s\n", ring->name,
+			r == 0 ? "successfully" : "failed");
+	/* Need the ring test to verify the pipe reset result.*/
+	return 0;
+}
+
+static int gfx_v12_0_reset_kcq(struct amdgpu_ring *ring,
+			       unsigned int vmid,
+			       struct amdgpu_fence *timedout_fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int r;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+
+	r = amdgpu_mes_reset_legacy_queue(ring->adev, ring, vmid, true);
+	if (r) {
+		dev_warn(adev->dev, "fail(%d) to reset kcq  and try pipe reset\n", r);
+		r = gfx_v12_0_reset_compute_pipe(ring);
+		if (r)
+			return r;
+	}
+
+	r = gfx_v12_0_kcq_init_queue(ring, true);
+	if (r) {
+		dev_err(adev->dev, "failed to init kcq\n");
+		return r;
+	}
+	r = amdgpu_mes_map_legacy_queue(adev, ring);
+	if (r) {
+		dev_err(adev->dev, "failed to remap kcq\n");
+		return r;
+	}
+
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static void gfx_v12_0_ring_begin_use(struct amdgpu_ring *ring)
+{
+	amdgpu_gfx_profile_ring_begin_use(ring);
+
+	amdgpu_gfx_enforce_isolation_ring_begin_use(ring);
+}
+
+static void gfx_v12_0_ring_end_use(struct amdgpu_ring *ring)
+{
+	amdgpu_gfx_profile_ring_end_use(ring);
+
+	amdgpu_gfx_enforce_isolation_ring_end_use(ring);
+}
+
+static const struct amd_ip_funcs gfx_v12_0_ip_funcs = {
+	.name = "gfx_v12_0",
+	.early_init = gfx_v12_0_early_init,
+	.late_init = gfx_v12_0_late_init,
+	.sw_init = gfx_v12_0_sw_init,
+	.sw_fini = gfx_v12_0_sw_fini,
+	.hw_init = gfx_v12_0_hw_init,
+	.hw_fini = gfx_v12_0_hw_fini,
+	.suspend = gfx_v12_0_suspend,
+	.resume = gfx_v12_0_resume,
+	.is_idle = gfx_v12_0_is_idle,
+	.wait_for_idle = gfx_v12_0_wait_for_idle,
+	.set_clockgating_state = gfx_v12_0_set_clockgating_state,
+	.set_powergating_state = gfx_v12_0_set_powergating_state,
+	.get_clockgating_state = gfx_v12_0_get_clockgating_state,
+	.dump_ip_state = gfx_v12_ip_dump,
+	.print_ip_state = gfx_v12_ip_print,
+};
+
+static const struct amdgpu_ring_funcs gfx_v12_0_ring_funcs_gfx = {
+	.type = AMDGPU_RING_TYPE_GFX,
+	.align_mask = 0xff,
+	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
+	.support_64bit_ptrs = true,
+	.secure_submission_supported = true,
+	.get_rptr = gfx_v12_0_ring_get_rptr_gfx,
+	.get_wptr = gfx_v12_0_ring_get_wptr_gfx,
+	.set_wptr = gfx_v12_0_ring_set_wptr_gfx,
+	.emit_frame_size = /* totally 242 maximum if 16 IBs */
+		5 + /* COND_EXEC */
+		7 + /* PIPELINE_SYNC */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 5 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 +
+		2 + /* VM_FLUSH */
+		8 + /* FENCE for VM_FLUSH */
+		5 + /* COND_EXEC */
+		7 + /* HDP_flush */
+		4 + /* VGT_flush */
+		31 + /*	DE_META */
+		3 + /* CNTX_CTRL */
+		5 + /* HDP_INVL */
+		8 + 8 + /* FENCE x2 */
+		8 + /* gfx_v12_0_emit_mem_sync */
+		2, /* gfx_v12_0_ring_emit_cleaner_shader */
+	.emit_ib_size =	4, /* gfx_v12_0_ring_emit_ib_gfx */
+	.emit_ib = gfx_v12_0_ring_emit_ib_gfx,
+	.emit_fence = gfx_v12_0_ring_emit_fence,
+	.emit_pipeline_sync = gfx_v12_0_ring_emit_pipeline_sync,
+	.emit_vm_flush = gfx_v12_0_ring_emit_vm_flush,
+	.emit_hdp_flush = gfx_v12_0_ring_emit_hdp_flush,
+	.test_ring = gfx_v12_0_ring_test_ring,
+	.test_ib = gfx_v12_0_ring_test_ib,
+	.insert_nop = gfx_v12_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_cntxcntl = gfx_v12_0_ring_emit_cntxcntl,
+	.init_cond_exec = gfx_v12_0_ring_emit_init_cond_exec,
+	.preempt_ib = gfx_v12_0_ring_preempt_ib,
+	.emit_frame_cntl = gfx_v12_0_ring_emit_frame_cntl,
+	.emit_wreg = gfx_v12_0_ring_emit_wreg,
+	.emit_reg_wait = gfx_v12_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = gfx_v12_0_ring_emit_reg_write_reg_wait,
+	.emit_mem_sync = gfx_v12_0_emit_mem_sync,
+	.reset = gfx_v12_0_reset_kgq,
+	.emit_cleaner_shader = gfx_v12_0_ring_emit_cleaner_shader,
+	.begin_use = gfx_v12_0_ring_begin_use,
+	.end_use = gfx_v12_0_ring_end_use,
+};
+
+static const struct amdgpu_ring_funcs gfx_v12_0_ring_funcs_compute = {
+	.type = AMDGPU_RING_TYPE_COMPUTE,
+	.align_mask = 0xff,
+	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
+	.support_64bit_ptrs = true,
+	.get_rptr = gfx_v12_0_ring_get_rptr_compute,
+	.get_wptr = gfx_v12_0_ring_get_wptr_compute,
+	.set_wptr = gfx_v12_0_ring_set_wptr_compute,
+	.emit_frame_size =
+		7 + /* gfx_v12_0_ring_emit_hdp_flush */
+		5 + /* hdp invalidate */
+		7 + /* gfx_v12_0_ring_emit_pipeline_sync */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 5 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 +
+		2 + /* gfx_v12_0_ring_emit_vm_flush */
+		8 + 8 + 8 + /* gfx_v12_0_ring_emit_fence x3 for user fence, vm fence */
+		8 + /* gfx_v12_0_emit_mem_sync */
+		2, /* gfx_v12_0_ring_emit_cleaner_shader */
+	.emit_ib_size =	7, /* gfx_v12_0_ring_emit_ib_compute */
+	.emit_ib = gfx_v12_0_ring_emit_ib_compute,
+	.emit_fence = gfx_v12_0_ring_emit_fence,
+	.emit_pipeline_sync = gfx_v12_0_ring_emit_pipeline_sync,
+	.emit_vm_flush = gfx_v12_0_ring_emit_vm_flush,
+	.emit_hdp_flush = gfx_v12_0_ring_emit_hdp_flush,
+	.test_ring = gfx_v12_0_ring_test_ring,
+	.test_ib = gfx_v12_0_ring_test_ib,
+	.insert_nop = gfx_v12_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_wreg = gfx_v12_0_ring_emit_wreg,
+	.emit_reg_wait = gfx_v12_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = gfx_v12_0_ring_emit_reg_write_reg_wait,
+	.emit_mem_sync = gfx_v12_0_emit_mem_sync,
+	.reset = gfx_v12_0_reset_kcq,
+	.emit_cleaner_shader = gfx_v12_0_ring_emit_cleaner_shader,
+	.begin_use = gfx_v12_0_ring_begin_use,
+	.end_use = gfx_v12_0_ring_end_use,
+};
+
+static const struct amdgpu_ring_funcs gfx_v12_0_ring_funcs_kiq = {
+	.type = AMDGPU_RING_TYPE_KIQ,
+	.align_mask = 0xff,
+	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
+	.support_64bit_ptrs = true,
+	.get_rptr = gfx_v12_0_ring_get_rptr_compute,
+	.get_wptr = gfx_v12_0_ring_get_wptr_compute,
+	.set_wptr = gfx_v12_0_ring_set_wptr_compute,
+	.emit_frame_size =
+		7 + /* gfx_v12_0_ring_emit_hdp_flush */
+		5 + /*hdp invalidate */
+		7 + /* gfx_v12_0_ring_emit_pipeline_sync */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 5 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 +
+		2 + /* gfx_v12_0_ring_emit_vm_flush */
+		8 + 8 + 8, /* gfx_v12_0_ring_emit_fence_kiq x3 for user fence, vm fence */
+	.emit_ib_size =	7, /* gfx_v12_0_ring_emit_ib_compute */
+	.emit_ib = gfx_v12_0_ring_emit_ib_compute,
+	.emit_fence = gfx_v12_0_ring_emit_fence_kiq,
+	.test_ring = gfx_v12_0_ring_test_ring,
+	.test_ib = gfx_v12_0_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_rreg = gfx_v12_0_ring_emit_rreg,
+	.emit_wreg = gfx_v12_0_ring_emit_wreg,
+	.emit_reg_wait = gfx_v12_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = gfx_v12_0_ring_emit_reg_write_reg_wait,
+};
+
+static void gfx_v12_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	adev->gfx.kiq[0].ring.funcs = &gfx_v12_0_ring_funcs_kiq;
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+		adev->gfx.gfx_ring[i].funcs = &gfx_v12_0_ring_funcs_gfx;
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++)
+		adev->gfx.compute_ring[i].funcs = &gfx_v12_0_ring_funcs_compute;
+}
+
+static const struct amdgpu_irq_src_funcs gfx_v12_0_eop_irq_funcs = {
+	.set = gfx_v12_0_set_eop_interrupt_state,
+	.process = gfx_v12_0_eop_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v12_0_priv_reg_irq_funcs = {
+	.set = gfx_v12_0_set_priv_reg_fault_state,
+	.process = gfx_v12_0_priv_reg_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v12_0_bad_op_irq_funcs = {
+	.set = gfx_v12_0_set_bad_op_fault_state,
+	.process = gfx_v12_0_bad_op_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v12_0_priv_inst_irq_funcs = {
+	.set = gfx_v12_0_set_priv_inst_fault_state,
+	.process = gfx_v12_0_priv_inst_irq,
+};
+
+static void gfx_v12_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->gfx.eop_irq.num_types = AMDGPU_CP_IRQ_LAST;
+	adev->gfx.eop_irq.funcs = &gfx_v12_0_eop_irq_funcs;
+
+	adev->gfx.priv_reg_irq.num_types = 1;
+	adev->gfx.priv_reg_irq.funcs = &gfx_v12_0_priv_reg_irq_funcs;
+
+	adev->gfx.bad_op_irq.num_types = 1;
+	adev->gfx.bad_op_irq.funcs = &gfx_v12_0_bad_op_irq_funcs;
+
+	adev->gfx.priv_inst_irq.num_types = 1;
+	adev->gfx.priv_inst_irq.funcs = &gfx_v12_0_priv_inst_irq_funcs;
+}
+
+static void gfx_v12_0_set_imu_funcs(struct amdgpu_device *adev)
+{
+	if (adev->flags & AMD_IS_APU)
+		adev->gfx.imu.mode = MISSION_MODE;
+	else
+		adev->gfx.imu.mode = DEBUG_MODE;
+
+	adev->gfx.imu.funcs = &gfx_v12_0_imu_funcs;
+}
+
+static void gfx_v12_0_set_rlc_funcs(struct amdgpu_device *adev)
+{
+	adev->gfx.rlc.funcs = &gfx_v12_0_rlc_funcs;
+}
+
+static void gfx_v12_0_set_mqd_funcs(struct amdgpu_device *adev)
+{
+	/* set gfx eng mqd */
+	adev->mqds[AMDGPU_HW_IP_GFX].mqd_size =
+		sizeof(struct v12_gfx_mqd);
+	adev->mqds[AMDGPU_HW_IP_GFX].init_mqd =
+		gfx_v12_0_gfx_mqd_init;
+	/* set compute eng mqd */
+	adev->mqds[AMDGPU_HW_IP_COMPUTE].mqd_size =
+		sizeof(struct v12_compute_mqd);
+	adev->mqds[AMDGPU_HW_IP_COMPUTE].init_mqd =
+		gfx_v12_0_compute_mqd_init;
+}
+
+static void gfx_v12_0_set_user_wgp_inactive_bitmap_per_sh(struct amdgpu_device *adev,
+							  u32 bitmap)
+{
+	u32 data;
+
+	if (!bitmap)
+		return;
+
+	data = bitmap << GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_WGPS__SHIFT;
+	data &= GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_WGPS_MASK;
+
+	WREG32_SOC15(GC, 0, regGC_USER_SHADER_ARRAY_CONFIG, data);
+}
+
+static u32 gfx_v12_0_get_wgp_active_bitmap_per_sh(struct amdgpu_device *adev)
+{
+	u32 data, wgp_bitmask;
+	data = RREG32_SOC15(GC, 0, regCC_GC_SHADER_ARRAY_CONFIG);
+	data |= RREG32_SOC15(GC, 0, regGC_USER_SHADER_ARRAY_CONFIG);
+
+	data &= CC_GC_SHADER_ARRAY_CONFIG__INACTIVE_WGPS_MASK;
+	data >>= CC_GC_SHADER_ARRAY_CONFIG__INACTIVE_WGPS__SHIFT;
+
+	wgp_bitmask =
+		amdgpu_gfx_create_bitmask(adev->gfx.config.max_cu_per_sh >> 1);
+
+	return (~data) & wgp_bitmask;
+}
+
+static u32 gfx_v12_0_get_cu_active_bitmap_per_sh(struct amdgpu_device *adev)
+{
+	u32 wgp_idx, wgp_active_bitmap;
+	u32 cu_bitmap_per_wgp, cu_active_bitmap;
+
+	wgp_active_bitmap = gfx_v12_0_get_wgp_active_bitmap_per_sh(adev);
+	cu_active_bitmap = 0;
+
+	for (wgp_idx = 0; wgp_idx < 16; wgp_idx++) {
+		/* if there is one WGP enabled, it means 2 CUs will be enabled */
+		cu_bitmap_per_wgp = 3 << (2 * wgp_idx);
+		if (wgp_active_bitmap & (1 << wgp_idx))
+			cu_active_bitmap |= cu_bitmap_per_wgp;
+	}
+
+	return cu_active_bitmap;
+}
+
+static int gfx_v12_0_get_cu_info(struct amdgpu_device *adev,
+				 struct amdgpu_cu_info *cu_info)
+{
+	int i, j, k, counter, active_cu_number = 0;
+	u32 mask, bitmap;
+	unsigned disable_masks[8 * 2];
+
+	if (!adev || !cu_info)
+		return -EINVAL;
+
+	amdgpu_gfx_parse_disable_cu(disable_masks, 8, 2);
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			bitmap = i * adev->gfx.config.max_sh_per_se + j;
+			if (!((gfx_v12_0_get_sa_active_bitmap(adev) >> bitmap) & 1))
+				continue;
+			mask = 1;
+			counter = 0;
+			gfx_v12_0_select_se_sh(adev, i, j, 0xffffffff, 0);
+			if (i < 8 && j < 2)
+				gfx_v12_0_set_user_wgp_inactive_bitmap_per_sh(
+					adev, disable_masks[i * 2 + j]);
+			bitmap = gfx_v12_0_get_cu_active_bitmap_per_sh(adev);
+
+			/**
+			 * GFX12 could support more than 4 SEs, while the bitmap
+			 * in cu_info struct is 4x4 and ioctl interface struct
+			 * drm_amdgpu_info_device should keep stable.
+			 * So we use last two columns of bitmap to store cu mask for
+			 * SEs 4 to 7, the layout of the bitmap is as below:
+			 *    SE0: {SH0,SH1} --> {bitmap[0][0], bitmap[0][1]}
+			 *    SE1: {SH0,SH1} --> {bitmap[1][0], bitmap[1][1]}
+			 *    SE2: {SH0,SH1} --> {bitmap[2][0], bitmap[2][1]}
+			 *    SE3: {SH0,SH1} --> {bitmap[3][0], bitmap[3][1]}
+			 *    SE4: {SH0,SH1} --> {bitmap[0][2], bitmap[0][3]}
+			 *    SE5: {SH0,SH1} --> {bitmap[1][2], bitmap[1][3]}
+			 *    SE6: {SH0,SH1} --> {bitmap[2][2], bitmap[2][3]}
+			 *    SE7: {SH0,SH1} --> {bitmap[3][2], bitmap[3][3]}
+			 */
+			cu_info->bitmap[0][i % 4][j + (i / 4) * 2] = bitmap;
+
+			for (k = 0; k < adev->gfx.config.max_cu_per_sh; k++) {
+				if (bitmap & mask)
+					counter++;
+
+				mask <<= 1;
+			}
+			active_cu_number += counter;
+		}
+	}
+	gfx_v12_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	cu_info->number = active_cu_number;
+	cu_info->simd_per_cu = NUM_SIMD_PER_CU;
+
+	return 0;
+}
+
+const struct amdgpu_ip_block_version gfx_v12_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_GFX,
+	.major = 12,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &gfx_v12_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.h b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.h
new file mode 100644
index 000000000000..f7184b2dc4e8
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2023 dvanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFX_V12_0_H__
+#define __GFX_V12_0_H__
+
+extern const struct amdgpu_ip_block_version gfx_v12_0_ip_block;
+
+int gfx_v12_0_request_gfx_index_mutex(struct amdgpu_device *adev,
+				      bool req);
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c
index d228f5a99044..7693b7953426 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c
@@ -21,55 +21,72 @@
  *
  */
 #include <linux/firmware.h>
+#include <linux/module.h>
+
 #include "amdgpu.h"
 #include "amdgpu_ih.h"
 #include "amdgpu_gfx.h"
 #include "amdgpu_ucode.h"
 #include "clearstate_si.h"
+#include "si.h"
+#include "sid.h"
+
 #include "bif/bif_3_0_d.h"
 #include "bif/bif_3_0_sh_mask.h"
+
 #include "oss/oss_1_0_d.h"
 #include "oss/oss_1_0_sh_mask.h"
+
 #include "gca/gfx_6_0_d.h"
 #include "gca/gfx_6_0_sh_mask.h"
+#include "gca/gfx_7_2_enum.h"
+
 #include "gmc/gmc_6_0_d.h"
 #include "gmc/gmc_6_0_sh_mask.h"
+
 #include "dce/dce_6_0_d.h"
 #include "dce/dce_6_0_sh_mask.h"
-#include "gca/gfx_7_2_enum.h"
+
 #include "si_enums.h"
 
+#define TAHITI_GB_ADDR_CONFIG_GOLDEN        0x12011003
+#define VERDE_GB_ADDR_CONFIG_GOLDEN         0x12010002
+#define HAINAN_GB_ADDR_CONFIG_GOLDEN        0x02010001
+
+#define GFX6_NUM_GFX_RINGS     1
+#define GFX6_NUM_COMPUTE_RINGS 2
+
 static void gfx_v6_0_set_ring_funcs(struct amdgpu_device *adev);
 static void gfx_v6_0_set_irq_funcs(struct amdgpu_device *adev);
 static void gfx_v6_0_get_cu_info(struct amdgpu_device *adev);
 
-MODULE_FIRMWARE("radeon/tahiti_pfp.bin");
-MODULE_FIRMWARE("radeon/tahiti_me.bin");
-MODULE_FIRMWARE("radeon/tahiti_ce.bin");
-MODULE_FIRMWARE("radeon/tahiti_rlc.bin");
+MODULE_FIRMWARE("amdgpu/tahiti_pfp.bin");
+MODULE_FIRMWARE("amdgpu/tahiti_me.bin");
+MODULE_FIRMWARE("amdgpu/tahiti_ce.bin");
+MODULE_FIRMWARE("amdgpu/tahiti_rlc.bin");
 
-MODULE_FIRMWARE("radeon/pitcairn_pfp.bin");
-MODULE_FIRMWARE("radeon/pitcairn_me.bin");
-MODULE_FIRMWARE("radeon/pitcairn_ce.bin");
-MODULE_FIRMWARE("radeon/pitcairn_rlc.bin");
+MODULE_FIRMWARE("amdgpu/pitcairn_pfp.bin");
+MODULE_FIRMWARE("amdgpu/pitcairn_me.bin");
+MODULE_FIRMWARE("amdgpu/pitcairn_ce.bin");
+MODULE_FIRMWARE("amdgpu/pitcairn_rlc.bin");
 
-MODULE_FIRMWARE("radeon/verde_pfp.bin");
-MODULE_FIRMWARE("radeon/verde_me.bin");
-MODULE_FIRMWARE("radeon/verde_ce.bin");
-MODULE_FIRMWARE("radeon/verde_rlc.bin");
+MODULE_FIRMWARE("amdgpu/verde_pfp.bin");
+MODULE_FIRMWARE("amdgpu/verde_me.bin");
+MODULE_FIRMWARE("amdgpu/verde_ce.bin");
+MODULE_FIRMWARE("amdgpu/verde_rlc.bin");
 
-MODULE_FIRMWARE("radeon/oland_pfp.bin");
-MODULE_FIRMWARE("radeon/oland_me.bin");
-MODULE_FIRMWARE("radeon/oland_ce.bin");
-MODULE_FIRMWARE("radeon/oland_rlc.bin");
+MODULE_FIRMWARE("amdgpu/oland_pfp.bin");
+MODULE_FIRMWARE("amdgpu/oland_me.bin");
+MODULE_FIRMWARE("amdgpu/oland_ce.bin");
+MODULE_FIRMWARE("amdgpu/oland_rlc.bin");
 
-MODULE_FIRMWARE("radeon/hainan_pfp.bin");
-MODULE_FIRMWARE("radeon/hainan_me.bin");
-MODULE_FIRMWARE("radeon/hainan_ce.bin");
-MODULE_FIRMWARE("radeon/hainan_rlc.bin");
+MODULE_FIRMWARE("amdgpu/hainan_pfp.bin");
+MODULE_FIRMWARE("amdgpu/hainan_me.bin");
+MODULE_FIRMWARE("amdgpu/hainan_ce.bin");
+MODULE_FIRMWARE("amdgpu/hainan_rlc.bin");
 
 static u32 gfx_v6_0_get_csb_size(struct amdgpu_device *adev);
-static void gfx_v6_0_get_csb_buffer(struct amdgpu_device *adev, volatile u32 *buffer);
+static void gfx_v6_0_get_csb_buffer(struct amdgpu_device *adev, u32 *buffer);
 //static void gfx_v6_0_init_cp_pg_table(struct amdgpu_device *adev);
 static void gfx_v6_0_init_pg(struct amdgpu_device *adev);
 
@@ -308,7 +325,6 @@ static const u32 verde_rlc_save_restore_register_list[] =
 static int gfx_v6_0_init_microcode(struct amdgpu_device *adev)
 {
 	const char *chip_name;
-	char fw_name[30];
 	int err;
 	const struct gfx_firmware_header_v1_0 *cp_hdr;
 	const struct rlc_firmware_header_v1_0 *rlc_hdr;
@@ -334,59 +350,49 @@ static int gfx_v6_0_init_microcode(struct amdgpu_device *adev)
 	default: BUG();
 	}
 
-	snprintf(fw_name, sizeof(fw_name), "radeon/%s_pfp.bin", chip_name);
-	err = request_firmware(&adev->gfx.pfp_fw, fw_name, adev->dev);
-	if (err)
-		goto out;
-	err = amdgpu_ucode_validate(adev->gfx.pfp_fw);
+	err = amdgpu_ucode_request(adev, &adev->gfx.pfp_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_pfp.bin", chip_name);
 	if (err)
 		goto out;
 	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data;
 	adev->gfx.pfp_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
 	adev->gfx.pfp_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
 
-	snprintf(fw_name, sizeof(fw_name), "radeon/%s_me.bin", chip_name);
-	err = request_firmware(&adev->gfx.me_fw, fw_name, adev->dev);
-	if (err)
-		goto out;
-	err = amdgpu_ucode_validate(adev->gfx.me_fw);
+	err = amdgpu_ucode_request(adev, &adev->gfx.me_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_me.bin", chip_name);
 	if (err)
 		goto out;
 	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data;
 	adev->gfx.me_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
 	adev->gfx.me_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
 
-	snprintf(fw_name, sizeof(fw_name), "radeon/%s_ce.bin", chip_name);
-	err = request_firmware(&adev->gfx.ce_fw, fw_name, adev->dev);
-	if (err)
-		goto out;
-	err = amdgpu_ucode_validate(adev->gfx.ce_fw);
+	err = amdgpu_ucode_request(adev, &adev->gfx.ce_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_ce.bin", chip_name);
 	if (err)
 		goto out;
 	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data;
 	adev->gfx.ce_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
 	adev->gfx.ce_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
 
-	snprintf(fw_name, sizeof(fw_name), "radeon/%s_rlc.bin", chip_name);
-	err = request_firmware(&adev->gfx.rlc_fw, fw_name, adev->dev);
+	err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_rlc.bin", chip_name);
 	if (err)
 		goto out;
-	err = amdgpu_ucode_validate(adev->gfx.rlc_fw);
 	rlc_hdr = (const struct rlc_firmware_header_v1_0 *)adev->gfx.rlc_fw->data;
 	adev->gfx.rlc_fw_version = le32_to_cpu(rlc_hdr->header.ucode_version);
 	adev->gfx.rlc_feature_version = le32_to_cpu(rlc_hdr->ucode_feature_version);
 
 out:
 	if (err) {
-		pr_err("gfx6: Failed to load firmware \"%s\"\n", fw_name);
-		release_firmware(adev->gfx.pfp_fw);
-		adev->gfx.pfp_fw = NULL;
-		release_firmware(adev->gfx.me_fw);
-		adev->gfx.me_fw = NULL;
-		release_firmware(adev->gfx.ce_fw);
-		adev->gfx.ce_fw = NULL;
-		release_firmware(adev->gfx.rlc_fw);
-		adev->gfx.rlc_fw = NULL;
+		pr_err("gfx6: Failed to load firmware %s gfx firmware\n", chip_name);
+		amdgpu_ucode_release(&adev->gfx.pfp_fw);
+		amdgpu_ucode_release(&adev->gfx.me_fw);
+		amdgpu_ucode_release(&adev->gfx.ce_fw);
+		amdgpu_ucode_release(&adev->gfx.rlc_fw);
 	}
 	return err;
 }
@@ -636,7 +642,213 @@ static void gfx_v6_0_tiling_mode_table_init(struct amdgpu_device *adev)
 				NUM_BANKS(ADDR_SURF_2_BANK);
 		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++)
 			WREG32(mmGB_TILE_MODE0 + reg_offset, tilemode[reg_offset]);
-	} else if (adev->asic_type == CHIP_OLAND || adev->asic_type == CHIP_HAINAN) {
+	} else if (adev->asic_type == CHIP_OLAND) {
+		tilemode[0] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4);
+		tilemode[1] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4);
+		tilemode[2] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4);
+		tilemode[3] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4);
+		tilemode[4] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[5] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(split_equal_to_row_size) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[6] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(split_equal_to_row_size) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[7] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(split_equal_to_row_size) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4);
+		tilemode[8] =   MICRO_TILE_MODE(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_LINEAR_ALIGNED) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[9] =   MICRO_TILE_MODE(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[10] =  MICRO_TILE_MODE(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4);
+		tilemode[11] =  MICRO_TILE_MODE(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[12] =  MICRO_TILE_MODE(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[13] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[14] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[15] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[16] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[17] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(split_equal_to_row_size) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[18] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16);
+		tilemode[19] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_XTHICK) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				TILE_SPLIT(split_equal_to_row_size);
+		tilemode[20] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				TILE_SPLIT(split_equal_to_row_size);
+		tilemode[21] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[22] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4);
+		tilemode[23] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[24] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[25] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_1KB) |
+				NUM_BANKS(ADDR_SURF_8_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1);
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++)
+			WREG32(mmGB_TILE_MODE0 + reg_offset, tilemode[reg_offset]);
+	} else if (adev->asic_type == CHIP_HAINAN) {
 		tilemode[0] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
 				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
 				PIPE_CONFIG(ADDR_SURF_P2) |
@@ -1090,7 +1302,7 @@ static void gfx_v6_0_tiling_mode_table_init(struct amdgpu_device *adev)
 }
 
 static void gfx_v6_0_select_se_sh(struct amdgpu_device *adev, u32 se_num,
-				  u32 sh_num, u32 instance)
+				  u32 sh_num, u32 instance, int xcc_id)
 {
 	u32 data;
 
@@ -1243,12 +1455,12 @@ static void gfx_v6_0_write_harvested_raster_configs(struct amdgpu_device *adev,
 		}
 
 		/* GRBM_GFX_INDEX has a different offset on SI */
-		gfx_v6_0_select_se_sh(adev, se, 0xffffffff, 0xffffffff);
+		gfx_v6_0_select_se_sh(adev, se, 0xffffffff, 0xffffffff, 0);
 		WREG32(mmPA_SC_RASTER_CONFIG, raster_config_se);
 	}
 
 	/* GRBM_GFX_INDEX has a different offset on SI */
-	gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 }
 
 static void gfx_v6_0_setup_rb(struct amdgpu_device *adev)
@@ -1264,14 +1476,14 @@ static void gfx_v6_0_setup_rb(struct amdgpu_device *adev)
 	mutex_lock(&adev->grbm_idx_mutex);
 	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
 		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
-			gfx_v6_0_select_se_sh(adev, i, j, 0xffffffff);
+			gfx_v6_0_select_se_sh(adev, i, j, 0xffffffff, 0);
 			data = gfx_v6_0_get_rb_active_bitmap(adev);
 			active_rbs |= data <<
 				((i * adev->gfx.config.max_sh_per_se + j) *
 				 rb_bitmap_width_per_sh);
 		}
 	}
-	gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 
 	adev->gfx.config.backend_enable_mask = active_rbs;
 	adev->gfx.config.num_rbs = hweight32(active_rbs);
@@ -1292,7 +1504,7 @@ static void gfx_v6_0_setup_rb(struct amdgpu_device *adev)
 	/* cache the values for userspace */
 	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
 		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
-			gfx_v6_0_select_se_sh(adev, i, j, 0xffffffff);
+			gfx_v6_0_select_se_sh(adev, i, j, 0xffffffff, 0);
 			adev->gfx.config.rb_config[i][j].rb_backend_disable =
 				RREG32(mmCC_RB_BACKEND_DISABLE);
 			adev->gfx.config.rb_config[i][j].user_rb_backend_disable =
@@ -1301,7 +1513,7 @@ static void gfx_v6_0_setup_rb(struct amdgpu_device *adev)
 				RREG32(mmPA_SC_RASTER_CONFIG);
 		}
 	}
-	gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 	mutex_unlock(&adev->grbm_idx_mutex);
 }
 
@@ -1340,7 +1552,7 @@ static void gfx_v6_0_setup_spi(struct amdgpu_device *adev)
 	mutex_lock(&adev->grbm_idx_mutex);
 	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
 		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
-			gfx_v6_0_select_se_sh(adev, i, j, 0xffffffff);
+			gfx_v6_0_select_se_sh(adev, i, j, 0xffffffff, 0);
 			data = RREG32(mmSPI_STATIC_THREAD_MGMT_3);
 			active_cu = gfx_v6_0_get_cu_enabled(adev);
 
@@ -1355,7 +1567,7 @@ static void gfx_v6_0_setup_spi(struct amdgpu_device *adev)
 			}
 		}
 	}
-	gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 	mutex_unlock(&adev->grbm_idx_mutex);
 }
 
@@ -1364,10 +1576,10 @@ static void gfx_v6_0_config_init(struct amdgpu_device *adev)
 	adev->gfx.config.double_offchip_lds_buf = 0;
 }
 
-static void gfx_v6_0_gpu_init(struct amdgpu_device *adev)
+static void gfx_v6_0_constants_init(struct amdgpu_device *adev)
 {
 	u32 gb_addr_config = 0;
-	u32 mc_shared_chmap, mc_arb_ramcfg;
+	u32 mc_arb_ramcfg;
 	u32 sx_debug_1;
 	u32 hdp_host_path_cntl;
 	u32 tmp;
@@ -1469,7 +1681,6 @@ static void gfx_v6_0_gpu_init(struct amdgpu_device *adev)
 
 	WREG32(mmBIF_FB_EN, BIF_FB_EN__FB_READ_EN_MASK | BIF_FB_EN__FB_WRITE_EN_MASK);
 
-	mc_shared_chmap = RREG32(mmMC_SHARED_CHMAP);
 	adev->gfx.config.mc_arb_ramcfg = RREG32(mmMC_ARB_RAMCFG);
 	mc_arb_ramcfg = adev->gfx.config.mc_arb_ramcfg;
 
@@ -1524,10 +1735,14 @@ static void gfx_v6_0_gpu_init(struct amdgpu_device *adev)
 	gfx_v6_0_get_cu_info(adev);
 	gfx_v6_0_config_init(adev);
 
-	WREG32(mmCP_QUEUE_THRESHOLDS, ((0x16 << CP_QUEUE_THRESHOLDS__ROQ_IB1_START__SHIFT) |
-				       (0x2b << CP_QUEUE_THRESHOLDS__ROQ_IB2_START__SHIFT)));
-	WREG32(mmCP_MEQ_THRESHOLDS, (0x30 << CP_MEQ_THRESHOLDS__MEQ1_START__SHIFT) |
-				    (0x60 << CP_MEQ_THRESHOLDS__MEQ2_START__SHIFT));
+	WREG32(mmCP_QUEUE_THRESHOLDS,
+		((0x16 << CP_QUEUE_THRESHOLDS__ROQ_IB1_START__SHIFT) |
+		(0x2b << CP_QUEUE_THRESHOLDS__ROQ_IB2_START__SHIFT)));
+
+	/* set HW defaults for 3D engine */
+	WREG32(mmCP_MEQ_THRESHOLDS,
+		(0x30 << CP_MEQ_THRESHOLDS__MEQ1_START__SHIFT) |
+		(0x60 << CP_MEQ_THRESHOLDS__MEQ2_START__SHIFT));
 
 	sx_debug_1 = RREG32(mmSX_DEBUG_1);
 	WREG32(mmSX_DEBUG_1, sx_debug_1);
@@ -1570,66 +1785,34 @@ static void gfx_v6_0_gpu_init(struct amdgpu_device *adev)
 	udelay(50);
 }
 
-
-static void gfx_v6_0_scratch_init(struct amdgpu_device *adev)
-{
-	adev->gfx.scratch.num_reg = 8;
-	adev->gfx.scratch.reg_base = mmSCRATCH_REG0;
-	adev->gfx.scratch.free_mask = (1u << adev->gfx.scratch.num_reg) - 1;
-}
-
 static int gfx_v6_0_ring_test_ring(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
-	uint32_t scratch;
 	uint32_t tmp = 0;
 	unsigned i;
 	int r;
 
-	r = amdgpu_gfx_scratch_get(adev, &scratch);
-	if (r) {
-		DRM_ERROR("amdgpu: cp failed to get scratch reg (%d).\n", r);
-		return r;
-	}
-	WREG32(scratch, 0xCAFEDEAD);
+	WREG32(mmSCRATCH_REG0, 0xCAFEDEAD);
 
 	r = amdgpu_ring_alloc(ring, 3);
-	if (r) {
-		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n", ring->idx, r);
-		amdgpu_gfx_scratch_free(adev, scratch);
+	if (r)
 		return r;
-	}
+
 	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1));
-	amdgpu_ring_write(ring, (scratch - PACKET3_SET_CONFIG_REG_START));
+	amdgpu_ring_write(ring, mmSCRATCH_REG0 - PACKET3_SET_CONFIG_REG_START);
 	amdgpu_ring_write(ring, 0xDEADBEEF);
 	amdgpu_ring_commit(ring);
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		tmp = RREG32(scratch);
+		tmp = RREG32(mmSCRATCH_REG0);
 		if (tmp == 0xDEADBEEF)
 			break;
-		DRM_UDELAY(1);
-	}
-	if (i < adev->usec_timeout) {
-		DRM_INFO("ring test on %d succeeded in %d usecs\n", ring->idx, i);
-	} else {
-		DRM_ERROR("amdgpu: ring %d test failed (scratch(0x%04X)=0x%08X)\n",
-			  ring->idx, scratch, tmp);
-		r = -EINVAL;
+		udelay(1);
 	}
-	amdgpu_gfx_scratch_free(adev, scratch);
-	return r;
-}
 
-static void gfx_v6_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
-{
-	/* flush hdp cache */
-	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
-	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(1) |
-				 WRITE_DATA_DST_SEL(0)));
-	amdgpu_ring_write(ring, mmHDP_MEM_COHERENCY_FLUSH_CNTL);
-	amdgpu_ring_write(ring, 0);
-	amdgpu_ring_write(ring, 0x1);
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+	return r;
 }
 
 static void gfx_v6_0_ring_emit_vgt_flush(struct amdgpu_ring *ring)
@@ -1639,24 +1822,6 @@ static void gfx_v6_0_ring_emit_vgt_flush(struct amdgpu_ring *ring)
 		EVENT_INDEX(0));
 }
 
-/**
- * gfx_v6_0_ring_emit_hdp_invalidate - emit an hdp invalidate on the cp
- *
- * @adev: amdgpu_device pointer
- * @ridx: amdgpu ring index
- *
- * Emits an hdp invalidate on the cp.
- */
-static void gfx_v6_0_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
-{
-	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
-	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(1) |
-				 WRITE_DATA_DST_SEL(0)));
-	amdgpu_ring_write(ring, mmHDP_DEBUG0);
-	amdgpu_ring_write(ring, 0);
-	amdgpu_ring_write(ring, 0x1);
-}
-
 static void gfx_v6_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
 				     u64 seq, unsigned flags)
 {
@@ -1686,13 +1851,15 @@ static void gfx_v6_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
 }
 
 static void gfx_v6_0_ring_emit_ib(struct amdgpu_ring *ring,
+				  struct amdgpu_job *job,
 				  struct amdgpu_ib *ib,
-				  unsigned vm_id, bool ctx_switch)
+				  uint32_t flags)
 {
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
 	u32 header, control = 0;
 
 	/* insert SWITCH_BUFFER packet before first IB in the ring frame */
-	if (ctx_switch) {
+	if (flags & AMDGPU_HAVE_CTX_SWITCH) {
 		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
 		amdgpu_ring_write(ring, 0);
 	}
@@ -1702,7 +1869,7 @@ static void gfx_v6_0_ring_emit_ib(struct amdgpu_ring *ring,
 	else
 		header = PACKET3(PACKET3_INDIRECT_BUFFER, 2);
 
-	control |= ib->length_dw | (vm_id << 24);
+	control |= ib->length_dw | (vmid << 24);
 
 	amdgpu_ring_write(ring, header);
 	amdgpu_ring_write(ring,
@@ -1718,6 +1885,7 @@ static void gfx_v6_0_ring_emit_ib(struct amdgpu_ring *ring,
  * gfx_v6_0_ring_test_ib - basic ring IB test
  *
  * @ring: amdgpu_ring structure holding ring information
+ * @timeout: timeout value in jiffies, or MAX_SCHEDULE_TIMEOUT
  *
  * Allocate an IB and execute it on the gfx ring (SI).
  * Provides a basic gfx ring test to verify that IBs are working.
@@ -1726,63 +1894,47 @@ static void gfx_v6_0_ring_emit_ib(struct amdgpu_ring *ring,
 static int gfx_v6_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 {
 	struct amdgpu_device *adev = ring->adev;
-	struct amdgpu_ib ib;
 	struct dma_fence *f = NULL;
-	uint32_t scratch;
+	struct amdgpu_ib ib;
 	uint32_t tmp = 0;
 	long r;
 
-	r = amdgpu_gfx_scratch_get(adev, &scratch);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get scratch reg (%ld).\n", r);
-		return r;
-	}
-	WREG32(scratch, 0xCAFEDEAD);
+	WREG32(mmSCRATCH_REG0, 0xCAFEDEAD);
 	memset(&ib, 0, sizeof(ib));
-	r = amdgpu_ib_get(adev, NULL, 256, &ib);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
-		goto err1;
-	}
+	r = amdgpu_ib_get(adev, NULL, 256, AMDGPU_IB_POOL_DIRECT, &ib);
+	if (r)
+		return r;
+
 	ib.ptr[0] = PACKET3(PACKET3_SET_CONFIG_REG, 1);
-	ib.ptr[1] = ((scratch - PACKET3_SET_CONFIG_REG_START));
+	ib.ptr[1] = mmSCRATCH_REG0 - PACKET3_SET_CONFIG_REG_START;
 	ib.ptr[2] = 0xDEADBEEF;
 	ib.length_dw = 3;
 
 	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
 	if (r)
-		goto err2;
+		goto error;
 
 	r = dma_fence_wait_timeout(f, false, timeout);
 	if (r == 0) {
-		DRM_ERROR("amdgpu: IB test timed out\n");
 		r = -ETIMEDOUT;
-		goto err2;
+		goto error;
 	} else if (r < 0) {
-		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
-		goto err2;
+		goto error;
 	}
-	tmp = RREG32(scratch);
-	if (tmp == 0xDEADBEEF) {
-		DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
+	tmp = RREG32(mmSCRATCH_REG0);
+	if (tmp == 0xDEADBEEF)
 		r = 0;
-	} else {
-		DRM_ERROR("amdgpu: ib test failed (scratch(0x%04X)=0x%08X)\n",
-			  scratch, tmp);
+	else
 		r = -EINVAL;
-	}
 
-err2:
-	amdgpu_ib_free(adev, &ib, NULL);
+error:
+	amdgpu_ib_free(&ib, NULL);
 	dma_fence_put(f);
-err1:
-	amdgpu_gfx_scratch_free(adev, scratch);
 	return r;
 }
 
 static void gfx_v6_0_cp_gfx_enable(struct amdgpu_device *adev, bool enable)
 {
-	int i;
 	if (enable) {
 		WREG32(mmCP_ME_CNTL, 0);
 	} else {
@@ -1790,10 +1942,6 @@ static void gfx_v6_0_cp_gfx_enable(struct amdgpu_device *adev, bool enable)
 				      CP_ME_CNTL__PFP_HALT_MASK |
 				      CP_ME_CNTL__CE_HALT_MASK));
 		WREG32(mmSCRATCH_UMSK, 0);
-		for (i = 0; i < adev->gfx.num_gfx_rings; i++)
-			adev->gfx.gfx_ring[i].ready = false;
-		for (i = 0; i < adev->gfx.num_compute_rings; i++)
-			adev->gfx.compute_ring[i].ready = false;
 	}
 	udelay(50);
 }
@@ -1952,7 +2100,7 @@ static int gfx_v6_0_cp_gfx_resume(struct amdgpu_device *adev)
 	WREG32(mmCP_RB0_WPTR, ring->wptr);
 
 	/* set the wb address whether it's enabled or not */
-	rptr_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+	rptr_addr = ring->rptr_gpu_addr;
 	WREG32(mmCP_RB0_RPTR_ADDR, lower_32_bits(rptr_addr));
 	WREG32(mmCP_RB0_RPTR_ADDR_HI, upper_32_bits(rptr_addr) & 0xFF);
 
@@ -1965,19 +2113,16 @@ static int gfx_v6_0_cp_gfx_resume(struct amdgpu_device *adev)
 
 	/* start the rings */
 	gfx_v6_0_cp_gfx_start(adev);
-	ring->ready = true;
-	r = amdgpu_ring_test_ring(ring);
-	if (r) {
-		ring->ready = false;
+	r = amdgpu_ring_test_helper(ring);
+	if (r)
 		return r;
-	}
 
 	return 0;
 }
 
 static u64 gfx_v6_0_ring_get_rptr(struct amdgpu_ring *ring)
 {
-	return ring->adev->wb.wb[ring->rptr_offs];
+	return *ring->rptr_cpu_addr;
 }
 
 static u64 gfx_v6_0_ring_get_wptr(struct amdgpu_ring *ring)
@@ -2041,7 +2186,7 @@ static int gfx_v6_0_cp_compute_resume(struct amdgpu_device *adev)
 	ring->wptr = 0;
 	WREG32(mmCP_RB1_WPTR, ring->wptr);
 
-	rptr_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+	rptr_addr = ring->rptr_gpu_addr;
 	WREG32(mmCP_RB1_RPTR_ADDR, lower_32_bits(rptr_addr));
 	WREG32(mmCP_RB1_RPTR_ADDR_HI, upper_32_bits(rptr_addr) & 0xFF);
 
@@ -2060,7 +2205,7 @@ static int gfx_v6_0_cp_compute_resume(struct amdgpu_device *adev)
 	WREG32(mmCP_RB2_CNTL, tmp | CP_RB2_CNTL__RB_RPTR_WR_ENA_MASK);
 	ring->wptr = 0;
 	WREG32(mmCP_RB2_WPTR, ring->wptr);
-	rptr_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+	rptr_addr = ring->rptr_gpu_addr;
 	WREG32(mmCP_RB2_RPTR_ADDR, lower_32_bits(rptr_addr));
 	WREG32(mmCP_RB2_RPTR_ADDR_HI, upper_32_bits(rptr_addr) & 0xFF);
 
@@ -2068,14 +2213,11 @@ static int gfx_v6_0_cp_compute_resume(struct amdgpu_device *adev)
 	WREG32(mmCP_RB2_CNTL, tmp);
 	WREG32(mmCP_RB2_BASE, ring->gpu_addr >> 8);
 
-	adev->gfx.compute_ring[0].ready = false;
-	adev->gfx.compute_ring[1].ready = false;
 
 	for (i = 0; i < 2; i++) {
-		r = amdgpu_ring_test_ring(&adev->gfx.compute_ring[i]);
+		r = amdgpu_ring_test_helper(&adev->gfx.compute_ring[i]);
 		if (r)
 			return r;
-		adev->gfx.compute_ring[i].ready = true;
 	}
 
 	return 0;
@@ -2167,29 +2309,11 @@ static void gfx_v6_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
 }
 
 static void gfx_v6_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
-					unsigned vm_id, uint64_t pd_addr)
+					unsigned vmid, uint64_t pd_addr)
 {
 	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
 
-	/* write new base address */
-	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
-	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(1) |
-				 WRITE_DATA_DST_SEL(0)));
-	if (vm_id < 8) {
-		amdgpu_ring_write(ring, (mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + vm_id ));
-	} else {
-		amdgpu_ring_write(ring, (mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + (vm_id - 8)));
-	}
-	amdgpu_ring_write(ring, 0);
-	amdgpu_ring_write(ring, pd_addr >> 12);
-
-	/* bits 0-15 are the VM contexts0-15 */
-	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
-	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(1) |
-				 WRITE_DATA_DST_SEL(0)));
-	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST);
-	amdgpu_ring_write(ring, 0);
-	amdgpu_ring_write(ring, 1 << vm_id);
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
 
 	/* wait for the invalidate to complete */
 	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
@@ -2214,19 +2338,24 @@ static void gfx_v6_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
 	}
 }
 
-
-static void gfx_v6_0_rlc_fini(struct amdgpu_device *adev)
+static void gfx_v6_0_ring_emit_wreg(struct amdgpu_ring *ring,
+				    uint32_t reg, uint32_t val)
 {
-	amdgpu_bo_free_kernel(&adev->gfx.rlc.save_restore_obj, NULL, NULL);
-	amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj, NULL, NULL);
-	amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj, NULL, NULL);
+	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(usepfp) |
+				 WRITE_DATA_DST_SEL(0)));
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, val);
 }
 
 static int gfx_v6_0_rlc_init(struct amdgpu_device *adev)
 {
 	const u32 *src_ptr;
-	volatile u32 *dst_ptr;
-	u32 dws, i;
+	u32 *dst_ptr;
+	u32 dws;
 	u64 reg_list_mc_addr;
 	const struct cs_section_def *cs_data;
 	int r;
@@ -2241,26 +2370,10 @@ static int gfx_v6_0_rlc_init(struct amdgpu_device *adev)
 	cs_data = adev->gfx.rlc.cs_data;
 
 	if (src_ptr) {
-		/* save restore block */
-		r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
-					      AMDGPU_GEM_DOMAIN_VRAM,
-					      &adev->gfx.rlc.save_restore_obj,
-					      &adev->gfx.rlc.save_restore_gpu_addr,
-					      (void **)&adev->gfx.rlc.sr_ptr);
-		if (r) {
-			dev_warn(adev->dev, "(%d) create RLC sr bo failed\n",
-				 r);
-			gfx_v6_0_rlc_fini(adev);
+		/* init save restore block */
+		r = amdgpu_gfx_rlc_init_sr(adev, dws);
+		if (r)
 			return r;
-		}
-
-		/* write the sr buffer */
-		dst_ptr = adev->gfx.rlc.sr_ptr;
-		for (i = 0; i < adev->gfx.rlc.reg_list_size; i++)
-			dst_ptr[i] = cpu_to_le32(src_ptr[i]);
-
-		amdgpu_bo_kunmap(adev->gfx.rlc.save_restore_obj);
-		amdgpu_bo_unreserve(adev->gfx.rlc.save_restore_obj);
 	}
 
 	if (cs_data) {
@@ -2269,13 +2382,14 @@ static int gfx_v6_0_rlc_init(struct amdgpu_device *adev)
 		dws = adev->gfx.rlc.clear_state_size + (256 / 4);
 
 		r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
-					      AMDGPU_GEM_DOMAIN_VRAM,
+					      AMDGPU_GEM_DOMAIN_VRAM |
+					      AMDGPU_GEM_DOMAIN_GTT,
 					      &adev->gfx.rlc.clear_state_obj,
 					      &adev->gfx.rlc.clear_state_gpu_addr,
 					      (void **)&adev->gfx.rlc.cs_ptr);
 		if (r) {
 			dev_warn(adev->dev, "(%d) create RLC c bo failed\n", r);
-			gfx_v6_0_rlc_fini(adev);
+			amdgpu_gfx_rlc_fini(adev);
 			return r;
 		}
 
@@ -2298,7 +2412,7 @@ static void gfx_v6_0_enable_lbpw(struct amdgpu_device *adev, bool enable)
 	WREG32_FIELD(RLC_LB_CNTL, LOAD_BALANCE_ENABLE, enable ? 1 : 0);
 
 	if (!enable) {
-		gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+		gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 		WREG32(mmSPI_LB_CU_MASK, 0x00ff);
 	}
 }
@@ -2396,8 +2510,8 @@ static int gfx_v6_0_rlc_resume(struct amdgpu_device *adev)
 	if (!adev->gfx.rlc_fw)
 		return -EINVAL;
 
-	gfx_v6_0_rlc_stop(adev);
-	gfx_v6_0_rlc_reset(adev);
+	adev->gfx.rlc.funcs->stop(adev);
+	adev->gfx.rlc.funcs->reset(adev);
 	gfx_v6_0_init_pg(adev);
 	gfx_v6_0_init_cg(adev);
 
@@ -2425,7 +2539,7 @@ static int gfx_v6_0_rlc_resume(struct amdgpu_device *adev)
 	WREG32(mmRLC_UCODE_ADDR, 0);
 
 	gfx_v6_0_enable_lbpw(adev, gfx_v6_0_lbpw_supported(adev));
-	gfx_v6_0_rlc_start(adev);
+	adev->gfx.rlc.funcs->start(adev);
 
 	return 0;
 }
@@ -2741,65 +2855,23 @@ static u32 gfx_v6_0_get_csb_size(struct amdgpu_device *adev)
 	return count;
 }
 
-static void gfx_v6_0_get_csb_buffer(struct amdgpu_device *adev,
-				    volatile u32 *buffer)
+static void gfx_v6_0_get_csb_buffer(struct amdgpu_device *adev, u32 *buffer)
 {
-	u32 count = 0, i;
-	const struct cs_section_def *sect = NULL;
-	const struct cs_extent_def *ext = NULL;
+	u32 count = 0;
 
 	if (adev->gfx.rlc.cs_data == NULL)
 		return;
 	if (buffer == NULL)
 		return;
 
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
-	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CONTEXT_CONTROL, 1));
-	buffer[count++] = cpu_to_le32(0x80000000);
-	buffer[count++] = cpu_to_le32(0x80000000);
-
-	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
-		for (ext = sect->section; ext->extent != NULL; ++ext) {
-			if (sect->id == SECT_CONTEXT) {
-				buffer[count++] =
-					cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, ext->reg_count));
-				buffer[count++] = cpu_to_le32(ext->reg_index - 0xa000);
-				for (i = 0; i < ext->reg_count; i++)
-					buffer[count++] = cpu_to_le32(ext->extent[i]);
-			} else {
-				return;
-			}
-		}
-	}
+	count = amdgpu_gfx_csb_preamble_start(buffer);
+	count = amdgpu_gfx_csb_data_parser(adev, buffer, count);
 
 	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, 1));
 	buffer[count++] = cpu_to_le32(mmPA_SC_RASTER_CONFIG - PACKET3_SET_CONTEXT_REG_START);
+	buffer[count++] = cpu_to_le32(adev->gfx.config.rb_config[0][0].raster_config);
 
-	switch (adev->asic_type) {
-	case CHIP_TAHITI:
-	case CHIP_PITCAIRN:
-		buffer[count++] = cpu_to_le32(0x2a00126a);
-		break;
-	case CHIP_VERDE:
-		buffer[count++] = cpu_to_le32(0x0000124a);
-		break;
-	case CHIP_OLAND:
-		buffer[count++] = cpu_to_le32(0x00000082);
-		break;
-	case CHIP_HAINAN:
-		buffer[count++] = cpu_to_le32(0x00000000);
-		break;
-	default:
-		buffer[count++] = cpu_to_le32(0x00000000);
-		break;
-	}
-
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
-	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_END_CLEAR_STATE);
-
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CLEAR_STATE, 0));
-	buffer[count++] = cpu_to_le32(0);
+	amdgpu_gfx_csb_preamble_end(buffer, count);
 }
 
 static void gfx_v6_0_init_pg(struct amdgpu_device *adev)
@@ -2893,7 +2965,7 @@ static void wave_read_regs(struct amdgpu_device *adev, uint32_t simd,
 		*(out++) = RREG32(mmSQ_IND_DATA);
 }
 
-static void gfx_v6_0_read_wave_data(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t *dst, int *no_fields)
+static void gfx_v6_0_read_wave_data(struct amdgpu_device *adev, uint32_t xcc_id, uint32_t simd, uint32_t wave, uint32_t *dst, int *no_fields)
 {
 	/* type 0 wave data */
 	dst[(*no_fields)++] = 0;
@@ -2915,9 +2987,10 @@ static void gfx_v6_0_read_wave_data(struct amdgpu_device *adev, uint32_t simd, u
 	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TMA_HI);
 	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_IB_DBG0);
 	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_M0);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_MODE);
 }
 
-static void gfx_v6_0_read_wave_sgprs(struct amdgpu_device *adev, uint32_t simd,
+static void gfx_v6_0_read_wave_sgprs(struct amdgpu_device *adev, uint32_t xcc_id, uint32_t simd,
 				     uint32_t wave, uint32_t start,
 				     uint32_t size, uint32_t *dst)
 {
@@ -2926,53 +2999,69 @@ static void gfx_v6_0_read_wave_sgprs(struct amdgpu_device *adev, uint32_t simd,
 		start + SQIND_WAVE_SGPRS_OFFSET, size, dst);
 }
 
+static void gfx_v6_0_select_me_pipe_q(struct amdgpu_device *adev,
+				  u32 me, u32 pipe, u32 q, u32 vm, u32 xcc_id)
+{
+	DRM_INFO("Not implemented\n");
+}
+
 static const struct amdgpu_gfx_funcs gfx_v6_0_gfx_funcs = {
 	.get_gpu_clock_counter = &gfx_v6_0_get_gpu_clock_counter,
 	.select_se_sh = &gfx_v6_0_select_se_sh,
 	.read_wave_data = &gfx_v6_0_read_wave_data,
 	.read_wave_sgprs = &gfx_v6_0_read_wave_sgprs,
+	.select_me_pipe_q = &gfx_v6_0_select_me_pipe_q
+};
+
+static const struct amdgpu_rlc_funcs gfx_v6_0_rlc_funcs = {
+	.init = gfx_v6_0_rlc_init,
+	.resume = gfx_v6_0_rlc_resume,
+	.stop = gfx_v6_0_rlc_stop,
+	.reset = gfx_v6_0_rlc_reset,
+	.start = gfx_v6_0_rlc_start
 };
 
-static int gfx_v6_0_early_init(void *handle)
+static int gfx_v6_0_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
+	adev->gfx.xcc_mask = 1;
 	adev->gfx.num_gfx_rings = GFX6_NUM_GFX_RINGS;
-	adev->gfx.num_compute_rings = GFX6_NUM_COMPUTE_RINGS;
+	adev->gfx.num_compute_rings = min(amdgpu_gfx_get_num_kcq(adev),
+					  GFX6_NUM_COMPUTE_RINGS);
 	adev->gfx.funcs = &gfx_v6_0_gfx_funcs;
+	adev->gfx.rlc.funcs = &gfx_v6_0_rlc_funcs;
 	gfx_v6_0_set_ring_funcs(adev);
 	gfx_v6_0_set_irq_funcs(adev);
 
 	return 0;
 }
 
-static int gfx_v6_0_sw_init(void *handle)
+static int gfx_v6_0_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_ring *ring;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int i, r;
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 181, &adev->gfx.eop_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 181, &adev->gfx.eop_irq);
 	if (r)
 		return r;
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 184, &adev->gfx.priv_reg_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 184, &adev->gfx.priv_reg_irq);
 	if (r)
 		return r;
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 185, &adev->gfx.priv_inst_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 185, &adev->gfx.priv_inst_irq);
 	if (r)
 		return r;
 
-	gfx_v6_0_scratch_init(adev);
-
 	r = gfx_v6_0_init_microcode(adev);
 	if (r) {
 		DRM_ERROR("Failed to load gfx firmware!\n");
 		return r;
 	}
 
-	r = gfx_v6_0_rlc_init(adev);
+	r = adev->gfx.rlc.funcs->init(adev);
 	if (r) {
 		DRM_ERROR("Failed to init rlc BOs!\n");
 		return r;
@@ -2982,8 +3071,10 @@ static int gfx_v6_0_sw_init(void *handle)
 		ring = &adev->gfx.gfx_ring[i];
 		ring->ring_obj = NULL;
 		sprintf(ring->name, "gfx");
-		r = amdgpu_ring_init(adev, ring, 1024,
-				     &adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_EOP);
+		r = amdgpu_ring_init(adev, ring, 2048,
+				     &adev->gfx.eop_irq,
+				     AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP,
+				     AMDGPU_RING_PRIO_DEFAULT, NULL);
 		if (r)
 			return r;
 	}
@@ -3005,7 +3096,8 @@ static int gfx_v6_0_sw_init(void *handle)
 		sprintf(ring->name, "comp_%d.%d.%d", ring->me, ring->pipe, ring->queue);
 		irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP + ring->pipe;
 		r = amdgpu_ring_init(adev, ring, 1024,
-				     &adev->gfx.eop_irq, irq_type);
+				     &adev->gfx.eop_irq, irq_type,
+				     AMDGPU_RING_PRIO_DEFAULT, NULL);
 		if (r)
 			return r;
 	}
@@ -3013,29 +3105,29 @@ static int gfx_v6_0_sw_init(void *handle)
 	return r;
 }
 
-static int gfx_v6_0_sw_fini(void *handle)
+static int gfx_v6_0_sw_fini(struct amdgpu_ip_block *ip_block)
 {
 	int i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
 		amdgpu_ring_fini(&adev->gfx.gfx_ring[i]);
 	for (i = 0; i < adev->gfx.num_compute_rings; i++)
 		amdgpu_ring_fini(&adev->gfx.compute_ring[i]);
 
-	gfx_v6_0_rlc_fini(adev);
+	amdgpu_gfx_rlc_fini(adev);
 
 	return 0;
 }
 
-static int gfx_v6_0_hw_init(void *handle)
+static int gfx_v6_0_hw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	gfx_v6_0_gpu_init(adev);
+	gfx_v6_0_constants_init(adev);
 
-	r = gfx_v6_0_rlc_resume(adev);
+	r = adev->gfx.rlc.funcs->resume(adev);
 	if (r)
 		return r;
 
@@ -3048,34 +3140,30 @@ static int gfx_v6_0_hw_init(void *handle)
 	return r;
 }
 
-static int gfx_v6_0_hw_fini(void *handle)
+static int gfx_v6_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	gfx_v6_0_cp_enable(adev, false);
-	gfx_v6_0_rlc_stop(adev);
+	adev->gfx.rlc.funcs->stop(adev);
 	gfx_v6_0_fini_pg(adev);
 
 	return 0;
 }
 
-static int gfx_v6_0_suspend(void *handle)
+static int gfx_v6_0_suspend(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return gfx_v6_0_hw_fini(adev);
+	return gfx_v6_0_hw_fini(ip_block);
 }
 
-static int gfx_v6_0_resume(void *handle)
+static int gfx_v6_0_resume(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return gfx_v6_0_hw_init(adev);
+	return gfx_v6_0_hw_init(ip_block);
 }
 
-static bool gfx_v6_0_is_idle(void *handle)
+static bool gfx_v6_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (RREG32(mmGRBM_STATUS) & GRBM_STATUS__GUI_ACTIVE_MASK)
 		return false;
@@ -3083,24 +3171,19 @@ static bool gfx_v6_0_is_idle(void *handle)
 		return true;
 }
 
-static int gfx_v6_0_wait_for_idle(void *handle)
+static int gfx_v6_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
 	unsigned i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		if (gfx_v6_0_is_idle(handle))
+		if (gfx_v6_0_is_idle(ip_block))
 			return 0;
 		udelay(1);
 	}
 	return -ETIMEDOUT;
 }
 
-static int gfx_v6_0_soft_reset(void *handle)
-{
-	return 0;
-}
-
 static void gfx_v6_0_set_gfx_eop_interrupt_state(struct amdgpu_device *adev,
 						 enum amdgpu_interrupt_state state)
 {
@@ -3218,7 +3301,7 @@ static int gfx_v6_0_set_eop_interrupt_state(struct amdgpu_device *adev,
 					    enum amdgpu_interrupt_state state)
 {
 	switch (type) {
-	case AMDGPU_CP_IRQ_GFX_EOP:
+	case AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP:
 		gfx_v6_0_set_gfx_eop_interrupt_state(adev, state);
 		break;
 	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP:
@@ -3251,12 +3334,31 @@ static int gfx_v6_0_eop_irq(struct amdgpu_device *adev,
 	return 0;
 }
 
+static void gfx_v6_0_fault(struct amdgpu_device *adev,
+			   struct amdgpu_iv_entry *entry)
+{
+	struct amdgpu_ring *ring;
+
+	switch (entry->ring_id) {
+	case 0:
+		ring = &adev->gfx.gfx_ring[0];
+		break;
+	case 1:
+	case 2:
+		ring = &adev->gfx.compute_ring[entry->ring_id - 1];
+		break;
+	default:
+		return;
+	}
+	drm_sched_fault(&ring->sched);
+}
+
 static int gfx_v6_0_priv_reg_irq(struct amdgpu_device *adev,
 				 struct amdgpu_irq_src *source,
 				 struct amdgpu_iv_entry *entry)
 {
 	DRM_ERROR("Illegal register access in command stream\n");
-	schedule_work(&adev->reset_work);
+	gfx_v6_0_fault(adev, entry);
 	return 0;
 }
 
@@ -3265,15 +3367,15 @@ static int gfx_v6_0_priv_inst_irq(struct amdgpu_device *adev,
 				  struct amdgpu_iv_entry *entry)
 {
 	DRM_ERROR("Illegal instruction in command stream\n");
-	schedule_work(&adev->reset_work);
+	gfx_v6_0_fault(adev, entry);
 	return 0;
 }
 
-static int gfx_v6_0_set_clockgating_state(void *handle,
+static int gfx_v6_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
 	bool gate = false;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (state == AMD_CG_STATE_GATE)
 		gate = true;
@@ -3291,11 +3393,11 @@ static int gfx_v6_0_set_clockgating_state(void *handle,
 	return 0;
 }
 
-static int gfx_v6_0_set_powergating_state(void *handle,
+static int gfx_v6_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	bool gate = false;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (state == AMD_PG_STATE_GATE)
 		gate = true;
@@ -3316,10 +3418,21 @@ static int gfx_v6_0_set_powergating_state(void *handle,
 	return 0;
 }
 
+static void gfx_v6_0_emit_mem_sync(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SURFACE_SYNC, 3));
+	amdgpu_ring_write(ring, PACKET3_TCL1_ACTION_ENA |
+			  PACKET3_TC_ACTION_ENA |
+			  PACKET3_SH_KCACHE_ACTION_ENA |
+			  PACKET3_SH_ICACHE_ACTION_ENA);  /* CP_COHER_CNTL */
+	amdgpu_ring_write(ring, 0xffffffff);  /* CP_COHER_SIZE */
+	amdgpu_ring_write(ring, 0);  /* CP_COHER_BASE */
+	amdgpu_ring_write(ring, 0x0000000A); /* poll interval */
+}
+
 static const struct amd_ip_funcs gfx_v6_0_ip_funcs = {
 	.name = "gfx_v6_0",
 	.early_init = gfx_v6_0_early_init,
-	.late_init = NULL,
 	.sw_init = gfx_v6_0_sw_init,
 	.sw_fini = gfx_v6_0_sw_fini,
 	.hw_init = gfx_v6_0_hw_init,
@@ -3328,7 +3441,6 @@ static const struct amd_ip_funcs gfx_v6_0_ip_funcs = {
 	.resume = gfx_v6_0_resume,
 	.is_idle = gfx_v6_0_is_idle,
 	.wait_for_idle = gfx_v6_0_wait_for_idle,
-	.soft_reset = gfx_v6_0_soft_reset,
 	.set_clockgating_state = gfx_v6_0_set_clockgating_state,
 	.set_powergating_state = gfx_v6_0_set_powergating_state,
 };
@@ -3342,23 +3454,23 @@ static const struct amdgpu_ring_funcs gfx_v6_0_ring_funcs_gfx = {
 	.get_wptr = gfx_v6_0_ring_get_wptr,
 	.set_wptr = gfx_v6_0_ring_set_wptr_gfx,
 	.emit_frame_size =
-		5 + /* gfx_v6_0_ring_emit_hdp_flush */
-		5 + /* gfx_v6_0_ring_emit_hdp_invalidate */
+		5 + 5 + /* hdp flush / invalidate */
 		14 + 14 + 14 + /* gfx_v6_0_ring_emit_fence x3 for user fence, vm fence */
 		7 + 4 + /* gfx_v6_0_ring_emit_pipeline_sync */
-		17 + 6 + /* gfx_v6_0_ring_emit_vm_flush */
-		3 + 2, /* gfx_v6_ring_emit_cntxcntl including vgt flush */
+		SI_FLUSH_GPU_TLB_NUM_WREG * 5 + 7 + 6 + /* gfx_v6_0_ring_emit_vm_flush */
+		3 + 2 + /* gfx_v6_ring_emit_cntxcntl including vgt flush */
+		5, /* SURFACE_SYNC */
 	.emit_ib_size = 6, /* gfx_v6_0_ring_emit_ib */
 	.emit_ib = gfx_v6_0_ring_emit_ib,
 	.emit_fence = gfx_v6_0_ring_emit_fence,
 	.emit_pipeline_sync = gfx_v6_0_ring_emit_pipeline_sync,
 	.emit_vm_flush = gfx_v6_0_ring_emit_vm_flush,
-	.emit_hdp_flush = gfx_v6_0_ring_emit_hdp_flush,
-	.emit_hdp_invalidate = gfx_v6_0_ring_emit_hdp_invalidate,
 	.test_ring = gfx_v6_0_ring_test_ring,
 	.test_ib = gfx_v6_0_ring_test_ib,
 	.insert_nop = amdgpu_ring_insert_nop,
 	.emit_cntxcntl = gfx_v6_ring_emit_cntxcntl,
+	.emit_wreg = gfx_v6_0_ring_emit_wreg,
+	.emit_mem_sync = gfx_v6_0_emit_mem_sync,
 };
 
 static const struct amdgpu_ring_funcs gfx_v6_0_ring_funcs_compute = {
@@ -3369,21 +3481,21 @@ static const struct amdgpu_ring_funcs gfx_v6_0_ring_funcs_compute = {
 	.get_wptr = gfx_v6_0_ring_get_wptr,
 	.set_wptr = gfx_v6_0_ring_set_wptr_compute,
 	.emit_frame_size =
-		5 + /* gfx_v6_0_ring_emit_hdp_flush */
-		5 + /* gfx_v6_0_ring_emit_hdp_invalidate */
+		5 + 5 + /* hdp flush / invalidate */
 		7 + /* gfx_v6_0_ring_emit_pipeline_sync */
-		17 + /* gfx_v6_0_ring_emit_vm_flush */
-		14 + 14 + 14, /* gfx_v6_0_ring_emit_fence x3 for user fence, vm fence */
+		SI_FLUSH_GPU_TLB_NUM_WREG * 5 + 7 + /* gfx_v6_0_ring_emit_vm_flush */
+		14 + 14 + 14 + /* gfx_v6_0_ring_emit_fence x3 for user fence, vm fence */
+		5, /* SURFACE_SYNC */
 	.emit_ib_size = 6, /* gfx_v6_0_ring_emit_ib */
 	.emit_ib = gfx_v6_0_ring_emit_ib,
 	.emit_fence = gfx_v6_0_ring_emit_fence,
 	.emit_pipeline_sync = gfx_v6_0_ring_emit_pipeline_sync,
 	.emit_vm_flush = gfx_v6_0_ring_emit_vm_flush,
-	.emit_hdp_flush = gfx_v6_0_ring_emit_hdp_flush,
-	.emit_hdp_invalidate = gfx_v6_0_ring_emit_hdp_invalidate,
 	.test_ring = gfx_v6_0_ring_test_ring,
 	.test_ib = gfx_v6_0_ring_test_ib,
 	.insert_nop = amdgpu_ring_insert_nop,
+	.emit_wreg = gfx_v6_0_ring_emit_wreg,
+	.emit_mem_sync = gfx_v6_0_emit_mem_sync,
 };
 
 static void gfx_v6_0_set_ring_funcs(struct amdgpu_device *adev)
@@ -3446,12 +3558,12 @@ static void gfx_v6_0_get_cu_info(struct amdgpu_device *adev)
 			mask = 1;
 			ao_bitmap = 0;
 			counter = 0;
-			gfx_v6_0_select_se_sh(adev, i, j, 0xffffffff);
+			gfx_v6_0_select_se_sh(adev, i, j, 0xffffffff, 0);
 			if (i < 4 && j < 2)
 				gfx_v6_0_set_user_cu_inactive_bitmap(
 					adev, disable_masks[i * 2 + j]);
 			bitmap = gfx_v6_0_get_cu_enabled(adev);
-			cu_info->bitmap[i][j] = bitmap;
+			cu_info->bitmap[0][i][j] = bitmap;
 
 			for (k = 0; k < adev->gfx.config.max_cu_per_sh; k++) {
 				if (bitmap & mask) {
@@ -3468,7 +3580,7 @@ static void gfx_v6_0_get_cu_info(struct amdgpu_device *adev)
 		}
 	}
 
-	gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 	mutex_unlock(&adev->grbm_idx_mutex);
 
 	cu_info->number = active_cu_number;
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
index 00868764a0dd..5976ed55d9db 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
@@ -20,8 +20,10 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  *
  */
+
 #include <linux/firmware.h>
-#include <drm/drmP.h>
+#include <linux/module.h>
+
 #include "amdgpu.h"
 #include "amdgpu_ih.h"
 #include "amdgpu_gfx.h"
@@ -48,46 +50,50 @@
 #include "oss/oss_2_0_d.h"
 #include "oss/oss_2_0_sh_mask.h"
 
+#define NUM_SIMD_PER_CU 0x4 /* missing from the gfx_7 IP headers */
+
 #define GFX7_NUM_GFX_RINGS     1
 #define GFX7_MEC_HPD_SIZE      2048
 
+#define BONAIRE_GB_ADDR_CONFIG_GOLDEN        0x12010001
+#define HAWAII_GB_ADDR_CONFIG_GOLDEN         0x12011003
+
 static void gfx_v7_0_set_ring_funcs(struct amdgpu_device *adev);
 static void gfx_v7_0_set_irq_funcs(struct amdgpu_device *adev);
 static void gfx_v7_0_set_gds_init(struct amdgpu_device *adev);
 
-MODULE_FIRMWARE("radeon/bonaire_pfp.bin");
-MODULE_FIRMWARE("radeon/bonaire_me.bin");
-MODULE_FIRMWARE("radeon/bonaire_ce.bin");
-MODULE_FIRMWARE("radeon/bonaire_rlc.bin");
-MODULE_FIRMWARE("radeon/bonaire_mec.bin");
-
-MODULE_FIRMWARE("radeon/hawaii_pfp.bin");
-MODULE_FIRMWARE("radeon/hawaii_me.bin");
-MODULE_FIRMWARE("radeon/hawaii_ce.bin");
-MODULE_FIRMWARE("radeon/hawaii_rlc.bin");
-MODULE_FIRMWARE("radeon/hawaii_mec.bin");
-
-MODULE_FIRMWARE("radeon/kaveri_pfp.bin");
-MODULE_FIRMWARE("radeon/kaveri_me.bin");
-MODULE_FIRMWARE("radeon/kaveri_ce.bin");
-MODULE_FIRMWARE("radeon/kaveri_rlc.bin");
-MODULE_FIRMWARE("radeon/kaveri_mec.bin");
-MODULE_FIRMWARE("radeon/kaveri_mec2.bin");
-
-MODULE_FIRMWARE("radeon/kabini_pfp.bin");
-MODULE_FIRMWARE("radeon/kabini_me.bin");
-MODULE_FIRMWARE("radeon/kabini_ce.bin");
-MODULE_FIRMWARE("radeon/kabini_rlc.bin");
-MODULE_FIRMWARE("radeon/kabini_mec.bin");
-
-MODULE_FIRMWARE("radeon/mullins_pfp.bin");
-MODULE_FIRMWARE("radeon/mullins_me.bin");
-MODULE_FIRMWARE("radeon/mullins_ce.bin");
-MODULE_FIRMWARE("radeon/mullins_rlc.bin");
-MODULE_FIRMWARE("radeon/mullins_mec.bin");
-
-static const struct amdgpu_gds_reg_offset amdgpu_gds_reg_offset[] =
-{
+MODULE_FIRMWARE("amdgpu/bonaire_pfp.bin");
+MODULE_FIRMWARE("amdgpu/bonaire_me.bin");
+MODULE_FIRMWARE("amdgpu/bonaire_ce.bin");
+MODULE_FIRMWARE("amdgpu/bonaire_rlc.bin");
+MODULE_FIRMWARE("amdgpu/bonaire_mec.bin");
+
+MODULE_FIRMWARE("amdgpu/hawaii_pfp.bin");
+MODULE_FIRMWARE("amdgpu/hawaii_me.bin");
+MODULE_FIRMWARE("amdgpu/hawaii_ce.bin");
+MODULE_FIRMWARE("amdgpu/hawaii_rlc.bin");
+MODULE_FIRMWARE("amdgpu/hawaii_mec.bin");
+
+MODULE_FIRMWARE("amdgpu/kaveri_pfp.bin");
+MODULE_FIRMWARE("amdgpu/kaveri_me.bin");
+MODULE_FIRMWARE("amdgpu/kaveri_ce.bin");
+MODULE_FIRMWARE("amdgpu/kaveri_rlc.bin");
+MODULE_FIRMWARE("amdgpu/kaveri_mec.bin");
+MODULE_FIRMWARE("amdgpu/kaveri_mec2.bin");
+
+MODULE_FIRMWARE("amdgpu/kabini_pfp.bin");
+MODULE_FIRMWARE("amdgpu/kabini_me.bin");
+MODULE_FIRMWARE("amdgpu/kabini_ce.bin");
+MODULE_FIRMWARE("amdgpu/kabini_rlc.bin");
+MODULE_FIRMWARE("amdgpu/kabini_mec.bin");
+
+MODULE_FIRMWARE("amdgpu/mullins_pfp.bin");
+MODULE_FIRMWARE("amdgpu/mullins_me.bin");
+MODULE_FIRMWARE("amdgpu/mullins_ce.bin");
+MODULE_FIRMWARE("amdgpu/mullins_rlc.bin");
+MODULE_FIRMWARE("amdgpu/mullins_mec.bin");
+
+static const struct amdgpu_gds_reg_offset amdgpu_gds_reg_offset[] = {
 	{mmGDS_VMID0_BASE, mmGDS_VMID0_SIZE, mmGDS_GWS_VMID0, mmGDS_OA_VMID0},
 	{mmGDS_VMID1_BASE, mmGDS_VMID1_SIZE, mmGDS_GWS_VMID1, mmGDS_OA_VMID1},
 	{mmGDS_VMID2_BASE, mmGDS_VMID2_SIZE, mmGDS_GWS_VMID2, mmGDS_OA_VMID2},
@@ -106,8 +112,7 @@ static const struct amdgpu_gds_reg_offset amdgpu_gds_reg_offset[] =
 	{mmGDS_VMID15_BASE, mmGDS_VMID15_SIZE, mmGDS_GWS_VMID15, mmGDS_OA_VMID15}
 };
 
-static const u32 spectre_rlc_save_restore_register_list[] =
-{
+static const u32 spectre_rlc_save_restore_register_list[] = {
 	(0x0e00 << 16) | (0xc12c >> 2),
 	0x00000000,
 	(0x0e00 << 16) | (0xc140 >> 2),
@@ -553,8 +558,7 @@ static const u32 spectre_rlc_save_restore_register_list[] =
 	(0x0e00 << 16) | (0x9600 >> 2),
 };
 
-static const u32 kalindi_rlc_save_restore_register_list[] =
-{
+static const u32 kalindi_rlc_save_restore_register_list[] = {
 	(0x0e00 << 16) | (0xc12c >> 2),
 	0x00000000,
 	(0x0e00 << 16) | (0xc140 >> 2),
@@ -879,11 +883,20 @@ static const u32 kalindi_rlc_save_restore_register_list[] =
 };
 
 static u32 gfx_v7_0_get_csb_size(struct amdgpu_device *adev);
-static void gfx_v7_0_get_csb_buffer(struct amdgpu_device *adev, volatile u32 *buffer);
-static void gfx_v7_0_init_cp_pg_table(struct amdgpu_device *adev);
+static void gfx_v7_0_get_csb_buffer(struct amdgpu_device *adev, u32 *buffer);
 static void gfx_v7_0_init_pg(struct amdgpu_device *adev);
 static void gfx_v7_0_get_cu_info(struct amdgpu_device *adev);
 
+static void gfx_v7_0_free_microcode(struct amdgpu_device *adev)
+{
+	amdgpu_ucode_release(&adev->gfx.pfp_fw);
+	amdgpu_ucode_release(&adev->gfx.me_fw);
+	amdgpu_ucode_release(&adev->gfx.ce_fw);
+	amdgpu_ucode_release(&adev->gfx.mec_fw);
+	amdgpu_ucode_release(&adev->gfx.mec2_fw);
+	amdgpu_ucode_release(&adev->gfx.rlc_fw);
+}
+
 /*
  * Core functions
  */
@@ -899,7 +912,6 @@ static void gfx_v7_0_get_cu_info(struct amdgpu_device *adev);
 static int gfx_v7_0_init_microcode(struct amdgpu_device *adev)
 {
 	const char *chip_name;
-	char fw_name[30];
 	int err;
 
 	DRM_DEBUG("\n");
@@ -920,92 +932,53 @@ static int gfx_v7_0_init_microcode(struct amdgpu_device *adev)
 	case CHIP_MULLINS:
 		chip_name = "mullins";
 		break;
-	default: BUG();
+	default:
+		BUG();
 	}
 
-	snprintf(fw_name, sizeof(fw_name), "radeon/%s_pfp.bin", chip_name);
-	err = request_firmware(&adev->gfx.pfp_fw, fw_name, adev->dev);
-	if (err)
-		goto out;
-	err = amdgpu_ucode_validate(adev->gfx.pfp_fw);
+	err = amdgpu_ucode_request(adev, &adev->gfx.pfp_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_pfp.bin", chip_name);
 	if (err)
 		goto out;
 
-	snprintf(fw_name, sizeof(fw_name), "radeon/%s_me.bin", chip_name);
-	err = request_firmware(&adev->gfx.me_fw, fw_name, adev->dev);
-	if (err)
-		goto out;
-	err = amdgpu_ucode_validate(adev->gfx.me_fw);
+	err = amdgpu_ucode_request(adev, &adev->gfx.me_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_me.bin", chip_name);
 	if (err)
 		goto out;
 
-	snprintf(fw_name, sizeof(fw_name), "radeon/%s_ce.bin", chip_name);
-	err = request_firmware(&adev->gfx.ce_fw, fw_name, adev->dev);
-	if (err)
-		goto out;
-	err = amdgpu_ucode_validate(adev->gfx.ce_fw);
+	err = amdgpu_ucode_request(adev, &adev->gfx.ce_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_ce.bin", chip_name);
 	if (err)
 		goto out;
 
-	snprintf(fw_name, sizeof(fw_name), "radeon/%s_mec.bin", chip_name);
-	err = request_firmware(&adev->gfx.mec_fw, fw_name, adev->dev);
-	if (err)
-		goto out;
-	err = amdgpu_ucode_validate(adev->gfx.mec_fw);
+	err = amdgpu_ucode_request(adev, &adev->gfx.mec_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_mec.bin", chip_name);
 	if (err)
 		goto out;
 
 	if (adev->asic_type == CHIP_KAVERI) {
-		snprintf(fw_name, sizeof(fw_name), "radeon/%s_mec2.bin", chip_name);
-		err = request_firmware(&adev->gfx.mec2_fw, fw_name, adev->dev);
-		if (err)
-			goto out;
-		err = amdgpu_ucode_validate(adev->gfx.mec2_fw);
+		err = amdgpu_ucode_request(adev, &adev->gfx.mec2_fw,
+					   AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_mec2.bin", chip_name);
 		if (err)
 			goto out;
 	}
 
-	snprintf(fw_name, sizeof(fw_name), "radeon/%s_rlc.bin", chip_name);
-	err = request_firmware(&adev->gfx.rlc_fw, fw_name, adev->dev);
-	if (err)
-		goto out;
-	err = amdgpu_ucode_validate(adev->gfx.rlc_fw);
-
+	err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_rlc.bin", chip_name);
 out:
 	if (err) {
-		pr_err("gfx7: Failed to load firmware \"%s\"\n", fw_name);
-		release_firmware(adev->gfx.pfp_fw);
-		adev->gfx.pfp_fw = NULL;
-		release_firmware(adev->gfx.me_fw);
-		adev->gfx.me_fw = NULL;
-		release_firmware(adev->gfx.ce_fw);
-		adev->gfx.ce_fw = NULL;
-		release_firmware(adev->gfx.mec_fw);
-		adev->gfx.mec_fw = NULL;
-		release_firmware(adev->gfx.mec2_fw);
-		adev->gfx.mec2_fw = NULL;
-		release_firmware(adev->gfx.rlc_fw);
-		adev->gfx.rlc_fw = NULL;
+		pr_err("gfx7: Failed to load firmware %s gfx firmware\n", chip_name);
+		gfx_v7_0_free_microcode(adev);
 	}
 	return err;
 }
 
-static void gfx_v7_0_free_microcode(struct amdgpu_device *adev)
-{
-	release_firmware(adev->gfx.pfp_fw);
-	adev->gfx.pfp_fw = NULL;
-	release_firmware(adev->gfx.me_fw);
-	adev->gfx.me_fw = NULL;
-	release_firmware(adev->gfx.ce_fw);
-	adev->gfx.ce_fw = NULL;
-	release_firmware(adev->gfx.mec_fw);
-	adev->gfx.mec_fw = NULL;
-	release_firmware(adev->gfx.mec2_fw);
-	adev->gfx.mec2_fw = NULL;
-	release_firmware(adev->gfx.rlc_fw);
-	adev->gfx.rlc_fw = NULL;
-}
-
 /**
  * gfx_v7_0_tiling_mode_table_init - init the hw tiling table
  *
@@ -1577,13 +1550,14 @@ static void gfx_v7_0_tiling_mode_table_init(struct amdgpu_device *adev)
  * @adev: amdgpu_device pointer
  * @se_num: shader engine to address
  * @sh_num: sh block to address
- *
- * Select which SE, SH combinations to address. Certain
- * registers are instanced per SE or SH.  0xffffffff means
- * broadcast to all SEs or SHs (CIK).
+ * @instance: Certain registers are instanced per SE or SH.
+ *            0xffffffff means broadcast to all SEs or SHs (CIK).
+ * @xcc_id: xcc accelerated compute core id
+ * Select which SE, SH combinations to address.
  */
 static void gfx_v7_0_select_se_sh(struct amdgpu_device *adev,
-				  u32 se_num, u32 sh_num, u32 instance)
+				  u32 se_num, u32 sh_num, u32 instance,
+				  int xcc_id)
 {
 	u32 data;
 
@@ -1763,21 +1737,19 @@ gfx_v7_0_write_harvested_raster_configs(struct amdgpu_device *adev,
 		}
 
 		/* GRBM_GFX_INDEX has a different offset on CI+ */
-		gfx_v7_0_select_se_sh(adev, se, 0xffffffff, 0xffffffff);
+		gfx_v7_0_select_se_sh(adev, se, 0xffffffff, 0xffffffff, 0);
 		WREG32(mmPA_SC_RASTER_CONFIG, raster_config_se);
 		WREG32(mmPA_SC_RASTER_CONFIG_1, raster_config_1);
 	}
 
 	/* GRBM_GFX_INDEX has a different offset on CI+ */
-	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 }
 
 /**
  * gfx_v7_0_setup_rb - setup the RBs on the asic
  *
  * @adev: amdgpu_device pointer
- * @se_num: number of SEs (shader engines) for the asic
- * @sh_per_se: number of SH blocks per SE for the asic
  *
  * Configures per-SE/SH RB registers (CIK).
  */
@@ -1794,13 +1766,13 @@ static void gfx_v7_0_setup_rb(struct amdgpu_device *adev)
 	mutex_lock(&adev->grbm_idx_mutex);
 	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
 		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
-			gfx_v7_0_select_se_sh(adev, i, j, 0xffffffff);
+			gfx_v7_0_select_se_sh(adev, i, j, 0xffffffff, 0);
 			data = gfx_v7_0_get_rb_active_bitmap(adev);
 			active_rbs |= data << ((i * adev->gfx.config.max_sh_per_se + j) *
 					       rb_bitmap_width_per_sh);
 		}
 	}
-	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 
 	adev->gfx.config.backend_enable_mask = active_rbs;
 	adev->gfx.config.num_rbs = hweight32(active_rbs);
@@ -1819,9 +1791,26 @@ static void gfx_v7_0_setup_rb(struct amdgpu_device *adev)
 							adev->gfx.config.backend_enable_mask,
 							num_rb_pipes);
 	}
+
+	/* cache the values for userspace */
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			gfx_v7_0_select_se_sh(adev, i, j, 0xffffffff, 0);
+			adev->gfx.config.rb_config[i][j].rb_backend_disable =
+				RREG32(mmCC_RB_BACKEND_DISABLE);
+			adev->gfx.config.rb_config[i][j].user_rb_backend_disable =
+				RREG32(mmGC_USER_RB_BACKEND_DISABLE);
+			adev->gfx.config.rb_config[i][j].raster_config =
+				RREG32(mmPA_SC_RASTER_CONFIG);
+			adev->gfx.config.rb_config[i][j].raster_config_1 =
+				RREG32(mmPA_SC_RASTER_CONFIG_1);
+		}
+	}
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 	mutex_unlock(&adev->grbm_idx_mutex);
 }
 
+#define DEFAULT_SH_MEM_BASES	(0x6000)
 /**
  * gfx_v7_0_init_compute_vmid - gart enable
  *
@@ -1830,9 +1819,6 @@ static void gfx_v7_0_setup_rb(struct amdgpu_device *adev)
  * Initialize compute vmid sh_mem registers
  *
  */
-#define DEFAULT_SH_MEM_BASES	(0x6000)
-#define FIRST_COMPUTE_VMID	(8)
-#define LAST_COMPUTE_VMID	(16)
 static void gfx_v7_0_init_compute_vmid(struct amdgpu_device *adev)
 {
 	int i;
@@ -1850,7 +1836,7 @@ static void gfx_v7_0_init_compute_vmid(struct amdgpu_device *adev)
 			SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT;
 	sh_mem_config |= MTYPE_NONCACHED << SH_MEM_CONFIG__DEFAULT_MTYPE__SHIFT;
 	mutex_lock(&adev->srbm_mutex);
-	for (i = FIRST_COMPUTE_VMID; i < LAST_COMPUTE_VMID; i++) {
+	for (i = adev->vm_manager.first_kfd_vmid; i < AMDGPU_NUM_VMID; i++) {
 		cik_srbm_select(adev, 0, 0, 0, i);
 		/* CP and shaders */
 		WREG32(mmSH_MEM_CONFIG, sh_mem_config);
@@ -1860,6 +1846,33 @@ static void gfx_v7_0_init_compute_vmid(struct amdgpu_device *adev)
 	}
 	cik_srbm_select(adev, 0, 0, 0, 0);
 	mutex_unlock(&adev->srbm_mutex);
+
+	/* Initialize all compute VMIDs to have no GDS, GWS, or OA
+	   access. These should be enabled by FW for target VMIDs. */
+	for (i = adev->vm_manager.first_kfd_vmid; i < AMDGPU_NUM_VMID; i++) {
+		WREG32(amdgpu_gds_reg_offset[i].mem_base, 0);
+		WREG32(amdgpu_gds_reg_offset[i].mem_size, 0);
+		WREG32(amdgpu_gds_reg_offset[i].gws, 0);
+		WREG32(amdgpu_gds_reg_offset[i].oa, 0);
+	}
+}
+
+static void gfx_v7_0_init_gds_vmid(struct amdgpu_device *adev)
+{
+	int vmid;
+
+	/*
+	 * Initialize all compute and user-gfx VMIDs to have no GDS, GWS, or OA
+	 * access. Compute VMIDs should be enabled by FW for target VMIDs,
+	 * the driver can enable them for graphics. VMID0 should maintain
+	 * access so that HWS firmware can save/restore entries.
+	 */
+	for (vmid = 1; vmid < AMDGPU_NUM_VMID; vmid++) {
+		WREG32(amdgpu_gds_reg_offset[vmid].mem_base, 0);
+		WREG32(amdgpu_gds_reg_offset[vmid].mem_size, 0);
+		WREG32(amdgpu_gds_reg_offset[vmid].gws, 0);
+		WREG32(amdgpu_gds_reg_offset[vmid].oa, 0);
+	}
 }
 
 static void gfx_v7_0_config_init(struct amdgpu_device *adev)
@@ -1868,14 +1881,14 @@ static void gfx_v7_0_config_init(struct amdgpu_device *adev)
 }
 
 /**
- * gfx_v7_0_gpu_init - setup the 3D engine
+ * gfx_v7_0_constants_init - setup the 3D engine
  *
  * @adev: amdgpu_device pointer
  *
- * Configures the 3D engine and tiling configuration
- * registers so that the 3D engine is usable.
+ * init the gfx constants such as the 3D engine, tiling configuration
+ * registers, maximum number of quad pipes, render backends...
  */
-static void gfx_v7_0_gpu_init(struct amdgpu_device *adev)
+static void gfx_v7_0_constants_init(struct amdgpu_device *adev)
 {
 	u32 sh_mem_cfg, sh_static_mem_cfg, sh_mem_base;
 	u32 tmp;
@@ -1903,7 +1916,7 @@ static void gfx_v7_0_gpu_init(struct amdgpu_device *adev)
 	 * making sure that the following register writes will be broadcasted
 	 * to all the shaders
 	 */
-	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 
 	/* XXX SH_MEM regs */
 	/* where to put LDS, scratch, GPUVM in FSA64 space */
@@ -1928,7 +1941,7 @@ static void gfx_v7_0_gpu_init(struct amdgpu_device *adev)
 		if (i == 0)
 			sh_mem_base = 0;
 		else
-			sh_mem_base = adev->mc.shared_aperture_start >> 48;
+			sh_mem_base = adev->gmc.shared_aperture_start >> 48;
 		cik_srbm_select(adev, 0, 0, 0, i);
 		/* CP and shaders */
 		WREG32(mmSH_MEM_CONFIG, sh_mem_cfg);
@@ -1940,6 +1953,7 @@ static void gfx_v7_0_gpu_init(struct amdgpu_device *adev)
 	mutex_unlock(&adev->srbm_mutex);
 
 	gfx_v7_0_init_compute_vmid(adev);
+	gfx_v7_0_init_gds_vmid(adev);
 
 	WREG32(mmSX_DEBUG_1, 0x20);
 
@@ -2006,30 +2020,9 @@ static void gfx_v7_0_gpu_init(struct amdgpu_device *adev)
 	udelay(50);
 }
 
-/*
- * GPU scratch registers helpers function.
- */
-/**
- * gfx_v7_0_scratch_init - setup driver info for CP scratch regs
- *
- * @adev: amdgpu_device pointer
- *
- * Set up the number and offset of the CP scratch registers.
- * NOTE: use of CP scratch registers is a legacy inferface and
- * is not used by default on newer asics (r6xx+).  On newer asics,
- * memory buffers are used for fences rather than scratch regs.
- */
-static void gfx_v7_0_scratch_init(struct amdgpu_device *adev)
-{
-	adev->gfx.scratch.num_reg = 8;
-	adev->gfx.scratch.reg_base = mmSCRATCH_REG0;
-	adev->gfx.scratch.free_mask = (1u << adev->gfx.scratch.num_reg) - 1;
-}
-
 /**
  * gfx_v7_0_ring_test_ring - basic gfx ring test
  *
- * @adev: amdgpu_device pointer
  * @ring: amdgpu_ring structure holding ring information
  *
  * Allocate a scratch register and write to it using the gfx ring (CIK).
@@ -2040,50 +2033,35 @@ static void gfx_v7_0_scratch_init(struct amdgpu_device *adev)
 static int gfx_v7_0_ring_test_ring(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
-	uint32_t scratch;
 	uint32_t tmp = 0;
 	unsigned i;
 	int r;
 
-	r = amdgpu_gfx_scratch_get(adev, &scratch);
-	if (r) {
-		DRM_ERROR("amdgpu: cp failed to get scratch reg (%d).\n", r);
-		return r;
-	}
-	WREG32(scratch, 0xCAFEDEAD);
+	WREG32(mmSCRATCH_REG0, 0xCAFEDEAD);
 	r = amdgpu_ring_alloc(ring, 3);
-	if (r) {
-		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n", ring->idx, r);
-		amdgpu_gfx_scratch_free(adev, scratch);
+	if (r)
 		return r;
-	}
+
 	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_UCONFIG_REG, 1));
-	amdgpu_ring_write(ring, (scratch - PACKET3_SET_UCONFIG_REG_START));
+	amdgpu_ring_write(ring, mmSCRATCH_REG0 - PACKET3_SET_UCONFIG_REG_START);
 	amdgpu_ring_write(ring, 0xDEADBEEF);
 	amdgpu_ring_commit(ring);
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		tmp = RREG32(scratch);
+		tmp = RREG32(mmSCRATCH_REG0);
 		if (tmp == 0xDEADBEEF)
 			break;
-		DRM_UDELAY(1);
-	}
-	if (i < adev->usec_timeout) {
-		DRM_INFO("ring test on %d succeeded in %d usecs\n", ring->idx, i);
-	} else {
-		DRM_ERROR("amdgpu: ring %d test failed (scratch(0x%04X)=0x%08X)\n",
-			  ring->idx, scratch, tmp);
-		r = -EINVAL;
+		udelay(1);
 	}
-	amdgpu_gfx_scratch_free(adev, scratch);
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
 	return r;
 }
 
 /**
- * gfx_v7_0_ring_emit_hdp - emit an hdp flush on the cp
+ * gfx_v7_0_ring_emit_hdp_flush - emit an hdp flush on the cp
  *
- * @adev: amdgpu_device pointer
- * @ridx: amdgpu ring index
+ * @ring: amdgpu_ring structure holding ring information
  *
  * Emits an hdp flush on the cp.
  */
@@ -2129,33 +2107,15 @@ static void gfx_v7_0_ring_emit_vgt_flush(struct amdgpu_ring *ring)
 		EVENT_INDEX(0));
 }
 
-
-/**
- * gfx_v7_0_ring_emit_hdp_invalidate - emit an hdp invalidate on the cp
- *
- * @adev: amdgpu_device pointer
- * @ridx: amdgpu ring index
- *
- * Emits an hdp invalidate on the cp.
- */
-static void gfx_v7_0_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
-{
-	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
-	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
-				 WRITE_DATA_DST_SEL(0) |
-				 WR_CONFIRM));
-	amdgpu_ring_write(ring, mmHDP_DEBUG0);
-	amdgpu_ring_write(ring, 0);
-	amdgpu_ring_write(ring, 1);
-}
-
 /**
  * gfx_v7_0_ring_emit_fence_gfx - emit a fence on the gfx ring
  *
- * @adev: amdgpu_device pointer
- * @fence: amdgpu fence object
+ * @ring: amdgpu_ring structure holding ring information
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
  *
- * Emits a fence sequnce number on the gfx ring and flushes
+ * Emits a fence sequence number on the gfx ring and flushes
  * GPU caches.
  */
 static void gfx_v7_0_ring_emit_fence_gfx(struct amdgpu_ring *ring, u64 addr,
@@ -2163,6 +2123,8 @@ static void gfx_v7_0_ring_emit_fence_gfx(struct amdgpu_ring *ring, u64 addr,
 {
 	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
 	bool int_sel = flags & AMDGPU_FENCE_FLAG_INT;
+	bool exec = flags & AMDGPU_FENCE_FLAG_EXEC;
+
 	/* Workaround for cache flush problems. First send a dummy EOP
 	 * event down the pipe with seq one below.
 	 */
@@ -2182,7 +2144,8 @@ static void gfx_v7_0_ring_emit_fence_gfx(struct amdgpu_ring *ring, u64 addr,
 	amdgpu_ring_write(ring, (EOP_TCL1_ACTION_EN |
 				 EOP_TC_ACTION_EN |
 				 EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) |
-				 EVENT_INDEX(5)));
+				 EVENT_INDEX(5) |
+				 (exec ? EOP_EXEC : 0)));
 	amdgpu_ring_write(ring, addr & 0xfffffffc);
 	amdgpu_ring_write(ring, (upper_32_bits(addr) & 0xffff) |
 				DATA_SEL(write64bit ? 2 : 1) | INT_SEL(int_sel ? 2 : 0));
@@ -2193,10 +2156,12 @@ static void gfx_v7_0_ring_emit_fence_gfx(struct amdgpu_ring *ring, u64 addr,
 /**
  * gfx_v7_0_ring_emit_fence_compute - emit a fence on the compute ring
  *
- * @adev: amdgpu_device pointer
- * @fence: amdgpu fence object
+ * @ring: amdgpu_ring structure holding ring information
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
  *
- * Emits a fence sequnce number on the compute ring and flushes
+ * Emits a fence sequence number on the compute ring and flushes
  * GPU caches.
  */
 static void gfx_v7_0_ring_emit_fence_compute(struct amdgpu_ring *ring,
@@ -2223,25 +2188,29 @@ static void gfx_v7_0_ring_emit_fence_compute(struct amdgpu_ring *ring,
  * IB stuff
  */
 /**
- * gfx_v7_0_ring_emit_ib - emit an IB (Indirect Buffer) on the ring
+ * gfx_v7_0_ring_emit_ib_gfx - emit an IB (Indirect Buffer) on the ring
  *
  * @ring: amdgpu_ring structure holding ring information
+ * @job: job to retrieve vmid from
  * @ib: amdgpu indirect buffer object
+ * @flags: options (AMDGPU_HAVE_CTX_SWITCH)
  *
  * Emits an DE (drawing engine) or CE (constant engine) IB
  * on the gfx ring.  IBs are usually generated by userspace
  * acceleration drivers and submitted to the kernel for
- * sheduling on the ring.  This function schedules the IB
+ * scheduling on the ring.  This function schedules the IB
  * on the gfx ring for execution by the GPU.
  */
 static void gfx_v7_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
-				      struct amdgpu_ib *ib,
-				      unsigned vm_id, bool ctx_switch)
+					struct amdgpu_job *job,
+					struct amdgpu_ib *ib,
+					uint32_t flags)
 {
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
 	u32 header, control = 0;
 
 	/* insert SWITCH_BUFFER packet before first IB in the ring frame */
-	if (ctx_switch) {
+	if (flags & AMDGPU_HAVE_CTX_SWITCH) {
 		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
 		amdgpu_ring_write(ring, 0);
 	}
@@ -2251,7 +2220,7 @@ static void gfx_v7_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
 	else
 		header = PACKET3(PACKET3_INDIRECT_BUFFER, 2);
 
-	control |= ib->length_dw | (vm_id << 24);
+	control |= ib->length_dw | (vmid << 24);
 
 	amdgpu_ring_write(ring, header);
 	amdgpu_ring_write(ring,
@@ -2264,10 +2233,28 @@ static void gfx_v7_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
 }
 
 static void gfx_v7_0_ring_emit_ib_compute(struct amdgpu_ring *ring,
+					  struct amdgpu_job *job,
 					  struct amdgpu_ib *ib,
-					  unsigned vm_id, bool ctx_switch)
+					  uint32_t flags)
 {
-	u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vm_id << 24);
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+	u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vmid << 24);
+
+	/* Currently, there is a high possibility to get wave ID mismatch
+	 * between ME and GDS, leading to a hw deadlock, because ME generates
+	 * different wave IDs than the GDS expects. This situation happens
+	 * randomly when at least 5 compute pipes use GDS ordered append.
+	 * The wave IDs generated by ME are also wrong after suspend/resume.
+	 * Those are probably bugs somewhere else in the kernel driver.
+	 *
+	 * Writing GDS_COMPUTE_MAX_WAVE_ID resets wave ID counters in ME and
+	 * GDS to 0 for this ring (me/pipe).
+	 */
+	if (ib->flags & AMDGPU_IB_FLAG_RESET_GDS_MAX_WAVE_ID) {
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1));
+		amdgpu_ring_write(ring, mmGDS_COMPUTE_MAX_WAVE_ID - PACKET3_SET_CONFIG_REG_START);
+		amdgpu_ring_write(ring, ring->adev->gds.gds_compute_max_wave_id);
+	}
 
 	amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2));
 	amdgpu_ring_write(ring,
@@ -2303,6 +2290,7 @@ static void gfx_v7_ring_emit_cntxcntl(struct amdgpu_ring *ring, uint32_t flags)
  * gfx_v7_0_ring_test_ib - basic ring IB test
  *
  * @ring: amdgpu_ring structure holding ring information
+ * @timeout: timeout value in jiffies, or MAX_SCHEDULE_TIMEOUT
  *
  * Allocate an IB and execute it on the gfx ring (CIK).
  * Provides a basic gfx ring test to verify that IBs are working.
@@ -2313,61 +2301,46 @@ static int gfx_v7_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 	struct amdgpu_device *adev = ring->adev;
 	struct amdgpu_ib ib;
 	struct dma_fence *f = NULL;
-	uint32_t scratch;
 	uint32_t tmp = 0;
 	long r;
 
-	r = amdgpu_gfx_scratch_get(adev, &scratch);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get scratch reg (%ld).\n", r);
-		return r;
-	}
-	WREG32(scratch, 0xCAFEDEAD);
+	WREG32(mmSCRATCH_REG0, 0xCAFEDEAD);
 	memset(&ib, 0, sizeof(ib));
-	r = amdgpu_ib_get(adev, NULL, 256, &ib);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
-		goto err1;
-	}
+	r = amdgpu_ib_get(adev, NULL, 256, AMDGPU_IB_POOL_DIRECT, &ib);
+	if (r)
+		return r;
+
 	ib.ptr[0] = PACKET3(PACKET3_SET_UCONFIG_REG, 1);
-	ib.ptr[1] = ((scratch - PACKET3_SET_UCONFIG_REG_START));
+	ib.ptr[1] = mmSCRATCH_REG0 - PACKET3_SET_UCONFIG_REG_START;
 	ib.ptr[2] = 0xDEADBEEF;
 	ib.length_dw = 3;
 
 	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
 	if (r)
-		goto err2;
+		goto error;
 
 	r = dma_fence_wait_timeout(f, false, timeout);
 	if (r == 0) {
-		DRM_ERROR("amdgpu: IB test timed out\n");
 		r = -ETIMEDOUT;
-		goto err2;
+		goto error;
 	} else if (r < 0) {
-		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
-		goto err2;
+		goto error;
 	}
-	tmp = RREG32(scratch);
-	if (tmp == 0xDEADBEEF) {
-		DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
+	tmp = RREG32(mmSCRATCH_REG0);
+	if (tmp == 0xDEADBEEF)
 		r = 0;
-	} else {
-		DRM_ERROR("amdgpu: ib test failed (scratch(0x%04X)=0x%08X)\n",
-			  scratch, tmp);
+	else
 		r = -EINVAL;
-	}
 
-err2:
-	amdgpu_ib_free(adev, &ib, NULL);
+error:
+	amdgpu_ib_free(&ib, NULL);
 	dma_fence_put(f);
-err1:
-	amdgpu_gfx_scratch_free(adev, scratch);
 	return r;
 }
 
 /*
  * CP.
- * On CIK, gfx and compute now have independant command processors.
+ * On CIK, gfx and compute now have independent command processors.
  *
  * GFX
  * Gfx consists of a single ring and can process both gfx jobs and
@@ -2398,15 +2371,12 @@ err1:
  */
 static void gfx_v7_0_cp_gfx_enable(struct amdgpu_device *adev, bool enable)
 {
-	int i;
-
-	if (enable) {
+	if (enable)
 		WREG32(mmCP_ME_CNTL, 0);
-	} else {
-		WREG32(mmCP_ME_CNTL, (CP_ME_CNTL__ME_HALT_MASK | CP_ME_CNTL__PFP_HALT_MASK | CP_ME_CNTL__CE_HALT_MASK));
-		for (i = 0; i < adev->gfx.num_gfx_rings; i++)
-			adev->gfx.gfx_ring[i].ready = false;
-	}
+	else
+		WREG32(mmCP_ME_CNTL, (CP_ME_CNTL__ME_HALT_MASK |
+				      CP_ME_CNTL__PFP_HALT_MASK |
+				      CP_ME_CNTL__CE_HALT_MASK));
 	udelay(50);
 }
 
@@ -2535,29 +2505,8 @@ static int gfx_v7_0_cp_gfx_start(struct amdgpu_device *adev)
 
 	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONTEXT_REG, 2));
 	amdgpu_ring_write(ring, mmPA_SC_RASTER_CONFIG - PACKET3_SET_CONTEXT_REG_START);
-	switch (adev->asic_type) {
-	case CHIP_BONAIRE:
-		amdgpu_ring_write(ring, 0x16000012);
-		amdgpu_ring_write(ring, 0x00000000);
-		break;
-	case CHIP_KAVERI:
-		amdgpu_ring_write(ring, 0x00000000); /* XXX */
-		amdgpu_ring_write(ring, 0x00000000);
-		break;
-	case CHIP_KABINI:
-	case CHIP_MULLINS:
-		amdgpu_ring_write(ring, 0x00000000); /* XXX */
-		amdgpu_ring_write(ring, 0x00000000);
-		break;
-	case CHIP_HAWAII:
-		amdgpu_ring_write(ring, 0x3a00161a);
-		amdgpu_ring_write(ring, 0x0000002e);
-		break;
-	default:
-		amdgpu_ring_write(ring, 0x00000000);
-		amdgpu_ring_write(ring, 0x00000000);
-		break;
-	}
+	amdgpu_ring_write(ring, adev->gfx.config.rb_config[0][0].raster_config);
+	amdgpu_ring_write(ring, adev->gfx.config.rb_config[0][0].raster_config_1);
 
 	amdgpu_ring_write(ring, PACKET3(PACKET3_PREAMBLE_CNTL, 0));
 	amdgpu_ring_write(ring, PACKET3_PREAMBLE_END_CLEAR_STATE);
@@ -2619,8 +2568,8 @@ static int gfx_v7_0_cp_gfx_resume(struct amdgpu_device *adev)
 	ring->wptr = 0;
 	WREG32(mmCP_RB0_WPTR, lower_32_bits(ring->wptr));
 
-	/* set the wb address wether it's enabled or not */
-	rptr_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+	/* set the wb address whether it's enabled or not */
+	rptr_addr = ring->rptr_gpu_addr;
 	WREG32(mmCP_RB0_RPTR_ADDR, lower_32_bits(rptr_addr));
 	WREG32(mmCP_RB0_RPTR_ADDR_HI, upper_32_bits(rptr_addr) & 0xFF);
 
@@ -2636,19 +2585,16 @@ static int gfx_v7_0_cp_gfx_resume(struct amdgpu_device *adev)
 
 	/* start the ring */
 	gfx_v7_0_cp_gfx_start(adev);
-	ring->ready = true;
-	r = amdgpu_ring_test_ring(ring);
-	if (r) {
-		ring->ready = false;
+	r = amdgpu_ring_test_helper(ring);
+	if (r)
 		return r;
-	}
 
 	return 0;
 }
 
 static u64 gfx_v7_0_ring_get_rptr(struct amdgpu_ring *ring)
 {
-	return ring->adev->wb.wb[ring->rptr_offs];
+	return *ring->rptr_cpu_addr;
 }
 
 static u64 gfx_v7_0_ring_get_wptr_gfx(struct amdgpu_ring *ring)
@@ -2669,7 +2615,7 @@ static void gfx_v7_0_ring_set_wptr_gfx(struct amdgpu_ring *ring)
 static u64 gfx_v7_0_ring_get_wptr_compute(struct amdgpu_ring *ring)
 {
 	/* XXX check if swapping is necessary on BE */
-	return ring->adev->wb.wb[ring->wptr_offs];
+	return *ring->wptr_cpu_addr;
 }
 
 static void gfx_v7_0_ring_set_wptr_compute(struct amdgpu_ring *ring)
@@ -2677,7 +2623,7 @@ static void gfx_v7_0_ring_set_wptr_compute(struct amdgpu_ring *ring)
 	struct amdgpu_device *adev = ring->adev;
 
 	/* XXX check if swapping is necessary on BE */
-	adev->wb.wb[ring->wptr_offs] = lower_32_bits(ring->wptr);
+	*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
 	WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
 }
 
@@ -2691,15 +2637,11 @@ static void gfx_v7_0_ring_set_wptr_compute(struct amdgpu_ring *ring)
  */
 static void gfx_v7_0_cp_compute_enable(struct amdgpu_device *adev, bool enable)
 {
-	int i;
-
-	if (enable) {
+	if (enable)
 		WREG32(mmCP_MEC_CNTL, 0);
-	} else {
-		WREG32(mmCP_MEC_CNTL, (CP_MEC_CNTL__MEC_ME1_HALT_MASK | CP_MEC_CNTL__MEC_ME2_HALT_MASK));
-		for (i = 0; i < adev->gfx.num_compute_rings; i++)
-			adev->gfx.compute_ring[i].ready = false;
-	}
+	else
+		WREG32(mmCP_MEC_CNTL, (CP_MEC_CNTL__MEC_ME1_HALT_MASK |
+				       CP_MEC_CNTL__MEC_ME2_HALT_MASK));
 	udelay(50);
 }
 
@@ -2794,7 +2736,7 @@ static int gfx_v7_0_mec_init(struct amdgpu_device *adev)
 	u32 *hpd;
 	size_t mec_hpd_size;
 
-	bitmap_zero(adev->gfx.mec.queue_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
+	bitmap_zero(adev->gfx.mec_bitmap[0].queue_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
 
 	/* take ownership of the relevant compute queues */
 	amdgpu_gfx_compute_queue_acquire(adev);
@@ -2804,6 +2746,7 @@ static int gfx_v7_0_mec_init(struct amdgpu_device *adev)
 		* GFX7_MEC_HPD_SIZE * 2;
 
 	r = amdgpu_bo_create_reserved(adev, mec_hpd_size, PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_VRAM |
 				      AMDGPU_GEM_DOMAIN_GTT,
 				      &adev->gfx.mec.hpd_eop_obj,
 				      &adev->gfx.mec.hpd_eop_gpu_addr,
@@ -2823,45 +2766,6 @@ static int gfx_v7_0_mec_init(struct amdgpu_device *adev)
 	return 0;
 }
 
-struct hqd_registers
-{
-	u32 cp_mqd_base_addr;
-	u32 cp_mqd_base_addr_hi;
-	u32 cp_hqd_active;
-	u32 cp_hqd_vmid;
-	u32 cp_hqd_persistent_state;
-	u32 cp_hqd_pipe_priority;
-	u32 cp_hqd_queue_priority;
-	u32 cp_hqd_quantum;
-	u32 cp_hqd_pq_base;
-	u32 cp_hqd_pq_base_hi;
-	u32 cp_hqd_pq_rptr;
-	u32 cp_hqd_pq_rptr_report_addr;
-	u32 cp_hqd_pq_rptr_report_addr_hi;
-	u32 cp_hqd_pq_wptr_poll_addr;
-	u32 cp_hqd_pq_wptr_poll_addr_hi;
-	u32 cp_hqd_pq_doorbell_control;
-	u32 cp_hqd_pq_wptr;
-	u32 cp_hqd_pq_control;
-	u32 cp_hqd_ib_base_addr;
-	u32 cp_hqd_ib_base_addr_hi;
-	u32 cp_hqd_ib_rptr;
-	u32 cp_hqd_ib_control;
-	u32 cp_hqd_iq_timer;
-	u32 cp_hqd_iq_rptr;
-	u32 cp_hqd_dequeue_request;
-	u32 cp_hqd_dma_offload;
-	u32 cp_hqd_sema_cmd;
-	u32 cp_hqd_msg_type;
-	u32 cp_hqd_atomic0_preop_lo;
-	u32 cp_hqd_atomic0_preop_hi;
-	u32 cp_hqd_atomic1_preop_lo;
-	u32 cp_hqd_atomic1_preop_hi;
-	u32 cp_hqd_hq_scheduler0;
-	u32 cp_hqd_hq_scheduler1;
-	u32 cp_mqd_control;
-};
-
 static void gfx_v7_0_compute_pipe_init(struct amdgpu_device *adev,
 				       int mec, int pipe)
 {
@@ -2977,12 +2881,12 @@ static void gfx_v7_0_mqd_init(struct amdgpu_device *adev,
 		CP_HQD_PQ_CONTROL__KMD_QUEUE_MASK; /* assuming kernel queue control */
 
 	/* only used if CP_PQ_WPTR_POLL_CNTL.CP_PQ_WPTR_POLL_CNTL__EN_MASK=1 */
-	wb_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
+	wb_gpu_addr = ring->wptr_gpu_addr;
 	mqd->cp_hqd_pq_wptr_poll_addr_lo = wb_gpu_addr & 0xfffffffc;
 	mqd->cp_hqd_pq_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr) & 0xffff;
 
-	/* set the wb address wether it's enabled or not */
-	wb_gpu_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+	/* set the wb address whether it's enabled or not */
+	wb_gpu_addr = ring->rptr_gpu_addr;
 	mqd->cp_hqd_pq_rptr_report_addr_lo = wb_gpu_addr & 0xfffffffc;
 	mqd->cp_hqd_pq_rptr_report_addr_hi =
 		upper_32_bits(wb_gpu_addr) & 0xffff;
@@ -3036,7 +2940,7 @@ static void gfx_v7_0_mqd_init(struct amdgpu_device *adev,
 	mqd->cp_hqd_active = 1;
 }
 
-int gfx_v7_0_mqd_commit(struct amdgpu_device *adev, struct cik_mqd *mqd)
+static int gfx_v7_0_mqd_commit(struct amdgpu_device *adev, struct cik_mqd *mqd)
 {
 	uint32_t tmp;
 	uint32_t mqd_reg;
@@ -3129,10 +3033,7 @@ static int gfx_v7_0_cp_compute_resume(struct amdgpu_device *adev)
 
 	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
 		ring = &adev->gfx.compute_ring[i];
-		ring->ready = true;
-		r = amdgpu_ring_test_ring(ring);
-		if (r)
-			ring->ready = false;
+		amdgpu_ring_test_helper(ring);
 	}
 
 	return 0;
@@ -3195,9 +3096,9 @@ static int gfx_v7_0_cp_resume(struct amdgpu_device *adev)
 }
 
 /**
- * gfx_v7_0_ring_emit_vm_flush - cik vm flush using the CP
+ * gfx_v7_0_ring_emit_pipeline_sync - cik vm flush using the CP
  *
- * @ring: the ring to emmit the commands to
+ * @ring: the ring to emit the commands to
  *
  * Sync the command pipeline with the PFP. E.g. wait for everything
  * to be completed.
@@ -3219,7 +3120,7 @@ static void gfx_v7_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
 	amdgpu_ring_write(ring, 4); /* poll interval */
 
 	if (usepfp) {
-		/* synce CE with ME to prevent CE fetch CEIB before context switch done */
+		/* sync CE with ME to prevent CE fetch CEIB before context switch done */
 		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
 		amdgpu_ring_write(ring, 0);
 		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
@@ -3236,36 +3137,19 @@ static void gfx_v7_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
 /**
  * gfx_v7_0_ring_emit_vm_flush - cik vm flush using the CP
  *
- * @adev: amdgpu_device pointer
+ * @ring: amdgpu_ring pointer
+ * @vmid: vmid number to use
+ * @pd_addr: address
  *
  * Update the page table base and flush the VM TLB
  * using the CP (CIK).
  */
 static void gfx_v7_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
-					unsigned vm_id, uint64_t pd_addr)
+					unsigned vmid, uint64_t pd_addr)
 {
 	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
 
-	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
-	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(usepfp) |
-				 WRITE_DATA_DST_SEL(0)));
-	if (vm_id < 8) {
-		amdgpu_ring_write(ring,
-				  (mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + vm_id));
-	} else {
-		amdgpu_ring_write(ring,
-				  (mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vm_id - 8));
-	}
-	amdgpu_ring_write(ring, 0);
-	amdgpu_ring_write(ring, pd_addr >> 12);
-
-	/* bits 0-15 are the VM contexts0-15 */
-	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
-	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
-				 WRITE_DATA_DST_SEL(0)));
-	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST);
-	amdgpu_ring_write(ring, 0);
-	amdgpu_ring_write(ring, 1 << vm_id);
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
 
 	/* wait for the invalidate to complete */
 	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
@@ -3292,23 +3176,28 @@ static void gfx_v7_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
 	}
 }
 
+static void gfx_v7_0_ring_emit_wreg(struct amdgpu_ring *ring,
+				    uint32_t reg, uint32_t val)
+{
+	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(usepfp) |
+				 WRITE_DATA_DST_SEL(0)));
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, val);
+}
+
 /*
  * RLC
  * The RLC is a multi-purpose microengine that handles a
  * variety of functions.
  */
-static void gfx_v7_0_rlc_fini(struct amdgpu_device *adev)
-{
-	amdgpu_bo_free_kernel(&adev->gfx.rlc.save_restore_obj, NULL, NULL);
-	amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj, NULL, NULL);
-	amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj, NULL, NULL);
-}
-
 static int gfx_v7_0_rlc_init(struct amdgpu_device *adev)
 {
 	const u32 *src_ptr;
-	volatile u32 *dst_ptr;
-	u32 dws, i;
+	u32 dws;
 	const struct cs_section_def *cs_data;
 	int r;
 
@@ -3335,68 +3224,29 @@ static int gfx_v7_0_rlc_init(struct amdgpu_device *adev)
 	cs_data = adev->gfx.rlc.cs_data;
 
 	if (src_ptr) {
-		/* save restore block */
-		r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
-					      AMDGPU_GEM_DOMAIN_VRAM,
-					      &adev->gfx.rlc.save_restore_obj,
-					      &adev->gfx.rlc.save_restore_gpu_addr,
-					      (void **)&adev->gfx.rlc.sr_ptr);
-		if (r) {
-			dev_warn(adev->dev, "(%d) create, pin or map of RLC sr bo failed\n", r);
-			gfx_v7_0_rlc_fini(adev);
+		/* init save restore block */
+		r = amdgpu_gfx_rlc_init_sr(adev, dws);
+		if (r)
 			return r;
-		}
-
-		/* write the sr buffer */
-		dst_ptr = adev->gfx.rlc.sr_ptr;
-		for (i = 0; i < adev->gfx.rlc.reg_list_size; i++)
-			dst_ptr[i] = cpu_to_le32(src_ptr[i]);
-		amdgpu_bo_kunmap(adev->gfx.rlc.save_restore_obj);
-		amdgpu_bo_unreserve(adev->gfx.rlc.save_restore_obj);
 	}
 
 	if (cs_data) {
-		/* clear state block */
-		adev->gfx.rlc.clear_state_size = dws = gfx_v7_0_get_csb_size(adev);
-
-		r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
-					      AMDGPU_GEM_DOMAIN_VRAM,
-					      &adev->gfx.rlc.clear_state_obj,
-					      &adev->gfx.rlc.clear_state_gpu_addr,
-					      (void **)&adev->gfx.rlc.cs_ptr);
-		if (r) {
-			dev_warn(adev->dev, "(%d) create RLC c bo failed\n", r);
-			gfx_v7_0_rlc_fini(adev);
+		/* init clear state block */
+		r = amdgpu_gfx_rlc_init_csb(adev);
+		if (r)
 			return r;
-		}
-
-		/* set up the cs buffer */
-		dst_ptr = adev->gfx.rlc.cs_ptr;
-		gfx_v7_0_get_csb_buffer(adev, dst_ptr);
-		amdgpu_bo_kunmap(adev->gfx.rlc.clear_state_obj);
-		amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
 	}
 
 	if (adev->gfx.rlc.cp_table_size) {
-
-		r = amdgpu_bo_create_reserved(adev, adev->gfx.rlc.cp_table_size,
-					      PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
-					      &adev->gfx.rlc.cp_table_obj,
-					      &adev->gfx.rlc.cp_table_gpu_addr,
-					      (void **)&adev->gfx.rlc.cp_table_ptr);
-		if (r) {
-			dev_warn(adev->dev, "(%d) create RLC cp table bo failed\n", r);
-			gfx_v7_0_rlc_fini(adev);
+		r = amdgpu_gfx_rlc_init_cpt(adev);
+		if (r)
 			return r;
-		}
-
-		gfx_v7_0_init_cp_pg_table(adev);
-
-		amdgpu_bo_kunmap(adev->gfx.rlc.cp_table_obj);
-		amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj);
-
 	}
 
+	/* init spm vmid with 0xf */
+	if (adev->gfx.rlc.funcs->update_spm_vmid)
+		adev->gfx.rlc.funcs->update_spm_vmid(adev, NULL, 0xf);
+
 	return 0;
 }
 
@@ -3420,7 +3270,7 @@ static void gfx_v7_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
 	mutex_lock(&adev->grbm_idx_mutex);
 	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
 		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
-			gfx_v7_0_select_se_sh(adev, i, j, 0xffffffff);
+			gfx_v7_0_select_se_sh(adev, i, j, 0xffffffff, 0);
 			for (k = 0; k < adev->usec_timeout; k++) {
 				if (RREG32(mmRLC_SERDES_CU_MASTER_BUSY) == 0)
 					break;
@@ -3428,7 +3278,7 @@ static void gfx_v7_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
 			}
 		}
 	}
-	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 	mutex_unlock(&adev->grbm_idx_mutex);
 
 	mask = RLC_SERDES_NONCU_MASTER_BUSY__SE_MASTER_BUSY_MASK |
@@ -3475,7 +3325,12 @@ static u32 gfx_v7_0_halt_rlc(struct amdgpu_device *adev)
 	return orig;
 }
 
-static void gfx_v7_0_enter_rlc_safe_mode(struct amdgpu_device *adev)
+static bool gfx_v7_0_is_rlc_enabled(struct amdgpu_device *adev)
+{
+	return true;
+}
+
+static void gfx_v7_0_set_safe_mode(struct amdgpu_device *adev, int xcc_id)
 {
 	u32 tmp, i, mask;
 
@@ -3497,7 +3352,7 @@ static void gfx_v7_0_enter_rlc_safe_mode(struct amdgpu_device *adev)
 	}
 }
 
-static void gfx_v7_0_exit_rlc_safe_mode(struct amdgpu_device *adev)
+static void gfx_v7_0_unset_safe_mode(struct amdgpu_device *adev, int xcc_id)
 {
 	u32 tmp;
 
@@ -3574,13 +3429,13 @@ static int gfx_v7_0_rlc_resume(struct amdgpu_device *adev)
 	adev->gfx.rlc_feature_version = le32_to_cpu(
 					hdr->ucode_feature_version);
 
-	gfx_v7_0_rlc_stop(adev);
+	adev->gfx.rlc.funcs->stop(adev);
 
 	/* disable CG */
 	tmp = RREG32(mmRLC_CGCG_CGLS_CTRL) & 0xfffffffc;
 	WREG32(mmRLC_CGCG_CGLS_CTRL, tmp);
 
-	gfx_v7_0_rlc_reset(adev);
+	adev->gfx.rlc.funcs->reset(adev);
 
 	gfx_v7_0_init_pg(adev);
 
@@ -3588,7 +3443,7 @@ static int gfx_v7_0_rlc_resume(struct amdgpu_device *adev)
 	WREG32(mmRLC_LB_CNTR_MAX, 0x00008000);
 
 	mutex_lock(&adev->grbm_idx_mutex);
-	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 	WREG32(mmRLC_LB_INIT_CU_MASK, 0xffffffff);
 	WREG32(mmRLC_LB_PARAMS, 0x00600408);
 	WREG32(mmRLC_LB_CNTL, 0x80000004);
@@ -3611,11 +3466,27 @@ static int gfx_v7_0_rlc_resume(struct amdgpu_device *adev)
 	if (adev->asic_type == CHIP_BONAIRE)
 		WREG32(mmRLC_DRIVER_CPDMA_STATUS, 0);
 
-	gfx_v7_0_rlc_start(adev);
+	adev->gfx.rlc.funcs->start(adev);
 
 	return 0;
 }
 
+static void gfx_v7_0_update_spm_vmid(struct amdgpu_device *adev, struct amdgpu_ring *ring, unsigned vmid)
+{
+	u32 data;
+
+	amdgpu_gfx_off_ctrl(adev, false);
+
+	data = RREG32(mmRLC_SPM_VMID);
+
+	data &= ~RLC_SPM_VMID__RLC_SPM_VMID_MASK;
+	data |= (vmid & RLC_SPM_VMID__RLC_SPM_VMID_MASK) << RLC_SPM_VMID__RLC_SPM_VMID__SHIFT;
+
+	WREG32(mmRLC_SPM_VMID, data);
+
+	amdgpu_gfx_off_ctrl(adev, true);
+}
+
 static void gfx_v7_0_enable_cgcg(struct amdgpu_device *adev, bool enable)
 {
 	u32 data, orig, tmp, tmp2;
@@ -3628,7 +3499,7 @@ static void gfx_v7_0_enable_cgcg(struct amdgpu_device *adev, bool enable)
 		tmp = gfx_v7_0_halt_rlc(adev);
 
 		mutex_lock(&adev->grbm_idx_mutex);
-		gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+		gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 		WREG32(mmRLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff);
 		WREG32(mmRLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff);
 		tmp2 = RLC_SERDES_WR_CTRL__BPM_ADDR_MASK |
@@ -3682,7 +3553,7 @@ static void gfx_v7_0_enable_mgcg(struct amdgpu_device *adev, bool enable)
 		tmp = gfx_v7_0_halt_rlc(adev);
 
 		mutex_lock(&adev->grbm_idx_mutex);
-		gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+		gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 		WREG32(mmRLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff);
 		WREG32(mmRLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff);
 		data = RLC_SERDES_WR_CTRL__BPM_ADDR_MASK |
@@ -3733,7 +3604,7 @@ static void gfx_v7_0_enable_mgcg(struct amdgpu_device *adev, bool enable)
 		tmp = gfx_v7_0_halt_rlc(adev);
 
 		mutex_lock(&adev->grbm_idx_mutex);
-		gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+		gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 		WREG32(mmRLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff);
 		WREG32(mmRLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff);
 		data = RLC_SERDES_WR_CTRL__BPM_ADDR_MASK | RLC_SERDES_WR_CTRL__MGCG_OVERRIDE_1_MASK;
@@ -3813,72 +3684,12 @@ static void gfx_v7_0_enable_gds_pg(struct amdgpu_device *adev, bool enable)
 		WREG32(mmRLC_PG_CNTL, data);
 }
 
-static void gfx_v7_0_init_cp_pg_table(struct amdgpu_device *adev)
+static int gfx_v7_0_cp_pg_table_num(struct amdgpu_device *adev)
 {
-	const __le32 *fw_data;
-	volatile u32 *dst_ptr;
-	int me, i, max_me = 4;
-	u32 bo_offset = 0;
-	u32 table_offset, table_size;
-
 	if (adev->asic_type == CHIP_KAVERI)
-		max_me = 5;
-
-	if (adev->gfx.rlc.cp_table_ptr == NULL)
-		return;
-
-	/* write the cp table buffer */
-	dst_ptr = adev->gfx.rlc.cp_table_ptr;
-	for (me = 0; me < max_me; me++) {
-		if (me == 0) {
-			const struct gfx_firmware_header_v1_0 *hdr =
-				(const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data;
-			fw_data = (const __le32 *)
-				(adev->gfx.ce_fw->data +
-				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
-			table_offset = le32_to_cpu(hdr->jt_offset);
-			table_size = le32_to_cpu(hdr->jt_size);
-		} else if (me == 1) {
-			const struct gfx_firmware_header_v1_0 *hdr =
-				(const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data;
-			fw_data = (const __le32 *)
-				(adev->gfx.pfp_fw->data +
-				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
-			table_offset = le32_to_cpu(hdr->jt_offset);
-			table_size = le32_to_cpu(hdr->jt_size);
-		} else if (me == 2) {
-			const struct gfx_firmware_header_v1_0 *hdr =
-				(const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data;
-			fw_data = (const __le32 *)
-				(adev->gfx.me_fw->data +
-				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
-			table_offset = le32_to_cpu(hdr->jt_offset);
-			table_size = le32_to_cpu(hdr->jt_size);
-		} else if (me == 3) {
-			const struct gfx_firmware_header_v1_0 *hdr =
-				(const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
-			fw_data = (const __le32 *)
-				(adev->gfx.mec_fw->data +
-				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
-			table_offset = le32_to_cpu(hdr->jt_offset);
-			table_size = le32_to_cpu(hdr->jt_size);
-		} else {
-			const struct gfx_firmware_header_v1_0 *hdr =
-				(const struct gfx_firmware_header_v1_0 *)adev->gfx.mec2_fw->data;
-			fw_data = (const __le32 *)
-				(adev->gfx.mec2_fw->data +
-				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
-			table_offset = le32_to_cpu(hdr->jt_offset);
-			table_size = le32_to_cpu(hdr->jt_size);
-		}
-
-		for (i = 0; i < table_size; i ++) {
-			dst_ptr[bo_offset + i] =
-				cpu_to_le32(le32_to_cpu(fw_data[table_offset + i]));
-		}
-
-		bo_offset += table_size;
-	}
+		return 5;
+	else
+		return 4;
 }
 
 static void gfx_v7_0_enable_gfx_cgpg(struct amdgpu_device *adev,
@@ -4071,70 +3882,24 @@ static u32 gfx_v7_0_get_csb_size(struct amdgpu_device *adev)
 	return count;
 }
 
-static void gfx_v7_0_get_csb_buffer(struct amdgpu_device *adev,
-				    volatile u32 *buffer)
+static void gfx_v7_0_get_csb_buffer(struct amdgpu_device *adev, u32 *buffer)
 {
-	u32 count = 0, i;
-	const struct cs_section_def *sect = NULL;
-	const struct cs_extent_def *ext = NULL;
+	u32 count = 0;
 
 	if (adev->gfx.rlc.cs_data == NULL)
 		return;
 	if (buffer == NULL)
 		return;
 
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
-	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
-
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CONTEXT_CONTROL, 1));
-	buffer[count++] = cpu_to_le32(0x80000000);
-	buffer[count++] = cpu_to_le32(0x80000000);
-
-	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
-		for (ext = sect->section; ext->extent != NULL; ++ext) {
-			if (sect->id == SECT_CONTEXT) {
-				buffer[count++] =
-					cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, ext->reg_count));
-				buffer[count++] = cpu_to_le32(ext->reg_index - PACKET3_SET_CONTEXT_REG_START);
-				for (i = 0; i < ext->reg_count; i++)
-					buffer[count++] = cpu_to_le32(ext->extent[i]);
-			} else {
-				return;
-			}
-		}
-	}
+	count = amdgpu_gfx_csb_preamble_start(buffer);
+	count = amdgpu_gfx_csb_data_parser(adev, buffer, count);
 
 	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, 2));
 	buffer[count++] = cpu_to_le32(mmPA_SC_RASTER_CONFIG - PACKET3_SET_CONTEXT_REG_START);
-	switch (adev->asic_type) {
-	case CHIP_BONAIRE:
-		buffer[count++] = cpu_to_le32(0x16000012);
-		buffer[count++] = cpu_to_le32(0x00000000);
-		break;
-	case CHIP_KAVERI:
-		buffer[count++] = cpu_to_le32(0x00000000); /* XXX */
-		buffer[count++] = cpu_to_le32(0x00000000);
-		break;
-	case CHIP_KABINI:
-	case CHIP_MULLINS:
-		buffer[count++] = cpu_to_le32(0x00000000); /* XXX */
-		buffer[count++] = cpu_to_le32(0x00000000);
-		break;
-	case CHIP_HAWAII:
-		buffer[count++] = cpu_to_le32(0x3a00161a);
-		buffer[count++] = cpu_to_le32(0x0000002e);
-		break;
-	default:
-		buffer[count++] = cpu_to_le32(0x00000000);
-		buffer[count++] = cpu_to_le32(0x00000000);
-		break;
-	}
-
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
-	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_END_CLEAR_STATE);
+	buffer[count++] = cpu_to_le32(adev->gfx.config.rb_config[0][0].raster_config);
+	buffer[count++] = cpu_to_le32(adev->gfx.config.rb_config[0][0].raster_config_1);
 
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CLEAR_STATE, 0));
-	buffer[count++] = cpu_to_le32(0);
+	amdgpu_gfx_csb_preamble_end(buffer, count);
 }
 
 static void gfx_v7_0_init_pg(struct amdgpu_device *adev)
@@ -4199,15 +3964,6 @@ static void gfx_v7_0_ring_emit_gds_switch(struct amdgpu_ring *ring,
 					  uint32_t gws_base, uint32_t gws_size,
 					  uint32_t oa_base, uint32_t oa_size)
 {
-	gds_base = gds_base >> AMDGPU_GDS_SHIFT;
-	gds_size = gds_size >> AMDGPU_GDS_SHIFT;
-
-	gws_base = gws_base >> AMDGPU_GWS_SHIFT;
-	gws_size = gws_size >> AMDGPU_GWS_SHIFT;
-
-	oa_base = oa_base >> AMDGPU_OA_SHIFT;
-	oa_size = oa_size >> AMDGPU_OA_SHIFT;
-
 	/* GDS Base */
 	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
 	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
@@ -4241,6 +3997,18 @@ static void gfx_v7_0_ring_emit_gds_switch(struct amdgpu_ring *ring,
 	amdgpu_ring_write(ring, (1 << (oa_size + oa_base)) - (1 << oa_base));
 }
 
+static void gfx_v7_0_ring_soft_recovery(struct amdgpu_ring *ring, unsigned vmid)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t value = 0;
+
+	value = REG_SET_FIELD(value, SQ_CMD, CMD, 0x03);
+	value = REG_SET_FIELD(value, SQ_CMD, MODE, 0x01);
+	value = REG_SET_FIELD(value, SQ_CMD, CHECK_VMID, 1);
+	value = REG_SET_FIELD(value, SQ_CMD, VM_ID, vmid);
+	WREG32(mmSQ_CMD, value);
+}
+
 static uint32_t wave_read_ind(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t address)
 {
 	WREG32(mmSQ_IND_INDEX,
@@ -4266,7 +4034,7 @@ static void wave_read_regs(struct amdgpu_device *adev, uint32_t simd,
 		*(out++) = RREG32(mmSQ_IND_DATA);
 }
 
-static void gfx_v7_0_read_wave_data(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t *dst, int *no_fields)
+static void gfx_v7_0_read_wave_data(struct amdgpu_device *adev, uint32_t xcc_id, uint32_t simd, uint32_t wave, uint32_t *dst, int *no_fields)
 {
 	/* type 0 wave data */
 	dst[(*no_fields)++] = 0;
@@ -4288,9 +4056,10 @@ static void gfx_v7_0_read_wave_data(struct amdgpu_device *adev, uint32_t simd, u
 	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TMA_HI);
 	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_IB_DBG0);
 	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_M0);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_MODE);
 }
 
-static void gfx_v7_0_read_wave_sgprs(struct amdgpu_device *adev, uint32_t simd,
+static void gfx_v7_0_read_wave_sgprs(struct amdgpu_device *adev, uint32_t xcc_id, uint32_t simd,
 				     uint32_t wave, uint32_t start,
 				     uint32_t size, uint32_t *dst)
 {
@@ -4299,24 +4068,43 @@ static void gfx_v7_0_read_wave_sgprs(struct amdgpu_device *adev, uint32_t simd,
 		start + SQIND_WAVE_SGPRS_OFFSET, size, dst);
 }
 
+static void gfx_v7_0_select_me_pipe_q(struct amdgpu_device *adev,
+				  u32 me, u32 pipe, u32 q, u32 vm, u32 xcc_id)
+{
+	cik_srbm_select(adev, me, pipe, q, vm);
+}
+
 static const struct amdgpu_gfx_funcs gfx_v7_0_gfx_funcs = {
 	.get_gpu_clock_counter = &gfx_v7_0_get_gpu_clock_counter,
 	.select_se_sh = &gfx_v7_0_select_se_sh,
 	.read_wave_data = &gfx_v7_0_read_wave_data,
 	.read_wave_sgprs = &gfx_v7_0_read_wave_sgprs,
+	.select_me_pipe_q = &gfx_v7_0_select_me_pipe_q
 };
 
 static const struct amdgpu_rlc_funcs gfx_v7_0_rlc_funcs = {
-	.enter_safe_mode = gfx_v7_0_enter_rlc_safe_mode,
-	.exit_safe_mode = gfx_v7_0_exit_rlc_safe_mode
+	.is_rlc_enabled = gfx_v7_0_is_rlc_enabled,
+	.set_safe_mode = gfx_v7_0_set_safe_mode,
+	.unset_safe_mode = gfx_v7_0_unset_safe_mode,
+	.init = gfx_v7_0_rlc_init,
+	.get_csb_size = gfx_v7_0_get_csb_size,
+	.get_csb_buffer = gfx_v7_0_get_csb_buffer,
+	.get_cp_table_num = gfx_v7_0_cp_pg_table_num,
+	.resume = gfx_v7_0_rlc_resume,
+	.stop = gfx_v7_0_rlc_stop,
+	.reset = gfx_v7_0_rlc_reset,
+	.start = gfx_v7_0_rlc_start,
+	.update_spm_vmid = gfx_v7_0_update_spm_vmid
 };
 
-static int gfx_v7_0_early_init(void *handle)
+static int gfx_v7_0_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
+	adev->gfx.xcc_mask = 1;
 	adev->gfx.num_gfx_rings = GFX7_NUM_GFX_RINGS;
-	adev->gfx.num_compute_rings = AMDGPU_MAX_COMPUTE_RINGS;
+	adev->gfx.num_compute_rings = min(amdgpu_gfx_get_num_kcq(adev),
+					  AMDGPU_MAX_COMPUTE_RINGS);
 	adev->gfx.funcs = &gfx_v7_0_gfx_funcs;
 	adev->gfx.rlc.funcs = &gfx_v7_0_rlc_funcs;
 	gfx_v7_0_set_ring_funcs(adev);
@@ -4326,9 +4114,9 @@ static int gfx_v7_0_early_init(void *handle)
 	return 0;
 }
 
-static int gfx_v7_0_late_init(void *handle)
+static int gfx_v7_0_late_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int r;
 
 	r = amdgpu_irq_get(adev, &adev->gfx.priv_reg_irq, 0);
@@ -4345,7 +4133,7 @@ static int gfx_v7_0_late_init(void *handle)
 static void gfx_v7_0_gpu_early_init(struct amdgpu_device *adev)
 {
 	u32 gb_addr_config;
-	u32 mc_shared_chmap, mc_arb_ramcfg;
+	u32 mc_arb_ramcfg;
 	u32 dimm00_addr_map, dimm01_addr_map, dimm10_addr_map, dimm11_addr_map;
 	u32 tmp;
 
@@ -4387,34 +4175,8 @@ static void gfx_v7_0_gpu_early_init(struct amdgpu_device *adev)
 	case CHIP_KAVERI:
 		adev->gfx.config.max_shader_engines = 1;
 		adev->gfx.config.max_tile_pipes = 4;
-		if ((adev->pdev->device == 0x1304) ||
-		    (adev->pdev->device == 0x1305) ||
-		    (adev->pdev->device == 0x130C) ||
-		    (adev->pdev->device == 0x130F) ||
-		    (adev->pdev->device == 0x1310) ||
-		    (adev->pdev->device == 0x1311) ||
-		    (adev->pdev->device == 0x131C)) {
-			adev->gfx.config.max_cu_per_sh = 8;
-			adev->gfx.config.max_backends_per_se = 2;
-		} else if ((adev->pdev->device == 0x1309) ||
-			   (adev->pdev->device == 0x130A) ||
-			   (adev->pdev->device == 0x130D) ||
-			   (adev->pdev->device == 0x1313) ||
-			   (adev->pdev->device == 0x131D)) {
-			adev->gfx.config.max_cu_per_sh = 6;
-			adev->gfx.config.max_backends_per_se = 2;
-		} else if ((adev->pdev->device == 0x1306) ||
-			   (adev->pdev->device == 0x1307) ||
-			   (adev->pdev->device == 0x130B) ||
-			   (adev->pdev->device == 0x130E) ||
-			   (adev->pdev->device == 0x1315) ||
-			   (adev->pdev->device == 0x131B)) {
-			adev->gfx.config.max_cu_per_sh = 4;
-			adev->gfx.config.max_backends_per_se = 1;
-		} else {
-			adev->gfx.config.max_cu_per_sh = 3;
-			adev->gfx.config.max_backends_per_se = 1;
-		}
+		adev->gfx.config.max_cu_per_sh = 8;
+		adev->gfx.config.max_backends_per_se = 2;
 		adev->gfx.config.max_sh_per_se = 1;
 		adev->gfx.config.max_texture_channel_caches = 4;
 		adev->gfx.config.max_gprs = 256;
@@ -4448,10 +4210,14 @@ static void gfx_v7_0_gpu_early_init(struct amdgpu_device *adev)
 		break;
 	}
 
-	mc_shared_chmap = RREG32(mmMC_SHARED_CHMAP);
 	adev->gfx.config.mc_arb_ramcfg = RREG32(mmMC_ARB_RAMCFG);
 	mc_arb_ramcfg = adev->gfx.config.mc_arb_ramcfg;
 
+	adev->gfx.config.num_banks = REG_GET_FIELD(mc_arb_ramcfg,
+				MC_ARB_RAMCFG, NOOFBANK);
+	adev->gfx.config.num_ranks = REG_GET_FIELD(mc_arb_ramcfg,
+				MC_ARB_RAMCFG, NOOFRANKS);
+
 	adev->gfx.config.num_tile_pipes = adev->gfx.config.max_tile_pipes;
 	adev->gfx.config.mem_max_burst_length_bytes = 256;
 	if (adev->flags & AMD_IS_APU) {
@@ -4522,7 +4288,7 @@ static int gfx_v7_0_compute_ring_init(struct amdgpu_device *adev, int ring_id,
 
 	ring->ring_obj = NULL;
 	ring->use_doorbell = true;
-	ring->doorbell_index = AMDGPU_DOORBELL_MEC_RING0 + ring_id;
+	ring->doorbell_index = adev->doorbell_index.mec_ring0 + ring_id;
 	sprintf(ring->name, "comp_%d.%d.%d", ring->me, ring->pipe, ring->queue);
 
 	irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP
@@ -4531,7 +4297,8 @@ static int gfx_v7_0_compute_ring_init(struct amdgpu_device *adev, int ring_id,
 
 	/* type-2 packets are deprecated on MEC, use type-3 instead */
 	r = amdgpu_ring_init(adev, ring, 1024,
-			&adev->gfx.eop_irq, irq_type);
+			     &adev->gfx.eop_irq, irq_type,
+			     AMDGPU_RING_PRIO_DEFAULT, NULL);
 	if (r)
 		return r;
 
@@ -4539,10 +4306,10 @@ static int gfx_v7_0_compute_ring_init(struct amdgpu_device *adev, int ring_id,
 	return 0;
 }
 
-static int gfx_v7_0_sw_init(void *handle)
+static int gfx_v7_0_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_ring *ring;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int i, j, k, r, ring_id;
 
 	switch (adev->asic_type) {
@@ -4561,31 +4328,29 @@ static int gfx_v7_0_sw_init(void *handle)
 	adev->gfx.mec.num_queue_per_pipe = 8;
 
 	/* EOP Event */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 181, &adev->gfx.eop_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 181, &adev->gfx.eop_irq);
 	if (r)
 		return r;
 
 	/* Privileged reg */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 184,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 184,
 			      &adev->gfx.priv_reg_irq);
 	if (r)
 		return r;
 
 	/* Privileged inst */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 185,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 185,
 			      &adev->gfx.priv_inst_irq);
 	if (r)
 		return r;
 
-	gfx_v7_0_scratch_init(adev);
-
 	r = gfx_v7_0_init_microcode(adev);
 	if (r) {
 		DRM_ERROR("Failed to load gfx firmware!\n");
 		return r;
 	}
 
-	r = gfx_v7_0_rlc_init(adev);
+	r = adev->gfx.rlc.funcs->init(adev);
 	if (r) {
 		DRM_ERROR("Failed to init rlc BOs!\n");
 		return r;
@@ -4603,7 +4368,9 @@ static int gfx_v7_0_sw_init(void *handle)
 		ring->ring_obj = NULL;
 		sprintf(ring->name, "gfx");
 		r = amdgpu_ring_init(adev, ring, 1024,
-				     &adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_EOP);
+				     &adev->gfx.eop_irq,
+				     AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP,
+				     AMDGPU_RING_PRIO_DEFAULT, NULL);
 		if (r)
 			return r;
 	}
@@ -4613,7 +4380,8 @@ static int gfx_v7_0_sw_init(void *handle)
 	for (i = 0; i < adev->gfx.mec.num_mec; ++i) {
 		for (j = 0; j < adev->gfx.mec.num_queue_per_pipe; j++) {
 			for (k = 0; k < adev->gfx.mec.num_pipe_per_mec; k++) {
-				if (!amdgpu_gfx_is_mec_queue_enabled(adev, i, k, j))
+				if (!amdgpu_gfx_is_mec_queue_enabled(adev, 0, i,
+								     k, j))
 					continue;
 
 				r = gfx_v7_0_compute_ring_init(adev,
@@ -4627,25 +4395,6 @@ static int gfx_v7_0_sw_init(void *handle)
 		}
 	}
 
-	/* reserve GDS, GWS and OA resource for gfx */
-	r = amdgpu_bo_create_kernel(adev, adev->gds.mem.gfx_partition_size,
-				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GDS,
-				    &adev->gds.gds_gfx_bo, NULL, NULL);
-	if (r)
-		return r;
-
-	r = amdgpu_bo_create_kernel(adev, adev->gds.gws.gfx_partition_size,
-				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GWS,
-				    &adev->gds.gws_gfx_bo, NULL, NULL);
-	if (r)
-		return r;
-
-	r = amdgpu_bo_create_kernel(adev, adev->gds.oa.gfx_partition_size,
-				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_OA,
-				    &adev->gds.oa_gfx_bo, NULL, NULL);
-	if (r)
-		return r;
-
 	adev->gfx.ce_ram_size = 0x8000;
 
 	gfx_v7_0_gpu_early_init(adev);
@@ -4653,14 +4402,10 @@ static int gfx_v7_0_sw_init(void *handle)
 	return r;
 }
 
-static int gfx_v7_0_sw_fini(void *handle)
+static int gfx_v7_0_sw_fini(struct amdgpu_ip_block *ip_block)
 {
+	struct amdgpu_device *adev = ip_block->adev;
 	int i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	amdgpu_bo_free_kernel(&adev->gds.oa_gfx_bo, NULL, NULL);
-	amdgpu_bo_free_kernel(&adev->gds.gws_gfx_bo, NULL, NULL);
-	amdgpu_bo_free_kernel(&adev->gds.gds_gfx_bo, NULL, NULL);
 
 	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
 		amdgpu_ring_fini(&adev->gfx.gfx_ring[i]);
@@ -4668,22 +4413,32 @@ static int gfx_v7_0_sw_fini(void *handle)
 		amdgpu_ring_fini(&adev->gfx.compute_ring[i]);
 
 	gfx_v7_0_cp_compute_fini(adev);
-	gfx_v7_0_rlc_fini(adev);
+	amdgpu_gfx_rlc_fini(adev);
 	gfx_v7_0_mec_fini(adev);
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj,
+				&adev->gfx.rlc.clear_state_gpu_addr,
+				(void **)&adev->gfx.rlc.cs_ptr);
+	if (adev->gfx.rlc.cp_table_size) {
+		amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj,
+				&adev->gfx.rlc.cp_table_gpu_addr,
+				(void **)&adev->gfx.rlc.cp_table_ptr);
+	}
 	gfx_v7_0_free_microcode(adev);
 
 	return 0;
 }
 
-static int gfx_v7_0_hw_init(void *handle)
+static int gfx_v7_0_hw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	gfx_v7_0_gpu_init(adev);
+	gfx_v7_0_constants_init(adev);
 
+	/* init CSB */
+	adev->gfx.rlc.funcs->get_csb_buffer(adev, adev->gfx.rlc.cs_ptr);
 	/* init rlc */
-	r = gfx_v7_0_rlc_resume(adev);
+	r = adev->gfx.rlc.funcs->resume(adev);
 	if (r)
 		return r;
 
@@ -4694,36 +4449,32 @@ static int gfx_v7_0_hw_init(void *handle)
 	return r;
 }
 
-static int gfx_v7_0_hw_fini(void *handle)
+static int gfx_v7_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	amdgpu_irq_put(adev, &adev->gfx.priv_reg_irq, 0);
 	amdgpu_irq_put(adev, &adev->gfx.priv_inst_irq, 0);
 	gfx_v7_0_cp_enable(adev, false);
-	gfx_v7_0_rlc_stop(adev);
+	adev->gfx.rlc.funcs->stop(adev);
 	gfx_v7_0_fini_pg(adev);
 
 	return 0;
 }
 
-static int gfx_v7_0_suspend(void *handle)
+static int gfx_v7_0_suspend(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return gfx_v7_0_hw_fini(adev);
+	return gfx_v7_0_hw_fini(ip_block);
 }
 
-static int gfx_v7_0_resume(void *handle)
+static int gfx_v7_0_resume(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return gfx_v7_0_hw_init(adev);
+	return gfx_v7_0_hw_init(ip_block);
 }
 
-static bool gfx_v7_0_is_idle(void *handle)
+static bool gfx_v7_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (RREG32(mmGRBM_STATUS) & GRBM_STATUS__GUI_ACTIVE_MASK)
 		return false;
@@ -4731,11 +4482,11 @@ static bool gfx_v7_0_is_idle(void *handle)
 		return true;
 }
 
-static int gfx_v7_0_wait_for_idle(void *handle)
+static int gfx_v7_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
 	unsigned i;
 	u32 tmp;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
 		/* read MC_STATUS */
@@ -4748,11 +4499,11 @@ static int gfx_v7_0_wait_for_idle(void *handle)
 	return -ETIMEDOUT;
 }
 
-static int gfx_v7_0_soft_reset(void *handle)
+static int gfx_v7_0_soft_reset(struct amdgpu_ip_block *ip_block)
 {
 	u32 grbm_soft_reset = 0, srbm_soft_reset = 0;
 	u32 tmp;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	/* GRBM_STATUS */
 	tmp = RREG32(mmGRBM_STATUS);
@@ -4786,7 +4537,7 @@ static int gfx_v7_0_soft_reset(void *handle)
 		gfx_v7_0_update_cg(adev, false);
 
 		/* stop the rlc */
-		gfx_v7_0_rlc_stop(adev);
+		adev->gfx.rlc.funcs->stop(adev);
 
 		/* Disable GFX parsing/prefetching */
 		WREG32(mmCP_ME_CNTL, CP_ME_CNTL__ME_HALT_MASK | CP_ME_CNTL__PFP_HALT_MASK | CP_ME_CNTL__CE_HALT_MASK);
@@ -4955,7 +4706,7 @@ static int gfx_v7_0_set_eop_interrupt_state(struct amdgpu_device *adev,
 					    enum amdgpu_interrupt_state state)
 {
 	switch (type) {
-	case AMDGPU_CP_IRQ_GFX_EOP:
+	case AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP:
 		gfx_v7_0_set_gfx_eop_interrupt_state(adev, state);
 		break;
 	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP:
@@ -5015,12 +4766,36 @@ static int gfx_v7_0_eop_irq(struct amdgpu_device *adev,
 	return 0;
 }
 
+static void gfx_v7_0_fault(struct amdgpu_device *adev,
+			   struct amdgpu_iv_entry *entry)
+{
+	struct amdgpu_ring *ring;
+	u8 me_id, pipe_id;
+	int i;
+
+	me_id = (entry->ring_id & 0x0c) >> 2;
+	pipe_id = (entry->ring_id & 0x03) >> 0;
+	switch (me_id) {
+	case 0:
+		drm_sched_fault(&adev->gfx.gfx_ring[0].sched);
+		break;
+	case 1:
+	case 2:
+		for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+			ring = &adev->gfx.compute_ring[i];
+			if ((ring->me == me_id) && (ring->pipe == pipe_id))
+				drm_sched_fault(&ring->sched);
+		}
+		break;
+	}
+}
+
 static int gfx_v7_0_priv_reg_irq(struct amdgpu_device *adev,
 				 struct amdgpu_irq_src *source,
 				 struct amdgpu_iv_entry *entry)
 {
 	DRM_ERROR("Illegal register access in command stream\n");
-	schedule_work(&adev->reset_work);
+	gfx_v7_0_fault(adev, entry);
 	return 0;
 }
 
@@ -5030,15 +4805,15 @@ static int gfx_v7_0_priv_inst_irq(struct amdgpu_device *adev,
 {
 	DRM_ERROR("Illegal instruction in command stream\n");
 	// XXX soft reset the gfx block only
-	schedule_work(&adev->reset_work);
+	gfx_v7_0_fault(adev, entry);
 	return 0;
 }
 
-static int gfx_v7_0_set_clockgating_state(void *handle,
+static int gfx_v7_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
 	bool gate = false;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (state == AMD_CG_STATE_GATE)
 		gate = true;
@@ -5057,11 +4832,11 @@ static int gfx_v7_0_set_clockgating_state(void *handle,
 	return 0;
 }
 
-static int gfx_v7_0_set_powergating_state(void *handle,
+static int gfx_v7_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	bool gate = false;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (state == AMD_PG_STATE_GATE)
 		gate = true;
@@ -5082,6 +4857,32 @@ static int gfx_v7_0_set_powergating_state(void *handle,
 	return 0;
 }
 
+static void gfx_v7_0_emit_mem_sync(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SURFACE_SYNC, 3));
+	amdgpu_ring_write(ring, PACKET3_TCL1_ACTION_ENA |
+			  PACKET3_TC_ACTION_ENA |
+			  PACKET3_SH_KCACHE_ACTION_ENA |
+			  PACKET3_SH_ICACHE_ACTION_ENA);  /* CP_COHER_CNTL */
+	amdgpu_ring_write(ring, 0xffffffff);  /* CP_COHER_SIZE */
+	amdgpu_ring_write(ring, 0);  /* CP_COHER_BASE */
+	amdgpu_ring_write(ring, 0x0000000A); /* poll interval */
+}
+
+static void gfx_v7_0_emit_mem_sync_compute(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_ACQUIRE_MEM, 5));
+	amdgpu_ring_write(ring, PACKET3_TCL1_ACTION_ENA |
+			  PACKET3_TC_ACTION_ENA |
+			  PACKET3_SH_KCACHE_ACTION_ENA |
+			  PACKET3_SH_ICACHE_ACTION_ENA);  /* CP_COHER_CNTL */
+	amdgpu_ring_write(ring, 0xffffffff);	/* CP_COHER_SIZE */
+	amdgpu_ring_write(ring, 0xff);		/* CP_COHER_SIZE_HI */
+	amdgpu_ring_write(ring, 0);		/* CP_COHER_BASE */
+	amdgpu_ring_write(ring, 0);		/* CP_COHER_BASE_HI */
+	amdgpu_ring_write(ring, 0x0000000A);	/* poll interval */
+}
+
 static const struct amd_ip_funcs gfx_v7_0_ip_funcs = {
 	.name = "gfx_v7_0",
 	.early_init = gfx_v7_0_early_init,
@@ -5110,11 +4911,12 @@ static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_gfx = {
 	.emit_frame_size =
 		20 + /* gfx_v7_0_ring_emit_gds_switch */
 		7 + /* gfx_v7_0_ring_emit_hdp_flush */
-		5 + /* gfx_v7_0_ring_emit_hdp_invalidate */
+		5 + /* hdp invalidate */
 		12 + 12 + 12 + /* gfx_v7_0_ring_emit_fence_gfx x3 for user fence, vm fence */
 		7 + 4 + /* gfx_v7_0_ring_emit_pipeline_sync */
-		17 + 6 + /* gfx_v7_0_ring_emit_vm_flush */
-		3 + 4, /* gfx_v7_ring_emit_cntxcntl including vgt flush*/
+		CIK_FLUSH_GPU_TLB_NUM_WREG * 5 + 7 + 6 + /* gfx_v7_0_ring_emit_vm_flush */
+		3 + 4 + /* gfx_v7_ring_emit_cntxcntl including vgt flush*/
+		5, /* SURFACE_SYNC */
 	.emit_ib_size = 4, /* gfx_v7_0_ring_emit_ib_gfx */
 	.emit_ib = gfx_v7_0_ring_emit_ib_gfx,
 	.emit_fence = gfx_v7_0_ring_emit_fence_gfx,
@@ -5122,12 +4924,14 @@ static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_gfx = {
 	.emit_vm_flush = gfx_v7_0_ring_emit_vm_flush,
 	.emit_gds_switch = gfx_v7_0_ring_emit_gds_switch,
 	.emit_hdp_flush = gfx_v7_0_ring_emit_hdp_flush,
-	.emit_hdp_invalidate = gfx_v7_0_ring_emit_hdp_invalidate,
 	.test_ring = gfx_v7_0_ring_test_ring,
 	.test_ib = gfx_v7_0_ring_test_ib,
 	.insert_nop = amdgpu_ring_insert_nop,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
 	.emit_cntxcntl = gfx_v7_ring_emit_cntxcntl,
+	.emit_wreg = gfx_v7_0_ring_emit_wreg,
+	.soft_recovery = gfx_v7_0_ring_soft_recovery,
+	.emit_mem_sync = gfx_v7_0_emit_mem_sync,
 };
 
 static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_compute = {
@@ -5141,22 +4945,25 @@ static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_compute = {
 	.emit_frame_size =
 		20 + /* gfx_v7_0_ring_emit_gds_switch */
 		7 + /* gfx_v7_0_ring_emit_hdp_flush */
-		5 + /* gfx_v7_0_ring_emit_hdp_invalidate */
+		5 + /* hdp invalidate */
 		7 + /* gfx_v7_0_ring_emit_pipeline_sync */
-		17 + /* gfx_v7_0_ring_emit_vm_flush */
-		7 + 7 + 7, /* gfx_v7_0_ring_emit_fence_compute x3 for user fence, vm fence */
-	.emit_ib_size =	4, /* gfx_v7_0_ring_emit_ib_compute */
+		CIK_FLUSH_GPU_TLB_NUM_WREG * 5 + 7 + /* gfx_v7_0_ring_emit_vm_flush */
+		7 + 7 + 7 + /* gfx_v7_0_ring_emit_fence_compute x3 for user fence, vm fence */
+		7, /* gfx_v7_0_emit_mem_sync_compute */
+	.emit_ib_size =	7, /* gfx_v7_0_ring_emit_ib_compute */
 	.emit_ib = gfx_v7_0_ring_emit_ib_compute,
 	.emit_fence = gfx_v7_0_ring_emit_fence_compute,
 	.emit_pipeline_sync = gfx_v7_0_ring_emit_pipeline_sync,
 	.emit_vm_flush = gfx_v7_0_ring_emit_vm_flush,
 	.emit_gds_switch = gfx_v7_0_ring_emit_gds_switch,
 	.emit_hdp_flush = gfx_v7_0_ring_emit_hdp_flush,
-	.emit_hdp_invalidate = gfx_v7_0_ring_emit_hdp_invalidate,
 	.test_ring = gfx_v7_0_ring_test_ring,
 	.test_ib = gfx_v7_0_ring_test_ib,
 	.insert_nop = amdgpu_ring_insert_nop,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_wreg = gfx_v7_0_ring_emit_wreg,
+	.soft_recovery = gfx_v7_0_ring_soft_recovery,
+	.emit_mem_sync = gfx_v7_0_emit_mem_sync_compute,
 };
 
 static void gfx_v7_0_set_ring_funcs(struct amdgpu_device *adev)
@@ -5199,29 +5006,10 @@ static void gfx_v7_0_set_irq_funcs(struct amdgpu_device *adev)
 static void gfx_v7_0_set_gds_init(struct amdgpu_device *adev)
 {
 	/* init asci gds info */
-	adev->gds.mem.total_size = RREG32(mmGDS_VMID0_SIZE);
-	adev->gds.gws.total_size = 64;
-	adev->gds.oa.total_size = 16;
-
-	if (adev->gds.mem.total_size == 64 * 1024) {
-		adev->gds.mem.gfx_partition_size = 4096;
-		adev->gds.mem.cs_partition_size = 4096;
-
-		adev->gds.gws.gfx_partition_size = 4;
-		adev->gds.gws.cs_partition_size = 4;
-
-		adev->gds.oa.gfx_partition_size = 4;
-		adev->gds.oa.cs_partition_size = 1;
-	} else {
-		adev->gds.mem.gfx_partition_size = 1024;
-		adev->gds.mem.cs_partition_size = 1024;
-
-		adev->gds.gws.gfx_partition_size = 16;
-		adev->gds.gws.cs_partition_size = 16;
-
-		adev->gds.oa.gfx_partition_size = 4;
-		adev->gds.oa.cs_partition_size = 4;
-	}
+	adev->gds.gds_size = RREG32(mmGDS_VMID0_SIZE);
+	adev->gds.gws_size = 64;
+	adev->gds.oa_size = 16;
+	adev->gds.gds_compute_max_wave_id = RREG32(mmGDS_COMPUTE_MAX_WAVE_ID);
 }
 
 
@@ -5248,18 +5036,18 @@ static void gfx_v7_0_get_cu_info(struct amdgpu_device *adev)
 			mask = 1;
 			ao_bitmap = 0;
 			counter = 0;
-			gfx_v7_0_select_se_sh(adev, i, j, 0xffffffff);
+			gfx_v7_0_select_se_sh(adev, i, j, 0xffffffff, 0);
 			if (i < 4 && j < 2)
 				gfx_v7_0_set_user_cu_inactive_bitmap(
 					adev, disable_masks[i * 2 + j]);
 			bitmap = gfx_v7_0_get_cu_active_bitmap(adev);
-			cu_info->bitmap[i][j] = bitmap;
+			cu_info->bitmap[0][i][j] = bitmap;
 
-			for (k = 0; k < adev->gfx.config.max_cu_per_sh; k ++) {
+			for (k = 0; k < adev->gfx.config.max_cu_per_sh; k++) {
 				if (bitmap & mask) {
 					if (counter < ao_cu_num)
 						ao_bitmap |= mask;
-					counter ++;
+					counter++;
 				}
 				mask <<= 1;
 			}
@@ -5269,24 +5057,19 @@ static void gfx_v7_0_get_cu_info(struct amdgpu_device *adev)
 			cu_info->ao_cu_bitmap[i][j] = ao_bitmap;
 		}
 	}
-	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 	mutex_unlock(&adev->grbm_idx_mutex);
 
 	cu_info->number = active_cu_number;
 	cu_info->ao_cu_mask = ao_cu_mask;
+	cu_info->simd_per_cu = NUM_SIMD_PER_CU;
+	cu_info->max_waves_per_simd = 10;
+	cu_info->max_scratch_slots_per_cu = 32;
+	cu_info->wave_front_size = 64;
+	cu_info->lds_size = 64;
 }
 
-const struct amdgpu_ip_block_version gfx_v7_0_ip_block =
-{
-	.type = AMD_IP_BLOCK_TYPE_GFX,
-	.major = 7,
-	.minor = 0,
-	.rev = 0,
-	.funcs = &gfx_v7_0_ip_funcs,
-};
-
-const struct amdgpu_ip_block_version gfx_v7_1_ip_block =
-{
+const struct amdgpu_ip_block_version gfx_v7_1_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_GFX,
 	.major = 7,
 	.minor = 1,
@@ -5294,8 +5077,7 @@ const struct amdgpu_ip_block_version gfx_v7_1_ip_block =
 	.funcs = &gfx_v7_0_ip_funcs,
 };
 
-const struct amdgpu_ip_block_version gfx_v7_2_ip_block =
-{
+const struct amdgpu_ip_block_version gfx_v7_2_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_GFX,
 	.major = 7,
 	.minor = 2,
@@ -5303,8 +5085,7 @@ const struct amdgpu_ip_block_version gfx_v7_2_ip_block =
 	.funcs = &gfx_v7_0_ip_funcs,
 };
 
-const struct amdgpu_ip_block_version gfx_v7_3_ip_block =
-{
+const struct amdgpu_ip_block_version gfx_v7_3_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_GFX,
 	.major = 7,
 	.minor = 3,
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.h b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.h
index 6fb9c1524691..eedce7d007f1 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.h
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.h
@@ -24,7 +24,6 @@
 #ifndef __GFX_V7_0_H__
 #define __GFX_V7_0_H__
 
-extern const struct amdgpu_ip_block_version gfx_v7_0_ip_block;
 extern const struct amdgpu_ip_block_version gfx_v7_1_ip_block;
 extern const struct amdgpu_ip_block_version gfx_v7_2_ip_block;
 extern const struct amdgpu_ip_block_version gfx_v7_3_ip_block;
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
index fc260c13b1da..0856ff65288c 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
@@ -20,10 +20,16 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  *
  */
+
+#include <linux/delay.h>
+#include <linux/kernel.h>
 #include <linux/firmware.h>
-#include <drm/drmP.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+
 #include "amdgpu.h"
 #include "amdgpu_gfx.h"
+#include "amdgpu_ring.h"
 #include "vi.h"
 #include "vi_structs.h"
 #include "vid.h"
@@ -43,15 +49,16 @@
 #include "gca/gfx_8_0_d.h"
 #include "gca/gfx_8_0_enum.h"
 #include "gca/gfx_8_0_sh_mask.h"
-#include "gca/gfx_8_0_enum.h"
 
 #include "dce/dce_10_0_d.h"
 #include "dce/dce_10_0_sh_mask.h"
 
 #include "smu/smu_7_1_3_d.h"
 
+#include "ivsrcid/ivsrcid_vislands30.h"
+
 #define GFX8_NUM_GFX_RINGS     1
-#define GFX8_MEC_HPD_SIZE 2048
+#define GFX8_MEC_HPD_SIZE 4096
 
 #define TOPAZ_GB_ADDR_CONFIG_GOLDEN 0x22010001
 #define CARRIZO_GB_ADDR_CONFIG_GOLDEN 0x22010001
@@ -124,27 +131,49 @@ MODULE_FIRMWARE("amdgpu/fiji_mec.bin");
 MODULE_FIRMWARE("amdgpu/fiji_mec2.bin");
 MODULE_FIRMWARE("amdgpu/fiji_rlc.bin");
 
-MODULE_FIRMWARE("amdgpu/polaris11_ce.bin");
-MODULE_FIRMWARE("amdgpu/polaris11_pfp.bin");
-MODULE_FIRMWARE("amdgpu/polaris11_me.bin");
-MODULE_FIRMWARE("amdgpu/polaris11_mec.bin");
-MODULE_FIRMWARE("amdgpu/polaris11_mec2.bin");
-MODULE_FIRMWARE("amdgpu/polaris11_rlc.bin");
-
 MODULE_FIRMWARE("amdgpu/polaris10_ce.bin");
+MODULE_FIRMWARE("amdgpu/polaris10_ce_2.bin");
 MODULE_FIRMWARE("amdgpu/polaris10_pfp.bin");
+MODULE_FIRMWARE("amdgpu/polaris10_pfp_2.bin");
 MODULE_FIRMWARE("amdgpu/polaris10_me.bin");
+MODULE_FIRMWARE("amdgpu/polaris10_me_2.bin");
 MODULE_FIRMWARE("amdgpu/polaris10_mec.bin");
+MODULE_FIRMWARE("amdgpu/polaris10_mec_2.bin");
 MODULE_FIRMWARE("amdgpu/polaris10_mec2.bin");
+MODULE_FIRMWARE("amdgpu/polaris10_mec2_2.bin");
 MODULE_FIRMWARE("amdgpu/polaris10_rlc.bin");
 
+MODULE_FIRMWARE("amdgpu/polaris11_ce.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_ce_2.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_pfp.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_pfp_2.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_me.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_me_2.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_mec.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_mec_2.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_mec2.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_mec2_2.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_rlc.bin");
+
 MODULE_FIRMWARE("amdgpu/polaris12_ce.bin");
+MODULE_FIRMWARE("amdgpu/polaris12_ce_2.bin");
 MODULE_FIRMWARE("amdgpu/polaris12_pfp.bin");
+MODULE_FIRMWARE("amdgpu/polaris12_pfp_2.bin");
 MODULE_FIRMWARE("amdgpu/polaris12_me.bin");
+MODULE_FIRMWARE("amdgpu/polaris12_me_2.bin");
 MODULE_FIRMWARE("amdgpu/polaris12_mec.bin");
+MODULE_FIRMWARE("amdgpu/polaris12_mec_2.bin");
 MODULE_FIRMWARE("amdgpu/polaris12_mec2.bin");
+MODULE_FIRMWARE("amdgpu/polaris12_mec2_2.bin");
 MODULE_FIRMWARE("amdgpu/polaris12_rlc.bin");
 
+MODULE_FIRMWARE("amdgpu/vegam_ce.bin");
+MODULE_FIRMWARE("amdgpu/vegam_pfp.bin");
+MODULE_FIRMWARE("amdgpu/vegam_me.bin");
+MODULE_FIRMWARE("amdgpu/vegam_mec.bin");
+MODULE_FIRMWARE("amdgpu/vegam_mec2.bin");
+MODULE_FIRMWARE("amdgpu/vegam_rlc.bin");
+
 static const struct amdgpu_gds_reg_offset amdgpu_gds_reg_offset[] =
 {
 	{mmGDS_VMID0_BASE, mmGDS_VMID0_SIZE, mmGDS_GWS_VMID0, mmGDS_OA_VMID0},
@@ -276,6 +305,37 @@ static const u32 tonga_mgcg_cgcg_init[] =
 	mmCP_MEM_SLP_CNTL, 0x00000001, 0x00000001,
 };
 
+static const u32 golden_settings_vegam_a11[] =
+{
+	mmCB_HW_CONTROL, 0x0001f3cf, 0x00007208,
+	mmCB_HW_CONTROL_2, 0x0f000000, 0x0d000000,
+	mmCB_HW_CONTROL_3, 0x000001ff, 0x00000040,
+	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
+	mmPA_SC_ENHANCE, 0xffffffff, 0x20000001,
+	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
+	mmPA_SC_RASTER_CONFIG, 0x3f3fffff, 0x3a00161a,
+	mmPA_SC_RASTER_CONFIG_1, 0x0000003f, 0x0000002e,
+	mmRLC_CGCG_CGLS_CTRL, 0x00000003, 0x0001003c,
+	mmRLC_CGCG_CGLS_CTRL_3D, 0xffffffff, 0x0001003c,
+	mmSQ_CONFIG, 0x07f80000, 0x01180000,
+	mmTA_CNTL_AUX, 0x000f000f, 0x000b0000,
+	mmTCC_CTRL, 0x00100000, 0xf31fff7f,
+	mmTCP_ADDR_CONFIG, 0x000003ff, 0x000000f7,
+	mmTCP_CHAN_STEER_HI, 0xffffffff, 0x00000000,
+	mmTCP_CHAN_STEER_LO, 0xffffffff, 0x32761054,
+	mmVGT_RESET_DEBUG, 0x00000004, 0x00000004,
+};
+
+static const u32 vegam_golden_common_all[] =
+{
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmGB_ADDR_CONFIG, 0xffffffff, 0x22011003,
+	mmSPI_RESOURCE_RESERVE_CU_0, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_CU_1, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00FF7FBF,
+	mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00FF7FAF,
+};
+
 static const u32 golden_settings_polaris11_a11[] =
 {
 	mmCB_HW_CONTROL, 0x0000f3cf, 0x00007208,
@@ -650,6 +710,17 @@ static const u32 stoney_mgcg_cgcg_init[] =
 	mmCGTS_SM_CTRL_REG, 0xffffffff, 0x96940200,
 };
 
+
+static const char * const sq_edc_source_names[] = {
+	"SQ_EDC_INFO_SOURCE_INVALID: No EDC error has occurred",
+	"SQ_EDC_INFO_SOURCE_INST: EDC source is Instruction Fetch",
+	"SQ_EDC_INFO_SOURCE_SGPR: EDC source is SGPR or SQC data return",
+	"SQ_EDC_INFO_SOURCE_VGPR: EDC source is VGPR",
+	"SQ_EDC_INFO_SOURCE_LDS: EDC source is LDS",
+	"SQ_EDC_INFO_SOURCE_GDS: EDC source is GDS",
+	"SQ_EDC_INFO_SOURCE_TA: EDC source is TA",
+};
+
 static void gfx_v8_0_set_ring_funcs(struct amdgpu_device *adev);
 static void gfx_v8_0_set_irq_funcs(struct amdgpu_device *adev);
 static void gfx_v8_0_set_gds_init(struct amdgpu_device *adev);
@@ -659,143 +730,138 @@ static void gfx_v8_0_get_cu_info(struct amdgpu_device *adev);
 static void gfx_v8_0_ring_emit_ce_meta(struct amdgpu_ring *ring);
 static void gfx_v8_0_ring_emit_de_meta(struct amdgpu_ring *ring);
 
+#define CG_ACLK_CNTL__ACLK_DIVIDER_MASK                    0x0000007fL
+#define CG_ACLK_CNTL__ACLK_DIVIDER__SHIFT                  0x00000000L
+
 static void gfx_v8_0_init_golden_registers(struct amdgpu_device *adev)
 {
+	uint32_t data;
+
 	switch (adev->asic_type) {
 	case CHIP_TOPAZ:
-		amdgpu_program_register_sequence(adev,
-						 iceland_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(iceland_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_iceland_a11,
-						 (const u32)ARRAY_SIZE(golden_settings_iceland_a11));
-		amdgpu_program_register_sequence(adev,
-						 iceland_golden_common_all,
-						 (const u32)ARRAY_SIZE(iceland_golden_common_all));
+		amdgpu_device_program_register_sequence(adev,
+							iceland_mgcg_cgcg_init,
+							ARRAY_SIZE(iceland_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_iceland_a11,
+							ARRAY_SIZE(golden_settings_iceland_a11));
+		amdgpu_device_program_register_sequence(adev,
+							iceland_golden_common_all,
+							ARRAY_SIZE(iceland_golden_common_all));
 		break;
 	case CHIP_FIJI:
-		amdgpu_program_register_sequence(adev,
-						 fiji_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(fiji_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_fiji_a10,
-						 (const u32)ARRAY_SIZE(golden_settings_fiji_a10));
-		amdgpu_program_register_sequence(adev,
-						 fiji_golden_common_all,
-						 (const u32)ARRAY_SIZE(fiji_golden_common_all));
+		amdgpu_device_program_register_sequence(adev,
+							fiji_mgcg_cgcg_init,
+							ARRAY_SIZE(fiji_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_fiji_a10,
+							ARRAY_SIZE(golden_settings_fiji_a10));
+		amdgpu_device_program_register_sequence(adev,
+							fiji_golden_common_all,
+							ARRAY_SIZE(fiji_golden_common_all));
 		break;
 
 	case CHIP_TONGA:
-		amdgpu_program_register_sequence(adev,
-						 tonga_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(tonga_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_tonga_a11,
-						 (const u32)ARRAY_SIZE(golden_settings_tonga_a11));
-		amdgpu_program_register_sequence(adev,
-						 tonga_golden_common_all,
-						 (const u32)ARRAY_SIZE(tonga_golden_common_all));
+		amdgpu_device_program_register_sequence(adev,
+							tonga_mgcg_cgcg_init,
+							ARRAY_SIZE(tonga_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_tonga_a11,
+							ARRAY_SIZE(golden_settings_tonga_a11));
+		amdgpu_device_program_register_sequence(adev,
+							tonga_golden_common_all,
+							ARRAY_SIZE(tonga_golden_common_all));
+		break;
+	case CHIP_VEGAM:
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_vegam_a11,
+							ARRAY_SIZE(golden_settings_vegam_a11));
+		amdgpu_device_program_register_sequence(adev,
+							vegam_golden_common_all,
+							ARRAY_SIZE(vegam_golden_common_all));
 		break;
 	case CHIP_POLARIS11:
 	case CHIP_POLARIS12:
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_polaris11_a11,
-						 (const u32)ARRAY_SIZE(golden_settings_polaris11_a11));
-		amdgpu_program_register_sequence(adev,
-						 polaris11_golden_common_all,
-						 (const u32)ARRAY_SIZE(polaris11_golden_common_all));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_polaris11_a11,
+							ARRAY_SIZE(golden_settings_polaris11_a11));
+		amdgpu_device_program_register_sequence(adev,
+							polaris11_golden_common_all,
+							ARRAY_SIZE(polaris11_golden_common_all));
 		break;
 	case CHIP_POLARIS10:
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_polaris10_a11,
-						 (const u32)ARRAY_SIZE(golden_settings_polaris10_a11));
-		amdgpu_program_register_sequence(adev,
-						 polaris10_golden_common_all,
-						 (const u32)ARRAY_SIZE(polaris10_golden_common_all));
-		WREG32_SMC(ixCG_ACLK_CNTL, 0x0000001C);
-		if (adev->pdev->revision == 0xc7 &&
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_polaris10_a11,
+							ARRAY_SIZE(golden_settings_polaris10_a11));
+		amdgpu_device_program_register_sequence(adev,
+							polaris10_golden_common_all,
+							ARRAY_SIZE(polaris10_golden_common_all));
+		data = RREG32_SMC(ixCG_ACLK_CNTL);
+		data &= ~CG_ACLK_CNTL__ACLK_DIVIDER_MASK;
+		data |= 0x18 << CG_ACLK_CNTL__ACLK_DIVIDER__SHIFT;
+		WREG32_SMC(ixCG_ACLK_CNTL, data);
+		if ((adev->pdev->device == 0x67DF) && (adev->pdev->revision == 0xc7) &&
 		    ((adev->pdev->subsystem_device == 0xb37 && adev->pdev->subsystem_vendor == 0x1002) ||
 		     (adev->pdev->subsystem_device == 0x4a8 && adev->pdev->subsystem_vendor == 0x1043) ||
-		     (adev->pdev->subsystem_device == 0x9480 && adev->pdev->subsystem_vendor == 0x1682))) {
+		     (adev->pdev->subsystem_device == 0x9480 && adev->pdev->subsystem_vendor == 0x1680))) {
 			amdgpu_atombios_i2c_channel_trans(adev, 0x10, 0x96, 0x1E, 0xDD);
 			amdgpu_atombios_i2c_channel_trans(adev, 0x10, 0x96, 0x1F, 0xD0);
 		}
 		break;
 	case CHIP_CARRIZO:
-		amdgpu_program_register_sequence(adev,
-						 cz_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(cz_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 cz_golden_settings_a11,
-						 (const u32)ARRAY_SIZE(cz_golden_settings_a11));
-		amdgpu_program_register_sequence(adev,
-						 cz_golden_common_all,
-						 (const u32)ARRAY_SIZE(cz_golden_common_all));
+		amdgpu_device_program_register_sequence(adev,
+							cz_mgcg_cgcg_init,
+							ARRAY_SIZE(cz_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							cz_golden_settings_a11,
+							ARRAY_SIZE(cz_golden_settings_a11));
+		amdgpu_device_program_register_sequence(adev,
+							cz_golden_common_all,
+							ARRAY_SIZE(cz_golden_common_all));
 		break;
 	case CHIP_STONEY:
-		amdgpu_program_register_sequence(adev,
-						 stoney_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(stoney_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 stoney_golden_settings_a11,
-						 (const u32)ARRAY_SIZE(stoney_golden_settings_a11));
-		amdgpu_program_register_sequence(adev,
-						 stoney_golden_common_all,
-						 (const u32)ARRAY_SIZE(stoney_golden_common_all));
+		amdgpu_device_program_register_sequence(adev,
+							stoney_mgcg_cgcg_init,
+							ARRAY_SIZE(stoney_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							stoney_golden_settings_a11,
+							ARRAY_SIZE(stoney_golden_settings_a11));
+		amdgpu_device_program_register_sequence(adev,
+							stoney_golden_common_all,
+							ARRAY_SIZE(stoney_golden_common_all));
 		break;
 	default:
 		break;
 	}
 }
 
-static void gfx_v8_0_scratch_init(struct amdgpu_device *adev)
-{
-	adev->gfx.scratch.num_reg = 8;
-	adev->gfx.scratch.reg_base = mmSCRATCH_REG0;
-	adev->gfx.scratch.free_mask = (1u << adev->gfx.scratch.num_reg) - 1;
-}
-
 static int gfx_v8_0_ring_test_ring(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
-	uint32_t scratch;
 	uint32_t tmp = 0;
 	unsigned i;
 	int r;
 
-	r = amdgpu_gfx_scratch_get(adev, &scratch);
-	if (r) {
-		DRM_ERROR("amdgpu: cp failed to get scratch reg (%d).\n", r);
-		return r;
-	}
-	WREG32(scratch, 0xCAFEDEAD);
+	WREG32(mmSCRATCH_REG0, 0xCAFEDEAD);
 	r = amdgpu_ring_alloc(ring, 3);
-	if (r) {
-		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n",
-			  ring->idx, r);
-		amdgpu_gfx_scratch_free(adev, scratch);
+	if (r)
 		return r;
-	}
+
 	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_UCONFIG_REG, 1));
-	amdgpu_ring_write(ring, (scratch - PACKET3_SET_UCONFIG_REG_START));
+	amdgpu_ring_write(ring, mmSCRATCH_REG0 - PACKET3_SET_UCONFIG_REG_START);
 	amdgpu_ring_write(ring, 0xDEADBEEF);
 	amdgpu_ring_commit(ring);
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		tmp = RREG32(scratch);
+		tmp = RREG32(mmSCRATCH_REG0);
 		if (tmp == 0xDEADBEEF)
 			break;
-		DRM_UDELAY(1);
-	}
-	if (i < adev->usec_timeout) {
-		DRM_INFO("ring test on %d succeeded in %d usecs\n",
-			 ring->idx, i);
-	} else {
-		DRM_ERROR("amdgpu: ring %d test failed (scratch(0x%04X)=0x%08X)\n",
-			  ring->idx, scratch, tmp);
-		r = -EINVAL;
+		udelay(1);
 	}
-	amdgpu_gfx_scratch_free(adev, scratch);
+
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+
 	return r;
 }
 
@@ -804,26 +870,30 @@ static int gfx_v8_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 	struct amdgpu_device *adev = ring->adev;
 	struct amdgpu_ib ib;
 	struct dma_fence *f = NULL;
-	uint32_t scratch;
-	uint32_t tmp = 0;
+
+	unsigned int index;
+	uint64_t gpu_addr;
+	uint32_t tmp;
 	long r;
 
-	r = amdgpu_gfx_scratch_get(adev, &scratch);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get scratch reg (%ld).\n", r);
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r)
 		return r;
-	}
-	WREG32(scratch, 0xCAFEDEAD);
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	adev->wb.wb[index] = cpu_to_le32(0xCAFEDEAD);
 	memset(&ib, 0, sizeof(ib));
-	r = amdgpu_ib_get(adev, NULL, 256, &ib);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
+
+	r = amdgpu_ib_get(adev, NULL, 20, AMDGPU_IB_POOL_DIRECT, &ib);
+	if (r)
 		goto err1;
-	}
-	ib.ptr[0] = PACKET3(PACKET3_SET_UCONFIG_REG, 1);
-	ib.ptr[1] = ((scratch - PACKET3_SET_UCONFIG_REG_START));
-	ib.ptr[2] = 0xDEADBEEF;
-	ib.length_dw = 3;
+
+	ib.ptr[0] = PACKET3(PACKET3_WRITE_DATA, 3);
+	ib.ptr[1] = WRITE_DATA_DST_SEL(5) | WR_CONFIRM;
+	ib.ptr[2] = lower_32_bits(gpu_addr);
+	ib.ptr[3] = upper_32_bits(gpu_addr);
+	ib.ptr[4] = 0xDEADBEEF;
+	ib.length_dw = 5;
 
 	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
 	if (r)
@@ -831,47 +901,37 @@ static int gfx_v8_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 
 	r = dma_fence_wait_timeout(f, false, timeout);
 	if (r == 0) {
-		DRM_ERROR("amdgpu: IB test timed out.\n");
 		r = -ETIMEDOUT;
 		goto err2;
 	} else if (r < 0) {
-		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
 		goto err2;
 	}
-	tmp = RREG32(scratch);
-	if (tmp == 0xDEADBEEF) {
-		DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
+
+	tmp = adev->wb.wb[index];
+	if (tmp == 0xDEADBEEF)
 		r = 0;
-	} else {
-		DRM_ERROR("amdgpu: ib test failed (scratch(0x%04X)=0x%08X)\n",
-			  scratch, tmp);
+	else
 		r = -EINVAL;
-	}
+
 err2:
-	amdgpu_ib_free(adev, &ib, NULL);
+	amdgpu_ib_free(&ib, NULL);
 	dma_fence_put(f);
 err1:
-	amdgpu_gfx_scratch_free(adev, scratch);
+	amdgpu_device_wb_free(adev, index);
 	return r;
 }
 
 
 static void gfx_v8_0_free_microcode(struct amdgpu_device *adev)
 {
-	release_firmware(adev->gfx.pfp_fw);
-	adev->gfx.pfp_fw = NULL;
-	release_firmware(adev->gfx.me_fw);
-	adev->gfx.me_fw = NULL;
-	release_firmware(adev->gfx.ce_fw);
-	adev->gfx.ce_fw = NULL;
-	release_firmware(adev->gfx.rlc_fw);
-	adev->gfx.rlc_fw = NULL;
-	release_firmware(adev->gfx.mec_fw);
-	adev->gfx.mec_fw = NULL;
+	amdgpu_ucode_release(&adev->gfx.pfp_fw);
+	amdgpu_ucode_release(&adev->gfx.me_fw);
+	amdgpu_ucode_release(&adev->gfx.ce_fw);
+	amdgpu_ucode_release(&adev->gfx.rlc_fw);
+	amdgpu_ucode_release(&adev->gfx.mec_fw);
 	if ((adev->asic_type != CHIP_STONEY) &&
 	    (adev->asic_type != CHIP_TOPAZ))
-		release_firmware(adev->gfx.mec2_fw);
-	adev->gfx.mec2_fw = NULL;
+		amdgpu_ucode_release(&adev->gfx.mec2_fw);
 
 	kfree(adev->gfx.rlc.register_list_format);
 }
@@ -879,7 +939,6 @@ static void gfx_v8_0_free_microcode(struct amdgpu_device *adev)
 static int gfx_v8_0_init_microcode(struct amdgpu_device *adev)
 {
 	const char *chip_name;
-	char fw_name[30];
 	int err;
 	struct amdgpu_firmware_info *info = NULL;
 	const struct common_firmware_header *header = NULL;
@@ -902,38 +961,59 @@ static int gfx_v8_0_init_microcode(struct amdgpu_device *adev)
 	case CHIP_FIJI:
 		chip_name = "fiji";
 		break;
-	case CHIP_POLARIS11:
-		chip_name = "polaris11";
+	case CHIP_STONEY:
+		chip_name = "stoney";
 		break;
 	case CHIP_POLARIS10:
 		chip_name = "polaris10";
 		break;
+	case CHIP_POLARIS11:
+		chip_name = "polaris11";
+		break;
 	case CHIP_POLARIS12:
 		chip_name = "polaris12";
 		break;
-	case CHIP_STONEY:
-		chip_name = "stoney";
+	case CHIP_VEGAM:
+		chip_name = "vegam";
 		break;
 	default:
 		BUG();
 	}
 
-	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_pfp.bin", chip_name);
-	err = request_firmware(&adev->gfx.pfp_fw, fw_name, adev->dev);
-	if (err)
-		goto out;
-	err = amdgpu_ucode_validate(adev->gfx.pfp_fw);
+	if (adev->asic_type >= CHIP_POLARIS10 && adev->asic_type <= CHIP_POLARIS12) {
+		err = amdgpu_ucode_request(adev, &adev->gfx.pfp_fw,
+					   AMDGPU_UCODE_OPTIONAL,
+					   "amdgpu/%s_pfp_2.bin", chip_name);
+		if (err == -ENODEV) {
+			err = amdgpu_ucode_request(adev, &adev->gfx.pfp_fw,
+						   AMDGPU_UCODE_REQUIRED,
+						   "amdgpu/%s_pfp.bin", chip_name);
+		}
+	} else {
+		err = amdgpu_ucode_request(adev, &adev->gfx.pfp_fw,
+					   AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_pfp.bin", chip_name);
+	}
 	if (err)
 		goto out;
 	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data;
 	adev->gfx.pfp_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
 	adev->gfx.pfp_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
 
-	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_me.bin", chip_name);
-	err = request_firmware(&adev->gfx.me_fw, fw_name, adev->dev);
-	if (err)
-		goto out;
-	err = amdgpu_ucode_validate(adev->gfx.me_fw);
+	if (adev->asic_type >= CHIP_POLARIS10 && adev->asic_type <= CHIP_POLARIS12) {
+		err = amdgpu_ucode_request(adev, &adev->gfx.me_fw,
+					   AMDGPU_UCODE_OPTIONAL,
+					   "amdgpu/%s_me_2.bin", chip_name);
+		if (err == -ENODEV) {
+			err = amdgpu_ucode_request(adev, &adev->gfx.me_fw,
+						   AMDGPU_UCODE_REQUIRED,
+						   "amdgpu/%s_me.bin", chip_name);
+		}
+	} else {
+		err = amdgpu_ucode_request(adev, &adev->gfx.me_fw,
+					   AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_me.bin", chip_name);
+	}
 	if (err)
 		goto out;
 	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data;
@@ -941,11 +1021,20 @@ static int gfx_v8_0_init_microcode(struct amdgpu_device *adev)
 
 	adev->gfx.me_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
 
-	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_ce.bin", chip_name);
-	err = request_firmware(&adev->gfx.ce_fw, fw_name, adev->dev);
-	if (err)
-		goto out;
-	err = amdgpu_ucode_validate(adev->gfx.ce_fw);
+	if (adev->asic_type >= CHIP_POLARIS10 && adev->asic_type <= CHIP_POLARIS12) {
+		err = amdgpu_ucode_request(adev, &adev->gfx.ce_fw,
+					   AMDGPU_UCODE_OPTIONAL,
+					   "amdgpu/%s_ce_2.bin", chip_name);
+		if (err == -ENODEV) {
+			err = amdgpu_ucode_request(adev, &adev->gfx.ce_fw,
+						   AMDGPU_UCODE_REQUIRED,
+						   "amdgpu/%s_ce.bin", chip_name);
+		}
+	} else {
+		err = amdgpu_ucode_request(adev, &adev->gfx.ce_fw,
+					   AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_ce.bin", chip_name);
+	}
 	if (err)
 		goto out;
 	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data;
@@ -963,11 +1052,11 @@ static int gfx_v8_0_init_microcode(struct amdgpu_device *adev)
 	} else
 		adev->virt.chained_ib_support = false;
 
-	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_rlc.bin", chip_name);
-	err = request_firmware(&adev->gfx.rlc_fw, fw_name, adev->dev);
+	err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_rlc.bin", chip_name);
 	if (err)
 		goto out;
-	err = amdgpu_ucode_validate(adev->gfx.rlc_fw);
 	rlc_hdr = (const struct rlc_firmware_header_v2_0 *)adev->gfx.rlc_fw->data;
 	adev->gfx.rlc_fw_version = le32_to_cpu(rlc_hdr->header.ucode_version);
 	adev->gfx.rlc_feature_version = le32_to_cpu(rlc_hdr->ucode_feature_version);
@@ -1002,21 +1091,30 @@ static int gfx_v8_0_init_microcode(struct amdgpu_device *adev)
 
 	tmp = (unsigned int *)((uintptr_t)rlc_hdr +
 			le32_to_cpu(rlc_hdr->reg_list_format_array_offset_bytes));
-	for (i = 0 ; i < (rlc_hdr->reg_list_format_size_bytes >> 2); i++)
+	for (i = 0 ; i < (adev->gfx.rlc.reg_list_format_size_bytes >> 2); i++)
 		adev->gfx.rlc.register_list_format[i] =	le32_to_cpu(tmp[i]);
 
 	adev->gfx.rlc.register_restore = adev->gfx.rlc.register_list_format + i;
 
 	tmp = (unsigned int *)((uintptr_t)rlc_hdr +
 			le32_to_cpu(rlc_hdr->reg_list_array_offset_bytes));
-	for (i = 0 ; i < (rlc_hdr->reg_list_size_bytes >> 2); i++)
+	for (i = 0 ; i < (adev->gfx.rlc.reg_list_size_bytes >> 2); i++)
 		adev->gfx.rlc.register_restore[i] = le32_to_cpu(tmp[i]);
 
-	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mec.bin", chip_name);
-	err = request_firmware(&adev->gfx.mec_fw, fw_name, adev->dev);
-	if (err)
-		goto out;
-	err = amdgpu_ucode_validate(adev->gfx.mec_fw);
+	if (adev->asic_type >= CHIP_POLARIS10 && adev->asic_type <= CHIP_POLARIS12) {
+		err = amdgpu_ucode_request(adev, &adev->gfx.mec_fw,
+					   AMDGPU_UCODE_OPTIONAL,
+					   "amdgpu/%s_mec_2.bin", chip_name);
+		if (err == -ENODEV) {
+			err = amdgpu_ucode_request(adev, &adev->gfx.mec_fw,
+						   AMDGPU_UCODE_REQUIRED,
+						   "amdgpu/%s_mec.bin", chip_name);
+		}
+	} else {
+		err = amdgpu_ucode_request(adev, &adev->gfx.mec_fw,
+					   AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_mec.bin", chip_name);
+	}
 	if (err)
 		goto out;
 	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
@@ -1025,12 +1123,21 @@ static int gfx_v8_0_init_microcode(struct amdgpu_device *adev)
 
 	if ((adev->asic_type != CHIP_STONEY) &&
 	    (adev->asic_type != CHIP_TOPAZ)) {
-		snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mec2.bin", chip_name);
-		err = request_firmware(&adev->gfx.mec2_fw, fw_name, adev->dev);
+		if (adev->asic_type >= CHIP_POLARIS10 && adev->asic_type <= CHIP_POLARIS12) {
+			err = amdgpu_ucode_request(adev, &adev->gfx.mec2_fw,
+						   AMDGPU_UCODE_OPTIONAL,
+						   "amdgpu/%s_mec2_2.bin", chip_name);
+			if (err == -ENODEV) {
+				err = amdgpu_ucode_request(adev, &adev->gfx.mec2_fw,
+							   AMDGPU_UCODE_REQUIRED,
+							   "amdgpu/%s_mec2.bin", chip_name);
+			}
+		} else {
+			err = amdgpu_ucode_request(adev, &adev->gfx.mec2_fw,
+						   AMDGPU_UCODE_REQUIRED,
+						   "amdgpu/%s_mec2.bin", chip_name);
+		}
 		if (!err) {
-			err = amdgpu_ucode_validate(adev->gfx.mec2_fw);
-			if (err)
-				goto out;
 			cp_hdr = (const struct gfx_firmware_header_v1_0 *)
 				adev->gfx.mec2_fw->data;
 			adev->gfx.mec2_fw_version =
@@ -1043,209 +1150,106 @@ static int gfx_v8_0_init_microcode(struct amdgpu_device *adev)
 		}
 	}
 
-	if (adev->firmware.load_type == AMDGPU_FW_LOAD_SMU) {
-		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_PFP];
-		info->ucode_id = AMDGPU_UCODE_ID_CP_PFP;
-		info->fw = adev->gfx.pfp_fw;
-		header = (const struct common_firmware_header *)info->fw->data;
-		adev->firmware.fw_size +=
-			ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
-
-		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_ME];
-		info->ucode_id = AMDGPU_UCODE_ID_CP_ME;
-		info->fw = adev->gfx.me_fw;
-		header = (const struct common_firmware_header *)info->fw->data;
-		adev->firmware.fw_size +=
-			ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
-
-		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_CE];
-		info->ucode_id = AMDGPU_UCODE_ID_CP_CE;
-		info->fw = adev->gfx.ce_fw;
-		header = (const struct common_firmware_header *)info->fw->data;
-		adev->firmware.fw_size +=
-			ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+	info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_PFP];
+	info->ucode_id = AMDGPU_UCODE_ID_CP_PFP;
+	info->fw = adev->gfx.pfp_fw;
+	header = (const struct common_firmware_header *)info->fw->data;
+	adev->firmware.fw_size +=
+		ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+
+	info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_ME];
+	info->ucode_id = AMDGPU_UCODE_ID_CP_ME;
+	info->fw = adev->gfx.me_fw;
+	header = (const struct common_firmware_header *)info->fw->data;
+	adev->firmware.fw_size +=
+		ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+
+	info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_CE];
+	info->ucode_id = AMDGPU_UCODE_ID_CP_CE;
+	info->fw = adev->gfx.ce_fw;
+	header = (const struct common_firmware_header *)info->fw->data;
+	adev->firmware.fw_size +=
+		ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+
+	info = &adev->firmware.ucode[AMDGPU_UCODE_ID_RLC_G];
+	info->ucode_id = AMDGPU_UCODE_ID_RLC_G;
+	info->fw = adev->gfx.rlc_fw;
+	header = (const struct common_firmware_header *)info->fw->data;
+	adev->firmware.fw_size +=
+		ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+
+	info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_MEC1];
+	info->ucode_id = AMDGPU_UCODE_ID_CP_MEC1;
+	info->fw = adev->gfx.mec_fw;
+	header = (const struct common_firmware_header *)info->fw->data;
+	adev->firmware.fw_size +=
+		ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+
+	/* we need account JT in */
+	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+	adev->firmware.fw_size +=
+		ALIGN(le32_to_cpu(cp_hdr->jt_size) << 2, PAGE_SIZE);
 
-		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_RLC_G];
-		info->ucode_id = AMDGPU_UCODE_ID_RLC_G;
-		info->fw = adev->gfx.rlc_fw;
-		header = (const struct common_firmware_header *)info->fw->data;
+	if (amdgpu_sriov_vf(adev)) {
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_STORAGE];
+		info->ucode_id = AMDGPU_UCODE_ID_STORAGE;
+		info->fw = adev->gfx.mec_fw;
 		adev->firmware.fw_size +=
-			ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+			ALIGN(le32_to_cpu(64 * PAGE_SIZE), PAGE_SIZE);
+	}
 
-		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_MEC1];
-		info->ucode_id = AMDGPU_UCODE_ID_CP_MEC1;
-		info->fw = adev->gfx.mec_fw;
+	if (adev->gfx.mec2_fw) {
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_MEC2];
+		info->ucode_id = AMDGPU_UCODE_ID_CP_MEC2;
+		info->fw = adev->gfx.mec2_fw;
 		header = (const struct common_firmware_header *)info->fw->data;
 		adev->firmware.fw_size +=
 			ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
-
-		/* we need account JT in */
-		cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
-		adev->firmware.fw_size +=
-			ALIGN(le32_to_cpu(cp_hdr->jt_size) << 2, PAGE_SIZE);
-
-		if (amdgpu_sriov_vf(adev)) {
-			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_STORAGE];
-			info->ucode_id = AMDGPU_UCODE_ID_STORAGE;
-			info->fw = adev->gfx.mec_fw;
-			adev->firmware.fw_size +=
-				ALIGN(le32_to_cpu(64 * PAGE_SIZE), PAGE_SIZE);
-		}
-
-		if (adev->gfx.mec2_fw) {
-			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_MEC2];
-			info->ucode_id = AMDGPU_UCODE_ID_CP_MEC2;
-			info->fw = adev->gfx.mec2_fw;
-			header = (const struct common_firmware_header *)info->fw->data;
-			adev->firmware.fw_size +=
-				ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
-		}
-
 	}
 
 out:
 	if (err) {
-		dev_err(adev->dev,
-			"gfx8: Failed to load firmware \"%s\"\n",
-			fw_name);
-		release_firmware(adev->gfx.pfp_fw);
-		adev->gfx.pfp_fw = NULL;
-		release_firmware(adev->gfx.me_fw);
-		adev->gfx.me_fw = NULL;
-		release_firmware(adev->gfx.ce_fw);
-		adev->gfx.ce_fw = NULL;
-		release_firmware(adev->gfx.rlc_fw);
-		adev->gfx.rlc_fw = NULL;
-		release_firmware(adev->gfx.mec_fw);
-		adev->gfx.mec_fw = NULL;
-		release_firmware(adev->gfx.mec2_fw);
-		adev->gfx.mec2_fw = NULL;
+		dev_err(adev->dev, "gfx8: Failed to load firmware %s gfx firmware\n", chip_name);
+		amdgpu_ucode_release(&adev->gfx.pfp_fw);
+		amdgpu_ucode_release(&adev->gfx.me_fw);
+		amdgpu_ucode_release(&adev->gfx.ce_fw);
+		amdgpu_ucode_release(&adev->gfx.rlc_fw);
+		amdgpu_ucode_release(&adev->gfx.mec_fw);
+		amdgpu_ucode_release(&adev->gfx.mec2_fw);
 	}
 	return err;
 }
 
-static void gfx_v8_0_get_csb_buffer(struct amdgpu_device *adev,
-				    volatile u32 *buffer)
+static void gfx_v8_0_get_csb_buffer(struct amdgpu_device *adev, u32 *buffer)
 {
-	u32 count = 0, i;
-	const struct cs_section_def *sect = NULL;
-	const struct cs_extent_def *ext = NULL;
+	u32 count = 0;
 
 	if (adev->gfx.rlc.cs_data == NULL)
 		return;
 	if (buffer == NULL)
 		return;
 
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
-	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
-
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CONTEXT_CONTROL, 1));
-	buffer[count++] = cpu_to_le32(0x80000000);
-	buffer[count++] = cpu_to_le32(0x80000000);
-
-	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
-		for (ext = sect->section; ext->extent != NULL; ++ext) {
-			if (sect->id == SECT_CONTEXT) {
-				buffer[count++] =
-					cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, ext->reg_count));
-				buffer[count++] = cpu_to_le32(ext->reg_index -
-						PACKET3_SET_CONTEXT_REG_START);
-				for (i = 0; i < ext->reg_count; i++)
-					buffer[count++] = cpu_to_le32(ext->extent[i]);
-			} else {
-				return;
-			}
-		}
-	}
+	count = amdgpu_gfx_csb_preamble_start(buffer);
+	count = amdgpu_gfx_csb_data_parser(adev, buffer, count);
 
 	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, 2));
-	buffer[count++] = cpu_to_le32(mmPA_SC_RASTER_CONFIG -
-			PACKET3_SET_CONTEXT_REG_START);
+	buffer[count++] = cpu_to_le32(mmPA_SC_RASTER_CONFIG - PACKET3_SET_CONTEXT_REG_START);
 	buffer[count++] = cpu_to_le32(adev->gfx.config.rb_config[0][0].raster_config);
 	buffer[count++] = cpu_to_le32(adev->gfx.config.rb_config[0][0].raster_config_1);
 
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
-	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_END_CLEAR_STATE);
-
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CLEAR_STATE, 0));
-	buffer[count++] = cpu_to_le32(0);
+	amdgpu_gfx_csb_preamble_end(buffer, count);
 }
 
-static void cz_init_cp_jump_table(struct amdgpu_device *adev)
+static int gfx_v8_0_cp_jump_table_num(struct amdgpu_device *adev)
 {
-	const __le32 *fw_data;
-	volatile u32 *dst_ptr;
-	int me, i, max_me = 4;
-	u32 bo_offset = 0;
-	u32 table_offset, table_size;
-
 	if (adev->asic_type == CHIP_CARRIZO)
-		max_me = 5;
-
-	/* write the cp table buffer */
-	dst_ptr = adev->gfx.rlc.cp_table_ptr;
-	for (me = 0; me < max_me; me++) {
-		if (me == 0) {
-			const struct gfx_firmware_header_v1_0 *hdr =
-				(const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data;
-			fw_data = (const __le32 *)
-				(adev->gfx.ce_fw->data +
-				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
-			table_offset = le32_to_cpu(hdr->jt_offset);
-			table_size = le32_to_cpu(hdr->jt_size);
-		} else if (me == 1) {
-			const struct gfx_firmware_header_v1_0 *hdr =
-				(const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data;
-			fw_data = (const __le32 *)
-				(adev->gfx.pfp_fw->data +
-				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
-			table_offset = le32_to_cpu(hdr->jt_offset);
-			table_size = le32_to_cpu(hdr->jt_size);
-		} else if (me == 2) {
-			const struct gfx_firmware_header_v1_0 *hdr =
-				(const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data;
-			fw_data = (const __le32 *)
-				(adev->gfx.me_fw->data +
-				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
-			table_offset = le32_to_cpu(hdr->jt_offset);
-			table_size = le32_to_cpu(hdr->jt_size);
-		} else if (me == 3) {
-			const struct gfx_firmware_header_v1_0 *hdr =
-				(const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
-			fw_data = (const __le32 *)
-				(adev->gfx.mec_fw->data +
-				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
-			table_offset = le32_to_cpu(hdr->jt_offset);
-			table_size = le32_to_cpu(hdr->jt_size);
-		} else  if (me == 4) {
-			const struct gfx_firmware_header_v1_0 *hdr =
-				(const struct gfx_firmware_header_v1_0 *)adev->gfx.mec2_fw->data;
-			fw_data = (const __le32 *)
-				(adev->gfx.mec2_fw->data +
-				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
-			table_offset = le32_to_cpu(hdr->jt_offset);
-			table_size = le32_to_cpu(hdr->jt_size);
-		}
-
-		for (i = 0; i < table_size; i ++) {
-			dst_ptr[bo_offset + i] =
-				cpu_to_le32(le32_to_cpu(fw_data[table_offset + i]));
-		}
-
-		bo_offset += table_size;
-	}
-}
-
-static void gfx_v8_0_rlc_fini(struct amdgpu_device *adev)
-{
-	amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj, NULL, NULL);
-	amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj, NULL, NULL);
+		return 5;
+	else
+		return 4;
 }
 
 static int gfx_v8_0_rlc_init(struct amdgpu_device *adev)
 {
-	volatile u32 *dst_ptr;
-	u32 dws;
 	const struct cs_section_def *cs_data;
 	int r;
 
@@ -1254,46 +1258,24 @@ static int gfx_v8_0_rlc_init(struct amdgpu_device *adev)
 	cs_data = adev->gfx.rlc.cs_data;
 
 	if (cs_data) {
-		/* clear state block */
-		adev->gfx.rlc.clear_state_size = dws = gfx_v8_0_get_csb_size(adev);
-
-		r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
-					      AMDGPU_GEM_DOMAIN_VRAM,
-					      &adev->gfx.rlc.clear_state_obj,
-					      &adev->gfx.rlc.clear_state_gpu_addr,
-					      (void **)&adev->gfx.rlc.cs_ptr);
-		if (r) {
-			dev_warn(adev->dev, "(%d) create RLC c bo failed\n", r);
-			gfx_v8_0_rlc_fini(adev);
+		/* init clear state block */
+		r = amdgpu_gfx_rlc_init_csb(adev);
+		if (r)
 			return r;
-		}
-
-		/* set up the cs buffer */
-		dst_ptr = adev->gfx.rlc.cs_ptr;
-		gfx_v8_0_get_csb_buffer(adev, dst_ptr);
-		amdgpu_bo_kunmap(adev->gfx.rlc.clear_state_obj);
-		amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
 	}
 
 	if ((adev->asic_type == CHIP_CARRIZO) ||
 	    (adev->asic_type == CHIP_STONEY)) {
 		adev->gfx.rlc.cp_table_size = ALIGN(96 * 5 * 4, 2048) + (64 * 1024); /* JT + GDS */
-		r = amdgpu_bo_create_reserved(adev, adev->gfx.rlc.cp_table_size,
-					      PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
-					      &adev->gfx.rlc.cp_table_obj,
-					      &adev->gfx.rlc.cp_table_gpu_addr,
-					      (void **)&adev->gfx.rlc.cp_table_ptr);
-		if (r) {
-			dev_warn(adev->dev, "(%d) create RLC cp table bo failed\n", r);
+		r = amdgpu_gfx_rlc_init_cpt(adev);
+		if (r)
 			return r;
-		}
-
-		cz_init_cp_jump_table(adev);
-
-		amdgpu_bo_kunmap(adev->gfx.rlc.cp_table_obj);
-		amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj);
 	}
 
+	/* init spm vmid with 0xf */
+	if (adev->gfx.rlc.funcs->update_spm_vmid)
+		adev->gfx.rlc.funcs->update_spm_vmid(adev, NULL, 0xf);
+
 	return 0;
 }
 
@@ -1308,27 +1290,29 @@ static int gfx_v8_0_mec_init(struct amdgpu_device *adev)
 	u32 *hpd;
 	size_t mec_hpd_size;
 
-	bitmap_zero(adev->gfx.mec.queue_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
+	bitmap_zero(adev->gfx.mec_bitmap[0].queue_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
 
 	/* take ownership of the relevant compute queues */
 	amdgpu_gfx_compute_queue_acquire(adev);
 
 	mec_hpd_size = adev->gfx.num_compute_rings * GFX8_MEC_HPD_SIZE;
+	if (mec_hpd_size) {
+		r = amdgpu_bo_create_reserved(adev, mec_hpd_size, PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_VRAM |
+					      AMDGPU_GEM_DOMAIN_GTT,
+					      &adev->gfx.mec.hpd_eop_obj,
+					      &adev->gfx.mec.hpd_eop_gpu_addr,
+					      (void **)&hpd);
+		if (r) {
+			dev_warn(adev->dev, "(%d) create HDP EOP bo failed\n", r);
+			return r;
+		}
 
-	r = amdgpu_bo_create_reserved(adev, mec_hpd_size, PAGE_SIZE,
-				      AMDGPU_GEM_DOMAIN_GTT,
-				      &adev->gfx.mec.hpd_eop_obj,
-				      &adev->gfx.mec.hpd_eop_gpu_addr,
-				      (void **)&hpd);
-	if (r) {
-		dev_warn(adev->dev, "(%d) create HDP EOP bo failed\n", r);
-		return r;
-	}
-
-	memset(hpd, 0, mec_hpd_size);
+		memset(hpd, 0, mec_hpd_size);
 
-	amdgpu_bo_kunmap(adev->gfx.mec.hpd_eop_obj);
-	amdgpu_bo_unreserve(adev->gfx.mec.hpd_eop_obj);
+		amdgpu_bo_kunmap(adev->gfx.mec.hpd_eop_obj);
+		amdgpu_bo_unreserve(adev->gfx.mec.hpd_eop_obj);
+	}
 
 	return 0;
 }
@@ -1398,10 +1382,11 @@ static const u32 sgpr_init_compute_shader[] =
 static const u32 vgpr_init_regs[] =
 {
 	mmCOMPUTE_STATIC_THREAD_MGMT_SE0, 0xffffffff,
-	mmCOMPUTE_RESOURCE_LIMITS, 0,
+	mmCOMPUTE_RESOURCE_LIMITS, 0x1000000, /* CU_GROUP_COUNT=1 */
 	mmCOMPUTE_NUM_THREAD_X, 256*4,
 	mmCOMPUTE_NUM_THREAD_Y, 1,
 	mmCOMPUTE_NUM_THREAD_Z, 1,
+	mmCOMPUTE_PGM_RSRC1, 0x100004f, /* VGPRS=15 (64 logical VGPRs), SGPRS=1 (16 SGPRs), BULKY=1 */
 	mmCOMPUTE_PGM_RSRC2, 20,
 	mmCOMPUTE_USER_DATA_0, 0xedcedc00,
 	mmCOMPUTE_USER_DATA_1, 0xedcedc01,
@@ -1418,10 +1403,11 @@ static const u32 vgpr_init_regs[] =
 static const u32 sgpr1_init_regs[] =
 {
 	mmCOMPUTE_STATIC_THREAD_MGMT_SE0, 0x0f,
-	mmCOMPUTE_RESOURCE_LIMITS, 0x1000000,
+	mmCOMPUTE_RESOURCE_LIMITS, 0x1000000, /* CU_GROUP_COUNT=1 */
 	mmCOMPUTE_NUM_THREAD_X, 256*5,
 	mmCOMPUTE_NUM_THREAD_Y, 1,
 	mmCOMPUTE_NUM_THREAD_Z, 1,
+	mmCOMPUTE_PGM_RSRC1, 0x240, /* SGPRS=9 (80 GPRS) */
 	mmCOMPUTE_PGM_RSRC2, 20,
 	mmCOMPUTE_USER_DATA_0, 0xedcedc00,
 	mmCOMPUTE_USER_DATA_1, 0xedcedc01,
@@ -1442,6 +1428,7 @@ static const u32 sgpr2_init_regs[] =
 	mmCOMPUTE_NUM_THREAD_X, 256*5,
 	mmCOMPUTE_NUM_THREAD_Y, 1,
 	mmCOMPUTE_NUM_THREAD_Z, 1,
+	mmCOMPUTE_PGM_RSRC1, 0x240, /* SGPRS=9 (80 GPRS) */
 	mmCOMPUTE_PGM_RSRC2, 20,
 	mmCOMPUTE_USER_DATA_0, 0xedcedc00,
 	mmCOMPUTE_USER_DATA_1, 0xedcedc01,
@@ -1499,7 +1486,7 @@ static int gfx_v8_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
 		return 0;
 
 	/* bail if the compute ring is not ready */
-	if (!ring->ready)
+	if (!ring->sched.ready)
 		return 0;
 
 	tmp = RREG32(mmGB_EDC_MODE);
@@ -1519,7 +1506,8 @@ static int gfx_v8_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
 
 	/* allocate an indirect buffer to put the commands in */
 	memset(&ib, 0, sizeof(ib));
-	r = amdgpu_ib_get(adev, NULL, total_size, &ib);
+	r = amdgpu_ib_get(adev, NULL, total_size,
+					AMDGPU_IB_POOL_DIRECT, &ib);
 	if (r) {
 		DRM_ERROR("amdgpu: failed to get ib (%d).\n", r);
 		return r;
@@ -1641,7 +1629,7 @@ static int gfx_v8_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
 		RREG32(sec_ded_counter_registers[i]);
 
 fail:
-	amdgpu_ib_free(adev, &ib, NULL);
+	amdgpu_ib_free(&ib, NULL);
 	dma_fence_put(f);
 
 	return r;
@@ -1650,7 +1638,7 @@ fail:
 static int gfx_v8_0_gpu_early_init(struct amdgpu_device *adev)
 {
 	u32 gb_addr_config;
-	u32 mc_shared_chmap, mc_arb_ramcfg;
+	u32 mc_arb_ramcfg;
 	u32 dimm00_addr_map, dimm01_addr_map, dimm10_addr_map, dimm11_addr_map;
 	u32 tmp;
 	int ret;
@@ -1706,6 +1694,7 @@ static int gfx_v8_0_gpu_early_init(struct amdgpu_device *adev)
 		gb_addr_config = POLARIS11_GB_ADDR_CONFIG_GOLDEN;
 		break;
 	case CHIP_POLARIS10:
+	case CHIP_VEGAM:
 		ret = amdgpu_atombios_get_gfx_info(adev);
 		if (ret)
 			return ret;
@@ -1789,10 +1778,14 @@ static int gfx_v8_0_gpu_early_init(struct amdgpu_device *adev)
 		break;
 	}
 
-	mc_shared_chmap = RREG32(mmMC_SHARED_CHMAP);
 	adev->gfx.config.mc_arb_ramcfg = RREG32(mmMC_ARB_RAMCFG);
 	mc_arb_ramcfg = adev->gfx.config.mc_arb_ramcfg;
 
+	adev->gfx.config.num_banks = REG_GET_FIELD(mc_arb_ramcfg,
+				MC_ARB_RAMCFG, NOOFBANK);
+	adev->gfx.config.num_ranks = REG_GET_FIELD(mc_arb_ramcfg,
+				MC_ARB_RAMCFG, NOOFRANKS);
+
 	adev->gfx.config.num_tile_pipes = adev->gfx.config.max_tile_pipes;
 	adev->gfx.config.mem_max_burst_length_bytes = 256;
 	if (adev->flags & AMD_IS_APU) {
@@ -1856,6 +1849,7 @@ static int gfx_v8_0_compute_ring_init(struct amdgpu_device *adev, int ring_id,
 	int r;
 	unsigned irq_type;
 	struct amdgpu_ring *ring = &adev->gfx.compute_ring[ring_id];
+	unsigned int hw_prio;
 
 	ring = &adev->gfx.compute_ring[ring_id];
 
@@ -1866,7 +1860,7 @@ static int gfx_v8_0_compute_ring_init(struct amdgpu_device *adev, int ring_id,
 
 	ring->ring_obj = NULL;
 	ring->use_doorbell = true;
-	ring->doorbell_index = AMDGPU_DOORBELL_MEC_RING0 + ring_id;
+	ring->doorbell_index = adev->doorbell_index.mec_ring0 + ring_id;
 	ring->eop_gpu_addr = adev->gfx.mec.hpd_eop_gpu_addr
 				+ (ring_id * GFX8_MEC_HPD_SIZE);
 	sprintf(ring->name, "comp_%d.%d.%d", ring->me, ring->pipe, ring->queue);
@@ -1875,9 +1869,11 @@ static int gfx_v8_0_compute_ring_init(struct amdgpu_device *adev, int ring_id,
 		+ ((ring->me - 1) * adev->gfx.mec.num_pipe_per_mec)
 		+ ring->pipe;
 
+	hw_prio = amdgpu_gfx_is_high_priority_compute_queue(adev, ring) ?
+			AMDGPU_RING_PRIO_2 : AMDGPU_RING_PRIO_DEFAULT;
 	/* type-2 packets are deprecated on MEC, use type-3 instead */
-	r = amdgpu_ring_init(adev, ring, 1024,
-			&adev->gfx.eop_irq, irq_type);
+	r = amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq, irq_type,
+			     hw_prio, NULL);
 	if (r)
 		return r;
 
@@ -1885,20 +1881,23 @@ static int gfx_v8_0_compute_ring_init(struct amdgpu_device *adev, int ring_id,
 	return 0;
 }
 
-static int gfx_v8_0_sw_init(void *handle)
+static void gfx_v8_0_sq_irq_work_func(struct work_struct *work);
+
+static int gfx_v8_0_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	int i, j, k, r, ring_id;
+	int xcc_id = 0;
 	struct amdgpu_ring *ring;
-	struct amdgpu_kiq *kiq;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	switch (adev->asic_type) {
-	case CHIP_FIJI:
 	case CHIP_TONGA:
+	case CHIP_CARRIZO:
+	case CHIP_FIJI:
+	case CHIP_POLARIS10:
 	case CHIP_POLARIS11:
 	case CHIP_POLARIS12:
-	case CHIP_POLARIS10:
-	case CHIP_CARRIZO:
+	case CHIP_VEGAM:
 		adev->gfx.mec.num_mec = 2;
 		break;
 	case CHIP_TOPAZ:
@@ -1911,31 +1910,40 @@ static int gfx_v8_0_sw_init(void *handle)
 	adev->gfx.mec.num_pipe_per_mec = 4;
 	adev->gfx.mec.num_queue_per_pipe = 8;
 
-	/* KIQ event */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 178, &adev->gfx.kiq.irq);
-	if (r)
-		return r;
-
 	/* EOP Event */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 181, &adev->gfx.eop_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_CP_END_OF_PIPE, &adev->gfx.eop_irq);
 	if (r)
 		return r;
 
 	/* Privileged reg */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 184,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_CP_PRIV_REG_FAULT,
 			      &adev->gfx.priv_reg_irq);
 	if (r)
 		return r;
 
 	/* Privileged inst */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 185,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_CP_PRIV_INSTR_FAULT,
 			      &adev->gfx.priv_inst_irq);
 	if (r)
 		return r;
 
-	adev->gfx.gfx_current_status = AMDGPU_GFX_NORMAL_MODE;
+	/* Add CP EDC/ECC irq  */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_CP_ECC_ERROR,
+			      &adev->gfx.cp_ecc_error_irq);
+	if (r)
+		return r;
 
-	gfx_v8_0_scratch_init(adev);
+	/* SQ interrupts. */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SQ_INTERRUPT_MSG,
+			      &adev->gfx.sq_irq);
+	if (r) {
+		DRM_ERROR("amdgpu_irq_add() for SQ failed: %d\n", r);
+		return r;
+	}
+
+	INIT_WORK(&adev->gfx.sq_work.work, gfx_v8_0_sq_irq_work_func);
+
+	adev->gfx.gfx_current_status = AMDGPU_GFX_NORMAL_MODE;
 
 	r = gfx_v8_0_init_microcode(adev);
 	if (r) {
@@ -1943,7 +1951,7 @@ static int gfx_v8_0_sw_init(void *handle)
 		return r;
 	}
 
-	r = gfx_v8_0_rlc_init(adev);
+	r = adev->gfx.rlc.funcs->init(adev);
 	if (r) {
 		DRM_ERROR("Failed to init rlc BOs!\n");
 		return r;
@@ -1963,11 +1971,12 @@ static int gfx_v8_0_sw_init(void *handle)
 		/* no gfx doorbells on iceland */
 		if (adev->asic_type != CHIP_TOPAZ) {
 			ring->use_doorbell = true;
-			ring->doorbell_index = AMDGPU_DOORBELL_GFX_RING0;
+			ring->doorbell_index = adev->doorbell_index.gfx_ring0;
 		}
 
 		r = amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq,
-				     AMDGPU_CP_IRQ_GFX_EOP);
+				     AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP,
+				     AMDGPU_RING_PRIO_DEFAULT, NULL);
 		if (r)
 			return r;
 	}
@@ -1978,7 +1987,8 @@ static int gfx_v8_0_sw_init(void *handle)
 	for (i = 0; i < adev->gfx.mec.num_mec; ++i) {
 		for (j = 0; j < adev->gfx.mec.num_queue_per_pipe; j++) {
 			for (k = 0; k < adev->gfx.mec.num_pipe_per_mec; k++) {
-				if (!amdgpu_gfx_is_mec_queue_enabled(adev, i, k, j))
+				if (!amdgpu_gfx_is_mec_queue_enabled(adev, 0, i,
+								     k, j))
 					continue;
 
 				r = gfx_v8_0_compute_ring_init(adev,
@@ -1992,38 +2002,18 @@ static int gfx_v8_0_sw_init(void *handle)
 		}
 	}
 
-	r = amdgpu_gfx_kiq_init(adev, GFX8_MEC_HPD_SIZE);
+	r = amdgpu_gfx_kiq_init(adev, GFX8_MEC_HPD_SIZE, 0);
 	if (r) {
 		DRM_ERROR("Failed to init KIQ BOs!\n");
 		return r;
 	}
 
-	kiq = &adev->gfx.kiq;
-	r = amdgpu_gfx_kiq_init_ring(adev, &kiq->ring, &kiq->irq);
+	r = amdgpu_gfx_kiq_init_ring(adev, xcc_id);
 	if (r)
 		return r;
 
 	/* create MQD for all compute queues as well as KIQ for SRIOV case */
-	r = amdgpu_gfx_compute_mqd_sw_init(adev, sizeof(struct vi_mqd_allocation));
-	if (r)
-		return r;
-
-	/* reserve GDS, GWS and OA resource for gfx */
-	r = amdgpu_bo_create_kernel(adev, adev->gds.mem.gfx_partition_size,
-				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GDS,
-				    &adev->gds.gds_gfx_bo, NULL, NULL);
-	if (r)
-		return r;
-
-	r = amdgpu_bo_create_kernel(adev, adev->gds.gws.gfx_partition_size,
-				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GWS,
-				    &adev->gds.gws_gfx_bo, NULL, NULL);
-	if (r)
-		return r;
-
-	r = amdgpu_bo_create_kernel(adev, adev->gds.oa.gfx_partition_size,
-				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_OA,
-				    &adev->gds.oa_gfx_bo, NULL, NULL);
+	r = amdgpu_gfx_mqd_sw_init(adev, sizeof(struct vi_mqd_allocation), 0);
 	if (r)
 		return r;
 
@@ -2036,26 +2026,31 @@ static int gfx_v8_0_sw_init(void *handle)
 	return 0;
 }
 
-static int gfx_v8_0_sw_fini(void *handle)
+static int gfx_v8_0_sw_fini(struct amdgpu_ip_block *ip_block)
 {
+	struct amdgpu_device *adev = ip_block->adev;
 	int i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	amdgpu_bo_free_kernel(&adev->gds.oa_gfx_bo, NULL, NULL);
-	amdgpu_bo_free_kernel(&adev->gds.gws_gfx_bo, NULL, NULL);
-	amdgpu_bo_free_kernel(&adev->gds.gds_gfx_bo, NULL, NULL);
 
 	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
 		amdgpu_ring_fini(&adev->gfx.gfx_ring[i]);
 	for (i = 0; i < adev->gfx.num_compute_rings; i++)
 		amdgpu_ring_fini(&adev->gfx.compute_ring[i]);
 
-	amdgpu_gfx_compute_mqd_sw_fini(adev);
-	amdgpu_gfx_kiq_free_ring(&adev->gfx.kiq.ring, &adev->gfx.kiq.irq);
-	amdgpu_gfx_kiq_fini(adev);
+	amdgpu_gfx_mqd_sw_fini(adev, 0);
+	amdgpu_gfx_kiq_free_ring(&adev->gfx.kiq[0].ring);
+	amdgpu_gfx_kiq_fini(adev, 0);
 
 	gfx_v8_0_mec_fini(adev);
-	gfx_v8_0_rlc_fini(adev);
+	amdgpu_gfx_rlc_fini(adev);
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj,
+				&adev->gfx.rlc.clear_state_gpu_addr,
+				(void **)&adev->gfx.rlc.cs_ptr);
+	if ((adev->asic_type == CHIP_CARRIZO) ||
+	    (adev->asic_type == CHIP_STONEY)) {
+		amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj,
+				&adev->gfx.rlc.cp_table_gpu_addr,
+				(void **)&adev->gfx.rlc.cp_table_ptr);
+	}
 	gfx_v8_0_free_microcode(adev);
 
 	return 0;
@@ -2250,6 +2245,7 @@ static void gfx_v8_0_tiling_mode_table_init(struct amdgpu_device *adev)
 
 		break;
 	case CHIP_FIJI:
+	case CHIP_VEGAM:
 		modearray[0] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
 				PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
 				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
@@ -3207,6 +3203,7 @@ static void gfx_v8_0_tiling_mode_table_init(struct amdgpu_device *adev)
 		dev_warn(adev->dev,
 			 "Unknown chip type (%d) in function gfx_v8_0_tiling_mode_table_init() falling through to CHIP_CARRIZO\n",
 			 adev->asic_type);
+		fallthrough;
 
 	case CHIP_CARRIZO:
 		modearray[0] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
@@ -3383,7 +3380,8 @@ static void gfx_v8_0_tiling_mode_table_init(struct amdgpu_device *adev)
 }
 
 static void gfx_v8_0_select_se_sh(struct amdgpu_device *adev,
-				  u32 se_num, u32 sh_num, u32 instance)
+				  u32 se_num, u32 sh_num, u32 instance,
+				  int xcc_id)
 {
 	u32 data;
 
@@ -3405,6 +3403,12 @@ static void gfx_v8_0_select_se_sh(struct amdgpu_device *adev,
 	WREG32(mmGRBM_GFX_INDEX, data);
 }
 
+static void gfx_v8_0_select_me_pipe_q(struct amdgpu_device *adev,
+				  u32 me, u32 pipe, u32 q, u32 vm, u32 xcc_id)
+{
+	vi_srbm_select(adev, me, pipe, q, vm);
+}
+
 static u32 gfx_v8_0_get_rb_active_bitmap(struct amdgpu_device *adev)
 {
 	u32 data, mask;
@@ -3425,6 +3429,7 @@ gfx_v8_0_raster_config(struct amdgpu_device *adev, u32 *rconf, u32 *rconf1)
 {
 	switch (adev->asic_type) {
 	case CHIP_FIJI:
+	case CHIP_VEGAM:
 		*rconf |= RB_MAP_PKR0(2) | RB_MAP_PKR1(2) |
 			  RB_XSEL2(1) | PKR_MAP(2) |
 			  PKR_XSEL(1) | PKR_YSEL(1) |
@@ -3560,13 +3565,13 @@ gfx_v8_0_write_harvested_raster_configs(struct amdgpu_device *adev,
 		}
 
 		/* GRBM_GFX_INDEX has a different offset on VI */
-		gfx_v8_0_select_se_sh(adev, se, 0xffffffff, 0xffffffff);
+		gfx_v8_0_select_se_sh(adev, se, 0xffffffff, 0xffffffff, 0);
 		WREG32(mmPA_SC_RASTER_CONFIG, raster_config_se);
 		WREG32(mmPA_SC_RASTER_CONFIG_1, raster_config_1);
 	}
 
 	/* GRBM_GFX_INDEX has a different offset on VI */
-	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 }
 
 static void gfx_v8_0_setup_rb(struct amdgpu_device *adev)
@@ -3582,13 +3587,13 @@ static void gfx_v8_0_setup_rb(struct amdgpu_device *adev)
 	mutex_lock(&adev->grbm_idx_mutex);
 	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
 		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
-			gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff);
+			gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff, 0);
 			data = gfx_v8_0_get_rb_active_bitmap(adev);
 			active_rbs |= data << ((i * adev->gfx.config.max_sh_per_se + j) *
 					       rb_bitmap_width_per_sh);
 		}
 	}
-	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 
 	adev->gfx.config.backend_enable_mask = active_rbs;
 	adev->gfx.config.num_rbs = hweight32(active_rbs);
@@ -3611,7 +3616,7 @@ static void gfx_v8_0_setup_rb(struct amdgpu_device *adev)
 	/* cache the values for userspace */
 	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
 		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
-			gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff);
+			gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff, 0);
 			adev->gfx.config.rb_config[i][j].rb_backend_disable =
 				RREG32(mmCC_RB_BACKEND_DISABLE);
 			adev->gfx.config.rb_config[i][j].user_rb_backend_disable =
@@ -3622,10 +3627,11 @@ static void gfx_v8_0_setup_rb(struct amdgpu_device *adev)
 				RREG32(mmPA_SC_RASTER_CONFIG_1);
 		}
 	}
-	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 	mutex_unlock(&adev->grbm_idx_mutex);
 }
 
+#define DEFAULT_SH_MEM_BASES	(0x6000)
 /**
  * gfx_v8_0_init_compute_vmid - gart enable
  *
@@ -3634,9 +3640,6 @@ static void gfx_v8_0_setup_rb(struct amdgpu_device *adev)
  * Initialize compute vmid sh_mem registers
  *
  */
-#define DEFAULT_SH_MEM_BASES	(0x6000)
-#define FIRST_COMPUTE_VMID	(8)
-#define LAST_COMPUTE_VMID	(16)
 static void gfx_v8_0_init_compute_vmid(struct amdgpu_device *adev)
 {
 	int i;
@@ -3659,7 +3662,7 @@ static void gfx_v8_0_init_compute_vmid(struct amdgpu_device *adev)
 			SH_MEM_CONFIG__PRIVATE_ATC_MASK;
 
 	mutex_lock(&adev->srbm_mutex);
-	for (i = FIRST_COMPUTE_VMID; i < LAST_COMPUTE_VMID; i++) {
+	for (i = adev->vm_manager.first_kfd_vmid; i < AMDGPU_NUM_VMID; i++) {
 		vi_srbm_select(adev, 0, 0, 0, i);
 		/* CP and shaders */
 		WREG32(mmSH_MEM_CONFIG, sh_mem_config);
@@ -3669,6 +3672,33 @@ static void gfx_v8_0_init_compute_vmid(struct amdgpu_device *adev)
 	}
 	vi_srbm_select(adev, 0, 0, 0, 0);
 	mutex_unlock(&adev->srbm_mutex);
+
+	/* Initialize all compute VMIDs to have no GDS, GWS, or OA
+	   access. These should be enabled by FW for target VMIDs. */
+	for (i = adev->vm_manager.first_kfd_vmid; i < AMDGPU_NUM_VMID; i++) {
+		WREG32(amdgpu_gds_reg_offset[i].mem_base, 0);
+		WREG32(amdgpu_gds_reg_offset[i].mem_size, 0);
+		WREG32(amdgpu_gds_reg_offset[i].gws, 0);
+		WREG32(amdgpu_gds_reg_offset[i].oa, 0);
+	}
+}
+
+static void gfx_v8_0_init_gds_vmid(struct amdgpu_device *adev)
+{
+	int vmid;
+
+	/*
+	 * Initialize all compute and user-gfx VMIDs to have no GDS, GWS, or OA
+	 * access. Compute VMIDs should be enabled by FW for target VMIDs,
+	 * the driver can enable them for graphics. VMID0 should maintain
+	 * access so that HWS firmware can save/restore entries.
+	 */
+	for (vmid = 1; vmid < AMDGPU_NUM_VMID; vmid++) {
+		WREG32(amdgpu_gds_reg_offset[vmid].mem_base, 0);
+		WREG32(amdgpu_gds_reg_offset[vmid].mem_size, 0);
+		WREG32(amdgpu_gds_reg_offset[vmid].gws, 0);
+		WREG32(amdgpu_gds_reg_offset[vmid].oa, 0);
+	}
 }
 
 static void gfx_v8_0_config_init(struct amdgpu_device *adev)
@@ -3684,7 +3714,7 @@ static void gfx_v8_0_config_init(struct amdgpu_device *adev)
 	}
 }
 
-static void gfx_v8_0_gpu_init(struct amdgpu_device *adev)
+static void gfx_v8_0_constants_init(struct amdgpu_device *adev)
 {
 	u32 tmp, sh_static_mem_cfg;
 	int i;
@@ -3726,7 +3756,7 @@ static void gfx_v8_0_gpu_init(struct amdgpu_device *adev)
 			tmp = REG_SET_FIELD(tmp, SH_MEM_CONFIG, ALIGNMENT_MODE,
 					    SH_MEM_ALIGNMENT_MODE_UNALIGNED);
 			WREG32(mmSH_MEM_CONFIG, tmp);
-			tmp = adev->mc.shared_aperture_start >> 48;
+			tmp = adev->gmc.shared_aperture_start >> 48;
 			WREG32(mmSH_MEM_BASES, tmp);
 		}
 
@@ -3737,13 +3767,14 @@ static void gfx_v8_0_gpu_init(struct amdgpu_device *adev)
 	mutex_unlock(&adev->srbm_mutex);
 
 	gfx_v8_0_init_compute_vmid(adev);
+	gfx_v8_0_init_gds_vmid(adev);
 
 	mutex_lock(&adev->grbm_idx_mutex);
 	/*
 	 * making sure that the following register writes will be broadcasted
 	 * to all the shaders
 	 */
-	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 
 	WREG32(mmPA_SC_FIFO_SIZE,
 		   (adev->gfx.config.sc_prim_fifo_size_frontend <<
@@ -3774,15 +3805,23 @@ static void gfx_v8_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
 	mutex_lock(&adev->grbm_idx_mutex);
 	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
 		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
-			gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff);
+			gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff, 0);
 			for (k = 0; k < adev->usec_timeout; k++) {
 				if (RREG32(mmRLC_SERDES_CU_MASTER_BUSY) == 0)
 					break;
 				udelay(1);
 			}
+			if (k == adev->usec_timeout) {
+				gfx_v8_0_select_se_sh(adev, 0xffffffff,
+						      0xffffffff, 0xffffffff, 0);
+				mutex_unlock(&adev->grbm_idx_mutex);
+				DRM_INFO("Timeout wait for RLC serdes %u,%u\n",
+					 i, j);
+				return;
+			}
 		}
 	}
-	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 	mutex_unlock(&adev->grbm_idx_mutex);
 
 	mask = RLC_SERDES_NONCU_MASTER_BUSY__SE_MASTER_BUSY_MASK |
@@ -3811,6 +3850,7 @@ static void gfx_v8_0_enable_gui_idle_interrupt(struct amdgpu_device *adev,
 
 static void gfx_v8_0_init_csb(struct amdgpu_device *adev)
 {
+	adev->gfx.rlc.funcs->get_csb_buffer(adev, adev->gfx.rlc.cs_ptr);
 	/* csib */
 	WREG32(mmRLC_CSIB_ADDR_HI,
 			adev->gfx.rlc.clear_state_gpu_addr >> 32);
@@ -3880,21 +3920,20 @@ static int gfx_v8_0_init_save_restore_list(struct amdgpu_device *adev)
 
 	int list_size;
 	unsigned int *register_list_format =
-		kmalloc(adev->gfx.rlc.reg_list_format_size_bytes, GFP_KERNEL);
+		kmemdup(adev->gfx.rlc.register_list_format,
+			adev->gfx.rlc.reg_list_format_size_bytes, GFP_KERNEL);
 	if (!register_list_format)
 		return -ENOMEM;
-	memcpy(register_list_format, adev->gfx.rlc.register_list_format,
-			adev->gfx.rlc.reg_list_format_size_bytes);
 
 	gfx_v8_0_parse_ind_reg_list(register_list_format,
 				RLC_FormatDirectRegListLength,
 				adev->gfx.rlc.reg_list_format_size_bytes >> 2,
 				unique_indices,
 				&indices_count,
-				sizeof(unique_indices) / sizeof(int),
+				ARRAY_SIZE(unique_indices),
 				indirect_start_offsets,
 				&offset_count,
-				sizeof(indirect_start_offsets)/sizeof(int));
+				ARRAY_SIZE(indirect_start_offsets));
 
 	/* save and restore list */
 	WREG32_FIELD(RLC_SRM_CNTL, AUTO_INCR_ADDR, 1);
@@ -3916,14 +3955,14 @@ static int gfx_v8_0_init_save_restore_list(struct amdgpu_device *adev)
 	/* starting offsets starts */
 	WREG32(mmRLC_GPM_SCRATCH_ADDR,
 		adev->gfx.rlc.starting_offsets_start);
-	for (i = 0; i < sizeof(indirect_start_offsets)/sizeof(int); i++)
+	for (i = 0; i < ARRAY_SIZE(indirect_start_offsets); i++)
 		WREG32(mmRLC_GPM_SCRATCH_DATA,
 				indirect_start_offsets[i]);
 
 	/* unique indices */
 	temp = mmRLC_SRM_INDEX_CNTL_ADDR_0;
 	data = mmRLC_SRM_INDEX_CNTL_DATA_0;
-	for (i = 0; i < sizeof(unique_indices) / sizeof(int); i++) {
+	for (i = 0; i < ARRAY_SIZE(unique_indices); i++) {
 		if (unique_indices[i] != 0) {
 			WREG32(temp + i, unique_indices[i] & 0x3FFFF);
 			WREG32(data + i, unique_indices[i] >> 20);
@@ -3984,7 +4023,8 @@ static void gfx_v8_0_init_pg(struct amdgpu_device *adev)
 		gfx_v8_0_init_power_gating(adev);
 		WREG32(mmRLC_PG_ALWAYS_ON_CU_MASK, adev->gfx.cu_info.ao_cu_mask);
 	} else if ((adev->asic_type == CHIP_POLARIS11) ||
-		   (adev->asic_type == CHIP_POLARIS12)) {
+		   (adev->asic_type == CHIP_POLARIS12) ||
+		   (adev->asic_type == CHIP_VEGAM)) {
 		gfx_v8_0_init_csb(adev);
 		gfx_v8_0_init_save_restore_list(adev);
 		gfx_v8_0_enable_save_restore_machine(adev);
@@ -4021,78 +4061,23 @@ static void gfx_v8_0_rlc_start(struct amdgpu_device *adev)
 	udelay(50);
 }
 
-static int gfx_v8_0_rlc_load_microcode(struct amdgpu_device *adev)
-{
-	const struct rlc_firmware_header_v2_0 *hdr;
-	const __le32 *fw_data;
-	unsigned i, fw_size;
-
-	if (!adev->gfx.rlc_fw)
-		return -EINVAL;
-
-	hdr = (const struct rlc_firmware_header_v2_0 *)adev->gfx.rlc_fw->data;
-	amdgpu_ucode_print_rlc_hdr(&hdr->header);
-
-	fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
-			   le32_to_cpu(hdr->header.ucode_array_offset_bytes));
-	fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
-
-	WREG32(mmRLC_GPM_UCODE_ADDR, 0);
-	for (i = 0; i < fw_size; i++)
-		WREG32(mmRLC_GPM_UCODE_DATA, le32_to_cpup(fw_data++));
-	WREG32(mmRLC_GPM_UCODE_ADDR, adev->gfx.rlc_fw_version);
-
-	return 0;
-}
-
 static int gfx_v8_0_rlc_resume(struct amdgpu_device *adev)
 {
-	int r;
-	u32 tmp;
-
-	gfx_v8_0_rlc_stop(adev);
-
-	/* disable CG */
-	tmp = RREG32(mmRLC_CGCG_CGLS_CTRL);
-	tmp &= ~(RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK |
-		 RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK);
-	WREG32(mmRLC_CGCG_CGLS_CTRL, tmp);
-	if (adev->asic_type == CHIP_POLARIS11 ||
-	    adev->asic_type == CHIP_POLARIS10 ||
-	    adev->asic_type == CHIP_POLARIS12) {
-		tmp = RREG32(mmRLC_CGCG_CGLS_CTRL_3D);
-		tmp &= ~0x3;
-		WREG32(mmRLC_CGCG_CGLS_CTRL_3D, tmp);
+	if (amdgpu_sriov_vf(adev)) {
+		gfx_v8_0_init_csb(adev);
+		return 0;
 	}
 
-	/* disable PG */
-	WREG32(mmRLC_PG_CNTL, 0);
-
-	gfx_v8_0_rlc_reset(adev);
+	adev->gfx.rlc.funcs->stop(adev);
+	adev->gfx.rlc.funcs->reset(adev);
 	gfx_v8_0_init_pg(adev);
-
-	if (!adev->pp_enabled) {
-		if (adev->firmware.load_type != AMDGPU_FW_LOAD_SMU) {
-			/* legacy rlc firmware loading */
-			r = gfx_v8_0_rlc_load_microcode(adev);
-			if (r)
-				return r;
-		} else {
-			r = adev->smu.smumgr_funcs->check_fw_load_finish(adev,
-							AMDGPU_UCODE_ID_RLC_G);
-			if (r)
-				return -EINVAL;
-		}
-	}
-
-	gfx_v8_0_rlc_start(adev);
+	adev->gfx.rlc.funcs->start(adev);
 
 	return 0;
 }
 
 static void gfx_v8_0_cp_gfx_enable(struct amdgpu_device *adev, bool enable)
 {
-	int i;
 	u32 tmp = RREG32(mmCP_ME_CNTL);
 
 	if (enable) {
@@ -4103,70 +4088,11 @@ static void gfx_v8_0_cp_gfx_enable(struct amdgpu_device *adev, bool enable)
 		tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_HALT, 1);
 		tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_HALT, 1);
 		tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, CE_HALT, 1);
-		for (i = 0; i < adev->gfx.num_gfx_rings; i++)
-			adev->gfx.gfx_ring[i].ready = false;
 	}
 	WREG32(mmCP_ME_CNTL, tmp);
 	udelay(50);
 }
 
-static int gfx_v8_0_cp_gfx_load_microcode(struct amdgpu_device *adev)
-{
-	const struct gfx_firmware_header_v1_0 *pfp_hdr;
-	const struct gfx_firmware_header_v1_0 *ce_hdr;
-	const struct gfx_firmware_header_v1_0 *me_hdr;
-	const __le32 *fw_data;
-	unsigned i, fw_size;
-
-	if (!adev->gfx.me_fw || !adev->gfx.pfp_fw || !adev->gfx.ce_fw)
-		return -EINVAL;
-
-	pfp_hdr = (const struct gfx_firmware_header_v1_0 *)
-		adev->gfx.pfp_fw->data;
-	ce_hdr = (const struct gfx_firmware_header_v1_0 *)
-		adev->gfx.ce_fw->data;
-	me_hdr = (const struct gfx_firmware_header_v1_0 *)
-		adev->gfx.me_fw->data;
-
-	amdgpu_ucode_print_gfx_hdr(&pfp_hdr->header);
-	amdgpu_ucode_print_gfx_hdr(&ce_hdr->header);
-	amdgpu_ucode_print_gfx_hdr(&me_hdr->header);
-
-	gfx_v8_0_cp_gfx_enable(adev, false);
-
-	/* PFP */
-	fw_data = (const __le32 *)
-		(adev->gfx.pfp_fw->data +
-		 le32_to_cpu(pfp_hdr->header.ucode_array_offset_bytes));
-	fw_size = le32_to_cpu(pfp_hdr->header.ucode_size_bytes) / 4;
-	WREG32(mmCP_PFP_UCODE_ADDR, 0);
-	for (i = 0; i < fw_size; i++)
-		WREG32(mmCP_PFP_UCODE_DATA, le32_to_cpup(fw_data++));
-	WREG32(mmCP_PFP_UCODE_ADDR, adev->gfx.pfp_fw_version);
-
-	/* CE */
-	fw_data = (const __le32 *)
-		(adev->gfx.ce_fw->data +
-		 le32_to_cpu(ce_hdr->header.ucode_array_offset_bytes));
-	fw_size = le32_to_cpu(ce_hdr->header.ucode_size_bytes) / 4;
-	WREG32(mmCP_CE_UCODE_ADDR, 0);
-	for (i = 0; i < fw_size; i++)
-		WREG32(mmCP_CE_UCODE_DATA, le32_to_cpup(fw_data++));
-	WREG32(mmCP_CE_UCODE_ADDR, adev->gfx.ce_fw_version);
-
-	/* ME */
-	fw_data = (const __le32 *)
-		(adev->gfx.me_fw->data +
-		 le32_to_cpu(me_hdr->header.ucode_array_offset_bytes));
-	fw_size = le32_to_cpu(me_hdr->header.ucode_size_bytes) / 4;
-	WREG32(mmCP_ME_RAM_WADDR, 0);
-	for (i = 0; i < fw_size; i++)
-		WREG32(mmCP_ME_RAM_DATA, le32_to_cpup(fw_data++));
-	WREG32(mmCP_ME_RAM_WADDR, adev->gfx.me_fw_version);
-
-	return 0;
-}
-
 static u32 gfx_v8_0_get_csb_size(struct amdgpu_device *adev)
 {
 	u32 count = 0;
@@ -4240,37 +4166,8 @@ static int gfx_v8_0_cp_gfx_start(struct amdgpu_device *adev)
 
 	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONTEXT_REG, 2));
 	amdgpu_ring_write(ring, mmPA_SC_RASTER_CONFIG - PACKET3_SET_CONTEXT_REG_START);
-	switch (adev->asic_type) {
-	case CHIP_TONGA:
-	case CHIP_POLARIS10:
-		amdgpu_ring_write(ring, 0x16000012);
-		amdgpu_ring_write(ring, 0x0000002A);
-		break;
-	case CHIP_POLARIS11:
-	case CHIP_POLARIS12:
-		amdgpu_ring_write(ring, 0x16000012);
-		amdgpu_ring_write(ring, 0x00000000);
-		break;
-	case CHIP_FIJI:
-		amdgpu_ring_write(ring, 0x3a00161a);
-		amdgpu_ring_write(ring, 0x0000002e);
-		break;
-	case CHIP_CARRIZO:
-		amdgpu_ring_write(ring, 0x00000002);
-		amdgpu_ring_write(ring, 0x00000000);
-		break;
-	case CHIP_TOPAZ:
-		amdgpu_ring_write(ring, adev->gfx.config.num_rbs == 1 ?
-				0x00000000 : 0x00000002);
-		amdgpu_ring_write(ring, 0x00000000);
-		break;
-	case CHIP_STONEY:
-		amdgpu_ring_write(ring, 0x00000000);
-		amdgpu_ring_write(ring, 0x00000000);
-		break;
-	default:
-		BUG();
-	}
+	amdgpu_ring_write(ring, adev->gfx.config.rb_config[0][0].raster_config);
+	amdgpu_ring_write(ring, adev->gfx.config.rb_config[0][0].raster_config_1);
 
 	amdgpu_ring_write(ring, PACKET3(PACKET3_PREAMBLE_CNTL, 0));
 	amdgpu_ring_write(ring, PACKET3_PREAMBLE_END_CLEAR_STATE);
@@ -4315,7 +4212,7 @@ static void gfx_v8_0_set_cpg_door_bell(struct amdgpu_device *adev, struct amdgpu
 
 	tmp = REG_SET_FIELD(0, CP_RB_DOORBELL_RANGE_LOWER,
 					DOORBELL_RANGE_LOWER,
-					AMDGPU_DOORBELL_GFX_RING0);
+					adev->doorbell_index.gfx_ring0);
 	WREG32(mmCP_RB_DOORBELL_RANGE_LOWER, tmp);
 
 	WREG32(mmCP_RB_DOORBELL_RANGE_UPPER,
@@ -4328,7 +4225,6 @@ static int gfx_v8_0_cp_gfx_resume(struct amdgpu_device *adev)
 	u32 tmp;
 	u32 rb_bufsz;
 	u64 rb_addr, rptr_addr, wptr_gpu_addr;
-	int r;
 
 	/* Set the write pointer delay */
 	WREG32(mmCP_RB_WPTR_DELAY, 0);
@@ -4353,12 +4249,12 @@ static int gfx_v8_0_cp_gfx_resume(struct amdgpu_device *adev)
 	ring->wptr = 0;
 	WREG32(mmCP_RB0_WPTR, lower_32_bits(ring->wptr));
 
-	/* set the wb address wether it's enabled or not */
-	rptr_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+	/* set the wb address whether it's enabled or not */
+	rptr_addr = ring->rptr_gpu_addr;
 	WREG32(mmCP_RB0_RPTR_ADDR, lower_32_bits(rptr_addr));
 	WREG32(mmCP_RB0_RPTR_ADDR_HI, upper_32_bits(rptr_addr) & 0xFF);
 
-	wptr_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
+	wptr_gpu_addr = ring->wptr_gpu_addr;
 	WREG32(mmCP_RB_WPTR_POLL_ADDR_LO, lower_32_bits(wptr_gpu_addr));
 	WREG32(mmCP_RB_WPTR_POLL_ADDR_HI, upper_32_bits(wptr_gpu_addr));
 	mdelay(1);
@@ -4372,75 +4268,21 @@ static int gfx_v8_0_cp_gfx_resume(struct amdgpu_device *adev)
 	/* start the ring */
 	amdgpu_ring_clear_ring(ring);
 	gfx_v8_0_cp_gfx_start(adev);
-	ring->ready = true;
-	r = amdgpu_ring_test_ring(ring);
-	if (r)
-		ring->ready = false;
 
-	return r;
+	return 0;
 }
 
 static void gfx_v8_0_cp_compute_enable(struct amdgpu_device *adev, bool enable)
 {
-	int i;
-
 	if (enable) {
 		WREG32(mmCP_MEC_CNTL, 0);
 	} else {
 		WREG32(mmCP_MEC_CNTL, (CP_MEC_CNTL__MEC_ME1_HALT_MASK | CP_MEC_CNTL__MEC_ME2_HALT_MASK));
-		for (i = 0; i < adev->gfx.num_compute_rings; i++)
-			adev->gfx.compute_ring[i].ready = false;
-		adev->gfx.kiq.ring.ready = false;
+		adev->gfx.kiq[0].ring.sched.ready = false;
 	}
 	udelay(50);
 }
 
-static int gfx_v8_0_cp_compute_load_microcode(struct amdgpu_device *adev)
-{
-	const struct gfx_firmware_header_v1_0 *mec_hdr;
-	const __le32 *fw_data;
-	unsigned i, fw_size;
-
-	if (!adev->gfx.mec_fw)
-		return -EINVAL;
-
-	gfx_v8_0_cp_compute_enable(adev, false);
-
-	mec_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
-	amdgpu_ucode_print_gfx_hdr(&mec_hdr->header);
-
-	fw_data = (const __le32 *)
-		(adev->gfx.mec_fw->data +
-		 le32_to_cpu(mec_hdr->header.ucode_array_offset_bytes));
-	fw_size = le32_to_cpu(mec_hdr->header.ucode_size_bytes) / 4;
-
-	/* MEC1 */
-	WREG32(mmCP_MEC_ME1_UCODE_ADDR, 0);
-	for (i = 0; i < fw_size; i++)
-		WREG32(mmCP_MEC_ME1_UCODE_DATA, le32_to_cpup(fw_data+i));
-	WREG32(mmCP_MEC_ME1_UCODE_ADDR, adev->gfx.mec_fw_version);
-
-	/* Loading MEC2 firmware is only necessary if MEC2 should run different microcode than MEC1. */
-	if (adev->gfx.mec2_fw) {
-		const struct gfx_firmware_header_v1_0 *mec2_hdr;
-
-		mec2_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec2_fw->data;
-		amdgpu_ucode_print_gfx_hdr(&mec2_hdr->header);
-
-		fw_data = (const __le32 *)
-			(adev->gfx.mec2_fw->data +
-			 le32_to_cpu(mec2_hdr->header.ucode_array_offset_bytes));
-		fw_size = le32_to_cpu(mec2_hdr->header.ucode_size_bytes) / 4;
-
-		WREG32(mmCP_MEC_ME2_UCODE_ADDR, 0);
-		for (i = 0; i < fw_size; i++)
-			WREG32(mmCP_MEC_ME2_UCODE_DATA, le32_to_cpup(fw_data+i));
-		WREG32(mmCP_MEC_ME2_UCODE_ADDR, adev->gfx.mec2_fw_version);
-	}
-
-	return 0;
-}
-
 /* KIQ functions */
 static void gfx_v8_0_kiq_setting(struct amdgpu_ring *ring)
 {
@@ -4451,20 +4293,17 @@ static void gfx_v8_0_kiq_setting(struct amdgpu_ring *ring)
 	tmp = RREG32(mmRLC_CP_SCHEDULERS);
 	tmp &= 0xffffff00;
 	tmp |= (ring->me << 5) | (ring->pipe << 3) | (ring->queue);
-	WREG32(mmRLC_CP_SCHEDULERS, tmp);
-	tmp |= 0x80;
-	WREG32(mmRLC_CP_SCHEDULERS, tmp);
+	WREG32(mmRLC_CP_SCHEDULERS, tmp | 0x80);
 }
 
 static int gfx_v8_0_kiq_kcq_enable(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *kiq_ring = &adev->gfx.kiq.ring;
-	uint32_t scratch, tmp = 0;
+	struct amdgpu_ring *kiq_ring = &adev->gfx.kiq[0].ring;
 	uint64_t queue_mask = 0;
 	int r, i;
 
 	for (i = 0; i < AMDGPU_MAX_COMPUTE_QUEUES; ++i) {
-		if (!test_bit(i, adev->gfx.mec.queue_bitmap))
+		if (!test_bit(i, adev->gfx.mec_bitmap[0].queue_bitmap))
 			continue;
 
 		/* This situation may be hit in the future if a new HW
@@ -4478,17 +4317,9 @@ static int gfx_v8_0_kiq_kcq_enable(struct amdgpu_device *adev)
 		queue_mask |= (1ull << i);
 	}
 
-	r = amdgpu_gfx_scratch_get(adev, &scratch);
-	if (r) {
-		DRM_ERROR("Failed to get scratch reg (%d).\n", r);
-		return r;
-	}
-	WREG32(scratch, 0xCAFEDEAD);
-
-	r = amdgpu_ring_alloc(kiq_ring, (8 * adev->gfx.num_compute_rings) + 11);
+	r = amdgpu_ring_alloc(kiq_ring, (8 * adev->gfx.num_compute_rings) + 8);
 	if (r) {
 		DRM_ERROR("Failed to lock KIQ (%d).\n", r);
-		amdgpu_gfx_scratch_free(adev, scratch);
 		return r;
 	}
 	/* set resources */
@@ -4503,7 +4334,7 @@ static int gfx_v8_0_kiq_kcq_enable(struct amdgpu_device *adev)
 	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
 		struct amdgpu_ring *ring = &adev->gfx.compute_ring[i];
 		uint64_t mqd_addr = amdgpu_bo_gpu_offset(ring->mqd_obj);
-		uint64_t wptr_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
+		uint64_t wptr_addr = ring->wptr_gpu_addr;
 
 		/* map queues */
 		amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_MAP_QUEUES, 5));
@@ -4520,26 +4351,10 @@ static int gfx_v8_0_kiq_kcq_enable(struct amdgpu_device *adev)
 		amdgpu_ring_write(kiq_ring, lower_32_bits(wptr_addr));
 		amdgpu_ring_write(kiq_ring, upper_32_bits(wptr_addr));
 	}
-	/* write to scratch for completion */
-	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_SET_UCONFIG_REG, 1));
-	amdgpu_ring_write(kiq_ring, (scratch - PACKET3_SET_UCONFIG_REG_START));
-	amdgpu_ring_write(kiq_ring, 0xDEADBEEF);
-	amdgpu_ring_commit(kiq_ring);
 
-	for (i = 0; i < adev->usec_timeout; i++) {
-		tmp = RREG32(scratch);
-		if (tmp == 0xDEADBEEF)
-			break;
-		DRM_UDELAY(1);
-	}
-	if (i >= adev->usec_timeout) {
-		DRM_ERROR("KCQ enable failed (scratch(0x%04X)=0x%08X)\n",
-			  scratch, tmp);
-		r = -EINVAL;
-	}
-	amdgpu_gfx_scratch_free(adev, scratch);
+	amdgpu_ring_commit(kiq_ring);
 
-	return r;
+	return 0;
 }
 
 static int gfx_v8_0_deactivate_hqd(struct amdgpu_device *adev, u32 req)
@@ -4563,6 +4378,19 @@ static int gfx_v8_0_deactivate_hqd(struct amdgpu_device *adev, u32 req)
 	return r;
 }
 
+static void gfx_v8_0_mqd_set_priority(struct amdgpu_ring *ring, struct vi_mqd *mqd)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE) {
+		if (amdgpu_gfx_is_high_priority_compute_queue(adev, ring)) {
+			mqd->cp_hqd_pipe_priority = AMDGPU_GFX_PIPE_PRIO_HIGH;
+			mqd->cp_hqd_queue_priority =
+				AMDGPU_GFX_QUEUE_PRIORITY_MAXIMUM;
+		}
+	}
+}
+
 static int gfx_v8_0_mqd_init(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
@@ -4577,12 +4405,10 @@ static int gfx_v8_0_mqd_init(struct amdgpu_ring *ring)
 	mqd->compute_static_thread_mgmt_se2 = 0xffffffff;
 	mqd->compute_static_thread_mgmt_se3 = 0xffffffff;
 	mqd->compute_misc_reserved = 0x00000003;
-	if (!(adev->flags & AMD_IS_APU)) {
-		mqd->dynamic_cu_mask_addr_lo = lower_32_bits(ring->mqd_gpu_addr
-					     + offsetof(struct vi_mqd_allocation, dynamic_cu_mask));
-		mqd->dynamic_cu_mask_addr_hi = upper_32_bits(ring->mqd_gpu_addr
-					     + offsetof(struct vi_mqd_allocation, dynamic_cu_mask));
-	}
+	mqd->dynamic_cu_mask_addr_lo = lower_32_bits(ring->mqd_gpu_addr
+						     + offsetof(struct vi_mqd_allocation, dynamic_cu_mask));
+	mqd->dynamic_cu_mask_addr_hi = upper_32_bits(ring->mqd_gpu_addr
+						     + offsetof(struct vi_mqd_allocation, dynamic_cu_mask));
 	eop_base_addr = ring->eop_gpu_addr >> 8;
 	mqd->cp_hqd_eop_base_addr_lo = eop_base_addr;
 	mqd->cp_hqd_eop_base_addr_hi = upper_32_bits(eop_base_addr);
@@ -4621,7 +4447,7 @@ static int gfx_v8_0_mqd_init(struct amdgpu_ring *ring)
 	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, QUEUE_SIZE,
 			    (order_base_2(ring->ring_size / 4) - 1));
 	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, RPTR_BLOCK_SIZE,
-			((order_base_2(AMDGPU_GPU_PAGE_SIZE / 4) - 1) << 8));
+			(order_base_2(AMDGPU_GPU_PAGE_SIZE / 4) - 1));
 #ifdef __BIG_ENDIAN
 	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, ENDIAN_SWAP, 1);
 #endif
@@ -4632,13 +4458,13 @@ static int gfx_v8_0_mqd_init(struct amdgpu_ring *ring)
 	mqd->cp_hqd_pq_control = tmp;
 
 	/* set the wb address whether it's enabled or not */
-	wb_gpu_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+	wb_gpu_addr = ring->rptr_gpu_addr;
 	mqd->cp_hqd_pq_rptr_report_addr_lo = wb_gpu_addr & 0xfffffffc;
 	mqd->cp_hqd_pq_rptr_report_addr_hi =
 		upper_32_bits(wb_gpu_addr) & 0xffff;
 
 	/* only used if CP_PQ_WPTR_POLL_CNTL.CP_PQ_WPTR_POLL_CNTL__EN_MASK=1 */
-	wb_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
+	wb_gpu_addr = ring->wptr_gpu_addr;
 	mqd->cp_hqd_pq_wptr_poll_addr_lo = wb_gpu_addr & 0xfffffffc;
 	mqd->cp_hqd_pq_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr) & 0xffff;
 
@@ -4688,9 +4514,6 @@ static int gfx_v8_0_mqd_init(struct amdgpu_ring *ring)
 	/* defaults */
 	mqd->cp_hqd_eop_rptr = RREG32(mmCP_HQD_EOP_RPTR);
 	mqd->cp_hqd_eop_wptr = RREG32(mmCP_HQD_EOP_WPTR);
-	mqd->cp_hqd_pipe_priority = RREG32(mmCP_HQD_PIPE_PRIORITY);
-	mqd->cp_hqd_queue_priority = RREG32(mmCP_HQD_QUEUE_PRIORITY);
-	mqd->cp_hqd_quantum = RREG32(mmCP_HQD_QUANTUM);
 	mqd->cp_hqd_ctx_save_base_addr_lo = RREG32(mmCP_HQD_CTX_SAVE_BASE_ADDR_LO);
 	mqd->cp_hqd_ctx_save_base_addr_hi = RREG32(mmCP_HQD_CTX_SAVE_BASE_ADDR_HI);
 	mqd->cp_hqd_cntl_stack_offset = RREG32(mmCP_HQD_CNTL_STACK_OFFSET);
@@ -4702,13 +4525,20 @@ static int gfx_v8_0_mqd_init(struct amdgpu_ring *ring)
 	mqd->cp_hqd_eop_wptr_mem = RREG32(mmCP_HQD_EOP_WPTR_MEM);
 	mqd->cp_hqd_eop_dones = RREG32(mmCP_HQD_EOP_DONES);
 
-	/* activate the queue */
-	mqd->cp_hqd_active = 1;
+	/* set static priority for a queue/ring */
+	gfx_v8_0_mqd_set_priority(ring, mqd);
+	mqd->cp_hqd_quantum = RREG32(mmCP_HQD_QUANTUM);
+
+	/* map_queues packet doesn't need activate the queue,
+	 * so only kiq need set this field.
+	 */
+	if (ring->funcs->type == AMDGPU_RING_TYPE_KIQ)
+		mqd->cp_hqd_active = 1;
 
 	return 0;
 }
 
-int gfx_v8_0_mqd_commit(struct amdgpu_device *adev,
+static int gfx_v8_0_mqd_commit(struct amdgpu_device *adev,
 			struct vi_mqd *mqd)
 {
 	uint32_t mqd_reg;
@@ -4749,14 +4579,13 @@ static int gfx_v8_0_kiq_init_queue(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
 	struct vi_mqd *mqd = ring->mqd_ptr;
-	int mqd_idx = AMDGPU_MAX_COMPUTE_RINGS;
 
 	gfx_v8_0_kiq_setting(ring);
 
-	if (adev->gfx.in_reset) { /* for GPU_RESET case */
+	if (amdgpu_in_reset(adev)) { /* for GPU_RESET case */
 		/* reset MQD to a clean status */
-		if (adev->gfx.mec.mqd_backup[mqd_idx])
-			memcpy(mqd, adev->gfx.mec.mqd_backup[mqd_idx], sizeof(struct vi_mqd_allocation));
+		if (adev->gfx.kiq[0].mqd_backup)
+			memcpy(mqd, adev->gfx.kiq[0].mqd_backup, sizeof(struct vi_mqd_allocation));
 
 		/* reset ring buffer */
 		ring->wptr = 0;
@@ -4770,6 +4599,8 @@ static int gfx_v8_0_kiq_init_queue(struct amdgpu_ring *ring)
 		memset((void *)mqd, 0, sizeof(struct vi_mqd_allocation));
 		((struct vi_mqd_allocation *)mqd)->dynamic_cu_mask = 0xFFFFFFFF;
 		((struct vi_mqd_allocation *)mqd)->dynamic_rb_mask = 0xFFFFFFFF;
+		if (amdgpu_sriov_vf(adev) && adev->in_suspend)
+			amdgpu_ring_clear_ring(ring);
 		mutex_lock(&adev->srbm_mutex);
 		vi_srbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
 		gfx_v8_0_mqd_init(ring);
@@ -4777,8 +4608,8 @@ static int gfx_v8_0_kiq_init_queue(struct amdgpu_ring *ring)
 		vi_srbm_select(adev, 0, 0, 0, 0);
 		mutex_unlock(&adev->srbm_mutex);
 
-		if (adev->gfx.mec.mqd_backup[mqd_idx])
-			memcpy(adev->gfx.mec.mqd_backup[mqd_idx], mqd, sizeof(struct vi_mqd_allocation));
+		if (adev->gfx.kiq[0].mqd_backup)
+			memcpy(adev->gfx.kiq[0].mqd_backup, mqd, sizeof(struct vi_mqd_allocation));
 	}
 
 	return 0;
@@ -4790,7 +4621,7 @@ static int gfx_v8_0_kcq_init_queue(struct amdgpu_ring *ring)
 	struct vi_mqd *mqd = ring->mqd_ptr;
 	int mqd_idx = ring - &adev->gfx.compute_ring[0];
 
-	if (!adev->gfx.in_reset && !adev->gfx.in_suspend) {
+	if (!amdgpu_in_reset(adev) && !adev->in_suspend) {
 		memset((void *)mqd, 0, sizeof(struct vi_mqd_allocation));
 		((struct vi_mqd_allocation *)mqd)->dynamic_cu_mask = 0xFFFFFFFF;
 		((struct vi_mqd_allocation *)mqd)->dynamic_rb_mask = 0xFFFFFFFF;
@@ -4802,14 +4633,13 @@ static int gfx_v8_0_kcq_init_queue(struct amdgpu_ring *ring)
 
 		if (adev->gfx.mec.mqd_backup[mqd_idx])
 			memcpy(adev->gfx.mec.mqd_backup[mqd_idx], mqd, sizeof(struct vi_mqd_allocation));
-	} else if (adev->gfx.in_reset) { /* for GPU_RESET case */
-		/* reset MQD to a clean status */
+	} else {
+		/* restore MQD to a clean status */
 		if (adev->gfx.mec.mqd_backup[mqd_idx])
 			memcpy(mqd, adev->gfx.mec.mqd_backup[mqd_idx], sizeof(struct vi_mqd_allocation));
 		/* reset ring buffer */
 		ring->wptr = 0;
-		amdgpu_ring_clear_ring(ring);
-	} else {
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr, 0);
 		amdgpu_ring_clear_ring(ring);
 	}
 	return 0;
@@ -4818,8 +4648,8 @@ static int gfx_v8_0_kcq_init_queue(struct amdgpu_ring *ring)
 static void gfx_v8_0_set_mec_doorbell_range(struct amdgpu_device *adev)
 {
 	if (adev->asic_type > CHIP_TONGA) {
-		WREG32(mmCP_MEC_DOORBELL_RANGE_LOWER, AMDGPU_DOORBELL_KIQ << 2);
-		WREG32(mmCP_MEC_DOORBELL_RANGE_UPPER, AMDGPU_DOORBELL_MEC_RING7 << 2);
+		WREG32(mmCP_MEC_DOORBELL_RANGE_LOWER, adev->doorbell_index.kiq << 2);
+		WREG32(mmCP_MEC_DOORBELL_RANGE_UPPER, adev->doorbell_index.mec_ring7 << 2);
 	}
 	/* enable doorbells */
 	WREG32_FIELD(CP_PQ_STATUS, DOORBELL_ENABLE, 1);
@@ -4827,70 +4657,49 @@ static void gfx_v8_0_set_mec_doorbell_range(struct amdgpu_device *adev)
 
 static int gfx_v8_0_kiq_resume(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *ring = NULL;
-	int r = 0, i;
-
-	gfx_v8_0_cp_compute_enable(adev, true);
-
-	ring = &adev->gfx.kiq.ring;
+	gfx_v8_0_kiq_init_queue(&adev->gfx.kiq[0].ring);
+	return 0;
+}
 
-	r = amdgpu_bo_reserve(ring->mqd_obj, false);
-	if (unlikely(r != 0))
-		goto done;
+static int gfx_v8_0_kcq_resume(struct amdgpu_device *adev)
+{
+	int i, r;
 
-	r = amdgpu_bo_kmap(ring->mqd_obj, &ring->mqd_ptr);
-	if (!r) {
-		r = gfx_v8_0_kiq_init_queue(ring);
-		amdgpu_bo_kunmap(ring->mqd_obj);
-		ring->mqd_ptr = NULL;
-	}
-	amdgpu_bo_unreserve(ring->mqd_obj);
-	if (r)
-		goto done;
+	gfx_v8_0_cp_compute_enable(adev, true);
 
 	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
-		ring = &adev->gfx.compute_ring[i];
-
-		r = amdgpu_bo_reserve(ring->mqd_obj, false);
-		if (unlikely(r != 0))
-			goto done;
-		r = amdgpu_bo_kmap(ring->mqd_obj, &ring->mqd_ptr);
-		if (!r) {
-			r = gfx_v8_0_kcq_init_queue(ring);
-			amdgpu_bo_kunmap(ring->mqd_obj);
-			ring->mqd_ptr = NULL;
-		}
-		amdgpu_bo_unreserve(ring->mqd_obj);
+		r = gfx_v8_0_kcq_init_queue(&adev->gfx.compute_ring[i]);
 		if (r)
-			goto done;
+			return r;
 	}
 
 	gfx_v8_0_set_mec_doorbell_range(adev);
 
-	r = gfx_v8_0_kiq_kcq_enable(adev);
+	return gfx_v8_0_kiq_kcq_enable(adev);
+}
+
+static int gfx_v8_0_cp_test_all_rings(struct amdgpu_device *adev)
+{
+	int r, i;
+	struct amdgpu_ring *ring;
+
+	/* collect all the ring_tests here, gfx, kiq, compute */
+	ring = &adev->gfx.gfx_ring[0];
+	r = amdgpu_ring_test_helper(ring);
 	if (r)
-		goto done;
+		return r;
 
-	/* Test KIQ */
-	ring = &adev->gfx.kiq.ring;
-	ring->ready = true;
-	r = amdgpu_ring_test_ring(ring);
-	if (r) {
-		ring->ready = false;
-		goto done;
-	}
+	ring = &adev->gfx.kiq[0].ring;
+	r = amdgpu_ring_test_helper(ring);
+	if (r)
+		return r;
 
-	/* Test KCQs */
 	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
 		ring = &adev->gfx.compute_ring[i];
-		ring->ready = true;
-		r = amdgpu_ring_test_ring(ring);
-		if (r)
-			ring->ready = false;
+		amdgpu_ring_test_helper(ring);
 	}
 
-done:
-	return r;
+	return 0;
 }
 
 static int gfx_v8_0_cp_resume(struct amdgpu_device *adev)
@@ -4900,50 +4709,19 @@ static int gfx_v8_0_cp_resume(struct amdgpu_device *adev)
 	if (!(adev->flags & AMD_IS_APU))
 		gfx_v8_0_enable_gui_idle_interrupt(adev, false);
 
-	if (!adev->pp_enabled) {
-		if (adev->firmware.load_type != AMDGPU_FW_LOAD_SMU) {
-			/* legacy firmware loading */
-			r = gfx_v8_0_cp_gfx_load_microcode(adev);
-			if (r)
-				return r;
-
-			r = gfx_v8_0_cp_compute_load_microcode(adev);
-			if (r)
-				return r;
-		} else {
-			r = adev->smu.smumgr_funcs->check_fw_load_finish(adev,
-							AMDGPU_UCODE_ID_CP_CE);
-			if (r)
-				return -EINVAL;
-
-			r = adev->smu.smumgr_funcs->check_fw_load_finish(adev,
-							AMDGPU_UCODE_ID_CP_PFP);
-			if (r)
-				return -EINVAL;
-
-			r = adev->smu.smumgr_funcs->check_fw_load_finish(adev,
-							AMDGPU_UCODE_ID_CP_ME);
-			if (r)
-				return -EINVAL;
-
-			if (adev->asic_type == CHIP_TOPAZ) {
-				r = gfx_v8_0_cp_compute_load_microcode(adev);
-				if (r)
-					return r;
-			} else {
-				r = adev->smu.smumgr_funcs->check_fw_load_finish(adev,
-										 AMDGPU_UCODE_ID_CP_MEC1);
-				if (r)
-					return -EINVAL;
-			}
-		}
-	}
+	r = gfx_v8_0_kiq_resume(adev);
+	if (r)
+		return r;
 
 	r = gfx_v8_0_cp_gfx_resume(adev);
 	if (r)
 		return r;
 
-	r = gfx_v8_0_kiq_resume(adev);
+	r = gfx_v8_0_kcq_resume(adev);
+	if (r)
+		return r;
+
+	r = gfx_v8_0_cp_test_all_rings(adev);
 	if (r)
 		return r;
 
@@ -4958,15 +4736,15 @@ static void gfx_v8_0_cp_enable(struct amdgpu_device *adev, bool enable)
 	gfx_v8_0_cp_compute_enable(adev, enable);
 }
 
-static int gfx_v8_0_hw_init(void *handle)
+static int gfx_v8_0_hw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	gfx_v8_0_init_golden_registers(adev);
-	gfx_v8_0_gpu_init(adev);
+	gfx_v8_0_constants_init(adev);
 
-	r = gfx_v8_0_rlc_resume(adev);
+	r = adev->gfx.rlc.funcs->resume(adev);
 	if (r)
 		return r;
 
@@ -4975,59 +4753,71 @@ static int gfx_v8_0_hw_init(void *handle)
 	return r;
 }
 
-static int gfx_v8_0_hw_fini(void *handle)
+static int gfx_v8_0_kcq_disable(struct amdgpu_device *adev)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int r, i;
+	struct amdgpu_ring *kiq_ring = &adev->gfx.kiq[0].ring;
 
-	amdgpu_irq_put(adev, &adev->gfx.priv_reg_irq, 0);
-	amdgpu_irq_put(adev, &adev->gfx.priv_inst_irq, 0);
-	if (amdgpu_sriov_vf(adev)) {
-		pr_debug("For SRIOV client, shouldn't do anything.\n");
-		return 0;
-	}
-	gfx_v8_0_cp_enable(adev, false);
-	gfx_v8_0_rlc_stop(adev);
+	r = amdgpu_ring_alloc(kiq_ring, 6 * adev->gfx.num_compute_rings);
+	if (r)
+		DRM_ERROR("Failed to lock KIQ (%d).\n", r);
 
-	amdgpu_set_powergating_state(adev,
-			AMD_IP_BLOCK_TYPE_GFX, AMD_PG_STATE_UNGATE);
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		struct amdgpu_ring *ring = &adev->gfx.compute_ring[i];
 
-	return 0;
-}
+		amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_UNMAP_QUEUES, 4));
+		amdgpu_ring_write(kiq_ring, /* Q_sel: 0, vmid: 0, engine: 0, num_Q: 1 */
+						PACKET3_UNMAP_QUEUES_ACTION(1) | /* RESET_QUEUES */
+						PACKET3_UNMAP_QUEUES_QUEUE_SEL(0) |
+						PACKET3_UNMAP_QUEUES_ENGINE_SEL(0) |
+						PACKET3_UNMAP_QUEUES_NUM_QUEUES(1));
+		amdgpu_ring_write(kiq_ring, PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET0(ring->doorbell_index));
+		amdgpu_ring_write(kiq_ring, 0);
+		amdgpu_ring_write(kiq_ring, 0);
+		amdgpu_ring_write(kiq_ring, 0);
+	}
+	/* Submit unmap queue packet */
+	amdgpu_ring_commit(kiq_ring);
+	/*
+	 * Ring test will do a basic scratch register change check. Just run
+	 * this to ensure that unmap queues that is submitted before got
+	 * processed successfully before returning.
+	 */
+	r = amdgpu_ring_test_helper(kiq_ring);
+	if (r)
+		DRM_ERROR("KCQ disable failed\n");
 
-static int gfx_v8_0_suspend(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	adev->gfx.in_suspend = true;
-	return gfx_v8_0_hw_fini(adev);
+	return r;
 }
 
-static int gfx_v8_0_resume(void *handle)
+static bool gfx_v8_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	r = gfx_v8_0_hw_init(adev);
-	adev->gfx.in_suspend = false;
-	return r;
+	if (REG_GET_FIELD(RREG32(mmGRBM_STATUS), GRBM_STATUS, GUI_ACTIVE)
+		|| RREG32(mmGRBM_STATUS2) != 0x8)
+		return false;
+	else
+		return true;
 }
 
-static bool gfx_v8_0_is_idle(void *handle)
+static bool gfx_v8_0_rlc_is_idle(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	if (REG_GET_FIELD(RREG32(mmGRBM_STATUS), GRBM_STATUS, GUI_ACTIVE))
+	if (RREG32(mmGRBM_STATUS2) != 0x8)
 		return false;
 	else
 		return true;
 }
 
-static int gfx_v8_0_wait_for_idle(void *handle)
+static int gfx_v8_0_wait_for_rlc_idle(void *handle)
 {
-	unsigned i;
+	unsigned int i;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		if (gfx_v8_0_is_idle(handle))
+		if (gfx_v8_0_rlc_is_idle(handle))
 			return 0;
 
 		udelay(1);
@@ -5035,9 +4825,66 @@ static int gfx_v8_0_wait_for_idle(void *handle)
 	return -ETIMEDOUT;
 }
 
-static bool gfx_v8_0_check_soft_reset(void *handle)
+static int gfx_v8_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	unsigned int i;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (gfx_v8_0_is_idle(ip_block))
+			return 0;
+
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static int gfx_v8_0_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	amdgpu_irq_put(adev, &adev->gfx.priv_reg_irq, 0);
+	amdgpu_irq_put(adev, &adev->gfx.priv_inst_irq, 0);
+
+	amdgpu_irq_put(adev, &adev->gfx.cp_ecc_error_irq, 0);
+
+	amdgpu_irq_put(adev, &adev->gfx.sq_irq, 0);
+
+	/* disable KCQ to avoid CPC touch memory not valid anymore */
+	gfx_v8_0_kcq_disable(adev);
+
+	if (amdgpu_sriov_vf(adev)) {
+		pr_debug("For SRIOV client, shouldn't do anything.\n");
+		return 0;
+	}
+
+	amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
+	if (!gfx_v8_0_wait_for_idle(ip_block))
+		gfx_v8_0_cp_enable(adev, false);
+	else
+		pr_err("cp is busy, skip halt cp\n");
+	if (!gfx_v8_0_wait_for_rlc_idle(adev))
+		adev->gfx.rlc.funcs->stop(adev);
+	else
+		pr_err("rlc is busy, skip halt rlc\n");
+	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
+
+	return 0;
+}
+
+static int gfx_v8_0_suspend(struct amdgpu_ip_block *ip_block)
+{
+	return gfx_v8_0_hw_fini(ip_block);
+}
+
+static int gfx_v8_0_resume(struct amdgpu_ip_block *ip_block)
+{
+	return gfx_v8_0_hw_init(ip_block);
+}
+
+static bool gfx_v8_0_check_soft_reset(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 grbm_soft_reset = 0, srbm_soft_reset = 0;
 	u32 tmp;
 
@@ -5097,20 +4944,19 @@ static bool gfx_v8_0_check_soft_reset(void *handle)
 	}
 }
 
-static int gfx_v8_0_pre_soft_reset(void *handle)
+static int gfx_v8_0_pre_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	u32 grbm_soft_reset = 0, srbm_soft_reset = 0;
+	struct amdgpu_device *adev = ip_block->adev;
+	u32 grbm_soft_reset = 0;
 
 	if ((!adev->gfx.grbm_soft_reset) &&
 	    (!adev->gfx.srbm_soft_reset))
 		return 0;
 
 	grbm_soft_reset = adev->gfx.grbm_soft_reset;
-	srbm_soft_reset = adev->gfx.srbm_soft_reset;
 
 	/* stop the rlc */
-	gfx_v8_0_rlc_stop(adev);
+	adev->gfx.rlc.funcs->stop(adev);
 
 	if (REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CP) ||
 	    REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_GFX))
@@ -5136,12 +4982,12 @@ static int gfx_v8_0_pre_soft_reset(void *handle)
 		gfx_v8_0_cp_compute_enable(adev, false);
 	}
 
-       return 0;
+	return 0;
 }
 
-static int gfx_v8_0_soft_reset(void *handle)
+static int gfx_v8_0_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 grbm_soft_reset = 0, srbm_soft_reset = 0;
 	u32 tmp;
 
@@ -5201,21 +5047,16 @@ static int gfx_v8_0_soft_reset(void *handle)
 	return 0;
 }
 
-static int gfx_v8_0_post_soft_reset(void *handle)
+static int gfx_v8_0_post_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	u32 grbm_soft_reset = 0, srbm_soft_reset = 0;
+	struct amdgpu_device *adev = ip_block->adev;
+	u32 grbm_soft_reset = 0;
 
 	if ((!adev->gfx.grbm_soft_reset) &&
 	    (!adev->gfx.srbm_soft_reset))
 		return 0;
 
 	grbm_soft_reset = adev->gfx.grbm_soft_reset;
-	srbm_soft_reset = adev->gfx.srbm_soft_reset;
-
-	if (REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CP) ||
-	    REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_GFX))
-		gfx_v8_0_cp_gfx_resume(adev);
 
 	if (REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CP) ||
 	    REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CPF) ||
@@ -5233,8 +5074,16 @@ static int gfx_v8_0_post_soft_reset(void *handle)
 			mutex_unlock(&adev->srbm_mutex);
 		}
 		gfx_v8_0_kiq_resume(adev);
+		gfx_v8_0_kcq_resume(adev);
 	}
-	gfx_v8_0_rlc_start(adev);
+
+	if (REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CP) ||
+	    REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_GFX))
+		gfx_v8_0_cp_gfx_resume(adev);
+
+	gfx_v8_0_cp_test_all_rings(adev);
+
+	adev->gfx.rlc.funcs->start(adev);
 
 	return 0;
 }
@@ -5265,15 +5114,6 @@ static void gfx_v8_0_ring_emit_gds_switch(struct amdgpu_ring *ring,
 					  uint32_t gws_base, uint32_t gws_size,
 					  uint32_t oa_base, uint32_t oa_size)
 {
-	gds_base = gds_base >> AMDGPU_GDS_SHIFT;
-	gds_size = gds_size >> AMDGPU_GDS_SHIFT;
-
-	gws_base = gws_base >> AMDGPU_GWS_SHIFT;
-	gws_size = gws_size >> AMDGPU_GWS_SHIFT;
-
-	oa_base = oa_base >> AMDGPU_OA_SHIFT;
-	oa_size = oa_size >> AMDGPU_OA_SHIFT;
-
 	/* GDS Base */
 	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
 	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
@@ -5332,7 +5172,7 @@ static void wave_read_regs(struct amdgpu_device *adev, uint32_t simd,
 		*(out++) = RREG32(mmSQ_IND_DATA);
 }
 
-static void gfx_v8_0_read_wave_data(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t *dst, int *no_fields)
+static void gfx_v8_0_read_wave_data(struct amdgpu_device *adev, uint32_t xcc_id, uint32_t simd, uint32_t wave, uint32_t *dst, int *no_fields)
 {
 	/* type 0 wave data */
 	dst[(*no_fields)++] = 0;
@@ -5354,9 +5194,10 @@ static void gfx_v8_0_read_wave_data(struct amdgpu_device *adev, uint32_t simd, u
 	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TMA_HI);
 	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_IB_DBG0);
 	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_M0);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_MODE);
 }
 
-static void gfx_v8_0_read_wave_sgprs(struct amdgpu_device *adev, uint32_t simd,
+static void gfx_v8_0_read_wave_sgprs(struct amdgpu_device *adev, uint32_t xcc_id, uint32_t simd,
 				     uint32_t wave, uint32_t start,
 				     uint32_t size, uint32_t *dst)
 {
@@ -5371,14 +5212,17 @@ static const struct amdgpu_gfx_funcs gfx_v8_0_gfx_funcs = {
 	.select_se_sh = &gfx_v8_0_select_se_sh,
 	.read_wave_data = &gfx_v8_0_read_wave_data,
 	.read_wave_sgprs = &gfx_v8_0_read_wave_sgprs,
+	.select_me_pipe_q = &gfx_v8_0_select_me_pipe_q
 };
 
-static int gfx_v8_0_early_init(void *handle)
+static int gfx_v8_0_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
+	adev->gfx.xcc_mask = 1;
 	adev->gfx.num_gfx_rings = GFX8_NUM_GFX_RINGS;
-	adev->gfx.num_compute_rings = AMDGPU_MAX_COMPUTE_RINGS;
+	adev->gfx.num_compute_rings = min(amdgpu_gfx_get_num_kcq(adev),
+					  AMDGPU_MAX_COMPUTE_RINGS);
 	adev->gfx.funcs = &gfx_v8_0_gfx_funcs;
 	gfx_v8_0_set_ring_funcs(adev);
 	gfx_v8_0_set_irq_funcs(adev);
@@ -5388,9 +5232,9 @@ static int gfx_v8_0_early_init(void *handle)
 	return 0;
 }
 
-static int gfx_v8_0_late_init(void *handle)
+static int gfx_v8_0_late_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int r;
 
 	r = amdgpu_irq_get(adev, &adev->gfx.priv_reg_irq, 0);
@@ -5406,8 +5250,19 @@ static int gfx_v8_0_late_init(void *handle)
 	if (r)
 		return r;
 
-	amdgpu_set_powergating_state(adev,
-			AMD_IP_BLOCK_TYPE_GFX, AMD_PG_STATE_GATE);
+	r = amdgpu_irq_get(adev, &adev->gfx.cp_ecc_error_irq, 0);
+	if (r) {
+		DRM_ERROR("amdgpu_irq_get() failed to get IRQ for EDC, r: %d.\n", r);
+		return r;
+	}
+
+	r = amdgpu_irq_get(adev, &adev->gfx.sq_irq, 0);
+	if (r) {
+		DRM_ERROR(
+			"amdgpu_irq_get() failed to get IRQ for SQ, r: %d.\n",
+			r);
+		return r;
+	}
 
 	return 0;
 }
@@ -5416,12 +5271,10 @@ static void gfx_v8_0_enable_gfx_static_mg_power_gating(struct amdgpu_device *ade
 						       bool enable)
 {
 	if ((adev->asic_type == CHIP_POLARIS11) ||
-	    (adev->asic_type == CHIP_POLARIS12))
+	    (adev->asic_type == CHIP_POLARIS12) ||
+	    (adev->asic_type == CHIP_VEGAM))
 		/* Send msg to SMU via Powerplay */
-		amdgpu_set_powergating_state(adev,
-					     AMD_IP_BLOCK_TYPE_SMC,
-					     enable ?
-					     AMD_PG_STATE_GATE : AMD_PG_STATE_UNGATE);
+		amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_GFX, enable, 0);
 
 	WREG32_FIELD(RLC_PG_CNTL, STATIC_PER_CU_PG_ENABLE, enable ? 1 : 0);
 }
@@ -5467,15 +5320,20 @@ static void cz_update_gfx_cg_power_gating(struct amdgpu_device *adev,
 	}
 }
 
-static int gfx_v8_0_set_powergating_state(void *handle,
+static int gfx_v8_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	bool enable = (state == AMD_PG_STATE_GATE);
 
 	if (amdgpu_sriov_vf(adev))
 		return 0;
 
+	if (adev->pg_flags & (AMD_PG_SUPPORT_GFX_SMG |
+				AMD_PG_SUPPORT_RLC_SMU_HS |
+				AMD_PG_SUPPORT_CP |
+				AMD_PG_SUPPORT_GFX_DMG))
+		amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
 	switch (adev->asic_type) {
 	case CHIP_CARRIZO:
 	case CHIP_STONEY:
@@ -5506,6 +5364,7 @@ static int gfx_v8_0_set_powergating_state(void *handle,
 		break;
 	case CHIP_POLARIS11:
 	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
 		if ((adev->pg_flags & AMD_PG_SUPPORT_GFX_SMG) && enable)
 			gfx_v8_0_enable_gfx_static_mg_power_gating(adev, true);
 		else
@@ -5524,13 +5383,17 @@ static int gfx_v8_0_set_powergating_state(void *handle,
 	default:
 		break;
 	}
-
+	if (adev->pg_flags & (AMD_PG_SUPPORT_GFX_SMG |
+				AMD_PG_SUPPORT_RLC_SMU_HS |
+				AMD_PG_SUPPORT_CP |
+				AMD_PG_SUPPORT_GFX_DMG))
+		amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
 	return 0;
 }
 
-static void gfx_v8_0_get_clockgating_state(void *handle, u32 *flags)
+static void gfx_v8_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int data;
 
 	if (amdgpu_sriov_vf(adev))
@@ -5575,7 +5438,7 @@ static void gfx_v8_0_send_serdes_cmd(struct amdgpu_device *adev,
 {
 	uint32_t data;
 
-	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 
 	WREG32(mmRLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff);
 	WREG32(mmRLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff);
@@ -5618,57 +5481,53 @@ static void gfx_v8_0_send_serdes_cmd(struct amdgpu_device *adev,
 #define RLC_GPR_REG2__MESSAGE__SHIFT 0x00000001
 #define RLC_GPR_REG2__MESSAGE_MASK 0x0000001e
 
-static void iceland_enter_rlc_safe_mode(struct amdgpu_device *adev)
+static bool gfx_v8_0_is_rlc_enabled(struct amdgpu_device *adev)
 {
-	u32 data;
-	unsigned i;
+	uint32_t rlc_setting;
 
-	data = RREG32(mmRLC_CNTL);
-	if (!(data & RLC_CNTL__RLC_ENABLE_F32_MASK))
-		return;
+	rlc_setting = RREG32(mmRLC_CNTL);
+	if (!(rlc_setting & RLC_CNTL__RLC_ENABLE_F32_MASK))
+		return false;
 
-	if (adev->cg_flags & (AMD_CG_SUPPORT_GFX_CGCG | AMD_CG_SUPPORT_GFX_MGCG)) {
-		data |= RLC_SAFE_MODE__CMD_MASK;
-		data &= ~RLC_SAFE_MODE__MESSAGE_MASK;
-		data |= (1 << RLC_SAFE_MODE__MESSAGE__SHIFT);
-		WREG32(mmRLC_SAFE_MODE, data);
+	return true;
+}
 
-		for (i = 0; i < adev->usec_timeout; i++) {
-			if ((RREG32(mmRLC_GPM_STAT) &
-			     (RLC_GPM_STAT__GFX_CLOCK_STATUS_MASK |
-			      RLC_GPM_STAT__GFX_POWER_STATUS_MASK)) ==
-			    (RLC_GPM_STAT__GFX_CLOCK_STATUS_MASK |
-			     RLC_GPM_STAT__GFX_POWER_STATUS_MASK))
-				break;
-			udelay(1);
-		}
+static void gfx_v8_0_set_safe_mode(struct amdgpu_device *adev, int xcc_id)
+{
+	uint32_t data;
+	unsigned i;
+	data = RREG32(mmRLC_CNTL);
+	data |= RLC_SAFE_MODE__CMD_MASK;
+	data &= ~RLC_SAFE_MODE__MESSAGE_MASK;
+	data |= (1 << RLC_SAFE_MODE__MESSAGE__SHIFT);
+	WREG32(mmRLC_SAFE_MODE, data);
 
-		for (i = 0; i < adev->usec_timeout; i++) {
-			if (!REG_GET_FIELD(RREG32(mmRLC_SAFE_MODE), RLC_SAFE_MODE, CMD))
-				break;
-			udelay(1);
-		}
-		adev->gfx.rlc.in_safe_mode = true;
+	/* wait for RLC_SAFE_MODE */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if ((RREG32(mmRLC_GPM_STAT) &
+		     (RLC_GPM_STAT__GFX_CLOCK_STATUS_MASK |
+		      RLC_GPM_STAT__GFX_POWER_STATUS_MASK)) ==
+		    (RLC_GPM_STAT__GFX_CLOCK_STATUS_MASK |
+		     RLC_GPM_STAT__GFX_POWER_STATUS_MASK))
+			break;
+		udelay(1);
+	}
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (!REG_GET_FIELD(RREG32(mmRLC_SAFE_MODE), RLC_SAFE_MODE, CMD))
+			break;
+		udelay(1);
 	}
 }
 
-static void iceland_exit_rlc_safe_mode(struct amdgpu_device *adev)
+static void gfx_v8_0_unset_safe_mode(struct amdgpu_device *adev, int xcc_id)
 {
-	u32 data = 0;
+	uint32_t data;
 	unsigned i;
 
 	data = RREG32(mmRLC_CNTL);
-	if (!(data & RLC_CNTL__RLC_ENABLE_F32_MASK))
-		return;
-
-	if (adev->cg_flags & (AMD_CG_SUPPORT_GFX_CGCG | AMD_CG_SUPPORT_GFX_MGCG)) {
-		if (adev->gfx.rlc.in_safe_mode) {
-			data |= RLC_SAFE_MODE__CMD_MASK;
-			data &= ~RLC_SAFE_MODE__MESSAGE_MASK;
-			WREG32(mmRLC_SAFE_MODE, data);
-			adev->gfx.rlc.in_safe_mode = false;
-		}
-	}
+	data |= RLC_SAFE_MODE__CMD_MASK;
+	data &= ~RLC_SAFE_MODE__MESSAGE_MASK;
+	WREG32(mmRLC_SAFE_MODE, data);
 
 	for (i = 0; i < adev->usec_timeout; i++) {
 		if (!REG_GET_FIELD(RREG32(mmRLC_SAFE_MODE), RLC_SAFE_MODE, CMD))
@@ -5677,9 +5536,41 @@ static void iceland_exit_rlc_safe_mode(struct amdgpu_device *adev)
 	}
 }
 
+static void gfx_v8_0_update_spm_vmid(struct amdgpu_device *adev, struct amdgpu_ring *ring, unsigned vmid)
+{
+	u32 data;
+
+	amdgpu_gfx_off_ctrl(adev, false);
+
+	if (amdgpu_sriov_is_pp_one_vf(adev))
+		data = RREG32_NO_KIQ(mmRLC_SPM_VMID);
+	else
+		data = RREG32(mmRLC_SPM_VMID);
+
+	data &= ~RLC_SPM_VMID__RLC_SPM_VMID_MASK;
+	data |= (vmid & RLC_SPM_VMID__RLC_SPM_VMID_MASK) << RLC_SPM_VMID__RLC_SPM_VMID__SHIFT;
+
+	if (amdgpu_sriov_is_pp_one_vf(adev))
+		WREG32_NO_KIQ(mmRLC_SPM_VMID, data);
+	else
+		WREG32(mmRLC_SPM_VMID, data);
+
+	amdgpu_gfx_off_ctrl(adev, true);
+}
+
 static const struct amdgpu_rlc_funcs iceland_rlc_funcs = {
-	.enter_safe_mode = iceland_enter_rlc_safe_mode,
-	.exit_safe_mode = iceland_exit_rlc_safe_mode
+	.is_rlc_enabled = gfx_v8_0_is_rlc_enabled,
+	.set_safe_mode = gfx_v8_0_set_safe_mode,
+	.unset_safe_mode = gfx_v8_0_unset_safe_mode,
+	.init = gfx_v8_0_rlc_init,
+	.get_csb_size = gfx_v8_0_get_csb_size,
+	.get_csb_buffer = gfx_v8_0_get_csb_buffer,
+	.get_cp_table_num = gfx_v8_0_cp_jump_table_num,
+	.resume = gfx_v8_0_rlc_resume,
+	.stop = gfx_v8_0_rlc_stop,
+	.reset = gfx_v8_0_rlc_reset,
+	.start = gfx_v8_0_rlc_start,
+	.update_spm_vmid = gfx_v8_0_update_spm_vmid
 };
 
 static void gfx_v8_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
@@ -5687,8 +5578,6 @@ static void gfx_v8_0_update_medium_grain_clock_gating(struct amdgpu_device *adev
 {
 	uint32_t temp, data;
 
-	adev->gfx.rlc.funcs->enter_safe_mode(adev);
-
 	/* It is disabled by HW by default */
 	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGCG)) {
 		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGLS) {
@@ -5782,8 +5671,6 @@ static void gfx_v8_0_update_medium_grain_clock_gating(struct amdgpu_device *adev
 		/* 7- wait for RLC_SERDES_CU_MASTER & RLC_SERDES_NONCU_MASTER idle */
 		gfx_v8_0_wait_for_rlc_serdes(adev);
 	}
-
-	adev->gfx.rlc.funcs->exit_safe_mode(adev);
 }
 
 static void gfx_v8_0_update_coarse_grain_clock_gating(struct amdgpu_device *adev,
@@ -5793,8 +5680,6 @@ static void gfx_v8_0_update_coarse_grain_clock_gating(struct amdgpu_device *adev
 
 	temp = data = RREG32(mmRLC_CGCG_CGLS_CTRL);
 
-	adev->gfx.rlc.funcs->enter_safe_mode(adev);
-
 	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGCG)) {
 		temp1 = data1 =	RREG32(mmRLC_CGTT_MGCG_OVERRIDE);
 		data1 &= ~RLC_CGTT_MGCG_OVERRIDE__CGCG_MASK;
@@ -5856,7 +5741,7 @@ static void gfx_v8_0_update_coarse_grain_clock_gating(struct amdgpu_device *adev
 		/* wait for RLC_SERDES_CU_MASTER & RLC_SERDES_NONCU_MASTER idle */
 		gfx_v8_0_wait_for_rlc_serdes(adev);
 
-		/* write cmd to Set CGCG Overrride */
+		/* write cmd to Set CGCG Override */
 		gfx_v8_0_send_serdes_cmd(adev, BPM_REG_CGCG_OVERRIDE, SET_BPM_SERDES_CMD);
 
 		/* wait for RLC_SERDES_CU_MASTER & RLC_SERDES_NONCU_MASTER idle */
@@ -5875,12 +5760,12 @@ static void gfx_v8_0_update_coarse_grain_clock_gating(struct amdgpu_device *adev
 	}
 
 	gfx_v8_0_wait_for_rlc_serdes(adev);
-
-	adev->gfx.rlc.funcs->exit_safe_mode(adev);
 }
 static int gfx_v8_0_update_gfx_clock_gating(struct amdgpu_device *adev,
 					    bool enable)
 {
+	amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
+
 	if (enable) {
 		/* CGCG/CGLS should be enabled after MGCG/MGLS/TS(CG/LS)
 		 * ===  MGCG + MGLS + TS(CG/LS) ===
@@ -5894,6 +5779,8 @@ static int gfx_v8_0_update_gfx_clock_gating(struct amdgpu_device *adev,
 		gfx_v8_0_update_coarse_grain_clock_gating(adev, enable);
 		gfx_v8_0_update_medium_grain_clock_gating(adev, enable);
 	}
+
+	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
 	return 0;
 }
 
@@ -5902,7 +5789,6 @@ static int gfx_v8_0_tonga_update_gfx_clock_gating(struct amdgpu_device *adev,
 {
 	uint32_t msg_id, pp_state = 0;
 	uint32_t pp_support_state = 0;
-	void *pp_handle = adev->powerplay.pp_handle;
 
 	if (adev->cg_flags & (AMD_CG_SUPPORT_GFX_CGCG | AMD_CG_SUPPORT_GFX_CGLS)) {
 		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGLS) {
@@ -5920,7 +5806,7 @@ static int gfx_v8_0_tonga_update_gfx_clock_gating(struct amdgpu_device *adev,
 				PP_BLOCK_GFX_CG,
 				pp_support_state,
 				pp_state);
-		amd_set_clockgating_by_smu(pp_handle, msg_id);
+		amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
 	}
 
 	if (adev->cg_flags & (AMD_CG_SUPPORT_GFX_MGCG | AMD_CG_SUPPORT_GFX_MGLS)) {
@@ -5941,7 +5827,7 @@ static int gfx_v8_0_tonga_update_gfx_clock_gating(struct amdgpu_device *adev,
 				PP_BLOCK_GFX_MG,
 				pp_support_state,
 				pp_state);
-		amd_set_clockgating_by_smu(pp_handle, msg_id);
+		amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
 	}
 
 	return 0;
@@ -5953,7 +5839,6 @@ static int gfx_v8_0_polaris_update_gfx_clock_gating(struct amdgpu_device *adev,
 
 	uint32_t msg_id, pp_state = 0;
 	uint32_t pp_support_state = 0;
-	void *pp_handle = adev->powerplay.pp_handle;
 
 	if (adev->cg_flags & (AMD_CG_SUPPORT_GFX_CGCG | AMD_CG_SUPPORT_GFX_CGLS)) {
 		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGLS) {
@@ -5971,7 +5856,7 @@ static int gfx_v8_0_polaris_update_gfx_clock_gating(struct amdgpu_device *adev,
 				PP_BLOCK_GFX_CG,
 				pp_support_state,
 				pp_state);
-		amd_set_clockgating_by_smu(pp_handle, msg_id);
+		amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
 	}
 
 	if (adev->cg_flags & (AMD_CG_SUPPORT_GFX_3D_CGCG | AMD_CG_SUPPORT_GFX_3D_CGLS)) {
@@ -5990,7 +5875,7 @@ static int gfx_v8_0_polaris_update_gfx_clock_gating(struct amdgpu_device *adev,
 				PP_BLOCK_GFX_3D,
 				pp_support_state,
 				pp_state);
-		amd_set_clockgating_by_smu(pp_handle, msg_id);
+		amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
 	}
 
 	if (adev->cg_flags & (AMD_CG_SUPPORT_GFX_MGCG | AMD_CG_SUPPORT_GFX_MGLS)) {
@@ -6011,7 +5896,7 @@ static int gfx_v8_0_polaris_update_gfx_clock_gating(struct amdgpu_device *adev,
 				PP_BLOCK_GFX_MG,
 				pp_support_state,
 				pp_state);
-		amd_set_clockgating_by_smu(pp_handle, msg_id);
+		amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
 	}
 
 	if (adev->cg_flags & AMD_CG_SUPPORT_GFX_RLC_LS) {
@@ -6026,7 +5911,7 @@ static int gfx_v8_0_polaris_update_gfx_clock_gating(struct amdgpu_device *adev,
 				PP_BLOCK_GFX_RLC,
 				pp_support_state,
 				pp_state);
-		amd_set_clockgating_by_smu(pp_handle, msg_id);
+		amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
 	}
 
 	if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CP_LS) {
@@ -6040,16 +5925,16 @@ static int gfx_v8_0_polaris_update_gfx_clock_gating(struct amdgpu_device *adev,
 			PP_BLOCK_GFX_CP,
 			pp_support_state,
 			pp_state);
-		amd_set_clockgating_by_smu(pp_handle, msg_id);
+		amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
 	}
 
 	return 0;
 }
 
-static int gfx_v8_0_set_clockgating_state(void *handle,
+static int gfx_v8_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (amdgpu_sriov_vf(adev))
 		return 0;
@@ -6067,6 +5952,7 @@ static int gfx_v8_0_set_clockgating_state(void *handle,
 	case CHIP_POLARIS10:
 	case CHIP_POLARIS11:
 	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
 		gfx_v8_0_polaris_update_gfx_clock_gating(adev, state);
 		break;
 	default:
@@ -6077,7 +5963,7 @@ static int gfx_v8_0_set_clockgating_state(void *handle,
 
 static u64 gfx_v8_0_ring_get_rptr(struct amdgpu_ring *ring)
 {
-	return ring->adev->wb.wb[ring->rptr_offs];
+	return *ring->rptr_cpu_addr;
 }
 
 static u64 gfx_v8_0_ring_get_wptr_gfx(struct amdgpu_ring *ring)
@@ -6086,7 +5972,7 @@ static u64 gfx_v8_0_ring_get_wptr_gfx(struct amdgpu_ring *ring)
 
 	if (ring->use_doorbell)
 		/* XXX check if swapping is necessary on BE */
-		return ring->adev->wb.wb[ring->wptr_offs];
+		return *ring->wptr_cpu_addr;
 	else
 		return RREG32(mmCP_RB0_WPTR);
 }
@@ -6097,7 +5983,7 @@ static void gfx_v8_0_ring_set_wptr_gfx(struct amdgpu_ring *ring)
 
 	if (ring->use_doorbell) {
 		/* XXX check if swapping is necessary on BE */
-		adev->wb.wb[ring->wptr_offs] = lower_32_bits(ring->wptr);
+		*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
 		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
 	} else {
 		WREG32(mmCP_RB0_WPTR, lower_32_bits(ring->wptr));
@@ -6149,23 +6035,12 @@ static void gfx_v8_0_ring_emit_vgt_flush(struct amdgpu_ring *ring)
 		EVENT_INDEX(0));
 }
 
-
-static void gfx_v8_0_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
-{
-	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
-	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
-				 WRITE_DATA_DST_SEL(0) |
-				 WR_CONFIRM));
-	amdgpu_ring_write(ring, mmHDP_DEBUG0);
-	amdgpu_ring_write(ring, 0);
-	amdgpu_ring_write(ring, 1);
-
-}
-
 static void gfx_v8_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
-				      struct amdgpu_ib *ib,
-				      unsigned vm_id, bool ctx_switch)
+					struct amdgpu_job *job,
+					struct amdgpu_ib *ib,
+					uint32_t flags)
 {
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
 	u32 header, control = 0;
 
 	if (ib->flags & AMDGPU_IB_FLAG_CE)
@@ -6173,12 +6048,12 @@ static void gfx_v8_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
 	else
 		header = PACKET3(PACKET3_INDIRECT_BUFFER, 2);
 
-	control |= ib->length_dw | (vm_id << 24);
+	control |= ib->length_dw | (vmid << 24);
 
 	if (amdgpu_sriov_vf(ring->adev) && (ib->flags & AMDGPU_IB_FLAG_PREEMPT)) {
 		control |= INDIRECT_BUFFER_PRE_ENB(1);
 
-		if (!(ib->flags & AMDGPU_IB_FLAG_CE))
+		if (!(ib->flags & AMDGPU_IB_FLAG_CE) && vmid)
 			gfx_v8_0_ring_emit_de_meta(ring);
 	}
 
@@ -6193,10 +6068,28 @@ static void gfx_v8_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
 }
 
 static void gfx_v8_0_ring_emit_ib_compute(struct amdgpu_ring *ring,
+					  struct amdgpu_job *job,
 					  struct amdgpu_ib *ib,
-					  unsigned vm_id, bool ctx_switch)
-{
-	u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vm_id << 24);
+					  uint32_t flags)
+{
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+	u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vmid << 24);
+
+	/* Currently, there is a high possibility to get wave ID mismatch
+	 * between ME and GDS, leading to a hw deadlock, because ME generates
+	 * different wave IDs than the GDS expects. This situation happens
+	 * randomly when at least 5 compute pipes use GDS ordered append.
+	 * The wave IDs generated by ME are also wrong after suspend/resume.
+	 * Those are probably bugs somewhere else in the kernel driver.
+	 *
+	 * Writing GDS_COMPUTE_MAX_WAVE_ID resets wave ID counters in ME and
+	 * GDS to 0 for this ring (me/pipe).
+	 */
+	if (ib->flags & AMDGPU_IB_FLAG_RESET_GDS_MAX_WAVE_ID) {
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1));
+		amdgpu_ring_write(ring, mmGDS_COMPUTE_MAX_WAVE_ID - PACKET3_SET_CONFIG_REG_START);
+		amdgpu_ring_write(ring, ring->adev->gds.gds_compute_max_wave_id);
+	}
 
 	amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2));
 	amdgpu_ring_write(ring,
@@ -6213,8 +6106,11 @@ static void gfx_v8_0_ring_emit_fence_gfx(struct amdgpu_ring *ring, u64 addr,
 {
 	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
 	bool int_sel = flags & AMDGPU_FENCE_FLAG_INT;
+	bool exec = flags & AMDGPU_FENCE_FLAG_EXEC;
 
-	/* EVENT_WRITE_EOP - flush caches, send int */
+	/* Workaround for cache flush problems. First send a dummy EOP
+	 * event down the pipe with seq one below.
+	 */
 	amdgpu_ring_write(ring, PACKET3(PACKET3_EVENT_WRITE_EOP, 4));
 	amdgpu_ring_write(ring, (EOP_TCL1_ACTION_EN |
 				 EOP_TC_ACTION_EN |
@@ -6223,6 +6119,21 @@ static void gfx_v8_0_ring_emit_fence_gfx(struct amdgpu_ring *ring, u64 addr,
 				 EVENT_INDEX(5)));
 	amdgpu_ring_write(ring, addr & 0xfffffffc);
 	amdgpu_ring_write(ring, (upper_32_bits(addr) & 0xffff) |
+				DATA_SEL(1) | INT_SEL(0));
+	amdgpu_ring_write(ring, lower_32_bits(seq - 1));
+	amdgpu_ring_write(ring, upper_32_bits(seq - 1));
+
+	/* Then send the real EOP event down the pipe:
+	 * EVENT_WRITE_EOP - flush caches, send int */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_EVENT_WRITE_EOP, 4));
+	amdgpu_ring_write(ring, (EOP_TCL1_ACTION_EN |
+				 EOP_TC_ACTION_EN |
+				 EOP_TC_WB_ACTION_EN |
+				 EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) |
+				 EVENT_INDEX(5) |
+				 (exec ? EOP_EXEC : 0)));
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, (upper_32_bits(addr) & 0xffff) |
 			  DATA_SEL(write64bit ? 2 : 1) | INT_SEL(int_sel ? 2 : 0));
 	amdgpu_ring_write(ring, lower_32_bits(seq));
 	amdgpu_ring_write(ring, upper_32_bits(seq));
@@ -6247,32 +6158,11 @@ static void gfx_v8_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
 }
 
 static void gfx_v8_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
-					unsigned vm_id, uint64_t pd_addr)
+					unsigned vmid, uint64_t pd_addr)
 {
 	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
 
-	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
-	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(usepfp) |
-				 WRITE_DATA_DST_SEL(0)) |
-				 WR_CONFIRM);
-	if (vm_id < 8) {
-		amdgpu_ring_write(ring,
-				  (mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + vm_id));
-	} else {
-		amdgpu_ring_write(ring,
-				  (mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vm_id - 8));
-	}
-	amdgpu_ring_write(ring, 0);
-	amdgpu_ring_write(ring, pd_addr >> 12);
-
-	/* bits 0-15 are the VM contexts0-15 */
-	/* invalidate the cache */
-	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
-	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
-				 WRITE_DATA_DST_SEL(0)));
-	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST);
-	amdgpu_ring_write(ring, 0);
-	amdgpu_ring_write(ring, 1 << vm_id);
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
 
 	/* wait for the invalidate to complete */
 	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
@@ -6295,7 +6185,7 @@ static void gfx_v8_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
 
 static u64 gfx_v8_0_ring_get_wptr_compute(struct amdgpu_ring *ring)
 {
-	return ring->adev->wb.wb[ring->wptr_offs];
+	return *ring->wptr_cpu_addr;
 }
 
 static void gfx_v8_0_ring_set_wptr_compute(struct amdgpu_ring *ring)
@@ -6303,7 +6193,7 @@ static void gfx_v8_0_ring_set_wptr_compute(struct amdgpu_ring *ring)
 	struct amdgpu_device *adev = ring->adev;
 
 	/* XXX check if swapping is necessary on BE */
-	adev->wb.wb[ring->wptr_offs] = lower_32_bits(ring->wptr);
+	*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
 	WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
 }
 
@@ -6392,34 +6282,24 @@ static void gfx_v8_ring_emit_cntxcntl(struct amdgpu_ring *ring, uint32_t flags)
 	amdgpu_ring_write(ring, 0);
 }
 
-static unsigned gfx_v8_0_ring_emit_init_cond_exec(struct amdgpu_ring *ring)
+static unsigned gfx_v8_0_ring_emit_init_cond_exec(struct amdgpu_ring *ring,
+						  uint64_t addr)
 {
 	unsigned ret;
 
 	amdgpu_ring_write(ring, PACKET3(PACKET3_COND_EXEC, 3));
-	amdgpu_ring_write(ring, lower_32_bits(ring->cond_exe_gpu_addr));
-	amdgpu_ring_write(ring, upper_32_bits(ring->cond_exe_gpu_addr));
-	amdgpu_ring_write(ring, 0); /* discard following DWs if *cond_exec_gpu_addr==0 */
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	/* discard following DWs if *cond_exec_gpu_addr==0 */
+	amdgpu_ring_write(ring, 0);
 	ret = ring->wptr & ring->buf_mask;
-	amdgpu_ring_write(ring, 0x55aa55aa); /* patch dummy value later */
+	/* patch dummy value later */
+	amdgpu_ring_write(ring, 0);
 	return ret;
 }
 
-static void gfx_v8_0_ring_emit_patch_cond_exec(struct amdgpu_ring *ring, unsigned offset)
-{
-	unsigned cur;
-
-	BUG_ON(offset > ring->buf_mask);
-	BUG_ON(ring->ring[offset] != 0x55aa55aa);
-
-	cur = (ring->wptr & ring->buf_mask) - 1;
-	if (likely(cur > offset))
-		ring->ring[offset] = cur - offset;
-	else
-		ring->ring[offset] = (ring->ring_size >> 2) - offset + cur;
-}
-
-static void gfx_v8_0_ring_emit_rreg(struct amdgpu_ring *ring, uint32_t reg)
+static void gfx_v8_0_ring_emit_rreg(struct amdgpu_ring *ring, uint32_t reg,
+				    uint32_t reg_val_offs)
 {
 	struct amdgpu_device *adev = ring->adev;
 
@@ -6430,21 +6310,47 @@ static void gfx_v8_0_ring_emit_rreg(struct amdgpu_ring *ring, uint32_t reg)
 	amdgpu_ring_write(ring, reg);
 	amdgpu_ring_write(ring, 0);
 	amdgpu_ring_write(ring, lower_32_bits(adev->wb.gpu_addr +
-				adev->virt.reg_val_offs * 4));
+				reg_val_offs * 4));
 	amdgpu_ring_write(ring, upper_32_bits(adev->wb.gpu_addr +
-				adev->virt.reg_val_offs * 4));
+				reg_val_offs * 4));
 }
 
 static void gfx_v8_0_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg,
 				  uint32_t val)
 {
+	uint32_t cmd;
+
+	switch (ring->funcs->type) {
+	case AMDGPU_RING_TYPE_GFX:
+		cmd = WRITE_DATA_ENGINE_SEL(1) | WR_CONFIRM;
+		break;
+	case AMDGPU_RING_TYPE_KIQ:
+		cmd = 1 << 16; /* no inc addr */
+		break;
+	default:
+		cmd = WR_CONFIRM;
+		break;
+	}
+
 	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
-	amdgpu_ring_write(ring, (1 << 16)); /* no inc addr */
+	amdgpu_ring_write(ring, cmd);
 	amdgpu_ring_write(ring, reg);
 	amdgpu_ring_write(ring, 0);
 	amdgpu_ring_write(ring, val);
 }
 
+static void gfx_v8_0_ring_soft_recovery(struct amdgpu_ring *ring, unsigned vmid)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t value = 0;
+
+	value = REG_SET_FIELD(value, SQ_CMD, CMD, 0x03);
+	value = REG_SET_FIELD(value, SQ_CMD, MODE, 0x01);
+	value = REG_SET_FIELD(value, SQ_CMD, CHECK_VMID, 1);
+	value = REG_SET_FIELD(value, SQ_CMD, VM_ID, vmid);
+	WREG32(mmSQ_CMD, value);
+}
+
 static void gfx_v8_0_set_gfx_eop_interrupt_state(struct amdgpu_device *adev,
 						 enum amdgpu_interrupt_state state)
 {
@@ -6531,7 +6437,7 @@ static int gfx_v8_0_set_eop_interrupt_state(struct amdgpu_device *adev,
 					    enum amdgpu_interrupt_state state)
 {
 	switch (type) {
-	case AMDGPU_CP_IRQ_GFX_EOP:
+	case AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP:
 		gfx_v8_0_set_gfx_eop_interrupt_state(adev, state);
 		break;
 	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP:
@@ -6564,6 +6470,77 @@ static int gfx_v8_0_set_eop_interrupt_state(struct amdgpu_device *adev,
 	return 0;
 }
 
+static int gfx_v8_0_set_cp_ecc_int_state(struct amdgpu_device *adev,
+					 struct amdgpu_irq_src *source,
+					 unsigned int type,
+					 enum amdgpu_interrupt_state state)
+{
+	int enable_flag;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		enable_flag = 0;
+		break;
+
+	case AMDGPU_IRQ_STATE_ENABLE:
+		enable_flag = 1;
+		break;
+
+	default:
+		return -EINVAL;
+	}
+
+	WREG32_FIELD(CP_INT_CNTL, CP_ECC_ERROR_INT_ENABLE, enable_flag);
+	WREG32_FIELD(CP_INT_CNTL_RING0, CP_ECC_ERROR_INT_ENABLE, enable_flag);
+	WREG32_FIELD(CP_INT_CNTL_RING1, CP_ECC_ERROR_INT_ENABLE, enable_flag);
+	WREG32_FIELD(CP_INT_CNTL_RING2, CP_ECC_ERROR_INT_ENABLE, enable_flag);
+	WREG32_FIELD(CPC_INT_CNTL, CP_ECC_ERROR_INT_ENABLE, enable_flag);
+	WREG32_FIELD(CP_ME1_PIPE0_INT_CNTL, CP_ECC_ERROR_INT_ENABLE,
+		     enable_flag);
+	WREG32_FIELD(CP_ME1_PIPE1_INT_CNTL, CP_ECC_ERROR_INT_ENABLE,
+		     enable_flag);
+	WREG32_FIELD(CP_ME1_PIPE2_INT_CNTL, CP_ECC_ERROR_INT_ENABLE,
+		     enable_flag);
+	WREG32_FIELD(CP_ME1_PIPE3_INT_CNTL, CP_ECC_ERROR_INT_ENABLE,
+		     enable_flag);
+	WREG32_FIELD(CP_ME2_PIPE0_INT_CNTL, CP_ECC_ERROR_INT_ENABLE,
+		     enable_flag);
+	WREG32_FIELD(CP_ME2_PIPE1_INT_CNTL, CP_ECC_ERROR_INT_ENABLE,
+		     enable_flag);
+	WREG32_FIELD(CP_ME2_PIPE2_INT_CNTL, CP_ECC_ERROR_INT_ENABLE,
+		     enable_flag);
+	WREG32_FIELD(CP_ME2_PIPE3_INT_CNTL, CP_ECC_ERROR_INT_ENABLE,
+		     enable_flag);
+
+	return 0;
+}
+
+static int gfx_v8_0_set_sq_int_state(struct amdgpu_device *adev,
+				     struct amdgpu_irq_src *source,
+				     unsigned int type,
+				     enum amdgpu_interrupt_state state)
+{
+	int enable_flag;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		enable_flag = 1;
+		break;
+
+	case AMDGPU_IRQ_STATE_ENABLE:
+		enable_flag = 0;
+		break;
+
+	default:
+		return -EINVAL;
+	}
+
+	WREG32_FIELD(SQ_INTERRUPT_MSG_CTRL, STALL,
+		     enable_flag);
+
+	return 0;
+}
+
 static int gfx_v8_0_eop_irq(struct amdgpu_device *adev,
 			    struct amdgpu_irq_src *source,
 			    struct amdgpu_iv_entry *entry)
@@ -6596,12 +6573,39 @@ static int gfx_v8_0_eop_irq(struct amdgpu_device *adev,
 	return 0;
 }
 
+static void gfx_v8_0_fault(struct amdgpu_device *adev,
+			   struct amdgpu_iv_entry *entry)
+{
+	u8 me_id, pipe_id, queue_id;
+	struct amdgpu_ring *ring;
+	int i;
+
+	me_id = (entry->ring_id & 0x0c) >> 2;
+	pipe_id = (entry->ring_id & 0x03) >> 0;
+	queue_id = (entry->ring_id & 0x70) >> 4;
+
+	switch (me_id) {
+	case 0:
+		drm_sched_fault(&adev->gfx.gfx_ring[0].sched);
+		break;
+	case 1:
+	case 2:
+		for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+			ring = &adev->gfx.compute_ring[i];
+			if (ring->me == me_id && ring->pipe == pipe_id &&
+			    ring->queue == queue_id)
+				drm_sched_fault(&ring->sched);
+		}
+		break;
+	}
+}
+
 static int gfx_v8_0_priv_reg_irq(struct amdgpu_device *adev,
 				 struct amdgpu_irq_src *source,
 				 struct amdgpu_iv_entry *entry)
 {
 	DRM_ERROR("Illegal register access in command stream\n");
-	schedule_work(&adev->reset_work);
+	gfx_v8_0_fault(adev, entry);
 	return 0;
 }
 
@@ -6610,54 +6614,205 @@ static int gfx_v8_0_priv_inst_irq(struct amdgpu_device *adev,
 				  struct amdgpu_iv_entry *entry)
 {
 	DRM_ERROR("Illegal instruction in command stream\n");
-	schedule_work(&adev->reset_work);
+	gfx_v8_0_fault(adev, entry);
 	return 0;
 }
 
-static int gfx_v8_0_kiq_set_interrupt_state(struct amdgpu_device *adev,
-					    struct amdgpu_irq_src *src,
-					    unsigned int type,
-					    enum amdgpu_interrupt_state state)
+static int gfx_v8_0_cp_ecc_error_irq(struct amdgpu_device *adev,
+				     struct amdgpu_irq_src *source,
+				     struct amdgpu_iv_entry *entry)
 {
-	struct amdgpu_ring *ring = &(adev->gfx.kiq.ring);
+	DRM_ERROR("CP EDC/ECC error detected.");
+	return 0;
+}
 
-	switch (type) {
-	case AMDGPU_CP_KIQ_IRQ_DRIVER0:
-		WREG32_FIELD(CPC_INT_CNTL, GENERIC2_INT_ENABLE,
-			     state == AMDGPU_IRQ_STATE_DISABLE ? 0 : 1);
-		if (ring->me == 1)
-			WREG32_FIELD_OFFSET(CP_ME1_PIPE0_INT_CNTL,
-				     ring->pipe,
-				     GENERIC2_INT_ENABLE,
-				     state == AMDGPU_IRQ_STATE_DISABLE ? 0 : 1);
-		else
-			WREG32_FIELD_OFFSET(CP_ME2_PIPE0_INT_CNTL,
-				     ring->pipe,
-				     GENERIC2_INT_ENABLE,
-				     state == AMDGPU_IRQ_STATE_DISABLE ? 0 : 1);
+static void gfx_v8_0_parse_sq_irq(struct amdgpu_device *adev, unsigned ih_data,
+				  bool from_wq)
+{
+	u32 enc, se_id, sh_id, cu_id;
+	char type[20];
+	int sq_edc_source = -1;
+
+	enc = REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_CMN, ENCODING);
+	se_id = REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_CMN, SE_ID);
+
+	switch (enc) {
+		case 0:
+			DRM_INFO("SQ general purpose intr detected:"
+					"se_id %d, immed_overflow %d, host_reg_overflow %d,"
+					"host_cmd_overflow %d, cmd_timestamp %d,"
+					"reg_timestamp %d, thread_trace_buff_full %d,"
+					"wlt %d, thread_trace %d.\n",
+					se_id,
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_AUTO, IMMED_OVERFLOW),
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_AUTO, HOST_REG_OVERFLOW),
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_AUTO, HOST_CMD_OVERFLOW),
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_AUTO, CMD_TIMESTAMP),
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_AUTO, REG_TIMESTAMP),
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_AUTO, THREAD_TRACE_BUF_FULL),
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_AUTO, WLT),
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_AUTO, THREAD_TRACE)
+					);
+			break;
+		case 1:
+		case 2:
+
+			cu_id = REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_WAVE, CU_ID);
+			sh_id = REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_WAVE, SH_ID);
+
+			/*
+			 * This function can be called either directly from ISR
+			 * or from BH in which case we can access SQ_EDC_INFO
+			 * instance
+			 */
+			if (from_wq) {
+				mutex_lock(&adev->grbm_idx_mutex);
+				gfx_v8_0_select_se_sh(adev, se_id, sh_id, cu_id, 0);
+
+				sq_edc_source = REG_GET_FIELD(RREG32(mmSQ_EDC_INFO), SQ_EDC_INFO, SOURCE);
+
+				gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
+				mutex_unlock(&adev->grbm_idx_mutex);
+			}
+
+			if (enc == 1)
+				sprintf(type, "instruction intr");
+			else
+				sprintf(type, "EDC/ECC error");
+
+			DRM_INFO(
+				"SQ %s detected: "
+					"se_id %d, sh_id %d, cu_id %d, simd_id %d, wave_id %d, vm_id %d "
+					"trap %s, sq_ed_info.source %s.\n",
+					type, se_id, sh_id, cu_id,
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_WAVE, SIMD_ID),
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_WAVE, WAVE_ID),
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_WAVE, VM_ID),
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_WAVE, PRIV) ? "true" : "false",
+					(sq_edc_source != -1) ? sq_edc_source_names[sq_edc_source] : "unavailable"
+				);
+			break;
+		default:
+			DRM_ERROR("SQ invalid encoding type\n.");
+	}
+}
+
+static void gfx_v8_0_sq_irq_work_func(struct work_struct *work)
+{
+
+	struct amdgpu_device *adev = container_of(work, struct amdgpu_device, gfx.sq_work.work);
+	struct sq_work *sq_work = container_of(work, struct sq_work, work);
+
+	gfx_v8_0_parse_sq_irq(adev, sq_work->ih_data, true);
+}
+
+static int gfx_v8_0_sq_irq(struct amdgpu_device *adev,
+			   struct amdgpu_irq_src *source,
+			   struct amdgpu_iv_entry *entry)
+{
+	unsigned ih_data = entry->src_data[0];
+
+	/*
+	 * Try to submit work so SQ_EDC_INFO can be accessed from
+	 * BH. If previous work submission hasn't finished yet
+	 * just print whatever info is possible directly from the ISR.
+	 */
+	if (work_pending(&adev->gfx.sq_work.work)) {
+		gfx_v8_0_parse_sq_irq(adev, ih_data, false);
+	} else {
+		adev->gfx.sq_work.ih_data = ih_data;
+		schedule_work(&adev->gfx.sq_work.work);
+	}
+
+	return 0;
+}
+
+static void gfx_v8_0_emit_mem_sync(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SURFACE_SYNC, 3));
+	amdgpu_ring_write(ring, PACKET3_TCL1_ACTION_ENA |
+			  PACKET3_TC_ACTION_ENA |
+			  PACKET3_SH_KCACHE_ACTION_ENA |
+			  PACKET3_SH_ICACHE_ACTION_ENA |
+			  PACKET3_TC_WB_ACTION_ENA);  /* CP_COHER_CNTL */
+	amdgpu_ring_write(ring, 0xffffffff);  /* CP_COHER_SIZE */
+	amdgpu_ring_write(ring, 0);  /* CP_COHER_BASE */
+	amdgpu_ring_write(ring, 0x0000000A); /* poll interval */
+}
+
+static void gfx_v8_0_emit_mem_sync_compute(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_ACQUIRE_MEM, 5));
+	amdgpu_ring_write(ring, PACKET3_TCL1_ACTION_ENA |
+			  PACKET3_TC_ACTION_ENA |
+			  PACKET3_SH_KCACHE_ACTION_ENA |
+			  PACKET3_SH_ICACHE_ACTION_ENA |
+			  PACKET3_TC_WB_ACTION_ENA);  /* CP_COHER_CNTL */
+	amdgpu_ring_write(ring, 0xffffffff);	/* CP_COHER_SIZE */
+	amdgpu_ring_write(ring, 0xff);		/* CP_COHER_SIZE_HI */
+	amdgpu_ring_write(ring, 0);		/* CP_COHER_BASE */
+	amdgpu_ring_write(ring, 0);		/* CP_COHER_BASE_HI */
+	amdgpu_ring_write(ring, 0x0000000A);	/* poll interval */
+}
+
+
+/* mmSPI_WCL_PIPE_PERCENT_CS[0-7]_DEFAULT values are same */
+#define mmSPI_WCL_PIPE_PERCENT_CS_DEFAULT	0x0000007f
+static void gfx_v8_0_emit_wave_limit_cs(struct amdgpu_ring *ring,
+					uint32_t pipe, bool enable)
+{
+	uint32_t val;
+	uint32_t wcl_cs_reg;
+
+	val = enable ? 0x1 : mmSPI_WCL_PIPE_PERCENT_CS_DEFAULT;
+
+	switch (pipe) {
+	case 0:
+		wcl_cs_reg = mmSPI_WCL_PIPE_PERCENT_CS0;
 		break;
-	default:
-		BUG(); /* kiq only support GENERIC2_INT now */
+	case 1:
+		wcl_cs_reg = mmSPI_WCL_PIPE_PERCENT_CS1;
+		break;
+	case 2:
+		wcl_cs_reg = mmSPI_WCL_PIPE_PERCENT_CS2;
+		break;
+	case 3:
+		wcl_cs_reg = mmSPI_WCL_PIPE_PERCENT_CS3;
 		break;
+	default:
+		DRM_DEBUG("invalid pipe %d\n", pipe);
+		return;
 	}
-	return 0;
+
+	amdgpu_ring_emit_wreg(ring, wcl_cs_reg, val);
+
 }
 
-static int gfx_v8_0_kiq_irq(struct amdgpu_device *adev,
-			    struct amdgpu_irq_src *source,
-			    struct amdgpu_iv_entry *entry)
+#define mmSPI_WCL_PIPE_PERCENT_GFX_DEFAULT	0x07ffffff
+static void gfx_v8_0_emit_wave_limit(struct amdgpu_ring *ring, bool enable)
 {
-	u8 me_id, pipe_id, queue_id;
-	struct amdgpu_ring *ring = &(adev->gfx.kiq.ring);
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t val;
+	int i;
 
-	me_id = (entry->ring_id & 0x0c) >> 2;
-	pipe_id = (entry->ring_id & 0x03) >> 0;
-	queue_id = (entry->ring_id & 0x70) >> 4;
-	DRM_DEBUG("IH: CPC GENERIC2_INT, me:%d, pipe:%d, queue:%d\n",
-		   me_id, pipe_id, queue_id);
+	/* mmSPI_WCL_PIPE_PERCENT_GFX is 7 bit multiplier register to limit
+	 * number of gfx waves. Setting 5 bit will make sure gfx only gets
+	 * around 25% of gpu resources.
+	 */
+	val = enable ? 0x1f : mmSPI_WCL_PIPE_PERCENT_GFX_DEFAULT;
+	amdgpu_ring_emit_wreg(ring, mmSPI_WCL_PIPE_PERCENT_GFX, val);
+
+	/* Restrict waves for normal/low priority compute queues as well
+	 * to get best QoS for high priority compute jobs.
+	 *
+	 * amdgpu controls only 1st ME(0-3 CS pipes).
+	 */
+	for (i = 0; i < adev->gfx.mec.num_pipe_per_mec; i++) {
+		if (i != ring->pipe)
+			gfx_v8_0_emit_wave_limit_cs(ring, i, enable);
+
+	}
 
-	amdgpu_fence_process(ring);
-	return 0;
 }
 
 static const struct amd_ip_funcs gfx_v8_0_ip_funcs = {
@@ -6692,8 +6847,8 @@ static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_gfx = {
 	.emit_frame_size = /* maximum 215dw if count 16 IBs in */
 		5 +  /* COND_EXEC */
 		7 +  /* PIPELINE_SYNC */
-		19 + /* VM_FLUSH */
-		8 +  /* FENCE for VM_FLUSH */
+		VI_FLUSH_GPU_TLB_NUM_WREG * 5 + 9 + /* VM_FLUSH */
+		12 +  /* FENCE for VM_FLUSH */
 		20 + /* GDS switch */
 		4 + /* double SWITCH_BUFFER,
 		       the first COND_EXEC jump to the place just
@@ -6705,8 +6860,9 @@ static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_gfx = {
 		31 + /*	DE_META */
 		3 + /* CNTX_CTRL */
 		5 + /* HDP_INVL */
-		8 + 8 + /* FENCE x2 */
-		2, /* SWITCH_BUFFER */
+		12 + 12 + /* FENCE x2 */
+		2 + /* SWITCH_BUFFER */
+		5, /* SURFACE_SYNC */
 	.emit_ib_size =	4, /* gfx_v8_0_ring_emit_ib_gfx */
 	.emit_ib = gfx_v8_0_ring_emit_ib_gfx,
 	.emit_fence = gfx_v8_0_ring_emit_fence_gfx,
@@ -6714,7 +6870,6 @@ static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_gfx = {
 	.emit_vm_flush = gfx_v8_0_ring_emit_vm_flush,
 	.emit_gds_switch = gfx_v8_0_ring_emit_gds_switch,
 	.emit_hdp_flush = gfx_v8_0_ring_emit_hdp_flush,
-	.emit_hdp_invalidate = gfx_v8_0_ring_emit_hdp_invalidate,
 	.test_ring = gfx_v8_0_ring_test_ring,
 	.test_ib = gfx_v8_0_ring_test_ib,
 	.insert_nop = amdgpu_ring_insert_nop,
@@ -6722,7 +6877,9 @@ static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_gfx = {
 	.emit_switch_buffer = gfx_v8_ring_emit_sb,
 	.emit_cntxcntl = gfx_v8_ring_emit_cntxcntl,
 	.init_cond_exec = gfx_v8_0_ring_emit_init_cond_exec,
-	.patch_cond_exec = gfx_v8_0_ring_emit_patch_cond_exec,
+	.emit_wreg = gfx_v8_0_ring_emit_wreg,
+	.soft_recovery = gfx_v8_0_ring_soft_recovery,
+	.emit_mem_sync = gfx_v8_0_emit_mem_sync,
 };
 
 static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_compute = {
@@ -6736,22 +6893,28 @@ static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_compute = {
 	.emit_frame_size =
 		20 + /* gfx_v8_0_ring_emit_gds_switch */
 		7 + /* gfx_v8_0_ring_emit_hdp_flush */
-		5 + /* gfx_v8_0_ring_emit_hdp_invalidate */
+		5 + /* hdp_invalidate */
 		7 + /* gfx_v8_0_ring_emit_pipeline_sync */
-		17 + /* gfx_v8_0_ring_emit_vm_flush */
-		7 + 7 + 7, /* gfx_v8_0_ring_emit_fence_compute x3 for user fence, vm fence */
-	.emit_ib_size =	4, /* gfx_v8_0_ring_emit_ib_compute */
+		VI_FLUSH_GPU_TLB_NUM_WREG * 5 + 7 + /* gfx_v8_0_ring_emit_vm_flush */
+		7 + 7 + 7 + /* gfx_v8_0_ring_emit_fence_compute x3 for user fence, vm fence */
+		7 + /* gfx_v8_0_emit_mem_sync_compute */
+		5 + /* gfx_v8_0_emit_wave_limit for updating mmSPI_WCL_PIPE_PERCENT_GFX register */
+		15, /* for updating 3 mmSPI_WCL_PIPE_PERCENT_CS registers */
+	.emit_ib_size =	7, /* gfx_v8_0_ring_emit_ib_compute */
 	.emit_ib = gfx_v8_0_ring_emit_ib_compute,
 	.emit_fence = gfx_v8_0_ring_emit_fence_compute,
 	.emit_pipeline_sync = gfx_v8_0_ring_emit_pipeline_sync,
 	.emit_vm_flush = gfx_v8_0_ring_emit_vm_flush,
 	.emit_gds_switch = gfx_v8_0_ring_emit_gds_switch,
 	.emit_hdp_flush = gfx_v8_0_ring_emit_hdp_flush,
-	.emit_hdp_invalidate = gfx_v8_0_ring_emit_hdp_invalidate,
 	.test_ring = gfx_v8_0_ring_test_ring,
 	.test_ib = gfx_v8_0_ring_test_ib,
 	.insert_nop = amdgpu_ring_insert_nop,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_wreg = gfx_v8_0_ring_emit_wreg,
+	.soft_recovery = gfx_v8_0_ring_soft_recovery,
+	.emit_mem_sync = gfx_v8_0_emit_mem_sync_compute,
+	.emit_wave_limit = gfx_v8_0_emit_wave_limit,
 };
 
 static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_kiq = {
@@ -6765,15 +6928,13 @@ static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_kiq = {
 	.emit_frame_size =
 		20 + /* gfx_v8_0_ring_emit_gds_switch */
 		7 + /* gfx_v8_0_ring_emit_hdp_flush */
-		5 + /* gfx_v8_0_ring_emit_hdp_invalidate */
+		5 + /* hdp_invalidate */
 		7 + /* gfx_v8_0_ring_emit_pipeline_sync */
 		17 + /* gfx_v8_0_ring_emit_vm_flush */
 		7 + 7 + 7, /* gfx_v8_0_ring_emit_fence_kiq x3 for user fence, vm fence */
-	.emit_ib_size =	4, /* gfx_v8_0_ring_emit_ib_compute */
-	.emit_ib = gfx_v8_0_ring_emit_ib_compute,
+	.emit_ib_size =	7, /* gfx_v8_0_ring_emit_ib_compute */
 	.emit_fence = gfx_v8_0_ring_emit_fence_kiq,
 	.test_ring = gfx_v8_0_ring_test_ring,
-	.test_ib = gfx_v8_0_ring_test_ib,
 	.insert_nop = amdgpu_ring_insert_nop,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
 	.emit_rreg = gfx_v8_0_ring_emit_rreg,
@@ -6784,7 +6945,7 @@ static void gfx_v8_0_set_ring_funcs(struct amdgpu_device *adev)
 {
 	int i;
 
-	adev->gfx.kiq.ring.funcs = &gfx_v8_0_ring_funcs_kiq;
+	adev->gfx.kiq[0].ring.funcs = &gfx_v8_0_ring_funcs_kiq;
 
 	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
 		adev->gfx.gfx_ring[i].funcs = &gfx_v8_0_ring_funcs_gfx;
@@ -6808,9 +6969,14 @@ static const struct amdgpu_irq_src_funcs gfx_v8_0_priv_inst_irq_funcs = {
 	.process = gfx_v8_0_priv_inst_irq,
 };
 
-static const struct amdgpu_irq_src_funcs gfx_v8_0_kiq_irq_funcs = {
-	.set = gfx_v8_0_kiq_set_interrupt_state,
-	.process = gfx_v8_0_kiq_irq,
+static const struct amdgpu_irq_src_funcs gfx_v8_0_cp_ecc_error_irq_funcs = {
+	.set = gfx_v8_0_set_cp_ecc_int_state,
+	.process = gfx_v8_0_cp_ecc_error_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v8_0_sq_irq_funcs = {
+	.set = gfx_v8_0_set_sq_int_state,
+	.process = gfx_v8_0_sq_irq,
 };
 
 static void gfx_v8_0_set_irq_funcs(struct amdgpu_device *adev)
@@ -6824,8 +6990,11 @@ static void gfx_v8_0_set_irq_funcs(struct amdgpu_device *adev)
 	adev->gfx.priv_inst_irq.num_types = 1;
 	adev->gfx.priv_inst_irq.funcs = &gfx_v8_0_priv_inst_irq_funcs;
 
-	adev->gfx.kiq.irq.num_types = AMDGPU_CP_KIQ_IRQ_LAST;
-	adev->gfx.kiq.irq.funcs = &gfx_v8_0_kiq_irq_funcs;
+	adev->gfx.cp_ecc_error_irq.num_types = 1;
+	adev->gfx.cp_ecc_error_irq.funcs = &gfx_v8_0_cp_ecc_error_irq_funcs;
+
+	adev->gfx.sq_irq.num_types = 1;
+	adev->gfx.sq_irq.funcs = &gfx_v8_0_sq_irq_funcs;
 }
 
 static void gfx_v8_0_set_rlc_funcs(struct amdgpu_device *adev)
@@ -6836,29 +7005,10 @@ static void gfx_v8_0_set_rlc_funcs(struct amdgpu_device *adev)
 static void gfx_v8_0_set_gds_init(struct amdgpu_device *adev)
 {
 	/* init asci gds info */
-	adev->gds.mem.total_size = RREG32(mmGDS_VMID0_SIZE);
-	adev->gds.gws.total_size = 64;
-	adev->gds.oa.total_size = 16;
-
-	if (adev->gds.mem.total_size == 64 * 1024) {
-		adev->gds.mem.gfx_partition_size = 4096;
-		adev->gds.mem.cs_partition_size = 4096;
-
-		adev->gds.gws.gfx_partition_size = 4;
-		adev->gds.gws.cs_partition_size = 4;
-
-		adev->gds.oa.gfx_partition_size = 4;
-		adev->gds.oa.cs_partition_size = 1;
-	} else {
-		adev->gds.mem.gfx_partition_size = 1024;
-		adev->gds.mem.cs_partition_size = 1024;
-
-		adev->gds.gws.gfx_partition_size = 16;
-		adev->gds.gws.cs_partition_size = 16;
-
-		adev->gds.oa.gfx_partition_size = 4;
-		adev->gds.oa.cs_partition_size = 4;
-	}
+	adev->gds.gds_size = RREG32(mmGDS_VMID0_SIZE);
+	adev->gds.gws_size = 64;
+	adev->gds.oa_size = 16;
+	adev->gds.gds_compute_max_wave_id = RREG32(mmGDS_COMPUTE_MAX_WAVE_ID);
 }
 
 static void gfx_v8_0_set_user_cu_inactive_bitmap(struct amdgpu_device *adev,
@@ -6910,12 +7060,12 @@ static void gfx_v8_0_get_cu_info(struct amdgpu_device *adev)
 			mask = 1;
 			ao_bitmap = 0;
 			counter = 0;
-			gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff);
+			gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff, 0);
 			if (i < 4 && j < 2)
 				gfx_v8_0_set_user_cu_inactive_bitmap(
 					adev, disable_masks[i * 2 + j]);
 			bitmap = gfx_v8_0_get_cu_active_bitmap(adev);
-			cu_info->bitmap[i][j] = bitmap;
+			cu_info->bitmap[0][i][j] = bitmap;
 
 			for (k = 0; k < adev->gfx.config.max_cu_per_sh; k ++) {
 				if (bitmap & mask) {
@@ -6931,11 +7081,16 @@ static void gfx_v8_0_get_cu_info(struct amdgpu_device *adev)
 			cu_info->ao_cu_bitmap[i][j] = ao_bitmap;
 		}
 	}
-	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 	mutex_unlock(&adev->grbm_idx_mutex);
 
 	cu_info->number = active_cu_number;
 	cu_info->ao_cu_mask = ao_cu_mask;
+	cu_info->simd_per_cu = NUM_SIMD_PER_CU;
+	cu_info->max_waves_per_simd = 10;
+	cu_info->max_scratch_slots_per_cu = 32;
+	cu_info->wave_front_size = 64;
+	cu_info->lds_size = 64;
 }
 
 const struct amdgpu_ip_block_version gfx_v8_0_ip_block =
@@ -6960,18 +7115,18 @@ static void gfx_v8_0_ring_emit_ce_meta(struct amdgpu_ring *ring)
 {
 	uint64_t ce_payload_addr;
 	int cnt_ce;
-	static union {
+	union {
 		struct vi_ce_ib_state regular;
 		struct vi_ce_ib_state_chained_ib chained;
 	} ce_payload = {};
 
 	if (ring->adev->virt.chained_ib_support) {
-		ce_payload_addr = AMDGPU_VA_RESERVED_SIZE - 2 * 4096 +
-						  offsetof(struct vi_gfx_meta_data_chained_ib, ce_payload);
+		ce_payload_addr = amdgpu_csa_vaddr(ring->adev) +
+			offsetof(struct vi_gfx_meta_data_chained_ib, ce_payload);
 		cnt_ce = (sizeof(ce_payload.chained) >> 2) + 4 - 2;
 	} else {
-		ce_payload_addr = AMDGPU_VA_RESERVED_SIZE - 2 * 4096 +
-						  offsetof(struct vi_gfx_meta_data, ce_payload);
+		ce_payload_addr = amdgpu_csa_vaddr(ring->adev) +
+			offsetof(struct vi_gfx_meta_data, ce_payload);
 		cnt_ce = (sizeof(ce_payload.regular) >> 2) + 4 - 2;
 	}
 
@@ -6989,12 +7144,12 @@ static void gfx_v8_0_ring_emit_de_meta(struct amdgpu_ring *ring)
 {
 	uint64_t de_payload_addr, gds_addr, csa_addr;
 	int cnt_de;
-	static union {
+	union {
 		struct vi_de_ib_state regular;
 		struct vi_de_ib_state_chained_ib chained;
 	} de_payload = {};
 
-	csa_addr = AMDGPU_VA_RESERVED_SIZE - 2 * 4096;
+	csa_addr = amdgpu_csa_vaddr(ring->adev);
 	gds_addr = csa_addr + 4096;
 	if (ring->adev->virt.chained_ib_support) {
 		de_payload.chained.gds_backup_addrlo = lower_32_bits(gds_addr);
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
index 69182eeca264..dd19a97436db 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
@@ -20,35 +20,51 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  *
  */
+
+#include <linux/delay.h>
+#include <linux/kernel.h>
 #include <linux/firmware.h>
-#include <drm/drmP.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+
 #include "amdgpu.h"
 #include "amdgpu_gfx.h"
 #include "soc15.h"
 #include "soc15d.h"
+#include "amdgpu_atomfirmware.h"
+#include "amdgpu_pm.h"
 
-#include "vega10/soc15ip.h"
-#include "vega10/GC/gc_9_0_offset.h"
-#include "vega10/GC/gc_9_0_sh_mask.h"
-#include "vega10/vega10_enum.h"
-#include "vega10/HDP/hdp_4_0_offset.h"
+#include "gc/gc_9_0_offset.h"
+#include "gc/gc_9_0_sh_mask.h"
+
+#include "vega10_enum.h"
 
 #include "soc15_common.h"
 #include "clearstate_gfx9.h"
 #include "v9_structs.h"
 
+#include "ivsrcid/gfx/irqsrcs_gfx_9_0.h"
+
+#include "amdgpu_ras.h"
+
+#include "amdgpu_ring_mux.h"
+#include "gfx_v9_4.h"
+#include "gfx_v9_0.h"
+#include "gfx_v9_0_cleaner_shader.h"
+#include "gfx_v9_4_2.h"
+
+#include "asic_reg/pwr/pwr_10_0_offset.h"
+#include "asic_reg/pwr/pwr_10_0_sh_mask.h"
+#include "asic_reg/gc/gc_9_0_default.h"
+
 #define GFX9_NUM_GFX_RINGS     1
-#define GFX9_MEC_HPD_SIZE 2048
+#define GFX9_NUM_SW_GFX_RINGS  2
+#define GFX9_MEC_HPD_SIZE 4096
 #define RLCG_UCODE_LOADING_START_ADDRESS 0x00002000L
 #define RLC_SAVE_RESTORE_ADDR_STARTING_OFFSET 0x00000000L
-#define GFX9_RLC_FORMAT_DIRECT_REG_LIST_LENGTH 34
 
-#define mmPWR_MISC_CNTL_STATUS					0x0183
-#define mmPWR_MISC_CNTL_STATUS_BASE_IDX				0
-#define PWR_MISC_CNTL_STATUS__PWR_GFX_RLC_CGPG_EN__SHIFT	0x0
-#define PWR_MISC_CNTL_STATUS__PWR_GFXOFF_STATUS__SHIFT		0x1
-#define PWR_MISC_CNTL_STATUS__PWR_GFX_RLC_CGPG_EN_MASK		0x00000001L
-#define PWR_MISC_CNTL_STATUS__PWR_GFXOFF_STATUS_MASK		0x00000006L
+#define mmGCEA_PROBE_MAP                        0x070c
+#define mmGCEA_PROBE_MAP_BASE_IDX               0
 
 MODULE_FIRMWARE("amdgpu/vega10_ce.bin");
 MODULE_FIRMWARE("amdgpu/vega10_pfp.bin");
@@ -57,6 +73,20 @@ MODULE_FIRMWARE("amdgpu/vega10_mec.bin");
 MODULE_FIRMWARE("amdgpu/vega10_mec2.bin");
 MODULE_FIRMWARE("amdgpu/vega10_rlc.bin");
 
+MODULE_FIRMWARE("amdgpu/vega12_ce.bin");
+MODULE_FIRMWARE("amdgpu/vega12_pfp.bin");
+MODULE_FIRMWARE("amdgpu/vega12_me.bin");
+MODULE_FIRMWARE("amdgpu/vega12_mec.bin");
+MODULE_FIRMWARE("amdgpu/vega12_mec2.bin");
+MODULE_FIRMWARE("amdgpu/vega12_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/vega20_ce.bin");
+MODULE_FIRMWARE("amdgpu/vega20_pfp.bin");
+MODULE_FIRMWARE("amdgpu/vega20_me.bin");
+MODULE_FIRMWARE("amdgpu/vega20_mec.bin");
+MODULE_FIRMWARE("amdgpu/vega20_mec2.bin");
+MODULE_FIRMWARE("amdgpu/vega20_rlc.bin");
+
 MODULE_FIRMWARE("amdgpu/raven_ce.bin");
 MODULE_FIRMWARE("amdgpu/raven_pfp.bin");
 MODULE_FIRMWARE("amdgpu/raven_me.bin");
@@ -64,158 +94,1067 @@ MODULE_FIRMWARE("amdgpu/raven_mec.bin");
 MODULE_FIRMWARE("amdgpu/raven_mec2.bin");
 MODULE_FIRMWARE("amdgpu/raven_rlc.bin");
 
-static const struct amdgpu_gds_reg_offset amdgpu_gds_reg_offset[] =
-{
-	{SOC15_REG_OFFSET(GC, 0, mmGDS_VMID0_BASE), SOC15_REG_OFFSET(GC, 0, mmGDS_VMID0_SIZE),
-		SOC15_REG_OFFSET(GC, 0, mmGDS_GWS_VMID0), SOC15_REG_OFFSET(GC, 0, mmGDS_OA_VMID0)},
-	{SOC15_REG_OFFSET(GC, 0, mmGDS_VMID1_BASE), SOC15_REG_OFFSET(GC, 0, mmGDS_VMID1_SIZE),
-		SOC15_REG_OFFSET(GC, 0, mmGDS_GWS_VMID1), SOC15_REG_OFFSET(GC, 0, mmGDS_OA_VMID1)},
-	{SOC15_REG_OFFSET(GC, 0, mmGDS_VMID2_BASE), SOC15_REG_OFFSET(GC, 0, mmGDS_VMID2_SIZE),
-		SOC15_REG_OFFSET(GC, 0, mmGDS_GWS_VMID2), SOC15_REG_OFFSET(GC, 0, mmGDS_OA_VMID2)},
-	{SOC15_REG_OFFSET(GC, 0, mmGDS_VMID3_BASE), SOC15_REG_OFFSET(GC, 0, mmGDS_VMID3_SIZE),
-		SOC15_REG_OFFSET(GC, 0, mmGDS_GWS_VMID3), SOC15_REG_OFFSET(GC, 0, mmGDS_OA_VMID3)},
-	{SOC15_REG_OFFSET(GC, 0, mmGDS_VMID4_BASE), SOC15_REG_OFFSET(GC, 0, mmGDS_VMID4_SIZE),
-		SOC15_REG_OFFSET(GC, 0, mmGDS_GWS_VMID4), SOC15_REG_OFFSET(GC, 0, mmGDS_OA_VMID4)},
-	{SOC15_REG_OFFSET(GC, 0, mmGDS_VMID5_BASE), SOC15_REG_OFFSET(GC, 0, mmGDS_VMID5_SIZE),
-		SOC15_REG_OFFSET(GC, 0, mmGDS_GWS_VMID5), SOC15_REG_OFFSET(GC, 0, mmGDS_OA_VMID5)},
-	{SOC15_REG_OFFSET(GC, 0, mmGDS_VMID6_BASE), SOC15_REG_OFFSET(GC, 0, mmGDS_VMID6_SIZE),
-		SOC15_REG_OFFSET(GC, 0, mmGDS_GWS_VMID6), SOC15_REG_OFFSET(GC, 0, mmGDS_OA_VMID6)},
-	{SOC15_REG_OFFSET(GC, 0, mmGDS_VMID7_BASE), SOC15_REG_OFFSET(GC, 0, mmGDS_VMID7_SIZE),
-		SOC15_REG_OFFSET(GC, 0, mmGDS_GWS_VMID7), SOC15_REG_OFFSET(GC, 0, mmGDS_OA_VMID7)},
-	{SOC15_REG_OFFSET(GC, 0, mmGDS_VMID8_BASE), SOC15_REG_OFFSET(GC, 0, mmGDS_VMID8_SIZE),
-		SOC15_REG_OFFSET(GC, 0, mmGDS_GWS_VMID8), SOC15_REG_OFFSET(GC, 0, mmGDS_OA_VMID8)},
-	{SOC15_REG_OFFSET(GC, 0, mmGDS_VMID9_BASE), SOC15_REG_OFFSET(GC, 0, mmGDS_VMID9_SIZE),
-		SOC15_REG_OFFSET(GC, 0, mmGDS_GWS_VMID9), SOC15_REG_OFFSET(GC, 0, mmGDS_OA_VMID9)},
-	{SOC15_REG_OFFSET(GC, 0, mmGDS_VMID10_BASE), SOC15_REG_OFFSET(GC, 0, mmGDS_VMID10_SIZE),
-		SOC15_REG_OFFSET(GC, 0, mmGDS_GWS_VMID10), SOC15_REG_OFFSET(GC, 0, mmGDS_OA_VMID10)},
-	{SOC15_REG_OFFSET(GC, 0, mmGDS_VMID11_BASE), SOC15_REG_OFFSET(GC, 0, mmGDS_VMID11_SIZE),
-	       	SOC15_REG_OFFSET(GC, 0, mmGDS_GWS_VMID11), SOC15_REG_OFFSET(GC, 0, mmGDS_OA_VMID11)},
-	{SOC15_REG_OFFSET(GC, 0, mmGDS_VMID12_BASE), SOC15_REG_OFFSET(GC, 0, mmGDS_VMID12_SIZE),
-		SOC15_REG_OFFSET(GC, 0, mmGDS_GWS_VMID12), SOC15_REG_OFFSET(GC, 0, mmGDS_OA_VMID12)},
-	{SOC15_REG_OFFSET(GC, 0, mmGDS_VMID13_BASE), SOC15_REG_OFFSET(GC, 0, mmGDS_VMID13_SIZE),
-		SOC15_REG_OFFSET(GC, 0, mmGDS_GWS_VMID13), SOC15_REG_OFFSET(GC, 0, mmGDS_OA_VMID13)},
-	{SOC15_REG_OFFSET(GC, 0, mmGDS_VMID14_BASE), SOC15_REG_OFFSET(GC, 0, mmGDS_VMID14_SIZE),
-		SOC15_REG_OFFSET(GC, 0, mmGDS_GWS_VMID14), SOC15_REG_OFFSET(GC, 0, mmGDS_OA_VMID14)},
-	{SOC15_REG_OFFSET(GC, 0, mmGDS_VMID15_BASE), SOC15_REG_OFFSET(GC, 0, mmGDS_VMID15_SIZE),
-		SOC15_REG_OFFSET(GC, 0, mmGDS_GWS_VMID15), SOC15_REG_OFFSET(GC, 0, mmGDS_OA_VMID15)}
+MODULE_FIRMWARE("amdgpu/picasso_ce.bin");
+MODULE_FIRMWARE("amdgpu/picasso_pfp.bin");
+MODULE_FIRMWARE("amdgpu/picasso_me.bin");
+MODULE_FIRMWARE("amdgpu/picasso_mec.bin");
+MODULE_FIRMWARE("amdgpu/picasso_mec2.bin");
+MODULE_FIRMWARE("amdgpu/picasso_rlc.bin");
+MODULE_FIRMWARE("amdgpu/picasso_rlc_am4.bin");
+
+MODULE_FIRMWARE("amdgpu/raven2_ce.bin");
+MODULE_FIRMWARE("amdgpu/raven2_pfp.bin");
+MODULE_FIRMWARE("amdgpu/raven2_me.bin");
+MODULE_FIRMWARE("amdgpu/raven2_mec.bin");
+MODULE_FIRMWARE("amdgpu/raven2_mec2.bin");
+MODULE_FIRMWARE("amdgpu/raven2_rlc.bin");
+MODULE_FIRMWARE("amdgpu/raven_kicker_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/arcturus_mec.bin");
+MODULE_FIRMWARE("amdgpu/arcturus_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/renoir_ce.bin");
+MODULE_FIRMWARE("amdgpu/renoir_pfp.bin");
+MODULE_FIRMWARE("amdgpu/renoir_me.bin");
+MODULE_FIRMWARE("amdgpu/renoir_mec.bin");
+MODULE_FIRMWARE("amdgpu/renoir_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/green_sardine_ce.bin");
+MODULE_FIRMWARE("amdgpu/green_sardine_pfp.bin");
+MODULE_FIRMWARE("amdgpu/green_sardine_me.bin");
+MODULE_FIRMWARE("amdgpu/green_sardine_mec.bin");
+MODULE_FIRMWARE("amdgpu/green_sardine_mec2.bin");
+MODULE_FIRMWARE("amdgpu/green_sardine_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/aldebaran_mec.bin");
+MODULE_FIRMWARE("amdgpu/aldebaran_mec2.bin");
+MODULE_FIRMWARE("amdgpu/aldebaran_rlc.bin");
+MODULE_FIRMWARE("amdgpu/aldebaran_sjt_mec.bin");
+MODULE_FIRMWARE("amdgpu/aldebaran_sjt_mec2.bin");
+
+#define mmTCP_CHAN_STEER_0_ARCT								0x0b03
+#define mmTCP_CHAN_STEER_0_ARCT_BASE_IDX							0
+#define mmTCP_CHAN_STEER_1_ARCT								0x0b04
+#define mmTCP_CHAN_STEER_1_ARCT_BASE_IDX							0
+#define mmTCP_CHAN_STEER_2_ARCT								0x0b09
+#define mmTCP_CHAN_STEER_2_ARCT_BASE_IDX							0
+#define mmTCP_CHAN_STEER_3_ARCT								0x0b0a
+#define mmTCP_CHAN_STEER_3_ARCT_BASE_IDX							0
+#define mmTCP_CHAN_STEER_4_ARCT								0x0b0b
+#define mmTCP_CHAN_STEER_4_ARCT_BASE_IDX							0
+#define mmTCP_CHAN_STEER_5_ARCT								0x0b0c
+#define mmTCP_CHAN_STEER_5_ARCT_BASE_IDX							0
+
+#define mmGOLDEN_TSC_COUNT_UPPER_Renoir                0x0025
+#define mmGOLDEN_TSC_COUNT_UPPER_Renoir_BASE_IDX       1
+#define mmGOLDEN_TSC_COUNT_LOWER_Renoir                0x0026
+#define mmGOLDEN_TSC_COUNT_LOWER_Renoir_BASE_IDX       1
+
+static const struct amdgpu_hwip_reg_entry gc_reg_list_9[] = {
+	SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS2),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_STALLED_STAT1),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_STALLED_STAT2),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CPC_STALLED_STAT1),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CPF_STALLED_STAT1),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_BUSY_STAT),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CPC_BUSY_STAT),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CPF_BUSY_STAT),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CPF_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_ERROR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB_BASE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB0_BASE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB0_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB0_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB1_BASE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB1_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB1_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB2_BASE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB2_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB2_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB1_CMD_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB2_CMD_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB1_CMD_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB2_CMD_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB1_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB1_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB1_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB2_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB2_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB2_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB1_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB1_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB1_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB2_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB2_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB2_BUFSZ),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCPF_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCPC_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCPG_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmGDS_PROTECTION_FAULT),
+	SOC15_REG_ENTRY_STR(GC, 0, mmGDS_VM_PROTECTION_FAULT),
+	SOC15_REG_ENTRY_STR(GC, 0, mmIA_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmIA_UTCL1_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmPA_CL_CNTL_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRMI_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSQC_DCACHE_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSQC_ICACHE_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSQ_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmTCP_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmWD_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmVM_L2_PROTECTION_FAULT_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmVM_L2_PROTECTION_FAULT_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_DEBUG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_INSTR_PNTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC1_INSTR_PNTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC2_INSTR_PNTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_INSTR_PNTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_INSTR_PNTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CPC_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_STAT),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SMU_COMMAND),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SMU_MESSAGE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SMU_ARGUMENT_1),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SMU_ARGUMENT_2),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSMU_RLC_RESPONSE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SAFE_MODE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SMU_SAFE_MODE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_INT_STAT),
+	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_GPM_GENERAL_6),
+	/* SE status registers */
+	SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS_SE0),
+	SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS_SE1),
+	SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS_SE2),
+	SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS_SE3),
+	/* packet headers */
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP)
+};
+
+static const struct amdgpu_hwip_reg_entry gc_cp_reg_list_9[] = {
+	/* compute queue registers */
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_VMID),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_ACTIVE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PERSISTENT_STATE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PIPE_PRIORITY),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_QUEUE_PRIORITY),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_QUANTUM),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PQ_BASE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PQ_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PQ_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PQ_WPTR_POLL_ADDR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PQ_WPTR_POLL_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PQ_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_IB_BASE_ADDR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_IB_BASE_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_IB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_IB_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_DEQUEUE_REQUEST),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_EOP_BASE_ADDR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_EOP_BASE_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_EOP_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_EOP_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_EOP_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_EOP_EVENTS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_CTX_SAVE_BASE_ADDR_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_CTX_SAVE_BASE_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_CTX_SAVE_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_CNTL_STACK_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_CNTL_STACK_SIZE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_WG_STATE_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_CTX_SAVE_SIZE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_GDS_RESOURCE_STATE),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_ERROR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_EOP_WPTR_MEM),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PQ_WPTR_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PQ_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_GFX_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP)
+};
+
+enum ta_ras_gfx_subblock {
+	/*CPC*/
+	TA_RAS_BLOCK__GFX_CPC_INDEX_START = 0,
+	TA_RAS_BLOCK__GFX_CPC_SCRATCH = TA_RAS_BLOCK__GFX_CPC_INDEX_START,
+	TA_RAS_BLOCK__GFX_CPC_UCODE,
+	TA_RAS_BLOCK__GFX_DC_STATE_ME1,
+	TA_RAS_BLOCK__GFX_DC_CSINVOC_ME1,
+	TA_RAS_BLOCK__GFX_DC_RESTORE_ME1,
+	TA_RAS_BLOCK__GFX_DC_STATE_ME2,
+	TA_RAS_BLOCK__GFX_DC_CSINVOC_ME2,
+	TA_RAS_BLOCK__GFX_DC_RESTORE_ME2,
+	TA_RAS_BLOCK__GFX_CPC_INDEX_END = TA_RAS_BLOCK__GFX_DC_RESTORE_ME2,
+	/* CPF*/
+	TA_RAS_BLOCK__GFX_CPF_INDEX_START,
+	TA_RAS_BLOCK__GFX_CPF_ROQ_ME2 = TA_RAS_BLOCK__GFX_CPF_INDEX_START,
+	TA_RAS_BLOCK__GFX_CPF_ROQ_ME1,
+	TA_RAS_BLOCK__GFX_CPF_TAG,
+	TA_RAS_BLOCK__GFX_CPF_INDEX_END = TA_RAS_BLOCK__GFX_CPF_TAG,
+	/* CPG*/
+	TA_RAS_BLOCK__GFX_CPG_INDEX_START,
+	TA_RAS_BLOCK__GFX_CPG_DMA_ROQ = TA_RAS_BLOCK__GFX_CPG_INDEX_START,
+	TA_RAS_BLOCK__GFX_CPG_DMA_TAG,
+	TA_RAS_BLOCK__GFX_CPG_TAG,
+	TA_RAS_BLOCK__GFX_CPG_INDEX_END = TA_RAS_BLOCK__GFX_CPG_TAG,
+	/* GDS*/
+	TA_RAS_BLOCK__GFX_GDS_INDEX_START,
+	TA_RAS_BLOCK__GFX_GDS_MEM = TA_RAS_BLOCK__GFX_GDS_INDEX_START,
+	TA_RAS_BLOCK__GFX_GDS_INPUT_QUEUE,
+	TA_RAS_BLOCK__GFX_GDS_OA_PHY_CMD_RAM_MEM,
+	TA_RAS_BLOCK__GFX_GDS_OA_PHY_DATA_RAM_MEM,
+	TA_RAS_BLOCK__GFX_GDS_OA_PIPE_MEM,
+	TA_RAS_BLOCK__GFX_GDS_INDEX_END = TA_RAS_BLOCK__GFX_GDS_OA_PIPE_MEM,
+	/* SPI*/
+	TA_RAS_BLOCK__GFX_SPI_SR_MEM,
+	/* SQ*/
+	TA_RAS_BLOCK__GFX_SQ_INDEX_START,
+	TA_RAS_BLOCK__GFX_SQ_SGPR = TA_RAS_BLOCK__GFX_SQ_INDEX_START,
+	TA_RAS_BLOCK__GFX_SQ_LDS_D,
+	TA_RAS_BLOCK__GFX_SQ_LDS_I,
+	TA_RAS_BLOCK__GFX_SQ_VGPR, /* VGPR = SP*/
+	TA_RAS_BLOCK__GFX_SQ_INDEX_END = TA_RAS_BLOCK__GFX_SQ_VGPR,
+	/* SQC (3 ranges)*/
+	TA_RAS_BLOCK__GFX_SQC_INDEX_START,
+	/* SQC range 0*/
+	TA_RAS_BLOCK__GFX_SQC_INDEX0_START = TA_RAS_BLOCK__GFX_SQC_INDEX_START,
+	TA_RAS_BLOCK__GFX_SQC_INST_UTCL1_LFIFO =
+		TA_RAS_BLOCK__GFX_SQC_INDEX0_START,
+	TA_RAS_BLOCK__GFX_SQC_DATA_CU0_WRITE_DATA_BUF,
+	TA_RAS_BLOCK__GFX_SQC_DATA_CU0_UTCL1_LFIFO,
+	TA_RAS_BLOCK__GFX_SQC_DATA_CU1_WRITE_DATA_BUF,
+	TA_RAS_BLOCK__GFX_SQC_DATA_CU1_UTCL1_LFIFO,
+	TA_RAS_BLOCK__GFX_SQC_DATA_CU2_WRITE_DATA_BUF,
+	TA_RAS_BLOCK__GFX_SQC_DATA_CU2_UTCL1_LFIFO,
+	TA_RAS_BLOCK__GFX_SQC_INDEX0_END =
+		TA_RAS_BLOCK__GFX_SQC_DATA_CU2_UTCL1_LFIFO,
+	/* SQC range 1*/
+	TA_RAS_BLOCK__GFX_SQC_INDEX1_START,
+	TA_RAS_BLOCK__GFX_SQC_INST_BANKA_TAG_RAM =
+		TA_RAS_BLOCK__GFX_SQC_INDEX1_START,
+	TA_RAS_BLOCK__GFX_SQC_INST_BANKA_UTCL1_MISS_FIFO,
+	TA_RAS_BLOCK__GFX_SQC_INST_BANKA_MISS_FIFO,
+	TA_RAS_BLOCK__GFX_SQC_INST_BANKA_BANK_RAM,
+	TA_RAS_BLOCK__GFX_SQC_DATA_BANKA_TAG_RAM,
+	TA_RAS_BLOCK__GFX_SQC_DATA_BANKA_HIT_FIFO,
+	TA_RAS_BLOCK__GFX_SQC_DATA_BANKA_MISS_FIFO,
+	TA_RAS_BLOCK__GFX_SQC_DATA_BANKA_DIRTY_BIT_RAM,
+	TA_RAS_BLOCK__GFX_SQC_DATA_BANKA_BANK_RAM,
+	TA_RAS_BLOCK__GFX_SQC_INDEX1_END =
+		TA_RAS_BLOCK__GFX_SQC_DATA_BANKA_BANK_RAM,
+	/* SQC range 2*/
+	TA_RAS_BLOCK__GFX_SQC_INDEX2_START,
+	TA_RAS_BLOCK__GFX_SQC_INST_BANKB_TAG_RAM =
+		TA_RAS_BLOCK__GFX_SQC_INDEX2_START,
+	TA_RAS_BLOCK__GFX_SQC_INST_BANKB_UTCL1_MISS_FIFO,
+	TA_RAS_BLOCK__GFX_SQC_INST_BANKB_MISS_FIFO,
+	TA_RAS_BLOCK__GFX_SQC_INST_BANKB_BANK_RAM,
+	TA_RAS_BLOCK__GFX_SQC_DATA_BANKB_TAG_RAM,
+	TA_RAS_BLOCK__GFX_SQC_DATA_BANKB_HIT_FIFO,
+	TA_RAS_BLOCK__GFX_SQC_DATA_BANKB_MISS_FIFO,
+	TA_RAS_BLOCK__GFX_SQC_DATA_BANKB_DIRTY_BIT_RAM,
+	TA_RAS_BLOCK__GFX_SQC_DATA_BANKB_BANK_RAM,
+	TA_RAS_BLOCK__GFX_SQC_INDEX2_END =
+		TA_RAS_BLOCK__GFX_SQC_DATA_BANKB_BANK_RAM,
+	TA_RAS_BLOCK__GFX_SQC_INDEX_END = TA_RAS_BLOCK__GFX_SQC_INDEX2_END,
+	/* TA*/
+	TA_RAS_BLOCK__GFX_TA_INDEX_START,
+	TA_RAS_BLOCK__GFX_TA_FS_DFIFO = TA_RAS_BLOCK__GFX_TA_INDEX_START,
+	TA_RAS_BLOCK__GFX_TA_FS_AFIFO,
+	TA_RAS_BLOCK__GFX_TA_FL_LFIFO,
+	TA_RAS_BLOCK__GFX_TA_FX_LFIFO,
+	TA_RAS_BLOCK__GFX_TA_FS_CFIFO,
+	TA_RAS_BLOCK__GFX_TA_INDEX_END = TA_RAS_BLOCK__GFX_TA_FS_CFIFO,
+	/* TCA*/
+	TA_RAS_BLOCK__GFX_TCA_INDEX_START,
+	TA_RAS_BLOCK__GFX_TCA_HOLE_FIFO = TA_RAS_BLOCK__GFX_TCA_INDEX_START,
+	TA_RAS_BLOCK__GFX_TCA_REQ_FIFO,
+	TA_RAS_BLOCK__GFX_TCA_INDEX_END = TA_RAS_BLOCK__GFX_TCA_REQ_FIFO,
+	/* TCC (5 sub-ranges)*/
+	TA_RAS_BLOCK__GFX_TCC_INDEX_START,
+	/* TCC range 0*/
+	TA_RAS_BLOCK__GFX_TCC_INDEX0_START = TA_RAS_BLOCK__GFX_TCC_INDEX_START,
+	TA_RAS_BLOCK__GFX_TCC_CACHE_DATA = TA_RAS_BLOCK__GFX_TCC_INDEX0_START,
+	TA_RAS_BLOCK__GFX_TCC_CACHE_DATA_BANK_0_1,
+	TA_RAS_BLOCK__GFX_TCC_CACHE_DATA_BANK_1_0,
+	TA_RAS_BLOCK__GFX_TCC_CACHE_DATA_BANK_1_1,
+	TA_RAS_BLOCK__GFX_TCC_CACHE_DIRTY_BANK_0,
+	TA_RAS_BLOCK__GFX_TCC_CACHE_DIRTY_BANK_1,
+	TA_RAS_BLOCK__GFX_TCC_HIGH_RATE_TAG,
+	TA_RAS_BLOCK__GFX_TCC_LOW_RATE_TAG,
+	TA_RAS_BLOCK__GFX_TCC_INDEX0_END = TA_RAS_BLOCK__GFX_TCC_LOW_RATE_TAG,
+	/* TCC range 1*/
+	TA_RAS_BLOCK__GFX_TCC_INDEX1_START,
+	TA_RAS_BLOCK__GFX_TCC_IN_USE_DEC = TA_RAS_BLOCK__GFX_TCC_INDEX1_START,
+	TA_RAS_BLOCK__GFX_TCC_IN_USE_TRANSFER,
+	TA_RAS_BLOCK__GFX_TCC_INDEX1_END =
+		TA_RAS_BLOCK__GFX_TCC_IN_USE_TRANSFER,
+	/* TCC range 2*/
+	TA_RAS_BLOCK__GFX_TCC_INDEX2_START,
+	TA_RAS_BLOCK__GFX_TCC_RETURN_DATA = TA_RAS_BLOCK__GFX_TCC_INDEX2_START,
+	TA_RAS_BLOCK__GFX_TCC_RETURN_CONTROL,
+	TA_RAS_BLOCK__GFX_TCC_UC_ATOMIC_FIFO,
+	TA_RAS_BLOCK__GFX_TCC_WRITE_RETURN,
+	TA_RAS_BLOCK__GFX_TCC_WRITE_CACHE_READ,
+	TA_RAS_BLOCK__GFX_TCC_SRC_FIFO,
+	TA_RAS_BLOCK__GFX_TCC_SRC_FIFO_NEXT_RAM,
+	TA_RAS_BLOCK__GFX_TCC_CACHE_TAG_PROBE_FIFO,
+	TA_RAS_BLOCK__GFX_TCC_INDEX2_END =
+		TA_RAS_BLOCK__GFX_TCC_CACHE_TAG_PROBE_FIFO,
+	/* TCC range 3*/
+	TA_RAS_BLOCK__GFX_TCC_INDEX3_START,
+	TA_RAS_BLOCK__GFX_TCC_LATENCY_FIFO = TA_RAS_BLOCK__GFX_TCC_INDEX3_START,
+	TA_RAS_BLOCK__GFX_TCC_LATENCY_FIFO_NEXT_RAM,
+	TA_RAS_BLOCK__GFX_TCC_INDEX3_END =
+		TA_RAS_BLOCK__GFX_TCC_LATENCY_FIFO_NEXT_RAM,
+	/* TCC range 4*/
+	TA_RAS_BLOCK__GFX_TCC_INDEX4_START,
+	TA_RAS_BLOCK__GFX_TCC_WRRET_TAG_WRITE_RETURN =
+		TA_RAS_BLOCK__GFX_TCC_INDEX4_START,
+	TA_RAS_BLOCK__GFX_TCC_ATOMIC_RETURN_BUFFER,
+	TA_RAS_BLOCK__GFX_TCC_INDEX4_END =
+		TA_RAS_BLOCK__GFX_TCC_ATOMIC_RETURN_BUFFER,
+	TA_RAS_BLOCK__GFX_TCC_INDEX_END = TA_RAS_BLOCK__GFX_TCC_INDEX4_END,
+	/* TCI*/
+	TA_RAS_BLOCK__GFX_TCI_WRITE_RAM,
+	/* TCP*/
+	TA_RAS_BLOCK__GFX_TCP_INDEX_START,
+	TA_RAS_BLOCK__GFX_TCP_CACHE_RAM = TA_RAS_BLOCK__GFX_TCP_INDEX_START,
+	TA_RAS_BLOCK__GFX_TCP_LFIFO_RAM,
+	TA_RAS_BLOCK__GFX_TCP_CMD_FIFO,
+	TA_RAS_BLOCK__GFX_TCP_VM_FIFO,
+	TA_RAS_BLOCK__GFX_TCP_DB_RAM,
+	TA_RAS_BLOCK__GFX_TCP_UTCL1_LFIFO0,
+	TA_RAS_BLOCK__GFX_TCP_UTCL1_LFIFO1,
+	TA_RAS_BLOCK__GFX_TCP_INDEX_END = TA_RAS_BLOCK__GFX_TCP_UTCL1_LFIFO1,
+	/* TD*/
+	TA_RAS_BLOCK__GFX_TD_INDEX_START,
+	TA_RAS_BLOCK__GFX_TD_SS_FIFO_LO = TA_RAS_BLOCK__GFX_TD_INDEX_START,
+	TA_RAS_BLOCK__GFX_TD_SS_FIFO_HI,
+	TA_RAS_BLOCK__GFX_TD_CS_FIFO,
+	TA_RAS_BLOCK__GFX_TD_INDEX_END = TA_RAS_BLOCK__GFX_TD_CS_FIFO,
+	/* EA (3 sub-ranges)*/
+	TA_RAS_BLOCK__GFX_EA_INDEX_START,
+	/* EA range 0*/
+	TA_RAS_BLOCK__GFX_EA_INDEX0_START = TA_RAS_BLOCK__GFX_EA_INDEX_START,
+	TA_RAS_BLOCK__GFX_EA_DRAMRD_CMDMEM = TA_RAS_BLOCK__GFX_EA_INDEX0_START,
+	TA_RAS_BLOCK__GFX_EA_DRAMWR_CMDMEM,
+	TA_RAS_BLOCK__GFX_EA_DRAMWR_DATAMEM,
+	TA_RAS_BLOCK__GFX_EA_RRET_TAGMEM,
+	TA_RAS_BLOCK__GFX_EA_WRET_TAGMEM,
+	TA_RAS_BLOCK__GFX_EA_GMIRD_CMDMEM,
+	TA_RAS_BLOCK__GFX_EA_GMIWR_CMDMEM,
+	TA_RAS_BLOCK__GFX_EA_GMIWR_DATAMEM,
+	TA_RAS_BLOCK__GFX_EA_INDEX0_END = TA_RAS_BLOCK__GFX_EA_GMIWR_DATAMEM,
+	/* EA range 1*/
+	TA_RAS_BLOCK__GFX_EA_INDEX1_START,
+	TA_RAS_BLOCK__GFX_EA_DRAMRD_PAGEMEM = TA_RAS_BLOCK__GFX_EA_INDEX1_START,
+	TA_RAS_BLOCK__GFX_EA_DRAMWR_PAGEMEM,
+	TA_RAS_BLOCK__GFX_EA_IORD_CMDMEM,
+	TA_RAS_BLOCK__GFX_EA_IOWR_CMDMEM,
+	TA_RAS_BLOCK__GFX_EA_IOWR_DATAMEM,
+	TA_RAS_BLOCK__GFX_EA_GMIRD_PAGEMEM,
+	TA_RAS_BLOCK__GFX_EA_GMIWR_PAGEMEM,
+	TA_RAS_BLOCK__GFX_EA_INDEX1_END = TA_RAS_BLOCK__GFX_EA_GMIWR_PAGEMEM,
+	/* EA range 2*/
+	TA_RAS_BLOCK__GFX_EA_INDEX2_START,
+	TA_RAS_BLOCK__GFX_EA_MAM_D0MEM = TA_RAS_BLOCK__GFX_EA_INDEX2_START,
+	TA_RAS_BLOCK__GFX_EA_MAM_D1MEM,
+	TA_RAS_BLOCK__GFX_EA_MAM_D2MEM,
+	TA_RAS_BLOCK__GFX_EA_MAM_D3MEM,
+	TA_RAS_BLOCK__GFX_EA_INDEX2_END = TA_RAS_BLOCK__GFX_EA_MAM_D3MEM,
+	TA_RAS_BLOCK__GFX_EA_INDEX_END = TA_RAS_BLOCK__GFX_EA_INDEX2_END,
+	/* UTC VM L2 bank*/
+	TA_RAS_BLOCK__UTC_VML2_BANK_CACHE,
+	/* UTC VM walker*/
+	TA_RAS_BLOCK__UTC_VML2_WALKER,
+	/* UTC ATC L2 2MB cache*/
+	TA_RAS_BLOCK__UTC_ATCL2_CACHE_2M_BANK,
+	/* UTC ATC L2 4KB cache*/
+	TA_RAS_BLOCK__UTC_ATCL2_CACHE_4K_BANK,
+	TA_RAS_BLOCK__GFX_MAX
+};
+
+struct ras_gfx_subblock {
+	unsigned char *name;
+	int ta_subblock;
+	int hw_supported_error_type;
+	int sw_supported_error_type;
+};
+
+#define AMDGPU_RAS_SUB_BLOCK(subblock, a, b, c, d, e, f, g, h)                             \
+	[AMDGPU_RAS_BLOCK__##subblock] = {                                     \
+		#subblock,                                                     \
+		TA_RAS_BLOCK__##subblock,                                      \
+		((a) | ((b) << 1) | ((c) << 2) | ((d) << 3)),                  \
+		(((e) << 1) | ((f) << 3) | (g) | ((h) << 2)),                  \
+	}
+
+static const struct ras_gfx_subblock ras_gfx_subblocks[] = {
+	AMDGPU_RAS_SUB_BLOCK(GFX_CPC_SCRATCH, 0, 1, 1, 1, 1, 0, 0, 1),
+	AMDGPU_RAS_SUB_BLOCK(GFX_CPC_UCODE, 0, 1, 1, 1, 1, 0, 0, 1),
+	AMDGPU_RAS_SUB_BLOCK(GFX_DC_STATE_ME1, 1, 0, 0, 1, 0, 0, 1, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_DC_CSINVOC_ME1, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_DC_RESTORE_ME1, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_DC_STATE_ME2, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_DC_CSINVOC_ME2, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_DC_RESTORE_ME2, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_CPF_ROQ_ME2, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_CPF_ROQ_ME1, 1, 0, 0, 1, 0, 0, 1, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_CPF_TAG, 0, 1, 1, 1, 1, 0, 0, 1),
+	AMDGPU_RAS_SUB_BLOCK(GFX_CPG_DMA_ROQ, 1, 0, 0, 1, 0, 0, 1, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_CPG_DMA_TAG, 0, 1, 1, 1, 0, 1, 0, 1),
+	AMDGPU_RAS_SUB_BLOCK(GFX_CPG_TAG, 0, 1, 1, 1, 1, 1, 0, 1),
+	AMDGPU_RAS_SUB_BLOCK(GFX_GDS_MEM, 0, 1, 1, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_GDS_INPUT_QUEUE, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_GDS_OA_PHY_CMD_RAM_MEM, 0, 1, 1, 1, 0, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_GDS_OA_PHY_DATA_RAM_MEM, 1, 0, 0, 1, 0, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_GDS_OA_PIPE_MEM, 0, 1, 1, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SPI_SR_MEM, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQ_SGPR, 0, 1, 1, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQ_LDS_D, 0, 1, 1, 1, 1, 0, 0, 1),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQ_LDS_I, 0, 1, 1, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQ_VGPR, 0, 1, 1, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_INST_UTCL1_LFIFO, 0, 1, 1, 1, 0, 0, 0, 1),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_DATA_CU0_WRITE_DATA_BUF, 0, 1, 1, 1, 0, 0,
+			     0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_DATA_CU0_UTCL1_LFIFO, 0, 1, 1, 1, 0, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_DATA_CU1_WRITE_DATA_BUF, 0, 1, 1, 1, 0, 0,
+			     0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_DATA_CU1_UTCL1_LFIFO, 0, 1, 1, 1, 1, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_DATA_CU2_WRITE_DATA_BUF, 0, 1, 1, 1, 0, 0,
+			     0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_DATA_CU2_UTCL1_LFIFO, 0, 1, 1, 1, 0, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_INST_BANKA_TAG_RAM, 0, 1, 1, 1, 1, 0, 0,
+			     1),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_INST_BANKA_UTCL1_MISS_FIFO, 1, 0, 0, 1, 0,
+			     0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_INST_BANKA_MISS_FIFO, 1, 0, 0, 1, 0, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_INST_BANKA_BANK_RAM, 0, 1, 1, 1, 0, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_DATA_BANKA_TAG_RAM, 0, 1, 1, 1, 0, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_DATA_BANKA_HIT_FIFO, 1, 0, 0, 1, 0, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_DATA_BANKA_MISS_FIFO, 1, 0, 0, 1, 0, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_DATA_BANKA_DIRTY_BIT_RAM, 1, 0, 0, 1, 0, 0,
+			     0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_DATA_BANKA_BANK_RAM, 0, 1, 1, 1, 0, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_INST_BANKB_TAG_RAM, 0, 1, 1, 1, 1, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_INST_BANKB_UTCL1_MISS_FIFO, 1, 0, 0, 1, 0,
+			     0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_INST_BANKB_MISS_FIFO, 1, 0, 0, 1, 0, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_INST_BANKB_BANK_RAM, 0, 1, 1, 1, 0, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_DATA_BANKB_TAG_RAM, 0, 1, 1, 1, 0, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_DATA_BANKB_HIT_FIFO, 1, 0, 0, 1, 0, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_DATA_BANKB_MISS_FIFO, 1, 0, 0, 1, 0, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_DATA_BANKB_DIRTY_BIT_RAM, 1, 0, 0, 1, 0, 0,
+			     0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_SQC_DATA_BANKB_BANK_RAM, 0, 1, 1, 1, 0, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TA_FS_DFIFO, 0, 1, 1, 1, 1, 0, 0, 1),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TA_FS_AFIFO, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TA_FL_LFIFO, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TA_FX_LFIFO, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TA_FS_CFIFO, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCA_HOLE_FIFO, 1, 0, 0, 1, 0, 1, 1, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCA_REQ_FIFO, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCC_CACHE_DATA, 0, 1, 1, 1, 1, 0, 0, 1),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCC_CACHE_DATA_BANK_0_1, 0, 1, 1, 1, 1, 0, 0,
+			     1),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCC_CACHE_DATA_BANK_1_0, 0, 1, 1, 1, 1, 0, 0,
+			     1),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCC_CACHE_DATA_BANK_1_1, 0, 1, 1, 1, 1, 0, 0,
+			     1),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCC_CACHE_DIRTY_BANK_0, 0, 1, 1, 1, 0, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCC_CACHE_DIRTY_BANK_1, 0, 1, 1, 1, 0, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCC_HIGH_RATE_TAG, 0, 1, 1, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCC_LOW_RATE_TAG, 0, 1, 1, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCC_IN_USE_DEC, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCC_IN_USE_TRANSFER, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCC_RETURN_DATA, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCC_RETURN_CONTROL, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCC_UC_ATOMIC_FIFO, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCC_WRITE_RETURN, 1, 0, 0, 1, 0, 1, 1, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCC_WRITE_CACHE_READ, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCC_SRC_FIFO, 0, 1, 1, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCC_SRC_FIFO_NEXT_RAM, 1, 0, 0, 1, 0, 0, 1, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCC_CACHE_TAG_PROBE_FIFO, 1, 0, 0, 1, 0, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCC_LATENCY_FIFO, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCC_LATENCY_FIFO_NEXT_RAM, 1, 0, 0, 1, 0, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCC_WRRET_TAG_WRITE_RETURN, 1, 0, 0, 1, 0, 0,
+			     0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCC_ATOMIC_RETURN_BUFFER, 1, 0, 0, 1, 0, 0, 0,
+			     0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCI_WRITE_RAM, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCP_CACHE_RAM, 0, 1, 1, 1, 1, 0, 0, 1),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCP_LFIFO_RAM, 0, 1, 1, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCP_CMD_FIFO, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCP_VM_FIFO, 0, 1, 1, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCP_DB_RAM, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCP_UTCL1_LFIFO0, 0, 1, 1, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TCP_UTCL1_LFIFO1, 0, 1, 1, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TD_SS_FIFO_LO, 0, 1, 1, 1, 1, 0, 0, 1),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TD_SS_FIFO_HI, 0, 1, 1, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_TD_CS_FIFO, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_EA_DRAMRD_CMDMEM, 0, 1, 1, 1, 1, 0, 0, 1),
+	AMDGPU_RAS_SUB_BLOCK(GFX_EA_DRAMWR_CMDMEM, 0, 1, 1, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_EA_DRAMWR_DATAMEM, 0, 1, 1, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_EA_RRET_TAGMEM, 0, 1, 1, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_EA_WRET_TAGMEM, 0, 1, 1, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_EA_GMIRD_CMDMEM, 0, 1, 1, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_EA_GMIWR_CMDMEM, 0, 1, 1, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_EA_GMIWR_DATAMEM, 0, 1, 1, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_EA_DRAMRD_PAGEMEM, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_EA_DRAMWR_PAGEMEM, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_EA_IORD_CMDMEM, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_EA_IOWR_CMDMEM, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_EA_IOWR_DATAMEM, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_EA_GMIRD_PAGEMEM, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_EA_GMIWR_PAGEMEM, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_EA_MAM_D0MEM, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_EA_MAM_D1MEM, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_EA_MAM_D2MEM, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(GFX_EA_MAM_D3MEM, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(UTC_VML2_BANK_CACHE, 0, 1, 1, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(UTC_VML2_WALKER, 0, 1, 1, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(UTC_ATCL2_CACHE_2M_BANK, 1, 0, 0, 1, 0, 0, 0, 0),
+	AMDGPU_RAS_SUB_BLOCK(UTC_ATCL2_CACHE_4K_BANK, 0, 1, 1, 1, 0, 0, 0, 0),
+};
+
+static const struct soc15_reg_golden golden_settings_gc_9_0[] =
+{
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG2, 0xf00fffff, 0x00000400),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG3, 0x80000000, 0x80000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_GPU_ID, 0x0000000f, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_EVENT_CNTL_3, 0x00000003, 0x82400024),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE, 0x3fffffff, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSH_MEM_CONFIG, 0x00001000, 0x00001000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_RESOURCE_RESERVE_CU_0, 0x0007ffff, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_RESOURCE_RESERVE_CU_1, 0x0007ffff, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_RESOURCE_RESERVE_EN_CU_0, 0x01ffffff, 0x00ffff87),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_RESOURCE_RESERVE_EN_CU_1, 0x01ffffff, 0x00ffff8f),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQC_CONFIG, 0x03000000, 0x020a2000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfffffeef, 0x010b0000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_HI, 0xffffffff, 0x4a2c0e68),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_LO, 0xffffffff, 0xb5d3f197),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_CACHE_INVALIDATION, 0x3fff3af3, 0x19200000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_GS_MAX_WAVE_ID, 0x00000fff, 0x000003ff),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCP_MEC1_F32_INT_DIS, 0x00000800, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCP_MEC2_F32_INT_DIS, 0x00000800, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCP_DEBUG, 0x00008000, 0x00008000)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_9_0_vg10[] =
+{
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL, 0x0000f000, 0x00012107),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL_3, 0x30000000, 0x10000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPC_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPF_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPG_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG, 0xffff77ff, 0x2a114042),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG_READ, 0xffff77ff, 0x2a114042),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmIA_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_1, 0x00008000, 0x00048000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_GPM_UTCL1_CNTL_0, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_GPM_UTCL1_CNTL_1, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_GPM_UTCL1_CNTL_2, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_PREWALKER_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRMI_UTCL1_CNTL2, 0x00030000, 0x00020000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_CONFIG_CNTL_1, 0x0000000f, 0x01000107),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTD_CNTL, 0x00001800, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmWD_UTCL1_CNTL, 0x08000000, 0x08000080)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_9_0_vg20[] =
+{
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_DCC_CONFIG, 0x0f000080, 0x04000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL_2, 0x0f000000, 0x0a000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL_3, 0x30000000, 0x10000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG, 0xf3e777ff, 0x22014042),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG_READ, 0xf3e777ff, 0x22014042),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG2, 0x00003e00, 0x00000400),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_1, 0xff840000, 0x04040000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRMI_UTCL1_CNTL2, 0x00030000, 0x00030000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_CONFIG_CNTL_1, 0xffff010f, 0x01000107),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0x000b0000, 0x000b0000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTD_CNTL, 0x01000000, 0x01000000)
 };
 
-static const u32 golden_settings_gc_9_0[] =
-{
-	SOC15_REG_OFFSET(GC, 0, mmCPC_UTCL1_CNTL), 0x08000000, 0x08000080,
-	SOC15_REG_OFFSET(GC, 0, mmCPF_UTCL1_CNTL), 0x08000000, 0x08000080,
-	SOC15_REG_OFFSET(GC, 0, mmCPG_UTCL1_CNTL), 0x08000000, 0x08000080,
-	SOC15_REG_OFFSET(GC, 0, mmDB_DEBUG2), 0xf00fffff, 0x00000420,
-	SOC15_REG_OFFSET(GC, 0, mmGB_GPU_ID), 0x0000000f, 0x00000000,
-	SOC15_REG_OFFSET(GC, 0, mmIA_UTCL1_CNTL), 0x08000000, 0x08000080,
-	SOC15_REG_OFFSET(GC, 0, mmPA_SC_BINNER_EVENT_CNTL_3), 0x00000003, 0x82400024,
-	SOC15_REG_OFFSET(GC, 0, mmPA_SC_ENHANCE), 0x3fffffff, 0x00000001,
-	SOC15_REG_OFFSET(GC, 0, mmPA_SC_LINE_STIPPLE_STATE), 0x0000ff0f, 0x00000000,
-	SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_UTCL1_CNTL_0), 0x08000000, 0x08000080,
-	SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_UTCL1_CNTL_1), 0x08000000, 0x08000080,
-	SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_UTCL1_CNTL_2), 0x08000000, 0x08000080,
-	SOC15_REG_OFFSET(GC, 0, mmRLC_PREWALKER_UTCL1_CNTL), 0x08000000, 0x08000080,
-	SOC15_REG_OFFSET(GC, 0, mmRLC_SPM_UTCL1_CNTL), 0x08000000, 0x08000080,
-	SOC15_REG_OFFSET(GC, 0, mmSH_MEM_CONFIG), 0x00001000, 0x00001000,
-	SOC15_REG_OFFSET(GC, 0, mmSPI_CONFIG_CNTL_1), 0x0000000f, 0x01000107,
-	SOC15_REG_OFFSET(GC, 0, mmSQC_CONFIG), 0x03000000, 0x020a2000,
-	SOC15_REG_OFFSET(GC, 0, mmTA_CNTL_AUX), 0xfffffeef, 0x010b0000,
-	SOC15_REG_OFFSET(GC, 0, mmTCP_CHAN_STEER_HI), 0xffffffff, 0x4a2c0e68,
-	SOC15_REG_OFFSET(GC, 0, mmTCP_CHAN_STEER_LO), 0xffffffff, 0xb5d3f197,
-	SOC15_REG_OFFSET(GC, 0, mmVGT_CACHE_INVALIDATION), 0x3fff3af3, 0x19200000,
-	SOC15_REG_OFFSET(GC, 0, mmVGT_GS_MAX_WAVE_ID), 0x00000fff, 0x000003ff,
-	SOC15_REG_OFFSET(GC, 0, mmWD_UTCL1_CNTL), 0x08000000, 0x08000080
+static const struct soc15_reg_golden golden_settings_gc_9_1[] =
+{
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL, 0xfffdf3cf, 0x00014104),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPC_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPF_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPG_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG2, 0xf00fffff, 0x00000420),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_GPU_ID, 0x0000000f, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmIA_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_EVENT_CNTL_3, 0x00000003, 0x82400024),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE, 0x3fffffff, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_GPM_UTCL1_CNTL_0, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_GPM_UTCL1_CNTL_1, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_GPM_UTCL1_CNTL_2, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_PREWALKER_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfffffeef, 0x010b0000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_HI, 0xffffffff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_LO, 0xffffffff, 0x00003120),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_CACHE_INVALIDATION, 0x3fff3af3, 0x19200000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_GS_MAX_WAVE_ID, 0x00000fff, 0x000000ff),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmWD_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCP_MEC1_F32_INT_DIS, 0x00000800, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCP_MEC2_F32_INT_DIS, 0x00000800, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCP_DEBUG, 0x00008000, 0x00008000)
 };
 
-static const u32 golden_settings_gc_9_0_vg10[] =
+static const struct soc15_reg_golden golden_settings_gc_9_1_rv1[] =
 {
-	SOC15_REG_OFFSET(GC, 0, mmCB_HW_CONTROL), 0x0000f000, 0x00012107,
-	SOC15_REG_OFFSET(GC, 0, mmCB_HW_CONTROL_3), 0x30000000, 0x10000000,
-	SOC15_REG_OFFSET(GC, 0, mmGB_ADDR_CONFIG), 0xffff77ff, 0x2a114042,
-	SOC15_REG_OFFSET(GC, 0, mmGB_ADDR_CONFIG_READ), 0xffff77ff, 0x2a114042,
-	SOC15_REG_OFFSET(GC, 0, mmPA_SC_ENHANCE_1), 0x00008000, 0x00048000,
-	SOC15_REG_OFFSET(GC, 0, mmRMI_UTCL1_CNTL2), 0x00030000, 0x00020000,
-	SOC15_REG_OFFSET(GC, 0, mmTD_CNTL), 0x00001800, 0x00000800
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL_3, 0x30000000, 0x10000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG, 0xffff77ff, 0x24000042),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG_READ, 0xffff77ff, 0x24000042),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_1, 0xffffffff, 0x04048000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_MODE_CNTL_1, 0x06000000, 0x06000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRMI_UTCL1_CNTL2, 0x00030000, 0x00020000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTD_CNTL, 0x01bd9f33, 0x00000800)
 };
 
-static const u32 golden_settings_gc_9_1[] =
-{
-	SOC15_REG_OFFSET(GC, 0, mmCB_HW_CONTROL), 0xfffdf3cf, 0x00014104,
-	SOC15_REG_OFFSET(GC, 0, mmCPC_UTCL1_CNTL), 0x08000000, 0x08000080,
-	SOC15_REG_OFFSET(GC, 0, mmCPF_UTCL1_CNTL), 0x08000000, 0x08000080,
-	SOC15_REG_OFFSET(GC, 0, mmCPG_UTCL1_CNTL), 0x08000000, 0x08000080,
-	SOC15_REG_OFFSET(GC, 0, mmDB_DEBUG2), 0xf00fffff, 0x00000420,
-	SOC15_REG_OFFSET(GC, 0, mmGB_GPU_ID), 0x0000000f, 0x00000000,
-	SOC15_REG_OFFSET(GC, 0, mmIA_UTCL1_CNTL), 0x08000000, 0x08000080,
-	SOC15_REG_OFFSET(GC, 0, mmPA_SC_BINNER_EVENT_CNTL_3), 0x00000003, 0x82400024,
-	SOC15_REG_OFFSET(GC, 0, mmPA_SC_ENHANCE), 0x3fffffff, 0x00000001,
-	SOC15_REG_OFFSET(GC, 0, mmPA_SC_LINE_STIPPLE_STATE), 0x0000ff0f, 0x00000000,
-	SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_UTCL1_CNTL_0), 0x08000000, 0x08000080,
-	SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_UTCL1_CNTL_1), 0x08000000, 0x08000080,
-	SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_UTCL1_CNTL_2), 0x08000000, 0x08000080,
-	SOC15_REG_OFFSET(GC, 0, mmRLC_PREWALKER_UTCL1_CNTL), 0x08000000, 0x08000080,
-	SOC15_REG_OFFSET(GC, 0, mmRLC_SPM_UTCL1_CNTL), 0x08000000, 0x08000080,
-	SOC15_REG_OFFSET(GC, 0, mmTA_CNTL_AUX), 0xfffffeef, 0x010b0000,
-	SOC15_REG_OFFSET(GC, 0, mmTCP_CHAN_STEER_HI), 0xffffffff, 0x00000000,
-	SOC15_REG_OFFSET(GC, 0, mmTCP_CHAN_STEER_LO), 0xffffffff, 0x00003120,
-	SOC15_REG_OFFSET(GC, 0, mmVGT_CACHE_INVALIDATION), 0x3fff3af3, 0x19200000,
-	SOC15_REG_OFFSET(GC, 0, mmVGT_GS_MAX_WAVE_ID), 0x00000fff, 0x000000ff,
-	SOC15_REG_OFFSET(GC, 0, mmWD_UTCL1_CNTL), 0x08000000, 0x08000080
+static const struct soc15_reg_golden golden_settings_gc_9_1_rv2[] =
+{
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_DCC_CONFIG, 0xff7fffff, 0x04000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL, 0xfffdf3cf, 0x00014104),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL_2, 0xff7fffff, 0x0a000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPC_UTCL1_CNTL, 0x7f0fffff, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPF_UTCL1_CNTL, 0xff8fffff, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPG_UTCL1_CNTL, 0x7f8fffff, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG, 0xffff77ff, 0x26013041),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG_READ, 0xffff77ff, 0x26013041),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmIA_UTCL1_CNTL, 0x3f8fffff, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_1, 0xffffffff, 0x04040000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_GPM_UTCL1_CNTL_0, 0xff0fffff, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_GPM_UTCL1_CNTL_1, 0xff0fffff, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_GPM_UTCL1_CNTL_2, 0xff0fffff, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_PREWALKER_UTCL1_CNTL, 0xff0fffff, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_UTCL1_CNTL, 0xff0fffff, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_HI, 0xffffffff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_LO, 0xffffffff, 0x00000010),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTD_CNTL, 0x01bd9f33, 0x01000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmWD_UTCL1_CNTL, 0x3f8fffff, 0x08000080),
 };
 
-static const u32 golden_settings_gc_9_1_rv1[] =
+static const struct soc15_reg_golden golden_settings_gc_9_1_rn[] =
+{
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL, 0xfffdf3cf, 0x00014104),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL_2, 0xff7fffff, 0x0a000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG2, 0xf00fffff, 0x00000400),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG, 0xf3e777ff, 0x24000042),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG_READ, 0xf3e777ff, 0x24000042),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE, 0x3fffffff, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_1, 0xffffffff, 0x04040000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfffffeef, 0x010b0000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_HI, 0xffffffff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_LO, 0xffffffff, 0x00003120),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCEA_PROBE_MAP, 0xffffffff, 0x0000cccc),
+};
+
+static const struct soc15_reg_golden golden_settings_gc_9_x_common[] =
+{
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCP_SD_CNTL, 0xffffffff, 0x000001ff),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_CAM_INDEX, 0xffffffff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_CAM_DATA, 0xffffffff, 0x2544c382)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_9_2_1[] =
+{
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG2, 0xf00fffff, 0x00000420),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_GPU_ID, 0x0000000f, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_EVENT_CNTL_3, 0x00000003, 0x82400024),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE, 0x3fffffff, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSH_MEM_CONFIG, 0x00001000, 0x00001000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_RESOURCE_RESERVE_CU_0, 0x0007ffff, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_RESOURCE_RESERVE_CU_1, 0x0007ffff, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_RESOURCE_RESERVE_EN_CU_0, 0x01ffffff, 0x0000ff87),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_RESOURCE_RESERVE_EN_CU_1, 0x01ffffff, 0x0000ff8f),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQC_CONFIG, 0x03000000, 0x020a2000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfffffeef, 0x010b0000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_HI, 0xffffffff, 0x4a2c0e68),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_LO, 0xffffffff, 0xb5d3f197),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_CACHE_INVALIDATION, 0x3fff3af3, 0x19200000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_GS_MAX_WAVE_ID, 0x00000fff, 0x000003ff)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_9_2_1_vg12[] =
+{
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_DCC_CONFIG, 0x00000080, 0x04000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL, 0xfffdf3cf, 0x00014104),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL_2, 0x0f000000, 0x0a000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG, 0xffff77ff, 0x24104041),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG_READ, 0xffff77ff, 0x24104041),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_1, 0xffffffff, 0x04040000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_CONFIG_CNTL_1, 0xffff03ff, 0x01000107),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_HI, 0xffffffff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_LO, 0xffffffff, 0x76325410),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTD_CNTL, 0x01bd9f33, 0x01000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCP_MEC1_F32_INT_DIS, 0x00000800, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCP_MEC2_F32_INT_DIS, 0x00000800, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCP_DEBUG, 0x00008000, 0x00008000)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_9_4_1_arct[] =
+{
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG, 0xffff77ff, 0x2a114042),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfffffeef, 0x10b0000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_0_ARCT, 0x3fffffff, 0x346f0a4e),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_1_ARCT, 0x3fffffff, 0x1c642ca),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_2_ARCT, 0x3fffffff, 0x26f45098),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_3_ARCT, 0x3fffffff, 0x2ebd9fe3),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_4_ARCT, 0x3fffffff, 0xb90f5b1),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_5_ARCT, 0x3ff, 0x135),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_CONFIG, 0xffffffff, 0x011A0000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_FIFO_SIZES, 0xffffffff, 0x00000f00),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_UTCL1_CNTL1, 0x30000000, 0x30000000)
+};
+
+static const struct soc15_reg_rlcg rlcg_access_gc_9_0[] = {
+	{SOC15_REG_ENTRY(GC, 0, mmGRBM_GFX_INDEX)},
+	{SOC15_REG_ENTRY(GC, 0, mmSQ_IND_INDEX)},
+};
+
+static const u32 GFX_RLC_SRM_INDEX_CNTL_ADDR_OFFSETS[] =
+{
+	mmRLC_SRM_INDEX_CNTL_ADDR_0 - mmRLC_SRM_INDEX_CNTL_ADDR_0,
+	mmRLC_SRM_INDEX_CNTL_ADDR_1 - mmRLC_SRM_INDEX_CNTL_ADDR_0,
+	mmRLC_SRM_INDEX_CNTL_ADDR_2 - mmRLC_SRM_INDEX_CNTL_ADDR_0,
+	mmRLC_SRM_INDEX_CNTL_ADDR_3 - mmRLC_SRM_INDEX_CNTL_ADDR_0,
+	mmRLC_SRM_INDEX_CNTL_ADDR_4 - mmRLC_SRM_INDEX_CNTL_ADDR_0,
+	mmRLC_SRM_INDEX_CNTL_ADDR_5 - mmRLC_SRM_INDEX_CNTL_ADDR_0,
+	mmRLC_SRM_INDEX_CNTL_ADDR_6 - mmRLC_SRM_INDEX_CNTL_ADDR_0,
+	mmRLC_SRM_INDEX_CNTL_ADDR_7 - mmRLC_SRM_INDEX_CNTL_ADDR_0,
+};
+
+static const u32 GFX_RLC_SRM_INDEX_CNTL_DATA_OFFSETS[] =
 {
-	SOC15_REG_OFFSET(GC, 0, mmCB_HW_CONTROL_3), 0x30000000, 0x10000000,
-	SOC15_REG_OFFSET(GC, 0, mmGB_ADDR_CONFIG), 0xffff77ff, 0x24000042,
-	SOC15_REG_OFFSET(GC, 0, mmGB_ADDR_CONFIG_READ), 0xffff77ff, 0x24000042,
-	SOC15_REG_OFFSET(GC, 0, mmPA_SC_ENHANCE_1), 0xffffffff, 0x04048000,
-	SOC15_REG_OFFSET(GC, 0, mmPA_SC_MODE_CNTL_1), 0x06000000, 0x06000000,
-	SOC15_REG_OFFSET(GC, 0, mmRMI_UTCL1_CNTL2), 0x00030000, 0x00020000,
-	SOC15_REG_OFFSET(GC, 0, mmTD_CNTL), 0x01bd9f33, 0x00000800
+	mmRLC_SRM_INDEX_CNTL_DATA_0 - mmRLC_SRM_INDEX_CNTL_DATA_0,
+	mmRLC_SRM_INDEX_CNTL_DATA_1 - mmRLC_SRM_INDEX_CNTL_DATA_0,
+	mmRLC_SRM_INDEX_CNTL_DATA_2 - mmRLC_SRM_INDEX_CNTL_DATA_0,
+	mmRLC_SRM_INDEX_CNTL_DATA_3 - mmRLC_SRM_INDEX_CNTL_DATA_0,
+	mmRLC_SRM_INDEX_CNTL_DATA_4 - mmRLC_SRM_INDEX_CNTL_DATA_0,
+	mmRLC_SRM_INDEX_CNTL_DATA_5 - mmRLC_SRM_INDEX_CNTL_DATA_0,
+	mmRLC_SRM_INDEX_CNTL_DATA_6 - mmRLC_SRM_INDEX_CNTL_DATA_0,
+	mmRLC_SRM_INDEX_CNTL_DATA_7 - mmRLC_SRM_INDEX_CNTL_DATA_0,
 };
 
 #define VEGA10_GB_ADDR_CONFIG_GOLDEN 0x2a114042
+#define VEGA12_GB_ADDR_CONFIG_GOLDEN 0x24104041
 #define RAVEN_GB_ADDR_CONFIG_GOLDEN 0x24000042
+#define RAVEN2_GB_ADDR_CONFIG_GOLDEN 0x26013041
 
 static void gfx_v9_0_set_ring_funcs(struct amdgpu_device *adev);
 static void gfx_v9_0_set_irq_funcs(struct amdgpu_device *adev);
 static void gfx_v9_0_set_gds_init(struct amdgpu_device *adev);
 static void gfx_v9_0_set_rlc_funcs(struct amdgpu_device *adev);
 static int gfx_v9_0_get_cu_info(struct amdgpu_device *adev,
-                                 struct amdgpu_cu_info *cu_info);
+				struct amdgpu_cu_info *cu_info);
 static uint64_t gfx_v9_0_get_gpu_clock_counter(struct amdgpu_device *adev);
-static void gfx_v9_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num, u32 instance);
-static void gfx_v9_0_ring_emit_de_meta(struct amdgpu_ring *ring);
+static void gfx_v9_0_ring_emit_de_meta(struct amdgpu_ring *ring, bool resume, bool usegds);
+static u64 gfx_v9_0_ring_get_rptr_compute(struct amdgpu_ring *ring);
+static void gfx_v9_0_query_ras_error_count(struct amdgpu_device *adev,
+					  void *ras_error_status);
+static int gfx_v9_0_ras_error_inject(struct amdgpu_device *adev,
+				     void *inject_if, uint32_t instance_mask);
+static void gfx_v9_0_reset_ras_error_count(struct amdgpu_device *adev);
+static void gfx_v9_0_update_spm_vmid_internal(struct amdgpu_device *adev,
+					      unsigned int vmid);
+static void gfx_v9_0_set_safe_mode(struct amdgpu_device *adev, int xcc_id);
+static void gfx_v9_0_unset_safe_mode(struct amdgpu_device *adev, int xcc_id);
+
+static void gfx_v9_0_kiq_set_resources(struct amdgpu_ring *kiq_ring,
+				uint64_t queue_mask)
+{
+	struct amdgpu_device *adev = kiq_ring->adev;
+	u64 shader_mc_addr;
+
+	/* Cleaner shader MC address */
+	shader_mc_addr = adev->gfx.cleaner_shader_gpu_addr >> 8;
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_SET_RESOURCES, 6));
+	amdgpu_ring_write(kiq_ring,
+		PACKET3_SET_RESOURCES_VMID_MASK(0) |
+		/* vmid_mask:0* queue_type:0 (KIQ) */
+		PACKET3_SET_RESOURCES_QUEUE_TYPE(0));
+	amdgpu_ring_write(kiq_ring,
+			lower_32_bits(queue_mask));	/* queue mask lo */
+	amdgpu_ring_write(kiq_ring,
+			upper_32_bits(queue_mask));	/* queue mask hi */
+	amdgpu_ring_write(kiq_ring, lower_32_bits(shader_mc_addr)); /* cleaner shader addr lo */
+	amdgpu_ring_write(kiq_ring, upper_32_bits(shader_mc_addr)); /* cleaner shader addr hi */
+	amdgpu_ring_write(kiq_ring, 0);	/* oac mask */
+	amdgpu_ring_write(kiq_ring, 0);	/* gds heap base:0, gds heap size:0 */
+}
+
+static void gfx_v9_0_kiq_map_queues(struct amdgpu_ring *kiq_ring,
+				 struct amdgpu_ring *ring)
+{
+	uint64_t mqd_addr = amdgpu_bo_gpu_offset(ring->mqd_obj);
+	uint64_t wptr_addr = ring->wptr_gpu_addr;
+	uint32_t eng_sel = ring->funcs->type == AMDGPU_RING_TYPE_GFX ? 4 : 0;
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_MAP_QUEUES, 5));
+	/* Q_sel:0, vmid:0, vidmem: 1, engine:0, num_Q:1*/
+	amdgpu_ring_write(kiq_ring, /* Q_sel: 0, vmid: 0, engine: 0, num_Q: 1 */
+			 PACKET3_MAP_QUEUES_QUEUE_SEL(0) | /* Queue_Sel */
+			 PACKET3_MAP_QUEUES_VMID(0) | /* VMID */
+			 PACKET3_MAP_QUEUES_QUEUE(ring->queue) |
+			 PACKET3_MAP_QUEUES_PIPE(ring->pipe) |
+			 PACKET3_MAP_QUEUES_ME((ring->me == 1 ? 0 : 1)) |
+			 /*queue_type: normal compute queue */
+			 PACKET3_MAP_QUEUES_QUEUE_TYPE(0) |
+			 /* alloc format: all_on_one_pipe */
+			 PACKET3_MAP_QUEUES_ALLOC_FORMAT(0) |
+			 PACKET3_MAP_QUEUES_ENGINE_SEL(eng_sel) |
+			 /* num_queues: must be 1 */
+			 PACKET3_MAP_QUEUES_NUM_QUEUES(1));
+	amdgpu_ring_write(kiq_ring,
+			PACKET3_MAP_QUEUES_DOORBELL_OFFSET(ring->doorbell_index));
+	amdgpu_ring_write(kiq_ring, lower_32_bits(mqd_addr));
+	amdgpu_ring_write(kiq_ring, upper_32_bits(mqd_addr));
+	amdgpu_ring_write(kiq_ring, lower_32_bits(wptr_addr));
+	amdgpu_ring_write(kiq_ring, upper_32_bits(wptr_addr));
+}
+
+static void gfx_v9_0_kiq_unmap_queues(struct amdgpu_ring *kiq_ring,
+				   struct amdgpu_ring *ring,
+				   enum amdgpu_unmap_queues_action action,
+				   u64 gpu_addr, u64 seq)
+{
+	uint32_t eng_sel = ring->funcs->type == AMDGPU_RING_TYPE_GFX ? 4 : 0;
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_UNMAP_QUEUES, 4));
+	amdgpu_ring_write(kiq_ring, /* Q_sel: 0, vmid: 0, engine: 0, num_Q: 1 */
+			  PACKET3_UNMAP_QUEUES_ACTION(action) |
+			  PACKET3_UNMAP_QUEUES_QUEUE_SEL(0) |
+			  PACKET3_UNMAP_QUEUES_ENGINE_SEL(eng_sel) |
+			  PACKET3_UNMAP_QUEUES_NUM_QUEUES(1));
+	amdgpu_ring_write(kiq_ring,
+			PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET0(ring->doorbell_index));
+
+	if (action == PREEMPT_QUEUES_NO_UNMAP) {
+		amdgpu_ring_write(kiq_ring, lower_32_bits(ring->wptr & ring->buf_mask));
+		amdgpu_ring_write(kiq_ring, 0);
+		amdgpu_ring_write(kiq_ring, 0);
+
+	} else {
+		amdgpu_ring_write(kiq_ring, 0);
+		amdgpu_ring_write(kiq_ring, 0);
+		amdgpu_ring_write(kiq_ring, 0);
+	}
+}
+
+static void gfx_v9_0_kiq_query_status(struct amdgpu_ring *kiq_ring,
+				   struct amdgpu_ring *ring,
+				   u64 addr,
+				   u64 seq)
+{
+	uint32_t eng_sel = ring->funcs->type == AMDGPU_RING_TYPE_GFX ? 4 : 0;
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_QUERY_STATUS, 5));
+	amdgpu_ring_write(kiq_ring,
+			  PACKET3_QUERY_STATUS_CONTEXT_ID(0) |
+			  PACKET3_QUERY_STATUS_INTERRUPT_SEL(0) |
+			  PACKET3_QUERY_STATUS_COMMAND(2));
+	/* Q_sel: 0, vmid: 0, engine: 0, num_Q: 1 */
+	amdgpu_ring_write(kiq_ring,
+			PACKET3_QUERY_STATUS_DOORBELL_OFFSET(ring->doorbell_index) |
+			PACKET3_QUERY_STATUS_ENG_SEL(eng_sel));
+	amdgpu_ring_write(kiq_ring, lower_32_bits(addr));
+	amdgpu_ring_write(kiq_ring, upper_32_bits(addr));
+	amdgpu_ring_write(kiq_ring, lower_32_bits(seq));
+	amdgpu_ring_write(kiq_ring, upper_32_bits(seq));
+}
+
+static void gfx_v9_0_kiq_invalidate_tlbs(struct amdgpu_ring *kiq_ring,
+				uint16_t pasid, uint32_t flush_type,
+				bool all_hub)
+{
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_INVALIDATE_TLBS, 0));
+	amdgpu_ring_write(kiq_ring,
+			PACKET3_INVALIDATE_TLBS_DST_SEL(1) |
+			PACKET3_INVALIDATE_TLBS_ALL_HUB(all_hub) |
+			PACKET3_INVALIDATE_TLBS_PASID(pasid) |
+			PACKET3_INVALIDATE_TLBS_FLUSH_TYPE(flush_type));
+}
+
+
+static void gfx_v9_0_kiq_reset_hw_queue(struct amdgpu_ring *kiq_ring, uint32_t queue_type,
+					uint32_t me_id, uint32_t pipe_id, uint32_t queue_id,
+					uint32_t xcc_id, uint32_t vmid)
+{
+	struct amdgpu_device *adev = kiq_ring->adev;
+	unsigned i;
+
+	/* enter save mode */
+	amdgpu_gfx_rlc_enter_safe_mode(adev, xcc_id);
+	mutex_lock(&adev->srbm_mutex);
+	soc15_grbm_select(adev, me_id, pipe_id, queue_id, 0, 0);
+
+	if (queue_type == AMDGPU_RING_TYPE_COMPUTE) {
+		WREG32_SOC15(GC, 0, mmCP_HQD_DEQUEUE_REQUEST, 0x2);
+		WREG32_SOC15(GC, 0, mmSPI_COMPUTE_QUEUE_RESET, 0x1);
+		/* wait till dequeue take effects */
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (!(RREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE) & 1))
+				break;
+			udelay(1);
+		}
+		if (i >= adev->usec_timeout)
+			dev_err(adev->dev, "fail to wait on hqd deactive\n");
+	} else {
+		dev_err(adev->dev, "reset queue_type(%d) not supported\n", queue_type);
+	}
+
+	soc15_grbm_select(adev, 0, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+	/* exit safe mode */
+	amdgpu_gfx_rlc_exit_safe_mode(adev, xcc_id);
+}
+
+static const struct kiq_pm4_funcs gfx_v9_0_kiq_pm4_funcs = {
+	.kiq_set_resources = gfx_v9_0_kiq_set_resources,
+	.kiq_map_queues = gfx_v9_0_kiq_map_queues,
+	.kiq_unmap_queues = gfx_v9_0_kiq_unmap_queues,
+	.kiq_query_status = gfx_v9_0_kiq_query_status,
+	.kiq_invalidate_tlbs = gfx_v9_0_kiq_invalidate_tlbs,
+	.kiq_reset_hw_queue = gfx_v9_0_kiq_reset_hw_queue,
+	.set_resources_size = 8,
+	.map_queues_size = 7,
+	.unmap_queues_size = 6,
+	.query_status_size = 7,
+	.invalidate_tlbs_size = 2,
+};
+
+static void gfx_v9_0_set_kiq_pm4_funcs(struct amdgpu_device *adev)
+{
+	adev->gfx.kiq[0].pmf = &gfx_v9_0_kiq_pm4_funcs;
+}
 
 static void gfx_v9_0_init_golden_registers(struct amdgpu_device *adev)
 {
-	switch (adev->asic_type) {
-	case CHIP_VEGA10:
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_gc_9_0,
-						 (const u32)ARRAY_SIZE(golden_settings_gc_9_0));
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_gc_9_0_vg10,
-						 (const u32)ARRAY_SIZE(golden_settings_gc_9_0_vg10));
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 0, 1):
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_9_0,
+						ARRAY_SIZE(golden_settings_gc_9_0));
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_9_0_vg10,
+						ARRAY_SIZE(golden_settings_gc_9_0_vg10));
+		break;
+	case IP_VERSION(9, 2, 1):
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_9_2_1,
+						ARRAY_SIZE(golden_settings_gc_9_2_1));
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_9_2_1_vg12,
+						ARRAY_SIZE(golden_settings_gc_9_2_1_vg12));
 		break;
-	case CHIP_RAVEN:
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_gc_9_1,
-						 (const u32)ARRAY_SIZE(golden_settings_gc_9_1));
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_gc_9_1_rv1,
-						 (const u32)ARRAY_SIZE(golden_settings_gc_9_1_rv1));
+	case IP_VERSION(9, 4, 0):
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_9_0,
+						ARRAY_SIZE(golden_settings_gc_9_0));
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_9_0_vg20,
+						ARRAY_SIZE(golden_settings_gc_9_0_vg20));
+		break;
+	case IP_VERSION(9, 4, 1):
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_9_4_1_arct,
+						ARRAY_SIZE(golden_settings_gc_9_4_1_arct));
+		break;
+	case IP_VERSION(9, 2, 2):
+	case IP_VERSION(9, 1, 0):
+		soc15_program_register_sequence(adev, golden_settings_gc_9_1,
+						ARRAY_SIZE(golden_settings_gc_9_1));
+		if (adev->apu_flags & AMD_APU_IS_RAVEN2)
+			soc15_program_register_sequence(adev,
+							golden_settings_gc_9_1_rv2,
+							ARRAY_SIZE(golden_settings_gc_9_1_rv2));
+		else
+			soc15_program_register_sequence(adev,
+							golden_settings_gc_9_1_rv1,
+							ARRAY_SIZE(golden_settings_gc_9_1_rv1));
+		break;
+	 case IP_VERSION(9, 3, 0):
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_9_1_rn,
+						ARRAY_SIZE(golden_settings_gc_9_1_rn));
+		return; /* for renoir, don't need common goldensetting */
+	case IP_VERSION(9, 4, 2):
+		gfx_v9_4_2_init_golden_registers(adev,
+						 adev->smuio.funcs->get_die_id(adev));
 		break;
 	default:
 		break;
 	}
-}
 
-static void gfx_v9_0_scratch_init(struct amdgpu_device *adev)
-{
-	adev->gfx.scratch.num_reg = 8;
-	adev->gfx.scratch.reg_base = SOC15_REG_OFFSET(GC, 0, mmSCRATCH_REG0);
-	adev->gfx.scratch.free_mask = (1u << adev->gfx.scratch.num_reg) - 1;
+	if ((amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 1)) &&
+	    (amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 2)))
+		soc15_program_register_sequence(adev, golden_settings_gc_9_x_common,
+						(const u32)ARRAY_SIZE(golden_settings_gc_9_x_common));
 }
 
 static void gfx_v9_0_write_data_to_reg(struct amdgpu_ring *ring, int eng_sel,
@@ -255,26 +1194,18 @@ static void gfx_v9_0_wait_reg_mem(struct amdgpu_ring *ring, int eng_sel,
 static int gfx_v9_0_ring_test_ring(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
-	uint32_t scratch;
+	uint32_t scratch = SOC15_REG_OFFSET(GC, 0, mmSCRATCH_REG0);
 	uint32_t tmp = 0;
 	unsigned i;
 	int r;
 
-	r = amdgpu_gfx_scratch_get(adev, &scratch);
-	if (r) {
-		DRM_ERROR("amdgpu: cp failed to get scratch reg (%d).\n", r);
-		return r;
-	}
 	WREG32(scratch, 0xCAFEDEAD);
 	r = amdgpu_ring_alloc(ring, 3);
-	if (r) {
-		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n",
-			  ring->idx, r);
-		amdgpu_gfx_scratch_free(adev, scratch);
+	if (r)
 		return r;
-	}
+
 	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_UCONFIG_REG, 1));
-	amdgpu_ring_write(ring, (scratch - PACKET3_SET_UCONFIG_REG_START));
+	amdgpu_ring_write(ring, scratch - PACKET3_SET_UCONFIG_REG_START);
 	amdgpu_ring_write(ring, 0xDEADBEEF);
 	amdgpu_ring_commit(ring);
 
@@ -282,290 +1213,413 @@ static int gfx_v9_0_ring_test_ring(struct amdgpu_ring *ring)
 		tmp = RREG32(scratch);
 		if (tmp == 0xDEADBEEF)
 			break;
-		DRM_UDELAY(1);
-	}
-	if (i < adev->usec_timeout) {
-		DRM_INFO("ring test on %d succeeded in %d usecs\n",
-			 ring->idx, i);
-	} else {
-		DRM_ERROR("amdgpu: ring %d test failed (scratch(0x%04X)=0x%08X)\n",
-			  ring->idx, scratch, tmp);
-		r = -EINVAL;
+		udelay(1);
 	}
-	amdgpu_gfx_scratch_free(adev, scratch);
+
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
 	return r;
 }
 
 static int gfx_v9_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 {
-        struct amdgpu_device *adev = ring->adev;
-        struct amdgpu_ib ib;
-        struct dma_fence *f = NULL;
-        uint32_t scratch;
-        uint32_t tmp = 0;
-        long r;
-
-        r = amdgpu_gfx_scratch_get(adev, &scratch);
-        if (r) {
-                DRM_ERROR("amdgpu: failed to get scratch reg (%ld).\n", r);
-                return r;
-        }
-        WREG32(scratch, 0xCAFEDEAD);
-        memset(&ib, 0, sizeof(ib));
-        r = amdgpu_ib_get(adev, NULL, 256, &ib);
-        if (r) {
-                DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
-                goto err1;
-        }
-        ib.ptr[0] = PACKET3(PACKET3_SET_UCONFIG_REG, 1);
-        ib.ptr[1] = ((scratch - PACKET3_SET_UCONFIG_REG_START));
-        ib.ptr[2] = 0xDEADBEEF;
-        ib.length_dw = 3;
-
-        r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
-        if (r)
-                goto err2;
-
-        r = dma_fence_wait_timeout(f, false, timeout);
-        if (r == 0) {
-                DRM_ERROR("amdgpu: IB test timed out.\n");
-                r = -ETIMEDOUT;
-                goto err2;
-        } else if (r < 0) {
-                DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
-                goto err2;
-        }
-        tmp = RREG32(scratch);
-        if (tmp == 0xDEADBEEF) {
-                DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
-                r = 0;
-        } else {
-                DRM_ERROR("amdgpu: ib test failed (scratch(0x%04X)=0x%08X)\n",
-                          scratch, tmp);
-                r = -EINVAL;
-        }
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct dma_fence *f = NULL;
+
+	unsigned index;
+	uint64_t gpu_addr;
+	uint32_t tmp;
+	long r;
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r)
+		return r;
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	adev->wb.wb[index] = cpu_to_le32(0xCAFEDEAD);
+	memset(&ib, 0, sizeof(ib));
+
+	r = amdgpu_ib_get(adev, NULL, 20, AMDGPU_IB_POOL_DIRECT, &ib);
+	if (r)
+		goto err1;
+
+	ib.ptr[0] = PACKET3(PACKET3_WRITE_DATA, 3);
+	ib.ptr[1] = WRITE_DATA_DST_SEL(5) | WR_CONFIRM;
+	ib.ptr[2] = lower_32_bits(gpu_addr);
+	ib.ptr[3] = upper_32_bits(gpu_addr);
+	ib.ptr[4] = 0xDEADBEEF;
+	ib.length_dw = 5;
+
+	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
+	if (r)
+		goto err2;
+
+	r = dma_fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		r = -ETIMEDOUT;
+		goto err2;
+	} else if (r < 0) {
+		goto err2;
+	}
+
+	tmp = adev->wb.wb[index];
+	if (tmp == 0xDEADBEEF)
+		r = 0;
+	else
+		r = -EINVAL;
+
 err2:
-        amdgpu_ib_free(adev, &ib, NULL);
-        dma_fence_put(f);
+	amdgpu_ib_free(&ib, NULL);
+	dma_fence_put(f);
 err1:
-        amdgpu_gfx_scratch_free(adev, scratch);
-        return r;
+	amdgpu_device_wb_free(adev, index);
+	return r;
 }
 
-static int gfx_v9_0_init_microcode(struct amdgpu_device *adev)
+
+static void gfx_v9_0_free_microcode(struct amdgpu_device *adev)
 {
-	const char *chip_name;
-	char fw_name[30];
-	int err;
-	struct amdgpu_firmware_info *info = NULL;
-	const struct common_firmware_header *header = NULL;
-	const struct gfx_firmware_header_v1_0 *cp_hdr;
-	const struct rlc_firmware_header_v2_0 *rlc_hdr;
-	unsigned int *tmp = NULL;
-	unsigned int i = 0;
+	amdgpu_ucode_release(&adev->gfx.pfp_fw);
+	amdgpu_ucode_release(&adev->gfx.me_fw);
+	amdgpu_ucode_release(&adev->gfx.ce_fw);
+	amdgpu_ucode_release(&adev->gfx.rlc_fw);
+	amdgpu_ucode_release(&adev->gfx.mec_fw);
+	amdgpu_ucode_release(&adev->gfx.mec2_fw);
 
-	DRM_DEBUG("\n");
+	kfree(adev->gfx.rlc.register_list_format);
+}
 
-	switch (adev->asic_type) {
-	case CHIP_VEGA10:
-		chip_name = "vega10";
+static void gfx_v9_0_check_fw_write_wait(struct amdgpu_device *adev)
+{
+	adev->gfx.me_fw_write_wait = false;
+	adev->gfx.mec_fw_write_wait = false;
+
+	if ((amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 1)) &&
+	    (amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 2)) &&
+	    ((adev->gfx.mec_fw_version < 0x000001a5) ||
+	     (adev->gfx.mec_feature_version < 46) ||
+	     (adev->gfx.pfp_fw_version < 0x000000b7) ||
+	     (adev->gfx.pfp_feature_version < 46)))
+		DRM_WARN_ONCE("CP firmware version too old, please update!");
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 0, 1):
+		if ((adev->gfx.me_fw_version >= 0x0000009c) &&
+		    (adev->gfx.me_feature_version >= 42) &&
+		    (adev->gfx.pfp_fw_version >=  0x000000b1) &&
+		    (adev->gfx.pfp_feature_version >= 42))
+			adev->gfx.me_fw_write_wait = true;
+
+		if ((adev->gfx.mec_fw_version >=  0x00000193) &&
+		    (adev->gfx.mec_feature_version >= 42))
+			adev->gfx.mec_fw_write_wait = true;
 		break;
-	case CHIP_RAVEN:
-		chip_name = "raven";
+	case IP_VERSION(9, 2, 1):
+		if ((adev->gfx.me_fw_version >= 0x0000009c) &&
+		    (adev->gfx.me_feature_version >= 44) &&
+		    (adev->gfx.pfp_fw_version >=  0x000000b2) &&
+		    (adev->gfx.pfp_feature_version >= 44))
+			adev->gfx.me_fw_write_wait = true;
+
+		if ((adev->gfx.mec_fw_version >=  0x00000196) &&
+		    (adev->gfx.mec_feature_version >= 44))
+			adev->gfx.mec_fw_write_wait = true;
+		break;
+	case IP_VERSION(9, 4, 0):
+		if ((adev->gfx.me_fw_version >= 0x0000009c) &&
+		    (adev->gfx.me_feature_version >= 44) &&
+		    (adev->gfx.pfp_fw_version >=  0x000000b2) &&
+		    (adev->gfx.pfp_feature_version >= 44))
+			adev->gfx.me_fw_write_wait = true;
+
+		if ((adev->gfx.mec_fw_version >=  0x00000197) &&
+		    (adev->gfx.mec_feature_version >= 44))
+			adev->gfx.mec_fw_write_wait = true;
+		break;
+	case IP_VERSION(9, 1, 0):
+	case IP_VERSION(9, 2, 2):
+		if ((adev->gfx.me_fw_version >= 0x0000009c) &&
+		    (adev->gfx.me_feature_version >= 42) &&
+		    (adev->gfx.pfp_fw_version >=  0x000000b1) &&
+		    (adev->gfx.pfp_feature_version >= 42))
+			adev->gfx.me_fw_write_wait = true;
+
+		if ((adev->gfx.mec_fw_version >=  0x00000192) &&
+		    (adev->gfx.mec_feature_version >= 42))
+			adev->gfx.mec_fw_write_wait = true;
 		break;
 	default:
-		BUG();
+		adev->gfx.me_fw_write_wait = true;
+		adev->gfx.mec_fw_write_wait = true;
+		break;
 	}
+}
 
-	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_pfp.bin", chip_name);
-	err = request_firmware(&adev->gfx.pfp_fw, fw_name, adev->dev);
-	if (err)
-		goto out;
-	err = amdgpu_ucode_validate(adev->gfx.pfp_fw);
-	if (err)
-		goto out;
-	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data;
-	adev->gfx.pfp_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
-	adev->gfx.pfp_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
+struct amdgpu_gfxoff_quirk {
+	u16 chip_vendor;
+	u16 chip_device;
+	u16 subsys_vendor;
+	u16 subsys_device;
+	u8 revision;
+};
 
-	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_me.bin", chip_name);
-	err = request_firmware(&adev->gfx.me_fw, fw_name, adev->dev);
-	if (err)
-		goto out;
-	err = amdgpu_ucode_validate(adev->gfx.me_fw);
+static const struct amdgpu_gfxoff_quirk amdgpu_gfxoff_quirk_list[] = {
+	/* https://bugzilla.kernel.org/show_bug.cgi?id=204689 */
+	{ 0x1002, 0x15dd, 0x1002, 0x15dd, 0xc8 },
+	/* https://bugzilla.kernel.org/show_bug.cgi?id=207171 */
+	{ 0x1002, 0x15dd, 0x103c, 0x83e7, 0xd3 },
+	/* GFXOFF is unstable on C6 parts with a VBIOS 113-RAVEN-114 */
+	{ 0x1002, 0x15dd, 0x1002, 0x15dd, 0xc6 },
+	/* Apple MacBook Pro (15-inch, 2019) Radeon Pro Vega 20 4 GB */
+	{ 0x1002, 0x69af, 0x106b, 0x019a, 0xc0 },
+	/* https://bbs.openkylin.top/t/topic/171497 */
+	{ 0x1002, 0x15d8, 0x19e5, 0x3e14, 0xc2 },
+	/* HP 705G4 DM with R5 2400G */
+	{ 0x1002, 0x15dd, 0x103c, 0x8464, 0xd6 },
+	{ 0, 0, 0, 0, 0 },
+};
+
+static bool gfx_v9_0_should_disable_gfxoff(struct pci_dev *pdev)
+{
+	const struct amdgpu_gfxoff_quirk *p = amdgpu_gfxoff_quirk_list;
+
+	while (p && p->chip_device != 0) {
+		if (pdev->vendor == p->chip_vendor &&
+		    pdev->device == p->chip_device &&
+		    pdev->subsystem_vendor == p->subsys_vendor &&
+		    pdev->subsystem_device == p->subsys_device &&
+		    pdev->revision == p->revision) {
+			return true;
+		}
+		++p;
+	}
+	return false;
+}
+
+static bool is_raven_kicker(struct amdgpu_device *adev)
+{
+	if (adev->pm.fw_version >= 0x41e2b)
+		return true;
+	else
+		return false;
+}
+
+static bool check_if_enlarge_doorbell_range(struct amdgpu_device *adev)
+{
+	if ((amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 3, 0)) &&
+	    (adev->gfx.me_fw_version >= 0x000000a5) &&
+	    (adev->gfx.me_feature_version >= 52))
+		return true;
+	else
+		return false;
+}
+
+static void gfx_v9_0_check_if_need_gfxoff(struct amdgpu_device *adev)
+{
+	if (gfx_v9_0_should_disable_gfxoff(adev->pdev))
+		adev->pm.pp_feature &= ~PP_GFXOFF_MASK;
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 0, 1):
+	case IP_VERSION(9, 2, 1):
+	case IP_VERSION(9, 4, 0):
+		break;
+	case IP_VERSION(9, 2, 2):
+	case IP_VERSION(9, 1, 0):
+		if (!((adev->apu_flags & AMD_APU_IS_RAVEN2) ||
+		      (adev->apu_flags & AMD_APU_IS_PICASSO)) &&
+		    ((!is_raven_kicker(adev) &&
+		      adev->gfx.rlc_fw_version < 531) ||
+		     (adev->gfx.rlc_feature_version < 1) ||
+		     !adev->gfx.rlc.is_rlc_v2_1))
+			adev->pm.pp_feature &= ~PP_GFXOFF_MASK;
+
+		if (adev->pm.pp_feature & PP_GFXOFF_MASK)
+			adev->pg_flags |= AMD_PG_SUPPORT_GFX_PG |
+				AMD_PG_SUPPORT_CP |
+				AMD_PG_SUPPORT_RLC_SMU_HS;
+		break;
+	case IP_VERSION(9, 3, 0):
+		if (adev->pm.pp_feature & PP_GFXOFF_MASK)
+			adev->pg_flags |= AMD_PG_SUPPORT_GFX_PG |
+				AMD_PG_SUPPORT_CP |
+				AMD_PG_SUPPORT_RLC_SMU_HS;
+		break;
+	default:
+		break;
+	}
+}
+
+static int gfx_v9_0_init_cp_gfx_microcode(struct amdgpu_device *adev,
+					  char *chip_name)
+{
+	int err;
+
+	err = amdgpu_ucode_request(adev, &adev->gfx.pfp_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_pfp.bin", chip_name);
 	if (err)
 		goto out;
-	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data;
-	adev->gfx.me_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
-	adev->gfx.me_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
+	amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_PFP);
 
-	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_ce.bin", chip_name);
-	err = request_firmware(&adev->gfx.ce_fw, fw_name, adev->dev);
+	err = amdgpu_ucode_request(adev, &adev->gfx.me_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_me.bin", chip_name);
 	if (err)
 		goto out;
-	err = amdgpu_ucode_validate(adev->gfx.ce_fw);
+	amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_ME);
+
+	err = amdgpu_ucode_request(adev, &adev->gfx.ce_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_ce.bin", chip_name);
 	if (err)
 		goto out;
-	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data;
-	adev->gfx.ce_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
-	adev->gfx.ce_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
+	amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_CE);
+
+out:
+	if (err) {
+		amdgpu_ucode_release(&adev->gfx.pfp_fw);
+		amdgpu_ucode_release(&adev->gfx.me_fw);
+		amdgpu_ucode_release(&adev->gfx.ce_fw);
+	}
+	return err;
+}
+
+static int gfx_v9_0_init_rlc_microcode(struct amdgpu_device *adev,
+				       char *chip_name)
+{
+	int err;
+	const struct rlc_firmware_header_v2_0 *rlc_hdr;
+	uint16_t version_major;
+	uint16_t version_minor;
+	uint32_t smu_version;
 
-	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_rlc.bin", chip_name);
-	err = request_firmware(&adev->gfx.rlc_fw, fw_name, adev->dev);
+	/*
+	 * For Picasso && AM4 SOCKET board, we use picasso_rlc_am4.bin
+	 * instead of picasso_rlc.bin.
+	 * Judgment method:
+	 * PCO AM4: revision >= 0xC8 && revision <= 0xCF
+	 *          or revision >= 0xD8 && revision <= 0xDF
+	 * otherwise is PCO FP5
+	 */
+	if (!strcmp(chip_name, "picasso") &&
+		(((adev->pdev->revision >= 0xC8) && (adev->pdev->revision <= 0xCF)) ||
+		((adev->pdev->revision >= 0xD8) && (adev->pdev->revision <= 0xDF))))
+		err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw,
+					   AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_rlc_am4.bin", chip_name);
+	else if (!strcmp(chip_name, "raven") && (amdgpu_pm_load_smu_firmware(adev, &smu_version) == 0) &&
+		(smu_version >= 0x41e2b))
+		/**
+		*SMC is loaded by SBIOS on APU and it's able to get the SMU version directly.
+		*/
+		err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw,
+					   AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_kicker_rlc.bin", chip_name);
+	else
+		err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw,
+					   AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_rlc.bin", chip_name);
 	if (err)
 		goto out;
-	err = amdgpu_ucode_validate(adev->gfx.rlc_fw);
+
 	rlc_hdr = (const struct rlc_firmware_header_v2_0 *)adev->gfx.rlc_fw->data;
-	adev->gfx.rlc_fw_version = le32_to_cpu(rlc_hdr->header.ucode_version);
-	adev->gfx.rlc_feature_version = le32_to_cpu(rlc_hdr->ucode_feature_version);
-	adev->gfx.rlc.save_and_restore_offset =
-			le32_to_cpu(rlc_hdr->save_and_restore_offset);
-	adev->gfx.rlc.clear_state_descriptor_offset =
-			le32_to_cpu(rlc_hdr->clear_state_descriptor_offset);
-	adev->gfx.rlc.avail_scratch_ram_locations =
-			le32_to_cpu(rlc_hdr->avail_scratch_ram_locations);
-	adev->gfx.rlc.reg_restore_list_size =
-			le32_to_cpu(rlc_hdr->reg_restore_list_size);
-	adev->gfx.rlc.reg_list_format_start =
-			le32_to_cpu(rlc_hdr->reg_list_format_start);
-	adev->gfx.rlc.reg_list_format_separate_start =
-			le32_to_cpu(rlc_hdr->reg_list_format_separate_start);
-	adev->gfx.rlc.starting_offsets_start =
-			le32_to_cpu(rlc_hdr->starting_offsets_start);
-	adev->gfx.rlc.reg_list_format_size_bytes =
-			le32_to_cpu(rlc_hdr->reg_list_format_size_bytes);
-	adev->gfx.rlc.reg_list_size_bytes =
-			le32_to_cpu(rlc_hdr->reg_list_size_bytes);
-	adev->gfx.rlc.register_list_format =
-			kmalloc(adev->gfx.rlc.reg_list_format_size_bytes +
-				adev->gfx.rlc.reg_list_size_bytes, GFP_KERNEL);
-	if (!adev->gfx.rlc.register_list_format) {
-		err = -ENOMEM;
-		goto out;
-	}
+	version_major = le16_to_cpu(rlc_hdr->header.header_version_major);
+	version_minor = le16_to_cpu(rlc_hdr->header.header_version_minor);
+	err = amdgpu_gfx_rlc_init_microcode(adev, version_major, version_minor);
+out:
+	if (err)
+		amdgpu_ucode_release(&adev->gfx.rlc_fw);
 
-	tmp = (unsigned int *)((uintptr_t)rlc_hdr +
-			le32_to_cpu(rlc_hdr->reg_list_format_array_offset_bytes));
-	for (i = 0 ; i < (rlc_hdr->reg_list_format_size_bytes >> 2); i++)
-		adev->gfx.rlc.register_list_format[i] =	le32_to_cpu(tmp[i]);
+	return err;
+}
 
-	adev->gfx.rlc.register_restore = adev->gfx.rlc.register_list_format + i;
+static bool gfx_v9_0_load_mec2_fw_bin_support(struct amdgpu_device *adev)
+{
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 1) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 3, 0))
+		return false;
 
-	tmp = (unsigned int *)((uintptr_t)rlc_hdr +
-			le32_to_cpu(rlc_hdr->reg_list_array_offset_bytes));
-	for (i = 0 ; i < (rlc_hdr->reg_list_size_bytes >> 2); i++)
-		adev->gfx.rlc.register_restore[i] = le32_to_cpu(tmp[i]);
+	return true;
+}
 
-	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mec.bin", chip_name);
-	err = request_firmware(&adev->gfx.mec_fw, fw_name, adev->dev);
-	if (err)
-		goto out;
-	err = amdgpu_ucode_validate(adev->gfx.mec_fw);
+static int gfx_v9_0_init_cp_compute_microcode(struct amdgpu_device *adev,
+					      char *chip_name)
+{
+	int err;
+
+	if (amdgpu_sriov_vf(adev) && (adev->asic_type == CHIP_ALDEBARAN))
+		err = amdgpu_ucode_request(adev, &adev->gfx.mec_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_sjt_mec.bin", chip_name);
+	else
+		err = amdgpu_ucode_request(adev, &adev->gfx.mec_fw,
+					   AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_mec.bin", chip_name);
 	if (err)
 		goto out;
-	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
-	adev->gfx.mec_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
-	adev->gfx.mec_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
 
+	amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_MEC1);
+	amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_MEC1_JT);
 
-	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mec2.bin", chip_name);
-	err = request_firmware(&adev->gfx.mec2_fw, fw_name, adev->dev);
-	if (!err) {
-		err = amdgpu_ucode_validate(adev->gfx.mec2_fw);
-		if (err)
-			goto out;
-		cp_hdr = (const struct gfx_firmware_header_v1_0 *)
-		adev->gfx.mec2_fw->data;
-		adev->gfx.mec2_fw_version =
-		le32_to_cpu(cp_hdr->header.ucode_version);
-		adev->gfx.mec2_feature_version =
-		le32_to_cpu(cp_hdr->ucode_feature_version);
-	} else {
-		err = 0;
-		adev->gfx.mec2_fw = NULL;
-	}
-
-	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
-		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_PFP];
-		info->ucode_id = AMDGPU_UCODE_ID_CP_PFP;
-		info->fw = adev->gfx.pfp_fw;
-		header = (const struct common_firmware_header *)info->fw->data;
-		adev->firmware.fw_size +=
-			ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
-
-		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_ME];
-		info->ucode_id = AMDGPU_UCODE_ID_CP_ME;
-		info->fw = adev->gfx.me_fw;
-		header = (const struct common_firmware_header *)info->fw->data;
-		adev->firmware.fw_size +=
-			ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
-
-		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_CE];
-		info->ucode_id = AMDGPU_UCODE_ID_CP_CE;
-		info->fw = adev->gfx.ce_fw;
-		header = (const struct common_firmware_header *)info->fw->data;
-		adev->firmware.fw_size +=
-			ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
-
-		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_RLC_G];
-		info->ucode_id = AMDGPU_UCODE_ID_RLC_G;
-		info->fw = adev->gfx.rlc_fw;
-		header = (const struct common_firmware_header *)info->fw->data;
-		adev->firmware.fw_size +=
-			ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
-
-		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_MEC1];
-		info->ucode_id = AMDGPU_UCODE_ID_CP_MEC1;
-		info->fw = adev->gfx.mec_fw;
-		header = (const struct common_firmware_header *)info->fw->data;
-		cp_hdr = (const struct gfx_firmware_header_v1_0 *)info->fw->data;
-		adev->firmware.fw_size +=
-			ALIGN(le32_to_cpu(header->ucode_size_bytes) - le32_to_cpu(cp_hdr->jt_size) * 4, PAGE_SIZE);
-
-		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_MEC1_JT];
-		info->ucode_id = AMDGPU_UCODE_ID_CP_MEC1_JT;
-		info->fw = adev->gfx.mec_fw;
-		adev->firmware.fw_size +=
-			ALIGN(le32_to_cpu(cp_hdr->jt_size) * 4, PAGE_SIZE);
-
-		if (adev->gfx.mec2_fw) {
-			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_MEC2];
-			info->ucode_id = AMDGPU_UCODE_ID_CP_MEC2;
-			info->fw = adev->gfx.mec2_fw;
-			header = (const struct common_firmware_header *)info->fw->data;
-			cp_hdr = (const struct gfx_firmware_header_v1_0 *)info->fw->data;
-			adev->firmware.fw_size +=
-				ALIGN(le32_to_cpu(header->ucode_size_bytes) - le32_to_cpu(cp_hdr->jt_size) * 4, PAGE_SIZE);
-			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_MEC2_JT];
-			info->ucode_id = AMDGPU_UCODE_ID_CP_MEC2_JT;
-			info->fw = adev->gfx.mec2_fw;
-			adev->firmware.fw_size +=
-				ALIGN(le32_to_cpu(cp_hdr->jt_size) * 4, PAGE_SIZE);
+	if (gfx_v9_0_load_mec2_fw_bin_support(adev)) {
+		if (amdgpu_sriov_vf(adev) && (adev->asic_type == CHIP_ALDEBARAN))
+			err = amdgpu_ucode_request(adev, &adev->gfx.mec2_fw,
+						   AMDGPU_UCODE_REQUIRED,
+						   "amdgpu/%s_sjt_mec2.bin", chip_name);
+		else
+			err = amdgpu_ucode_request(adev, &adev->gfx.mec2_fw,
+						   AMDGPU_UCODE_REQUIRED,
+						   "amdgpu/%s_mec2.bin", chip_name);
+		if (!err) {
+			amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_MEC2);
+			amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_MEC2_JT);
+		} else {
+			err = 0;
+			amdgpu_ucode_release(&adev->gfx.mec2_fw);
 		}
-
+	} else {
+		adev->gfx.mec2_fw_version = adev->gfx.mec_fw_version;
+		adev->gfx.mec2_feature_version = adev->gfx.mec_feature_version;
 	}
 
+	gfx_v9_0_check_if_need_gfxoff(adev);
+	gfx_v9_0_check_fw_write_wait(adev);
+
 out:
-	if (err) {
-		dev_err(adev->dev,
-			"gfx9: Failed to load firmware \"%s\"\n",
-			fw_name);
-		release_firmware(adev->gfx.pfp_fw);
-		adev->gfx.pfp_fw = NULL;
-		release_firmware(adev->gfx.me_fw);
-		adev->gfx.me_fw = NULL;
-		release_firmware(adev->gfx.ce_fw);
-		adev->gfx.ce_fw = NULL;
-		release_firmware(adev->gfx.rlc_fw);
-		adev->gfx.rlc_fw = NULL;
-		release_firmware(adev->gfx.mec_fw);
-		adev->gfx.mec_fw = NULL;
-		release_firmware(adev->gfx.mec2_fw);
-		adev->gfx.mec2_fw = NULL;
-	}
+	if (err)
+		amdgpu_ucode_release(&adev->gfx.mec_fw);
 	return err;
 }
 
+static int gfx_v9_0_init_microcode(struct amdgpu_device *adev)
+{
+	char ucode_prefix[30];
+	int r;
+
+	DRM_DEBUG("\n");
+	amdgpu_ucode_ip_version_decode(adev, GC_HWIP, ucode_prefix, sizeof(ucode_prefix));
+
+	/* No CPG in Arcturus */
+	if (adev->gfx.num_gfx_rings) {
+		r = gfx_v9_0_init_cp_gfx_microcode(adev, ucode_prefix);
+		if (r)
+			return r;
+	}
+
+	r = gfx_v9_0_init_rlc_microcode(adev, ucode_prefix);
+	if (r)
+		return r;
+
+	r = gfx_v9_0_init_cp_compute_microcode(adev, ucode_prefix);
+	if (r)
+		return r;
+
+	return r;
+}
+
 static u32 gfx_v9_0_get_csb_size(struct amdgpu_device *adev)
 {
 	u32 count = 0;
@@ -594,45 +1648,62 @@ static u32 gfx_v9_0_get_csb_size(struct amdgpu_device *adev)
 	return count;
 }
 
-static void gfx_v9_0_get_csb_buffer(struct amdgpu_device *adev,
-				    volatile u32 *buffer)
+static void gfx_v9_0_get_csb_buffer(struct amdgpu_device *adev, u32 *buffer)
 {
-	u32 count = 0, i;
-	const struct cs_section_def *sect = NULL;
-	const struct cs_extent_def *ext = NULL;
+	u32 count = 0;
 
 	if (adev->gfx.rlc.cs_data == NULL)
 		return;
 	if (buffer == NULL)
 		return;
 
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
-	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
+	count = amdgpu_gfx_csb_preamble_start(buffer);
+	count = amdgpu_gfx_csb_data_parser(adev, buffer, count);
+	amdgpu_gfx_csb_preamble_end(buffer, count);
+}
 
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CONTEXT_CONTROL, 1));
-	buffer[count++] = cpu_to_le32(0x80000000);
-	buffer[count++] = cpu_to_le32(0x80000000);
+static void gfx_v9_0_init_always_on_cu_mask(struct amdgpu_device *adev)
+{
+	struct amdgpu_cu_info *cu_info = &adev->gfx.cu_info;
+	uint32_t pg_always_on_cu_num = 2;
+	uint32_t always_on_cu_num;
+	uint32_t i, j, k;
+	uint32_t mask, cu_bitmap, counter;
 
-	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
-		for (ext = sect->section; ext->extent != NULL; ++ext) {
-			if (sect->id == SECT_CONTEXT) {
-				buffer[count++] =
-					cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, ext->reg_count));
-				buffer[count++] = cpu_to_le32(ext->reg_index -
-						PACKET3_SET_CONTEXT_REG_START);
-				for (i = 0; i < ext->reg_count; i++)
-					buffer[count++] = cpu_to_le32(ext->extent[i]);
-			} else {
-				return;
+	if (adev->flags & AMD_IS_APU)
+		always_on_cu_num = 4;
+	else if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 2, 1))
+		always_on_cu_num = 8;
+	else
+		always_on_cu_num = 12;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			mask = 1;
+			cu_bitmap = 0;
+			counter = 0;
+			amdgpu_gfx_select_se_sh(adev, i, j, 0xffffffff, 0);
+
+			for (k = 0; k < adev->gfx.config.max_cu_per_sh; k ++) {
+				if (cu_info->bitmap[0][i][j] & mask) {
+					if (counter == pg_always_on_cu_num)
+						WREG32_SOC15(GC, 0, mmRLC_PG_ALWAYS_ON_CU_MASK, cu_bitmap);
+					if (counter < always_on_cu_num)
+						cu_bitmap |= mask;
+					else
+						break;
+					counter++;
+				}
+				mask <<= 1;
 			}
+
+			WREG32_SOC15(GC, 0, mmRLC_LB_ALWAYS_ACTIVE_CU_MASK, cu_bitmap);
+			cu_info->ao_cu_bitmap[i][j] = cu_bitmap;
 		}
 	}
-
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
-	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_END_CLEAR_STATE);
-
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CLEAR_STATE, 0));
-	buffer[count++] = cpu_to_le32(0);
+	amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
+	mutex_unlock(&adev->grbm_idx_mutex);
 }
 
 static void gfx_v9_0_init_lbpw(struct amdgpu_device *adev)
@@ -653,7 +1724,7 @@ static void gfx_v9_0_init_lbpw(struct amdgpu_device *adev)
 
 	mutex_lock(&adev->grbm_idx_mutex);
 	/* set mmRLC_LB_INIT_CU_MASK thru broadcast mode to enable all SE/SH*/
-	gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 	WREG32_SOC15(GC, 0, mmRLC_LB_INIT_CU_MASK, 0xffffffff);
 
 	/* set mmRLC_LB_PARAMS = 0x003F_1006 */
@@ -668,8 +1739,10 @@ static void gfx_v9_0_init_lbpw(struct amdgpu_device *adev)
 	data |= 0x00C00000;
 	WREG32_SOC15(GC, 0, mmRLC_GPM_GENERAL_7, data);
 
-	/* set RLC_LB_ALWAYS_ACTIVE_CU_MASK = 0xFFF */
-	WREG32_SOC15(GC, 0, mmRLC_LB_ALWAYS_ACTIVE_CU_MASK, 0xFFF);
+	/*
+	 * RLC_LB_ALWAYS_ACTIVE_CU_MASK = 0xF (4 CUs AON for Raven),
+	 * programmed in gfx_v9_0_init_always_on_cu_mask()
+	 */
 
 	/* set RLC_LB_CNTL = 0x8000_0095, 31 bit is reserved,
 	 * but used for RLC_LB_CNTL configuration */
@@ -678,6 +1751,57 @@ static void gfx_v9_0_init_lbpw(struct amdgpu_device *adev)
 	data |= REG_SET_FIELD(data, RLC_LB_CNTL, RESERVED, 0x80000);
 	WREG32_SOC15(GC, 0, mmRLC_LB_CNTL, data);
 	mutex_unlock(&adev->grbm_idx_mutex);
+
+	gfx_v9_0_init_always_on_cu_mask(adev);
+}
+
+static void gfx_v9_4_init_lbpw(struct amdgpu_device *adev)
+{
+	uint32_t data;
+
+	/* set mmRLC_LB_THR_CONFIG_1/2/3/4 */
+	WREG32_SOC15(GC, 0, mmRLC_LB_THR_CONFIG_1, 0x0000007F);
+	WREG32_SOC15(GC, 0, mmRLC_LB_THR_CONFIG_2, 0x033388F8);
+	WREG32_SOC15(GC, 0, mmRLC_LB_THR_CONFIG_3, 0x00000077);
+	WREG32_SOC15(GC, 0, mmRLC_LB_THR_CONFIG_4, (0x10 | 0x27 << 8 | 0x02FA << 16));
+
+	/* set mmRLC_LB_CNTR_INIT = 0x0000_0000 */
+	WREG32_SOC15(GC, 0, mmRLC_LB_CNTR_INIT, 0x00000000);
+
+	/* set mmRLC_LB_CNTR_MAX = 0x0000_0500 */
+	WREG32_SOC15(GC, 0, mmRLC_LB_CNTR_MAX, 0x00000800);
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	/* set mmRLC_LB_INIT_CU_MASK thru broadcast mode to enable all SE/SH*/
+	amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
+	WREG32_SOC15(GC, 0, mmRLC_LB_INIT_CU_MASK, 0xffffffff);
+
+	/* set mmRLC_LB_PARAMS = 0x003F_1006 */
+	data = REG_SET_FIELD(0, RLC_LB_PARAMS, FIFO_SAMPLES, 0x0003);
+	data |= REG_SET_FIELD(data, RLC_LB_PARAMS, PG_IDLE_SAMPLES, 0x0010);
+	data |= REG_SET_FIELD(data, RLC_LB_PARAMS, PG_IDLE_SAMPLE_INTERVAL, 0x033F);
+	WREG32_SOC15(GC, 0, mmRLC_LB_PARAMS, data);
+
+	/* set mmRLC_GPM_GENERAL_7[31-16] = 0x00C0 */
+	data = RREG32_SOC15(GC, 0, mmRLC_GPM_GENERAL_7);
+	data &= 0x0000FFFF;
+	data |= 0x00C00000;
+	WREG32_SOC15(GC, 0, mmRLC_GPM_GENERAL_7, data);
+
+	/*
+	 * RLC_LB_ALWAYS_ACTIVE_CU_MASK = 0xFFF (12 CUs AON),
+	 * programmed in gfx_v9_0_init_always_on_cu_mask()
+	 */
+
+	/* set RLC_LB_CNTL = 0x8000_0095, 31 bit is reserved,
+	 * but used for RLC_LB_CNTL configuration */
+	data = RLC_LB_CNTL__LB_CNT_SPIM_ACTIVE_MASK;
+	data |= REG_SET_FIELD(data, RLC_LB_CNTL, CU_MASK_USED_OFF_HYST, 0x09);
+	data |= REG_SET_FIELD(data, RLC_LB_CNTL, RESERVED, 0x80000);
+	WREG32_SOC15(GC, 0, mmRLC_LB_CNTL, data);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	gfx_v9_0_init_always_on_cu_mask(adev);
 }
 
 static void gfx_v9_0_enable_lbpw(struct amdgpu_device *adev, bool enable)
@@ -685,85 +1809,31 @@ static void gfx_v9_0_enable_lbpw(struct amdgpu_device *adev, bool enable)
 	WREG32_FIELD15(GC, 0, RLC_LB_CNTL, LOAD_BALANCE_ENABLE, enable ? 1 : 0);
 }
 
-static void rv_init_cp_jump_table(struct amdgpu_device *adev)
+static int gfx_v9_0_cp_jump_table_num(struct amdgpu_device *adev)
 {
-	const __le32 *fw_data;
-	volatile u32 *dst_ptr;
-	int me, i, max_me = 5;
-	u32 bo_offset = 0;
-	u32 table_offset, table_size;
-
-	/* write the cp table buffer */
-	dst_ptr = adev->gfx.rlc.cp_table_ptr;
-	for (me = 0; me < max_me; me++) {
-		if (me == 0) {
-			const struct gfx_firmware_header_v1_0 *hdr =
-				(const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data;
-			fw_data = (const __le32 *)
-				(adev->gfx.ce_fw->data +
-				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
-			table_offset = le32_to_cpu(hdr->jt_offset);
-			table_size = le32_to_cpu(hdr->jt_size);
-		} else if (me == 1) {
-			const struct gfx_firmware_header_v1_0 *hdr =
-				(const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data;
-			fw_data = (const __le32 *)
-				(adev->gfx.pfp_fw->data +
-				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
-			table_offset = le32_to_cpu(hdr->jt_offset);
-			table_size = le32_to_cpu(hdr->jt_size);
-		} else if (me == 2) {
-			const struct gfx_firmware_header_v1_0 *hdr =
-				(const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data;
-			fw_data = (const __le32 *)
-				(adev->gfx.me_fw->data +
-				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
-			table_offset = le32_to_cpu(hdr->jt_offset);
-			table_size = le32_to_cpu(hdr->jt_size);
-		} else if (me == 3) {
-			const struct gfx_firmware_header_v1_0 *hdr =
-				(const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
-			fw_data = (const __le32 *)
-				(adev->gfx.mec_fw->data +
-				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
-			table_offset = le32_to_cpu(hdr->jt_offset);
-			table_size = le32_to_cpu(hdr->jt_size);
-		} else  if (me == 4) {
-			const struct gfx_firmware_header_v1_0 *hdr =
-				(const struct gfx_firmware_header_v1_0 *)adev->gfx.mec2_fw->data;
-			fw_data = (const __le32 *)
-				(adev->gfx.mec2_fw->data +
-				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
-			table_offset = le32_to_cpu(hdr->jt_offset);
-			table_size = le32_to_cpu(hdr->jt_size);
-		}
-
-		for (i = 0; i < table_size; i ++) {
-			dst_ptr[bo_offset + i] =
-				cpu_to_le32(le32_to_cpu(fw_data[table_offset + i]));
-		}
-
-		bo_offset += table_size;
-	}
+	if (gfx_v9_0_load_mec2_fw_bin_support(adev))
+		return 5;
+	else
+		return 4;
 }
 
-static void gfx_v9_0_rlc_fini(struct amdgpu_device *adev)
+static void gfx_v9_0_init_rlcg_reg_access_ctrl(struct amdgpu_device *adev)
 {
-	/* clear state block */
-	amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj,
-			&adev->gfx.rlc.clear_state_gpu_addr,
-			(void **)&adev->gfx.rlc.cs_ptr);
+	struct amdgpu_rlcg_reg_access_ctrl *reg_access_ctrl;
 
-	/* jump table block */
-	amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj,
-			&adev->gfx.rlc.cp_table_gpu_addr,
-			(void **)&adev->gfx.rlc.cp_table_ptr);
+	reg_access_ctrl = &adev->gfx.rlc.reg_access_ctrl[0];
+	reg_access_ctrl->scratch_reg0 = SOC15_REG_OFFSET(GC, 0, mmSCRATCH_REG0);
+	reg_access_ctrl->scratch_reg1 = SOC15_REG_OFFSET(GC, 0, mmSCRATCH_REG1);
+	reg_access_ctrl->scratch_reg2 = SOC15_REG_OFFSET(GC, 0, mmSCRATCH_REG2);
+	reg_access_ctrl->scratch_reg3 = SOC15_REG_OFFSET(GC, 0, mmSCRATCH_REG3);
+	reg_access_ctrl->grbm_cntl = SOC15_REG_OFFSET(GC, 0, mmGRBM_GFX_CNTL);
+	reg_access_ctrl->grbm_idx = SOC15_REG_OFFSET(GC, 0, mmGRBM_GFX_INDEX);
+	reg_access_ctrl->spare_int = SOC15_REG_OFFSET(GC, 0, mmRLC_SPARE_INT);
+	adev->gfx.rlc.rlcg_reg_access_supported = true;
 }
 
 static int gfx_v9_0_rlc_init(struct amdgpu_device *adev)
 {
-	volatile u32 *dst_ptr;
-	u32 dws;
 	const struct cs_section_def *cs_data;
 	int r;
 
@@ -772,46 +1842,18 @@ static int gfx_v9_0_rlc_init(struct amdgpu_device *adev)
 	cs_data = adev->gfx.rlc.cs_data;
 
 	if (cs_data) {
-		/* clear state block */
-		adev->gfx.rlc.clear_state_size = dws = gfx_v9_0_get_csb_size(adev);
-		r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
-					      AMDGPU_GEM_DOMAIN_VRAM,
-					      &adev->gfx.rlc.clear_state_obj,
-					      &adev->gfx.rlc.clear_state_gpu_addr,
-					      (void **)&adev->gfx.rlc.cs_ptr);
-		if (r) {
-			dev_err(adev->dev, "(%d) failed to create rlc csb bo\n",
-				r);
-			gfx_v9_0_rlc_fini(adev);
+		/* init clear state block */
+		r = amdgpu_gfx_rlc_init_csb(adev);
+		if (r)
 			return r;
-		}
-		/* set up the cs buffer */
-		dst_ptr = adev->gfx.rlc.cs_ptr;
-		gfx_v9_0_get_csb_buffer(adev, dst_ptr);
-		amdgpu_bo_kunmap(adev->gfx.rlc.clear_state_obj);
-		amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
 	}
 
-	if (adev->asic_type == CHIP_RAVEN) {
+	if (adev->flags & AMD_IS_APU) {
 		/* TODO: double check the cp_table_size for RV */
 		adev->gfx.rlc.cp_table_size = ALIGN(96 * 5 * 4, 2048) + (64 * 1024); /* JT + GDS */
-		r = amdgpu_bo_create_reserved(adev, adev->gfx.rlc.cp_table_size,
-					      PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
-					      &adev->gfx.rlc.cp_table_obj,
-					      &adev->gfx.rlc.cp_table_gpu_addr,
-					      (void **)&adev->gfx.rlc.cp_table_ptr);
-		if (r) {
-			dev_err(adev->dev,
-				"(%d) failed to create cp table bo\n", r);
-			gfx_v9_0_rlc_fini(adev);
+		r = amdgpu_gfx_rlc_init_cpt(adev);
+		if (r)
 			return r;
-		}
-
-		rv_init_cp_jump_table(adev);
-		amdgpu_bo_kunmap(adev->gfx.rlc.cp_table_obj);
-		amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj);
-
-		gfx_v9_0_init_lbpw(adev);
 	}
 
 	return 0;
@@ -834,34 +1876,36 @@ static int gfx_v9_0_mec_init(struct amdgpu_device *adev)
 
 	const struct gfx_firmware_header_v1_0 *mec_hdr;
 
-	bitmap_zero(adev->gfx.mec.queue_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
+	bitmap_zero(adev->gfx.mec_bitmap[0].queue_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
 
 	/* take ownership of the relevant compute queues */
 	amdgpu_gfx_compute_queue_acquire(adev);
 	mec_hpd_size = adev->gfx.num_compute_rings * GFX9_MEC_HPD_SIZE;
+	if (mec_hpd_size) {
+		r = amdgpu_bo_create_reserved(adev, mec_hpd_size, PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_VRAM |
+					      AMDGPU_GEM_DOMAIN_GTT,
+					      &adev->gfx.mec.hpd_eop_obj,
+					      &adev->gfx.mec.hpd_eop_gpu_addr,
+					      (void **)&hpd);
+		if (r) {
+			dev_warn(adev->dev, "(%d) create HDP EOP bo failed\n", r);
+			gfx_v9_0_mec_fini(adev);
+			return r;
+		}
 
-	r = amdgpu_bo_create_reserved(adev, mec_hpd_size, PAGE_SIZE,
-				      AMDGPU_GEM_DOMAIN_GTT,
-				      &adev->gfx.mec.hpd_eop_obj,
-				      &adev->gfx.mec.hpd_eop_gpu_addr,
-				      (void **)&hpd);
-	if (r) {
-		dev_warn(adev->dev, "(%d) create HDP EOP bo failed\n", r);
-		gfx_v9_0_mec_fini(adev);
-		return r;
-	}
-
-	memset(hpd, 0, adev->gfx.mec.hpd_eop_obj->tbo.mem.size);
+		memset(hpd, 0, mec_hpd_size);
 
-	amdgpu_bo_kunmap(adev->gfx.mec.hpd_eop_obj);
-	amdgpu_bo_unreserve(adev->gfx.mec.hpd_eop_obj);
+		amdgpu_bo_kunmap(adev->gfx.mec.hpd_eop_obj);
+		amdgpu_bo_unreserve(adev->gfx.mec.hpd_eop_obj);
+	}
 
 	mec_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
 
 	fw_data = (const __le32 *)
 		(adev->gfx.mec_fw->data +
 		 le32_to_cpu(mec_hdr->header.ucode_array_offset_bytes));
-	fw_size = le32_to_cpu(mec_hdr->header.ucode_size_bytes) / 4;
+	fw_size = le32_to_cpu(mec_hdr->header.ucode_size_bytes);
 
 	r = amdgpu_bo_create_reserved(adev, mec_hdr->header.ucode_size_bytes,
 				      PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
@@ -884,7 +1928,7 @@ static int gfx_v9_0_mec_init(struct amdgpu_device *adev)
 
 static uint32_t wave_read_ind(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t address)
 {
-	WREG32_SOC15(GC, 0, mmSQ_IND_INDEX,
+	WREG32_SOC15_RLC(GC, 0, mmSQ_IND_INDEX,
 		(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
 		(simd << SQ_IND_INDEX__SIMD_ID__SHIFT) |
 		(address << SQ_IND_INDEX__INDEX__SHIFT) |
@@ -896,7 +1940,7 @@ static void wave_read_regs(struct amdgpu_device *adev, uint32_t simd,
 			   uint32_t wave, uint32_t thread,
 			   uint32_t regno, uint32_t num, uint32_t *out)
 {
-	WREG32_SOC15(GC, 0, mmSQ_IND_INDEX,
+	WREG32_SOC15_RLC(GC, 0, mmSQ_IND_INDEX,
 		(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
 		(simd << SQ_IND_INDEX__SIMD_ID__SHIFT) |
 		(regno << SQ_IND_INDEX__INDEX__SHIFT) |
@@ -907,7 +1951,7 @@ static void wave_read_regs(struct amdgpu_device *adev, uint32_t simd,
 		*(out++) = RREG32_SOC15(GC, 0, mmSQ_IND_DATA);
 }
 
-static void gfx_v9_0_read_wave_data(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t *dst, int *no_fields)
+static void gfx_v9_0_read_wave_data(struct amdgpu_device *adev, uint32_t xcc_id, uint32_t simd, uint32_t wave, uint32_t *dst, int *no_fields)
 {
 	/* type 1 wave data */
 	dst[(*no_fields)++] = 1;
@@ -925,9 +1969,10 @@ static void gfx_v9_0_read_wave_data(struct amdgpu_device *adev, uint32_t simd, u
 	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_IB_STS);
 	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_IB_DBG0);
 	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_M0);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_MODE);
 }
 
-static void gfx_v9_0_read_wave_sgprs(struct amdgpu_device *adev, uint32_t simd,
+static void gfx_v9_0_read_wave_sgprs(struct amdgpu_device *adev, uint32_t xcc_id, uint32_t simd,
 				     uint32_t wave, uint32_t start,
 				     uint32_t size, uint32_t *dst)
 {
@@ -936,22 +1981,50 @@ static void gfx_v9_0_read_wave_sgprs(struct amdgpu_device *adev, uint32_t simd,
 		start + SQIND_WAVE_SGPRS_OFFSET, size, dst);
 }
 
+static void gfx_v9_0_read_wave_vgprs(struct amdgpu_device *adev, uint32_t xcc_id, uint32_t simd,
+				     uint32_t wave, uint32_t thread,
+				     uint32_t start, uint32_t size,
+				     uint32_t *dst)
+{
+	wave_read_regs(
+		adev, simd, wave, thread,
+		start + SQIND_WAVE_VGPRS_OFFSET, size, dst);
+}
+
+static void gfx_v9_0_select_me_pipe_q(struct amdgpu_device *adev,
+				  u32 me, u32 pipe, u32 q, u32 vm, u32 xcc_id)
+{
+	soc15_grbm_select(adev, me, pipe, q, vm, 0);
+}
 
 static const struct amdgpu_gfx_funcs gfx_v9_0_gfx_funcs = {
-	.get_gpu_clock_counter = &gfx_v9_0_get_gpu_clock_counter,
-	.select_se_sh = &gfx_v9_0_select_se_sh,
-	.read_wave_data = &gfx_v9_0_read_wave_data,
-	.read_wave_sgprs = &gfx_v9_0_read_wave_sgprs,
+        .get_gpu_clock_counter = &gfx_v9_0_get_gpu_clock_counter,
+        .select_se_sh = &gfx_v9_0_select_se_sh,
+        .read_wave_data = &gfx_v9_0_read_wave_data,
+        .read_wave_sgprs = &gfx_v9_0_read_wave_sgprs,
+        .read_wave_vgprs = &gfx_v9_0_read_wave_vgprs,
+        .select_me_pipe_q = &gfx_v9_0_select_me_pipe_q,
+};
+
+const struct amdgpu_ras_block_hw_ops  gfx_v9_0_ras_ops = {
+		.ras_error_inject = &gfx_v9_0_ras_error_inject,
+		.query_ras_error_count = &gfx_v9_0_query_ras_error_count,
+		.reset_ras_error_count = &gfx_v9_0_reset_ras_error_count,
+};
+
+static struct amdgpu_gfx_ras gfx_v9_0_ras = {
+	.ras_block = {
+		.hw_ops = &gfx_v9_0_ras_ops,
+	},
 };
 
-static void gfx_v9_0_gpu_early_init(struct amdgpu_device *adev)
+static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev)
 {
 	u32 gb_addr_config;
+	int err;
 
-	adev->gfx.funcs = &gfx_v9_0_gfx_funcs;
-
-	switch (adev->asic_type) {
-	case CHIP_VEGA10:
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 0, 1):
 		adev->gfx.config.max_hw_contexts = 8;
 		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
 		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
@@ -959,13 +2032,77 @@ static void gfx_v9_0_gpu_early_init(struct amdgpu_device *adev)
 		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x4C0;
 		gb_addr_config = VEGA10_GB_ADDR_CONFIG_GOLDEN;
 		break;
-	case CHIP_RAVEN:
+	case IP_VERSION(9, 2, 1):
+		adev->gfx.config.max_hw_contexts = 8;
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x4C0;
+		gb_addr_config = VEGA12_GB_ADDR_CONFIG_GOLDEN;
+		DRM_INFO("fix gfx.config for vega12\n");
+		break;
+	case IP_VERSION(9, 4, 0):
+		adev->gfx.ras = &gfx_v9_0_ras;
+		adev->gfx.config.max_hw_contexts = 8;
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x4C0;
+		gb_addr_config = RREG32_SOC15(GC, 0, mmGB_ADDR_CONFIG);
+		gb_addr_config &= ~0xf3e777ff;
+		gb_addr_config |= 0x22014042;
+		/* check vbios table if gpu info is not available */
+		err = amdgpu_atomfirmware_get_gfx_info(adev);
+		if (err)
+			return err;
+		break;
+	case IP_VERSION(9, 2, 2):
+	case IP_VERSION(9, 1, 0):
+		adev->gfx.config.max_hw_contexts = 8;
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x4C0;
+		if (adev->apu_flags & AMD_APU_IS_RAVEN2)
+			gb_addr_config = RAVEN2_GB_ADDR_CONFIG_GOLDEN;
+		else
+			gb_addr_config = RAVEN_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	case IP_VERSION(9, 4, 1):
+		adev->gfx.ras = &gfx_v9_4_ras;
+		adev->gfx.config.max_hw_contexts = 8;
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x4C0;
+		gb_addr_config = RREG32_SOC15(GC, 0, mmGB_ADDR_CONFIG);
+		gb_addr_config &= ~0xf3e777ff;
+		gb_addr_config |= 0x22014042;
+		break;
+	case IP_VERSION(9, 3, 0):
+		adev->gfx.config.max_hw_contexts = 8;
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x80;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x4C0;
+		gb_addr_config = RREG32_SOC15(GC, 0, mmGB_ADDR_CONFIG);
+		gb_addr_config &= ~0xf3e777ff;
+		gb_addr_config |= 0x22010042;
+		break;
+	case IP_VERSION(9, 4, 2):
+		adev->gfx.ras = &gfx_v9_4_2_ras;
 		adev->gfx.config.max_hw_contexts = 8;
 		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
 		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
 		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
 		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x4C0;
-		gb_addr_config = RAVEN_GB_ADDR_CONFIG_GOLDEN;
+		gb_addr_config = RREG32_SOC15(GC, 0, mmGB_ADDR_CONFIG);
+		gb_addr_config &= ~0xf3e777ff;
+		gb_addr_config |= 0x22014042;
+		/* check vbios table if gpu info is not available */
+		err = amdgpu_atomfirmware_get_gfx_info(adev);
+		if (err)
+			return err;
 		break;
 	default:
 		BUG();
@@ -1008,197 +2145,6 @@ static void gfx_v9_0_gpu_early_init(struct amdgpu_device *adev)
 					adev->gfx.config.gb_addr_config,
 					GB_ADDR_CONFIG,
 					PIPE_INTERLEAVE_SIZE));
-}
-
-static int gfx_v9_0_ngg_create_buf(struct amdgpu_device *adev,
-				   struct amdgpu_ngg_buf *ngg_buf,
-				   int size_se,
-				   int default_size_se)
-{
-	int r;
-
-	if (size_se < 0) {
-		dev_err(adev->dev, "Buffer size is invalid: %d\n", size_se);
-		return -EINVAL;
-	}
-	size_se = size_se ? size_se : default_size_se;
-
-	ngg_buf->size = size_se * adev->gfx.config.max_shader_engines;
-	r = amdgpu_bo_create_kernel(adev, ngg_buf->size,
-				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
-				    &ngg_buf->bo,
-				    &ngg_buf->gpu_addr,
-				    NULL);
-	if (r) {
-		dev_err(adev->dev, "(%d) failed to create NGG buffer\n", r);
-		return r;
-	}
-	ngg_buf->bo_size = amdgpu_bo_size(ngg_buf->bo);
-
-	return r;
-}
-
-static int gfx_v9_0_ngg_fini(struct amdgpu_device *adev)
-{
-	int i;
-
-	for (i = 0; i < NGG_BUF_MAX; i++)
-		amdgpu_bo_free_kernel(&adev->gfx.ngg.buf[i].bo,
-				      &adev->gfx.ngg.buf[i].gpu_addr,
-				      NULL);
-
-	memset(&adev->gfx.ngg.buf[0], 0,
-			sizeof(struct amdgpu_ngg_buf) * NGG_BUF_MAX);
-
-	adev->gfx.ngg.init = false;
-
-	return 0;
-}
-
-static int gfx_v9_0_ngg_init(struct amdgpu_device *adev)
-{
-	int r;
-
-	if (!amdgpu_ngg || adev->gfx.ngg.init == true)
-		return 0;
-
-	/* GDS reserve memory: 64 bytes alignment */
-	adev->gfx.ngg.gds_reserve_size = ALIGN(5 * 4, 0x40);
-	adev->gds.mem.total_size -= adev->gfx.ngg.gds_reserve_size;
-	adev->gds.mem.gfx_partition_size -= adev->gfx.ngg.gds_reserve_size;
-	adev->gfx.ngg.gds_reserve_addr = amdgpu_gds_reg_offset[0].mem_base;
-	adev->gfx.ngg.gds_reserve_addr += adev->gds.mem.gfx_partition_size;
-
-	/* Primitive Buffer */
-	r = gfx_v9_0_ngg_create_buf(adev, &adev->gfx.ngg.buf[NGG_PRIM],
-				    amdgpu_prim_buf_per_se,
-				    64 * 1024);
-	if (r) {
-		dev_err(adev->dev, "Failed to create Primitive Buffer\n");
-		goto err;
-	}
-
-	/* Position Buffer */
-	r = gfx_v9_0_ngg_create_buf(adev, &adev->gfx.ngg.buf[NGG_POS],
-				    amdgpu_pos_buf_per_se,
-				    256 * 1024);
-	if (r) {
-		dev_err(adev->dev, "Failed to create Position Buffer\n");
-		goto err;
-	}
-
-	/* Control Sideband */
-	r = gfx_v9_0_ngg_create_buf(adev, &adev->gfx.ngg.buf[NGG_CNTL],
-				    amdgpu_cntl_sb_buf_per_se,
-				    256);
-	if (r) {
-		dev_err(adev->dev, "Failed to create Control Sideband Buffer\n");
-		goto err;
-	}
-
-	/* Parameter Cache, not created by default */
-	if (amdgpu_param_buf_per_se <= 0)
-		goto out;
-
-	r = gfx_v9_0_ngg_create_buf(adev, &adev->gfx.ngg.buf[NGG_PARAM],
-				    amdgpu_param_buf_per_se,
-				    512 * 1024);
-	if (r) {
-		dev_err(adev->dev, "Failed to create Parameter Cache\n");
-		goto err;
-	}
-
-out:
-	adev->gfx.ngg.init = true;
-	return 0;
-err:
-	gfx_v9_0_ngg_fini(adev);
-	return r;
-}
-
-static int gfx_v9_0_ngg_en(struct amdgpu_device *adev)
-{
-	struct amdgpu_ring *ring = &adev->gfx.gfx_ring[0];
-	int r;
-	u32 data;
-	u32 size;
-	u32 base;
-
-	if (!amdgpu_ngg)
-		return 0;
-
-	/* Program buffer size */
-	data = 0;
-	size = adev->gfx.ngg.buf[NGG_PRIM].size / 256;
-	data = REG_SET_FIELD(data, WD_BUF_RESOURCE_1, INDEX_BUF_SIZE, size);
-
-	size = adev->gfx.ngg.buf[NGG_POS].size / 256;
-	data = REG_SET_FIELD(data, WD_BUF_RESOURCE_1, POS_BUF_SIZE, size);
-
-	WREG32_SOC15(GC, 0, mmWD_BUF_RESOURCE_1, data);
-
-	data = 0;
-	size = adev->gfx.ngg.buf[NGG_CNTL].size / 256;
-	data = REG_SET_FIELD(data, WD_BUF_RESOURCE_2, CNTL_SB_BUF_SIZE, size);
-
-	size = adev->gfx.ngg.buf[NGG_PARAM].size / 1024;
-	data = REG_SET_FIELD(data, WD_BUF_RESOURCE_2, PARAM_BUF_SIZE, size);
-
-	WREG32_SOC15(GC, 0, mmWD_BUF_RESOURCE_2, data);
-
-	/* Program buffer base address */
-	base = lower_32_bits(adev->gfx.ngg.buf[NGG_PRIM].gpu_addr);
-	data = REG_SET_FIELD(0, WD_INDEX_BUF_BASE, BASE, base);
-	WREG32_SOC15(GC, 0, mmWD_INDEX_BUF_BASE, data);
-
-	base = upper_32_bits(adev->gfx.ngg.buf[NGG_PRIM].gpu_addr);
-	data = REG_SET_FIELD(0, WD_INDEX_BUF_BASE_HI, BASE_HI, base);
-	WREG32_SOC15(GC, 0, mmWD_INDEX_BUF_BASE_HI, data);
-
-	base = lower_32_bits(adev->gfx.ngg.buf[NGG_POS].gpu_addr);
-	data = REG_SET_FIELD(0, WD_POS_BUF_BASE, BASE, base);
-	WREG32_SOC15(GC, 0, mmWD_POS_BUF_BASE, data);
-
-	base = upper_32_bits(adev->gfx.ngg.buf[NGG_POS].gpu_addr);
-	data = REG_SET_FIELD(0, WD_POS_BUF_BASE_HI, BASE_HI, base);
-	WREG32_SOC15(GC, 0, mmWD_POS_BUF_BASE_HI, data);
-
-	base = lower_32_bits(adev->gfx.ngg.buf[NGG_CNTL].gpu_addr);
-	data = REG_SET_FIELD(0, WD_CNTL_SB_BUF_BASE, BASE, base);
-	WREG32_SOC15(GC, 0, mmWD_CNTL_SB_BUF_BASE, data);
-
-	base = upper_32_bits(adev->gfx.ngg.buf[NGG_CNTL].gpu_addr);
-	data = REG_SET_FIELD(0, WD_CNTL_SB_BUF_BASE_HI, BASE_HI, base);
-	WREG32_SOC15(GC, 0, mmWD_CNTL_SB_BUF_BASE_HI, data);
-
-	/* Clear GDS reserved memory */
-	r = amdgpu_ring_alloc(ring, 17);
-	if (r) {
-		DRM_ERROR("amdgpu: NGG failed to lock ring %d (%d).\n",
-			  ring->idx, r);
-		return r;
-	}
-
-	gfx_v9_0_write_data_to_reg(ring, 0, false,
-				   amdgpu_gds_reg_offset[0].mem_size,
-			           (adev->gds.mem.total_size +
-				    adev->gfx.ngg.gds_reserve_size) >>
-				   AMDGPU_GDS_SHIFT);
-
-	amdgpu_ring_write(ring, PACKET3(PACKET3_DMA_DATA, 5));
-	amdgpu_ring_write(ring, (PACKET3_DMA_DATA_CP_SYNC |
-				PACKET3_DMA_DATA_SRC_SEL(2)));
-	amdgpu_ring_write(ring, 0);
-	amdgpu_ring_write(ring, 0);
-	amdgpu_ring_write(ring, adev->gfx.ngg.gds_reserve_addr);
-	amdgpu_ring_write(ring, 0);
-	amdgpu_ring_write(ring, adev->gfx.ngg.gds_reserve_size);
-
-
-	gfx_v9_0_write_data_to_reg(ring, 0, false,
-				   amdgpu_gds_reg_offset[0].mem_size, 0);
-
-	amdgpu_ring_commit(ring);
 
 	return 0;
 }
@@ -1206,9 +2152,9 @@ static int gfx_v9_0_ngg_en(struct amdgpu_device *adev)
 static int gfx_v9_0_compute_ring_init(struct amdgpu_device *adev, int ring_id,
 				      int mec, int pipe, int queue)
 {
-	int r;
 	unsigned irq_type;
 	struct amdgpu_ring *ring = &adev->gfx.compute_ring[ring_id];
+	unsigned int hw_prio;
 
 	ring = &adev->gfx.compute_ring[ring_id];
 
@@ -1219,35 +2165,67 @@ static int gfx_v9_0_compute_ring_init(struct amdgpu_device *adev, int ring_id,
 
 	ring->ring_obj = NULL;
 	ring->use_doorbell = true;
-	ring->doorbell_index = (AMDGPU_DOORBELL_MEC_RING0 + ring_id) << 1;
+	ring->doorbell_index = (adev->doorbell_index.mec_ring0 + ring_id) << 1;
 	ring->eop_gpu_addr = adev->gfx.mec.hpd_eop_gpu_addr
 				+ (ring_id * GFX9_MEC_HPD_SIZE);
+	ring->vm_hub = AMDGPU_GFXHUB(0);
 	sprintf(ring->name, "comp_%d.%d.%d", ring->me, ring->pipe, ring->queue);
 
 	irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP
 		+ ((ring->me - 1) * adev->gfx.mec.num_pipe_per_mec)
 		+ ring->pipe;
-
+	hw_prio = amdgpu_gfx_is_high_priority_compute_queue(adev, ring) ?
+			AMDGPU_RING_PRIO_2 : AMDGPU_RING_PRIO_DEFAULT;
 	/* type-2 packets are deprecated on MEC, use type-3 instead */
-	r = amdgpu_ring_init(adev, ring, 1024,
-			     &adev->gfx.eop_irq, irq_type);
-	if (r)
-		return r;
+	return amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq, irq_type,
+				hw_prio, NULL);
+}
 
+static void gfx_v9_0_alloc_ip_dump(struct amdgpu_device *adev)
+{
+	uint32_t reg_count = ARRAY_SIZE(gc_reg_list_9);
+	uint32_t *ptr;
+	uint32_t inst;
 
-	return 0;
+	ptr = kcalloc(reg_count, sizeof(uint32_t), GFP_KERNEL);
+	if (!ptr) {
+		DRM_ERROR("Failed to allocate memory for GFX IP Dump\n");
+		adev->gfx.ip_dump_core = NULL;
+	} else {
+		adev->gfx.ip_dump_core = ptr;
+	}
+
+	/* Allocate memory for compute queue registers for all the instances */
+	reg_count = ARRAY_SIZE(gc_cp_reg_list_9);
+	inst = adev->gfx.mec.num_mec * adev->gfx.mec.num_pipe_per_mec *
+		adev->gfx.mec.num_queue_per_pipe;
+
+	ptr = kcalloc(reg_count * inst, sizeof(uint32_t), GFP_KERNEL);
+	if (!ptr) {
+		DRM_ERROR("Failed to allocate memory for Compute Queues IP Dump\n");
+		adev->gfx.ip_dump_compute_queues = NULL;
+	} else {
+		adev->gfx.ip_dump_compute_queues = ptr;
+	}
 }
 
-static int gfx_v9_0_sw_init(void *handle)
+static int gfx_v9_0_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	int i, j, k, r, ring_id;
+	int xcc_id = 0;
 	struct amdgpu_ring *ring;
-	struct amdgpu_kiq *kiq;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	switch (adev->asic_type) {
-	case CHIP_VEGA10:
-	case CHIP_RAVEN:
+	struct amdgpu_device *adev = ip_block->adev;
+	unsigned int hw_prio;
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 0, 1):
+	case IP_VERSION(9, 2, 1):
+	case IP_VERSION(9, 4, 0):
+	case IP_VERSION(9, 2, 2):
+	case IP_VERSION(9, 1, 0):
+	case IP_VERSION(9, 4, 1):
+	case IP_VERSION(9, 3, 0):
+	case IP_VERSION(9, 4, 2):
 		adev->gfx.mec.num_mec = 2;
 		break;
 	default:
@@ -1255,45 +2233,92 @@ static int gfx_v9_0_sw_init(void *handle)
 		break;
 	}
 
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 0, 1):
+	case IP_VERSION(9, 2, 1):
+	case IP_VERSION(9, 4, 0):
+	case IP_VERSION(9, 2, 2):
+	case IP_VERSION(9, 1, 0):
+	case IP_VERSION(9, 3, 0):
+		adev->gfx.cleaner_shader_ptr = gfx_9_4_2_cleaner_shader_hex;
+		adev->gfx.cleaner_shader_size = sizeof(gfx_9_4_2_cleaner_shader_hex);
+		if (adev->gfx.me_fw_version  >= 167 &&
+		    adev->gfx.pfp_fw_version >= 196 &&
+		    adev->gfx.mec_fw_version >= 474) {
+			adev->gfx.enable_cleaner_shader = true;
+			r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);
+			if (r) {
+				adev->gfx.enable_cleaner_shader = false;
+				dev_err(adev->dev, "Failed to initialize cleaner shader\n");
+			}
+		}
+		break;
+	case IP_VERSION(9, 4, 2):
+		adev->gfx.cleaner_shader_ptr = gfx_9_4_2_cleaner_shader_hex;
+		adev->gfx.cleaner_shader_size = sizeof(gfx_9_4_2_cleaner_shader_hex);
+		if (adev->gfx.mec_fw_version >= 88) {
+			adev->gfx.enable_cleaner_shader = true;
+			r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);
+			if (r) {
+				adev->gfx.enable_cleaner_shader = false;
+				dev_err(adev->dev, "Failed to initialize cleaner shader\n");
+			}
+		}
+		break;
+	default:
+		adev->gfx.enable_cleaner_shader = false;
+		break;
+	}
+
 	adev->gfx.mec.num_pipe_per_mec = 4;
 	adev->gfx.mec.num_queue_per_pipe = 8;
 
-	/* KIQ event */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_GRBM_CP, 178, &adev->gfx.kiq.irq);
+	/* EOP Event */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_GRBM_CP, GFX_9_0__SRCID__CP_EOP_INTERRUPT, &adev->gfx.eop_irq);
 	if (r)
 		return r;
 
-	/* EOP Event */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_GRBM_CP, 181, &adev->gfx.eop_irq);
+	/* Bad opcode Event */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_GRBM_CP,
+			      GFX_9_0__SRCID__CP_BAD_OPCODE_ERROR,
+			      &adev->gfx.bad_op_irq);
 	if (r)
 		return r;
 
 	/* Privileged reg */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_GRBM_CP, 184,
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_GRBM_CP, GFX_9_0__SRCID__CP_PRIV_REG_FAULT,
 			      &adev->gfx.priv_reg_irq);
 	if (r)
 		return r;
 
 	/* Privileged inst */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_GRBM_CP, 185,
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_GRBM_CP, GFX_9_0__SRCID__CP_PRIV_INSTR_FAULT,
 			      &adev->gfx.priv_inst_irq);
 	if (r)
 		return r;
 
-	adev->gfx.gfx_current_status = AMDGPU_GFX_NORMAL_MODE;
-
-	gfx_v9_0_scratch_init(adev);
-
-	r = gfx_v9_0_init_microcode(adev);
-	if (r) {
-		DRM_ERROR("Failed to load gfx firmware!\n");
+	/* ECC error */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_GRBM_CP, GFX_9_0__SRCID__CP_ECC_ERROR,
+			      &adev->gfx.cp_ecc_error_irq);
+	if (r)
 		return r;
-	}
 
-	r = gfx_v9_0_rlc_init(adev);
-	if (r) {
-		DRM_ERROR("Failed to init rlc BOs!\n");
+	/* FUE error */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_GRBM_CP, GFX_9_0__SRCID__CP_FUE_ERROR,
+			      &adev->gfx.cp_ecc_error_irq);
+	if (r)
 		return r;
+
+	adev->gfx.gfx_current_status = AMDGPU_GFX_NORMAL_MODE;
+
+	if (adev->gfx.rlc.funcs) {
+		if (adev->gfx.rlc.funcs->init) {
+			r = adev->gfx.rlc.funcs->init(adev);
+			if (r) {
+				dev_err(adev->dev, "Failed to init rlc BOs!\n");
+				return r;
+			}
+		}
 	}
 
 	r = gfx_v9_0_mec_init(adev);
@@ -1306,21 +2331,66 @@ static int gfx_v9_0_sw_init(void *handle)
 	for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
 		ring = &adev->gfx.gfx_ring[i];
 		ring->ring_obj = NULL;
-		sprintf(ring->name, "gfx");
+		if (!i)
+			sprintf(ring->name, "gfx");
+		else
+			sprintf(ring->name, "gfx_%d", i);
 		ring->use_doorbell = true;
-		ring->doorbell_index = AMDGPU_DOORBELL64_GFX_RING0 << 1;
-		r = amdgpu_ring_init(adev, ring, 1024,
-				     &adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_EOP);
+		ring->doorbell_index = adev->doorbell_index.gfx_ring0 << 1;
+
+		/* disable scheduler on the real ring */
+		ring->no_scheduler = adev->gfx.mcbp;
+		ring->vm_hub = AMDGPU_GFXHUB(0);
+		r = amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq,
+				     AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP,
+				     AMDGPU_RING_PRIO_DEFAULT, NULL);
 		if (r)
 			return r;
 	}
 
+	/* set up the software rings */
+	if (adev->gfx.mcbp && adev->gfx.num_gfx_rings) {
+		for (i = 0; i < GFX9_NUM_SW_GFX_RINGS; i++) {
+			ring = &adev->gfx.sw_gfx_ring[i];
+			ring->ring_obj = NULL;
+			sprintf(ring->name, amdgpu_sw_ring_name(i));
+			ring->use_doorbell = true;
+			ring->doorbell_index = adev->doorbell_index.gfx_ring0 << 1;
+			ring->is_sw_ring = true;
+			hw_prio = amdgpu_sw_ring_priority(i);
+			ring->vm_hub = AMDGPU_GFXHUB(0);
+			r = amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq,
+					     AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP, hw_prio,
+					     NULL);
+			if (r)
+				return r;
+			ring->wptr = 0;
+		}
+
+		/* init the muxer and add software rings */
+		r = amdgpu_ring_mux_init(&adev->gfx.muxer, &adev->gfx.gfx_ring[0],
+					 GFX9_NUM_SW_GFX_RINGS);
+		if (r) {
+			DRM_ERROR("amdgpu_ring_mux_init failed(%d)\n", r);
+			return r;
+		}
+		for (i = 0; i < GFX9_NUM_SW_GFX_RINGS; i++) {
+			r = amdgpu_ring_mux_add_sw_ring(&adev->gfx.muxer,
+							&adev->gfx.sw_gfx_ring[i]);
+			if (r) {
+				DRM_ERROR("amdgpu_ring_mux_add_sw_ring failed(%d)\n", r);
+				return r;
+			}
+		}
+	}
+
 	/* set up the compute queues - allocate horizontally across pipes */
 	ring_id = 0;
 	for (i = 0; i < adev->gfx.mec.num_mec; ++i) {
 		for (j = 0; j < adev->gfx.mec.num_queue_per_pipe; j++) {
 			for (k = 0; k < adev->gfx.mec.num_pipe_per_mec; k++) {
-				if (!amdgpu_gfx_is_mec_queue_enabled(adev, i, k, j))
+				if (!amdgpu_gfx_is_mec_queue_enabled(adev, 0, i,
+								     k, j))
 					continue;
 
 				r = gfx_v9_0_compute_ring_init(adev,
@@ -1334,46 +2404,43 @@ static int gfx_v9_0_sw_init(void *handle)
 		}
 	}
 
-	r = amdgpu_gfx_kiq_init(adev, GFX9_MEC_HPD_SIZE);
+	/* TODO: Add queue reset mask when FW fully supports it */
+	adev->gfx.gfx_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.gfx_ring[0]);
+	adev->gfx.compute_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.compute_ring[0]);
+	if (!amdgpu_sriov_vf(adev))
+		adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+
+	r = amdgpu_gfx_kiq_init(adev, GFX9_MEC_HPD_SIZE, 0);
 	if (r) {
 		DRM_ERROR("Failed to init KIQ BOs!\n");
 		return r;
 	}
 
-	kiq = &adev->gfx.kiq;
-	r = amdgpu_gfx_kiq_init_ring(adev, &kiq->ring, &kiq->irq);
+	r = amdgpu_gfx_kiq_init_ring(adev, xcc_id);
 	if (r)
 		return r;
 
 	/* create MQD for all compute queues as wel as KIQ for SRIOV case */
-	r = amdgpu_gfx_compute_mqd_sw_init(adev, sizeof(struct v9_mqd));
-	if (r)
-		return r;
-
-	/* reserve GDS, GWS and OA resource for gfx */
-	r = amdgpu_bo_create_kernel(adev, adev->gds.mem.gfx_partition_size,
-				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GDS,
-				    &adev->gds.gds_gfx_bo, NULL, NULL);
+	r = amdgpu_gfx_mqd_sw_init(adev, sizeof(struct v9_mqd_allocation), 0);
 	if (r)
 		return r;
 
-	r = amdgpu_bo_create_kernel(adev, adev->gds.gws.gfx_partition_size,
-				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GWS,
-				    &adev->gds.gws_gfx_bo, NULL, NULL);
-	if (r)
-		return r;
+	adev->gfx.ce_ram_size = 0x8000;
 
-	r = amdgpu_bo_create_kernel(adev, adev->gds.oa.gfx_partition_size,
-				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_OA,
-				    &adev->gds.oa_gfx_bo, NULL, NULL);
+	r = gfx_v9_0_gpu_early_init(adev);
 	if (r)
 		return r;
 
-	adev->gfx.ce_ram_size = 0x8000;
+	if (amdgpu_gfx_ras_sw_init(adev)) {
+		dev_err(adev->dev, "Failed to initialize gfx ras block!\n");
+		return -EINVAL;
+	}
 
-	gfx_v9_0_gpu_early_init(adev);
+	gfx_v9_0_alloc_ip_dump(adev);
 
-	r = gfx_v9_0_ngg_init(adev);
+	r = amdgpu_gfx_sysfs_init(adev);
 	if (r)
 		return r;
 
@@ -1381,26 +2448,43 @@ static int gfx_v9_0_sw_init(void *handle)
 }
 
 
-static int gfx_v9_0_sw_fini(void *handle)
+static int gfx_v9_0_sw_fini(struct amdgpu_ip_block *ip_block)
 {
 	int i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	amdgpu_bo_free_kernel(&adev->gds.oa_gfx_bo, NULL, NULL);
-	amdgpu_bo_free_kernel(&adev->gds.gws_gfx_bo, NULL, NULL);
-	amdgpu_bo_free_kernel(&adev->gds.gds_gfx_bo, NULL, NULL);
+	if (adev->gfx.mcbp && adev->gfx.num_gfx_rings) {
+		for (i = 0; i < GFX9_NUM_SW_GFX_RINGS; i++)
+			amdgpu_ring_fini(&adev->gfx.sw_gfx_ring[i]);
+		amdgpu_ring_mux_fini(&adev->gfx.muxer);
+	}
 
 	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
 		amdgpu_ring_fini(&adev->gfx.gfx_ring[i]);
 	for (i = 0; i < adev->gfx.num_compute_rings; i++)
 		amdgpu_ring_fini(&adev->gfx.compute_ring[i]);
 
-	amdgpu_gfx_compute_mqd_sw_fini(adev);
-	amdgpu_gfx_kiq_free_ring(&adev->gfx.kiq.ring, &adev->gfx.kiq.irq);
-	amdgpu_gfx_kiq_fini(adev);
+	amdgpu_gfx_mqd_sw_fini(adev, 0);
+	amdgpu_gfx_kiq_free_ring(&adev->gfx.kiq[0].ring);
+	amdgpu_gfx_kiq_fini(adev, 0);
+
+	amdgpu_gfx_cleaner_shader_sw_fini(adev);
 
 	gfx_v9_0_mec_fini(adev);
-	gfx_v9_0_ngg_fini(adev);
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj,
+				&adev->gfx.rlc.clear_state_gpu_addr,
+				(void **)&adev->gfx.rlc.cs_ptr);
+	if (adev->flags & AMD_IS_APU) {
+		amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj,
+				&adev->gfx.rlc.cp_table_gpu_addr,
+				(void **)&adev->gfx.rlc.cp_table_ptr);
+	}
+	gfx_v9_0_free_microcode(adev);
+
+	amdgpu_gfx_sysfs_fini(adev);
+
+	kfree(adev->gfx.ip_dump_core);
+	kfree(adev->gfx.ip_dump_compute_queues);
 
 	return 0;
 }
@@ -1411,7 +2495,8 @@ static void gfx_v9_0_tiling_mode_table_init(struct amdgpu_device *adev)
 	/* TODO */
 }
 
-static void gfx_v9_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num, u32 instance)
+void gfx_v9_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num,
+			   u32 instance, int xcc_id)
 {
 	u32 data;
 
@@ -1430,7 +2515,7 @@ static void gfx_v9_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh
 	else
 		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_INDEX, sh_num);
 
-	WREG32_SOC15(GC, 0, mmGRBM_GFX_INDEX, data);
+	WREG32_SOC15_RLC_SHADOW(GC, 0, mmGRBM_GFX_INDEX, data);
 }
 
 static u32 gfx_v9_0_get_rb_active_bitmap(struct amdgpu_device *adev)
@@ -1460,22 +2545,43 @@ static void gfx_v9_0_setup_rb(struct amdgpu_device *adev)
 	mutex_lock(&adev->grbm_idx_mutex);
 	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
 		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
-			gfx_v9_0_select_se_sh(adev, i, j, 0xffffffff);
+			amdgpu_gfx_select_se_sh(adev, i, j, 0xffffffff, 0);
 			data = gfx_v9_0_get_rb_active_bitmap(adev);
 			active_rbs |= data << ((i * adev->gfx.config.max_sh_per_se + j) *
 					       rb_bitmap_width_per_sh);
 		}
 	}
-	gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 	mutex_unlock(&adev->grbm_idx_mutex);
 
 	adev->gfx.config.backend_enable_mask = active_rbs;
 	adev->gfx.config.num_rbs = hweight32(active_rbs);
 }
 
+static void gfx_v9_0_debug_trap_config_init(struct amdgpu_device *adev,
+				uint32_t first_vmid,
+				uint32_t last_vmid)
+{
+	uint32_t data;
+	uint32_t trap_config_vmid_mask = 0;
+	int i;
+
+	/* Calculate trap config vmid mask */
+	for (i = first_vmid; i < last_vmid; i++)
+		trap_config_vmid_mask |= (1 << i);
+
+	data = REG_SET_FIELD(0, SPI_GDBG_TRAP_CONFIG,
+			VMID_SEL, trap_config_vmid_mask);
+	data = REG_SET_FIELD(data, SPI_GDBG_TRAP_CONFIG,
+			TRAP_EN, 1);
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_TRAP_CONFIG), data);
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_TRAP_MASK), 0);
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_TRAP_DATA0), 0);
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_TRAP_DATA1), 0);
+}
+
 #define DEFAULT_SH_MEM_BASES	(0x6000)
-#define FIRST_COMPUTE_VMID	(8)
-#define LAST_COMPUTE_VMID	(16)
 static void gfx_v9_0_init_compute_vmid(struct amdgpu_device *adev)
 {
 	int i;
@@ -1495,64 +2601,112 @@ static void gfx_v9_0_init_compute_vmid(struct amdgpu_device *adev)
 			SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT;
 
 	mutex_lock(&adev->srbm_mutex);
-	for (i = FIRST_COMPUTE_VMID; i < LAST_COMPUTE_VMID; i++) {
-		soc15_grbm_select(adev, 0, 0, 0, i);
+	for (i = adev->vm_manager.first_kfd_vmid; i < AMDGPU_NUM_VMID; i++) {
+		soc15_grbm_select(adev, 0, 0, 0, i, 0);
 		/* CP and shaders */
-		WREG32_SOC15(GC, 0, mmSH_MEM_CONFIG, sh_mem_config);
-		WREG32_SOC15(GC, 0, mmSH_MEM_BASES, sh_mem_bases);
+		WREG32_SOC15_RLC(GC, 0, mmSH_MEM_CONFIG, sh_mem_config);
+		WREG32_SOC15_RLC(GC, 0, mmSH_MEM_BASES, sh_mem_bases);
 	}
-	soc15_grbm_select(adev, 0, 0, 0, 0);
+	soc15_grbm_select(adev, 0, 0, 0, 0, 0);
 	mutex_unlock(&adev->srbm_mutex);
+
+	/* Initialize all compute VMIDs to have no GDS, GWS, or OA
+	   access. These should be enabled by FW for target VMIDs. */
+	for (i = adev->vm_manager.first_kfd_vmid; i < AMDGPU_NUM_VMID; i++) {
+		WREG32_SOC15_OFFSET(GC, 0, mmGDS_VMID0_BASE, 2 * i, 0);
+		WREG32_SOC15_OFFSET(GC, 0, mmGDS_VMID0_SIZE, 2 * i, 0);
+		WREG32_SOC15_OFFSET(GC, 0, mmGDS_GWS_VMID0, i, 0);
+		WREG32_SOC15_OFFSET(GC, 0, mmGDS_OA_VMID0, i, 0);
+	}
+}
+
+static void gfx_v9_0_init_gds_vmid(struct amdgpu_device *adev)
+{
+	int vmid;
+
+	/*
+	 * Initialize all compute and user-gfx VMIDs to have no GDS, GWS, or OA
+	 * access. Compute VMIDs should be enabled by FW for target VMIDs,
+	 * the driver can enable them for graphics. VMID0 should maintain
+	 * access so that HWS firmware can save/restore entries.
+	 */
+	for (vmid = 1; vmid < AMDGPU_NUM_VMID; vmid++) {
+		WREG32_SOC15_OFFSET(GC, 0, mmGDS_VMID0_BASE, 2 * vmid, 0);
+		WREG32_SOC15_OFFSET(GC, 0, mmGDS_VMID0_SIZE, 2 * vmid, 0);
+		WREG32_SOC15_OFFSET(GC, 0, mmGDS_GWS_VMID0, vmid, 0);
+		WREG32_SOC15_OFFSET(GC, 0, mmGDS_OA_VMID0, vmid, 0);
+	}
+}
+
+static void gfx_v9_0_init_sq_config(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 4, 1):
+		tmp = RREG32_SOC15(GC, 0, mmSQ_CONFIG);
+		tmp = REG_SET_FIELD(tmp, SQ_CONFIG, DISABLE_BARRIER_WAITCNT,
+				!READ_ONCE(adev->barrier_has_auto_waitcnt));
+		WREG32_SOC15(GC, 0, mmSQ_CONFIG, tmp);
+		break;
+	case IP_VERSION(9, 4, 2):
+		gfx_v9_4_2_init_sq(adev);
+		break;
+	default:
+		break;
+	}
 }
 
-static void gfx_v9_0_gpu_init(struct amdgpu_device *adev)
+static void gfx_v9_0_constants_init(struct amdgpu_device *adev)
 {
 	u32 tmp;
 	int i;
 
-	WREG32_FIELD15(GC, 0, GRBM_CNTL, READ_TIMEOUT, 0xff);
+	if (!amdgpu_sriov_vf(adev) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 2)) {
+		WREG32_FIELD15_RLC(GC, 0, GRBM_CNTL, READ_TIMEOUT, 0xff);
+	}
 
 	gfx_v9_0_tiling_mode_table_init(adev);
 
-	gfx_v9_0_setup_rb(adev);
+	if (adev->gfx.num_gfx_rings)
+		gfx_v9_0_setup_rb(adev);
 	gfx_v9_0_get_cu_info(adev, &adev->gfx.cu_info);
+	adev->gfx.config.db_debug2 = RREG32_SOC15(GC, 0, mmDB_DEBUG2);
 
 	/* XXX SH_MEM regs */
 	/* where to put LDS, scratch, GPUVM in FSA64 space */
 	mutex_lock(&adev->srbm_mutex);
-	for (i = 0; i < 16; i++) {
-		soc15_grbm_select(adev, 0, 0, 0, i);
+	for (i = 0; i < adev->vm_manager.id_mgr[AMDGPU_GFXHUB(0)].num_ids; i++) {
+		soc15_grbm_select(adev, 0, 0, 0, i, 0);
 		/* CP and shaders */
-		tmp = 0;
-		tmp = REG_SET_FIELD(tmp, SH_MEM_CONFIG, ALIGNMENT_MODE,
-				    SH_MEM_ALIGNMENT_MODE_UNALIGNED);
-		WREG32_SOC15(GC, 0, mmSH_MEM_CONFIG, tmp);
-		WREG32_SOC15(GC, 0, mmSH_MEM_BASES, 0);
+		if (i == 0) {
+			tmp = REG_SET_FIELD(0, SH_MEM_CONFIG, ALIGNMENT_MODE,
+					    SH_MEM_ALIGNMENT_MODE_UNALIGNED);
+			tmp = REG_SET_FIELD(tmp, SH_MEM_CONFIG, RETRY_DISABLE,
+					    !!adev->gmc.noretry);
+			WREG32_SOC15_RLC(GC, 0, mmSH_MEM_CONFIG, tmp);
+			WREG32_SOC15_RLC(GC, 0, mmSH_MEM_BASES, 0);
+		} else {
+			tmp = REG_SET_FIELD(0, SH_MEM_CONFIG, ALIGNMENT_MODE,
+					    SH_MEM_ALIGNMENT_MODE_UNALIGNED);
+			tmp = REG_SET_FIELD(tmp, SH_MEM_CONFIG, RETRY_DISABLE,
+					    !!adev->gmc.noretry);
+			WREG32_SOC15_RLC(GC, 0, mmSH_MEM_CONFIG, tmp);
+			tmp = REG_SET_FIELD(0, SH_MEM_BASES, PRIVATE_BASE,
+				(adev->gmc.private_aperture_start >> 48));
+			tmp = REG_SET_FIELD(tmp, SH_MEM_BASES, SHARED_BASE,
+				(adev->gmc.shared_aperture_start >> 48));
+			WREG32_SOC15_RLC(GC, 0, mmSH_MEM_BASES, tmp);
+		}
 	}
-	soc15_grbm_select(adev, 0, 0, 0, 0);
+	soc15_grbm_select(adev, 0, 0, 0, 0, 0);
 
 	mutex_unlock(&adev->srbm_mutex);
 
 	gfx_v9_0_init_compute_vmid(adev);
-
-	mutex_lock(&adev->grbm_idx_mutex);
-	/*
-	 * making sure that the following register writes will be broadcasted
-	 * to all the shaders
-	 */
-	gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
-
-	WREG32_SOC15(GC, 0, mmPA_SC_FIFO_SIZE,
-		   (adev->gfx.config.sc_prim_fifo_size_frontend <<
-			PA_SC_FIFO_SIZE__SC_FRONTEND_PRIM_FIFO_SIZE__SHIFT) |
-		   (adev->gfx.config.sc_prim_fifo_size_backend <<
-			PA_SC_FIFO_SIZE__SC_BACKEND_PRIM_FIFO_SIZE__SHIFT) |
-		   (adev->gfx.config.sc_hiz_tile_fifo_size <<
-			PA_SC_FIFO_SIZE__SC_HIZ_TILE_FIFO_SIZE__SHIFT) |
-		   (adev->gfx.config.sc_earlyz_tile_fifo_size <<
-			PA_SC_FIFO_SIZE__SC_EARLYZ_TILE_FIFO_SIZE__SHIFT));
-	mutex_unlock(&adev->grbm_idx_mutex);
-
+	gfx_v9_0_init_gds_vmid(adev);
+	gfx_v9_0_init_sq_config(adev);
 }
 
 static void gfx_v9_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
@@ -1563,15 +2717,23 @@ static void gfx_v9_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
 	mutex_lock(&adev->grbm_idx_mutex);
 	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
 		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
-			gfx_v9_0_select_se_sh(adev, i, j, 0xffffffff);
+			amdgpu_gfx_select_se_sh(adev, i, j, 0xffffffff, 0);
 			for (k = 0; k < adev->usec_timeout; k++) {
 				if (RREG32_SOC15(GC, 0, mmRLC_SERDES_CU_MASTER_BUSY) == 0)
 					break;
 				udelay(1);
 			}
+			if (k == adev->usec_timeout) {
+				amdgpu_gfx_select_se_sh(adev, 0xffffffff,
+						      0xffffffff, 0xffffffff, 0);
+				mutex_unlock(&adev->grbm_idx_mutex);
+				DRM_INFO("Timeout wait for RLC serdes %u,%u\n",
+					 i, j);
+				return;
+			}
 		}
 	}
-	gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 	mutex_unlock(&adev->grbm_idx_mutex);
 
 	mask = RLC_SERDES_NONCU_MASTER_BUSY__SE_MASTER_BUSY_MASK |
@@ -1588,76 +2750,71 @@ static void gfx_v9_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
 static void gfx_v9_0_enable_gui_idle_interrupt(struct amdgpu_device *adev,
 					       bool enable)
 {
-	u32 tmp = RREG32_SOC15(GC, 0, mmCP_INT_CNTL_RING0);
+	u32 tmp;
+
+	/* These interrupts should be enabled to drive DS clock */
+
+	tmp= RREG32_SOC15(GC, 0, mmCP_INT_CNTL_RING0);
 
 	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_BUSY_INT_ENABLE, enable ? 1 : 0);
 	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_EMPTY_INT_ENABLE, enable ? 1 : 0);
 	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CMP_BUSY_INT_ENABLE, enable ? 1 : 0);
-	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, GFX_IDLE_INT_ENABLE, enable ? 1 : 0);
+	if (adev->gfx.num_gfx_rings)
+		tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, GFX_IDLE_INT_ENABLE, enable ? 1 : 0);
 
 	WREG32_SOC15(GC, 0, mmCP_INT_CNTL_RING0, tmp);
 }
 
 static void gfx_v9_0_init_csb(struct amdgpu_device *adev)
 {
+	adev->gfx.rlc.funcs->get_csb_buffer(adev, adev->gfx.rlc.cs_ptr);
 	/* csib */
-	WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_CSIB_ADDR_HI),
+	WREG32_RLC(SOC15_REG_OFFSET(GC, 0, mmRLC_CSIB_ADDR_HI),
 			adev->gfx.rlc.clear_state_gpu_addr >> 32);
-	WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_CSIB_ADDR_LO),
+	WREG32_RLC(SOC15_REG_OFFSET(GC, 0, mmRLC_CSIB_ADDR_LO),
 			adev->gfx.rlc.clear_state_gpu_addr & 0xfffffffc);
-	WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_CSIB_LENGTH),
+	WREG32_RLC(SOC15_REG_OFFSET(GC, 0, mmRLC_CSIB_LENGTH),
 			adev->gfx.rlc.clear_state_size);
 }
 
-static void gfx_v9_0_parse_ind_reg_list(int *register_list_format,
+static void gfx_v9_1_parse_ind_reg_list(int *register_list_format,
 				int indirect_offset,
 				int list_size,
 				int *unique_indirect_regs,
-				int *unique_indirect_reg_count,
-				int max_indirect_reg_count,
+				int unique_indirect_reg_count,
 				int *indirect_start_offsets,
 				int *indirect_start_offsets_count,
-				int max_indirect_start_offsets_count)
+				int max_start_offsets_count)
 {
 	int idx;
-	bool new_entry = true;
 
 	for (; indirect_offset < list_size; indirect_offset++) {
+		WARN_ON(*indirect_start_offsets_count >= max_start_offsets_count);
+		indirect_start_offsets[*indirect_start_offsets_count] = indirect_offset;
+		*indirect_start_offsets_count = *indirect_start_offsets_count + 1;
+
+		while (register_list_format[indirect_offset] != 0xFFFFFFFF) {
+			indirect_offset += 2;
+
+			/* look for the matching indice */
+			for (idx = 0; idx < unique_indirect_reg_count; idx++) {
+				if (unique_indirect_regs[idx] ==
+					register_list_format[indirect_offset] ||
+					!unique_indirect_regs[idx])
+					break;
+			}
 
-		if (new_entry) {
-			new_entry = false;
-			indirect_start_offsets[*indirect_start_offsets_count] = indirect_offset;
-			*indirect_start_offsets_count = *indirect_start_offsets_count + 1;
-			BUG_ON(*indirect_start_offsets_count >= max_indirect_start_offsets_count);
-		}
-
-		if (register_list_format[indirect_offset] == 0xFFFFFFFF) {
-			new_entry = true;
-			continue;
-		}
-
-		indirect_offset += 2;
+			BUG_ON(idx >= unique_indirect_reg_count);
 
-		/* look for the matching indice */
-		for (idx = 0; idx < *unique_indirect_reg_count; idx++) {
-			if (unique_indirect_regs[idx] ==
-				register_list_format[indirect_offset])
-				break;
-		}
+			if (!unique_indirect_regs[idx])
+				unique_indirect_regs[idx] = register_list_format[indirect_offset];
 
-		if (idx >= *unique_indirect_reg_count) {
-			unique_indirect_regs[*unique_indirect_reg_count] =
-				register_list_format[indirect_offset];
-			idx = *unique_indirect_reg_count;
-			*unique_indirect_reg_count = *unique_indirect_reg_count + 1;
-			BUG_ON(*unique_indirect_reg_count >= max_indirect_reg_count);
+			indirect_offset++;
 		}
-
-		register_list_format[indirect_offset] = idx;
 	}
 }
 
-static int gfx_v9_0_init_rlc_save_restore_list(struct amdgpu_device *adev)
+static int gfx_v9_1_init_rlc_save_restore_list(struct amdgpu_device *adev)
 {
 	int unique_indirect_regs[] = {0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0};
 	int unique_indirect_reg_count = 0;
@@ -1666,26 +2823,25 @@ static int gfx_v9_0_init_rlc_save_restore_list(struct amdgpu_device *adev)
 	int indirect_start_offsets_count = 0;
 
 	int list_size = 0;
-	int i = 0;
+	int i = 0, j = 0;
 	u32 tmp = 0;
 
 	u32 *register_list_format =
-		kmalloc(adev->gfx.rlc.reg_list_format_size_bytes, GFP_KERNEL);
+		kmemdup(adev->gfx.rlc.register_list_format,
+			adev->gfx.rlc.reg_list_format_size_bytes, GFP_KERNEL);
 	if (!register_list_format)
 		return -ENOMEM;
-	memcpy(register_list_format, adev->gfx.rlc.register_list_format,
-		adev->gfx.rlc.reg_list_format_size_bytes);
 
 	/* setup unique_indirect_regs array and indirect_start_offsets array */
-	gfx_v9_0_parse_ind_reg_list(register_list_format,
-				GFX9_RLC_FORMAT_DIRECT_REG_LIST_LENGTH,
-				adev->gfx.rlc.reg_list_format_size_bytes >> 2,
-				unique_indirect_regs,
-				&unique_indirect_reg_count,
-				sizeof(unique_indirect_regs)/sizeof(int),
-				indirect_start_offsets,
-				&indirect_start_offsets_count,
-				sizeof(indirect_start_offsets)/sizeof(int));
+	unique_indirect_reg_count = ARRAY_SIZE(unique_indirect_regs);
+	gfx_v9_1_parse_ind_reg_list(register_list_format,
+				    adev->gfx.rlc.reg_list_format_direct_reg_list_length,
+				    adev->gfx.rlc.reg_list_format_size_bytes >> 2,
+				    unique_indirect_regs,
+				    unique_indirect_reg_count,
+				    indirect_start_offsets,
+				    &indirect_start_offsets_count,
+				    ARRAY_SIZE(indirect_start_offsets));
 
 	/* enable auto inc in case it is disabled */
 	tmp = RREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_SRM_CNTL));
@@ -1699,19 +2855,37 @@ static int gfx_v9_0_init_rlc_save_restore_list(struct amdgpu_device *adev)
 		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_SRM_ARAM_DATA),
 			adev->gfx.rlc.register_restore[i]);
 
-	/* load direct register */
-	WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_SRM_ARAM_ADDR), 0);
-	for (i = 0; i < adev->gfx.rlc.reg_list_size_bytes >> 2; i++)
-		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_SRM_ARAM_DATA),
-			adev->gfx.rlc.register_restore[i]);
-
 	/* load indirect register */
 	WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_SCRATCH_ADDR),
 		adev->gfx.rlc.reg_list_format_start);
-	for (i = 0; i < adev->gfx.rlc.reg_list_format_size_bytes >> 2; i++)
+
+	/* direct register portion */
+	for (i = 0; i < adev->gfx.rlc.reg_list_format_direct_reg_list_length; i++)
 		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_SCRATCH_DATA),
 			register_list_format[i]);
 
+	/* indirect register portion */
+	while (i < (adev->gfx.rlc.reg_list_format_size_bytes >> 2)) {
+		if (register_list_format[i] == 0xFFFFFFFF) {
+			WREG32_SOC15(GC, 0, mmRLC_GPM_SCRATCH_DATA, register_list_format[i++]);
+			continue;
+		}
+
+		WREG32_SOC15(GC, 0, mmRLC_GPM_SCRATCH_DATA, register_list_format[i++]);
+		WREG32_SOC15(GC, 0, mmRLC_GPM_SCRATCH_DATA, register_list_format[i++]);
+
+		for (j = 0; j < unique_indirect_reg_count; j++) {
+			if (register_list_format[i] == unique_indirect_regs[j]) {
+				WREG32_SOC15(GC, 0, mmRLC_GPM_SCRATCH_DATA, j);
+				break;
+			}
+		}
+
+		BUG_ON(j >= unique_indirect_reg_count);
+
+		i++;
+	}
+
 	/* set save/restore list size */
 	list_size = adev->gfx.rlc.reg_list_size_bytes >> 2;
 	list_size = list_size >> 1;
@@ -1722,16 +2896,21 @@ static int gfx_v9_0_init_rlc_save_restore_list(struct amdgpu_device *adev)
 	/* write the starting offsets to RLC scratch ram */
 	WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_SCRATCH_ADDR),
 		adev->gfx.rlc.starting_offsets_start);
-	for (i = 0; i < sizeof(indirect_start_offsets)/sizeof(int); i++)
+	for (i = 0; i < ARRAY_SIZE(indirect_start_offsets); i++)
 		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_SCRATCH_DATA),
-			indirect_start_offsets[i]);
+		       indirect_start_offsets[i]);
 
 	/* load unique indirect regs*/
-	for (i = 0; i < sizeof(unique_indirect_regs)/sizeof(int); i++) {
-		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_SRM_INDEX_CNTL_ADDR_0) + i,
-			unique_indirect_regs[i] & 0x3FFFF);
-		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_SRM_INDEX_CNTL_DATA_0) + i,
-			unique_indirect_regs[i] >> 20);
+	for (i = 0; i < ARRAY_SIZE(unique_indirect_regs); i++) {
+		if (unique_indirect_regs[i] != 0) {
+			WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_SRM_INDEX_CNTL_ADDR_0)
+			       + GFX_RLC_SRM_INDEX_CNTL_ADDR_OFFSETS[i],
+			       unique_indirect_regs[i] & 0x3FFFF);
+
+			WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_SRM_INDEX_CNTL_DATA_0)
+			       + GFX_RLC_SRM_INDEX_CNTL_DATA_OFFSETS[i],
+			       unique_indirect_regs[i] >> 20);
+		}
 	}
 
 	kfree(register_list_format);
@@ -1740,11 +2919,7 @@ static int gfx_v9_0_init_rlc_save_restore_list(struct amdgpu_device *adev)
 
 static void gfx_v9_0_enable_save_restore_machine(struct amdgpu_device *adev)
 {
-	u32 tmp = 0;
-
-	tmp = RREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_SRM_CNTL));
-	tmp |= RLC_SRM_CNTL__SRM_ENABLE_MASK;
-	WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_SRM_CNTL), tmp);
+	WREG32_FIELD15(GC, 0, RLC_SRM_CNTL, SRM_ENABLE, 1);
 }
 
 static void pwr_10_0_gfxip_control_over_cgpg(struct amdgpu_device *adev,
@@ -1754,7 +2929,7 @@ static void pwr_10_0_gfxip_control_over_cgpg(struct amdgpu_device *adev,
 	uint32_t default_data = 0;
 
 	default_data = data = RREG32(SOC15_REG_OFFSET(PWR, 0, mmPWR_MISC_CNTL_STATUS));
-	if (enable == true) {
+	if (enable) {
 		/* enable GFXIP control over CGPG */
 		data |= PWR_MISC_CNTL_STATUS__PWR_GFX_RLC_CGPG_EN_MASK;
 		if(default_data != data)
@@ -1810,8 +2985,8 @@ static void gfx_v9_0_init_gfx_power_gating(struct amdgpu_device *adev)
 		/* program GRBM_REG_SAVE_GFX_IDLE_THRESHOLD to 0x55f0 */
 		data |= (0x55f0 << RLC_AUTO_PG_CTRL__GRBM_REG_SAVE_GFX_IDLE_THRESHOLD__SHIFT);
 		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_AUTO_PG_CTRL), data);
-
-		pwr_10_0_gfxip_control_over_cgpg(adev, true);
+		if (amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 3, 0))
+			pwr_10_0_gfxip_control_over_cgpg(adev, true);
 	}
 }
 
@@ -1822,16 +2997,11 @@ static void gfx_v9_0_enable_sck_slow_down_on_power_up(struct amdgpu_device *adev
 	uint32_t default_data = 0;
 
 	default_data = data = RREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL));
-
-	if (enable == true) {
-		data |= RLC_PG_CNTL__SMU_CLK_SLOWDOWN_ON_PU_ENABLE_MASK;
-		if (default_data != data)
-			WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL), data);
-	} else {
-		data &= ~RLC_PG_CNTL__SMU_CLK_SLOWDOWN_ON_PU_ENABLE_MASK;
-		if(default_data != data)
-			WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL), data);
-	}
+	data = REG_SET_FIELD(data, RLC_PG_CNTL,
+			     SMU_CLK_SLOWDOWN_ON_PU_ENABLE,
+			     enable ? 1 : 0);
+	if (default_data != data)
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL), data);
 }
 
 static void gfx_v9_0_enable_sck_slow_down_on_power_down(struct amdgpu_device *adev,
@@ -1841,16 +3011,11 @@ static void gfx_v9_0_enable_sck_slow_down_on_power_down(struct amdgpu_device *ad
 	uint32_t default_data = 0;
 
 	default_data = data = RREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL));
-
-	if (enable == true) {
-		data |= RLC_PG_CNTL__SMU_CLK_SLOWDOWN_ON_PD_ENABLE_MASK;
-		if(default_data != data)
-			WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL), data);
-	} else {
-		data &= ~RLC_PG_CNTL__SMU_CLK_SLOWDOWN_ON_PD_ENABLE_MASK;
-		if(default_data != data)
-			WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL), data);
-	}
+	data = REG_SET_FIELD(data, RLC_PG_CNTL,
+			     SMU_CLK_SLOWDOWN_ON_PD_ENABLE,
+			     enable ? 1 : 0);
+	if(default_data != data)
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL), data);
 }
 
 static void gfx_v9_0_enable_cp_power_gating(struct amdgpu_device *adev,
@@ -1860,16 +3025,11 @@ static void gfx_v9_0_enable_cp_power_gating(struct amdgpu_device *adev,
 	uint32_t default_data = 0;
 
 	default_data = data = RREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL));
-
-	if (enable == true) {
-		data &= ~RLC_PG_CNTL__CP_PG_DISABLE_MASK;
-		if(default_data != data)
-			WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL), data);
-	} else {
-		data |= RLC_PG_CNTL__CP_PG_DISABLE_MASK;
-		if(default_data != data)
-			WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL), data);
-	}
+	data = REG_SET_FIELD(data, RLC_PG_CNTL,
+			     CP_PG_DISABLE,
+			     enable ? 0 : 1);
+	if(default_data != data)
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL), data);
 }
 
 static void gfx_v9_0_enable_gfx_cg_power_gating(struct amdgpu_device *adev,
@@ -1878,10 +3038,9 @@ static void gfx_v9_0_enable_gfx_cg_power_gating(struct amdgpu_device *adev,
 	uint32_t data, default_data;
 
 	default_data = data = RREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL));
-	if (enable == true)
-		data |= RLC_PG_CNTL__GFX_POWER_GATING_ENABLE_MASK;
-	else
-		data &= ~RLC_PG_CNTL__GFX_POWER_GATING_ENABLE_MASK;
+	data = REG_SET_FIELD(data, RLC_PG_CNTL,
+			     GFX_POWER_GATING_ENABLE,
+			     enable ? 1 : 0);
 	if(default_data != data)
 		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL), data);
 }
@@ -1892,10 +3051,9 @@ static void gfx_v9_0_enable_gfx_pipeline_powergating(struct amdgpu_device *adev,
 	uint32_t data, default_data;
 
 	default_data = data = RREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL));
-	if (enable == true)
-		data |= RLC_PG_CNTL__GFX_PIPELINE_PG_ENABLE_MASK;
-	else
-		data &= ~RLC_PG_CNTL__GFX_PIPELINE_PG_ENABLE_MASK;
+	data = REG_SET_FIELD(data, RLC_PG_CNTL,
+			     GFX_PIPELINE_PG_ENABLE,
+			     enable ? 1 : 0);
 	if(default_data != data)
 		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL), data);
 
@@ -1910,10 +3068,9 @@ static void gfx_v9_0_enable_gfx_static_mg_power_gating(struct amdgpu_device *ade
 	uint32_t data, default_data;
 
 	default_data = data = RREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL));
-	if (enable == true)
-		data |= RLC_PG_CNTL__STATIC_PER_CU_PG_ENABLE_MASK;
-	else
-		data &= ~RLC_PG_CNTL__STATIC_PER_CU_PG_ENABLE_MASK;
+	data = REG_SET_FIELD(data, RLC_PG_CNTL,
+			     STATIC_PER_CU_PG_ENABLE,
+			     enable ? 1 : 0);
 	if(default_data != data)
 		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL), data);
 }
@@ -1924,56 +3081,45 @@ static void gfx_v9_0_enable_gfx_dynamic_mg_power_gating(struct amdgpu_device *ad
 	uint32_t data, default_data;
 
 	default_data = data = RREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL));
-	if (enable == true)
-		data |= RLC_PG_CNTL__DYN_PER_CU_PG_ENABLE_MASK;
-	else
-		data &= ~RLC_PG_CNTL__DYN_PER_CU_PG_ENABLE_MASK;
+	data = REG_SET_FIELD(data, RLC_PG_CNTL,
+			     DYN_PER_CU_PG_ENABLE,
+			     enable ? 1 : 0);
 	if(default_data != data)
 		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL), data);
 }
 
 static void gfx_v9_0_init_pg(struct amdgpu_device *adev)
 {
+	gfx_v9_0_init_csb(adev);
+
+	/*
+	 * Rlc save restore list is workable since v2_1.
+	 * And it's needed by gfxoff feature.
+	 */
+	if (adev->gfx.rlc.is_rlc_v2_1) {
+		if (amdgpu_ip_version(adev, GC_HWIP, 0) ==
+			    IP_VERSION(9, 2, 1) ||
+		    (adev->apu_flags & AMD_APU_IS_RAVEN2))
+			gfx_v9_1_init_rlc_save_restore_list(adev);
+		gfx_v9_0_enable_save_restore_machine(adev);
+	}
+
 	if (adev->pg_flags & (AMD_PG_SUPPORT_GFX_PG |
 			      AMD_PG_SUPPORT_GFX_SMG |
 			      AMD_PG_SUPPORT_GFX_DMG |
 			      AMD_PG_SUPPORT_CP |
 			      AMD_PG_SUPPORT_GDS |
 			      AMD_PG_SUPPORT_RLC_SMU_HS)) {
-		gfx_v9_0_init_csb(adev);
-		gfx_v9_0_init_rlc_save_restore_list(adev);
-		gfx_v9_0_enable_save_restore_machine(adev);
-
-		if (adev->asic_type == CHIP_RAVEN) {
-			WREG32(mmRLC_JUMP_TABLE_RESTORE,
-				adev->gfx.rlc.cp_table_gpu_addr >> 8);
-			gfx_v9_0_init_gfx_power_gating(adev);
-
-			if (adev->pg_flags & AMD_PG_SUPPORT_RLC_SMU_HS) {
-				gfx_v9_0_enable_sck_slow_down_on_power_up(adev, true);
-				gfx_v9_0_enable_sck_slow_down_on_power_down(adev, true);
-			} else {
-				gfx_v9_0_enable_sck_slow_down_on_power_up(adev, false);
-				gfx_v9_0_enable_sck_slow_down_on_power_down(adev, false);
-			}
-
-			if (adev->pg_flags & AMD_PG_SUPPORT_CP)
-				gfx_v9_0_enable_cp_power_gating(adev, true);
-			else
-				gfx_v9_0_enable_cp_power_gating(adev, false);
-		}
+		WREG32_SOC15(GC, 0, mmRLC_JUMP_TABLE_RESTORE,
+			     adev->gfx.rlc.cp_table_gpu_addr >> 8);
+		gfx_v9_0_init_gfx_power_gating(adev);
 	}
 }
 
-void gfx_v9_0_rlc_stop(struct amdgpu_device *adev)
+static void gfx_v9_0_rlc_stop(struct amdgpu_device *adev)
 {
-	u32 tmp = RREG32_SOC15(GC, 0, mmRLC_CNTL);
-
-	tmp = REG_SET_FIELD(tmp, RLC_CNTL, RLC_ENABLE_F32, 0);
-	WREG32_SOC15(GC, 0, mmRLC_CNTL, tmp);
-
+	WREG32_FIELD15(GC, 0, RLC_CNTL, RLC_ENABLE_F32, 0);
 	gfx_v9_0_enable_gui_idle_interrupt(adev, false);
-
 	gfx_v9_0_wait_for_rlc_serdes(adev);
 }
 
@@ -1992,12 +3138,13 @@ static void gfx_v9_0_rlc_start(struct amdgpu_device *adev)
 #endif
 
 	WREG32_FIELD15(GC, 0, RLC_CNTL, RLC_ENABLE_F32, 1);
+	udelay(50);
 
 	/* carrizo do enable cp interrupt after cp inited */
-	if (!(adev->flags & AMD_IS_APU))
+	if (!(adev->flags & AMD_IS_APU)) {
 		gfx_v9_0_enable_gui_idle_interrupt(adev, true);
-
-	udelay(50);
+		udelay(50);
+	}
 
 #ifdef AMDGPU_RLC_DEBUG_RETRY
 	/* RLC_GPM_GENERAL_6 : RLC Ucode version */
@@ -2045,19 +3192,16 @@ static int gfx_v9_0_rlc_resume(struct amdgpu_device *adev)
 {
 	int r;
 
-	if (amdgpu_sriov_vf(adev))
+	if (amdgpu_sriov_vf(adev)) {
+		gfx_v9_0_init_csb(adev);
 		return 0;
+	}
 
-	gfx_v9_0_rlc_stop(adev);
+	adev->gfx.rlc.funcs->stop(adev);
 
 	/* disable CG */
 	WREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL, 0);
 
-	/* disable PG */
-	WREG32_SOC15(GC, 0, mmRLC_PG_CNTL, 0);
-
-	gfx_v9_0_rlc_reset(adev);
-
 	gfx_v9_0_init_pg(adev);
 
 	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
@@ -2067,31 +3211,50 @@ static int gfx_v9_0_rlc_resume(struct amdgpu_device *adev)
 			return r;
 	}
 
-	if (adev->asic_type == CHIP_RAVEN) {
-		if (amdgpu_lbpw != 0)
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 2, 2):
+	case IP_VERSION(9, 1, 0):
+		gfx_v9_0_init_lbpw(adev);
+		if (amdgpu_lbpw == 0)
+			gfx_v9_0_enable_lbpw(adev, false);
+		else
+			gfx_v9_0_enable_lbpw(adev, true);
+		break;
+	case IP_VERSION(9, 4, 0):
+		gfx_v9_4_init_lbpw(adev);
+		if (amdgpu_lbpw > 0)
 			gfx_v9_0_enable_lbpw(adev, true);
 		else
 			gfx_v9_0_enable_lbpw(adev, false);
+		break;
+	default:
+		break;
 	}
 
-	gfx_v9_0_rlc_start(adev);
+	gfx_v9_0_update_spm_vmid_internal(adev, 0xf);
+
+	adev->gfx.rlc.funcs->start(adev);
 
 	return 0;
 }
 
 static void gfx_v9_0_cp_gfx_enable(struct amdgpu_device *adev, bool enable)
 {
-	int i;
 	u32 tmp = RREG32_SOC15(GC, 0, mmCP_ME_CNTL);
 
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, CE_INVALIDATE_ICACHE, enable ? 0 : 1);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_INVALIDATE_ICACHE, enable ? 0 : 1);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_INVALIDATE_ICACHE, enable ? 0 : 1);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, CE_PIPE0_RESET, enable ? 0 : 1);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, CE_PIPE1_RESET, enable ? 0 : 1);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_PIPE0_RESET, enable ? 0 : 1);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_PIPE1_RESET, enable ? 0 : 1);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_PIPE0_RESET, enable ? 0 : 1);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_PIPE1_RESET, enable ? 0 : 1);
 	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_HALT, enable ? 0 : 1);
 	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_HALT, enable ? 0 : 1);
 	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, CE_HALT, enable ? 0 : 1);
-	if (!enable) {
-		for (i = 0; i < adev->gfx.num_gfx_rings; i++)
-			adev->gfx.gfx_ring[i].ready = false;
-	}
-	WREG32_SOC15(GC, 0, mmCP_ME_CNTL, tmp);
+	WREG32_SOC15_RLC(GC, 0, mmCP_ME_CNTL, tmp);
 	udelay(50);
 }
 
@@ -2165,6 +3328,14 @@ static int gfx_v9_0_cp_gfx_start(struct amdgpu_device *adev)
 
 	gfx_v9_0_cp_gfx_enable(adev, true);
 
+	/* Now only limit the quirk on the APU gfx9 series and already
+	 * confirmed that the APU gfx10/gfx11 needn't such update.
+	 */
+	if (adev->flags & AMD_IS_APU &&
+			adev->in_s3 && !pm_resume_via_firmware()) {
+		DRM_INFO("Will skip the CSB packet resubmit\n");
+		return 0;
+	}
 	r = amdgpu_ring_alloc(ring, gfx_v9_0_get_csb_size(adev) + 4 + 3);
 	if (r) {
 		DRM_ERROR("amdgpu: cp failed to lock ring (%d).\n", r);
@@ -2242,12 +3413,12 @@ static int gfx_v9_0_cp_gfx_resume(struct amdgpu_device *adev)
 	WREG32_SOC15(GC, 0, mmCP_RB0_WPTR, lower_32_bits(ring->wptr));
 	WREG32_SOC15(GC, 0, mmCP_RB0_WPTR_HI, upper_32_bits(ring->wptr));
 
-	/* set the wb address wether it's enabled or not */
-	rptr_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+	/* set the wb address whether it's enabled or not */
+	rptr_addr = ring->rptr_gpu_addr;
 	WREG32_SOC15(GC, 0, mmCP_RB0_RPTR_ADDR, lower_32_bits(rptr_addr));
 	WREG32_SOC15(GC, 0, mmCP_RB0_RPTR_ADDR_HI, upper_32_bits(rptr_addr) & CP_RB_RPTR_ADDR_HI__RB_RPTR_ADDR_HI_MASK);
 
-	wptr_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
+	wptr_gpu_addr = ring->wptr_gpu_addr;
 	WREG32_SOC15(GC, 0, mmCP_RB_WPTR_POLL_ADDR_LO, lower_32_bits(wptr_gpu_addr));
 	WREG32_SOC15(GC, 0, mmCP_RB_WPTR_POLL_ADDR_HI, upper_32_bits(wptr_gpu_addr));
 
@@ -2279,23 +3450,26 @@ static int gfx_v9_0_cp_gfx_resume(struct amdgpu_device *adev)
 
 	/* start the ring */
 	gfx_v9_0_cp_gfx_start(adev);
-	ring->ready = true;
 
 	return 0;
 }
 
 static void gfx_v9_0_cp_compute_enable(struct amdgpu_device *adev, bool enable)
 {
-	int i;
-
 	if (enable) {
-		WREG32_SOC15(GC, 0, mmCP_MEC_CNTL, 0);
+		WREG32_SOC15_RLC(GC, 0, mmCP_MEC_CNTL, 0);
 	} else {
-		WREG32_SOC15(GC, 0, mmCP_MEC_CNTL,
-			(CP_MEC_CNTL__MEC_ME1_HALT_MASK | CP_MEC_CNTL__MEC_ME2_HALT_MASK));
-		for (i = 0; i < adev->gfx.num_compute_rings; i++)
-			adev->gfx.compute_ring[i].ready = false;
-		adev->gfx.kiq.ring.ready = false;
+		WREG32_SOC15_RLC(GC, 0, mmCP_MEC_CNTL,
+				 (CP_MEC_CNTL__MEC_INVALIDATE_ICACHE_MASK |
+				  CP_MEC_CNTL__MEC_ME1_PIPE0_RESET_MASK |
+				  CP_MEC_CNTL__MEC_ME1_PIPE1_RESET_MASK |
+				  CP_MEC_CNTL__MEC_ME1_PIPE2_RESET_MASK |
+				  CP_MEC_CNTL__MEC_ME1_PIPE3_RESET_MASK |
+				  CP_MEC_CNTL__MEC_ME2_PIPE0_RESET_MASK |
+				  CP_MEC_CNTL__MEC_ME2_PIPE1_RESET_MASK |
+				  CP_MEC_CNTL__MEC_ME1_HALT_MASK |
+				  CP_MEC_CNTL__MEC_ME2_HALT_MASK));
+		adev->gfx.kiq[0].ring.sched.ready = false;
 	}
 	udelay(50);
 }
@@ -2352,100 +3526,20 @@ static void gfx_v9_0_kiq_setting(struct amdgpu_ring *ring)
 	tmp = RREG32_SOC15(GC, 0, mmRLC_CP_SCHEDULERS);
 	tmp &= 0xffffff00;
 	tmp |= (ring->me << 5) | (ring->pipe << 3) | (ring->queue);
-	WREG32_SOC15(GC, 0, mmRLC_CP_SCHEDULERS, tmp);
-	tmp |= 0x80;
-	WREG32_SOC15(GC, 0, mmRLC_CP_SCHEDULERS, tmp);
+	WREG32_SOC15_RLC(GC, 0, mmRLC_CP_SCHEDULERS, tmp | 0x80);
 }
 
-static int gfx_v9_0_kiq_kcq_enable(struct amdgpu_device *adev)
+static void gfx_v9_0_mqd_set_priority(struct amdgpu_ring *ring, struct v9_mqd *mqd)
 {
-	struct amdgpu_ring *kiq_ring = &adev->gfx.kiq.ring;
-	uint32_t scratch, tmp = 0;
-	uint64_t queue_mask = 0;
-	int r, i;
-
-	for (i = 0; i < AMDGPU_MAX_COMPUTE_QUEUES; ++i) {
-		if (!test_bit(i, adev->gfx.mec.queue_bitmap))
-			continue;
+	struct amdgpu_device *adev = ring->adev;
 
-		/* This situation may be hit in the future if a new HW
-		 * generation exposes more than 64 queues. If so, the
-		 * definition of queue_mask needs updating */
-		if (WARN_ON(i >= (sizeof(queue_mask)*8))) {
-			DRM_ERROR("Invalid KCQ enabled: %d\n", i);
-			break;
+	if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE) {
+		if (amdgpu_gfx_is_high_priority_compute_queue(adev, ring)) {
+			mqd->cp_hqd_pipe_priority = AMDGPU_GFX_PIPE_PRIO_HIGH;
+			mqd->cp_hqd_queue_priority =
+				AMDGPU_GFX_QUEUE_PRIORITY_MAXIMUM;
 		}
-
-		queue_mask |= (1ull << i);
 	}
-
-	r = amdgpu_gfx_scratch_get(adev, &scratch);
-	if (r) {
-		DRM_ERROR("Failed to get scratch reg (%d).\n", r);
-		return r;
-	}
-	WREG32(scratch, 0xCAFEDEAD);
-
-	r = amdgpu_ring_alloc(kiq_ring, (7 * adev->gfx.num_compute_rings) + 11);
-	if (r) {
-		DRM_ERROR("Failed to lock KIQ (%d).\n", r);
-		amdgpu_gfx_scratch_free(adev, scratch);
-		return r;
-	}
-
-	/* set resources */
-	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_SET_RESOURCES, 6));
-	amdgpu_ring_write(kiq_ring, PACKET3_SET_RESOURCES_VMID_MASK(0) |
-			  PACKET3_SET_RESOURCES_QUEUE_TYPE(0));	/* vmid_mask:0 queue_type:0 (KIQ) */
-	amdgpu_ring_write(kiq_ring, lower_32_bits(queue_mask));	/* queue mask lo */
-	amdgpu_ring_write(kiq_ring, upper_32_bits(queue_mask));	/* queue mask hi */
-	amdgpu_ring_write(kiq_ring, 0);	/* gws mask lo */
-	amdgpu_ring_write(kiq_ring, 0);	/* gws mask hi */
-	amdgpu_ring_write(kiq_ring, 0);	/* oac mask */
-	amdgpu_ring_write(kiq_ring, 0);	/* gds heap base:0, gds heap size:0 */
-	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
-		struct amdgpu_ring *ring = &adev->gfx.compute_ring[i];
-		uint64_t mqd_addr = amdgpu_bo_gpu_offset(ring->mqd_obj);
-		uint64_t wptr_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
-
-		amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_MAP_QUEUES, 5));
-		/* Q_sel:0, vmid:0, vidmem: 1, engine:0, num_Q:1*/
-		amdgpu_ring_write(kiq_ring, /* Q_sel: 0, vmid: 0, engine: 0, num_Q: 1 */
-				  PACKET3_MAP_QUEUES_QUEUE_SEL(0) | /* Queue_Sel */
-				  PACKET3_MAP_QUEUES_VMID(0) | /* VMID */
-				  PACKET3_MAP_QUEUES_QUEUE(ring->queue) |
-				  PACKET3_MAP_QUEUES_PIPE(ring->pipe) |
-				  PACKET3_MAP_QUEUES_ME((ring->me == 1 ? 0 : 1)) |
-				  PACKET3_MAP_QUEUES_QUEUE_TYPE(0) | /*queue_type: normal compute queue */
-				  PACKET3_MAP_QUEUES_ALLOC_FORMAT(1) | /* alloc format: all_on_one_pipe */
-				  PACKET3_MAP_QUEUES_ENGINE_SEL(0) | /* engine_sel: compute */
-				  PACKET3_MAP_QUEUES_NUM_QUEUES(1)); /* num_queues: must be 1 */
-		amdgpu_ring_write(kiq_ring, PACKET3_MAP_QUEUES_DOORBELL_OFFSET(ring->doorbell_index));
-		amdgpu_ring_write(kiq_ring, lower_32_bits(mqd_addr));
-		amdgpu_ring_write(kiq_ring, upper_32_bits(mqd_addr));
-		amdgpu_ring_write(kiq_ring, lower_32_bits(wptr_addr));
-		amdgpu_ring_write(kiq_ring, upper_32_bits(wptr_addr));
-	}
-	/* write to scratch for completion */
-	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_SET_UCONFIG_REG, 1));
-	amdgpu_ring_write(kiq_ring, (scratch - PACKET3_SET_UCONFIG_REG_START));
-	amdgpu_ring_write(kiq_ring, 0xDEADBEEF);
-	amdgpu_ring_commit(kiq_ring);
-
-	for (i = 0; i < adev->usec_timeout; i++) {
-		tmp = RREG32(scratch);
-		if (tmp == 0xDEADBEEF)
-			break;
-		DRM_UDELAY(1);
-	}
-	if (i >= adev->usec_timeout) {
-		DRM_ERROR("KCQ enable failed (scratch(0x%04X)=0x%08X)\n",
-			  scratch, tmp);
-		r = -EINVAL;
-	}
-	amdgpu_gfx_scratch_free(adev, scratch);
-
-	return r;
 }
 
 static int gfx_v9_0_mqd_init(struct amdgpu_ring *ring)
@@ -2461,8 +3555,19 @@ static int gfx_v9_0_mqd_init(struct amdgpu_ring *ring)
 	mqd->compute_static_thread_mgmt_se1 = 0xffffffff;
 	mqd->compute_static_thread_mgmt_se2 = 0xffffffff;
 	mqd->compute_static_thread_mgmt_se3 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se4 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se5 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se6 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se7 = 0xffffffff;
 	mqd->compute_misc_reserved = 0x00000003;
 
+	mqd->dynamic_cu_mask_addr_lo =
+		lower_32_bits(ring->mqd_gpu_addr
+			      + offsetof(struct v9_mqd_allocation, dynamic_cu_mask));
+	mqd->dynamic_cu_mask_addr_hi =
+		upper_32_bits(ring->mqd_gpu_addr
+			      + offsetof(struct v9_mqd_allocation, dynamic_cu_mask));
+
 	eop_base_addr = ring->eop_gpu_addr >> 8;
 	mqd->cp_hqd_eop_base_addr_lo = eop_base_addr;
 	mqd->cp_hqd_eop_base_addr_hi = upper_32_bits(eop_base_addr);
@@ -2486,10 +3591,10 @@ static int gfx_v9_0_mqd_init(struct amdgpu_ring *ring)
 				    DOORBELL_SOURCE, 0);
 		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
 				    DOORBELL_HIT, 0);
-	}
-	else
+	} else {
 		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
 					 DOORBELL_EN, 0);
+	}
 
 	mqd->cp_hqd_pq_doorbell_control = tmp;
 
@@ -2519,7 +3624,7 @@ static int gfx_v9_0_mqd_init(struct amdgpu_ring *ring)
 	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, QUEUE_SIZE,
 			    (order_base_2(ring->ring_size / 4) - 1));
 	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, RPTR_BLOCK_SIZE,
-			((order_base_2(AMDGPU_GPU_PAGE_SIZE / 4) - 1) << 8));
+			(order_base_2(AMDGPU_GPU_PAGE_SIZE / 4) - 1));
 #ifdef __BIG_ENDIAN
 	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, ENDIAN_SWAP, 1);
 #endif
@@ -2530,33 +3635,16 @@ static int gfx_v9_0_mqd_init(struct amdgpu_ring *ring)
 	mqd->cp_hqd_pq_control = tmp;
 
 	/* set the wb address whether it's enabled or not */
-	wb_gpu_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+	wb_gpu_addr = ring->rptr_gpu_addr;
 	mqd->cp_hqd_pq_rptr_report_addr_lo = wb_gpu_addr & 0xfffffffc;
 	mqd->cp_hqd_pq_rptr_report_addr_hi =
 		upper_32_bits(wb_gpu_addr) & 0xffff;
 
 	/* only used if CP_PQ_WPTR_POLL_CNTL.CP_PQ_WPTR_POLL_CNTL__EN_MASK=1 */
-	wb_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
+	wb_gpu_addr = ring->wptr_gpu_addr;
 	mqd->cp_hqd_pq_wptr_poll_addr_lo = wb_gpu_addr & 0xfffffffc;
 	mqd->cp_hqd_pq_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr) & 0xffff;
 
-	tmp = 0;
-	/* enable the doorbell if requested */
-	if (ring->use_doorbell) {
-		tmp = RREG32_SOC15(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL);
-		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
-				DOORBELL_OFFSET, ring->doorbell_index);
-
-		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
-					 DOORBELL_EN, 1);
-		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
-					 DOORBELL_SOURCE, 0);
-		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
-					 DOORBELL_HIT, 0);
-	}
-
-	mqd->cp_hqd_pq_doorbell_control = tmp;
-
 	/* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */
 	ring->wptr = 0;
 	mqd->cp_hqd_pq_rptr = RREG32_SOC15(GC, 0, mmCP_HQD_PQ_RPTR);
@@ -2573,8 +3661,15 @@ static int gfx_v9_0_mqd_init(struct amdgpu_ring *ring)
 	tmp = REG_SET_FIELD(tmp, CP_HQD_IB_CONTROL, MIN_IB_AVAIL_SIZE, 3);
 	mqd->cp_hqd_ib_control = tmp;
 
-	/* activate the queue */
-	mqd->cp_hqd_active = 1;
+	/* set static priority for a queue/ring */
+	gfx_v9_0_mqd_set_priority(ring, mqd);
+	mqd->cp_hqd_quantum = RREG32_SOC15(GC, 0, mmCP_HQD_QUANTUM);
+
+	/* map_queues packet doesn't need activate the queue,
+	 * so only kiq need set this field.
+	 */
+	if (ring->funcs->type == AMDGPU_RING_TYPE_KIQ)
+		mqd->cp_hqd_active = 1;
 
 	return 0;
 }
@@ -2588,94 +3683,103 @@ static int gfx_v9_0_kiq_init_register(struct amdgpu_ring *ring)
 	/* disable wptr polling */
 	WREG32_FIELD15(GC, 0, CP_PQ_WPTR_POLL_CNTL, EN, 0);
 
-	WREG32_SOC15(GC, 0, mmCP_HQD_EOP_BASE_ADDR,
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_EOP_BASE_ADDR,
 	       mqd->cp_hqd_eop_base_addr_lo);
-	WREG32_SOC15(GC, 0, mmCP_HQD_EOP_BASE_ADDR_HI,
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_EOP_BASE_ADDR_HI,
 	       mqd->cp_hqd_eop_base_addr_hi);
 
 	/* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */
-	WREG32_SOC15(GC, 0, mmCP_HQD_EOP_CONTROL,
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_EOP_CONTROL,
 	       mqd->cp_hqd_eop_control);
 
 	/* enable doorbell? */
-	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL,
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL,
 	       mqd->cp_hqd_pq_doorbell_control);
 
 	/* disable the queue if it's active */
 	if (RREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE) & 1) {
-		WREG32_SOC15(GC, 0, mmCP_HQD_DEQUEUE_REQUEST, 1);
+		WREG32_SOC15_RLC(GC, 0, mmCP_HQD_DEQUEUE_REQUEST, 1);
 		for (j = 0; j < adev->usec_timeout; j++) {
 			if (!(RREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE) & 1))
 				break;
 			udelay(1);
 		}
-		WREG32_SOC15(GC, 0, mmCP_HQD_DEQUEUE_REQUEST,
+		WREG32_SOC15_RLC(GC, 0, mmCP_HQD_DEQUEUE_REQUEST,
 		       mqd->cp_hqd_dequeue_request);
-		WREG32_SOC15(GC, 0, mmCP_HQD_PQ_RPTR,
+		WREG32_SOC15_RLC(GC, 0, mmCP_HQD_PQ_RPTR,
 		       mqd->cp_hqd_pq_rptr);
-		WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_LO,
+		WREG32_SOC15_RLC(GC, 0, mmCP_HQD_PQ_WPTR_LO,
 		       mqd->cp_hqd_pq_wptr_lo);
-		WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_HI,
+		WREG32_SOC15_RLC(GC, 0, mmCP_HQD_PQ_WPTR_HI,
 		       mqd->cp_hqd_pq_wptr_hi);
 	}
 
 	/* set the pointer to the MQD */
-	WREG32_SOC15(GC, 0, mmCP_MQD_BASE_ADDR,
+	WREG32_SOC15_RLC(GC, 0, mmCP_MQD_BASE_ADDR,
 	       mqd->cp_mqd_base_addr_lo);
-	WREG32_SOC15(GC, 0, mmCP_MQD_BASE_ADDR_HI,
+	WREG32_SOC15_RLC(GC, 0, mmCP_MQD_BASE_ADDR_HI,
 	       mqd->cp_mqd_base_addr_hi);
 
 	/* set MQD vmid to 0 */
-	WREG32_SOC15(GC, 0, mmCP_MQD_CONTROL,
+	WREG32_SOC15_RLC(GC, 0, mmCP_MQD_CONTROL,
 	       mqd->cp_mqd_control);
 
 	/* set the pointer to the HQD, this is similar CP_RB0_BASE/_HI */
-	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_BASE,
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_PQ_BASE,
 	       mqd->cp_hqd_pq_base_lo);
-	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_BASE_HI,
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_PQ_BASE_HI,
 	       mqd->cp_hqd_pq_base_hi);
 
 	/* set up the HQD, this is similar to CP_RB0_CNTL */
-	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_CONTROL,
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_PQ_CONTROL,
 	       mqd->cp_hqd_pq_control);
 
 	/* set the wb address whether it's enabled or not */
-	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_RPTR_REPORT_ADDR,
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_PQ_RPTR_REPORT_ADDR,
 				mqd->cp_hqd_pq_rptr_report_addr_lo);
-	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_RPTR_REPORT_ADDR_HI,
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_PQ_RPTR_REPORT_ADDR_HI,
 				mqd->cp_hqd_pq_rptr_report_addr_hi);
 
 	/* only used if CP_PQ_WPTR_POLL_CNTL.CP_PQ_WPTR_POLL_CNTL__EN_MASK=1 */
-	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_POLL_ADDR,
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_PQ_WPTR_POLL_ADDR,
 	       mqd->cp_hqd_pq_wptr_poll_addr_lo);
-	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_POLL_ADDR_HI,
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_PQ_WPTR_POLL_ADDR_HI,
 	       mqd->cp_hqd_pq_wptr_poll_addr_hi);
 
 	/* enable the doorbell if requested */
 	if (ring->use_doorbell) {
 		WREG32_SOC15(GC, 0, mmCP_MEC_DOORBELL_RANGE_LOWER,
-					(AMDGPU_DOORBELL64_KIQ *2) << 2);
-		WREG32_SOC15(GC, 0, mmCP_MEC_DOORBELL_RANGE_UPPER,
-					(AMDGPU_DOORBELL64_USERQUEUE_END * 2) << 2);
+					(adev->doorbell_index.kiq * 2) << 2);
+		/* If GC has entered CGPG, ringing doorbell > first page
+		 * doesn't wakeup GC. Enlarge CP_MEC_DOORBELL_RANGE_UPPER to
+		 * workaround this issue. And this change has to align with firmware
+		 * update.
+		 */
+		if (check_if_enlarge_doorbell_range(adev))
+			WREG32_SOC15(GC, 0, mmCP_MEC_DOORBELL_RANGE_UPPER,
+					(adev->doorbell.size - 4));
+		else
+			WREG32_SOC15(GC, 0, mmCP_MEC_DOORBELL_RANGE_UPPER,
+					(adev->doorbell_index.userqueue_end * 2) << 2);
 	}
 
-	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL,
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL,
 	       mqd->cp_hqd_pq_doorbell_control);
 
 	/* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */
-	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_LO,
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_PQ_WPTR_LO,
 	       mqd->cp_hqd_pq_wptr_lo);
-	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_HI,
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_PQ_WPTR_HI,
 	       mqd->cp_hqd_pq_wptr_hi);
 
 	/* set the vmid for the queue */
-	WREG32_SOC15(GC, 0, mmCP_HQD_VMID, mqd->cp_hqd_vmid);
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_VMID, mqd->cp_hqd_vmid);
 
-	WREG32_SOC15(GC, 0, mmCP_HQD_PERSISTENT_STATE,
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_PERSISTENT_STATE,
 	       mqd->cp_hqd_persistent_state);
 
 	/* activate the queue */
-	WREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE,
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_ACTIVE,
 	       mqd->cp_hqd_active);
 
 	if (ring->use_doorbell)
@@ -2684,69 +3788,125 @@ static int gfx_v9_0_kiq_init_register(struct amdgpu_ring *ring)
 	return 0;
 }
 
+static int gfx_v9_0_kiq_fini_register(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int j;
+
+	/* disable the queue if it's active */
+	if (RREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE) & 1) {
+
+		WREG32_SOC15_RLC(GC, 0, mmCP_HQD_DEQUEUE_REQUEST, 1);
+
+		for (j = 0; j < adev->usec_timeout; j++) {
+			if (!(RREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE) & 1))
+				break;
+			udelay(1);
+		}
+
+		if (j == AMDGPU_MAX_USEC_TIMEOUT) {
+			DRM_DEBUG("KIQ dequeue request failed.\n");
+
+			/* Manual disable if dequeue request times out */
+			WREG32_SOC15_RLC(GC, 0, mmCP_HQD_ACTIVE, 0);
+		}
+
+		WREG32_SOC15_RLC(GC, 0, mmCP_HQD_DEQUEUE_REQUEST,
+		      0);
+	}
+
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_IQ_TIMER, 0);
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_IB_CONTROL, 0);
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_PERSISTENT_STATE, 0);
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL, 0x40000000);
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL, 0);
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_PQ_RPTR, 0);
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_PQ_WPTR_HI, 0);
+	WREG32_SOC15_RLC(GC, 0, mmCP_HQD_PQ_WPTR_LO, 0);
+
+	return 0;
+}
+
 static int gfx_v9_0_kiq_init_queue(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
 	struct v9_mqd *mqd = ring->mqd_ptr;
-	int mqd_idx = AMDGPU_MAX_COMPUTE_RINGS;
+	struct v9_mqd *tmp_mqd;
 
 	gfx_v9_0_kiq_setting(ring);
 
-	if (adev->gfx.in_reset) { /* for GPU_RESET case */
-		/* reset MQD to a clean status */
-		if (adev->gfx.mec.mqd_backup[mqd_idx])
-			memcpy(mqd, adev->gfx.mec.mqd_backup[mqd_idx], sizeof(*mqd));
+	/* GPU could be in bad state during probe, driver trigger the reset
+	 * after load the SMU, in this case , the mqd is not be initialized.
+	 * driver need to re-init the mqd.
+	 * check mqd->cp_hqd_pq_control since this value should not be 0
+	 */
+	tmp_mqd = (struct v9_mqd *)adev->gfx.kiq[0].mqd_backup;
+	if (amdgpu_in_reset(adev) && tmp_mqd->cp_hqd_pq_control){
+		/* for GPU_RESET case , reset MQD to a clean status */
+		if (adev->gfx.kiq[0].mqd_backup)
+			memcpy(mqd, adev->gfx.kiq[0].mqd_backup, sizeof(struct v9_mqd_allocation));
 
 		/* reset ring buffer */
 		ring->wptr = 0;
 		amdgpu_ring_clear_ring(ring);
 
 		mutex_lock(&adev->srbm_mutex);
-		soc15_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		soc15_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0, 0);
 		gfx_v9_0_kiq_init_register(ring);
-		soc15_grbm_select(adev, 0, 0, 0, 0);
+		soc15_grbm_select(adev, 0, 0, 0, 0, 0);
 		mutex_unlock(&adev->srbm_mutex);
 	} else {
-		memset((void *)mqd, 0, sizeof(*mqd));
+		memset((void *)mqd, 0, sizeof(struct v9_mqd_allocation));
+		((struct v9_mqd_allocation *)mqd)->dynamic_cu_mask = 0xFFFFFFFF;
+		((struct v9_mqd_allocation *)mqd)->dynamic_rb_mask = 0xFFFFFFFF;
+		if (amdgpu_sriov_vf(adev) && adev->in_suspend)
+			amdgpu_ring_clear_ring(ring);
 		mutex_lock(&adev->srbm_mutex);
-		soc15_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		soc15_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0, 0);
 		gfx_v9_0_mqd_init(ring);
 		gfx_v9_0_kiq_init_register(ring);
-		soc15_grbm_select(adev, 0, 0, 0, 0);
+		soc15_grbm_select(adev, 0, 0, 0, 0, 0);
 		mutex_unlock(&adev->srbm_mutex);
 
-		if (adev->gfx.mec.mqd_backup[mqd_idx])
-			memcpy(adev->gfx.mec.mqd_backup[mqd_idx], mqd, sizeof(*mqd));
+		if (adev->gfx.kiq[0].mqd_backup)
+			memcpy(adev->gfx.kiq[0].mqd_backup, mqd, sizeof(struct v9_mqd_allocation));
 	}
 
 	return 0;
 }
 
-static int gfx_v9_0_kcq_init_queue(struct amdgpu_ring *ring)
+static int gfx_v9_0_kcq_init_queue(struct amdgpu_ring *ring, bool restore)
 {
 	struct amdgpu_device *adev = ring->adev;
 	struct v9_mqd *mqd = ring->mqd_ptr;
 	int mqd_idx = ring - &adev->gfx.compute_ring[0];
+	struct v9_mqd *tmp_mqd;
+
+	/* Same as above kiq init, driver need to re-init the mqd if mqd->cp_hqd_pq_control
+	 * is not be initialized before
+	 */
+	tmp_mqd = (struct v9_mqd *)adev->gfx.mec.mqd_backup[mqd_idx];
 
-	if (!adev->gfx.in_reset && !adev->gfx.in_suspend) {
-		memset((void *)mqd, 0, sizeof(*mqd));
+	if (!restore && (!tmp_mqd->cp_hqd_pq_control ||
+	    (!amdgpu_in_reset(adev) && !adev->in_suspend))) {
+		memset((void *)mqd, 0, sizeof(struct v9_mqd_allocation));
+		((struct v9_mqd_allocation *)mqd)->dynamic_cu_mask = 0xFFFFFFFF;
+		((struct v9_mqd_allocation *)mqd)->dynamic_rb_mask = 0xFFFFFFFF;
 		mutex_lock(&adev->srbm_mutex);
-		soc15_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		soc15_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0, 0);
 		gfx_v9_0_mqd_init(ring);
-		soc15_grbm_select(adev, 0, 0, 0, 0);
+		soc15_grbm_select(adev, 0, 0, 0, 0, 0);
 		mutex_unlock(&adev->srbm_mutex);
 
 		if (adev->gfx.mec.mqd_backup[mqd_idx])
-			memcpy(adev->gfx.mec.mqd_backup[mqd_idx], mqd, sizeof(*mqd));
-	} else if (adev->gfx.in_reset) { /* for GPU_RESET case */
-		/* reset MQD to a clean status */
+			memcpy(adev->gfx.mec.mqd_backup[mqd_idx], mqd, sizeof(struct v9_mqd_allocation));
+	} else {
+		/* restore MQD to a clean status */
 		if (adev->gfx.mec.mqd_backup[mqd_idx])
-			memcpy(mqd, adev->gfx.mec.mqd_backup[mqd_idx], sizeof(*mqd));
-
+			memcpy(mqd, adev->gfx.mec.mqd_backup[mqd_idx], sizeof(struct v9_mqd_allocation));
 		/* reset ring buffer */
 		ring->wptr = 0;
-		amdgpu_ring_clear_ring(ring);
-	} else {
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr, 0);
 		amdgpu_ring_clear_ring(ring);
 	}
 
@@ -2755,47 +3915,23 @@ static int gfx_v9_0_kcq_init_queue(struct amdgpu_ring *ring)
 
 static int gfx_v9_0_kiq_resume(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *ring = NULL;
-	int r = 0, i;
-
-	gfx_v9_0_cp_compute_enable(adev, true);
-
-	ring = &adev->gfx.kiq.ring;
+	gfx_v9_0_kiq_init_queue(&adev->gfx.kiq[0].ring);
+	return 0;
+}
 
-	r = amdgpu_bo_reserve(ring->mqd_obj, false);
-	if (unlikely(r != 0))
-		goto done;
+static int gfx_v9_0_kcq_resume(struct amdgpu_device *adev)
+{
+	int i, r;
 
-	r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-	if (!r) {
-		r = gfx_v9_0_kiq_init_queue(ring);
-		amdgpu_bo_kunmap(ring->mqd_obj);
-		ring->mqd_ptr = NULL;
-	}
-	amdgpu_bo_unreserve(ring->mqd_obj);
-	if (r)
-		goto done;
+	gfx_v9_0_cp_compute_enable(adev, true);
 
 	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
-		ring = &adev->gfx.compute_ring[i];
-
-		r = amdgpu_bo_reserve(ring->mqd_obj, false);
-		if (unlikely(r != 0))
-			goto done;
-		r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-		if (!r) {
-			r = gfx_v9_0_kcq_init_queue(ring);
-			amdgpu_bo_kunmap(ring->mqd_obj);
-			ring->mqd_ptr = NULL;
-		}
-		amdgpu_bo_unreserve(ring->mqd_obj);
+		r = gfx_v9_0_kcq_init_queue(&adev->gfx.compute_ring[i], false);
 		if (r)
-			goto done;
+			return r;
 	}
 
-	r = gfx_v9_0_kiq_kcq_enable(adev);
-done:
-	return r;
+	return amdgpu_gfx_enable_kcq(adev, 0);
 }
 
 static int gfx_v9_0_cp_resume(struct amdgpu_device *adev)
@@ -2807,44 +3943,46 @@ static int gfx_v9_0_cp_resume(struct amdgpu_device *adev)
 		gfx_v9_0_enable_gui_idle_interrupt(adev, false);
 
 	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
-		/* legacy firmware loading */
-		r = gfx_v9_0_cp_gfx_load_microcode(adev);
-		if (r)
-			return r;
+		if (adev->gfx.num_gfx_rings) {
+			/* legacy firmware loading */
+			r = gfx_v9_0_cp_gfx_load_microcode(adev);
+			if (r)
+				return r;
+		}
 
 		r = gfx_v9_0_cp_compute_load_microcode(adev);
 		if (r)
 			return r;
 	}
 
-	r = gfx_v9_0_cp_gfx_resume(adev);
-	if (r)
-		return r;
+	if (adev->gfx.num_gfx_rings)
+		gfx_v9_0_cp_gfx_enable(adev, false);
+	gfx_v9_0_cp_compute_enable(adev, false);
 
 	r = gfx_v9_0_kiq_resume(adev);
 	if (r)
 		return r;
 
-	ring = &adev->gfx.gfx_ring[0];
-	r = amdgpu_ring_test_ring(ring);
-	if (r) {
-		ring->ready = false;
-		return r;
+	if (adev->gfx.num_gfx_rings) {
+		r = gfx_v9_0_cp_gfx_resume(adev);
+		if (r)
+			return r;
 	}
 
-	ring = &adev->gfx.kiq.ring;
-	ring->ready = true;
-	r = amdgpu_ring_test_ring(ring);
+	r = gfx_v9_0_kcq_resume(adev);
 	if (r)
-		ring->ready = false;
+		return r;
+
+	if (adev->gfx.num_gfx_rings) {
+		ring = &adev->gfx.gfx_ring[0];
+		r = amdgpu_ring_test_helper(ring);
+		if (r)
+			return r;
+	}
 
 	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
 		ring = &adev->gfx.compute_ring[i];
-
-		ring->ready = true;
-		r = amdgpu_ring_test_ring(ring);
-		if (r)
-			ring->ready = false;
+		amdgpu_ring_test_helper(ring);
 	}
 
 	gfx_v9_0_enable_gui_idle_interrupt(adev, true);
@@ -2852,22 +3990,47 @@ static int gfx_v9_0_cp_resume(struct amdgpu_device *adev)
 	return 0;
 }
 
+static void gfx_v9_0_init_tcp_config(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 1) &&
+	    amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 2))
+		return;
+
+	tmp = RREG32_SOC15(GC, 0, mmTCP_ADDR_CONFIG);
+	tmp = REG_SET_FIELD(tmp, TCP_ADDR_CONFIG, ENABLE64KHASH,
+				adev->df.hash_status.hash_64k);
+	tmp = REG_SET_FIELD(tmp, TCP_ADDR_CONFIG, ENABLE2MHASH,
+				adev->df.hash_status.hash_2m);
+	tmp = REG_SET_FIELD(tmp, TCP_ADDR_CONFIG, ENABLE1GHASH,
+				adev->df.hash_status.hash_1g);
+	WREG32_SOC15(GC, 0, mmTCP_ADDR_CONFIG, tmp);
+}
+
 static void gfx_v9_0_cp_enable(struct amdgpu_device *adev, bool enable)
 {
-	gfx_v9_0_cp_gfx_enable(adev, enable);
+	if (adev->gfx.num_gfx_rings)
+		gfx_v9_0_cp_gfx_enable(adev, enable);
 	gfx_v9_0_cp_compute_enable(adev, enable);
 }
 
-static int gfx_v9_0_hw_init(void *handle)
+static int gfx_v9_0_hw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	amdgpu_gfx_cleaner_shader_init(adev, adev->gfx.cleaner_shader_size,
+				       adev->gfx.cleaner_shader_ptr);
 
-	gfx_v9_0_init_golden_registers(adev);
+	if (!amdgpu_sriov_vf(adev))
+		gfx_v9_0_init_golden_registers(adev);
 
-	gfx_v9_0_gpu_init(adev);
+	gfx_v9_0_constants_init(adev);
 
-	r = gfx_v9_0_rlc_resume(adev);
+	gfx_v9_0_init_tcp_config(adev);
+
+	r = adev->gfx.rlc.funcs->resume(adev);
 	if (r)
 		return r;
 
@@ -2875,50 +4038,78 @@ static int gfx_v9_0_hw_init(void *handle)
 	if (r)
 		return r;
 
-	r = gfx_v9_0_ngg_en(adev);
-	if (r)
-		return r;
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2) &&
+	    !amdgpu_sriov_vf(adev))
+		gfx_v9_4_2_set_power_brake_sequence(adev);
 
 	return r;
 }
 
-static int gfx_v9_0_hw_fini(void *handle)
+static int gfx_v9_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
+		amdgpu_irq_put(adev, &adev->gfx.cp_ecc_error_irq, 0);
 	amdgpu_irq_put(adev, &adev->gfx.priv_reg_irq, 0);
 	amdgpu_irq_put(adev, &adev->gfx.priv_inst_irq, 0);
+	amdgpu_irq_put(adev, &adev->gfx.bad_op_irq, 0);
+
+	/* DF freeze and kcq disable will fail */
+	if (!amdgpu_ras_intr_triggered())
+		/* disable KCQ to avoid CPC touch memory not valid anymore */
+		amdgpu_gfx_disable_kcq(adev, 0);
+
 	if (amdgpu_sriov_vf(adev)) {
-		pr_debug("For SRIOV client, shouldn't do anything.\n");
+		gfx_v9_0_cp_gfx_enable(adev, false);
+		/* must disable polling for SRIOV when hw finished, otherwise
+		 * CPC engine may still keep fetching WB address which is already
+		 * invalid after sw finished and trigger DMAR reading error in
+		 * hypervisor side.
+		 */
+		WREG32_FIELD15(GC, 0, CP_PQ_WPTR_POLL_CNTL, EN, 0);
 		return 0;
 	}
+
+	/* Use deinitialize sequence from CAIL when unbinding device from driver,
+	 * otherwise KIQ is hanging when binding back
+	 */
+	if (!amdgpu_in_reset(adev) && !adev->in_suspend) {
+		mutex_lock(&adev->srbm_mutex);
+		soc15_grbm_select(adev, adev->gfx.kiq[0].ring.me,
+				adev->gfx.kiq[0].ring.pipe,
+				adev->gfx.kiq[0].ring.queue, 0, 0);
+		gfx_v9_0_kiq_fini_register(&adev->gfx.kiq[0].ring);
+		soc15_grbm_select(adev, 0, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+	}
+
 	gfx_v9_0_cp_enable(adev, false);
-	gfx_v9_0_rlc_stop(adev);
 
+	/* Skip stopping RLC with A+A reset or when RLC controls GFX clock */
+	if ((adev->gmc.xgmi.connected_to_cpu && amdgpu_in_reset(adev)) ||
+	    (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(9, 4, 2))) {
+		dev_dbg(adev->dev, "Skipping RLC halt\n");
+		return 0;
+	}
+
+	adev->gfx.rlc.funcs->stop(adev);
 	return 0;
 }
 
-static int gfx_v9_0_suspend(void *handle)
+static int gfx_v9_0_suspend(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	adev->gfx.in_suspend = true;
-	return gfx_v9_0_hw_fini(adev);
+	return gfx_v9_0_hw_fini(ip_block);
 }
 
-static int gfx_v9_0_resume(void *handle)
+static int gfx_v9_0_resume(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	int r;
-
-	r = gfx_v9_0_hw_init(adev);
-	adev->gfx.in_suspend = false;
-	return r;
+	return gfx_v9_0_hw_init(ip_block);
 }
 
-static bool gfx_v9_0_is_idle(void *handle)
+static bool gfx_v9_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (REG_GET_FIELD(RREG32_SOC15(GC, 0, mmGRBM_STATUS),
 				GRBM_STATUS, GUI_ACTIVE))
@@ -2927,29 +4118,24 @@ static bool gfx_v9_0_is_idle(void *handle)
 		return true;
 }
 
-static int gfx_v9_0_wait_for_idle(void *handle)
+static int gfx_v9_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
 	unsigned i;
-	u32 tmp;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		/* read MC_STATUS */
-		tmp = RREG32_SOC15(GC, 0, mmGRBM_STATUS) &
-			GRBM_STATUS__GUI_ACTIVE_MASK;
-
-		if (!REG_GET_FIELD(tmp, GRBM_STATUS, GUI_ACTIVE))
+		if (gfx_v9_0_is_idle(ip_block))
 			return 0;
 		udelay(1);
 	}
 	return -ETIMEDOUT;
 }
 
-static int gfx_v9_0_soft_reset(void *handle)
+static int gfx_v9_0_soft_reset(struct amdgpu_ip_block *ip_block)
 {
 	u32 grbm_soft_reset = 0;
 	u32 tmp;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	/* GRBM_STATUS */
 	tmp = RREG32_SOC15(GC, 0, mmGRBM_STATUS);
@@ -2979,27 +4165,26 @@ static int gfx_v9_0_soft_reset(void *handle)
 
 	if (grbm_soft_reset) {
 		/* stop the rlc */
-		gfx_v9_0_rlc_stop(adev);
+		adev->gfx.rlc.funcs->stop(adev);
 
-		/* Disable GFX parsing/prefetching */
-		gfx_v9_0_cp_gfx_enable(adev, false);
+		if (adev->gfx.num_gfx_rings)
+			/* Disable GFX parsing/prefetching */
+			gfx_v9_0_cp_gfx_enable(adev, false);
 
 		/* Disable MEC parsing/prefetching */
 		gfx_v9_0_cp_compute_enable(adev, false);
 
-		if (grbm_soft_reset) {
-			tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
-			tmp |= grbm_soft_reset;
-			dev_info(adev->dev, "GRBM_SOFT_RESET=0x%08X\n", tmp);
-			WREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET, tmp);
-			tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
+		tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
+		tmp |= grbm_soft_reset;
+		dev_info(adev->dev, "GRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET, tmp);
+		tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
 
-			udelay(50);
+		udelay(50);
 
-			tmp &= ~grbm_soft_reset;
-			WREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET, tmp);
-			tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
-		}
+		tmp &= ~grbm_soft_reset;
+		WREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET, tmp);
+		tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
 
 		/* Wait a little for things to settle down */
 		udelay(50);
@@ -3007,15 +4192,116 @@ static int gfx_v9_0_soft_reset(void *handle)
 	return 0;
 }
 
+static uint64_t gfx_v9_0_kiq_read_clock(struct amdgpu_device *adev)
+{
+	signed long r, cnt = 0;
+	unsigned long flags;
+	uint32_t seq, reg_val_offs = 0;
+	uint64_t value = 0;
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[0];
+	struct amdgpu_ring *ring = &kiq->ring;
+
+	BUG_ON(!ring->funcs->emit_rreg);
+
+	spin_lock_irqsave(&kiq->ring_lock, flags);
+	if (amdgpu_device_wb_get(adev, &reg_val_offs)) {
+		pr_err("critical bug! too many kiq readers\n");
+		goto failed_unlock;
+	}
+	amdgpu_ring_alloc(ring, 32);
+	amdgpu_ring_write(ring, PACKET3(PACKET3_COPY_DATA, 4));
+	amdgpu_ring_write(ring, 9 |	/* src: register*/
+				(5 << 8) |	/* dst: memory */
+				(1 << 16) |	/* count sel */
+				(1 << 20));	/* write confirm */
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, lower_32_bits(adev->wb.gpu_addr +
+				reg_val_offs * 4));
+	amdgpu_ring_write(ring, upper_32_bits(adev->wb.gpu_addr +
+				reg_val_offs * 4));
+	r = amdgpu_fence_emit_polling(ring, &seq, MAX_KIQ_REG_WAIT);
+	if (r)
+		goto failed_undo;
+
+	amdgpu_ring_commit(ring);
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+
+	r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
+
+	/* don't wait anymore for gpu reset case because this way may
+	 * block gpu_recover() routine forever, e.g. this virt_kiq_rreg
+	 * is triggered in TTM and ttm_bo_lock_delayed_workqueue() will
+	 * never return if we keep waiting in virt_kiq_rreg, which cause
+	 * gpu_recover() hang there.
+	 *
+	 * also don't wait anymore for IRQ context
+	 * */
+	if (r < 1 && (amdgpu_in_reset(adev)))
+		goto failed_kiq_read;
+
+	might_sleep();
+	while (r < 1 && cnt++ < MAX_KIQ_REG_TRY) {
+		msleep(MAX_KIQ_REG_BAILOUT_INTERVAL);
+		r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
+	}
+
+	if (cnt > MAX_KIQ_REG_TRY)
+		goto failed_kiq_read;
+
+	mb();
+	value = (uint64_t)adev->wb.wb[reg_val_offs] |
+		(uint64_t)adev->wb.wb[reg_val_offs + 1 ] << 32ULL;
+	amdgpu_device_wb_free(adev, reg_val_offs);
+	return value;
+
+failed_undo:
+	amdgpu_ring_undo(ring);
+failed_unlock:
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+failed_kiq_read:
+	if (reg_val_offs)
+		amdgpu_device_wb_free(adev, reg_val_offs);
+	pr_err("failed to read gpu clock\n");
+	return ~0;
+}
+
 static uint64_t gfx_v9_0_get_gpu_clock_counter(struct amdgpu_device *adev)
 {
-	uint64_t clock;
+	uint64_t clock, clock_lo, clock_hi, hi_check;
 
-	mutex_lock(&adev->gfx.gpu_clock_mutex);
-	WREG32_SOC15(GC, 0, mmRLC_CAPTURE_GPU_CLOCK_COUNT, 1);
-	clock = (uint64_t)RREG32_SOC15(GC, 0, mmRLC_GPU_CLOCK_COUNT_LSB) |
-		((uint64_t)RREG32_SOC15(GC, 0, mmRLC_GPU_CLOCK_COUNT_MSB) << 32ULL);
-	mutex_unlock(&adev->gfx.gpu_clock_mutex);
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 3, 0):
+		preempt_disable();
+		clock_hi = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_UPPER_Renoir);
+		clock_lo = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_LOWER_Renoir);
+		hi_check = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_UPPER_Renoir);
+		/* The SMUIO TSC clock frequency is 100MHz, which sets 32-bit carry over
+		 * roughly every 42 seconds.
+		 */
+		if (hi_check != clock_hi) {
+			clock_lo = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_LOWER_Renoir);
+			clock_hi = hi_check;
+		}
+		preempt_enable();
+		clock = clock_lo | (clock_hi << 32ULL);
+		break;
+	default:
+		amdgpu_gfx_off_ctrl(adev, false);
+		mutex_lock(&adev->gfx.gpu_clock_mutex);
+		if (amdgpu_ip_version(adev, GC_HWIP, 0) ==
+			    IP_VERSION(9, 0, 1) &&
+		    amdgpu_sriov_runtime(adev)) {
+			clock = gfx_v9_0_kiq_read_clock(adev);
+		} else {
+			WREG32_SOC15(GC, 0, mmRLC_CAPTURE_GPU_CLOCK_COUNT, 1);
+			clock = (uint64_t)RREG32_SOC15(GC, 0, mmRLC_GPU_CLOCK_COUNT_LSB) |
+				((uint64_t)RREG32_SOC15(GC, 0, mmRLC_GPU_CLOCK_COUNT_MSB) << 32ULL);
+		}
+		mutex_unlock(&adev->gfx.gpu_clock_mutex);
+		amdgpu_gfx_off_ctrl(adev, true);
+		break;
+	}
 	return clock;
 }
 
@@ -3025,53 +4311,547 @@ static void gfx_v9_0_ring_emit_gds_switch(struct amdgpu_ring *ring,
 					  uint32_t gws_base, uint32_t gws_size,
 					  uint32_t oa_base, uint32_t oa_size)
 {
-	gds_base = gds_base >> AMDGPU_GDS_SHIFT;
-	gds_size = gds_size >> AMDGPU_GDS_SHIFT;
-
-	gws_base = gws_base >> AMDGPU_GWS_SHIFT;
-	gws_size = gws_size >> AMDGPU_GWS_SHIFT;
-
-	oa_base = oa_base >> AMDGPU_OA_SHIFT;
-	oa_size = oa_size >> AMDGPU_OA_SHIFT;
+	struct amdgpu_device *adev = ring->adev;
 
 	/* GDS Base */
 	gfx_v9_0_write_data_to_reg(ring, 0, false,
-				   amdgpu_gds_reg_offset[vmid].mem_base,
+				   SOC15_REG_OFFSET(GC, 0, mmGDS_VMID0_BASE) + 2 * vmid,
 				   gds_base);
 
 	/* GDS Size */
 	gfx_v9_0_write_data_to_reg(ring, 0, false,
-				   amdgpu_gds_reg_offset[vmid].mem_size,
+				   SOC15_REG_OFFSET(GC, 0, mmGDS_VMID0_SIZE) + 2 * vmid,
 				   gds_size);
 
 	/* GWS */
 	gfx_v9_0_write_data_to_reg(ring, 0, false,
-				   amdgpu_gds_reg_offset[vmid].gws,
+				   SOC15_REG_OFFSET(GC, 0, mmGDS_GWS_VMID0) + vmid,
 				   gws_size << GDS_GWS_VMID0__SIZE__SHIFT | gws_base);
 
 	/* OA */
 	gfx_v9_0_write_data_to_reg(ring, 0, false,
-				   amdgpu_gds_reg_offset[vmid].oa,
+				   SOC15_REG_OFFSET(GC, 0, mmGDS_OA_VMID0) + vmid,
 				   (1 << (oa_size + oa_base)) - (1 << oa_base));
 }
 
-static int gfx_v9_0_early_init(void *handle)
+static const u32 vgpr_init_compute_shader[] =
+{
+	0xb07c0000, 0xbe8000ff,
+	0x000000f8, 0xbf110800,
+	0x7e000280, 0x7e020280,
+	0x7e040280, 0x7e060280,
+	0x7e080280, 0x7e0a0280,
+	0x7e0c0280, 0x7e0e0280,
+	0x80808800, 0xbe803200,
+	0xbf84fff5, 0xbf9c0000,
+	0xd28c0001, 0x0001007f,
+	0xd28d0001, 0x0002027e,
+	0x10020288, 0xb8810904,
+	0xb7814000, 0xd1196a01,
+	0x00000301, 0xbe800087,
+	0xbefc00c1, 0xd89c4000,
+	0x00020201, 0xd89cc080,
+	0x00040401, 0x320202ff,
+	0x00000800, 0x80808100,
+	0xbf84fff8, 0x7e020280,
+	0xbf810000, 0x00000000,
+};
+
+static const u32 sgpr_init_compute_shader[] =
+{
+	0xb07c0000, 0xbe8000ff,
+	0x0000005f, 0xbee50080,
+	0xbe812c65, 0xbe822c65,
+	0xbe832c65, 0xbe842c65,
+	0xbe852c65, 0xb77c0005,
+	0x80808500, 0xbf84fff8,
+	0xbe800080, 0xbf810000,
+};
+
+static const u32 vgpr_init_compute_shader_arcturus[] = {
+	0xd3d94000, 0x18000080, 0xd3d94001, 0x18000080, 0xd3d94002, 0x18000080,
+	0xd3d94003, 0x18000080, 0xd3d94004, 0x18000080, 0xd3d94005, 0x18000080,
+	0xd3d94006, 0x18000080, 0xd3d94007, 0x18000080, 0xd3d94008, 0x18000080,
+	0xd3d94009, 0x18000080, 0xd3d9400a, 0x18000080, 0xd3d9400b, 0x18000080,
+	0xd3d9400c, 0x18000080, 0xd3d9400d, 0x18000080, 0xd3d9400e, 0x18000080,
+	0xd3d9400f, 0x18000080, 0xd3d94010, 0x18000080, 0xd3d94011, 0x18000080,
+	0xd3d94012, 0x18000080, 0xd3d94013, 0x18000080, 0xd3d94014, 0x18000080,
+	0xd3d94015, 0x18000080, 0xd3d94016, 0x18000080, 0xd3d94017, 0x18000080,
+	0xd3d94018, 0x18000080, 0xd3d94019, 0x18000080, 0xd3d9401a, 0x18000080,
+	0xd3d9401b, 0x18000080, 0xd3d9401c, 0x18000080, 0xd3d9401d, 0x18000080,
+	0xd3d9401e, 0x18000080, 0xd3d9401f, 0x18000080, 0xd3d94020, 0x18000080,
+	0xd3d94021, 0x18000080, 0xd3d94022, 0x18000080, 0xd3d94023, 0x18000080,
+	0xd3d94024, 0x18000080, 0xd3d94025, 0x18000080, 0xd3d94026, 0x18000080,
+	0xd3d94027, 0x18000080, 0xd3d94028, 0x18000080, 0xd3d94029, 0x18000080,
+	0xd3d9402a, 0x18000080, 0xd3d9402b, 0x18000080, 0xd3d9402c, 0x18000080,
+	0xd3d9402d, 0x18000080, 0xd3d9402e, 0x18000080, 0xd3d9402f, 0x18000080,
+	0xd3d94030, 0x18000080, 0xd3d94031, 0x18000080, 0xd3d94032, 0x18000080,
+	0xd3d94033, 0x18000080, 0xd3d94034, 0x18000080, 0xd3d94035, 0x18000080,
+	0xd3d94036, 0x18000080, 0xd3d94037, 0x18000080, 0xd3d94038, 0x18000080,
+	0xd3d94039, 0x18000080, 0xd3d9403a, 0x18000080, 0xd3d9403b, 0x18000080,
+	0xd3d9403c, 0x18000080, 0xd3d9403d, 0x18000080, 0xd3d9403e, 0x18000080,
+	0xd3d9403f, 0x18000080, 0xd3d94040, 0x18000080, 0xd3d94041, 0x18000080,
+	0xd3d94042, 0x18000080, 0xd3d94043, 0x18000080, 0xd3d94044, 0x18000080,
+	0xd3d94045, 0x18000080, 0xd3d94046, 0x18000080, 0xd3d94047, 0x18000080,
+	0xd3d94048, 0x18000080, 0xd3d94049, 0x18000080, 0xd3d9404a, 0x18000080,
+	0xd3d9404b, 0x18000080, 0xd3d9404c, 0x18000080, 0xd3d9404d, 0x18000080,
+	0xd3d9404e, 0x18000080, 0xd3d9404f, 0x18000080, 0xd3d94050, 0x18000080,
+	0xd3d94051, 0x18000080, 0xd3d94052, 0x18000080, 0xd3d94053, 0x18000080,
+	0xd3d94054, 0x18000080, 0xd3d94055, 0x18000080, 0xd3d94056, 0x18000080,
+	0xd3d94057, 0x18000080, 0xd3d94058, 0x18000080, 0xd3d94059, 0x18000080,
+	0xd3d9405a, 0x18000080, 0xd3d9405b, 0x18000080, 0xd3d9405c, 0x18000080,
+	0xd3d9405d, 0x18000080, 0xd3d9405e, 0x18000080, 0xd3d9405f, 0x18000080,
+	0xd3d94060, 0x18000080, 0xd3d94061, 0x18000080, 0xd3d94062, 0x18000080,
+	0xd3d94063, 0x18000080, 0xd3d94064, 0x18000080, 0xd3d94065, 0x18000080,
+	0xd3d94066, 0x18000080, 0xd3d94067, 0x18000080, 0xd3d94068, 0x18000080,
+	0xd3d94069, 0x18000080, 0xd3d9406a, 0x18000080, 0xd3d9406b, 0x18000080,
+	0xd3d9406c, 0x18000080, 0xd3d9406d, 0x18000080, 0xd3d9406e, 0x18000080,
+	0xd3d9406f, 0x18000080, 0xd3d94070, 0x18000080, 0xd3d94071, 0x18000080,
+	0xd3d94072, 0x18000080, 0xd3d94073, 0x18000080, 0xd3d94074, 0x18000080,
+	0xd3d94075, 0x18000080, 0xd3d94076, 0x18000080, 0xd3d94077, 0x18000080,
+	0xd3d94078, 0x18000080, 0xd3d94079, 0x18000080, 0xd3d9407a, 0x18000080,
+	0xd3d9407b, 0x18000080, 0xd3d9407c, 0x18000080, 0xd3d9407d, 0x18000080,
+	0xd3d9407e, 0x18000080, 0xd3d9407f, 0x18000080, 0xd3d94080, 0x18000080,
+	0xd3d94081, 0x18000080, 0xd3d94082, 0x18000080, 0xd3d94083, 0x18000080,
+	0xd3d94084, 0x18000080, 0xd3d94085, 0x18000080, 0xd3d94086, 0x18000080,
+	0xd3d94087, 0x18000080, 0xd3d94088, 0x18000080, 0xd3d94089, 0x18000080,
+	0xd3d9408a, 0x18000080, 0xd3d9408b, 0x18000080, 0xd3d9408c, 0x18000080,
+	0xd3d9408d, 0x18000080, 0xd3d9408e, 0x18000080, 0xd3d9408f, 0x18000080,
+	0xd3d94090, 0x18000080, 0xd3d94091, 0x18000080, 0xd3d94092, 0x18000080,
+	0xd3d94093, 0x18000080, 0xd3d94094, 0x18000080, 0xd3d94095, 0x18000080,
+	0xd3d94096, 0x18000080, 0xd3d94097, 0x18000080, 0xd3d94098, 0x18000080,
+	0xd3d94099, 0x18000080, 0xd3d9409a, 0x18000080, 0xd3d9409b, 0x18000080,
+	0xd3d9409c, 0x18000080, 0xd3d9409d, 0x18000080, 0xd3d9409e, 0x18000080,
+	0xd3d9409f, 0x18000080, 0xd3d940a0, 0x18000080, 0xd3d940a1, 0x18000080,
+	0xd3d940a2, 0x18000080, 0xd3d940a3, 0x18000080, 0xd3d940a4, 0x18000080,
+	0xd3d940a5, 0x18000080, 0xd3d940a6, 0x18000080, 0xd3d940a7, 0x18000080,
+	0xd3d940a8, 0x18000080, 0xd3d940a9, 0x18000080, 0xd3d940aa, 0x18000080,
+	0xd3d940ab, 0x18000080, 0xd3d940ac, 0x18000080, 0xd3d940ad, 0x18000080,
+	0xd3d940ae, 0x18000080, 0xd3d940af, 0x18000080, 0xd3d940b0, 0x18000080,
+	0xd3d940b1, 0x18000080, 0xd3d940b2, 0x18000080, 0xd3d940b3, 0x18000080,
+	0xd3d940b4, 0x18000080, 0xd3d940b5, 0x18000080, 0xd3d940b6, 0x18000080,
+	0xd3d940b7, 0x18000080, 0xd3d940b8, 0x18000080, 0xd3d940b9, 0x18000080,
+	0xd3d940ba, 0x18000080, 0xd3d940bb, 0x18000080, 0xd3d940bc, 0x18000080,
+	0xd3d940bd, 0x18000080, 0xd3d940be, 0x18000080, 0xd3d940bf, 0x18000080,
+	0xd3d940c0, 0x18000080, 0xd3d940c1, 0x18000080, 0xd3d940c2, 0x18000080,
+	0xd3d940c3, 0x18000080, 0xd3d940c4, 0x18000080, 0xd3d940c5, 0x18000080,
+	0xd3d940c6, 0x18000080, 0xd3d940c7, 0x18000080, 0xd3d940c8, 0x18000080,
+	0xd3d940c9, 0x18000080, 0xd3d940ca, 0x18000080, 0xd3d940cb, 0x18000080,
+	0xd3d940cc, 0x18000080, 0xd3d940cd, 0x18000080, 0xd3d940ce, 0x18000080,
+	0xd3d940cf, 0x18000080, 0xd3d940d0, 0x18000080, 0xd3d940d1, 0x18000080,
+	0xd3d940d2, 0x18000080, 0xd3d940d3, 0x18000080, 0xd3d940d4, 0x18000080,
+	0xd3d940d5, 0x18000080, 0xd3d940d6, 0x18000080, 0xd3d940d7, 0x18000080,
+	0xd3d940d8, 0x18000080, 0xd3d940d9, 0x18000080, 0xd3d940da, 0x18000080,
+	0xd3d940db, 0x18000080, 0xd3d940dc, 0x18000080, 0xd3d940dd, 0x18000080,
+	0xd3d940de, 0x18000080, 0xd3d940df, 0x18000080, 0xd3d940e0, 0x18000080,
+	0xd3d940e1, 0x18000080, 0xd3d940e2, 0x18000080, 0xd3d940e3, 0x18000080,
+	0xd3d940e4, 0x18000080, 0xd3d940e5, 0x18000080, 0xd3d940e6, 0x18000080,
+	0xd3d940e7, 0x18000080, 0xd3d940e8, 0x18000080, 0xd3d940e9, 0x18000080,
+	0xd3d940ea, 0x18000080, 0xd3d940eb, 0x18000080, 0xd3d940ec, 0x18000080,
+	0xd3d940ed, 0x18000080, 0xd3d940ee, 0x18000080, 0xd3d940ef, 0x18000080,
+	0xd3d940f0, 0x18000080, 0xd3d940f1, 0x18000080, 0xd3d940f2, 0x18000080,
+	0xd3d940f3, 0x18000080, 0xd3d940f4, 0x18000080, 0xd3d940f5, 0x18000080,
+	0xd3d940f6, 0x18000080, 0xd3d940f7, 0x18000080, 0xd3d940f8, 0x18000080,
+	0xd3d940f9, 0x18000080, 0xd3d940fa, 0x18000080, 0xd3d940fb, 0x18000080,
+	0xd3d940fc, 0x18000080, 0xd3d940fd, 0x18000080, 0xd3d940fe, 0x18000080,
+	0xd3d940ff, 0x18000080, 0xb07c0000, 0xbe8a00ff, 0x000000f8, 0xbf11080a,
+	0x7e000280, 0x7e020280, 0x7e040280, 0x7e060280, 0x7e080280, 0x7e0a0280,
+	0x7e0c0280, 0x7e0e0280, 0x808a880a, 0xbe80320a, 0xbf84fff5, 0xbf9c0000,
+	0xd28c0001, 0x0001007f, 0xd28d0001, 0x0002027e, 0x10020288, 0xb88b0904,
+	0xb78b4000, 0xd1196a01, 0x00001701, 0xbe8a0087, 0xbefc00c1, 0xd89c4000,
+	0x00020201, 0xd89cc080, 0x00040401, 0x320202ff, 0x00000800, 0x808a810a,
+	0xbf84fff8, 0xbf810000,
+};
+
+/* When below register arrays changed, please update gpr_reg_size,
+  and sec_ded_counter_reg_size in function gfx_v9_0_do_edc_gpr_workarounds,
+  to cover all gfx9 ASICs */
+static const struct soc15_reg_entry vgpr_init_regs[] = {
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_RESOURCE_LIMITS), 0x0000000 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_X), 0x40 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Y), 4 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Z), 1 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC1), 0x3f },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC2), 0x400000 },  /* 64KB LDS */
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE0), 0xffffffff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE1), 0xffffffff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE2), 0xffffffff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE3), 0xffffffff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE4), 0xffffffff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE5), 0xffffffff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE6), 0xffffffff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE7), 0xffffffff },
+};
+
+static const struct soc15_reg_entry vgpr_init_regs_arcturus[] = {
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_RESOURCE_LIMITS), 0x0000000 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_X), 0x40 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Y), 4 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Z), 1 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC1), 0xbf },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC2), 0x400000 },  /* 64KB LDS */
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE0), 0xffffffff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE1), 0xffffffff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE2), 0xffffffff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE3), 0xffffffff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE4), 0xffffffff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE5), 0xffffffff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE6), 0xffffffff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE7), 0xffffffff },
+};
+
+static const struct soc15_reg_entry sgpr1_init_regs[] = {
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_RESOURCE_LIMITS), 0x0000000 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_X), 0x40 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Y), 8 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Z), 1 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC1), 0x240 }, /* (80 GPRS) */
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC2), 0x0 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE0), 0x000000ff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE1), 0x000000ff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE2), 0x000000ff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE3), 0x000000ff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE4), 0x000000ff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE5), 0x000000ff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE6), 0x000000ff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE7), 0x000000ff },
+};
+
+static const struct soc15_reg_entry sgpr2_init_regs[] = {
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_RESOURCE_LIMITS), 0x0000000 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_X), 0x40 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Y), 8 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Z), 1 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC1), 0x240 }, /* (80 GPRS) */
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC2), 0x0 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE0), 0x0000ff00 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE1), 0x0000ff00 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE2), 0x0000ff00 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE3), 0x0000ff00 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE4), 0x0000ff00 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE5), 0x0000ff00 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE6), 0x0000ff00 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE7), 0x0000ff00 },
+};
+
+static const struct soc15_reg_entry gfx_v9_0_edc_counter_regs[] = {
+   { SOC15_REG_ENTRY(GC, 0, mmCPC_EDC_SCRATCH_CNT), 0, 1, 1},
+   { SOC15_REG_ENTRY(GC, 0, mmCPC_EDC_UCODE_CNT), 0, 1, 1},
+   { SOC15_REG_ENTRY(GC, 0, mmCPF_EDC_ROQ_CNT), 0, 1, 1},
+   { SOC15_REG_ENTRY(GC, 0, mmCPF_EDC_TAG_CNT), 0, 1, 1},
+   { SOC15_REG_ENTRY(GC, 0, mmCPG_EDC_DMA_CNT), 0, 1, 1},
+   { SOC15_REG_ENTRY(GC, 0, mmCPG_EDC_TAG_CNT), 0, 1, 1},
+   { SOC15_REG_ENTRY(GC, 0, mmDC_EDC_CSINVOC_CNT), 0, 1, 1},
+   { SOC15_REG_ENTRY(GC, 0, mmDC_EDC_RESTORE_CNT), 0, 1, 1},
+   { SOC15_REG_ENTRY(GC, 0, mmDC_EDC_STATE_CNT), 0, 1, 1},
+   { SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_CNT), 0, 1, 1},
+   { SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_GRBM_CNT), 0, 1, 1},
+   { SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_DED), 0, 1, 1},
+   { SOC15_REG_ENTRY(GC, 0, mmSPI_EDC_CNT), 0, 4, 1},
+   { SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT), 0, 4, 6},
+   { SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_DED_CNT), 0, 4, 16},
+   { SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_INFO), 0, 4, 16},
+   { SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_SEC_CNT), 0, 4, 16},
+   { SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT), 0, 1, 16},
+   { SOC15_REG_ENTRY(GC, 0, mmTCP_ATC_EDC_GATCL1_CNT), 0, 4, 16},
+   { SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT), 0, 4, 16},
+   { SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW), 0, 4, 16},
+   { SOC15_REG_ENTRY(GC, 0, mmTD_EDC_CNT), 0, 4, 16},
+   { SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2), 0, 4, 6},
+   { SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT), 0, 4, 16},
+   { SOC15_REG_ENTRY(GC, 0, mmTA_EDC_CNT), 0, 4, 16},
+   { SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PHY_CNT), 0, 1, 1},
+   { SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PIPE_CNT), 0, 1, 1},
+   { SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT), 0, 1, 32},
+   { SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2), 0, 1, 32},
+   { SOC15_REG_ENTRY(GC, 0, mmTCI_EDC_CNT), 0, 1, 72},
+   { SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2), 0, 1, 16},
+   { SOC15_REG_ENTRY(GC, 0, mmTCA_EDC_CNT), 0, 1, 2},
+   { SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3), 0, 4, 6},
+};
+
+static int gfx_v9_0_do_edc_gds_workarounds(struct amdgpu_device *adev)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_ring *ring = &adev->gfx.compute_ring[0];
+	int i, r;
+
+	/* only support when RAS is enabled */
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
+		return 0;
+
+	r = amdgpu_ring_alloc(ring, 7);
+	if (r) {
+		DRM_ERROR("amdgpu: GDS workarounds failed to lock ring %s (%d).\n",
+			ring->name, r);
+		return r;
+	}
+
+	WREG32_SOC15(GC, 0, mmGDS_VMID0_BASE, 0x00000000);
+	WREG32_SOC15(GC, 0, mmGDS_VMID0_SIZE, adev->gds.gds_size);
 
-	adev->gfx.num_gfx_rings = GFX9_NUM_GFX_RINGS;
-	adev->gfx.num_compute_rings = AMDGPU_MAX_COMPUTE_RINGS;
+	amdgpu_ring_write(ring, PACKET3(PACKET3_DMA_DATA, 5));
+	amdgpu_ring_write(ring, (PACKET3_DMA_DATA_CP_SYNC |
+				PACKET3_DMA_DATA_DST_SEL(1) |
+				PACKET3_DMA_DATA_SRC_SEL(2) |
+				PACKET3_DMA_DATA_ENGINE(0)));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, PACKET3_DMA_DATA_CMD_RAW_WAIT |
+				adev->gds.gds_size);
+
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (ring->wptr == gfx_v9_0_ring_get_rptr_compute(ring))
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+
+	WREG32_SOC15(GC, 0, mmGDS_VMID0_SIZE, 0x00000000);
+
+	return r;
+}
+
+static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = &adev->gfx.compute_ring[0];
+	struct amdgpu_ib ib;
+	struct dma_fence *f = NULL;
+	int r, i;
+	unsigned total_size, vgpr_offset, sgpr_offset;
+	u64 gpu_addr;
+
+	int compute_dim_x = adev->gfx.config.max_shader_engines *
+						adev->gfx.config.max_cu_per_sh *
+						adev->gfx.config.max_sh_per_se;
+	int sgpr_work_group_size = 5;
+	int gpr_reg_size = adev->gfx.config.max_shader_engines + 6;
+	int vgpr_init_shader_size;
+	const u32 *vgpr_init_shader_ptr;
+	const struct soc15_reg_entry *vgpr_init_regs_ptr;
+
+	/* only support when RAS is enabled */
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
+		return 0;
+
+	/* bail if the compute ring is not ready */
+	if (!ring->sched.ready)
+		return 0;
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 1)) {
+		vgpr_init_shader_ptr = vgpr_init_compute_shader_arcturus;
+		vgpr_init_shader_size = sizeof(vgpr_init_compute_shader_arcturus);
+		vgpr_init_regs_ptr = vgpr_init_regs_arcturus;
+	} else {
+		vgpr_init_shader_ptr = vgpr_init_compute_shader;
+		vgpr_init_shader_size = sizeof(vgpr_init_compute_shader);
+		vgpr_init_regs_ptr = vgpr_init_regs;
+	}
+
+	total_size =
+		(gpr_reg_size * 3 + 4 + 5 + 2) * 4; /* VGPRS */
+	total_size +=
+		(gpr_reg_size * 3 + 4 + 5 + 2) * 4; /* SGPRS1 */
+	total_size +=
+		(gpr_reg_size * 3 + 4 + 5 + 2) * 4; /* SGPRS2 */
+	total_size = ALIGN(total_size, 256);
+	vgpr_offset = total_size;
+	total_size += ALIGN(vgpr_init_shader_size, 256);
+	sgpr_offset = total_size;
+	total_size += sizeof(sgpr_init_compute_shader);
+
+	/* allocate an indirect buffer to put the commands in */
+	memset(&ib, 0, sizeof(ib));
+	r = amdgpu_ib_get(adev, NULL, total_size,
+					AMDGPU_IB_POOL_DIRECT, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%d).\n", r);
+		return r;
+	}
+
+	/* load the compute shaders */
+	for (i = 0; i < vgpr_init_shader_size/sizeof(u32); i++)
+		ib.ptr[i + (vgpr_offset / 4)] = vgpr_init_shader_ptr[i];
+
+	for (i = 0; i < ARRAY_SIZE(sgpr_init_compute_shader); i++)
+		ib.ptr[i + (sgpr_offset / 4)] = sgpr_init_compute_shader[i];
+
+	/* init the ib length to 0 */
+	ib.length_dw = 0;
+
+	/* VGPR */
+	/* write the register state for the compute dispatch */
+	for (i = 0; i < gpr_reg_size; i++) {
+		ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 1);
+		ib.ptr[ib.length_dw++] = SOC15_REG_ENTRY_OFFSET(vgpr_init_regs_ptr[i])
+								- PACKET3_SET_SH_REG_START;
+		ib.ptr[ib.length_dw++] = vgpr_init_regs_ptr[i].reg_value;
+	}
+	/* write the shader start address: mmCOMPUTE_PGM_LO, mmCOMPUTE_PGM_HI */
+	gpu_addr = (ib.gpu_addr + (u64)vgpr_offset) >> 8;
+	ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 2);
+	ib.ptr[ib.length_dw++] = SOC15_REG_OFFSET(GC, 0, mmCOMPUTE_PGM_LO)
+							- PACKET3_SET_SH_REG_START;
+	ib.ptr[ib.length_dw++] = lower_32_bits(gpu_addr);
+	ib.ptr[ib.length_dw++] = upper_32_bits(gpu_addr);
+
+	/* write dispatch packet */
+	ib.ptr[ib.length_dw++] = PACKET3(PACKET3_DISPATCH_DIRECT, 3);
+	ib.ptr[ib.length_dw++] = compute_dim_x * 2; /* x */
+	ib.ptr[ib.length_dw++] = 1; /* y */
+	ib.ptr[ib.length_dw++] = 1; /* z */
+	ib.ptr[ib.length_dw++] =
+		REG_SET_FIELD(0, COMPUTE_DISPATCH_INITIATOR, COMPUTE_SHADER_EN, 1);
+
+	/* write CS partial flush packet */
+	ib.ptr[ib.length_dw++] = PACKET3(PACKET3_EVENT_WRITE, 0);
+	ib.ptr[ib.length_dw++] = EVENT_TYPE(7) | EVENT_INDEX(4);
+
+	/* SGPR1 */
+	/* write the register state for the compute dispatch */
+	for (i = 0; i < gpr_reg_size; i++) {
+		ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 1);
+		ib.ptr[ib.length_dw++] = SOC15_REG_ENTRY_OFFSET(sgpr1_init_regs[i])
+								- PACKET3_SET_SH_REG_START;
+		ib.ptr[ib.length_dw++] = sgpr1_init_regs[i].reg_value;
+	}
+	/* write the shader start address: mmCOMPUTE_PGM_LO, mmCOMPUTE_PGM_HI */
+	gpu_addr = (ib.gpu_addr + (u64)sgpr_offset) >> 8;
+	ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 2);
+	ib.ptr[ib.length_dw++] = SOC15_REG_OFFSET(GC, 0, mmCOMPUTE_PGM_LO)
+							- PACKET3_SET_SH_REG_START;
+	ib.ptr[ib.length_dw++] = lower_32_bits(gpu_addr);
+	ib.ptr[ib.length_dw++] = upper_32_bits(gpu_addr);
+
+	/* write dispatch packet */
+	ib.ptr[ib.length_dw++] = PACKET3(PACKET3_DISPATCH_DIRECT, 3);
+	ib.ptr[ib.length_dw++] = compute_dim_x / 2 * sgpr_work_group_size; /* x */
+	ib.ptr[ib.length_dw++] = 1; /* y */
+	ib.ptr[ib.length_dw++] = 1; /* z */
+	ib.ptr[ib.length_dw++] =
+		REG_SET_FIELD(0, COMPUTE_DISPATCH_INITIATOR, COMPUTE_SHADER_EN, 1);
+
+	/* write CS partial flush packet */
+	ib.ptr[ib.length_dw++] = PACKET3(PACKET3_EVENT_WRITE, 0);
+	ib.ptr[ib.length_dw++] = EVENT_TYPE(7) | EVENT_INDEX(4);
+
+	/* SGPR2 */
+	/* write the register state for the compute dispatch */
+	for (i = 0; i < gpr_reg_size; i++) {
+		ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 1);
+		ib.ptr[ib.length_dw++] = SOC15_REG_ENTRY_OFFSET(sgpr2_init_regs[i])
+								- PACKET3_SET_SH_REG_START;
+		ib.ptr[ib.length_dw++] = sgpr2_init_regs[i].reg_value;
+	}
+	/* write the shader start address: mmCOMPUTE_PGM_LO, mmCOMPUTE_PGM_HI */
+	gpu_addr = (ib.gpu_addr + (u64)sgpr_offset) >> 8;
+	ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 2);
+	ib.ptr[ib.length_dw++] = SOC15_REG_OFFSET(GC, 0, mmCOMPUTE_PGM_LO)
+							- PACKET3_SET_SH_REG_START;
+	ib.ptr[ib.length_dw++] = lower_32_bits(gpu_addr);
+	ib.ptr[ib.length_dw++] = upper_32_bits(gpu_addr);
+
+	/* write dispatch packet */
+	ib.ptr[ib.length_dw++] = PACKET3(PACKET3_DISPATCH_DIRECT, 3);
+	ib.ptr[ib.length_dw++] = compute_dim_x / 2 * sgpr_work_group_size; /* x */
+	ib.ptr[ib.length_dw++] = 1; /* y */
+	ib.ptr[ib.length_dw++] = 1; /* z */
+	ib.ptr[ib.length_dw++] =
+		REG_SET_FIELD(0, COMPUTE_DISPATCH_INITIATOR, COMPUTE_SHADER_EN, 1);
+
+	/* write CS partial flush packet */
+	ib.ptr[ib.length_dw++] = PACKET3(PACKET3_EVENT_WRITE, 0);
+	ib.ptr[ib.length_dw++] = EVENT_TYPE(7) | EVENT_INDEX(4);
+
+	/* shedule the ib on the ring */
+	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
+	if (r) {
+		DRM_ERROR("amdgpu: ib submit failed (%d).\n", r);
+		goto fail;
+	}
+
+	/* wait for the GPU to finish processing the IB */
+	r = dma_fence_wait(f, false);
+	if (r) {
+		DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
+		goto fail;
+	}
+
+fail:
+	amdgpu_ib_free(&ib, NULL);
+	dma_fence_put(f);
+
+	return r;
+}
+
+static int gfx_v9_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	adev->gfx.funcs = &gfx_v9_0_gfx_funcs;
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 1) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2))
+		adev->gfx.num_gfx_rings = 0;
+	else
+		adev->gfx.num_gfx_rings = GFX9_NUM_GFX_RINGS;
+	adev->gfx.xcc_mask = 1;
+	adev->gfx.num_compute_rings = min(amdgpu_gfx_get_num_kcq(adev),
+					  AMDGPU_MAX_COMPUTE_RINGS);
+	gfx_v9_0_set_kiq_pm4_funcs(adev);
 	gfx_v9_0_set_ring_funcs(adev);
 	gfx_v9_0_set_irq_funcs(adev);
 	gfx_v9_0_set_gds_init(adev);
 	gfx_v9_0_set_rlc_funcs(adev);
 
+	/* init rlcg reg access ctrl */
+	gfx_v9_0_init_rlcg_reg_access_ctrl(adev);
+
+	return gfx_v9_0_init_microcode(adev);
+}
+
+static int gfx_v9_0_ecc_late_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	/*
+	 * Temp workaround to fix the issue that CP firmware fails to
+	 * update read pointer when CPDMA is writing clearing operation
+	 * to GDS in suspend/resume sequence on several cards. So just
+	 * limit this operation in cold boot sequence.
+	 */
+	if ((!adev->in_suspend) &&
+	    (adev->gds.gds_size)) {
+		r = gfx_v9_0_do_edc_gds_workarounds(adev);
+		if (r)
+			return r;
+	}
+
+	/* requires IBs so do in late init after IB pool is initialized */
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2))
+		r = gfx_v9_4_2_do_edc_gpr_workarounds(adev);
+	else
+		r = gfx_v9_0_do_edc_gpr_workarounds(adev);
+
+	if (r)
+		return r;
+
+	if (adev->gfx.ras &&
+	    adev->gfx.ras->enable_watchdog_timer)
+		adev->gfx.ras->enable_watchdog_timer(adev);
+
 	return 0;
 }
 
-static int gfx_v9_0_late_init(void *handle)
+static int gfx_v9_0_late_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int r;
 
 	r = amdgpu_irq_get(adev, &adev->gfx.priv_reg_irq, 0);
@@ -3082,68 +4862,65 @@ static int gfx_v9_0_late_init(void *handle)
 	if (r)
 		return r;
 
+	r = amdgpu_irq_get(adev, &adev->gfx.bad_op_irq, 0);
+	if (r)
+		return r;
+
+	r = gfx_v9_0_ecc_late_init(ip_block);
+	if (r)
+		return r;
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2))
+		gfx_v9_4_2_debug_trap_config_init(adev,
+			adev->vm_manager.first_kfd_vmid, AMDGPU_NUM_VMID);
+	else
+		gfx_v9_0_debug_trap_config_init(adev,
+			adev->vm_manager.first_kfd_vmid, AMDGPU_NUM_VMID);
+
 	return 0;
 }
 
-static void gfx_v9_0_enter_rlc_safe_mode(struct amdgpu_device *adev)
+static bool gfx_v9_0_is_rlc_enabled(struct amdgpu_device *adev)
 {
-	uint32_t rlc_setting, data;
-	unsigned i;
-
-	if (adev->gfx.rlc.in_safe_mode)
-		return;
+	uint32_t rlc_setting;
 
 	/* if RLC is not enabled, do nothing */
 	rlc_setting = RREG32_SOC15(GC, 0, mmRLC_CNTL);
 	if (!(rlc_setting & RLC_CNTL__RLC_ENABLE_F32_MASK))
-		return;
-
-	if (adev->cg_flags &
-	    (AMD_CG_SUPPORT_GFX_CGCG | AMD_CG_SUPPORT_GFX_MGCG |
-	     AMD_CG_SUPPORT_GFX_3D_CGCG)) {
-		data = RLC_SAFE_MODE__CMD_MASK;
-		data |= (1 << RLC_SAFE_MODE__MESSAGE__SHIFT);
-		WREG32_SOC15(GC, 0, mmRLC_SAFE_MODE, data);
+		return false;
 
-		/* wait for RLC_SAFE_MODE */
-		for (i = 0; i < adev->usec_timeout; i++) {
-			if (!REG_GET_FIELD(SOC15_REG_OFFSET(GC, 0, mmRLC_SAFE_MODE), RLC_SAFE_MODE, CMD))
-				break;
-			udelay(1);
-		}
-		adev->gfx.rlc.in_safe_mode = true;
-	}
+	return true;
 }
 
-static void gfx_v9_0_exit_rlc_safe_mode(struct amdgpu_device *adev)
+static void gfx_v9_0_set_safe_mode(struct amdgpu_device *adev, int xcc_id)
 {
-	uint32_t rlc_setting, data;
-
-	if (!adev->gfx.rlc.in_safe_mode)
-		return;
+	uint32_t data;
+	unsigned i;
 
-	/* if RLC is not enabled, do nothing */
-	rlc_setting = RREG32_SOC15(GC, 0, mmRLC_CNTL);
-	if (!(rlc_setting & RLC_CNTL__RLC_ENABLE_F32_MASK))
-		return;
+	data = RLC_SAFE_MODE__CMD_MASK;
+	data |= (1 << RLC_SAFE_MODE__MESSAGE__SHIFT);
+	WREG32_SOC15(GC, 0, mmRLC_SAFE_MODE, data);
 
-	if (adev->cg_flags &
-	    (AMD_CG_SUPPORT_GFX_CGCG | AMD_CG_SUPPORT_GFX_MGCG)) {
-		/*
-		 * Try to exit safe mode only if it is already in safe
-		 * mode.
-		 */
-		data = RLC_SAFE_MODE__CMD_MASK;
-		WREG32_SOC15(GC, 0, mmRLC_SAFE_MODE, data);
-		adev->gfx.rlc.in_safe_mode = false;
+	/* wait for RLC_SAFE_MODE */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (!REG_GET_FIELD(RREG32_SOC15(GC, 0, mmRLC_SAFE_MODE), RLC_SAFE_MODE, CMD))
+			break;
+		udelay(1);
 	}
 }
 
+static void gfx_v9_0_unset_safe_mode(struct amdgpu_device *adev, int xcc_id)
+{
+	uint32_t data;
+
+	data = RLC_SAFE_MODE__CMD_MASK;
+	WREG32_SOC15(GC, 0, mmRLC_SAFE_MODE, data);
+}
+
 static void gfx_v9_0_update_gfx_cg_power_gating(struct amdgpu_device *adev,
 						bool enable)
 {
-	/* TODO: double check if we need to perform under safe mdoe */
-	/* gfx_v9_0_enter_rlc_safe_mode(adev); */
+	amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
 
 	if ((adev->pg_flags & AMD_PG_SUPPORT_GFX_PG) && enable) {
 		gfx_v9_0_enable_gfx_cg_power_gating(adev, true);
@@ -3151,10 +4928,11 @@ static void gfx_v9_0_update_gfx_cg_power_gating(struct amdgpu_device *adev,
 			gfx_v9_0_enable_gfx_pipeline_powergating(adev, true);
 	} else {
 		gfx_v9_0_enable_gfx_cg_power_gating(adev, false);
-		gfx_v9_0_enable_gfx_pipeline_powergating(adev, false);
+		if (adev->pg_flags & AMD_PG_SUPPORT_GFX_PIPELINE)
+			gfx_v9_0_enable_gfx_pipeline_powergating(adev, false);
 	}
 
-	/* gfx_v9_0_exit_rlc_safe_mode(adev); */
+	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
 }
 
 static void gfx_v9_0_update_gfx_mg_power_gating(struct amdgpu_device *adev,
@@ -3185,8 +4963,11 @@ static void gfx_v9_0_update_medium_grain_clock_gating(struct amdgpu_device *adev
 	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGCG)) {
 		/* 1 - RLC_CGTT_MGCG_OVERRIDE */
 		def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE);
-		data &= ~(RLC_CGTT_MGCG_OVERRIDE__CPF_CGTT_SCLK_OVERRIDE_MASK |
-			  RLC_CGTT_MGCG_OVERRIDE__GRBM_CGTT_SCLK_OVERRIDE_MASK |
+
+		if (amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 2, 1))
+			data &= ~RLC_CGTT_MGCG_OVERRIDE__CPF_CGTT_SCLK_OVERRIDE_MASK;
+
+		data &= ~(RLC_CGTT_MGCG_OVERRIDE__GRBM_CGTT_SCLK_OVERRIDE_MASK |
 			  RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGCG_OVERRIDE_MASK |
 			  RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGLS_OVERRIDE_MASK);
 
@@ -3216,11 +4997,15 @@ static void gfx_v9_0_update_medium_grain_clock_gating(struct amdgpu_device *adev
 	} else {
 		/* 1 - MGCG_OVERRIDE */
 		def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE);
-		data |= (RLC_CGTT_MGCG_OVERRIDE__CPF_CGTT_SCLK_OVERRIDE_MASK |
-			 RLC_CGTT_MGCG_OVERRIDE__RLC_CGTT_SCLK_OVERRIDE_MASK |
+
+		if (amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 2, 1))
+			data |= RLC_CGTT_MGCG_OVERRIDE__CPF_CGTT_SCLK_OVERRIDE_MASK;
+
+		data |= (RLC_CGTT_MGCG_OVERRIDE__RLC_CGTT_SCLK_OVERRIDE_MASK |
 			 RLC_CGTT_MGCG_OVERRIDE__GRBM_CGTT_SCLK_OVERRIDE_MASK |
 			 RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGCG_OVERRIDE_MASK |
 			 RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGLS_OVERRIDE_MASK);
+
 		if (def != data)
 			WREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE, data);
 
@@ -3245,10 +5030,11 @@ static void gfx_v9_0_update_3d_clock_gating(struct amdgpu_device *adev,
 {
 	uint32_t data, def;
 
-	adev->gfx.rlc.funcs->enter_safe_mode(adev);
+	if (!adev->gfx.num_gfx_rings)
+		return;
 
 	/* Enable 3D CGCG/CGLS */
-	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGCG)) {
+	if (enable) {
 		/* write cmd to clear cgcg/cgls ov */
 		def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE);
 		/* unset CGCG override */
@@ -3256,10 +5042,16 @@ static void gfx_v9_0_update_3d_clock_gating(struct amdgpu_device *adev,
 		/* update CGCG and CGLS override bits */
 		if (def != data)
 			WREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE, data);
-		/* enable 3Dcgcg FSM(0x0020003f) */
+
+		/* enable 3Dcgcg FSM(0x0000363f) */
 		def = RREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL_3D);
-		data = (0x2000 << RLC_CGCG_CGLS_CTRL_3D__CGCG_GFX_IDLE_THRESHOLD__SHIFT) |
-			RLC_CGCG_CGLS_CTRL_3D__CGCG_EN_MASK;
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGCG)
+			data = (0x36 << RLC_CGCG_CGLS_CTRL_3D__CGCG_GFX_IDLE_THRESHOLD__SHIFT) |
+				RLC_CGCG_CGLS_CTRL_3D__CGCG_EN_MASK;
+		else
+			data = 0x0 << RLC_CGCG_CGLS_CTRL_3D__CGCG_GFX_IDLE_THRESHOLD__SHIFT;
+
 		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGLS)
 			data |= (0x000F << RLC_CGCG_CGLS_CTRL_3D__CGLS_REP_COMPANSAT_DELAY__SHIFT) |
 				RLC_CGCG_CGLS_CTRL_3D__CGLS_EN_MASK;
@@ -3282,8 +5074,6 @@ static void gfx_v9_0_update_3d_clock_gating(struct amdgpu_device *adev,
 		if (def != data)
 			WREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL_3D, data);
 	}
-
-	adev->gfx.rlc.funcs->exit_safe_mode(adev);
 }
 
 static void gfx_v9_0_update_coarse_grain_clock_gating(struct amdgpu_device *adev,
@@ -3291,8 +5081,6 @@ static void gfx_v9_0_update_coarse_grain_clock_gating(struct amdgpu_device *adev
 {
 	uint32_t def, data;
 
-	adev->gfx.rlc.funcs->enter_safe_mode(adev);
-
 	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGCG)) {
 		def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE);
 		/* unset CGCG override */
@@ -3305,10 +5093,15 @@ static void gfx_v9_0_update_coarse_grain_clock_gating(struct amdgpu_device *adev
 		if (def != data)
 			WREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE, data);
 
-		/* enable cgcg FSM(0x0020003F) */
+		/* enable cgcg FSM(0x0000363F) */
 		def = RREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL);
-		data = (0x2000 << RLC_CGCG_CGLS_CTRL__CGCG_GFX_IDLE_THRESHOLD__SHIFT) |
-			RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK;
+
+		if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 1))
+			data = (0x2000 << RLC_CGCG_CGLS_CTRL__CGCG_GFX_IDLE_THRESHOLD__SHIFT) |
+				RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK;
+		else
+			data = (0x36 << RLC_CGCG_CGLS_CTRL__CGCG_GFX_IDLE_THRESHOLD__SHIFT) |
+				RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK;
 		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGLS)
 			data |= (0x000F << RLC_CGCG_CGLS_CTRL__CGLS_REP_COMPANSAT_DELAY__SHIFT) |
 				RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK;
@@ -3329,13 +5122,12 @@ static void gfx_v9_0_update_coarse_grain_clock_gating(struct amdgpu_device *adev
 		if (def != data)
 			WREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL, data);
 	}
-
-	adev->gfx.rlc.funcs->exit_safe_mode(adev);
 }
 
 static int gfx_v9_0_update_gfx_clock_gating(struct amdgpu_device *adev,
 					    bool enable)
 {
+	amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
 	if (enable) {
 		/* CGCG/CGLS should be enabled after MGCG/MGLS
 		 * ===  MGCG + MGLS ===
@@ -3355,22 +5147,97 @@ static int gfx_v9_0_update_gfx_clock_gating(struct amdgpu_device *adev,
 		/* ===  MGCG + MGLS === */
 		gfx_v9_0_update_medium_grain_clock_gating(adev, enable);
 	}
+	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
 	return 0;
 }
 
+static void gfx_v9_0_update_spm_vmid_internal(struct amdgpu_device *adev,
+					      unsigned int vmid)
+{
+	u32 reg, data;
+
+	reg = SOC15_REG_OFFSET(GC, 0, mmRLC_SPM_MC_CNTL);
+	if (amdgpu_sriov_is_pp_one_vf(adev))
+		data = RREG32_NO_KIQ(reg);
+	else
+		data = RREG32_SOC15(GC, 0, mmRLC_SPM_MC_CNTL);
+
+	data &= ~RLC_SPM_MC_CNTL__RLC_SPM_VMID_MASK;
+	data |= (vmid & RLC_SPM_MC_CNTL__RLC_SPM_VMID_MASK) << RLC_SPM_MC_CNTL__RLC_SPM_VMID__SHIFT;
+
+	if (amdgpu_sriov_is_pp_one_vf(adev))
+		WREG32_SOC15_NO_KIQ(GC, 0, mmRLC_SPM_MC_CNTL, data);
+	else
+		WREG32_SOC15(GC, 0, mmRLC_SPM_MC_CNTL, data);
+}
+
+static void gfx_v9_0_update_spm_vmid(struct amdgpu_device *adev, struct amdgpu_ring *ring, unsigned int vmid)
+{
+	amdgpu_gfx_off_ctrl(adev, false);
+
+	gfx_v9_0_update_spm_vmid_internal(adev, vmid);
+
+	amdgpu_gfx_off_ctrl(adev, true);
+}
+
+static bool gfx_v9_0_check_rlcg_range(struct amdgpu_device *adev,
+					uint32_t offset,
+					struct soc15_reg_rlcg *entries, int arr_size)
+{
+	int i;
+	uint32_t reg;
+
+	if (!entries)
+		return false;
+
+	for (i = 0; i < arr_size; i++) {
+		const struct soc15_reg_rlcg *entry;
+
+		entry = &entries[i];
+		reg = adev->reg_offset[entry->hwip][entry->instance][entry->segment] + entry->reg;
+		if (offset == reg)
+			return true;
+	}
+
+	return false;
+}
+
+static bool gfx_v9_0_is_rlcg_access_range(struct amdgpu_device *adev, u32 offset)
+{
+	return gfx_v9_0_check_rlcg_range(adev, offset,
+					(void *)rlcg_access_gc_9_0,
+					ARRAY_SIZE(rlcg_access_gc_9_0));
+}
+
 static const struct amdgpu_rlc_funcs gfx_v9_0_rlc_funcs = {
-	.enter_safe_mode = gfx_v9_0_enter_rlc_safe_mode,
-	.exit_safe_mode = gfx_v9_0_exit_rlc_safe_mode
+	.is_rlc_enabled = gfx_v9_0_is_rlc_enabled,
+	.set_safe_mode = gfx_v9_0_set_safe_mode,
+	.unset_safe_mode = gfx_v9_0_unset_safe_mode,
+	.init = gfx_v9_0_rlc_init,
+	.get_csb_size = gfx_v9_0_get_csb_size,
+	.get_csb_buffer = gfx_v9_0_get_csb_buffer,
+	.get_cp_table_num = gfx_v9_0_cp_jump_table_num,
+	.resume = gfx_v9_0_rlc_resume,
+	.stop = gfx_v9_0_rlc_stop,
+	.reset = gfx_v9_0_rlc_reset,
+	.start = gfx_v9_0_rlc_start,
+	.update_spm_vmid = gfx_v9_0_update_spm_vmid,
+	.is_rlcg_access_range = gfx_v9_0_is_rlcg_access_range,
 };
 
-static int gfx_v9_0_set_powergating_state(void *handle,
+static int gfx_v9_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	bool enable = (state == AMD_PG_STATE_GATE) ? true : false;
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = (state == AMD_PG_STATE_GATE);
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 2, 2):
+	case IP_VERSION(9, 1, 0):
+	case IP_VERSION(9, 3, 0):
+		if (!enable)
+			amdgpu_gfx_off_ctrl_immediate(adev, false);
 
-	switch (adev->asic_type) {
-	case CHIP_RAVEN:
 		if (adev->pg_flags & AMD_PG_SUPPORT_RLC_SMU_HS) {
 			gfx_v9_0_enable_sck_slow_down_on_power_up(adev, true);
 			gfx_v9_0_enable_sck_slow_down_on_power_down(adev, true);
@@ -3389,6 +5256,12 @@ static int gfx_v9_0_set_powergating_state(void *handle,
 
 		/* update mgcg state */
 		gfx_v9_0_update_gfx_mg_power_gating(adev, enable);
+
+		if (enable)
+			amdgpu_gfx_off_ctrl_immediate(adev, true);
+		break;
+	case IP_VERSION(9, 2, 1):
+		amdgpu_gfx_off_ctrl_immediate(adev, enable);
 		break;
 	default:
 		break;
@@ -3397,19 +5270,25 @@ static int gfx_v9_0_set_powergating_state(void *handle,
 	return 0;
 }
 
-static int gfx_v9_0_set_clockgating_state(void *handle,
+static int gfx_v9_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (amdgpu_sriov_vf(adev))
 		return 0;
 
-	switch (adev->asic_type) {
-	case CHIP_VEGA10:
-	case CHIP_RAVEN:
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 0, 1):
+	case IP_VERSION(9, 2, 1):
+	case IP_VERSION(9, 4, 0):
+	case IP_VERSION(9, 2, 2):
+	case IP_VERSION(9, 1, 0):
+	case IP_VERSION(9, 4, 1):
+	case IP_VERSION(9, 3, 0):
+	case IP_VERSION(9, 4, 2):
 		gfx_v9_0_update_gfx_clock_gating(adev,
-						 state == AMD_CG_STATE_GATE ? true : false);
+						 state == AMD_CG_STATE_GATE);
 		break;
 	default:
 		break;
@@ -3417,21 +5296,21 @@ static int gfx_v9_0_set_clockgating_state(void *handle,
 	return 0;
 }
 
-static void gfx_v9_0_get_clockgating_state(void *handle, u32 *flags)
+static void gfx_v9_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int data;
 
 	if (amdgpu_sriov_vf(adev))
 		*flags = 0;
 
 	/* AMD_CG_SUPPORT_GFX_MGCG */
-	data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE);
+	data = RREG32_KIQ(SOC15_REG_OFFSET(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE));
 	if (!(data & RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGCG_OVERRIDE_MASK))
 		*flags |= AMD_CG_SUPPORT_GFX_MGCG;
 
 	/* AMD_CG_SUPPORT_GFX_CGCG */
-	data = RREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL);
+	data = RREG32_KIQ(SOC15_REG_OFFSET(GC, 0, mmRLC_CGCG_CGLS_CTRL));
 	if (data & RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK)
 		*flags |= AMD_CG_SUPPORT_GFX_CGCG;
 
@@ -3440,28 +5319,30 @@ static void gfx_v9_0_get_clockgating_state(void *handle, u32 *flags)
 		*flags |= AMD_CG_SUPPORT_GFX_CGLS;
 
 	/* AMD_CG_SUPPORT_GFX_RLC_LS */
-	data = RREG32_SOC15(GC, 0, mmRLC_MEM_SLP_CNTL);
+	data = RREG32_KIQ(SOC15_REG_OFFSET(GC, 0, mmRLC_MEM_SLP_CNTL));
 	if (data & RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK)
 		*flags |= AMD_CG_SUPPORT_GFX_RLC_LS | AMD_CG_SUPPORT_GFX_MGLS;
 
 	/* AMD_CG_SUPPORT_GFX_CP_LS */
-	data = RREG32_SOC15(GC, 0, mmCP_MEM_SLP_CNTL);
+	data = RREG32_KIQ(SOC15_REG_OFFSET(GC, 0, mmCP_MEM_SLP_CNTL));
 	if (data & CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK)
 		*flags |= AMD_CG_SUPPORT_GFX_CP_LS | AMD_CG_SUPPORT_GFX_MGLS;
 
-	/* AMD_CG_SUPPORT_GFX_3D_CGCG */
-	data = RREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL_3D);
-	if (data & RLC_CGCG_CGLS_CTRL_3D__CGCG_EN_MASK)
-		*flags |= AMD_CG_SUPPORT_GFX_3D_CGCG;
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 1)) {
+		/* AMD_CG_SUPPORT_GFX_3D_CGCG */
+		data = RREG32_KIQ(SOC15_REG_OFFSET(GC, 0, mmRLC_CGCG_CGLS_CTRL_3D));
+		if (data & RLC_CGCG_CGLS_CTRL_3D__CGCG_EN_MASK)
+			*flags |= AMD_CG_SUPPORT_GFX_3D_CGCG;
 
-	/* AMD_CG_SUPPORT_GFX_3D_CGLS */
-	if (data & RLC_CGCG_CGLS_CTRL_3D__CGLS_EN_MASK)
-		*flags |= AMD_CG_SUPPORT_GFX_3D_CGLS;
+		/* AMD_CG_SUPPORT_GFX_3D_CGLS */
+		if (data & RLC_CGCG_CGLS_CTRL_3D__CGLS_EN_MASK)
+			*flags |= AMD_CG_SUPPORT_GFX_3D_CGLS;
+	}
 }
 
 static u64 gfx_v9_0_ring_get_rptr_gfx(struct amdgpu_ring *ring)
 {
-	return ring->adev->wb.wb[ring->rptr_offs]; /* gfx9 is 32bit rptr*/
+	return *ring->rptr_cpu_addr; /* gfx9 is 32bit rptr*/
 }
 
 static u64 gfx_v9_0_ring_get_wptr_gfx(struct amdgpu_ring *ring)
@@ -3471,7 +5352,7 @@ static u64 gfx_v9_0_ring_get_wptr_gfx(struct amdgpu_ring *ring)
 
 	/* XXX check if swapping is necessary on BE */
 	if (ring->use_doorbell) {
-		wptr = atomic64_read((atomic64_t *)&adev->wb.wb[ring->wptr_offs]);
+		wptr = atomic64_read((atomic64_t *)ring->wptr_cpu_addr);
 	} else {
 		wptr = RREG32_SOC15(GC, 0, mmCP_RB0_WPTR);
 		wptr += (u64)RREG32_SOC15(GC, 0, mmCP_RB0_WPTR_HI) << 32;
@@ -3486,7 +5367,7 @@ static void gfx_v9_0_ring_set_wptr_gfx(struct amdgpu_ring *ring)
 
 	if (ring->use_doorbell) {
 		/* XXX check if swapping is necessary on BE */
-		atomic64_set((atomic64_t*)&adev->wb.wb[ring->wptr_offs], ring->wptr);
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr, ring->wptr);
 		WDOORBELL64(ring->doorbell_index, ring->wptr);
 	} else {
 		WREG32_SOC15(GC, 0, mmCP_RB0_WPTR, lower_32_bits(ring->wptr));
@@ -3496,11 +5377,9 @@ static void gfx_v9_0_ring_set_wptr_gfx(struct amdgpu_ring *ring)
 
 static void gfx_v9_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
 {
+	struct amdgpu_device *adev = ring->adev;
 	u32 ref_and_mask, reg_mem_engine;
-	struct nbio_hdp_flush_reg *nbio_hf_reg;
-
-	if (ring->adev->asic_type == CHIP_VEGA10)
-		nbio_hf_reg = &nbio_v6_1_hdp_flush_reg;
+	const struct nbio_hdp_flush_reg *nbio_hf_reg = adev->nbio.hdp_flush_reg;
 
 	if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE) {
 		switch (ring->me) {
@@ -3520,21 +5399,17 @@ static void gfx_v9_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
 	}
 
 	gfx_v9_0_wait_reg_mem(ring, reg_mem_engine, 0, 1,
-			      nbio_hf_reg->hdp_flush_req_offset,
-			      nbio_hf_reg->hdp_flush_done_offset,
+			      adev->nbio.funcs->get_hdp_flush_req_offset(adev),
+			      adev->nbio.funcs->get_hdp_flush_done_offset(adev),
 			      ref_and_mask, ref_and_mask, 0x20);
 }
 
-static void gfx_v9_0_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
-{
-	gfx_v9_0_write_data_to_reg(ring, 0, true,
-				   SOC15_REG_OFFSET(HDP, 0, mmHDP_DEBUG0), 1);
-}
-
 static void gfx_v9_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
-                                      struct amdgpu_ib *ib,
-                                      unsigned vm_id, bool ctx_switch)
+					struct amdgpu_job *job,
+					struct amdgpu_ib *ib,
+					uint32_t flags)
 {
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
 	u32 header, control = 0;
 
 	if (ib->flags & AMDGPU_IB_FLAG_CE)
@@ -3542,41 +5417,127 @@ static void gfx_v9_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
 	else
 		header = PACKET3(PACKET3_INDIRECT_BUFFER, 2);
 
-	control |= ib->length_dw | (vm_id << 24);
+	control |= ib->length_dw | (vmid << 24);
 
-	if (amdgpu_sriov_vf(ring->adev) && (ib->flags & AMDGPU_IB_FLAG_PREEMPT)) {
+	if (ib->flags & AMDGPU_IB_FLAG_PREEMPT) {
 		control |= INDIRECT_BUFFER_PRE_ENB(1);
 
-		if (!(ib->flags & AMDGPU_IB_FLAG_CE))
-			gfx_v9_0_ring_emit_de_meta(ring);
+		if (flags & AMDGPU_IB_PREEMPTED)
+			control |= INDIRECT_BUFFER_PRE_RESUME(1);
+
+		if (!(ib->flags & AMDGPU_IB_FLAG_CE) && vmid)
+			gfx_v9_0_ring_emit_de_meta(ring,
+						   (!amdgpu_sriov_vf(ring->adev) &&
+						   flags & AMDGPU_IB_PREEMPTED) ?
+						   true : false,
+						   job->gds_size > 0 && job->gds_base != 0);
 	}
 
 	amdgpu_ring_write(ring, header);
-BUG_ON(ib->gpu_addr & 0x3); /* Dword align */
+	BUG_ON(ib->gpu_addr & 0x3); /* Dword align */
 	amdgpu_ring_write(ring,
 #ifdef __BIG_ENDIAN
 		(2 << 0) |
 #endif
 		lower_32_bits(ib->gpu_addr));
 	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_ib_on_emit_cntl(ring);
 	amdgpu_ring_write(ring, control);
 }
 
-static void gfx_v9_0_ring_emit_ib_compute(struct amdgpu_ring *ring,
-                                          struct amdgpu_ib *ib,
-                                          unsigned vm_id, bool ctx_switch)
+static void gfx_v9_0_ring_patch_cntl(struct amdgpu_ring *ring,
+				     unsigned offset)
+{
+	u32 control = ring->ring[offset];
+
+	control |= INDIRECT_BUFFER_PRE_RESUME(1);
+	ring->ring[offset] = control;
+}
+
+static void gfx_v9_0_ring_patch_ce_meta(struct amdgpu_ring *ring,
+					unsigned offset)
+{
+	struct amdgpu_device *adev = ring->adev;
+	void *ce_payload_cpu_addr;
+	uint64_t payload_offset, payload_size;
+
+	payload_size = sizeof(struct v9_ce_ib_state);
+
+	payload_offset = offsetof(struct v9_gfx_meta_data, ce_payload);
+	ce_payload_cpu_addr = adev->virt.csa_cpu_addr + payload_offset;
+
+	if (offset + (payload_size >> 2) <= ring->buf_mask + 1) {
+		memcpy((void *)&ring->ring[offset], ce_payload_cpu_addr, payload_size);
+	} else {
+		memcpy((void *)&ring->ring[offset], ce_payload_cpu_addr,
+		       (ring->buf_mask + 1 - offset) << 2);
+		payload_size -= (ring->buf_mask + 1 - offset) << 2;
+		memcpy((void *)&ring->ring[0],
+		       ce_payload_cpu_addr + ((ring->buf_mask + 1 - offset) << 2),
+		       payload_size);
+	}
+}
+
+static void gfx_v9_0_ring_patch_de_meta(struct amdgpu_ring *ring,
+					unsigned offset)
 {
-        u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vm_id << 24);
+	struct amdgpu_device *adev = ring->adev;
+	void *de_payload_cpu_addr;
+	uint64_t payload_offset, payload_size;
+
+	payload_size = sizeof(struct v9_de_ib_state);
+
+	payload_offset = offsetof(struct v9_gfx_meta_data, de_payload);
+	de_payload_cpu_addr = adev->virt.csa_cpu_addr + payload_offset;
+
+	((struct v9_de_ib_state *)de_payload_cpu_addr)->ib_completion_status =
+		IB_COMPLETION_STATUS_PREEMPTED;
+
+	if (offset + (payload_size >> 2) <= ring->buf_mask + 1) {
+		memcpy((void *)&ring->ring[offset], de_payload_cpu_addr, payload_size);
+	} else {
+		memcpy((void *)&ring->ring[offset], de_payload_cpu_addr,
+		       (ring->buf_mask + 1 - offset) << 2);
+		payload_size -= (ring->buf_mask + 1 - offset) << 2;
+		memcpy((void *)&ring->ring[0],
+		       de_payload_cpu_addr + ((ring->buf_mask + 1 - offset) << 2),
+		       payload_size);
+	}
+}
 
-        amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2));
+static void gfx_v9_0_ring_emit_ib_compute(struct amdgpu_ring *ring,
+					  struct amdgpu_job *job,
+					  struct amdgpu_ib *ib,
+					  uint32_t flags)
+{
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+	u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vmid << 24);
+
+	/* Currently, there is a high possibility to get wave ID mismatch
+	 * between ME and GDS, leading to a hw deadlock, because ME generates
+	 * different wave IDs than the GDS expects. This situation happens
+	 * randomly when at least 5 compute pipes use GDS ordered append.
+	 * The wave IDs generated by ME are also wrong after suspend/resume.
+	 * Those are probably bugs somewhere else in the kernel driver.
+	 *
+	 * Writing GDS_COMPUTE_MAX_WAVE_ID resets wave ID counters in ME and
+	 * GDS to 0 for this ring (me/pipe).
+	 */
+	if (ib->flags & AMDGPU_IB_FLAG_RESET_GDS_MAX_WAVE_ID) {
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1));
+		amdgpu_ring_write(ring, mmGDS_COMPUTE_MAX_WAVE_ID);
+		amdgpu_ring_write(ring, ring->adev->gds.gds_compute_max_wave_id);
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2));
 	BUG_ON(ib->gpu_addr & 0x3); /* Dword align */
-        amdgpu_ring_write(ring,
+	amdgpu_ring_write(ring,
 #ifdef __BIG_ENDIAN
-                                (2 << 0) |
+				(2 << 0) |
 #endif
-                                lower_32_bits(ib->gpu_addr));
-        amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
-        amdgpu_ring_write(ring, control);
+				lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, control);
 }
 
 static void gfx_v9_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
@@ -3584,15 +5545,25 @@ static void gfx_v9_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
 {
 	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
 	bool int_sel = flags & AMDGPU_FENCE_FLAG_INT;
+	bool writeback = flags & AMDGPU_FENCE_FLAG_TC_WB_ONLY;
+	bool exec = flags & AMDGPU_FENCE_FLAG_EXEC;
+	uint32_t dw2 = 0;
 
 	/* RELEASE_MEM - flush caches, send int */
 	amdgpu_ring_write(ring, PACKET3(PACKET3_RELEASE_MEM, 6));
-	amdgpu_ring_write(ring, (EOP_TCL1_ACTION_EN |
-				 EOP_TC_ACTION_EN |
-				 EOP_TC_WB_ACTION_EN |
-				 EOP_TC_MD_ACTION_EN |
-				 EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) |
-				 EVENT_INDEX(5)));
+
+	if (writeback) {
+		dw2 = EOP_TC_NC_ACTION_EN;
+	} else {
+		dw2 = EOP_TCL1_ACTION_EN | EOP_TC_ACTION_EN |
+				EOP_TC_MD_ACTION_EN;
+	}
+	dw2 |= EOP_TC_WB_ACTION_EN | EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) |
+				EVENT_INDEX(5);
+	if (exec)
+		dw2 |= EOP_EXEC;
+
+	amdgpu_ring_write(ring, dw2);
 	amdgpu_ring_write(ring, DATA_SEL(write64bit ? 2 : 1) | INT_SEL(int_sel ? 2 : 0));
 
 	/*
@@ -3622,33 +5593,12 @@ static void gfx_v9_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
 }
 
 static void gfx_v9_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
-					unsigned vm_id, uint64_t pd_addr)
+					unsigned vmid, uint64_t pd_addr)
 {
-	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
-	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
-	uint32_t req = ring->adev->gart.gart_funcs->get_invalidate_req(vm_id);
-	unsigned eng = ring->vm_inv_eng;
-
-	pd_addr = amdgpu_gart_get_vm_pde(ring->adev, pd_addr);
-	pd_addr |= AMDGPU_PTE_VALID;
-
-	gfx_v9_0_write_data_to_reg(ring, usepfp, true,
-				   hub->ctx0_ptb_addr_lo32 + (2 * vm_id),
-				   lower_32_bits(pd_addr));
-
-	gfx_v9_0_write_data_to_reg(ring, usepfp, true,
-				   hub->ctx0_ptb_addr_hi32 + (2 * vm_id),
-				   upper_32_bits(pd_addr));
-
-	gfx_v9_0_write_data_to_reg(ring, usepfp, true,
-				   hub->vm_inv_eng0_req + eng, req);
-
-	/* wait for the invalidate to complete */
-	gfx_v9_0_wait_reg_mem(ring, 0, 0, 0, hub->vm_inv_eng0_ack +
-			      eng, 0, 1 << vm_id, 1 << vm_id, 0x20);
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
 
 	/* compute doesn't have PFP */
-	if (usepfp) {
+	if (ring->funcs->type == AMDGPU_RING_TYPE_GFX) {
 		/* sync PFP to ME, otherwise we might get invalid PFP reads */
 		amdgpu_ring_write(ring, PACKET3(PACKET3_PFP_SYNC_ME, 0));
 		amdgpu_ring_write(ring, 0x0);
@@ -3657,7 +5607,7 @@ static void gfx_v9_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
 
 static u64 gfx_v9_0_ring_get_rptr_compute(struct amdgpu_ring *ring)
 {
-	return ring->adev->wb.wb[ring->rptr_offs]; /* gfx9 hardware is 32bit rptr */
+	return *ring->rptr_cpu_addr; /* gfx9 hardware is 32bit rptr */
 }
 
 static u64 gfx_v9_0_ring_get_wptr_compute(struct amdgpu_ring *ring)
@@ -3666,7 +5616,7 @@ static u64 gfx_v9_0_ring_get_wptr_compute(struct amdgpu_ring *ring)
 
 	/* XXX check if swapping is necessary on BE */
 	if (ring->use_doorbell)
-		wptr = atomic64_read((atomic64_t *)&ring->adev->wb.wb[ring->wptr_offs]);
+		wptr = atomic64_read((atomic64_t *)ring->wptr_cpu_addr);
 	else
 		BUG();
 	return wptr;
@@ -3678,7 +5628,7 @@ static void gfx_v9_0_ring_set_wptr_compute(struct amdgpu_ring *ring)
 
 	/* XXX check if swapping is necessary on BE */
 	if (ring->use_doorbell) {
-		atomic64_set((atomic64_t*)&adev->wb.wb[ring->wptr_offs], ring->wptr);
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr, ring->wptr);
 		WDOORBELL64(ring->doorbell_index, ring->wptr);
 	} else{
 		BUG(); /* only DOORBELL method supported on gfx9 now */
@@ -3688,6 +5638,8 @@ static void gfx_v9_0_ring_set_wptr_compute(struct amdgpu_ring *ring)
 static void gfx_v9_0_ring_emit_fence_kiq(struct amdgpu_ring *ring, u64 addr,
 					 u64 seq, unsigned int flags)
 {
+	struct amdgpu_device *adev = ring->adev;
+
 	/* we only allocate 32bit for each seq wb address */
 	BUG_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
 
@@ -3716,35 +5668,116 @@ static void gfx_v9_ring_emit_sb(struct amdgpu_ring *ring)
 	amdgpu_ring_write(ring, 0);
 }
 
-static void gfx_v9_0_ring_emit_ce_meta(struct amdgpu_ring *ring)
+static void gfx_v9_0_ring_emit_ce_meta(struct amdgpu_ring *ring, bool resume)
 {
-	static struct v9_ce_ib_state ce_payload = {0};
-	uint64_t csa_addr;
+	struct amdgpu_device *adev = ring->adev;
+	struct v9_ce_ib_state ce_payload = {0};
+	uint64_t offset, ce_payload_gpu_addr;
+	void *ce_payload_cpu_addr;
 	int cnt;
 
 	cnt = (sizeof(ce_payload) >> 2) + 4 - 2;
-	csa_addr = AMDGPU_VA_RESERVED_SIZE - 2 * 4096;
+
+	offset = offsetof(struct v9_gfx_meta_data, ce_payload);
+	ce_payload_gpu_addr = amdgpu_csa_vaddr(ring->adev) + offset;
+	ce_payload_cpu_addr = adev->virt.csa_cpu_addr + offset;
 
 	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, cnt));
 	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(2) |
 				 WRITE_DATA_DST_SEL(8) |
 				 WR_CONFIRM) |
 				 WRITE_DATA_CACHE_POLICY(0));
-	amdgpu_ring_write(ring, lower_32_bits(csa_addr + offsetof(struct v9_gfx_meta_data, ce_payload)));
-	amdgpu_ring_write(ring, upper_32_bits(csa_addr + offsetof(struct v9_gfx_meta_data, ce_payload)));
-	amdgpu_ring_write_multiple(ring, (void *)&ce_payload, sizeof(ce_payload) >> 2);
+	amdgpu_ring_write(ring, lower_32_bits(ce_payload_gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ce_payload_gpu_addr));
+
+	amdgpu_ring_ib_on_emit_ce(ring);
+
+	if (resume)
+		amdgpu_ring_write_multiple(ring, ce_payload_cpu_addr,
+					   sizeof(ce_payload) >> 2);
+	else
+		amdgpu_ring_write_multiple(ring, (void *)&ce_payload,
+					   sizeof(ce_payload) >> 2);
+}
+
+static int gfx_v9_0_ring_preempt_ib(struct amdgpu_ring *ring)
+{
+	int i, r = 0;
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[0];
+	struct amdgpu_ring *kiq_ring = &kiq->ring;
+	unsigned long flags;
+
+	if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
+		return -EINVAL;
+
+	spin_lock_irqsave(&kiq->ring_lock, flags);
+
+	if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->unmap_queues_size)) {
+		spin_unlock_irqrestore(&kiq->ring_lock, flags);
+		return -ENOMEM;
+	}
+
+	/* assert preemption condition */
+	amdgpu_ring_set_preempt_cond_exec(ring, false);
+
+	ring->trail_seq += 1;
+	amdgpu_ring_alloc(ring, 13);
+	gfx_v9_0_ring_emit_fence(ring, ring->trail_fence_gpu_addr,
+				 ring->trail_seq, AMDGPU_FENCE_FLAG_EXEC | AMDGPU_FENCE_FLAG_INT);
+
+	/* assert IB preemption, emit the trailing fence */
+	kiq->pmf->kiq_unmap_queues(kiq_ring, ring, PREEMPT_QUEUES_NO_UNMAP,
+				   ring->trail_fence_gpu_addr,
+				   ring->trail_seq);
+
+	amdgpu_ring_commit(kiq_ring);
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+
+	/* poll the trailing fence */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (ring->trail_seq ==
+			le32_to_cpu(*ring->trail_fence_cpu_addr))
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout) {
+		r = -EINVAL;
+		DRM_WARN("ring %d timeout to preempt ib\n", ring->idx);
+	}
+
+	/*reset the CP_VMID_PREEMPT after trailing fence*/
+	amdgpu_ring_emit_wreg(ring,
+			      SOC15_REG_OFFSET(GC, 0, mmCP_VMID_PREEMPT),
+			      0x0);
+	amdgpu_ring_commit(ring);
+
+	/* deassert preemption condition */
+	amdgpu_ring_set_preempt_cond_exec(ring, true);
+	return r;
 }
 
-static void gfx_v9_0_ring_emit_de_meta(struct amdgpu_ring *ring)
+static void gfx_v9_0_ring_emit_de_meta(struct amdgpu_ring *ring, bool resume, bool usegds)
 {
-	static struct v9_de_ib_state de_payload = {0};
-	uint64_t csa_addr, gds_addr;
+	struct amdgpu_device *adev = ring->adev;
+	struct v9_de_ib_state de_payload = {0};
+	uint64_t offset, gds_addr, de_payload_gpu_addr;
+	void *de_payload_cpu_addr;
 	int cnt;
 
-	csa_addr = AMDGPU_VA_RESERVED_SIZE - 2 * 4096;
-	gds_addr = csa_addr + 4096;
-	de_payload.gds_backup_addrlo = lower_32_bits(gds_addr);
-	de_payload.gds_backup_addrhi = upper_32_bits(gds_addr);
+	offset = offsetof(struct v9_gfx_meta_data, de_payload);
+	de_payload_gpu_addr = amdgpu_csa_vaddr(ring->adev) + offset;
+	de_payload_cpu_addr = adev->virt.csa_cpu_addr + offset;
+
+	gds_addr = ALIGN(amdgpu_csa_vaddr(ring->adev) +
+			 AMDGPU_CSA_SIZE - adev->gds.gds_size,
+			 PAGE_SIZE);
+
+	if (usegds) {
+		de_payload.gds_backup_addrlo = lower_32_bits(gds_addr);
+		de_payload.gds_backup_addrhi = upper_32_bits(gds_addr);
+	}
 
 	cnt = (sizeof(de_payload) >> 2) + 4 - 2;
 	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, cnt));
@@ -3752,17 +5785,34 @@ static void gfx_v9_0_ring_emit_de_meta(struct amdgpu_ring *ring)
 				 WRITE_DATA_DST_SEL(8) |
 				 WR_CONFIRM) |
 				 WRITE_DATA_CACHE_POLICY(0));
-	amdgpu_ring_write(ring, lower_32_bits(csa_addr + offsetof(struct v9_gfx_meta_data, de_payload)));
-	amdgpu_ring_write(ring, upper_32_bits(csa_addr + offsetof(struct v9_gfx_meta_data, de_payload)));
-	amdgpu_ring_write_multiple(ring, (void *)&de_payload, sizeof(de_payload) >> 2);
+	amdgpu_ring_write(ring, lower_32_bits(de_payload_gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(de_payload_gpu_addr));
+
+	amdgpu_ring_ib_on_emit_de(ring);
+	if (resume)
+		amdgpu_ring_write_multiple(ring, de_payload_cpu_addr,
+					   sizeof(de_payload) >> 2);
+	else
+		amdgpu_ring_write_multiple(ring, (void *)&de_payload,
+					   sizeof(de_payload) >> 2);
+}
+
+static void gfx_v9_0_ring_emit_frame_cntl(struct amdgpu_ring *ring, bool start,
+				   bool secure)
+{
+	uint32_t v = secure ? FRAME_TMZ : 0;
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_FRAME_CONTROL, 0));
+	amdgpu_ring_write(ring, v | FRAME_CMD(start ? 0 : 1));
 }
 
 static void gfx_v9_ring_emit_cntxcntl(struct amdgpu_ring *ring, uint32_t flags)
 {
 	uint32_t dw2 = 0;
 
-	if (amdgpu_sriov_vf(ring->adev))
-		gfx_v9_0_ring_emit_ce_meta(ring);
+	gfx_v9_0_ring_emit_ce_meta(ring,
+				   (!amdgpu_sriov_vf(ring->adev) &&
+				   flags & AMDGPU_IB_PREEMPTED) ? true : false);
 
 	dw2 |= 0x80000000; /* set load_enable otherwise this package is just NOPs */
 	if (flags & AMDGPU_HAVE_CTX_SWITCH) {
@@ -3789,38 +5839,23 @@ static void gfx_v9_ring_emit_cntxcntl(struct amdgpu_ring *ring, uint32_t flags)
 	amdgpu_ring_write(ring, 0);
 }
 
-static unsigned gfx_v9_0_ring_emit_init_cond_exec(struct amdgpu_ring *ring)
+static unsigned gfx_v9_0_ring_emit_init_cond_exec(struct amdgpu_ring *ring,
+						  uint64_t addr)
 {
 	unsigned ret;
 	amdgpu_ring_write(ring, PACKET3(PACKET3_COND_EXEC, 3));
-	amdgpu_ring_write(ring, lower_32_bits(ring->cond_exe_gpu_addr));
-	amdgpu_ring_write(ring, upper_32_bits(ring->cond_exe_gpu_addr));
-	amdgpu_ring_write(ring, 0); /* discard following DWs if *cond_exec_gpu_addr==0 */
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	/* discard following DWs if *cond_exec_gpu_addr==0 */
+	amdgpu_ring_write(ring, 0);
 	ret = ring->wptr & ring->buf_mask;
-	amdgpu_ring_write(ring, 0x55aa55aa); /* patch dummy value later */
+	/* patch dummy value later */
+	amdgpu_ring_write(ring, 0);
 	return ret;
 }
 
-static void gfx_v9_0_ring_emit_patch_cond_exec(struct amdgpu_ring *ring, unsigned offset)
-{
-	unsigned cur;
-	BUG_ON(offset > ring->buf_mask);
-	BUG_ON(ring->ring[offset] != 0x55aa55aa);
-
-	cur = (ring->wptr & ring->buf_mask) - 1;
-	if (likely(cur > offset))
-		ring->ring[offset] = cur - offset;
-	else
-		ring->ring[offset] = (ring->ring_size>>2) - offset + cur;
-}
-
-static void gfx_v9_0_ring_emit_tmz(struct amdgpu_ring *ring, bool start)
-{
-	amdgpu_ring_write(ring, PACKET3(PACKET3_FRAME_CONTROL, 0));
-	amdgpu_ring_write(ring, FRAME_CMD(start ? 0 : 1)); /* frame_end */
-}
-
-static void gfx_v9_0_ring_emit_rreg(struct amdgpu_ring *ring, uint32_t reg)
+static void gfx_v9_0_ring_emit_rreg(struct amdgpu_ring *ring, uint32_t reg,
+				    uint32_t reg_val_offs)
 {
 	struct amdgpu_device *adev = ring->adev;
 
@@ -3831,21 +5866,71 @@ static void gfx_v9_0_ring_emit_rreg(struct amdgpu_ring *ring, uint32_t reg)
 	amdgpu_ring_write(ring, reg);
 	amdgpu_ring_write(ring, 0);
 	amdgpu_ring_write(ring, lower_32_bits(adev->wb.gpu_addr +
-				adev->virt.reg_val_offs * 4));
+				reg_val_offs * 4));
 	amdgpu_ring_write(ring, upper_32_bits(adev->wb.gpu_addr +
-				adev->virt.reg_val_offs * 4));
+				reg_val_offs * 4));
 }
 
 static void gfx_v9_0_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg,
-				  uint32_t val)
+				    uint32_t val)
 {
+	uint32_t cmd = 0;
+
+	switch (ring->funcs->type) {
+	case AMDGPU_RING_TYPE_GFX:
+		cmd = WRITE_DATA_ENGINE_SEL(1) | WR_CONFIRM;
+		break;
+	case AMDGPU_RING_TYPE_KIQ:
+		cmd = (1 << 16); /* no inc addr */
+		break;
+	default:
+		cmd = WR_CONFIRM;
+		break;
+	}
 	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
-	amdgpu_ring_write(ring, (1 << 16)); /* no inc addr */
+	amdgpu_ring_write(ring, cmd);
 	amdgpu_ring_write(ring, reg);
 	amdgpu_ring_write(ring, 0);
 	amdgpu_ring_write(ring, val);
 }
 
+static void gfx_v9_0_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+					uint32_t val, uint32_t mask)
+{
+	gfx_v9_0_wait_reg_mem(ring, 0, 0, 0, reg, 0, val, mask, 0x20);
+}
+
+static void gfx_v9_0_ring_emit_reg_write_reg_wait(struct amdgpu_ring *ring,
+						  uint32_t reg0, uint32_t reg1,
+						  uint32_t ref, uint32_t mask)
+{
+	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
+	struct amdgpu_device *adev = ring->adev;
+	bool fw_version_ok = (ring->funcs->type == AMDGPU_RING_TYPE_GFX) ?
+		adev->gfx.me_fw_write_wait : adev->gfx.mec_fw_write_wait;
+
+	if (fw_version_ok)
+		gfx_v9_0_wait_reg_mem(ring, usepfp, 0, 1, reg0, reg1,
+				      ref, mask, 0x20);
+	else
+		amdgpu_ring_emit_reg_write_reg_wait_helper(ring, reg0, reg1,
+							   ref, mask);
+}
+
+static void gfx_v9_0_ring_soft_recovery(struct amdgpu_ring *ring, unsigned vmid)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t value = 0;
+
+	value = REG_SET_FIELD(value, SQ_CMD, CMD, 0x03);
+	value = REG_SET_FIELD(value, SQ_CMD, MODE, 0x01);
+	value = REG_SET_FIELD(value, SQ_CMD, CHECK_VMID, 1);
+	value = REG_SET_FIELD(value, SQ_CMD, VM_ID, vmid);
+	amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
+	WREG32_SOC15(GC, 0, mmSQ_CMD, value);
+	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
+}
+
 static void gfx_v9_0_set_gfx_eop_interrupt_state(struct amdgpu_device *adev,
 						 enum amdgpu_interrupt_state state)
 {
@@ -3898,33 +5983,111 @@ static void gfx_v9_0_set_compute_eop_interrupt_state(struct amdgpu_device *adev,
 
 	switch (state) {
 	case AMDGPU_IRQ_STATE_DISABLE:
-		mec_int_cntl = RREG32(mec_int_cntl_reg);
+		mec_int_cntl = RREG32_SOC15_IP(GC,mec_int_cntl_reg);
 		mec_int_cntl = REG_SET_FIELD(mec_int_cntl, CP_ME1_PIPE0_INT_CNTL,
 					     TIME_STAMP_INT_ENABLE, 0);
-		WREG32(mec_int_cntl_reg, mec_int_cntl);
+		WREG32_SOC15_IP(GC, mec_int_cntl_reg, mec_int_cntl);
 		break;
 	case AMDGPU_IRQ_STATE_ENABLE:
-		mec_int_cntl = RREG32(mec_int_cntl_reg);
+		mec_int_cntl = RREG32_SOC15_IP(GC, mec_int_cntl_reg);
 		mec_int_cntl = REG_SET_FIELD(mec_int_cntl, CP_ME1_PIPE0_INT_CNTL,
 					     TIME_STAMP_INT_ENABLE, 1);
-		WREG32(mec_int_cntl_reg, mec_int_cntl);
+		WREG32_SOC15_IP(GC, mec_int_cntl_reg, mec_int_cntl);
 		break;
 	default:
 		break;
 	}
 }
 
+static u32 gfx_v9_0_get_cpc_int_cntl(struct amdgpu_device *adev,
+				     int me, int pipe)
+{
+	/*
+	 * amdgpu controls only the first MEC. That's why this function only
+	 * handles the setting of interrupts for this specific MEC. All other
+	 * pipes' interrupts are set by amdkfd.
+	 */
+	if (me != 1)
+		return 0;
+
+	switch (pipe) {
+	case 0:
+		return SOC15_REG_OFFSET(GC, 0, mmCP_ME1_PIPE0_INT_CNTL);
+	case 1:
+		return SOC15_REG_OFFSET(GC, 0, mmCP_ME1_PIPE1_INT_CNTL);
+	case 2:
+		return SOC15_REG_OFFSET(GC, 0, mmCP_ME1_PIPE2_INT_CNTL);
+	case 3:
+		return SOC15_REG_OFFSET(GC, 0, mmCP_ME1_PIPE3_INT_CNTL);
+	default:
+		return 0;
+	}
+}
+
 static int gfx_v9_0_set_priv_reg_fault_state(struct amdgpu_device *adev,
 					     struct amdgpu_irq_src *source,
 					     unsigned type,
 					     enum amdgpu_interrupt_state state)
 {
+	u32 cp_int_cntl_reg, cp_int_cntl;
+	int i, j;
+
 	switch (state) {
 	case AMDGPU_IRQ_STATE_DISABLE:
 	case AMDGPU_IRQ_STATE_ENABLE:
 		WREG32_FIELD15(GC, 0, CP_INT_CNTL_RING0,
 			       PRIV_REG_INT_ENABLE,
 			       state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+		for (i = 0; i < adev->gfx.mec.num_mec; i++) {
+			for (j = 0; j < adev->gfx.mec.num_pipe_per_mec; j++) {
+				/* MECs start at 1 */
+				cp_int_cntl_reg = gfx_v9_0_get_cpc_int_cntl(adev, i + 1, j);
+
+				if (cp_int_cntl_reg) {
+					cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+					cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+								    PRIV_REG_INT_ENABLE,
+								    state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+					WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+				}
+			}
+		}
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gfx_v9_0_set_bad_op_fault_state(struct amdgpu_device *adev,
+					   struct amdgpu_irq_src *source,
+					   unsigned type,
+					   enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl_reg, cp_int_cntl;
+	int i, j;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+	case AMDGPU_IRQ_STATE_ENABLE:
+		WREG32_FIELD15(GC, 0, CP_INT_CNTL_RING0,
+			       OPCODE_ERROR_INT_ENABLE,
+			       state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+		for (i = 0; i < adev->gfx.mec.num_mec; i++) {
+			for (j = 0; j < adev->gfx.mec.num_pipe_per_mec; j++) {
+				/* MECs start at 1 */
+				cp_int_cntl_reg = gfx_v9_0_get_cpc_int_cntl(adev, i + 1, j);
+
+				if (cp_int_cntl_reg) {
+					cp_int_cntl = RREG32_SOC15_IP(GC, cp_int_cntl_reg);
+					cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+								    OPCODE_ERROR_INT_ENABLE,
+								    state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+					WREG32_SOC15_IP(GC, cp_int_cntl_reg, cp_int_cntl);
+				}
+			}
+		}
 		break;
 	default:
 		break;
@@ -3944,6 +6107,7 @@ static int gfx_v9_0_set_priv_inst_fault_state(struct amdgpu_device *adev,
 		WREG32_FIELD15(GC, 0, CP_INT_CNTL_RING0,
 			       PRIV_INSTR_INT_ENABLE,
 			       state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+		break;
 	default:
 		break;
 	}
@@ -3951,13 +6115,52 @@ static int gfx_v9_0_set_priv_inst_fault_state(struct amdgpu_device *adev,
 	return 0;
 }
 
+#define ENABLE_ECC_ON_ME_PIPE(me, pipe)				\
+	WREG32_FIELD15(GC, 0, CP_ME##me##_PIPE##pipe##_INT_CNTL,\
+			CP_ECC_ERROR_INT_ENABLE, 1)
+
+#define DISABLE_ECC_ON_ME_PIPE(me, pipe)			\
+	WREG32_FIELD15(GC, 0, CP_ME##me##_PIPE##pipe##_INT_CNTL,\
+			CP_ECC_ERROR_INT_ENABLE, 0)
+
+static int gfx_v9_0_set_cp_ecc_error_state(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      unsigned type,
+					      enum amdgpu_interrupt_state state)
+{
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		WREG32_FIELD15(GC, 0, CP_INT_CNTL_RING0,
+				CP_ECC_ERROR_INT_ENABLE, 0);
+		DISABLE_ECC_ON_ME_PIPE(1, 0);
+		DISABLE_ECC_ON_ME_PIPE(1, 1);
+		DISABLE_ECC_ON_ME_PIPE(1, 2);
+		DISABLE_ECC_ON_ME_PIPE(1, 3);
+		break;
+
+	case AMDGPU_IRQ_STATE_ENABLE:
+		WREG32_FIELD15(GC, 0, CP_INT_CNTL_RING0,
+				CP_ECC_ERROR_INT_ENABLE, 1);
+		ENABLE_ECC_ON_ME_PIPE(1, 0);
+		ENABLE_ECC_ON_ME_PIPE(1, 1);
+		ENABLE_ECC_ON_ME_PIPE(1, 2);
+		ENABLE_ECC_ON_ME_PIPE(1, 3);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+
 static int gfx_v9_0_set_eop_interrupt_state(struct amdgpu_device *adev,
 					    struct amdgpu_irq_src *src,
 					    unsigned type,
 					    enum amdgpu_interrupt_state state)
 {
 	switch (type) {
-	case AMDGPU_CP_IRQ_GFX_EOP:
+	case AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP:
 		gfx_v9_0_set_gfx_eop_interrupt_state(adev, state);
 		break;
 	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP:
@@ -4005,7 +6208,15 @@ static int gfx_v9_0_eop_irq(struct amdgpu_device *adev,
 
 	switch (me_id) {
 	case 0:
-		amdgpu_fence_process(&adev->gfx.gfx_ring[0]);
+		if (adev->gfx.num_gfx_rings) {
+			if (!adev->gfx.mcbp) {
+				amdgpu_fence_process(&adev->gfx.gfx_ring[0]);
+			} else if (!amdgpu_mcbp_handle_trailing_fence_irq(&adev->gfx.muxer)) {
+				/* Fence signals are handled on the software rings*/
+				for (i = 0; i < GFX9_NUM_SW_GFX_RINGS; i++)
+					amdgpu_fence_process(&adev->gfx.sw_gfx_ring[i]);
+			}
+		}
 		break;
 	case 1:
 	case 2:
@@ -4022,12 +6233,48 @@ static int gfx_v9_0_eop_irq(struct amdgpu_device *adev,
 	return 0;
 }
 
+static void gfx_v9_0_fault(struct amdgpu_device *adev,
+			   struct amdgpu_iv_entry *entry)
+{
+	u8 me_id, pipe_id, queue_id;
+	struct amdgpu_ring *ring;
+	int i;
+
+	me_id = (entry->ring_id & 0x0c) >> 2;
+	pipe_id = (entry->ring_id & 0x03) >> 0;
+	queue_id = (entry->ring_id & 0x70) >> 4;
+
+	switch (me_id) {
+	case 0:
+		drm_sched_fault(&adev->gfx.gfx_ring[0].sched);
+		break;
+	case 1:
+	case 2:
+		for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+			ring = &adev->gfx.compute_ring[i];
+			if (ring->me == me_id && ring->pipe == pipe_id &&
+			    ring->queue == queue_id)
+				drm_sched_fault(&ring->sched);
+		}
+		break;
+	}
+}
+
 static int gfx_v9_0_priv_reg_irq(struct amdgpu_device *adev,
 				 struct amdgpu_irq_src *source,
 				 struct amdgpu_iv_entry *entry)
 {
 	DRM_ERROR("Illegal register access in command stream\n");
-	schedule_work(&adev->reset_work);
+	gfx_v9_0_fault(adev, entry);
+	return 0;
+}
+
+static int gfx_v9_0_bad_op_irq(struct amdgpu_device *adev,
+			       struct amdgpu_irq_src *source,
+			       struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal opcode in command stream\n");
+	gfx_v9_0_fault(adev, entry);
 	return 0;
 }
 
@@ -4036,72 +6283,1105 @@ static int gfx_v9_0_priv_inst_irq(struct amdgpu_device *adev,
 				  struct amdgpu_iv_entry *entry)
 {
 	DRM_ERROR("Illegal instruction in command stream\n");
-	schedule_work(&adev->reset_work);
+	gfx_v9_0_fault(adev, entry);
 	return 0;
 }
 
-static int gfx_v9_0_kiq_set_interrupt_state(struct amdgpu_device *adev,
-					    struct amdgpu_irq_src *src,
-					    unsigned int type,
-					    enum amdgpu_interrupt_state state)
+
+static const struct soc15_ras_field_entry gfx_v9_0_ras_fields[] = {
+	{ "CPC_SCRATCH", SOC15_REG_ENTRY(GC, 0, mmCPC_EDC_SCRATCH_CNT),
+	  SOC15_REG_FIELD(CPC_EDC_SCRATCH_CNT, SEC_COUNT),
+	  SOC15_REG_FIELD(CPC_EDC_SCRATCH_CNT, DED_COUNT)
+	},
+	{ "CPC_UCODE", SOC15_REG_ENTRY(GC, 0, mmCPC_EDC_UCODE_CNT),
+	  SOC15_REG_FIELD(CPC_EDC_UCODE_CNT, SEC_COUNT),
+	  SOC15_REG_FIELD(CPC_EDC_UCODE_CNT, DED_COUNT)
+	},
+	{ "CPF_ROQ_ME1", SOC15_REG_ENTRY(GC, 0, mmCPF_EDC_ROQ_CNT),
+	  SOC15_REG_FIELD(CPF_EDC_ROQ_CNT, COUNT_ME1),
+	  0, 0
+	},
+	{ "CPF_ROQ_ME2", SOC15_REG_ENTRY(GC, 0, mmCPF_EDC_ROQ_CNT),
+	  SOC15_REG_FIELD(CPF_EDC_ROQ_CNT, COUNT_ME2),
+	  0, 0
+	},
+	{ "CPF_TAG", SOC15_REG_ENTRY(GC, 0, mmCPF_EDC_TAG_CNT),
+	  SOC15_REG_FIELD(CPF_EDC_TAG_CNT, SEC_COUNT),
+	  SOC15_REG_FIELD(CPF_EDC_TAG_CNT, DED_COUNT)
+	},
+	{ "CPG_DMA_ROQ", SOC15_REG_ENTRY(GC, 0, mmCPG_EDC_DMA_CNT),
+	  SOC15_REG_FIELD(CPG_EDC_DMA_CNT, ROQ_COUNT),
+	  0, 0
+	},
+	{ "CPG_DMA_TAG", SOC15_REG_ENTRY(GC, 0, mmCPG_EDC_DMA_CNT),
+	  SOC15_REG_FIELD(CPG_EDC_DMA_CNT, TAG_SEC_COUNT),
+	  SOC15_REG_FIELD(CPG_EDC_DMA_CNT, TAG_DED_COUNT)
+	},
+	{ "CPG_TAG", SOC15_REG_ENTRY(GC, 0, mmCPG_EDC_TAG_CNT),
+	  SOC15_REG_FIELD(CPG_EDC_TAG_CNT, SEC_COUNT),
+	  SOC15_REG_FIELD(CPG_EDC_TAG_CNT, DED_COUNT)
+	},
+	{ "DC_CSINVOC", SOC15_REG_ENTRY(GC, 0, mmDC_EDC_CSINVOC_CNT),
+	  SOC15_REG_FIELD(DC_EDC_CSINVOC_CNT, COUNT_ME1),
+	  0, 0
+	},
+	{ "DC_RESTORE", SOC15_REG_ENTRY(GC, 0, mmDC_EDC_RESTORE_CNT),
+	  SOC15_REG_FIELD(DC_EDC_RESTORE_CNT, COUNT_ME1),
+	  0, 0
+	},
+	{ "DC_STATE", SOC15_REG_ENTRY(GC, 0, mmDC_EDC_STATE_CNT),
+	  SOC15_REG_FIELD(DC_EDC_STATE_CNT, COUNT_ME1),
+	  0, 0
+	},
+	{ "GDS_MEM", SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_CNT, GDS_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_CNT, GDS_MEM_DED)
+	},
+	{ "GDS_INPUT_QUEUE", SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_CNT, GDS_INPUT_QUEUE_SED),
+	  0, 0
+	},
+	{ "GDS_ME0_CS_PIPE_MEM", SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PHY_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, ME0_CS_PIPE_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, ME0_CS_PIPE_MEM_DED)
+	},
+	{ "GDS_OA_PHY_PHY_CMD_RAM_MEM",
+	  SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PHY_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, PHY_CMD_RAM_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, PHY_CMD_RAM_MEM_DED)
+	},
+	{ "GDS_OA_PHY_PHY_DATA_RAM_MEM",
+	  SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PHY_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, PHY_DATA_RAM_MEM_SED),
+	  0, 0
+	},
+	{ "GDS_OA_PIPE_ME1_PIPE0_PIPE_MEM",
+	  SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PIPE_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE0_PIPE_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE0_PIPE_MEM_DED)
+	},
+	{ "GDS_OA_PIPE_ME1_PIPE1_PIPE_MEM",
+	  SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PIPE_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE1_PIPE_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE1_PIPE_MEM_DED)
+	},
+	{ "GDS_OA_PIPE_ME1_PIPE2_PIPE_MEM",
+	  SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PIPE_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE2_PIPE_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE2_PIPE_MEM_DED)
+	},
+	{ "GDS_OA_PIPE_ME1_PIPE3_PIPE_MEM",
+	  SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PIPE_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE3_PIPE_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE3_PIPE_MEM_DED)
+	},
+	{ "SPI_SR_MEM", SOC15_REG_ENTRY(GC, 0, mmSPI_EDC_CNT),
+	  SOC15_REG_FIELD(SPI_EDC_CNT, SPI_SR_MEM_SED_COUNT),
+	  0, 0
+	},
+	{ "TA_FS_DFIFO", SOC15_REG_ENTRY(GC, 0, mmTA_EDC_CNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_DFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_DFIFO_DED_COUNT)
+	},
+	{ "TA_FS_AFIFO", SOC15_REG_ENTRY(GC, 0, mmTA_EDC_CNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_AFIFO_SED_COUNT),
+	  0, 0
+	},
+	{ "TA_FL_LFIFO", SOC15_REG_ENTRY(GC, 0, mmTA_EDC_CNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FL_LFIFO_SED_COUNT),
+	  0, 0
+	},
+	{ "TA_FX_LFIFO", SOC15_REG_ENTRY(GC, 0, mmTA_EDC_CNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FX_LFIFO_SED_COUNT),
+	  0, 0
+	},
+	{ "TA_FS_CFIFO", SOC15_REG_ENTRY(GC, 0, mmTA_EDC_CNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_CFIFO_SED_COUNT),
+	  0, 0
+	},
+	{ "TCA_HOLE_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCA_EDC_CNT),
+	  SOC15_REG_FIELD(TCA_EDC_CNT, HOLE_FIFO_SED_COUNT),
+	  0, 0
+	},
+	{ "TCA_REQ_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCA_EDC_CNT),
+	  SOC15_REG_FIELD(TCA_EDC_CNT, REQ_FIFO_SED_COUNT),
+	  0, 0
+	},
+	{ "TCC_CACHE_DATA", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, CACHE_DATA_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, CACHE_DATA_DED_COUNT)
+	},
+	{ "TCC_CACHE_DIRTY", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, CACHE_DIRTY_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, CACHE_DIRTY_DED_COUNT)
+	},
+	{ "TCC_HIGH_RATE_TAG", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, HIGH_RATE_TAG_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, HIGH_RATE_TAG_DED_COUNT)
+	},
+	{ "TCC_LOW_RATE_TAG", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, LOW_RATE_TAG_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, LOW_RATE_TAG_DED_COUNT)
+	},
+	{ "TCC_SRC_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, SRC_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, SRC_FIFO_DED_COUNT)
+	},
+	{ "TCC_IN_USE_DEC", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, IN_USE_DEC_SED_COUNT),
+	  0, 0
+	},
+	{ "TCC_IN_USE_TRANSFER", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, IN_USE_TRANSFER_SED_COUNT),
+	  0, 0
+	},
+	{ "TCC_LATENCY_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, LATENCY_FIFO_SED_COUNT),
+	  0, 0
+	},
+	{ "TCC_RETURN_DATA", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, RETURN_DATA_SED_COUNT),
+	  0, 0
+	},
+	{ "TCC_RETURN_CONTROL", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, RETURN_CONTROL_SED_COUNT),
+	  0, 0
+	},
+	{ "TCC_UC_ATOMIC_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, UC_ATOMIC_FIFO_SED_COUNT),
+	  0, 0
+	},
+	{ "TCC_WRITE_RETURN", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, WRITE_RETURN_SED_COUNT),
+	  0, 0
+	},
+	{ "TCC_WRITE_CACHE_READ", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, WRITE_CACHE_READ_SED_COUNT),
+	  0, 0
+	},
+	{ "TCC_SRC_FIFO_NEXT_RAM", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, SRC_FIFO_NEXT_RAM_SED_COUNT),
+	  0, 0
+	},
+	{ "TCC_LATENCY_FIFO_NEXT_RAM", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, LATENCY_FIFO_NEXT_RAM_SED_COUNT),
+	  0, 0
+	},
+	{ "TCC_CACHE_TAG_PROBE_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, CACHE_TAG_PROBE_FIFO_SED_COUNT),
+	  0, 0
+	},
+	{ "TCC_WRRET_TAG_WRITE_RETURN", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, WRRET_TAG_WRITE_RETURN_SED_COUNT),
+	  0, 0
+	},
+	{ "TCC_ATOMIC_RETURN_BUFFER", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, ATOMIC_RETURN_BUFFER_SED_COUNT),
+	  0, 0
+	},
+	{ "TCI_WRITE_RAM", SOC15_REG_ENTRY(GC, 0, mmTCI_EDC_CNT),
+	  SOC15_REG_FIELD(TCI_EDC_CNT, WRITE_RAM_SED_COUNT),
+	  0, 0
+	},
+	{ "TCP_CACHE_RAM", SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, CACHE_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, CACHE_RAM_DED_COUNT)
+	},
+	{ "TCP_LFIFO_RAM", SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, LFIFO_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, LFIFO_RAM_DED_COUNT)
+	},
+	{ "TCP_CMD_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, CMD_FIFO_SED_COUNT),
+	  0, 0
+	},
+	{ "TCP_VM_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, VM_FIFO_SEC_COUNT),
+	  0, 0
+	},
+	{ "TCP_DB_RAM", SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, DB_RAM_SED_COUNT),
+	  0, 0
+	},
+	{ "TCP_UTCL1_LFIFO0", SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, UTCL1_LFIFO0_SEC_COUNT),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, UTCL1_LFIFO0_DED_COUNT)
+	},
+	{ "TCP_UTCL1_LFIFO1", SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, UTCL1_LFIFO1_SEC_COUNT),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, UTCL1_LFIFO1_DED_COUNT)
+	},
+	{ "TD_SS_FIFO_LO", SOC15_REG_ENTRY(GC, 0, mmTD_EDC_CNT),
+	  SOC15_REG_FIELD(TD_EDC_CNT, SS_FIFO_LO_SEC_COUNT),
+	  SOC15_REG_FIELD(TD_EDC_CNT, SS_FIFO_LO_DED_COUNT)
+	},
+	{ "TD_SS_FIFO_HI", SOC15_REG_ENTRY(GC, 0, mmTD_EDC_CNT),
+	  SOC15_REG_FIELD(TD_EDC_CNT, SS_FIFO_HI_SEC_COUNT),
+	  SOC15_REG_FIELD(TD_EDC_CNT, SS_FIFO_HI_DED_COUNT)
+	},
+	{ "TD_CS_FIFO", SOC15_REG_ENTRY(GC, 0, mmTD_EDC_CNT),
+	  SOC15_REG_FIELD(TD_EDC_CNT, CS_FIFO_SED_COUNT),
+	  0, 0
+	},
+	{ "SQ_LDS_D", SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, LDS_D_SEC_COUNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, LDS_D_DED_COUNT)
+	},
+	{ "SQ_LDS_I", SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, LDS_I_SEC_COUNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, LDS_I_DED_COUNT)
+	},
+	{ "SQ_SGPR", SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, SGPR_SEC_COUNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, SGPR_DED_COUNT)
+	},
+	{ "SQ_VGPR0", SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR0_SEC_COUNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR0_DED_COUNT)
+	},
+	{ "SQ_VGPR1", SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR1_SEC_COUNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR1_DED_COUNT)
+	},
+	{ "SQ_VGPR2", SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR2_SEC_COUNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR2_DED_COUNT)
+	},
+	{ "SQ_VGPR3", SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR3_SEC_COUNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR3_DED_COUNT)
+	},
+	{ "SQC_DATA_CU0_WRITE_DATA_BUF", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU0_WRITE_DATA_BUF_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU0_WRITE_DATA_BUF_DED_COUNT)
+	},
+	{ "SQC_DATA_CU0_UTCL1_LFIFO", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU0_UTCL1_LFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU0_UTCL1_LFIFO_DED_COUNT)
+	},
+	{ "SQC_DATA_CU1_WRITE_DATA_BUF", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU1_WRITE_DATA_BUF_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU1_WRITE_DATA_BUF_DED_COUNT)
+	},
+	{ "SQC_DATA_CU1_UTCL1_LFIFO", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU1_UTCL1_LFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU1_UTCL1_LFIFO_DED_COUNT)
+	},
+	{ "SQC_DATA_CU2_WRITE_DATA_BUF", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU2_WRITE_DATA_BUF_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU2_WRITE_DATA_BUF_DED_COUNT)
+	},
+	{ "SQC_DATA_CU2_UTCL1_LFIFO", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU2_UTCL1_LFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU2_UTCL1_LFIFO_DED_COUNT)
+	},
+	{ "SQC_INST_BANKA_TAG_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, INST_BANKA_TAG_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, INST_BANKA_TAG_RAM_DED_COUNT)
+	},
+	{ "SQC_INST_BANKA_BANK_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, INST_BANKA_BANK_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, INST_BANKA_BANK_RAM_DED_COUNT)
+	},
+	{ "SQC_DATA_BANKA_TAG_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_TAG_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_TAG_RAM_DED_COUNT)
+	},
+	{ "SQC_DATA_BANKA_BANK_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_BANK_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_BANK_RAM_DED_COUNT)
+	},
+	{ "SQC_INST_BANKA_UTCL1_MISS_FIFO", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, INST_BANKA_UTCL1_MISS_FIFO_SED_COUNT),
+	  0, 0
+	},
+	{ "SQC_INST_BANKA_MISS_FIFO", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, INST_BANKA_MISS_FIFO_SED_COUNT),
+	  0, 0
+	},
+	{ "SQC_DATA_BANKA_HIT_FIFO", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_HIT_FIFO_SED_COUNT),
+	  0, 0
+	},
+	{ "SQC_DATA_BANKA_MISS_FIFO", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_MISS_FIFO_SED_COUNT),
+	  0, 0
+	},
+	{ "SQC_DATA_BANKA_DIRTY_BIT_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_DIRTY_BIT_RAM_SED_COUNT),
+	  0, 0
+	},
+	{ "SQC_INST_UTCL1_LFIFO", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, INST_UTCL1_LFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, INST_UTCL1_LFIFO_DED_COUNT)
+	},
+	{ "SQC_INST_BANKB_TAG_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, INST_BANKB_TAG_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, INST_BANKB_TAG_RAM_DED_COUNT)
+	},
+	{ "SQC_INST_BANKB_BANK_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, INST_BANKB_BANK_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, INST_BANKB_BANK_RAM_DED_COUNT)
+	},
+	{ "SQC_DATA_BANKB_TAG_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_TAG_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_TAG_RAM_DED_COUNT)
+	},
+	{ "SQC_DATA_BANKB_BANK_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_BANK_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_BANK_RAM_DED_COUNT)
+	},
+	{ "SQC_INST_BANKB_UTCL1_MISS_FIFO", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, INST_BANKB_UTCL1_MISS_FIFO_SED_COUNT),
+	  0, 0
+	},
+	{ "SQC_INST_BANKB_MISS_FIFO", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, INST_BANKB_MISS_FIFO_SED_COUNT),
+	  0, 0
+	},
+	{ "SQC_DATA_BANKB_HIT_FIFO", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_HIT_FIFO_SED_COUNT),
+	  0, 0
+	},
+	{ "SQC_DATA_BANKB_MISS_FIFO", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_MISS_FIFO_SED_COUNT),
+	  0, 0
+	},
+	{ "SQC_DATA_BANKB_DIRTY_BIT_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_DIRTY_BIT_RAM_SED_COUNT),
+	  0, 0
+	},
+	{ "EA_DRAMRD_CMDMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT)
+	},
+	{ "EA_DRAMWR_CMDMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT)
+	},
+	{ "EA_DRAMWR_DATAMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT)
+	},
+	{ "EA_RRET_TAGMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, RRET_TAGMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, RRET_TAGMEM_DED_COUNT)
+	},
+	{ "EA_WRET_TAGMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, WRET_TAGMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, WRET_TAGMEM_DED_COUNT)
+	},
+	{ "EA_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT),
+	  0, 0
+	},
+	{ "EA_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT),
+	  0, 0
+	},
+	{ "EA_IORD_CMDMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, IORD_CMDMEM_SED_COUNT),
+	  0, 0
+	},
+	{ "EA_IOWR_CMDMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, IOWR_CMDMEM_SED_COUNT),
+	  0, 0
+	},
+	{ "EA_IOWR_DATAMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, IOWR_DATAMEM_SED_COUNT),
+	  0, 0
+	},
+	{ "GMIRD_CMDMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT)
+	},
+	{ "GMIWR_CMDMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT)
+	},
+	{ "GMIWR_DATAMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT)
+	},
+	{ "GMIRD_PAGEMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT),
+	  0, 0
+	},
+	{ "GMIWR_PAGEMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT),
+	  0, 0
+	},
+	{ "MAM_D0MEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D0MEM_SED_COUNT),
+	  0, 0
+	},
+	{ "MAM_D1MEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D1MEM_SED_COUNT),
+	  0, 0
+	},
+	{ "MAM_D2MEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D2MEM_SED_COUNT),
+	  0, 0
+	},
+	{ "MAM_D3MEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D3MEM_SED_COUNT),
+	  0, 0
+	}
+};
+
+static int gfx_v9_0_ras_error_inject(struct amdgpu_device *adev,
+				     void *inject_if, uint32_t instance_mask)
 {
-	uint32_t tmp, target;
-	struct amdgpu_ring *ring = &(adev->gfx.kiq.ring);
+	struct ras_inject_if *info = (struct ras_inject_if *)inject_if;
+	int ret;
+	struct ta_ras_trigger_error_input block_info = { 0 };
 
-	if (ring->me == 1)
-		target = SOC15_REG_OFFSET(GC, 0, mmCP_ME1_PIPE0_INT_CNTL);
-	else
-		target = SOC15_REG_OFFSET(GC, 0, mmCP_ME2_PIPE0_INT_CNTL);
-	target += ring->pipe;
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
+		return -EINVAL;
 
-	switch (type) {
-	case AMDGPU_CP_KIQ_IRQ_DRIVER0:
-		if (state == AMDGPU_IRQ_STATE_DISABLE) {
-			tmp = RREG32_SOC15(GC, 0, mmCPC_INT_CNTL);
-			tmp = REG_SET_FIELD(tmp, CPC_INT_CNTL,
-						 GENERIC2_INT_ENABLE, 0);
-			WREG32_SOC15(GC, 0, mmCPC_INT_CNTL, tmp);
-
-			tmp = RREG32(target);
-			tmp = REG_SET_FIELD(tmp, CP_ME2_PIPE0_INT_CNTL,
-						 GENERIC2_INT_ENABLE, 0);
-			WREG32(target, tmp);
-		} else {
-			tmp = RREG32_SOC15(GC, 0, mmCPC_INT_CNTL);
-			tmp = REG_SET_FIELD(tmp, CPC_INT_CNTL,
-						 GENERIC2_INT_ENABLE, 1);
-			WREG32_SOC15(GC, 0, mmCPC_INT_CNTL, tmp);
-
-			tmp = RREG32(target);
-			tmp = REG_SET_FIELD(tmp, CP_ME2_PIPE0_INT_CNTL,
-						 GENERIC2_INT_ENABLE, 1);
-			WREG32(target, tmp);
+	if (info->head.sub_block_index >= ARRAY_SIZE(ras_gfx_subblocks))
+		return -EINVAL;
+
+	if (!ras_gfx_subblocks[info->head.sub_block_index].name)
+		return -EPERM;
+
+	if (!(ras_gfx_subblocks[info->head.sub_block_index].hw_supported_error_type &
+	      info->head.type)) {
+		DRM_ERROR("GFX Subblock %s, hardware do not support type 0x%x\n",
+			ras_gfx_subblocks[info->head.sub_block_index].name,
+			info->head.type);
+		return -EPERM;
+	}
+
+	if (!(ras_gfx_subblocks[info->head.sub_block_index].sw_supported_error_type &
+	      info->head.type)) {
+		DRM_ERROR("GFX Subblock %s, driver do not support type 0x%x\n",
+			ras_gfx_subblocks[info->head.sub_block_index].name,
+			info->head.type);
+		return -EPERM;
+	}
+
+	block_info.block_id = amdgpu_ras_block_to_ta(info->head.block);
+	block_info.sub_block_index =
+		ras_gfx_subblocks[info->head.sub_block_index].ta_subblock;
+	block_info.inject_error_type = amdgpu_ras_error_to_ta(info->head.type);
+	block_info.address = info->address;
+	block_info.value = info->value;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	ret = psp_ras_trigger_error(&adev->psp, &block_info, instance_mask);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return ret;
+}
+
+static const char * const vml2_mems[] = {
+	"UTC_VML2_BANK_CACHE_0_BIGK_MEM0",
+	"UTC_VML2_BANK_CACHE_0_BIGK_MEM1",
+	"UTC_VML2_BANK_CACHE_0_4K_MEM0",
+	"UTC_VML2_BANK_CACHE_0_4K_MEM1",
+	"UTC_VML2_BANK_CACHE_1_BIGK_MEM0",
+	"UTC_VML2_BANK_CACHE_1_BIGK_MEM1",
+	"UTC_VML2_BANK_CACHE_1_4K_MEM0",
+	"UTC_VML2_BANK_CACHE_1_4K_MEM1",
+	"UTC_VML2_BANK_CACHE_2_BIGK_MEM0",
+	"UTC_VML2_BANK_CACHE_2_BIGK_MEM1",
+	"UTC_VML2_BANK_CACHE_2_4K_MEM0",
+	"UTC_VML2_BANK_CACHE_2_4K_MEM1",
+	"UTC_VML2_BANK_CACHE_3_BIGK_MEM0",
+	"UTC_VML2_BANK_CACHE_3_BIGK_MEM1",
+	"UTC_VML2_BANK_CACHE_3_4K_MEM0",
+	"UTC_VML2_BANK_CACHE_3_4K_MEM1",
+};
+
+static const char * const vml2_walker_mems[] = {
+	"UTC_VML2_CACHE_PDE0_MEM0",
+	"UTC_VML2_CACHE_PDE0_MEM1",
+	"UTC_VML2_CACHE_PDE1_MEM0",
+	"UTC_VML2_CACHE_PDE1_MEM1",
+	"UTC_VML2_CACHE_PDE2_MEM0",
+	"UTC_VML2_CACHE_PDE2_MEM1",
+	"UTC_VML2_RDIF_LOG_FIFO",
+};
+
+static const char * const atc_l2_cache_2m_mems[] = {
+	"UTC_ATCL2_CACHE_2M_BANK0_WAY0_MEM",
+	"UTC_ATCL2_CACHE_2M_BANK0_WAY1_MEM",
+	"UTC_ATCL2_CACHE_2M_BANK1_WAY0_MEM",
+	"UTC_ATCL2_CACHE_2M_BANK1_WAY1_MEM",
+};
+
+static const char *atc_l2_cache_4k_mems[] = {
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM0",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM1",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM2",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM3",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM4",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM5",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM6",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM7",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM0",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM1",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM2",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM3",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM4",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM5",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM6",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM7",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM0",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM1",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM2",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM3",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM4",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM5",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM6",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM7",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM0",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM1",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM2",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM3",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM4",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM5",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM6",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM7",
+};
+
+static int gfx_v9_0_query_utc_edc_status(struct amdgpu_device *adev,
+					 struct ras_err_data *err_data)
+{
+	uint32_t i, data;
+	uint32_t sec_count, ded_count;
+
+	WREG32_SOC15(GC, 0, mmVM_L2_MEM_ECC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmVM_L2_MEM_ECC_CNT, 0);
+	WREG32_SOC15(GC, 0, mmVM_L2_WALKER_MEM_ECC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmVM_L2_WALKER_MEM_ECC_CNT, 0);
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_EDC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_EDC_CNT, 0);
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_EDC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_EDC_CNT, 0);
+
+	for (i = 0; i < ARRAY_SIZE(vml2_mems); i++) {
+		WREG32_SOC15(GC, 0, mmVM_L2_MEM_ECC_INDEX, i);
+		data = RREG32_SOC15(GC, 0, mmVM_L2_MEM_ECC_CNT);
+
+		sec_count = REG_GET_FIELD(data, VM_L2_MEM_ECC_CNT, SEC_COUNT);
+		if (sec_count) {
+			dev_info(adev->dev, "Instance[%d]: SubBlock %s, "
+				"SEC %d\n", i, vml2_mems[i], sec_count);
+			err_data->ce_count += sec_count;
+		}
+
+		ded_count = REG_GET_FIELD(data, VM_L2_MEM_ECC_CNT, DED_COUNT);
+		if (ded_count) {
+			dev_info(adev->dev, "Instance[%d]: SubBlock %s, "
+				"DED %d\n", i, vml2_mems[i], ded_count);
+			err_data->ue_count += ded_count;
+		}
+	}
+
+	for (i = 0; i < ARRAY_SIZE(vml2_walker_mems); i++) {
+		WREG32_SOC15(GC, 0, mmVM_L2_WALKER_MEM_ECC_INDEX, i);
+		data = RREG32_SOC15(GC, 0, mmVM_L2_WALKER_MEM_ECC_CNT);
+
+		sec_count = REG_GET_FIELD(data, VM_L2_WALKER_MEM_ECC_CNT,
+						SEC_COUNT);
+		if (sec_count) {
+			dev_info(adev->dev, "Instance[%d]: SubBlock %s, "
+				"SEC %d\n", i, vml2_walker_mems[i], sec_count);
+			err_data->ce_count += sec_count;
+		}
+
+		ded_count = REG_GET_FIELD(data, VM_L2_WALKER_MEM_ECC_CNT,
+						DED_COUNT);
+		if (ded_count) {
+			dev_info(adev->dev, "Instance[%d]: SubBlock %s, "
+				"DED %d\n", i, vml2_walker_mems[i], ded_count);
+			err_data->ue_count += ded_count;
+		}
+	}
+
+	for (i = 0; i < ARRAY_SIZE(atc_l2_cache_2m_mems); i++) {
+		WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_EDC_INDEX, i);
+		data = RREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_EDC_CNT);
+
+		sec_count = (data & 0x00006000L) >> 0xd;
+		if (sec_count) {
+			dev_info(adev->dev, "Instance[%d]: SubBlock %s, "
+				"SEC %d\n", i, atc_l2_cache_2m_mems[i],
+				sec_count);
+			err_data->ce_count += sec_count;
+		}
+	}
+
+	for (i = 0; i < ARRAY_SIZE(atc_l2_cache_4k_mems); i++) {
+		WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_EDC_INDEX, i);
+		data = RREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_EDC_CNT);
+
+		sec_count = (data & 0x00006000L) >> 0xd;
+		if (sec_count) {
+			dev_info(adev->dev, "Instance[%d]: SubBlock %s, "
+				"SEC %d\n", i, atc_l2_cache_4k_mems[i],
+				sec_count);
+			err_data->ce_count += sec_count;
+		}
+
+		ded_count = (data & 0x00018000L) >> 0xf;
+		if (ded_count) {
+			dev_info(adev->dev, "Instance[%d]: SubBlock %s, "
+				"DED %d\n", i, atc_l2_cache_4k_mems[i],
+				ded_count);
+			err_data->ue_count += ded_count;
 		}
-		break;
-	default:
-		BUG(); /* kiq only support GENERIC2_INT now */
-		break;
 	}
+
+	WREG32_SOC15(GC, 0, mmVM_L2_MEM_ECC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmVM_L2_WALKER_MEM_ECC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_EDC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_EDC_INDEX, 255);
+
 	return 0;
 }
 
-static int gfx_v9_0_kiq_irq(struct amdgpu_device *adev,
-			    struct amdgpu_irq_src *source,
-			    struct amdgpu_iv_entry *entry)
+static int gfx_v9_0_ras_error_count(struct amdgpu_device *adev,
+	const struct soc15_reg_entry *reg,
+	uint32_t se_id, uint32_t inst_id, uint32_t value,
+	uint32_t *sec_count, uint32_t *ded_count)
 {
-	u8 me_id, pipe_id, queue_id;
-	struct amdgpu_ring *ring = &(adev->gfx.kiq.ring);
+	uint32_t i;
+	uint32_t sec_cnt, ded_cnt;
 
-	me_id = (entry->ring_id & 0x0c) >> 2;
-	pipe_id = (entry->ring_id & 0x03) >> 0;
-	queue_id = (entry->ring_id & 0x70) >> 4;
-	DRM_DEBUG("IH: CPC GENERIC2_INT, me:%d, pipe:%d, queue:%d\n",
-		   me_id, pipe_id, queue_id);
+	for (i = 0; i < ARRAY_SIZE(gfx_v9_0_ras_fields); i++) {
+		if(gfx_v9_0_ras_fields[i].reg_offset != reg->reg_offset ||
+			gfx_v9_0_ras_fields[i].seg != reg->seg ||
+			gfx_v9_0_ras_fields[i].inst != reg->inst)
+			continue;
+
+		sec_cnt = (value &
+				gfx_v9_0_ras_fields[i].sec_count_mask) >>
+				gfx_v9_0_ras_fields[i].sec_count_shift;
+		if (sec_cnt) {
+			dev_info(adev->dev, "GFX SubBlock %s, "
+				"Instance[%d][%d], SEC %d\n",
+				gfx_v9_0_ras_fields[i].name,
+				se_id, inst_id,
+				sec_cnt);
+			*sec_count += sec_cnt;
+		}
+
+		ded_cnt = (value &
+				gfx_v9_0_ras_fields[i].ded_count_mask) >>
+				gfx_v9_0_ras_fields[i].ded_count_shift;
+		if (ded_cnt) {
+			dev_info(adev->dev, "GFX SubBlock %s, "
+				"Instance[%d][%d], DED %d\n",
+				gfx_v9_0_ras_fields[i].name,
+				se_id, inst_id,
+				ded_cnt);
+			*ded_count += ded_cnt;
+		}
+	}
 
-	amdgpu_fence_process(ring);
 	return 0;
 }
 
+static void gfx_v9_0_reset_ras_error_count(struct amdgpu_device *adev)
+{
+	int i, j, k;
+
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
+		return;
+
+	/* read back registers to clear the counters */
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < ARRAY_SIZE(gfx_v9_0_edc_counter_regs); i++) {
+		for (j = 0; j < gfx_v9_0_edc_counter_regs[i].se_num; j++) {
+			for (k = 0; k < gfx_v9_0_edc_counter_regs[i].instance; k++) {
+				amdgpu_gfx_select_se_sh(adev, j, 0x0, k, 0);
+				RREG32(SOC15_REG_ENTRY_OFFSET(gfx_v9_0_edc_counter_regs[i]));
+			}
+		}
+	}
+	WREG32_SOC15(GC, 0, mmGRBM_GFX_INDEX, 0xe0000000);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	WREG32_SOC15(GC, 0, mmVM_L2_MEM_ECC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmVM_L2_MEM_ECC_CNT, 0);
+	WREG32_SOC15(GC, 0, mmVM_L2_WALKER_MEM_ECC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmVM_L2_WALKER_MEM_ECC_CNT, 0);
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_EDC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_EDC_CNT, 0);
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_EDC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_EDC_CNT, 0);
+
+	for (i = 0; i < ARRAY_SIZE(vml2_mems); i++) {
+		WREG32_SOC15(GC, 0, mmVM_L2_MEM_ECC_INDEX, i);
+		RREG32_SOC15(GC, 0, mmVM_L2_MEM_ECC_CNT);
+	}
+
+	for (i = 0; i < ARRAY_SIZE(vml2_walker_mems); i++) {
+		WREG32_SOC15(GC, 0, mmVM_L2_WALKER_MEM_ECC_INDEX, i);
+		RREG32_SOC15(GC, 0, mmVM_L2_WALKER_MEM_ECC_CNT);
+	}
+
+	for (i = 0; i < ARRAY_SIZE(atc_l2_cache_2m_mems); i++) {
+		WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_EDC_INDEX, i);
+		RREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_EDC_CNT);
+	}
+
+	for (i = 0; i < ARRAY_SIZE(atc_l2_cache_4k_mems); i++) {
+		WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_EDC_INDEX, i);
+		RREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_EDC_CNT);
+	}
+
+	WREG32_SOC15(GC, 0, mmVM_L2_MEM_ECC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmVM_L2_WALKER_MEM_ECC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_EDC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_EDC_INDEX, 255);
+}
+
+static void gfx_v9_0_query_ras_error_count(struct amdgpu_device *adev,
+					  void *ras_error_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+	uint32_t sec_count = 0, ded_count = 0;
+	uint32_t i, j, k;
+	uint32_t reg_value;
+
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
+		return;
+
+	err_data->ue_count = 0;
+	err_data->ce_count = 0;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	for (i = 0; i < ARRAY_SIZE(gfx_v9_0_edc_counter_regs); i++) {
+		for (j = 0; j < gfx_v9_0_edc_counter_regs[i].se_num; j++) {
+			for (k = 0; k < gfx_v9_0_edc_counter_regs[i].instance; k++) {
+				amdgpu_gfx_select_se_sh(adev, j, 0, k, 0);
+				reg_value =
+					RREG32(SOC15_REG_ENTRY_OFFSET(gfx_v9_0_edc_counter_regs[i]));
+				if (reg_value)
+					gfx_v9_0_ras_error_count(adev,
+						&gfx_v9_0_edc_counter_regs[i],
+						j, k, reg_value,
+						&sec_count, &ded_count);
+			}
+		}
+	}
+
+	err_data->ce_count += sec_count;
+	err_data->ue_count += ded_count;
+
+	amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	gfx_v9_0_query_utc_edc_status(adev, err_data);
+}
+
+static void gfx_v9_0_emit_mem_sync(struct amdgpu_ring *ring)
+{
+	const unsigned int cp_coher_cntl =
+			PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_SH_ICACHE_ACTION_ENA(1) |
+			PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_SH_KCACHE_ACTION_ENA(1) |
+			PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_TC_ACTION_ENA(1) |
+			PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_TCL1_ACTION_ENA(1) |
+			PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_TC_WB_ACTION_ENA(1);
+
+	/* ACQUIRE_MEM -make one or more surfaces valid for use by the subsequent operations */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_ACQUIRE_MEM, 5));
+	amdgpu_ring_write(ring, cp_coher_cntl); /* CP_COHER_CNTL */
+	amdgpu_ring_write(ring, 0xffffffff);  /* CP_COHER_SIZE */
+	amdgpu_ring_write(ring, 0xffffff);  /* CP_COHER_SIZE_HI */
+	amdgpu_ring_write(ring, 0); /* CP_COHER_BASE */
+	amdgpu_ring_write(ring, 0);  /* CP_COHER_BASE_HI */
+	amdgpu_ring_write(ring, 0x0000000A); /* POLL_INTERVAL */
+}
+
+static void gfx_v9_0_emit_wave_limit_cs(struct amdgpu_ring *ring,
+					uint32_t pipe, bool enable)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t val;
+	uint32_t wcl_cs_reg;
+
+	/* mmSPI_WCL_PIPE_PERCENT_CS[0-7]_DEFAULT values are same */
+	val = enable ? 0x1 : mmSPI_WCL_PIPE_PERCENT_CS0_DEFAULT;
+
+	switch (pipe) {
+	case 0:
+		wcl_cs_reg = SOC15_REG_OFFSET(GC, 0, mmSPI_WCL_PIPE_PERCENT_CS0);
+		break;
+	case 1:
+		wcl_cs_reg = SOC15_REG_OFFSET(GC, 0, mmSPI_WCL_PIPE_PERCENT_CS1);
+		break;
+	case 2:
+		wcl_cs_reg = SOC15_REG_OFFSET(GC, 0, mmSPI_WCL_PIPE_PERCENT_CS2);
+		break;
+	case 3:
+		wcl_cs_reg = SOC15_REG_OFFSET(GC, 0, mmSPI_WCL_PIPE_PERCENT_CS3);
+		break;
+	default:
+		DRM_DEBUG("invalid pipe %d\n", pipe);
+		return;
+	}
+
+	amdgpu_ring_emit_wreg(ring, wcl_cs_reg, val);
+
+}
+static void gfx_v9_0_emit_wave_limit(struct amdgpu_ring *ring, bool enable)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t val;
+	int i;
+
+
+	/* mmSPI_WCL_PIPE_PERCENT_GFX is 7 bit multiplier register to limit
+	 * number of gfx waves. Setting 5 bit will make sure gfx only gets
+	 * around 25% of gpu resources.
+	 */
+	val = enable ? 0x1f : mmSPI_WCL_PIPE_PERCENT_GFX_DEFAULT;
+	amdgpu_ring_emit_wreg(ring,
+			      SOC15_REG_OFFSET(GC, 0, mmSPI_WCL_PIPE_PERCENT_GFX),
+			      val);
+
+	/* Restrict waves for normal/low priority compute queues as well
+	 * to get best QoS for high priority compute jobs.
+	 *
+	 * amdgpu controls only 1st ME(0-3 CS pipes).
+	 */
+	for (i = 0; i < adev->gfx.mec.num_pipe_per_mec; i++) {
+		if (i != ring->pipe)
+			gfx_v9_0_emit_wave_limit_cs(ring, i, enable);
+
+	}
+}
+
+static void gfx_v9_ring_insert_nop(struct amdgpu_ring *ring, uint32_t num_nop)
+{
+	/* Header itself is a NOP packet */
+	if (num_nop == 1) {
+		amdgpu_ring_write(ring, ring->funcs->nop);
+		return;
+	}
+
+	/* Max HW optimization till 0x3ffe, followed by remaining one NOP at a time*/
+	amdgpu_ring_write(ring, PACKET3(PACKET3_NOP, min(num_nop - 2, 0x3ffe)));
+
+	/* Header is at index 0, followed by num_nops - 1 NOP packet's */
+	amdgpu_ring_insert_nop(ring, num_nop - 1);
+}
+
+static int gfx_v9_0_reset_kcq(struct amdgpu_ring *ring,
+			      unsigned int vmid,
+			      struct amdgpu_fence *timedout_fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[0];
+	struct amdgpu_ring *kiq_ring = &kiq->ring;
+	unsigned long flags;
+	int i, r;
+
+	if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
+		return -EINVAL;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+
+	spin_lock_irqsave(&kiq->ring_lock, flags);
+
+	if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->unmap_queues_size)) {
+		spin_unlock_irqrestore(&kiq->ring_lock, flags);
+		return -ENOMEM;
+	}
+
+	kiq->pmf->kiq_unmap_queues(kiq_ring, ring, RESET_QUEUES,
+				   0, 0);
+	amdgpu_ring_commit(kiq_ring);
+
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+
+	r = amdgpu_ring_test_ring(kiq_ring);
+	if (r)
+		return r;
+
+	/* make sure dequeue is complete*/
+	amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
+	mutex_lock(&adev->srbm_mutex);
+	soc15_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0, 0);
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (!(RREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE) & 1))
+			break;
+		udelay(1);
+	}
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+	soc15_grbm_select(adev, 0, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
+	if (r) {
+		dev_err(adev->dev, "fail to wait on hqd deactive\n");
+		return r;
+	}
+
+	r = gfx_v9_0_kcq_init_queue(ring, true);
+	if (r) {
+		dev_err(adev->dev, "fail to init kcq\n");
+		return r;
+	}
+	spin_lock_irqsave(&kiq->ring_lock, flags);
+	r = amdgpu_ring_alloc(kiq_ring, kiq->pmf->map_queues_size);
+	if (r) {
+		spin_unlock_irqrestore(&kiq->ring_lock, flags);
+		return -ENOMEM;
+	}
+	kiq->pmf->kiq_map_queues(kiq_ring, ring);
+	amdgpu_ring_commit(kiq_ring);
+	r = amdgpu_ring_test_ring(kiq_ring);
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+	if (r) {
+		DRM_ERROR("fail to remap queue\n");
+		return r;
+	}
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static void gfx_v9_ip_print(struct amdgpu_ip_block *ip_block, struct drm_printer *p)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	uint32_t i, j, k, reg, index = 0;
+	uint32_t reg_count = ARRAY_SIZE(gc_reg_list_9);
+
+	if (!adev->gfx.ip_dump_core)
+		return;
+
+	for (i = 0; i < reg_count; i++)
+		drm_printf(p, "%-50s \t 0x%08x\n",
+			   gc_reg_list_9[i].reg_name,
+			   adev->gfx.ip_dump_core[i]);
+
+	/* print compute queue registers for all instances */
+	if (!adev->gfx.ip_dump_compute_queues)
+		return;
+
+	reg_count = ARRAY_SIZE(gc_cp_reg_list_9);
+	drm_printf(p, "\nnum_mec: %d num_pipe: %d num_queue: %d\n",
+		   adev->gfx.mec.num_mec,
+		   adev->gfx.mec.num_pipe_per_mec,
+		   adev->gfx.mec.num_queue_per_pipe);
+
+	for (i = 0; i < adev->gfx.mec.num_mec; i++) {
+		for (j = 0; j < adev->gfx.mec.num_pipe_per_mec; j++) {
+			for (k = 0; k < adev->gfx.mec.num_queue_per_pipe; k++) {
+				drm_printf(p, "\nmec %d, pipe %d, queue %d\n", i, j, k);
+				for (reg = 0; reg < reg_count; reg++) {
+					if (i && gc_cp_reg_list_9[reg].reg_offset == mmCP_MEC_ME1_HEADER_DUMP)
+						drm_printf(p, "%-50s \t 0x%08x\n",
+							   "mmCP_MEC_ME2_HEADER_DUMP",
+							   adev->gfx.ip_dump_compute_queues[index + reg]);
+					else
+						drm_printf(p, "%-50s \t 0x%08x\n",
+							   gc_cp_reg_list_9[reg].reg_name,
+							   adev->gfx.ip_dump_compute_queues[index + reg]);
+				}
+				index += reg_count;
+			}
+		}
+	}
+
+}
+
+static void gfx_v9_ip_dump(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	uint32_t i, j, k, reg, index = 0;
+	uint32_t reg_count = ARRAY_SIZE(gc_reg_list_9);
+
+	if (!adev->gfx.ip_dump_core || !adev->gfx.num_gfx_rings)
+		return;
+
+	amdgpu_gfx_off_ctrl(adev, false);
+	for (i = 0; i < reg_count; i++)
+		adev->gfx.ip_dump_core[i] = RREG32(SOC15_REG_ENTRY_OFFSET(gc_reg_list_9[i]));
+	amdgpu_gfx_off_ctrl(adev, true);
+
+	/* dump compute queue registers for all instances */
+	if (!adev->gfx.ip_dump_compute_queues)
+		return;
+
+	reg_count = ARRAY_SIZE(gc_cp_reg_list_9);
+	amdgpu_gfx_off_ctrl(adev, false);
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < adev->gfx.mec.num_mec; i++) {
+		for (j = 0; j < adev->gfx.mec.num_pipe_per_mec; j++) {
+			for (k = 0; k < adev->gfx.mec.num_queue_per_pipe; k++) {
+				/* ME0 is for GFX so start from 1 for CP */
+				soc15_grbm_select(adev, 1 + i, j, k, 0, 0);
+
+				for (reg = 0; reg < reg_count; reg++) {
+					if (i && gc_cp_reg_list_9[reg].reg_offset == mmCP_MEC_ME1_HEADER_DUMP)
+						adev->gfx.ip_dump_compute_queues[index + reg] =
+							RREG32(SOC15_REG_OFFSET(GC, 0, mmCP_MEC_ME2_HEADER_DUMP));
+					else
+						adev->gfx.ip_dump_compute_queues[index + reg] =
+							RREG32(SOC15_REG_ENTRY_OFFSET(
+								       gc_cp_reg_list_9[reg]));
+				}
+				index += reg_count;
+			}
+		}
+	}
+	soc15_grbm_select(adev, 0, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+	amdgpu_gfx_off_ctrl(adev, true);
+
+}
+
+static void gfx_v9_0_ring_emit_cleaner_shader(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	/* Emit the cleaner shader */
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2))
+		amdgpu_ring_write(ring, PACKET3(PACKET3_RUN_CLEANER_SHADER, 0));
+	else
+		amdgpu_ring_write(ring, PACKET3(PACKET3_RUN_CLEANER_SHADER_9_0, 0));
+
+	amdgpu_ring_write(ring, 0);  /* RESERVED field, programmed to zero */
+}
+
+static void gfx_v9_0_ring_begin_use_compute(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ip_block *gfx_block =
+		amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_GFX);
+
+	amdgpu_gfx_enforce_isolation_ring_begin_use(ring);
+
+	/* Raven and PCO APUs seem to have stability issues
+	 * with compute and gfxoff and gfx pg.  Disable gfx pg during
+	 * submission and allow again afterwards.
+	 */
+	if (gfx_block && amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 1, 0))
+		gfx_v9_0_set_powergating_state(gfx_block, AMD_PG_STATE_UNGATE);
+}
+
+static void gfx_v9_0_ring_end_use_compute(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ip_block *gfx_block =
+		amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_GFX);
+
+	/* Raven and PCO APUs seem to have stability issues
+	 * with compute and gfxoff and gfx pg.  Disable gfx pg during
+	 * submission and allow again afterwards.
+	 */
+	if (gfx_block && amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 1, 0))
+		gfx_v9_0_set_powergating_state(gfx_block, AMD_PG_STATE_GATE);
+
+	amdgpu_gfx_enforce_isolation_ring_end_use(ring);
+}
+
 static const struct amd_ip_funcs gfx_v9_0_ip_funcs = {
 	.name = "gfx_v9_0",
 	.early_init = gfx_v9_0_early_init,
@@ -4118,6 +7398,8 @@ static const struct amd_ip_funcs gfx_v9_0_ip_funcs = {
 	.set_clockgating_state = gfx_v9_0_set_clockgating_state,
 	.set_powergating_state = gfx_v9_0_set_powergating_state,
 	.get_clockgating_state = gfx_v9_0_get_clockgating_state,
+	.dump_ip_state = gfx_v9_ip_dump,
+	.print_ip_state = gfx_v9_ip_print,
 };
 
 static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_gfx = {
@@ -4125,14 +7407,16 @@ static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_gfx = {
 	.align_mask = 0xff,
 	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
 	.support_64bit_ptrs = true,
-	.vmhub = AMDGPU_GFXHUB,
+	.secure_submission_supported = true,
 	.get_rptr = gfx_v9_0_ring_get_rptr_gfx,
 	.get_wptr = gfx_v9_0_ring_get_wptr_gfx,
 	.set_wptr = gfx_v9_0_ring_set_wptr_gfx,
 	.emit_frame_size = /* totally 242 maximum if 16 IBs */
 		5 +  /* COND_EXEC */
 		7 +  /* PIPELINE_SYNC */
-		24 + /* VM_FLUSH */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 5 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 +
+		2 + /* VM_FLUSH */
 		8 +  /* FENCE for VM_FLUSH */
 		20 + /* GDS switch */
 		4 + /* double SWITCH_BUFFER,
@@ -4146,7 +7430,66 @@ static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_gfx = {
 		3 + /* CNTX_CTRL */
 		5 + /* HDP_INVL */
 		8 + 8 + /* FENCE x2 */
-		2, /* SWITCH_BUFFER */
+		2 + /* SWITCH_BUFFER */
+		7 + /* gfx_v9_0_emit_mem_sync */
+		2, /* gfx_v9_0_ring_emit_cleaner_shader */
+	.emit_ib_size =	4, /* gfx_v9_0_ring_emit_ib_gfx */
+	.emit_ib = gfx_v9_0_ring_emit_ib_gfx,
+	.emit_fence = gfx_v9_0_ring_emit_fence,
+	.emit_pipeline_sync = gfx_v9_0_ring_emit_pipeline_sync,
+	.emit_vm_flush = gfx_v9_0_ring_emit_vm_flush,
+	.emit_gds_switch = gfx_v9_0_ring_emit_gds_switch,
+	.emit_hdp_flush = gfx_v9_0_ring_emit_hdp_flush,
+	.test_ring = gfx_v9_0_ring_test_ring,
+	.insert_nop = gfx_v9_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_switch_buffer = gfx_v9_ring_emit_sb,
+	.emit_cntxcntl = gfx_v9_ring_emit_cntxcntl,
+	.init_cond_exec = gfx_v9_0_ring_emit_init_cond_exec,
+	.preempt_ib = gfx_v9_0_ring_preempt_ib,
+	.emit_frame_cntl = gfx_v9_0_ring_emit_frame_cntl,
+	.emit_wreg = gfx_v9_0_ring_emit_wreg,
+	.emit_reg_wait = gfx_v9_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = gfx_v9_0_ring_emit_reg_write_reg_wait,
+	.soft_recovery = gfx_v9_0_ring_soft_recovery,
+	.emit_mem_sync = gfx_v9_0_emit_mem_sync,
+	.emit_cleaner_shader = gfx_v9_0_ring_emit_cleaner_shader,
+	.begin_use = amdgpu_gfx_enforce_isolation_ring_begin_use,
+	.end_use = amdgpu_gfx_enforce_isolation_ring_end_use,
+};
+
+static const struct amdgpu_ring_funcs gfx_v9_0_sw_ring_funcs_gfx = {
+	.type = AMDGPU_RING_TYPE_GFX,
+	.align_mask = 0xff,
+	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
+	.support_64bit_ptrs = true,
+	.secure_submission_supported = true,
+	.get_rptr = amdgpu_sw_ring_get_rptr_gfx,
+	.get_wptr = amdgpu_sw_ring_get_wptr_gfx,
+	.set_wptr = amdgpu_sw_ring_set_wptr_gfx,
+	.emit_frame_size = /* totally 242 maximum if 16 IBs */
+		5 +  /* COND_EXEC */
+		7 +  /* PIPELINE_SYNC */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 5 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 +
+		2 + /* VM_FLUSH */
+		8 +  /* FENCE for VM_FLUSH */
+		20 + /* GDS switch */
+		4 + /* double SWITCH_BUFFER,
+		     * the first COND_EXEC jump to the place just
+		     * prior to this double SWITCH_BUFFER
+		     */
+		5 + /* COND_EXEC */
+		7 +	 /*	HDP_flush */
+		4 +	 /*	VGT_flush */
+		14 + /*	CE_META */
+		31 + /*	DE_META */
+		3 + /* CNTX_CTRL */
+		5 + /* HDP_INVL */
+		8 + 8 + /* FENCE x2 */
+		2 + /* SWITCH_BUFFER */
+		7 + /* gfx_v9_0_emit_mem_sync */
+		2, /* gfx_v9_0_ring_emit_cleaner_shader */
 	.emit_ib_size =	4, /* gfx_v9_0_ring_emit_ib_gfx */
 	.emit_ib = gfx_v9_0_ring_emit_ib_gfx,
 	.emit_fence = gfx_v9_0_ring_emit_fence,
@@ -4154,16 +7497,25 @@ static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_gfx = {
 	.emit_vm_flush = gfx_v9_0_ring_emit_vm_flush,
 	.emit_gds_switch = gfx_v9_0_ring_emit_gds_switch,
 	.emit_hdp_flush = gfx_v9_0_ring_emit_hdp_flush,
-	.emit_hdp_invalidate = gfx_v9_0_ring_emit_hdp_invalidate,
 	.test_ring = gfx_v9_0_ring_test_ring,
 	.test_ib = gfx_v9_0_ring_test_ib,
-	.insert_nop = amdgpu_ring_insert_nop,
+	.insert_nop = gfx_v9_ring_insert_nop,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
 	.emit_switch_buffer = gfx_v9_ring_emit_sb,
 	.emit_cntxcntl = gfx_v9_ring_emit_cntxcntl,
 	.init_cond_exec = gfx_v9_0_ring_emit_init_cond_exec,
-	.patch_cond_exec = gfx_v9_0_ring_emit_patch_cond_exec,
-	.emit_tmz = gfx_v9_0_ring_emit_tmz,
+	.emit_frame_cntl = gfx_v9_0_ring_emit_frame_cntl,
+	.emit_wreg = gfx_v9_0_ring_emit_wreg,
+	.emit_reg_wait = gfx_v9_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = gfx_v9_0_ring_emit_reg_write_reg_wait,
+	.soft_recovery = gfx_v9_0_ring_soft_recovery,
+	.emit_mem_sync = gfx_v9_0_emit_mem_sync,
+	.patch_cntl = gfx_v9_0_ring_patch_cntl,
+	.patch_de = gfx_v9_0_ring_patch_de_meta,
+	.patch_ce = gfx_v9_0_ring_patch_ce_meta,
+	.emit_cleaner_shader = gfx_v9_0_ring_emit_cleaner_shader,
+	.begin_use = amdgpu_gfx_enforce_isolation_ring_begin_use,
+	.end_use = amdgpu_gfx_enforce_isolation_ring_end_use,
 };
 
 static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_compute = {
@@ -4171,29 +7523,42 @@ static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_compute = {
 	.align_mask = 0xff,
 	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
 	.support_64bit_ptrs = true,
-	.vmhub = AMDGPU_GFXHUB,
 	.get_rptr = gfx_v9_0_ring_get_rptr_compute,
 	.get_wptr = gfx_v9_0_ring_get_wptr_compute,
 	.set_wptr = gfx_v9_0_ring_set_wptr_compute,
 	.emit_frame_size =
 		20 + /* gfx_v9_0_ring_emit_gds_switch */
 		7 + /* gfx_v9_0_ring_emit_hdp_flush */
-		5 + /* gfx_v9_0_ring_emit_hdp_invalidate */
+		5 + /* hdp invalidate */
 		7 + /* gfx_v9_0_ring_emit_pipeline_sync */
-		24 + /* gfx_v9_0_ring_emit_vm_flush */
-		8 + 8 + 8, /* gfx_v9_0_ring_emit_fence x3 for user fence, vm fence */
-	.emit_ib_size =	4, /* gfx_v9_0_ring_emit_ib_compute */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 5 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 +
+		8 + 8 + 8 + /* gfx_v9_0_ring_emit_fence x3 for user fence, vm fence */
+		7 + /* gfx_v9_0_emit_mem_sync */
+		5 + /* gfx_v9_0_emit_wave_limit for updating mmSPI_WCL_PIPE_PERCENT_GFX register */
+		15 + /* for updating 3 mmSPI_WCL_PIPE_PERCENT_CS registers */
+		2, /* gfx_v9_0_ring_emit_cleaner_shader */
+	.emit_ib_size =	7, /* gfx_v9_0_ring_emit_ib_compute */
 	.emit_ib = gfx_v9_0_ring_emit_ib_compute,
 	.emit_fence = gfx_v9_0_ring_emit_fence,
 	.emit_pipeline_sync = gfx_v9_0_ring_emit_pipeline_sync,
 	.emit_vm_flush = gfx_v9_0_ring_emit_vm_flush,
 	.emit_gds_switch = gfx_v9_0_ring_emit_gds_switch,
 	.emit_hdp_flush = gfx_v9_0_ring_emit_hdp_flush,
-	.emit_hdp_invalidate = gfx_v9_0_ring_emit_hdp_invalidate,
 	.test_ring = gfx_v9_0_ring_test_ring,
 	.test_ib = gfx_v9_0_ring_test_ib,
-	.insert_nop = amdgpu_ring_insert_nop,
+	.insert_nop = gfx_v9_ring_insert_nop,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_wreg = gfx_v9_0_ring_emit_wreg,
+	.emit_reg_wait = gfx_v9_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = gfx_v9_0_ring_emit_reg_write_reg_wait,
+	.soft_recovery = gfx_v9_0_ring_soft_recovery,
+	.emit_mem_sync = gfx_v9_0_emit_mem_sync,
+	.emit_wave_limit = gfx_v9_0_emit_wave_limit,
+	.reset = gfx_v9_0_reset_kcq,
+	.emit_cleaner_shader = gfx_v9_0_ring_emit_cleaner_shader,
+	.begin_use = gfx_v9_0_ring_begin_use_compute,
+	.end_use = gfx_v9_0_ring_end_use_compute,
 };
 
 static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_kiq = {
@@ -4201,46 +7566,46 @@ static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_kiq = {
 	.align_mask = 0xff,
 	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
 	.support_64bit_ptrs = true,
-	.vmhub = AMDGPU_GFXHUB,
 	.get_rptr = gfx_v9_0_ring_get_rptr_compute,
 	.get_wptr = gfx_v9_0_ring_get_wptr_compute,
 	.set_wptr = gfx_v9_0_ring_set_wptr_compute,
 	.emit_frame_size =
 		20 + /* gfx_v9_0_ring_emit_gds_switch */
 		7 + /* gfx_v9_0_ring_emit_hdp_flush */
-		5 + /* gfx_v9_0_ring_emit_hdp_invalidate */
+		5 + /* hdp invalidate */
 		7 + /* gfx_v9_0_ring_emit_pipeline_sync */
-		24 + /* gfx_v9_0_ring_emit_vm_flush */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 5 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 +
 		8 + 8 + 8, /* gfx_v9_0_ring_emit_fence_kiq x3 for user fence, vm fence */
-	.emit_ib_size =	4, /* gfx_v9_0_ring_emit_ib_compute */
-	.emit_ib = gfx_v9_0_ring_emit_ib_compute,
+	.emit_ib_size =	7, /* gfx_v9_0_ring_emit_ib_compute */
 	.emit_fence = gfx_v9_0_ring_emit_fence_kiq,
 	.test_ring = gfx_v9_0_ring_test_ring,
-	.test_ib = gfx_v9_0_ring_test_ib,
 	.insert_nop = amdgpu_ring_insert_nop,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
 	.emit_rreg = gfx_v9_0_ring_emit_rreg,
 	.emit_wreg = gfx_v9_0_ring_emit_wreg,
+	.emit_reg_wait = gfx_v9_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = gfx_v9_0_ring_emit_reg_write_reg_wait,
 };
 
 static void gfx_v9_0_set_ring_funcs(struct amdgpu_device *adev)
 {
 	int i;
 
-	adev->gfx.kiq.ring.funcs = &gfx_v9_0_ring_funcs_kiq;
+	adev->gfx.kiq[0].ring.funcs = &gfx_v9_0_ring_funcs_kiq;
 
 	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
 		adev->gfx.gfx_ring[i].funcs = &gfx_v9_0_ring_funcs_gfx;
 
+	if (adev->gfx.mcbp && adev->gfx.num_gfx_rings) {
+		for (i = 0; i < GFX9_NUM_SW_GFX_RINGS; i++)
+			adev->gfx.sw_gfx_ring[i].funcs = &gfx_v9_0_sw_ring_funcs_gfx;
+	}
+
 	for (i = 0; i < adev->gfx.num_compute_rings; i++)
 		adev->gfx.compute_ring[i].funcs = &gfx_v9_0_ring_funcs_compute;
 }
 
-static const struct amdgpu_irq_src_funcs gfx_v9_0_kiq_irq_funcs = {
-	.set = gfx_v9_0_kiq_set_interrupt_state,
-	.process = gfx_v9_0_kiq_irq,
-};
-
 static const struct amdgpu_irq_src_funcs gfx_v9_0_eop_irq_funcs = {
 	.set = gfx_v9_0_set_eop_interrupt_state,
 	.process = gfx_v9_0_eop_irq,
@@ -4251,11 +7616,22 @@ static const struct amdgpu_irq_src_funcs gfx_v9_0_priv_reg_irq_funcs = {
 	.process = gfx_v9_0_priv_reg_irq,
 };
 
+static const struct amdgpu_irq_src_funcs gfx_v9_0_bad_op_irq_funcs = {
+	.set = gfx_v9_0_set_bad_op_fault_state,
+	.process = gfx_v9_0_bad_op_irq,
+};
+
 static const struct amdgpu_irq_src_funcs gfx_v9_0_priv_inst_irq_funcs = {
 	.set = gfx_v9_0_set_priv_inst_fault_state,
 	.process = gfx_v9_0_priv_inst_irq,
 };
 
+static const struct amdgpu_irq_src_funcs gfx_v9_0_cp_ecc_error_irq_funcs = {
+	.set = gfx_v9_0_set_cp_ecc_error_state,
+	.process = amdgpu_gfx_cp_ecc_error_irq,
+};
+
+
 static void gfx_v9_0_set_irq_funcs(struct amdgpu_device *adev)
 {
 	adev->gfx.eop_irq.num_types = AMDGPU_CP_IRQ_LAST;
@@ -4264,18 +7640,27 @@ static void gfx_v9_0_set_irq_funcs(struct amdgpu_device *adev)
 	adev->gfx.priv_reg_irq.num_types = 1;
 	adev->gfx.priv_reg_irq.funcs = &gfx_v9_0_priv_reg_irq_funcs;
 
+	adev->gfx.bad_op_irq.num_types = 1;
+	adev->gfx.bad_op_irq.funcs = &gfx_v9_0_bad_op_irq_funcs;
+
 	adev->gfx.priv_inst_irq.num_types = 1;
 	adev->gfx.priv_inst_irq.funcs = &gfx_v9_0_priv_inst_irq_funcs;
 
-	adev->gfx.kiq.irq.num_types = AMDGPU_CP_KIQ_IRQ_LAST;
-	adev->gfx.kiq.irq.funcs = &gfx_v9_0_kiq_irq_funcs;
+	adev->gfx.cp_ecc_error_irq.num_types = 2; /*C5 ECC error and C9 FUE error*/
+	adev->gfx.cp_ecc_error_irq.funcs = &gfx_v9_0_cp_ecc_error_irq_funcs;
 }
 
 static void gfx_v9_0_set_rlc_funcs(struct amdgpu_device *adev)
 {
-	switch (adev->asic_type) {
-	case CHIP_VEGA10:
-	case CHIP_RAVEN:
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 0, 1):
+	case IP_VERSION(9, 2, 1):
+	case IP_VERSION(9, 4, 0):
+	case IP_VERSION(9, 2, 2):
+	case IP_VERSION(9, 1, 0):
+	case IP_VERSION(9, 4, 1):
+	case IP_VERSION(9, 3, 0):
+	case IP_VERSION(9, 4, 2):
 		adev->gfx.rlc.funcs = &gfx_v9_0_rlc_funcs;
 		break;
 	default:
@@ -4286,29 +7671,58 @@ static void gfx_v9_0_set_rlc_funcs(struct amdgpu_device *adev)
 static void gfx_v9_0_set_gds_init(struct amdgpu_device *adev)
 {
 	/* init asci gds info */
-	adev->gds.mem.total_size = RREG32_SOC15(GC, 0, mmGDS_VMID0_SIZE);
-	adev->gds.gws.total_size = 64;
-	adev->gds.oa.total_size = 16;
-
-	if (adev->gds.mem.total_size == 64 * 1024) {
-		adev->gds.mem.gfx_partition_size = 4096;
-		adev->gds.mem.cs_partition_size = 4096;
-
-		adev->gds.gws.gfx_partition_size = 4;
-		adev->gds.gws.cs_partition_size = 4;
-
-		adev->gds.oa.gfx_partition_size = 4;
-		adev->gds.oa.cs_partition_size = 1;
-	} else {
-		adev->gds.mem.gfx_partition_size = 1024;
-		adev->gds.mem.cs_partition_size = 1024;
-
-		adev->gds.gws.gfx_partition_size = 16;
-		adev->gds.gws.cs_partition_size = 16;
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 0, 1):
+	case IP_VERSION(9, 2, 1):
+	case IP_VERSION(9, 4, 0):
+		adev->gds.gds_size = 0x10000;
+		break;
+	case IP_VERSION(9, 2, 2):
+	case IP_VERSION(9, 1, 0):
+	case IP_VERSION(9, 4, 1):
+		adev->gds.gds_size = 0x1000;
+		break;
+	case IP_VERSION(9, 4, 2):
+		/* aldebaran removed all the GDS internal memory,
+		 * only support GWS opcode in kernel, like barrier
+		 * semaphore.etc */
+		adev->gds.gds_size = 0;
+		break;
+	default:
+		adev->gds.gds_size = 0x10000;
+		break;
+	}
 
-		adev->gds.oa.gfx_partition_size = 4;
-		adev->gds.oa.cs_partition_size = 4;
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 0, 1):
+	case IP_VERSION(9, 4, 0):
+		adev->gds.gds_compute_max_wave_id = 0x7ff;
+		break;
+	case IP_VERSION(9, 2, 1):
+		adev->gds.gds_compute_max_wave_id = 0x27f;
+		break;
+	case IP_VERSION(9, 2, 2):
+	case IP_VERSION(9, 1, 0):
+		if (adev->apu_flags & AMD_APU_IS_RAVEN2)
+			adev->gds.gds_compute_max_wave_id = 0x77; /* raven2 */
+		else
+			adev->gds.gds_compute_max_wave_id = 0x15f; /* raven1 */
+		break;
+	case IP_VERSION(9, 4, 1):
+		adev->gds.gds_compute_max_wave_id = 0xfff;
+		break;
+	case IP_VERSION(9, 4, 2):
+		/* deprecated for Aldebaran, no usage at all */
+		adev->gds.gds_compute_max_wave_id = 0;
+		break;
+	default:
+		/* this really depends on the chip */
+		adev->gds.gds_compute_max_wave_id = 0x7ff;
+		break;
 	}
+
+	adev->gds.gws_size = 64;
+	adev->gds.oa_size = 16;
 }
 
 static void gfx_v9_0_set_user_cu_inactive_bitmap(struct amdgpu_device *adev,
@@ -4345,12 +7759,21 @@ static int gfx_v9_0_get_cu_info(struct amdgpu_device *adev,
 {
 	int i, j, k, counter, active_cu_number = 0;
 	u32 mask, bitmap, ao_bitmap, ao_cu_mask = 0;
-	unsigned disable_masks[4 * 2];
+	unsigned disable_masks[4 * 4];
 
 	if (!adev || !cu_info)
 		return -EINVAL;
 
-	amdgpu_gfx_parse_disable_cu(disable_masks, 4, 2);
+	/*
+	 * 16 comes from bitmap array size 4*4, and it can cover all gfx9 ASICs
+	 */
+	if (adev->gfx.config.max_shader_engines *
+		adev->gfx.config.max_sh_per_se > 16)
+		return -EINVAL;
+
+	amdgpu_gfx_parse_disable_cu(disable_masks,
+				    adev->gfx.config.max_shader_engines,
+				    adev->gfx.config.max_sh_per_se);
 
 	mutex_lock(&adev->grbm_idx_mutex);
 	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
@@ -4358,12 +7781,24 @@ static int gfx_v9_0_get_cu_info(struct amdgpu_device *adev,
 			mask = 1;
 			ao_bitmap = 0;
 			counter = 0;
-			gfx_v9_0_select_se_sh(adev, i, j, 0xffffffff);
-			if (i < 4 && j < 2)
-				gfx_v9_0_set_user_cu_inactive_bitmap(
-					adev, disable_masks[i * 2 + j]);
+			amdgpu_gfx_select_se_sh(adev, i, j, 0xffffffff, 0);
+			gfx_v9_0_set_user_cu_inactive_bitmap(
+				adev, disable_masks[i * adev->gfx.config.max_sh_per_se + j]);
 			bitmap = gfx_v9_0_get_cu_active_bitmap(adev);
-			cu_info->bitmap[i][j] = bitmap;
+
+			/*
+			 * The bitmap(and ao_cu_bitmap) in cu_info structure is
+			 * 4x4 size array, and it's usually suitable for Vega
+			 * ASICs which has 4*2 SE/SH layout.
+			 * But for Arcturus, SE/SH layout is changed to 8*1.
+			 * To mostly reduce the impact, we make it compatible
+			 * with current bitmap array as below:
+			 *    SE4,SH0 --> bitmap[0][1]
+			 *    SE5,SH0 --> bitmap[1][1]
+			 *    SE6,SH0 --> bitmap[2][1]
+			 *    SE7,SH0 --> bitmap[3][1]
+			 */
+			cu_info->bitmap[0][i % 4][j + i / 4] = bitmap;
 
 			for (k = 0; k < adev->gfx.config.max_cu_per_sh; k ++) {
 				if (bitmap & mask) {
@@ -4376,14 +7811,15 @@ static int gfx_v9_0_get_cu_info(struct amdgpu_device *adev,
 			active_cu_number += counter;
 			if (i < 2 && j < 2)
 				ao_cu_mask |= (ao_bitmap << (i * 16 + j * 8));
-			cu_info->ao_cu_bitmap[i][j] = ao_bitmap;
+			cu_info->ao_cu_bitmap[i % 4][j + i / 4] = ao_bitmap;
 		}
 	}
-	gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 	mutex_unlock(&adev->grbm_idx_mutex);
 
 	cu_info->number = active_cu_number;
 	cu_info->ao_cu_mask = ao_cu_mask;
+	cu_info->simd_per_cu = NUM_SIMD_PER_CU;
 
 	return 0;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.h b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.h
index fa5a3fbaf6ab..f9f6edc5e558 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.h
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.h
@@ -26,9 +26,7 @@
 
 extern const struct amdgpu_ip_block_version gfx_v9_0_ip_block;
 
-void gfx_v9_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num);
-
-uint64_t gfx_v9_0_get_gpu_clock_counter(struct amdgpu_device *adev);
-int gfx_v9_0_get_cu_info(struct amdgpu_device *adev, struct amdgpu_cu_info *cu_info);
+void gfx_v9_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num,
+			   u32 instance, int xcc_id);
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0_cleaner_shader.h b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0_cleaner_shader.h
new file mode 100644
index 000000000000..0b6bd09b7529
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0_cleaner_shader.h
@@ -0,0 +1,68 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+/* Define the cleaner shader gfx_9_0 */
+static const u32 __maybe_unused gfx_9_0_cleaner_shader_hex[] = {
+	/* Add the cleaner shader code here */
+};
+
+/* Define the cleaner shader gfx_9_4_2 */
+static const u32 gfx_9_4_2_cleaner_shader_hex[] = {
+	0xbf068100, 0xbf84003b,
+	0xbf8a0000, 0xb07c0000,
+	0xbe8200ff, 0x00000078,
+	0xbf110802, 0x7e000280,
+	0x7e020280, 0x7e040280,
+	0x7e060280, 0x7e080280,
+	0x7e0a0280, 0x7e0c0280,
+	0x7e0e0280, 0x80828802,
+	0xbe803202, 0xbf84fff5,
+	0xbf9c0000, 0xbe8200ff,
+	0x80000000, 0x86020102,
+	0xbf840011, 0xbefe00c1,
+	0xbeff00c1, 0xd28c0001,
+	0x0001007f, 0xd28d0001,
+	0x0002027e, 0x10020288,
+	0xbe8200bf, 0xbefc00c1,
+	0xd89c2000, 0x00020201,
+	0xd89c6040, 0x00040401,
+	0x320202ff, 0x00000400,
+	0x80828102, 0xbf84fff8,
+	0xbefc00ff, 0x0000005c,
+	0xbf800000, 0xbe802c80,
+	0xbe812c80, 0xbe822c80,
+	0xbe832c80, 0x80fc847c,
+	0xbf84fffa, 0xbee60080,
+	0xbee70080, 0xbeea0180,
+	0xbeec0180, 0xbeee0180,
+	0xbef00180, 0xbef20180,
+	0xbef40180, 0xbef60180,
+	0xbef80180, 0xbefa0180,
+	0xbf810000, 0xbf8d0001,
+	0xbefc00ff, 0x0000005c,
+	0xbf800000, 0xbe802c80,
+	0xbe812c80, 0xbe822c80,
+	0xbe832c80, 0x80fc847c,
+	0xbf84fffa, 0xbee60080,
+	0xbee70080, 0xbeea01ff,
+	0x000000ee, 0xbf810000,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c
new file mode 100644
index 000000000000..6028afd81690
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c
@@ -0,0 +1,1019 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/kernel.h>
+
+#include "amdgpu.h"
+#include "amdgpu_gfx.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "amdgpu_pm.h"
+
+#include "gc/gc_9_4_1_offset.h"
+#include "gc/gc_9_4_1_sh_mask.h"
+#include "soc15_common.h"
+
+#include "gfx_v9_4.h"
+#include "amdgpu_ras.h"
+
+static const struct soc15_reg_entry gfx_v9_4_edc_counter_regs[] = {
+	/* CPC */
+	{ SOC15_REG_ENTRY(GC, 0, mmCPC_EDC_SCRATCH_CNT), 0, 1, 1 },
+	{ SOC15_REG_ENTRY(GC, 0, mmCPC_EDC_UCODE_CNT), 0, 1, 1 },
+	/* DC */
+	{ SOC15_REG_ENTRY(GC, 0, mmDC_EDC_STATE_CNT), 0, 1, 1 },
+	{ SOC15_REG_ENTRY(GC, 0, mmDC_EDC_CSINVOC_CNT), 0, 1, 1 },
+	{ SOC15_REG_ENTRY(GC, 0, mmDC_EDC_RESTORE_CNT), 0, 1, 1 },
+	/* CPF */
+	{ SOC15_REG_ENTRY(GC, 0, mmCPF_EDC_ROQ_CNT), 0, 1, 1 },
+	{ SOC15_REG_ENTRY(GC, 0, mmCPF_EDC_TAG_CNT), 0, 1, 1 },
+	/* GDS */
+	{ SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_CNT), 0, 1, 1 },
+	{ SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_GRBM_CNT), 0, 1, 1 },
+	{ SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_DED), 0, 1, 1 },
+	{ SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PHY_CNT), 0, 1, 1 },
+	{ SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PIPE_CNT), 0, 1, 1 },
+	/* SPI */
+	{ SOC15_REG_ENTRY(GC, 0, mmSPI_EDC_CNT), 0, 4, 1 },
+	/* SQ */
+	{ SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT), 0, 8, 16 },
+	{ SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_DED_CNT), 0, 8, 16 },
+	{ SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_INFO), 0, 8, 16 },
+	{ SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_SEC_CNT), 0, 8, 16 },
+	/* SQC */
+	{ SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT), 0, 4, 6 },
+	{ SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2), 0, 4, 6 },
+	{ SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3), 0, 4, 6 },
+	{ SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_PARITY_CNT3), 0, 4, 6 },
+	/* TA */
+	{ SOC15_REG_ENTRY(GC, 0, mmTA_EDC_CNT), 0, 4, 16 },
+	/* TCA */
+	{ SOC15_REG_ENTRY(GC, 0, mmTCA_EDC_CNT), 0, 1, 2 },
+	/* TCC */
+	{ SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT), 0, 1, 16 },
+	{ SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2), 0, 1, 16 },
+	/* TCI */
+	{ SOC15_REG_ENTRY(GC, 0, mmTCI_EDC_CNT), 0, 1, 72 },
+	/* TCP */
+	{ SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW), 0, 4, 16 },
+	{ SOC15_REG_ENTRY(GC, 0, mmTCP_ATC_EDC_GATCL1_CNT), 0, 4, 16 },
+	{ SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT), 0, 4, 16 },
+	/* TD */
+	{ SOC15_REG_ENTRY(GC, 0, mmTD_EDC_CNT), 0, 4, 16 },
+	/* GCEA */
+	{ SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT), 0, 1, 32 },
+	{ SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2), 0, 1, 32 },
+	{ SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3), 0, 1, 32 },
+	/* RLC */
+	{ SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT), 0, 1, 1 },
+	{ SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT2), 0, 1, 1 },
+};
+
+static void gfx_v9_4_select_se_sh(struct amdgpu_device *adev, u32 se_num,
+				  u32 sh_num, u32 instance)
+{
+	u32 data;
+
+	if (instance == 0xffffffff)
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX,
+				     INSTANCE_BROADCAST_WRITES, 1);
+	else
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_INDEX,
+				     instance);
+
+	if (se_num == 0xffffffff)
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_BROADCAST_WRITES,
+				     1);
+	else
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_INDEX, se_num);
+
+	if (sh_num == 0xffffffff)
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_BROADCAST_WRITES,
+				     1);
+	else
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_INDEX, sh_num);
+
+	WREG32_SOC15_RLC_SHADOW(GC, 0, mmGRBM_GFX_INDEX, data);
+}
+
+static const struct soc15_ras_field_entry gfx_v9_4_ras_fields[] = {
+	/* CPC */
+	{ "CPC_SCRATCH", SOC15_REG_ENTRY(GC, 0, mmCPC_EDC_SCRATCH_CNT),
+	  SOC15_REG_FIELD(CPC_EDC_SCRATCH_CNT, SEC_COUNT),
+	  SOC15_REG_FIELD(CPC_EDC_SCRATCH_CNT, DED_COUNT) },
+	{ "CPC_UCODE", SOC15_REG_ENTRY(GC, 0, mmCPC_EDC_UCODE_CNT),
+	  SOC15_REG_FIELD(CPC_EDC_UCODE_CNT, SEC_COUNT),
+	  SOC15_REG_FIELD(CPC_EDC_UCODE_CNT, DED_COUNT) },
+	{ "CPC_DC_STATE_RAM_ME1", SOC15_REG_ENTRY(GC, 0, mmDC_EDC_STATE_CNT),
+	  SOC15_REG_FIELD(DC_EDC_STATE_CNT, SEC_COUNT_ME1),
+	  SOC15_REG_FIELD(DC_EDC_STATE_CNT, DED_COUNT_ME1) },
+	{ "CPC_DC_CSINVOC_RAM_ME1",
+	  SOC15_REG_ENTRY(GC, 0, mmDC_EDC_CSINVOC_CNT),
+	  SOC15_REG_FIELD(DC_EDC_CSINVOC_CNT, SEC_COUNT_ME1),
+	  SOC15_REG_FIELD(DC_EDC_CSINVOC_CNT, DED_COUNT_ME1) },
+	{ "CPC_DC_RESTORE_RAM_ME1",
+	  SOC15_REG_ENTRY(GC, 0, mmDC_EDC_RESTORE_CNT),
+	  SOC15_REG_FIELD(DC_EDC_RESTORE_CNT, SEC_COUNT_ME1),
+	  SOC15_REG_FIELD(DC_EDC_RESTORE_CNT, DED_COUNT_ME1) },
+	{ "CPC_DC_CSINVOC_RAM1_ME1",
+	  SOC15_REG_ENTRY(GC, 0, mmDC_EDC_CSINVOC_CNT),
+	  SOC15_REG_FIELD(DC_EDC_CSINVOC_CNT, SEC_COUNT1_ME1),
+	  SOC15_REG_FIELD(DC_EDC_CSINVOC_CNT, DED_COUNT1_ME1) },
+	{ "CPC_DC_RESTORE_RAM1_ME1",
+	  SOC15_REG_ENTRY(GC, 0, mmDC_EDC_RESTORE_CNT),
+	  SOC15_REG_FIELD(DC_EDC_RESTORE_CNT, SEC_COUNT1_ME1),
+	  SOC15_REG_FIELD(DC_EDC_RESTORE_CNT, DED_COUNT1_ME1) },
+
+	/* CPF */
+	{ "CPF_ROQ_ME2", SOC15_REG_ENTRY(GC, 0, mmCPF_EDC_ROQ_CNT),
+	  SOC15_REG_FIELD(CPF_EDC_ROQ_CNT, SEC_COUNT_ME2),
+	  SOC15_REG_FIELD(CPF_EDC_ROQ_CNT, DED_COUNT_ME2) },
+	{ "CPF_ROQ_ME1", SOC15_REG_ENTRY(GC, 0, mmCPF_EDC_ROQ_CNT),
+	  SOC15_REG_FIELD(CPF_EDC_ROQ_CNT, SEC_COUNT_ME1),
+	  SOC15_REG_FIELD(CPF_EDC_ROQ_CNT, DED_COUNT_ME1) },
+	{ "CPF_TCIU_TAG", SOC15_REG_ENTRY(GC, 0, mmCPF_EDC_TAG_CNT),
+	  SOC15_REG_FIELD(CPF_EDC_TAG_CNT, SEC_COUNT),
+	  SOC15_REG_FIELD(CPF_EDC_TAG_CNT, DED_COUNT) },
+
+	/* GDS */
+	{ "GDS_GRBM", SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_GRBM_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_GRBM_CNT, SEC),
+	  SOC15_REG_FIELD(GDS_EDC_GRBM_CNT, DED) },
+	{ "GDS_MEM", SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_CNT, GDS_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_CNT, GDS_MEM_DED) },
+	{ "GDS_PHY_CMD_RAM_MEM", SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PHY_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, PHY_CMD_RAM_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, PHY_CMD_RAM_MEM_DED) },
+	{ "GDS_PHY_DATA_RAM_MEM", SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PHY_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, PHY_DATA_RAM_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, PHY_DATA_RAM_MEM_DED) },
+	{ "GDS_ME0_CS_PIPE_MEM", SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PHY_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, ME0_CS_PIPE_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, ME0_CS_PIPE_MEM_DED) },
+	{ "GDS_ME1_PIPE0_PIPE_MEM",
+	  SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PIPE_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE0_PIPE_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE0_PIPE_MEM_DED) },
+	{ "GDS_ME1_PIPE1_PIPE_MEM",
+	  SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PIPE_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE1_PIPE_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE1_PIPE_MEM_DED) },
+	{ "GDS_ME1_PIPE2_PIPE_MEM",
+	  SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PIPE_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE2_PIPE_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE2_PIPE_MEM_DED) },
+	{ "GDS_ME1_PIPE3_PIPE_MEM",
+	  SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PIPE_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE3_PIPE_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE3_PIPE_MEM_DED) },
+
+	/* SPI */
+	{ "SPI_SR_MEM", SOC15_REG_ENTRY(GC, 0, mmSPI_EDC_CNT),
+	  SOC15_REG_FIELD(SPI_EDC_CNT, SPI_SR_MEM_SEC_COUNT),
+	  SOC15_REG_FIELD(SPI_EDC_CNT, SPI_SR_MEM_DED_COUNT) },
+	{ "SPI_GDS_EXPREQ", SOC15_REG_ENTRY(GC, 0, mmSPI_EDC_CNT),
+	  SOC15_REG_FIELD(SPI_EDC_CNT, SPI_GDS_EXPREQ_SEC_COUNT),
+	  SOC15_REG_FIELD(SPI_EDC_CNT, SPI_GDS_EXPREQ_DED_COUNT) },
+	{ "SPI_WB_GRANT_30", SOC15_REG_ENTRY(GC, 0, mmSPI_EDC_CNT),
+	  SOC15_REG_FIELD(SPI_EDC_CNT, SPI_WB_GRANT_30_SEC_COUNT),
+	  SOC15_REG_FIELD(SPI_EDC_CNT, SPI_WB_GRANT_30_DED_COUNT) },
+	{ "SPI_WB_GRANT_61", SOC15_REG_ENTRY(GC, 0, mmSPI_EDC_CNT),
+	  SOC15_REG_FIELD(SPI_EDC_CNT, SPI_WB_GRANT_61_SEC_COUNT),
+	  SOC15_REG_FIELD(SPI_EDC_CNT, SPI_WB_GRANT_61_DED_COUNT) },
+	{ "SPI_LIFE_CNT", SOC15_REG_ENTRY(GC, 0, mmSPI_EDC_CNT),
+	  SOC15_REG_FIELD(SPI_EDC_CNT, SPI_LIFE_CNT_SEC_COUNT),
+	  SOC15_REG_FIELD(SPI_EDC_CNT, SPI_LIFE_CNT_DED_COUNT) },
+
+	/* SQ */
+	{ "SQ_SGPR", SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, SGPR_SEC_COUNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, SGPR_DED_COUNT) },
+	{ "SQ_LDS_D", SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, LDS_D_SEC_COUNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, LDS_D_DED_COUNT) },
+	{ "SQ_LDS_I", SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, LDS_I_SEC_COUNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, LDS_I_DED_COUNT) },
+	{ "SQ_VGPR0", SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR0_SEC_COUNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR0_DED_COUNT) },
+	{ "SQ_VGPR1", SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR1_SEC_COUNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR1_DED_COUNT) },
+	{ "SQ_VGPR2", SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR2_SEC_COUNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR2_DED_COUNT) },
+	{ "SQ_VGPR3", SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR3_SEC_COUNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR3_DED_COUNT) },
+
+	/* SQC */
+	{ "SQC_INST_UTCL1_LFIFO", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, INST_UTCL1_LFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, INST_UTCL1_LFIFO_DED_COUNT) },
+	{ "SQC_DATA_CU0_WRITE_DATA_BUF", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU0_WRITE_DATA_BUF_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU0_WRITE_DATA_BUF_DED_COUNT) },
+	{ "SQC_DATA_CU0_UTCL1_LFIFO", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU0_UTCL1_LFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU0_UTCL1_LFIFO_DED_COUNT) },
+	{ "SQC_DATA_CU1_WRITE_DATA_BUF", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU1_WRITE_DATA_BUF_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU1_WRITE_DATA_BUF_DED_COUNT) },
+	{ "SQC_DATA_CU1_UTCL1_LFIFO", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU1_UTCL1_LFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU1_UTCL1_LFIFO_DED_COUNT) },
+	{ "SQC_DATA_CU2_WRITE_DATA_BUF", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU2_WRITE_DATA_BUF_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU2_WRITE_DATA_BUF_DED_COUNT) },
+	{ "SQC_DATA_CU2_UTCL1_LFIFO", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU2_UTCL1_LFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU2_UTCL1_LFIFO_DED_COUNT) },
+	{ "SQC_INST_BANKA_TAG_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, INST_BANKA_TAG_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, INST_BANKA_TAG_RAM_DED_COUNT) },
+	{ "SQC_INST_BANKA_UTCL1_MISS_FIFO",
+	  SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_PARITY_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3,
+			  INST_BANKA_UTCL1_MISS_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3,
+			  INST_BANKA_UTCL1_MISS_FIFO_DED_COUNT) },
+	{ "SQC_INST_BANKA_MISS_FIFO",
+	  SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_PARITY_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, INST_BANKA_MISS_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3,
+			  INST_BANKA_MISS_FIFO_DED_COUNT) },
+	{ "SQC_INST_BANKA_BANK_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, INST_BANKA_BANK_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, INST_BANKA_BANK_RAM_DED_COUNT) },
+	{ "SQC_DATA_BANKA_TAG_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_TAG_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_TAG_RAM_DED_COUNT) },
+	{ "SQC_DATA_BANKA_HIT_FIFO",
+	  SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_PARITY_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKA_HIT_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKA_HIT_FIFO_DED_COUNT) },
+	{ "SQC_DATA_BANKA_MISS_FIFO",
+	  SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_PARITY_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKA_MISS_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3,
+			  DATA_BANKA_MISS_FIFO_DED_COUNT) },
+	{ "SQC_DATA_BANKA_BANK_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_BANK_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_BANK_RAM_DED_COUNT) },
+	{ "SQC_INST_BANKB_TAG_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, INST_BANKB_TAG_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, INST_BANKB_TAG_RAM_DED_COUNT) },
+	{ "SQC_INST_BANKB_UTCL1_MISS_FIFO",
+	  SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_PARITY_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3,
+			  INST_BANKB_UTCL1_MISS_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3,
+			  INST_BANKB_UTCL1_MISS_FIFO_DED_COUNT) },
+	{ "SQC_INST_BANKB_MISS_FIFO",
+	  SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_PARITY_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, INST_BANKB_MISS_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3,
+			  INST_BANKB_MISS_FIFO_DED_COUNT) },
+	{ "SQC_INST_BANKB_BANK_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, INST_BANKB_BANK_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, INST_BANKB_BANK_RAM_DED_COUNT) },
+	{ "SQC_DATA_BANKB_TAG_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_TAG_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_TAG_RAM_DED_COUNT) },
+	{ "SQC_DATA_BANKB_HIT_FIFO",
+	  SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_PARITY_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKB_HIT_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKB_HIT_FIFO_DED_COUNT) },
+	{ "SQC_DATA_BANKB_MISS_FIFO",
+	  SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_PARITY_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKB_MISS_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3,
+			  DATA_BANKB_MISS_FIFO_DED_COUNT) },
+	{ "SQC_DATA_BANKB_BANK_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_BANK_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_BANK_RAM_DED_COUNT) },
+
+	/* TA */
+	{ "TA_FS_DFIFO", SOC15_REG_ENTRY(GC, 0, mmTA_EDC_CNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_DFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_DFIFO_DED_COUNT) },
+	{ "TA_FS_AFIFO", SOC15_REG_ENTRY(GC, 0, mmTA_EDC_CNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_AFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_AFIFO_DED_COUNT) },
+	{ "TA_FL_LFIFO", SOC15_REG_ENTRY(GC, 0, mmTA_EDC_CNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FL_LFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FL_LFIFO_DED_COUNT) },
+	{ "TA_FX_LFIFO", SOC15_REG_ENTRY(GC, 0, mmTA_EDC_CNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FX_LFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FX_LFIFO_DED_COUNT) },
+	{ "TA_FS_CFIFO", SOC15_REG_ENTRY(GC, 0, mmTA_EDC_CNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_CFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_CFIFO_DED_COUNT) },
+
+	/* TCA */
+	{ "TCA_HOLE_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCA_EDC_CNT),
+	  SOC15_REG_FIELD(TCA_EDC_CNT, HOLE_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TCA_EDC_CNT, HOLE_FIFO_DED_COUNT) },
+	{ "TCA_REQ_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCA_EDC_CNT),
+	  SOC15_REG_FIELD(TCA_EDC_CNT, REQ_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TCA_EDC_CNT, REQ_FIFO_DED_COUNT) },
+
+	/* TCC */
+	{ "TCC_CACHE_DATA", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, CACHE_DATA_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, CACHE_DATA_DED_COUNT) },
+	{ "TCC_CACHE_DIRTY", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, CACHE_DIRTY_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, CACHE_DIRTY_DED_COUNT) },
+	{ "TCC_HIGH_RATE_TAG", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, HIGH_RATE_TAG_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, HIGH_RATE_TAG_DED_COUNT) },
+	{ "TCC_LOW_RATE_TAG", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, LOW_RATE_TAG_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, LOW_RATE_TAG_DED_COUNT) },
+	{ "TCC_IN_USE_DEC", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, IN_USE_DEC_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, IN_USE_DEC_DED_COUNT) },
+	{ "TCC_IN_USE_TRANSFER", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, IN_USE_TRANSFER_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, IN_USE_TRANSFER_DED_COUNT) },
+	{ "TCC_RETURN_DATA", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, RETURN_DATA_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, RETURN_DATA_DED_COUNT) },
+	{ "TCC_RETURN_CONTROL", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, RETURN_CONTROL_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, RETURN_CONTROL_DED_COUNT) },
+	{ "TCC_UC_ATOMIC_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, UC_ATOMIC_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, UC_ATOMIC_FIFO_DED_COUNT) },
+	{ "TCC_WRITE_RETURN", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, WRITE_RETURN_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, WRITE_RETURN_DED_COUNT) },
+	{ "TCC_WRITE_CACHE_READ", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, WRITE_CACHE_READ_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, WRITE_CACHE_READ_DED_COUNT) },
+	{ "TCC_SRC_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, SRC_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, SRC_FIFO_DED_COUNT) },
+	{ "TCC_CACHE_TAG_PROBE_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, CACHE_TAG_PROBE_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, CACHE_TAG_PROBE_FIFO_DED_COUNT) },
+	{ "TCC_LATENCY_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, LATENCY_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, LATENCY_FIFO_DED_COUNT) },
+	{ "TCC_LATENCY_FIFO_NEXT_RAM", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, LATENCY_FIFO_NEXT_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, LATENCY_FIFO_NEXT_RAM_DED_COUNT) },
+
+	/* TCI */
+	{ "TCI_WRITE_RAM", SOC15_REG_ENTRY(GC, 0, mmTCI_EDC_CNT),
+	  SOC15_REG_FIELD(TCI_EDC_CNT, WRITE_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(TCI_EDC_CNT, WRITE_RAM_DED_COUNT) },
+
+	/* TCP */
+	{ "TCP_CACHE_RAM", SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, CACHE_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, CACHE_RAM_DED_COUNT) },
+	{ "TCP_LFIFO_RAM", SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, LFIFO_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, LFIFO_RAM_DED_COUNT) },
+	{ "TCP_CMD_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, CMD_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, CMD_FIFO_DED_COUNT) },
+	{ "TCP_VM_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, VM_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, VM_FIFO_DED_COUNT) },
+	{ "TCP_DB_RAM", SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, DB_RAM_SED_COUNT), 0, 0 },
+	{ "TCP_UTCL1_LFIFO0", SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, UTCL1_LFIFO0_SEC_COUNT),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, UTCL1_LFIFO0_DED_COUNT) },
+	{ "TCP_UTCL1_LFIFO1", SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, UTCL1_LFIFO1_SEC_COUNT),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, UTCL1_LFIFO1_DED_COUNT) },
+
+	/* TD */
+	{ "TD_SS_FIFO_LO", SOC15_REG_ENTRY(GC, 0, mmTD_EDC_CNT),
+	  SOC15_REG_FIELD(TD_EDC_CNT, SS_FIFO_LO_SEC_COUNT),
+	  SOC15_REG_FIELD(TD_EDC_CNT, SS_FIFO_LO_DED_COUNT) },
+	{ "TD_SS_FIFO_HI", SOC15_REG_ENTRY(GC, 0, mmTD_EDC_CNT),
+	  SOC15_REG_FIELD(TD_EDC_CNT, SS_FIFO_HI_SEC_COUNT),
+	  SOC15_REG_FIELD(TD_EDC_CNT, SS_FIFO_HI_DED_COUNT) },
+	{ "TD_CS_FIFO", SOC15_REG_ENTRY(GC, 0, mmTD_EDC_CNT),
+	  SOC15_REG_FIELD(TD_EDC_CNT, CS_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TD_EDC_CNT, CS_FIFO_DED_COUNT) },
+
+	/* EA */
+	{ "EA_DRAMRD_CMDMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT) },
+	{ "EA_DRAMWR_CMDMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT) },
+	{ "EA_DRAMWR_DATAMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT) },
+	{ "EA_RRET_TAGMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, RRET_TAGMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, RRET_TAGMEM_DED_COUNT) },
+	{ "EA_WRET_TAGMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, WRET_TAGMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, WRET_TAGMEM_DED_COUNT) },
+	{ "EA_GMIRD_CMDMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT) },
+	{ "EA_GMIWR_CMDMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT) },
+	{ "EA_GMIWR_DATAMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT) },
+	{ "EA_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT), 0, 0 },
+	{ "EA_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3), 0, 0,
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT) },
+	{ "EA_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT), 0, 0 },
+	{ "EA_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3), 0, 0,
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT) },
+	{ "EA_IORD_CMDMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, IORD_CMDMEM_SED_COUNT), 0, 0 },
+	{ "EA_IORD_CMDMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3), 0, 0,
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, IORD_CMDMEM_DED_COUNT) },
+	{ "EA_IOWR_CMDMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, IOWR_CMDMEM_SED_COUNT), 0, 0 },
+	{ "EA_IOWR_CMDMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3), 0, 0,
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, IOWR_CMDMEM_DED_COUNT) },
+	{ "EA_IOWR_DATAMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, IOWR_DATAMEM_SED_COUNT), 0, 0 },
+	{ "EA_IOWR_DATAMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3), 0, 0,
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, IOWR_DATAMEM_DED_COUNT) },
+	{ "EA_GMIRD_PAGEMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT), 0, 0 },
+	{ "EA_GMIRD_PAGEMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3), 0, 0,
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT) },
+	{ "EA_GMIWR_PAGEMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT), 0, 0 },
+	{ "EA_GMIWR_PAGEMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3), 0, 0,
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT) },
+	{ "EA_MAM_D0MEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D0MEM_SED_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D0MEM_DED_COUNT) },
+	{ "EA_MAM_D1MEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D1MEM_SED_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D1MEM_DED_COUNT) },
+	{ "EA_MAM_D2MEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D2MEM_SED_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D2MEM_DED_COUNT) },
+	{ "EA_MAM_D3MEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D3MEM_SED_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D3MEM_DED_COUNT) },
+	{ "EA_MAM_A0MEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A0MEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A0MEM_DED_COUNT) },
+	{ "EA_MAM_A1MEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A1MEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A1MEM_DED_COUNT) },
+	{ "EA_MAM_A2MEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A2MEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A2MEM_DED_COUNT) },
+	{ "EA_MAM_A3MEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A3MEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A3MEM_DED_COUNT) },
+	{ "EA_MAM_AFMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, MAM_AFMEM_SEC_COUNT), 0, 0 },
+	{ "EA_MAM_AFMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3), 0, 0,
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_AFMEM_DED_COUNT) },
+
+	/* RLC */
+	{ "RLCG_INSTR_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLCG_INSTR_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLCG_INSTR_RAM_DED_COUNT) },
+	{ "RLCG_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLCG_SCRATCH_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLCG_SCRATCH_RAM_DED_COUNT) },
+	{ "RLCV_INSTR_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLCV_INSTR_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLCV_INSTR_RAM_DED_COUNT) },
+	{ "RLCV_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLCV_SCRATCH_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLCV_SCRATCH_RAM_DED_COUNT) },
+	{ "RLC_TCTAG_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLC_TCTAG_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLC_TCTAG_RAM_DED_COUNT) },
+	{ "RLC_SPM_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SPM_SCRATCH_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SPM_SCRATCH_RAM_DED_COUNT) },
+	{ "RLC_SRM_DATA_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SRM_DATA_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SRM_DATA_RAM_DED_COUNT) },
+	{ "RLC_SRM_ADDR_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SRM_ADDR_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SRM_ADDR_RAM_DED_COUNT) },
+	{ "RLC_SPM_SE0_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT2),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE0_SCRATCH_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE0_SCRATCH_RAM_DED_COUNT) },
+	{ "RLC_SPM_SE1_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT2),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE1_SCRATCH_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE1_SCRATCH_RAM_DED_COUNT) },
+	{ "RLC_SPM_SE2_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT2),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE2_SCRATCH_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE2_SCRATCH_RAM_DED_COUNT) },
+	{ "RLC_SPM_SE3_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT2),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE3_SCRATCH_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE3_SCRATCH_RAM_DED_COUNT) },
+	{ "RLC_SPM_SE4_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT2),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE4_SCRATCH_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE4_SCRATCH_RAM_DED_COUNT) },
+	{ "RLC_SPM_SE5_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT2),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE5_SCRATCH_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE5_SCRATCH_RAM_DED_COUNT) },
+	{ "RLC_SPM_SE6_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT2),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE6_SCRATCH_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE6_SCRATCH_RAM_DED_COUNT) },
+	{ "RLC_SPM_SE7_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT2),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE7_SCRATCH_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE7_SCRATCH_RAM_DED_COUNT) },
+};
+
+static const char * const vml2_mems[] = {
+	"UTC_VML2_BANK_CACHE_0_BIGK_MEM0",
+	"UTC_VML2_BANK_CACHE_0_BIGK_MEM1",
+	"UTC_VML2_BANK_CACHE_0_4K_MEM0",
+	"UTC_VML2_BANK_CACHE_0_4K_MEM1",
+	"UTC_VML2_BANK_CACHE_1_BIGK_MEM0",
+	"UTC_VML2_BANK_CACHE_1_BIGK_MEM1",
+	"UTC_VML2_BANK_CACHE_1_4K_MEM0",
+	"UTC_VML2_BANK_CACHE_1_4K_MEM1",
+	"UTC_VML2_BANK_CACHE_2_BIGK_MEM0",
+	"UTC_VML2_BANK_CACHE_2_BIGK_MEM1",
+	"UTC_VML2_BANK_CACHE_2_4K_MEM0",
+	"UTC_VML2_BANK_CACHE_2_4K_MEM1",
+	"UTC_VML2_BANK_CACHE_3_BIGK_MEM0",
+	"UTC_VML2_BANK_CACHE_3_BIGK_MEM1",
+	"UTC_VML2_BANK_CACHE_3_4K_MEM0",
+	"UTC_VML2_BANK_CACHE_3_4K_MEM1",
+	"UTC_VML2_IFIFO_GROUP0",
+	"UTC_VML2_IFIFO_GROUP1",
+	"UTC_VML2_IFIFO_GROUP2",
+	"UTC_VML2_IFIFO_GROUP3",
+	"UTC_VML2_IFIFO_GROUP4",
+	"UTC_VML2_IFIFO_GROUP5",
+	"UTC_VML2_IFIFO_GROUP6",
+	"UTC_VML2_IFIFO_GROUP7",
+	"UTC_VML2_IFIFO_GROUP8",
+	"UTC_VML2_IFIFO_GROUP9",
+	"UTC_VML2_IFIFO_GROUP10",
+	"UTC_VML2_IFIFO_GROUP11",
+	"UTC_VML2_IFIFO_GROUP12",
+	"UTC_VML2_IFIFO_GROUP13",
+	"UTC_VML2_IFIFO_GROUP14",
+	"UTC_VML2_IFIFO_GROUP15",
+	"UTC_VML2_IFIFO_GROUP16",
+	"UTC_VML2_IFIFO_GROUP17",
+	"UTC_VML2_IFIFO_GROUP18",
+	"UTC_VML2_IFIFO_GROUP19",
+	"UTC_VML2_IFIFO_GROUP20",
+	"UTC_VML2_IFIFO_GROUP21",
+	"UTC_VML2_IFIFO_GROUP22",
+	"UTC_VML2_IFIFO_GROUP23",
+	"UTC_VML2_IFIFO_GROUP24",
+};
+
+static const char * const vml2_walker_mems[] = {
+	"UTC_VML2_CACHE_PDE0_MEM0",
+	"UTC_VML2_CACHE_PDE0_MEM1",
+	"UTC_VML2_CACHE_PDE1_MEM0",
+	"UTC_VML2_CACHE_PDE1_MEM1",
+	"UTC_VML2_CACHE_PDE2_MEM0",
+	"UTC_VML2_CACHE_PDE2_MEM1",
+	"UTC_VML2_RDIF_ARADDRS",
+	"UTC_VML2_RDIF_LOG_FIFO",
+	"UTC_VML2_QUEUE_REQ",
+	"UTC_VML2_QUEUE_RET",
+};
+
+static const char * const utcl2_router_mems[] = {
+	"UTCL2_ROUTER_GROUP0_VML2_REQ_FIFO0",
+	"UTCL2_ROUTER_GROUP1_VML2_REQ_FIFO1",
+	"UTCL2_ROUTER_GROUP2_VML2_REQ_FIFO2",
+	"UTCL2_ROUTER_GROUP3_VML2_REQ_FIFO3",
+	"UTCL2_ROUTER_GROUP4_VML2_REQ_FIFO4",
+	"UTCL2_ROUTER_GROUP5_VML2_REQ_FIFO5",
+	"UTCL2_ROUTER_GROUP6_VML2_REQ_FIFO6",
+	"UTCL2_ROUTER_GROUP7_VML2_REQ_FIFO7",
+	"UTCL2_ROUTER_GROUP8_VML2_REQ_FIFO8",
+	"UTCL2_ROUTER_GROUP9_VML2_REQ_FIFO9",
+	"UTCL2_ROUTER_GROUP10_VML2_REQ_FIFO10",
+	"UTCL2_ROUTER_GROUP11_VML2_REQ_FIFO11",
+	"UTCL2_ROUTER_GROUP12_VML2_REQ_FIFO12",
+	"UTCL2_ROUTER_GROUP13_VML2_REQ_FIFO13",
+	"UTCL2_ROUTER_GROUP14_VML2_REQ_FIFO14",
+	"UTCL2_ROUTER_GROUP15_VML2_REQ_FIFO15",
+	"UTCL2_ROUTER_GROUP16_VML2_REQ_FIFO16",
+	"UTCL2_ROUTER_GROUP17_VML2_REQ_FIFO17",
+	"UTCL2_ROUTER_GROUP18_VML2_REQ_FIFO18",
+	"UTCL2_ROUTER_GROUP19_VML2_REQ_FIFO19",
+	"UTCL2_ROUTER_GROUP20_VML2_REQ_FIFO20",
+	"UTCL2_ROUTER_GROUP21_VML2_REQ_FIFO21",
+	"UTCL2_ROUTER_GROUP22_VML2_REQ_FIFO22",
+	"UTCL2_ROUTER_GROUP23_VML2_REQ_FIFO23",
+	"UTCL2_ROUTER_GROUP24_VML2_REQ_FIFO24",
+};
+
+static const char * const atc_l2_cache_2m_mems[] = {
+	"UTC_ATCL2_CACHE_2M_BANK0_WAY0_MEM",
+	"UTC_ATCL2_CACHE_2M_BANK0_WAY1_MEM",
+	"UTC_ATCL2_CACHE_2M_BANK1_WAY0_MEM",
+	"UTC_ATCL2_CACHE_2M_BANK1_WAY1_MEM",
+};
+
+static const char * const atc_l2_cache_4k_mems[] = {
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM0",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM1",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM2",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM3",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM4",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM5",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM6",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM7",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM0",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM1",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM2",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM3",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM4",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM5",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM6",
+	"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM7",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM0",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM1",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM2",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM3",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM4",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM5",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM6",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM7",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM0",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM1",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM2",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM3",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM4",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM5",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM6",
+	"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM7",
+};
+
+static int gfx_v9_4_query_utc_edc_status(struct amdgpu_device *adev,
+					 struct ras_err_data *err_data)
+{
+	uint32_t i, data;
+	uint32_t sec_count, ded_count;
+
+	WREG32_SOC15(GC, 0, mmVML2_MEM_ECC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmVML2_MEM_ECC_CNTL, 0);
+	WREG32_SOC15(GC, 0, mmVML2_WALKER_MEM_ECC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmVML2_WALKER_MEM_ECC_CNTL, 0);
+	WREG32_SOC15(GC, 0, mmUTCL2_MEM_ECC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmUTCL2_MEM_ECC_CNTL, 0);
+
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_CNTL, 0);
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_DSM_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_CNTL, 0);
+
+	for (i = 0; i < ARRAY_SIZE(vml2_mems); i++) {
+		WREG32_SOC15(GC, 0, mmVML2_MEM_ECC_INDEX, i);
+		data = RREG32_SOC15(GC, 0, mmVML2_MEM_ECC_CNTL);
+
+		sec_count = REG_GET_FIELD(data, VML2_MEM_ECC_CNTL, SEC_COUNT);
+		if (sec_count) {
+			dev_info(adev->dev,
+				 "Instance[%d]: SubBlock %s, SEC %d\n", i,
+				 vml2_mems[i], sec_count);
+			err_data->ce_count += sec_count;
+		}
+
+		ded_count = REG_GET_FIELD(data, VML2_MEM_ECC_CNTL, DED_COUNT);
+		if (ded_count) {
+			dev_info(adev->dev,
+				 "Instance[%d]: SubBlock %s, DED %d\n", i,
+				 vml2_mems[i], ded_count);
+			err_data->ue_count += ded_count;
+		}
+	}
+
+	for (i = 0; i < ARRAY_SIZE(vml2_walker_mems); i++) {
+		WREG32_SOC15(GC, 0, mmVML2_WALKER_MEM_ECC_INDEX, i);
+		data = RREG32_SOC15(GC, 0, mmVML2_WALKER_MEM_ECC_CNTL);
+
+		sec_count = REG_GET_FIELD(data, VML2_WALKER_MEM_ECC_CNTL,
+					  SEC_COUNT);
+		if (sec_count) {
+			dev_info(adev->dev,
+				 "Instance[%d]: SubBlock %s, SEC %d\n", i,
+				 vml2_walker_mems[i], sec_count);
+			err_data->ce_count += sec_count;
+		}
+
+		ded_count = REG_GET_FIELD(data, VML2_WALKER_MEM_ECC_CNTL,
+					  DED_COUNT);
+		if (ded_count) {
+			dev_info(adev->dev,
+				 "Instance[%d]: SubBlock %s, DED %d\n", i,
+				 vml2_walker_mems[i], ded_count);
+			err_data->ue_count += ded_count;
+		}
+	}
+
+	for (i = 0; i < ARRAY_SIZE(utcl2_router_mems); i++) {
+		WREG32_SOC15(GC, 0, mmUTCL2_MEM_ECC_INDEX, i);
+		data = RREG32_SOC15(GC, 0, mmUTCL2_MEM_ECC_CNTL);
+
+		sec_count = REG_GET_FIELD(data, UTCL2_MEM_ECC_CNTL, SEC_COUNT);
+		if (sec_count) {
+			dev_info(adev->dev,
+				 "Instance[%d]: SubBlock %s, SEC %d\n", i,
+				 utcl2_router_mems[i], sec_count);
+			err_data->ce_count += sec_count;
+		}
+
+		ded_count = REG_GET_FIELD(data, UTCL2_MEM_ECC_CNTL, DED_COUNT);
+		if (ded_count) {
+			dev_info(adev->dev,
+				 "Instance[%d]: SubBlock %s, DED %d\n", i,
+				 utcl2_router_mems[i], ded_count);
+			err_data->ue_count += ded_count;
+		}
+	}
+
+	for (i = 0; i < ARRAY_SIZE(atc_l2_cache_2m_mems); i++) {
+		WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_INDEX, i);
+		data = RREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_CNTL);
+
+		sec_count = REG_GET_FIELD(data, ATC_L2_CACHE_2M_DSM_CNTL,
+					  SEC_COUNT);
+		if (sec_count) {
+			dev_info(adev->dev,
+				 "Instance[%d]: SubBlock %s, SEC %d\n", i,
+				 atc_l2_cache_2m_mems[i], sec_count);
+			err_data->ce_count += sec_count;
+		}
+
+		ded_count = REG_GET_FIELD(data, ATC_L2_CACHE_2M_DSM_CNTL,
+					  DED_COUNT);
+		if (ded_count) {
+			dev_info(adev->dev,
+				 "Instance[%d]: SubBlock %s, DED %d\n", i,
+				 atc_l2_cache_2m_mems[i], ded_count);
+			err_data->ue_count += ded_count;
+		}
+	}
+
+	for (i = 0; i < ARRAY_SIZE(atc_l2_cache_4k_mems); i++) {
+		WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_DSM_INDEX, i);
+		data = RREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_DSM_CNTL);
+
+		sec_count = REG_GET_FIELD(data, ATC_L2_CACHE_4K_DSM_CNTL,
+					  SEC_COUNT);
+		if (sec_count) {
+			dev_info(adev->dev,
+				 "Instance[%d]: SubBlock %s, SEC %d\n", i,
+				 atc_l2_cache_4k_mems[i], sec_count);
+			err_data->ce_count += sec_count;
+		}
+
+		ded_count = REG_GET_FIELD(data, ATC_L2_CACHE_4K_DSM_CNTL,
+					  DED_COUNT);
+		if (ded_count) {
+			dev_info(adev->dev,
+				 "Instance[%d]: SubBlock %s, DED %d\n", i,
+				 atc_l2_cache_4k_mems[i], ded_count);
+			err_data->ue_count += ded_count;
+		}
+	}
+
+	WREG32_SOC15(GC, 0, mmVML2_MEM_ECC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmVML2_WALKER_MEM_ECC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmUTCL2_MEM_ECC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_DSM_INDEX, 255);
+
+	return 0;
+}
+
+static int gfx_v9_4_ras_error_count(struct amdgpu_device *adev,
+				    const struct soc15_reg_entry *reg,
+				    uint32_t se_id, uint32_t inst_id,
+				    uint32_t value, uint32_t *sec_count,
+				    uint32_t *ded_count)
+{
+	uint32_t i;
+	uint32_t sec_cnt, ded_cnt;
+
+	for (i = 0; i < ARRAY_SIZE(gfx_v9_4_ras_fields); i++) {
+		if (gfx_v9_4_ras_fields[i].reg_offset != reg->reg_offset ||
+		    gfx_v9_4_ras_fields[i].seg != reg->seg ||
+		    gfx_v9_4_ras_fields[i].inst != reg->inst)
+			continue;
+
+		sec_cnt = (value & gfx_v9_4_ras_fields[i].sec_count_mask) >>
+			  gfx_v9_4_ras_fields[i].sec_count_shift;
+		if (sec_cnt) {
+			dev_info(adev->dev,
+				 "GFX SubBlock %s, Instance[%d][%d], SEC %d\n",
+				 gfx_v9_4_ras_fields[i].name, se_id, inst_id,
+				 sec_cnt);
+			*sec_count += sec_cnt;
+		}
+
+		ded_cnt = (value & gfx_v9_4_ras_fields[i].ded_count_mask) >>
+			  gfx_v9_4_ras_fields[i].ded_count_shift;
+		if (ded_cnt) {
+			dev_info(adev->dev,
+				 "GFX SubBlock %s, Instance[%d][%d], DED %d\n",
+				 gfx_v9_4_ras_fields[i].name, se_id, inst_id,
+				 ded_cnt);
+			*ded_count += ded_cnt;
+		}
+	}
+
+	return 0;
+}
+
+static void gfx_v9_4_query_ras_error_count(struct amdgpu_device *adev,
+					  void *ras_error_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+	uint32_t sec_count = 0, ded_count = 0;
+	uint32_t i, j, k;
+	uint32_t reg_value;
+
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
+		return;
+
+	err_data->ue_count = 0;
+	err_data->ce_count = 0;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	for (i = 0; i < ARRAY_SIZE(gfx_v9_4_edc_counter_regs); i++) {
+		for (j = 0; j < gfx_v9_4_edc_counter_regs[i].se_num; j++) {
+			for (k = 0; k < gfx_v9_4_edc_counter_regs[i].instance;
+			     k++) {
+				gfx_v9_4_select_se_sh(adev, j, 0, k);
+				reg_value = RREG32(SOC15_REG_ENTRY_OFFSET(
+					gfx_v9_4_edc_counter_regs[i]));
+				if (reg_value)
+					gfx_v9_4_ras_error_count(adev,
+						&gfx_v9_4_edc_counter_regs[i],
+						j, k, reg_value, &sec_count,
+						&ded_count);
+			}
+		}
+	}
+
+	err_data->ce_count += sec_count;
+	err_data->ue_count += ded_count;
+
+	gfx_v9_4_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	gfx_v9_4_query_utc_edc_status(adev, err_data);
+
+}
+
+static void gfx_v9_4_reset_ras_error_count(struct amdgpu_device *adev)
+{
+	int i, j, k;
+
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
+		return;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < ARRAY_SIZE(gfx_v9_4_edc_counter_regs); i++) {
+		for (j = 0; j < gfx_v9_4_edc_counter_regs[i].se_num; j++) {
+			for (k = 0; k < gfx_v9_4_edc_counter_regs[i].instance;
+			     k++) {
+				gfx_v9_4_select_se_sh(adev, j, 0x0, k);
+				RREG32(SOC15_REG_ENTRY_OFFSET(
+					gfx_v9_4_edc_counter_regs[i]));
+			}
+		}
+	}
+	WREG32_SOC15(GC, 0, mmGRBM_GFX_INDEX, 0xe0000000);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	WREG32_SOC15(GC, 0, mmVML2_MEM_ECC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmVML2_MEM_ECC_CNTL, 0);
+	WREG32_SOC15(GC, 0, mmVML2_WALKER_MEM_ECC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmVML2_WALKER_MEM_ECC_CNTL, 0);
+	WREG32_SOC15(GC, 0, mmUTCL2_MEM_ECC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmUTCL2_MEM_ECC_CNTL, 0);
+
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_CNTL, 0);
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_DSM_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_CNTL, 0);
+
+	for (i = 0; i < ARRAY_SIZE(vml2_mems); i++) {
+		WREG32_SOC15(GC, 0, mmVML2_MEM_ECC_INDEX, i);
+		RREG32_SOC15(GC, 0, mmVML2_MEM_ECC_CNTL);
+	}
+
+	for (i = 0; i < ARRAY_SIZE(vml2_walker_mems); i++) {
+		WREG32_SOC15(GC, 0, mmVML2_WALKER_MEM_ECC_INDEX, i);
+		RREG32_SOC15(GC, 0, mmVML2_WALKER_MEM_ECC_CNTL);
+	}
+
+	for (i = 0; i < ARRAY_SIZE(utcl2_router_mems); i++) {
+		WREG32_SOC15(GC, 0, mmUTCL2_MEM_ECC_INDEX, i);
+		RREG32_SOC15(GC, 0, mmUTCL2_MEM_ECC_CNTL);
+	}
+
+	for (i = 0; i < ARRAY_SIZE(atc_l2_cache_2m_mems); i++) {
+		WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_INDEX, i);
+		RREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_CNTL);
+	}
+
+	for (i = 0; i < ARRAY_SIZE(atc_l2_cache_4k_mems); i++) {
+		WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_DSM_INDEX, i);
+		RREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_DSM_CNTL);
+	}
+
+	WREG32_SOC15(GC, 0, mmVML2_MEM_ECC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmVML2_WALKER_MEM_ECC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmUTCL2_MEM_ECC_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_INDEX, 255);
+	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_DSM_INDEX, 255);
+}
+
+static const struct soc15_reg_entry gfx_v9_4_ea_err_status_regs = {
+	SOC15_REG_ENTRY(GC, 0, mmGCEA_ERR_STATUS), 0, 1, 32
+};
+
+static void gfx_v9_4_query_ras_error_status(struct amdgpu_device *adev)
+{
+	uint32_t i, j;
+	uint32_t reg_value;
+
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
+		return;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	for (i = 0; i < gfx_v9_4_ea_err_status_regs.se_num; i++) {
+		for (j = 0; j < gfx_v9_4_ea_err_status_regs.instance;
+		     j++) {
+			gfx_v9_4_select_se_sh(adev, i, 0, j);
+			reg_value = RREG32(SOC15_REG_ENTRY_OFFSET(
+				gfx_v9_4_ea_err_status_regs));
+			if (REG_GET_FIELD(reg_value, GCEA_ERR_STATUS, SDP_RDRSP_STATUS) ||
+			    REG_GET_FIELD(reg_value, GCEA_ERR_STATUS, SDP_WRRSP_STATUS) ||
+			    REG_GET_FIELD(reg_value, GCEA_ERR_STATUS, SDP_RDRSP_DATAPARITY_ERROR)) {
+				/* SDP read/write error/parity error in FUE_IS_FATAL mode
+				 * can cause system fatal error in arcturas. Harvest the error
+				 * status before GPU reset */
+				dev_warn(adev->dev, "GCEA err detected at instance: %d, status: 0x%x!\n",
+						j, reg_value);
+			}
+		}
+	}
+
+	gfx_v9_4_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+}
+
+
+const struct amdgpu_ras_block_hw_ops  gfx_v9_4_ras_ops = {
+	.query_ras_error_count = &gfx_v9_4_query_ras_error_count,
+	.reset_ras_error_count = &gfx_v9_4_reset_ras_error_count,
+	.query_ras_error_status = &gfx_v9_4_query_ras_error_status,
+};
+
+struct amdgpu_gfx_ras gfx_v9_4_ras = {
+	.ras_block = {
+		.hw_ops = &gfx_v9_4_ras_ops,
+	},
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.h b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.h
new file mode 100644
index 000000000000..ca520a767267
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFX_V9_4_H__
+#define __GFX_V9_4_H__
+
+extern struct amdgpu_gfx_ras gfx_v9_4_ras;
+
+#endif /* __GFX_V9_4_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c
new file mode 100644
index 000000000000..8058ea91ecaf
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c
@@ -0,0 +1,1938 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "soc15.h"
+#include "soc15d.h"
+
+#include "gc/gc_9_4_2_offset.h"
+#include "gc/gc_9_4_2_sh_mask.h"
+#include "gfx_v9_0.h"
+
+#include "gfx_v9_4_2.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_gfx.h"
+
+#define SE_ID_MAX 8
+#define CU_ID_MAX 16
+#define SIMD_ID_MAX 4
+#define WAVE_ID_MAX 10
+
+enum gfx_v9_4_2_utc_type {
+	VML2_MEM,
+	VML2_WALKER_MEM,
+	UTCL2_MEM,
+	ATC_L2_CACHE_2M,
+	ATC_L2_CACHE_32K,
+	ATC_L2_CACHE_4K
+};
+
+struct gfx_v9_4_2_utc_block {
+	enum gfx_v9_4_2_utc_type type;
+	uint32_t num_banks;
+	uint32_t num_ways;
+	uint32_t num_mem_blocks;
+	struct soc15_reg idx_reg;
+	struct soc15_reg data_reg;
+	uint32_t sec_count_mask;
+	uint32_t sec_count_shift;
+	uint32_t ded_count_mask;
+	uint32_t ded_count_shift;
+	uint32_t clear;
+};
+
+static const struct soc15_reg_golden golden_settings_gc_9_4_2_alde_die_0[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_0, 0x3fffffff, 0x141dc920),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_1, 0x3fffffff, 0x3b458b93),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_2, 0x3fffffff, 0x1a4f5583),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_3, 0x3fffffff, 0x317717f6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_4, 0x3fffffff, 0x107cc1e6),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_5, 0x3ff, 0x351),
+};
+
+static const struct soc15_reg_golden golden_settings_gc_9_4_2_alde_die_1[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_0, 0x3fffffff, 0x2591aa38),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_1, 0x3fffffff, 0xac9e88b),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_2, 0x3fffffff, 0x2bc3369b),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_3, 0x3fffffff, 0xfb74ee),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_4, 0x3fffffff, 0x21f0a2fe),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_5, 0x3ff, 0x49),
+};
+
+static const struct soc15_reg_golden golden_settings_gc_9_4_2_alde[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regGB_ADDR_CONFIG, 0xffff77ff, 0x2a114042),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regTA_CNTL_AUX, 0xfffffeef, 0x10b0000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_UTCL1_CNTL1, 0xffffffff, 0x30800400),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, regTCI_CNTL_3, 0xff, 0x20),
+};
+
+/*
+ * This shader is used to clear VGPRS and LDS, and also write the input
+ * pattern into the write back buffer, which will be used by driver to
+ * check whether all SIMDs have been covered.
+*/
+static const u32 vgpr_init_compute_shader_aldebaran[] = {
+	0xb8840904, 0xb8851a04, 0xb8861344, 0xb8831804, 0x9208ff06, 0x00000280,
+	0x9209a805, 0x920a8a04, 0x81080908, 0x81080a08, 0x81080308, 0x8e078208,
+	0x81078407, 0xc0410080, 0x00000007, 0xbf8c0000, 0xbf8a0000, 0xd3d94000,
+	0x18000080, 0xd3d94001, 0x18000080, 0xd3d94002, 0x18000080, 0xd3d94003,
+	0x18000080, 0xd3d94004, 0x18000080, 0xd3d94005, 0x18000080, 0xd3d94006,
+	0x18000080, 0xd3d94007, 0x18000080, 0xd3d94008, 0x18000080, 0xd3d94009,
+	0x18000080, 0xd3d9400a, 0x18000080, 0xd3d9400b, 0x18000080, 0xd3d9400c,
+	0x18000080, 0xd3d9400d, 0x18000080, 0xd3d9400e, 0x18000080, 0xd3d9400f,
+	0x18000080, 0xd3d94010, 0x18000080, 0xd3d94011, 0x18000080, 0xd3d94012,
+	0x18000080, 0xd3d94013, 0x18000080, 0xd3d94014, 0x18000080, 0xd3d94015,
+	0x18000080, 0xd3d94016, 0x18000080, 0xd3d94017, 0x18000080, 0xd3d94018,
+	0x18000080, 0xd3d94019, 0x18000080, 0xd3d9401a, 0x18000080, 0xd3d9401b,
+	0x18000080, 0xd3d9401c, 0x18000080, 0xd3d9401d, 0x18000080, 0xd3d9401e,
+	0x18000080, 0xd3d9401f, 0x18000080, 0xd3d94020, 0x18000080, 0xd3d94021,
+	0x18000080, 0xd3d94022, 0x18000080, 0xd3d94023, 0x18000080, 0xd3d94024,
+	0x18000080, 0xd3d94025, 0x18000080, 0xd3d94026, 0x18000080, 0xd3d94027,
+	0x18000080, 0xd3d94028, 0x18000080, 0xd3d94029, 0x18000080, 0xd3d9402a,
+	0x18000080, 0xd3d9402b, 0x18000080, 0xd3d9402c, 0x18000080, 0xd3d9402d,
+	0x18000080, 0xd3d9402e, 0x18000080, 0xd3d9402f, 0x18000080, 0xd3d94030,
+	0x18000080, 0xd3d94031, 0x18000080, 0xd3d94032, 0x18000080, 0xd3d94033,
+	0x18000080, 0xd3d94034, 0x18000080, 0xd3d94035, 0x18000080, 0xd3d94036,
+	0x18000080, 0xd3d94037, 0x18000080, 0xd3d94038, 0x18000080, 0xd3d94039,
+	0x18000080, 0xd3d9403a, 0x18000080, 0xd3d9403b, 0x18000080, 0xd3d9403c,
+	0x18000080, 0xd3d9403d, 0x18000080, 0xd3d9403e, 0x18000080, 0xd3d9403f,
+	0x18000080, 0xd3d94040, 0x18000080, 0xd3d94041, 0x18000080, 0xd3d94042,
+	0x18000080, 0xd3d94043, 0x18000080, 0xd3d94044, 0x18000080, 0xd3d94045,
+	0x18000080, 0xd3d94046, 0x18000080, 0xd3d94047, 0x18000080, 0xd3d94048,
+	0x18000080, 0xd3d94049, 0x18000080, 0xd3d9404a, 0x18000080, 0xd3d9404b,
+	0x18000080, 0xd3d9404c, 0x18000080, 0xd3d9404d, 0x18000080, 0xd3d9404e,
+	0x18000080, 0xd3d9404f, 0x18000080, 0xd3d94050, 0x18000080, 0xd3d94051,
+	0x18000080, 0xd3d94052, 0x18000080, 0xd3d94053, 0x18000080, 0xd3d94054,
+	0x18000080, 0xd3d94055, 0x18000080, 0xd3d94056, 0x18000080, 0xd3d94057,
+	0x18000080, 0xd3d94058, 0x18000080, 0xd3d94059, 0x18000080, 0xd3d9405a,
+	0x18000080, 0xd3d9405b, 0x18000080, 0xd3d9405c, 0x18000080, 0xd3d9405d,
+	0x18000080, 0xd3d9405e, 0x18000080, 0xd3d9405f, 0x18000080, 0xd3d94060,
+	0x18000080, 0xd3d94061, 0x18000080, 0xd3d94062, 0x18000080, 0xd3d94063,
+	0x18000080, 0xd3d94064, 0x18000080, 0xd3d94065, 0x18000080, 0xd3d94066,
+	0x18000080, 0xd3d94067, 0x18000080, 0xd3d94068, 0x18000080, 0xd3d94069,
+	0x18000080, 0xd3d9406a, 0x18000080, 0xd3d9406b, 0x18000080, 0xd3d9406c,
+	0x18000080, 0xd3d9406d, 0x18000080, 0xd3d9406e, 0x18000080, 0xd3d9406f,
+	0x18000080, 0xd3d94070, 0x18000080, 0xd3d94071, 0x18000080, 0xd3d94072,
+	0x18000080, 0xd3d94073, 0x18000080, 0xd3d94074, 0x18000080, 0xd3d94075,
+	0x18000080, 0xd3d94076, 0x18000080, 0xd3d94077, 0x18000080, 0xd3d94078,
+	0x18000080, 0xd3d94079, 0x18000080, 0xd3d9407a, 0x18000080, 0xd3d9407b,
+	0x18000080, 0xd3d9407c, 0x18000080, 0xd3d9407d, 0x18000080, 0xd3d9407e,
+	0x18000080, 0xd3d9407f, 0x18000080, 0xd3d94080, 0x18000080, 0xd3d94081,
+	0x18000080, 0xd3d94082, 0x18000080, 0xd3d94083, 0x18000080, 0xd3d94084,
+	0x18000080, 0xd3d94085, 0x18000080, 0xd3d94086, 0x18000080, 0xd3d94087,
+	0x18000080, 0xd3d94088, 0x18000080, 0xd3d94089, 0x18000080, 0xd3d9408a,
+	0x18000080, 0xd3d9408b, 0x18000080, 0xd3d9408c, 0x18000080, 0xd3d9408d,
+	0x18000080, 0xd3d9408e, 0x18000080, 0xd3d9408f, 0x18000080, 0xd3d94090,
+	0x18000080, 0xd3d94091, 0x18000080, 0xd3d94092, 0x18000080, 0xd3d94093,
+	0x18000080, 0xd3d94094, 0x18000080, 0xd3d94095, 0x18000080, 0xd3d94096,
+	0x18000080, 0xd3d94097, 0x18000080, 0xd3d94098, 0x18000080, 0xd3d94099,
+	0x18000080, 0xd3d9409a, 0x18000080, 0xd3d9409b, 0x18000080, 0xd3d9409c,
+	0x18000080, 0xd3d9409d, 0x18000080, 0xd3d9409e, 0x18000080, 0xd3d9409f,
+	0x18000080, 0xd3d940a0, 0x18000080, 0xd3d940a1, 0x18000080, 0xd3d940a2,
+	0x18000080, 0xd3d940a3, 0x18000080, 0xd3d940a4, 0x18000080, 0xd3d940a5,
+	0x18000080, 0xd3d940a6, 0x18000080, 0xd3d940a7, 0x18000080, 0xd3d940a8,
+	0x18000080, 0xd3d940a9, 0x18000080, 0xd3d940aa, 0x18000080, 0xd3d940ab,
+	0x18000080, 0xd3d940ac, 0x18000080, 0xd3d940ad, 0x18000080, 0xd3d940ae,
+	0x18000080, 0xd3d940af, 0x18000080, 0xd3d940b0, 0x18000080, 0xd3d940b1,
+	0x18000080, 0xd3d940b2, 0x18000080, 0xd3d940b3, 0x18000080, 0xd3d940b4,
+	0x18000080, 0xd3d940b5, 0x18000080, 0xd3d940b6, 0x18000080, 0xd3d940b7,
+	0x18000080, 0xd3d940b8, 0x18000080, 0xd3d940b9, 0x18000080, 0xd3d940ba,
+	0x18000080, 0xd3d940bb, 0x18000080, 0xd3d940bc, 0x18000080, 0xd3d940bd,
+	0x18000080, 0xd3d940be, 0x18000080, 0xd3d940bf, 0x18000080, 0xd3d940c0,
+	0x18000080, 0xd3d940c1, 0x18000080, 0xd3d940c2, 0x18000080, 0xd3d940c3,
+	0x18000080, 0xd3d940c4, 0x18000080, 0xd3d940c5, 0x18000080, 0xd3d940c6,
+	0x18000080, 0xd3d940c7, 0x18000080, 0xd3d940c8, 0x18000080, 0xd3d940c9,
+	0x18000080, 0xd3d940ca, 0x18000080, 0xd3d940cb, 0x18000080, 0xd3d940cc,
+	0x18000080, 0xd3d940cd, 0x18000080, 0xd3d940ce, 0x18000080, 0xd3d940cf,
+	0x18000080, 0xd3d940d0, 0x18000080, 0xd3d940d1, 0x18000080, 0xd3d940d2,
+	0x18000080, 0xd3d940d3, 0x18000080, 0xd3d940d4, 0x18000080, 0xd3d940d5,
+	0x18000080, 0xd3d940d6, 0x18000080, 0xd3d940d7, 0x18000080, 0xd3d940d8,
+	0x18000080, 0xd3d940d9, 0x18000080, 0xd3d940da, 0x18000080, 0xd3d940db,
+	0x18000080, 0xd3d940dc, 0x18000080, 0xd3d940dd, 0x18000080, 0xd3d940de,
+	0x18000080, 0xd3d940df, 0x18000080, 0xd3d940e0, 0x18000080, 0xd3d940e1,
+	0x18000080, 0xd3d940e2, 0x18000080, 0xd3d940e3, 0x18000080, 0xd3d940e4,
+	0x18000080, 0xd3d940e5, 0x18000080, 0xd3d940e6, 0x18000080, 0xd3d940e7,
+	0x18000080, 0xd3d940e8, 0x18000080, 0xd3d940e9, 0x18000080, 0xd3d940ea,
+	0x18000080, 0xd3d940eb, 0x18000080, 0xd3d940ec, 0x18000080, 0xd3d940ed,
+	0x18000080, 0xd3d940ee, 0x18000080, 0xd3d940ef, 0x18000080, 0xd3d940f0,
+	0x18000080, 0xd3d940f1, 0x18000080, 0xd3d940f2, 0x18000080, 0xd3d940f3,
+	0x18000080, 0xd3d940f4, 0x18000080, 0xd3d940f5, 0x18000080, 0xd3d940f6,
+	0x18000080, 0xd3d940f7, 0x18000080, 0xd3d940f8, 0x18000080, 0xd3d940f9,
+	0x18000080, 0xd3d940fa, 0x18000080, 0xd3d940fb, 0x18000080, 0xd3d940fc,
+	0x18000080, 0xd3d940fd, 0x18000080, 0xd3d940fe, 0x18000080, 0xd3d940ff,
+	0x18000080, 0xb07c0000, 0xbe8a00ff, 0x000000f8, 0xbf11080a, 0x7e000280,
+	0x7e020280, 0x7e040280, 0x7e060280, 0x7e080280, 0x7e0a0280, 0x7e0c0280,
+	0x7e0e0280, 0x808a880a, 0xbe80320a, 0xbf84fff5, 0xbf9c0000, 0xd28c0001,
+	0x0001007f, 0xd28d0001, 0x0002027e, 0x10020288, 0xbe8b0004, 0xb78b4000,
+	0xd1196a01, 0x00001701, 0xbe8a0087, 0xbefc00c1, 0xd89c4000, 0x00020201,
+	0xd89cc080, 0x00040401, 0x320202ff, 0x00000800, 0x808a810a, 0xbf84fff8,
+	0xbf810000,
+};
+
+const struct soc15_reg_entry vgpr_init_regs_aldebaran[] = {
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_RESOURCE_LIMITS), 0x0000000 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_NUM_THREAD_X), 0x40 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_NUM_THREAD_Y), 4 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_NUM_THREAD_Z), 1 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_PGM_RSRC1), 0xbf },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_PGM_RSRC2), 0x400006 },  /* 64KB LDS */
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_PGM_RSRC3), 0x3F }, /*  63 - accum-offset = 256 */
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE0), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE1), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE2), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE3), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE4), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE5), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE6), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE7), 0xffffffff },
+};
+
+/*
+ * The below shaders are used to clear SGPRS, and also write the input
+ * pattern into the write back buffer. The first two dispatch should be
+ * scheduled simultaneously which make sure that all SGPRS could be
+ * allocated, so the dispatch 1 need check write back buffer before scheduled,
+ * make sure that waves of dispatch 0 are all dispacthed to all simds
+ * balanced. both dispatch 0 and dispatch 1 should be halted until all waves
+ * are dispatched, and then driver write a pattern to the shared memory to make
+ * all waves continue.
+*/
+static const u32 sgpr112_init_compute_shader_aldebaran[] = {
+	0xb8840904, 0xb8851a04, 0xb8861344, 0xb8831804, 0x9208ff06, 0x00000280,
+	0x9209a805, 0x920a8a04, 0x81080908, 0x81080a08, 0x81080308, 0x8e078208,
+	0x81078407, 0xc0410080, 0x00000007, 0xbf8c0000, 0xbf8e003f, 0xc0030200,
+	0x00000000, 0xbf8c0000, 0xbf06ff08, 0xdeadbeaf, 0xbf84fff9, 0x81028102,
+	0xc0410080, 0x00000007, 0xbf8c0000, 0xbf8a0000, 0xbefc0080, 0xbeea0080,
+	0xbeeb0080, 0xbf00f280, 0xbee60080, 0xbee70080, 0xbee80080, 0xbee90080,
+	0xbefe0080, 0xbeff0080, 0xbe880080, 0xbe890080, 0xbe8a0080, 0xbe8b0080,
+	0xbe8c0080, 0xbe8d0080, 0xbe8e0080, 0xbe8f0080, 0xbe900080, 0xbe910080,
+	0xbe920080, 0xbe930080, 0xbe940080, 0xbe950080, 0xbe960080, 0xbe970080,
+	0xbe980080, 0xbe990080, 0xbe9a0080, 0xbe9b0080, 0xbe9c0080, 0xbe9d0080,
+	0xbe9e0080, 0xbe9f0080, 0xbea00080, 0xbea10080, 0xbea20080, 0xbea30080,
+	0xbea40080, 0xbea50080, 0xbea60080, 0xbea70080, 0xbea80080, 0xbea90080,
+	0xbeaa0080, 0xbeab0080, 0xbeac0080, 0xbead0080, 0xbeae0080, 0xbeaf0080,
+	0xbeb00080, 0xbeb10080, 0xbeb20080, 0xbeb30080, 0xbeb40080, 0xbeb50080,
+	0xbeb60080, 0xbeb70080, 0xbeb80080, 0xbeb90080, 0xbeba0080, 0xbebb0080,
+	0xbebc0080, 0xbebd0080, 0xbebe0080, 0xbebf0080, 0xbec00080, 0xbec10080,
+	0xbec20080, 0xbec30080, 0xbec40080, 0xbec50080, 0xbec60080, 0xbec70080,
+	0xbec80080, 0xbec90080, 0xbeca0080, 0xbecb0080, 0xbecc0080, 0xbecd0080,
+	0xbece0080, 0xbecf0080, 0xbed00080, 0xbed10080, 0xbed20080, 0xbed30080,
+	0xbed40080, 0xbed50080, 0xbed60080, 0xbed70080, 0xbed80080, 0xbed90080,
+	0xbeda0080, 0xbedb0080, 0xbedc0080, 0xbedd0080, 0xbede0080, 0xbedf0080,
+	0xbee00080, 0xbee10080, 0xbee20080, 0xbee30080, 0xbee40080, 0xbee50080,
+	0xbf810000
+};
+
+const struct soc15_reg_entry sgpr112_init_regs_aldebaran[] = {
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_RESOURCE_LIMITS), 0x0000000 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_NUM_THREAD_X), 0x40 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_NUM_THREAD_Y), 8 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_NUM_THREAD_Z), 1 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_PGM_RSRC1), 0x340 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_PGM_RSRC2), 0x6 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_PGM_RSRC3), 0x0 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE0), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE1), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE2), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE3), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE4), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE5), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE6), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE7), 0xffffffff },
+};
+
+static const u32 sgpr96_init_compute_shader_aldebaran[] = {
+	0xb8840904, 0xb8851a04, 0xb8861344, 0xb8831804, 0x9208ff06, 0x00000280,
+	0x9209a805, 0x920a8a04, 0x81080908, 0x81080a08, 0x81080308, 0x8e078208,
+	0x81078407, 0xc0410080, 0x00000007, 0xbf8c0000, 0xbf8e003f, 0xc0030200,
+	0x00000000, 0xbf8c0000, 0xbf06ff08, 0xdeadbeaf, 0xbf84fff9, 0x81028102,
+	0xc0410080, 0x00000007, 0xbf8c0000, 0xbf8a0000, 0xbefc0080, 0xbeea0080,
+	0xbeeb0080, 0xbf00f280, 0xbee60080, 0xbee70080, 0xbee80080, 0xbee90080,
+	0xbefe0080, 0xbeff0080, 0xbe880080, 0xbe890080, 0xbe8a0080, 0xbe8b0080,
+	0xbe8c0080, 0xbe8d0080, 0xbe8e0080, 0xbe8f0080, 0xbe900080, 0xbe910080,
+	0xbe920080, 0xbe930080, 0xbe940080, 0xbe950080, 0xbe960080, 0xbe970080,
+	0xbe980080, 0xbe990080, 0xbe9a0080, 0xbe9b0080, 0xbe9c0080, 0xbe9d0080,
+	0xbe9e0080, 0xbe9f0080, 0xbea00080, 0xbea10080, 0xbea20080, 0xbea30080,
+	0xbea40080, 0xbea50080, 0xbea60080, 0xbea70080, 0xbea80080, 0xbea90080,
+	0xbeaa0080, 0xbeab0080, 0xbeac0080, 0xbead0080, 0xbeae0080, 0xbeaf0080,
+	0xbeb00080, 0xbeb10080, 0xbeb20080, 0xbeb30080, 0xbeb40080, 0xbeb50080,
+	0xbeb60080, 0xbeb70080, 0xbeb80080, 0xbeb90080, 0xbeba0080, 0xbebb0080,
+	0xbebc0080, 0xbebd0080, 0xbebe0080, 0xbebf0080, 0xbec00080, 0xbec10080,
+	0xbec20080, 0xbec30080, 0xbec40080, 0xbec50080, 0xbec60080, 0xbec70080,
+	0xbec80080, 0xbec90080, 0xbeca0080, 0xbecb0080, 0xbecc0080, 0xbecd0080,
+	0xbece0080, 0xbecf0080, 0xbed00080, 0xbed10080, 0xbed20080, 0xbed30080,
+	0xbed40080, 0xbed50080, 0xbed60080, 0xbed70080, 0xbed80080, 0xbed90080,
+	0xbf810000,
+};
+
+const struct soc15_reg_entry sgpr96_init_regs_aldebaran[] = {
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_RESOURCE_LIMITS), 0x0000000 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_NUM_THREAD_X), 0x40 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_NUM_THREAD_Y), 0xc },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_NUM_THREAD_Z), 1 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_PGM_RSRC1), 0x2c0 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_PGM_RSRC2), 0x6 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_PGM_RSRC3), 0x0 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE0), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE1), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE2), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE3), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE4), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE5), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE6), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE7), 0xffffffff },
+};
+
+/*
+ * This shader is used to clear the uninitiated sgprs after the above
+ * two dispatches, because of hardware feature, dispath 0 couldn't clear
+ * top hole sgprs. Therefore need 4 waves per SIMD to cover these sgprs
+*/
+static const u32 sgpr64_init_compute_shader_aldebaran[] = {
+	0xb8840904, 0xb8851a04, 0xb8861344, 0xb8831804, 0x9208ff06, 0x00000280,
+	0x9209a805, 0x920a8a04, 0x81080908, 0x81080a08, 0x81080308, 0x8e078208,
+	0x81078407, 0xc0410080, 0x00000007, 0xbf8c0000, 0xbf8e003f, 0xc0030200,
+	0x00000000, 0xbf8c0000, 0xbf06ff08, 0xdeadbeaf, 0xbf84fff9, 0x81028102,
+	0xc0410080, 0x00000007, 0xbf8c0000, 0xbf8a0000, 0xbefc0080, 0xbeea0080,
+	0xbeeb0080, 0xbf00f280, 0xbee60080, 0xbee70080, 0xbee80080, 0xbee90080,
+	0xbefe0080, 0xbeff0080, 0xbe880080, 0xbe890080, 0xbe8a0080, 0xbe8b0080,
+	0xbe8c0080, 0xbe8d0080, 0xbe8e0080, 0xbe8f0080, 0xbe900080, 0xbe910080,
+	0xbe920080, 0xbe930080, 0xbe940080, 0xbe950080, 0xbe960080, 0xbe970080,
+	0xbe980080, 0xbe990080, 0xbe9a0080, 0xbe9b0080, 0xbe9c0080, 0xbe9d0080,
+	0xbe9e0080, 0xbe9f0080, 0xbea00080, 0xbea10080, 0xbea20080, 0xbea30080,
+	0xbea40080, 0xbea50080, 0xbea60080, 0xbea70080, 0xbea80080, 0xbea90080,
+	0xbeaa0080, 0xbeab0080, 0xbeac0080, 0xbead0080, 0xbeae0080, 0xbeaf0080,
+	0xbeb00080, 0xbeb10080, 0xbeb20080, 0xbeb30080, 0xbeb40080, 0xbeb50080,
+	0xbeb60080, 0xbeb70080, 0xbeb80080, 0xbeb90080, 0xbf810000,
+};
+
+const struct soc15_reg_entry sgpr64_init_regs_aldebaran[] = {
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_RESOURCE_LIMITS), 0x0000000 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_NUM_THREAD_X), 0x40 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_NUM_THREAD_Y), 0x10 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_NUM_THREAD_Z), 1 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_PGM_RSRC1), 0x1c0 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_PGM_RSRC2), 0x6 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_PGM_RSRC3), 0x0 },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE0), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE1), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE2), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE3), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE4), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE5), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE6), 0xffffffff },
+	{ SOC15_REG_ENTRY(GC, 0, regCOMPUTE_STATIC_THREAD_MGMT_SE7), 0xffffffff },
+};
+
+static int gfx_v9_4_2_run_shader(struct amdgpu_device *adev,
+				 struct amdgpu_ring *ring,
+				 struct amdgpu_ib *ib,
+				 const u32 *shader_ptr, u32 shader_size,
+				 const struct soc15_reg_entry *init_regs, u32 regs_size,
+				 u32 compute_dim_x, u64 wb_gpu_addr, u32 pattern,
+				 struct dma_fence **fence_ptr)
+{
+	int r, i;
+	uint32_t total_size, shader_offset;
+	u64 gpu_addr;
+
+	total_size = (regs_size * 3 + 4 + 5 + 5) * 4;
+	total_size = ALIGN(total_size, 256);
+	shader_offset = total_size;
+	total_size += ALIGN(shader_size, 256);
+
+	/* allocate an indirect buffer to put the commands in */
+	memset(ib, 0, sizeof(*ib));
+	r = amdgpu_ib_get(adev, NULL, total_size,
+					AMDGPU_IB_POOL_DIRECT, ib);
+	if (r) {
+		dev_err(adev->dev, "failed to get ib (%d).\n", r);
+		return r;
+	}
+
+	/* load the compute shaders */
+	for (i = 0; i < shader_size/sizeof(u32); i++)
+		ib->ptr[i + (shader_offset / 4)] = shader_ptr[i];
+
+	/* init the ib length to 0 */
+	ib->length_dw = 0;
+
+	/* write the register state for the compute dispatch */
+	for (i = 0; i < regs_size; i++) {
+		ib->ptr[ib->length_dw++] = PACKET3(PACKET3_SET_SH_REG, 1);
+		ib->ptr[ib->length_dw++] = SOC15_REG_ENTRY_OFFSET(init_regs[i])
+								- PACKET3_SET_SH_REG_START;
+		ib->ptr[ib->length_dw++] = init_regs[i].reg_value;
+	}
+
+	/* write the shader start address: mmCOMPUTE_PGM_LO, mmCOMPUTE_PGM_HI */
+	gpu_addr = (ib->gpu_addr + (u64)shader_offset) >> 8;
+	ib->ptr[ib->length_dw++] = PACKET3(PACKET3_SET_SH_REG, 2);
+	ib->ptr[ib->length_dw++] = SOC15_REG_OFFSET(GC, 0, regCOMPUTE_PGM_LO)
+							- PACKET3_SET_SH_REG_START;
+	ib->ptr[ib->length_dw++] = lower_32_bits(gpu_addr);
+	ib->ptr[ib->length_dw++] = upper_32_bits(gpu_addr);
+
+	/* write the wb buffer address */
+	ib->ptr[ib->length_dw++] = PACKET3(PACKET3_SET_SH_REG, 3);
+	ib->ptr[ib->length_dw++] = SOC15_REG_OFFSET(GC, 0, regCOMPUTE_USER_DATA_0)
+							- PACKET3_SET_SH_REG_START;
+	ib->ptr[ib->length_dw++] = lower_32_bits(wb_gpu_addr);
+	ib->ptr[ib->length_dw++] = upper_32_bits(wb_gpu_addr);
+	ib->ptr[ib->length_dw++] = pattern;
+
+	/* write dispatch packet */
+	ib->ptr[ib->length_dw++] = PACKET3(PACKET3_DISPATCH_DIRECT, 3);
+	ib->ptr[ib->length_dw++] = compute_dim_x; /* x */
+	ib->ptr[ib->length_dw++] = 1; /* y */
+	ib->ptr[ib->length_dw++] = 1; /* z */
+	ib->ptr[ib->length_dw++] =
+		REG_SET_FIELD(0, COMPUTE_DISPATCH_INITIATOR, COMPUTE_SHADER_EN, 1);
+
+	/* shedule the ib on the ring */
+	r = amdgpu_ib_schedule(ring, 1, ib, NULL, fence_ptr);
+	if (r) {
+		dev_err(adev->dev, "ib submit failed (%d).\n", r);
+		amdgpu_ib_free(ib, NULL);
+	}
+	return r;
+}
+
+static void gfx_v9_4_2_log_wave_assignment(struct amdgpu_device *adev, uint32_t *wb_ptr)
+{
+	uint32_t se, cu, simd, wave;
+	uint32_t offset = 0;
+	char *str;
+	int size;
+
+	str = kmalloc(256, GFP_KERNEL);
+	if (!str)
+		return;
+
+	dev_dbg(adev->dev, "wave assignment:\n");
+
+	for (se = 0; se < adev->gfx.config.max_shader_engines; se++) {
+		for (cu = 0; cu < CU_ID_MAX; cu++) {
+			memset(str, 0, 256);
+			size = sprintf(str, "SE[%02d]CU[%02d]: ", se, cu);
+			for (simd = 0; simd < SIMD_ID_MAX; simd++) {
+				size += sprintf(str + size, "[");
+				for (wave = 0; wave < WAVE_ID_MAX; wave++) {
+					size += sprintf(str + size, "%x", wb_ptr[offset]);
+					offset++;
+				}
+				size += sprintf(str + size, "]  ");
+			}
+			dev_dbg(adev->dev, "%s\n", str);
+		}
+	}
+
+	kfree(str);
+}
+
+static int gfx_v9_4_2_wait_for_waves_assigned(struct amdgpu_device *adev,
+					      uint32_t *wb_ptr, uint32_t mask,
+					      uint32_t pattern, uint32_t num_wave, bool wait)
+{
+	uint32_t se, cu, simd, wave;
+	uint32_t loop = 0;
+	uint32_t wave_cnt;
+	uint32_t offset;
+
+	do {
+		wave_cnt = 0;
+		offset = 0;
+
+		for (se = 0; se < adev->gfx.config.max_shader_engines; se++)
+			for (cu = 0; cu < CU_ID_MAX; cu++)
+				for (simd = 0; simd < SIMD_ID_MAX; simd++)
+					for (wave = 0; wave < WAVE_ID_MAX; wave++) {
+						if (((1 << wave) & mask) &&
+						    (wb_ptr[offset] == pattern))
+							wave_cnt++;
+
+						offset++;
+					}
+
+		if (wave_cnt == num_wave)
+			return 0;
+
+		mdelay(1);
+	} while (++loop < 2000 && wait);
+
+	dev_err(adev->dev, "actual wave num: %d, expected wave num: %d\n",
+		wave_cnt, num_wave);
+
+	gfx_v9_4_2_log_wave_assignment(adev, wb_ptr);
+
+	return -EBADSLT;
+}
+
+static int gfx_v9_4_2_do_sgprs_init(struct amdgpu_device *adev)
+{
+	int r;
+	int wb_size = adev->gfx.config.max_shader_engines *
+			 CU_ID_MAX * SIMD_ID_MAX * WAVE_ID_MAX;
+	struct amdgpu_ib wb_ib;
+	struct amdgpu_ib disp_ibs[3];
+	struct dma_fence *fences[3];
+	u32 pattern[3] = { 0x1, 0x5, 0xa };
+
+	/* bail if the compute ring is not ready */
+	if (!adev->gfx.compute_ring[0].sched.ready ||
+		 !adev->gfx.compute_ring[1].sched.ready)
+		return 0;
+
+	/* allocate the write-back buffer from IB */
+	memset(&wb_ib, 0, sizeof(wb_ib));
+	r = amdgpu_ib_get(adev, NULL, (1 + wb_size) * sizeof(uint32_t),
+			  AMDGPU_IB_POOL_DIRECT, &wb_ib);
+	if (r) {
+		dev_err(adev->dev, "failed to get ib (%d) for wb\n", r);
+		return r;
+	}
+	memset(wb_ib.ptr, 0, (1 + wb_size) * sizeof(uint32_t));
+
+	r = gfx_v9_4_2_run_shader(adev,
+			&adev->gfx.compute_ring[0],
+			&disp_ibs[0],
+			sgpr112_init_compute_shader_aldebaran,
+			sizeof(sgpr112_init_compute_shader_aldebaran),
+			sgpr112_init_regs_aldebaran,
+			ARRAY_SIZE(sgpr112_init_regs_aldebaran),
+			adev->gfx.cu_info.number,
+			wb_ib.gpu_addr, pattern[0], &fences[0]);
+	if (r) {
+		dev_err(adev->dev, "failed to clear first 224 sgprs\n");
+		goto pro_end;
+	}
+
+	r = gfx_v9_4_2_wait_for_waves_assigned(adev,
+			&wb_ib.ptr[1], 0b11,
+			pattern[0],
+			adev->gfx.cu_info.number * SIMD_ID_MAX * 2,
+			true);
+	if (r) {
+		dev_err(adev->dev, "wave coverage failed when clear first 224 sgprs\n");
+		wb_ib.ptr[0] = 0xdeadbeaf; /* stop waves */
+		goto disp0_failed;
+	}
+
+	r = gfx_v9_4_2_run_shader(adev,
+			&adev->gfx.compute_ring[1],
+			&disp_ibs[1],
+			sgpr96_init_compute_shader_aldebaran,
+			sizeof(sgpr96_init_compute_shader_aldebaran),
+			sgpr96_init_regs_aldebaran,
+			ARRAY_SIZE(sgpr96_init_regs_aldebaran),
+			adev->gfx.cu_info.number * 2,
+			wb_ib.gpu_addr, pattern[1], &fences[1]);
+	if (r) {
+		dev_err(adev->dev, "failed to clear next 576 sgprs\n");
+		goto disp0_failed;
+	}
+
+	r = gfx_v9_4_2_wait_for_waves_assigned(adev,
+			&wb_ib.ptr[1], 0b11111100,
+			pattern[1], adev->gfx.cu_info.number * SIMD_ID_MAX * 6,
+			true);
+	if (r) {
+		dev_err(adev->dev, "wave coverage failed when clear first 576 sgprs\n");
+		wb_ib.ptr[0] = 0xdeadbeaf; /* stop waves */
+		goto disp1_failed;
+	}
+
+	wb_ib.ptr[0] = 0xdeadbeaf; /* stop waves */
+
+	/* wait for the GPU to finish processing the IB */
+	r = dma_fence_wait(fences[0], false);
+	if (r) {
+		dev_err(adev->dev, "timeout to clear first 224 sgprs\n");
+		goto disp1_failed;
+	}
+
+	r = dma_fence_wait(fences[1], false);
+	if (r) {
+		dev_err(adev->dev, "timeout to clear first 576 sgprs\n");
+		goto disp1_failed;
+	}
+
+	memset(wb_ib.ptr, 0, (1 + wb_size) * sizeof(uint32_t));
+	r = gfx_v9_4_2_run_shader(adev,
+			&adev->gfx.compute_ring[0],
+			&disp_ibs[2],
+			sgpr64_init_compute_shader_aldebaran,
+			sizeof(sgpr64_init_compute_shader_aldebaran),
+			sgpr64_init_regs_aldebaran,
+			ARRAY_SIZE(sgpr64_init_regs_aldebaran),
+			adev->gfx.cu_info.number,
+			wb_ib.gpu_addr, pattern[2], &fences[2]);
+	if (r) {
+		dev_err(adev->dev, "failed to clear first 256 sgprs\n");
+		goto disp1_failed;
+	}
+
+	r = gfx_v9_4_2_wait_for_waves_assigned(adev,
+			&wb_ib.ptr[1], 0b1111,
+			pattern[2],
+			adev->gfx.cu_info.number * SIMD_ID_MAX * 4,
+			true);
+	if (r) {
+		dev_err(adev->dev, "wave coverage failed when clear first 256 sgprs\n");
+		wb_ib.ptr[0] = 0xdeadbeaf; /* stop waves */
+		goto disp2_failed;
+	}
+
+	wb_ib.ptr[0] = 0xdeadbeaf; /* stop waves */
+
+	r = dma_fence_wait(fences[2], false);
+	if (r) {
+		dev_err(adev->dev, "timeout to clear first 256 sgprs\n");
+		goto disp2_failed;
+	}
+
+disp2_failed:
+	amdgpu_ib_free(&disp_ibs[2], NULL);
+	dma_fence_put(fences[2]);
+disp1_failed:
+	amdgpu_ib_free(&disp_ibs[1], NULL);
+	dma_fence_put(fences[1]);
+disp0_failed:
+	amdgpu_ib_free(&disp_ibs[0], NULL);
+	dma_fence_put(fences[0]);
+pro_end:
+	amdgpu_ib_free(&wb_ib, NULL);
+
+	if (r)
+		dev_info(adev->dev, "Init SGPRS Failed\n");
+	else
+		dev_info(adev->dev, "Init SGPRS Successfully\n");
+
+	return r;
+}
+
+static int gfx_v9_4_2_do_vgprs_init(struct amdgpu_device *adev)
+{
+	int r;
+	/* CU_ID: 0~15, SIMD_ID: 0~3, WAVE_ID: 0 ~ 9 */
+	int wb_size = adev->gfx.config.max_shader_engines *
+			 CU_ID_MAX * SIMD_ID_MAX * WAVE_ID_MAX;
+	struct amdgpu_ib wb_ib;
+	struct amdgpu_ib disp_ib;
+	struct dma_fence *fence;
+	u32 pattern = 0xa;
+
+	/* bail if the compute ring is not ready */
+	if (!adev->gfx.compute_ring[0].sched.ready)
+		return 0;
+
+	/* allocate the write-back buffer from IB */
+	memset(&wb_ib, 0, sizeof(wb_ib));
+	r = amdgpu_ib_get(adev, NULL, (1 + wb_size) * sizeof(uint32_t),
+			  AMDGPU_IB_POOL_DIRECT, &wb_ib);
+	if (r) {
+		dev_err(adev->dev, "failed to get ib (%d) for wb.\n", r);
+		return r;
+	}
+	memset(wb_ib.ptr, 0, (1 + wb_size) * sizeof(uint32_t));
+
+	r = gfx_v9_4_2_run_shader(adev,
+			&adev->gfx.compute_ring[0],
+			&disp_ib,
+			vgpr_init_compute_shader_aldebaran,
+			sizeof(vgpr_init_compute_shader_aldebaran),
+			vgpr_init_regs_aldebaran,
+			ARRAY_SIZE(vgpr_init_regs_aldebaran),
+			adev->gfx.cu_info.number,
+			wb_ib.gpu_addr, pattern, &fence);
+	if (r) {
+		dev_err(adev->dev, "failed to clear vgprs\n");
+		goto pro_end;
+	}
+
+	/* wait for the GPU to finish processing the IB */
+	r = dma_fence_wait(fence, false);
+	if (r) {
+		dev_err(adev->dev, "timeout to clear vgprs\n");
+		goto disp_failed;
+	}
+
+	r = gfx_v9_4_2_wait_for_waves_assigned(adev,
+			&wb_ib.ptr[1], 0b1,
+			pattern,
+			adev->gfx.cu_info.number * SIMD_ID_MAX,
+			false);
+	if (r) {
+		dev_err(adev->dev, "failed to cover all simds when clearing vgprs\n");
+		goto disp_failed;
+	}
+
+disp_failed:
+	amdgpu_ib_free(&disp_ib, NULL);
+	dma_fence_put(fence);
+pro_end:
+	amdgpu_ib_free(&wb_ib, NULL);
+
+	if (r)
+		dev_info(adev->dev, "Init VGPRS Failed\n");
+	else
+		dev_info(adev->dev, "Init VGPRS Successfully\n");
+
+	return r;
+}
+
+int gfx_v9_4_2_do_edc_gpr_workarounds(struct amdgpu_device *adev)
+{
+	/* only support when RAS is enabled */
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
+		return 0;
+
+	/* Workaround for ALDEBARAN, skip GPRs init in GPU reset.
+	   Will remove it once GPRs init algorithm works for all CU settings. */
+	if (amdgpu_in_reset(adev))
+		return 0;
+
+	gfx_v9_4_2_do_sgprs_init(adev);
+
+	gfx_v9_4_2_do_vgprs_init(adev);
+
+	return 0;
+}
+
+static void gfx_v9_4_2_query_sq_timeout_status(struct amdgpu_device *adev);
+static void gfx_v9_4_2_reset_sq_timeout_status(struct amdgpu_device *adev);
+
+void gfx_v9_4_2_init_golden_registers(struct amdgpu_device *adev,
+				      uint32_t die_id)
+{
+	soc15_program_register_sequence(adev,
+					golden_settings_gc_9_4_2_alde,
+					ARRAY_SIZE(golden_settings_gc_9_4_2_alde));
+
+	/* apply golden settings per die */
+	switch (die_id) {
+	case 0:
+		soc15_program_register_sequence(adev,
+				golden_settings_gc_9_4_2_alde_die_0,
+				ARRAY_SIZE(golden_settings_gc_9_4_2_alde_die_0));
+		break;
+	case 1:
+		soc15_program_register_sequence(adev,
+				golden_settings_gc_9_4_2_alde_die_1,
+				ARRAY_SIZE(golden_settings_gc_9_4_2_alde_die_1));
+		break;
+	default:
+		dev_warn(adev->dev,
+			 "invalid die id %d, ignore channel fabricid remap settings\n",
+			 die_id);
+		break;
+	}
+}
+
+void gfx_v9_4_2_init_sq(struct amdgpu_device *adev)
+{
+	uint32_t data;
+
+	if (adev->gfx.mec_fw_version >= 98) {
+		adev->gmc.xnack_flags |= AMDGPU_GMC_XNACK_FLAG_CHAIN;
+		data = RREG32_SOC15(GC, 0, regSQ_CONFIG1);
+		data = REG_SET_FIELD(data, SQ_CONFIG1, DISABLE_XNACK_CHECK_IN_RETRY_DISABLE, 1);
+		WREG32_SOC15(GC, 0, regSQ_CONFIG1, data);
+	}
+}
+
+void gfx_v9_4_2_debug_trap_config_init(struct amdgpu_device *adev,
+				uint32_t first_vmid,
+				uint32_t last_vmid)
+{
+	uint32_t data;
+	int i;
+
+	mutex_lock(&adev->srbm_mutex);
+
+	for (i = first_vmid; i < last_vmid; i++) {
+		data = 0;
+		soc15_grbm_select(adev, 0, 0, 0, i, 0);
+		data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, TRAP_EN, 1);
+		data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_EN, 0);
+		data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_REPLACE,
+					0);
+		WREG32(SOC15_REG_OFFSET(GC, 0, regSPI_GDBG_PER_VMID_CNTL), data);
+	}
+
+	soc15_grbm_select(adev, 0, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, regSPI_GDBG_TRAP_DATA0), 0);
+	WREG32(SOC15_REG_OFFSET(GC, 0, regSPI_GDBG_TRAP_DATA1), 0);
+}
+
+void gfx_v9_4_2_set_power_brake_sequence(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
+
+	tmp = 0;
+	tmp = REG_SET_FIELD(tmp, GC_THROTTLE_CTRL, PATTERN_MODE, 1);
+	WREG32_SOC15(GC, 0, regGC_THROTTLE_CTRL, tmp);
+
+	tmp = 0;
+	tmp = REG_SET_FIELD(tmp, GC_THROTTLE_CTRL1, PWRBRK_STALL_EN, 1);
+	WREG32_SOC15(GC, 0, regGC_THROTTLE_CTRL1, tmp);
+
+	WREG32_SOC15(GC, 0, regGC_CAC_IND_INDEX, ixPWRBRK_STALL_PATTERN_CTRL);
+	tmp = 0;
+	tmp = REG_SET_FIELD(tmp, PWRBRK_STALL_PATTERN_CTRL, PWRBRK_END_STEP, 0x12);
+	WREG32_SOC15(GC, 0, regGC_CAC_IND_DATA, tmp);
+}
+
+static const struct soc15_reg_entry gfx_v9_4_2_edc_counter_regs[] = {
+	/* CPF */
+	{ SOC15_REG_ENTRY(GC, 0, regCPF_EDC_ROQ_CNT), 0, 1, 1 },
+	{ SOC15_REG_ENTRY(GC, 0, regCPF_EDC_TAG_CNT), 0, 1, 1 },
+	/* CPC */
+	{ SOC15_REG_ENTRY(GC, 0, regCPC_EDC_SCRATCH_CNT), 0, 1, 1 },
+	{ SOC15_REG_ENTRY(GC, 0, regCPC_EDC_UCODE_CNT), 0, 1, 1 },
+	{ SOC15_REG_ENTRY(GC, 0, regDC_EDC_STATE_CNT), 0, 1, 1 },
+	{ SOC15_REG_ENTRY(GC, 0, regDC_EDC_CSINVOC_CNT), 0, 1, 1 },
+	{ SOC15_REG_ENTRY(GC, 0, regDC_EDC_RESTORE_CNT), 0, 1, 1 },
+	/* GDS */
+	{ SOC15_REG_ENTRY(GC, 0, regGDS_EDC_CNT), 0, 1, 1 },
+	{ SOC15_REG_ENTRY(GC, 0, regGDS_EDC_GRBM_CNT), 0, 1, 1 },
+	{ SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 1, 1 },
+	{ SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_PHY_CNT), 0, 1, 1 },
+	{ SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_PIPE_CNT), 0, 1, 1 },
+	/* RLC */
+	{ SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT), 0, 1, 1 },
+	{ SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT2), 0, 1, 1 },
+	/* SPI */
+	{ SOC15_REG_ENTRY(GC, 0, regSPI_EDC_CNT), 0, 8, 1 },
+	/* SQC */
+	{ SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT), 0, 8, 7 },
+	{ SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT2), 0, 8, 7 },
+	{ SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT3), 0, 8, 7 },
+	{ SOC15_REG_ENTRY(GC, 0, regSQC_EDC_PARITY_CNT3), 0, 8, 7 },
+	/* SQ */
+	{ SOC15_REG_ENTRY(GC, 0, regSQ_EDC_CNT), 0, 8, 14 },
+	/* TCP */
+	{ SOC15_REG_ENTRY(GC, 0, regTCP_EDC_CNT_NEW), 0, 8, 14 },
+	/* TCI */
+	{ SOC15_REG_ENTRY(GC, 0, regTCI_EDC_CNT), 0, 1, 69 },
+	/* TCC */
+	{ SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT), 0, 1, 16 },
+	{ SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT2), 0, 1, 16 },
+	/* TCA */
+	{ SOC15_REG_ENTRY(GC, 0, regTCA_EDC_CNT), 0, 1, 2 },
+	/* TCX */
+	{ SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT), 0, 1, 2 },
+	{ SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT2), 0, 1, 2 },
+	/* TD */
+	{ SOC15_REG_ENTRY(GC, 0, regTD_EDC_CNT), 0, 8, 14 },
+	/* TA */
+	{ SOC15_REG_ENTRY(GC, 0, regTA_EDC_CNT), 0, 8, 14 },
+	/* GCEA */
+	{ SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT), 0, 1, 16 },
+	{ SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT2), 0, 1, 16 },
+	{ SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3), 0, 1, 16 },
+};
+
+static void gfx_v9_4_2_select_se_sh(struct amdgpu_device *adev, u32 se_num,
+				  u32 sh_num, u32 instance)
+{
+	u32 data;
+
+	if (instance == 0xffffffff)
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX,
+				     INSTANCE_BROADCAST_WRITES, 1);
+	else
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_INDEX,
+				     instance);
+
+	if (se_num == 0xffffffff)
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_BROADCAST_WRITES,
+				     1);
+	else
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_INDEX, se_num);
+
+	if (sh_num == 0xffffffff)
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_BROADCAST_WRITES,
+				     1);
+	else
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_INDEX, sh_num);
+
+	WREG32_SOC15_RLC_SHADOW_EX(reg, GC, 0, regGRBM_GFX_INDEX, data);
+}
+
+static const struct soc15_ras_field_entry gfx_v9_4_2_ras_fields[] = {
+	/* CPF */
+	{ "CPF_ROQ_ME2", SOC15_REG_ENTRY(GC, 0, regCPF_EDC_ROQ_CNT),
+	  SOC15_REG_FIELD(CPF_EDC_ROQ_CNT, SEC_COUNT_ME2),
+	  SOC15_REG_FIELD(CPF_EDC_ROQ_CNT, DED_COUNT_ME2) },
+	{ "CPF_ROQ_ME1", SOC15_REG_ENTRY(GC, 0, regCPF_EDC_ROQ_CNT),
+	  SOC15_REG_FIELD(CPF_EDC_ROQ_CNT, SEC_COUNT_ME1),
+	  SOC15_REG_FIELD(CPF_EDC_ROQ_CNT, DED_COUNT_ME1) },
+	{ "CPF_TCIU_TAG", SOC15_REG_ENTRY(GC, 0, regCPF_EDC_TAG_CNT),
+	  SOC15_REG_FIELD(CPF_EDC_TAG_CNT, SEC_COUNT),
+	  SOC15_REG_FIELD(CPF_EDC_TAG_CNT, DED_COUNT) },
+
+	/* CPC */
+	{ "CPC_SCRATCH", SOC15_REG_ENTRY(GC, 0, regCPC_EDC_SCRATCH_CNT),
+	  SOC15_REG_FIELD(CPC_EDC_SCRATCH_CNT, SEC_COUNT),
+	  SOC15_REG_FIELD(CPC_EDC_SCRATCH_CNT, DED_COUNT) },
+	{ "CPC_UCODE", SOC15_REG_ENTRY(GC, 0, regCPC_EDC_UCODE_CNT),
+	  SOC15_REG_FIELD(CPC_EDC_UCODE_CNT, SEC_COUNT),
+	  SOC15_REG_FIELD(CPC_EDC_UCODE_CNT, DED_COUNT) },
+	{ "CPC_DC_STATE_RAM_ME1", SOC15_REG_ENTRY(GC, 0, regDC_EDC_STATE_CNT),
+	  SOC15_REG_FIELD(DC_EDC_STATE_CNT, SEC_COUNT_ME1),
+	  SOC15_REG_FIELD(DC_EDC_STATE_CNT, DED_COUNT_ME1) },
+	{ "CPC_DC_CSINVOC_RAM_ME1",
+	  SOC15_REG_ENTRY(GC, 0, regDC_EDC_CSINVOC_CNT),
+	  SOC15_REG_FIELD(DC_EDC_CSINVOC_CNT, SEC_COUNT_ME1),
+	  SOC15_REG_FIELD(DC_EDC_CSINVOC_CNT, DED_COUNT_ME1) },
+	{ "CPC_DC_RESTORE_RAM_ME1",
+	  SOC15_REG_ENTRY(GC, 0, regDC_EDC_RESTORE_CNT),
+	  SOC15_REG_FIELD(DC_EDC_RESTORE_CNT, SEC_COUNT_ME1),
+	  SOC15_REG_FIELD(DC_EDC_RESTORE_CNT, DED_COUNT_ME1) },
+	{ "CPC_DC_CSINVOC_RAM1_ME1",
+	  SOC15_REG_ENTRY(GC, 0, regDC_EDC_CSINVOC_CNT),
+	  SOC15_REG_FIELD(DC_EDC_CSINVOC_CNT, SEC_COUNT1_ME1),
+	  SOC15_REG_FIELD(DC_EDC_CSINVOC_CNT, DED_COUNT1_ME1) },
+	{ "CPC_DC_RESTORE_RAM1_ME1",
+	  SOC15_REG_ENTRY(GC, 0, regDC_EDC_RESTORE_CNT),
+	  SOC15_REG_FIELD(DC_EDC_RESTORE_CNT, SEC_COUNT1_ME1),
+	  SOC15_REG_FIELD(DC_EDC_RESTORE_CNT, DED_COUNT1_ME1) },
+
+	/* GDS */
+	{ "GDS_GRBM", SOC15_REG_ENTRY(GC, 0, regGDS_EDC_GRBM_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_GRBM_CNT, SEC),
+	  SOC15_REG_FIELD(GDS_EDC_GRBM_CNT, DED) },
+	{ "GDS_MEM", SOC15_REG_ENTRY(GC, 0, regGDS_EDC_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_CNT, GDS_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_CNT, GDS_MEM_DED) },
+	{ "GDS_PHY_CMD_RAM_MEM", SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_PHY_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, PHY_CMD_RAM_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, PHY_CMD_RAM_MEM_DED) },
+	{ "GDS_PHY_DATA_RAM_MEM", SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_PHY_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, PHY_DATA_RAM_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, PHY_DATA_RAM_MEM_DED) },
+	{ "GDS_ME0_CS_PIPE_MEM", SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_PHY_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, ME0_CS_PIPE_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, ME0_CS_PIPE_MEM_DED) },
+	{ "GDS_ME1_PIPE0_PIPE_MEM",
+	  SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_PIPE_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE0_PIPE_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE0_PIPE_MEM_DED) },
+	{ "GDS_ME1_PIPE1_PIPE_MEM",
+	  SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_PIPE_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE1_PIPE_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE1_PIPE_MEM_DED) },
+	{ "GDS_ME1_PIPE2_PIPE_MEM",
+	  SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_PIPE_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE2_PIPE_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE2_PIPE_MEM_DED) },
+	{ "GDS_ME1_PIPE3_PIPE_MEM",
+	  SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_PIPE_CNT),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE3_PIPE_MEM_SEC),
+	  SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE3_PIPE_MEM_DED) },
+	{ "GDS_ME0_GFXHP3D_PIX_DED",
+	  SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0,
+	  SOC15_REG_FIELD(GDS_EDC_OA_DED, ME0_GFXHP3D_PIX_DED) },
+	{ "GDS_ME0_GFXHP3D_VTX_DED",
+	  SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0,
+	  SOC15_REG_FIELD(GDS_EDC_OA_DED, ME0_GFXHP3D_VTX_DED) },
+	{ "GDS_ME0_CS_DED",
+	  SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0,
+	  SOC15_REG_FIELD(GDS_EDC_OA_DED, ME0_CS_DED) },
+	{ "GDS_ME0_GFXHP3D_GS_DED",
+	  SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0,
+	  SOC15_REG_FIELD(GDS_EDC_OA_DED, ME0_GFXHP3D_GS_DED) },
+	{ "GDS_ME1_PIPE0_DED",
+	  SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0,
+	  SOC15_REG_FIELD(GDS_EDC_OA_DED, ME1_PIPE0_DED) },
+	{ "GDS_ME1_PIPE1_DED",
+	  SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0,
+	  SOC15_REG_FIELD(GDS_EDC_OA_DED, ME1_PIPE1_DED) },
+	{ "GDS_ME1_PIPE2_DED",
+	  SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0,
+	  SOC15_REG_FIELD(GDS_EDC_OA_DED, ME1_PIPE2_DED) },
+	{ "GDS_ME1_PIPE3_DED",
+	  SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0,
+	  SOC15_REG_FIELD(GDS_EDC_OA_DED, ME1_PIPE3_DED) },
+	{ "GDS_ME2_PIPE0_DED",
+	  SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0,
+	  SOC15_REG_FIELD(GDS_EDC_OA_DED, ME2_PIPE0_DED) },
+	{ "GDS_ME2_PIPE1_DED",
+	  SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0,
+	  SOC15_REG_FIELD(GDS_EDC_OA_DED, ME2_PIPE1_DED) },
+	{ "GDS_ME2_PIPE2_DED",
+	  SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0,
+	  SOC15_REG_FIELD(GDS_EDC_OA_DED, ME2_PIPE2_DED) },
+	{ "GDS_ME2_PIPE3_DED",
+	  SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0,
+	  SOC15_REG_FIELD(GDS_EDC_OA_DED, ME2_PIPE3_DED) },
+
+	/* RLC */
+	{ "RLCG_INSTR_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLCG_INSTR_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLCG_INSTR_RAM_DED_COUNT) },
+	{ "RLCG_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLCG_SCRATCH_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLCG_SCRATCH_RAM_DED_COUNT) },
+	{ "RLCV_INSTR_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLCV_INSTR_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLCV_INSTR_RAM_DED_COUNT) },
+	{ "RLCV_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLCV_SCRATCH_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLCV_SCRATCH_RAM_DED_COUNT) },
+	{ "RLC_TCTAG_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLC_TCTAG_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLC_TCTAG_RAM_DED_COUNT) },
+	{ "RLC_SPM_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SPM_SCRATCH_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SPM_SCRATCH_RAM_DED_COUNT) },
+	{ "RLC_SRM_DATA_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SRM_DATA_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SRM_DATA_RAM_DED_COUNT) },
+	{ "RLC_SRM_ADDR_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SRM_ADDR_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SRM_ADDR_RAM_DED_COUNT) },
+	{ "RLC_SPM_SE0_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT2),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE0_SCRATCH_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE0_SCRATCH_RAM_DED_COUNT) },
+	{ "RLC_SPM_SE1_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT2),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE1_SCRATCH_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE1_SCRATCH_RAM_DED_COUNT) },
+	{ "RLC_SPM_SE2_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT2),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE2_SCRATCH_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE2_SCRATCH_RAM_DED_COUNT) },
+	{ "RLC_SPM_SE3_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT2),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE3_SCRATCH_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE3_SCRATCH_RAM_DED_COUNT) },
+	{ "RLC_SPM_SE4_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT2),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE4_SCRATCH_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE4_SCRATCH_RAM_DED_COUNT) },
+	{ "RLC_SPM_SE5_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT2),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE5_SCRATCH_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE5_SCRATCH_RAM_DED_COUNT) },
+	{ "RLC_SPM_SE6_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT2),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE6_SCRATCH_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE6_SCRATCH_RAM_DED_COUNT) },
+	{ "RLC_SPM_SE7_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT2),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE7_SCRATCH_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE7_SCRATCH_RAM_DED_COUNT) },
+
+	/* SPI */
+	{ "SPI_SR_MEM", SOC15_REG_ENTRY(GC, 0, regSPI_EDC_CNT),
+	  SOC15_REG_FIELD(SPI_EDC_CNT, SPI_SR_MEM_SEC_COUNT),
+	  SOC15_REG_FIELD(SPI_EDC_CNT, SPI_SR_MEM_DED_COUNT) },
+	{ "SPI_GDS_EXPREQ", SOC15_REG_ENTRY(GC, 0, regSPI_EDC_CNT),
+	  SOC15_REG_FIELD(SPI_EDC_CNT, SPI_GDS_EXPREQ_SEC_COUNT),
+	  SOC15_REG_FIELD(SPI_EDC_CNT, SPI_GDS_EXPREQ_DED_COUNT) },
+	{ "SPI_WB_GRANT_30", SOC15_REG_ENTRY(GC, 0, regSPI_EDC_CNT),
+	  SOC15_REG_FIELD(SPI_EDC_CNT, SPI_WB_GRANT_30_SEC_COUNT),
+	  SOC15_REG_FIELD(SPI_EDC_CNT, SPI_WB_GRANT_30_DED_COUNT) },
+	{ "SPI_LIFE_CNT", SOC15_REG_ENTRY(GC, 0, regSPI_EDC_CNT),
+	  SOC15_REG_FIELD(SPI_EDC_CNT, SPI_LIFE_CNT_SEC_COUNT),
+	  SOC15_REG_FIELD(SPI_EDC_CNT, SPI_LIFE_CNT_DED_COUNT) },
+
+	/* SQC - regSQC_EDC_CNT */
+	{ "SQC_DATA_CU0_WRITE_DATA_BUF", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU0_WRITE_DATA_BUF_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU0_WRITE_DATA_BUF_DED_COUNT) },
+	{ "SQC_DATA_CU0_UTCL1_LFIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU0_UTCL1_LFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU0_UTCL1_LFIFO_DED_COUNT) },
+	{ "SQC_DATA_CU1_WRITE_DATA_BUF", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU1_WRITE_DATA_BUF_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU1_WRITE_DATA_BUF_DED_COUNT) },
+	{ "SQC_DATA_CU1_UTCL1_LFIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU1_UTCL1_LFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU1_UTCL1_LFIFO_DED_COUNT) },
+	{ "SQC_DATA_CU2_WRITE_DATA_BUF", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU2_WRITE_DATA_BUF_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU2_WRITE_DATA_BUF_DED_COUNT) },
+	{ "SQC_DATA_CU2_UTCL1_LFIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU2_UTCL1_LFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU2_UTCL1_LFIFO_DED_COUNT) },
+	{ "SQC_DATA_CU3_WRITE_DATA_BUF", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU3_WRITE_DATA_BUF_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU3_WRITE_DATA_BUF_DED_COUNT) },
+	{ "SQC_DATA_CU3_UTCL1_LFIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU3_UTCL1_LFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU3_UTCL1_LFIFO_DED_COUNT) },
+
+	/* SQC - regSQC_EDC_CNT2 */
+	{ "SQC_INST_BANKA_TAG_RAM", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT2),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, INST_BANKA_TAG_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, INST_BANKA_TAG_RAM_DED_COUNT) },
+	{ "SQC_INST_BANKA_BANK_RAM", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT2),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, INST_BANKA_BANK_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, INST_BANKA_BANK_RAM_DED_COUNT) },
+	{ "SQC_DATA_BANKA_TAG_RAM", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT2),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_TAG_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_TAG_RAM_DED_COUNT) },
+	{ "SQC_DATA_BANKA_BANK_RAM", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT2),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_BANK_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_BANK_RAM_DED_COUNT) },
+	{ "SQC_INST_UTCL1_LFIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT2),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, INST_UTCL1_LFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, INST_UTCL1_LFIFO_DED_COUNT) },
+	{ "SQC_DATA_BANKA_DIRTY_BIT_RAM", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT2),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_DIRTY_BIT_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_DIRTY_BIT_RAM_DED_COUNT) },
+
+	/* SQC - regSQC_EDC_CNT3 */
+	{ "SQC_INST_BANKB_TAG_RAM", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, INST_BANKB_TAG_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, INST_BANKB_TAG_RAM_DED_COUNT) },
+	{ "SQC_INST_BANKB_BANK_RAM", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, INST_BANKB_BANK_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, INST_BANKB_BANK_RAM_DED_COUNT) },
+	{ "SQC_DATA_BANKB_TAG_RAM", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_TAG_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_TAG_RAM_DED_COUNT) },
+	{ "SQC_DATA_BANKB_BANK_RAM", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_BANK_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_BANK_RAM_DED_COUNT) },
+	{ "SQC_DATA_BANKB_DIRTY_BIT_RAM", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_DIRTY_BIT_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_DIRTY_BIT_RAM_DED_COUNT) },
+
+	/* SQC - regSQC_EDC_PARITY_CNT3 */
+	{ "SQC_INST_BANKA_UTCL1_MISS_FIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_PARITY_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, INST_BANKA_UTCL1_MISS_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, INST_BANKA_UTCL1_MISS_FIFO_DED_COUNT) },
+	{ "SQC_INST_BANKA_MISS_FIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_PARITY_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, INST_BANKA_MISS_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, INST_BANKA_MISS_FIFO_DED_COUNT) },
+	{ "SQC_DATA_BANKA_HIT_FIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_PARITY_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKA_HIT_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKA_HIT_FIFO_DED_COUNT) },
+	{ "SQC_DATA_BANKA_MISS_FIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_PARITY_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKA_MISS_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKA_MISS_FIFO_DED_COUNT) },
+	{ "SQC_INST_BANKB_UTCL1_MISS_FIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_PARITY_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, INST_BANKB_UTCL1_MISS_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, INST_BANKB_UTCL1_MISS_FIFO_DED_COUNT) },
+	{ "SQC_INST_BANKB_MISS_FIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_PARITY_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, INST_BANKB_MISS_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, INST_BANKB_MISS_FIFO_DED_COUNT) },
+	{ "SQC_DATA_BANKB_HIT_FIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_PARITY_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKB_HIT_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKB_HIT_FIFO_DED_COUNT) },
+	{ "SQC_DATA_BANKB_MISS_FIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_PARITY_CNT3),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKB_MISS_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKB_MISS_FIFO_DED_COUNT) },
+
+	/* SQ */
+	{ "SQ_LDS_D", SOC15_REG_ENTRY(GC, 0, regSQ_EDC_CNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, LDS_D_SEC_COUNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, LDS_D_DED_COUNT) },
+	{ "SQ_LDS_I", SOC15_REG_ENTRY(GC, 0, regSQ_EDC_CNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, LDS_I_SEC_COUNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, LDS_I_DED_COUNT) },
+	{ "SQ_SGPR", SOC15_REG_ENTRY(GC, 0, regSQ_EDC_CNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, SGPR_SEC_COUNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, SGPR_DED_COUNT) },
+	{ "SQ_VGPR0", SOC15_REG_ENTRY(GC, 0, regSQ_EDC_CNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR0_SEC_COUNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR0_DED_COUNT) },
+	{ "SQ_VGPR1", SOC15_REG_ENTRY(GC, 0, regSQ_EDC_CNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR1_SEC_COUNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR1_DED_COUNT) },
+	{ "SQ_VGPR2", SOC15_REG_ENTRY(GC, 0, regSQ_EDC_CNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR2_SEC_COUNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR2_DED_COUNT) },
+	{ "SQ_VGPR3", SOC15_REG_ENTRY(GC, 0, regSQ_EDC_CNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR3_SEC_COUNT),
+	  SOC15_REG_FIELD(SQ_EDC_CNT, VGPR3_DED_COUNT) },
+
+	/* TCP */
+	{ "TCP_CACHE_RAM", SOC15_REG_ENTRY(GC, 0, regTCP_EDC_CNT_NEW),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, CACHE_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, CACHE_RAM_DED_COUNT) },
+	{ "TCP_LFIFO_RAM", SOC15_REG_ENTRY(GC, 0, regTCP_EDC_CNT_NEW),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, LFIFO_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, LFIFO_RAM_DED_COUNT) },
+	{ "TCP_CMD_FIFO", SOC15_REG_ENTRY(GC, 0, regTCP_EDC_CNT_NEW),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, CMD_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, CMD_FIFO_DED_COUNT) },
+	{ "TCP_VM_FIFO", SOC15_REG_ENTRY(GC, 0, regTCP_EDC_CNT_NEW),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, VM_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, VM_FIFO_DED_COUNT) },
+	{ "TCP_DB_RAM", SOC15_REG_ENTRY(GC, 0, regTCP_EDC_CNT_NEW),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, DB_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, DB_RAM_DED_COUNT) },
+	{ "TCP_UTCL1_LFIFO0", SOC15_REG_ENTRY(GC, 0, regTCP_EDC_CNT_NEW),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, UTCL1_LFIFO0_SEC_COUNT),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, UTCL1_LFIFO0_DED_COUNT) },
+	{ "TCP_UTCL1_LFIFO1", SOC15_REG_ENTRY(GC, 0, regTCP_EDC_CNT_NEW),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, UTCL1_LFIFO1_SEC_COUNT),
+	  SOC15_REG_FIELD(TCP_EDC_CNT_NEW, UTCL1_LFIFO1_DED_COUNT) },
+
+	/* TCI */
+	{ "TCI_WRITE_RAM", SOC15_REG_ENTRY(GC, 0, regTCI_EDC_CNT),
+	  SOC15_REG_FIELD(TCI_EDC_CNT, WRITE_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(TCI_EDC_CNT, WRITE_RAM_DED_COUNT) },
+
+	/* TCC */
+	{ "TCC_CACHE_DATA", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, CACHE_DATA_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, CACHE_DATA_DED_COUNT) },
+	{ "TCC_CACHE_DIRTY", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, CACHE_DIRTY_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, CACHE_DIRTY_DED_COUNT) },
+	{ "TCC_HIGH_RATE_TAG", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, HIGH_RATE_TAG_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, HIGH_RATE_TAG_DED_COUNT) },
+	{ "TCC_LOW_RATE_TAG", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, LOW_RATE_TAG_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, LOW_RATE_TAG_DED_COUNT) },
+	{ "TCC_SRC_FIFO", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, SRC_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, SRC_FIFO_DED_COUNT) },
+	{ "TCC_LATENCY_FIFO", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, LATENCY_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, LATENCY_FIFO_DED_COUNT) },
+	{ "TCC_LATENCY_FIFO_NEXT_RAM", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, LATENCY_FIFO_NEXT_RAM_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT, LATENCY_FIFO_NEXT_RAM_DED_COUNT) },
+	{ "TCC_CACHE_TAG_PROBE_FIFO", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, CACHE_TAG_PROBE_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, CACHE_TAG_PROBE_FIFO_DED_COUNT) },
+	{ "TCC_UC_ATOMIC_FIFO", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, UC_ATOMIC_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, UC_ATOMIC_FIFO_DED_COUNT) },
+	{ "TCC_WRITE_CACHE_READ", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, WRITE_CACHE_READ_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, WRITE_CACHE_READ_DED_COUNT) },
+	{ "TCC_RETURN_CONTROL", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, RETURN_CONTROL_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, RETURN_CONTROL_DED_COUNT) },
+	{ "TCC_IN_USE_TRANSFER", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, IN_USE_TRANSFER_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, IN_USE_TRANSFER_DED_COUNT) },
+	{ "TCC_IN_USE_DEC", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, IN_USE_DEC_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, IN_USE_DEC_DED_COUNT) },
+	{ "TCC_WRITE_RETURN", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, WRITE_RETURN_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, WRITE_RETURN_DED_COUNT) },
+	{ "TCC_RETURN_DATA", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT2),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, RETURN_DATA_SEC_COUNT),
+	  SOC15_REG_FIELD(TCC_EDC_CNT2, RETURN_DATA_DED_COUNT) },
+
+	/* TCA */
+	{ "TCA_HOLE_FIFO", SOC15_REG_ENTRY(GC, 0, regTCA_EDC_CNT),
+	  SOC15_REG_FIELD(TCA_EDC_CNT, HOLE_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TCA_EDC_CNT, HOLE_FIFO_DED_COUNT) },
+	{ "TCA_REQ_FIFO", SOC15_REG_ENTRY(GC, 0, regTCA_EDC_CNT),
+	  SOC15_REG_FIELD(TCA_EDC_CNT, REQ_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TCA_EDC_CNT, REQ_FIFO_DED_COUNT) },
+
+	/* TCX */
+	{ "TCX_GROUP0", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT),
+	  SOC15_REG_FIELD(TCX_EDC_CNT, GROUP0_SEC_COUNT),
+	  SOC15_REG_FIELD(TCX_EDC_CNT, GROUP0_DED_COUNT) },
+	{ "TCX_GROUP1", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT),
+	  SOC15_REG_FIELD(TCX_EDC_CNT, GROUP1_SEC_COUNT),
+	  SOC15_REG_FIELD(TCX_EDC_CNT, GROUP1_DED_COUNT) },
+	{ "TCX_GROUP2", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT),
+	  SOC15_REG_FIELD(TCX_EDC_CNT, GROUP2_SEC_COUNT),
+	  SOC15_REG_FIELD(TCX_EDC_CNT, GROUP2_DED_COUNT) },
+	{ "TCX_GROUP3", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT),
+	  SOC15_REG_FIELD(TCX_EDC_CNT, GROUP3_SEC_COUNT),
+	  SOC15_REG_FIELD(TCX_EDC_CNT, GROUP3_DED_COUNT) },
+	{ "TCX_GROUP4", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT),
+	  SOC15_REG_FIELD(TCX_EDC_CNT, GROUP4_SEC_COUNT),
+	  SOC15_REG_FIELD(TCX_EDC_CNT, GROUP4_DED_COUNT) },
+	{ "TCX_GROUP5", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT),
+	  SOC15_REG_FIELD(TCX_EDC_CNT, GROUP5_SED_COUNT), 0, 0 },
+	{ "TCX_GROUP6", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT),
+	  SOC15_REG_FIELD(TCX_EDC_CNT, GROUP6_SED_COUNT), 0, 0 },
+	{ "TCX_GROUP7", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT),
+	  SOC15_REG_FIELD(TCX_EDC_CNT, GROUP7_SED_COUNT), 0, 0 },
+	{ "TCX_GROUP8", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT),
+	  SOC15_REG_FIELD(TCX_EDC_CNT, GROUP8_SED_COUNT), 0, 0 },
+	{ "TCX_GROUP9", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT),
+	  SOC15_REG_FIELD(TCX_EDC_CNT, GROUP9_SED_COUNT), 0, 0 },
+	{ "TCX_GROUP10", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT),
+	  SOC15_REG_FIELD(TCX_EDC_CNT, GROUP10_SED_COUNT), 0, 0 },
+	{ "TCX_GROUP11", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT2),
+	  SOC15_REG_FIELD(TCX_EDC_CNT2, GROUP11_SED_COUNT), 0, 0 },
+	{ "TCX_GROUP12", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT2),
+	  SOC15_REG_FIELD(TCX_EDC_CNT2, GROUP12_SED_COUNT), 0, 0 },
+	{ "TCX_GROUP13", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT2),
+	  SOC15_REG_FIELD(TCX_EDC_CNT2, GROUP13_SED_COUNT), 0, 0 },
+	{ "TCX_GROUP14", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT2),
+	  SOC15_REG_FIELD(TCX_EDC_CNT2, GROUP14_SED_COUNT), 0, 0 },
+
+	/* TD */
+	{ "TD_SS_FIFO_LO", SOC15_REG_ENTRY(GC, 0, regTD_EDC_CNT),
+	  SOC15_REG_FIELD(TD_EDC_CNT, SS_FIFO_LO_SEC_COUNT),
+	  SOC15_REG_FIELD(TD_EDC_CNT, SS_FIFO_LO_DED_COUNT) },
+	{ "TD_SS_FIFO_HI", SOC15_REG_ENTRY(GC, 0, regTD_EDC_CNT),
+	  SOC15_REG_FIELD(TD_EDC_CNT, SS_FIFO_HI_SEC_COUNT),
+	  SOC15_REG_FIELD(TD_EDC_CNT, SS_FIFO_HI_DED_COUNT) },
+	{ "TD_CS_FIFO", SOC15_REG_ENTRY(GC, 0, regTD_EDC_CNT),
+	  SOC15_REG_FIELD(TD_EDC_CNT, CS_FIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TD_EDC_CNT, CS_FIFO_DED_COUNT) },
+
+	/* TA */
+	{ "TA_FS_DFIFO", SOC15_REG_ENTRY(GC, 0, regTA_EDC_CNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_DFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_DFIFO_DED_COUNT) },
+	{ "TA_FS_AFIFO_LO", SOC15_REG_ENTRY(GC, 0, regTA_EDC_CNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_AFIFO_LO_SEC_COUNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_AFIFO_LO_DED_COUNT) },
+	{ "TA_FL_LFIFO", SOC15_REG_ENTRY(GC, 0, regTA_EDC_CNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FL_LFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FL_LFIFO_DED_COUNT) },
+	{ "TA_FX_LFIFO", SOC15_REG_ENTRY(GC, 0, regTA_EDC_CNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FX_LFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FX_LFIFO_DED_COUNT) },
+	{ "TA_FS_CFIFO", SOC15_REG_ENTRY(GC, 0, regTA_EDC_CNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_CFIFO_SEC_COUNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_CFIFO_DED_COUNT) },
+	{ "TA_FS_AFIFO_HI", SOC15_REG_ENTRY(GC, 0, regTA_EDC_CNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_AFIFO_HI_SEC_COUNT),
+	  SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_AFIFO_HI_DED_COUNT) },
+
+	/* EA - regGCEA_EDC_CNT */
+	{ "EA_DRAMRD_CMDMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT) },
+	{ "EA_DRAMWR_CMDMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT) },
+	{ "EA_DRAMWR_DATAMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT) },
+	{ "EA_RRET_TAGMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, RRET_TAGMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, RRET_TAGMEM_DED_COUNT) },
+	{ "EA_WRET_TAGMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, WRET_TAGMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, WRET_TAGMEM_DED_COUNT) },
+	{ "EA_IOWR_DATAMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, IOWR_DATAMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, IOWR_DATAMEM_DED_COUNT) },
+	{ "EA_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT), 0, 0 },
+	{ "EA_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT), 0, 0 },
+	{ "EA_IORD_CMDMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, IORD_CMDMEM_SED_COUNT), 0, 0 },
+	{ "EA_IOWR_CMDMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT, IOWR_CMDMEM_SED_COUNT), 0, 0 },
+
+	/* EA - regGCEA_EDC_CNT2 */
+	{ "EA_GMIRD_CMDMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT) },
+	{ "EA_GMIWR_CMDMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT) },
+	{ "EA_GMIWR_DATAMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT) },
+	{ "EA_GMIRD_PAGEMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT), 0, 0 },
+	{ "EA_GMIWR_PAGEMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT), 0, 0 },
+	{ "EA_MAM_D0MEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D0MEM_SED_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D0MEM_DED_COUNT) },
+	{ "EA_MAM_D1MEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D1MEM_SED_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D1MEM_DED_COUNT) },
+	{ "EA_MAM_D2MEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D2MEM_SED_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D2MEM_DED_COUNT) },
+	{ "EA_MAM_D3MEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT2),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D3MEM_SED_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D3MEM_DED_COUNT) },
+
+	/* EA - regGCEA_EDC_CNT3 */
+	{ "EA_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3), 0, 0,
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT) },
+	{ "EA_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3), 0, 0,
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT) },
+	{ "EA_IORD_CMDMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3), 0, 0,
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, IORD_CMDMEM_DED_COUNT) },
+	{ "EA_IOWR_CMDMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3), 0, 0,
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, IOWR_CMDMEM_DED_COUNT) },
+	{ "EA_GMIRD_PAGEMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3), 0, 0,
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT) },
+	{ "EA_GMIWR_PAGEMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3), 0, 0,
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT) },
+	{ "EA_MAM_A0MEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A0MEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A0MEM_DED_COUNT) },
+	{ "EA_MAM_A1MEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A1MEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A1MEM_DED_COUNT) },
+	{ "EA_MAM_A2MEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A2MEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A2MEM_DED_COUNT) },
+	{ "EA_MAM_A3MEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A3MEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A3MEM_DED_COUNT) },
+	{ "EA_MAM_AFMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_AFMEM_SEC_COUNT),
+	  SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_AFMEM_DED_COUNT) },
+};
+
+static const char * const vml2_walker_mems[] = {
+	"UTC_VML2_CACHE_PDE0_MEM0",
+	"UTC_VML2_CACHE_PDE0_MEM1",
+	"UTC_VML2_CACHE_PDE1_MEM0",
+	"UTC_VML2_CACHE_PDE1_MEM1",
+	"UTC_VML2_CACHE_PDE2_MEM0",
+	"UTC_VML2_CACHE_PDE2_MEM1",
+	"UTC_VML2_RDIF_ARADDRS",
+	"UTC_VML2_RDIF_LOG_FIFO",
+	"UTC_VML2_QUEUE_REQ",
+	"UTC_VML2_QUEUE_RET",
+};
+
+static struct gfx_v9_4_2_utc_block gfx_v9_4_2_utc_blocks[] = {
+	{ VML2_MEM, 8, 2, 2,
+	  { SOC15_REG_ENTRY(GC, 0, regVML2_MEM_ECC_INDEX) },
+	  { SOC15_REG_ENTRY(GC, 0, regVML2_MEM_ECC_CNTL) },
+	  SOC15_REG_FIELD(VML2_MEM_ECC_CNTL, SEC_COUNT),
+	  SOC15_REG_FIELD(VML2_MEM_ECC_CNTL, DED_COUNT),
+	  REG_SET_FIELD(0, VML2_MEM_ECC_CNTL, WRITE_COUNTERS, 1) },
+	{ VML2_WALKER_MEM, ARRAY_SIZE(vml2_walker_mems), 1, 1,
+	  { SOC15_REG_ENTRY(GC, 0, regVML2_WALKER_MEM_ECC_INDEX) },
+	  { SOC15_REG_ENTRY(GC, 0, regVML2_WALKER_MEM_ECC_CNTL) },
+	  SOC15_REG_FIELD(VML2_WALKER_MEM_ECC_CNTL, SEC_COUNT),
+	  SOC15_REG_FIELD(VML2_WALKER_MEM_ECC_CNTL, DED_COUNT),
+	  REG_SET_FIELD(0, VML2_WALKER_MEM_ECC_CNTL, WRITE_COUNTERS, 1) },
+	{ UTCL2_MEM, 18, 1, 2,
+	  { SOC15_REG_ENTRY(GC, 0, regUTCL2_MEM_ECC_INDEX) },
+	  { SOC15_REG_ENTRY(GC, 0, regUTCL2_MEM_ECC_CNTL) },
+	  SOC15_REG_FIELD(UTCL2_MEM_ECC_CNTL, SEC_COUNT),
+	  SOC15_REG_FIELD(UTCL2_MEM_ECC_CNTL, DED_COUNT),
+	  REG_SET_FIELD(0, UTCL2_MEM_ECC_CNTL, WRITE_COUNTERS, 1) },
+	{ ATC_L2_CACHE_2M, 8, 2, 1,
+	  { SOC15_REG_ENTRY(GC, 0, regATC_L2_CACHE_2M_DSM_INDEX) },
+	  { SOC15_REG_ENTRY(GC, 0, regATC_L2_CACHE_2M_DSM_CNTL) },
+	  SOC15_REG_FIELD(ATC_L2_CACHE_2M_DSM_CNTL, SEC_COUNT),
+	  SOC15_REG_FIELD(ATC_L2_CACHE_2M_DSM_CNTL, DED_COUNT),
+	  REG_SET_FIELD(0, ATC_L2_CACHE_2M_DSM_CNTL, WRITE_COUNTERS, 1) },
+	{ ATC_L2_CACHE_32K, 8, 2, 2,
+	  { SOC15_REG_ENTRY(GC, 0, regATC_L2_CACHE_32K_DSM_INDEX) },
+	  { SOC15_REG_ENTRY(GC, 0, regATC_L2_CACHE_32K_DSM_CNTL) },
+	  SOC15_REG_FIELD(ATC_L2_CACHE_32K_DSM_CNTL, SEC_COUNT),
+	  SOC15_REG_FIELD(ATC_L2_CACHE_32K_DSM_CNTL, DED_COUNT),
+	  REG_SET_FIELD(0, ATC_L2_CACHE_32K_DSM_CNTL, WRITE_COUNTERS, 1) },
+	{ ATC_L2_CACHE_4K, 8, 2, 8,
+	  { SOC15_REG_ENTRY(GC, 0, regATC_L2_CACHE_4K_DSM_INDEX) },
+	  { SOC15_REG_ENTRY(GC, 0, regATC_L2_CACHE_4K_DSM_CNTL) },
+	  SOC15_REG_FIELD(ATC_L2_CACHE_4K_DSM_CNTL, SEC_COUNT),
+	  SOC15_REG_FIELD(ATC_L2_CACHE_4K_DSM_CNTL, DED_COUNT),
+	  REG_SET_FIELD(0, ATC_L2_CACHE_4K_DSM_CNTL, WRITE_COUNTERS, 1) },
+};
+
+static const struct soc15_reg_entry gfx_v9_4_2_ea_err_status_regs = {
+	SOC15_REG_ENTRY(GC, 0, regGCEA_ERR_STATUS), 0, 1, 16
+};
+
+static int gfx_v9_4_2_get_reg_error_count(struct amdgpu_device *adev,
+					  const struct soc15_reg_entry *reg,
+					  uint32_t se_id, uint32_t inst_id,
+					  uint32_t value, uint32_t *sec_count,
+					  uint32_t *ded_count)
+{
+	uint32_t i;
+	uint32_t sec_cnt, ded_cnt;
+
+	for (i = 0; i < ARRAY_SIZE(gfx_v9_4_2_ras_fields); i++) {
+		if (gfx_v9_4_2_ras_fields[i].reg_offset != reg->reg_offset ||
+		    gfx_v9_4_2_ras_fields[i].seg != reg->seg ||
+		    gfx_v9_4_2_ras_fields[i].inst != reg->inst)
+			continue;
+
+		sec_cnt = SOC15_RAS_REG_FIELD_VAL(
+			value, gfx_v9_4_2_ras_fields[i], sec);
+		if (sec_cnt) {
+			dev_info(adev->dev,
+				 "GFX SubBlock %s, Instance[%d][%d], SEC %d\n",
+				 gfx_v9_4_2_ras_fields[i].name, se_id, inst_id,
+				 sec_cnt);
+			*sec_count += sec_cnt;
+		}
+
+		ded_cnt = SOC15_RAS_REG_FIELD_VAL(
+			value, gfx_v9_4_2_ras_fields[i], ded);
+		if (ded_cnt) {
+			dev_info(adev->dev,
+				 "GFX SubBlock %s, Instance[%d][%d], DED %d\n",
+				 gfx_v9_4_2_ras_fields[i].name, se_id, inst_id,
+				 ded_cnt);
+			*ded_count += ded_cnt;
+		}
+	}
+
+	return 0;
+}
+
+static int gfx_v9_4_2_query_sram_edc_count(struct amdgpu_device *adev,
+				uint32_t *sec_count, uint32_t *ded_count)
+{
+	uint32_t i, j, k, data;
+	uint32_t sec_cnt = 0, ded_cnt = 0;
+
+	if (sec_count && ded_count) {
+		*sec_count = 0;
+		*ded_count = 0;
+	}
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	for (i = 0; i < ARRAY_SIZE(gfx_v9_4_2_edc_counter_regs); i++) {
+		for (j = 0; j < gfx_v9_4_2_edc_counter_regs[i].se_num; j++) {
+			for (k = 0; k < gfx_v9_4_2_edc_counter_regs[i].instance;
+			     k++) {
+				gfx_v9_4_2_select_se_sh(adev, j, 0, k);
+
+				/* if sec/ded_count is null, just clear counter */
+				if (!sec_count || !ded_count) {
+					WREG32(SOC15_REG_ENTRY_OFFSET(
+						gfx_v9_4_2_edc_counter_regs[i]), 0);
+					continue;
+				}
+
+				data = RREG32(SOC15_REG_ENTRY_OFFSET(
+					gfx_v9_4_2_edc_counter_regs[i]));
+
+				if (!data)
+					continue;
+
+				gfx_v9_4_2_get_reg_error_count(adev,
+					&gfx_v9_4_2_edc_counter_regs[i],
+					j, k, data, &sec_cnt, &ded_cnt);
+
+				/* clear counter after read */
+				WREG32(SOC15_REG_ENTRY_OFFSET(
+					gfx_v9_4_2_edc_counter_regs[i]), 0);
+			}
+		}
+	}
+
+	if (sec_count && ded_count) {
+		*sec_count += sec_cnt;
+		*ded_count += ded_cnt;
+	}
+
+	gfx_v9_4_2_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+static void gfx_v9_4_2_log_utc_edc_count(struct amdgpu_device *adev,
+					 struct gfx_v9_4_2_utc_block *blk,
+					 uint32_t instance, uint32_t sec_cnt,
+					 uint32_t ded_cnt)
+{
+	uint32_t bank, way, mem;
+	static const char * const vml2_way_str[] = { "BIGK", "4K" };
+	static const char * const utcl2_router_str[] = { "VMC", "APT" };
+
+	mem = instance % blk->num_mem_blocks;
+	way = (instance / blk->num_mem_blocks) % blk->num_ways;
+	bank = instance / (blk->num_mem_blocks * blk->num_ways);
+
+	switch (blk->type) {
+	case VML2_MEM:
+		dev_info(
+			adev->dev,
+			"GFX SubBlock UTC_VML2_BANK_CACHE_%d_%s_MEM%d, SED %d, DED %d\n",
+			bank, vml2_way_str[way], mem, sec_cnt, ded_cnt);
+		break;
+	case VML2_WALKER_MEM:
+		dev_info(adev->dev, "GFX SubBlock %s, SED %d, DED %d\n",
+			 vml2_walker_mems[bank], sec_cnt, ded_cnt);
+		break;
+	case UTCL2_MEM:
+		dev_info(
+			adev->dev,
+			"GFX SubBlock UTCL2_ROUTER_IFIF%d_GROUP0_%s, SED %d, DED %d\n",
+			bank, utcl2_router_str[mem], sec_cnt, ded_cnt);
+		break;
+	case ATC_L2_CACHE_2M:
+		dev_info(
+			adev->dev,
+			"GFX SubBlock UTC_ATCL2_CACHE_2M_BANK%d_WAY%d_MEM, SED %d, DED %d\n",
+			bank, way, sec_cnt, ded_cnt);
+		break;
+	case ATC_L2_CACHE_32K:
+		dev_info(
+			adev->dev,
+			"GFX SubBlock UTC_ATCL2_CACHE_32K_BANK%d_WAY%d_MEM%d, SED %d, DED %d\n",
+			bank, way, mem, sec_cnt, ded_cnt);
+		break;
+	case ATC_L2_CACHE_4K:
+		dev_info(
+			adev->dev,
+			"GFX SubBlock UTC_ATCL2_CACHE_4K_BANK%d_WAY%d_MEM%d, SED %d, DED %d\n",
+			bank, way, mem, sec_cnt, ded_cnt);
+		break;
+	}
+}
+
+static int gfx_v9_4_2_query_utc_edc_count(struct amdgpu_device *adev,
+					  uint32_t *sec_count,
+					  uint32_t *ded_count)
+{
+	uint32_t i, j, data;
+	uint32_t sec_cnt, ded_cnt;
+	uint32_t num_instances;
+	struct gfx_v9_4_2_utc_block *blk;
+
+	if (sec_count && ded_count) {
+		*sec_count = 0;
+		*ded_count = 0;
+	}
+
+	for (i = 0; i < ARRAY_SIZE(gfx_v9_4_2_utc_blocks); i++) {
+		blk = &gfx_v9_4_2_utc_blocks[i];
+		num_instances =
+			blk->num_banks * blk->num_ways * blk->num_mem_blocks;
+		for (j = 0; j < num_instances; j++) {
+			WREG32(SOC15_REG_ENTRY_OFFSET(blk->idx_reg), j);
+
+			/* if sec/ded_count is NULL, just clear counter */
+			if (!sec_count || !ded_count) {
+				WREG32(SOC15_REG_ENTRY_OFFSET(blk->data_reg),
+				       blk->clear);
+				continue;
+			}
+
+			data = RREG32(SOC15_REG_ENTRY_OFFSET(blk->data_reg));
+			if (!data)
+				continue;
+
+			sec_cnt = SOC15_RAS_REG_FIELD_VAL(data, *blk, sec);
+			*sec_count += sec_cnt;
+			ded_cnt = SOC15_RAS_REG_FIELD_VAL(data, *blk, ded);
+			*ded_count += ded_cnt;
+
+			/* clear counter after read */
+			WREG32(SOC15_REG_ENTRY_OFFSET(blk->data_reg),
+			       blk->clear);
+
+			/* print the edc count */
+			if (sec_cnt || ded_cnt)
+				gfx_v9_4_2_log_utc_edc_count(adev, blk, j, sec_cnt,
+							     ded_cnt);
+		}
+	}
+
+	return 0;
+}
+
+static void gfx_v9_4_2_query_ras_error_count(struct amdgpu_device *adev,
+					    void *ras_error_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+	uint32_t sec_count = 0, ded_count = 0;
+
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
+		return;
+
+	err_data->ue_count = 0;
+	err_data->ce_count = 0;
+
+	gfx_v9_4_2_query_sram_edc_count(adev, &sec_count, &ded_count);
+	err_data->ce_count += sec_count;
+	err_data->ue_count += ded_count;
+
+	gfx_v9_4_2_query_utc_edc_count(adev, &sec_count, &ded_count);
+	err_data->ce_count += sec_count;
+	err_data->ue_count += ded_count;
+
+}
+
+static void gfx_v9_4_2_reset_utc_err_status(struct amdgpu_device *adev)
+{
+	WREG32_SOC15(GC, 0, regUTCL2_MEM_ECC_STATUS, 0x3);
+	WREG32_SOC15(GC, 0, regVML2_MEM_ECC_STATUS, 0x3);
+	WREG32_SOC15(GC, 0, regVML2_WALKER_MEM_ECC_STATUS, 0x3);
+}
+
+static void gfx_v9_4_2_reset_ea_err_status(struct amdgpu_device *adev)
+{
+	uint32_t i, j;
+	uint32_t value;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < gfx_v9_4_2_ea_err_status_regs.se_num; i++) {
+		for (j = 0; j < gfx_v9_4_2_ea_err_status_regs.instance;
+		     j++) {
+			gfx_v9_4_2_select_se_sh(adev, i, 0, j);
+			value = RREG32(SOC15_REG_ENTRY_OFFSET(
+				gfx_v9_4_2_ea_err_status_regs));
+			value = REG_SET_FIELD(value, GCEA_ERR_STATUS, CLEAR_ERROR_STATUS, 0x1);
+			WREG32(SOC15_REG_ENTRY_OFFSET(gfx_v9_4_2_ea_err_status_regs), value);
+		}
+	}
+	gfx_v9_4_2_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+}
+
+static void gfx_v9_4_2_reset_ras_error_count(struct amdgpu_device *adev)
+{
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
+		return;
+
+	gfx_v9_4_2_query_sram_edc_count(adev, NULL, NULL);
+	gfx_v9_4_2_query_utc_edc_count(adev, NULL, NULL);
+}
+
+static void gfx_v9_4_2_query_ea_err_status(struct amdgpu_device *adev)
+{
+	uint32_t i, j;
+	uint32_t reg_value;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	for (i = 0; i < gfx_v9_4_2_ea_err_status_regs.se_num; i++) {
+		for (j = 0; j < gfx_v9_4_2_ea_err_status_regs.instance;
+		     j++) {
+			gfx_v9_4_2_select_se_sh(adev, i, 0, j);
+			reg_value = RREG32(SOC15_REG_ENTRY_OFFSET(
+				gfx_v9_4_2_ea_err_status_regs));
+
+			if (REG_GET_FIELD(reg_value, GCEA_ERR_STATUS, SDP_RDRSP_STATUS) ||
+			    REG_GET_FIELD(reg_value, GCEA_ERR_STATUS, SDP_WRRSP_STATUS) ||
+			    REG_GET_FIELD(reg_value, GCEA_ERR_STATUS, SDP_RDRSP_DATAPARITY_ERROR)) {
+				dev_warn(adev->dev, "GCEA err detected at instance: %d, status: 0x%x!\n",
+						j, reg_value);
+			}
+			/* clear after read */
+			reg_value = REG_SET_FIELD(reg_value, GCEA_ERR_STATUS,
+						  CLEAR_ERROR_STATUS, 0x1);
+			WREG32(SOC15_REG_ENTRY_OFFSET(gfx_v9_4_2_ea_err_status_regs), reg_value);
+		}
+	}
+
+	gfx_v9_4_2_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+}
+
+static void gfx_v9_4_2_query_utc_err_status(struct amdgpu_device *adev)
+{
+	uint32_t data;
+
+	data = RREG32_SOC15(GC, 0, regUTCL2_MEM_ECC_STATUS);
+	if (data) {
+		dev_warn(adev->dev, "GFX UTCL2 Mem Ecc Status: 0x%x!\n", data);
+		WREG32_SOC15(GC, 0, regUTCL2_MEM_ECC_STATUS, 0x3);
+	}
+
+	data = RREG32_SOC15(GC, 0, regVML2_MEM_ECC_STATUS);
+	if (data) {
+		dev_warn(adev->dev, "GFX VML2 Mem Ecc Status: 0x%x!\n", data);
+		WREG32_SOC15(GC, 0, regVML2_MEM_ECC_STATUS, 0x3);
+	}
+
+	data = RREG32_SOC15(GC, 0, regVML2_WALKER_MEM_ECC_STATUS);
+	if (data) {
+		dev_warn(adev->dev, "GFX VML2 Walker Mem Ecc Status: 0x%x!\n", data);
+		WREG32_SOC15(GC, 0, regVML2_WALKER_MEM_ECC_STATUS, 0x3);
+	}
+}
+
+static void gfx_v9_4_2_query_ras_error_status(struct amdgpu_device *adev)
+{
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
+		return;
+
+	gfx_v9_4_2_query_ea_err_status(adev);
+	gfx_v9_4_2_query_utc_err_status(adev);
+	gfx_v9_4_2_query_sq_timeout_status(adev);
+}
+
+static void gfx_v9_4_2_reset_ras_error_status(struct amdgpu_device *adev)
+{
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
+		return;
+
+	gfx_v9_4_2_reset_utc_err_status(adev);
+	gfx_v9_4_2_reset_ea_err_status(adev);
+	gfx_v9_4_2_reset_sq_timeout_status(adev);
+}
+
+static void gfx_v9_4_2_enable_watchdog_timer(struct amdgpu_device *adev)
+{
+	uint32_t i;
+	uint32_t data;
+
+	data = REG_SET_FIELD(0, SQ_TIMEOUT_CONFIG, TIMEOUT_FATAL_DISABLE,
+			     amdgpu_watchdog_timer.timeout_fatal_disable ? 1 :
+									   0);
+
+	if (amdgpu_watchdog_timer.timeout_fatal_disable &&
+	    (amdgpu_watchdog_timer.period < 1 ||
+	     amdgpu_watchdog_timer.period > 0x23)) {
+		dev_warn(adev->dev, "Watchdog period range is 1 to 0x23\n");
+		amdgpu_watchdog_timer.period = 0x23;
+	}
+	data = REG_SET_FIELD(data, SQ_TIMEOUT_CONFIG, PERIOD_SEL,
+			     amdgpu_watchdog_timer.period);
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		gfx_v9_4_2_select_se_sh(adev, i, 0xffffffff, 0xffffffff);
+		WREG32_SOC15(GC, 0, regSQ_TIMEOUT_CONFIG, data);
+	}
+	gfx_v9_4_2_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+}
+
+static uint32_t wave_read_ind(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t address)
+{
+	WREG32_SOC15_RLC_EX(reg, GC, 0, regSQ_IND_INDEX,
+		(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
+		(simd << SQ_IND_INDEX__SIMD_ID__SHIFT) |
+		(address << SQ_IND_INDEX__INDEX__SHIFT) |
+		(SQ_IND_INDEX__FORCE_READ_MASK));
+	return RREG32_SOC15(GC, 0, regSQ_IND_DATA);
+}
+
+static void gfx_v9_4_2_log_cu_timeout_status(struct amdgpu_device *adev,
+					uint32_t status)
+{
+	struct amdgpu_cu_info *cu_info = &adev->gfx.cu_info;
+	uint32_t i, simd, wave;
+	uint32_t wave_status;
+	uint32_t wave_pc_lo, wave_pc_hi;
+	uint32_t wave_exec_lo, wave_exec_hi;
+	uint32_t wave_inst_dw0, wave_inst_dw1;
+	uint32_t wave_ib_sts;
+
+	for (i = 0; i < 32; i++) {
+		if (!((i << 1) & status))
+			continue;
+
+		simd = i / cu_info->max_waves_per_simd;
+		wave = i % cu_info->max_waves_per_simd;
+
+		wave_status = wave_read_ind(adev, simd, wave, ixSQ_WAVE_STATUS);
+		wave_pc_lo = wave_read_ind(adev, simd, wave, ixSQ_WAVE_PC_LO);
+		wave_pc_hi = wave_read_ind(adev, simd, wave, ixSQ_WAVE_PC_HI);
+		wave_exec_lo =
+			wave_read_ind(adev, simd, wave, ixSQ_WAVE_EXEC_LO);
+		wave_exec_hi =
+			wave_read_ind(adev, simd, wave, ixSQ_WAVE_EXEC_HI);
+		wave_inst_dw0 =
+			wave_read_ind(adev, simd, wave, ixSQ_WAVE_INST_DW0);
+		wave_inst_dw1 =
+			wave_read_ind(adev, simd, wave, ixSQ_WAVE_INST_DW1);
+		wave_ib_sts = wave_read_ind(adev, simd, wave, ixSQ_WAVE_IB_STS);
+
+		dev_info(
+			adev->dev,
+			"\t SIMD %d, Wave %d: status 0x%x, pc 0x%llx, exec 0x%llx, inst 0x%llx, ib_sts 0x%x\n",
+			simd, wave, wave_status,
+			((uint64_t)wave_pc_hi << 32 | wave_pc_lo),
+			((uint64_t)wave_exec_hi << 32 | wave_exec_lo),
+			((uint64_t)wave_inst_dw1 << 32 | wave_inst_dw0),
+			wave_ib_sts);
+	}
+}
+
+static void gfx_v9_4_2_query_sq_timeout_status(struct amdgpu_device *adev)
+{
+	uint32_t se_idx, sh_idx, cu_idx;
+	uint32_t status;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (se_idx = 0; se_idx < adev->gfx.config.max_shader_engines;
+	     se_idx++) {
+		for (sh_idx = 0; sh_idx < adev->gfx.config.max_sh_per_se;
+		     sh_idx++) {
+			for (cu_idx = 0;
+			     cu_idx < adev->gfx.config.max_cu_per_sh;
+			     cu_idx++) {
+				gfx_v9_4_2_select_se_sh(adev, se_idx, sh_idx,
+							cu_idx);
+				status = RREG32_SOC15(GC, 0,
+						      regSQ_TIMEOUT_STATUS);
+				if (status != 0) {
+					dev_info(
+						adev->dev,
+						"GFX Watchdog Timeout: SE %d, SH %d, CU %d\n",
+						se_idx, sh_idx, cu_idx);
+					gfx_v9_4_2_log_cu_timeout_status(
+						adev, status);
+				}
+				/* clear old status */
+				WREG32_SOC15(GC, 0, regSQ_TIMEOUT_STATUS, 0);
+			}
+		}
+	}
+	gfx_v9_4_2_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+}
+
+static void gfx_v9_4_2_reset_sq_timeout_status(struct amdgpu_device *adev)
+{
+	uint32_t se_idx, sh_idx, cu_idx;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (se_idx = 0; se_idx < adev->gfx.config.max_shader_engines;
+	     se_idx++) {
+		for (sh_idx = 0; sh_idx < adev->gfx.config.max_sh_per_se;
+		     sh_idx++) {
+			for (cu_idx = 0;
+			     cu_idx < adev->gfx.config.max_cu_per_sh;
+			     cu_idx++) {
+				gfx_v9_4_2_select_se_sh(adev, se_idx, sh_idx,
+							cu_idx);
+				WREG32_SOC15(GC, 0, regSQ_TIMEOUT_STATUS, 0);
+			}
+		}
+	}
+	gfx_v9_4_2_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+}
+
+
+
+struct amdgpu_ras_block_hw_ops  gfx_v9_4_2_ras_ops = {
+		.query_ras_error_count = &gfx_v9_4_2_query_ras_error_count,
+		.reset_ras_error_count = &gfx_v9_4_2_reset_ras_error_count,
+		.query_ras_error_status = &gfx_v9_4_2_query_ras_error_status,
+		.reset_ras_error_status = &gfx_v9_4_2_reset_ras_error_status,
+};
+
+struct amdgpu_gfx_ras gfx_v9_4_2_ras = {
+	.ras_block = {
+		.hw_ops = &gfx_v9_4_2_ras_ops,
+	},
+	.enable_watchdog_timer = &gfx_v9_4_2_enable_watchdog_timer,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.h b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.h
new file mode 100644
index 000000000000..a603724c1dfc
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.h
@@ -0,0 +1,37 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFX_V9_4_2_H__
+#define __GFX_V9_4_2_H__
+
+void gfx_v9_4_2_debug_trap_config_init(struct amdgpu_device *adev,
+				uint32_t first_vmid, uint32_t last_vmid);
+void gfx_v9_4_2_init_golden_registers(struct amdgpu_device *adev,
+				      uint32_t die_id);
+void gfx_v9_4_2_init_sq(struct amdgpu_device *adev);
+void gfx_v9_4_2_set_power_brake_sequence(struct amdgpu_device *adev);
+int gfx_v9_4_2_do_edc_gpr_workarounds(struct amdgpu_device *adev);
+
+extern struct amdgpu_gfx_ras gfx_v9_4_2_ras;
+
+#endif /* __GFX_V9_4_2_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2_cleaner_shader.asm b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2_cleaner_shader.asm
new file mode 100644
index 000000000000..35b8cf9070bd
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2_cleaner_shader.asm
@@ -0,0 +1,153 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+// This shader is to clean LDS, SGPRs and VGPRs. It is  first 64 Dwords or 256 bytes of 192 Dwords cleaner shader.
+//To turn this shader program on for complitaion change this to main and lower shader main to main_1
+
+// MI200 : Clear SGPRs, VGPRs and LDS
+//   Uses two kernels launched separately:
+//   1. Clean VGPRs, LDS, and lower SGPRs
+//        Launches one workgroup per CU, each workgroup with 4x wave64 per SIMD in the CU
+//        Waves are "wave64" and have 128 VGPRs each, which uses all 512 VGPRs per SIMD
+//        Waves in the workgroup share the 64KB of LDS
+//        Each wave clears SGPRs 0 - 95. Because there are 4 waves/SIMD, this is physical SGPRs 0-383
+//        Each wave clears 128 VGPRs, so all 512 in the SIMD
+//        The first wave of the workgroup clears its 64KB of LDS
+//        The shader starts with "S_BARRIER" to ensure SPI has launched all waves of the workgroup
+//          before any wave in the workgroup could end.  Without this, it is possible not all SGPRs get cleared.
+//    2. Clean remaining SGPRs
+//        Launches a workgroup with 24 waves per workgroup, yielding 6 waves per SIMD in each CU
+//        Waves are allocating 96 SGPRs
+//          CP sets up SPI_RESOURCE_RESERVE_* registers to prevent these waves from allocating SGPRs 0-223.
+//          As such, these 6 waves per SIMD are allocated physical SGPRs 224-799
+//        Barriers do not work for >16 waves per workgroup, so we cannot start with S_BARRIER
+//          Instead, the shader starts with an S_SETHALT 1. Once all waves are launched CP will send unhalt command
+//        The shader then clears all SGPRs allocated to it, cleaning out physical SGPRs 224-799
+
+shader main
+  asic(MI200)
+  type(CS)
+  wave_size(64)
+// Note: original source code from SQ team
+
+//   (theorhetical fastest = ~512clks vgpr + 1536 lds + ~128 sgpr  = 2176 clks)
+
+  s_cmp_eq_u32 s0, 1                                // Bit0 is set, sgpr0 is set then clear VGPRS and LDS as FW set COMPUTE_USER_DATA_3
+  s_cbranch_scc0  label_0023                        // Clean VGPRs and LDS if sgpr0 of wave is set, scc = (s3 == 1)
+  S_BARRIER
+
+  s_movk_i32    m0, 0x0000
+  s_mov_b32     s2, 0x00000078  // Loop 128/8=16 times  (loop unrolled for performance)
+  //
+  // CLEAR VGPRs
+  //
+  s_set_gpr_idx_on  s2, 0x8    // enable Dest VGPR indexing
+label_0005:
+  v_mov_b32     v0, 0
+  v_mov_b32     v1, 0
+  v_mov_b32     v2, 0
+  v_mov_b32     v3, 0
+  v_mov_b32     v4, 0
+  v_mov_b32     v5, 0
+  v_mov_b32     v6, 0
+  v_mov_b32     v7, 0
+  s_sub_u32     s2, s2, 8
+  s_set_gpr_idx_idx  s2
+  s_cbranch_scc0  label_0005
+  s_set_gpr_idx_off
+
+  //
+  //
+
+  s_mov_b32     s2, 0x80000000                      // Bit31 is first_wave
+  s_and_b32     s2, s2, s1                          // sgpr0 has tg_size (first_wave) term as in ucode only COMPUTE_PGM_RSRC2.tg_size_en is set
+  s_cbranch_scc0  label_clean_sgpr_1                // Clean LDS if its first wave of ThreadGroup/WorkGroup
+  // CLEAR LDS
+  //
+  s_mov_b32 exec_lo, 0xffffffff
+  s_mov_b32 exec_hi, 0xffffffff
+  v_mbcnt_lo_u32_b32  v1, exec_hi, 0          // Set V1 to thread-ID (0..63)
+  v_mbcnt_hi_u32_b32  v1, exec_lo, v1         // Set V1 to thread-ID (0..63)
+  v_mul_u32_u24  v1, 0x00000008, v1           // * 8, so each thread is a double-dword address (8byte)
+  s_mov_b32     s2, 0x00000003f               // 64 loop iterations
+  s_mov_b32     m0, 0xffffffff
+  // Clear all of LDS space
+  // Each FirstWave of WorkGroup clears 64kbyte block
+
+label_001F:
+  ds_write2_b64  v1, v[2:3], v[2:3] offset1:32
+  ds_write2_b64  v1, v[4:5], v[4:5] offset0:64 offset1:96
+  v_add_co_u32     v1, vcc, 0x00000400, v1
+  s_sub_u32     s2, s2, 1
+  s_cbranch_scc0  label_001F
+  //
+  // CLEAR SGPRs
+  //
+label_clean_sgpr_1:
+  s_mov_b32     m0, 0x0000005c   // Loop 96/4=24 times  (loop unrolled for performance)
+  s_nop 0
+label_sgpr_loop:
+  s_movreld_b32     s0, 0
+  s_movreld_b32     s1, 0
+  s_movreld_b32     s2, 0
+  s_movreld_b32     s3, 0
+  s_sub_u32         m0, m0, 4
+  s_cbranch_scc0  label_sgpr_loop
+
+  //clear vcc, flat scratch
+  s_mov_b32 flat_scratch_lo, 0   //clear  flat scratch lo SGPR
+  s_mov_b32 flat_scratch_hi, 0   //clear  flat scratch hi SGPR
+  s_mov_b64 vcc, 0               //clear vcc
+  s_mov_b64 ttmp0, 0             //Clear ttmp0 and ttmp1
+  s_mov_b64 ttmp2, 0             //Clear ttmp2 and ttmp3
+  s_mov_b64 ttmp4, 0             //Clear ttmp4 and ttmp5
+  s_mov_b64 ttmp6, 0             //Clear ttmp6 and ttmp7
+  s_mov_b64 ttmp8, 0             //Clear ttmp8 and ttmp9
+  s_mov_b64 ttmp10, 0            //Clear ttmp10 and ttmp11
+  s_mov_b64 ttmp12, 0            //Clear ttmp12 and ttmp13
+  s_mov_b64 ttmp14, 0            //Clear ttmp14 and ttmp15
+s_endpgm
+
+label_0023:
+
+  s_sethalt 1
+
+  s_mov_b32     m0, 0x0000005c   // Loop 96/4=24 times  (loop unrolled for performance)
+  s_nop 0
+label_sgpr_loop1:
+
+  s_movreld_b32     s0, 0
+  s_movreld_b32     s1, 0
+  s_movreld_b32     s2, 0
+  s_movreld_b32     s3, 0
+  s_sub_u32         m0, m0, 4
+  s_cbranch_scc0  label_sgpr_loop1
+
+  //clear vcc, flat scratch
+  s_mov_b32 flat_scratch_lo, 0   //clear  flat scratch lo SGPR
+  s_mov_b32 flat_scratch_hi, 0   //clear  flat scratch hi SGPR
+  s_mov_b64 vcc, 0xee            //clear vcc
+
+s_endpgm
+end
+
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
new file mode 100644
index 000000000000..77f9d5b9a556
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
@@ -0,0 +1,5063 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/firmware.h>
+
+#include "amdgpu.h"
+#include "amdgpu_gfx.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "soc15_common.h"
+#include "vega10_enum.h"
+
+#include "v9_structs.h"
+
+#include "ivsrcid/gfx/irqsrcs_gfx_9_0.h"
+
+#include "gc/gc_9_4_3_offset.h"
+#include "gc/gc_9_4_3_sh_mask.h"
+
+#include "gfx_v9_4_3.h"
+#include "gfx_v9_4_3_cleaner_shader.h"
+#include "amdgpu_xcp.h"
+#include "amdgpu_aca.h"
+
+MODULE_FIRMWARE("amdgpu/gc_9_4_3_mec.bin");
+MODULE_FIRMWARE("amdgpu/gc_9_4_4_mec.bin");
+MODULE_FIRMWARE("amdgpu/gc_9_5_0_mec.bin");
+MODULE_FIRMWARE("amdgpu/gc_9_4_3_rlc.bin");
+MODULE_FIRMWARE("amdgpu/gc_9_4_4_rlc.bin");
+MODULE_FIRMWARE("amdgpu/gc_9_5_0_rlc.bin");
+MODULE_FIRMWARE("amdgpu/gc_9_4_3_sjt_mec.bin");
+MODULE_FIRMWARE("amdgpu/gc_9_4_4_sjt_mec.bin");
+
+#define GFX9_MEC_HPD_SIZE 4096
+#define RLCG_UCODE_LOADING_START_ADDRESS 0x00002000L
+
+#define GOLDEN_GB_ADDR_CONFIG 0x2a114042
+#define CP_HQD_PERSISTENT_STATE_DEFAULT 0xbe05301
+
+#define XCC_REG_RANGE_0_LOW  0x2000     /* XCC gfxdec0 lower Bound */
+#define XCC_REG_RANGE_0_HIGH 0x3400     /* XCC gfxdec0 upper Bound */
+#define XCC_REG_RANGE_1_LOW  0xA000     /* XCC gfxdec1 lower Bound */
+#define XCC_REG_RANGE_1_HIGH 0x10000    /* XCC gfxdec1 upper Bound */
+
+#define NORMALIZE_XCC_REG_OFFSET(offset) \
+	(offset & 0xFFFF)
+
+static const struct amdgpu_hwip_reg_entry gc_reg_list_9_4_3[] = {
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS2),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_STALLED_STAT1),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_STALLED_STAT2),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPC_STALLED_STAT1),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPF_STALLED_STAT1),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_BUSY_STAT),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPC_BUSY_STAT),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPF_BUSY_STAT),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPF_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_ERROR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCPF_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regCPC_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regCPG_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regGDS_PROTECTION_FAULT),
+	SOC15_REG_ENTRY_STR(GC, 0, regGDS_VM_PROTECTION_FAULT),
+	SOC15_REG_ENTRY_STR(GC, 0, regRLC_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regRMI_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regSQC_DCACHE_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regSQC_ICACHE_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regSQ_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regTCP_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regWD_UTCL1_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regVM_L2_PROTECTION_FAULT_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regVM_L2_PROTECTION_FAULT_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_DEBUG),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC1_INSTR_PNTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC2_INSTR_PNTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPC_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regRLC_STAT),
+	SOC15_REG_ENTRY_STR(GC, 0, regRLC_SMU_COMMAND),
+	SOC15_REG_ENTRY_STR(GC, 0, regRLC_SMU_MESSAGE),
+	SOC15_REG_ENTRY_STR(GC, 0, regRLC_SMU_ARGUMENT_1),
+	SOC15_REG_ENTRY_STR(GC, 0, regRLC_SMU_ARGUMENT_2),
+	SOC15_REG_ENTRY_STR(GC, 0, regSMU_RLC_RESPONSE),
+	SOC15_REG_ENTRY_STR(GC, 0, regRLC_SAFE_MODE),
+	SOC15_REG_ENTRY_STR(GC, 0, regRLC_SMU_SAFE_MODE),
+	SOC15_REG_ENTRY_STR(GC, 0, regRLC_INT_STAT),
+	SOC15_REG_ENTRY_STR(GC, 0, regRLC_GPM_GENERAL_6),
+	/* SE status registers */
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS_SE0),
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS_SE1),
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS_SE2),
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS_SE3)
+};
+
+static const struct amdgpu_hwip_reg_entry gc_cp_reg_list_9_4_3[] = {
+	/* compute queue registers */
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_VMID),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_ACTIVE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PERSISTENT_STATE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PIPE_PRIORITY),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_QUEUE_PRIORITY),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_QUANTUM),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_BASE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_WPTR_POLL_ADDR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_WPTR_POLL_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_IB_BASE_ADDR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_IB_BASE_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_IB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_IB_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_DEQUEUE_REQUEST),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_BASE_ADDR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_BASE_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_EVENTS),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CTX_SAVE_BASE_ADDR_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CTX_SAVE_BASE_ADDR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CTX_SAVE_CONTROL),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CNTL_STACK_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CNTL_STACK_SIZE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_WG_STATE_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CTX_SAVE_SIZE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_GDS_RESOURCE_STATE),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_ERROR),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_WPTR_MEM),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_WPTR_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_GFX_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+};
+
+struct amdgpu_gfx_ras gfx_v9_4_3_ras;
+
+static void gfx_v9_4_3_set_ring_funcs(struct amdgpu_device *adev);
+static void gfx_v9_4_3_set_irq_funcs(struct amdgpu_device *adev);
+static void gfx_v9_4_3_set_gds_init(struct amdgpu_device *adev);
+static void gfx_v9_4_3_set_rlc_funcs(struct amdgpu_device *adev);
+static int gfx_v9_4_3_get_cu_info(struct amdgpu_device *adev,
+				struct amdgpu_cu_info *cu_info);
+static void gfx_v9_4_3_xcc_set_safe_mode(struct amdgpu_device *adev, int xcc_id);
+static void gfx_v9_4_3_xcc_unset_safe_mode(struct amdgpu_device *adev, int xcc_id);
+
+static void gfx_v9_4_3_kiq_set_resources(struct amdgpu_ring *kiq_ring,
+				uint64_t queue_mask)
+{
+	struct amdgpu_device *adev = kiq_ring->adev;
+	u64 shader_mc_addr;
+
+	/* Cleaner shader MC address */
+	shader_mc_addr = adev->gfx.cleaner_shader_gpu_addr >> 8;
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_SET_RESOURCES, 6));
+	amdgpu_ring_write(kiq_ring,
+		PACKET3_SET_RESOURCES_VMID_MASK(0) |
+		/* vmid_mask:0* queue_type:0 (KIQ) */
+		PACKET3_SET_RESOURCES_QUEUE_TYPE(0));
+	amdgpu_ring_write(kiq_ring,
+			lower_32_bits(queue_mask));	/* queue mask lo */
+	amdgpu_ring_write(kiq_ring,
+			upper_32_bits(queue_mask));	/* queue mask hi */
+	amdgpu_ring_write(kiq_ring, lower_32_bits(shader_mc_addr)); /* cleaner shader addr lo */
+	amdgpu_ring_write(kiq_ring, upper_32_bits(shader_mc_addr)); /* cleaner shader addr hi */
+	amdgpu_ring_write(kiq_ring, 0);	/* oac mask */
+	amdgpu_ring_write(kiq_ring, 0);	/* gds heap base:0, gds heap size:0 */
+}
+
+static void gfx_v9_4_3_kiq_map_queues(struct amdgpu_ring *kiq_ring,
+				 struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = kiq_ring->adev;
+	uint64_t mqd_addr = amdgpu_bo_gpu_offset(ring->mqd_obj);
+	uint64_t wptr_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
+	uint32_t eng_sel = ring->funcs->type == AMDGPU_RING_TYPE_GFX ? 4 : 0;
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_MAP_QUEUES, 5));
+	/* Q_sel:0, vmid:0, vidmem: 1, engine:0, num_Q:1*/
+	amdgpu_ring_write(kiq_ring, /* Q_sel: 0, vmid: 0, engine: 0, num_Q: 1 */
+			 PACKET3_MAP_QUEUES_QUEUE_SEL(0) | /* Queue_Sel */
+			 PACKET3_MAP_QUEUES_VMID(0) | /* VMID */
+			 PACKET3_MAP_QUEUES_QUEUE(ring->queue) |
+			 PACKET3_MAP_QUEUES_PIPE(ring->pipe) |
+			 PACKET3_MAP_QUEUES_ME((ring->me == 1 ? 0 : 1)) |
+			 /*queue_type: normal compute queue */
+			 PACKET3_MAP_QUEUES_QUEUE_TYPE(0) |
+			 /* alloc format: all_on_one_pipe */
+			 PACKET3_MAP_QUEUES_ALLOC_FORMAT(0) |
+			 PACKET3_MAP_QUEUES_ENGINE_SEL(eng_sel) |
+			 /* num_queues: must be 1 */
+			 PACKET3_MAP_QUEUES_NUM_QUEUES(1));
+	amdgpu_ring_write(kiq_ring,
+			PACKET3_MAP_QUEUES_DOORBELL_OFFSET(ring->doorbell_index));
+	amdgpu_ring_write(kiq_ring, lower_32_bits(mqd_addr));
+	amdgpu_ring_write(kiq_ring, upper_32_bits(mqd_addr));
+	amdgpu_ring_write(kiq_ring, lower_32_bits(wptr_addr));
+	amdgpu_ring_write(kiq_ring, upper_32_bits(wptr_addr));
+}
+
+static void gfx_v9_4_3_kiq_unmap_queues(struct amdgpu_ring *kiq_ring,
+				   struct amdgpu_ring *ring,
+				   enum amdgpu_unmap_queues_action action,
+				   u64 gpu_addr, u64 seq)
+{
+	uint32_t eng_sel = ring->funcs->type == AMDGPU_RING_TYPE_GFX ? 4 : 0;
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_UNMAP_QUEUES, 4));
+	amdgpu_ring_write(kiq_ring, /* Q_sel: 0, vmid: 0, engine: 0, num_Q: 1 */
+			  PACKET3_UNMAP_QUEUES_ACTION(action) |
+			  PACKET3_UNMAP_QUEUES_QUEUE_SEL(0) |
+			  PACKET3_UNMAP_QUEUES_ENGINE_SEL(eng_sel) |
+			  PACKET3_UNMAP_QUEUES_NUM_QUEUES(1));
+	amdgpu_ring_write(kiq_ring,
+			PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET0(ring->doorbell_index));
+
+	if (action == PREEMPT_QUEUES_NO_UNMAP) {
+		amdgpu_ring_write(kiq_ring, lower_32_bits(gpu_addr));
+		amdgpu_ring_write(kiq_ring, upper_32_bits(gpu_addr));
+		amdgpu_ring_write(kiq_ring, seq);
+	} else {
+		amdgpu_ring_write(kiq_ring, 0);
+		amdgpu_ring_write(kiq_ring, 0);
+		amdgpu_ring_write(kiq_ring, 0);
+	}
+}
+
+static void gfx_v9_4_3_kiq_query_status(struct amdgpu_ring *kiq_ring,
+				   struct amdgpu_ring *ring,
+				   u64 addr,
+				   u64 seq)
+{
+	uint32_t eng_sel = ring->funcs->type == AMDGPU_RING_TYPE_GFX ? 4 : 0;
+
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_QUERY_STATUS, 5));
+	amdgpu_ring_write(kiq_ring,
+			  PACKET3_QUERY_STATUS_CONTEXT_ID(0) |
+			  PACKET3_QUERY_STATUS_INTERRUPT_SEL(0) |
+			  PACKET3_QUERY_STATUS_COMMAND(2));
+	/* Q_sel: 0, vmid: 0, engine: 0, num_Q: 1 */
+	amdgpu_ring_write(kiq_ring,
+			PACKET3_QUERY_STATUS_DOORBELL_OFFSET(ring->doorbell_index) |
+			PACKET3_QUERY_STATUS_ENG_SEL(eng_sel));
+	amdgpu_ring_write(kiq_ring, lower_32_bits(addr));
+	amdgpu_ring_write(kiq_ring, upper_32_bits(addr));
+	amdgpu_ring_write(kiq_ring, lower_32_bits(seq));
+	amdgpu_ring_write(kiq_ring, upper_32_bits(seq));
+}
+
+static void gfx_v9_4_3_kiq_invalidate_tlbs(struct amdgpu_ring *kiq_ring,
+				uint16_t pasid, uint32_t flush_type,
+				bool all_hub)
+{
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_INVALIDATE_TLBS, 0));
+	amdgpu_ring_write(kiq_ring,
+			PACKET3_INVALIDATE_TLBS_DST_SEL(1) |
+			PACKET3_INVALIDATE_TLBS_ALL_HUB(all_hub) |
+			PACKET3_INVALIDATE_TLBS_PASID(pasid) |
+			PACKET3_INVALIDATE_TLBS_FLUSH_TYPE(flush_type));
+}
+
+static void gfx_v9_4_3_kiq_reset_hw_queue(struct amdgpu_ring *kiq_ring, uint32_t queue_type,
+					  uint32_t me_id, uint32_t pipe_id, uint32_t queue_id,
+					  uint32_t xcc_id, uint32_t vmid)
+{
+	struct amdgpu_device *adev = kiq_ring->adev;
+	unsigned i;
+
+	/* enter save mode */
+	amdgpu_gfx_rlc_enter_safe_mode(adev, xcc_id);
+	mutex_lock(&adev->srbm_mutex);
+	soc15_grbm_select(adev, me_id, pipe_id, queue_id, 0, xcc_id);
+
+	if (queue_type == AMDGPU_RING_TYPE_COMPUTE) {
+		WREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_HQD_DEQUEUE_REQUEST, 0x2);
+		WREG32_SOC15(GC, GET_INST(GC, xcc_id), regSPI_COMPUTE_QUEUE_RESET, 0x1);
+		/* wait till dequeue take effects */
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (!(RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_HQD_ACTIVE) & 1))
+				break;
+			udelay(1);
+		}
+		if (i >= adev->usec_timeout)
+			dev_err(adev->dev, "fail to wait on hqd deactive\n");
+	} else {
+		dev_err(adev->dev, "reset queue_type(%d) not supported\n\n", queue_type);
+	}
+
+	soc15_grbm_select(adev, 0, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+	/* exit safe mode */
+	amdgpu_gfx_rlc_exit_safe_mode(adev, xcc_id);
+}
+
+static const struct kiq_pm4_funcs gfx_v9_4_3_kiq_pm4_funcs = {
+	.kiq_set_resources = gfx_v9_4_3_kiq_set_resources,
+	.kiq_map_queues = gfx_v9_4_3_kiq_map_queues,
+	.kiq_unmap_queues = gfx_v9_4_3_kiq_unmap_queues,
+	.kiq_query_status = gfx_v9_4_3_kiq_query_status,
+	.kiq_invalidate_tlbs = gfx_v9_4_3_kiq_invalidate_tlbs,
+	.kiq_reset_hw_queue = gfx_v9_4_3_kiq_reset_hw_queue,
+	.set_resources_size = 8,
+	.map_queues_size = 7,
+	.unmap_queues_size = 6,
+	.query_status_size = 7,
+	.invalidate_tlbs_size = 2,
+};
+
+static void gfx_v9_4_3_set_kiq_pm4_funcs(struct amdgpu_device *adev)
+{
+	int i, num_xcc;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	for (i = 0; i < num_xcc; i++)
+		adev->gfx.kiq[i].pmf = &gfx_v9_4_3_kiq_pm4_funcs;
+}
+
+static void gfx_v9_4_3_init_golden_registers(struct amdgpu_device *adev)
+{
+	int i, num_xcc, dev_inst;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	for (i = 0; i < num_xcc; i++) {
+		dev_inst = GET_INST(GC, i);
+
+		WREG32_SOC15(GC, dev_inst, regGB_ADDR_CONFIG,
+			     GOLDEN_GB_ADDR_CONFIG);
+		WREG32_FIELD15_PREREG(GC, dev_inst, TCP_UTCL1_CNTL2, SPARE, 0x1);
+	}
+}
+
+static uint32_t gfx_v9_4_3_normalize_xcc_reg_offset(uint32_t reg)
+{
+	uint32_t normalized_reg = NORMALIZE_XCC_REG_OFFSET(reg);
+
+	/* If it is an XCC reg, normalize the reg to keep
+	   lower 16 bits in local xcc */
+
+	if (((normalized_reg >= XCC_REG_RANGE_0_LOW) && (normalized_reg < XCC_REG_RANGE_0_HIGH)) ||
+		((normalized_reg >= XCC_REG_RANGE_1_LOW) && (normalized_reg < XCC_REG_RANGE_1_HIGH)))
+		return normalized_reg;
+	else
+		return reg;
+}
+
+static void gfx_v9_4_3_write_data_to_reg(struct amdgpu_ring *ring, int eng_sel,
+				       bool wc, uint32_t reg, uint32_t val)
+{
+	reg = gfx_v9_4_3_normalize_xcc_reg_offset(reg);
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, WRITE_DATA_ENGINE_SEL(eng_sel) |
+				WRITE_DATA_DST_SEL(0) |
+				(wc ? WR_CONFIRM : 0));
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, val);
+}
+
+static void gfx_v9_4_3_wait_reg_mem(struct amdgpu_ring *ring, int eng_sel,
+				  int mem_space, int opt, uint32_t addr0,
+				  uint32_t addr1, uint32_t ref, uint32_t mask,
+				  uint32_t inv)
+{
+	/* Only do the normalization on regspace */
+	if (mem_space == 0) {
+		addr0 = gfx_v9_4_3_normalize_xcc_reg_offset(addr0);
+		addr1 = gfx_v9_4_3_normalize_xcc_reg_offset(addr1);
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+	amdgpu_ring_write(ring,
+				 /* memory (1) or register (0) */
+				 (WAIT_REG_MEM_MEM_SPACE(mem_space) |
+				 WAIT_REG_MEM_OPERATION(opt) | /* wait */
+				 WAIT_REG_MEM_FUNCTION(3) |  /* equal */
+				 WAIT_REG_MEM_ENGINE(eng_sel)));
+
+	if (mem_space)
+		BUG_ON(addr0 & 0x3); /* Dword align */
+	amdgpu_ring_write(ring, addr0);
+	amdgpu_ring_write(ring, addr1);
+	amdgpu_ring_write(ring, ref);
+	amdgpu_ring_write(ring, mask);
+	amdgpu_ring_write(ring, inv); /* poll interval */
+}
+
+static int gfx_v9_4_3_ring_test_ring(struct amdgpu_ring *ring)
+{
+	uint32_t scratch_reg0_offset, xcc_offset;
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	/* Use register offset which is local to XCC in the packet */
+	xcc_offset = SOC15_REG_OFFSET(GC, 0, regSCRATCH_REG0);
+	scratch_reg0_offset = SOC15_REG_OFFSET(GC, GET_INST(GC, ring->xcc_id), regSCRATCH_REG0);
+	WREG32(scratch_reg0_offset, 0xCAFEDEAD);
+	tmp = RREG32(scratch_reg0_offset);
+
+	r = amdgpu_ring_alloc(ring, 3);
+	if (r)
+		return r;
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_UCONFIG_REG, 1));
+	amdgpu_ring_write(ring, xcc_offset - PACKET3_SET_UCONFIG_REG_START);
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(scratch_reg0_offset);
+		if (tmp == 0xDEADBEEF)
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+	return r;
+}
+
+static int gfx_v9_4_3_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct dma_fence *f = NULL;
+
+	unsigned index;
+	uint64_t gpu_addr;
+	uint32_t tmp;
+	long r;
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r)
+		return r;
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	adev->wb.wb[index] = cpu_to_le32(0xCAFEDEAD);
+	memset(&ib, 0, sizeof(ib));
+
+	r = amdgpu_ib_get(adev, NULL, 20, AMDGPU_IB_POOL_DIRECT, &ib);
+	if (r)
+		goto err1;
+
+	ib.ptr[0] = PACKET3(PACKET3_WRITE_DATA, 3);
+	ib.ptr[1] = WRITE_DATA_DST_SEL(5) | WR_CONFIRM;
+	ib.ptr[2] = lower_32_bits(gpu_addr);
+	ib.ptr[3] = upper_32_bits(gpu_addr);
+	ib.ptr[4] = 0xDEADBEEF;
+	ib.length_dw = 5;
+
+	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
+	if (r)
+		goto err2;
+
+	r = dma_fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		r = -ETIMEDOUT;
+		goto err2;
+	} else if (r < 0) {
+		goto err2;
+	}
+
+	tmp = adev->wb.wb[index];
+	if (tmp == 0xDEADBEEF)
+		r = 0;
+	else
+		r = -EINVAL;
+
+err2:
+	amdgpu_ib_free(&ib, NULL);
+	dma_fence_put(f);
+err1:
+	amdgpu_device_wb_free(adev, index);
+	return r;
+}
+
+
+/* This value might differs per partition */
+static uint64_t gfx_v9_4_3_get_gpu_clock_counter(struct amdgpu_device *adev)
+{
+	uint64_t clock;
+
+	mutex_lock(&adev->gfx.gpu_clock_mutex);
+	WREG32_SOC15(GC, GET_INST(GC, 0), regRLC_CAPTURE_GPU_CLOCK_COUNT, 1);
+	clock = (uint64_t)RREG32_SOC15(GC, GET_INST(GC, 0), regRLC_GPU_CLOCK_COUNT_LSB) |
+		((uint64_t)RREG32_SOC15(GC, GET_INST(GC, 0), regRLC_GPU_CLOCK_COUNT_MSB) << 32ULL);
+	mutex_unlock(&adev->gfx.gpu_clock_mutex);
+
+	return clock;
+}
+
+static void gfx_v9_4_3_free_microcode(struct amdgpu_device *adev)
+{
+	amdgpu_ucode_release(&adev->gfx.pfp_fw);
+	amdgpu_ucode_release(&adev->gfx.me_fw);
+	amdgpu_ucode_release(&adev->gfx.ce_fw);
+	amdgpu_ucode_release(&adev->gfx.rlc_fw);
+	amdgpu_ucode_release(&adev->gfx.mec_fw);
+	amdgpu_ucode_release(&adev->gfx.mec2_fw);
+
+	kfree(adev->gfx.rlc.register_list_format);
+}
+
+static int gfx_v9_4_3_init_rlc_microcode(struct amdgpu_device *adev,
+					  const char *chip_name)
+{
+	int err;
+	const struct rlc_firmware_header_v2_0 *rlc_hdr;
+	uint16_t version_major;
+	uint16_t version_minor;
+
+
+	err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw,
+				   AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_rlc.bin", chip_name);
+	if (err)
+		goto out;
+	rlc_hdr = (const struct rlc_firmware_header_v2_0 *)adev->gfx.rlc_fw->data;
+
+	version_major = le16_to_cpu(rlc_hdr->header.header_version_major);
+	version_minor = le16_to_cpu(rlc_hdr->header.header_version_minor);
+	err = amdgpu_gfx_rlc_init_microcode(adev, version_major, version_minor);
+out:
+	if (err)
+		amdgpu_ucode_release(&adev->gfx.rlc_fw);
+
+	return err;
+}
+
+static int gfx_v9_4_3_init_cp_compute_microcode(struct amdgpu_device *adev,
+					  const char *chip_name)
+{
+	int err;
+
+	if (amdgpu_sriov_vf(adev)) {
+		err = amdgpu_ucode_request(adev, &adev->gfx.mec_fw,
+					   AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_sjt_mec.bin", chip_name);
+
+		if (err)
+			err = amdgpu_ucode_request(adev, &adev->gfx.mec_fw,
+							AMDGPU_UCODE_REQUIRED,
+							"amdgpu/%s_mec.bin", chip_name);
+	} else
+		err = amdgpu_ucode_request(adev, &adev->gfx.mec_fw,
+					   AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_mec.bin", chip_name);
+	if (err)
+		goto out;
+	amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_MEC1);
+	amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_MEC1_JT);
+
+	adev->gfx.mec2_fw_version = adev->gfx.mec_fw_version;
+	adev->gfx.mec2_feature_version = adev->gfx.mec_feature_version;
+
+out:
+	if (err)
+		amdgpu_ucode_release(&adev->gfx.mec_fw);
+	return err;
+}
+
+static int gfx_v9_4_3_init_microcode(struct amdgpu_device *adev)
+{
+	char ucode_prefix[15];
+	int r;
+
+	amdgpu_ucode_ip_version_decode(adev, GC_HWIP, ucode_prefix, sizeof(ucode_prefix));
+
+	r = gfx_v9_4_3_init_rlc_microcode(adev, ucode_prefix);
+	if (r)
+		return r;
+
+	r = gfx_v9_4_3_init_cp_compute_microcode(adev, ucode_prefix);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static void gfx_v9_4_3_mec_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->gfx.mec.hpd_eop_obj, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gfx.mec.mec_fw_obj, NULL, NULL);
+}
+
+static int gfx_v9_4_3_mec_init(struct amdgpu_device *adev)
+{
+	int r, i, num_xcc;
+	u32 *hpd;
+	const __le32 *fw_data;
+	unsigned fw_size;
+	u32 *fw;
+	size_t mec_hpd_size;
+
+	const struct gfx_firmware_header_v1_0 *mec_hdr;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	for (i = 0; i < num_xcc; i++)
+		bitmap_zero(adev->gfx.mec_bitmap[i].queue_bitmap,
+			AMDGPU_MAX_COMPUTE_QUEUES);
+
+	/* take ownership of the relevant compute queues */
+	amdgpu_gfx_compute_queue_acquire(adev);
+	mec_hpd_size =
+		adev->gfx.num_compute_rings * num_xcc * GFX9_MEC_HPD_SIZE;
+	if (mec_hpd_size) {
+		r = amdgpu_bo_create_reserved(adev, mec_hpd_size, PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_VRAM |
+					      AMDGPU_GEM_DOMAIN_GTT,
+					      &adev->gfx.mec.hpd_eop_obj,
+					      &adev->gfx.mec.hpd_eop_gpu_addr,
+					      (void **)&hpd);
+		if (r) {
+			dev_warn(adev->dev, "(%d) create HDP EOP bo failed\n", r);
+			gfx_v9_4_3_mec_fini(adev);
+			return r;
+		}
+
+		if (amdgpu_emu_mode == 1) {
+			for (i = 0; i < mec_hpd_size / 4; i++) {
+				memset((void *)(hpd + i), 0, 4);
+				if (i % 50 == 0)
+					msleep(1);
+			}
+		} else {
+			memset(hpd, 0, mec_hpd_size);
+		}
+
+		amdgpu_bo_kunmap(adev->gfx.mec.hpd_eop_obj);
+		amdgpu_bo_unreserve(adev->gfx.mec.hpd_eop_obj);
+	}
+
+	mec_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+
+	fw_data = (const __le32 *)
+		(adev->gfx.mec_fw->data +
+		 le32_to_cpu(mec_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(mec_hdr->header.ucode_size_bytes);
+
+	r = amdgpu_bo_create_reserved(adev, mec_hdr->header.ucode_size_bytes,
+				      PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.mec.mec_fw_obj,
+				      &adev->gfx.mec.mec_fw_gpu_addr,
+				      (void **)&fw);
+	if (r) {
+		dev_warn(adev->dev, "(%d) create mec firmware bo failed\n", r);
+		gfx_v9_4_3_mec_fini(adev);
+		return r;
+	}
+
+	memcpy(fw, fw_data, fw_size);
+
+	amdgpu_bo_kunmap(adev->gfx.mec.mec_fw_obj);
+	amdgpu_bo_unreserve(adev->gfx.mec.mec_fw_obj);
+
+	return 0;
+}
+
+static void gfx_v9_4_3_xcc_select_se_sh(struct amdgpu_device *adev, u32 se_num,
+					u32 sh_num, u32 instance, int xcc_id)
+{
+	u32 data;
+
+	if (instance == 0xffffffff)
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX,
+				     INSTANCE_BROADCAST_WRITES, 1);
+	else
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX,
+				     INSTANCE_INDEX, instance);
+
+	if (se_num == 0xffffffff)
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+				     SE_BROADCAST_WRITES, 1);
+	else
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_INDEX, se_num);
+
+	if (sh_num == 0xffffffff)
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+				     SH_BROADCAST_WRITES, 1);
+	else
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_INDEX, sh_num);
+
+	WREG32_SOC15_RLC_SHADOW_EX(reg, GC, GET_INST(GC, xcc_id), regGRBM_GFX_INDEX, data);
+}
+
+static uint32_t wave_read_ind(struct amdgpu_device *adev, uint32_t xcc_id, uint32_t simd, uint32_t wave, uint32_t address)
+{
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regSQ_IND_INDEX,
+		(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
+		(simd << SQ_IND_INDEX__SIMD_ID__SHIFT) |
+		(address << SQ_IND_INDEX__INDEX__SHIFT) |
+		(SQ_IND_INDEX__FORCE_READ_MASK));
+	return RREG32_SOC15(GC, GET_INST(GC, xcc_id), regSQ_IND_DATA);
+}
+
+static void wave_read_regs(struct amdgpu_device *adev, uint32_t xcc_id, uint32_t simd,
+			   uint32_t wave, uint32_t thread,
+			   uint32_t regno, uint32_t num, uint32_t *out)
+{
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regSQ_IND_INDEX,
+		(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
+		(simd << SQ_IND_INDEX__SIMD_ID__SHIFT) |
+		(regno << SQ_IND_INDEX__INDEX__SHIFT) |
+		(thread << SQ_IND_INDEX__THREAD_ID__SHIFT) |
+		(SQ_IND_INDEX__FORCE_READ_MASK) |
+		(SQ_IND_INDEX__AUTO_INCR_MASK));
+	while (num--)
+		*(out++) = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regSQ_IND_DATA);
+}
+
+static void gfx_v9_4_3_read_wave_data(struct amdgpu_device *adev,
+				      uint32_t xcc_id, uint32_t simd, uint32_t wave,
+				      uint32_t *dst, int *no_fields)
+{
+	/* type 1 wave data */
+	dst[(*no_fields)++] = 1;
+	dst[(*no_fields)++] = wave_read_ind(adev, xcc_id, simd, wave, ixSQ_WAVE_STATUS);
+	dst[(*no_fields)++] = wave_read_ind(adev, xcc_id, simd, wave, ixSQ_WAVE_PC_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, xcc_id, simd, wave, ixSQ_WAVE_PC_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, xcc_id, simd, wave, ixSQ_WAVE_EXEC_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, xcc_id, simd, wave, ixSQ_WAVE_EXEC_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, xcc_id, simd, wave, ixSQ_WAVE_HW_ID);
+	dst[(*no_fields)++] = wave_read_ind(adev, xcc_id, simd, wave, ixSQ_WAVE_INST_DW0);
+	dst[(*no_fields)++] = wave_read_ind(adev, xcc_id, simd, wave, ixSQ_WAVE_INST_DW1);
+	dst[(*no_fields)++] = wave_read_ind(adev, xcc_id, simd, wave, ixSQ_WAVE_GPR_ALLOC);
+	dst[(*no_fields)++] = wave_read_ind(adev, xcc_id, simd, wave, ixSQ_WAVE_LDS_ALLOC);
+	dst[(*no_fields)++] = wave_read_ind(adev, xcc_id, simd, wave, ixSQ_WAVE_TRAPSTS);
+	dst[(*no_fields)++] = wave_read_ind(adev, xcc_id, simd, wave, ixSQ_WAVE_IB_STS);
+	dst[(*no_fields)++] = wave_read_ind(adev, xcc_id, simd, wave, ixSQ_WAVE_IB_DBG0);
+	dst[(*no_fields)++] = wave_read_ind(adev, xcc_id, simd, wave, ixSQ_WAVE_M0);
+	dst[(*no_fields)++] = wave_read_ind(adev, xcc_id, simd, wave, ixSQ_WAVE_MODE);
+}
+
+static void gfx_v9_4_3_read_wave_sgprs(struct amdgpu_device *adev, uint32_t xcc_id, uint32_t simd,
+				       uint32_t wave, uint32_t start,
+				       uint32_t size, uint32_t *dst)
+{
+	wave_read_regs(adev, xcc_id, simd, wave, 0,
+		       start + SQIND_WAVE_SGPRS_OFFSET, size, dst);
+}
+
+static void gfx_v9_4_3_read_wave_vgprs(struct amdgpu_device *adev, uint32_t xcc_id, uint32_t simd,
+				       uint32_t wave, uint32_t thread,
+				       uint32_t start, uint32_t size,
+				       uint32_t *dst)
+{
+	wave_read_regs(adev, xcc_id, simd, wave, thread,
+		       start + SQIND_WAVE_VGPRS_OFFSET, size, dst);
+}
+
+static void gfx_v9_4_3_select_me_pipe_q(struct amdgpu_device *adev,
+					u32 me, u32 pipe, u32 q, u32 vm, u32 xcc_id)
+{
+	soc15_grbm_select(adev, me, pipe, q, vm, GET_INST(GC, xcc_id));
+}
+
+static int gfx_v9_4_3_get_xccs_per_xcp(struct amdgpu_device *adev)
+{
+	u32 xcp_ctl;
+
+	/* Value is expected to be the same on all, fetch from first instance */
+	xcp_ctl = RREG32_SOC15(GC, GET_INST(GC, 0), regCP_HYP_XCP_CTL);
+
+	return REG_GET_FIELD(xcp_ctl, CP_HYP_XCP_CTL, NUM_XCC_IN_XCP);
+}
+
+static int gfx_v9_4_3_switch_compute_partition(struct amdgpu_device *adev,
+						int num_xccs_per_xcp)
+{
+	int ret, i, num_xcc;
+	u32 tmp = 0;
+
+	if (adev->psp.funcs) {
+		ret = psp_spatial_partition(&adev->psp,
+					    NUM_XCC(adev->gfx.xcc_mask) /
+						    num_xccs_per_xcp);
+		if (ret)
+			return ret;
+	} else {
+		num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+
+		for (i = 0; i < num_xcc; i++) {
+			tmp = REG_SET_FIELD(tmp, CP_HYP_XCP_CTL, NUM_XCC_IN_XCP,
+					    num_xccs_per_xcp);
+			tmp = REG_SET_FIELD(tmp, CP_HYP_XCP_CTL, VIRTUAL_XCC_ID,
+					    i % num_xccs_per_xcp);
+			WREG32_SOC15(GC, GET_INST(GC, i), regCP_HYP_XCP_CTL,
+				     tmp);
+		}
+		ret = 0;
+	}
+
+	adev->gfx.num_xcc_per_xcp = num_xccs_per_xcp;
+
+	return ret;
+}
+
+static int gfx_v9_4_3_ih_to_xcc_inst(struct amdgpu_device *adev, int ih_node)
+{
+	int xcc;
+
+	xcc = hweight8(adev->gfx.xcc_mask & GENMASK(ih_node / 2, 0));
+	if (!xcc) {
+		dev_err(adev->dev, "Couldn't find xcc mapping from IH node");
+		return -EINVAL;
+	}
+
+	return xcc - 1;
+}
+
+static const struct amdgpu_gfx_funcs gfx_v9_4_3_gfx_funcs = {
+	.get_gpu_clock_counter = &gfx_v9_4_3_get_gpu_clock_counter,
+	.select_se_sh = &gfx_v9_4_3_xcc_select_se_sh,
+	.read_wave_data = &gfx_v9_4_3_read_wave_data,
+	.read_wave_sgprs = &gfx_v9_4_3_read_wave_sgprs,
+	.read_wave_vgprs = &gfx_v9_4_3_read_wave_vgprs,
+	.select_me_pipe_q = &gfx_v9_4_3_select_me_pipe_q,
+	.switch_partition_mode = &gfx_v9_4_3_switch_compute_partition,
+	.ih_node_to_logical_xcc = &gfx_v9_4_3_ih_to_xcc_inst,
+	.get_xccs_per_xcp = &gfx_v9_4_3_get_xccs_per_xcp,
+};
+
+static int gfx_v9_4_3_aca_bank_parser(struct aca_handle *handle,
+				      struct aca_bank *bank, enum aca_smu_type type,
+				      void *data)
+{
+	struct aca_bank_info info;
+	u64 misc0;
+	u32 instlo;
+	int ret;
+
+	ret = aca_bank_info_decode(bank, &info);
+	if (ret)
+		return ret;
+
+	/* NOTE: overwrite info.die_id with xcd id for gfx */
+	instlo = ACA_REG__IPID__INSTANCEIDLO(bank->regs[ACA_REG_IDX_IPID]);
+	instlo &= GENMASK(31, 1);
+	info.die_id = instlo == mmSMNAID_XCD0_MCA_SMU ? 0 : 1;
+
+	misc0 = bank->regs[ACA_REG_IDX_MISC0];
+
+	switch (type) {
+	case ACA_SMU_TYPE_UE:
+		bank->aca_err_type = ACA_ERROR_TYPE_UE;
+		ret = aca_error_cache_log_bank_error(handle, &info, bank->aca_err_type, 1ULL);
+		break;
+	case ACA_SMU_TYPE_CE:
+		bank->aca_err_type = ACA_ERROR_TYPE_CE;
+		ret = aca_error_cache_log_bank_error(handle, &info, bank->aca_err_type,
+						     ACA_REG__MISC0__ERRCNT(misc0));
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return ret;
+}
+
+static bool gfx_v9_4_3_aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank,
+					 enum aca_smu_type type, void *data)
+{
+	u32 instlo;
+
+	instlo = ACA_REG__IPID__INSTANCEIDLO(bank->regs[ACA_REG_IDX_IPID]);
+	instlo &= GENMASK(31, 1);
+	switch (instlo) {
+	case mmSMNAID_XCD0_MCA_SMU:
+	case mmSMNAID_XCD1_MCA_SMU:
+	case mmSMNXCD_XCD0_MCA_SMU:
+		return true;
+	default:
+		break;
+	}
+
+	return false;
+}
+
+static const struct aca_bank_ops gfx_v9_4_3_aca_bank_ops = {
+	.aca_bank_parser = gfx_v9_4_3_aca_bank_parser,
+	.aca_bank_is_valid = gfx_v9_4_3_aca_bank_is_valid,
+};
+
+static const struct aca_info gfx_v9_4_3_aca_info = {
+	.hwip = ACA_HWIP_TYPE_SMU,
+	.mask = ACA_ERROR_UE_MASK | ACA_ERROR_CE_MASK,
+	.bank_ops = &gfx_v9_4_3_aca_bank_ops,
+};
+
+static int gfx_v9_4_3_gpu_early_init(struct amdgpu_device *adev)
+{
+	adev->gfx.funcs = &gfx_v9_4_3_gfx_funcs;
+	adev->gfx.ras = &gfx_v9_4_3_ras;
+
+	adev->gfx.config.max_hw_contexts = 8;
+	adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+	adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+	adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+	adev->gfx.config.sc_earlyz_tile_fifo_size = 0x4C0;
+	adev->gfx.config.gb_addr_config = GOLDEN_GB_ADDR_CONFIG;
+
+	adev->gfx.config.gb_addr_config_fields.num_pipes = 1 <<
+			REG_GET_FIELD(
+					adev->gfx.config.gb_addr_config,
+					GB_ADDR_CONFIG,
+					NUM_PIPES);
+
+	adev->gfx.config.max_tile_pipes =
+		adev->gfx.config.gb_addr_config_fields.num_pipes;
+
+	adev->gfx.config.gb_addr_config_fields.num_banks = 1 <<
+			REG_GET_FIELD(
+					adev->gfx.config.gb_addr_config,
+					GB_ADDR_CONFIG,
+					NUM_BANKS);
+	adev->gfx.config.gb_addr_config_fields.max_compress_frags = 1 <<
+			REG_GET_FIELD(
+					adev->gfx.config.gb_addr_config,
+					GB_ADDR_CONFIG,
+					MAX_COMPRESSED_FRAGS);
+	adev->gfx.config.gb_addr_config_fields.num_rb_per_se = 1 <<
+			REG_GET_FIELD(
+					adev->gfx.config.gb_addr_config,
+					GB_ADDR_CONFIG,
+					NUM_RB_PER_SE);
+	adev->gfx.config.gb_addr_config_fields.num_se = 1 <<
+			REG_GET_FIELD(
+					adev->gfx.config.gb_addr_config,
+					GB_ADDR_CONFIG,
+					NUM_SHADER_ENGINES);
+	adev->gfx.config.gb_addr_config_fields.pipe_interleave_size = 1 << (8 +
+			REG_GET_FIELD(
+					adev->gfx.config.gb_addr_config,
+					GB_ADDR_CONFIG,
+					PIPE_INTERLEAVE_SIZE));
+
+	return 0;
+}
+
+static int gfx_v9_4_3_compute_ring_init(struct amdgpu_device *adev, int ring_id,
+				        int xcc_id, int mec, int pipe, int queue)
+{
+	unsigned irq_type;
+	struct amdgpu_ring *ring = &adev->gfx.compute_ring[ring_id];
+	unsigned int hw_prio;
+	uint32_t xcc_doorbell_start;
+
+	ring = &adev->gfx.compute_ring[xcc_id * adev->gfx.num_compute_rings +
+				       ring_id];
+
+	/* mec0 is me1 */
+	ring->xcc_id = xcc_id;
+	ring->me = mec + 1;
+	ring->pipe = pipe;
+	ring->queue = queue;
+
+	ring->ring_obj = NULL;
+	ring->use_doorbell = true;
+	xcc_doorbell_start = adev->doorbell_index.mec_ring0 +
+			     xcc_id * adev->doorbell_index.xcc_doorbell_range;
+	ring->doorbell_index = (xcc_doorbell_start + ring_id) << 1;
+	ring->eop_gpu_addr = adev->gfx.mec.hpd_eop_gpu_addr +
+			     (ring_id + xcc_id * adev->gfx.num_compute_rings) *
+				     GFX9_MEC_HPD_SIZE;
+	ring->vm_hub = AMDGPU_GFXHUB(xcc_id);
+	sprintf(ring->name, "comp_%d.%d.%d.%d",
+			ring->xcc_id, ring->me, ring->pipe, ring->queue);
+
+	irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP
+		+ ((ring->me - 1) * adev->gfx.mec.num_pipe_per_mec)
+		+ ring->pipe;
+	hw_prio = amdgpu_gfx_is_high_priority_compute_queue(adev, ring) ?
+			AMDGPU_GFX_PIPE_PRIO_HIGH : AMDGPU_GFX_PIPE_PRIO_NORMAL;
+	/* type-2 packets are deprecated on MEC, use type-3 instead */
+	return amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq, irq_type,
+				hw_prio, NULL);
+}
+
+static void gfx_v9_4_3_alloc_ip_dump(struct amdgpu_device *adev)
+{
+	uint32_t reg_count = ARRAY_SIZE(gc_reg_list_9_4_3);
+	uint32_t *ptr, num_xcc, inst;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+
+	ptr = kcalloc(reg_count * num_xcc, sizeof(uint32_t), GFP_KERNEL);
+	if (!ptr) {
+		DRM_ERROR("Failed to allocate memory for GFX IP Dump\n");
+		adev->gfx.ip_dump_core = NULL;
+	} else {
+		adev->gfx.ip_dump_core = ptr;
+	}
+
+	/* Allocate memory for compute queue registers for all the instances */
+	reg_count = ARRAY_SIZE(gc_cp_reg_list_9_4_3);
+	inst = adev->gfx.mec.num_mec * adev->gfx.mec.num_pipe_per_mec *
+		adev->gfx.mec.num_queue_per_pipe;
+
+	ptr = kcalloc(reg_count * inst * num_xcc, sizeof(uint32_t), GFP_KERNEL);
+	if (!ptr) {
+		DRM_ERROR("Failed to allocate memory for Compute Queues IP Dump\n");
+		adev->gfx.ip_dump_compute_queues = NULL;
+	} else {
+		adev->gfx.ip_dump_compute_queues = ptr;
+	}
+}
+
+static int gfx_v9_4_3_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	int i, j, k, r, ring_id, xcc_id, num_xcc;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 4, 3):
+	case IP_VERSION(9, 4, 4):
+		adev->gfx.cleaner_shader_ptr = gfx_9_4_3_cleaner_shader_hex;
+		adev->gfx.cleaner_shader_size = sizeof(gfx_9_4_3_cleaner_shader_hex);
+		if (adev->gfx.mec_fw_version >= 153) {
+			adev->gfx.enable_cleaner_shader = true;
+			r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);
+			if (r) {
+				adev->gfx.enable_cleaner_shader = false;
+				dev_err(adev->dev, "Failed to initialize cleaner shader\n");
+			}
+		}
+		break;
+	default:
+		adev->gfx.enable_cleaner_shader = false;
+		break;
+	}
+
+	adev->gfx.mec.num_mec = 2;
+	adev->gfx.mec.num_pipe_per_mec = 4;
+	adev->gfx.mec.num_queue_per_pipe = 8;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+
+	/* EOP Event */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_GRBM_CP, GFX_9_0__SRCID__CP_EOP_INTERRUPT, &adev->gfx.eop_irq);
+	if (r)
+		return r;
+
+	/* Bad opcode Event */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_GRBM_CP,
+			      GFX_9_0__SRCID__CP_BAD_OPCODE_ERROR,
+			      &adev->gfx.bad_op_irq);
+	if (r)
+		return r;
+
+	/* Privileged reg */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_GRBM_CP, GFX_9_0__SRCID__CP_PRIV_REG_FAULT,
+			      &adev->gfx.priv_reg_irq);
+	if (r)
+		return r;
+
+	/* Privileged inst */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_GRBM_CP, GFX_9_0__SRCID__CP_PRIV_INSTR_FAULT,
+			      &adev->gfx.priv_inst_irq);
+	if (r)
+		return r;
+
+	adev->gfx.gfx_current_status = AMDGPU_GFX_NORMAL_MODE;
+
+	r = adev->gfx.rlc.funcs->init(adev);
+	if (r) {
+		DRM_ERROR("Failed to init rlc BOs!\n");
+		return r;
+	}
+
+	r = gfx_v9_4_3_mec_init(adev);
+	if (r) {
+		DRM_ERROR("Failed to init MEC BOs!\n");
+		return r;
+	}
+
+	/* set up the compute queues - allocate horizontally across pipes */
+	for (xcc_id = 0; xcc_id < num_xcc; xcc_id++) {
+		ring_id = 0;
+		for (i = 0; i < adev->gfx.mec.num_mec; ++i) {
+			for (j = 0; j < adev->gfx.mec.num_queue_per_pipe; j++) {
+				for (k = 0; k < adev->gfx.mec.num_pipe_per_mec;
+				     k++) {
+					if (!amdgpu_gfx_is_mec_queue_enabled(
+							adev, xcc_id, i, k, j))
+						continue;
+
+					r = gfx_v9_4_3_compute_ring_init(adev,
+								       ring_id,
+								       xcc_id,
+								       i, k, j);
+					if (r)
+						return r;
+
+					ring_id++;
+				}
+			}
+		}
+
+		r = amdgpu_gfx_kiq_init(adev, GFX9_MEC_HPD_SIZE, xcc_id);
+		if (r) {
+			DRM_ERROR("Failed to init KIQ BOs!\n");
+			return r;
+		}
+
+		r = amdgpu_gfx_kiq_init_ring(adev, xcc_id);
+		if (r)
+			return r;
+
+		/* create MQD for all compute queues as wel as KIQ for SRIOV case */
+		r = amdgpu_gfx_mqd_sw_init(adev,
+				sizeof(struct v9_mqd_allocation), xcc_id);
+		if (r)
+			return r;
+	}
+
+	adev->gfx.compute_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.compute_ring[0]);
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 4, 3):
+	case IP_VERSION(9, 4, 4):
+		if ((adev->gfx.mec_fw_version >= 155) &&
+		    !amdgpu_sriov_vf(adev)) {
+			adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+			adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_PIPE;
+		}
+		break;
+	case IP_VERSION(9, 5, 0):
+		if ((adev->gfx.mec_fw_version >= 21) &&
+		    !amdgpu_sriov_vf(adev)) {
+			adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+			adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_PIPE;
+		}
+		break;
+	default:
+		break;
+	}
+	r = gfx_v9_4_3_gpu_early_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_gfx_ras_sw_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_gfx_sysfs_init(adev);
+	if (r)
+		return r;
+
+	gfx_v9_4_3_alloc_ip_dump(adev);
+
+	return 0;
+}
+
+static int gfx_v9_4_3_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	int i, num_xcc;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	for (i = 0; i < adev->gfx.num_compute_rings * num_xcc; i++)
+		amdgpu_ring_fini(&adev->gfx.compute_ring[i]);
+
+	for (i = 0; i < num_xcc; i++) {
+		amdgpu_gfx_mqd_sw_fini(adev, i);
+		amdgpu_gfx_kiq_free_ring(&adev->gfx.kiq[i].ring);
+		amdgpu_gfx_kiq_fini(adev, i);
+	}
+
+	amdgpu_gfx_cleaner_shader_sw_fini(adev);
+
+	gfx_v9_4_3_mec_fini(adev);
+	amdgpu_bo_unref(&adev->gfx.rlc.clear_state_obj);
+	gfx_v9_4_3_free_microcode(adev);
+	amdgpu_gfx_sysfs_fini(adev);
+
+	kfree(adev->gfx.ip_dump_core);
+	kfree(adev->gfx.ip_dump_compute_queues);
+
+	return 0;
+}
+
+#define DEFAULT_SH_MEM_BASES	(0x6000)
+static void gfx_v9_4_3_xcc_init_compute_vmid(struct amdgpu_device *adev,
+					     int xcc_id)
+{
+	int i;
+	uint32_t sh_mem_config;
+	uint32_t sh_mem_bases;
+	uint32_t data;
+
+	/*
+	 * Configure apertures:
+	 * LDS:         0x60000000'00000000 - 0x60000001'00000000 (4GB)
+	 * Scratch:     0x60000001'00000000 - 0x60000002'00000000 (4GB)
+	 * GPUVM:       0x60010000'00000000 - 0x60020000'00000000 (1TB)
+	 */
+	sh_mem_bases = DEFAULT_SH_MEM_BASES | (DEFAULT_SH_MEM_BASES << 16);
+
+	sh_mem_config = SH_MEM_ADDRESS_MODE_64 |
+			SH_MEM_ALIGNMENT_MODE_UNALIGNED <<
+			SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT;
+
+	mutex_lock(&adev->srbm_mutex);
+	for (i = adev->vm_manager.first_kfd_vmid; i < AMDGPU_NUM_VMID; i++) {
+		soc15_grbm_select(adev, 0, 0, 0, i, GET_INST(GC, xcc_id));
+		/* CP and shaders */
+		WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regSH_MEM_CONFIG, sh_mem_config);
+		WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regSH_MEM_BASES, sh_mem_bases);
+
+		/* Enable trap for each kfd vmid. */
+		data = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regSPI_GDBG_PER_VMID_CNTL);
+		data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, TRAP_EN, 1);
+		WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regSPI_GDBG_PER_VMID_CNTL, data);
+	}
+	soc15_grbm_select(adev, 0, 0, 0, 0, GET_INST(GC, xcc_id));
+	mutex_unlock(&adev->srbm_mutex);
+
+	/*
+	 * Initialize all compute VMIDs to have no GDS, GWS, or OA
+	 * access. These should be enabled by FW for target VMIDs.
+	 */
+	for (i = adev->vm_manager.first_kfd_vmid; i < AMDGPU_NUM_VMID; i++) {
+		WREG32_SOC15_OFFSET(GC, GET_INST(GC, xcc_id), regGDS_VMID0_BASE, 2 * i, 0);
+		WREG32_SOC15_OFFSET(GC, GET_INST(GC, xcc_id), regGDS_VMID0_SIZE, 2 * i, 0);
+		WREG32_SOC15_OFFSET(GC, GET_INST(GC, xcc_id), regGDS_GWS_VMID0, i, 0);
+		WREG32_SOC15_OFFSET(GC, GET_INST(GC, xcc_id), regGDS_OA_VMID0, i, 0);
+	}
+}
+
+static void gfx_v9_4_3_xcc_init_gds_vmid(struct amdgpu_device *adev, int xcc_id)
+{
+	int vmid;
+
+	/*
+	 * Initialize all compute and user-gfx VMIDs to have no GDS, GWS, or OA
+	 * access. Compute VMIDs should be enabled by FW for target VMIDs,
+	 * the driver can enable them for graphics. VMID0 should maintain
+	 * access so that HWS firmware can save/restore entries.
+	 */
+	for (vmid = 1; vmid < AMDGPU_NUM_VMID; vmid++) {
+		WREG32_SOC15_OFFSET(GC, GET_INST(GC, xcc_id), regGDS_VMID0_BASE, 2 * vmid, 0);
+		WREG32_SOC15_OFFSET(GC, GET_INST(GC, xcc_id), regGDS_VMID0_SIZE, 2 * vmid, 0);
+		WREG32_SOC15_OFFSET(GC, GET_INST(GC, xcc_id), regGDS_GWS_VMID0, vmid, 0);
+		WREG32_SOC15_OFFSET(GC, GET_INST(GC, xcc_id), regGDS_OA_VMID0, vmid, 0);
+	}
+}
+
+/* For ASICs that needs xnack chain and MEC version supports, set SG_CONFIG1
+ * DISABLE_XNACK_CHECK_IN_RETRY_DISABLE bit and inform KFD to set xnack_chain
+ * bit in SET_RESOURCES
+ */
+static void gfx_v9_4_3_xcc_init_sq(struct amdgpu_device *adev, int xcc_id)
+{
+	uint32_t data;
+
+	if (!(adev->gmc.xnack_flags & AMDGPU_GMC_XNACK_FLAG_CHAIN))
+		return;
+
+	data = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regSQ_CONFIG1);
+	data = REG_SET_FIELD(data, SQ_CONFIG1, DISABLE_XNACK_CHECK_IN_RETRY_DISABLE, 1);
+	WREG32_SOC15(GC, xcc_id, regSQ_CONFIG1, data);
+}
+
+static void gfx_v9_4_3_xcc_constants_init(struct amdgpu_device *adev,
+					  int xcc_id)
+{
+	u32 tmp;
+	int i;
+
+	/* XXX SH_MEM regs */
+	/* where to put LDS, scratch, GPUVM in FSA64 space */
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < adev->vm_manager.id_mgr[AMDGPU_GFXHUB(0)].num_ids; i++) {
+		soc15_grbm_select(adev, 0, 0, 0, i, GET_INST(GC, xcc_id));
+		/* CP and shaders */
+		if (i == 0) {
+			tmp = REG_SET_FIELD(0, SH_MEM_CONFIG, ALIGNMENT_MODE,
+					    SH_MEM_ALIGNMENT_MODE_UNALIGNED);
+			tmp = REG_SET_FIELD(tmp, SH_MEM_CONFIG, RETRY_DISABLE,
+					    !!adev->gmc.noretry);
+			WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id),
+					 regSH_MEM_CONFIG, tmp);
+			WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id),
+					 regSH_MEM_BASES, 0);
+		} else {
+			tmp = REG_SET_FIELD(0, SH_MEM_CONFIG, ALIGNMENT_MODE,
+					    SH_MEM_ALIGNMENT_MODE_UNALIGNED);
+			tmp = REG_SET_FIELD(tmp, SH_MEM_CONFIG, RETRY_DISABLE,
+					    !!adev->gmc.noretry);
+			WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id),
+					 regSH_MEM_CONFIG, tmp);
+			tmp = REG_SET_FIELD(0, SH_MEM_BASES, PRIVATE_BASE,
+					    (adev->gmc.private_aperture_start >>
+					     48));
+			tmp = REG_SET_FIELD(tmp, SH_MEM_BASES, SHARED_BASE,
+					    (adev->gmc.shared_aperture_start >>
+					     48));
+			WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id),
+					 regSH_MEM_BASES, tmp);
+		}
+	}
+	soc15_grbm_select(adev, 0, 0, 0, 0, GET_INST(GC, 0));
+
+	mutex_unlock(&adev->srbm_mutex);
+
+	gfx_v9_4_3_xcc_init_compute_vmid(adev, xcc_id);
+	gfx_v9_4_3_xcc_init_gds_vmid(adev, xcc_id);
+	gfx_v9_4_3_xcc_init_sq(adev, xcc_id);
+}
+
+static void gfx_v9_4_3_constants_init(struct amdgpu_device *adev)
+{
+	int i, num_xcc;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+
+	gfx_v9_4_3_get_cu_info(adev, &adev->gfx.cu_info);
+	adev->gfx.config.db_debug2 =
+		RREG32_SOC15(GC, GET_INST(GC, 0), regDB_DEBUG2);
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	/* ToDo: GC 9.4.4 */
+	case IP_VERSION(9, 4, 3):
+		if (adev->gfx.mec_fw_version >= 184 &&
+		    (amdgpu_sriov_reg_access_sq_config(adev) ||
+		     !amdgpu_sriov_vf(adev)))
+			adev->gmc.xnack_flags |= AMDGPU_GMC_XNACK_FLAG_CHAIN;
+		break;
+	case IP_VERSION(9, 5, 0):
+		if (adev->gfx.mec_fw_version >= 23)
+			adev->gmc.xnack_flags |= AMDGPU_GMC_XNACK_FLAG_CHAIN;
+		break;
+	default:
+		break;
+	}
+
+	for (i = 0; i < num_xcc; i++)
+		gfx_v9_4_3_xcc_constants_init(adev, i);
+}
+
+static void
+gfx_v9_4_3_xcc_enable_save_restore_machine(struct amdgpu_device *adev,
+					   int xcc_id)
+{
+	WREG32_FIELD15_PREREG(GC, GET_INST(GC, xcc_id), RLC_SRM_CNTL, SRM_ENABLE, 1);
+}
+
+static void gfx_v9_4_3_xcc_init_pg(struct amdgpu_device *adev, int xcc_id)
+{
+	/*
+	 * Rlc save restore list is workable since v2_1.
+	 */
+	gfx_v9_4_3_xcc_enable_save_restore_machine(adev, xcc_id);
+}
+
+static void gfx_v9_4_3_xcc_disable_gpa_mode(struct amdgpu_device *adev, int xcc_id)
+{
+	uint32_t data;
+
+	data = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCPC_PSP_DEBUG);
+	data |= CPC_PSP_DEBUG__UTCL2IUGPAOVERRIDE_MASK;
+	WREG32_SOC15(GC, GET_INST(GC, xcc_id), regCPC_PSP_DEBUG, data);
+}
+
+static bool gfx_v9_4_3_is_rlc_enabled(struct amdgpu_device *adev)
+{
+	uint32_t rlc_setting;
+
+	/* if RLC is not enabled, do nothing */
+	rlc_setting = RREG32_SOC15(GC, GET_INST(GC, 0), regRLC_CNTL);
+	if (!(rlc_setting & RLC_CNTL__RLC_ENABLE_F32_MASK))
+		return false;
+
+	return true;
+}
+
+static void gfx_v9_4_3_xcc_set_safe_mode(struct amdgpu_device *adev, int xcc_id)
+{
+	uint32_t data;
+	unsigned i;
+
+	data = RLC_SAFE_MODE__CMD_MASK;
+	data |= (1 << RLC_SAFE_MODE__MESSAGE__SHIFT);
+	WREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_SAFE_MODE, data);
+
+	/* wait for RLC_SAFE_MODE */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (!REG_GET_FIELD(RREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_SAFE_MODE), RLC_SAFE_MODE, CMD))
+			break;
+		udelay(1);
+	}
+}
+
+static void gfx_v9_4_3_xcc_unset_safe_mode(struct amdgpu_device *adev,
+					   int xcc_id)
+{
+	uint32_t data;
+
+	data = RLC_SAFE_MODE__CMD_MASK;
+	WREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_SAFE_MODE, data);
+}
+
+static void gfx_v9_4_3_init_rlcg_reg_access_ctrl(struct amdgpu_device *adev)
+{
+	int xcc_id, num_xcc;
+	struct amdgpu_rlcg_reg_access_ctrl *reg_access_ctrl;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	for (xcc_id = 0; xcc_id < num_xcc; xcc_id++) {
+		reg_access_ctrl = &adev->gfx.rlc.reg_access_ctrl[GET_INST(GC, xcc_id)];
+		reg_access_ctrl->scratch_reg0 = SOC15_REG_OFFSET(GC, GET_INST(GC, xcc_id), regSCRATCH_REG0);
+		reg_access_ctrl->scratch_reg1 = SOC15_REG_OFFSET(GC, GET_INST(GC, xcc_id), regSCRATCH_REG1);
+		reg_access_ctrl->scratch_reg2 = SOC15_REG_OFFSET(GC, GET_INST(GC, xcc_id), regSCRATCH_REG2);
+		reg_access_ctrl->scratch_reg3 = SOC15_REG_OFFSET(GC, GET_INST(GC, xcc_id), regSCRATCH_REG3);
+		reg_access_ctrl->grbm_cntl = SOC15_REG_OFFSET(GC, GET_INST(GC, xcc_id), regGRBM_GFX_CNTL);
+		reg_access_ctrl->grbm_idx = SOC15_REG_OFFSET(GC, GET_INST(GC, xcc_id), regGRBM_GFX_INDEX);
+		reg_access_ctrl->spare_int = SOC15_REG_OFFSET(GC, GET_INST(GC, xcc_id), regRLC_SPARE_INT);
+	}
+	adev->gfx.rlc.rlcg_reg_access_supported = true;
+}
+
+static int gfx_v9_4_3_rlc_init(struct amdgpu_device *adev)
+{
+	/* init spm vmid with 0xf */
+	if (adev->gfx.rlc.funcs->update_spm_vmid)
+		adev->gfx.rlc.funcs->update_spm_vmid(adev, NULL, 0xf);
+
+	return 0;
+}
+
+static void gfx_v9_4_3_xcc_wait_for_rlc_serdes(struct amdgpu_device *adev,
+					       int xcc_id)
+{
+	u32 i, j, k;
+	u32 mask;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			gfx_v9_4_3_xcc_select_se_sh(adev, i, j, 0xffffffff,
+						    xcc_id);
+			for (k = 0; k < adev->usec_timeout; k++) {
+				if (RREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_SERDES_CU_MASTER_BUSY) == 0)
+					break;
+				udelay(1);
+			}
+			if (k == adev->usec_timeout) {
+				gfx_v9_4_3_xcc_select_se_sh(adev, 0xffffffff,
+							    0xffffffff,
+							    0xffffffff, xcc_id);
+				mutex_unlock(&adev->grbm_idx_mutex);
+				DRM_INFO("Timeout wait for RLC serdes %u,%u\n",
+					 i, j);
+				return;
+			}
+		}
+	}
+	gfx_v9_4_3_xcc_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff,
+				    xcc_id);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	mask = RLC_SERDES_NONCU_MASTER_BUSY__SE_MASTER_BUSY_MASK |
+		RLC_SERDES_NONCU_MASTER_BUSY__GC_MASTER_BUSY_MASK |
+		RLC_SERDES_NONCU_MASTER_BUSY__TC0_MASTER_BUSY_MASK |
+		RLC_SERDES_NONCU_MASTER_BUSY__TC1_MASTER_BUSY_MASK;
+	for (k = 0; k < adev->usec_timeout; k++) {
+		if ((RREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_SERDES_NONCU_MASTER_BUSY) & mask) == 0)
+			break;
+		udelay(1);
+	}
+}
+
+static void gfx_v9_4_3_xcc_enable_gui_idle_interrupt(struct amdgpu_device *adev,
+						     bool enable, int xcc_id)
+{
+	u32 tmp;
+
+	/* These interrupts should be enabled to drive DS clock */
+
+	tmp = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_INT_CNTL_RING0);
+
+	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_BUSY_INT_ENABLE, enable ? 1 : 0);
+	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_EMPTY_INT_ENABLE, enable ? 1 : 0);
+	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CMP_BUSY_INT_ENABLE, enable ? 1 : 0);
+
+	WREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_INT_CNTL_RING0, tmp);
+}
+
+static void gfx_v9_4_3_xcc_rlc_stop(struct amdgpu_device *adev, int xcc_id)
+{
+	WREG32_FIELD15_PREREG(GC, GET_INST(GC, xcc_id), RLC_CNTL,
+			      RLC_ENABLE_F32, 0);
+	gfx_v9_4_3_xcc_enable_gui_idle_interrupt(adev, false, xcc_id);
+	gfx_v9_4_3_xcc_wait_for_rlc_serdes(adev, xcc_id);
+}
+
+static void gfx_v9_4_3_rlc_stop(struct amdgpu_device *adev)
+{
+	int i, num_xcc;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	for (i = 0; i < num_xcc; i++)
+		gfx_v9_4_3_xcc_rlc_stop(adev, i);
+}
+
+static void gfx_v9_4_3_xcc_rlc_reset(struct amdgpu_device *adev, int xcc_id)
+{
+	WREG32_FIELD15_PREREG(GC, GET_INST(GC, xcc_id), GRBM_SOFT_RESET,
+			      SOFT_RESET_RLC, 1);
+	udelay(50);
+	WREG32_FIELD15_PREREG(GC, GET_INST(GC, xcc_id), GRBM_SOFT_RESET,
+			      SOFT_RESET_RLC, 0);
+	udelay(50);
+}
+
+static void gfx_v9_4_3_rlc_reset(struct amdgpu_device *adev)
+{
+	int i, num_xcc;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	for (i = 0; i < num_xcc; i++)
+		gfx_v9_4_3_xcc_rlc_reset(adev, i);
+}
+
+static void gfx_v9_4_3_xcc_rlc_start(struct amdgpu_device *adev, int xcc_id)
+{
+	WREG32_FIELD15_PREREG(GC, GET_INST(GC, xcc_id), RLC_CNTL,
+			      RLC_ENABLE_F32, 1);
+	udelay(50);
+
+	/* carrizo do enable cp interrupt after cp inited */
+	if (!(adev->flags & AMD_IS_APU)) {
+		gfx_v9_4_3_xcc_enable_gui_idle_interrupt(adev, true, xcc_id);
+		udelay(50);
+	}
+}
+
+static void gfx_v9_4_3_rlc_start(struct amdgpu_device *adev)
+{
+#ifdef AMDGPU_RLC_DEBUG_RETRY
+	u32 rlc_ucode_ver;
+#endif
+	int i, num_xcc;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	for (i = 0; i < num_xcc; i++) {
+		gfx_v9_4_3_xcc_rlc_start(adev, i);
+#ifdef AMDGPU_RLC_DEBUG_RETRY
+		/* RLC_GPM_GENERAL_6 : RLC Ucode version */
+		rlc_ucode_ver = RREG32_SOC15(GC, GET_INST(GC, i), regRLC_GPM_GENERAL_6);
+		if (rlc_ucode_ver == 0x108) {
+			dev_info(adev->dev,
+				 "Using rlc debug ucode. regRLC_GPM_GENERAL_6 ==0x08%x / fw_ver == %i \n",
+				 rlc_ucode_ver, adev->gfx.rlc_fw_version);
+			/* RLC_GPM_TIMER_INT_3 : Timer interval in RefCLK cycles,
+			 * default is 0x9C4 to create a 100us interval */
+			WREG32_SOC15(GC, GET_INST(GC, i), regRLC_GPM_TIMER_INT_3, 0x9C4);
+			/* RLC_GPM_GENERAL_12 : Minimum gap between wptr and rptr
+			 * to disable the page fault retry interrupts, default is
+			 * 0x100 (256) */
+			WREG32_SOC15(GC, GET_INST(GC, i), regRLC_GPM_GENERAL_12, 0x100);
+		}
+#endif
+	}
+}
+
+static int gfx_v9_4_3_xcc_rlc_load_microcode(struct amdgpu_device *adev,
+					     int xcc_id)
+{
+	const struct rlc_firmware_header_v2_0 *hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+
+	if (!adev->gfx.rlc_fw)
+		return -EINVAL;
+
+	hdr = (const struct rlc_firmware_header_v2_0 *)adev->gfx.rlc_fw->data;
+	amdgpu_ucode_print_rlc_hdr(&hdr->header);
+
+	fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+			   le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+
+	WREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_GPM_UCODE_ADDR,
+			RLCG_UCODE_LOADING_START_ADDRESS);
+	for (i = 0; i < fw_size; i++) {
+		if (amdgpu_emu_mode == 1 && i % 100 == 0) {
+			dev_info(adev->dev, "Write RLC ucode data %u DWs\n", i);
+			msleep(1);
+		}
+		WREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_GPM_UCODE_DATA, le32_to_cpup(fw_data++));
+	}
+	WREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_GPM_UCODE_ADDR, adev->gfx.rlc_fw_version);
+
+	return 0;
+}
+
+static int gfx_v9_4_3_xcc_rlc_resume(struct amdgpu_device *adev, int xcc_id)
+{
+	int r;
+
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
+		gfx_v9_4_3_xcc_rlc_stop(adev, xcc_id);
+		/* legacy rlc firmware loading */
+		r = gfx_v9_4_3_xcc_rlc_load_microcode(adev, xcc_id);
+		if (r)
+			return r;
+		gfx_v9_4_3_xcc_rlc_start(adev, xcc_id);
+	}
+
+	amdgpu_gfx_rlc_enter_safe_mode(adev, xcc_id);
+	/* disable CG */
+	WREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_CGCG_CGLS_CTRL, 0);
+	gfx_v9_4_3_xcc_init_pg(adev, xcc_id);
+	amdgpu_gfx_rlc_exit_safe_mode(adev, xcc_id);
+
+	return 0;
+}
+
+static int gfx_v9_4_3_rlc_resume(struct amdgpu_device *adev)
+{
+	int r, i, num_xcc;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	for (i = 0; i < num_xcc; i++) {
+		r = gfx_v9_4_3_xcc_rlc_resume(adev, i);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+static void gfx_v9_4_3_update_spm_vmid(struct amdgpu_device *adev, struct amdgpu_ring *ring,
+				       unsigned vmid)
+{
+	u32 reg, pre_data, data;
+
+	reg = SOC15_REG_OFFSET(GC, GET_INST(GC, 0), regRLC_SPM_MC_CNTL);
+	if (amdgpu_sriov_is_pp_one_vf(adev) && !amdgpu_sriov_runtime(adev))
+		pre_data = RREG32_NO_KIQ(reg);
+	else
+		pre_data = RREG32(reg);
+
+	data =	pre_data & (~RLC_SPM_MC_CNTL__RLC_SPM_VMID_MASK);
+	data |= (vmid & RLC_SPM_MC_CNTL__RLC_SPM_VMID_MASK) << RLC_SPM_MC_CNTL__RLC_SPM_VMID__SHIFT;
+
+	if (pre_data != data) {
+		if (amdgpu_sriov_is_pp_one_vf(adev) && !amdgpu_sriov_runtime(adev)) {
+			WREG32_SOC15_NO_KIQ(GC, GET_INST(GC, 0), regRLC_SPM_MC_CNTL, data);
+		} else
+			WREG32_SOC15(GC, GET_INST(GC, 0), regRLC_SPM_MC_CNTL, data);
+	}
+}
+
+static const struct soc15_reg_rlcg rlcg_access_gc_9_4_3[] = {
+	{SOC15_REG_ENTRY(GC, 0, regGRBM_GFX_INDEX)},
+	{SOC15_REG_ENTRY(GC, 0, regSQ_IND_INDEX)},
+};
+
+static bool gfx_v9_4_3_check_rlcg_range(struct amdgpu_device *adev,
+					uint32_t offset,
+					struct soc15_reg_rlcg *entries, int arr_size)
+{
+	int i, inst;
+	uint32_t reg;
+
+	if (!entries)
+		return false;
+
+	for (i = 0; i < arr_size; i++) {
+		const struct soc15_reg_rlcg *entry;
+
+		entry = &entries[i];
+		inst = adev->ip_map.logical_to_dev_inst ?
+			       adev->ip_map.logical_to_dev_inst(
+				       adev, entry->hwip, entry->instance) :
+			       entry->instance;
+		reg = adev->reg_offset[entry->hwip][inst][entry->segment] +
+		      entry->reg;
+		if (offset == reg)
+			return true;
+	}
+
+	return false;
+}
+
+static bool gfx_v9_4_3_is_rlcg_access_range(struct amdgpu_device *adev, u32 offset)
+{
+	return gfx_v9_4_3_check_rlcg_range(adev, offset,
+					(void *)rlcg_access_gc_9_4_3,
+					ARRAY_SIZE(rlcg_access_gc_9_4_3));
+}
+
+static void gfx_v9_4_3_xcc_cp_compute_enable(struct amdgpu_device *adev,
+					     bool enable, int xcc_id)
+{
+	if (enable) {
+		WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_MEC_CNTL, 0);
+	} else {
+		WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_MEC_CNTL,
+			(CP_MEC_CNTL__MEC_INVALIDATE_ICACHE_MASK |
+			 CP_MEC_CNTL__MEC_ME1_PIPE0_RESET_MASK |
+			 CP_MEC_CNTL__MEC_ME1_PIPE1_RESET_MASK |
+			 CP_MEC_CNTL__MEC_ME1_PIPE2_RESET_MASK |
+			 CP_MEC_CNTL__MEC_ME1_PIPE3_RESET_MASK |
+			 CP_MEC_CNTL__MEC_ME2_PIPE0_RESET_MASK |
+			 CP_MEC_CNTL__MEC_ME2_PIPE1_RESET_MASK |
+			 CP_MEC_CNTL__MEC_ME1_HALT_MASK |
+			 CP_MEC_CNTL__MEC_ME2_HALT_MASK));
+		adev->gfx.kiq[xcc_id].ring.sched.ready = false;
+	}
+	udelay(50);
+}
+
+static int gfx_v9_4_3_xcc_cp_compute_load_microcode(struct amdgpu_device *adev,
+						    int xcc_id)
+{
+	const struct gfx_firmware_header_v1_0 *mec_hdr;
+	const __le32 *fw_data;
+	unsigned i;
+	u32 tmp;
+	u32 mec_ucode_addr_offset;
+	u32 mec_ucode_data_offset;
+
+	if (!adev->gfx.mec_fw)
+		return -EINVAL;
+
+	gfx_v9_4_3_xcc_cp_compute_enable(adev, false, xcc_id);
+
+	mec_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+	amdgpu_ucode_print_gfx_hdr(&mec_hdr->header);
+
+	fw_data = (const __le32 *)
+		(adev->gfx.mec_fw->data +
+		 le32_to_cpu(mec_hdr->header.ucode_array_offset_bytes));
+	tmp = 0;
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, CACHE_POLICY, 0);
+	WREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_CPC_IC_BASE_CNTL, tmp);
+
+	WREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_CPC_IC_BASE_LO,
+		adev->gfx.mec.mec_fw_gpu_addr & 0xFFFFF000);
+	WREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_CPC_IC_BASE_HI,
+		upper_32_bits(adev->gfx.mec.mec_fw_gpu_addr));
+
+	mec_ucode_addr_offset =
+		SOC15_REG_OFFSET(GC, GET_INST(GC, xcc_id), regCP_MEC_ME1_UCODE_ADDR);
+	mec_ucode_data_offset =
+		SOC15_REG_OFFSET(GC, GET_INST(GC, xcc_id), regCP_MEC_ME1_UCODE_DATA);
+
+	/* MEC1 */
+	WREG32(mec_ucode_addr_offset, mec_hdr->jt_offset);
+	for (i = 0; i < mec_hdr->jt_size; i++)
+		WREG32(mec_ucode_data_offset,
+		       le32_to_cpup(fw_data + mec_hdr->jt_offset + i));
+
+	WREG32(mec_ucode_addr_offset, adev->gfx.mec_fw_version);
+	/* Todo : Loading MEC2 firmware is only necessary if MEC2 should run different microcode than MEC1. */
+
+	return 0;
+}
+
+/* KIQ functions */
+static void gfx_v9_4_3_xcc_kiq_setting(struct amdgpu_ring *ring, int xcc_id)
+{
+	uint32_t tmp;
+	struct amdgpu_device *adev = ring->adev;
+
+	/* tell RLC which is KIQ queue */
+	tmp = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_CP_SCHEDULERS);
+	tmp &= 0xffffff00;
+	tmp |= (ring->me << 5) | (ring->pipe << 3) | (ring->queue);
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regRLC_CP_SCHEDULERS, tmp | 0x80);
+}
+
+static void gfx_v9_4_3_mqd_set_priority(struct amdgpu_ring *ring, struct v9_mqd *mqd)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE) {
+		if (amdgpu_gfx_is_high_priority_compute_queue(adev, ring)) {
+			mqd->cp_hqd_pipe_priority = AMDGPU_GFX_PIPE_PRIO_HIGH;
+			mqd->cp_hqd_queue_priority =
+				AMDGPU_GFX_QUEUE_PRIORITY_MAXIMUM;
+		}
+	}
+}
+
+static int gfx_v9_4_3_xcc_mqd_init(struct amdgpu_ring *ring, int xcc_id)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v9_mqd *mqd = ring->mqd_ptr;
+	uint64_t hqd_gpu_addr, wb_gpu_addr, eop_base_addr;
+	uint32_t tmp;
+
+	mqd->header = 0xC0310800;
+	mqd->compute_pipelinestat_enable = 0x00000001;
+	mqd->compute_static_thread_mgmt_se0 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se1 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se2 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se3 = 0xffffffff;
+	mqd->compute_misc_reserved = 0x00000003;
+
+	mqd->dynamic_cu_mask_addr_lo =
+		lower_32_bits(ring->mqd_gpu_addr
+			      + offsetof(struct v9_mqd_allocation, dynamic_cu_mask));
+	mqd->dynamic_cu_mask_addr_hi =
+		upper_32_bits(ring->mqd_gpu_addr
+			      + offsetof(struct v9_mqd_allocation, dynamic_cu_mask));
+
+	eop_base_addr = ring->eop_gpu_addr >> 8;
+	mqd->cp_hqd_eop_base_addr_lo = eop_base_addr;
+	mqd->cp_hqd_eop_base_addr_hi = upper_32_bits(eop_base_addr);
+
+	/* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */
+	tmp = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_HQD_EOP_CONTROL);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_EOP_CONTROL, EOP_SIZE,
+			(order_base_2(GFX9_MEC_HPD_SIZE / 4) - 1));
+
+	mqd->cp_hqd_eop_control = tmp;
+
+	/* enable doorbell? */
+	tmp = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_HQD_PQ_DOORBELL_CONTROL);
+
+	if (ring->use_doorbell) {
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_OFFSET, ring->doorbell_index);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_EN, 1);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_SOURCE, 0);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_HIT, 0);
+		if (amdgpu_sriov_multi_vf_mode(adev))
+			tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+					    DOORBELL_MODE, 1);
+	} else {
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+					 DOORBELL_EN, 0);
+	}
+
+	mqd->cp_hqd_pq_doorbell_control = tmp;
+
+	/* disable the queue if it's active */
+	ring->wptr = 0;
+	mqd->cp_hqd_dequeue_request = 0;
+	mqd->cp_hqd_pq_rptr = 0;
+	mqd->cp_hqd_pq_wptr_lo = 0;
+	mqd->cp_hqd_pq_wptr_hi = 0;
+
+	/* set the pointer to the MQD */
+	mqd->cp_mqd_base_addr_lo = ring->mqd_gpu_addr & 0xfffffffc;
+	mqd->cp_mqd_base_addr_hi = upper_32_bits(ring->mqd_gpu_addr);
+
+	/* set MQD vmid to 0 */
+	tmp = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_MQD_CONTROL);
+	tmp = REG_SET_FIELD(tmp, CP_MQD_CONTROL, VMID, 0);
+	mqd->cp_mqd_control = tmp;
+
+	/* set the pointer to the HQD, this is similar CP_RB0_BASE/_HI */
+	hqd_gpu_addr = ring->gpu_addr >> 8;
+	mqd->cp_hqd_pq_base_lo = hqd_gpu_addr;
+	mqd->cp_hqd_pq_base_hi = upper_32_bits(hqd_gpu_addr);
+
+	/* set up the HQD, this is similar to CP_RB0_CNTL */
+	tmp = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_HQD_PQ_CONTROL);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, QUEUE_SIZE,
+			    (order_base_2(ring->ring_size / 4) - 1));
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, RPTR_BLOCK_SIZE,
+			((order_base_2(AMDGPU_GPU_PAGE_SIZE / 4) - 1) << 8));
+#ifdef __BIG_ENDIAN
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, ENDIAN_SWAP, 1);
+#endif
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, UNORD_DISPATCH, 0);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, ROQ_PQ_IB_FLIP, 0);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, PRIV_STATE, 1);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, KMD_QUEUE, 1);
+	mqd->cp_hqd_pq_control = tmp;
+
+	/* set the wb address whether it's enabled or not */
+	wb_gpu_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+	mqd->cp_hqd_pq_rptr_report_addr_lo = wb_gpu_addr & 0xfffffffc;
+	mqd->cp_hqd_pq_rptr_report_addr_hi =
+		upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	/* only used if CP_PQ_WPTR_POLL_CNTL.CP_PQ_WPTR_POLL_CNTL__EN_MASK=1 */
+	wb_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
+	mqd->cp_hqd_pq_wptr_poll_addr_lo = wb_gpu_addr & 0xfffffffc;
+	mqd->cp_hqd_pq_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	/* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */
+	ring->wptr = 0;
+	mqd->cp_hqd_pq_rptr = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_HQD_PQ_RPTR);
+
+	/* set the vmid for the queue */
+	mqd->cp_hqd_vmid = 0;
+
+	tmp = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_HQD_PERSISTENT_STATE);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PERSISTENT_STATE, PRELOAD_SIZE, 0x53);
+	mqd->cp_hqd_persistent_state = tmp;
+
+	/* set MIN_IB_AVAIL_SIZE */
+	tmp = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_HQD_IB_CONTROL);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_IB_CONTROL, MIN_IB_AVAIL_SIZE, 3);
+	mqd->cp_hqd_ib_control = tmp;
+
+	/* set static priority for a queue/ring */
+	gfx_v9_4_3_mqd_set_priority(ring, mqd);
+	mqd->cp_hqd_quantum = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_HQD_QUANTUM);
+
+	/* map_queues packet doesn't need activate the queue,
+	 * so only kiq need set this field.
+	 */
+	if (ring->funcs->type == AMDGPU_RING_TYPE_KIQ)
+		mqd->cp_hqd_active = 1;
+
+	return 0;
+}
+
+static int gfx_v9_4_3_xcc_kiq_init_register(struct amdgpu_ring *ring,
+					    int xcc_id)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v9_mqd *mqd = ring->mqd_ptr;
+	int j;
+
+	/* disable wptr polling */
+	WREG32_FIELD15_PREREG(GC, GET_INST(GC, xcc_id), CP_PQ_WPTR_POLL_CNTL, EN, 0);
+
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_EOP_BASE_ADDR,
+	       mqd->cp_hqd_eop_base_addr_lo);
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_EOP_BASE_ADDR_HI,
+	       mqd->cp_hqd_eop_base_addr_hi);
+
+	/* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_EOP_CONTROL,
+	       mqd->cp_hqd_eop_control);
+
+	/* enable doorbell? */
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_PQ_DOORBELL_CONTROL,
+	       mqd->cp_hqd_pq_doorbell_control);
+
+	/* disable the queue if it's active */
+	if (RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_HQD_ACTIVE) & 1) {
+		WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_DEQUEUE_REQUEST, 1);
+		for (j = 0; j < adev->usec_timeout; j++) {
+			if (!(RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_HQD_ACTIVE) & 1))
+				break;
+			udelay(1);
+		}
+		WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_DEQUEUE_REQUEST,
+		       mqd->cp_hqd_dequeue_request);
+		WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_PQ_RPTR,
+		       mqd->cp_hqd_pq_rptr);
+		WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_PQ_WPTR_LO,
+		       mqd->cp_hqd_pq_wptr_lo);
+		WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_PQ_WPTR_HI,
+		       mqd->cp_hqd_pq_wptr_hi);
+	}
+
+	/* set the pointer to the MQD */
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_MQD_BASE_ADDR,
+	       mqd->cp_mqd_base_addr_lo);
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_MQD_BASE_ADDR_HI,
+	       mqd->cp_mqd_base_addr_hi);
+
+	/* set MQD vmid to 0 */
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_MQD_CONTROL,
+	       mqd->cp_mqd_control);
+
+	/* set the pointer to the HQD, this is similar CP_RB0_BASE/_HI */
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_PQ_BASE,
+	       mqd->cp_hqd_pq_base_lo);
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_PQ_BASE_HI,
+	       mqd->cp_hqd_pq_base_hi);
+
+	/* set up the HQD, this is similar to CP_RB0_CNTL */
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_PQ_CONTROL,
+	       mqd->cp_hqd_pq_control);
+
+	/* set the wb address whether it's enabled or not */
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_PQ_RPTR_REPORT_ADDR,
+				mqd->cp_hqd_pq_rptr_report_addr_lo);
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_PQ_RPTR_REPORT_ADDR_HI,
+				mqd->cp_hqd_pq_rptr_report_addr_hi);
+
+	/* only used if CP_PQ_WPTR_POLL_CNTL.CP_PQ_WPTR_POLL_CNTL__EN_MASK=1 */
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_PQ_WPTR_POLL_ADDR,
+	       mqd->cp_hqd_pq_wptr_poll_addr_lo);
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_PQ_WPTR_POLL_ADDR_HI,
+	       mqd->cp_hqd_pq_wptr_poll_addr_hi);
+
+	/* enable the doorbell if requested */
+	if (ring->use_doorbell) {
+		WREG32_SOC15(
+			GC, GET_INST(GC, xcc_id),
+			regCP_MEC_DOORBELL_RANGE_LOWER,
+			((adev->doorbell_index.kiq +
+			  xcc_id * adev->doorbell_index.xcc_doorbell_range) *
+			 2) << 2);
+		WREG32_SOC15(
+			GC, GET_INST(GC, xcc_id),
+			regCP_MEC_DOORBELL_RANGE_UPPER,
+			((adev->doorbell_index.userqueue_end +
+			  xcc_id * adev->doorbell_index.xcc_doorbell_range) *
+			 2) << 2);
+	}
+
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_PQ_DOORBELL_CONTROL,
+	       mqd->cp_hqd_pq_doorbell_control);
+
+	/* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_PQ_WPTR_LO,
+	       mqd->cp_hqd_pq_wptr_lo);
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_PQ_WPTR_HI,
+	       mqd->cp_hqd_pq_wptr_hi);
+
+	/* set the vmid for the queue */
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_VMID, mqd->cp_hqd_vmid);
+
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_PERSISTENT_STATE,
+	       mqd->cp_hqd_persistent_state);
+
+	/* activate the queue */
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_ACTIVE,
+	       mqd->cp_hqd_active);
+
+	if (ring->use_doorbell)
+		WREG32_FIELD15_PREREG(GC, GET_INST(GC, xcc_id), CP_PQ_STATUS, DOORBELL_ENABLE, 1);
+
+	return 0;
+}
+
+static int gfx_v9_4_3_xcc_q_fini_register(struct amdgpu_ring *ring,
+					    int xcc_id)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int j;
+
+	/* disable the queue if it's active */
+	if (RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_HQD_ACTIVE) & 1) {
+
+		WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_DEQUEUE_REQUEST, 1);
+
+		for (j = 0; j < adev->usec_timeout; j++) {
+			if (!(RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_HQD_ACTIVE) & 1))
+				break;
+			udelay(1);
+		}
+
+		if (j == AMDGPU_MAX_USEC_TIMEOUT) {
+			DRM_DEBUG("%s dequeue request failed.\n", ring->name);
+
+			/* Manual disable if dequeue request times out */
+			WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_ACTIVE, 0);
+		}
+
+		WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_DEQUEUE_REQUEST,
+		      0);
+	}
+
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_IQ_TIMER, 0);
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_IB_CONTROL, 0);
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_PERSISTENT_STATE, CP_HQD_PERSISTENT_STATE_DEFAULT);
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_PQ_DOORBELL_CONTROL, 0x40000000);
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_PQ_DOORBELL_CONTROL, 0);
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_PQ_RPTR, 0);
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_PQ_WPTR_HI, 0);
+	WREG32_SOC15_RLC(GC, GET_INST(GC, xcc_id), regCP_HQD_PQ_WPTR_LO, 0);
+
+	return 0;
+}
+
+static int gfx_v9_4_3_xcc_kiq_init_queue(struct amdgpu_ring *ring, int xcc_id)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v9_mqd *mqd = ring->mqd_ptr;
+	struct v9_mqd *tmp_mqd;
+
+	gfx_v9_4_3_xcc_kiq_setting(ring, xcc_id);
+
+	/* GPU could be in bad state during probe, driver trigger the reset
+	 * after load the SMU, in this case , the mqd is not be initialized.
+	 * driver need to re-init the mqd.
+	 * check mqd->cp_hqd_pq_control since this value should not be 0
+	 */
+	tmp_mqd = (struct v9_mqd *)adev->gfx.kiq[xcc_id].mqd_backup;
+	if (amdgpu_in_reset(adev) && tmp_mqd->cp_hqd_pq_control) {
+		/* for GPU_RESET case , reset MQD to a clean status */
+		if (adev->gfx.kiq[xcc_id].mqd_backup)
+			memcpy(mqd, adev->gfx.kiq[xcc_id].mqd_backup, sizeof(struct v9_mqd_allocation));
+
+		/* reset ring buffer */
+		ring->wptr = 0;
+		amdgpu_ring_clear_ring(ring);
+		mutex_lock(&adev->srbm_mutex);
+		soc15_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0, GET_INST(GC, xcc_id));
+		gfx_v9_4_3_xcc_kiq_init_register(ring, xcc_id);
+		soc15_grbm_select(adev, 0, 0, 0, 0, GET_INST(GC, xcc_id));
+		mutex_unlock(&adev->srbm_mutex);
+	} else {
+		memset((void *)mqd, 0, sizeof(struct v9_mqd_allocation));
+		((struct v9_mqd_allocation *)mqd)->dynamic_cu_mask = 0xFFFFFFFF;
+		((struct v9_mqd_allocation *)mqd)->dynamic_rb_mask = 0xFFFFFFFF;
+		mutex_lock(&adev->srbm_mutex);
+		if (amdgpu_sriov_vf(adev) && adev->in_suspend)
+			amdgpu_ring_clear_ring(ring);
+		soc15_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0, GET_INST(GC, xcc_id));
+		gfx_v9_4_3_xcc_mqd_init(ring, xcc_id);
+		gfx_v9_4_3_xcc_kiq_init_register(ring, xcc_id);
+		soc15_grbm_select(adev, 0, 0, 0, 0, GET_INST(GC, xcc_id));
+		mutex_unlock(&adev->srbm_mutex);
+
+		if (adev->gfx.kiq[xcc_id].mqd_backup)
+			memcpy(adev->gfx.kiq[xcc_id].mqd_backup, mqd, sizeof(struct v9_mqd_allocation));
+	}
+
+	return 0;
+}
+
+static int gfx_v9_4_3_xcc_kcq_init_queue(struct amdgpu_ring *ring, int xcc_id, bool restore)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v9_mqd *mqd = ring->mqd_ptr;
+	int mqd_idx = ring - &adev->gfx.compute_ring[0];
+	struct v9_mqd *tmp_mqd;
+
+	/* Same as above kiq init, driver need to re-init the mqd if mqd->cp_hqd_pq_control
+	 * is not be initialized before
+	 */
+	tmp_mqd = (struct v9_mqd *)adev->gfx.mec.mqd_backup[mqd_idx];
+
+	if (!restore && (!tmp_mqd->cp_hqd_pq_control ||
+	    (!amdgpu_in_reset(adev) && !adev->in_suspend))) {
+		memset((void *)mqd, 0, sizeof(struct v9_mqd_allocation));
+		((struct v9_mqd_allocation *)mqd)->dynamic_cu_mask = 0xFFFFFFFF;
+		((struct v9_mqd_allocation *)mqd)->dynamic_rb_mask = 0xFFFFFFFF;
+		mutex_lock(&adev->srbm_mutex);
+		soc15_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0, GET_INST(GC, xcc_id));
+		gfx_v9_4_3_xcc_mqd_init(ring, xcc_id);
+		soc15_grbm_select(adev, 0, 0, 0, 0, GET_INST(GC, xcc_id));
+		mutex_unlock(&adev->srbm_mutex);
+
+		if (adev->gfx.mec.mqd_backup[mqd_idx])
+			memcpy(adev->gfx.mec.mqd_backup[mqd_idx], mqd, sizeof(struct v9_mqd_allocation));
+	} else {
+		/* restore MQD to a clean status */
+		if (adev->gfx.mec.mqd_backup[mqd_idx])
+			memcpy(mqd, adev->gfx.mec.mqd_backup[mqd_idx], sizeof(struct v9_mqd_allocation));
+		/* reset ring buffer */
+		ring->wptr = 0;
+		atomic64_set((atomic64_t *)&adev->wb.wb[ring->wptr_offs], 0);
+		amdgpu_ring_clear_ring(ring);
+	}
+
+	return 0;
+}
+
+static int gfx_v9_4_3_xcc_kcq_fini_register(struct amdgpu_device *adev, int xcc_id)
+{
+	struct amdgpu_ring *ring;
+	int j;
+
+	for (j = 0; j < adev->gfx.num_compute_rings; j++) {
+		ring = &adev->gfx.compute_ring[j +  xcc_id * adev->gfx.num_compute_rings];
+		if (!amdgpu_in_reset(adev) && !adev->in_suspend) {
+			mutex_lock(&adev->srbm_mutex);
+			soc15_grbm_select(adev, ring->me,
+					ring->pipe,
+					ring->queue, 0, GET_INST(GC, xcc_id));
+			gfx_v9_4_3_xcc_q_fini_register(ring, xcc_id);
+			soc15_grbm_select(adev, 0, 0, 0, 0, GET_INST(GC, xcc_id));
+			mutex_unlock(&adev->srbm_mutex);
+		}
+	}
+
+	return 0;
+}
+
+static int gfx_v9_4_3_xcc_kiq_resume(struct amdgpu_device *adev, int xcc_id)
+{
+	gfx_v9_4_3_xcc_kiq_init_queue(&adev->gfx.kiq[xcc_id].ring, xcc_id);
+	return 0;
+}
+
+static int gfx_v9_4_3_xcc_kcq_resume(struct amdgpu_device *adev, int xcc_id)
+{
+	struct amdgpu_ring *ring;
+	int i, r;
+
+	gfx_v9_4_3_xcc_cp_compute_enable(adev, true, xcc_id);
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		ring = &adev->gfx.compute_ring[i + xcc_id *
+			adev->gfx.num_compute_rings];
+
+		r = gfx_v9_4_3_xcc_kcq_init_queue(ring, xcc_id, false);
+		if (r)
+			return r;
+	}
+
+	return amdgpu_gfx_enable_kcq(adev, xcc_id);
+}
+
+static int gfx_v9_4_3_xcc_cp_resume(struct amdgpu_device *adev, int xcc_id)
+{
+	struct amdgpu_ring *ring;
+	int r, j;
+
+	gfx_v9_4_3_xcc_enable_gui_idle_interrupt(adev, false, xcc_id);
+
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
+		gfx_v9_4_3_xcc_disable_gpa_mode(adev, xcc_id);
+
+		r = gfx_v9_4_3_xcc_cp_compute_load_microcode(adev, xcc_id);
+		if (r)
+			return r;
+	} else {
+		gfx_v9_4_3_xcc_cp_compute_enable(adev, false, xcc_id);
+	}
+
+	r = gfx_v9_4_3_xcc_kiq_resume(adev, xcc_id);
+	if (r)
+		return r;
+
+	r = gfx_v9_4_3_xcc_kcq_resume(adev, xcc_id);
+	if (r)
+		return r;
+
+	for (j = 0; j < adev->gfx.num_compute_rings; j++) {
+		ring = &adev->gfx.compute_ring
+				[j + xcc_id * adev->gfx.num_compute_rings];
+		r = amdgpu_ring_test_helper(ring);
+		if (r)
+			return r;
+	}
+
+	gfx_v9_4_3_xcc_enable_gui_idle_interrupt(adev, true, xcc_id);
+
+	return 0;
+}
+
+static int gfx_v9_4_3_cp_resume(struct amdgpu_device *adev)
+{
+	int r = 0, i, num_xcc, num_xcp, num_xcc_per_xcp;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	if (amdgpu_sriov_vf(adev)) {
+		enum amdgpu_gfx_partition mode;
+
+		mode = amdgpu_xcp_query_partition_mode(adev->xcp_mgr,
+						       AMDGPU_XCP_FL_NONE);
+		if (mode == AMDGPU_UNKNOWN_COMPUTE_PARTITION_MODE)
+			return -EINVAL;
+		num_xcc_per_xcp = gfx_v9_4_3_get_xccs_per_xcp(adev);
+		adev->gfx.num_xcc_per_xcp = num_xcc_per_xcp;
+		num_xcp = num_xcc / num_xcc_per_xcp;
+		r = amdgpu_xcp_init(adev->xcp_mgr, num_xcp, mode);
+
+	} else {
+		if (amdgpu_xcp_query_partition_mode(adev->xcp_mgr,
+						    AMDGPU_XCP_FL_NONE) ==
+		    AMDGPU_UNKNOWN_COMPUTE_PARTITION_MODE)
+			r = amdgpu_xcp_switch_partition_mode(
+				adev->xcp_mgr, amdgpu_user_partt_mode);
+	}
+	if (r)
+		return r;
+
+	for (i = 0; i < num_xcc; i++) {
+		r = gfx_v9_4_3_xcc_cp_resume(adev, i);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+static void gfx_v9_4_3_xcc_fini(struct amdgpu_device *adev, int xcc_id)
+{
+	if (amdgpu_gfx_disable_kcq(adev, xcc_id))
+		DRM_ERROR("XCD %d KCQ disable failed\n", xcc_id);
+
+	if (amdgpu_sriov_vf(adev)) {
+		/* must disable polling for SRIOV when hw finished, otherwise
+		 * CPC engine may still keep fetching WB address which is already
+		 * invalid after sw finished and trigger DMAR reading error in
+		 * hypervisor side.
+		 */
+		WREG32_FIELD15_PREREG(GC, GET_INST(GC, xcc_id), CP_PQ_WPTR_POLL_CNTL, EN, 0);
+		return;
+	}
+
+	/* Use deinitialize sequence from CAIL when unbinding device
+	 * from driver, otherwise KIQ is hanging when binding back
+	 */
+	if (!amdgpu_in_reset(adev) && !adev->in_suspend) {
+		mutex_lock(&adev->srbm_mutex);
+		soc15_grbm_select(adev, adev->gfx.kiq[xcc_id].ring.me,
+				  adev->gfx.kiq[xcc_id].ring.pipe,
+				  adev->gfx.kiq[xcc_id].ring.queue, 0,
+				  GET_INST(GC, xcc_id));
+		gfx_v9_4_3_xcc_q_fini_register(&adev->gfx.kiq[xcc_id].ring,
+						 xcc_id);
+		soc15_grbm_select(adev, 0, 0, 0, 0, GET_INST(GC, xcc_id));
+		mutex_unlock(&adev->srbm_mutex);
+	}
+
+	gfx_v9_4_3_xcc_kcq_fini_register(adev, xcc_id);
+	gfx_v9_4_3_xcc_cp_compute_enable(adev, false, xcc_id);
+}
+
+static int gfx_v9_4_3_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	amdgpu_gfx_cleaner_shader_init(adev, adev->gfx.cleaner_shader_size,
+				       adev->gfx.cleaner_shader_ptr);
+
+	if (!amdgpu_sriov_vf(adev))
+		gfx_v9_4_3_init_golden_registers(adev);
+
+	gfx_v9_4_3_constants_init(adev);
+
+	r = adev->gfx.rlc.funcs->resume(adev);
+	if (r)
+		return r;
+
+	r = gfx_v9_4_3_cp_resume(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int gfx_v9_4_3_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, num_xcc;
+
+	amdgpu_irq_put(adev, &adev->gfx.priv_reg_irq, 0);
+	amdgpu_irq_put(adev, &adev->gfx.priv_inst_irq, 0);
+	amdgpu_irq_put(adev, &adev->gfx.bad_op_irq, 0);
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	for (i = 0; i < num_xcc; i++) {
+		gfx_v9_4_3_xcc_fini(adev, i);
+	}
+
+	return 0;
+}
+
+static int gfx_v9_4_3_suspend(struct amdgpu_ip_block *ip_block)
+{
+	return gfx_v9_4_3_hw_fini(ip_block);
+}
+
+static int gfx_v9_4_3_resume(struct amdgpu_ip_block *ip_block)
+{
+	return gfx_v9_4_3_hw_init(ip_block);
+}
+
+static bool gfx_v9_4_3_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, num_xcc;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	for (i = 0; i < num_xcc; i++) {
+		if (REG_GET_FIELD(RREG32_SOC15(GC, GET_INST(GC, i), regGRBM_STATUS),
+					GRBM_STATUS, GUI_ACTIVE))
+			return false;
+	}
+	return true;
+}
+
+static int gfx_v9_4_3_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	unsigned i;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (gfx_v9_4_3_is_idle(ip_block))
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static int gfx_v9_4_3_soft_reset(struct amdgpu_ip_block *ip_block)
+{
+	u32 grbm_soft_reset = 0;
+	u32 tmp;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	/* GRBM_STATUS */
+	tmp = RREG32_SOC15(GC, GET_INST(GC, 0), regGRBM_STATUS);
+	if (tmp & (GRBM_STATUS__PA_BUSY_MASK | GRBM_STATUS__SC_BUSY_MASK |
+		   GRBM_STATUS__BCI_BUSY_MASK | GRBM_STATUS__SX_BUSY_MASK |
+		   GRBM_STATUS__TA_BUSY_MASK | GRBM_STATUS__VGT_BUSY_MASK |
+		   GRBM_STATUS__DB_BUSY_MASK | GRBM_STATUS__CB_BUSY_MASK |
+		   GRBM_STATUS__GDS_BUSY_MASK | GRBM_STATUS__SPI_BUSY_MASK |
+		   GRBM_STATUS__IA_BUSY_MASK | GRBM_STATUS__IA_BUSY_NO_DMA_MASK)) {
+		grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset,
+						GRBM_SOFT_RESET, SOFT_RESET_CP, 1);
+		grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset,
+						GRBM_SOFT_RESET, SOFT_RESET_GFX, 1);
+	}
+
+	if (tmp & (GRBM_STATUS__CP_BUSY_MASK | GRBM_STATUS__CP_COHERENCY_BUSY_MASK)) {
+		grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset,
+						GRBM_SOFT_RESET, SOFT_RESET_CP, 1);
+	}
+
+	/* GRBM_STATUS2 */
+	tmp = RREG32_SOC15(GC, GET_INST(GC, 0), regGRBM_STATUS2);
+	if (REG_GET_FIELD(tmp, GRBM_STATUS2, RLC_BUSY))
+		grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset,
+						GRBM_SOFT_RESET, SOFT_RESET_RLC, 1);
+
+
+	if (grbm_soft_reset) {
+		/* stop the rlc */
+		adev->gfx.rlc.funcs->stop(adev);
+
+		/* Disable MEC parsing/prefetching */
+		gfx_v9_4_3_xcc_cp_compute_enable(adev, false, 0);
+
+		tmp = RREG32_SOC15(GC, GET_INST(GC, 0), regGRBM_SOFT_RESET);
+		tmp |= grbm_soft_reset;
+		dev_info(adev->dev, "GRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32_SOC15(GC, GET_INST(GC, 0), regGRBM_SOFT_RESET, tmp);
+		tmp = RREG32_SOC15(GC, GET_INST(GC, 0), regGRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~grbm_soft_reset;
+		WREG32_SOC15(GC, GET_INST(GC, 0), regGRBM_SOFT_RESET, tmp);
+		tmp = RREG32_SOC15(GC, GET_INST(GC, 0), regGRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+	}
+	return 0;
+}
+
+static void gfx_v9_4_3_ring_emit_gds_switch(struct amdgpu_ring *ring,
+					  uint32_t vmid,
+					  uint32_t gds_base, uint32_t gds_size,
+					  uint32_t gws_base, uint32_t gws_size,
+					  uint32_t oa_base, uint32_t oa_size)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	/* GDS Base */
+	gfx_v9_4_3_write_data_to_reg(ring, 0, false,
+				   SOC15_REG_OFFSET(GC, GET_INST(GC, 0), regGDS_VMID0_BASE) + 2 * vmid,
+				   gds_base);
+
+	/* GDS Size */
+	gfx_v9_4_3_write_data_to_reg(ring, 0, false,
+				   SOC15_REG_OFFSET(GC, GET_INST(GC, 0), regGDS_VMID0_SIZE) + 2 * vmid,
+				   gds_size);
+
+	/* GWS */
+	gfx_v9_4_3_write_data_to_reg(ring, 0, false,
+				   SOC15_REG_OFFSET(GC, GET_INST(GC, 0), regGDS_GWS_VMID0) + vmid,
+				   gws_size << GDS_GWS_VMID0__SIZE__SHIFT | gws_base);
+
+	/* OA */
+	gfx_v9_4_3_write_data_to_reg(ring, 0, false,
+				   SOC15_REG_OFFSET(GC, GET_INST(GC, 0), regGDS_OA_VMID0) + vmid,
+				   (1 << (oa_size + oa_base)) - (1 << oa_base));
+}
+
+static int gfx_v9_4_3_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	adev->gfx.num_compute_rings = min(amdgpu_gfx_get_num_kcq(adev),
+					  AMDGPU_MAX_COMPUTE_RINGS);
+	gfx_v9_4_3_set_kiq_pm4_funcs(adev);
+	gfx_v9_4_3_set_ring_funcs(adev);
+	gfx_v9_4_3_set_irq_funcs(adev);
+	gfx_v9_4_3_set_gds_init(adev);
+	gfx_v9_4_3_set_rlc_funcs(adev);
+
+	/* init rlcg reg access ctrl */
+	gfx_v9_4_3_init_rlcg_reg_access_ctrl(adev);
+
+	return gfx_v9_4_3_init_microcode(adev);
+}
+
+static int gfx_v9_4_3_late_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = amdgpu_irq_get(adev, &adev->gfx.priv_reg_irq, 0);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_get(adev, &adev->gfx.priv_inst_irq, 0);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_get(adev, &adev->gfx.bad_op_irq, 0);
+	if (r)
+		return r;
+
+	if (adev->gfx.ras &&
+	    adev->gfx.ras->enable_watchdog_timer)
+		adev->gfx.ras->enable_watchdog_timer(adev);
+
+	return 0;
+}
+
+static void gfx_v9_4_3_xcc_update_sram_fgcg(struct amdgpu_device *adev,
+					    bool enable, int xcc_id)
+{
+	uint32_t def, data;
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_GFX_FGCG))
+		return;
+
+	def = data = RREG32_SOC15(GC, GET_INST(GC, xcc_id),
+				  regRLC_CGTT_MGCG_OVERRIDE);
+
+	if (enable)
+		data &= ~RLC_CGTT_MGCG_OVERRIDE__GFXIP_FGCG_OVERRIDE_MASK;
+	else
+		data |= RLC_CGTT_MGCG_OVERRIDE__GFXIP_FGCG_OVERRIDE_MASK;
+
+	if (def != data)
+		WREG32_SOC15(GC, GET_INST(GC, xcc_id),
+			     regRLC_CGTT_MGCG_OVERRIDE, data);
+
+}
+
+static void gfx_v9_4_3_xcc_update_repeater_fgcg(struct amdgpu_device *adev,
+						bool enable, int xcc_id)
+{
+	uint32_t def, data;
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_REPEATER_FGCG))
+		return;
+
+	def = data = RREG32_SOC15(GC, GET_INST(GC, xcc_id),
+				  regRLC_CGTT_MGCG_OVERRIDE);
+
+	if (enable)
+		data &= ~RLC_CGTT_MGCG_OVERRIDE__GFXIP_REP_FGCG_OVERRIDE_MASK;
+	else
+		data |= RLC_CGTT_MGCG_OVERRIDE__GFXIP_REP_FGCG_OVERRIDE_MASK;
+
+	if (def != data)
+		WREG32_SOC15(GC, GET_INST(GC, xcc_id),
+			     regRLC_CGTT_MGCG_OVERRIDE, data);
+}
+
+static void
+gfx_v9_4_3_xcc_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						bool enable, int xcc_id)
+{
+	uint32_t data, def;
+
+	/* It is disabled by HW by default */
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGCG)) {
+		/* 1 - RLC_CGTT_MGCG_OVERRIDE */
+		def = data = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_CGTT_MGCG_OVERRIDE);
+
+		data &= ~(RLC_CGTT_MGCG_OVERRIDE__GRBM_CGTT_SCLK_OVERRIDE_MASK |
+			  RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGCG_OVERRIDE_MASK |
+			  RLC_CGTT_MGCG_OVERRIDE__RLC_CGTT_SCLK_OVERRIDE_MASK |
+			  RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGLS_OVERRIDE_MASK);
+
+		if (def != data)
+			WREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_CGTT_MGCG_OVERRIDE, data);
+
+		/* MGLS is a global flag to control all MGLS in GFX */
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGLS) {
+			/* 2 - RLC memory Light sleep */
+			if (adev->cg_flags & AMD_CG_SUPPORT_GFX_RLC_LS) {
+				def = data = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_MEM_SLP_CNTL);
+				data |= RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK;
+				if (def != data)
+					WREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_MEM_SLP_CNTL, data);
+			}
+			/* 3 - CP memory Light sleep */
+			if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CP_LS) {
+				def = data = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_MEM_SLP_CNTL);
+				data |= CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK;
+				if (def != data)
+					WREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_MEM_SLP_CNTL, data);
+			}
+		}
+	} else {
+		/* 1 - MGCG_OVERRIDE */
+		def = data = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_CGTT_MGCG_OVERRIDE);
+
+		data |= (RLC_CGTT_MGCG_OVERRIDE__RLC_CGTT_SCLK_OVERRIDE_MASK |
+			 RLC_CGTT_MGCG_OVERRIDE__GRBM_CGTT_SCLK_OVERRIDE_MASK |
+			 RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGCG_OVERRIDE_MASK |
+			 RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGLS_OVERRIDE_MASK);
+
+		if (def != data)
+			WREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_CGTT_MGCG_OVERRIDE, data);
+
+		/* 2 - disable MGLS in RLC */
+		data = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_MEM_SLP_CNTL);
+		if (data & RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK) {
+			data &= ~RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK;
+			WREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_MEM_SLP_CNTL, data);
+		}
+
+		/* 3 - disable MGLS in CP */
+		data = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_MEM_SLP_CNTL);
+		if (data & CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK) {
+			data &= ~CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK;
+			WREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_MEM_SLP_CNTL, data);
+		}
+	}
+
+}
+
+static void
+gfx_v9_4_3_xcc_update_coarse_grain_clock_gating(struct amdgpu_device *adev,
+						bool enable, int xcc_id)
+{
+	uint32_t def, data;
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGCG)) {
+
+		def = data = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_CGTT_MGCG_OVERRIDE);
+		/* unset CGCG override */
+		data &= ~RLC_CGTT_MGCG_OVERRIDE__GFXIP_CGCG_OVERRIDE_MASK;
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGLS)
+			data &= ~RLC_CGTT_MGCG_OVERRIDE__GFXIP_CGLS_OVERRIDE_MASK;
+		else
+			data |= RLC_CGTT_MGCG_OVERRIDE__GFXIP_CGLS_OVERRIDE_MASK;
+		/* update CGCG and CGLS override bits */
+		if (def != data)
+			WREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_CGTT_MGCG_OVERRIDE, data);
+
+		/* CGCG Hysteresis: 400us */
+		def = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_CGCG_CGLS_CTRL);
+
+		data = (0x2710
+			<< RLC_CGCG_CGLS_CTRL__CGCG_GFX_IDLE_THRESHOLD__SHIFT) |
+		       RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK;
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGLS)
+			data |= (0x000F << RLC_CGCG_CGLS_CTRL__CGLS_REP_COMPANSAT_DELAY__SHIFT) |
+				RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK;
+		if (def != data)
+			WREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_CGCG_CGLS_CTRL, data);
+
+		/* set IDLE_POLL_COUNT(0x33450100)*/
+		def = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_RB_WPTR_POLL_CNTL);
+		data = (0x0100 << CP_RB_WPTR_POLL_CNTL__POLL_FREQUENCY__SHIFT) |
+			(0x3345 << CP_RB_WPTR_POLL_CNTL__IDLE_POLL_COUNT__SHIFT);
+		if (def != data)
+			WREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_RB_WPTR_POLL_CNTL, data);
+	} else {
+		def = data = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_CGCG_CGLS_CTRL);
+		/* reset CGCG/CGLS bits */
+		data &= ~(RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK | RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK);
+		/* disable cgcg and cgls in FSM */
+		if (def != data)
+			WREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_CGCG_CGLS_CTRL, data);
+	}
+
+}
+
+static int gfx_v9_4_3_xcc_update_gfx_clock_gating(struct amdgpu_device *adev,
+						  bool enable, int xcc_id)
+{
+	amdgpu_gfx_rlc_enter_safe_mode(adev, xcc_id);
+
+	if (enable) {
+		/* FGCG */
+		gfx_v9_4_3_xcc_update_sram_fgcg(adev, enable, xcc_id);
+		gfx_v9_4_3_xcc_update_repeater_fgcg(adev, enable, xcc_id);
+
+		/* CGCG/CGLS should be enabled after MGCG/MGLS
+		 * ===  MGCG + MGLS ===
+		 */
+		gfx_v9_4_3_xcc_update_medium_grain_clock_gating(adev, enable,
+								xcc_id);
+		/* ===  CGCG + CGLS === */
+		gfx_v9_4_3_xcc_update_coarse_grain_clock_gating(adev, enable,
+								xcc_id);
+	} else {
+		/* CGCG/CGLS should be disabled before MGCG/MGLS
+		 * ===  CGCG + CGLS ===
+		 */
+		gfx_v9_4_3_xcc_update_coarse_grain_clock_gating(adev, enable,
+								xcc_id);
+		/* ===  MGCG + MGLS === */
+		gfx_v9_4_3_xcc_update_medium_grain_clock_gating(adev, enable,
+								xcc_id);
+
+		/* FGCG */
+		gfx_v9_4_3_xcc_update_sram_fgcg(adev, enable, xcc_id);
+		gfx_v9_4_3_xcc_update_repeater_fgcg(adev, enable, xcc_id);
+	}
+
+	amdgpu_gfx_rlc_exit_safe_mode(adev, xcc_id);
+
+	return 0;
+}
+
+static const struct amdgpu_rlc_funcs gfx_v9_4_3_rlc_funcs = {
+	.is_rlc_enabled = gfx_v9_4_3_is_rlc_enabled,
+	.set_safe_mode = gfx_v9_4_3_xcc_set_safe_mode,
+	.unset_safe_mode = gfx_v9_4_3_xcc_unset_safe_mode,
+	.init = gfx_v9_4_3_rlc_init,
+	.resume = gfx_v9_4_3_rlc_resume,
+	.stop = gfx_v9_4_3_rlc_stop,
+	.reset = gfx_v9_4_3_rlc_reset,
+	.start = gfx_v9_4_3_rlc_start,
+	.update_spm_vmid = gfx_v9_4_3_update_spm_vmid,
+	.is_rlcg_access_range = gfx_v9_4_3_is_rlcg_access_range,
+};
+
+static int gfx_v9_4_3_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static int gfx_v9_4_3_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, num_xcc;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	for (i = 0; i < num_xcc; i++)
+		gfx_v9_4_3_xcc_update_gfx_clock_gating(
+			adev, state == AMD_CG_STATE_GATE, i);
+
+	return 0;
+}
+
+static void gfx_v9_4_3_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int data;
+
+	if (amdgpu_sriov_vf(adev))
+		*flags = 0;
+
+	/* AMD_CG_SUPPORT_GFX_MGCG */
+	data = RREG32_KIQ(SOC15_REG_OFFSET(GC, GET_INST(GC, 0), regRLC_CGTT_MGCG_OVERRIDE));
+	if (!(data & RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGCG_OVERRIDE_MASK))
+		*flags |= AMD_CG_SUPPORT_GFX_MGCG;
+
+	/* AMD_CG_SUPPORT_GFX_CGCG */
+	data = RREG32_KIQ(SOC15_REG_OFFSET(GC, GET_INST(GC, 0), regRLC_CGCG_CGLS_CTRL));
+	if (data & RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_CGCG;
+
+	/* AMD_CG_SUPPORT_GFX_CGLS */
+	if (data & RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_CGLS;
+
+	/* AMD_CG_SUPPORT_GFX_RLC_LS */
+	data = RREG32_KIQ(SOC15_REG_OFFSET(GC, GET_INST(GC, 0), regRLC_MEM_SLP_CNTL));
+	if (data & RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_RLC_LS | AMD_CG_SUPPORT_GFX_MGLS;
+
+	/* AMD_CG_SUPPORT_GFX_CP_LS */
+	data = RREG32_KIQ(SOC15_REG_OFFSET(GC, GET_INST(GC, 0), regCP_MEM_SLP_CNTL));
+	if (data & CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_CP_LS | AMD_CG_SUPPORT_GFX_MGLS;
+}
+
+static void gfx_v9_4_3_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 ref_and_mask, reg_mem_engine;
+	const struct nbio_hdp_flush_reg *nbio_hf_reg = adev->nbio.hdp_flush_reg;
+
+	if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE) {
+		switch (ring->me) {
+		case 1:
+			ref_and_mask = nbio_hf_reg->ref_and_mask_cp2 << ring->pipe;
+			break;
+		case 2:
+			ref_and_mask = nbio_hf_reg->ref_and_mask_cp6 << ring->pipe;
+			break;
+		default:
+			return;
+		}
+		reg_mem_engine = 0;
+	} else {
+		ref_and_mask = nbio_hf_reg->ref_and_mask_cp0;
+		reg_mem_engine = 1; /* pfp */
+	}
+
+	gfx_v9_4_3_wait_reg_mem(ring, reg_mem_engine, 0, 1,
+			      adev->nbio.funcs->get_hdp_flush_req_offset(adev),
+			      adev->nbio.funcs->get_hdp_flush_done_offset(adev),
+			      ref_and_mask, ref_and_mask, 0x20);
+}
+
+static void gfx_v9_4_3_ring_emit_ib_compute(struct amdgpu_ring *ring,
+					  struct amdgpu_job *job,
+					  struct amdgpu_ib *ib,
+					  uint32_t flags)
+{
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+	u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vmid << 24);
+
+	/* Currently, there is a high possibility to get wave ID mismatch
+	 * between ME and GDS, leading to a hw deadlock, because ME generates
+	 * different wave IDs than the GDS expects. This situation happens
+	 * randomly when at least 5 compute pipes use GDS ordered append.
+	 * The wave IDs generated by ME are also wrong after suspend/resume.
+	 * Those are probably bugs somewhere else in the kernel driver.
+	 *
+	 * Writing GDS_COMPUTE_MAX_WAVE_ID resets wave ID counters in ME and
+	 * GDS to 0 for this ring (me/pipe).
+	 */
+	if (ib->flags & AMDGPU_IB_FLAG_RESET_GDS_MAX_WAVE_ID) {
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1));
+		amdgpu_ring_write(ring, regGDS_COMPUTE_MAX_WAVE_ID);
+		amdgpu_ring_write(ring, ring->adev->gds.gds_compute_max_wave_id);
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2));
+	BUG_ON(ib->gpu_addr & 0x3); /* Dword align */
+	amdgpu_ring_write(ring,
+#ifdef __BIG_ENDIAN
+				(2 << 0) |
+#endif
+				lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, control);
+}
+
+static void gfx_v9_4_3_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
+				     u64 seq, unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	bool int_sel = flags & AMDGPU_FENCE_FLAG_INT;
+	bool writeback = flags & AMDGPU_FENCE_FLAG_TC_WB_ONLY;
+
+	/* RELEASE_MEM - flush caches, send int */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_RELEASE_MEM, 6));
+	amdgpu_ring_write(ring, ((writeback ? (EOP_TC_WB_ACTION_EN |
+					       EOP_TC_NC_ACTION_EN) :
+					      (EOP_TCL1_ACTION_EN |
+					       EOP_TC_ACTION_EN |
+					       EOP_TC_WB_ACTION_EN |
+					       EOP_TC_MD_ACTION_EN)) |
+				 EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) |
+				 EVENT_INDEX(5)));
+	amdgpu_ring_write(ring, DATA_SEL(write64bit ? 2 : 1) | INT_SEL(int_sel ? 2 : 0));
+
+	/*
+	 * the address should be Qword aligned if 64bit write, Dword
+	 * aligned if only send 32bit data low (discard data high)
+	 */
+	if (write64bit)
+		BUG_ON(addr & 0x7);
+	else
+		BUG_ON(addr & 0x3);
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+	amdgpu_ring_write(ring, upper_32_bits(seq));
+	amdgpu_ring_write(ring, 0);
+}
+
+static void gfx_v9_4_3_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	gfx_v9_4_3_wait_reg_mem(ring, usepfp, 1, 0,
+			      lower_32_bits(addr), upper_32_bits(addr),
+			      seq, 0xffffffff, 4);
+}
+
+static void gfx_v9_4_3_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					unsigned vmid, uint64_t pd_addr)
+{
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+}
+
+static u64 gfx_v9_4_3_ring_get_rptr_compute(struct amdgpu_ring *ring)
+{
+	return ring->adev->wb.wb[ring->rptr_offs]; /* gfx9 hardware is 32bit rptr */
+}
+
+static u64 gfx_v9_4_3_ring_get_wptr_compute(struct amdgpu_ring *ring)
+{
+	u64 wptr;
+
+	/* XXX check if swapping is necessary on BE */
+	if (ring->use_doorbell)
+		wptr = atomic64_read((atomic64_t *)&ring->adev->wb.wb[ring->wptr_offs]);
+	else
+		BUG();
+	return wptr;
+}
+
+static void gfx_v9_4_3_ring_set_wptr_compute(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	/* XXX check if swapping is necessary on BE */
+	if (ring->use_doorbell) {
+		atomic64_set((atomic64_t *)&adev->wb.wb[ring->wptr_offs], ring->wptr);
+		WDOORBELL64(ring->doorbell_index, ring->wptr);
+	} else {
+		BUG(); /* only DOORBELL method supported on gfx9 now */
+	}
+}
+
+static void gfx_v9_4_3_ring_emit_fence_kiq(struct amdgpu_ring *ring, u64 addr,
+					 u64 seq, unsigned int flags)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	/* we only allocate 32bit for each seq wb address */
+	BUG_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	/* write fence seq to the "addr" */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				 WRITE_DATA_DST_SEL(5) | WR_CONFIRM));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+
+	if (flags & AMDGPU_FENCE_FLAG_INT) {
+		/* set register to trigger INT */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+		amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+					 WRITE_DATA_DST_SEL(0) | WR_CONFIRM));
+		amdgpu_ring_write(ring, SOC15_REG_OFFSET(GC, GET_INST(GC, 0), regCPC_INT_STATUS));
+		amdgpu_ring_write(ring, 0);
+		amdgpu_ring_write(ring, 0x20000000); /* src_id is 178 */
+	}
+}
+
+static void gfx_v9_4_3_ring_emit_rreg(struct amdgpu_ring *ring, uint32_t reg,
+				    uint32_t reg_val_offs)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	reg = gfx_v9_4_3_normalize_xcc_reg_offset(reg);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_COPY_DATA, 4));
+	amdgpu_ring_write(ring, 0 |	/* src: register*/
+				(5 << 8) |	/* dst: memory */
+				(1 << 20));	/* write confirm */
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, lower_32_bits(adev->wb.gpu_addr +
+				reg_val_offs * 4));
+	amdgpu_ring_write(ring, upper_32_bits(adev->wb.gpu_addr +
+				reg_val_offs * 4));
+}
+
+static void gfx_v9_4_3_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg,
+				    uint32_t val)
+{
+	uint32_t cmd = 0;
+
+	reg = gfx_v9_4_3_normalize_xcc_reg_offset(reg);
+
+	switch (ring->funcs->type) {
+	case AMDGPU_RING_TYPE_GFX:
+		cmd = WRITE_DATA_ENGINE_SEL(1) | WR_CONFIRM;
+		break;
+	case AMDGPU_RING_TYPE_KIQ:
+		cmd = (1 << 16); /* no inc addr */
+		break;
+	default:
+		cmd = WR_CONFIRM;
+		break;
+	}
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, cmd);
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, val);
+}
+
+static void gfx_v9_4_3_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+					uint32_t val, uint32_t mask)
+{
+	gfx_v9_4_3_wait_reg_mem(ring, 0, 0, 0, reg, 0, val, mask, 0x20);
+}
+
+static void gfx_v9_4_3_ring_emit_reg_write_reg_wait(struct amdgpu_ring *ring,
+						  uint32_t reg0, uint32_t reg1,
+						  uint32_t ref, uint32_t mask)
+{
+	amdgpu_ring_emit_reg_write_reg_wait_helper(ring, reg0, reg1,
+						   ref, mask);
+}
+
+static void gfx_v9_4_3_ring_soft_recovery(struct amdgpu_ring *ring,
+					  unsigned vmid)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t value = 0;
+
+	value = REG_SET_FIELD(value, SQ_CMD, CMD, 0x03);
+	value = REG_SET_FIELD(value, SQ_CMD, MODE, 0x01);
+	value = REG_SET_FIELD(value, SQ_CMD, CHECK_VMID, 1);
+	value = REG_SET_FIELD(value, SQ_CMD, VM_ID, vmid);
+	amdgpu_gfx_rlc_enter_safe_mode(adev, ring->xcc_id);
+	WREG32_SOC15(GC, GET_INST(GC, ring->xcc_id), regSQ_CMD, value);
+	amdgpu_gfx_rlc_exit_safe_mode(adev, ring->xcc_id);
+}
+
+static void gfx_v9_4_3_xcc_set_compute_eop_interrupt_state(
+	struct amdgpu_device *adev, int me, int pipe,
+	enum amdgpu_interrupt_state state, int xcc_id)
+{
+	u32 mec_int_cntl, mec_int_cntl_reg;
+
+	/*
+	 * amdgpu controls only the first MEC. That's why this function only
+	 * handles the setting of interrupts for this specific MEC. All other
+	 * pipes' interrupts are set by amdkfd.
+	 */
+
+	if (me == 1) {
+		switch (pipe) {
+		case 0:
+			mec_int_cntl_reg = SOC15_REG_OFFSET(GC, GET_INST(GC, xcc_id), regCP_ME1_PIPE0_INT_CNTL);
+			break;
+		case 1:
+			mec_int_cntl_reg = SOC15_REG_OFFSET(GC, GET_INST(GC, xcc_id), regCP_ME1_PIPE1_INT_CNTL);
+			break;
+		case 2:
+			mec_int_cntl_reg = SOC15_REG_OFFSET(GC, GET_INST(GC, xcc_id), regCP_ME1_PIPE2_INT_CNTL);
+			break;
+		case 3:
+			mec_int_cntl_reg = SOC15_REG_OFFSET(GC, GET_INST(GC, xcc_id), regCP_ME1_PIPE3_INT_CNTL);
+			break;
+		default:
+			DRM_DEBUG("invalid pipe %d\n", pipe);
+			return;
+		}
+	} else {
+		DRM_DEBUG("invalid me %d\n", me);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		mec_int_cntl = RREG32_XCC(mec_int_cntl_reg, xcc_id);
+		mec_int_cntl = REG_SET_FIELD(mec_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+					     TIME_STAMP_INT_ENABLE, 0);
+		WREG32_XCC(mec_int_cntl_reg, mec_int_cntl, xcc_id);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		mec_int_cntl = RREG32_XCC(mec_int_cntl_reg, xcc_id);
+		mec_int_cntl = REG_SET_FIELD(mec_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+					     TIME_STAMP_INT_ENABLE, 1);
+		WREG32_XCC(mec_int_cntl_reg, mec_int_cntl, xcc_id);
+		break;
+	default:
+		break;
+	}
+}
+
+static u32 gfx_v9_4_3_get_cpc_int_cntl(struct amdgpu_device *adev,
+				     int xcc_id, int me, int pipe)
+{
+	/*
+	 * amdgpu controls only the first MEC. That's why this function only
+	 * handles the setting of interrupts for this specific MEC. All other
+	 * pipes' interrupts are set by amdkfd.
+	 */
+	if (me != 1)
+		return 0;
+
+	switch (pipe) {
+	case 0:
+		return SOC15_REG_OFFSET(GC, GET_INST(GC, xcc_id), regCP_ME1_PIPE0_INT_CNTL);
+	case 1:
+		return SOC15_REG_OFFSET(GC, GET_INST(GC, xcc_id), regCP_ME1_PIPE1_INT_CNTL);
+	case 2:
+		return SOC15_REG_OFFSET(GC, GET_INST(GC, xcc_id), regCP_ME1_PIPE2_INT_CNTL);
+	case 3:
+		return SOC15_REG_OFFSET(GC, GET_INST(GC, xcc_id), regCP_ME1_PIPE3_INT_CNTL);
+	default:
+		return 0;
+	}
+}
+
+static int gfx_v9_4_3_set_priv_reg_fault_state(struct amdgpu_device *adev,
+					     struct amdgpu_irq_src *source,
+					     unsigned type,
+					     enum amdgpu_interrupt_state state)
+{
+	u32 mec_int_cntl_reg, mec_int_cntl;
+	int i, j, k, num_xcc;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+	case AMDGPU_IRQ_STATE_ENABLE:
+		for (i = 0; i < num_xcc; i++) {
+			WREG32_FIELD15_PREREG(GC, GET_INST(GC, i), CP_INT_CNTL_RING0,
+					      PRIV_REG_INT_ENABLE,
+					      state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+			for (j = 0; j < adev->gfx.mec.num_mec; j++) {
+				for (k = 0; k < adev->gfx.mec.num_pipe_per_mec; k++) {
+					/* MECs start at 1 */
+					mec_int_cntl_reg = gfx_v9_4_3_get_cpc_int_cntl(adev, i, j + 1, k);
+
+					if (mec_int_cntl_reg) {
+						mec_int_cntl = RREG32_XCC(mec_int_cntl_reg, i);
+						mec_int_cntl = REG_SET_FIELD(mec_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+									     PRIV_REG_INT_ENABLE,
+									     state == AMDGPU_IRQ_STATE_ENABLE ?
+									     1 : 0);
+						WREG32_XCC(mec_int_cntl_reg, mec_int_cntl, i);
+					}
+				}
+			}
+		}
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gfx_v9_4_3_set_bad_op_fault_state(struct amdgpu_device *adev,
+					     struct amdgpu_irq_src *source,
+					     unsigned type,
+					     enum amdgpu_interrupt_state state)
+{
+	u32 mec_int_cntl_reg, mec_int_cntl;
+	int i, j, k, num_xcc;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+	case AMDGPU_IRQ_STATE_ENABLE:
+		for (i = 0; i < num_xcc; i++) {
+			WREG32_FIELD15_PREREG(GC, GET_INST(GC, i), CP_INT_CNTL_RING0,
+					      OPCODE_ERROR_INT_ENABLE,
+					      state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+			for (j = 0; j < adev->gfx.mec.num_mec; j++) {
+				for (k = 0; k < adev->gfx.mec.num_pipe_per_mec; k++) {
+					/* MECs start at 1 */
+					mec_int_cntl_reg = gfx_v9_4_3_get_cpc_int_cntl(adev, i, j + 1, k);
+
+					if (mec_int_cntl_reg) {
+						mec_int_cntl = RREG32_XCC(mec_int_cntl_reg, i);
+						mec_int_cntl = REG_SET_FIELD(mec_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+									     OPCODE_ERROR_INT_ENABLE,
+									     state == AMDGPU_IRQ_STATE_ENABLE ?
+									     1 : 0);
+						WREG32_XCC(mec_int_cntl_reg, mec_int_cntl, i);
+					}
+				}
+			}
+		}
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gfx_v9_4_3_set_priv_inst_fault_state(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      unsigned type,
+					      enum amdgpu_interrupt_state state)
+{
+	int i, num_xcc;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+	case AMDGPU_IRQ_STATE_ENABLE:
+		for (i = 0; i < num_xcc; i++)
+			WREG32_FIELD15_PREREG(GC, GET_INST(GC, i), CP_INT_CNTL_RING0,
+				PRIV_INSTR_INT_ENABLE,
+				state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gfx_v9_4_3_set_eop_interrupt_state(struct amdgpu_device *adev,
+					    struct amdgpu_irq_src *src,
+					    unsigned type,
+					    enum amdgpu_interrupt_state state)
+{
+	int i, num_xcc;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	for (i = 0; i < num_xcc; i++) {
+		switch (type) {
+		case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP:
+			gfx_v9_4_3_xcc_set_compute_eop_interrupt_state(
+				adev, 1, 0, state, i);
+			break;
+		case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE1_EOP:
+			gfx_v9_4_3_xcc_set_compute_eop_interrupt_state(
+				adev, 1, 1, state, i);
+			break;
+		case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE2_EOP:
+			gfx_v9_4_3_xcc_set_compute_eop_interrupt_state(
+				adev, 1, 2, state, i);
+			break;
+		case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE3_EOP:
+			gfx_v9_4_3_xcc_set_compute_eop_interrupt_state(
+				adev, 1, 3, state, i);
+			break;
+		case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE0_EOP:
+			gfx_v9_4_3_xcc_set_compute_eop_interrupt_state(
+				adev, 2, 0, state, i);
+			break;
+		case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE1_EOP:
+			gfx_v9_4_3_xcc_set_compute_eop_interrupt_state(
+				adev, 2, 1, state, i);
+			break;
+		case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE2_EOP:
+			gfx_v9_4_3_xcc_set_compute_eop_interrupt_state(
+				adev, 2, 2, state, i);
+			break;
+		case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE3_EOP:
+			gfx_v9_4_3_xcc_set_compute_eop_interrupt_state(
+				adev, 2, 3, state, i);
+			break;
+		default:
+			break;
+		}
+	}
+
+	return 0;
+}
+
+static int gfx_v9_4_3_eop_irq(struct amdgpu_device *adev,
+			    struct amdgpu_irq_src *source,
+			    struct amdgpu_iv_entry *entry)
+{
+	int i, xcc_id;
+	u8 me_id, pipe_id, queue_id;
+	struct amdgpu_ring *ring;
+
+	DRM_DEBUG("IH: CP EOP\n");
+	me_id = (entry->ring_id & 0x0c) >> 2;
+	pipe_id = (entry->ring_id & 0x03) >> 0;
+	queue_id = (entry->ring_id & 0x70) >> 4;
+
+	xcc_id = gfx_v9_4_3_ih_to_xcc_inst(adev, entry->node_id);
+
+	if (xcc_id == -EINVAL)
+		return -EINVAL;
+
+	switch (me_id) {
+	case 0:
+	case 1:
+	case 2:
+		for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+			ring = &adev->gfx.compute_ring
+					[i +
+					 xcc_id * adev->gfx.num_compute_rings];
+			/* Per-queue interrupt is supported for MEC starting from VI.
+			  * The interrupt can only be enabled/disabled per pipe instead of per queue.
+			  */
+
+			if ((ring->me == me_id) && (ring->pipe == pipe_id) && (ring->queue == queue_id))
+				amdgpu_fence_process(ring);
+		}
+		break;
+	}
+	return 0;
+}
+
+static void gfx_v9_4_3_fault(struct amdgpu_device *adev,
+			   struct amdgpu_iv_entry *entry)
+{
+	u8 me_id, pipe_id, queue_id;
+	struct amdgpu_ring *ring;
+	int i, xcc_id;
+
+	me_id = (entry->ring_id & 0x0c) >> 2;
+	pipe_id = (entry->ring_id & 0x03) >> 0;
+	queue_id = (entry->ring_id & 0x70) >> 4;
+
+	xcc_id = gfx_v9_4_3_ih_to_xcc_inst(adev, entry->node_id);
+
+	if (xcc_id == -EINVAL)
+		return;
+
+	switch (me_id) {
+	case 0:
+	case 1:
+	case 2:
+		for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+			ring = &adev->gfx.compute_ring
+					[i +
+					 xcc_id * adev->gfx.num_compute_rings];
+			if (ring->me == me_id && ring->pipe == pipe_id &&
+			    ring->queue == queue_id)
+				drm_sched_fault(&ring->sched);
+		}
+		break;
+	}
+}
+
+static int gfx_v9_4_3_priv_reg_irq(struct amdgpu_device *adev,
+				 struct amdgpu_irq_src *source,
+				 struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal register access in command stream\n");
+	gfx_v9_4_3_fault(adev, entry);
+	return 0;
+}
+
+static int gfx_v9_4_3_bad_op_irq(struct amdgpu_device *adev,
+				 struct amdgpu_irq_src *source,
+				 struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal opcode in command stream\n");
+	gfx_v9_4_3_fault(adev, entry);
+	return 0;
+}
+
+static int gfx_v9_4_3_priv_inst_irq(struct amdgpu_device *adev,
+				  struct amdgpu_irq_src *source,
+				  struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal instruction in command stream\n");
+	gfx_v9_4_3_fault(adev, entry);
+	return 0;
+}
+
+static void gfx_v9_4_3_emit_mem_sync(struct amdgpu_ring *ring)
+{
+	const unsigned int cp_coher_cntl =
+			PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_SH_ICACHE_ACTION_ENA(1) |
+			PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_SH_KCACHE_ACTION_ENA(1) |
+			PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_TC_ACTION_ENA(1) |
+			PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_TCL1_ACTION_ENA(1) |
+			PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_TC_WB_ACTION_ENA(1);
+
+	/* ACQUIRE_MEM -make one or more surfaces valid for use by the subsequent operations */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_ACQUIRE_MEM, 5));
+	amdgpu_ring_write(ring, cp_coher_cntl); /* CP_COHER_CNTL */
+	amdgpu_ring_write(ring, 0xffffffff);  /* CP_COHER_SIZE */
+	amdgpu_ring_write(ring, 0xffffff);  /* CP_COHER_SIZE_HI */
+	amdgpu_ring_write(ring, 0); /* CP_COHER_BASE */
+	amdgpu_ring_write(ring, 0);  /* CP_COHER_BASE_HI */
+	amdgpu_ring_write(ring, 0x0000000A); /* POLL_INTERVAL */
+}
+
+static void gfx_v9_4_3_emit_wave_limit_cs(struct amdgpu_ring *ring,
+					uint32_t pipe, bool enable)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t val;
+	uint32_t wcl_cs_reg;
+
+	/* regSPI_WCL_PIPE_PERCENT_CS[0-7]_DEFAULT values are same */
+	val = enable ? 0x1 : 0x7f;
+
+	switch (pipe) {
+	case 0:
+		wcl_cs_reg = SOC15_REG_OFFSET(GC, GET_INST(GC, 0), regSPI_WCL_PIPE_PERCENT_CS0);
+		break;
+	case 1:
+		wcl_cs_reg = SOC15_REG_OFFSET(GC, GET_INST(GC, 0), regSPI_WCL_PIPE_PERCENT_CS1);
+		break;
+	case 2:
+		wcl_cs_reg = SOC15_REG_OFFSET(GC, GET_INST(GC, 0), regSPI_WCL_PIPE_PERCENT_CS2);
+		break;
+	case 3:
+		wcl_cs_reg = SOC15_REG_OFFSET(GC, GET_INST(GC, 0), regSPI_WCL_PIPE_PERCENT_CS3);
+		break;
+	default:
+		DRM_DEBUG("invalid pipe %d\n", pipe);
+		return;
+	}
+
+	amdgpu_ring_emit_wreg(ring, wcl_cs_reg, val);
+
+}
+static void gfx_v9_4_3_emit_wave_limit(struct amdgpu_ring *ring, bool enable)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t val;
+	int i;
+
+	/* regSPI_WCL_PIPE_PERCENT_GFX is 7 bit multiplier register to limit
+	 * number of gfx waves. Setting 5 bit will make sure gfx only gets
+	 * around 25% of gpu resources.
+	 */
+	val = enable ? 0x1f : 0x07ffffff;
+	amdgpu_ring_emit_wreg(ring,
+			      SOC15_REG_OFFSET(GC, GET_INST(GC, 0), regSPI_WCL_PIPE_PERCENT_GFX),
+			      val);
+
+	/* Restrict waves for normal/low priority compute queues as well
+	 * to get best QoS for high priority compute jobs.
+	 *
+	 * amdgpu controls only 1st ME(0-3 CS pipes).
+	 */
+	for (i = 0; i < adev->gfx.mec.num_pipe_per_mec; i++) {
+		if (i != ring->pipe)
+			gfx_v9_4_3_emit_wave_limit_cs(ring, i, enable);
+
+	}
+}
+
+static int gfx_v9_4_3_unmap_done(struct amdgpu_device *adev, uint32_t me,
+				uint32_t pipe, uint32_t queue,
+				uint32_t xcc_id)
+{
+	int i, r;
+	/* make sure dequeue is complete*/
+	gfx_v9_4_3_xcc_set_safe_mode(adev, xcc_id);
+	mutex_lock(&adev->srbm_mutex);
+	soc15_grbm_select(adev, me, pipe, queue, 0, GET_INST(GC, xcc_id));
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (!(RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_HQD_ACTIVE) & 1))
+			break;
+		udelay(1);
+	}
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+	else
+		r = 0;
+	soc15_grbm_select(adev, 0, 0, 0, 0, GET_INST(GC, xcc_id));
+	mutex_unlock(&adev->srbm_mutex);
+	gfx_v9_4_3_xcc_unset_safe_mode(adev, xcc_id);
+
+	return r;
+
+}
+
+static bool gfx_v9_4_3_pipe_reset_support(struct amdgpu_device *adev)
+{
+	if (!!(adev->gfx.compute_supported_reset & AMDGPU_RESET_TYPE_PER_PIPE))
+		return true;
+	else
+		dev_warn_once(adev->dev, "Please use the latest MEC version to see whether support pipe reset\n");
+
+	return false;
+}
+
+static int gfx_v9_4_3_reset_hw_pipe(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t reset_pipe, clean_pipe;
+	int r;
+
+	if (!gfx_v9_4_3_pipe_reset_support(adev))
+		return -EINVAL;
+
+	gfx_v9_4_3_xcc_set_safe_mode(adev, ring->xcc_id);
+	mutex_lock(&adev->srbm_mutex);
+
+	reset_pipe = RREG32_SOC15(GC, GET_INST(GC, ring->xcc_id), regCP_MEC_CNTL);
+	clean_pipe = reset_pipe;
+
+	if (ring->me == 1) {
+		switch (ring->pipe) {
+		case 0:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+						   MEC_ME1_PIPE0_RESET, 1);
+			break;
+		case 1:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+						   MEC_ME1_PIPE1_RESET, 1);
+			break;
+		case 2:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+						   MEC_ME1_PIPE2_RESET, 1);
+			break;
+		case 3:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+						   MEC_ME1_PIPE3_RESET, 1);
+			break;
+		default:
+			break;
+		}
+	} else {
+		if (ring->pipe)
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+						   MEC_ME2_PIPE1_RESET, 1);
+		else
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+						   MEC_ME2_PIPE0_RESET, 1);
+	}
+
+	WREG32_SOC15(GC, GET_INST(GC, ring->xcc_id), regCP_MEC_CNTL, reset_pipe);
+	WREG32_SOC15(GC, GET_INST(GC, ring->xcc_id), regCP_MEC_CNTL, clean_pipe);
+	mutex_unlock(&adev->srbm_mutex);
+	gfx_v9_4_3_xcc_unset_safe_mode(adev, ring->xcc_id);
+
+	r = gfx_v9_4_3_unmap_done(adev, ring->me, ring->pipe, ring->queue, ring->xcc_id);
+	return r;
+}
+
+static int gfx_v9_4_3_reset_kcq(struct amdgpu_ring *ring,
+				unsigned int vmid,
+				struct amdgpu_fence *timedout_fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[ring->xcc_id];
+	struct amdgpu_ring *kiq_ring = &kiq->ring;
+	int reset_mode = AMDGPU_RESET_TYPE_PER_QUEUE;
+	unsigned long flags;
+	int r;
+
+	if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
+		return -EINVAL;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+
+	spin_lock_irqsave(&kiq->ring_lock, flags);
+
+	if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->unmap_queues_size)) {
+		spin_unlock_irqrestore(&kiq->ring_lock, flags);
+		return -ENOMEM;
+	}
+
+	kiq->pmf->kiq_unmap_queues(kiq_ring, ring, RESET_QUEUES,
+				   0, 0);
+	amdgpu_ring_commit(kiq_ring);
+
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+
+	r = amdgpu_ring_test_ring(kiq_ring);
+	if (r) {
+		dev_err(adev->dev, "kiq ring test failed after ring: %s queue reset\n",
+				ring->name);
+		goto pipe_reset;
+	}
+
+	r = gfx_v9_4_3_unmap_done(adev, ring->me, ring->pipe, ring->queue, ring->xcc_id);
+	if (r)
+		dev_err(adev->dev, "fail to wait on hqd deactive and will try pipe reset\n");
+
+pipe_reset:
+	if (r) {
+		if (!(adev->gfx.compute_supported_reset & AMDGPU_RESET_TYPE_PER_PIPE))
+			return -EOPNOTSUPP;
+		r = gfx_v9_4_3_reset_hw_pipe(ring);
+		reset_mode = AMDGPU_RESET_TYPE_PER_PIPE;
+		dev_info(adev->dev, "ring: %s pipe reset :%s\n", ring->name,
+				r ? "failed" : "successfully");
+		if (r)
+			return r;
+	}
+
+	r = gfx_v9_4_3_xcc_kcq_init_queue(ring, ring->xcc_id, true);
+	if (r) {
+		dev_err(adev->dev, "fail to init kcq\n");
+		return r;
+	}
+	spin_lock_irqsave(&kiq->ring_lock, flags);
+	r = amdgpu_ring_alloc(kiq_ring, kiq->pmf->map_queues_size);
+	if (r) {
+		spin_unlock_irqrestore(&kiq->ring_lock, flags);
+		return -ENOMEM;
+	}
+	kiq->pmf->kiq_map_queues(kiq_ring, ring);
+	amdgpu_ring_commit(kiq_ring);
+	r = amdgpu_ring_test_ring(kiq_ring);
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+	if (r) {
+		if (reset_mode == AMDGPU_RESET_TYPE_PER_QUEUE)
+			goto pipe_reset;
+
+		dev_err(adev->dev, "fail to remap queue\n");
+		return r;
+	}
+
+	if (reset_mode == AMDGPU_RESET_TYPE_PER_QUEUE) {
+		r = amdgpu_ring_test_ring(ring);
+		if (r)
+			goto pipe_reset;
+	}
+
+
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+enum amdgpu_gfx_cp_ras_mem_id {
+	AMDGPU_GFX_CP_MEM1 = 1,
+	AMDGPU_GFX_CP_MEM2,
+	AMDGPU_GFX_CP_MEM3,
+	AMDGPU_GFX_CP_MEM4,
+	AMDGPU_GFX_CP_MEM5,
+};
+
+enum amdgpu_gfx_gcea_ras_mem_id {
+	AMDGPU_GFX_GCEA_IOWR_CMDMEM = 4,
+	AMDGPU_GFX_GCEA_IORD_CMDMEM,
+	AMDGPU_GFX_GCEA_GMIWR_CMDMEM,
+	AMDGPU_GFX_GCEA_GMIRD_CMDMEM,
+	AMDGPU_GFX_GCEA_DRAMWR_CMDMEM,
+	AMDGPU_GFX_GCEA_DRAMRD_CMDMEM,
+	AMDGPU_GFX_GCEA_MAM_DMEM0,
+	AMDGPU_GFX_GCEA_MAM_DMEM1,
+	AMDGPU_GFX_GCEA_MAM_DMEM2,
+	AMDGPU_GFX_GCEA_MAM_DMEM3,
+	AMDGPU_GFX_GCEA_MAM_AMEM0,
+	AMDGPU_GFX_GCEA_MAM_AMEM1,
+	AMDGPU_GFX_GCEA_MAM_AMEM2,
+	AMDGPU_GFX_GCEA_MAM_AMEM3,
+	AMDGPU_GFX_GCEA_MAM_AFLUSH_BUFFER,
+	AMDGPU_GFX_GCEA_WRET_TAGMEM,
+	AMDGPU_GFX_GCEA_RRET_TAGMEM,
+	AMDGPU_GFX_GCEA_IOWR_DATAMEM,
+	AMDGPU_GFX_GCEA_GMIWR_DATAMEM,
+	AMDGPU_GFX_GCEA_DRAM_DATAMEM,
+};
+
+enum amdgpu_gfx_gc_cane_ras_mem_id {
+	AMDGPU_GFX_GC_CANE_MEM0 = 0,
+};
+
+enum amdgpu_gfx_gcutcl2_ras_mem_id {
+	AMDGPU_GFX_GCUTCL2_MEM2P512X95 = 160,
+};
+
+enum amdgpu_gfx_gds_ras_mem_id {
+	AMDGPU_GFX_GDS_MEM0 = 0,
+};
+
+enum amdgpu_gfx_lds_ras_mem_id {
+	AMDGPU_GFX_LDS_BANK0 = 0,
+	AMDGPU_GFX_LDS_BANK1,
+	AMDGPU_GFX_LDS_BANK2,
+	AMDGPU_GFX_LDS_BANK3,
+	AMDGPU_GFX_LDS_BANK4,
+	AMDGPU_GFX_LDS_BANK5,
+	AMDGPU_GFX_LDS_BANK6,
+	AMDGPU_GFX_LDS_BANK7,
+	AMDGPU_GFX_LDS_BANK8,
+	AMDGPU_GFX_LDS_BANK9,
+	AMDGPU_GFX_LDS_BANK10,
+	AMDGPU_GFX_LDS_BANK11,
+	AMDGPU_GFX_LDS_BANK12,
+	AMDGPU_GFX_LDS_BANK13,
+	AMDGPU_GFX_LDS_BANK14,
+	AMDGPU_GFX_LDS_BANK15,
+	AMDGPU_GFX_LDS_BANK16,
+	AMDGPU_GFX_LDS_BANK17,
+	AMDGPU_GFX_LDS_BANK18,
+	AMDGPU_GFX_LDS_BANK19,
+	AMDGPU_GFX_LDS_BANK20,
+	AMDGPU_GFX_LDS_BANK21,
+	AMDGPU_GFX_LDS_BANK22,
+	AMDGPU_GFX_LDS_BANK23,
+	AMDGPU_GFX_LDS_BANK24,
+	AMDGPU_GFX_LDS_BANK25,
+	AMDGPU_GFX_LDS_BANK26,
+	AMDGPU_GFX_LDS_BANK27,
+	AMDGPU_GFX_LDS_BANK28,
+	AMDGPU_GFX_LDS_BANK29,
+	AMDGPU_GFX_LDS_BANK30,
+	AMDGPU_GFX_LDS_BANK31,
+	AMDGPU_GFX_LDS_SP_BUFFER_A,
+	AMDGPU_GFX_LDS_SP_BUFFER_B,
+};
+
+enum amdgpu_gfx_rlc_ras_mem_id {
+	AMDGPU_GFX_RLC_GPMF32 = 1,
+	AMDGPU_GFX_RLC_RLCVF32,
+	AMDGPU_GFX_RLC_SCRATCH,
+	AMDGPU_GFX_RLC_SRM_ARAM,
+	AMDGPU_GFX_RLC_SRM_DRAM,
+	AMDGPU_GFX_RLC_TCTAG,
+	AMDGPU_GFX_RLC_SPM_SE,
+	AMDGPU_GFX_RLC_SPM_GRBMT,
+};
+
+enum amdgpu_gfx_sp_ras_mem_id {
+	AMDGPU_GFX_SP_SIMDID0 = 0,
+};
+
+enum amdgpu_gfx_spi_ras_mem_id {
+	AMDGPU_GFX_SPI_MEM0 = 0,
+	AMDGPU_GFX_SPI_MEM1,
+	AMDGPU_GFX_SPI_MEM2,
+	AMDGPU_GFX_SPI_MEM3,
+};
+
+enum amdgpu_gfx_sqc_ras_mem_id {
+	AMDGPU_GFX_SQC_INST_CACHE_A = 100,
+	AMDGPU_GFX_SQC_INST_CACHE_B = 101,
+	AMDGPU_GFX_SQC_INST_CACHE_TAG_A = 102,
+	AMDGPU_GFX_SQC_INST_CACHE_TAG_B = 103,
+	AMDGPU_GFX_SQC_INST_CACHE_MISS_FIFO_A = 104,
+	AMDGPU_GFX_SQC_INST_CACHE_MISS_FIFO_B = 105,
+	AMDGPU_GFX_SQC_INST_CACHE_GATCL1_MISS_FIFO_A = 106,
+	AMDGPU_GFX_SQC_INST_CACHE_GATCL1_MISS_FIFO_B = 107,
+	AMDGPU_GFX_SQC_DATA_CACHE_A = 200,
+	AMDGPU_GFX_SQC_DATA_CACHE_B = 201,
+	AMDGPU_GFX_SQC_DATA_CACHE_TAG_A = 202,
+	AMDGPU_GFX_SQC_DATA_CACHE_TAG_B = 203,
+	AMDGPU_GFX_SQC_DATA_CACHE_MISS_FIFO_A = 204,
+	AMDGPU_GFX_SQC_DATA_CACHE_MISS_FIFO_B = 205,
+	AMDGPU_GFX_SQC_DATA_CACHE_HIT_FIFO_A = 206,
+	AMDGPU_GFX_SQC_DATA_CACHE_HIT_FIFO_B = 207,
+	AMDGPU_GFX_SQC_DIRTY_BIT_A = 208,
+	AMDGPU_GFX_SQC_DIRTY_BIT_B = 209,
+	AMDGPU_GFX_SQC_WRITE_DATA_BUFFER_CU0 = 210,
+	AMDGPU_GFX_SQC_WRITE_DATA_BUFFER_CU1 = 211,
+	AMDGPU_GFX_SQC_UTCL1_MISS_LFIFO_DATA_CACHE_A = 212,
+	AMDGPU_GFX_SQC_UTCL1_MISS_LFIFO_DATA_CACHE_B = 213,
+	AMDGPU_GFX_SQC_UTCL1_MISS_LFIFO_INST_CACHE = 108,
+};
+
+enum amdgpu_gfx_sq_ras_mem_id {
+	AMDGPU_GFX_SQ_SGPR_MEM0 = 0,
+	AMDGPU_GFX_SQ_SGPR_MEM1,
+	AMDGPU_GFX_SQ_SGPR_MEM2,
+	AMDGPU_GFX_SQ_SGPR_MEM3,
+};
+
+enum amdgpu_gfx_ta_ras_mem_id {
+	AMDGPU_GFX_TA_FS_AFIFO_RAM_LO = 1,
+	AMDGPU_GFX_TA_FS_AFIFO_RAM_HI,
+	AMDGPU_GFX_TA_FS_CFIFO_RAM,
+	AMDGPU_GFX_TA_FSX_LFIFO,
+	AMDGPU_GFX_TA_FS_DFIFO_RAM,
+};
+
+enum amdgpu_gfx_tcc_ras_mem_id {
+	AMDGPU_GFX_TCC_MEM1 = 1,
+};
+
+enum amdgpu_gfx_tca_ras_mem_id {
+	AMDGPU_GFX_TCA_MEM1 = 1,
+};
+
+enum amdgpu_gfx_tci_ras_mem_id {
+	AMDGPU_GFX_TCIW_MEM = 1,
+};
+
+enum amdgpu_gfx_tcp_ras_mem_id {
+	AMDGPU_GFX_TCP_LFIFO0 = 1,
+	AMDGPU_GFX_TCP_SET0BANK0_RAM,
+	AMDGPU_GFX_TCP_SET0BANK1_RAM,
+	AMDGPU_GFX_TCP_SET0BANK2_RAM,
+	AMDGPU_GFX_TCP_SET0BANK3_RAM,
+	AMDGPU_GFX_TCP_SET1BANK0_RAM,
+	AMDGPU_GFX_TCP_SET1BANK1_RAM,
+	AMDGPU_GFX_TCP_SET1BANK2_RAM,
+	AMDGPU_GFX_TCP_SET1BANK3_RAM,
+	AMDGPU_GFX_TCP_SET2BANK0_RAM,
+	AMDGPU_GFX_TCP_SET2BANK1_RAM,
+	AMDGPU_GFX_TCP_SET2BANK2_RAM,
+	AMDGPU_GFX_TCP_SET2BANK3_RAM,
+	AMDGPU_GFX_TCP_SET3BANK0_RAM,
+	AMDGPU_GFX_TCP_SET3BANK1_RAM,
+	AMDGPU_GFX_TCP_SET3BANK2_RAM,
+	AMDGPU_GFX_TCP_SET3BANK3_RAM,
+	AMDGPU_GFX_TCP_VM_FIFO,
+	AMDGPU_GFX_TCP_DB_TAGRAM0,
+	AMDGPU_GFX_TCP_DB_TAGRAM1,
+	AMDGPU_GFX_TCP_DB_TAGRAM2,
+	AMDGPU_GFX_TCP_DB_TAGRAM3,
+	AMDGPU_GFX_TCP_UTCL1_LFIFO_PROBE0,
+	AMDGPU_GFX_TCP_UTCL1_LFIFO_PROBE1,
+	AMDGPU_GFX_TCP_CMD_FIFO,
+};
+
+enum amdgpu_gfx_td_ras_mem_id {
+	AMDGPU_GFX_TD_UTD_CS_FIFO_MEM = 1,
+	AMDGPU_GFX_TD_UTD_SS_FIFO_LO_MEM,
+	AMDGPU_GFX_TD_UTD_SS_FIFO_HI_MEM,
+};
+
+enum amdgpu_gfx_tcx_ras_mem_id {
+	AMDGPU_GFX_TCX_FIFOD0 = 0,
+	AMDGPU_GFX_TCX_FIFOD1,
+	AMDGPU_GFX_TCX_FIFOD2,
+	AMDGPU_GFX_TCX_FIFOD3,
+	AMDGPU_GFX_TCX_FIFOD4,
+	AMDGPU_GFX_TCX_FIFOD5,
+	AMDGPU_GFX_TCX_FIFOD6,
+	AMDGPU_GFX_TCX_FIFOD7,
+	AMDGPU_GFX_TCX_FIFOB0,
+	AMDGPU_GFX_TCX_FIFOB1,
+	AMDGPU_GFX_TCX_FIFOB2,
+	AMDGPU_GFX_TCX_FIFOB3,
+	AMDGPU_GFX_TCX_FIFOB4,
+	AMDGPU_GFX_TCX_FIFOB5,
+	AMDGPU_GFX_TCX_FIFOB6,
+	AMDGPU_GFX_TCX_FIFOB7,
+	AMDGPU_GFX_TCX_FIFOA0,
+	AMDGPU_GFX_TCX_FIFOA1,
+	AMDGPU_GFX_TCX_FIFOA2,
+	AMDGPU_GFX_TCX_FIFOA3,
+	AMDGPU_GFX_TCX_FIFOA4,
+	AMDGPU_GFX_TCX_FIFOA5,
+	AMDGPU_GFX_TCX_FIFOA6,
+	AMDGPU_GFX_TCX_FIFOA7,
+	AMDGPU_GFX_TCX_CFIFO0,
+	AMDGPU_GFX_TCX_CFIFO1,
+	AMDGPU_GFX_TCX_CFIFO2,
+	AMDGPU_GFX_TCX_CFIFO3,
+	AMDGPU_GFX_TCX_CFIFO4,
+	AMDGPU_GFX_TCX_CFIFO5,
+	AMDGPU_GFX_TCX_CFIFO6,
+	AMDGPU_GFX_TCX_CFIFO7,
+	AMDGPU_GFX_TCX_FIFO_ACKB0,
+	AMDGPU_GFX_TCX_FIFO_ACKB1,
+	AMDGPU_GFX_TCX_FIFO_ACKB2,
+	AMDGPU_GFX_TCX_FIFO_ACKB3,
+	AMDGPU_GFX_TCX_FIFO_ACKB4,
+	AMDGPU_GFX_TCX_FIFO_ACKB5,
+	AMDGPU_GFX_TCX_FIFO_ACKB6,
+	AMDGPU_GFX_TCX_FIFO_ACKB7,
+	AMDGPU_GFX_TCX_FIFO_ACKD0,
+	AMDGPU_GFX_TCX_FIFO_ACKD1,
+	AMDGPU_GFX_TCX_FIFO_ACKD2,
+	AMDGPU_GFX_TCX_FIFO_ACKD3,
+	AMDGPU_GFX_TCX_FIFO_ACKD4,
+	AMDGPU_GFX_TCX_FIFO_ACKD5,
+	AMDGPU_GFX_TCX_FIFO_ACKD6,
+	AMDGPU_GFX_TCX_FIFO_ACKD7,
+	AMDGPU_GFX_TCX_DST_FIFOA0,
+	AMDGPU_GFX_TCX_DST_FIFOA1,
+	AMDGPU_GFX_TCX_DST_FIFOA2,
+	AMDGPU_GFX_TCX_DST_FIFOA3,
+	AMDGPU_GFX_TCX_DST_FIFOA4,
+	AMDGPU_GFX_TCX_DST_FIFOA5,
+	AMDGPU_GFX_TCX_DST_FIFOA6,
+	AMDGPU_GFX_TCX_DST_FIFOA7,
+	AMDGPU_GFX_TCX_DST_FIFOB0,
+	AMDGPU_GFX_TCX_DST_FIFOB1,
+	AMDGPU_GFX_TCX_DST_FIFOB2,
+	AMDGPU_GFX_TCX_DST_FIFOB3,
+	AMDGPU_GFX_TCX_DST_FIFOB4,
+	AMDGPU_GFX_TCX_DST_FIFOB5,
+	AMDGPU_GFX_TCX_DST_FIFOB6,
+	AMDGPU_GFX_TCX_DST_FIFOB7,
+	AMDGPU_GFX_TCX_DST_FIFOD0,
+	AMDGPU_GFX_TCX_DST_FIFOD1,
+	AMDGPU_GFX_TCX_DST_FIFOD2,
+	AMDGPU_GFX_TCX_DST_FIFOD3,
+	AMDGPU_GFX_TCX_DST_FIFOD4,
+	AMDGPU_GFX_TCX_DST_FIFOD5,
+	AMDGPU_GFX_TCX_DST_FIFOD6,
+	AMDGPU_GFX_TCX_DST_FIFOD7,
+	AMDGPU_GFX_TCX_DST_FIFO_ACKB0,
+	AMDGPU_GFX_TCX_DST_FIFO_ACKB1,
+	AMDGPU_GFX_TCX_DST_FIFO_ACKB2,
+	AMDGPU_GFX_TCX_DST_FIFO_ACKB3,
+	AMDGPU_GFX_TCX_DST_FIFO_ACKB4,
+	AMDGPU_GFX_TCX_DST_FIFO_ACKB5,
+	AMDGPU_GFX_TCX_DST_FIFO_ACKB6,
+	AMDGPU_GFX_TCX_DST_FIFO_ACKB7,
+	AMDGPU_GFX_TCX_DST_FIFO_ACKD0,
+	AMDGPU_GFX_TCX_DST_FIFO_ACKD1,
+	AMDGPU_GFX_TCX_DST_FIFO_ACKD2,
+	AMDGPU_GFX_TCX_DST_FIFO_ACKD3,
+	AMDGPU_GFX_TCX_DST_FIFO_ACKD4,
+	AMDGPU_GFX_TCX_DST_FIFO_ACKD5,
+	AMDGPU_GFX_TCX_DST_FIFO_ACKD6,
+	AMDGPU_GFX_TCX_DST_FIFO_ACKD7,
+};
+
+enum amdgpu_gfx_atc_l2_ras_mem_id {
+	AMDGPU_GFX_ATC_L2_MEM0 = 0,
+};
+
+enum amdgpu_gfx_utcl2_ras_mem_id {
+	AMDGPU_GFX_UTCL2_MEM0 = 0,
+};
+
+enum amdgpu_gfx_vml2_ras_mem_id {
+	AMDGPU_GFX_VML2_MEM0 = 0,
+};
+
+enum amdgpu_gfx_vml2_walker_ras_mem_id {
+	AMDGPU_GFX_VML2_WALKER_MEM0 = 0,
+};
+
+static const struct amdgpu_ras_memory_id_entry gfx_v9_4_3_ras_cp_mem_list[] = {
+	{AMDGPU_GFX_CP_MEM1, "CP_MEM1"},
+	{AMDGPU_GFX_CP_MEM2, "CP_MEM2"},
+	{AMDGPU_GFX_CP_MEM3, "CP_MEM3"},
+	{AMDGPU_GFX_CP_MEM4, "CP_MEM4"},
+	{AMDGPU_GFX_CP_MEM5, "CP_MEM5"},
+};
+
+static const struct amdgpu_ras_memory_id_entry gfx_v9_4_3_ras_gcea_mem_list[] = {
+	{AMDGPU_GFX_GCEA_IOWR_CMDMEM, "GCEA_IOWR_CMDMEM"},
+	{AMDGPU_GFX_GCEA_IORD_CMDMEM, "GCEA_IORD_CMDMEM"},
+	{AMDGPU_GFX_GCEA_GMIWR_CMDMEM, "GCEA_GMIWR_CMDMEM"},
+	{AMDGPU_GFX_GCEA_GMIRD_CMDMEM, "GCEA_GMIRD_CMDMEM"},
+	{AMDGPU_GFX_GCEA_DRAMWR_CMDMEM, "GCEA_DRAMWR_CMDMEM"},
+	{AMDGPU_GFX_GCEA_DRAMRD_CMDMEM, "GCEA_DRAMRD_CMDMEM"},
+	{AMDGPU_GFX_GCEA_MAM_DMEM0, "GCEA_MAM_DMEM0"},
+	{AMDGPU_GFX_GCEA_MAM_DMEM1, "GCEA_MAM_DMEM1"},
+	{AMDGPU_GFX_GCEA_MAM_DMEM2, "GCEA_MAM_DMEM2"},
+	{AMDGPU_GFX_GCEA_MAM_DMEM3, "GCEA_MAM_DMEM3"},
+	{AMDGPU_GFX_GCEA_MAM_AMEM0, "GCEA_MAM_AMEM0"},
+	{AMDGPU_GFX_GCEA_MAM_AMEM1, "GCEA_MAM_AMEM1"},
+	{AMDGPU_GFX_GCEA_MAM_AMEM2, "GCEA_MAM_AMEM2"},
+	{AMDGPU_GFX_GCEA_MAM_AMEM3, "GCEA_MAM_AMEM3"},
+	{AMDGPU_GFX_GCEA_MAM_AFLUSH_BUFFER, "GCEA_MAM_AFLUSH_BUFFER"},
+	{AMDGPU_GFX_GCEA_WRET_TAGMEM, "GCEA_WRET_TAGMEM"},
+	{AMDGPU_GFX_GCEA_RRET_TAGMEM, "GCEA_RRET_TAGMEM"},
+	{AMDGPU_GFX_GCEA_IOWR_DATAMEM, "GCEA_IOWR_DATAMEM"},
+	{AMDGPU_GFX_GCEA_GMIWR_DATAMEM, "GCEA_GMIWR_DATAMEM"},
+	{AMDGPU_GFX_GCEA_DRAM_DATAMEM, "GCEA_DRAM_DATAMEM"},
+};
+
+static const struct amdgpu_ras_memory_id_entry gfx_v9_4_3_ras_gc_cane_mem_list[] = {
+	{AMDGPU_GFX_GC_CANE_MEM0, "GC_CANE_MEM0"},
+};
+
+static const struct amdgpu_ras_memory_id_entry gfx_v9_4_3_ras_gcutcl2_mem_list[] = {
+	{AMDGPU_GFX_GCUTCL2_MEM2P512X95, "GCUTCL2_MEM2P512X95"},
+};
+
+static const struct amdgpu_ras_memory_id_entry gfx_v9_4_3_ras_gds_mem_list[] = {
+	{AMDGPU_GFX_GDS_MEM0, "GDS_MEM"},
+};
+
+static const struct amdgpu_ras_memory_id_entry gfx_v9_4_3_ras_lds_mem_list[] = {
+	{AMDGPU_GFX_LDS_BANK0, "LDS_BANK0"},
+	{AMDGPU_GFX_LDS_BANK1, "LDS_BANK1"},
+	{AMDGPU_GFX_LDS_BANK2, "LDS_BANK2"},
+	{AMDGPU_GFX_LDS_BANK3, "LDS_BANK3"},
+	{AMDGPU_GFX_LDS_BANK4, "LDS_BANK4"},
+	{AMDGPU_GFX_LDS_BANK5, "LDS_BANK5"},
+	{AMDGPU_GFX_LDS_BANK6, "LDS_BANK6"},
+	{AMDGPU_GFX_LDS_BANK7, "LDS_BANK7"},
+	{AMDGPU_GFX_LDS_BANK8, "LDS_BANK8"},
+	{AMDGPU_GFX_LDS_BANK9, "LDS_BANK9"},
+	{AMDGPU_GFX_LDS_BANK10, "LDS_BANK10"},
+	{AMDGPU_GFX_LDS_BANK11, "LDS_BANK11"},
+	{AMDGPU_GFX_LDS_BANK12, "LDS_BANK12"},
+	{AMDGPU_GFX_LDS_BANK13, "LDS_BANK13"},
+	{AMDGPU_GFX_LDS_BANK14, "LDS_BANK14"},
+	{AMDGPU_GFX_LDS_BANK15, "LDS_BANK15"},
+	{AMDGPU_GFX_LDS_BANK16, "LDS_BANK16"},
+	{AMDGPU_GFX_LDS_BANK17, "LDS_BANK17"},
+	{AMDGPU_GFX_LDS_BANK18, "LDS_BANK18"},
+	{AMDGPU_GFX_LDS_BANK19, "LDS_BANK19"},
+	{AMDGPU_GFX_LDS_BANK20, "LDS_BANK20"},
+	{AMDGPU_GFX_LDS_BANK21, "LDS_BANK21"},
+	{AMDGPU_GFX_LDS_BANK22, "LDS_BANK22"},
+	{AMDGPU_GFX_LDS_BANK23, "LDS_BANK23"},
+	{AMDGPU_GFX_LDS_BANK24, "LDS_BANK24"},
+	{AMDGPU_GFX_LDS_BANK25, "LDS_BANK25"},
+	{AMDGPU_GFX_LDS_BANK26, "LDS_BANK26"},
+	{AMDGPU_GFX_LDS_BANK27, "LDS_BANK27"},
+	{AMDGPU_GFX_LDS_BANK28, "LDS_BANK28"},
+	{AMDGPU_GFX_LDS_BANK29, "LDS_BANK29"},
+	{AMDGPU_GFX_LDS_BANK30, "LDS_BANK30"},
+	{AMDGPU_GFX_LDS_BANK31, "LDS_BANK31"},
+	{AMDGPU_GFX_LDS_SP_BUFFER_A, "LDS_SP_BUFFER_A"},
+	{AMDGPU_GFX_LDS_SP_BUFFER_B, "LDS_SP_BUFFER_B"},
+};
+
+static const struct amdgpu_ras_memory_id_entry gfx_v9_4_3_ras_rlc_mem_list[] = {
+	{AMDGPU_GFX_RLC_GPMF32, "RLC_GPMF32"},
+	{AMDGPU_GFX_RLC_RLCVF32, "RLC_RLCVF32"},
+	{AMDGPU_GFX_RLC_SCRATCH, "RLC_SCRATCH"},
+	{AMDGPU_GFX_RLC_SRM_ARAM, "RLC_SRM_ARAM"},
+	{AMDGPU_GFX_RLC_SRM_DRAM, "RLC_SRM_DRAM"},
+	{AMDGPU_GFX_RLC_TCTAG, "RLC_TCTAG"},
+	{AMDGPU_GFX_RLC_SPM_SE, "RLC_SPM_SE"},
+	{AMDGPU_GFX_RLC_SPM_GRBMT, "RLC_SPM_GRBMT"},
+};
+
+static const struct amdgpu_ras_memory_id_entry gfx_v9_4_3_ras_sp_mem_list[] = {
+	{AMDGPU_GFX_SP_SIMDID0, "SP_SIMDID0"},
+};
+
+static const struct amdgpu_ras_memory_id_entry gfx_v9_4_3_ras_spi_mem_list[] = {
+	{AMDGPU_GFX_SPI_MEM0, "SPI_MEM0"},
+	{AMDGPU_GFX_SPI_MEM1, "SPI_MEM1"},
+	{AMDGPU_GFX_SPI_MEM2, "SPI_MEM2"},
+	{AMDGPU_GFX_SPI_MEM3, "SPI_MEM3"},
+};
+
+static const struct amdgpu_ras_memory_id_entry gfx_v9_4_3_ras_sqc_mem_list[] = {
+	{AMDGPU_GFX_SQC_INST_CACHE_A, "SQC_INST_CACHE_A"},
+	{AMDGPU_GFX_SQC_INST_CACHE_B, "SQC_INST_CACHE_B"},
+	{AMDGPU_GFX_SQC_INST_CACHE_TAG_A, "SQC_INST_CACHE_TAG_A"},
+	{AMDGPU_GFX_SQC_INST_CACHE_TAG_B, "SQC_INST_CACHE_TAG_B"},
+	{AMDGPU_GFX_SQC_INST_CACHE_MISS_FIFO_A, "SQC_INST_CACHE_MISS_FIFO_A"},
+	{AMDGPU_GFX_SQC_INST_CACHE_MISS_FIFO_B, "SQC_INST_CACHE_MISS_FIFO_B"},
+	{AMDGPU_GFX_SQC_INST_CACHE_GATCL1_MISS_FIFO_A, "SQC_INST_CACHE_GATCL1_MISS_FIFO_A"},
+	{AMDGPU_GFX_SQC_INST_CACHE_GATCL1_MISS_FIFO_B, "SQC_INST_CACHE_GATCL1_MISS_FIFO_B"},
+	{AMDGPU_GFX_SQC_DATA_CACHE_A, "SQC_DATA_CACHE_A"},
+	{AMDGPU_GFX_SQC_DATA_CACHE_B, "SQC_DATA_CACHE_B"},
+	{AMDGPU_GFX_SQC_DATA_CACHE_TAG_A, "SQC_DATA_CACHE_TAG_A"},
+	{AMDGPU_GFX_SQC_DATA_CACHE_TAG_B, "SQC_DATA_CACHE_TAG_B"},
+	{AMDGPU_GFX_SQC_DATA_CACHE_MISS_FIFO_A, "SQC_DATA_CACHE_MISS_FIFO_A"},
+	{AMDGPU_GFX_SQC_DATA_CACHE_MISS_FIFO_B, "SQC_DATA_CACHE_MISS_FIFO_B"},
+	{AMDGPU_GFX_SQC_DATA_CACHE_HIT_FIFO_A, "SQC_DATA_CACHE_HIT_FIFO_A"},
+	{AMDGPU_GFX_SQC_DATA_CACHE_HIT_FIFO_B, "SQC_DATA_CACHE_HIT_FIFO_B"},
+	{AMDGPU_GFX_SQC_DIRTY_BIT_A, "SQC_DIRTY_BIT_A"},
+	{AMDGPU_GFX_SQC_DIRTY_BIT_B, "SQC_DIRTY_BIT_B"},
+	{AMDGPU_GFX_SQC_WRITE_DATA_BUFFER_CU0, "SQC_WRITE_DATA_BUFFER_CU0"},
+	{AMDGPU_GFX_SQC_WRITE_DATA_BUFFER_CU1, "SQC_WRITE_DATA_BUFFER_CU1"},
+	{AMDGPU_GFX_SQC_UTCL1_MISS_LFIFO_DATA_CACHE_A, "SQC_UTCL1_MISS_LFIFO_DATA_CACHE_A"},
+	{AMDGPU_GFX_SQC_UTCL1_MISS_LFIFO_DATA_CACHE_B, "SQC_UTCL1_MISS_LFIFO_DATA_CACHE_B"},
+	{AMDGPU_GFX_SQC_UTCL1_MISS_LFIFO_INST_CACHE, "SQC_UTCL1_MISS_LFIFO_INST_CACHE"},
+};
+
+static const struct amdgpu_ras_memory_id_entry gfx_v9_4_3_ras_sq_mem_list[] = {
+	{AMDGPU_GFX_SQ_SGPR_MEM0, "SQ_SGPR_MEM0"},
+	{AMDGPU_GFX_SQ_SGPR_MEM1, "SQ_SGPR_MEM1"},
+	{AMDGPU_GFX_SQ_SGPR_MEM2, "SQ_SGPR_MEM2"},
+	{AMDGPU_GFX_SQ_SGPR_MEM3, "SQ_SGPR_MEM3"},
+};
+
+static const struct amdgpu_ras_memory_id_entry gfx_v9_4_3_ras_ta_mem_list[] = {
+	{AMDGPU_GFX_TA_FS_AFIFO_RAM_LO, "TA_FS_AFIFO_RAM_LO"},
+	{AMDGPU_GFX_TA_FS_AFIFO_RAM_HI, "TA_FS_AFIFO_RAM_HI"},
+	{AMDGPU_GFX_TA_FS_CFIFO_RAM, "TA_FS_CFIFO_RAM"},
+	{AMDGPU_GFX_TA_FSX_LFIFO, "TA_FSX_LFIFO"},
+	{AMDGPU_GFX_TA_FS_DFIFO_RAM, "TA_FS_DFIFO_RAM"},
+};
+
+static const struct amdgpu_ras_memory_id_entry gfx_v9_4_3_ras_tcc_mem_list[] = {
+	{AMDGPU_GFX_TCC_MEM1, "TCC_MEM1"},
+};
+
+static const struct amdgpu_ras_memory_id_entry gfx_v9_4_3_ras_tca_mem_list[] = {
+	{AMDGPU_GFX_TCA_MEM1, "TCA_MEM1"},
+};
+
+static const struct amdgpu_ras_memory_id_entry gfx_v9_4_3_ras_tci_mem_list[] = {
+	{AMDGPU_GFX_TCIW_MEM, "TCIW_MEM"},
+};
+
+static const struct amdgpu_ras_memory_id_entry gfx_v9_4_3_ras_tcp_mem_list[] = {
+	{AMDGPU_GFX_TCP_LFIFO0, "TCP_LFIFO0"},
+	{AMDGPU_GFX_TCP_SET0BANK0_RAM, "TCP_SET0BANK0_RAM"},
+	{AMDGPU_GFX_TCP_SET0BANK1_RAM, "TCP_SET0BANK1_RAM"},
+	{AMDGPU_GFX_TCP_SET0BANK2_RAM, "TCP_SET0BANK2_RAM"},
+	{AMDGPU_GFX_TCP_SET0BANK3_RAM, "TCP_SET0BANK3_RAM"},
+	{AMDGPU_GFX_TCP_SET1BANK0_RAM, "TCP_SET1BANK0_RAM"},
+	{AMDGPU_GFX_TCP_SET1BANK1_RAM, "TCP_SET1BANK1_RAM"},
+	{AMDGPU_GFX_TCP_SET1BANK2_RAM, "TCP_SET1BANK2_RAM"},
+	{AMDGPU_GFX_TCP_SET1BANK3_RAM, "TCP_SET1BANK3_RAM"},
+	{AMDGPU_GFX_TCP_SET2BANK0_RAM, "TCP_SET2BANK0_RAM"},
+	{AMDGPU_GFX_TCP_SET2BANK1_RAM, "TCP_SET2BANK1_RAM"},
+	{AMDGPU_GFX_TCP_SET2BANK2_RAM, "TCP_SET2BANK2_RAM"},
+	{AMDGPU_GFX_TCP_SET2BANK3_RAM, "TCP_SET2BANK3_RAM"},
+	{AMDGPU_GFX_TCP_SET3BANK0_RAM, "TCP_SET3BANK0_RAM"},
+	{AMDGPU_GFX_TCP_SET3BANK1_RAM, "TCP_SET3BANK1_RAM"},
+	{AMDGPU_GFX_TCP_SET3BANK2_RAM, "TCP_SET3BANK2_RAM"},
+	{AMDGPU_GFX_TCP_SET3BANK3_RAM, "TCP_SET3BANK3_RAM"},
+	{AMDGPU_GFX_TCP_VM_FIFO, "TCP_VM_FIFO"},
+	{AMDGPU_GFX_TCP_DB_TAGRAM0, "TCP_DB_TAGRAM0"},
+	{AMDGPU_GFX_TCP_DB_TAGRAM1, "TCP_DB_TAGRAM1"},
+	{AMDGPU_GFX_TCP_DB_TAGRAM2, "TCP_DB_TAGRAM2"},
+	{AMDGPU_GFX_TCP_DB_TAGRAM3, "TCP_DB_TAGRAM3"},
+	{AMDGPU_GFX_TCP_UTCL1_LFIFO_PROBE0, "TCP_UTCL1_LFIFO_PROBE0"},
+	{AMDGPU_GFX_TCP_UTCL1_LFIFO_PROBE1, "TCP_UTCL1_LFIFO_PROBE1"},
+	{AMDGPU_GFX_TCP_CMD_FIFO, "TCP_CMD_FIFO"},
+};
+
+static const struct amdgpu_ras_memory_id_entry gfx_v9_4_3_ras_td_mem_list[] = {
+	{AMDGPU_GFX_TD_UTD_CS_FIFO_MEM, "TD_UTD_CS_FIFO_MEM"},
+	{AMDGPU_GFX_TD_UTD_SS_FIFO_LO_MEM, "TD_UTD_SS_FIFO_LO_MEM"},
+	{AMDGPU_GFX_TD_UTD_SS_FIFO_HI_MEM, "TD_UTD_SS_FIFO_HI_MEM"},
+};
+
+static const struct amdgpu_ras_memory_id_entry gfx_v9_4_3_ras_tcx_mem_list[] = {
+	{AMDGPU_GFX_TCX_FIFOD0, "TCX_FIFOD0"},
+	{AMDGPU_GFX_TCX_FIFOD1, "TCX_FIFOD1"},
+	{AMDGPU_GFX_TCX_FIFOD2, "TCX_FIFOD2"},
+	{AMDGPU_GFX_TCX_FIFOD3, "TCX_FIFOD3"},
+	{AMDGPU_GFX_TCX_FIFOD4, "TCX_FIFOD4"},
+	{AMDGPU_GFX_TCX_FIFOD5, "TCX_FIFOD5"},
+	{AMDGPU_GFX_TCX_FIFOD6, "TCX_FIFOD6"},
+	{AMDGPU_GFX_TCX_FIFOD7, "TCX_FIFOD7"},
+	{AMDGPU_GFX_TCX_FIFOB0, "TCX_FIFOB0"},
+	{AMDGPU_GFX_TCX_FIFOB1, "TCX_FIFOB1"},
+	{AMDGPU_GFX_TCX_FIFOB2, "TCX_FIFOB2"},
+	{AMDGPU_GFX_TCX_FIFOB3, "TCX_FIFOB3"},
+	{AMDGPU_GFX_TCX_FIFOB4, "TCX_FIFOB4"},
+	{AMDGPU_GFX_TCX_FIFOB5, "TCX_FIFOB5"},
+	{AMDGPU_GFX_TCX_FIFOB6, "TCX_FIFOB6"},
+	{AMDGPU_GFX_TCX_FIFOB7, "TCX_FIFOB7"},
+	{AMDGPU_GFX_TCX_FIFOA0, "TCX_FIFOA0"},
+	{AMDGPU_GFX_TCX_FIFOA1, "TCX_FIFOA1"},
+	{AMDGPU_GFX_TCX_FIFOA2, "TCX_FIFOA2"},
+	{AMDGPU_GFX_TCX_FIFOA3, "TCX_FIFOA3"},
+	{AMDGPU_GFX_TCX_FIFOA4, "TCX_FIFOA4"},
+	{AMDGPU_GFX_TCX_FIFOA5, "TCX_FIFOA5"},
+	{AMDGPU_GFX_TCX_FIFOA6, "TCX_FIFOA6"},
+	{AMDGPU_GFX_TCX_FIFOA7, "TCX_FIFOA7"},
+	{AMDGPU_GFX_TCX_CFIFO0, "TCX_CFIFO0"},
+	{AMDGPU_GFX_TCX_CFIFO1, "TCX_CFIFO1"},
+	{AMDGPU_GFX_TCX_CFIFO2, "TCX_CFIFO2"},
+	{AMDGPU_GFX_TCX_CFIFO3, "TCX_CFIFO3"},
+	{AMDGPU_GFX_TCX_CFIFO4, "TCX_CFIFO4"},
+	{AMDGPU_GFX_TCX_CFIFO5, "TCX_CFIFO5"},
+	{AMDGPU_GFX_TCX_CFIFO6, "TCX_CFIFO6"},
+	{AMDGPU_GFX_TCX_CFIFO7, "TCX_CFIFO7"},
+	{AMDGPU_GFX_TCX_FIFO_ACKB0, "TCX_FIFO_ACKB0"},
+	{AMDGPU_GFX_TCX_FIFO_ACKB1, "TCX_FIFO_ACKB1"},
+	{AMDGPU_GFX_TCX_FIFO_ACKB2, "TCX_FIFO_ACKB2"},
+	{AMDGPU_GFX_TCX_FIFO_ACKB3, "TCX_FIFO_ACKB3"},
+	{AMDGPU_GFX_TCX_FIFO_ACKB4, "TCX_FIFO_ACKB4"},
+	{AMDGPU_GFX_TCX_FIFO_ACKB5, "TCX_FIFO_ACKB5"},
+	{AMDGPU_GFX_TCX_FIFO_ACKB6, "TCX_FIFO_ACKB6"},
+	{AMDGPU_GFX_TCX_FIFO_ACKB7, "TCX_FIFO_ACKB7"},
+	{AMDGPU_GFX_TCX_FIFO_ACKD0, "TCX_FIFO_ACKD0"},
+	{AMDGPU_GFX_TCX_FIFO_ACKD1, "TCX_FIFO_ACKD1"},
+	{AMDGPU_GFX_TCX_FIFO_ACKD2, "TCX_FIFO_ACKD2"},
+	{AMDGPU_GFX_TCX_FIFO_ACKD3, "TCX_FIFO_ACKD3"},
+	{AMDGPU_GFX_TCX_FIFO_ACKD4, "TCX_FIFO_ACKD4"},
+	{AMDGPU_GFX_TCX_FIFO_ACKD5, "TCX_FIFO_ACKD5"},
+	{AMDGPU_GFX_TCX_FIFO_ACKD6, "TCX_FIFO_ACKD6"},
+	{AMDGPU_GFX_TCX_FIFO_ACKD7, "TCX_FIFO_ACKD7"},
+	{AMDGPU_GFX_TCX_DST_FIFOA0, "TCX_DST_FIFOA0"},
+	{AMDGPU_GFX_TCX_DST_FIFOA1, "TCX_DST_FIFOA1"},
+	{AMDGPU_GFX_TCX_DST_FIFOA2, "TCX_DST_FIFOA2"},
+	{AMDGPU_GFX_TCX_DST_FIFOA3, "TCX_DST_FIFOA3"},
+	{AMDGPU_GFX_TCX_DST_FIFOA4, "TCX_DST_FIFOA4"},
+	{AMDGPU_GFX_TCX_DST_FIFOA5, "TCX_DST_FIFOA5"},
+	{AMDGPU_GFX_TCX_DST_FIFOA6, "TCX_DST_FIFOA6"},
+	{AMDGPU_GFX_TCX_DST_FIFOA7, "TCX_DST_FIFOA7"},
+	{AMDGPU_GFX_TCX_DST_FIFOB0, "TCX_DST_FIFOB0"},
+	{AMDGPU_GFX_TCX_DST_FIFOB1, "TCX_DST_FIFOB1"},
+	{AMDGPU_GFX_TCX_DST_FIFOB2, "TCX_DST_FIFOB2"},
+	{AMDGPU_GFX_TCX_DST_FIFOB3, "TCX_DST_FIFOB3"},
+	{AMDGPU_GFX_TCX_DST_FIFOB4, "TCX_DST_FIFOB4"},
+	{AMDGPU_GFX_TCX_DST_FIFOB5, "TCX_DST_FIFOB5"},
+	{AMDGPU_GFX_TCX_DST_FIFOB6, "TCX_DST_FIFOB6"},
+	{AMDGPU_GFX_TCX_DST_FIFOB7, "TCX_DST_FIFOB7"},
+	{AMDGPU_GFX_TCX_DST_FIFOD0, "TCX_DST_FIFOD0"},
+	{AMDGPU_GFX_TCX_DST_FIFOD1, "TCX_DST_FIFOD1"},
+	{AMDGPU_GFX_TCX_DST_FIFOD2, "TCX_DST_FIFOD2"},
+	{AMDGPU_GFX_TCX_DST_FIFOD3, "TCX_DST_FIFOD3"},
+	{AMDGPU_GFX_TCX_DST_FIFOD4, "TCX_DST_FIFOD4"},
+	{AMDGPU_GFX_TCX_DST_FIFOD5, "TCX_DST_FIFOD5"},
+	{AMDGPU_GFX_TCX_DST_FIFOD6, "TCX_DST_FIFOD6"},
+	{AMDGPU_GFX_TCX_DST_FIFOD7, "TCX_DST_FIFOD7"},
+	{AMDGPU_GFX_TCX_DST_FIFO_ACKB0, "TCX_DST_FIFO_ACKB0"},
+	{AMDGPU_GFX_TCX_DST_FIFO_ACKB1, "TCX_DST_FIFO_ACKB1"},
+	{AMDGPU_GFX_TCX_DST_FIFO_ACKB2, "TCX_DST_FIFO_ACKB2"},
+	{AMDGPU_GFX_TCX_DST_FIFO_ACKB3, "TCX_DST_FIFO_ACKB3"},
+	{AMDGPU_GFX_TCX_DST_FIFO_ACKB4, "TCX_DST_FIFO_ACKB4"},
+	{AMDGPU_GFX_TCX_DST_FIFO_ACKB5, "TCX_DST_FIFO_ACKB5"},
+	{AMDGPU_GFX_TCX_DST_FIFO_ACKB6, "TCX_DST_FIFO_ACKB6"},
+	{AMDGPU_GFX_TCX_DST_FIFO_ACKB7, "TCX_DST_FIFO_ACKB7"},
+	{AMDGPU_GFX_TCX_DST_FIFO_ACKD0, "TCX_DST_FIFO_ACKD0"},
+	{AMDGPU_GFX_TCX_DST_FIFO_ACKD1, "TCX_DST_FIFO_ACKD1"},
+	{AMDGPU_GFX_TCX_DST_FIFO_ACKD2, "TCX_DST_FIFO_ACKD2"},
+	{AMDGPU_GFX_TCX_DST_FIFO_ACKD3, "TCX_DST_FIFO_ACKD3"},
+	{AMDGPU_GFX_TCX_DST_FIFO_ACKD4, "TCX_DST_FIFO_ACKD4"},
+	{AMDGPU_GFX_TCX_DST_FIFO_ACKD5, "TCX_DST_FIFO_ACKD5"},
+	{AMDGPU_GFX_TCX_DST_FIFO_ACKD6, "TCX_DST_FIFO_ACKD6"},
+	{AMDGPU_GFX_TCX_DST_FIFO_ACKD7, "TCX_DST_FIFO_ACKD7"},
+};
+
+static const struct amdgpu_ras_memory_id_entry gfx_v9_4_3_ras_atc_l2_mem_list[] = {
+	{AMDGPU_GFX_ATC_L2_MEM, "ATC_L2_MEM"},
+};
+
+static const struct amdgpu_ras_memory_id_entry gfx_v9_4_3_ras_utcl2_mem_list[] = {
+	{AMDGPU_GFX_UTCL2_MEM, "UTCL2_MEM"},
+};
+
+static const struct amdgpu_ras_memory_id_entry gfx_v9_4_3_ras_vml2_mem_list[] = {
+	{AMDGPU_GFX_VML2_MEM, "VML2_MEM"},
+};
+
+static const struct amdgpu_ras_memory_id_entry gfx_v9_4_3_ras_vml2_walker_mem_list[] = {
+	{AMDGPU_GFX_VML2_WALKER_MEM, "VML2_WALKER_MEM"},
+};
+
+static const struct amdgpu_gfx_ras_mem_id_entry gfx_v9_4_3_ras_mem_list_array[AMDGPU_GFX_MEM_TYPE_NUM] = {
+	AMDGPU_GFX_MEMID_ENT(gfx_v9_4_3_ras_cp_mem_list)
+	AMDGPU_GFX_MEMID_ENT(gfx_v9_4_3_ras_gcea_mem_list)
+	AMDGPU_GFX_MEMID_ENT(gfx_v9_4_3_ras_gc_cane_mem_list)
+	AMDGPU_GFX_MEMID_ENT(gfx_v9_4_3_ras_gcutcl2_mem_list)
+	AMDGPU_GFX_MEMID_ENT(gfx_v9_4_3_ras_gds_mem_list)
+	AMDGPU_GFX_MEMID_ENT(gfx_v9_4_3_ras_lds_mem_list)
+	AMDGPU_GFX_MEMID_ENT(gfx_v9_4_3_ras_rlc_mem_list)
+	AMDGPU_GFX_MEMID_ENT(gfx_v9_4_3_ras_sp_mem_list)
+	AMDGPU_GFX_MEMID_ENT(gfx_v9_4_3_ras_spi_mem_list)
+	AMDGPU_GFX_MEMID_ENT(gfx_v9_4_3_ras_sqc_mem_list)
+	AMDGPU_GFX_MEMID_ENT(gfx_v9_4_3_ras_sq_mem_list)
+	AMDGPU_GFX_MEMID_ENT(gfx_v9_4_3_ras_ta_mem_list)
+	AMDGPU_GFX_MEMID_ENT(gfx_v9_4_3_ras_tcc_mem_list)
+	AMDGPU_GFX_MEMID_ENT(gfx_v9_4_3_ras_tca_mem_list)
+	AMDGPU_GFX_MEMID_ENT(gfx_v9_4_3_ras_tci_mem_list)
+	AMDGPU_GFX_MEMID_ENT(gfx_v9_4_3_ras_tcp_mem_list)
+	AMDGPU_GFX_MEMID_ENT(gfx_v9_4_3_ras_td_mem_list)
+	AMDGPU_GFX_MEMID_ENT(gfx_v9_4_3_ras_tcx_mem_list)
+	AMDGPU_GFX_MEMID_ENT(gfx_v9_4_3_ras_atc_l2_mem_list)
+	AMDGPU_GFX_MEMID_ENT(gfx_v9_4_3_ras_utcl2_mem_list)
+	AMDGPU_GFX_MEMID_ENT(gfx_v9_4_3_ras_vml2_mem_list)
+	AMDGPU_GFX_MEMID_ENT(gfx_v9_4_3_ras_vml2_walker_mem_list)
+};
+
+static const struct amdgpu_gfx_ras_reg_entry gfx_v9_4_3_ce_reg_list[] = {
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regRLC_CE_ERR_STATUS_LOW, regRLC_CE_ERR_STATUS_HIGH),
+	    1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "RLC"},
+	    AMDGPU_GFX_RLC_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regCPC_CE_ERR_STATUS_LO, regCPC_CE_ERR_STATUS_HI),
+	    1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "CPC"},
+	    AMDGPU_GFX_CP_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regCPF_CE_ERR_STATUS_LO, regCPF_CE_ERR_STATUS_HI),
+	    1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "CPF"},
+	    AMDGPU_GFX_CP_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regCPG_CE_ERR_STATUS_LO, regCPG_CE_ERR_STATUS_HI),
+	    1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "CPG"},
+	    AMDGPU_GFX_CP_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regGDS_CE_ERR_STATUS_LO, regGDS_CE_ERR_STATUS_HI),
+	    1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "GDS"},
+	    AMDGPU_GFX_GDS_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regGC_CANE_CE_ERR_STATUS_LO, regGC_CANE_CE_ERR_STATUS_HI),
+	    1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "CANE"},
+	    AMDGPU_GFX_GC_CANE_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regSPI_CE_ERR_STATUS_LO, regSPI_CE_ERR_STATUS_HI),
+	    1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "SPI"},
+	    AMDGPU_GFX_SPI_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regSP0_CE_ERR_STATUS_LO, regSP0_CE_ERR_STATUS_HI),
+	    10, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "SP0"},
+	    AMDGPU_GFX_SP_MEM, 4},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regSP1_CE_ERR_STATUS_LO, regSP1_CE_ERR_STATUS_HI),
+	    10, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "SP1"},
+	    AMDGPU_GFX_SP_MEM, 4},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regSQ_CE_ERR_STATUS_LO, regSQ_CE_ERR_STATUS_HI),
+	    10, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "SQ"},
+	    AMDGPU_GFX_SQ_MEM, 4},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regSQC_CE_EDC_LO, regSQC_CE_EDC_HI),
+	    5, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "SQC"},
+	    AMDGPU_GFX_SQC_MEM, 4},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regTCX_CE_ERR_STATUS_LO, regTCX_CE_ERR_STATUS_HI),
+	    2, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "TCX"},
+	    AMDGPU_GFX_TCX_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regTCC_CE_ERR_STATUS_LO, regTCC_CE_ERR_STATUS_HI),
+	    16, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "TCC"},
+	    AMDGPU_GFX_TCC_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regTA_CE_EDC_LO, regTA_CE_EDC_HI),
+	    10, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "TA"},
+	    AMDGPU_GFX_TA_MEM, 4},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regTCI_CE_EDC_LO_REG, regTCI_CE_EDC_HI_REG),
+	    27, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "TCI"},
+	    AMDGPU_GFX_TCI_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regTCP_CE_EDC_LO_REG, regTCP_CE_EDC_HI_REG),
+	    10, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "TCP"},
+	    AMDGPU_GFX_TCP_MEM, 4},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regTD_CE_EDC_LO, regTD_CE_EDC_HI),
+	    10, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "TD"},
+	    AMDGPU_GFX_TD_MEM, 4},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regGCEA_CE_ERR_STATUS_LO, regGCEA_CE_ERR_STATUS_HI),
+	    16, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "GCEA"},
+	    AMDGPU_GFX_GCEA_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regLDS_CE_ERR_STATUS_LO, regLDS_CE_ERR_STATUS_HI),
+	    10, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "LDS"},
+	    AMDGPU_GFX_LDS_MEM, 4},
+};
+
+static const struct amdgpu_gfx_ras_reg_entry gfx_v9_4_3_ue_reg_list[] = {
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regRLC_UE_ERR_STATUS_LOW, regRLC_UE_ERR_STATUS_HIGH),
+	    1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "RLC"},
+	    AMDGPU_GFX_RLC_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regCPC_UE_ERR_STATUS_LO, regCPC_UE_ERR_STATUS_HI),
+	    1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "CPC"},
+	    AMDGPU_GFX_CP_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regCPF_UE_ERR_STATUS_LO, regCPF_UE_ERR_STATUS_HI),
+	    1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "CPF"},
+	    AMDGPU_GFX_CP_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regCPG_UE_ERR_STATUS_LO, regCPG_UE_ERR_STATUS_HI),
+	    1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "CPG"},
+	    AMDGPU_GFX_CP_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regGDS_UE_ERR_STATUS_LO, regGDS_UE_ERR_STATUS_HI),
+	    1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "GDS"},
+	    AMDGPU_GFX_GDS_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regGC_CANE_UE_ERR_STATUS_LO, regGC_CANE_UE_ERR_STATUS_HI),
+	    1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "CANE"},
+	    AMDGPU_GFX_GC_CANE_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regSPI_UE_ERR_STATUS_LO, regSPI_UE_ERR_STATUS_HI),
+	    1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "SPI"},
+	    AMDGPU_GFX_SPI_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regSP0_UE_ERR_STATUS_LO, regSP0_UE_ERR_STATUS_HI),
+	    10, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "SP0"},
+	    AMDGPU_GFX_SP_MEM, 4},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regSP1_UE_ERR_STATUS_LO, regSP1_UE_ERR_STATUS_HI),
+	    10, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "SP1"},
+	    AMDGPU_GFX_SP_MEM, 4},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regSQ_UE_ERR_STATUS_LO, regSQ_UE_ERR_STATUS_HI),
+	    10, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "SQ"},
+	    AMDGPU_GFX_SQ_MEM, 4},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regSQC_UE_EDC_LO, regSQC_UE_EDC_HI),
+	    5, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "SQC"},
+	    AMDGPU_GFX_SQC_MEM, 4},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regTCX_UE_ERR_STATUS_LO, regTCX_UE_ERR_STATUS_HI),
+	    2, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "TCX"},
+	    AMDGPU_GFX_TCX_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regTCC_UE_ERR_STATUS_LO, regTCC_UE_ERR_STATUS_HI),
+	    16, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "TCC"},
+	    AMDGPU_GFX_TCC_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regTA_UE_EDC_LO, regTA_UE_EDC_HI),
+	    10, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "TA"},
+	    AMDGPU_GFX_TA_MEM, 4},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regTCI_UE_EDC_LO_REG, regTCI_UE_EDC_HI_REG),
+	    27, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "TCI"},
+	    AMDGPU_GFX_TCI_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regTCP_UE_EDC_LO_REG, regTCP_UE_EDC_HI_REG),
+	    10, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "TCP"},
+	    AMDGPU_GFX_TCP_MEM, 4},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regTD_UE_EDC_LO, regTD_UE_EDC_HI),
+	    10, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "TD"},
+	    AMDGPU_GFX_TD_MEM, 4},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regTCA_UE_ERR_STATUS_LO, regTCA_UE_ERR_STATUS_HI),
+	    2, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "TCA"},
+	    AMDGPU_GFX_TCA_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regGCEA_UE_ERR_STATUS_LO, regGCEA_UE_ERR_STATUS_HI),
+	    16, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "GCEA"},
+	    AMDGPU_GFX_GCEA_MEM, 1},
+	{{AMDGPU_RAS_REG_ENTRY(GC, 0, regLDS_UE_ERR_STATUS_LO, regLDS_UE_ERR_STATUS_HI),
+	    10, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "LDS"},
+	    AMDGPU_GFX_LDS_MEM, 4},
+};
+
+static void gfx_v9_4_3_inst_query_ras_err_count(struct amdgpu_device *adev,
+					void *ras_error_status, int xcc_id)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+	unsigned long ce_count = 0, ue_count = 0;
+	uint32_t i, j, k;
+
+	/* NOTE: convert xcc_id to physical XCD ID (XCD0 or XCD1) */
+	struct amdgpu_smuio_mcm_config_info mcm_info = {
+		.socket_id = adev->smuio.funcs->get_socket_id(adev),
+		.die_id = xcc_id & 0x01 ? 1 : 0,
+	};
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	for (i = 0; i < ARRAY_SIZE(gfx_v9_4_3_ce_reg_list); i++) {
+		for (j = 0; j < gfx_v9_4_3_ce_reg_list[i].se_num; j++) {
+			for (k = 0; k < gfx_v9_4_3_ce_reg_list[i].reg_entry.reg_inst; k++) {
+				/* no need to select if instance number is 1 */
+				if (gfx_v9_4_3_ce_reg_list[i].se_num > 1 ||
+				    gfx_v9_4_3_ce_reg_list[i].reg_entry.reg_inst > 1)
+					gfx_v9_4_3_xcc_select_se_sh(adev, j, 0, k, xcc_id);
+
+				amdgpu_ras_inst_query_ras_error_count(adev,
+					&(gfx_v9_4_3_ce_reg_list[i].reg_entry),
+					1,
+					gfx_v9_4_3_ras_mem_list_array[gfx_v9_4_3_ce_reg_list[i].mem_id_type].mem_id_ent,
+					gfx_v9_4_3_ras_mem_list_array[gfx_v9_4_3_ce_reg_list[i].mem_id_type].size,
+					GET_INST(GC, xcc_id),
+					AMDGPU_RAS_ERROR__SINGLE_CORRECTABLE,
+					&ce_count);
+
+				amdgpu_ras_inst_query_ras_error_count(adev,
+					&(gfx_v9_4_3_ue_reg_list[i].reg_entry),
+					1,
+					gfx_v9_4_3_ras_mem_list_array[gfx_v9_4_3_ue_reg_list[i].mem_id_type].mem_id_ent,
+					gfx_v9_4_3_ras_mem_list_array[gfx_v9_4_3_ue_reg_list[i].mem_id_type].size,
+					GET_INST(GC, xcc_id),
+					AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE,
+					&ue_count);
+			}
+		}
+	}
+
+	/* handle extra register entries of UE */
+	for (; i < ARRAY_SIZE(gfx_v9_4_3_ue_reg_list); i++) {
+		for (j = 0; j < gfx_v9_4_3_ue_reg_list[i].se_num; j++) {
+			for (k = 0; k < gfx_v9_4_3_ue_reg_list[i].reg_entry.reg_inst; k++) {
+				/* no need to select if instance number is 1 */
+				if (gfx_v9_4_3_ue_reg_list[i].se_num > 1 ||
+					gfx_v9_4_3_ue_reg_list[i].reg_entry.reg_inst > 1)
+					gfx_v9_4_3_xcc_select_se_sh(adev, j, 0, k, xcc_id);
+
+				amdgpu_ras_inst_query_ras_error_count(adev,
+					&(gfx_v9_4_3_ue_reg_list[i].reg_entry),
+					1,
+					gfx_v9_4_3_ras_mem_list_array[gfx_v9_4_3_ue_reg_list[i].mem_id_type].mem_id_ent,
+					gfx_v9_4_3_ras_mem_list_array[gfx_v9_4_3_ue_reg_list[i].mem_id_type].size,
+					GET_INST(GC, xcc_id),
+					AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE,
+					&ue_count);
+			}
+		}
+	}
+
+	gfx_v9_4_3_xcc_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff,
+			xcc_id);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	/* the caller should make sure initialize value of
+	 * err_data->ue_count and err_data->ce_count
+	 */
+	amdgpu_ras_error_statistic_ue_count(err_data, &mcm_info, ue_count);
+	amdgpu_ras_error_statistic_ce_count(err_data, &mcm_info, ce_count);
+}
+
+static void gfx_v9_4_3_inst_reset_ras_err_count(struct amdgpu_device *adev,
+					void *ras_error_status, int xcc_id)
+{
+	uint32_t i, j, k;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	for (i = 0; i < ARRAY_SIZE(gfx_v9_4_3_ce_reg_list); i++) {
+		for (j = 0; j < gfx_v9_4_3_ce_reg_list[i].se_num; j++) {
+			for (k = 0; k < gfx_v9_4_3_ce_reg_list[i].reg_entry.reg_inst; k++) {
+				/* no need to select if instance number is 1 */
+				if (gfx_v9_4_3_ce_reg_list[i].se_num > 1 ||
+				    gfx_v9_4_3_ce_reg_list[i].reg_entry.reg_inst > 1)
+					gfx_v9_4_3_xcc_select_se_sh(adev, j, 0, k, xcc_id);
+
+				amdgpu_ras_inst_reset_ras_error_count(adev,
+					&(gfx_v9_4_3_ce_reg_list[i].reg_entry),
+					1,
+					GET_INST(GC, xcc_id));
+
+				amdgpu_ras_inst_reset_ras_error_count(adev,
+					&(gfx_v9_4_3_ue_reg_list[i].reg_entry),
+					1,
+					GET_INST(GC, xcc_id));
+			}
+		}
+	}
+
+	/* handle extra register entries of UE */
+	for (; i < ARRAY_SIZE(gfx_v9_4_3_ue_reg_list); i++) {
+		for (j = 0; j < gfx_v9_4_3_ue_reg_list[i].se_num; j++) {
+			for (k = 0; k < gfx_v9_4_3_ue_reg_list[i].reg_entry.reg_inst; k++) {
+				/* no need to select if instance number is 1 */
+				if (gfx_v9_4_3_ue_reg_list[i].se_num > 1 ||
+					gfx_v9_4_3_ue_reg_list[i].reg_entry.reg_inst > 1)
+					gfx_v9_4_3_xcc_select_se_sh(adev, j, 0, k, xcc_id);
+
+				amdgpu_ras_inst_reset_ras_error_count(adev,
+					&(gfx_v9_4_3_ue_reg_list[i].reg_entry),
+					1,
+					GET_INST(GC, xcc_id));
+			}
+		}
+	}
+
+	gfx_v9_4_3_xcc_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff,
+			xcc_id);
+	mutex_unlock(&adev->grbm_idx_mutex);
+}
+
+static void gfx_v9_4_3_inst_enable_watchdog_timer(struct amdgpu_device *adev,
+					void *ras_error_status, int xcc_id)
+{
+	uint32_t i;
+	uint32_t data;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	data = RREG32_SOC15(GC, GET_INST(GC, 0), regSQ_TIMEOUT_CONFIG);
+	data = REG_SET_FIELD(data, SQ_TIMEOUT_CONFIG, TIMEOUT_FATAL_DISABLE,
+			     amdgpu_watchdog_timer.timeout_fatal_disable ? 1 : 0);
+
+	if (amdgpu_watchdog_timer.timeout_fatal_disable &&
+	    (amdgpu_watchdog_timer.period < 1 ||
+	     amdgpu_watchdog_timer.period > 0x23)) {
+		dev_warn(adev->dev, "Watchdog period range is 1 to 0x23\n");
+		amdgpu_watchdog_timer.period = 0x23;
+	}
+	data = REG_SET_FIELD(data, SQ_TIMEOUT_CONFIG, PERIOD_SEL,
+			     amdgpu_watchdog_timer.period);
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		gfx_v9_4_3_xcc_select_se_sh(adev, i, 0xffffffff, 0xffffffff, xcc_id);
+		WREG32_SOC15(GC, GET_INST(GC, xcc_id), regSQ_TIMEOUT_CONFIG, data);
+	}
+	gfx_v9_4_3_xcc_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff,
+			xcc_id);
+	mutex_unlock(&adev->grbm_idx_mutex);
+}
+
+static void gfx_v9_4_3_query_ras_error_count(struct amdgpu_device *adev,
+					void *ras_error_status)
+{
+	amdgpu_gfx_ras_error_func(adev, ras_error_status,
+			gfx_v9_4_3_inst_query_ras_err_count);
+}
+
+static void gfx_v9_4_3_reset_ras_error_count(struct amdgpu_device *adev)
+{
+	amdgpu_gfx_ras_error_func(adev, NULL, gfx_v9_4_3_inst_reset_ras_err_count);
+}
+
+static void gfx_v9_4_3_enable_watchdog_timer(struct amdgpu_device *adev)
+{
+	amdgpu_gfx_ras_error_func(adev, NULL, gfx_v9_4_3_inst_enable_watchdog_timer);
+}
+
+static void gfx_v9_4_3_ring_insert_nop(struct amdgpu_ring *ring, uint32_t num_nop)
+{
+	/* Header itself is a NOP packet */
+	if (num_nop == 1) {
+		amdgpu_ring_write(ring, ring->funcs->nop);
+		return;
+	}
+
+	/* Max HW optimization till 0x3ffe, followed by remaining one NOP at a time*/
+	amdgpu_ring_write(ring, PACKET3(PACKET3_NOP, min(num_nop - 2, 0x3ffe)));
+
+	/* Header is at index 0, followed by num_nops - 1 NOP packet's */
+	amdgpu_ring_insert_nop(ring, num_nop - 1);
+}
+
+static void gfx_v9_4_3_ip_print(struct amdgpu_ip_block *ip_block, struct drm_printer *p)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	uint32_t i, j, k;
+	uint32_t xcc_id, xcc_offset, inst_offset;
+	uint32_t num_xcc, reg, num_inst;
+	uint32_t reg_count = ARRAY_SIZE(gc_reg_list_9_4_3);
+
+	if (!adev->gfx.ip_dump_core)
+		return;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	drm_printf(p, "Number of Instances:%d\n", num_xcc);
+	for (xcc_id = 0; xcc_id < num_xcc; xcc_id++) {
+		xcc_offset = xcc_id * reg_count;
+		drm_printf(p, "\nInstance id:%d\n", xcc_id);
+		for (i = 0; i < reg_count; i++)
+			drm_printf(p, "%-50s \t 0x%08x\n",
+				   gc_reg_list_9_4_3[i].reg_name,
+				   adev->gfx.ip_dump_core[xcc_offset + i]);
+	}
+
+	/* print compute queue registers for all instances */
+	if (!adev->gfx.ip_dump_compute_queues)
+		return;
+
+	num_inst = adev->gfx.mec.num_mec * adev->gfx.mec.num_pipe_per_mec *
+		adev->gfx.mec.num_queue_per_pipe;
+
+	reg_count = ARRAY_SIZE(gc_cp_reg_list_9_4_3);
+	drm_printf(p, "\nnum_xcc: %d num_mec: %d num_pipe: %d num_queue: %d\n",
+		   num_xcc,
+		   adev->gfx.mec.num_mec,
+		   adev->gfx.mec.num_pipe_per_mec,
+		   adev->gfx.mec.num_queue_per_pipe);
+
+	for (xcc_id = 0; xcc_id < num_xcc; xcc_id++) {
+		xcc_offset = xcc_id * reg_count * num_inst;
+		inst_offset = 0;
+		for (i = 0; i < adev->gfx.mec.num_mec; i++) {
+			for (j = 0; j < adev->gfx.mec.num_pipe_per_mec; j++) {
+				for (k = 0; k < adev->gfx.mec.num_queue_per_pipe; k++) {
+					drm_printf(p,
+						   "\nxcc:%d mec:%d, pipe:%d, queue:%d\n",
+						    xcc_id, i, j, k);
+					for (reg = 0; reg < reg_count; reg++) {
+						if (i && gc_cp_reg_list_9_4_3[reg].reg_offset ==
+						    regCP_MEC_ME1_HEADER_DUMP)
+							drm_printf(p,
+								   "%-50s \t 0x%08x\n",
+								   "regCP_MEC_ME2_HEADER_DUMP",
+								   adev->gfx.ip_dump_compute_queues
+								   [xcc_offset + inst_offset +
+								    reg]);
+						else
+							drm_printf(p,
+								   "%-50s \t 0x%08x\n",
+								   gc_cp_reg_list_9_4_3[reg].reg_name,
+								   adev->gfx.ip_dump_compute_queues
+								   [xcc_offset + inst_offset +
+								    reg]);
+					}
+					inst_offset += reg_count;
+				}
+			}
+		}
+	}
+}
+
+static void gfx_v9_4_3_ip_dump(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	uint32_t i, j, k;
+	uint32_t num_xcc, reg, num_inst;
+	uint32_t xcc_id, xcc_offset, inst_offset;
+	uint32_t reg_count = ARRAY_SIZE(gc_reg_list_9_4_3);
+
+	if (!adev->gfx.ip_dump_core)
+		return;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+
+	for (xcc_id = 0; xcc_id < num_xcc; xcc_id++) {
+		xcc_offset = xcc_id * reg_count;
+		for (i = 0; i < reg_count; i++)
+			adev->gfx.ip_dump_core[xcc_offset + i] =
+				RREG32(SOC15_REG_ENTRY_OFFSET_INST(gc_reg_list_9_4_3[i],
+								   GET_INST(GC, xcc_id)));
+	}
+
+	/* dump compute queue registers for all instances */
+	if (!adev->gfx.ip_dump_compute_queues)
+		return;
+
+	num_inst = adev->gfx.mec.num_mec * adev->gfx.mec.num_pipe_per_mec *
+		adev->gfx.mec.num_queue_per_pipe;
+	reg_count = ARRAY_SIZE(gc_cp_reg_list_9_4_3);
+	mutex_lock(&adev->srbm_mutex);
+	for (xcc_id = 0; xcc_id < num_xcc; xcc_id++) {
+		xcc_offset = xcc_id * reg_count * num_inst;
+		inst_offset = 0;
+		for (i = 0; i < adev->gfx.mec.num_mec; i++) {
+			for (j = 0; j < adev->gfx.mec.num_pipe_per_mec; j++) {
+				for (k = 0; k < adev->gfx.mec.num_queue_per_pipe; k++) {
+					/* ME0 is for GFX so start from 1 for CP */
+					soc15_grbm_select(adev, 1 + i, j, k, 0,
+							  GET_INST(GC, xcc_id));
+
+					for (reg = 0; reg < reg_count; reg++) {
+						if (i && gc_cp_reg_list_9_4_3[reg].reg_offset ==
+						    regCP_MEC_ME1_HEADER_DUMP)
+							adev->gfx.ip_dump_compute_queues
+								[xcc_offset +
+								 inst_offset + reg] =
+								RREG32(SOC15_REG_OFFSET(GC, GET_INST(GC, xcc_id),
+											regCP_MEC_ME2_HEADER_DUMP));
+						else
+							adev->gfx.ip_dump_compute_queues
+								[xcc_offset +
+								 inst_offset + reg] =
+								RREG32(SOC15_REG_ENTRY_OFFSET_INST(
+									       gc_cp_reg_list_9_4_3[reg],
+									       GET_INST(GC, xcc_id)));
+					}
+					inst_offset += reg_count;
+				}
+			}
+		}
+	}
+	soc15_grbm_select(adev, 0, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static void gfx_v9_4_3_ring_emit_cleaner_shader(struct amdgpu_ring *ring)
+{
+	/* Emit the cleaner shader */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_RUN_CLEANER_SHADER, 0));
+	amdgpu_ring_write(ring, 0);  /* RESERVED field, programmed to zero */
+}
+
+static const struct amd_ip_funcs gfx_v9_4_3_ip_funcs = {
+	.name = "gfx_v9_4_3",
+	.early_init = gfx_v9_4_3_early_init,
+	.late_init = gfx_v9_4_3_late_init,
+	.sw_init = gfx_v9_4_3_sw_init,
+	.sw_fini = gfx_v9_4_3_sw_fini,
+	.hw_init = gfx_v9_4_3_hw_init,
+	.hw_fini = gfx_v9_4_3_hw_fini,
+	.suspend = gfx_v9_4_3_suspend,
+	.resume = gfx_v9_4_3_resume,
+	.is_idle = gfx_v9_4_3_is_idle,
+	.wait_for_idle = gfx_v9_4_3_wait_for_idle,
+	.soft_reset = gfx_v9_4_3_soft_reset,
+	.set_clockgating_state = gfx_v9_4_3_set_clockgating_state,
+	.set_powergating_state = gfx_v9_4_3_set_powergating_state,
+	.get_clockgating_state = gfx_v9_4_3_get_clockgating_state,
+	.dump_ip_state = gfx_v9_4_3_ip_dump,
+	.print_ip_state = gfx_v9_4_3_ip_print,
+};
+
+static const struct amdgpu_ring_funcs gfx_v9_4_3_ring_funcs_compute = {
+	.type = AMDGPU_RING_TYPE_COMPUTE,
+	.align_mask = 0xff,
+	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
+	.support_64bit_ptrs = true,
+	.get_rptr = gfx_v9_4_3_ring_get_rptr_compute,
+	.get_wptr = gfx_v9_4_3_ring_get_wptr_compute,
+	.set_wptr = gfx_v9_4_3_ring_set_wptr_compute,
+	.emit_frame_size =
+		20 + /* gfx_v9_4_3_ring_emit_gds_switch */
+		7 + /* gfx_v9_4_3_ring_emit_hdp_flush */
+		5 + /* hdp invalidate */
+		7 + /* gfx_v9_4_3_ring_emit_pipeline_sync */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 5 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 +
+		2 + /* gfx_v9_4_3_ring_emit_vm_flush */
+		8 + 8 + 8 + /* gfx_v9_4_3_ring_emit_fence x3 for user fence, vm fence */
+		7 + /* gfx_v9_4_3_emit_mem_sync */
+		5 + /* gfx_v9_4_3_emit_wave_limit for updating regSPI_WCL_PIPE_PERCENT_GFX register */
+		15 + /* for updating 3 regSPI_WCL_PIPE_PERCENT_CS registers */
+		2, /* gfx_v9_4_3_ring_emit_cleaner_shader */
+	.emit_ib_size =	7, /* gfx_v9_4_3_ring_emit_ib_compute */
+	.emit_ib = gfx_v9_4_3_ring_emit_ib_compute,
+	.emit_fence = gfx_v9_4_3_ring_emit_fence,
+	.emit_pipeline_sync = gfx_v9_4_3_ring_emit_pipeline_sync,
+	.emit_vm_flush = gfx_v9_4_3_ring_emit_vm_flush,
+	.emit_gds_switch = gfx_v9_4_3_ring_emit_gds_switch,
+	.emit_hdp_flush = gfx_v9_4_3_ring_emit_hdp_flush,
+	.test_ring = gfx_v9_4_3_ring_test_ring,
+	.test_ib = gfx_v9_4_3_ring_test_ib,
+	.insert_nop = gfx_v9_4_3_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_wreg = gfx_v9_4_3_ring_emit_wreg,
+	.emit_reg_wait = gfx_v9_4_3_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = gfx_v9_4_3_ring_emit_reg_write_reg_wait,
+	.soft_recovery = gfx_v9_4_3_ring_soft_recovery,
+	.emit_mem_sync = gfx_v9_4_3_emit_mem_sync,
+	.emit_wave_limit = gfx_v9_4_3_emit_wave_limit,
+	.reset = gfx_v9_4_3_reset_kcq,
+	.emit_cleaner_shader = gfx_v9_4_3_ring_emit_cleaner_shader,
+	.begin_use = amdgpu_gfx_enforce_isolation_ring_begin_use,
+	.end_use = amdgpu_gfx_enforce_isolation_ring_end_use,
+};
+
+static const struct amdgpu_ring_funcs gfx_v9_4_3_ring_funcs_kiq = {
+	.type = AMDGPU_RING_TYPE_KIQ,
+	.align_mask = 0xff,
+	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
+	.support_64bit_ptrs = true,
+	.get_rptr = gfx_v9_4_3_ring_get_rptr_compute,
+	.get_wptr = gfx_v9_4_3_ring_get_wptr_compute,
+	.set_wptr = gfx_v9_4_3_ring_set_wptr_compute,
+	.emit_frame_size =
+		20 + /* gfx_v9_4_3_ring_emit_gds_switch */
+		7 + /* gfx_v9_4_3_ring_emit_hdp_flush */
+		5 + /* hdp invalidate */
+		7 + /* gfx_v9_4_3_ring_emit_pipeline_sync */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 5 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 +
+		2 + /* gfx_v9_4_3_ring_emit_vm_flush */
+		8 + 8 + 8, /* gfx_v9_4_3_ring_emit_fence_kiq x3 for user fence, vm fence */
+	.emit_ib_size =	7, /* gfx_v9_4_3_ring_emit_ib_compute */
+	.emit_fence = gfx_v9_4_3_ring_emit_fence_kiq,
+	.test_ring = gfx_v9_4_3_ring_test_ring,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_rreg = gfx_v9_4_3_ring_emit_rreg,
+	.emit_wreg = gfx_v9_4_3_ring_emit_wreg,
+	.emit_reg_wait = gfx_v9_4_3_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = gfx_v9_4_3_ring_emit_reg_write_reg_wait,
+};
+
+static void gfx_v9_4_3_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i, j, num_xcc;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	for (i = 0; i < num_xcc; i++) {
+		adev->gfx.kiq[i].ring.funcs = &gfx_v9_4_3_ring_funcs_kiq;
+
+		for (j = 0; j < adev->gfx.num_compute_rings; j++)
+			adev->gfx.compute_ring[j + i * adev->gfx.num_compute_rings].funcs
+					= &gfx_v9_4_3_ring_funcs_compute;
+	}
+}
+
+static const struct amdgpu_irq_src_funcs gfx_v9_4_3_eop_irq_funcs = {
+	.set = gfx_v9_4_3_set_eop_interrupt_state,
+	.process = gfx_v9_4_3_eop_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v9_4_3_priv_reg_irq_funcs = {
+	.set = gfx_v9_4_3_set_priv_reg_fault_state,
+	.process = gfx_v9_4_3_priv_reg_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v9_4_3_bad_op_irq_funcs = {
+	.set = gfx_v9_4_3_set_bad_op_fault_state,
+	.process = gfx_v9_4_3_bad_op_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v9_4_3_priv_inst_irq_funcs = {
+	.set = gfx_v9_4_3_set_priv_inst_fault_state,
+	.process = gfx_v9_4_3_priv_inst_irq,
+};
+
+static void gfx_v9_4_3_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->gfx.eop_irq.num_types = AMDGPU_CP_IRQ_LAST;
+	adev->gfx.eop_irq.funcs = &gfx_v9_4_3_eop_irq_funcs;
+
+	adev->gfx.priv_reg_irq.num_types = 1;
+	adev->gfx.priv_reg_irq.funcs = &gfx_v9_4_3_priv_reg_irq_funcs;
+
+	adev->gfx.bad_op_irq.num_types = 1;
+	adev->gfx.bad_op_irq.funcs = &gfx_v9_4_3_bad_op_irq_funcs;
+
+	adev->gfx.priv_inst_irq.num_types = 1;
+	adev->gfx.priv_inst_irq.funcs = &gfx_v9_4_3_priv_inst_irq_funcs;
+}
+
+static void gfx_v9_4_3_set_rlc_funcs(struct amdgpu_device *adev)
+{
+	adev->gfx.rlc.funcs = &gfx_v9_4_3_rlc_funcs;
+}
+
+
+static void gfx_v9_4_3_set_gds_init(struct amdgpu_device *adev)
+{
+	/* 9.4.3 variants removed all the GDS internal memory,
+	 * only support GWS opcode in kernel, like barrier
+	 * semaphore.etc */
+
+	/* init asic gds info */
+	adev->gds.gds_size = 0;
+	adev->gds.gds_compute_max_wave_id = 0;
+	adev->gds.gws_size = 64;
+	adev->gds.oa_size = 16;
+}
+
+static void gfx_v9_4_3_set_user_cu_inactive_bitmap(struct amdgpu_device *adev,
+						 u32 bitmap, int xcc_id)
+{
+	u32 data;
+
+	if (!bitmap)
+		return;
+
+	data = bitmap << GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_CUS__SHIFT;
+	data &= GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_CUS_MASK;
+
+	WREG32_SOC15(GC, GET_INST(GC, xcc_id), regGC_USER_SHADER_ARRAY_CONFIG, data);
+}
+
+static u32 gfx_v9_4_3_get_cu_active_bitmap(struct amdgpu_device *adev, int xcc_id)
+{
+	u32 data, mask;
+
+	data = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCC_GC_SHADER_ARRAY_CONFIG);
+	data |= RREG32_SOC15(GC, GET_INST(GC, xcc_id), regGC_USER_SHADER_ARRAY_CONFIG);
+
+	data &= CC_GC_SHADER_ARRAY_CONFIG__INACTIVE_CUS_MASK;
+	data >>= CC_GC_SHADER_ARRAY_CONFIG__INACTIVE_CUS__SHIFT;
+
+	mask = amdgpu_gfx_create_bitmask(adev->gfx.config.max_cu_per_sh);
+
+	return (~data) & mask;
+}
+
+static int gfx_v9_4_3_get_cu_info(struct amdgpu_device *adev,
+				 struct amdgpu_cu_info *cu_info)
+{
+	int i, j, k, prev_counter, counter, xcc_id, active_cu_number = 0;
+	u32 mask, bitmap, ao_bitmap, ao_cu_mask = 0, tmp;
+	unsigned disable_masks[4 * 4];
+	bool is_symmetric_cus;
+
+	if (!adev || !cu_info)
+		return -EINVAL;
+
+	/*
+	 * 16 comes from bitmap array size 4*4, and it can cover all gfx9 ASICs
+	 */
+	if (adev->gfx.config.max_shader_engines *
+		adev->gfx.config.max_sh_per_se > 16)
+		return -EINVAL;
+
+	amdgpu_gfx_parse_disable_cu(disable_masks,
+				    adev->gfx.config.max_shader_engines,
+				    adev->gfx.config.max_sh_per_se);
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (xcc_id = 0; xcc_id < NUM_XCC(adev->gfx.xcc_mask); xcc_id++) {
+		is_symmetric_cus = true;
+		for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+			for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+				mask = 1;
+				ao_bitmap = 0;
+				counter = 0;
+				gfx_v9_4_3_xcc_select_se_sh(adev, i, j, 0xffffffff, xcc_id);
+				gfx_v9_4_3_set_user_cu_inactive_bitmap(
+					adev,
+					disable_masks[i * adev->gfx.config.max_sh_per_se + j],
+					xcc_id);
+				bitmap = gfx_v9_4_3_get_cu_active_bitmap(adev, xcc_id);
+
+				cu_info->bitmap[xcc_id][i][j] = bitmap;
+
+				for (k = 0; k < adev->gfx.config.max_cu_per_sh; k++) {
+					if (bitmap & mask) {
+						if (counter < adev->gfx.config.max_cu_per_sh)
+							ao_bitmap |= mask;
+						counter++;
+					}
+					mask <<= 1;
+				}
+				active_cu_number += counter;
+				if (i < 2 && j < 2)
+					ao_cu_mask |= (ao_bitmap << (i * 16 + j * 8));
+				cu_info->ao_cu_bitmap[i][j] = ao_bitmap;
+			}
+			if (i && is_symmetric_cus && prev_counter != counter)
+				is_symmetric_cus = false;
+			prev_counter = counter;
+		}
+		if (is_symmetric_cus) {
+			tmp = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_CPC_DEBUG);
+			tmp = REG_SET_FIELD(tmp, CP_CPC_DEBUG, CPC_HARVESTING_RELAUNCH_DISABLE, 1);
+			tmp = REG_SET_FIELD(tmp, CP_CPC_DEBUG, CPC_HARVESTING_DISPATCH_DISABLE, 1);
+			WREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_CPC_DEBUG, tmp);
+		}
+		gfx_v9_4_3_xcc_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff,
+					    xcc_id);
+	}
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	cu_info->number = active_cu_number;
+	cu_info->ao_cu_mask = ao_cu_mask;
+	cu_info->simd_per_cu = NUM_SIMD_PER_CU;
+
+	return 0;
+}
+
+const struct amdgpu_ip_block_version gfx_v9_4_3_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_GFX,
+	.major = 9,
+	.minor = 4,
+	.rev = 3,
+	.funcs = &gfx_v9_4_3_ip_funcs,
+};
+
+static int gfx_v9_4_3_xcp_resume(void *handle, uint32_t inst_mask)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	uint32_t tmp_mask;
+	int i, r;
+
+	/* TODO : Initialize golden regs */
+	/* gfx_v9_4_3_init_golden_registers(adev); */
+
+	tmp_mask = inst_mask;
+	for_each_inst(i, tmp_mask)
+		gfx_v9_4_3_xcc_constants_init(adev, i);
+
+	if (!amdgpu_sriov_vf(adev)) {
+		tmp_mask = inst_mask;
+		for_each_inst(i, tmp_mask) {
+			r = gfx_v9_4_3_xcc_rlc_resume(adev, i);
+			if (r)
+				return r;
+		}
+	}
+
+	tmp_mask = inst_mask;
+	for_each_inst(i, tmp_mask) {
+		r = gfx_v9_4_3_xcc_cp_resume(adev, i);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+static int gfx_v9_4_3_xcp_suspend(void *handle, uint32_t inst_mask)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int i;
+
+	for_each_inst(i, inst_mask)
+		gfx_v9_4_3_xcc_fini(adev, i);
+
+	return 0;
+}
+
+struct amdgpu_xcp_ip_funcs gfx_v9_4_3_xcp_funcs = {
+	.suspend = &gfx_v9_4_3_xcp_suspend,
+	.resume = &gfx_v9_4_3_xcp_resume
+};
+
+struct amdgpu_ras_block_hw_ops  gfx_v9_4_3_ras_ops = {
+	.query_ras_error_count = &gfx_v9_4_3_query_ras_error_count,
+	.reset_ras_error_count = &gfx_v9_4_3_reset_ras_error_count,
+};
+
+static int gfx_v9_4_3_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
+{
+	int r;
+
+	r = amdgpu_ras_block_late_init(adev, ras_block);
+	if (r)
+		return r;
+
+	r = amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__GFX,
+				&gfx_v9_4_3_aca_info,
+				NULL);
+	if (r)
+		goto late_fini;
+
+	return 0;
+
+late_fini:
+	amdgpu_ras_block_late_fini(adev, ras_block);
+
+	return r;
+}
+
+struct amdgpu_gfx_ras gfx_v9_4_3_ras = {
+	.ras_block = {
+		.hw_ops = &gfx_v9_4_3_ras_ops,
+		.ras_late_init = &gfx_v9_4_3_ras_late_init,
+	},
+	.enable_watchdog_timer = &gfx_v9_4_3_enable_watchdog_timer,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.h b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.h
new file mode 100644
index 000000000000..42d67ee0e7ef
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFX_V9_4_3_H__
+#define __GFX_V9_4_3_H__
+
+extern const struct amdgpu_ip_block_version gfx_v9_4_3_ip_block;
+
+extern struct amdgpu_xcp_ip_funcs gfx_v9_4_3_xcp_funcs;
+
+#endif /* __GFX_V9_4_3_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3_cleaner_shader.asm b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3_cleaner_shader.asm
new file mode 100644
index 000000000000..d5325ef80ab0
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3_cleaner_shader.asm
@@ -0,0 +1,153 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+// This shader is to clean LDS, SGPRs and VGPRs. It is  first 64 Dwords or 256 bytes of 192 Dwords cleaner shader.
+//To turn this shader program on for complitaion change this to main and lower shader main to main_1
+ 
+// MI300 : Clear SGPRs, VGPRs and LDS
+//   Uses two kernels launched separately:
+//   1. Clean VGPRs, LDS, and lower SGPRs
+//        Launches one workgroup per CU, each workgroup with 4x wave64 per SIMD in the CU
+//        Waves are "wave64" and have 128 VGPRs each, which uses all 512 VGPRs per SIMD
+//        Waves in the workgroup share the 64KB of LDS
+//        Each wave clears SGPRs 0 - 95. Because there are 4 waves/SIMD, this is physical SGPRs 0-383
+//        Each wave clears 128 VGPRs, so all 512 in the SIMD
+//        The first wave of the workgroup clears its 64KB of LDS
+//        The shader starts with "S_BARRIER" to ensure SPI has launched all waves of the workgroup
+//          before any wave in the workgroup could end.  Without this, it is possible not all SGPRs get cleared.
+//    2. Clean remaining SGPRs
+//        Launches a workgroup with 24 waves per workgroup, yielding 6 waves per SIMD in each CU
+//        Waves are allocating 96 SGPRs
+//          CP sets up SPI_RESOURCE_RESERVE_* registers to prevent these waves from allocating SGPRs 0-223.
+//          As such, these 6 waves per SIMD are allocated physical SGPRs 224-799
+//        Barriers do not work for >16 waves per workgroup, so we cannot start with S_BARRIER
+//          Instead, the shader starts with an S_SETHALT 1. Once all waves are launched CP will send unhalt command
+//        The shader then clears all SGPRs allocated to it, cleaning out physical SGPRs 224-799
+ 
+shader main
+  asic(MI300)
+  type(CS)
+  wave_size(64)
+// Note: original source code from SQ team
+
+//   (theorhetical fastest = ~512clks vgpr + 1536 lds + ~128 sgpr  = 2176 clks)
+
+  s_cmp_eq_u32 s0, 1                                // Bit0 is set, sgpr0 is set then clear VGPRS and LDS as FW set COMPUTE_USER_DATA_3
+  s_cbranch_scc0  label_0023                        // Clean VGPRs and LDS if sgpr0 of wave is set, scc = (s3 == 1)
+  S_BARRIER
+
+  s_movk_i32    m0, 0x0000
+  s_mov_b32     s2, 0x00000078  // Loop 128/8=16 times  (loop unrolled for performance)
+  //
+  // CLEAR VGPRs
+  //
+  s_set_gpr_idx_on  s2, 0x8    // enable Dest VGPR indexing
+label_0005:
+  v_mov_b32     v0, 0
+  v_mov_b32     v1, 0
+  v_mov_b32     v2, 0
+  v_mov_b32     v3, 0
+  v_mov_b32     v4, 0
+  v_mov_b32     v5, 0
+  v_mov_b32     v6, 0
+  v_mov_b32     v7, 0
+  s_sub_u32     s2, s2, 8
+  s_set_gpr_idx_idx  s2
+  s_cbranch_scc0  label_0005
+  s_set_gpr_idx_off
+ 
+  //
+  //
+ 
+  s_mov_b32     s2, 0x80000000                      // Bit31 is first_wave
+  s_and_b32     s2, s2, s1                          // sgpr0 has tg_size (first_wave) term as in ucode only COMPUTE_PGM_RSRC2.tg_size_en is set
+  s_cbranch_scc0  label_clean_sgpr_1                // Clean LDS if its first wave of ThreadGroup/WorkGroup
+  // CLEAR LDS
+  //
+  s_mov_b32 exec_lo, 0xffffffff
+  s_mov_b32 exec_hi, 0xffffffff
+  v_mbcnt_lo_u32_b32  v1, exec_hi, 0          // Set V1 to thread-ID (0..63)
+  v_mbcnt_hi_u32_b32  v1, exec_lo, v1         // Set V1 to thread-ID (0..63)
+  v_mul_u32_u24  v1, 0x00000008, v1           // * 8, so each thread is a double-dword address (8byte)
+  s_mov_b32     s2, 0x00000003f               // 64 loop iteraions
+  s_mov_b32     m0, 0xffffffff
+  // Clear all of LDS space
+  // Each FirstWave of WorkGroup clears 64kbyte block
+ 
+label_001F:
+  ds_write2_b64  v1, v[2:3], v[2:3] offset1:32
+  ds_write2_b64  v1, v[4:5], v[4:5] offset0:64 offset1:96
+  v_add_co_u32     v1, vcc, 0x00000400, v1
+  s_sub_u32     s2, s2, 1
+  s_cbranch_scc0  label_001F
+  //
+  // CLEAR SGPRs
+  //
+label_clean_sgpr_1:
+  s_mov_b32     m0, 0x0000005c   // Loop 96/4=24 times  (loop unrolled for performance)
+  s_nop 0
+label_sgpr_loop:
+  s_movreld_b32     s0, 0
+  s_movreld_b32     s1, 0
+  s_movreld_b32     s2, 0
+  s_movreld_b32     s3, 0
+  s_sub_u32         m0, m0, 4
+  s_cbranch_scc0  label_sgpr_loop
+ 
+  //clear vcc, flat scratch
+  s_mov_b32 flat_scratch_lo, 0   //clear  flat scratch lo SGPR 
+  s_mov_b32 flat_scratch_hi, 0   //clear  flat scratch hi SGPR 
+  s_mov_b64 vcc, 0               //clear vcc
+  s_mov_b64 ttmp0, 0             //Clear ttmp0 and ttmp1 
+  s_mov_b64 ttmp2, 0             //Clear ttmp2 and ttmp3 
+  s_mov_b64 ttmp4, 0             //Clear ttmp4 and ttmp5 
+  s_mov_b64 ttmp6, 0             //Clear ttmp6 and ttmp7 
+  s_mov_b64 ttmp8, 0             //Clear ttmp8 and ttmp9 
+  s_mov_b64 ttmp10, 0            //Clear ttmp10 and ttmp11 
+  s_mov_b64 ttmp12, 0            //Clear ttmp12 and ttmp13 
+  s_mov_b64 ttmp14, 0            //Clear ttmp14 and ttmp15 
+s_endpgm
+
+label_0023:
+
+  s_sethalt 1
+
+  s_mov_b32     m0, 0x0000005c   // Loop 96/4=24 times  (loop unrolled for performance)
+  s_nop 0
+label_sgpr_loop1:
+
+  s_movreld_b32     s0, 0
+  s_movreld_b32     s1, 0
+  s_movreld_b32     s2, 0
+  s_movreld_b32     s3, 0
+  s_sub_u32         m0, m0, 4
+  s_cbranch_scc0  label_sgpr_loop1
+ 
+  //clear vcc, flat scratch
+  s_mov_b32 flat_scratch_lo, 0   //clear  flat scratch lo SGPR 
+  s_mov_b32 flat_scratch_hi, 0   //clear  flat scratch hi SGPR 
+  s_mov_b64 vcc, 0xee            //clear vcc
+
+s_endpgm
+end  
+  
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3_cleaner_shader.h b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3_cleaner_shader.h
new file mode 100644
index 000000000000..69aa567c6c1d
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3_cleaner_shader.h
@@ -0,0 +1,64 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/* Define the cleaner shader gfx_9_4_3 */
+static const u32 gfx_9_4_3_cleaner_shader_hex[] = {
+	0xbf068100, 0xbf84003b,
+	0xbf8a0000, 0xb07c0000,
+	0xbe8200ff, 0x00000078,
+	0xbf110802, 0x7e000280,
+	0x7e020280, 0x7e040280,
+	0x7e060280, 0x7e080280,
+	0x7e0a0280, 0x7e0c0280,
+	0x7e0e0280, 0x80828802,
+	0xbe803202, 0xbf84fff5,
+	0xbf9c0000, 0xbe8200ff,
+	0x80000000, 0x86020102,
+	0xbf840011, 0xbefe00c1,
+	0xbeff00c1, 0xd28c0001,
+	0x0001007f, 0xd28d0001,
+	0x0002027e, 0x10020288,
+	0xbe8200bf, 0xbefc00c1,
+	0xd89c2000, 0x00020201,
+	0xd89c6040, 0x00040401,
+	0x320202ff, 0x00000400,
+	0x80828102, 0xbf84fff8,
+	0xbefc00ff, 0x0000005c,
+	0xbf800000, 0xbe802c80,
+	0xbe812c80, 0xbe822c80,
+	0xbe832c80, 0x80fc847c,
+	0xbf84fffa, 0xbee60080,
+	0xbee70080, 0xbeea0180,
+	0xbeec0180, 0xbeee0180,
+	0xbef00180, 0xbef20180,
+	0xbef40180, 0xbef60180,
+	0xbef80180, 0xbefa0180,
+	0xbf810000, 0xbf8d0001,
+	0xbefc00ff, 0x0000005c,
+	0xbf800000, 0xbe802c80,
+	0xbe812c80, 0xbe822c80,
+	0xbe832c80, 0x80fc847c,
+	0xbf84fffa, 0xbee60080,
+	0xbee70080, 0xbeea01ff,
+	0x000000ee, 0xbf810000,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v11_5_0.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v11_5_0.c
new file mode 100644
index 000000000000..f9949fedfbb9
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v11_5_0.c
@@ -0,0 +1,516 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "gfxhub_v11_5_0.h"
+
+#include "gc/gc_11_5_0_offset.h"
+#include "gc/gc_11_5_0_sh_mask.h"
+
+#include "navi10_enum.h"
+#include "soc15_common.h"
+
+#define regGCVM_L2_CNTL3_DEFAULT                0x80100007
+#define regGCVM_L2_CNTL4_DEFAULT                0x000000c1
+#define regGCVM_L2_CNTL5_DEFAULT                0x00003fe0
+
+
+static const char *gfxhub_client_ids[] = {
+	"CB/DB",
+	"Reserved",
+	"GE1",
+	"GE2",
+	"CPF",
+	"CPC",
+	"CPG",
+	"RLC",
+	"TCP",
+	"SQC (inst)",
+	"SQC (data)",
+	"SQG",
+	"Reserved",
+	"SDMA0",
+	"SDMA1",
+	"GCR",
+	"SDMA2",
+	"SDMA3",
+};
+
+static uint32_t gfxhub_v11_5_0_get_invalidate_req(unsigned int vmid,
+					       uint32_t flush_type)
+{
+	u32 req = 0;
+
+	/* invalidate using legacy mode on vmid*/
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ,
+			    PER_VMID_INVALIDATE_REQ, 1 << vmid);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, FLUSH_TYPE, flush_type);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PTES, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE0, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE1, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE2, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L1_PTES, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ,
+			    CLEAR_PROTECTION_FAULT_STATUS_ADDR,	0);
+
+	return req;
+}
+
+static void
+gfxhub_v11_5_0_print_l2_protection_fault_status(struct amdgpu_device *adev,
+					     uint32_t status)
+{
+	u32 cid = REG_GET_FIELD(status,
+				GCVM_L2_PROTECTION_FAULT_STATUS, CID);
+
+	dev_err(adev->dev,
+		"GCVM_L2_PROTECTION_FAULT_STATUS:0x%08X\n",
+		status);
+	dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n",
+		cid >= ARRAY_SIZE(gfxhub_client_ids) ? "unknown" : gfxhub_client_ids[cid],
+		cid);
+	dev_err(adev->dev, "\t MORE_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, MORE_FAULTS));
+	dev_err(adev->dev, "\t WALKER_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, WALKER_ERROR));
+	dev_err(adev->dev, "\t PERMISSION_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, PERMISSION_FAULTS));
+	dev_err(adev->dev, "\t MAPPING_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, MAPPING_ERROR));
+	dev_err(adev->dev, "\t RW: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, RW));
+}
+
+static u64 gfxhub_v11_5_0_get_fb_location(struct amdgpu_device *adev)
+{
+	u64 base = RREG32_SOC15(GC, 0, regGCMC_VM_FB_LOCATION_BASE);
+
+	base &= GCMC_VM_FB_LOCATION_BASE__FB_BASE_MASK;
+	base <<= 24;
+
+	return base;
+}
+
+static u64 gfxhub_v11_5_0_get_mc_fb_offset(struct amdgpu_device *adev)
+{
+	return (u64)RREG32_SOC15(GC, 0, regGCMC_VM_FB_OFFSET) << 24;
+}
+
+static void gfxhub_v11_5_0_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid,
+				uint64_t page_table_base)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+
+	WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+			    hub->ctx_addr_distance * vmid,
+			    lower_32_bits(page_table_base));
+
+	WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+			    hub->ctx_addr_distance * vmid,
+			    upper_32_bits(page_table_base));
+}
+
+static void gfxhub_v11_5_0_init_gart_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t pt_base = amdgpu_gmc_pd_addr(adev->gart.bo);
+
+	gfxhub_v11_5_0_setup_vm_pt_regs(adev, 0, pt_base);
+
+	WREG32_SOC15(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+		     (u32)(adev->gmc.gart_start >> 12));
+	WREG32_SOC15(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+		     (u32)(adev->gmc.gart_start >> 44));
+
+	WREG32_SOC15(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+		     (u32)(adev->gmc.gart_end >> 12));
+	WREG32_SOC15(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+		     (u32)(adev->gmc.gart_end >> 44));
+}
+
+static void gfxhub_v11_5_0_init_system_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t value;
+
+	WREG32_SOC15(GC, 0, regGCMC_VM_AGP_BASE, 0);
+	WREG32_SOC15(GC, 0, regGCMC_VM_AGP_BOT, adev->gmc.agp_start >> 24);
+	WREG32_SOC15(GC, 0, regGCMC_VM_AGP_TOP, adev->gmc.agp_end >> 24);
+
+	/* Program the system aperture low logical page number. */
+	WREG32_SOC15(GC, 0, regGCMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+		     min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
+
+	WREG32_SOC15(GC, 0, regGCMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+		     max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
+
+	/* Set default page address. */
+	value = amdgpu_gmc_vram_mc2pa(adev, adev->mem_scratch.gpu_addr);
+	WREG32_SOC15(GC, 0, regGCMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
+		     (u32)(value >> 12));
+	WREG32_SOC15(GC, 0, regGCMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
+		     (u32)(value >> 44));
+
+	/* Program "protection fault". */
+	WREG32_SOC15(GC, 0, regGCVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
+		     (u32)(adev->dummy_page_addr >> 12));
+	WREG32_SOC15(GC, 0, regGCVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+		     (u32)((u64)adev->dummy_page_addr >> 44));
+
+	WREG32_FIELD15_PREREG(GC, 0, GCVM_L2_PROTECTION_FAULT_CNTL2,
+		       ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+}
+
+static void gfxhub_v11_5_0_init_tlb_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 1);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, ECO_BITS, 0);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    MTYPE, MTYPE_UC); /* UC, uncached */
+
+	WREG32_SOC15(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL, tmp);
+}
+
+static void gfxhub_v11_5_0_init_cache_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(GC, 0, regGCVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, ENABLE_L2_CACHE, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL,
+			    ENABLE_DEFAULT_PAGE_OUT_TO_SYSTEM_MEMORY, 1);
+	/* XXX for emulation, Refer to closed source code.*/
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL,
+			    L2_PDE0_CACHE_TAG_GENERATION_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL, tmp);
+
+	tmp = RREG32_SOC15(GC, 0, regGCVM_L2_CNTL2);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL2, tmp);
+
+	tmp = regGCVM_L2_CNTL3_DEFAULT;
+	if (adev->gmc.translate_further) {
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3, BANK_SELECT, 12);
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+	} else {
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3, BANK_SELECT, 9);
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+	}
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL3, tmp);
+
+	tmp = regGCVM_L2_CNTL4_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL4, tmp);
+
+	tmp = regGCVM_L2_CNTL5_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL5, L2_CACHE_SMALLK_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL5, tmp);
+}
+
+static void gfxhub_v11_5_0_enable_system_domain(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	tmp = RREG32_SOC15(GC, 0, regGCVM_CONTEXT0_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT0_CNTL,
+			    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
+	WREG32_SOC15(GC, 0, regGCVM_CONTEXT0_CNTL, tmp);
+}
+
+static void gfxhub_v11_5_0_disable_identity_aperture(struct amdgpu_device *adev)
+{
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32,
+		     0xFFFFFFFF);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32,
+		     0x0000000F);
+
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32,
+		     0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32,
+		     0);
+
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32, 0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32, 0);
+
+}
+
+static void gfxhub_v11_5_0_setup_vmid_config(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	int i;
+	uint32_t tmp;
+
+	for (i = 0; i <= 14; i++) {
+		tmp = RREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_CNTL, i * hub->ctx_distance);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH,
+				    adev->vm_manager.num_level);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				PAGE_TABLE_BLOCK_SIZE,
+				adev->vm_manager.block_size - 9);
+		/* Send no-retry XNACK on fault to suppress VM fault storm. */
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
+				    !amdgpu_noretry);
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_CNTL,
+				    i * hub->ctx_distance, tmp);
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32,
+				    i * hub->ctx_addr_distance,
+				    lower_32_bits(adev->vm_manager.max_pfn - 1));
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32,
+				    i * hub->ctx_addr_distance,
+				    upper_32_bits(adev->vm_manager.max_pfn - 1));
+	}
+
+	hub->vm_cntx_cntl = tmp;
+}
+
+static void gfxhub_v11_5_0_program_invalidation(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	unsigned i;
+
+	for (i = 0 ; i < 18; ++i) {
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
+				    i * hub->eng_addr_distance, 0xffffffff);
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
+				    i * hub->eng_addr_distance, 0x1f);
+	}
+}
+
+static int gfxhub_v11_5_0_gart_enable(struct amdgpu_device *adev)
+{
+	if (amdgpu_sriov_vf(adev)) {
+		/*
+		 * GCMC_VM_FB_LOCATION_BASE/TOP is NULL for VF, becuase they are
+		 * VF copy registers so vbios post doesn't program them, for
+		 * SRIOV driver need to program them
+		 */
+		WREG32_SOC15(GC, 0, regGCMC_VM_FB_LOCATION_BASE,
+			     adev->gmc.vram_start >> 24);
+		WREG32_SOC15(GC, 0, regGCMC_VM_FB_LOCATION_TOP,
+			     adev->gmc.vram_end >> 24);
+	}
+
+	/* GART Enable. */
+	gfxhub_v11_5_0_init_gart_aperture_regs(adev);
+	gfxhub_v11_5_0_init_system_aperture_regs(adev);
+	gfxhub_v11_5_0_init_tlb_regs(adev);
+	gfxhub_v11_5_0_init_cache_regs(adev);
+
+	gfxhub_v11_5_0_enable_system_domain(adev);
+	gfxhub_v11_5_0_disable_identity_aperture(adev);
+	gfxhub_v11_5_0_setup_vmid_config(adev);
+	gfxhub_v11_5_0_program_invalidation(adev);
+
+	return 0;
+}
+
+static void gfxhub_v11_5_0_gart_disable(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	u32 tmp;
+	u32 i;
+
+	/* Disable all tables */
+	for (i = 0; i < 16; i++)
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT0_CNTL,
+				    i * hub->ctx_distance, 0);
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 0);
+	WREG32_SOC15(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL, tmp);
+
+	/* Setup L2 cache */
+	WREG32_FIELD15_PREREG(GC, 0, GCVM_L2_CNTL, ENABLE_L2_CACHE, 0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL3, 0);
+}
+
+/**
+ * gfxhub_v11_5_0_set_fault_enable_default - update GART/VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+static void gfxhub_v11_5_0_set_fault_enable_default(struct amdgpu_device *adev,
+					  bool value)
+{
+	u32 tmp;
+
+	/* NO halt CP when page fault */
+	tmp = RREG32_SOC15(GC, 0, regCP_DEBUG);
+	tmp = REG_SET_FIELD(tmp, CP_DEBUG, CPG_UTCL1_ERROR_HALT_DISABLE, 1);
+	WREG32_SOC15(GC, 0, regCP_DEBUG, tmp);
+
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	tmp = RREG32_SOC15(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE1_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE2_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT,
+			    value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    NACK_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	if (!value) {
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_NO_RETRY_FAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_RETRY_FAULT, 1);
+	}
+	WREG32_SOC15(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL, tmp);
+}
+
+static const struct amdgpu_vmhub_funcs gfxhub_v11_5_0_vmhub_funcs = {
+	.print_l2_protection_fault_status = gfxhub_v11_5_0_print_l2_protection_fault_status,
+	.get_invalidate_req = gfxhub_v11_5_0_get_invalidate_req,
+};
+
+static void gfxhub_v11_5_0_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+
+	hub->ctx0_ptb_addr_lo32 =
+		SOC15_REG_OFFSET(GC, 0,
+				 regGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32);
+	hub->ctx0_ptb_addr_hi32 =
+		SOC15_REG_OFFSET(GC, 0,
+				 regGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32);
+	hub->vm_inv_eng0_sem =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_INVALIDATE_ENG0_SEM);
+	hub->vm_inv_eng0_req =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_INVALIDATE_ENG0_REQ);
+	hub->vm_inv_eng0_ack =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_INVALIDATE_ENG0_ACK);
+	hub->vm_context0_cntl =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_CONTEXT0_CNTL);
+	hub->vm_l2_pro_fault_status =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_L2_PROTECTION_FAULT_STATUS);
+	hub->vm_l2_pro_fault_cntl =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL);
+
+	hub->ctx_distance = regGCVM_CONTEXT1_CNTL - regGCVM_CONTEXT0_CNTL;
+	hub->ctx_addr_distance = regGCVM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 -
+		regGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32;
+	hub->eng_distance = regGCVM_INVALIDATE_ENG1_REQ -
+		regGCVM_INVALIDATE_ENG0_REQ;
+	hub->eng_addr_distance = regGCVM_INVALIDATE_ENG1_ADDR_RANGE_LO32 -
+		regGCVM_INVALIDATE_ENG0_ADDR_RANGE_LO32;
+
+	hub->vm_cntx_cntl_vm_fault = GCVM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__EXECUTE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK;
+
+	hub->vmhub_funcs = &gfxhub_v11_5_0_vmhub_funcs;
+}
+
+const struct amdgpu_gfxhub_funcs gfxhub_v11_5_0_funcs = {
+	.get_fb_location = gfxhub_v11_5_0_get_fb_location,
+	.get_mc_fb_offset = gfxhub_v11_5_0_get_mc_fb_offset,
+	.setup_vm_pt_regs = gfxhub_v11_5_0_setup_vm_pt_regs,
+	.gart_enable = gfxhub_v11_5_0_gart_enable,
+	.gart_disable = gfxhub_v11_5_0_gart_disable,
+	.set_fault_enable_default = gfxhub_v11_5_0_set_fault_enable_default,
+	.init = gfxhub_v11_5_0_init,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v11_5_0.h b/drivers/gpu/drm/amd/amdgpu/gfxhub_v11_5_0.h
new file mode 100644
index 000000000000..265ab631b3d0
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v11_5_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFXHUB_V11_5_0_H__
+#define __GFXHUB_V11_5_0_H__
+
+extern const struct amdgpu_gfxhub_funcs gfxhub_v11_5_0_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v12_0.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v12_0.c
new file mode 100644
index 000000000000..7609b9cecae8
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v12_0.c
@@ -0,0 +1,521 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "gfxhub_v12_0.h"
+
+#include "gc/gc_12_0_0_offset.h"
+#include "gc/gc_12_0_0_sh_mask.h"
+#include "soc24_enum.h"
+#include "soc15_common.h"
+
+#define regGCVM_L2_CNTL3_DEFAULT	0x80120007
+#define regGCVM_L2_CNTL4_DEFAULT	0x000000c1
+#define regGCVM_L2_CNTL5_DEFAULT	0x00003fe0
+#define regGRBM_GFX_INDEX_DEFAULT	0xe0000000
+
+static const char *gfxhub_client_ids[] = {
+	"CB",
+	"DB",
+	"GE1",
+	"GE2",
+	"CPF",
+	"CPC",
+	"CPG",
+	"RLC",
+	"TCP",
+	"SQC (inst)",
+	"SQC (data)",
+	"SQG/PC/SC",
+	"Reserved",
+	"SDMA0",
+	"SDMA1",
+	"GCR",
+	"Reserved",
+	"Reserved",
+	"WGS",
+	"DSM",
+	"PA"
+};
+
+static uint32_t gfxhub_v12_0_get_invalidate_req(unsigned int vmid,
+					        uint32_t flush_type)
+{
+	u32 req = 0;
+
+	/* invalidate using legacy mode on vmid*/
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ,
+			    PER_VMID_INVALIDATE_REQ, 1 << vmid);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, FLUSH_TYPE, flush_type);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PTES, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE0, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE1, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE2, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L1_PTES, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ,
+			    CLEAR_PROTECTION_FAULT_STATUS_ADDR,	0);
+
+	return req;
+}
+
+static void
+gfxhub_v12_0_print_l2_protection_fault_status(struct amdgpu_device *adev,
+					      uint32_t status)
+{
+	u32 cid = REG_GET_FIELD(status,
+				GCVM_L2_PROTECTION_FAULT_STATUS_LO32, CID);
+
+	dev_err(adev->dev,
+		"GCVM_L2_PROTECTION_FAULT_STATUS:0x%08X\n",
+		status);
+	dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n",
+		cid >= ARRAY_SIZE(gfxhub_client_ids) ? "unknown" : gfxhub_client_ids[cid],
+		cid);
+	dev_err(adev->dev, "\t MORE_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS_LO32, MORE_FAULTS));
+	dev_err(adev->dev, "\t WALKER_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS_LO32, WALKER_ERROR));
+	dev_err(adev->dev, "\t PERMISSION_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS_LO32, PERMISSION_FAULTS));
+	dev_err(adev->dev, "\t MAPPING_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS_LO32, MAPPING_ERROR));
+	dev_err(adev->dev, "\t RW: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS_LO32, RW));
+}
+
+static u64 gfxhub_v12_0_get_fb_location(struct amdgpu_device *adev)
+{
+	u64 base = RREG32_SOC15(GC, 0, regGCMC_VM_FB_LOCATION_BASE);
+
+	base &= GCMC_VM_FB_LOCATION_BASE__FB_BASE_MASK;
+	base <<= 24;
+
+	return base;
+}
+
+static u64 gfxhub_v12_0_get_mc_fb_offset(struct amdgpu_device *adev)
+{
+	return (u64)RREG32_SOC15(GC, 0, regGCMC_VM_FB_OFFSET) << 24;
+}
+
+static void gfxhub_v12_0_setup_vm_pt_regs(struct amdgpu_device *adev,
+					  uint32_t vmid,
+					  uint64_t page_table_base)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+
+	WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+			    hub->ctx_addr_distance * vmid,
+			    lower_32_bits(page_table_base));
+
+	WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+			    hub->ctx_addr_distance * vmid,
+			    upper_32_bits(page_table_base));
+}
+
+static void gfxhub_v12_0_init_gart_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t pt_base = amdgpu_gmc_pd_addr(adev->gart.bo);
+
+	gfxhub_v12_0_setup_vm_pt_regs(adev, 0, pt_base);
+
+	WREG32_SOC15(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+		     (u32)(adev->gmc.gart_start >> 12));
+	WREG32_SOC15(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+		     (u32)(adev->gmc.gart_start >> 44));
+
+	WREG32_SOC15(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+		     (u32)(adev->gmc.gart_end >> 12));
+	WREG32_SOC15(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+		     (u32)(adev->gmc.gart_end >> 44));
+}
+
+static void gfxhub_v12_0_init_system_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t value;
+
+	/* Program the AGP BAR */
+	WREG32_SOC15(GC, 0, regGCMC_VM_AGP_BASE, 0);
+	WREG32_SOC15(GC, 0, regGCMC_VM_AGP_BOT, adev->gmc.agp_start >> 24);
+	WREG32_SOC15(GC, 0, regGCMC_VM_AGP_TOP, adev->gmc.agp_end >> 24);
+
+	/* Program the system aperture low logical page number. */
+	WREG32_SOC15(GC, 0, regGCMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+		     min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
+	WREG32_SOC15(GC, 0, regGCMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+		     max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
+
+	/* Set default page address. */
+	value = adev->mem_scratch.gpu_addr - adev->gmc.vram_start
+		+ adev->vm_manager.vram_base_offset;
+	WREG32_SOC15(GC, 0, regGCMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
+		     (u32)(value >> 12));
+	WREG32_SOC15(GC, 0, regGCMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
+		     (u32)(value >> 44));
+
+	/* Program "protection fault". */
+	WREG32_SOC15(GC, 0, regGCVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
+		     (u32)(adev->dummy_page_addr >> 12));
+	WREG32_SOC15(GC, 0, regGCVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+		     (u32)((u64)adev->dummy_page_addr >> 44));
+
+	WREG32_FIELD15_PREREG(GC, 0, GCVM_L2_PROTECTION_FAULT_CNTL2,
+		       ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+}
+
+
+static void gfxhub_v12_0_init_tlb_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 1);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, ECO_BITS, 0);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    MTYPE, MTYPE_UC); /* UC, uncached */
+
+	WREG32_SOC15(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL, tmp);
+}
+
+static void gfxhub_v12_0_init_cache_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(GC, 0, regGCVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, ENABLE_L2_CACHE, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL,
+			    ENABLE_DEFAULT_PAGE_OUT_TO_SYSTEM_MEMORY, 1);
+	/* XXX for emulation, Refer to closed source code.*/
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL,
+			    L2_PDE0_CACHE_TAG_GENERATION_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL, tmp);
+
+	tmp = RREG32_SOC15(GC, 0, regGCVM_L2_CNTL2);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL2, tmp);
+
+	tmp = regGCVM_L2_CNTL3_DEFAULT;
+	if (adev->gmc.translate_further) {
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3, BANK_SELECT, 12);
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+	} else {
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3, BANK_SELECT, 9);
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+	}
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL3, tmp);
+
+	tmp = regGCVM_L2_CNTL4_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL4, tmp);
+
+	tmp = regGCVM_L2_CNTL5_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL5, L2_CACHE_SMALLK_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL5, tmp);
+}
+
+static void gfxhub_v12_0_enable_system_domain(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	tmp = RREG32_SOC15(GC, 0, regGCVM_CONTEXT0_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT0_CNTL,
+			    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
+	WREG32_SOC15(GC, 0, regGCVM_CONTEXT0_CNTL, tmp);
+}
+
+static void gfxhub_v12_0_disable_identity_aperture(struct amdgpu_device *adev)
+{
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32,
+		     0xFFFFFFFF);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32,
+		     0x0000000F);
+
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32,
+		     0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32,
+		     0);
+
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32, 0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32, 0);
+
+}
+
+static void gfxhub_v12_0_setup_vmid_config(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	int i;
+	uint32_t tmp;
+
+	for (i = 0; i <= 14; i++) {
+		tmp = RREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_CNTL, i);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH,
+				    adev->vm_manager.num_level);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				PAGE_TABLE_BLOCK_SIZE,
+				adev->vm_manager.block_size - 9);
+		/* Send no-retry XNACK on fault to suppress VM fault storm. */
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
+				    !amdgpu_noretry);
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_CNTL,
+				    i * hub->ctx_distance, tmp);
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32,
+				    i * hub->ctx_addr_distance,
+				    lower_32_bits(adev->vm_manager.max_pfn - 1));
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32,
+				    i * hub->ctx_addr_distance,
+				    upper_32_bits(adev->vm_manager.max_pfn - 1));
+	}
+
+	hub->vm_cntx_cntl = tmp;
+}
+
+static void gfxhub_v12_0_program_invalidation(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	unsigned i;
+
+	for (i = 0 ; i < 18; ++i) {
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
+				    i * hub->eng_addr_distance, 0xffffffff);
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
+				    i * hub->eng_addr_distance, 0x1f);
+	}
+}
+
+static int gfxhub_v12_0_gart_enable(struct amdgpu_device *adev)
+{
+	if (amdgpu_sriov_vf(adev)) {
+		/*
+		 * GCMC_VM_FB_LOCATION_BASE/TOP is NULL for VF, becuase they are
+		 * VF copy registers so vbios post doesn't program them, for
+		 * SRIOV driver need to program them
+		 */
+		WREG32_SOC15(GC, 0, regGCMC_VM_FB_LOCATION_BASE,
+			     adev->gmc.vram_start >> 24);
+		WREG32_SOC15(GC, 0, regGCMC_VM_FB_LOCATION_TOP,
+			     adev->gmc.vram_end >> 24);
+	}
+
+	/* GART Enable. */
+	gfxhub_v12_0_init_gart_aperture_regs(adev);
+	gfxhub_v12_0_init_system_aperture_regs(adev);
+	gfxhub_v12_0_init_tlb_regs(adev);
+	gfxhub_v12_0_init_cache_regs(adev);
+
+	gfxhub_v12_0_enable_system_domain(adev);
+	gfxhub_v12_0_disable_identity_aperture(adev);
+	gfxhub_v12_0_setup_vmid_config(adev);
+	gfxhub_v12_0_program_invalidation(adev);
+
+	return 0;
+}
+
+static void gfxhub_v12_0_gart_disable(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	u32 tmp;
+	u32 i;
+
+	/* Disable all tables */
+	for (i = 0; i < 16; i++)
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT0_CNTL,
+				    i * hub->ctx_distance, 0);
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 0);
+	WREG32_SOC15(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL, tmp);
+
+	/* Setup L2 cache */
+	WREG32_FIELD15_PREREG(GC, 0, GCVM_L2_CNTL, ENABLE_L2_CACHE, 0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL3, 0);
+}
+
+/**
+ * gfxhub_v12_0_set_fault_enable_default - update GART/VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+static void gfxhub_v12_0_set_fault_enable_default(struct amdgpu_device *adev,
+						  bool value)
+{
+	u32 tmp;
+
+	/* NO halt CP when page fault */
+	tmp = RREG32_SOC15(GC, 0, regCP_DEBUG);
+	tmp = REG_SET_FIELD(tmp, CP_DEBUG, CPG_UTCL1_ERROR_HALT_DISABLE, 1);
+	WREG32_SOC15(GC, 0, regCP_DEBUG, tmp);
+
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	tmp = RREG32_SOC15(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE1_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE2_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT,
+			    value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    NACK_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	if (!value) {
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_NO_RETRY_FAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_RETRY_FAULT, 1);
+	}
+	WREG32_SOC15(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL, tmp);
+}
+
+static const struct amdgpu_vmhub_funcs gfxhub_v12_0_vmhub_funcs = {
+	.print_l2_protection_fault_status = gfxhub_v12_0_print_l2_protection_fault_status,
+	.get_invalidate_req = gfxhub_v12_0_get_invalidate_req,
+};
+
+static void gfxhub_v12_0_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+
+	hub->ctx0_ptb_addr_lo32 =
+		SOC15_REG_OFFSET(GC, 0,
+				 regGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32);
+	hub->ctx0_ptb_addr_hi32 =
+		SOC15_REG_OFFSET(GC, 0,
+				 regGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32);
+	hub->vm_inv_eng0_sem =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_INVALIDATE_ENG0_SEM);
+	hub->vm_inv_eng0_req =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_INVALIDATE_ENG0_REQ);
+	hub->vm_inv_eng0_ack =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_INVALIDATE_ENG0_ACK);
+	hub->vm_context0_cntl =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_CONTEXT0_CNTL);
+	hub->vm_l2_pro_fault_status =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_L2_PROTECTION_FAULT_STATUS_LO32);
+	hub->vm_l2_pro_fault_cntl =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL);
+
+	hub->ctx_distance = regGCVM_CONTEXT1_CNTL - regGCVM_CONTEXT0_CNTL;
+	hub->ctx_addr_distance = regGCVM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 -
+		regGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32;
+	hub->eng_distance = regGCVM_INVALIDATE_ENG1_REQ -
+		regGCVM_INVALIDATE_ENG0_REQ;
+	hub->eng_addr_distance = regGCVM_INVALIDATE_ENG1_ADDR_RANGE_LO32 -
+		regGCVM_INVALIDATE_ENG0_ADDR_RANGE_LO32;
+
+	hub->vm_cntx_cntl_vm_fault = GCVM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__EXECUTE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK;
+
+	hub->vmhub_funcs = &gfxhub_v12_0_vmhub_funcs;
+}
+
+const struct amdgpu_gfxhub_funcs gfxhub_v12_0_funcs = {
+	.get_fb_location = gfxhub_v12_0_get_fb_location,
+	.get_mc_fb_offset = gfxhub_v12_0_get_mc_fb_offset,
+	.setup_vm_pt_regs = gfxhub_v12_0_setup_vm_pt_regs,
+	.gart_enable = gfxhub_v12_0_gart_enable,
+	.gart_disable = gfxhub_v12_0_gart_disable,
+	.set_fault_enable_default = gfxhub_v12_0_set_fault_enable_default,
+	.init = gfxhub_v12_0_init,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v12_0.h b/drivers/gpu/drm/amd/amdgpu/gfxhub_v12_0.h
new file mode 100644
index 000000000000..f1258265f802
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v12_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFXHUB_V12_0_H__
+#define __GFXHUB_V12_0_H__
+
+extern const struct amdgpu_gfxhub_funcs gfxhub_v12_0_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c
index 6c8040e616c4..a7bfc9f41d0e 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c
@@ -22,83 +22,133 @@
  */
 #include "amdgpu.h"
 #include "gfxhub_v1_0.h"
+#include "gfxhub_v1_1.h"
 
-#include "vega10/soc15ip.h"
-#include "vega10/GC/gc_9_0_offset.h"
-#include "vega10/GC/gc_9_0_sh_mask.h"
-#include "vega10/GC/gc_9_0_default.h"
-#include "vega10/vega10_enum.h"
+#include "gc/gc_9_0_offset.h"
+#include "gc/gc_9_0_sh_mask.h"
+#include "gc/gc_9_0_default.h"
+#include "vega10_enum.h"
 
 #include "soc15_common.h"
 
-u64 gfxhub_v1_0_get_mc_fb_offset(struct amdgpu_device *adev)
+static u64 gfxhub_v1_0_get_mc_fb_offset(struct amdgpu_device *adev)
 {
 	return (u64)RREG32_SOC15(GC, 0, mmMC_VM_FB_OFFSET) << 24;
 }
 
-static void gfxhub_v1_0_init_gart_pt_regs(struct amdgpu_device *adev)
+static void gfxhub_v1_0_setup_vm_pt_regs(struct amdgpu_device *adev,
+					 uint32_t vmid,
+					 uint64_t page_table_base)
 {
-	uint64_t value;
-
-	BUG_ON(adev->gart.table_addr & (~0x0000FFFFFFFFF000ULL));
-	value = adev->gart.table_addr - adev->mc.vram_start
-		+ adev->vm_manager.vram_base_offset;
-	value &= 0x0000FFFFFFFFF000ULL;
-	value |= 0x1; /*valid bit*/
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
 
-	WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
-		     lower_32_bits(value));
+	WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+			    hub->ctx_addr_distance * vmid,
+			    lower_32_bits(page_table_base));
 
-	WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
-		     upper_32_bits(value));
+	WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+			    hub->ctx_addr_distance * vmid,
+			    upper_32_bits(page_table_base));
 }
 
 static void gfxhub_v1_0_init_gart_aperture_regs(struct amdgpu_device *adev)
 {
-	gfxhub_v1_0_init_gart_pt_regs(adev);
-
-	WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
-		     (u32)(adev->mc.gart_start >> 12));
-	WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
-		     (u32)(adev->mc.gart_start >> 44));
-
-	WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
-		     (u32)(adev->mc.gart_end >> 12));
-	WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
-		     (u32)(adev->mc.gart_end >> 44));
+	uint64_t pt_base;
+
+	if (adev->gmc.pdb0_bo)
+		pt_base = amdgpu_gmc_pd_addr(adev->gmc.pdb0_bo);
+	else
+		pt_base = amdgpu_gmc_pd_addr(adev->gart.bo);
+
+	gfxhub_v1_0_setup_vm_pt_regs(adev, 0, pt_base);
+
+	/* If use GART for FB translation, vmid0 page table covers both
+	 * vram and system memory (gart)
+	 */
+	if (adev->gmc.pdb0_bo) {
+		WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+				(u32)(adev->gmc.fb_start >> 12));
+		WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+				(u32)(adev->gmc.fb_start >> 44));
+
+		WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+				(u32)(adev->gmc.gart_end >> 12));
+		WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+				(u32)(adev->gmc.gart_end >> 44));
+	} else {
+		WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+				(u32)(adev->gmc.gart_start >> 12));
+		WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+				(u32)(adev->gmc.gart_start >> 44));
+
+		WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+				(u32)(adev->gmc.gart_end >> 12));
+		WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+				(u32)(adev->gmc.gart_end >> 44));
+	}
 }
 
 static void gfxhub_v1_0_init_system_aperture_regs(struct amdgpu_device *adev)
 {
 	uint64_t value;
 
-	/* Disable AGP. */
-	WREG32_SOC15(GC, 0, mmMC_VM_AGP_BASE, 0);
-	WREG32_SOC15(GC, 0, mmMC_VM_AGP_TOP, 0);
-	WREG32_SOC15(GC, 0, mmMC_VM_AGP_BOT, 0xFFFFFFFF);
-
-	/* Program the system aperture low logical page number. */
-	WREG32_SOC15(GC, 0, mmMC_VM_SYSTEM_APERTURE_LOW_ADDR,
-		     adev->mc.vram_start >> 18);
-	WREG32_SOC15(GC, 0, mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
-		     adev->mc.vram_end >> 18);
-
-	/* Set default page address. */
-	value = adev->vram_scratch.gpu_addr - adev->mc.vram_start
-		+ adev->vm_manager.vram_base_offset;
-	WREG32_SOC15(GC, 0, mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
-		     (u32)(value >> 12));
-	WREG32_SOC15(GC, 0, mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
-		     (u32)(value >> 44));
-
-	/* Program "protection fault". */
-	WREG32_SOC15(GC, 0, mmVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
-		     (u32)(adev->dummy_page.addr >> 12));
-	WREG32_SOC15(GC, 0, mmVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
-		     (u32)((u64)adev->dummy_page.addr >> 44));
-
-	WREG32_FIELD15(GC, 0, VM_L2_PROTECTION_FAULT_CNTL2,
-		       ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+	if (!amdgpu_sriov_vf(adev) || adev->asic_type <= CHIP_VEGA10) {
+		/* Program the AGP BAR */
+		WREG32_SOC15_RLC(GC, 0, mmMC_VM_AGP_BASE, 0);
+		WREG32_SOC15_RLC(GC, 0, mmMC_VM_AGP_BOT, adev->gmc.agp_start >> 24);
+		WREG32_SOC15_RLC(GC, 0, mmMC_VM_AGP_TOP, adev->gmc.agp_end >> 24);
+
+		/* Program the system aperture low logical page number. */
+		WREG32_SOC15_RLC(GC, 0, mmMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+			min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
+
+		if (adev->apu_flags & (AMD_APU_IS_RAVEN2 |
+				       AMD_APU_IS_RENOIR |
+				       AMD_APU_IS_GREEN_SARDINE))
+		       /*
+			* Raven2 has a HW issue that it is unable to use the
+			* vram which is out of MC_VM_SYSTEM_APERTURE_HIGH_ADDR.
+			* So here is the workaround that increase system
+			* aperture high address (add 1) to get rid of the VM
+			* fault and hardware hang.
+			*/
+			WREG32_SOC15_RLC(GC, 0,
+					 mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+					 max((adev->gmc.fb_end >> 18) + 0x1,
+					     adev->gmc.agp_end >> 18));
+		else
+			WREG32_SOC15_RLC(
+				GC, 0, mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+				max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
+
+		/* Set default page address. */
+		value = amdgpu_gmc_vram_mc2pa(adev, adev->mem_scratch.gpu_addr);
+		WREG32_SOC15(GC, 0, mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
+			     (u32)(value >> 12));
+		WREG32_SOC15(GC, 0, mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
+			     (u32)(value >> 44));
+
+		/* Program "protection fault". */
+		WREG32_SOC15(GC, 0, mmVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
+			     (u32)(adev->dummy_page_addr >> 12));
+		WREG32_SOC15(GC, 0, mmVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+			     (u32)((u64)adev->dummy_page_addr >> 44));
+
+		WREG32_FIELD15(GC, 0, VM_L2_PROTECTION_FAULT_CNTL2,
+			       ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+	}
+
+	/* In the case squeezing vram into GART aperture, we don't use
+	 * FB aperture and AGP aperture. Disable them.
+	 */
+	if (adev->gmc.pdb0_bo) {
+		WREG32_SOC15(GC, 0, mmMC_VM_FB_LOCATION_TOP, 0);
+		WREG32_SOC15(GC, 0, mmMC_VM_FB_LOCATION_BASE, 0x00FFFFFF);
+		WREG32_SOC15(GC, 0, mmMC_VM_AGP_TOP, 0);
+		WREG32_SOC15(GC, 0, mmMC_VM_AGP_BOT, 0xFFFFFF);
+		WREG32_SOC15(GC, 0, mmMC_VM_SYSTEM_APERTURE_LOW_ADDR, 0x3FFFFFFF);
+		WREG32_SOC15(GC, 0, mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR, 0);
+	}
 }
 
 static void gfxhub_v1_0_init_tlb_regs(struct amdgpu_device *adev)
@@ -114,12 +164,11 @@ static void gfxhub_v1_0_init_tlb_regs(struct amdgpu_device *adev)
 			    ENABLE_ADVANCED_DRIVER_MODEL, 1);
 	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
 			    SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
-	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ECO_BITS, 0);
 	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
 			    MTYPE, MTYPE_UC);/* XXX for emulation. */
 	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ATC_EN, 1);
 
-	WREG32_SOC15(GC, 0, mmMC_VM_MX_L1_TLB_CNTL, tmp);
+	WREG32_SOC15_RLC(GC, 0, mmMC_VM_MX_L1_TLB_CNTL, tmp);
 }
 
 static void gfxhub_v1_0_init_cache_regs(struct amdgpu_device *adev)
@@ -133,25 +182,37 @@ static void gfxhub_v1_0_init_cache_regs(struct amdgpu_device *adev)
 	/* XXX for emulation, Refer to closed source code.*/
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, L2_PDE0_CACHE_TAG_GENERATION_MODE,
 			    0);
-	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 0);
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0);
-	WREG32_SOC15(GC, 0, mmVM_L2_CNTL, tmp);
+	WREG32_SOC15_RLC(GC, 0, mmVM_L2_CNTL, tmp);
 
 	tmp = RREG32_SOC15(GC, 0, mmVM_L2_CNTL2);
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
-	WREG32_SOC15(GC, 0, mmVM_L2_CNTL2, tmp);
+	WREG32_SOC15_RLC(GC, 0, mmVM_L2_CNTL2, tmp);
 
 	tmp = mmVM_L2_CNTL3_DEFAULT;
-	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 9);
-	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
-	WREG32_SOC15(GC, 0, mmVM_L2_CNTL3, tmp);
+	if (adev->gmc.translate_further) {
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 12);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+	} else {
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 9);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+	}
+	WREG32_SOC15_RLC(GC, 0, mmVM_L2_CNTL3, tmp);
 
 	tmp = mmVM_L2_CNTL4_DEFAULT;
-	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
-	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
-	WREG32_SOC15(GC, 0, mmVM_L2_CNTL4, tmp);
+	if (adev->gmc.xgmi.connected_to_cpu) {
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 1);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 1);
+	} else {
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
+	}
+	WREG32_SOC15_RLC(GC, 0, mmVM_L2_CNTL4, tmp);
 }
 
 static void gfxhub_v1_0_enable_system_domain(struct amdgpu_device *adev)
@@ -160,7 +221,12 @@ static void gfxhub_v1_0_enable_system_domain(struct amdgpu_device *adev)
 
 	tmp = RREG32_SOC15(GC, 0, mmVM_CONTEXT0_CNTL);
 	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
-	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH,
+			adev->gmc.vmid0_page_table_depth);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, PAGE_TABLE_BLOCK_SIZE,
+			adev->gmc.vmid0_page_table_block_size);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL,
+			    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
 	WREG32_SOC15(GC, 0, mmVM_CONTEXT0_CNTL, tmp);
 }
 
@@ -183,92 +249,109 @@ static void gfxhub_v1_0_disable_identity_aperture(struct amdgpu_device *adev)
 
 static void gfxhub_v1_0_setup_vmid_config(struct amdgpu_device *adev)
 {
-	int i;
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	unsigned int num_level, block_size;
 	uint32_t tmp;
+	int i;
+
+	num_level = adev->vm_manager.num_level;
+	block_size = adev->vm_manager.block_size;
+	if (adev->gmc.translate_further)
+		num_level -= 1;
+	else
+		block_size -= 9;
 
 	for (i = 0; i <= 14; i++) {
-		tmp = RREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_CNTL, i);
+		tmp = RREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_CNTL, i * hub->ctx_distance);
 		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
 		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH,
-				    adev->vm_manager.num_level);
+				    num_level);
 		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
-				RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+				    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
 		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
-				DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+				    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    1);
 		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
-				PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+				    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
 		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
-				VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+				    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
 		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
-				READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+				    READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
 		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
-				WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+				    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
 		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
-				EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+				    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
 		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
-				PAGE_TABLE_BLOCK_SIZE,
-				adev->vm_manager.block_size - 9);
-		/* Send no-retry XNACK on fault to suppress VM fault storm. */
+				    PAGE_TABLE_BLOCK_SIZE,
+				    block_size);
+		/* Send no-retry XNACK on fault to suppress VM fault storm.
+		 * On Aldebaran, XNACK can be enabled in the SQ per-process.
+		 * Retry faults need to be enabled for that to work.
+		 */
 		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
-				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
-		WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_CNTL, i, tmp);
-		WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32, i*2, 0);
-		WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32, i*2, 0);
-		WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32,  i*2,
-			lower_32_bits(adev->vm_manager.max_pfn - 1));
-		WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32, i*2,
-			upper_32_bits(adev->vm_manager.max_pfn - 1));
+				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
+				    !adev->gmc.noretry ||
+				    adev->asic_type == CHIP_ALDEBARAN);
+		WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_CNTL,
+				    i * hub->ctx_distance, tmp);
+		WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32,
+				    i * hub->ctx_addr_distance,
+				    lower_32_bits(adev->vm_manager.max_pfn - 1));
+		WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32,
+				    i * hub->ctx_addr_distance,
+				    upper_32_bits(adev->vm_manager.max_pfn - 1));
 	}
 }
 
 static void gfxhub_v1_0_program_invalidation(struct amdgpu_device *adev)
 {
-	unsigned i;
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	unsigned int i;
 
 	for (i = 0 ; i < 18; ++i) {
 		WREG32_SOC15_OFFSET(GC, 0, mmVM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
-				    2 * i, 0xffffffff);
+				    i * hub->eng_addr_distance, 0xffffffff);
 		WREG32_SOC15_OFFSET(GC, 0, mmVM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
-				    2 * i, 0x1f);
+				    i * hub->eng_addr_distance, 0x1f);
 	}
 }
 
-int gfxhub_v1_0_gart_enable(struct amdgpu_device *adev)
+static int gfxhub_v1_0_gart_enable(struct amdgpu_device *adev)
 {
-	if (amdgpu_sriov_vf(adev)) {
-		/*
-		 * MC_VM_FB_LOCATION_BASE/TOP is NULL for VF, becuase they are
-		 * VF copy registers so vbios post doesn't program them, for
-		 * SRIOV driver need to program them
-		 */
-		WREG32_SOC15(GC, 0, mmMC_VM_FB_LOCATION_BASE,
-			     adev->mc.vram_start >> 24);
-		WREG32_SOC15(GC, 0, mmMC_VM_FB_LOCATION_TOP,
-			     adev->mc.vram_end >> 24);
-	}
-
 	/* GART Enable. */
 	gfxhub_v1_0_init_gart_aperture_regs(adev);
 	gfxhub_v1_0_init_system_aperture_regs(adev);
 	gfxhub_v1_0_init_tlb_regs(adev);
-	gfxhub_v1_0_init_cache_regs(adev);
+	if (!amdgpu_sriov_vf(adev))
+		gfxhub_v1_0_init_cache_regs(adev);
 
 	gfxhub_v1_0_enable_system_domain(adev);
-	gfxhub_v1_0_disable_identity_aperture(adev);
+	if (!amdgpu_sriov_vf(adev))
+		gfxhub_v1_0_disable_identity_aperture(adev);
 	gfxhub_v1_0_setup_vmid_config(adev);
 	gfxhub_v1_0_program_invalidation(adev);
 
 	return 0;
 }
 
-void gfxhub_v1_0_gart_disable(struct amdgpu_device *adev)
+static void gfxhub_v1_0_gart_disable(struct amdgpu_device *adev)
 {
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
 	u32 tmp;
 	u32 i;
 
 	/* Disable all tables */
 	for (i = 0; i < 16; i++)
-		WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT0_CNTL, i, 0);
+		WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT0_CNTL,
+				    i * hub->ctx_distance, 0);
+
+	if (amdgpu_sriov_vf(adev))
+		/* Avoid write to GMC registers */
+		return;
 
 	/* Setup TLB control */
 	tmp = RREG32_SOC15(GC, 0, mmMC_VM_MX_L1_TLB_CNTL);
@@ -277,7 +360,7 @@ void gfxhub_v1_0_gart_disable(struct amdgpu_device *adev)
 				MC_VM_MX_L1_TLB_CNTL,
 				ENABLE_ADVANCED_DRIVER_MODEL,
 				0);
-	WREG32_SOC15(GC, 0, mmMC_VM_MX_L1_TLB_CNTL, tmp);
+	WREG32_SOC15_RLC(GC, 0, mmMC_VM_MX_L1_TLB_CNTL, tmp);
 
 	/* Setup L2 cache */
 	WREG32_FIELD15(GC, 0, VM_L2_CNTL, ENABLE_L2_CACHE, 0);
@@ -290,10 +373,11 @@ void gfxhub_v1_0_gart_disable(struct amdgpu_device *adev)
  * @adev: amdgpu_device pointer
  * @value: true redirects VM faults to the default page
  */
-void gfxhub_v1_0_set_fault_enable_default(struct amdgpu_device *adev,
-					  bool value)
+static void gfxhub_v1_0_set_fault_enable_default(struct amdgpu_device *adev,
+						 bool value)
 {
 	u32 tmp;
+
 	tmp = RREG32_SOC15(GC, 0, mmVM_L2_PROTECTION_FAULT_CNTL);
 	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
 			RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
@@ -319,12 +403,18 @@ void gfxhub_v1_0_set_fault_enable_default(struct amdgpu_device *adev,
 			WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
 	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
 			EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	if (!value) {
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_NO_RETRY_FAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_RETRY_FAULT, 1);
+	}
 	WREG32_SOC15(GC, 0, mmVM_L2_PROTECTION_FAULT_CNTL, tmp);
 }
 
-void gfxhub_v1_0_init(struct amdgpu_device *adev)
+static void gfxhub_v1_0_init(struct amdgpu_device *adev)
 {
-	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB];
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
 
 	hub->ctx0_ptb_addr_lo32 =
 		SOC15_REG_OFFSET(GC, 0,
@@ -332,6 +422,8 @@ void gfxhub_v1_0_init(struct amdgpu_device *adev)
 	hub->ctx0_ptb_addr_hi32 =
 		SOC15_REG_OFFSET(GC, 0,
 				 mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32);
+	hub->vm_inv_eng0_sem =
+		SOC15_REG_OFFSET(GC, 0, mmVM_INVALIDATE_ENG0_SEM);
 	hub->vm_inv_eng0_req =
 		SOC15_REG_OFFSET(GC, 0, mmVM_INVALIDATE_ENG0_REQ);
 	hub->vm_inv_eng0_ack =
@@ -342,4 +434,21 @@ void gfxhub_v1_0_init(struct amdgpu_device *adev)
 		SOC15_REG_OFFSET(GC, 0, mmVM_L2_PROTECTION_FAULT_STATUS);
 	hub->vm_l2_pro_fault_cntl =
 		SOC15_REG_OFFSET(GC, 0, mmVM_L2_PROTECTION_FAULT_CNTL);
+
+	hub->ctx_distance = mmVM_CONTEXT1_CNTL - mmVM_CONTEXT0_CNTL;
+	hub->ctx_addr_distance = mmVM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 -
+		mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32;
+	hub->eng_distance = mmVM_INVALIDATE_ENG1_REQ - mmVM_INVALIDATE_ENG0_REQ;
+	hub->eng_addr_distance = mmVM_INVALIDATE_ENG1_ADDR_RANGE_LO32 -
+		mmVM_INVALIDATE_ENG0_ADDR_RANGE_LO32;
 }
+
+const struct amdgpu_gfxhub_funcs gfxhub_v1_0_funcs = {
+	.get_mc_fb_offset = gfxhub_v1_0_get_mc_fb_offset,
+	.setup_vm_pt_regs = gfxhub_v1_0_setup_vm_pt_regs,
+	.gart_enable = gfxhub_v1_0_gart_enable,
+	.gart_disable = gfxhub_v1_0_gart_disable,
+	.set_fault_enable_default = gfxhub_v1_0_set_fault_enable_default,
+	.init = gfxhub_v1_0_init,
+	.get_xgmi_info = gfxhub_v1_1_get_xgmi_info,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.h b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.h
index 206e29cad753..3174bc5766fd 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.h
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.h
@@ -24,11 +24,6 @@
 #ifndef __GFXHUB_V1_0_H__
 #define __GFXHUB_V1_0_H__
 
-int gfxhub_v1_0_gart_enable(struct amdgpu_device *adev);
-void gfxhub_v1_0_gart_disable(struct amdgpu_device *adev);
-void gfxhub_v1_0_set_fault_enable_default(struct amdgpu_device *adev,
-					  bool value);
-void gfxhub_v1_0_init(struct amdgpu_device *adev);
-u64 gfxhub_v1_0_get_mc_fb_offset(struct amdgpu_device *adev);
+extern const struct amdgpu_gfxhub_funcs gfxhub_v1_0_funcs;
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.c
new file mode 100644
index 000000000000..90f0aefbdb39
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.c
@@ -0,0 +1,111 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "gfxhub_v1_1.h"
+
+#include "gc/gc_9_2_1_offset.h"
+#include "gc/gc_9_2_1_sh_mask.h"
+
+#include "soc15_common.h"
+
+#define mmMC_VM_XGMI_LFB_CNTL_ALDE			0x0978
+#define mmMC_VM_XGMI_LFB_CNTL_ALDE_BASE_IDX		0
+#define mmMC_VM_XGMI_LFB_SIZE_ALDE			0x0979
+#define mmMC_VM_XGMI_LFB_SIZE_ALDE_BASE_IDX		0
+//MC_VM_XGMI_LFB_CNTL
+#define MC_VM_XGMI_LFB_CNTL_ALDE__PF_LFB_REGION__SHIFT	0x0
+#define MC_VM_XGMI_LFB_CNTL_ALDE__PF_MAX_REGION__SHIFT	0x4
+#define MC_VM_XGMI_LFB_CNTL_ALDE__PF_LFB_REGION_MASK	0x0000000FL
+#define MC_VM_XGMI_LFB_CNTL_ALDE__PF_MAX_REGION_MASK	0x000000F0L
+//MC_VM_XGMI_LFB_SIZE
+#define MC_VM_XGMI_LFB_SIZE_ALDE__PF_LFB_SIZE__SHIFT	0x0
+#define MC_VM_XGMI_LFB_SIZE_ALDE__PF_LFB_SIZE_MASK	0x0001FFFFL
+
+int gfxhub_v1_1_get_xgmi_info(struct amdgpu_device *adev)
+{
+	u32 max_num_physical_nodes;
+	u32 max_physical_node_id;
+	u32 xgmi_lfb_cntl;
+	u32 max_region;
+	u64 seg_size;
+
+	if (adev->asic_type == CHIP_ALDEBARAN) {
+		xgmi_lfb_cntl = RREG32_SOC15(GC, 0, mmMC_VM_XGMI_LFB_CNTL_ALDE);
+		seg_size = REG_GET_FIELD(
+			RREG32_SOC15(GC, 0, mmMC_VM_XGMI_LFB_SIZE_ALDE),
+			MC_VM_XGMI_LFB_SIZE, PF_LFB_SIZE) << 24;
+		max_region =
+			REG_GET_FIELD(xgmi_lfb_cntl, MC_VM_XGMI_LFB_CNTL_ALDE, PF_MAX_REGION);
+	} else {
+		xgmi_lfb_cntl = RREG32_SOC15(GC, 0, mmMC_VM_XGMI_LFB_CNTL);
+		seg_size = REG_GET_FIELD(
+			RREG32_SOC15(GC, 0, mmMC_VM_XGMI_LFB_SIZE),
+			MC_VM_XGMI_LFB_SIZE, PF_LFB_SIZE) << 24;
+		max_region =
+			REG_GET_FIELD(xgmi_lfb_cntl, MC_VM_XGMI_LFB_CNTL, PF_MAX_REGION);
+	}
+
+
+
+	switch (adev->asic_type) {
+	case CHIP_VEGA20:
+		max_num_physical_nodes   = 4;
+		max_physical_node_id     = 3;
+		break;
+	case CHIP_ARCTURUS:
+		max_num_physical_nodes   = 8;
+		max_physical_node_id     = 7;
+		break;
+	case CHIP_ALDEBARAN:
+		max_num_physical_nodes   = 16;
+		max_physical_node_id     = 15;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	/* PF_MAX_REGION=0 means xgmi is disabled */
+	if (max_region || adev->gmc.xgmi.connected_to_cpu) {
+		adev->gmc.xgmi.num_physical_nodes = max_region + 1;
+
+		if (adev->gmc.xgmi.num_physical_nodes > max_num_physical_nodes)
+			return -EINVAL;
+
+		if (adev->asic_type == CHIP_ALDEBARAN) {
+			adev->gmc.xgmi.physical_node_id =
+				REG_GET_FIELD(xgmi_lfb_cntl, MC_VM_XGMI_LFB_CNTL_ALDE,
+						PF_LFB_REGION);
+		} else {
+			adev->gmc.xgmi.physical_node_id =
+				REG_GET_FIELD(xgmi_lfb_cntl, MC_VM_XGMI_LFB_CNTL,
+						PF_LFB_REGION);
+		}
+
+		if (adev->gmc.xgmi.physical_node_id > max_physical_node_id)
+			return -EINVAL;
+
+		adev->gmc.xgmi.node_segment_size = seg_size;
+	}
+
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.h b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.h
new file mode 100644
index 000000000000..d753cf28a0a6
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFXHUB_V1_1_H__
+#define __GFXHUB_V1_1_H__
+
+int gfxhub_v1_1_get_xgmi_info(struct amdgpu_device *adev);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.c
new file mode 100644
index 000000000000..6c03bf9f1ae8
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.c
@@ -0,0 +1,674 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "amdgpu_xcp.h"
+#include "gfxhub_v1_2.h"
+#include "gfxhub_v1_1.h"
+
+#include "gc/gc_9_4_3_offset.h"
+#include "gc/gc_9_4_3_sh_mask.h"
+#include "vega10_enum.h"
+
+#include "soc15_common.h"
+
+#define regVM_L2_CNTL3_DEFAULT	0x80100007
+#define regVM_L2_CNTL4_DEFAULT	0x000000c1
+
+static u64 gfxhub_v1_2_get_mc_fb_offset(struct amdgpu_device *adev)
+{
+	return (u64)RREG32_SOC15(GC, GET_INST(GC, 0), regMC_VM_FB_OFFSET) << 24;
+}
+
+static void gfxhub_v1_2_xcc_setup_vm_pt_regs(struct amdgpu_device *adev,
+					     uint32_t vmid,
+					     uint64_t page_table_base,
+					     uint32_t xcc_mask)
+{
+	struct amdgpu_vmhub *hub;
+	int i;
+
+	for_each_inst(i, xcc_mask) {
+		hub = &adev->vmhub[AMDGPU_GFXHUB(i)];
+		WREG32_SOC15_OFFSET(GC, GET_INST(GC, i),
+				    regVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+				    hub->ctx_addr_distance * vmid,
+				    lower_32_bits(page_table_base));
+
+		WREG32_SOC15_OFFSET(GC, GET_INST(GC, i),
+				    regVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+				    hub->ctx_addr_distance * vmid,
+				    upper_32_bits(page_table_base));
+	}
+}
+
+static void gfxhub_v1_2_setup_vm_pt_regs(struct amdgpu_device *adev,
+					 uint32_t vmid,
+					 uint64_t page_table_base)
+{
+	uint32_t xcc_mask;
+
+	xcc_mask = GENMASK(NUM_XCC(adev->gfx.xcc_mask) - 1, 0);
+	gfxhub_v1_2_xcc_setup_vm_pt_regs(adev, vmid, page_table_base, xcc_mask);
+}
+
+static void gfxhub_v1_2_xcc_init_gart_aperture_regs(struct amdgpu_device *adev,
+						    uint32_t xcc_mask)
+{
+	uint64_t gart_start = amdgpu_virt_xgmi_migrate_enabled(adev) ?
+			adev->gmc.vram_start : adev->gmc.fb_start;
+	uint64_t pt_base;
+	int i;
+
+	if (adev->gmc.pdb0_bo)
+		pt_base = amdgpu_gmc_pd_addr(adev->gmc.pdb0_bo);
+	else
+		pt_base = amdgpu_gmc_pd_addr(adev->gart.bo);
+
+	gfxhub_v1_2_xcc_setup_vm_pt_regs(adev, 0, pt_base, xcc_mask);
+
+	/* If use GART for FB translation, vmid0 page table covers both
+	 * vram and system memory (gart)
+	 */
+	for_each_inst(i, xcc_mask) {
+		if (adev->gmc.pdb0_bo) {
+			WREG32_SOC15(GC, GET_INST(GC, i),
+				     regVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+				     (u32)(gart_start >> 12));
+			WREG32_SOC15(GC, GET_INST(GC, i),
+				     regVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+				     (u32)(gart_start >> 44));
+
+			WREG32_SOC15(GC, GET_INST(GC, i),
+				     regVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+				     (u32)(adev->gmc.gart_end >> 12));
+			WREG32_SOC15(GC, GET_INST(GC, i),
+				     regVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+				     (u32)(adev->gmc.gart_end >> 44));
+		} else {
+			WREG32_SOC15(GC, GET_INST(GC, i),
+				     regVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+				     (u32)(adev->gmc.gart_start >> 12));
+			WREG32_SOC15(GC, GET_INST(GC, i),
+				     regVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+				     (u32)(adev->gmc.gart_start >> 44));
+
+			WREG32_SOC15(GC, GET_INST(GC, i),
+				     regVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+				     (u32)(adev->gmc.gart_end >> 12));
+			WREG32_SOC15(GC, GET_INST(GC, i),
+				     regVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+				     (u32)(adev->gmc.gart_end >> 44));
+		}
+	}
+}
+
+static void
+gfxhub_v1_2_xcc_init_system_aperture_regs(struct amdgpu_device *adev,
+					  uint32_t xcc_mask)
+{
+	uint64_t value;
+	uint32_t tmp;
+	int i;
+
+	for_each_inst(i, xcc_mask) {
+		/* Program the AGP BAR */
+		WREG32_SOC15_RLC(GC, GET_INST(GC, i), regMC_VM_AGP_BASE, 0);
+		WREG32_SOC15_RLC(GC, GET_INST(GC, i), regMC_VM_AGP_BOT, adev->gmc.agp_start >> 24);
+		WREG32_SOC15_RLC(GC, GET_INST(GC, i), regMC_VM_AGP_TOP, adev->gmc.agp_end >> 24);
+
+		if (!amdgpu_sriov_vf(adev) || adev->asic_type <= CHIP_VEGA10) {
+			/* Program the system aperture low logical page number. */
+			WREG32_SOC15_RLC(GC, GET_INST(GC, i), regMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+				min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
+
+			if (adev->apu_flags & (AMD_APU_IS_RAVEN2 |
+					       AMD_APU_IS_RENOIR |
+					       AMD_APU_IS_GREEN_SARDINE))
+			       /*
+				* Raven2 has a HW issue that it is unable to use the
+				* vram which is out of MC_VM_SYSTEM_APERTURE_HIGH_ADDR.
+				* So here is the workaround that increase system
+				* aperture high address (add 1) to get rid of the VM
+				* fault and hardware hang.
+				*/
+				WREG32_SOC15_RLC(GC, GET_INST(GC, i),
+						 regMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+						 max((adev->gmc.fb_end >> 18) + 0x1,
+						     adev->gmc.agp_end >> 18));
+			else
+				WREG32_SOC15_RLC(GC, GET_INST(GC, i),
+					regMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+					max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
+
+			/* Set default page address. */
+			value = amdgpu_gmc_vram_mc2pa(adev, adev->mem_scratch.gpu_addr);
+			WREG32_SOC15(GC, GET_INST(GC, i), regMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
+				     (u32)(value >> 12));
+			WREG32_SOC15(GC, GET_INST(GC, i), regMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
+				     (u32)(value >> 44));
+
+			/* Program "protection fault". */
+			WREG32_SOC15(GC, GET_INST(GC, i), regVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
+				     (u32)(adev->dummy_page_addr >> 12));
+			WREG32_SOC15(GC, GET_INST(GC, i), regVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+				     (u32)((u64)adev->dummy_page_addr >> 44));
+
+			tmp = RREG32_SOC15(GC, GET_INST(GC, i), regVM_L2_PROTECTION_FAULT_CNTL2);
+			tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL2,
+					    ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+			WREG32_SOC15(GC, GET_INST(GC, i), regVM_L2_PROTECTION_FAULT_CNTL2, tmp);
+		}
+
+		/* In the case squeezing vram into GART aperture, we don't use
+		 * FB aperture and AGP aperture. Disable them.
+		 */
+		if (adev->gmc.pdb0_bo && adev->gmc.xgmi.connected_to_cpu) {
+			WREG32_SOC15(GC, GET_INST(GC, i), regMC_VM_FB_LOCATION_TOP, 0);
+			WREG32_SOC15(GC, GET_INST(GC, i), regMC_VM_FB_LOCATION_BASE, 0x00FFFFFF);
+			WREG32_SOC15(GC, GET_INST(GC, i), regMC_VM_AGP_TOP, 0);
+			WREG32_SOC15(GC, GET_INST(GC, i), regMC_VM_AGP_BOT, 0xFFFFFF);
+			WREG32_SOC15(GC, GET_INST(GC, i), regMC_VM_SYSTEM_APERTURE_LOW_ADDR, 0x3FFFFFFF);
+			WREG32_SOC15(GC, GET_INST(GC, i), regMC_VM_SYSTEM_APERTURE_HIGH_ADDR, 0);
+		}
+	}
+}
+
+static void gfxhub_v1_2_xcc_init_tlb_regs(struct amdgpu_device *adev,
+					  uint32_t xcc_mask)
+{
+	uint32_t tmp;
+	int i;
+
+	for_each_inst(i, xcc_mask) {
+		/* Setup TLB control */
+		tmp = RREG32_SOC15(GC, GET_INST(GC, i), regMC_VM_MX_L1_TLB_CNTL);
+
+		tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+				    ENABLE_L1_TLB, 1);
+		tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+				    SYSTEM_ACCESS_MODE, 3);
+		tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+				    ENABLE_ADVANCED_DRIVER_MODEL, 1);
+		tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+				    SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+		tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+				    MTYPE, MTYPE_UC);/* XXX for emulation. */
+		tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ATC_EN, 1);
+
+		WREG32_SOC15_RLC(GC, GET_INST(GC, i), regMC_VM_MX_L1_TLB_CNTL, tmp);
+	}
+}
+
+static void gfxhub_v1_2_xcc_init_cache_regs(struct amdgpu_device *adev,
+					    uint32_t xcc_mask)
+{
+	uint32_t tmp;
+	int i;
+
+	for_each_inst(i, xcc_mask) {
+		/* Setup L2 cache */
+		tmp = RREG32_SOC15(GC, GET_INST(GC, i), regVM_L2_CNTL);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 1);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 1);
+		/* XXX for emulation, Refer to closed source code.*/
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, L2_PDE0_CACHE_TAG_GENERATION_MODE,
+				    0);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 0);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0);
+		WREG32_SOC15_RLC(GC, GET_INST(GC, i), regVM_L2_CNTL, tmp);
+
+		tmp = RREG32_SOC15(GC, GET_INST(GC, i), regVM_L2_CNTL2);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
+		WREG32_SOC15_RLC(GC, GET_INST(GC, i), regVM_L2_CNTL2, tmp);
+
+		tmp = regVM_L2_CNTL3_DEFAULT;
+		if (adev->gmc.translate_further) {
+			tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 12);
+			tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3,
+					    L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+		} else {
+			tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 9);
+			tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3,
+					    L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+		}
+		WREG32_SOC15_RLC(GC, GET_INST(GC, i), regVM_L2_CNTL3, tmp);
+
+		tmp = regVM_L2_CNTL4_DEFAULT;
+		/* For AMD APP APUs setup WC memory */
+		if (adev->gmc.xgmi.connected_to_cpu || adev->gmc.is_app_apu) {
+			tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 1);
+			tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 1);
+		} else {
+			tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
+			tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
+		}
+		WREG32_SOC15_RLC(GC, GET_INST(GC, i), regVM_L2_CNTL4, tmp);
+	}
+}
+
+static void gfxhub_v1_2_xcc_enable_system_domain(struct amdgpu_device *adev,
+						 uint32_t xcc_mask)
+{
+	uint32_t tmp;
+	int i;
+
+	for_each_inst(i, xcc_mask) {
+		tmp = RREG32_SOC15(GC, GET_INST(GC, i), regVM_CONTEXT0_CNTL);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH,
+				adev->gmc.vmid0_page_table_depth);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, PAGE_TABLE_BLOCK_SIZE,
+				adev->gmc.vmid0_page_table_block_size);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL,
+				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
+		WREG32_SOC15(GC, GET_INST(GC, i), regVM_CONTEXT0_CNTL, tmp);
+	}
+}
+
+static void
+gfxhub_v1_2_xcc_disable_identity_aperture(struct amdgpu_device *adev,
+					  uint32_t xcc_mask)
+{
+	int i;
+
+	for_each_inst(i, xcc_mask) {
+		WREG32_SOC15(GC, GET_INST(GC, i),
+			     regVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32,
+			     0XFFFFFFFF);
+		WREG32_SOC15(GC, GET_INST(GC, i),
+			     regVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32,
+			     0x0000000F);
+
+		WREG32_SOC15(GC, GET_INST(GC, i),
+			     regVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32,
+			     0);
+		WREG32_SOC15(GC, GET_INST(GC, i),
+			     regVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32,
+			     0);
+
+		WREG32_SOC15(GC, GET_INST(GC, i),
+			     regVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32, 0);
+		WREG32_SOC15(GC, GET_INST(GC, i),
+			     regVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32, 0);
+	}
+}
+
+static inline bool
+gfxhub_v1_2_per_process_xnack_support(struct amdgpu_device *adev)
+{
+	/*
+	 * TODO: Check if this function is really needed, so far only 9.4.3
+	 * variants use GFXHUB 1.2
+	 */
+	return !!adev->aid_mask;
+}
+
+static void gfxhub_v1_2_xcc_setup_vmid_config(struct amdgpu_device *adev,
+					      uint32_t xcc_mask)
+{
+	struct amdgpu_vmhub *hub;
+	unsigned int num_level, block_size;
+	uint32_t tmp;
+	int i, j;
+
+	num_level = adev->vm_manager.num_level;
+	block_size = adev->vm_manager.block_size;
+	if (adev->gmc.translate_further)
+		num_level -= 1;
+	else
+		block_size -= 9;
+
+	for_each_inst(j, xcc_mask) {
+		hub = &adev->vmhub[AMDGPU_GFXHUB(j)];
+		for (i = 0; i <= 14; i++) {
+			tmp = RREG32_SOC15_OFFSET(GC, GET_INST(GC, j), regVM_CONTEXT1_CNTL,
+					i * hub->ctx_distance);
+			tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
+			tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH,
+					    num_level);
+			tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+					    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+			tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+					    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT,
+					    1);
+			tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+					    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+			tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+					    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+			tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+					    READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+			tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+					    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+			tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+					    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+			tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+					    PAGE_TABLE_BLOCK_SIZE,
+					    block_size);
+			/* Send no-retry XNACK on fault to suppress VM fault storm.
+			 * On 9.4.3 variants, XNACK can be enabled in
+			 * the SQ per-process.
+			 * Retry faults need to be enabled for that to work.
+			 */
+			tmp = REG_SET_FIELD(
+				tmp, VM_CONTEXT1_CNTL,
+				RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
+				!adev->gmc.noretry ||
+					gfxhub_v1_2_per_process_xnack_support(
+						adev));
+			WREG32_SOC15_OFFSET(GC, GET_INST(GC, j), regVM_CONTEXT1_CNTL,
+					    i * hub->ctx_distance, tmp);
+			WREG32_SOC15_OFFSET(GC, GET_INST(GC, j),
+					    regVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,
+					    i * hub->ctx_addr_distance, 0);
+			WREG32_SOC15_OFFSET(GC, GET_INST(GC, j),
+					    regVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32,
+					    i * hub->ctx_addr_distance, 0);
+			WREG32_SOC15_OFFSET(GC, GET_INST(GC, j),
+					    regVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32,
+					    i * hub->ctx_addr_distance,
+					    lower_32_bits(adev->vm_manager.max_pfn - 1));
+			WREG32_SOC15_OFFSET(GC, GET_INST(GC, j),
+					    regVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32,
+					    i * hub->ctx_addr_distance,
+					    upper_32_bits(adev->vm_manager.max_pfn - 1));
+		}
+	}
+}
+
+static void gfxhub_v1_2_xcc_program_invalidation(struct amdgpu_device *adev,
+						 uint32_t xcc_mask)
+{
+	struct amdgpu_vmhub *hub;
+	unsigned int i, j;
+
+	for_each_inst(j, xcc_mask) {
+		hub = &adev->vmhub[AMDGPU_GFXHUB(j)];
+
+		for (i = 0 ; i < 18; ++i) {
+			WREG32_SOC15_OFFSET(GC, GET_INST(GC, j), regVM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
+					    i * hub->eng_addr_distance, 0xffffffff);
+			WREG32_SOC15_OFFSET(GC, GET_INST(GC, j), regVM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
+					    i * hub->eng_addr_distance, 0x1f);
+		}
+	}
+}
+
+static int gfxhub_v1_2_xcc_gart_enable(struct amdgpu_device *adev,
+				       uint32_t xcc_mask)
+{
+	/* GART Enable. */
+	gfxhub_v1_2_xcc_init_gart_aperture_regs(adev, xcc_mask);
+	gfxhub_v1_2_xcc_init_system_aperture_regs(adev, xcc_mask);
+	gfxhub_v1_2_xcc_init_tlb_regs(adev, xcc_mask);
+	if (!amdgpu_sriov_vf(adev))
+		gfxhub_v1_2_xcc_init_cache_regs(adev, xcc_mask);
+
+	gfxhub_v1_2_xcc_enable_system_domain(adev, xcc_mask);
+	if (!amdgpu_sriov_vf(adev))
+		gfxhub_v1_2_xcc_disable_identity_aperture(adev, xcc_mask);
+	gfxhub_v1_2_xcc_setup_vmid_config(adev, xcc_mask);
+	gfxhub_v1_2_xcc_program_invalidation(adev, xcc_mask);
+
+	return 0;
+}
+
+static int gfxhub_v1_2_gart_enable(struct amdgpu_device *adev)
+{
+	uint32_t xcc_mask;
+
+	xcc_mask = GENMASK(NUM_XCC(adev->gfx.xcc_mask) - 1, 0);
+	return gfxhub_v1_2_xcc_gart_enable(adev, xcc_mask);
+}
+
+static void gfxhub_v1_2_xcc_gart_disable(struct amdgpu_device *adev,
+					 uint32_t xcc_mask)
+{
+	struct amdgpu_vmhub *hub;
+	u32 tmp;
+	u32 i, j;
+
+	for_each_inst(j, xcc_mask) {
+		hub = &adev->vmhub[AMDGPU_GFXHUB(j)];
+		/* Disable all tables */
+		for (i = 0; i < 16; i++)
+			WREG32_SOC15_OFFSET(GC, GET_INST(GC, j), regVM_CONTEXT0_CNTL,
+					    i * hub->ctx_distance, 0);
+
+		/* Setup TLB control */
+		tmp = RREG32_SOC15(GC, GET_INST(GC, j), regMC_VM_MX_L1_TLB_CNTL);
+		tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
+		tmp = REG_SET_FIELD(tmp,
+					MC_VM_MX_L1_TLB_CNTL,
+					ENABLE_ADVANCED_DRIVER_MODEL,
+					0);
+		WREG32_SOC15_RLC(GC, GET_INST(GC, j), regMC_VM_MX_L1_TLB_CNTL, tmp);
+
+		/* Setup L2 cache */
+		if (!amdgpu_sriov_vf(adev)) {
+			tmp = RREG32_SOC15(GC, GET_INST(GC, j), regVM_L2_CNTL);
+			tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 0);
+			WREG32_SOC15(GC, GET_INST(GC, j), regVM_L2_CNTL, tmp);
+			WREG32_SOC15(GC, GET_INST(GC, j), regVM_L2_CNTL3, 0);
+		}
+	}
+}
+
+static void gfxhub_v1_2_gart_disable(struct amdgpu_device *adev)
+{
+	uint32_t xcc_mask;
+
+	xcc_mask = GENMASK(NUM_XCC(adev->gfx.xcc_mask) - 1, 0);
+	gfxhub_v1_2_xcc_gart_disable(adev, xcc_mask);
+}
+
+static void gfxhub_v1_2_xcc_set_fault_enable_default(struct amdgpu_device *adev,
+						     bool value,
+						     uint32_t xcc_mask)
+{
+	u32 tmp;
+	int i;
+
+	for_each_inst(i, xcc_mask) {
+		tmp = RREG32_SOC15(GC, GET_INST(GC, i), regVM_L2_PROTECTION_FAULT_CNTL);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				PDE1_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				PDE2_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+		tmp = REG_SET_FIELD(tmp,
+				VM_L2_PROTECTION_FAULT_CNTL,
+				TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT,
+				value);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				NACK_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+		if (!value) {
+			tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+					CRASH_ON_NO_RETRY_FAULT, 1);
+			tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+					CRASH_ON_RETRY_FAULT, 1);
+		}
+		WREG32_SOC15(GC, GET_INST(GC, i), regVM_L2_PROTECTION_FAULT_CNTL, tmp);
+	}
+}
+
+/**
+ * gfxhub_v1_2_set_fault_enable_default - update GART/VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+static void gfxhub_v1_2_set_fault_enable_default(struct amdgpu_device *adev,
+						 bool value)
+{
+	uint32_t xcc_mask;
+
+	xcc_mask = GENMASK(NUM_XCC(adev->gfx.xcc_mask) - 1, 0);
+	gfxhub_v1_2_xcc_set_fault_enable_default(adev, value, xcc_mask);
+}
+
+static void gfxhub_v1_2_xcc_init(struct amdgpu_device *adev, uint32_t xcc_mask)
+{
+	struct amdgpu_vmhub *hub;
+	int i;
+
+	for_each_inst(i, xcc_mask) {
+		hub = &adev->vmhub[AMDGPU_GFXHUB(i)];
+
+		hub->ctx0_ptb_addr_lo32 =
+			SOC15_REG_OFFSET(GC, GET_INST(GC, i),
+				regVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32);
+		hub->ctx0_ptb_addr_hi32 =
+			SOC15_REG_OFFSET(GC, GET_INST(GC, i),
+				regVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32);
+		hub->vm_inv_eng0_sem =
+			SOC15_REG_OFFSET(GC, GET_INST(GC, i), regVM_INVALIDATE_ENG0_SEM);
+		hub->vm_inv_eng0_req =
+			SOC15_REG_OFFSET(GC, GET_INST(GC, i), regVM_INVALIDATE_ENG0_REQ);
+		hub->vm_inv_eng0_ack =
+			SOC15_REG_OFFSET(GC, GET_INST(GC, i), regVM_INVALIDATE_ENG0_ACK);
+		hub->vm_context0_cntl =
+			SOC15_REG_OFFSET(GC, GET_INST(GC, i), regVM_CONTEXT0_CNTL);
+		hub->vm_l2_pro_fault_status =
+			SOC15_REG_OFFSET(GC, GET_INST(GC, i),
+				regVM_L2_PROTECTION_FAULT_STATUS);
+		hub->vm_l2_pro_fault_cntl =
+			SOC15_REG_OFFSET(GC, GET_INST(GC, i), regVM_L2_PROTECTION_FAULT_CNTL);
+
+		hub->ctx_distance = regVM_CONTEXT1_CNTL -
+				regVM_CONTEXT0_CNTL;
+		hub->ctx_addr_distance =
+				regVM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 -
+				regVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32;
+		hub->eng_distance = regVM_INVALIDATE_ENG1_REQ -
+				regVM_INVALIDATE_ENG0_REQ;
+		hub->eng_addr_distance =
+				regVM_INVALIDATE_ENG1_ADDR_RANGE_LO32 -
+				regVM_INVALIDATE_ENG0_ADDR_RANGE_LO32;
+	}
+}
+
+static void gfxhub_v1_2_init(struct amdgpu_device *adev)
+{
+	uint32_t xcc_mask;
+
+	xcc_mask = GENMASK(NUM_XCC(adev->gfx.xcc_mask) - 1, 0);
+	gfxhub_v1_2_xcc_init(adev, xcc_mask);
+}
+
+static int gfxhub_v1_2_get_xgmi_info(struct amdgpu_device *adev)
+{
+	u32 max_num_physical_nodes;
+	u32 max_physical_node_id;
+	u32 xgmi_lfb_cntl;
+	u32 max_region;
+	u64 seg_size;
+
+	xgmi_lfb_cntl = RREG32_SOC15(GC, GET_INST(GC, 0), regMC_VM_XGMI_LFB_CNTL);
+	seg_size = REG_GET_FIELD(
+		RREG32_SOC15(GC, GET_INST(GC, 0), regMC_VM_XGMI_LFB_SIZE),
+		MC_VM_XGMI_LFB_SIZE, PF_LFB_SIZE) << 24;
+	max_region =
+		REG_GET_FIELD(xgmi_lfb_cntl, MC_VM_XGMI_LFB_CNTL, PF_MAX_REGION);
+
+
+
+	max_num_physical_nodes   = 8;
+	max_physical_node_id     = 7;
+
+	/* PF_MAX_REGION=0 means xgmi is disabled */
+	if (max_region || adev->gmc.xgmi.connected_to_cpu) {
+		adev->gmc.xgmi.num_physical_nodes = max_region + 1;
+
+		if (adev->gmc.xgmi.num_physical_nodes > max_num_physical_nodes)
+			return -EINVAL;
+
+		adev->gmc.xgmi.physical_node_id =
+			REG_GET_FIELD(xgmi_lfb_cntl, MC_VM_XGMI_LFB_CNTL,
+					PF_LFB_REGION);
+
+		if (adev->gmc.xgmi.physical_node_id > max_physical_node_id)
+			return -EINVAL;
+
+		adev->gmc.xgmi.node_segment_size = seg_size;
+	}
+
+	return 0;
+}
+
+const struct amdgpu_gfxhub_funcs gfxhub_v1_2_funcs = {
+	.get_mc_fb_offset = gfxhub_v1_2_get_mc_fb_offset,
+	.setup_vm_pt_regs = gfxhub_v1_2_setup_vm_pt_regs,
+	.gart_enable = gfxhub_v1_2_gart_enable,
+	.gart_disable = gfxhub_v1_2_gart_disable,
+	.set_fault_enable_default = gfxhub_v1_2_set_fault_enable_default,
+	.init = gfxhub_v1_2_init,
+	.get_xgmi_info = gfxhub_v1_2_get_xgmi_info,
+};
+
+static int gfxhub_v1_2_xcp_resume(void *handle, uint32_t inst_mask)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	bool value;
+
+	if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS)
+		value = false;
+	else
+		value = true;
+
+	gfxhub_v1_2_xcc_set_fault_enable_default(adev, value, inst_mask);
+
+	if (!amdgpu_sriov_vf(adev))
+		return gfxhub_v1_2_xcc_gart_enable(adev, inst_mask);
+
+	return 0;
+}
+
+static int gfxhub_v1_2_xcp_suspend(void *handle, uint32_t inst_mask)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!amdgpu_sriov_vf(adev))
+		gfxhub_v1_2_xcc_gart_disable(adev, inst_mask);
+
+	return 0;
+}
+
+struct amdgpu_xcp_ip_funcs gfxhub_v1_2_xcp_funcs = {
+	.suspend = &gfxhub_v1_2_xcp_suspend,
+	.resume = &gfxhub_v1_2_xcp_resume
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.h b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.h
new file mode 100644
index 000000000000..997e9f90c990
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFXHUB_V1_2_H__
+#define __GFXHUB_V1_2_H__
+
+extern const struct amdgpu_gfxhub_funcs gfxhub_v1_2_funcs;
+
+extern struct amdgpu_xcp_ip_funcs gfxhub_v1_2_xcp_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_0.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_0.c
new file mode 100644
index 000000000000..793faf62cb07
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_0.c
@@ -0,0 +1,488 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "gfxhub_v2_0.h"
+
+#include "gc/gc_10_1_0_offset.h"
+#include "gc/gc_10_1_0_sh_mask.h"
+#include "gc/gc_10_1_0_default.h"
+#include "navi10_enum.h"
+
+#include "soc15_common.h"
+
+static const char * const gfxhub_client_ids[] = {
+	"CB/DB",
+	"Reserved",
+	"GE1",
+	"GE2",
+	"CPF",
+	"CPC",
+	"CPG",
+	"RLC",
+	"TCP",
+	"SQC (inst)",
+	"SQC (data)",
+	"SQG",
+	"Reserved",
+	"SDMA0",
+	"SDMA1",
+	"GCR",
+	"SDMA2",
+	"SDMA3",
+};
+
+static uint32_t gfxhub_v2_0_get_invalidate_req(unsigned int vmid,
+					       uint32_t flush_type)
+{
+	u32 req = 0;
+
+	/* invalidate using legacy mode on vmid*/
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ,
+			    PER_VMID_INVALIDATE_REQ, 1 << vmid);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, FLUSH_TYPE, flush_type);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PTES, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE0, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE1, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE2, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L1_PTES, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ,
+			    CLEAR_PROTECTION_FAULT_STATUS_ADDR,	0);
+
+	return req;
+}
+
+static void
+gfxhub_v2_0_print_l2_protection_fault_status(struct amdgpu_device *adev,
+					     uint32_t status)
+{
+	u32 cid = REG_GET_FIELD(status,
+				GCVM_L2_PROTECTION_FAULT_STATUS, CID);
+
+	dev_err(adev->dev,
+		"GCVM_L2_PROTECTION_FAULT_STATUS:0x%08X\n",
+		status);
+	dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n",
+		cid >= ARRAY_SIZE(gfxhub_client_ids) ? "unknown" : gfxhub_client_ids[cid],
+		cid);
+	dev_err(adev->dev, "\t MORE_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, MORE_FAULTS));
+	dev_err(adev->dev, "\t WALKER_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, WALKER_ERROR));
+	dev_err(adev->dev, "\t PERMISSION_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, PERMISSION_FAULTS));
+	dev_err(adev->dev, "\t MAPPING_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, MAPPING_ERROR));
+	dev_err(adev->dev, "\t RW: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, RW));
+}
+
+static u64 gfxhub_v2_0_get_fb_location(struct amdgpu_device *adev)
+{
+	u64 base = RREG32_SOC15(GC, 0, mmGCMC_VM_FB_LOCATION_BASE);
+
+	base &= GCMC_VM_FB_LOCATION_BASE__FB_BASE_MASK;
+	base <<= 24;
+
+	return base;
+}
+
+static u64 gfxhub_v2_0_get_mc_fb_offset(struct amdgpu_device *adev)
+{
+	return (u64)RREG32_SOC15(GC, 0, mmGCMC_VM_FB_OFFSET) << 24;
+}
+
+static void gfxhub_v2_0_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid,
+				uint64_t page_table_base)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+
+	WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+			    hub->ctx_addr_distance * vmid,
+			    lower_32_bits(page_table_base));
+
+	WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+			    hub->ctx_addr_distance * vmid,
+			    upper_32_bits(page_table_base));
+}
+
+static void gfxhub_v2_0_init_gart_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t pt_base = amdgpu_gmc_pd_addr(adev->gart.bo);
+
+	gfxhub_v2_0_setup_vm_pt_regs(adev, 0, pt_base);
+
+	WREG32_SOC15(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+		     (u32)(adev->gmc.gart_start >> 12));
+	WREG32_SOC15(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+		     (u32)(adev->gmc.gart_start >> 44));
+
+	WREG32_SOC15(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+		     (u32)(adev->gmc.gart_end >> 12));
+	WREG32_SOC15(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+		     (u32)(adev->gmc.gart_end >> 44));
+}
+
+static void gfxhub_v2_0_init_system_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t value;
+
+	if (!amdgpu_sriov_vf(adev)) {
+		/* Program the AGP BAR */
+		WREG32_SOC15(GC, 0, mmGCMC_VM_AGP_BASE, 0);
+		WREG32_SOC15(GC, 0, mmGCMC_VM_AGP_BOT, adev->gmc.agp_start >> 24);
+		WREG32_SOC15(GC, 0, mmGCMC_VM_AGP_TOP, adev->gmc.agp_end >> 24);
+
+		/* Program the system aperture low logical page number. */
+		WREG32_SOC15(GC, 0, mmGCMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+			     min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
+		WREG32_SOC15(GC, 0, mmGCMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+			     max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
+
+		/* Set default page address. */
+		value = amdgpu_gmc_vram_mc2pa(adev, adev->mem_scratch.gpu_addr);
+		WREG32_SOC15(GC, 0, mmGCMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
+			     (u32)(value >> 12));
+		WREG32_SOC15(GC, 0, mmGCMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
+			     (u32)(value >> 44));
+	}
+
+	/* Program "protection fault". */
+	WREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
+		     (u32)(adev->dummy_page_addr >> 12));
+	WREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+		     (u32)((u64)adev->dummy_page_addr >> 44));
+
+	WREG32_FIELD15(GC, 0, GCVM_L2_PROTECTION_FAULT_CNTL2,
+		       ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+}
+
+
+static void gfxhub_v2_0_init_tlb_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(GC, 0, mmGCMC_VM_MX_L1_TLB_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 1);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    MTYPE, MTYPE_UC); /* UC, uncached */
+
+	WREG32_SOC15(GC, 0, mmGCMC_VM_MX_L1_TLB_CNTL, tmp);
+}
+
+static void gfxhub_v2_0_init_cache_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* These regs are not accessible for VF, PF will program these in SRIOV */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(GC, 0, mmGCVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, ENABLE_L2_CACHE, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL,
+			    ENABLE_DEFAULT_PAGE_OUT_TO_SYSTEM_MEMORY, 1);
+	/* XXX for emulation, Refer to closed source code.*/
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL,
+			    L2_PDE0_CACHE_TAG_GENERATION_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CNTL, tmp);
+
+	tmp = RREG32_SOC15(GC, 0, mmGCVM_L2_CNTL2);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CNTL2, tmp);
+
+	tmp = mmGCVM_L2_CNTL3_DEFAULT;
+	if (adev->gmc.translate_further) {
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3, BANK_SELECT, 12);
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+	} else {
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3, BANK_SELECT, 9);
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+	}
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CNTL3, tmp);
+
+	tmp = mmGCVM_L2_CNTL4_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CNTL4, tmp);
+
+	tmp = mmGCVM_L2_CNTL5_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL5, L2_CACHE_SMALLK_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CNTL5, tmp);
+}
+
+static void gfxhub_v2_0_enable_system_domain(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	tmp = RREG32_SOC15(GC, 0, mmGCVM_CONTEXT0_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT0_CNTL,
+			    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
+	WREG32_SOC15(GC, 0, mmGCVM_CONTEXT0_CNTL, tmp);
+}
+
+static void gfxhub_v2_0_disable_identity_aperture(struct amdgpu_device *adev)
+{
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32,
+		     0xFFFFFFFF);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32,
+		     0x0000000F);
+
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32,
+		     0);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32,
+		     0);
+
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32, 0);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32, 0);
+
+}
+
+static void gfxhub_v2_0_setup_vmid_config(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	int i;
+	uint32_t tmp;
+
+	for (i = 0; i <= 14; i++) {
+		tmp = RREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_CNTL, i * hub->ctx_distance);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH,
+				    adev->vm_manager.num_level);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				PAGE_TABLE_BLOCK_SIZE,
+				adev->vm_manager.block_size - 9);
+		/* Send no-retry XNACK on fault to suppress VM fault storm. */
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
+				    !adev->gmc.noretry);
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_CNTL,
+				    i * hub->ctx_distance, tmp);
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32,
+				    i * hub->ctx_addr_distance,
+				    lower_32_bits(adev->vm_manager.max_pfn - 1));
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32,
+				    i * hub->ctx_addr_distance,
+				    upper_32_bits(adev->vm_manager.max_pfn - 1));
+	}
+
+	hub->vm_cntx_cntl = tmp;
+}
+
+static void gfxhub_v2_0_program_invalidation(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	unsigned int i;
+
+	for (i = 0 ; i < 18; ++i) {
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
+				    i * hub->eng_addr_distance, 0xffffffff);
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
+				    i * hub->eng_addr_distance, 0x1f);
+	}
+}
+
+static int gfxhub_v2_0_gart_enable(struct amdgpu_device *adev)
+{
+	/* GART Enable. */
+	gfxhub_v2_0_init_gart_aperture_regs(adev);
+	gfxhub_v2_0_init_system_aperture_regs(adev);
+	gfxhub_v2_0_init_tlb_regs(adev);
+	gfxhub_v2_0_init_cache_regs(adev);
+
+	gfxhub_v2_0_enable_system_domain(adev);
+	gfxhub_v2_0_disable_identity_aperture(adev);
+	gfxhub_v2_0_setup_vmid_config(adev);
+	gfxhub_v2_0_program_invalidation(adev);
+
+	return 0;
+}
+
+static void gfxhub_v2_0_gart_disable(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	u32 tmp;
+	u32 i;
+
+	/* Disable all tables */
+	for (i = 0; i < 16; i++)
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT0_CNTL,
+				    i * hub->ctx_distance, 0);
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(GC, 0, mmGCMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 0);
+	WREG32_SOC15(GC, 0, mmGCMC_VM_MX_L1_TLB_CNTL, tmp);
+
+	if (!amdgpu_sriov_vf(adev)) {
+		/* Setup L2 cache */
+		WREG32_FIELD15(GC, 0, GCVM_L2_CNTL, ENABLE_L2_CACHE, 0);
+		WREG32_SOC15(GC, 0, mmGCVM_L2_CNTL3, 0);
+	}
+}
+
+/**
+ * gfxhub_v2_0_set_fault_enable_default - update GART/VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+static void gfxhub_v2_0_set_fault_enable_default(struct amdgpu_device *adev,
+					  bool value)
+{
+	u32 tmp;
+
+	tmp = RREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE1_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE2_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT,
+			    value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    NACK_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	if (!value) {
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_NO_RETRY_FAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_RETRY_FAULT, 1);
+	}
+	WREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_CNTL, tmp);
+}
+
+static const struct amdgpu_vmhub_funcs gfxhub_v2_0_vmhub_funcs = {
+	.print_l2_protection_fault_status = gfxhub_v2_0_print_l2_protection_fault_status,
+	.get_invalidate_req = gfxhub_v2_0_get_invalidate_req,
+};
+
+static void gfxhub_v2_0_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+
+	hub->ctx0_ptb_addr_lo32 =
+		SOC15_REG_OFFSET(GC, 0,
+				 mmGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32);
+	hub->ctx0_ptb_addr_hi32 =
+		SOC15_REG_OFFSET(GC, 0,
+				 mmGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32);
+	hub->vm_inv_eng0_sem =
+		SOC15_REG_OFFSET(GC, 0, mmGCVM_INVALIDATE_ENG0_SEM);
+	hub->vm_inv_eng0_req =
+		SOC15_REG_OFFSET(GC, 0, mmGCVM_INVALIDATE_ENG0_REQ);
+	hub->vm_inv_eng0_ack =
+		SOC15_REG_OFFSET(GC, 0, mmGCVM_INVALIDATE_ENG0_ACK);
+	hub->vm_context0_cntl =
+		SOC15_REG_OFFSET(GC, 0, mmGCVM_CONTEXT0_CNTL);
+	hub->vm_l2_pro_fault_status =
+		SOC15_REG_OFFSET(GC, 0, mmGCVM_L2_PROTECTION_FAULT_STATUS);
+	hub->vm_l2_pro_fault_cntl =
+		SOC15_REG_OFFSET(GC, 0, mmGCVM_L2_PROTECTION_FAULT_CNTL);
+
+	hub->ctx_distance = mmGCVM_CONTEXT1_CNTL - mmGCVM_CONTEXT0_CNTL;
+	hub->ctx_addr_distance = mmGCVM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 -
+		mmGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32;
+	hub->eng_distance = mmGCVM_INVALIDATE_ENG1_REQ -
+		mmGCVM_INVALIDATE_ENG0_REQ;
+	hub->eng_addr_distance = mmGCVM_INVALIDATE_ENG1_ADDR_RANGE_LO32 -
+		mmGCVM_INVALIDATE_ENG0_ADDR_RANGE_LO32;
+
+	hub->vm_cntx_cntl_vm_fault = GCVM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__EXECUTE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK;
+
+	/* TODO: This is only needed on some Navi 1x revisions */
+	hub->sdma_invalidation_workaround = true;
+
+	hub->vmhub_funcs = &gfxhub_v2_0_vmhub_funcs;
+}
+
+const struct amdgpu_gfxhub_funcs gfxhub_v2_0_funcs = {
+	.get_fb_location = gfxhub_v2_0_get_fb_location,
+	.get_mc_fb_offset = gfxhub_v2_0_get_mc_fb_offset,
+	.setup_vm_pt_regs = gfxhub_v2_0_setup_vm_pt_regs,
+	.gart_enable = gfxhub_v2_0_gart_enable,
+	.gart_disable = gfxhub_v2_0_gart_disable,
+	.set_fault_enable_default = gfxhub_v2_0_set_fault_enable_default,
+	.init = gfxhub_v2_0_init,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_0.h b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_0.h
new file mode 100644
index 000000000000..9ddc35cd53d4
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFXHUB_V2_0_H__
+#define __GFXHUB_V2_0_H__
+
+extern const struct amdgpu_gfxhub_funcs gfxhub_v2_0_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c
new file mode 100644
index 000000000000..deb95fab02df
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c
@@ -0,0 +1,667 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "gfxhub_v2_1.h"
+
+#include "gc/gc_10_3_0_offset.h"
+#include "gc/gc_10_3_0_sh_mask.h"
+#include "gc/gc_10_3_0_default.h"
+#include "navi10_enum.h"
+
+#include "soc15_common.h"
+
+#define mmGCUTCL2_HARVEST_BYPASS_GROUPS_YELLOW_CARP				0x16f8
+#define mmGCUTCL2_HARVEST_BYPASS_GROUPS_YELLOW_CARP_BASE_IDX	0
+
+static const char * const gfxhub_client_ids[] = {
+	"CB/DB",
+	"Reserved",
+	"GE1",
+	"GE2",
+	"CPF",
+	"CPC",
+	"CPG",
+	"RLC",
+	"TCP",
+	"SQC (inst)",
+	"SQC (data)",
+	"SQG",
+	"Reserved",
+	"SDMA0",
+	"SDMA1",
+	"GCR",
+	"SDMA2",
+	"SDMA3",
+};
+
+static uint32_t gfxhub_v2_1_get_invalidate_req(unsigned int vmid,
+					       uint32_t flush_type)
+{
+	u32 req = 0;
+
+	/* invalidate using legacy mode on vmid*/
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ,
+			    PER_VMID_INVALIDATE_REQ, 1 << vmid);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, FLUSH_TYPE, flush_type);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PTES, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE0, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE1, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE2, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L1_PTES, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ,
+			    CLEAR_PROTECTION_FAULT_STATUS_ADDR,	0);
+
+	return req;
+}
+
+static void
+gfxhub_v2_1_print_l2_protection_fault_status(struct amdgpu_device *adev,
+					     uint32_t status)
+{
+	u32 cid = REG_GET_FIELD(status,
+				GCVM_L2_PROTECTION_FAULT_STATUS, CID);
+
+	dev_err(adev->dev,
+		"GCVM_L2_PROTECTION_FAULT_STATUS:0x%08X\n",
+		status);
+	dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n",
+		cid >= ARRAY_SIZE(gfxhub_client_ids) ? "unknown" : gfxhub_client_ids[cid],
+		cid);
+	dev_err(adev->dev, "\t MORE_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, MORE_FAULTS));
+	dev_err(adev->dev, "\t WALKER_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, WALKER_ERROR));
+	dev_err(adev->dev, "\t PERMISSION_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, PERMISSION_FAULTS));
+	dev_err(adev->dev, "\t MAPPING_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, MAPPING_ERROR));
+	dev_err(adev->dev, "\t RW: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, RW));
+}
+
+static u64 gfxhub_v2_1_get_fb_location(struct amdgpu_device *adev)
+{
+	u64 base = RREG32_SOC15(GC, 0, mmGCMC_VM_FB_LOCATION_BASE);
+
+	base &= GCMC_VM_FB_LOCATION_BASE__FB_BASE_MASK;
+	base <<= 24;
+
+	return base;
+}
+
+static u64 gfxhub_v2_1_get_mc_fb_offset(struct amdgpu_device *adev)
+{
+	return (u64)RREG32_SOC15(GC, 0, mmGCMC_VM_FB_OFFSET) << 24;
+}
+
+static void gfxhub_v2_1_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid,
+				uint64_t page_table_base)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+
+	WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+			    hub->ctx_addr_distance * vmid,
+			    lower_32_bits(page_table_base));
+
+	WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+			    hub->ctx_addr_distance * vmid,
+			    upper_32_bits(page_table_base));
+}
+
+static void gfxhub_v2_1_init_gart_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t pt_base = amdgpu_gmc_pd_addr(adev->gart.bo);
+
+	gfxhub_v2_1_setup_vm_pt_regs(adev, 0, pt_base);
+
+	WREG32_SOC15(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+		     (u32)(adev->gmc.gart_start >> 12));
+	WREG32_SOC15(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+		     (u32)(adev->gmc.gart_start >> 44));
+
+	WREG32_SOC15(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+		     (u32)(adev->gmc.gart_end >> 12));
+	WREG32_SOC15(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+		     (u32)(adev->gmc.gart_end >> 44));
+}
+
+static void gfxhub_v2_1_init_system_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t value;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/* Program the AGP BAR */
+	WREG32_SOC15(GC, 0, mmGCMC_VM_AGP_BASE, 0);
+	WREG32_SOC15(GC, 0, mmGCMC_VM_AGP_BOT, adev->gmc.agp_start >> 24);
+	WREG32_SOC15(GC, 0, mmGCMC_VM_AGP_TOP, adev->gmc.agp_end >> 24);
+
+	/* Program the system aperture low logical page number. */
+	WREG32_SOC15(GC, 0, mmGCMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+		     min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
+	WREG32_SOC15(GC, 0, mmGCMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+		     max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
+
+	/* Set default page address. */
+	value = amdgpu_gmc_vram_mc2pa(adev, adev->mem_scratch.gpu_addr);
+	WREG32_SOC15(GC, 0, mmGCMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
+		     (u32)(value >> 12));
+	WREG32_SOC15(GC, 0, mmGCMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
+		     (u32)(value >> 44));
+
+	/* Program "protection fault". */
+	WREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
+		     (u32)(adev->dummy_page_addr >> 12));
+	WREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+		     (u32)((u64)adev->dummy_page_addr >> 44));
+
+	WREG32_FIELD15(GC, 0, GCVM_L2_PROTECTION_FAULT_CNTL2,
+		       ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+}
+
+
+static void gfxhub_v2_1_init_tlb_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(GC, 0, mmGCMC_VM_MX_L1_TLB_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 1);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    MTYPE, MTYPE_UC); /* UC, uncached */
+
+	WREG32_SOC15(GC, 0, mmGCMC_VM_MX_L1_TLB_CNTL, tmp);
+}
+
+static void gfxhub_v2_1_init_cache_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(GC, 0, mmGCVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, ENABLE_L2_CACHE, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL,
+			    ENABLE_DEFAULT_PAGE_OUT_TO_SYSTEM_MEMORY, 1);
+	/* XXX for emulation, Refer to closed source code.*/
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL,
+			    L2_PDE0_CACHE_TAG_GENERATION_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CNTL, tmp);
+
+	tmp = RREG32_SOC15(GC, 0, mmGCVM_L2_CNTL2);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CNTL2, tmp);
+
+	tmp = mmGCVM_L2_CNTL3_DEFAULT;
+	if (adev->gmc.translate_further) {
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3, BANK_SELECT, 12);
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+	} else {
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3, BANK_SELECT, 9);
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+	}
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CNTL3, tmp);
+
+	tmp = mmGCVM_L2_CNTL4_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CNTL4, tmp);
+
+	tmp = mmGCVM_L2_CNTL5_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL5, L2_CACHE_SMALLK_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CNTL5, tmp);
+}
+
+static void gfxhub_v2_1_enable_system_domain(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	tmp = RREG32_SOC15(GC, 0, mmGCVM_CONTEXT0_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT0_CNTL,
+			    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
+	WREG32_SOC15(GC, 0, mmGCVM_CONTEXT0_CNTL, tmp);
+}
+
+static void gfxhub_v2_1_disable_identity_aperture(struct amdgpu_device *adev)
+{
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32,
+		     0xFFFFFFFF);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32,
+		     0x0000000F);
+
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32,
+		     0);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32,
+		     0);
+
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32, 0);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32, 0);
+
+}
+
+static void gfxhub_v2_1_setup_vmid_config(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	int i;
+	uint32_t tmp;
+
+	for (i = 0; i <= 14; i++) {
+		tmp = RREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_CNTL, i * hub->ctx_distance);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH,
+				    adev->vm_manager.num_level);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				PAGE_TABLE_BLOCK_SIZE,
+				adev->vm_manager.block_size - 9);
+		/* Send no-retry XNACK on fault to suppress VM fault storm. */
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
+				    !adev->gmc.noretry);
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_CNTL,
+				    i * hub->ctx_distance, tmp);
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32,
+				    i * hub->ctx_addr_distance,
+				    lower_32_bits(adev->vm_manager.max_pfn - 1));
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32,
+				    i * hub->ctx_addr_distance,
+				    upper_32_bits(adev->vm_manager.max_pfn - 1));
+	}
+
+	hub->vm_cntx_cntl = tmp;
+}
+
+static void gfxhub_v2_1_program_invalidation(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	unsigned int i;
+
+	for (i = 0 ; i < 18; ++i) {
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
+				    i * hub->eng_addr_distance, 0xffffffff);
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
+				    i * hub->eng_addr_distance, 0x1f);
+	}
+}
+
+static int gfxhub_v2_1_gart_enable(struct amdgpu_device *adev)
+{
+	if (amdgpu_sriov_vf(adev)) {
+		/*
+		 * GCMC_VM_FB_LOCATION_BASE/TOP is NULL for VF, becuase they are
+		 * VF copy registers so vbios post doesn't program them, for
+		 * SRIOV driver need to program them
+		 */
+		WREG32_SOC15(GC, 0, mmGCMC_VM_FB_LOCATION_BASE,
+			     adev->gmc.vram_start >> 24);
+		WREG32_SOC15(GC, 0, mmGCMC_VM_FB_LOCATION_TOP,
+			     adev->gmc.vram_end >> 24);
+	}
+
+	/* GART Enable. */
+	gfxhub_v2_1_init_gart_aperture_regs(adev);
+	gfxhub_v2_1_init_system_aperture_regs(adev);
+	gfxhub_v2_1_init_tlb_regs(adev);
+	gfxhub_v2_1_init_cache_regs(adev);
+
+	gfxhub_v2_1_enable_system_domain(adev);
+	gfxhub_v2_1_disable_identity_aperture(adev);
+	gfxhub_v2_1_setup_vmid_config(adev);
+	gfxhub_v2_1_program_invalidation(adev);
+
+	return 0;
+}
+
+static void gfxhub_v2_1_gart_disable(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	u32 tmp;
+	u32 i;
+
+	/* Disable all tables */
+	for (i = 0; i < 16; i++)
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT0_CNTL,
+				    i * hub->ctx_distance, 0);
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(GC, 0, mmGCMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 0);
+	WREG32_SOC15(GC, 0, mmGCMC_VM_MX_L1_TLB_CNTL, tmp);
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/* Setup L2 cache */
+	WREG32_FIELD15(GC, 0, GCVM_L2_CNTL, ENABLE_L2_CACHE, 0);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CNTL3, 0);
+}
+
+/**
+ * gfxhub_v2_1_set_fault_enable_default - update GART/VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+static void gfxhub_v2_1_set_fault_enable_default(struct amdgpu_device *adev,
+					  bool value)
+{
+	u32 tmp;
+
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	tmp = RREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE1_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE2_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT,
+			    value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    NACK_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	if (!value) {
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_NO_RETRY_FAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_RETRY_FAULT, 1);
+	}
+	WREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_CNTL, tmp);
+}
+
+static const struct amdgpu_vmhub_funcs gfxhub_v2_1_vmhub_funcs = {
+	.print_l2_protection_fault_status = gfxhub_v2_1_print_l2_protection_fault_status,
+	.get_invalidate_req = gfxhub_v2_1_get_invalidate_req,
+};
+
+static void gfxhub_v2_1_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+
+	hub->ctx0_ptb_addr_lo32 =
+		SOC15_REG_OFFSET(GC, 0,
+				 mmGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32);
+	hub->ctx0_ptb_addr_hi32 =
+		SOC15_REG_OFFSET(GC, 0,
+				 mmGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32);
+	hub->vm_inv_eng0_sem =
+		SOC15_REG_OFFSET(GC, 0, mmGCVM_INVALIDATE_ENG0_SEM);
+	hub->vm_inv_eng0_req =
+		SOC15_REG_OFFSET(GC, 0, mmGCVM_INVALIDATE_ENG0_REQ);
+	hub->vm_inv_eng0_ack =
+		SOC15_REG_OFFSET(GC, 0, mmGCVM_INVALIDATE_ENG0_ACK);
+	hub->vm_context0_cntl =
+		SOC15_REG_OFFSET(GC, 0, mmGCVM_CONTEXT0_CNTL);
+	hub->vm_l2_pro_fault_status =
+		SOC15_REG_OFFSET(GC, 0, mmGCVM_L2_PROTECTION_FAULT_STATUS);
+	hub->vm_l2_pro_fault_cntl =
+		SOC15_REG_OFFSET(GC, 0, mmGCVM_L2_PROTECTION_FAULT_CNTL);
+
+	hub->ctx_distance = mmGCVM_CONTEXT1_CNTL - mmGCVM_CONTEXT0_CNTL;
+	hub->ctx_addr_distance = mmGCVM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 -
+		mmGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32;
+	hub->eng_distance = mmGCVM_INVALIDATE_ENG1_REQ -
+		mmGCVM_INVALIDATE_ENG0_REQ;
+	hub->eng_addr_distance = mmGCVM_INVALIDATE_ENG1_ADDR_RANGE_LO32 -
+		mmGCVM_INVALIDATE_ENG0_ADDR_RANGE_LO32;
+
+	hub->vm_cntx_cntl_vm_fault = GCVM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__EXECUTE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK;
+
+	hub->vmhub_funcs = &gfxhub_v2_1_vmhub_funcs;
+}
+
+static void gfxhub_v2_1_utcl2_harvest(struct amdgpu_device *adev)
+{
+	int i;
+	u32 tmp = 0, disabled_sa = 0;
+	u32 efuse_setting, vbios_setting;
+
+	u32 max_sa_mask = amdgpu_gfx_create_bitmask(
+		adev->gfx.config.max_sh_per_se *
+		adev->gfx.config.max_shader_engines);
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 3):
+		/* Get SA disabled bitmap from eFuse setting */
+		efuse_setting = RREG32_SOC15(GC, 0, mmCC_GC_SA_UNIT_DISABLE);
+		efuse_setting &= CC_GC_SA_UNIT_DISABLE__SA_DISABLE_MASK;
+		efuse_setting >>= CC_GC_SA_UNIT_DISABLE__SA_DISABLE__SHIFT;
+
+		/* Get SA disabled bitmap from VBIOS setting */
+		vbios_setting = RREG32_SOC15(GC, 0, mmGC_USER_SA_UNIT_DISABLE);
+		vbios_setting &= GC_USER_SA_UNIT_DISABLE__SA_DISABLE_MASK;
+		vbios_setting >>= GC_USER_SA_UNIT_DISABLE__SA_DISABLE__SHIFT;
+
+		disabled_sa |= efuse_setting | vbios_setting;
+		/* Make sure not to report harvested SAs beyond the max SA count */
+		disabled_sa &= max_sa_mask;
+
+		for (i = 0; disabled_sa > 0; i++) {
+			if (disabled_sa & 1)
+				tmp |= 0x3 << (i * 2);
+			disabled_sa >>= 1;
+		}
+		disabled_sa = tmp;
+
+		WREG32_SOC15(GC, 0, mmGCUTCL2_HARVEST_BYPASS_GROUPS_YELLOW_CARP, disabled_sa);
+		break;
+	default:
+		break;
+	}
+}
+
+static void gfxhub_v2_1_save_regs(struct amdgpu_device *adev)
+{
+	int i;
+
+	adev->gmc.VM_L2_CNTL = RREG32_SOC15(GC, 0, mmGCVM_L2_CNTL);
+	adev->gmc.VM_L2_CNTL2 = RREG32_SOC15(GC, 0, mmGCVM_L2_CNTL2);
+	adev->gmc.VM_DUMMY_PAGE_FAULT_CNTL = RREG32_SOC15(GC, 0, mmGCVM_DUMMY_PAGE_FAULT_CNTL);
+	adev->gmc.VM_DUMMY_PAGE_FAULT_ADDR_LO32 = RREG32_SOC15(GC, 0, mmGCVM_DUMMY_PAGE_FAULT_ADDR_LO32);
+	adev->gmc.VM_DUMMY_PAGE_FAULT_ADDR_HI32 = RREG32_SOC15(GC, 0, mmGCVM_DUMMY_PAGE_FAULT_ADDR_HI32);
+	adev->gmc.VM_L2_PROTECTION_FAULT_CNTL = RREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_CNTL);
+	adev->gmc.VM_L2_PROTECTION_FAULT_CNTL2 = RREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_CNTL2);
+	adev->gmc.VM_L2_PROTECTION_FAULT_MM_CNTL3 = RREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_MM_CNTL3);
+	adev->gmc.VM_L2_PROTECTION_FAULT_MM_CNTL4 = RREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_MM_CNTL4);
+	adev->gmc.VM_L2_PROTECTION_FAULT_ADDR_LO32 = RREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_ADDR_LO32);
+	adev->gmc.VM_L2_PROTECTION_FAULT_ADDR_HI32 = RREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_ADDR_HI32);
+	adev->gmc.VM_DEBUG = RREG32_SOC15(GC, 0, mmGCVM_DEBUG);
+	adev->gmc.VM_L2_MM_GROUP_RT_CLASSES = RREG32_SOC15(GC, 0, mmGCVM_L2_MM_GROUP_RT_CLASSES);
+	adev->gmc.VM_L2_BANK_SELECT_RESERVED_CID = RREG32_SOC15(GC, 0, mmGCVM_L2_BANK_SELECT_RESERVED_CID);
+	adev->gmc.VM_L2_BANK_SELECT_RESERVED_CID2 = RREG32_SOC15(GC, 0, mmGCVM_L2_BANK_SELECT_RESERVED_CID2);
+	adev->gmc.VM_L2_CACHE_PARITY_CNTL = RREG32_SOC15(GC, 0, mmGCVM_L2_CACHE_PARITY_CNTL);
+	adev->gmc.VM_L2_IH_LOG_CNTL = RREG32_SOC15(GC, 0, mmGCVM_L2_IH_LOG_CNTL);
+
+	for (i = 0; i <= 15; i++) {
+		adev->gmc.VM_CONTEXT_CNTL[i] = RREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT0_CNTL, i);
+		adev->gmc.VM_CONTEXT_PAGE_TABLE_BASE_ADDR_LO32[i] = RREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32, i * 2);
+		adev->gmc.VM_CONTEXT_PAGE_TABLE_BASE_ADDR_HI32[i] = RREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32, i * 2);
+		adev->gmc.VM_CONTEXT_PAGE_TABLE_START_ADDR_LO32[i] = RREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32, i * 2);
+		adev->gmc.VM_CONTEXT_PAGE_TABLE_START_ADDR_HI32[i] = RREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32, i * 2);
+		adev->gmc.VM_CONTEXT_PAGE_TABLE_END_ADDR_LO32[i] = RREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32, i * 2);
+		adev->gmc.VM_CONTEXT_PAGE_TABLE_END_ADDR_HI32[i] = RREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32, i * 2);
+	}
+
+	adev->gmc.MC_VM_MX_L1_TLB_CNTL = RREG32_SOC15(GC, 0, mmGCMC_VM_MX_L1_TLB_CNTL);
+}
+
+static void gfxhub_v2_1_restore_regs(struct amdgpu_device *adev)
+{
+	int i;
+
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CNTL, adev->gmc.VM_L2_CNTL);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CNTL2, adev->gmc.VM_L2_CNTL2);
+	WREG32_SOC15(GC, 0, mmGCVM_DUMMY_PAGE_FAULT_CNTL, adev->gmc.VM_DUMMY_PAGE_FAULT_CNTL);
+	WREG32_SOC15(GC, 0, mmGCVM_DUMMY_PAGE_FAULT_ADDR_LO32, adev->gmc.VM_DUMMY_PAGE_FAULT_ADDR_LO32);
+	WREG32_SOC15(GC, 0, mmGCVM_DUMMY_PAGE_FAULT_ADDR_HI32, adev->gmc.VM_DUMMY_PAGE_FAULT_ADDR_HI32);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_CNTL, adev->gmc.VM_L2_PROTECTION_FAULT_CNTL);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_CNTL2, adev->gmc.VM_L2_PROTECTION_FAULT_CNTL2);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_MM_CNTL3, adev->gmc.VM_L2_PROTECTION_FAULT_MM_CNTL3);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_MM_CNTL4, adev->gmc.VM_L2_PROTECTION_FAULT_MM_CNTL4);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_ADDR_LO32, adev->gmc.VM_L2_PROTECTION_FAULT_ADDR_LO32);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_ADDR_HI32, adev->gmc.VM_L2_PROTECTION_FAULT_ADDR_HI32);
+	WREG32_SOC15(GC, 0, mmGCVM_DEBUG, adev->gmc.VM_DEBUG);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_MM_GROUP_RT_CLASSES, adev->gmc.VM_L2_MM_GROUP_RT_CLASSES);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_BANK_SELECT_RESERVED_CID, adev->gmc.VM_L2_BANK_SELECT_RESERVED_CID);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_BANK_SELECT_RESERVED_CID2, adev->gmc.VM_L2_BANK_SELECT_RESERVED_CID2);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_CACHE_PARITY_CNTL, adev->gmc.VM_L2_CACHE_PARITY_CNTL);
+	WREG32_SOC15(GC, 0, mmGCVM_L2_IH_LOG_CNTL, adev->gmc.VM_L2_IH_LOG_CNTL);
+
+	for (i = 0; i <= 15; i++) {
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT0_CNTL, i, adev->gmc.VM_CONTEXT_CNTL[i]);
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32, i * 2, adev->gmc.VM_CONTEXT_PAGE_TABLE_BASE_ADDR_LO32[i]);
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32, i * 2, adev->gmc.VM_CONTEXT_PAGE_TABLE_BASE_ADDR_HI32[i]);
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32, i * 2, adev->gmc.VM_CONTEXT_PAGE_TABLE_START_ADDR_LO32[i]);
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32, i * 2, adev->gmc.VM_CONTEXT_PAGE_TABLE_START_ADDR_HI32[i]);
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32, i * 2, adev->gmc.VM_CONTEXT_PAGE_TABLE_END_ADDR_LO32[i]);
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32, i * 2, adev->gmc.VM_CONTEXT_PAGE_TABLE_END_ADDR_HI32[i]);
+	}
+
+	WREG32_SOC15(GC, 0, mmGCMC_VM_FB_LOCATION_BASE, adev->gmc.vram_start >> 24);
+	WREG32_SOC15(GC, 0, mmGCMC_VM_FB_LOCATION_TOP, adev->gmc.vram_end >> 24);
+	WREG32_SOC15(GC, 0, mmGCMC_VM_MX_L1_TLB_CNTL, adev->gmc.MC_VM_MX_L1_TLB_CNTL);
+}
+
+static void gfxhub_v2_1_halt(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	int i;
+	uint32_t tmp;
+	int time = 1000;
+
+	gfxhub_v2_1_set_fault_enable_default(adev, false);
+
+	for (i = 0; i <= 14; i++) {
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,
+				    i * hub->ctx_addr_distance, ~0);
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32,
+				    i * hub->ctx_addr_distance, ~0);
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32,
+				    i * hub->ctx_addr_distance,
+				    0);
+		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32,
+				    i * hub->ctx_addr_distance,
+				    0);
+	}
+	tmp = RREG32_SOC15(GC, 0, mmGRBM_STATUS2);
+	while ((tmp & (GRBM_STATUS2__EA_BUSY_MASK |
+		      GRBM_STATUS2__EA_LINK_BUSY_MASK)) != 0 &&
+	       time) {
+		udelay(100);
+		time--;
+		tmp = RREG32_SOC15(GC, 0, mmGRBM_STATUS2);
+	}
+
+	if (!time)
+		DRM_WARN("failed to wait for GRBM(EA) idle\n");
+}
+
+const struct amdgpu_gfxhub_funcs gfxhub_v2_1_funcs = {
+	.get_fb_location = gfxhub_v2_1_get_fb_location,
+	.get_mc_fb_offset = gfxhub_v2_1_get_mc_fb_offset,
+	.setup_vm_pt_regs = gfxhub_v2_1_setup_vm_pt_regs,
+	.gart_enable = gfxhub_v2_1_gart_enable,
+	.gart_disable = gfxhub_v2_1_gart_disable,
+	.set_fault_enable_default = gfxhub_v2_1_set_fault_enable_default,
+	.init = gfxhub_v2_1_init,
+	.utcl2_harvest = gfxhub_v2_1_utcl2_harvest,
+	.mode2_save_regs = gfxhub_v2_1_save_regs,
+	.mode2_restore_regs = gfxhub_v2_1_restore_regs,
+	.halt = gfxhub_v2_1_halt,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.h b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.h
new file mode 100644
index 000000000000..f75c2eccfad9
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFXHUB_V2_1_H__
+#define __GFXHUB_V2_1_H__
+
+extern const struct amdgpu_gfxhub_funcs gfxhub_v2_1_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v3_0.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v3_0.c
new file mode 100644
index 000000000000..abe30c8bd2ba
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v3_0.c
@@ -0,0 +1,513 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "gfxhub_v3_0.h"
+
+#include "gc/gc_11_0_0_offset.h"
+#include "gc/gc_11_0_0_sh_mask.h"
+#include "gc/gc_11_0_0_default.h"
+#include "navi10_enum.h"
+#include "soc15_common.h"
+
+static const char * const gfxhub_client_ids[] = {
+	"CB/DB",
+	"Reserved",
+	"GE1",
+	"GE2",
+	"CPF",
+	"CPC",
+	"CPG",
+	"RLC",
+	"TCP",
+	"SQC (inst)",
+	"SQC (data)",
+	"SQG",
+	"Reserved",
+	"SDMA0",
+	"SDMA1",
+	"GCR",
+	"SDMA2",
+	"SDMA3",
+};
+
+static uint32_t gfxhub_v3_0_get_invalidate_req(unsigned int vmid,
+					       uint32_t flush_type)
+{
+	u32 req = 0;
+
+	/* invalidate using legacy mode on vmid*/
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ,
+			    PER_VMID_INVALIDATE_REQ, 1 << vmid);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, FLUSH_TYPE, flush_type);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PTES, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE0, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE1, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE2, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L1_PTES, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ,
+			    CLEAR_PROTECTION_FAULT_STATUS_ADDR,	0);
+
+	return req;
+}
+
+static void
+gfxhub_v3_0_print_l2_protection_fault_status(struct amdgpu_device *adev,
+					     uint32_t status)
+{
+	u32 cid = REG_GET_FIELD(status,
+				GCVM_L2_PROTECTION_FAULT_STATUS, CID);
+
+	dev_err(adev->dev,
+		"GCVM_L2_PROTECTION_FAULT_STATUS:0x%08X\n",
+		status);
+	dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n",
+		cid >= ARRAY_SIZE(gfxhub_client_ids) ? "unknown" : gfxhub_client_ids[cid],
+		cid);
+	dev_err(adev->dev, "\t MORE_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, MORE_FAULTS));
+	dev_err(adev->dev, "\t WALKER_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, WALKER_ERROR));
+	dev_err(adev->dev, "\t PERMISSION_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, PERMISSION_FAULTS));
+	dev_err(adev->dev, "\t MAPPING_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, MAPPING_ERROR));
+	dev_err(adev->dev, "\t RW: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, RW));
+}
+
+static u64 gfxhub_v3_0_get_fb_location(struct amdgpu_device *adev)
+{
+	u64 base = RREG32_SOC15(GC, 0, regGCMC_VM_FB_LOCATION_BASE);
+
+	base &= GCMC_VM_FB_LOCATION_BASE__FB_BASE_MASK;
+	base <<= 24;
+
+	return base;
+}
+
+static u64 gfxhub_v3_0_get_mc_fb_offset(struct amdgpu_device *adev)
+{
+	return (u64)RREG32_SOC15(GC, 0, regGCMC_VM_FB_OFFSET) << 24;
+}
+
+static void gfxhub_v3_0_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid,
+				uint64_t page_table_base)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+
+	WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+			    hub->ctx_addr_distance * vmid,
+			    lower_32_bits(page_table_base));
+
+	WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+			    hub->ctx_addr_distance * vmid,
+			    upper_32_bits(page_table_base));
+}
+
+static void gfxhub_v3_0_init_gart_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t pt_base = amdgpu_gmc_pd_addr(adev->gart.bo);
+
+	gfxhub_v3_0_setup_vm_pt_regs(adev, 0, pt_base);
+
+	WREG32_SOC15(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+		     (u32)(adev->gmc.gart_start >> 12));
+	WREG32_SOC15(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+		     (u32)(adev->gmc.gart_start >> 44));
+
+	WREG32_SOC15(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+		     (u32)(adev->gmc.gart_end >> 12));
+	WREG32_SOC15(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+		     (u32)(adev->gmc.gart_end >> 44));
+}
+
+static void gfxhub_v3_0_init_system_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t value;
+
+	/* Program the AGP BAR */
+	WREG32_SOC15(GC, 0, regGCMC_VM_AGP_BASE, 0);
+	WREG32_SOC15(GC, 0, regGCMC_VM_AGP_BOT, adev->gmc.agp_start >> 24);
+	WREG32_SOC15(GC, 0, regGCMC_VM_AGP_TOP, adev->gmc.agp_end >> 24);
+
+
+	/* Program the system aperture low logical page number. */
+	WREG32_SOC15(GC, 0, regGCMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+		     min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
+	WREG32_SOC15(GC, 0, regGCMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+		     max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
+
+	/* Set default page address. */
+	value = amdgpu_gmc_vram_mc2pa(adev, adev->mem_scratch.gpu_addr);
+	WREG32_SOC15(GC, 0, regGCMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
+		     (u32)(value >> 12));
+	WREG32_SOC15(GC, 0, regGCMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
+		     (u32)(value >> 44));
+
+	/* Program "protection fault". */
+	WREG32_SOC15(GC, 0, regGCVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
+		     (u32)(adev->dummy_page_addr >> 12));
+	WREG32_SOC15(GC, 0, regGCVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+		     (u32)((u64)adev->dummy_page_addr >> 44));
+
+	WREG32_FIELD15_PREREG(GC, 0, GCVM_L2_PROTECTION_FAULT_CNTL2,
+		       ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+}
+
+
+static void gfxhub_v3_0_init_tlb_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 1);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, ECO_BITS, 0);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    MTYPE, MTYPE_UC); /* UC, uncached */
+
+	WREG32_SOC15(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL, tmp);
+}
+
+static void gfxhub_v3_0_init_cache_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(GC, 0, regGCVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, ENABLE_L2_CACHE, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL,
+			    ENABLE_DEFAULT_PAGE_OUT_TO_SYSTEM_MEMORY, 1);
+	/* XXX for emulation, Refer to closed source code.*/
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL,
+			    L2_PDE0_CACHE_TAG_GENERATION_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL, tmp);
+
+	tmp = RREG32_SOC15(GC, 0, regGCVM_L2_CNTL2);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL2, tmp);
+
+	tmp = regGCVM_L2_CNTL3_DEFAULT;
+	if (adev->gmc.translate_further) {
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3, BANK_SELECT, 12);
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+	} else {
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3, BANK_SELECT, 9);
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+	}
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL3, tmp);
+
+	tmp = regGCVM_L2_CNTL4_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL4, tmp);
+
+	tmp = regGCVM_L2_CNTL5_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL5, L2_CACHE_SMALLK_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL5, tmp);
+}
+
+static void gfxhub_v3_0_enable_system_domain(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	tmp = RREG32_SOC15(GC, 0, regGCVM_CONTEXT0_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT0_CNTL,
+			    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
+	WREG32_SOC15(GC, 0, regGCVM_CONTEXT0_CNTL, tmp);
+}
+
+static void gfxhub_v3_0_disable_identity_aperture(struct amdgpu_device *adev)
+{
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32,
+		     0xFFFFFFFF);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32,
+		     0x0000000F);
+
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32,
+		     0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32,
+		     0);
+
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32, 0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32, 0);
+
+}
+
+static void gfxhub_v3_0_setup_vmid_config(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	int i;
+	uint32_t tmp;
+
+	for (i = 0; i <= 14; i++) {
+		tmp = RREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_CNTL, i * hub->ctx_distance);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH,
+				    adev->vm_manager.num_level);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				PAGE_TABLE_BLOCK_SIZE,
+				adev->vm_manager.block_size - 9);
+		/* Send no-retry XNACK on fault to suppress VM fault storm. */
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
+				    !amdgpu_noretry);
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_CNTL,
+				    i * hub->ctx_distance, tmp);
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32,
+				    i * hub->ctx_addr_distance,
+				    lower_32_bits(adev->vm_manager.max_pfn - 1));
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32,
+				    i * hub->ctx_addr_distance,
+				    upper_32_bits(adev->vm_manager.max_pfn - 1));
+	}
+
+	hub->vm_cntx_cntl = tmp;
+}
+
+static void gfxhub_v3_0_program_invalidation(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	unsigned int i;
+
+	for (i = 0 ; i < 18; ++i) {
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
+				    i * hub->eng_addr_distance, 0xffffffff);
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
+				    i * hub->eng_addr_distance, 0x1f);
+	}
+}
+
+static int gfxhub_v3_0_gart_enable(struct amdgpu_device *adev)
+{
+	if (amdgpu_sriov_vf(adev)) {
+		/*
+		 * GCMC_VM_FB_LOCATION_BASE/TOP is NULL for VF, becuase they are
+		 * VF copy registers so vbios post doesn't program them, for
+		 * SRIOV driver need to program them
+		 */
+		WREG32_SOC15(GC, 0, regGCMC_VM_FB_LOCATION_BASE,
+			     adev->gmc.vram_start >> 24);
+		WREG32_SOC15(GC, 0, regGCMC_VM_FB_LOCATION_TOP,
+			     adev->gmc.vram_end >> 24);
+	}
+
+	/* GART Enable. */
+	gfxhub_v3_0_init_gart_aperture_regs(adev);
+	gfxhub_v3_0_init_system_aperture_regs(adev);
+	gfxhub_v3_0_init_tlb_regs(adev);
+	gfxhub_v3_0_init_cache_regs(adev);
+
+	gfxhub_v3_0_enable_system_domain(adev);
+	gfxhub_v3_0_disable_identity_aperture(adev);
+	gfxhub_v3_0_setup_vmid_config(adev);
+	gfxhub_v3_0_program_invalidation(adev);
+
+	return 0;
+}
+
+static void gfxhub_v3_0_gart_disable(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	u32 tmp;
+	u32 i;
+
+	/* Disable all tables */
+	for (i = 0; i < 16; i++)
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT0_CNTL,
+				    i * hub->ctx_distance, 0);
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 0);
+	WREG32_SOC15(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL, tmp);
+
+	/* Setup L2 cache */
+	WREG32_FIELD15_PREREG(GC, 0, GCVM_L2_CNTL, ENABLE_L2_CACHE, 0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL3, 0);
+}
+
+/**
+ * gfxhub_v3_0_set_fault_enable_default - update GART/VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+static void gfxhub_v3_0_set_fault_enable_default(struct amdgpu_device *adev,
+					  bool value)
+{
+	u32 tmp;
+
+	/* NO halt CP when page fault */
+	tmp = RREG32_SOC15(GC, 0, regCP_DEBUG);
+	tmp = REG_SET_FIELD(tmp, CP_DEBUG, CPG_UTCL1_ERROR_HALT_DISABLE, 1);
+	WREG32_SOC15(GC, 0, regCP_DEBUG, tmp);
+
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	tmp = RREG32_SOC15(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE1_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE2_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT,
+			    value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    NACK_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	if (!value) {
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_NO_RETRY_FAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_RETRY_FAULT, 1);
+	}
+	WREG32_SOC15(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL, tmp);
+}
+
+static const struct amdgpu_vmhub_funcs gfxhub_v3_0_vmhub_funcs = {
+	.print_l2_protection_fault_status = gfxhub_v3_0_print_l2_protection_fault_status,
+	.get_invalidate_req = gfxhub_v3_0_get_invalidate_req,
+};
+
+static void gfxhub_v3_0_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+
+	hub->ctx0_ptb_addr_lo32 =
+		SOC15_REG_OFFSET(GC, 0,
+				 regGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32);
+	hub->ctx0_ptb_addr_hi32 =
+		SOC15_REG_OFFSET(GC, 0,
+				 regGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32);
+	hub->vm_inv_eng0_sem =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_INVALIDATE_ENG0_SEM);
+	hub->vm_inv_eng0_req =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_INVALIDATE_ENG0_REQ);
+	hub->vm_inv_eng0_ack =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_INVALIDATE_ENG0_ACK);
+	hub->vm_context0_cntl =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_CONTEXT0_CNTL);
+	hub->vm_l2_pro_fault_status =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_L2_PROTECTION_FAULT_STATUS);
+	hub->vm_l2_pro_fault_cntl =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL);
+
+	hub->ctx_distance = regGCVM_CONTEXT1_CNTL - regGCVM_CONTEXT0_CNTL;
+	hub->ctx_addr_distance = regGCVM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 -
+		regGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32;
+	hub->eng_distance = regGCVM_INVALIDATE_ENG1_REQ -
+		regGCVM_INVALIDATE_ENG0_REQ;
+	hub->eng_addr_distance = regGCVM_INVALIDATE_ENG1_ADDR_RANGE_LO32 -
+		regGCVM_INVALIDATE_ENG0_ADDR_RANGE_LO32;
+
+	hub->vm_cntx_cntl_vm_fault = GCVM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__EXECUTE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK;
+
+	hub->vmhub_funcs = &gfxhub_v3_0_vmhub_funcs;
+}
+
+const struct amdgpu_gfxhub_funcs gfxhub_v3_0_funcs = {
+	.get_fb_location = gfxhub_v3_0_get_fb_location,
+	.get_mc_fb_offset = gfxhub_v3_0_get_mc_fb_offset,
+	.setup_vm_pt_regs = gfxhub_v3_0_setup_vm_pt_regs,
+	.gart_enable = gfxhub_v3_0_gart_enable,
+	.gart_disable = gfxhub_v3_0_gart_disable,
+	.set_fault_enable_default = gfxhub_v3_0_set_fault_enable_default,
+	.init = gfxhub_v3_0_init,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v3_0.h b/drivers/gpu/drm/amd/amdgpu/gfxhub_v3_0.h
new file mode 100644
index 000000000000..ea345e4e072a
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v3_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFXHUB_V3_0_H__
+#define __GFXHUB_V3_0_H__
+
+extern const struct amdgpu_gfxhub_funcs gfxhub_v3_0_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v3_0_3.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v3_0_3.c
new file mode 100644
index 000000000000..b3ef6e71811f
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v3_0_3.c
@@ -0,0 +1,501 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "gfxhub_v3_0_3.h"
+
+#include "gc/gc_11_0_3_offset.h"
+#include "gc/gc_11_0_3_sh_mask.h"
+#include "navi10_enum.h"
+#include "soc15_common.h"
+
+#define regGCVM_L2_CNTL3_DEFAULT		0x80100007
+#define regGCVM_L2_CNTL4_DEFAULT		0x000000c1
+#define regGCVM_L2_CNTL5_DEFAULT		0x00003fe0
+
+static const char * const gfxhub_client_ids[] = {
+	"CB/DB",
+	"Reserved",
+	"GE1",
+	"GE2",
+	"CPF",
+	"CPC",
+	"CPG",
+	"RLC",
+	"TCP",
+	"SQC (inst)",
+	"SQC (data)",
+	"SQG",
+	"Reserved",
+	"SDMA0",
+	"SDMA1",
+	"GCR",
+	"SDMA2",
+	"SDMA3",
+};
+
+static uint32_t gfxhub_v3_0_3_get_invalidate_req(unsigned int vmid,
+					       uint32_t flush_type)
+{
+	u32 req = 0;
+
+	/* invalidate using legacy mode on vmid*/
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ,
+			    PER_VMID_INVALIDATE_REQ, 1 << vmid);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, FLUSH_TYPE, flush_type);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PTES, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE0, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE1, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE2, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L1_PTES, 1);
+	req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ,
+			    CLEAR_PROTECTION_FAULT_STATUS_ADDR,	0);
+
+	return req;
+}
+
+static void
+gfxhub_v3_0_3_print_l2_protection_fault_status(struct amdgpu_device *adev,
+					     uint32_t status)
+{
+	u32 cid = REG_GET_FIELD(status,
+				GCVM_L2_PROTECTION_FAULT_STATUS, CID);
+
+	dev_err(adev->dev,
+		"GCVM_L2_PROTECTION_FAULT_STATUS:0x%08X\n",
+		status);
+	dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n",
+		cid >= ARRAY_SIZE(gfxhub_client_ids) ? "unknown" : gfxhub_client_ids[cid],
+		cid);
+	dev_err(adev->dev, "\t MORE_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, MORE_FAULTS));
+	dev_err(adev->dev, "\t WALKER_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, WALKER_ERROR));
+	dev_err(adev->dev, "\t PERMISSION_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, PERMISSION_FAULTS));
+	dev_err(adev->dev, "\t MAPPING_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, MAPPING_ERROR));
+	dev_err(adev->dev, "\t RW: 0x%lx\n",
+		REG_GET_FIELD(status,
+		GCVM_L2_PROTECTION_FAULT_STATUS, RW));
+}
+
+static u64 gfxhub_v3_0_3_get_fb_location(struct amdgpu_device *adev)
+{
+	u64 base = RREG32_SOC15(GC, 0, regGCMC_VM_FB_LOCATION_BASE);
+
+	base &= GCMC_VM_FB_LOCATION_BASE__FB_BASE_MASK;
+	base <<= 24;
+
+	return base;
+}
+
+static u64 gfxhub_v3_0_3_get_mc_fb_offset(struct amdgpu_device *adev)
+{
+	return (u64)RREG32_SOC15(GC, 0, regGCMC_VM_FB_OFFSET) << 24;
+}
+
+static void gfxhub_v3_0_3_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid,
+				uint64_t page_table_base)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+
+	WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+			    hub->ctx_addr_distance * vmid,
+			    lower_32_bits(page_table_base));
+
+	WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+			    hub->ctx_addr_distance * vmid,
+			    upper_32_bits(page_table_base));
+}
+
+static void gfxhub_v3_0_3_init_gart_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t pt_base = amdgpu_gmc_pd_addr(adev->gart.bo);
+
+	gfxhub_v3_0_3_setup_vm_pt_regs(adev, 0, pt_base);
+
+	WREG32_SOC15(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+		     (u32)(adev->gmc.gart_start >> 12));
+	WREG32_SOC15(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+		     (u32)(adev->gmc.gart_start >> 44));
+
+	WREG32_SOC15(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+		     (u32)(adev->gmc.gart_end >> 12));
+	WREG32_SOC15(GC, 0, regGCVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+		     (u32)(adev->gmc.gart_end >> 44));
+}
+
+static void gfxhub_v3_0_3_init_system_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t value;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/* Disable AGP. */
+	WREG32_SOC15(GC, 0, regGCMC_VM_AGP_BASE, 0);
+	WREG32_SOC15(GC, 0, regGCMC_VM_AGP_BOT, adev->gmc.agp_start >> 24);
+	WREG32_SOC15(GC, 0, regGCMC_VM_AGP_TOP, adev->gmc.agp_end >> 24);
+
+	/* Program the system aperture low logical page number. */
+	WREG32_SOC15(GC, 0, regGCMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+		     min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
+	WREG32_SOC15(GC, 0, regGCMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+		     max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
+
+	/* Set default page address. */
+	value = amdgpu_gmc_vram_mc2pa(adev, adev->mem_scratch.gpu_addr);
+	WREG32_SOC15(GC, 0, regGCMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
+		     (u32)(value >> 12));
+	WREG32_SOC15(GC, 0, regGCMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
+		     (u32)(value >> 44));
+
+	/* Program "protection fault". */
+	WREG32_SOC15(GC, 0, regGCVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
+		     (u32)(adev->dummy_page_addr >> 12));
+	WREG32_SOC15(GC, 0, regGCVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+		     (u32)((u64)adev->dummy_page_addr >> 44));
+
+	WREG32_FIELD15_PREREG(GC, 0, GCVM_L2_PROTECTION_FAULT_CNTL2,
+		       ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+}
+
+
+static void gfxhub_v3_0_3_init_tlb_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 1);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, ECO_BITS, 0);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    MTYPE, MTYPE_UC); /* UC, uncached */
+
+	WREG32_SOC15(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL, tmp);
+}
+
+static void gfxhub_v3_0_3_init_cache_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(GC, 0, regGCVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, ENABLE_L2_CACHE, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL,
+			    ENABLE_DEFAULT_PAGE_OUT_TO_SYSTEM_MEMORY, 1);
+	/* XXX for emulation, Refer to closed source code.*/
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL,
+			    L2_PDE0_CACHE_TAG_GENERATION_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL, tmp);
+
+	tmp = RREG32_SOC15(GC, 0, regGCVM_L2_CNTL2);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL2, tmp);
+
+	tmp = regGCVM_L2_CNTL3_DEFAULT;
+	if (adev->gmc.translate_further) {
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3, BANK_SELECT, 12);
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+	} else {
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3, BANK_SELECT, 9);
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+	}
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL3, tmp);
+
+	tmp = regGCVM_L2_CNTL4_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL4, tmp);
+
+	tmp = regGCVM_L2_CNTL5_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL5, L2_CACHE_SMALLK_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL5, tmp);
+}
+
+static void gfxhub_v3_0_3_enable_system_domain(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	tmp = RREG32_SOC15(GC, 0, regGCVM_CONTEXT0_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT0_CNTL,
+			    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
+	WREG32_SOC15(GC, 0, regGCVM_CONTEXT0_CNTL, tmp);
+}
+
+static void gfxhub_v3_0_3_disable_identity_aperture(struct amdgpu_device *adev)
+{
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32,
+		     0xFFFFFFFF);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32,
+		     0x0000000F);
+
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32,
+		     0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32,
+		     0);
+
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32, 0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32, 0);
+
+}
+
+static void gfxhub_v3_0_3_setup_vmid_config(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	int i;
+	uint32_t tmp;
+
+	for (i = 0; i <= 14; i++) {
+		tmp = RREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_CNTL, i * hub->ctx_distance);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH,
+				    adev->vm_manager.num_level);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				PAGE_TABLE_BLOCK_SIZE,
+				adev->vm_manager.block_size - 9);
+		/* Send no-retry XNACK on fault to suppress VM fault storm. */
+		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
+				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
+				    !amdgpu_noretry);
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_CNTL,
+				    i * hub->ctx_distance, tmp);
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32,
+				    i * hub->ctx_addr_distance,
+				    lower_32_bits(adev->vm_manager.max_pfn - 1));
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32,
+				    i * hub->ctx_addr_distance,
+				    upper_32_bits(adev->vm_manager.max_pfn - 1));
+	}
+
+	hub->vm_cntx_cntl = tmp;
+}
+
+static void gfxhub_v3_0_3_program_invalidation(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	unsigned int i;
+
+	for (i = 0 ; i < 18; ++i) {
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
+				    i * hub->eng_addr_distance, 0xffffffff);
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
+				    i * hub->eng_addr_distance, 0x1f);
+	}
+}
+
+static int gfxhub_v3_0_3_gart_enable(struct amdgpu_device *adev)
+{
+	/* GART Enable. */
+	gfxhub_v3_0_3_init_gart_aperture_regs(adev);
+	gfxhub_v3_0_3_init_system_aperture_regs(adev);
+	gfxhub_v3_0_3_init_tlb_regs(adev);
+	gfxhub_v3_0_3_init_cache_regs(adev);
+
+	gfxhub_v3_0_3_enable_system_domain(adev);
+	gfxhub_v3_0_3_disable_identity_aperture(adev);
+	gfxhub_v3_0_3_setup_vmid_config(adev);
+	gfxhub_v3_0_3_program_invalidation(adev);
+
+	return 0;
+}
+
+static void gfxhub_v3_0_3_gart_disable(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	u32 tmp;
+	u32 i;
+
+	/* Disable all tables */
+	for (i = 0; i < 16; i++)
+		WREG32_SOC15_OFFSET(GC, 0, regGCVM_CONTEXT0_CNTL,
+				    i * hub->ctx_distance, 0);
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
+	tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 0);
+	WREG32_SOC15(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL, tmp);
+
+	/* Setup L2 cache */
+	WREG32_FIELD15_PREREG(GC, 0, GCVM_L2_CNTL, ENABLE_L2_CACHE, 0);
+	WREG32_SOC15(GC, 0, regGCVM_L2_CNTL3, 0);
+}
+
+/**
+ * gfxhub_v3_0_3_set_fault_enable_default - update GART/VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+static void gfxhub_v3_0_3_set_fault_enable_default(struct amdgpu_device *adev,
+					  bool value)
+{
+	u32 tmp;
+
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	tmp = RREG32_SOC15(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE1_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE2_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT,
+			    value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    NACK_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+			    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	if (!value) {
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_NO_RETRY_FAULT, 1);
+		tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_RETRY_FAULT, 1);
+	}
+	WREG32_SOC15(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL, tmp);
+}
+
+static const struct amdgpu_vmhub_funcs gfxhub_v3_0_3_vmhub_funcs = {
+	.print_l2_protection_fault_status = gfxhub_v3_0_3_print_l2_protection_fault_status,
+	.get_invalidate_req = gfxhub_v3_0_3_get_invalidate_req,
+};
+
+static void gfxhub_v3_0_3_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+
+	hub->ctx0_ptb_addr_lo32 =
+		SOC15_REG_OFFSET(GC, 0,
+				 regGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32);
+	hub->ctx0_ptb_addr_hi32 =
+		SOC15_REG_OFFSET(GC, 0,
+				 regGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32);
+	hub->vm_inv_eng0_sem =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_INVALIDATE_ENG0_SEM);
+	hub->vm_inv_eng0_req =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_INVALIDATE_ENG0_REQ);
+	hub->vm_inv_eng0_ack =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_INVALIDATE_ENG0_ACK);
+	hub->vm_context0_cntl =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_CONTEXT0_CNTL);
+	hub->vm_l2_pro_fault_status =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_L2_PROTECTION_FAULT_STATUS);
+	hub->vm_l2_pro_fault_cntl =
+		SOC15_REG_OFFSET(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL);
+
+	hub->ctx_distance = regGCVM_CONTEXT1_CNTL - regGCVM_CONTEXT0_CNTL;
+	hub->ctx_addr_distance = regGCVM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 -
+		regGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32;
+	hub->eng_distance = regGCVM_INVALIDATE_ENG1_REQ -
+		regGCVM_INVALIDATE_ENG0_REQ;
+	hub->eng_addr_distance = regGCVM_INVALIDATE_ENG1_ADDR_RANGE_LO32 -
+		regGCVM_INVALIDATE_ENG0_ADDR_RANGE_LO32;
+
+	hub->vm_cntx_cntl_vm_fault = GCVM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		GCVM_CONTEXT1_CNTL__EXECUTE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK;
+
+	hub->vmhub_funcs = &gfxhub_v3_0_3_vmhub_funcs;
+}
+
+const struct amdgpu_gfxhub_funcs gfxhub_v3_0_3_funcs = {
+	.get_fb_location = gfxhub_v3_0_3_get_fb_location,
+	.get_mc_fb_offset = gfxhub_v3_0_3_get_mc_fb_offset,
+	.setup_vm_pt_regs = gfxhub_v3_0_3_setup_vm_pt_regs,
+	.gart_enable = gfxhub_v3_0_3_gart_enable,
+	.gart_disable = gfxhub_v3_0_3_gart_disable,
+	.set_fault_enable_default = gfxhub_v3_0_3_set_fault_enable_default,
+	.init = gfxhub_v3_0_3_init,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v3_0_3.h b/drivers/gpu/drm/amd/amdgpu/gfxhub_v3_0_3.h
new file mode 100644
index 000000000000..6153bd5e3083
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v3_0_3.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFXHUB_V3_0_3_H__
+#define __GFXHUB_V3_0_3_H__
+
+extern const struct amdgpu_gfxhub_funcs gfxhub_v3_0_3_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
new file mode 100644
index 000000000000..d7499be8c4bf
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
@@ -0,0 +1,1154 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/firmware.h>
+#include <linux/pci.h>
+
+#include <drm/drm_cache.h>
+
+#include "amdgpu.h"
+#include "amdgpu_atomfirmware.h"
+#include "gmc_v10_0.h"
+#include "umc_v8_7.h"
+
+#include "athub/athub_2_0_0_sh_mask.h"
+#include "athub/athub_2_0_0_offset.h"
+#include "dcn/dcn_2_0_0_offset.h"
+#include "dcn/dcn_2_0_0_sh_mask.h"
+#include "oss/osssys_5_0_0_offset.h"
+#include "ivsrcid/vmc/irqsrcs_vmc_1_0.h"
+#include "navi10_enum.h"
+
+#include "soc15.h"
+#include "soc15d.h"
+#include "soc15_common.h"
+
+#include "nbio_v2_3.h"
+
+#include "gfxhub_v2_0.h"
+#include "gfxhub_v2_1.h"
+#include "mmhub_v2_0.h"
+#include "mmhub_v2_3.h"
+#include "athub_v2_0.h"
+#include "athub_v2_1.h"
+
+static int gmc_v10_0_ecc_interrupt_state(struct amdgpu_device *adev,
+					 struct amdgpu_irq_src *src,
+					 unsigned int type,
+					 enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+static int
+gmc_v10_0_vm_fault_interrupt_state(struct amdgpu_device *adev,
+				   struct amdgpu_irq_src *src, unsigned int type,
+				   enum amdgpu_interrupt_state state)
+{
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		/* MM HUB */
+		amdgpu_gmc_set_vm_fault_masks(adev, AMDGPU_MMHUB0(0), false);
+		/* GFX HUB */
+		/* This works because this interrupt is only
+		 * enabled at init/resume and disabled in
+		 * fini/suspend, so the overall state doesn't
+		 * change over the course of suspend/resume.
+		 */
+		if (!adev->in_s0ix)
+			amdgpu_gmc_set_vm_fault_masks(adev, AMDGPU_GFXHUB(0), false);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		/* MM HUB */
+		amdgpu_gmc_set_vm_fault_masks(adev, AMDGPU_MMHUB0(0), true);
+		/* GFX HUB */
+		/* This works because this interrupt is only
+		 * enabled at init/resume and disabled in
+		 * fini/suspend, so the overall state doesn't
+		 * change over the course of suspend/resume.
+		 */
+		if (!adev->in_s0ix)
+			amdgpu_gmc_set_vm_fault_masks(adev, AMDGPU_GFXHUB(0), true);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gmc_v10_0_process_interrupt(struct amdgpu_device *adev,
+				       struct amdgpu_irq_src *source,
+				       struct amdgpu_iv_entry *entry)
+{
+	uint32_t vmhub_index = entry->client_id == SOC15_IH_CLIENTID_VMC ?
+			       AMDGPU_MMHUB0(0) : AMDGPU_GFXHUB(0);
+	struct amdgpu_vmhub *hub = &adev->vmhub[vmhub_index];
+	bool retry_fault = !!(entry->src_data[1] & 0x80);
+	bool write_fault = !!(entry->src_data[1] & 0x20);
+	struct amdgpu_task_info *task_info;
+	uint32_t status = 0;
+	u64 addr;
+
+	addr = (u64)entry->src_data[0] << 12;
+	addr |= ((u64)entry->src_data[1] & 0xf) << 44;
+
+	if (retry_fault) {
+		/* Returning 1 here also prevents sending the IV to the KFD */
+
+		/* Process it onyl if it's the first fault for this address */
+		if (entry->ih != &adev->irq.ih_soft &&
+		    amdgpu_gmc_filter_faults(adev, entry->ih, addr, entry->pasid,
+					     entry->timestamp))
+			return 1;
+
+		/* Delegate it to a different ring if the hardware hasn't
+		 * already done it.
+		 */
+		if (entry->ih == &adev->irq.ih) {
+			amdgpu_irq_delegate(adev, entry, 8);
+			return 1;
+		}
+
+		/* Try to handle the recoverable page faults by filling page
+		 * tables
+		 */
+		if (amdgpu_vm_handle_fault(adev, entry->pasid, 0, 0, addr,
+					   entry->timestamp, write_fault))
+			return 1;
+	}
+
+	if (!amdgpu_sriov_vf(adev)) {
+		/*
+		 * Issue a dummy read to wait for the status register to
+		 * be updated to avoid reading an incorrect value due to
+		 * the new fast GRBM interface.
+		 */
+		if ((entry->vmid_src == AMDGPU_GFXHUB(0)) &&
+		    (amdgpu_ip_version(adev, GC_HWIP, 0) <
+		     IP_VERSION(10, 3, 0)))
+			RREG32(hub->vm_l2_pro_fault_status);
+
+		status = RREG32(hub->vm_l2_pro_fault_status);
+		WREG32_P(hub->vm_l2_pro_fault_cntl, 1, ~1);
+
+		amdgpu_vm_update_fault_cache(adev, entry->pasid, addr, status,
+					     entry->vmid_src ? AMDGPU_MMHUB0(0) : AMDGPU_GFXHUB(0));
+	}
+
+	if (!printk_ratelimit())
+		return 0;
+
+	dev_err(adev->dev,
+		"[%s] page fault (src_id:%u ring:%u vmid:%u pasid:%u)\n",
+		entry->vmid_src ? "mmhub" : "gfxhub",
+		entry->src_id, entry->ring_id, entry->vmid, entry->pasid);
+	task_info = amdgpu_vm_get_task_info_pasid(adev, entry->pasid);
+	if (task_info) {
+		amdgpu_vm_print_task_info(adev, task_info);
+		amdgpu_vm_put_task_info(task_info);
+	}
+
+	dev_err(adev->dev, "  in page starting at address 0x%016llx from client 0x%x (%s)\n",
+			addr, entry->client_id,
+			soc15_ih_clientid_name[entry->client_id]);
+
+	/* Only print L2 fault status if the status register could be read and
+	 * contains useful information
+	 */
+	if (status != 0)
+		hub->vmhub_funcs->print_l2_protection_fault_status(adev,
+								   status);
+
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs gmc_v10_0_irq_funcs = {
+	.set = gmc_v10_0_vm_fault_interrupt_state,
+	.process = gmc_v10_0_process_interrupt,
+};
+
+static const struct amdgpu_irq_src_funcs gmc_v10_0_ecc_funcs = {
+	.set = gmc_v10_0_ecc_interrupt_state,
+	.process = amdgpu_umc_process_ecc_irq,
+};
+
+static void gmc_v10_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->gmc.vm_fault.num_types = 1;
+	adev->gmc.vm_fault.funcs = &gmc_v10_0_irq_funcs;
+
+	if (!amdgpu_sriov_vf(adev)) {
+		adev->gmc.ecc_irq.num_types = 1;
+		adev->gmc.ecc_irq.funcs = &gmc_v10_0_ecc_funcs;
+	}
+}
+
+/**
+ * gmc_v10_0_use_invalidate_semaphore - judge whether to use semaphore
+ *
+ * @adev: amdgpu_device pointer
+ * @vmhub: vmhub type
+ *
+ */
+static bool gmc_v10_0_use_invalidate_semaphore(struct amdgpu_device *adev,
+				       uint32_t vmhub)
+{
+	return ((vmhub == AMDGPU_MMHUB0(0)) &&
+		(!amdgpu_sriov_vf(adev)));
+}
+
+static bool gmc_v10_0_get_atc_vmid_pasid_mapping_info(
+					struct amdgpu_device *adev,
+					uint8_t vmid, uint16_t *p_pasid)
+{
+	uint32_t value;
+
+	value = RREG32(SOC15_REG_OFFSET(ATHUB, 0, mmATC_VMID0_PASID_MAPPING)
+		     + vmid);
+	*p_pasid = value & ATC_VMID0_PASID_MAPPING__PASID_MASK;
+
+	return !!(value & ATC_VMID0_PASID_MAPPING__VALID_MASK);
+}
+
+/*
+ * GART
+ * VMID 0 is the physical GPU addresses as used by the kernel.
+ * VMIDs 1-15 are used for userspace clients and are handled
+ * by the amdgpu vm/hsa code.
+ */
+
+/**
+ * gmc_v10_0_flush_gpu_tlb - gart tlb flush callback
+ *
+ * @adev: amdgpu_device pointer
+ * @vmid: vm instance to flush
+ * @vmhub: vmhub type
+ * @flush_type: the flush type
+ *
+ * Flush the TLB for the requested page table.
+ */
+static void gmc_v10_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid,
+					uint32_t vmhub, uint32_t flush_type)
+{
+	bool use_semaphore = gmc_v10_0_use_invalidate_semaphore(adev, vmhub);
+	struct amdgpu_vmhub *hub = &adev->vmhub[vmhub];
+	u32 inv_req = hub->vmhub_funcs->get_invalidate_req(vmid, flush_type);
+	/* Use register 17 for GART */
+	const unsigned int eng = 17;
+	unsigned char hub_ip = 0;
+	u32 sem, req, ack;
+	unsigned int i;
+	u32 tmp;
+
+	sem = hub->vm_inv_eng0_sem + hub->eng_distance * eng;
+	req = hub->vm_inv_eng0_req + hub->eng_distance * eng;
+	ack = hub->vm_inv_eng0_ack + hub->eng_distance * eng;
+
+	/* flush hdp cache */
+	amdgpu_device_flush_hdp(adev, NULL);
+
+	/* This is necessary for SRIOV as well as for GFXOFF to function
+	 * properly under bare metal
+	 */
+	if (adev->gfx.kiq[0].ring.sched.ready && !adev->enable_mes &&
+	    (amdgpu_sriov_runtime(adev) || !amdgpu_sriov_vf(adev))) {
+		amdgpu_gmc_fw_reg_write_reg_wait(adev, req, ack, inv_req,
+						 1 << vmid, GET_INST(GC, 0));
+		return;
+	}
+
+	/* This path is needed before KIQ/MES/GFXOFF are set up */
+	hub_ip = (vmhub == AMDGPU_GFXHUB(0)) ? GC_HWIP : MMHUB_HWIP;
+
+	spin_lock(&adev->gmc.invalidate_lock);
+	/*
+	 * It may lose gpuvm invalidate acknowldege state across power-gating
+	 * off cycle, add semaphore acquire before invalidation and semaphore
+	 * release after invalidation to avoid entering power gated state
+	 * to WA the Issue
+	 */
+
+	/* TODO: It needs to continue working on debugging with semaphore for GFXHUB as well. */
+	if (use_semaphore) {
+		for (i = 0; i < adev->usec_timeout; i++) {
+			/* a read return value of 1 means semaphore acuqire */
+			tmp = RREG32_RLC_NO_KIQ(sem, hub_ip);
+			if (tmp & 0x1)
+				break;
+			udelay(1);
+		}
+
+		if (i >= adev->usec_timeout)
+			DRM_ERROR("Timeout waiting for sem acquire in VM flush!\n");
+	}
+
+	WREG32_RLC_NO_KIQ(req, inv_req, hub_ip);
+
+	/*
+	 * Issue a dummy read to wait for the ACK register to be cleared
+	 * to avoid a false ACK due to the new fast GRBM interface.
+	 */
+	if ((vmhub == AMDGPU_GFXHUB(0)) &&
+	    (amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(10, 3, 0)))
+		RREG32_RLC_NO_KIQ(req, hub_ip);
+
+	/* Wait for ACK with a delay.*/
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32_RLC_NO_KIQ(ack, hub_ip);
+		tmp &= 1 << vmid;
+		if (tmp)
+			break;
+
+		udelay(1);
+	}
+
+	/* TODO: It needs to continue working on debugging with semaphore for GFXHUB as well. */
+	if (use_semaphore)
+		WREG32_RLC_NO_KIQ(sem, 0, hub_ip);
+
+	spin_unlock(&adev->gmc.invalidate_lock);
+
+	if (i >= adev->usec_timeout)
+		dev_err(adev->dev, "Timeout waiting for VM flush hub: %d!\n",
+			vmhub);
+}
+
+/**
+ * gmc_v10_0_flush_gpu_tlb_pasid - tlb flush via pasid
+ *
+ * @adev: amdgpu_device pointer
+ * @pasid: pasid to be flush
+ * @flush_type: the flush type
+ * @all_hub: Used with PACKET3_INVALIDATE_TLBS_ALL_HUB()
+ * @inst: is used to select which instance of KIQ to use for the invalidation
+ *
+ * Flush the TLB for the requested pasid.
+ */
+static void gmc_v10_0_flush_gpu_tlb_pasid(struct amdgpu_device *adev,
+					  uint16_t pasid, uint32_t flush_type,
+					  bool all_hub, uint32_t inst)
+{
+	uint16_t queried;
+	int vmid, i;
+
+	for (vmid = 1; vmid < AMDGPU_NUM_VMID; vmid++) {
+		bool valid;
+
+		valid = gmc_v10_0_get_atc_vmid_pasid_mapping_info(adev, vmid,
+								  &queried);
+		if (!valid || queried != pasid)
+			continue;
+
+		if (all_hub) {
+			for_each_set_bit(i, adev->vmhubs_mask,
+					 AMDGPU_MAX_VMHUBS)
+				gmc_v10_0_flush_gpu_tlb(adev, vmid, i,
+							flush_type);
+		} else {
+			gmc_v10_0_flush_gpu_tlb(adev, vmid, AMDGPU_GFXHUB(0),
+						flush_type);
+		}
+	}
+}
+
+static uint64_t gmc_v10_0_emit_flush_gpu_tlb(struct amdgpu_ring *ring,
+					     unsigned int vmid, uint64_t pd_addr)
+{
+	bool use_semaphore = gmc_v10_0_use_invalidate_semaphore(ring->adev, ring->vm_hub);
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
+	uint32_t req = hub->vmhub_funcs->get_invalidate_req(vmid, 0);
+	unsigned int eng = ring->vm_inv_eng;
+
+	/*
+	 * It may lose gpuvm invalidate acknowldege state across power-gating
+	 * off cycle, add semaphore acquire before invalidation and semaphore
+	 * release after invalidation to avoid entering power gated state
+	 * to WA the Issue
+	 */
+
+	/* TODO: It needs to continue working on debugging with semaphore for GFXHUB as well. */
+	if (use_semaphore)
+		/* a read return value of 1 means semaphore acuqire */
+		amdgpu_ring_emit_reg_wait(ring,
+					  hub->vm_inv_eng0_sem +
+					  hub->eng_distance * eng, 0x1, 0x1);
+
+	amdgpu_ring_emit_wreg(ring, hub->ctx0_ptb_addr_lo32 +
+			      (hub->ctx_addr_distance * vmid),
+			      lower_32_bits(pd_addr));
+
+	amdgpu_ring_emit_wreg(ring, hub->ctx0_ptb_addr_hi32 +
+			      (hub->ctx_addr_distance * vmid),
+			      upper_32_bits(pd_addr));
+
+	amdgpu_ring_emit_reg_write_reg_wait(ring, hub->vm_inv_eng0_req +
+					    hub->eng_distance * eng,
+					    hub->vm_inv_eng0_ack +
+					    hub->eng_distance * eng,
+					    req, 1 << vmid);
+
+	/* TODO: It needs to continue working on debugging with semaphore for GFXHUB as well. */
+	if (use_semaphore)
+		/*
+		 * add semaphore release after invalidation,
+		 * write with 0 means semaphore release
+		 */
+		amdgpu_ring_emit_wreg(ring, hub->vm_inv_eng0_sem +
+				      hub->eng_distance * eng, 0);
+
+	return pd_addr;
+}
+
+static void gmc_v10_0_emit_pasid_mapping(struct amdgpu_ring *ring, unsigned int vmid,
+					 unsigned int pasid)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t reg;
+
+	if (ring->vm_hub == AMDGPU_GFXHUB(0))
+		reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_VMID_0_LUT) + vmid;
+	else
+		reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_VMID_0_LUT_MM) + vmid;
+
+	amdgpu_ring_emit_wreg(ring, reg, pasid);
+}
+
+/*
+ * PTE format on NAVI 10:
+ * 63:59 reserved
+ * 58 reserved and for sienna_cichlid is used for MALL noalloc
+ * 57 reserved
+ * 56 F
+ * 55 L
+ * 54 reserved
+ * 53:52 SW
+ * 51 T
+ * 50:48 mtype
+ * 47:12 4k physical page base address
+ * 11:7 fragment
+ * 6 write
+ * 5 read
+ * 4 exe
+ * 3 Z
+ * 2 snooped
+ * 1 system
+ * 0 valid
+ *
+ * PDE format on NAVI 10:
+ * 63:59 block fragment size
+ * 58:55 reserved
+ * 54 P
+ * 53:48 reserved
+ * 47:6 physical base address of PD or PTE
+ * 5:3 reserved
+ * 2 C
+ * 1 system
+ * 0 valid
+ */
+
+static void gmc_v10_0_get_vm_pde(struct amdgpu_device *adev, int level,
+				 uint64_t *addr, uint64_t *flags)
+{
+	if (!(*flags & AMDGPU_PDE_PTE) && !(*flags & AMDGPU_PTE_SYSTEM))
+		*addr = amdgpu_gmc_vram_mc2pa(adev, *addr);
+	BUG_ON(*addr & 0xFFFF00000000003FULL);
+
+	if (!adev->gmc.translate_further)
+		return;
+
+	if (level == AMDGPU_VM_PDB1) {
+		/* Set the block fragment size */
+		if (!(*flags & AMDGPU_PDE_PTE))
+			*flags |= AMDGPU_PDE_BFS(0x9);
+
+	} else if (level == AMDGPU_VM_PDB0) {
+		if (*flags & AMDGPU_PDE_PTE)
+			*flags &= ~AMDGPU_PDE_PTE;
+		else
+			*flags |= AMDGPU_PTE_TF;
+	}
+}
+
+static void gmc_v10_0_get_vm_pte(struct amdgpu_device *adev,
+				 struct amdgpu_vm *vm,
+				 struct amdgpu_bo *bo,
+				 uint32_t vm_flags,
+				 uint64_t *flags)
+{
+	if (vm_flags & AMDGPU_VM_PAGE_EXECUTABLE)
+		*flags |= AMDGPU_PTE_EXECUTABLE;
+	else
+		*flags &= ~AMDGPU_PTE_EXECUTABLE;
+
+	switch (vm_flags & AMDGPU_VM_MTYPE_MASK) {
+	case AMDGPU_VM_MTYPE_DEFAULT:
+	case AMDGPU_VM_MTYPE_NC:
+	default:
+		*flags = AMDGPU_PTE_MTYPE_NV10(*flags, MTYPE_NC);
+		break;
+	case AMDGPU_VM_MTYPE_WC:
+		*flags = AMDGPU_PTE_MTYPE_NV10(*flags, MTYPE_WC);
+		break;
+	case AMDGPU_VM_MTYPE_CC:
+		*flags = AMDGPU_PTE_MTYPE_NV10(*flags, MTYPE_CC);
+		break;
+	case AMDGPU_VM_MTYPE_UC:
+		*flags = AMDGPU_PTE_MTYPE_NV10(*flags, MTYPE_UC);
+		break;
+	}
+
+	if (vm_flags & AMDGPU_VM_PAGE_NOALLOC)
+		*flags |= AMDGPU_PTE_NOALLOC;
+	else
+		*flags &= ~AMDGPU_PTE_NOALLOC;
+
+	if (vm_flags & AMDGPU_VM_PAGE_PRT) {
+		*flags |= AMDGPU_PTE_PRT;
+		*flags |= AMDGPU_PTE_SNOOPED;
+		*flags |= AMDGPU_PTE_LOG;
+		*flags |= AMDGPU_PTE_SYSTEM;
+		*flags &= ~AMDGPU_PTE_VALID;
+	}
+
+	if (bo && bo->flags & (AMDGPU_GEM_CREATE_COHERENT |
+			       AMDGPU_GEM_CREATE_EXT_COHERENT |
+			       AMDGPU_GEM_CREATE_UNCACHED))
+		*flags = AMDGPU_PTE_MTYPE_NV10(*flags, MTYPE_UC);
+}
+
+static unsigned int gmc_v10_0_get_vbios_fb_size(struct amdgpu_device *adev)
+{
+	u32 d1vga_control = RREG32_SOC15(DCE, 0, mmD1VGA_CONTROL);
+	unsigned int size;
+
+	if (REG_GET_FIELD(d1vga_control, D1VGA_CONTROL, D1VGA_MODE_ENABLE)) {
+		size = AMDGPU_VBIOS_VGA_ALLOCATION;
+	} else {
+		u32 viewport;
+		u32 pitch;
+
+		viewport = RREG32_SOC15(DCE, 0, mmHUBP0_DCSURF_PRI_VIEWPORT_DIMENSION);
+		pitch = RREG32_SOC15(DCE, 0, mmHUBPREQ0_DCSURF_SURFACE_PITCH);
+		size = (REG_GET_FIELD(viewport,
+					HUBP0_DCSURF_PRI_VIEWPORT_DIMENSION, PRI_VIEWPORT_HEIGHT) *
+				REG_GET_FIELD(pitch, HUBPREQ0_DCSURF_SURFACE_PITCH, PITCH) *
+				4);
+	}
+
+	return size;
+}
+
+static const struct amdgpu_gmc_funcs gmc_v10_0_gmc_funcs = {
+	.flush_gpu_tlb = gmc_v10_0_flush_gpu_tlb,
+	.flush_gpu_tlb_pasid = gmc_v10_0_flush_gpu_tlb_pasid,
+	.emit_flush_gpu_tlb = gmc_v10_0_emit_flush_gpu_tlb,
+	.emit_pasid_mapping = gmc_v10_0_emit_pasid_mapping,
+	.get_vm_pde = gmc_v10_0_get_vm_pde,
+	.get_vm_pte = gmc_v10_0_get_vm_pte,
+	.get_vbios_fb_size = gmc_v10_0_get_vbios_fb_size,
+};
+
+static void gmc_v10_0_set_gmc_funcs(struct amdgpu_device *adev)
+{
+	if (adev->gmc.gmc_funcs == NULL)
+		adev->gmc.gmc_funcs = &gmc_v10_0_gmc_funcs;
+}
+
+static void gmc_v10_0_set_umc_funcs(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, UMC_HWIP, 0)) {
+	case IP_VERSION(8, 7, 0):
+		adev->umc.max_ras_err_cnt_per_query = UMC_V8_7_TOTAL_CHANNEL_NUM;
+		adev->umc.channel_inst_num = UMC_V8_7_CHANNEL_INSTANCE_NUM;
+		adev->umc.umc_inst_num = UMC_V8_7_UMC_INSTANCE_NUM;
+		adev->umc.channel_offs = UMC_V8_7_PER_CHANNEL_OFFSET_SIENNA;
+		adev->umc.retire_unit = 1;
+		adev->umc.channel_idx_tbl = &umc_v8_7_channel_idx_tbl[0][0];
+		adev->umc.ras = &umc_v8_7_ras;
+		break;
+	default:
+		break;
+	}
+}
+
+static void gmc_v10_0_set_mmhub_funcs(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) {
+	case IP_VERSION(2, 3, 0):
+	case IP_VERSION(2, 4, 0):
+	case IP_VERSION(2, 4, 1):
+		adev->mmhub.funcs = &mmhub_v2_3_funcs;
+		break;
+	default:
+		adev->mmhub.funcs = &mmhub_v2_0_funcs;
+		break;
+	}
+}
+
+static void gmc_v10_0_set_gfxhub_funcs(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+	case IP_VERSION(10, 3, 6):
+	case IP_VERSION(10, 3, 3):
+	case IP_VERSION(10, 3, 7):
+		adev->gfxhub.funcs = &gfxhub_v2_1_funcs;
+		break;
+	default:
+		adev->gfxhub.funcs = &gfxhub_v2_0_funcs;
+		break;
+	}
+}
+
+
+static int gmc_v10_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	gmc_v10_0_set_mmhub_funcs(adev);
+	gmc_v10_0_set_gfxhub_funcs(adev);
+	gmc_v10_0_set_gmc_funcs(adev);
+	gmc_v10_0_set_irq_funcs(adev);
+	gmc_v10_0_set_umc_funcs(adev);
+
+	adev->gmc.shared_aperture_start = 0x2000000000000000ULL;
+	adev->gmc.shared_aperture_end =
+		adev->gmc.shared_aperture_start + (4ULL << 30) - 1;
+	adev->gmc.private_aperture_start = 0x1000000000000000ULL;
+	adev->gmc.private_aperture_end =
+		adev->gmc.private_aperture_start + (4ULL << 30) - 1;
+	adev->gmc.noretry_flags = AMDGPU_VM_NORETRY_FLAGS_TF;
+
+	return 0;
+}
+
+static int gmc_v10_0_late_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = amdgpu_gmc_allocate_vm_inv_eng(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_gmc_ras_late_init(adev);
+	if (r)
+		return r;
+
+	return amdgpu_irq_get(adev, &adev->gmc.vm_fault, 0);
+}
+
+static void gmc_v10_0_vram_gtt_location(struct amdgpu_device *adev,
+					struct amdgpu_gmc *mc)
+{
+	u64 base = 0;
+
+	base = adev->gfxhub.funcs->get_fb_location(adev);
+
+	/* add the xgmi offset of the physical node */
+	base += adev->gmc.xgmi.physical_node_id * adev->gmc.xgmi.node_segment_size;
+
+	amdgpu_gmc_set_agp_default(adev, mc);
+	amdgpu_gmc_vram_location(adev, &adev->gmc, base);
+	amdgpu_gmc_gart_location(adev, mc, AMDGPU_GART_PLACEMENT_BEST_FIT);
+	if (!amdgpu_sriov_vf(adev) && (amdgpu_agp == 1))
+		amdgpu_gmc_agp_location(adev, mc);
+
+	/* base offset of vram pages */
+	adev->vm_manager.vram_base_offset = adev->gfxhub.funcs->get_mc_fb_offset(adev);
+
+	/* add the xgmi offset of the physical node */
+	adev->vm_manager.vram_base_offset +=
+		adev->gmc.xgmi.physical_node_id * adev->gmc.xgmi.node_segment_size;
+}
+
+/**
+ * gmc_v10_0_mc_init - initialize the memory controller driver params
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Look up the amount of vram, vram width, and decide how to place
+ * vram and gart within the GPU's physical address space.
+ * Returns 0 for success.
+ */
+static int gmc_v10_0_mc_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	/* size in MB on si */
+	adev->gmc.mc_vram_size =
+		adev->nbio.funcs->get_memsize(adev) * 1024ULL * 1024ULL;
+	adev->gmc.real_vram_size = adev->gmc.mc_vram_size;
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		r = amdgpu_device_resize_fb_bar(adev);
+		if (r)
+			return r;
+	}
+	adev->gmc.aper_base = pci_resource_start(adev->pdev, 0);
+	adev->gmc.aper_size = pci_resource_len(adev->pdev, 0);
+
+#ifdef CONFIG_X86_64
+	if ((adev->flags & AMD_IS_APU) && !amdgpu_passthrough(adev)) {
+		adev->gmc.aper_base = adev->gfxhub.funcs->get_mc_fb_offset(adev);
+		adev->gmc.aper_size = adev->gmc.real_vram_size;
+	}
+#endif
+
+	adev->gmc.visible_vram_size = adev->gmc.aper_size;
+
+	/* set the gart size */
+	if (amdgpu_gart_size == -1) {
+		switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+		default:
+			adev->gmc.gart_size = 512ULL << 20;
+			break;
+		case IP_VERSION(10, 3, 1):   /* DCE SG support */
+		case IP_VERSION(10, 3, 3):   /* DCE SG support */
+		case IP_VERSION(10, 3, 6):   /* DCE SG support */
+		case IP_VERSION(10, 3, 7):   /* DCE SG support */
+			adev->gmc.gart_size = 1024ULL << 20;
+			break;
+		}
+	} else {
+		adev->gmc.gart_size = (u64)amdgpu_gart_size << 20;
+	}
+
+	gmc_v10_0_vram_gtt_location(adev, &adev->gmc);
+
+	return 0;
+}
+
+static int gmc_v10_0_gart_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->gart.bo) {
+		WARN(1, "NAVI10 PCIE GART already initialized\n");
+		return 0;
+	}
+
+	/* Initialize common gart structure */
+	r = amdgpu_gart_init(adev);
+	if (r)
+		return r;
+
+	adev->gart.table_size = adev->gart.num_gpu_pages * 8;
+	adev->gart.gart_pte_flags = AMDGPU_PTE_MTYPE_NV10(0ULL, MTYPE_UC) |
+				 AMDGPU_PTE_EXECUTABLE;
+
+	return amdgpu_gart_table_vram_alloc(adev);
+}
+
+static int gmc_v10_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	int r, vram_width = 0, vram_type = 0, vram_vendor = 0;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	adev->gfxhub.funcs->init(adev);
+
+	adev->mmhub.funcs->init(adev);
+
+	spin_lock_init(&adev->gmc.invalidate_lock);
+
+	if ((adev->flags & AMD_IS_APU) && amdgpu_emu_mode == 1) {
+		adev->gmc.vram_type = AMDGPU_VRAM_TYPE_DDR4;
+		adev->gmc.vram_width = 64;
+	} else if (amdgpu_emu_mode == 1) {
+		adev->gmc.vram_type = AMDGPU_VRAM_TYPE_GDDR6;
+		adev->gmc.vram_width = 1 * 128; /* numchan * chansize */
+	} else {
+		r = amdgpu_atomfirmware_get_vram_info(adev,
+				&vram_width, &vram_type, &vram_vendor);
+		adev->gmc.vram_width = vram_width;
+
+		adev->gmc.vram_type = vram_type;
+		adev->gmc.vram_vendor = vram_vendor;
+	}
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 3, 0):
+		adev->gmc.mall_size = 128 * 1024 * 1024;
+		break;
+	case IP_VERSION(10, 3, 2):
+		adev->gmc.mall_size = 96 * 1024 * 1024;
+		break;
+	case IP_VERSION(10, 3, 4):
+		adev->gmc.mall_size = 32 * 1024 * 1024;
+		break;
+	case IP_VERSION(10, 3, 5):
+		adev->gmc.mall_size = 16 * 1024 * 1024;
+		break;
+	default:
+		adev->gmc.mall_size = 0;
+		break;
+	}
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 1, 10):
+	case IP_VERSION(10, 1, 1):
+	case IP_VERSION(10, 1, 2):
+	case IP_VERSION(10, 1, 3):
+	case IP_VERSION(10, 1, 4):
+	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+	case IP_VERSION(10, 3, 6):
+	case IP_VERSION(10, 3, 3):
+	case IP_VERSION(10, 3, 7):
+		set_bit(AMDGPU_GFXHUB(0), adev->vmhubs_mask);
+		set_bit(AMDGPU_MMHUB0(0), adev->vmhubs_mask);
+		/*
+		 * To fulfill 4-level page support,
+		 * vm size is 256TB (48bit), maximum size of Navi10/Navi14/Navi12,
+		 * block size 512 (9bit)
+		 */
+		amdgpu_vm_adjust_size(adev, 256 * 1024, 9, 3, 48);
+		break;
+	default:
+		break;
+	}
+
+	/* This interrupt is VMC page fault.*/
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VMC,
+			      VMC_1_0__SRCID__VM_FAULT,
+			      &adev->gmc.vm_fault);
+
+	if (r)
+		return r;
+
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_UTCL2,
+			      UTCL2_1_0__SRCID__FAULT,
+			      &adev->gmc.vm_fault);
+	if (r)
+		return r;
+
+	if (!amdgpu_sriov_vf(adev)) {
+		/* interrupt sent to DF. */
+		r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_DF, 0,
+				      &adev->gmc.ecc_irq);
+		if (r)
+			return r;
+	}
+
+	/*
+	 * Set the internal MC address mask This is the max address of the GPU's
+	 * internal address space.
+	 */
+	adev->gmc.mc_mask = 0xffffffffffffULL; /* 48 bit MC */
+
+	r = dma_set_mask_and_coherent(adev->dev, DMA_BIT_MASK(44));
+	if (r) {
+		dev_warn(adev->dev, "amdgpu: No suitable DMA available.\n");
+		return r;
+	}
+
+	adev->need_swiotlb = drm_need_swiotlb(44);
+
+	r = gmc_v10_0_mc_init(adev);
+	if (r)
+		return r;
+
+	amdgpu_gmc_get_vbios_allocations(adev);
+
+	/* Memory manager */
+	r = amdgpu_bo_init(adev);
+	if (r)
+		return r;
+
+	r = gmc_v10_0_gart_init(adev);
+	if (r)
+		return r;
+
+	/*
+	 * number of VMs
+	 * VMID 0 is reserved for System
+	 * amdgpu graphics/compute will use VMIDs 1-7
+	 * amdkfd will use VMIDs 8-15
+	 */
+	adev->vm_manager.first_kfd_vmid = 8;
+
+	amdgpu_vm_manager_init(adev);
+
+	r = amdgpu_gmc_ras_sw_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+/**
+ * gmc_v10_0_gart_fini - vm fini callback
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tears down the driver GART/VM setup (CIK).
+ */
+static void gmc_v10_0_gart_fini(struct amdgpu_device *adev)
+{
+	amdgpu_gart_table_vram_free(adev);
+}
+
+static int gmc_v10_0_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	amdgpu_vm_manager_fini(adev);
+	gmc_v10_0_gart_fini(adev);
+	amdgpu_gem_force_release(adev);
+	amdgpu_bo_fini(adev);
+
+	return 0;
+}
+
+static void gmc_v10_0_init_golden_registers(struct amdgpu_device *adev)
+{
+}
+
+/**
+ * gmc_v10_0_gart_enable - gart enable
+ *
+ * @adev: amdgpu_device pointer
+ */
+static int gmc_v10_0_gart_enable(struct amdgpu_device *adev)
+{
+	int r;
+	bool value;
+
+	if (adev->gart.bo == NULL) {
+		dev_err(adev->dev, "No VRAM object for PCIE GART.\n");
+		return -EINVAL;
+	}
+
+	amdgpu_gtt_mgr_recover(&adev->mman.gtt_mgr);
+
+	if (!adev->in_s0ix) {
+		r = adev->gfxhub.funcs->gart_enable(adev);
+		if (r)
+			return r;
+	}
+
+	r = adev->mmhub.funcs->gart_enable(adev);
+	if (r)
+		return r;
+
+	adev->hdp.funcs->init_registers(adev);
+
+	/* Flush HDP after it is initialized */
+	amdgpu_device_flush_hdp(adev, NULL);
+
+	value = amdgpu_vm_fault_stop != AMDGPU_VM_FAULT_STOP_ALWAYS;
+
+	if (!adev->in_s0ix)
+		adev->gfxhub.funcs->set_fault_enable_default(adev, value);
+	adev->mmhub.funcs->set_fault_enable_default(adev, value);
+	gmc_v10_0_flush_gpu_tlb(adev, 0, AMDGPU_MMHUB0(0), 0);
+	if (!adev->in_s0ix)
+		gmc_v10_0_flush_gpu_tlb(adev, 0, AMDGPU_GFXHUB(0), 0);
+
+	DRM_INFO("PCIE GART of %uM enabled (table at 0x%016llX).\n",
+		 (unsigned int)(adev->gmc.gart_size >> 20),
+		 (unsigned long long)amdgpu_bo_gpu_offset(adev->gart.bo));
+
+	return 0;
+}
+
+static int gmc_v10_0_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	adev->gmc.flush_pasid_uses_kiq = !amdgpu_emu_mode;
+
+	/* The sequence of these two function calls matters.*/
+	gmc_v10_0_init_golden_registers(adev);
+
+	/*
+	 * harvestable groups in gc_utcl2 need to be programmed before any GFX block
+	 * register setup within GMC, or else system hang when harvesting SA.
+	 */
+	if (!adev->in_s0ix && adev->gfxhub.funcs && adev->gfxhub.funcs->utcl2_harvest)
+		adev->gfxhub.funcs->utcl2_harvest(adev);
+
+	r = gmc_v10_0_gart_enable(adev);
+	if (r)
+		return r;
+
+	if (amdgpu_emu_mode == 1) {
+		r = amdgpu_gmc_vram_checking(adev);
+		if (r)
+			return r;
+	}
+
+	if (adev->umc.funcs && adev->umc.funcs->init_registers)
+		adev->umc.funcs->init_registers(adev);
+
+	return 0;
+}
+
+/**
+ * gmc_v10_0_gart_disable - gart disable
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * This disables all VM page table.
+ */
+static void gmc_v10_0_gart_disable(struct amdgpu_device *adev)
+{
+	if (!adev->in_s0ix)
+		adev->gfxhub.funcs->gart_disable(adev);
+	adev->mmhub.funcs->gart_disable(adev);
+}
+
+static int gmc_v10_0_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	gmc_v10_0_gart_disable(adev);
+
+	if (amdgpu_sriov_vf(adev)) {
+		/* full access mode, so don't touch any GMC register */
+		DRM_DEBUG("For SRIOV client, shouldn't do anything.\n");
+		return 0;
+	}
+
+	amdgpu_irq_put(adev, &adev->gmc.vm_fault, 0);
+
+	if (adev->gmc.ecc_irq.funcs &&
+		amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__UMC))
+		amdgpu_irq_put(adev, &adev->gmc.ecc_irq, 0);
+
+	return 0;
+}
+
+static int gmc_v10_0_suspend(struct amdgpu_ip_block *ip_block)
+{
+	gmc_v10_0_hw_fini(ip_block);
+
+	return 0;
+}
+
+static int gmc_v10_0_resume(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	r = gmc_v10_0_hw_init(ip_block);
+	if (r)
+		return r;
+
+	amdgpu_vmid_reset_all(ip_block->adev);
+
+	return 0;
+}
+
+static bool gmc_v10_0_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	/* MC is always ready in GMC v10.*/
+	return true;
+}
+
+static int gmc_v10_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	/* There is no need to wait for MC idle in GMC v10.*/
+	return 0;
+}
+
+static int gmc_v10_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					   enum amd_clockgating_state state)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	/*
+	 * The issue mmhub can't disconnect from DF with MMHUB clock gating being disabled
+	 * is a new problem observed at DF 3.0.3, however with the same suspend sequence not
+	 * seen any issue on the DF 3.0.2 series platform.
+	 */
+	if (adev->in_s0ix &&
+	    amdgpu_ip_version(adev, DF_HWIP, 0) > IP_VERSION(3, 0, 2)) {
+		dev_dbg(adev->dev, "keep mmhub clock gating being enabled for s0ix\n");
+		return 0;
+	}
+
+	r = adev->mmhub.funcs->set_clockgating(adev, state);
+	if (r)
+		return r;
+
+	if (amdgpu_ip_version(adev, ATHUB_HWIP, 0) >= IP_VERSION(2, 1, 0))
+		return athub_v2_1_set_clockgating(adev, state);
+	else
+		return athub_v2_0_set_clockgating(adev, state);
+}
+
+static void gmc_v10_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 1, 3) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 1, 4))
+		return;
+
+	adev->mmhub.funcs->get_clockgating(adev, flags);
+
+	if (amdgpu_ip_version(adev, ATHUB_HWIP, 0) >= IP_VERSION(2, 1, 0))
+		athub_v2_1_get_clockgating(adev, flags);
+	else
+		athub_v2_0_get_clockgating(adev, flags);
+}
+
+static int gmc_v10_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					   enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs gmc_v10_0_ip_funcs = {
+	.name = "gmc_v10_0",
+	.early_init = gmc_v10_0_early_init,
+	.late_init = gmc_v10_0_late_init,
+	.sw_init = gmc_v10_0_sw_init,
+	.sw_fini = gmc_v10_0_sw_fini,
+	.hw_init = gmc_v10_0_hw_init,
+	.hw_fini = gmc_v10_0_hw_fini,
+	.suspend = gmc_v10_0_suspend,
+	.resume = gmc_v10_0_resume,
+	.is_idle = gmc_v10_0_is_idle,
+	.wait_for_idle = gmc_v10_0_wait_for_idle,
+	.set_clockgating_state = gmc_v10_0_set_clockgating_state,
+	.set_powergating_state = gmc_v10_0_set_powergating_state,
+	.get_clockgating_state = gmc_v10_0_get_clockgating_state,
+};
+
+const struct amdgpu_ip_block_version gmc_v10_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_GMC,
+	.major = 10,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &gmc_v10_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.h b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.h
new file mode 100644
index 000000000000..7daa53d8996c
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GMC_V10_0_H__
+#define __GMC_V10_0_H__
+
+extern const struct amd_ip_funcs gmc_v10_0_ip_funcs;
+extern const struct amdgpu_ip_block_version gmc_v10_0_ip_block;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c
new file mode 100644
index 000000000000..7bc389d9f5c4
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c
@@ -0,0 +1,1056 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/firmware.h>
+#include <linux/pci.h>
+
+#include <drm/drm_cache.h>
+
+#include "amdgpu.h"
+#include "amdgpu_atomfirmware.h"
+#include "gmc_v11_0.h"
+#include "umc_v8_10.h"
+#include "athub/athub_3_0_0_sh_mask.h"
+#include "athub/athub_3_0_0_offset.h"
+#include "dcn/dcn_3_2_0_offset.h"
+#include "dcn/dcn_3_2_0_sh_mask.h"
+#include "oss/osssys_6_0_0_offset.h"
+#include "ivsrcid/vmc/irqsrcs_vmc_1_0.h"
+#include "navi10_enum.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "soc15_common.h"
+#include "nbio_v4_3.h"
+#include "gfxhub_v3_0.h"
+#include "gfxhub_v3_0_3.h"
+#include "gfxhub_v11_5_0.h"
+#include "mmhub_v3_0.h"
+#include "mmhub_v3_0_1.h"
+#include "mmhub_v3_0_2.h"
+#include "mmhub_v3_3.h"
+#include "athub_v3_0.h"
+
+
+static int gmc_v11_0_ecc_interrupt_state(struct amdgpu_device *adev,
+					 struct amdgpu_irq_src *src,
+					 unsigned int type,
+					 enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+static int
+gmc_v11_0_vm_fault_interrupt_state(struct amdgpu_device *adev,
+				   struct amdgpu_irq_src *src, unsigned int type,
+				   enum amdgpu_interrupt_state state)
+{
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		/* MM HUB */
+		amdgpu_gmc_set_vm_fault_masks(adev, AMDGPU_MMHUB0(0), false);
+		/* GFX HUB */
+		/* This works because this interrupt is only
+		 * enabled at init/resume and disabled in
+		 * fini/suspend, so the overall state doesn't
+		 * change over the course of suspend/resume.
+		 */
+		if (!adev->in_s0ix && (adev->in_runpm || adev->in_suspend ||
+							   amdgpu_in_reset(adev)))
+			amdgpu_gmc_set_vm_fault_masks(adev, AMDGPU_GFXHUB(0), false);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		/* MM HUB */
+		amdgpu_gmc_set_vm_fault_masks(adev, AMDGPU_MMHUB0(0), true);
+		/* GFX HUB */
+		/* This works because this interrupt is only
+		 * enabled at init/resume and disabled in
+		 * fini/suspend, so the overall state doesn't
+		 * change over the course of suspend/resume.
+		 */
+		if (!adev->in_s0ix)
+			amdgpu_gmc_set_vm_fault_masks(adev, AMDGPU_GFXHUB(0), true);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gmc_v11_0_process_interrupt(struct amdgpu_device *adev,
+				       struct amdgpu_irq_src *source,
+				       struct amdgpu_iv_entry *entry)
+{
+	uint32_t vmhub_index = entry->client_id == SOC21_IH_CLIENTID_VMC ?
+			       AMDGPU_MMHUB0(0) : AMDGPU_GFXHUB(0);
+	struct amdgpu_vmhub *hub = &adev->vmhub[vmhub_index];
+	uint32_t status = 0;
+	u64 addr;
+
+	addr = (u64)entry->src_data[0] << 12;
+	addr |= ((u64)entry->src_data[1] & 0xf) << 44;
+
+	if (!amdgpu_sriov_vf(adev)) {
+		/*
+		 * Issue a dummy read to wait for the status register to
+		 * be updated to avoid reading an incorrect value due to
+		 * the new fast GRBM interface.
+		 */
+		if (entry->vmid_src == AMDGPU_GFXHUB(0))
+			RREG32(hub->vm_l2_pro_fault_status);
+
+		status = RREG32(hub->vm_l2_pro_fault_status);
+		WREG32_P(hub->vm_l2_pro_fault_cntl, 1, ~1);
+
+		amdgpu_vm_update_fault_cache(adev, entry->pasid, addr, status,
+					     entry->vmid_src ? AMDGPU_MMHUB0(0) : AMDGPU_GFXHUB(0));
+	}
+
+	if (printk_ratelimit()) {
+		struct amdgpu_task_info *task_info;
+
+		dev_err(adev->dev,
+			"[%s] page fault (src_id:%u ring:%u vmid:%u pasid:%u)\n",
+			entry->vmid_src ? "mmhub" : "gfxhub",
+			entry->src_id, entry->ring_id, entry->vmid, entry->pasid);
+		task_info = amdgpu_vm_get_task_info_pasid(adev, entry->pasid);
+		if (task_info) {
+			amdgpu_vm_print_task_info(adev, task_info);
+			amdgpu_vm_put_task_info(task_info);
+		}
+
+		dev_err(adev->dev, "  in page starting at address 0x%016llx from client %d\n",
+				addr, entry->client_id);
+
+		/* Only print L2 fault status if the status register could be read and
+		 * contains useful information
+		 */
+		if (status != 0)
+			hub->vmhub_funcs->print_l2_protection_fault_status(adev, status);
+	}
+
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs gmc_v11_0_irq_funcs = {
+	.set = gmc_v11_0_vm_fault_interrupt_state,
+	.process = gmc_v11_0_process_interrupt,
+};
+
+static const struct amdgpu_irq_src_funcs gmc_v11_0_ecc_funcs = {
+	.set = gmc_v11_0_ecc_interrupt_state,
+	.process = amdgpu_umc_process_ecc_irq,
+};
+
+static void gmc_v11_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->gmc.vm_fault.num_types = 1;
+	adev->gmc.vm_fault.funcs = &gmc_v11_0_irq_funcs;
+
+	if (!amdgpu_sriov_vf(adev)) {
+		adev->gmc.ecc_irq.num_types = 1;
+		adev->gmc.ecc_irq.funcs = &gmc_v11_0_ecc_funcs;
+	}
+}
+
+/**
+ * gmc_v11_0_use_invalidate_semaphore - judge whether to use semaphore
+ *
+ * @adev: amdgpu_device pointer
+ * @vmhub: vmhub type
+ *
+ */
+static bool gmc_v11_0_use_invalidate_semaphore(struct amdgpu_device *adev,
+				       uint32_t vmhub)
+{
+	return ((vmhub == AMDGPU_MMHUB0(0)) &&
+		(!amdgpu_sriov_vf(adev)));
+}
+
+static bool gmc_v11_0_get_vmid_pasid_mapping_info(
+					struct amdgpu_device *adev,
+					uint8_t vmid, uint16_t *p_pasid)
+{
+	*p_pasid = RREG32(SOC15_REG_OFFSET(OSSSYS, 0, regIH_VMID_0_LUT) + vmid) & 0xffff;
+
+	return !!(*p_pasid);
+}
+
+/**
+ * gmc_v11_0_flush_gpu_tlb - gart tlb flush callback
+ *
+ * @adev: amdgpu_device pointer
+ * @vmid: vm instance to flush
+ * @vmhub: which hub to flush
+ * @flush_type: the flush type
+ *
+ * Flush the TLB for the requested page table.
+ */
+static void gmc_v11_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid,
+					uint32_t vmhub, uint32_t flush_type)
+{
+	bool use_semaphore = gmc_v11_0_use_invalidate_semaphore(adev, vmhub);
+	struct amdgpu_vmhub *hub = &adev->vmhub[vmhub];
+	u32 inv_req = hub->vmhub_funcs->get_invalidate_req(vmid, flush_type);
+	/* Use register 17 for GART */
+	const unsigned int eng = 17;
+	unsigned char hub_ip;
+	u32 sem, req, ack;
+	unsigned int i;
+	u32 tmp;
+
+	if ((vmhub == AMDGPU_GFXHUB(0)) && !adev->gfx.is_poweron)
+		return;
+
+	sem = hub->vm_inv_eng0_sem + hub->eng_distance * eng;
+	req = hub->vm_inv_eng0_req + hub->eng_distance * eng;
+	ack = hub->vm_inv_eng0_ack + hub->eng_distance * eng;
+
+	/* flush hdp cache */
+	amdgpu_device_flush_hdp(adev, NULL);
+
+	/* This is necessary for SRIOV as well as for GFXOFF to function
+	 * properly under bare metal
+	 */
+	if ((adev->gfx.kiq[0].ring.sched.ready || adev->mes.ring[0].sched.ready) &&
+	    (amdgpu_sriov_runtime(adev) || !amdgpu_sriov_vf(adev))) {
+		amdgpu_gmc_fw_reg_write_reg_wait(adev, req, ack, inv_req,
+						 1 << vmid, GET_INST(GC, 0));
+		return;
+	}
+
+	/* This path is needed before KIQ/MES/GFXOFF are set up */
+	hub_ip = (vmhub == AMDGPU_GFXHUB(0)) ? GC_HWIP : MMHUB_HWIP;
+
+	spin_lock(&adev->gmc.invalidate_lock);
+	/*
+	 * It may lose gpuvm invalidate acknowldege state across power-gating
+	 * off cycle, add semaphore acquire before invalidation and semaphore
+	 * release after invalidation to avoid entering power gated state
+	 * to WA the Issue
+	 */
+
+	/* TODO: It needs to continue working on debugging with semaphore for GFXHUB as well. */
+	if (use_semaphore) {
+		for (i = 0; i < adev->usec_timeout; i++) {
+			/* a read return value of 1 means semaphore acuqire */
+			tmp = RREG32_RLC_NO_KIQ(sem, hub_ip);
+			if (tmp & 0x1)
+				break;
+			udelay(1);
+		}
+
+		if (i >= adev->usec_timeout)
+			DRM_ERROR("Timeout waiting for sem acquire in VM flush!\n");
+	}
+
+	WREG32_RLC_NO_KIQ(req, inv_req, hub_ip);
+
+	/* Wait for ACK with a delay.*/
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32_RLC_NO_KIQ(ack, hub_ip);
+		tmp &= 1 << vmid;
+		if (tmp)
+			break;
+
+		udelay(1);
+	}
+
+	/* TODO: It needs to continue working on debugging with semaphore for GFXHUB as well. */
+	if (use_semaphore)
+		WREG32_RLC_NO_KIQ(sem, 0, hub_ip);
+
+	/* Issue additional private vm invalidation to MMHUB */
+	if ((vmhub != AMDGPU_GFXHUB(0)) &&
+	    (hub->vm_l2_bank_select_reserved_cid2) &&
+		!amdgpu_sriov_vf(adev)) {
+		inv_req = RREG32_NO_KIQ(hub->vm_l2_bank_select_reserved_cid2);
+		/* bit 25: RSERVED_CACHE_PRIVATE_INVALIDATION */
+		inv_req |= (1 << 25);
+		/* Issue private invalidation */
+		WREG32_NO_KIQ(hub->vm_l2_bank_select_reserved_cid2, inv_req);
+		/* Read back to ensure invalidation is done*/
+		RREG32_NO_KIQ(hub->vm_l2_bank_select_reserved_cid2);
+	}
+
+	spin_unlock(&adev->gmc.invalidate_lock);
+
+	if (i >= adev->usec_timeout)
+		dev_err(adev->dev, "Timeout waiting for VM flush ACK!\n");
+}
+
+/**
+ * gmc_v11_0_flush_gpu_tlb_pasid - tlb flush via pasid
+ *
+ * @adev: amdgpu_device pointer
+ * @pasid: pasid to be flush
+ * @flush_type: the flush type
+ * @all_hub: flush all hubs
+ * @inst: is used to select which instance of KIQ to use for the invalidation
+ *
+ * Flush the TLB for the requested pasid.
+ */
+static void gmc_v11_0_flush_gpu_tlb_pasid(struct amdgpu_device *adev,
+					  uint16_t pasid, uint32_t flush_type,
+					  bool all_hub, uint32_t inst)
+{
+	uint16_t queried;
+	int vmid, i;
+
+	for (vmid = 1; vmid < 16; vmid++) {
+		bool valid;
+
+		valid = gmc_v11_0_get_vmid_pasid_mapping_info(adev, vmid,
+							      &queried);
+		if (!valid || queried != pasid)
+			continue;
+
+		if (all_hub) {
+			for_each_set_bit(i, adev->vmhubs_mask,
+					 AMDGPU_MAX_VMHUBS)
+				gmc_v11_0_flush_gpu_tlb(adev, vmid, i,
+							flush_type);
+		} else {
+			gmc_v11_0_flush_gpu_tlb(adev, vmid, AMDGPU_GFXHUB(0),
+						flush_type);
+		}
+	}
+}
+
+static uint64_t gmc_v11_0_emit_flush_gpu_tlb(struct amdgpu_ring *ring,
+					     unsigned int vmid, uint64_t pd_addr)
+{
+	bool use_semaphore = gmc_v11_0_use_invalidate_semaphore(ring->adev, ring->vm_hub);
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
+	uint32_t req = hub->vmhub_funcs->get_invalidate_req(vmid, 0);
+	unsigned int eng = ring->vm_inv_eng;
+
+	/*
+	 * It may lose gpuvm invalidate acknowldege state across power-gating
+	 * off cycle, add semaphore acquire before invalidation and semaphore
+	 * release after invalidation to avoid entering power gated state
+	 * to WA the Issue
+	 */
+
+	/* TODO: It needs to continue working on debugging with semaphore for GFXHUB as well. */
+	if (use_semaphore)
+		/* a read return value of 1 means semaphore acuqire */
+		amdgpu_ring_emit_reg_wait(ring,
+					  hub->vm_inv_eng0_sem +
+					  hub->eng_distance * eng, 0x1, 0x1);
+
+	amdgpu_ring_emit_wreg(ring, hub->ctx0_ptb_addr_lo32 +
+			      (hub->ctx_addr_distance * vmid),
+			      lower_32_bits(pd_addr));
+
+	amdgpu_ring_emit_wreg(ring, hub->ctx0_ptb_addr_hi32 +
+			      (hub->ctx_addr_distance * vmid),
+			      upper_32_bits(pd_addr));
+
+	amdgpu_ring_emit_reg_write_reg_wait(ring, hub->vm_inv_eng0_req +
+					    hub->eng_distance * eng,
+					    hub->vm_inv_eng0_ack +
+					    hub->eng_distance * eng,
+					    req, 1 << vmid);
+
+	/* TODO: It needs to continue working on debugging with semaphore for GFXHUB as well. */
+	if (use_semaphore)
+		/*
+		 * add semaphore release after invalidation,
+		 * write with 0 means semaphore release
+		 */
+		amdgpu_ring_emit_wreg(ring, hub->vm_inv_eng0_sem +
+				      hub->eng_distance * eng, 0);
+
+	return pd_addr;
+}
+
+static void gmc_v11_0_emit_pasid_mapping(struct amdgpu_ring *ring, unsigned int vmid,
+					 unsigned int pasid)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t reg;
+
+	if (ring->vm_hub == AMDGPU_GFXHUB(0))
+		reg = SOC15_REG_OFFSET(OSSSYS, 0, regIH_VMID_0_LUT) + vmid;
+	else
+		reg = SOC15_REG_OFFSET(OSSSYS, 0, regIH_VMID_0_LUT_MM) + vmid;
+
+	amdgpu_ring_emit_wreg(ring, reg, pasid);
+}
+
+/*
+ * PTE format:
+ * 63:59 reserved
+ * 58:57 reserved
+ * 56 F
+ * 55 L
+ * 54 reserved
+ * 53:52 SW
+ * 51 T
+ * 50:48 mtype
+ * 47:12 4k physical page base address
+ * 11:7 fragment
+ * 6 write
+ * 5 read
+ * 4 exe
+ * 3 Z
+ * 2 snooped
+ * 1 system
+ * 0 valid
+ *
+ * PDE format:
+ * 63:59 block fragment size
+ * 58:55 reserved
+ * 54 P
+ * 53:48 reserved
+ * 47:6 physical base address of PD or PTE
+ * 5:3 reserved
+ * 2 C
+ * 1 system
+ * 0 valid
+ */
+
+static void gmc_v11_0_get_vm_pde(struct amdgpu_device *adev, int level,
+				 uint64_t *addr, uint64_t *flags)
+{
+	if (!(*flags & AMDGPU_PDE_PTE) && !(*flags & AMDGPU_PTE_SYSTEM))
+		*addr = amdgpu_gmc_vram_mc2pa(adev, *addr);
+	BUG_ON(*addr & 0xFFFF00000000003FULL);
+
+	if (!adev->gmc.translate_further)
+		return;
+
+	if (level == AMDGPU_VM_PDB1) {
+		/* Set the block fragment size */
+		if (!(*flags & AMDGPU_PDE_PTE))
+			*flags |= AMDGPU_PDE_BFS(0x9);
+
+	} else if (level == AMDGPU_VM_PDB0) {
+		if (*flags & AMDGPU_PDE_PTE)
+			*flags &= ~AMDGPU_PDE_PTE;
+		else
+			*flags |= AMDGPU_PTE_TF;
+	}
+}
+
+static void gmc_v11_0_get_vm_pte(struct amdgpu_device *adev,
+				 struct amdgpu_vm *vm,
+				 struct amdgpu_bo *bo,
+				 uint32_t vm_flags,
+				 uint64_t *flags)
+{
+	if (vm_flags & AMDGPU_VM_PAGE_EXECUTABLE)
+		*flags |= AMDGPU_PTE_EXECUTABLE;
+	else
+		*flags &= ~AMDGPU_PTE_EXECUTABLE;
+
+	switch (vm_flags & AMDGPU_VM_MTYPE_MASK) {
+	case AMDGPU_VM_MTYPE_DEFAULT:
+	case AMDGPU_VM_MTYPE_NC:
+	default:
+		*flags = AMDGPU_PTE_MTYPE_NV10(*flags, MTYPE_NC);
+		break;
+	case AMDGPU_VM_MTYPE_WC:
+		*flags = AMDGPU_PTE_MTYPE_NV10(*flags, MTYPE_WC);
+		break;
+	case AMDGPU_VM_MTYPE_CC:
+		*flags = AMDGPU_PTE_MTYPE_NV10(*flags, MTYPE_CC);
+		break;
+	case AMDGPU_VM_MTYPE_UC:
+		*flags = AMDGPU_PTE_MTYPE_NV10(*flags, MTYPE_UC);
+		break;
+	}
+
+	if (vm_flags & AMDGPU_VM_PAGE_NOALLOC)
+		*flags |= AMDGPU_PTE_NOALLOC;
+	else
+		*flags &= ~AMDGPU_PTE_NOALLOC;
+
+	if (vm_flags & AMDGPU_VM_PAGE_PRT) {
+		*flags |= AMDGPU_PTE_PRT;
+		*flags |= AMDGPU_PTE_SNOOPED;
+		*flags |= AMDGPU_PTE_LOG;
+		*flags |= AMDGPU_PTE_SYSTEM;
+		*flags &= ~AMDGPU_PTE_VALID;
+	}
+
+	if (bo && bo->flags & (AMDGPU_GEM_CREATE_COHERENT |
+			       AMDGPU_GEM_CREATE_EXT_COHERENT |
+			       AMDGPU_GEM_CREATE_UNCACHED))
+		*flags = AMDGPU_PTE_MTYPE_NV10(*flags, MTYPE_UC);
+}
+
+static unsigned int gmc_v11_0_get_vbios_fb_size(struct amdgpu_device *adev)
+{
+	u32 d1vga_control = RREG32_SOC15(DCE, 0, regD1VGA_CONTROL);
+	unsigned int size;
+
+	if (REG_GET_FIELD(d1vga_control, D1VGA_CONTROL, D1VGA_MODE_ENABLE)) {
+		size = AMDGPU_VBIOS_VGA_ALLOCATION;
+	} else {
+		u32 viewport;
+		u32 pitch;
+
+		viewport = RREG32_SOC15(DCE, 0, regHUBP0_DCSURF_PRI_VIEWPORT_DIMENSION);
+		pitch = RREG32_SOC15(DCE, 0, regHUBPREQ0_DCSURF_SURFACE_PITCH);
+		size = (REG_GET_FIELD(viewport,
+					HUBP0_DCSURF_PRI_VIEWPORT_DIMENSION, PRI_VIEWPORT_HEIGHT) *
+				REG_GET_FIELD(pitch, HUBPREQ0_DCSURF_SURFACE_PITCH, PITCH) *
+				4);
+	}
+
+	return size;
+}
+
+static const struct amdgpu_gmc_funcs gmc_v11_0_gmc_funcs = {
+	.flush_gpu_tlb = gmc_v11_0_flush_gpu_tlb,
+	.flush_gpu_tlb_pasid = gmc_v11_0_flush_gpu_tlb_pasid,
+	.emit_flush_gpu_tlb = gmc_v11_0_emit_flush_gpu_tlb,
+	.emit_pasid_mapping = gmc_v11_0_emit_pasid_mapping,
+	.get_vm_pde = gmc_v11_0_get_vm_pde,
+	.get_vm_pte = gmc_v11_0_get_vm_pte,
+	.get_vbios_fb_size = gmc_v11_0_get_vbios_fb_size,
+};
+
+static void gmc_v11_0_set_gmc_funcs(struct amdgpu_device *adev)
+{
+	adev->gmc.gmc_funcs = &gmc_v11_0_gmc_funcs;
+}
+
+static void gmc_v11_0_set_umc_funcs(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, UMC_HWIP, 0)) {
+	case IP_VERSION(8, 10, 0):
+		adev->umc.channel_inst_num = UMC_V8_10_CHANNEL_INSTANCE_NUM;
+		adev->umc.umc_inst_num = UMC_V8_10_UMC_INSTANCE_NUM;
+		adev->umc.max_ras_err_cnt_per_query = UMC_V8_10_TOTAL_CHANNEL_NUM(adev);
+		adev->umc.channel_offs = UMC_V8_10_PER_CHANNEL_OFFSET;
+		adev->umc.retire_unit = UMC_V8_10_NA_COL_2BITS_POWER_OF_2_NUM;
+		if (adev->umc.node_inst_num == 4)
+			adev->umc.channel_idx_tbl = &umc_v8_10_channel_idx_tbl_ext0[0][0][0];
+		else
+			adev->umc.channel_idx_tbl = &umc_v8_10_channel_idx_tbl[0][0][0];
+		adev->umc.ras = &umc_v8_10_ras;
+		break;
+	case IP_VERSION(8, 11, 0):
+		break;
+	default:
+		break;
+	}
+}
+
+
+static void gmc_v11_0_set_mmhub_funcs(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) {
+	case IP_VERSION(3, 0, 1):
+		adev->mmhub.funcs = &mmhub_v3_0_1_funcs;
+		break;
+	case IP_VERSION(3, 0, 2):
+		adev->mmhub.funcs = &mmhub_v3_0_2_funcs;
+		break;
+	case IP_VERSION(3, 3, 0):
+	case IP_VERSION(3, 3, 1):
+	case IP_VERSION(3, 3, 2):
+		adev->mmhub.funcs = &mmhub_v3_3_funcs;
+		break;
+	default:
+		adev->mmhub.funcs = &mmhub_v3_0_funcs;
+		break;
+	}
+}
+
+static void gmc_v11_0_set_gfxhub_funcs(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(11, 0, 3):
+		adev->gfxhub.funcs = &gfxhub_v3_0_3_funcs;
+		break;
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 1):
+	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
+		adev->gfxhub.funcs = &gfxhub_v11_5_0_funcs;
+		break;
+	default:
+		adev->gfxhub.funcs = &gfxhub_v3_0_funcs;
+		break;
+	}
+}
+
+static int gmc_v11_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	gmc_v11_0_set_gfxhub_funcs(adev);
+	gmc_v11_0_set_mmhub_funcs(adev);
+	gmc_v11_0_set_gmc_funcs(adev);
+	gmc_v11_0_set_irq_funcs(adev);
+	gmc_v11_0_set_umc_funcs(adev);
+
+	adev->gmc.shared_aperture_start = 0x2000000000000000ULL;
+	adev->gmc.shared_aperture_end =
+		adev->gmc.shared_aperture_start + (4ULL << 30) - 1;
+	adev->gmc.private_aperture_start = 0x1000000000000000ULL;
+	adev->gmc.private_aperture_end =
+		adev->gmc.private_aperture_start + (4ULL << 30) - 1;
+	adev->gmc.noretry_flags = AMDGPU_VM_NORETRY_FLAGS_TF;
+
+	return 0;
+}
+
+static int gmc_v11_0_late_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = amdgpu_gmc_allocate_vm_inv_eng(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_gmc_ras_late_init(adev);
+	if (r)
+		return r;
+
+	return amdgpu_irq_get(adev, &adev->gmc.vm_fault, 0);
+}
+
+static void gmc_v11_0_vram_gtt_location(struct amdgpu_device *adev,
+					struct amdgpu_gmc *mc)
+{
+	u64 base = 0;
+
+	base = adev->mmhub.funcs->get_fb_location(adev);
+
+	amdgpu_gmc_set_agp_default(adev, mc);
+	amdgpu_gmc_vram_location(adev, &adev->gmc, base);
+	amdgpu_gmc_gart_location(adev, mc, AMDGPU_GART_PLACEMENT_HIGH);
+	if (!amdgpu_sriov_vf(adev) &&
+	    (amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(11, 5, 0)) &&
+	    (amdgpu_agp == 1))
+		amdgpu_gmc_agp_location(adev, mc);
+
+	/* base offset of vram pages */
+	if (amdgpu_sriov_vf(adev))
+		adev->vm_manager.vram_base_offset = 0;
+	else
+		adev->vm_manager.vram_base_offset = adev->mmhub.funcs->get_mc_fb_offset(adev);
+}
+
+/**
+ * gmc_v11_0_mc_init - initialize the memory controller driver params
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Look up the amount of vram, vram width, and decide how to place
+ * vram and gart within the GPU's physical address space.
+ * Returns 0 for success.
+ */
+static int gmc_v11_0_mc_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	/* size in MB on si */
+	adev->gmc.mc_vram_size =
+		adev->nbio.funcs->get_memsize(adev) * 1024ULL * 1024ULL;
+	adev->gmc.real_vram_size = adev->gmc.mc_vram_size;
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		r = amdgpu_device_resize_fb_bar(adev);
+		if (r)
+			return r;
+	}
+	adev->gmc.aper_base = pci_resource_start(adev->pdev, 0);
+	adev->gmc.aper_size = pci_resource_len(adev->pdev, 0);
+
+#ifdef CONFIG_X86_64
+	if ((adev->flags & AMD_IS_APU) && !amdgpu_passthrough(adev)) {
+		adev->gmc.aper_base = adev->mmhub.funcs->get_mc_fb_offset(adev);
+		adev->gmc.aper_size = adev->gmc.real_vram_size;
+	}
+#endif
+	/* In case the PCI BAR is larger than the actual amount of vram */
+	adev->gmc.visible_vram_size = adev->gmc.aper_size;
+	if (adev->gmc.visible_vram_size > adev->gmc.real_vram_size)
+		adev->gmc.visible_vram_size = adev->gmc.real_vram_size;
+
+	/* set the gart size */
+	if (amdgpu_gart_size == -1)
+		adev->gmc.gart_size = 512ULL << 20;
+	else
+		adev->gmc.gart_size = (u64)amdgpu_gart_size << 20;
+
+	gmc_v11_0_vram_gtt_location(adev, &adev->gmc);
+
+	return 0;
+}
+
+static int gmc_v11_0_gart_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->gart.bo) {
+		WARN(1, "PCIE GART already initialized\n");
+		return 0;
+	}
+
+	/* Initialize common gart structure */
+	r = amdgpu_gart_init(adev);
+	if (r)
+		return r;
+
+	adev->gart.table_size = adev->gart.num_gpu_pages * 8;
+	adev->gart.gart_pte_flags = AMDGPU_PTE_MTYPE_NV10(0ULL, MTYPE_UC) |
+				 AMDGPU_PTE_EXECUTABLE;
+
+	return amdgpu_gart_table_vram_alloc(adev);
+}
+
+static int gmc_v11_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	int r, vram_width = 0, vram_type = 0, vram_vendor = 0;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	adev->mmhub.funcs->init(adev);
+
+	adev->gfxhub.funcs->init(adev);
+
+	spin_lock_init(&adev->gmc.invalidate_lock);
+
+	r = amdgpu_atomfirmware_get_vram_info(adev,
+					      &vram_width, &vram_type, &vram_vendor);
+	adev->gmc.vram_width = vram_width;
+
+	adev->gmc.vram_type = vram_type;
+	adev->gmc.vram_vendor = vram_vendor;
+
+	/* The mall_size is already calculated as mall_size_per_umc * num_umc.
+	 * However, for gfx1151, which features a 2-to-1 UMC mapping,
+	 * the result must be multiplied by 2 to determine the actual mall size.
+	 */
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(11, 5, 1):
+		adev->gmc.mall_size *= 2;
+		break;
+	default:
+		break;
+	}
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 1):
+	case IP_VERSION(11, 0, 2):
+	case IP_VERSION(11, 0, 3):
+	case IP_VERSION(11, 0, 4):
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 1):
+	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
+		set_bit(AMDGPU_GFXHUB(0), adev->vmhubs_mask);
+		set_bit(AMDGPU_MMHUB0(0), adev->vmhubs_mask);
+		/*
+		 * To fulfill 4-level page support,
+		 * vm size is 256TB (48bit), maximum size,
+		 * block size 512 (9bit)
+		 */
+		amdgpu_vm_adjust_size(adev, 256 * 1024, 9, 3, 48);
+		break;
+	default:
+		break;
+	}
+
+	/* This interrupt is VMC page fault.*/
+	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_VMC,
+			      VMC_1_0__SRCID__VM_FAULT,
+			      &adev->gmc.vm_fault);
+
+	if (r)
+		return r;
+
+	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GFX,
+			      UTCL2_1_0__SRCID__FAULT,
+			      &adev->gmc.vm_fault);
+	if (r)
+		return r;
+
+	if (!amdgpu_sriov_vf(adev)) {
+		/* interrupt sent to DF. */
+		r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_DF, 0,
+				      &adev->gmc.ecc_irq);
+		if (r)
+			return r;
+	}
+
+	/*
+	 * Set the internal MC address mask This is the max address of the GPU's
+	 * internal address space.
+	 */
+	adev->gmc.mc_mask = 0xffffffffffffULL; /* 48 bit MC */
+
+	r = dma_set_mask_and_coherent(adev->dev, DMA_BIT_MASK(44));
+	if (r) {
+		dev_warn(adev->dev, "amdgpu: No suitable DMA available.\n");
+		return r;
+	}
+
+	adev->need_swiotlb = drm_need_swiotlb(44);
+
+	r = gmc_v11_0_mc_init(adev);
+	if (r)
+		return r;
+
+	amdgpu_gmc_get_vbios_allocations(adev);
+
+	/* Memory manager */
+	r = amdgpu_bo_init(adev);
+	if (r)
+		return r;
+
+	r = gmc_v11_0_gart_init(adev);
+	if (r)
+		return r;
+
+	/*
+	 * number of VMs
+	 * VMID 0 is reserved for System
+	 * amdgpu graphics/compute will use VMIDs 1-7
+	 * amdkfd will use VMIDs 8-15
+	 */
+	adev->vm_manager.first_kfd_vmid = adev->gfx.disable_kq ? 1 : 8;
+
+	amdgpu_vm_manager_init(adev);
+
+	r = amdgpu_gmc_ras_sw_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+/**
+ * gmc_v11_0_gart_fini - vm fini callback
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tears down the driver GART/VM setup (CIK).
+ */
+static void gmc_v11_0_gart_fini(struct amdgpu_device *adev)
+{
+	amdgpu_gart_table_vram_free(adev);
+}
+
+static int gmc_v11_0_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	amdgpu_vm_manager_fini(adev);
+	gmc_v11_0_gart_fini(adev);
+	amdgpu_gem_force_release(adev);
+	amdgpu_bo_fini(adev);
+
+	return 0;
+}
+
+static void gmc_v11_0_init_golden_registers(struct amdgpu_device *adev)
+{
+	if (amdgpu_sriov_vf(adev)) {
+		struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+
+		WREG32(hub->vm_contexts_disable, 0);
+		return;
+	}
+}
+
+/**
+ * gmc_v11_0_gart_enable - gart enable
+ *
+ * @adev: amdgpu_device pointer
+ */
+static int gmc_v11_0_gart_enable(struct amdgpu_device *adev)
+{
+	int r;
+	bool value;
+
+	if (adev->gart.bo == NULL) {
+		dev_err(adev->dev, "No VRAM object for PCIE GART.\n");
+		return -EINVAL;
+	}
+
+	amdgpu_gtt_mgr_recover(&adev->mman.gtt_mgr);
+
+	r = adev->mmhub.funcs->gart_enable(adev);
+	if (r)
+		return r;
+
+	/* Flush HDP after it is initialized */
+	amdgpu_device_flush_hdp(adev, NULL);
+
+	value = amdgpu_vm_fault_stop != AMDGPU_VM_FAULT_STOP_ALWAYS;
+
+	adev->mmhub.funcs->set_fault_enable_default(adev, value);
+	gmc_v11_0_flush_gpu_tlb(adev, 0, AMDGPU_MMHUB0(0), 0);
+
+	DRM_INFO("PCIE GART of %uM enabled (table at 0x%016llX).\n",
+		 (unsigned int)(adev->gmc.gart_size >> 20),
+		 (unsigned long long)amdgpu_bo_gpu_offset(adev->gart.bo));
+
+	return 0;
+}
+
+static int gmc_v11_0_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	adev->gmc.flush_pasid_uses_kiq = !amdgpu_emu_mode;
+
+	/* The sequence of these two function calls matters.*/
+	gmc_v11_0_init_golden_registers(adev);
+
+	r = gmc_v11_0_gart_enable(adev);
+	if (r)
+		return r;
+
+	if (adev->umc.funcs && adev->umc.funcs->init_registers)
+		adev->umc.funcs->init_registers(adev);
+
+	return 0;
+}
+
+/**
+ * gmc_v11_0_gart_disable - gart disable
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * This disables all VM page table.
+ */
+static void gmc_v11_0_gart_disable(struct amdgpu_device *adev)
+{
+	adev->mmhub.funcs->gart_disable(adev);
+}
+
+static int gmc_v11_0_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_sriov_vf(adev)) {
+		/* full access mode, so don't touch any GMC register */
+		DRM_DEBUG("For SRIOV client, shouldn't do anything.\n");
+		return 0;
+	}
+
+	amdgpu_irq_put(adev, &adev->gmc.vm_fault, 0);
+
+	if (adev->gmc.ecc_irq.funcs &&
+		amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__UMC))
+		amdgpu_irq_put(adev, &adev->gmc.ecc_irq, 0);
+
+	gmc_v11_0_gart_disable(adev);
+
+	return 0;
+}
+
+static int gmc_v11_0_suspend(struct amdgpu_ip_block *ip_block)
+{
+	gmc_v11_0_hw_fini(ip_block);
+
+	return 0;
+}
+
+static int gmc_v11_0_resume(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	r = gmc_v11_0_hw_init(ip_block);
+	if (r)
+		return r;
+
+	amdgpu_vmid_reset_all(ip_block->adev);
+
+	return 0;
+}
+
+static bool gmc_v11_0_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	/* MC is always ready in GMC v11.*/
+	return true;
+}
+
+static int gmc_v11_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	/* There is no need to wait for MC idle in GMC v11.*/
+	return 0;
+}
+
+static int gmc_v11_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					   enum amd_clockgating_state state)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	r = adev->mmhub.funcs->set_clockgating(adev, state);
+	if (r)
+		return r;
+
+	return athub_v3_0_set_clockgating(adev, state);
+}
+
+static void gmc_v11_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	adev->mmhub.funcs->get_clockgating(adev, flags);
+
+	athub_v3_0_get_clockgating(adev, flags);
+}
+
+static int gmc_v11_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					   enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs gmc_v11_0_ip_funcs = {
+	.name = "gmc_v11_0",
+	.early_init = gmc_v11_0_early_init,
+	.sw_init = gmc_v11_0_sw_init,
+	.hw_init = gmc_v11_0_hw_init,
+	.late_init = gmc_v11_0_late_init,
+	.sw_fini = gmc_v11_0_sw_fini,
+	.hw_fini = gmc_v11_0_hw_fini,
+	.suspend = gmc_v11_0_suspend,
+	.resume = gmc_v11_0_resume,
+	.is_idle = gmc_v11_0_is_idle,
+	.wait_for_idle = gmc_v11_0_wait_for_idle,
+	.set_clockgating_state = gmc_v11_0_set_clockgating_state,
+	.set_powergating_state = gmc_v11_0_set_powergating_state,
+	.get_clockgating_state = gmc_v11_0_get_clockgating_state,
+};
+
+const struct amdgpu_ip_block_version gmc_v11_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_GMC,
+	.major = 11,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &gmc_v11_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.h b/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.h
new file mode 100644
index 000000000000..def4d5516f82
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GMC_V11_0_H__
+#define __GMC_V11_0_H__
+
+extern const struct amd_ip_funcs gmc_v11_0_ip_funcs;
+extern const struct amdgpu_ip_block_version gmc_v11_0_ip_block;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v12_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v12_0.c
new file mode 100644
index 000000000000..f4a19357ccbc
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v12_0.c
@@ -0,0 +1,1043 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/firmware.h>
+#include <linux/pci.h>
+
+#include <drm/drm_cache.h>
+
+#include "amdgpu.h"
+#include "amdgpu_atomfirmware.h"
+#include "gmc_v12_0.h"
+#include "athub/athub_4_1_0_sh_mask.h"
+#include "athub/athub_4_1_0_offset.h"
+#include "oss/osssys_7_0_0_offset.h"
+#include "ivsrcid/vmc/irqsrcs_vmc_1_0.h"
+#include "soc24_enum.h"
+#include "soc24.h"
+#include "soc15d.h"
+#include "soc15_common.h"
+#include "nbif_v6_3_1.h"
+#include "gfxhub_v12_0.h"
+#include "mmhub_v4_1_0.h"
+#include "athub_v4_1_0.h"
+#include "umc_v8_14.h"
+
+static int gmc_v12_0_ecc_interrupt_state(struct amdgpu_device *adev,
+					 struct amdgpu_irq_src *src,
+					 unsigned type,
+					 enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+static int gmc_v12_0_vm_fault_interrupt_state(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *src, unsigned type,
+					      enum amdgpu_interrupt_state state)
+{
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		/* MM HUB */
+		amdgpu_gmc_set_vm_fault_masks(adev, AMDGPU_MMHUB0(0), false);
+		/* GFX HUB */
+		/* This works because this interrupt is only
+		 * enabled at init/resume and disabled in
+		 * fini/suspend, so the overall state doesn't
+		 * change over the course of suspend/resume.
+		 */
+		if (!adev->in_s0ix)
+			amdgpu_gmc_set_vm_fault_masks(adev, AMDGPU_GFXHUB(0), false);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		/* MM HUB */
+		amdgpu_gmc_set_vm_fault_masks(adev, AMDGPU_MMHUB0(0), true);
+		/* GFX HUB */
+		/* This works because this interrupt is only
+		 * enabled at init/resume and disabled in
+		 * fini/suspend, so the overall state doesn't
+		 * change over the course of suspend/resume.
+		 */
+		if (!adev->in_s0ix)
+			amdgpu_gmc_set_vm_fault_masks(adev, AMDGPU_GFXHUB(0), true);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gmc_v12_0_process_interrupt(struct amdgpu_device *adev,
+				       struct amdgpu_irq_src *source,
+				       struct amdgpu_iv_entry *entry)
+{
+	struct amdgpu_vmhub *hub;
+	uint32_t status = 0;
+	u64 addr;
+
+	addr = (u64)entry->src_data[0] << 12;
+	addr |= ((u64)entry->src_data[1] & 0xf) << 44;
+
+	if (entry->client_id == SOC21_IH_CLIENTID_VMC)
+		hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	else
+		hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+
+	if (!amdgpu_sriov_vf(adev)) {
+		/*
+		 * Issue a dummy read to wait for the status register to
+		 * be updated to avoid reading an incorrect value due to
+		 * the new fast GRBM interface.
+		 */
+		if (entry->vmid_src == AMDGPU_GFXHUB(0))
+			RREG32(hub->vm_l2_pro_fault_status);
+
+		status = RREG32(hub->vm_l2_pro_fault_status);
+		WREG32_P(hub->vm_l2_pro_fault_cntl, 1, ~1);
+
+		amdgpu_vm_update_fault_cache(adev, entry->pasid, addr, status,
+					     entry->vmid_src ? AMDGPU_MMHUB0(0) : AMDGPU_GFXHUB(0));
+	}
+
+	if (printk_ratelimit()) {
+		struct amdgpu_task_info *task_info;
+
+		dev_err(adev->dev,
+			"[%s] page fault (src_id:%u ring:%u vmid:%u pasid:%u)\n",
+			entry->vmid_src ? "mmhub" : "gfxhub",
+			entry->src_id, entry->ring_id, entry->vmid, entry->pasid);
+		task_info = amdgpu_vm_get_task_info_pasid(adev, entry->pasid);
+		if (task_info) {
+			amdgpu_vm_print_task_info(adev, task_info);
+			amdgpu_vm_put_task_info(task_info);
+		}
+
+		dev_err(adev->dev, "  in page starting at address 0x%016llx from client %d\n",
+				addr, entry->client_id);
+
+		/* Only print L2 fault status if the status register could be read and
+		 * contains useful information
+		 */
+		if (status != 0)
+			hub->vmhub_funcs->print_l2_protection_fault_status(adev, status);
+	}
+
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs gmc_v12_0_irq_funcs = {
+	.set = gmc_v12_0_vm_fault_interrupt_state,
+	.process = gmc_v12_0_process_interrupt,
+};
+
+static const struct amdgpu_irq_src_funcs gmc_v12_0_ecc_funcs = {
+	.set = gmc_v12_0_ecc_interrupt_state,
+	.process = amdgpu_umc_process_ecc_irq,
+};
+
+static void gmc_v12_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->gmc.vm_fault.num_types = 1;
+	adev->gmc.vm_fault.funcs = &gmc_v12_0_irq_funcs;
+
+	if (!amdgpu_sriov_vf(adev)) {
+		adev->gmc.ecc_irq.num_types = 1;
+		adev->gmc.ecc_irq.funcs = &gmc_v12_0_ecc_funcs;
+	}
+}
+
+/**
+ * gmc_v12_0_use_invalidate_semaphore - judge whether to use semaphore
+ *
+ * @adev: amdgpu_device pointer
+ * @vmhub: vmhub type
+ *
+ */
+static bool gmc_v12_0_use_invalidate_semaphore(struct amdgpu_device *adev,
+				       uint32_t vmhub)
+{
+	return ((vmhub == AMDGPU_MMHUB0(0)) &&
+		(!amdgpu_sriov_vf(adev)));
+}
+
+static bool gmc_v12_0_get_vmid_pasid_mapping_info(
+					struct amdgpu_device *adev,
+					uint8_t vmid, uint16_t *p_pasid)
+{
+	*p_pasid = RREG32(SOC15_REG_OFFSET(OSSSYS, 0, regIH_VMID_0_LUT) + vmid) & 0xffff;
+
+	return !!(*p_pasid);
+}
+
+/*
+ * GART
+ * VMID 0 is the physical GPU addresses as used by the kernel.
+ * VMIDs 1-15 are used for userspace clients and are handled
+ * by the amdgpu vm/hsa code.
+ */
+
+static void gmc_v12_0_flush_vm_hub(struct amdgpu_device *adev, uint32_t vmid,
+				   unsigned int vmhub, uint32_t flush_type)
+{
+	bool use_semaphore = gmc_v12_0_use_invalidate_semaphore(adev, vmhub);
+	struct amdgpu_vmhub *hub = &adev->vmhub[vmhub];
+	u32 inv_req = hub->vmhub_funcs->get_invalidate_req(vmid, flush_type);
+	u32 tmp;
+	/* Use register 17 for GART */
+	const unsigned eng = 17;
+	unsigned int i;
+	unsigned char hub_ip = 0;
+
+	hub_ip = (vmhub == AMDGPU_GFXHUB(0)) ?
+		   GC_HWIP : MMHUB_HWIP;
+
+	spin_lock(&adev->gmc.invalidate_lock);
+	/*
+	 * It may lose gpuvm invalidate acknowldege state across power-gating
+	 * off cycle, add semaphore acquire before invalidation and semaphore
+	 * release after invalidation to avoid entering power gated state
+	 * to WA the Issue
+	 */
+
+	/* TODO: It needs to continue working on debugging with semaphore for GFXHUB as well. */
+	if (use_semaphore) {
+		for (i = 0; i < adev->usec_timeout; i++) {
+			/* a read return value of 1 means semaphore acuqire */
+			tmp = RREG32_RLC_NO_KIQ(hub->vm_inv_eng0_sem +
+					    hub->eng_distance * eng, hub_ip);
+			if (tmp & 0x1)
+				break;
+			udelay(1);
+		}
+
+		if (i >= adev->usec_timeout)
+			dev_err(adev->dev,
+				"Timeout waiting for sem acquire in VM flush!\n");
+	}
+
+	WREG32_RLC_NO_KIQ(hub->vm_inv_eng0_req + hub->eng_distance * eng, inv_req, hub_ip);
+
+	/* Wait for ACK with a delay.*/
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32_RLC_NO_KIQ(hub->vm_inv_eng0_ack +
+				    hub->eng_distance * eng, hub_ip);
+		tmp &= 1 << vmid;
+		if (tmp)
+			break;
+
+		udelay(1);
+	}
+
+	/* TODO: It needs to continue working on debugging with semaphore for GFXHUB as well. */
+	if (use_semaphore)
+		/*
+		 * add semaphore release after invalidation,
+		 * write with 0 means semaphore release
+		 */
+		WREG32_RLC_NO_KIQ(hub->vm_inv_eng0_sem +
+			      hub->eng_distance * eng, 0, hub_ip);
+
+	/* Issue additional private vm invalidation to MMHUB */
+	if ((vmhub != AMDGPU_GFXHUB(0)) &&
+	    (hub->vm_l2_bank_select_reserved_cid2) &&
+		!amdgpu_sriov_vf(adev)) {
+		inv_req = RREG32_NO_KIQ(hub->vm_l2_bank_select_reserved_cid2);
+		/* bit 25: RSERVED_CACHE_PRIVATE_INVALIDATION */
+		inv_req |= (1 << 25);
+		/* Issue private invalidation */
+		WREG32_NO_KIQ(hub->vm_l2_bank_select_reserved_cid2, inv_req);
+		/* Read back to ensure invalidation is done*/
+		RREG32_NO_KIQ(hub->vm_l2_bank_select_reserved_cid2);
+	}
+
+	spin_unlock(&adev->gmc.invalidate_lock);
+
+	if (i < adev->usec_timeout)
+		return;
+
+	dev_err(adev->dev, "Timeout waiting for VM flush ACK!\n");
+}
+
+/**
+ * gmc_v12_0_flush_gpu_tlb - gart tlb flush callback
+ *
+ * @adev: amdgpu_device pointer
+ * @vmid: vm instance to flush
+ * @vmhub: which hub to flush
+ * @flush_type: the flush type
+ *
+ * Flush the TLB for the requested page table.
+ */
+static void gmc_v12_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid,
+					uint32_t vmhub, uint32_t flush_type)
+{
+	if ((vmhub == AMDGPU_GFXHUB(0)) && !adev->gfx.is_poweron)
+		return;
+
+	/* flush hdp cache */
+	amdgpu_device_flush_hdp(adev, NULL);
+
+	/* This is necessary for SRIOV as well as for GFXOFF to function
+	 * properly under bare metal
+	 */
+	if ((adev->gfx.kiq[0].ring.sched.ready || adev->mes.ring[0].sched.ready) &&
+	    (amdgpu_sriov_runtime(adev) || !amdgpu_sriov_vf(adev))) {
+		struct amdgpu_vmhub *hub = &adev->vmhub[vmhub];
+		const unsigned eng = 17;
+		u32 inv_req = hub->vmhub_funcs->get_invalidate_req(vmid, flush_type);
+		u32 req = hub->vm_inv_eng0_req + hub->eng_distance * eng;
+		u32 ack = hub->vm_inv_eng0_ack + hub->eng_distance * eng;
+
+		amdgpu_gmc_fw_reg_write_reg_wait(adev, req, ack, inv_req,
+				1 << vmid, GET_INST(GC, 0));
+		return;
+	}
+
+	mutex_lock(&adev->mman.gtt_window_lock);
+	gmc_v12_0_flush_vm_hub(adev, vmid, vmhub, 0);
+	mutex_unlock(&adev->mman.gtt_window_lock);
+	return;
+}
+
+/**
+ * gmc_v12_0_flush_gpu_tlb_pasid - tlb flush via pasid
+ *
+ * @adev: amdgpu_device pointer
+ * @pasid: pasid to be flush
+ * @flush_type: the flush type
+ * @all_hub: flush all hubs
+ * @inst: is used to select which instance of KIQ to use for the invalidation
+ *
+ * Flush the TLB for the requested pasid.
+ */
+static void gmc_v12_0_flush_gpu_tlb_pasid(struct amdgpu_device *adev,
+					  uint16_t pasid, uint32_t flush_type,
+					  bool all_hub, uint32_t inst)
+{
+	uint16_t queried;
+	int vmid, i;
+
+	if (adev->enable_uni_mes && adev->mes.ring[AMDGPU_MES_SCHED_PIPE].sched.ready &&
+	    (adev->mes.sched_version & AMDGPU_MES_VERSION_MASK) >= 0x84) {
+		struct mes_inv_tlbs_pasid_input input = {0};
+		input.pasid = pasid;
+		input.flush_type = flush_type;
+		input.hub_id = AMDGPU_GFXHUB(0);
+		/* MES will invalidate all gc_hub for the device from master */
+		adev->mes.funcs->invalidate_tlbs_pasid(&adev->mes, &input);
+		if (all_hub) {
+			/* Only need to invalidate mm_hub now, gfx12 only support one mmhub */
+			input.hub_id = AMDGPU_MMHUB0(0);
+			adev->mes.funcs->invalidate_tlbs_pasid(&adev->mes, &input);
+		}
+		return;
+	}
+
+	for (vmid = 1; vmid < 16; vmid++) {
+		bool valid;
+
+		valid = gmc_v12_0_get_vmid_pasid_mapping_info(adev, vmid,
+							      &queried);
+		if (!valid || queried != pasid)
+			continue;
+
+		if (all_hub) {
+			for_each_set_bit(i, adev->vmhubs_mask,
+					 AMDGPU_MAX_VMHUBS)
+				gmc_v12_0_flush_gpu_tlb(adev, vmid, i,
+							flush_type);
+		} else {
+			gmc_v12_0_flush_gpu_tlb(adev, vmid, AMDGPU_GFXHUB(0),
+						flush_type);
+		}
+	}
+}
+
+static uint64_t gmc_v12_0_emit_flush_gpu_tlb(struct amdgpu_ring *ring,
+					     unsigned vmid, uint64_t pd_addr)
+{
+	bool use_semaphore = gmc_v12_0_use_invalidate_semaphore(ring->adev, ring->vm_hub);
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
+	uint32_t req = hub->vmhub_funcs->get_invalidate_req(vmid, 0);
+	unsigned eng = ring->vm_inv_eng;
+
+	/*
+	 * It may lose gpuvm invalidate acknowldege state across power-gating
+	 * off cycle, add semaphore acquire before invalidation and semaphore
+	 * release after invalidation to avoid entering power gated state
+	 * to WA the Issue
+	 */
+
+	/* TODO: It needs to continue working on debugging with semaphore for GFXHUB as well. */
+	if (use_semaphore)
+		/* a read return value of 1 means semaphore acuqire */
+		amdgpu_ring_emit_reg_wait(ring,
+					  hub->vm_inv_eng0_sem +
+					  hub->eng_distance * eng, 0x1, 0x1);
+
+	amdgpu_ring_emit_wreg(ring, hub->ctx0_ptb_addr_lo32 +
+			      (hub->ctx_addr_distance * vmid),
+			      lower_32_bits(pd_addr));
+
+	amdgpu_ring_emit_wreg(ring, hub->ctx0_ptb_addr_hi32 +
+			      (hub->ctx_addr_distance * vmid),
+			      upper_32_bits(pd_addr));
+
+	amdgpu_ring_emit_reg_write_reg_wait(ring, hub->vm_inv_eng0_req +
+					    hub->eng_distance * eng,
+					    hub->vm_inv_eng0_ack +
+					    hub->eng_distance * eng,
+					    req, 1 << vmid);
+
+	/* TODO: It needs to continue working on debugging with semaphore for GFXHUB as well. */
+	if (use_semaphore)
+		/*
+		 * add semaphore release after invalidation,
+		 * write with 0 means semaphore release
+		 */
+		amdgpu_ring_emit_wreg(ring, hub->vm_inv_eng0_sem +
+				      hub->eng_distance * eng, 0);
+
+	return pd_addr;
+}
+
+static void gmc_v12_0_emit_pasid_mapping(struct amdgpu_ring *ring, unsigned vmid,
+					 unsigned pasid)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t reg;
+
+	if (ring->vm_hub == AMDGPU_GFXHUB(0))
+		reg = SOC15_REG_OFFSET(OSSSYS, 0, regIH_VMID_0_LUT) + vmid;
+	else
+		reg = SOC15_REG_OFFSET(OSSSYS, 0, regIH_VMID_0_LUT_MM) + vmid;
+
+	amdgpu_ring_emit_wreg(ring, reg, pasid);
+}
+
+/*
+ * PTE format:
+ * 63 P
+ * 62:59 reserved
+ * 58 D
+ * 57 G
+ * 56 T
+ * 55:54 M
+ * 53:52 SW
+ * 51:48 reserved for future
+ * 47:12 4k physical page base address
+ * 11:7 fragment
+ * 6 write
+ * 5 read
+ * 4 exe
+ * 3 Z
+ * 2 snooped
+ * 1 system
+ * 0 valid
+ *
+ * PDE format:
+ * 63 P
+ * 62:58 block fragment size
+ * 57 reserved
+ * 56 A
+ * 55:54 M
+ * 53:52 reserved
+ * 51:48 reserved for future
+ * 47:6 physical base address of PD or PTE
+ * 5:3 reserved
+ * 2 C
+ * 1 system
+ * 0 valid
+ */
+
+static void gmc_v12_0_get_vm_pde(struct amdgpu_device *adev, int level,
+				 uint64_t *addr, uint64_t *flags)
+{
+	if (!(*flags & AMDGPU_PDE_PTE_GFX12) && !(*flags & AMDGPU_PTE_SYSTEM))
+		*addr = adev->vm_manager.vram_base_offset + *addr -
+			adev->gmc.vram_start;
+	BUG_ON(*addr & 0xFFFF00000000003FULL);
+
+	if (!adev->gmc.translate_further)
+		return;
+
+	if (level == AMDGPU_VM_PDB1) {
+		/* Set the block fragment size */
+		if (!(*flags & AMDGPU_PDE_PTE_GFX12))
+			*flags |= AMDGPU_PDE_BFS_GFX12(0x9);
+
+	} else if (level == AMDGPU_VM_PDB0) {
+		if (*flags & AMDGPU_PDE_PTE_GFX12)
+			*flags &= ~AMDGPU_PDE_PTE_GFX12;
+	}
+}
+
+static void gmc_v12_0_get_vm_pte(struct amdgpu_device *adev,
+				 struct amdgpu_vm *vm,
+				 struct amdgpu_bo *bo,
+				 uint32_t vm_flags,
+				 uint64_t *flags)
+{
+	if (vm_flags & AMDGPU_VM_PAGE_EXECUTABLE)
+		*flags |= AMDGPU_PTE_EXECUTABLE;
+	else
+		*flags &= ~AMDGPU_PTE_EXECUTABLE;
+
+	switch (vm_flags & AMDGPU_VM_MTYPE_MASK) {
+	case AMDGPU_VM_MTYPE_DEFAULT:
+		*flags = AMDGPU_PTE_MTYPE_GFX12(*flags, MTYPE_NC);
+		break;
+	case AMDGPU_VM_MTYPE_NC:
+	default:
+		*flags = AMDGPU_PTE_MTYPE_GFX12(*flags, MTYPE_NC);
+		break;
+	case AMDGPU_VM_MTYPE_UC:
+		*flags = AMDGPU_PTE_MTYPE_GFX12(*flags, MTYPE_UC);
+		break;
+	}
+
+	if (vm_flags & AMDGPU_VM_PAGE_NOALLOC)
+		*flags |= AMDGPU_PTE_NOALLOC;
+	else
+		*flags &= ~AMDGPU_PTE_NOALLOC;
+
+	if (vm_flags & AMDGPU_VM_PAGE_PRT) {
+		*flags |= AMDGPU_PTE_PRT_GFX12;
+		*flags |= AMDGPU_PTE_SNOOPED;
+		*flags |= AMDGPU_PTE_SYSTEM;
+		*flags |= AMDGPU_PTE_IS_PTE;
+		*flags &= ~AMDGPU_PTE_VALID;
+	}
+
+	if (bo && bo->flags & AMDGPU_GEM_CREATE_GFX12_DCC)
+		*flags |= AMDGPU_PTE_DCC;
+
+	if (bo && bo->flags & AMDGPU_GEM_CREATE_UNCACHED)
+		*flags = AMDGPU_PTE_MTYPE_GFX12(*flags, MTYPE_UC);
+}
+
+static unsigned gmc_v12_0_get_vbios_fb_size(struct amdgpu_device *adev)
+{
+	return 0;
+}
+
+static unsigned int gmc_v12_0_get_dcc_alignment(struct amdgpu_device *adev)
+{
+	unsigned int max_tex_channel_caches, alignment;
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(12, 0, 0) &&
+	    amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(12, 0, 1))
+		return 0;
+
+	max_tex_channel_caches = adev->gfx.config.max_texture_channel_caches;
+	if (is_power_of_2(max_tex_channel_caches))
+		alignment = (unsigned int)(max_tex_channel_caches / SZ_4);
+	else
+		alignment = roundup_pow_of_two(max_tex_channel_caches);
+
+	return (unsigned int)(alignment * max_tex_channel_caches * SZ_1K);
+}
+
+static const struct amdgpu_gmc_funcs gmc_v12_0_gmc_funcs = {
+	.flush_gpu_tlb = gmc_v12_0_flush_gpu_tlb,
+	.flush_gpu_tlb_pasid = gmc_v12_0_flush_gpu_tlb_pasid,
+	.emit_flush_gpu_tlb = gmc_v12_0_emit_flush_gpu_tlb,
+	.emit_pasid_mapping = gmc_v12_0_emit_pasid_mapping,
+	.get_vm_pde = gmc_v12_0_get_vm_pde,
+	.get_vm_pte = gmc_v12_0_get_vm_pte,
+	.get_vbios_fb_size = gmc_v12_0_get_vbios_fb_size,
+	.get_dcc_alignment = gmc_v12_0_get_dcc_alignment,
+};
+
+static void gmc_v12_0_set_gmc_funcs(struct amdgpu_device *adev)
+{
+	adev->gmc.gmc_funcs = &gmc_v12_0_gmc_funcs;
+}
+
+static void gmc_v12_0_set_umc_funcs(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, UMC_HWIP, 0)) {
+	case IP_VERSION(8, 14, 0):
+		adev->umc.channel_inst_num = UMC_V8_14_CHANNEL_INSTANCE_NUM;
+		adev->umc.umc_inst_num = UMC_V8_14_UMC_INSTANCE_NUM(adev);
+		adev->umc.node_inst_num = 0;
+		adev->umc.max_ras_err_cnt_per_query = UMC_V8_14_TOTAL_CHANNEL_NUM(adev);
+		adev->umc.channel_offs = UMC_V8_14_PER_CHANNEL_OFFSET;
+		adev->umc.ras = &umc_v8_14_ras;
+		break;
+	default:
+		break;
+	}
+}
+
+
+static void gmc_v12_0_set_mmhub_funcs(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) {
+	case IP_VERSION(4, 1, 0):
+		adev->mmhub.funcs = &mmhub_v4_1_0_funcs;
+		break;
+	default:
+		break;
+	}
+}
+
+static void gmc_v12_0_set_gfxhub_funcs(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 0, 1):
+		adev->gfxhub.funcs = &gfxhub_v12_0_funcs;
+		break;
+	default:
+		break;
+	}
+}
+
+static int gmc_v12_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	gmc_v12_0_set_gfxhub_funcs(adev);
+	gmc_v12_0_set_mmhub_funcs(adev);
+	gmc_v12_0_set_gmc_funcs(adev);
+	gmc_v12_0_set_irq_funcs(adev);
+	gmc_v12_0_set_umc_funcs(adev);
+
+	adev->gmc.shared_aperture_start = 0x2000000000000000ULL;
+	adev->gmc.shared_aperture_end =
+		adev->gmc.shared_aperture_start + (4ULL << 30) - 1;
+	adev->gmc.private_aperture_start = 0x1000000000000000ULL;
+	adev->gmc.private_aperture_end =
+		adev->gmc.private_aperture_start + (4ULL << 30) - 1;
+
+	return 0;
+}
+
+static int gmc_v12_0_late_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = amdgpu_gmc_allocate_vm_inv_eng(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_gmc_ras_late_init(adev);
+	if (r)
+		return r;
+
+	return amdgpu_irq_get(adev, &adev->gmc.vm_fault, 0);
+}
+
+static void gmc_v12_0_vram_gtt_location(struct amdgpu_device *adev,
+					struct amdgpu_gmc *mc)
+{
+	u64 base = 0;
+
+	base = adev->mmhub.funcs->get_fb_location(adev);
+
+	amdgpu_gmc_set_agp_default(adev, mc);
+	amdgpu_gmc_vram_location(adev, &adev->gmc, base);
+	amdgpu_gmc_gart_location(adev, mc, AMDGPU_GART_PLACEMENT_LOW);
+	if (!amdgpu_sriov_vf(adev) && (amdgpu_agp == 1))
+		amdgpu_gmc_agp_location(adev, mc);
+
+	/* base offset of vram pages */
+	if (amdgpu_sriov_vf(adev))
+		adev->vm_manager.vram_base_offset = 0;
+	else
+		adev->vm_manager.vram_base_offset = adev->mmhub.funcs->get_mc_fb_offset(adev);
+}
+
+/**
+ * gmc_v12_0_mc_init - initialize the memory controller driver params
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Look up the amount of vram, vram width, and decide how to place
+ * vram and gart within the GPU's physical address space.
+ * Returns 0 for success.
+ */
+static int gmc_v12_0_mc_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	/* size in MB on si */
+	adev->gmc.mc_vram_size =
+		adev->nbio.funcs->get_memsize(adev) * 1024ULL * 1024ULL;
+	adev->gmc.real_vram_size = adev->gmc.mc_vram_size;
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		r = amdgpu_device_resize_fb_bar(adev);
+		if (r)
+			return r;
+	}
+
+	adev->gmc.aper_base = pci_resource_start(adev->pdev, 0);
+	adev->gmc.aper_size = pci_resource_len(adev->pdev, 0);
+
+#ifdef CONFIG_X86_64
+	if ((adev->flags & AMD_IS_APU) && !amdgpu_passthrough(adev)) {
+		adev->gmc.aper_base = adev->mmhub.funcs->get_mc_fb_offset(adev);
+		adev->gmc.aper_size = adev->gmc.real_vram_size;
+	}
+#endif
+	/* In case the PCI BAR is larger than the actual amount of vram */
+	adev->gmc.visible_vram_size = adev->gmc.aper_size;
+	if (adev->gmc.visible_vram_size > adev->gmc.real_vram_size)
+		adev->gmc.visible_vram_size = adev->gmc.real_vram_size;
+
+	/* set the gart size */
+	if (amdgpu_gart_size == -1) {
+		adev->gmc.gart_size = 512ULL << 20;
+	} else
+		adev->gmc.gart_size = (u64)amdgpu_gart_size << 20;
+
+	gmc_v12_0_vram_gtt_location(adev, &adev->gmc);
+
+	return 0;
+}
+
+static int gmc_v12_0_gart_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->gart.bo) {
+		WARN(1, "PCIE GART already initialized\n");
+		return 0;
+	}
+
+	/* Initialize common gart structure */
+	r = amdgpu_gart_init(adev);
+	if (r)
+		return r;
+
+	adev->gart.table_size = adev->gart.num_gpu_pages * 8;
+	adev->gart.gart_pte_flags = AMDGPU_PTE_MTYPE_GFX12(0ULL, MTYPE_UC) |
+				    AMDGPU_PTE_EXECUTABLE |
+				    AMDGPU_PTE_IS_PTE;
+
+	return amdgpu_gart_table_vram_alloc(adev);
+}
+
+static int gmc_v12_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	int r, vram_width = 0, vram_type = 0, vram_vendor = 0;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	adev->mmhub.funcs->init(adev);
+
+	adev->gfxhub.funcs->init(adev);
+
+	spin_lock_init(&adev->gmc.invalidate_lock);
+
+	r = amdgpu_atomfirmware_get_vram_info(adev,
+					      &vram_width, &vram_type, &vram_vendor);
+	adev->gmc.vram_width = vram_width;
+
+	adev->gmc.vram_type = vram_type;
+	adev->gmc.vram_vendor = vram_vendor;
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 0, 1):
+		set_bit(AMDGPU_GFXHUB(0), adev->vmhubs_mask);
+		set_bit(AMDGPU_MMHUB0(0), adev->vmhubs_mask);
+		/*
+		 * To fulfill 4-level page support,
+		 * vm size is 256TB (48bit), maximum size,
+		 * block size 512 (9bit)
+		 */
+		amdgpu_vm_adjust_size(adev, 256 * 1024, 9, 3, 48);
+		break;
+	default:
+		break;
+	}
+
+	/* This interrupt is VMC page fault.*/
+	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_VMC,
+			      VMC_1_0__SRCID__VM_FAULT,
+			      &adev->gmc.vm_fault);
+
+	if (r)
+		return r;
+
+	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GFX,
+			      UTCL2_1_0__SRCID__FAULT,
+			      &adev->gmc.vm_fault);
+	if (r)
+		return r;
+
+	if (!amdgpu_sriov_vf(adev)) {
+		/* interrupt sent to DF. */
+		r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_DF, 0,
+				      &adev->gmc.ecc_irq);
+		if (r)
+			return r;
+	}
+
+	/*
+	 * Set the internal MC address mask This is the max address of the GPU's
+	 * internal address space.
+	 */
+	adev->gmc.mc_mask = 0xffffffffffffULL; /* 48 bit MC */
+
+	r = dma_set_mask_and_coherent(adev->dev, DMA_BIT_MASK(44));
+	if (r) {
+		printk(KERN_WARNING "amdgpu: No suitable DMA available.\n");
+		return r;
+	}
+
+	adev->need_swiotlb = drm_need_swiotlb(44);
+
+	r = gmc_v12_0_mc_init(adev);
+	if (r)
+		return r;
+
+	amdgpu_gmc_get_vbios_allocations(adev);
+
+	/* Memory manager */
+	r = amdgpu_bo_init(adev);
+	if (r)
+		return r;
+
+	r = gmc_v12_0_gart_init(adev);
+	if (r)
+		return r;
+
+	/*
+	 * number of VMs
+	 * VMID 0 is reserved for System
+	 * amdgpu graphics/compute will use VMIDs 1-7
+	 * amdkfd will use VMIDs 8-15
+	 */
+	adev->vm_manager.first_kfd_vmid = adev->gfx.disable_kq ? 1 : 8;
+
+	amdgpu_vm_manager_init(adev);
+
+	r = amdgpu_gmc_ras_sw_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+/**
+ * gmc_v12_0_gart_fini - vm fini callback
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tears down the driver GART/VM setup (CIK).
+ */
+static void gmc_v12_0_gart_fini(struct amdgpu_device *adev)
+{
+	amdgpu_gart_table_vram_free(adev);
+}
+
+static int gmc_v12_0_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	amdgpu_vm_manager_fini(adev);
+	gmc_v12_0_gart_fini(adev);
+	amdgpu_gem_force_release(adev);
+	amdgpu_bo_fini(adev);
+
+	return 0;
+}
+
+static void gmc_v12_0_init_golden_registers(struct amdgpu_device *adev)
+{
+}
+
+/**
+ * gmc_v12_0_gart_enable - gart enable
+ *
+ * @adev: amdgpu_device pointer
+ */
+static int gmc_v12_0_gart_enable(struct amdgpu_device *adev)
+{
+	int r;
+	bool value;
+
+	if (adev->gart.bo == NULL) {
+		dev_err(adev->dev, "No VRAM object for PCIE GART.\n");
+		return -EINVAL;
+	}
+
+	amdgpu_gtt_mgr_recover(&adev->mman.gtt_mgr);
+
+	r = adev->mmhub.funcs->gart_enable(adev);
+	if (r)
+		return r;
+
+	/* Flush HDP after it is initialized */
+	amdgpu_device_flush_hdp(adev, NULL);
+
+	value = amdgpu_vm_fault_stop != AMDGPU_VM_FAULT_STOP_ALWAYS;
+
+	adev->mmhub.funcs->set_fault_enable_default(adev, value);
+	gmc_v12_0_flush_gpu_tlb(adev, 0, AMDGPU_MMHUB0(0), 0);
+
+	dev_info(adev->dev, "PCIE GART of %uM enabled (table at 0x%016llX).\n",
+		 (unsigned)(adev->gmc.gart_size >> 20),
+		 (unsigned long long)amdgpu_bo_gpu_offset(adev->gart.bo));
+
+	return 0;
+}
+
+static int gmc_v12_0_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	/* The sequence of these two function calls matters.*/
+	gmc_v12_0_init_golden_registers(adev);
+
+	r = gmc_v12_0_gart_enable(adev);
+	if (r)
+		return r;
+
+	if (adev->umc.funcs && adev->umc.funcs->init_registers)
+		adev->umc.funcs->init_registers(adev);
+
+	return 0;
+}
+
+/**
+ * gmc_v12_0_gart_disable - gart disable
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * This disables all VM page table.
+ */
+static void gmc_v12_0_gart_disable(struct amdgpu_device *adev)
+{
+	adev->mmhub.funcs->gart_disable(adev);
+}
+
+static int gmc_v12_0_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_sriov_vf(adev)) {
+		/* full access mode, so don't touch any GMC register */
+		DRM_DEBUG("For SRIOV client, shouldn't do anything.\n");
+		return 0;
+	}
+
+	amdgpu_irq_put(adev, &adev->gmc.vm_fault, 0);
+
+	if (adev->gmc.ecc_irq.funcs &&
+		amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__UMC))
+		amdgpu_irq_put(adev, &adev->gmc.ecc_irq, 0);
+
+	gmc_v12_0_gart_disable(adev);
+
+	return 0;
+}
+
+static int gmc_v12_0_suspend(struct amdgpu_ip_block *ip_block)
+{
+	gmc_v12_0_hw_fini(ip_block);
+
+	return 0;
+}
+
+static int gmc_v12_0_resume(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	r = gmc_v12_0_hw_init(ip_block);
+	if (r)
+		return r;
+
+	amdgpu_vmid_reset_all(ip_block->adev);
+
+	return 0;
+}
+
+static bool gmc_v12_0_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	/* MC is always ready in GMC v11.*/
+	return true;
+}
+
+static int gmc_v12_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	/* There is no need to wait for MC idle in GMC v11.*/
+	return 0;
+}
+
+static int gmc_v12_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					   enum amd_clockgating_state state)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	r = adev->mmhub.funcs->set_clockgating(adev, state);
+	if (r)
+		return r;
+
+	return athub_v4_1_0_set_clockgating(adev, state);
+}
+
+static void gmc_v12_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	adev->mmhub.funcs->get_clockgating(adev, flags);
+
+	athub_v4_1_0_get_clockgating(adev, flags);
+}
+
+static int gmc_v12_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					   enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs gmc_v12_0_ip_funcs = {
+	.name = "gmc_v12_0",
+	.early_init = gmc_v12_0_early_init,
+	.sw_init = gmc_v12_0_sw_init,
+	.hw_init = gmc_v12_0_hw_init,
+	.late_init = gmc_v12_0_late_init,
+	.sw_fini = gmc_v12_0_sw_fini,
+	.hw_fini = gmc_v12_0_hw_fini,
+	.suspend = gmc_v12_0_suspend,
+	.resume = gmc_v12_0_resume,
+	.is_idle = gmc_v12_0_is_idle,
+	.wait_for_idle = gmc_v12_0_wait_for_idle,
+	.set_clockgating_state = gmc_v12_0_set_clockgating_state,
+	.set_powergating_state = gmc_v12_0_set_powergating_state,
+	.get_clockgating_state = gmc_v12_0_get_clockgating_state,
+};
+
+const struct amdgpu_ip_block_version gmc_v12_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_GMC,
+	.major = 12,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &gmc_v12_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v12_0.h b/drivers/gpu/drm/amd/amdgpu/gmc_v12_0.h
new file mode 100644
index 000000000000..deca93e4a156
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v12_0.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GMC_V12_0_H__
+#define __GMC_V12_0_H__
+
+extern const struct amd_ip_funcs gmc_v12_0_ip_funcs;
+extern const struct amdgpu_ip_block_version gmc_v12_0_ip_block;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c
index 5be9c83dfcf7..f6ad7911f1e6 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c
@@ -20,11 +20,16 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  *
  */
+
 #include <linux/firmware.h>
-#include <drm/drmP.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+
+#include <drm/drm_cache.h>
 #include "amdgpu.h"
 #include "gmc_v6_0.h"
 #include "amdgpu_ucode.h"
+#include "amdgpu_gem.h"
 
 #include "bif/bif_3_0_d.h"
 #include "bif/bif_3_0_sh_mask.h"
@@ -36,15 +41,16 @@
 #include "dce/dce_6_0_sh_mask.h"
 #include "si_enums.h"
 
-static void gmc_v6_0_set_gart_funcs(struct amdgpu_device *adev);
+static void gmc_v6_0_set_gmc_funcs(struct amdgpu_device *adev);
 static void gmc_v6_0_set_irq_funcs(struct amdgpu_device *adev);
-static int gmc_v6_0_wait_for_idle(void *handle);
+static int gmc_v6_0_wait_for_idle(struct amdgpu_ip_block *ip_block);
 
-MODULE_FIRMWARE("radeon/tahiti_mc.bin");
-MODULE_FIRMWARE("radeon/pitcairn_mc.bin");
-MODULE_FIRMWARE("radeon/verde_mc.bin");
-MODULE_FIRMWARE("radeon/oland_mc.bin");
-MODULE_FIRMWARE("radeon/si58_mc.bin");
+MODULE_FIRMWARE("amdgpu/tahiti_mc.bin");
+MODULE_FIRMWARE("amdgpu/pitcairn_mc.bin");
+MODULE_FIRMWARE("amdgpu/verde_mc.bin");
+MODULE_FIRMWARE("amdgpu/oland_mc.bin");
+MODULE_FIRMWARE("amdgpu/hainan_mc.bin");
+MODULE_FIRMWARE("amdgpu/si58_mc.bin");
 
 #define MC_SEQ_MISC0__MT__MASK   0xf0000000
 #define MC_SEQ_MISC0__MT__GDDR1  0x10000000
@@ -55,22 +61,16 @@ MODULE_FIRMWARE("radeon/si58_mc.bin");
 #define MC_SEQ_MISC0__MT__HBM    0x60000000
 #define MC_SEQ_MISC0__MT__DDR3   0xB0000000
 
-
-static const u32 crtc_offsets[6] =
-{
-	SI_CRTC0_REGISTER_OFFSET,
-	SI_CRTC1_REGISTER_OFFSET,
-	SI_CRTC2_REGISTER_OFFSET,
-	SI_CRTC3_REGISTER_OFFSET,
-	SI_CRTC4_REGISTER_OFFSET,
-	SI_CRTC5_REGISTER_OFFSET
-};
-
 static void gmc_v6_0_mc_stop(struct amdgpu_device *adev)
 {
 	u32 blackout;
+	struct amdgpu_ip_block *ip_block;
+
+	ip_block = amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_GMC);
+	if (!ip_block)
+		return;
 
-	gmc_v6_0_wait_for_idle((void *)adev);
+	gmc_v6_0_wait_for_idle(ip_block);
 
 	blackout = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
 	if (REG_GET_FIELD(blackout, MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE) != 1) {
@@ -103,9 +103,7 @@ static void gmc_v6_0_mc_resume(struct amdgpu_device *adev)
 static int gmc_v6_0_init_microcode(struct amdgpu_device *adev)
 {
 	const char *chip_name;
-	char fw_name[30];
 	int err;
-	bool is_58_fw = false;
 
 	DRM_DEBUG("\n");
 
@@ -125,30 +123,21 @@ static int gmc_v6_0_init_microcode(struct amdgpu_device *adev)
 	case CHIP_HAINAN:
 		chip_name = "hainan";
 		break;
-	default: BUG();
+	default:
+		BUG();
 	}
 
 	/* this memory configuration requires special firmware */
 	if (((RREG32(mmMC_SEQ_MISC0) & 0xff000000) >> 24) == 0x58)
-		is_58_fw = true;
-
-	if (is_58_fw)
-		snprintf(fw_name, sizeof(fw_name), "radeon/si58_mc.bin");
-	else
-		snprintf(fw_name, sizeof(fw_name), "radeon/%s_mc.bin", chip_name);
-	err = request_firmware(&adev->mc.fw, fw_name, adev->dev);
-	if (err)
-		goto out;
+		chip_name = "si58";
 
-	err = amdgpu_ucode_validate(adev->mc.fw);
-
-out:
+	err = amdgpu_ucode_request(adev, &adev->gmc.fw, AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_mc.bin", chip_name);
 	if (err) {
 		dev_err(adev->dev,
-		       "si_mc: Failed to load firmware \"%s\"\n",
-		       fw_name);
-		release_firmware(adev->mc.fw);
-		adev->mc.fw = NULL;
+		       "si_mc: Failed to load firmware \"%s_mc.bin\"\n",
+		       chip_name);
+		amdgpu_ucode_release(&adev->gmc.fw);
 	}
 	return err;
 }
@@ -161,20 +150,20 @@ static int gmc_v6_0_mc_load_microcode(struct amdgpu_device *adev)
 	int i, regs_size, ucode_size;
 	const struct mc_firmware_header_v1_0 *hdr;
 
-	if (!adev->mc.fw)
+	if (!adev->gmc.fw)
 		return -EINVAL;
 
-	hdr = (const struct mc_firmware_header_v1_0 *)adev->mc.fw->data;
+	hdr = (const struct mc_firmware_header_v1_0 *)adev->gmc.fw->data;
 
 	amdgpu_ucode_print_mc_hdr(&hdr->header);
 
-	adev->mc.fw_version = le32_to_cpu(hdr->header.ucode_version);
+	adev->gmc.fw_version = le32_to_cpu(hdr->header.ucode_version);
 	regs_size = le32_to_cpu(hdr->io_debug_size_bytes) / (4 * 2);
 	new_io_mc_regs = (const __le32 *)
-		(adev->mc.fw->data + le32_to_cpu(hdr->io_debug_array_offset_bytes));
+		(adev->gmc.fw->data + le32_to_cpu(hdr->io_debug_array_offset_bytes));
 	ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
 	new_fw_data = (const __le32 *)
-		(adev->mc.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+		(adev->gmc.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
 
 	running = RREG32(mmMC_SEQ_SUP_CNTL) & MC_SEQ_SUP_CNTL__RUN_MASK;
 
@@ -190,9 +179,8 @@ static int gmc_v6_0_mc_load_microcode(struct amdgpu_device *adev)
 			WREG32(mmMC_SEQ_IO_DEBUG_DATA, le32_to_cpup(new_io_mc_regs++));
 		}
 		/* load the MC ucode */
-		for (i = 0; i < ucode_size; i++) {
+		for (i = 0; i < ucode_size; i++)
 			WREG32(mmMC_SEQ_SUP_PGM, le32_to_cpup(new_fw_data++));
-		}
 
 		/* put the engine back into the active state */
 		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000008);
@@ -217,23 +205,22 @@ static int gmc_v6_0_mc_load_microcode(struct amdgpu_device *adev)
 }
 
 static void gmc_v6_0_vram_gtt_location(struct amdgpu_device *adev,
-				       struct amdgpu_mc *mc)
+				       struct amdgpu_gmc *mc)
 {
 	u64 base = RREG32(mmMC_VM_FB_LOCATION) & 0xFFFF;
+
 	base <<= 24;
 
-	if (mc->mc_vram_size > 0xFFC0000000ULL) {
-		dev_warn(adev->dev, "limiting VRAM\n");
-		mc->real_vram_size = 0xFFC0000000ULL;
-		mc->mc_vram_size = 0xFFC0000000ULL;
-	}
-	amdgpu_vram_location(adev, &adev->mc, base);
-	amdgpu_gart_location(adev, mc);
+	amdgpu_gmc_set_agp_default(adev, mc);
+	amdgpu_gmc_vram_location(adev, mc, base);
+	amdgpu_gmc_gart_location(adev, mc, AMDGPU_GART_PLACEMENT_BEST_FIT);
 }
 
 static void gmc_v6_0_mc_program(struct amdgpu_device *adev)
 {
 	int i, j;
+	struct amdgpu_ip_block *ip_block;
+
 
 	/* Initialize HDP */
 	for (i = 0, j = 0; i < 32; i++, j += 0x6) {
@@ -245,9 +232,12 @@ static void gmc_v6_0_mc_program(struct amdgpu_device *adev)
 	}
 	WREG32(mmHDP_REG_COHERENCY_FLUSH_CNTL, 0);
 
-	if (gmc_v6_0_wait_for_idle((void *)adev)) {
+	ip_block = amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_GMC);
+	if (!ip_block)
+		return;
+
+	if (gmc_v6_0_wait_for_idle(ip_block))
 		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
-	}
 
 	if (adev->mode_info.num_crtc) {
 		u32 tmp;
@@ -259,23 +249,22 @@ static void gmc_v6_0_mc_program(struct amdgpu_device *adev)
 
 		/* disable VGA render */
 		tmp = RREG32(mmVGA_RENDER_CONTROL);
-		tmp &= ~VGA_VSTATUS_CNTL;
+		tmp &= VGA_RENDER_CONTROL__VGA_VSTATUS_CNTL_MASK;
 		WREG32(mmVGA_RENDER_CONTROL, tmp);
 	}
 	/* Update configuration */
 	WREG32(mmMC_VM_SYSTEM_APERTURE_LOW_ADDR,
-	       adev->mc.vram_start >> 12);
+	       adev->gmc.vram_start >> 12);
 	WREG32(mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
-	       adev->mc.vram_end >> 12);
+	       adev->gmc.vram_end >> 12);
 	WREG32(mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR,
-	       adev->vram_scratch.gpu_addr >> 12);
+	       adev->mem_scratch.gpu_addr >> 12);
 	WREG32(mmMC_VM_AGP_BASE, 0);
-	WREG32(mmMC_VM_AGP_TOP, 0x0FFFFFFF);
-	WREG32(mmMC_VM_AGP_BOT, 0x0FFFFFFF);
+	WREG32(mmMC_VM_AGP_TOP, adev->gmc.agp_end >> 22);
+	WREG32(mmMC_VM_AGP_BOT, adev->gmc.agp_start >> 22);
 
-	if (gmc_v6_0_wait_for_idle((void *)adev)) {
+	if (gmc_v6_0_wait_for_idle(ip_block))
 		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
-	}
 }
 
 static int gmc_v6_0_mc_init(struct amdgpu_device *adev)
@@ -283,15 +272,16 @@ static int gmc_v6_0_mc_init(struct amdgpu_device *adev)
 
 	u32 tmp;
 	int chansize, numchan;
+	int r;
 
 	tmp = RREG32(mmMC_ARB_RAMCFG);
-	if (tmp & (1 << 11)) {
+	if (tmp & (1 << 11))
 		chansize = 16;
-	} else if (tmp & MC_ARB_RAMCFG__CHANSIZE_MASK) {
+	else if (tmp & MC_ARB_RAMCFG__CHANSIZE_MASK)
 		chansize = 64;
-	} else {
+	else
 		chansize = 32;
-	}
+
 	tmp = RREG32(mmMC_SHARED_CHMAP);
 	switch ((tmp & MC_SHARED_CHMAP__NOOFCHAN_MASK) >> MC_SHARED_CHMAP__NOOFCHAN__SHIFT) {
 	case 0:
@@ -323,81 +313,82 @@ static int gmc_v6_0_mc_init(struct amdgpu_device *adev)
 		numchan = 16;
 		break;
 	}
-	adev->mc.vram_width = numchan * chansize;
-	/* Could aper size report 0 ? */
-	adev->mc.aper_base = pci_resource_start(adev->pdev, 0);
-	adev->mc.aper_size = pci_resource_len(adev->pdev, 0);
+	adev->gmc.vram_width = numchan * chansize;
 	/* size in MB on si */
-	adev->mc.mc_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
-	adev->mc.real_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
-	adev->mc.visible_vram_size = adev->mc.aper_size;
+	adev->gmc.mc_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
+	adev->gmc.real_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		r = amdgpu_device_resize_fb_bar(adev);
+		if (r)
+			return r;
+	}
+	adev->gmc.aper_base = pci_resource_start(adev->pdev, 0);
+	adev->gmc.aper_size = pci_resource_len(adev->pdev, 0);
+	adev->gmc.visible_vram_size = adev->gmc.aper_size;
 
 	/* set the gart size */
 	if (amdgpu_gart_size == -1) {
 		switch (adev->asic_type) {
 		case CHIP_HAINAN:    /* no MM engines */
 		default:
-			adev->mc.gart_size = 256ULL << 20;
+			adev->gmc.gart_size = 256ULL << 20;
 			break;
 		case CHIP_VERDE:    /* UVD, VCE do not support GPUVM */
 		case CHIP_TAHITI:   /* UVD, VCE do not support GPUVM */
 		case CHIP_PITCAIRN: /* UVD, VCE do not support GPUVM */
 		case CHIP_OLAND:    /* UVD, VCE do not support GPUVM */
-			adev->mc.gart_size = 1024ULL << 20;
+			adev->gmc.gart_size = 1024ULL << 20;
 			break;
 		}
 	} else {
-		adev->mc.gart_size = (u64)amdgpu_gart_size << 20;
+		adev->gmc.gart_size = (u64)amdgpu_gart_size << 20;
 	}
 
-	gmc_v6_0_vram_gtt_location(adev, &adev->mc);
+	adev->gmc.gart_size += adev->pm.smu_prv_buffer_size;
+	gmc_v6_0_vram_gtt_location(adev, &adev->gmc);
 
 	return 0;
 }
 
-static void gmc_v6_0_gart_flush_gpu_tlb(struct amdgpu_device *adev,
-					uint32_t vmid)
+static void gmc_v6_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid,
+					uint32_t vmhub, uint32_t flush_type)
 {
-	WREG32(mmHDP_MEM_COHERENCY_FLUSH_CNTL, 0);
-
 	WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
 }
 
-static int gmc_v6_0_gart_set_pte_pde(struct amdgpu_device *adev,
-				     void *cpu_pt_addr,
-				     uint32_t gpu_page_idx,
-				     uint64_t addr,
-				     uint64_t flags)
+static uint64_t gmc_v6_0_emit_flush_gpu_tlb(struct amdgpu_ring *ring,
+					    unsigned int vmid, uint64_t pd_addr)
 {
-	void __iomem *ptr = (void *)cpu_pt_addr;
-	uint64_t value;
+	uint32_t reg;
 
-	value = addr & 0xFFFFFFFFFFFFF000ULL;
-	value |= flags;
-	writeq(value, ptr + (gpu_page_idx * 8));
+	/* write new base address */
+	if (vmid < 8)
+		reg = mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + vmid;
+	else
+		reg = mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + (vmid - 8);
+	amdgpu_ring_emit_wreg(ring, reg, pd_addr >> 12);
 
-	return 0;
+	/* bits 0-15 are the VM contexts0-15 */
+	amdgpu_ring_emit_wreg(ring, mmVM_INVALIDATE_REQUEST, 1 << vmid);
+
+	return pd_addr;
 }
 
-static uint64_t gmc_v6_0_get_vm_pte_flags(struct amdgpu_device *adev,
-					  uint32_t flags)
+static void gmc_v6_0_get_vm_pde(struct amdgpu_device *adev, int level,
+				uint64_t *addr, uint64_t *flags)
 {
-	uint64_t pte_flag = 0;
-
-	if (flags & AMDGPU_VM_PAGE_READABLE)
-		pte_flag |= AMDGPU_PTE_READABLE;
-	if (flags & AMDGPU_VM_PAGE_WRITEABLE)
-		pte_flag |= AMDGPU_PTE_WRITEABLE;
-	if (flags & AMDGPU_VM_PAGE_PRT)
-		pte_flag |= AMDGPU_PTE_PRT;
-
-	return pte_flag;
+	BUG_ON(*addr & 0xFFFFFF0000000FFFULL);
 }
 
-static uint64_t gmc_v6_0_get_vm_pde(struct amdgpu_device *adev, uint64_t addr)
+static void gmc_v6_0_get_vm_pte(struct amdgpu_device *adev,
+				struct amdgpu_vm *vm,
+				struct amdgpu_bo *bo,
+				uint32_t vm_flags,
+				uint64_t *flags)
 {
-	BUG_ON(addr & 0xFFFFFF0000000FFFULL);
-	return addr;
+	*flags &= ~AMDGPU_PTE_EXECUTABLE;
+	*flags &= ~AMDGPU_PTE_PRT;
 }
 
 static void gmc_v6_0_set_fault_enable_default(struct amdgpu_device *adev,
@@ -422,18 +413,18 @@ static void gmc_v6_0_set_fault_enable_default(struct amdgpu_device *adev,
 }
 
  /**
-   + * gmc_v8_0_set_prt - set PRT VM fault
-   + *
-   + * @adev: amdgpu_device pointer
-   + * @enable: enable/disable VM fault handling for PRT
-   +*/
+  * gmc_v8_0_set_prt() - set PRT VM fault
+  *
+  * @adev: amdgpu_device pointer
+  * @enable: enable/disable VM fault handling for PRT
+  */
 static void gmc_v6_0_set_prt(struct amdgpu_device *adev, bool enable)
 {
 	u32 tmp;
 
-	if (enable && !adev->mc.prt_warning) {
+	if (enable && !adev->gmc.prt_warning) {
 		dev_warn(adev->dev, "Disabling VM faults because of PRT request!\n");
-		adev->mc.prt_warning = true;
+		adev->gmc.prt_warning = true;
 	}
 
 	tmp = RREG32(mmVM_PRT_CNTL);
@@ -452,8 +443,10 @@ static void gmc_v6_0_set_prt(struct amdgpu_device *adev, bool enable)
 	WREG32(mmVM_PRT_CNTL, tmp);
 
 	if (enable) {
-		uint32_t low = AMDGPU_VA_RESERVED_SIZE >> AMDGPU_GPU_PAGE_SHIFT;
-		uint32_t high = adev->vm_manager.max_pfn;
+		uint32_t low = AMDGPU_VA_RESERVED_BOTTOM >>
+			AMDGPU_GPU_PAGE_SHIFT;
+		uint32_t high = adev->vm_manager.max_pfn -
+			(AMDGPU_VA_RESERVED_TOP >> AMDGPU_GPU_PAGE_SHIFT);
 
 		WREG32(mmVM_PRT_APERTURE0_LOW_ADDR, low);
 		WREG32(mmVM_PRT_APERTURE1_LOW_ADDR, low);
@@ -477,16 +470,18 @@ static void gmc_v6_0_set_prt(struct amdgpu_device *adev, bool enable)
 
 static int gmc_v6_0_gart_enable(struct amdgpu_device *adev)
 {
-	int r, i;
+	uint64_t table_addr;
 	u32 field;
+	int i;
 
-	if (adev->gart.robj == NULL) {
+	if (adev->gart.bo == NULL) {
 		dev_err(adev->dev, "No VRAM object for PCIE GART.\n");
 		return -EINVAL;
 	}
-	r = amdgpu_gart_table_vram_pin(adev);
-	if (r)
-		return r;
+	amdgpu_gtt_mgr_recover(&adev->mman.gtt_mgr);
+
+	table_addr = amdgpu_bo_gpu_offset(adev->gart.bo);
+
 	/* Setup TLB control */
 	WREG32(mmMC_VM_MX_L1_TLB_CNTL,
 	       (0xA << 7) |
@@ -513,11 +508,11 @@ static int gmc_v6_0_gart_enable(struct amdgpu_device *adev)
 	       (field << VM_L2_CNTL3__BANK_SELECT__SHIFT) |
 	       (field << VM_L2_CNTL3__L2_CACHE_BIGK_FRAGMENT_SIZE__SHIFT));
 	/* setup context0 */
-	WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->mc.gart_start >> 12);
-	WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, adev->mc.gart_end >> 12);
-	WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR, adev->gart.table_addr >> 12);
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->gmc.gart_start >> 12);
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, adev->gmc.gart_end >> 12);
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR, table_addr >> 12);
 	WREG32(mmVM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR,
-			(u32)(adev->dummy_page.addr >> 12));
+			(u32)(adev->dummy_page_addr >> 12));
 	WREG32(mmVM_CONTEXT0_CNTL2, 0);
 	WREG32(mmVM_CONTEXT0_CNTL,
 	       VM_CONTEXT0_CNTL__ENABLE_CONTEXT_MASK |
@@ -536,18 +531,18 @@ static int gmc_v6_0_gart_enable(struct amdgpu_device *adev)
 	 * the VMs are determined by the application and setup and assigned
 	 * on the fly in the vm part of radeon_gart.c
 	 */
-	for (i = 1; i < 16; i++) {
+	for (i = 1; i < AMDGPU_NUM_VMID; i++) {
 		if (i < 8)
 			WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + i,
-			       adev->gart.table_addr >> 12);
+			       table_addr >> 12);
 		else
 			WREG32(mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + i - 8,
-			       adev->gart.table_addr >> 12);
+			       table_addr >> 12);
 	}
 
 	/* enable context1-15 */
 	WREG32(mmVM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR,
-	       (u32)(adev->dummy_page.addr >> 12));
+	       (u32)(adev->dummy_page_addr >> 12));
 	WREG32(mmVM_CONTEXT1_CNTL2, 4);
 	WREG32(mmVM_CONTEXT1_CNTL,
 	       VM_CONTEXT1_CNTL__ENABLE_CONTEXT_MASK |
@@ -559,11 +554,10 @@ static int gmc_v6_0_gart_enable(struct amdgpu_device *adev)
 	else
 		gmc_v6_0_set_fault_enable_default(adev, true);
 
-	gmc_v6_0_gart_flush_gpu_tlb(adev, 0);
+	gmc_v6_0_flush_gpu_tlb(adev, 0, 0, 0);
 	dev_info(adev->dev, "PCIE GART of %uM enabled (table at 0x%016llX).\n",
-		 (unsigned)(adev->mc.gart_size >> 20),
-		 (unsigned long long)adev->gart.table_addr);
-	adev->gart.ready = true;
+		 (unsigned int)(adev->gmc.gart_size >> 20),
+		 (unsigned long long)table_addr);
 	return 0;
 }
 
@@ -571,7 +565,7 @@ static int gmc_v6_0_gart_init(struct amdgpu_device *adev)
 {
 	int r;
 
-	if (adev->gart.robj) {
+	if (adev->gart.bo) {
 		dev_warn(adev->dev, "gmc_v6_0 PCIE GART already initialized\n");
 		return 0;
 	}
@@ -613,13 +607,6 @@ static void gmc_v6_0_gart_disable(struct amdgpu_device *adev)
 	WREG32(mmVM_L2_CNTL3,
 	       VM_L2_CNTL3__L2_CACHE_BIGK_ASSOCIATIVITY_MASK |
 	       (0UL << VM_L2_CNTL3__L2_CACHE_BIGK_FRAGMENT_SIZE__SHIFT));
-	amdgpu_gart_table_vram_unpin(adev);
-}
-
-static void gmc_v6_0_gart_fini(struct amdgpu_device *adev)
-{
-	amdgpu_gart_table_vram_free(adev);
-	amdgpu_gart_fini(adev);
 }
 
 static void gmc_v6_0_vm_decode_fault(struct amdgpu_device *adev,
@@ -642,17 +629,16 @@ static void gmc_v6_0_vm_decode_fault(struct amdgpu_device *adev,
 	       "write" : "read", block, mc_client, mc_id);
 }
 
-/*
 static const u32 mc_cg_registers[] = {
-	MC_HUB_MISC_HUB_CG,
-	MC_HUB_MISC_SIP_CG,
-	MC_HUB_MISC_VM_CG,
-	MC_XPB_CLK_GAT,
-	ATC_MISC_CG,
-	MC_CITF_MISC_WR_CG,
-	MC_CITF_MISC_RD_CG,
-	MC_CITF_MISC_VM_CG,
-	VM_L2_CG,
+	mmMC_HUB_MISC_HUB_CG,
+	mmMC_HUB_MISC_SIP_CG,
+	mmMC_HUB_MISC_VM_CG,
+	mmMC_XPB_CLK_GAT,
+	mmATC_MISC_CG,
+	mmMC_CITF_MISC_WR_CG,
+	mmMC_CITF_MISC_RD_CG,
+	mmMC_CITF_MISC_VM_CG,
+	mmVM_L2_CG,
 };
 
 static const u32 mc_cg_ls_en[] = {
@@ -687,7 +673,7 @@ static void gmc_v6_0_enable_mc_ls(struct amdgpu_device *adev,
 
 	for (i = 0; i < ARRAY_SIZE(mc_cg_registers); i++) {
 		orig = data = RREG32(mc_cg_registers[i]);
-		if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_MC_LS))
+		if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_LS))
 			data |= mc_cg_ls_en[i];
 		else
 			data &= ~mc_cg_ls_en[i];
@@ -704,7 +690,7 @@ static void gmc_v6_0_enable_mc_mgcg(struct amdgpu_device *adev,
 
 	for (i = 0; i < ARRAY_SIZE(mc_cg_registers); i++) {
 		orig = data = RREG32(mc_cg_registers[i]);
-		if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_MC_MGCG))
+		if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG))
 			data |= mc_cg_en[i];
 		else
 			data &= ~mc_cg_en[i];
@@ -720,7 +706,7 @@ static void gmc_v6_0_enable_bif_mgls(struct amdgpu_device *adev,
 
 	orig = data = RREG32_PCIE(ixPCIE_CNTL2);
 
-	if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_BIF_LS)) {
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_BIF_LS)) {
 		data = REG_SET_FIELD(data, PCIE_CNTL2, SLV_MEM_LS_EN, 1);
 		data = REG_SET_FIELD(data, PCIE_CNTL2, MST_MEM_LS_EN, 1);
 		data = REG_SET_FIELD(data, PCIE_CNTL2, REPLAY_MEM_LS_EN, 1);
@@ -743,7 +729,7 @@ static void gmc_v6_0_enable_hdp_mgcg(struct amdgpu_device *adev,
 
 	orig = data = RREG32(mmHDP_HOST_PATH_CNTL);
 
-	if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_HDP_MGCG))
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_HDP_MGCG))
 		data = REG_SET_FIELD(data, HDP_HOST_PATH_CNTL, CLOCK_GATING_DIS, 0);
 	else
 		data = REG_SET_FIELD(data, HDP_HOST_PATH_CNTL, CLOCK_GATING_DIS, 1);
@@ -759,7 +745,7 @@ static void gmc_v6_0_enable_hdp_ls(struct amdgpu_device *adev,
 
 	orig = data = RREG32(mmHDP_MEM_POWER_LS);
 
-	if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_HDP_LS))
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS))
 		data = REG_SET_FIELD(data, HDP_MEM_POWER_LS, LS_ENABLE, 1);
 	else
 		data = REG_SET_FIELD(data, HDP_MEM_POWER_LS, LS_ENABLE, 0);
@@ -767,7 +753,6 @@ static void gmc_v6_0_enable_hdp_ls(struct amdgpu_device *adev,
 	if (orig != data)
 		WREG32(mmHDP_MEM_POWER_LS, data);
 }
-*/
 
 static int gmc_v6_0_convert_vram_type(int mc_seq_vram_type)
 {
@@ -789,68 +774,77 @@ static int gmc_v6_0_convert_vram_type(int mc_seq_vram_type)
 	}
 }
 
-static int gmc_v6_0_early_init(void *handle)
+static int gmc_v6_0_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	gmc_v6_0_set_gart_funcs(adev);
+	gmc_v6_0_set_gmc_funcs(adev);
 	gmc_v6_0_set_irq_funcs(adev);
 
 	return 0;
 }
 
-static int gmc_v6_0_late_init(void *handle)
+static int gmc_v6_0_late_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (amdgpu_vm_fault_stop != AMDGPU_VM_FAULT_STOP_ALWAYS)
-		return amdgpu_irq_get(adev, &adev->mc.vm_fault, 0);
+		return amdgpu_irq_get(adev, &adev->gmc.vm_fault, 0);
 	else
 		return 0;
 }
 
-static int gmc_v6_0_sw_init(void *handle)
+static unsigned int gmc_v6_0_get_vbios_fb_size(struct amdgpu_device *adev)
+{
+	u32 d1vga_control = RREG32(mmD1VGA_CONTROL);
+	unsigned int size;
+
+	if (REG_GET_FIELD(d1vga_control, D1VGA_CONTROL, D1VGA_MODE_ENABLE)) {
+		size = AMDGPU_VBIOS_VGA_ALLOCATION;
+	} else {
+		u32 viewport = RREG32(mmVIEWPORT_SIZE);
+
+		size = (REG_GET_FIELD(viewport, VIEWPORT_SIZE, VIEWPORT_HEIGHT) *
+			REG_GET_FIELD(viewport, VIEWPORT_SIZE, VIEWPORT_WIDTH) *
+			4);
+	}
+	return size;
+}
+
+static int gmc_v6_0_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	int dma_bits;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	set_bit(AMDGPU_GFXHUB(0), adev->vmhubs_mask);
 
 	if (adev->flags & AMD_IS_APU) {
-		adev->mc.vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
+		adev->gmc.vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
 	} else {
 		u32 tmp = RREG32(mmMC_SEQ_MISC0);
+
 		tmp &= MC_SEQ_MISC0__MT__MASK;
-		adev->mc.vram_type = gmc_v6_0_convert_vram_type(tmp);
+		adev->gmc.vram_type = gmc_v6_0_convert_vram_type(tmp);
 	}
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 146, &adev->mc.vm_fault);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 146, &adev->gmc.vm_fault);
 	if (r)
 		return r;
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 147, &adev->mc.vm_fault);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 147, &adev->gmc.vm_fault);
 	if (r)
 		return r;
 
-	amdgpu_vm_adjust_size(adev, 64, 4);
-	adev->vm_manager.max_pfn = adev->vm_manager.vm_size << 18;
+	amdgpu_vm_adjust_size(adev, 64, 9, 1, 40);
 
-	adev->mc.mc_mask = 0xffffffffffULL;
+	adev->gmc.mc_mask = 0xffffffffffULL;
 
-	adev->mc.stolen_size = 256 * 1024;
-
-	adev->need_dma32 = false;
-	dma_bits = adev->need_dma32 ? 32 : 40;
-	r = pci_set_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
-	if (r) {
-		adev->need_dma32 = true;
-		dma_bits = 32;
-		dev_warn(adev->dev, "amdgpu: No suitable DMA available.\n");
-	}
-	r = pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
+	r = dma_set_mask_and_coherent(adev->dev, DMA_BIT_MASK(40));
 	if (r) {
-		pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(32));
-		dev_warn(adev->dev, "amdgpu: No coherent DMA available.\n");
+		dev_warn(adev->dev, "No suitable DMA available.\n");
+		return r;
 	}
+	adev->need_swiotlb = drm_need_swiotlb(40);
 
 	r = gmc_v6_0_init_microcode(adev);
 	if (r) {
@@ -862,6 +856,8 @@ static int gmc_v6_0_sw_init(void *handle)
 	if (r)
 		return r;
 
+	amdgpu_gmc_get_vbios_allocations(adev);
+
 	r = amdgpu_bo_init(adev);
 	if (r)
 		return r;
@@ -876,8 +872,7 @@ static int gmc_v6_0_sw_init(void *handle)
 	 * amdgpu graphics/compute will use VMIDs 1-7
 	 * amdkfd will use VMIDs 8-15
 	 */
-	adev->vm_manager.id_mgr[0].num_ids = AMDGPU_NUM_OF_VMIDS;
-	adev->vm_manager.num_level = 1;
+	adev->vm_manager.first_kfd_vmid = 8;
 	amdgpu_vm_manager_init(adev);
 
 	/* base offset of vram pages */
@@ -893,22 +888,23 @@ static int gmc_v6_0_sw_init(void *handle)
 	return 0;
 }
 
-static int gmc_v6_0_sw_fini(void *handle)
+static int gmc_v6_0_sw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	amdgpu_vm_manager_fini(adev);
-	gmc_v6_0_gart_fini(adev);
 	amdgpu_gem_force_release(adev);
+	amdgpu_vm_manager_fini(adev);
+	amdgpu_gart_table_vram_free(adev);
 	amdgpu_bo_fini(adev);
+	amdgpu_ucode_release(&adev->gmc.fw);
 
 	return 0;
 }
 
-static int gmc_v6_0_hw_init(void *handle)
+static int gmc_v6_0_hw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	gmc_v6_0_mc_program(adev);
 
@@ -924,45 +920,47 @@ static int gmc_v6_0_hw_init(void *handle)
 	if (r)
 		return r;
 
-	return r;
+	if (amdgpu_emu_mode == 1)
+		return amdgpu_gmc_vram_checking(adev);
+
+	return 0;
 }
 
-static int gmc_v6_0_hw_fini(void *handle)
+static int gmc_v6_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	amdgpu_irq_put(adev, &adev->mc.vm_fault, 0);
+	amdgpu_irq_put(adev, &adev->gmc.vm_fault, 0);
 	gmc_v6_0_gart_disable(adev);
 
 	return 0;
 }
 
-static int gmc_v6_0_suspend(void *handle)
+static int gmc_v6_0_suspend(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	gmc_v6_0_hw_fini(adev);
+	gmc_v6_0_hw_fini(ip_block);
 
 	return 0;
 }
 
-static int gmc_v6_0_resume(void *handle)
+static int gmc_v6_0_resume(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	r = gmc_v6_0_hw_init(adev);
+	r = gmc_v6_0_hw_init(ip_block);
 	if (r)
 		return r;
 
-	amdgpu_vm_reset_all_ids(adev);
+	amdgpu_vmid_reset_all(adev);
 
 	return 0;
 }
 
-static bool gmc_v6_0_is_idle(void *handle)
+static bool gmc_v6_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+
 	u32 tmp = RREG32(mmSRBM_STATUS);
 
 	if (tmp & (SRBM_STATUS__MCB_BUSY_MASK | SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
@@ -972,13 +970,13 @@ static bool gmc_v6_0_is_idle(void *handle)
 	return true;
 }
 
-static int gmc_v6_0_wait_for_idle(void *handle)
+static int gmc_v6_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
-	unsigned i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	unsigned int i;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		if (gmc_v6_0_is_idle(handle))
+		if (gmc_v6_0_is_idle(ip_block))
 			return 0;
 		udelay(1);
 	}
@@ -986,9 +984,10 @@ static int gmc_v6_0_wait_for_idle(void *handle)
 
 }
 
-static int gmc_v6_0_soft_reset(void *handle)
+static int gmc_v6_0_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+
 	u32 srbm_soft_reset = 0;
 	u32 tmp = RREG32(mmSRBM_STATUS);
 
@@ -1005,10 +1004,9 @@ static int gmc_v6_0_soft_reset(void *handle)
 
 	if (srbm_soft_reset) {
 		gmc_v6_0_mc_stop(adev);
-		if (gmc_v6_0_wait_for_idle(adev)) {
-			dev_warn(adev->dev, "Wait for GMC idle timed out !\n");
-		}
 
+		if (gmc_v6_0_wait_for_idle(ip_block))
+			dev_warn(adev->dev, "Wait for GMC idle timed out !\n");
 
 		tmp = RREG32(mmSRBM_SOFT_RESET);
 		tmp |= srbm_soft_reset;
@@ -1033,7 +1031,7 @@ static int gmc_v6_0_soft_reset(void *handle)
 
 static int gmc_v6_0_vm_fault_interrupt_state(struct amdgpu_device *adev,
 					     struct amdgpu_irq_src *src,
-					     unsigned type,
+					     unsigned int type,
 					     enum amdgpu_interrupt_state state)
 {
 	u32 tmp;
@@ -1097,13 +1095,27 @@ static int gmc_v6_0_process_interrupt(struct amdgpu_device *adev,
 	return 0;
 }
 
-static int gmc_v6_0_set_clockgating_state(void *handle,
+static int gmc_v6_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
+	struct amdgpu_device *adev = ip_block->adev;
+	bool gate = false;
+
+	if (state == AMD_CG_STATE_GATE)
+		gate = true;
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		gmc_v6_0_enable_mc_mgcg(adev, gate);
+		gmc_v6_0_enable_mc_ls(adev, gate);
+	}
+	gmc_v6_0_enable_bif_mgls(adev, gate);
+	gmc_v6_0_enable_hdp_mgcg(adev, gate);
+	gmc_v6_0_enable_hdp_ls(adev, gate);
+
 	return 0;
 }
 
-static int gmc_v6_0_set_powergating_state(void *handle,
+static int gmc_v6_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	return 0;
@@ -1126,12 +1138,13 @@ static const struct amd_ip_funcs gmc_v6_0_ip_funcs = {
 	.set_powergating_state = gmc_v6_0_set_powergating_state,
 };
 
-static const struct amdgpu_gart_funcs gmc_v6_0_gart_funcs = {
-	.flush_gpu_tlb = gmc_v6_0_gart_flush_gpu_tlb,
-	.set_pte_pde = gmc_v6_0_gart_set_pte_pde,
+static const struct amdgpu_gmc_funcs gmc_v6_0_gmc_funcs = {
+	.flush_gpu_tlb = gmc_v6_0_flush_gpu_tlb,
+	.emit_flush_gpu_tlb = gmc_v6_0_emit_flush_gpu_tlb,
 	.set_prt = gmc_v6_0_set_prt,
 	.get_vm_pde = gmc_v6_0_get_vm_pde,
-	.get_vm_pte_flags = gmc_v6_0_get_vm_pte_flags
+	.get_vm_pte = gmc_v6_0_get_vm_pte,
+	.get_vbios_fb_size = gmc_v6_0_get_vbios_fb_size,
 };
 
 static const struct amdgpu_irq_src_funcs gmc_v6_0_irq_funcs = {
@@ -1139,20 +1152,18 @@ static const struct amdgpu_irq_src_funcs gmc_v6_0_irq_funcs = {
 	.process = gmc_v6_0_process_interrupt,
 };
 
-static void gmc_v6_0_set_gart_funcs(struct amdgpu_device *adev)
+static void gmc_v6_0_set_gmc_funcs(struct amdgpu_device *adev)
 {
-	if (adev->gart.gart_funcs == NULL)
-		adev->gart.gart_funcs = &gmc_v6_0_gart_funcs;
+	adev->gmc.gmc_funcs = &gmc_v6_0_gmc_funcs;
 }
 
 static void gmc_v6_0_set_irq_funcs(struct amdgpu_device *adev)
 {
-	adev->mc.vm_fault.num_types = 1;
-	adev->mc.vm_fault.funcs = &gmc_v6_0_irq_funcs;
+	adev->gmc.vm_fault.num_types = 1;
+	adev->gmc.vm_fault.funcs = &gmc_v6_0_irq_funcs;
 }
 
-const struct amdgpu_ip_block_version gmc_v6_0_ip_block =
-{
+const struct amdgpu_ip_block_version gmc_v6_0_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_GMC,
 	.major = 6,
 	.minor = 0,
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
index eace9e7182c8..93d7ccb7d013 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
@@ -20,13 +20,19 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  *
  */
+
 #include <linux/firmware.h>
-#include <drm/drmP.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+
+#include <drm/drm_cache.h>
 #include "amdgpu.h"
 #include "cikd.h"
 #include "cik.h"
 #include "gmc_v7_0.h"
 #include "amdgpu_ucode.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_gem.h"
 
 #include "bif/bif_4_1_d.h"
 #include "bif/bif_4_1_sh_mask.h"
@@ -42,24 +48,24 @@
 
 #include "amdgpu_atombios.h"
 
-static void gmc_v7_0_set_gart_funcs(struct amdgpu_device *adev);
+#include "ivsrcid/ivsrcid_vislands30.h"
+
+static void gmc_v7_0_set_gmc_funcs(struct amdgpu_device *adev);
 static void gmc_v7_0_set_irq_funcs(struct amdgpu_device *adev);
-static int gmc_v7_0_wait_for_idle(void *handle);
+static int gmc_v7_0_wait_for_idle(struct amdgpu_ip_block *ip_block);
 
-MODULE_FIRMWARE("radeon/bonaire_mc.bin");
-MODULE_FIRMWARE("radeon/hawaii_mc.bin");
+MODULE_FIRMWARE("amdgpu/bonaire_mc.bin");
+MODULE_FIRMWARE("amdgpu/hawaii_mc.bin");
 MODULE_FIRMWARE("amdgpu/topaz_mc.bin");
 
-static const u32 golden_settings_iceland_a11[] =
-{
+static const u32 golden_settings_iceland_a11[] = {
 	mmVM_PRT_APERTURE0_LOW_ADDR, 0x0fffffff, 0x0fffffff,
 	mmVM_PRT_APERTURE1_LOW_ADDR, 0x0fffffff, 0x0fffffff,
 	mmVM_PRT_APERTURE2_LOW_ADDR, 0x0fffffff, 0x0fffffff,
 	mmVM_PRT_APERTURE3_LOW_ADDR, 0x0fffffff, 0x0fffffff
 };
 
-static const u32 iceland_mgcg_cgcg_init[] =
-{
+static const u32 iceland_mgcg_cgcg_init[] = {
 	mmMC_MEM_POWER_LS, 0xffffffff, 0x00000104
 };
 
@@ -67,12 +73,12 @@ static void gmc_v7_0_init_golden_registers(struct amdgpu_device *adev)
 {
 	switch (adev->asic_type) {
 	case CHIP_TOPAZ:
-		amdgpu_program_register_sequence(adev,
-						 iceland_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(iceland_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_iceland_a11,
-						 (const u32)ARRAY_SIZE(golden_settings_iceland_a11));
+		amdgpu_device_program_register_sequence(adev,
+							iceland_mgcg_cgcg_init,
+							ARRAY_SIZE(iceland_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_iceland_a11,
+							ARRAY_SIZE(golden_settings_iceland_a11));
 		break;
 	default:
 		break;
@@ -81,9 +87,14 @@ static void gmc_v7_0_init_golden_registers(struct amdgpu_device *adev)
 
 static void gmc_v7_0_mc_stop(struct amdgpu_device *adev)
 {
+	struct amdgpu_ip_block *ip_block;
 	u32 blackout;
 
-	gmc_v7_0_wait_for_idle((void *)adev);
+	ip_block = amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_GMC);
+	if (!ip_block)
+		return;
+
+	gmc_v7_0_wait_for_idle(ip_block);
 
 	blackout = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
 	if (REG_GET_FIELD(blackout, MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE) != 1) {
@@ -124,7 +135,6 @@ static void gmc_v7_0_mc_resume(struct amdgpu_device *adev)
 static int gmc_v7_0_init_microcode(struct amdgpu_device *adev)
 {
 	const char *chip_name;
-	char fw_name[30];
 	int err;
 
 	DRM_DEBUG("\n");
@@ -143,24 +153,15 @@ static int gmc_v7_0_init_microcode(struct amdgpu_device *adev)
 	case CHIP_KABINI:
 	case CHIP_MULLINS:
 		return 0;
-	default: BUG();
+	default:
+		return -EINVAL;
 	}
 
-	if (adev->asic_type == CHIP_TOPAZ)
-		snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mc.bin", chip_name);
-	else
-		snprintf(fw_name, sizeof(fw_name), "radeon/%s_mc.bin", chip_name);
-
-	err = request_firmware(&adev->mc.fw, fw_name, adev->dev);
-	if (err)
-		goto out;
-	err = amdgpu_ucode_validate(adev->mc.fw);
-
-out:
+	err = amdgpu_ucode_request(adev, &adev->gmc.fw, AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_mc.bin", chip_name);
 	if (err) {
-		pr_err("cik_mc: Failed to load firmware \"%s\"\n", fw_name);
-		release_firmware(adev->mc.fw);
-		adev->mc.fw = NULL;
+		pr_err("cik_mc: Failed to load firmware \"%s_mc.bin\"\n", chip_name);
+		amdgpu_ucode_release(&adev->gmc.fw);
 	}
 	return err;
 }
@@ -181,19 +182,19 @@ static int gmc_v7_0_mc_load_microcode(struct amdgpu_device *adev)
 	u32 running;
 	int i, ucode_size, regs_size;
 
-	if (!adev->mc.fw)
+	if (!adev->gmc.fw)
 		return -EINVAL;
 
-	hdr = (const struct mc_firmware_header_v1_0 *)adev->mc.fw->data;
+	hdr = (const struct mc_firmware_header_v1_0 *)adev->gmc.fw->data;
 	amdgpu_ucode_print_mc_hdr(&hdr->header);
 
-	adev->mc.fw_version = le32_to_cpu(hdr->header.ucode_version);
+	adev->gmc.fw_version = le32_to_cpu(hdr->header.ucode_version);
 	regs_size = le32_to_cpu(hdr->io_debug_size_bytes) / (4 * 2);
 	io_mc_regs = (const __le32 *)
-		(adev->mc.fw->data + le32_to_cpu(hdr->io_debug_array_offset_bytes));
+		(adev->gmc.fw->data + le32_to_cpu(hdr->io_debug_array_offset_bytes));
 	ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
 	fw_data = (const __le32 *)
-		(adev->mc.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+		(adev->gmc.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
 
 	running = REG_GET_FIELD(RREG32(mmMC_SEQ_SUP_CNTL), MC_SEQ_SUP_CNTL, RUN);
 
@@ -235,19 +236,15 @@ static int gmc_v7_0_mc_load_microcode(struct amdgpu_device *adev)
 }
 
 static void gmc_v7_0_vram_gtt_location(struct amdgpu_device *adev,
-				       struct amdgpu_mc *mc)
+				       struct amdgpu_gmc *mc)
 {
 	u64 base = RREG32(mmMC_VM_FB_LOCATION) & 0xFFFF;
+
 	base <<= 24;
 
-	if (mc->mc_vram_size > 0xFFC0000000ULL) {
-		/* leave room for at least 1024M GTT */
-		dev_warn(adev->dev, "limiting VRAM\n");
-		mc->real_vram_size = 0xFFC0000000ULL;
-		mc->mc_vram_size = 0xFFC0000000ULL;
-	}
-	amdgpu_vram_location(adev, &adev->mc, base);
-	amdgpu_gart_location(adev, mc);
+	amdgpu_gmc_set_agp_default(adev, mc);
+	amdgpu_gmc_vram_location(adev, mc, base);
+	amdgpu_gmc_gart_location(adev, mc, AMDGPU_GART_PLACEMENT_BEST_FIT);
 }
 
 /**
@@ -260,9 +257,14 @@ static void gmc_v7_0_vram_gtt_location(struct amdgpu_device *adev,
  */
 static void gmc_v7_0_mc_program(struct amdgpu_device *adev)
 {
+	struct amdgpu_ip_block *ip_block;
 	u32 tmp;
 	int i, j;
 
+	ip_block = amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_GMC);
+	if (!ip_block)
+		return;
+
 	/* Initialize HDP */
 	for (i = 0, j = 0; i < 32; i++, j += 0x6) {
 		WREG32((0xb05 + j), 0x00000000);
@@ -273,9 +275,9 @@ static void gmc_v7_0_mc_program(struct amdgpu_device *adev)
 	}
 	WREG32(mmHDP_REG_COHERENCY_FLUSH_CNTL, 0);
 
-	if (gmc_v7_0_wait_for_idle((void *)adev)) {
+	if (gmc_v7_0_wait_for_idle(ip_block))
 		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
-	}
+
 	if (adev->mode_info.num_crtc) {
 		/* Lockout access through VGA aperture*/
 		tmp = RREG32(mmVGA_HDP_CONTROL);
@@ -289,17 +291,16 @@ static void gmc_v7_0_mc_program(struct amdgpu_device *adev)
 	}
 	/* Update configuration */
 	WREG32(mmMC_VM_SYSTEM_APERTURE_LOW_ADDR,
-	       adev->mc.vram_start >> 12);
+	       adev->gmc.vram_start >> 12);
 	WREG32(mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
-	       adev->mc.vram_end >> 12);
+	       adev->gmc.vram_end >> 12);
 	WREG32(mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR,
-	       adev->vram_scratch.gpu_addr >> 12);
+	       adev->mem_scratch.gpu_addr >> 12);
 	WREG32(mmMC_VM_AGP_BASE, 0);
-	WREG32(mmMC_VM_AGP_TOP, 0x0FFFFFFF);
-	WREG32(mmMC_VM_AGP_BOT, 0x0FFFFFFF);
-	if (gmc_v7_0_wait_for_idle((void *)adev)) {
+	WREG32(mmMC_VM_AGP_TOP, adev->gmc.agp_end >> 22);
+	WREG32(mmMC_VM_AGP_BOT, adev->gmc.agp_start >> 22);
+	if (gmc_v7_0_wait_for_idle(ip_block))
 		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
-	}
 
 	WREG32(mmBIF_FB_EN, BIF_FB_EN__FB_READ_EN_MASK | BIF_FB_EN__FB_WRITE_EN_MASK);
 
@@ -322,18 +323,20 @@ static void gmc_v7_0_mc_program(struct amdgpu_device *adev)
  */
 static int gmc_v7_0_mc_init(struct amdgpu_device *adev)
 {
-	adev->mc.vram_width = amdgpu_atombios_get_vram_width(adev);
-	if (!adev->mc.vram_width) {
+	int r;
+
+	adev->gmc.vram_width = amdgpu_atombios_get_vram_width(adev);
+	if (!adev->gmc.vram_width) {
 		u32 tmp;
 		int chansize, numchan;
 
 		/* Get VRAM informations */
 		tmp = RREG32(mmMC_ARB_RAMCFG);
-		if (REG_GET_FIELD(tmp, MC_ARB_RAMCFG, CHANSIZE)) {
+		if (REG_GET_FIELD(tmp, MC_ARB_RAMCFG, CHANSIZE))
 			chansize = 64;
-		} else {
+		else
 			chansize = 32;
-		}
+
 		tmp = RREG32(mmMC_SHARED_CHMAP);
 		switch (REG_GET_FIELD(tmp, MC_SHARED_CHMAP, NOOFCHAN)) {
 		case 0:
@@ -365,33 +368,37 @@ static int gmc_v7_0_mc_init(struct amdgpu_device *adev)
 			numchan = 16;
 			break;
 		}
-		adev->mc.vram_width = numchan * chansize;
+		adev->gmc.vram_width = numchan * chansize;
 	}
-	/* Could aper size report 0 ? */
-	adev->mc.aper_base = pci_resource_start(adev->pdev, 0);
-	adev->mc.aper_size = pci_resource_len(adev->pdev, 0);
 	/* size in MB on si */
-	adev->mc.mc_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
-	adev->mc.real_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
+	adev->gmc.mc_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
+	adev->gmc.real_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		r = amdgpu_device_resize_fb_bar(adev);
+		if (r)
+			return r;
+	}
+	adev->gmc.aper_base = pci_resource_start(adev->pdev, 0);
+	adev->gmc.aper_size = pci_resource_len(adev->pdev, 0);
 
 #ifdef CONFIG_X86_64
-	if (adev->flags & AMD_IS_APU) {
-		adev->mc.aper_base = ((u64)RREG32(mmMC_VM_FB_OFFSET)) << 22;
-		adev->mc.aper_size = adev->mc.real_vram_size;
+	if ((adev->flags & AMD_IS_APU) &&
+	    adev->gmc.real_vram_size > adev->gmc.aper_size &&
+	    !amdgpu_passthrough(adev)) {
+		adev->gmc.aper_base = ((u64)RREG32(mmMC_VM_FB_OFFSET)) << 22;
+		adev->gmc.aper_size = adev->gmc.real_vram_size;
 	}
 #endif
 
-	/* In case the PCI BAR is larger than the actual amount of vram */
-	adev->mc.visible_vram_size = adev->mc.aper_size;
-	if (adev->mc.visible_vram_size > adev->mc.real_vram_size)
-		adev->mc.visible_vram_size = adev->mc.real_vram_size;
+	adev->gmc.visible_vram_size = adev->gmc.aper_size;
 
 	/* set the gart size */
 	if (amdgpu_gart_size == -1) {
 		switch (adev->asic_type) {
 		case CHIP_TOPAZ:     /* no MM engines */
 		default:
-			adev->mc.gart_size = 256ULL << 20;
+			adev->gmc.gart_size = 256ULL << 20;
 			break;
 #ifdef CONFIG_DRM_AMDGPU_CIK
 		case CHIP_BONAIRE: /* UVD, VCE do not support GPUVM */
@@ -399,19 +406,50 @@ static int gmc_v7_0_mc_init(struct amdgpu_device *adev)
 		case CHIP_KAVERI:  /* UVD, VCE do not support GPUVM */
 		case CHIP_KABINI:  /* UVD, VCE do not support GPUVM */
 		case CHIP_MULLINS: /* UVD, VCE do not support GPUVM */
-			adev->mc.gart_size = 1024ULL << 20;
+			adev->gmc.gart_size = 1024ULL << 20;
 			break;
 #endif
 		}
 	} else {
-		adev->mc.gart_size = (u64)amdgpu_gart_size << 20;
+		adev->gmc.gart_size = (u64)amdgpu_gart_size << 20;
 	}
 
-	gmc_v7_0_vram_gtt_location(adev, &adev->mc);
+	adev->gmc.gart_size += adev->pm.smu_prv_buffer_size;
+	gmc_v7_0_vram_gtt_location(adev, &adev->gmc);
 
 	return 0;
 }
 
+/**
+ * gmc_v7_0_flush_gpu_tlb_pasid - tlb flush via pasid
+ *
+ * @adev: amdgpu_device pointer
+ * @pasid: pasid to be flush
+ * @flush_type: type of flush
+ * @all_hub: flush all hubs
+ * @inst: is used to select which instance of KIQ to use for the invalidation
+ *
+ * Flush the TLB for the requested pasid.
+ */
+static void gmc_v7_0_flush_gpu_tlb_pasid(struct amdgpu_device *adev,
+					 uint16_t pasid, uint32_t flush_type,
+					 bool all_hub, uint32_t inst)
+{
+	u32 mask = 0x0;
+	int vmid;
+
+	for (vmid = 1; vmid < 16; vmid++) {
+		u32 tmp = RREG32(mmATC_VMID0_PASID_MAPPING + vmid);
+
+		if ((tmp & ATC_VMID0_PASID_MAPPING__VALID_MASK) &&
+		    (tmp & ATC_VMID0_PASID_MAPPING__PASID_MASK) == pasid)
+			mask |= 1 << vmid;
+	}
+
+	WREG32(mmVM_INVALIDATE_REQUEST, mask);
+	RREG32(mmVM_INVALIDATE_RESPONSE);
+}
+
 /*
  * GART
  * VMID 0 is the physical GPU addresses as used by the kernel.
@@ -420,73 +458,63 @@ static int gmc_v7_0_mc_init(struct amdgpu_device *adev)
  */
 
 /**
- * gmc_v7_0_gart_flush_gpu_tlb - gart tlb flush callback
+ * gmc_v7_0_flush_gpu_tlb - gart tlb flush callback
  *
  * @adev: amdgpu_device pointer
  * @vmid: vm instance to flush
- *
+ * @vmhub: which hub to flush
+ * @flush_type: type of flush
+ * *
  * Flush the TLB for the requested page table (CIK).
  */
-static void gmc_v7_0_gart_flush_gpu_tlb(struct amdgpu_device *adev,
-					uint32_t vmid)
+static void gmc_v7_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid,
+					uint32_t vmhub, uint32_t flush_type)
 {
-	/* flush hdp cache */
-	WREG32(mmHDP_MEM_COHERENCY_FLUSH_CNTL, 0);
-
 	/* bits 0-15 are the VM contexts0-15 */
 	WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
 }
 
-/**
- * gmc_v7_0_gart_set_pte_pde - update the page tables using MMIO
- *
- * @adev: amdgpu_device pointer
- * @cpu_pt_addr: cpu address of the page table
- * @gpu_page_idx: entry in the page table to update
- * @addr: dst addr to write into pte/pde
- * @flags: access flags
- *
- * Update the page tables using the CPU.
- */
-static int gmc_v7_0_gart_set_pte_pde(struct amdgpu_device *adev,
-				     void *cpu_pt_addr,
-				     uint32_t gpu_page_idx,
-				     uint64_t addr,
-				     uint64_t flags)
+static uint64_t gmc_v7_0_emit_flush_gpu_tlb(struct amdgpu_ring *ring,
+					    unsigned int vmid, uint64_t pd_addr)
 {
-	void __iomem *ptr = (void *)cpu_pt_addr;
-	uint64_t value;
+	uint32_t reg;
 
-	value = addr & 0xFFFFFFFFFFFFF000ULL;
-	value |= flags;
-	writeq(value, ptr + (gpu_page_idx * 8));
+	if (vmid < 8)
+		reg = mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + vmid;
+	else
+		reg = mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vmid - 8;
+	amdgpu_ring_emit_wreg(ring, reg, pd_addr >> 12);
 
-	return 0;
+	/* bits 0-15 are the VM contexts0-15 */
+	amdgpu_ring_emit_wreg(ring, mmVM_INVALIDATE_REQUEST, 1 << vmid);
+
+	return pd_addr;
 }
 
-static uint64_t gmc_v7_0_get_vm_pte_flags(struct amdgpu_device *adev,
-					  uint32_t flags)
+static void gmc_v7_0_emit_pasid_mapping(struct amdgpu_ring *ring, unsigned int vmid,
+					unsigned int pasid)
 {
-	uint64_t pte_flag = 0;
-
-	if (flags & AMDGPU_VM_PAGE_READABLE)
-		pte_flag |= AMDGPU_PTE_READABLE;
-	if (flags & AMDGPU_VM_PAGE_WRITEABLE)
-		pte_flag |= AMDGPU_PTE_WRITEABLE;
-	if (flags & AMDGPU_VM_PAGE_PRT)
-		pte_flag |= AMDGPU_PTE_PRT;
+	amdgpu_ring_emit_wreg(ring, mmIH_VMID_0_LUT + vmid, pasid);
+}
 
-	return pte_flag;
+static void gmc_v7_0_get_vm_pde(struct amdgpu_device *adev, int level,
+				uint64_t *addr, uint64_t *flags)
+{
+	BUG_ON(*addr & 0xFFFFFF0000000FFFULL);
 }
 
-static uint64_t gmc_v7_0_get_vm_pde(struct amdgpu_device *adev, uint64_t addr)
+static void gmc_v7_0_get_vm_pte(struct amdgpu_device *adev,
+				struct amdgpu_vm *vm,
+				struct amdgpu_bo *bo,
+				uint32_t vm_flags,
+				uint64_t *flags)
 {
-	BUG_ON(addr & 0xFFFFFF0000000FFFULL);
-	return addr;
+	*flags &= ~AMDGPU_PTE_EXECUTABLE;
+	*flags &= ~AMDGPU_PTE_PRT;
 }
 
 /**
- * gmc_v8_0_set_fault_enable_default - update VM fault handling
+ * gmc_v7_0_set_fault_enable_default - update VM fault handling
  *
  * @adev: amdgpu_device pointer
  * @value: true redirects VM faults to the default page
@@ -522,9 +550,9 @@ static void gmc_v7_0_set_prt(struct amdgpu_device *adev, bool enable)
 {
 	uint32_t tmp;
 
-	if (enable && !adev->mc.prt_warning) {
+	if (enable && !adev->gmc.prt_warning) {
 		dev_warn(adev->dev, "Disabling VM faults because of PRT request!\n");
-		adev->mc.prt_warning = true;
+		adev->gmc.prt_warning = true;
 	}
 
 	tmp = RREG32(mmVM_PRT_CNTL);
@@ -545,8 +573,10 @@ static void gmc_v7_0_set_prt(struct amdgpu_device *adev, bool enable)
 	WREG32(mmVM_PRT_CNTL, tmp);
 
 	if (enable) {
-		uint32_t low = AMDGPU_VA_RESERVED_SIZE >> AMDGPU_GPU_PAGE_SHIFT;
-		uint32_t high = adev->vm_manager.max_pfn;
+		uint32_t low = AMDGPU_VA_RESERVED_BOTTOM >>
+			AMDGPU_GPU_PAGE_SHIFT;
+		uint32_t high = adev->vm_manager.max_pfn -
+			(AMDGPU_VA_RESERVED_TOP >> AMDGPU_GPU_PAGE_SHIFT);
 
 		WREG32(mmVM_PRT_APERTURE0_LOW_ADDR, low);
 		WREG32(mmVM_PRT_APERTURE1_LOW_ADDR, low);
@@ -581,16 +611,17 @@ static void gmc_v7_0_set_prt(struct amdgpu_device *adev, bool enable)
  */
 static int gmc_v7_0_gart_enable(struct amdgpu_device *adev)
 {
-	int r, i;
+	uint64_t table_addr;
 	u32 tmp, field;
+	int i;
 
-	if (adev->gart.robj == NULL) {
+	if (adev->gart.bo == NULL) {
 		dev_err(adev->dev, "No VRAM object for PCIE GART.\n");
 		return -EINVAL;
 	}
-	r = amdgpu_gart_table_vram_pin(adev);
-	if (r)
-		return r;
+	amdgpu_gtt_mgr_recover(&adev->mman.gtt_mgr);
+	table_addr = amdgpu_bo_gpu_offset(adev->gart.bo);
+
 	/* Setup TLB control */
 	tmp = RREG32(mmMC_VM_MX_L1_TLB_CNTL);
 	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
@@ -620,11 +651,11 @@ static int gmc_v7_0_gart_enable(struct amdgpu_device *adev)
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_FRAGMENT_SIZE, field);
 	WREG32(mmVM_L2_CNTL3, tmp);
 	/* setup context0 */
-	WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->mc.gart_start >> 12);
-	WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, adev->mc.gart_end >> 12);
-	WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR, adev->gart.table_addr >> 12);
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->gmc.gart_start >> 12);
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, adev->gmc.gart_end >> 12);
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR, table_addr >> 12);
 	WREG32(mmVM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR,
-			(u32)(adev->dummy_page.addr >> 12));
+			(u32)(adev->dummy_page_addr >> 12));
 	WREG32(mmVM_CONTEXT0_CNTL2, 0);
 	tmp = RREG32(mmVM_CONTEXT0_CNTL);
 	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
@@ -643,18 +674,18 @@ static int gmc_v7_0_gart_enable(struct amdgpu_device *adev)
 	/* set vm size, must be a multiple of 4 */
 	WREG32(mmVM_CONTEXT1_PAGE_TABLE_START_ADDR, 0);
 	WREG32(mmVM_CONTEXT1_PAGE_TABLE_END_ADDR, adev->vm_manager.max_pfn - 1);
-	for (i = 1; i < 16; i++) {
+	for (i = 1; i < AMDGPU_NUM_VMID; i++) {
 		if (i < 8)
 			WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + i,
-			       adev->gart.table_addr >> 12);
+			       table_addr >> 12);
 		else
 			WREG32(mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + i - 8,
-			       adev->gart.table_addr >> 12);
+			       table_addr >> 12);
 	}
 
 	/* enable context1-15 */
 	WREG32(mmVM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR,
-	       (u32)(adev->dummy_page.addr >> 12));
+	       (u32)(adev->dummy_page_addr >> 12));
 	WREG32(mmVM_CONTEXT1_CNTL2, 4);
 	tmp = RREG32(mmVM_CONTEXT1_CNTL);
 	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
@@ -673,11 +704,10 @@ static int gmc_v7_0_gart_enable(struct amdgpu_device *adev)
 		WREG32(mmCHUB_CONTROL, tmp);
 	}
 
-	gmc_v7_0_gart_flush_gpu_tlb(adev, 0);
+	gmc_v7_0_flush_gpu_tlb(adev, 0, 0, 0);
 	DRM_INFO("PCIE GART of %uM enabled (table at 0x%016llX).\n",
-		 (unsigned)(adev->mc.gart_size >> 20),
-		 (unsigned long long)adev->gart.table_addr);
-	adev->gart.ready = true;
+		 (unsigned int)(adev->gmc.gart_size >> 20),
+		 (unsigned long long)table_addr);
 	return 0;
 }
 
@@ -685,7 +715,7 @@ static int gmc_v7_0_gart_init(struct amdgpu_device *adev)
 {
 	int r;
 
-	if (adev->gart.robj) {
+	if (adev->gart.bo) {
 		WARN(1, "R600 PCIE GART already initialized\n");
 		return 0;
 	}
@@ -723,20 +753,6 @@ static void gmc_v7_0_gart_disable(struct amdgpu_device *adev)
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 0);
 	WREG32(mmVM_L2_CNTL, tmp);
 	WREG32(mmVM_L2_CNTL2, 0);
-	amdgpu_gart_table_vram_unpin(adev);
-}
-
-/**
- * gmc_v7_0_gart_fini - vm fini callback
- *
- * @adev: amdgpu_device pointer
- *
- * Tears down the driver GART/VM setup (CIK).
- */
-static void gmc_v7_0_gart_fini(struct amdgpu_device *adev)
-{
-	amdgpu_gart_table_vram_free(adev);
-	amdgpu_gart_fini(adev);
 }
 
 /**
@@ -745,24 +761,26 @@ static void gmc_v7_0_gart_fini(struct amdgpu_device *adev)
  * @adev: amdgpu_device pointer
  * @status: VM_CONTEXT1_PROTECTION_FAULT_STATUS register value
  * @addr: VM_CONTEXT1_PROTECTION_FAULT_ADDR register value
+ * @mc_client: VM_CONTEXT1_PROTECTION_FAULT_MCCLIENT register value
+ * @pasid: debug logging only - no functional use
  *
  * Print human readable fault information (CIK).
  */
-static void gmc_v7_0_vm_decode_fault(struct amdgpu_device *adev,
-				     u32 status, u32 addr, u32 mc_client)
+static void gmc_v7_0_vm_decode_fault(struct amdgpu_device *adev, u32 status,
+				     u32 addr, u32 mc_client, unsigned int pasid)
 {
-	u32 mc_id;
 	u32 vmid = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS, VMID);
 	u32 protections = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
 					PROTECTIONS);
 	char block[5] = { mc_client >> 24, (mc_client >> 16) & 0xff,
 		(mc_client >> 8) & 0xff, mc_client & 0xff, 0 };
+	u32 mc_id;
 
 	mc_id = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
 			      MEMORY_CLIENT_ID);
 
-	dev_err(adev->dev, "VM fault (0x%02x, vmid %d) at page %u, %s from '%s' (0x%08x) (%d)\n",
-	       protections, vmid, addr,
+	dev_err(adev->dev, "VM fault (0x%02x, vmid %d, pasid %d) at page %u, %s from '%s' (0x%08x) (%d)\n",
+	       protections, vmid, pasid, addr,
 	       REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
 			     MEMORY_CLIENT_RW) ?
 	       "write" : "read", block, mc_client, mc_id);
@@ -916,53 +934,74 @@ static int gmc_v7_0_convert_vram_type(int mc_seq_vram_type)
 	}
 }
 
-static int gmc_v7_0_early_init(void *handle)
+static int gmc_v7_0_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	gmc_v7_0_set_gart_funcs(adev);
+	gmc_v7_0_set_gmc_funcs(adev);
 	gmc_v7_0_set_irq_funcs(adev);
 
-	adev->mc.shared_aperture_start = 0x2000000000000000ULL;
-	adev->mc.shared_aperture_end =
-		adev->mc.shared_aperture_start + (4ULL << 30) - 1;
-	adev->mc.private_aperture_start =
-		adev->mc.shared_aperture_end + 1;
-	adev->mc.private_aperture_end =
-		adev->mc.private_aperture_start + (4ULL << 30) - 1;
+	adev->gmc.shared_aperture_start = 0x2000000000000000ULL;
+	adev->gmc.shared_aperture_end =
+		adev->gmc.shared_aperture_start + (4ULL << 30) - 1;
+	adev->gmc.private_aperture_start =
+		adev->gmc.shared_aperture_end + 1;
+	adev->gmc.private_aperture_end =
+		adev->gmc.private_aperture_start + (4ULL << 30) - 1;
+	adev->gmc.noretry_flags = AMDGPU_VM_NORETRY_FLAGS_TF;
 
 	return 0;
 }
 
-static int gmc_v7_0_late_init(void *handle)
+static int gmc_v7_0_late_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (amdgpu_vm_fault_stop != AMDGPU_VM_FAULT_STOP_ALWAYS)
-		return amdgpu_irq_get(adev, &adev->mc.vm_fault, 0);
+		return amdgpu_irq_get(adev, &adev->gmc.vm_fault, 0);
 	else
 		return 0;
 }
 
-static int gmc_v7_0_sw_init(void *handle)
+static unsigned int gmc_v7_0_get_vbios_fb_size(struct amdgpu_device *adev)
+{
+	u32 d1vga_control = RREG32(mmD1VGA_CONTROL);
+	unsigned int size;
+
+	if (REG_GET_FIELD(d1vga_control, D1VGA_CONTROL, D1VGA_MODE_ENABLE)) {
+		size = AMDGPU_VBIOS_VGA_ALLOCATION;
+	} else {
+		u32 viewport = RREG32(mmVIEWPORT_SIZE);
+
+		size = (REG_GET_FIELD(viewport, VIEWPORT_SIZE, VIEWPORT_HEIGHT) *
+			REG_GET_FIELD(viewport, VIEWPORT_SIZE, VIEWPORT_WIDTH) *
+			4);
+	}
+
+	return size;
+}
+
+static int gmc_v7_0_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	int dma_bits;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	set_bit(AMDGPU_GFXHUB(0), adev->vmhubs_mask);
 
 	if (adev->flags & AMD_IS_APU) {
-		adev->mc.vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
+		adev->gmc.vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
 	} else {
 		u32 tmp = RREG32(mmMC_SEQ_MISC0);
+
 		tmp &= MC_SEQ_MISC0__MT__MASK;
-		adev->mc.vram_type = gmc_v7_0_convert_vram_type(tmp);
+		adev->gmc.vram_type = gmc_v7_0_convert_vram_type(tmp);
 	}
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 146, &adev->mc.vm_fault);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_GFX_PAGE_INV_FAULT, &adev->gmc.vm_fault);
 	if (r)
 		return r;
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 147, &adev->mc.vm_fault);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_GFX_MEM_PROT_FAULT, &adev->gmc.vm_fault);
 	if (r)
 		return r;
 
@@ -970,35 +1009,20 @@ static int gmc_v7_0_sw_init(void *handle)
 	 * Currently set to 4GB ((1 << 20) 4k pages).
 	 * Max GPUVM size for cayman and SI is 40 bits.
 	 */
-	amdgpu_vm_adjust_size(adev, 64, 4);
-	adev->vm_manager.max_pfn = adev->vm_manager.vm_size << 18;
+	amdgpu_vm_adjust_size(adev, 64, 9, 1, 40);
 
 	/* Set the internal MC address mask
 	 * This is the max address of the GPU's
 	 * internal address space.
 	 */
-	adev->mc.mc_mask = 0xffffffffffULL; /* 40 bit MC */
+	adev->gmc.mc_mask = 0xffffffffffULL; /* 40 bit MC */
 
-	adev->mc.stolen_size = 256 * 1024;
-
-	/* set DMA mask + need_dma32 flags.
-	 * PCIE - can handle 40-bits.
-	 * IGP - can handle 40-bits
-	 * PCI - dma32 for legacy pci gart, 40 bits on newer asics
-	 */
-	adev->need_dma32 = false;
-	dma_bits = adev->need_dma32 ? 32 : 40;
-	r = pci_set_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
+	r = dma_set_mask_and_coherent(adev->dev, DMA_BIT_MASK(40));
 	if (r) {
-		adev->need_dma32 = true;
-		dma_bits = 32;
-		pr_warn("amdgpu: No suitable DMA available\n");
-	}
-	r = pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
-	if (r) {
-		pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(32));
-		pr_warn("amdgpu: No coherent DMA available\n");
+		pr_warn("No suitable DMA available\n");
+		return r;
 	}
+	adev->need_swiotlb = drm_need_swiotlb(40);
 
 	r = gmc_v7_0_init_microcode(adev);
 	if (r) {
@@ -1010,6 +1034,8 @@ static int gmc_v7_0_sw_init(void *handle)
 	if (r)
 		return r;
 
+	amdgpu_gmc_get_vbios_allocations(adev);
+
 	/* Memory manager */
 	r = amdgpu_bo_init(adev);
 	if (r)
@@ -1025,8 +1051,7 @@ static int gmc_v7_0_sw_init(void *handle)
 	 * amdgpu graphics/compute will use VMIDs 1-7
 	 * amdkfd will use VMIDs 8-15
 	 */
-	adev->vm_manager.id_mgr[0].num_ids = AMDGPU_NUM_OF_VMIDS;
-	adev->vm_manager.num_level = 1;
+	adev->vm_manager.first_kfd_vmid = 8;
 	amdgpu_vm_manager_init(adev);
 
 	/* base offset of vram pages */
@@ -1039,25 +1064,33 @@ static int gmc_v7_0_sw_init(void *handle)
 		adev->vm_manager.vram_base_offset = 0;
 	}
 
+	adev->gmc.vm_fault_info = kmalloc(sizeof(struct kfd_vm_fault_info),
+					GFP_KERNEL);
+	if (!adev->gmc.vm_fault_info)
+		return -ENOMEM;
+	atomic_set(&adev->gmc.vm_fault_info_updated, 0);
+
 	return 0;
 }
 
-static int gmc_v7_0_sw_fini(void *handle)
+static int gmc_v7_0_sw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	amdgpu_vm_manager_fini(adev);
-	gmc_v7_0_gart_fini(adev);
 	amdgpu_gem_force_release(adev);
+	amdgpu_vm_manager_fini(adev);
+	kfree(adev->gmc.vm_fault_info);
+	amdgpu_gart_table_vram_free(adev);
 	amdgpu_bo_fini(adev);
+	amdgpu_ucode_release(&adev->gmc.fw);
 
 	return 0;
 }
 
-static int gmc_v7_0_hw_init(void *handle)
+static int gmc_v7_0_hw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	gmc_v7_0_init_golden_registers(adev);
 
@@ -1075,45 +1108,45 @@ static int gmc_v7_0_hw_init(void *handle)
 	if (r)
 		return r;
 
-	return r;
+	if (amdgpu_emu_mode == 1)
+		return amdgpu_gmc_vram_checking(adev);
+
+	return 0;
 }
 
-static int gmc_v7_0_hw_fini(void *handle)
+static int gmc_v7_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	amdgpu_irq_put(adev, &adev->mc.vm_fault, 0);
+	amdgpu_irq_put(adev, &adev->gmc.vm_fault, 0);
 	gmc_v7_0_gart_disable(adev);
 
 	return 0;
 }
 
-static int gmc_v7_0_suspend(void *handle)
+static int gmc_v7_0_suspend(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	gmc_v7_0_hw_fini(adev);
+	gmc_v7_0_hw_fini(ip_block);
 
 	return 0;
 }
 
-static int gmc_v7_0_resume(void *handle)
+static int gmc_v7_0_resume(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	r = gmc_v7_0_hw_init(adev);
+	r = gmc_v7_0_hw_init(ip_block);
 	if (r)
 		return r;
 
-	amdgpu_vm_reset_all_ids(adev);
+	amdgpu_vmid_reset_all(ip_block->adev);
 
 	return 0;
 }
 
-static bool gmc_v7_0_is_idle(void *handle)
+static bool gmc_v7_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 tmp = RREG32(mmSRBM_STATUS);
 
 	if (tmp & (SRBM_STATUS__MCB_BUSY_MASK | SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
@@ -1123,20 +1156,13 @@ static bool gmc_v7_0_is_idle(void *handle)
 	return true;
 }
 
-static int gmc_v7_0_wait_for_idle(void *handle)
+static int gmc_v7_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
-	unsigned i;
-	u32 tmp;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	unsigned int i;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		/* read MC_STATUS */
-		tmp = RREG32(mmSRBM_STATUS) & (SRBM_STATUS__MCB_BUSY_MASK |
-					       SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
-					       SRBM_STATUS__MCC_BUSY_MASK |
-					       SRBM_STATUS__MCD_BUSY_MASK |
-					       SRBM_STATUS__VMC_BUSY_MASK);
-		if (!tmp)
+		if (gmc_v7_0_is_idle(ip_block))
 			return 0;
 		udelay(1);
 	}
@@ -1144,9 +1170,9 @@ static int gmc_v7_0_wait_for_idle(void *handle)
 
 }
 
-static int gmc_v7_0_soft_reset(void *handle)
+static int gmc_v7_0_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 srbm_soft_reset = 0;
 	u32 tmp = RREG32(mmSRBM_STATUS);
 
@@ -1163,10 +1189,8 @@ static int gmc_v7_0_soft_reset(void *handle)
 
 	if (srbm_soft_reset) {
 		gmc_v7_0_mc_stop(adev);
-		if (gmc_v7_0_wait_for_idle((void *)adev)) {
+		if (gmc_v7_0_wait_for_idle(ip_block))
 			dev_warn(adev->dev, "Wait for GMC idle timed out !\n");
-		}
-
 
 		tmp = RREG32(mmSRBM_SOFT_RESET);
 		tmp |= srbm_soft_reset;
@@ -1192,7 +1216,7 @@ static int gmc_v7_0_soft_reset(void *handle)
 
 static int gmc_v7_0_vm_fault_interrupt_state(struct amdgpu_device *adev,
 					     struct amdgpu_irq_src *src,
-					     unsigned type,
+					     unsigned int type,
 					     enum amdgpu_interrupt_state state)
 {
 	u32 tmp;
@@ -1235,7 +1259,7 @@ static int gmc_v7_0_process_interrupt(struct amdgpu_device *adev,
 				      struct amdgpu_irq_src *source,
 				      struct amdgpu_iv_entry *entry)
 {
-	u32 addr, status, mc_client;
+	u32 addr, status, mc_client, vmid;
 
 	addr = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_ADDR);
 	status = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS);
@@ -1246,6 +1270,9 @@ static int gmc_v7_0_process_interrupt(struct amdgpu_device *adev,
 	if (!addr && !status)
 		return 0;
 
+	amdgpu_vm_update_fault_cache(adev, entry->pasid,
+				     ((u64)addr) << AMDGPU_GPU_PAGE_SHIFT, status, AMDGPU_GFXHUB(0));
+
 	if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_FIRST)
 		gmc_v7_0_set_fault_enable_default(adev, false);
 
@@ -1256,17 +1283,41 @@ static int gmc_v7_0_process_interrupt(struct amdgpu_device *adev,
 			addr);
 		dev_err(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x%08X\n",
 			status);
-		gmc_v7_0_vm_decode_fault(adev, status, addr, mc_client);
+		gmc_v7_0_vm_decode_fault(adev, status, addr, mc_client,
+					 entry->pasid);
+	}
+
+	vmid = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+			     VMID);
+	if (amdgpu_amdkfd_is_kfd_vmid(adev, vmid)
+		&& !atomic_read(&adev->gmc.vm_fault_info_updated)) {
+		struct kfd_vm_fault_info *info = adev->gmc.vm_fault_info;
+		u32 protections = REG_GET_FIELD(status,
+					VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+					PROTECTIONS);
+
+		info->vmid = vmid;
+		info->mc_id = REG_GET_FIELD(status,
+					    VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+					    MEMORY_CLIENT_ID);
+		info->status = status;
+		info->page_addr = addr;
+		info->prot_valid = protections & 0x7 ? true : false;
+		info->prot_read = protections & 0x8 ? true : false;
+		info->prot_write = protections & 0x10 ? true : false;
+		info->prot_exec = protections & 0x20 ? true : false;
+		mb();
+		atomic_set(&adev->gmc.vm_fault_info_updated, 1);
 	}
 
 	return 0;
 }
 
-static int gmc_v7_0_set_clockgating_state(void *handle,
+static int gmc_v7_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
 	bool gate = false;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (state == AMD_CG_STATE_GATE)
 		gate = true;
@@ -1282,7 +1333,7 @@ static int gmc_v7_0_set_clockgating_state(void *handle,
 	return 0;
 }
 
-static int gmc_v7_0_set_powergating_state(void *handle,
+static int gmc_v7_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	return 0;
@@ -1305,12 +1356,15 @@ static const struct amd_ip_funcs gmc_v7_0_ip_funcs = {
 	.set_powergating_state = gmc_v7_0_set_powergating_state,
 };
 
-static const struct amdgpu_gart_funcs gmc_v7_0_gart_funcs = {
-	.flush_gpu_tlb = gmc_v7_0_gart_flush_gpu_tlb,
-	.set_pte_pde = gmc_v7_0_gart_set_pte_pde,
+static const struct amdgpu_gmc_funcs gmc_v7_0_gmc_funcs = {
+	.flush_gpu_tlb = gmc_v7_0_flush_gpu_tlb,
+	.flush_gpu_tlb_pasid = gmc_v7_0_flush_gpu_tlb_pasid,
+	.emit_flush_gpu_tlb = gmc_v7_0_emit_flush_gpu_tlb,
+	.emit_pasid_mapping = gmc_v7_0_emit_pasid_mapping,
 	.set_prt = gmc_v7_0_set_prt,
-	.get_vm_pte_flags = gmc_v7_0_get_vm_pte_flags,
-	.get_vm_pde = gmc_v7_0_get_vm_pde
+	.get_vm_pde = gmc_v7_0_get_vm_pde,
+	.get_vm_pte = gmc_v7_0_get_vm_pte,
+	.get_vbios_fb_size = gmc_v7_0_get_vbios_fb_size,
 };
 
 static const struct amdgpu_irq_src_funcs gmc_v7_0_irq_funcs = {
@@ -1318,20 +1372,18 @@ static const struct amdgpu_irq_src_funcs gmc_v7_0_irq_funcs = {
 	.process = gmc_v7_0_process_interrupt,
 };
 
-static void gmc_v7_0_set_gart_funcs(struct amdgpu_device *adev)
+static void gmc_v7_0_set_gmc_funcs(struct amdgpu_device *adev)
 {
-	if (adev->gart.gart_funcs == NULL)
-		adev->gart.gart_funcs = &gmc_v7_0_gart_funcs;
+	adev->gmc.gmc_funcs = &gmc_v7_0_gmc_funcs;
 }
 
 static void gmc_v7_0_set_irq_funcs(struct amdgpu_device *adev)
 {
-	adev->mc.vm_fault.num_types = 1;
-	adev->mc.vm_fault.funcs = &gmc_v7_0_irq_funcs;
+	adev->gmc.vm_fault.num_types = 1;
+	adev->gmc.vm_fault.funcs = &gmc_v7_0_irq_funcs;
 }
 
-const struct amdgpu_ip_block_version gmc_v7_0_ip_block =
-{
+const struct amdgpu_ip_block_version gmc_v7_0_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_GMC,
 	.major = 7,
 	.minor = 0,
@@ -1339,8 +1391,7 @@ const struct amdgpu_ip_block_version gmc_v7_0_ip_block =
 	.funcs = &gmc_v7_0_ip_funcs,
 };
 
-const struct amdgpu_ip_block_version gmc_v7_4_ip_block =
-{
+const struct amdgpu_ip_block_version gmc_v7_4_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_GMC,
 	.major = 7,
 	.minor = 4,
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
index 3b3326daf32b..c5e2a2c41e06 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
@@ -20,11 +20,17 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  *
  */
+
 #include <linux/firmware.h>
-#include <drm/drmP.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+
+#include <drm/drm_cache.h>
 #include "amdgpu.h"
 #include "gmc_v8_0.h"
 #include "amdgpu_ucode.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_gem.h"
 
 #include "gmc/gmc_8_1_d.h"
 #include "gmc/gmc_8_1_sh_mask.h"
@@ -43,18 +49,22 @@
 
 #include "amdgpu_atombios.h"
 
+#include "ivsrcid/ivsrcid_vislands30.h"
 
-static void gmc_v8_0_set_gart_funcs(struct amdgpu_device *adev);
+static void gmc_v8_0_set_gmc_funcs(struct amdgpu_device *adev);
 static void gmc_v8_0_set_irq_funcs(struct amdgpu_device *adev);
-static int gmc_v8_0_wait_for_idle(void *handle);
+static int gmc_v8_0_wait_for_idle(struct amdgpu_ip_block *ip_block);
 
 MODULE_FIRMWARE("amdgpu/tonga_mc.bin");
 MODULE_FIRMWARE("amdgpu/polaris11_mc.bin");
 MODULE_FIRMWARE("amdgpu/polaris10_mc.bin");
 MODULE_FIRMWARE("amdgpu/polaris12_mc.bin");
+MODULE_FIRMWARE("amdgpu/polaris12_32_mc.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_k_mc.bin");
+MODULE_FIRMWARE("amdgpu/polaris10_k_mc.bin");
+MODULE_FIRMWARE("amdgpu/polaris12_k_mc.bin");
 
-static const u32 golden_settings_tonga_a11[] =
-{
+static const u32 golden_settings_tonga_a11[] = {
 	mmMC_ARB_WTM_GRPWT_RD, 0x00000003, 0x00000000,
 	mmMC_HUB_RDREQ_DMIF_LIMIT, 0x0000007f, 0x00000028,
 	mmMC_HUB_WDP_UMC, 0x00007fb6, 0x00000991,
@@ -64,34 +74,29 @@ static const u32 golden_settings_tonga_a11[] =
 	mmVM_PRT_APERTURE3_LOW_ADDR, 0x0fffffff, 0x0fffffff,
 };
 
-static const u32 tonga_mgcg_cgcg_init[] =
-{
+static const u32 tonga_mgcg_cgcg_init[] = {
 	mmMC_MEM_POWER_LS, 0xffffffff, 0x00000104
 };
 
-static const u32 golden_settings_fiji_a10[] =
-{
+static const u32 golden_settings_fiji_a10[] = {
 	mmVM_PRT_APERTURE0_LOW_ADDR, 0x0fffffff, 0x0fffffff,
 	mmVM_PRT_APERTURE1_LOW_ADDR, 0x0fffffff, 0x0fffffff,
 	mmVM_PRT_APERTURE2_LOW_ADDR, 0x0fffffff, 0x0fffffff,
 	mmVM_PRT_APERTURE3_LOW_ADDR, 0x0fffffff, 0x0fffffff,
 };
 
-static const u32 fiji_mgcg_cgcg_init[] =
-{
+static const u32 fiji_mgcg_cgcg_init[] = {
 	mmMC_MEM_POWER_LS, 0xffffffff, 0x00000104
 };
 
-static const u32 golden_settings_polaris11_a11[] =
-{
+static const u32 golden_settings_polaris11_a11[] = {
 	mmVM_PRT_APERTURE0_LOW_ADDR, 0x0fffffff, 0x0fffffff,
 	mmVM_PRT_APERTURE1_LOW_ADDR, 0x0fffffff, 0x0fffffff,
 	mmVM_PRT_APERTURE2_LOW_ADDR, 0x0fffffff, 0x0fffffff,
 	mmVM_PRT_APERTURE3_LOW_ADDR, 0x0fffffff, 0x0fffffff
 };
 
-static const u32 golden_settings_polaris10_a11[] =
-{
+static const u32 golden_settings_polaris10_a11[] = {
 	mmMC_ARB_WTM_GRPWT_RD, 0x00000003, 0x00000000,
 	mmVM_PRT_APERTURE0_LOW_ADDR, 0x0fffffff, 0x0fffffff,
 	mmVM_PRT_APERTURE1_LOW_ADDR, 0x0fffffff, 0x0fffffff,
@@ -99,19 +104,16 @@ static const u32 golden_settings_polaris10_a11[] =
 	mmVM_PRT_APERTURE3_LOW_ADDR, 0x0fffffff, 0x0fffffff
 };
 
-static const u32 cz_mgcg_cgcg_init[] =
-{
+static const u32 cz_mgcg_cgcg_init[] = {
 	mmMC_MEM_POWER_LS, 0xffffffff, 0x00000104
 };
 
-static const u32 stoney_mgcg_cgcg_init[] =
-{
+static const u32 stoney_mgcg_cgcg_init[] = {
 	mmATC_MISC_CG, 0xffffffff, 0x000c0200,
 	mmMC_MEM_POWER_LS, 0xffffffff, 0x00000104
 };
 
-static const u32 golden_settings_stoney_common[] =
-{
+static const u32 golden_settings_stoney_common[] = {
 	mmMC_HUB_RDREQ_UVD, MC_HUB_RDREQ_UVD__PRESCALE_MASK, 0x00000004,
 	mmMC_RD_GRP_OTH, MC_RD_GRP_OTH__UVD_MASK, 0x00600000
 };
@@ -120,44 +122,45 @@ static void gmc_v8_0_init_golden_registers(struct amdgpu_device *adev)
 {
 	switch (adev->asic_type) {
 	case CHIP_FIJI:
-		amdgpu_program_register_sequence(adev,
-						 fiji_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(fiji_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_fiji_a10,
-						 (const u32)ARRAY_SIZE(golden_settings_fiji_a10));
+		amdgpu_device_program_register_sequence(adev,
+							fiji_mgcg_cgcg_init,
+							ARRAY_SIZE(fiji_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_fiji_a10,
+							ARRAY_SIZE(golden_settings_fiji_a10));
 		break;
 	case CHIP_TONGA:
-		amdgpu_program_register_sequence(adev,
-						 tonga_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(tonga_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_tonga_a11,
-						 (const u32)ARRAY_SIZE(golden_settings_tonga_a11));
+		amdgpu_device_program_register_sequence(adev,
+							tonga_mgcg_cgcg_init,
+							ARRAY_SIZE(tonga_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_tonga_a11,
+							ARRAY_SIZE(golden_settings_tonga_a11));
 		break;
 	case CHIP_POLARIS11:
 	case CHIP_POLARIS12:
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_polaris11_a11,
-						 (const u32)ARRAY_SIZE(golden_settings_polaris11_a11));
+	case CHIP_VEGAM:
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_polaris11_a11,
+							ARRAY_SIZE(golden_settings_polaris11_a11));
 		break;
 	case CHIP_POLARIS10:
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_polaris10_a11,
-						 (const u32)ARRAY_SIZE(golden_settings_polaris10_a11));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_polaris10_a11,
+							ARRAY_SIZE(golden_settings_polaris10_a11));
 		break;
 	case CHIP_CARRIZO:
-		amdgpu_program_register_sequence(adev,
-						 cz_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(cz_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							cz_mgcg_cgcg_init,
+							ARRAY_SIZE(cz_mgcg_cgcg_init));
 		break;
 	case CHIP_STONEY:
-		amdgpu_program_register_sequence(adev,
-						 stoney_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(stoney_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_stoney_common,
-						 (const u32)ARRAY_SIZE(golden_settings_stoney_common));
+		amdgpu_device_program_register_sequence(adev,
+							stoney_mgcg_cgcg_init,
+							ARRAY_SIZE(stoney_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_stoney_common,
+							ARRAY_SIZE(golden_settings_stoney_common));
 		break;
 	default:
 		break;
@@ -167,8 +170,13 @@ static void gmc_v8_0_init_golden_registers(struct amdgpu_device *adev)
 static void gmc_v8_0_mc_stop(struct amdgpu_device *adev)
 {
 	u32 blackout;
+	struct amdgpu_ip_block *ip_block;
 
-	gmc_v8_0_wait_for_idle(adev);
+	ip_block = amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_GMC);
+	if (!ip_block)
+		return;
+
+	gmc_v8_0_wait_for_idle(ip_block);
 
 	blackout = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
 	if (REG_GET_FIELD(blackout, MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE) != 1) {
@@ -209,7 +217,6 @@ static void gmc_v8_0_mc_resume(struct amdgpu_device *adev)
 static int gmc_v8_0_init_microcode(struct amdgpu_device *adev)
 {
 	const char *chip_name;
-	char fw_name[30];
 	int err;
 
 	DRM_DEBUG("\n");
@@ -219,32 +226,44 @@ static int gmc_v8_0_init_microcode(struct amdgpu_device *adev)
 		chip_name = "tonga";
 		break;
 	case CHIP_POLARIS11:
-		chip_name = "polaris11";
+		if (ASICID_IS_P21(adev->pdev->device, adev->pdev->revision) ||
+		    ASICID_IS_P31(adev->pdev->device, adev->pdev->revision))
+			chip_name = "polaris11_k";
+		else
+			chip_name = "polaris11";
 		break;
 	case CHIP_POLARIS10:
-		chip_name = "polaris10";
+		if (ASICID_IS_P30(adev->pdev->device, adev->pdev->revision))
+			chip_name = "polaris10_k";
+		else
+			chip_name = "polaris10";
 		break;
 	case CHIP_POLARIS12:
-		chip_name = "polaris12";
+		if (ASICID_IS_P23(adev->pdev->device, adev->pdev->revision)) {
+			chip_name = "polaris12_k";
+		} else {
+			WREG32(mmMC_SEQ_IO_DEBUG_INDEX, ixMC_IO_DEBUG_UP_159);
+			/* Polaris12 32bit ASIC needs a special MC firmware */
+			if (RREG32(mmMC_SEQ_IO_DEBUG_DATA) == 0x05b4dc40)
+				chip_name = "polaris12_32";
+			else
+				chip_name = "polaris12";
+		}
 		break;
 	case CHIP_FIJI:
 	case CHIP_CARRIZO:
 	case CHIP_STONEY:
+	case CHIP_VEGAM:
 		return 0;
-	default: BUG();
+	default:
+		return -EINVAL;
 	}
 
-	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mc.bin", chip_name);
-	err = request_firmware(&adev->mc.fw, fw_name, adev->dev);
-	if (err)
-		goto out;
-	err = amdgpu_ucode_validate(adev->mc.fw);
-
-out:
+	err = amdgpu_ucode_request(adev, &adev->gmc.fw, AMDGPU_UCODE_REQUIRED,
+				   "amdgpu/%s_mc.bin", chip_name);
 	if (err) {
-		pr_err("mc: Failed to load firmware \"%s\"\n", fw_name);
-		release_firmware(adev->mc.fw);
-		adev->mc.fw = NULL;
+		pr_err("mc: Failed to load firmware \"%s_mc.bin\"\n", chip_name);
+		amdgpu_ucode_release(&adev->gmc.fw);
 	}
 	return err;
 }
@@ -254,7 +273,7 @@ out:
  *
  * @adev: amdgpu_device pointer
  *
- * Load the GDDR MC ucode into the hw (CIK).
+ * Load the GDDR MC ucode into the hw (VI).
  * Returns 0 on success, error on failure.
  */
 static int gmc_v8_0_tonga_mc_load_microcode(struct amdgpu_device *adev)
@@ -273,19 +292,19 @@ static int gmc_v8_0_tonga_mc_load_microcode(struct amdgpu_device *adev)
 	if (amdgpu_sriov_bios(adev))
 		return 0;
 
-	if (!adev->mc.fw)
+	if (!adev->gmc.fw)
 		return -EINVAL;
 
-	hdr = (const struct mc_firmware_header_v1_0 *)adev->mc.fw->data;
+	hdr = (const struct mc_firmware_header_v1_0 *)adev->gmc.fw->data;
 	amdgpu_ucode_print_mc_hdr(&hdr->header);
 
-	adev->mc.fw_version = le32_to_cpu(hdr->header.ucode_version);
+	adev->gmc.fw_version = le32_to_cpu(hdr->header.ucode_version);
 	regs_size = le32_to_cpu(hdr->io_debug_size_bytes) / (4 * 2);
 	io_mc_regs = (const __le32 *)
-		(adev->mc.fw->data + le32_to_cpu(hdr->io_debug_array_offset_bytes));
+		(adev->gmc.fw->data + le32_to_cpu(hdr->io_debug_array_offset_bytes));
 	ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
 	fw_data = (const __le32 *)
-		(adev->mc.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+		(adev->gmc.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
 
 	running = REG_GET_FIELD(RREG32(mmMC_SEQ_SUP_CNTL), MC_SEQ_SUP_CNTL, RUN);
 
@@ -331,7 +350,7 @@ static int gmc_v8_0_polaris_mc_load_microcode(struct amdgpu_device *adev)
 	const struct mc_firmware_header_v1_0 *hdr;
 	const __le32 *fw_data = NULL;
 	const __le32 *io_mc_regs = NULL;
-	u32 data, vbios_version;
+	u32 data;
 	int i, ucode_size, regs_size;
 
 	/* Skip MC ucode loading on SR-IOV capable boards.
@@ -342,26 +361,19 @@ static int gmc_v8_0_polaris_mc_load_microcode(struct amdgpu_device *adev)
 	if (amdgpu_sriov_bios(adev))
 		return 0;
 
-	WREG32(mmMC_SEQ_IO_DEBUG_INDEX, 0x9F);
-	data = RREG32(mmMC_SEQ_IO_DEBUG_DATA);
-	vbios_version = data & 0xf;
-
-	if (vbios_version == 0)
-		return 0;
-
-	if (!adev->mc.fw)
+	if (!adev->gmc.fw)
 		return -EINVAL;
 
-	hdr = (const struct mc_firmware_header_v1_0 *)adev->mc.fw->data;
+	hdr = (const struct mc_firmware_header_v1_0 *)adev->gmc.fw->data;
 	amdgpu_ucode_print_mc_hdr(&hdr->header);
 
-	adev->mc.fw_version = le32_to_cpu(hdr->header.ucode_version);
+	adev->gmc.fw_version = le32_to_cpu(hdr->header.ucode_version);
 	regs_size = le32_to_cpu(hdr->io_debug_size_bytes) / (4 * 2);
 	io_mc_regs = (const __le32 *)
-		(adev->mc.fw->data + le32_to_cpu(hdr->io_debug_array_offset_bytes));
+		(adev->gmc.fw->data + le32_to_cpu(hdr->io_debug_array_offset_bytes));
 	ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
 	fw_data = (const __le32 *)
-		(adev->mc.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+		(adev->gmc.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
 
 	data = RREG32(mmMC_SEQ_MISC0);
 	data &= ~(0x40);
@@ -397,7 +409,7 @@ static int gmc_v8_0_polaris_mc_load_microcode(struct amdgpu_device *adev)
 }
 
 static void gmc_v8_0_vram_gtt_location(struct amdgpu_device *adev,
-				       struct amdgpu_mc *mc)
+				       struct amdgpu_gmc *mc)
 {
 	u64 base = 0;
 
@@ -405,14 +417,9 @@ static void gmc_v8_0_vram_gtt_location(struct amdgpu_device *adev,
 		base = RREG32(mmMC_VM_FB_LOCATION) & 0xFFFF;
 	base <<= 24;
 
-	if (mc->mc_vram_size > 0xFFC0000000ULL) {
-		/* leave room for at least 1024M GTT */
-		dev_warn(adev->dev, "limiting VRAM\n");
-		mc->real_vram_size = 0xFFC0000000ULL;
-		mc->mc_vram_size = 0xFFC0000000ULL;
-	}
-	amdgpu_vram_location(adev, &adev->mc, base);
-	amdgpu_gart_location(adev, mc);
+	amdgpu_gmc_set_agp_default(adev, mc);
+	amdgpu_gmc_vram_location(adev, mc, base);
+	amdgpu_gmc_gart_location(adev, mc, AMDGPU_GART_PLACEMENT_BEST_FIT);
 }
 
 /**
@@ -421,10 +428,11 @@ static void gmc_v8_0_vram_gtt_location(struct amdgpu_device *adev,
  * @adev: amdgpu_device pointer
  *
  * Set the location of vram, gart, and AGP in the GPU's
- * physical address space (CIK).
+ * physical address space (VI).
  */
 static void gmc_v8_0_mc_program(struct amdgpu_device *adev)
 {
+	struct amdgpu_ip_block *ip_block;
 	u32 tmp;
 	int i, j;
 
@@ -438,9 +446,13 @@ static void gmc_v8_0_mc_program(struct amdgpu_device *adev)
 	}
 	WREG32(mmHDP_REG_COHERENCY_FLUSH_CNTL, 0);
 
-	if (gmc_v8_0_wait_for_idle((void *)adev)) {
+	ip_block = amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_GMC);
+	if (!ip_block)
+		return;
+
+	if (gmc_v8_0_wait_for_idle(ip_block))
 		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
-	}
+
 	if (adev->mode_info.num_crtc) {
 		/* Lockout access through VGA aperture*/
 		tmp = RREG32(mmVGA_HDP_CONTROL);
@@ -454,28 +466,27 @@ static void gmc_v8_0_mc_program(struct amdgpu_device *adev)
 	}
 	/* Update configuration */
 	WREG32(mmMC_VM_SYSTEM_APERTURE_LOW_ADDR,
-	       adev->mc.vram_start >> 12);
+	       adev->gmc.vram_start >> 12);
 	WREG32(mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
-	       adev->mc.vram_end >> 12);
+	       adev->gmc.vram_end >> 12);
 	WREG32(mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR,
-	       adev->vram_scratch.gpu_addr >> 12);
+	       adev->mem_scratch.gpu_addr >> 12);
 
 	if (amdgpu_sriov_vf(adev)) {
-		tmp = ((adev->mc.vram_end >> 24) & 0xFFFF) << 16;
-		tmp |= ((adev->mc.vram_start >> 24) & 0xFFFF);
+		tmp = ((adev->gmc.vram_end >> 24) & 0xFFFF) << 16;
+		tmp |= ((adev->gmc.vram_start >> 24) & 0xFFFF);
 		WREG32(mmMC_VM_FB_LOCATION, tmp);
 		/* XXX double check these! */
-		WREG32(mmHDP_NONSURFACE_BASE, (adev->mc.vram_start >> 8));
+		WREG32(mmHDP_NONSURFACE_BASE, (adev->gmc.vram_start >> 8));
 		WREG32(mmHDP_NONSURFACE_INFO, (2 << 7) | (1 << 30));
 		WREG32(mmHDP_NONSURFACE_SIZE, 0x3FFFFFFF);
 	}
 
 	WREG32(mmMC_VM_AGP_BASE, 0);
-	WREG32(mmMC_VM_AGP_TOP, 0x0FFFFFFF);
-	WREG32(mmMC_VM_AGP_BOT, 0x0FFFFFFF);
-	if (gmc_v8_0_wait_for_idle((void *)adev)) {
+	WREG32(mmMC_VM_AGP_TOP, adev->gmc.agp_end >> 22);
+	WREG32(mmMC_VM_AGP_BOT, adev->gmc.agp_start >> 22);
+	if (gmc_v8_0_wait_for_idle(ip_block))
 		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
-	}
 
 	WREG32(mmBIF_FB_EN, BIF_FB_EN__FB_READ_EN_MASK | BIF_FB_EN__FB_WRITE_EN_MASK);
 
@@ -493,23 +504,25 @@ static void gmc_v8_0_mc_program(struct amdgpu_device *adev)
  * @adev: amdgpu_device pointer
  *
  * Look up the amount of vram, vram width, and decide how to place
- * vram and gart within the GPU's physical address space (CIK).
+ * vram and gart within the GPU's physical address space (VI).
  * Returns 0 for success.
  */
 static int gmc_v8_0_mc_init(struct amdgpu_device *adev)
 {
-	adev->mc.vram_width = amdgpu_atombios_get_vram_width(adev);
-	if (!adev->mc.vram_width) {
-		u32 tmp;
+	int r;
+	u32 tmp;
+
+	adev->gmc.vram_width = amdgpu_atombios_get_vram_width(adev);
+	if (!adev->gmc.vram_width) {
 		int chansize, numchan;
 
 		/* Get VRAM informations */
 		tmp = RREG32(mmMC_ARB_RAMCFG);
-		if (REG_GET_FIELD(tmp, MC_ARB_RAMCFG, CHANSIZE)) {
+		if (REG_GET_FIELD(tmp, MC_ARB_RAMCFG, CHANSIZE))
 			chansize = 64;
-		} else {
+		else
 			chansize = 32;
-		}
+
 		tmp = RREG32(mmMC_SHARED_CHMAP);
 		switch (REG_GET_FIELD(tmp, MC_SHARED_CHMAP, NOOFCHAN)) {
 		case 0:
@@ -541,52 +554,93 @@ static int gmc_v8_0_mc_init(struct amdgpu_device *adev)
 			numchan = 16;
 			break;
 		}
-		adev->mc.vram_width = numchan * chansize;
+		adev->gmc.vram_width = numchan * chansize;
 	}
-	/* Could aper size report 0 ? */
-	adev->mc.aper_base = pci_resource_start(adev->pdev, 0);
-	adev->mc.aper_size = pci_resource_len(adev->pdev, 0);
 	/* size in MB on si */
-	adev->mc.mc_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
-	adev->mc.real_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
+	tmp = RREG32(mmCONFIG_MEMSIZE);
+	/* some boards may have garbage in the upper 16 bits */
+	if (tmp & 0xffff0000) {
+		DRM_INFO("Probable bad vram size: 0x%08x\n", tmp);
+		if (tmp & 0xffff)
+			tmp &= 0xffff;
+	}
+	adev->gmc.mc_vram_size = tmp * 1024ULL * 1024ULL;
+	adev->gmc.real_vram_size = adev->gmc.mc_vram_size;
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		r = amdgpu_device_resize_fb_bar(adev);
+		if (r)
+			return r;
+	}
+	adev->gmc.aper_base = pci_resource_start(adev->pdev, 0);
+	adev->gmc.aper_size = pci_resource_len(adev->pdev, 0);
 
 #ifdef CONFIG_X86_64
-	if (adev->flags & AMD_IS_APU) {
-		adev->mc.aper_base = ((u64)RREG32(mmMC_VM_FB_OFFSET)) << 22;
-		adev->mc.aper_size = adev->mc.real_vram_size;
+	if ((adev->flags & AMD_IS_APU) && !amdgpu_passthrough(adev)) {
+		adev->gmc.aper_base = ((u64)RREG32(mmMC_VM_FB_OFFSET)) << 22;
+		adev->gmc.aper_size = adev->gmc.real_vram_size;
 	}
 #endif
 
-	/* In case the PCI BAR is larger than the actual amount of vram */
-	adev->mc.visible_vram_size = adev->mc.aper_size;
-	if (adev->mc.visible_vram_size > adev->mc.real_vram_size)
-		adev->mc.visible_vram_size = adev->mc.real_vram_size;
+	adev->gmc.visible_vram_size = adev->gmc.aper_size;
 
 	/* set the gart size */
 	if (amdgpu_gart_size == -1) {
 		switch (adev->asic_type) {
-		case CHIP_POLARIS11: /* all engines support GPUVM */
 		case CHIP_POLARIS10: /* all engines support GPUVM */
+		case CHIP_POLARIS11: /* all engines support GPUVM */
 		case CHIP_POLARIS12: /* all engines support GPUVM */
+		case CHIP_VEGAM:     /* all engines support GPUVM */
 		default:
-			adev->mc.gart_size = 256ULL << 20;
+			adev->gmc.gart_size = 256ULL << 20;
 			break;
 		case CHIP_TONGA:   /* UVD, VCE do not support GPUVM */
 		case CHIP_FIJI:    /* UVD, VCE do not support GPUVM */
 		case CHIP_CARRIZO: /* UVD, VCE do not support GPUVM, DCE SG support */
 		case CHIP_STONEY:  /* UVD does not support GPUVM, DCE SG support */
-			adev->mc.gart_size = 1024ULL << 20;
+			adev->gmc.gart_size = 1024ULL << 20;
 			break;
 		}
 	} else {
-		adev->mc.gart_size = (u64)amdgpu_gart_size << 20;
+		adev->gmc.gart_size = (u64)amdgpu_gart_size << 20;
 	}
 
-	gmc_v8_0_vram_gtt_location(adev, &adev->mc);
+	adev->gmc.gart_size += adev->pm.smu_prv_buffer_size;
+	gmc_v8_0_vram_gtt_location(adev, &adev->gmc);
 
 	return 0;
 }
 
+/**
+ * gmc_v8_0_flush_gpu_tlb_pasid - tlb flush via pasid
+ *
+ * @adev: amdgpu_device pointer
+ * @pasid: pasid to be flush
+ * @flush_type: type of flush
+ * @all_hub: flush all hubs
+ * @inst: is used to select which instance of KIQ to use for the invalidation
+ *
+ * Flush the TLB for the requested pasid.
+ */
+static void gmc_v8_0_flush_gpu_tlb_pasid(struct amdgpu_device *adev,
+					 uint16_t pasid, uint32_t flush_type,
+					 bool all_hub, uint32_t inst)
+{
+	u32 mask = 0x0;
+	int vmid;
+
+	for (vmid = 1; vmid < 16; vmid++) {
+		u32 tmp = RREG32(mmATC_VMID0_PASID_MAPPING + vmid);
+
+		if ((tmp & ATC_VMID0_PASID_MAPPING__VALID_MASK) &&
+		    (tmp & ATC_VMID0_PASID_MAPPING__PASID_MASK) == pasid)
+			mask |= 1 << vmid;
+	}
+
+	WREG32(mmVM_INVALIDATE_REQUEST, mask);
+	RREG32(mmVM_INVALIDATE_RESPONSE);
+}
+
 /*
  * GART
  * VMID 0 is the physical GPU addresses as used by the kernel.
@@ -595,91 +649,83 @@ static int gmc_v8_0_mc_init(struct amdgpu_device *adev)
  */
 
 /**
- * gmc_v8_0_gart_flush_gpu_tlb - gart tlb flush callback
+ * gmc_v8_0_flush_gpu_tlb - gart tlb flush callback
  *
  * @adev: amdgpu_device pointer
  * @vmid: vm instance to flush
+ * @vmhub: which hub to flush
+ * @flush_type: type of flush
  *
- * Flush the TLB for the requested page table (CIK).
+ * Flush the TLB for the requested page table (VI).
  */
-static void gmc_v8_0_gart_flush_gpu_tlb(struct amdgpu_device *adev,
-					uint32_t vmid)
+static void gmc_v8_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid,
+					uint32_t vmhub, uint32_t flush_type)
 {
-	/* flush hdp cache */
-	WREG32(mmHDP_MEM_COHERENCY_FLUSH_CNTL, 0);
-
 	/* bits 0-15 are the VM contexts0-15 */
 	WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
 }
 
-/**
- * gmc_v8_0_gart_set_pte_pde - update the page tables using MMIO
- *
- * @adev: amdgpu_device pointer
- * @cpu_pt_addr: cpu address of the page table
- * @gpu_page_idx: entry in the page table to update
- * @addr: dst addr to write into pte/pde
- * @flags: access flags
- *
- * Update the page tables using the CPU.
- */
-static int gmc_v8_0_gart_set_pte_pde(struct amdgpu_device *adev,
-				     void *cpu_pt_addr,
-				     uint32_t gpu_page_idx,
-				     uint64_t addr,
-				     uint64_t flags)
+static uint64_t gmc_v8_0_emit_flush_gpu_tlb(struct amdgpu_ring *ring,
+					    unsigned int vmid, uint64_t pd_addr)
 {
-	void __iomem *ptr = (void *)cpu_pt_addr;
-	uint64_t value;
+	uint32_t reg;
 
-	/*
-	 * PTE format on VI:
-	 * 63:40 reserved
-	 * 39:12 4k physical page base address
-	 * 11:7 fragment
-	 * 6 write
-	 * 5 read
-	 * 4 exe
-	 * 3 reserved
-	 * 2 snooped
-	 * 1 system
-	 * 0 valid
-	 *
-	 * PDE format on VI:
-	 * 63:59 block fragment size
-	 * 58:40 reserved
-	 * 39:1 physical base address of PTE
-	 * bits 5:1 must be 0.
-	 * 0 valid
-	 */
-	value = addr & 0x000000FFFFFFF000ULL;
-	value |= flags;
-	writeq(value, ptr + (gpu_page_idx * 8));
+	if (vmid < 8)
+		reg = mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + vmid;
+	else
+		reg = mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vmid - 8;
+	amdgpu_ring_emit_wreg(ring, reg, pd_addr >> 12);
 
-	return 0;
+	/* bits 0-15 are the VM contexts0-15 */
+	amdgpu_ring_emit_wreg(ring, mmVM_INVALIDATE_REQUEST, 1 << vmid);
+
+	return pd_addr;
 }
 
-static uint64_t gmc_v8_0_get_vm_pte_flags(struct amdgpu_device *adev,
-					  uint32_t flags)
+static void gmc_v8_0_emit_pasid_mapping(struct amdgpu_ring *ring, unsigned int vmid,
+					unsigned int pasid)
 {
-	uint64_t pte_flag = 0;
-
-	if (flags & AMDGPU_VM_PAGE_EXECUTABLE)
-		pte_flag |= AMDGPU_PTE_EXECUTABLE;
-	if (flags & AMDGPU_VM_PAGE_READABLE)
-		pte_flag |= AMDGPU_PTE_READABLE;
-	if (flags & AMDGPU_VM_PAGE_WRITEABLE)
-		pte_flag |= AMDGPU_PTE_WRITEABLE;
-	if (flags & AMDGPU_VM_PAGE_PRT)
-		pte_flag |= AMDGPU_PTE_PRT;
-
-	return pte_flag;
+	amdgpu_ring_emit_wreg(ring, mmIH_VMID_0_LUT + vmid, pasid);
 }
 
-static uint64_t gmc_v8_0_get_vm_pde(struct amdgpu_device *adev, uint64_t addr)
+/*
+ * PTE format on VI:
+ * 63:40 reserved
+ * 39:12 4k physical page base address
+ * 11:7 fragment
+ * 6 write
+ * 5 read
+ * 4 exe
+ * 3 reserved
+ * 2 snooped
+ * 1 system
+ * 0 valid
+ *
+ * PDE format on VI:
+ * 63:59 block fragment size
+ * 58:40 reserved
+ * 39:1 physical base address of PTE
+ * bits 5:1 must be 0.
+ * 0 valid
+ */
+
+static void gmc_v8_0_get_vm_pde(struct amdgpu_device *adev, int level,
+				uint64_t *addr, uint64_t *flags)
 {
-	BUG_ON(addr & 0xFFFFFF0000000FFFULL);
-	return addr;
+	BUG_ON(*addr & 0xFFFFFF0000000FFFULL);
+}
+
+static void gmc_v8_0_get_vm_pte(struct amdgpu_device *adev,
+				struct amdgpu_vm *vm,
+				struct amdgpu_bo *bo,
+				uint32_t vm_flags,
+				uint64_t *flags)
+{
+	if (vm_flags & AMDGPU_VM_PAGE_EXECUTABLE)
+		*flags |= AMDGPU_PTE_EXECUTABLE;
+	else
+		*flags &= ~AMDGPU_PTE_EXECUTABLE;
+	*flags &= ~AMDGPU_PTE_PRT;
 }
 
 /**
@@ -712,18 +758,18 @@ static void gmc_v8_0_set_fault_enable_default(struct amdgpu_device *adev,
 }
 
 /**
- * gmc_v8_0_set_prt - set PRT VM fault
+ * gmc_v8_0_set_prt() - set PRT VM fault
  *
  * @adev: amdgpu_device pointer
  * @enable: enable/disable VM fault handling for PRT
-*/
+ */
 static void gmc_v8_0_set_prt(struct amdgpu_device *adev, bool enable)
 {
 	u32 tmp;
 
-	if (enable && !adev->mc.prt_warning) {
+	if (enable && !adev->gmc.prt_warning) {
 		dev_warn(adev->dev, "Disabling VM faults because of PRT request!\n");
-		adev->mc.prt_warning = true;
+		adev->gmc.prt_warning = true;
 	}
 
 	tmp = RREG32(mmVM_PRT_CNTL);
@@ -744,8 +790,10 @@ static void gmc_v8_0_set_prt(struct amdgpu_device *adev, bool enable)
 	WREG32(mmVM_PRT_CNTL, tmp);
 
 	if (enable) {
-		uint32_t low = AMDGPU_VA_RESERVED_SIZE >> AMDGPU_GPU_PAGE_SHIFT;
-		uint32_t high = adev->vm_manager.max_pfn;
+		uint32_t low = AMDGPU_VA_RESERVED_BOTTOM >>
+			AMDGPU_GPU_PAGE_SHIFT;
+		uint32_t high = adev->vm_manager.max_pfn -
+			(AMDGPU_VA_RESERVED_TOP >> AMDGPU_GPU_PAGE_SHIFT);
 
 		WREG32(mmVM_PRT_APERTURE0_LOW_ADDR, low);
 		WREG32(mmVM_PRT_APERTURE1_LOW_ADDR, low);
@@ -775,21 +823,22 @@ static void gmc_v8_0_set_prt(struct amdgpu_device *adev, bool enable)
  * This sets up the TLBs, programs the page tables for VMID0,
  * sets up the hw for VMIDs 1-15 which are allocated on
  * demand, and sets up the global locations for the LDS, GDS,
- * and GPUVM for FSA64 clients (CIK).
+ * and GPUVM for FSA64 clients (VI).
  * Returns 0 for success, errors for failure.
  */
 static int gmc_v8_0_gart_enable(struct amdgpu_device *adev)
 {
-	int r, i;
+	uint64_t table_addr;
 	u32 tmp, field;
+	int i;
 
-	if (adev->gart.robj == NULL) {
+	if (adev->gart.bo == NULL) {
 		dev_err(adev->dev, "No VRAM object for PCIE GART.\n");
 		return -EINVAL;
 	}
-	r = amdgpu_gart_table_vram_pin(adev);
-	if (r)
-		return r;
+	amdgpu_gtt_mgr_recover(&adev->mman.gtt_mgr);
+	table_addr = amdgpu_bo_gpu_offset(adev->gart.bo);
+
 	/* Setup TLB control */
 	tmp = RREG32(mmMC_VM_MX_L1_TLB_CNTL);
 	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
@@ -835,11 +884,11 @@ static int gmc_v8_0_gart_enable(struct amdgpu_device *adev)
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT1_PTE_REQUEST_SNOOP, 0);
 	WREG32(mmVM_L2_CNTL4, tmp);
 	/* setup context0 */
-	WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->mc.gart_start >> 12);
-	WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, adev->mc.gart_end >> 12);
-	WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR, adev->gart.table_addr >> 12);
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->gmc.gart_start >> 12);
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, adev->gmc.gart_end >> 12);
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR, table_addr >> 12);
 	WREG32(mmVM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR,
-			(u32)(adev->dummy_page.addr >> 12));
+			(u32)(adev->dummy_page_addr >> 12));
 	WREG32(mmVM_CONTEXT0_CNTL2, 0);
 	tmp = RREG32(mmVM_CONTEXT0_CNTL);
 	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
@@ -858,18 +907,18 @@ static int gmc_v8_0_gart_enable(struct amdgpu_device *adev)
 	/* set vm size, must be a multiple of 4 */
 	WREG32(mmVM_CONTEXT1_PAGE_TABLE_START_ADDR, 0);
 	WREG32(mmVM_CONTEXT1_PAGE_TABLE_END_ADDR, adev->vm_manager.max_pfn - 1);
-	for (i = 1; i < 16; i++) {
+	for (i = 1; i < AMDGPU_NUM_VMID; i++) {
 		if (i < 8)
 			WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + i,
-			       adev->gart.table_addr >> 12);
+			       table_addr >> 12);
 		else
 			WREG32(mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + i - 8,
-			       adev->gart.table_addr >> 12);
+			       table_addr >> 12);
 	}
 
 	/* enable context1-15 */
 	WREG32(mmVM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR,
-	       (u32)(adev->dummy_page.addr >> 12));
+	       (u32)(adev->dummy_page_addr >> 12));
 	WREG32(mmVM_CONTEXT1_CNTL2, 4);
 	tmp = RREG32(mmVM_CONTEXT1_CNTL);
 	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
@@ -889,11 +938,10 @@ static int gmc_v8_0_gart_enable(struct amdgpu_device *adev)
 	else
 		gmc_v8_0_set_fault_enable_default(adev, true);
 
-	gmc_v8_0_gart_flush_gpu_tlb(adev, 0);
+	gmc_v8_0_flush_gpu_tlb(adev, 0, 0, 0);
 	DRM_INFO("PCIE GART of %uM enabled (table at 0x%016llX).\n",
-		 (unsigned)(adev->mc.gart_size >> 20),
-		 (unsigned long long)adev->gart.table_addr);
-	adev->gart.ready = true;
+		 (unsigned int)(adev->gmc.gart_size >> 20),
+		 (unsigned long long)table_addr);
 	return 0;
 }
 
@@ -901,7 +949,7 @@ static int gmc_v8_0_gart_init(struct amdgpu_device *adev)
 {
 	int r;
 
-	if (adev->gart.robj) {
+	if (adev->gart.bo) {
 		WARN(1, "R600 PCIE GART already initialized\n");
 		return 0;
 	}
@@ -919,7 +967,7 @@ static int gmc_v8_0_gart_init(struct amdgpu_device *adev)
  *
  * @adev: amdgpu_device pointer
  *
- * This disables all VM page table (CIK).
+ * This disables all VM page table (VI).
  */
 static void gmc_v8_0_gart_disable(struct amdgpu_device *adev)
 {
@@ -939,20 +987,6 @@ static void gmc_v8_0_gart_disable(struct amdgpu_device *adev)
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 0);
 	WREG32(mmVM_L2_CNTL, tmp);
 	WREG32(mmVM_L2_CNTL2, 0);
-	amdgpu_gart_table_vram_unpin(adev);
-}
-
-/**
- * gmc_v8_0_gart_fini - vm fini callback
- *
- * @adev: amdgpu_device pointer
- *
- * Tears down the driver GART/VM setup (CIK).
- */
-static void gmc_v8_0_gart_fini(struct amdgpu_device *adev)
-{
-	amdgpu_gart_table_vram_free(adev);
-	amdgpu_gart_fini(adev);
 }
 
 /**
@@ -961,24 +995,26 @@ static void gmc_v8_0_gart_fini(struct amdgpu_device *adev)
  * @adev: amdgpu_device pointer
  * @status: VM_CONTEXT1_PROTECTION_FAULT_STATUS register value
  * @addr: VM_CONTEXT1_PROTECTION_FAULT_ADDR register value
+ * @mc_client: VM_CONTEXT1_PROTECTION_FAULT_MCCLIENT register value
+ * @pasid: debug logging only - no functional use
  *
- * Print human readable fault information (CIK).
+ * Print human readable fault information (VI).
  */
-static void gmc_v8_0_vm_decode_fault(struct amdgpu_device *adev,
-				     u32 status, u32 addr, u32 mc_client)
+static void gmc_v8_0_vm_decode_fault(struct amdgpu_device *adev, u32 status,
+				     u32 addr, u32 mc_client, unsigned int pasid)
 {
-	u32 mc_id;
 	u32 vmid = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS, VMID);
 	u32 protections = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
 					PROTECTIONS);
 	char block[5] = { mc_client >> 24, (mc_client >> 16) & 0xff,
 		(mc_client >> 8) & 0xff, mc_client & 0xff, 0 };
+	u32 mc_id;
 
 	mc_id = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
 			      MEMORY_CLIENT_ID);
 
-	dev_err(adev->dev, "VM fault (0x%02x, vmid %d) at page %u, %s from '%s' (0x%08x) (%d)\n",
-	       protections, vmid, addr,
+	dev_err(adev->dev, "VM fault (0x%02x, vmid %d, pasid %d) at page %u, %s from '%s' (0x%08x) (%d)\n",
+	       protections, vmid, pasid, addr,
 	       REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
 			     MEMORY_CLIENT_RW) ?
 	       "write" : "read", block, mc_client, mc_id);
@@ -1006,60 +1042,81 @@ static int gmc_v8_0_convert_vram_type(int mc_seq_vram_type)
 	}
 }
 
-static int gmc_v8_0_early_init(void *handle)
+static int gmc_v8_0_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	gmc_v8_0_set_gart_funcs(adev);
+	gmc_v8_0_set_gmc_funcs(adev);
 	gmc_v8_0_set_irq_funcs(adev);
 
-	adev->mc.shared_aperture_start = 0x2000000000000000ULL;
-	adev->mc.shared_aperture_end =
-		adev->mc.shared_aperture_start + (4ULL << 30) - 1;
-	adev->mc.private_aperture_start =
-		adev->mc.shared_aperture_end + 1;
-	adev->mc.private_aperture_end =
-		adev->mc.private_aperture_start + (4ULL << 30) - 1;
+	adev->gmc.shared_aperture_start = 0x2000000000000000ULL;
+	adev->gmc.shared_aperture_end =
+		adev->gmc.shared_aperture_start + (4ULL << 30) - 1;
+	adev->gmc.private_aperture_start =
+		adev->gmc.shared_aperture_end + 1;
+	adev->gmc.private_aperture_end =
+		adev->gmc.private_aperture_start + (4ULL << 30) - 1;
+	adev->gmc.noretry_flags = AMDGPU_VM_NORETRY_FLAGS_TF;
 
 	return 0;
 }
 
-static int gmc_v8_0_late_init(void *handle)
+static int gmc_v8_0_late_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (amdgpu_vm_fault_stop != AMDGPU_VM_FAULT_STOP_ALWAYS)
-		return amdgpu_irq_get(adev, &adev->mc.vm_fault, 0);
+		return amdgpu_irq_get(adev, &adev->gmc.vm_fault, 0);
 	else
 		return 0;
 }
 
+static unsigned int gmc_v8_0_get_vbios_fb_size(struct amdgpu_device *adev)
+{
+	u32 d1vga_control = RREG32(mmD1VGA_CONTROL);
+	unsigned int size;
+
+	if (REG_GET_FIELD(d1vga_control, D1VGA_CONTROL, D1VGA_MODE_ENABLE)) {
+		size = AMDGPU_VBIOS_VGA_ALLOCATION;
+	} else {
+		u32 viewport = RREG32(mmVIEWPORT_SIZE);
+
+		size = (REG_GET_FIELD(viewport, VIEWPORT_SIZE, VIEWPORT_HEIGHT) *
+			REG_GET_FIELD(viewport, VIEWPORT_SIZE, VIEWPORT_WIDTH) *
+			4);
+	}
+
+	return size;
+}
+
 #define mmMC_SEQ_MISC0_FIJI 0xA71
 
-static int gmc_v8_0_sw_init(void *handle)
+static int gmc_v8_0_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	int dma_bits;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	set_bit(AMDGPU_GFXHUB(0), adev->vmhubs_mask);
 
 	if (adev->flags & AMD_IS_APU) {
-		adev->mc.vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
+		adev->gmc.vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
 	} else {
 		u32 tmp;
 
-		if (adev->asic_type == CHIP_FIJI)
+		if ((adev->asic_type == CHIP_FIJI) ||
+		    (adev->asic_type == CHIP_VEGAM))
 			tmp = RREG32(mmMC_SEQ_MISC0_FIJI);
 		else
 			tmp = RREG32(mmMC_SEQ_MISC0);
 		tmp &= MC_SEQ_MISC0__MT__MASK;
-		adev->mc.vram_type = gmc_v8_0_convert_vram_type(tmp);
+		adev->gmc.vram_type = gmc_v8_0_convert_vram_type(tmp);
 	}
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 146, &adev->mc.vm_fault);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_GFX_PAGE_INV_FAULT, &adev->gmc.vm_fault);
 	if (r)
 		return r;
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 147, &adev->mc.vm_fault);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_GFX_MEM_PROT_FAULT, &adev->gmc.vm_fault);
 	if (r)
 		return r;
 
@@ -1067,35 +1124,20 @@ static int gmc_v8_0_sw_init(void *handle)
 	 * Currently set to 4GB ((1 << 20) 4k pages).
 	 * Max GPUVM size for cayman and SI is 40 bits.
 	 */
-	amdgpu_vm_adjust_size(adev, 64, 4);
-	adev->vm_manager.max_pfn = adev->vm_manager.vm_size << 18;
+	amdgpu_vm_adjust_size(adev, 64, 9, 1, 40);
 
 	/* Set the internal MC address mask
 	 * This is the max address of the GPU's
 	 * internal address space.
 	 */
-	adev->mc.mc_mask = 0xffffffffffULL; /* 40 bit MC */
+	adev->gmc.mc_mask = 0xffffffffffULL; /* 40 bit MC */
 
-	adev->mc.stolen_size = 256 * 1024;
-
-	/* set DMA mask + need_dma32 flags.
-	 * PCIE - can handle 40-bits.
-	 * IGP - can handle 40-bits
-	 * PCI - dma32 for legacy pci gart, 40 bits on newer asics
-	 */
-	adev->need_dma32 = false;
-	dma_bits = adev->need_dma32 ? 32 : 40;
-	r = pci_set_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
-	if (r) {
-		adev->need_dma32 = true;
-		dma_bits = 32;
-		pr_warn("amdgpu: No suitable DMA available\n");
-	}
-	r = pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
+	r = dma_set_mask_and_coherent(adev->dev, DMA_BIT_MASK(40));
 	if (r) {
-		pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(32));
-		pr_warn("amdgpu: No coherent DMA available\n");
+		pr_warn("No suitable DMA available\n");
+		return r;
 	}
+	adev->need_swiotlb = drm_need_swiotlb(40);
 
 	r = gmc_v8_0_init_microcode(adev);
 	if (r) {
@@ -1107,6 +1149,8 @@ static int gmc_v8_0_sw_init(void *handle)
 	if (r)
 		return r;
 
+	amdgpu_gmc_get_vbios_allocations(adev);
+
 	/* Memory manager */
 	r = amdgpu_bo_init(adev);
 	if (r)
@@ -1122,8 +1166,7 @@ static int gmc_v8_0_sw_init(void *handle)
 	 * amdgpu graphics/compute will use VMIDs 1-7
 	 * amdkfd will use VMIDs 8-15
 	 */
-	adev->vm_manager.id_mgr[0].num_ids = AMDGPU_NUM_OF_VMIDS;
-	adev->vm_manager.num_level = 1;
+	adev->vm_manager.first_kfd_vmid = 8;
 	amdgpu_vm_manager_init(adev);
 
 	/* base offset of vram pages */
@@ -1136,25 +1179,33 @@ static int gmc_v8_0_sw_init(void *handle)
 		adev->vm_manager.vram_base_offset = 0;
 	}
 
+	adev->gmc.vm_fault_info = kmalloc(sizeof(struct kfd_vm_fault_info),
+					GFP_KERNEL);
+	if (!adev->gmc.vm_fault_info)
+		return -ENOMEM;
+	atomic_set(&adev->gmc.vm_fault_info_updated, 0);
+
 	return 0;
 }
 
-static int gmc_v8_0_sw_fini(void *handle)
+static int gmc_v8_0_sw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	amdgpu_vm_manager_fini(adev);
-	gmc_v8_0_gart_fini(adev);
 	amdgpu_gem_force_release(adev);
+	amdgpu_vm_manager_fini(adev);
+	kfree(adev->gmc.vm_fault_info);
+	amdgpu_gart_table_vram_free(adev);
 	amdgpu_bo_fini(adev);
+	amdgpu_ucode_release(&adev->gmc.fw);
 
 	return 0;
 }
 
-static int gmc_v8_0_hw_init(void *handle)
+static int gmc_v8_0_hw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	gmc_v8_0_init_golden_registers(adev);
 
@@ -1180,45 +1231,45 @@ static int gmc_v8_0_hw_init(void *handle)
 	if (r)
 		return r;
 
-	return r;
+	if (amdgpu_emu_mode == 1)
+		return amdgpu_gmc_vram_checking(adev);
+
+	return 0;
 }
 
-static int gmc_v8_0_hw_fini(void *handle)
+static int gmc_v8_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	amdgpu_irq_put(adev, &adev->mc.vm_fault, 0);
+	amdgpu_irq_put(adev, &adev->gmc.vm_fault, 0);
 	gmc_v8_0_gart_disable(adev);
 
 	return 0;
 }
 
-static int gmc_v8_0_suspend(void *handle)
+static int gmc_v8_0_suspend(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	gmc_v8_0_hw_fini(adev);
+	gmc_v8_0_hw_fini(ip_block);
 
 	return 0;
 }
 
-static int gmc_v8_0_resume(void *handle)
+static int gmc_v8_0_resume(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	r = gmc_v8_0_hw_init(adev);
+	r = gmc_v8_0_hw_init(ip_block);
 	if (r)
 		return r;
 
-	amdgpu_vm_reset_all_ids(adev);
+	amdgpu_vmid_reset_all(ip_block->adev);
 
 	return 0;
 }
 
-static bool gmc_v8_0_is_idle(void *handle)
+static bool gmc_v8_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 tmp = RREG32(mmSRBM_STATUS);
 
 	if (tmp & (SRBM_STATUS__MCB_BUSY_MASK | SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
@@ -1228,11 +1279,11 @@ static bool gmc_v8_0_is_idle(void *handle)
 	return true;
 }
 
-static int gmc_v8_0_wait_for_idle(void *handle)
+static int gmc_v8_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
-	unsigned i;
+	unsigned int i;
 	u32 tmp;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
 		/* read MC_STATUS */
@@ -1250,10 +1301,10 @@ static int gmc_v8_0_wait_for_idle(void *handle)
 
 }
 
-static bool gmc_v8_0_check_soft_reset(void *handle)
+static bool gmc_v8_0_check_soft_reset(struct amdgpu_ip_block *ip_block)
 {
 	u32 srbm_soft_reset = 0;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 tmp = RREG32(mmSRBM_STATUS);
 
 	if (tmp & SRBM_STATUS__VMC_BUSY_MASK)
@@ -1266,38 +1317,39 @@ static bool gmc_v8_0_check_soft_reset(void *handle)
 			srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset,
 							SRBM_SOFT_RESET, SOFT_RESET_MC, 1);
 	}
+
 	if (srbm_soft_reset) {
-		adev->mc.srbm_soft_reset = srbm_soft_reset;
+		adev->gmc.srbm_soft_reset = srbm_soft_reset;
 		return true;
-	} else {
-		adev->mc.srbm_soft_reset = 0;
-		return false;
 	}
+
+	adev->gmc.srbm_soft_reset = 0;
+
+	return false;
 }
 
-static int gmc_v8_0_pre_soft_reset(void *handle)
+static int gmc_v8_0_pre_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	if (!adev->mc.srbm_soft_reset)
+	if (!adev->gmc.srbm_soft_reset)
 		return 0;
 
 	gmc_v8_0_mc_stop(adev);
-	if (gmc_v8_0_wait_for_idle(adev)) {
+	if (gmc_v8_0_wait_for_idle(ip_block))
 		dev_warn(adev->dev, "Wait for GMC idle timed out !\n");
-	}
 
 	return 0;
 }
 
-static int gmc_v8_0_soft_reset(void *handle)
+static int gmc_v8_0_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 srbm_soft_reset;
 
-	if (!adev->mc.srbm_soft_reset)
+	if (!adev->gmc.srbm_soft_reset)
 		return 0;
-	srbm_soft_reset = adev->mc.srbm_soft_reset;
+	srbm_soft_reset = adev->gmc.srbm_soft_reset;
 
 	if (srbm_soft_reset) {
 		u32 tmp;
@@ -1321,11 +1373,11 @@ static int gmc_v8_0_soft_reset(void *handle)
 	return 0;
 }
 
-static int gmc_v8_0_post_soft_reset(void *handle)
+static int gmc_v8_0_post_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	if (!adev->mc.srbm_soft_reset)
+	if (!adev->gmc.srbm_soft_reset)
 		return 0;
 
 	gmc_v8_0_mc_resume(adev);
@@ -1334,7 +1386,7 @@ static int gmc_v8_0_post_soft_reset(void *handle)
 
 static int gmc_v8_0_vm_fault_interrupt_state(struct amdgpu_device *adev,
 					     struct amdgpu_irq_src *src,
-					     unsigned type,
+					     unsigned int type,
 					     enum amdgpu_interrupt_state state)
 {
 	u32 tmp;
@@ -1378,7 +1430,7 @@ static int gmc_v8_0_process_interrupt(struct amdgpu_device *adev,
 				      struct amdgpu_irq_src *source,
 				      struct amdgpu_iv_entry *entry)
 {
-	u32 addr, status, mc_client;
+	u32 addr, status, mc_client, vmid;
 
 	if (amdgpu_sriov_vf(adev)) {
 		dev_err(adev->dev, "GPU fault detected: %d 0x%08x\n",
@@ -1396,17 +1448,54 @@ static int gmc_v8_0_process_interrupt(struct amdgpu_device *adev,
 	if (!addr && !status)
 		return 0;
 
+	amdgpu_vm_update_fault_cache(adev, entry->pasid,
+				     ((u64)addr) << AMDGPU_GPU_PAGE_SHIFT, status, AMDGPU_GFXHUB(0));
+
 	if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_FIRST)
 		gmc_v8_0_set_fault_enable_default(adev, false);
 
 	if (printk_ratelimit()) {
+		struct amdgpu_task_info *task_info;
+
 		dev_err(adev->dev, "GPU fault detected: %d 0x%08x\n",
 			entry->src_id, entry->src_data[0]);
+
+		task_info = amdgpu_vm_get_task_info_pasid(adev, entry->pasid);
+		if (task_info) {
+			amdgpu_vm_print_task_info(adev, task_info);
+			amdgpu_vm_put_task_info(task_info);
+		}
+
 		dev_err(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x%08X\n",
-			addr);
+				addr);
 		dev_err(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x%08X\n",
 			status);
-		gmc_v8_0_vm_decode_fault(adev, status, addr, mc_client);
+
+		gmc_v8_0_vm_decode_fault(adev, status, addr, mc_client,
+					 entry->pasid);
+	}
+
+	vmid = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+			     VMID);
+	if (amdgpu_amdkfd_is_kfd_vmid(adev, vmid)
+		&& !atomic_read(&adev->gmc.vm_fault_info_updated)) {
+		struct kfd_vm_fault_info *info = adev->gmc.vm_fault_info;
+		u32 protections = REG_GET_FIELD(status,
+					VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+					PROTECTIONS);
+
+		info->vmid = vmid;
+		info->mc_id = REG_GET_FIELD(status,
+					    VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+					    MEMORY_CLIENT_ID);
+		info->status = status;
+		info->page_addr = addr;
+		info->prot_valid = protections & 0x7 ? true : false;
+		info->prot_read = protections & 0x8 ? true : false;
+		info->prot_write = protections & 0x10 ? true : false;
+		info->prot_exec = protections & 0x20 ? true : false;
+		mb();
+		atomic_set(&adev->gmc.vm_fault_info_updated, 1);
 	}
 
 	return 0;
@@ -1572,10 +1661,10 @@ static void fiji_update_mc_light_sleep(struct amdgpu_device *adev,
 	}
 }
 
-static int gmc_v8_0_set_clockgating_state(void *handle,
+static int gmc_v8_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (amdgpu_sriov_vf(adev))
 		return 0;
@@ -1593,15 +1682,15 @@ static int gmc_v8_0_set_clockgating_state(void *handle,
 	return 0;
 }
 
-static int gmc_v8_0_set_powergating_state(void *handle,
+static int gmc_v8_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	return 0;
 }
 
-static void gmc_v8_0_get_clockgating_state(void *handle, u32 *flags)
+static void gmc_v8_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int data;
 
 	if (amdgpu_sriov_vf(adev))
@@ -1638,12 +1727,15 @@ static const struct amd_ip_funcs gmc_v8_0_ip_funcs = {
 	.get_clockgating_state = gmc_v8_0_get_clockgating_state,
 };
 
-static const struct amdgpu_gart_funcs gmc_v8_0_gart_funcs = {
-	.flush_gpu_tlb = gmc_v8_0_gart_flush_gpu_tlb,
-	.set_pte_pde = gmc_v8_0_gart_set_pte_pde,
+static const struct amdgpu_gmc_funcs gmc_v8_0_gmc_funcs = {
+	.flush_gpu_tlb = gmc_v8_0_flush_gpu_tlb,
+	.flush_gpu_tlb_pasid = gmc_v8_0_flush_gpu_tlb_pasid,
+	.emit_flush_gpu_tlb = gmc_v8_0_emit_flush_gpu_tlb,
+	.emit_pasid_mapping = gmc_v8_0_emit_pasid_mapping,
 	.set_prt = gmc_v8_0_set_prt,
-	.get_vm_pte_flags = gmc_v8_0_get_vm_pte_flags,
-	.get_vm_pde = gmc_v8_0_get_vm_pde
+	.get_vm_pde = gmc_v8_0_get_vm_pde,
+	.get_vm_pte = gmc_v8_0_get_vm_pte,
+	.get_vbios_fb_size = gmc_v8_0_get_vbios_fb_size,
 };
 
 static const struct amdgpu_irq_src_funcs gmc_v8_0_irq_funcs = {
@@ -1651,20 +1743,18 @@ static const struct amdgpu_irq_src_funcs gmc_v8_0_irq_funcs = {
 	.process = gmc_v8_0_process_interrupt,
 };
 
-static void gmc_v8_0_set_gart_funcs(struct amdgpu_device *adev)
+static void gmc_v8_0_set_gmc_funcs(struct amdgpu_device *adev)
 {
-	if (adev->gart.gart_funcs == NULL)
-		adev->gart.gart_funcs = &gmc_v8_0_gart_funcs;
+	adev->gmc.gmc_funcs = &gmc_v8_0_gmc_funcs;
 }
 
 static void gmc_v8_0_set_irq_funcs(struct amdgpu_device *adev)
 {
-	adev->mc.vm_fault.num_types = 1;
-	adev->mc.vm_fault.funcs = &gmc_v8_0_irq_funcs;
+	adev->gmc.vm_fault.num_types = 1;
+	adev->gmc.vm_fault.funcs = &gmc_v8_0_irq_funcs;
 }
 
-const struct amdgpu_ip_block_version gmc_v8_0_ip_block =
-{
+const struct amdgpu_ip_block_version gmc_v8_0_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_GMC,
 	.major = 8,
 	.minor = 0,
@@ -1672,8 +1762,7 @@ const struct amdgpu_ip_block_version gmc_v8_0_ip_block =
 	.funcs = &gmc_v8_0_ip_funcs,
 };
 
-const struct amdgpu_ip_block_version gmc_v8_1_ip_block =
-{
+const struct amdgpu_ip_block_version gmc_v8_1_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_GMC,
 	.major = 8,
 	.minor = 1,
@@ -1681,8 +1770,7 @@ const struct amdgpu_ip_block_version gmc_v8_1_ip_block =
 	.funcs = &gmc_v8_0_ip_funcs,
 };
 
-const struct amdgpu_ip_block_version gmc_v8_5_ip_block =
-{
+const struct amdgpu_ip_block_version gmc_v8_5_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_GMC,
 	.major = 8,
 	.minor = 5,
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
index d04d0b123212..0d1dd587db5f 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
@@ -20,107 +20,434 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  *
  */
+
 #include <linux/firmware.h>
+#include <linux/pci.h>
+
+#include <drm/drm_cache.h>
+
 #include "amdgpu.h"
 #include "gmc_v9_0.h"
 #include "amdgpu_atomfirmware.h"
-
-#include "vega10/soc15ip.h"
-#include "vega10/HDP/hdp_4_0_offset.h"
-#include "vega10/HDP/hdp_4_0_sh_mask.h"
-#include "vega10/GC/gc_9_0_sh_mask.h"
-#include "vega10/DC/dce_12_0_offset.h"
-#include "vega10/DC/dce_12_0_sh_mask.h"
-#include "vega10/vega10_enum.h"
-
+#include "amdgpu_gem.h"
+
+#include "gc/gc_9_0_sh_mask.h"
+#include "dce/dce_12_0_offset.h"
+#include "dce/dce_12_0_sh_mask.h"
+#include "vega10_enum.h"
+#include "mmhub/mmhub_1_0_offset.h"
+#include "athub/athub_1_0_sh_mask.h"
+#include "athub/athub_1_0_offset.h"
+#include "oss/osssys_4_0_offset.h"
+
+#include "soc15.h"
+#include "soc15d.h"
 #include "soc15_common.h"
+#include "umc/umc_6_0_sh_mask.h"
 
-#include "nbio_v6_1.h"
-#include "nbio_v7_0.h"
 #include "gfxhub_v1_0.h"
 #include "mmhub_v1_0.h"
+#include "athub_v1_0.h"
+#include "gfxhub_v1_1.h"
+#include "gfxhub_v1_2.h"
+#include "mmhub_v9_4.h"
+#include "mmhub_v1_7.h"
+#include "mmhub_v1_8.h"
+#include "umc_v6_1.h"
+#include "umc_v6_0.h"
+#include "umc_v6_7.h"
+#include "umc_v12_0.h"
+#include "hdp_v4_0.h"
+#include "mca_v3_0.h"
+
+#include "ivsrcid/vmc/irqsrcs_vmc_1_0.h"
+
+#include "amdgpu_ras.h"
+#include "amdgpu_xgmi.h"
+
+/* add these here since we already include dce12 headers and these are for DCN */
+#define mmHUBP0_DCSURF_PRI_VIEWPORT_DIMENSION                                                          0x055d
+#define mmHUBP0_DCSURF_PRI_VIEWPORT_DIMENSION_BASE_IDX                                                 2
+#define HUBP0_DCSURF_PRI_VIEWPORT_DIMENSION__PRI_VIEWPORT_WIDTH__SHIFT                                        0x0
+#define HUBP0_DCSURF_PRI_VIEWPORT_DIMENSION__PRI_VIEWPORT_HEIGHT__SHIFT                                       0x10
+#define HUBP0_DCSURF_PRI_VIEWPORT_DIMENSION__PRI_VIEWPORT_WIDTH_MASK                                          0x00003FFFL
+#define HUBP0_DCSURF_PRI_VIEWPORT_DIMENSION__PRI_VIEWPORT_HEIGHT_MASK                                         0x3FFF0000L
+#define mmDCHUBBUB_SDPIF_MMIO_CNTRL_0                                                                  0x049d
+#define mmDCHUBBUB_SDPIF_MMIO_CNTRL_0_BASE_IDX                                                         2
+
+#define mmHUBP0_DCSURF_PRI_VIEWPORT_DIMENSION_DCN2                                                          0x05ea
+#define mmHUBP0_DCSURF_PRI_VIEWPORT_DIMENSION_DCN2_BASE_IDX                                                 2
+
+static const char * const gfxhub_client_ids[] = {
+	"CB",
+	"DB",
+	"IA",
+	"WD",
+	"CPF",
+	"CPC",
+	"CPG",
+	"RLC",
+	"TCP",
+	"SQC (inst)",
+	"SQC (data)",
+	"SQG",
+	"PA",
+};
 
-#define mmDF_CS_AON0_DramBaseAddress0                                                                  0x0044
-#define mmDF_CS_AON0_DramBaseAddress0_BASE_IDX                                                         0
-//DF_CS_AON0_DramBaseAddress0
-#define DF_CS_AON0_DramBaseAddress0__AddrRngVal__SHIFT                                                        0x0
-#define DF_CS_AON0_DramBaseAddress0__LgcyMmioHoleEn__SHIFT                                                    0x1
-#define DF_CS_AON0_DramBaseAddress0__IntLvNumChan__SHIFT                                                      0x4
-#define DF_CS_AON0_DramBaseAddress0__IntLvAddrSel__SHIFT                                                      0x8
-#define DF_CS_AON0_DramBaseAddress0__DramBaseAddr__SHIFT                                                      0xc
-#define DF_CS_AON0_DramBaseAddress0__AddrRngVal_MASK                                                          0x00000001L
-#define DF_CS_AON0_DramBaseAddress0__LgcyMmioHoleEn_MASK                                                      0x00000002L
-#define DF_CS_AON0_DramBaseAddress0__IntLvNumChan_MASK                                                        0x000000F0L
-#define DF_CS_AON0_DramBaseAddress0__IntLvAddrSel_MASK                                                        0x00000700L
-#define DF_CS_AON0_DramBaseAddress0__DramBaseAddr_MASK                                                        0xFFFFF000L
-
-/* XXX Move this macro to VEGA10 header file, which is like vid.h for VI.*/
-#define AMDGPU_NUM_OF_VMIDS			8
-
-static const u32 golden_settings_vega10_hdp[] =
-{
-	0xf64, 0x0fffffff, 0x00000000,
-	0xf65, 0x0fffffff, 0x00000000,
-	0xf66, 0x0fffffff, 0x00000000,
-	0xf67, 0x0fffffff, 0x00000000,
-	0xf68, 0x0fffffff, 0x00000000,
-	0xf6a, 0x0fffffff, 0x00000000,
-	0xf6b, 0x0fffffff, 0x00000000,
-	0xf6c, 0x0fffffff, 0x00000000,
-	0xf6d, 0x0fffffff, 0x00000000,
-	0xf6e, 0x0fffffff, 0x00000000,
+static const char *mmhub_client_ids_raven[][2] = {
+	[0][0] = "MP1",
+	[1][0] = "MP0",
+	[2][0] = "VCN",
+	[3][0] = "VCNU",
+	[4][0] = "HDP",
+	[5][0] = "DCE",
+	[13][0] = "UTCL2",
+	[19][0] = "TLS",
+	[26][0] = "OSS",
+	[27][0] = "SDMA0",
+	[0][1] = "MP1",
+	[1][1] = "MP0",
+	[2][1] = "VCN",
+	[3][1] = "VCNU",
+	[4][1] = "HDP",
+	[5][1] = "XDP",
+	[6][1] = "DBGU0",
+	[7][1] = "DCE",
+	[8][1] = "DCEDWB0",
+	[9][1] = "DCEDWB1",
+	[26][1] = "OSS",
+	[27][1] = "SDMA0",
 };
 
-static int gmc_v9_0_vm_fault_interrupt_state(struct amdgpu_device *adev,
-					struct amdgpu_irq_src *src,
-					unsigned type,
-					enum amdgpu_interrupt_state state)
+static const char *mmhub_client_ids_renoir[][2] = {
+	[0][0] = "MP1",
+	[1][0] = "MP0",
+	[2][0] = "HDP",
+	[4][0] = "DCEDMC",
+	[5][0] = "DCEVGA",
+	[13][0] = "UTCL2",
+	[19][0] = "TLS",
+	[26][0] = "OSS",
+	[27][0] = "SDMA0",
+	[28][0] = "VCN",
+	[29][0] = "VCNU",
+	[30][0] = "JPEG",
+	[0][1] = "MP1",
+	[1][1] = "MP0",
+	[2][1] = "HDP",
+	[3][1] = "XDP",
+	[6][1] = "DBGU0",
+	[7][1] = "DCEDMC",
+	[8][1] = "DCEVGA",
+	[9][1] = "DCEDWB",
+	[26][1] = "OSS",
+	[27][1] = "SDMA0",
+	[28][1] = "VCN",
+	[29][1] = "VCNU",
+	[30][1] = "JPEG",
+};
+
+static const char *mmhub_client_ids_vega10[][2] = {
+	[0][0] = "MP0",
+	[1][0] = "UVD",
+	[2][0] = "UVDU",
+	[3][0] = "HDP",
+	[13][0] = "UTCL2",
+	[14][0] = "OSS",
+	[15][0] = "SDMA1",
+	[32+0][0] = "VCE0",
+	[32+1][0] = "VCE0U",
+	[32+2][0] = "XDMA",
+	[32+3][0] = "DCE",
+	[32+4][0] = "MP1",
+	[32+14][0] = "SDMA0",
+	[0][1] = "MP0",
+	[1][1] = "UVD",
+	[2][1] = "UVDU",
+	[3][1] = "DBGU0",
+	[4][1] = "HDP",
+	[5][1] = "XDP",
+	[14][1] = "OSS",
+	[15][1] = "SDMA0",
+	[32+0][1] = "VCE0",
+	[32+1][1] = "VCE0U",
+	[32+2][1] = "XDMA",
+	[32+3][1] = "DCE",
+	[32+4][1] = "DCEDWB",
+	[32+5][1] = "MP1",
+	[32+6][1] = "DBGU1",
+	[32+14][1] = "SDMA1",
+};
+
+static const char *mmhub_client_ids_vega12[][2] = {
+	[0][0] = "MP0",
+	[1][0] = "VCE0",
+	[2][0] = "VCE0U",
+	[3][0] = "HDP",
+	[13][0] = "UTCL2",
+	[14][0] = "OSS",
+	[15][0] = "SDMA1",
+	[32+0][0] = "DCE",
+	[32+1][0] = "XDMA",
+	[32+2][0] = "UVD",
+	[32+3][0] = "UVDU",
+	[32+4][0] = "MP1",
+	[32+15][0] = "SDMA0",
+	[0][1] = "MP0",
+	[1][1] = "VCE0",
+	[2][1] = "VCE0U",
+	[3][1] = "DBGU0",
+	[4][1] = "HDP",
+	[5][1] = "XDP",
+	[14][1] = "OSS",
+	[15][1] = "SDMA0",
+	[32+0][1] = "DCE",
+	[32+1][1] = "DCEDWB",
+	[32+2][1] = "XDMA",
+	[32+3][1] = "UVD",
+	[32+4][1] = "UVDU",
+	[32+5][1] = "MP1",
+	[32+6][1] = "DBGU1",
+	[32+15][1] = "SDMA1",
+};
+
+static const char *mmhub_client_ids_vega20[][2] = {
+	[0][0] = "XDMA",
+	[1][0] = "DCE",
+	[2][0] = "VCE0",
+	[3][0] = "VCE0U",
+	[4][0] = "UVD",
+	[5][0] = "UVD1U",
+	[13][0] = "OSS",
+	[14][0] = "HDP",
+	[15][0] = "SDMA0",
+	[32+0][0] = "UVD",
+	[32+1][0] = "UVDU",
+	[32+2][0] = "MP1",
+	[32+3][0] = "MP0",
+	[32+12][0] = "UTCL2",
+	[32+14][0] = "SDMA1",
+	[0][1] = "XDMA",
+	[1][1] = "DCE",
+	[2][1] = "DCEDWB",
+	[3][1] = "VCE0",
+	[4][1] = "VCE0U",
+	[5][1] = "UVD1",
+	[6][1] = "UVD1U",
+	[7][1] = "DBGU0",
+	[8][1] = "XDP",
+	[13][1] = "OSS",
+	[14][1] = "HDP",
+	[15][1] = "SDMA0",
+	[32+0][1] = "UVD",
+	[32+1][1] = "UVDU",
+	[32+2][1] = "DBGU1",
+	[32+3][1] = "MP1",
+	[32+4][1] = "MP0",
+	[32+14][1] = "SDMA1",
+};
+
+static const char *mmhub_client_ids_arcturus[][2] = {
+	[0][0] = "DBGU1",
+	[1][0] = "XDP",
+	[2][0] = "MP1",
+	[14][0] = "HDP",
+	[171][0] = "JPEG",
+	[172][0] = "VCN",
+	[173][0] = "VCNU",
+	[203][0] = "JPEG1",
+	[204][0] = "VCN1",
+	[205][0] = "VCN1U",
+	[256][0] = "SDMA0",
+	[257][0] = "SDMA1",
+	[258][0] = "SDMA2",
+	[259][0] = "SDMA3",
+	[260][0] = "SDMA4",
+	[261][0] = "SDMA5",
+	[262][0] = "SDMA6",
+	[263][0] = "SDMA7",
+	[384][0] = "OSS",
+	[0][1] = "DBGU1",
+	[1][1] = "XDP",
+	[2][1] = "MP1",
+	[14][1] = "HDP",
+	[171][1] = "JPEG",
+	[172][1] = "VCN",
+	[173][1] = "VCNU",
+	[203][1] = "JPEG1",
+	[204][1] = "VCN1",
+	[205][1] = "VCN1U",
+	[256][1] = "SDMA0",
+	[257][1] = "SDMA1",
+	[258][1] = "SDMA2",
+	[259][1] = "SDMA3",
+	[260][1] = "SDMA4",
+	[261][1] = "SDMA5",
+	[262][1] = "SDMA6",
+	[263][1] = "SDMA7",
+	[384][1] = "OSS",
+};
+
+static const char *mmhub_client_ids_aldebaran[][2] = {
+	[2][0] = "MP1",
+	[3][0] = "MP0",
+	[32+1][0] = "DBGU_IO0",
+	[32+2][0] = "DBGU_IO2",
+	[32+4][0] = "MPIO",
+	[96+11][0] = "JPEG0",
+	[96+12][0] = "VCN0",
+	[96+13][0] = "VCNU0",
+	[128+11][0] = "JPEG1",
+	[128+12][0] = "VCN1",
+	[128+13][0] = "VCNU1",
+	[160+1][0] = "XDP",
+	[160+14][0] = "HDP",
+	[256+0][0] = "SDMA0",
+	[256+1][0] = "SDMA1",
+	[256+2][0] = "SDMA2",
+	[256+3][0] = "SDMA3",
+	[256+4][0] = "SDMA4",
+	[384+0][0] = "OSS",
+	[2][1] = "MP1",
+	[3][1] = "MP0",
+	[32+1][1] = "DBGU_IO0",
+	[32+2][1] = "DBGU_IO2",
+	[32+4][1] = "MPIO",
+	[96+11][1] = "JPEG0",
+	[96+12][1] = "VCN0",
+	[96+13][1] = "VCNU0",
+	[128+11][1] = "JPEG1",
+	[128+12][1] = "VCN1",
+	[128+13][1] = "VCNU1",
+	[160+1][1] = "XDP",
+	[160+14][1] = "HDP",
+	[256+0][1] = "SDMA0",
+	[256+1][1] = "SDMA1",
+	[256+2][1] = "SDMA2",
+	[256+3][1] = "SDMA3",
+	[256+4][1] = "SDMA4",
+	[384+0][1] = "OSS",
+};
+
+static const struct soc15_reg_golden golden_settings_mmhub_1_0_0[] = {
+	SOC15_REG_GOLDEN_VALUE(MMHUB, 0, mmDAGB1_WRCLI2, 0x00000007, 0xfe5fe0fa),
+	SOC15_REG_GOLDEN_VALUE(MMHUB, 0, mmMMEA1_DRAM_WR_CLI2GRP_MAP0, 0x00000030, 0x55555565)
+};
+
+static const struct soc15_reg_golden golden_settings_athub_1_0_0[] = {
+	SOC15_REG_GOLDEN_VALUE(ATHUB, 0, mmRPB_ARB_CNTL, 0x0000ff00, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(ATHUB, 0, mmRPB_ARB_CNTL2, 0x00ff00ff, 0x00080008)
+};
+
+static const uint32_t ecc_umc_mcumc_ctrl_addrs[] = {
+	(0x000143c0 + 0x00000000),
+	(0x000143c0 + 0x00000800),
+	(0x000143c0 + 0x00001000),
+	(0x000143c0 + 0x00001800),
+	(0x000543c0 + 0x00000000),
+	(0x000543c0 + 0x00000800),
+	(0x000543c0 + 0x00001000),
+	(0x000543c0 + 0x00001800),
+	(0x000943c0 + 0x00000000),
+	(0x000943c0 + 0x00000800),
+	(0x000943c0 + 0x00001000),
+	(0x000943c0 + 0x00001800),
+	(0x000d43c0 + 0x00000000),
+	(0x000d43c0 + 0x00000800),
+	(0x000d43c0 + 0x00001000),
+	(0x000d43c0 + 0x00001800),
+	(0x001143c0 + 0x00000000),
+	(0x001143c0 + 0x00000800),
+	(0x001143c0 + 0x00001000),
+	(0x001143c0 + 0x00001800),
+	(0x001543c0 + 0x00000000),
+	(0x001543c0 + 0x00000800),
+	(0x001543c0 + 0x00001000),
+	(0x001543c0 + 0x00001800),
+	(0x001943c0 + 0x00000000),
+	(0x001943c0 + 0x00000800),
+	(0x001943c0 + 0x00001000),
+	(0x001943c0 + 0x00001800),
+	(0x001d43c0 + 0x00000000),
+	(0x001d43c0 + 0x00000800),
+	(0x001d43c0 + 0x00001000),
+	(0x001d43c0 + 0x00001800),
+};
+
+static const uint32_t ecc_umc_mcumc_ctrl_mask_addrs[] = {
+	(0x000143e0 + 0x00000000),
+	(0x000143e0 + 0x00000800),
+	(0x000143e0 + 0x00001000),
+	(0x000143e0 + 0x00001800),
+	(0x000543e0 + 0x00000000),
+	(0x000543e0 + 0x00000800),
+	(0x000543e0 + 0x00001000),
+	(0x000543e0 + 0x00001800),
+	(0x000943e0 + 0x00000000),
+	(0x000943e0 + 0x00000800),
+	(0x000943e0 + 0x00001000),
+	(0x000943e0 + 0x00001800),
+	(0x000d43e0 + 0x00000000),
+	(0x000d43e0 + 0x00000800),
+	(0x000d43e0 + 0x00001000),
+	(0x000d43e0 + 0x00001800),
+	(0x001143e0 + 0x00000000),
+	(0x001143e0 + 0x00000800),
+	(0x001143e0 + 0x00001000),
+	(0x001143e0 + 0x00001800),
+	(0x001543e0 + 0x00000000),
+	(0x001543e0 + 0x00000800),
+	(0x001543e0 + 0x00001000),
+	(0x001543e0 + 0x00001800),
+	(0x001943e0 + 0x00000000),
+	(0x001943e0 + 0x00000800),
+	(0x001943e0 + 0x00001000),
+	(0x001943e0 + 0x00001800),
+	(0x001d43e0 + 0x00000000),
+	(0x001d43e0 + 0x00000800),
+	(0x001d43e0 + 0x00001000),
+	(0x001d43e0 + 0x00001800),
+};
+
+static int gmc_v9_0_ecc_interrupt_state(struct amdgpu_device *adev,
+		struct amdgpu_irq_src *src,
+		unsigned int type,
+		enum amdgpu_interrupt_state state)
 {
-	struct amdgpu_vmhub *hub;
-	u32 tmp, reg, bits, i;
+	u32 bits, i, tmp, reg;
 
-	bits = VM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
-		VM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
-		VM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
-		VM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
-		VM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
-		VM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
-		VM_CONTEXT1_CNTL__EXECUTE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK;
+	/* Devices newer then VEGA10/12 shall have these programming
+	 * sequences performed by PSP BL
+	 */
+	if (adev->asic_type >= CHIP_VEGA20)
+		return 0;
+
+	bits = 0x7f;
 
 	switch (state) {
 	case AMDGPU_IRQ_STATE_DISABLE:
-		/* MM HUB */
-		hub = &adev->vmhub[AMDGPU_MMHUB];
-		for (i = 0; i< 16; i++) {
-			reg = hub->vm_context0_cntl + i;
+		for (i = 0; i < ARRAY_SIZE(ecc_umc_mcumc_ctrl_addrs); i++) {
+			reg = ecc_umc_mcumc_ctrl_addrs[i];
 			tmp = RREG32(reg);
 			tmp &= ~bits;
 			WREG32(reg, tmp);
 		}
-
-		/* GFX HUB */
-		hub = &adev->vmhub[AMDGPU_GFXHUB];
-		for (i = 0; i < 16; i++) {
-			reg = hub->vm_context0_cntl + i;
+		for (i = 0; i < ARRAY_SIZE(ecc_umc_mcumc_ctrl_mask_addrs); i++) {
+			reg = ecc_umc_mcumc_ctrl_mask_addrs[i];
 			tmp = RREG32(reg);
 			tmp &= ~bits;
 			WREG32(reg, tmp);
 		}
 		break;
 	case AMDGPU_IRQ_STATE_ENABLE:
-		/* MM HUB */
-		hub = &adev->vmhub[AMDGPU_MMHUB];
-		for (i = 0; i< 16; i++) {
-			reg = hub->vm_context0_cntl + i;
+		for (i = 0; i < ARRAY_SIZE(ecc_umc_mcumc_ctrl_addrs); i++) {
+			reg = ecc_umc_mcumc_ctrl_addrs[i];
 			tmp = RREG32(reg);
 			tmp |= bits;
 			WREG32(reg, tmp);
 		}
-
-		/* GFX HUB */
-		hub = &adev->vmhub[AMDGPU_GFXHUB];
-		for (i = 0; i < 16; i++) {
-			reg = hub->vm_context0_cntl + i;
+		for (i = 0; i < ARRAY_SIZE(ecc_umc_mcumc_ctrl_mask_addrs); i++) {
+			reg = ecc_umc_mcumc_ctrl_mask_addrs[i];
 			tmp = RREG32(reg);
 			tmp |= bits;
 			WREG32(reg, tmp);
@@ -133,36 +460,280 @@ static int gmc_v9_0_vm_fault_interrupt_state(struct amdgpu_device *adev,
 	return 0;
 }
 
+static int gmc_v9_0_vm_fault_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *src,
+					unsigned int type,
+					enum amdgpu_interrupt_state state)
+{
+	struct amdgpu_vmhub *hub;
+	u32 tmp, reg, bits, i, j;
+
+	bits = VM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		VM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		VM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		VM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		VM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		VM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		VM_CONTEXT1_CNTL__EXECUTE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		for_each_set_bit(j, adev->vmhubs_mask, AMDGPU_MAX_VMHUBS) {
+			hub = &adev->vmhub[j];
+			for (i = 0; i < 16; i++) {
+				reg = hub->vm_context0_cntl + i;
+
+				/* This works because this interrupt is only
+				 * enabled at init/resume and disabled in
+				 * fini/suspend, so the overall state doesn't
+				 * change over the course of suspend/resume.
+				 */
+				if (adev->in_s0ix && (j == AMDGPU_GFXHUB(0)))
+					continue;
+
+				if (j >= AMDGPU_MMHUB0(0))
+					tmp = RREG32_SOC15_IP(MMHUB, reg);
+				else
+					tmp = RREG32_XCC(reg, j);
+
+				tmp &= ~bits;
+
+				if (j >= AMDGPU_MMHUB0(0))
+					WREG32_SOC15_IP(MMHUB, reg, tmp);
+				else
+					WREG32_XCC(reg, tmp, j);
+			}
+		}
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		for_each_set_bit(j, adev->vmhubs_mask, AMDGPU_MAX_VMHUBS) {
+			hub = &adev->vmhub[j];
+			for (i = 0; i < 16; i++) {
+				reg = hub->vm_context0_cntl + i;
+
+				/* This works because this interrupt is only
+				 * enabled at init/resume and disabled in
+				 * fini/suspend, so the overall state doesn't
+				 * change over the course of suspend/resume.
+				 */
+				if (adev->in_s0ix && (j == AMDGPU_GFXHUB(0)))
+					continue;
+
+				if (j >= AMDGPU_MMHUB0(0))
+					tmp = RREG32_SOC15_IP(MMHUB, reg);
+				else
+					tmp = RREG32_XCC(reg, j);
+
+				tmp |= bits;
+
+				if (j >= AMDGPU_MMHUB0(0))
+					WREG32_SOC15_IP(MMHUB, reg, tmp);
+				else
+					WREG32_XCC(reg, tmp, j);
+			}
+		}
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
 static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev,
-				struct amdgpu_irq_src *source,
-				struct amdgpu_iv_entry *entry)
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
 {
-	struct amdgpu_vmhub *hub = &adev->vmhub[entry->vm_id_src];
-	uint32_t status = 0;
+	bool retry_fault = !!(entry->src_data[1] & 0x80);
+	bool write_fault = !!(entry->src_data[1] & 0x20);
+	uint32_t status = 0, cid = 0, rw = 0, fed = 0;
+	struct amdgpu_task_info *task_info;
+	struct amdgpu_vmhub *hub;
+	const char *mmhub_cid;
+	const char *hub_name;
+	unsigned int vmhub;
 	u64 addr;
+	uint32_t cam_index = 0;
+	int ret, xcc_id = 0;
+	uint32_t node_id;
+
+	node_id = entry->node_id;
 
 	addr = (u64)entry->src_data[0] << 12;
 	addr |= ((u64)entry->src_data[1] & 0xf) << 44;
 
-	if (!amdgpu_sriov_vf(adev)) {
-		status = RREG32(hub->vm_l2_pro_fault_status);
-		WREG32_P(hub->vm_l2_pro_fault_cntl, 1, ~1);
+	if (entry->client_id == SOC15_IH_CLIENTID_VMC) {
+		hub_name = "mmhub0";
+		vmhub = AMDGPU_MMHUB0(node_id / 4);
+	} else if (entry->client_id == SOC15_IH_CLIENTID_VMC1) {
+		hub_name = "mmhub1";
+		vmhub = AMDGPU_MMHUB1(0);
+	} else {
+		hub_name = "gfxhub0";
+		if (adev->gfx.funcs->ih_node_to_logical_xcc) {
+			xcc_id = adev->gfx.funcs->ih_node_to_logical_xcc(adev,
+				node_id);
+			if (xcc_id < 0)
+				xcc_id = 0;
+		}
+		vmhub = xcc_id;
 	}
+	hub = &adev->vmhub[vmhub];
+
+	if (retry_fault) {
+		if (adev->irq.retry_cam_enabled) {
+			/* Delegate it to a different ring if the hardware hasn't
+			 * already done it.
+			 */
+			if (entry->ih == &adev->irq.ih) {
+				amdgpu_irq_delegate(adev, entry, 8);
+				return 1;
+			}
+
+			cam_index = entry->src_data[2] & 0x3ff;
+
+			ret = amdgpu_vm_handle_fault(adev, entry->pasid, entry->vmid, node_id,
+						     addr, entry->timestamp, write_fault);
+			WDOORBELL32(adev->irq.retry_cam_doorbell_index, cam_index);
+			if (ret)
+				return 1;
+		} else {
+			/* Process it onyl if it's the first fault for this address */
+			if (entry->ih != &adev->irq.ih_soft &&
+			    amdgpu_gmc_filter_faults(adev, entry->ih, addr, entry->pasid,
+					     entry->timestamp))
+				return 1;
+
+			/* Delegate it to a different ring if the hardware hasn't
+			 * already done it.
+			 */
+			if (entry->ih == &adev->irq.ih) {
+				amdgpu_irq_delegate(adev, entry, 8);
+				return 1;
+			}
+
+			/* Try to handle the recoverable page faults by filling page
+			 * tables
+			 */
+			if (amdgpu_vm_handle_fault(adev, entry->pasid, entry->vmid, node_id,
+						   addr, entry->timestamp, write_fault))
+				return 1;
+		}
+	}
+
+	if (kgd2kfd_vmfault_fast_path(adev, entry, retry_fault))
+		return 1;
 
-	if (printk_ratelimit()) {
-		dev_err(adev->dev,
-			"[%s] VMC page fault (src_id:%u ring:%u vm_id:%u pas_id:%u)\n",
-			entry->vm_id_src ? "mmhub" : "gfxhub",
-			entry->src_id, entry->ring_id, entry->vm_id,
-			entry->pas_id);
-		dev_err(adev->dev, "  at page 0x%016llx from %d\n",
-			addr, entry->client_id);
-		if (!amdgpu_sriov_vf(adev))
-			dev_err(adev->dev,
-				"VM_L2_PROTECTION_FAULT_STATUS:0x%08X\n",
-				status);
+	if (!printk_ratelimit())
+		return 0;
+
+	dev_err(adev->dev,
+		"[%s] %s page fault (src_id:%u ring:%u vmid:%u pasid:%u)\n", hub_name,
+		retry_fault ? "retry" : "no-retry",
+		entry->src_id, entry->ring_id, entry->vmid, entry->pasid);
+
+	task_info = amdgpu_vm_get_task_info_pasid(adev, entry->pasid);
+	if (task_info) {
+		amdgpu_vm_print_task_info(adev, task_info);
+		amdgpu_vm_put_task_info(task_info);
 	}
 
+	dev_err(adev->dev, "  in page starting at address 0x%016llx from IH client 0x%x (%s)\n",
+		addr, entry->client_id,
+		soc15_ih_clientid_name[entry->client_id]);
+
+	if (amdgpu_is_multi_aid(adev))
+		dev_err(adev->dev, "  cookie node_id %d fault from die %s%d%s\n",
+			node_id, node_id % 4 == 3 ? "RSV" : "AID", node_id / 4,
+			node_id % 4 == 1 ? ".XCD0" : node_id % 4 == 2 ? ".XCD1" : "");
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	/*
+	 * Issue a dummy read to wait for the status register to
+	 * be updated to avoid reading an incorrect value due to
+	 * the new fast GRBM interface.
+	 */
+	if ((entry->vmid_src == AMDGPU_GFXHUB(0)) &&
+	    (amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(9, 4, 2)))
+		RREG32(hub->vm_l2_pro_fault_status);
+
+	status = RREG32(hub->vm_l2_pro_fault_status);
+	cid = REG_GET_FIELD(status, VM_L2_PROTECTION_FAULT_STATUS, CID);
+	rw = REG_GET_FIELD(status, VM_L2_PROTECTION_FAULT_STATUS, RW);
+	fed = REG_GET_FIELD(status, VM_L2_PROTECTION_FAULT_STATUS, FED);
+
+	/* for fed error, kfd will handle it, return directly */
+	if (fed && amdgpu_ras_is_poison_mode_supported(adev) &&
+	    (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(9, 4, 2)))
+		return 0;
+
+	/* Only print L2 fault status if the status register could be read and
+	 * contains useful information
+	 */
+	if (!status)
+		return 0;
+
+	if (!amdgpu_sriov_vf(adev))
+		WREG32_P(hub->vm_l2_pro_fault_cntl, 1, ~1);
+
+	amdgpu_vm_update_fault_cache(adev, entry->pasid, addr, status, vmhub);
+
+	dev_err(adev->dev,
+		"VM_L2_PROTECTION_FAULT_STATUS:0x%08X\n",
+		status);
+	if (entry->vmid_src == AMDGPU_GFXHUB(0)) {
+		dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n",
+			cid >= ARRAY_SIZE(gfxhub_client_ids) ? "unknown" :
+			gfxhub_client_ids[cid],
+			cid);
+	} else {
+		switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) {
+		case IP_VERSION(9, 0, 0):
+			mmhub_cid = mmhub_client_ids_vega10[cid][rw];
+			break;
+		case IP_VERSION(9, 3, 0):
+			mmhub_cid = mmhub_client_ids_vega12[cid][rw];
+			break;
+		case IP_VERSION(9, 4, 0):
+			mmhub_cid = mmhub_client_ids_vega20[cid][rw];
+			break;
+		case IP_VERSION(9, 4, 1):
+			mmhub_cid = mmhub_client_ids_arcturus[cid][rw];
+			break;
+		case IP_VERSION(9, 1, 0):
+		case IP_VERSION(9, 2, 0):
+			mmhub_cid = mmhub_client_ids_raven[cid][rw];
+			break;
+		case IP_VERSION(1, 5, 0):
+		case IP_VERSION(2, 4, 0):
+			mmhub_cid = mmhub_client_ids_renoir[cid][rw];
+			break;
+		case IP_VERSION(1, 8, 0):
+		case IP_VERSION(9, 4, 2):
+			mmhub_cid = mmhub_client_ids_aldebaran[cid][rw];
+			break;
+		default:
+			mmhub_cid = NULL;
+			break;
+		}
+		dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n",
+			mmhub_cid ? mmhub_cid : "unknown", cid);
+	}
+	dev_err(adev->dev, "\t MORE_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		VM_L2_PROTECTION_FAULT_STATUS, MORE_FAULTS));
+	dev_err(adev->dev, "\t WALKER_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		VM_L2_PROTECTION_FAULT_STATUS, WALKER_ERROR));
+	dev_err(adev->dev, "\t PERMISSION_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		VM_L2_PROTECTION_FAULT_STATUS, PERMISSION_FAULTS));
+	dev_err(adev->dev, "\t MAPPING_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		VM_L2_PROTECTION_FAULT_STATUS, MAPPING_ERROR));
+	dev_err(adev->dev, "\t RW: 0x%x\n", rw);
 	return 0;
 }
 
@@ -171,20 +742,33 @@ static const struct amdgpu_irq_src_funcs gmc_v9_0_irq_funcs = {
 	.process = gmc_v9_0_process_interrupt,
 };
 
+
+static const struct amdgpu_irq_src_funcs gmc_v9_0_ecc_funcs = {
+	.set = gmc_v9_0_ecc_interrupt_state,
+	.process = amdgpu_umc_process_ecc_irq,
+};
+
 static void gmc_v9_0_set_irq_funcs(struct amdgpu_device *adev)
 {
-	adev->mc.vm_fault.num_types = 1;
-	adev->mc.vm_fault.funcs = &gmc_v9_0_irq_funcs;
+	adev->gmc.vm_fault.num_types = 1;
+	adev->gmc.vm_fault.funcs = &gmc_v9_0_irq_funcs;
+
+	if (!amdgpu_sriov_vf(adev) &&
+	    !adev->gmc.xgmi.connected_to_cpu &&
+	    !adev->gmc.is_app_apu) {
+		adev->gmc.ecc_irq.num_types = 1;
+		adev->gmc.ecc_irq.funcs = &gmc_v9_0_ecc_funcs;
+	}
 }
 
-static uint32_t gmc_v9_0_get_invalidate_req(unsigned int vm_id)
+static uint32_t gmc_v9_0_get_invalidate_req(unsigned int vmid,
+					uint32_t flush_type)
 {
 	u32 req = 0;
 
-	/* invalidate using legacy mode on vm_id*/
 	req = REG_SET_FIELD(req, VM_INVALIDATE_ENG0_REQ,
-			    PER_VMID_INVALIDATE_REQ, 1 << vm_id);
-	req = REG_SET_FIELD(req, VM_INVALIDATE_ENG0_REQ, FLUSH_TYPE, 0);
+			    PER_VMID_INVALIDATE_REQ, 1 << vmid);
+	req = REG_SET_FIELD(req, VM_INVALIDATE_ENG0_REQ, FLUSH_TYPE, flush_type);
 	req = REG_SET_FIELD(req, VM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PTES, 1);
 	req = REG_SET_FIELD(req, VM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE0, 1);
 	req = REG_SET_FIELD(req, VM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE1, 1);
@@ -196,6 +780,39 @@ static uint32_t gmc_v9_0_get_invalidate_req(unsigned int vm_id)
 	return req;
 }
 
+/**
+ * gmc_v9_0_use_invalidate_semaphore - judge whether to use semaphore
+ *
+ * @adev: amdgpu_device pointer
+ * @vmhub: vmhub type
+ *
+ */
+static bool gmc_v9_0_use_invalidate_semaphore(struct amdgpu_device *adev,
+				       uint32_t vmhub)
+{
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2) ||
+	    amdgpu_is_multi_aid(adev))
+		return false;
+
+	return ((vmhub == AMDGPU_MMHUB0(0) ||
+		 vmhub == AMDGPU_MMHUB1(0)) &&
+		(!amdgpu_sriov_vf(adev)) &&
+		(!(!(adev->apu_flags & AMD_APU_IS_RAVEN2) &&
+		   (adev->apu_flags & AMD_APU_IS_PICASSO))));
+}
+
+static bool gmc_v9_0_get_atc_vmid_pasid_mapping_info(struct amdgpu_device *adev,
+					uint8_t vmid, uint16_t *p_pasid)
+{
+	uint32_t value;
+
+	value = RREG32(SOC15_REG_OFFSET(ATHUB, 0, mmATC_VMID0_PASID_MAPPING)
+		     + vmid);
+	*p_pasid = value & ATC_VMID0_PASID_MAPPING__PASID_MASK;
+
+	return !!(value & ATC_VMID0_PASID_MAPPING__VALID_MASK);
+}
+
 /*
  * GART
  * VMID 0 is the physical GPU addresses as used by the kernel.
@@ -204,230 +821,868 @@ static uint32_t gmc_v9_0_get_invalidate_req(unsigned int vm_id)
  */
 
 /**
- * gmc_v9_0_gart_flush_gpu_tlb - gart tlb flush callback
+ * gmc_v9_0_flush_gpu_tlb - tlb flush with certain type
  *
  * @adev: amdgpu_device pointer
  * @vmid: vm instance to flush
+ * @vmhub: which hub to flush
+ * @flush_type: the flush type
  *
- * Flush the TLB for the requested page table.
+ * Flush the TLB for the requested page table using certain type.
  */
-static void gmc_v9_0_gart_flush_gpu_tlb(struct amdgpu_device *adev,
-					uint32_t vmid)
+static void gmc_v9_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid,
+					uint32_t vmhub, uint32_t flush_type)
 {
-	/* Use register 17 for GART */
-	const unsigned eng = 17;
-	unsigned i, j;
+	bool use_semaphore = gmc_v9_0_use_invalidate_semaphore(adev, vmhub);
+	u32 j, inv_req, tmp, sem, req, ack, inst;
+	const unsigned int eng = 17;
+	struct amdgpu_vmhub *hub;
 
-	/* flush hdp cache */
-	if (adev->flags & AMD_IS_APU)
-		nbio_v7_0_hdp_flush(adev);
-	else
-		nbio_v6_1_hdp_flush(adev);
+	BUG_ON(vmhub >= AMDGPU_MAX_VMHUBS);
 
-	spin_lock(&adev->mc.invalidate_lock);
+	hub = &adev->vmhub[vmhub];
+	inv_req = gmc_v9_0_get_invalidate_req(vmid, flush_type);
+	sem = hub->vm_inv_eng0_sem + hub->eng_distance * eng;
+	req = hub->vm_inv_eng0_req + hub->eng_distance * eng;
+	ack = hub->vm_inv_eng0_ack + hub->eng_distance * eng;
 
-	for (i = 0; i < AMDGPU_MAX_VMHUBS; ++i) {
-		struct amdgpu_vmhub *hub = &adev->vmhub[i];
-		u32 tmp = gmc_v9_0_get_invalidate_req(vmid);
+	if (vmhub >= AMDGPU_MMHUB0(0))
+		inst = 0;
+	else
+		inst = vmhub;
 
-		WREG32_NO_KIQ(hub->vm_inv_eng0_req + eng, tmp);
+	/* This is necessary for SRIOV as well as for GFXOFF to function
+	 * properly under bare metal
+	 */
+	if (adev->gfx.kiq[inst].ring.sched.ready &&
+	    (amdgpu_sriov_runtime(adev) || !amdgpu_sriov_vf(adev))) {
+		uint32_t req = hub->vm_inv_eng0_req + hub->eng_distance * eng;
+		uint32_t ack = hub->vm_inv_eng0_ack + hub->eng_distance * eng;
+
+		amdgpu_gmc_fw_reg_write_reg_wait(adev, req, ack, inv_req,
+						 1 << vmid, inst);
+		return;
+	}
 
-		/* Busy wait for ACK.*/
-		for (j = 0; j < 100; j++) {
-			tmp = RREG32_NO_KIQ(hub->vm_inv_eng0_ack + eng);
-			tmp &= 1 << vmid;
-			if (tmp)
-				break;
-			cpu_relax();
-		}
-		if (j < 100)
-			continue;
+	/* This path is needed before KIQ/MES/GFXOFF are set up */
+	spin_lock(&adev->gmc.invalidate_lock);
 
-		/* Wait for ACK with a delay.*/
+	/*
+	 * It may lose gpuvm invalidate acknowldege state across power-gating
+	 * off cycle, add semaphore acquire before invalidation and semaphore
+	 * release after invalidation to avoid entering power gated state
+	 * to WA the Issue
+	 */
+
+	/* TODO: It needs to continue working on debugging with semaphore for GFXHUB as well. */
+	if (use_semaphore) {
 		for (j = 0; j < adev->usec_timeout; j++) {
-			tmp = RREG32_NO_KIQ(hub->vm_inv_eng0_ack + eng);
-			tmp &= 1 << vmid;
-			if (tmp)
+			/* a read return value of 1 means semaphore acquire */
+			if (vmhub >= AMDGPU_MMHUB0(0))
+				tmp = RREG32_SOC15_IP_NO_KIQ(MMHUB, sem, GET_INST(GC, inst));
+			else
+				tmp = RREG32_SOC15_IP_NO_KIQ(GC, sem, GET_INST(GC, inst));
+			if (tmp & 0x1)
 				break;
 			udelay(1);
 		}
-		if (j < adev->usec_timeout)
-			continue;
 
-		DRM_ERROR("Timeout waiting for VM flush ACK!\n");
+		if (j >= adev->usec_timeout)
+			DRM_ERROR("Timeout waiting for sem acquire in VM flush!\n");
+	}
+
+	if (vmhub >= AMDGPU_MMHUB0(0))
+		WREG32_SOC15_IP_NO_KIQ(MMHUB, req, inv_req, GET_INST(GC, inst));
+	else
+		WREG32_SOC15_IP_NO_KIQ(GC, req, inv_req, GET_INST(GC, inst));
+
+	/*
+	 * Issue a dummy read to wait for the ACK register to
+	 * be cleared to avoid a false ACK due to the new fast
+	 * GRBM interface.
+	 */
+	if ((vmhub == AMDGPU_GFXHUB(0)) &&
+	    (amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(9, 4, 2)))
+		RREG32_NO_KIQ(req);
+
+	for (j = 0; j < adev->usec_timeout; j++) {
+		if (vmhub >= AMDGPU_MMHUB0(0))
+			tmp = RREG32_SOC15_IP_NO_KIQ(MMHUB, ack, GET_INST(GC, inst));
+		else
+			tmp = RREG32_SOC15_IP_NO_KIQ(GC, ack, GET_INST(GC, inst));
+		if (tmp & (1 << vmid))
+			break;
+		udelay(1);
+	}
+
+	/* TODO: It needs to continue working on debugging with semaphore for GFXHUB as well. */
+	if (use_semaphore) {
+		/*
+		 * add semaphore release after invalidation,
+		 * write with 0 means semaphore release
+		 */
+		if (vmhub >= AMDGPU_MMHUB0(0))
+			WREG32_SOC15_IP_NO_KIQ(MMHUB, sem, 0, GET_INST(GC, inst));
+		else
+			WREG32_SOC15_IP_NO_KIQ(GC, sem, 0, GET_INST(GC, inst));
 	}
 
-	spin_unlock(&adev->mc.invalidate_lock);
+	spin_unlock(&adev->gmc.invalidate_lock);
+
+	if (j < adev->usec_timeout)
+		return;
+
+	DRM_ERROR("Timeout waiting for VM flush ACK!\n");
 }
 
 /**
- * gmc_v9_0_gart_set_pte_pde - update the page tables using MMIO
+ * gmc_v9_0_flush_gpu_tlb_pasid - tlb flush via pasid
  *
  * @adev: amdgpu_device pointer
- * @cpu_pt_addr: cpu address of the page table
- * @gpu_page_idx: entry in the page table to update
- * @addr: dst addr to write into pte/pde
- * @flags: access flags
+ * @pasid: pasid to be flush
+ * @flush_type: the flush type
+ * @all_hub: flush all hubs
+ * @inst: is used to select which instance of KIQ to use for the invalidation
  *
- * Update the page tables using the CPU.
+ * Flush the TLB for the requested pasid.
  */
-static int gmc_v9_0_gart_set_pte_pde(struct amdgpu_device *adev,
-					void *cpu_pt_addr,
-					uint32_t gpu_page_idx,
-					uint64_t addr,
-					uint64_t flags)
+static void gmc_v9_0_flush_gpu_tlb_pasid(struct amdgpu_device *adev,
+					 uint16_t pasid, uint32_t flush_type,
+					 bool all_hub, uint32_t inst)
+{
+	uint16_t queried;
+	int i, vmid;
+
+	for (vmid = 1; vmid < 16; vmid++) {
+		bool valid;
+
+		valid = gmc_v9_0_get_atc_vmid_pasid_mapping_info(adev, vmid,
+								 &queried);
+		if (!valid || queried != pasid)
+			continue;
+
+		if (all_hub) {
+			for_each_set_bit(i, adev->vmhubs_mask,
+					 AMDGPU_MAX_VMHUBS)
+				gmc_v9_0_flush_gpu_tlb(adev, vmid, i,
+						       flush_type);
+		} else {
+			gmc_v9_0_flush_gpu_tlb(adev, vmid,
+					       AMDGPU_GFXHUB(0),
+					       flush_type);
+		}
+	}
+}
+
+static uint64_t gmc_v9_0_emit_flush_gpu_tlb(struct amdgpu_ring *ring,
+					    unsigned int vmid, uint64_t pd_addr)
 {
-	void __iomem *ptr = (void *)cpu_pt_addr;
-	uint64_t value;
+	bool use_semaphore = gmc_v9_0_use_invalidate_semaphore(ring->adev, ring->vm_hub);
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_vmhub *hub = &adev->vmhub[ring->vm_hub];
+	uint32_t req = gmc_v9_0_get_invalidate_req(vmid, 0);
+	unsigned int eng = ring->vm_inv_eng;
 
 	/*
-	 * PTE format on VEGA 10:
-	 * 63:59 reserved
-	 * 58:57 mtype
-	 * 56 F
-	 * 55 L
-	 * 54 P
-	 * 53 SW
-	 * 52 T
-	 * 50:48 reserved
-	 * 47:12 4k physical page base address
-	 * 11:7 fragment
-	 * 6 write
-	 * 5 read
-	 * 4 exe
-	 * 3 Z
-	 * 2 snooped
-	 * 1 system
-	 * 0 valid
-	 *
-	 * PDE format on VEGA 10:
-	 * 63:59 block fragment size
-	 * 58:55 reserved
-	 * 54 P
-	 * 53:48 reserved
-	 * 47:6 physical base address of PD or PTE
-	 * 5:3 reserved
-	 * 2 C
-	 * 1 system
-	 * 0 valid
+	 * It may lose gpuvm invalidate acknowldege state across power-gating
+	 * off cycle, add semaphore acquire before invalidation and semaphore
+	 * release after invalidation to avoid entering power gated state
+	 * to WA the Issue
 	 */
 
-	/*
-	 * The following is for PTE only. GART does not have PDEs.
-	*/
-	value = addr & 0x0000FFFFFFFFF000ULL;
-	value |= flags;
-	writeq(value, ptr + (gpu_page_idx * 8));
-	return 0;
+	/* TODO: It needs to continue working on debugging with semaphore for GFXHUB as well. */
+	if (use_semaphore)
+		/* a read return value of 1 means semaphore acuqire */
+		amdgpu_ring_emit_reg_wait(ring,
+					  hub->vm_inv_eng0_sem +
+					  hub->eng_distance * eng, 0x1, 0x1);
+
+	amdgpu_ring_emit_wreg(ring, hub->ctx0_ptb_addr_lo32 +
+			      (hub->ctx_addr_distance * vmid),
+			      lower_32_bits(pd_addr));
+
+	amdgpu_ring_emit_wreg(ring, hub->ctx0_ptb_addr_hi32 +
+			      (hub->ctx_addr_distance * vmid),
+			      upper_32_bits(pd_addr));
+
+	amdgpu_ring_emit_reg_write_reg_wait(ring, hub->vm_inv_eng0_req +
+					    hub->eng_distance * eng,
+					    hub->vm_inv_eng0_ack +
+					    hub->eng_distance * eng,
+					    req, 1 << vmid);
+
+	/* TODO: It needs to continue working on debugging with semaphore for GFXHUB as well. */
+	if (use_semaphore)
+		/*
+		 * add semaphore release after invalidation,
+		 * write with 0 means semaphore release
+		 */
+		amdgpu_ring_emit_wreg(ring, hub->vm_inv_eng0_sem +
+				      hub->eng_distance * eng, 0);
+
+	return pd_addr;
+}
+
+static void gmc_v9_0_emit_pasid_mapping(struct amdgpu_ring *ring, unsigned int vmid,
+					unsigned int pasid)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t reg;
+
+	/* Do nothing because there's no lut register for mmhub1. */
+	if (ring->vm_hub == AMDGPU_MMHUB1(0))
+		return;
+
+	if (ring->vm_hub == AMDGPU_GFXHUB(0))
+		reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_VMID_0_LUT) + vmid;
+	else
+		reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_VMID_0_LUT_MM) + vmid;
+
+	amdgpu_ring_emit_wreg(ring, reg, pasid);
 }
 
-static uint64_t gmc_v9_0_get_vm_pte_flags(struct amdgpu_device *adev,
-						uint32_t flags)
+/*
+ * PTE format on VEGA 10:
+ * 63:59 reserved
+ * 58:57 mtype
+ * 56 F
+ * 55 L
+ * 54 P
+ * 53 SW
+ * 52 T
+ * 50:48 reserved
+ * 47:12 4k physical page base address
+ * 11:7 fragment
+ * 6 write
+ * 5 read
+ * 4 exe
+ * 3 Z
+ * 2 snooped
+ * 1 system
+ * 0 valid
+ *
+ * PDE format on VEGA 10:
+ * 63:59 block fragment size
+ * 58:55 reserved
+ * 54 P
+ * 53:48 reserved
+ * 47:6 physical base address of PD or PTE
+ * 5:3 reserved
+ * 2 C
+ * 1 system
+ * 0 valid
+ */
 
+static void gmc_v9_0_get_vm_pde(struct amdgpu_device *adev, int level,
+				uint64_t *addr, uint64_t *flags)
 {
-	uint64_t pte_flag = 0;
+	if (!(*flags & AMDGPU_PDE_PTE) && !(*flags & AMDGPU_PTE_SYSTEM))
+		*addr = amdgpu_gmc_vram_mc2pa(adev, *addr);
+	BUG_ON(*addr & 0xFFFF00000000003FULL);
+
+	if (!adev->gmc.translate_further)
+		return;
+
+	if (level == AMDGPU_VM_PDB1) {
+		/* Set the block fragment size */
+		if (!(*flags & AMDGPU_PDE_PTE))
+			*flags |= AMDGPU_PDE_BFS(0x9);
+
+	} else if (level == AMDGPU_VM_PDB0) {
+		if (*flags & AMDGPU_PDE_PTE) {
+			*flags &= ~AMDGPU_PDE_PTE;
+			if (!(*flags & AMDGPU_PTE_VALID))
+				*addr |= 1 << PAGE_SHIFT;
+		} else {
+			*flags |= AMDGPU_PTE_TF;
+		}
+	}
+}
 
-	if (flags & AMDGPU_VM_PAGE_EXECUTABLE)
-		pte_flag |= AMDGPU_PTE_EXECUTABLE;
-	if (flags & AMDGPU_VM_PAGE_READABLE)
-		pte_flag |= AMDGPU_PTE_READABLE;
-	if (flags & AMDGPU_VM_PAGE_WRITEABLE)
-		pte_flag |= AMDGPU_PTE_WRITEABLE;
+static void gmc_v9_0_get_coherence_flags(struct amdgpu_device *adev,
+					 struct amdgpu_vm *vm,
+					 struct amdgpu_bo *bo,
+					 uint32_t vm_flags,
+					 uint64_t *flags)
+{
+	struct amdgpu_device *bo_adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	bool is_vram = bo->tbo.resource &&
+		bo->tbo.resource->mem_type == TTM_PL_VRAM;
+	bool coherent = bo->flags & (AMDGPU_GEM_CREATE_COHERENT |
+				     AMDGPU_GEM_CREATE_EXT_COHERENT);
+	bool ext_coherent = bo->flags & AMDGPU_GEM_CREATE_EXT_COHERENT;
+	bool uncached = bo->flags & AMDGPU_GEM_CREATE_UNCACHED;
+	unsigned int mtype_local, mtype;
+	uint32_t gc_ip_version = amdgpu_ip_version(adev, GC_HWIP, 0);
+	bool snoop = false;
+	bool is_local;
+
+	dma_resv_assert_held(bo->tbo.base.resv);
+
+	switch (gc_ip_version) {
+	case IP_VERSION(9, 4, 1):
+	case IP_VERSION(9, 4, 2):
+		if (is_vram) {
+			if (bo_adev == adev) {
+				if (uncached)
+					mtype = MTYPE_UC;
+				else if (coherent)
+					mtype = MTYPE_CC;
+				else
+					mtype = MTYPE_RW;
+				/* FIXME: is this still needed? Or does
+				 * amdgpu_ttm_tt_pde_flags already handle this?
+				 */
+				if (gc_ip_version == IP_VERSION(9, 4, 2) &&
+				    adev->gmc.xgmi.connected_to_cpu)
+					snoop = true;
+			} else {
+				if (uncached || coherent)
+					mtype = MTYPE_UC;
+				else
+					mtype = MTYPE_NC;
+				if (amdgpu_xgmi_same_hive(adev, bo_adev))
+					snoop = true;
+			}
+		} else {
+			if (uncached || coherent)
+				mtype = MTYPE_UC;
+			else
+				mtype = MTYPE_NC;
+			/* FIXME: is this still needed? Or does
+			 * amdgpu_ttm_tt_pde_flags already handle this?
+			 */
+			snoop = true;
+		}
+		break;
+	case IP_VERSION(9, 4, 3):
+	case IP_VERSION(9, 4, 4):
+	case IP_VERSION(9, 5, 0):
+		/* Only local VRAM BOs or system memory on non-NUMA APUs
+		 * can be assumed to be local in their entirety. Choose
+		 * MTYPE_NC as safe fallback for all system memory BOs on
+		 * NUMA systems. Their MTYPE can be overridden per-page in
+		 * gmc_v9_0_override_vm_pte_flags.
+		 */
+		mtype_local = MTYPE_RW;
+		if (amdgpu_mtype_local == 1) {
+			DRM_INFO_ONCE("Using MTYPE_NC for local memory\n");
+			mtype_local = MTYPE_NC;
+		} else if (amdgpu_mtype_local == 2) {
+			DRM_INFO_ONCE("Using MTYPE_CC for local memory\n");
+			mtype_local = MTYPE_CC;
+		} else {
+			DRM_INFO_ONCE("Using MTYPE_RW for local memory\n");
+		}
+		is_local = (!is_vram && (adev->flags & AMD_IS_APU) &&
+			    num_possible_nodes() <= 1) ||
+			   (is_vram && adev == bo_adev &&
+			    KFD_XCP_MEM_ID(adev, bo->xcp_id) == vm->mem_id);
+		snoop = true;
+		if (uncached) {
+			mtype = MTYPE_UC;
+		} else if (ext_coherent) {
+			mtype = is_local ? MTYPE_CC : MTYPE_UC;
+		} else if (adev->flags & AMD_IS_APU) {
+			mtype = is_local ? mtype_local : MTYPE_NC;
+		} else {
+			/* dGPU */
+			if (is_local)
+				mtype = mtype_local;
+			else if (gc_ip_version < IP_VERSION(9, 5, 0) && !is_vram)
+				mtype = MTYPE_UC;
+			else
+				mtype = MTYPE_NC;
+		}
 
-	switch (flags & AMDGPU_VM_MTYPE_MASK) {
-	case AMDGPU_VM_MTYPE_DEFAULT:
-		pte_flag |= AMDGPU_PTE_MTYPE(MTYPE_NC);
 		break;
+	default:
+		if (uncached || coherent)
+			mtype = MTYPE_UC;
+		else
+			mtype = MTYPE_NC;
+
+		/* FIXME: is this still needed? Or does
+		 * amdgpu_ttm_tt_pde_flags already handle this?
+		 */
+		if (!is_vram)
+			snoop = true;
+	}
+
+	if (mtype != MTYPE_NC)
+		*flags = AMDGPU_PTE_MTYPE_VG10(*flags, mtype);
+
+	*flags |= snoop ? AMDGPU_PTE_SNOOPED : 0;
+}
+
+static void gmc_v9_0_get_vm_pte(struct amdgpu_device *adev,
+				struct amdgpu_vm *vm,
+				struct amdgpu_bo *bo,
+				uint32_t vm_flags,
+				uint64_t *flags)
+{
+	if (vm_flags & AMDGPU_VM_PAGE_EXECUTABLE)
+		*flags |= AMDGPU_PTE_EXECUTABLE;
+	else
+		*flags &= ~AMDGPU_PTE_EXECUTABLE;
+
+	switch (vm_flags & AMDGPU_VM_MTYPE_MASK) {
+	case AMDGPU_VM_MTYPE_DEFAULT:
 	case AMDGPU_VM_MTYPE_NC:
-		pte_flag |= AMDGPU_PTE_MTYPE(MTYPE_NC);
+	default:
+		*flags = AMDGPU_PTE_MTYPE_VG10(*flags, MTYPE_NC);
 		break;
 	case AMDGPU_VM_MTYPE_WC:
-		pte_flag |= AMDGPU_PTE_MTYPE(MTYPE_WC);
+		*flags |= AMDGPU_PTE_MTYPE_VG10(*flags, MTYPE_WC);
+		break;
+	case AMDGPU_VM_MTYPE_RW:
+		*flags |= AMDGPU_PTE_MTYPE_VG10(*flags, MTYPE_RW);
 		break;
 	case AMDGPU_VM_MTYPE_CC:
-		pte_flag |= AMDGPU_PTE_MTYPE(MTYPE_CC);
+		*flags |= AMDGPU_PTE_MTYPE_VG10(*flags, MTYPE_CC);
 		break;
 	case AMDGPU_VM_MTYPE_UC:
-		pte_flag |= AMDGPU_PTE_MTYPE(MTYPE_UC);
+		*flags |= AMDGPU_PTE_MTYPE_VG10(*flags, MTYPE_UC);
+		break;
+	}
+
+	if (vm_flags & AMDGPU_VM_PAGE_PRT) {
+		*flags |= AMDGPU_PTE_PRT;
+		*flags &= ~AMDGPU_PTE_VALID;
+	}
+
+	if ((*flags & AMDGPU_PTE_VALID) && bo)
+		gmc_v9_0_get_coherence_flags(adev, vm, bo, vm_flags, flags);
+}
+
+static void gmc_v9_0_override_vm_pte_flags(struct amdgpu_device *adev,
+					   struct amdgpu_vm *vm,
+					   uint64_t addr, uint64_t *flags)
+{
+	int local_node, nid;
+
+	/* Only GFX 9.4.3 APUs associate GPUs with NUMA nodes. Local system
+	 * memory can use more efficient MTYPEs.
+	 */
+	if (!(adev->flags & AMD_IS_APU) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 3))
+		return;
+
+	/* Only direct-mapped memory allows us to determine the NUMA node from
+	 * the DMA address.
+	 */
+	if (!adev->ram_is_direct_mapped) {
+		dev_dbg_ratelimited(adev->dev, "RAM is not direct mapped\n");
+		return;
+	}
+
+	/* MTYPE_NC is the same default and can be overridden.
+	 * MTYPE_UC will be present if the memory is extended-coherent
+	 * and can also be overridden.
+	 */
+	if ((*flags & AMDGPU_PTE_MTYPE_VG10_MASK) !=
+	    AMDGPU_PTE_MTYPE_VG10(0ULL, MTYPE_NC) &&
+	    (*flags & AMDGPU_PTE_MTYPE_VG10_MASK) !=
+	    AMDGPU_PTE_MTYPE_VG10(0ULL, MTYPE_UC)) {
+		dev_dbg_ratelimited(adev->dev, "MTYPE is not NC or UC\n");
+		return;
+	}
+
+	/* FIXME: Only supported on native mode for now. For carve-out, the
+	 * NUMA affinity of the GPU/VM needs to come from the PCI info because
+	 * memory partitions are not associated with different NUMA nodes.
+	 */
+	if (adev->gmc.is_app_apu && vm->mem_id >= 0) {
+		local_node = adev->gmc.mem_partitions[vm->mem_id].numa.node;
+	} else {
+		dev_dbg_ratelimited(adev->dev, "Only native mode APU is supported.\n");
+		return;
+	}
+
+	/* Only handle real RAM. Mappings of PCIe resources don't have struct
+	 * page or NUMA nodes.
+	 */
+	if (!page_is_ram(addr >> PAGE_SHIFT)) {
+		dev_dbg_ratelimited(adev->dev, "Page is not RAM.\n");
+		return;
+	}
+	nid = pfn_to_nid(addr >> PAGE_SHIFT);
+	dev_dbg_ratelimited(adev->dev, "vm->mem_id=%d, local_node=%d, nid=%d\n",
+			    vm->mem_id, local_node, nid);
+	if (nid == local_node) {
+		uint64_t old_flags = *flags;
+		if ((*flags & AMDGPU_PTE_MTYPE_VG10_MASK) ==
+			AMDGPU_PTE_MTYPE_VG10(0ULL, MTYPE_NC)) {
+			unsigned int mtype_local = MTYPE_RW;
+
+			if (amdgpu_mtype_local == 1)
+				mtype_local = MTYPE_NC;
+			else if (amdgpu_mtype_local == 2)
+				mtype_local = MTYPE_CC;
+
+			*flags = AMDGPU_PTE_MTYPE_VG10(*flags, mtype_local);
+		} else {
+			/* MTYPE_UC case */
+			*flags = AMDGPU_PTE_MTYPE_VG10(*flags, MTYPE_CC);
+		}
+
+		dev_dbg_ratelimited(adev->dev, "flags updated from %llx to %llx\n",
+				    old_flags, *flags);
+	}
+}
+
+static unsigned int gmc_v9_0_get_vbios_fb_size(struct amdgpu_device *adev)
+{
+	u32 d1vga_control = RREG32_SOC15(DCE, 0, mmD1VGA_CONTROL);
+	unsigned int size;
+
+	/* TODO move to DC so GMC doesn't need to hard-code DCN registers */
+
+	if (REG_GET_FIELD(d1vga_control, D1VGA_CONTROL, D1VGA_MODE_ENABLE)) {
+		size = AMDGPU_VBIOS_VGA_ALLOCATION;
+	} else {
+		u32 viewport;
+
+		switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) {
+		case IP_VERSION(1, 0, 0):
+		case IP_VERSION(1, 0, 1):
+			viewport = RREG32_SOC15(DCE, 0, mmHUBP0_DCSURF_PRI_VIEWPORT_DIMENSION);
+			size = (REG_GET_FIELD(viewport,
+					      HUBP0_DCSURF_PRI_VIEWPORT_DIMENSION, PRI_VIEWPORT_HEIGHT) *
+				REG_GET_FIELD(viewport,
+					      HUBP0_DCSURF_PRI_VIEWPORT_DIMENSION, PRI_VIEWPORT_WIDTH) *
+				4);
+			break;
+		case IP_VERSION(2, 1, 0):
+			viewport = RREG32_SOC15(DCE, 0, mmHUBP0_DCSURF_PRI_VIEWPORT_DIMENSION_DCN2);
+			size = (REG_GET_FIELD(viewport,
+					      HUBP0_DCSURF_PRI_VIEWPORT_DIMENSION, PRI_VIEWPORT_HEIGHT) *
+				REG_GET_FIELD(viewport,
+					      HUBP0_DCSURF_PRI_VIEWPORT_DIMENSION, PRI_VIEWPORT_WIDTH) *
+				4);
+			break;
+		default:
+			viewport = RREG32_SOC15(DCE, 0, mmSCL0_VIEWPORT_SIZE);
+			size = (REG_GET_FIELD(viewport, SCL0_VIEWPORT_SIZE, VIEWPORT_HEIGHT) *
+				REG_GET_FIELD(viewport, SCL0_VIEWPORT_SIZE, VIEWPORT_WIDTH) *
+				4);
+			break;
+		}
+	}
+
+	return size;
+}
+
+static bool gmc_v9_0_need_reset_on_init(struct amdgpu_device *adev)
+{
+	if (adev->nbio.funcs && adev->nbio.funcs->is_nps_switch_requested &&
+	    adev->nbio.funcs->is_nps_switch_requested(adev)) {
+		adev->gmc.reset_flags |= AMDGPU_GMC_INIT_RESET_NPS;
+		return true;
+	}
+
+	return false;
+}
+
+static const struct amdgpu_gmc_funcs gmc_v9_0_gmc_funcs = {
+	.flush_gpu_tlb = gmc_v9_0_flush_gpu_tlb,
+	.flush_gpu_tlb_pasid = gmc_v9_0_flush_gpu_tlb_pasid,
+	.emit_flush_gpu_tlb = gmc_v9_0_emit_flush_gpu_tlb,
+	.emit_pasid_mapping = gmc_v9_0_emit_pasid_mapping,
+	.get_vm_pde = gmc_v9_0_get_vm_pde,
+	.get_vm_pte = gmc_v9_0_get_vm_pte,
+	.override_vm_pte_flags = gmc_v9_0_override_vm_pte_flags,
+	.get_vbios_fb_size = gmc_v9_0_get_vbios_fb_size,
+	.query_mem_partition_mode = &amdgpu_gmc_query_memory_partition,
+	.request_mem_partition_mode = &amdgpu_gmc_request_memory_partition,
+	.need_reset_on_init = &gmc_v9_0_need_reset_on_init,
+};
+
+static void gmc_v9_0_set_gmc_funcs(struct amdgpu_device *adev)
+{
+	adev->gmc.gmc_funcs = &gmc_v9_0_gmc_funcs;
+}
+
+static void gmc_v9_0_set_umc_funcs(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, UMC_HWIP, 0)) {
+	case IP_VERSION(6, 0, 0):
+		adev->umc.funcs = &umc_v6_0_funcs;
+		break;
+	case IP_VERSION(6, 1, 1):
+		adev->umc.max_ras_err_cnt_per_query = UMC_V6_1_TOTAL_CHANNEL_NUM;
+		adev->umc.channel_inst_num = UMC_V6_1_CHANNEL_INSTANCE_NUM;
+		adev->umc.umc_inst_num = UMC_V6_1_UMC_INSTANCE_NUM;
+		adev->umc.channel_offs = UMC_V6_1_PER_CHANNEL_OFFSET_VG20;
+		adev->umc.retire_unit = 1;
+		adev->umc.channel_idx_tbl = &umc_v6_1_channel_idx_tbl[0][0];
+		adev->umc.ras = &umc_v6_1_ras;
+		break;
+	case IP_VERSION(6, 1, 2):
+		adev->umc.max_ras_err_cnt_per_query = UMC_V6_1_TOTAL_CHANNEL_NUM;
+		adev->umc.channel_inst_num = UMC_V6_1_CHANNEL_INSTANCE_NUM;
+		adev->umc.umc_inst_num = UMC_V6_1_UMC_INSTANCE_NUM;
+		adev->umc.channel_offs = UMC_V6_1_PER_CHANNEL_OFFSET_ARCT;
+		adev->umc.retire_unit = 1;
+		adev->umc.channel_idx_tbl = &umc_v6_1_channel_idx_tbl[0][0];
+		adev->umc.ras = &umc_v6_1_ras;
+		break;
+	case IP_VERSION(6, 7, 0):
+		adev->umc.max_ras_err_cnt_per_query =
+			UMC_V6_7_TOTAL_CHANNEL_NUM * UMC_V6_7_BAD_PAGE_NUM_PER_CHANNEL;
+		adev->umc.channel_inst_num = UMC_V6_7_CHANNEL_INSTANCE_NUM;
+		adev->umc.umc_inst_num = UMC_V6_7_UMC_INSTANCE_NUM;
+		adev->umc.channel_offs = UMC_V6_7_PER_CHANNEL_OFFSET;
+		adev->umc.retire_unit = (UMC_V6_7_NA_MAP_PA_NUM * 2);
+		if (!adev->gmc.xgmi.connected_to_cpu)
+			adev->umc.ras = &umc_v6_7_ras;
+		if (1 & adev->smuio.funcs->get_die_id(adev))
+			adev->umc.channel_idx_tbl = &umc_v6_7_channel_idx_tbl_first[0][0];
+		else
+			adev->umc.channel_idx_tbl = &umc_v6_7_channel_idx_tbl_second[0][0];
+		break;
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 5, 0):
+		adev->umc.max_ras_err_cnt_per_query =
+			UMC_V12_0_TOTAL_CHANNEL_NUM(adev) * UMC_V12_0_BAD_PAGE_NUM_PER_CHANNEL;
+		adev->umc.channel_inst_num = UMC_V12_0_CHANNEL_INSTANCE_NUM;
+		adev->umc.umc_inst_num = UMC_V12_0_UMC_INSTANCE_NUM;
+		adev->umc.node_inst_num /= UMC_V12_0_UMC_INSTANCE_NUM;
+		adev->umc.channel_offs = UMC_V12_0_PER_CHANNEL_OFFSET;
+		if (!adev->gmc.xgmi.connected_to_cpu && !adev->gmc.is_app_apu)
+			adev->umc.ras = &umc_v12_0_ras;
+		break;
+	default:
+		break;
+	}
+}
+
+static void gmc_v9_0_set_mmhub_funcs(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) {
+	case IP_VERSION(9, 4, 1):
+		adev->mmhub.funcs = &mmhub_v9_4_funcs;
+		break;
+	case IP_VERSION(9, 4, 2):
+		adev->mmhub.funcs = &mmhub_v1_7_funcs;
+		break;
+	case IP_VERSION(1, 8, 0):
+	case IP_VERSION(1, 8, 1):
+		adev->mmhub.funcs = &mmhub_v1_8_funcs;
 		break;
 	default:
-		pte_flag |= AMDGPU_PTE_MTYPE(MTYPE_NC);
+		adev->mmhub.funcs = &mmhub_v1_0_funcs;
 		break;
 	}
+}
 
-	if (flags & AMDGPU_VM_PAGE_PRT)
-		pte_flag |= AMDGPU_PTE_PRT;
+static void gmc_v9_0_set_mmhub_ras_funcs(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) {
+	case IP_VERSION(9, 4, 0):
+		adev->mmhub.ras = &mmhub_v1_0_ras;
+		break;
+	case IP_VERSION(9, 4, 1):
+		adev->mmhub.ras = &mmhub_v9_4_ras;
+		break;
+	case IP_VERSION(9, 4, 2):
+		adev->mmhub.ras = &mmhub_v1_7_ras;
+		break;
+	case IP_VERSION(1, 8, 0):
+	case IP_VERSION(1, 8, 1):
+		adev->mmhub.ras = &mmhub_v1_8_ras;
+		break;
+	default:
+		/* mmhub ras is not available */
+		break;
+	}
+}
 
-	return pte_flag;
+static void gmc_v9_0_set_gfxhub_funcs(struct amdgpu_device *adev)
+{
+	if (amdgpu_is_multi_aid(adev))
+		adev->gfxhub.funcs = &gfxhub_v1_2_funcs;
+	else
+		adev->gfxhub.funcs = &gfxhub_v1_0_funcs;
 }
 
-static u64 gmc_v9_0_get_vm_pde(struct amdgpu_device *adev, u64 addr)
+static void gmc_v9_0_set_hdp_ras_funcs(struct amdgpu_device *adev)
 {
-	addr = adev->vm_manager.vram_base_offset + addr - adev->mc.vram_start;
-	BUG_ON(addr & 0xFFFF00000000003FULL);
-	return addr;
+	adev->hdp.ras = &hdp_v4_0_ras;
 }
 
-static const struct amdgpu_gart_funcs gmc_v9_0_gart_funcs = {
-	.flush_gpu_tlb = gmc_v9_0_gart_flush_gpu_tlb,
-	.set_pte_pde = gmc_v9_0_gart_set_pte_pde,
-	.get_invalidate_req = gmc_v9_0_get_invalidate_req,
-	.get_vm_pte_flags = gmc_v9_0_get_vm_pte_flags,
-	.get_vm_pde = gmc_v9_0_get_vm_pde
-};
+static void gmc_v9_0_set_mca_ras_funcs(struct amdgpu_device *adev)
+{
+	struct amdgpu_mca *mca = &adev->mca;
+
+	/* is UMC the right IP to check for MCA?  Maybe DF? */
+	switch (amdgpu_ip_version(adev, UMC_HWIP, 0)) {
+	case IP_VERSION(6, 7, 0):
+		if (!adev->gmc.xgmi.connected_to_cpu) {
+			mca->mp0.ras = &mca_v3_0_mp0_ras;
+			mca->mp1.ras = &mca_v3_0_mp1_ras;
+			mca->mpio.ras = &mca_v3_0_mpio_ras;
+		}
+		break;
+	default:
+		break;
+	}
+}
+
+static void gmc_v9_0_set_xgmi_ras_funcs(struct amdgpu_device *adev)
+{
+	if (!adev->gmc.xgmi.connected_to_cpu)
+		adev->gmc.xgmi.ras = &xgmi_ras;
+}
 
-static void gmc_v9_0_set_gart_funcs(struct amdgpu_device *adev)
+static void gmc_v9_0_init_nps_details(struct amdgpu_device *adev)
 {
-	if (adev->gart.gart_funcs == NULL)
-		adev->gart.gart_funcs = &gmc_v9_0_gart_funcs;
+	enum amdgpu_memory_partition mode;
+	uint32_t supp_modes;
+	int i;
+
+	adev->gmc.supported_nps_modes = 0;
+
+	if (amdgpu_sriov_vf(adev) || (adev->flags & AMD_IS_APU))
+		return;
+
+	mode = amdgpu_gmc_get_memory_partition(adev, &supp_modes);
+
+	/* Mode detected by hardware and supported modes available */
+	if ((mode != UNKNOWN_MEMORY_PARTITION_MODE) && supp_modes) {
+		while ((i = ffs(supp_modes))) {
+			if (AMDGPU_ALL_NPS_MASK & BIT(i))
+				adev->gmc.supported_nps_modes |= BIT(i);
+			supp_modes &= supp_modes - 1;
+		}
+	} else {
+		/*TODO: Check PSP version also which supports NPS switch. Otherwise keep
+	 * supported modes as 0.
+	 */
+		switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+		case IP_VERSION(9, 4, 3):
+		case IP_VERSION(9, 4, 4):
+			adev->gmc.supported_nps_modes =
+				BIT(AMDGPU_NPS1_PARTITION_MODE) |
+				BIT(AMDGPU_NPS4_PARTITION_MODE);
+			break;
+		default:
+			break;
+		}
+	}
 }
 
-static int gmc_v9_0_early_init(void *handle)
+static int gmc_v9_0_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	/*
+	 * 9.4.0, 9.4.1 and 9.4.3 don't have XGMI defined
+	 * in their IP discovery tables
+	 */
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 0) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 1) ||
+	    amdgpu_is_multi_aid(adev))
+		adev->gmc.xgmi.supported = true;
+
+	if (amdgpu_ip_version(adev, XGMI_HWIP, 0) == IP_VERSION(6, 1, 0)) {
+		adev->gmc.xgmi.supported = true;
+		adev->gmc.xgmi.connected_to_cpu =
+			adev->smuio.funcs->is_host_gpu_xgmi_supported(adev);
+	}
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3)) {
+		enum amdgpu_pkg_type pkg_type =
+			adev->smuio.funcs->get_pkg_type(adev);
+		/* On GFXIP 9.4.3. APU, there is no physical VRAM domain present
+		 * and the APU, can be in used two possible modes:
+		 *  - carveout mode
+		 *  - native APU mode
+		 * "is_app_apu" can be used to identify the APU in the native
+		 * mode.
+		 */
+		adev->gmc.is_app_apu = (pkg_type == AMDGPU_PKG_TYPE_APU &&
+					!pci_resource_len(adev->pdev, 0));
+	}
 
-	gmc_v9_0_set_gart_funcs(adev);
+	gmc_v9_0_set_gmc_funcs(adev);
 	gmc_v9_0_set_irq_funcs(adev);
+	gmc_v9_0_set_umc_funcs(adev);
+	gmc_v9_0_set_mmhub_funcs(adev);
+	gmc_v9_0_set_mmhub_ras_funcs(adev);
+	gmc_v9_0_set_gfxhub_funcs(adev);
+	gmc_v9_0_set_hdp_ras_funcs(adev);
+	gmc_v9_0_set_mca_ras_funcs(adev);
+	gmc_v9_0_set_xgmi_ras_funcs(adev);
+
+	adev->gmc.shared_aperture_start = 0x2000000000000000ULL;
+	adev->gmc.shared_aperture_end =
+		adev->gmc.shared_aperture_start + (4ULL << 30) - 1;
+	adev->gmc.private_aperture_start = 0x1000000000000000ULL;
+	adev->gmc.private_aperture_end =
+		adev->gmc.private_aperture_start + (4ULL << 30) - 1;
+	adev->gmc.noretry_flags = AMDGPU_VM_NORETRY_FLAGS_TF;
 
 	return 0;
 }
 
-static int gmc_v9_0_late_init(void *handle)
+static int gmc_v9_0_late_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	unsigned vm_inv_eng[AMDGPU_MAX_VMHUBS] = { 3, 3 };
-	unsigned i;
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
 
-	for(i = 0; i < adev->num_rings; ++i) {
-		struct amdgpu_ring *ring = adev->rings[i];
-		unsigned vmhub = ring->funcs->vmhub;
+	r = amdgpu_gmc_allocate_vm_inv_eng(adev);
+	if (r)
+		return r;
 
-		ring->vm_inv_eng = vm_inv_eng[vmhub]++;
-		dev_info(adev->dev, "ring %u(%s) uses VM inv eng %u on hub %u\n",
-			 ring->idx, ring->name, ring->vm_inv_eng,
-			 ring->funcs->vmhub);
+	/*
+	 * Workaround performance drop issue with VBIOS enables partial
+	 * writes, while disables HBM ECC for vega10.
+	 */
+	if (!amdgpu_sriov_vf(adev) &&
+	    (amdgpu_ip_version(adev, UMC_HWIP, 0) == IP_VERSION(6, 0, 0))) {
+		if (!(adev->ras_enabled & (1 << AMDGPU_RAS_BLOCK__UMC))) {
+			if (adev->df.funcs &&
+			    adev->df.funcs->enable_ecc_force_par_wr_rmw)
+				adev->df.funcs->enable_ecc_force_par_wr_rmw(adev, false);
+		}
+	}
+
+	if (!amdgpu_persistent_edc_harvesting_supported(adev)) {
+		amdgpu_ras_reset_error_count(adev, AMDGPU_RAS_BLOCK__MMHUB);
+		amdgpu_ras_reset_error_count(adev, AMDGPU_RAS_BLOCK__HDP);
 	}
 
-	/* Engine 17 is used for GART flushes */
-	for(i = 0; i < AMDGPU_MAX_VMHUBS; ++i)
-		BUG_ON(vm_inv_eng[i] > 17);
+	r = amdgpu_gmc_ras_late_init(adev);
+	if (r)
+		return r;
 
-	return amdgpu_irq_get(adev, &adev->mc.vm_fault, 0);
+	return amdgpu_irq_get(adev, &adev->gmc.vm_fault, 0);
 }
 
 static void gmc_v9_0_vram_gtt_location(struct amdgpu_device *adev,
-					struct amdgpu_mc *mc)
+					struct amdgpu_gmc *mc)
 {
-	u64 base = 0;
-	if (!amdgpu_sriov_vf(adev))
-		base = mmhub_v1_0_get_fb_location(adev);
-	amdgpu_vram_location(adev, &adev->mc, base);
-	amdgpu_gart_location(adev, mc);
+	u64 base = adev->mmhub.funcs->get_fb_location(adev);
+
+	amdgpu_gmc_set_agp_default(adev, mc);
+
+	/* add the xgmi offset of the physical node */
+	base += adev->gmc.xgmi.physical_node_id * adev->gmc.xgmi.node_segment_size;
+	if (amdgpu_gmc_is_pdb0_enabled(adev)) {
+		amdgpu_gmc_sysvm_location(adev, mc);
+	} else {
+		amdgpu_gmc_vram_location(adev, mc, base);
+		amdgpu_gmc_gart_location(adev, mc, AMDGPU_GART_PLACEMENT_BEST_FIT);
+		if (!amdgpu_sriov_vf(adev) && (amdgpu_agp == 1))
+			amdgpu_gmc_agp_location(adev, mc);
+	}
 	/* base offset of vram pages */
-	if (adev->flags & AMD_IS_APU)
-		adev->vm_manager.vram_base_offset = gfxhub_v1_0_get_mc_fb_offset(adev);
-	else
-		adev->vm_manager.vram_base_offset = 0;
+	adev->vm_manager.vram_base_offset = adev->gfxhub.funcs->get_mc_fb_offset(adev);
+
+	/* XXX: add the xgmi offset of the physical node? */
+	adev->vm_manager.vram_base_offset +=
+		adev->gmc.xgmi.physical_node_id * adev->gmc.xgmi.node_segment_size;
 }
 
 /**
@@ -441,80 +1696,81 @@ static void gmc_v9_0_vram_gtt_location(struct amdgpu_device *adev,
  */
 static int gmc_v9_0_mc_init(struct amdgpu_device *adev)
 {
-	u32 tmp;
-	int chansize, numchan;
+	int r;
 
-	adev->mc.vram_width = amdgpu_atomfirmware_get_vram_width(adev);
-	if (!adev->mc.vram_width) {
-		/* hbm memory channel size */
-		chansize = 128;
+	/* size in MB on si */
+	if (!adev->gmc.is_app_apu) {
+		adev->gmc.mc_vram_size =
+			adev->nbio.funcs->get_memsize(adev) * 1024ULL * 1024ULL;
+	} else {
+		DRM_DEBUG("Set mc_vram_size = 0 for APP APU\n");
+		adev->gmc.mc_vram_size = 0;
+	}
+	adev->gmc.real_vram_size = adev->gmc.mc_vram_size;
 
-		tmp = RREG32_SOC15(DF, 0, mmDF_CS_AON0_DramBaseAddress0);
-		tmp &= DF_CS_AON0_DramBaseAddress0__IntLvNumChan_MASK;
-		tmp >>= DF_CS_AON0_DramBaseAddress0__IntLvNumChan__SHIFT;
-		switch (tmp) {
-		case 0:
-		default:
-			numchan = 1;
-			break;
-		case 1:
-			numchan = 2;
-			break;
-		case 2:
-			numchan = 0;
-			break;
-		case 3:
-			numchan = 4;
-			break;
-		case 4:
-			numchan = 0;
-			break;
-		case 5:
-			numchan = 8;
-			break;
-		case 6:
-			numchan = 0;
-			break;
-		case 7:
-			numchan = 16;
-			break;
-		case 8:
-			numchan = 2;
-			break;
-		}
-		adev->mc.vram_width = numchan * chansize;
+	if (!(adev->flags & AMD_IS_APU) &&
+	    !adev->gmc.xgmi.connected_to_cpu) {
+		r = amdgpu_device_resize_fb_bar(adev);
+		if (r)
+			return r;
 	}
+	adev->gmc.aper_base = pci_resource_start(adev->pdev, 0);
+	adev->gmc.aper_size = pci_resource_len(adev->pdev, 0);
 
-	/* Could aper size report 0 ? */
-	adev->mc.aper_base = pci_resource_start(adev->pdev, 0);
-	adev->mc.aper_size = pci_resource_len(adev->pdev, 0);
-	/* size in MB on si */
-	adev->mc.mc_vram_size =
-		((adev->flags & AMD_IS_APU) ? nbio_v7_0_get_memsize(adev) :
-		 nbio_v6_1_get_memsize(adev)) * 1024ULL * 1024ULL;
-	adev->mc.real_vram_size = adev->mc.mc_vram_size;
-	adev->mc.visible_vram_size = adev->mc.aper_size;
+#ifdef CONFIG_X86_64
+	/*
+	 * AMD Accelerated Processing Platform (APP) supporting GPU-HOST xgmi
+	 * interface can use VRAM through here as it appears system reserved
+	 * memory in host address space.
+	 *
+	 * For APUs, VRAM is just the stolen system memory and can be accessed
+	 * directly.
+	 *
+	 * Otherwise, use the legacy Host Data Path (HDP) through PCIe BAR.
+	 */
+
+	/* check whether both host-gpu and gpu-gpu xgmi links exist */
+	if ((!amdgpu_sriov_vf(adev) &&
+		(adev->flags & AMD_IS_APU) && !amdgpu_passthrough(adev)) ||
+	    (adev->gmc.xgmi.supported &&
+	     adev->gmc.xgmi.connected_to_cpu)) {
+		adev->gmc.aper_base =
+			adev->gfxhub.funcs->get_mc_fb_offset(adev) +
+			adev->gmc.xgmi.physical_node_id *
+			adev->gmc.xgmi.node_segment_size;
+		adev->gmc.aper_size = adev->gmc.real_vram_size;
+	}
 
-	/* In case the PCI BAR is larger than the actual amount of vram */
-	if (adev->mc.visible_vram_size > adev->mc.real_vram_size)
-		adev->mc.visible_vram_size = adev->mc.real_vram_size;
+#endif
+	adev->gmc.visible_vram_size = adev->gmc.aper_size;
 
 	/* set the gart size */
 	if (amdgpu_gart_size == -1) {
-		switch (adev->asic_type) {
-		case CHIP_VEGA10:  /* all engines support GPUVM */
+		switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+		case IP_VERSION(9, 0, 1):  /* all engines support GPUVM */
+		case IP_VERSION(9, 2, 1):  /* all engines support GPUVM */
+		case IP_VERSION(9, 4, 0):
+		case IP_VERSION(9, 4, 1):
+		case IP_VERSION(9, 4, 2):
+		case IP_VERSION(9, 4, 3):
+		case IP_VERSION(9, 4, 4):
+		case IP_VERSION(9, 5, 0):
 		default:
-			adev->mc.gart_size = 256ULL << 20;
+			adev->gmc.gart_size = 512ULL << 20;
 			break;
-		case CHIP_RAVEN:   /* DCE SG support */
-			adev->mc.gart_size = 1024ULL << 20;
+		case IP_VERSION(9, 1, 0):   /* DCE SG support */
+		case IP_VERSION(9, 2, 2):   /* DCE SG support */
+		case IP_VERSION(9, 3, 0):
+			adev->gmc.gart_size = 1024ULL << 20;
 			break;
 		}
 	} else {
-		adev->mc.gart_size = (u64)amdgpu_gart_size << 20;
+		adev->gmc.gart_size = (u64)amdgpu_gart_size << 20;
 	}
 
-	gmc_v9_0_vram_gtt_location(adev, &adev->mc);
+	adev->gmc.gart_size += adev->pm.smu_prv_buffer_size;
+
+	gmc_v9_0_vram_gtt_location(adev, &adev->gmc);
 
 	return 0;
 }
@@ -523,113 +1779,244 @@ static int gmc_v9_0_gart_init(struct amdgpu_device *adev)
 {
 	int r;
 
-	if (adev->gart.robj) {
+	if (adev->gart.bo) {
 		WARN(1, "VEGA10 PCIE GART already initialized\n");
 		return 0;
 	}
+
+	if (amdgpu_gmc_is_pdb0_enabled(adev)) {
+		adev->gmc.vmid0_page_table_depth = 1;
+		adev->gmc.vmid0_page_table_block_size = 12;
+	} else {
+		adev->gmc.vmid0_page_table_depth = 0;
+		adev->gmc.vmid0_page_table_block_size = 0;
+	}
+
 	/* Initialize common gart structure */
 	r = amdgpu_gart_init(adev);
 	if (r)
 		return r;
 	adev->gart.table_size = adev->gart.num_gpu_pages * 8;
-	adev->gart.gart_pte_flags = AMDGPU_PTE_MTYPE(MTYPE_UC) |
+	adev->gart.gart_pte_flags = AMDGPU_PTE_MTYPE_VG10(0ULL, MTYPE_UC) |
 				 AMDGPU_PTE_EXECUTABLE;
-	return amdgpu_gart_table_vram_alloc(adev);
+
+	if (!adev->gmc.real_vram_size) {
+		dev_info(adev->dev, "Put GART in system memory for APU\n");
+		r = amdgpu_gart_table_ram_alloc(adev);
+		if (r)
+			dev_err(adev->dev, "Failed to allocate GART in system memory\n");
+	} else {
+		r = amdgpu_gart_table_vram_alloc(adev);
+		if (r)
+			return r;
+
+		if (amdgpu_gmc_is_pdb0_enabled(adev))
+			r = amdgpu_gmc_pdb0_alloc(adev);
+	}
+
+	return r;
 }
 
-static int gmc_v9_0_sw_init(void *handle)
+/**
+ * gmc_v9_0_save_registers - saves regs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * This saves potential register values that should be
+ * restored upon resume
+ */
+static void gmc_v9_0_save_registers(struct amdgpu_device *adev)
 {
-	int r;
-	int dma_bits;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	if ((amdgpu_ip_version(adev, DCE_HWIP, 0) == IP_VERSION(1, 0, 0)) ||
+	    (amdgpu_ip_version(adev, DCE_HWIP, 0) == IP_VERSION(1, 0, 1)))
+		adev->gmc.sdpif_register = RREG32_SOC15(DCE, 0, mmDCHUBBUB_SDPIF_MMIO_CNTRL_0);
+}
+
+static void gmc_v9_4_3_init_vram_info(struct amdgpu_device *adev)
+{
+	static const u32 regBIF_BIOS_SCRATCH_4 = 0x50;
+	u32 vram_info;
+
+	adev->gmc.vram_type = AMDGPU_VRAM_TYPE_HBM;
+	adev->gmc.vram_width = 128 * 64;
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0))
+		adev->gmc.vram_type = AMDGPU_VRAM_TYPE_HBM3E;
+
+	if (!(adev->flags & AMD_IS_APU) && !amdgpu_sriov_vf(adev)) {
+		vram_info = RREG32(regBIF_BIOS_SCRATCH_4);
+		adev->gmc.vram_vendor = vram_info & 0xF;
+	}
+}
+
+static int gmc_v9_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	int r, vram_width = 0, vram_type = 0, vram_vendor = 0, dma_addr_bits;
+	struct amdgpu_device *adev = ip_block->adev;
+	unsigned long inst_mask = adev->aid_mask;
+
+	adev->gfxhub.funcs->init(adev);
 
-	gfxhub_v1_0_init(adev);
-	mmhub_v1_0_init(adev);
+	adev->mmhub.funcs->init(adev);
 
-	spin_lock_init(&adev->mc.invalidate_lock);
+	spin_lock_init(&adev->gmc.invalidate_lock);
+
+	if (amdgpu_is_multi_aid(adev)) {
+		gmc_v9_4_3_init_vram_info(adev);
+	} else if (!adev->bios) {
+		if (adev->flags & AMD_IS_APU) {
+			adev->gmc.vram_type = AMDGPU_VRAM_TYPE_DDR4;
+			adev->gmc.vram_width = 64 * 64;
+		} else {
+			adev->gmc.vram_type = AMDGPU_VRAM_TYPE_HBM;
+			adev->gmc.vram_width = 128 * 64;
+		}
+	} else {
+		r = amdgpu_atomfirmware_get_vram_info(adev,
+			&vram_width, &vram_type, &vram_vendor);
+		if (amdgpu_sriov_vf(adev))
+			/* For Vega10 SR-IOV, vram_width can't be read from ATOM as RAVEN,
+			 * and DF related registers is not readable, seems hardcord is the
+			 * only way to set the correct vram_width
+			 */
+			adev->gmc.vram_width = 2048;
+		else if (amdgpu_emu_mode != 1)
+			adev->gmc.vram_width = vram_width;
+
+		if (!adev->gmc.vram_width) {
+			int chansize, numchan;
+
+			/* hbm memory channel size */
+			if (adev->flags & AMD_IS_APU)
+				chansize = 64;
+			else
+				chansize = 128;
+			if (adev->df.funcs &&
+			    adev->df.funcs->get_hbm_channel_number) {
+				numchan = adev->df.funcs->get_hbm_channel_number(adev);
+				adev->gmc.vram_width = numchan * chansize;
+			}
+		}
+
+		adev->gmc.vram_type = vram_type;
+		adev->gmc.vram_vendor = vram_vendor;
+	}
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 1, 0):
+	case IP_VERSION(9, 2, 2):
+		set_bit(AMDGPU_GFXHUB(0), adev->vmhubs_mask);
+		set_bit(AMDGPU_MMHUB0(0), adev->vmhubs_mask);
 
-	switch (adev->asic_type) {
-	case CHIP_RAVEN:
-		adev->mc.vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
 		if (adev->rev_id == 0x0 || adev->rev_id == 0x1) {
-			adev->vm_manager.vm_size = 1U << 18;
-			adev->vm_manager.block_size = 9;
-			adev->vm_manager.num_level = 3;
-			amdgpu_vm_set_fragment_size(adev, 9);
+			amdgpu_vm_adjust_size(adev, 256 * 1024, 9, 3, 48);
 		} else {
-			/* vm_size is 64GB for legacy 2-level page support */
-			amdgpu_vm_adjust_size(adev, 64, 9);
-			adev->vm_manager.num_level = 1;
+			/* vm_size is 128TB + 512GB for legacy 3-level page support */
+			amdgpu_vm_adjust_size(adev, 128 * 1024 + 512, 9, 2, 48);
+			adev->gmc.translate_further =
+				adev->vm_manager.num_level > 1;
 		}
 		break;
-	case CHIP_VEGA10:
-		/* XXX Don't know how to get VRAM type yet. */
-		adev->mc.vram_type = AMDGPU_VRAM_TYPE_HBM;
+	case IP_VERSION(9, 0, 1):
+	case IP_VERSION(9, 2, 1):
+	case IP_VERSION(9, 4, 0):
+	case IP_VERSION(9, 3, 0):
+	case IP_VERSION(9, 4, 2):
+		set_bit(AMDGPU_GFXHUB(0), adev->vmhubs_mask);
+		set_bit(AMDGPU_MMHUB0(0), adev->vmhubs_mask);
+
 		/*
 		 * To fulfill 4-level page support,
 		 * vm size is 256TB (48bit), maximum size of Vega10,
 		 * block size 512 (9bit)
 		 */
-		adev->vm_manager.vm_size = 1U << 18;
-		adev->vm_manager.block_size = 9;
-		adev->vm_manager.num_level = 3;
-		amdgpu_vm_set_fragment_size(adev, 9);
+
+		amdgpu_vm_adjust_size(adev, 256 * 1024, 9, 3, 48);
+		if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2))
+			adev->gmc.translate_further = adev->vm_manager.num_level > 1;
+		break;
+	case IP_VERSION(9, 4, 1):
+		set_bit(AMDGPU_GFXHUB(0), adev->vmhubs_mask);
+		set_bit(AMDGPU_MMHUB0(0), adev->vmhubs_mask);
+		set_bit(AMDGPU_MMHUB1(0), adev->vmhubs_mask);
+
+		/* Keep the vm size same with Vega20 */
+		amdgpu_vm_adjust_size(adev, 256 * 1024, 9, 3, 48);
+		adev->gmc.translate_further = adev->vm_manager.num_level > 1;
+		break;
+	case IP_VERSION(9, 4, 3):
+	case IP_VERSION(9, 4, 4):
+	case IP_VERSION(9, 5, 0):
+		bitmap_set(adev->vmhubs_mask, AMDGPU_GFXHUB(0),
+				  NUM_XCC(adev->gfx.xcc_mask));
+
+		inst_mask <<= AMDGPU_MMHUB0(0);
+		bitmap_or(adev->vmhubs_mask, adev->vmhubs_mask, &inst_mask, 32);
+
+		amdgpu_vm_adjust_size(adev, 256 * 1024, 9, 3, 48);
+		adev->gmc.translate_further = adev->vm_manager.num_level > 1;
 		break;
 	default:
 		break;
 	}
 
-	DRM_INFO("vm size is %llu GB, block size is %u-bit,fragment size is %u-bit\n",
-			adev->vm_manager.vm_size,
-			adev->vm_manager.block_size,
-			adev->vm_manager.fragment_size);
-
 	/* This interrupt is VMC page fault.*/
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_VMC, 0,
-				&adev->mc.vm_fault);
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_UTCL2, 0,
-				&adev->mc.vm_fault);
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VMC, VMC_1_0__SRCID__VM_FAULT,
+				&adev->gmc.vm_fault);
+	if (r)
+		return r;
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 1)) {
+		r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VMC1, VMC_1_0__SRCID__VM_FAULT,
+					&adev->gmc.vm_fault);
+		if (r)
+			return r;
+	}
+
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_UTCL2, UTCL2_1_0__SRCID__FAULT,
+				&adev->gmc.vm_fault);
 
 	if (r)
 		return r;
 
-	adev->vm_manager.max_pfn = adev->vm_manager.vm_size << 18;
+	if (!amdgpu_sriov_vf(adev) &&
+	    !adev->gmc.xgmi.connected_to_cpu &&
+	    !adev->gmc.is_app_apu) {
+		/* interrupt sent to DF. */
+		r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_DF, 0,
+				      &adev->gmc.ecc_irq);
+		if (r)
+			return r;
+	}
 
 	/* Set the internal MC address mask
 	 * This is the max address of the GPU's
 	 * internal address space.
 	 */
-	adev->mc.mc_mask = 0xffffffffffffULL; /* 48 bit MC */
+	adev->gmc.mc_mask = 0xffffffffffffULL; /* 48 bit MC */
 
-	/*
-	 * It needs to reserve 8M stolen memory for vega10
-	 * TODO: Figure out how to avoid that...
-	 */
-	adev->mc.stolen_size = 8 * 1024 * 1024;
-
-	/* set DMA mask + need_dma32 flags.
-	 * PCIE - can handle 44-bits.
-	 * IGP - can handle 44-bits
-	 * PCI - dma32 for legacy pci gart, 44 bits on vega10
-	 */
-	adev->need_dma32 = false;
-	dma_bits = adev->need_dma32 ? 32 : 44;
-	r = pci_set_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
+	dma_addr_bits = amdgpu_ip_version(adev, GC_HWIP, 0) >=
+					IP_VERSION(9, 4, 2) ?
+				48 :
+				44;
+	r = dma_set_mask_and_coherent(adev->dev, DMA_BIT_MASK(dma_addr_bits));
 	if (r) {
-		adev->need_dma32 = true;
-		dma_bits = 32;
-		printk(KERN_WARNING "amdgpu: No suitable DMA available.\n");
-	}
-	r = pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
-	if (r) {
-		pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(32));
-		printk(KERN_WARNING "amdgpu: No coherent DMA available.\n");
+		dev_warn(adev->dev, "amdgpu: No suitable DMA available.\n");
+		return r;
 	}
+	adev->need_swiotlb = drm_need_swiotlb(dma_addr_bits);
 
 	r = gmc_v9_0_mc_init(adev);
 	if (r)
 		return r;
 
+	amdgpu_gmc_get_vbios_allocations(adev);
+
+	if (amdgpu_is_multi_aid(adev)) {
+		r = amdgpu_gmc_init_mem_ranges(adev);
+		if (r)
+			return r;
+	}
+
 	/* Memory manager */
 	r = amdgpu_bo_init(adev);
 	if (r)
@@ -639,51 +2026,84 @@ static int gmc_v9_0_sw_init(void *handle)
 	if (r)
 		return r;
 
+	gmc_v9_0_init_nps_details(adev);
 	/*
 	 * number of VMs
 	 * VMID 0 is reserved for System
-	 * amdgpu graphics/compute will use VMIDs 1-7
-	 * amdkfd will use VMIDs 8-15
+	 * amdgpu graphics/compute will use VMIDs 1..n-1
+	 * amdkfd will use VMIDs n..15
+	 *
+	 * The first KFD VMID is 8 for GPUs with graphics, 3 for
+	 * compute-only GPUs. On compute-only GPUs that leaves 2 VMIDs
+	 * for video processing.
 	 */
-	adev->vm_manager.id_mgr[AMDGPU_GFXHUB].num_ids = AMDGPU_NUM_OF_VMIDS;
-	adev->vm_manager.id_mgr[AMDGPU_MMHUB].num_ids = AMDGPU_NUM_OF_VMIDS;
+	adev->vm_manager.first_kfd_vmid =
+		(amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 1) ||
+		 amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2) ||
+		 amdgpu_is_multi_aid(adev)) ?
+			3 :
+			8;
 
 	amdgpu_vm_manager_init(adev);
 
+	gmc_v9_0_save_registers(adev);
+
+	r = amdgpu_gmc_ras_sw_init(adev);
+	if (r)
+		return r;
+
+	if (amdgpu_is_multi_aid(adev))
+		amdgpu_gmc_sysfs_init(adev);
+
 	return 0;
 }
 
-/**
- * gmc_v8_0_gart_fini - vm fini callback
- *
- * @adev: amdgpu_device pointer
- *
- * Tears down the driver GART/VM setup (CIK).
- */
-static void gmc_v9_0_gart_fini(struct amdgpu_device *adev)
+static int gmc_v9_0_sw_fini(struct amdgpu_ip_block *ip_block)
 {
-	amdgpu_gart_table_vram_free(adev);
-	amdgpu_gart_fini(adev);
-}
+	struct amdgpu_device *adev = ip_block->adev;
 
-static int gmc_v9_0_sw_fini(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	if (amdgpu_is_multi_aid(adev))
+		amdgpu_gmc_sysfs_fini(adev);
 
-	amdgpu_vm_manager_fini(adev);
-	gmc_v9_0_gart_fini(adev);
+	amdgpu_gmc_ras_fini(adev);
 	amdgpu_gem_force_release(adev);
+	amdgpu_vm_manager_fini(adev);
+	if (!adev->gmc.real_vram_size) {
+		dev_info(adev->dev, "Put GART in system memory for APU free\n");
+		amdgpu_gart_table_ram_free(adev);
+	} else {
+		amdgpu_gart_table_vram_free(adev);
+	}
+	amdgpu_bo_free_kernel(&adev->gmc.pdb0_bo, NULL, &adev->gmc.ptr_pdb0);
 	amdgpu_bo_fini(adev);
 
+	adev->gmc.num_mem_partitions = 0;
+	kfree(adev->gmc.mem_partitions);
+
 	return 0;
 }
 
 static void gmc_v9_0_init_golden_registers(struct amdgpu_device *adev)
 {
-	switch (adev->asic_type) {
-	case CHIP_VEGA10:
+	switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) {
+	case IP_VERSION(9, 0, 0):
+		if (amdgpu_sriov_vf(adev))
+			break;
+		fallthrough;
+	case IP_VERSION(9, 4, 0):
+		soc15_program_register_sequence(adev,
+						golden_settings_mmhub_1_0_0,
+						ARRAY_SIZE(golden_settings_mmhub_1_0_0));
+		soc15_program_register_sequence(adev,
+						golden_settings_athub_1_0_0,
+						ARRAY_SIZE(golden_settings_athub_1_0_0));
 		break;
-	case CHIP_RAVEN:
+	case IP_VERSION(9, 1, 0):
+	case IP_VERSION(9, 2, 0):
+		/* TODO for renoir */
+		soc15_program_register_sequence(adev,
+						golden_settings_athub_1_0_0,
+						ARRAY_SIZE(golden_settings_athub_1_0_0));
 		break;
 	default:
 		break;
@@ -691,6 +2111,23 @@ static void gmc_v9_0_init_golden_registers(struct amdgpu_device *adev)
 }
 
 /**
+ * gmc_v9_0_restore_registers - restores regs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * This restores register values, saved at suspend.
+ */
+void gmc_v9_0_restore_registers(struct amdgpu_device *adev)
+{
+	if ((amdgpu_ip_version(adev, DCE_HWIP, 0) == IP_VERSION(1, 0, 0)) ||
+	    (amdgpu_ip_version(adev, DCE_HWIP, 0) == IP_VERSION(1, 0, 1))) {
+		WREG32_SOC15(DCE, 0, mmDCHUBBUB_SDPIF_MMIO_CNTRL_0, adev->gmc.sdpif_register);
+		WARN_ON(adev->gmc.sdpif_register !=
+			RREG32_SOC15(DCE, 0, mmDCHUBBUB_SDPIF_MMIO_CNTRL_0));
+	}
+}
+
+/**
  * gmc_v9_0_gart_enable - gart enable
  *
  * @adev: amdgpu_device pointer
@@ -698,94 +2135,100 @@ static void gmc_v9_0_init_golden_registers(struct amdgpu_device *adev)
 static int gmc_v9_0_gart_enable(struct amdgpu_device *adev)
 {
 	int r;
-	bool value;
-	u32 tmp;
 
-	amdgpu_program_register_sequence(adev,
-		golden_settings_vega10_hdp,
-		(const u32)ARRAY_SIZE(golden_settings_vega10_hdp));
+	if (amdgpu_gmc_is_pdb0_enabled(adev))
+		amdgpu_gmc_init_pdb0(adev);
 
-	if (adev->gart.robj == NULL) {
+	if (adev->gart.bo == NULL) {
 		dev_err(adev->dev, "No VRAM object for PCIE GART.\n");
 		return -EINVAL;
 	}
-	r = amdgpu_gart_table_vram_pin(adev);
-	if (r)
-		return r;
 
-	/* After HDP is initialized, flush HDP.*/
-	if (adev->flags & AMD_IS_APU)
-		nbio_v7_0_hdp_flush(adev);
-	else
-		nbio_v6_1_hdp_flush(adev);
+	amdgpu_gtt_mgr_recover(&adev->mman.gtt_mgr);
 
-	switch (adev->asic_type) {
-	case CHIP_RAVEN:
-		mmhub_v1_0_initialize_power_gating(adev);
-		mmhub_v1_0_update_power_gating(adev, true);
-		break;
-	default:
-		break;
+	if (!adev->in_s0ix) {
+		r = adev->gfxhub.funcs->gart_enable(adev);
+		if (r)
+			return r;
 	}
 
-	r = gfxhub_v1_0_gart_enable(adev);
-	if (r)
-		return r;
-
-	r = mmhub_v1_0_gart_enable(adev);
+	r = adev->mmhub.funcs->gart_enable(adev);
 	if (r)
 		return r;
 
-	tmp = RREG32_SOC15(HDP, 0, mmHDP_MISC_CNTL);
-	tmp |= HDP_MISC_CNTL__FLUSH_INVALIDATE_CACHE_MASK;
-	WREG32_SOC15(HDP, 0, mmHDP_MISC_CNTL, tmp);
-
-	tmp = RREG32_SOC15(HDP, 0, mmHDP_HOST_PATH_CNTL);
-	WREG32_SOC15(HDP, 0, mmHDP_HOST_PATH_CNTL, tmp);
-
+	DRM_INFO("PCIE GART of %uM enabled.\n",
+		 (unsigned int)(adev->gmc.gart_size >> 20));
+	if (adev->gmc.pdb0_bo)
+		DRM_INFO("PDB0 located at 0x%016llX\n",
+				(unsigned long long)amdgpu_bo_gpu_offset(adev->gmc.pdb0_bo));
+	DRM_INFO("PTB located at 0x%016llX\n",
+			(unsigned long long)amdgpu_bo_gpu_offset(adev->gart.bo));
 
-	if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS)
-		value = false;
-	else
-		value = true;
-
-	gfxhub_v1_0_set_fault_enable_default(adev, value);
-	mmhub_v1_0_set_fault_enable_default(adev, value);
-
-	gmc_v9_0_gart_flush_gpu_tlb(adev, 0);
-
-	DRM_INFO("PCIE GART of %uM enabled (table at 0x%016llX).\n",
-		 (unsigned)(adev->mc.gart_size >> 20),
-		 (unsigned long long)adev->gart.table_addr);
-	adev->gart.ready = true;
 	return 0;
 }
 
-static int gmc_v9_0_hw_init(void *handle)
+static int gmc_v9_0_hw_init(struct amdgpu_ip_block *ip_block)
 {
-	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+	bool value;
+	int i, r;
+
+	adev->gmc.flush_pasid_uses_kiq = true;
+
+	/* Vega20+XGMI caches PTEs in TC and TLB. Add a heavy-weight TLB flush
+	 * (type 2), which flushes both. Due to a race condition with
+	 * concurrent memory accesses using the same TLB cache line, we still
+	 * need a second TLB flush after this.
+	 */
+	adev->gmc.flush_tlb_needs_extra_type_2 =
+		amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 0) &&
+		adev->gmc.xgmi.num_physical_nodes;
 
 	/* The sequence of these two function calls matters.*/
 	gmc_v9_0_init_golden_registers(adev);
 
 	if (adev->mode_info.num_crtc) {
-		u32 tmp;
-
 		/* Lockout access through VGA aperture*/
-		tmp = RREG32_SOC15(DCE, 0, mmVGA_HDP_CONTROL);
-		tmp = REG_SET_FIELD(tmp, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 1);
-		WREG32_SOC15(DCE, 0, mmVGA_HDP_CONTROL, tmp);
-
+		WREG32_FIELD15(DCE, 0, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 1);
 		/* disable VGA render */
-		tmp = RREG32_SOC15(DCE, 0, mmVGA_RENDER_CONTROL);
-		tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 0);
-		WREG32_SOC15(DCE, 0, mmVGA_RENDER_CONTROL, tmp);
+		WREG32_FIELD15(DCE, 0, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 0);
 	}
 
+	if (adev->mmhub.funcs->update_power_gating)
+		adev->mmhub.funcs->update_power_gating(adev, true);
+
+	adev->hdp.funcs->init_registers(adev);
+
+	/* After HDP is initialized, flush HDP.*/
+	amdgpu_device_flush_hdp(adev, NULL);
+
+	if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS)
+		value = false;
+	else
+		value = true;
+
+	if (!amdgpu_sriov_vf(adev)) {
+		if (!adev->in_s0ix)
+			adev->gfxhub.funcs->set_fault_enable_default(adev, value);
+		adev->mmhub.funcs->set_fault_enable_default(adev, value);
+	}
+	for_each_set_bit(i, adev->vmhubs_mask, AMDGPU_MAX_VMHUBS) {
+		if (adev->in_s0ix && (i == AMDGPU_GFXHUB(0)))
+			continue;
+		gmc_v9_0_flush_gpu_tlb(adev, 0, i, 0);
+	}
+
+	if (adev->umc.funcs && adev->umc.funcs->init_registers)
+		adev->umc.funcs->init_registers(adev);
+
 	r = gmc_v9_0_gart_enable(adev);
+	if (r)
+		return r;
 
-	return r;
+	if (amdgpu_emu_mode == 1)
+		return amdgpu_gmc_vram_checking(adev);
+
+	return 0;
 }
 
 /**
@@ -797,14 +2240,16 @@ static int gmc_v9_0_hw_init(void *handle)
  */
 static void gmc_v9_0_gart_disable(struct amdgpu_device *adev)
 {
-	gfxhub_v1_0_gart_disable(adev);
-	mmhub_v1_0_gart_disable(adev);
-	amdgpu_gart_table_vram_unpin(adev);
+	if (!adev->in_s0ix)
+		adev->gfxhub.funcs->gart_disable(adev);
+	adev->mmhub.funcs->gart_disable(adev);
 }
 
-static int gmc_v9_0_hw_fini(void *handle)
+static int gmc_v9_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	gmc_v9_0_gart_disable(adev);
 
 	if (amdgpu_sriov_vf(adev)) {
 		/* full access mode, so don't touch any GMC register */
@@ -812,69 +2257,96 @@ static int gmc_v9_0_hw_fini(void *handle)
 		return 0;
 	}
 
-	amdgpu_irq_put(adev, &adev->mc.vm_fault, 0);
-	gmc_v9_0_gart_disable(adev);
+	/*
+	 * Pair the operations did in gmc_v9_0_hw_init and thus maintain
+	 * a correct cached state for GMC. Otherwise, the "gate" again
+	 * operation on S3 resuming will fail due to wrong cached state.
+	 */
+	if (adev->mmhub.funcs->update_power_gating)
+		adev->mmhub.funcs->update_power_gating(adev, false);
+
+	/*
+	 * For minimal init, late_init is not called, hence VM fault/RAS irqs
+	 * are not enabled.
+	 */
+	if (adev->init_lvl->level != AMDGPU_INIT_LEVEL_MINIMAL_XGMI) {
+		amdgpu_irq_put(adev, &adev->gmc.vm_fault, 0);
+
+		if (adev->gmc.ecc_irq.funcs &&
+		    amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__UMC))
+			amdgpu_irq_put(adev, &adev->gmc.ecc_irq, 0);
+	}
 
 	return 0;
 }
 
-static int gmc_v9_0_suspend(void *handle)
+static int gmc_v9_0_suspend(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	gmc_v9_0_hw_fini(adev);
-
-	return 0;
+	return gmc_v9_0_hw_fini(ip_block);
 }
 
-static int gmc_v9_0_resume(void *handle)
+static int gmc_v9_0_resume(struct amdgpu_ip_block *ip_block)
 {
+	struct amdgpu_device *adev = ip_block->adev;
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	r = gmc_v9_0_hw_init(adev);
+	/* If a reset is done for NPS mode switch, read the memory range
+	 * information again.
+	 */
+	if (adev->gmc.reset_flags & AMDGPU_GMC_INIT_RESET_NPS) {
+		amdgpu_gmc_init_sw_mem_ranges(adev, adev->gmc.mem_partitions);
+		adev->gmc.reset_flags &= ~AMDGPU_GMC_INIT_RESET_NPS;
+	}
+
+	r = gmc_v9_0_hw_init(ip_block);
 	if (r)
 		return r;
 
-	amdgpu_vm_reset_all_ids(adev);
+	amdgpu_vmid_reset_all(ip_block->adev);
 
 	return 0;
 }
 
-static bool gmc_v9_0_is_idle(void *handle)
+static bool gmc_v9_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	/* MC is always ready in GMC v9.*/
 	return true;
 }
 
-static int gmc_v9_0_wait_for_idle(void *handle)
+static int gmc_v9_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
 	/* There is no need to wait for MC idle in GMC v9.*/
 	return 0;
 }
 
-static int gmc_v9_0_soft_reset(void *handle)
+static int gmc_v9_0_soft_reset(struct amdgpu_ip_block *ip_block)
 {
 	/* XXX for emulation.*/
 	return 0;
 }
 
-static int gmc_v9_0_set_clockgating_state(void *handle,
+static int gmc_v9_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					enum amd_clockgating_state state)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	adev->mmhub.funcs->set_clockgating(adev, state);
+
+	athub_v1_0_set_clockgating(adev, state);
 
-	return mmhub_v1_0_set_clockgating(adev, state);
+	return 0;
 }
 
-static void gmc_v9_0_get_clockgating_state(void *handle, u32 *flags)
+static void gmc_v9_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	adev->mmhub.funcs->get_clockgating(adev, flags);
 
-	mmhub_v1_0_get_clockgating(adev, flags);
+	athub_v1_0_get_clockgating(adev, flags);
 }
 
-static int gmc_v9_0_set_powergating_state(void *handle,
+static int gmc_v9_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					enum amd_powergating_state state)
 {
 	return 0;
@@ -898,8 +2370,7 @@ const struct amd_ip_funcs gmc_v9_0_ip_funcs = {
 	.get_clockgating_state = gmc_v9_0_get_clockgating_state,
 };
 
-const struct amdgpu_ip_block_version gmc_v9_0_ip_block =
-{
+const struct amdgpu_ip_block_version gmc_v9_0_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_GMC,
 	.major = 9,
 	.minor = 0,
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.h b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.h
index b030ca5ea107..c415c439f690 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.h
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.h
@@ -27,4 +27,5 @@
 extern const struct amd_ip_funcs gmc_v9_0_ip_funcs;
 extern const struct amdgpu_ip_block_version gmc_v9_0_ip_block;
 
+void gmc_v9_0_restore_registers(struct amdgpu_device *adev);
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
new file mode 100644
index 000000000000..e6c0d86d3486
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
@@ -0,0 +1,177 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "hdp_v4_0.h"
+#include "amdgpu_ras.h"
+
+#include "hdp/hdp_4_0_offset.h"
+#include "hdp/hdp_4_0_sh_mask.h"
+#include <uapi/linux/kfd_ioctl.h>
+
+/* for Vega20 register name change */
+#define mmHDP_MEM_POWER_CTRL    0x00d4
+#define HDP_MEM_POWER_CTRL__IPH_MEM_POWER_CTRL_EN_MASK  0x00000001L
+#define HDP_MEM_POWER_CTRL__IPH_MEM_POWER_LS_EN_MASK    0x00000002L
+#define HDP_MEM_POWER_CTRL__RC_MEM_POWER_CTRL_EN_MASK   0x00010000L
+#define HDP_MEM_POWER_CTRL__RC_MEM_POWER_LS_EN_MASK     0x00020000L
+#define mmHDP_MEM_POWER_CTRL_BASE_IDX   0
+
+static void hdp_v4_0_invalidate_hdp(struct amdgpu_device *adev,
+				    struct amdgpu_ring *ring)
+{
+	if (amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(4, 4, 0) ||
+	    amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(4, 4, 2) ||
+	    amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(4, 4, 5))
+		return;
+
+	if (!ring || !ring->funcs->emit_wreg) {
+		WREG32_SOC15_NO_KIQ(HDP, 0, mmHDP_READ_CACHE_INVALIDATE, 1);
+		RREG32_SOC15_NO_KIQ(HDP, 0, mmHDP_READ_CACHE_INVALIDATE);
+	} else {
+		amdgpu_ring_emit_wreg(ring, SOC15_REG_OFFSET(
+			HDP, 0, mmHDP_READ_CACHE_INVALIDATE), 1);
+	}
+}
+
+static void hdp_v4_0_query_ras_error_count(struct amdgpu_device *adev,
+					   void *ras_error_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+
+	err_data->ue_count = 0;
+	err_data->ce_count = 0;
+
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__HDP))
+		return;
+
+	/* HDP SRAM errors are uncorrectable ones (i.e. fatal errors) */
+	err_data->ue_count += RREG32_SOC15(HDP, 0, mmHDP_EDC_CNT);
+};
+
+static void hdp_v4_0_reset_ras_error_count(struct amdgpu_device *adev)
+{
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__HDP))
+		return;
+
+	if (amdgpu_ip_version(adev, HDP_HWIP, 0) >= IP_VERSION(4, 4, 0))
+		WREG32_SOC15(HDP, 0, mmHDP_EDC_CNT, 0);
+	else
+		/*read back hdp ras counter to reset it to 0 */
+		RREG32_SOC15(HDP, 0, mmHDP_EDC_CNT);
+}
+
+static void hdp_v4_0_update_clock_gating(struct amdgpu_device *adev,
+					 bool enable)
+{
+	uint32_t def, data;
+
+	if (amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(4, 0, 0) ||
+	    amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(4, 0, 1) ||
+	    amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(4, 1, 1) ||
+	    amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(4, 1, 0)) {
+		def = data = RREG32(SOC15_REG_OFFSET(HDP, 0, mmHDP_MEM_POWER_LS));
+
+		if (enable && (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS))
+			data |= HDP_MEM_POWER_LS__LS_ENABLE_MASK;
+		else
+			data &= ~HDP_MEM_POWER_LS__LS_ENABLE_MASK;
+
+		if (def != data)
+			WREG32(SOC15_REG_OFFSET(HDP, 0, mmHDP_MEM_POWER_LS), data);
+	} else {
+		def = data = RREG32(SOC15_REG_OFFSET(HDP, 0, mmHDP_MEM_POWER_CTRL));
+
+		if (enable && (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS))
+			data |= HDP_MEM_POWER_CTRL__IPH_MEM_POWER_CTRL_EN_MASK |
+				HDP_MEM_POWER_CTRL__IPH_MEM_POWER_LS_EN_MASK |
+				HDP_MEM_POWER_CTRL__RC_MEM_POWER_CTRL_EN_MASK |
+				HDP_MEM_POWER_CTRL__RC_MEM_POWER_LS_EN_MASK;
+		else
+			data &= ~(HDP_MEM_POWER_CTRL__IPH_MEM_POWER_CTRL_EN_MASK |
+				  HDP_MEM_POWER_CTRL__IPH_MEM_POWER_LS_EN_MASK |
+				  HDP_MEM_POWER_CTRL__RC_MEM_POWER_CTRL_EN_MASK |
+				  HDP_MEM_POWER_CTRL__RC_MEM_POWER_LS_EN_MASK);
+
+		if (def != data)
+			WREG32(SOC15_REG_OFFSET(HDP, 0, mmHDP_MEM_POWER_CTRL), data);
+	}
+}
+
+static void hdp_v4_0_get_clockgating_state(struct amdgpu_device *adev,
+					    u64 *flags)
+{
+	int data;
+
+	if (amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(4, 4, 2) ||
+	    amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(4, 4, 5)) {
+		/* Default enabled */
+		*flags |= AMD_CG_SUPPORT_HDP_MGCG;
+		return;
+	}
+	/* AMD_CG_SUPPORT_HDP_LS */
+	data = RREG32(SOC15_REG_OFFSET(HDP, 0, mmHDP_MEM_POWER_LS));
+	if (data & HDP_MEM_POWER_LS__LS_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_HDP_LS;
+}
+
+static void hdp_v4_0_init_registers(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, HDP_HWIP, 0)) {
+	case IP_VERSION(4, 2, 1):
+		WREG32_FIELD15(HDP, 0, HDP_MMHUB_CNTL, HDP_MMHUB_GCC, 1);
+		break;
+	default:
+		break;
+	}
+
+	/* Do not program registers if VF */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	WREG32_FIELD15(HDP, 0, HDP_MISC_CNTL, FLUSH_INVALIDATE_CACHE, 1);
+
+	if (amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(4, 4, 0))
+		WREG32_FIELD15(HDP, 0, HDP_MISC_CNTL, READ_BUFFER_WATERMARK, 2);
+
+	WREG32_SOC15(HDP, 0, mmHDP_NONSURFACE_BASE, (adev->gmc.vram_start >> 8));
+	WREG32_SOC15(HDP, 0, mmHDP_NONSURFACE_BASE_HI, (adev->gmc.vram_start >> 40));
+}
+
+struct amdgpu_ras_block_hw_ops hdp_v4_0_ras_hw_ops = {
+	.query_ras_error_count = hdp_v4_0_query_ras_error_count,
+	.reset_ras_error_count = hdp_v4_0_reset_ras_error_count,
+};
+
+struct amdgpu_hdp_ras hdp_v4_0_ras = {
+	.ras_block = {
+		.hw_ops = &hdp_v4_0_ras_hw_ops,
+	},
+};
+
+const struct amdgpu_hdp_funcs hdp_v4_0_funcs = {
+	.flush_hdp = amdgpu_hdp_generic_flush,
+	.invalidate_hdp = hdp_v4_0_invalidate_hdp,
+	.update_clock_gating = hdp_v4_0_update_clock_gating,
+	.get_clock_gating_state = hdp_v4_0_get_clockgating_state,
+	.init_registers = hdp_v4_0_init_registers,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.h b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.h
new file mode 100644
index 000000000000..c44eee9282ab
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __HDP_V4_0_H__
+#define __HDP_V4_0_H__
+
+#include "soc15_common.h"
+
+extern const struct amdgpu_hdp_funcs hdp_v4_0_funcs;
+extern struct amdgpu_hdp_ras  hdp_v4_0_ras;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v5_0.c b/drivers/gpu/drm/amd/amdgpu/hdp_v5_0.c
new file mode 100644
index 000000000000..8bc001dc9f63
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v5_0.c
@@ -0,0 +1,214 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "hdp_v5_0.h"
+
+#include "hdp/hdp_5_0_0_offset.h"
+#include "hdp/hdp_5_0_0_sh_mask.h"
+#include <uapi/linux/kfd_ioctl.h>
+
+static void hdp_v5_0_invalidate_hdp(struct amdgpu_device *adev,
+				    struct amdgpu_ring *ring)
+{
+	if (!ring || !ring->funcs->emit_wreg) {
+		WREG32_SOC15_NO_KIQ(HDP, 0, mmHDP_READ_CACHE_INVALIDATE, 1);
+		RREG32_SOC15_NO_KIQ(HDP, 0, mmHDP_READ_CACHE_INVALIDATE);
+	} else {
+		amdgpu_ring_emit_wreg(ring, SOC15_REG_OFFSET(
+					HDP, 0, mmHDP_READ_CACHE_INVALIDATE), 1);
+	}
+}
+
+static void hdp_v5_0_update_mem_power_gating(struct amdgpu_device *adev,
+					  bool enable)
+{
+	uint32_t hdp_clk_cntl, hdp_clk_cntl1;
+	uint32_t hdp_mem_pwr_cntl;
+
+	if (!(adev->cg_flags & (AMD_CG_SUPPORT_HDP_LS |
+				AMD_CG_SUPPORT_HDP_DS |
+				AMD_CG_SUPPORT_HDP_SD)))
+		return;
+
+	hdp_clk_cntl = hdp_clk_cntl1 = RREG32_SOC15(HDP, 0, mmHDP_CLK_CNTL);
+	hdp_mem_pwr_cntl = RREG32_SOC15(HDP, 0, mmHDP_MEM_POWER_CTRL);
+
+	/* Before doing clock/power mode switch,
+	 * forced on IPH & RC clock */
+	hdp_clk_cntl = REG_SET_FIELD(hdp_clk_cntl, HDP_CLK_CNTL,
+				     IPH_MEM_CLK_SOFT_OVERRIDE, 1);
+	hdp_clk_cntl = REG_SET_FIELD(hdp_clk_cntl, HDP_CLK_CNTL,
+				     RC_MEM_CLK_SOFT_OVERRIDE, 1);
+	WREG32_SOC15(HDP, 0, mmHDP_CLK_CNTL, hdp_clk_cntl);
+
+	/* HDP 5.0 doesn't support dynamic power mode switch,
+	 * disable clock and power gating before any changing */
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 IPH_MEM_POWER_CTRL_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 IPH_MEM_POWER_LS_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 IPH_MEM_POWER_DS_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 IPH_MEM_POWER_SD_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 RC_MEM_POWER_CTRL_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 RC_MEM_POWER_LS_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 RC_MEM_POWER_DS_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 RC_MEM_POWER_SD_EN, 0);
+	WREG32_SOC15(HDP, 0, mmHDP_MEM_POWER_CTRL, hdp_mem_pwr_cntl);
+
+	/* Already disabled above. The actions below are for "enabled" only */
+	if (enable) {
+		/* only one clock gating mode (LS/DS/SD) can be enabled */
+		if (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS) {
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 IPH_MEM_POWER_LS_EN, 1);
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 RC_MEM_POWER_LS_EN, 1);
+		} else if (adev->cg_flags & AMD_CG_SUPPORT_HDP_DS) {
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 IPH_MEM_POWER_DS_EN, 1);
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 RC_MEM_POWER_DS_EN, 1);
+		} else if (adev->cg_flags & AMD_CG_SUPPORT_HDP_SD) {
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 IPH_MEM_POWER_SD_EN, 1);
+			/* RC should not use shut down mode, fallback to ds  or ls if allowed */
+			if (adev->cg_flags & AMD_CG_SUPPORT_HDP_DS)
+				hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+								 HDP_MEM_POWER_CTRL,
+								 RC_MEM_POWER_DS_EN, 1);
+			else if (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS)
+				hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+								 HDP_MEM_POWER_CTRL,
+								 RC_MEM_POWER_LS_EN, 1);
+		}
+
+		/* confirmed that IPH_MEM_POWER_CTRL_EN and RC_MEM_POWER_CTRL_EN have to
+		 * be set for SRAM LS/DS/SD */
+		if (adev->cg_flags & (AMD_CG_SUPPORT_HDP_LS | AMD_CG_SUPPORT_HDP_DS |
+				      AMD_CG_SUPPORT_HDP_SD)) {
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+							 IPH_MEM_POWER_CTRL_EN, 1);
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+							 RC_MEM_POWER_CTRL_EN, 1);
+			WREG32_SOC15(HDP, 0, mmHDP_MEM_POWER_CTRL, hdp_mem_pwr_cntl);
+		}
+	}
+
+	/* disable IPH & RC clock override after clock/power mode changing */
+	hdp_clk_cntl = REG_SET_FIELD(hdp_clk_cntl, HDP_CLK_CNTL,
+				     IPH_MEM_CLK_SOFT_OVERRIDE, 0);
+	hdp_clk_cntl = REG_SET_FIELD(hdp_clk_cntl, HDP_CLK_CNTL,
+				     RC_MEM_CLK_SOFT_OVERRIDE, 0);
+	WREG32_SOC15(HDP, 0, mmHDP_CLK_CNTL, hdp_clk_cntl);
+}
+
+static void hdp_v5_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						      bool enable)
+{
+	uint32_t hdp_clk_cntl;
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_HDP_MGCG))
+		return;
+
+	hdp_clk_cntl = RREG32_SOC15(HDP, 0, mmHDP_CLK_CNTL);
+
+	if (enable) {
+		hdp_clk_cntl &=
+			~(uint32_t)
+			(HDP_CLK_CNTL__IPH_MEM_CLK_SOFT_OVERRIDE_MASK |
+			 HDP_CLK_CNTL__RC_MEM_CLK_SOFT_OVERRIDE_MASK |
+			 HDP_CLK_CNTL__DBUS_CLK_SOFT_OVERRIDE_MASK |
+			 HDP_CLK_CNTL__DYN_CLK_SOFT_OVERRIDE_MASK |
+			 HDP_CLK_CNTL__XDP_REG_CLK_SOFT_OVERRIDE_MASK |
+			 HDP_CLK_CNTL__HDP_REG_CLK_SOFT_OVERRIDE_MASK);
+	} else {
+		hdp_clk_cntl |= HDP_CLK_CNTL__IPH_MEM_CLK_SOFT_OVERRIDE_MASK |
+			HDP_CLK_CNTL__RC_MEM_CLK_SOFT_OVERRIDE_MASK |
+			HDP_CLK_CNTL__DBUS_CLK_SOFT_OVERRIDE_MASK |
+			HDP_CLK_CNTL__DYN_CLK_SOFT_OVERRIDE_MASK |
+			HDP_CLK_CNTL__XDP_REG_CLK_SOFT_OVERRIDE_MASK |
+			HDP_CLK_CNTL__HDP_REG_CLK_SOFT_OVERRIDE_MASK;
+	}
+
+	WREG32_SOC15(HDP, 0, mmHDP_CLK_CNTL, hdp_clk_cntl);
+}
+
+static void hdp_v5_0_update_clock_gating(struct amdgpu_device *adev,
+					      bool enable)
+{
+	hdp_v5_0_update_mem_power_gating(adev, enable);
+	hdp_v5_0_update_medium_grain_clock_gating(adev, enable);
+}
+
+static void hdp_v5_0_get_clockgating_state(struct amdgpu_device *adev,
+					    u64 *flags)
+{
+	uint32_t tmp;
+
+	/* AMD_CG_SUPPORT_HDP_MGCG */
+	tmp = RREG32_SOC15(HDP, 0, mmHDP_CLK_CNTL);
+	if (!(tmp & (HDP_CLK_CNTL__IPH_MEM_CLK_SOFT_OVERRIDE_MASK |
+		     HDP_CLK_CNTL__RC_MEM_CLK_SOFT_OVERRIDE_MASK |
+		     HDP_CLK_CNTL__DBUS_CLK_SOFT_OVERRIDE_MASK |
+		     HDP_CLK_CNTL__DYN_CLK_SOFT_OVERRIDE_MASK |
+		     HDP_CLK_CNTL__XDP_REG_CLK_SOFT_OVERRIDE_MASK |
+		     HDP_CLK_CNTL__HDP_REG_CLK_SOFT_OVERRIDE_MASK)))
+		*flags |= AMD_CG_SUPPORT_HDP_MGCG;
+
+	/* AMD_CG_SUPPORT_HDP_LS/DS/SD */
+	tmp = RREG32_SOC15(HDP, 0, mmHDP_MEM_POWER_CTRL);
+	if (tmp & HDP_MEM_POWER_CTRL__IPH_MEM_POWER_LS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_HDP_LS;
+	else if (tmp & HDP_MEM_POWER_CTRL__IPH_MEM_POWER_DS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_HDP_DS;
+	else if (tmp & HDP_MEM_POWER_CTRL__IPH_MEM_POWER_SD_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_HDP_SD;
+}
+
+static void hdp_v5_0_init_registers(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	tmp = RREG32_SOC15(HDP, 0, mmHDP_MISC_CNTL);
+	tmp |= HDP_MISC_CNTL__FLUSH_INVALIDATE_CACHE_MASK;
+	WREG32_SOC15(HDP, 0, mmHDP_MISC_CNTL, tmp);
+}
+
+const struct amdgpu_hdp_funcs hdp_v5_0_funcs = {
+	.flush_hdp = amdgpu_hdp_generic_flush,
+	.invalidate_hdp = hdp_v5_0_invalidate_hdp,
+	.update_clock_gating = hdp_v5_0_update_clock_gating,
+	.get_clock_gating_state = hdp_v5_0_get_clockgating_state,
+	.init_registers = hdp_v5_0_init_registers,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v5_0.h b/drivers/gpu/drm/amd/amdgpu/hdp_v5_0.h
new file mode 100644
index 000000000000..2d5ec2b419f3
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v5_0.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __HDP_V5_0_H__
+#define __HDP_V5_0_H__
+
+#include "soc15_common.h"
+
+extern const struct amdgpu_hdp_funcs hdp_v5_0_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v5_2.c b/drivers/gpu/drm/amd/amdgpu/hdp_v5_2.c
new file mode 100644
index 000000000000..40940b4ab400
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v5_2.c
@@ -0,0 +1,206 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "hdp_v5_2.h"
+
+#include "hdp/hdp_5_2_1_offset.h"
+#include "hdp/hdp_5_2_1_sh_mask.h"
+#include <uapi/linux/kfd_ioctl.h>
+
+static void hdp_v5_2_flush_hdp(struct amdgpu_device *adev,
+				struct amdgpu_ring *ring)
+{
+	if (!ring || !ring->funcs->emit_wreg) {
+		WREG32_NO_KIQ((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2,
+			0);
+		if (amdgpu_sriov_vf(adev)) {
+			/* this is fine because SR_IOV doesn't remap the register */
+			RREG32_NO_KIQ((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2);
+		} else {
+			/* We just need to read back a register to post the write.
+			 * Reading back the remapped register causes problems on
+			 * some platforms so just read back the memory size register.
+			 */
+			if (adev->nbio.funcs->get_memsize)
+				adev->nbio.funcs->get_memsize(adev);
+		}
+	} else {
+		amdgpu_ring_emit_wreg(ring,
+			(adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2,
+			0);
+	}
+}
+
+static void hdp_v5_2_update_mem_power_gating(struct amdgpu_device *adev,
+					     bool enable)
+{
+	uint32_t hdp_clk_cntl;
+	uint32_t hdp_mem_pwr_cntl;
+
+	if (!(adev->cg_flags & (AMD_CG_SUPPORT_HDP_LS |
+				AMD_CG_SUPPORT_HDP_DS |
+				AMD_CG_SUPPORT_HDP_SD)))
+		return;
+
+	hdp_clk_cntl = RREG32_SOC15(HDP, 0, regHDP_CLK_CNTL);
+	hdp_mem_pwr_cntl = RREG32_SOC15(HDP, 0, regHDP_MEM_POWER_CTRL);
+
+	/* Before doing clock/power mode switch, forced on MEM clock */
+	hdp_clk_cntl = REG_SET_FIELD(hdp_clk_cntl, HDP_CLK_CNTL,
+				     ATOMIC_MEM_CLK_SOFT_OVERRIDE, 1);
+	hdp_clk_cntl = REG_SET_FIELD(hdp_clk_cntl, HDP_CLK_CNTL,
+				     RC_MEM_CLK_SOFT_OVERRIDE, 1);
+	WREG32_SOC15(HDP, 0, regHDP_CLK_CNTL, hdp_clk_cntl);
+
+	/* disable clock and power gating before any changing */
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 ATOMIC_MEM_POWER_CTRL_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 ATOMIC_MEM_POWER_LS_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 ATOMIC_MEM_POWER_DS_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 ATOMIC_MEM_POWER_SD_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 RC_MEM_POWER_CTRL_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 RC_MEM_POWER_LS_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 RC_MEM_POWER_DS_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 RC_MEM_POWER_SD_EN, 0);
+	WREG32_SOC15(HDP, 0, regHDP_MEM_POWER_CTRL, hdp_mem_pwr_cntl);
+
+	/* Already disabled above. The actions below are for "enabled" only */
+	if (enable) {
+		/* only one clock gating mode (LS/DS/SD) can be enabled */
+		if (adev->cg_flags & AMD_CG_SUPPORT_HDP_SD) {
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 ATOMIC_MEM_POWER_SD_EN, 1);
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 RC_MEM_POWER_SD_EN, 1);
+		} else if (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS) {
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 ATOMIC_MEM_POWER_LS_EN, 1);
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 RC_MEM_POWER_LS_EN, 1);
+		} else if (adev->cg_flags & AMD_CG_SUPPORT_HDP_DS) {
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 ATOMIC_MEM_POWER_DS_EN, 1);
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 RC_MEM_POWER_DS_EN, 1);
+		}
+
+		/* confirmed that ATOMIC/RC_MEM_POWER_CTRL_EN have to be set for SRAM LS/DS/SD */
+		if (adev->cg_flags & (AMD_CG_SUPPORT_HDP_LS | AMD_CG_SUPPORT_HDP_DS |
+				      AMD_CG_SUPPORT_HDP_SD)) {
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+							 ATOMIC_MEM_POWER_CTRL_EN, 1);
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+							 RC_MEM_POWER_CTRL_EN, 1);
+			WREG32_SOC15(HDP, 0, regHDP_MEM_POWER_CTRL, hdp_mem_pwr_cntl);
+		}
+	}
+
+	/* disable MEM clock override after clock/power mode changing */
+	hdp_clk_cntl = REG_SET_FIELD(hdp_clk_cntl, HDP_CLK_CNTL,
+				     ATOMIC_MEM_CLK_SOFT_OVERRIDE, 0);
+	hdp_clk_cntl = REG_SET_FIELD(hdp_clk_cntl, HDP_CLK_CNTL,
+				     RC_MEM_CLK_SOFT_OVERRIDE, 0);
+	WREG32_SOC15(HDP, 0, regHDP_CLK_CNTL, hdp_clk_cntl);
+}
+
+static void hdp_v5_2_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						      bool enable)
+{
+	uint32_t hdp_clk_cntl;
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_HDP_MGCG))
+		return;
+
+	hdp_clk_cntl = RREG32_SOC15(HDP, 0, regHDP_CLK_CNTL);
+
+	if (enable) {
+		hdp_clk_cntl &=
+			~(uint32_t)
+			(HDP_CLK_CNTL__ATOMIC_MEM_CLK_SOFT_OVERRIDE_MASK |
+			 HDP_CLK_CNTL__RC_MEM_CLK_SOFT_OVERRIDE_MASK |
+			 HDP_CLK_CNTL__DBUS_CLK_SOFT_OVERRIDE_MASK |
+			 HDP_CLK_CNTL__DYN_CLK_SOFT_OVERRIDE_MASK |
+			 HDP_CLK_CNTL__XDP_REG_CLK_SOFT_OVERRIDE_MASK |
+			 HDP_CLK_CNTL__HDP_REG_CLK_SOFT_OVERRIDE_MASK);
+	} else {
+		hdp_clk_cntl |= HDP_CLK_CNTL__ATOMIC_MEM_CLK_SOFT_OVERRIDE_MASK |
+			HDP_CLK_CNTL__RC_MEM_CLK_SOFT_OVERRIDE_MASK |
+			HDP_CLK_CNTL__DBUS_CLK_SOFT_OVERRIDE_MASK |
+			HDP_CLK_CNTL__DYN_CLK_SOFT_OVERRIDE_MASK |
+			HDP_CLK_CNTL__XDP_REG_CLK_SOFT_OVERRIDE_MASK |
+			HDP_CLK_CNTL__HDP_REG_CLK_SOFT_OVERRIDE_MASK;
+	}
+
+	WREG32_SOC15(HDP, 0, regHDP_CLK_CNTL, hdp_clk_cntl);
+}
+
+static void hdp_v5_2_get_clockgating_state(struct amdgpu_device *adev,
+					   u64 *flags)
+{
+	uint32_t tmp;
+
+	/* AMD_CG_SUPPORT_HDP_MGCG */
+	tmp = RREG32_SOC15(HDP, 0, regHDP_CLK_CNTL);
+	if (!(tmp & (HDP_CLK_CNTL__ATOMIC_MEM_CLK_SOFT_OVERRIDE_MASK |
+		     HDP_CLK_CNTL__RC_MEM_CLK_SOFT_OVERRIDE_MASK |
+		     HDP_CLK_CNTL__DBUS_CLK_SOFT_OVERRIDE_MASK |
+		     HDP_CLK_CNTL__DYN_CLK_SOFT_OVERRIDE_MASK |
+		     HDP_CLK_CNTL__XDP_REG_CLK_SOFT_OVERRIDE_MASK |
+		     HDP_CLK_CNTL__HDP_REG_CLK_SOFT_OVERRIDE_MASK)))
+		*flags |= AMD_CG_SUPPORT_HDP_MGCG;
+
+	/* AMD_CG_SUPPORT_HDP_LS/DS/SD */
+	tmp = RREG32_SOC15(HDP, 0, regHDP_MEM_POWER_CTRL);
+	if (tmp & HDP_MEM_POWER_CTRL__ATOMIC_MEM_POWER_LS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_HDP_LS;
+	else if (tmp & HDP_MEM_POWER_CTRL__ATOMIC_MEM_POWER_DS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_HDP_DS;
+	else if (tmp & HDP_MEM_POWER_CTRL__ATOMIC_MEM_POWER_SD_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_HDP_SD;
+}
+
+static void hdp_v5_2_update_clock_gating(struct amdgpu_device *adev,
+					      bool enable)
+{
+	hdp_v5_2_update_mem_power_gating(adev, enable);
+	hdp_v5_2_update_medium_grain_clock_gating(adev, enable);
+}
+
+const struct amdgpu_hdp_funcs hdp_v5_2_funcs = {
+	.flush_hdp = hdp_v5_2_flush_hdp,
+	.update_clock_gating = hdp_v5_2_update_clock_gating,
+	.get_clock_gating_state = hdp_v5_2_get_clockgating_state,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v5_2.h b/drivers/gpu/drm/amd/amdgpu/hdp_v5_2.h
new file mode 100644
index 000000000000..cb2abc0c80ee
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v5_2.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __HDP_V5_2_H__
+#define __HDP_V5_2_H__
+
+#include "soc15_common.h"
+
+extern const struct amdgpu_hdp_funcs hdp_v5_2_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v6_0.c b/drivers/gpu/drm/amd/amdgpu/hdp_v6_0.c
new file mode 100644
index 000000000000..ec20daf4272c
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v6_0.c
@@ -0,0 +1,144 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "hdp_v6_0.h"
+
+#include "hdp/hdp_6_0_0_offset.h"
+#include "hdp/hdp_6_0_0_sh_mask.h"
+#include <uapi/linux/kfd_ioctl.h>
+
+#define regHDP_CLK_CNTL_V6_1	0xd5
+#define regHDP_CLK_CNTL_V6_1_BASE_IDX 0
+
+static void hdp_v6_0_update_clock_gating(struct amdgpu_device *adev,
+					 bool enable)
+{
+	uint32_t hdp_clk_cntl;
+	uint32_t hdp_mem_pwr_cntl;
+
+	if (!(adev->cg_flags & (AMD_CG_SUPPORT_HDP_LS |
+				AMD_CG_SUPPORT_HDP_DS |
+				AMD_CG_SUPPORT_HDP_SD)))
+		return;
+
+	if (amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(6, 1, 0))
+		hdp_clk_cntl = RREG32_SOC15(HDP, 0, regHDP_CLK_CNTL_V6_1);
+	else
+		hdp_clk_cntl = RREG32_SOC15(HDP, 0, regHDP_CLK_CNTL);
+	hdp_mem_pwr_cntl = RREG32_SOC15(HDP, 0, regHDP_MEM_POWER_CTRL);
+
+	/* Before doing clock/power mode switch,
+	 * forced on IPH & RC clock */
+	hdp_clk_cntl = REG_SET_FIELD(hdp_clk_cntl, HDP_CLK_CNTL,
+				     RC_MEM_CLK_SOFT_OVERRIDE, 1);
+	if (amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(6, 1, 0))
+		WREG32_SOC15(HDP, 0, regHDP_CLK_CNTL_V6_1, hdp_clk_cntl);
+	else
+		WREG32_SOC15(HDP, 0, regHDP_CLK_CNTL, hdp_clk_cntl);
+
+	/* disable clock and power gating before any changing */
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 ATOMIC_MEM_POWER_CTRL_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 ATOMIC_MEM_POWER_LS_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 ATOMIC_MEM_POWER_DS_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 ATOMIC_MEM_POWER_SD_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 RC_MEM_POWER_CTRL_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 RC_MEM_POWER_LS_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 RC_MEM_POWER_DS_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 RC_MEM_POWER_SD_EN, 0);
+	WREG32_SOC15(HDP, 0, regHDP_MEM_POWER_CTRL, hdp_mem_pwr_cntl);
+
+	/* Already disabled above. The actions below are for "enabled" only */
+	if (enable) {
+		/* only one clock gating mode (LS/DS/SD) can be enabled */
+		if (adev->cg_flags & AMD_CG_SUPPORT_HDP_SD) {
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 ATOMIC_MEM_POWER_SD_EN, 1);
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 RC_MEM_POWER_SD_EN, 1);
+		} else if (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS) {
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 ATOMIC_MEM_POWER_LS_EN, 1);
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 RC_MEM_POWER_LS_EN, 1);
+		} else if (adev->cg_flags & AMD_CG_SUPPORT_HDP_DS) {
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 ATOMIC_MEM_POWER_DS_EN, 1);
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 RC_MEM_POWER_DS_EN, 1);
+		}
+
+		/* confirmed that IPH_MEM_POWER_CTRL_EN and RC_MEM_POWER_CTRL_EN have to
+		 * be set for SRAM LS/DS/SD */
+		if (adev->cg_flags & (AMD_CG_SUPPORT_HDP_LS | AMD_CG_SUPPORT_HDP_DS |
+				      AMD_CG_SUPPORT_HDP_SD)) {
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+							 ATOMIC_MEM_POWER_CTRL_EN, 1);
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+							 RC_MEM_POWER_CTRL_EN, 1);
+			WREG32_SOC15(HDP, 0, regHDP_MEM_POWER_CTRL, hdp_mem_pwr_cntl);
+		}
+	}
+
+	/* disable IPH & RC clock override after clock/power mode changing */
+	hdp_clk_cntl = REG_SET_FIELD(hdp_clk_cntl, HDP_CLK_CNTL,
+				     RC_MEM_CLK_SOFT_OVERRIDE, 0);
+	if (amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(6, 1, 0))
+		WREG32_SOC15(HDP, 0, regHDP_CLK_CNTL_V6_1, hdp_clk_cntl);
+	else
+		WREG32_SOC15(HDP, 0, regHDP_CLK_CNTL, hdp_clk_cntl);
+}
+
+static void hdp_v6_0_get_clockgating_state(struct amdgpu_device *adev,
+					    u64 *flags)
+{
+	uint32_t tmp;
+
+	/* AMD_CG_SUPPORT_HDP_LS/DS/SD */
+	tmp = RREG32_SOC15(HDP, 0, regHDP_MEM_POWER_CTRL);
+	if (tmp & HDP_MEM_POWER_CTRL__ATOMIC_MEM_POWER_LS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_HDP_LS;
+	else if (tmp & HDP_MEM_POWER_CTRL__ATOMIC_MEM_POWER_DS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_HDP_DS;
+	else if (tmp & HDP_MEM_POWER_CTRL__ATOMIC_MEM_POWER_SD_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_HDP_SD;
+}
+
+const struct amdgpu_hdp_funcs hdp_v6_0_funcs = {
+	.flush_hdp = amdgpu_hdp_generic_flush,
+	.update_clock_gating = hdp_v6_0_update_clock_gating,
+	.get_clock_gating_state = hdp_v6_0_get_clockgating_state,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v6_0.h b/drivers/gpu/drm/amd/amdgpu/hdp_v6_0.h
new file mode 100644
index 000000000000..533ecd8c0800
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v6_0.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __HDP_V6_0_H__
+#define __HDP_V6_0_H__
+
+#include "soc15_common.h"
+
+extern const struct amdgpu_hdp_funcs hdp_v6_0_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v7_0.c b/drivers/gpu/drm/amd/amdgpu/hdp_v7_0.c
new file mode 100644
index 000000000000..ed1debc03507
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v7_0.c
@@ -0,0 +1,132 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "hdp_v7_0.h"
+
+#include "hdp/hdp_7_0_0_offset.h"
+#include "hdp/hdp_7_0_0_sh_mask.h"
+#include <uapi/linux/kfd_ioctl.h>
+
+static void hdp_v7_0_update_clock_gating(struct amdgpu_device *adev,
+					 bool enable)
+{
+	uint32_t hdp_clk_cntl, hdp_clk_cntl1;
+	uint32_t hdp_mem_pwr_cntl;
+
+	if (!(adev->cg_flags & (AMD_CG_SUPPORT_HDP_LS |
+				AMD_CG_SUPPORT_HDP_DS |
+				AMD_CG_SUPPORT_HDP_SD)))
+		return;
+
+	hdp_clk_cntl = hdp_clk_cntl1 = RREG32_SOC15(HDP, 0,regHDP_CLK_CNTL);
+	hdp_mem_pwr_cntl = RREG32_SOC15(HDP, 0, regHDP_MEM_POWER_CTRL);
+
+	/* Before doing clock/power mode switch,
+	 * forced on IPH & RC clock */
+	hdp_clk_cntl = REG_SET_FIELD(hdp_clk_cntl, HDP_CLK_CNTL,
+				     RC_MEM_CLK_SOFT_OVERRIDE, 1);
+	WREG32_SOC15(HDP, 0, regHDP_CLK_CNTL, hdp_clk_cntl);
+
+	/* disable clock and power gating before any changing */
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 ATOMIC_MEM_POWER_CTRL_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 ATOMIC_MEM_POWER_LS_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 ATOMIC_MEM_POWER_DS_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 ATOMIC_MEM_POWER_SD_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 RC_MEM_POWER_CTRL_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 RC_MEM_POWER_LS_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 RC_MEM_POWER_DS_EN, 0);
+	hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+					 RC_MEM_POWER_SD_EN, 0);
+	WREG32_SOC15(HDP, 0, regHDP_MEM_POWER_CTRL, hdp_mem_pwr_cntl);
+
+	/* Already disabled above. The actions below are for "enabled" only */
+	if (enable) {
+		/* only one clock gating mode (LS/DS/SD) can be enabled */
+		if (adev->cg_flags & AMD_CG_SUPPORT_HDP_SD) {
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 ATOMIC_MEM_POWER_SD_EN, 1);
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 RC_MEM_POWER_SD_EN, 1);
+		} else if (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS) {
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 ATOMIC_MEM_POWER_LS_EN, 1);
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 RC_MEM_POWER_LS_EN, 1);
+		} else if (adev->cg_flags & AMD_CG_SUPPORT_HDP_DS) {
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 ATOMIC_MEM_POWER_DS_EN, 1);
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+							 HDP_MEM_POWER_CTRL,
+							 RC_MEM_POWER_DS_EN, 1);
+		}
+
+		/* confirmed that IPH_MEM_POWER_CTRL_EN and RC_MEM_POWER_CTRL_EN have to
+		 * be set for SRAM LS/DS/SD */
+		if (adev->cg_flags & (AMD_CG_SUPPORT_HDP_LS | AMD_CG_SUPPORT_HDP_DS |
+				      AMD_CG_SUPPORT_HDP_SD)) {
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+							 ATOMIC_MEM_POWER_CTRL_EN, 1);
+			hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+							 RC_MEM_POWER_CTRL_EN, 1);
+			WREG32_SOC15(HDP, 0, regHDP_MEM_POWER_CTRL, hdp_mem_pwr_cntl);
+		}
+	}
+
+	/* disable IPH & RC clock override after clock/power mode changing */
+	hdp_clk_cntl = REG_SET_FIELD(hdp_clk_cntl, HDP_CLK_CNTL,
+				     RC_MEM_CLK_SOFT_OVERRIDE, 0);
+	WREG32_SOC15(HDP, 0, regHDP_CLK_CNTL, hdp_clk_cntl);
+}
+
+static void hdp_v7_0_get_clockgating_state(struct amdgpu_device *adev,
+					    u64 *flags)
+{
+	uint32_t tmp;
+
+	/* AMD_CG_SUPPORT_HDP_LS/DS/SD */
+	tmp = RREG32_SOC15(HDP, 0, regHDP_MEM_POWER_CTRL);
+	if (tmp & HDP_MEM_POWER_CTRL__ATOMIC_MEM_POWER_LS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_HDP_LS;
+	else if (tmp & HDP_MEM_POWER_CTRL__ATOMIC_MEM_POWER_DS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_HDP_DS;
+	else if (tmp & HDP_MEM_POWER_CTRL__ATOMIC_MEM_POWER_SD_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_HDP_SD;
+}
+
+const struct amdgpu_hdp_funcs hdp_v7_0_funcs = {
+	.flush_hdp = amdgpu_hdp_generic_flush,
+	.update_clock_gating = hdp_v7_0_update_clock_gating,
+	.get_clock_gating_state = hdp_v7_0_get_clockgating_state,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v7_0.h b/drivers/gpu/drm/amd/amdgpu/hdp_v7_0.h
new file mode 100644
index 000000000000..25b69201402d
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v7_0.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __HDP_V7_0_H__
+#define __HDP_V7_0_H__
+
+#include "soc15_common.h"
+
+extern const struct amdgpu_hdp_funcs hdp_v7_0_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/iceland_ih.c b/drivers/gpu/drm/amd/amdgpu/iceland_ih.c
index 7a0ea27ac429..1317ede131b6 100644
--- a/drivers/gpu/drm/amd/amdgpu/iceland_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/iceland_ih.c
@@ -20,7 +20,9 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  *
  */
-#include <drm/drmP.h>
+
+#include <linux/pci.h>
+
 #include "amdgpu.h"
 #include "amdgpu_ih.h"
 #include "vid.h"
@@ -103,15 +105,15 @@ static void iceland_ih_disable_interrupts(struct amdgpu_device *adev)
  */
 static int iceland_ih_irq_init(struct amdgpu_device *adev)
 {
+	struct amdgpu_ih_ring *ih = &adev->irq.ih;
 	int rb_bufsz;
 	u32 interrupt_cntl, ih_cntl, ih_rb_cntl;
-	u64 wptr_off;
 
 	/* disable irqs */
 	iceland_ih_disable_interrupts(adev);
 
 	/* setup interrupt control */
-	WREG32(mmINTERRUPT_CNTL2, adev->dummy_page.addr >> 8);
+	WREG32(mmINTERRUPT_CNTL2, adev->dummy_page_addr >> 8);
 	interrupt_cntl = RREG32(mmINTERRUPT_CNTL);
 	/* INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
 	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
@@ -133,9 +135,8 @@ static int iceland_ih_irq_init(struct amdgpu_device *adev)
 	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_WRITEBACK_ENABLE, 1);
 
 	/* set the writeback address whether it's enabled or not */
-	wptr_off = adev->wb.gpu_addr + (adev->irq.ih.wptr_offs * 4);
-	WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(wptr_off));
-	WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(wptr_off) & 0xFF);
+	WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(ih->wptr_addr));
+	WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(ih->wptr_addr) & 0xFF);
 
 	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
 
@@ -178,6 +179,7 @@ static void iceland_ih_irq_disable(struct amdgpu_device *adev)
  * iceland_ih_get_wptr - get the IH ring buffer wptr
  *
  * @adev: amdgpu_device pointer
+ * @ih: IH ring buffer to fetch wptr
  *
  * Get the IH ring buffer wptr from either the register
  * or the writeback memory buffer (VI).  Also check for
@@ -185,74 +187,95 @@ static void iceland_ih_irq_disable(struct amdgpu_device *adev)
  * Used by cz_irq_process(VI).
  * Returns the value of the wptr.
  */
-static u32 iceland_ih_get_wptr(struct amdgpu_device *adev)
+static u32 iceland_ih_get_wptr(struct amdgpu_device *adev,
+			       struct amdgpu_ih_ring *ih)
 {
 	u32 wptr, tmp;
 
-	wptr = le32_to_cpu(adev->wb.wb[adev->irq.ih.wptr_offs]);
-
-	if (REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW)) {
-		wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
-		/* When a ring buffer overflow happen start parsing interrupt
-		 * from the last not overwritten vector (wptr + 16). Hopefully
-		 * this should allow us to catchup.
-		 */
-		dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
-			wptr, adev->irq.ih.rptr, (wptr + 16) & adev->irq.ih.ptr_mask);
-		adev->irq.ih.rptr = (wptr + 16) & adev->irq.ih.ptr_mask;
-		tmp = RREG32(mmIH_RB_CNTL);
-		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
-		WREG32(mmIH_RB_CNTL, tmp);
-	}
-	return (wptr & adev->irq.ih.ptr_mask);
+	wptr = le32_to_cpu(*ih->wptr_cpu);
+
+	if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
+		goto out;
+
+	/* Double check that the overflow wasn't already cleared. */
+	wptr = RREG32(mmIH_RB_WPTR);
+
+	if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
+		goto out;
+
+	wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
+	/* When a ring buffer overflow happen start parsing interrupt
+	 * from the last not overwritten vector (wptr + 16). Hopefully
+	 * this should allow us to catchup.
+	 */
+	dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
+		wptr, ih->rptr, (wptr + 16) & ih->ptr_mask);
+	ih->rptr = (wptr + 16) & ih->ptr_mask;
+	tmp = RREG32(mmIH_RB_CNTL);
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+	WREG32(mmIH_RB_CNTL, tmp);
+
+	/* Unset the CLEAR_OVERFLOW bit immediately so new overflows
+	 * can be detected.
+	 */
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 0);
+	WREG32(mmIH_RB_CNTL, tmp);
+
+out:
+	return (wptr & ih->ptr_mask);
 }
 
 /**
  * iceland_ih_decode_iv - decode an interrupt vector
  *
  * @adev: amdgpu_device pointer
+ * @ih: IH ring buffer to decode
+ * @entry: IV entry to place decoded information into
  *
  * Decodes the interrupt vector at the current rptr
  * position and also advance the position.
  */
 static void iceland_ih_decode_iv(struct amdgpu_device *adev,
+				 struct amdgpu_ih_ring *ih,
 				 struct amdgpu_iv_entry *entry)
 {
 	/* wptr/rptr are in bytes! */
-	u32 ring_index = adev->irq.ih.rptr >> 2;
+	u32 ring_index = ih->rptr >> 2;
 	uint32_t dw[4];
 
-	dw[0] = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
-	dw[1] = le32_to_cpu(adev->irq.ih.ring[ring_index + 1]);
-	dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
-	dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
+	dw[0] = le32_to_cpu(ih->ring[ring_index + 0]);
+	dw[1] = le32_to_cpu(ih->ring[ring_index + 1]);
+	dw[2] = le32_to_cpu(ih->ring[ring_index + 2]);
+	dw[3] = le32_to_cpu(ih->ring[ring_index + 3]);
 
-	entry->client_id = AMDGPU_IH_CLIENTID_LEGACY;
+	entry->client_id = AMDGPU_IRQ_CLIENTID_LEGACY;
 	entry->src_id = dw[0] & 0xff;
 	entry->src_data[0] = dw[1] & 0xfffffff;
 	entry->ring_id = dw[2] & 0xff;
-	entry->vm_id = (dw[2] >> 8) & 0xff;
-	entry->pas_id = (dw[2] >> 16) & 0xffff;
+	entry->vmid = (dw[2] >> 8) & 0xff;
+	entry->pasid = (dw[2] >> 16) & 0xffff;
 
 	/* wptr/rptr are in bytes! */
-	adev->irq.ih.rptr += 16;
+	ih->rptr += 16;
 }
 
 /**
  * iceland_ih_set_rptr - set the IH ring buffer rptr
  *
  * @adev: amdgpu_device pointer
+ * @ih: IH ring buffer to set rptr
  *
  * Set the IH ring buffer rptr.
  */
-static void iceland_ih_set_rptr(struct amdgpu_device *adev)
+static void iceland_ih_set_rptr(struct amdgpu_device *adev,
+				struct amdgpu_ih_ring *ih)
 {
-	WREG32(mmIH_RB_RPTR, adev->irq.ih.rptr);
+	WREG32(mmIH_RB_RPTR, ih->rptr);
 }
 
-static int iceland_ih_early_init(void *handle)
+static int iceland_ih_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int ret;
 
 	ret = amdgpu_irq_add_domain(adev);
@@ -264,12 +287,12 @@ static int iceland_ih_early_init(void *handle)
 	return 0;
 }
 
-static int iceland_ih_sw_init(void *handle)
+static int iceland_ih_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	r = amdgpu_ih_ring_init(adev, 64 * 1024, false);
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih, 64 * 1024, false);
 	if (r)
 		return r;
 
@@ -278,55 +301,43 @@ static int iceland_ih_sw_init(void *handle)
 	return r;
 }
 
-static int iceland_ih_sw_fini(void *handle)
+static int iceland_ih_sw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	amdgpu_irq_fini(adev);
-	amdgpu_ih_ring_fini(adev);
+	amdgpu_irq_fini_sw(adev);
 	amdgpu_irq_remove_domain(adev);
 
 	return 0;
 }
 
-static int iceland_ih_hw_init(void *handle)
+static int iceland_ih_hw_init(struct amdgpu_ip_block *ip_block)
 {
-	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	r = iceland_ih_irq_init(adev);
-	if (r)
-		return r;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	return 0;
+	return iceland_ih_irq_init(adev);
 }
 
-static int iceland_ih_hw_fini(void *handle)
+static int iceland_ih_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	iceland_ih_irq_disable(adev);
+	iceland_ih_irq_disable(ip_block->adev);
 
 	return 0;
 }
 
-static int iceland_ih_suspend(void *handle)
+static int iceland_ih_suspend(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return iceland_ih_hw_fini(adev);
+	return iceland_ih_hw_fini(ip_block);
 }
 
-static int iceland_ih_resume(void *handle)
+static int iceland_ih_resume(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return iceland_ih_hw_init(adev);
+	return iceland_ih_hw_init(ip_block);
 }
 
-static bool iceland_ih_is_idle(void *handle)
+static bool iceland_ih_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 tmp = RREG32(mmSRBM_STATUS);
 
 	if (REG_GET_FIELD(tmp, SRBM_STATUS, IH_BUSY))
@@ -335,11 +346,11 @@ static bool iceland_ih_is_idle(void *handle)
 	return true;
 }
 
-static int iceland_ih_wait_for_idle(void *handle)
+static int iceland_ih_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
 	unsigned i;
 	u32 tmp;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
 		/* read MC_STATUS */
@@ -351,10 +362,10 @@ static int iceland_ih_wait_for_idle(void *handle)
 	return -ETIMEDOUT;
 }
 
-static int iceland_ih_soft_reset(void *handle)
+static int iceland_ih_soft_reset(struct amdgpu_ip_block *ip_block)
 {
 	u32 srbm_soft_reset = 0;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 tmp = RREG32(mmSRBM_STATUS);
 
 	if (tmp & SRBM_STATUS__IH_BUSY_MASK)
@@ -381,13 +392,13 @@ static int iceland_ih_soft_reset(void *handle)
 	return 0;
 }
 
-static int iceland_ih_set_clockgating_state(void *handle,
+static int iceland_ih_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
 	return 0;
 }
 
-static int iceland_ih_set_powergating_state(void *handle,
+static int iceland_ih_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	return 0;
@@ -396,7 +407,6 @@ static int iceland_ih_set_powergating_state(void *handle,
 static const struct amd_ip_funcs iceland_ih_ip_funcs = {
 	.name = "iceland_ih",
 	.early_init = iceland_ih_early_init,
-	.late_init = NULL,
 	.sw_init = iceland_ih_sw_init,
 	.sw_fini = iceland_ih_sw_fini,
 	.hw_init = iceland_ih_hw_init,
@@ -418,8 +428,7 @@ static const struct amdgpu_ih_funcs iceland_ih_funcs = {
 
 static void iceland_ih_set_interrupt_funcs(struct amdgpu_device *adev)
 {
-	if (adev->irq.ih_funcs == NULL)
-		adev->irq.ih_funcs = &iceland_ih_funcs;
+	adev->irq.ih_funcs = &iceland_ih_funcs;
 }
 
 const struct amdgpu_ip_block_version iceland_ih_ip_block =
diff --git a/drivers/gpu/drm/amd/amdgpu/ih_v6_0.c b/drivers/gpu/drm/amd/amdgpu/ih_v6_0.c
new file mode 100644
index 000000000000..333e9c30c091
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/ih_v6_0.c
@@ -0,0 +1,817 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/pci.h>
+
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+
+#include "oss/osssys_6_0_0_offset.h"
+#include "oss/osssys_6_0_0_sh_mask.h"
+
+#include "soc15_common.h"
+#include "ih_v6_0.h"
+
+#define MAX_REARM_RETRY 10
+
+static void ih_v6_0_set_interrupt_funcs(struct amdgpu_device *adev);
+
+/**
+ * ih_v6_0_init_register_offset - Initialize register offset for ih rings
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialize register offset ih rings (IH_V6_0).
+ */
+static void ih_v6_0_init_register_offset(struct amdgpu_device *adev)
+{
+	struct amdgpu_ih_regs *ih_regs;
+
+	/* ih ring 2 is removed
+	 * ih ring and ih ring 1 are available */
+	if (adev->irq.ih.ring_size) {
+		ih_regs = &adev->irq.ih.ih_regs;
+		ih_regs->ih_rb_base = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_BASE);
+		ih_regs->ih_rb_base_hi = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_BASE_HI);
+		ih_regs->ih_rb_cntl = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_CNTL);
+		ih_regs->ih_rb_wptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_WPTR);
+		ih_regs->ih_rb_rptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_RPTR);
+		ih_regs->ih_doorbell_rptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_DOORBELL_RPTR);
+		ih_regs->ih_rb_wptr_addr_lo = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_WPTR_ADDR_LO);
+		ih_regs->ih_rb_wptr_addr_hi = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_WPTR_ADDR_HI);
+		ih_regs->psp_reg_id = PSP_REG_IH_RB_CNTL;
+	}
+
+	if (adev->irq.ih1.ring_size) {
+		ih_regs = &adev->irq.ih1.ih_regs;
+		ih_regs->ih_rb_base = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_BASE_RING1);
+		ih_regs->ih_rb_base_hi = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_BASE_HI_RING1);
+		ih_regs->ih_rb_cntl = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_CNTL_RING1);
+		ih_regs->ih_rb_wptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_WPTR_RING1);
+		ih_regs->ih_rb_rptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_RPTR_RING1);
+		ih_regs->ih_doorbell_rptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_DOORBELL_RPTR_RING1);
+		ih_regs->psp_reg_id = PSP_REG_IH_RB_CNTL_RING1;
+	}
+}
+
+/**
+ * force_update_wptr_for_self_int - Force update the wptr for self interrupt
+ *
+ * @adev: amdgpu_device pointer
+ * @threshold: threshold to trigger the wptr reporting
+ * @timeout: timeout to trigger the wptr reporting
+ * @enabled: Enable/disable timeout flush mechanism
+ *
+ * threshold input range: 0 ~ 15, default 0,
+ * real_threshold = 2^threshold
+ * timeout input range: 0 ~ 20, default 8,
+ * real_timeout = (2^timeout) * 1024 / (socclk_freq)
+ *
+ * Force update wptr for self interrupt ( >= SIENNA_CICHLID).
+ */
+static void
+force_update_wptr_for_self_int(struct amdgpu_device *adev,
+			       u32 threshold, u32 timeout, bool enabled)
+{
+	u32 ih_cntl, ih_rb_cntl;
+
+	ih_cntl = RREG32_SOC15(OSSSYS, 0, regIH_CNTL2);
+	ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, regIH_RB_CNTL_RING1);
+
+	ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL2,
+				SELF_IV_FORCE_WPTR_UPDATE_TIMEOUT, timeout);
+	ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL2,
+				SELF_IV_FORCE_WPTR_UPDATE_ENABLE, enabled);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL_RING1,
+				   RB_USED_INT_THRESHOLD, threshold);
+
+	if (amdgpu_sriov_vf(adev) && amdgpu_sriov_reg_indirect_ih(adev)) {
+		if (psp_reg_program(&adev->psp, PSP_REG_IH_RB_CNTL_RING1, ih_rb_cntl))
+			return;
+	} else {
+		WREG32_SOC15(OSSSYS, 0, regIH_RB_CNTL_RING1, ih_rb_cntl);
+	}
+
+	WREG32_SOC15(OSSSYS, 0, regIH_CNTL2, ih_cntl);
+}
+
+/**
+ * ih_v6_0_toggle_ring_interrupts - toggle the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointer
+ * @enable: true - enable the interrupts, false - disable the interrupts
+ *
+ * Toggle the interrupt ring buffer (IH_V6_0)
+ */
+static int ih_v6_0_toggle_ring_interrupts(struct amdgpu_device *adev,
+					  struct amdgpu_ih_ring *ih,
+					  bool enable)
+{
+	struct amdgpu_ih_regs *ih_regs;
+	uint32_t tmp;
+
+	ih_regs = &ih->ih_regs;
+
+	tmp = RREG32(ih_regs->ih_rb_cntl);
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RB_ENABLE, (enable ? 1 : 0));
+
+	if (enable) {
+		/* Unset the CLEAR_OVERFLOW bit to make sure the next step
+		 * is switching the bit from 0 to 1
+		 */
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 0);
+		if (amdgpu_sriov_vf(adev) && amdgpu_sriov_reg_indirect_ih(adev)) {
+			if (psp_reg_program(&adev->psp, ih_regs->psp_reg_id, tmp))
+				return -ETIMEDOUT;
+		} else {
+			WREG32_NO_KIQ(ih_regs->ih_rb_cntl, tmp);
+		}
+
+		/* Clear RB_OVERFLOW bit */
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+		if (amdgpu_sriov_vf(adev) && amdgpu_sriov_reg_indirect_ih(adev)) {
+			if (psp_reg_program(&adev->psp, ih_regs->psp_reg_id, tmp))
+				return -ETIMEDOUT;
+		} else {
+			WREG32_NO_KIQ(ih_regs->ih_rb_cntl, tmp);
+		}
+
+		/* Unset the CLEAR_OVERFLOW bit immediately so new overflows
+		 * can be detected.
+		 */
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 0);
+	}
+
+	/* enable_intr field is only valid in ring0 */
+	if (ih == &adev->irq.ih)
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, ENABLE_INTR, (enable ? 1 : 0));
+
+	if (amdgpu_sriov_vf(adev) && amdgpu_sriov_reg_indirect_ih(adev)) {
+		if (psp_reg_program(&adev->psp, ih_regs->psp_reg_id, tmp))
+			return -ETIMEDOUT;
+	} else {
+		WREG32(ih_regs->ih_rb_cntl, tmp);
+	}
+
+	if (enable) {
+		ih->enabled = true;
+	} else {
+		/* set rptr, wptr to 0 */
+		WREG32(ih_regs->ih_rb_rptr, 0);
+		WREG32(ih_regs->ih_rb_wptr, 0);
+		ih->enabled = false;
+		ih->rptr = 0;
+	}
+
+	return 0;
+}
+
+/**
+ * ih_v6_0_toggle_interrupts - Toggle all the available interrupt ring buffers
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable or disable interrupt ring buffers
+ *
+ * Toggle all the available interrupt ring buffers (IH_V6_0).
+ */
+static int ih_v6_0_toggle_interrupts(struct amdgpu_device *adev, bool enable)
+{
+	struct amdgpu_ih_ring *ih[] = {&adev->irq.ih, &adev->irq.ih1};
+	int i;
+	int r;
+
+	for (i = 0; i < ARRAY_SIZE(ih); i++) {
+		if (ih[i]->ring_size) {
+			r = ih_v6_0_toggle_ring_interrupts(adev, ih[i], enable);
+			if (r)
+				return r;
+		}
+	}
+
+	return 0;
+}
+
+static uint32_t ih_v6_0_rb_cntl(struct amdgpu_ih_ring *ih, uint32_t ih_rb_cntl)
+{
+	int rb_bufsz = order_base_2(ih->ring_size / 4);
+
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   MC_SPACE, ih->use_bus_addr ? 2 : 4);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   WPTR_OVERFLOW_CLEAR, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   WPTR_OVERFLOW_ENABLE, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_SIZE, rb_bufsz);
+	/* Ring Buffer write pointer writeback. If enabled, IH_RB_WPTR register
+	 * value is written to memory
+	 */
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   WPTR_WRITEBACK_ENABLE, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_SNOOP, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_RO, 0);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_VMID, 0);
+
+	return ih_rb_cntl;
+}
+
+static uint32_t ih_v6_0_doorbell_rptr(struct amdgpu_ih_ring *ih)
+{
+	u32 ih_doorbell_rtpr = 0;
+
+	if (ih->use_doorbell) {
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+						 IH_DOORBELL_RPTR, OFFSET,
+						 ih->doorbell_index);
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+						 IH_DOORBELL_RPTR,
+						 ENABLE, 1);
+	} else {
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+						 IH_DOORBELL_RPTR,
+						 ENABLE, 0);
+	}
+	return ih_doorbell_rtpr;
+}
+
+/**
+ * ih_v6_0_enable_ring - enable an ih ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointer
+ *
+ * Enable an ih ring buffer (IH_V6_0)
+ */
+static int ih_v6_0_enable_ring(struct amdgpu_device *adev,
+				      struct amdgpu_ih_ring *ih)
+{
+	struct amdgpu_ih_regs *ih_regs;
+	uint32_t tmp;
+
+	ih_regs = &ih->ih_regs;
+
+	/* Ring Buffer base. [39:8] of 40-bit address of the beginning of the ring buffer*/
+	WREG32(ih_regs->ih_rb_base, ih->gpu_addr >> 8);
+	WREG32(ih_regs->ih_rb_base_hi, (ih->gpu_addr >> 40) & 0xff);
+
+	tmp = RREG32(ih_regs->ih_rb_cntl);
+	tmp = ih_v6_0_rb_cntl(ih, tmp);
+	if (ih == &adev->irq.ih)
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RPTR_REARM, !!adev->irq.msi_enabled);
+	if (ih == &adev->irq.ih1) {
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_ENABLE, 0);
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RB_FULL_DRAIN_ENABLE, 1);
+	}
+
+	if (amdgpu_sriov_vf(adev) && amdgpu_sriov_reg_indirect_ih(adev)) {
+		if (psp_reg_program(&adev->psp, ih_regs->psp_reg_id, tmp)) {
+			DRM_ERROR("PSP program IH_RB_CNTL failed!\n");
+			return -ETIMEDOUT;
+		}
+	} else {
+		WREG32(ih_regs->ih_rb_cntl, tmp);
+	}
+
+	if (ih == &adev->irq.ih) {
+		/* set the ih ring 0 writeback address whether it's enabled or not */
+		WREG32(ih_regs->ih_rb_wptr_addr_lo, lower_32_bits(ih->wptr_addr));
+		WREG32(ih_regs->ih_rb_wptr_addr_hi, upper_32_bits(ih->wptr_addr) & 0xFFFF);
+	}
+
+	/* set rptr, wptr to 0 */
+	WREG32(ih_regs->ih_rb_wptr, 0);
+	WREG32(ih_regs->ih_rb_rptr, 0);
+
+	WREG32(ih_regs->ih_doorbell_rptr, ih_v6_0_doorbell_rptr(ih));
+
+	return 0;
+}
+
+/**
+ * ih_v6_0_irq_init - init and enable the interrupt ring
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate a ring buffer for the interrupt controller,
+ * enable the RLC, disable interrupts, enable the IH
+ * ring buffer and enable it.
+ * Called at device load and reume.
+ * Returns 0 for success, errors for failure.
+ */
+static int ih_v6_0_irq_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_ih_ring *ih[] = {&adev->irq.ih, &adev->irq.ih1};
+	u32 ih_chicken;
+	u32 tmp;
+	int ret;
+	int i;
+
+	/* disable irqs */
+	ret = ih_v6_0_toggle_interrupts(adev, false);
+	if (ret)
+		return ret;
+
+	adev->nbio.funcs->ih_control(adev);
+
+	if (unlikely((adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) ||
+		     (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO))) {
+		if (ih[0]->use_bus_addr) {
+			ih_chicken = RREG32_SOC15(OSSSYS, 0, regIH_CHICKEN);
+			ih_chicken = REG_SET_FIELD(ih_chicken,
+					IH_CHICKEN, MC_SPACE_GPA_ENABLE, 1);
+			WREG32_SOC15(OSSSYS, 0, regIH_CHICKEN, ih_chicken);
+		}
+	}
+
+	for (i = 0; i < ARRAY_SIZE(ih); i++) {
+		if (ih[i]->ring_size) {
+			ret = ih_v6_0_enable_ring(adev, ih[i]);
+			if (ret)
+				return ret;
+		}
+		ih[i]->overflow = false;
+	}
+
+	/* update doorbell range for ih ring 0 */
+	adev->nbio.funcs->ih_doorbell_range(adev, ih[0]->use_doorbell,
+					    ih[0]->doorbell_index);
+
+	tmp = RREG32_SOC15(OSSSYS, 0, regIH_STORM_CLIENT_LIST_CNTL);
+	tmp = REG_SET_FIELD(tmp, IH_STORM_CLIENT_LIST_CNTL,
+			    CLIENT18_IS_STORM_CLIENT, 1);
+	WREG32_SOC15(OSSSYS, 0, regIH_STORM_CLIENT_LIST_CNTL, tmp);
+
+	tmp = RREG32_SOC15(OSSSYS, 0, regIH_INT_FLOOD_CNTL);
+	tmp = REG_SET_FIELD(tmp, IH_INT_FLOOD_CNTL, FLOOD_CNTL_ENABLE, 1);
+	WREG32_SOC15(OSSSYS, 0, regIH_INT_FLOOD_CNTL, tmp);
+
+	/* GC/MMHUB UTCL2 page fault interrupts are configured as
+	 * MSI storm capable interrupts by deafult. The delay is
+	 * used to avoid ISR being called too frequently
+	 * when page fault happens on several continuous page
+	 * and thus avoid MSI storm */
+	tmp = RREG32_SOC15(OSSSYS, 0, regIH_MSI_STORM_CTRL);
+	tmp = REG_SET_FIELD(tmp, IH_MSI_STORM_CTRL,
+			    DELAY, 3);
+	WREG32_SOC15(OSSSYS, 0, regIH_MSI_STORM_CTRL, tmp);
+
+	/* Redirect the interrupts to IH RB1 for dGPU */
+	if (adev->irq.ih1.ring_size) {
+		tmp = RREG32_SOC15(OSSSYS, 0, regIH_RING1_CLIENT_CFG_INDEX);
+		tmp = REG_SET_FIELD(tmp, IH_RING1_CLIENT_CFG_INDEX, INDEX, 0);
+		WREG32_SOC15(OSSSYS, 0, regIH_RING1_CLIENT_CFG_INDEX, tmp);
+
+		tmp = RREG32_SOC15(OSSSYS, 0, regIH_RING1_CLIENT_CFG_DATA);
+		tmp = REG_SET_FIELD(tmp, IH_RING1_CLIENT_CFG_DATA, CLIENT_ID, 0xa);
+		tmp = REG_SET_FIELD(tmp, IH_RING1_CLIENT_CFG_DATA, SOURCE_ID, 0x0);
+		tmp = REG_SET_FIELD(tmp, IH_RING1_CLIENT_CFG_DATA,
+				    SOURCE_ID_MATCH_ENABLE, 0x1);
+
+		WREG32_SOC15(OSSSYS, 0, regIH_RING1_CLIENT_CFG_DATA, tmp);
+	}
+
+	pci_set_master(adev->pdev);
+
+	/* enable interrupts */
+	ret = ih_v6_0_toggle_interrupts(adev, true);
+	if (ret)
+		return ret;
+	/* enable wptr force update for self int */
+	force_update_wptr_for_self_int(adev, 0, 8, true);
+
+	if (adev->irq.ih_soft.ring_size)
+		adev->irq.ih_soft.enabled = true;
+
+	return 0;
+}
+
+/**
+ * ih_v6_0_irq_disable - disable interrupts
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable interrupts on the hw.
+ */
+static void ih_v6_0_irq_disable(struct amdgpu_device *adev)
+{
+	force_update_wptr_for_self_int(adev, 0, 8, false);
+	ih_v6_0_toggle_interrupts(adev, false);
+
+	/* Wait and acknowledge irq */
+	mdelay(1);
+}
+
+/**
+ * ih_v6_0_get_wptr - get the IH ring buffer wptr
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointer
+ *
+ * Get the IH ring buffer wptr from either the register
+ * or the writeback memory buffer.  Also check for
+ * ring buffer overflow and deal with it.
+ * Returns the value of the wptr.
+ */
+static u32 ih_v6_0_get_wptr(struct amdgpu_device *adev,
+			      struct amdgpu_ih_ring *ih)
+{
+	u32 wptr, tmp;
+	struct amdgpu_ih_regs *ih_regs;
+
+	wptr = le32_to_cpu(*ih->wptr_cpu);
+	ih_regs = &ih->ih_regs;
+
+	if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
+		goto out;
+
+	wptr = RREG32_NO_KIQ(ih_regs->ih_rb_wptr);
+	if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
+		goto out;
+	if (!amdgpu_sriov_vf(adev))
+		wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
+	else
+		ih->overflow = true;
+
+	/* When a ring buffer overflow happen start parsing interrupt
+	 * from the last not overwritten vector (wptr + 32). Hopefully
+	 * this should allow us to catch up.
+	 */
+	tmp = (wptr + 32) & ih->ptr_mask;
+	dev_warn(adev->dev, "IH ring buffer overflow "
+		 "(0x%08X, 0x%08X, 0x%08X)\n",
+		 wptr, ih->rptr, tmp);
+	ih->rptr = tmp;
+
+	tmp = RREG32_NO_KIQ(ih_regs->ih_rb_cntl);
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+	WREG32_NO_KIQ(ih_regs->ih_rb_cntl, tmp);
+
+	/* Unset the CLEAR_OVERFLOW bit immediately so new overflows
+	 * can be detected.
+	 */
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 0);
+	WREG32_NO_KIQ(ih_regs->ih_rb_cntl, tmp);
+out:
+	return (wptr & ih->ptr_mask);
+}
+
+/**
+ * ih_v6_0_irq_rearm - rearm IRQ if lost
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointer
+ *
+ */
+static void ih_v6_0_irq_rearm(struct amdgpu_device *adev,
+			       struct amdgpu_ih_ring *ih)
+{
+	uint32_t v = 0;
+	uint32_t i = 0;
+	struct amdgpu_ih_regs *ih_regs;
+
+	ih_regs = &ih->ih_regs;
+
+	/* Rearm IRQ / re-write doorbell if doorbell write is lost */
+	for (i = 0; i < MAX_REARM_RETRY; i++) {
+		v = RREG32_NO_KIQ(ih_regs->ih_rb_rptr);
+		if ((v < ih->ring_size) && (v != ih->rptr))
+			WDOORBELL32(ih->doorbell_index, ih->rptr);
+		else
+			break;
+	}
+}
+
+/**
+ * ih_v6_0_set_rptr - set the IH ring buffer rptr
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointer
+ *
+ * Set the IH ring buffer rptr.
+ */
+static void ih_v6_0_set_rptr(struct amdgpu_device *adev,
+			       struct amdgpu_ih_ring *ih)
+{
+	struct amdgpu_ih_regs *ih_regs;
+
+	if (ih->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		*ih->rptr_cpu = ih->rptr;
+		WDOORBELL32(ih->doorbell_index, ih->rptr);
+
+		if (amdgpu_sriov_vf(adev))
+			ih_v6_0_irq_rearm(adev, ih);
+	} else {
+		ih_regs = &ih->ih_regs;
+		WREG32(ih_regs->ih_rb_rptr, ih->rptr);
+	}
+}
+
+/**
+ * ih_v6_0_self_irq - dispatch work for ring 1
+ *
+ * @adev: amdgpu_device pointer
+ * @source: irq source
+ * @entry: IV with WPTR update
+ *
+ * Update the WPTR from the IV and schedule work to handle the entries.
+ */
+static int ih_v6_0_self_irq(struct amdgpu_device *adev,
+			      struct amdgpu_irq_src *source,
+			      struct amdgpu_iv_entry *entry)
+{
+	uint32_t wptr = cpu_to_le32(entry->src_data[0]);
+
+	switch (entry->ring_id) {
+	case 1:
+		*adev->irq.ih1.wptr_cpu = wptr;
+		schedule_work(&adev->irq.ih1_work);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs ih_v6_0_self_irq_funcs = {
+	.process = ih_v6_0_self_irq,
+};
+
+static void ih_v6_0_set_self_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->irq.self_irq.num_types = 0;
+	adev->irq.self_irq.funcs = &ih_v6_0_self_irq_funcs;
+}
+
+static int ih_v6_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	ih_v6_0_set_interrupt_funcs(adev);
+	ih_v6_0_set_self_irq_funcs(adev);
+	return 0;
+}
+
+static int ih_v6_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+	bool use_bus_addr;
+
+	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_IH, 0,
+			      &adev->irq.self_irq);
+
+	if (r)
+		return r;
+
+	/* use gpu virtual address for ih ring
+	 * until ih_checken is programmed to allow
+	 * use bus address for ih ring by psp bl */
+	use_bus_addr = adev->firmware.load_type != AMDGPU_FW_LOAD_PSP;
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih, IH_RING_SIZE, use_bus_addr);
+	if (r)
+		return r;
+
+	adev->irq.ih.use_doorbell = true;
+	adev->irq.ih.doorbell_index = adev->doorbell_index.ih << 1;
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		r = amdgpu_ih_ring_init(adev, &adev->irq.ih1, IH_RING_SIZE,
+					use_bus_addr);
+		if (r)
+			return r;
+
+		adev->irq.ih1.use_doorbell = true;
+		adev->irq.ih1.doorbell_index = (adev->doorbell_index.ih + 1) << 1;
+	}
+
+	/* initialize ih control register offset */
+	ih_v6_0_init_register_offset(adev);
+
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih_soft, IH_SW_RING_SIZE, true);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_init(adev);
+
+	return r;
+}
+
+static int ih_v6_0_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	amdgpu_irq_fini_sw(adev);
+
+	return 0;
+}
+
+static int ih_v6_0_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	r = ih_v6_0_irq_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static int ih_v6_0_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	ih_v6_0_irq_disable(ip_block->adev);
+
+	return 0;
+}
+
+static int ih_v6_0_suspend(struct amdgpu_ip_block *ip_block)
+{
+	return ih_v6_0_hw_fini(ip_block);
+}
+
+static int ih_v6_0_resume(struct amdgpu_ip_block *ip_block)
+{
+	return ih_v6_0_hw_init(ip_block);
+}
+
+static bool ih_v6_0_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	/* todo */
+	return true;
+}
+
+static int ih_v6_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	/* todo */
+	return -ETIMEDOUT;
+}
+
+static int ih_v6_0_soft_reset(struct amdgpu_ip_block *ip_block)
+{
+	/* todo */
+	return 0;
+}
+
+static void ih_v6_0_update_clockgating_state(struct amdgpu_device *adev,
+					       bool enable)
+{
+	uint32_t data, def, field_val;
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_IH_CG) {
+		def = data = RREG32_SOC15(OSSSYS, 0, regIH_CLK_CTRL);
+		field_val = enable ? 0 : 1;
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     DBUS_MUX_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     OSSSYS_SHARE_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     LIMIT_SMN_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     DYN_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     REG_CLK_SOFT_OVERRIDE, field_val);
+		if (def != data)
+			WREG32_SOC15(OSSSYS, 0, regIH_CLK_CTRL, data);
+	}
+}
+
+static int ih_v6_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					   enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	ih_v6_0_update_clockgating_state(adev,
+				state == AMD_CG_STATE_GATE);
+	return 0;
+}
+
+static void ih_v6_0_update_ih_mem_power_gating(struct amdgpu_device *adev,
+					       bool enable)
+{
+	uint32_t ih_mem_pwr_cntl;
+
+	/* Disable ih sram power cntl before switch powergating mode */
+	ih_mem_pwr_cntl = RREG32_SOC15(OSSSYS, 0, regIH_MEM_POWER_CTRL);
+	ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+					IH_BUFFER_MEM_POWER_CTRL_EN, 0);
+	WREG32_SOC15(OSSSYS, 0, regIH_MEM_POWER_CTRL, ih_mem_pwr_cntl);
+
+	/* It is recommended to set mem powergating mode to DS mode */
+	if (enable) {
+		/* mem power mode */
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_LS_EN, 0);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_DS_EN, 1);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_SD_EN, 0);
+		/* cam mem power mode */
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_RETRY_INT_CAM_MEM_POWER_LS_EN, 0);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_RETRY_INT_CAM_MEM_POWER_DS_EN, 1);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_RETRY_INT_CAM_MEM_POWER_SD_EN, 0);
+		/* re-enable power cntl */
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_CTRL_EN, 1);
+	} else {
+		/* mem power mode */
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_LS_EN, 0);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_DS_EN, 0);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_SD_EN, 0);
+		/* cam mem power mode */
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_RETRY_INT_CAM_MEM_POWER_LS_EN, 0);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_RETRY_INT_CAM_MEM_POWER_DS_EN, 0);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_RETRY_INT_CAM_MEM_POWER_SD_EN, 0);
+		/* re-enable power cntl*/
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_CTRL_EN, 1);
+	}
+
+	WREG32_SOC15(OSSSYS, 0, regIH_MEM_POWER_CTRL, ih_mem_pwr_cntl);
+}
+
+static int ih_v6_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					 enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = (state == AMD_PG_STATE_GATE);
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_IH_SRAM_PG)
+		ih_v6_0_update_ih_mem_power_gating(adev, enable);
+
+	return 0;
+}
+
+static void ih_v6_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (!RREG32_SOC15(OSSSYS, 0, regIH_CLK_CTRL))
+		*flags |= AMD_CG_SUPPORT_IH_CG;
+}
+
+static const struct amd_ip_funcs ih_v6_0_ip_funcs = {
+	.name = "ih_v6_0",
+	.early_init = ih_v6_0_early_init,
+	.sw_init = ih_v6_0_sw_init,
+	.sw_fini = ih_v6_0_sw_fini,
+	.hw_init = ih_v6_0_hw_init,
+	.hw_fini = ih_v6_0_hw_fini,
+	.suspend = ih_v6_0_suspend,
+	.resume = ih_v6_0_resume,
+	.is_idle = ih_v6_0_is_idle,
+	.wait_for_idle = ih_v6_0_wait_for_idle,
+	.soft_reset = ih_v6_0_soft_reset,
+	.set_clockgating_state = ih_v6_0_set_clockgating_state,
+	.set_powergating_state = ih_v6_0_set_powergating_state,
+	.get_clockgating_state = ih_v6_0_get_clockgating_state,
+};
+
+static const struct amdgpu_ih_funcs ih_v6_0_funcs = {
+	.get_wptr = ih_v6_0_get_wptr,
+	.decode_iv = amdgpu_ih_decode_iv_helper,
+	.decode_iv_ts = amdgpu_ih_decode_iv_ts_helper,
+	.set_rptr = ih_v6_0_set_rptr
+};
+
+static void ih_v6_0_set_interrupt_funcs(struct amdgpu_device *adev)
+{
+	adev->irq.ih_funcs = &ih_v6_0_funcs;
+}
+
+const struct amdgpu_ip_block_version ih_v6_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_IH,
+	.major = 6,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &ih_v6_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/ih_v6_0.h b/drivers/gpu/drm/amd/amdgpu/ih_v6_0.h
new file mode 100644
index 000000000000..f27b55580716
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/ih_v6_0.h
@@ -0,0 +1,28 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __IH_V6_0_IH_H__
+#define __IH_V6_0_IH_H__
+
+extern const struct amdgpu_ip_block_version ih_v6_0_ip_block;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/ih_v6_1.c b/drivers/gpu/drm/amd/amdgpu/ih_v6_1.c
new file mode 100644
index 000000000000..95b3f4e55ec3
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/ih_v6_1.c
@@ -0,0 +1,796 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/pci.h>
+
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+
+#include "oss/osssys_6_1_0_offset.h"
+#include "oss/osssys_6_1_0_sh_mask.h"
+
+#include "soc15_common.h"
+#include "ih_v6_1.h"
+
+#define MAX_REARM_RETRY 10
+
+static void ih_v6_1_set_interrupt_funcs(struct amdgpu_device *adev);
+
+/**
+ * ih_v6_1_init_register_offset - Initialize register offset for ih rings
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialize register offset ih rings (IH_V6_0).
+ */
+static void ih_v6_1_init_register_offset(struct amdgpu_device *adev)
+{
+	struct amdgpu_ih_regs *ih_regs;
+
+	/* ih ring 2 is removed
+	 * ih ring and ih ring 1 are available */
+	if (adev->irq.ih.ring_size) {
+		ih_regs = &adev->irq.ih.ih_regs;
+		ih_regs->ih_rb_base = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_BASE);
+		ih_regs->ih_rb_base_hi = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_BASE_HI);
+		ih_regs->ih_rb_cntl = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_CNTL);
+		ih_regs->ih_rb_wptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_WPTR);
+		ih_regs->ih_rb_rptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_RPTR);
+		ih_regs->ih_doorbell_rptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_DOORBELL_RPTR);
+		ih_regs->ih_rb_wptr_addr_lo = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_WPTR_ADDR_LO);
+		ih_regs->ih_rb_wptr_addr_hi = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_WPTR_ADDR_HI);
+		ih_regs->psp_reg_id = PSP_REG_IH_RB_CNTL;
+	}
+
+	if (adev->irq.ih1.ring_size) {
+		ih_regs = &adev->irq.ih1.ih_regs;
+		ih_regs->ih_rb_base = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_BASE_RING1);
+		ih_regs->ih_rb_base_hi = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_BASE_HI_RING1);
+		ih_regs->ih_rb_cntl = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_CNTL_RING1);
+		ih_regs->ih_rb_wptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_WPTR_RING1);
+		ih_regs->ih_rb_rptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_RPTR_RING1);
+		ih_regs->ih_doorbell_rptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_DOORBELL_RPTR_RING1);
+		ih_regs->psp_reg_id = PSP_REG_IH_RB_CNTL_RING1;
+	}
+}
+
+/**
+ * force_update_wptr_for_self_int - Force update the wptr for self interrupt
+ *
+ * @adev: amdgpu_device pointer
+ * @threshold: threshold to trigger the wptr reporting
+ * @timeout: timeout to trigger the wptr reporting
+ * @enabled: Enable/disable timeout flush mechanism
+ *
+ * threshold input range: 0 ~ 15, default 0,
+ * real_threshold = 2^threshold
+ * timeout input range: 0 ~ 20, default 8,
+ * real_timeout = (2^timeout) * 1024 / (socclk_freq)
+ *
+ * Force update wptr for self interrupt ( >= SIENNA_CICHLID).
+ */
+static void
+force_update_wptr_for_self_int(struct amdgpu_device *adev,
+			       u32 threshold, u32 timeout, bool enabled)
+{
+	u32 ih_cntl, ih_rb_cntl;
+
+	ih_cntl = RREG32_SOC15(OSSSYS, 0, regIH_CNTL2);
+	ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, regIH_RB_CNTL_RING1);
+
+	ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL2,
+				SELF_IV_FORCE_WPTR_UPDATE_TIMEOUT, timeout);
+	ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL2,
+				SELF_IV_FORCE_WPTR_UPDATE_ENABLE, enabled);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL_RING1,
+				   RB_USED_INT_THRESHOLD, threshold);
+
+	if (amdgpu_sriov_vf(adev) && amdgpu_sriov_reg_indirect_ih(adev)) {
+		if (psp_reg_program(&adev->psp, PSP_REG_IH_RB_CNTL_RING1, ih_rb_cntl))
+			return;
+	} else {
+		WREG32_SOC15(OSSSYS, 0, regIH_RB_CNTL_RING1, ih_rb_cntl);
+	}
+
+	WREG32_SOC15(OSSSYS, 0, regIH_CNTL2, ih_cntl);
+}
+
+/**
+ * ih_v6_1_toggle_ring_interrupts - toggle the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointer
+ * @enable: true - enable the interrupts, false - disable the interrupts
+ *
+ * Toggle the interrupt ring buffer (IH_V6_0)
+ */
+static int ih_v6_1_toggle_ring_interrupts(struct amdgpu_device *adev,
+					  struct amdgpu_ih_ring *ih,
+					  bool enable)
+{
+	struct amdgpu_ih_regs *ih_regs;
+	uint32_t tmp;
+
+	ih_regs = &ih->ih_regs;
+
+	tmp = RREG32(ih_regs->ih_rb_cntl);
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RB_ENABLE, (enable ? 1 : 0));
+	/* enable_intr field is only valid in ring0 */
+	if (ih == &adev->irq.ih)
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, ENABLE_INTR, (enable ? 1 : 0));
+
+	if (amdgpu_sriov_vf(adev) && amdgpu_sriov_reg_indirect_ih(adev)) {
+		if (psp_reg_program(&adev->psp, ih_regs->psp_reg_id, tmp))
+			return -ETIMEDOUT;
+	} else {
+		WREG32(ih_regs->ih_rb_cntl, tmp);
+	}
+
+	if (enable) {
+		ih->enabled = true;
+	} else {
+		/* set rptr, wptr to 0 */
+		WREG32(ih_regs->ih_rb_rptr, 0);
+		WREG32(ih_regs->ih_rb_wptr, 0);
+		ih->enabled = false;
+		ih->rptr = 0;
+	}
+
+	return 0;
+}
+
+/**
+ * ih_v6_1_toggle_interrupts - Toggle all the available interrupt ring buffers
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable or disable interrupt ring buffers
+ *
+ * Toggle all the available interrupt ring buffers (IH_V6_0).
+ */
+static int ih_v6_1_toggle_interrupts(struct amdgpu_device *adev, bool enable)
+{
+	struct amdgpu_ih_ring *ih[] = {&adev->irq.ih, &adev->irq.ih1};
+	int i;
+	int r;
+
+	for (i = 0; i < ARRAY_SIZE(ih); i++) {
+		if (ih[i]->ring_size) {
+			r = ih_v6_1_toggle_ring_interrupts(adev, ih[i], enable);
+			if (r)
+				return r;
+		}
+	}
+
+	return 0;
+}
+
+static uint32_t ih_v6_1_rb_cntl(struct amdgpu_ih_ring *ih, uint32_t ih_rb_cntl)
+{
+	int rb_bufsz = order_base_2(ih->ring_size / 4);
+
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   MC_SPACE, ih->use_bus_addr ? 2 : 4);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   WPTR_OVERFLOW_CLEAR, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   WPTR_OVERFLOW_ENABLE, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_SIZE, rb_bufsz);
+	/* Ring Buffer write pointer writeback. If enabled, IH_RB_WPTR register
+	 * value is written to memory
+	 */
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   WPTR_WRITEBACK_ENABLE, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_SNOOP, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_RO, 0);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_VMID, 0);
+
+	return ih_rb_cntl;
+}
+
+static uint32_t ih_v6_1_doorbell_rptr(struct amdgpu_ih_ring *ih)
+{
+	u32 ih_doorbell_rtpr = 0;
+
+	if (ih->use_doorbell) {
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+						 IH_DOORBELL_RPTR, OFFSET,
+						 ih->doorbell_index);
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+						 IH_DOORBELL_RPTR,
+						 ENABLE, 1);
+	} else {
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+						 IH_DOORBELL_RPTR,
+						 ENABLE, 0);
+	}
+	return ih_doorbell_rtpr;
+}
+
+/**
+ * ih_v6_1_enable_ring - enable an ih ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointer
+ *
+ * Enable an ih ring buffer (IH_V6_0)
+ */
+static int ih_v6_1_enable_ring(struct amdgpu_device *adev,
+				      struct amdgpu_ih_ring *ih)
+{
+	struct amdgpu_ih_regs *ih_regs;
+	uint32_t tmp;
+
+	ih_regs = &ih->ih_regs;
+
+	/* Ring Buffer base. [39:8] of 40-bit address of the beginning of the ring buffer*/
+	WREG32(ih_regs->ih_rb_base, ih->gpu_addr >> 8);
+	WREG32(ih_regs->ih_rb_base_hi, (ih->gpu_addr >> 40) & 0xff);
+
+	tmp = RREG32(ih_regs->ih_rb_cntl);
+	tmp = ih_v6_1_rb_cntl(ih, tmp);
+	if (ih == &adev->irq.ih)
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RPTR_REARM, !!adev->irq.msi_enabled);
+	if (ih == &adev->irq.ih1) {
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_ENABLE, 0);
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RB_FULL_DRAIN_ENABLE, 1);
+	}
+
+	if (amdgpu_sriov_vf(adev) && amdgpu_sriov_reg_indirect_ih(adev)) {
+		if (psp_reg_program(&adev->psp, ih_regs->psp_reg_id, tmp)) {
+			DRM_ERROR("PSP program IH_RB_CNTL failed!\n");
+			return -ETIMEDOUT;
+		}
+	} else {
+		WREG32(ih_regs->ih_rb_cntl, tmp);
+	}
+
+	if (ih == &adev->irq.ih) {
+		/* set the ih ring 0 writeback address whether it's enabled or not */
+		WREG32(ih_regs->ih_rb_wptr_addr_lo, lower_32_bits(ih->wptr_addr));
+		WREG32(ih_regs->ih_rb_wptr_addr_hi, upper_32_bits(ih->wptr_addr) & 0xFFFF);
+	}
+
+	/* set rptr, wptr to 0 */
+	WREG32(ih_regs->ih_rb_wptr, 0);
+	WREG32(ih_regs->ih_rb_rptr, 0);
+
+	WREG32(ih_regs->ih_doorbell_rptr, ih_v6_1_doorbell_rptr(ih));
+
+	return 0;
+}
+
+/**
+ * ih_v6_1_irq_init - init and enable the interrupt ring
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate a ring buffer for the interrupt controller,
+ * enable the RLC, disable interrupts, enable the IH
+ * ring buffer and enable it.
+ * Called at device load and reume.
+ * Returns 0 for success, errors for failure.
+ */
+static int ih_v6_1_irq_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_ih_ring *ih[] = {&adev->irq.ih, &adev->irq.ih1};
+	u32 ih_chicken;
+	u32 tmp;
+	int ret;
+	int i;
+
+	/* disable irqs */
+	ret = ih_v6_1_toggle_interrupts(adev, false);
+	if (ret)
+		return ret;
+
+	adev->nbio.funcs->ih_control(adev);
+
+	if (unlikely((adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) ||
+		     (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO))) {
+		if (ih[0]->use_bus_addr) {
+			ih_chicken = RREG32_SOC15(OSSSYS, 0, regIH_CHICKEN);
+			ih_chicken = REG_SET_FIELD(ih_chicken,
+					IH_CHICKEN, MC_SPACE_GPA_ENABLE, 1);
+			WREG32_SOC15(OSSSYS, 0, regIH_CHICKEN, ih_chicken);
+		}
+	}
+
+	for (i = 0; i < ARRAY_SIZE(ih); i++) {
+		if (ih[i]->ring_size) {
+			ret = ih_v6_1_enable_ring(adev, ih[i]);
+			if (ret)
+				return ret;
+		}
+	}
+
+	/* update doorbell range for ih ring 0 */
+	adev->nbio.funcs->ih_doorbell_range(adev, ih[0]->use_doorbell,
+					    ih[0]->doorbell_index);
+
+	tmp = RREG32_SOC15(OSSSYS, 0, regIH_STORM_CLIENT_LIST_CNTL);
+	tmp = REG_SET_FIELD(tmp, IH_STORM_CLIENT_LIST_CNTL,
+			    CLIENT18_IS_STORM_CLIENT, 1);
+	WREG32_SOC15(OSSSYS, 0, regIH_STORM_CLIENT_LIST_CNTL, tmp);
+
+	tmp = RREG32_SOC15(OSSSYS, 0, regIH_INT_FLOOD_CNTL);
+	tmp = REG_SET_FIELD(tmp, IH_INT_FLOOD_CNTL, FLOOD_CNTL_ENABLE, 1);
+	WREG32_SOC15(OSSSYS, 0, regIH_INT_FLOOD_CNTL, tmp);
+
+	/* GC/MMHUB UTCL2 page fault interrupts are configured as
+	 * MSI storm capable interrupts by deafult. The delay is
+	 * used to avoid ISR being called too frequently
+	 * when page fault happens on several continuous page
+	 * and thus avoid MSI storm */
+	tmp = RREG32_SOC15(OSSSYS, 0, regIH_MSI_STORM_CTRL);
+	tmp = REG_SET_FIELD(tmp, IH_MSI_STORM_CTRL,
+			    DELAY, 3);
+	WREG32_SOC15(OSSSYS, 0, regIH_MSI_STORM_CTRL, tmp);
+
+	/* Redirect the interrupts to IH RB1 for dGPU */
+	if (adev->irq.ih1.ring_size) {
+		tmp = RREG32_SOC15(OSSSYS, 0, regIH_RING1_CLIENT_CFG_INDEX);
+		tmp = REG_SET_FIELD(tmp, IH_RING1_CLIENT_CFG_INDEX, INDEX, 0);
+		WREG32_SOC15(OSSSYS, 0, regIH_RING1_CLIENT_CFG_INDEX, tmp);
+
+		tmp = RREG32_SOC15(OSSSYS, 0, regIH_RING1_CLIENT_CFG_DATA);
+		tmp = REG_SET_FIELD(tmp, IH_RING1_CLIENT_CFG_DATA, CLIENT_ID, 0xa);
+		tmp = REG_SET_FIELD(tmp, IH_RING1_CLIENT_CFG_DATA, SOURCE_ID, 0x0);
+		tmp = REG_SET_FIELD(tmp, IH_RING1_CLIENT_CFG_DATA,
+				    SOURCE_ID_MATCH_ENABLE, 0x1);
+
+		WREG32_SOC15(OSSSYS, 0, regIH_RING1_CLIENT_CFG_DATA, tmp);
+	}
+
+	pci_set_master(adev->pdev);
+
+	/* enable interrupts */
+	ret = ih_v6_1_toggle_interrupts(adev, true);
+	if (ret)
+		return ret;
+	/* enable wptr force update for self int */
+	force_update_wptr_for_self_int(adev, 0, 8, true);
+
+	if (adev->irq.ih_soft.ring_size)
+		adev->irq.ih_soft.enabled = true;
+
+	return 0;
+}
+
+/**
+ * ih_v6_1_irq_disable - disable interrupts
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable interrupts on the hw.
+ */
+static void ih_v6_1_irq_disable(struct amdgpu_device *adev)
+{
+	force_update_wptr_for_self_int(adev, 0, 8, false);
+	ih_v6_1_toggle_interrupts(adev, false);
+
+	/* Wait and acknowledge irq */
+	mdelay(1);
+}
+
+/**
+ * ih_v6_1_get_wptr - get the IH ring buffer wptr
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointer
+ *
+ * Get the IH ring buffer wptr from either the register
+ * or the writeback memory buffer.  Also check for
+ * ring buffer overflow and deal with it.
+ * Returns the value of the wptr.
+ */
+static u32 ih_v6_1_get_wptr(struct amdgpu_device *adev,
+			      struct amdgpu_ih_ring *ih)
+{
+	u32 wptr, tmp;
+	struct amdgpu_ih_regs *ih_regs;
+
+	wptr = le32_to_cpu(*ih->wptr_cpu);
+	ih_regs = &ih->ih_regs;
+
+	if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
+		goto out;
+
+	wptr = RREG32_NO_KIQ(ih_regs->ih_rb_wptr);
+	if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
+		goto out;
+	wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
+
+	/* When a ring buffer overflow happen start parsing interrupt
+	 * from the last not overwritten vector (wptr + 32). Hopefully
+	 * this should allow us to catch up.
+	 */
+	tmp = (wptr + 32) & ih->ptr_mask;
+	dev_warn(adev->dev, "IH ring buffer overflow "
+		 "(0x%08X, 0x%08X, 0x%08X)\n",
+		 wptr, ih->rptr, tmp);
+	ih->rptr = tmp;
+
+	tmp = RREG32_NO_KIQ(ih_regs->ih_rb_cntl);
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+	WREG32_NO_KIQ(ih_regs->ih_rb_cntl, tmp);
+
+	/* Unset the CLEAR_OVERFLOW bit immediately so new overflows
+	 * can be detected.
+	 */
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 0);
+	WREG32_NO_KIQ(ih_regs->ih_rb_cntl, tmp);
+
+out:
+	return (wptr & ih->ptr_mask);
+}
+
+/**
+ * ih_v6_1_irq_rearm - rearm IRQ if lost
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointer
+ *
+ */
+static void ih_v6_1_irq_rearm(struct amdgpu_device *adev,
+			       struct amdgpu_ih_ring *ih)
+{
+	uint32_t v = 0;
+	uint32_t i = 0;
+	struct amdgpu_ih_regs *ih_regs;
+
+	ih_regs = &ih->ih_regs;
+
+	/* Rearm IRQ / re-write doorbell if doorbell write is lost */
+	for (i = 0; i < MAX_REARM_RETRY; i++) {
+		v = RREG32_NO_KIQ(ih_regs->ih_rb_rptr);
+		if ((v < ih->ring_size) && (v != ih->rptr))
+			WDOORBELL32(ih->doorbell_index, ih->rptr);
+		else
+			break;
+	}
+}
+
+/**
+ * ih_v6_1_set_rptr - set the IH ring buffer rptr
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointer
+ *
+ * Set the IH ring buffer rptr.
+ */
+static void ih_v6_1_set_rptr(struct amdgpu_device *adev,
+			       struct amdgpu_ih_ring *ih)
+{
+	struct amdgpu_ih_regs *ih_regs;
+
+	if (ih->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		*ih->rptr_cpu = ih->rptr;
+		WDOORBELL32(ih->doorbell_index, ih->rptr);
+
+		if (amdgpu_sriov_vf(adev))
+			ih_v6_1_irq_rearm(adev, ih);
+	} else {
+		ih_regs = &ih->ih_regs;
+		WREG32(ih_regs->ih_rb_rptr, ih->rptr);
+	}
+}
+
+/**
+ * ih_v6_1_self_irq - dispatch work for ring 1
+ *
+ * @adev: amdgpu_device pointer
+ * @source: irq source
+ * @entry: IV with WPTR update
+ *
+ * Update the WPTR from the IV and schedule work to handle the entries.
+ */
+static int ih_v6_1_self_irq(struct amdgpu_device *adev,
+			      struct amdgpu_irq_src *source,
+			      struct amdgpu_iv_entry *entry)
+{
+	uint32_t wptr = cpu_to_le32(entry->src_data[0]);
+
+	switch (entry->ring_id) {
+	case 1:
+		*adev->irq.ih1.wptr_cpu = wptr;
+		schedule_work(&adev->irq.ih1_work);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs ih_v6_1_self_irq_funcs = {
+	.process = ih_v6_1_self_irq,
+};
+
+static void ih_v6_1_set_self_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->irq.self_irq.num_types = 0;
+	adev->irq.self_irq.funcs = &ih_v6_1_self_irq_funcs;
+}
+
+static int ih_v6_1_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int ret;
+
+	ret = amdgpu_irq_add_domain(adev);
+	if (ret) {
+		return ret;
+	}
+
+	ih_v6_1_set_interrupt_funcs(adev);
+	ih_v6_1_set_self_irq_funcs(adev);
+	return 0;
+}
+
+static int ih_v6_1_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+	bool use_bus_addr;
+
+	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_IH, 0,
+			      &adev->irq.self_irq);
+
+	if (r)
+		return r;
+
+	/* use gpu virtual address for ih ring
+	 * until ih_checken is programmed to allow
+	 * use bus address for ih ring by psp bl */
+	use_bus_addr = adev->firmware.load_type != AMDGPU_FW_LOAD_PSP;
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih, 256 * 1024, use_bus_addr);
+	if (r)
+		return r;
+
+	adev->irq.ih.use_doorbell = true;
+	adev->irq.ih.doorbell_index = adev->doorbell_index.ih << 1;
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		r = amdgpu_ih_ring_init(adev, &adev->irq.ih1, IH_RING_SIZE,
+					use_bus_addr);
+		if (r)
+			return r;
+
+		adev->irq.ih1.use_doorbell = true;
+		adev->irq.ih1.doorbell_index = (adev->doorbell_index.ih + 1) << 1;
+	}
+
+	/* initialize ih control register offset */
+	ih_v6_1_init_register_offset(adev);
+
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih_soft, PAGE_SIZE, true);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_init(adev);
+
+	return r;
+}
+
+static int ih_v6_1_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	amdgpu_irq_fini_sw(adev);
+
+	return 0;
+}
+
+static int ih_v6_1_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	r = ih_v6_1_irq_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static int ih_v6_1_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	ih_v6_1_irq_disable(ip_block->adev);
+
+	return 0;
+}
+
+static int ih_v6_1_suspend(struct amdgpu_ip_block *ip_block)
+{
+	return ih_v6_1_hw_fini(ip_block);
+}
+
+static int ih_v6_1_resume(struct amdgpu_ip_block *ip_block)
+{
+	return ih_v6_1_hw_init(ip_block);
+}
+
+static bool ih_v6_1_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	/* todo */
+	return true;
+}
+
+static int ih_v6_1_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	/* todo */
+	return -ETIMEDOUT;
+}
+
+static int ih_v6_1_soft_reset(struct amdgpu_ip_block *ip_block)
+{
+	/* todo */
+	return 0;
+}
+
+static void ih_v6_1_update_clockgating_state(struct amdgpu_device *adev,
+					       bool enable)
+{
+	uint32_t data, def, field_val;
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_IH_CG) {
+		def = data = RREG32_SOC15(OSSSYS, 0, regIH_CLK_CTRL);
+		field_val = enable ? 0 : 1;
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     DBUS_MUX_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     OSSSYS_SHARE_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     LIMIT_SMN_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     DYN_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     REG_CLK_SOFT_OVERRIDE, field_val);
+		if (def != data)
+			WREG32_SOC15(OSSSYS, 0, regIH_CLK_CTRL, data);
+	}
+
+	return;
+}
+
+static int ih_v6_1_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					   enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	ih_v6_1_update_clockgating_state(adev,
+				state == AMD_CG_STATE_GATE);
+	return 0;
+}
+
+static void ih_v6_1_update_ih_mem_power_gating(struct amdgpu_device *adev,
+					       bool enable)
+{
+	uint32_t ih_mem_pwr_cntl;
+
+	/* Disable ih sram power cntl before switch powergating mode */
+	ih_mem_pwr_cntl = RREG32_SOC15(OSSSYS, 0, regIH_MEM_POWER_CTRL);
+	ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+					IH_BUFFER_MEM_POWER_CTRL_EN, 0);
+	WREG32_SOC15(OSSSYS, 0, regIH_MEM_POWER_CTRL, ih_mem_pwr_cntl);
+
+	/* It is recommended to set mem powergating mode to DS mode */
+	if (enable) {
+		/* mem power mode */
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_LS_EN, 0);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_DS_EN, 1);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_SD_EN, 0);
+		/* cam mem power mode */
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_RETRY_INT_CAM_MEM_POWER_LS_EN, 0);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_RETRY_INT_CAM_MEM_POWER_DS_EN, 1);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_RETRY_INT_CAM_MEM_POWER_SD_EN, 0);
+		/* re-enable power cntl */
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_CTRL_EN, 1);
+	} else {
+		/* mem power mode */
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_LS_EN, 0);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_DS_EN, 0);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_SD_EN, 0);
+		/* cam mem power mode */
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_RETRY_INT_CAM_MEM_POWER_LS_EN, 0);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_RETRY_INT_CAM_MEM_POWER_DS_EN, 0);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_RETRY_INT_CAM_MEM_POWER_SD_EN, 0);
+		/* re-enable power cntl*/
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_CTRL_EN, 1);
+	}
+
+	WREG32_SOC15(OSSSYS, 0, regIH_MEM_POWER_CTRL, ih_mem_pwr_cntl);
+}
+
+static int ih_v6_1_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					 enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = (state == AMD_PG_STATE_GATE);
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_IH_SRAM_PG)
+		ih_v6_1_update_ih_mem_power_gating(adev, enable);
+
+	return 0;
+}
+
+static void ih_v6_1_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (!RREG32_SOC15(OSSSYS, 0, regIH_CLK_CTRL))
+		*flags |= AMD_CG_SUPPORT_IH_CG;
+
+	return;
+}
+
+static const struct amd_ip_funcs ih_v6_1_ip_funcs = {
+	.name = "ih_v6_1",
+	.early_init = ih_v6_1_early_init,
+	.sw_init = ih_v6_1_sw_init,
+	.sw_fini = ih_v6_1_sw_fini,
+	.hw_init = ih_v6_1_hw_init,
+	.hw_fini = ih_v6_1_hw_fini,
+	.suspend = ih_v6_1_suspend,
+	.resume = ih_v6_1_resume,
+	.is_idle = ih_v6_1_is_idle,
+	.wait_for_idle = ih_v6_1_wait_for_idle,
+	.soft_reset = ih_v6_1_soft_reset,
+	.set_clockgating_state = ih_v6_1_set_clockgating_state,
+	.set_powergating_state = ih_v6_1_set_powergating_state,
+	.get_clockgating_state = ih_v6_1_get_clockgating_state,
+};
+
+static const struct amdgpu_ih_funcs ih_v6_1_funcs = {
+	.get_wptr = ih_v6_1_get_wptr,
+	.decode_iv = amdgpu_ih_decode_iv_helper,
+	.decode_iv_ts = amdgpu_ih_decode_iv_ts_helper,
+	.set_rptr = ih_v6_1_set_rptr
+};
+
+static void ih_v6_1_set_interrupt_funcs(struct amdgpu_device *adev)
+{
+	adev->irq.ih_funcs = &ih_v6_1_funcs;
+}
+
+const struct amdgpu_ip_block_version ih_v6_1_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_IH,
+	.major = 6,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &ih_v6_1_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/ih_v6_1.h b/drivers/gpu/drm/amd/amdgpu/ih_v6_1.h
new file mode 100644
index 000000000000..2232bc5cbd09
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/ih_v6_1.h
@@ -0,0 +1,28 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __IH_V6_1_IH_H__
+#define __IH_V6_1_IH_H__
+
+extern const struct amdgpu_ip_block_version ih_v6_1_ip_block;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/ih_v7_0.c b/drivers/gpu/drm/amd/amdgpu/ih_v7_0.c
new file mode 100644
index 000000000000..b32ea4129c61
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/ih_v7_0.c
@@ -0,0 +1,787 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/pci.h>
+
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+
+#include "oss/osssys_7_0_0_offset.h"
+#include "oss/osssys_7_0_0_sh_mask.h"
+
+#include "soc15_common.h"
+#include "ih_v7_0.h"
+
+#define MAX_REARM_RETRY 10
+
+static void ih_v7_0_set_interrupt_funcs(struct amdgpu_device *adev);
+
+/**
+ * ih_v7_0_init_register_offset - Initialize register offset for ih rings
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialize register offset ih rings (IH_V7_0).
+ */
+static void ih_v7_0_init_register_offset(struct amdgpu_device *adev)
+{
+	struct amdgpu_ih_regs *ih_regs;
+
+	/* ih ring 2 is removed
+	 * ih ring and ih ring 1 are available */
+	if (adev->irq.ih.ring_size) {
+		ih_regs = &adev->irq.ih.ih_regs;
+		ih_regs->ih_rb_base = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_BASE);
+		ih_regs->ih_rb_base_hi = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_BASE_HI);
+		ih_regs->ih_rb_cntl = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_CNTL);
+		ih_regs->ih_rb_wptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_WPTR);
+		ih_regs->ih_rb_rptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_RPTR);
+		ih_regs->ih_doorbell_rptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_DOORBELL_RPTR);
+		ih_regs->ih_rb_wptr_addr_lo = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_WPTR_ADDR_LO);
+		ih_regs->ih_rb_wptr_addr_hi = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_WPTR_ADDR_HI);
+		ih_regs->psp_reg_id = PSP_REG_IH_RB_CNTL;
+	}
+
+	if (adev->irq.ih1.ring_size) {
+		ih_regs = &adev->irq.ih1.ih_regs;
+		ih_regs->ih_rb_base = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_BASE_RING1);
+		ih_regs->ih_rb_base_hi = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_BASE_HI_RING1);
+		ih_regs->ih_rb_cntl = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_CNTL_RING1);
+		ih_regs->ih_rb_wptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_WPTR_RING1);
+		ih_regs->ih_rb_rptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_RPTR_RING1);
+		ih_regs->ih_doorbell_rptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_DOORBELL_RPTR_RING1);
+		ih_regs->psp_reg_id = PSP_REG_IH_RB_CNTL_RING1;
+	}
+}
+
+/**
+ * force_update_wptr_for_self_int - Force update the wptr for self interrupt
+ *
+ * @adev: amdgpu_device pointer
+ * @threshold: threshold to trigger the wptr reporting
+ * @timeout: timeout to trigger the wptr reporting
+ * @enabled: Enable/disable timeout flush mechanism
+ *
+ * threshold input range: 0 ~ 15, default 0,
+ * real_threshold = 2^threshold
+ * timeout input range: 0 ~ 20, default 8,
+ * real_timeout = (2^timeout) * 1024 / (socclk_freq)
+ *
+ * Force update wptr for self interrupt ( >= SIENNA_CICHLID).
+ */
+static void
+force_update_wptr_for_self_int(struct amdgpu_device *adev,
+			       u32 threshold, u32 timeout, bool enabled)
+{
+	u32 ih_cntl, ih_rb_cntl;
+
+	ih_cntl = RREG32_SOC15(OSSSYS, 0, regIH_CNTL2);
+	ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, regIH_RB_CNTL_RING1);
+
+	ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL2,
+				SELF_IV_FORCE_WPTR_UPDATE_TIMEOUT, timeout);
+	ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL2,
+				SELF_IV_FORCE_WPTR_UPDATE_ENABLE, enabled);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL_RING1,
+				   RB_USED_INT_THRESHOLD, threshold);
+
+	if (amdgpu_sriov_vf(adev) && amdgpu_sriov_reg_indirect_ih(adev)) {
+		if (psp_reg_program(&adev->psp, PSP_REG_IH_RB_CNTL_RING1, ih_rb_cntl))
+			return;
+	} else {
+		WREG32_SOC15(OSSSYS, 0, regIH_RB_CNTL_RING1, ih_rb_cntl);
+	}
+
+	WREG32_SOC15(OSSSYS, 0, regIH_CNTL2, ih_cntl);
+}
+
+/**
+ * ih_v7_0_toggle_ring_interrupts - toggle the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointet
+ * @enable: true - enable the interrupts, false - disable the interrupts
+ *
+ * Toggle the interrupt ring buffer (IH_V7_0)
+ */
+static int ih_v7_0_toggle_ring_interrupts(struct amdgpu_device *adev,
+					  struct amdgpu_ih_ring *ih,
+					  bool enable)
+{
+	struct amdgpu_ih_regs *ih_regs;
+	uint32_t tmp;
+
+	ih_regs = &ih->ih_regs;
+
+	tmp = RREG32(ih_regs->ih_rb_cntl);
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RB_ENABLE, (enable ? 1 : 0));
+	/* enable_intr field is only valid in ring0 */
+	if (ih == &adev->irq.ih)
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, ENABLE_INTR, (enable ? 1 : 0));
+
+	if (amdgpu_sriov_vf(adev) && amdgpu_sriov_reg_indirect_ih(adev)) {
+		if (psp_reg_program(&adev->psp, ih_regs->psp_reg_id, tmp))
+			return -ETIMEDOUT;
+	} else {
+		WREG32(ih_regs->ih_rb_cntl, tmp);
+	}
+
+	if (enable) {
+		ih->enabled = true;
+	} else {
+		/* set rptr, wptr to 0 */
+		WREG32(ih_regs->ih_rb_rptr, 0);
+		WREG32(ih_regs->ih_rb_wptr, 0);
+		ih->enabled = false;
+		ih->rptr = 0;
+	}
+
+	return 0;
+}
+
+/**
+ * ih_v7_0_toggle_interrupts - Toggle all the available interrupt ring buffers
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable or disable interrupt ring buffers
+ *
+ * Toggle all the available interrupt ring buffers (IH_V7_0).
+ */
+static int ih_v7_0_toggle_interrupts(struct amdgpu_device *adev, bool enable)
+{
+	struct amdgpu_ih_ring *ih[] = {&adev->irq.ih, &adev->irq.ih1};
+	int i;
+	int r;
+
+	for (i = 0; i < ARRAY_SIZE(ih); i++) {
+		if (ih[i]->ring_size) {
+			r = ih_v7_0_toggle_ring_interrupts(adev, ih[i], enable);
+			if (r)
+				return r;
+		}
+	}
+
+	return 0;
+}
+
+static uint32_t ih_v7_0_rb_cntl(struct amdgpu_ih_ring *ih, uint32_t ih_rb_cntl)
+{
+	int rb_bufsz = order_base_2(ih->ring_size / 4);
+
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   MC_SPACE, ih->use_bus_addr ? 2 : 4);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   WPTR_OVERFLOW_CLEAR, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   WPTR_OVERFLOW_ENABLE, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_SIZE, rb_bufsz);
+	/* Ring Buffer write pointer writeback. If enabled, IH_RB_WPTR register
+	 * value is written to memory
+	 */
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   WPTR_WRITEBACK_ENABLE, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_SNOOP, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_RO, 0);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_VMID, 0);
+
+	return ih_rb_cntl;
+}
+
+static uint32_t ih_v7_0_doorbell_rptr(struct amdgpu_ih_ring *ih)
+{
+	u32 ih_doorbell_rtpr = 0;
+
+	if (ih->use_doorbell) {
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+						 IH_DOORBELL_RPTR, OFFSET,
+						 ih->doorbell_index);
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+						 IH_DOORBELL_RPTR,
+						 ENABLE, 1);
+	} else {
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+						 IH_DOORBELL_RPTR,
+						 ENABLE, 0);
+	}
+	return ih_doorbell_rtpr;
+}
+
+/**
+ * ih_v7_0_enable_ring - enable an ih ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointer
+ *
+ * Enable an ih ring buffer (IH_V7_0)
+ */
+static int ih_v7_0_enable_ring(struct amdgpu_device *adev,
+				      struct amdgpu_ih_ring *ih)
+{
+	struct amdgpu_ih_regs *ih_regs;
+	uint32_t tmp;
+
+	ih_regs = &ih->ih_regs;
+
+	/* Ring Buffer base. [39:8] of 40-bit address of the beginning of the ring buffer*/
+	WREG32(ih_regs->ih_rb_base, ih->gpu_addr >> 8);
+	WREG32(ih_regs->ih_rb_base_hi, (ih->gpu_addr >> 40) & 0xff);
+
+	tmp = RREG32(ih_regs->ih_rb_cntl);
+	tmp = ih_v7_0_rb_cntl(ih, tmp);
+	if (ih == &adev->irq.ih)
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RPTR_REARM, !!adev->irq.msi_enabled);
+	if (ih == &adev->irq.ih1) {
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_ENABLE, 0);
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RB_FULL_DRAIN_ENABLE, 1);
+	}
+
+	if (amdgpu_sriov_vf(adev) && amdgpu_sriov_reg_indirect_ih(adev)) {
+		if (psp_reg_program(&adev->psp, ih_regs->psp_reg_id, tmp)) {
+			DRM_ERROR("PSP program IH_RB_CNTL failed!\n");
+			return -ETIMEDOUT;
+		}
+	} else {
+		WREG32(ih_regs->ih_rb_cntl, tmp);
+	}
+
+	if (ih == &adev->irq.ih) {
+		/* set the ih ring 0 writeback address whether it's enabled or not */
+		WREG32(ih_regs->ih_rb_wptr_addr_lo, lower_32_bits(ih->wptr_addr));
+		WREG32(ih_regs->ih_rb_wptr_addr_hi, upper_32_bits(ih->wptr_addr) & 0xFFFF);
+	}
+
+	/* set rptr, wptr to 0 */
+	WREG32(ih_regs->ih_rb_wptr, 0);
+	WREG32(ih_regs->ih_rb_rptr, 0);
+
+	WREG32(ih_regs->ih_doorbell_rptr, ih_v7_0_doorbell_rptr(ih));
+
+	return 0;
+}
+
+/**
+ * ih_v7_0_irq_init - init and enable the interrupt ring
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate a ring buffer for the interrupt controller,
+ * enable the RLC, disable interrupts, enable the IH
+ * ring buffer and enable it.
+ * Called at device load and reume.
+ * Returns 0 for success, errors for failure.
+ */
+static int ih_v7_0_irq_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_ih_ring *ih[] = {&adev->irq.ih, &adev->irq.ih1};
+	u32 ih_chicken;
+	u32 tmp;
+	int ret;
+	int i;
+
+	/* disable irqs */
+	ret = ih_v7_0_toggle_interrupts(adev, false);
+	if (ret)
+		return ret;
+
+	adev->nbio.funcs->ih_control(adev);
+
+	if (unlikely((adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) ||
+		     (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO))) {
+		if (ih[0]->use_bus_addr) {
+			ih_chicken = RREG32_SOC15(OSSSYS, 0, regIH_CHICKEN);
+			ih_chicken = REG_SET_FIELD(ih_chicken,
+					IH_CHICKEN, MC_SPACE_GPA_ENABLE, 1);
+			WREG32_SOC15(OSSSYS, 0, regIH_CHICKEN, ih_chicken);
+		}
+	}
+
+	for (i = 0; i < ARRAY_SIZE(ih); i++) {
+		if (ih[i]->ring_size) {
+			ret = ih_v7_0_enable_ring(adev, ih[i]);
+			if (ret)
+				return ret;
+		}
+	}
+
+	/* update doorbell range for ih ring 0 */
+	adev->nbio.funcs->ih_doorbell_range(adev, ih[0]->use_doorbell,
+					    ih[0]->doorbell_index);
+
+	tmp = RREG32_SOC15(OSSSYS, 0, regIH_STORM_CLIENT_LIST_CNTL);
+	tmp = REG_SET_FIELD(tmp, IH_STORM_CLIENT_LIST_CNTL,
+			    CLIENT18_IS_STORM_CLIENT, 1);
+	WREG32_SOC15(OSSSYS, 0, regIH_STORM_CLIENT_LIST_CNTL, tmp);
+
+	tmp = RREG32_SOC15(OSSSYS, 0, regIH_INT_FLOOD_CNTL);
+	tmp = REG_SET_FIELD(tmp, IH_INT_FLOOD_CNTL, FLOOD_CNTL_ENABLE, 1);
+	WREG32_SOC15(OSSSYS, 0, regIH_INT_FLOOD_CNTL, tmp);
+
+	/* GC/MMHUB UTCL2 page fault interrupts are configured as
+	 * MSI storm capable interrupts by deafult. The delay is
+	 * used to avoid ISR being called too frequently
+	 * when page fault happens on several continuous page
+	 * and thus avoid MSI storm */
+	tmp = RREG32_SOC15(OSSSYS, 0, regIH_MSI_STORM_CTRL);
+	tmp = REG_SET_FIELD(tmp, IH_MSI_STORM_CTRL,
+			    DELAY, 3);
+	WREG32_SOC15(OSSSYS, 0, regIH_MSI_STORM_CTRL, tmp);
+
+	/* Redirect the interrupts to IH RB1 for dGPU */
+	if (adev->irq.ih1.ring_size) {
+		tmp = RREG32_SOC15(OSSSYS, 0, regIH_RING1_CLIENT_CFG_INDEX);
+		tmp = REG_SET_FIELD(tmp, IH_RING1_CLIENT_CFG_INDEX, INDEX, 0);
+		WREG32_SOC15(OSSSYS, 0, regIH_RING1_CLIENT_CFG_INDEX, tmp);
+
+		tmp = RREG32_SOC15(OSSSYS, 0, regIH_RING1_CLIENT_CFG_DATA);
+		tmp = REG_SET_FIELD(tmp, IH_RING1_CLIENT_CFG_DATA, CLIENT_ID, 0xa);
+		tmp = REG_SET_FIELD(tmp, IH_RING1_CLIENT_CFG_DATA, SOURCE_ID, 0x0);
+		tmp = REG_SET_FIELD(tmp, IH_RING1_CLIENT_CFG_DATA,
+				    SOURCE_ID_MATCH_ENABLE, 0x1);
+
+		WREG32_SOC15(OSSSYS, 0, regIH_RING1_CLIENT_CFG_DATA, tmp);
+	}
+
+	pci_set_master(adev->pdev);
+
+	/* enable interrupts */
+	ret = ih_v7_0_toggle_interrupts(adev, true);
+	if (ret)
+		return ret;
+	/* enable wptr force update for self int */
+	force_update_wptr_for_self_int(adev, 0, 8, true);
+
+	if (adev->irq.ih_soft.ring_size)
+		adev->irq.ih_soft.enabled = true;
+
+	return 0;
+}
+
+/**
+ * ih_v7_0_irq_disable - disable interrupts
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable interrupts on the hw.
+ */
+static void ih_v7_0_irq_disable(struct amdgpu_device *adev)
+{
+	force_update_wptr_for_self_int(adev, 0, 8, false);
+	ih_v7_0_toggle_interrupts(adev, false);
+
+	/* Wait and acknowledge irq */
+	mdelay(1);
+}
+
+/**
+ * ih_v7_0_get_wptr() - get the IH ring buffer wptr
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: IH ring buffer to fetch wptr
+ *
+ * Get the IH ring buffer wptr from either the register
+ * or the writeback memory buffer.  Also check for
+ * ring buffer overflow and deal with it.
+ * Returns the value of the wptr.
+ */
+static u32 ih_v7_0_get_wptr(struct amdgpu_device *adev,
+			      struct amdgpu_ih_ring *ih)
+{
+	u32 wptr, tmp;
+	struct amdgpu_ih_regs *ih_regs;
+
+	wptr = le32_to_cpu(*ih->wptr_cpu);
+	ih_regs = &ih->ih_regs;
+
+	if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
+		goto out;
+
+	wptr = RREG32_NO_KIQ(ih_regs->ih_rb_wptr);
+	if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
+		goto out;
+	wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
+
+	/* When a ring buffer overflow happen start parsing interrupt
+	 * from the last not overwritten vector (wptr + 32). Hopefully
+	 * this should allow us to catch up.
+	 */
+	tmp = (wptr + 32) & ih->ptr_mask;
+	dev_warn(adev->dev, "IH ring buffer overflow "
+		 "(0x%08X, 0x%08X, 0x%08X)\n",
+		 wptr, ih->rptr, tmp);
+	ih->rptr = tmp;
+
+	tmp = RREG32_NO_KIQ(ih_regs->ih_rb_cntl);
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+	WREG32_NO_KIQ(ih_regs->ih_rb_cntl, tmp);
+
+	/* Unset the CLEAR_OVERFLOW bit immediately so new overflows
+	 * can be detected.
+	 */
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 0);
+	WREG32_NO_KIQ(ih_regs->ih_rb_cntl, tmp);
+out:
+	return (wptr & ih->ptr_mask);
+}
+
+/**
+ * ih_v7_0_irq_rearm - rearm IRQ if lost
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: IH ring to match
+ *
+ */
+static void ih_v7_0_irq_rearm(struct amdgpu_device *adev,
+			       struct amdgpu_ih_ring *ih)
+{
+	uint32_t v = 0;
+	uint32_t i = 0;
+	struct amdgpu_ih_regs *ih_regs;
+
+	ih_regs = &ih->ih_regs;
+
+	/* Rearm IRQ / re-write doorbell if doorbell write is lost */
+	for (i = 0; i < MAX_REARM_RETRY; i++) {
+		v = RREG32_NO_KIQ(ih_regs->ih_rb_rptr);
+		if ((v < ih->ring_size) && (v != ih->rptr))
+			WDOORBELL32(ih->doorbell_index, ih->rptr);
+		else
+			break;
+	}
+}
+
+/**
+ * ih_v7_0_set_rptr - set the IH ring buffer rptr
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: IH ring buffer to set rptr
+ */
+static void ih_v7_0_set_rptr(struct amdgpu_device *adev,
+			       struct amdgpu_ih_ring *ih)
+{
+	struct amdgpu_ih_regs *ih_regs;
+
+	if (ih->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		*ih->rptr_cpu = ih->rptr;
+		WDOORBELL32(ih->doorbell_index, ih->rptr);
+
+		if (amdgpu_sriov_vf(adev))
+			ih_v7_0_irq_rearm(adev, ih);
+	} else {
+		ih_regs = &ih->ih_regs;
+		WREG32(ih_regs->ih_rb_rptr, ih->rptr);
+	}
+}
+
+/**
+ * ih_v7_0_self_irq - dispatch work for ring 1
+ *
+ * @adev: amdgpu_device pointer
+ * @source: irq source
+ * @entry: IV with WPTR update
+ *
+ * Update the WPTR from the IV and schedule work to handle the entries.
+ */
+static int ih_v7_0_self_irq(struct amdgpu_device *adev,
+			      struct amdgpu_irq_src *source,
+			      struct amdgpu_iv_entry *entry)
+{
+	uint32_t wptr = cpu_to_le32(entry->src_data[0]);
+
+	switch (entry->ring_id) {
+	case 1:
+		*adev->irq.ih1.wptr_cpu = wptr;
+		schedule_work(&adev->irq.ih1_work);
+		break;
+	default: break;
+	}
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs ih_v7_0_self_irq_funcs = {
+	.process = ih_v7_0_self_irq,
+};
+
+static void ih_v7_0_set_self_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->irq.self_irq.num_types = 0;
+	adev->irq.self_irq.funcs = &ih_v7_0_self_irq_funcs;
+}
+
+static int ih_v7_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	ih_v7_0_set_interrupt_funcs(adev);
+	ih_v7_0_set_self_irq_funcs(adev);
+	return 0;
+}
+
+static int ih_v7_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+	bool use_bus_addr;
+
+	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_IH, 0,
+			      &adev->irq.self_irq);
+
+	if (r)
+		return r;
+
+	/* use gpu virtual address for ih ring
+	 * until ih_checken is programmed to allow
+	 * use bus address for ih ring by psp bl */
+	use_bus_addr = adev->firmware.load_type != AMDGPU_FW_LOAD_PSP;
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih, 256 * 1024, use_bus_addr);
+	if (r)
+		return r;
+
+	adev->irq.ih.use_doorbell = true;
+	adev->irq.ih.doorbell_index = adev->doorbell_index.ih << 1;
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		r = amdgpu_ih_ring_init(adev, &adev->irq.ih1, IH_RING_SIZE,
+					use_bus_addr);
+		if (r)
+			return r;
+
+		adev->irq.ih1.use_doorbell = true;
+		adev->irq.ih1.doorbell_index = (adev->doorbell_index.ih + 1) << 1;
+	}
+
+	/* initialize ih control register offset */
+	ih_v7_0_init_register_offset(adev);
+
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih_soft, PAGE_SIZE, true);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_init(adev);
+
+	return r;
+}
+
+static int ih_v7_0_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	amdgpu_irq_fini_sw(adev);
+
+	return 0;
+}
+
+static int ih_v7_0_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	r = ih_v7_0_irq_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static int ih_v7_0_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	ih_v7_0_irq_disable(ip_block->adev);
+
+	return 0;
+}
+
+static int ih_v7_0_suspend(struct amdgpu_ip_block *ip_block)
+{
+	return ih_v7_0_hw_fini(ip_block);
+}
+
+static int ih_v7_0_resume(struct amdgpu_ip_block *ip_block)
+{
+	return ih_v7_0_hw_init(ip_block);
+}
+
+static bool ih_v7_0_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	/* todo */
+	return true;
+}
+
+static int ih_v7_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	/* todo */
+	return -ETIMEDOUT;
+}
+
+static int ih_v7_0_soft_reset(struct amdgpu_ip_block *ip_block)
+{
+	/* todo */
+	return 0;
+}
+
+static void ih_v7_0_update_clockgating_state(struct amdgpu_device *adev,
+					       bool enable)
+{
+	uint32_t data, def, field_val;
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_IH_CG) {
+		def = data = RREG32_SOC15(OSSSYS, 0, regIH_CLK_CTRL);
+		field_val = enable ? 0 : 1;
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     DBUS_MUX_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     OSSSYS_SHARE_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     LIMIT_SMN_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     DYN_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     REG_CLK_SOFT_OVERRIDE, field_val);
+		if (def != data)
+			WREG32_SOC15(OSSSYS, 0, regIH_CLK_CTRL, data);
+	}
+
+	return;
+}
+
+static int ih_v7_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					   enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	ih_v7_0_update_clockgating_state(adev,
+				state == AMD_CG_STATE_GATE);
+	return 0;
+}
+
+static void ih_v7_0_update_ih_mem_power_gating(struct amdgpu_device *adev,
+					       bool enable)
+{
+	uint32_t ih_mem_pwr_cntl;
+
+	/* Disable ih sram power cntl before switch powergating mode */
+	ih_mem_pwr_cntl = RREG32_SOC15(OSSSYS, 0, regIH_MEM_POWER_CTRL);
+	ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+					IH_BUFFER_MEM_POWER_CTRL_EN, 0);
+	WREG32_SOC15(OSSSYS, 0, regIH_MEM_POWER_CTRL, ih_mem_pwr_cntl);
+
+	/* It is recommended to set mem powergating mode to DS mode */
+	if (enable) {
+		/* mem power mode */
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_LS_EN, 0);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_DS_EN, 1);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_SD_EN, 0);
+		/* cam mem power mode */
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_RETRY_INT_CAM_MEM_POWER_LS_EN, 0);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_RETRY_INT_CAM_MEM_POWER_DS_EN, 1);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_RETRY_INT_CAM_MEM_POWER_SD_EN, 0);
+		/* re-enable power cntl */
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_CTRL_EN, 1);
+	} else {
+		/* mem power mode */
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_LS_EN, 0);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_DS_EN, 0);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_SD_EN, 0);
+		/* cam mem power mode */
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_RETRY_INT_CAM_MEM_POWER_LS_EN, 0);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_RETRY_INT_CAM_MEM_POWER_DS_EN, 0);
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_RETRY_INT_CAM_MEM_POWER_SD_EN, 0);
+		/* re-enable power cntl*/
+		ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+						IH_BUFFER_MEM_POWER_CTRL_EN, 1);
+	}
+
+	WREG32_SOC15(OSSSYS, 0, regIH_MEM_POWER_CTRL, ih_mem_pwr_cntl);
+}
+
+static int ih_v7_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					 enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = (state == AMD_PG_STATE_GATE);
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_IH_SRAM_PG)
+		ih_v7_0_update_ih_mem_power_gating(adev, enable);
+
+	return 0;
+}
+
+static void ih_v7_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (!RREG32_SOC15(OSSSYS, 0, regIH_CLK_CTRL))
+		*flags |= AMD_CG_SUPPORT_IH_CG;
+
+	return;
+}
+
+static const struct amd_ip_funcs ih_v7_0_ip_funcs = {
+	.name = "ih_v7_0",
+	.early_init = ih_v7_0_early_init,
+	.sw_init = ih_v7_0_sw_init,
+	.sw_fini = ih_v7_0_sw_fini,
+	.hw_init = ih_v7_0_hw_init,
+	.hw_fini = ih_v7_0_hw_fini,
+	.suspend = ih_v7_0_suspend,
+	.resume = ih_v7_0_resume,
+	.is_idle = ih_v7_0_is_idle,
+	.wait_for_idle = ih_v7_0_wait_for_idle,
+	.soft_reset = ih_v7_0_soft_reset,
+	.set_clockgating_state = ih_v7_0_set_clockgating_state,
+	.set_powergating_state = ih_v7_0_set_powergating_state,
+	.get_clockgating_state = ih_v7_0_get_clockgating_state,
+};
+
+static const struct amdgpu_ih_funcs ih_v7_0_funcs = {
+	.get_wptr = ih_v7_0_get_wptr,
+	.decode_iv = amdgpu_ih_decode_iv_helper,
+	.decode_iv_ts = amdgpu_ih_decode_iv_ts_helper,
+	.set_rptr = ih_v7_0_set_rptr
+};
+
+static void ih_v7_0_set_interrupt_funcs(struct amdgpu_device *adev)
+{
+	adev->irq.ih_funcs = &ih_v7_0_funcs;
+}
+
+const struct amdgpu_ip_block_version ih_v7_0_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_IH,
+	.major = 7,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &ih_v7_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/ih_v7_0.h b/drivers/gpu/drm/amd/amdgpu/ih_v7_0.h
new file mode 100644
index 000000000000..af9dcbc451fd
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/ih_v7_0.h
@@ -0,0 +1,28 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __IH_V7_0_IH_H__
+#define __IH_V7_0_IH_H__
+
+extern const struct amdgpu_ip_block_version ih_v7_0_ip_block;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/imu_v11_0.c b/drivers/gpu/drm/amd/amdgpu/imu_v11_0.c
new file mode 100644
index 000000000000..cc626036ed9c
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/imu_v11_0.c
@@ -0,0 +1,393 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include "amdgpu.h"
+#include "amdgpu_imu.h"
+#include "amdgpu_dpm.h"
+
+#include "imu_v11_0_3.h"
+
+#include "gc/gc_11_0_0_offset.h"
+#include "gc/gc_11_0_0_sh_mask.h"
+
+MODULE_FIRMWARE("amdgpu/gc_11_0_0_imu.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_0_imu_kicker.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_1_imu.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_2_imu.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_3_imu.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_4_imu.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_0_imu.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_1_imu.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_2_imu.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_3_imu.bin");
+
+static int imu_v11_0_init_microcode(struct amdgpu_device *adev)
+{
+	char ucode_prefix[30];
+	int err;
+	const struct imu_firmware_header_v1_0 *imu_hdr;
+	struct amdgpu_firmware_info *info = NULL;
+
+	DRM_DEBUG("\n");
+
+	amdgpu_ucode_ip_version_decode(adev, GC_HWIP, ucode_prefix, sizeof(ucode_prefix));
+	if (amdgpu_is_kicker_fw(adev))
+		err = amdgpu_ucode_request(adev, &adev->gfx.imu_fw, AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_imu_kicker.bin", ucode_prefix);
+	else
+		err = amdgpu_ucode_request(adev, &adev->gfx.imu_fw, AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_imu.bin", ucode_prefix);
+	if (err)
+		goto out;
+
+	imu_hdr = (const struct imu_firmware_header_v1_0 *)adev->gfx.imu_fw->data;
+	//adev->gfx.imu_feature_version = le32_to_cpu(imu_hdr->ucode_feature_version);
+	
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_IMU_I];
+		info->ucode_id = AMDGPU_UCODE_ID_IMU_I;
+		info->fw = adev->gfx.imu_fw;
+		adev->firmware.fw_size +=
+			ALIGN(le32_to_cpu(imu_hdr->imu_iram_ucode_size_bytes), PAGE_SIZE);
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_IMU_D];
+		info->ucode_id = AMDGPU_UCODE_ID_IMU_D;
+		info->fw = adev->gfx.imu_fw;
+		adev->firmware.fw_size +=
+			ALIGN(le32_to_cpu(imu_hdr->imu_dram_ucode_size_bytes), PAGE_SIZE);
+	} else
+		adev->gfx.imu_fw_version = le32_to_cpu(imu_hdr->header.ucode_version);
+
+out:
+	if (err) {
+		dev_err(adev->dev,
+			"gfx11: Failed to load firmware \"%s_imu.bin\"\n",
+			ucode_prefix);
+		amdgpu_ucode_release(&adev->gfx.imu_fw);
+	}
+
+	return err;
+}
+
+static int imu_v11_0_load_microcode(struct amdgpu_device *adev)
+{
+	const struct imu_firmware_header_v1_0 *hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+
+	if (!adev->gfx.imu_fw)
+		return -EINVAL;
+
+	hdr = (const struct imu_firmware_header_v1_0 *)adev->gfx.imu_fw->data;
+	//amdgpu_ucode_print_rlc_hdr(&hdr->header);
+
+	fw_data = (const __le32 *)(adev->gfx.imu_fw->data +
+			le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(hdr->imu_iram_ucode_size_bytes) / 4;
+
+	WREG32_SOC15(GC, 0, regGFX_IMU_I_RAM_ADDR, 0);
+
+	for (i = 0; i < fw_size; i++)
+		WREG32_SOC15(GC, 0, regGFX_IMU_I_RAM_DATA, le32_to_cpup(fw_data++));
+
+	WREG32_SOC15(GC, 0, regGFX_IMU_I_RAM_ADDR, adev->gfx.imu_fw_version);
+
+	fw_data = (const __le32 *)(adev->gfx.imu_fw->data +
+			le32_to_cpu(hdr->header.ucode_array_offset_bytes) +
+			le32_to_cpu(hdr->imu_iram_ucode_size_bytes));
+	fw_size = le32_to_cpu(hdr->imu_dram_ucode_size_bytes) / 4;
+
+	WREG32_SOC15(GC, 0, regGFX_IMU_D_RAM_ADDR, 0);
+
+	for (i = 0; i < fw_size; i++)
+		WREG32_SOC15(GC, 0, regGFX_IMU_D_RAM_DATA, le32_to_cpup(fw_data++));
+
+	WREG32_SOC15(GC, 0, regGFX_IMU_D_RAM_ADDR, adev->gfx.imu_fw_version);
+
+	return 0;
+}
+
+static int imu_v11_0_wait_for_reset_status(struct amdgpu_device *adev)
+{
+	int i, imu_reg_val = 0;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		imu_reg_val = RREG32_SOC15(GC, 0, regGFX_IMU_GFX_RESET_CTRL);
+		if ((imu_reg_val & 0x1f) == 0x1f)
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout) {
+		dev_err(adev->dev, "init imu: IMU start timeout\n");
+		return -ETIMEDOUT;
+	}
+
+	return 0;
+}
+
+static void imu_v11_0_setup(struct amdgpu_device *adev)
+{
+	int imu_reg_val;
+
+	//enable IMU debug mode
+	WREG32_SOC15(GC, 0, regGFX_IMU_C2PMSG_ACCESS_CTRL0, 0xffffff);
+	WREG32_SOC15(GC, 0, regGFX_IMU_C2PMSG_ACCESS_CTRL1, 0xffff);
+
+	if (adev->gfx.imu.mode == DEBUG_MODE) {
+		imu_reg_val = RREG32_SOC15(GC, 0, regGFX_IMU_C2PMSG_16);
+		imu_reg_val |= 0x1;
+		WREG32_SOC15(GC, 0, regGFX_IMU_C2PMSG_16, imu_reg_val);
+	}
+
+	//disable imu Rtavfs, SmsRepair, DfllBTC, and ClkB
+	imu_reg_val = RREG32_SOC15(GC, 0, regGFX_IMU_SCRATCH_10);
+	imu_reg_val |= 0x10007;
+	WREG32_SOC15(GC, 0, regGFX_IMU_SCRATCH_10, imu_reg_val);
+}
+
+static int imu_v11_0_start(struct amdgpu_device *adev)
+{
+	int imu_reg_val;
+
+	//Start IMU by set GFX_IMU_CORE_CTRL.CRESET = 0
+	imu_reg_val = RREG32_SOC15(GC, 0, regGFX_IMU_CORE_CTRL);
+	imu_reg_val &= 0xfffffffe;
+	WREG32_SOC15(GC, 0, regGFX_IMU_CORE_CTRL, imu_reg_val);
+
+	if (adev->flags & AMD_IS_APU)
+		amdgpu_dpm_set_gfx_power_up_by_imu(adev);
+
+	return imu_v11_0_wait_for_reset_status(adev);
+}
+
+static const struct imu_rlc_ram_golden imu_rlc_ram_golden_11[] =
+{
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_IO_RD_COMBINE_FLUSH, 0x00055555, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_IO_WR_COMBINE_FLUSH, 0x00055555, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_DRAM_COMBINE_FLUSH, 0x00555555, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_MISC2, 0x00001ffe, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_SDP_CREDITS , 0x003f3fff, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_SDP_TAG_RESERVE1, 0x00000000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_SDP_VCC_RESERVE0, 0x00041000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_SDP_VCC_RESERVE1, 0x00000000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_SDP_VCD_RESERVE0, 0x00040000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_SDP_VCD_RESERVE1, 0x00000000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_MISC, 0x00000017, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_SDP_ENABLE, 0x00000001, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_CREDITS , 0x003f3fbf, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_TAG_RESERVE0, 0x10201000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_TAG_RESERVE1, 0x00000080, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_VCC_RESERVE0, 0x1d041040, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_VCC_RESERVE1, 0x80000000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_IO_PRIORITY, 0x88888888, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_MAM_CTRL, 0x0000d800, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_ARB_FINAL, 0x000003f7, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_ENABLE, 0x00000001, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL2, 0x00020000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_APT_CNTL, 0x0000000c, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_CACHEABLE_DRAM_ADDRESS_END, 0x000fffff, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_MISC, 0x0c48bff0, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCC_GC_SA_UNIT_DISABLE, 0x00fffc01, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCC_GC_PRIM_CONFIG, 0x000fffe1, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCC_RB_BACKEND_DISABLE, 0x0fffff01, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCC_GC_SHADER_ARRAY_CONFIG, 0xfffe0001, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL, 0x00000500, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_SYSTEM_APERTURE_LOW_ADDR, 0x00000001, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_SYSTEM_APERTURE_HIGH_ADDR, 0x00000000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_LOCAL_FB_ADDRESS_START, 0x00000000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_LOCAL_FB_ADDRESS_END, 0x000fffff, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_CONTEXT0_CNTL, 0x00000000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_CONTEXT1_CNTL, 0x00000000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_NB_TOP_OF_DRAM_SLOT1, 0xff800000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_NB_LOWER_TOP_OF_DRAM2, 0x00000001, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_NB_UPPER_TOP_OF_DRAM2, 0x00000fff, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL, 0x00001ffc, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL, 0x00000501, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_CNTL, 0x00080603, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_CNTL2, 0x00000003, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_CNTL3, 0x00100003, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_CNTL5, 0x00003fe0, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_CONTEXT0_CNTL, 0x00000001, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_CONTEXT0_PER_PFVF_PTE_CACHE_FRAGMENT_SIZES, 0x00000c00, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_CONTEXT1_CNTL, 0x00000001, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_CONTEXT1_PER_PFVF_PTE_CACHE_FRAGMENT_SIZES, 0x00000c00, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGB_ADDR_CONFIG, 0x00000545, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGL2_PIPE_STEER_0, 0x13455431, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGL2_PIPE_STEER_1, 0x13455431, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGL2_PIPE_STEER_2, 0x76027602, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGL2_PIPE_STEER_3, 0x76207620, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGB_ADDR_CONFIG, 0x00000345, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCUTCL2_HARVEST_BYPASS_GROUPS, 0x0000003e, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_FB_LOCATION_BASE, 0x00006000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_FB_LOCATION_TOP, 0x000061ff, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_APT_CNTL, 0x0000000c, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_AGP_BASE, 0x00000000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_AGP_BOT, 0x00000002, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_AGP_TOP, 0x00000000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL2, 0x00020000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regSDMA0_UCODE_SELFLOAD_CONTROL, 0x00000210, 0), 
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regSDMA1_UCODE_SELFLOAD_CONTROL, 0x00000210, 0), 
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCPC_PSP_DEBUG, CPC_PSP_DEBUG__GPA_OVERRIDE_MASK, 0), 
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCPG_PSP_DEBUG, CPG_PSP_DEBUG__GPA_OVERRIDE_MASK, 0)
+};
+
+static const struct imu_rlc_ram_golden imu_rlc_ram_golden_11_0_2[] =
+{
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_MISC, 0x0c48bff0, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_CREDITS, 0x003f3fbf, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_TAG_RESERVE0, 0x10200800, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_TAG_RESERVE1, 0x00000088, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_VCC_RESERVE0, 0x1d041040, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_VCC_RESERVE1, 0x80000000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_IO_PRIORITY, 0x88888888, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_MAM_CTRL, 0x0000d800, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_ARB_FINAL, 0x000007ef, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_DRAM_PAGE_BURST, 0x20080200, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_ENABLE, 0x00000001, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_APT_CNTL, 0x0000000c, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_CACHEABLE_DRAM_ADDRESS_END, 0x000fffff, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_IO_RD_COMBINE_FLUSH, 0x00055555, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_IO_WR_COMBINE_FLUSH, 0x00055555, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_DRAM_COMBINE_FLUSH, 0x00555555, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_MISC2, 0x00001ffe, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_SDP_CREDITS, 0x003f3fff, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_SDP_TAG_RESERVE1, 0x00000000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_SDP_VCC_RESERVE0, 0x00041000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_SDP_VCC_RESERVE1, 0x00000000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_SDP_VCD_RESERVE0, 0x00040000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_SDP_VCD_RESERVE1, 0x00000000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_MISC, 0x00000017, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_SDP_ENABLE, 0x00000001, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCC_GC_SA_UNIT_DISABLE, 0x00fffc01, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCC_GC_PRIM_CONFIG, 0x000fffe1, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCC_RB_BACKEND_DISABLE, 0x00000f01, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCC_GC_SHADER_ARRAY_CONFIG, 0xfffe0001, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGL1_PIPE_STEER, 0x000000e4, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCH_PIPE_STEER, 0x000000e4, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGL2_PIPE_STEER_0, 0x01231023, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGB_ADDR_CONFIG, 0x00000243, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCUTCL2_HARVEST_BYPASS_GROUPS, 0x00000002, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL, 0x00000500, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_SYSTEM_APERTURE_LOW_ADDR, 0x00000001, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_SYSTEM_APERTURE_HIGH_ADDR, 0x00000000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_LOCAL_FB_ADDRESS_START, 0x00000000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_LOCAL_FB_ADDRESS_END, 0x000001ff, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_FB_LOCATION_BASE, 0x00006000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_FB_LOCATION_TOP, 0x000061ff, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_CONTEXT0_CNTL, 0x00000000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_CONTEXT1_CNTL, 0x00000000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_APT_CNTL, 0x0000000c, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_NB_TOP_OF_DRAM_SLOT1, 0xff800000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_NB_LOWER_TOP_OF_DRAM2, 0x00000001, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_NB_UPPER_TOP_OF_DRAM2, 0x00000fff, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_AGP_BASE, 0x00000000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_AGP_BOT, 0x00000002, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_AGP_TOP, 0x00000000, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL, 0x00001ffc, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL2, 0x00002825, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL, 0x00000501, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_CNTL, 0x00080603, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_CNTL2, 0x00000003, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_CNTL3, 0x00100003, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_CNTL5, 0x00003fe0, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_CONTEXT0_CNTL, 0x00000001, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_CONTEXT0_PER_PFVF_PTE_CACHE_FRAGMENT_SIZES, 0x00000c00, 0xe0000000),
+        IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_CONTEXT1_CNTL, 0x00000001, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_CONTEXT1_PER_PFVF_PTE_CACHE_FRAGMENT_SIZES, 0x00000c00, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regSDMA0_UCODE_SELFLOAD_CONTROL, 0x00000210, 0),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regSDMA1_UCODE_SELFLOAD_CONTROL, 0x00000210, 0),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCPC_PSP_DEBUG, CPC_PSP_DEBUG__GPA_OVERRIDE_MASK, 0),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCPG_PSP_DEBUG, CPG_PSP_DEBUG__GPA_OVERRIDE_MASK, 0)
+};
+
+static void program_imu_rlc_ram(struct amdgpu_device *adev,
+				const struct imu_rlc_ram_golden *regs,
+				const u32 array_size)
+{
+	const struct imu_rlc_ram_golden *entry;
+	u32 reg, data;
+	int i;
+
+	for (i = 0; i < array_size; ++i) {
+		entry = &regs[i];
+		reg =  adev->reg_offset[entry->hwip][entry->instance][entry->segment] + entry->reg;
+		reg |= entry->addr_mask;
+
+		data = entry->data;
+		if (entry->reg == regGCMC_VM_AGP_BASE)
+			data = 0x00ffffff;
+		else if (entry->reg == regGCMC_VM_AGP_TOP)
+			data = 0x0;
+		else if (entry->reg == regGCMC_VM_FB_LOCATION_BASE)
+			data = adev->gmc.vram_start >> 24;
+		else if (entry->reg == regGCMC_VM_FB_LOCATION_TOP)
+			data = adev->gmc.vram_end >> 24;
+
+		WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_ADDR_HIGH, 0);
+		WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_ADDR_LOW, reg);
+		WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_DATA, data);
+	}
+	//Indicate the latest entry
+	WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_ADDR_HIGH, 0);
+	WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_ADDR_LOW, 0);
+	WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_DATA, 0);
+}
+
+static void imu_v11_0_program_rlc_ram(struct amdgpu_device *adev)
+{
+	u32 reg_data;
+
+	WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_INDEX, 0x2);
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(11, 0, 0):
+		program_imu_rlc_ram(adev, imu_rlc_ram_golden_11,
+				(const u32)ARRAY_SIZE(imu_rlc_ram_golden_11));
+		break;
+	case IP_VERSION(11, 0, 2):
+		program_imu_rlc_ram(adev, imu_rlc_ram_golden_11_0_2,
+				(const u32)ARRAY_SIZE(imu_rlc_ram_golden_11_0_2));
+		break;
+	case IP_VERSION(11, 0, 3):
+		imu_v11_0_3_program_rlc_ram(adev);
+		break;
+	default:
+		BUG();
+		break;
+	}
+
+	//Indicate the contents of the RAM are valid
+	reg_data = RREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_INDEX);
+	reg_data |= GFX_IMU_RLC_RAM_INDEX__RAM_VALID_MASK;
+	WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_INDEX, reg_data);
+}
+
+const struct amdgpu_imu_funcs gfx_v11_0_imu_funcs = {
+	.init_microcode = imu_v11_0_init_microcode,
+	.load_microcode = imu_v11_0_load_microcode,
+	.setup_imu = imu_v11_0_setup,
+	.start_imu = imu_v11_0_start,
+	.program_rlc_ram = imu_v11_0_program_rlc_ram,
+	.wait_for_reset_status = imu_v11_0_wait_for_reset_status,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/imu_v11_0.h b/drivers/gpu/drm/amd/amdgpu/imu_v11_0.h
new file mode 100644
index 000000000000..e71f96fc2f06
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/imu_v11_0.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __IMU_V11_0_H__
+#define __IMU_V11_0_H__
+
+extern const struct amdgpu_imu_funcs gfx_v11_0_imu_funcs;
+
+#endif
+
diff --git a/drivers/gpu/drm/amd/amdgpu/imu_v11_0_3.c b/drivers/gpu/drm/amd/amdgpu/imu_v11_0_3.c
new file mode 100644
index 000000000000..fc69c1a29e23
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/imu_v11_0_3.c
@@ -0,0 +1,145 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "amdgpu_imu.h"
+#include "imu_v11_0_3.h"
+
+#include "gc/gc_11_0_3_offset.h"
+#include "gc/gc_11_0_3_sh_mask.h"
+
+static const struct imu_rlc_ram_golden imu_rlc_ram_golden_11_0_3[] = {
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_IO_RD_COMBINE_FLUSH, 0x00055555, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_IO_WR_COMBINE_FLUSH, 0x00055555, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_DRAM_COMBINE_FLUSH, 0x00555555, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_MISC2, 0x00001ffe, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_SDP_CREDITS, 0x003f3fff, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_SDP_TAG_RESERVE1, 0x00000000, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_SDP_VCC_RESERVE0, 0x00041000, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_SDP_VCC_RESERVE1, 0x00000000, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_SDP_VCD_RESERVE0, 0x00040000, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_SDP_VCD_RESERVE1, 0x00000000, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_MISC, 0x00000017, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGUS_SDP_ENABLE, 0x00000001, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_CREDITS, 0x003f3fbf, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_TAG_RESERVE0, 0x10200800, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_TAG_RESERVE1, 0x00000088, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_VCC_RESERVE0, 0x1d041040, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_VCC_RESERVE1, 0x80000000, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_IO_PRIORITY, 0x88888888, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_MAM_CTRL, 0x0000d800, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_ARB_FINAL, 0x000007ff, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_DRAM_PAGE_BURST, 0x20080200, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_SDP_ENABLE, 0x00000001, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL2, 0x00020000, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_APT_CNTL, 0x0000000c, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_CACHEABLE_DRAM_ADDRESS_END, 0x000fffff, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCEA_MISC, 0x0c48bff0, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCC_GC_SA_UNIT_DISABLE, 0x00fffc01, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCC_GC_PRIM_CONFIG, 0x000fffe1, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCC_RB_BACKEND_DISABLE, 0xffffff01, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCC_GC_SHADER_ARRAY_CONFIG, 0xfffe0001, 0x40000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCC_GC_SHADER_ARRAY_CONFIG, 0xfffe0001, 0x42000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCC_GC_SHADER_ARRAY_CONFIG, 0xffff0001, 0x44000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCC_GC_SHADER_ARRAY_CONFIG, 0xffff0001, 0x46000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCC_GC_SHADER_ARRAY_CONFIG, 0xffff0001, 0x48000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCC_GC_SHADER_ARRAY_CONFIG, 0xffff0001, 0x4A000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCGTS_TCC_DISABLE, 0x00000001, 0x00000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCC_GC_SHADER_RATE_CONFIG, 0x00000001, 0x00000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCC_GC_EDC_CONFIG, 0x00000001, 0x00000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL, 0x00000500, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_SYSTEM_APERTURE_LOW_ADDR, 0x00000001, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_SYSTEM_APERTURE_HIGH_ADDR, 0x00000000, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_LOCAL_FB_ADDRESS_START, 0x00000000, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_LOCAL_FB_ADDRESS_END, 0x000005ff, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_FB_LOCATION_BASE, 0x00006000, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_FB_LOCATION_TOP, 0x000065ff, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_CONTEXT0_CNTL, 0x00000000, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_CONTEXT1_CNTL, 0x00000000, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_NB_TOP_OF_DRAM_SLOT1, 0xff800000, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_NB_LOWER_TOP_OF_DRAM2, 0x00000001, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_NB_UPPER_TOP_OF_DRAM2, 0x00000fff, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL, 0x00001ffc, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL, 0x00000551, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_CNTL, 0x00080603, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_CNTL2, 0x00000003, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_CNTL3, 0x00100003, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_CNTL5, 0x00003fe0, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_CONTEXT0_CNTL, 0x00000001, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_CONTEXT0_PER_PFVF_PTE_CACHE_FRAGMENT_SIZES, 0x00000c00, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_CONTEXT1_CNTL, 0x00000001, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_CONTEXT1_PER_PFVF_PTE_CACHE_FRAGMENT_SIZES, 0x00000c00, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGB_ADDR_CONFIG, 0x00000444, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGL2_PIPE_STEER_0, 0x54105410, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGL2_PIPE_STEER_2, 0x76323276, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGB_ADDR_CONFIG, 0x00000244, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCUTCL2_HARVEST_BYPASS_GROUPS, 0x00000006, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_APT_CNTL, 0x0000000c, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_AGP_BASE, 0x00000000, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_AGP_BOT, 0x00000002, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_AGP_TOP, 0x00000000, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL2, 0x00020000, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regSDMA0_UCODE_SELFLOAD_CONTROL, 0x00000210, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regSDMA1_UCODE_SELFLOAD_CONTROL, 0x00000210, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCPC_PSP_DEBUG, CPC_PSP_DEBUG__GPA_OVERRIDE_MASK, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCPG_PSP_DEBUG, CPG_PSP_DEBUG__GPA_OVERRIDE_MASK, 0xe0000000),
+};
+
+static void program_rlc_ram_register_setting(struct amdgpu_device *adev,
+					     const struct imu_rlc_ram_golden *regs,
+					     const u32 array_size)
+{
+	const struct imu_rlc_ram_golden *entry;
+	u32 reg, data;
+	int i;
+
+	for (i = 0; i < array_size; ++i) {
+		entry = &regs[i];
+		reg =  adev->reg_offset[entry->hwip][entry->instance][entry->segment] + entry->reg;
+		reg |= entry->addr_mask;
+
+		data = entry->data;
+		if (entry->reg == regGCMC_VM_AGP_BASE)
+			data = 0x00ffffff;
+		else if (entry->reg == regGCMC_VM_AGP_TOP)
+			data = 0x0;
+		else if (entry->reg == regGCMC_VM_FB_LOCATION_BASE)
+			data = adev->gmc.vram_start >> 24;
+		else if (entry->reg == regGCMC_VM_FB_LOCATION_TOP)
+			data = adev->gmc.vram_end >> 24;
+
+		WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_ADDR_HIGH, 0);
+		WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_ADDR_LOW, reg);
+		WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_DATA, data);
+	}
+	//Indicate the latest entry
+	WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_ADDR_HIGH, 0);
+	WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_ADDR_LOW, 0);
+	WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_DATA, 0);
+}
+
+void imu_v11_0_3_program_rlc_ram(struct amdgpu_device *adev)
+{
+	program_rlc_ram_register_setting(adev,
+					 imu_rlc_ram_golden_11_0_3,
+					 (const u32)ARRAY_SIZE(imu_rlc_ram_golden_11_0_3));
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/imu_v11_0_3.h b/drivers/gpu/drm/amd/amdgpu/imu_v11_0_3.h
new file mode 100644
index 000000000000..702be568f26b
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/imu_v11_0_3.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __IMU_V11_0_3_H__
+#define __IMU_V11_0_3_H__
+
+void imu_v11_0_3_program_rlc_ram(struct amdgpu_device *adev);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/imu_v12_0.c b/drivers/gpu/drm/amd/amdgpu/imu_v12_0.c
new file mode 100644
index 000000000000..58cd87db8061
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/imu_v12_0.c
@@ -0,0 +1,406 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include "amdgpu.h"
+#include "amdgpu_imu.h"
+#include "amdgpu_dpm.h"
+
+#include "imu_v12_0.h"
+
+#include "gc/gc_12_0_0_offset.h"
+#include "gc/gc_12_0_0_sh_mask.h"
+#include "mmhub/mmhub_4_1_0_offset.h"
+
+MODULE_FIRMWARE("amdgpu/gc_12_0_0_imu.bin");
+MODULE_FIRMWARE("amdgpu/gc_12_0_1_imu.bin");
+MODULE_FIRMWARE("amdgpu/gc_12_0_1_imu_kicker.bin");
+
+#define TRANSFER_RAM_MASK	0x001c0000
+
+static int imu_v12_0_init_microcode(struct amdgpu_device *adev)
+{
+	char ucode_prefix[30];
+	int err;
+	const struct imu_firmware_header_v1_0 *imu_hdr;
+	struct amdgpu_firmware_info *info = NULL;
+
+	DRM_DEBUG("\n");
+
+	amdgpu_ucode_ip_version_decode(adev, GC_HWIP, ucode_prefix, sizeof(ucode_prefix));
+	if (amdgpu_is_kicker_fw(adev))
+		err = amdgpu_ucode_request(adev, &adev->gfx.imu_fw, AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_imu_kicker.bin", ucode_prefix);
+	else
+		err = amdgpu_ucode_request(adev, &adev->gfx.imu_fw, AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_imu.bin", ucode_prefix);
+	if (err)
+		goto out;
+
+	imu_hdr = (const struct imu_firmware_header_v1_0 *)adev->gfx.imu_fw->data;
+	adev->gfx.imu_fw_version = le32_to_cpu(imu_hdr->header.ucode_version);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_IMU_I];
+		info->ucode_id = AMDGPU_UCODE_ID_IMU_I;
+		info->fw = adev->gfx.imu_fw;
+		adev->firmware.fw_size +=
+			ALIGN(le32_to_cpu(imu_hdr->imu_iram_ucode_size_bytes), PAGE_SIZE);
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_IMU_D];
+		info->ucode_id = AMDGPU_UCODE_ID_IMU_D;
+		info->fw = adev->gfx.imu_fw;
+		adev->firmware.fw_size +=
+			ALIGN(le32_to_cpu(imu_hdr->imu_dram_ucode_size_bytes), PAGE_SIZE);
+	}
+
+out:
+	if (err) {
+		dev_err(adev->dev,
+			"gfx12: Failed to load firmware \"%s_imu.bin\"\n",
+			ucode_prefix);
+		amdgpu_ucode_release(&adev->gfx.imu_fw);
+	}
+
+	return err;
+}
+
+static int imu_v12_0_load_microcode(struct amdgpu_device *adev)
+{
+	const struct imu_firmware_header_v1_0 *hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+
+	if (!adev->gfx.imu_fw)
+		return -EINVAL;
+
+	hdr = (const struct imu_firmware_header_v1_0 *)adev->gfx.imu_fw->data;
+
+	fw_data = (const __le32 *)(adev->gfx.imu_fw->data +
+			le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(hdr->imu_iram_ucode_size_bytes) / 4;
+
+	WREG32_SOC15(GC, 0, regGFX_IMU_I_RAM_ADDR, 0);
+
+	for (i = 0; i < fw_size; i++)
+		WREG32_SOC15(GC, 0, regGFX_IMU_I_RAM_DATA, le32_to_cpup(fw_data++));
+
+	WREG32_SOC15(GC, 0, regGFX_IMU_I_RAM_ADDR, adev->gfx.imu_fw_version);
+
+	fw_data = (const __le32 *)(adev->gfx.imu_fw->data +
+			le32_to_cpu(hdr->header.ucode_array_offset_bytes) +
+			le32_to_cpu(hdr->imu_iram_ucode_size_bytes));
+	fw_size = le32_to_cpu(hdr->imu_dram_ucode_size_bytes) / 4;
+
+	WREG32_SOC15(GC, 0, regGFX_IMU_D_RAM_ADDR, 0);
+
+	for (i = 0; i < fw_size; i++)
+		WREG32_SOC15(GC, 0, regGFX_IMU_D_RAM_DATA, le32_to_cpup(fw_data++));
+
+	WREG32_SOC15(GC, 0, regGFX_IMU_D_RAM_ADDR, adev->gfx.imu_fw_version);
+
+	return 0;
+}
+
+static int imu_v12_0_wait_for_reset_status(struct amdgpu_device *adev)
+{
+	u32 imu_reg_val = 0;
+	int i;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		imu_reg_val = RREG32_SOC15(GC, 0, regGFX_IMU_GFX_RESET_CTRL);
+		if ((imu_reg_val & 0x1f) == 0x1f)
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout) {
+		dev_err(adev->dev, "init imu: IMU start timeout\n");
+		return -ETIMEDOUT;
+	}
+
+	return 0;
+}
+
+static void imu_v12_0_setup(struct amdgpu_device *adev)
+{
+	u32 imu_reg_val;
+
+	WREG32_SOC15(GC, 0, regGFX_IMU_C2PMSG_ACCESS_CTRL0, 0xffffff);
+	WREG32_SOC15(GC, 0, regGFX_IMU_C2PMSG_ACCESS_CTRL1, 0xffff);
+
+	if (adev->gfx.imu.mode == DEBUG_MODE) {
+		imu_reg_val = RREG32_SOC15(GC, 0, regGFX_IMU_C2PMSG_16);
+		imu_reg_val |= 0x1;
+		WREG32_SOC15(GC, 0, regGFX_IMU_C2PMSG_16, imu_reg_val);
+
+		imu_reg_val = RREG32_SOC15(GC, 0, regGFX_IMU_SCRATCH_10);
+		imu_reg_val |= 0x20010007;
+		WREG32_SOC15(GC, 0, regGFX_IMU_SCRATCH_10, imu_reg_val);
+
+	}
+}
+
+static int imu_v12_0_start(struct amdgpu_device *adev)
+{
+	u32 imu_reg_val;
+
+	imu_reg_val = RREG32_SOC15(GC, 0, regGFX_IMU_CORE_CTRL);
+	imu_reg_val &= 0xfffffffe;
+	WREG32_SOC15(GC, 0, regGFX_IMU_CORE_CTRL, imu_reg_val);
+
+	if (adev->flags & AMD_IS_APU)
+		amdgpu_dpm_set_gfx_power_up_by_imu(adev);
+
+	return imu_v12_0_wait_for_reset_status(adev);
+}
+
+static const struct imu_rlc_ram_golden imu_rlc_ram_golden_12_0_1[] = {
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCH_PIPE_STEER, 0x1e4, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGL1X_PIPE_STEER, 0x1e4, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGL1_PIPE_STEER, 0x1e4, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGL2_PIPE_STEER_0, 0x13571357, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGL2_PIPE_STEER_1, 0x64206420, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGL2_PIPE_STEER_2, 0x2460246, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGL2_PIPE_STEER_3, 0x75317531, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGL2C_CTRL3, 0xc0d41183, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regSDMA0_CHICKEN_BITS, 0x507d1c0, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regSDMA1_CHICKEN_BITS, 0x507d1c0, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCP_RB_WPTR_POLL_CNTL, 0x600100, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGC_EA_CPWD_SDP_CREDITS, 0x3f7fff, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGC_EA_SE_SDP_CREDITS, 0x3f7ebf, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGC_EA_CPWD_SDP_TAG_RESERVE0, 0x2e00000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGC_EA_CPWD_SDP_TAG_RESERVE1, 0x1a078, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGC_EA_CPWD_SDP_TAG_RESERVE2, 0x0, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGC_EA_SE_SDP_TAG_RESERVE0, 0x0, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGC_EA_SE_SDP_TAG_RESERVE1, 0x12030, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGC_EA_SE_SDP_TAG_RESERVE2, 0x0, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGC_EA_CPWD_SDP_VCC_RESERVE0, 0x19041000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGC_EA_CPWD_SDP_VCC_RESERVE1, 0x80000000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGC_EA_SE_SDP_VCC_RESERVE0, 0x1e080000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGC_EA_SE_SDP_VCC_RESERVE1, 0x80000000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGC_EA_CPWD_SDP_PRIORITY, 0x880, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGC_EA_SE_SDP_PRIORITY, 0x8880, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGC_EA_CPWD_SDP_ARB_FINAL, 0x17, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGC_EA_SE_SDP_ARB_FINAL, 0x77, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGC_EA_CPWD_SDP_ENABLE, 0x00000001, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGC_EA_SE_SDP_ENABLE, 0x00000001, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL2, 0x20000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_APT_CNTL, 0x0c, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_CACHEABLE_DRAM_ADDRESS_END, 0xfffff, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGC_EA_CPWD_MISC, 0x0091, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGC_EA_SE_MISC, 0x0091, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGRBM_GFX_INDEX, 0xe0000000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCR_GENERAL_CNTL, 0x00008500, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regPA_CL_ENHANCE, 0x00880007, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regTD_CNTL, 0x00000001, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGRBM_GFX_INDEX, 0x00000000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regRMI_GENERAL_CNTL, 0x01e00000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGRBM_GFX_INDEX, 0x00000001, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regRMI_GENERAL_CNTL, 0x01e00000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGRBM_GFX_INDEX, 0x00000100, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regRMI_GENERAL_CNTL, 0x01e00000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGRBM_GFX_INDEX, 0x00000101, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regRMI_GENERAL_CNTL, 0x01e00000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGRBM_GFX_INDEX, 0xe0000000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGB_ADDR_CONFIG, 0x08200545, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGRBMH_CP_PERFMON_CNTL, 0x00000000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCB_PERFCOUNTER0_SELECT1, 0x000fffff, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCP_DEBUG_2, 0x00020000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regCP_CPC_DEBUG, 0x00500010, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL, 0x00000500, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_SYSTEM_APERTURE_LOW_ADDR, 0x00000001, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_SYSTEM_APERTURE_HIGH_ADDR, 0x00000000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_LOCAL_FB_ADDRESS_START, 0x00000000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_LOCAL_FB_ADDRESS_END, 0x0000000f, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_FB_LOCATION_BASE, 0x00006000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_FB_LOCATION_TOP, 0x0000600f, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_CONTEXT0_CNTL, 0x00000000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_CONTEXT1_CNTL, 0x00000000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_NB_TOP_OF_DRAM_SLOT1, 0xff800000, 0xe0000000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_NB_LOWER_TOP_OF_DRAM2, 0x00000001, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_NB_UPPER_TOP_OF_DRAM2, 0x0000ffff, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_AGP_BASE, 0x00000000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_AGP_BOT, 0x00000002, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_AGP_TOP, 0x00000000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL, 0x00001ffc, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL, 0x00000551, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_CNTL, 0x00080603, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_CNTL2, 0x00000003, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_CNTL3, 0x00100003, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCVM_L2_CNTL5, 0x00003fe0, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_SYSTEM_APERTURE_LOW_ADDR, 0x0003d000, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_SYSTEM_APERTURE_HIGH_ADDR, 0x0003d7ff, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB, 0, 0x1c0000),
+	IMU_RLC_RAM_GOLDEN_VALUE(GC, 0, regGCMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB, 0, 0x1c0000)
+};
+
+static void program_imu_rlc_ram_old(struct amdgpu_device *adev,
+				    const struct imu_rlc_ram_golden *regs,
+				    const u32 array_size)
+{
+	const struct imu_rlc_ram_golden *entry;
+	u32 reg, data;
+	int i;
+
+	for (i = 0; i < array_size; ++i) {
+		entry = &regs[i];
+		reg =  adev->reg_offset[entry->hwip][entry->instance][entry->segment] + entry->reg;
+		reg |= entry->addr_mask;
+		data = entry->data;
+		if (entry->reg == regGCMC_VM_AGP_BASE)
+			data = 0x00ffffff;
+		else if (entry->reg == regGCMC_VM_AGP_TOP)
+			data = 0x0;
+		else if (entry->reg == regGCMC_VM_FB_LOCATION_BASE)
+			data = adev->gmc.vram_start >> 24;
+		else if (entry->reg == regGCMC_VM_FB_LOCATION_TOP)
+			data = adev->gmc.vram_end >> 24;
+
+		WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_ADDR_HIGH, 0);
+		WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_ADDR_LOW, reg);
+		WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_DATA, data);
+	}
+}
+
+static u32 imu_v12_0_grbm_gfx_index_remap(struct amdgpu_device *adev,
+					  u32 data, bool high)
+{
+	u32 val, inst_index;
+
+	inst_index = REG_GET_FIELD(data, GRBM_GFX_INDEX, INSTANCE_INDEX);
+
+	if (high)
+		val = inst_index >> 5;
+	else
+		val = REG_GET_FIELD(data, GRBM_GFX_INDEX, SE_BROADCAST_WRITES) << 18 |
+		      REG_GET_FIELD(data, GRBM_GFX_INDEX, SA_BROADCAST_WRITES) << 19 |
+		      REG_GET_FIELD(data, GRBM_GFX_INDEX, INSTANCE_BROADCAST_WRITES) << 20 |
+		      REG_GET_FIELD(data, GRBM_GFX_INDEX, SE_INDEX) << 21 |
+		      REG_GET_FIELD(data, GRBM_GFX_INDEX, SA_INDEX) << 25 |
+		      (inst_index & 0x1f);
+
+	return val;
+}
+
+static u32 imu_v12_init_gfxhub_settings(struct amdgpu_device *adev,
+					u32 reg, u32 data)
+{
+	if (reg == SOC15_REG_OFFSET(GC, 0, regGCMC_VM_FB_LOCATION_BASE))
+		return RREG32_SOC15(MMHUB, 0, regMMMC_VM_FB_LOCATION_BASE);
+	else if (reg == SOC15_REG_OFFSET(GC, 0, regGCMC_VM_FB_LOCATION_TOP))
+		return RREG32_SOC15(MMHUB, 0, regMMMC_VM_FB_LOCATION_TOP);
+	else if (reg == SOC15_REG_OFFSET(GC, 0, regGCMC_VM_FB_OFFSET))
+		return RREG32_SOC15(MMHUB, 0, regMMMC_VM_FB_OFFSET);
+	else if (reg == SOC15_REG_OFFSET(GC, 0, regGCMC_VM_AGP_BASE))
+		return RREG32_SOC15(MMHUB, 0, regMMMC_VM_AGP_BASE);
+	else if (reg == SOC15_REG_OFFSET(GC, 0, regGCMC_VM_AGP_BOT))
+		return RREG32_SOC15(MMHUB, 0, regMMMC_VM_AGP_BOT);
+	else if (reg == SOC15_REG_OFFSET(GC, 0, regGCMC_VM_AGP_TOP))
+		return RREG32_SOC15(MMHUB, 0, regMMMC_VM_AGP_TOP);
+	else if (reg == SOC15_REG_OFFSET(GC, 0, regGCMC_VM_MX_L1_TLB_CNTL))
+		return RREG32_SOC15(MMHUB, 0, regMMMC_VM_MX_L1_TLB_CNTL);
+	else if (reg == SOC15_REG_OFFSET(GC, 0, regGCMC_VM_SYSTEM_APERTURE_LOW_ADDR))
+		return RREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_LOW_ADDR);
+	else if (reg == SOC15_REG_OFFSET(GC, 0, regGCMC_VM_SYSTEM_APERTURE_HIGH_ADDR))
+		return RREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_HIGH_ADDR);
+	else if (reg == SOC15_REG_OFFSET(GC, 0, regGCMC_VM_LOCAL_FB_ADDRESS_START))
+		return RREG32_SOC15(MMHUB, 0, regMMMC_VM_LOCAL_FB_ADDRESS_START);
+	else if (reg == SOC15_REG_OFFSET(GC, 0, regGCMC_VM_LOCAL_FB_ADDRESS_END))
+		return RREG32_SOC15(MMHUB, 0, regMMMC_VM_LOCAL_FB_ADDRESS_END);
+	else if (reg == SOC15_REG_OFFSET(GC, 0, regGCMC_VM_LOCAL_SYSMEM_ADDRESS_START))
+		return RREG32_SOC15(MMHUB, 0, regMMMC_VM_LOCAL_SYSMEM_ADDRESS_START);
+	else if (reg == SOC15_REG_OFFSET(GC, 0, regGCMC_VM_LOCAL_SYSMEM_ADDRESS_END))
+		return RREG32_SOC15(MMHUB, 0, regMMMC_VM_LOCAL_SYSMEM_ADDRESS_END);
+	else if (reg == SOC15_REG_OFFSET(GC, 0, regGCMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB))
+		return RREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB);
+	else if (reg == SOC15_REG_OFFSET(GC, 0, regGCMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB))
+		return RREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB);
+	else
+		return data;
+}
+
+static void program_imu_rlc_ram(struct amdgpu_device *adev,
+				const u32 *regs,
+				const u32 array_size)
+{
+	u32 reg, data, val_h = 0, val_l = TRANSFER_RAM_MASK;
+	int i;
+
+	if (array_size % 3)
+		return;
+
+	for (i = 0; i < array_size; i += 3) {
+		reg = regs[i + 0];
+		data = regs[i + 2];
+		data = imu_v12_init_gfxhub_settings(adev, reg, data);
+		if (reg == SOC15_REG_OFFSET(GC, 0, regGRBM_GFX_INDEX)) {
+			val_l = imu_v12_0_grbm_gfx_index_remap(adev, data, false);
+			val_h = imu_v12_0_grbm_gfx_index_remap(adev, data, true);
+		} else {
+			WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_ADDR_HIGH, val_h);
+			WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_ADDR_LOW, reg | val_l);
+			WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_DATA, data);
+		}
+	}
+}
+
+static void imu_v12_0_program_rlc_ram(struct amdgpu_device *adev)
+{
+	u32 reg_data, size = 0;
+	const u32 *data = NULL;
+	int r = -EINVAL;
+
+	WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_INDEX, 0x2);
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 0, 1):
+		if (!r)
+			program_imu_rlc_ram(adev, data, (const u32)size);
+		else
+			program_imu_rlc_ram_old(adev, imu_rlc_ram_golden_12_0_1,
+				(const u32)ARRAY_SIZE(imu_rlc_ram_golden_12_0_1));
+		break;
+	default:
+		BUG();
+		break;
+	}
+
+	//Indicate the latest entry
+	WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_ADDR_HIGH, 0);
+	WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_ADDR_LOW, 0);
+	WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_DATA, 0);
+
+	reg_data = RREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_INDEX);
+	reg_data |= GFX_IMU_RLC_RAM_INDEX__RAM_VALID_MASK;
+	WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_INDEX, reg_data);
+}
+
+const struct amdgpu_imu_funcs gfx_v12_0_imu_funcs = {
+	.init_microcode = imu_v12_0_init_microcode,
+	.load_microcode = imu_v12_0_load_microcode,
+	.setup_imu = imu_v12_0_setup,
+	.start_imu = imu_v12_0_start,
+	.program_rlc_ram = imu_v12_0_program_rlc_ram,
+	.wait_for_reset_status = imu_v12_0_wait_for_reset_status,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/imu_v12_0.h b/drivers/gpu/drm/amd/amdgpu/imu_v12_0.h
new file mode 100644
index 000000000000..a1f50cb1aeab
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/imu_v12_0.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __IMU_V12_0_H__
+#define __IMU_V12_0_H__
+
+extern const struct amdgpu_imu_funcs gfx_v12_0_imu_funcs;
+
+#endif
+
diff --git a/drivers/gpu/drm/amd/amdgpu/isp_v4_1_0.c b/drivers/gpu/drm/amd/amdgpu/isp_v4_1_0.c
new file mode 100644
index 000000000000..0027a639c7e6
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/isp_v4_1_0.c
@@ -0,0 +1,191 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright (C) 2024 Advanced Micro Devices, Inc. All rights reserved.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+
+#include "amdgpu.h"
+#include "isp_v4_1_0.h"
+
+static const unsigned int isp_4_1_0_int_srcid[MAX_ISP410_INT_SRC] = {
+	ISP_4_1__SRCID__ISP_RINGBUFFER_WPT9,
+	ISP_4_1__SRCID__ISP_RINGBUFFER_WPT10,
+	ISP_4_1__SRCID__ISP_RINGBUFFER_WPT11,
+	ISP_4_1__SRCID__ISP_RINGBUFFER_WPT12,
+	ISP_4_1__SRCID__ISP_RINGBUFFER_WPT13,
+	ISP_4_1__SRCID__ISP_RINGBUFFER_WPT14,
+	ISP_4_1__SRCID__ISP_RINGBUFFER_WPT15,
+	ISP_4_1__SRCID__ISP_RINGBUFFER_WPT16
+};
+
+static int isp_v4_1_0_hw_init(struct amdgpu_isp *isp)
+{
+	struct amdgpu_device *adev = isp->adev;
+	int idx, int_idx, num_res, r;
+	u64 isp_base;
+
+	if (adev->rmmio_size == 0 || adev->rmmio_size < 0x5289)
+		return -EINVAL;
+
+	isp_base = adev->rmmio_base;
+
+	isp->isp_cell = kcalloc(3, sizeof(struct mfd_cell), GFP_KERNEL);
+	if (!isp->isp_cell) {
+		r = -ENOMEM;
+		drm_err(&adev->ddev,
+			"%s: isp mfd cell alloc failed\n", __func__);
+		goto failure;
+	}
+
+	num_res = MAX_ISP410_MEM_RES + MAX_ISP410_INT_SRC;
+	isp->isp_res = kcalloc(num_res, sizeof(struct resource),
+			       GFP_KERNEL);
+	if (!isp->isp_res) {
+		r = -ENOMEM;
+		drm_err(&adev->ddev,
+			"%s: isp mfd res alloc failed\n", __func__);
+		goto failure;
+	}
+
+	isp->isp_pdata = kzalloc(sizeof(*isp->isp_pdata), GFP_KERNEL);
+	if (!isp->isp_pdata) {
+		r = -ENOMEM;
+		drm_err(&adev->ddev,
+			"%s: isp platform data alloc failed\n", __func__);
+		goto failure;
+	}
+
+	/* initialize isp platform data */
+	isp->isp_pdata->adev = (void *)adev;
+	isp->isp_pdata->asic_type = adev->asic_type;
+	isp->isp_pdata->base_rmmio_size = adev->rmmio_size;
+
+	isp->isp_res[0].name = "isp_4_1_0_reg";
+	isp->isp_res[0].flags = IORESOURCE_MEM;
+	isp->isp_res[0].start = isp_base;
+	isp->isp_res[0].end = isp_base + ISP_REGS_OFFSET_END;
+
+	isp->isp_res[1].name = "isp_4_1_phy0_reg";
+	isp->isp_res[1].flags = IORESOURCE_MEM;
+	isp->isp_res[1].start = isp_base + ISP410_PHY0_OFFSET;
+	isp->isp_res[1].end = isp_base + ISP410_PHY0_OFFSET + ISP410_PHY0_SIZE;
+
+	for (idx = MAX_ISP410_MEM_RES, int_idx = 0; idx < num_res; idx++, int_idx++) {
+		isp->isp_res[idx].name = "isp_4_1_0_irq";
+		isp->isp_res[idx].flags = IORESOURCE_IRQ;
+		isp->isp_res[idx].start =
+			amdgpu_irq_create_mapping(adev, isp_4_1_0_int_srcid[int_idx]);
+		isp->isp_res[idx].end =
+			isp->isp_res[idx].start;
+	}
+
+	isp->isp_cell[0].name = "amd_isp_capture";
+	isp->isp_cell[0].num_resources = num_res;
+	isp->isp_cell[0].resources = &isp->isp_res[0];
+	isp->isp_cell[0].platform_data = isp->isp_pdata;
+	isp->isp_cell[0].pdata_size = sizeof(struct isp_platform_data);
+
+	/* initialize isp i2c platform data */
+	isp->isp_i2c_res = kcalloc(1, sizeof(struct resource), GFP_KERNEL);
+	if (!isp->isp_i2c_res) {
+		r = -ENOMEM;
+		drm_err(&adev->ddev,
+			"%s: isp mfd res alloc failed\n", __func__);
+		goto failure;
+	}
+
+	isp->isp_i2c_res[0].name = "isp_i2c0_reg";
+	isp->isp_i2c_res[0].flags = IORESOURCE_MEM;
+	isp->isp_i2c_res[0].start = isp_base + ISP410_I2C0_OFFSET;
+	isp->isp_i2c_res[0].end = isp_base + ISP410_I2C0_OFFSET + ISP410_I2C0_SIZE;
+
+	isp->isp_cell[1].name = "amd_isp_i2c_designware";
+	isp->isp_cell[1].num_resources = 1;
+	isp->isp_cell[1].resources = &isp->isp_i2c_res[0];
+	isp->isp_cell[1].platform_data = isp->isp_pdata;
+	isp->isp_cell[1].pdata_size = sizeof(struct isp_platform_data);
+
+	/* initialize isp gpiochip platform data */
+	isp->isp_gpio_res = kcalloc(1, sizeof(struct resource), GFP_KERNEL);
+	if (!isp->isp_gpio_res) {
+		r = -ENOMEM;
+		drm_err(&adev->ddev,
+			"%s: isp gpio res alloc failed\n", __func__);
+		goto failure;
+	}
+
+	isp->isp_gpio_res[0].name = "isp_gpio_reg";
+	isp->isp_gpio_res[0].flags = IORESOURCE_MEM;
+	isp->isp_gpio_res[0].start = isp_base + ISP410_GPIO_SENSOR_OFFSET;
+	isp->isp_gpio_res[0].end = isp_base + ISP410_GPIO_SENSOR_OFFSET +
+				   ISP410_GPIO_SENSOR_SIZE;
+
+	isp->isp_cell[2].name = "amdisp-pinctrl";
+	isp->isp_cell[2].num_resources = 1;
+	isp->isp_cell[2].resources = &isp->isp_gpio_res[0];
+	isp->isp_cell[2].platform_data = isp->isp_pdata;
+	isp->isp_cell[2].pdata_size = sizeof(struct isp_platform_data);
+
+	r = mfd_add_hotplug_devices(isp->parent, isp->isp_cell, 3);
+	if (r) {
+		drm_err(&adev->ddev,
+			"%s: add mfd hotplug device failed\n", __func__);
+		goto failure;
+	}
+
+	return 0;
+
+failure:
+
+	kfree(isp->isp_pdata);
+	kfree(isp->isp_res);
+	kfree(isp->isp_cell);
+	kfree(isp->isp_i2c_res);
+	kfree(isp->isp_gpio_res);
+
+	return r;
+}
+
+static int isp_v4_1_0_hw_fini(struct amdgpu_isp *isp)
+{
+	mfd_remove_devices(isp->parent);
+
+	kfree(isp->isp_res);
+	kfree(isp->isp_cell);
+	kfree(isp->isp_pdata);
+	kfree(isp->isp_i2c_res);
+	kfree(isp->isp_gpio_res);
+
+	return 0;
+}
+
+static const struct isp_funcs isp_v4_1_0_funcs = {
+	.hw_init = isp_v4_1_0_hw_init,
+	.hw_fini = isp_v4_1_0_hw_fini,
+};
+
+void isp_v4_1_0_set_isp_funcs(struct amdgpu_isp *isp)
+{
+	isp->funcs = &isp_v4_1_0_funcs;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/isp_v4_1_0.h b/drivers/gpu/drm/amd/amdgpu/isp_v4_1_0.h
new file mode 100644
index 000000000000..4d239198edd0
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/isp_v4_1_0.h
@@ -0,0 +1,50 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright (C) 2024 Advanced Micro Devices, Inc. All rights reserved.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+
+#ifndef __ISP_V4_1_0_H__
+#define __ISP_V4_1_0_H__
+
+#include "amdgpu_isp.h"
+
+#include "ivsrcid/isp/irqsrcs_isp_4_1.h"
+
+#define MAX_ISP410_MEM_RES 2
+#define MAX_ISP410_SENSOR_RES 1
+#define MAX_ISP410_INT_SRC 8
+
+#define ISP410_PHY0_OFFSET 0x66700
+#define ISP410_PHY0_SIZE   0xD30
+
+#define ISP410_I2C0_OFFSET 0x66400
+#define ISP410_I2C0_SIZE 0x100
+
+#define ISP410_GPIO_SENSOR_OFFSET 0x6613C
+#define ISP410_GPIO_SENSOR_SIZE 0x54
+
+void isp_v4_1_0_set_isp_funcs(struct amdgpu_isp *isp);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.c b/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.c
new file mode 100644
index 000000000000..4258d3e0b706
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.c
@@ -0,0 +1,377 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright (C) 2024 Advanced Micro Devices, Inc. All rights reserved.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+
+#include <linux/gpio/machine.h>
+#include "amdgpu.h"
+#include "isp_v4_1_1.h"
+
+MODULE_FIRMWARE("amdgpu/isp_4_1_1.bin");
+
+#define ISP_PERFORMANCE_STATE_LOW 0
+#define ISP_PERFORMANCE_STATE_HIGH 1
+
+#define ISP_HIGH_PERFORMANC_XCLK 788
+#define ISP_HIGH_PERFORMANC_ICLK 788
+
+static const unsigned int isp_4_1_1_int_srcid[MAX_ISP411_INT_SRC] = {
+	ISP_4_1__SRCID__ISP_RINGBUFFER_WPT9,
+	ISP_4_1__SRCID__ISP_RINGBUFFER_WPT10,
+	ISP_4_1__SRCID__ISP_RINGBUFFER_WPT11,
+	ISP_4_1__SRCID__ISP_RINGBUFFER_WPT12,
+	ISP_4_1__SRCID__ISP_RINGBUFFER_WPT13,
+	ISP_4_1__SRCID__ISP_RINGBUFFER_WPT14,
+	ISP_4_1__SRCID__ISP_RINGBUFFER_WPT15,
+	ISP_4_1__SRCID__ISP_RINGBUFFER_WPT16
+};
+
+static struct gpiod_lookup_table isp_gpio_table = {
+	.dev_id = "amd_isp_capture",
+	.table = {
+		GPIO_LOOKUP("AMDI0030:00", 85, "enable_isp", GPIO_ACTIVE_HIGH),
+		{ }
+	},
+};
+
+static struct gpiod_lookup_table isp_sensor_gpio_table = {
+	.dev_id = "i2c-ov05c10",
+	.table = {
+		GPIO_LOOKUP("amdisp-pinctrl", 0, "enable", GPIO_ACTIVE_HIGH),
+		{ }
+	},
+};
+
+static int isp_poweroff(struct generic_pm_domain *genpd)
+{
+	struct amdgpu_isp *isp = container_of(genpd, struct amdgpu_isp, ispgpd);
+	struct amdgpu_device *adev = isp->adev;
+
+	return amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_ISP, true, 0);
+}
+
+static int isp_poweron(struct generic_pm_domain *genpd)
+{
+	struct amdgpu_isp *isp = container_of(genpd, struct amdgpu_isp, ispgpd);
+	struct amdgpu_device *adev = isp->adev;
+
+	return amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_ISP, false, 0);
+}
+
+static int isp_set_performance_state(struct generic_pm_domain *genpd,
+				     unsigned int state)
+{
+	struct amdgpu_isp *isp = container_of(genpd, struct amdgpu_isp, ispgpd);
+	struct amdgpu_device *adev = isp->adev;
+	u32 iclk, xclk;
+	int ret;
+
+	switch (state) {
+	case ISP_PERFORMANCE_STATE_HIGH:
+		xclk = ISP_HIGH_PERFORMANC_XCLK;
+		iclk = ISP_HIGH_PERFORMANC_ICLK;
+		break;
+	case ISP_PERFORMANCE_STATE_LOW:
+		/* isp runs at default lowest clock-rate on power-on, do nothing */
+		return 0;
+	default:
+		return -EINVAL;
+	}
+
+	ret = amdgpu_dpm_set_soft_freq_range(adev, PP_ISPXCLK, xclk, 0);
+	if (ret) {
+		drm_err(&adev->ddev, "failed to set xclk %u to %u: %d\n",
+			xclk, state, ret);
+		return ret;
+	}
+
+	ret = amdgpu_dpm_set_soft_freq_range(adev, PP_ISPICLK, iclk, 0);
+	if (ret) {
+		drm_err(&adev->ddev, "failed to set iclk %u to %u: %d\n",
+			iclk, state, ret);
+		return ret;
+	}
+
+	return 0;
+}
+
+static int isp_genpd_add_device(struct device *dev, void *data)
+{
+	struct generic_pm_domain *gpd = data;
+	struct platform_device *pdev = container_of(dev, struct platform_device, dev);
+	struct amdgpu_isp *isp = container_of(gpd, struct amdgpu_isp, ispgpd);
+	struct amdgpu_device *adev = isp->adev;
+	int ret;
+
+	if (!pdev)
+		return -EINVAL;
+
+	if (!dev->type->name) {
+		drm_dbg(&adev->ddev, "Invalid device type to add\n");
+		goto exit;
+	}
+
+	if (strcmp(dev->type->name, "mfd_device")) {
+		drm_dbg(&adev->ddev, "Invalid isp mfd device %s to add\n", pdev->mfd_cell->name);
+		goto exit;
+	}
+
+	ret = pm_genpd_add_device(gpd, dev);
+	if (ret) {
+		drm_err(&adev->ddev, "Failed to add dev %s to genpd %d\n",
+			pdev->mfd_cell->name, ret);
+		return -ENODEV;
+	}
+
+exit:
+	/* Continue to add */
+	return 0;
+}
+
+static int isp_genpd_remove_device(struct device *dev, void *data)
+{
+	struct generic_pm_domain *gpd = data;
+	struct platform_device *pdev = container_of(dev, struct platform_device, dev);
+	struct amdgpu_isp *isp = container_of(gpd, struct amdgpu_isp, ispgpd);
+	struct amdgpu_device *adev = isp->adev;
+	int ret;
+
+	if (!pdev)
+		return -EINVAL;
+
+	if (!dev->type->name) {
+		drm_dbg(&adev->ddev, "Invalid device type to remove\n");
+		goto exit;
+	}
+
+	if (strcmp(dev->type->name, "mfd_device")) {
+		drm_dbg(&adev->ddev, "Invalid isp mfd device %s to remove\n",
+			pdev->mfd_cell->name);
+		goto exit;
+	}
+
+	ret = pm_genpd_remove_device(dev);
+	if (ret) {
+		drm_err(&adev->ddev, "Failed to remove dev from genpd %d\n", ret);
+		return -ENODEV;
+	}
+
+exit:
+	/* Continue to remove */
+	return 0;
+}
+
+static int isp_v4_1_1_hw_init(struct amdgpu_isp *isp)
+{
+	const struct software_node *amd_camera_node, *isp4_node;
+	struct amdgpu_device *adev = isp->adev;
+	struct acpi_device *acpi_dev;
+	int idx, int_idx, num_res, r;
+	u64 isp_base;
+
+	if (adev->rmmio_size == 0 || adev->rmmio_size < 0x5289)
+		return -EINVAL;
+
+	r = amdgpu_acpi_get_isp4_dev(&acpi_dev);
+	if (r) {
+		drm_dbg(&adev->ddev, "Invalid isp platform detected (%d)", r);
+		/* allow GPU init to progress */
+		return 0;
+	}
+
+	/* add GPIO resources required for OMNI5C10 sensor */
+	if (!strcmp("OMNI5C10", acpi_device_hid(acpi_dev))) {
+		gpiod_add_lookup_table(&isp_gpio_table);
+		gpiod_add_lookup_table(&isp_sensor_gpio_table);
+	}
+
+	isp_base = adev->rmmio_base;
+
+	isp->ispgpd.name = "ISP_v_4_1_1";
+	isp->ispgpd.power_off = isp_poweroff;
+	isp->ispgpd.power_on = isp_poweron;
+	isp->ispgpd.set_performance_state = isp_set_performance_state;
+
+	r = pm_genpd_init(&isp->ispgpd, NULL, true);
+	if (r) {
+		drm_err(&adev->ddev, "failed to initialize genpd (%d)\n", r);
+		return -EINVAL;
+	}
+
+	isp->isp_cell = kcalloc(3, sizeof(struct mfd_cell), GFP_KERNEL);
+	if (!isp->isp_cell) {
+		r = -ENOMEM;
+		drm_err(&adev->ddev, "isp mfd cell alloc failed (%d)\n", r);
+		goto failure;
+	}
+
+	num_res = MAX_ISP411_MEM_RES + MAX_ISP411_INT_SRC;
+
+	isp->isp_res = kcalloc(num_res, sizeof(struct resource),
+			       GFP_KERNEL);
+	if (!isp->isp_res) {
+		r = -ENOMEM;
+		drm_err(&adev->ddev, "isp mfd resource alloc failed (%d)\n", r);
+		goto failure;
+	}
+
+	isp->isp_pdata = kzalloc(sizeof(*isp->isp_pdata), GFP_KERNEL);
+	if (!isp->isp_pdata) {
+		r = -ENOMEM;
+		drm_err(&adev->ddev, "isp platform data alloc failed (%d)\n", r);
+		goto failure;
+	}
+
+	amd_camera_node = (const struct software_node *)acpi_dev->driver_data;
+	isp4_node = software_node_find_by_name(amd_camera_node, "isp4");
+
+	/* initialize isp platform data */
+	isp->isp_pdata->adev = (void *)adev;
+	isp->isp_pdata->asic_type = adev->asic_type;
+	isp->isp_pdata->base_rmmio_size = adev->rmmio_size;
+
+	isp->isp_res[0].name = "isp_4_1_1_reg";
+	isp->isp_res[0].flags = IORESOURCE_MEM;
+	isp->isp_res[0].start = isp_base;
+	isp->isp_res[0].end = isp_base + ISP_REGS_OFFSET_END;
+
+	isp->isp_res[1].name = "isp_4_1_1_phy0_reg";
+	isp->isp_res[1].flags = IORESOURCE_MEM;
+	isp->isp_res[1].start = isp_base + ISP411_PHY0_OFFSET;
+	isp->isp_res[1].end = isp_base + ISP411_PHY0_OFFSET + ISP411_PHY0_SIZE;
+
+	for (idx = MAX_ISP411_MEM_RES, int_idx = 0; idx < num_res; idx++, int_idx++) {
+		isp->isp_res[idx].name = "isp_4_1_1_irq";
+		isp->isp_res[idx].flags = IORESOURCE_IRQ;
+		isp->isp_res[idx].start =
+			amdgpu_irq_create_mapping(adev, isp_4_1_1_int_srcid[int_idx]);
+		isp->isp_res[idx].end =
+			isp->isp_res[idx].start;
+	}
+
+	isp->isp_cell[0].name = "amd_isp_capture";
+	isp->isp_cell[0].num_resources = num_res;
+	isp->isp_cell[0].resources = &isp->isp_res[0];
+	isp->isp_cell[0].platform_data = isp->isp_pdata;
+	isp->isp_cell[0].swnode = isp4_node;
+	isp->isp_cell[0].pdata_size = sizeof(struct isp_platform_data);
+
+	/* initialize isp i2c platform data */
+	isp->isp_i2c_res = kcalloc(1, sizeof(struct resource), GFP_KERNEL);
+	if (!isp->isp_i2c_res) {
+		r = -ENOMEM;
+		drm_err(&adev->ddev, "isp mfd res alloc failed (%d)\n", r);
+		goto failure;
+	}
+
+	isp->isp_i2c_res[0].name = "isp_i2c0_reg";
+	isp->isp_i2c_res[0].flags = IORESOURCE_MEM;
+	isp->isp_i2c_res[0].start = isp_base + ISP411_I2C0_OFFSET;
+	isp->isp_i2c_res[0].end = isp_base + ISP411_I2C0_OFFSET + ISP411_I2C0_SIZE;
+
+	isp->isp_cell[1].name = "amd_isp_i2c_designware";
+	isp->isp_cell[1].num_resources = 1;
+	isp->isp_cell[1].resources = &isp->isp_i2c_res[0];
+	isp->isp_cell[1].platform_data = isp->isp_pdata;
+	isp->isp_cell[1].pdata_size = sizeof(struct isp_platform_data);
+
+	/* initialize isp gpiochip platform data */
+	isp->isp_gpio_res = kcalloc(1, sizeof(struct resource), GFP_KERNEL);
+	if (!isp->isp_gpio_res) {
+		r = -ENOMEM;
+		drm_err(&adev->ddev, "isp gpio resource alloc failed (%d)\n", r);
+		goto failure;
+	}
+
+	isp->isp_gpio_res[0].name = "isp_gpio_reg";
+	isp->isp_gpio_res[0].flags = IORESOURCE_MEM;
+	isp->isp_gpio_res[0].start = isp_base + ISP411_GPIO_SENSOR_OFFSET;
+	isp->isp_gpio_res[0].end = isp_base + ISP411_GPIO_SENSOR_OFFSET +
+				   ISP411_GPIO_SENSOR_SIZE;
+
+	isp->isp_cell[2].name = "amdisp-pinctrl";
+	isp->isp_cell[2].num_resources = 1;
+	isp->isp_cell[2].resources = &isp->isp_gpio_res[0];
+	isp->isp_cell[2].platform_data = isp->isp_pdata;
+	isp->isp_cell[2].pdata_size = sizeof(struct isp_platform_data);
+
+	/* add only amd_isp_capture and amd_isp_i2c_designware to genpd */
+	r = mfd_add_hotplug_devices(isp->parent, isp->isp_cell, 2);
+	if (r) {
+		drm_err(&adev->ddev, "add mfd hotplug device failed (%d)\n", r);
+		goto failure;
+	}
+
+	r = device_for_each_child(isp->parent, &isp->ispgpd,
+				  isp_genpd_add_device);
+	if (r) {
+		drm_err(&adev->ddev, "failed to add devices to genpd (%d)\n", r);
+		goto failure;
+	}
+
+	r = mfd_add_hotplug_devices(isp->parent, &isp->isp_cell[2], 1);
+	if (r) {
+		drm_err(&adev->ddev, "add pinctl hotplug device failed (%d)\n", r);
+		goto failure;
+	}
+
+	return 0;
+
+failure:
+
+	kfree(isp->isp_pdata);
+	kfree(isp->isp_res);
+	kfree(isp->isp_cell);
+	kfree(isp->isp_i2c_res);
+	kfree(isp->isp_gpio_res);
+
+	return r;
+}
+
+static int isp_v4_1_1_hw_fini(struct amdgpu_isp *isp)
+{
+	device_for_each_child(isp->parent, NULL,
+			      isp_genpd_remove_device);
+
+	mfd_remove_devices(isp->parent);
+
+	kfree(isp->isp_res);
+	kfree(isp->isp_cell);
+	kfree(isp->isp_pdata);
+	kfree(isp->isp_i2c_res);
+	kfree(isp->isp_gpio_res);
+
+	return 0;
+}
+
+static const struct isp_funcs isp_v4_1_1_funcs = {
+	.hw_init = isp_v4_1_1_hw_init,
+	.hw_fini = isp_v4_1_1_hw_fini,
+};
+
+void isp_v4_1_1_set_isp_funcs(struct amdgpu_isp *isp)
+{
+	isp->funcs = &isp_v4_1_1_funcs;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.h b/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.h
new file mode 100644
index 000000000000..fe45d70d87f1
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.h
@@ -0,0 +1,49 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright (C) 2024 Advanced Micro Devices, Inc. All rights reserved.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+
+#ifndef __ISP_V4_1_1_H__
+#define __ISP_V4_1_1_H__
+
+#include "amdgpu_isp.h"
+
+#include "ivsrcid/isp/irqsrcs_isp_4_1.h"
+
+#define MAX_ISP411_MEM_RES 2
+#define MAX_ISP411_INT_SRC 8
+
+#define ISP411_PHY0_OFFSET 0x66700
+#define ISP411_PHY0_SIZE   0xD30
+
+#define ISP411_I2C0_OFFSET 0x66400
+#define ISP411_I2C0_SIZE 0x100
+
+#define ISP411_GPIO_SENSOR_OFFSET 0x6613C
+#define ISP411_GPIO_SENSOR_SIZE 0x54
+
+void isp_v4_1_1_set_isp_funcs(struct amdgpu_isp *isp);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.c
new file mode 100644
index 000000000000..b5bb7f4d607c
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.c
@@ -0,0 +1,687 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "amdgpu_jpeg.h"
+#include "amdgpu_cs.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "vcn_v1_0.h"
+#include "jpeg_v1_0.h"
+
+#include "vcn/vcn_1_0_offset.h"
+#include "vcn/vcn_1_0_sh_mask.h"
+
+static void jpeg_v1_0_set_dec_ring_funcs(struct amdgpu_device *adev);
+static void jpeg_v1_0_set_irq_funcs(struct amdgpu_device *adev);
+static void jpeg_v1_0_ring_begin_use(struct amdgpu_ring *ring);
+static int jpeg_v1_dec_ring_parse_cs(struct amdgpu_cs_parser *parser,
+				     struct amdgpu_job *job,
+				     struct amdgpu_ib *ib);
+
+static void jpeg_v1_0_decode_ring_patch_wreg(struct amdgpu_ring *ring, uint32_t *ptr, uint32_t reg_offset, uint32_t val)
+{
+	struct amdgpu_device *adev = ring->adev;
+	ring->ring[(*ptr)++] = PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JRBC_EXTERNAL_REG_BASE), 0, 0, PACKETJ_TYPE0);
+	if (((reg_offset >= 0x1f800) && (reg_offset <= 0x21fff)) ||
+		((reg_offset >= 0x1e000) && (reg_offset <= 0x1e1ff))) {
+		ring->ring[(*ptr)++] = 0;
+		ring->ring[(*ptr)++] = PACKETJ((reg_offset >> 2), 0, 0, PACKETJ_TYPE0);
+	} else {
+		ring->ring[(*ptr)++] = reg_offset;
+		ring->ring[(*ptr)++] = PACKETJ(0, 0, 0, PACKETJ_TYPE0);
+	}
+	ring->ring[(*ptr)++] = val;
+}
+
+static void jpeg_v1_0_decode_ring_set_patch_ring(struct amdgpu_ring *ring, uint32_t ptr)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	uint32_t reg, reg_offset, val, mask, i;
+
+	// 1st: program mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_LOW
+	reg = SOC15_REG_OFFSET(JPEG, 0, mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_LOW);
+	reg_offset = (reg << 2);
+	val = lower_32_bits(ring->gpu_addr);
+	jpeg_v1_0_decode_ring_patch_wreg(ring, &ptr, reg_offset, val);
+
+	// 2nd: program mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_HIGH
+	reg = SOC15_REG_OFFSET(JPEG, 0, mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_HIGH);
+	reg_offset = (reg << 2);
+	val = upper_32_bits(ring->gpu_addr);
+	jpeg_v1_0_decode_ring_patch_wreg(ring, &ptr, reg_offset, val);
+
+	// 3rd to 5th: issue MEM_READ commands
+	for (i = 0; i <= 2; i++) {
+		ring->ring[ptr++] = PACKETJ(0, 0, 0, PACKETJ_TYPE2);
+		ring->ring[ptr++] = 0;
+	}
+
+	// 6th: program mmUVD_JRBC_RB_CNTL register to enable NO_FETCH and RPTR write ability
+	reg = SOC15_REG_OFFSET(JPEG, 0, mmUVD_JRBC_RB_CNTL);
+	reg_offset = (reg << 2);
+	val = 0x13;
+	jpeg_v1_0_decode_ring_patch_wreg(ring, &ptr, reg_offset, val);
+
+	// 7th: program mmUVD_JRBC_RB_REF_DATA
+	reg = SOC15_REG_OFFSET(JPEG, 0, mmUVD_JRBC_RB_REF_DATA);
+	reg_offset = (reg << 2);
+	val = 0x1;
+	jpeg_v1_0_decode_ring_patch_wreg(ring, &ptr, reg_offset, val);
+
+	// 8th: issue conditional register read mmUVD_JRBC_RB_CNTL
+	reg = SOC15_REG_OFFSET(JPEG, 0, mmUVD_JRBC_RB_CNTL);
+	reg_offset = (reg << 2);
+	val = 0x1;
+	mask = 0x1;
+
+	ring->ring[ptr++] = PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JRBC_RB_COND_RD_TIMER), 0, 0, PACKETJ_TYPE0);
+	ring->ring[ptr++] = 0x01400200;
+	ring->ring[ptr++] = PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JRBC_RB_REF_DATA), 0, 0, PACKETJ_TYPE0);
+	ring->ring[ptr++] = val;
+	ring->ring[ptr++] = PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JRBC_EXTERNAL_REG_BASE), 0, 0, PACKETJ_TYPE0);
+	if (((reg_offset >= 0x1f800) && (reg_offset <= 0x21fff)) ||
+		((reg_offset >= 0x1e000) && (reg_offset <= 0x1e1ff))) {
+		ring->ring[ptr++] = 0;
+		ring->ring[ptr++] = PACKETJ((reg_offset >> 2), 0, 0, PACKETJ_TYPE3);
+	} else {
+		ring->ring[ptr++] = reg_offset;
+		ring->ring[ptr++] = PACKETJ(0, 0, 0, PACKETJ_TYPE3);
+	}
+	ring->ring[ptr++] = mask;
+
+	//9th to 21st: insert no-op
+	for (i = 0; i <= 12; i++) {
+		ring->ring[ptr++] = PACKETJ(0, 0, 0, PACKETJ_TYPE6);
+		ring->ring[ptr++] = 0;
+	}
+
+	//22nd: reset mmUVD_JRBC_RB_RPTR
+	reg = SOC15_REG_OFFSET(JPEG, 0, mmUVD_JRBC_RB_RPTR);
+	reg_offset = (reg << 2);
+	val = 0;
+	jpeg_v1_0_decode_ring_patch_wreg(ring, &ptr, reg_offset, val);
+
+	//23rd: program mmUVD_JRBC_RB_CNTL to disable no_fetch
+	reg = SOC15_REG_OFFSET(JPEG, 0, mmUVD_JRBC_RB_CNTL);
+	reg_offset = (reg << 2);
+	val = 0x12;
+	jpeg_v1_0_decode_ring_patch_wreg(ring, &ptr, reg_offset, val);
+}
+
+/**
+ * jpeg_v1_0_decode_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t jpeg_v1_0_decode_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_RPTR);
+}
+
+/**
+ * jpeg_v1_0_decode_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t jpeg_v1_0_decode_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_WPTR);
+}
+
+/**
+ * jpeg_v1_0_decode_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void jpeg_v1_0_decode_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	WREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_WPTR, lower_32_bits(ring->wptr));
+}
+
+/**
+ * jpeg_v1_0_decode_ring_insert_start - insert a start command
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Write a start command to the ring.
+ */
+static void jpeg_v1_0_decode_ring_insert_start(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JRBC_EXTERNAL_REG_BASE), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x68e04);
+
+	amdgpu_ring_write(ring, PACKETJ(0, 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x80010000);
+}
+
+/**
+ * jpeg_v1_0_decode_ring_insert_end - insert a end command
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Write a end command to the ring.
+ */
+static void jpeg_v1_0_decode_ring_insert_end(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JRBC_EXTERNAL_REG_BASE), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x68e04);
+
+	amdgpu_ring_write(ring, PACKETJ(0, 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x00010000);
+}
+
+/**
+ * jpeg_v1_0_decode_ring_emit_fence - emit an fence & trap command
+ *
+ * @ring: amdgpu_ring pointer
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
+ *
+ * Write a fence and a trap command to the ring.
+ */
+static void jpeg_v1_0_decode_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				     unsigned flags)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JPEG_GPCOM_DATA0), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, seq);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JPEG_GPCOM_DATA1), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, seq);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_LMI_JRBC_RB_MEM_WR_64BIT_BAR_LOW), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_LMI_JRBC_RB_MEM_WR_64BIT_BAR_HIGH), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JPEG_GPCOM_CMD), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x8);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JPEG_GPCOM_CMD), 0, PACKETJ_CONDITION_CHECK0, PACKETJ_TYPE4));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JRBC_RB_COND_RD_TIMER), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x01400200);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JRBC_RB_REF_DATA), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, seq);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_LOW), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_HIGH), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+
+	amdgpu_ring_write(ring,
+		PACKETJ(0, 0, PACKETJ_CONDITION_CHECK3, PACKETJ_TYPE2));
+	amdgpu_ring_write(ring, 0xffffffff);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JRBC_EXTERNAL_REG_BASE), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x3fbc);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(0, 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x1);
+
+	/* emit trap */
+	amdgpu_ring_write(ring, PACKETJ(0, 0, 0, PACKETJ_TYPE7));
+	amdgpu_ring_write(ring, 0);
+}
+
+/**
+ * jpeg_v1_0_decode_ring_emit_ib - execute indirect buffer
+ *
+ * @ring: amdgpu_ring pointer
+ * @job: job to retrieve vmid from
+ * @ib: indirect buffer to execute
+ * @flags: unused
+ *
+ * Write ring commands to execute the indirect buffer.
+ */
+static void jpeg_v1_0_decode_ring_emit_ib(struct amdgpu_ring *ring,
+					struct amdgpu_job *job,
+					struct amdgpu_ib *ib,
+					uint32_t flags)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_LMI_JRBC_IB_VMID), 0, 0, PACKETJ_TYPE0));
+	if (ring->funcs->parse_cs)
+		amdgpu_ring_write(ring, 0);
+	else
+		amdgpu_ring_write(ring, (vmid | (vmid << 4)));
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_LMI_JPEG_VMID), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, (vmid | (vmid << 4)));
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_LMI_JRBC_IB_64BIT_BAR_LOW), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_LMI_JRBC_IB_64BIT_BAR_HIGH), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JRBC_IB_SIZE), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, ib->length_dw);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_LOW), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, lower_32_bits(ring->gpu_addr));
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_HIGH), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, upper_32_bits(ring->gpu_addr));
+
+	amdgpu_ring_write(ring,
+		PACKETJ(0, 0, PACKETJ_CONDITION_CHECK0, PACKETJ_TYPE2));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JRBC_RB_COND_RD_TIMER), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x01400200);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JRBC_RB_REF_DATA), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x2);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JRBC_STATUS), 0, PACKETJ_CONDITION_CHECK3, PACKETJ_TYPE3));
+	amdgpu_ring_write(ring, 0x2);
+}
+
+static void jpeg_v1_0_decode_ring_emit_reg_wait(struct amdgpu_ring *ring,
+					    uint32_t reg, uint32_t val,
+					    uint32_t mask)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t reg_offset = (reg << 2);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JRBC_RB_COND_RD_TIMER), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x01400200);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JRBC_RB_REF_DATA), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, val);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JRBC_EXTERNAL_REG_BASE), 0, 0, PACKETJ_TYPE0));
+	if (((reg_offset >= 0x1f800) && (reg_offset <= 0x21fff)) ||
+		((reg_offset >= 0x1e000) && (reg_offset <= 0x1e1ff))) {
+		amdgpu_ring_write(ring, 0);
+		amdgpu_ring_write(ring,
+			PACKETJ((reg_offset >> 2), 0, 0, PACKETJ_TYPE3));
+	} else {
+		amdgpu_ring_write(ring, reg_offset);
+		amdgpu_ring_write(ring,
+			PACKETJ(0, 0, 0, PACKETJ_TYPE3));
+	}
+	amdgpu_ring_write(ring, mask);
+}
+
+static void jpeg_v1_0_decode_ring_emit_vm_flush(struct amdgpu_ring *ring,
+		unsigned vmid, uint64_t pd_addr)
+{
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
+	uint32_t data0, data1, mask;
+
+	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* wait for register write */
+	data0 = hub->ctx0_ptb_addr_lo32 + vmid * hub->ctx_addr_distance;
+	data1 = lower_32_bits(pd_addr);
+	mask = 0xffffffff;
+	jpeg_v1_0_decode_ring_emit_reg_wait(ring, data0, data1, mask);
+}
+
+static void jpeg_v1_0_decode_ring_emit_wreg(struct amdgpu_ring *ring,
+					uint32_t reg, uint32_t val)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t reg_offset = (reg << 2);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JRBC_EXTERNAL_REG_BASE), 0, 0, PACKETJ_TYPE0));
+	if (((reg_offset >= 0x1f800) && (reg_offset <= 0x21fff)) ||
+			((reg_offset >= 0x1e000) && (reg_offset <= 0x1e1ff))) {
+		amdgpu_ring_write(ring, 0);
+		amdgpu_ring_write(ring,
+			PACKETJ((reg_offset >> 2), 0, 0, PACKETJ_TYPE0));
+	} else {
+		amdgpu_ring_write(ring, reg_offset);
+		amdgpu_ring_write(ring,
+			PACKETJ(0, 0, 0, PACKETJ_TYPE0));
+	}
+	amdgpu_ring_write(ring, val);
+}
+
+static void jpeg_v1_0_decode_ring_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	int i;
+
+	WARN_ON(ring->wptr % 2 || count % 2);
+
+	for (i = 0; i < count / 2; i++) {
+		amdgpu_ring_write(ring, PACKETJ(0, 0, 0, PACKETJ_TYPE6));
+		amdgpu_ring_write(ring, 0);
+	}
+}
+
+static int jpeg_v1_0_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+static int jpeg_v1_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	DRM_DEBUG("IH: JPEG decode TRAP\n");
+
+	switch (entry->src_id) {
+	case 126:
+		amdgpu_fence_process(adev->jpeg.inst->ring_dec);
+		break;
+	default:
+		DRM_ERROR("Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+/**
+ * jpeg_v1_0_early_init - set function pointers
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Set ring and irq function pointers
+ */
+int jpeg_v1_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	adev->jpeg.num_jpeg_inst = 1;
+	adev->jpeg.num_jpeg_rings = 1;
+
+	jpeg_v1_0_set_dec_ring_funcs(adev);
+	jpeg_v1_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+/**
+ * jpeg_v1_0_sw_init - sw init for JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ */
+int jpeg_v1_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	int r;
+
+	/* JPEG TRAP */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN, 126, &adev->jpeg.inst->irq);
+	if (r)
+		return r;
+
+	ring = adev->jpeg.inst->ring_dec;
+	ring->vm_hub = AMDGPU_MMHUB0(0);
+	sprintf(ring->name, "jpeg_dec");
+	r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst->irq,
+			     0, AMDGPU_RING_PRIO_DEFAULT, NULL);
+	if (r)
+		return r;
+
+	adev->jpeg.internal.jpeg_pitch[0] = adev->jpeg.inst->external.jpeg_pitch[0] =
+		SOC15_REG_OFFSET(JPEG, 0, mmUVD_JPEG_PITCH);
+
+	return 0;
+}
+
+/**
+ * jpeg_v1_0_sw_fini - sw fini for JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * JPEG free up sw allocation
+ */
+void jpeg_v1_0_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	amdgpu_ring_fini(adev->jpeg.inst->ring_dec);
+}
+
+/**
+ * jpeg_v1_0_start - start JPEG block
+ *
+ * @adev: amdgpu_device pointer
+ * @mode: SPG or DPG mode
+ *
+ * Setup and start the JPEG block
+ */
+void jpeg_v1_0_start(struct amdgpu_device *adev, int mode)
+{
+	struct amdgpu_ring *ring = adev->jpeg.inst->ring_dec;
+
+	if (mode == 0) {
+		WREG32_SOC15(JPEG, 0, mmUVD_LMI_JRBC_RB_VMID, 0);
+		WREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_CNTL, UVD_JRBC_RB_CNTL__RB_NO_FETCH_MASK |
+				UVD_JRBC_RB_CNTL__RB_RPTR_WR_EN_MASK);
+		WREG32_SOC15(JPEG, 0, mmUVD_LMI_JRBC_RB_64BIT_BAR_LOW, lower_32_bits(ring->gpu_addr));
+		WREG32_SOC15(JPEG, 0, mmUVD_LMI_JRBC_RB_64BIT_BAR_HIGH, upper_32_bits(ring->gpu_addr));
+		WREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_RPTR, 0);
+		WREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_WPTR, 0);
+		WREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_CNTL, UVD_JRBC_RB_CNTL__RB_RPTR_WR_EN_MASK);
+	}
+
+	/* initialize wptr */
+	ring->wptr = RREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_WPTR);
+
+	/* copy patch commands to the jpeg ring */
+	jpeg_v1_0_decode_ring_set_patch_ring(ring,
+		(ring->wptr + ring->max_dw * amdgpu_sched_hw_submission));
+}
+
+static const struct amdgpu_ring_funcs jpeg_v1_0_decode_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_JPEG,
+	.align_mask = 0xf,
+	.nop = PACKET0(0x81ff, 0),
+	.support_64bit_ptrs = false,
+	.no_user_fence = true,
+	.extra_bytes = 256,
+	.get_rptr = jpeg_v1_0_decode_ring_get_rptr,
+	.get_wptr = jpeg_v1_0_decode_ring_get_wptr,
+	.set_wptr = jpeg_v1_0_decode_ring_set_wptr,
+	.parse_cs = jpeg_v1_dec_ring_parse_cs,
+	.emit_frame_size =
+		6 + 6 + /* hdp invalidate / flush */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
+		8 + /* jpeg_v1_0_decode_ring_emit_vm_flush */
+		26 + 26 + /* jpeg_v1_0_decode_ring_emit_fence x2 vm fence */
+		6,
+	.emit_ib_size = 22, /* jpeg_v1_0_decode_ring_emit_ib */
+	.emit_ib = jpeg_v1_0_decode_ring_emit_ib,
+	.emit_fence = jpeg_v1_0_decode_ring_emit_fence,
+	.emit_vm_flush = jpeg_v1_0_decode_ring_emit_vm_flush,
+	.test_ring = amdgpu_jpeg_dec_ring_test_ring,
+	.test_ib = amdgpu_jpeg_dec_ring_test_ib,
+	.insert_nop = jpeg_v1_0_decode_ring_nop,
+	.insert_start = jpeg_v1_0_decode_ring_insert_start,
+	.insert_end = jpeg_v1_0_decode_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = jpeg_v1_0_ring_begin_use,
+	.end_use = vcn_v1_0_ring_end_use,
+	.emit_wreg = jpeg_v1_0_decode_ring_emit_wreg,
+	.emit_reg_wait = jpeg_v1_0_decode_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+};
+
+static void jpeg_v1_0_set_dec_ring_funcs(struct amdgpu_device *adev)
+{
+	adev->jpeg.inst->ring_dec->funcs = &jpeg_v1_0_decode_ring_vm_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs jpeg_v1_0_irq_funcs = {
+	.set = jpeg_v1_0_set_interrupt_state,
+	.process = jpeg_v1_0_process_interrupt,
+};
+
+static void jpeg_v1_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->jpeg.inst->irq.funcs = &jpeg_v1_0_irq_funcs;
+}
+
+static void jpeg_v1_0_ring_begin_use(struct amdgpu_ring *ring)
+{
+	struct	amdgpu_device *adev = ring->adev;
+	bool	set_clocks = !cancel_delayed_work_sync(&adev->vcn.inst[0].idle_work);
+	int		cnt = 0;
+
+	mutex_lock(&adev->vcn.inst[0].vcn1_jpeg1_workaround);
+
+	if (amdgpu_fence_wait_empty(&adev->vcn.inst->ring_dec))
+		DRM_ERROR("JPEG dec: vcn dec ring may not be empty\n");
+
+	for (cnt = 0; cnt < adev->vcn.inst[0].num_enc_rings; cnt++) {
+		if (amdgpu_fence_wait_empty(&adev->vcn.inst->ring_enc[cnt]))
+			DRM_ERROR("JPEG dec: vcn enc ring[%d] may not be empty\n", cnt);
+	}
+
+	vcn_v1_0_set_pg_for_begin_use(ring, set_clocks);
+}
+
+/**
+ * jpeg_v1_dec_ring_parse_cs - command submission parser
+ *
+ * @parser: Command submission parser context
+ * @job: the job to parse
+ * @ib: the IB to parse
+ *
+ * Parse the command stream, return -EINVAL for invalid packet,
+ * 0 otherwise
+ */
+static int jpeg_v1_dec_ring_parse_cs(struct amdgpu_cs_parser *parser,
+				     struct amdgpu_job *job,
+				     struct amdgpu_ib *ib)
+{
+	u32 i, reg, res, cond, type;
+	int ret = 0;
+	struct amdgpu_device *adev = parser->adev;
+
+	for (i = 0; i < ib->length_dw ; i += 2) {
+		reg  = CP_PACKETJ_GET_REG(ib->ptr[i]);
+		res  = CP_PACKETJ_GET_RES(ib->ptr[i]);
+		cond = CP_PACKETJ_GET_COND(ib->ptr[i]);
+		type = CP_PACKETJ_GET_TYPE(ib->ptr[i]);
+
+		if (res || cond != PACKETJ_CONDITION_CHECK0) /* only allow 0 for now */
+			return -EINVAL;
+
+		if (reg >= JPEG_V1_REG_RANGE_START && reg <= JPEG_V1_REG_RANGE_END)
+			continue;
+
+		switch (type) {
+		case PACKETJ_TYPE0:
+			if (reg != JPEG_V1_LMI_JPEG_WRITE_64BIT_BAR_HIGH &&
+			    reg != JPEG_V1_LMI_JPEG_WRITE_64BIT_BAR_LOW &&
+			    reg != JPEG_V1_LMI_JPEG_READ_64BIT_BAR_HIGH &&
+			    reg != JPEG_V1_LMI_JPEG_READ_64BIT_BAR_LOW &&
+			    reg != JPEG_V1_REG_CTX_INDEX &&
+			    reg != JPEG_V1_REG_CTX_DATA) {
+				ret = -EINVAL;
+			}
+			break;
+		case PACKETJ_TYPE1:
+			if (reg != JPEG_V1_REG_CTX_DATA)
+				ret = -EINVAL;
+			break;
+		case PACKETJ_TYPE3:
+			if (reg != JPEG_V1_REG_SOFT_RESET)
+				ret = -EINVAL;
+			break;
+		case PACKETJ_TYPE6:
+			if (ib->ptr[i] != CP_PACKETJ_NOP)
+				ret = -EINVAL;
+			break;
+		default:
+			ret = -EINVAL;
+		}
+
+		if (ret) {
+			dev_err(adev->dev, "Invalid packet [0x%08x]!\n", ib->ptr[i]);
+			break;
+		}
+	}
+
+	return ret;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.h b/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.h
new file mode 100644
index 000000000000..097328635083
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.h
@@ -0,0 +1,43 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __JPEG_V1_0_H__
+#define __JPEG_V1_0_H__
+
+int jpeg_v1_0_early_init(struct amdgpu_ip_block *ip_block);
+int jpeg_v1_0_sw_init(struct amdgpu_ip_block *ip_block);
+void jpeg_v1_0_sw_fini(struct amdgpu_ip_block *ip_block);
+void jpeg_v1_0_start(struct amdgpu_device *adev, int mode);
+
+#define JPEG_V1_REG_RANGE_START	0x8000
+#define JPEG_V1_REG_RANGE_END	0x803f
+
+#define JPEG_V1_LMI_JPEG_WRITE_64BIT_BAR_HIGH	0x8238
+#define JPEG_V1_LMI_JPEG_WRITE_64BIT_BAR_LOW	0x8239
+#define JPEG_V1_LMI_JPEG_READ_64BIT_BAR_HIGH	0x825a
+#define JPEG_V1_LMI_JPEG_READ_64BIT_BAR_LOW	0x825b
+#define JPEG_V1_REG_CTX_INDEX			0x8328
+#define JPEG_V1_REG_CTX_DATA			0x8329
+#define JPEG_V1_REG_SOFT_RESET			0x83a0
+
+#endif /*__JPEG_V1_0_H__*/
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c
new file mode 100644
index 000000000000..27c76bd424cf
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c
@@ -0,0 +1,855 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "amdgpu_jpeg.h"
+#include "amdgpu_pm.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "jpeg_v2_0.h"
+
+#include "vcn/vcn_2_0_0_offset.h"
+#include "vcn/vcn_2_0_0_sh_mask.h"
+#include "ivsrcid/vcn/irqsrcs_vcn_2_0.h"
+
+static const struct amdgpu_hwip_reg_entry jpeg_reg_list_2_0[] = {
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_INT_STAT),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_CNTL),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_SIZE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_ADDR_MODE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_GFX10_ADDR_CONFIG),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_Y_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_UV_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_PITCH),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_UV_PITCH),
+};
+
+static void jpeg_v2_0_set_dec_ring_funcs(struct amdgpu_device *adev);
+static void jpeg_v2_0_set_irq_funcs(struct amdgpu_device *adev);
+static int jpeg_v2_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
+				enum amd_powergating_state state);
+
+/**
+ * jpeg_v2_0_early_init - set function pointers
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Set ring and irq function pointers
+ */
+static int jpeg_v2_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	adev->jpeg.num_jpeg_inst = 1;
+	adev->jpeg.num_jpeg_rings = 1;
+
+	jpeg_v2_0_set_dec_ring_funcs(adev);
+	jpeg_v2_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+/**
+ * jpeg_v2_0_sw_init - sw init for JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Load firmware and sw initialization
+ */
+static int jpeg_v2_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	int r;
+
+	/* JPEG TRAP */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+		VCN_2_0__SRCID__JPEG_DECODE, &adev->jpeg.inst->irq);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_sw_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_resume(adev);
+	if (r)
+		return r;
+
+	ring = adev->jpeg.inst->ring_dec;
+	ring->use_doorbell = true;
+	ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 1;
+	ring->vm_hub = AMDGPU_MMHUB0(0);
+	sprintf(ring->name, "jpeg_dec");
+	r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst->irq,
+			     0, AMDGPU_RING_PRIO_DEFAULT, NULL);
+	if (r)
+		return r;
+
+	adev->jpeg.internal.jpeg_pitch[0] = mmUVD_JPEG_PITCH_INTERNAL_OFFSET;
+	adev->jpeg.inst->external.jpeg_pitch[0] = SOC15_REG_OFFSET(JPEG, 0, mmUVD_JPEG_PITCH);
+
+	r = amdgpu_jpeg_reg_dump_init(adev, jpeg_reg_list_2_0, ARRAY_SIZE(jpeg_reg_list_2_0));
+	if (r)
+		return r;
+
+	adev->jpeg.supported_reset =
+		amdgpu_get_soft_full_reset_mask(adev->jpeg.inst[0].ring_dec);
+	if (!amdgpu_sriov_vf(adev))
+		adev->jpeg.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+	r = amdgpu_jpeg_sysfs_reset_mask_init(adev);
+
+	return r;
+}
+
+/**
+ * jpeg_v2_0_sw_fini - sw fini for JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * JPEG suspend and free up sw allocation
+ */
+static int jpeg_v2_0_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	r = amdgpu_jpeg_suspend(adev);
+	if (r)
+		return r;
+
+	amdgpu_jpeg_sysfs_reset_mask_fini(adev);
+
+	r = amdgpu_jpeg_sw_fini(adev);
+
+	return r;
+}
+
+/**
+ * jpeg_v2_0_hw_init - start and test JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ */
+static int jpeg_v2_0_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring = adev->jpeg.inst->ring_dec;
+
+	adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
+		(adev->doorbell_index.vcn.vcn_ring0_1 << 1), 0);
+
+	return amdgpu_ring_test_helper(ring);
+}
+
+/**
+ * jpeg_v2_0_hw_fini - stop the hardware block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Stop the JPEG block, mark ring as not ready any more
+ */
+static int jpeg_v2_0_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	cancel_delayed_work_sync(&adev->jpeg.idle_work);
+
+	if (adev->jpeg.cur_state != AMD_PG_STATE_GATE &&
+	      RREG32_SOC15(JPEG, 0, mmUVD_JRBC_STATUS))
+		jpeg_v2_0_set_powergating_state(ip_block, AMD_PG_STATE_GATE);
+
+	return 0;
+}
+
+/**
+ * jpeg_v2_0_suspend - suspend JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * HW fini and suspend JPEG block
+ */
+static int jpeg_v2_0_suspend(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	r = jpeg_v2_0_hw_fini(ip_block);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_suspend(ip_block->adev);
+
+	return r;
+}
+
+/**
+ * jpeg_v2_0_resume - resume JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Resume firmware and hw init JPEG block
+ */
+static int jpeg_v2_0_resume(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	r = amdgpu_jpeg_resume(ip_block->adev);
+	if (r)
+		return r;
+
+	r = jpeg_v2_0_hw_init(ip_block);
+
+	return r;
+}
+
+static int jpeg_v2_0_disable_power_gating(struct amdgpu_device *adev)
+{
+	uint32_t data;
+	int r = 0;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_JPEG) {
+		data = 1 << UVD_PGFSM_CONFIG__UVDJ_PWR_CONFIG__SHIFT;
+		WREG32(SOC15_REG_OFFSET(JPEG, 0, mmUVD_PGFSM_CONFIG), data);
+
+		r = SOC15_WAIT_ON_RREG(JPEG, 0,
+			mmUVD_PGFSM_STATUS, UVD_PGFSM_STATUS_UVDJ_PWR_ON,
+			UVD_PGFSM_STATUS__UVDJ_PWR_STATUS_MASK);
+
+		if (r) {
+			DRM_ERROR("amdgpu: JPEG disable power gating failed\n");
+			return r;
+		}
+	}
+
+	/* Removing the anti hang mechanism to indicate the UVDJ tile is ON */
+	data = RREG32(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JPEG_POWER_STATUS)) & ~0x1;
+	WREG32(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JPEG_POWER_STATUS), data);
+
+	return 0;
+}
+
+static int jpeg_v2_0_enable_power_gating(struct amdgpu_device *adev)
+{
+	if (adev->pg_flags & AMD_PG_SUPPORT_JPEG) {
+		uint32_t data;
+		int r = 0;
+
+		data = RREG32(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JPEG_POWER_STATUS));
+		data &= ~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK;
+		data |=  0x1; //UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_TILES_OFF;
+		WREG32(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JPEG_POWER_STATUS), data);
+
+		data = 2 << UVD_PGFSM_CONFIG__UVDJ_PWR_CONFIG__SHIFT;
+		WREG32(SOC15_REG_OFFSET(JPEG, 0, mmUVD_PGFSM_CONFIG), data);
+
+		r = SOC15_WAIT_ON_RREG(JPEG, 0, mmUVD_PGFSM_STATUS,
+			(2 << UVD_PGFSM_STATUS__UVDJ_PWR_STATUS__SHIFT),
+			UVD_PGFSM_STATUS__UVDJ_PWR_STATUS_MASK);
+
+		if (r) {
+			DRM_ERROR("amdgpu: JPEG enable power gating failed\n");
+			return r;
+		}
+	}
+
+	return 0;
+}
+
+static void jpeg_v2_0_disable_clock_gating(struct amdgpu_device *adev)
+{
+	uint32_t data;
+
+	data = RREG32_SOC15(JPEG, 0, mmJPEG_CGC_CTRL);
+	if (adev->cg_flags & AMD_CG_SUPPORT_JPEG_MGCG)
+		data |= 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		data &= ~JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+
+	data |= 1 << JPEG_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << JPEG_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(JPEG, 0, mmJPEG_CGC_CTRL, data);
+
+	data = RREG32_SOC15(JPEG, 0, mmJPEG_CGC_GATE);
+	data &= ~(JPEG_CGC_GATE__JPEG_DEC_MASK
+		| JPEG_CGC_GATE__JPEG2_DEC_MASK
+		| JPEG_CGC_GATE__JPEG_ENC_MASK
+		| JPEG_CGC_GATE__JMCIF_MASK
+		| JPEG_CGC_GATE__JRBBM_MASK);
+	WREG32_SOC15(JPEG, 0, mmJPEG_CGC_GATE, data);
+}
+
+static void jpeg_v2_0_enable_clock_gating(struct amdgpu_device *adev)
+{
+	uint32_t data;
+
+	data = RREG32_SOC15(JPEG, 0, mmJPEG_CGC_CTRL);
+	if (adev->cg_flags & AMD_CG_SUPPORT_JPEG_MGCG)
+		data |= 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		data |= 0 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+
+	data |= 1 << JPEG_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << JPEG_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(JPEG, 0, mmJPEG_CGC_CTRL, data);
+
+	data = RREG32_SOC15(JPEG, 0, mmJPEG_CGC_GATE);
+	data |= (JPEG_CGC_GATE__JPEG_DEC_MASK
+		|JPEG_CGC_GATE__JPEG2_DEC_MASK
+		|JPEG_CGC_GATE__JPEG_ENC_MASK
+		|JPEG_CGC_GATE__JMCIF_MASK
+		|JPEG_CGC_GATE__JRBBM_MASK);
+	WREG32_SOC15(JPEG, 0, mmJPEG_CGC_GATE, data);
+}
+
+/**
+ * jpeg_v2_0_start - start JPEG block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the JPEG block
+ */
+static int jpeg_v2_0_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = adev->jpeg.inst->ring_dec;
+	int r;
+
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_jpeg(adev, true);
+
+	/* disable power gating */
+	r = jpeg_v2_0_disable_power_gating(adev);
+	if (r)
+		return r;
+
+	/* JPEG disable CGC */
+	jpeg_v2_0_disable_clock_gating(adev);
+
+	WREG32_SOC15(JPEG, 0, mmJPEG_DEC_GFX10_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+
+	/* enable JMI channel */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JMI_CNTL), 0,
+		~UVD_JMI_CNTL__SOFT_RESET_MASK);
+
+	/* enable System Interrupt for JRBC */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, 0, mmJPEG_SYS_INT_EN),
+		JPEG_SYS_INT_EN__DJRBC_MASK,
+		~JPEG_SYS_INT_EN__DJRBC_MASK);
+
+	WREG32_SOC15(JPEG, 0, mmUVD_LMI_JRBC_RB_VMID, 0);
+	WREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_CNTL, (0x00000001L | 0x00000002L));
+	WREG32_SOC15(JPEG, 0, mmUVD_LMI_JRBC_RB_64BIT_BAR_LOW,
+		lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15(JPEG, 0, mmUVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
+		upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_RPTR, 0);
+	WREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_WPTR, 0);
+	WREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_CNTL, 0x00000002L);
+	WREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_SIZE, ring->ring_size / 4);
+	ring->wptr = RREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_WPTR);
+
+	return 0;
+}
+
+/**
+ * jpeg_v2_0_stop - stop JPEG block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * stop the JPEG block
+ */
+static int jpeg_v2_0_stop(struct amdgpu_device *adev)
+{
+	int r;
+
+	/* reset JMI */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JMI_CNTL),
+		UVD_JMI_CNTL__SOFT_RESET_MASK,
+		~UVD_JMI_CNTL__SOFT_RESET_MASK);
+
+	/* enable JPEG CGC */
+	jpeg_v2_0_enable_clock_gating(adev);
+
+	/* enable power gating */
+	r = jpeg_v2_0_enable_power_gating(adev);
+	if (r)
+		return r;
+
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_jpeg(adev, false);
+
+	return 0;
+}
+
+/**
+ * jpeg_v2_0_dec_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t jpeg_v2_0_dec_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_RPTR);
+}
+
+/**
+ * jpeg_v2_0_dec_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t jpeg_v2_0_dec_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell)
+		return *ring->wptr_cpu_addr;
+	else
+		return RREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_WPTR);
+}
+
+/**
+ * jpeg_v2_0_dec_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void jpeg_v2_0_dec_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
+		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+	} else {
+		WREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_WPTR, lower_32_bits(ring->wptr));
+	}
+}
+
+/**
+ * jpeg_v2_0_dec_ring_insert_start - insert a start command
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Write a start command to the ring.
+ */
+void jpeg_v2_0_dec_ring_insert_start(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, PACKETJ(mmUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x68e04);
+
+	amdgpu_ring_write(ring, PACKETJ(JRBC_DEC_EXTERNAL_REG_WRITE_ADDR,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x80010000);
+}
+
+/**
+ * jpeg_v2_0_dec_ring_insert_end - insert a end command
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Write a end command to the ring.
+ */
+void jpeg_v2_0_dec_ring_insert_end(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, PACKETJ(mmUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x68e04);
+
+	amdgpu_ring_write(ring, PACKETJ(JRBC_DEC_EXTERNAL_REG_WRITE_ADDR,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x00010000);
+}
+
+/**
+ * jpeg_v2_0_dec_ring_emit_fence - emit an fence & trap command
+ *
+ * @ring: amdgpu_ring pointer
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
+ *
+ * Write a fence and a trap command to the ring.
+ */
+void jpeg_v2_0_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				unsigned flags)
+{
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	amdgpu_ring_write(ring, PACKETJ(mmUVD_JPEG_GPCOM_DATA0_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, seq);
+
+	amdgpu_ring_write(ring,	PACKETJ(mmUVD_JPEG_GPCOM_DATA1_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, seq);
+
+	amdgpu_ring_write(ring,	PACKETJ(mmUVD_LMI_JRBC_RB_MEM_WR_64BIT_BAR_LOW_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+
+	amdgpu_ring_write(ring,	PACKETJ(mmUVD_LMI_JRBC_RB_MEM_WR_64BIT_BAR_HIGH_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+
+	amdgpu_ring_write(ring,	PACKETJ(mmUVD_JPEG_GPCOM_CMD_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x8);
+
+	amdgpu_ring_write(ring,	PACKETJ(mmUVD_JPEG_GPCOM_CMD_INTERNAL_OFFSET,
+		0, PACKETJ_CONDITION_CHECK0, PACKETJ_TYPE4));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring,	PACKETJ(mmUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x3fbc);
+
+	amdgpu_ring_write(ring, PACKETJ(JRBC_DEC_EXTERNAL_REG_WRITE_ADDR,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x1);
+
+	amdgpu_ring_write(ring, PACKETJ(0, 0, 0, PACKETJ_TYPE7));
+	amdgpu_ring_write(ring, 0);
+}
+
+/**
+ * jpeg_v2_0_dec_ring_emit_ib - execute indirect buffer
+ *
+ * @ring: amdgpu_ring pointer
+ * @job: job to retrieve vmid from
+ * @ib: indirect buffer to execute
+ * @flags: unused
+ *
+ * Write ring commands to execute the indirect buffer.
+ */
+void jpeg_v2_0_dec_ring_emit_ib(struct amdgpu_ring *ring,
+				struct amdgpu_job *job,
+				struct amdgpu_ib *ib,
+				uint32_t flags)
+{
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+
+	amdgpu_ring_write(ring,	PACKETJ(mmUVD_JPEG_IH_CTRL_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, (vmid << JPEG_IH_CTRL__IH_VMID__SHIFT));
+
+	amdgpu_ring_write(ring, PACKETJ(mmUVD_LMI_JRBC_IB_VMID_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+
+	if (ring->funcs->parse_cs)
+		amdgpu_ring_write(ring, 0);
+	else
+		amdgpu_ring_write(ring, (vmid | (vmid << 4) | (vmid << 8)));
+
+	amdgpu_ring_write(ring, PACKETJ(mmUVD_LMI_JPEG_VMID_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, (vmid | (vmid << 4) | (vmid << 8)));
+
+	amdgpu_ring_write(ring,	PACKETJ(mmUVD_LMI_JRBC_IB_64BIT_BAR_LOW_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
+
+	amdgpu_ring_write(ring,	PACKETJ(mmUVD_LMI_JRBC_IB_64BIT_BAR_HIGH_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+
+	amdgpu_ring_write(ring,	PACKETJ(mmUVD_JRBC_IB_SIZE_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, ib->length_dw);
+
+	amdgpu_ring_write(ring,	PACKETJ(mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_LOW_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, lower_32_bits(ring->gpu_addr));
+
+	amdgpu_ring_write(ring,	PACKETJ(mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_HIGH_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, upper_32_bits(ring->gpu_addr));
+
+	amdgpu_ring_write(ring,	PACKETJ(0, 0, PACKETJ_CONDITION_CHECK0, PACKETJ_TYPE2));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring,	PACKETJ(mmUVD_JRBC_RB_COND_RD_TIMER_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x01400200);
+
+	amdgpu_ring_write(ring, PACKETJ(mmUVD_JRBC_RB_REF_DATA_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x2);
+
+	amdgpu_ring_write(ring,	PACKETJ(mmUVD_JRBC_STATUS_INTERNAL_OFFSET,
+		0, PACKETJ_CONDITION_CHECK3, PACKETJ_TYPE3));
+	amdgpu_ring_write(ring, 0x2);
+}
+
+void jpeg_v2_0_dec_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+				uint32_t val, uint32_t mask)
+{
+	uint32_t reg_offset = (reg << 2);
+
+	amdgpu_ring_write(ring, PACKETJ(mmUVD_JRBC_RB_COND_RD_TIMER_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x01400200);
+
+	amdgpu_ring_write(ring,	PACKETJ(mmUVD_JRBC_RB_REF_DATA_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, val);
+
+	amdgpu_ring_write(ring, PACKETJ(mmUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	if (reg_offset >= 0x10000 && reg_offset <= 0x105ff) {
+		amdgpu_ring_write(ring, 0);
+		amdgpu_ring_write(ring,
+			PACKETJ((reg_offset >> 2), 0, 0, PACKETJ_TYPE3));
+	} else {
+		amdgpu_ring_write(ring, reg_offset);
+		amdgpu_ring_write(ring,	PACKETJ(JRBC_DEC_EXTERNAL_REG_WRITE_ADDR,
+			0, 0, PACKETJ_TYPE3));
+	}
+	amdgpu_ring_write(ring, mask);
+}
+
+void jpeg_v2_0_dec_ring_emit_vm_flush(struct amdgpu_ring *ring,
+				unsigned vmid, uint64_t pd_addr)
+{
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
+	uint32_t data0, data1, mask;
+
+	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* wait for register write */
+	data0 = hub->ctx0_ptb_addr_lo32 + vmid * hub->ctx_addr_distance;
+	data1 = lower_32_bits(pd_addr);
+	mask = 0xffffffff;
+	jpeg_v2_0_dec_ring_emit_reg_wait(ring, data0, data1, mask);
+}
+
+void jpeg_v2_0_dec_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg, uint32_t val)
+{
+	uint32_t reg_offset = (reg << 2);
+
+	amdgpu_ring_write(ring,	PACKETJ(mmUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	if (reg_offset >= 0x10000 && reg_offset <= 0x105ff) {
+		amdgpu_ring_write(ring, 0);
+		amdgpu_ring_write(ring,
+			PACKETJ((reg_offset >> 2), 0, 0, PACKETJ_TYPE0));
+	} else {
+		amdgpu_ring_write(ring, reg_offset);
+		amdgpu_ring_write(ring,	PACKETJ(JRBC_DEC_EXTERNAL_REG_WRITE_ADDR,
+			0, 0, PACKETJ_TYPE0));
+	}
+	amdgpu_ring_write(ring, val);
+}
+
+void jpeg_v2_0_dec_ring_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	int i;
+
+	WARN_ON(ring->wptr % 2 || count % 2);
+
+	for (i = 0; i < count / 2; i++) {
+		amdgpu_ring_write(ring, PACKETJ(0, 0, 0, PACKETJ_TYPE6));
+		amdgpu_ring_write(ring, 0);
+	}
+}
+
+static bool jpeg_v2_0_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	return ((RREG32_SOC15(JPEG, 0, mmUVD_JRBC_STATUS) &
+		UVD_JRBC_STATUS__RB_JOB_DONE_MASK) ==
+		UVD_JRBC_STATUS__RB_JOB_DONE_MASK);
+}
+
+static int jpeg_v2_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int ret;
+
+	ret = SOC15_WAIT_ON_RREG(JPEG, 0, mmUVD_JRBC_STATUS, UVD_JRBC_STATUS__RB_JOB_DONE_MASK,
+		UVD_JRBC_STATUS__RB_JOB_DONE_MASK);
+
+	return ret;
+}
+
+static int jpeg_v2_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = (state == AMD_CG_STATE_GATE);
+
+	if (enable) {
+		if (!jpeg_v2_0_is_idle(ip_block))
+			return -EBUSY;
+		jpeg_v2_0_enable_clock_gating(adev);
+	} else {
+		jpeg_v2_0_disable_clock_gating(adev);
+	}
+
+	return 0;
+}
+
+static int jpeg_v2_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int ret;
+
+	if (state == adev->jpeg.cur_state)
+		return 0;
+
+	if (state == AMD_PG_STATE_GATE)
+		ret = jpeg_v2_0_stop(adev);
+	else
+		ret = jpeg_v2_0_start(adev);
+
+	if (!ret)
+		adev->jpeg.cur_state = state;
+
+	return ret;
+}
+
+static int jpeg_v2_0_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+static int jpeg_v2_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	DRM_DEBUG("IH: JPEG TRAP\n");
+
+	switch (entry->src_id) {
+	case VCN_2_0__SRCID__JPEG_DECODE:
+		amdgpu_fence_process(adev->jpeg.inst->ring_dec);
+		break;
+	default:
+		DRM_ERROR("Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+static int jpeg_v2_0_ring_reset(struct amdgpu_ring *ring,
+				unsigned int vmid,
+				struct amdgpu_fence *timedout_fence)
+{
+	int r;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+	r = jpeg_v2_0_stop(ring->adev);
+	if (r)
+		return r;
+	r = jpeg_v2_0_start(ring->adev);
+	if (r)
+		return r;
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static const struct amd_ip_funcs jpeg_v2_0_ip_funcs = {
+	.name = "jpeg_v2_0",
+	.early_init = jpeg_v2_0_early_init,
+	.sw_init = jpeg_v2_0_sw_init,
+	.sw_fini = jpeg_v2_0_sw_fini,
+	.hw_init = jpeg_v2_0_hw_init,
+	.hw_fini = jpeg_v2_0_hw_fini,
+	.suspend = jpeg_v2_0_suspend,
+	.resume = jpeg_v2_0_resume,
+	.is_idle = jpeg_v2_0_is_idle,
+	.wait_for_idle = jpeg_v2_0_wait_for_idle,
+	.set_clockgating_state = jpeg_v2_0_set_clockgating_state,
+	.set_powergating_state = jpeg_v2_0_set_powergating_state,
+	.dump_ip_state = amdgpu_jpeg_dump_ip_state,
+	.print_ip_state = amdgpu_jpeg_print_ip_state,
+};
+
+static const struct amdgpu_ring_funcs jpeg_v2_0_dec_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_JPEG,
+	.align_mask = 0xf,
+	.get_rptr = jpeg_v2_0_dec_ring_get_rptr,
+	.get_wptr = jpeg_v2_0_dec_ring_get_wptr,
+	.set_wptr = jpeg_v2_0_dec_ring_set_wptr,
+	.parse_cs = amdgpu_jpeg_dec_parse_cs,
+	.emit_frame_size =
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
+		8 + /* jpeg_v2_0_dec_ring_emit_vm_flush */
+		18 + 18 + /* jpeg_v2_0_dec_ring_emit_fence x2 vm fence */
+		8 + 16,
+	.emit_ib_size = 24, /* jpeg_v2_0_dec_ring_emit_ib */
+	.emit_ib = jpeg_v2_0_dec_ring_emit_ib,
+	.emit_fence = jpeg_v2_0_dec_ring_emit_fence,
+	.emit_vm_flush = jpeg_v2_0_dec_ring_emit_vm_flush,
+	.test_ring = amdgpu_jpeg_dec_ring_test_ring,
+	.test_ib = amdgpu_jpeg_dec_ring_test_ib,
+	.insert_nop = jpeg_v2_0_dec_ring_nop,
+	.insert_start = jpeg_v2_0_dec_ring_insert_start,
+	.insert_end = jpeg_v2_0_dec_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_jpeg_ring_begin_use,
+	.end_use = amdgpu_jpeg_ring_end_use,
+	.emit_wreg = jpeg_v2_0_dec_ring_emit_wreg,
+	.emit_reg_wait = jpeg_v2_0_dec_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = jpeg_v2_0_ring_reset,
+};
+
+static void jpeg_v2_0_set_dec_ring_funcs(struct amdgpu_device *adev)
+{
+	adev->jpeg.inst->ring_dec->funcs = &jpeg_v2_0_dec_ring_vm_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs jpeg_v2_0_irq_funcs = {
+	.set = jpeg_v2_0_set_interrupt_state,
+	.process = jpeg_v2_0_process_interrupt,
+};
+
+static void jpeg_v2_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->jpeg.inst->irq.num_types = 1;
+	adev->jpeg.inst->irq.funcs = &jpeg_v2_0_irq_funcs;
+}
+
+const struct amdgpu_ip_block_version jpeg_v2_0_ip_block = {
+		.type = AMD_IP_BLOCK_TYPE_JPEG,
+		.major = 2,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &jpeg_v2_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.h b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.h
new file mode 100644
index 000000000000..654e43e83e2c
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.h
@@ -0,0 +1,63 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __JPEG_V2_0_H__
+#define __JPEG_V2_0_H__
+
+#define mmUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET				0x1bfff
+#define mmUVD_JPEG_GPCOM_CMD_INTERNAL_OFFSET				0x4029
+#define mmUVD_JPEG_GPCOM_DATA0_INTERNAL_OFFSET				0x402a
+#define mmUVD_JPEG_GPCOM_DATA1_INTERNAL_OFFSET				0x402b
+#define mmUVD_LMI_JRBC_RB_MEM_WR_64BIT_BAR_LOW_INTERNAL_OFFSET		0x40ea
+#define mmUVD_LMI_JRBC_RB_MEM_WR_64BIT_BAR_HIGH_INTERNAL_OFFSET		0x40eb
+#define mmUVD_LMI_JRBC_IB_VMID_INTERNAL_OFFSET				0x40cf
+#define mmUVD_LMI_JPEG_VMID_INTERNAL_OFFSET				0x40d1
+#define mmUVD_LMI_JRBC_IB_64BIT_BAR_LOW_INTERNAL_OFFSET			0x40e8
+#define mmUVD_LMI_JRBC_IB_64BIT_BAR_HIGH_INTERNAL_OFFSET		0x40e9
+#define mmUVD_JRBC_IB_SIZE_INTERNAL_OFFSET				0x4082
+#define mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_LOW_INTERNAL_OFFSET		0x40ec
+#define mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_HIGH_INTERNAL_OFFSET		0x40ed
+#define mmUVD_JRBC_RB_COND_RD_TIMER_INTERNAL_OFFSET			0x4085
+#define mmUVD_JRBC_RB_REF_DATA_INTERNAL_OFFSET				0x4084
+#define mmUVD_JRBC_STATUS_INTERNAL_OFFSET				0x4089
+#define mmUVD_JPEG_PITCH_INTERNAL_OFFSET				0x401f
+#define mmUVD_JPEG_IH_CTRL_INTERNAL_OFFSET				0x4149
+
+#define JRBC_DEC_EXTERNAL_REG_WRITE_ADDR				0x18000
+
+void jpeg_v2_0_dec_ring_insert_start(struct amdgpu_ring *ring);
+void jpeg_v2_0_dec_ring_insert_end(struct amdgpu_ring *ring);
+void jpeg_v2_0_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				unsigned flags);
+void jpeg_v2_0_dec_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_job *job,
+				struct amdgpu_ib *ib, uint32_t flags);
+void jpeg_v2_0_dec_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+				uint32_t val, uint32_t mask);
+void jpeg_v2_0_dec_ring_emit_vm_flush(struct amdgpu_ring *ring,
+				unsigned vmid, uint64_t pd_addr);
+void jpeg_v2_0_dec_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg, uint32_t val);
+void jpeg_v2_0_dec_ring_nop(struct amdgpu_ring *ring, uint32_t count);
+
+extern const struct amdgpu_ip_block_version jpeg_v2_0_ip_block;
+
+#endif /* __JPEG_V2_0_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c
new file mode 100644
index 000000000000..20983f126b49
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c
@@ -0,0 +1,869 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "amdgpu_jpeg.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "jpeg_v2_0.h"
+#include "jpeg_v2_5.h"
+
+#include "vcn/vcn_2_5_offset.h"
+#include "vcn/vcn_2_5_sh_mask.h"
+#include "ivsrcid/vcn/irqsrcs_vcn_2_0.h"
+
+#define mmUVD_JPEG_PITCH_INTERNAL_OFFSET			0x401f
+
+#define JPEG25_MAX_HW_INSTANCES_ARCTURUS			2
+
+static const struct amdgpu_hwip_reg_entry jpeg_reg_list_2_5[] = {
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_INT_STAT),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_CNTL),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_SIZE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_ADDR_MODE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_GFX10_ADDR_CONFIG),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_Y_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_UV_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_PITCH),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_UV_PITCH),
+};
+
+static void jpeg_v2_5_set_dec_ring_funcs(struct amdgpu_device *adev);
+static void jpeg_v2_5_set_irq_funcs(struct amdgpu_device *adev);
+static int jpeg_v2_5_set_powergating_state(struct amdgpu_ip_block *ip_block,
+				enum amd_powergating_state state);
+static void jpeg_v2_5_set_ras_funcs(struct amdgpu_device *adev);
+
+static int amdgpu_ih_clientid_jpeg[] = {
+	SOC15_IH_CLIENTID_VCN,
+	SOC15_IH_CLIENTID_VCN1
+};
+
+/**
+ * jpeg_v2_5_early_init - set function pointers
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Set ring and irq function pointers
+ */
+static int jpeg_v2_5_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	u32 harvest;
+	int i;
+
+	adev->jpeg.num_jpeg_rings = 1;
+	adev->jpeg.num_jpeg_inst = JPEG25_MAX_HW_INSTANCES_ARCTURUS;
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; i++) {
+		harvest = RREG32_SOC15(JPEG, i, mmCC_UVD_HARVESTING);
+		if (harvest & CC_UVD_HARVESTING__UVD_DISABLE_MASK)
+			adev->jpeg.harvest_config |= 1 << i;
+	}
+	if (adev->jpeg.harvest_config == (AMDGPU_JPEG_HARVEST_JPEG0 |
+					 AMDGPU_JPEG_HARVEST_JPEG1))
+		return -ENOENT;
+
+	jpeg_v2_5_set_dec_ring_funcs(adev);
+	jpeg_v2_5_set_irq_funcs(adev);
+	jpeg_v2_5_set_ras_funcs(adev);
+
+	return 0;
+}
+
+/**
+ * jpeg_v2_5_sw_init - sw init for JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Load firmware and sw initialization
+ */
+static int jpeg_v2_5_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_ring *ring;
+	int i, r;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+
+		/* JPEG TRAP */
+		r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_jpeg[i],
+				VCN_2_0__SRCID__JPEG_DECODE, &adev->jpeg.inst[i].irq);
+		if (r)
+			return r;
+
+		/* JPEG DJPEG POISON EVENT */
+		r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_jpeg[i],
+			VCN_2_6__SRCID_DJPEG0_POISON, &adev->jpeg.inst[i].ras_poison_irq);
+		if (r)
+			return r;
+
+		/* JPEG EJPEG POISON EVENT */
+		r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_jpeg[i],
+			VCN_2_6__SRCID_EJPEG0_POISON, &adev->jpeg.inst[i].ras_poison_irq);
+		if (r)
+			return r;
+	}
+
+	r = amdgpu_jpeg_sw_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_resume(adev);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+
+		ring = adev->jpeg.inst[i].ring_dec;
+		ring->use_doorbell = true;
+		if (amdgpu_ip_version(adev, UVD_HWIP, 0) == IP_VERSION(2, 5, 0))
+			ring->vm_hub = AMDGPU_MMHUB1(0);
+		else
+			ring->vm_hub = AMDGPU_MMHUB0(0);
+		ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 1 + 8 * i;
+		sprintf(ring->name, "jpeg_dec_%d", i);
+		r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst[i].irq,
+				     0, AMDGPU_RING_PRIO_DEFAULT, NULL);
+		if (r)
+			return r;
+
+		adev->jpeg.internal.jpeg_pitch[0] = mmUVD_JPEG_PITCH_INTERNAL_OFFSET;
+		adev->jpeg.inst[i].external.jpeg_pitch[0] = SOC15_REG_OFFSET(JPEG, i, mmUVD_JPEG_PITCH);
+	}
+
+	r = amdgpu_jpeg_ras_sw_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_reg_dump_init(adev, jpeg_reg_list_2_5, ARRAY_SIZE(jpeg_reg_list_2_5));
+	if (r)
+		return r;
+
+	adev->jpeg.supported_reset =
+		amdgpu_get_soft_full_reset_mask(adev->jpeg.inst[0].ring_dec);
+	if (!amdgpu_sriov_vf(adev))
+		adev->jpeg.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+	r = amdgpu_jpeg_sysfs_reset_mask_init(adev);
+
+	return r;
+}
+
+/**
+ * jpeg_v2_5_sw_fini - sw fini for JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * JPEG suspend and free up sw allocation
+ */
+static int jpeg_v2_5_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	r = amdgpu_jpeg_suspend(adev);
+	if (r)
+		return r;
+
+	amdgpu_jpeg_sysfs_reset_mask_fini(adev);
+
+	r = amdgpu_jpeg_sw_fini(adev);
+
+	return r;
+}
+
+/**
+ * jpeg_v2_5_hw_init - start and test JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ */
+static int jpeg_v2_5_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	int i, r;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+
+		ring = adev->jpeg.inst[i].ring_dec;
+		adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
+			(adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 8 * i, i);
+
+		r = amdgpu_ring_test_helper(ring);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+/**
+ * jpeg_v2_5_hw_fini - stop the hardware block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Stop the JPEG block, mark ring as not ready any more
+ */
+static int jpeg_v2_5_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i;
+
+	cancel_delayed_work_sync(&adev->jpeg.idle_work);
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+
+		if (adev->jpeg.cur_state != AMD_PG_STATE_GATE &&
+		      RREG32_SOC15(JPEG, i, mmUVD_JRBC_STATUS))
+			jpeg_v2_5_set_powergating_state(ip_block, AMD_PG_STATE_GATE);
+
+		if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__JPEG))
+			amdgpu_irq_put(adev, &adev->jpeg.inst[i].ras_poison_irq, 0);
+	}
+
+	return 0;
+}
+
+/**
+ * jpeg_v2_5_suspend - suspend JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * HW fini and suspend JPEG block
+ */
+static int jpeg_v2_5_suspend(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	r = jpeg_v2_5_hw_fini(ip_block);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_suspend(ip_block->adev);
+
+	return r;
+}
+
+/**
+ * jpeg_v2_5_resume - resume JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Resume firmware and hw init JPEG block
+ */
+static int jpeg_v2_5_resume(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	r = amdgpu_jpeg_resume(ip_block->adev);
+	if (r)
+		return r;
+
+	r = jpeg_v2_5_hw_init(ip_block);
+
+	return r;
+}
+
+static void jpeg_v2_5_disable_clock_gating(struct amdgpu_device *adev, int inst)
+{
+	uint32_t data;
+
+	data = RREG32_SOC15(JPEG, inst, mmJPEG_CGC_CTRL);
+	if (adev->cg_flags & AMD_CG_SUPPORT_JPEG_MGCG)
+		data |= 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		data &= ~JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+
+	data |= 1 << JPEG_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << JPEG_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(JPEG, inst, mmJPEG_CGC_CTRL, data);
+
+	data = RREG32_SOC15(JPEG, inst, mmJPEG_CGC_GATE);
+	data &= ~(JPEG_CGC_GATE__JPEG_DEC_MASK
+		| JPEG_CGC_GATE__JPEG2_DEC_MASK
+		| JPEG_CGC_GATE__JMCIF_MASK
+		| JPEG_CGC_GATE__JRBBM_MASK);
+	WREG32_SOC15(JPEG, inst, mmJPEG_CGC_GATE, data);
+
+	data = RREG32_SOC15(JPEG, inst, mmJPEG_CGC_CTRL);
+	data &= ~(JPEG_CGC_CTRL__JPEG_DEC_MODE_MASK
+		| JPEG_CGC_CTRL__JPEG2_DEC_MODE_MASK
+		| JPEG_CGC_CTRL__JMCIF_MODE_MASK
+		| JPEG_CGC_CTRL__JRBBM_MODE_MASK);
+	WREG32_SOC15(JPEG, inst, mmJPEG_CGC_CTRL, data);
+}
+
+static void jpeg_v2_5_enable_clock_gating(struct amdgpu_device *adev, int inst)
+{
+	uint32_t data;
+
+	data = RREG32_SOC15(JPEG, inst, mmJPEG_CGC_GATE);
+	data |= (JPEG_CGC_GATE__JPEG_DEC_MASK
+		|JPEG_CGC_GATE__JPEG2_DEC_MASK
+		|JPEG_CGC_GATE__JPEG_ENC_MASK
+		|JPEG_CGC_GATE__JMCIF_MASK
+		|JPEG_CGC_GATE__JRBBM_MASK);
+	WREG32_SOC15(JPEG, inst, mmJPEG_CGC_GATE, data);
+}
+
+static void jpeg_v2_5_start_inst(struct amdgpu_device *adev, int i)
+{
+	struct amdgpu_ring *ring = adev->jpeg.inst[i].ring_dec;
+	/* disable anti hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, i, mmUVD_JPEG_POWER_STATUS), 0,
+		~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+
+	/* JPEG disable CGC */
+	jpeg_v2_5_disable_clock_gating(adev, i);
+
+	/* MJPEG global tiling registers */
+	WREG32_SOC15(JPEG, i, mmJPEG_DEC_GFX8_ADDR_CONFIG,
+		adev->gfx.config.gb_addr_config);
+	WREG32_SOC15(JPEG, i, mmJPEG_DEC_GFX10_ADDR_CONFIG,
+		adev->gfx.config.gb_addr_config);
+
+	/* enable JMI channel */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, i, mmUVD_JMI_CNTL), 0,
+		~UVD_JMI_CNTL__SOFT_RESET_MASK);
+
+	/* enable System Interrupt for JRBC */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, i, mmJPEG_SYS_INT_EN),
+		JPEG_SYS_INT_EN__DJRBC_MASK,
+		~JPEG_SYS_INT_EN__DJRBC_MASK);
+
+	WREG32_SOC15(JPEG, i, mmUVD_LMI_JRBC_RB_VMID, 0);
+	WREG32_SOC15(JPEG, i, mmUVD_JRBC_RB_CNTL, (0x00000001L | 0x00000002L));
+	WREG32_SOC15(JPEG, i, mmUVD_LMI_JRBC_RB_64BIT_BAR_LOW,
+		lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15(JPEG, i, mmUVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
+		upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(JPEG, i, mmUVD_JRBC_RB_RPTR, 0);
+	WREG32_SOC15(JPEG, i, mmUVD_JRBC_RB_WPTR, 0);
+	WREG32_SOC15(JPEG, i, mmUVD_JRBC_RB_CNTL, 0x00000002L);
+	WREG32_SOC15(JPEG, i, mmUVD_JRBC_RB_SIZE, ring->ring_size / 4);
+	ring->wptr = RREG32_SOC15(JPEG, i, mmUVD_JRBC_RB_WPTR);
+}
+
+/**
+ * jpeg_v2_5_start - start JPEG block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the JPEG block
+ */
+static int jpeg_v2_5_start(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+		jpeg_v2_5_start_inst(adev, i);
+
+	}
+
+	return 0;
+}
+
+static void jpeg_v2_5_stop_inst(struct amdgpu_device *adev, int i)
+{
+	/* reset JMI */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, i, mmUVD_JMI_CNTL),
+		UVD_JMI_CNTL__SOFT_RESET_MASK,
+		~UVD_JMI_CNTL__SOFT_RESET_MASK);
+
+	jpeg_v2_5_enable_clock_gating(adev, i);
+
+	/* enable anti hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, i, mmUVD_JPEG_POWER_STATUS),
+		UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK,
+		~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+}
+
+/**
+ * jpeg_v2_5_stop - stop JPEG block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * stop the JPEG block
+ */
+static int jpeg_v2_5_stop(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+		jpeg_v2_5_stop_inst(adev, i);
+	}
+
+	return 0;
+}
+
+/**
+ * jpeg_v2_5_dec_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t jpeg_v2_5_dec_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32_SOC15(JPEG, ring->me, mmUVD_JRBC_RB_RPTR);
+}
+
+/**
+ * jpeg_v2_5_dec_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t jpeg_v2_5_dec_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell)
+		return *ring->wptr_cpu_addr;
+	else
+		return RREG32_SOC15(JPEG, ring->me, mmUVD_JRBC_RB_WPTR);
+}
+
+/**
+ * jpeg_v2_5_dec_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void jpeg_v2_5_dec_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
+		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+	} else {
+		WREG32_SOC15(JPEG, ring->me, mmUVD_JRBC_RB_WPTR, lower_32_bits(ring->wptr));
+	}
+}
+
+/**
+ * jpeg_v2_6_dec_ring_insert_start - insert a start command
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Write a start command to the ring.
+ */
+static void jpeg_v2_6_dec_ring_insert_start(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, PACKETJ(mmUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x6aa04); /* PCTL0_MMHUB_DEEPSLEEP_IB */
+
+	amdgpu_ring_write(ring, PACKETJ(JRBC_DEC_EXTERNAL_REG_WRITE_ADDR,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x80000000 | (1 << (ring->me * 2 + 14)));
+}
+
+/**
+ * jpeg_v2_6_dec_ring_insert_end - insert a end command
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Write a end command to the ring.
+ */
+static void jpeg_v2_6_dec_ring_insert_end(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, PACKETJ(mmUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x6aa04); /* PCTL0_MMHUB_DEEPSLEEP_IB */
+
+	amdgpu_ring_write(ring, PACKETJ(JRBC_DEC_EXTERNAL_REG_WRITE_ADDR,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, (1 << (ring->me * 2 + 14)));
+}
+
+static bool jpeg_v2_5_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, ret = 1;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+
+		ret &= (((RREG32_SOC15(JPEG, i, mmUVD_JRBC_STATUS) &
+			UVD_JRBC_STATUS__RB_JOB_DONE_MASK) ==
+			UVD_JRBC_STATUS__RB_JOB_DONE_MASK));
+	}
+
+	return ret;
+}
+
+static int jpeg_v2_5_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, ret;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+
+		ret = SOC15_WAIT_ON_RREG(JPEG, i, mmUVD_JRBC_STATUS,
+			UVD_JRBC_STATUS__RB_JOB_DONE_MASK,
+			UVD_JRBC_STATUS__RB_JOB_DONE_MASK);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int jpeg_v2_5_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = (state == AMD_CG_STATE_GATE);
+	int i;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+
+		if (enable) {
+			if (!jpeg_v2_5_is_idle(ip_block))
+				return -EBUSY;
+			jpeg_v2_5_enable_clock_gating(adev, i);
+		} else {
+			jpeg_v2_5_disable_clock_gating(adev, i);
+		}
+	}
+
+	return 0;
+}
+
+static int jpeg_v2_5_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int ret;
+
+	if (state == adev->jpeg.cur_state)
+		return 0;
+
+	if (state == AMD_PG_STATE_GATE)
+		ret = jpeg_v2_5_stop(adev);
+	else
+		ret = jpeg_v2_5_start(adev);
+
+	if (!ret)
+		adev->jpeg.cur_state = state;
+
+	return ret;
+}
+
+static int jpeg_v2_5_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+static int jpeg_v2_6_set_ras_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned int type,
+					enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+static int jpeg_v2_5_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	uint32_t ip_instance;
+
+	switch (entry->client_id) {
+	case SOC15_IH_CLIENTID_VCN:
+		ip_instance = 0;
+		break;
+	case SOC15_IH_CLIENTID_VCN1:
+		ip_instance = 1;
+		break;
+	default:
+		DRM_ERROR("Unhandled client id: %d\n", entry->client_id);
+		return 0;
+	}
+
+	DRM_DEBUG("IH: JPEG TRAP\n");
+
+	switch (entry->src_id) {
+	case VCN_2_0__SRCID__JPEG_DECODE:
+		amdgpu_fence_process(adev->jpeg.inst[ip_instance].ring_dec);
+		break;
+	default:
+		DRM_ERROR("Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+static int jpeg_v2_5_ring_reset(struct amdgpu_ring *ring,
+				unsigned int vmid,
+				struct amdgpu_fence *timedout_fence)
+{
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+	jpeg_v2_5_stop_inst(ring->adev, ring->me);
+	jpeg_v2_5_start_inst(ring->adev, ring->me);
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static const struct amd_ip_funcs jpeg_v2_5_ip_funcs = {
+	.name = "jpeg_v2_5",
+	.early_init = jpeg_v2_5_early_init,
+	.sw_init = jpeg_v2_5_sw_init,
+	.sw_fini = jpeg_v2_5_sw_fini,
+	.hw_init = jpeg_v2_5_hw_init,
+	.hw_fini = jpeg_v2_5_hw_fini,
+	.suspend = jpeg_v2_5_suspend,
+	.resume = jpeg_v2_5_resume,
+	.is_idle = jpeg_v2_5_is_idle,
+	.wait_for_idle = jpeg_v2_5_wait_for_idle,
+	.set_clockgating_state = jpeg_v2_5_set_clockgating_state,
+	.set_powergating_state = jpeg_v2_5_set_powergating_state,
+	.dump_ip_state = amdgpu_jpeg_dump_ip_state,
+	.print_ip_state = amdgpu_jpeg_print_ip_state,
+};
+
+static const struct amd_ip_funcs jpeg_v2_6_ip_funcs = {
+	.name = "jpeg_v2_6",
+	.early_init = jpeg_v2_5_early_init,
+	.sw_init = jpeg_v2_5_sw_init,
+	.sw_fini = jpeg_v2_5_sw_fini,
+	.hw_init = jpeg_v2_5_hw_init,
+	.hw_fini = jpeg_v2_5_hw_fini,
+	.suspend = jpeg_v2_5_suspend,
+	.resume = jpeg_v2_5_resume,
+	.is_idle = jpeg_v2_5_is_idle,
+	.wait_for_idle = jpeg_v2_5_wait_for_idle,
+	.set_clockgating_state = jpeg_v2_5_set_clockgating_state,
+	.set_powergating_state = jpeg_v2_5_set_powergating_state,
+	.dump_ip_state = amdgpu_jpeg_dump_ip_state,
+	.print_ip_state = amdgpu_jpeg_print_ip_state,
+};
+
+static const struct amdgpu_ring_funcs jpeg_v2_5_dec_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_JPEG,
+	.align_mask = 0xf,
+	.get_rptr = jpeg_v2_5_dec_ring_get_rptr,
+	.get_wptr = jpeg_v2_5_dec_ring_get_wptr,
+	.set_wptr = jpeg_v2_5_dec_ring_set_wptr,
+	.parse_cs = amdgpu_jpeg_dec_parse_cs,
+	.emit_frame_size =
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
+		8 + /* jpeg_v2_5_dec_ring_emit_vm_flush */
+		18 + 18 + /* jpeg_v2_5_dec_ring_emit_fence x2 vm fence */
+		8 + 16,
+	.emit_ib_size = 22, /* jpeg_v2_5_dec_ring_emit_ib */
+	.emit_ib = jpeg_v2_0_dec_ring_emit_ib,
+	.emit_fence = jpeg_v2_0_dec_ring_emit_fence,
+	.emit_vm_flush = jpeg_v2_0_dec_ring_emit_vm_flush,
+	.test_ring = amdgpu_jpeg_dec_ring_test_ring,
+	.test_ib = amdgpu_jpeg_dec_ring_test_ib,
+	.insert_nop = jpeg_v2_0_dec_ring_nop,
+	.insert_start = jpeg_v2_0_dec_ring_insert_start,
+	.insert_end = jpeg_v2_0_dec_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_jpeg_ring_begin_use,
+	.end_use = amdgpu_jpeg_ring_end_use,
+	.emit_wreg = jpeg_v2_0_dec_ring_emit_wreg,
+	.emit_reg_wait = jpeg_v2_0_dec_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = jpeg_v2_5_ring_reset,
+};
+
+static const struct amdgpu_ring_funcs jpeg_v2_6_dec_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_JPEG,
+	.align_mask = 0xf,
+	.get_rptr = jpeg_v2_5_dec_ring_get_rptr,
+	.get_wptr = jpeg_v2_5_dec_ring_get_wptr,
+	.set_wptr = jpeg_v2_5_dec_ring_set_wptr,
+	.parse_cs = amdgpu_jpeg_dec_parse_cs,
+	.emit_frame_size =
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
+		8 + /* jpeg_v2_5_dec_ring_emit_vm_flush */
+		18 + 18 + /* jpeg_v2_5_dec_ring_emit_fence x2 vm fence */
+		8 + 16,
+	.emit_ib_size = 22, /* jpeg_v2_5_dec_ring_emit_ib */
+	.emit_ib = jpeg_v2_0_dec_ring_emit_ib,
+	.emit_fence = jpeg_v2_0_dec_ring_emit_fence,
+	.emit_vm_flush = jpeg_v2_0_dec_ring_emit_vm_flush,
+	.test_ring = amdgpu_jpeg_dec_ring_test_ring,
+	.test_ib = amdgpu_jpeg_dec_ring_test_ib,
+	.insert_nop = jpeg_v2_0_dec_ring_nop,
+	.insert_start = jpeg_v2_6_dec_ring_insert_start,
+	.insert_end = jpeg_v2_6_dec_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_jpeg_ring_begin_use,
+	.end_use = amdgpu_jpeg_ring_end_use,
+	.emit_wreg = jpeg_v2_0_dec_ring_emit_wreg,
+	.emit_reg_wait = jpeg_v2_0_dec_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = jpeg_v2_5_ring_reset,
+};
+
+static void jpeg_v2_5_set_dec_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+		if (adev->asic_type == CHIP_ARCTURUS)
+			adev->jpeg.inst[i].ring_dec->funcs = &jpeg_v2_5_dec_ring_vm_funcs;
+		else  /* CHIP_ALDEBARAN */
+			adev->jpeg.inst[i].ring_dec->funcs = &jpeg_v2_6_dec_ring_vm_funcs;
+		adev->jpeg.inst[i].ring_dec->me = i;
+	}
+}
+
+static const struct amdgpu_irq_src_funcs jpeg_v2_5_irq_funcs = {
+	.set = jpeg_v2_5_set_interrupt_state,
+	.process = jpeg_v2_5_process_interrupt,
+};
+
+static const struct amdgpu_irq_src_funcs jpeg_v2_6_ras_irq_funcs = {
+	.set = jpeg_v2_6_set_ras_interrupt_state,
+	.process = amdgpu_jpeg_process_poison_irq,
+};
+
+static void jpeg_v2_5_set_irq_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+
+		adev->jpeg.inst[i].irq.num_types = 1;
+		adev->jpeg.inst[i].irq.funcs = &jpeg_v2_5_irq_funcs;
+
+		adev->jpeg.inst[i].ras_poison_irq.num_types = 1;
+		adev->jpeg.inst[i].ras_poison_irq.funcs = &jpeg_v2_6_ras_irq_funcs;
+	}
+}
+
+const struct amdgpu_ip_block_version jpeg_v2_5_ip_block = {
+		.type = AMD_IP_BLOCK_TYPE_JPEG,
+		.major = 2,
+		.minor = 5,
+		.rev = 0,
+		.funcs = &jpeg_v2_5_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version jpeg_v2_6_ip_block = {
+		.type = AMD_IP_BLOCK_TYPE_JPEG,
+		.major = 2,
+		.minor = 6,
+		.rev = 0,
+		.funcs = &jpeg_v2_6_ip_funcs,
+};
+
+static uint32_t jpeg_v2_6_query_poison_by_instance(struct amdgpu_device *adev,
+		uint32_t instance, uint32_t sub_block)
+{
+	uint32_t poison_stat = 0, reg_value = 0;
+
+	switch (sub_block) {
+	case AMDGPU_JPEG_V2_6_JPEG0:
+		reg_value = RREG32_SOC15(JPEG, instance, mmUVD_RAS_JPEG0_STATUS);
+		poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_JPEG0_STATUS, POISONED_PF);
+		break;
+	case AMDGPU_JPEG_V2_6_JPEG1:
+		reg_value = RREG32_SOC15(JPEG, instance, mmUVD_RAS_JPEG1_STATUS);
+		poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_JPEG1_STATUS, POISONED_PF);
+		break;
+	default:
+		break;
+	}
+
+	if (poison_stat)
+		dev_info(adev->dev, "Poison detected in JPEG%d sub_block%d\n",
+			instance, sub_block);
+
+	return poison_stat;
+}
+
+static bool jpeg_v2_6_query_ras_poison_status(struct amdgpu_device *adev)
+{
+	uint32_t inst = 0, sub = 0, poison_stat = 0;
+
+	for (inst = 0; inst < adev->jpeg.num_jpeg_inst; inst++)
+		for (sub = 0; sub < AMDGPU_JPEG_V2_6_MAX_SUB_BLOCK; sub++)
+			poison_stat +=
+			jpeg_v2_6_query_poison_by_instance(adev, inst, sub);
+
+	return !!poison_stat;
+}
+
+const struct amdgpu_ras_block_hw_ops jpeg_v2_6_ras_hw_ops = {
+	.query_poison_status = jpeg_v2_6_query_ras_poison_status,
+};
+
+static struct amdgpu_jpeg_ras jpeg_v2_6_ras = {
+	.ras_block = {
+		.hw_ops = &jpeg_v2_6_ras_hw_ops,
+		.ras_late_init = amdgpu_jpeg_ras_late_init,
+	},
+};
+
+static void jpeg_v2_5_set_ras_funcs(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, JPEG_HWIP, 0)) {
+	case IP_VERSION(2, 6, 0):
+		adev->jpeg.ras = &jpeg_v2_6_ras;
+		break;
+	default:
+		break;
+	}
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.h b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.h
new file mode 100644
index 000000000000..1e858c6cdf13
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.h
@@ -0,0 +1,37 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __JPEG_V2_5_H__
+#define __JPEG_V2_5_H__
+
+enum amdgpu_jpeg_v2_6_sub_block {
+	AMDGPU_JPEG_V2_6_JPEG0 = 0,
+	AMDGPU_JPEG_V2_6_JPEG1,
+
+	AMDGPU_JPEG_V2_6_MAX_SUB_BLOCK,
+};
+
+extern const struct amdgpu_ip_block_version jpeg_v2_5_ip_block;
+extern const struct amdgpu_ip_block_version jpeg_v2_6_ip_block;
+
+#endif /* __JPEG_V2_5_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c
new file mode 100644
index 000000000000..d1a011c40ba2
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c
@@ -0,0 +1,648 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "amdgpu_jpeg.h"
+#include "amdgpu_pm.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "jpeg_v2_0.h"
+
+#include "vcn/vcn_3_0_0_offset.h"
+#include "vcn/vcn_3_0_0_sh_mask.h"
+#include "ivsrcid/vcn/irqsrcs_vcn_2_0.h"
+
+#define mmUVD_JPEG_PITCH_INTERNAL_OFFSET	0x401f
+
+static const struct amdgpu_hwip_reg_entry jpeg_reg_list_3_0[] = {
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_INT_STAT),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_CNTL),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_SIZE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_ADDR_MODE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_GFX10_ADDR_CONFIG),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_Y_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_UV_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_PITCH),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_UV_PITCH),
+};
+
+static void jpeg_v3_0_set_dec_ring_funcs(struct amdgpu_device *adev);
+static void jpeg_v3_0_set_irq_funcs(struct amdgpu_device *adev);
+static int jpeg_v3_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
+				enum amd_powergating_state state);
+
+/**
+ * jpeg_v3_0_early_init - set function pointers
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Set ring and irq function pointers
+ */
+static int jpeg_v3_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	u32 harvest;
+
+	switch (amdgpu_ip_version(adev, UVD_HWIP, 0)) {
+	case IP_VERSION(3, 1, 1):
+	case IP_VERSION(3, 1, 2):
+		break;
+	default:
+		harvest = RREG32_SOC15(JPEG, 0, mmCC_UVD_HARVESTING);
+		if (harvest & CC_UVD_HARVESTING__UVD_DISABLE_MASK)
+			return -ENOENT;
+		break;
+	}
+
+	adev->jpeg.num_jpeg_inst = 1;
+	adev->jpeg.num_jpeg_rings = 1;
+
+	jpeg_v3_0_set_dec_ring_funcs(adev);
+	jpeg_v3_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+/**
+ * jpeg_v3_0_sw_init - sw init for JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Load firmware and sw initialization
+ */
+static int jpeg_v3_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	int r;
+
+	/* JPEG TRAP */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+		VCN_2_0__SRCID__JPEG_DECODE, &adev->jpeg.inst->irq);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_sw_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_resume(adev);
+	if (r)
+		return r;
+
+	ring = adev->jpeg.inst->ring_dec;
+	ring->use_doorbell = true;
+	ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 1;
+	ring->vm_hub = AMDGPU_MMHUB0(0);
+	sprintf(ring->name, "jpeg_dec");
+	r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst->irq, 0,
+			     AMDGPU_RING_PRIO_DEFAULT, NULL);
+	if (r)
+		return r;
+
+	adev->jpeg.internal.jpeg_pitch[0] = mmUVD_JPEG_PITCH_INTERNAL_OFFSET;
+	adev->jpeg.inst->external.jpeg_pitch[0] = SOC15_REG_OFFSET(JPEG, 0, mmUVD_JPEG_PITCH);
+
+	r = amdgpu_jpeg_reg_dump_init(adev, jpeg_reg_list_3_0, ARRAY_SIZE(jpeg_reg_list_3_0));
+	if (r)
+		return r;
+
+	adev->jpeg.supported_reset =
+		amdgpu_get_soft_full_reset_mask(adev->jpeg.inst[0].ring_dec);
+	if (!amdgpu_sriov_vf(adev))
+		adev->jpeg.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+	r = amdgpu_jpeg_sysfs_reset_mask_init(adev);
+
+	return r;
+}
+
+/**
+ * jpeg_v3_0_sw_fini - sw fini for JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * JPEG suspend and free up sw allocation
+ */
+static int jpeg_v3_0_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = amdgpu_jpeg_suspend(adev);
+	if (r)
+		return r;
+
+	amdgpu_jpeg_sysfs_reset_mask_fini(adev);
+
+	r = amdgpu_jpeg_sw_fini(adev);
+
+	return r;
+}
+
+/**
+ * jpeg_v3_0_hw_init - start and test JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ */
+static int jpeg_v3_0_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring = adev->jpeg.inst->ring_dec;
+
+	adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
+		(adev->doorbell_index.vcn.vcn_ring0_1 << 1), 0);
+
+	return amdgpu_ring_test_helper(ring);
+}
+
+/**
+ * jpeg_v3_0_hw_fini - stop the hardware block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Stop the JPEG block, mark ring as not ready any more
+ */
+static int jpeg_v3_0_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	cancel_delayed_work_sync(&adev->jpeg.idle_work);
+
+	if (adev->jpeg.cur_state != AMD_PG_STATE_GATE &&
+	      RREG32_SOC15(JPEG, 0, mmUVD_JRBC_STATUS))
+		jpeg_v3_0_set_powergating_state(ip_block, AMD_PG_STATE_GATE);
+
+	return 0;
+}
+
+/**
+ * jpeg_v3_0_suspend - suspend JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * HW fini and suspend JPEG block
+ */
+static int jpeg_v3_0_suspend(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	r = jpeg_v3_0_hw_fini(ip_block);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_suspend(ip_block->adev);
+
+	return r;
+}
+
+/**
+ * jpeg_v3_0_resume - resume JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Resume firmware and hw init JPEG block
+ */
+static int jpeg_v3_0_resume(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	r = amdgpu_jpeg_resume(ip_block->adev);
+	if (r)
+		return r;
+
+	r = jpeg_v3_0_hw_init(ip_block);
+
+	return r;
+}
+
+static void jpeg_v3_0_disable_clock_gating(struct amdgpu_device *adev)
+{
+	uint32_t data = 0;
+
+	data = RREG32_SOC15(JPEG, 0, mmJPEG_CGC_CTRL);
+	if (adev->cg_flags & AMD_CG_SUPPORT_JPEG_MGCG)
+		data |= 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		data &= ~JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+
+	data |= 1 << JPEG_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << JPEG_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(JPEG, 0, mmJPEG_CGC_CTRL, data);
+
+	data = RREG32_SOC15(JPEG, 0, mmJPEG_CGC_GATE);
+	data &= ~(JPEG_CGC_GATE__JPEG_DEC_MASK
+		| JPEG_CGC_GATE__JPEG2_DEC_MASK
+		| JPEG_CGC_GATE__JPEG_ENC_MASK
+		| JPEG_CGC_GATE__JMCIF_MASK
+		| JPEG_CGC_GATE__JRBBM_MASK);
+	WREG32_SOC15(JPEG, 0, mmJPEG_CGC_GATE, data);
+
+	data = RREG32_SOC15(JPEG, 0, mmJPEG_CGC_CTRL);
+	data &= ~(JPEG_CGC_CTRL__JPEG_DEC_MODE_MASK
+		| JPEG_CGC_CTRL__JPEG2_DEC_MODE_MASK
+		| JPEG_CGC_CTRL__JMCIF_MODE_MASK
+		| JPEG_CGC_CTRL__JRBBM_MODE_MASK);
+	WREG32_SOC15(JPEG, 0, mmJPEG_CGC_CTRL, data);
+}
+
+static void jpeg_v3_0_enable_clock_gating(struct amdgpu_device *adev)
+{
+	uint32_t data = 0;
+
+	data = RREG32_SOC15(JPEG, 0, mmJPEG_CGC_GATE);
+	data |= (JPEG_CGC_GATE__JPEG_DEC_MASK
+		|JPEG_CGC_GATE__JPEG2_DEC_MASK
+		|JPEG_CGC_GATE__JPEG_ENC_MASK
+		|JPEG_CGC_GATE__JMCIF_MASK
+		|JPEG_CGC_GATE__JRBBM_MASK);
+	WREG32_SOC15(JPEG, 0, mmJPEG_CGC_GATE, data);
+}
+
+static int jpeg_v3_0_disable_static_power_gating(struct amdgpu_device *adev)
+{
+	if (adev->pg_flags & AMD_PG_SUPPORT_JPEG) {
+		uint32_t data = 0;
+		int r = 0;
+
+		data = 1 << UVD_PGFSM_CONFIG__UVDJ_PWR_CONFIG__SHIFT;
+		WREG32(SOC15_REG_OFFSET(JPEG, 0, mmUVD_PGFSM_CONFIG), data);
+
+		r = SOC15_WAIT_ON_RREG(JPEG, 0,
+			mmUVD_PGFSM_STATUS, UVD_PGFSM_STATUS_UVDJ_PWR_ON,
+			UVD_PGFSM_STATUS__UVDJ_PWR_STATUS_MASK);
+
+		if (r) {
+			DRM_ERROR("amdgpu: JPEG disable power gating failed\n");
+			return r;
+		}
+	}
+
+	/* disable anti hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JPEG_POWER_STATUS), 0,
+		~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+
+	/* keep the JPEG in static PG mode */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JPEG_POWER_STATUS), 0,
+		~UVD_JPEG_POWER_STATUS__JPEG_PG_MODE_MASK);
+
+	return 0;
+}
+
+static int jpeg_v3_0_enable_static_power_gating(struct amdgpu_device *adev)
+{
+	/* enable anti hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JPEG_POWER_STATUS),
+		UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK,
+		~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_JPEG) {
+		uint32_t data = 0;
+		int r = 0;
+
+		data = 2 << UVD_PGFSM_CONFIG__UVDJ_PWR_CONFIG__SHIFT;
+		WREG32(SOC15_REG_OFFSET(JPEG, 0, mmUVD_PGFSM_CONFIG), data);
+
+		r = SOC15_WAIT_ON_RREG(JPEG, 0, mmUVD_PGFSM_STATUS,
+			(2 << UVD_PGFSM_STATUS__UVDJ_PWR_STATUS__SHIFT),
+			UVD_PGFSM_STATUS__UVDJ_PWR_STATUS_MASK);
+
+		if (r) {
+			DRM_ERROR("amdgpu: JPEG enable power gating failed\n");
+			return r;
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * jpeg_v3_0_start - start JPEG block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the JPEG block
+ */
+static int jpeg_v3_0_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = adev->jpeg.inst->ring_dec;
+	int r;
+
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_jpeg(adev, true);
+
+	/* disable power gating */
+	r = jpeg_v3_0_disable_static_power_gating(adev);
+	if (r)
+		return r;
+
+	/* JPEG disable CGC */
+	jpeg_v3_0_disable_clock_gating(adev);
+
+	/* MJPEG global tiling registers */
+	WREG32_SOC15(JPEG, 0, mmJPEG_DEC_GFX10_ADDR_CONFIG,
+		adev->gfx.config.gb_addr_config);
+	WREG32_SOC15(JPEG, 0, mmJPEG_ENC_GFX10_ADDR_CONFIG,
+		adev->gfx.config.gb_addr_config);
+
+	/* enable JMI channel */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JMI_CNTL), 0,
+		~UVD_JMI_CNTL__SOFT_RESET_MASK);
+
+	/* enable System Interrupt for JRBC */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, 0, mmJPEG_SYS_INT_EN),
+		JPEG_SYS_INT_EN__DJRBC_MASK,
+		~JPEG_SYS_INT_EN__DJRBC_MASK);
+
+	WREG32_SOC15(JPEG, 0, mmUVD_LMI_JRBC_RB_VMID, 0);
+	WREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_CNTL, (0x00000001L | 0x00000002L));
+	WREG32_SOC15(JPEG, 0, mmUVD_LMI_JRBC_RB_64BIT_BAR_LOW,
+		lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15(JPEG, 0, mmUVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
+		upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_RPTR, 0);
+	WREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_WPTR, 0);
+	WREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_CNTL, 0x00000002L);
+	WREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_SIZE, ring->ring_size / 4);
+	ring->wptr = RREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_WPTR);
+
+	return 0;
+}
+
+/**
+ * jpeg_v3_0_stop - stop JPEG block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * stop the JPEG block
+ */
+static int jpeg_v3_0_stop(struct amdgpu_device *adev)
+{
+	int r;
+
+	/* reset JMI */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JMI_CNTL),
+		UVD_JMI_CNTL__SOFT_RESET_MASK,
+		~UVD_JMI_CNTL__SOFT_RESET_MASK);
+
+	jpeg_v3_0_enable_clock_gating(adev);
+
+	/* enable power gating */
+	r = jpeg_v3_0_enable_static_power_gating(adev);
+	if (r)
+		return r;
+
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_jpeg(adev, false);
+
+	return 0;
+}
+
+/**
+ * jpeg_v3_0_dec_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t jpeg_v3_0_dec_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_RPTR);
+}
+
+/**
+ * jpeg_v3_0_dec_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t jpeg_v3_0_dec_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell)
+		return *ring->wptr_cpu_addr;
+	else
+		return RREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_WPTR);
+}
+
+/**
+ * jpeg_v3_0_dec_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void jpeg_v3_0_dec_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
+		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+	} else {
+		WREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_WPTR, lower_32_bits(ring->wptr));
+	}
+}
+
+static bool jpeg_v3_0_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int ret = 1;
+
+	ret &= (((RREG32_SOC15(JPEG, 0, mmUVD_JRBC_STATUS) &
+		UVD_JRBC_STATUS__RB_JOB_DONE_MASK) ==
+		UVD_JRBC_STATUS__RB_JOB_DONE_MASK));
+
+	return ret;
+}
+
+static int jpeg_v3_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	return SOC15_WAIT_ON_RREG(JPEG, 0, mmUVD_JRBC_STATUS,
+		UVD_JRBC_STATUS__RB_JOB_DONE_MASK,
+		UVD_JRBC_STATUS__RB_JOB_DONE_MASK);
+}
+
+static int jpeg_v3_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = state == AMD_CG_STATE_GATE;
+
+	if (enable) {
+		if (!jpeg_v3_0_is_idle(ip_block))
+			return -EBUSY;
+		jpeg_v3_0_enable_clock_gating(adev);
+	} else {
+		jpeg_v3_0_disable_clock_gating(adev);
+	}
+
+	return 0;
+}
+
+static int jpeg_v3_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int ret;
+
+	if(state == adev->jpeg.cur_state)
+		return 0;
+
+	if (state == AMD_PG_STATE_GATE)
+		ret = jpeg_v3_0_stop(adev);
+	else
+		ret = jpeg_v3_0_start(adev);
+
+	if(!ret)
+		adev->jpeg.cur_state = state;
+
+	return ret;
+}
+
+static int jpeg_v3_0_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+static int jpeg_v3_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	DRM_DEBUG("IH: JPEG TRAP\n");
+
+	switch (entry->src_id) {
+	case VCN_2_0__SRCID__JPEG_DECODE:
+		amdgpu_fence_process(adev->jpeg.inst->ring_dec);
+		break;
+	default:
+		DRM_ERROR("Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+static int jpeg_v3_0_ring_reset(struct amdgpu_ring *ring,
+				unsigned int vmid,
+				struct amdgpu_fence *timedout_fence)
+{
+	int r;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+	r = jpeg_v3_0_stop(ring->adev);
+	if (r)
+		return r;
+	r = jpeg_v3_0_start(ring->adev);
+	if (r)
+		return r;
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static const struct amd_ip_funcs jpeg_v3_0_ip_funcs = {
+	.name = "jpeg_v3_0",
+	.early_init = jpeg_v3_0_early_init,
+	.sw_init = jpeg_v3_0_sw_init,
+	.sw_fini = jpeg_v3_0_sw_fini,
+	.hw_init = jpeg_v3_0_hw_init,
+	.hw_fini = jpeg_v3_0_hw_fini,
+	.suspend = jpeg_v3_0_suspend,
+	.resume = jpeg_v3_0_resume,
+	.is_idle = jpeg_v3_0_is_idle,
+	.wait_for_idle = jpeg_v3_0_wait_for_idle,
+	.set_clockgating_state = jpeg_v3_0_set_clockgating_state,
+	.set_powergating_state = jpeg_v3_0_set_powergating_state,
+	.dump_ip_state = amdgpu_jpeg_dump_ip_state,
+	.print_ip_state = amdgpu_jpeg_print_ip_state,
+};
+
+static const struct amdgpu_ring_funcs jpeg_v3_0_dec_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_JPEG,
+	.align_mask = 0xf,
+	.get_rptr = jpeg_v3_0_dec_ring_get_rptr,
+	.get_wptr = jpeg_v3_0_dec_ring_get_wptr,
+	.set_wptr = jpeg_v3_0_dec_ring_set_wptr,
+	.parse_cs = amdgpu_jpeg_dec_parse_cs,
+	.emit_frame_size =
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
+		8 + /* jpeg_v3_0_dec_ring_emit_vm_flush */
+		18 + 18 + /* jpeg_v3_0_dec_ring_emit_fence x2 vm fence */
+		8 + 16,
+	.emit_ib_size = 22, /* jpeg_v3_0_dec_ring_emit_ib */
+	.emit_ib = jpeg_v2_0_dec_ring_emit_ib,
+	.emit_fence = jpeg_v2_0_dec_ring_emit_fence,
+	.emit_vm_flush = jpeg_v2_0_dec_ring_emit_vm_flush,
+	.test_ring = amdgpu_jpeg_dec_ring_test_ring,
+	.test_ib = amdgpu_jpeg_dec_ring_test_ib,
+	.insert_nop = jpeg_v2_0_dec_ring_nop,
+	.insert_start = jpeg_v2_0_dec_ring_insert_start,
+	.insert_end = jpeg_v2_0_dec_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_jpeg_ring_begin_use,
+	.end_use = amdgpu_jpeg_ring_end_use,
+	.emit_wreg = jpeg_v2_0_dec_ring_emit_wreg,
+	.emit_reg_wait = jpeg_v2_0_dec_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = jpeg_v3_0_ring_reset,
+};
+
+static void jpeg_v3_0_set_dec_ring_funcs(struct amdgpu_device *adev)
+{
+	adev->jpeg.inst->ring_dec->funcs = &jpeg_v3_0_dec_ring_vm_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs jpeg_v3_0_irq_funcs = {
+	.set = jpeg_v3_0_set_interrupt_state,
+	.process = jpeg_v3_0_process_interrupt,
+};
+
+static void jpeg_v3_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->jpeg.inst->irq.num_types = 1;
+	adev->jpeg.inst->irq.funcs = &jpeg_v3_0_irq_funcs;
+}
+
+const struct amdgpu_ip_block_version jpeg_v3_0_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_JPEG,
+	.major = 3,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &jpeg_v3_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.h b/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.h
new file mode 100644
index 000000000000..ce775a0c742f
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __JPEG_V3_0_H__
+#define __JPEG_V3_0_H__
+
+extern const struct amdgpu_ip_block_version jpeg_v3_0_ip_block;
+
+#endif /* __JPEG_V3_0_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c
new file mode 100644
index 000000000000..33db2c1ae6cc
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c
@@ -0,0 +1,878 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "amdgpu_jpeg.h"
+#include "amdgpu_pm.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "jpeg_v2_0.h"
+#include "jpeg_v4_0.h"
+#include "mmsch_v4_0.h"
+
+#include "vcn/vcn_4_0_0_offset.h"
+#include "vcn/vcn_4_0_0_sh_mask.h"
+#include "ivsrcid/vcn/irqsrcs_vcn_4_0.h"
+
+#define regUVD_JPEG_PITCH_INTERNAL_OFFSET                  0x401f
+
+static const struct amdgpu_hwip_reg_entry jpeg_reg_list_4_0[] = {
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_INT_STAT),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_CNTL),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_SIZE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_ADDR_MODE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_GFX10_ADDR_CONFIG),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_Y_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_UV_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_PITCH),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_UV_PITCH),
+};
+
+static int jpeg_v4_0_start_sriov(struct amdgpu_device *adev);
+static void jpeg_v4_0_set_dec_ring_funcs(struct amdgpu_device *adev);
+static void jpeg_v4_0_set_irq_funcs(struct amdgpu_device *adev);
+static int jpeg_v4_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
+				enum amd_powergating_state state);
+static void jpeg_v4_0_set_ras_funcs(struct amdgpu_device *adev);
+static void jpeg_v4_0_dec_ring_set_wptr(struct amdgpu_ring *ring);
+
+/**
+ * jpeg_v4_0_early_init - set function pointers
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Set ring and irq function pointers
+ */
+static int jpeg_v4_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+
+	adev->jpeg.num_jpeg_inst = 1;
+	adev->jpeg.num_jpeg_rings = 1;
+
+	jpeg_v4_0_set_dec_ring_funcs(adev);
+	jpeg_v4_0_set_irq_funcs(adev);
+	jpeg_v4_0_set_ras_funcs(adev);
+
+	return 0;
+}
+
+/**
+ * jpeg_v4_0_sw_init - sw init for JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Load firmware and sw initialization
+ */
+static int jpeg_v4_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	int r;
+
+	/* JPEG TRAP */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+		VCN_4_0__SRCID__JPEG_DECODE, &adev->jpeg.inst->irq);
+	if (r)
+		return r;
+
+	/* JPEG DJPEG POISON EVENT */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+			VCN_4_0__SRCID_DJPEG0_POISON, &adev->jpeg.inst->ras_poison_irq);
+	if (r)
+		return r;
+
+	/* JPEG EJPEG POISON EVENT */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+			VCN_4_0__SRCID_EJPEG0_POISON, &adev->jpeg.inst->ras_poison_irq);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_sw_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_resume(adev);
+	if (r)
+		return r;
+
+	ring = adev->jpeg.inst->ring_dec;
+	ring->use_doorbell = true;
+	ring->doorbell_index = amdgpu_sriov_vf(adev) ? (((adev->doorbell_index.vcn.vcn_ring0_1) << 1) + 4) : ((adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 1);
+	ring->vm_hub = AMDGPU_MMHUB0(0);
+
+	sprintf(ring->name, "jpeg_dec");
+	r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst->irq, 0,
+			     AMDGPU_RING_PRIO_DEFAULT, NULL);
+	if (r)
+		return r;
+
+	adev->jpeg.internal.jpeg_pitch[0] = regUVD_JPEG_PITCH_INTERNAL_OFFSET;
+	adev->jpeg.inst->external.jpeg_pitch[0] = SOC15_REG_OFFSET(JPEG, 0, regUVD_JPEG_PITCH);
+
+	r = amdgpu_jpeg_ras_sw_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_reg_dump_init(adev, jpeg_reg_list_4_0, ARRAY_SIZE(jpeg_reg_list_4_0));
+	if (r)
+		return r;
+
+	adev->jpeg.supported_reset =
+		amdgpu_get_soft_full_reset_mask(adev->jpeg.inst[0].ring_dec);
+	if (!amdgpu_sriov_vf(adev))
+		adev->jpeg.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+	r = amdgpu_jpeg_sysfs_reset_mask_init(adev);
+
+	return r;
+}
+
+/**
+ * jpeg_v4_0_sw_fini - sw fini for JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * JPEG suspend and free up sw allocation
+ */
+static int jpeg_v4_0_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = amdgpu_jpeg_suspend(adev);
+	if (r)
+		return r;
+
+	amdgpu_jpeg_sysfs_reset_mask_fini(adev);
+	r = amdgpu_jpeg_sw_fini(adev);
+
+	return r;
+}
+
+/**
+ * jpeg_v4_0_hw_init - start and test JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ */
+static int jpeg_v4_0_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring = adev->jpeg.inst->ring_dec;
+	int r;
+
+	if (amdgpu_sriov_vf(adev)) {
+		r = jpeg_v4_0_start_sriov(adev);
+		if (r)
+			return r;
+		ring->wptr = 0;
+		ring->wptr_old = 0;
+		jpeg_v4_0_dec_ring_set_wptr(ring);
+		ring->sched.ready = true;
+	} else {
+		adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
+						(adev->doorbell_index.vcn.vcn_ring0_1 << 1), 0);
+
+		WREG32_SOC15(VCN, 0, regVCN_JPEG_DB_CTRL,
+			ring->doorbell_index << VCN_JPEG_DB_CTRL__OFFSET__SHIFT |
+			VCN_JPEG_DB_CTRL__EN_MASK);
+
+		r = amdgpu_ring_test_helper(ring);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+/**
+ * jpeg_v4_0_hw_fini - stop the hardware block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Stop the JPEG block, mark ring as not ready any more
+ */
+static int jpeg_v4_0_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	cancel_delayed_work_sync(&adev->jpeg.idle_work);
+	if (!amdgpu_sriov_vf(adev)) {
+		if (adev->jpeg.cur_state != AMD_PG_STATE_GATE &&
+			RREG32_SOC15(JPEG, 0, regUVD_JRBC_STATUS))
+			jpeg_v4_0_set_powergating_state(ip_block, AMD_PG_STATE_GATE);
+	}
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__JPEG))
+		amdgpu_irq_put(adev, &adev->jpeg.inst->ras_poison_irq, 0);
+
+	return 0;
+}
+
+/**
+ * jpeg_v4_0_suspend - suspend JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * HW fini and suspend JPEG block
+ */
+static int jpeg_v4_0_suspend(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	r = jpeg_v4_0_hw_fini(ip_block);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_suspend(ip_block->adev);
+
+	return r;
+}
+
+/**
+ * jpeg_v4_0_resume - resume JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Resume firmware and hw init JPEG block
+ */
+static int jpeg_v4_0_resume(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	r = amdgpu_jpeg_resume(ip_block->adev);
+	if (r)
+		return r;
+
+	r = jpeg_v4_0_hw_init(ip_block);
+
+	return r;
+}
+
+static void jpeg_v4_0_disable_clock_gating(struct amdgpu_device *adev)
+{
+	uint32_t data = 0;
+
+	data = RREG32_SOC15(JPEG, 0, regJPEG_CGC_CTRL);
+	if (adev->cg_flags & AMD_CG_SUPPORT_JPEG_MGCG) {
+		data |= 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+		data &= (~JPEG_CGC_CTRL__JPEG_DEC_MODE_MASK);
+	} else {
+		data &= ~JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	}
+
+	data |= 1 << JPEG_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << JPEG_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(JPEG, 0, regJPEG_CGC_CTRL, data);
+
+	data = RREG32_SOC15(JPEG, 0, regJPEG_CGC_GATE);
+	data &= ~(JPEG_CGC_GATE__JPEG_DEC_MASK
+		| JPEG_CGC_GATE__JPEG2_DEC_MASK
+		| JPEG_CGC_GATE__JMCIF_MASK
+		| JPEG_CGC_GATE__JRBBM_MASK);
+	WREG32_SOC15(JPEG, 0, regJPEG_CGC_GATE, data);
+}
+
+static void jpeg_v4_0_enable_clock_gating(struct amdgpu_device *adev)
+{
+	uint32_t data = 0;
+
+	data = RREG32_SOC15(JPEG, 0, regJPEG_CGC_CTRL);
+	if (adev->cg_flags & AMD_CG_SUPPORT_JPEG_MGCG) {
+		data |= 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+		data |= JPEG_CGC_CTRL__JPEG_DEC_MODE_MASK;
+	} else {
+		data &= ~JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	}
+
+	data |= 1 << JPEG_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << JPEG_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(JPEG, 0, regJPEG_CGC_CTRL, data);
+
+	data = RREG32_SOC15(JPEG, 0, regJPEG_CGC_GATE);
+	data |= (JPEG_CGC_GATE__JPEG_DEC_MASK
+		|JPEG_CGC_GATE__JPEG2_DEC_MASK
+		|JPEG_CGC_GATE__JMCIF_MASK
+		|JPEG_CGC_GATE__JRBBM_MASK);
+	WREG32_SOC15(JPEG, 0, regJPEG_CGC_GATE, data);
+}
+
+static int jpeg_v4_0_disable_static_power_gating(struct amdgpu_device *adev)
+{
+	if (adev->pg_flags & AMD_PG_SUPPORT_JPEG) {
+		uint32_t data = 0;
+		int r = 0;
+
+		data = 1 << UVD_PGFSM_CONFIG__UVDJ_PWR_CONFIG__SHIFT;
+		WREG32(SOC15_REG_OFFSET(JPEG, 0, regUVD_PGFSM_CONFIG), data);
+
+		r = SOC15_WAIT_ON_RREG(JPEG, 0,
+			regUVD_PGFSM_STATUS, UVD_PGFSM_STATUS_UVDJ_PWR_ON,
+			UVD_PGFSM_STATUS__UVDJ_PWR_STATUS_MASK);
+
+		if (r) {
+			DRM_DEV_ERROR(adev->dev, "amdgpu: JPEG disable power gating failed\n");
+			return r;
+		}
+	}
+
+	/* disable anti hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regUVD_JPEG_POWER_STATUS), 0,
+		~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+
+	/* keep the JPEG in static PG mode */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regUVD_JPEG_POWER_STATUS), 0,
+		~UVD_JPEG_POWER_STATUS__JPEG_PG_MODE_MASK);
+
+	return 0;
+}
+
+static int jpeg_v4_0_enable_static_power_gating(struct amdgpu_device *adev)
+{
+	/* enable anti hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regUVD_JPEG_POWER_STATUS),
+		UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK,
+		~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_JPEG) {
+		uint32_t data = 0;
+		int r = 0;
+
+		data = 2 << UVD_PGFSM_CONFIG__UVDJ_PWR_CONFIG__SHIFT;
+		WREG32(SOC15_REG_OFFSET(JPEG, 0, regUVD_PGFSM_CONFIG), data);
+
+		r = SOC15_WAIT_ON_RREG(JPEG, 0, regUVD_PGFSM_STATUS,
+			(2 << UVD_PGFSM_STATUS__UVDJ_PWR_STATUS__SHIFT),
+			UVD_PGFSM_STATUS__UVDJ_PWR_STATUS_MASK);
+
+		if (r) {
+			DRM_DEV_ERROR(adev->dev, "amdgpu: JPEG enable power gating failed\n");
+			return r;
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * jpeg_v4_0_start - start JPEG block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the JPEG block
+ */
+static int jpeg_v4_0_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = adev->jpeg.inst->ring_dec;
+	int r;
+
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_jpeg(adev, true);
+
+	/* disable power gating */
+	r = jpeg_v4_0_disable_static_power_gating(adev);
+	if (r)
+		return r;
+
+	/* JPEG disable CGC */
+	jpeg_v4_0_disable_clock_gating(adev);
+
+	/* MJPEG global tiling registers */
+	WREG32_SOC15(JPEG, 0, regJPEG_DEC_GFX10_ADDR_CONFIG,
+		adev->gfx.config.gb_addr_config);
+
+
+	/* enable JMI channel */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regUVD_JMI_CNTL), 0,
+		~UVD_JMI_CNTL__SOFT_RESET_MASK);
+
+	/* enable System Interrupt for JRBC */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regJPEG_SYS_INT_EN),
+		JPEG_SYS_INT_EN__DJRBC_MASK,
+		~JPEG_SYS_INT_EN__DJRBC_MASK);
+
+	WREG32_SOC15(JPEG, 0, regUVD_LMI_JRBC_RB_VMID, 0);
+	WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_CNTL, (0x00000001L | 0x00000002L));
+	WREG32_SOC15(JPEG, 0, regUVD_LMI_JRBC_RB_64BIT_BAR_LOW,
+		lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15(JPEG, 0, regUVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
+		upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_RPTR, 0);
+	WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_WPTR, 0);
+	WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_CNTL, 0x00000002L);
+	WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_SIZE, ring->ring_size / 4);
+	ring->wptr = RREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_WPTR);
+
+	return 0;
+}
+
+static int jpeg_v4_0_start_sriov(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	uint64_t ctx_addr;
+	uint32_t param, resp, expected;
+	uint32_t tmp, timeout;
+
+	struct amdgpu_mm_table *table = &adev->virt.mm_table;
+	uint32_t *table_loc;
+	uint32_t table_size;
+	uint32_t size, size_dw;
+	uint32_t init_status;
+
+	struct mmsch_v4_0_cmd_direct_write
+		direct_wt = { {0} };
+	struct mmsch_v4_0_cmd_end end = { {0} };
+	struct mmsch_v4_0_init_header header;
+
+	direct_wt.cmd_header.command_type =
+		MMSCH_COMMAND__DIRECT_REG_WRITE;
+	end.cmd_header.command_type =
+		MMSCH_COMMAND__END;
+
+	size = sizeof(struct mmsch_v4_0_init_header);
+	table_loc = (uint32_t *)table->cpu_addr;
+	memcpy(&header, (void *)table_loc, size);
+
+	header.version = MMSCH_VERSION;
+	header.total_size = RREG32_SOC15(VCN, 0, regMMSCH_VF_CTX_SIZE);
+
+	header.jpegdec.init_status = 0;
+	header.jpegdec.table_offset = 0;
+	header.jpegdec.table_size = 0;
+
+	table_loc = (uint32_t *)table->cpu_addr;
+	table_loc += header.total_size;
+
+	table_size = 0;
+
+	ring = adev->jpeg.inst->ring_dec;
+
+	MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(JPEG, 0,
+		regUVD_LMI_JRBC_RB_64BIT_BAR_LOW),
+		lower_32_bits(ring->gpu_addr));
+	MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(JPEG, 0,
+		regUVD_LMI_JRBC_RB_64BIT_BAR_HIGH),
+		upper_32_bits(ring->gpu_addr));
+	MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(JPEG, 0,
+		regUVD_JRBC_RB_SIZE), ring->ring_size / 4);
+
+	/* add end packet */
+	MMSCH_V4_0_INSERT_END();
+
+	/* refine header */
+	header.jpegdec.init_status = 0;
+	header.jpegdec.table_offset = header.total_size;
+	header.jpegdec.table_size = table_size;
+	header.total_size += table_size;
+
+	/* Update init table header in memory */
+	size = sizeof(struct mmsch_v4_0_init_header);
+	table_loc = (uint32_t *)table->cpu_addr;
+	memcpy((void *)table_loc, &header, size);
+
+	/* Perform HDP flush before writing to MMSCH registers */
+	amdgpu_device_flush_hdp(adev, NULL);
+
+	/* message MMSCH (in VCN[0]) to initialize this client
+	 * 1, write to mmsch_vf_ctx_addr_lo/hi register with GPU mc addr
+	 * of memory descriptor location
+	 */
+	ctx_addr = table->gpu_addr;
+	WREG32_SOC15(VCN, 0, regMMSCH_VF_CTX_ADDR_LO, lower_32_bits(ctx_addr));
+	WREG32_SOC15(VCN, 0, regMMSCH_VF_CTX_ADDR_HI, upper_32_bits(ctx_addr));
+
+	/* 2, update vmid of descriptor */
+	tmp = RREG32_SOC15(VCN, 0, regMMSCH_VF_VMID);
+	tmp &= ~MMSCH_VF_VMID__VF_CTX_VMID_MASK;
+	/* use domain0 for MM scheduler */
+	tmp |= (0 << MMSCH_VF_VMID__VF_CTX_VMID__SHIFT);
+	WREG32_SOC15(VCN, 0, regMMSCH_VF_VMID, tmp);
+
+	/* 3, notify mmsch about the size of this descriptor */
+	size = header.total_size;
+	WREG32_SOC15(VCN, 0, regMMSCH_VF_CTX_SIZE, size);
+
+	/* 4, set resp to zero */
+	WREG32_SOC15(VCN, 0, regMMSCH_VF_MAILBOX_RESP, 0);
+
+	/* 5, kick off the initialization and wait until
+	 * MMSCH_VF_MAILBOX_RESP becomes non-zero
+	 */
+	param = 0x00000001;
+	WREG32_SOC15(VCN, 0, regMMSCH_VF_MAILBOX_HOST, param);
+	tmp = 0;
+	timeout = 1000;
+	resp = 0;
+	expected = MMSCH_VF_MAILBOX_RESP__OK;
+	init_status = ((struct mmsch_v4_0_init_header *)(table_loc))->jpegdec.init_status;
+	while (resp != expected) {
+		resp = RREG32_SOC15(VCN, 0, regMMSCH_VF_MAILBOX_RESP);
+
+		if (resp != 0)
+			break;
+		udelay(10);
+		tmp = tmp + 10;
+		if (tmp >= timeout) {
+			DRM_ERROR("failed to init MMSCH. TIME-OUT after %d usec"\
+				" waiting for regMMSCH_VF_MAILBOX_RESP "\
+				"(expected=0x%08x, readback=0x%08x)\n",
+				tmp, expected, resp);
+			return -EBUSY;
+		}
+	}
+	if (resp != expected && resp != MMSCH_VF_MAILBOX_RESP__INCOMPLETE
+			&& init_status != MMSCH_VF_ENGINE_STATUS__PASS) {
+		DRM_ERROR("MMSCH init status is incorrect! readback=0x%08x, header init status for jpeg: %x\n", resp, init_status);
+		return -EINVAL;
+	}
+
+	return 0;
+
+}
+
+/**
+ * jpeg_v4_0_stop - stop JPEG block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * stop the JPEG block
+ */
+static int jpeg_v4_0_stop(struct amdgpu_device *adev)
+{
+	int r;
+
+	/* reset JMI */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regUVD_JMI_CNTL),
+		UVD_JMI_CNTL__SOFT_RESET_MASK,
+		~UVD_JMI_CNTL__SOFT_RESET_MASK);
+
+	jpeg_v4_0_enable_clock_gating(adev);
+
+	/* enable power gating */
+	r = jpeg_v4_0_enable_static_power_gating(adev);
+	if (r)
+		return r;
+
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_jpeg(adev, false);
+
+	return 0;
+}
+
+/**
+ * jpeg_v4_0_dec_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t jpeg_v4_0_dec_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_RPTR);
+}
+
+/**
+ * jpeg_v4_0_dec_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t jpeg_v4_0_dec_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell)
+		return *ring->wptr_cpu_addr;
+	else
+		return RREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_WPTR);
+}
+
+/**
+ * jpeg_v4_0_dec_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void jpeg_v4_0_dec_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
+		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+	} else {
+		WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_WPTR, lower_32_bits(ring->wptr));
+	}
+}
+
+static bool jpeg_v4_0_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int ret = 1;
+
+	ret &= (((RREG32_SOC15(JPEG, 0, regUVD_JRBC_STATUS) &
+		UVD_JRBC_STATUS__RB_JOB_DONE_MASK) ==
+		UVD_JRBC_STATUS__RB_JOB_DONE_MASK));
+
+	return ret;
+}
+
+static int jpeg_v4_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	return SOC15_WAIT_ON_RREG(JPEG, 0, regUVD_JRBC_STATUS,
+		UVD_JRBC_STATUS__RB_JOB_DONE_MASK,
+		UVD_JRBC_STATUS__RB_JOB_DONE_MASK);
+}
+
+static int jpeg_v4_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = state == AMD_CG_STATE_GATE;
+
+	if (enable) {
+		if (!jpeg_v4_0_is_idle(ip_block))
+			return -EBUSY;
+		jpeg_v4_0_enable_clock_gating(adev);
+	} else {
+		jpeg_v4_0_disable_clock_gating(adev);
+	}
+
+	return 0;
+}
+
+static int jpeg_v4_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int ret;
+
+	if (amdgpu_sriov_vf(adev)) {
+		adev->jpeg.cur_state = AMD_PG_STATE_UNGATE;
+		return 0;
+	}
+
+	if (state == adev->jpeg.cur_state)
+		return 0;
+
+	if (state == AMD_PG_STATE_GATE)
+		ret = jpeg_v4_0_stop(adev);
+	else
+		ret = jpeg_v4_0_start(adev);
+
+	if (!ret)
+		adev->jpeg.cur_state = state;
+
+	return ret;
+}
+
+static int jpeg_v4_0_set_ras_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned int type,
+					enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+static int jpeg_v4_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	DRM_DEBUG("IH: JPEG TRAP\n");
+
+	switch (entry->src_id) {
+	case VCN_4_0__SRCID__JPEG_DECODE:
+		amdgpu_fence_process(adev->jpeg.inst->ring_dec);
+		break;
+	default:
+		DRM_DEV_ERROR(adev->dev, "Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+static int jpeg_v4_0_ring_reset(struct amdgpu_ring *ring,
+				unsigned int vmid,
+				struct amdgpu_fence *timedout_fence)
+{
+	int r;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+	r = jpeg_v4_0_stop(ring->adev);
+	if (r)
+		return r;
+	r = jpeg_v4_0_start(ring->adev);
+	if (r)
+		return r;
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static const struct amd_ip_funcs jpeg_v4_0_ip_funcs = {
+	.name = "jpeg_v4_0",
+	.early_init = jpeg_v4_0_early_init,
+	.sw_init = jpeg_v4_0_sw_init,
+	.sw_fini = jpeg_v4_0_sw_fini,
+	.hw_init = jpeg_v4_0_hw_init,
+	.hw_fini = jpeg_v4_0_hw_fini,
+	.suspend = jpeg_v4_0_suspend,
+	.resume = jpeg_v4_0_resume,
+	.is_idle = jpeg_v4_0_is_idle,
+	.wait_for_idle = jpeg_v4_0_wait_for_idle,
+	.set_clockgating_state = jpeg_v4_0_set_clockgating_state,
+	.set_powergating_state = jpeg_v4_0_set_powergating_state,
+	.dump_ip_state = amdgpu_jpeg_dump_ip_state,
+	.print_ip_state = amdgpu_jpeg_print_ip_state,
+};
+
+static const struct amdgpu_ring_funcs jpeg_v4_0_dec_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_JPEG,
+	.align_mask = 0xf,
+	.get_rptr = jpeg_v4_0_dec_ring_get_rptr,
+	.get_wptr = jpeg_v4_0_dec_ring_get_wptr,
+	.set_wptr = jpeg_v4_0_dec_ring_set_wptr,
+	.parse_cs = amdgpu_jpeg_dec_parse_cs,
+	.emit_frame_size =
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
+		8 + /* jpeg_v4_0_dec_ring_emit_vm_flush */
+		18 + 18 + /* jpeg_v4_0_dec_ring_emit_fence x2 vm fence */
+		8 + 16,
+	.emit_ib_size = 22, /* jpeg_v4_0_dec_ring_emit_ib */
+	.emit_ib = jpeg_v2_0_dec_ring_emit_ib,
+	.emit_fence = jpeg_v2_0_dec_ring_emit_fence,
+	.emit_vm_flush = jpeg_v2_0_dec_ring_emit_vm_flush,
+	.test_ring = amdgpu_jpeg_dec_ring_test_ring,
+	.test_ib = amdgpu_jpeg_dec_ring_test_ib,
+	.insert_nop = jpeg_v2_0_dec_ring_nop,
+	.insert_start = jpeg_v2_0_dec_ring_insert_start,
+	.insert_end = jpeg_v2_0_dec_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_jpeg_ring_begin_use,
+	.end_use = amdgpu_jpeg_ring_end_use,
+	.emit_wreg = jpeg_v2_0_dec_ring_emit_wreg,
+	.emit_reg_wait = jpeg_v2_0_dec_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = jpeg_v4_0_ring_reset,
+};
+
+static void jpeg_v4_0_set_dec_ring_funcs(struct amdgpu_device *adev)
+{
+	adev->jpeg.inst->ring_dec->funcs = &jpeg_v4_0_dec_ring_vm_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs jpeg_v4_0_irq_funcs = {
+	.process = jpeg_v4_0_process_interrupt,
+};
+
+static const struct amdgpu_irq_src_funcs jpeg_v4_0_ras_irq_funcs = {
+	.set = jpeg_v4_0_set_ras_interrupt_state,
+	.process = amdgpu_jpeg_process_poison_irq,
+};
+
+static void jpeg_v4_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->jpeg.inst->irq.num_types = 1;
+	adev->jpeg.inst->irq.funcs = &jpeg_v4_0_irq_funcs;
+
+	adev->jpeg.inst->ras_poison_irq.num_types = 1;
+	adev->jpeg.inst->ras_poison_irq.funcs = &jpeg_v4_0_ras_irq_funcs;
+}
+
+const struct amdgpu_ip_block_version jpeg_v4_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_JPEG,
+	.major = 4,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &jpeg_v4_0_ip_funcs,
+};
+
+static uint32_t jpeg_v4_0_query_poison_by_instance(struct amdgpu_device *adev,
+		uint32_t instance, uint32_t sub_block)
+{
+	uint32_t poison_stat = 0, reg_value = 0;
+
+	switch (sub_block) {
+	case AMDGPU_JPEG_V4_0_JPEG0:
+		reg_value = RREG32_SOC15(JPEG, instance, regUVD_RAS_JPEG0_STATUS);
+		poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_JPEG0_STATUS, POISONED_PF);
+		break;
+	case AMDGPU_JPEG_V4_0_JPEG1:
+		reg_value = RREG32_SOC15(JPEG, instance, regUVD_RAS_JPEG1_STATUS);
+		poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_JPEG1_STATUS, POISONED_PF);
+		break;
+	default:
+		break;
+	}
+
+	if (poison_stat)
+		dev_info(adev->dev, "Poison detected in JPEG%d sub_block%d\n",
+			instance, sub_block);
+
+	return poison_stat;
+}
+
+static bool jpeg_v4_0_query_ras_poison_status(struct amdgpu_device *adev)
+{
+	uint32_t inst = 0, sub = 0, poison_stat = 0;
+
+	for (inst = 0; inst < adev->jpeg.num_jpeg_inst; inst++)
+		for (sub = 0; sub < AMDGPU_JPEG_V4_0_MAX_SUB_BLOCK; sub++)
+			poison_stat +=
+				jpeg_v4_0_query_poison_by_instance(adev, inst, sub);
+
+	return !!poison_stat;
+}
+
+const struct amdgpu_ras_block_hw_ops jpeg_v4_0_ras_hw_ops = {
+	.query_poison_status = jpeg_v4_0_query_ras_poison_status,
+};
+
+static struct amdgpu_jpeg_ras jpeg_v4_0_ras = {
+	.ras_block = {
+		.hw_ops = &jpeg_v4_0_ras_hw_ops,
+		.ras_late_init = amdgpu_jpeg_ras_late_init,
+	},
+};
+
+static void jpeg_v4_0_set_ras_funcs(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, JPEG_HWIP, 0)) {
+	case IP_VERSION(4, 0, 0):
+		adev->jpeg.ras = &jpeg_v4_0_ras;
+		break;
+	default:
+		break;
+	}
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.h b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.h
new file mode 100644
index 000000000000..47638fd4d4e2
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.h
@@ -0,0 +1,35 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __JPEG_V4_0_H__
+#define __JPEG_V4_0_H__
+
+enum amdgpu_jpeg_v4_0_sub_block {
+	AMDGPU_JPEG_V4_0_JPEG0 = 0,
+	AMDGPU_JPEG_V4_0_JPEG1,
+
+	AMDGPU_JPEG_V4_0_MAX_SUB_BLOCK,
+};
+
+extern const struct amdgpu_ip_block_version jpeg_v4_0_ip_block;
+#endif /* __JPEG_V4_0_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c
new file mode 100644
index 000000000000..aae7328973d1
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c
@@ -0,0 +1,1486 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "amdgpu_jpeg.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "jpeg_v2_0.h"
+#include "jpeg_v4_0_3.h"
+#include "mmsch_v4_0_3.h"
+
+#include "vcn/vcn_4_0_3_offset.h"
+#include "vcn/vcn_4_0_3_sh_mask.h"
+#include "ivsrcid/vcn/irqsrcs_vcn_4_0.h"
+
+#define NORMALIZE_JPEG_REG_OFFSET(offset) \
+		(offset & 0x1FFFF)
+
+enum jpeg_engin_status {
+	UVD_PGFSM_STATUS__UVDJ_PWR_ON  = 0,
+	UVD_PGFSM_STATUS__UVDJ_PWR_OFF = 2,
+};
+
+static void jpeg_v4_0_3_set_dec_ring_funcs(struct amdgpu_device *adev);
+static void jpeg_v4_0_3_set_irq_funcs(struct amdgpu_device *adev);
+static int jpeg_v4_0_3_set_powergating_state(struct amdgpu_ip_block *ip_block,
+				enum amd_powergating_state state);
+static void jpeg_v4_0_3_set_ras_funcs(struct amdgpu_device *adev);
+static void jpeg_v4_0_3_dec_ring_set_wptr(struct amdgpu_ring *ring);
+
+static int amdgpu_ih_srcid_jpeg[] = {
+	VCN_4_0__SRCID__JPEG_DECODE,
+	VCN_4_0__SRCID__JPEG1_DECODE,
+	VCN_4_0__SRCID__JPEG2_DECODE,
+	VCN_4_0__SRCID__JPEG3_DECODE,
+	VCN_4_0__SRCID__JPEG4_DECODE,
+	VCN_4_0__SRCID__JPEG5_DECODE,
+	VCN_4_0__SRCID__JPEG6_DECODE,
+	VCN_4_0__SRCID__JPEG7_DECODE
+};
+
+static const struct amdgpu_hwip_reg_entry jpeg_reg_list_4_0_3[] = {
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_INT_STAT),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_SYS_INT_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC0_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC0_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC0_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_ADDR_MODE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_GFX10_ADDR_CONFIG),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_Y_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_UV_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_PITCH),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_UV_PITCH),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC1_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC1_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC1_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC2_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC2_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC2_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC3_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC3_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC3_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC4_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC4_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC4_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC5_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC5_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC5_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC6_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC6_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC6_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC7_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC7_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC7_UVD_JRBC_STATUS),
+};
+
+static inline bool jpeg_v4_0_3_normalizn_reqd(struct amdgpu_device *adev)
+{
+	return (adev->jpeg.caps & AMDGPU_JPEG_CAPS(RRMT_ENABLED)) == 0;
+}
+
+static inline int jpeg_v4_0_3_core_reg_offset(u32 pipe)
+{
+	if (pipe)
+		return ((0x40 * pipe) - 0xc80);
+	else
+		return 0;
+}
+
+/**
+ * jpeg_v4_0_3_early_init - set function pointers
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Set ring and irq function pointers
+ */
+static int jpeg_v4_0_3_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	adev->jpeg.num_jpeg_rings = AMDGPU_MAX_JPEG_RINGS_4_0_3;
+
+	jpeg_v4_0_3_set_dec_ring_funcs(adev);
+	jpeg_v4_0_3_set_irq_funcs(adev);
+	jpeg_v4_0_3_set_ras_funcs(adev);
+
+	return 0;
+}
+
+/**
+ * jpeg_v4_0_3_sw_init - sw init for JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Load firmware and sw initialization
+ */
+static int jpeg_v4_0_3_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	int i, j, r, jpeg_inst;
+
+	for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
+		/* JPEG TRAP */
+		r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+				amdgpu_ih_srcid_jpeg[j], &adev->jpeg.inst->irq);
+		if (r)
+			return r;
+	}
+
+	/* JPEG DJPEG POISON EVENT */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+			VCN_4_0__SRCID_DJPEG0_POISON, &adev->jpeg.inst->ras_poison_irq);
+	if (r)
+		return r;
+
+	/* JPEG EJPEG POISON EVENT */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+			VCN_4_0__SRCID_EJPEG0_POISON, &adev->jpeg.inst->ras_poison_irq);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_sw_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_resume(adev);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		jpeg_inst = GET_INST(JPEG, i);
+
+		for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
+			ring = &adev->jpeg.inst[i].ring_dec[j];
+			ring->use_doorbell = true;
+			ring->vm_hub = AMDGPU_MMHUB0(adev->jpeg.inst[i].aid_id);
+			if (!amdgpu_sriov_vf(adev)) {
+				ring->doorbell_index =
+					(adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
+					1 + j + 9 * jpeg_inst;
+			} else {
+				if (j < 4)
+					ring->doorbell_index =
+						(adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
+						4 + j + 32 * jpeg_inst;
+				else
+					ring->doorbell_index =
+						(adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
+						8 + j + 32 * jpeg_inst;
+			}
+			sprintf(ring->name, "jpeg_dec_%d.%d", adev->jpeg.inst[i].aid_id, j);
+			r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst->irq, 0,
+						AMDGPU_RING_PRIO_DEFAULT, NULL);
+			if (r)
+				return r;
+
+			adev->jpeg.internal.jpeg_pitch[j] =
+				regUVD_JRBC0_UVD_JRBC_SCRATCH0_INTERNAL_OFFSET;
+			adev->jpeg.inst[i].external.jpeg_pitch[j] =
+				SOC15_REG_OFFSET1(JPEG, jpeg_inst, regUVD_JRBC0_UVD_JRBC_SCRATCH0,
+						  jpeg_v4_0_3_core_reg_offset(j));
+		}
+	}
+
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__JPEG)) {
+		r = amdgpu_jpeg_ras_sw_init(adev);
+		if (r) {
+			dev_err(adev->dev, "Failed to initialize jpeg ras block!\n");
+			return r;
+		}
+	}
+
+	r = amdgpu_jpeg_reg_dump_init(adev, jpeg_reg_list_4_0_3, ARRAY_SIZE(jpeg_reg_list_4_0_3));
+	if (r)
+		return r;
+
+	adev->jpeg.supported_reset =
+		amdgpu_get_soft_full_reset_mask(adev->jpeg.inst[0].ring_dec);
+	if (!amdgpu_sriov_vf(adev))
+		adev->jpeg.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+	r = amdgpu_jpeg_sysfs_reset_mask_init(adev);
+
+	return 0;
+}
+
+/**
+ * jpeg_v4_0_3_sw_fini - sw fini for JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * JPEG suspend and free up sw allocation
+ */
+static int jpeg_v4_0_3_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = amdgpu_jpeg_suspend(adev);
+	if (r)
+		return r;
+
+	amdgpu_jpeg_sysfs_reset_mask_fini(adev);
+
+	r = amdgpu_jpeg_sw_fini(adev);
+
+	return r;
+}
+
+static int jpeg_v4_0_3_start_sriov(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	uint64_t ctx_addr;
+	uint32_t param, resp, expected;
+	uint32_t tmp, timeout;
+
+	struct amdgpu_mm_table *table = &adev->virt.mm_table;
+	uint32_t *table_loc;
+	uint32_t table_size;
+	uint32_t size, size_dw, item_offset;
+	uint32_t init_status;
+	int i, j, jpeg_inst;
+
+	struct mmsch_v4_0_cmd_direct_write
+		direct_wt = { {0} };
+	struct mmsch_v4_0_cmd_end end = { {0} };
+	struct mmsch_v4_0_3_init_header header;
+
+	direct_wt.cmd_header.command_type =
+		MMSCH_COMMAND__DIRECT_REG_WRITE;
+	end.cmd_header.command_type =
+		MMSCH_COMMAND__END;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; i++) {
+		jpeg_inst = GET_INST(JPEG, i);
+
+		memset(&header, 0, sizeof(struct mmsch_v4_0_3_init_header));
+		header.version = MMSCH_VERSION;
+		header.total_size = sizeof(struct mmsch_v4_0_3_init_header) >> 2;
+
+		table_loc = (uint32_t *)table->cpu_addr;
+		table_loc += header.total_size;
+
+		item_offset = header.total_size;
+
+		for (j = 0; j < adev->jpeg.num_jpeg_rings; j++) {
+			ring = &adev->jpeg.inst[i].ring_dec[j];
+			table_size = 0;
+
+			tmp = SOC15_REG_OFFSET(JPEG, 0, regUVD_JMI0_UVD_LMI_JRBC_RB_64BIT_BAR_LOW);
+			MMSCH_V4_0_INSERT_DIRECT_WT(tmp, lower_32_bits(ring->gpu_addr));
+			tmp = SOC15_REG_OFFSET(JPEG, 0, regUVD_JMI0_UVD_LMI_JRBC_RB_64BIT_BAR_HIGH);
+			MMSCH_V4_0_INSERT_DIRECT_WT(tmp, upper_32_bits(ring->gpu_addr));
+			tmp = SOC15_REG_OFFSET(JPEG, 0, regUVD_JRBC0_UVD_JRBC_RB_SIZE);
+			MMSCH_V4_0_INSERT_DIRECT_WT(tmp, ring->ring_size / 4);
+
+			if (j <= 3) {
+				header.mjpegdec0[j].table_offset = item_offset;
+				header.mjpegdec0[j].init_status = 0;
+				header.mjpegdec0[j].table_size = table_size;
+			} else {
+				header.mjpegdec1[j - 4].table_offset = item_offset;
+				header.mjpegdec1[j - 4].init_status = 0;
+				header.mjpegdec1[j - 4].table_size = table_size;
+			}
+			header.total_size += table_size;
+			item_offset += table_size;
+		}
+
+		MMSCH_V4_0_INSERT_END();
+
+		/* send init table to MMSCH */
+		size = sizeof(struct mmsch_v4_0_3_init_header);
+		table_loc = (uint32_t *)table->cpu_addr;
+		memcpy((void *)table_loc, &header, size);
+
+		ctx_addr = table->gpu_addr;
+		WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_CTX_ADDR_LO, lower_32_bits(ctx_addr));
+		WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_CTX_ADDR_HI, upper_32_bits(ctx_addr));
+
+		tmp = RREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_VMID);
+		tmp &= ~MMSCH_VF_VMID__VF_CTX_VMID_MASK;
+		tmp |= (0 << MMSCH_VF_VMID__VF_CTX_VMID__SHIFT);
+		WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_VMID, tmp);
+
+		size = header.total_size;
+		WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_CTX_SIZE, size);
+
+		WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_MAILBOX_RESP, 0);
+
+		param = 0x00000001;
+		WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_MAILBOX_HOST, param);
+		tmp = 0;
+		timeout = 1000;
+		resp = 0;
+		expected = MMSCH_VF_MAILBOX_RESP__OK;
+		init_status =
+			((struct mmsch_v4_0_3_init_header *)(table_loc))->mjpegdec0[i].init_status;
+		while (resp != expected) {
+			resp = RREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_MAILBOX_RESP);
+
+			if (resp != 0)
+				break;
+			udelay(10);
+			tmp = tmp + 10;
+			if (tmp >= timeout) {
+				DRM_ERROR("failed to init MMSCH. TIME-OUT after %d usec"\
+					" waiting for regMMSCH_VF_MAILBOX_RESP "\
+					"(expected=0x%08x, readback=0x%08x)\n",
+					tmp, expected, resp);
+				return -EBUSY;
+			}
+		}
+		if (resp != expected && resp != MMSCH_VF_MAILBOX_RESP__INCOMPLETE &&
+				init_status != MMSCH_VF_ENGINE_STATUS__PASS)
+			DRM_ERROR("MMSCH init status is incorrect! readback=0x%08x, header init status for jpeg: %x\n",
+					resp, init_status);
+
+	}
+	return 0;
+}
+
+/**
+ * jpeg_v4_0_3_hw_init - start and test JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ */
+static int jpeg_v4_0_3_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	int i, j, r, jpeg_inst;
+
+	if (amdgpu_sriov_vf(adev)) {
+		r = jpeg_v4_0_3_start_sriov(adev);
+		if (r)
+			return r;
+
+		for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+			for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
+				ring = &adev->jpeg.inst[i].ring_dec[j];
+				ring->wptr = 0;
+				ring->wptr_old = 0;
+				jpeg_v4_0_3_dec_ring_set_wptr(ring);
+				ring->sched.ready = true;
+			}
+		}
+	} else {
+		/* This flag is not set for VF, assumed to be disabled always */
+		if (RREG32_SOC15(VCN, GET_INST(VCN, 0), regVCN_RRMT_CNTL) &
+		    0x100)
+			adev->jpeg.caps |= AMDGPU_JPEG_CAPS(RRMT_ENABLED);
+
+		for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+			jpeg_inst = GET_INST(JPEG, i);
+
+			ring = adev->jpeg.inst[i].ring_dec;
+
+			if (ring->use_doorbell)
+				adev->nbio.funcs->vcn_doorbell_range(
+					adev, ring->use_doorbell,
+					(adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
+						9 * jpeg_inst,
+					adev->jpeg.inst[i].aid_id);
+
+			for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
+				ring = &adev->jpeg.inst[i].ring_dec[j];
+				if (ring->use_doorbell)
+					WREG32_SOC15_OFFSET(
+						VCN, GET_INST(VCN, i),
+						regVCN_JPEG_DB_CTRL,
+						(ring->pipe ? (ring->pipe - 0x15) : 0),
+						ring->doorbell_index
+							<< VCN_JPEG_DB_CTRL__OFFSET__SHIFT |
+							VCN_JPEG_DB_CTRL__EN_MASK);
+				r = amdgpu_ring_test_helper(ring);
+				if (r)
+					return r;
+			}
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * jpeg_v4_0_3_hw_fini - stop the hardware block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Stop the JPEG block, mark ring as not ready any more
+ */
+static int jpeg_v4_0_3_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int ret = 0;
+
+	cancel_delayed_work_sync(&adev->jpeg.idle_work);
+
+	if (!amdgpu_sriov_vf(adev)) {
+		if (adev->jpeg.cur_state != AMD_PG_STATE_GATE)
+			ret = jpeg_v4_0_3_set_powergating_state(ip_block, AMD_PG_STATE_GATE);
+	}
+
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__JPEG) && !amdgpu_sriov_vf(adev))
+		amdgpu_irq_put(adev, &adev->jpeg.inst->ras_poison_irq, 0);
+
+	return ret;
+}
+
+/**
+ * jpeg_v4_0_3_suspend - suspend JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * HW fini and suspend JPEG block
+ */
+static int jpeg_v4_0_3_suspend(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	r = jpeg_v4_0_3_hw_fini(ip_block);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_suspend(ip_block->adev);
+
+	return r;
+}
+
+/**
+ * jpeg_v4_0_3_resume - resume JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Resume firmware and hw init JPEG block
+ */
+static int jpeg_v4_0_3_resume(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	r = amdgpu_jpeg_resume(ip_block->adev);
+	if (r)
+		return r;
+
+	r = jpeg_v4_0_3_hw_init(ip_block);
+
+	return r;
+}
+
+static void jpeg_v4_0_3_disable_clock_gating(struct amdgpu_device *adev, int inst_idx)
+{
+	int i, jpeg_inst;
+	uint32_t data;
+
+	jpeg_inst = GET_INST(JPEG, inst_idx);
+	data = RREG32_SOC15(JPEG, jpeg_inst, regJPEG_CGC_CTRL);
+	if (adev->cg_flags & AMD_CG_SUPPORT_JPEG_MGCG) {
+		data |= 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+		data &= (~(JPEG_CGC_CTRL__JPEG0_DEC_MODE_MASK << 1));
+	} else {
+		data &= ~JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	}
+
+	data |= 1 << JPEG_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << JPEG_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(JPEG, jpeg_inst, regJPEG_CGC_CTRL, data);
+
+	data = RREG32_SOC15(JPEG, jpeg_inst, regJPEG_CGC_GATE);
+	data &= ~(JPEG_CGC_GATE__JMCIF_MASK | JPEG_CGC_GATE__JRBBM_MASK);
+	for (i = 0; i < adev->jpeg.num_jpeg_rings; ++i)
+		data &= ~(JPEG_CGC_GATE__JPEG0_DEC_MASK << i);
+	WREG32_SOC15(JPEG, jpeg_inst, regJPEG_CGC_GATE, data);
+}
+
+static void jpeg_v4_0_3_enable_clock_gating(struct amdgpu_device *adev, int inst_idx)
+{
+	int i, jpeg_inst;
+	uint32_t data;
+
+	jpeg_inst = GET_INST(JPEG, inst_idx);
+	data = RREG32_SOC15(JPEG, jpeg_inst, regJPEG_CGC_CTRL);
+	if (adev->cg_flags & AMD_CG_SUPPORT_JPEG_MGCG) {
+		data |= 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+		data |= (JPEG_CGC_CTRL__JPEG0_DEC_MODE_MASK << 1);
+	} else {
+		data &= ~JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	}
+
+	data |= 1 << JPEG_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << JPEG_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(JPEG, jpeg_inst, regJPEG_CGC_CTRL, data);
+
+	data = RREG32_SOC15(JPEG, jpeg_inst, regJPEG_CGC_GATE);
+	data |= (JPEG_CGC_GATE__JMCIF_MASK | JPEG_CGC_GATE__JRBBM_MASK);
+	for (i = 0; i < adev->jpeg.num_jpeg_rings; ++i)
+		data |= (JPEG_CGC_GATE__JPEG0_DEC_MASK << i);
+	WREG32_SOC15(JPEG, jpeg_inst, regJPEG_CGC_GATE, data);
+}
+
+static void jpeg_v4_0_3_start_inst(struct amdgpu_device *adev, int inst)
+{
+	int jpeg_inst = GET_INST(JPEG, inst);
+
+	WREG32_SOC15(JPEG, jpeg_inst, regUVD_PGFSM_CONFIG,
+		     1 << UVD_PGFSM_CONFIG__UVDJ_PWR_CONFIG__SHIFT);
+	SOC15_WAIT_ON_RREG(JPEG, jpeg_inst, regUVD_PGFSM_STATUS,
+			   UVD_PGFSM_STATUS__UVDJ_PWR_ON <<
+			   UVD_PGFSM_STATUS__UVDJ_PWR_STATUS__SHIFT,
+			   UVD_PGFSM_STATUS__UVDJ_PWR_STATUS_MASK);
+
+	/* disable anti hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JPEG_POWER_STATUS),
+		 0, ~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+
+	/* JPEG disable CGC */
+	jpeg_v4_0_3_disable_clock_gating(adev, inst);
+
+	/* MJPEG global tiling registers */
+	WREG32_SOC15(JPEG, jpeg_inst, regJPEG_DEC_GFX8_ADDR_CONFIG,
+		     adev->gfx.config.gb_addr_config);
+	WREG32_SOC15(JPEG, jpeg_inst, regJPEG_DEC_GFX10_ADDR_CONFIG,
+		     adev->gfx.config.gb_addr_config);
+
+	/* enable JMI channel */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JMI_CNTL), 0,
+		 ~UVD_JMI_CNTL__SOFT_RESET_MASK);
+}
+
+static void jpeg_v4_0_3_start_jrbc(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int jpeg_inst = GET_INST(JPEG, ring->me);
+	int reg_offset = jpeg_v4_0_3_core_reg_offset(ring->pipe);
+
+	/* enable System Interrupt for JRBC */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regJPEG_SYS_INT_EN),
+		 JPEG_SYS_INT_EN__DJRBC0_MASK << ring->pipe,
+		 ~(JPEG_SYS_INT_EN__DJRBC0_MASK << ring->pipe));
+
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JMI0_UVD_LMI_JRBC_RB_VMID,
+			    reg_offset, 0);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JRBC0_UVD_JRBC_RB_CNTL,
+			    reg_offset,
+			    (0x00000001L | 0x00000002L));
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JMI0_UVD_LMI_JRBC_RB_64BIT_BAR_LOW,
+			    reg_offset, lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JMI0_UVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
+			    reg_offset, upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JRBC0_UVD_JRBC_RB_RPTR,
+			    reg_offset, 0);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JRBC0_UVD_JRBC_RB_WPTR,
+			    reg_offset, 0);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JRBC0_UVD_JRBC_RB_CNTL,
+			    reg_offset, 0x00000002L);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JRBC0_UVD_JRBC_RB_SIZE,
+			    reg_offset, ring->ring_size / 4);
+	ring->wptr = RREG32_SOC15_OFFSET(JPEG, jpeg_inst, regUVD_JRBC0_UVD_JRBC_RB_WPTR,
+					 reg_offset);
+}
+
+/**
+ * jpeg_v4_0_3_start - start JPEG block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the JPEG block
+ */
+static int jpeg_v4_0_3_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	int i, j;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		jpeg_v4_0_3_start_inst(adev, i);
+		for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
+			ring = &adev->jpeg.inst[i].ring_dec[j];
+			jpeg_v4_0_3_start_jrbc(ring);
+		}
+	}
+
+	return 0;
+}
+
+static void jpeg_v4_0_3_stop_inst(struct amdgpu_device *adev, int inst)
+{
+	int jpeg_inst = GET_INST(JPEG, inst);
+	/* reset JMI */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JMI_CNTL),
+		 UVD_JMI_CNTL__SOFT_RESET_MASK,
+		 ~UVD_JMI_CNTL__SOFT_RESET_MASK);
+
+	jpeg_v4_0_3_enable_clock_gating(adev, inst);
+
+	/* enable anti hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JPEG_POWER_STATUS),
+		 UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK,
+		 ~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+
+}
+
+/**
+ * jpeg_v4_0_3_stop - stop JPEG block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * stop the JPEG block
+ */
+static int jpeg_v4_0_3_stop(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i)
+		jpeg_v4_0_3_stop_inst(adev, i);
+
+	return 0;
+}
+
+/**
+ * jpeg_v4_0_3_dec_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t jpeg_v4_0_3_dec_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32_SOC15_OFFSET(JPEG, GET_INST(JPEG, ring->me), regUVD_JRBC0_UVD_JRBC_RB_RPTR,
+				   jpeg_v4_0_3_core_reg_offset(ring->pipe));
+}
+
+/**
+ * jpeg_v4_0_3_dec_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t jpeg_v4_0_3_dec_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell)
+		return adev->wb.wb[ring->wptr_offs];
+
+	return RREG32_SOC15_OFFSET(JPEG, GET_INST(JPEG, ring->me), regUVD_JRBC0_UVD_JRBC_RB_WPTR,
+				   jpeg_v4_0_3_core_reg_offset(ring->pipe));
+}
+
+void jpeg_v4_0_3_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	/* JPEG engine access for HDP flush doesn't work when RRMT is enabled.
+	 * This is a workaround to avoid any HDP flush through JPEG ring.
+	 */
+}
+
+/**
+ * jpeg_v4_0_3_dec_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void jpeg_v4_0_3_dec_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		adev->wb.wb[ring->wptr_offs] = lower_32_bits(ring->wptr);
+		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+	} else {
+		WREG32_SOC15_OFFSET(JPEG, GET_INST(JPEG, ring->me), regUVD_JRBC0_UVD_JRBC_RB_WPTR,
+				    jpeg_v4_0_3_core_reg_offset(ring->pipe),
+				    lower_32_bits(ring->wptr));
+	}
+}
+
+/**
+ * jpeg_v4_0_3_dec_ring_insert_start - insert a start command
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Write a start command to the ring.
+ */
+void jpeg_v4_0_3_dec_ring_insert_start(struct amdgpu_ring *ring)
+{
+	if (!amdgpu_sriov_vf(ring->adev)) {
+		amdgpu_ring_write(ring, PACKETJ(regUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET,
+			0, 0, PACKETJ_TYPE0));
+		amdgpu_ring_write(ring, 0x62a04); /* PCTL0_MMHUB_DEEPSLEEP_IB */
+
+		amdgpu_ring_write(ring,
+				  PACKETJ(JRBC_DEC_EXTERNAL_REG_WRITE_ADDR, 0,
+					  0, PACKETJ_TYPE0));
+		amdgpu_ring_write(ring, 0x80004000);
+	}
+}
+
+/**
+ * jpeg_v4_0_3_dec_ring_insert_end - insert a end command
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Write a end command to the ring.
+ */
+void jpeg_v4_0_3_dec_ring_insert_end(struct amdgpu_ring *ring)
+{
+	if (!amdgpu_sriov_vf(ring->adev)) {
+		amdgpu_ring_write(ring, PACKETJ(regUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET,
+			0, 0, PACKETJ_TYPE0));
+		amdgpu_ring_write(ring, 0x62a04);
+
+		amdgpu_ring_write(ring,
+				  PACKETJ(JRBC_DEC_EXTERNAL_REG_WRITE_ADDR, 0,
+					  0, PACKETJ_TYPE0));
+		amdgpu_ring_write(ring, 0x00004000);
+	}
+}
+
+/**
+ * jpeg_v4_0_3_dec_ring_emit_fence - emit an fence & trap command
+ *
+ * @ring: amdgpu_ring pointer
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
+ *
+ * Write a fence and a trap command to the ring.
+ */
+void jpeg_v4_0_3_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				unsigned int flags)
+{
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	amdgpu_ring_write(ring, PACKETJ(regUVD_JPEG_GPCOM_DATA0_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, seq);
+
+	amdgpu_ring_write(ring,	PACKETJ(regUVD_JPEG_GPCOM_DATA1_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, seq);
+
+	amdgpu_ring_write(ring,	PACKETJ(regUVD_LMI_JRBC_RB_MEM_WR_64BIT_BAR_LOW_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+
+	amdgpu_ring_write(ring,	PACKETJ(regUVD_LMI_JRBC_RB_MEM_WR_64BIT_BAR_HIGH_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+
+	amdgpu_ring_write(ring,	PACKETJ(regUVD_JPEG_GPCOM_CMD_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x8);
+
+	amdgpu_ring_write(ring,	PACKETJ(regUVD_JPEG_GPCOM_CMD_INTERNAL_OFFSET,
+		0, PACKETJ_CONDITION_CHECK0, PACKETJ_TYPE4));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring, PACKETJ(0, 0, 0, PACKETJ_TYPE6));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring, PACKETJ(0, 0, 0, PACKETJ_TYPE6));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring, PACKETJ(0, 0, 0, PACKETJ_TYPE7));
+	amdgpu_ring_write(ring, 0);
+}
+
+/**
+ * jpeg_v4_0_3_dec_ring_emit_ib - execute indirect buffer
+ *
+ * @ring: amdgpu_ring pointer
+ * @job: job to retrieve vmid from
+ * @ib: indirect buffer to execute
+ * @flags: unused
+ *
+ * Write ring commands to execute the indirect buffer.
+ */
+void jpeg_v4_0_3_dec_ring_emit_ib(struct amdgpu_ring *ring,
+				struct amdgpu_job *job,
+				struct amdgpu_ib *ib,
+				uint32_t flags)
+{
+	unsigned int vmid = AMDGPU_JOB_GET_VMID(job);
+
+	amdgpu_ring_write(ring, PACKETJ(regUVD_LMI_JRBC_IB_VMID_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+
+	if (ring->funcs->parse_cs)
+		amdgpu_ring_write(ring, 0);
+	else
+		amdgpu_ring_write(ring, (vmid | (vmid << 4) | (vmid << 8)));
+
+	amdgpu_ring_write(ring, PACKETJ(regUVD_LMI_JPEG_VMID_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, (vmid | (vmid << 4) | (vmid << 8)));
+
+	amdgpu_ring_write(ring,	PACKETJ(regUVD_LMI_JRBC_IB_64BIT_BAR_LOW_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
+
+	amdgpu_ring_write(ring,	PACKETJ(regUVD_LMI_JRBC_IB_64BIT_BAR_HIGH_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+
+	amdgpu_ring_write(ring,	PACKETJ(regUVD_JRBC_IB_SIZE_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, ib->length_dw);
+
+	amdgpu_ring_write(ring,	PACKETJ(regUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_LOW_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, lower_32_bits(ring->gpu_addr));
+
+	amdgpu_ring_write(ring,	PACKETJ(regUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_HIGH_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, upper_32_bits(ring->gpu_addr));
+
+	amdgpu_ring_write(ring,	PACKETJ(0, 0, PACKETJ_CONDITION_CHECK0, PACKETJ_TYPE2));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring,	PACKETJ(regUVD_JRBC_RB_COND_RD_TIMER_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x01400200);
+
+	amdgpu_ring_write(ring, PACKETJ(regUVD_JRBC_RB_REF_DATA_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x2);
+
+	amdgpu_ring_write(ring,	PACKETJ(regUVD_JRBC_STATUS_INTERNAL_OFFSET,
+		0, PACKETJ_CONDITION_CHECK3, PACKETJ_TYPE3));
+	amdgpu_ring_write(ring, 0x2);
+}
+
+void jpeg_v4_0_3_dec_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+				uint32_t val, uint32_t mask)
+{
+	uint32_t reg_offset;
+
+	/* Use normalized offsets if required */
+	if (jpeg_v4_0_3_normalizn_reqd(ring->adev))
+		reg = NORMALIZE_JPEG_REG_OFFSET(reg);
+
+	reg_offset = (reg << 2);
+
+	amdgpu_ring_write(ring, PACKETJ(regUVD_JRBC_RB_COND_RD_TIMER_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x01400200);
+
+	amdgpu_ring_write(ring,	PACKETJ(regUVD_JRBC_RB_REF_DATA_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, val);
+
+	amdgpu_ring_write(ring, PACKETJ(regUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	if (reg_offset >= 0x10000 && reg_offset <= 0x105ff) {
+		amdgpu_ring_write(ring, 0);
+		amdgpu_ring_write(ring,
+			PACKETJ((reg_offset >> 2), 0, 0, PACKETJ_TYPE3));
+	} else {
+		amdgpu_ring_write(ring, reg_offset);
+		amdgpu_ring_write(ring,	PACKETJ(JRBC_DEC_EXTERNAL_REG_WRITE_ADDR,
+			0, 0, PACKETJ_TYPE3));
+	}
+	amdgpu_ring_write(ring, mask);
+}
+
+void jpeg_v4_0_3_dec_ring_emit_vm_flush(struct amdgpu_ring *ring,
+				unsigned int vmid, uint64_t pd_addr)
+{
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
+	uint32_t data0, data1, mask;
+
+	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* wait for register write */
+	data0 = hub->ctx0_ptb_addr_lo32 + vmid * hub->ctx_addr_distance;
+	data1 = lower_32_bits(pd_addr);
+	mask = 0xffffffff;
+	jpeg_v4_0_3_dec_ring_emit_reg_wait(ring, data0, data1, mask);
+}
+
+void jpeg_v4_0_3_dec_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg, uint32_t val)
+{
+	uint32_t reg_offset;
+
+	/* Use normalized offsets if required */
+	if (jpeg_v4_0_3_normalizn_reqd(ring->adev))
+		reg = NORMALIZE_JPEG_REG_OFFSET(reg);
+
+	reg_offset = (reg << 2);
+
+	amdgpu_ring_write(ring,	PACKETJ(regUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET,
+		0, 0, PACKETJ_TYPE0));
+	if (reg_offset >= 0x10000 && reg_offset <= 0x105ff) {
+		amdgpu_ring_write(ring, 0);
+		amdgpu_ring_write(ring,
+			PACKETJ((reg_offset >> 2), 0, 0, PACKETJ_TYPE0));
+	} else {
+		amdgpu_ring_write(ring, reg_offset);
+		amdgpu_ring_write(ring,	PACKETJ(JRBC_DEC_EXTERNAL_REG_WRITE_ADDR,
+			0, 0, PACKETJ_TYPE0));
+	}
+	amdgpu_ring_write(ring, val);
+}
+
+void jpeg_v4_0_3_dec_ring_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	int i;
+
+	WARN_ON(ring->wptr % 2 || count % 2);
+
+	for (i = 0; i < count / 2; i++) {
+		amdgpu_ring_write(ring, PACKETJ(0, 0, 0, PACKETJ_TYPE6));
+		amdgpu_ring_write(ring, 0);
+	}
+}
+
+static bool jpeg_v4_0_3_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool ret = false;
+	int i, j;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
+			ret &= ((RREG32_SOC15_OFFSET(JPEG, GET_INST(JPEG, i),
+				regUVD_JRBC0_UVD_JRBC_STATUS, jpeg_v4_0_3_core_reg_offset(j)) &
+				UVD_JRBC0_UVD_JRBC_STATUS__RB_JOB_DONE_MASK) ==
+				UVD_JRBC0_UVD_JRBC_STATUS__RB_JOB_DONE_MASK);
+		}
+	}
+
+	return ret;
+}
+
+static int jpeg_v4_0_3_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int ret = 0;
+	int i, j;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
+			ret &= (SOC15_WAIT_ON_RREG_OFFSET(JPEG, GET_INST(JPEG, i),
+				regUVD_JRBC0_UVD_JRBC_STATUS, jpeg_v4_0_3_core_reg_offset(j),
+				UVD_JRBC0_UVD_JRBC_STATUS__RB_JOB_DONE_MASK,
+				UVD_JRBC0_UVD_JRBC_STATUS__RB_JOB_DONE_MASK));
+		}
+	}
+	return ret;
+}
+
+static int jpeg_v4_0_3_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = state == AMD_CG_STATE_GATE;
+	int i;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (enable) {
+			if (!jpeg_v4_0_3_is_idle(ip_block))
+				return -EBUSY;
+			jpeg_v4_0_3_enable_clock_gating(adev, i);
+		} else {
+			jpeg_v4_0_3_disable_clock_gating(adev, i);
+		}
+	}
+	return 0;
+}
+
+static int jpeg_v4_0_3_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int ret;
+
+	if (amdgpu_sriov_vf(adev)) {
+		adev->jpeg.cur_state = AMD_PG_STATE_UNGATE;
+		return 0;
+	}
+
+	if (state == adev->jpeg.cur_state)
+		return 0;
+
+	if (state == AMD_PG_STATE_GATE)
+		ret = jpeg_v4_0_3_stop(adev);
+	else
+		ret = jpeg_v4_0_3_start(adev);
+
+	if (!ret)
+		adev->jpeg.cur_state = state;
+
+	return ret;
+}
+
+static int jpeg_v4_0_3_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned int type,
+					enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+static int jpeg_v4_0_3_set_ras_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned int type,
+					enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+static int jpeg_v4_0_3_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	uint32_t i, inst;
+
+	i = node_id_to_phys_map[entry->node_id];
+	DRM_DEV_DEBUG(adev->dev, "IH: JPEG TRAP\n");
+
+	for (inst = 0; inst < adev->jpeg.num_jpeg_inst; ++inst)
+		if (adev->jpeg.inst[inst].aid_id == i)
+			break;
+
+	if (inst >= adev->jpeg.num_jpeg_inst) {
+		dev_WARN_ONCE(adev->dev, 1,
+			      "Interrupt received for unknown JPEG instance %d",
+			      entry->node_id);
+		return 0;
+	}
+
+	switch (entry->src_id) {
+	case VCN_4_0__SRCID__JPEG_DECODE:
+		amdgpu_fence_process(&adev->jpeg.inst[inst].ring_dec[0]);
+		break;
+	case VCN_4_0__SRCID__JPEG1_DECODE:
+		amdgpu_fence_process(&adev->jpeg.inst[inst].ring_dec[1]);
+		break;
+	case VCN_4_0__SRCID__JPEG2_DECODE:
+		amdgpu_fence_process(&adev->jpeg.inst[inst].ring_dec[2]);
+		break;
+	case VCN_4_0__SRCID__JPEG3_DECODE:
+		amdgpu_fence_process(&adev->jpeg.inst[inst].ring_dec[3]);
+		break;
+	case VCN_4_0__SRCID__JPEG4_DECODE:
+		amdgpu_fence_process(&adev->jpeg.inst[inst].ring_dec[4]);
+		break;
+	case VCN_4_0__SRCID__JPEG5_DECODE:
+		amdgpu_fence_process(&adev->jpeg.inst[inst].ring_dec[5]);
+		break;
+	case VCN_4_0__SRCID__JPEG6_DECODE:
+		amdgpu_fence_process(&adev->jpeg.inst[inst].ring_dec[6]);
+		break;
+	case VCN_4_0__SRCID__JPEG7_DECODE:
+		amdgpu_fence_process(&adev->jpeg.inst[inst].ring_dec[7]);
+		break;
+	default:
+		DRM_DEV_ERROR(adev->dev, "Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+static void jpeg_v4_0_3_core_stall_reset(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int jpeg_inst = GET_INST(JPEG, ring->me);
+	int reg_offset = jpeg_v4_0_3_core_reg_offset(ring->pipe);
+
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JMI0_UVD_JMI_CLIENT_STALL,
+			    reg_offset, 0x1F);
+	SOC15_WAIT_ON_RREG_OFFSET(JPEG, jpeg_inst,
+				  regUVD_JMI0_UVD_JMI_CLIENT_CLEAN_STATUS,
+				  reg_offset, 0x1F, 0x1F);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JMI0_JPEG_LMI_DROP,
+			    reg_offset, 0x1F);
+	WREG32_SOC15(JPEG, jpeg_inst, regJPEG_CORE_RST_CTRL, 1 << ring->pipe);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JMI0_UVD_JMI_CLIENT_STALL,
+			    reg_offset, 0x00);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JMI0_JPEG_LMI_DROP,
+			    reg_offset, 0x00);
+	WREG32_SOC15(JPEG, jpeg_inst, regJPEG_CORE_RST_CTRL, 0x00);
+}
+
+static int jpeg_v4_0_3_ring_reset(struct amdgpu_ring *ring,
+				  unsigned int vmid,
+				  struct amdgpu_fence *timedout_fence)
+{
+	if (amdgpu_sriov_vf(ring->adev))
+		return -EOPNOTSUPP;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+	jpeg_v4_0_3_core_stall_reset(ring);
+	jpeg_v4_0_3_start_jrbc(ring);
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static const struct amd_ip_funcs jpeg_v4_0_3_ip_funcs = {
+	.name = "jpeg_v4_0_3",
+	.early_init = jpeg_v4_0_3_early_init,
+	.sw_init = jpeg_v4_0_3_sw_init,
+	.sw_fini = jpeg_v4_0_3_sw_fini,
+	.hw_init = jpeg_v4_0_3_hw_init,
+	.hw_fini = jpeg_v4_0_3_hw_fini,
+	.suspend = jpeg_v4_0_3_suspend,
+	.resume = jpeg_v4_0_3_resume,
+	.is_idle = jpeg_v4_0_3_is_idle,
+	.wait_for_idle = jpeg_v4_0_3_wait_for_idle,
+	.set_clockgating_state = jpeg_v4_0_3_set_clockgating_state,
+	.set_powergating_state = jpeg_v4_0_3_set_powergating_state,
+	.dump_ip_state = amdgpu_jpeg_dump_ip_state,
+	.print_ip_state = amdgpu_jpeg_print_ip_state,
+};
+
+static const struct amdgpu_ring_funcs jpeg_v4_0_3_dec_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_JPEG,
+	.align_mask = 0xf,
+	.get_rptr = jpeg_v4_0_3_dec_ring_get_rptr,
+	.get_wptr = jpeg_v4_0_3_dec_ring_get_wptr,
+	.set_wptr = jpeg_v4_0_3_dec_ring_set_wptr,
+	.parse_cs = amdgpu_jpeg_dec_parse_cs,
+	.emit_frame_size =
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
+		8 + /* jpeg_v4_0_3_dec_ring_emit_vm_flush */
+		18 + 18 + /* jpeg_v4_0_3_dec_ring_emit_fence x2 vm fence */
+		8 + 16,
+	.emit_ib_size = 22, /* jpeg_v4_0_3_dec_ring_emit_ib */
+	.emit_ib = jpeg_v4_0_3_dec_ring_emit_ib,
+	.emit_fence = jpeg_v4_0_3_dec_ring_emit_fence,
+	.emit_vm_flush = jpeg_v4_0_3_dec_ring_emit_vm_flush,
+	.emit_hdp_flush = jpeg_v4_0_3_ring_emit_hdp_flush,
+	.test_ring = amdgpu_jpeg_dec_ring_test_ring,
+	.test_ib = amdgpu_jpeg_dec_ring_test_ib,
+	.insert_nop = jpeg_v4_0_3_dec_ring_nop,
+	.insert_start = jpeg_v4_0_3_dec_ring_insert_start,
+	.insert_end = jpeg_v4_0_3_dec_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_jpeg_ring_begin_use,
+	.end_use = amdgpu_jpeg_ring_end_use,
+	.emit_wreg = jpeg_v4_0_3_dec_ring_emit_wreg,
+	.emit_reg_wait = jpeg_v4_0_3_dec_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = jpeg_v4_0_3_ring_reset,
+};
+
+static void jpeg_v4_0_3_set_dec_ring_funcs(struct amdgpu_device *adev)
+{
+	int i, j, jpeg_inst;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
+			adev->jpeg.inst[i].ring_dec[j].funcs = &jpeg_v4_0_3_dec_ring_vm_funcs;
+			adev->jpeg.inst[i].ring_dec[j].me = i;
+			adev->jpeg.inst[i].ring_dec[j].pipe = j;
+		}
+		jpeg_inst = GET_INST(JPEG, i);
+		adev->jpeg.inst[i].aid_id =
+			jpeg_inst / adev->jpeg.num_inst_per_aid;
+	}
+}
+
+static const struct amdgpu_irq_src_funcs jpeg_v4_0_3_irq_funcs = {
+	.set = jpeg_v4_0_3_set_interrupt_state,
+	.process = jpeg_v4_0_3_process_interrupt,
+};
+
+static const struct amdgpu_irq_src_funcs jpeg_v4_0_3_ras_irq_funcs = {
+	.set = jpeg_v4_0_3_set_ras_interrupt_state,
+	.process = amdgpu_jpeg_process_poison_irq,
+};
+
+static void jpeg_v4_0_3_set_irq_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		adev->jpeg.inst->irq.num_types += adev->jpeg.num_jpeg_rings;
+	}
+	adev->jpeg.inst->irq.funcs = &jpeg_v4_0_3_irq_funcs;
+
+	adev->jpeg.inst->ras_poison_irq.num_types = 1;
+	adev->jpeg.inst->ras_poison_irq.funcs = &jpeg_v4_0_3_ras_irq_funcs;
+}
+
+const struct amdgpu_ip_block_version jpeg_v4_0_3_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_JPEG,
+	.major = 4,
+	.minor = 0,
+	.rev = 3,
+	.funcs = &jpeg_v4_0_3_ip_funcs,
+};
+
+static const struct amdgpu_ras_err_status_reg_entry jpeg_v4_0_3_ue_reg_list[] = {
+	{AMDGPU_RAS_REG_ENTRY(JPEG, 0, regVCN_UE_ERR_STATUS_LO_JPEG0S, regVCN_UE_ERR_STATUS_HI_JPEG0S),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "JPEG0S"},
+	{AMDGPU_RAS_REG_ENTRY(JPEG, 0, regVCN_UE_ERR_STATUS_LO_JPEG0D, regVCN_UE_ERR_STATUS_HI_JPEG0D),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "JPEG0D"},
+	{AMDGPU_RAS_REG_ENTRY(JPEG, 0, regVCN_UE_ERR_STATUS_LO_JPEG1S, regVCN_UE_ERR_STATUS_HI_JPEG1S),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "JPEG1S"},
+	{AMDGPU_RAS_REG_ENTRY(JPEG, 0, regVCN_UE_ERR_STATUS_LO_JPEG1D, regVCN_UE_ERR_STATUS_HI_JPEG1D),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "JPEG1D"},
+	{AMDGPU_RAS_REG_ENTRY(JPEG, 0, regVCN_UE_ERR_STATUS_LO_JPEG2S, regVCN_UE_ERR_STATUS_HI_JPEG2S),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "JPEG2S"},
+	{AMDGPU_RAS_REG_ENTRY(JPEG, 0, regVCN_UE_ERR_STATUS_LO_JPEG2D, regVCN_UE_ERR_STATUS_HI_JPEG2D),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "JPEG2D"},
+	{AMDGPU_RAS_REG_ENTRY(JPEG, 0, regVCN_UE_ERR_STATUS_LO_JPEG3S, regVCN_UE_ERR_STATUS_HI_JPEG3S),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "JPEG3S"},
+	{AMDGPU_RAS_REG_ENTRY(JPEG, 0, regVCN_UE_ERR_STATUS_LO_JPEG3D, regVCN_UE_ERR_STATUS_HI_JPEG3D),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "JPEG3D"},
+	{AMDGPU_RAS_REG_ENTRY(JPEG, 0, regVCN_UE_ERR_STATUS_LO_JPEG4S, regVCN_UE_ERR_STATUS_HI_JPEG4S),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "JPEG4S"},
+	{AMDGPU_RAS_REG_ENTRY(JPEG, 0, regVCN_UE_ERR_STATUS_LO_JPEG4D, regVCN_UE_ERR_STATUS_HI_JPEG4D),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "JPEG4D"},
+	{AMDGPU_RAS_REG_ENTRY(JPEG, 0, regVCN_UE_ERR_STATUS_LO_JPEG5S, regVCN_UE_ERR_STATUS_HI_JPEG5S),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "JPEG5S"},
+	{AMDGPU_RAS_REG_ENTRY(JPEG, 0, regVCN_UE_ERR_STATUS_LO_JPEG5D, regVCN_UE_ERR_STATUS_HI_JPEG5D),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "JPEG5D"},
+	{AMDGPU_RAS_REG_ENTRY(JPEG, 0, regVCN_UE_ERR_STATUS_LO_JPEG6S, regVCN_UE_ERR_STATUS_HI_JPEG6S),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "JPEG6S"},
+	{AMDGPU_RAS_REG_ENTRY(JPEG, 0, regVCN_UE_ERR_STATUS_LO_JPEG6D, regVCN_UE_ERR_STATUS_HI_JPEG6D),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "JPEG6D"},
+	{AMDGPU_RAS_REG_ENTRY(JPEG, 0, regVCN_UE_ERR_STATUS_LO_JPEG7S, regVCN_UE_ERR_STATUS_HI_JPEG7S),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "JPEG7S"},
+	{AMDGPU_RAS_REG_ENTRY(JPEG, 0, regVCN_UE_ERR_STATUS_LO_JPEG7D, regVCN_UE_ERR_STATUS_HI_JPEG7D),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "JPEG7D"},
+};
+
+static void jpeg_v4_0_3_inst_query_ras_error_count(struct amdgpu_device *adev,
+						   uint32_t jpeg_inst,
+						   void *ras_err_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_err_status;
+
+	/* jpeg v4_0_3 only support uncorrectable errors */
+	amdgpu_ras_inst_query_ras_error_count(adev,
+			jpeg_v4_0_3_ue_reg_list,
+			ARRAY_SIZE(jpeg_v4_0_3_ue_reg_list),
+			NULL, 0, GET_INST(VCN, jpeg_inst),
+			AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE,
+			&err_data->ue_count);
+}
+
+static void jpeg_v4_0_3_query_ras_error_count(struct amdgpu_device *adev,
+					      void *ras_err_status)
+{
+	uint32_t i;
+
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__JPEG)) {
+		dev_warn(adev->dev, "JPEG RAS is not supported\n");
+		return;
+	}
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; i++)
+		jpeg_v4_0_3_inst_query_ras_error_count(adev, i, ras_err_status);
+}
+
+static void jpeg_v4_0_3_inst_reset_ras_error_count(struct amdgpu_device *adev,
+						   uint32_t jpeg_inst)
+{
+	amdgpu_ras_inst_reset_ras_error_count(adev,
+			jpeg_v4_0_3_ue_reg_list,
+			ARRAY_SIZE(jpeg_v4_0_3_ue_reg_list),
+			GET_INST(VCN, jpeg_inst));
+}
+
+static void jpeg_v4_0_3_reset_ras_error_count(struct amdgpu_device *adev)
+{
+	uint32_t i;
+
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__JPEG)) {
+		dev_warn(adev->dev, "JPEG RAS is not supported\n");
+		return;
+	}
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; i++)
+		jpeg_v4_0_3_inst_reset_ras_error_count(adev, i);
+}
+
+static uint32_t jpeg_v4_0_3_query_poison_by_instance(struct amdgpu_device *adev,
+		uint32_t instance, uint32_t sub_block)
+{
+	uint32_t poison_stat = 0, reg_value = 0;
+
+	switch (sub_block) {
+	case AMDGPU_JPEG_V4_0_3_JPEG0:
+		reg_value = RREG32_SOC15(JPEG, instance, regUVD_RAS_JPEG0_STATUS);
+		poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_JPEG0_STATUS, POISONED_PF);
+		break;
+	case AMDGPU_JPEG_V4_0_3_JPEG1:
+		reg_value = RREG32_SOC15(JPEG, instance, regUVD_RAS_JPEG1_STATUS);
+		poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_JPEG1_STATUS, POISONED_PF);
+		break;
+	default:
+		break;
+	}
+
+	if (poison_stat)
+		dev_info(adev->dev, "Poison detected in JPEG%d sub_block%d\n",
+			instance, sub_block);
+
+	return poison_stat;
+}
+
+static bool jpeg_v4_0_3_query_ras_poison_status(struct amdgpu_device *adev)
+{
+	uint32_t inst = 0, sub = 0, poison_stat = 0;
+
+	for (inst = 0; inst < adev->jpeg.num_jpeg_inst; inst++)
+		for (sub = 0; sub < AMDGPU_JPEG_V4_0_3_MAX_SUB_BLOCK; sub++)
+			poison_stat +=
+			jpeg_v4_0_3_query_poison_by_instance(adev, inst, sub);
+
+	return !!poison_stat;
+}
+
+static const struct amdgpu_ras_block_hw_ops jpeg_v4_0_3_ras_hw_ops = {
+	.query_ras_error_count = jpeg_v4_0_3_query_ras_error_count,
+	.reset_ras_error_count = jpeg_v4_0_3_reset_ras_error_count,
+	.query_poison_status = jpeg_v4_0_3_query_ras_poison_status,
+};
+
+static int jpeg_v4_0_3_aca_bank_parser(struct aca_handle *handle, struct aca_bank *bank,
+				      enum aca_smu_type type, void *data)
+{
+	struct aca_bank_info info;
+	u64 misc0;
+	int ret;
+
+	ret = aca_bank_info_decode(bank, &info);
+	if (ret)
+		return ret;
+
+	misc0 = bank->regs[ACA_REG_IDX_MISC0];
+	switch (type) {
+	case ACA_SMU_TYPE_UE:
+		bank->aca_err_type = ACA_ERROR_TYPE_UE;
+		ret = aca_error_cache_log_bank_error(handle, &info, ACA_ERROR_TYPE_UE,
+						     1ULL);
+		break;
+	case ACA_SMU_TYPE_CE:
+		bank->aca_err_type = ACA_ERROR_TYPE_CE;
+		ret = aca_error_cache_log_bank_error(handle, &info, bank->aca_err_type,
+						     ACA_REG__MISC0__ERRCNT(misc0));
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return ret;
+}
+
+/* reference to smu driver if header file */
+static int jpeg_v4_0_3_err_codes[] = {
+	16, 17, 18, 19, 20, 21, 22, 23, /* JPEG[0-7][S|D] */
+	24, 25, 26, 27, 28, 29, 30, 31
+};
+
+static bool jpeg_v4_0_3_aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank,
+					 enum aca_smu_type type, void *data)
+{
+	u32 instlo;
+
+	instlo = ACA_REG__IPID__INSTANCEIDLO(bank->regs[ACA_REG_IDX_IPID]);
+	instlo &= GENMASK(31, 1);
+
+	if (instlo != mmSMNAID_AID0_MCA_SMU)
+		return false;
+
+	if (aca_bank_check_error_codes(handle->adev, bank,
+				       jpeg_v4_0_3_err_codes,
+				       ARRAY_SIZE(jpeg_v4_0_3_err_codes)))
+		return false;
+
+	return true;
+}
+
+static const struct aca_bank_ops jpeg_v4_0_3_aca_bank_ops = {
+	.aca_bank_parser = jpeg_v4_0_3_aca_bank_parser,
+	.aca_bank_is_valid = jpeg_v4_0_3_aca_bank_is_valid,
+};
+
+static const struct aca_info jpeg_v4_0_3_aca_info = {
+	.hwip = ACA_HWIP_TYPE_SMU,
+	.mask = ACA_ERROR_UE_MASK,
+	.bank_ops = &jpeg_v4_0_3_aca_bank_ops,
+};
+
+static int jpeg_v4_0_3_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
+{
+	int r;
+
+	r = amdgpu_ras_block_late_init(adev, ras_block);
+	if (r)
+		return r;
+
+	if (amdgpu_ras_is_supported(adev, ras_block->block) &&
+		adev->jpeg.inst->ras_poison_irq.funcs) {
+		r = amdgpu_irq_get(adev, &adev->jpeg.inst->ras_poison_irq, 0);
+		if (r)
+			goto late_fini;
+	}
+
+	r = amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__JPEG,
+				&jpeg_v4_0_3_aca_info, NULL);
+	if (r)
+		goto late_fini;
+
+	return 0;
+
+late_fini:
+	amdgpu_ras_block_late_fini(adev, ras_block);
+
+	return r;
+}
+
+static struct amdgpu_jpeg_ras jpeg_v4_0_3_ras = {
+	.ras_block = {
+		.hw_ops = &jpeg_v4_0_3_ras_hw_ops,
+		.ras_late_init = jpeg_v4_0_3_ras_late_init,
+	},
+};
+
+static void jpeg_v4_0_3_set_ras_funcs(struct amdgpu_device *adev)
+{
+	adev->jpeg.ras = &jpeg_v4_0_3_ras;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.h b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.h
new file mode 100644
index 000000000000..2e110d04af84
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.h
@@ -0,0 +1,74 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __JPEG_V4_0_3_H__
+#define __JPEG_V4_0_3_H__
+
+#define regUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET			0x1bfff
+#define regUVD_JPEG_GPCOM_CMD_INTERNAL_OFFSET				0x404d
+#define regUVD_JPEG_GPCOM_DATA0_INTERNAL_OFFSET				0x404e
+#define regUVD_JPEG_GPCOM_DATA1_INTERNAL_OFFSET				0x404f
+#define regUVD_LMI_JRBC_RB_MEM_WR_64BIT_BAR_LOW_INTERNAL_OFFSET		0x40ab
+#define regUVD_LMI_JRBC_RB_MEM_WR_64BIT_BAR_HIGH_INTERNAL_OFFSET	0x40ac
+#define regUVD_LMI_JRBC_IB_VMID_INTERNAL_OFFSET				0x40a4
+#define regUVD_LMI_JPEG_VMID_INTERNAL_OFFSET				0x40a6
+#define regUVD_LMI_JRBC_IB_64BIT_BAR_LOW_INTERNAL_OFFSET		0x40b6
+#define regUVD_LMI_JRBC_IB_64BIT_BAR_HIGH_INTERNAL_OFFSET		0x40b7
+#define regUVD_JRBC_IB_SIZE_INTERNAL_OFFSET				0x4082
+#define regUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_LOW_INTERNAL_OFFSET		0x42d4
+#define regUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_HIGH_INTERNAL_OFFSET	0x42d5
+#define regUVD_JRBC_RB_COND_RD_TIMER_INTERNAL_OFFSET			0x4085
+#define regUVD_JRBC_RB_REF_DATA_INTERNAL_OFFSET				0x4084
+#define regUVD_JRBC_STATUS_INTERNAL_OFFSET				0x4089
+#define regUVD_JPEG_PITCH_INTERNAL_OFFSET				0x4043
+#define regUVD_JRBC0_UVD_JRBC_SCRATCH0_INTERNAL_OFFSET			0x4094
+#define regUVD_JRBC_EXTERNAL_MCM_ADDR_INTERNAL_OFFSET			0x1bffe
+
+#define JRBC_DEC_EXTERNAL_REG_WRITE_ADDR				0x18000
+
+enum amdgpu_jpeg_v4_0_3_sub_block {
+	AMDGPU_JPEG_V4_0_3_JPEG0 = 0,
+	AMDGPU_JPEG_V4_0_3_JPEG1,
+
+	AMDGPU_JPEG_V4_0_3_MAX_SUB_BLOCK,
+};
+
+extern const struct amdgpu_ip_block_version jpeg_v4_0_3_ip_block;
+
+void jpeg_v4_0_3_dec_ring_emit_ib(struct amdgpu_ring *ring,
+				struct amdgpu_job *job,
+				struct amdgpu_ib *ib,
+				uint32_t flags);
+void jpeg_v4_0_3_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				unsigned int flags);
+void jpeg_v4_0_3_dec_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					unsigned int vmid, uint64_t pd_addr);
+void jpeg_v4_0_3_ring_emit_hdp_flush(struct amdgpu_ring *ring);
+void jpeg_v4_0_3_dec_ring_nop(struct amdgpu_ring *ring, uint32_t count);
+void jpeg_v4_0_3_dec_ring_insert_start(struct amdgpu_ring *ring);
+void jpeg_v4_0_3_dec_ring_insert_end(struct amdgpu_ring *ring);
+void jpeg_v4_0_3_dec_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg, uint32_t val);
+void jpeg_v4_0_3_dec_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+					uint32_t val, uint32_t mask);
+
+#endif /* __JPEG_V4_0_3_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.c
new file mode 100644
index 000000000000..54fd9c800c40
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.c
@@ -0,0 +1,872 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "amdgpu_jpeg.h"
+#include "amdgpu_pm.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "jpeg_v2_0.h"
+#include "jpeg_v4_0_5.h"
+#include "mmsch_v4_0.h"
+
+#include "vcn/vcn_4_0_5_offset.h"
+#include "vcn/vcn_4_0_5_sh_mask.h"
+#include "ivsrcid/vcn/irqsrcs_vcn_4_0.h"
+
+#define mmUVD_DPG_LMA_CTL						regUVD_DPG_LMA_CTL
+#define mmUVD_DPG_LMA_CTL_BASE_IDX					regUVD_DPG_LMA_CTL_BASE_IDX
+#define mmUVD_DPG_LMA_DATA						regUVD_DPG_LMA_DATA
+#define mmUVD_DPG_LMA_DATA_BASE_IDX					regUVD_DPG_LMA_DATA_BASE_IDX
+
+#define regUVD_JPEG_PITCH_INTERNAL_OFFSET		0x401f
+#define regJPEG_DEC_GFX10_ADDR_CONFIG_INTERNAL_OFFSET	0x4026
+#define regJPEG_SYS_INT_EN_INTERNAL_OFFSET		0x4141
+#define regJPEG_CGC_CTRL_INTERNAL_OFFSET		0x4161
+#define regJPEG_CGC_GATE_INTERNAL_OFFSET		0x4160
+#define regUVD_NO_OP_INTERNAL_OFFSET			0x0029
+
+static const struct amdgpu_hwip_reg_entry jpeg_reg_list_4_0_5[] = {
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_INT_STAT),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_CNTL),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_SIZE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_ADDR_MODE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_GFX10_ADDR_CONFIG),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_Y_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_UV_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_PITCH),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_UV_PITCH),
+};
+
+static void jpeg_v4_0_5_set_dec_ring_funcs(struct amdgpu_device *adev);
+static void jpeg_v4_0_5_set_irq_funcs(struct amdgpu_device *adev);
+static int jpeg_v4_0_5_set_powergating_state(struct amdgpu_ip_block *ip_block,
+				enum amd_powergating_state state);
+static void jpeg_v4_0_5_dec_ring_set_wptr(struct amdgpu_ring *ring);
+
+static int amdgpu_ih_clientid_jpeg[] = {
+	SOC15_IH_CLIENTID_VCN,
+	SOC15_IH_CLIENTID_VCN1
+};
+
+
+
+/**
+ * jpeg_v4_0_5_early_init - set function pointers
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Set ring and irq function pointers
+ */
+static int jpeg_v4_0_5_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	switch (amdgpu_ip_version(adev, UVD_HWIP, 0)) {
+	case IP_VERSION(4, 0, 5):
+		adev->jpeg.num_jpeg_inst = 1;
+		break;
+	case IP_VERSION(4, 0, 6):
+		adev->jpeg.num_jpeg_inst = 2;
+		break;
+	default:
+		DRM_DEV_ERROR(adev->dev,
+			"Failed to init vcn ip block(UVD_HWIP:0x%x)\n",
+			amdgpu_ip_version(adev, UVD_HWIP, 0));
+		return -EINVAL;
+	}
+
+	adev->jpeg.num_jpeg_rings = 1;
+
+	jpeg_v4_0_5_set_dec_ring_funcs(adev);
+	jpeg_v4_0_5_set_irq_funcs(adev);
+
+	return 0;
+}
+
+/**
+ * jpeg_v4_0_5_sw_init - sw init for JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Load firmware and sw initialization
+ */
+static int jpeg_v4_0_5_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	int r, i;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+
+		/* JPEG TRAP */
+		r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_jpeg[i],
+				VCN_4_0__SRCID__JPEG_DECODE, &adev->jpeg.inst[i].irq);
+		if (r)
+			return r;
+
+		/* JPEG DJPEG POISON EVENT */
+		r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_jpeg[i],
+			VCN_4_0__SRCID_DJPEG0_POISON, &adev->jpeg.inst[i].irq);
+		if (r)
+			return r;
+
+		/* JPEG EJPEG POISON EVENT */
+		r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_jpeg[i],
+			VCN_4_0__SRCID_EJPEG0_POISON, &adev->jpeg.inst[i].irq);
+		if (r)
+			return r;
+	}
+
+	r = amdgpu_jpeg_sw_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_resume(adev);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+
+		ring = adev->jpeg.inst[i].ring_dec;
+		ring->use_doorbell = true;
+		ring->vm_hub = AMDGPU_MMHUB0(0);
+		ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 1 + 8 * i;
+		sprintf(ring->name, "jpeg_dec_%d", i);
+		r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst[i].irq,
+				     0, AMDGPU_RING_PRIO_DEFAULT, NULL);
+		if (r)
+			return r;
+
+		adev->jpeg.internal.jpeg_pitch[0] = regUVD_JPEG_PITCH_INTERNAL_OFFSET;
+		adev->jpeg.inst[i].external.jpeg_pitch[0] = SOC15_REG_OFFSET(JPEG, i, regUVD_JPEG_PITCH);
+	}
+
+	r = amdgpu_jpeg_reg_dump_init(adev, jpeg_reg_list_4_0_5, ARRAY_SIZE(jpeg_reg_list_4_0_5));
+	if (r)
+		return r;
+
+	adev->jpeg.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->jpeg.inst[0].ring_dec[0]);
+	if (!amdgpu_sriov_vf(adev))
+		adev->jpeg.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+	r = amdgpu_jpeg_sysfs_reset_mask_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+/**
+ * jpeg_v4_0_5_sw_fini - sw fini for JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * JPEG suspend and free up sw allocation
+ */
+static int jpeg_v4_0_5_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = amdgpu_jpeg_suspend(adev);
+	if (r)
+		return r;
+
+	amdgpu_jpeg_sysfs_reset_mask_fini(adev);
+	r = amdgpu_jpeg_sw_fini(adev);
+
+	return r;
+}
+
+/**
+ * jpeg_v4_0_5_hw_init - start and test JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ */
+static int jpeg_v4_0_5_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	int i, r = 0;
+
+	// TODO: Enable ring test with DPG support
+	if (adev->pg_flags & AMD_PG_SUPPORT_JPEG_DPG) {
+		return 0;
+	}
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+
+		ring = adev->jpeg.inst[i].ring_dec;
+		r = amdgpu_ring_test_helper(ring);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+/**
+ * jpeg_v4_0_5_hw_fini - stop the hardware block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Stop the JPEG block, mark ring as not ready any more
+ */
+static int jpeg_v4_0_5_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i;
+
+	cancel_delayed_work_sync(&adev->jpeg.idle_work);
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+
+		if (!amdgpu_sriov_vf(adev)) {
+			if (adev->jpeg.cur_state != AMD_PG_STATE_GATE &&
+			    RREG32_SOC15(JPEG, i, regUVD_JRBC_STATUS))
+				jpeg_v4_0_5_set_powergating_state(ip_block, AMD_PG_STATE_GATE);
+		}
+	}
+	return 0;
+}
+
+/**
+ * jpeg_v4_0_5_suspend - suspend JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * HW fini and suspend JPEG block
+ */
+static int jpeg_v4_0_5_suspend(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	r = jpeg_v4_0_5_hw_fini(ip_block);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_suspend(ip_block->adev);
+
+	return r;
+}
+
+/**
+ * jpeg_v4_0_5_resume - resume JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Resume firmware and hw init JPEG block
+ */
+static int jpeg_v4_0_5_resume(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	r = amdgpu_jpeg_resume(ip_block->adev);
+	if (r)
+		return r;
+
+	r = jpeg_v4_0_5_hw_init(ip_block);
+
+	return r;
+}
+
+static void jpeg_v4_0_5_disable_clock_gating(struct amdgpu_device *adev, int inst)
+{
+	uint32_t data = 0;
+
+	data = RREG32_SOC15(JPEG, inst, regJPEG_CGC_CTRL);
+	if (adev->cg_flags & AMD_CG_SUPPORT_JPEG_MGCG) {
+		data |= 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+		data &= (~JPEG_CGC_CTRL__JPEG_DEC_MODE_MASK);
+	} else {
+		data &= ~JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	}
+
+	data |= 1 << JPEG_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << JPEG_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(JPEG, inst, regJPEG_CGC_CTRL, data);
+
+	data = RREG32_SOC15(JPEG, inst, regJPEG_CGC_GATE);
+	data &= ~(JPEG_CGC_GATE__JPEG_DEC_MASK
+		| JPEG_CGC_GATE__JPEG2_DEC_MASK
+		| JPEG_CGC_GATE__JMCIF_MASK
+		| JPEG_CGC_GATE__JRBBM_MASK);
+	WREG32_SOC15(JPEG, inst, regJPEG_CGC_GATE, data);
+}
+
+static void jpeg_v4_0_5_enable_clock_gating(struct amdgpu_device *adev, int inst)
+{
+	uint32_t data = 0;
+
+	data = RREG32_SOC15(JPEG, inst, regJPEG_CGC_CTRL);
+	if (adev->cg_flags & AMD_CG_SUPPORT_JPEG_MGCG) {
+		data |= 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+		data |= JPEG_CGC_CTRL__JPEG_DEC_MODE_MASK;
+	} else {
+		data &= ~JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	}
+
+	data |= 1 << JPEG_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << JPEG_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(JPEG, inst, regJPEG_CGC_CTRL, data);
+
+	data = RREG32_SOC15(JPEG, inst, regJPEG_CGC_GATE);
+	data |= (JPEG_CGC_GATE__JPEG_DEC_MASK
+		|JPEG_CGC_GATE__JPEG2_DEC_MASK
+		|JPEG_CGC_GATE__JMCIF_MASK
+		|JPEG_CGC_GATE__JRBBM_MASK);
+	WREG32_SOC15(JPEG, inst, regJPEG_CGC_GATE, data);
+}
+
+static void jpeg_engine_4_0_5_dpg_clock_gating_mode(struct amdgpu_device *adev,
+			int inst_idx, uint8_t indirect)
+{
+	uint32_t data = 0;
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_JPEG_MGCG)
+		data |= 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		data |= 0 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+
+	data |= 1 << JPEG_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << JPEG_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15_JPEG_DPG_MODE(inst_idx, regJPEG_CGC_CTRL_INTERNAL_OFFSET, data, indirect);
+
+	data = 0;
+	WREG32_SOC15_JPEG_DPG_MODE(inst_idx, regJPEG_CGC_GATE_INTERNAL_OFFSET,
+				data, indirect);
+}
+
+static int jpeg_v4_0_5_disable_static_power_gating(struct amdgpu_device *adev, int inst)
+{
+	if (adev->pg_flags & AMD_PG_SUPPORT_JPEG) {
+		WREG32(SOC15_REG_OFFSET(JPEG, inst, regUVD_IPX_DLDO_CONFIG),
+			1 << UVD_IPX_DLDO_CONFIG__ONO1_PWR_CONFIG__SHIFT);
+		SOC15_WAIT_ON_RREG(JPEG, inst, regUVD_IPX_DLDO_STATUS,
+			0, UVD_IPX_DLDO_STATUS__ONO1_PWR_STATUS_MASK);
+	}
+
+	/* disable anti hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, inst, regUVD_JPEG_POWER_STATUS), 0,
+		~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+
+	/* keep the JPEG in static PG mode */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, inst, regUVD_JPEG_POWER_STATUS), 0,
+		~UVD_JPEG_POWER_STATUS__JPEG_PG_MODE_MASK);
+
+	return 0;
+}
+
+static int jpeg_v4_0_5_enable_static_power_gating(struct amdgpu_device *adev, int inst)
+{
+	/* enable anti hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, inst, regUVD_JPEG_POWER_STATUS),
+		UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK,
+		~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_JPEG) {
+		WREG32(SOC15_REG_OFFSET(JPEG, inst, regUVD_IPX_DLDO_CONFIG),
+			2 << UVD_IPX_DLDO_CONFIG__ONO1_PWR_CONFIG__SHIFT);
+		SOC15_WAIT_ON_RREG(JPEG, inst, regUVD_IPX_DLDO_STATUS,
+			1 << UVD_IPX_DLDO_STATUS__ONO1_PWR_STATUS__SHIFT,
+			UVD_IPX_DLDO_STATUS__ONO1_PWR_STATUS_MASK);
+	}
+
+	return 0;
+}
+
+/**
+ * jpeg_v4_0_5_start_dpg_mode - Jpeg start with dpg mode
+ *
+ * @adev: amdgpu_device pointer
+ * @inst_idx: instance number index
+ * @indirect: indirectly write sram
+ *
+ * Start JPEG block with dpg mode
+ */
+static void jpeg_v4_0_5_start_dpg_mode(struct amdgpu_device *adev, int inst_idx, bool indirect)
+{
+	struct amdgpu_ring *ring = adev->jpeg.inst[inst_idx].ring_dec;
+	uint32_t reg_data = 0;
+
+	/* enable anti hang mechanism */
+	reg_data = RREG32_SOC15(JPEG, inst_idx, regUVD_JPEG_POWER_STATUS);
+	reg_data &= ~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK;
+	reg_data |=  0x1;
+	WREG32_SOC15(JPEG, inst_idx, regUVD_JPEG_POWER_STATUS, reg_data);
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_JPEG) {
+		WREG32(SOC15_REG_OFFSET(JPEG, inst_idx, regUVD_IPX_DLDO_CONFIG),
+			2 << UVD_IPX_DLDO_CONFIG__ONO1_PWR_CONFIG__SHIFT);
+		SOC15_WAIT_ON_RREG(JPEG, inst_idx, regUVD_IPX_DLDO_STATUS,
+			1 << UVD_IPX_DLDO_STATUS__ONO1_PWR_STATUS__SHIFT,
+			UVD_IPX_DLDO_STATUS__ONO1_PWR_STATUS_MASK);
+	}
+
+	reg_data = RREG32_SOC15(JPEG, inst_idx, regUVD_JPEG_POWER_STATUS);
+	reg_data |= UVD_JPEG_POWER_STATUS__JPEG_PG_MODE_MASK;
+	WREG32_SOC15(JPEG, inst_idx, regUVD_JPEG_POWER_STATUS, reg_data);
+
+	if (indirect)
+		adev->jpeg.inst[inst_idx].dpg_sram_curr_addr =
+					(uint32_t *)adev->jpeg.inst[inst_idx].dpg_sram_cpu_addr;
+
+	jpeg_engine_4_0_5_dpg_clock_gating_mode(adev, inst_idx, indirect);
+
+	/* MJPEG global tiling registers */
+	WREG32_SOC15_JPEG_DPG_MODE(inst_idx, regJPEG_DEC_GFX10_ADDR_CONFIG_INTERNAL_OFFSET,
+	adev->gfx.config.gb_addr_config, indirect);
+	/* enable System Interrupt for JRBC */
+	WREG32_SOC15_JPEG_DPG_MODE(inst_idx, regJPEG_SYS_INT_EN_INTERNAL_OFFSET,
+	JPEG_SYS_INT_EN__DJRBC_MASK, indirect);
+
+	/* add nop to workaround PSP size check */
+	WREG32_SOC15_JPEG_DPG_MODE(inst_idx, regUVD_NO_OP_INTERNAL_OFFSET, 0, indirect);
+
+	if (indirect)
+		amdgpu_jpeg_psp_update_sram(adev, inst_idx, 0);
+
+	WREG32_SOC15(JPEG, inst_idx, regUVD_LMI_JRBC_RB_VMID, 0);
+	WREG32_SOC15(JPEG, inst_idx, regUVD_JRBC_RB_CNTL, (0x00000001L | 0x00000002L));
+	WREG32_SOC15(JPEG, inst_idx, regUVD_LMI_JRBC_RB_64BIT_BAR_LOW,
+		lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15(JPEG, inst_idx, regUVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
+		upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(JPEG, inst_idx, regUVD_JRBC_RB_RPTR, 0);
+	WREG32_SOC15(JPEG, inst_idx, regUVD_JRBC_RB_WPTR, 0);
+	WREG32_SOC15(JPEG, inst_idx, regUVD_JRBC_RB_CNTL, 0x00000002L);
+	WREG32_SOC15(JPEG, inst_idx, regUVD_JRBC_RB_SIZE, ring->ring_size / 4);
+	ring->wptr = RREG32_SOC15(JPEG, inst_idx, regUVD_JRBC_RB_WPTR);
+}
+
+/**
+ * jpeg_v4_0_5_stop_dpg_mode - Jpeg stop with dpg mode
+ *
+ * @adev: amdgpu_device pointer
+ * @inst_idx: instance number index
+ *
+ * Stop JPEG block with dpg mode
+ */
+static void jpeg_v4_0_5_stop_dpg_mode(struct amdgpu_device *adev, int inst_idx)
+{
+	uint32_t reg_data = 0;
+
+	reg_data = RREG32_SOC15(JPEG, inst_idx, regUVD_JPEG_POWER_STATUS);
+	reg_data &= ~UVD_JPEG_POWER_STATUS__JPEG_PG_MODE_MASK;
+	WREG32_SOC15(JPEG, inst_idx, regUVD_JPEG_POWER_STATUS, reg_data);
+
+}
+
+/**
+ * jpeg_v4_0_5_start - start JPEG block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the JPEG block
+ */
+static int jpeg_v4_0_5_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	int r, i;
+
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_jpeg(adev, true);
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+
+		ring = adev->jpeg.inst[i].ring_dec;
+		/* doorbell programming is done for every playback */
+		adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
+				(adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 8 * i, i);
+
+		WREG32_SOC15(VCN, i, regVCN_JPEG_DB_CTRL,
+			ring->doorbell_index << VCN_JPEG_DB_CTRL__OFFSET__SHIFT |
+			VCN_JPEG_DB_CTRL__EN_MASK);
+
+		if (adev->pg_flags & AMD_PG_SUPPORT_JPEG_DPG) {
+			jpeg_v4_0_5_start_dpg_mode(adev, i, adev->jpeg.indirect_sram);
+			continue;
+		}
+
+		/* disable power gating */
+		r = jpeg_v4_0_5_disable_static_power_gating(adev, i);
+		if (r)
+			return r;
+
+		/* JPEG disable CGC */
+		jpeg_v4_0_5_disable_clock_gating(adev, i);
+
+		/* MJPEG global tiling registers */
+		WREG32_SOC15(JPEG, i, regJPEG_DEC_GFX10_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config);
+
+		/* enable JMI channel */
+		WREG32_P(SOC15_REG_OFFSET(JPEG, i, regUVD_JMI_CNTL), 0,
+			~UVD_JMI_CNTL__SOFT_RESET_MASK);
+
+		/* enable System Interrupt for JRBC */
+		WREG32_P(SOC15_REG_OFFSET(JPEG, i, regJPEG_SYS_INT_EN),
+			JPEG_SYS_INT_EN__DJRBC_MASK,
+			~JPEG_SYS_INT_EN__DJRBC_MASK);
+
+		WREG32_SOC15(JPEG, i, regUVD_LMI_JRBC_RB_VMID, 0);
+		WREG32_SOC15(JPEG, i, regUVD_JRBC_RB_CNTL, (0x00000001L | 0x00000002L));
+		WREG32_SOC15(JPEG, i, regUVD_LMI_JRBC_RB_64BIT_BAR_LOW,
+			lower_32_bits(ring->gpu_addr));
+		WREG32_SOC15(JPEG, i, regUVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
+			upper_32_bits(ring->gpu_addr));
+		WREG32_SOC15(JPEG, i, regUVD_JRBC_RB_RPTR, 0);
+		WREG32_SOC15(JPEG, i, regUVD_JRBC_RB_WPTR, 0);
+		WREG32_SOC15(JPEG, i, regUVD_JRBC_RB_CNTL, 0x00000002L);
+		WREG32_SOC15(JPEG, i, regUVD_JRBC_RB_SIZE, ring->ring_size / 4);
+		ring->wptr = RREG32_SOC15(JPEG, i, regUVD_JRBC_RB_WPTR);
+	}
+
+	return 0;
+}
+
+/**
+ * jpeg_v4_0_5_stop - stop JPEG block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * stop the JPEG block
+ */
+static int jpeg_v4_0_5_stop(struct amdgpu_device *adev)
+{
+	int r, i;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+
+		if (adev->pg_flags & AMD_PG_SUPPORT_JPEG_DPG) {
+			jpeg_v4_0_5_stop_dpg_mode(adev, i);
+			continue;
+		}
+
+		/* reset JMI */
+		WREG32_P(SOC15_REG_OFFSET(JPEG, i, regUVD_JMI_CNTL),
+			UVD_JMI_CNTL__SOFT_RESET_MASK,
+			~UVD_JMI_CNTL__SOFT_RESET_MASK);
+
+		jpeg_v4_0_5_enable_clock_gating(adev, i);
+
+		/* enable power gating */
+		r = jpeg_v4_0_5_enable_static_power_gating(adev, i);
+		if (r)
+			return r;
+	}
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_jpeg(adev, false);
+
+	return 0;
+}
+
+/**
+ * jpeg_v4_0_5_dec_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t jpeg_v4_0_5_dec_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32_SOC15(JPEG, ring->me, regUVD_JRBC_RB_RPTR);
+}
+
+/**
+ * jpeg_v4_0_5_dec_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t jpeg_v4_0_5_dec_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell)
+		return *ring->wptr_cpu_addr;
+	else
+		return RREG32_SOC15(JPEG, ring->me, regUVD_JRBC_RB_WPTR);
+}
+
+/**
+ * jpeg_v4_0_5_dec_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void jpeg_v4_0_5_dec_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
+		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+	} else {
+		WREG32_SOC15(JPEG, ring->me, regUVD_JRBC_RB_WPTR, lower_32_bits(ring->wptr));
+	}
+}
+
+static bool jpeg_v4_0_5_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, ret = 1;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+
+		ret &= (((RREG32_SOC15(JPEG, i, regUVD_JRBC_STATUS) &
+			UVD_JRBC_STATUS__RB_JOB_DONE_MASK) ==
+			UVD_JRBC_STATUS__RB_JOB_DONE_MASK));
+	}
+	return ret;
+}
+
+static int jpeg_v4_0_5_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+
+		return SOC15_WAIT_ON_RREG(JPEG, i, regUVD_JRBC_STATUS,
+			UVD_JRBC_STATUS__RB_JOB_DONE_MASK,
+			UVD_JRBC_STATUS__RB_JOB_DONE_MASK);
+	}
+
+	return 0;
+}
+
+static int jpeg_v4_0_5_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = state == AMD_CG_STATE_GATE;
+	int i;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+
+		if (enable) {
+			if (!jpeg_v4_0_5_is_idle(ip_block))
+				return -EBUSY;
+
+			jpeg_v4_0_5_enable_clock_gating(adev, i);
+		} else {
+			jpeg_v4_0_5_disable_clock_gating(adev, i);
+		}
+	}
+
+	return 0;
+}
+
+static int jpeg_v4_0_5_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int ret;
+
+	if (amdgpu_sriov_vf(adev)) {
+		adev->jpeg.cur_state = AMD_PG_STATE_UNGATE;
+		return 0;
+	}
+
+	if (state == adev->jpeg.cur_state)
+		return 0;
+
+	if (state == AMD_PG_STATE_GATE)
+		ret = jpeg_v4_0_5_stop(adev);
+	else
+		ret = jpeg_v4_0_5_start(adev);
+
+	if (!ret)
+		adev->jpeg.cur_state = state;
+
+	return ret;
+}
+
+static int jpeg_v4_0_5_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	uint32_t ip_instance;
+
+	DRM_DEBUG("IH: JPEG TRAP\n");
+
+	switch (entry->client_id) {
+	case SOC15_IH_CLIENTID_VCN:
+		ip_instance = 0;
+		break;
+	case SOC15_IH_CLIENTID_VCN1:
+		ip_instance = 1;
+		break;
+	default:
+		DRM_ERROR("Unhandled client id: %d\n", entry->client_id);
+		return 0;
+	}
+
+	switch (entry->src_id) {
+	case VCN_4_0__SRCID__JPEG_DECODE:
+		amdgpu_fence_process(adev->jpeg.inst[ip_instance].ring_dec);
+		break;
+	case VCN_4_0__SRCID_DJPEG0_POISON:
+	case VCN_4_0__SRCID_EJPEG0_POISON:
+		amdgpu_jpeg_process_poison_irq(adev, source, entry);
+		break;
+	default:
+		DRM_DEV_ERROR(adev->dev, "Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+static int jpeg_v4_0_5_ring_reset(struct amdgpu_ring *ring,
+				  unsigned int vmid,
+				  struct amdgpu_fence *timedout_fence)
+{
+	int r;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+	r = jpeg_v4_0_5_stop(ring->adev);
+	if (r)
+		return r;
+	r = jpeg_v4_0_5_start(ring->adev);
+	if (r)
+		return r;
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static const struct amd_ip_funcs jpeg_v4_0_5_ip_funcs = {
+	.name = "jpeg_v4_0_5",
+	.early_init = jpeg_v4_0_5_early_init,
+	.sw_init = jpeg_v4_0_5_sw_init,
+	.sw_fini = jpeg_v4_0_5_sw_fini,
+	.hw_init = jpeg_v4_0_5_hw_init,
+	.hw_fini = jpeg_v4_0_5_hw_fini,
+	.suspend = jpeg_v4_0_5_suspend,
+	.resume = jpeg_v4_0_5_resume,
+	.is_idle = jpeg_v4_0_5_is_idle,
+	.wait_for_idle = jpeg_v4_0_5_wait_for_idle,
+	.set_clockgating_state = jpeg_v4_0_5_set_clockgating_state,
+	.set_powergating_state = jpeg_v4_0_5_set_powergating_state,
+	.dump_ip_state = amdgpu_jpeg_dump_ip_state,
+	.print_ip_state = amdgpu_jpeg_print_ip_state,
+};
+
+static const struct amdgpu_ring_funcs jpeg_v4_0_5_dec_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_JPEG,
+	.align_mask = 0xf,
+	.get_rptr = jpeg_v4_0_5_dec_ring_get_rptr,
+	.get_wptr = jpeg_v4_0_5_dec_ring_get_wptr,
+	.set_wptr = jpeg_v4_0_5_dec_ring_set_wptr,
+	.parse_cs = amdgpu_jpeg_dec_parse_cs,
+	.emit_frame_size =
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
+		8 + /* jpeg_v4_0_5_dec_ring_emit_vm_flush */
+		18 + 18 + /* jpeg_v4_0_5_dec_ring_emit_fence x2 vm fence */
+		8 + 16,
+	.emit_ib_size = 22, /* jpeg_v4_0_5_dec_ring_emit_ib */
+	.emit_ib = jpeg_v2_0_dec_ring_emit_ib,
+	.emit_fence = jpeg_v2_0_dec_ring_emit_fence,
+	.emit_vm_flush = jpeg_v2_0_dec_ring_emit_vm_flush,
+	.test_ring = amdgpu_jpeg_dec_ring_test_ring,
+	.test_ib = amdgpu_jpeg_dec_ring_test_ib,
+	.insert_nop = jpeg_v2_0_dec_ring_nop,
+	.insert_start = jpeg_v2_0_dec_ring_insert_start,
+	.insert_end = jpeg_v2_0_dec_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_jpeg_ring_begin_use,
+	.end_use = amdgpu_jpeg_ring_end_use,
+	.emit_wreg = jpeg_v2_0_dec_ring_emit_wreg,
+	.emit_reg_wait = jpeg_v2_0_dec_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = jpeg_v4_0_5_ring_reset,
+};
+
+static void jpeg_v4_0_5_set_dec_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+
+		adev->jpeg.inst[i].ring_dec->funcs = &jpeg_v4_0_5_dec_ring_vm_funcs;
+		adev->jpeg.inst[i].ring_dec->me = i;
+	}
+}
+
+static const struct amdgpu_irq_src_funcs jpeg_v4_0_5_irq_funcs = {
+	.process = jpeg_v4_0_5_process_interrupt,
+};
+
+static void jpeg_v4_0_5_set_irq_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+
+		adev->jpeg.inst[i].irq.num_types = 1;
+		adev->jpeg.inst[i].irq.funcs = &jpeg_v4_0_5_irq_funcs;
+	}
+}
+
+const struct amdgpu_ip_block_version jpeg_v4_0_5_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_JPEG,
+	.major = 4,
+	.minor = 0,
+	.rev = 5,
+	.funcs = &jpeg_v4_0_5_ip_funcs,
+};
+
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.h b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.h
new file mode 100644
index 000000000000..c5eee572079c
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.h
@@ -0,0 +1,35 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __JPEG_V4_0_5_H__
+#define __JPEG_V4_0_5_H__
+
+enum amdgpu_jpeg_v4_0_5_sub_block {
+	AMDGPU_JPEG_V4_0_5_JPEG0 = 0,
+
+	AMDGPU_JPEG_V4_0_5_MAX_SUB_BLOCK,
+};
+
+extern const struct amdgpu_ip_block_version jpeg_v4_0_5_ip_block;
+
+#endif /* __JPEG_V4_0_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.c
new file mode 100644
index 000000000000..46bf15dce2bd
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.c
@@ -0,0 +1,733 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "amdgpu_jpeg.h"
+#include "amdgpu_pm.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "jpeg_v2_0.h"
+#include "jpeg_v4_0_3.h"
+
+#include "vcn/vcn_5_0_0_offset.h"
+#include "vcn/vcn_5_0_0_sh_mask.h"
+#include "ivsrcid/vcn/irqsrcs_vcn_5_0.h"
+#include "jpeg_v5_0_0.h"
+
+static const struct amdgpu_hwip_reg_entry jpeg_reg_list_5_0[] = {
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_INT_STAT),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_CNTL),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_SIZE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_ADDR_MODE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_GFX10_ADDR_CONFIG),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_Y_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_UV_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_PITCH),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_UV_PITCH),
+};
+
+static void jpeg_v5_0_0_set_dec_ring_funcs(struct amdgpu_device *adev);
+static void jpeg_v5_0_0_set_irq_funcs(struct amdgpu_device *adev);
+static int jpeg_v5_0_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
+				enum amd_powergating_state state);
+
+/**
+ * jpeg_v5_0_0_early_init - set function pointers
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Set ring and irq function pointers
+ */
+static int jpeg_v5_0_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	adev->jpeg.num_jpeg_inst = 1;
+	adev->jpeg.num_jpeg_rings = 1;
+
+	jpeg_v5_0_0_set_dec_ring_funcs(adev);
+	jpeg_v5_0_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+/**
+ * jpeg_v5_0_0_sw_init - sw init for JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Load firmware and sw initialization
+ */
+static int jpeg_v5_0_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	int r;
+
+	/* JPEG TRAP */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+		VCN_5_0__SRCID__JPEG_DECODE, &adev->jpeg.inst->irq);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_sw_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_resume(adev);
+	if (r)
+		return r;
+
+	ring = adev->jpeg.inst->ring_dec;
+	ring->use_doorbell = true;
+	ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 1;
+	ring->vm_hub = AMDGPU_MMHUB0(0);
+
+	sprintf(ring->name, "jpeg_dec");
+	r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst->irq, 0,
+			     AMDGPU_RING_PRIO_DEFAULT, NULL);
+	if (r)
+		return r;
+
+	adev->jpeg.internal.jpeg_pitch[0] = regUVD_JPEG_PITCH_INTERNAL_OFFSET;
+	adev->jpeg.inst->external.jpeg_pitch[0] = SOC15_REG_OFFSET(JPEG, 0, regUVD_JPEG_PITCH);
+
+	r = amdgpu_jpeg_reg_dump_init(adev, jpeg_reg_list_5_0, ARRAY_SIZE(jpeg_reg_list_5_0));
+	if (r)
+		return r;
+
+	adev->jpeg.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->jpeg.inst[0].ring_dec[0]);
+	if (!amdgpu_sriov_vf(adev))
+		adev->jpeg.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+	r = amdgpu_jpeg_sysfs_reset_mask_init(adev);
+
+	return r;
+}
+
+/**
+ * jpeg_v5_0_0_sw_fini - sw fini for JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * JPEG suspend and free up sw allocation
+ */
+static int jpeg_v5_0_0_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = amdgpu_jpeg_suspend(adev);
+	if (r)
+		return r;
+
+	amdgpu_jpeg_sysfs_reset_mask_fini(adev);
+	r = amdgpu_jpeg_sw_fini(adev);
+
+	return r;
+}
+
+/**
+ * jpeg_v5_0_0_hw_init - start and test JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ */
+static int jpeg_v5_0_0_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring = adev->jpeg.inst->ring_dec;
+	int r;
+
+	adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
+			(adev->doorbell_index.vcn.vcn_ring0_1 << 1), 0);
+
+	/* Skip ring test because pause DPG is not implemented. */
+	if (adev->pg_flags & AMD_PG_SUPPORT_JPEG_DPG)
+		return 0;
+
+	r = amdgpu_ring_test_helper(ring);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+/**
+ * jpeg_v5_0_0_hw_fini - stop the hardware block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Stop the JPEG block, mark ring as not ready any more
+ */
+static int jpeg_v5_0_0_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	cancel_delayed_work_sync(&adev->jpeg.idle_work);
+
+	if (adev->jpeg.cur_state != AMD_PG_STATE_GATE &&
+	      RREG32_SOC15(JPEG, 0, regUVD_JRBC_STATUS))
+		jpeg_v5_0_0_set_powergating_state(ip_block, AMD_PG_STATE_GATE);
+
+	return 0;
+}
+
+/**
+ * jpeg_v5_0_0_suspend - suspend JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * HW fini and suspend JPEG block
+ */
+static int jpeg_v5_0_0_suspend(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	r = jpeg_v5_0_0_hw_fini(ip_block);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_suspend(ip_block->adev);
+
+	return r;
+}
+
+/**
+ * jpeg_v5_0_0_resume - resume JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Resume firmware and hw init JPEG block
+ */
+static int jpeg_v5_0_0_resume(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	r = amdgpu_jpeg_resume(ip_block->adev);
+	if (r)
+		return r;
+
+	r = jpeg_v5_0_0_hw_init(ip_block);
+
+	return r;
+}
+
+static void jpeg_v5_0_0_disable_clock_gating(struct amdgpu_device *adev)
+{
+	uint32_t data = 0;
+
+	WREG32_SOC15(JPEG, 0, regJPEG_CGC_GATE, data);
+
+	data = RREG32_SOC15(JPEG, 0, regJPEG_CGC_CTRL);
+	data &= ~(JPEG_CGC_CTRL__JPEG0_DEC_MODE_MASK
+		| JPEG_CGC_CTRL__JPEG_ENC_MODE_MASK);
+	WREG32_SOC15(JPEG, 0, regJPEG_CGC_CTRL, data);
+}
+
+static void jpeg_v5_0_0_enable_clock_gating(struct amdgpu_device *adev)
+{
+	uint32_t data = 0;
+
+	data = RREG32_SOC15(JPEG, 0, regJPEG_CGC_CTRL);
+
+	data |= 1 << JPEG_CGC_CTRL__JPEG0_DEC_MODE__SHIFT;
+	WREG32_SOC15(JPEG, 0, regJPEG_CGC_CTRL, data);
+
+	data = RREG32_SOC15(JPEG, 0, regJPEG_CGC_GATE);
+	data |= (JPEG_CGC_GATE__JPEG0_DEC_MASK
+		|JPEG_CGC_GATE__JPEG_ENC_MASK
+		|JPEG_CGC_GATE__JMCIF_MASK
+		|JPEG_CGC_GATE__JRBBM_MASK);
+	WREG32_SOC15(JPEG, 0, regJPEG_CGC_GATE, data);
+}
+
+static int jpeg_v5_0_0_disable_power_gating(struct amdgpu_device *adev)
+{
+	uint32_t data = 0;
+
+	data = 1 << UVD_IPX_DLDO_CONFIG__ONO1_PWR_CONFIG__SHIFT;
+	WREG32_SOC15(JPEG, 0, regUVD_IPX_DLDO_CONFIG, data);
+	SOC15_WAIT_ON_RREG(JPEG, 0, regUVD_IPX_DLDO_STATUS, 0,
+			UVD_IPX_DLDO_STATUS__ONO1_PWR_STATUS_MASK);
+
+	/* disable anti hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regUVD_JPEG_POWER_STATUS), 0,
+		~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+
+	return 0;
+}
+
+static int jpeg_v5_0_0_enable_power_gating(struct amdgpu_device *adev)
+{
+	/* enable anti hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regUVD_JPEG_POWER_STATUS),
+		UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK,
+		~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_JPEG) {
+		WREG32(SOC15_REG_OFFSET(JPEG, 0, regUVD_IPX_DLDO_CONFIG),
+			2 << UVD_IPX_DLDO_CONFIG__ONO1_PWR_CONFIG__SHIFT);
+		SOC15_WAIT_ON_RREG(JPEG, 0, regUVD_IPX_DLDO_STATUS,
+			1 << UVD_IPX_DLDO_STATUS__ONO1_PWR_STATUS__SHIFT,
+			UVD_IPX_DLDO_STATUS__ONO1_PWR_STATUS_MASK);
+	}
+
+	return 0;
+}
+
+static void jpeg_engine_5_0_0_dpg_clock_gating_mode(struct amdgpu_device *adev,
+	       int inst_idx, uint8_t indirect)
+{
+	uint32_t data = 0;
+
+	// JPEG disable CGC
+	if (adev->cg_flags & AMD_CG_SUPPORT_JPEG_MGCG)
+		data = 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		data = 0 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+
+	data |= 1 << JPEG_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << JPEG_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+
+	if (indirect) {
+		ADD_SOC24_JPEG_TO_DPG_SRAM(inst_idx, vcnipJPEG_CGC_CTRL, data, indirect);
+
+		// Turn on All JPEG clocks
+		data = 0;
+		ADD_SOC24_JPEG_TO_DPG_SRAM(inst_idx, vcnipJPEG_CGC_GATE, data, indirect);
+	} else {
+		WREG32_SOC24_JPEG_DPG_MODE(inst_idx, vcnipJPEG_CGC_CTRL, data, indirect);
+
+		// Turn on All JPEG clocks
+		data = 0;
+		WREG32_SOC24_JPEG_DPG_MODE(inst_idx, vcnipJPEG_CGC_GATE, data, indirect);
+	}
+}
+
+/**
+ * jpeg_v5_0_0_start_dpg_mode - Jpeg start with dpg mode
+ *
+ * @adev: amdgpu_device pointer
+ * @inst_idx: instance number index
+ * @indirect: indirectly write sram
+ *
+ * Start JPEG block with dpg mode
+ */
+static int jpeg_v5_0_0_start_dpg_mode(struct amdgpu_device *adev, int inst_idx, bool indirect)
+{
+	struct amdgpu_ring *ring = adev->jpeg.inst[inst_idx].ring_dec;
+	uint32_t reg_data = 0;
+
+	jpeg_v5_0_0_enable_power_gating(adev);
+
+	// enable dynamic power gating mode
+	reg_data = RREG32_SOC15(JPEG, inst_idx, regUVD_JPEG_POWER_STATUS);
+	reg_data |= UVD_JPEG_POWER_STATUS__JPEG_PG_MODE_MASK;
+	WREG32_SOC15(JPEG, inst_idx, regUVD_JPEG_POWER_STATUS, reg_data);
+
+	if (indirect)
+		adev->jpeg.inst[inst_idx].dpg_sram_curr_addr =
+			(uint32_t *)adev->jpeg.inst[inst_idx].dpg_sram_cpu_addr;
+
+	jpeg_engine_5_0_0_dpg_clock_gating_mode(adev, inst_idx, indirect);
+
+	/* MJPEG global tiling registers */
+	if (indirect)
+		ADD_SOC24_JPEG_TO_DPG_SRAM(inst_idx, vcnipJPEG_DEC_GFX10_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config, indirect);
+	else
+		WREG32_SOC24_JPEG_DPG_MODE(inst_idx, vcnipJPEG_DEC_GFX10_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config, 1);
+
+	/* enable System Interrupt for JRBC */
+	if (indirect)
+		ADD_SOC24_JPEG_TO_DPG_SRAM(inst_idx, vcnipJPEG_SYS_INT_EN,
+			JPEG_SYS_INT_EN__DJRBC0_MASK, indirect);
+	else
+		WREG32_SOC24_JPEG_DPG_MODE(inst_idx, vcnipJPEG_SYS_INT_EN,
+			JPEG_SYS_INT_EN__DJRBC0_MASK, 1);
+
+	if (indirect) {
+		/* add nop to workaround PSP size check */
+		ADD_SOC24_JPEG_TO_DPG_SRAM(inst_idx, vcnipUVD_NO_OP, 0, indirect);
+
+		amdgpu_jpeg_psp_update_sram(adev, inst_idx, 0);
+	}
+
+	WREG32_SOC15(VCN, 0, regVCN_JPEG_DB_CTRL,
+		ring->doorbell_index << VCN_JPEG_DB_CTRL__OFFSET__SHIFT |
+		VCN_JPEG_DB_CTRL__EN_MASK);
+
+	WREG32_SOC15(JPEG, inst_idx, regUVD_LMI_JRBC_RB_VMID, 0);
+	WREG32_SOC15(JPEG, inst_idx, regUVD_JRBC_RB_CNTL, (0x00000001L | 0x00000002L));
+	WREG32_SOC15(JPEG, inst_idx, regUVD_LMI_JRBC_RB_64BIT_BAR_LOW,
+		lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15(JPEG, inst_idx, regUVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
+		upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(JPEG, inst_idx, regUVD_JRBC_RB_RPTR, 0);
+	WREG32_SOC15(JPEG, inst_idx, regUVD_JRBC_RB_WPTR, 0);
+	WREG32_SOC15(JPEG, inst_idx, regUVD_JRBC_RB_CNTL, 0x00000002L);
+	WREG32_SOC15(JPEG, inst_idx, regUVD_JRBC_RB_SIZE, ring->ring_size / 4);
+	ring->wptr = RREG32_SOC15(JPEG, inst_idx, regUVD_JRBC_RB_WPTR);
+
+	return 0;
+}
+
+/**
+ * jpeg_v5_0_0_stop_dpg_mode - Jpeg stop with dpg mode
+ *
+ * @adev: amdgpu_device pointer
+ * @inst_idx: instance number index
+ *
+ * Stop JPEG block with dpg mode
+ */
+static void jpeg_v5_0_0_stop_dpg_mode(struct amdgpu_device *adev, int inst_idx)
+{
+	uint32_t reg_data = 0;
+
+	reg_data = RREG32_SOC15(JPEG, inst_idx, regUVD_JPEG_POWER_STATUS);
+	reg_data &= ~UVD_JPEG_POWER_STATUS__JPEG_PG_MODE_MASK;
+	WREG32_SOC15(JPEG, inst_idx, regUVD_JPEG_POWER_STATUS, reg_data);
+}
+
+/**
+ * jpeg_v5_0_0_start - start JPEG block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the JPEG block
+ */
+static int jpeg_v5_0_0_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = adev->jpeg.inst->ring_dec;
+	int r;
+
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_jpeg(adev, true);
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_JPEG_DPG) {
+		r = jpeg_v5_0_0_start_dpg_mode(adev, 0, adev->jpeg.indirect_sram);
+		return r;
+	}
+
+	/* disable power gating */
+	r = jpeg_v5_0_0_disable_power_gating(adev);
+	if (r)
+		return r;
+
+	/* JPEG disable CGC */
+	jpeg_v5_0_0_disable_clock_gating(adev);
+
+	/* MJPEG global tiling registers */
+	WREG32_SOC15(JPEG, 0, regJPEG_DEC_GFX10_ADDR_CONFIG,
+		adev->gfx.config.gb_addr_config);
+
+	/* enable JMI channel */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regUVD_JMI_CNTL), 0,
+		~UVD_JMI_CNTL__SOFT_RESET_MASK);
+
+	/* enable System Interrupt for JRBC */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regJPEG_SYS_INT_EN),
+		JPEG_SYS_INT_EN__DJRBC0_MASK,
+		~JPEG_SYS_INT_EN__DJRBC0_MASK);
+
+	WREG32_SOC15(VCN, 0, regVCN_JPEG_DB_CTRL,
+		ring->doorbell_index << VCN_JPEG_DB_CTRL__OFFSET__SHIFT |
+		VCN_JPEG_DB_CTRL__EN_MASK);
+
+	WREG32_SOC15(JPEG, 0, regUVD_LMI_JRBC_RB_VMID, 0);
+	WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_CNTL, (0x00000001L | 0x00000002L));
+	WREG32_SOC15(JPEG, 0, regUVD_LMI_JRBC_RB_64BIT_BAR_LOW,
+		lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15(JPEG, 0, regUVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
+		upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_RPTR, 0);
+	WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_WPTR, 0);
+	WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_CNTL, 0x00000002L);
+	WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_SIZE, ring->ring_size / 4);
+	ring->wptr = RREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_WPTR);
+
+	return 0;
+}
+
+/**
+ * jpeg_v5_0_0_stop - stop JPEG block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * stop the JPEG block
+ */
+static int jpeg_v5_0_0_stop(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_JPEG_DPG) {
+		jpeg_v5_0_0_stop_dpg_mode(adev, 0);
+	} else {
+
+		/* reset JMI */
+		WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regUVD_JMI_CNTL),
+			UVD_JMI_CNTL__SOFT_RESET_MASK,
+			~UVD_JMI_CNTL__SOFT_RESET_MASK);
+
+		jpeg_v5_0_0_enable_clock_gating(adev);
+
+		/* enable power gating */
+		r = jpeg_v5_0_0_enable_power_gating(adev);
+		if (r)
+			return r;
+	}
+
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_jpeg(adev, false);
+
+	return 0;
+}
+
+/**
+ * jpeg_v5_0_0_dec_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t jpeg_v5_0_0_dec_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_RPTR);
+}
+
+/**
+ * jpeg_v5_0_0_dec_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t jpeg_v5_0_0_dec_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell)
+		return *ring->wptr_cpu_addr;
+	else
+		return RREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_WPTR);
+}
+
+/**
+ * jpeg_v5_0_0_dec_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void jpeg_v5_0_0_dec_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
+		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+	} else {
+		WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_WPTR, lower_32_bits(ring->wptr));
+	}
+}
+
+static bool jpeg_v5_0_0_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int ret = 1;
+
+	ret &= (((RREG32_SOC15(JPEG, 0, regUVD_JRBC_STATUS) &
+		UVD_JRBC_STATUS__RB_JOB_DONE_MASK) ==
+		UVD_JRBC_STATUS__RB_JOB_DONE_MASK));
+
+	return ret;
+}
+
+static int jpeg_v5_0_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	return SOC15_WAIT_ON_RREG(JPEG, 0, regUVD_JRBC_STATUS,
+		UVD_JRBC_STATUS__RB_JOB_DONE_MASK,
+		UVD_JRBC_STATUS__RB_JOB_DONE_MASK);
+}
+
+static int jpeg_v5_0_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = state == AMD_CG_STATE_GATE;
+
+	if (enable) {
+		if (!jpeg_v5_0_0_is_idle(ip_block))
+			return -EBUSY;
+		jpeg_v5_0_0_enable_clock_gating(adev);
+	} else {
+		jpeg_v5_0_0_disable_clock_gating(adev);
+	}
+
+	return 0;
+}
+
+static int jpeg_v5_0_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int ret;
+
+	if (state == adev->jpeg.cur_state)
+		return 0;
+
+	if (state == AMD_PG_STATE_GATE)
+		ret = jpeg_v5_0_0_stop(adev);
+	else
+		ret = jpeg_v5_0_0_start(adev);
+
+	if (!ret)
+		adev->jpeg.cur_state = state;
+
+	return ret;
+}
+
+static int jpeg_v5_0_0_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned int type,
+					enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+static int jpeg_v5_0_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	DRM_DEBUG("IH: JPEG TRAP\n");
+
+	switch (entry->src_id) {
+	case VCN_5_0__SRCID__JPEG_DECODE:
+		amdgpu_fence_process(adev->jpeg.inst->ring_dec);
+		break;
+	default:
+		DRM_DEV_ERROR(adev->dev, "Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+static int jpeg_v5_0_0_ring_reset(struct amdgpu_ring *ring,
+				  unsigned int vmid,
+				  struct amdgpu_fence *timedout_fence)
+{
+	int r;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+	r = jpeg_v5_0_0_stop(ring->adev);
+	if (r)
+		return r;
+	r = jpeg_v5_0_0_start(ring->adev);
+	if (r)
+		return r;
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static const struct amd_ip_funcs jpeg_v5_0_0_ip_funcs = {
+	.name = "jpeg_v5_0_0",
+	.early_init = jpeg_v5_0_0_early_init,
+	.sw_init = jpeg_v5_0_0_sw_init,
+	.sw_fini = jpeg_v5_0_0_sw_fini,
+	.hw_init = jpeg_v5_0_0_hw_init,
+	.hw_fini = jpeg_v5_0_0_hw_fini,
+	.suspend = jpeg_v5_0_0_suspend,
+	.resume = jpeg_v5_0_0_resume,
+	.is_idle = jpeg_v5_0_0_is_idle,
+	.wait_for_idle = jpeg_v5_0_0_wait_for_idle,
+	.set_clockgating_state = jpeg_v5_0_0_set_clockgating_state,
+	.set_powergating_state = jpeg_v5_0_0_set_powergating_state,
+	.dump_ip_state = amdgpu_jpeg_dump_ip_state,
+	.print_ip_state = amdgpu_jpeg_print_ip_state,
+};
+
+static const struct amdgpu_ring_funcs jpeg_v5_0_0_dec_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_JPEG,
+	.align_mask = 0xf,
+	.get_rptr = jpeg_v5_0_0_dec_ring_get_rptr,
+	.get_wptr = jpeg_v5_0_0_dec_ring_get_wptr,
+	.set_wptr = jpeg_v5_0_0_dec_ring_set_wptr,
+	.parse_cs = amdgpu_jpeg_dec_parse_cs,
+	.emit_frame_size =
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
+		8 + /* jpeg_v5_0_0_dec_ring_emit_vm_flush */
+		22 + 22 + /* jpeg_v5_0_0_dec_ring_emit_fence x2 vm fence */
+		8 + 16,
+	.emit_ib_size = 22, /* jpeg_v5_0_0_dec_ring_emit_ib */
+	.emit_ib = jpeg_v4_0_3_dec_ring_emit_ib,
+	.emit_fence = jpeg_v4_0_3_dec_ring_emit_fence,
+	.emit_vm_flush = jpeg_v4_0_3_dec_ring_emit_vm_flush,
+	.test_ring = amdgpu_jpeg_dec_ring_test_ring,
+	.test_ib = amdgpu_jpeg_dec_ring_test_ib,
+	.insert_nop = jpeg_v4_0_3_dec_ring_nop,
+	.insert_start = jpeg_v4_0_3_dec_ring_insert_start,
+	.insert_end = jpeg_v4_0_3_dec_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_jpeg_ring_begin_use,
+	.end_use = amdgpu_jpeg_ring_end_use,
+	.emit_wreg = jpeg_v4_0_3_dec_ring_emit_wreg,
+	.emit_reg_wait = jpeg_v4_0_3_dec_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = jpeg_v5_0_0_ring_reset,
+};
+
+static void jpeg_v5_0_0_set_dec_ring_funcs(struct amdgpu_device *adev)
+{
+	adev->jpeg.inst->ring_dec->funcs = &jpeg_v5_0_0_dec_ring_vm_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs jpeg_v5_0_0_irq_funcs = {
+	.set = jpeg_v5_0_0_set_interrupt_state,
+	.process = jpeg_v5_0_0_process_interrupt,
+};
+
+static void jpeg_v5_0_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->jpeg.inst->irq.num_types = 1;
+	adev->jpeg.inst->irq.funcs = &jpeg_v5_0_0_irq_funcs;
+}
+
+const struct amdgpu_ip_block_version jpeg_v5_0_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_JPEG,
+	.major = 5,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &jpeg_v5_0_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.h b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.h
new file mode 100644
index 000000000000..5abb96159814
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.h
@@ -0,0 +1,35 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __JPEG_V5_0_0_H__
+#define __JPEG_V5_0_0_H__
+
+#define vcnipJPEG_CGC_GATE                                 0x4160
+#define vcnipJPEG_CGC_CTRL                                 0x4161
+#define vcnipJPEG_SYS_INT_EN                               0x4141
+#define vcnipUVD_NO_OP                                     0x0029
+#define vcnipJPEG_DEC_GFX10_ADDR_CONFIG                    0x404A
+
+extern const struct amdgpu_ip_block_version jpeg_v5_0_0_ip_block;
+
+#endif /* __JPEG_V5_0_0_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.c
new file mode 100644
index 000000000000..baf097d2e1ac
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.c
@@ -0,0 +1,1100 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
+/*
+ * Copyright 2014-2024 Advanced Micro Devices, Inc. All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include "amdgpu.h"
+#include "amdgpu_jpeg.h"
+#include "amdgpu_pm.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "jpeg_v4_0_3.h"
+#include "jpeg_v5_0_1.h"
+#include "mmsch_v5_0.h"
+
+#include "vcn/vcn_5_0_0_offset.h"
+#include "vcn/vcn_5_0_0_sh_mask.h"
+#include "ivsrcid/vcn/irqsrcs_vcn_5_0.h"
+
+static int jpeg_v5_0_1_start_sriov(struct amdgpu_device *adev);
+static void jpeg_v5_0_1_set_dec_ring_funcs(struct amdgpu_device *adev);
+static void jpeg_v5_0_1_set_irq_funcs(struct amdgpu_device *adev);
+static int jpeg_v5_0_1_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					     enum amd_powergating_state state);
+static void jpeg_v5_0_1_set_ras_funcs(struct amdgpu_device *adev);
+static void jpeg_v5_0_1_dec_ring_set_wptr(struct amdgpu_ring *ring);
+
+static int amdgpu_ih_srcid_jpeg[] = {
+	VCN_5_0__SRCID__JPEG_DECODE,
+	VCN_5_0__SRCID__JPEG1_DECODE,
+	VCN_5_0__SRCID__JPEG2_DECODE,
+	VCN_5_0__SRCID__JPEG3_DECODE,
+	VCN_5_0__SRCID__JPEG4_DECODE,
+	VCN_5_0__SRCID__JPEG5_DECODE,
+	VCN_5_0__SRCID__JPEG6_DECODE,
+	VCN_5_0__SRCID__JPEG7_DECODE,
+	VCN_5_0__SRCID__JPEG8_DECODE,
+	VCN_5_0__SRCID__JPEG9_DECODE,
+};
+
+static const struct amdgpu_hwip_reg_entry jpeg_reg_list_5_0_1[] = {
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_INT_STAT),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC0_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC0_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC0_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_ADDR_MODE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_GFX10_ADDR_CONFIG),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_Y_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_UV_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_PITCH),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_UV_PITCH),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC1_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC1_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC1_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC2_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC2_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC2_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC3_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC3_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC3_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC4_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC4_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC4_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC5_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC5_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC5_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC6_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC6_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC6_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC7_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC7_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC7_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC8_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC8_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC8_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC9_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC9_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC9_UVD_JRBC_STATUS),
+};
+
+static int jpeg_v5_0_1_core_reg_offset(u32 pipe)
+{
+	if (pipe <= AMDGPU_MAX_JPEG_RINGS_4_0_3)
+		return ((0x40 * pipe) - 0xc80);
+	else
+		return ((0x40 * pipe) - 0x440);
+}
+
+/**
+ * jpeg_v5_0_1_early_init - set function pointers
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Set ring and irq function pointers
+ */
+static int jpeg_v5_0_1_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (!adev->jpeg.num_jpeg_inst || adev->jpeg.num_jpeg_inst > AMDGPU_MAX_JPEG_INSTANCES)
+		return -ENOENT;
+
+	adev->jpeg.num_jpeg_rings = AMDGPU_MAX_JPEG_RINGS;
+	jpeg_v5_0_1_set_dec_ring_funcs(adev);
+	jpeg_v5_0_1_set_irq_funcs(adev);
+	jpeg_v5_0_1_set_ras_funcs(adev);
+
+	return 0;
+}
+
+/**
+ * jpeg_v5_0_1_sw_init - sw init for JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Load firmware and sw initialization
+ */
+static int jpeg_v5_0_1_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	int i, j, r, jpeg_inst;
+
+	for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
+		/* JPEG TRAP */
+		r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+				      amdgpu_ih_srcid_jpeg[j], &adev->jpeg.inst->irq);
+		if (r)
+			return r;
+	}
+	/* JPEG DJPEG POISON EVENT */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+			VCN_5_0__SRCID_DJPEG0_POISON, &adev->jpeg.inst->ras_poison_irq);
+	if (r)
+		return r;
+
+	/* JPEG EJPEG POISON EVENT */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+			VCN_5_0__SRCID_EJPEG0_POISON, &adev->jpeg.inst->ras_poison_irq);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_sw_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_resume(adev);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		jpeg_inst = GET_INST(JPEG, i);
+
+		for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
+			ring = &adev->jpeg.inst[i].ring_dec[j];
+			ring->use_doorbell = true;
+			ring->vm_hub = AMDGPU_MMHUB0(adev->jpeg.inst[i].aid_id);
+			if (!amdgpu_sriov_vf(adev)) {
+				ring->doorbell_index =
+					(adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
+					1 + j + 11 * jpeg_inst;
+			} else {
+				ring->doorbell_index =
+					(adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
+					2 + j + 32 * jpeg_inst;
+			}
+			sprintf(ring->name, "jpeg_dec_%d.%d", adev->jpeg.inst[i].aid_id, j);
+			r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst->irq, 0,
+					     AMDGPU_RING_PRIO_DEFAULT, NULL);
+			if (r)
+				return r;
+
+			adev->jpeg.internal.jpeg_pitch[j] =
+				regUVD_JRBC0_UVD_JRBC_SCRATCH0_INTERNAL_OFFSET;
+			adev->jpeg.inst[i].external.jpeg_pitch[j] =
+				SOC15_REG_OFFSET1(JPEG, jpeg_inst, regUVD_JRBC_SCRATCH0,
+						  (j ? jpeg_v5_0_1_core_reg_offset(j) : 0));
+		}
+	}
+
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__JPEG)) {
+		r = amdgpu_jpeg_ras_sw_init(adev);
+		if (r) {
+			dev_err(adev->dev, "Failed to initialize jpeg ras block!\n");
+			return r;
+		}
+	}
+
+	r = amdgpu_jpeg_reg_dump_init(adev, jpeg_reg_list_5_0_1, ARRAY_SIZE(jpeg_reg_list_5_0_1));
+	if (r)
+		return r;
+
+	adev->jpeg.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->jpeg.inst[0].ring_dec[0]);
+	if (!amdgpu_sriov_vf(adev))
+		adev->jpeg.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+	r = amdgpu_jpeg_sysfs_reset_mask_init(adev);
+
+	return r;
+}
+
+/**
+ * jpeg_v5_0_1_sw_fini - sw fini for JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * JPEG suspend and free up sw allocation
+ */
+static int jpeg_v5_0_1_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = amdgpu_jpeg_suspend(adev);
+	if (r)
+		return r;
+
+	amdgpu_jpeg_sysfs_reset_mask_fini(adev);
+
+	r = amdgpu_jpeg_sw_fini(adev);
+
+	return r;
+}
+
+/**
+ * jpeg_v5_0_1_hw_init - start and test JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ */
+static int jpeg_v5_0_1_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	int i, j, r, jpeg_inst;
+
+	if (amdgpu_sriov_vf(adev)) {
+		r = jpeg_v5_0_1_start_sriov(adev);
+		if (r)
+			return r;
+
+		for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+			for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
+				ring = &adev->jpeg.inst[i].ring_dec[j];
+				ring->wptr = 0;
+				ring->wptr_old = 0;
+				jpeg_v5_0_1_dec_ring_set_wptr(ring);
+				ring->sched.ready = true;
+			}
+		}
+		return 0;
+	}
+	if (RREG32_SOC15(VCN, GET_INST(VCN, 0), regVCN_RRMT_CNTL) & 0x100)
+		adev->jpeg.caps |= AMDGPU_JPEG_CAPS(RRMT_ENABLED);
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		jpeg_inst = GET_INST(JPEG, i);
+		ring = adev->jpeg.inst[i].ring_dec;
+		if (ring->use_doorbell)
+			adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
+				 (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 11 * jpeg_inst,
+				 adev->jpeg.inst[i].aid_id);
+
+		for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
+			ring = &adev->jpeg.inst[i].ring_dec[j];
+			if (ring->use_doorbell)
+				WREG32_SOC15_OFFSET(VCN, GET_INST(VCN, i), regVCN_JPEG_DB_CTRL,
+						    ring->pipe,
+						    ring->doorbell_index <<
+						    VCN_JPEG_DB_CTRL__OFFSET__SHIFT |
+						    VCN_JPEG_DB_CTRL__EN_MASK);
+			r = amdgpu_ring_test_helper(ring);
+			if (r)
+				return r;
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * jpeg_v5_0_1_hw_fini - stop the hardware block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Stop the JPEG block, mark ring as not ready any more
+ */
+static int jpeg_v5_0_1_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int ret = 0;
+
+	cancel_delayed_work_sync(&adev->jpeg.idle_work);
+
+	if (!amdgpu_sriov_vf(adev)) {
+		if (adev->jpeg.cur_state != AMD_PG_STATE_GATE)
+			ret = jpeg_v5_0_1_set_powergating_state(ip_block, AMD_PG_STATE_GATE);
+	}
+
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__JPEG) && !amdgpu_sriov_vf(adev))
+		amdgpu_irq_put(adev, &adev->jpeg.inst->ras_poison_irq, 0);
+
+	return ret;
+}
+
+/**
+ * jpeg_v5_0_1_suspend - suspend JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * HW fini and suspend JPEG block
+ */
+static int jpeg_v5_0_1_suspend(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = jpeg_v5_0_1_hw_fini(ip_block);
+	if (r)
+		return r;
+
+	r = amdgpu_jpeg_suspend(adev);
+
+	return r;
+}
+
+/**
+ * jpeg_v5_0_1_resume - resume JPEG block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Resume firmware and hw init JPEG block
+ */
+static int jpeg_v5_0_1_resume(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = amdgpu_jpeg_resume(adev);
+	if (r)
+		return r;
+
+	r = jpeg_v5_0_1_hw_init(ip_block);
+
+	return r;
+}
+
+static void jpeg_v5_0_1_init_inst(struct amdgpu_device *adev, int i)
+{
+	int jpeg_inst = GET_INST(JPEG, i);
+
+	/* disable anti hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JPEG_POWER_STATUS), 0,
+		 ~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+
+	/* keep the JPEG in static PG mode */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JPEG_POWER_STATUS), 0,
+		 ~UVD_JPEG_POWER_STATUS__JPEG_PG_MODE_MASK);
+
+	/* MJPEG global tiling registers */
+	WREG32_SOC15(JPEG, 0, regJPEG_DEC_GFX10_ADDR_CONFIG,
+		     adev->gfx.config.gb_addr_config);
+
+	/* enable JMI channel */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JMI_CNTL), 0,
+		 ~UVD_JMI_CNTL__SOFT_RESET_MASK);
+}
+
+static void jpeg_v5_0_1_deinit_inst(struct amdgpu_device *adev, int i)
+{
+	int jpeg_inst = GET_INST(JPEG, i);
+	/* reset JMI */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JMI_CNTL),
+		 UVD_JMI_CNTL__SOFT_RESET_MASK,
+		 ~UVD_JMI_CNTL__SOFT_RESET_MASK);
+
+	/* enable anti hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JPEG_POWER_STATUS),
+		 UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK,
+		 ~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+}
+
+static void jpeg_v5_0_1_init_jrbc(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 reg, data, mask;
+	int jpeg_inst = GET_INST(JPEG, ring->me);
+	int reg_offset = ring->pipe ? jpeg_v5_0_1_core_reg_offset(ring->pipe) : 0;
+
+	/* enable System Interrupt for JRBC */
+	reg = SOC15_REG_OFFSET(JPEG, jpeg_inst, regJPEG_SYS_INT_EN);
+	if (ring->pipe < AMDGPU_MAX_JPEG_RINGS_4_0_3) {
+		data = JPEG_SYS_INT_EN__DJRBC0_MASK << ring->pipe;
+		mask = ~(JPEG_SYS_INT_EN__DJRBC0_MASK << ring->pipe);
+		WREG32_P(reg, data, mask);
+	} else {
+		data = JPEG_SYS_INT_EN__DJRBC0_MASK << (ring->pipe+12);
+		mask = ~(JPEG_SYS_INT_EN__DJRBC0_MASK << (ring->pipe+12));
+		WREG32_P(reg, data, mask);
+	}
+
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_LMI_JRBC_RB_VMID,
+			    reg_offset, 0);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JRBC_RB_CNTL,
+			    reg_offset,
+			    (0x00000001L | 0x00000002L));
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_LMI_JRBC_RB_64BIT_BAR_LOW,
+			    reg_offset, lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
+			    reg_offset, upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JRBC_RB_RPTR,
+			    reg_offset, 0);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JRBC_RB_WPTR,
+			    reg_offset, 0);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JRBC_RB_CNTL,
+			    reg_offset, 0x00000002L);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JRBC_RB_SIZE,
+			    reg_offset, ring->ring_size / 4);
+	ring->wptr = RREG32_SOC15_OFFSET(JPEG, jpeg_inst, regUVD_JRBC_RB_WPTR,
+					 reg_offset);
+}
+
+static int jpeg_v5_0_1_start_sriov(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	uint64_t ctx_addr;
+	uint32_t param, resp, expected;
+	uint32_t tmp, timeout;
+
+	struct amdgpu_mm_table *table = &adev->virt.mm_table;
+	uint32_t *table_loc;
+	uint32_t table_size;
+	uint32_t size, size_dw, item_offset;
+	uint32_t init_status;
+	int i, j, jpeg_inst;
+
+	struct mmsch_v5_0_cmd_direct_write
+		direct_wt = { {0} };
+	struct mmsch_v5_0_cmd_end end = { {0} };
+	struct mmsch_v5_0_init_header header;
+
+	direct_wt.cmd_header.command_type =
+		MMSCH_COMMAND__DIRECT_REG_WRITE;
+	end.cmd_header.command_type =
+		MMSCH_COMMAND__END;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; i++) {
+		jpeg_inst = GET_INST(JPEG, i);
+
+		memset(&header, 0, sizeof(struct mmsch_v5_0_init_header));
+		header.version = MMSCH_VERSION;
+		header.total_size = sizeof(struct mmsch_v5_0_init_header) >> 2;
+
+		table_loc = (uint32_t *)table->cpu_addr;
+		table_loc += header.total_size;
+
+		item_offset = header.total_size;
+
+		for (j = 0; j < adev->jpeg.num_jpeg_rings; j++) {
+			ring = &adev->jpeg.inst[i].ring_dec[j];
+			table_size = 0;
+
+			tmp = SOC15_REG_OFFSET(JPEG, 0, regUVD_LMI_JRBC_RB_64BIT_BAR_LOW);
+			MMSCH_V5_0_INSERT_DIRECT_WT(tmp, lower_32_bits(ring->gpu_addr));
+			tmp = SOC15_REG_OFFSET(JPEG, 0, regUVD_LMI_JRBC_RB_64BIT_BAR_HIGH);
+			MMSCH_V5_0_INSERT_DIRECT_WT(tmp, upper_32_bits(ring->gpu_addr));
+			tmp = SOC15_REG_OFFSET(JPEG, 0, regUVD_JRBC_RB_SIZE);
+			MMSCH_V5_0_INSERT_DIRECT_WT(tmp, ring->ring_size / 4);
+
+			if (j < 5) {
+				header.mjpegdec0[j].table_offset = item_offset;
+				header.mjpegdec0[j].init_status = 0;
+				header.mjpegdec0[j].table_size = table_size;
+			} else {
+				header.mjpegdec1[j - 5].table_offset = item_offset;
+				header.mjpegdec1[j - 5].init_status = 0;
+				header.mjpegdec1[j - 5].table_size = table_size;
+			}
+			header.total_size += table_size;
+			item_offset += table_size;
+		}
+
+		MMSCH_V5_0_INSERT_END();
+
+		/* send init table to MMSCH */
+		size = sizeof(struct mmsch_v5_0_init_header);
+		table_loc = (uint32_t *)table->cpu_addr;
+		memcpy((void *)table_loc, &header, size);
+
+		ctx_addr = table->gpu_addr;
+		WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_CTX_ADDR_LO, lower_32_bits(ctx_addr));
+		WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_CTX_ADDR_HI, upper_32_bits(ctx_addr));
+
+		tmp = RREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_VMID);
+		tmp &= ~MMSCH_VF_VMID__VF_CTX_VMID_MASK;
+		tmp |= (0 << MMSCH_VF_VMID__VF_CTX_VMID__SHIFT);
+		WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_VMID, tmp);
+
+		size = header.total_size;
+		WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_CTX_SIZE, size);
+
+		WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_MAILBOX_RESP, 0);
+
+		param = 0x00000001;
+		WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_MAILBOX_HOST, param);
+		tmp = 0;
+		timeout = 1000;
+		resp = 0;
+		expected = MMSCH_VF_MAILBOX_RESP__OK;
+		init_status =
+			((struct mmsch_v5_0_init_header *)(table_loc))->mjpegdec0[i].init_status;
+		while (resp != expected) {
+			resp = RREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_MAILBOX_RESP);
+
+			if (resp != 0)
+				break;
+			udelay(10);
+			tmp = tmp + 10;
+			if (tmp >= timeout) {
+				DRM_ERROR("failed to init MMSCH. TIME-OUT after %d usec"\
+					" waiting for regMMSCH_VF_MAILBOX_RESP "\
+					"(expected=0x%08x, readback=0x%08x)\n",
+					tmp, expected, resp);
+				return -EBUSY;
+			}
+		}
+		if (resp != expected && resp != MMSCH_VF_MAILBOX_RESP__INCOMPLETE &&
+				init_status != MMSCH_VF_ENGINE_STATUS__PASS)
+			DRM_ERROR("MMSCH init status is incorrect! readback=0x%08x, header init status for jpeg: %x\n",
+					resp, init_status);
+
+	}
+	return 0;
+}
+
+/**
+ * jpeg_v5_0_1_start - start JPEG block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the JPEG block
+ */
+static int jpeg_v5_0_1_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	int i, j;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		jpeg_v5_0_1_init_inst(adev, i);
+		for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
+			ring = &adev->jpeg.inst[i].ring_dec[j];
+			jpeg_v5_0_1_init_jrbc(ring);
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * jpeg_v5_0_1_stop - stop JPEG block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * stop the JPEG block
+ */
+static int jpeg_v5_0_1_stop(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i)
+		jpeg_v5_0_1_deinit_inst(adev, i);
+
+	return 0;
+}
+
+/**
+ * jpeg_v5_0_1_dec_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t jpeg_v5_0_1_dec_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32_SOC15_OFFSET(JPEG, GET_INST(JPEG, ring->me), regUVD_JRBC_RB_RPTR,
+				   ring->pipe ? jpeg_v5_0_1_core_reg_offset(ring->pipe) : 0);
+}
+
+/**
+ * jpeg_v5_0_1_dec_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t jpeg_v5_0_1_dec_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell)
+		return adev->wb.wb[ring->wptr_offs];
+
+	return RREG32_SOC15_OFFSET(JPEG, GET_INST(JPEG, ring->me), regUVD_JRBC_RB_WPTR,
+				   ring->pipe ? jpeg_v5_0_1_core_reg_offset(ring->pipe) : 0);
+}
+
+/**
+ * jpeg_v5_0_1_dec_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void jpeg_v5_0_1_dec_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		adev->wb.wb[ring->wptr_offs] = lower_32_bits(ring->wptr);
+		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+	} else {
+		WREG32_SOC15_OFFSET(JPEG, GET_INST(JPEG, ring->me),
+				    regUVD_JRBC_RB_WPTR,
+				    (ring->pipe ? jpeg_v5_0_1_core_reg_offset(ring->pipe) : 0),
+				    lower_32_bits(ring->wptr));
+	}
+}
+
+static bool jpeg_v5_0_1_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool ret = false;
+	int i, j;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
+			int reg_offset = (j ? jpeg_v5_0_1_core_reg_offset(j) : 0);
+
+			ret &= ((RREG32_SOC15_OFFSET(JPEG, GET_INST(JPEG, i),
+				regUVD_JRBC_STATUS, reg_offset) &
+				UVD_JRBC_STATUS__RB_JOB_DONE_MASK) ==
+				UVD_JRBC_STATUS__RB_JOB_DONE_MASK);
+		}
+	}
+
+	return ret;
+}
+
+static int jpeg_v5_0_1_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int ret = 0;
+	int i, j;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
+			int reg_offset = (j ? jpeg_v5_0_1_core_reg_offset(j) : 0);
+
+			ret &= SOC15_WAIT_ON_RREG_OFFSET(JPEG, GET_INST(JPEG, i),
+							 regUVD_JRBC_STATUS, reg_offset,
+							 UVD_JRBC_STATUS__RB_JOB_DONE_MASK,
+							 UVD_JRBC_STATUS__RB_JOB_DONE_MASK);
+		}
+	}
+	return ret;
+}
+
+static int jpeg_v5_0_1_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					     enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = state == AMD_CG_STATE_GATE;
+
+	int i;
+
+	if (!enable)
+		return 0;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (!jpeg_v5_0_1_is_idle(ip_block))
+			return -EBUSY;
+	}
+
+	return 0;
+}
+
+static int jpeg_v5_0_1_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					     enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int ret;
+
+	if (amdgpu_sriov_vf(adev)) {
+		adev->jpeg.cur_state = AMD_PG_STATE_UNGATE;
+		return 0;
+	}
+
+	if (state == adev->jpeg.cur_state)
+		return 0;
+
+	if (state == AMD_PG_STATE_GATE)
+		ret = jpeg_v5_0_1_stop(adev);
+	else
+		ret = jpeg_v5_0_1_start(adev);
+
+	if (!ret)
+		adev->jpeg.cur_state = state;
+
+	return ret;
+}
+
+static int jpeg_v5_0_1_set_interrupt_state(struct amdgpu_device *adev,
+					   struct amdgpu_irq_src *source,
+					   unsigned int type,
+					   enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+static int jpeg_v5_0_1_set_ras_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned int type,
+					enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+
+
+static int jpeg_v5_0_1_process_interrupt(struct amdgpu_device *adev,
+					 struct amdgpu_irq_src *source,
+					 struct amdgpu_iv_entry *entry)
+{
+	u32 i, inst;
+
+	i = node_id_to_phys_map[entry->node_id];
+	DRM_DEV_DEBUG(adev->dev, "IH: JPEG TRAP\n");
+
+	for (inst = 0; inst < adev->jpeg.num_jpeg_inst; ++inst)
+		if (adev->jpeg.inst[inst].aid_id == i)
+			break;
+
+	if (inst >= adev->jpeg.num_jpeg_inst) {
+		dev_WARN_ONCE(adev->dev, 1,
+			      "Interrupt received for unknown JPEG instance %d",
+			      entry->node_id);
+		return 0;
+	}
+
+	switch (entry->src_id) {
+	case VCN_5_0__SRCID__JPEG_DECODE:
+		amdgpu_fence_process(&adev->jpeg.inst[inst].ring_dec[0]);
+		break;
+	case VCN_5_0__SRCID__JPEG1_DECODE:
+		amdgpu_fence_process(&adev->jpeg.inst[inst].ring_dec[1]);
+		break;
+	case VCN_5_0__SRCID__JPEG2_DECODE:
+		amdgpu_fence_process(&adev->jpeg.inst[inst].ring_dec[2]);
+		break;
+	case VCN_5_0__SRCID__JPEG3_DECODE:
+		amdgpu_fence_process(&adev->jpeg.inst[inst].ring_dec[3]);
+		break;
+	case VCN_5_0__SRCID__JPEG4_DECODE:
+		amdgpu_fence_process(&adev->jpeg.inst[inst].ring_dec[4]);
+		break;
+	case VCN_5_0__SRCID__JPEG5_DECODE:
+		amdgpu_fence_process(&adev->jpeg.inst[inst].ring_dec[5]);
+		break;
+	case VCN_5_0__SRCID__JPEG6_DECODE:
+		amdgpu_fence_process(&adev->jpeg.inst[inst].ring_dec[6]);
+		break;
+	case VCN_5_0__SRCID__JPEG7_DECODE:
+		amdgpu_fence_process(&adev->jpeg.inst[inst].ring_dec[7]);
+		break;
+	case VCN_5_0__SRCID__JPEG8_DECODE:
+		amdgpu_fence_process(&adev->jpeg.inst[inst].ring_dec[8]);
+		break;
+	case VCN_5_0__SRCID__JPEG9_DECODE:
+		amdgpu_fence_process(&adev->jpeg.inst[inst].ring_dec[9]);
+		break;
+	default:
+		DRM_DEV_ERROR(adev->dev, "Unhandled interrupt: %d %d\n",
+			      entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+static void jpeg_v5_0_1_core_stall_reset(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int jpeg_inst = GET_INST(JPEG, ring->me);
+	int reg_offset = ring->pipe ? jpeg_v5_0_1_core_reg_offset(ring->pipe) : 0;
+
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JMI0_UVD_JMI_CLIENT_STALL,
+			    reg_offset, 0x1F);
+	SOC15_WAIT_ON_RREG_OFFSET(JPEG, jpeg_inst,
+				  regUVD_JMI0_UVD_JMI_CLIENT_CLEAN_STATUS,
+				  reg_offset, 0x1F, 0x1F);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JMI0_JPEG_LMI_DROP,
+			    reg_offset, 0x1F);
+	WREG32_SOC15(JPEG, jpeg_inst, regJPEG_CORE_RST_CTRL, 1 << ring->pipe);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JMI0_UVD_JMI_CLIENT_STALL,
+			    reg_offset, 0x00);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JMI0_JPEG_LMI_DROP,
+			    reg_offset, 0x00);
+	WREG32_SOC15(JPEG, jpeg_inst, regJPEG_CORE_RST_CTRL, 0x00);
+}
+
+static int jpeg_v5_0_1_ring_reset(struct amdgpu_ring *ring,
+				  unsigned int vmid,
+				  struct amdgpu_fence *timedout_fence)
+{
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+	jpeg_v5_0_1_core_stall_reset(ring);
+	jpeg_v5_0_1_init_jrbc(ring);
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static const struct amd_ip_funcs jpeg_v5_0_1_ip_funcs = {
+	.name = "jpeg_v5_0_1",
+	.early_init = jpeg_v5_0_1_early_init,
+	.late_init = NULL,
+	.sw_init = jpeg_v5_0_1_sw_init,
+	.sw_fini = jpeg_v5_0_1_sw_fini,
+	.hw_init = jpeg_v5_0_1_hw_init,
+	.hw_fini = jpeg_v5_0_1_hw_fini,
+	.suspend = jpeg_v5_0_1_suspend,
+	.resume = jpeg_v5_0_1_resume,
+	.is_idle = jpeg_v5_0_1_is_idle,
+	.wait_for_idle = jpeg_v5_0_1_wait_for_idle,
+	.check_soft_reset = NULL,
+	.pre_soft_reset = NULL,
+	.soft_reset = NULL,
+	.post_soft_reset = NULL,
+	.set_clockgating_state = jpeg_v5_0_1_set_clockgating_state,
+	.set_powergating_state = jpeg_v5_0_1_set_powergating_state,
+	.dump_ip_state = amdgpu_jpeg_dump_ip_state,
+	.print_ip_state = amdgpu_jpeg_print_ip_state,
+};
+
+static const struct amdgpu_ring_funcs jpeg_v5_0_1_dec_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_JPEG,
+	.align_mask = 0xf,
+	.get_rptr = jpeg_v5_0_1_dec_ring_get_rptr,
+	.get_wptr = jpeg_v5_0_1_dec_ring_get_wptr,
+	.set_wptr = jpeg_v5_0_1_dec_ring_set_wptr,
+	.emit_frame_size =
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
+		8 + /* jpeg_v5_0_1_dec_ring_emit_vm_flush */
+		22 + 22 + /* jpeg_v5_0_1_dec_ring_emit_fence x2 vm fence */
+		8 + 16,
+	.emit_ib_size = 22, /* jpeg_v5_0_1_dec_ring_emit_ib */
+	.emit_ib = jpeg_v4_0_3_dec_ring_emit_ib,
+	.emit_fence = jpeg_v4_0_3_dec_ring_emit_fence,
+	.emit_vm_flush = jpeg_v4_0_3_dec_ring_emit_vm_flush,
+	.emit_hdp_flush = jpeg_v4_0_3_ring_emit_hdp_flush,
+	.test_ring = amdgpu_jpeg_dec_ring_test_ring,
+	.test_ib = amdgpu_jpeg_dec_ring_test_ib,
+	.insert_nop = jpeg_v4_0_3_dec_ring_nop,
+	.insert_start = jpeg_v4_0_3_dec_ring_insert_start,
+	.insert_end = jpeg_v4_0_3_dec_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_jpeg_ring_begin_use,
+	.end_use = amdgpu_jpeg_ring_end_use,
+	.emit_wreg = jpeg_v4_0_3_dec_ring_emit_wreg,
+	.emit_reg_wait = jpeg_v4_0_3_dec_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = jpeg_v5_0_1_ring_reset,
+};
+
+static void jpeg_v5_0_1_set_dec_ring_funcs(struct amdgpu_device *adev)
+{
+	int i, j, jpeg_inst;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
+			adev->jpeg.inst[i].ring_dec[j].funcs = &jpeg_v5_0_1_dec_ring_vm_funcs;
+			adev->jpeg.inst[i].ring_dec[j].me = i;
+			adev->jpeg.inst[i].ring_dec[j].pipe = j;
+		}
+		jpeg_inst = GET_INST(JPEG, i);
+		adev->jpeg.inst[i].aid_id =
+			jpeg_inst / adev->jpeg.num_inst_per_aid;
+	}
+}
+
+static const struct amdgpu_irq_src_funcs jpeg_v5_0_1_irq_funcs = {
+	.set = jpeg_v5_0_1_set_interrupt_state,
+	.process = jpeg_v5_0_1_process_interrupt,
+};
+
+static const struct amdgpu_irq_src_funcs jpeg_v5_0_1_ras_irq_funcs = {
+	.set = jpeg_v5_0_1_set_ras_interrupt_state,
+	.process = amdgpu_jpeg_process_poison_irq,
+};
+
+static void jpeg_v5_0_1_set_irq_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i)
+		adev->jpeg.inst->irq.num_types += adev->jpeg.num_jpeg_rings;
+
+	adev->jpeg.inst->irq.funcs = &jpeg_v5_0_1_irq_funcs;
+
+	adev->jpeg.inst->ras_poison_irq.num_types = 1;
+	adev->jpeg.inst->ras_poison_irq.funcs = &jpeg_v5_0_1_ras_irq_funcs;
+
+}
+
+const struct amdgpu_ip_block_version jpeg_v5_0_1_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_JPEG,
+	.major = 5,
+	.minor = 0,
+	.rev = 1,
+	.funcs = &jpeg_v5_0_1_ip_funcs,
+};
+
+static uint32_t jpeg_v5_0_1_query_poison_by_instance(struct amdgpu_device *adev,
+		uint32_t instance, uint32_t sub_block)
+{
+	uint32_t poison_stat = 0, reg_value = 0;
+
+	switch (sub_block) {
+	case AMDGPU_JPEG_V5_0_1_JPEG0:
+		reg_value = RREG32_SOC15(JPEG, instance, regUVD_RAS_JPEG0_STATUS);
+		poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_JPEG0_STATUS, POISONED_PF);
+		break;
+	case AMDGPU_JPEG_V5_0_1_JPEG1:
+		reg_value = RREG32_SOC15(JPEG, instance, regUVD_RAS_JPEG1_STATUS);
+		poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_JPEG1_STATUS, POISONED_PF);
+		break;
+	default:
+		break;
+	}
+
+	if (poison_stat)
+		dev_info(adev->dev, "Poison detected in JPEG%d sub_block%d\n",
+			instance, sub_block);
+
+	return poison_stat;
+}
+
+static bool jpeg_v5_0_1_query_ras_poison_status(struct amdgpu_device *adev)
+{
+	uint32_t inst = 0, sub = 0, poison_stat = 0;
+
+	for (inst = 0; inst < adev->jpeg.num_jpeg_inst; inst++)
+		for (sub = 0; sub < AMDGPU_JPEG_V5_0_1_MAX_SUB_BLOCK; sub++)
+			poison_stat +=
+			jpeg_v5_0_1_query_poison_by_instance(adev, inst, sub);
+
+	return !!poison_stat;
+}
+
+static const struct amdgpu_ras_block_hw_ops jpeg_v5_0_1_ras_hw_ops = {
+	.query_poison_status = jpeg_v5_0_1_query_ras_poison_status,
+};
+
+static int jpeg_v5_0_1_aca_bank_parser(struct aca_handle *handle, struct aca_bank *bank,
+				      enum aca_smu_type type, void *data)
+{
+	struct aca_bank_info info;
+	u64 misc0;
+	int ret;
+
+	ret = aca_bank_info_decode(bank, &info);
+	if (ret)
+		return ret;
+
+	misc0 = bank->regs[ACA_REG_IDX_MISC0];
+	switch (type) {
+	case ACA_SMU_TYPE_UE:
+		bank->aca_err_type = ACA_ERROR_TYPE_UE;
+		ret = aca_error_cache_log_bank_error(handle, &info, ACA_ERROR_TYPE_UE,
+						     1ULL);
+		break;
+	case ACA_SMU_TYPE_CE:
+		bank->aca_err_type = ACA_ERROR_TYPE_CE;
+		ret = aca_error_cache_log_bank_error(handle, &info, bank->aca_err_type,
+						     ACA_REG__MISC0__ERRCNT(misc0));
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return ret;
+}
+
+/* reference to smu driver if header file */
+static int jpeg_v5_0_1_err_codes[] = {
+	16, 17, 18, 19, 20, 21, 22, 23, /* JPEG[0-9][S|D] */
+	24, 25, 26, 27, 28, 29, 30, 31,
+	48, 49, 50, 51,
+};
+
+static bool jpeg_v5_0_1_aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank,
+					 enum aca_smu_type type, void *data)
+{
+	u32 instlo;
+
+	instlo = ACA_REG__IPID__INSTANCEIDLO(bank->regs[ACA_REG_IDX_IPID]);
+	instlo &= GENMASK(31, 1);
+
+	if (instlo != mmSMNAID_AID0_MCA_SMU)
+		return false;
+
+	if (aca_bank_check_error_codes(handle->adev, bank,
+				       jpeg_v5_0_1_err_codes,
+				       ARRAY_SIZE(jpeg_v5_0_1_err_codes)))
+		return false;
+
+	return true;
+}
+
+static const struct aca_bank_ops jpeg_v5_0_1_aca_bank_ops = {
+	.aca_bank_parser = jpeg_v5_0_1_aca_bank_parser,
+	.aca_bank_is_valid = jpeg_v5_0_1_aca_bank_is_valid,
+};
+
+static const struct aca_info jpeg_v5_0_1_aca_info = {
+	.hwip = ACA_HWIP_TYPE_SMU,
+	.mask = ACA_ERROR_UE_MASK,
+	.bank_ops = &jpeg_v5_0_1_aca_bank_ops,
+};
+
+static int jpeg_v5_0_1_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
+{
+	int r;
+
+	r = amdgpu_ras_block_late_init(adev, ras_block);
+	if (r)
+		return r;
+
+	r = amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__JPEG,
+				&jpeg_v5_0_1_aca_info, NULL);
+	if (r)
+		goto late_fini;
+
+	if (amdgpu_ras_is_supported(adev, ras_block->block) &&
+		adev->jpeg.inst->ras_poison_irq.funcs) {
+		r = amdgpu_irq_get(adev, &adev->jpeg.inst->ras_poison_irq, 0);
+		if (r)
+			goto late_fini;
+	}
+
+	return 0;
+
+late_fini:
+	amdgpu_ras_block_late_fini(adev, ras_block);
+
+	return r;
+}
+
+static struct amdgpu_jpeg_ras jpeg_v5_0_1_ras = {
+	.ras_block = {
+		.hw_ops = &jpeg_v5_0_1_ras_hw_ops,
+		.ras_late_init = jpeg_v5_0_1_ras_late_init,
+	},
+};
+
+static void jpeg_v5_0_1_set_ras_funcs(struct amdgpu_device *adev)
+{
+	adev->jpeg.ras = &jpeg_v5_0_1_ras;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.h b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.h
new file mode 100644
index 000000000000..a7e58d5fb246
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.h
@@ -0,0 +1,111 @@
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __JPEG_V5_0_1_H__
+#define __JPEG_V5_0_1_H__
+
+extern const struct amdgpu_ip_block_version jpeg_v5_0_1_ip_block;
+
+#define regUVD_JRBC0_UVD_JRBC_SCRATCH0_INTERNAL_OFFSET			0x4094
+#define regUVD_JRBC_EXTERNAL_MCM_ADDR_INTERNAL_OFFSET			0x1bffe
+
+#define regUVD_JRBC0_UVD_JRBC_RB_WPTR                                                         0x0640
+#define regUVD_JRBC0_UVD_JRBC_RB_WPTR_BASE_IDX                                                1
+#define regUVD_JRBC0_UVD_JRBC_STATUS                                                          0x0649
+#define regUVD_JRBC0_UVD_JRBC_STATUS_BASE_IDX                                                 1
+#define regUVD_JRBC0_UVD_JRBC_RB_RPTR                                                         0x064a
+#define regUVD_JRBC0_UVD_JRBC_RB_RPTR_BASE_IDX                                                1
+#define regUVD_JRBC1_UVD_JRBC_RB_WPTR                                                         0x0000
+#define regUVD_JRBC1_UVD_JRBC_RB_WPTR_BASE_IDX                                                0
+#define regUVD_JRBC1_UVD_JRBC_STATUS                                                          0x0009
+#define regUVD_JRBC1_UVD_JRBC_STATUS_BASE_IDX                                                 0
+#define regUVD_JRBC1_UVD_JRBC_RB_RPTR                                                         0x000a
+#define regUVD_JRBC1_UVD_JRBC_RB_RPTR_BASE_IDX                                                0
+#define regUVD_JRBC2_UVD_JRBC_RB_WPTR                                                         0x0040
+#define regUVD_JRBC2_UVD_JRBC_RB_WPTR_BASE_IDX                                                0
+#define regUVD_JRBC2_UVD_JRBC_STATUS                                                          0x0049
+#define regUVD_JRBC2_UVD_JRBC_STATUS_BASE_IDX                                                 0
+#define regUVD_JRBC2_UVD_JRBC_RB_RPTR                                                         0x004a
+#define regUVD_JRBC2_UVD_JRBC_RB_RPTR_BASE_IDX                                                0
+#define regUVD_JRBC3_UVD_JRBC_RB_WPTR                                                         0x0080
+#define regUVD_JRBC3_UVD_JRBC_RB_WPTR_BASE_IDX                                                0
+#define regUVD_JRBC3_UVD_JRBC_STATUS                                                          0x0089
+#define regUVD_JRBC3_UVD_JRBC_STATUS_BASE_IDX                                                 0
+#define regUVD_JRBC3_UVD_JRBC_RB_RPTR                                                         0x008a
+#define regUVD_JRBC3_UVD_JRBC_RB_RPTR_BASE_IDX                                                0
+#define regUVD_JRBC4_UVD_JRBC_RB_WPTR                                                         0x00c0
+#define regUVD_JRBC4_UVD_JRBC_RB_WPTR_BASE_IDX                                                0
+#define regUVD_JRBC4_UVD_JRBC_STATUS                                                          0x00c9
+#define regUVD_JRBC4_UVD_JRBC_STATUS_BASE_IDX                                                 0
+#define regUVD_JRBC4_UVD_JRBC_RB_RPTR                                                         0x00ca
+#define regUVD_JRBC4_UVD_JRBC_RB_RPTR_BASE_IDX                                                0
+#define regUVD_JRBC5_UVD_JRBC_RB_WPTR                                                         0x0100
+#define regUVD_JRBC5_UVD_JRBC_RB_WPTR_BASE_IDX                                                0
+#define regUVD_JRBC5_UVD_JRBC_STATUS                                                          0x0109
+#define regUVD_JRBC5_UVD_JRBC_STATUS_BASE_IDX                                                 0
+#define regUVD_JRBC5_UVD_JRBC_RB_RPTR                                                         0x010a
+#define regUVD_JRBC5_UVD_JRBC_RB_RPTR_BASE_IDX                                                0
+#define regUVD_JRBC6_UVD_JRBC_RB_WPTR                                                         0x0140
+#define regUVD_JRBC6_UVD_JRBC_RB_WPTR_BASE_IDX                                                0
+#define regUVD_JRBC6_UVD_JRBC_STATUS                                                          0x0149
+#define regUVD_JRBC6_UVD_JRBC_STATUS_BASE_IDX                                                 0
+#define regUVD_JRBC6_UVD_JRBC_RB_RPTR                                                         0x014a
+#define regUVD_JRBC6_UVD_JRBC_RB_RPTR_BASE_IDX                                                0
+#define regUVD_JRBC7_UVD_JRBC_RB_WPTR                                                         0x0180
+#define regUVD_JRBC7_UVD_JRBC_RB_WPTR_BASE_IDX                                                0
+#define regUVD_JRBC7_UVD_JRBC_STATUS                                                          0x0189
+#define regUVD_JRBC7_UVD_JRBC_STATUS_BASE_IDX                                                 0
+#define regUVD_JRBC7_UVD_JRBC_RB_RPTR                                                         0x018a
+#define regUVD_JRBC7_UVD_JRBC_RB_RPTR_BASE_IDX                                                0
+#define regUVD_JRBC8_UVD_JRBC_RB_WPTR                                                         0x01c0
+#define regUVD_JRBC8_UVD_JRBC_RB_WPTR_BASE_IDX                                                0
+#define regUVD_JRBC8_UVD_JRBC_STATUS                                                          0x01c9
+#define regUVD_JRBC8_UVD_JRBC_STATUS_BASE_IDX                                                 0
+#define regUVD_JRBC8_UVD_JRBC_RB_RPTR                                                         0x01ca
+#define regUVD_JRBC8_UVD_JRBC_RB_RPTR_BASE_IDX                                                0
+#define regUVD_JRBC9_UVD_JRBC_RB_WPTR                                                         0x0440
+#define regUVD_JRBC9_UVD_JRBC_RB_WPTR_BASE_IDX                                                1
+#define regUVD_JRBC9_UVD_JRBC_STATUS                                                          0x0449
+#define regUVD_JRBC9_UVD_JRBC_STATUS_BASE_IDX                                                 1
+#define regUVD_JRBC9_UVD_JRBC_RB_RPTR                                                         0x044a
+#define regUVD_JRBC9_UVD_JRBC_RB_RPTR_BASE_IDX                                                1
+#define regUVD_JMI0_JPEG_LMI_DROP                                                             0x0663
+#define regUVD_JMI0_JPEG_LMI_DROP_BASE_IDX                                                    1
+#define regUVD_JMI0_UVD_JMI_CLIENT_STALL                                                      0x067a
+#define regUVD_JMI0_UVD_JMI_CLIENT_STALL_BASE_IDX                                             1
+#define regUVD_JMI0_UVD_JMI_CLIENT_CLEAN_STATUS                                               0x067b
+#define regUVD_JMI0_UVD_JMI_CLIENT_CLEAN_STATUS_BASE_IDX                                      1
+#define regJPEG_CORE_RST_CTRL                                                                 0x072e
+#define regJPEG_CORE_RST_CTRL_BASE_IDX                                                        1
+
+#define regVCN_RRMT_CNTL                          0x0940
+#define regVCN_RRMT_CNTL_BASE_IDX                 1
+
+enum amdgpu_jpeg_v5_0_1_sub_block {
+	AMDGPU_JPEG_V5_0_1_JPEG0 = 0,
+	AMDGPU_JPEG_V5_0_1_JPEG1,
+
+	AMDGPU_JPEG_V5_0_1_MAX_SUB_BLOCK,
+};
+
+#endif /* __JPEG_V5_0_1_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/lsdma_v6_0.c b/drivers/gpu/drm/amd/amdgpu/lsdma_v6_0.c
new file mode 100644
index 000000000000..1a285b531881
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/lsdma_v6_0.c
@@ -0,0 +1,121 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/delay.h>
+#include "amdgpu.h"
+#include "lsdma_v6_0.h"
+#include "amdgpu_lsdma.h"
+
+#include "lsdma/lsdma_6_0_0_offset.h"
+#include "lsdma/lsdma_6_0_0_sh_mask.h"
+
+static int lsdma_v6_0_wait_pio_status(struct amdgpu_device *adev)
+{
+	return amdgpu_lsdma_wait_for(adev, SOC15_REG_OFFSET(LSDMA, 0, regLSDMA_PIO_STATUS),
+			LSDMA_PIO_STATUS__PIO_IDLE_MASK | LSDMA_PIO_STATUS__PIO_FIFO_EMPTY_MASK,
+			LSDMA_PIO_STATUS__PIO_IDLE_MASK | LSDMA_PIO_STATUS__PIO_FIFO_EMPTY_MASK);
+}
+
+static int lsdma_v6_0_copy_mem(struct amdgpu_device *adev,
+			       uint64_t src_addr,
+			       uint64_t dst_addr,
+			       uint64_t size)
+{
+	int ret;
+	uint32_t tmp;
+
+	WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_SRC_ADDR_LO, lower_32_bits(src_addr));
+	WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_SRC_ADDR_HI, upper_32_bits(src_addr));
+
+	WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_DST_ADDR_LO, lower_32_bits(dst_addr));
+	WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_DST_ADDR_HI, upper_32_bits(dst_addr));
+
+	WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_CONTROL, 0x0);
+
+	tmp = RREG32_SOC15(LSDMA, 0, regLSDMA_PIO_COMMAND);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, BYTE_COUNT, size);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, SRC_LOCATION, 0);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, DST_LOCATION, 0);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, SRC_ADDR_INC, 0);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, DST_ADDR_INC, 0);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, OVERLAP_DISABLE, 0);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, CONSTANT_FILL, 0);
+	WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_COMMAND, tmp);
+
+	ret = lsdma_v6_0_wait_pio_status(adev);
+	if (ret)
+		dev_err(adev->dev, "LSDMA PIO failed to copy memory!\n");
+
+	return ret;
+}
+
+static int lsdma_v6_0_fill_mem(struct amdgpu_device *adev,
+			       uint64_t dst_addr,
+			       uint32_t data,
+			       uint64_t size)
+{
+	int ret;
+	uint32_t tmp;
+
+	WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_CONSTFILL_DATA, data);
+
+	WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_DST_ADDR_LO, lower_32_bits(dst_addr));
+	WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_DST_ADDR_HI, upper_32_bits(dst_addr));
+
+	WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_CONTROL, 0x0);
+
+	tmp = RREG32_SOC15(LSDMA, 0, regLSDMA_PIO_COMMAND);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, BYTE_COUNT, size);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, SRC_LOCATION, 0);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, DST_LOCATION, 0);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, SRC_ADDR_INC, 0);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, DST_ADDR_INC, 0);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, OVERLAP_DISABLE, 0);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, CONSTANT_FILL, 1);
+	WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_COMMAND, tmp);
+
+	ret = lsdma_v6_0_wait_pio_status(adev);
+	if (ret)
+		dev_err(adev->dev, "LSDMA PIO failed to fill memory!\n");
+
+	return ret;
+}
+
+static void lsdma_v6_0_update_memory_power_gating(struct amdgpu_device *adev,
+						 bool enable)
+{
+	uint32_t tmp;
+
+	tmp = RREG32_SOC15(LSDMA, 0, regLSDMA_MEM_POWER_CTRL);
+	tmp = REG_SET_FIELD(tmp, LSDMA_MEM_POWER_CTRL, MEM_POWER_CTRL_EN, 0);
+	WREG32_SOC15(LSDMA, 0, regLSDMA_MEM_POWER_CTRL, tmp);
+
+	tmp = REG_SET_FIELD(tmp, LSDMA_MEM_POWER_CTRL, MEM_POWER_CTRL_EN, enable);
+	WREG32_SOC15(LSDMA, 0, regLSDMA_MEM_POWER_CTRL, tmp);
+}
+
+const struct amdgpu_lsdma_funcs lsdma_v6_0_funcs = {
+	.copy_mem = lsdma_v6_0_copy_mem,
+	.fill_mem = lsdma_v6_0_fill_mem,
+	.update_memory_power_gating = lsdma_v6_0_update_memory_power_gating
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/lsdma_v6_0.h b/drivers/gpu/drm/amd/amdgpu/lsdma_v6_0.h
new file mode 100644
index 000000000000..3ef79be1a9bf
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/lsdma_v6_0.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __LSDMA_V6_0_H__
+#define __LSDMA_V6_0_H__
+
+#include "soc15_common.h"
+
+extern const struct amdgpu_lsdma_funcs lsdma_v6_0_funcs;
+
+#endif /* __LSDMA_V6_0_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/lsdma_v7_0.c b/drivers/gpu/drm/amd/amdgpu/lsdma_v7_0.c
new file mode 100644
index 000000000000..396262044ea8
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/lsdma_v7_0.c
@@ -0,0 +1,121 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/delay.h>
+#include "amdgpu.h"
+#include "lsdma_v7_0.h"
+#include "amdgpu_lsdma.h"
+
+#include "lsdma/lsdma_7_0_0_offset.h"
+#include "lsdma/lsdma_7_0_0_sh_mask.h"
+
+static int lsdma_v7_0_wait_pio_status(struct amdgpu_device *adev)
+{
+	return amdgpu_lsdma_wait_for(adev, SOC15_REG_OFFSET(LSDMA, 0, regLSDMA_PIO_STATUS),
+			LSDMA_PIO_STATUS__PIO_IDLE_MASK | LSDMA_PIO_STATUS__PIO_FIFO_EMPTY_MASK,
+			LSDMA_PIO_STATUS__PIO_IDLE_MASK | LSDMA_PIO_STATUS__PIO_FIFO_EMPTY_MASK);
+}
+
+static int lsdma_v7_0_copy_mem(struct amdgpu_device *adev,
+			       uint64_t src_addr,
+			       uint64_t dst_addr,
+			       uint64_t size)
+{
+	int ret;
+	uint32_t tmp;
+
+	WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_SRC_ADDR_LO, lower_32_bits(src_addr));
+	WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_SRC_ADDR_HI, upper_32_bits(src_addr));
+
+	WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_DST_ADDR_LO, lower_32_bits(dst_addr));
+	WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_DST_ADDR_HI, upper_32_bits(dst_addr));
+
+	WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_CONTROL, 0x0);
+
+	tmp = RREG32_SOC15(LSDMA, 0, regLSDMA_PIO_COMMAND);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, BYTE_COUNT, size);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, SRC_LOCATION, 0);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, DST_LOCATION, 0);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, SRC_ADDR_INC, 0);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, DST_ADDR_INC, 0);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, OVERLAP_DISABLE, 0);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, CONSTANT_FILL, 0);
+	WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_COMMAND, tmp);
+
+	ret = lsdma_v7_0_wait_pio_status(adev);
+	if (ret)
+		dev_err(adev->dev, "LSDMA PIO failed to copy memory!\n");
+
+	return ret;
+}
+
+static int lsdma_v7_0_fill_mem(struct amdgpu_device *adev,
+			       uint64_t dst_addr,
+			       uint32_t data,
+			       uint64_t size)
+{
+	int ret;
+	uint32_t tmp;
+
+	WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_CONSTFILL_DATA, data);
+
+	WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_DST_ADDR_LO, lower_32_bits(dst_addr));
+	WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_DST_ADDR_HI, upper_32_bits(dst_addr));
+
+	WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_CONTROL, 0x0);
+
+	tmp = RREG32_SOC15(LSDMA, 0, regLSDMA_PIO_COMMAND);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, BYTE_COUNT, size);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, SRC_LOCATION, 0);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, DST_LOCATION, 0);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, SRC_ADDR_INC, 0);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, DST_ADDR_INC, 0);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, OVERLAP_DISABLE, 0);
+	tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, CONSTANT_FILL, 1);
+	WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_COMMAND, tmp);
+
+	ret = lsdma_v7_0_wait_pio_status(adev);
+	if (ret)
+		dev_err(adev->dev, "LSDMA PIO failed to fill memory!\n");
+
+	return ret;
+}
+
+static void lsdma_v7_0_update_memory_power_gating(struct amdgpu_device *adev,
+						 bool enable)
+{
+	uint32_t tmp;
+
+	tmp = RREG32_SOC15(LSDMA, 0, regLSDMA_MEM_POWER_CTRL);
+	tmp = REG_SET_FIELD(tmp, LSDMA_MEM_POWER_CTRL, MEM_POWER_CTRL_EN, 0);
+	WREG32_SOC15(LSDMA, 0, regLSDMA_MEM_POWER_CTRL, tmp);
+
+	tmp = REG_SET_FIELD(tmp, LSDMA_MEM_POWER_CTRL, MEM_POWER_CTRL_EN, enable);
+	WREG32_SOC15(LSDMA, 0, regLSDMA_MEM_POWER_CTRL, tmp);
+}
+
+const struct amdgpu_lsdma_funcs lsdma_v7_0_funcs = {
+	.copy_mem = lsdma_v7_0_copy_mem,
+	.fill_mem = lsdma_v7_0_fill_mem,
+	.update_memory_power_gating = lsdma_v7_0_update_memory_power_gating
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/lsdma_v7_0.h b/drivers/gpu/drm/amd/amdgpu/lsdma_v7_0.h
new file mode 100644
index 000000000000..52b4485cdd98
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/lsdma_v7_0.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __LSDMA_V7_0_H__
+#define __LSDMA_V7_0_H__
+
+#include "soc15_common.h"
+
+extern const struct amdgpu_lsdma_funcs lsdma_v7_0_funcs;
+
+#endif /* __LSDMA_V7_0_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/mca_v3_0.c b/drivers/gpu/drm/amd/amdgpu/mca_v3_0.c
new file mode 100644
index 000000000000..6dae4a2e2767
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mca_v3_0.c
@@ -0,0 +1,104 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu_ras.h"
+#include "amdgpu.h"
+#include "amdgpu_mca.h"
+
+#define smnMCMP0_STATUST0 	0x03830408
+#define smnMCMP1_STATUST0 	0x03b30408
+#define smnMCMPIO_STATUST0 	0x0c930408
+
+
+static void mca_v3_0_mp0_query_ras_error_count(struct amdgpu_device *adev,
+					       void *ras_error_status)
+{
+	amdgpu_mca_query_ras_error_count(adev,
+				         smnMCMP0_STATUST0,
+				         ras_error_status);
+}
+
+static int mca_v3_0_ras_block_match(struct amdgpu_ras_block_object *block_obj,
+				enum amdgpu_ras_block block, uint32_t sub_block_index)
+{
+	if (!block_obj)
+		return -EINVAL;
+
+	if ((block_obj->ras_comm.block == block) &&
+		(block_obj->ras_comm.sub_block_index == sub_block_index)) {
+		return 0;
+	}
+
+	return -EINVAL;
+}
+
+static const struct amdgpu_ras_block_hw_ops mca_v3_0_mp0_hw_ops = {
+	.query_ras_error_count = mca_v3_0_mp0_query_ras_error_count,
+	.query_ras_error_address = NULL,
+};
+
+struct amdgpu_mca_ras_block mca_v3_0_mp0_ras = {
+	.ras_block = {
+		.hw_ops = &mca_v3_0_mp0_hw_ops,
+		.ras_block_match = mca_v3_0_ras_block_match,
+	},
+};
+
+static void mca_v3_0_mp1_query_ras_error_count(struct amdgpu_device *adev,
+					       void *ras_error_status)
+{
+	amdgpu_mca_query_ras_error_count(adev,
+				         smnMCMP1_STATUST0,
+				         ras_error_status);
+}
+
+static const struct amdgpu_ras_block_hw_ops mca_v3_0_mp1_hw_ops = {
+	.query_ras_error_count = mca_v3_0_mp1_query_ras_error_count,
+	.query_ras_error_address = NULL,
+};
+
+struct amdgpu_mca_ras_block mca_v3_0_mp1_ras = {
+	.ras_block = {
+		.hw_ops = &mca_v3_0_mp1_hw_ops,
+		.ras_block_match = mca_v3_0_ras_block_match,
+	},
+};
+
+static void mca_v3_0_mpio_query_ras_error_count(struct amdgpu_device *adev,
+					       void *ras_error_status)
+{
+	amdgpu_mca_query_ras_error_count(adev,
+				         smnMCMPIO_STATUST0,
+				         ras_error_status);
+}
+
+static const struct amdgpu_ras_block_hw_ops mca_v3_0_mpio_hw_ops = {
+	.query_ras_error_count = mca_v3_0_mpio_query_ras_error_count,
+	.query_ras_error_address = NULL,
+};
+
+struct amdgpu_mca_ras_block mca_v3_0_mpio_ras = {
+	.ras_block = {
+		.hw_ops = &mca_v3_0_mpio_hw_ops,
+		.ras_block_match = mca_v3_0_ras_block_match,
+	},
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/mca_v3_0.h b/drivers/gpu/drm/amd/amdgpu/mca_v3_0.h
new file mode 100644
index 000000000000..d3eaef0d7f2d
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mca_v3_0.h
@@ -0,0 +1,28 @@
+/*
+ * Copyright (C) 2021  Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+#ifndef __MCA_V3_0_H__
+#define __MCA_V3_0_H__
+
+extern struct amdgpu_mca_ras_block mca_v3_0_mp0_ras;
+extern struct amdgpu_mca_ras_block mca_v3_0_mp1_ras;
+extern struct amdgpu_mca_ras_block mca_v3_0_mpio_ras;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/mes_userqueue.c b/drivers/gpu/drm/amd/amdgpu/mes_userqueue.c
new file mode 100644
index 000000000000..2db9b2c63693
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mes_userqueue.c
@@ -0,0 +1,485 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <drm/drm_drv.h>
+#include "amdgpu.h"
+#include "amdgpu_gfx.h"
+#include "mes_userqueue.h"
+#include "amdgpu_userq_fence.h"
+
+#define AMDGPU_USERQ_PROC_CTX_SZ PAGE_SIZE
+#define AMDGPU_USERQ_GANG_CTX_SZ PAGE_SIZE
+
+static int
+mes_userq_map_gtt_bo_to_gart(struct amdgpu_bo *bo)
+{
+	int ret;
+
+	ret = amdgpu_bo_reserve(bo, true);
+	if (ret) {
+		DRM_ERROR("Failed to reserve bo. ret %d\n", ret);
+		goto err_reserve_bo_failed;
+	}
+
+	ret = amdgpu_ttm_alloc_gart(&bo->tbo);
+	if (ret) {
+		DRM_ERROR("Failed to bind bo to GART. ret %d\n", ret);
+		goto err_map_bo_gart_failed;
+	}
+
+	amdgpu_bo_unreserve(bo);
+	bo = amdgpu_bo_ref(bo);
+
+	return 0;
+
+err_map_bo_gart_failed:
+	amdgpu_bo_unreserve(bo);
+err_reserve_bo_failed:
+	return ret;
+}
+
+static int
+mes_userq_create_wptr_mapping(struct amdgpu_userq_mgr *uq_mgr,
+			      struct amdgpu_usermode_queue *queue,
+			      uint64_t wptr)
+{
+	struct amdgpu_bo_va_mapping *wptr_mapping;
+	struct amdgpu_vm *wptr_vm;
+	struct amdgpu_userq_obj *wptr_obj = &queue->wptr_obj;
+	int ret;
+
+	wptr_vm = queue->vm;
+	ret = amdgpu_bo_reserve(wptr_vm->root.bo, false);
+	if (ret)
+		return ret;
+
+	wptr &= AMDGPU_GMC_HOLE_MASK;
+	wptr_mapping = amdgpu_vm_bo_lookup_mapping(wptr_vm, wptr >> PAGE_SHIFT);
+	amdgpu_bo_unreserve(wptr_vm->root.bo);
+	if (!wptr_mapping) {
+		DRM_ERROR("Failed to lookup wptr bo\n");
+		return -EINVAL;
+	}
+
+	wptr_obj->obj = wptr_mapping->bo_va->base.bo;
+	if (wptr_obj->obj->tbo.base.size > PAGE_SIZE) {
+		DRM_ERROR("Requested GART mapping for wptr bo larger than one page\n");
+		return -EINVAL;
+	}
+
+	ret = mes_userq_map_gtt_bo_to_gart(wptr_obj->obj);
+	if (ret) {
+		DRM_ERROR("Failed to map wptr bo to GART\n");
+		return ret;
+	}
+
+	queue->wptr_obj.gpu_addr = amdgpu_bo_gpu_offset_no_check(wptr_obj->obj);
+	return 0;
+}
+
+static int convert_to_mes_priority(int priority)
+{
+	switch (priority) {
+	case AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_NORMAL_LOW:
+	default:
+		return AMDGPU_MES_PRIORITY_LEVEL_NORMAL;
+	case AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_LOW:
+		return AMDGPU_MES_PRIORITY_LEVEL_LOW;
+	case AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_NORMAL_HIGH:
+		return AMDGPU_MES_PRIORITY_LEVEL_MEDIUM;
+	case AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_HIGH:
+		return AMDGPU_MES_PRIORITY_LEVEL_HIGH;
+	}
+}
+
+static int mes_userq_map(struct amdgpu_userq_mgr *uq_mgr,
+			 struct amdgpu_usermode_queue *queue)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	struct amdgpu_userq_obj *ctx = &queue->fw_obj;
+	struct amdgpu_mqd_prop *userq_props = queue->userq_prop;
+	struct mes_add_queue_input queue_input;
+	int r;
+
+	memset(&queue_input, 0x0, sizeof(struct mes_add_queue_input));
+
+	queue_input.process_va_start = 0;
+	queue_input.process_va_end = adev->vm_manager.max_pfn - 1;
+
+	/* set process quantum to 10 ms and gang quantum to 1 ms as default */
+	queue_input.process_quantum = 100000;
+	queue_input.gang_quantum = 10000;
+	queue_input.paging = false;
+
+	queue_input.process_context_addr = ctx->gpu_addr;
+	queue_input.gang_context_addr = ctx->gpu_addr + AMDGPU_USERQ_PROC_CTX_SZ;
+	queue_input.inprocess_gang_priority = AMDGPU_MES_PRIORITY_LEVEL_NORMAL;
+	queue_input.gang_global_priority_level = convert_to_mes_priority(queue->priority);
+
+	queue_input.process_id = queue->vm->pasid;
+	queue_input.queue_type = queue->queue_type;
+	queue_input.mqd_addr = queue->mqd.gpu_addr;
+	queue_input.wptr_addr = userq_props->wptr_gpu_addr;
+	queue_input.queue_size = userq_props->queue_size >> 2;
+	queue_input.doorbell_offset = userq_props->doorbell_index;
+	queue_input.page_table_base_addr = amdgpu_gmc_pd_addr(queue->vm->root.bo);
+	queue_input.wptr_mc_addr = queue->wptr_obj.gpu_addr;
+
+	amdgpu_mes_lock(&adev->mes);
+	r = adev->mes.funcs->add_hw_queue(&adev->mes, &queue_input);
+	amdgpu_mes_unlock(&adev->mes);
+	if (r) {
+		DRM_ERROR("Failed to map queue in HW, err (%d)\n", r);
+		return r;
+	}
+
+	DRM_DEBUG_DRIVER("Queue (doorbell:%d) mapped successfully\n", userq_props->doorbell_index);
+	return 0;
+}
+
+static int mes_userq_unmap(struct amdgpu_userq_mgr *uq_mgr,
+			   struct amdgpu_usermode_queue *queue)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	struct mes_remove_queue_input queue_input;
+	struct amdgpu_userq_obj *ctx = &queue->fw_obj;
+	int r;
+
+	memset(&queue_input, 0x0, sizeof(struct mes_remove_queue_input));
+	queue_input.doorbell_offset = queue->doorbell_index;
+	queue_input.gang_context_addr = ctx->gpu_addr + AMDGPU_USERQ_PROC_CTX_SZ;
+
+	amdgpu_mes_lock(&adev->mes);
+	r = adev->mes.funcs->remove_hw_queue(&adev->mes, &queue_input);
+	amdgpu_mes_unlock(&adev->mes);
+	if (r)
+		DRM_ERROR("Failed to unmap queue in HW, err (%d)\n", r);
+	return r;
+}
+
+static int mes_userq_create_ctx_space(struct amdgpu_userq_mgr *uq_mgr,
+				      struct amdgpu_usermode_queue *queue,
+				      struct drm_amdgpu_userq_in *mqd_user)
+{
+	struct amdgpu_userq_obj *ctx = &queue->fw_obj;
+	int r, size;
+
+	/*
+	 * The FW expects at least one page space allocated for
+	 * process ctx and gang ctx each. Create an object
+	 * for the same.
+	 */
+	size = AMDGPU_USERQ_PROC_CTX_SZ + AMDGPU_USERQ_GANG_CTX_SZ;
+	r = amdgpu_userq_create_object(uq_mgr, ctx, size);
+	if (r) {
+		DRM_ERROR("Failed to allocate ctx space bo for userqueue, err:%d\n", r);
+		return r;
+	}
+
+	return 0;
+}
+
+static int mes_userq_detect_and_reset(struct amdgpu_device *adev,
+				      int queue_type)
+{
+	int db_array_size = amdgpu_mes_get_hung_queue_db_array_size(adev);
+	struct mes_detect_and_reset_queue_input input;
+	struct amdgpu_usermode_queue *queue;
+	struct amdgpu_userq_mgr *uqm, *tmp;
+	unsigned int hung_db_num = 0;
+	int queue_id, r, i;
+	u32 db_array[4];
+
+	if (db_array_size > 4) {
+		dev_err(adev->dev, "DB array size (%d vs 4) too small\n",
+			db_array_size);
+		return -EINVAL;
+	}
+
+	memset(&input, 0x0, sizeof(struct mes_detect_and_reset_queue_input));
+
+	input.queue_type = queue_type;
+
+	amdgpu_mes_lock(&adev->mes);
+	r = amdgpu_mes_detect_and_reset_hung_queues(adev, queue_type, false,
+						    &hung_db_num, db_array);
+	amdgpu_mes_unlock(&adev->mes);
+	if (r) {
+		dev_err(adev->dev, "Failed to detect and reset queues, err (%d)\n", r);
+	} else if (hung_db_num) {
+		list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) {
+			idr_for_each_entry(&uqm->userq_idr, queue, queue_id) {
+				if (queue->queue_type == queue_type) {
+					for (i = 0; i < hung_db_num; i++) {
+						if (queue->doorbell_index == db_array[i]) {
+							queue->state = AMDGPU_USERQ_STATE_HUNG;
+							atomic_inc(&adev->gpu_reset_counter);
+							amdgpu_userq_fence_driver_force_completion(queue);
+							drm_dev_wedged_event(adev_to_drm(adev), DRM_WEDGE_RECOVERY_NONE, NULL);
+						}
+					}
+				}
+			}
+		}
+	}
+
+	return r;
+}
+
+static int mes_userq_mqd_create(struct amdgpu_userq_mgr *uq_mgr,
+				struct drm_amdgpu_userq_in *args_in,
+				struct amdgpu_usermode_queue *queue)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	struct amdgpu_mqd *mqd_hw_default = &adev->mqds[queue->queue_type];
+	struct drm_amdgpu_userq_in *mqd_user = args_in;
+	struct amdgpu_mqd_prop *userq_props;
+	struct amdgpu_gfx_shadow_info shadow_info;
+	int r;
+
+	/* Structure to initialize MQD for userqueue using generic MQD init function */
+	userq_props = kzalloc(sizeof(struct amdgpu_mqd_prop), GFP_KERNEL);
+	if (!userq_props) {
+		DRM_ERROR("Failed to allocate memory for userq_props\n");
+		return -ENOMEM;
+	}
+
+	r = amdgpu_userq_create_object(uq_mgr, &queue->mqd, mqd_hw_default->mqd_size);
+	if (r) {
+		DRM_ERROR("Failed to create MQD object for userqueue\n");
+		goto free_props;
+	}
+
+	/* Initialize the MQD BO with user given values */
+	userq_props->wptr_gpu_addr = mqd_user->wptr_va;
+	userq_props->rptr_gpu_addr = mqd_user->rptr_va;
+	userq_props->queue_size = mqd_user->queue_size;
+	userq_props->hqd_base_gpu_addr = mqd_user->queue_va;
+	userq_props->mqd_gpu_addr = queue->mqd.gpu_addr;
+	userq_props->use_doorbell = true;
+	userq_props->doorbell_index = queue->doorbell_index;
+	userq_props->fence_address = queue->fence_drv->gpu_addr;
+
+	if (adev->gfx.funcs->get_gfx_shadow_info)
+		adev->gfx.funcs->get_gfx_shadow_info(adev, &shadow_info, true);
+	if (queue->queue_type == AMDGPU_HW_IP_COMPUTE) {
+		struct drm_amdgpu_userq_mqd_compute_gfx11 *compute_mqd;
+
+		if (mqd_user->mqd_size != sizeof(*compute_mqd)) {
+			DRM_ERROR("Invalid compute IP MQD size\n");
+			r = -EINVAL;
+			goto free_mqd;
+		}
+
+		compute_mqd = memdup_user(u64_to_user_ptr(mqd_user->mqd), mqd_user->mqd_size);
+		if (IS_ERR(compute_mqd)) {
+			DRM_ERROR("Failed to read user MQD\n");
+			r = -ENOMEM;
+			goto free_mqd;
+		}
+
+		if (amdgpu_userq_input_va_validate(queue->vm, compute_mqd->eop_va,
+		    max_t(u32, PAGE_SIZE, AMDGPU_GPU_PAGE_SIZE)))
+			goto free_mqd;
+
+		userq_props->eop_gpu_addr = compute_mqd->eop_va;
+		userq_props->hqd_pipe_priority = AMDGPU_GFX_PIPE_PRIO_NORMAL;
+		userq_props->hqd_queue_priority = AMDGPU_GFX_QUEUE_PRIORITY_MINIMUM;
+		userq_props->hqd_active = false;
+		userq_props->tmz_queue =
+			mqd_user->flags & AMDGPU_USERQ_CREATE_FLAGS_QUEUE_SECURE;
+		kfree(compute_mqd);
+	} else if (queue->queue_type == AMDGPU_HW_IP_GFX) {
+		struct drm_amdgpu_userq_mqd_gfx11 *mqd_gfx_v11;
+
+		if (mqd_user->mqd_size != sizeof(*mqd_gfx_v11) || !mqd_user->mqd) {
+			DRM_ERROR("Invalid GFX MQD\n");
+			r = -EINVAL;
+			goto free_mqd;
+		}
+
+		mqd_gfx_v11 = memdup_user(u64_to_user_ptr(mqd_user->mqd), mqd_user->mqd_size);
+		if (IS_ERR(mqd_gfx_v11)) {
+			DRM_ERROR("Failed to read user MQD\n");
+			r = -ENOMEM;
+			goto free_mqd;
+		}
+
+		userq_props->shadow_addr = mqd_gfx_v11->shadow_va;
+		userq_props->csa_addr = mqd_gfx_v11->csa_va;
+		userq_props->tmz_queue =
+			mqd_user->flags & AMDGPU_USERQ_CREATE_FLAGS_QUEUE_SECURE;
+
+		if (amdgpu_userq_input_va_validate(queue->vm, mqd_gfx_v11->shadow_va,
+		    shadow_info.shadow_size))
+			goto free_mqd;
+
+		kfree(mqd_gfx_v11);
+	} else if (queue->queue_type == AMDGPU_HW_IP_DMA) {
+		struct drm_amdgpu_userq_mqd_sdma_gfx11 *mqd_sdma_v11;
+
+		if (mqd_user->mqd_size != sizeof(*mqd_sdma_v11) || !mqd_user->mqd) {
+			DRM_ERROR("Invalid SDMA MQD\n");
+			r = -EINVAL;
+			goto free_mqd;
+		}
+
+		mqd_sdma_v11 = memdup_user(u64_to_user_ptr(mqd_user->mqd), mqd_user->mqd_size);
+		if (IS_ERR(mqd_sdma_v11)) {
+			DRM_ERROR("Failed to read sdma user MQD\n");
+			r = -ENOMEM;
+			goto free_mqd;
+		}
+
+		if (amdgpu_userq_input_va_validate(queue->vm, mqd_sdma_v11->csa_va,
+		    shadow_info.csa_size))
+			goto free_mqd;
+
+		userq_props->csa_addr = mqd_sdma_v11->csa_va;
+		kfree(mqd_sdma_v11);
+	}
+
+	queue->userq_prop = userq_props;
+
+	r = mqd_hw_default->init_mqd(adev, (void *)queue->mqd.cpu_ptr, userq_props);
+	if (r) {
+		DRM_ERROR("Failed to initialize MQD for userqueue\n");
+		goto free_mqd;
+	}
+
+	/* Create BO for FW operations */
+	r = mes_userq_create_ctx_space(uq_mgr, queue, mqd_user);
+	if (r) {
+		DRM_ERROR("Failed to allocate BO for userqueue (%d)", r);
+		goto free_mqd;
+	}
+
+	/* FW expects WPTR BOs to be mapped into GART */
+	r = mes_userq_create_wptr_mapping(uq_mgr, queue, userq_props->wptr_gpu_addr);
+	if (r) {
+		DRM_ERROR("Failed to create WPTR mapping\n");
+		goto free_ctx;
+	}
+
+	return 0;
+
+free_ctx:
+	amdgpu_userq_destroy_object(uq_mgr, &queue->fw_obj);
+
+free_mqd:
+	amdgpu_userq_destroy_object(uq_mgr, &queue->mqd);
+
+free_props:
+	kfree(userq_props);
+
+	return r;
+}
+
+static void
+mes_userq_mqd_destroy(struct amdgpu_userq_mgr *uq_mgr,
+		      struct amdgpu_usermode_queue *queue)
+{
+	amdgpu_userq_destroy_object(uq_mgr, &queue->fw_obj);
+	kfree(queue->userq_prop);
+	amdgpu_userq_destroy_object(uq_mgr, &queue->mqd);
+}
+
+static int mes_userq_preempt(struct amdgpu_userq_mgr *uq_mgr,
+				struct amdgpu_usermode_queue *queue)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	struct mes_suspend_gang_input queue_input;
+	struct amdgpu_userq_obj *ctx = &queue->fw_obj;
+	signed long timeout = 2100000; /* 2100 ms */
+	u64 fence_gpu_addr;
+	u32 fence_offset;
+	u64 *fence_ptr;
+	int i, r;
+
+	if (queue->state != AMDGPU_USERQ_STATE_MAPPED)
+		return 0;
+	r = amdgpu_device_wb_get(adev, &fence_offset);
+	if (r)
+		return r;
+
+	fence_gpu_addr = adev->wb.gpu_addr + (fence_offset * 4);
+	fence_ptr = (u64 *)&adev->wb.wb[fence_offset];
+	*fence_ptr = 0;
+
+	memset(&queue_input, 0x0, sizeof(struct mes_suspend_gang_input));
+	queue_input.gang_context_addr = ctx->gpu_addr + AMDGPU_USERQ_PROC_CTX_SZ;
+	queue_input.suspend_fence_addr = fence_gpu_addr;
+	queue_input.suspend_fence_value = 1;
+	amdgpu_mes_lock(&adev->mes);
+	r = adev->mes.funcs->suspend_gang(&adev->mes, &queue_input);
+	amdgpu_mes_unlock(&adev->mes);
+	if (r) {
+		DRM_ERROR("Failed to suspend gang: %d\n", r);
+		goto out;
+	}
+
+	for (i = 0; i < timeout; i++) {
+		if (*fence_ptr == 1)
+			goto out;
+		udelay(1);
+	}
+	r = -ETIMEDOUT;
+
+out:
+	amdgpu_device_wb_free(adev, fence_offset);
+	return r;
+}
+
+static int mes_userq_restore(struct amdgpu_userq_mgr *uq_mgr,
+				struct amdgpu_usermode_queue *queue)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	struct mes_resume_gang_input queue_input;
+	struct amdgpu_userq_obj *ctx = &queue->fw_obj;
+	int r;
+
+	if (queue->state == AMDGPU_USERQ_STATE_HUNG)
+		return -EINVAL;
+	if (queue->state != AMDGPU_USERQ_STATE_PREEMPTED)
+		return 0;
+
+	memset(&queue_input, 0x0, sizeof(struct mes_resume_gang_input));
+	queue_input.gang_context_addr = ctx->gpu_addr + AMDGPU_USERQ_PROC_CTX_SZ;
+
+	amdgpu_mes_lock(&adev->mes);
+	r = adev->mes.funcs->resume_gang(&adev->mes, &queue_input);
+	amdgpu_mes_unlock(&adev->mes);
+	if (r)
+		dev_err(adev->dev, "Failed to resume queue, err (%d)\n", r);
+	return r;
+}
+
+const struct amdgpu_userq_funcs userq_mes_funcs = {
+	.mqd_create = mes_userq_mqd_create,
+	.mqd_destroy = mes_userq_mqd_destroy,
+	.unmap = mes_userq_unmap,
+	.map = mes_userq_map,
+	.detect_and_reset = mes_userq_detect_and_reset,
+	.preempt = mes_userq_preempt,
+	.restore = mes_userq_restore,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/mes_userqueue.h b/drivers/gpu/drm/amd/amdgpu/mes_userqueue.h
new file mode 100644
index 000000000000..090ae8897770
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mes_userqueue.h
@@ -0,0 +1,30 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef MES_USERQ_H
+#define MES_USERQ_H
+#include "amdgpu_userq.h"
+
+extern const struct amdgpu_userq_funcs userq_mes_funcs;
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c
new file mode 100644
index 000000000000..e82188431f79
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c
@@ -0,0 +1,1754 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include "amdgpu.h"
+#include "soc15_common.h"
+#include "soc21.h"
+#include "gfx_v11_0.h"
+#include "gc/gc_11_0_0_offset.h"
+#include "gc/gc_11_0_0_sh_mask.h"
+#include "gc/gc_11_0_0_default.h"
+#include "v11_structs.h"
+#include "mes_v11_api_def.h"
+
+MODULE_FIRMWARE("amdgpu/gc_11_0_0_mes.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_0_mes_2.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_0_mes1.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_1_mes.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_1_mes_2.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_1_mes1.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_2_mes.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_2_mes_2.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_2_mes1.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_3_mes.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_3_mes_2.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_3_mes1.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_4_mes.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_4_mes_2.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_4_mes1.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_0_mes_2.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_0_mes1.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_1_mes_2.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_1_mes1.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_2_mes_2.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_2_mes1.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_3_mes_2.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_3_mes1.bin");
+
+static int mes_v11_0_hw_init(struct amdgpu_ip_block *ip_block);
+static int mes_v11_0_hw_fini(struct amdgpu_ip_block *ip_block);
+static int mes_v11_0_kiq_hw_init(struct amdgpu_device *adev);
+static int mes_v11_0_kiq_hw_fini(struct amdgpu_device *adev);
+
+#define MES_EOP_SIZE   2048
+#define GFX_MES_DRAM_SIZE	0x80000
+#define MES11_HW_RESOURCE_1_SIZE (128 * AMDGPU_GPU_PAGE_SIZE)
+
+#define MES11_HUNG_DB_OFFSET_ARRAY_SIZE 4
+
+static void mes_v11_0_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr,
+			     ring->wptr);
+		WDOORBELL64(ring->doorbell_index, ring->wptr);
+	} else {
+		BUG();
+	}
+}
+
+static u64 mes_v11_0_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	return *ring->rptr_cpu_addr;
+}
+
+static u64 mes_v11_0_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	u64 wptr;
+
+	if (ring->use_doorbell)
+		wptr = atomic64_read((atomic64_t *)ring->wptr_cpu_addr);
+	else
+		BUG();
+	return wptr;
+}
+
+static const struct amdgpu_ring_funcs mes_v11_0_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_MES,
+	.align_mask = 1,
+	.nop = 0,
+	.support_64bit_ptrs = true,
+	.get_rptr = mes_v11_0_ring_get_rptr,
+	.get_wptr = mes_v11_0_ring_get_wptr,
+	.set_wptr = mes_v11_0_ring_set_wptr,
+	.insert_nop = amdgpu_ring_insert_nop,
+};
+
+static const char *mes_v11_0_opcodes[] = {
+	"SET_HW_RSRC",
+	"SET_SCHEDULING_CONFIG",
+	"ADD_QUEUE",
+	"REMOVE_QUEUE",
+	"PERFORM_YIELD",
+	"SET_GANG_PRIORITY_LEVEL",
+	"SUSPEND",
+	"RESUME",
+	"RESET",
+	"SET_LOG_BUFFER",
+	"CHANGE_GANG_PRORITY",
+	"QUERY_SCHEDULER_STATUS",
+	"PROGRAM_GDS",
+	"SET_DEBUG_VMID",
+	"MISC",
+	"UPDATE_ROOT_PAGE_TABLE",
+	"AMD_LOG",
+	"unused",
+	"unused",
+	"SET_HW_RSRC_1",
+};
+
+static const char *mes_v11_0_misc_opcodes[] = {
+	"WRITE_REG",
+	"INV_GART",
+	"QUERY_STATUS",
+	"READ_REG",
+	"WAIT_REG_MEM",
+	"SET_SHADER_DEBUGGER",
+};
+
+static const char *mes_v11_0_get_op_string(union MESAPI__MISC *x_pkt)
+{
+	const char *op_str = NULL;
+
+	if (x_pkt->header.opcode < ARRAY_SIZE(mes_v11_0_opcodes))
+		op_str = mes_v11_0_opcodes[x_pkt->header.opcode];
+
+	return op_str;
+}
+
+static const char *mes_v11_0_get_misc_op_string(union MESAPI__MISC *x_pkt)
+{
+	const char *op_str = NULL;
+
+	if ((x_pkt->header.opcode == MES_SCH_API_MISC) &&
+	    (x_pkt->opcode < ARRAY_SIZE(mes_v11_0_misc_opcodes)))
+		op_str = mes_v11_0_misc_opcodes[x_pkt->opcode];
+
+	return op_str;
+}
+
+static int mes_v11_0_submit_pkt_and_poll_completion(struct amdgpu_mes *mes,
+						    void *pkt, int size,
+						    int api_status_off)
+{
+	union MESAPI__QUERY_MES_STATUS mes_status_pkt;
+	signed long timeout = 2100000; /* 2100 ms */
+	struct amdgpu_device *adev = mes->adev;
+	struct amdgpu_ring *ring = &mes->ring[0];
+	struct MES_API_STATUS *api_status;
+	union MESAPI__MISC *x_pkt = pkt;
+	const char *op_str, *misc_op_str;
+	unsigned long flags;
+	u64 status_gpu_addr;
+	u32 seq, status_offset;
+	u64 *status_ptr;
+	signed long r;
+	int ret;
+
+	if (x_pkt->header.opcode >= MES_SCH_API_MAX)
+		return -EINVAL;
+
+	if (amdgpu_emu_mode) {
+		timeout *= 100;
+	} else if (amdgpu_sriov_vf(adev)) {
+		/* Worst case in sriov where all other 15 VF timeout, each VF needs about 600ms */
+		timeout = 15 * 600 * 1000;
+	}
+
+	ret = amdgpu_device_wb_get(adev, &status_offset);
+	if (ret)
+		return ret;
+
+	status_gpu_addr = adev->wb.gpu_addr + (status_offset * 4);
+	status_ptr = (u64 *)&adev->wb.wb[status_offset];
+	*status_ptr = 0;
+
+	spin_lock_irqsave(&mes->ring_lock[0], flags);
+	r = amdgpu_ring_alloc(ring, (size + sizeof(mes_status_pkt)) / 4);
+	if (r)
+		goto error_unlock_free;
+
+	seq = ++ring->fence_drv.sync_seq;
+	r = amdgpu_fence_wait_polling(ring,
+				      seq - ring->fence_drv.num_fences_mask,
+				      timeout);
+	if (r < 1)
+		goto error_undo;
+
+	api_status = (struct MES_API_STATUS *)((char *)pkt + api_status_off);
+	api_status->api_completion_fence_addr = status_gpu_addr;
+	api_status->api_completion_fence_value = 1;
+
+	amdgpu_ring_write_multiple(ring, pkt, size / 4);
+
+	memset(&mes_status_pkt, 0, sizeof(mes_status_pkt));
+	mes_status_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_status_pkt.header.opcode = MES_SCH_API_QUERY_SCHEDULER_STATUS;
+	mes_status_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+	mes_status_pkt.api_status.api_completion_fence_addr =
+		ring->fence_drv.gpu_addr;
+	mes_status_pkt.api_status.api_completion_fence_value = seq;
+
+	amdgpu_ring_write_multiple(ring, &mes_status_pkt,
+				   sizeof(mes_status_pkt) / 4);
+
+	amdgpu_ring_commit(ring);
+	spin_unlock_irqrestore(&mes->ring_lock[0], flags);
+
+	op_str = mes_v11_0_get_op_string(x_pkt);
+	misc_op_str = mes_v11_0_get_misc_op_string(x_pkt);
+
+	if (misc_op_str)
+		dev_dbg(adev->dev, "MES msg=%s (%s) was emitted\n", op_str,
+			misc_op_str);
+	else if (op_str)
+		dev_dbg(adev->dev, "MES msg=%s was emitted\n", op_str);
+	else
+		dev_dbg(adev->dev, "MES msg=%d was emitted\n",
+			x_pkt->header.opcode);
+
+	r = amdgpu_fence_wait_polling(ring, seq, timeout);
+	if (r < 1 || !*status_ptr) {
+
+		if (misc_op_str)
+			dev_err(adev->dev, "MES failed to respond to msg=%s (%s)\n",
+				op_str, misc_op_str);
+		else if (op_str)
+			dev_err(adev->dev, "MES failed to respond to msg=%s\n",
+				op_str);
+		else
+			dev_err(adev->dev, "MES failed to respond to msg=%d\n",
+				x_pkt->header.opcode);
+
+		while (halt_if_hws_hang)
+			schedule();
+
+		r = -ETIMEDOUT;
+		goto error_wb_free;
+	}
+
+	amdgpu_device_wb_free(adev, status_offset);
+	return 0;
+
+error_undo:
+	dev_err(adev->dev, "MES ring buffer is full.\n");
+	amdgpu_ring_undo(ring);
+
+error_unlock_free:
+	spin_unlock_irqrestore(&mes->ring_lock[0], flags);
+
+error_wb_free:
+	amdgpu_device_wb_free(adev, status_offset);
+	return r;
+}
+
+static int convert_to_mes_queue_type(int queue_type)
+{
+	if (queue_type == AMDGPU_RING_TYPE_GFX)
+		return MES_QUEUE_TYPE_GFX;
+	else if (queue_type == AMDGPU_RING_TYPE_COMPUTE)
+		return MES_QUEUE_TYPE_COMPUTE;
+	else if (queue_type == AMDGPU_RING_TYPE_SDMA)
+		return MES_QUEUE_TYPE_SDMA;
+	else
+		BUG();
+	return -1;
+}
+
+static int convert_to_mes_priority_level(int priority_level)
+{
+	switch (priority_level) {
+	case AMDGPU_MES_PRIORITY_LEVEL_LOW:
+		return AMD_PRIORITY_LEVEL_LOW;
+	case AMDGPU_MES_PRIORITY_LEVEL_NORMAL:
+	default:
+		return AMD_PRIORITY_LEVEL_NORMAL;
+	case AMDGPU_MES_PRIORITY_LEVEL_MEDIUM:
+		return AMD_PRIORITY_LEVEL_MEDIUM;
+	case AMDGPU_MES_PRIORITY_LEVEL_HIGH:
+		return AMD_PRIORITY_LEVEL_HIGH;
+	case AMDGPU_MES_PRIORITY_LEVEL_REALTIME:
+		return AMD_PRIORITY_LEVEL_REALTIME;
+	}
+}
+
+static int mes_v11_0_add_hw_queue(struct amdgpu_mes *mes,
+				  struct mes_add_queue_input *input)
+{
+	struct amdgpu_device *adev = mes->adev;
+	union MESAPI__ADD_QUEUE mes_add_queue_pkt;
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	uint32_t vm_cntx_cntl = hub->vm_cntx_cntl;
+
+	memset(&mes_add_queue_pkt, 0, sizeof(mes_add_queue_pkt));
+
+	mes_add_queue_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_add_queue_pkt.header.opcode = MES_SCH_API_ADD_QUEUE;
+	mes_add_queue_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	mes_add_queue_pkt.process_id = input->process_id;
+	mes_add_queue_pkt.page_table_base_addr = input->page_table_base_addr;
+	mes_add_queue_pkt.process_va_start = input->process_va_start;
+	mes_add_queue_pkt.process_va_end = input->process_va_end;
+	mes_add_queue_pkt.process_quantum = input->process_quantum;
+	mes_add_queue_pkt.process_context_addr = input->process_context_addr;
+	mes_add_queue_pkt.gang_quantum = input->gang_quantum;
+	mes_add_queue_pkt.gang_context_addr = input->gang_context_addr;
+	mes_add_queue_pkt.inprocess_gang_priority =
+		convert_to_mes_priority_level(input->inprocess_gang_priority);
+	mes_add_queue_pkt.gang_global_priority_level =
+		convert_to_mes_priority_level(input->gang_global_priority_level);
+	mes_add_queue_pkt.doorbell_offset = input->doorbell_offset;
+	mes_add_queue_pkt.mqd_addr = input->mqd_addr;
+
+	if (((adev->mes.sched_version & AMDGPU_MES_API_VERSION_MASK) >>
+			AMDGPU_MES_API_VERSION_SHIFT) >= 2)
+		mes_add_queue_pkt.wptr_addr = input->wptr_mc_addr;
+	else
+		mes_add_queue_pkt.wptr_addr = input->wptr_addr;
+
+	mes_add_queue_pkt.queue_type =
+		convert_to_mes_queue_type(input->queue_type);
+	mes_add_queue_pkt.paging = input->paging;
+	mes_add_queue_pkt.vm_context_cntl = vm_cntx_cntl;
+	mes_add_queue_pkt.gws_base = input->gws_base;
+	mes_add_queue_pkt.gws_size = input->gws_size;
+	mes_add_queue_pkt.trap_handler_addr = input->tba_addr;
+	mes_add_queue_pkt.tma_addr = input->tma_addr;
+	mes_add_queue_pkt.trap_en = input->trap_en;
+	mes_add_queue_pkt.skip_process_ctx_clear = input->skip_process_ctx_clear;
+	mes_add_queue_pkt.is_kfd_process = input->is_kfd_process;
+
+	/* For KFD, gds_size is re-used for queue size (needed in MES for AQL queues) */
+	mes_add_queue_pkt.is_aql_queue = input->is_aql_queue;
+	mes_add_queue_pkt.gds_size = input->queue_size;
+
+	mes_add_queue_pkt.exclusively_scheduled = input->exclusively_scheduled;
+
+	return mes_v11_0_submit_pkt_and_poll_completion(mes,
+			&mes_add_queue_pkt, sizeof(mes_add_queue_pkt),
+			offsetof(union MESAPI__ADD_QUEUE, api_status));
+}
+
+static int mes_v11_0_remove_hw_queue(struct amdgpu_mes *mes,
+				     struct mes_remove_queue_input *input)
+{
+	union MESAPI__REMOVE_QUEUE mes_remove_queue_pkt;
+
+	memset(&mes_remove_queue_pkt, 0, sizeof(mes_remove_queue_pkt));
+
+	mes_remove_queue_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_remove_queue_pkt.header.opcode = MES_SCH_API_REMOVE_QUEUE;
+	mes_remove_queue_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	mes_remove_queue_pkt.doorbell_offset = input->doorbell_offset;
+	mes_remove_queue_pkt.gang_context_addr = input->gang_context_addr;
+
+	return mes_v11_0_submit_pkt_and_poll_completion(mes,
+			&mes_remove_queue_pkt, sizeof(mes_remove_queue_pkt),
+			offsetof(union MESAPI__REMOVE_QUEUE, api_status));
+}
+
+static int mes_v11_0_reset_queue_mmio(struct amdgpu_mes *mes, uint32_t queue_type,
+				      uint32_t me_id, uint32_t pipe_id,
+				      uint32_t queue_id, uint32_t vmid)
+{
+	struct amdgpu_device *adev = mes->adev;
+	uint32_t value, reg;
+	int i, r = 0;
+
+	amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
+
+	if (queue_type == AMDGPU_RING_TYPE_GFX) {
+		dev_info(adev->dev, "reset gfx queue (%d:%d:%d: vmid:%d)\n",
+			 me_id, pipe_id, queue_id, vmid);
+
+		mutex_lock(&adev->gfx.reset_sem_mutex);
+		gfx_v11_0_request_gfx_index_mutex(adev, true);
+		/* all se allow writes */
+		WREG32_SOC15(GC, 0, regGRBM_GFX_INDEX,
+			     (uint32_t)(0x1 << GRBM_GFX_INDEX__SE_BROADCAST_WRITES__SHIFT));
+		value = REG_SET_FIELD(0, CP_VMID_RESET, RESET_REQUEST, 1 << vmid);
+		if (pipe_id == 0)
+			value = REG_SET_FIELD(value, CP_VMID_RESET, PIPE0_QUEUES, 1 << queue_id);
+		else
+			value = REG_SET_FIELD(value, CP_VMID_RESET, PIPE1_QUEUES, 1 << queue_id);
+		WREG32_SOC15(GC, 0, regCP_VMID_RESET, value);
+		gfx_v11_0_request_gfx_index_mutex(adev, false);
+		mutex_unlock(&adev->gfx.reset_sem_mutex);
+
+		mutex_lock(&adev->srbm_mutex);
+		soc21_grbm_select(adev, me_id, pipe_id, queue_id, 0);
+		/* wait till dequeue take effects */
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (!(RREG32_SOC15(GC, 0, regCP_GFX_HQD_ACTIVE) & 1))
+				break;
+			udelay(1);
+		}
+		if (i >= adev->usec_timeout) {
+			dev_err(adev->dev, "failed to wait on gfx hqd deactivate\n");
+			r = -ETIMEDOUT;
+		}
+
+		soc21_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+	} else if (queue_type == AMDGPU_RING_TYPE_COMPUTE) {
+		dev_info(adev->dev, "reset compute queue (%d:%d:%d)\n",
+			 me_id, pipe_id, queue_id);
+		mutex_lock(&adev->srbm_mutex);
+		soc21_grbm_select(adev, me_id, pipe_id, queue_id, 0);
+		WREG32_SOC15(GC, 0, regCP_HQD_DEQUEUE_REQUEST, 0x2);
+		WREG32_SOC15(GC, 0, regSPI_COMPUTE_QUEUE_RESET, 0x1);
+
+		/* wait till dequeue take effects */
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (!(RREG32_SOC15(GC, 0, regCP_HQD_ACTIVE) & 1))
+				break;
+			udelay(1);
+		}
+		if (i >= adev->usec_timeout) {
+			dev_err(adev->dev, "failed to wait on hqd deactivate\n");
+			r = -ETIMEDOUT;
+		}
+		soc21_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+	} else if (queue_type == AMDGPU_RING_TYPE_SDMA) {
+		dev_info(adev->dev, "reset sdma queue (%d:%d:%d)\n",
+			 me_id, pipe_id, queue_id);
+		switch (me_id) {
+		case 1:
+			reg = SOC15_REG_OFFSET(GC, 0, regSDMA1_QUEUE_RESET_REQ);
+			break;
+		case 0:
+		default:
+			reg = SOC15_REG_OFFSET(GC, 0, regSDMA0_QUEUE_RESET_REQ);
+			break;
+		}
+
+		value = 1 << queue_id;
+		WREG32(reg, value);
+		/* wait for queue reset done */
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (!(RREG32(reg) & value))
+				break;
+			udelay(1);
+		}
+		if (i >= adev->usec_timeout) {
+			dev_err(adev->dev, "failed to wait on sdma queue reset done\n");
+			r = -ETIMEDOUT;
+		}
+	}
+
+	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
+	return r;
+}
+
+static int mes_v11_0_map_legacy_queue(struct amdgpu_mes *mes,
+				      struct mes_map_legacy_queue_input *input)
+{
+	union MESAPI__ADD_QUEUE mes_add_queue_pkt;
+
+	memset(&mes_add_queue_pkt, 0, sizeof(mes_add_queue_pkt));
+
+	mes_add_queue_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_add_queue_pkt.header.opcode = MES_SCH_API_ADD_QUEUE;
+	mes_add_queue_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	mes_add_queue_pkt.pipe_id = input->pipe_id;
+	mes_add_queue_pkt.queue_id = input->queue_id;
+	mes_add_queue_pkt.doorbell_offset = input->doorbell_offset;
+	mes_add_queue_pkt.mqd_addr = input->mqd_addr;
+	mes_add_queue_pkt.wptr_addr = input->wptr_addr;
+	mes_add_queue_pkt.queue_type =
+		convert_to_mes_queue_type(input->queue_type);
+	mes_add_queue_pkt.map_legacy_kq = 1;
+
+	return mes_v11_0_submit_pkt_and_poll_completion(mes,
+			&mes_add_queue_pkt, sizeof(mes_add_queue_pkt),
+			offsetof(union MESAPI__ADD_QUEUE, api_status));
+}
+
+static int mes_v11_0_unmap_legacy_queue(struct amdgpu_mes *mes,
+			struct mes_unmap_legacy_queue_input *input)
+{
+	union MESAPI__REMOVE_QUEUE mes_remove_queue_pkt;
+
+	memset(&mes_remove_queue_pkt, 0, sizeof(mes_remove_queue_pkt));
+
+	mes_remove_queue_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_remove_queue_pkt.header.opcode = MES_SCH_API_REMOVE_QUEUE;
+	mes_remove_queue_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	mes_remove_queue_pkt.doorbell_offset = input->doorbell_offset;
+	mes_remove_queue_pkt.gang_context_addr = 0;
+
+	mes_remove_queue_pkt.pipe_id = input->pipe_id;
+	mes_remove_queue_pkt.queue_id = input->queue_id;
+
+	if (input->action == PREEMPT_QUEUES_NO_UNMAP) {
+		mes_remove_queue_pkt.preempt_legacy_gfx_queue = 1;
+		mes_remove_queue_pkt.tf_addr = input->trail_fence_addr;
+		mes_remove_queue_pkt.tf_data =
+			lower_32_bits(input->trail_fence_data);
+	} else {
+		mes_remove_queue_pkt.unmap_legacy_queue = 1;
+		mes_remove_queue_pkt.queue_type =
+			convert_to_mes_queue_type(input->queue_type);
+	}
+
+	return mes_v11_0_submit_pkt_and_poll_completion(mes,
+			&mes_remove_queue_pkt, sizeof(mes_remove_queue_pkt),
+			offsetof(union MESAPI__REMOVE_QUEUE, api_status));
+}
+
+static int mes_v11_0_suspend_gang(struct amdgpu_mes *mes,
+				  struct mes_suspend_gang_input *input)
+{
+	union MESAPI__SUSPEND mes_suspend_gang_pkt;
+
+	memset(&mes_suspend_gang_pkt, 0, sizeof(mes_suspend_gang_pkt));
+
+	mes_suspend_gang_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_suspend_gang_pkt.header.opcode = MES_SCH_API_SUSPEND;
+	mes_suspend_gang_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	mes_suspend_gang_pkt.suspend_all_gangs = input->suspend_all_gangs;
+	mes_suspend_gang_pkt.gang_context_addr = input->gang_context_addr;
+	mes_suspend_gang_pkt.suspend_fence_addr = input->suspend_fence_addr;
+	mes_suspend_gang_pkt.suspend_fence_value = input->suspend_fence_value;
+
+	return mes_v11_0_submit_pkt_and_poll_completion(mes,
+			&mes_suspend_gang_pkt, sizeof(mes_suspend_gang_pkt),
+			offsetof(union MESAPI__SUSPEND, api_status));
+}
+
+static int mes_v11_0_resume_gang(struct amdgpu_mes *mes,
+				 struct mes_resume_gang_input *input)
+{
+	union MESAPI__RESUME mes_resume_gang_pkt;
+
+	memset(&mes_resume_gang_pkt, 0, sizeof(mes_resume_gang_pkt));
+
+	mes_resume_gang_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_resume_gang_pkt.header.opcode = MES_SCH_API_RESUME;
+	mes_resume_gang_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	mes_resume_gang_pkt.resume_all_gangs = input->resume_all_gangs;
+	mes_resume_gang_pkt.gang_context_addr = input->gang_context_addr;
+
+	return mes_v11_0_submit_pkt_and_poll_completion(mes,
+			&mes_resume_gang_pkt, sizeof(mes_resume_gang_pkt),
+			offsetof(union MESAPI__RESUME, api_status));
+}
+
+static int mes_v11_0_query_sched_status(struct amdgpu_mes *mes)
+{
+	union MESAPI__QUERY_MES_STATUS mes_status_pkt;
+
+	memset(&mes_status_pkt, 0, sizeof(mes_status_pkt));
+
+	mes_status_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_status_pkt.header.opcode = MES_SCH_API_QUERY_SCHEDULER_STATUS;
+	mes_status_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	return mes_v11_0_submit_pkt_and_poll_completion(mes,
+			&mes_status_pkt, sizeof(mes_status_pkt),
+			offsetof(union MESAPI__QUERY_MES_STATUS, api_status));
+}
+
+static int mes_v11_0_misc_op(struct amdgpu_mes *mes,
+			     struct mes_misc_op_input *input)
+{
+	union MESAPI__MISC misc_pkt;
+
+	memset(&misc_pkt, 0, sizeof(misc_pkt));
+
+	misc_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	misc_pkt.header.opcode = MES_SCH_API_MISC;
+	misc_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	switch (input->op) {
+	case MES_MISC_OP_READ_REG:
+		misc_pkt.opcode = MESAPI_MISC__READ_REG;
+		misc_pkt.read_reg.reg_offset = input->read_reg.reg_offset;
+		misc_pkt.read_reg.buffer_addr = input->read_reg.buffer_addr;
+		break;
+	case MES_MISC_OP_WRITE_REG:
+		misc_pkt.opcode = MESAPI_MISC__WRITE_REG;
+		misc_pkt.write_reg.reg_offset = input->write_reg.reg_offset;
+		misc_pkt.write_reg.reg_value = input->write_reg.reg_value;
+		break;
+	case MES_MISC_OP_WRM_REG_WAIT:
+		misc_pkt.opcode = MESAPI_MISC__WAIT_REG_MEM;
+		misc_pkt.wait_reg_mem.op = WRM_OPERATION__WAIT_REG_MEM;
+		misc_pkt.wait_reg_mem.reference = input->wrm_reg.ref;
+		misc_pkt.wait_reg_mem.mask = input->wrm_reg.mask;
+		misc_pkt.wait_reg_mem.reg_offset1 = input->wrm_reg.reg0;
+		misc_pkt.wait_reg_mem.reg_offset2 = 0;
+		break;
+	case MES_MISC_OP_WRM_REG_WR_WAIT:
+		misc_pkt.opcode = MESAPI_MISC__WAIT_REG_MEM;
+		misc_pkt.wait_reg_mem.op = WRM_OPERATION__WR_WAIT_WR_REG;
+		misc_pkt.wait_reg_mem.reference = input->wrm_reg.ref;
+		misc_pkt.wait_reg_mem.mask = input->wrm_reg.mask;
+		misc_pkt.wait_reg_mem.reg_offset1 = input->wrm_reg.reg0;
+		misc_pkt.wait_reg_mem.reg_offset2 = input->wrm_reg.reg1;
+		break;
+	case MES_MISC_OP_SET_SHADER_DEBUGGER:
+		misc_pkt.opcode = MESAPI_MISC__SET_SHADER_DEBUGGER;
+		misc_pkt.set_shader_debugger.process_context_addr =
+				input->set_shader_debugger.process_context_addr;
+		misc_pkt.set_shader_debugger.flags.u32all =
+				input->set_shader_debugger.flags.u32all;
+		misc_pkt.set_shader_debugger.spi_gdbg_per_vmid_cntl =
+				input->set_shader_debugger.spi_gdbg_per_vmid_cntl;
+		memcpy(misc_pkt.set_shader_debugger.tcp_watch_cntl,
+				input->set_shader_debugger.tcp_watch_cntl,
+				sizeof(misc_pkt.set_shader_debugger.tcp_watch_cntl));
+		misc_pkt.set_shader_debugger.trap_en = input->set_shader_debugger.trap_en;
+		break;
+	case MES_MISC_OP_CHANGE_CONFIG:
+		if ((mes->adev->mes.sched_version & AMDGPU_MES_VERSION_MASK) < 0x63) {
+			dev_warn_once(mes->adev->dev,
+				      "MES FW version must be larger than 0x63 to support limit single process feature.\n");
+			return 0;
+		}
+		misc_pkt.opcode = MESAPI_MISC__CHANGE_CONFIG;
+		misc_pkt.change_config.opcode =
+				MESAPI_MISC__CHANGE_CONFIG_OPTION_LIMIT_SINGLE_PROCESS;
+		misc_pkt.change_config.option.bits.limit_single_process =
+				input->change_config.option.limit_single_process;
+		break;
+
+	default:
+		DRM_ERROR("unsupported misc op (%d) \n", input->op);
+		return -EINVAL;
+	}
+
+	return mes_v11_0_submit_pkt_and_poll_completion(mes,
+			&misc_pkt, sizeof(misc_pkt),
+			offsetof(union MESAPI__MISC, api_status));
+}
+
+static int mes_v11_0_set_hw_resources(struct amdgpu_mes *mes)
+{
+	int i;
+	struct amdgpu_device *adev = mes->adev;
+	union MESAPI_SET_HW_RESOURCES mes_set_hw_res_pkt;
+
+	memset(&mes_set_hw_res_pkt, 0, sizeof(mes_set_hw_res_pkt));
+
+	mes_set_hw_res_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_set_hw_res_pkt.header.opcode = MES_SCH_API_SET_HW_RSRC;
+	mes_set_hw_res_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	mes_set_hw_res_pkt.vmid_mask_mmhub = mes->vmid_mask_mmhub;
+	mes_set_hw_res_pkt.vmid_mask_gfxhub = mes->vmid_mask_gfxhub;
+	mes_set_hw_res_pkt.gds_size = adev->gds.gds_size;
+	mes_set_hw_res_pkt.paging_vmid = 0;
+	mes_set_hw_res_pkt.g_sch_ctx_gpu_mc_ptr = mes->sch_ctx_gpu_addr[0];
+	mes_set_hw_res_pkt.query_status_fence_gpu_mc_ptr =
+		mes->query_status_fence_gpu_addr[0];
+
+	for (i = 0; i < MAX_COMPUTE_PIPES; i++)
+		mes_set_hw_res_pkt.compute_hqd_mask[i] =
+			mes->compute_hqd_mask[i];
+
+	for (i = 0; i < MAX_GFX_PIPES; i++)
+		mes_set_hw_res_pkt.gfx_hqd_mask[i] =
+			mes->gfx_hqd_mask[i];
+
+	for (i = 0; i < MAX_SDMA_PIPES; i++)
+		mes_set_hw_res_pkt.sdma_hqd_mask[i] = mes->sdma_hqd_mask[i];
+
+	for (i = 0; i < AMD_PRIORITY_NUM_LEVELS; i++)
+		mes_set_hw_res_pkt.aggregated_doorbells[i] =
+			mes->aggregated_doorbells[i];
+
+	for (i = 0; i < 5; i++) {
+		mes_set_hw_res_pkt.gc_base[i] = adev->reg_offset[GC_HWIP][0][i];
+		mes_set_hw_res_pkt.mmhub_base[i] =
+				adev->reg_offset[MMHUB_HWIP][0][i];
+		mes_set_hw_res_pkt.osssys_base[i] =
+		adev->reg_offset[OSSSYS_HWIP][0][i];
+	}
+
+	mes_set_hw_res_pkt.disable_reset = 1;
+	mes_set_hw_res_pkt.disable_mes_log = 1;
+	mes_set_hw_res_pkt.use_different_vmid_compute = 1;
+	mes_set_hw_res_pkt.enable_reg_active_poll = 1;
+	mes_set_hw_res_pkt.enable_level_process_quantum_check = 1;
+	mes_set_hw_res_pkt.oversubscription_timer = 50;
+	if ((mes->adev->mes.sched_version & AMDGPU_MES_VERSION_MASK) >= 0x7f)
+		mes_set_hw_res_pkt.enable_lr_compute_wa = 1;
+	else
+		dev_info_once(mes->adev->dev,
+			      "MES FW version must be >= 0x7f to enable LR compute workaround.\n");
+
+	if (amdgpu_mes_log_enable) {
+		mes_set_hw_res_pkt.enable_mes_event_int_logging = 1;
+		mes_set_hw_res_pkt.event_intr_history_gpu_mc_ptr =
+					mes->event_log_gpu_addr;
+	}
+
+	if (adev->enforce_isolation[0] == AMDGPU_ENFORCE_ISOLATION_ENABLE)
+		mes_set_hw_res_pkt.limit_single_process = 1;
+
+	return mes_v11_0_submit_pkt_and_poll_completion(mes,
+			&mes_set_hw_res_pkt, sizeof(mes_set_hw_res_pkt),
+			offsetof(union MESAPI_SET_HW_RESOURCES, api_status));
+}
+
+static int mes_v11_0_set_hw_resources_1(struct amdgpu_mes *mes)
+{
+	union MESAPI_SET_HW_RESOURCES_1 mes_set_hw_res_pkt;
+	memset(&mes_set_hw_res_pkt, 0, sizeof(mes_set_hw_res_pkt));
+
+	mes_set_hw_res_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_set_hw_res_pkt.header.opcode = MES_SCH_API_SET_HW_RSRC_1;
+	mes_set_hw_res_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+	mes_set_hw_res_pkt.enable_mes_info_ctx = 1;
+
+	mes_set_hw_res_pkt.cleaner_shader_fence_mc_addr = mes->resource_1_gpu_addr[0];
+	if (amdgpu_sriov_is_mes_info_enable(mes->adev)) {
+		mes_set_hw_res_pkt.mes_info_ctx_mc_addr =
+			mes->resource_1_gpu_addr[0] + AMDGPU_GPU_PAGE_SIZE;
+		mes_set_hw_res_pkt.mes_info_ctx_size = MES11_HW_RESOURCE_1_SIZE;
+	}
+
+	return mes_v11_0_submit_pkt_and_poll_completion(mes,
+			&mes_set_hw_res_pkt, sizeof(mes_set_hw_res_pkt),
+			offsetof(union MESAPI_SET_HW_RESOURCES_1, api_status));
+}
+
+static int mes_v11_0_reset_hw_queue(struct amdgpu_mes *mes,
+				    struct mes_reset_queue_input *input)
+{
+	union MESAPI__RESET mes_reset_queue_pkt;
+
+	if (input->use_mmio)
+		return mes_v11_0_reset_queue_mmio(mes, input->queue_type,
+						  input->me_id, input->pipe_id,
+						  input->queue_id, input->vmid);
+
+	memset(&mes_reset_queue_pkt, 0, sizeof(mes_reset_queue_pkt));
+
+	mes_reset_queue_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_reset_queue_pkt.header.opcode = MES_SCH_API_RESET;
+	mes_reset_queue_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	mes_reset_queue_pkt.queue_type =
+		convert_to_mes_queue_type(input->queue_type);
+
+	if (input->legacy_gfx) {
+		mes_reset_queue_pkt.reset_legacy_gfx = 1;
+		mes_reset_queue_pkt.pipe_id_lp = input->pipe_id;
+		mes_reset_queue_pkt.queue_id_lp = input->queue_id;
+		mes_reset_queue_pkt.mqd_mc_addr_lp = input->mqd_addr;
+		mes_reset_queue_pkt.doorbell_offset_lp = input->doorbell_offset;
+		mes_reset_queue_pkt.wptr_addr_lp = input->wptr_addr;
+		mes_reset_queue_pkt.vmid_id_lp = input->vmid;
+	} else {
+		mes_reset_queue_pkt.reset_queue_only = 1;
+		mes_reset_queue_pkt.doorbell_offset = input->doorbell_offset;
+	}
+
+	return mes_v11_0_submit_pkt_and_poll_completion(mes,
+			&mes_reset_queue_pkt, sizeof(mes_reset_queue_pkt),
+			offsetof(union MESAPI__RESET, api_status));
+}
+
+static int mes_v11_0_detect_and_reset_hung_queues(struct amdgpu_mes *mes,
+						  struct mes_detect_and_reset_queue_input *input)
+{
+	union MESAPI__RESET mes_reset_queue_pkt;
+
+	memset(&mes_reset_queue_pkt, 0, sizeof(mes_reset_queue_pkt));
+
+	mes_reset_queue_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_reset_queue_pkt.header.opcode = MES_SCH_API_RESET;
+	mes_reset_queue_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	mes_reset_queue_pkt.queue_type =
+		convert_to_mes_queue_type(input->queue_type);
+	mes_reset_queue_pkt.doorbell_offset_addr =
+		mes->hung_queue_db_array_gpu_addr;
+
+	if (input->detect_only)
+		mes_reset_queue_pkt.hang_detect_only = 1;
+	else
+		mes_reset_queue_pkt.hang_detect_then_reset = 1;
+
+	return mes_v11_0_submit_pkt_and_poll_completion(mes,
+			&mes_reset_queue_pkt, sizeof(mes_reset_queue_pkt),
+			offsetof(union MESAPI__RESET, api_status));
+}
+
+static const struct amdgpu_mes_funcs mes_v11_0_funcs = {
+	.add_hw_queue = mes_v11_0_add_hw_queue,
+	.remove_hw_queue = mes_v11_0_remove_hw_queue,
+	.map_legacy_queue = mes_v11_0_map_legacy_queue,
+	.unmap_legacy_queue = mes_v11_0_unmap_legacy_queue,
+	.suspend_gang = mes_v11_0_suspend_gang,
+	.resume_gang = mes_v11_0_resume_gang,
+	.misc_op = mes_v11_0_misc_op,
+	.reset_hw_queue = mes_v11_0_reset_hw_queue,
+	.detect_and_reset_hung_queues = mes_v11_0_detect_and_reset_hung_queues,
+};
+
+static int mes_v11_0_allocate_ucode_buffer(struct amdgpu_device *adev,
+					   enum amdgpu_mes_pipe pipe)
+{
+	int r;
+	const struct mes_firmware_header_v1_0 *mes_hdr;
+	const __le32 *fw_data;
+	unsigned fw_size;
+
+	mes_hdr = (const struct mes_firmware_header_v1_0 *)
+		adev->mes.fw[pipe]->data;
+
+	fw_data = (const __le32 *)(adev->mes.fw[pipe]->data +
+		   le32_to_cpu(mes_hdr->mes_ucode_offset_bytes));
+	fw_size = le32_to_cpu(mes_hdr->mes_ucode_size_bytes);
+
+	r = amdgpu_bo_create_reserved(adev, fw_size,
+				      PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_VRAM |
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->mes.ucode_fw_obj[pipe],
+				      &adev->mes.ucode_fw_gpu_addr[pipe],
+				      (void **)&adev->mes.ucode_fw_ptr[pipe]);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create mes fw bo\n", r);
+		return r;
+	}
+
+	memcpy(adev->mes.ucode_fw_ptr[pipe], fw_data, fw_size);
+
+	amdgpu_bo_kunmap(adev->mes.ucode_fw_obj[pipe]);
+	amdgpu_bo_unreserve(adev->mes.ucode_fw_obj[pipe]);
+
+	return 0;
+}
+
+static int mes_v11_0_allocate_ucode_data_buffer(struct amdgpu_device *adev,
+						enum amdgpu_mes_pipe pipe)
+{
+	int r;
+	const struct mes_firmware_header_v1_0 *mes_hdr;
+	const __le32 *fw_data;
+	unsigned fw_size;
+
+	mes_hdr = (const struct mes_firmware_header_v1_0 *)
+		adev->mes.fw[pipe]->data;
+
+	fw_data = (const __le32 *)(adev->mes.fw[pipe]->data +
+		   le32_to_cpu(mes_hdr->mes_ucode_data_offset_bytes));
+	fw_size = le32_to_cpu(mes_hdr->mes_ucode_data_size_bytes);
+
+	if (fw_size > GFX_MES_DRAM_SIZE) {
+		dev_err(adev->dev, "PIPE%d ucode data fw size (%d) is greater than dram size (%d)\n",
+			pipe, fw_size, GFX_MES_DRAM_SIZE);
+		return -EINVAL;
+	}
+
+	r = amdgpu_bo_create_reserved(adev, GFX_MES_DRAM_SIZE,
+				      64 * 1024,
+				      AMDGPU_GEM_DOMAIN_VRAM |
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->mes.data_fw_obj[pipe],
+				      &adev->mes.data_fw_gpu_addr[pipe],
+				      (void **)&adev->mes.data_fw_ptr[pipe]);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create mes data fw bo\n", r);
+		return r;
+	}
+
+	memcpy(adev->mes.data_fw_ptr[pipe], fw_data, fw_size);
+
+	amdgpu_bo_kunmap(adev->mes.data_fw_obj[pipe]);
+	amdgpu_bo_unreserve(adev->mes.data_fw_obj[pipe]);
+
+	return 0;
+}
+
+static void mes_v11_0_free_ucode_buffers(struct amdgpu_device *adev,
+					 enum amdgpu_mes_pipe pipe)
+{
+	amdgpu_bo_free_kernel(&adev->mes.data_fw_obj[pipe],
+			      &adev->mes.data_fw_gpu_addr[pipe],
+			      (void **)&adev->mes.data_fw_ptr[pipe]);
+
+	amdgpu_bo_free_kernel(&adev->mes.ucode_fw_obj[pipe],
+			      &adev->mes.ucode_fw_gpu_addr[pipe],
+			      (void **)&adev->mes.ucode_fw_ptr[pipe]);
+}
+
+static void mes_v11_0_get_fw_version(struct amdgpu_device *adev)
+{
+	int pipe;
+
+	/* return early if we have already fetched these */
+	if (adev->mes.sched_version && adev->mes.kiq_version)
+		return;
+
+	/* get MES scheduler/KIQ versions */
+	mutex_lock(&adev->srbm_mutex);
+
+	for (pipe = 0; pipe < AMDGPU_MAX_MES_PIPES; pipe++) {
+		soc21_grbm_select(adev, 3, pipe, 0, 0);
+
+		if (pipe == AMDGPU_MES_SCHED_PIPE)
+			adev->mes.sched_version =
+				RREG32_SOC15(GC, 0, regCP_MES_GP3_LO);
+		else if (pipe == AMDGPU_MES_KIQ_PIPE && adev->enable_mes_kiq)
+			adev->mes.kiq_version =
+				RREG32_SOC15(GC, 0, regCP_MES_GP3_LO);
+	}
+
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static void mes_v11_0_enable(struct amdgpu_device *adev, bool enable)
+{
+	uint64_t ucode_addr;
+	uint32_t pipe, data = 0;
+
+	if (enable) {
+		if (amdgpu_mes_log_enable) {
+			WREG32_SOC15(GC, 0, regCP_MES_MSCRATCH_LO,
+				lower_32_bits(adev->mes.event_log_gpu_addr + AMDGPU_MES_LOG_BUFFER_SIZE));
+			WREG32_SOC15(GC, 0, regCP_MES_MSCRATCH_HI,
+				upper_32_bits(adev->mes.event_log_gpu_addr + AMDGPU_MES_LOG_BUFFER_SIZE));
+			dev_info(adev->dev, "Setup CP MES MSCRATCH address : 0x%x. 0x%x\n",
+				RREG32_SOC15(GC, 0, regCP_MES_MSCRATCH_HI),
+				RREG32_SOC15(GC, 0, regCP_MES_MSCRATCH_LO));
+		}
+
+		data = RREG32_SOC15(GC, 0, regCP_MES_CNTL);
+		data = REG_SET_FIELD(data, CP_MES_CNTL, MES_PIPE0_RESET, 1);
+		data = REG_SET_FIELD(data, CP_MES_CNTL,
+			     MES_PIPE1_RESET, adev->enable_mes_kiq ? 1 : 0);
+		WREG32_SOC15(GC, 0, regCP_MES_CNTL, data);
+
+		mutex_lock(&adev->srbm_mutex);
+		for (pipe = 0; pipe < AMDGPU_MAX_MES_PIPES; pipe++) {
+			if (!adev->enable_mes_kiq &&
+			    pipe == AMDGPU_MES_KIQ_PIPE)
+				continue;
+
+			soc21_grbm_select(adev, 3, pipe, 0, 0);
+
+			ucode_addr = adev->mes.uc_start_addr[pipe] >> 2;
+			WREG32_SOC15(GC, 0, regCP_MES_PRGRM_CNTR_START,
+				     lower_32_bits(ucode_addr));
+			WREG32_SOC15(GC, 0, regCP_MES_PRGRM_CNTR_START_HI,
+				     upper_32_bits(ucode_addr));
+		}
+		soc21_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+
+		/* unhalt MES and activate pipe0 */
+		data = REG_SET_FIELD(0, CP_MES_CNTL, MES_PIPE0_ACTIVE, 1);
+		data = REG_SET_FIELD(data, CP_MES_CNTL, MES_PIPE1_ACTIVE,
+				     adev->enable_mes_kiq ? 1 : 0);
+		WREG32_SOC15(GC, 0, regCP_MES_CNTL, data);
+
+		if (amdgpu_emu_mode)
+			msleep(100);
+		else
+			udelay(500);
+	} else {
+		data = RREG32_SOC15(GC, 0, regCP_MES_CNTL);
+		data = REG_SET_FIELD(data, CP_MES_CNTL, MES_PIPE0_ACTIVE, 0);
+		data = REG_SET_FIELD(data, CP_MES_CNTL, MES_PIPE1_ACTIVE, 0);
+		data = REG_SET_FIELD(data, CP_MES_CNTL,
+				     MES_INVALIDATE_ICACHE, 1);
+		data = REG_SET_FIELD(data, CP_MES_CNTL, MES_PIPE0_RESET, 1);
+		data = REG_SET_FIELD(data, CP_MES_CNTL, MES_PIPE1_RESET,
+				     adev->enable_mes_kiq ? 1 : 0);
+		data = REG_SET_FIELD(data, CP_MES_CNTL, MES_HALT, 1);
+		WREG32_SOC15(GC, 0, regCP_MES_CNTL, data);
+	}
+}
+
+/* This function is for backdoor MES firmware */
+static int mes_v11_0_load_microcode(struct amdgpu_device *adev,
+				    enum amdgpu_mes_pipe pipe, bool prime_icache)
+{
+	int r;
+	uint32_t data;
+	uint64_t ucode_addr;
+
+	mes_v11_0_enable(adev, false);
+
+	if (!adev->mes.fw[pipe])
+		return -EINVAL;
+
+	r = mes_v11_0_allocate_ucode_buffer(adev, pipe);
+	if (r)
+		return r;
+
+	r = mes_v11_0_allocate_ucode_data_buffer(adev, pipe);
+	if (r) {
+		mes_v11_0_free_ucode_buffers(adev, pipe);
+		return r;
+	}
+
+	mutex_lock(&adev->srbm_mutex);
+	/* me=3, pipe=0, queue=0 */
+	soc21_grbm_select(adev, 3, pipe, 0, 0);
+
+	WREG32_SOC15(GC, 0, regCP_MES_IC_BASE_CNTL, 0);
+
+	/* set ucode start address */
+	ucode_addr = adev->mes.uc_start_addr[pipe] >> 2;
+	WREG32_SOC15(GC, 0, regCP_MES_PRGRM_CNTR_START,
+		     lower_32_bits(ucode_addr));
+	WREG32_SOC15(GC, 0, regCP_MES_PRGRM_CNTR_START_HI,
+		     upper_32_bits(ucode_addr));
+
+	/* set ucode fimrware address */
+	WREG32_SOC15(GC, 0, regCP_MES_IC_BASE_LO,
+		     lower_32_bits(adev->mes.ucode_fw_gpu_addr[pipe]));
+	WREG32_SOC15(GC, 0, regCP_MES_IC_BASE_HI,
+		     upper_32_bits(adev->mes.ucode_fw_gpu_addr[pipe]));
+
+	/* set ucode instruction cache boundary to 2M-1 */
+	WREG32_SOC15(GC, 0, regCP_MES_MIBOUND_LO, 0x1FFFFF);
+
+	/* set ucode data firmware address */
+	WREG32_SOC15(GC, 0, regCP_MES_MDBASE_LO,
+		     lower_32_bits(adev->mes.data_fw_gpu_addr[pipe]));
+	WREG32_SOC15(GC, 0, regCP_MES_MDBASE_HI,
+		     upper_32_bits(adev->mes.data_fw_gpu_addr[pipe]));
+
+	/* Set 0x7FFFF (512K-1) to CP_MES_MDBOUND_LO */
+	WREG32_SOC15(GC, 0, regCP_MES_MDBOUND_LO, 0x7FFFF);
+
+	if (prime_icache) {
+		/* invalidate ICACHE */
+		data = RREG32_SOC15(GC, 0, regCP_MES_IC_OP_CNTL);
+		data = REG_SET_FIELD(data, CP_MES_IC_OP_CNTL, PRIME_ICACHE, 0);
+		data = REG_SET_FIELD(data, CP_MES_IC_OP_CNTL, INVALIDATE_CACHE, 1);
+		WREG32_SOC15(GC, 0, regCP_MES_IC_OP_CNTL, data);
+
+		/* prime the ICACHE. */
+		data = RREG32_SOC15(GC, 0, regCP_MES_IC_OP_CNTL);
+		data = REG_SET_FIELD(data, CP_MES_IC_OP_CNTL, PRIME_ICACHE, 1);
+		WREG32_SOC15(GC, 0, regCP_MES_IC_OP_CNTL, data);
+	}
+
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	return 0;
+}
+
+static int mes_v11_0_allocate_eop_buf(struct amdgpu_device *adev,
+				      enum amdgpu_mes_pipe pipe)
+{
+	int r;
+	u32 *eop;
+
+	r = amdgpu_bo_create_reserved(adev, MES_EOP_SIZE, PAGE_SIZE,
+			      AMDGPU_GEM_DOMAIN_GTT,
+			      &adev->mes.eop_gpu_obj[pipe],
+			      &adev->mes.eop_gpu_addr[pipe],
+			      (void **)&eop);
+	if (r) {
+		dev_warn(adev->dev, "(%d) create EOP bo failed\n", r);
+		return r;
+	}
+
+	memset(eop, 0,
+	       adev->mes.eop_gpu_obj[pipe]->tbo.base.size);
+
+	amdgpu_bo_kunmap(adev->mes.eop_gpu_obj[pipe]);
+	amdgpu_bo_unreserve(adev->mes.eop_gpu_obj[pipe]);
+
+	return 0;
+}
+
+static int mes_v11_0_mqd_init(struct amdgpu_ring *ring)
+{
+	struct v11_compute_mqd *mqd = ring->mqd_ptr;
+	uint64_t hqd_gpu_addr, wb_gpu_addr, eop_base_addr;
+	uint32_t tmp;
+
+	memset(mqd, 0, sizeof(*mqd));
+
+	mqd->header = 0xC0310800;
+	mqd->compute_pipelinestat_enable = 0x00000001;
+	mqd->compute_static_thread_mgmt_se0 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se1 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se2 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se3 = 0xffffffff;
+	mqd->compute_misc_reserved = 0x00000007;
+
+	eop_base_addr = ring->eop_gpu_addr >> 8;
+
+	/* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */
+	tmp = regCP_HQD_EOP_CONTROL_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_HQD_EOP_CONTROL, EOP_SIZE,
+			(order_base_2(MES_EOP_SIZE / 4) - 1));
+
+	mqd->cp_hqd_eop_base_addr_lo = lower_32_bits(eop_base_addr);
+	mqd->cp_hqd_eop_base_addr_hi = upper_32_bits(eop_base_addr);
+	mqd->cp_hqd_eop_control = tmp;
+
+	/* disable the queue if it's active */
+	ring->wptr = 0;
+	mqd->cp_hqd_pq_rptr = 0;
+	mqd->cp_hqd_pq_wptr_lo = 0;
+	mqd->cp_hqd_pq_wptr_hi = 0;
+
+	/* set the pointer to the MQD */
+	mqd->cp_mqd_base_addr_lo = ring->mqd_gpu_addr & 0xfffffffc;
+	mqd->cp_mqd_base_addr_hi = upper_32_bits(ring->mqd_gpu_addr);
+
+	/* set MQD vmid to 0 */
+	tmp = regCP_MQD_CONTROL_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_MQD_CONTROL, VMID, 0);
+	mqd->cp_mqd_control = tmp;
+
+	/* set the pointer to the HQD, this is similar CP_RB0_BASE/_HI */
+	hqd_gpu_addr = ring->gpu_addr >> 8;
+	mqd->cp_hqd_pq_base_lo = lower_32_bits(hqd_gpu_addr);
+	mqd->cp_hqd_pq_base_hi = upper_32_bits(hqd_gpu_addr);
+
+	/* set the wb address whether it's enabled or not */
+	wb_gpu_addr = ring->rptr_gpu_addr;
+	mqd->cp_hqd_pq_rptr_report_addr_lo = wb_gpu_addr & 0xfffffffc;
+	mqd->cp_hqd_pq_rptr_report_addr_hi =
+		upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	/* only used if CP_PQ_WPTR_POLL_CNTL.CP_PQ_WPTR_POLL_CNTL__EN_MASK=1 */
+	wb_gpu_addr = ring->wptr_gpu_addr;
+	mqd->cp_hqd_pq_wptr_poll_addr_lo = wb_gpu_addr & 0xfffffff8;
+	mqd->cp_hqd_pq_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	/* set up the HQD, this is similar to CP_RB0_CNTL */
+	tmp = regCP_HQD_PQ_CONTROL_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, QUEUE_SIZE,
+			    (order_base_2(ring->ring_size / 4) - 1));
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, RPTR_BLOCK_SIZE,
+			    ((order_base_2(AMDGPU_GPU_PAGE_SIZE / 4) - 1) << 8));
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, UNORD_DISPATCH, 1);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, TUNNEL_DISPATCH, 0);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, PRIV_STATE, 1);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, KMD_QUEUE, 1);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, NO_UPDATE_RPTR, 1);
+	mqd->cp_hqd_pq_control = tmp;
+
+	/* enable doorbell */
+	tmp = 0;
+	if (ring->use_doorbell) {
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_OFFSET, ring->doorbell_index);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_EN, 1);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_SOURCE, 0);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_HIT, 0);
+	} else
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_EN, 0);
+	mqd->cp_hqd_pq_doorbell_control = tmp;
+
+	mqd->cp_hqd_vmid = 0;
+	/* activate the queue */
+	mqd->cp_hqd_active = 1;
+
+	tmp = regCP_HQD_PERSISTENT_STATE_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PERSISTENT_STATE,
+			    PRELOAD_SIZE, 0x55);
+	mqd->cp_hqd_persistent_state = tmp;
+
+	mqd->cp_hqd_ib_control = regCP_HQD_IB_CONTROL_DEFAULT;
+	mqd->cp_hqd_iq_timer = regCP_HQD_IQ_TIMER_DEFAULT;
+	mqd->cp_hqd_quantum = regCP_HQD_QUANTUM_DEFAULT;
+
+	amdgpu_device_flush_hdp(ring->adev, NULL);
+	return 0;
+}
+
+static void mes_v11_0_queue_init_register(struct amdgpu_ring *ring)
+{
+	struct v11_compute_mqd *mqd = ring->mqd_ptr;
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t data = 0;
+
+	mutex_lock(&adev->srbm_mutex);
+	soc21_grbm_select(adev, 3, ring->pipe, 0, 0);
+
+	/* set CP_HQD_VMID.VMID = 0. */
+	data = RREG32_SOC15(GC, 0, regCP_HQD_VMID);
+	data = REG_SET_FIELD(data, CP_HQD_VMID, VMID, 0);
+	WREG32_SOC15(GC, 0, regCP_HQD_VMID, data);
+
+	/* set CP_HQD_PQ_DOORBELL_CONTROL.DOORBELL_EN=0 */
+	data = RREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL);
+	data = REG_SET_FIELD(data, CP_HQD_PQ_DOORBELL_CONTROL,
+			     DOORBELL_EN, 0);
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL, data);
+
+	/* set CP_MQD_BASE_ADDR/HI with the MQD base address */
+	WREG32_SOC15(GC, 0, regCP_MQD_BASE_ADDR, mqd->cp_mqd_base_addr_lo);
+	WREG32_SOC15(GC, 0, regCP_MQD_BASE_ADDR_HI, mqd->cp_mqd_base_addr_hi);
+
+	/* set CP_MQD_CONTROL.VMID=0 */
+	data = RREG32_SOC15(GC, 0, regCP_MQD_CONTROL);
+	data = REG_SET_FIELD(data, CP_MQD_CONTROL, VMID, 0);
+	WREG32_SOC15(GC, 0, regCP_MQD_CONTROL, 0);
+
+	/* set CP_HQD_PQ_BASE/HI with the ring buffer base address */
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_BASE, mqd->cp_hqd_pq_base_lo);
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_BASE_HI, mqd->cp_hqd_pq_base_hi);
+
+	/* set CP_HQD_PQ_RPTR_REPORT_ADDR/HI */
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_RPTR_REPORT_ADDR,
+		     mqd->cp_hqd_pq_rptr_report_addr_lo);
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_RPTR_REPORT_ADDR_HI,
+		     mqd->cp_hqd_pq_rptr_report_addr_hi);
+
+	/* set CP_HQD_PQ_CONTROL */
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_CONTROL, mqd->cp_hqd_pq_control);
+
+	/* set CP_HQD_PQ_WPTR_POLL_ADDR/HI */
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_WPTR_POLL_ADDR,
+		     mqd->cp_hqd_pq_wptr_poll_addr_lo);
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_WPTR_POLL_ADDR_HI,
+		     mqd->cp_hqd_pq_wptr_poll_addr_hi);
+
+	/* set CP_HQD_PQ_DOORBELL_CONTROL */
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL,
+		     mqd->cp_hqd_pq_doorbell_control);
+
+	/* set CP_HQD_PERSISTENT_STATE.PRELOAD_SIZE=0x53 */
+	WREG32_SOC15(GC, 0, regCP_HQD_PERSISTENT_STATE, mqd->cp_hqd_persistent_state);
+
+	/* set CP_HQD_ACTIVE.ACTIVE=1 */
+	WREG32_SOC15(GC, 0, regCP_HQD_ACTIVE, mqd->cp_hqd_active);
+
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static int mes_v11_0_kiq_enable_queue(struct amdgpu_device *adev)
+{
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[0];
+	struct amdgpu_ring *kiq_ring = &adev->gfx.kiq[0].ring;
+	int r;
+
+	if (!kiq->pmf || !kiq->pmf->kiq_map_queues)
+		return -EINVAL;
+
+	r = amdgpu_ring_alloc(kiq_ring, kiq->pmf->map_queues_size);
+	if (r) {
+		DRM_ERROR("Failed to lock KIQ (%d).\n", r);
+		return r;
+	}
+
+	kiq->pmf->kiq_map_queues(kiq_ring, &adev->mes.ring[0]);
+
+	return amdgpu_ring_test_helper(kiq_ring);
+}
+
+static int mes_v11_0_queue_init(struct amdgpu_device *adev,
+				enum amdgpu_mes_pipe pipe)
+{
+	struct amdgpu_ring *ring;
+	int r;
+
+	if (pipe == AMDGPU_MES_KIQ_PIPE)
+		ring = &adev->gfx.kiq[0].ring;
+	else if (pipe == AMDGPU_MES_SCHED_PIPE)
+		ring = &adev->mes.ring[0];
+	else
+		BUG();
+
+	if ((pipe == AMDGPU_MES_SCHED_PIPE) &&
+	    (amdgpu_in_reset(adev) || adev->in_suspend)) {
+		*(ring->wptr_cpu_addr) = 0;
+		*(ring->rptr_cpu_addr) = 0;
+		amdgpu_ring_clear_ring(ring);
+	}
+
+	r = mes_v11_0_mqd_init(ring);
+	if (r)
+		return r;
+
+	if (pipe == AMDGPU_MES_SCHED_PIPE) {
+		r = mes_v11_0_kiq_enable_queue(adev);
+		if (r)
+			return r;
+	} else {
+		mes_v11_0_queue_init_register(ring);
+	}
+
+	return 0;
+}
+
+static int mes_v11_0_ring_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+
+	ring = &adev->mes.ring[0];
+
+	ring->funcs = &mes_v11_0_ring_funcs;
+
+	ring->me = 3;
+	ring->pipe = 0;
+	ring->queue = 0;
+
+	ring->ring_obj = NULL;
+	ring->use_doorbell = true;
+	ring->doorbell_index = adev->doorbell_index.mes_ring0 << 1;
+	ring->eop_gpu_addr = adev->mes.eop_gpu_addr[AMDGPU_MES_SCHED_PIPE];
+	ring->no_scheduler = true;
+	sprintf(ring->name, "mes_%d.%d.%d", ring->me, ring->pipe, ring->queue);
+
+	return amdgpu_ring_init(adev, ring, 1024, NULL, 0,
+				AMDGPU_RING_PRIO_DEFAULT, NULL);
+}
+
+static int mes_v11_0_kiq_ring_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+
+	spin_lock_init(&adev->gfx.kiq[0].ring_lock);
+
+	ring = &adev->gfx.kiq[0].ring;
+
+	ring->me = 3;
+	ring->pipe = 1;
+	ring->queue = 0;
+
+	ring->adev = NULL;
+	ring->ring_obj = NULL;
+	ring->use_doorbell = true;
+	ring->doorbell_index = adev->doorbell_index.mes_ring1 << 1;
+	ring->eop_gpu_addr = adev->mes.eop_gpu_addr[AMDGPU_MES_KIQ_PIPE];
+	ring->no_scheduler = true;
+	sprintf(ring->name, "mes_kiq_%d.%d.%d",
+		ring->me, ring->pipe, ring->queue);
+
+	return amdgpu_ring_init(adev, ring, 1024, NULL, 0,
+				AMDGPU_RING_PRIO_DEFAULT, NULL);
+}
+
+static int mes_v11_0_mqd_sw_init(struct amdgpu_device *adev,
+				 enum amdgpu_mes_pipe pipe)
+{
+	int r, mqd_size = sizeof(struct v11_compute_mqd);
+	struct amdgpu_ring *ring;
+
+	if (pipe == AMDGPU_MES_KIQ_PIPE)
+		ring = &adev->gfx.kiq[0].ring;
+	else if (pipe == AMDGPU_MES_SCHED_PIPE)
+		ring = &adev->mes.ring[0];
+	else
+		BUG();
+
+	if (ring->mqd_obj)
+		return 0;
+
+	r = amdgpu_bo_create_kernel(adev, mqd_size, PAGE_SIZE,
+				    AMDGPU_GEM_DOMAIN_VRAM |
+				    AMDGPU_GEM_DOMAIN_GTT, &ring->mqd_obj,
+				    &ring->mqd_gpu_addr, &ring->mqd_ptr);
+	if (r) {
+		dev_warn(adev->dev, "failed to create ring mqd bo (%d)", r);
+		return r;
+	}
+
+	memset(ring->mqd_ptr, 0, mqd_size);
+
+	/* prepare MQD backup */
+	adev->mes.mqd_backup[pipe] = kmalloc(mqd_size, GFP_KERNEL);
+	if (!adev->mes.mqd_backup[pipe]) {
+		dev_warn(adev->dev,
+			 "no memory to create MQD backup for ring %s\n",
+			 ring->name);
+		return -ENOMEM;
+	}
+
+	return 0;
+}
+
+static int mes_v11_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int pipe, r, bo_size;
+
+	adev->mes.funcs = &mes_v11_0_funcs;
+	adev->mes.kiq_hw_init = &mes_v11_0_kiq_hw_init;
+	adev->mes.kiq_hw_fini = &mes_v11_0_kiq_hw_fini;
+
+	adev->mes.event_log_size = AMDGPU_MES_LOG_BUFFER_SIZE + AMDGPU_MES_MSCRATCH_SIZE;
+
+	r = amdgpu_mes_init(adev);
+	if (r)
+		return r;
+
+	for (pipe = 0; pipe < AMDGPU_MAX_MES_PIPES; pipe++) {
+		if (!adev->enable_mes_kiq && pipe == AMDGPU_MES_KIQ_PIPE)
+			continue;
+
+		r = mes_v11_0_allocate_eop_buf(adev, pipe);
+		if (r)
+			return r;
+
+		r = mes_v11_0_mqd_sw_init(adev, pipe);
+		if (r)
+			return r;
+	}
+
+	if (adev->enable_mes_kiq) {
+		r = mes_v11_0_kiq_ring_init(adev);
+		if (r)
+			return r;
+	}
+
+	r = mes_v11_0_ring_init(adev);
+	if (r)
+		return r;
+
+	bo_size = AMDGPU_GPU_PAGE_SIZE;
+	if (amdgpu_sriov_is_mes_info_enable(adev))
+		bo_size += MES11_HW_RESOURCE_1_SIZE;
+
+	/* Only needed for AMDGPU_MES_SCHED_PIPE on MES 11*/
+	r = amdgpu_bo_create_kernel(adev,
+				    bo_size,
+				    PAGE_SIZE,
+				    AMDGPU_GEM_DOMAIN_VRAM,
+				    &adev->mes.resource_1[0],
+				    &adev->mes.resource_1_gpu_addr[0],
+				    &adev->mes.resource_1_addr[0]);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create mes resource_1 bo\n", r);
+		return r;
+	}
+
+	return 0;
+}
+
+static int mes_v11_0_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int pipe;
+
+	amdgpu_bo_free_kernel(&adev->mes.resource_1[0], &adev->mes.resource_1_gpu_addr[0],
+			      &adev->mes.resource_1_addr[0]);
+
+	for (pipe = 0; pipe < AMDGPU_MAX_MES_PIPES; pipe++) {
+		kfree(adev->mes.mqd_backup[pipe]);
+
+		amdgpu_bo_free_kernel(&adev->mes.eop_gpu_obj[pipe],
+				      &adev->mes.eop_gpu_addr[pipe],
+				      NULL);
+		amdgpu_ucode_release(&adev->mes.fw[pipe]);
+	}
+
+	amdgpu_bo_free_kernel(&adev->gfx.kiq[0].ring.mqd_obj,
+			      &adev->gfx.kiq[0].ring.mqd_gpu_addr,
+			      &adev->gfx.kiq[0].ring.mqd_ptr);
+
+	amdgpu_bo_free_kernel(&adev->mes.ring[0].mqd_obj,
+			      &adev->mes.ring[0].mqd_gpu_addr,
+			      &adev->mes.ring[0].mqd_ptr);
+
+	amdgpu_ring_fini(&adev->gfx.kiq[0].ring);
+	amdgpu_ring_fini(&adev->mes.ring[0]);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+		mes_v11_0_free_ucode_buffers(adev, AMDGPU_MES_KIQ_PIPE);
+		mes_v11_0_free_ucode_buffers(adev, AMDGPU_MES_SCHED_PIPE);
+	}
+
+	amdgpu_mes_fini(adev);
+	return 0;
+}
+
+static void mes_v11_0_kiq_dequeue(struct amdgpu_ring *ring)
+{
+	uint32_t data;
+	int i;
+	struct amdgpu_device *adev = ring->adev;
+
+	mutex_lock(&adev->srbm_mutex);
+	soc21_grbm_select(adev, 3, ring->pipe, 0, 0);
+
+	/* disable the queue if it's active */
+	if (RREG32_SOC15(GC, 0, regCP_HQD_ACTIVE) & 1) {
+		WREG32_SOC15(GC, 0, regCP_HQD_DEQUEUE_REQUEST, 1);
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (!(RREG32_SOC15(GC, 0, regCP_HQD_ACTIVE) & 1))
+				break;
+			udelay(1);
+		}
+	}
+	data = RREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL);
+	data = REG_SET_FIELD(data, CP_HQD_PQ_DOORBELL_CONTROL,
+				DOORBELL_EN, 0);
+	data = REG_SET_FIELD(data, CP_HQD_PQ_DOORBELL_CONTROL,
+				DOORBELL_HIT, 1);
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL, data);
+
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL, 0);
+
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_WPTR_LO, 0);
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_WPTR_HI, 0);
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_RPTR, 0);
+
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static void mes_v11_0_kiq_setting(struct amdgpu_ring *ring)
+{
+	uint32_t tmp;
+	struct amdgpu_device *adev = ring->adev;
+
+	/* tell RLC which is KIQ queue */
+	tmp = RREG32_SOC15(GC, 0, regRLC_CP_SCHEDULERS);
+	tmp &= 0xffffff00;
+	tmp |= (ring->me << 5) | (ring->pipe << 3) | (ring->queue);
+	WREG32_SOC15(GC, 0, regRLC_CP_SCHEDULERS, tmp | 0x80);
+}
+
+static void mes_v11_0_kiq_clear(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* tell RLC which is KIQ dequeue */
+	tmp = RREG32_SOC15(GC, 0, regRLC_CP_SCHEDULERS);
+	tmp &= ~RLC_CP_SCHEDULERS__scheduler0_MASK;
+	WREG32_SOC15(GC, 0, regRLC_CP_SCHEDULERS, tmp);
+}
+
+static int mes_v11_0_kiq_hw_init(struct amdgpu_device *adev)
+{
+	int r = 0;
+	struct amdgpu_ip_block *ip_block;
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+
+		r = mes_v11_0_load_microcode(adev, AMDGPU_MES_SCHED_PIPE, false);
+		if (r) {
+			DRM_ERROR("failed to load MES fw, r=%d\n", r);
+			return r;
+		}
+
+		r = mes_v11_0_load_microcode(adev, AMDGPU_MES_KIQ_PIPE, true);
+		if (r) {
+			DRM_ERROR("failed to load MES kiq fw, r=%d\n", r);
+			return r;
+		}
+
+	}
+
+	mes_v11_0_enable(adev, true);
+
+	mes_v11_0_get_fw_version(adev);
+
+	mes_v11_0_kiq_setting(&adev->gfx.kiq[0].ring);
+
+	ip_block = amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_MES);
+	if (unlikely(!ip_block)) {
+		dev_err(adev->dev, "Failed to get MES handle\n");
+		return -EINVAL;
+	}
+
+	r = mes_v11_0_queue_init(adev, AMDGPU_MES_KIQ_PIPE);
+	if (r)
+		goto failure;
+
+	if ((adev->mes.sched_version & AMDGPU_MES_VERSION_MASK) >= 0x47)
+		adev->mes.enable_legacy_queue_map = true;
+	else
+		adev->mes.enable_legacy_queue_map = false;
+
+	if (adev->mes.enable_legacy_queue_map) {
+		r = mes_v11_0_hw_init(ip_block);
+		if (r)
+			goto failure;
+	}
+
+	return r;
+
+failure:
+	mes_v11_0_hw_fini(ip_block);
+	return r;
+}
+
+static int mes_v11_0_kiq_hw_fini(struct amdgpu_device *adev)
+{
+	if (adev->mes.ring[0].sched.ready) {
+		mes_v11_0_kiq_dequeue(&adev->mes.ring[0]);
+		adev->mes.ring[0].sched.ready = false;
+	}
+
+	if (amdgpu_sriov_vf(adev)) {
+		mes_v11_0_kiq_dequeue(&adev->gfx.kiq[0].ring);
+		mes_v11_0_kiq_clear(adev);
+	}
+
+	mes_v11_0_enable(adev, false);
+
+	return 0;
+}
+
+static int mes_v11_0_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (adev->mes.ring[0].sched.ready)
+		goto out;
+
+	if (!adev->enable_mes_kiq) {
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+			r = mes_v11_0_load_microcode(adev,
+					     AMDGPU_MES_SCHED_PIPE, true);
+			if (r) {
+				DRM_ERROR("failed to MES fw, r=%d\n", r);
+				return r;
+			}
+		}
+
+		mes_v11_0_enable(adev, true);
+	}
+
+	r = mes_v11_0_queue_init(adev, AMDGPU_MES_SCHED_PIPE);
+	if (r)
+		goto failure;
+
+	r = mes_v11_0_set_hw_resources(&adev->mes);
+	if (r)
+		goto failure;
+
+	if ((adev->mes.sched_version & AMDGPU_MES_VERSION_MASK) >= 0x50) {
+		r = mes_v11_0_set_hw_resources_1(&adev->mes);
+		if (r) {
+			DRM_ERROR("failed mes_v11_0_set_hw_resources_1, r=%d\n", r);
+			goto failure;
+		}
+	}
+
+	r = mes_v11_0_query_sched_status(&adev->mes);
+	if (r) {
+		DRM_ERROR("MES is busy\n");
+		goto failure;
+	}
+
+	r = amdgpu_mes_update_enforce_isolation(adev);
+	if (r)
+		goto failure;
+
+out:
+	/*
+	 * Disable KIQ ring usage from the driver once MES is enabled.
+	 * MES uses KIQ ring exclusively so driver cannot access KIQ ring
+	 * with MES enabled.
+	 */
+	adev->gfx.kiq[0].ring.sched.ready = false;
+	adev->mes.ring[0].sched.ready = true;
+
+	return 0;
+
+failure:
+	mes_v11_0_hw_fini(ip_block);
+	return r;
+}
+
+static int mes_v11_0_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	return 0;
+}
+
+static int mes_v11_0_suspend(struct amdgpu_ip_block *ip_block)
+{
+	return mes_v11_0_hw_fini(ip_block);
+}
+
+static int mes_v11_0_resume(struct amdgpu_ip_block *ip_block)
+{
+	return mes_v11_0_hw_init(ip_block);
+}
+
+static int mes_v11_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int pipe, r;
+
+	adev->mes.hung_queue_db_array_size =
+		MES11_HUNG_DB_OFFSET_ARRAY_SIZE;
+	for (pipe = 0; pipe < AMDGPU_MAX_MES_PIPES; pipe++) {
+		if (!adev->enable_mes_kiq && pipe == AMDGPU_MES_KIQ_PIPE)
+			continue;
+		r = amdgpu_mes_init_microcode(adev, pipe);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+static const struct amd_ip_funcs mes_v11_0_ip_funcs = {
+	.name = "mes_v11_0",
+	.early_init = mes_v11_0_early_init,
+	.late_init = NULL,
+	.sw_init = mes_v11_0_sw_init,
+	.sw_fini = mes_v11_0_sw_fini,
+	.hw_init = mes_v11_0_hw_init,
+	.hw_fini = mes_v11_0_hw_fini,
+	.suspend = mes_v11_0_suspend,
+	.resume = mes_v11_0_resume,
+};
+
+const struct amdgpu_ip_block_version mes_v11_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_MES,
+	.major = 11,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &mes_v11_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v11_0.h b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.h
new file mode 100644
index 000000000000..b3519e1df2b2
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __MES_V11_0_H__
+#define __MES_V11_0_H__
+
+extern const struct amdgpu_ip_block_version mes_v11_0_ip_block;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v12_0.c b/drivers/gpu/drm/amd/amdgpu/mes_v12_0.c
new file mode 100644
index 000000000000..aff06f06aeee
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mes_v12_0.c
@@ -0,0 +1,1931 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include "amdgpu.h"
+#include "gfx_v12_0.h"
+#include "soc15_common.h"
+#include "soc21.h"
+#include "gc/gc_12_0_0_offset.h"
+#include "gc/gc_12_0_0_sh_mask.h"
+#include "gc/gc_11_0_0_default.h"
+#include "v12_structs.h"
+#include "mes_v12_api_def.h"
+
+MODULE_FIRMWARE("amdgpu/gc_12_0_0_mes.bin");
+MODULE_FIRMWARE("amdgpu/gc_12_0_0_mes1.bin");
+MODULE_FIRMWARE("amdgpu/gc_12_0_0_uni_mes.bin");
+MODULE_FIRMWARE("amdgpu/gc_12_0_1_mes.bin");
+MODULE_FIRMWARE("amdgpu/gc_12_0_1_mes1.bin");
+MODULE_FIRMWARE("amdgpu/gc_12_0_1_uni_mes.bin");
+
+static int mes_v12_0_hw_init(struct amdgpu_ip_block *ip_block);
+static int mes_v12_0_hw_fini(struct amdgpu_ip_block *ip_block);
+static int mes_v12_0_kiq_hw_init(struct amdgpu_device *adev);
+static int mes_v12_0_kiq_hw_fini(struct amdgpu_device *adev);
+
+#define MES_EOP_SIZE   2048
+
+#define MES12_HUNG_DB_OFFSET_ARRAY_SIZE 4
+
+static void mes_v12_0_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr,
+			     ring->wptr);
+		WDOORBELL64(ring->doorbell_index, ring->wptr);
+	} else {
+		BUG();
+	}
+}
+
+static u64 mes_v12_0_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	return *ring->rptr_cpu_addr;
+}
+
+static u64 mes_v12_0_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	u64 wptr;
+
+	if (ring->use_doorbell)
+		wptr = atomic64_read((atomic64_t *)ring->wptr_cpu_addr);
+	else
+		BUG();
+	return wptr;
+}
+
+static const struct amdgpu_ring_funcs mes_v12_0_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_MES,
+	.align_mask = 1,
+	.nop = 0,
+	.support_64bit_ptrs = true,
+	.get_rptr = mes_v12_0_ring_get_rptr,
+	.get_wptr = mes_v12_0_ring_get_wptr,
+	.set_wptr = mes_v12_0_ring_set_wptr,
+	.insert_nop = amdgpu_ring_insert_nop,
+};
+
+static const char *mes_v12_0_opcodes[] = {
+	"SET_HW_RSRC",
+	"SET_SCHEDULING_CONFIG",
+	"ADD_QUEUE",
+	"REMOVE_QUEUE",
+	"PERFORM_YIELD",
+	"SET_GANG_PRIORITY_LEVEL",
+	"SUSPEND",
+	"RESUME",
+	"RESET",
+	"SET_LOG_BUFFER",
+	"CHANGE_GANG_PRORITY",
+	"QUERY_SCHEDULER_STATUS",
+	"unused",
+	"SET_DEBUG_VMID",
+	"MISC",
+	"UPDATE_ROOT_PAGE_TABLE",
+	"AMD_LOG",
+	"SET_SE_MODE",
+	"SET_GANG_SUBMIT",
+	"SET_HW_RSRC_1",
+	"INVALIDATE_TLBS",
+};
+
+static const char *mes_v12_0_misc_opcodes[] = {
+	"WRITE_REG",
+	"INV_GART",
+	"QUERY_STATUS",
+	"READ_REG",
+	"WAIT_REG_MEM",
+	"SET_SHADER_DEBUGGER",
+	"NOTIFY_WORK_ON_UNMAPPED_QUEUE",
+	"NOTIFY_TO_UNMAP_PROCESSES",
+};
+
+static const char *mes_v12_0_get_op_string(union MESAPI__MISC *x_pkt)
+{
+	const char *op_str = NULL;
+
+	if (x_pkt->header.opcode < ARRAY_SIZE(mes_v12_0_opcodes))
+		op_str = mes_v12_0_opcodes[x_pkt->header.opcode];
+
+	return op_str;
+}
+
+static const char *mes_v12_0_get_misc_op_string(union MESAPI__MISC *x_pkt)
+{
+	const char *op_str = NULL;
+
+	if ((x_pkt->header.opcode == MES_SCH_API_MISC) &&
+	    (x_pkt->opcode < ARRAY_SIZE(mes_v12_0_misc_opcodes)))
+		op_str = mes_v12_0_misc_opcodes[x_pkt->opcode];
+
+	return op_str;
+}
+
+static int mes_v12_0_submit_pkt_and_poll_completion(struct amdgpu_mes *mes,
+					    int pipe, void *pkt, int size,
+					    int api_status_off)
+{
+	union MESAPI__QUERY_MES_STATUS mes_status_pkt;
+	signed long timeout = 2100000; /* 2100 ms */
+	struct amdgpu_device *adev = mes->adev;
+	struct amdgpu_ring *ring = &mes->ring[pipe];
+	spinlock_t *ring_lock = &mes->ring_lock[pipe];
+	struct MES_API_STATUS *api_status;
+	union MESAPI__MISC *x_pkt = pkt;
+	const char *op_str, *misc_op_str;
+	unsigned long flags;
+	u64 status_gpu_addr;
+	u32 seq, status_offset;
+	u64 *status_ptr;
+	signed long r;
+	int ret;
+
+	if (x_pkt->header.opcode >= MES_SCH_API_MAX)
+		return -EINVAL;
+
+	if (amdgpu_emu_mode) {
+		timeout *= 100;
+	} else if (amdgpu_sriov_vf(adev)) {
+		/* Worst case in sriov where all other 15 VF timeout, each VF needs about 600ms */
+		timeout = 15 * 600 * 1000;
+	}
+
+	ret = amdgpu_device_wb_get(adev, &status_offset);
+	if (ret)
+		return ret;
+
+	status_gpu_addr = adev->wb.gpu_addr + (status_offset * 4);
+	status_ptr = (u64 *)&adev->wb.wb[status_offset];
+	*status_ptr = 0;
+
+	spin_lock_irqsave(ring_lock, flags);
+	r = amdgpu_ring_alloc(ring, (size + sizeof(mes_status_pkt)) / 4);
+	if (r)
+		goto error_unlock_free;
+
+	seq = ++ring->fence_drv.sync_seq;
+	r = amdgpu_fence_wait_polling(ring,
+				      seq - ring->fence_drv.num_fences_mask,
+				      timeout);
+	if (r < 1)
+		goto error_undo;
+
+	api_status = (struct MES_API_STATUS *)((char *)pkt + api_status_off);
+	api_status->api_completion_fence_addr = status_gpu_addr;
+	api_status->api_completion_fence_value = 1;
+
+	amdgpu_ring_write_multiple(ring, pkt, size / 4);
+
+	memset(&mes_status_pkt, 0, sizeof(mes_status_pkt));
+	mes_status_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_status_pkt.header.opcode = MES_SCH_API_QUERY_SCHEDULER_STATUS;
+	mes_status_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+	mes_status_pkt.api_status.api_completion_fence_addr =
+		ring->fence_drv.gpu_addr;
+	mes_status_pkt.api_status.api_completion_fence_value = seq;
+
+	amdgpu_ring_write_multiple(ring, &mes_status_pkt,
+				   sizeof(mes_status_pkt) / 4);
+
+	amdgpu_ring_commit(ring);
+	spin_unlock_irqrestore(ring_lock, flags);
+
+	op_str = mes_v12_0_get_op_string(x_pkt);
+	misc_op_str = mes_v12_0_get_misc_op_string(x_pkt);
+
+	if (misc_op_str)
+		dev_dbg(adev->dev, "MES(%d) msg=%s (%s) was emitted\n",
+			pipe, op_str, misc_op_str);
+	else if (op_str)
+		dev_dbg(adev->dev, "MES(%d) msg=%s was emitted\n",
+			pipe, op_str);
+	else
+		dev_dbg(adev->dev, "MES(%d) msg=%d was emitted\n",
+			pipe, x_pkt->header.opcode);
+
+	r = amdgpu_fence_wait_polling(ring, seq, timeout);
+	if (r < 1 || !*status_ptr) {
+
+		if (misc_op_str)
+			dev_err(adev->dev, "MES(%d) failed to respond to msg=%s (%s)\n",
+				pipe, op_str, misc_op_str);
+		else if (op_str)
+			dev_err(adev->dev, "MES(%d) failed to respond to msg=%s\n",
+				pipe, op_str);
+		else
+			dev_err(adev->dev, "MES(%d) failed to respond to msg=%d\n",
+				pipe, x_pkt->header.opcode);
+
+		while (halt_if_hws_hang)
+			schedule();
+
+		r = -ETIMEDOUT;
+		goto error_wb_free;
+	}
+
+	amdgpu_device_wb_free(adev, status_offset);
+	return 0;
+
+error_undo:
+	dev_err(adev->dev, "MES ring buffer is full.\n");
+	amdgpu_ring_undo(ring);
+
+error_unlock_free:
+	spin_unlock_irqrestore(ring_lock, flags);
+
+error_wb_free:
+	amdgpu_device_wb_free(adev, status_offset);
+	return r;
+}
+
+static int convert_to_mes_queue_type(int queue_type)
+{
+	if (queue_type == AMDGPU_RING_TYPE_GFX)
+		return MES_QUEUE_TYPE_GFX;
+	else if (queue_type == AMDGPU_RING_TYPE_COMPUTE)
+		return MES_QUEUE_TYPE_COMPUTE;
+	else if (queue_type == AMDGPU_RING_TYPE_SDMA)
+		return MES_QUEUE_TYPE_SDMA;
+	else if (queue_type == AMDGPU_RING_TYPE_MES)
+		return MES_QUEUE_TYPE_SCHQ;
+	else
+		BUG();
+	return -1;
+}
+
+static int convert_to_mes_priority_level(int priority_level)
+{
+	switch (priority_level) {
+	case AMDGPU_MES_PRIORITY_LEVEL_LOW:
+		return AMD_PRIORITY_LEVEL_LOW;
+	case AMDGPU_MES_PRIORITY_LEVEL_NORMAL:
+	default:
+		return AMD_PRIORITY_LEVEL_NORMAL;
+	case AMDGPU_MES_PRIORITY_LEVEL_MEDIUM:
+		return AMD_PRIORITY_LEVEL_MEDIUM;
+	case AMDGPU_MES_PRIORITY_LEVEL_HIGH:
+		return AMD_PRIORITY_LEVEL_HIGH;
+	case AMDGPU_MES_PRIORITY_LEVEL_REALTIME:
+		return AMD_PRIORITY_LEVEL_REALTIME;
+	}
+}
+
+static int mes_v12_0_add_hw_queue(struct amdgpu_mes *mes,
+				  struct mes_add_queue_input *input)
+{
+	struct amdgpu_device *adev = mes->adev;
+	union MESAPI__ADD_QUEUE mes_add_queue_pkt;
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
+	uint32_t vm_cntx_cntl = hub->vm_cntx_cntl;
+
+	memset(&mes_add_queue_pkt, 0, sizeof(mes_add_queue_pkt));
+
+	mes_add_queue_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_add_queue_pkt.header.opcode = MES_SCH_API_ADD_QUEUE;
+	mes_add_queue_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	mes_add_queue_pkt.process_id = input->process_id;
+	mes_add_queue_pkt.page_table_base_addr = input->page_table_base_addr;
+	mes_add_queue_pkt.process_va_start = input->process_va_start;
+	mes_add_queue_pkt.process_va_end = input->process_va_end;
+	mes_add_queue_pkt.process_quantum = input->process_quantum;
+	mes_add_queue_pkt.process_context_addr = input->process_context_addr;
+	mes_add_queue_pkt.gang_quantum = input->gang_quantum;
+	mes_add_queue_pkt.gang_context_addr = input->gang_context_addr;
+	mes_add_queue_pkt.inprocess_gang_priority =
+		convert_to_mes_priority_level(input->inprocess_gang_priority);
+	mes_add_queue_pkt.gang_global_priority_level =
+		convert_to_mes_priority_level(input->gang_global_priority_level);
+	mes_add_queue_pkt.doorbell_offset = input->doorbell_offset;
+	mes_add_queue_pkt.mqd_addr = input->mqd_addr;
+
+	mes_add_queue_pkt.wptr_addr = input->wptr_mc_addr;
+
+	mes_add_queue_pkt.queue_type =
+		convert_to_mes_queue_type(input->queue_type);
+	mes_add_queue_pkt.paging = input->paging;
+	mes_add_queue_pkt.vm_context_cntl = vm_cntx_cntl;
+	mes_add_queue_pkt.gws_base = input->gws_base;
+	mes_add_queue_pkt.gws_size = input->gws_size;
+	mes_add_queue_pkt.trap_handler_addr = input->tba_addr;
+	mes_add_queue_pkt.tma_addr = input->tma_addr;
+	mes_add_queue_pkt.trap_en = input->trap_en;
+	mes_add_queue_pkt.skip_process_ctx_clear = input->skip_process_ctx_clear;
+	mes_add_queue_pkt.is_kfd_process = input->is_kfd_process;
+
+	/* For KFD, gds_size is re-used for queue size (needed in MES for AQL queues) */
+	mes_add_queue_pkt.is_aql_queue = input->is_aql_queue;
+	mes_add_queue_pkt.gds_size = input->queue_size;
+
+	/* For KFD, gds_size is re-used for queue size (needed in MES for AQL queues) */
+	mes_add_queue_pkt.is_aql_queue = input->is_aql_queue;
+	mes_add_queue_pkt.gds_size = input->queue_size;
+
+	return mes_v12_0_submit_pkt_and_poll_completion(mes,
+			AMDGPU_MES_SCHED_PIPE,
+			&mes_add_queue_pkt, sizeof(mes_add_queue_pkt),
+			offsetof(union MESAPI__ADD_QUEUE, api_status));
+}
+
+static int mes_v12_0_remove_hw_queue(struct amdgpu_mes *mes,
+				     struct mes_remove_queue_input *input)
+{
+	union MESAPI__REMOVE_QUEUE mes_remove_queue_pkt;
+
+	memset(&mes_remove_queue_pkt, 0, sizeof(mes_remove_queue_pkt));
+
+	mes_remove_queue_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_remove_queue_pkt.header.opcode = MES_SCH_API_REMOVE_QUEUE;
+	mes_remove_queue_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	mes_remove_queue_pkt.doorbell_offset = input->doorbell_offset;
+	mes_remove_queue_pkt.gang_context_addr = input->gang_context_addr;
+
+	return mes_v12_0_submit_pkt_and_poll_completion(mes,
+			AMDGPU_MES_SCHED_PIPE,
+			&mes_remove_queue_pkt, sizeof(mes_remove_queue_pkt),
+			offsetof(union MESAPI__REMOVE_QUEUE, api_status));
+}
+
+int gfx_v12_0_request_gfx_index_mutex(struct amdgpu_device *adev,
+				      bool req)
+{
+	u32 i, tmp, val;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		/* Request with MeId=2, PipeId=0 */
+		tmp = REG_SET_FIELD(0, CP_GFX_INDEX_MUTEX, REQUEST, req);
+		tmp = REG_SET_FIELD(tmp, CP_GFX_INDEX_MUTEX, CLIENTID, 4);
+		WREG32_SOC15(GC, 0, regCP_GFX_INDEX_MUTEX, tmp);
+
+		val = RREG32_SOC15(GC, 0, regCP_GFX_INDEX_MUTEX);
+		if (req) {
+			if (val == tmp)
+				break;
+		} else {
+			tmp = REG_SET_FIELD(tmp, CP_GFX_INDEX_MUTEX,
+					    REQUEST, 1);
+
+			/* unlocked or locked by firmware */
+			if (val != tmp)
+				break;
+		}
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout)
+		return -EINVAL;
+
+	return 0;
+}
+
+static int mes_v12_0_reset_queue_mmio(struct amdgpu_mes *mes, uint32_t queue_type,
+				      uint32_t me_id, uint32_t pipe_id,
+				      uint32_t queue_id, uint32_t vmid)
+{
+	struct amdgpu_device *adev = mes->adev;
+	uint32_t value, reg;
+	int i, r = 0;
+
+	amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
+
+	if (queue_type == AMDGPU_RING_TYPE_GFX) {
+		dev_info(adev->dev, "reset gfx queue (%d:%d:%d: vmid:%d)\n",
+			 me_id, pipe_id, queue_id, vmid);
+
+		mutex_lock(&adev->gfx.reset_sem_mutex);
+		gfx_v12_0_request_gfx_index_mutex(adev, true);
+		/* all se allow writes */
+		WREG32_SOC15(GC, 0, regGRBM_GFX_INDEX,
+			     (uint32_t)(0x1 << GRBM_GFX_INDEX__SE_BROADCAST_WRITES__SHIFT));
+		value = REG_SET_FIELD(0, CP_VMID_RESET, RESET_REQUEST, 1 << vmid);
+		if (pipe_id == 0)
+			value = REG_SET_FIELD(value, CP_VMID_RESET, PIPE0_QUEUES, 1 << queue_id);
+		else
+			value = REG_SET_FIELD(value, CP_VMID_RESET, PIPE1_QUEUES, 1 << queue_id);
+		WREG32_SOC15(GC, 0, regCP_VMID_RESET, value);
+		gfx_v12_0_request_gfx_index_mutex(adev, false);
+		mutex_unlock(&adev->gfx.reset_sem_mutex);
+
+		mutex_lock(&adev->srbm_mutex);
+		soc21_grbm_select(adev, me_id, pipe_id, queue_id, 0);
+		/* wait till dequeue take effects */
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (!(RREG32_SOC15(GC, 0, regCP_GFX_HQD_ACTIVE) & 1))
+				break;
+			udelay(1);
+		}
+		if (i >= adev->usec_timeout) {
+			dev_err(adev->dev, "failed to wait on gfx hqd deactivate\n");
+			r = -ETIMEDOUT;
+		}
+
+		soc21_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+	} else if (queue_type == AMDGPU_RING_TYPE_COMPUTE) {
+		dev_info(adev->dev, "reset compute queue (%d:%d:%d)\n",
+			 me_id, pipe_id, queue_id);
+		mutex_lock(&adev->srbm_mutex);
+		soc21_grbm_select(adev, me_id, pipe_id, queue_id, 0);
+		WREG32_SOC15(GC, 0, regCP_HQD_DEQUEUE_REQUEST, 0x2);
+		WREG32_SOC15(GC, 0, regSPI_COMPUTE_QUEUE_RESET, 0x1);
+
+		/* wait till dequeue take effects */
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (!(RREG32_SOC15(GC, 0, regCP_HQD_ACTIVE) & 1))
+				break;
+			udelay(1);
+		}
+		if (i >= adev->usec_timeout) {
+			dev_err(adev->dev, "failed to wait on hqd deactivate\n");
+			r = -ETIMEDOUT;
+		}
+		soc21_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+	} else if (queue_type == AMDGPU_RING_TYPE_SDMA) {
+		dev_info(adev->dev, "reset sdma queue (%d:%d:%d)\n",
+			 me_id, pipe_id, queue_id);
+		switch (me_id) {
+		case 1:
+			reg = SOC15_REG_OFFSET(GC, 0, regSDMA1_QUEUE_RESET_REQ);
+			break;
+		case 0:
+		default:
+			reg = SOC15_REG_OFFSET(GC, 0, regSDMA0_QUEUE_RESET_REQ);
+			break;
+		}
+
+		value = 1 << queue_id;
+		WREG32(reg, value);
+		/* wait for queue reset done */
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (!(RREG32(reg) & value))
+				break;
+			udelay(1);
+		}
+		if (i >= adev->usec_timeout) {
+			dev_err(adev->dev, "failed to wait on sdma queue reset done\n");
+			r = -ETIMEDOUT;
+		}
+	}
+
+	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
+	return r;
+}
+
+static int mes_v12_0_map_legacy_queue(struct amdgpu_mes *mes,
+				      struct mes_map_legacy_queue_input *input)
+{
+	union MESAPI__ADD_QUEUE mes_add_queue_pkt;
+	int pipe;
+
+	memset(&mes_add_queue_pkt, 0, sizeof(mes_add_queue_pkt));
+
+	mes_add_queue_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_add_queue_pkt.header.opcode = MES_SCH_API_ADD_QUEUE;
+	mes_add_queue_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	mes_add_queue_pkt.pipe_id = input->pipe_id;
+	mes_add_queue_pkt.queue_id = input->queue_id;
+	mes_add_queue_pkt.doorbell_offset = input->doorbell_offset;
+	mes_add_queue_pkt.mqd_addr = input->mqd_addr;
+	mes_add_queue_pkt.wptr_addr = input->wptr_addr;
+	mes_add_queue_pkt.queue_type =
+		convert_to_mes_queue_type(input->queue_type);
+	mes_add_queue_pkt.map_legacy_kq = 1;
+
+	if (mes->adev->enable_uni_mes)
+		pipe = AMDGPU_MES_KIQ_PIPE;
+	else
+		pipe = AMDGPU_MES_SCHED_PIPE;
+
+	return mes_v12_0_submit_pkt_and_poll_completion(mes, pipe,
+			&mes_add_queue_pkt, sizeof(mes_add_queue_pkt),
+			offsetof(union MESAPI__ADD_QUEUE, api_status));
+}
+
+static int mes_v12_0_unmap_legacy_queue(struct amdgpu_mes *mes,
+			struct mes_unmap_legacy_queue_input *input)
+{
+	union MESAPI__REMOVE_QUEUE mes_remove_queue_pkt;
+	int pipe;
+
+	memset(&mes_remove_queue_pkt, 0, sizeof(mes_remove_queue_pkt));
+
+	mes_remove_queue_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_remove_queue_pkt.header.opcode = MES_SCH_API_REMOVE_QUEUE;
+	mes_remove_queue_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	mes_remove_queue_pkt.doorbell_offset = input->doorbell_offset;
+	mes_remove_queue_pkt.gang_context_addr = 0;
+
+	mes_remove_queue_pkt.pipe_id = input->pipe_id;
+	mes_remove_queue_pkt.queue_id = input->queue_id;
+
+	if (input->action == PREEMPT_QUEUES_NO_UNMAP) {
+		mes_remove_queue_pkt.preempt_legacy_gfx_queue = 1;
+		mes_remove_queue_pkt.tf_addr = input->trail_fence_addr;
+		mes_remove_queue_pkt.tf_data =
+			lower_32_bits(input->trail_fence_data);
+	} else {
+		mes_remove_queue_pkt.unmap_legacy_queue = 1;
+		mes_remove_queue_pkt.queue_type =
+			convert_to_mes_queue_type(input->queue_type);
+	}
+
+	if (mes->adev->enable_uni_mes)
+		pipe = AMDGPU_MES_KIQ_PIPE;
+	else
+		pipe = AMDGPU_MES_SCHED_PIPE;
+
+	return mes_v12_0_submit_pkt_and_poll_completion(mes, pipe,
+			&mes_remove_queue_pkt, sizeof(mes_remove_queue_pkt),
+			offsetof(union MESAPI__REMOVE_QUEUE, api_status));
+}
+
+static int mes_v12_0_suspend_gang(struct amdgpu_mes *mes,
+				  struct mes_suspend_gang_input *input)
+{
+	union MESAPI__SUSPEND mes_suspend_gang_pkt;
+
+	memset(&mes_suspend_gang_pkt, 0, sizeof(mes_suspend_gang_pkt));
+
+	mes_suspend_gang_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_suspend_gang_pkt.header.opcode = MES_SCH_API_SUSPEND;
+	mes_suspend_gang_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	mes_suspend_gang_pkt.suspend_all_gangs = input->suspend_all_gangs;
+	mes_suspend_gang_pkt.gang_context_addr = input->gang_context_addr;
+	mes_suspend_gang_pkt.suspend_fence_addr = input->suspend_fence_addr;
+	mes_suspend_gang_pkt.suspend_fence_value = input->suspend_fence_value;
+
+	return mes_v12_0_submit_pkt_and_poll_completion(mes, AMDGPU_MES_SCHED_PIPE,
+			&mes_suspend_gang_pkt, sizeof(mes_suspend_gang_pkt),
+			offsetof(union MESAPI__SUSPEND, api_status));
+}
+
+static int mes_v12_0_resume_gang(struct amdgpu_mes *mes,
+				 struct mes_resume_gang_input *input)
+{
+	union MESAPI__RESUME mes_resume_gang_pkt;
+
+	memset(&mes_resume_gang_pkt, 0, sizeof(mes_resume_gang_pkt));
+
+	mes_resume_gang_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_resume_gang_pkt.header.opcode = MES_SCH_API_RESUME;
+	mes_resume_gang_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	mes_resume_gang_pkt.resume_all_gangs = input->resume_all_gangs;
+	mes_resume_gang_pkt.gang_context_addr = input->gang_context_addr;
+
+	return mes_v12_0_submit_pkt_and_poll_completion(mes, AMDGPU_MES_SCHED_PIPE,
+			&mes_resume_gang_pkt, sizeof(mes_resume_gang_pkt),
+			offsetof(union MESAPI__RESUME, api_status));
+}
+
+static int mes_v12_0_query_sched_status(struct amdgpu_mes *mes, int pipe)
+{
+	union MESAPI__QUERY_MES_STATUS mes_status_pkt;
+
+	memset(&mes_status_pkt, 0, sizeof(mes_status_pkt));
+
+	mes_status_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_status_pkt.header.opcode = MES_SCH_API_QUERY_SCHEDULER_STATUS;
+	mes_status_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	return mes_v12_0_submit_pkt_and_poll_completion(mes, pipe,
+			&mes_status_pkt, sizeof(mes_status_pkt),
+			offsetof(union MESAPI__QUERY_MES_STATUS, api_status));
+}
+
+static int mes_v12_0_misc_op(struct amdgpu_mes *mes,
+			     struct mes_misc_op_input *input)
+{
+	union MESAPI__MISC misc_pkt;
+	int pipe;
+
+	if (mes->adev->enable_uni_mes)
+		pipe = AMDGPU_MES_KIQ_PIPE;
+	else
+		pipe = AMDGPU_MES_SCHED_PIPE;
+
+	memset(&misc_pkt, 0, sizeof(misc_pkt));
+
+	misc_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	misc_pkt.header.opcode = MES_SCH_API_MISC;
+	misc_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	switch (input->op) {
+	case MES_MISC_OP_READ_REG:
+		misc_pkt.opcode = MESAPI_MISC__READ_REG;
+		misc_pkt.read_reg.reg_offset = input->read_reg.reg_offset;
+		misc_pkt.read_reg.buffer_addr = input->read_reg.buffer_addr;
+		break;
+	case MES_MISC_OP_WRITE_REG:
+		misc_pkt.opcode = MESAPI_MISC__WRITE_REG;
+		misc_pkt.write_reg.reg_offset = input->write_reg.reg_offset;
+		misc_pkt.write_reg.reg_value = input->write_reg.reg_value;
+		break;
+	case MES_MISC_OP_WRM_REG_WAIT:
+		misc_pkt.opcode = MESAPI_MISC__WAIT_REG_MEM;
+		misc_pkt.wait_reg_mem.op = WRM_OPERATION__WAIT_REG_MEM;
+		misc_pkt.wait_reg_mem.reference = input->wrm_reg.ref;
+		misc_pkt.wait_reg_mem.mask = input->wrm_reg.mask;
+		misc_pkt.wait_reg_mem.reg_offset1 = input->wrm_reg.reg0;
+		misc_pkt.wait_reg_mem.reg_offset2 = 0;
+		break;
+	case MES_MISC_OP_WRM_REG_WR_WAIT:
+		misc_pkt.opcode = MESAPI_MISC__WAIT_REG_MEM;
+		misc_pkt.wait_reg_mem.op = WRM_OPERATION__WR_WAIT_WR_REG;
+		misc_pkt.wait_reg_mem.reference = input->wrm_reg.ref;
+		misc_pkt.wait_reg_mem.mask = input->wrm_reg.mask;
+		misc_pkt.wait_reg_mem.reg_offset1 = input->wrm_reg.reg0;
+		misc_pkt.wait_reg_mem.reg_offset2 = input->wrm_reg.reg1;
+		break;
+	case MES_MISC_OP_SET_SHADER_DEBUGGER:
+		pipe = AMDGPU_MES_SCHED_PIPE;
+		misc_pkt.opcode = MESAPI_MISC__SET_SHADER_DEBUGGER;
+		misc_pkt.set_shader_debugger.process_context_addr =
+				input->set_shader_debugger.process_context_addr;
+		misc_pkt.set_shader_debugger.flags.u32all =
+				input->set_shader_debugger.flags.u32all;
+		misc_pkt.set_shader_debugger.spi_gdbg_per_vmid_cntl =
+				input->set_shader_debugger.spi_gdbg_per_vmid_cntl;
+		memcpy(misc_pkt.set_shader_debugger.tcp_watch_cntl,
+				input->set_shader_debugger.tcp_watch_cntl,
+				sizeof(misc_pkt.set_shader_debugger.tcp_watch_cntl));
+		misc_pkt.set_shader_debugger.trap_en = input->set_shader_debugger.trap_en;
+		break;
+	case MES_MISC_OP_CHANGE_CONFIG:
+		misc_pkt.opcode = MESAPI_MISC__CHANGE_CONFIG;
+		misc_pkt.change_config.opcode =
+				MESAPI_MISC__CHANGE_CONFIG_OPTION_LIMIT_SINGLE_PROCESS;
+		misc_pkt.change_config.option.bits.limit_single_process =
+				input->change_config.option.limit_single_process;
+		break;
+
+	default:
+		DRM_ERROR("unsupported misc op (%d) \n", input->op);
+		return -EINVAL;
+	}
+
+	return mes_v12_0_submit_pkt_and_poll_completion(mes, pipe,
+			&misc_pkt, sizeof(misc_pkt),
+			offsetof(union MESAPI__MISC, api_status));
+}
+
+static int mes_v12_0_set_hw_resources_1(struct amdgpu_mes *mes, int pipe)
+{
+	union MESAPI_SET_HW_RESOURCES_1 mes_set_hw_res_1_pkt;
+
+	memset(&mes_set_hw_res_1_pkt, 0, sizeof(mes_set_hw_res_1_pkt));
+
+	mes_set_hw_res_1_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_set_hw_res_1_pkt.header.opcode = MES_SCH_API_SET_HW_RSRC_1;
+	mes_set_hw_res_1_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+	mes_set_hw_res_1_pkt.mes_kiq_unmap_timeout = 0xa;
+	mes_set_hw_res_1_pkt.cleaner_shader_fence_mc_addr =
+		mes->resource_1_gpu_addr[pipe];
+
+	return mes_v12_0_submit_pkt_and_poll_completion(mes, pipe,
+			&mes_set_hw_res_1_pkt, sizeof(mes_set_hw_res_1_pkt),
+			offsetof(union MESAPI_SET_HW_RESOURCES_1, api_status));
+}
+
+static int mes_v12_0_set_hw_resources(struct amdgpu_mes *mes, int pipe)
+{
+	int i;
+	struct amdgpu_device *adev = mes->adev;
+	union MESAPI_SET_HW_RESOURCES mes_set_hw_res_pkt;
+
+	memset(&mes_set_hw_res_pkt, 0, sizeof(mes_set_hw_res_pkt));
+
+	mes_set_hw_res_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_set_hw_res_pkt.header.opcode = MES_SCH_API_SET_HW_RSRC;
+	mes_set_hw_res_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	if (pipe == AMDGPU_MES_SCHED_PIPE) {
+		mes_set_hw_res_pkt.vmid_mask_mmhub = mes->vmid_mask_mmhub;
+		mes_set_hw_res_pkt.vmid_mask_gfxhub = mes->vmid_mask_gfxhub;
+		mes_set_hw_res_pkt.gds_size = adev->gds.gds_size;
+		mes_set_hw_res_pkt.paging_vmid = 0;
+
+		for (i = 0; i < MAX_COMPUTE_PIPES; i++)
+			mes_set_hw_res_pkt.compute_hqd_mask[i] =
+				mes->compute_hqd_mask[i];
+
+		for (i = 0; i < MAX_GFX_PIPES; i++)
+			mes_set_hw_res_pkt.gfx_hqd_mask[i] =
+				mes->gfx_hqd_mask[i];
+
+		for (i = 0; i < MAX_SDMA_PIPES; i++)
+			mes_set_hw_res_pkt.sdma_hqd_mask[i] =
+				mes->sdma_hqd_mask[i];
+
+		for (i = 0; i < AMD_PRIORITY_NUM_LEVELS; i++)
+			mes_set_hw_res_pkt.aggregated_doorbells[i] =
+				mes->aggregated_doorbells[i];
+	}
+
+	mes_set_hw_res_pkt.g_sch_ctx_gpu_mc_ptr =
+		mes->sch_ctx_gpu_addr[pipe];
+	mes_set_hw_res_pkt.query_status_fence_gpu_mc_ptr =
+		mes->query_status_fence_gpu_addr[pipe];
+
+	for (i = 0; i < 5; i++) {
+		mes_set_hw_res_pkt.gc_base[i] = adev->reg_offset[GC_HWIP][0][i];
+		mes_set_hw_res_pkt.mmhub_base[i] =
+				adev->reg_offset[MMHUB_HWIP][0][i];
+		mes_set_hw_res_pkt.osssys_base[i] =
+		adev->reg_offset[OSSSYS_HWIP][0][i];
+	}
+
+	mes_set_hw_res_pkt.disable_reset = 1;
+	mes_set_hw_res_pkt.disable_mes_log = 1;
+	mes_set_hw_res_pkt.use_different_vmid_compute = 1;
+	mes_set_hw_res_pkt.enable_reg_active_poll = 1;
+	mes_set_hw_res_pkt.enable_level_process_quantum_check = 1;
+	if ((mes->adev->mes.sched_version & AMDGPU_MES_VERSION_MASK) >= 0x82)
+		mes_set_hw_res_pkt.enable_lr_compute_wa = 1;
+	else
+		dev_info_once(adev->dev,
+			      "MES FW version must be >= 0x82 to enable LR compute workaround.\n");
+
+	/*
+	 * Keep oversubscribe timer for sdma . When we have unmapped doorbell
+	 * handling support, other queue will not use the oversubscribe timer.
+	 * handling  mode - 0: disabled; 1: basic version; 2: basic+ version
+	 */
+	mes_set_hw_res_pkt.oversubscription_timer = 50;
+	mes_set_hw_res_pkt.unmapped_doorbell_handling = 1;
+
+	if (amdgpu_mes_log_enable) {
+		mes_set_hw_res_pkt.enable_mes_event_int_logging = 1;
+		mes_set_hw_res_pkt.event_intr_history_gpu_mc_ptr = mes->event_log_gpu_addr +
+				pipe * (AMDGPU_MES_LOG_BUFFER_SIZE + AMDGPU_MES_MSCRATCH_SIZE);
+	}
+
+	if (adev->enforce_isolation[0] == AMDGPU_ENFORCE_ISOLATION_ENABLE)
+		mes_set_hw_res_pkt.limit_single_process = 1;
+
+	return mes_v12_0_submit_pkt_and_poll_completion(mes, pipe,
+			&mes_set_hw_res_pkt, sizeof(mes_set_hw_res_pkt),
+			offsetof(union MESAPI_SET_HW_RESOURCES, api_status));
+}
+
+static void mes_v12_0_init_aggregated_doorbell(struct amdgpu_mes *mes)
+{
+	struct amdgpu_device *adev = mes->adev;
+	uint32_t data;
+
+	data = RREG32_SOC15(GC, 0, regCP_MES_DOORBELL_CONTROL1);
+	data &= ~(CP_MES_DOORBELL_CONTROL1__DOORBELL_OFFSET_MASK |
+		  CP_MES_DOORBELL_CONTROL1__DOORBELL_EN_MASK |
+		  CP_MES_DOORBELL_CONTROL1__DOORBELL_HIT_MASK);
+	data |= mes->aggregated_doorbells[AMDGPU_MES_PRIORITY_LEVEL_LOW] <<
+		CP_MES_DOORBELL_CONTROL1__DOORBELL_OFFSET__SHIFT;
+	data |= 1 << CP_MES_DOORBELL_CONTROL1__DOORBELL_EN__SHIFT;
+	WREG32_SOC15(GC, 0, regCP_MES_DOORBELL_CONTROL1, data);
+
+	data = RREG32_SOC15(GC, 0, regCP_MES_DOORBELL_CONTROL2);
+	data &= ~(CP_MES_DOORBELL_CONTROL2__DOORBELL_OFFSET_MASK |
+		  CP_MES_DOORBELL_CONTROL2__DOORBELL_EN_MASK |
+		  CP_MES_DOORBELL_CONTROL2__DOORBELL_HIT_MASK);
+	data |= mes->aggregated_doorbells[AMDGPU_MES_PRIORITY_LEVEL_NORMAL] <<
+		CP_MES_DOORBELL_CONTROL2__DOORBELL_OFFSET__SHIFT;
+	data |= 1 << CP_MES_DOORBELL_CONTROL2__DOORBELL_EN__SHIFT;
+	WREG32_SOC15(GC, 0, regCP_MES_DOORBELL_CONTROL2, data);
+
+	data = RREG32_SOC15(GC, 0, regCP_MES_DOORBELL_CONTROL3);
+	data &= ~(CP_MES_DOORBELL_CONTROL3__DOORBELL_OFFSET_MASK |
+		  CP_MES_DOORBELL_CONTROL3__DOORBELL_EN_MASK |
+		  CP_MES_DOORBELL_CONTROL3__DOORBELL_HIT_MASK);
+	data |= mes->aggregated_doorbells[AMDGPU_MES_PRIORITY_LEVEL_MEDIUM] <<
+		CP_MES_DOORBELL_CONTROL3__DOORBELL_OFFSET__SHIFT;
+	data |= 1 << CP_MES_DOORBELL_CONTROL3__DOORBELL_EN__SHIFT;
+	WREG32_SOC15(GC, 0, regCP_MES_DOORBELL_CONTROL3, data);
+
+	data = RREG32_SOC15(GC, 0, regCP_MES_DOORBELL_CONTROL4);
+	data &= ~(CP_MES_DOORBELL_CONTROL4__DOORBELL_OFFSET_MASK |
+		  CP_MES_DOORBELL_CONTROL4__DOORBELL_EN_MASK |
+		  CP_MES_DOORBELL_CONTROL4__DOORBELL_HIT_MASK);
+	data |= mes->aggregated_doorbells[AMDGPU_MES_PRIORITY_LEVEL_HIGH] <<
+		CP_MES_DOORBELL_CONTROL4__DOORBELL_OFFSET__SHIFT;
+	data |= 1 << CP_MES_DOORBELL_CONTROL4__DOORBELL_EN__SHIFT;
+	WREG32_SOC15(GC, 0, regCP_MES_DOORBELL_CONTROL4, data);
+
+	data = RREG32_SOC15(GC, 0, regCP_MES_DOORBELL_CONTROL5);
+	data &= ~(CP_MES_DOORBELL_CONTROL5__DOORBELL_OFFSET_MASK |
+		  CP_MES_DOORBELL_CONTROL5__DOORBELL_EN_MASK |
+		  CP_MES_DOORBELL_CONTROL5__DOORBELL_HIT_MASK);
+	data |= mes->aggregated_doorbells[AMDGPU_MES_PRIORITY_LEVEL_REALTIME] <<
+		CP_MES_DOORBELL_CONTROL5__DOORBELL_OFFSET__SHIFT;
+	data |= 1 << CP_MES_DOORBELL_CONTROL5__DOORBELL_EN__SHIFT;
+	WREG32_SOC15(GC, 0, regCP_MES_DOORBELL_CONTROL5, data);
+
+	data = 1 << CP_HQD_GFX_CONTROL__DB_UPDATED_MSG_EN__SHIFT;
+	WREG32_SOC15(GC, 0, regCP_HQD_GFX_CONTROL, data);
+}
+
+
+static void mes_v12_0_enable_unmapped_doorbell_handling(
+		struct amdgpu_mes *mes, bool enable)
+{
+	struct amdgpu_device *adev = mes->adev;
+	uint32_t data = RREG32_SOC15(GC, 0, regCP_UNMAPPED_DOORBELL);
+
+	/*
+	 * The default PROC_LSB settng is 0xc which means doorbell
+	 * addr[16:12] gives the doorbell page number. For kfd, each
+	 * process will use 2 pages of doorbell, we need to change the
+	 * setting to 0xd
+	 */
+	data &= ~CP_UNMAPPED_DOORBELL__PROC_LSB_MASK;
+	data |= 0xd <<  CP_UNMAPPED_DOORBELL__PROC_LSB__SHIFT;
+
+	data |= (enable ? 1 : 0) << CP_UNMAPPED_DOORBELL__ENABLE__SHIFT;
+
+	WREG32_SOC15(GC, 0, regCP_UNMAPPED_DOORBELL, data);
+}
+
+static int mes_v12_0_reset_hw_queue(struct amdgpu_mes *mes,
+				    struct mes_reset_queue_input *input)
+{
+	union MESAPI__RESET mes_reset_queue_pkt;
+	int pipe;
+
+	if (input->use_mmio)
+		return mes_v12_0_reset_queue_mmio(mes, input->queue_type,
+						  input->me_id, input->pipe_id,
+						  input->queue_id, input->vmid);
+
+	memset(&mes_reset_queue_pkt, 0, sizeof(mes_reset_queue_pkt));
+
+	mes_reset_queue_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_reset_queue_pkt.header.opcode = MES_SCH_API_RESET;
+	mes_reset_queue_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	mes_reset_queue_pkt.queue_type =
+		convert_to_mes_queue_type(input->queue_type);
+
+	if (input->legacy_gfx) {
+		mes_reset_queue_pkt.reset_legacy_gfx = 1;
+		mes_reset_queue_pkt.pipe_id_lp = input->pipe_id;
+		mes_reset_queue_pkt.queue_id_lp = input->queue_id;
+		mes_reset_queue_pkt.mqd_mc_addr_lp = input->mqd_addr;
+		mes_reset_queue_pkt.doorbell_offset_lp = input->doorbell_offset;
+		mes_reset_queue_pkt.wptr_addr_lp = input->wptr_addr;
+		mes_reset_queue_pkt.vmid_id_lp = input->vmid;
+	} else {
+		mes_reset_queue_pkt.reset_queue_only = 1;
+		mes_reset_queue_pkt.doorbell_offset = input->doorbell_offset;
+	}
+
+	if (input->is_kq)
+		pipe = AMDGPU_MES_KIQ_PIPE;
+	else
+		pipe = AMDGPU_MES_SCHED_PIPE;
+
+	return mes_v12_0_submit_pkt_and_poll_completion(mes, pipe,
+			&mes_reset_queue_pkt, sizeof(mes_reset_queue_pkt),
+			offsetof(union MESAPI__RESET, api_status));
+}
+
+static int mes_v12_0_detect_and_reset_hung_queues(struct amdgpu_mes *mes,
+						  struct mes_detect_and_reset_queue_input *input)
+{
+	union MESAPI__RESET mes_reset_queue_pkt;
+
+	memset(&mes_reset_queue_pkt, 0, sizeof(mes_reset_queue_pkt));
+
+	mes_reset_queue_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_reset_queue_pkt.header.opcode = MES_SCH_API_RESET;
+	mes_reset_queue_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	mes_reset_queue_pkt.queue_type =
+		convert_to_mes_queue_type(input->queue_type);
+	mes_reset_queue_pkt.doorbell_offset_addr =
+		mes->hung_queue_db_array_gpu_addr;
+
+	if (input->detect_only)
+		mes_reset_queue_pkt.hang_detect_only = 1;
+	else
+		mes_reset_queue_pkt.hang_detect_then_reset = 1;
+
+	return mes_v12_0_submit_pkt_and_poll_completion(mes, AMDGPU_MES_SCHED_PIPE,
+			&mes_reset_queue_pkt, sizeof(mes_reset_queue_pkt),
+			offsetof(union MESAPI__RESET, api_status));
+}
+
+static int mes_v12_inv_tlb_convert_hub_id(uint8_t id)
+{
+	/*
+	 * MES doesn't support invalidate gc_hub on slave xcc individually
+	 * master xcc will invalidate all gc_hub for the partition
+	 */
+	if (AMDGPU_IS_GFXHUB(id))
+		return 0;
+	else if (AMDGPU_IS_MMHUB0(id))
+		return 1;
+	else
+		return -EINVAL;
+
+}
+
+static int mes_v12_0_inv_tlbs_pasid(struct amdgpu_mes *mes,
+				    struct mes_inv_tlbs_pasid_input *input)
+{
+	union MESAPI__INV_TLBS mes_inv_tlbs;
+	int ret;
+
+	memset(&mes_inv_tlbs, 0, sizeof(mes_inv_tlbs));
+
+	mes_inv_tlbs.header.type = MES_API_TYPE_SCHEDULER;
+	mes_inv_tlbs.header.opcode = MES_SCH_API_INV_TLBS;
+	mes_inv_tlbs.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	mes_inv_tlbs.invalidate_tlbs.inv_sel = 0;
+	mes_inv_tlbs.invalidate_tlbs.flush_type = input->flush_type;
+	mes_inv_tlbs.invalidate_tlbs.inv_sel_id = input->pasid;
+
+	/*convert amdgpu_mes_hub_id to mes expected hub_id */
+	ret = mes_v12_inv_tlb_convert_hub_id(input->hub_id);
+	if (ret < 0)
+		return -EINVAL;
+	mes_inv_tlbs.invalidate_tlbs.hub_id = ret;
+	return mes_v12_0_submit_pkt_and_poll_completion(mes, AMDGPU_MES_KIQ_PIPE,
+			&mes_inv_tlbs, sizeof(mes_inv_tlbs),
+			offsetof(union MESAPI__INV_TLBS, api_status));
+
+}
+
+static const struct amdgpu_mes_funcs mes_v12_0_funcs = {
+	.add_hw_queue = mes_v12_0_add_hw_queue,
+	.remove_hw_queue = mes_v12_0_remove_hw_queue,
+	.map_legacy_queue = mes_v12_0_map_legacy_queue,
+	.unmap_legacy_queue = mes_v12_0_unmap_legacy_queue,
+	.suspend_gang = mes_v12_0_suspend_gang,
+	.resume_gang = mes_v12_0_resume_gang,
+	.misc_op = mes_v12_0_misc_op,
+	.reset_hw_queue = mes_v12_0_reset_hw_queue,
+	.invalidate_tlbs_pasid = mes_v12_0_inv_tlbs_pasid,
+	.detect_and_reset_hung_queues = mes_v12_0_detect_and_reset_hung_queues,
+};
+
+static int mes_v12_0_allocate_ucode_buffer(struct amdgpu_device *adev,
+					   enum amdgpu_mes_pipe pipe)
+{
+	int r;
+	const struct mes_firmware_header_v1_0 *mes_hdr;
+	const __le32 *fw_data;
+	unsigned fw_size;
+
+	mes_hdr = (const struct mes_firmware_header_v1_0 *)
+		adev->mes.fw[pipe]->data;
+
+	fw_data = (const __le32 *)(adev->mes.fw[pipe]->data +
+		   le32_to_cpu(mes_hdr->mes_ucode_offset_bytes));
+	fw_size = le32_to_cpu(mes_hdr->mes_ucode_size_bytes);
+
+	r = amdgpu_bo_create_reserved(adev, fw_size,
+				      PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_VRAM,
+				      &adev->mes.ucode_fw_obj[pipe],
+				      &adev->mes.ucode_fw_gpu_addr[pipe],
+				      (void **)&adev->mes.ucode_fw_ptr[pipe]);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create mes fw bo\n", r);
+		return r;
+	}
+
+	memcpy(adev->mes.ucode_fw_ptr[pipe], fw_data, fw_size);
+
+	amdgpu_bo_kunmap(adev->mes.ucode_fw_obj[pipe]);
+	amdgpu_bo_unreserve(adev->mes.ucode_fw_obj[pipe]);
+
+	return 0;
+}
+
+static int mes_v12_0_allocate_ucode_data_buffer(struct amdgpu_device *adev,
+						enum amdgpu_mes_pipe pipe)
+{
+	int r;
+	const struct mes_firmware_header_v1_0 *mes_hdr;
+	const __le32 *fw_data;
+	unsigned fw_size;
+
+	mes_hdr = (const struct mes_firmware_header_v1_0 *)
+		adev->mes.fw[pipe]->data;
+
+	fw_data = (const __le32 *)(adev->mes.fw[pipe]->data +
+		   le32_to_cpu(mes_hdr->mes_ucode_data_offset_bytes));
+	fw_size = le32_to_cpu(mes_hdr->mes_ucode_data_size_bytes);
+
+	r = amdgpu_bo_create_reserved(adev, fw_size,
+				      64 * 1024,
+				      AMDGPU_GEM_DOMAIN_VRAM,
+				      &adev->mes.data_fw_obj[pipe],
+				      &adev->mes.data_fw_gpu_addr[pipe],
+				      (void **)&adev->mes.data_fw_ptr[pipe]);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create mes data fw bo\n", r);
+		return r;
+	}
+
+	memcpy(adev->mes.data_fw_ptr[pipe], fw_data, fw_size);
+
+	amdgpu_bo_kunmap(adev->mes.data_fw_obj[pipe]);
+	amdgpu_bo_unreserve(adev->mes.data_fw_obj[pipe]);
+
+	return 0;
+}
+
+static void mes_v12_0_free_ucode_buffers(struct amdgpu_device *adev,
+					 enum amdgpu_mes_pipe pipe)
+{
+	amdgpu_bo_free_kernel(&adev->mes.data_fw_obj[pipe],
+			      &adev->mes.data_fw_gpu_addr[pipe],
+			      (void **)&adev->mes.data_fw_ptr[pipe]);
+
+	amdgpu_bo_free_kernel(&adev->mes.ucode_fw_obj[pipe],
+			      &adev->mes.ucode_fw_gpu_addr[pipe],
+			      (void **)&adev->mes.ucode_fw_ptr[pipe]);
+}
+
+static void mes_v12_0_enable(struct amdgpu_device *adev, bool enable)
+{
+	uint64_t ucode_addr;
+	uint32_t pipe, data = 0;
+
+	if (enable) {
+		mutex_lock(&adev->srbm_mutex);
+		for (pipe = 0; pipe < AMDGPU_MAX_MES_PIPES; pipe++) {
+			soc21_grbm_select(adev, 3, pipe, 0, 0);
+			if (amdgpu_mes_log_enable) {
+				u32 log_size = AMDGPU_MES_LOG_BUFFER_SIZE + AMDGPU_MES_MSCRATCH_SIZE;
+				/* In case uni mes is not enabled, only program for pipe 0 */
+				if (adev->mes.event_log_size >= (pipe + 1) * log_size) {
+					WREG32_SOC15(GC, 0, regCP_MES_MSCRATCH_LO,
+						     lower_32_bits(adev->mes.event_log_gpu_addr +
+						     pipe * log_size + AMDGPU_MES_LOG_BUFFER_SIZE));
+					WREG32_SOC15(GC, 0, regCP_MES_MSCRATCH_HI,
+						     upper_32_bits(adev->mes.event_log_gpu_addr +
+						     pipe * log_size + AMDGPU_MES_LOG_BUFFER_SIZE));
+					dev_info(adev->dev, "Setup CP MES MSCRATCH address : 0x%x. 0x%x\n",
+						 RREG32_SOC15(GC, 0, regCP_MES_MSCRATCH_HI),
+						 RREG32_SOC15(GC, 0, regCP_MES_MSCRATCH_LO));
+				}
+			}
+
+			data = RREG32_SOC15(GC, 0, regCP_MES_CNTL);
+			if (pipe == 0)
+				data = REG_SET_FIELD(data, CP_MES_CNTL, MES_PIPE0_RESET, 1);
+			else
+				data = REG_SET_FIELD(data, CP_MES_CNTL, MES_PIPE1_RESET, 1);
+			WREG32_SOC15(GC, 0, regCP_MES_CNTL, data);
+
+			ucode_addr = adev->mes.uc_start_addr[pipe] >> 2;
+			WREG32_SOC15(GC, 0, regCP_MES_PRGRM_CNTR_START,
+				     lower_32_bits(ucode_addr));
+			WREG32_SOC15(GC, 0, regCP_MES_PRGRM_CNTR_START_HI,
+				     upper_32_bits(ucode_addr));
+
+			/* unhalt MES and activate one pipe each loop */
+			data = REG_SET_FIELD(0, CP_MES_CNTL, MES_PIPE0_ACTIVE, 1);
+			if (pipe)
+				data = REG_SET_FIELD(data, CP_MES_CNTL, MES_PIPE1_ACTIVE, 1);
+			dev_info(adev->dev, "program CP_MES_CNTL : 0x%x\n", data);
+
+			WREG32_SOC15(GC, 0, regCP_MES_CNTL, data);
+
+		}
+		soc21_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+
+		if (amdgpu_emu_mode)
+			msleep(100);
+		else if (adev->enable_uni_mes)
+			udelay(500);
+		else
+			udelay(50);
+	} else {
+		data = RREG32_SOC15(GC, 0, regCP_MES_CNTL);
+		data = REG_SET_FIELD(data, CP_MES_CNTL, MES_PIPE0_ACTIVE, 0);
+		data = REG_SET_FIELD(data, CP_MES_CNTL, MES_PIPE1_ACTIVE, 0);
+		data = REG_SET_FIELD(data, CP_MES_CNTL,
+				     MES_INVALIDATE_ICACHE, 1);
+		data = REG_SET_FIELD(data, CP_MES_CNTL, MES_PIPE0_RESET, 1);
+		data = REG_SET_FIELD(data, CP_MES_CNTL, MES_PIPE1_RESET, 1);
+		data = REG_SET_FIELD(data, CP_MES_CNTL, MES_HALT, 1);
+		WREG32_SOC15(GC, 0, regCP_MES_CNTL, data);
+	}
+}
+
+static void mes_v12_0_set_ucode_start_addr(struct amdgpu_device *adev)
+{
+	uint64_t ucode_addr;
+	int pipe;
+
+	mes_v12_0_enable(adev, false);
+
+	mutex_lock(&adev->srbm_mutex);
+	for (pipe = 0; pipe < AMDGPU_MAX_MES_PIPES; pipe++) {
+		/* me=3, queue=0 */
+		soc21_grbm_select(adev, 3, pipe, 0, 0);
+
+		/* set ucode start address */
+		ucode_addr = adev->mes.uc_start_addr[pipe] >> 2;
+		WREG32_SOC15(GC, 0, regCP_MES_PRGRM_CNTR_START,
+				lower_32_bits(ucode_addr));
+		WREG32_SOC15(GC, 0, regCP_MES_PRGRM_CNTR_START_HI,
+				upper_32_bits(ucode_addr));
+
+		soc21_grbm_select(adev, 0, 0, 0, 0);
+	}
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+/* This function is for backdoor MES firmware */
+static int mes_v12_0_load_microcode(struct amdgpu_device *adev,
+				    enum amdgpu_mes_pipe pipe, bool prime_icache)
+{
+	int r;
+	uint32_t data;
+
+	mes_v12_0_enable(adev, false);
+
+	if (!adev->mes.fw[pipe])
+		return -EINVAL;
+
+	r = mes_v12_0_allocate_ucode_buffer(adev, pipe);
+	if (r)
+		return r;
+
+	r = mes_v12_0_allocate_ucode_data_buffer(adev, pipe);
+	if (r) {
+		mes_v12_0_free_ucode_buffers(adev, pipe);
+		return r;
+	}
+
+	mutex_lock(&adev->srbm_mutex);
+	/* me=3, pipe=0, queue=0 */
+	soc21_grbm_select(adev, 3, pipe, 0, 0);
+
+	WREG32_SOC15(GC, 0, regCP_MES_IC_BASE_CNTL, 0);
+
+	/* set ucode fimrware address */
+	WREG32_SOC15(GC, 0, regCP_MES_IC_BASE_LO,
+		     lower_32_bits(adev->mes.ucode_fw_gpu_addr[pipe]));
+	WREG32_SOC15(GC, 0, regCP_MES_IC_BASE_HI,
+		     upper_32_bits(adev->mes.ucode_fw_gpu_addr[pipe]));
+
+	/* set ucode instruction cache boundary to 2M-1 */
+	WREG32_SOC15(GC, 0, regCP_MES_MIBOUND_LO, 0x1FFFFF);
+
+	/* set ucode data firmware address */
+	WREG32_SOC15(GC, 0, regCP_MES_MDBASE_LO,
+		     lower_32_bits(adev->mes.data_fw_gpu_addr[pipe]));
+	WREG32_SOC15(GC, 0, regCP_MES_MDBASE_HI,
+		     upper_32_bits(adev->mes.data_fw_gpu_addr[pipe]));
+
+	/* Set data cache boundary CP_MES_MDBOUND_LO */
+	WREG32_SOC15(GC, 0, regCP_MES_MDBOUND_LO, 0x7FFFF);
+
+	if (prime_icache) {
+		/* invalidate ICACHE */
+		data = RREG32_SOC15(GC, 0, regCP_MES_IC_OP_CNTL);
+		data = REG_SET_FIELD(data, CP_MES_IC_OP_CNTL, PRIME_ICACHE, 0);
+		data = REG_SET_FIELD(data, CP_MES_IC_OP_CNTL, INVALIDATE_CACHE, 1);
+		WREG32_SOC15(GC, 0, regCP_MES_IC_OP_CNTL, data);
+
+		/* prime the ICACHE. */
+		data = RREG32_SOC15(GC, 0, regCP_MES_IC_OP_CNTL);
+		data = REG_SET_FIELD(data, CP_MES_IC_OP_CNTL, PRIME_ICACHE, 1);
+		WREG32_SOC15(GC, 0, regCP_MES_IC_OP_CNTL, data);
+	}
+
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	return 0;
+}
+
+static int mes_v12_0_allocate_eop_buf(struct amdgpu_device *adev,
+				      enum amdgpu_mes_pipe pipe)
+{
+	int r;
+	u32 *eop;
+
+	r = amdgpu_bo_create_reserved(adev, MES_EOP_SIZE, PAGE_SIZE,
+			      AMDGPU_GEM_DOMAIN_GTT,
+			      &adev->mes.eop_gpu_obj[pipe],
+			      &adev->mes.eop_gpu_addr[pipe],
+			      (void **)&eop);
+	if (r) {
+		dev_warn(adev->dev, "(%d) create EOP bo failed\n", r);
+		return r;
+	}
+
+	memset(eop, 0,
+	       adev->mes.eop_gpu_obj[pipe]->tbo.base.size);
+
+	amdgpu_bo_kunmap(adev->mes.eop_gpu_obj[pipe]);
+	amdgpu_bo_unreserve(adev->mes.eop_gpu_obj[pipe]);
+
+	return 0;
+}
+
+static int mes_v12_0_mqd_init(struct amdgpu_ring *ring)
+{
+	struct v12_compute_mqd *mqd = ring->mqd_ptr;
+	uint64_t hqd_gpu_addr, wb_gpu_addr, eop_base_addr;
+	uint32_t tmp;
+
+	mqd->header = 0xC0310800;
+	mqd->compute_pipelinestat_enable = 0x00000001;
+	mqd->compute_static_thread_mgmt_se0 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se1 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se2 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se3 = 0xffffffff;
+	mqd->compute_misc_reserved = 0x00000007;
+
+	eop_base_addr = ring->eop_gpu_addr >> 8;
+
+	/* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */
+	tmp = regCP_HQD_EOP_CONTROL_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_HQD_EOP_CONTROL, EOP_SIZE,
+			(order_base_2(MES_EOP_SIZE / 4) - 1));
+
+	mqd->cp_hqd_eop_base_addr_lo = lower_32_bits(eop_base_addr);
+	mqd->cp_hqd_eop_base_addr_hi = upper_32_bits(eop_base_addr);
+	mqd->cp_hqd_eop_control = tmp;
+
+	/* disable the queue if it's active */
+	ring->wptr = 0;
+	mqd->cp_hqd_pq_rptr = 0;
+	mqd->cp_hqd_pq_wptr_lo = 0;
+	mqd->cp_hqd_pq_wptr_hi = 0;
+
+	/* set the pointer to the MQD */
+	mqd->cp_mqd_base_addr_lo = ring->mqd_gpu_addr & 0xfffffffc;
+	mqd->cp_mqd_base_addr_hi = upper_32_bits(ring->mqd_gpu_addr);
+
+	/* set MQD vmid to 0 */
+	tmp = regCP_MQD_CONTROL_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_MQD_CONTROL, VMID, 0);
+	mqd->cp_mqd_control = tmp;
+
+	/* set the pointer to the HQD, this is similar CP_RB0_BASE/_HI */
+	hqd_gpu_addr = ring->gpu_addr >> 8;
+	mqd->cp_hqd_pq_base_lo = lower_32_bits(hqd_gpu_addr);
+	mqd->cp_hqd_pq_base_hi = upper_32_bits(hqd_gpu_addr);
+
+	/* set the wb address whether it's enabled or not */
+	wb_gpu_addr = ring->rptr_gpu_addr;
+	mqd->cp_hqd_pq_rptr_report_addr_lo = wb_gpu_addr & 0xfffffffc;
+	mqd->cp_hqd_pq_rptr_report_addr_hi =
+		upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	/* only used if CP_PQ_WPTR_POLL_CNTL.CP_PQ_WPTR_POLL_CNTL__EN_MASK=1 */
+	wb_gpu_addr = ring->wptr_gpu_addr;
+	mqd->cp_hqd_pq_wptr_poll_addr_lo = wb_gpu_addr & 0xfffffff8;
+	mqd->cp_hqd_pq_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	/* set up the HQD, this is similar to CP_RB0_CNTL */
+	tmp = regCP_HQD_PQ_CONTROL_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, QUEUE_SIZE,
+			    (order_base_2(ring->ring_size / 4) - 1));
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, RPTR_BLOCK_SIZE,
+			    ((order_base_2(AMDGPU_GPU_PAGE_SIZE / 4) - 1) << 8));
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, UNORD_DISPATCH, 1);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, TUNNEL_DISPATCH, 0);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, PRIV_STATE, 1);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, KMD_QUEUE, 1);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, NO_UPDATE_RPTR, 1);
+	mqd->cp_hqd_pq_control = tmp;
+
+	/* enable doorbell */
+	tmp = 0;
+	if (ring->use_doorbell) {
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_OFFSET, ring->doorbell_index);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_EN, 1);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_SOURCE, 0);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_HIT, 0);
+	} else {
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_EN, 0);
+	}
+	mqd->cp_hqd_pq_doorbell_control = tmp;
+
+	mqd->cp_hqd_vmid = 0;
+	/* activate the queue */
+	mqd->cp_hqd_active = 1;
+
+	tmp = regCP_HQD_PERSISTENT_STATE_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PERSISTENT_STATE,
+			    PRELOAD_SIZE, 0x55);
+	mqd->cp_hqd_persistent_state = tmp;
+
+	mqd->cp_hqd_ib_control = regCP_HQD_IB_CONTROL_DEFAULT;
+	mqd->cp_hqd_iq_timer = regCP_HQD_IQ_TIMER_DEFAULT;
+	mqd->cp_hqd_quantum = regCP_HQD_QUANTUM_DEFAULT;
+
+	/*
+	 * Set CP_HQD_GFX_CONTROL.DB_UPDATED_MSG_EN[15] to enable unmapped
+	 * doorbell handling. This is a reserved CP internal register can
+	 * not be accesss by others
+	 */
+	mqd->reserved_184 = BIT(15);
+
+	return 0;
+}
+
+static void mes_v12_0_queue_init_register(struct amdgpu_ring *ring)
+{
+	struct v12_compute_mqd *mqd = ring->mqd_ptr;
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t data = 0;
+
+	mutex_lock(&adev->srbm_mutex);
+	soc21_grbm_select(adev, 3, ring->pipe, 0, 0);
+
+	/* set CP_HQD_VMID.VMID = 0. */
+	data = RREG32_SOC15(GC, 0, regCP_HQD_VMID);
+	data = REG_SET_FIELD(data, CP_HQD_VMID, VMID, 0);
+	WREG32_SOC15(GC, 0, regCP_HQD_VMID, data);
+
+	/* set CP_HQD_PQ_DOORBELL_CONTROL.DOORBELL_EN=0 */
+	data = RREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL);
+	data = REG_SET_FIELD(data, CP_HQD_PQ_DOORBELL_CONTROL,
+			     DOORBELL_EN, 0);
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL, data);
+
+	/* set CP_MQD_BASE_ADDR/HI with the MQD base address */
+	WREG32_SOC15(GC, 0, regCP_MQD_BASE_ADDR, mqd->cp_mqd_base_addr_lo);
+	WREG32_SOC15(GC, 0, regCP_MQD_BASE_ADDR_HI, mqd->cp_mqd_base_addr_hi);
+
+	/* set CP_MQD_CONTROL.VMID=0 */
+	data = RREG32_SOC15(GC, 0, regCP_MQD_CONTROL);
+	data = REG_SET_FIELD(data, CP_MQD_CONTROL, VMID, 0);
+	WREG32_SOC15(GC, 0, regCP_MQD_CONTROL, 0);
+
+	/* set CP_HQD_PQ_BASE/HI with the ring buffer base address */
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_BASE, mqd->cp_hqd_pq_base_lo);
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_BASE_HI, mqd->cp_hqd_pq_base_hi);
+
+	/* set CP_HQD_PQ_RPTR_REPORT_ADDR/HI */
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_RPTR_REPORT_ADDR,
+		     mqd->cp_hqd_pq_rptr_report_addr_lo);
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_RPTR_REPORT_ADDR_HI,
+		     mqd->cp_hqd_pq_rptr_report_addr_hi);
+
+	/* set CP_HQD_PQ_CONTROL */
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_CONTROL, mqd->cp_hqd_pq_control);
+
+	/* set CP_HQD_PQ_WPTR_POLL_ADDR/HI */
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_WPTR_POLL_ADDR,
+		     mqd->cp_hqd_pq_wptr_poll_addr_lo);
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_WPTR_POLL_ADDR_HI,
+		     mqd->cp_hqd_pq_wptr_poll_addr_hi);
+
+	/* set CP_HQD_PQ_DOORBELL_CONTROL */
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL,
+		     mqd->cp_hqd_pq_doorbell_control);
+
+	/* set CP_HQD_PERSISTENT_STATE.PRELOAD_SIZE=0x53 */
+	WREG32_SOC15(GC, 0, regCP_HQD_PERSISTENT_STATE, mqd->cp_hqd_persistent_state);
+
+	/* set CP_HQD_ACTIVE.ACTIVE=1 */
+	WREG32_SOC15(GC, 0, regCP_HQD_ACTIVE, mqd->cp_hqd_active);
+
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static int mes_v12_0_kiq_enable_queue(struct amdgpu_device *adev)
+{
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[0];
+	struct amdgpu_ring *kiq_ring = &adev->gfx.kiq[0].ring;
+	int r;
+
+	if (!kiq->pmf || !kiq->pmf->kiq_map_queues)
+		return -EINVAL;
+
+	r = amdgpu_ring_alloc(kiq_ring, kiq->pmf->map_queues_size);
+	if (r) {
+		DRM_ERROR("Failed to lock KIQ (%d).\n", r);
+		return r;
+	}
+
+	kiq->pmf->kiq_map_queues(kiq_ring, &adev->mes.ring[0]);
+
+	r = amdgpu_ring_test_ring(kiq_ring);
+	if (r) {
+		DRM_ERROR("kfq enable failed\n");
+		kiq_ring->sched.ready = false;
+	}
+	return r;
+}
+
+static int mes_v12_0_queue_init(struct amdgpu_device *adev,
+				enum amdgpu_mes_pipe pipe)
+{
+	struct amdgpu_ring *ring;
+	int r;
+
+	if (!adev->enable_uni_mes && pipe == AMDGPU_MES_KIQ_PIPE)
+		ring = &adev->gfx.kiq[0].ring;
+	else
+		ring = &adev->mes.ring[pipe];
+
+	if ((adev->enable_uni_mes || pipe == AMDGPU_MES_SCHED_PIPE) &&
+	    (amdgpu_in_reset(adev) || adev->in_suspend)) {
+		*(ring->wptr_cpu_addr) = 0;
+		*(ring->rptr_cpu_addr) = 0;
+		amdgpu_ring_clear_ring(ring);
+	}
+
+	r = mes_v12_0_mqd_init(ring);
+	if (r)
+		return r;
+
+	if (pipe == AMDGPU_MES_SCHED_PIPE) {
+		if (adev->enable_uni_mes)
+			r = amdgpu_mes_map_legacy_queue(adev, ring);
+		else
+			r = mes_v12_0_kiq_enable_queue(adev);
+		if (r)
+			return r;
+	} else {
+		mes_v12_0_queue_init_register(ring);
+	}
+
+	if (((pipe == AMDGPU_MES_SCHED_PIPE) && !adev->mes.sched_version) ||
+	    ((pipe == AMDGPU_MES_KIQ_PIPE) && !adev->mes.kiq_version)) {
+		/* get MES scheduler/KIQ versions */
+		mutex_lock(&adev->srbm_mutex);
+		soc21_grbm_select(adev, 3, pipe, 0, 0);
+
+		if (pipe == AMDGPU_MES_SCHED_PIPE)
+			adev->mes.sched_version = RREG32_SOC15(GC, 0, regCP_MES_GP3_LO);
+		else if (pipe == AMDGPU_MES_KIQ_PIPE && adev->enable_mes_kiq)
+			adev->mes.kiq_version = RREG32_SOC15(GC, 0, regCP_MES_GP3_LO);
+
+		soc21_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+	}
+
+	return 0;
+}
+
+static int mes_v12_0_ring_init(struct amdgpu_device *adev, int pipe)
+{
+	struct amdgpu_ring *ring;
+
+	ring = &adev->mes.ring[pipe];
+
+	ring->funcs = &mes_v12_0_ring_funcs;
+
+	ring->me = 3;
+	ring->pipe = pipe;
+	ring->queue = 0;
+
+	ring->ring_obj = NULL;
+	ring->use_doorbell = true;
+	ring->eop_gpu_addr = adev->mes.eop_gpu_addr[pipe];
+	ring->no_scheduler = true;
+	sprintf(ring->name, "mes_%d.%d.%d", ring->me, ring->pipe, ring->queue);
+
+	if (pipe == AMDGPU_MES_SCHED_PIPE)
+		ring->doorbell_index = adev->doorbell_index.mes_ring0 << 1;
+	else
+		ring->doorbell_index = adev->doorbell_index.mes_ring1 << 1;
+
+	return amdgpu_ring_init(adev, ring, 1024, NULL, 0,
+				AMDGPU_RING_PRIO_DEFAULT, NULL);
+}
+
+static int mes_v12_0_kiq_ring_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+
+	spin_lock_init(&adev->gfx.kiq[0].ring_lock);
+
+	ring = &adev->gfx.kiq[0].ring;
+
+	ring->me = 3;
+	ring->pipe = 1;
+	ring->queue = 0;
+
+	ring->adev = NULL;
+	ring->ring_obj = NULL;
+	ring->use_doorbell = true;
+	ring->doorbell_index = adev->doorbell_index.mes_ring1 << 1;
+	ring->eop_gpu_addr = adev->mes.eop_gpu_addr[AMDGPU_MES_KIQ_PIPE];
+	ring->no_scheduler = true;
+	sprintf(ring->name, "mes_kiq_%d.%d.%d",
+		ring->me, ring->pipe, ring->queue);
+
+	return amdgpu_ring_init(adev, ring, 1024, NULL, 0,
+				AMDGPU_RING_PRIO_DEFAULT, NULL);
+}
+
+static int mes_v12_0_mqd_sw_init(struct amdgpu_device *adev,
+				 enum amdgpu_mes_pipe pipe)
+{
+	int r, mqd_size = sizeof(struct v12_compute_mqd);
+	struct amdgpu_ring *ring;
+
+	if (!adev->enable_uni_mes && pipe == AMDGPU_MES_KIQ_PIPE)
+		ring = &adev->gfx.kiq[0].ring;
+	else
+		ring = &adev->mes.ring[pipe];
+
+	if (ring->mqd_obj)
+		return 0;
+
+	r = amdgpu_bo_create_kernel(adev, mqd_size, PAGE_SIZE,
+				    AMDGPU_GEM_DOMAIN_GTT, &ring->mqd_obj,
+				    &ring->mqd_gpu_addr, &ring->mqd_ptr);
+	if (r) {
+		dev_warn(adev->dev, "failed to create ring mqd bo (%d)", r);
+		return r;
+	}
+
+	memset(ring->mqd_ptr, 0, mqd_size);
+
+	/* prepare MQD backup */
+	adev->mes.mqd_backup[pipe] = kmalloc(mqd_size, GFP_KERNEL);
+	if (!adev->mes.mqd_backup[pipe])
+		dev_warn(adev->dev,
+			 "no memory to create MQD backup for ring %s\n",
+			 ring->name);
+
+	return 0;
+}
+
+static int mes_v12_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int pipe, r;
+
+	adev->mes.funcs = &mes_v12_0_funcs;
+	adev->mes.kiq_hw_init = &mes_v12_0_kiq_hw_init;
+	adev->mes.kiq_hw_fini = &mes_v12_0_kiq_hw_fini;
+	adev->mes.enable_legacy_queue_map = true;
+
+	adev->mes.event_log_size = adev->enable_uni_mes ?
+		(AMDGPU_MAX_MES_PIPES * (AMDGPU_MES_LOG_BUFFER_SIZE + AMDGPU_MES_MSCRATCH_SIZE)) :
+		(AMDGPU_MES_LOG_BUFFER_SIZE + AMDGPU_MES_MSCRATCH_SIZE);
+	r = amdgpu_mes_init(adev);
+	if (r)
+		return r;
+
+	for (pipe = 0; pipe < AMDGPU_MAX_MES_PIPES; pipe++) {
+		r = mes_v12_0_allocate_eop_buf(adev, pipe);
+		if (r)
+			return r;
+
+		r = mes_v12_0_mqd_sw_init(adev, pipe);
+		if (r)
+			return r;
+
+		if (!adev->enable_uni_mes && pipe == AMDGPU_MES_KIQ_PIPE) {
+			r = mes_v12_0_kiq_ring_init(adev);
+		}
+		else {
+			r = mes_v12_0_ring_init(adev, pipe);
+			if (r)
+				return r;
+			r = amdgpu_bo_create_kernel(adev, AMDGPU_GPU_PAGE_SIZE, PAGE_SIZE,
+						    AMDGPU_GEM_DOMAIN_VRAM,
+						    &adev->mes.resource_1[pipe],
+						    &adev->mes.resource_1_gpu_addr[pipe],
+						    &adev->mes.resource_1_addr[pipe]);
+			if (r) {
+				dev_err(adev->dev, "(%d) failed to create mes resource_1 bo pipe[%d]\n", r, pipe);
+				return r;
+			}
+		}
+	}
+
+	return 0;
+}
+
+static int mes_v12_0_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int pipe;
+
+	for (pipe = 0; pipe < AMDGPU_MAX_MES_PIPES; pipe++) {
+		amdgpu_bo_free_kernel(&adev->mes.resource_1[pipe],
+				      &adev->mes.resource_1_gpu_addr[pipe],
+				      &adev->mes.resource_1_addr[pipe]);
+
+		kfree(adev->mes.mqd_backup[pipe]);
+
+		amdgpu_bo_free_kernel(&adev->mes.eop_gpu_obj[pipe],
+				      &adev->mes.eop_gpu_addr[pipe],
+				      NULL);
+		amdgpu_ucode_release(&adev->mes.fw[pipe]);
+
+		if (adev->enable_uni_mes || pipe == AMDGPU_MES_SCHED_PIPE) {
+			amdgpu_bo_free_kernel(&adev->mes.ring[pipe].mqd_obj,
+					      &adev->mes.ring[pipe].mqd_gpu_addr,
+					      &adev->mes.ring[pipe].mqd_ptr);
+			amdgpu_ring_fini(&adev->mes.ring[pipe]);
+		}
+	}
+
+	if (!adev->enable_uni_mes) {
+		amdgpu_bo_free_kernel(&adev->gfx.kiq[0].ring.mqd_obj,
+				      &adev->gfx.kiq[0].ring.mqd_gpu_addr,
+				      &adev->gfx.kiq[0].ring.mqd_ptr);
+		amdgpu_ring_fini(&adev->gfx.kiq[0].ring);
+	}
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+		mes_v12_0_free_ucode_buffers(adev, AMDGPU_MES_KIQ_PIPE);
+		mes_v12_0_free_ucode_buffers(adev, AMDGPU_MES_SCHED_PIPE);
+	}
+
+	amdgpu_mes_fini(adev);
+	return 0;
+}
+
+static void mes_v12_0_kiq_dequeue_sched(struct amdgpu_device *adev)
+{
+	uint32_t data;
+	int i;
+
+	mutex_lock(&adev->srbm_mutex);
+	soc21_grbm_select(adev, 3, AMDGPU_MES_SCHED_PIPE, 0, 0);
+
+	/* disable the queue if it's active */
+	if (RREG32_SOC15(GC, 0, regCP_HQD_ACTIVE) & 1) {
+		WREG32_SOC15(GC, 0, regCP_HQD_DEQUEUE_REQUEST, 1);
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (!(RREG32_SOC15(GC, 0, regCP_HQD_ACTIVE) & 1))
+				break;
+			udelay(1);
+		}
+	}
+	data = RREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL);
+	data = REG_SET_FIELD(data, CP_HQD_PQ_DOORBELL_CONTROL,
+				DOORBELL_EN, 0);
+	data = REG_SET_FIELD(data, CP_HQD_PQ_DOORBELL_CONTROL,
+				DOORBELL_HIT, 1);
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL, data);
+
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL, 0);
+
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_WPTR_LO, 0);
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_WPTR_HI, 0);
+	WREG32_SOC15(GC, 0, regCP_HQD_PQ_RPTR, 0);
+
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	adev->mes.ring[0].sched.ready = false;
+}
+
+static void mes_v12_0_kiq_setting(struct amdgpu_ring *ring)
+{
+	uint32_t tmp;
+	struct amdgpu_device *adev = ring->adev;
+
+	/* tell RLC which is KIQ queue */
+	tmp = RREG32_SOC15(GC, 0, regRLC_CP_SCHEDULERS);
+	tmp &= 0xffffff00;
+	tmp |= (ring->me << 5) | (ring->pipe << 3) | (ring->queue);
+	WREG32_SOC15(GC, 0, regRLC_CP_SCHEDULERS, tmp | 0x80);
+}
+
+static int mes_v12_0_kiq_hw_init(struct amdgpu_device *adev)
+{
+	int r = 0;
+	struct amdgpu_ip_block *ip_block;
+
+	if (adev->enable_uni_mes)
+		mes_v12_0_kiq_setting(&adev->mes.ring[AMDGPU_MES_KIQ_PIPE]);
+	else
+		mes_v12_0_kiq_setting(&adev->gfx.kiq[0].ring);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+
+		r = mes_v12_0_load_microcode(adev, AMDGPU_MES_SCHED_PIPE, false);
+		if (r) {
+			DRM_ERROR("failed to load MES fw, r=%d\n", r);
+			return r;
+		}
+
+		r = mes_v12_0_load_microcode(adev, AMDGPU_MES_KIQ_PIPE, true);
+		if (r) {
+			DRM_ERROR("failed to load MES kiq fw, r=%d\n", r);
+			return r;
+		}
+
+		mes_v12_0_set_ucode_start_addr(adev);
+
+	} else if (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO)
+		mes_v12_0_set_ucode_start_addr(adev);
+
+	mes_v12_0_enable(adev, true);
+
+	ip_block = amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_MES);
+	if (unlikely(!ip_block)) {
+		dev_err(adev->dev, "Failed to get MES handle\n");
+		return -EINVAL;
+	}
+
+	r = mes_v12_0_queue_init(adev, AMDGPU_MES_KIQ_PIPE);
+	if (r)
+		goto failure;
+
+	if (adev->enable_uni_mes) {
+		r = mes_v12_0_set_hw_resources(&adev->mes, AMDGPU_MES_KIQ_PIPE);
+		if (r)
+			goto failure;
+
+		mes_v12_0_set_hw_resources_1(&adev->mes, AMDGPU_MES_KIQ_PIPE);
+	}
+
+	if (adev->mes.enable_legacy_queue_map) {
+		r = mes_v12_0_hw_init(ip_block);
+		if (r)
+			goto failure;
+	}
+
+	return r;
+
+failure:
+	mes_v12_0_hw_fini(ip_block);
+	return r;
+}
+
+static int mes_v12_0_kiq_hw_fini(struct amdgpu_device *adev)
+{
+	if (adev->mes.ring[0].sched.ready) {
+		if (adev->enable_uni_mes)
+			amdgpu_mes_unmap_legacy_queue(adev,
+				      &adev->mes.ring[AMDGPU_MES_SCHED_PIPE],
+				      RESET_QUEUES, 0, 0);
+		else
+			mes_v12_0_kiq_dequeue_sched(adev);
+
+		adev->mes.ring[0].sched.ready = false;
+	}
+
+	mes_v12_0_enable(adev, false);
+
+	return 0;
+}
+
+static int mes_v12_0_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (adev->mes.ring[0].sched.ready)
+		goto out;
+
+	if (!adev->enable_mes_kiq) {
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+			r = mes_v12_0_load_microcode(adev,
+					     AMDGPU_MES_SCHED_PIPE, true);
+			if (r) {
+				DRM_ERROR("failed to MES fw, r=%d\n", r);
+				return r;
+			}
+
+			mes_v12_0_set_ucode_start_addr(adev);
+
+		} else if (adev->firmware.load_type ==
+			   AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO) {
+
+			mes_v12_0_set_ucode_start_addr(adev);
+		}
+
+		mes_v12_0_enable(adev, true);
+	}
+
+	/* Enable the MES to handle doorbell ring on unmapped queue */
+	mes_v12_0_enable_unmapped_doorbell_handling(&adev->mes, true);
+
+	r = mes_v12_0_queue_init(adev, AMDGPU_MES_SCHED_PIPE);
+	if (r)
+		goto failure;
+
+	r = mes_v12_0_set_hw_resources(&adev->mes, AMDGPU_MES_SCHED_PIPE);
+	if (r)
+		goto failure;
+
+	if ((adev->mes.sched_version & AMDGPU_MES_VERSION_MASK) >= 0x4b)
+		mes_v12_0_set_hw_resources_1(&adev->mes, AMDGPU_MES_SCHED_PIPE);
+
+	mes_v12_0_init_aggregated_doorbell(&adev->mes);
+
+	r = mes_v12_0_query_sched_status(&adev->mes, AMDGPU_MES_SCHED_PIPE);
+	if (r) {
+		DRM_ERROR("MES is busy\n");
+		goto failure;
+	}
+
+	r = amdgpu_mes_update_enforce_isolation(adev);
+	if (r)
+		goto failure;
+
+out:
+	/*
+	 * Disable KIQ ring usage from the driver once MES is enabled.
+	 * MES uses KIQ ring exclusively so driver cannot access KIQ ring
+	 * with MES enabled.
+	 */
+	adev->gfx.kiq[0].ring.sched.ready = false;
+	adev->mes.ring[0].sched.ready = true;
+
+	return 0;
+
+failure:
+	mes_v12_0_hw_fini(ip_block);
+	return r;
+}
+
+static int mes_v12_0_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	return 0;
+}
+
+static int mes_v12_0_suspend(struct amdgpu_ip_block *ip_block)
+{
+	return mes_v12_0_hw_fini(ip_block);
+}
+
+static int mes_v12_0_resume(struct amdgpu_ip_block *ip_block)
+{
+	return mes_v12_0_hw_init(ip_block);
+}
+
+static int mes_v12_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int pipe, r;
+
+	adev->mes.hung_queue_db_array_size =
+		MES12_HUNG_DB_OFFSET_ARRAY_SIZE;
+	for (pipe = 0; pipe < AMDGPU_MAX_MES_PIPES; pipe++) {
+		r = amdgpu_mes_init_microcode(adev, pipe);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+static const struct amd_ip_funcs mes_v12_0_ip_funcs = {
+	.name = "mes_v12_0",
+	.early_init = mes_v12_0_early_init,
+	.late_init = NULL,
+	.sw_init = mes_v12_0_sw_init,
+	.sw_fini = mes_v12_0_sw_fini,
+	.hw_init = mes_v12_0_hw_init,
+	.hw_fini = mes_v12_0_hw_fini,
+	.suspend = mes_v12_0_suspend,
+	.resume = mes_v12_0_resume,
+};
+
+const struct amdgpu_ip_block_version mes_v12_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_MES,
+	.major = 12,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &mes_v12_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v12_0.h b/drivers/gpu/drm/amd/amdgpu/mes_v12_0.h
new file mode 100644
index 000000000000..ac3740f353aa
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mes_v12_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __MES_V12_0_H__
+#define __MES_V12_0_H__
+
+extern const struct amdgpu_ip_block_version mes_v12_0_ip_block;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c
index 74cb647da30e..243eabda0607 100644
--- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c
@@ -21,62 +21,65 @@
  *
  */
 #include "amdgpu.h"
+#include "amdgpu_ras.h"
 #include "mmhub_v1_0.h"
 
-#include "vega10/soc15ip.h"
-#include "vega10/MMHUB/mmhub_1_0_offset.h"
-#include "vega10/MMHUB/mmhub_1_0_sh_mask.h"
-#include "vega10/MMHUB/mmhub_1_0_default.h"
-#include "vega10/ATHUB/athub_1_0_offset.h"
-#include "vega10/ATHUB/athub_1_0_sh_mask.h"
-#include "vega10/ATHUB/athub_1_0_default.h"
-#include "vega10/vega10_enum.h"
-
+#include "mmhub/mmhub_1_0_offset.h"
+#include "mmhub/mmhub_1_0_sh_mask.h"
+#include "mmhub/mmhub_1_0_default.h"
+#include "vega10_enum.h"
+#include "soc15.h"
 #include "soc15_common.h"
 
 #define mmDAGB0_CNTL_MISC2_RV 0x008f
 #define mmDAGB0_CNTL_MISC2_RV_BASE_IDX 0
 
-u64 mmhub_v1_0_get_fb_location(struct amdgpu_device *adev)
+static u64 mmhub_v1_0_get_fb_location(struct amdgpu_device *adev)
 {
 	u64 base = RREG32_SOC15(MMHUB, 0, mmMC_VM_FB_LOCATION_BASE);
+	u64 top = RREG32_SOC15(MMHUB, 0, mmMC_VM_FB_LOCATION_TOP);
 
 	base &= MC_VM_FB_LOCATION_BASE__FB_BASE_MASK;
 	base <<= 24;
 
+	top &= MC_VM_FB_LOCATION_TOP__FB_TOP_MASK;
+	top <<= 24;
+
+	adev->gmc.fb_start = base;
+	adev->gmc.fb_end = top;
+
 	return base;
 }
 
-static void mmhub_v1_0_init_gart_pt_regs(struct amdgpu_device *adev)
+static void mmhub_v1_0_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid,
+				uint64_t page_table_base)
 {
-	uint64_t value;
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
 
-	BUG_ON(adev->gart.table_addr & (~0x0000FFFFFFFFF000ULL));
-	value = adev->gart.table_addr - adev->mc.vram_start +
-		adev->vm_manager.vram_base_offset;
-	value &= 0x0000FFFFFFFFF000ULL;
-	value |= 0x1; /* valid bit */
+	WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+			    hub->ctx_addr_distance * vmid,
+			    lower_32_bits(page_table_base));
 
-	WREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
-		     lower_32_bits(value));
-
-	WREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
-		     upper_32_bits(value));
+	WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+			    hub->ctx_addr_distance * vmid,
+			    upper_32_bits(page_table_base));
 }
 
 static void mmhub_v1_0_init_gart_aperture_regs(struct amdgpu_device *adev)
 {
-	mmhub_v1_0_init_gart_pt_regs(adev);
+	uint64_t pt_base = amdgpu_gmc_pd_addr(adev->gart.bo);
+
+	mmhub_v1_0_setup_vm_pt_regs(adev, 0, pt_base);
 
 	WREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
-		     (u32)(adev->mc.gart_start >> 12));
+		     (u32)(adev->gmc.gart_start >> 12));
 	WREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
-		     (u32)(adev->mc.gart_start >> 44));
+		     (u32)(adev->gmc.gart_start >> 44));
 
 	WREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
-		     (u32)(adev->mc.gart_end >> 12));
+		     (u32)(adev->gmc.gart_end >> 12));
 	WREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
-		     (u32)(adev->mc.gart_end >> 44));
+		     (u32)(adev->gmc.gart_end >> 44));
 }
 
 static void mmhub_v1_0_init_system_aperture_regs(struct amdgpu_device *adev)
@@ -84,20 +87,36 @@ static void mmhub_v1_0_init_system_aperture_regs(struct amdgpu_device *adev)
 	uint64_t value;
 	uint32_t tmp;
 
-	/* Disable AGP. */
+	/* Program the AGP BAR */
 	WREG32_SOC15(MMHUB, 0, mmMC_VM_AGP_BASE, 0);
-	WREG32_SOC15(MMHUB, 0, mmMC_VM_AGP_TOP, 0);
-	WREG32_SOC15(MMHUB, 0, mmMC_VM_AGP_BOT, 0x00FFFFFF);
+	WREG32_SOC15(MMHUB, 0, mmMC_VM_AGP_BOT, adev->gmc.agp_start >> 24);
+	WREG32_SOC15(MMHUB, 0, mmMC_VM_AGP_TOP, adev->gmc.agp_end >> 24);
 
 	/* Program the system aperture low logical page number. */
 	WREG32_SOC15(MMHUB, 0, mmMC_VM_SYSTEM_APERTURE_LOW_ADDR,
-		     adev->mc.vram_start >> 18);
-	WREG32_SOC15(MMHUB, 0, mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
-		     adev->mc.vram_end >> 18);
+		     min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
+
+	if (adev->apu_flags & (AMD_APU_IS_RAVEN2 |
+			       AMD_APU_IS_RENOIR |
+			       AMD_APU_IS_GREEN_SARDINE))
+		/*
+		 * Raven2 has a HW issue that it is unable to use the vram which
+		 * is out of MC_VM_SYSTEM_APERTURE_HIGH_ADDR. So here is the
+		 * workaround that increase system aperture high address (add 1)
+		 * to get rid of the VM fault and hardware hang.
+		 */
+		WREG32_SOC15(MMHUB, 0, mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+			     max((adev->gmc.fb_end >> 18) + 0x1,
+				 adev->gmc.agp_end >> 18));
+	else
+		WREG32_SOC15(MMHUB, 0, mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+			     max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
+
+	if (amdgpu_sriov_vf(adev))
+		return;
 
 	/* Set default page address. */
-	value = adev->vram_scratch.gpu_addr - adev->mc.vram_start +
-		adev->vm_manager.vram_base_offset;
+	value = amdgpu_gmc_vram_mc2pa(adev, adev->mem_scratch.gpu_addr);
 	WREG32_SOC15(MMHUB, 0, mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
 		     (u32)(value >> 12));
 	WREG32_SOC15(MMHUB, 0, mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
@@ -105,9 +124,9 @@ static void mmhub_v1_0_init_system_aperture_regs(struct amdgpu_device *adev)
 
 	/* Program "protection fault". */
 	WREG32_SOC15(MMHUB, 0, mmVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
-		     (u32)(adev->dummy_page.addr >> 12));
+		     (u32)(adev->dummy_page_addr >> 12));
 	WREG32_SOC15(MMHUB, 0, mmVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
-		     (u32)((u64)adev->dummy_page.addr >> 44));
+		     (u32)((u64)adev->dummy_page_addr >> 44));
 
 	tmp = RREG32_SOC15(MMHUB, 0, mmVM_L2_PROTECTION_FAULT_CNTL2);
 	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL2,
@@ -128,7 +147,6 @@ static void mmhub_v1_0_init_tlb_regs(struct amdgpu_device *adev)
 			    ENABLE_ADVANCED_DRIVER_MODEL, 1);
 	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
 			    SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
-	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ECO_BITS, 0);
 	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
 			    MTYPE, MTYPE_UC);/* XXX for emulation. */
 	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ATC_EN, 1);
@@ -140,6 +158,9 @@ static void mmhub_v1_0_init_cache_regs(struct amdgpu_device *adev)
 {
 	uint32_t tmp;
 
+	if (amdgpu_sriov_vf(adev))
+		return;
+
 	/* Setup L2 cache */
 	tmp = RREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL);
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 1);
@@ -147,7 +168,7 @@ static void mmhub_v1_0_init_cache_regs(struct amdgpu_device *adev)
 	/* XXX for emulation, Refer to closed source code.*/
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, L2_PDE0_CACHE_TAG_GENERATION_MODE,
 			    0);
-	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 0);
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0);
 	WREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL, tmp);
@@ -158,8 +179,15 @@ static void mmhub_v1_0_init_cache_regs(struct amdgpu_device *adev)
 	WREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL2, tmp);
 
 	tmp = mmVM_L2_CNTL3_DEFAULT;
-	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 9);
-	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+	if (adev->gmc.translate_further) {
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 12);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+	} else {
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 9);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+	}
 	WREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL3, tmp);
 
 	tmp = mmVM_L2_CNTL4_DEFAULT;
@@ -175,11 +203,16 @@ static void mmhub_v1_0_enable_system_domain(struct amdgpu_device *adev)
 	tmp = RREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_CNTL);
 	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
 	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL,
+			    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
 	WREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_CNTL, tmp);
 }
 
 static void mmhub_v1_0_disable_identity_aperture(struct amdgpu_device *adev)
 {
+	if (amdgpu_sriov_vf(adev))
+		return;
+
 	WREG32_SOC15(MMHUB, 0, mmVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32,
 		     0XFFFFFFFF);
 	WREG32_SOC15(MMHUB, 0, mmVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32,
@@ -196,285 +229,137 @@ static void mmhub_v1_0_disable_identity_aperture(struct amdgpu_device *adev)
 		     0);
 }
 
+static void mmhub_v1_0_init_saw(struct amdgpu_device *adev)
+{
+	uint64_t pt_base = amdgpu_gmc_pd_addr(adev->gart.bo);
+	uint32_t tmp;
+
+	/* VM_9_X_REGISTER_VM_L2_SAW_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32 */
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_SAW_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+			    lower_32_bits(pt_base >> 12));
+
+	/* VM_9_X_REGISTER_VM_L2_SAW_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32 */
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_SAW_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+			    upper_32_bits(pt_base >> 12));
+
+	/* VM_9_X_REGISTER_VM_L2_SAW_CONTEXT0_PAGE_TABLE_START_ADDR_LO32 */
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_SAW_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+			    (u32)(adev->gmc.gart_start >> 12));
+
+	/* VM_9_X_REGISTER_VM_L2_SAW_CONTEXT0_PAGE_TABLE_START_ADDR_HI32 */
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_SAW_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+			    (u32)(adev->gmc.gart_start >> 44));
+
+	/* VM_9_X_REGISTER_VM_L2_SAW_CONTEXT0_PAGE_TABLE_END_ADDR_LO32 */
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_SAW_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+			    (u32)(adev->gmc.gart_end >> 12));
+
+	/* VM_9_X_REGISTER_VM_L2_SAW_CONTEXT0_PAGE_TABLE_END_ADDR_HI32 */
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_SAW_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+			    (u32)(adev->gmc.gart_end >> 44));
+
+	/* Program SAW CONTEXT0 CNTL */
+	tmp = RREG32_SOC15(MMHUB, 0, mmVM_L2_SAW_CONTEXT0_CNTL);
+	tmp |= 1 << CONTEXT0_CNTL_ENABLE_OFFSET;
+	tmp &= ~(3 << CONTEXT0_CNTL_PAGE_TABLE_DEPTH_OFFSET);
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_SAW_CONTEXT0_CNTL, tmp);
+
+	/* Disable all Contexts except Context0 */
+	tmp = 0xfffe;
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_SAW_CONTEXTS_DISABLE, tmp);
+
+	/* Program SAW CNTL4 */
+	tmp = RREG32_SOC15(MMHUB, 0, mmVM_L2_SAW_CNTL4);
+	tmp |= 1 << VMC_TAP_PDE_REQUEST_SNOOP_OFFSET;
+	tmp |= 1 << VMC_TAP_PTE_REQUEST_SNOOP_OFFSET;
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_SAW_CNTL4, tmp);
+}
+
 static void mmhub_v1_0_setup_vmid_config(struct amdgpu_device *adev)
 {
-	int i;
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	unsigned num_level, block_size;
 	uint32_t tmp;
+	int i;
+
+	num_level = adev->vm_manager.num_level;
+	block_size = adev->vm_manager.block_size;
+	if (adev->gmc.translate_further)
+		num_level -= 1;
+	else
+		block_size -= 9;
 
 	for (i = 0; i <= 14; i++) {
-		tmp = RREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_CNTL, i);
-		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
-				ENABLE_CONTEXT, 1);
-		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
-				PAGE_TABLE_DEPTH, adev->vm_manager.num_level);
+		tmp = RREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_CNTL, i * hub->ctx_distance);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH,
+				    num_level);
 		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
-				RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+				    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
 		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
-				DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+				    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    1);
 		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
-				PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+				    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
 		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
-				VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+				    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
 		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
-				READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+				    READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
 		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
-				WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+				    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
 		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
-				EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+				    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
 		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
-				PAGE_TABLE_BLOCK_SIZE,
-				adev->vm_manager.block_size - 9);
+				    PAGE_TABLE_BLOCK_SIZE,
+				    block_size);
 		/* Send no-retry XNACK on fault to suppress VM fault storm. */
 		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
-				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
-		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_CNTL, i, tmp);
-		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32, i*2, 0);
-		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32, i*2, 0);
-		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32, i*2,
-			lower_32_bits(adev->vm_manager.max_pfn - 1));
-		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32, i*2,
-			upper_32_bits(adev->vm_manager.max_pfn - 1));
+				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
+				    !adev->gmc.noretry);
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_CNTL,
+				    i * hub->ctx_distance, tmp);
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32,
+				    i * hub->ctx_addr_distance,
+				    lower_32_bits(adev->vm_manager.max_pfn - 1));
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32,
+				    i * hub->ctx_addr_distance,
+				    upper_32_bits(adev->vm_manager.max_pfn - 1));
 	}
+
+	if (amdgpu_ip_version(adev, ISP_HWIP, 0))
+		mmhub_v1_0_init_saw(adev);
 }
 
 static void mmhub_v1_0_program_invalidation(struct amdgpu_device *adev)
 {
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
 	unsigned i;
 
 	for (i = 0; i < 18; ++i) {
 		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
-				    2 * i, 0xffffffff);
+				    i * hub->eng_addr_distance, 0xffffffff);
 		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
-				    2 * i, 0x1f);
+				    i * hub->eng_addr_distance, 0x1f);
 	}
 }
 
-struct pctl_data {
-	uint32_t index;
-	uint32_t data;
-};
-
-static const struct pctl_data pctl0_data[] = {
-	{0x0, 0x7a640},
-	{0x9, 0x2a64a},
-	{0xd, 0x2a680},
-	{0x11, 0x6a684},
-	{0x19, 0xea68e},
-	{0x29, 0xa69e},
-	{0x2b, 0x34a6c0},
-	{0x61, 0x83a707},
-	{0xe6, 0x8a7a4},
-	{0xf0, 0x1a7b8},
-	{0xf3, 0xfa7cc},
-	{0x104, 0x17a7dd},
-	{0x11d, 0xa7dc},
-	{0x11f, 0x12a7f5},
-	{0x133, 0xa808},
-	{0x135, 0x12a810},
-	{0x149, 0x7a82c}
-};
-#define PCTL0_DATA_LEN (sizeof(pctl0_data)/sizeof(pctl0_data[0]))
-
-#define PCTL0_RENG_EXEC_END_PTR 0x151
-#define PCTL0_STCTRL_REG_SAVE_RANGE0_BASE  0xa640
-#define PCTL0_STCTRL_REG_SAVE_RANGE0_LIMIT 0xa833
-
-static const struct pctl_data pctl1_data[] = {
-	{0x0, 0x39a000},
-	{0x3b, 0x44a040},
-	{0x81, 0x2a08d},
-	{0x85, 0x6ba094},
-	{0xf2, 0x18a100},
-	{0x10c, 0x4a132},
-	{0x112, 0xca141},
-	{0x120, 0x2fa158},
-	{0x151, 0x17a1d0},
-	{0x16a, 0x1a1e9},
-	{0x16d, 0x13a1ec},
-	{0x182, 0x7a201},
-	{0x18b, 0x3a20a},
-	{0x190, 0x7a580},
-	{0x199, 0xa590},
-	{0x19b, 0x4a594},
-	{0x1a1, 0x1a59c},
-	{0x1a4, 0x7a82c},
-	{0x1ad, 0xfa7cc},
-	{0x1be, 0x17a7dd},
-	{0x1d7, 0x12a810},
-	{0x1eb, 0x4000a7e1},
-	{0x1ec, 0x5000a7f5},
-	{0x1ed, 0x4000a7e2},
-	{0x1ee, 0x5000a7dc},
-	{0x1ef, 0x4000a7e3},
-	{0x1f0, 0x5000a7f6},
-	{0x1f1, 0x5000a7e4}
-};
-#define PCTL1_DATA_LEN (sizeof(pctl1_data)/sizeof(pctl1_data[0]))
-
-#define PCTL1_RENG_EXEC_END_PTR 0x1f1
-#define PCTL1_STCTRL_REG_SAVE_RANGE0_BASE  0xa000
-#define PCTL1_STCTRL_REG_SAVE_RANGE0_LIMIT 0xa20d
-#define PCTL1_STCTRL_REG_SAVE_RANGE1_BASE  0xa580
-#define PCTL1_STCTRL_REG_SAVE_RANGE1_LIMIT 0xa59d
-#define PCTL1_STCTRL_REG_SAVE_RANGE2_BASE  0xa82c
-#define PCTL1_STCTRL_REG_SAVE_RANGE2_LIMIT 0xa833
-
-static void mmhub_v1_0_power_gating_write_save_ranges(struct amdgpu_device *adev)
-{
-	uint32_t tmp = 0;
-
-	/* PCTL0_STCTRL_REGISTER_SAVE_RANGE0 */
-	tmp = REG_SET_FIELD(tmp, PCTL0_STCTRL_REGISTER_SAVE_RANGE0,
-			STCTRL_REGISTER_SAVE_BASE,
-			PCTL0_STCTRL_REG_SAVE_RANGE0_BASE);
-	tmp = REG_SET_FIELD(tmp, PCTL0_STCTRL_REGISTER_SAVE_RANGE0,
-			STCTRL_REGISTER_SAVE_LIMIT,
-			PCTL0_STCTRL_REG_SAVE_RANGE0_LIMIT);
-	WREG32_SOC15(MMHUB, 0, mmPCTL0_STCTRL_REGISTER_SAVE_RANGE0, tmp);
-
-	/* PCTL1_STCTRL_REGISTER_SAVE_RANGE0 */
-	tmp = 0;
-	tmp = REG_SET_FIELD(tmp, PCTL1_STCTRL_REGISTER_SAVE_RANGE0,
-			STCTRL_REGISTER_SAVE_BASE,
-			PCTL1_STCTRL_REG_SAVE_RANGE0_BASE);
-	tmp = REG_SET_FIELD(tmp, PCTL1_STCTRL_REGISTER_SAVE_RANGE0,
-			STCTRL_REGISTER_SAVE_LIMIT,
-			PCTL1_STCTRL_REG_SAVE_RANGE0_LIMIT);
-	WREG32_SOC15(MMHUB, 0, mmPCTL1_STCTRL_REGISTER_SAVE_RANGE0, tmp);
-
-	/* PCTL1_STCTRL_REGISTER_SAVE_RANGE1 */
-	tmp = 0;
-	tmp = REG_SET_FIELD(tmp, PCTL1_STCTRL_REGISTER_SAVE_RANGE1,
-			STCTRL_REGISTER_SAVE_BASE,
-			PCTL1_STCTRL_REG_SAVE_RANGE1_BASE);
-	tmp = REG_SET_FIELD(tmp, PCTL1_STCTRL_REGISTER_SAVE_RANGE1,
-			STCTRL_REGISTER_SAVE_LIMIT,
-			PCTL1_STCTRL_REG_SAVE_RANGE1_LIMIT);
-	WREG32_SOC15(MMHUB, 0, mmPCTL1_STCTRL_REGISTER_SAVE_RANGE1, tmp);
-
-	/* PCTL1_STCTRL_REGISTER_SAVE_RANGE2 */
-	tmp = 0;
-	tmp = REG_SET_FIELD(tmp, PCTL1_STCTRL_REGISTER_SAVE_RANGE2,
-			STCTRL_REGISTER_SAVE_BASE,
-			PCTL1_STCTRL_REG_SAVE_RANGE2_BASE);
-	tmp = REG_SET_FIELD(tmp, PCTL1_STCTRL_REGISTER_SAVE_RANGE2,
-			STCTRL_REGISTER_SAVE_LIMIT,
-			PCTL1_STCTRL_REG_SAVE_RANGE2_LIMIT);
-	WREG32_SOC15(MMHUB, 0, mmPCTL1_STCTRL_REGISTER_SAVE_RANGE2, tmp);
-}
-
-void mmhub_v1_0_initialize_power_gating(struct amdgpu_device *adev)
-{
-	uint32_t pctl0_misc = 0;
-	uint32_t pctl0_reng_execute = 0;
-	uint32_t pctl1_misc = 0;
-	uint32_t pctl1_reng_execute = 0;
-	int i = 0;
-
-	if (amdgpu_sriov_vf(adev))
-		return;
-
-	pctl0_misc = RREG32_SOC15(MMHUB, 0, mmPCTL0_MISC);
-	pctl0_reng_execute = RREG32_SOC15(MMHUB, 0, mmPCTL0_RENG_EXECUTE);
-	pctl1_misc = RREG32_SOC15(MMHUB, 0, mmPCTL1_MISC);
-	pctl1_reng_execute = RREG32_SOC15(MMHUB, 0, mmPCTL1_RENG_EXECUTE);
-
-	/* Light sleep must be disabled before writing to pctl0 registers */
-	pctl0_misc &= ~PCTL0_MISC__RENG_MEM_LS_ENABLE_MASK;
-	WREG32_SOC15(MMHUB, 0, mmPCTL0_MISC, pctl0_misc);
-
-	/* Write data used to access ram of register engine */
-	for (i = 0; i < PCTL0_DATA_LEN; i++) {
-                WREG32_SOC15(MMHUB, 0, mmPCTL0_RENG_RAM_INDEX,
-			pctl0_data[i].index);
-                WREG32_SOC15(MMHUB, 0, mmPCTL0_RENG_RAM_DATA,
-			pctl0_data[i].data);
-        }
-
-	/* Set the reng execute end ptr for pctl0 */
-	pctl0_reng_execute = REG_SET_FIELD(pctl0_reng_execute,
-					PCTL0_RENG_EXECUTE,
-					RENG_EXECUTE_END_PTR,
-					PCTL0_RENG_EXEC_END_PTR);
-	WREG32_SOC15(MMHUB, 0, mmPCTL0_RENG_EXECUTE, pctl0_reng_execute);
-
-	/* Light sleep must be disabled before writing to pctl1 registers */
-	pctl1_misc &= ~PCTL1_MISC__RENG_MEM_LS_ENABLE_MASK;
-	WREG32_SOC15(MMHUB, 0, mmPCTL1_MISC, pctl1_misc);
-
-	/* Write data used to access ram of register engine */
-	for (i = 0; i < PCTL1_DATA_LEN; i++) {
-                WREG32_SOC15(MMHUB, 0, mmPCTL1_RENG_RAM_INDEX,
-			pctl1_data[i].index);
-                WREG32_SOC15(MMHUB, 0, mmPCTL1_RENG_RAM_DATA,
-			pctl1_data[i].data);
-        }
-
-	/* Set the reng execute end ptr for pctl1 */
-	pctl1_reng_execute = REG_SET_FIELD(pctl1_reng_execute,
-					PCTL1_RENG_EXECUTE,
-					RENG_EXECUTE_END_PTR,
-					PCTL1_RENG_EXEC_END_PTR);
-	WREG32_SOC15(MMHUB, 0, mmPCTL1_RENG_EXECUTE, pctl1_reng_execute);
-
-	mmhub_v1_0_power_gating_write_save_ranges(adev);
-
-	/* Re-enable light sleep */
-	pctl0_misc |= PCTL0_MISC__RENG_MEM_LS_ENABLE_MASK;
-	WREG32_SOC15(MMHUB, 0, mmPCTL0_MISC, pctl0_misc);
-	pctl1_misc |= PCTL1_MISC__RENG_MEM_LS_ENABLE_MASK;
-	WREG32_SOC15(MMHUB, 0, mmPCTL1_MISC, pctl1_misc);
-}
-
-void mmhub_v1_0_update_power_gating(struct amdgpu_device *adev,
+static void mmhub_v1_0_update_power_gating(struct amdgpu_device *adev,
 				bool enable)
 {
-	uint32_t pctl0_reng_execute = 0;
-	uint32_t pctl1_reng_execute = 0;
-
 	if (amdgpu_sriov_vf(adev))
 		return;
 
-	pctl0_reng_execute = RREG32_SOC15(MMHUB, 0, mmPCTL0_RENG_EXECUTE);
-	pctl1_reng_execute = RREG32_SOC15(MMHUB, 0, mmPCTL1_RENG_EXECUTE);
-
-	if (enable && adev->pg_flags & AMD_PG_SUPPORT_MMHUB) {
-		pctl0_reng_execute = REG_SET_FIELD(pctl0_reng_execute,
-						PCTL0_RENG_EXECUTE,
-						RENG_EXECUTE_ON_PWR_UP, 1);
-		pctl0_reng_execute = REG_SET_FIELD(pctl0_reng_execute,
-						PCTL0_RENG_EXECUTE,
-						RENG_EXECUTE_ON_REG_UPDATE, 1);
-		WREG32_SOC15(MMHUB, 0, mmPCTL0_RENG_EXECUTE, pctl0_reng_execute);
-
-		pctl1_reng_execute = REG_SET_FIELD(pctl1_reng_execute,
-						PCTL1_RENG_EXECUTE,
-						RENG_EXECUTE_ON_PWR_UP, 1);
-		pctl1_reng_execute = REG_SET_FIELD(pctl1_reng_execute,
-						PCTL1_RENG_EXECUTE,
-						RENG_EXECUTE_ON_REG_UPDATE, 1);
-		WREG32_SOC15(MMHUB, 0, mmPCTL1_RENG_EXECUTE, pctl1_reng_execute);
-
-	} else {
-		pctl0_reng_execute = REG_SET_FIELD(pctl0_reng_execute,
-						PCTL0_RENG_EXECUTE,
-						RENG_EXECUTE_ON_PWR_UP, 0);
-		pctl0_reng_execute = REG_SET_FIELD(pctl0_reng_execute,
-						PCTL0_RENG_EXECUTE,
-						RENG_EXECUTE_ON_REG_UPDATE, 0);
-		WREG32_SOC15(MMHUB, 0, mmPCTL0_RENG_EXECUTE, pctl0_reng_execute);
-
-		pctl1_reng_execute = REG_SET_FIELD(pctl1_reng_execute,
-						PCTL1_RENG_EXECUTE,
-						RENG_EXECUTE_ON_PWR_UP, 0);
-		pctl1_reng_execute = REG_SET_FIELD(pctl1_reng_execute,
-						PCTL1_RENG_EXECUTE,
-						RENG_EXECUTE_ON_REG_UPDATE, 0);
-		WREG32_SOC15(MMHUB, 0, mmPCTL1_RENG_EXECUTE, pctl1_reng_execute);
-	}
+	if (adev->pg_flags & AMD_PG_SUPPORT_MMHUB)
+		amdgpu_dpm_set_powergating_by_smu(adev,
+						  AMD_IP_BLOCK_TYPE_GMC,
+						  enable, 0);
 }
 
-int mmhub_v1_0_gart_enable(struct amdgpu_device *adev)
+static int mmhub_v1_0_gart_enable(struct amdgpu_device *adev)
 {
 	if (amdgpu_sriov_vf(adev)) {
 		/*
@@ -483,9 +368,9 @@ int mmhub_v1_0_gart_enable(struct amdgpu_device *adev)
 		 * SRIOV driver need to program them
 		 */
 		WREG32_SOC15(MMHUB, 0, mmMC_VM_FB_LOCATION_BASE,
-			     adev->mc.vram_start >> 24);
+			     adev->gmc.vram_start >> 24);
 		WREG32_SOC15(MMHUB, 0, mmMC_VM_FB_LOCATION_TOP,
-			     adev->mc.vram_end >> 24);
+			     adev->gmc.vram_end >> 24);
 	}
 
 	/* GART Enable. */
@@ -502,14 +387,16 @@ int mmhub_v1_0_gart_enable(struct amdgpu_device *adev)
 	return 0;
 }
 
-void mmhub_v1_0_gart_disable(struct amdgpu_device *adev)
+static void mmhub_v1_0_gart_disable(struct amdgpu_device *adev)
 {
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
 	u32 tmp;
 	u32 i;
 
 	/* Disable all tables */
-	for (i = 0; i < 16; i++)
-		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT0_CNTL, i, 0);
+	for (i = 0; i < AMDGPU_NUM_VMID; i++)
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT0_CNTL,
+				    i * hub->ctx_distance, 0);
 
 	/* Setup TLB control */
 	tmp = RREG32_SOC15(MMHUB, 0, mmMC_VM_MX_L1_TLB_CNTL);
@@ -520,11 +407,13 @@ void mmhub_v1_0_gart_disable(struct amdgpu_device *adev)
 				0);
 	WREG32_SOC15(MMHUB, 0, mmMC_VM_MX_L1_TLB_CNTL, tmp);
 
-	/* Setup L2 cache */
-	tmp = RREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL);
-	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 0);
-	WREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL, tmp);
-	WREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL3, 0);
+	if (!amdgpu_sriov_vf(adev)) {
+		/* Setup L2 cache */
+		tmp = RREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 0);
+		WREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL, tmp);
+		WREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL3, 0);
+	}
 }
 
 /**
@@ -533,9 +422,13 @@ void mmhub_v1_0_gart_disable(struct amdgpu_device *adev)
  * @adev: amdgpu_device pointer
  * @value: true redirects VM faults to the default page
  */
-void mmhub_v1_0_set_fault_enable_default(struct amdgpu_device *adev, bool value)
+static void mmhub_v1_0_set_fault_enable_default(struct amdgpu_device *adev, bool value)
 {
 	u32 tmp;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
 	tmp = RREG32_SOC15(MMHUB, 0, mmVM_L2_PROTECTION_FAULT_CNTL);
 	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
 			RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
@@ -561,12 +454,19 @@ void mmhub_v1_0_set_fault_enable_default(struct amdgpu_device *adev, bool value)
 			WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
 	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
 			EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	if (!value) {
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_NO_RETRY_FAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_RETRY_FAULT, 1);
+	}
+
 	WREG32_SOC15(MMHUB, 0, mmVM_L2_PROTECTION_FAULT_CNTL, tmp);
 }
 
-void mmhub_v1_0_init(struct amdgpu_device *adev)
+static void mmhub_v1_0_init(struct amdgpu_device *adev)
 {
-	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB];
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
 
 	hub->ctx0_ptb_addr_lo32 =
 		SOC15_REG_OFFSET(MMHUB, 0,
@@ -574,6 +474,8 @@ void mmhub_v1_0_init(struct amdgpu_device *adev)
 	hub->ctx0_ptb_addr_hi32 =
 		SOC15_REG_OFFSET(MMHUB, 0,
 				 mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32);
+	hub->vm_inv_eng0_sem =
+		SOC15_REG_OFFSET(MMHUB, 0, mmVM_INVALIDATE_ENG0_SEM);
 	hub->vm_inv_eng0_req =
 		SOC15_REG_OFFSET(MMHUB, 0, mmVM_INVALIDATE_ENG0_REQ);
 	hub->vm_inv_eng0_ack =
@@ -585,6 +487,12 @@ void mmhub_v1_0_init(struct amdgpu_device *adev)
 	hub->vm_l2_pro_fault_cntl =
 		SOC15_REG_OFFSET(MMHUB, 0, mmVM_L2_PROTECTION_FAULT_CNTL);
 
+	hub->ctx_distance = mmVM_CONTEXT1_CNTL - mmVM_CONTEXT0_CNTL;
+	hub->ctx_addr_distance = mmVM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 -
+		mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32;
+	hub->eng_distance = mmVM_INVALIDATE_ENG1_REQ - mmVM_INVALIDATE_ENG0_REQ;
+	hub->eng_addr_distance = mmVM_INVALIDATE_ENG1_ADDR_RANGE_LO32 -
+		mmVM_INVALIDATE_ENG0_ADDR_RANGE_LO32;
 }
 
 static void mmhub_v1_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
@@ -650,22 +558,6 @@ static void mmhub_v1_0_update_medium_grain_clock_gating(struct amdgpu_device *ad
 		WREG32_SOC15(MMHUB, 0, mmDAGB1_CNTL_MISC2, data2);
 }
 
-static void athub_update_medium_grain_clock_gating(struct amdgpu_device *adev,
-						   bool enable)
-{
-	uint32_t def, data;
-
-	def = data = RREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL);
-
-	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG))
-		data |= ATHUB_MISC_CNTL__CG_ENABLE_MASK;
-	else
-		data &= ~ATHUB_MISC_CNTL__CG_ENABLE_MASK;
-
-	if (def != data)
-		WREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL, data);
-}
-
 static void mmhub_v1_0_update_medium_grain_light_sleep(struct amdgpu_device *adev,
 						       bool enable)
 {
@@ -682,24 +574,7 @@ static void mmhub_v1_0_update_medium_grain_light_sleep(struct amdgpu_device *ade
 		WREG32_SOC15(MMHUB, 0, mmATC_L2_MISC_CG, data);
 }
 
-static void athub_update_medium_grain_light_sleep(struct amdgpu_device *adev,
-						  bool enable)
-{
-	uint32_t def, data;
-
-	def = data = RREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL);
-
-	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_LS) &&
-	    (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS))
-		data |= ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK;
-	else
-		data &= ~ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK;
-
-	if(def != data)
-		WREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL, data);
-}
-
-int mmhub_v1_0_set_clockgating(struct amdgpu_device *adev,
+static int mmhub_v1_0_set_clockgating(struct amdgpu_device *adev,
 			       enum amd_clockgating_state state)
 {
 	if (amdgpu_sriov_vf(adev))
@@ -707,15 +582,14 @@ int mmhub_v1_0_set_clockgating(struct amdgpu_device *adev,
 
 	switch (adev->asic_type) {
 	case CHIP_VEGA10:
+	case CHIP_VEGA12:
+	case CHIP_VEGA20:
 	case CHIP_RAVEN:
+	case CHIP_RENOIR:
 		mmhub_v1_0_update_medium_grain_clock_gating(adev,
-				state == AMD_CG_STATE_GATE ? true : false);
-		athub_update_medium_grain_clock_gating(adev,
-				state == AMD_CG_STATE_GATE ? true : false);
+				state == AMD_CG_STATE_GATE);
 		mmhub_v1_0_update_medium_grain_light_sleep(adev,
-				state == AMD_CG_STATE_GATE ? true : false);
-		athub_update_medium_grain_light_sleep(adev,
-				state == AMD_CG_STATE_GATE ? true : false);
+				state == AMD_CG_STATE_GATE);
 		break;
 	default:
 		break;
@@ -724,20 +598,253 @@ int mmhub_v1_0_set_clockgating(struct amdgpu_device *adev,
 	return 0;
 }
 
-void mmhub_v1_0_get_clockgating(struct amdgpu_device *adev, u32 *flags)
+static void mmhub_v1_0_get_clockgating(struct amdgpu_device *adev, u64 *flags)
 {
-	int data;
+	int data, data1;
 
 	if (amdgpu_sriov_vf(adev))
 		*flags = 0;
 
+	data = RREG32_SOC15(MMHUB, 0, mmATC_L2_MISC_CG);
+
+	data1 = RREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2);
+
 	/* AMD_CG_SUPPORT_MC_MGCG */
-	data = RREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL);
-	if (data & ATHUB_MISC_CNTL__CG_ENABLE_MASK)
+	if ((data & ATC_L2_MISC_CG__ENABLE_MASK) &&
+	    !(data1 & (DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+		       DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+		       DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+		       DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+		       DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+		       DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK)))
 		*flags |= AMD_CG_SUPPORT_MC_MGCG;
 
 	/* AMD_CG_SUPPORT_MC_LS */
-	data = RREG32_SOC15(MMHUB, 0, mmATC_L2_MISC_CG);
 	if (data & ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK)
 		*flags |= AMD_CG_SUPPORT_MC_LS;
 }
+
+static const struct soc15_ras_field_entry mmhub_v1_0_ras_fields[] = {
+	{ "MMEA0_DRAMRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT_VG20),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT_VG20, DRAMRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT_VG20, DRAMRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA0_DRAMWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT_VG20),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT_VG20, DRAMWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT_VG20, DRAMWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA0_DRAMWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT_VG20),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT_VG20, DRAMWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT_VG20, DRAMWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA0_RRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT_VG20),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT_VG20, RRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT_VG20, RRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA0_WRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT_VG20),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT_VG20, WRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT_VG20, WRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA0_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT_VG20),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT_VG20, DRAMRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA0_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT_VG20),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT_VG20, DRAMWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA0_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT_VG20),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT_VG20, IORD_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA0_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT_VG20),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT_VG20, IOWR_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA0_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT_VG20),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT_VG20, IOWR_DATAMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA0_GMIRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT2_VG20),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2_VG20, GMIRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2_VG20, GMIRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA0_GMIWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT2_VG20),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2_VG20, GMIWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2_VG20, GMIWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA0_GMIWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT2_VG20),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2_VG20, GMIWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2_VG20, GMIWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA0_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT2_VG20),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2_VG20, GMIRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA0_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT2_VG20),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2_VG20, GMIWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA1_DRAMRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT_VG20),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT_VG20, DRAMRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT_VG20, DRAMRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA1_DRAMWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT_VG20),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT_VG20, DRAMWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT_VG20, DRAMWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA1_DRAMWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT_VG20),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT_VG20, DRAMWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT_VG20, DRAMWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA1_RRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT_VG20),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT_VG20, RRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT_VG20, RRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA1_WRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT_VG20),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT_VG20, WRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT_VG20, WRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA1_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT_VG20),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT_VG20, DRAMRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA1_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT_VG20),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT_VG20, DRAMWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA1_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT_VG20),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT_VG20, IORD_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA1_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT_VG20),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT_VG20, IOWR_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA1_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT_VG20),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT_VG20, IOWR_DATAMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA1_GMIRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT2_VG20),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2_VG20, GMIRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2_VG20, GMIRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA1_GMIWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT2_VG20),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2_VG20, GMIWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2_VG20, GMIWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA1_GMIWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT2_VG20),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2_VG20, GMIWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2_VG20, GMIWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA1_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT2_VG20),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2_VG20, GMIRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA1_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT2_VG20),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2_VG20, GMIWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	}
+};
+
+static const struct soc15_reg_entry mmhub_v1_0_edc_cnt_regs[] = {
+   { SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT_VG20), 0, 0, 0},
+   { SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT2_VG20), 0, 0, 0},
+   { SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT_VG20), 0, 0, 0},
+   { SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT2_VG20), 0, 0, 0},
+};
+
+static int mmhub_v1_0_get_ras_error_count(struct amdgpu_device *adev,
+	const struct soc15_reg_entry *reg,
+	uint32_t value, uint32_t *sec_count, uint32_t *ded_count)
+{
+	uint32_t i;
+	uint32_t sec_cnt, ded_cnt;
+
+	for (i = 0; i < ARRAY_SIZE(mmhub_v1_0_ras_fields); i++) {
+		if (mmhub_v1_0_ras_fields[i].reg_offset != reg->reg_offset)
+			continue;
+
+		sec_cnt = (value &
+				mmhub_v1_0_ras_fields[i].sec_count_mask) >>
+				mmhub_v1_0_ras_fields[i].sec_count_shift;
+		if (sec_cnt) {
+			dev_info(adev->dev,
+				"MMHUB SubBlock %s, SEC %d\n",
+				mmhub_v1_0_ras_fields[i].name,
+				sec_cnt);
+			*sec_count += sec_cnt;
+		}
+
+		ded_cnt = (value &
+				mmhub_v1_0_ras_fields[i].ded_count_mask) >>
+				mmhub_v1_0_ras_fields[i].ded_count_shift;
+		if (ded_cnt) {
+			dev_info(adev->dev,
+				"MMHUB SubBlock %s, DED %d\n",
+				mmhub_v1_0_ras_fields[i].name,
+				ded_cnt);
+			*ded_count += ded_cnt;
+		}
+	}
+
+	return 0;
+}
+
+static void mmhub_v1_0_query_ras_error_count(struct amdgpu_device *adev,
+					   void *ras_error_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+	uint32_t sec_count = 0, ded_count = 0;
+	uint32_t i;
+	uint32_t reg_value;
+
+	err_data->ue_count = 0;
+	err_data->ce_count = 0;
+
+	for (i = 0; i < ARRAY_SIZE(mmhub_v1_0_edc_cnt_regs); i++) {
+		reg_value =
+			RREG32(SOC15_REG_ENTRY_OFFSET(mmhub_v1_0_edc_cnt_regs[i]));
+		if (reg_value)
+			mmhub_v1_0_get_ras_error_count(adev,
+				&mmhub_v1_0_edc_cnt_regs[i],
+				reg_value, &sec_count, &ded_count);
+	}
+
+	err_data->ce_count += sec_count;
+	err_data->ue_count += ded_count;
+}
+
+static void mmhub_v1_0_reset_ras_error_count(struct amdgpu_device *adev)
+{
+	uint32_t i;
+
+	/* read back edc counter registers to reset the counters to 0 */
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__MMHUB)) {
+		for (i = 0; i < ARRAY_SIZE(mmhub_v1_0_edc_cnt_regs); i++)
+			RREG32(SOC15_REG_ENTRY_OFFSET(mmhub_v1_0_edc_cnt_regs[i]));
+	}
+}
+
+struct amdgpu_ras_block_hw_ops mmhub_v1_0_ras_hw_ops = {
+	.query_ras_error_count = mmhub_v1_0_query_ras_error_count,
+	.reset_ras_error_count = mmhub_v1_0_reset_ras_error_count,
+};
+
+struct amdgpu_mmhub_ras mmhub_v1_0_ras = {
+	.ras_block = {
+		.hw_ops = &mmhub_v1_0_ras_hw_ops,
+	},
+};
+
+const struct amdgpu_mmhub_funcs mmhub_v1_0_funcs = {
+	.get_fb_location = mmhub_v1_0_get_fb_location,
+	.init = mmhub_v1_0_init,
+	.gart_enable = mmhub_v1_0_gart_enable,
+	.set_fault_enable_default = mmhub_v1_0_set_fault_enable_default,
+	.gart_disable = mmhub_v1_0_gart_disable,
+	.set_clockgating = mmhub_v1_0_set_clockgating,
+	.get_clockgating = mmhub_v1_0_get_clockgating,
+	.setup_vm_pt_regs = mmhub_v1_0_setup_vm_pt_regs,
+	.update_power_gating = mmhub_v1_0_update_power_gating,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h
index 5d38229baf69..dae7ca48bd8b 100644
--- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h
@@ -23,17 +23,7 @@
 #ifndef __MMHUB_V1_0_H__
 #define __MMHUB_V1_0_H__
 
-u64 mmhub_v1_0_get_fb_location(struct amdgpu_device *adev);
-int mmhub_v1_0_gart_enable(struct amdgpu_device *adev);
-void mmhub_v1_0_gart_disable(struct amdgpu_device *adev);
-void mmhub_v1_0_set_fault_enable_default(struct amdgpu_device *adev,
-					 bool value);
-void mmhub_v1_0_init(struct amdgpu_device *adev);
-int mmhub_v1_0_set_clockgating(struct amdgpu_device *adev,
-			       enum amd_clockgating_state state);
-void mmhub_v1_0_get_clockgating(struct amdgpu_device *adev, u32 *flags);
-void mmhub_v1_0_initialize_power_gating(struct amdgpu_device *adev);
-void mmhub_v1_0_update_power_gating(struct amdgpu_device *adev,
-                                bool enable);
+extern const struct amdgpu_mmhub_funcs mmhub_v1_0_funcs;
+extern struct amdgpu_mmhub_ras mmhub_v1_0_ras;
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.c
new file mode 100644
index 000000000000..2adee2b94c37
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.c
@@ -0,0 +1,1371 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "amdgpu_ras.h"
+#include "mmhub_v1_7.h"
+
+#include "mmhub/mmhub_1_7_offset.h"
+#include "mmhub/mmhub_1_7_sh_mask.h"
+#include "vega10_enum.h"
+
+#include "soc15_common.h"
+#include "soc15.h"
+
+#define regVM_L2_CNTL3_DEFAULT	0x80100007
+#define regVM_L2_CNTL4_DEFAULT	0x000000c1
+
+static u64 mmhub_v1_7_get_fb_location(struct amdgpu_device *adev)
+{
+	u64 base = RREG32_SOC15(MMHUB, 0, regMC_VM_FB_LOCATION_BASE);
+	u64 top = RREG32_SOC15(MMHUB, 0, regMC_VM_FB_LOCATION_TOP);
+
+	base &= MC_VM_FB_LOCATION_BASE__FB_BASE_MASK;
+	base <<= 24;
+
+	top &= MC_VM_FB_LOCATION_TOP__FB_TOP_MASK;
+	top <<= 24;
+
+	adev->gmc.fb_start = base;
+	adev->gmc.fb_end = top;
+
+	return base;
+}
+
+static void mmhub_v1_7_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid,
+				uint64_t page_table_base)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+
+	WREG32_SOC15_OFFSET(MMHUB, 0, regVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+			hub->ctx_addr_distance * vmid, lower_32_bits(page_table_base));
+
+	WREG32_SOC15_OFFSET(MMHUB, 0, regVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+			hub->ctx_addr_distance * vmid, upper_32_bits(page_table_base));
+}
+
+static void mmhub_v1_7_init_gart_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t pt_base;
+
+	if (adev->gmc.pdb0_bo)
+		pt_base = amdgpu_gmc_pd_addr(adev->gmc.pdb0_bo);
+	else
+		pt_base = amdgpu_gmc_pd_addr(adev->gart.bo);
+
+	mmhub_v1_7_setup_vm_pt_regs(adev, 0, pt_base);
+
+	/* If use GART for FB translation, vmid0 page table covers both
+	 * vram and system memory (gart)
+	 */
+	if (adev->gmc.pdb0_bo) {
+		WREG32_SOC15(MMHUB, 0, regVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+				(u32)(adev->gmc.fb_start >> 12));
+		WREG32_SOC15(MMHUB, 0, regVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+				(u32)(adev->gmc.fb_start >> 44));
+
+		WREG32_SOC15(MMHUB, 0, regVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+				(u32)(adev->gmc.gart_end >> 12));
+		WREG32_SOC15(MMHUB, 0, regVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+				(u32)(adev->gmc.gart_end >> 44));
+
+	} else {
+		WREG32_SOC15(MMHUB, 0, regVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+				(u32)(adev->gmc.gart_start >> 12));
+		WREG32_SOC15(MMHUB, 0, regVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+				(u32)(adev->gmc.gart_start >> 44));
+
+		WREG32_SOC15(MMHUB, 0, regVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+				(u32)(adev->gmc.gart_end >> 12));
+		WREG32_SOC15(MMHUB, 0, regVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+				(u32)(adev->gmc.gart_end >> 44));
+	}
+}
+
+static void mmhub_v1_7_init_system_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t value;
+	uint32_t tmp;
+
+	/* Program the AGP BAR */
+	WREG32_SOC15(MMHUB, 0, regMC_VM_AGP_BASE, 0);
+	WREG32_SOC15(MMHUB, 0, regMC_VM_AGP_BOT, adev->gmc.agp_start >> 24);
+	WREG32_SOC15(MMHUB, 0, regMC_VM_AGP_TOP, adev->gmc.agp_end >> 24);
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/* Program the system aperture low logical page number. */
+	WREG32_SOC15(MMHUB, 0, regMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+		     min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
+
+	WREG32_SOC15(MMHUB, 0, regMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+		     max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
+
+	/* In the case squeezing vram into GART aperture, we don't use
+	 * FB aperture and AGP aperture. Disable them.
+	 */
+	if (adev->gmc.pdb0_bo) {
+		WREG32_SOC15(MMHUB, 0, regMC_VM_AGP_BOT, 0xFFFFFF);
+		WREG32_SOC15(MMHUB, 0, regMC_VM_AGP_TOP, 0);
+		WREG32_SOC15(MMHUB, 0, regMC_VM_FB_LOCATION_TOP, 0);
+		WREG32_SOC15(MMHUB, 0, regMC_VM_FB_LOCATION_BASE, 0x00FFFFFF);
+		WREG32_SOC15(MMHUB, 0, regMC_VM_SYSTEM_APERTURE_LOW_ADDR, 0x3FFFFFFF);
+		WREG32_SOC15(MMHUB, 0, regMC_VM_SYSTEM_APERTURE_HIGH_ADDR, 0);
+	}
+
+	/* Set default page address. */
+	value = amdgpu_gmc_vram_mc2pa(adev, adev->mem_scratch.gpu_addr);
+	WREG32_SOC15(MMHUB, 0, regMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
+		     (u32)(value >> 12));
+	WREG32_SOC15(MMHUB, 0, regMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
+		     (u32)(value >> 44));
+
+	/* Program "protection fault". */
+	WREG32_SOC15(MMHUB, 0, regVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
+		     (u32)(adev->dummy_page_addr >> 12));
+	WREG32_SOC15(MMHUB, 0, regVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+		     (u32)((u64)adev->dummy_page_addr >> 44));
+
+	tmp = RREG32_SOC15(MMHUB, 0, regVM_L2_PROTECTION_FAULT_CNTL2);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL2,
+			    ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+	WREG32_SOC15(MMHUB, 0, regVM_L2_PROTECTION_FAULT_CNTL2, tmp);
+}
+
+static void mmhub_v1_7_init_tlb_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(MMHUB, 0, regMC_VM_MX_L1_TLB_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 1);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+			    SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+			    MTYPE, MTYPE_UC);/* XXX for emulation. */
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ATC_EN, 1);
+
+	WREG32_SOC15(MMHUB, 0, regMC_VM_MX_L1_TLB_CNTL, tmp);
+}
+
+/* Set snoop bit for SDMA so that SDMA writes probe-invalidates RW lines */
+static void mmhub_v1_7_init_snoop_override_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+	int i;
+	uint32_t distance = regDAGB1_WRCLI_GPU_SNOOP_OVERRIDE -
+			    regDAGB0_WRCLI_GPU_SNOOP_OVERRIDE;
+
+	for (i = 0; i < 5; i++) { /* DAGB instances */
+		tmp = RREG32_SOC15_OFFSET(MMHUB, 0,
+			regDAGB0_WRCLI_GPU_SNOOP_OVERRIDE, i * distance);
+		tmp |= (1 << 15); /* SDMA client is BIT15 */
+		WREG32_SOC15_OFFSET(MMHUB, 0,
+			regDAGB0_WRCLI_GPU_SNOOP_OVERRIDE, i * distance, tmp);
+
+		tmp = RREG32_SOC15_OFFSET(MMHUB, 0,
+			regDAGB0_WRCLI_GPU_SNOOP_OVERRIDE_VALUE, i * distance);
+		tmp |= (1 << 15);
+		WREG32_SOC15_OFFSET(MMHUB, 0,
+			regDAGB0_WRCLI_GPU_SNOOP_OVERRIDE_VALUE, i * distance, tmp);
+	}
+
+}
+
+static void mmhub_v1_7_init_cache_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(MMHUB, 0, regVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 1);
+	/* XXX for emulation, Refer to closed source code.*/
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, L2_PDE0_CACHE_TAG_GENERATION_MODE,
+			    0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(MMHUB, 0, regVM_L2_CNTL, tmp);
+
+	tmp = RREG32_SOC15(MMHUB, 0, regVM_L2_CNTL2);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
+	WREG32_SOC15(MMHUB, 0, regVM_L2_CNTL2, tmp);
+
+	tmp = regVM_L2_CNTL3_DEFAULT;
+	if (adev->gmc.translate_further) {
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 12);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+	} else {
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 9);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+	}
+	WREG32_SOC15(MMHUB, 0, regVM_L2_CNTL3, tmp);
+
+	tmp = regVM_L2_CNTL4_DEFAULT;
+	if (adev->gmc.xgmi.connected_to_cpu) {
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4,
+				    VMC_TAP_PDE_REQUEST_PHYSICAL, 1);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4,
+				    VMC_TAP_PTE_REQUEST_PHYSICAL, 1);
+	} else {
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4,
+				    VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4,
+				    VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
+	}
+	WREG32_SOC15(MMHUB, 0, regVM_L2_CNTL4, tmp);
+}
+
+static void mmhub_v1_7_enable_system_domain(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	tmp = RREG32_SOC15(MMHUB, 0, regVM_CONTEXT0_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH,
+			adev->gmc.vmid0_page_table_depth);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, PAGE_TABLE_BLOCK_SIZE,
+			adev->gmc.vmid0_page_table_block_size);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL,
+			    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
+	WREG32_SOC15(MMHUB, 0, regVM_CONTEXT0_CNTL, tmp);
+}
+
+static void mmhub_v1_7_disable_identity_aperture(struct amdgpu_device *adev)
+{
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	WREG32_SOC15(MMHUB, 0, regVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32,
+		     0XFFFFFFFF);
+	WREG32_SOC15(MMHUB, 0, regVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32,
+		     0x0000000F);
+
+	WREG32_SOC15(MMHUB, 0,
+		     regVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32, 0);
+	WREG32_SOC15(MMHUB, 0,
+		     regVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32, 0);
+
+	WREG32_SOC15(MMHUB, 0, regVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32,
+		     0);
+	WREG32_SOC15(MMHUB, 0, regVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32,
+		     0);
+}
+
+static void mmhub_v1_7_setup_vmid_config(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	unsigned num_level, block_size;
+	uint32_t tmp;
+	int i;
+
+	num_level = adev->vm_manager.num_level;
+	block_size = adev->vm_manager.block_size;
+	if (adev->gmc.translate_further)
+		num_level -= 1;
+	else
+		block_size -= 9;
+
+	for (i = 0; i <= 14; i++) {
+		tmp = RREG32_SOC15_OFFSET(MMHUB, 0, regVM_CONTEXT1_CNTL, i * hub->ctx_distance);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH,
+				    num_level);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    PAGE_TABLE_BLOCK_SIZE,
+				    block_size);
+		/* On Aldebaran, XNACK can be enabled in the SQ per-process.
+		 * Retry faults need to be enabled for that to work.
+		 */
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
+				    1);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regVM_CONTEXT1_CNTL,
+				    i * hub->ctx_distance, tmp);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32,
+				    i * hub->ctx_addr_distance,
+				    lower_32_bits(adev->vm_manager.max_pfn - 1));
+		WREG32_SOC15_OFFSET(MMHUB, 0, regVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32,
+				    i * hub->ctx_addr_distance,
+				    upper_32_bits(adev->vm_manager.max_pfn - 1));
+	}
+}
+
+static void mmhub_v1_7_program_invalidation(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	unsigned i;
+
+	for (i = 0; i < 18; ++i) {
+		WREG32_SOC15_OFFSET(MMHUB, 0, regVM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
+				    i * hub->eng_addr_distance, 0xffffffff);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regVM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
+				    i * hub->eng_addr_distance, 0x1f);
+	}
+}
+
+static int mmhub_v1_7_gart_enable(struct amdgpu_device *adev)
+{
+	/* GART Enable. */
+	mmhub_v1_7_init_gart_aperture_regs(adev);
+	mmhub_v1_7_init_system_aperture_regs(adev);
+	mmhub_v1_7_init_tlb_regs(adev);
+	mmhub_v1_7_init_cache_regs(adev);
+	mmhub_v1_7_init_snoop_override_regs(adev);
+
+	mmhub_v1_7_enable_system_domain(adev);
+	mmhub_v1_7_disable_identity_aperture(adev);
+	mmhub_v1_7_setup_vmid_config(adev);
+	mmhub_v1_7_program_invalidation(adev);
+
+	return 0;
+}
+
+static void mmhub_v1_7_gart_disable(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	u32 tmp;
+	u32 i;
+
+	/* Disable all tables */
+	for (i = 0; i < 16; i++)
+		WREG32_SOC15_OFFSET(MMHUB, 0, regVM_CONTEXT0_CNTL,
+				    i * hub->ctx_distance, 0);
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(MMHUB, 0, regMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
+	tmp = REG_SET_FIELD(tmp,
+				MC_VM_MX_L1_TLB_CNTL,
+				ENABLE_ADVANCED_DRIVER_MODEL,
+				0);
+	WREG32_SOC15(MMHUB, 0, regMC_VM_MX_L1_TLB_CNTL, tmp);
+
+	if (!amdgpu_sriov_vf(adev)) {
+		/* Setup L2 cache */
+		tmp = RREG32_SOC15(MMHUB, 0, regVM_L2_CNTL);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 0);
+		WREG32_SOC15(MMHUB, 0, regVM_L2_CNTL, tmp);
+		WREG32_SOC15(MMHUB, 0, regVM_L2_CNTL3, 0);
+	}
+}
+
+/**
+ * mmhub_v1_7_set_fault_enable_default - update GART/VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+static void mmhub_v1_7_set_fault_enable_default(struct amdgpu_device *adev, bool value)
+{
+	u32 tmp;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	tmp = RREG32_SOC15(MMHUB, 0, regVM_L2_PROTECTION_FAULT_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			PDE1_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			PDE2_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp,
+			VM_L2_PROTECTION_FAULT_CNTL,
+			TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT,
+			value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			NACK_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	if (!value) {
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_NO_RETRY_FAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_RETRY_FAULT, 1);
+    }
+
+	WREG32_SOC15(MMHUB, 0, regVM_L2_PROTECTION_FAULT_CNTL, tmp);
+}
+
+static void mmhub_v1_7_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+
+	hub->ctx0_ptb_addr_lo32 =
+		SOC15_REG_OFFSET(MMHUB, 0,
+				 regVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32);
+	hub->ctx0_ptb_addr_hi32 =
+		SOC15_REG_OFFSET(MMHUB, 0,
+				 regVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32);
+	hub->vm_inv_eng0_req =
+		SOC15_REG_OFFSET(MMHUB, 0, regVM_INVALIDATE_ENG0_REQ);
+	hub->vm_inv_eng0_ack =
+		SOC15_REG_OFFSET(MMHUB, 0, regVM_INVALIDATE_ENG0_ACK);
+	hub->vm_context0_cntl =
+		SOC15_REG_OFFSET(MMHUB, 0, regVM_CONTEXT0_CNTL);
+	hub->vm_l2_pro_fault_status =
+		SOC15_REG_OFFSET(MMHUB, 0, regVM_L2_PROTECTION_FAULT_STATUS);
+	hub->vm_l2_pro_fault_cntl =
+		SOC15_REG_OFFSET(MMHUB, 0, regVM_L2_PROTECTION_FAULT_CNTL);
+
+	hub->ctx_distance = regVM_CONTEXT1_CNTL - regVM_CONTEXT0_CNTL;
+	hub->ctx_addr_distance = regVM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 -
+		regVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32;
+	hub->eng_distance = regVM_INVALIDATE_ENG1_REQ - regVM_INVALIDATE_ENG0_REQ;
+	hub->eng_addr_distance = regVM_INVALIDATE_ENG1_ADDR_RANGE_LO32 -
+		regVM_INVALIDATE_ENG0_ADDR_RANGE_LO32;
+
+}
+
+static void mmhub_v1_7_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+							bool enable)
+{
+	uint32_t def, data, def1, data1, def2 = 0, data2 = 0;
+
+	def  = data  = RREG32_SOC15(MMHUB, 0, regATC_L2_MISC_CG);
+
+	def1 = data1 = RREG32_SOC15(MMHUB, 0, regDAGB0_CNTL_MISC2);
+	def2 = data2 = RREG32_SOC15(MMHUB, 0, regDAGB1_CNTL_MISC2);
+
+	if (enable) {
+		data |= ATC_L2_MISC_CG__ENABLE_MASK;
+
+		data1 &= ~(DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+		           DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+		           DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+		           DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+		           DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+		           DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK);
+
+		data2 &= ~(DAGB1_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+		           DAGB1_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+		           DAGB1_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+		           DAGB1_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+		           DAGB1_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+		           DAGB1_CNTL_MISC2__DISABLE_TLBRD_CG_MASK);
+	} else {
+		data &= ~ATC_L2_MISC_CG__ENABLE_MASK;
+
+		data1 |= (DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK);
+
+		data2 |= (DAGB1_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+		          DAGB1_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+		          DAGB1_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+		          DAGB1_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+		          DAGB1_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+		          DAGB1_CNTL_MISC2__DISABLE_TLBRD_CG_MASK);
+	}
+
+	if (def != data)
+		WREG32_SOC15(MMHUB, 0, regATC_L2_MISC_CG, data);
+
+	if (def1 != data1)
+		WREG32_SOC15(MMHUB, 0, regDAGB0_CNTL_MISC2, data1);
+
+	if (def2 != data2)
+		WREG32_SOC15(MMHUB, 0, regDAGB1_CNTL_MISC2, data2);
+}
+
+static void mmhub_v1_7_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+						       bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_SOC15(MMHUB, 0, regATC_L2_MISC_CG);
+
+	if (enable)
+		data |= ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK;
+	else
+		data &= ~ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK;
+
+	if (def != data)
+		WREG32_SOC15(MMHUB, 0, regATC_L2_MISC_CG, data);
+}
+
+static int mmhub_v1_7_set_clockgating(struct amdgpu_device *adev,
+			       enum amd_clockgating_state state)
+{
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	/* Change state only if MCCG support is enabled through driver */
+	if (adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG)
+		mmhub_v1_7_update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+
+	/* Change state only if LS support is enabled through driver */
+	if (adev->cg_flags & AMD_CG_SUPPORT_MC_LS)
+		mmhub_v1_7_update_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+
+	return 0;
+}
+
+static void mmhub_v1_7_get_clockgating(struct amdgpu_device *adev, u64 *flags)
+{
+	u32 data, data1;
+
+	if (amdgpu_sriov_vf(adev))
+		*flags = 0;
+
+	data = RREG32_SOC15(MMHUB, 0, regATC_L2_MISC_CG);
+
+	data1 = RREG32_SOC15(MMHUB, 0, regDAGB0_CNTL_MISC2);
+
+	/* AMD_CG_SUPPORT_MC_MGCG */
+	if ((data & ATC_L2_MISC_CG__ENABLE_MASK) &&
+	    !(data1 & (DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+		       DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+		       DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+		       DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+		       DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+		       DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK)))
+		*flags |= AMD_CG_SUPPORT_MC_MGCG;
+
+	/* AMD_CG_SUPPORT_MC_LS */
+	if (data & ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_MC_LS;
+}
+
+static const struct soc15_ras_field_entry mmhub_v1_7_ras_fields[] = {
+	/* MMHUB Range 0 */
+	{ "MMEA0_DRAMRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA0_DRAMWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA0_DRAMWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA0_RRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, RRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, RRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA0_WRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, WRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, WRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA0_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, IOWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, IOWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA0_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA0_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA0_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, IORD_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA0_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, IOWR_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA0_GMIRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA0_GMIWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA0_GMIWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA0_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA0_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA0_MAM_D0MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D0MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D0MEM_DED_COUNT),
+	},
+	{ "MMEA0_MAM_D1MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D1MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D1MEM_DED_COUNT),
+	},
+	{ "MMEA0_MAM_D2MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D2MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D2MEM_DED_COUNT),
+	},
+	{ "MMEA0_MAM_D3MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D3MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D3MEM_DED_COUNT),
+	},
+	{ "MMEA0_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA0_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA0_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA0_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA0_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA0_EDC_CNT3, IORD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA0_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA0_EDC_CNT3, IOWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA0_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA0_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA0_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA0_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT),
+	},
+
+	/* MMHUB Range 1 */
+	{ "MMEA1_DRAMRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA1_DRAMWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA1_DRAMWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA1_RRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, RRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, RRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA1_WRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, WRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, WRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA1_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, IOWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, IOWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA1_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA1_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA1_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, IORD_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA1_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, IOWR_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA1_GMIRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA1_GMIWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA1_GMIWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA1_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA1_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA1_MAM_D0MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D0MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D0MEM_DED_COUNT),
+	},
+	{ "MMEA1_MAM_D1MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D1MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D1MEM_DED_COUNT),
+	},
+	{ "MMEA1_MAM_D2MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D2MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D2MEM_DED_COUNT),
+	},
+	{ "MMEA1_MAM_D3MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D3MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D3MEM_DED_COUNT),
+	},
+	{ "MMEA1_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA1_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA1_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA1_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA1_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA1_EDC_CNT3, IORD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA1_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA1_EDC_CNT3, IOWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA1_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA1_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA1_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA1_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT),
+	},
+
+	/* MMHAB Range 2*/
+	{ "MMEA2_DRAMRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA2_DRAMWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA2_DRAMWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA2_RRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, RRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, RRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA2_WRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, WRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, WRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA2_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, IOWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, IOWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA2_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA2_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA2_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, IORD_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA2_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, IOWR_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA2_GMIRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA2_GMIWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA2_GMIWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA2_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA2_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA2_MAM_D0MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D0MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D0MEM_DED_COUNT),
+	},
+	{ "MMEA2_MAM_D1MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, MAM_D1MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, MAM_D1MEM_DED_COUNT),
+	},
+	{ "MMEA2_MAM_D2MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, MAM_D2MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, MAM_D2MEM_DED_COUNT),
+	},
+	{ "MMEA2_MAM_D3MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, MAM_D3MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, MAM_D3MEM_DED_COUNT),
+	},
+	{ "MMEA2_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA2_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA2_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA2_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA2_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA2_EDC_CNT3, IORD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA2_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA2_EDC_CNT3, IOWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA2_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA2_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA2_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA2_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT),
+	},
+
+	/* MMHUB Rang 3 */
+	{ "MMEA3_DRAMRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA3_DRAMWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA3_DRAMWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA3_RRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, RRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, RRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA3_WRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, WRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, WRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA3_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT),
+        SOC15_REG_FIELD(MMEA3_EDC_CNT, IOWR_DATAMEM_SEC_COUNT),
+        SOC15_REG_FIELD(MMEA3_EDC_CNT, IOWR_DATAMEM_DED_COUNT),
+        },
+	{ "MMEA3_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA3_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA3_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, IORD_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA3_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, IOWR_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA3_GMIRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA3_GMIWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA3_GMIWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA3_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA3_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA3_MAM_D0MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D0MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D0MEM_DED_COUNT),
+	},
+	{ "MMEA3_MAM_D1MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D1MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D1MEM_DED_COUNT),
+	},
+	{ "MMEA3_MAM_D2MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D2MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D2MEM_DED_COUNT),
+	},
+	{ "MMEA3_MAM_D3MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D3MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D3MEM_DED_COUNT),
+	},
+	{ "MMEA3_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA3_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA3_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA3_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA3_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA3_EDC_CNT3, IORD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA3_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA3_EDC_CNT3, IOWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA3_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA3_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA3_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA3_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT),
+	},
+
+	/* MMHUB Range 4 */
+	{ "MMEA4_DRAMRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA4_DRAMWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA4_DRAMWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA4_RRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, RRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, RRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA4_WRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, WRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, WRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA4_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, IOWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, IOWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA4_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA4_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA4_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, IORD_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA4_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, IOWR_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA4_GMIRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA4_GMIWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA4_GMIWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA4_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA4_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA4_MAM_D0MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D0MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D0MEM_DED_COUNT),
+	},
+	{ "MMEA4_MAM_D1MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D1MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D1MEM_DED_COUNT),
+	},
+	{ "MMEA4_MAM_D2MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D2MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D2MEM_DED_COUNT),
+	},
+	{ "MMEA4_MAM_D3MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D3MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D3MEM_DED_COUNT),
+	},
+	{ "MMEA4_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA4_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA4_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA4_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA4_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA4_EDC_CNT3, IORD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA4_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA4_EDC_CNT3, IOWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA4_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA4_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA4_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA4_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT),
+	},
+
+	/* MMHUAB Range 5 */
+	{ "MMEA5_DRAMRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA5_DRAMWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA5_DRAMWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA5_RRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, RRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, RRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA5_WRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, WRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, WRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA5_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, IOWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, IOWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA5_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA5_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA5_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, IORD_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA5_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, IOWR_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA5_GMIRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA5_GMIWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA5_GMIWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA5_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA5_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA5_MAM_D0MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D0MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D0MEM_DED_COUNT),
+	},
+	{ "MMEA5_MAM_D1MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D1MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D1MEM_DED_COUNT),
+	},
+	{ "MMEA5_MAM_D2MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D2MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D2MEM_DED_COUNT),
+	},
+	{ "MMEA5_MAM_D3MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D3MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D3MEM_DED_COUNT),
+	},
+	{ "MMEA5_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA5_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA5_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA5_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA5_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA5_EDC_CNT3, IORD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA5_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA5_EDC_CNT3, IOWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA5_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA5_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA5_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA5_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT),
+	},
+};
+
+static const struct soc15_reg_entry mmhub_v1_7_edc_cnt_regs[] = {
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT2), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT3), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT2), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT3), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT2), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT3), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT2), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT3), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT2), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT3), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT2), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT3), 0, 0, 0 },
+};
+
+static int mmhub_v1_7_get_ras_error_count(struct amdgpu_device *adev,
+					  const struct soc15_reg_entry *reg,
+					  uint32_t value,
+					  uint32_t *sec_count,
+					  uint32_t *ded_count)
+{
+	uint32_t i;
+	uint32_t sec_cnt, ded_cnt;
+
+	for (i = 0; i < ARRAY_SIZE(mmhub_v1_7_ras_fields); i++) {
+		if(mmhub_v1_7_ras_fields[i].reg_offset != reg->reg_offset)
+			continue;
+
+		sec_cnt = (value &
+				mmhub_v1_7_ras_fields[i].sec_count_mask) >>
+				mmhub_v1_7_ras_fields[i].sec_count_shift;
+		if (sec_cnt) {
+			dev_info(adev->dev, "MMHUB SubBlock %s, SEC %d\n",
+				 mmhub_v1_7_ras_fields[i].name,
+				 sec_cnt);
+			*sec_count += sec_cnt;
+		}
+
+		ded_cnt = (value &
+				mmhub_v1_7_ras_fields[i].ded_count_mask) >>
+				mmhub_v1_7_ras_fields[i].ded_count_shift;
+		if (ded_cnt) {
+			dev_info(adev->dev, "MMHUB SubBlock %s, DED %d\n",
+				 mmhub_v1_7_ras_fields[i].name,
+				 ded_cnt);
+			*ded_count += ded_cnt;
+		}
+	}
+
+	return 0;
+}
+
+static void mmhub_v1_7_query_ras_error_count(struct amdgpu_device *adev,
+					     void *ras_error_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+	uint32_t sec_count = 0, ded_count = 0;
+	uint32_t i;
+	uint32_t reg_value;
+
+	err_data->ue_count = 0;
+	err_data->ce_count = 0;
+
+	for (i = 0; i < ARRAY_SIZE(mmhub_v1_7_edc_cnt_regs); i++) {
+		reg_value =
+			RREG32(SOC15_REG_ENTRY_OFFSET(mmhub_v1_7_edc_cnt_regs[i]));
+		if (reg_value)
+			mmhub_v1_7_get_ras_error_count(adev, &mmhub_v1_7_edc_cnt_regs[i],
+				reg_value, &sec_count, &ded_count);
+	}
+
+	err_data->ce_count += sec_count;
+	err_data->ue_count += ded_count;
+}
+
+static void mmhub_v1_7_reset_ras_error_count(struct amdgpu_device *adev)
+{
+	uint32_t i;
+
+	/* write 0 to reset the edc counters */
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__MMHUB)) {
+		for (i = 0; i < ARRAY_SIZE(mmhub_v1_7_edc_cnt_regs); i++)
+			WREG32(SOC15_REG_ENTRY_OFFSET(mmhub_v1_7_edc_cnt_regs[i]), 0);
+	}
+}
+
+static const struct soc15_reg_entry mmhub_v1_7_ea_err_status_regs[] = {
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_ERR_STATUS), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_ERR_STATUS), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_ERR_STATUS), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_ERR_STATUS), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_ERR_STATUS), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_ERR_STATUS), 0, 0, 0 },
+};
+
+static void mmhub_v1_7_query_ras_error_status(struct amdgpu_device *adev)
+{
+	int i;
+	uint32_t reg_value;
+
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__MMHUB))
+		return;
+
+	for (i = 0; i < ARRAY_SIZE(mmhub_v1_7_ea_err_status_regs); i++) {
+		reg_value =
+			RREG32(SOC15_REG_ENTRY_OFFSET(mmhub_v1_7_ea_err_status_regs[i]));
+		if (REG_GET_FIELD(reg_value, MMEA0_ERR_STATUS, SDP_RDRSP_STATUS) ||
+		    REG_GET_FIELD(reg_value, MMEA0_ERR_STATUS, SDP_WRRSP_STATUS) ||
+		    REG_GET_FIELD(reg_value, MMEA0_ERR_STATUS, SDP_RDRSP_DATAPARITY_ERROR)) {
+			dev_warn(adev->dev, "MMHUB EA err detected at instance: %d, status: 0x%x!\n",
+					i, reg_value);
+		}
+	}
+}
+
+static void mmhub_v1_7_reset_ras_error_status(struct amdgpu_device *adev)
+{
+	int i;
+	uint32_t reg_value;
+
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__MMHUB))
+		return;
+
+	for (i = 0; i < ARRAY_SIZE(mmhub_v1_7_ea_err_status_regs); i++) {
+		reg_value = RREG32(SOC15_REG_ENTRY_OFFSET(
+			mmhub_v1_7_ea_err_status_regs[i]));
+		reg_value = REG_SET_FIELD(reg_value, MMEA0_ERR_STATUS,
+					  CLEAR_ERROR_STATUS, 0x01);
+		WREG32(SOC15_REG_ENTRY_OFFSET(mmhub_v1_7_ea_err_status_regs[i]),
+		       reg_value);
+	}
+}
+
+struct amdgpu_ras_block_hw_ops mmhub_v1_7_ras_hw_ops = {
+	.query_ras_error_count = mmhub_v1_7_query_ras_error_count,
+	.reset_ras_error_count = mmhub_v1_7_reset_ras_error_count,
+	.query_ras_error_status = mmhub_v1_7_query_ras_error_status,
+	.reset_ras_error_status = mmhub_v1_7_reset_ras_error_status,
+};
+
+struct amdgpu_mmhub_ras mmhub_v1_7_ras = {
+	.ras_block = {
+		.hw_ops = &mmhub_v1_7_ras_hw_ops,
+	},
+};
+
+const struct amdgpu_mmhub_funcs mmhub_v1_7_funcs = {
+	.get_fb_location = mmhub_v1_7_get_fb_location,
+	.init = mmhub_v1_7_init,
+	.gart_enable = mmhub_v1_7_gart_enable,
+	.set_fault_enable_default = mmhub_v1_7_set_fault_enable_default,
+	.gart_disable = mmhub_v1_7_gart_disable,
+	.set_clockgating = mmhub_v1_7_set_clockgating,
+	.get_clockgating = mmhub_v1_7_get_clockgating,
+	.setup_vm_pt_regs = mmhub_v1_7_setup_vm_pt_regs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.h
new file mode 100644
index 000000000000..629f49052137
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __MMHUB_V1_7_H__
+#define __MMHUB_V1_7_H__
+
+extern const struct amdgpu_mmhub_funcs mmhub_v1_7_funcs;
+extern struct amdgpu_mmhub_ras mmhub_v1_7_ras;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c
new file mode 100644
index 000000000000..cc688ae79e84
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c
@@ -0,0 +1,871 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "mmhub_v1_8.h"
+
+#include "mmhub/mmhub_1_8_0_offset.h"
+#include "mmhub/mmhub_1_8_0_sh_mask.h"
+#include "vega10_enum.h"
+
+#include "soc15_common.h"
+#include "soc15.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_psp.h"
+
+#define regVM_L2_CNTL3_DEFAULT	0x80100007
+#define regVM_L2_CNTL4_DEFAULT	0x000000c1
+
+static u64 mmhub_v1_8_get_fb_location(struct amdgpu_device *adev)
+{
+	u64 base = RREG32_SOC15(MMHUB, 0, regMC_VM_FB_LOCATION_BASE);
+	u64 top = RREG32_SOC15(MMHUB, 0, regMC_VM_FB_LOCATION_TOP);
+
+	base &= MC_VM_FB_LOCATION_BASE__FB_BASE_MASK;
+	base <<= 24;
+
+	top &= MC_VM_FB_LOCATION_TOP__FB_TOP_MASK;
+	top <<= 24;
+
+	adev->gmc.fb_start = base;
+	adev->gmc.fb_end = top;
+
+	return base;
+}
+
+static void mmhub_v1_8_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid,
+				uint64_t page_table_base)
+{
+	struct amdgpu_vmhub *hub;
+	u32 inst_mask;
+	int i;
+
+	inst_mask = adev->aid_mask;
+	for_each_inst(i, inst_mask) {
+		hub = &adev->vmhub[AMDGPU_MMHUB0(i)];
+		WREG32_SOC15_OFFSET(MMHUB, i,
+				    regVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+				    hub->ctx_addr_distance * vmid,
+				    lower_32_bits(page_table_base));
+
+		WREG32_SOC15_OFFSET(MMHUB, i,
+				    regVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+				    hub->ctx_addr_distance * vmid,
+				    upper_32_bits(page_table_base));
+	}
+}
+
+static void mmhub_v1_8_init_gart_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t gart_start = amdgpu_virt_xgmi_migrate_enabled(adev) ?
+			adev->gmc.vram_start : adev->gmc.fb_start;
+	uint64_t pt_base;
+	u32 inst_mask;
+	int i;
+
+	if (adev->gmc.pdb0_bo)
+		pt_base = amdgpu_gmc_pd_addr(adev->gmc.pdb0_bo);
+	else
+		pt_base = amdgpu_gmc_pd_addr(adev->gart.bo);
+
+	mmhub_v1_8_setup_vm_pt_regs(adev, 0, pt_base);
+
+	/* If use GART for FB translation, vmid0 page table covers both
+	 * vram and system memory (gart)
+	 */
+	inst_mask = adev->aid_mask;
+	for_each_inst(i, inst_mask) {
+		if (adev->gmc.pdb0_bo) {
+			WREG32_SOC15(MMHUB, i,
+				     regVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+				     (u32)(gart_start >> 12));
+			WREG32_SOC15(MMHUB, i,
+				     regVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+				     (u32)(gart_start >> 44));
+
+			WREG32_SOC15(MMHUB, i,
+				     regVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+				     (u32)(adev->gmc.gart_end >> 12));
+			WREG32_SOC15(MMHUB, i,
+				     regVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+				     (u32)(adev->gmc.gart_end >> 44));
+
+		} else {
+			WREG32_SOC15(MMHUB, i,
+				     regVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+				     (u32)(adev->gmc.gart_start >> 12));
+			WREG32_SOC15(MMHUB, i,
+				     regVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+				     (u32)(adev->gmc.gart_start >> 44));
+
+			WREG32_SOC15(MMHUB, i,
+				     regVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+				     (u32)(adev->gmc.gart_end >> 12));
+			WREG32_SOC15(MMHUB, i,
+				     regVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+				     (u32)(adev->gmc.gart_end >> 44));
+		}
+	}
+}
+
+static void mmhub_v1_8_init_system_aperture_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp, inst_mask;
+	uint64_t value;
+	int i;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	inst_mask = adev->aid_mask;
+	for_each_inst(i, inst_mask) {
+		/* Program the AGP BAR */
+		WREG32_SOC15(MMHUB, i, regMC_VM_AGP_BASE, 0);
+		WREG32_SOC15(MMHUB, i, regMC_VM_AGP_BOT,
+			     adev->gmc.agp_start >> 24);
+		WREG32_SOC15(MMHUB, i, regMC_VM_AGP_TOP,
+			     adev->gmc.agp_end >> 24);
+
+		/* Program the system aperture low logical page number. */
+		WREG32_SOC15(MMHUB, i, regMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+			min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
+
+		WREG32_SOC15(MMHUB, i, regMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+			max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
+
+		/* In the case squeezing vram into GART aperture, we don't use
+		 * FB aperture and AGP aperture. Disable them.
+		 */
+		if (adev->gmc.pdb0_bo) {
+			WREG32_SOC15(MMHUB, i, regMC_VM_AGP_BOT, 0xFFFFFF);
+			WREG32_SOC15(MMHUB, i, regMC_VM_AGP_TOP, 0);
+			WREG32_SOC15(MMHUB, i, regMC_VM_FB_LOCATION_TOP, 0);
+			WREG32_SOC15(MMHUB, i, regMC_VM_FB_LOCATION_BASE,
+				     0x00FFFFFF);
+			WREG32_SOC15(MMHUB, i,
+				     regMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+				     0x3FFFFFFF);
+			WREG32_SOC15(MMHUB, i,
+				     regMC_VM_SYSTEM_APERTURE_HIGH_ADDR, 0);
+		}
+
+		/* Set default page address. */
+		value = amdgpu_gmc_vram_mc2pa(adev, adev->mem_scratch.gpu_addr);
+		WREG32_SOC15(MMHUB, i, regMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
+			     (u32)(value >> 12));
+		WREG32_SOC15(MMHUB, i, regMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
+			     (u32)(value >> 44));
+
+		/* Program "protection fault". */
+		WREG32_SOC15(MMHUB, i,
+			     regVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
+			     (u32)(adev->dummy_page_addr >> 12));
+		WREG32_SOC15(MMHUB, i,
+			     regVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+			     (u32)((u64)adev->dummy_page_addr >> 44));
+
+		tmp = RREG32_SOC15(MMHUB, i, regVM_L2_PROTECTION_FAULT_CNTL2);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL2,
+				    ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+		WREG32_SOC15(MMHUB, i, regVM_L2_PROTECTION_FAULT_CNTL2, tmp);
+	}
+}
+
+static void mmhub_v1_8_init_tlb_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp, inst_mask;
+	int i;
+
+	if (amdgpu_sriov_reg_indirect_l1_tlb_cntl(adev)) {
+		tmp = RREG32_SOC15(MMHUB, 0, regMC_VM_MX_L1_TLB_CNTL);
+
+		tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB,
+				    1);
+		tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+				    SYSTEM_ACCESS_MODE, 3);
+		tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+				    ENABLE_ADVANCED_DRIVER_MODEL, 1);
+		tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+				    SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+		tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+				    MTYPE, MTYPE_UC);/* XXX for emulation. */
+		tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ATC_EN, 1);
+
+		psp_reg_program_no_ring(&adev->psp, tmp, PSP_REG_MMHUB_L1_TLB_CNTL);
+	} else {
+		inst_mask = adev->aid_mask;
+		for_each_inst(i, inst_mask) {
+			tmp = RREG32_SOC15(MMHUB, i, regMC_VM_MX_L1_TLB_CNTL);
+
+			tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB,
+					    1);
+			tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+					    SYSTEM_ACCESS_MODE, 3);
+			tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+					    ENABLE_ADVANCED_DRIVER_MODEL, 1);
+			tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+					    SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+			tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+					    MTYPE, MTYPE_UC);/* XXX for emulation. */
+			tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ATC_EN, 1);
+
+			WREG32_SOC15(MMHUB, i, regMC_VM_MX_L1_TLB_CNTL, tmp);
+		}
+	}
+}
+
+/* Set snoop bit for SDMA so that SDMA writes probe-invalidates RW lines */
+static void mmhub_v1_8_init_snoop_override_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp, inst_mask;
+	int i, j;
+	uint32_t distance = regDAGB1_WRCLI_GPU_SNOOP_OVERRIDE -
+			    regDAGB0_WRCLI_GPU_SNOOP_OVERRIDE;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	inst_mask = adev->aid_mask;
+	for_each_inst(i, inst_mask) {
+		for (j = 0; j < 5; j++) { /* DAGB instances */
+			tmp = RREG32_SOC15_OFFSET(MMHUB, i,
+				regDAGB0_WRCLI_GPU_SNOOP_OVERRIDE, j * distance);
+			tmp |= (1 << 15); /* SDMA client is BIT15 */
+			WREG32_SOC15_OFFSET(MMHUB, i,
+				regDAGB0_WRCLI_GPU_SNOOP_OVERRIDE, j * distance, tmp);
+
+			tmp = RREG32_SOC15_OFFSET(MMHUB, i,
+				regDAGB0_WRCLI_GPU_SNOOP_OVERRIDE_VALUE, j * distance);
+			tmp |= (1 << 15);
+			WREG32_SOC15_OFFSET(MMHUB, i,
+				regDAGB0_WRCLI_GPU_SNOOP_OVERRIDE_VALUE, j * distance, tmp);
+		}
+	}
+}
+
+static void mmhub_v1_8_init_cache_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp, inst_mask;
+	int i;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/* Setup L2 cache */
+	inst_mask = adev->aid_mask;
+	for_each_inst(i, inst_mask) {
+		tmp = RREG32_SOC15(MMHUB, i, regVM_L2_CNTL);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 1);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL,
+				    ENABLE_L2_FRAGMENT_PROCESSING, 1);
+		/* XXX for emulation, Refer to closed source code.*/
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL,
+				    L2_PDE0_CACHE_TAG_GENERATION_MODE, 0);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, PDE_FAULT_CLASSIFICATION,
+				    0);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL,
+				    CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL,
+				    IDENTITY_MODE_FRAGMENT_SIZE, 0);
+		WREG32_SOC15(MMHUB, i, regVM_L2_CNTL, tmp);
+
+		tmp = RREG32_SOC15(MMHUB, i, regVM_L2_CNTL2);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS,
+				    1);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
+		WREG32_SOC15(MMHUB, i, regVM_L2_CNTL2, tmp);
+
+		tmp = regVM_L2_CNTL3_DEFAULT;
+		if (adev->gmc.translate_further) {
+			tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 12);
+			tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3,
+					    L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+		} else {
+			tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 9);
+			tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3,
+					    L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+		}
+		WREG32_SOC15(MMHUB, i, regVM_L2_CNTL3, tmp);
+
+		tmp = regVM_L2_CNTL4_DEFAULT;
+		/* For AMD APP APUs setup WC memory */
+		if (adev->gmc.xgmi.connected_to_cpu || adev->gmc.is_app_apu) {
+			tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4,
+					    VMC_TAP_PDE_REQUEST_PHYSICAL, 1);
+			tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4,
+					    VMC_TAP_PTE_REQUEST_PHYSICAL, 1);
+		} else {
+			tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4,
+					    VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
+			tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4,
+					    VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
+		}
+		WREG32_SOC15(MMHUB, i, regVM_L2_CNTL4, tmp);
+	}
+}
+
+static void mmhub_v1_8_enable_system_domain(struct amdgpu_device *adev)
+{
+	uint32_t tmp, inst_mask;
+	int i;
+
+	inst_mask = adev->aid_mask;
+	for_each_inst(i, inst_mask) {
+		tmp = RREG32_SOC15(MMHUB, i, regVM_CONTEXT0_CNTL);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH,
+				adev->gmc.vmid0_page_table_depth);
+		tmp = REG_SET_FIELD(tmp,
+				    VM_CONTEXT0_CNTL, PAGE_TABLE_BLOCK_SIZE,
+				    adev->gmc.vmid0_page_table_block_size);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL,
+				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
+		WREG32_SOC15(MMHUB, i, regVM_CONTEXT0_CNTL, tmp);
+	}
+}
+
+static void mmhub_v1_8_disable_identity_aperture(struct amdgpu_device *adev)
+{
+	u32 inst_mask;
+	int i;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	inst_mask = adev->aid_mask;
+	for_each_inst(i, inst_mask) {
+		WREG32_SOC15(MMHUB, i,
+			     regVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32,
+			     0XFFFFFFFF);
+		WREG32_SOC15(MMHUB, i,
+			     regVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32,
+			     0x0000000F);
+
+		WREG32_SOC15(MMHUB, i,
+			     regVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32,
+			     0);
+		WREG32_SOC15(MMHUB, i,
+			     regVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32,
+			     0);
+
+		WREG32_SOC15(MMHUB, i,
+			     regVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32, 0);
+		WREG32_SOC15(MMHUB, i,
+			     regVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32, 0);
+	}
+}
+
+static void mmhub_v1_8_setup_vmid_config(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub;
+	unsigned int num_level, block_size;
+	uint32_t tmp, inst_mask;
+	int i, j;
+
+	num_level = adev->vm_manager.num_level;
+	block_size = adev->vm_manager.block_size;
+	if (adev->gmc.translate_further)
+		num_level -= 1;
+	else
+		block_size -= 9;
+
+	inst_mask = adev->aid_mask;
+	for_each_inst(j, inst_mask) {
+		hub = &adev->vmhub[AMDGPU_MMHUB0(j)];
+		for (i = 0; i <= 14; i++) {
+			tmp = RREG32_SOC15_OFFSET(MMHUB, j, regVM_CONTEXT1_CNTL,
+						  i * hub->ctx_distance);
+			tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+					    ENABLE_CONTEXT, 1);
+			tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+					    PAGE_TABLE_DEPTH, num_level);
+			tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+			tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+			tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+			tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+			tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+			tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+			tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+			tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+					    PAGE_TABLE_BLOCK_SIZE,
+					    block_size);
+			/* On 9.4.3, XNACK can be enabled in the SQ
+			 * per-process. Retry faults need to be enabled for
+			 * that to work.
+			 */
+			tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 1);
+			WREG32_SOC15_OFFSET(MMHUB, j, regVM_CONTEXT1_CNTL,
+					    i * hub->ctx_distance, tmp);
+			WREG32_SOC15_OFFSET(MMHUB, j,
+				regVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,
+				i * hub->ctx_addr_distance, 0);
+			WREG32_SOC15_OFFSET(MMHUB, j,
+				regVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32,
+				i * hub->ctx_addr_distance, 0);
+			WREG32_SOC15_OFFSET(MMHUB, j,
+				regVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32,
+				i * hub->ctx_addr_distance,
+				lower_32_bits(adev->vm_manager.max_pfn - 1));
+			WREG32_SOC15_OFFSET(MMHUB, j,
+				regVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32,
+				i * hub->ctx_addr_distance,
+				upper_32_bits(adev->vm_manager.max_pfn - 1));
+		}
+	}
+}
+
+static void mmhub_v1_8_program_invalidation(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub;
+	u32 i, j, inst_mask;
+
+	inst_mask = adev->aid_mask;
+	for_each_inst(j, inst_mask) {
+		hub = &adev->vmhub[AMDGPU_MMHUB0(j)];
+		for (i = 0; i < 18; ++i) {
+			WREG32_SOC15_OFFSET(MMHUB, j,
+					regVM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
+					i * hub->eng_addr_distance, 0xffffffff);
+			WREG32_SOC15_OFFSET(MMHUB, j,
+					regVM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
+					i * hub->eng_addr_distance, 0x1f);
+		}
+	}
+}
+
+static int mmhub_v1_8_gart_enable(struct amdgpu_device *adev)
+{
+	/* GART Enable. */
+	mmhub_v1_8_init_gart_aperture_regs(adev);
+	mmhub_v1_8_init_system_aperture_regs(adev);
+	mmhub_v1_8_init_tlb_regs(adev);
+	mmhub_v1_8_init_cache_regs(adev);
+	mmhub_v1_8_init_snoop_override_regs(adev);
+
+	mmhub_v1_8_enable_system_domain(adev);
+	mmhub_v1_8_disable_identity_aperture(adev);
+	mmhub_v1_8_setup_vmid_config(adev);
+	mmhub_v1_8_program_invalidation(adev);
+
+	return 0;
+}
+
+static void mmhub_v1_8_disable_l1_tlb(struct amdgpu_device *adev)
+{
+	u32 tmp;
+	u32 i, inst_mask;
+
+	if (amdgpu_sriov_reg_indirect_l1_tlb_cntl(adev)) {
+		tmp = RREG32_SOC15(MMHUB, 0, regMC_VM_MX_L1_TLB_CNTL);
+		tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
+		tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+				    ENABLE_ADVANCED_DRIVER_MODEL, 0);
+		psp_reg_program_no_ring(&adev->psp, tmp, PSP_REG_MMHUB_L1_TLB_CNTL);
+	} else {
+		inst_mask = adev->aid_mask;
+		for_each_inst(i, inst_mask) {
+			tmp = RREG32_SOC15(MMHUB, i, regMC_VM_MX_L1_TLB_CNTL);
+			tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB,
+					    0);
+			tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+					    ENABLE_ADVANCED_DRIVER_MODEL, 0);
+			WREG32_SOC15(MMHUB, i, regMC_VM_MX_L1_TLB_CNTL, tmp);
+		}
+	}
+}
+
+static void mmhub_v1_8_gart_disable(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub;
+	u32 tmp;
+	u32 i, j, inst_mask;
+
+	/* Disable all tables */
+	inst_mask = adev->aid_mask;
+	for_each_inst(j, inst_mask) {
+		hub = &adev->vmhub[AMDGPU_MMHUB0(j)];
+		for (i = 0; i < 16; i++)
+			WREG32_SOC15_OFFSET(MMHUB, j, regVM_CONTEXT0_CNTL,
+					    i * hub->ctx_distance, 0);
+		if (!amdgpu_sriov_vf(adev)) {
+			/* Setup L2 cache */
+			tmp = RREG32_SOC15(MMHUB, j, regVM_L2_CNTL);
+			tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE,
+					    0);
+			WREG32_SOC15(MMHUB, j, regVM_L2_CNTL, tmp);
+			WREG32_SOC15(MMHUB, j, regVM_L2_CNTL3, 0);
+		}
+	}
+
+	mmhub_v1_8_disable_l1_tlb(adev);
+}
+
+/**
+ * mmhub_v1_8_set_fault_enable_default - update GART/VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+static void mmhub_v1_8_set_fault_enable_default(struct amdgpu_device *adev, bool value)
+{
+	u32 tmp, inst_mask;
+	int i;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	inst_mask = adev->aid_mask;
+	for_each_inst(i, inst_mask) {
+		tmp = RREG32_SOC15(MMHUB, i, regVM_L2_PROTECTION_FAULT_CNTL);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				PDE1_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				PDE2_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT,
+			value);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				NACK_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+		if (!value) {
+			tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+					    CRASH_ON_NO_RETRY_FAULT, 1);
+			tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+					    CRASH_ON_RETRY_FAULT, 1);
+		}
+
+		WREG32_SOC15(MMHUB, i, regVM_L2_PROTECTION_FAULT_CNTL, tmp);
+	}
+}
+
+static void mmhub_v1_8_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub;
+	u32 inst_mask;
+	int i;
+
+	inst_mask = adev->aid_mask;
+	for_each_inst(i, inst_mask) {
+		hub = &adev->vmhub[AMDGPU_MMHUB0(i)];
+
+		hub->ctx0_ptb_addr_lo32 = SOC15_REG_OFFSET(MMHUB, i,
+			regVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32);
+		hub->ctx0_ptb_addr_hi32 = SOC15_REG_OFFSET(MMHUB, i,
+			regVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32);
+		hub->vm_inv_eng0_req =
+			SOC15_REG_OFFSET(MMHUB, i, regVM_INVALIDATE_ENG0_REQ);
+		hub->vm_inv_eng0_ack =
+			SOC15_REG_OFFSET(MMHUB, i, regVM_INVALIDATE_ENG0_ACK);
+		hub->vm_context0_cntl =
+			SOC15_REG_OFFSET(MMHUB, i, regVM_CONTEXT0_CNTL);
+		hub->vm_l2_pro_fault_status = SOC15_REG_OFFSET(MMHUB, i,
+			regVM_L2_PROTECTION_FAULT_STATUS);
+		hub->vm_l2_pro_fault_cntl = SOC15_REG_OFFSET(MMHUB, i,
+			regVM_L2_PROTECTION_FAULT_CNTL);
+
+		hub->ctx_distance = regVM_CONTEXT1_CNTL - regVM_CONTEXT0_CNTL;
+		hub->ctx_addr_distance =
+			regVM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 -
+			regVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32;
+		hub->eng_distance = regVM_INVALIDATE_ENG1_REQ -
+			regVM_INVALIDATE_ENG0_REQ;
+		hub->eng_addr_distance = regVM_INVALIDATE_ENG1_ADDR_RANGE_LO32 -
+			regVM_INVALIDATE_ENG0_ADDR_RANGE_LO32;
+	}
+}
+
+static int mmhub_v1_8_set_clockgating(struct amdgpu_device *adev,
+				      enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static void mmhub_v1_8_get_clockgating(struct amdgpu_device *adev, u64 *flags)
+{
+
+}
+
+const struct amdgpu_mmhub_funcs mmhub_v1_8_funcs = {
+	.get_fb_location = mmhub_v1_8_get_fb_location,
+	.init = mmhub_v1_8_init,
+	.gart_enable = mmhub_v1_8_gart_enable,
+	.set_fault_enable_default = mmhub_v1_8_set_fault_enable_default,
+	.gart_disable = mmhub_v1_8_gart_disable,
+	.setup_vm_pt_regs = mmhub_v1_8_setup_vm_pt_regs,
+	.set_clockgating = mmhub_v1_8_set_clockgating,
+	.get_clockgating = mmhub_v1_8_get_clockgating,
+};
+
+static const struct amdgpu_ras_err_status_reg_entry mmhub_v1_8_ce_reg_list[] = {
+	{AMDGPU_RAS_REG_ENTRY(MMHUB, 0, regMMEA0_CE_ERR_STATUS_LO, regMMEA0_CE_ERR_STATUS_HI),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "MMEA0"},
+	{AMDGPU_RAS_REG_ENTRY(MMHUB, 0, regMMEA1_CE_ERR_STATUS_LO, regMMEA1_CE_ERR_STATUS_HI),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "MMEA1"},
+	{AMDGPU_RAS_REG_ENTRY(MMHUB, 0, regMMEA2_CE_ERR_STATUS_LO, regMMEA2_CE_ERR_STATUS_HI),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "MMEA2"},
+	{AMDGPU_RAS_REG_ENTRY(MMHUB, 0, regMMEA3_CE_ERR_STATUS_LO, regMMEA3_CE_ERR_STATUS_HI),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "MMEA3"},
+	{AMDGPU_RAS_REG_ENTRY(MMHUB, 0, regMMEA4_CE_ERR_STATUS_LO, regMMEA4_CE_ERR_STATUS_HI),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "MMEA4"},
+	{AMDGPU_RAS_REG_ENTRY(MMHUB, 0, regMM_CANE_CE_ERR_STATUS_LO, regMM_CANE_CE_ERR_STATUS_HI),
+	1, 0, "MM_CANE"},
+};
+
+static const struct amdgpu_ras_err_status_reg_entry mmhub_v1_8_ue_reg_list[] = {
+	{AMDGPU_RAS_REG_ENTRY(MMHUB, 0, regMMEA0_UE_ERR_STATUS_LO, regMMEA0_UE_ERR_STATUS_HI),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "MMEA0"},
+	{AMDGPU_RAS_REG_ENTRY(MMHUB, 0, regMMEA1_UE_ERR_STATUS_LO, regMMEA1_UE_ERR_STATUS_HI),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "MMEA1"},
+	{AMDGPU_RAS_REG_ENTRY(MMHUB, 0, regMMEA2_UE_ERR_STATUS_LO, regMMEA2_UE_ERR_STATUS_HI),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "MMEA2"},
+	{AMDGPU_RAS_REG_ENTRY(MMHUB, 0, regMMEA3_UE_ERR_STATUS_LO, regMMEA3_UE_ERR_STATUS_HI),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "MMEA3"},
+	{AMDGPU_RAS_REG_ENTRY(MMHUB, 0, regMMEA4_UE_ERR_STATUS_LO, regMMEA4_UE_ERR_STATUS_HI),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "MMEA4"},
+	{AMDGPU_RAS_REG_ENTRY(MMHUB, 0, regMM_CANE_UE_ERR_STATUS_LO, regMM_CANE_UE_ERR_STATUS_HI),
+	1, 0, "MM_CANE"},
+};
+
+static const struct amdgpu_ras_memory_id_entry mmhub_v1_8_ras_memory_list[] = {
+	{AMDGPU_MMHUB_WGMI_PAGEMEM, "MMEA_WGMI_PAGEMEM"},
+	{AMDGPU_MMHUB_RGMI_PAGEMEM, "MMEA_RGMI_PAGEMEM"},
+	{AMDGPU_MMHUB_WDRAM_PAGEMEM, "MMEA_WDRAM_PAGEMEM"},
+	{AMDGPU_MMHUB_RDRAM_PAGEMEM, "MMEA_RDRAM_PAGEMEM"},
+	{AMDGPU_MMHUB_WIO_CMDMEM, "MMEA_WIO_CMDMEM"},
+	{AMDGPU_MMHUB_RIO_CMDMEM, "MMEA_RIO_CMDMEM"},
+	{AMDGPU_MMHUB_WGMI_CMDMEM, "MMEA_WGMI_CMDMEM"},
+	{AMDGPU_MMHUB_RGMI_CMDMEM, "MMEA_RGMI_CMDMEM"},
+	{AMDGPU_MMHUB_WDRAM_CMDMEM, "MMEA_WDRAM_CMDMEM"},
+	{AMDGPU_MMHUB_RDRAM_CMDMEM, "MMEA_RDRAM_CMDMEM"},
+	{AMDGPU_MMHUB_MAM_DMEM0, "MMEA_MAM_DMEM0"},
+	{AMDGPU_MMHUB_MAM_DMEM1, "MMEA_MAM_DMEM1"},
+	{AMDGPU_MMHUB_MAM_DMEM2, "MMEA_MAM_DMEM2"},
+	{AMDGPU_MMHUB_MAM_DMEM3, "MMEA_MAM_DMEM3"},
+	{AMDGPU_MMHUB_WRET_TAGMEM, "MMEA_WRET_TAGMEM"},
+	{AMDGPU_MMHUB_RRET_TAGMEM, "MMEA_RRET_TAGMEM"},
+	{AMDGPU_MMHUB_WIO_DATAMEM, "MMEA_WIO_DATAMEM"},
+	{AMDGPU_MMHUB_WGMI_DATAMEM, "MMEA_WGMI_DATAMEM"},
+	{AMDGPU_MMHUB_WDRAM_DATAMEM, "MMEA_WDRAM_DATAMEM"},
+};
+
+static void mmhub_v1_8_inst_query_ras_error_count(struct amdgpu_device *adev,
+						  uint32_t mmhub_inst,
+						  void *ras_err_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_err_status;
+	unsigned long ue_count = 0, ce_count = 0;
+
+	/* NOTE: mmhub is converted by aid_mask and the range is 0-3,
+	 * which can be used as die ID directly */
+	struct amdgpu_smuio_mcm_config_info mcm_info = {
+		.socket_id = adev->smuio.funcs->get_socket_id(adev),
+		.die_id = mmhub_inst,
+	};
+
+	amdgpu_ras_inst_query_ras_error_count(adev,
+					mmhub_v1_8_ce_reg_list,
+					ARRAY_SIZE(mmhub_v1_8_ce_reg_list),
+					mmhub_v1_8_ras_memory_list,
+					ARRAY_SIZE(mmhub_v1_8_ras_memory_list),
+					mmhub_inst,
+					AMDGPU_RAS_ERROR__SINGLE_CORRECTABLE,
+					&ce_count);
+	amdgpu_ras_inst_query_ras_error_count(adev,
+					mmhub_v1_8_ue_reg_list,
+					ARRAY_SIZE(mmhub_v1_8_ue_reg_list),
+					mmhub_v1_8_ras_memory_list,
+					ARRAY_SIZE(mmhub_v1_8_ras_memory_list),
+					mmhub_inst,
+					AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE,
+					&ue_count);
+
+	amdgpu_ras_error_statistic_ce_count(err_data, &mcm_info, ce_count);
+	amdgpu_ras_error_statistic_ue_count(err_data, &mcm_info, ue_count);
+}
+
+static void mmhub_v1_8_query_ras_error_count(struct amdgpu_device *adev,
+					     void *ras_err_status)
+{
+	uint32_t inst_mask;
+	uint32_t i;
+
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__MMHUB)) {
+		dev_warn(adev->dev, "MMHUB RAS is not supported\n");
+		return;
+	}
+
+	inst_mask = adev->aid_mask;
+	for_each_inst(i, inst_mask)
+		mmhub_v1_8_inst_query_ras_error_count(adev, i, ras_err_status);
+}
+
+static void mmhub_v1_8_inst_reset_ras_error_count(struct amdgpu_device *adev,
+						  uint32_t mmhub_inst)
+{
+	amdgpu_ras_inst_reset_ras_error_count(adev,
+					mmhub_v1_8_ce_reg_list,
+					ARRAY_SIZE(mmhub_v1_8_ce_reg_list),
+					mmhub_inst);
+	amdgpu_ras_inst_reset_ras_error_count(adev,
+					mmhub_v1_8_ue_reg_list,
+					ARRAY_SIZE(mmhub_v1_8_ue_reg_list),
+					mmhub_inst);
+}
+
+static void mmhub_v1_8_reset_ras_error_count(struct amdgpu_device *adev)
+{
+	uint32_t inst_mask;
+	uint32_t i;
+
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__MMHUB)) {
+		dev_warn(adev->dev, "MMHUB RAS is not supported\n");
+		return;
+	}
+
+	inst_mask = adev->aid_mask;
+	for_each_inst(i, inst_mask)
+		mmhub_v1_8_inst_reset_ras_error_count(adev, i);
+}
+
+static const struct amdgpu_ras_block_hw_ops mmhub_v1_8_ras_hw_ops = {
+	.query_ras_error_count = mmhub_v1_8_query_ras_error_count,
+	.reset_ras_error_count = mmhub_v1_8_reset_ras_error_count,
+};
+
+static int mmhub_v1_8_aca_bank_parser(struct aca_handle *handle, struct aca_bank *bank,
+				      enum aca_smu_type type, void *data)
+{
+	struct aca_bank_info info;
+	u64 misc0;
+	int ret;
+
+	ret = aca_bank_info_decode(bank, &info);
+	if (ret)
+		return ret;
+
+	misc0 = bank->regs[ACA_REG_IDX_MISC0];
+	switch (type) {
+	case ACA_SMU_TYPE_UE:
+		bank->aca_err_type = ACA_ERROR_TYPE_UE;
+		ret = aca_error_cache_log_bank_error(handle, &info, ACA_ERROR_TYPE_UE,
+						     1ULL);
+		break;
+	case ACA_SMU_TYPE_CE:
+		bank->aca_err_type = ACA_ERROR_TYPE_CE;
+		ret = aca_error_cache_log_bank_error(handle, &info, bank->aca_err_type,
+						     ACA_REG__MISC0__ERRCNT(misc0));
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return ret;
+}
+
+/* reference to smu driver if header file */
+static int mmhub_v1_8_err_codes[] = {
+	0, 1, 2, 3, 4, /* CODE_DAGB0 - 4 */
+	5, 6, 7, 8, 9, /* CODE_EA0 - 4 */
+	10, /* CODE_UTCL2_ROUTER */
+	11, /* CODE_VML2 */
+	12, /* CODE_VML2_WALKER */
+	13, /* CODE_MMCANE */
+};
+
+static bool mmhub_v1_8_aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank,
+					 enum aca_smu_type type, void *data)
+{
+	u32 instlo;
+
+	instlo = ACA_REG__IPID__INSTANCEIDLO(bank->regs[ACA_REG_IDX_IPID]);
+	instlo &= GENMASK(31, 1);
+
+	if (instlo != mmSMNAID_AID0_MCA_SMU)
+		return false;
+
+	if (aca_bank_check_error_codes(handle->adev, bank,
+				       mmhub_v1_8_err_codes,
+				       ARRAY_SIZE(mmhub_v1_8_err_codes)))
+		return false;
+
+	return true;
+}
+
+static const struct aca_bank_ops mmhub_v1_8_aca_bank_ops = {
+	.aca_bank_parser = mmhub_v1_8_aca_bank_parser,
+	.aca_bank_is_valid = mmhub_v1_8_aca_bank_is_valid,
+};
+
+static const struct aca_info mmhub_v1_8_aca_info = {
+	.hwip = ACA_HWIP_TYPE_SMU,
+	.mask = ACA_ERROR_UE_MASK,
+	.bank_ops = &mmhub_v1_8_aca_bank_ops,
+};
+
+static int mmhub_v1_8_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
+{
+	int r;
+
+	r = amdgpu_ras_block_late_init(adev, ras_block);
+	if (r)
+		return r;
+
+	r = amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__MMHUB,
+				&mmhub_v1_8_aca_info, NULL);
+	if (r)
+		goto late_fini;
+
+	return 0;
+
+late_fini:
+	amdgpu_ras_block_late_fini(adev, ras_block);
+
+	return r;
+}
+
+struct amdgpu_mmhub_ras mmhub_v1_8_ras = {
+	.ras_block = {
+		.hw_ops = &mmhub_v1_8_ras_hw_ops,
+		.ras_late_init = mmhub_v1_8_ras_late_init,
+	},
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.h
new file mode 100644
index 000000000000..126f0075ac50
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __MMHUB_V1_8_H__
+#define __MMHUB_V1_8_H__
+
+extern const struct amdgpu_mmhub_funcs mmhub_v1_8_funcs;
+extern struct amdgpu_mmhub_ras mmhub_v1_8_ras;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c
new file mode 100644
index 000000000000..a0cc8e218ca1
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c
@@ -0,0 +1,718 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "mmhub_v2_0.h"
+
+#include "mmhub/mmhub_2_0_0_offset.h"
+#include "mmhub/mmhub_2_0_0_sh_mask.h"
+#include "mmhub/mmhub_2_0_0_default.h"
+#include "navi10_enum.h"
+
+#include "gc/gc_10_1_0_offset.h"
+#include "soc15_common.h"
+
+#define mmDAGB0_CNTL_MISC2_Sienna_Cichlid                       0x0070
+#define mmDAGB0_CNTL_MISC2_Sienna_Cichlid_BASE_IDX              0
+
+static const char *mmhub_client_ids_navi1x[][2] = {
+	[3][0] = "DCEDMC",
+	[4][0] = "DCEVGA",
+	[5][0] = "MP0",
+	[6][0] = "MP1",
+	[13][0] = "VMC",
+	[14][0] = "HDP",
+	[15][0] = "OSS",
+	[16][0] = "VCNU",
+	[17][0] = "JPEG",
+	[18][0] = "VCN",
+	[3][1] = "DCEDMC",
+	[4][1] = "DCEXFC",
+	[5][1] = "DCEVGA",
+	[6][1] = "DCEDWB",
+	[7][1] = "MP0",
+	[8][1] = "MP1",
+	[9][1] = "DBGU1",
+	[10][1] = "DBGU0",
+	[11][1] = "XDP",
+	[14][1] = "HDP",
+	[15][1] = "OSS",
+	[16][1] = "VCNU",
+	[17][1] = "JPEG",
+	[18][1] = "VCN",
+};
+
+static const char *mmhub_client_ids_sienna_cichlid[][2] = {
+	[3][0] = "DCEDMC",
+	[4][0] = "DCEVGA",
+	[5][0] = "MP0",
+	[6][0] = "MP1",
+	[8][0] = "VMC",
+	[9][0] = "VCNU0",
+	[10][0] = "JPEG",
+	[12][0] = "VCNU1",
+	[13][0] = "VCN1",
+	[14][0] = "HDP",
+	[15][0] = "OSS",
+	[32+11][0] = "VCN0",
+	[0][1] = "DBGU0",
+	[1][1] = "DBGU1",
+	[2][1] = "DCEDWB",
+	[3][1] = "DCEDMC",
+	[4][1] = "DCEVGA",
+	[5][1] = "MP0",
+	[6][1] = "MP1",
+	[7][1] = "XDP",
+	[9][1] = "VCNU0",
+	[10][1] = "JPEG",
+	[11][1] = "VCN0",
+	[12][1] = "VCNU1",
+	[13][1] = "VCN1",
+	[14][1] = "HDP",
+	[15][1] = "OSS",
+};
+
+static const char *mmhub_client_ids_beige_goby[][2] = {
+	[3][0] = "DCEDMC",
+	[4][0] = "DCEVGA",
+	[5][0] = "MP0",
+	[6][0] = "MP1",
+	[8][0] = "VMC",
+	[9][0] = "VCNU0",
+	[11][0] = "VCN0",
+	[14][0] = "HDP",
+	[15][0] = "OSS",
+	[0][1] = "DBGU0",
+	[1][1] = "DBGU1",
+	[2][1] = "DCEDWB",
+	[3][1] = "DCEDMC",
+	[4][1] = "DCEVGA",
+	[5][1] = "MP0",
+	[6][1] = "MP1",
+	[7][1] = "XDP",
+	[9][1] = "VCNU0",
+	[11][1] = "VCN0",
+	[14][1] = "HDP",
+	[15][1] = "OSS",
+};
+
+static uint32_t mmhub_v2_0_get_invalidate_req(unsigned int vmid,
+					      uint32_t flush_type)
+{
+	u32 req = 0;
+
+	/* invalidate using legacy mode on vmid*/
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ,
+			    PER_VMID_INVALIDATE_REQ, 1 << vmid);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, FLUSH_TYPE, flush_type);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PTES, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE0, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE1, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE2, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L1_PTES, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ,
+			    CLEAR_PROTECTION_FAULT_STATUS_ADDR,	0);
+
+	return req;
+}
+
+static void
+mmhub_v2_0_print_l2_protection_fault_status(struct amdgpu_device *adev,
+					     uint32_t status)
+{
+	uint32_t cid, rw;
+	const char *mmhub_cid = NULL;
+
+	cid = REG_GET_FIELD(status,
+			    MMVM_L2_PROTECTION_FAULT_STATUS, CID);
+	rw = REG_GET_FIELD(status,
+			   MMVM_L2_PROTECTION_FAULT_STATUS, RW);
+
+	dev_err(adev->dev,
+		"MMVM_L2_PROTECTION_FAULT_STATUS:0x%08X\n",
+		status);
+	switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) {
+	case IP_VERSION(2, 0, 0):
+	case IP_VERSION(2, 0, 2):
+		mmhub_cid = mmhub_client_ids_navi1x[cid][rw];
+		break;
+	case IP_VERSION(2, 1, 0):
+	case IP_VERSION(2, 1, 1):
+		mmhub_cid = mmhub_client_ids_sienna_cichlid[cid][rw];
+		break;
+	case IP_VERSION(2, 1, 2):
+		mmhub_cid = mmhub_client_ids_beige_goby[cid][rw];
+		break;
+	default:
+		mmhub_cid = NULL;
+		break;
+	}
+	dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n",
+		mmhub_cid ? mmhub_cid : "unknown", cid);
+	dev_err(adev->dev, "\t MORE_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, MORE_FAULTS));
+	dev_err(adev->dev, "\t WALKER_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, WALKER_ERROR));
+	dev_err(adev->dev, "\t PERMISSION_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, PERMISSION_FAULTS));
+	dev_err(adev->dev, "\t MAPPING_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, MAPPING_ERROR));
+	dev_err(adev->dev, "\t RW: 0x%x\n", rw);
+}
+
+static void mmhub_v2_0_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid,
+				uint64_t page_table_base)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+
+	WREG32_SOC15_OFFSET_RLC(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+			    hub->ctx_addr_distance * vmid,
+			    lower_32_bits(page_table_base));
+
+	WREG32_SOC15_OFFSET_RLC(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+			    hub->ctx_addr_distance * vmid,
+			    upper_32_bits(page_table_base));
+}
+
+static void mmhub_v2_0_init_gart_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t pt_base = amdgpu_gmc_pd_addr(adev->gart.bo);
+
+	mmhub_v2_0_setup_vm_pt_regs(adev, 0, pt_base);
+
+	WREG32_SOC15_RLC(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+		     (u32)(adev->gmc.gart_start >> 12));
+	WREG32_SOC15_RLC(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+		     (u32)(adev->gmc.gart_start >> 44));
+
+	WREG32_SOC15_RLC(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+		     (u32)(adev->gmc.gart_end >> 12));
+	WREG32_SOC15_RLC(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+		     (u32)(adev->gmc.gart_end >> 44));
+}
+
+static void mmhub_v2_0_init_system_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t value;
+	uint32_t tmp;
+
+	if (!amdgpu_sriov_vf(adev)) {
+		/* Program the AGP BAR */
+		WREG32_SOC15_RLC(MMHUB, 0, mmMMMC_VM_AGP_BASE, 0);
+		WREG32_SOC15_RLC(MMHUB, 0, mmMMMC_VM_AGP_BOT, adev->gmc.agp_start >> 24);
+		WREG32_SOC15_RLC(MMHUB, 0, mmMMMC_VM_AGP_TOP, adev->gmc.agp_end >> 24);
+
+		/* Program the system aperture low logical page number. */
+		WREG32_SOC15(MMHUB, 0, mmMMMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+			     min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
+		WREG32_SOC15(MMHUB, 0, mmMMMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+			     max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
+	}
+
+	/* Set default page address. */
+	value = amdgpu_gmc_vram_mc2pa(adev, adev->mem_scratch.gpu_addr);
+	WREG32_SOC15(MMHUB, 0, mmMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
+		     (u32)(value >> 12));
+	WREG32_SOC15(MMHUB, 0, mmMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
+		     (u32)(value >> 44));
+
+	/* Program "protection fault". */
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
+		     (u32)(adev->dummy_page_addr >> 12));
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+		     (u32)((u64)adev->dummy_page_addr >> 44));
+
+	tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_CNTL2);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL2,
+			    ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_CNTL2, tmp);
+}
+
+static void mmhub_v2_0_init_tlb_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(MMHUB, 0, mmMMMC_VM_MX_L1_TLB_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 1);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    MTYPE, MTYPE_UC); /* UC, uncached */
+
+	WREG32_SOC15(MMHUB, 0, mmMMMC_VM_MX_L1_TLB_CNTL, tmp);
+}
+
+static void mmhub_v2_0_init_cache_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_CACHE, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL,
+			    ENABLE_DEFAULT_PAGE_OUT_TO_SYSTEM_MEMORY, 1);
+	/* XXX for emulation, Refer to closed source code.*/
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, L2_PDE0_CACHE_TAG_GENERATION_MODE,
+			    0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL, tmp);
+
+	tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL2);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL2, tmp);
+
+	tmp = mmMMVM_L2_CNTL3_DEFAULT;
+	if (adev->gmc.translate_further) {
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3, BANK_SELECT, 12);
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+	} else {
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3, BANK_SELECT, 9);
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+	}
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL3, tmp);
+
+	tmp = mmMMVM_L2_CNTL4_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL4, tmp);
+
+	tmp = mmMMVM_L2_CNTL5_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL5, L2_CACHE_SMALLK_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL5, tmp);
+}
+
+static void mmhub_v2_0_enable_system_domain(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_CONTEXT0_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL,
+			    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
+	WREG32_SOC15_RLC(MMHUB, 0, mmMMVM_CONTEXT0_CNTL, tmp);
+}
+
+static void mmhub_v2_0_disable_identity_aperture(struct amdgpu_device *adev)
+{
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	WREG32_SOC15(MMHUB, 0,
+		     mmMMVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32,
+		     0xFFFFFFFF);
+	WREG32_SOC15(MMHUB, 0,
+		     mmMMVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32,
+		     0x0000000F);
+
+	WREG32_SOC15(MMHUB, 0,
+		     mmMMVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32, 0);
+	WREG32_SOC15(MMHUB, 0,
+		     mmMMVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32, 0);
+
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32,
+		     0);
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32,
+		     0);
+}
+
+static void mmhub_v2_0_setup_vmid_config(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	int i;
+	uint32_t tmp;
+
+	for (i = 0; i <= 14; i++) {
+		tmp = RREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_CNTL, i * hub->ctx_distance);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH,
+				    adev->vm_manager.num_level);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    PAGE_TABLE_BLOCK_SIZE,
+				    adev->vm_manager.block_size - 9);
+		/* Send no-retry XNACK on fault to suppress VM fault storm. */
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
+				    !adev->gmc.noretry);
+		WREG32_SOC15_OFFSET_RLC(MMHUB, 0, mmMMVM_CONTEXT1_CNTL,
+				    i * hub->ctx_distance, tmp);
+		WREG32_SOC15_OFFSET_RLC(MMHUB, 0, mmMMVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET_RLC(MMHUB, 0, mmMMVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET_RLC(MMHUB, 0, mmMMVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32,
+				    i * hub->ctx_addr_distance,
+				    lower_32_bits(adev->vm_manager.max_pfn - 1));
+		WREG32_SOC15_OFFSET_RLC(MMHUB, 0, mmMMVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32,
+				    i * hub->ctx_addr_distance,
+				    upper_32_bits(adev->vm_manager.max_pfn - 1));
+	}
+
+	hub->vm_cntx_cntl = tmp;
+}
+
+static void mmhub_v2_0_program_invalidation(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	unsigned i;
+
+	for (i = 0; i < 18; ++i) {
+		WREG32_SOC15_OFFSET_RLC(MMHUB, 0, mmMMVM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
+				    i * hub->eng_addr_distance, 0xffffffff);
+		WREG32_SOC15_OFFSET_RLC(MMHUB, 0, mmMMVM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
+				    i * hub->eng_addr_distance, 0x1f);
+	}
+}
+
+static int mmhub_v2_0_gart_enable(struct amdgpu_device *adev)
+{
+	/* GART Enable. */
+	mmhub_v2_0_init_gart_aperture_regs(adev);
+	mmhub_v2_0_init_system_aperture_regs(adev);
+	mmhub_v2_0_init_tlb_regs(adev);
+	mmhub_v2_0_init_cache_regs(adev);
+
+	mmhub_v2_0_enable_system_domain(adev);
+	mmhub_v2_0_disable_identity_aperture(adev);
+	mmhub_v2_0_setup_vmid_config(adev);
+	mmhub_v2_0_program_invalidation(adev);
+
+	return 0;
+}
+
+static void mmhub_v2_0_gart_disable(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	u32 tmp;
+	u32 i;
+
+	/* Disable all tables */
+	for (i = 0; i < AMDGPU_NUM_VMID; i++)
+		WREG32_SOC15_OFFSET_RLC(MMHUB, 0, mmMMVM_CONTEXT0_CNTL,
+				    i * hub->ctx_distance, 0);
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(MMHUB, 0, mmMMMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 0);
+	WREG32_SOC15(MMHUB, 0, mmMMMC_VM_MX_L1_TLB_CNTL, tmp);
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_CACHE, 0);
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL, tmp);
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL3, 0);
+}
+
+/**
+ * mmhub_v2_0_set_fault_enable_default - update GART/VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+static void mmhub_v2_0_set_fault_enable_default(struct amdgpu_device *adev, bool value)
+{
+	u32 tmp;
+
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE1_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE2_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT,
+			    value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    NACK_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	if (!value) {
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_NO_RETRY_FAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_RETRY_FAULT, 1);
+	}
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_CNTL, tmp);
+}
+
+static const struct amdgpu_vmhub_funcs mmhub_v2_0_vmhub_funcs = {
+	.print_l2_protection_fault_status = mmhub_v2_0_print_l2_protection_fault_status,
+	.get_invalidate_req = mmhub_v2_0_get_invalidate_req,
+};
+
+static void mmhub_v2_0_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+
+	hub->ctx0_ptb_addr_lo32 =
+		SOC15_REG_OFFSET(MMHUB, 0,
+				 mmMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32);
+	hub->ctx0_ptb_addr_hi32 =
+		SOC15_REG_OFFSET(MMHUB, 0,
+				 mmMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32);
+	hub->vm_inv_eng0_sem =
+		SOC15_REG_OFFSET(MMHUB, 0, mmMMVM_INVALIDATE_ENG0_SEM);
+	hub->vm_inv_eng0_req =
+		SOC15_REG_OFFSET(MMHUB, 0, mmMMVM_INVALIDATE_ENG0_REQ);
+	hub->vm_inv_eng0_ack =
+		SOC15_REG_OFFSET(MMHUB, 0, mmMMVM_INVALIDATE_ENG0_ACK);
+	hub->vm_context0_cntl =
+		SOC15_REG_OFFSET(MMHUB, 0, mmMMVM_CONTEXT0_CNTL);
+	hub->vm_l2_pro_fault_status =
+		SOC15_REG_OFFSET(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_STATUS);
+	hub->vm_l2_pro_fault_cntl =
+		SOC15_REG_OFFSET(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_CNTL);
+
+	hub->ctx_distance = mmMMVM_CONTEXT1_CNTL - mmMMVM_CONTEXT0_CNTL;
+	hub->ctx_addr_distance = mmMMVM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 -
+		mmMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32;
+	hub->eng_distance = mmMMVM_INVALIDATE_ENG1_REQ -
+		mmMMVM_INVALIDATE_ENG0_REQ;
+	hub->eng_addr_distance = mmMMVM_INVALIDATE_ENG1_ADDR_RANGE_LO32 -
+		mmMMVM_INVALIDATE_ENG0_ADDR_RANGE_LO32;
+
+	hub->vm_cntx_cntl_vm_fault = MMVM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__EXECUTE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK;
+
+	hub->vmhub_funcs = &mmhub_v2_0_vmhub_funcs;
+}
+
+static void mmhub_v2_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+							bool enable)
+{
+	uint32_t def, data, def1, data1;
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG))
+		return;
+
+	switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) {
+	case IP_VERSION(2, 1, 0):
+	case IP_VERSION(2, 1, 1):
+	case IP_VERSION(2, 1, 2):
+		def1 = data1 = RREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2_Sienna_Cichlid);
+		break;
+	default:
+		def  = data  = RREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG);
+		def1 = data1 = RREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2);
+		break;
+	}
+
+	if (enable) {
+		data |= MM_ATC_L2_MISC_CG__ENABLE_MASK;
+
+		data1 &= ~(DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+		           DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+		           DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+		           DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+		           DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+		           DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK);
+
+	} else {
+		data &= ~MM_ATC_L2_MISC_CG__ENABLE_MASK;
+
+		data1 |= (DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK);
+	}
+
+	switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) {
+	case IP_VERSION(2, 1, 0):
+	case IP_VERSION(2, 1, 1):
+	case IP_VERSION(2, 1, 2):
+		if (def1 != data1)
+			WREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2_Sienna_Cichlid, data1);
+		break;
+	default:
+		if (def != data)
+			WREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG, data);
+		if (def1 != data1)
+			WREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2, data1);
+		break;
+	}
+}
+
+static void mmhub_v2_0_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+						       bool enable)
+{
+	uint32_t def, data;
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_MC_LS))
+		return;
+
+	switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) {
+	case IP_VERSION(2, 1, 0):
+	case IP_VERSION(2, 1, 1):
+	case IP_VERSION(2, 1, 2):
+		/* There is no ATCL2 in MMHUB for 2.1.x */
+		return;
+	default:
+		def  = data  = RREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG);
+		break;
+	}
+
+	if (enable)
+		data |= MM_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK;
+	else
+		data &= ~MM_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK;
+
+	if (def != data)
+		WREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG, data);
+}
+
+static int mmhub_v2_0_set_clockgating(struct amdgpu_device *adev,
+			       enum amd_clockgating_state state)
+{
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) {
+	case IP_VERSION(2, 0, 0):
+	case IP_VERSION(2, 0, 2):
+	case IP_VERSION(2, 1, 0):
+	case IP_VERSION(2, 1, 1):
+	case IP_VERSION(2, 1, 2):
+		mmhub_v2_0_update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		mmhub_v2_0_update_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static void mmhub_v2_0_get_clockgating(struct amdgpu_device *adev, u64 *flags)
+{
+	u32 data, data1;
+
+	if (amdgpu_sriov_vf(adev))
+		*flags = 0;
+
+	switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) {
+	case IP_VERSION(2, 1, 0):
+	case IP_VERSION(2, 1, 1):
+	case IP_VERSION(2, 1, 2):
+		/* There is no ATCL2 in MMHUB for 2.1.x. Keep the status
+		 * based on DAGB
+		 */
+		data = MM_ATC_L2_MISC_CG__ENABLE_MASK;
+		data1 = RREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2_Sienna_Cichlid);
+		break;
+	default:
+		data  = RREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG);
+		data1 = RREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2);
+		break;
+	}
+
+	/* AMD_CG_SUPPORT_MC_MGCG */
+	if ((data & MM_ATC_L2_MISC_CG__ENABLE_MASK) &&
+	    !(data1 & (DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+		       DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+		       DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+		       DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+		       DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+		       DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK)))
+		*flags |= AMD_CG_SUPPORT_MC_MGCG;
+
+	/* AMD_CG_SUPPORT_MC_LS */
+	if (data & MM_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_MC_LS;
+}
+
+const struct amdgpu_mmhub_funcs mmhub_v2_0_funcs = {
+	.init = mmhub_v2_0_init,
+	.gart_enable = mmhub_v2_0_gart_enable,
+	.set_fault_enable_default = mmhub_v2_0_set_fault_enable_default,
+	.gart_disable = mmhub_v2_0_gart_disable,
+	.set_clockgating = mmhub_v2_0_set_clockgating,
+	.get_clockgating = mmhub_v2_0_get_clockgating,
+	.setup_vm_pt_regs = mmhub_v2_0_setup_vm_pt_regs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.h
new file mode 100644
index 000000000000..f80f461d67da
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.h
@@ -0,0 +1,28 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __MMHUB_V2_0_H__
+#define __MMHUB_V2_0_H__
+
+extern const struct amdgpu_mmhub_funcs mmhub_v2_0_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c
new file mode 100644
index 000000000000..5eb8122e2746
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c
@@ -0,0 +1,629 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "mmhub_v2_3.h"
+
+#include "mmhub/mmhub_2_3_0_offset.h"
+#include "mmhub/mmhub_2_3_0_sh_mask.h"
+#include "mmhub/mmhub_2_3_0_default.h"
+#include "navi10_enum.h"
+
+#include "soc15_common.h"
+
+static const char *mmhub_client_ids_vangogh[][2] = {
+	[0][0] = "MP0",
+	[1][0] = "MP1",
+	[2][0] = "DCEDMC",
+	[3][0] = "DCEVGA",
+	[13][0] = "UTCL2",
+	[26][0] = "OSS",
+	[27][0] = "HDP",
+	[28][0] = "VCN",
+	[29][0] = "VCNU",
+	[30][0] = "JPEG",
+	[0][1] = "MP0",
+	[1][1] = "MP1",
+	[2][1] = "DCEDMC",
+	[3][1] = "DCEVGA",
+	[4][1] = "DCEDWB",
+	[5][1] = "XDP",
+	[26][1] = "OSS",
+	[27][1] = "HDP",
+	[28][1] = "VCN",
+	[29][1] = "VCNU",
+	[30][1] = "JPEG",
+};
+
+static uint32_t mmhub_v2_3_get_invalidate_req(unsigned int vmid,
+					      uint32_t flush_type)
+{
+	u32 req = 0;
+
+	/* invalidate using legacy mode on vmid*/
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ,
+			    PER_VMID_INVALIDATE_REQ, 1 << vmid);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, FLUSH_TYPE, flush_type);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PTES, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE0, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE1, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE2, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L1_PTES, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ,
+			    CLEAR_PROTECTION_FAULT_STATUS_ADDR,	0);
+
+	return req;
+}
+
+static void
+mmhub_v2_3_print_l2_protection_fault_status(struct amdgpu_device *adev,
+					     uint32_t status)
+{
+	uint32_t cid, rw;
+	const char *mmhub_cid = NULL;
+
+	cid = REG_GET_FIELD(status,
+			    MMVM_L2_PROTECTION_FAULT_STATUS, CID);
+	rw = REG_GET_FIELD(status,
+			   MMVM_L2_PROTECTION_FAULT_STATUS, RW);
+
+	dev_err(adev->dev,
+		"MMVM_L2_PROTECTION_FAULT_STATUS:0x%08X\n",
+		status);
+	switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) {
+	case IP_VERSION(2, 3, 0):
+	case IP_VERSION(2, 4, 0):
+	case IP_VERSION(2, 4, 1):
+		mmhub_cid = mmhub_client_ids_vangogh[cid][rw];
+		break;
+	default:
+		mmhub_cid = NULL;
+		break;
+	}
+	dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n",
+		mmhub_cid ? mmhub_cid : "unknown", cid);
+	dev_err(adev->dev, "\t MORE_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, MORE_FAULTS));
+	dev_err(adev->dev, "\t WALKER_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, WALKER_ERROR));
+	dev_err(adev->dev, "\t PERMISSION_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, PERMISSION_FAULTS));
+	dev_err(adev->dev, "\t MAPPING_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, MAPPING_ERROR));
+	dev_err(adev->dev, "\t RW: 0x%x\n", rw);
+}
+
+static void mmhub_v2_3_setup_vm_pt_regs(struct amdgpu_device *adev,
+					uint32_t vmid,
+					uint64_t page_table_base)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+
+	WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+			    hub->ctx_addr_distance * vmid, lower_32_bits(page_table_base));
+
+	WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+			    hub->ctx_addr_distance * vmid, upper_32_bits(page_table_base));
+}
+
+static void mmhub_v2_3_init_gart_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t pt_base = amdgpu_gmc_pd_addr(adev->gart.bo);
+
+	mmhub_v2_3_setup_vm_pt_regs(adev, 0, pt_base);
+
+	WREG32_SOC15(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+		     (u32)(adev->gmc.gart_start >> 12));
+	WREG32_SOC15(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+		     (u32)(adev->gmc.gart_start >> 44));
+
+	WREG32_SOC15(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+		     (u32)(adev->gmc.gart_end >> 12));
+	WREG32_SOC15(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+		     (u32)(adev->gmc.gart_end >> 44));
+}
+
+static void mmhub_v2_3_init_system_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t value;
+	uint32_t tmp;
+
+	/* Disable AGP. */
+	WREG32_SOC15(MMHUB, 0, mmMMMC_VM_AGP_BASE, 0);
+	WREG32_SOC15(MMHUB, 0, mmMMMC_VM_AGP_BOT, adev->gmc.agp_start >> 24);
+	WREG32_SOC15(MMHUB, 0, mmMMMC_VM_AGP_TOP, adev->gmc.agp_end >> 24);
+
+	/* Program the system aperture low logical page number. */
+	WREG32_SOC15(MMHUB, 0, mmMMMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+		     min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
+	WREG32_SOC15(MMHUB, 0, mmMMMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+		     max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
+
+	/* Set default page address. */
+	value = amdgpu_gmc_vram_mc2pa(adev, adev->mem_scratch.gpu_addr);
+	WREG32_SOC15(MMHUB, 0, mmMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
+		     (u32)(value >> 12));
+	WREG32_SOC15(MMHUB, 0, mmMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
+		     (u32)(value >> 44));
+
+	/* Program "protection fault". */
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
+		     (u32)(adev->dummy_page_addr >> 12));
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+		     (u32)((u64)adev->dummy_page_addr >> 44));
+
+	tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_CNTL2);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL2,
+			    ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_CNTL2, tmp);
+}
+
+static void mmhub_v2_3_init_tlb_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(MMHUB, 0, mmMMMC_VM_MX_L1_TLB_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 1);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    MTYPE, MTYPE_UC); /* UC, uncached */
+
+	WREG32_SOC15(MMHUB, 0, mmMMMC_VM_MX_L1_TLB_CNTL, tmp);
+}
+
+static void mmhub_v2_3_init_cache_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_CACHE, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL,
+			    ENABLE_DEFAULT_PAGE_OUT_TO_SYSTEM_MEMORY, 1);
+	/* XXX for emulation, Refer to closed source code.*/
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, L2_PDE0_CACHE_TAG_GENERATION_MODE,
+			    0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL, tmp);
+
+	tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL2);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL2, tmp);
+
+	tmp = mmMMVM_L2_CNTL3_DEFAULT;
+	if (adev->gmc.translate_further) {
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3, BANK_SELECT, 12);
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+	} else {
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3, BANK_SELECT, 9);
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+	}
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL3, tmp);
+
+	tmp = mmMMVM_L2_CNTL4_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL4, tmp);
+
+	tmp = mmMMVM_L2_CNTL5_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL5, L2_CACHE_SMALLK_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL5, tmp);
+}
+
+static void mmhub_v2_3_enable_system_domain(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_CONTEXT0_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL,
+			    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
+	WREG32_SOC15(MMHUB, 0, mmMMVM_CONTEXT0_CNTL, tmp);
+}
+
+static void mmhub_v2_3_disable_identity_aperture(struct amdgpu_device *adev)
+{
+	WREG32_SOC15(MMHUB, 0,
+		     mmMMVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32,
+		     0xFFFFFFFF);
+	WREG32_SOC15(MMHUB, 0,
+		     mmMMVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32,
+		     0x0000000F);
+
+	WREG32_SOC15(MMHUB, 0,
+		     mmMMVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32, 0);
+	WREG32_SOC15(MMHUB, 0,
+		     mmMMVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32, 0);
+
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32,
+		     0);
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32,
+		     0);
+}
+
+static void mmhub_v2_3_setup_vmid_config(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	int i;
+	uint32_t tmp;
+
+	for (i = 0; i <= 14; i++) {
+		tmp = RREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_CNTL, i * hub->ctx_distance);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH,
+				    adev->vm_manager.num_level);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    PAGE_TABLE_BLOCK_SIZE,
+				    adev->vm_manager.block_size - 9);
+		/* Send no-retry XNACK on fault to suppress VM fault storm. */
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
+				    !adev->gmc.noretry);
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_CNTL,
+				    i * hub->ctx_distance, tmp);
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32,
+				    i * hub->ctx_addr_distance,
+				    lower_32_bits(adev->vm_manager.max_pfn - 1));
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32,
+				    i * hub->ctx_addr_distance,
+				    upper_32_bits(adev->vm_manager.max_pfn - 1));
+	}
+
+	hub->vm_cntx_cntl = tmp;
+}
+
+static void mmhub_v2_3_program_invalidation(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	unsigned int i;
+
+	for (i = 0; i < 18; ++i) {
+		WREG32_SOC15_OFFSET(MMHUB, 0,
+				    mmMMVM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
+				    i * hub->eng_addr_distance, 0xffffffff);
+		WREG32_SOC15_OFFSET(MMHUB, 0,
+				    mmMMVM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
+				    i * hub->eng_addr_distance, 0x1f);
+	}
+}
+
+static int mmhub_v2_3_gart_enable(struct amdgpu_device *adev)
+{
+	if (amdgpu_sriov_vf(adev)) {
+		/*
+		 * MMMC_VM_FB_LOCATION_BASE/TOP is NULL for VF, becuase they are
+		 * VF copy registers so vbios post doesn't program them, for
+		 * SRIOV driver need to program them
+		 */
+		WREG32_SOC15(MMHUB, 0, mmMMMC_VM_FB_LOCATION_BASE,
+			     adev->gmc.vram_start >> 24);
+		WREG32_SOC15(MMHUB, 0, mmMMMC_VM_FB_LOCATION_TOP,
+			     adev->gmc.vram_end >> 24);
+	}
+
+	/* GART Enable. */
+	mmhub_v2_3_init_gart_aperture_regs(adev);
+	mmhub_v2_3_init_system_aperture_regs(adev);
+	mmhub_v2_3_init_tlb_regs(adev);
+	mmhub_v2_3_init_cache_regs(adev);
+
+	mmhub_v2_3_enable_system_domain(adev);
+	mmhub_v2_3_disable_identity_aperture(adev);
+	mmhub_v2_3_setup_vmid_config(adev);
+	mmhub_v2_3_program_invalidation(adev);
+
+	return 0;
+}
+
+static void mmhub_v2_3_gart_disable(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	u32 tmp;
+	u32 i;
+
+	/* Disable all tables */
+	for (i = 0; i < AMDGPU_NUM_VMID; i++)
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT0_CNTL,
+				    i * hub->ctx_distance, 0);
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(MMHUB, 0, mmMMMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 0);
+	WREG32_SOC15(MMHUB, 0, mmMMMC_VM_MX_L1_TLB_CNTL, tmp);
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_CACHE, 0);
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL, tmp);
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL3, 0);
+}
+
+/**
+ * mmhub_v2_3_set_fault_enable_default - update GART/VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+static void mmhub_v2_3_set_fault_enable_default(struct amdgpu_device *adev,
+						bool value)
+{
+	u32 tmp;
+
+	tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE1_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE2_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT,
+			    value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    NACK_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	if (!value) {
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_NO_RETRY_FAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_RETRY_FAULT, 1);
+	}
+	WREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_CNTL, tmp);
+}
+
+static const struct amdgpu_vmhub_funcs mmhub_v2_3_vmhub_funcs = {
+	.print_l2_protection_fault_status = mmhub_v2_3_print_l2_protection_fault_status,
+	.get_invalidate_req = mmhub_v2_3_get_invalidate_req,
+};
+
+static void mmhub_v2_3_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+
+	hub->ctx0_ptb_addr_lo32 =
+		SOC15_REG_OFFSET(MMHUB, 0,
+				 mmMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32);
+	hub->ctx0_ptb_addr_hi32 =
+		SOC15_REG_OFFSET(MMHUB, 0,
+				 mmMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32);
+	hub->vm_inv_eng0_sem =
+		SOC15_REG_OFFSET(MMHUB, 0,
+				 mmMMVM_INVALIDATE_ENG0_SEM);
+	hub->vm_inv_eng0_req =
+		SOC15_REG_OFFSET(MMHUB, 0, mmMMVM_INVALIDATE_ENG0_REQ);
+	hub->vm_inv_eng0_ack =
+		SOC15_REG_OFFSET(MMHUB, 0, mmMMVM_INVALIDATE_ENG0_ACK);
+	hub->vm_context0_cntl =
+		SOC15_REG_OFFSET(MMHUB, 0, mmMMVM_CONTEXT0_CNTL);
+	hub->vm_l2_pro_fault_status =
+		SOC15_REG_OFFSET(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_STATUS);
+	hub->vm_l2_pro_fault_cntl =
+		SOC15_REG_OFFSET(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_CNTL);
+
+	hub->ctx_distance = mmMMVM_CONTEXT1_CNTL - mmMMVM_CONTEXT0_CNTL;
+	hub->ctx_addr_distance = mmMMVM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 -
+		mmMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32;
+	hub->eng_distance = mmMMVM_INVALIDATE_ENG1_REQ -
+		mmMMVM_INVALIDATE_ENG0_REQ;
+	hub->eng_addr_distance = mmMMVM_INVALIDATE_ENG1_ADDR_RANGE_LO32 -
+		mmMMVM_INVALIDATE_ENG0_ADDR_RANGE_LO32;
+
+	hub->vm_cntx_cntl_vm_fault = MMVM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__EXECUTE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK;
+
+	hub->vmhub_funcs = &mmhub_v2_3_vmhub_funcs;
+}
+
+static void
+mmhub_v2_3_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+					    bool enable)
+{
+	uint32_t def, data, def1, data1;
+
+	def  = data  = RREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_CGTT_CLK_CTRL);
+	def1 = data1 = RREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG)) {
+		data &= ~MM_ATC_L2_CGTT_CLK_CTRL__SOFT_OVERRIDE_MASK;
+		data1 &= ~(DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+			   DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+			   DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+			   DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+			   DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+			   DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK);
+
+	} else {
+		data |= MM_ATC_L2_CGTT_CLK_CTRL__SOFT_OVERRIDE_MASK;
+		data1 |= (DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK);
+	}
+
+	if (def != data)
+		WREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_CGTT_CLK_CTRL, data);
+	if (def1 != data1)
+		WREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2, data1);
+}
+
+static void
+mmhub_v2_3_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+					   bool enable)
+{
+	uint32_t def, data, def1, data1, def2, data2;
+
+	def  = data  = RREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_CGTT_CLK_CTRL);
+	def1 = data1 = RREG32_SOC15(MMHUB, 0, mmDAGB0_WR_CGTT_CLK_CTRL);
+	def2 = data2 = RREG32_SOC15(MMHUB, 0, mmDAGB0_RD_CGTT_CLK_CTRL);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_LS)) {
+		data &= ~MM_ATC_L2_CGTT_CLK_CTRL__MGLS_OVERRIDE_MASK;
+		data1 &= ~(DAGB0_WR_CGTT_CLK_CTRL__LS_OVERRIDE_MASK |
+			DAGB0_WR_CGTT_CLK_CTRL__LS_OVERRIDE_WRITE_MASK |
+			DAGB0_WR_CGTT_CLK_CTRL__LS_OVERRIDE_READ_MASK |
+			DAGB0_WR_CGTT_CLK_CTRL__LS_OVERRIDE_RETURN_MASK |
+			DAGB0_WR_CGTT_CLK_CTRL__LS_OVERRIDE_REGISTER_MASK);
+		data2 &= ~(DAGB0_RD_CGTT_CLK_CTRL__LS_OVERRIDE_MASK |
+			DAGB0_RD_CGTT_CLK_CTRL__LS_OVERRIDE_WRITE_MASK |
+			DAGB0_RD_CGTT_CLK_CTRL__LS_OVERRIDE_READ_MASK |
+			DAGB0_RD_CGTT_CLK_CTRL__LS_OVERRIDE_RETURN_MASK |
+			DAGB0_RD_CGTT_CLK_CTRL__LS_OVERRIDE_REGISTER_MASK);
+	} else {
+		data |= MM_ATC_L2_CGTT_CLK_CTRL__MGLS_OVERRIDE_MASK;
+		data1 |= (DAGB0_WR_CGTT_CLK_CTRL__LS_OVERRIDE_MASK |
+			DAGB0_WR_CGTT_CLK_CTRL__LS_OVERRIDE_WRITE_MASK |
+			DAGB0_WR_CGTT_CLK_CTRL__LS_OVERRIDE_READ_MASK |
+			DAGB0_WR_CGTT_CLK_CTRL__LS_OVERRIDE_RETURN_MASK |
+			DAGB0_WR_CGTT_CLK_CTRL__LS_OVERRIDE_REGISTER_MASK);
+		data2 |= (DAGB0_RD_CGTT_CLK_CTRL__LS_OVERRIDE_MASK |
+			DAGB0_RD_CGTT_CLK_CTRL__LS_OVERRIDE_WRITE_MASK |
+			DAGB0_RD_CGTT_CLK_CTRL__LS_OVERRIDE_READ_MASK |
+			DAGB0_RD_CGTT_CLK_CTRL__LS_OVERRIDE_RETURN_MASK |
+			DAGB0_RD_CGTT_CLK_CTRL__LS_OVERRIDE_REGISTER_MASK);
+	}
+
+	if (def != data)
+		WREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_CGTT_CLK_CTRL, data);
+	if (def1 != data1)
+		WREG32_SOC15(MMHUB, 0, mmDAGB0_WR_CGTT_CLK_CTRL, data1);
+	if (def2 != data2)
+		WREG32_SOC15(MMHUB, 0, mmDAGB0_RD_CGTT_CLK_CTRL, data2);
+}
+
+static int mmhub_v2_3_set_clockgating(struct amdgpu_device *adev,
+				      enum amd_clockgating_state state)
+{
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	mmhub_v2_3_update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+	mmhub_v2_3_update_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+
+	return 0;
+}
+
+static void mmhub_v2_3_get_clockgating(struct amdgpu_device *adev, u64 *flags)
+{
+	int data, data1, data2, data3;
+
+	if (amdgpu_sriov_vf(adev))
+		*flags = 0;
+
+	data = RREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2);
+	data1  = RREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_CGTT_CLK_CTRL);
+	data2 = RREG32_SOC15(MMHUB, 0, mmDAGB0_WR_CGTT_CLK_CTRL);
+	data3 = RREG32_SOC15(MMHUB, 0, mmDAGB0_RD_CGTT_CLK_CTRL);
+
+	/* AMD_CG_SUPPORT_MC_MGCG */
+	if (!(data & (DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+			DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+			DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+			DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+			DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+			DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK))
+		&& !(data1 & MM_ATC_L2_CGTT_CLK_CTRL__SOFT_OVERRIDE_MASK)) {
+		*flags |= AMD_CG_SUPPORT_MC_MGCG;
+	}
+
+	/* AMD_CG_SUPPORT_MC_LS */
+	if (!(data1 & MM_ATC_L2_CGTT_CLK_CTRL__MGLS_OVERRIDE_MASK)
+		&& !(data2 & (DAGB0_WR_CGTT_CLK_CTRL__LS_OVERRIDE_MASK |
+				DAGB0_WR_CGTT_CLK_CTRL__LS_OVERRIDE_WRITE_MASK |
+				DAGB0_WR_CGTT_CLK_CTRL__LS_OVERRIDE_READ_MASK |
+				DAGB0_WR_CGTT_CLK_CTRL__LS_OVERRIDE_RETURN_MASK |
+				DAGB0_WR_CGTT_CLK_CTRL__LS_OVERRIDE_REGISTER_MASK))
+		&& !(data3 & (DAGB0_RD_CGTT_CLK_CTRL__LS_OVERRIDE_MASK |
+				DAGB0_RD_CGTT_CLK_CTRL__LS_OVERRIDE_WRITE_MASK |
+				DAGB0_RD_CGTT_CLK_CTRL__LS_OVERRIDE_READ_MASK |
+				DAGB0_RD_CGTT_CLK_CTRL__LS_OVERRIDE_RETURN_MASK |
+				DAGB0_RD_CGTT_CLK_CTRL__LS_OVERRIDE_REGISTER_MASK)))
+		*flags |= AMD_CG_SUPPORT_MC_LS;
+}
+
+const struct amdgpu_mmhub_funcs mmhub_v2_3_funcs = {
+	.init = mmhub_v2_3_init,
+	.gart_enable = mmhub_v2_3_gart_enable,
+	.set_fault_enable_default = mmhub_v2_3_set_fault_enable_default,
+	.gart_disable = mmhub_v2_3_gart_disable,
+	.set_clockgating = mmhub_v2_3_set_clockgating,
+	.get_clockgating = mmhub_v2_3_get_clockgating,
+	.setup_vm_pt_regs = mmhub_v2_3_setup_vm_pt_regs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.h
new file mode 100644
index 000000000000..2926d21dea8b
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.h
@@ -0,0 +1,28 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __MMHUB_V2_3_H__
+#define __MMHUB_V2_3_H__
+
+extern const struct amdgpu_mmhub_funcs mmhub_v2_3_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0.c
new file mode 100644
index 000000000000..7d5242df58a5
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0.c
@@ -0,0 +1,664 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "mmhub_v3_0.h"
+
+#include "mmhub/mmhub_3_0_0_offset.h"
+#include "mmhub/mmhub_3_0_0_sh_mask.h"
+#include "navi10_enum.h"
+
+#include "soc15_common.h"
+
+#define regMMVM_L2_CNTL3_DEFAULT				0x80100007
+#define regMMVM_L2_CNTL4_DEFAULT				0x000000c1
+#define regMMVM_L2_CNTL5_DEFAULT				0x00003fe0
+
+static const char *mmhub_client_ids_v3_0_0[][2] = {
+	[0][0] = "VMC",
+	[4][0] = "DCEDMC",
+	[5][0] = "DCEVGA",
+	[6][0] = "MP0",
+	[7][0] = "MP1",
+	[8][0] = "MPIO",
+	[16][0] = "HDP",
+	[17][0] = "LSDMA",
+	[18][0] = "JPEG",
+	[19][0] = "VCNU0",
+	[21][0] = "VSCH",
+	[22][0] = "VCNU1",
+	[23][0] = "VCN1",
+	[32+20][0] = "VCN0",
+	[2][1] = "DBGUNBIO",
+	[3][1] = "DCEDWB",
+	[4][1] = "DCEDMC",
+	[5][1] = "DCEVGA",
+	[6][1] = "MP0",
+	[7][1] = "MP1",
+	[8][1] = "MPIO",
+	[10][1] = "DBGU0",
+	[11][1] = "DBGU1",
+	[12][1] = "DBGU2",
+	[13][1] = "DBGU3",
+	[14][1] = "XDP",
+	[15][1] = "OSSSYS",
+	[16][1] = "HDP",
+	[17][1] = "LSDMA",
+	[18][1] = "JPEG",
+	[19][1] = "VCNU0",
+	[20][1] = "VCN0",
+	[21][1] = "VSCH",
+	[22][1] = "VCNU1",
+	[23][1] = "VCN1",
+};
+
+static uint32_t mmhub_v3_0_get_invalidate_req(unsigned int vmid,
+					      uint32_t flush_type)
+{
+	u32 req = 0;
+
+	/* invalidate using legacy mode on vmid*/
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ,
+			    PER_VMID_INVALIDATE_REQ, 1 << vmid);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, FLUSH_TYPE, flush_type);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PTES, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE0, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE1, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE2, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L1_PTES, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ,
+			    CLEAR_PROTECTION_FAULT_STATUS_ADDR,	0);
+
+	return req;
+}
+
+static void
+mmhub_v3_0_print_l2_protection_fault_status(struct amdgpu_device *adev,
+					     uint32_t status)
+{
+	uint32_t cid, rw;
+	const char *mmhub_cid = NULL;
+
+	cid = REG_GET_FIELD(status,
+			    MMVM_L2_PROTECTION_FAULT_STATUS, CID);
+	rw = REG_GET_FIELD(status,
+			   MMVM_L2_PROTECTION_FAULT_STATUS, RW);
+
+	dev_err(adev->dev,
+		"MMVM_L2_PROTECTION_FAULT_STATUS:0x%08X\n",
+		status);
+	switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) {
+	case IP_VERSION(3, 0, 0):
+	case IP_VERSION(3, 0, 1):
+		mmhub_cid = mmhub_client_ids_v3_0_0[cid][rw];
+		break;
+	default:
+		mmhub_cid = NULL;
+		break;
+	}
+	dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n",
+		mmhub_cid ? mmhub_cid : "unknown", cid);
+	dev_err(adev->dev, "\t MORE_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, MORE_FAULTS));
+	dev_err(adev->dev, "\t WALKER_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, WALKER_ERROR));
+	dev_err(adev->dev, "\t PERMISSION_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, PERMISSION_FAULTS));
+	dev_err(adev->dev, "\t MAPPING_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, MAPPING_ERROR));
+	dev_err(adev->dev, "\t RW: 0x%x\n", rw);
+}
+
+static void mmhub_v3_0_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid,
+				uint64_t page_table_base)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+
+	WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+			    hub->ctx_addr_distance * vmid,
+			    lower_32_bits(page_table_base));
+
+	WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+			    hub->ctx_addr_distance * vmid,
+			    upper_32_bits(page_table_base));
+}
+
+static void mmhub_v3_0_init_gart_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t pt_base = amdgpu_gmc_pd_addr(adev->gart.bo);
+
+	mmhub_v3_0_setup_vm_pt_regs(adev, 0, pt_base);
+
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+		     (u32)(adev->gmc.gart_start >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+		     (u32)(adev->gmc.gart_start >> 44));
+
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+		     (u32)(adev->gmc.gart_end >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+		     (u32)(adev->gmc.gart_end >> 44));
+}
+
+static void mmhub_v3_0_init_system_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t value;
+	uint32_t tmp;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/*
+	 * the new L1 policy will block SRIOV guest from writing
+	 * these regs, and they will be programed at host.
+	 * so skip programing these regs.
+	 */
+	/* Program the AGP BAR */
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_AGP_BASE, 0);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_AGP_BOT, adev->gmc.agp_start >> 24);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_AGP_TOP, adev->gmc.agp_end >> 24);
+
+	/* Program the system aperture low logical page number. */
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+		     min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+		     max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
+
+	/* Set default page address. */
+	value = amdgpu_gmc_vram_mc2pa(adev, adev->mem_scratch.gpu_addr);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
+		     (u32)(value >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
+		     (u32)(value >> 44));
+
+	/* Program "protection fault". */
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
+		     (u32)(adev->dummy_page_addr >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+		     (u32)((u64)adev->dummy_page_addr >> 44));
+
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL2);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL2,
+			    ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL2, tmp);
+}
+
+static void mmhub_v3_0_init_tlb_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(MMHUB, 0, regMMMC_VM_MX_L1_TLB_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 1);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ECO_BITS, 0);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    MTYPE, MTYPE_UC); /* UC, uncached */
+
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_MX_L1_TLB_CNTL, tmp);
+}
+
+static void mmhub_v3_0_init_cache_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_CACHE, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL,
+			    ENABLE_DEFAULT_PAGE_OUT_TO_SYSTEM_MEMORY, 1);
+	/* XXX for emulation, Refer to closed source code.*/
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, L2_PDE0_CACHE_TAG_GENERATION_MODE,
+			    0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL, tmp);
+
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL2);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL2, tmp);
+
+	tmp = regMMVM_L2_CNTL3_DEFAULT;
+	if (adev->gmc.translate_further) {
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3, BANK_SELECT, 12);
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+	} else {
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3, BANK_SELECT, 9);
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+	}
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL3, tmp);
+
+	tmp = regMMVM_L2_CNTL4_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL4, tmp);
+
+	tmp = regMMVM_L2_CNTL5_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL5, L2_CACHE_SMALLK_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL5, tmp);
+}
+
+static void mmhub_v3_0_enable_system_domain(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL,
+			    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_CNTL, tmp);
+}
+
+static void mmhub_v3_0_disable_identity_aperture(struct amdgpu_device *adev)
+{
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	WREG32_SOC15(MMHUB, 0,
+		     regMMVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32,
+		     0xFFFFFFFF);
+	WREG32_SOC15(MMHUB, 0,
+		     regMMVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32,
+		     0x0000000F);
+
+	WREG32_SOC15(MMHUB, 0,
+		     regMMVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32, 0);
+	WREG32_SOC15(MMHUB, 0,
+		     regMMVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32, 0);
+
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32,
+		     0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32,
+		     0);
+}
+
+static void mmhub_v3_0_setup_vmid_config(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	int i;
+	uint32_t tmp;
+
+	for (i = 0; i <= 14; i++) {
+		tmp = RREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_CNTL, i * hub->ctx_distance);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH,
+				    adev->vm_manager.num_level);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    PAGE_TABLE_BLOCK_SIZE,
+				    adev->vm_manager.block_size - 9);
+		/* Send no-retry XNACK on fault to suppress VM fault storm. */
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
+				    !amdgpu_noretry);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_CNTL,
+				    i * hub->ctx_distance, tmp);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32,
+				    i * hub->ctx_addr_distance,
+				    lower_32_bits(adev->vm_manager.max_pfn - 1));
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32,
+				    i * hub->ctx_addr_distance,
+				    upper_32_bits(adev->vm_manager.max_pfn - 1));
+	}
+
+	hub->vm_cntx_cntl = tmp;
+}
+
+static void mmhub_v3_0_program_invalidation(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	unsigned i;
+
+	for (i = 0; i < 18; ++i) {
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
+				    i * hub->eng_addr_distance, 0xffffffff);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
+				    i * hub->eng_addr_distance, 0x1f);
+	}
+}
+
+static int mmhub_v3_0_gart_enable(struct amdgpu_device *adev)
+{
+	/* GART Enable. */
+	mmhub_v3_0_init_gart_aperture_regs(adev);
+	mmhub_v3_0_init_system_aperture_regs(adev);
+	mmhub_v3_0_init_tlb_regs(adev);
+	mmhub_v3_0_init_cache_regs(adev);
+
+	mmhub_v3_0_enable_system_domain(adev);
+	mmhub_v3_0_disable_identity_aperture(adev);
+	mmhub_v3_0_setup_vmid_config(adev);
+	mmhub_v3_0_program_invalidation(adev);
+
+	return 0;
+}
+
+static void mmhub_v3_0_gart_disable(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	u32 tmp;
+	u32 i;
+
+	/* Disable all tables */
+	for (i = 0; i < 16; i++)
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT0_CNTL,
+				    i * hub->ctx_distance, 0);
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(MMHUB, 0, regMMMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 0);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_MX_L1_TLB_CNTL, tmp);
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_CACHE, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL, tmp);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL3, 0);
+}
+
+/**
+ * mmhub_v3_0_set_fault_enable_default - update GART/VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+static void mmhub_v3_0_set_fault_enable_default(struct amdgpu_device *adev, bool value)
+{
+	u32 tmp;
+
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE1_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE2_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT,
+			    value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    NACK_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	if (!value) {
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_NO_RETRY_FAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_RETRY_FAULT, 1);
+	}
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL, tmp);
+}
+
+static const struct amdgpu_vmhub_funcs mmhub_v3_0_vmhub_funcs = {
+	.print_l2_protection_fault_status = mmhub_v3_0_print_l2_protection_fault_status,
+	.get_invalidate_req = mmhub_v3_0_get_invalidate_req,
+};
+
+static void mmhub_v3_0_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+
+	hub->ctx0_ptb_addr_lo32 =
+		SOC15_REG_OFFSET(MMHUB, 0,
+				 regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32);
+	hub->ctx0_ptb_addr_hi32 =
+		SOC15_REG_OFFSET(MMHUB, 0,
+				 regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32);
+	hub->vm_inv_eng0_sem =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_SEM);
+	hub->vm_inv_eng0_req =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_REQ);
+	hub->vm_inv_eng0_ack =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_ACK);
+	hub->vm_context0_cntl =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_CONTEXT0_CNTL);
+	hub->vm_l2_pro_fault_status =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_STATUS);
+	hub->vm_l2_pro_fault_cntl =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL);
+
+	hub->ctx_distance = regMMVM_CONTEXT1_CNTL - regMMVM_CONTEXT0_CNTL;
+	hub->ctx_addr_distance = regMMVM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 -
+		regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32;
+	hub->eng_distance = regMMVM_INVALIDATE_ENG1_REQ -
+		regMMVM_INVALIDATE_ENG0_REQ;
+	hub->eng_addr_distance = regMMVM_INVALIDATE_ENG1_ADDR_RANGE_LO32 -
+		regMMVM_INVALIDATE_ENG0_ADDR_RANGE_LO32;
+
+	hub->vm_cntx_cntl_vm_fault = MMVM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__EXECUTE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK;
+
+	hub->vm_l2_bank_select_reserved_cid2 =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_L2_BANK_SELECT_RESERVED_CID2);
+
+	hub->vm_contexts_disable =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_CONTEXTS_DISABLE);
+
+	hub->vmhub_funcs = &mmhub_v3_0_vmhub_funcs;
+}
+
+static u64 mmhub_v3_0_get_fb_location(struct amdgpu_device *adev)
+{
+	u64 base;
+
+	base = RREG32_SOC15(MMHUB, 0, regMMMC_VM_FB_LOCATION_BASE);
+
+	base &= MMMC_VM_FB_LOCATION_BASE__FB_BASE_MASK;
+	base <<= 24;
+
+	return base;
+}
+
+static u64 mmhub_v3_0_get_mc_fb_offset(struct amdgpu_device *adev)
+{
+	return (u64)RREG32_SOC15(MMHUB, 0, regMMMC_VM_FB_OFFSET) << 24;
+}
+
+static void mmhub_v3_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+							bool enable)
+{
+	uint32_t def, data;
+#if 0
+	uint32_t def1, data1, def2 = 0, data2 = 0;
+#endif
+
+	def  = data  = RREG32_SOC15(MMHUB, 0, regMM_ATC_L2_MISC_CG);
+#if 0
+	def1 = data1 = RREG32_SOC15(MMHUB, 0, regDAGB0_CNTL_MISC2);
+	def2 = data2 = RREG32_SOC15(MMHUB, 0, regDAGB1_CNTL_MISC2);
+#endif
+
+	if (enable) {
+		data |= MM_ATC_L2_MISC_CG__ENABLE_MASK;
+#if 0
+		data1 &= ~(DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+		           DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+		           DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+		           DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+		           DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+		           DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK);
+
+		data2 &= ~(DAGB1_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+		           DAGB1_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+		           DAGB1_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+		           DAGB1_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+		           DAGB1_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+		           DAGB1_CNTL_MISC2__DISABLE_TLBRD_CG_MASK);
+#endif
+	} else {
+		data &= ~MM_ATC_L2_MISC_CG__ENABLE_MASK;
+#if 0
+		data1 |= (DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK);
+
+		data2 |= (DAGB1_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+		          DAGB1_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+		          DAGB1_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+		          DAGB1_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+		          DAGB1_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+		          DAGB1_CNTL_MISC2__DISABLE_TLBRD_CG_MASK);
+#endif
+	}
+
+	if (def != data)
+		WREG32_SOC15(MMHUB, 0, regMM_ATC_L2_MISC_CG, data);
+#if 0
+	if (def1 != data1)
+		WREG32_SOC15(MMHUB, 0, regDAGB0_CNTL_MISC2, data1);
+
+	if (def2 != data2)
+		WREG32_SOC15(MMHUB, 0, regDAGB1_CNTL_MISC2, data2);
+#endif
+}
+
+static void mmhub_v3_0_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+						       bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_SOC15(MMHUB, 0, regMM_ATC_L2_MISC_CG);
+
+	if (enable)
+		data |= MM_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK;
+	else
+		data &= ~MM_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK;
+
+	if (def != data)
+		WREG32_SOC15(MMHUB, 0, regMM_ATC_L2_MISC_CG, data);
+}
+
+static int mmhub_v3_0_set_clockgating(struct amdgpu_device *adev,
+			       enum amd_clockgating_state state)
+{
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG)
+		mmhub_v3_0_update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_MC_LS)
+		mmhub_v3_0_update_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+
+	return 0;
+}
+
+static void mmhub_v3_0_get_clockgating(struct amdgpu_device *adev, u64 *flags)
+{
+	int data;
+
+	if (amdgpu_sriov_vf(adev))
+		*flags = 0;
+
+	data = RREG32_SOC15(MMHUB, 0, regMM_ATC_L2_MISC_CG);
+
+	/* AMD_CG_SUPPORT_MC_MGCG */
+	if (data & MM_ATC_L2_MISC_CG__ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_MC_MGCG;
+
+	/* AMD_CG_SUPPORT_MC_LS */
+	if (data & MM_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_MC_LS;
+}
+
+const struct amdgpu_mmhub_funcs mmhub_v3_0_funcs = {
+	.init = mmhub_v3_0_init,
+	.get_fb_location = mmhub_v3_0_get_fb_location,
+	.get_mc_fb_offset = mmhub_v3_0_get_mc_fb_offset,
+	.gart_enable = mmhub_v3_0_gart_enable,
+	.set_fault_enable_default = mmhub_v3_0_set_fault_enable_default,
+	.gart_disable = mmhub_v3_0_gart_disable,
+	.set_clockgating = mmhub_v3_0_set_clockgating,
+	.get_clockgating = mmhub_v3_0_get_clockgating,
+	.setup_vm_pt_regs = mmhub_v3_0_setup_vm_pt_regs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0.h
new file mode 100644
index 000000000000..3ced20f350bb
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0.h
@@ -0,0 +1,28 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __MMHUB_V3_0_H__
+#define __MMHUB_V3_0_H__
+
+extern const struct amdgpu_mmhub_funcs mmhub_v3_0_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0_1.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0_1.c
new file mode 100644
index 000000000000..910337dc28d1
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0_1.c
@@ -0,0 +1,597 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "mmhub_v3_0_1.h"
+
+#include "mmhub/mmhub_3_0_1_offset.h"
+#include "mmhub/mmhub_3_0_1_sh_mask.h"
+#include "navi10_enum.h"
+
+#include "soc15_common.h"
+
+#define regMMVM_L2_CNTL3_DEFAULT				0x80100007
+#define regMMVM_L2_CNTL4_DEFAULT				0x000000c1
+#define regMMVM_L2_CNTL5_DEFAULT				0x00003fe0
+
+static const char *mmhub_client_ids_v3_0_1[][2] = {
+	[0][0] = "VMC",
+	[1][0] = "ISPXT",
+	[2][0] = "ISPIXT",
+	[4][0] = "DCEDMC",
+	[5][0] = "DCEVGA",
+	[6][0] = "MP0",
+	[7][0] = "MP1",
+	[8][0] = "MPM",
+	[12][0] = "ISPTNR",
+	[14][0] = "ISPCRD0",
+	[15][0] = "ISPCRD1",
+	[16][0] = "ISPCRD2",
+	[22][0] = "HDP",
+	[23][0] = "LSDMA",
+	[24][0] = "JPEG",
+	[27][0] = "VSCH",
+	[28][0] = "VCNU",
+	[29][0] = "VCN",
+	[1][1] = "ISPXT",
+	[2][1] = "ISPIXT",
+	[3][1] = "DCEDWB",
+	[4][1] = "DCEDMC",
+	[5][1] = "DCEVGA",
+	[6][1] = "MP0",
+	[7][1] = "MP1",
+	[8][1] = "MPM",
+	[10][1] = "ISPMWR0",
+	[11][1] = "ISPMWR1",
+	[12][1] = "ISPTNR",
+	[13][1] = "ISPSWR",
+	[14][1] = "ISPCWR0",
+	[15][1] = "ISPCWR1",
+	[16][1] = "ISPCWR2",
+	[17][1] = "ISPCWR3",
+	[18][1] = "XDP",
+	[21][1] = "OSSSYS",
+	[22][1] = "HDP",
+	[23][1] = "LSDMA",
+	[24][1] = "JPEG",
+	[27][1] = "VSCH",
+	[28][1] = "VCNU",
+	[29][1] = "VCN",
+};
+
+static uint32_t mmhub_v3_0_1_get_invalidate_req(unsigned int vmid,
+						uint32_t flush_type)
+{
+	u32 req = 0;
+
+	/* invalidate using legacy mode on vmid*/
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ,
+			    PER_VMID_INVALIDATE_REQ, 1 << vmid);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, FLUSH_TYPE, flush_type);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PTES, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE0, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE1, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE2, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L1_PTES, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ,
+			    CLEAR_PROTECTION_FAULT_STATUS_ADDR,	0);
+
+	return req;
+}
+
+static void
+mmhub_v3_0_1_print_l2_protection_fault_status(struct amdgpu_device *adev,
+					      uint32_t status)
+{
+	uint32_t cid, rw;
+	const char *mmhub_cid = NULL;
+
+	cid = REG_GET_FIELD(status,
+			    MMVM_L2_PROTECTION_FAULT_STATUS, CID);
+	rw = REG_GET_FIELD(status,
+			   MMVM_L2_PROTECTION_FAULT_STATUS, RW);
+
+	dev_err(adev->dev,
+		"MMVM_L2_PROTECTION_FAULT_STATUS:0x%08X\n",
+		status);
+
+	switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) {
+	case IP_VERSION(3, 0, 1):
+		mmhub_cid = mmhub_client_ids_v3_0_1[cid][rw];
+		break;
+	default:
+		mmhub_cid = NULL;
+		break;
+	}
+
+	dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n",
+		mmhub_cid ? mmhub_cid : "unknown", cid);
+	dev_err(adev->dev, "\t MORE_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, MORE_FAULTS));
+	dev_err(adev->dev, "\t WALKER_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, WALKER_ERROR));
+	dev_err(adev->dev, "\t PERMISSION_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, PERMISSION_FAULTS));
+	dev_err(adev->dev, "\t MAPPING_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, MAPPING_ERROR));
+	dev_err(adev->dev, "\t RW: 0x%x\n", rw);
+}
+
+static void mmhub_v3_0_1_setup_vm_pt_regs(struct amdgpu_device *adev,
+					  uint32_t vmid,
+					  uint64_t page_table_base)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+
+	WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+			    hub->ctx_addr_distance * vmid,
+			    lower_32_bits(page_table_base));
+
+	WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+			    hub->ctx_addr_distance * vmid,
+			    upper_32_bits(page_table_base));
+}
+
+static void mmhub_v3_0_1_init_gart_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t pt_base = amdgpu_gmc_pd_addr(adev->gart.bo);
+
+	mmhub_v3_0_1_setup_vm_pt_regs(adev, 0, pt_base);
+
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+		     (u32)(adev->gmc.gart_start >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+		     (u32)(adev->gmc.gart_start >> 44));
+
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+		     (u32)(adev->gmc.gart_end >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+		     (u32)(adev->gmc.gart_end >> 44));
+}
+
+static void mmhub_v3_0_1_init_system_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t value;
+	uint32_t tmp;
+
+	/* Program the AGP BAR */
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_AGP_BASE, 0);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_AGP_BOT, adev->gmc.agp_start >> 24);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_AGP_TOP, adev->gmc.agp_end >> 24);
+
+	/*
+	 * the new L1 policy will block SRIOV guest from writing
+	 * these regs, and they will be programed at host.
+	 * so skip programing these regs.
+	 */
+	/* Program the system aperture low logical page number. */
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+		     min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+		     max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
+
+	/* Set default page address. */
+	value = amdgpu_gmc_vram_mc2pa(adev, adev->mem_scratch.gpu_addr);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
+		     (u32)(value >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
+		     (u32)(value >> 44));
+
+	/* Program "protection fault". */
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
+		     (u32)(adev->dummy_page_addr >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+		     (u32)((u64)adev->dummy_page_addr >> 44));
+
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL2);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL2,
+			    ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL2, tmp);
+}
+
+static void mmhub_v3_0_1_init_tlb_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(MMHUB, 0, regMMMC_VM_MX_L1_TLB_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 1);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ECO_BITS, 0);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    MTYPE, MTYPE_UC); /* UC, uncached */
+
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_MX_L1_TLB_CNTL, tmp);
+}
+
+static void mmhub_v3_0_1_init_cache_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_CACHE, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL,
+			    ENABLE_DEFAULT_PAGE_OUT_TO_SYSTEM_MEMORY, 1);
+	/* XXX for emulation, Refer to closed source code.*/
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, L2_PDE0_CACHE_TAG_GENERATION_MODE,
+			    0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL, tmp);
+
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL2);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL2, tmp);
+
+	tmp = regMMVM_L2_CNTL3_DEFAULT;
+	if (adev->gmc.translate_further) {
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3, BANK_SELECT, 12);
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+	} else {
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3, BANK_SELECT, 9);
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+	}
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL3, tmp);
+
+	tmp = regMMVM_L2_CNTL4_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL4, tmp);
+
+	tmp = regMMVM_L2_CNTL5_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL5, L2_CACHE_SMALLK_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL5, tmp);
+}
+
+static void mmhub_v3_0_1_enable_system_domain(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL,
+			    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_CNTL, tmp);
+}
+
+static void mmhub_v3_0_1_disable_identity_aperture(struct amdgpu_device *adev)
+{
+	WREG32_SOC15(MMHUB, 0,
+		     regMMVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32,
+		     0xFFFFFFFF);
+	WREG32_SOC15(MMHUB, 0,
+		     regMMVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32,
+		     0x0000000F);
+
+	WREG32_SOC15(MMHUB, 0,
+		     regMMVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32, 0);
+	WREG32_SOC15(MMHUB, 0,
+		     regMMVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32, 0);
+
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32,
+		     0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32,
+		     0);
+}
+
+static void mmhub_v3_0_1_setup_vmid_config(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	int i;
+	uint32_t tmp;
+
+	for (i = 0; i <= 14; i++) {
+		tmp = RREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_CNTL, i * hub->ctx_distance);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH,
+				    adev->vm_manager.num_level);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    PAGE_TABLE_BLOCK_SIZE,
+				    adev->vm_manager.block_size - 9);
+		/* Send no-retry XNACK on fault to suppress VM fault storm. */
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
+				    !amdgpu_noretry);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_CNTL,
+				    i * hub->ctx_distance, tmp);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32,
+				    i * hub->ctx_addr_distance,
+				    lower_32_bits(adev->vm_manager.max_pfn - 1));
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32,
+				    i * hub->ctx_addr_distance,
+				    upper_32_bits(adev->vm_manager.max_pfn - 1));
+	}
+
+	hub->vm_cntx_cntl = tmp;
+}
+
+static void mmhub_v3_0_1_program_invalidation(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	unsigned i;
+
+	for (i = 0; i < 18; ++i) {
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
+				    i * hub->eng_addr_distance, 0xffffffff);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
+				    i * hub->eng_addr_distance, 0x1f);
+	}
+}
+
+static int mmhub_v3_0_1_gart_enable(struct amdgpu_device *adev)
+{
+	/* GART Enable. */
+	mmhub_v3_0_1_init_gart_aperture_regs(adev);
+	mmhub_v3_0_1_init_system_aperture_regs(adev);
+	mmhub_v3_0_1_init_tlb_regs(adev);
+	mmhub_v3_0_1_init_cache_regs(adev);
+
+	mmhub_v3_0_1_enable_system_domain(adev);
+	mmhub_v3_0_1_disable_identity_aperture(adev);
+	mmhub_v3_0_1_setup_vmid_config(adev);
+	mmhub_v3_0_1_program_invalidation(adev);
+
+	return 0;
+}
+
+static void mmhub_v3_0_1_gart_disable(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	u32 tmp;
+	u32 i;
+
+	/* Disable all tables */
+	for (i = 0; i < 16; i++)
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT0_CNTL,
+				    i * hub->ctx_distance, 0);
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(MMHUB, 0, regMMMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 0);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_MX_L1_TLB_CNTL, tmp);
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_CACHE, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL, tmp);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL3, 0);
+}
+
+/**
+ * mmhub_v3_0_1_set_fault_enable_default - update GART/VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+static void mmhub_v3_0_1_set_fault_enable_default(struct amdgpu_device *adev,
+						  bool value)
+{
+	u32 tmp;
+
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE1_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE2_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT,
+			    value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    NACK_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	if (!value) {
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_NO_RETRY_FAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_RETRY_FAULT, 1);
+	}
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL, tmp);
+}
+
+static const struct amdgpu_vmhub_funcs mmhub_v3_0_1_vmhub_funcs = {
+	.print_l2_protection_fault_status = mmhub_v3_0_1_print_l2_protection_fault_status,
+	.get_invalidate_req = mmhub_v3_0_1_get_invalidate_req,
+};
+
+static void mmhub_v3_0_1_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+
+	hub->ctx0_ptb_addr_lo32 =
+		SOC15_REG_OFFSET(MMHUB, 0,
+				 regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32);
+	hub->ctx0_ptb_addr_hi32 =
+		SOC15_REG_OFFSET(MMHUB, 0,
+				 regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32);
+	hub->vm_inv_eng0_sem =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_SEM);
+	hub->vm_inv_eng0_req =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_REQ);
+	hub->vm_inv_eng0_ack =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_ACK);
+	hub->vm_context0_cntl =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_CONTEXT0_CNTL);
+	hub->vm_l2_pro_fault_status =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_STATUS);
+	hub->vm_l2_pro_fault_cntl =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL);
+
+	hub->ctx_distance = regMMVM_CONTEXT1_CNTL - regMMVM_CONTEXT0_CNTL;
+	hub->ctx_addr_distance = regMMVM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 -
+		regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32;
+	hub->eng_distance = regMMVM_INVALIDATE_ENG1_REQ -
+		regMMVM_INVALIDATE_ENG0_REQ;
+	hub->eng_addr_distance = regMMVM_INVALIDATE_ENG1_ADDR_RANGE_LO32 -
+		regMMVM_INVALIDATE_ENG0_ADDR_RANGE_LO32;
+
+	hub->vm_cntx_cntl_vm_fault = MMVM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__EXECUTE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK;
+
+	hub->vmhub_funcs = &mmhub_v3_0_1_vmhub_funcs;
+}
+
+static u64 mmhub_v3_0_1_get_fb_location(struct amdgpu_device *adev)
+{
+	u64 base;
+
+	base = RREG32_SOC15(MMHUB, 0, regMMMC_VM_FB_LOCATION_BASE);
+	base &= MMMC_VM_FB_LOCATION_BASE__FB_BASE_MASK;
+	base <<= 24;
+
+	return base;
+}
+
+static u64 mmhub_v3_0_1_get_mc_fb_offset(struct amdgpu_device *adev)
+{
+	return (u64)RREG32_SOC15(MMHUB, 0, regMMMC_VM_FB_OFFSET) << 24;
+}
+
+static void mmhub_v3_0_1_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+							  bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_SOC15(MMHUB, 0, regMM_ATC_L2_MISC_CG);
+
+	if (enable)
+		data |= MM_ATC_L2_MISC_CG__ENABLE_MASK;
+	else
+		data &= ~MM_ATC_L2_MISC_CG__ENABLE_MASK;
+
+	if (def != data)
+		WREG32_SOC15(MMHUB, 0, regMM_ATC_L2_MISC_CG, data);
+}
+
+static void mmhub_v3_0_1_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+							 bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_SOC15(MMHUB, 0, regMM_ATC_L2_MISC_CG);
+
+	if (enable)
+		data |= MM_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK;
+	else
+		data &= ~MM_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK;
+
+	if (def != data)
+		WREG32_SOC15(MMHUB, 0, regMM_ATC_L2_MISC_CG, data);
+}
+
+static int mmhub_v3_0_1_set_clockgating(struct amdgpu_device *adev,
+					enum amd_clockgating_state state)
+{
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	mmhub_v3_0_1_update_medium_grain_clock_gating(adev,
+			state == AMD_CG_STATE_GATE);
+	mmhub_v3_0_1_update_medium_grain_light_sleep(adev,
+			state == AMD_CG_STATE_GATE);
+	return 0;
+}
+
+static void mmhub_v3_0_1_get_clockgating(struct amdgpu_device *adev, u64 *flags)
+{
+	int data;
+
+	if (amdgpu_sriov_vf(adev))
+		*flags = 0;
+
+	data = RREG32_SOC15(MMHUB, 0, regMM_ATC_L2_MISC_CG);
+
+	/* AMD_CG_SUPPORT_MC_MGCG */
+	if (data & MM_ATC_L2_MISC_CG__ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_MC_MGCG;
+
+	/* AMD_CG_SUPPORT_MC_LS */
+	if (data & MM_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_MC_LS;
+}
+
+const struct amdgpu_mmhub_funcs mmhub_v3_0_1_funcs = {
+	.init = mmhub_v3_0_1_init,
+	.get_fb_location = mmhub_v3_0_1_get_fb_location,
+	.get_mc_fb_offset = mmhub_v3_0_1_get_mc_fb_offset,
+	.gart_enable = mmhub_v3_0_1_gart_enable,
+	.set_fault_enable_default = mmhub_v3_0_1_set_fault_enable_default,
+	.gart_disable = mmhub_v3_0_1_gart_disable,
+	.set_clockgating = mmhub_v3_0_1_set_clockgating,
+	.get_clockgating = mmhub_v3_0_1_get_clockgating,
+	.setup_vm_pt_regs = mmhub_v3_0_1_setup_vm_pt_regs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0_1.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0_1.h
new file mode 100644
index 000000000000..4c1246735e7d
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0_1.h
@@ -0,0 +1,28 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __MMHUB_V3_0_1_H__
+#define __MMHUB_V3_0_1_H__
+
+extern const struct amdgpu_mmhub_funcs mmhub_v3_0_1_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0_2.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0_2.c
new file mode 100644
index 000000000000..f0f182f033b9
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0_2.c
@@ -0,0 +1,570 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "mmhub_v3_0_2.h"
+
+#include "mmhub/mmhub_3_0_2_offset.h"
+#include "mmhub/mmhub_3_0_2_sh_mask.h"
+#include "navi10_enum.h"
+
+#include "soc15_common.h"
+
+#define regMMVM_L2_CNTL3_DEFAULT				0x80100007
+#define regMMVM_L2_CNTL4_DEFAULT				0x000000c1
+#define regMMVM_L2_CNTL5_DEFAULT				0x00003fe0
+
+static const char *mmhub_client_ids_v3_0_2[][2] = {
+	[0][0] = "VMC",
+	[4][0] = "DCEDMC",
+	[5][0] = "DCEVGA",
+	[6][0] = "MP0",
+	[7][0] = "MP1",
+	[8][0] = "MPIO",
+	[16][0] = "HDP",
+	[17][0] = "LSDMA",
+	[18][0] = "JPEG",
+	[19][0] = "VCNU0",
+	[21][0] = "VSCH",
+	[22][0] = "VCNU1",
+	[23][0] = "VCN1",
+	[32+20][0] = "VCN0",
+	[2][1] = "DBGUNBIO",
+	[3][1] = "DCEDWB",
+	[4][1] = "DCEDMC",
+	[5][1] = "DCEVGA",
+	[6][1] = "MP0",
+	[7][1] = "MP1",
+	[8][1] = "MPIO",
+	[10][1] = "DBGU0",
+	[11][1] = "DBGU1",
+	[12][1] = "DBGU2",
+	[13][1] = "DBGU3",
+	[14][1] = "XDP",
+	[15][1] = "OSSSYS",
+	[16][1] = "HDP",
+	[17][1] = "LSDMA",
+	[18][1] = "JPEG",
+	[19][1] = "VCNU0",
+	[20][1] = "VCN0",
+	[21][1] = "VSCH",
+	[22][1] = "VCNU1",
+	[23][1] = "VCN1",
+};
+
+static uint32_t mmhub_v3_0_2_get_invalidate_req(unsigned int vmid,
+					      uint32_t flush_type)
+{
+	u32 req = 0;
+
+	/* invalidate using legacy mode on vmid*/
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ,
+			    PER_VMID_INVALIDATE_REQ, 1 << vmid);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, FLUSH_TYPE, flush_type);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PTES, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE0, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE1, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE2, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L1_PTES, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ,
+			    CLEAR_PROTECTION_FAULT_STATUS_ADDR,	0);
+
+	return req;
+}
+
+static void
+mmhub_v3_0_2_print_l2_protection_fault_status(struct amdgpu_device *adev,
+					     uint32_t status)
+{
+	uint32_t cid, rw;
+	const char *mmhub_cid = NULL;
+
+	cid = REG_GET_FIELD(status,
+			    MMVM_L2_PROTECTION_FAULT_STATUS, CID);
+	rw = REG_GET_FIELD(status,
+			   MMVM_L2_PROTECTION_FAULT_STATUS, RW);
+
+	dev_err(adev->dev,
+		"MMVM_L2_PROTECTION_FAULT_STATUS:0x%08X\n",
+		status);
+
+	mmhub_cid = mmhub_client_ids_v3_0_2[cid][rw];
+	dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n",
+		mmhub_cid ? mmhub_cid : "unknown", cid);
+	dev_err(adev->dev, "\t MORE_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, MORE_FAULTS));
+	dev_err(adev->dev, "\t WALKER_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, WALKER_ERROR));
+	dev_err(adev->dev, "\t PERMISSION_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, PERMISSION_FAULTS));
+	dev_err(adev->dev, "\t MAPPING_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, MAPPING_ERROR));
+	dev_err(adev->dev, "\t RW: 0x%x\n", rw);
+}
+
+static void mmhub_v3_0_2_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid,
+				uint64_t page_table_base)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+
+	WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+			    hub->ctx_addr_distance * vmid,
+			    lower_32_bits(page_table_base));
+
+	WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+			    hub->ctx_addr_distance * vmid,
+			    upper_32_bits(page_table_base));
+}
+
+static void mmhub_v3_0_2_init_gart_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t pt_base = amdgpu_gmc_pd_addr(adev->gart.bo);
+
+	mmhub_v3_0_2_setup_vm_pt_regs(adev, 0, pt_base);
+
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+		     (u32)(adev->gmc.gart_start >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+		     (u32)(adev->gmc.gart_start >> 44));
+
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+		     (u32)(adev->gmc.gart_end >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+		     (u32)(adev->gmc.gart_end >> 44));
+}
+
+static void mmhub_v3_0_2_init_system_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t value;
+	uint32_t tmp;
+
+	/* Program the AGP BAR */
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_AGP_BASE, 0);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_AGP_BOT, adev->gmc.agp_start >> 24);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_AGP_TOP, adev->gmc.agp_end >> 24);
+
+	if (!amdgpu_sriov_vf(adev)) {
+		/*
+		 * the new L1 policy will block SRIOV guest from writing
+		 * these regs, and they will be programed at host.
+		 * so skip programing these regs.
+		 */
+		/* Program the system aperture low logical page number. */
+		WREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+			     min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
+		WREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+			     max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
+	}
+
+	/* Set default page address. */
+	value = amdgpu_gmc_vram_mc2pa(adev, adev->mem_scratch.gpu_addr);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
+		     (u32)(value >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
+		     (u32)(value >> 44));
+
+	/* Program "protection fault". */
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
+		     (u32)(adev->dummy_page_addr >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+		     (u32)((u64)adev->dummy_page_addr >> 44));
+
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL2);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL2,
+			    ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL2, tmp);
+}
+
+static void mmhub_v3_0_2_init_tlb_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(MMHUB, 0, regMMMC_VM_MX_L1_TLB_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 1);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ECO_BITS, 0);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    MTYPE, MTYPE_UC); /* UC, uncached */
+
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_MX_L1_TLB_CNTL, tmp);
+}
+
+static void mmhub_v3_0_2_init_cache_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_CACHE, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL,
+			    ENABLE_DEFAULT_PAGE_OUT_TO_SYSTEM_MEMORY, 1);
+	/* XXX for emulation, Refer to closed source code.*/
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, L2_PDE0_CACHE_TAG_GENERATION_MODE,
+			    0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL, tmp);
+
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL2);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL2, tmp);
+
+	tmp = regMMVM_L2_CNTL3_DEFAULT;
+	if (adev->gmc.translate_further) {
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3, BANK_SELECT, 12);
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+	} else {
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3, BANK_SELECT, 9);
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+	}
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL3, tmp);
+
+	tmp = regMMVM_L2_CNTL4_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL4, tmp);
+
+	tmp = regMMVM_L2_CNTL5_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL5, L2_CACHE_SMALLK_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL5, tmp);
+}
+
+static void mmhub_v3_0_2_enable_system_domain(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL,
+			    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_CNTL, tmp);
+}
+
+static void mmhub_v3_0_2_disable_identity_aperture(struct amdgpu_device *adev)
+{
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	WREG32_SOC15(MMHUB, 0,
+		     regMMVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32,
+		     0xFFFFFFFF);
+	WREG32_SOC15(MMHUB, 0,
+		     regMMVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32,
+		     0x0000000F);
+
+	WREG32_SOC15(MMHUB, 0,
+		     regMMVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32, 0);
+	WREG32_SOC15(MMHUB, 0,
+		     regMMVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32, 0);
+
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32,
+		     0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32,
+		     0);
+}
+
+static void mmhub_v3_0_2_setup_vmid_config(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	int i;
+	uint32_t tmp;
+
+	for (i = 0; i <= 14; i++) {
+		tmp = RREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_CNTL, i * hub->ctx_distance);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH,
+				    adev->vm_manager.num_level);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    PAGE_TABLE_BLOCK_SIZE,
+				    adev->vm_manager.block_size - 9);
+		/* Send no-retry XNACK on fault to suppress VM fault storm. */
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
+				    !amdgpu_noretry);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_CNTL,
+				    i * hub->ctx_distance, tmp);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32,
+				    i * hub->ctx_addr_distance,
+				    lower_32_bits(adev->vm_manager.max_pfn - 1));
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32,
+				    i * hub->ctx_addr_distance,
+				    upper_32_bits(adev->vm_manager.max_pfn - 1));
+	}
+
+	hub->vm_cntx_cntl = tmp;
+}
+
+static void mmhub_v3_0_2_program_invalidation(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	unsigned i;
+
+	for (i = 0; i < 18; ++i) {
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
+				    i * hub->eng_addr_distance, 0xffffffff);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
+				    i * hub->eng_addr_distance, 0x1f);
+	}
+}
+
+static int mmhub_v3_0_2_gart_enable(struct amdgpu_device *adev)
+{
+	/* GART Enable. */
+	mmhub_v3_0_2_init_gart_aperture_regs(adev);
+	mmhub_v3_0_2_init_system_aperture_regs(adev);
+	mmhub_v3_0_2_init_tlb_regs(adev);
+	mmhub_v3_0_2_init_cache_regs(adev);
+
+	mmhub_v3_0_2_enable_system_domain(adev);
+	mmhub_v3_0_2_disable_identity_aperture(adev);
+	mmhub_v3_0_2_setup_vmid_config(adev);
+	mmhub_v3_0_2_program_invalidation(adev);
+
+	return 0;
+}
+
+static void mmhub_v3_0_2_gart_disable(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	u32 tmp;
+	u32 i;
+
+	/* Disable all tables */
+	for (i = 0; i < 16; i++)
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT0_CNTL,
+				    i * hub->ctx_distance, 0);
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(MMHUB, 0, regMMMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 0);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_MX_L1_TLB_CNTL, tmp);
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_CACHE, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL, tmp);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL3, 0);
+}
+
+/**
+ * mmhub_v3_0_2_set_fault_enable_default - update GART/VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+static void mmhub_v3_0_2_set_fault_enable_default(struct amdgpu_device *adev, bool value)
+{
+	u32 tmp;
+
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE1_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE2_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT,
+			    value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    NACK_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	if (!value) {
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_NO_RETRY_FAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_RETRY_FAULT, 1);
+	}
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL, tmp);
+}
+
+static const struct amdgpu_vmhub_funcs mmhub_v3_0_2_vmhub_funcs = {
+	.print_l2_protection_fault_status = mmhub_v3_0_2_print_l2_protection_fault_status,
+	.get_invalidate_req = mmhub_v3_0_2_get_invalidate_req,
+};
+
+static void mmhub_v3_0_2_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+
+	hub->ctx0_ptb_addr_lo32 =
+		SOC15_REG_OFFSET(MMHUB, 0,
+				 regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32);
+	hub->ctx0_ptb_addr_hi32 =
+		SOC15_REG_OFFSET(MMHUB, 0,
+				 regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32);
+	hub->vm_inv_eng0_sem =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_SEM);
+	hub->vm_inv_eng0_req =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_REQ);
+	hub->vm_inv_eng0_ack =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_ACK);
+	hub->vm_context0_cntl =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_CONTEXT0_CNTL);
+	hub->vm_l2_pro_fault_status =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_STATUS);
+	hub->vm_l2_pro_fault_cntl =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL);
+
+	hub->ctx_distance = regMMVM_CONTEXT1_CNTL - regMMVM_CONTEXT0_CNTL;
+	hub->ctx_addr_distance = regMMVM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 -
+		regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32;
+	hub->eng_distance = regMMVM_INVALIDATE_ENG1_REQ -
+		regMMVM_INVALIDATE_ENG0_REQ;
+	hub->eng_addr_distance = regMMVM_INVALIDATE_ENG1_ADDR_RANGE_LO32 -
+		regMMVM_INVALIDATE_ENG0_ADDR_RANGE_LO32;
+
+	hub->vm_cntx_cntl_vm_fault = MMVM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__EXECUTE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK;
+
+	hub->vm_l2_bank_select_reserved_cid2 =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_L2_BANK_SELECT_RESERVED_CID2);
+
+	hub->vmhub_funcs = &mmhub_v3_0_2_vmhub_funcs;
+}
+
+static u64 mmhub_v3_0_2_get_fb_location(struct amdgpu_device *adev)
+{
+	u64 base;
+
+	base = RREG32_SOC15(MMHUB, 0, regMMMC_VM_FB_LOCATION_BASE);
+	base &= MMMC_VM_FB_LOCATION_BASE__FB_BASE_MASK;
+	base <<= 24;
+
+	return base;
+}
+
+static u64 mmhub_v3_0_2_get_mc_fb_offset(struct amdgpu_device *adev)
+{
+	return (u64)RREG32_SOC15(MMHUB, 0, regMMMC_VM_FB_OFFSET) << 24;
+}
+
+static void mmhub_v3_0_2_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+							bool enable)
+{
+	//TODO
+}
+
+static void mmhub_v3_0_2_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+						       bool enable)
+{
+	//TODO
+}
+
+static int mmhub_v3_0_2_set_clockgating(struct amdgpu_device *adev,
+			       enum amd_clockgating_state state)
+{
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	mmhub_v3_0_2_update_medium_grain_clock_gating(adev,
+			state == AMD_CG_STATE_GATE);
+	mmhub_v3_0_2_update_medium_grain_light_sleep(adev,
+			state == AMD_CG_STATE_GATE);
+	return 0;
+}
+
+static void mmhub_v3_0_2_get_clockgating(struct amdgpu_device *adev, u64 *flags)
+{
+	//TODO
+}
+
+const struct amdgpu_mmhub_funcs mmhub_v3_0_2_funcs = {
+	.init = mmhub_v3_0_2_init,
+	.get_fb_location = mmhub_v3_0_2_get_fb_location,
+	.get_mc_fb_offset = mmhub_v3_0_2_get_mc_fb_offset,
+	.gart_enable = mmhub_v3_0_2_gart_enable,
+	.set_fault_enable_default = mmhub_v3_0_2_set_fault_enable_default,
+	.gart_disable = mmhub_v3_0_2_gart_disable,
+	.set_clockgating = mmhub_v3_0_2_set_clockgating,
+	.get_clockgating = mmhub_v3_0_2_get_clockgating,
+	.setup_vm_pt_regs = mmhub_v3_0_2_setup_vm_pt_regs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0_2.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0_2.h
new file mode 100644
index 000000000000..23ad7b156cdb
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0_2.h
@@ -0,0 +1,28 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __MMHUB_V3_0_2_H__
+#define __MMHUB_V3_0_2_H__
+
+extern const struct amdgpu_mmhub_funcs mmhub_v3_0_2_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v3_3.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_3.c
new file mode 100644
index 000000000000..f6fc9778bc30
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_3.c
@@ -0,0 +1,746 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "mmhub_v3_3.h"
+
+#include "mmhub/mmhub_3_3_0_offset.h"
+#include "mmhub/mmhub_3_3_0_sh_mask.h"
+
+#include "navi10_enum.h"
+#include "soc15_common.h"
+
+#define regMMVM_L2_CNTL3_DEFAULT				0x80100007
+#define regMMVM_L2_CNTL4_DEFAULT				0x000000c1
+#define regMMVM_L2_CNTL5_DEFAULT				0x00003fe0
+#define regDAGB0_L1TLB_REG_RW_3_3                   0x00a4
+#define regDAGB0_L1TLB_REG_RW_3_3_BASE_IDX          1
+#define regDAGB1_L1TLB_REG_RW_3_3                   0x0163
+#define regDAGB1_L1TLB_REG_RW_3_3_BASE_IDX          1
+
+static const char *mmhub_client_ids_v3_3[][2] = {
+	[0][0] = "VMC",
+	[1][0] = "ISPXT",
+	[2][0] = "ISPIXT",
+	[4][0] = "DCEDMC",
+	[6][0] = "MP0",
+	[7][0] = "MP1",
+	[8][0] = "MPM",
+	[9][0] = "ISPPDPRD",
+	[10][0] = "ISPCSTATRD",
+	[11][0] = "ISPBYRPRD",
+	[12][0] = "ISPRGBPRD",
+	[13][0] = "ISPMCFPRD",
+	[14][0] = "ISPMCFPRD1",
+	[15][0] = "ISPYUVPRD",
+	[16][0] = "ISPMCSCRD",
+	[17][0] = "ISPGDCRD",
+	[18][0] = "ISPLMERD",
+	[22][0] = "ISPXT1",
+	[23][0] = "ISPIXT1",
+	[24][0] = "HDP",
+	[25][0] = "LSDMA",
+	[26][0] = "JPEG",
+	[27][0] = "VPE",
+	[28][0] = "VSCH",
+	[29][0] = "VCNU",
+	[30][0] = "VCN",
+	[1][1] = "ISPXT",
+	[2][1] = "ISPIXT",
+	[3][1] = "DCEDWB",
+	[4][1] = "DCEDMC",
+	[5][1] = "ISPCSISWR",
+	[6][1] = "MP0",
+	[7][1] = "MP1",
+	[8][1] = "MPM",
+	[9][1] = "ISPPDPWR",
+	[10][1] = "ISPCSTATWR",
+	[11][1] = "ISPBYRPWR",
+	[12][1] = "ISPRGBPWR",
+	[13][1] = "ISPMCFPWR",
+	[14][1] = "ISPMWR0",
+	[15][1] = "ISPYUVPWR",
+	[16][1] = "ISPMCSCWR",
+	[17][1] = "ISPGDCWR",
+	[18][1] = "ISPLMEWR",
+	[20][1] = "ISPMWR2",
+	[21][1] = "OSSSYS",
+	[22][1] = "ISPXT1",
+	[23][1] = "ISPIXT1",
+	[24][1] = "HDP",
+	[25][1] = "LSDMA",
+	[26][1] = "JPEG",
+	[27][1] = "VPE",
+	[28][1] = "VSCH",
+	[29][1] = "VCNU",
+	[30][1] = "VCN",
+};
+
+static const char *mmhub_client_ids_v3_3_1[][2] = {
+	[0][0] = "VMC",
+	[4][0] = "DCEDMC",
+	[6][0] = "MP0",
+	[7][0] = "MP1",
+	[8][0] = "MPM",
+	[24][0] = "HDP",
+	[25][0] = "LSDMA",
+	[26][0] = "JPEG0",
+	[27][0] = "VPE0",
+	[28][0] = "VSCH",
+	[29][0] = "VCNU0",
+	[30][0] = "VCN0",
+	[32+1][0] = "ISPXT",
+	[32+2][0] = "ISPIXT",
+	[32+9][0] = "ISPPDPRD",
+	[32+10][0] = "ISPCSTATRD",
+	[32+11][0] = "ISPBYRPRD",
+	[32+12][0] = "ISPRGBPRD",
+	[32+13][0] = "ISPMCFPRD",
+	[32+14][0] = "ISPMCFPRD1",
+	[32+15][0] = "ISPYUVPRD",
+	[32+16][0] = "ISPMCSCRD",
+	[32+17][0] = "ISPGDCRD",
+	[32+18][0] = "ISPLMERD",
+	[32+22][0] = "ISPXT1",
+	[32+23][0] = "ISPIXT1",
+	[32+26][0] = "JPEG1",
+	[32+27][0] = "VPE1",
+	[32+29][0] = "VCNU1",
+	[32+30][0] = "VCN1",
+	[3][1] = "DCEDWB",
+	[4][1] = "DCEDMC",
+	[6][1] = "MP0",
+	[7][1] = "MP1",
+	[8][1] = "MPM",
+	[21][1] = "OSSSYS",
+	[24][1] = "HDP",
+	[25][1] = "LSDMA",
+	[26][1] = "JPEG0",
+	[27][1] = "VPE0",
+	[28][1] = "VSCH",
+	[29][1] = "VCNU0",
+	[30][1] = "VCN0",
+	[32+1][1] = "ISPXT",
+	[32+2][1] = "ISPIXT",
+	[32+5][1] = "ISPCSISWR",
+	[32+9][1] = "ISPPDPWR",
+	[32+10][1] = "ISPCSTATWR",
+	[32+11][1] = "ISPBYRPWR",
+	[32+12][1] = "ISPRGBPWR",
+	[32+13][1] = "ISPMCFPWR",
+	[32+14][1] = "ISPMWR0",
+	[32+15][1] = "ISPYUVPWR",
+	[32+16][1] = "ISPMCSCWR",
+	[32+17][1] = "ISPGDCWR",
+	[32+18][1] = "ISPLMEWR",
+	[32+19][1] = "ISPMWR1",
+	[32+20][1] = "ISPMWR2",
+	[32+22][1] = "ISPXT1",
+	[32+23][1] = "ISPIXT1",
+	[32+26][1] = "JPEG1",
+	[32+27][1] = "VPE1",
+	[32+29][1] = "VCNU1",
+	[32+30][1] = "VCN1",
+};
+
+static uint32_t mmhub_v3_3_get_invalidate_req(unsigned int vmid,
+						uint32_t flush_type)
+{
+	u32 req = 0;
+
+	/* invalidate using legacy mode on vmid*/
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ,
+			    PER_VMID_INVALIDATE_REQ, 1 << vmid);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, FLUSH_TYPE, flush_type ? : 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PTES, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE0, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE1, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE2, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L1_PTES, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ,
+			    CLEAR_PROTECTION_FAULT_STATUS_ADDR,	0);
+
+	return req;
+}
+
+static void
+mmhub_v3_3_print_l2_protection_fault_status(struct amdgpu_device *adev,
+					      uint32_t status)
+{
+	uint32_t cid, rw;
+	const char *mmhub_cid = NULL;
+
+	cid = REG_GET_FIELD(status,
+			    MMVM_L2_PROTECTION_FAULT_STATUS, CID);
+	rw = REG_GET_FIELD(status,
+			   MMVM_L2_PROTECTION_FAULT_STATUS, RW);
+
+	dev_err(adev->dev,
+		"MMVM_L2_PROTECTION_FAULT_STATUS:0x%08X\n",
+		status);
+
+	switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) {
+	case IP_VERSION(3, 3, 0):
+	case IP_VERSION(3, 3, 2):
+		mmhub_cid = cid < ARRAY_SIZE(mmhub_client_ids_v3_3) ?
+			    mmhub_client_ids_v3_3[cid][rw] :
+			    cid == 0x140 ? "UMSCH" : NULL;
+		break;
+	case IP_VERSION(3, 3, 1):
+		mmhub_cid = cid < ARRAY_SIZE(mmhub_client_ids_v3_3_1) ?
+			    mmhub_client_ids_v3_3_1[cid][rw] :
+			    cid == 0x140 ? "UMSCH" : NULL;
+		break;
+	default:
+		mmhub_cid = NULL;
+		break;
+	}
+
+	dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n",
+		mmhub_cid ? mmhub_cid : "unknown", cid);
+	dev_err(adev->dev, "\t MORE_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, MORE_FAULTS));
+	dev_err(adev->dev, "\t WALKER_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, WALKER_ERROR));
+	dev_err(adev->dev, "\t PERMISSION_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, PERMISSION_FAULTS));
+	dev_err(adev->dev, "\t MAPPING_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS, MAPPING_ERROR));
+	dev_err(adev->dev, "\t RW: 0x%x\n", rw);
+}
+
+static void mmhub_v3_3_setup_vm_pt_regs(struct amdgpu_device *adev,
+					  uint32_t vmid,
+					  uint64_t page_table_base)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+
+	WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+			    hub->ctx_addr_distance * vmid,
+			    lower_32_bits(page_table_base));
+
+	WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+			    hub->ctx_addr_distance * vmid,
+			    upper_32_bits(page_table_base));
+
+}
+
+static void mmhub_v3_3_init_gart_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t pt_base = amdgpu_gmc_pd_addr(adev->gart.bo);
+
+	mmhub_v3_3_setup_vm_pt_regs(adev, 0, pt_base);
+
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+		     (u32)(adev->gmc.gart_start >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+		     (u32)(adev->gmc.gart_start >> 44));
+
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+		     (u32)(adev->gmc.gart_end >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+		     (u32)(adev->gmc.gart_end >> 44));
+}
+
+static void mmhub_v3_3_init_system_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t value;
+	uint32_t tmp;
+
+	/* Program the AGP BAR */
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_AGP_BASE, 0);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_AGP_BOT, adev->gmc.agp_start >> 24);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_AGP_TOP, adev->gmc.agp_end >> 24);
+
+	/*
+	 * the new L1 policy will block SRIOV guest from writing
+	 * these regs, and they will be programed at host.
+	 * so skip programing these regs.
+	 */
+	/* Program the system aperture low logical page number. */
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+		     min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+		     max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
+
+	/* Set default page address. */
+	value = amdgpu_gmc_vram_mc2pa(adev, adev->mem_scratch.gpu_addr);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
+		     (u32)(value >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
+		     (u32)(value >> 44));
+
+	/* Program "protection fault". */
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
+		     (u32)(adev->dummy_page_addr >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+		     (u32)((u64)adev->dummy_page_addr >> 44));
+
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL2);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL2,
+			    ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL2, tmp);
+}
+
+static void mmhub_v3_3_init_tlb_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(MMHUB, 0, regMMMC_VM_MX_L1_TLB_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 1);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ECO_BITS, 0);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    MTYPE, MTYPE_UC); /* UC, uncached */
+
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_MX_L1_TLB_CNTL, tmp);
+}
+
+static void mmhub_v3_3_init_cache_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_CACHE, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL,
+			    ENABLE_DEFAULT_PAGE_OUT_TO_SYSTEM_MEMORY, 1);
+	/* XXX for emulation, Refer to closed source code.*/
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, L2_PDE0_CACHE_TAG_GENERATION_MODE,
+			    0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL, tmp);
+
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL2);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL2, tmp);
+
+	tmp = regMMVM_L2_CNTL3_DEFAULT;
+	if (adev->gmc.translate_further) {
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3, BANK_SELECT, 12);
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+	} else {
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3, BANK_SELECT, 9);
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+	}
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL3, tmp);
+
+	tmp = regMMVM_L2_CNTL4_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL4, tmp);
+
+	tmp = regMMVM_L2_CNTL5_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL5, L2_CACHE_SMALLK_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL5, tmp);
+}
+
+static void mmhub_v3_3_enable_system_domain(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL,
+			    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
+
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_CNTL, tmp);
+}
+
+static void mmhub_v3_3_disable_identity_aperture(struct amdgpu_device *adev)
+{
+	WREG32_SOC15(MMHUB, 0,
+		     regMMVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32,
+		     0xFFFFFFFF);
+	WREG32_SOC15(MMHUB, 0,
+		     regMMVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32,
+		     0x0000000F);
+
+	WREG32_SOC15(MMHUB, 0,
+		     regMMVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32, 0);
+	WREG32_SOC15(MMHUB, 0,
+		     regMMVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32, 0);
+
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32,
+		     0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32,
+		     0);
+}
+
+static void mmhub_v3_3_setup_vmid_config(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	int i;
+	uint32_t tmp;
+
+	for (i = 0; i <= 14; i++) {
+		tmp = RREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_CNTL, i * hub->ctx_distance);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH,
+				    adev->vm_manager.num_level);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    PAGE_TABLE_BLOCK_SIZE,
+				    adev->vm_manager.block_size - 9);
+		/* Send no-retry XNACK on fault to suppress VM fault storm. */
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
+				    !amdgpu_noretry);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_CNTL,
+				    i * hub->ctx_distance, tmp);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32,
+				    i * hub->ctx_addr_distance,
+				    lower_32_bits(adev->vm_manager.max_pfn - 1));
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32,
+				    i * hub->ctx_addr_distance,
+				    upper_32_bits(adev->vm_manager.max_pfn - 1));
+	}
+
+	hub->vm_cntx_cntl = tmp;
+}
+
+static void mmhub_v3_3_program_invalidation(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	unsigned int i;
+
+	for (i = 0; i < 18; ++i) {
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
+				    i * hub->eng_addr_distance, 0xffffffff);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
+				    i * hub->eng_addr_distance, 0x1f);
+	}
+}
+
+static void mmhub_v3_3_init_saw_regs(struct amdgpu_device *adev)
+{
+	uint64_t pt_base = amdgpu_gmc_pd_addr(adev->gart.bo);
+	uint32_t tmp;
+
+	/* Program page table base, gart start, gart end */
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_SAW_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+			lower_32_bits(pt_base >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_SAW_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+			upper_32_bits(pt_base >> 12));
+
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_SAW_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+		     (u32)(adev->gmc.gart_start >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_SAW_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+		     (u32)(adev->gmc.gart_start >> 44));
+
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_SAW_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+		     (u32)(adev->gmc.gart_end >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_SAW_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+		     (u32)(adev->gmc.gart_end >> 44));
+
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_SAW_CONTEXT0_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_SAW_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_SAW_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_SAW_CONTEXT0_CNTL, tmp);
+
+	/* Disable all contexts except context 0 */
+	tmp = 0xfffe;
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_SAW_CONTEXTS_DISABLE, tmp);
+
+	/* Program saw cntl4 */
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_SAW_CNTL4);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_SAW_CNTL4, VMC_TAP_CONTEXT0_PDE_REQUEST_SNOOP, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_SAW_CNTL4, VMC_TAP_CONTEXT0_PTE_REQUEST_SNOOP, 1);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_SAW_CNTL4, tmp);
+}
+
+static void mmhub_v3_3_enable_tls(struct amdgpu_device *adev)
+{
+	WREG32_SOC15(MMHUB, 0, regDAGB0_L1TLB_REG_RW_3_3, 0);
+	WREG32_SOC15(MMHUB, 0, regDAGB1_L1TLB_REG_RW_3_3, 3);
+}
+
+static int mmhub_v3_3_gart_enable(struct amdgpu_device *adev)
+{
+	/* GART Enable. */
+	mmhub_v3_3_init_gart_aperture_regs(adev);
+	mmhub_v3_3_init_system_aperture_regs(adev);
+	mmhub_v3_3_init_tlb_regs(adev);
+	mmhub_v3_3_init_cache_regs(adev);
+
+	mmhub_v3_3_enable_system_domain(adev);
+	mmhub_v3_3_disable_identity_aperture(adev);
+	mmhub_v3_3_setup_vmid_config(adev);
+	mmhub_v3_3_program_invalidation(adev);
+
+	/* standalone alone walker init */
+	mmhub_v3_3_init_saw_regs(adev);
+
+	/* enable mmhub tls */
+	mmhub_v3_3_enable_tls(adev);
+
+	return 0;
+}
+
+static void mmhub_v3_3_gart_disable(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	u32 tmp;
+	u32 i;
+
+	/* Disable all tables */
+	for (i = 0; i < 16; i++)
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT0_CNTL,
+				    i * hub->ctx_distance, 0);
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(MMHUB, 0, regMMMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 0);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_MX_L1_TLB_CNTL, tmp);
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_CACHE, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL, tmp);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL3, 0);
+}
+
+/**
+ * mmhub_v3_3_set_fault_enable_default - update GART/VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+static void mmhub_v3_3_set_fault_enable_default(struct amdgpu_device *adev,
+						  bool value)
+{
+	u32 tmp;
+
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE1_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE2_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT,
+			    value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    NACK_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	if (!value) {
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_NO_RETRY_FAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_RETRY_FAULT, 1);
+	}
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL, tmp);
+}
+
+static const struct amdgpu_vmhub_funcs mmhub_v3_3_vmhub_funcs = {
+	.print_l2_protection_fault_status = mmhub_v3_3_print_l2_protection_fault_status,
+	.get_invalidate_req = mmhub_v3_3_get_invalidate_req,
+};
+
+static void mmhub_v3_3_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+
+	hub->ctx0_ptb_addr_lo32 =
+		SOC15_REG_OFFSET(MMHUB, 0,
+				 regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32);
+	hub->ctx0_ptb_addr_hi32 =
+		SOC15_REG_OFFSET(MMHUB, 0,
+				 regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32);
+	hub->vm_inv_eng0_sem =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_SEM);
+	hub->vm_inv_eng0_req =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_REQ);
+	hub->vm_inv_eng0_ack =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_ACK);
+	hub->vm_context0_cntl =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_CONTEXT0_CNTL);
+	hub->vm_l2_pro_fault_status =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_STATUS);
+	hub->vm_l2_pro_fault_cntl =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL);
+
+	hub->ctx_distance = regMMVM_CONTEXT1_CNTL - regMMVM_CONTEXT0_CNTL;
+	hub->ctx_addr_distance = regMMVM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 -
+		regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32;
+	hub->eng_distance = regMMVM_INVALIDATE_ENG1_REQ -
+		regMMVM_INVALIDATE_ENG0_REQ;
+	hub->eng_addr_distance = regMMVM_INVALIDATE_ENG1_ADDR_RANGE_LO32 -
+		regMMVM_INVALIDATE_ENG0_ADDR_RANGE_LO32;
+
+	hub->vm_cntx_cntl_vm_fault = MMVM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__EXECUTE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK;
+
+	hub->vmhub_funcs = &mmhub_v3_3_vmhub_funcs;
+}
+
+static u64 mmhub_v3_3_get_fb_location(struct amdgpu_device *adev)
+{
+	u64 base;
+
+	base = RREG32_SOC15(MMHUB, 0, regMMMC_VM_FB_LOCATION_BASE);
+	base &= MMMC_VM_FB_LOCATION_BASE__FB_BASE_MASK;
+	base <<= 24;
+
+	return base;
+}
+
+static u64 mmhub_v3_3_get_mc_fb_offset(struct amdgpu_device *adev)
+{
+	u64 offset;
+
+	offset = RREG32_SOC15(MMHUB, 0, regMMMC_VM_FB_OFFSET);
+	offset &= MMMC_VM_FB_OFFSET__FB_OFFSET_MASK;
+	offset <<= 24;
+
+	return offset;
+}
+
+static void mmhub_v3_3_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+							  bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_SOC15(MMHUB, 0, regMM_ATC_L2_MISC_CG);
+
+	if (enable)
+		data |= MM_ATC_L2_MISC_CG__ENABLE_MASK;
+	else
+		data &= ~MM_ATC_L2_MISC_CG__ENABLE_MASK;
+
+	if (def != data)
+		WREG32_SOC15(MMHUB, 0, regMM_ATC_L2_MISC_CG, data);
+}
+
+static void mmhub_v3_3_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+							 bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_SOC15(MMHUB, 0, regMM_ATC_L2_MISC_CG);
+
+	if (enable)
+		data |= MM_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK;
+	else
+		data &= ~MM_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK;
+
+	if (def != data)
+		WREG32_SOC15(MMHUB, 0, regMM_ATC_L2_MISC_CG, data);
+}
+
+static int mmhub_v3_3_set_clockgating(struct amdgpu_device *adev,
+					enum amd_clockgating_state state)
+{
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	mmhub_v3_3_update_medium_grain_clock_gating(adev,
+			state == AMD_CG_STATE_GATE);
+	mmhub_v3_3_update_medium_grain_light_sleep(adev,
+			state == AMD_CG_STATE_GATE);
+	return 0;
+}
+
+static void mmhub_v3_3_get_clockgating(struct amdgpu_device *adev, u64 *flags)
+{
+	u32 data;
+
+	if (amdgpu_sriov_vf(adev))
+		*flags = 0;
+
+	data = RREG32_SOC15(MMHUB, 0, regMM_ATC_L2_MISC_CG);
+
+	/* AMD_CG_SUPPORT_MC_MGCG */
+	if (data & MM_ATC_L2_MISC_CG__ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_MC_MGCG;
+
+	/* AMD_CG_SUPPORT_MC_LS */
+	if (data & MM_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_MC_LS;
+}
+
+const struct amdgpu_mmhub_funcs mmhub_v3_3_funcs = {
+	.init = mmhub_v3_3_init,
+	.get_fb_location = mmhub_v3_3_get_fb_location,
+	.get_mc_fb_offset = mmhub_v3_3_get_mc_fb_offset,
+	.gart_enable = mmhub_v3_3_gart_enable,
+	.set_fault_enable_default = mmhub_v3_3_set_fault_enable_default,
+	.gart_disable = mmhub_v3_3_gart_disable,
+	.set_clockgating = mmhub_v3_3_set_clockgating,
+	.get_clockgating = mmhub_v3_3_get_clockgating,
+	.setup_vm_pt_regs = mmhub_v3_3_setup_vm_pt_regs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v3_3.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_3.h
new file mode 100644
index 000000000000..37b62c7e5a4a
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_3.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __MMHUB_V3_3_H__
+#define __MMHUB_V3_3_H__
+
+extern const struct amdgpu_mmhub_funcs mmhub_v3_3_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v4_1_0.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v4_1_0.c
new file mode 100644
index 000000000000..951998454b25
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v4_1_0.c
@@ -0,0 +1,647 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "mmhub_v4_1_0.h"
+
+#include "mmhub/mmhub_4_1_0_offset.h"
+#include "mmhub/mmhub_4_1_0_sh_mask.h"
+
+#include "soc15_common.h"
+#include "soc24_enum.h"
+
+#define regMMVM_L2_CNTL3_DEFAULT				0x80100007
+#define regMMVM_L2_CNTL4_DEFAULT				0x000000c1
+#define regMMVM_L2_CNTL5_DEFAULT				0x00003fe0
+
+static const char *mmhub_client_ids_v4_1_0[][2] = {
+	[0][0] = "VMC",
+	[4][0] = "DCEDMC",
+	[6][0] = "MP0",
+	[7][0] = "MP1",
+	[8][0] = "MPIO",
+	[16][0] = "LSDMA",
+	[17][0] = "JPEG",
+	[19][0] = "VCNU",
+	[22][0] = "VSCH",
+	[23][0] = "HDP",
+	[32+23][0] = "VCNRD",
+	[3][1] = "DCEDWB",
+	[4][1] = "DCEDMC",
+	[6][1] = "MP0",
+	[7][1] = "MP1",
+	[8][1] = "MPIO",
+	[10][1] = "DBGU0",
+	[11][1] = "DBGU1",
+	[12][1] = "DBGUNBIO",
+	[14][1] = "XDP",
+	[15][1] = "OSSSYS",
+	[16][1] = "LSDMA",
+	[17][1] = "JPEG",
+	[18][1] = "VCNWR",
+	[19][1] = "VCNU",
+	[22][1] = "VSCH",
+	[23][1] = "HDP",
+};
+
+static uint32_t mmhub_v4_1_0_get_invalidate_req(unsigned int vmid,
+						uint32_t flush_type)
+{
+	u32 req = 0;
+
+	/* invalidate using legacy mode on vmid*/
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ,
+			    PER_VMID_INVALIDATE_REQ, 1 << vmid);
+	/* Only use legacy inv on mmhub side */
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, FLUSH_TYPE, 0);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PTES, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE0, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE1, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE2, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L1_PTES, 1);
+	req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ,
+			    CLEAR_PROTECTION_FAULT_STATUS_ADDR,	0);
+
+	return req;
+}
+
+static void
+mmhub_v4_1_0_print_l2_protection_fault_status(struct amdgpu_device *adev,
+					      uint32_t status)
+{
+	uint32_t cid, rw;
+	const char *mmhub_cid = NULL;
+
+	cid = REG_GET_FIELD(status,
+			    MMVM_L2_PROTECTION_FAULT_STATUS_LO32, CID);
+	rw = REG_GET_FIELD(status,
+			   MMVM_L2_PROTECTION_FAULT_STATUS_LO32, RW);
+
+	dev_err(adev->dev,
+		"MMVM_L2_PROTECTION_FAULT_STATUS_LO32:0x%08X\n",
+		status);
+	switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) {
+	case IP_VERSION(4, 1, 0):
+		mmhub_cid = mmhub_client_ids_v4_1_0[cid][rw];
+		break;
+	default:
+		mmhub_cid = NULL;
+		break;
+	}
+	dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n",
+		mmhub_cid ? mmhub_cid : "unknown", cid);
+	dev_err(adev->dev, "\t MORE_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS_LO32, MORE_FAULTS));
+	dev_err(adev->dev, "\t WALKER_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS_LO32, WALKER_ERROR));
+	dev_err(adev->dev, "\t PERMISSION_FAULTS: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS_LO32, PERMISSION_FAULTS));
+	dev_err(adev->dev, "\t MAPPING_ERROR: 0x%lx\n",
+		REG_GET_FIELD(status,
+		MMVM_L2_PROTECTION_FAULT_STATUS_LO32, MAPPING_ERROR));
+	dev_err(adev->dev, "\t RW: 0x%x\n", rw);
+}
+
+static void mmhub_v4_1_0_setup_vm_pt_regs(struct amdgpu_device *adev,
+					  uint32_t vmid, uint64_t page_table_base)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+
+	WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+			    hub->ctx_addr_distance * vmid,
+			    lower_32_bits(page_table_base));
+
+	WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+			    hub->ctx_addr_distance * vmid,
+			    upper_32_bits(page_table_base));
+}
+
+static void mmhub_v4_1_0_init_gart_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t pt_base = amdgpu_gmc_pd_addr(adev->gart.bo);
+
+	mmhub_v4_1_0_setup_vm_pt_regs(adev, 0, pt_base);
+
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+		     (u32)(adev->gmc.gart_start >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+		     (u32)(adev->gmc.gart_start >> 44));
+
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+		     (u32)(adev->gmc.gart_end >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+		     (u32)(adev->gmc.gart_end >> 44));
+}
+
+static void mmhub_v4_1_0_init_system_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t value;
+	uint32_t tmp;
+
+	/*
+	 * the new L1 policy will block SRIOV guest from writing
+	 * these regs, and they will be programed at host.
+	 * so skip programing these regs.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/* Program the AGP BAR */
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_AGP_BASE, 0);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_AGP_BOT, adev->gmc.agp_start >> 24);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_AGP_TOP, adev->gmc.agp_end >> 24);
+
+	/* Program the system aperture low logical page number. */
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+		     min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+		     max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
+
+	/* Set default page address. */
+	value = adev->mem_scratch.gpu_addr - adev->gmc.vram_start +
+		adev->vm_manager.vram_base_offset;
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
+		     (u32)(value >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
+		     (u32)(value >> 44));
+
+	/* Program "protection fault". */
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
+		     (u32)(adev->dummy_page_addr >> 12));
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+		     (u32)((u64)adev->dummy_page_addr >> 44));
+
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL2);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL2,
+			    ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL2, tmp);
+}
+
+static void mmhub_v4_1_0_init_tlb_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(MMHUB, 0, regMMMC_VM_MX_L1_TLB_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 1);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ECO_BITS, 0);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    MTYPE, MTYPE_UC); /* UC, uncached */
+
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_MX_L1_TLB_CNTL, tmp);
+}
+
+static void mmhub_v4_1_0_init_cache_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_CACHE, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL,
+			    ENABLE_DEFAULT_PAGE_OUT_TO_SYSTEM_MEMORY, 1);
+	/* XXX for emulation, Refer to closed source code.*/
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, L2_PDE0_CACHE_TAG_GENERATION_MODE,
+			    0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL, tmp);
+
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL2);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL2, tmp);
+
+	tmp = regMMVM_L2_CNTL3_DEFAULT;
+	if (adev->gmc.translate_further) {
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3, BANK_SELECT, 12);
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+	} else {
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3, BANK_SELECT, 9);
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+	}
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL3, tmp);
+
+	tmp = regMMVM_L2_CNTL4_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL4, tmp);
+
+	tmp = regMMVM_L2_CNTL5_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL5, L2_CACHE_SMALLK_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL5, tmp);
+}
+
+static void mmhub_v4_1_0_enable_system_domain(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL,
+			    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_CONTEXT0_CNTL, tmp);
+}
+
+static void mmhub_v4_1_0_disable_identity_aperture(struct amdgpu_device *adev)
+{
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	WREG32_SOC15(MMHUB, 0,
+		     regMMVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32,
+		     0xFFFFFFFF);
+	WREG32_SOC15(MMHUB, 0,
+		     regMMVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32,
+		     0x0000000F);
+
+	WREG32_SOC15(MMHUB, 0,
+		     regMMVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32, 0);
+	WREG32_SOC15(MMHUB, 0,
+		     regMMVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32, 0);
+
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32,
+		     0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32,
+		     0);
+}
+
+static void mmhub_v4_1_0_setup_vmid_config(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	int i;
+	uint32_t tmp;
+
+	for (i = 0; i <= 14; i++) {
+		tmp = RREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_CNTL, i);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH,
+				    adev->vm_manager.num_level);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    PAGE_TABLE_BLOCK_SIZE,
+				    adev->vm_manager.block_size - 9);
+		/* Send no-retry XNACK on fault to suppress VM fault storm. */
+		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
+				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
+				    !amdgpu_noretry);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_CNTL,
+				    i * hub->ctx_distance, tmp);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32,
+				    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32,
+				    i * hub->ctx_addr_distance,
+				    lower_32_bits(adev->vm_manager.max_pfn - 1));
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32,
+				    i * hub->ctx_addr_distance,
+				    upper_32_bits(adev->vm_manager.max_pfn - 1));
+	}
+
+	hub->vm_cntx_cntl = tmp;
+}
+
+static void mmhub_v4_1_0_program_invalidation(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	unsigned i;
+
+	for (i = 0; i < 18; ++i) {
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
+				    i * hub->eng_addr_distance, 0xffffffff);
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
+				    i * hub->eng_addr_distance, 0x1f);
+	}
+}
+
+static int mmhub_v4_1_0_gart_enable(struct amdgpu_device *adev)
+{
+	/* GART Enable. */
+	mmhub_v4_1_0_init_gart_aperture_regs(adev);
+	mmhub_v4_1_0_init_system_aperture_regs(adev);
+	mmhub_v4_1_0_init_tlb_regs(adev);
+	mmhub_v4_1_0_init_cache_regs(adev);
+
+	mmhub_v4_1_0_enable_system_domain(adev);
+	mmhub_v4_1_0_disable_identity_aperture(adev);
+	mmhub_v4_1_0_setup_vmid_config(adev);
+	mmhub_v4_1_0_program_invalidation(adev);
+
+	return 0;
+}
+
+static void mmhub_v4_1_0_gart_disable(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	u32 tmp;
+	u32 i;
+
+	/* Disable all tables */
+	for (i = 0; i < 16; i++)
+		WREG32_SOC15_OFFSET(MMHUB, 0, regMMVM_CONTEXT0_CNTL,
+				    i * hub->ctx_distance, 0);
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(MMHUB, 0, regMMMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
+	tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 0);
+	WREG32_SOC15(MMHUB, 0, regMMMC_VM_MX_L1_TLB_CNTL, tmp);
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_CACHE, 0);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL, tmp);
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_CNTL3, 0);
+}
+
+/**
+ * mmhub_v4_1_0_set_fault_enable_default - update GART/VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+static void
+mmhub_v4_1_0_set_fault_enable_default(struct amdgpu_device *adev, bool value)
+{
+	u32 tmp;
+
+	/* These registers are not accessible to VF-SRIOV.
+	 * The PF will program them instead.
+	 */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	tmp = RREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE1_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    PDE2_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT,
+			    value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    NACK_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+			    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	if (!value) {
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_NO_RETRY_FAULT, 1);
+		tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_RETRY_FAULT, 1);
+	}
+	WREG32_SOC15(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL, tmp);
+}
+
+static const struct amdgpu_vmhub_funcs mmhub_v4_1_0_vmhub_funcs = {
+	.print_l2_protection_fault_status = mmhub_v4_1_0_print_l2_protection_fault_status,
+	.get_invalidate_req = mmhub_v4_1_0_get_invalidate_req,
+};
+
+static void mmhub_v4_1_0_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+
+	hub->ctx0_ptb_addr_lo32 =
+		SOC15_REG_OFFSET(MMHUB, 0,
+				 regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32);
+	hub->ctx0_ptb_addr_hi32 =
+		SOC15_REG_OFFSET(MMHUB, 0,
+				 regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32);
+	hub->vm_inv_eng0_sem =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_SEM);
+	hub->vm_inv_eng0_req =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_REQ);
+	hub->vm_inv_eng0_ack =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_INVALIDATE_ENG0_ACK);
+	hub->vm_context0_cntl =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_CONTEXT0_CNTL);
+	hub->vm_l2_pro_fault_status =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_STATUS_LO32);
+	hub->vm_l2_pro_fault_cntl =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_L2_PROTECTION_FAULT_CNTL);
+
+	hub->ctx_distance = regMMVM_CONTEXT1_CNTL - regMMVM_CONTEXT0_CNTL;
+	hub->ctx_addr_distance = regMMVM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 -
+		regMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32;
+	hub->eng_distance = regMMVM_INVALIDATE_ENG1_REQ -
+		regMMVM_INVALIDATE_ENG0_REQ;
+	hub->eng_addr_distance = regMMVM_INVALIDATE_ENG1_ADDR_RANGE_LO32 -
+		regMMVM_INVALIDATE_ENG0_ADDR_RANGE_LO32;
+
+	hub->vm_cntx_cntl_vm_fault = MMVM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		MMVM_CONTEXT1_CNTL__EXECUTE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK;
+
+	hub->vm_l2_bank_select_reserved_cid2 =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_L2_BANK_SELECT_RESERVED_CID2);
+
+	hub->vm_contexts_disable =
+		SOC15_REG_OFFSET(MMHUB, 0, regMMVM_CONTEXTS_DISABLE);
+
+	hub->vmhub_funcs = &mmhub_v4_1_0_vmhub_funcs;
+}
+
+static u64 mmhub_v4_1_0_get_fb_location(struct amdgpu_device *adev)
+{
+	u64 base;
+
+	base = RREG32_SOC15(MMHUB, 0, regMMMC_VM_FB_LOCATION_BASE);
+
+	base &= MMMC_VM_FB_LOCATION_BASE__FB_BASE_MASK;
+	base <<= 24;
+
+	return base;
+}
+
+static u64 mmhub_v4_1_0_get_mc_fb_offset(struct amdgpu_device *adev)
+{
+	return (u64)RREG32_SOC15(MMHUB, 0, regMMMC_VM_FB_OFFSET) << 24;
+}
+
+static void
+mmhub_v4_1_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+					      bool enable)
+{
+#if 0
+	uint32_t def, data;
+#endif
+	uint32_t def1, data1, def2 = 0, data2 = 0;
+#if 0
+	def  = data  = RREG32_SOC15(MMHUB, 0, regMM_ATC_L2_MISC_CG);
+#endif
+	def1 = data1 = RREG32_SOC15(MMHUB, 0, regDAGB0_CNTL_MISC2);
+	def2 = data2 = RREG32_SOC15(MMHUB, 0, regDAGB1_CNTL_MISC2);
+
+	if (enable) {
+#if 0
+		data |= MM_ATC_L2_MISC_CG__ENABLE_MASK;
+#endif
+		data1 &= ~(DAGB0_CNTL_MISC2__DISABLE_RDRET_TAP_CHAIN_FGCG_MASK |
+			   DAGB0_CNTL_MISC2__DISABLE_WRRET_TAP_CHAIN_FGCG_MASK);
+
+		data2 &= ~(DAGB1_CNTL_MISC2__DISABLE_RDRET_TAP_CHAIN_FGCG_MASK |
+			   DAGB1_CNTL_MISC2__DISABLE_WRRET_TAP_CHAIN_FGCG_MASK);
+	} else {
+#if 0
+		data &= ~MM_ATC_L2_MISC_CG__ENABLE_MASK;
+#endif
+		data1 |= (DAGB0_CNTL_MISC2__DISABLE_RDRET_TAP_CHAIN_FGCG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_WRRET_TAP_CHAIN_FGCG_MASK);
+
+		data2 |= (DAGB1_CNTL_MISC2__DISABLE_RDRET_TAP_CHAIN_FGCG_MASK |
+			  DAGB1_CNTL_MISC2__DISABLE_WRRET_TAP_CHAIN_FGCG_MASK);
+	}
+
+#if 0
+	if (def != data)
+		WREG32_SOC15(MMHUB, 0, regMM_ATC_L2_MISC_CG, data);
+#endif
+	if (def1 != data1)
+		WREG32_SOC15(MMHUB, 0, regDAGB0_CNTL_MISC2, data1);
+
+	if (def2 != data2)
+		WREG32_SOC15(MMHUB, 0, regDAGB1_CNTL_MISC2, data2);
+}
+
+static void
+mmhub_v4_1_0_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+					     bool enable)
+{
+#if 0
+	uint32_t def, data;
+
+	def = data = RREG32_SOC15(MMHUB, 0, regMM_ATC_L2_MISC_CG);
+
+	if (enable)
+		data |= MM_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK;
+	else
+		data &= ~MM_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK;
+
+	if (def != data)
+		WREG32_SOC15(MMHUB, 0, regMM_ATC_L2_MISC_CG, data);
+#endif
+}
+
+static int mmhub_v4_1_0_set_clockgating(struct amdgpu_device *adev,
+					enum amd_clockgating_state state)
+{
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG)
+		mmhub_v4_1_0_update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_MC_LS)
+		mmhub_v4_1_0_update_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+
+	return 0;
+}
+
+static void mmhub_v4_1_0_get_clockgating(struct amdgpu_device *adev, u64 *flags)
+{
+#if 0
+	int data;
+
+	if (amdgpu_sriov_vf(adev))
+		*flags = 0;
+
+	data = RREG32_SOC15(MMHUB, 0, regMM_ATC_L2_MISC_CG);
+
+	/* AMD_CG_SUPPORT_MC_MGCG */
+	if (data & MM_ATC_L2_MISC_CG__ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_MC_MGCG;
+
+	/* AMD_CG_SUPPORT_MC_LS */
+	if (data & MM_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_MC_LS;
+#endif
+}
+
+const struct amdgpu_mmhub_funcs mmhub_v4_1_0_funcs = {
+	.init = mmhub_v4_1_0_init,
+	.get_fb_location = mmhub_v4_1_0_get_fb_location,
+	.get_mc_fb_offset = mmhub_v4_1_0_get_mc_fb_offset,
+	.gart_enable = mmhub_v4_1_0_gart_enable,
+	.set_fault_enable_default = mmhub_v4_1_0_set_fault_enable_default,
+	.gart_disable = mmhub_v4_1_0_gart_disable,
+	.set_clockgating = mmhub_v4_1_0_set_clockgating,
+	.get_clockgating = mmhub_v4_1_0_get_clockgating,
+	.setup_vm_pt_regs = mmhub_v4_1_0_setup_vm_pt_regs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v4_1_0.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v4_1_0.h
new file mode 100644
index 000000000000..3902d653353c
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v4_1_0.h
@@ -0,0 +1,28 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __MMHUB_V4_1_0_H__
+#define __MMHUB_V4_1_0_H__
+
+extern const struct amdgpu_mmhub_funcs mmhub_v4_1_0_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c
new file mode 100644
index 000000000000..fe0710b55c3a
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c
@@ -0,0 +1,1718 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "amdgpu_ras.h"
+#include "mmhub_v9_4.h"
+
+#include "mmhub/mmhub_9_4_1_offset.h"
+#include "mmhub/mmhub_9_4_1_sh_mask.h"
+#include "mmhub/mmhub_9_4_1_default.h"
+#include "athub/athub_1_0_offset.h"
+#include "athub/athub_1_0_sh_mask.h"
+#include "vega10_enum.h"
+#include "soc15.h"
+#include "soc15_common.h"
+
+#define MMHUB_NUM_INSTANCES			2
+#define MMHUB_INSTANCE_REGISTER_OFFSET		0x3000
+
+static u64 mmhub_v9_4_get_fb_location(struct amdgpu_device *adev)
+{
+	/* The base should be same b/t 2 mmhubs on Acrturus. Read one here. */
+	u64 base = RREG32_SOC15(MMHUB, 0, mmVMSHAREDVC0_MC_VM_FB_LOCATION_BASE);
+	u64 top = RREG32_SOC15(MMHUB, 0, mmVMSHAREDVC0_MC_VM_FB_LOCATION_TOP);
+
+	base &= VMSHAREDVC0_MC_VM_FB_LOCATION_BASE__FB_BASE_MASK;
+	base <<= 24;
+
+	top &= VMSHAREDVC0_MC_VM_FB_LOCATION_TOP__FB_TOP_MASK;
+	top <<= 24;
+
+	adev->gmc.fb_start = base;
+	adev->gmc.fb_end = top;
+
+	return base;
+}
+
+static void mmhub_v9_4_setup_hubid_vm_pt_regs(struct amdgpu_device *adev, int hubid,
+				uint32_t vmid, uint64_t value)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+
+	WREG32_SOC15_OFFSET(MMHUB, 0,
+			    mmVML2VC0_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+			    hub->ctx_addr_distance * vmid + hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+			    lower_32_bits(value));
+
+	WREG32_SOC15_OFFSET(MMHUB, 0,
+			    mmVML2VC0_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+			    hub->ctx_addr_distance * vmid + hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+			    upper_32_bits(value));
+
+}
+
+static void mmhub_v9_4_init_gart_aperture_regs(struct amdgpu_device *adev,
+					       int hubid)
+{
+	uint64_t pt_base = amdgpu_gmc_pd_addr(adev->gart.bo);
+
+	mmhub_v9_4_setup_hubid_vm_pt_regs(adev, hubid, 0, pt_base);
+
+	WREG32_SOC15_OFFSET(MMHUB, 0,
+			    mmVML2VC0_VM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+			    hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+			    (u32)(adev->gmc.gart_start >> 12));
+	WREG32_SOC15_OFFSET(MMHUB, 0,
+			    mmVML2VC0_VM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+			    hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+			    (u32)(adev->gmc.gart_start >> 44));
+
+	WREG32_SOC15_OFFSET(MMHUB, 0,
+			    mmVML2VC0_VM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+			    hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+			    (u32)(adev->gmc.gart_end >> 12));
+	WREG32_SOC15_OFFSET(MMHUB, 0,
+			    mmVML2VC0_VM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+			    hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+			    (u32)(adev->gmc.gart_end >> 44));
+}
+
+static void mmhub_v9_4_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid,
+				uint64_t page_table_base)
+{
+	int i;
+
+	for (i = 0; i < MMHUB_NUM_INSTANCES; i++)
+		mmhub_v9_4_setup_hubid_vm_pt_regs(adev, i, vmid,
+				page_table_base);
+}
+
+static void mmhub_v9_4_init_system_aperture_regs(struct amdgpu_device *adev,
+						int hubid)
+{
+	uint64_t value;
+	uint32_t tmp;
+
+	/* Program the AGP BAR */
+	WREG32_SOC15_OFFSET(MMHUB, 0, mmVMSHAREDVC0_MC_VM_AGP_BASE,
+			    hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+			    0);
+	WREG32_SOC15_OFFSET(MMHUB, 0, mmVMSHAREDVC0_MC_VM_AGP_TOP,
+			    hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+			    adev->gmc.agp_end >> 24);
+	WREG32_SOC15_OFFSET(MMHUB, 0, mmVMSHAREDVC0_MC_VM_AGP_BOT,
+			    hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+			    adev->gmc.agp_start >> 24);
+
+	if (!amdgpu_sriov_vf(adev)) {
+		/* Program the system aperture low logical page number. */
+		WREG32_SOC15_OFFSET(
+			MMHUB, 0, mmVMSHAREDVC0_MC_VM_SYSTEM_APERTURE_LOW_ADDR,
+			hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+			min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
+		WREG32_SOC15_OFFSET(
+			MMHUB, 0, mmVMSHAREDVC0_MC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+			hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+			max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
+
+		/* Set default page address. */
+		value = amdgpu_gmc_vram_mc2pa(adev, adev->mem_scratch.gpu_addr);
+		WREG32_SOC15_OFFSET(
+			MMHUB, 0,
+			mmVMSHAREDPF0_MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
+			hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+			(u32)(value >> 12));
+		WREG32_SOC15_OFFSET(
+			MMHUB, 0,
+			mmVMSHAREDPF0_MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
+			hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+			(u32)(value >> 44));
+
+		/* Program "protection fault". */
+		WREG32_SOC15_OFFSET(
+			MMHUB, 0,
+			mmVML2PF0_VM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
+			hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+			(u32)(adev->dummy_page_addr >> 12));
+		WREG32_SOC15_OFFSET(
+			MMHUB, 0,
+			mmVML2PF0_VM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+			hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+			(u32)((u64)adev->dummy_page_addr >> 44));
+
+		tmp = RREG32_SOC15_OFFSET(
+			MMHUB, 0, mmVML2PF0_VM_L2_PROTECTION_FAULT_CNTL2,
+			hubid * MMHUB_INSTANCE_REGISTER_OFFSET);
+		tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_PROTECTION_FAULT_CNTL2,
+				    ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+		WREG32_SOC15_OFFSET(MMHUB, 0,
+				    mmVML2PF0_VM_L2_PROTECTION_FAULT_CNTL2,
+				    hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+				    tmp);
+	}
+}
+
+static void mmhub_v9_4_init_tlb_regs(struct amdgpu_device *adev, int hubid)
+{
+	uint32_t tmp;
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15_OFFSET(MMHUB, 0,
+			   mmVMSHAREDVC0_MC_VM_MX_L1_TLB_CNTL,
+			   hubid * MMHUB_INSTANCE_REGISTER_OFFSET);
+
+	tmp = REG_SET_FIELD(tmp, VMSHAREDVC0_MC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_L1_TLB, 1);
+	tmp = REG_SET_FIELD(tmp, VMSHAREDVC0_MC_VM_MX_L1_TLB_CNTL,
+			    SYSTEM_ACCESS_MODE, 3);
+	tmp = REG_SET_FIELD(tmp, VMSHAREDVC0_MC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 1);
+	tmp = REG_SET_FIELD(tmp, VMSHAREDVC0_MC_VM_MX_L1_TLB_CNTL,
+			    SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+	tmp = REG_SET_FIELD(tmp, VMSHAREDVC0_MC_VM_MX_L1_TLB_CNTL,
+			    MTYPE, MTYPE_UC);/* XXX for emulation. */
+	tmp = REG_SET_FIELD(tmp, VMSHAREDVC0_MC_VM_MX_L1_TLB_CNTL,
+			    ATC_EN, 1);
+
+	WREG32_SOC15_OFFSET(MMHUB, 0, mmVMSHAREDVC0_MC_VM_MX_L1_TLB_CNTL,
+			    hubid * MMHUB_INSTANCE_REGISTER_OFFSET, tmp);
+}
+
+/* Set snoop bit for SDMA so that SDMA writes probe-invalidates RW lines */
+static void mmhub_v9_4_init_snoop_override_regs(struct amdgpu_device *adev, int hubid)
+{
+	uint32_t tmp;
+	int i;
+	uint32_t distance = mmDAGB1_WRCLI_GPU_SNOOP_OVERRIDE -
+			    mmDAGB0_WRCLI_GPU_SNOOP_OVERRIDE;
+	uint32_t huboffset = hubid * MMHUB_INSTANCE_REGISTER_OFFSET;
+
+	for (i = 0; i < 5 - (2 * hubid); i++) {
+		/* DAGB instances 0 to 4 are in hub0 and 5 to 7 are in hub1 */
+		tmp = RREG32_SOC15_OFFSET(MMHUB, 0,
+			mmDAGB0_WRCLI_GPU_SNOOP_OVERRIDE,
+			huboffset + i * distance);
+		tmp |= (1 << 15); /* SDMA client is BIT15 */
+		WREG32_SOC15_OFFSET(MMHUB, 0,
+			mmDAGB0_WRCLI_GPU_SNOOP_OVERRIDE,
+			huboffset + i * distance, tmp);
+
+		tmp = RREG32_SOC15_OFFSET(MMHUB, 0,
+			mmDAGB0_WRCLI_GPU_SNOOP_OVERRIDE_VALUE,
+			huboffset + i * distance);
+		tmp |= (1 << 15);
+		WREG32_SOC15_OFFSET(MMHUB, 0,
+			mmDAGB0_WRCLI_GPU_SNOOP_OVERRIDE_VALUE,
+			huboffset + i * distance, tmp);
+	}
+
+}
+
+static void mmhub_v9_4_init_cache_regs(struct amdgpu_device *adev, int hubid)
+{
+	uint32_t tmp;
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15_OFFSET(MMHUB, 0, mmVML2PF0_VM_L2_CNTL,
+				  hubid * MMHUB_INSTANCE_REGISTER_OFFSET);
+	tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_CNTL,
+			    ENABLE_L2_CACHE, 1);
+	tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_CNTL,
+			    ENABLE_L2_FRAGMENT_PROCESSING, 1);
+	/* XXX for emulation, Refer to closed source code.*/
+	tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_CNTL,
+			    L2_PDE0_CACHE_TAG_GENERATION_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_CNTL,
+			    PDE_FAULT_CLASSIFICATION, 0);
+	tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_CNTL,
+			    CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+	tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_CNTL,
+			    IDENTITY_MODE_FRAGMENT_SIZE, 0);
+	WREG32_SOC15_OFFSET(MMHUB, 0, mmVML2PF0_VM_L2_CNTL,
+		     hubid * MMHUB_INSTANCE_REGISTER_OFFSET, tmp);
+
+	tmp = RREG32_SOC15_OFFSET(MMHUB, 0, mmVML2PF0_VM_L2_CNTL2,
+				  hubid * MMHUB_INSTANCE_REGISTER_OFFSET);
+	tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_CNTL2,
+			    INVALIDATE_ALL_L1_TLBS, 1);
+	tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_CNTL2,
+			    INVALIDATE_L2_CACHE, 1);
+	WREG32_SOC15_OFFSET(MMHUB, 0, mmVML2PF0_VM_L2_CNTL2,
+			    hubid * MMHUB_INSTANCE_REGISTER_OFFSET, tmp);
+
+	tmp = mmVML2PF0_VM_L2_CNTL3_DEFAULT;
+	if (adev->gmc.translate_further) {
+		tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_CNTL3, BANK_SELECT, 12);
+		tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+	} else {
+		tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_CNTL3, BANK_SELECT, 9);
+		tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+	}
+	WREG32_SOC15_OFFSET(MMHUB, 0, mmVML2PF0_VM_L2_CNTL3,
+			    hubid * MMHUB_INSTANCE_REGISTER_OFFSET, tmp);
+
+	tmp = mmVML2PF0_VM_L2_CNTL4_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_CNTL4,
+			    VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_CNTL4,
+			    VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
+	WREG32_SOC15_OFFSET(MMHUB, 0, mmVML2PF0_VM_L2_CNTL4,
+			    hubid * MMHUB_INSTANCE_REGISTER_OFFSET, tmp);
+}
+
+static void mmhub_v9_4_enable_system_domain(struct amdgpu_device *adev,
+					    int hubid)
+{
+	uint32_t tmp;
+
+	tmp = RREG32_SOC15_OFFSET(MMHUB, 0, mmVML2VC0_VM_CONTEXT0_CNTL,
+				  hubid * MMHUB_INSTANCE_REGISTER_OFFSET);
+	tmp = REG_SET_FIELD(tmp, VML2VC0_VM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, VML2VC0_VM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	tmp = REG_SET_FIELD(tmp, VML2VC0_VM_CONTEXT0_CNTL,
+			    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
+	WREG32_SOC15_OFFSET(MMHUB, 0, mmVML2VC0_VM_CONTEXT0_CNTL,
+			    hubid * MMHUB_INSTANCE_REGISTER_OFFSET, tmp);
+}
+
+static void mmhub_v9_4_disable_identity_aperture(struct amdgpu_device *adev,
+						 int hubid)
+{
+	WREG32_SOC15_OFFSET(MMHUB, 0,
+		    mmVML2PF0_VM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32,
+		    hubid * MMHUB_INSTANCE_REGISTER_OFFSET, 0XFFFFFFFF);
+	WREG32_SOC15_OFFSET(MMHUB, 0,
+		    mmVML2PF0_VM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32,
+		    hubid * MMHUB_INSTANCE_REGISTER_OFFSET, 0x0000000F);
+
+	WREG32_SOC15_OFFSET(MMHUB, 0,
+		    mmVML2PF0_VM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32,
+		    hubid * MMHUB_INSTANCE_REGISTER_OFFSET, 0);
+	WREG32_SOC15_OFFSET(MMHUB, 0,
+		    mmVML2PF0_VM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32,
+		    hubid * MMHUB_INSTANCE_REGISTER_OFFSET, 0);
+
+	WREG32_SOC15_OFFSET(MMHUB, 0,
+		    mmVML2PF0_VM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32,
+		    hubid * MMHUB_INSTANCE_REGISTER_OFFSET, 0);
+	WREG32_SOC15_OFFSET(MMHUB, 0,
+		    mmVML2PF0_VM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32,
+		    hubid * MMHUB_INSTANCE_REGISTER_OFFSET, 0);
+}
+
+static void mmhub_v9_4_setup_vmid_config(struct amdgpu_device *adev, int hubid)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	unsigned int num_level, block_size;
+	uint32_t tmp;
+	int i;
+
+	num_level = adev->vm_manager.num_level;
+	block_size = adev->vm_manager.block_size;
+	if (adev->gmc.translate_further)
+		num_level -= 1;
+	else
+		block_size -= 9;
+
+	for (i = 0; i <= 14; i++) {
+		tmp = RREG32_SOC15_OFFSET(MMHUB, 0, mmVML2VC0_VM_CONTEXT1_CNTL,
+				hubid * MMHUB_INSTANCE_REGISTER_OFFSET + i * hub->ctx_distance);
+		tmp = REG_SET_FIELD(tmp, VML2VC0_VM_CONTEXT1_CNTL,
+				    ENABLE_CONTEXT, 1);
+		tmp = REG_SET_FIELD(tmp, VML2VC0_VM_CONTEXT1_CNTL,
+				    PAGE_TABLE_DEPTH,
+				    num_level);
+		tmp = REG_SET_FIELD(tmp, VML2VC0_VM_CONTEXT1_CNTL,
+				    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VML2VC0_VM_CONTEXT1_CNTL,
+				    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    1);
+		tmp = REG_SET_FIELD(tmp, VML2VC0_VM_CONTEXT1_CNTL,
+				    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VML2VC0_VM_CONTEXT1_CNTL,
+				    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VML2VC0_VM_CONTEXT1_CNTL,
+				    READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VML2VC0_VM_CONTEXT1_CNTL,
+				    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VML2VC0_VM_CONTEXT1_CNTL,
+				    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VML2VC0_VM_CONTEXT1_CNTL,
+				    PAGE_TABLE_BLOCK_SIZE,
+				    block_size);
+		/* Send no-retry XNACK on fault to suppress VM fault storm. */
+		tmp = REG_SET_FIELD(tmp, VML2VC0_VM_CONTEXT1_CNTL,
+				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
+				    !adev->gmc.noretry);
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmVML2VC0_VM_CONTEXT1_CNTL,
+				    hubid * MMHUB_INSTANCE_REGISTER_OFFSET +
+				    i * hub->ctx_distance, tmp);
+		WREG32_SOC15_OFFSET(MMHUB, 0,
+			    mmVML2VC0_VM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,
+			    hubid * MMHUB_INSTANCE_REGISTER_OFFSET +
+			    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0,
+			    mmVML2VC0_VM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32,
+			    hubid * MMHUB_INSTANCE_REGISTER_OFFSET +
+			    i * hub->ctx_addr_distance, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0,
+				mmVML2VC0_VM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32,
+				hubid * MMHUB_INSTANCE_REGISTER_OFFSET +
+				i * hub->ctx_addr_distance,
+				lower_32_bits(adev->vm_manager.max_pfn - 1));
+		WREG32_SOC15_OFFSET(MMHUB, 0,
+				mmVML2VC0_VM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32,
+				hubid * MMHUB_INSTANCE_REGISTER_OFFSET +
+				i * hub->ctx_addr_distance,
+				upper_32_bits(adev->vm_manager.max_pfn - 1));
+	}
+}
+
+static void mmhub_v9_4_program_invalidation(struct amdgpu_device *adev,
+					    int hubid)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	unsigned i;
+
+	for (i = 0; i < 18; ++i) {
+		WREG32_SOC15_OFFSET(MMHUB, 0,
+				mmVML2VC0_VM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
+				hubid * MMHUB_INSTANCE_REGISTER_OFFSET +
+				i * hub->eng_addr_distance,
+				0xffffffff);
+		WREG32_SOC15_OFFSET(MMHUB, 0,
+				mmVML2VC0_VM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
+				hubid * MMHUB_INSTANCE_REGISTER_OFFSET +
+				i * hub->eng_addr_distance,
+				0x1f);
+	}
+}
+
+static int mmhub_v9_4_gart_enable(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < MMHUB_NUM_INSTANCES; i++) {
+		/* GART Enable. */
+		mmhub_v9_4_init_gart_aperture_regs(adev, i);
+		mmhub_v9_4_init_system_aperture_regs(adev, i);
+		mmhub_v9_4_init_tlb_regs(adev, i);
+		if (!amdgpu_sriov_vf(adev))
+			mmhub_v9_4_init_cache_regs(adev, i);
+
+		mmhub_v9_4_init_snoop_override_regs(adev, i);
+		mmhub_v9_4_enable_system_domain(adev, i);
+		if (!amdgpu_sriov_vf(adev))
+			mmhub_v9_4_disable_identity_aperture(adev, i);
+		mmhub_v9_4_setup_vmid_config(adev, i);
+		mmhub_v9_4_program_invalidation(adev, i);
+	}
+
+	return 0;
+}
+
+static void mmhub_v9_4_gart_disable(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
+	u32 tmp;
+	u32 i, j;
+
+	for (j = 0; j < MMHUB_NUM_INSTANCES; j++) {
+		/* Disable all tables */
+		for (i = 0; i < AMDGPU_NUM_VMID; i++)
+			WREG32_SOC15_OFFSET(MMHUB, 0,
+					    mmVML2VC0_VM_CONTEXT0_CNTL,
+					    j * MMHUB_INSTANCE_REGISTER_OFFSET +
+					    i * hub->ctx_distance, 0);
+
+		/* Setup TLB control */
+		tmp = RREG32_SOC15_OFFSET(MMHUB, 0,
+				   mmVMSHAREDVC0_MC_VM_MX_L1_TLB_CNTL,
+				   j * MMHUB_INSTANCE_REGISTER_OFFSET);
+		tmp = REG_SET_FIELD(tmp, VMSHAREDVC0_MC_VM_MX_L1_TLB_CNTL,
+				    ENABLE_L1_TLB, 0);
+		tmp = REG_SET_FIELD(tmp,
+				    VMSHAREDVC0_MC_VM_MX_L1_TLB_CNTL,
+				    ENABLE_ADVANCED_DRIVER_MODEL, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0,
+				    mmVMSHAREDVC0_MC_VM_MX_L1_TLB_CNTL,
+				    j * MMHUB_INSTANCE_REGISTER_OFFSET, tmp);
+
+		/* Setup L2 cache */
+		tmp = RREG32_SOC15_OFFSET(MMHUB, 0, mmVML2PF0_VM_L2_CNTL,
+					  j * MMHUB_INSTANCE_REGISTER_OFFSET);
+		tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_CNTL,
+				    ENABLE_L2_CACHE, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmVML2PF0_VM_L2_CNTL,
+				    j * MMHUB_INSTANCE_REGISTER_OFFSET, tmp);
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmVML2PF0_VM_L2_CNTL3,
+				    j * MMHUB_INSTANCE_REGISTER_OFFSET, 0);
+	}
+}
+
+/**
+ * mmhub_v9_4_set_fault_enable_default - update GART/VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+static void mmhub_v9_4_set_fault_enable_default(struct amdgpu_device *adev, bool value)
+{
+	u32 tmp;
+	int i;
+
+	for (i = 0; i < MMHUB_NUM_INSTANCES; i++) {
+		tmp = RREG32_SOC15_OFFSET(MMHUB, 0,
+					  mmVML2PF0_VM_L2_PROTECTION_FAULT_CNTL,
+					  i * MMHUB_INSTANCE_REGISTER_OFFSET);
+		tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_PROTECTION_FAULT_CNTL,
+				    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    value);
+		tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_PROTECTION_FAULT_CNTL,
+				    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    value);
+		tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_PROTECTION_FAULT_CNTL,
+				    PDE1_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    value);
+		tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_PROTECTION_FAULT_CNTL,
+				    PDE2_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    value);
+		tmp = REG_SET_FIELD(tmp,
+			    VML2PF0_VM_L2_PROTECTION_FAULT_CNTL,
+			    TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT,
+			    value);
+		tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_PROTECTION_FAULT_CNTL,
+				    NACK_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    value);
+		tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_PROTECTION_FAULT_CNTL,
+				    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    value);
+		tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_PROTECTION_FAULT_CNTL,
+				    VALID_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    value);
+		tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_PROTECTION_FAULT_CNTL,
+				    READ_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    value);
+		tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_PROTECTION_FAULT_CNTL,
+				    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    value);
+		tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_PROTECTION_FAULT_CNTL,
+				    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    value);
+		if (!value) {
+			tmp = REG_SET_FIELD(tmp,
+					    VML2PF0_VM_L2_PROTECTION_FAULT_CNTL,
+					    CRASH_ON_NO_RETRY_FAULT, 1);
+			tmp = REG_SET_FIELD(tmp,
+					    VML2PF0_VM_L2_PROTECTION_FAULT_CNTL,
+					    CRASH_ON_RETRY_FAULT, 1);
+		}
+
+		WREG32_SOC15_OFFSET(MMHUB, 0,
+				    mmVML2PF0_VM_L2_PROTECTION_FAULT_CNTL,
+				    i * MMHUB_INSTANCE_REGISTER_OFFSET, tmp);
+	}
+}
+
+static void mmhub_v9_4_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub[MMHUB_NUM_INSTANCES] = {
+		&adev->vmhub[AMDGPU_MMHUB0(0)], &adev->vmhub[AMDGPU_MMHUB1(0)]};
+	int i;
+
+	for (i = 0; i < MMHUB_NUM_INSTANCES; i++) {
+		hub[i]->ctx0_ptb_addr_lo32 =
+			SOC15_REG_OFFSET(MMHUB, 0,
+			    mmVML2VC0_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32) +
+			    i * MMHUB_INSTANCE_REGISTER_OFFSET;
+		hub[i]->ctx0_ptb_addr_hi32 =
+			SOC15_REG_OFFSET(MMHUB, 0,
+			    mmVML2VC0_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32) +
+			    i * MMHUB_INSTANCE_REGISTER_OFFSET;
+		hub[i]->vm_inv_eng0_sem =
+			SOC15_REG_OFFSET(MMHUB, 0,
+					 mmVML2VC0_VM_INVALIDATE_ENG0_SEM) +
+					 i * MMHUB_INSTANCE_REGISTER_OFFSET;
+		hub[i]->vm_inv_eng0_req =
+			SOC15_REG_OFFSET(MMHUB, 0,
+					 mmVML2VC0_VM_INVALIDATE_ENG0_REQ) +
+					 i * MMHUB_INSTANCE_REGISTER_OFFSET;
+		hub[i]->vm_inv_eng0_ack =
+			SOC15_REG_OFFSET(MMHUB, 0,
+					 mmVML2VC0_VM_INVALIDATE_ENG0_ACK) +
+					 i * MMHUB_INSTANCE_REGISTER_OFFSET;
+		hub[i]->vm_context0_cntl =
+			SOC15_REG_OFFSET(MMHUB, 0,
+					 mmVML2VC0_VM_CONTEXT0_CNTL) +
+					 i * MMHUB_INSTANCE_REGISTER_OFFSET;
+		hub[i]->vm_l2_pro_fault_status =
+			SOC15_REG_OFFSET(MMHUB, 0,
+				    mmVML2PF0_VM_L2_PROTECTION_FAULT_STATUS) +
+				    i * MMHUB_INSTANCE_REGISTER_OFFSET;
+		hub[i]->vm_l2_pro_fault_cntl =
+			SOC15_REG_OFFSET(MMHUB, 0,
+				    mmVML2PF0_VM_L2_PROTECTION_FAULT_CNTL) +
+				    i * MMHUB_INSTANCE_REGISTER_OFFSET;
+
+		hub[i]->ctx_distance = mmVML2VC0_VM_CONTEXT1_CNTL -
+			mmVML2VC0_VM_CONTEXT0_CNTL;
+		hub[i]->ctx_addr_distance = mmVML2VC0_VM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 -
+			mmVML2VC0_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32;
+		hub[i]->eng_distance = mmVML2VC0_VM_INVALIDATE_ENG1_REQ -
+			mmVML2VC0_VM_INVALIDATE_ENG0_REQ;
+		hub[i]->eng_addr_distance = mmVML2VC0_VM_INVALIDATE_ENG1_ADDR_RANGE_LO32 -
+			mmVML2VC0_VM_INVALIDATE_ENG0_ADDR_RANGE_LO32;
+	}
+}
+
+static void mmhub_v9_4_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+							bool enable)
+{
+	uint32_t def, data, def1, data1;
+	int i, j;
+	int dist = mmDAGB1_CNTL_MISC2 - mmDAGB0_CNTL_MISC2;
+
+	for (i = 0; i < MMHUB_NUM_INSTANCES; i++) {
+		def = data = RREG32_SOC15_OFFSET(MMHUB, 0,
+					mmATCL2_0_ATC_L2_MISC_CG,
+					i * MMHUB_INSTANCE_REGISTER_OFFSET);
+
+		if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG))
+			data |= ATCL2_0_ATC_L2_MISC_CG__ENABLE_MASK;
+		else
+			data &= ~ATCL2_0_ATC_L2_MISC_CG__ENABLE_MASK;
+
+		if (def != data)
+			WREG32_SOC15_OFFSET(MMHUB, 0, mmATCL2_0_ATC_L2_MISC_CG,
+				i * MMHUB_INSTANCE_REGISTER_OFFSET, data);
+
+		for (j = 0; j < 5; j++) {
+			def1 = data1 = RREG32_SOC15_OFFSET(MMHUB, 0,
+					mmDAGB0_CNTL_MISC2,
+					i * MMHUB_INSTANCE_REGISTER_OFFSET +
+					j * dist);
+			if (enable &&
+			    (adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG)) {
+				data1 &=
+				    ~(DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+				    DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+				    DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+				    DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+				    DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+				    DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK);
+			} else {
+				data1 |=
+				    (DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+				    DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+				    DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+				    DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+				    DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+				    DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK);
+			}
+
+			if (def1 != data1)
+				WREG32_SOC15_OFFSET(MMHUB, 0,
+					mmDAGB0_CNTL_MISC2,
+					i * MMHUB_INSTANCE_REGISTER_OFFSET +
+					j * dist, data1);
+
+			if (i == 1 && j == 3)
+				break;
+		}
+	}
+}
+
+static void mmhub_v9_4_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+						       bool enable)
+{
+	uint32_t def, data;
+	int i;
+
+	for (i = 0; i < MMHUB_NUM_INSTANCES; i++) {
+		def = data = RREG32_SOC15_OFFSET(MMHUB, 0,
+					mmATCL2_0_ATC_L2_MISC_CG,
+					i * MMHUB_INSTANCE_REGISTER_OFFSET);
+
+		if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_LS))
+			data |= ATCL2_0_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK;
+		else
+			data &= ~ATCL2_0_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK;
+
+		if (def != data)
+			WREG32_SOC15_OFFSET(MMHUB, 0, mmATCL2_0_ATC_L2_MISC_CG,
+				i * MMHUB_INSTANCE_REGISTER_OFFSET, data);
+	}
+}
+
+static int mmhub_v9_4_set_clockgating(struct amdgpu_device *adev,
+			       enum amd_clockgating_state state)
+{
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (adev->asic_type) {
+	case CHIP_ARCTURUS:
+		mmhub_v9_4_update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		mmhub_v9_4_update_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static void mmhub_v9_4_get_clockgating(struct amdgpu_device *adev, u64 *flags)
+{
+	u32 data, data1;
+
+	if (amdgpu_sriov_vf(adev))
+		*flags = 0;
+
+	/* AMD_CG_SUPPORT_MC_MGCG */
+	data = RREG32_SOC15(MMHUB, 0, mmATCL2_0_ATC_L2_MISC_CG);
+
+	data1 = RREG32_SOC15(MMHUB, 0, mmATCL2_0_ATC_L2_MISC_CG);
+
+	if ((data & ATCL2_0_ATC_L2_MISC_CG__ENABLE_MASK) &&
+	    !(data1 & (DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+		       DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+		       DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+		       DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+		       DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+		       DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK)))
+		*flags |= AMD_CG_SUPPORT_MC_MGCG;
+
+	/* AMD_CG_SUPPORT_MC_LS */
+	if (data & ATCL2_0_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_MC_LS;
+}
+
+static const struct soc15_ras_field_entry mmhub_v9_4_ras_fields[] = {
+	/* MMHUB Range 0 */
+	{ "MMEA0_DRAMRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA0_DRAMWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA0_DRAMWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA0_RRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, RRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, RRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA0_WRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, WRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, WRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA0_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA0_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA0_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, IORD_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA0_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, IOWR_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA0_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT, IOWR_DATAMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA0_GMIRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA0_GMIWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA0_GMIWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA0_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA0_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA0_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA0_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA0_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA0_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA0_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA0_EDC_CNT3, IORD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA0_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA0_EDC_CNT3, IOWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA0_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA0_EDC_CNT3, IOWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA0_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA0_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA0_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA0_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA0_MAM_D0MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D0MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D0MEM_DED_COUNT),
+	},
+	{ "MMEA0_MAM_D1MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D1MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D1MEM_DED_COUNT),
+	},
+	{ "MMEA0_MAM_D2MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D2MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D2MEM_DED_COUNT),
+	},
+	{ "MMEA0_MAM_D3MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D3MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D3MEM_DED_COUNT),
+	},
+
+	/* MMHUB Range 1 */
+	{ "MMEA1_DRAMRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA1_DRAMWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA1_DRAMWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA1_RRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, RRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, RRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA1_WRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, WRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, WRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA1_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA1_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA1_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, IORD_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA1_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, IOWR_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA1_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT, IOWR_DATAMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA1_GMIRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA1_GMIWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA1_GMIWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA1_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA1_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA1_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA1_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA1_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA1_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA1_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA1_EDC_CNT3, IORD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA1_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA1_EDC_CNT3, IOWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA1_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA1_EDC_CNT3, IOWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA1_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA1_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA1_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA1_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA1_MAM_D0MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D0MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D0MEM_DED_COUNT),
+	},
+	{ "MMEA1_MAM_D1MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D1MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D1MEM_DED_COUNT),
+	},
+	{ "MMEA1_MAM_D2MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D2MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D2MEM_DED_COUNT),
+	},
+	{ "MMEA1_MAM_D3MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D3MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D3MEM_DED_COUNT),
+	},
+
+	/* MMHAB Range 2*/
+	{ "MMEA2_DRAMRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA2_DRAMWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA2_DRAMWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA2_RRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, RRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, RRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA2_WRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, WRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, WRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA2_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA2_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA2_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, IORD_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA2_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, IOWR_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA2_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT, IOWR_DATAMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA2_GMIRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA2_GMIWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA2_GMIWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA2_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA2_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA2_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA2_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA2_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA2_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA2_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA2_EDC_CNT3, IORD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA2_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA2_EDC_CNT3, IOWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA2_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA2_EDC_CNT3, IOWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA2_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA2_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA2_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA2_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA2_MAM_D0MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, MAM_D0MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, MAM_D0MEM_DED_COUNT),
+	},
+	{ "MMEA2_MAM_D1MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, MAM_D1MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, MAM_D1MEM_DED_COUNT),
+	},
+	{ "MMEA2_MAM_D2MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, MAM_D2MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, MAM_D2MEM_DED_COUNT),
+	},
+	{ "MMEA2_MAM_D3MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, MAM_D3MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA2_EDC_CNT2, MAM_D3MEM_DED_COUNT),
+	},
+
+	/* MMHUB Rang 3 */
+	{ "MMEA3_DRAMRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA3_DRAMWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA3_DRAMWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA3_RRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, RRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, RRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA3_WRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, WRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, WRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA3_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA3_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA3_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, IORD_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA3_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, IOWR_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA3_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT, IOWR_DATAMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA3_GMIRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA3_GMIWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA3_GMIWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA3_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA3_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA3_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA3_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA3_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA3_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA3_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA3_EDC_CNT3, IORD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA3_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA3_EDC_CNT3, IOWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA3_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA3_EDC_CNT3, IOWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA3_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA3_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA3_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA3_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA3_MAM_D0MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D0MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D0MEM_DED_COUNT),
+	},
+	{ "MMEA3_MAM_D1MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D1MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D1MEM_DED_COUNT),
+	},
+	{ "MMEA3_MAM_D2MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D2MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D2MEM_DED_COUNT),
+	},
+	{ "MMEA3_MAM_D3MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D3MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D3MEM_DED_COUNT),
+	},
+
+	/* MMHUB Range 4 */
+	{ "MMEA4_DRAMRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA4_DRAMWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA4_DRAMWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA4_RRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, RRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, RRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA4_WRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, WRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, WRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA4_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA4_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA4_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, IORD_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA4_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, IOWR_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA4_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT, IOWR_DATAMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA4_GMIRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA4_GMIWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA4_GMIWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA4_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA4_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA4_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA4_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA4_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA4_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA4_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA4_EDC_CNT3, IORD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA4_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA4_EDC_CNT3, IOWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA4_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA4_EDC_CNT3, IOWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA4_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA4_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA4_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA4_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA4_MAM_D0MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D0MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D0MEM_DED_COUNT),
+	},
+	{ "MMEA4_MAM_D1MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D1MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D1MEM_DED_COUNT),
+	},
+	{ "MMEA4_MAM_D2MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D2MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D2MEM_DED_COUNT),
+	},
+	{ "MMEA4_MAM_D3MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D3MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D3MEM_DED_COUNT),
+	},
+
+	/* MMHUAB Range 5 */
+	{ "MMEA5_DRAMRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA5_DRAMWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA5_DRAMWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA5_RRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, RRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, RRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA5_WRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, WRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, WRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA5_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA5_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA5_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, IORD_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA5_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, IOWR_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA5_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT, IOWR_DATAMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA5_GMIRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA5_GMIWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA5_GMIWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA5_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA5_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA5_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA5_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA5_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA5_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA5_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA5_EDC_CNT3, IORD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA5_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA5_EDC_CNT3, IOWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA5_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA5_EDC_CNT3, IOWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA5_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA5_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA5_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA5_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA5_MAM_D0MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D0MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D0MEM_DED_COUNT),
+	},
+	{ "MMEA5_MAM_D1MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D1MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D1MEM_DED_COUNT),
+	},
+	{ "MMEA5_MAM_D2MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D2MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D2MEM_DED_COUNT),
+	},
+	{ "MMEA5_MAM_D3MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D3MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D3MEM_DED_COUNT),
+	},
+
+	/* MMHUB Range 6 */
+	{ "MMEA6_DRAMRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA6_DRAMWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA6_DRAMWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA6_RRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT, RRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT, RRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA6_WRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT, WRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT, WRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA6_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA6_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA6_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT, IORD_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA6_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT, IOWR_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA6_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT, IOWR_DATAMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA6_GMIRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA6_GMIWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA6_GMIWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA6_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA6_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA6_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA6_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA6_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA6_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA6_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA6_EDC_CNT3, IORD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA6_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA6_EDC_CNT3, IOWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA6_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA6_EDC_CNT3, IOWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA6_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA6_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA6_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA6_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA6_MAM_D0MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT2, MAM_D0MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT2, MAM_D0MEM_DED_COUNT),
+	},
+	{ "MMEA6_MAM_D1MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT2, MAM_D1MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT2, MAM_D1MEM_DED_COUNT),
+	},
+	{ "MMEA6_MAM_D2MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT2, MAM_D2MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT2, MAM_D2MEM_DED_COUNT),
+	},
+	{ "MMEA6_MAM_D3MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT2, MAM_D3MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA6_EDC_CNT2, MAM_D3MEM_DED_COUNT),
+	},
+
+	/* MMHUB Range 7*/
+	{ "MMEA7_DRAMRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA7_DRAMWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA7_DRAMWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA7_RRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT, RRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT, RRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA7_WRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT, WRET_TAGMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT, WRET_TAGMEM_DED_COUNT),
+	},
+	{ "MMEA7_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA7_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA7_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT, IORD_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA7_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT, IOWR_CMDMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA7_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT, IOWR_DATAMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA7_GMIRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA7_GMIWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA7_GMIWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA7_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA7_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT),
+	0, 0,
+	},
+	{ "MMEA7_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA7_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA7_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA7_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA7_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA7_EDC_CNT3, IORD_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA7_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA7_EDC_CNT3, IOWR_CMDMEM_DED_COUNT),
+	},
+	{ "MMEA7_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA7_EDC_CNT3, IOWR_DATAMEM_DED_COUNT),
+	},
+	{ "MMEA7_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA7_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA7_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT3),
+	0, 0,
+	SOC15_REG_FIELD(MMEA7_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT),
+	},
+	{ "MMEA7_MAM_D0MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT2, MAM_D0MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT2, MAM_D0MEM_DED_COUNT),
+	},
+	{ "MMEA7_MAM_D1MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT2, MAM_D1MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT2, MAM_D1MEM_DED_COUNT),
+	},
+	{ "MMEA7_MAM_D2MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT2, MAM_D2MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT2, MAM_D2MEM_DED_COUNT),
+	},
+	{ "MMEA7_MAM_D3MEM", SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT2),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT2, MAM_D3MEM_SED_COUNT),
+	SOC15_REG_FIELD(MMEA7_EDC_CNT2, MAM_D3MEM_DED_COUNT),
+	}
+};
+
+static const struct soc15_reg_entry mmhub_v9_4_edc_cnt_regs[] = {
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT2), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_EDC_CNT3), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT2), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_EDC_CNT3), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT2), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_EDC_CNT3), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT2), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_EDC_CNT3), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT2), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_EDC_CNT3), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT2), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_EDC_CNT3), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT2), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_EDC_CNT3), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT2), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_EDC_CNT3), 0, 0, 0 },
+};
+
+static int mmhub_v9_4_get_ras_error_count(struct amdgpu_device *adev,
+					  const struct soc15_reg_entry *reg,
+					  uint32_t value,
+					  uint32_t *sec_count,
+					  uint32_t *ded_count)
+{
+	uint32_t i;
+	uint32_t sec_cnt, ded_cnt;
+
+	for (i = 0; i < ARRAY_SIZE(mmhub_v9_4_ras_fields); i++) {
+		if (mmhub_v9_4_ras_fields[i].reg_offset != reg->reg_offset)
+			continue;
+
+		sec_cnt = (value &
+				mmhub_v9_4_ras_fields[i].sec_count_mask) >>
+				mmhub_v9_4_ras_fields[i].sec_count_shift;
+		if (sec_cnt) {
+			dev_info(adev->dev, "MMHUB SubBlock %s, SEC %d\n",
+				mmhub_v9_4_ras_fields[i].name,
+				sec_cnt);
+			*sec_count += sec_cnt;
+		}
+
+		ded_cnt = (value &
+				mmhub_v9_4_ras_fields[i].ded_count_mask) >>
+				mmhub_v9_4_ras_fields[i].ded_count_shift;
+		if (ded_cnt) {
+			dev_info(adev->dev, "MMHUB SubBlock %s, DED %d\n",
+				mmhub_v9_4_ras_fields[i].name,
+				ded_cnt);
+			*ded_count += ded_cnt;
+		}
+	}
+
+	return 0;
+}
+
+static void mmhub_v9_4_query_ras_error_count(struct amdgpu_device *adev,
+					   void *ras_error_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+	uint32_t sec_count = 0, ded_count = 0;
+	uint32_t i;
+	uint32_t reg_value;
+
+	err_data->ue_count = 0;
+	err_data->ce_count = 0;
+
+	for (i = 0; i < ARRAY_SIZE(mmhub_v9_4_edc_cnt_regs); i++) {
+		reg_value =
+			RREG32(SOC15_REG_ENTRY_OFFSET(mmhub_v9_4_edc_cnt_regs[i]));
+		if (reg_value)
+			mmhub_v9_4_get_ras_error_count(adev, &mmhub_v9_4_edc_cnt_regs[i],
+				reg_value, &sec_count, &ded_count);
+	}
+
+	err_data->ce_count += sec_count;
+	err_data->ue_count += ded_count;
+}
+
+static void mmhub_v9_4_reset_ras_error_count(struct amdgpu_device *adev)
+{
+	uint32_t i;
+
+	/* read back edc counter registers to reset the counters to 0 */
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__MMHUB)) {
+		for (i = 0; i < ARRAY_SIZE(mmhub_v9_4_edc_cnt_regs); i++)
+			RREG32(SOC15_REG_ENTRY_OFFSET(mmhub_v9_4_edc_cnt_regs[i]));
+	}
+}
+
+static const struct soc15_reg_entry mmhub_v9_4_err_status_regs[] = {
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_ERR_STATUS), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_ERR_STATUS), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_ERR_STATUS), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_ERR_STATUS), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_ERR_STATUS), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_ERR_STATUS), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_ERR_STATUS), 0, 0, 0 },
+	{ SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_ERR_STATUS), 0, 0, 0 },
+};
+
+static void mmhub_v9_4_query_ras_error_status(struct amdgpu_device *adev)
+{
+	int i;
+	uint32_t reg_value;
+
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__MMHUB))
+		return;
+
+	for (i = 0; i < ARRAY_SIZE(mmhub_v9_4_err_status_regs); i++) {
+		reg_value =
+			RREG32(SOC15_REG_ENTRY_OFFSET(mmhub_v9_4_err_status_regs[i]));
+		if (REG_GET_FIELD(reg_value, MMEA0_ERR_STATUS, SDP_RDRSP_STATUS) ||
+		    REG_GET_FIELD(reg_value, MMEA0_ERR_STATUS, SDP_WRRSP_STATUS) ||
+		    REG_GET_FIELD(reg_value, MMEA0_ERR_STATUS, SDP_RDRSP_DATAPARITY_ERROR)) {
+			/* SDP read/write error/parity error in FUE_IS_FATAL mode
+			 * can cause system fatal error in arcturas. Harvest the error
+			 * status before GPU reset */
+			dev_warn(adev->dev, "MMHUB EA err detected at instance: %d, status: 0x%x!\n",
+					i, reg_value);
+		}
+	}
+}
+
+const struct amdgpu_ras_block_hw_ops mmhub_v9_4_ras_hw_ops = {
+	.query_ras_error_count = mmhub_v9_4_query_ras_error_count,
+	.reset_ras_error_count = mmhub_v9_4_reset_ras_error_count,
+	.query_ras_error_status = mmhub_v9_4_query_ras_error_status,
+};
+
+struct amdgpu_mmhub_ras mmhub_v9_4_ras = {
+	.ras_block = {
+		.hw_ops = &mmhub_v9_4_ras_hw_ops,
+	},
+};
+
+const struct amdgpu_mmhub_funcs mmhub_v9_4_funcs = {
+	.get_fb_location = mmhub_v9_4_get_fb_location,
+	.init = mmhub_v9_4_init,
+	.gart_enable = mmhub_v9_4_gart_enable,
+	.set_fault_enable_default = mmhub_v9_4_set_fault_enable_default,
+	.gart_disable = mmhub_v9_4_gart_disable,
+	.set_clockgating = mmhub_v9_4_set_clockgating,
+	.get_clockgating = mmhub_v9_4_get_clockgating,
+	.setup_vm_pt_regs = mmhub_v9_4_setup_vm_pt_regs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.h
new file mode 100644
index 000000000000..a48329d95f71
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __MMHUB_V9_4_H__
+#define __MMHUB_V9_4_H__
+
+extern const struct amdgpu_mmhub_funcs mmhub_v9_4_funcs;
+extern struct amdgpu_mmhub_ras mmhub_v9_4_ras;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/mmsch_v1_0.h b/drivers/gpu/drm/amd/amdgpu/mmsch_v1_0.h
index 8af0bddf85e4..2cdab8062c86 100644
--- a/drivers/gpu/drm/amd/amdgpu/mmsch_v1_0.h
+++ b/drivers/gpu/drm/amd/amdgpu/mmsch_v1_0.h
@@ -24,9 +24,7 @@
 #ifndef __MMSCH_V1_0_H__
 #define __MMSCH_V1_0_H__
 
-#define MMSCH_VERSION_MAJOR	1
-#define MMSCH_VERSION_MINOR	0
-#define MMSCH_VERSION	(MMSCH_VERSION_MAJOR << 16 | MMSCH_VERSION_MINOR)
+#define MMSCH_VERSION	0x1
 
 enum mmsch_v1_0_command_type {
 	MMSCH_COMMAND__DIRECT_REG_WRITE = 0,
@@ -47,6 +45,18 @@ struct mmsch_v1_0_init_header {
 	uint32_t uvd_table_size;
 };
 
+struct mmsch_vf_eng_init_header {
+	uint32_t init_status;
+	uint32_t table_offset;
+	uint32_t table_size;
+};
+
+struct mmsch_v1_1_init_header {
+	uint32_t version;
+	uint32_t total_size;
+	struct mmsch_vf_eng_init_header eng[2];
+};
+
 struct mmsch_v1_0_cmd_direct_reg_header {
 	uint32_t reg_offset   : 28;
 	uint32_t command_type : 4;
diff --git a/drivers/gpu/drm/amd/amdgpu/mmsch_v2_0.h b/drivers/gpu/drm/amd/amdgpu/mmsch_v2_0.h
new file mode 100644
index 000000000000..1b5086c7d4e6
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmsch_v2_0.h
@@ -0,0 +1,338 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __MMSCH_V2_0_H__
+#define __MMSCH_V2_0_H__
+
+// addressBlock: uvd0_mmsch_dec
+// base address: 0x1e000
+#define mmMMSCH_UCODE_ADDR                                                                             0x0000
+#define mmMMSCH_UCODE_ADDR_BASE_IDX                                                                    0
+#define mmMMSCH_UCODE_DATA                                                                             0x0001
+#define mmMMSCH_UCODE_DATA_BASE_IDX                                                                    0
+#define mmMMSCH_SRAM_ADDR                                                                              0x0002
+#define mmMMSCH_SRAM_ADDR_BASE_IDX                                                                     0
+#define mmMMSCH_SRAM_DATA                                                                              0x0003
+#define mmMMSCH_SRAM_DATA_BASE_IDX                                                                     0
+#define mmMMSCH_VF_SRAM_OFFSET                                                                         0x0004
+#define mmMMSCH_VF_SRAM_OFFSET_BASE_IDX                                                                0
+#define mmMMSCH_DB_SRAM_OFFSET                                                                         0x0005
+#define mmMMSCH_DB_SRAM_OFFSET_BASE_IDX                                                                0
+#define mmMMSCH_CTX_SRAM_OFFSET                                                                        0x0006
+#define mmMMSCH_CTX_SRAM_OFFSET_BASE_IDX                                                               0
+#define mmMMSCH_CTL                                                                                    0x0007
+#define mmMMSCH_CTL_BASE_IDX                                                                           0
+#define mmMMSCH_INTR                                                                                   0x0008
+#define mmMMSCH_INTR_BASE_IDX                                                                          0
+#define mmMMSCH_INTR_ACK                                                                               0x0009
+#define mmMMSCH_INTR_ACK_BASE_IDX                                                                      0
+#define mmMMSCH_INTR_STATUS                                                                            0x000a
+#define mmMMSCH_INTR_STATUS_BASE_IDX                                                                   0
+#define mmMMSCH_VF_VMID                                                                                0x000b
+#define mmMMSCH_VF_VMID_BASE_IDX                                                                       0
+#define mmMMSCH_VF_CTX_ADDR_LO                                                                         0x000c
+#define mmMMSCH_VF_CTX_ADDR_LO_BASE_IDX                                                                0
+#define mmMMSCH_VF_CTX_ADDR_HI                                                                         0x000d
+#define mmMMSCH_VF_CTX_ADDR_HI_BASE_IDX                                                                0
+#define mmMMSCH_VF_CTX_SIZE                                                                            0x000e
+#define mmMMSCH_VF_CTX_SIZE_BASE_IDX                                                                   0
+#define mmMMSCH_VF_GPCOM_ADDR_LO                                                                       0x000f
+#define mmMMSCH_VF_GPCOM_ADDR_LO_BASE_IDX                                                              0
+#define mmMMSCH_VF_GPCOM_ADDR_HI                                                                       0x0010
+#define mmMMSCH_VF_GPCOM_ADDR_HI_BASE_IDX                                                              0
+#define mmMMSCH_VF_GPCOM_SIZE                                                                          0x0011
+#define mmMMSCH_VF_GPCOM_SIZE_BASE_IDX                                                                 0
+#define mmMMSCH_VF_MAILBOX_HOST                                                                        0x0012
+#define mmMMSCH_VF_MAILBOX_HOST_BASE_IDX                                                               0
+#define mmMMSCH_VF_MAILBOX_RESP                                                                        0x0013
+#define mmMMSCH_VF_MAILBOX_RESP_BASE_IDX                                                               0
+#define mmMMSCH_VF_MAILBOX_0                                                                           0x0014
+#define mmMMSCH_VF_MAILBOX_0_BASE_IDX                                                                  0
+#define mmMMSCH_VF_MAILBOX_0_RESP                                                                      0x0015
+#define mmMMSCH_VF_MAILBOX_0_RESP_BASE_IDX                                                             0
+#define mmMMSCH_VF_MAILBOX_1                                                                           0x0016
+#define mmMMSCH_VF_MAILBOX_1_BASE_IDX                                                                  0
+#define mmMMSCH_VF_MAILBOX_1_RESP                                                                      0x0017
+#define mmMMSCH_VF_MAILBOX_1_RESP_BASE_IDX                                                             0
+#define mmMMSCH_CNTL                                                                                   0x001c
+#define mmMMSCH_CNTL_BASE_IDX                                                                          0
+#define mmMMSCH_NONCACHE_OFFSET0                                                                       0x001d
+#define mmMMSCH_NONCACHE_OFFSET0_BASE_IDX                                                              0
+#define mmMMSCH_NONCACHE_SIZE0                                                                         0x001e
+#define mmMMSCH_NONCACHE_SIZE0_BASE_IDX                                                                0
+#define mmMMSCH_NONCACHE_OFFSET1                                                                       0x001f
+#define mmMMSCH_NONCACHE_OFFSET1_BASE_IDX                                                              0
+#define mmMMSCH_NONCACHE_SIZE1                                                                         0x0020
+#define mmMMSCH_NONCACHE_SIZE1_BASE_IDX                                                                0
+#define mmMMSCH_PDEBUG_STATUS                                                                          0x0021
+#define mmMMSCH_PDEBUG_STATUS_BASE_IDX                                                                 0
+#define mmMMSCH_PDEBUG_DATA_32UPPERBITS                                                                0x0022
+#define mmMMSCH_PDEBUG_DATA_32UPPERBITS_BASE_IDX                                                       0
+#define mmMMSCH_PDEBUG_DATA_32LOWERBITS                                                                0x0023
+#define mmMMSCH_PDEBUG_DATA_32LOWERBITS_BASE_IDX                                                       0
+#define mmMMSCH_PDEBUG_EPC                                                                             0x0024
+#define mmMMSCH_PDEBUG_EPC_BASE_IDX                                                                    0
+#define mmMMSCH_PDEBUG_EXCCAUSE                                                                        0x0025
+#define mmMMSCH_PDEBUG_EXCCAUSE_BASE_IDX                                                               0
+#define mmMMSCH_PROC_STATE1                                                                            0x0026
+#define mmMMSCH_PROC_STATE1_BASE_IDX                                                                   0
+#define mmMMSCH_LAST_MC_ADDR                                                                           0x0027
+#define mmMMSCH_LAST_MC_ADDR_BASE_IDX                                                                  0
+#define mmMMSCH_LAST_MEM_ACCESS_HI                                                                     0x0028
+#define mmMMSCH_LAST_MEM_ACCESS_HI_BASE_IDX                                                            0
+#define mmMMSCH_LAST_MEM_ACCESS_LO                                                                     0x0029
+#define mmMMSCH_LAST_MEM_ACCESS_LO_BASE_IDX                                                            0
+#define mmMMSCH_IOV_ACTIVE_FCN_ID                                                                      0x002a
+#define mmMMSCH_IOV_ACTIVE_FCN_ID_BASE_IDX                                                             0
+#define mmMMSCH_SCRATCH_0                                                                              0x002b
+#define mmMMSCH_SCRATCH_0_BASE_IDX                                                                     0
+#define mmMMSCH_SCRATCH_1                                                                              0x002c
+#define mmMMSCH_SCRATCH_1_BASE_IDX                                                                     0
+#define mmMMSCH_GPUIOV_SCH_BLOCK_0                                                                     0x002d
+#define mmMMSCH_GPUIOV_SCH_BLOCK_0_BASE_IDX                                                            0
+#define mmMMSCH_GPUIOV_CMD_CONTROL_0                                                                   0x002e
+#define mmMMSCH_GPUIOV_CMD_CONTROL_0_BASE_IDX                                                          0
+#define mmMMSCH_GPUIOV_CMD_STATUS_0                                                                    0x002f
+#define mmMMSCH_GPUIOV_CMD_STATUS_0_BASE_IDX                                                           0
+#define mmMMSCH_GPUIOV_VM_BUSY_STATUS_0                                                                0x0030
+#define mmMMSCH_GPUIOV_VM_BUSY_STATUS_0_BASE_IDX                                                       0
+#define mmMMSCH_GPUIOV_ACTIVE_FCNS_0                                                                   0x0031
+#define mmMMSCH_GPUIOV_ACTIVE_FCNS_0_BASE_IDX                                                          0
+#define mmMMSCH_GPUIOV_ACTIVE_FCN_ID_0                                                                 0x0032
+#define mmMMSCH_GPUIOV_ACTIVE_FCN_ID_0_BASE_IDX                                                        0
+#define mmMMSCH_GPUIOV_DW6_0                                                                           0x0033
+#define mmMMSCH_GPUIOV_DW6_0_BASE_IDX                                                                  0
+#define mmMMSCH_GPUIOV_DW7_0                                                                           0x0034
+#define mmMMSCH_GPUIOV_DW7_0_BASE_IDX                                                                  0
+#define mmMMSCH_GPUIOV_DW8_0                                                                           0x0035
+#define mmMMSCH_GPUIOV_DW8_0_BASE_IDX                                                                  0
+#define mmMMSCH_GPUIOV_SCH_BLOCK_1                                                                     0x0036
+#define mmMMSCH_GPUIOV_SCH_BLOCK_1_BASE_IDX                                                            0
+#define mmMMSCH_GPUIOV_CMD_CONTROL_1                                                                   0x0037
+#define mmMMSCH_GPUIOV_CMD_CONTROL_1_BASE_IDX                                                          0
+#define mmMMSCH_GPUIOV_CMD_STATUS_1                                                                    0x0038
+#define mmMMSCH_GPUIOV_CMD_STATUS_1_BASE_IDX                                                           0
+#define mmMMSCH_GPUIOV_VM_BUSY_STATUS_1                                                                0x0039
+#define mmMMSCH_GPUIOV_VM_BUSY_STATUS_1_BASE_IDX                                                       0
+#define mmMMSCH_GPUIOV_ACTIVE_FCNS_1                                                                   0x003a
+#define mmMMSCH_GPUIOV_ACTIVE_FCNS_1_BASE_IDX                                                          0
+#define mmMMSCH_GPUIOV_ACTIVE_FCN_ID_1                                                                 0x003b
+#define mmMMSCH_GPUIOV_ACTIVE_FCN_ID_1_BASE_IDX                                                        0
+#define mmMMSCH_GPUIOV_DW6_1                                                                           0x003c
+#define mmMMSCH_GPUIOV_DW6_1_BASE_IDX                                                                  0
+#define mmMMSCH_GPUIOV_DW7_1                                                                           0x003d
+#define mmMMSCH_GPUIOV_DW7_1_BASE_IDX                                                                  0
+#define mmMMSCH_GPUIOV_DW8_1                                                                           0x003e
+#define mmMMSCH_GPUIOV_DW8_1_BASE_IDX                                                                  0
+#define mmMMSCH_GPUIOV_CNTXT                                                                           0x003f
+#define mmMMSCH_GPUIOV_CNTXT_BASE_IDX                                                                  0
+#define mmMMSCH_SCRATCH_2                                                                              0x0040
+#define mmMMSCH_SCRATCH_2_BASE_IDX                                                                     0
+#define mmMMSCH_SCRATCH_3                                                                              0x0041
+#define mmMMSCH_SCRATCH_3_BASE_IDX                                                                     0
+#define mmMMSCH_SCRATCH_4                                                                              0x0042
+#define mmMMSCH_SCRATCH_4_BASE_IDX                                                                     0
+#define mmMMSCH_SCRATCH_5                                                                              0x0043
+#define mmMMSCH_SCRATCH_5_BASE_IDX                                                                     0
+#define mmMMSCH_SCRATCH_6                                                                              0x0044
+#define mmMMSCH_SCRATCH_6_BASE_IDX                                                                     0
+#define mmMMSCH_SCRATCH_7                                                                              0x0045
+#define mmMMSCH_SCRATCH_7_BASE_IDX                                                                     0
+#define mmMMSCH_VFID_FIFO_HEAD_0                                                                       0x0046
+#define mmMMSCH_VFID_FIFO_HEAD_0_BASE_IDX                                                              0
+#define mmMMSCH_VFID_FIFO_TAIL_0                                                                       0x0047
+#define mmMMSCH_VFID_FIFO_TAIL_0_BASE_IDX                                                              0
+#define mmMMSCH_VFID_FIFO_HEAD_1                                                                       0x0048
+#define mmMMSCH_VFID_FIFO_HEAD_1_BASE_IDX                                                              0
+#define mmMMSCH_VFID_FIFO_TAIL_1                                                                       0x0049
+#define mmMMSCH_VFID_FIFO_TAIL_1_BASE_IDX                                                              0
+#define mmMMSCH_NACK_STATUS                                                                            0x004a
+#define mmMMSCH_NACK_STATUS_BASE_IDX                                                                   0
+#define mmMMSCH_VF_MAILBOX0_DATA                                                                       0x004b
+#define mmMMSCH_VF_MAILBOX0_DATA_BASE_IDX                                                              0
+#define mmMMSCH_VF_MAILBOX1_DATA                                                                       0x004c
+#define mmMMSCH_VF_MAILBOX1_DATA_BASE_IDX                                                              0
+#define mmMMSCH_GPUIOV_SCH_BLOCK_IP_0                                                                  0x004d
+#define mmMMSCH_GPUIOV_SCH_BLOCK_IP_0_BASE_IDX                                                         0
+#define mmMMSCH_GPUIOV_CMD_STATUS_IP_0                                                                 0x004e
+#define mmMMSCH_GPUIOV_CMD_STATUS_IP_0_BASE_IDX                                                        0
+#define mmMMSCH_GPUIOV_ACTIVE_FCN_ID_IP_0                                                              0x004f
+#define mmMMSCH_GPUIOV_ACTIVE_FCN_ID_IP_0_BASE_IDX                                                     0
+#define mmMMSCH_GPUIOV_SCH_BLOCK_IP_1                                                                  0x0050
+#define mmMMSCH_GPUIOV_SCH_BLOCK_IP_1_BASE_IDX                                                         0
+#define mmMMSCH_GPUIOV_CMD_STATUS_IP_1                                                                 0x0051
+#define mmMMSCH_GPUIOV_CMD_STATUS_IP_1_BASE_IDX                                                        0
+#define mmMMSCH_GPUIOV_ACTIVE_FCN_ID_IP_1                                                              0x0052
+#define mmMMSCH_GPUIOV_ACTIVE_FCN_ID_IP_1_BASE_IDX                                                     0
+#define mmMMSCH_GPUIOV_CNTXT_IP                                                                        0x0053
+#define mmMMSCH_GPUIOV_CNTXT_IP_BASE_IDX                                                               0
+#define mmMMSCH_GPUIOV_SCH_BLOCK_2                                                                     0x0054
+#define mmMMSCH_GPUIOV_SCH_BLOCK_2_BASE_IDX                                                            0
+#define mmMMSCH_GPUIOV_CMD_CONTROL_2                                                                   0x0055
+#define mmMMSCH_GPUIOV_CMD_CONTROL_2_BASE_IDX                                                          0
+#define mmMMSCH_GPUIOV_CMD_STATUS_2                                                                    0x0056
+#define mmMMSCH_GPUIOV_CMD_STATUS_2_BASE_IDX                                                           0
+#define mmMMSCH_GPUIOV_VM_BUSY_STATUS_2                                                                0x0057
+#define mmMMSCH_GPUIOV_VM_BUSY_STATUS_2_BASE_IDX                                                       0
+#define mmMMSCH_GPUIOV_ACTIVE_FCNS_2                                                                   0x0058
+#define mmMMSCH_GPUIOV_ACTIVE_FCNS_2_BASE_IDX                                                          0
+#define mmMMSCH_GPUIOV_ACTIVE_FCN_ID_2                                                                 0x0059
+#define mmMMSCH_GPUIOV_ACTIVE_FCN_ID_2_BASE_IDX                                                        0
+#define mmMMSCH_GPUIOV_DW6_2                                                                           0x005a
+#define mmMMSCH_GPUIOV_DW6_2_BASE_IDX                                                                  0
+#define mmMMSCH_GPUIOV_DW7_2                                                                           0x005b
+#define mmMMSCH_GPUIOV_DW7_2_BASE_IDX                                                                  0
+#define mmMMSCH_GPUIOV_DW8_2                                                                           0x005c
+#define mmMMSCH_GPUIOV_DW8_2_BASE_IDX                                                                  0
+#define mmMMSCH_GPUIOV_SCH_BLOCK_IP_2                                                                  0x005d
+#define mmMMSCH_GPUIOV_SCH_BLOCK_IP_2_BASE_IDX                                                         0
+#define mmMMSCH_GPUIOV_CMD_STATUS_IP_2                                                                 0x005e
+#define mmMMSCH_GPUIOV_CMD_STATUS_IP_2_BASE_IDX                                                        0
+#define mmMMSCH_GPUIOV_ACTIVE_FCN_ID_IP_2                                                              0x005f
+#define mmMMSCH_GPUIOV_ACTIVE_FCN_ID_IP_2_BASE_IDX                                                     0
+#define mmMMSCH_VFID_FIFO_HEAD_2                                                                       0x0060
+#define mmMMSCH_VFID_FIFO_HEAD_2_BASE_IDX                                                              0
+#define mmMMSCH_VFID_FIFO_TAIL_2                                                                       0x0061
+#define mmMMSCH_VFID_FIFO_TAIL_2_BASE_IDX                                                              0
+#define mmMMSCH_VM_BUSY_STATUS_0                                                                       0x0062
+#define mmMMSCH_VM_BUSY_STATUS_0_BASE_IDX                                                              0
+#define mmMMSCH_VM_BUSY_STATUS_1                                                                       0x0063
+#define mmMMSCH_VM_BUSY_STATUS_1_BASE_IDX                                                              0
+#define mmMMSCH_VM_BUSY_STATUS_2                                                                       0x0064
+#define mmMMSCH_VM_BUSY_STATUS_2_BASE_IDX                                                              0
+
+#define MMSCH_VERSION_MAJOR	2
+#define MMSCH_VERSION_MINOR	0
+#define MMSCH_VERSION	(MMSCH_VERSION_MAJOR << 16 | MMSCH_VERSION_MINOR)
+
+enum mmsch_v2_0_command_type {
+	MMSCH_COMMAND__DIRECT_REG_WRITE = 0,
+	MMSCH_COMMAND__DIRECT_REG_POLLING = 2,
+	MMSCH_COMMAND__DIRECT_REG_READ_MODIFY_WRITE = 3,
+	MMSCH_COMMAND__INDIRECT_REG_WRITE = 8,
+	MMSCH_COMMAND__END = 0xf
+};
+
+struct mmsch_v2_0_init_header {
+	uint32_t version;
+	uint32_t header_size;
+	uint32_t vcn_init_status;
+	uint32_t vcn_table_offset;
+	uint32_t vcn_table_size;
+};
+
+struct mmsch_v2_0_cmd_direct_reg_header {
+	uint32_t reg_offset   : 28;
+	uint32_t command_type : 4;
+};
+
+struct mmsch_v2_0_cmd_indirect_reg_header {
+	uint32_t reg_offset    : 20;
+	uint32_t reg_idx_space : 8;
+	uint32_t command_type  : 4;
+};
+
+struct mmsch_v2_0_cmd_direct_write {
+	struct mmsch_v2_0_cmd_direct_reg_header cmd_header;
+	uint32_t reg_value;
+};
+
+struct mmsch_v2_0_cmd_direct_read_modify_write {
+	struct mmsch_v2_0_cmd_direct_reg_header cmd_header;
+	uint32_t write_data;
+	uint32_t mask_value;
+};
+
+struct mmsch_v2_0_cmd_direct_polling {
+	struct mmsch_v2_0_cmd_direct_reg_header cmd_header;
+	uint32_t mask_value;
+	uint32_t wait_value;
+};
+
+struct mmsch_v2_0_cmd_end {
+	struct mmsch_v2_0_cmd_direct_reg_header cmd_header;
+};
+
+struct mmsch_v2_0_cmd_indirect_write {
+	struct mmsch_v2_0_cmd_indirect_reg_header cmd_header;
+	uint32_t reg_value;
+};
+
+static inline void mmsch_v2_0_insert_direct_wt(struct mmsch_v2_0_cmd_direct_write *direct_wt,
+					       uint32_t *init_table,
+					       uint32_t reg_offset,
+					       uint32_t value)
+{
+	direct_wt->cmd_header.reg_offset = reg_offset;
+	direct_wt->reg_value = value;
+	memcpy((void *)init_table, direct_wt, sizeof(struct mmsch_v2_0_cmd_direct_write));
+}
+
+static inline void mmsch_v2_0_insert_direct_rd_mod_wt(struct mmsch_v2_0_cmd_direct_read_modify_write *direct_rd_mod_wt,
+						      uint32_t *init_table,
+						      uint32_t reg_offset,
+						      uint32_t mask, uint32_t data)
+{
+	direct_rd_mod_wt->cmd_header.reg_offset = reg_offset;
+	direct_rd_mod_wt->mask_value = mask;
+	direct_rd_mod_wt->write_data = data;
+	memcpy((void *)init_table, direct_rd_mod_wt,
+	       sizeof(struct mmsch_v2_0_cmd_direct_read_modify_write));
+}
+
+static inline void mmsch_v2_0_insert_direct_poll(struct mmsch_v2_0_cmd_direct_polling *direct_poll,
+						 uint32_t *init_table,
+						 uint32_t reg_offset,
+						 uint32_t mask, uint32_t wait)
+{
+	direct_poll->cmd_header.reg_offset = reg_offset;
+	direct_poll->mask_value = mask;
+	direct_poll->wait_value = wait;
+	memcpy((void *)init_table, direct_poll, sizeof(struct mmsch_v2_0_cmd_direct_polling));
+}
+
+#define MMSCH_V2_0_INSERT_DIRECT_RD_MOD_WT(reg, mask, data) { \
+	mmsch_v2_0_insert_direct_rd_mod_wt(&direct_rd_mod_wt, \
+					   init_table, (reg), \
+					   (mask), (data)); \
+	init_table += sizeof(struct mmsch_v2_0_cmd_direct_read_modify_write)/4; \
+	table_size += sizeof(struct mmsch_v2_0_cmd_direct_read_modify_write)/4; \
+}
+
+#define MMSCH_V2_0_INSERT_DIRECT_WT(reg, value) { \
+	mmsch_v2_0_insert_direct_wt(&direct_wt, \
+				    init_table, (reg), \
+				    (value)); \
+	init_table += sizeof(struct mmsch_v2_0_cmd_direct_write)/4; \
+	table_size += sizeof(struct mmsch_v2_0_cmd_direct_write)/4; \
+}
+
+#define MMSCH_V2_0_INSERT_DIRECT_POLL(reg, mask, wait) { \
+	mmsch_v2_0_insert_direct_poll(&direct_poll, \
+				      init_table, (reg), \
+				      (mask), (wait)); \
+	init_table += sizeof(struct mmsch_v2_0_cmd_direct_polling)/4; \
+	table_size += sizeof(struct mmsch_v2_0_cmd_direct_polling)/4; \
+}
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/mmsch_v3_0.h b/drivers/gpu/drm/amd/amdgpu/mmsch_v3_0.h
new file mode 100644
index 000000000000..a773ef61b78c
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmsch_v3_0.h
@@ -0,0 +1,132 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __MMSCH_V3_0_H__
+#define __MMSCH_V3_0_H__
+
+#include "amdgpu_vcn.h"
+
+#define MMSCH_VERSION_MAJOR	3
+#define MMSCH_VERSION_MINOR	0
+#define MMSCH_VERSION	(MMSCH_VERSION_MAJOR << 16 | MMSCH_VERSION_MINOR)
+
+#define MMSCH_V3_0_VCN_INSTANCES 0x2
+
+enum mmsch_v3_0_command_type {
+	MMSCH_COMMAND__DIRECT_REG_WRITE = 0,
+	MMSCH_COMMAND__DIRECT_REG_POLLING = 2,
+	MMSCH_COMMAND__DIRECT_REG_READ_MODIFY_WRITE = 3,
+	MMSCH_COMMAND__INDIRECT_REG_WRITE = 8,
+	MMSCH_COMMAND__END = 0xf
+};
+
+struct mmsch_v3_0_table_info {
+	uint32_t init_status;
+	uint32_t table_offset;
+	uint32_t table_size;
+};
+
+struct mmsch_v3_0_init_header {
+	uint32_t version;
+	uint32_t total_size;
+	struct mmsch_v3_0_table_info inst[MMSCH_V3_0_VCN_INSTANCES];
+};
+
+struct mmsch_v3_0_cmd_direct_reg_header {
+	uint32_t reg_offset   : 28;
+	uint32_t command_type : 4;
+};
+
+struct mmsch_v3_0_cmd_indirect_reg_header {
+	uint32_t reg_offset    : 20;
+	uint32_t reg_idx_space : 8;
+	uint32_t command_type  : 4;
+};
+
+struct mmsch_v3_0_cmd_direct_write {
+	struct mmsch_v3_0_cmd_direct_reg_header cmd_header;
+	uint32_t reg_value;
+};
+
+struct mmsch_v3_0_cmd_direct_read_modify_write {
+	struct mmsch_v3_0_cmd_direct_reg_header cmd_header;
+	uint32_t write_data;
+	uint32_t mask_value;
+};
+
+struct mmsch_v3_0_cmd_direct_polling {
+	struct mmsch_v3_0_cmd_direct_reg_header cmd_header;
+	uint32_t mask_value;
+	uint32_t wait_value;
+};
+
+struct mmsch_v3_0_cmd_end {
+	struct mmsch_v3_0_cmd_direct_reg_header cmd_header;
+};
+
+struct mmsch_v3_0_cmd_indirect_write {
+	struct mmsch_v3_0_cmd_indirect_reg_header cmd_header;
+	uint32_t reg_value;
+};
+
+#define MMSCH_V3_0_INSERT_DIRECT_RD_MOD_WT(reg, mask, data) { \
+	size = sizeof(struct mmsch_v3_0_cmd_direct_read_modify_write); \
+	size_dw = size / 4; \
+	direct_rd_mod_wt.cmd_header.reg_offset = reg; \
+	direct_rd_mod_wt.mask_value = mask; \
+	direct_rd_mod_wt.write_data = data; \
+	memcpy((void *)table_loc, &direct_rd_mod_wt, size); \
+	table_loc += size_dw; \
+	table_size += size_dw; \
+}
+
+#define MMSCH_V3_0_INSERT_DIRECT_WT(reg, value) { \
+	size = sizeof(struct mmsch_v3_0_cmd_direct_write); \
+	size_dw = size / 4; \
+	direct_wt.cmd_header.reg_offset = reg; \
+	direct_wt.reg_value = value; \
+	memcpy((void *)table_loc, &direct_wt, size); \
+	table_loc += size_dw; \
+	table_size += size_dw; \
+}
+
+#define MMSCH_V3_0_INSERT_DIRECT_POLL(reg, mask, wait) { \
+	size = sizeof(struct mmsch_v3_0_cmd_direct_polling); \
+	size_dw = size / 4; \
+	direct_poll.cmd_header.reg_offset = reg; \
+	direct_poll.mask_value = mask; \
+	direct_poll.wait_value = wait; \
+	memcpy((void *)table_loc, &direct_poll, size); \
+	table_loc += size_dw; \
+	table_size += size_dw; \
+}
+
+#define MMSCH_V3_0_INSERT_END() { \
+	size = sizeof(struct mmsch_v3_0_cmd_end); \
+	size_dw = size / 4; \
+	memcpy((void *)table_loc, &end, size); \
+	table_loc += size_dw; \
+	table_size += size_dw; \
+}
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/mmsch_v4_0.h b/drivers/gpu/drm/amd/amdgpu/mmsch_v4_0.h
new file mode 100644
index 000000000000..ced26cc5123a
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmsch_v4_0.h
@@ -0,0 +1,144 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __MMSCH_V4_0_H__
+#define __MMSCH_V4_0_H__
+
+#include "amdgpu_vcn.h"
+
+#define MMSCH_VERSION_MAJOR	4
+#define MMSCH_VERSION_MINOR	0
+#define MMSCH_VERSION	(MMSCH_VERSION_MAJOR << 16 | MMSCH_VERSION_MINOR)
+
+#define RB_ENABLED (1 << 0)
+#define RB4_ENABLED (1 << 1)
+
+#define MMSCH_VF_ENGINE_STATUS__PASS 0x1
+
+#define MMSCH_VF_MAILBOX_RESP__OK                      0x1
+#define MMSCH_VF_MAILBOX_RESP__INCOMPLETE              0x2
+#define MMSCH_VF_MAILBOX_RESP__FAILED                  0x3
+#define MMSCH_VF_MAILBOX_RESP__FAILED_SMALL_CTX_SIZE   0x4
+#define MMSCH_VF_MAILBOX_RESP__UNKNOWN_CMD             0x5
+
+#define MMSCH_V4_0_VCN_INSTANCES 0x2
+
+enum mmsch_v4_0_command_type {
+	MMSCH_COMMAND__DIRECT_REG_WRITE = 0,
+	MMSCH_COMMAND__DIRECT_REG_POLLING = 2,
+	MMSCH_COMMAND__DIRECT_REG_READ_MODIFY_WRITE = 3,
+	MMSCH_COMMAND__INDIRECT_REG_WRITE = 8,
+	MMSCH_COMMAND__END = 0xf
+};
+
+struct mmsch_v4_0_table_info {
+	uint32_t init_status;
+	uint32_t table_offset;
+	uint32_t table_size;
+};
+
+struct mmsch_v4_0_init_header {
+	uint32_t version;
+	uint32_t total_size;
+	struct mmsch_v4_0_table_info inst[MMSCH_V4_0_VCN_INSTANCES];
+	struct mmsch_v4_0_table_info jpegdec;
+};
+
+struct mmsch_v4_0_cmd_direct_reg_header {
+	uint32_t reg_offset   : 28;
+	uint32_t command_type : 4;
+};
+
+struct mmsch_v4_0_cmd_indirect_reg_header {
+	uint32_t reg_offset    : 20;
+	uint32_t reg_idx_space : 8;
+	uint32_t command_type  : 4;
+};
+
+struct mmsch_v4_0_cmd_direct_write {
+	struct mmsch_v4_0_cmd_direct_reg_header cmd_header;
+	uint32_t reg_value;
+};
+
+struct mmsch_v4_0_cmd_direct_read_modify_write {
+	struct mmsch_v4_0_cmd_direct_reg_header cmd_header;
+	uint32_t write_data;
+	uint32_t mask_value;
+};
+
+struct mmsch_v4_0_cmd_direct_polling {
+	struct mmsch_v4_0_cmd_direct_reg_header cmd_header;
+	uint32_t mask_value;
+	uint32_t wait_value;
+};
+
+struct mmsch_v4_0_cmd_end {
+	struct mmsch_v4_0_cmd_direct_reg_header cmd_header;
+};
+
+struct mmsch_v4_0_cmd_indirect_write {
+	struct mmsch_v4_0_cmd_indirect_reg_header cmd_header;
+	uint32_t reg_value;
+};
+
+#define MMSCH_V4_0_INSERT_DIRECT_RD_MOD_WT(reg, mask, data) { \
+	size = sizeof(struct mmsch_v4_0_cmd_direct_read_modify_write); \
+	size_dw = size / 4; \
+	direct_rd_mod_wt.cmd_header.reg_offset = reg; \
+	direct_rd_mod_wt.mask_value = mask; \
+	direct_rd_mod_wt.write_data = data; \
+	memcpy((void *)table_loc, &direct_rd_mod_wt, size); \
+	table_loc += size_dw; \
+	table_size += size_dw; \
+}
+
+#define MMSCH_V4_0_INSERT_DIRECT_WT(reg, value) { \
+	size = sizeof(struct mmsch_v4_0_cmd_direct_write); \
+	size_dw = size / 4; \
+	direct_wt.cmd_header.reg_offset = reg; \
+	direct_wt.reg_value = value; \
+	memcpy((void *)table_loc, &direct_wt, size); \
+	table_loc += size_dw; \
+	table_size += size_dw; \
+}
+
+#define MMSCH_V4_0_INSERT_DIRECT_POLL(reg, mask, wait) { \
+	size = sizeof(struct mmsch_v4_0_cmd_direct_polling); \
+	size_dw = size / 4; \
+	direct_poll.cmd_header.reg_offset = reg; \
+	direct_poll.mask_value = mask; \
+	direct_poll.wait_value = wait; \
+	memcpy((void *)table_loc, &direct_poll, size); \
+	table_loc += size_dw; \
+	table_size += size_dw; \
+}
+
+#define MMSCH_V4_0_INSERT_END() { \
+	size = sizeof(struct mmsch_v4_0_cmd_end); \
+	size_dw = size / 4; \
+	memcpy((void *)table_loc, &end, size); \
+	table_loc += size_dw; \
+	table_size += size_dw; \
+}
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/mmsch_v4_0_3.h b/drivers/gpu/drm/amd/amdgpu/mmsch_v4_0_3.h
new file mode 100644
index 000000000000..db7eb5260295
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmsch_v4_0_3.h
@@ -0,0 +1,37 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __MMSCH_V4_0_3_H__
+#define __MMSCH_V4_0_3_H__
+
+#include "amdgpu_vcn.h"
+#include "mmsch_v4_0.h"
+
+struct mmsch_v4_0_3_init_header {
+	uint32_t version;
+	uint32_t total_size;
+	struct mmsch_v4_0_table_info vcn0;
+	struct mmsch_v4_0_table_info mjpegdec0[4];
+	struct mmsch_v4_0_table_info mjpegdec1[4];
+};
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/mmsch_v5_0.h b/drivers/gpu/drm/amd/amdgpu/mmsch_v5_0.h
new file mode 100644
index 000000000000..6f749814929f
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mmsch_v5_0.h
@@ -0,0 +1,144 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __MMSCH_V5_0_H__
+#define __MMSCH_V5_0_H__
+
+#include "amdgpu_vcn.h"
+
+#define MMSCH_VERSION_MAJOR	5
+#define MMSCH_VERSION_MINOR	0
+#define MMSCH_VERSION	(MMSCH_VERSION_MAJOR << 16 | MMSCH_VERSION_MINOR)
+
+#define RB_ENABLED (1 << 0)
+#define RB4_ENABLED (1 << 1)
+
+#define MMSCH_VF_ENGINE_STATUS__PASS 0x1
+
+#define MMSCH_VF_MAILBOX_RESP__OK                      0x1
+#define MMSCH_VF_MAILBOX_RESP__INCOMPLETE              0x2
+#define MMSCH_VF_MAILBOX_RESP__FAILED                  0x3
+#define MMSCH_VF_MAILBOX_RESP__FAILED_SMALL_CTX_SIZE   0x4
+#define MMSCH_VF_MAILBOX_RESP__UNKNOWN_CMD             0x5
+
+enum mmsch_v5_0_command_type {
+	MMSCH_COMMAND__DIRECT_REG_WRITE = 0,
+	MMSCH_COMMAND__DIRECT_REG_POLLING = 2,
+	MMSCH_COMMAND__DIRECT_REG_READ_MODIFY_WRITE = 3,
+	MMSCH_COMMAND__INDIRECT_REG_WRITE = 8,
+	MMSCH_COMMAND__END = 0xf
+};
+
+struct mmsch_v5_0_table_info {
+	uint32_t init_status;
+	uint32_t table_offset;
+	uint32_t table_size;
+};
+
+struct mmsch_v5_0_init_header {
+	uint32_t version;
+	uint32_t total_size;
+	struct mmsch_v5_0_table_info vcn0;
+	struct mmsch_v5_0_table_info mjpegdec0[5];
+	struct mmsch_v5_0_table_info mjpegdec1[5];
+};
+
+struct mmsch_v5_0_cmd_direct_reg_header {
+	uint32_t reg_offset   : 28;
+	uint32_t command_type : 4;
+};
+
+struct mmsch_v5_0_cmd_indirect_reg_header {
+	uint32_t reg_offset    : 20;
+	uint32_t reg_idx_space : 8;
+	uint32_t command_type  : 4;
+};
+
+struct mmsch_v5_0_cmd_direct_write {
+	struct mmsch_v5_0_cmd_direct_reg_header cmd_header;
+	uint32_t reg_value;
+};
+
+struct mmsch_v5_0_cmd_direct_read_modify_write {
+	struct mmsch_v5_0_cmd_direct_reg_header cmd_header;
+	uint32_t write_data;
+	uint32_t mask_value;
+};
+
+struct mmsch_v5_0_cmd_direct_polling {
+	struct mmsch_v5_0_cmd_direct_reg_header cmd_header;
+	uint32_t mask_value;
+	uint32_t wait_value;
+};
+
+struct mmsch_v5_0_cmd_end {
+	struct mmsch_v5_0_cmd_direct_reg_header cmd_header;
+};
+
+struct mmsch_v5_0_cmd_indirect_write {
+	struct mmsch_v5_0_cmd_indirect_reg_header cmd_header;
+	uint32_t reg_value;
+};
+
+#define MMSCH_V5_0_INSERT_DIRECT_RD_MOD_WT(reg, mask, data) { \
+	size = sizeof(struct mmsch_v5_0_cmd_direct_read_modify_write); \
+	size_dw = size / 4; \
+	direct_rd_mod_wt.cmd_header.reg_offset = reg; \
+	direct_rd_mod_wt.mask_value = mask; \
+	direct_rd_mod_wt.write_data = data; \
+	memcpy((void *)table_loc, &direct_rd_mod_wt, size); \
+	table_loc += size_dw; \
+	table_size += size_dw; \
+}
+
+#define MMSCH_V5_0_INSERT_DIRECT_WT(reg, value) { \
+	size = sizeof(struct mmsch_v5_0_cmd_direct_write); \
+	size_dw = size / 4; \
+	direct_wt.cmd_header.reg_offset = reg; \
+	direct_wt.reg_value = value; \
+	memcpy((void *)table_loc, &direct_wt, size); \
+	table_loc += size_dw; \
+	table_size += size_dw; \
+}
+
+#define MMSCH_V5_0_INSERT_DIRECT_POLL(reg, mask, wait) { \
+	size = sizeof(struct mmsch_v5_0_cmd_direct_polling); \
+	size_dw = size / 4; \
+	direct_poll.cmd_header.reg_offset = reg; \
+	direct_poll.mask_value = mask; \
+	direct_poll.wait_value = wait; \
+	memcpy((void *)table_loc, &direct_poll, size); \
+	table_loc += size_dw; \
+	table_size += size_dw; \
+}
+
+#define MMSCH_V5_0_INSERT_END() { \
+	size = sizeof(struct mmsch_v5_0_cmd_end); \
+	size_dw = size / 4; \
+	memcpy((void *)table_loc, &end, size); \
+	table_loc += size_dw; \
+	table_size += size_dw; \
+}
+
+#endif
+
diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c b/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c
index 2812d88a8bdd..9a40107a0869 100644
--- a/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c
+++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c
@@ -22,68 +22,48 @@
  */
 
 #include "amdgpu.h"
-#include "vega10/soc15ip.h"
-#include "vega10/NBIO/nbio_6_1_offset.h"
-#include "vega10/NBIO/nbio_6_1_sh_mask.h"
-#include "vega10/GC/gc_9_0_offset.h"
-#include "vega10/GC/gc_9_0_sh_mask.h"
+#include "nbio/nbio_6_1_offset.h"
+#include "nbio/nbio_6_1_sh_mask.h"
+#include "gc/gc_9_0_offset.h"
+#include "gc/gc_9_0_sh_mask.h"
+#include "mp/mp_9_0_offset.h"
 #include "soc15.h"
 #include "vega10_ih.h"
 #include "soc15_common.h"
 #include "mxgpu_ai.h"
 
+#include "amdgpu_reset.h"
+
 static void xgpu_ai_mailbox_send_ack(struct amdgpu_device *adev)
 {
-	u32 reg;
-	int timeout = AI_MAILBOX_TIMEDOUT;
-	u32 mask = REG_FIELD_MASK(BIF_BX_PF0_MAILBOX_CONTROL, RCV_MSG_VALID);
-
-	reg = RREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0,
-					     mmBIF_BX_PF0_MAILBOX_CONTROL));
-	reg = REG_SET_FIELD(reg, BIF_BX_PF0_MAILBOX_CONTROL, RCV_MSG_ACK, 1);
-	WREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0,
-				       mmBIF_BX_PF0_MAILBOX_CONTROL), reg);
-
-	/*Wait for RCV_MSG_VALID to be 0*/
-	reg = RREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0,
-					     mmBIF_BX_PF0_MAILBOX_CONTROL));
-	while (reg & mask) {
-		if (timeout <= 0) {
-			pr_err("RCV_MSG_VALID is not cleared\n");
-			break;
-		}
-		mdelay(1);
-		timeout -=1;
-
-		reg = RREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0,
-						     mmBIF_BX_PF0_MAILBOX_CONTROL));
-	}
+	WREG8(AI_MAIBOX_CONTROL_RCV_OFFSET_BYTE, 2);
 }
 
 static void xgpu_ai_mailbox_set_valid(struct amdgpu_device *adev, bool val)
 {
-	u32 reg;
+	WREG8(AI_MAIBOX_CONTROL_TRN_OFFSET_BYTE, val ? 1 : 0);
+}
 
-	reg = RREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0,
-					     mmBIF_BX_PF0_MAILBOX_CONTROL));
-	reg = REG_SET_FIELD(reg, BIF_BX_PF0_MAILBOX_CONTROL,
-			    TRN_MSG_VALID, val ? 1 : 0);
-	WREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_MAILBOX_CONTROL),
-		      reg);
+/*
+ * this peek_msg could *only* be called in IRQ routine becuase in IRQ routine
+ * RCV_MSG_VALID filed of BIF_BX_PF0_MAILBOX_CONTROL must already be set to 1
+ * by host.
+ *
+ * if called no in IRQ routine, this peek_msg cannot guaranteed to return the
+ * correct value since it doesn't return the RCV_DW0 under the case that
+ * RCV_MSG_VALID is set by host.
+ */
+static enum idh_event xgpu_ai_mailbox_peek_msg(struct amdgpu_device *adev)
+{
+	return RREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0,
+				mmBIF_BX_PF0_MAILBOX_MSGBUF_RCV_DW0));
 }
 
+
 static int xgpu_ai_mailbox_rcv_msg(struct amdgpu_device *adev,
 				   enum idh_event event)
 {
 	u32 reg;
-	u32 mask = REG_FIELD_MASK(BIF_BX_PF0_MAILBOX_CONTROL, RCV_MSG_VALID);
-
-	if (event != IDH_FLR_NOTIFICATION_CMPL) {
-		reg = RREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0,
-						     mmBIF_BX_PF0_MAILBOX_CONTROL));
-		if (!(reg & mask))
-			return -ENOENT;
-	}
 
 	reg = RREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0,
 					     mmBIF_BX_PF0_MAILBOX_MSGBUF_RCV_DW0));
@@ -95,54 +75,67 @@ static int xgpu_ai_mailbox_rcv_msg(struct amdgpu_device *adev,
 	return 0;
 }
 
+static uint8_t xgpu_ai_peek_ack(struct amdgpu_device *adev) {
+	return RREG8(AI_MAIBOX_CONTROL_TRN_OFFSET_BYTE) & 2;
+}
+
 static int xgpu_ai_poll_ack(struct amdgpu_device *adev)
 {
-	int r = 0, timeout = AI_MAILBOX_TIMEDOUT;
-	u32 mask = REG_FIELD_MASK(BIF_BX_PF0_MAILBOX_CONTROL, TRN_MSG_ACK);
-	u32 reg;
+	int timeout  = AI_MAILBOX_POLL_ACK_TIMEDOUT;
+	u8 reg;
+
+	do {
+		reg = RREG8(AI_MAIBOX_CONTROL_TRN_OFFSET_BYTE);
+		if (reg & 2)
+			return 0;
 
-	reg = RREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0,
-					     mmBIF_BX_PF0_MAILBOX_CONTROL));
-	while (!(reg & mask)) {
-		if (timeout <= 0) {
-			pr_err("Doesn't get ack from pf.\n");
-			r = -ETIME;
-			break;
-		}
 		mdelay(5);
 		timeout -= 5;
+	} while (timeout > 1);
 
-		reg = RREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0,
-						     mmBIF_BX_PF0_MAILBOX_CONTROL));
-	}
+	dev_err(adev->dev, "Doesn't get TRN_MSG_ACK from pf in %d msec\n", AI_MAILBOX_POLL_ACK_TIMEDOUT);
 
-	return r;
+	return -ETIME;
 }
 
 static int xgpu_ai_poll_msg(struct amdgpu_device *adev, enum idh_event event)
 {
-	int r = 0, timeout = AI_MAILBOX_TIMEDOUT;
-
-	r = xgpu_ai_mailbox_rcv_msg(adev, event);
-	while (r) {
-		if (timeout <= 0) {
-			pr_err("Doesn't get msg:%d from pf.\n", event);
-			r = -ETIME;
-			break;
-		}
-		mdelay(5);
-		timeout -= 5;
+	int r, timeout = AI_MAILBOX_POLL_MSG_TIMEDOUT;
 
+	do {
 		r = xgpu_ai_mailbox_rcv_msg(adev, event);
-	}
+		if (!r)
+			return 0;
 
-	return r;
+		msleep(10);
+		timeout -= 10;
+	} while (timeout > 1);
+
+	dev_err(adev->dev, "Doesn't get msg:%d from pf, error=%d\n", event, r);
+
+	return -ETIME;
 }
 
 static void xgpu_ai_mailbox_trans_msg (struct amdgpu_device *adev,
 	      enum idh_request req, u32 data1, u32 data2, u32 data3) {
 	u32 reg;
 	int r;
+	uint8_t trn;
+
+	/* IMPORTANT:
+	 * clear TRN_MSG_VALID valid to clear host's RCV_MSG_ACK
+	 * and with host's RCV_MSG_ACK cleared hw automatically clear host's RCV_MSG_ACK
+	 * which lead to VF's TRN_MSG_ACK cleared, otherwise below xgpu_ai_poll_ack()
+	 * will return immediatly
+	 */
+	do {
+		xgpu_ai_mailbox_set_valid(adev, false);
+		trn = xgpu_ai_peek_ack(adev);
+		if (trn) {
+			dev_err_ratelimited(adev->dev, "trn=%x ACK should not assert! wait again !\n", trn);
+			msleep(1);
+		}
+	} while(trn);
 
 	reg = RREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0,
 					     mmBIF_BX_PF0_MAILBOX_MSGBUF_TRN_DW0));
@@ -162,7 +155,7 @@ static void xgpu_ai_mailbox_trans_msg (struct amdgpu_device *adev,
 	/* start to poll ack */
 	r = xgpu_ai_poll_ack(adev);
 	if (r)
-		pr_err("Doesn't get ack from pf, continue\n");
+		dev_err(adev->dev, "Doesn't get ack from pf, continue\n");
 
 	xgpu_ai_mailbox_set_valid(adev, false);
 }
@@ -180,9 +173,20 @@ static int xgpu_ai_send_access_requests(struct amdgpu_device *adev,
 		req == IDH_REQ_GPU_RESET_ACCESS) {
 		r = xgpu_ai_poll_msg(adev, IDH_READY_TO_ACCESS_GPU);
 		if (r) {
-			pr_err("Doesn't get READY_TO_ACCESS_GPU from pf, give up\n");
+			dev_err(adev->dev, "Doesn't get READY_TO_ACCESS_GPU from pf, give up\n");
 			return r;
 		}
+		/* Retrieve checksum from mailbox2 */
+		if (req == IDH_REQ_GPU_INIT_ACCESS || req == IDH_REQ_GPU_RESET_ACCESS) {
+			adev->virt.fw_reserve.checksum_key =
+				RREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0,
+					mmBIF_BX_PF0_MAILBOX_MSGBUF_RCV_DW2));
+		}
+	} else if (req == IDH_REQ_GPU_INIT_DATA){
+		/* Dummy REQ_GPU_INIT_DATA handling */
+		r = xgpu_ai_poll_msg(adev, IDH_REQ_GPU_INIT_DATA_READY);
+		/* version set to 0 since dummy */
+		adev->virt.req_init_data_ver = 0;	
 	}
 
 	return 0;
@@ -190,7 +194,16 @@ static int xgpu_ai_send_access_requests(struct amdgpu_device *adev,
 
 static int xgpu_ai_request_reset(struct amdgpu_device *adev)
 {
-	return xgpu_ai_send_access_requests(adev, IDH_REQ_GPU_RESET_ACCESS);
+	int ret, i = 0;
+
+	while (i < AI_MAILBOX_POLL_MSG_REP_MAX) {
+		ret = xgpu_ai_send_access_requests(adev, IDH_REQ_GPU_RESET_ACCESS);
+		if (!ret)
+			break;
+		i++;
+	}
+
+	return ret;
 }
 
 static int xgpu_ai_request_full_gpu_access(struct amdgpu_device *adev,
@@ -218,7 +231,7 @@ static int xgpu_ai_mailbox_ack_irq(struct amdgpu_device *adev,
 					struct amdgpu_irq_src *source,
 					struct amdgpu_iv_entry *entry)
 {
-	DRM_DEBUG("get ack intr and do nothing.\n");
+	dev_dbg(adev->dev, "get ack intr and do nothing.\n");
 	return 0;
 }
 
@@ -236,19 +249,78 @@ static int xgpu_ai_set_mailbox_ack_irq(struct amdgpu_device *adev,
 	return 0;
 }
 
+static void xgpu_ai_ready_to_reset(struct amdgpu_device *adev)
+{
+	xgpu_ai_mailbox_trans_msg(adev, IDH_READY_TO_RESET, 0, 0, 0);
+}
+
+static int xgpu_ai_wait_reset(struct amdgpu_device *adev)
+{
+	int timeout = AI_MAILBOX_POLL_FLR_TIMEDOUT;
+	do {
+		if (xgpu_ai_mailbox_peek_msg(adev) == IDH_FLR_NOTIFICATION_CMPL) {
+			dev_dbg(adev->dev, "Got AI IDH_FLR_NOTIFICATION_CMPL after %d ms\n", AI_MAILBOX_POLL_FLR_TIMEDOUT - timeout);
+			return 0;
+		}
+		msleep(10);
+		timeout -= 10;
+	} while (timeout > 1);
+
+	dev_dbg(adev->dev, "waiting AI IDH_FLR_NOTIFICATION_CMPL timeout\n");
+	return -ETIME;
+}
+
 static void xgpu_ai_mailbox_flr_work(struct work_struct *work)
 {
 	struct amdgpu_virt *virt = container_of(work, struct amdgpu_virt, flr_work);
 	struct amdgpu_device *adev = container_of(virt, struct amdgpu_device, virt);
+	struct amdgpu_reset_context reset_context = { 0 };
+
+	amdgpu_virt_fini_data_exchange(adev);
+
+	/* Trigger recovery for world switch failure if no TDR */
+	if (amdgpu_device_should_recover_gpu(adev)
+		&& (!amdgpu_device_has_job_running(adev) ||
+			adev->sdma_timeout == MAX_SCHEDULE_TIMEOUT)) {
+
+		reset_context.method = AMD_RESET_METHOD_NONE;
+		reset_context.reset_req_dev = adev;
+		clear_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
+		set_bit(AMDGPU_HOST_FLR, &reset_context.flags);
+
+		amdgpu_device_gpu_recover(adev, NULL, &reset_context);
+	}
+}
+
+static void xgpu_ai_mailbox_req_bad_pages_work(struct work_struct *work)
+{
+	struct amdgpu_virt *virt = container_of(work, struct amdgpu_virt, req_bad_pages_work);
+	struct amdgpu_device *adev = container_of(virt, struct amdgpu_device, virt);
 
-	/* wait until RCV_MSG become 3 */
-	if (xgpu_ai_poll_msg(adev, IDH_FLR_NOTIFICATION_CMPL)) {
-		pr_err("failed to recieve FLR_CMPL\n");
-		return;
+	if (down_read_trylock(&adev->reset_domain->sem)) {
+		amdgpu_virt_fini_data_exchange(adev);
+		amdgpu_virt_request_bad_pages(adev);
+		up_read(&adev->reset_domain->sem);
 	}
+}
+
+/**
+ * xgpu_ai_mailbox_handle_bad_pages_work - Reinitialize the data exchange region to get fresh bad page information
+ * @work: pointer to the work_struct
+ *
+ * This work handler is triggered when bad pages are ready, and it reinitializes
+ * the data exchange region to retrieve updated bad page information from the host.
+ */
+static void xgpu_ai_mailbox_handle_bad_pages_work(struct work_struct *work)
+{
+	struct amdgpu_virt *virt = container_of(work, struct amdgpu_virt, handle_bad_pages_work);
+	struct amdgpu_device *adev = container_of(virt, struct amdgpu_device, virt);
 
-	/* Trigger recovery due to world switch failure */
-	amdgpu_sriov_gpu_reset(adev, NULL);
+	if (down_read_trylock(&adev->reset_domain->sem)) {
+		amdgpu_virt_fini_data_exchange(adev);
+		amdgpu_virt_init_data_exchange(adev);
+		up_read(&adev->reset_domain->sem);
+	}
 }
 
 static int xgpu_ai_set_mailbox_rcv_irq(struct amdgpu_device *adev,
@@ -269,16 +341,49 @@ static int xgpu_ai_mailbox_rcv_irq(struct amdgpu_device *adev,
 				   struct amdgpu_irq_src *source,
 				   struct amdgpu_iv_entry *entry)
 {
-	int r;
-
-	/* trigger gpu-reset by hypervisor only if TDR disbaled */
-	if (amdgpu_lockup_timeout == 0) {
-		/* see what event we get */
-		r = xgpu_ai_mailbox_rcv_msg(adev, IDH_FLR_NOTIFICATION);
-
-		/* only handle FLR_NOTIFY now */
-		if (!r)
-			schedule_work(&adev->virt.flr_work);
+	enum idh_event event = xgpu_ai_mailbox_peek_msg(adev);
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+
+	switch (event) {
+	case IDH_RAS_BAD_PAGES_READY:
+		xgpu_ai_mailbox_send_ack(adev);
+		if (amdgpu_sriov_runtime(adev))
+			schedule_work(&adev->virt.handle_bad_pages_work);
+		break;
+	case IDH_RAS_BAD_PAGES_NOTIFICATION:
+		xgpu_ai_mailbox_send_ack(adev);
+		if (amdgpu_sriov_runtime(adev))
+			schedule_work(&adev->virt.req_bad_pages_work);
+		break;
+	case IDH_UNRECOV_ERR_NOTIFICATION:
+		xgpu_ai_mailbox_send_ack(adev);
+		ras->is_rma = true;
+		dev_err(adev->dev, "VF is in an unrecoverable state. Runtime Services are halted.\n");
+		if (amdgpu_sriov_runtime(adev))
+			WARN_ONCE(!amdgpu_reset_domain_schedule(adev->reset_domain,
+					&adev->virt.flr_work),
+					"Failed to queue work! at %s",
+					__func__);
+		break;
+	case IDH_FLR_NOTIFICATION:
+		if (amdgpu_sriov_runtime(adev))
+			WARN_ONCE(!amdgpu_reset_domain_schedule(adev->reset_domain,
+						&adev->virt.flr_work),
+					"Failed to queue work! at %s",
+					__func__);
+		break;
+	case IDH_QUERY_ALIVE:
+		xgpu_ai_mailbox_send_ack(adev);
+		break;
+	/* READY_TO_ACCESS_GPU is fetched by kernel polling, IRQ can ignore
+	 * it byfar since that polling thread will handle it,
+	 * other msg like flr complete is not handled here.
+	 */
+	case IDH_CLR_MSG_BUF:
+	case IDH_FLR_NOTIFICATION_CMPL:
+	case IDH_READY_TO_ACCESS_GPU:
+	default:
+		break;
 	}
 
 	return 0;
@@ -306,11 +411,11 @@ int xgpu_ai_mailbox_add_irq_id(struct amdgpu_device *adev)
 {
 	int r;
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_BIF, 135, &adev->virt.rcv_irq);
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_BIF, 135, &adev->virt.rcv_irq);
 	if (r)
 		return r;
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_BIF, 138, &adev->virt.ack_irq);
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_BIF, 138, &adev->virt.ack_irq);
 	if (r) {
 		amdgpu_irq_put(adev, &adev->virt.rcv_irq, 0);
 		return r;
@@ -333,6 +438,8 @@ int xgpu_ai_mailbox_get_irq(struct amdgpu_device *adev)
 	}
 
 	INIT_WORK(&adev->virt.flr_work, xgpu_ai_mailbox_flr_work);
+	INIT_WORK(&adev->virt.req_bad_pages_work, xgpu_ai_mailbox_req_bad_pages_work);
+	INIT_WORK(&adev->virt.handle_bad_pages_work, xgpu_ai_mailbox_handle_bad_pages_work);
 
 	return 0;
 }
@@ -343,9 +450,32 @@ void xgpu_ai_mailbox_put_irq(struct amdgpu_device *adev)
 	amdgpu_irq_put(adev, &adev->virt.rcv_irq, 0);
 }
 
+static int xgpu_ai_request_init_data(struct amdgpu_device *adev)
+{
+	return xgpu_ai_send_access_requests(adev, IDH_REQ_GPU_INIT_DATA);
+}
+
+static void xgpu_ai_ras_poison_handler(struct amdgpu_device *adev,
+					enum amdgpu_ras_block block)
+{
+	xgpu_ai_send_access_requests(adev, IDH_RAS_POISON);
+}
+
+static bool xgpu_ai_rcvd_ras_intr(struct amdgpu_device *adev)
+{
+	enum idh_event msg = xgpu_ai_mailbox_peek_msg(adev);
+
+	return (msg == IDH_RAS_ERROR_DETECTED || msg == 0xFFFFFFFF);
+}
+
 const struct amdgpu_virt_ops xgpu_ai_virt_ops = {
 	.req_full_gpu	= xgpu_ai_request_full_gpu_access,
 	.rel_full_gpu	= xgpu_ai_release_full_gpu_access,
 	.reset_gpu = xgpu_ai_request_reset,
+	.ready_to_reset = xgpu_ai_ready_to_reset,
+	.wait_reset = xgpu_ai_wait_reset,
 	.trans_msg = xgpu_ai_mailbox_trans_msg,
+	.req_init_data  = xgpu_ai_request_init_data,
+	.ras_poison_handler = xgpu_ai_ras_poison_handler,
+	.rcvd_ras_intr = xgpu_ai_rcvd_ras_intr,
 };
diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.h b/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.h
index 1e91b9a1c591..874b9f8f9804 100644
--- a/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.h
+++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.h
@@ -24,7 +24,10 @@
 #ifndef __MXGPU_AI_H__
 #define __MXGPU_AI_H__
 
-#define AI_MAILBOX_TIMEDOUT	5000
+#define AI_MAILBOX_POLL_ACK_TIMEDOUT	500
+#define AI_MAILBOX_POLL_MSG_TIMEDOUT	6000
+#define AI_MAILBOX_POLL_FLR_TIMEDOUT	10000
+#define AI_MAILBOX_POLL_MSG_REP_MAX	11
 
 enum idh_request {
 	IDH_REQ_GPU_INIT_ACCESS = 1,
@@ -32,8 +35,12 @@ enum idh_request {
 	IDH_REQ_GPU_FINI_ACCESS,
 	IDH_REL_GPU_FINI_ACCESS,
 	IDH_REQ_GPU_RESET_ACCESS,
+	IDH_REQ_GPU_INIT_DATA,
 
 	IDH_LOG_VF_ERROR       = 200,
+	IDH_READY_TO_RESET 	= 201,
+	IDH_RAS_POISON  = 202,
+	IDH_REQ_RAS_BAD_PAGES = 205,
 };
 
 enum idh_event {
@@ -41,7 +48,17 @@ enum idh_event {
 	IDH_READY_TO_ACCESS_GPU,
 	IDH_FLR_NOTIFICATION,
 	IDH_FLR_NOTIFICATION_CMPL,
-	IDH_EVENT_MAX
+	IDH_SUCCESS,
+	IDH_FAIL,
+	IDH_QUERY_ALIVE,
+	IDH_REQ_GPU_INIT_DATA_READY,
+	IDH_RAS_POISON_READY,
+	IDH_PF_SOFT_FLR_NOTIFICATION,
+	IDH_RAS_ERROR_DETECTED,
+	IDH_RAS_BAD_PAGES_READY = 15,
+	IDH_RAS_BAD_PAGES_NOTIFICATION = 16,
+	IDH_UNRECOV_ERR_NOTIFICATION = 17,
+	IDH_TEXT_MESSAGE = 255,
 };
 
 extern const struct amdgpu_virt_ops xgpu_ai_virt_ops;
@@ -51,4 +68,9 @@ int xgpu_ai_mailbox_add_irq_id(struct amdgpu_device *adev);
 int xgpu_ai_mailbox_get_irq(struct amdgpu_device *adev);
 void xgpu_ai_mailbox_put_irq(struct amdgpu_device *adev);
 
+#define AI_MAIBOX_CONTROL_TRN_OFFSET_BYTE \
+	(SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_MAILBOX_CONTROL) * 4)
+#define AI_MAIBOX_CONTROL_RCV_OFFSET_BYTE \
+	(SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_MAILBOX_CONTROL) * 4 + 1)
+
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c
new file mode 100644
index 000000000000..e5282a5d05d9
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c
@@ -0,0 +1,586 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "nbio/nbio_2_3_offset.h"
+#include "nbio/nbio_2_3_sh_mask.h"
+#include "gc/gc_10_1_0_offset.h"
+#include "gc/gc_10_1_0_sh_mask.h"
+#include "soc15.h"
+#include "navi10_ih.h"
+#include "soc15_common.h"
+#include "mxgpu_nv.h"
+
+#include "amdgpu_reset.h"
+
+static void xgpu_nv_mailbox_send_ack(struct amdgpu_device *adev)
+{
+	WREG8(NV_MAIBOX_CONTROL_RCV_OFFSET_BYTE, 2);
+}
+
+static void xgpu_nv_mailbox_set_valid(struct amdgpu_device *adev, bool val)
+{
+	WREG8(NV_MAIBOX_CONTROL_TRN_OFFSET_BYTE, val ? 1 : 0);
+}
+
+/*
+ * this peek_msg could *only* be called in IRQ routine becuase in IRQ routine
+ * RCV_MSG_VALID filed of BIF_BX_PF_MAILBOX_CONTROL must already be set to 1
+ * by host.
+ *
+ * if called no in IRQ routine, this peek_msg cannot guaranteed to return the
+ * correct value since it doesn't return the RCV_DW0 under the case that
+ * RCV_MSG_VALID is set by host.
+ */
+static enum idh_event xgpu_nv_mailbox_peek_msg(struct amdgpu_device *adev)
+{
+	return RREG32_NO_KIQ(mmMAILBOX_MSGBUF_RCV_DW0);
+}
+
+
+static int xgpu_nv_mailbox_rcv_msg(struct amdgpu_device *adev,
+				   enum idh_event event)
+{
+	int r = 0;
+	u32 reg;
+
+	reg = RREG32_NO_KIQ(mmMAILBOX_MSGBUF_RCV_DW0);
+	if (reg == IDH_FAIL)
+		r = -EINVAL;
+	if (reg == IDH_UNRECOV_ERR_NOTIFICATION)
+		r = -ENODEV;
+	else if (reg != event)
+		return -ENOENT;
+
+	xgpu_nv_mailbox_send_ack(adev);
+
+	return r;
+}
+
+static uint8_t xgpu_nv_peek_ack(struct amdgpu_device *adev)
+{
+	return RREG8(NV_MAIBOX_CONTROL_TRN_OFFSET_BYTE) & 2;
+}
+
+static int xgpu_nv_poll_ack(struct amdgpu_device *adev)
+{
+	int timeout  = NV_MAILBOX_POLL_ACK_TIMEDOUT;
+	u8 reg;
+
+	do {
+		reg = RREG8(NV_MAIBOX_CONTROL_TRN_OFFSET_BYTE);
+		if (reg & 2)
+			return 0;
+
+		mdelay(5);
+		timeout -= 5;
+	} while (timeout > 1);
+
+	dev_err(adev->dev, "Doesn't get TRN_MSG_ACK from pf in %d msec \n", NV_MAILBOX_POLL_ACK_TIMEDOUT);
+
+	return -ETIME;
+}
+
+static int xgpu_nv_poll_msg(struct amdgpu_device *adev, enum idh_event event)
+{
+	int r;
+	uint64_t timeout, now;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+
+	now = (uint64_t)ktime_to_ms(ktime_get());
+	timeout = now + NV_MAILBOX_POLL_MSG_TIMEDOUT;
+
+	do {
+		r = xgpu_nv_mailbox_rcv_msg(adev, event);
+		if (!r) {
+			dev_dbg(adev->dev, "rcv_msg 0x%x after %llu ms\n",
+					event, NV_MAILBOX_POLL_MSG_TIMEDOUT - timeout + now);
+			return 0;
+		} else if (r == -ENODEV) {
+			if (!amdgpu_ras_is_rma(adev)) {
+				ras->is_rma = true;
+				dev_err(adev->dev, "VF is in an unrecoverable state. "
+						"Runtime Services are halted.\n");
+			}
+			return r;
+		}
+
+		msleep(10);
+		now = (uint64_t)ktime_to_ms(ktime_get());
+	} while (timeout > now);
+
+	dev_dbg(adev->dev, "nv_poll_msg timed out\n");
+
+	return -ETIME;
+}
+
+static void xgpu_nv_mailbox_trans_msg (struct amdgpu_device *adev,
+	      enum idh_request req, u32 data1, u32 data2, u32 data3)
+{
+	int r;
+	uint8_t trn;
+
+	/* IMPORTANT:
+	 * clear TRN_MSG_VALID valid to clear host's RCV_MSG_ACK
+	 * and with host's RCV_MSG_ACK cleared hw automatically clear host's RCV_MSG_ACK
+	 * which lead to VF's TRN_MSG_ACK cleared, otherwise below xgpu_nv_poll_ack()
+	 * will return immediatly
+	 */
+	do {
+		xgpu_nv_mailbox_set_valid(adev, false);
+		trn = xgpu_nv_peek_ack(adev);
+		if (trn) {
+			dev_err_ratelimited(adev->dev, "trn=%x ACK should not assert! wait again !\n", trn);
+			msleep(1);
+		}
+	} while (trn);
+
+	dev_dbg(adev->dev, "trans_msg req = 0x%x, data1 = 0x%x\n", req, data1);
+	WREG32_NO_KIQ(mmMAILBOX_MSGBUF_TRN_DW0, req);
+	WREG32_NO_KIQ(mmMAILBOX_MSGBUF_TRN_DW1, data1);
+	WREG32_NO_KIQ(mmMAILBOX_MSGBUF_TRN_DW2, data2);
+	WREG32_NO_KIQ(mmMAILBOX_MSGBUF_TRN_DW3, data3);
+	xgpu_nv_mailbox_set_valid(adev, true);
+
+	/* start to poll ack */
+	r = xgpu_nv_poll_ack(adev);
+	if (r)
+		dev_err(adev->dev, "Doesn't get ack from pf, continue\n");
+
+	xgpu_nv_mailbox_set_valid(adev, false);
+}
+
+static int xgpu_nv_send_access_requests_with_param(struct amdgpu_device *adev,
+			enum idh_request req, u32 data1, u32 data2, u32 data3)
+{
+	int r, retry = 1;
+	enum idh_event event = -1;
+
+send_request:
+
+	if (amdgpu_ras_is_rma(adev))
+		return -ENODEV;
+
+	xgpu_nv_mailbox_trans_msg(adev, req, data1, data2, data3);
+
+	switch (req) {
+	case IDH_REQ_GPU_INIT_ACCESS:
+	case IDH_REQ_GPU_FINI_ACCESS:
+	case IDH_REQ_GPU_RESET_ACCESS:
+		event = IDH_READY_TO_ACCESS_GPU;
+		break;
+	case IDH_REQ_GPU_INIT_DATA:
+		event = IDH_REQ_GPU_INIT_DATA_READY;
+		break;
+	case IDH_RAS_POISON:
+		if (data1 != 0)
+			event = IDH_RAS_POISON_READY;
+		break;
+	case IDH_REQ_RAS_ERROR_COUNT:
+		event = IDH_RAS_ERROR_COUNT_READY;
+		break;
+	case IDH_REQ_RAS_CPER_DUMP:
+		event = IDH_RAS_CPER_DUMP_READY;
+		break;
+	case IDH_REQ_RAS_CHK_CRITI:
+		event = IDH_REQ_RAS_CHK_CRITI_READY;
+		break;
+	default:
+		break;
+	}
+
+	if (event != -1) {
+		r = xgpu_nv_poll_msg(adev, event);
+		if (r) {
+			if (retry++ < 5)
+				goto send_request;
+
+			if (req != IDH_REQ_GPU_INIT_DATA) {
+				dev_err(adev->dev, "Doesn't get msg:%d from pf, error=%d\n", event, r);
+				return r;
+			} else /* host doesn't support REQ_GPU_INIT_DATA handshake */
+				adev->virt.req_init_data_ver = 0;
+		} else {
+			if (req == IDH_REQ_GPU_INIT_DATA) {
+				adev->virt.req_init_data_ver =
+					RREG32_NO_KIQ(mmMAILBOX_MSGBUF_RCV_DW1);
+
+				/* assume V1 in case host doesn't set version number */
+				if (adev->virt.req_init_data_ver < 1)
+					adev->virt.req_init_data_ver = 1;
+			}
+		}
+
+		/* Retrieve checksum from mailbox2 */
+		if (req == IDH_REQ_GPU_INIT_ACCESS || req == IDH_REQ_GPU_RESET_ACCESS) {
+			adev->virt.fw_reserve.checksum_key =
+				RREG32_NO_KIQ(mmMAILBOX_MSGBUF_RCV_DW2);
+		}
+	}
+
+	return 0;
+}
+
+static int xgpu_nv_send_access_requests(struct amdgpu_device *adev,
+					enum idh_request req)
+{
+	return xgpu_nv_send_access_requests_with_param(adev,
+						req, 0, 0, 0);
+}
+
+static int xgpu_nv_request_reset(struct amdgpu_device *adev)
+{
+	int ret, i = 0;
+
+	while (i < NV_MAILBOX_POLL_MSG_REP_MAX) {
+		ret = xgpu_nv_send_access_requests(adev, IDH_REQ_GPU_RESET_ACCESS);
+		if (!ret)
+			break;
+		i++;
+	}
+
+	return ret;
+}
+
+static int xgpu_nv_request_full_gpu_access(struct amdgpu_device *adev,
+					   bool init)
+{
+	enum idh_request req;
+
+	req = init ? IDH_REQ_GPU_INIT_ACCESS : IDH_REQ_GPU_FINI_ACCESS;
+	return xgpu_nv_send_access_requests(adev, req);
+}
+
+static int xgpu_nv_release_full_gpu_access(struct amdgpu_device *adev,
+					   bool init)
+{
+	enum idh_request req;
+	int r = 0;
+
+	req = init ? IDH_REL_GPU_INIT_ACCESS : IDH_REL_GPU_FINI_ACCESS;
+	r = xgpu_nv_send_access_requests(adev, req);
+
+	return r;
+}
+
+static int xgpu_nv_request_init_data(struct amdgpu_device *adev)
+{
+	return xgpu_nv_send_access_requests(adev, IDH_REQ_GPU_INIT_DATA);
+}
+
+static int xgpu_nv_mailbox_ack_irq(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					struct amdgpu_iv_entry *entry)
+{
+	dev_dbg(adev->dev, "get ack intr and do nothing.\n");
+	return 0;
+}
+
+static int xgpu_nv_set_mailbox_ack_irq(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	u32 tmp = RREG32_NO_KIQ(mmMAILBOX_INT_CNTL);
+
+	if (state == AMDGPU_IRQ_STATE_ENABLE)
+		tmp |= 2;
+	else
+		tmp &= ~2;
+
+	WREG32_NO_KIQ(mmMAILBOX_INT_CNTL, tmp);
+
+	return 0;
+}
+
+static void xgpu_nv_ready_to_reset(struct amdgpu_device *adev)
+{
+	xgpu_nv_mailbox_trans_msg(adev, IDH_READY_TO_RESET, 0, 0, 0);
+}
+
+static int xgpu_nv_wait_reset(struct amdgpu_device *adev)
+{
+	int timeout = NV_MAILBOX_POLL_FLR_TIMEDOUT;
+	do {
+		if (xgpu_nv_mailbox_peek_msg(adev) == IDH_FLR_NOTIFICATION_CMPL) {
+			dev_dbg(adev->dev, "Got NV IDH_FLR_NOTIFICATION_CMPL after %d ms\n", NV_MAILBOX_POLL_FLR_TIMEDOUT - timeout);
+			return 0;
+		}
+		msleep(10);
+		timeout -= 10;
+	} while (timeout > 1);
+
+	dev_dbg(adev->dev, "waiting NV IDH_FLR_NOTIFICATION_CMPL timeout\n");
+	return -ETIME;
+}
+
+static void xgpu_nv_mailbox_flr_work(struct work_struct *work)
+{
+	struct amdgpu_virt *virt = container_of(work, struct amdgpu_virt, flr_work);
+	struct amdgpu_device *adev = container_of(virt, struct amdgpu_device, virt);
+	struct amdgpu_reset_context reset_context = { 0 };
+
+	amdgpu_virt_fini_data_exchange(adev);
+
+	/* Trigger recovery for world switch failure if no TDR */
+	if (amdgpu_device_should_recover_gpu(adev)
+		&& (!amdgpu_device_has_job_running(adev) ||
+		adev->sdma_timeout == MAX_SCHEDULE_TIMEOUT ||
+		adev->gfx_timeout == MAX_SCHEDULE_TIMEOUT ||
+		adev->compute_timeout == MAX_SCHEDULE_TIMEOUT ||
+		adev->video_timeout == MAX_SCHEDULE_TIMEOUT)) {
+
+		reset_context.method = AMD_RESET_METHOD_NONE;
+		reset_context.reset_req_dev = adev;
+		clear_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
+		set_bit(AMDGPU_HOST_FLR, &reset_context.flags);
+
+		amdgpu_device_gpu_recover(adev, NULL, &reset_context);
+	}
+}
+
+static void xgpu_nv_mailbox_req_bad_pages_work(struct work_struct *work)
+{
+	struct amdgpu_virt *virt = container_of(work, struct amdgpu_virt, req_bad_pages_work);
+	struct amdgpu_device *adev = container_of(virt, struct amdgpu_device, virt);
+
+	if (down_read_trylock(&adev->reset_domain->sem)) {
+		amdgpu_virt_fini_data_exchange(adev);
+		amdgpu_virt_request_bad_pages(adev);
+		up_read(&adev->reset_domain->sem);
+	}
+}
+
+/**
+ * xgpu_nv_mailbox_handle_bad_pages_work - Reinitialize the data exchange region to get fresh bad page information
+ * @work: pointer to the work_struct
+ *
+ * This work handler is triggered when bad pages are ready, and it reinitializes
+ * the data exchange region to retrieve updated bad page information from the host.
+ */
+static void xgpu_nv_mailbox_handle_bad_pages_work(struct work_struct *work)
+{
+	struct amdgpu_virt *virt = container_of(work, struct amdgpu_virt, handle_bad_pages_work);
+	struct amdgpu_device *adev = container_of(virt, struct amdgpu_device, virt);
+
+	if (down_read_trylock(&adev->reset_domain->sem)) {
+		amdgpu_virt_fini_data_exchange(adev);
+		amdgpu_virt_init_data_exchange(adev);
+		up_read(&adev->reset_domain->sem);
+	}
+}
+
+static int xgpu_nv_set_mailbox_rcv_irq(struct amdgpu_device *adev,
+				       struct amdgpu_irq_src *src,
+				       unsigned type,
+				       enum amdgpu_interrupt_state state)
+{
+	u32 tmp = RREG32_NO_KIQ(mmMAILBOX_INT_CNTL);
+
+	if (state == AMDGPU_IRQ_STATE_ENABLE)
+		tmp |= 1;
+	else
+		tmp &= ~1;
+
+	WREG32_NO_KIQ(mmMAILBOX_INT_CNTL, tmp);
+
+	return 0;
+}
+
+static int xgpu_nv_mailbox_rcv_irq(struct amdgpu_device *adev,
+				   struct amdgpu_irq_src *source,
+				   struct amdgpu_iv_entry *entry)
+{
+	enum idh_event event = xgpu_nv_mailbox_peek_msg(adev);
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+
+	switch (event) {
+	case IDH_RAS_BAD_PAGES_READY:
+		xgpu_nv_mailbox_send_ack(adev);
+		if (amdgpu_sriov_runtime(adev))
+			schedule_work(&adev->virt.handle_bad_pages_work);
+		break;
+	case IDH_RAS_BAD_PAGES_NOTIFICATION:
+		xgpu_nv_mailbox_send_ack(adev);
+		if (amdgpu_sriov_runtime(adev))
+			schedule_work(&adev->virt.req_bad_pages_work);
+		break;
+	case IDH_UNRECOV_ERR_NOTIFICATION:
+		xgpu_nv_mailbox_send_ack(adev);
+		if (!amdgpu_ras_is_rma(adev)) {
+			ras->is_rma = true;
+			dev_err(adev->dev, "VF is in an unrecoverable state. Runtime Services are halted.\n");
+		}
+
+		if (amdgpu_sriov_runtime(adev))
+			WARN_ONCE(!amdgpu_reset_domain_schedule(adev->reset_domain,
+						&adev->virt.flr_work),
+					"Failed to queue work! at %s",
+					__func__);
+		break;
+	case IDH_FLR_NOTIFICATION:
+		if (amdgpu_sriov_runtime(adev))
+			WARN_ONCE(!amdgpu_reset_domain_schedule(adev->reset_domain,
+				   &adev->virt.flr_work),
+				  "Failed to queue work! at %s",
+				  __func__);
+		break;
+		/* READY_TO_ACCESS_GPU is fetched by kernel polling, IRQ can ignore
+		 * it byfar since that polling thread will handle it,
+		 * other msg like flr complete is not handled here.
+		 */
+	case IDH_CLR_MSG_BUF:
+	case IDH_FLR_NOTIFICATION_CMPL:
+	case IDH_READY_TO_ACCESS_GPU:
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs xgpu_nv_mailbox_ack_irq_funcs = {
+	.set = xgpu_nv_set_mailbox_ack_irq,
+	.process = xgpu_nv_mailbox_ack_irq,
+};
+
+static const struct amdgpu_irq_src_funcs xgpu_nv_mailbox_rcv_irq_funcs = {
+	.set = xgpu_nv_set_mailbox_rcv_irq,
+	.process = xgpu_nv_mailbox_rcv_irq,
+};
+
+void xgpu_nv_mailbox_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->virt.ack_irq.num_types = 1;
+	adev->virt.ack_irq.funcs = &xgpu_nv_mailbox_ack_irq_funcs;
+	adev->virt.rcv_irq.num_types = 1;
+	adev->virt.rcv_irq.funcs = &xgpu_nv_mailbox_rcv_irq_funcs;
+}
+
+int xgpu_nv_mailbox_add_irq_id(struct amdgpu_device *adev)
+{
+	int r;
+
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_BIF, 135, &adev->virt.rcv_irq);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_BIF, 138, &adev->virt.ack_irq);
+	if (r) {
+		amdgpu_irq_put(adev, &adev->virt.rcv_irq, 0);
+		return r;
+	}
+
+	return 0;
+}
+
+int xgpu_nv_mailbox_get_irq(struct amdgpu_device *adev)
+{
+	int r;
+
+	r = amdgpu_irq_get(adev, &adev->virt.rcv_irq, 0);
+	if (r)
+		return r;
+	r = amdgpu_irq_get(adev, &adev->virt.ack_irq, 0);
+	if (r) {
+		amdgpu_irq_put(adev, &adev->virt.rcv_irq, 0);
+		return r;
+	}
+
+	INIT_WORK(&adev->virt.flr_work, xgpu_nv_mailbox_flr_work);
+	INIT_WORK(&adev->virt.req_bad_pages_work, xgpu_nv_mailbox_req_bad_pages_work);
+	INIT_WORK(&adev->virt.handle_bad_pages_work, xgpu_nv_mailbox_handle_bad_pages_work);
+
+	return 0;
+}
+
+void xgpu_nv_mailbox_put_irq(struct amdgpu_device *adev)
+{
+	amdgpu_irq_put(adev, &adev->virt.ack_irq, 0);
+	amdgpu_irq_put(adev, &adev->virt.rcv_irq, 0);
+}
+
+static void xgpu_nv_ras_poison_handler(struct amdgpu_device *adev,
+		enum amdgpu_ras_block block)
+{
+	if (amdgpu_ip_version(adev, UMC_HWIP, 0) < IP_VERSION(12, 0, 0)) {
+		xgpu_nv_send_access_requests(adev, IDH_RAS_POISON);
+	} else {
+		amdgpu_virt_fini_data_exchange(adev);
+		xgpu_nv_send_access_requests_with_param(adev,
+					IDH_RAS_POISON,	block, 0, 0);
+	}
+}
+
+static bool xgpu_nv_rcvd_ras_intr(struct amdgpu_device *adev)
+{
+	enum idh_event msg = xgpu_nv_mailbox_peek_msg(adev);
+
+	return (msg == IDH_RAS_ERROR_DETECTED || msg == 0xFFFFFFFF);
+}
+
+static int xgpu_nv_req_ras_err_count(struct amdgpu_device *adev)
+{
+	return xgpu_nv_send_access_requests(adev, IDH_REQ_RAS_ERROR_COUNT);
+}
+
+static int xgpu_nv_req_ras_cper_dump(struct amdgpu_device *adev, u64 vf_rptr)
+{
+	uint32_t vf_rptr_hi, vf_rptr_lo;
+
+	vf_rptr_hi = (uint32_t)(vf_rptr >> 32);
+	vf_rptr_lo = (uint32_t)(vf_rptr & 0xFFFFFFFF);
+	return xgpu_nv_send_access_requests_with_param(
+		adev, IDH_REQ_RAS_CPER_DUMP, vf_rptr_hi, vf_rptr_lo, 0);
+}
+
+static int xgpu_nv_req_ras_bad_pages(struct amdgpu_device *adev)
+{
+	return xgpu_nv_send_access_requests(adev, IDH_REQ_RAS_BAD_PAGES);
+}
+
+static int xgpu_nv_check_vf_critical_region(struct amdgpu_device *adev, u64 addr)
+{
+	uint32_t addr_hi, addr_lo;
+
+	addr_hi = (uint32_t)(addr >> 32);
+	addr_lo = (uint32_t)(addr & 0xFFFFFFFF);
+	return xgpu_nv_send_access_requests_with_param(
+		adev, IDH_REQ_RAS_CHK_CRITI, addr_hi, addr_lo, 0);
+}
+
+const struct amdgpu_virt_ops xgpu_nv_virt_ops = {
+	.req_full_gpu	= xgpu_nv_request_full_gpu_access,
+	.rel_full_gpu	= xgpu_nv_release_full_gpu_access,
+	.req_init_data  = xgpu_nv_request_init_data,
+	.reset_gpu = xgpu_nv_request_reset,
+	.ready_to_reset = xgpu_nv_ready_to_reset,
+	.wait_reset = xgpu_nv_wait_reset,
+	.trans_msg = xgpu_nv_mailbox_trans_msg,
+	.ras_poison_handler = xgpu_nv_ras_poison_handler,
+	.rcvd_ras_intr = xgpu_nv_rcvd_ras_intr,
+	.req_ras_err_count = xgpu_nv_req_ras_err_count,
+	.req_ras_cper_dump = xgpu_nv_req_ras_cper_dump,
+	.req_bad_pages = xgpu_nv_req_ras_bad_pages,
+	.req_ras_chk_criti = xgpu_nv_check_vf_critical_region
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.h b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.h
new file mode 100644
index 000000000000..c1083e5e41e0
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.h
@@ -0,0 +1,95 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __MXGPU_NV_H__
+#define __MXGPU_NV_H__
+
+#define NV_MAILBOX_POLL_ACK_TIMEDOUT	500
+#define NV_MAILBOX_POLL_MSG_TIMEDOUT	15000
+#define NV_MAILBOX_POLL_FLR_TIMEDOUT	10000
+#define NV_MAILBOX_POLL_MSG_REP_MAX	11
+
+enum idh_request {
+	IDH_REQ_GPU_INIT_ACCESS = 1,
+	IDH_REL_GPU_INIT_ACCESS,
+	IDH_REQ_GPU_FINI_ACCESS,
+	IDH_REL_GPU_FINI_ACCESS,
+	IDH_REQ_GPU_RESET_ACCESS,
+	IDH_REQ_GPU_INIT_DATA,
+
+	IDH_LOG_VF_ERROR	= 200,
+	IDH_READY_TO_RESET 	= 201,
+	IDH_RAS_POISON	= 202,
+	IDH_REQ_RAS_ERROR_COUNT = 203,
+	IDH_REQ_RAS_CPER_DUMP = 204,
+	IDH_REQ_RAS_BAD_PAGES = 205,
+	IDH_REQ_RAS_CHK_CRITI = 206
+};
+
+enum idh_event {
+	IDH_CLR_MSG_BUF	= 0,
+	IDH_READY_TO_ACCESS_GPU,
+	IDH_FLR_NOTIFICATION,
+	IDH_FLR_NOTIFICATION_CMPL,
+	IDH_SUCCESS,
+	IDH_FAIL,
+	IDH_QUERY_ALIVE,
+	IDH_REQ_GPU_INIT_DATA_READY,
+	IDH_RAS_POISON_READY,
+	IDH_PF_SOFT_FLR_NOTIFICATION,
+	IDH_RAS_ERROR_DETECTED,
+	IDH_RAS_ERROR_COUNT_READY = 11,
+	IDH_RAS_CPER_DUMP_READY = 14,
+	IDH_RAS_BAD_PAGES_READY = 15,
+	IDH_RAS_BAD_PAGES_NOTIFICATION = 16,
+	IDH_UNRECOV_ERR_NOTIFICATION = 17,
+	IDH_REQ_RAS_CHK_CRITI_READY = 18,
+
+	IDH_TEXT_MESSAGE = 255,
+};
+
+extern const struct amdgpu_virt_ops xgpu_nv_virt_ops;
+
+void xgpu_nv_mailbox_set_irq_funcs(struct amdgpu_device *adev);
+int xgpu_nv_mailbox_add_irq_id(struct amdgpu_device *adev);
+int xgpu_nv_mailbox_get_irq(struct amdgpu_device *adev);
+void xgpu_nv_mailbox_put_irq(struct amdgpu_device *adev);
+
+#define mmMAILBOX_CONTROL 0xE5E
+
+#define NV_MAIBOX_CONTROL_TRN_OFFSET_BYTE (mmMAILBOX_CONTROL * 4)
+#define NV_MAIBOX_CONTROL_RCV_OFFSET_BYTE (NV_MAIBOX_CONTROL_TRN_OFFSET_BYTE + 1)
+
+#define mmMAILBOX_MSGBUF_TRN_DW0 0xE56
+#define mmMAILBOX_MSGBUF_TRN_DW1 0xE57
+#define mmMAILBOX_MSGBUF_TRN_DW2 0xE58
+#define mmMAILBOX_MSGBUF_TRN_DW3 0xE59
+
+#define mmMAILBOX_MSGBUF_RCV_DW0 0xE5A
+#define mmMAILBOX_MSGBUF_RCV_DW1 0xE5B
+#define mmMAILBOX_MSGBUF_RCV_DW2 0xE5C
+#define mmMAILBOX_MSGBUF_RCV_DW3 0xE5D
+
+#define mmMAILBOX_INT_CNTL 0xE5F
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c b/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c
index c25a831f94ec..e1d63bed84bf 100644
--- a/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c
+++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c
@@ -37,12 +37,13 @@
 #include "gmc/gmc_8_2_sh_mask.h"
 #include "oss/oss_3_0_d.h"
 #include "oss/oss_3_0_sh_mask.h"
-#include "gca/gfx_8_0_sh_mask.h"
 #include "dce/dce_10_0_d.h"
 #include "dce/dce_10_0_sh_mask.h"
 #include "smu/smu_7_1_3_d.h"
 #include "mxgpu_vi.h"
 
+#include "amdgpu_reset.h"
+
 /* VI golden setting */
 static const u32 xgpu_fiji_mgcg_cgcg_init[] = {
 	mmRLC_CGTT_MGCG_OVERRIDE, 0xffffffff, 0xffffffff,
@@ -279,32 +280,32 @@ void xgpu_vi_init_golden_registers(struct amdgpu_device *adev)
 {
 	switch (adev->asic_type) {
 	case CHIP_FIJI:
-		amdgpu_program_register_sequence(adev,
-						 xgpu_fiji_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(
-						 xgpu_fiji_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 xgpu_fiji_golden_settings_a10,
-						 (const u32)ARRAY_SIZE(
-						 xgpu_fiji_golden_settings_a10));
-		amdgpu_program_register_sequence(adev,
-						 xgpu_fiji_golden_common_all,
-						 (const u32)ARRAY_SIZE(
-						 xgpu_fiji_golden_common_all));
+		amdgpu_device_program_register_sequence(adev,
+							xgpu_fiji_mgcg_cgcg_init,
+							ARRAY_SIZE(
+								xgpu_fiji_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							xgpu_fiji_golden_settings_a10,
+							ARRAY_SIZE(
+								xgpu_fiji_golden_settings_a10));
+		amdgpu_device_program_register_sequence(adev,
+							xgpu_fiji_golden_common_all,
+							ARRAY_SIZE(
+								xgpu_fiji_golden_common_all));
 		break;
 	case CHIP_TONGA:
-		amdgpu_program_register_sequence(adev,
-						 xgpu_tonga_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(
-						 xgpu_tonga_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 xgpu_tonga_golden_settings_a11,
-						 (const u32)ARRAY_SIZE(
-						 xgpu_tonga_golden_settings_a11));
-		amdgpu_program_register_sequence(adev,
-						 xgpu_tonga_golden_common_all,
-						 (const u32)ARRAY_SIZE(
-						 xgpu_tonga_golden_common_all));
+		amdgpu_device_program_register_sequence(adev,
+							xgpu_tonga_mgcg_cgcg_init,
+							ARRAY_SIZE(
+								xgpu_tonga_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							xgpu_tonga_golden_settings_a11,
+							ARRAY_SIZE(
+								xgpu_tonga_golden_settings_a11));
+		amdgpu_device_program_register_sequence(adev,
+							xgpu_tonga_golden_common_all,
+							ARRAY_SIZE(
+								xgpu_tonga_golden_common_all));
 		break;
 	default:
 		BUG_ON("Doesn't support chip type.\n");
@@ -333,7 +334,7 @@ static void xgpu_vi_mailbox_send_ack(struct amdgpu_device *adev)
 			break;
 		}
 		mdelay(1);
-		timeout -=1;
+		timeout -= 1;
 
 		reg = RREG32_NO_KIQ(mmMAILBOX_CONTROL);
 	}
@@ -446,8 +447,10 @@ static int xgpu_vi_send_access_requests(struct amdgpu_device *adev,
 		request == IDH_REQ_GPU_FINI_ACCESS ||
 		request == IDH_REQ_GPU_RESET_ACCESS) {
 		r = xgpu_vi_poll_msg(adev, IDH_READY_TO_ACCESS_GPU);
-		if (r)
-			pr_err("Doesn't get ack from pf, continue\n");
+		if (r) {
+			pr_err("Doesn't get ack from pf, give up\n");
+			return r;
+		}
 	}
 
 	return 0;
@@ -458,6 +461,11 @@ static int xgpu_vi_request_reset(struct amdgpu_device *adev)
 	return xgpu_vi_send_access_requests(adev, IDH_REQ_GPU_RESET_ACCESS);
 }
 
+static int xgpu_vi_wait_reset_cmpl(struct amdgpu_device *adev)
+{
+	return xgpu_vi_poll_msg(adev, IDH_FLR_NOTIFICATION_CMPL);
+}
+
 static int xgpu_vi_request_full_gpu_access(struct amdgpu_device *adev,
 					   bool init)
 {
@@ -507,14 +515,18 @@ static void xgpu_vi_mailbox_flr_work(struct work_struct *work)
 	struct amdgpu_virt *virt = container_of(work, struct amdgpu_virt, flr_work);
 	struct amdgpu_device *adev = container_of(virt, struct amdgpu_device, virt);
 
-	/* wait until RCV_MSG become 3 */
-	if (xgpu_vi_poll_msg(adev, IDH_FLR_NOTIFICATION_CMPL)) {
-		pr_err("failed to recieve FLR_CMPL\n");
-		return;
-	}
-
 	/* Trigger recovery due to world switch failure */
-	amdgpu_sriov_gpu_reset(adev, NULL);
+	if (amdgpu_device_should_recover_gpu(adev)) {
+		struct amdgpu_reset_context reset_context;
+		memset(&reset_context, 0, sizeof(reset_context));
+
+		reset_context.method = AMD_RESET_METHOD_NONE;
+		reset_context.reset_req_dev = adev;
+		clear_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
+		set_bit(AMDGPU_HOST_FLR, &reset_context.flags);
+
+		amdgpu_device_gpu_recover(adev, NULL, &reset_context);
+	}
 }
 
 static int xgpu_vi_set_mailbox_rcv_irq(struct amdgpu_device *adev,
@@ -537,14 +549,17 @@ static int xgpu_vi_mailbox_rcv_irq(struct amdgpu_device *adev,
 {
 	int r;
 
-	/* trigger gpu-reset by hypervisor only if TDR disbaled */
-	if (amdgpu_lockup_timeout == 0) {
+	/* trigger gpu-reset by hypervisor only if TDR disabled */
+	if (!amdgpu_gpu_recovery) {
 		/* see what event we get */
 		r = xgpu_vi_mailbox_rcv_msg(adev, IDH_FLR_NOTIFICATION);
 
 		/* only handle FLR_NOTIFY now */
 		if (!r)
-			schedule_work(&adev->virt.flr_work);
+			WARN_ONCE(!amdgpu_reset_domain_schedule(adev->reset_domain,
+								&adev->virt.flr_work),
+				  "Failed to queue work! at %s",
+				  __func__);
 	}
 
 	return 0;
@@ -572,11 +587,11 @@ int xgpu_vi_mailbox_add_irq_id(struct amdgpu_device *adev)
 {
 	int r;
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 135, &adev->virt.rcv_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 135, &adev->virt.rcv_irq);
 	if (r)
 		return r;
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 138, &adev->virt.ack_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 138, &adev->virt.ack_irq);
 	if (r) {
 		amdgpu_irq_put(adev, &adev->virt.rcv_irq, 0);
 		return r;
@@ -613,5 +628,6 @@ const struct amdgpu_virt_ops xgpu_vi_virt_ops = {
 	.req_full_gpu		= xgpu_vi_request_full_gpu_access,
 	.rel_full_gpu		= xgpu_vi_release_full_gpu_access,
 	.reset_gpu		= xgpu_vi_request_reset,
+	.wait_reset             = xgpu_vi_wait_reset_cmpl,
 	.trans_msg		= NULL, /* Does not need to trans VF errors to host. */
 };
diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.h b/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.h
index c791d73d2d54..713ee66a4d3e 100644
--- a/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.h
+++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.h
@@ -23,7 +23,7 @@
 #ifndef __MXGPU_VI_H__
 #define __MXGPU_VI_H__
 
-#define VI_MAILBOX_TIMEDOUT	5000
+#define VI_MAILBOX_TIMEDOUT	12000
 #define VI_MAILBOX_RESET_TIME	12
 
 /* VI mailbox messages request */
@@ -43,7 +43,8 @@ enum idh_event {
 	IDH_READY_TO_ACCESS_GPU,
 	IDH_FLR_NOTIFICATION,
 	IDH_FLR_NOTIFICATION_CMPL,
-	IDH_EVENT_MAX
+
+	IDH_TEXT_MESSAGE = 255
 };
 
 extern const struct amdgpu_virt_ops xgpu_vi_virt_ops;
diff --git a/drivers/gpu/drm/amd/amdgpu/navi10_ih.c b/drivers/gpu/drm/amd/amdgpu/navi10_ih.c
new file mode 100644
index 000000000000..4cd325149b63
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/navi10_ih.c
@@ -0,0 +1,729 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/pci.h>
+
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+
+#include "oss/osssys_5_0_0_offset.h"
+#include "oss/osssys_5_0_0_sh_mask.h"
+
+#include "soc15_common.h"
+#include "navi10_ih.h"
+
+#define MAX_REARM_RETRY 10
+
+#define mmIH_CHICKEN_Sienna_Cichlid                 0x018d
+#define mmIH_CHICKEN_Sienna_Cichlid_BASE_IDX        0
+
+static void navi10_ih_set_interrupt_funcs(struct amdgpu_device *adev);
+
+/**
+ * navi10_ih_init_register_offset - Initialize register offset for ih rings
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialize register offset ih rings (NAVI10).
+ */
+static void navi10_ih_init_register_offset(struct amdgpu_device *adev)
+{
+	struct amdgpu_ih_regs *ih_regs;
+
+	if (adev->irq.ih.ring_size) {
+		ih_regs = &adev->irq.ih.ih_regs;
+		ih_regs->ih_rb_base = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_BASE);
+		ih_regs->ih_rb_base_hi = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_BASE_HI);
+		ih_regs->ih_rb_cntl = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL);
+		ih_regs->ih_rb_wptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR);
+		ih_regs->ih_rb_rptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_RPTR);
+		ih_regs->ih_doorbell_rptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_DOORBELL_RPTR);
+		ih_regs->ih_rb_wptr_addr_lo = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR_ADDR_LO);
+		ih_regs->ih_rb_wptr_addr_hi = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR_ADDR_HI);
+		ih_regs->psp_reg_id = PSP_REG_IH_RB_CNTL;
+	}
+
+	if (adev->irq.ih1.ring_size) {
+		ih_regs = &adev->irq.ih1.ih_regs;
+		ih_regs->ih_rb_base = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_BASE_RING1);
+		ih_regs->ih_rb_base_hi = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_BASE_HI_RING1);
+		ih_regs->ih_rb_cntl = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL_RING1);
+		ih_regs->ih_rb_wptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR_RING1);
+		ih_regs->ih_rb_rptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_RPTR_RING1);
+		ih_regs->ih_doorbell_rptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_DOORBELL_RPTR_RING1);
+		ih_regs->psp_reg_id = PSP_REG_IH_RB_CNTL_RING1;
+	}
+
+	if (adev->irq.ih2.ring_size) {
+		ih_regs = &adev->irq.ih2.ih_regs;
+		ih_regs->ih_rb_base = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_BASE_RING2);
+		ih_regs->ih_rb_base_hi = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_BASE_HI_RING2);
+		ih_regs->ih_rb_cntl = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL_RING2);
+		ih_regs->ih_rb_wptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR_RING2);
+		ih_regs->ih_rb_rptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_RPTR_RING2);
+		ih_regs->ih_doorbell_rptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_DOORBELL_RPTR_RING2);
+		ih_regs->psp_reg_id = PSP_REG_IH_RB_CNTL_RING2;
+	}
+}
+
+/**
+ * force_update_wptr_for_self_int - Force update the wptr for self interrupt
+ *
+ * @adev: amdgpu_device pointer
+ * @threshold: threshold to trigger the wptr reporting
+ * @timeout: timeout to trigger the wptr reporting
+ * @enabled: Enable/disable timeout flush mechanism
+ *
+ * threshold input range: 0 ~ 15, default 0,
+ * real_threshold = 2^threshold
+ * timeout input range: 0 ~ 20, default 8,
+ * real_timeout = (2^timeout) * 1024 / (socclk_freq)
+ *
+ * Force update wptr for self interrupt ( >= SIENNA_CICHLID).
+ */
+static void
+force_update_wptr_for_self_int(struct amdgpu_device *adev,
+			       u32 threshold, u32 timeout, bool enabled)
+{
+	u32 ih_cntl, ih_rb_cntl;
+
+	if (amdgpu_ip_version(adev, OSSSYS_HWIP, 0) < IP_VERSION(5, 0, 3))
+		return;
+
+	ih_cntl = RREG32_SOC15(OSSSYS, 0, mmIH_CNTL2);
+	ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL_RING1);
+
+	ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL2,
+				SELF_IV_FORCE_WPTR_UPDATE_TIMEOUT, timeout);
+	ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL2,
+				SELF_IV_FORCE_WPTR_UPDATE_ENABLE, enabled);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL_RING1,
+				   RB_USED_INT_THRESHOLD, threshold);
+
+	if (amdgpu_sriov_vf(adev) && amdgpu_sriov_reg_indirect_ih(adev)) {
+		if (psp_reg_program(&adev->psp, PSP_REG_IH_RB_CNTL_RING1, ih_rb_cntl))
+			return;
+	} else {
+		WREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL_RING1, ih_rb_cntl);
+	}
+
+	ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL_RING2);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL_RING2,
+				   RB_USED_INT_THRESHOLD, threshold);
+	if (amdgpu_sriov_vf(adev) && amdgpu_sriov_reg_indirect_ih(adev)) {
+		if (psp_reg_program(&adev->psp, PSP_REG_IH_RB_CNTL_RING2, ih_rb_cntl))
+			return;
+	} else {
+		WREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL_RING2, ih_rb_cntl);
+	}
+
+	WREG32_SOC15(OSSSYS, 0, mmIH_CNTL2, ih_cntl);
+}
+
+/**
+ * navi10_ih_toggle_ring_interrupts - toggle the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointet
+ * @enable: true - enable the interrupts, false - disable the interrupts
+ *
+ * Toggle the interrupt ring buffer (NAVI10)
+ */
+static int navi10_ih_toggle_ring_interrupts(struct amdgpu_device *adev,
+					    struct amdgpu_ih_ring *ih,
+					    bool enable)
+{
+	struct amdgpu_ih_regs *ih_regs;
+	uint32_t tmp;
+
+	ih_regs = &ih->ih_regs;
+
+	tmp = RREG32(ih_regs->ih_rb_cntl);
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RB_ENABLE, (enable ? 1 : 0));
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RB_GPU_TS_ENABLE, 1);
+	/* enable_intr field is only valid in ring0 */
+	if (ih == &adev->irq.ih)
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, ENABLE_INTR, (enable ? 1 : 0));
+
+	if (amdgpu_sriov_vf(adev) && amdgpu_sriov_reg_indirect_ih(adev)) {
+		if (psp_reg_program(&adev->psp, ih_regs->psp_reg_id, tmp))
+			return -ETIMEDOUT;
+	} else {
+		WREG32(ih_regs->ih_rb_cntl, tmp);
+	}
+
+	if (enable) {
+		ih->enabled = true;
+	} else {
+		/* set rptr, wptr to 0 */
+		WREG32(ih_regs->ih_rb_rptr, 0);
+		WREG32(ih_regs->ih_rb_wptr, 0);
+		ih->enabled = false;
+		ih->rptr = 0;
+	}
+
+	return 0;
+}
+
+/**
+ * navi10_ih_toggle_interrupts - Toggle all the available interrupt ring buffers
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable or disable interrupt ring buffers
+ *
+ * Toggle all the available interrupt ring buffers (NAVI10).
+ */
+static int navi10_ih_toggle_interrupts(struct amdgpu_device *adev, bool enable)
+{
+	struct amdgpu_ih_ring *ih[] = {&adev->irq.ih, &adev->irq.ih1, &adev->irq.ih2};
+	int i;
+	int r;
+
+	for (i = 0; i < ARRAY_SIZE(ih); i++) {
+		if (ih[i]->ring_size) {
+			r = navi10_ih_toggle_ring_interrupts(adev, ih[i], enable);
+			if (r)
+				return r;
+		}
+	}
+
+	return 0;
+}
+
+static uint32_t navi10_ih_rb_cntl(struct amdgpu_ih_ring *ih, uint32_t ih_rb_cntl)
+{
+	int rb_bufsz = order_base_2(ih->ring_size / 4);
+
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   MC_SPACE, ih->use_bus_addr ? 1 : 4);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   WPTR_OVERFLOW_CLEAR, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   WPTR_OVERFLOW_ENABLE, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_SIZE, rb_bufsz);
+	/* Ring Buffer write pointer writeback. If enabled, IH_RB_WPTR register
+	 * value is written to memory
+	 */
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   WPTR_WRITEBACK_ENABLE, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_SNOOP, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_RO, 0);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_VMID, 0);
+
+	return ih_rb_cntl;
+}
+
+static uint32_t navi10_ih_doorbell_rptr(struct amdgpu_ih_ring *ih)
+{
+	u32 ih_doorbell_rtpr = 0;
+
+	if (ih->use_doorbell) {
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+						 IH_DOORBELL_RPTR, OFFSET,
+						 ih->doorbell_index);
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+						 IH_DOORBELL_RPTR,
+						 ENABLE, 1);
+	} else {
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+						 IH_DOORBELL_RPTR,
+						 ENABLE, 0);
+	}
+	return ih_doorbell_rtpr;
+}
+
+/**
+ * navi10_ih_enable_ring - enable an ih ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointer
+ *
+ * Enable an ih ring buffer (NAVI10)
+ */
+static int navi10_ih_enable_ring(struct amdgpu_device *adev,
+				 struct amdgpu_ih_ring *ih)
+{
+	struct amdgpu_ih_regs *ih_regs;
+	uint32_t tmp;
+
+	ih_regs = &ih->ih_regs;
+
+	/* Ring Buffer base. [39:8] of 40-bit address of the beginning of the ring buffer*/
+	WREG32(ih_regs->ih_rb_base, ih->gpu_addr >> 8);
+	WREG32(ih_regs->ih_rb_base_hi, (ih->gpu_addr >> 40) & 0xff);
+
+	tmp = RREG32(ih_regs->ih_rb_cntl);
+	tmp = navi10_ih_rb_cntl(ih, tmp);
+	if (ih == &adev->irq.ih)
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RPTR_REARM, !!adev->irq.msi_enabled);
+	if (ih == &adev->irq.ih1)
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RB_FULL_DRAIN_ENABLE, 1);
+
+	if (amdgpu_sriov_vf(adev) && amdgpu_sriov_reg_indirect_ih(adev)) {
+		if (psp_reg_program(&adev->psp, ih_regs->psp_reg_id, tmp)) {
+			DRM_ERROR("PSP program IH_RB_CNTL failed!\n");
+			return -ETIMEDOUT;
+		}
+	} else {
+		WREG32(ih_regs->ih_rb_cntl, tmp);
+	}
+
+	if (ih == &adev->irq.ih) {
+		/* set the ih ring 0 writeback address whether it's enabled or not */
+		WREG32(ih_regs->ih_rb_wptr_addr_lo, lower_32_bits(ih->wptr_addr));
+		WREG32(ih_regs->ih_rb_wptr_addr_hi, upper_32_bits(ih->wptr_addr) & 0xFFFF);
+	}
+
+	/* set rptr, wptr to 0 */
+	WREG32(ih_regs->ih_rb_wptr, 0);
+	WREG32(ih_regs->ih_rb_rptr, 0);
+
+	WREG32(ih_regs->ih_doorbell_rptr, navi10_ih_doorbell_rptr(ih));
+
+	return 0;
+}
+
+/**
+ * navi10_ih_irq_init - init and enable the interrupt ring
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate a ring buffer for the interrupt controller,
+ * enable the RLC, disable interrupts, enable the IH
+ * ring buffer and enable it (NAVI).
+ * Called at device load and reume.
+ * Returns 0 for success, errors for failure.
+ */
+static int navi10_ih_irq_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_ih_ring *ih[] = {&adev->irq.ih, &adev->irq.ih1, &adev->irq.ih2};
+	u32 ih_chicken;
+	int ret;
+	int i;
+
+	/* disable irqs */
+	ret = navi10_ih_toggle_interrupts(adev, false);
+	if (ret)
+		return ret;
+
+	adev->nbio.funcs->ih_control(adev);
+
+	if (unlikely(adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT)) {
+		if (ih[0]->use_bus_addr) {
+			switch (amdgpu_ip_version(adev, OSSSYS_HWIP, 0)) {
+			case IP_VERSION(5, 0, 3):
+			case IP_VERSION(5, 2, 0):
+			case IP_VERSION(5, 2, 1):
+				ih_chicken = RREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN_Sienna_Cichlid);
+				ih_chicken = REG_SET_FIELD(ih_chicken,
+						IH_CHICKEN, MC_SPACE_GPA_ENABLE, 1);
+				WREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN_Sienna_Cichlid, ih_chicken);
+				break;
+			default:
+				ih_chicken = RREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN);
+				ih_chicken = REG_SET_FIELD(ih_chicken,
+						IH_CHICKEN, MC_SPACE_GPA_ENABLE, 1);
+				WREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN, ih_chicken);
+				break;
+			}
+		}
+	}
+
+	for (i = 0; i < ARRAY_SIZE(ih); i++) {
+		if (ih[i]->ring_size) {
+			ret = navi10_ih_enable_ring(adev, ih[i]);
+			if (ret)
+				return ret;
+		}
+	}
+
+	/* update doorbell range for ih ring 0*/
+	adev->nbio.funcs->ih_doorbell_range(adev, ih[0]->use_doorbell,
+					    ih[0]->doorbell_index);
+
+	pci_set_master(adev->pdev);
+
+	/* enable interrupts */
+	ret = navi10_ih_toggle_interrupts(adev, true);
+	if (ret)
+		return ret;
+	/* enable wptr force update for self int */
+	force_update_wptr_for_self_int(adev, 0, 8, true);
+
+	if (adev->irq.ih_soft.ring_size)
+		adev->irq.ih_soft.enabled = true;
+
+	return 0;
+}
+
+/**
+ * navi10_ih_irq_disable - disable interrupts
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable interrupts on the hw (NAVI10).
+ */
+static void navi10_ih_irq_disable(struct amdgpu_device *adev)
+{
+	force_update_wptr_for_self_int(adev, 0, 8, false);
+	navi10_ih_toggle_interrupts(adev, false);
+
+	/* Wait and acknowledge irq */
+	mdelay(1);
+}
+
+/**
+ * navi10_ih_get_wptr - get the IH ring buffer wptr
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: IH ring buffer to fetch wptr
+ *
+ * Get the IH ring buffer wptr from either the register
+ * or the writeback memory buffer (NAVI10).  Also check for
+ * ring buffer overflow and deal with it.
+ * Returns the value of the wptr.
+ */
+static u32 navi10_ih_get_wptr(struct amdgpu_device *adev,
+			      struct amdgpu_ih_ring *ih)
+{
+	u32 wptr, tmp;
+	struct amdgpu_ih_regs *ih_regs;
+
+	if (ih == &adev->irq.ih || ih == &adev->irq.ih_soft) {
+		/* Only ring0 supports writeback. On other rings fall back
+		 * to register-based code with overflow checking below.
+		 * ih_soft ring doesn't have any backing hardware registers,
+		 * update wptr and return.
+		 */
+		wptr = le32_to_cpu(*ih->wptr_cpu);
+
+		if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
+			goto out;
+	}
+
+	ih_regs = &ih->ih_regs;
+
+	/* Double check that the overflow wasn't already cleared. */
+	wptr = RREG32_NO_KIQ(ih_regs->ih_rb_wptr);
+	if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
+		goto out;
+	wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
+
+	/* When a ring buffer overflow happen start parsing interrupt
+	 * from the last not overwritten vector (wptr + 32). Hopefully
+	 * this should allow us to catch up.
+	 */
+	tmp = (wptr + 32) & ih->ptr_mask;
+	dev_warn(adev->dev, "%s ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
+		 amdgpu_ih_ring_name(adev, ih), wptr, ih->rptr, tmp);
+	ih->rptr = tmp;
+
+	tmp = RREG32_NO_KIQ(ih_regs->ih_rb_cntl);
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+	WREG32_NO_KIQ(ih_regs->ih_rb_cntl, tmp);
+
+	/* Unset the CLEAR_OVERFLOW bit immediately so new overflows
+	 * can be detected.
+	 */
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 0);
+	WREG32_NO_KIQ(ih_regs->ih_rb_cntl, tmp);
+out:
+	return (wptr & ih->ptr_mask);
+}
+
+/**
+ * navi10_ih_irq_rearm - rearm IRQ if lost
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: IH ring to match
+ *
+ */
+static void navi10_ih_irq_rearm(struct amdgpu_device *adev,
+			       struct amdgpu_ih_ring *ih)
+{
+	uint32_t v = 0;
+	uint32_t i = 0;
+	struct amdgpu_ih_regs *ih_regs;
+
+	ih_regs = &ih->ih_regs;
+
+	/* Rearm IRQ / re-write doorbell if doorbell write is lost */
+	for (i = 0; i < MAX_REARM_RETRY; i++) {
+		v = RREG32_NO_KIQ(ih_regs->ih_rb_rptr);
+		if ((v < ih->ring_size) && (v != ih->rptr))
+			WDOORBELL32(ih->doorbell_index, ih->rptr);
+		else
+			break;
+	}
+}
+
+/**
+ * navi10_ih_set_rptr - set the IH ring buffer rptr
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * @ih: IH ring buffer to set rptr
+ * Set the IH ring buffer rptr.
+ */
+static void navi10_ih_set_rptr(struct amdgpu_device *adev,
+			       struct amdgpu_ih_ring *ih)
+{
+	struct amdgpu_ih_regs *ih_regs;
+
+	if (ih == &adev->irq.ih_soft)
+		return;
+
+	if (ih->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		*ih->rptr_cpu = ih->rptr;
+		WDOORBELL32(ih->doorbell_index, ih->rptr);
+
+		if (amdgpu_sriov_vf(adev))
+			navi10_ih_irq_rearm(adev, ih);
+	} else {
+		ih_regs = &ih->ih_regs;
+		WREG32(ih_regs->ih_rb_rptr, ih->rptr);
+	}
+}
+
+/**
+ * navi10_ih_self_irq - dispatch work for ring 1 and 2
+ *
+ * @adev: amdgpu_device pointer
+ * @source: irq source
+ * @entry: IV with WPTR update
+ *
+ * Update the WPTR from the IV and schedule work to handle the entries.
+ */
+static int navi10_ih_self_irq(struct amdgpu_device *adev,
+			      struct amdgpu_irq_src *source,
+			      struct amdgpu_iv_entry *entry)
+{
+	switch (entry->ring_id) {
+	case 1:
+		schedule_work(&adev->irq.ih1_work);
+		break;
+	case 2:
+		schedule_work(&adev->irq.ih2_work);
+		break;
+	default: break;
+	}
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs navi10_ih_self_irq_funcs = {
+	.process = navi10_ih_self_irq,
+};
+
+static void navi10_ih_set_self_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->irq.self_irq.num_types = 0;
+	adev->irq.self_irq.funcs = &navi10_ih_self_irq_funcs;
+}
+
+static int navi10_ih_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	navi10_ih_set_interrupt_funcs(adev);
+	navi10_ih_set_self_irq_funcs(adev);
+	return 0;
+}
+
+static int navi10_ih_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+	bool use_bus_addr;
+
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_IH, 0,
+				&adev->irq.self_irq);
+
+	if (r)
+		return r;
+
+	/* use gpu virtual address for ih ring
+	 * until ih_checken is programmed to allow
+	 * use bus address for ih ring by psp bl */
+	if ((adev->flags & AMD_IS_APU) ||
+	    (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP))
+		use_bus_addr = false;
+	else
+		use_bus_addr = true;
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih, IH_RING_SIZE, use_bus_addr);
+	if (r)
+		return r;
+
+	adev->irq.ih.use_doorbell = true;
+	adev->irq.ih.doorbell_index = adev->doorbell_index.ih << 1;
+
+	adev->irq.ih1.ring_size = 0;
+	adev->irq.ih2.ring_size = 0;
+
+	/* initialize ih control registers offset */
+	navi10_ih_init_register_offset(adev);
+
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih_soft, IH_SW_RING_SIZE, true);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_init(adev);
+
+	return r;
+}
+
+static int navi10_ih_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	amdgpu_irq_fini_sw(adev);
+
+	return 0;
+}
+
+static int navi10_ih_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	return navi10_ih_irq_init(adev);
+}
+
+static int navi10_ih_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	navi10_ih_irq_disable(ip_block->adev);
+
+	return 0;
+}
+
+static int navi10_ih_suspend(struct amdgpu_ip_block *ip_block)
+{
+	return navi10_ih_hw_fini(ip_block);
+}
+
+static int navi10_ih_resume(struct amdgpu_ip_block *ip_block)
+{
+	return navi10_ih_hw_init(ip_block);
+}
+
+static bool navi10_ih_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	/* todo */
+	return true;
+}
+
+static int navi10_ih_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	/* todo */
+	return -ETIMEDOUT;
+}
+
+static int navi10_ih_soft_reset(struct amdgpu_ip_block *ip_block)
+{
+	/* todo */
+	return 0;
+}
+
+static void navi10_ih_update_clockgating_state(struct amdgpu_device *adev,
+					       bool enable)
+{
+	uint32_t data, def, field_val;
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_IH_CG) {
+		def = data = RREG32_SOC15(OSSSYS, 0, mmIH_CLK_CTRL);
+		field_val = enable ? 0 : 1;
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     DBUS_MUX_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     OSSSYS_SHARE_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     LIMIT_SMN_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     DYN_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     REG_CLK_SOFT_OVERRIDE, field_val);
+		if (def != data)
+			WREG32_SOC15(OSSSYS, 0, mmIH_CLK_CTRL, data);
+	}
+}
+
+static int navi10_ih_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					   enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	navi10_ih_update_clockgating_state(adev,
+				state == AMD_CG_STATE_GATE);
+	return 0;
+}
+
+static int navi10_ih_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					   enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static void navi10_ih_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (!RREG32_SOC15(OSSSYS, 0, mmIH_CLK_CTRL))
+		*flags |= AMD_CG_SUPPORT_IH_CG;
+}
+
+static const struct amd_ip_funcs navi10_ih_ip_funcs = {
+	.name = "navi10_ih",
+	.early_init = navi10_ih_early_init,
+	.sw_init = navi10_ih_sw_init,
+	.sw_fini = navi10_ih_sw_fini,
+	.hw_init = navi10_ih_hw_init,
+	.hw_fini = navi10_ih_hw_fini,
+	.suspend = navi10_ih_suspend,
+	.resume = navi10_ih_resume,
+	.is_idle = navi10_ih_is_idle,
+	.wait_for_idle = navi10_ih_wait_for_idle,
+	.soft_reset = navi10_ih_soft_reset,
+	.set_clockgating_state = navi10_ih_set_clockgating_state,
+	.set_powergating_state = navi10_ih_set_powergating_state,
+	.get_clockgating_state = navi10_ih_get_clockgating_state,
+};
+
+static const struct amdgpu_ih_funcs navi10_ih_funcs = {
+	.get_wptr = navi10_ih_get_wptr,
+	.decode_iv = amdgpu_ih_decode_iv_helper,
+	.decode_iv_ts = amdgpu_ih_decode_iv_ts_helper,
+	.set_rptr = navi10_ih_set_rptr
+};
+
+static void navi10_ih_set_interrupt_funcs(struct amdgpu_device *adev)
+{
+	if (adev->irq.ih_funcs == NULL)
+		adev->irq.ih_funcs = &navi10_ih_funcs;
+}
+
+const struct amdgpu_ip_block_version navi10_ih_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_IH,
+	.major = 5,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &navi10_ih_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/navi10_ih.h b/drivers/gpu/drm/amd/amdgpu/navi10_ih.h
new file mode 100644
index 000000000000..140fbdaaed17
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/navi10_ih.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __NAVI10_IH_H__
+#define __NAVI10_IH_H__
+
+extern const struct amdgpu_ip_block_version navi10_ih_ip_block;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/navi10_sdma_pkt_open.h b/drivers/gpu/drm/amd/amdgpu/navi10_sdma_pkt_open.h
new file mode 100644
index 000000000000..c88284ff92d8
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/navi10_sdma_pkt_open.h
@@ -0,0 +1,4886 @@
+/*
+ * Copyright (C) 2019  Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __NAVI10_SDMA_PKT_OPEN_H_
+#define __NAVI10_SDMA_PKT_OPEN_H_
+
+#define SDMA_OP_NOP  0
+#define SDMA_OP_COPY  1
+#define SDMA_OP_WRITE  2
+#define SDMA_OP_INDIRECT  4
+#define SDMA_OP_FENCE  5
+#define SDMA_OP_TRAP  6
+#define SDMA_OP_SEM  7
+#define SDMA_OP_POLL_REGMEM  8
+#define SDMA_OP_COND_EXE  9
+#define SDMA_OP_ATOMIC  10
+#define SDMA_OP_CONST_FILL  11
+#define SDMA_OP_PTEPDE  12
+#define SDMA_OP_TIMESTAMP  13
+#define SDMA_OP_SRBM_WRITE  14
+#define SDMA_OP_PRE_EXE  15
+#define SDMA_OP_GPUVM_INV  16
+#define SDMA_OP_GCR_REQ  17
+#define SDMA_OP_DUMMY_TRAP  32
+#define SDMA_SUBOP_TIMESTAMP_SET  0
+#define SDMA_SUBOP_TIMESTAMP_GET  1
+#define SDMA_SUBOP_TIMESTAMP_GET_GLOBAL  2
+#define SDMA_SUBOP_COPY_LINEAR  0
+#define SDMA_SUBOP_COPY_LINEAR_SUB_WIND  4
+#define SDMA_SUBOP_COPY_TILED  1
+#define SDMA_SUBOP_COPY_TILED_SUB_WIND  5
+#define SDMA_SUBOP_COPY_T2T_SUB_WIND  6
+#define SDMA_SUBOP_COPY_SOA  3
+#define SDMA_SUBOP_COPY_DIRTY_PAGE  7
+#define SDMA_SUBOP_COPY_LINEAR_PHY  8
+#define SDMA_SUBOP_COPY_LINEAR_BC  16
+#define SDMA_SUBOP_COPY_TILED_BC  17
+#define SDMA_SUBOP_COPY_LINEAR_SUB_WIND_BC  20
+#define SDMA_SUBOP_COPY_TILED_SUB_WIND_BC  21
+#define SDMA_SUBOP_COPY_T2T_SUB_WIND_BC  22
+#define SDMA_SUBOP_WRITE_LINEAR  0
+#define SDMA_SUBOP_WRITE_TILED  1
+#define SDMA_SUBOP_WRITE_TILED_BC  17
+#define SDMA_SUBOP_PTEPDE_GEN  0
+#define SDMA_SUBOP_PTEPDE_COPY  1
+#define SDMA_SUBOP_PTEPDE_RMW  2
+#define SDMA_SUBOP_PTEPDE_COPY_BACKWARDS  3
+#define SDMA_SUBOP_DATA_FILL_MULTI  1
+#define SDMA_SUBOP_POLL_REG_WRITE_MEM  1
+#define SDMA_SUBOP_POLL_DBIT_WRITE_MEM  2
+#define SDMA_SUBOP_POLL_MEM_VERIFY  3
+#define SDMA_SUBOP_VM_INVALIDATION  4
+#define HEADER_AGENT_DISPATCH  4
+#define HEADER_BARRIER  5
+#define SDMA_OP_AQL_COPY  0
+#define SDMA_OP_AQL_BARRIER_OR  0
+
+#define SDMA_GCR_RANGE_IS_PA		(1 << 18)
+#define SDMA_GCR_SEQ(x)			(((x) & 0x3) << 16)
+#define SDMA_GCR_GL2_WB			(1 << 15)
+#define SDMA_GCR_GL2_INV		(1 << 14)
+#define SDMA_GCR_GL2_DISCARD		(1 << 13)
+#define SDMA_GCR_GL2_RANGE(x)		(((x) & 0x3) << 11)
+#define SDMA_GCR_GL2_US			(1 << 10)
+#define SDMA_GCR_GL1_INV		(1 << 9)
+#define SDMA_GCR_GLV_INV		(1 << 8)
+#define SDMA_GCR_GLK_INV		(1 << 7)
+#define SDMA_GCR_GLK_WB			(1 << 6)
+#define SDMA_GCR_GLM_INV		(1 << 5)
+#define SDMA_GCR_GLM_WB			(1 << 4)
+#define SDMA_GCR_GL1_RANGE(x)		(((x) & 0x3) << 2)
+#define SDMA_GCR_GLI_INV(x)		(((x) & 0x3) << 0)
+
+/*define for op field*/
+#define SDMA_PKT_HEADER_op_offset 0
+#define SDMA_PKT_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_HEADER_op_shift  0
+#define SDMA_PKT_HEADER_OP(x) (((x) & SDMA_PKT_HEADER_op_mask) << SDMA_PKT_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_HEADER_sub_op_offset 0
+#define SDMA_PKT_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_HEADER_sub_op_shift  8
+#define SDMA_PKT_HEADER_SUB_OP(x) (((x) & SDMA_PKT_HEADER_sub_op_mask) << SDMA_PKT_HEADER_sub_op_shift)
+
+/*
+** Definitions for SDMA_PKT_COPY_LINEAR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_HEADER_op_shift  0
+#define SDMA_PKT_COPY_LINEAR_HEADER_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_op_mask) << SDMA_PKT_COPY_LINEAR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_LINEAR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_sub_op_mask) << SDMA_PKT_COPY_LINEAR_HEADER_sub_op_shift)
+
+/*define for encrypt field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_encrypt_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_encrypt_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_HEADER_encrypt_shift  16
+#define SDMA_PKT_COPY_LINEAR_HEADER_ENCRYPT(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_encrypt_mask) << SDMA_PKT_COPY_LINEAR_HEADER_encrypt_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_LINEAR_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_tmz_mask) << SDMA_PKT_COPY_LINEAR_HEADER_tmz_shift)
+
+/*define for backwards field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_backwards_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_backwards_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_HEADER_backwards_shift  25
+#define SDMA_PKT_COPY_LINEAR_HEADER_BACKWARDS(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_backwards_mask) << SDMA_PKT_COPY_LINEAR_HEADER_backwards_shift)
+
+/*define for broadcast field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_broadcast_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_broadcast_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_HEADER_broadcast_shift  27
+#define SDMA_PKT_COPY_LINEAR_HEADER_BROADCAST(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_broadcast_mask) << SDMA_PKT_COPY_LINEAR_HEADER_broadcast_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_LINEAR_COUNT_count_offset 1
+#define SDMA_PKT_COPY_LINEAR_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_COPY_LINEAR_COUNT_count_shift  0
+#define SDMA_PKT_COPY_LINEAR_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_LINEAR_COUNT_count_mask) << SDMA_PKT_COPY_LINEAR_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_shift  16
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_DST_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_shift  24
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_SRC_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_offset 3
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_offset 4
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_offset 5
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_offset 6
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_LINEAR_BC packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_LINEAR_BC_HEADER_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_BC_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_BC_HEADER_op_shift  0
+#define SDMA_PKT_COPY_LINEAR_BC_HEADER_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_BC_HEADER_op_mask) << SDMA_PKT_COPY_LINEAR_BC_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_LINEAR_BC_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_BC_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_BC_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_LINEAR_BC_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_BC_HEADER_sub_op_mask) << SDMA_PKT_COPY_LINEAR_BC_HEADER_sub_op_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_LINEAR_BC_COUNT_count_offset 1
+#define SDMA_PKT_COPY_LINEAR_BC_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_COPY_LINEAR_BC_COUNT_count_shift  0
+#define SDMA_PKT_COPY_LINEAR_BC_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_LINEAR_BC_COUNT_count_mask) << SDMA_PKT_COPY_LINEAR_BC_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_dst_sw_offset 2
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_dst_sw_shift  16
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_DST_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_BC_PARAMETER_dst_sw_mask) << SDMA_PKT_COPY_LINEAR_BC_PARAMETER_dst_sw_shift)
+
+/*define for dst_ha field*/
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_dst_ha_offset 2
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_dst_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_dst_ha_shift  22
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_DST_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_BC_PARAMETER_dst_ha_mask) << SDMA_PKT_COPY_LINEAR_BC_PARAMETER_dst_ha_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_src_sw_offset 2
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_src_sw_shift  24
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_SRC_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_BC_PARAMETER_src_sw_mask) << SDMA_PKT_COPY_LINEAR_BC_PARAMETER_src_sw_shift)
+
+/*define for src_ha field*/
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_src_ha_offset 2
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_src_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_src_ha_shift  30
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_SRC_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_BC_PARAMETER_src_ha_mask) << SDMA_PKT_COPY_LINEAR_BC_PARAMETER_src_ha_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_LO_src_addr_31_0_offset 3
+#define SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_HI_src_addr_63_32_offset 4
+#define SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_LO_dst_addr_31_0_offset 5
+#define SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_HI_dst_addr_63_32_offset 6
+#define SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_HI_dst_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_DIRTY_PAGE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_op_offset 0
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_op_shift  0
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_OP(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_HEADER_op_mask) << SDMA_PKT_COPY_DIRTY_PAGE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_HEADER_sub_op_mask) << SDMA_PKT_COPY_DIRTY_PAGE_HEADER_sub_op_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_HEADER_tmz_mask) << SDMA_PKT_COPY_DIRTY_PAGE_HEADER_tmz_shift)
+
+/*define for all field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_all_offset 0
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_all_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_all_shift  31
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_ALL(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_HEADER_all_mask) << SDMA_PKT_COPY_DIRTY_PAGE_HEADER_all_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_COUNT_count_offset 1
+#define SDMA_PKT_COPY_DIRTY_PAGE_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_COPY_DIRTY_PAGE_COUNT_count_shift  0
+#define SDMA_PKT_COPY_DIRTY_PAGE_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_COUNT_count_mask) << SDMA_PKT_COPY_DIRTY_PAGE_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst_mtype field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_mtype_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_mtype_mask   0x00000007
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_mtype_shift  3
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_DST_MTYPE(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_mtype_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_mtype_shift)
+
+/*define for dst_l2_policy field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_l2_policy_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_l2_policy_mask   0x00000003
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_l2_policy_shift  6
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_DST_L2_POLICY(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_l2_policy_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_l2_policy_shift)
+
+/*define for src_mtype field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_mtype_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_mtype_mask   0x00000007
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_mtype_shift  11
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_SRC_MTYPE(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_mtype_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_mtype_shift)
+
+/*define for src_l2_policy field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_l2_policy_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_l2_policy_mask   0x00000003
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_l2_policy_shift  14
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_SRC_L2_POLICY(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_l2_policy_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_l2_policy_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sw_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sw_shift  16
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_DST_SW(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sw_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sw_shift)
+
+/*define for dst_gcc field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gcc_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gcc_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gcc_shift  19
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_DST_GCC(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gcc_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gcc_shift)
+
+/*define for dst_sys field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sys_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sys_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sys_shift  20
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_DST_SYS(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sys_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sys_shift)
+
+/*define for dst_snoop field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_snoop_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_snoop_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_snoop_shift  22
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_DST_SNOOP(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_snoop_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_snoop_shift)
+
+/*define for dst_gpa field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gpa_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gpa_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gpa_shift  23
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_DST_GPA(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gpa_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gpa_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sw_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sw_shift  24
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_SRC_SW(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sw_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sw_shift)
+
+/*define for src_sys field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sys_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sys_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sys_shift  28
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_SRC_SYS(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sys_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sys_shift)
+
+/*define for src_snoop field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_snoop_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_snoop_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_snoop_shift  30
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_SRC_SNOOP(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_snoop_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_snoop_shift)
+
+/*define for src_gpa field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_gpa_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_gpa_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_gpa_shift  31
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_SRC_GPA(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_gpa_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_gpa_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_LO_src_addr_31_0_offset 3
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_HI_src_addr_63_32_offset 4
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_LO_dst_addr_31_0_offset 5
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_HI_dst_addr_63_32_offset 6
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_HI_dst_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_PHYSICAL_LINEAR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_op_offset 0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_op_shift  0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_OP(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_op_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_sub_op_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_sub_op_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_tmz_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_tmz_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_count_offset 1
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_count_shift  0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_count_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst_mtype field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_mtype_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_mtype_mask   0x00000007
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_mtype_shift  3
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_MTYPE(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_mtype_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_mtype_shift)
+
+/*define for dst_l2_policy field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_l2_policy_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_l2_policy_mask   0x00000003
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_l2_policy_shift  6
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_L2_POLICY(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_l2_policy_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_l2_policy_shift)
+
+/*define for src_mtype field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_mtype_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_mtype_mask   0x00000007
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_mtype_shift  11
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_SRC_MTYPE(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_mtype_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_mtype_shift)
+
+/*define for src_l2_policy field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_l2_policy_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_l2_policy_mask   0x00000003
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_l2_policy_shift  14
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_SRC_L2_POLICY(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_l2_policy_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_l2_policy_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sw_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sw_shift  16
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_SW(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sw_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sw_shift)
+
+/*define for dst_gcc field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gcc_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gcc_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gcc_shift  19
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_GCC(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gcc_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gcc_shift)
+
+/*define for dst_sys field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sys_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sys_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sys_shift  20
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_SYS(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sys_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sys_shift)
+
+/*define for dst_log field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_log_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_log_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_log_shift  21
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_LOG(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_log_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_log_shift)
+
+/*define for dst_snoop field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_snoop_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_snoop_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_snoop_shift  22
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_SNOOP(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_snoop_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_snoop_shift)
+
+/*define for dst_gpa field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gpa_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gpa_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gpa_shift  23
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_GPA(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gpa_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gpa_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sw_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sw_shift  24
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_SRC_SW(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sw_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sw_shift)
+
+/*define for src_gcc field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gcc_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gcc_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gcc_shift  27
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_SRC_GCC(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gcc_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gcc_shift)
+
+/*define for src_sys field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sys_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sys_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sys_shift  28
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_SRC_SYS(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sys_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sys_shift)
+
+/*define for src_snoop field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_snoop_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_snoop_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_snoop_shift  30
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_SRC_SNOOP(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_snoop_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_snoop_shift)
+
+/*define for src_gpa field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gpa_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gpa_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gpa_shift  31
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_SRC_GPA(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gpa_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gpa_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_LO_src_addr_31_0_offset 3
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_HI_src_addr_63_32_offset 4
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_LO_dst_addr_31_0_offset 5
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_HI_dst_addr_63_32_offset 6
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_BROADCAST_LINEAR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_OP(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_shift)
+
+/*define for encrypt field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_encrypt_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_encrypt_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_encrypt_shift  16
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_ENCRYPT(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_encrypt_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_encrypt_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_tmz_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_tmz_shift)
+
+/*define for broadcast field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_shift  27
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_BROADCAST(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_offset 1
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst2_sw field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_mask   0x00000003
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_shift  8
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST2_SW(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_shift)
+
+/*define for dst1_sw field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_mask   0x00000003
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_shift  16
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST1_SW(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_shift  24
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_SRC_SW(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_offset 3
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_offset 4
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST1_ADDR_LO word*/
+/*define for dst1_addr_31_0 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_offset 5
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_DST1_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_shift)
+
+/*define for DST1_ADDR_HI word*/
+/*define for dst1_addr_63_32 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_offset 6
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_DST1_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_shift)
+
+/*define for DST2_ADDR_LO word*/
+/*define for dst2_addr_31_0 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_offset 7
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_DST2_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_shift)
+
+/*define for DST2_ADDR_HI word*/
+/*define for dst2_addr_63_32 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_offset 8
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_DST2_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_LINEAR_SUBWIN packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_tmz_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_tmz_shift)
+
+/*define for elementsize field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_mask   0x00000007
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_shift  29
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_ELEMENTSIZE(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for src_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_offset 3
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_SRC_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_shift)
+
+/*define for src_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_offset 3
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_SRC_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_shift)
+
+/*define for DW_4 word*/
+/*define for src_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_offset 4
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_mask   0x00001FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_SRC_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_shift)
+
+/*define for src_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_offset 4
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_mask   0x0007FFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_shift  13
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_SRC_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_shift)
+
+/*define for DW_5 word*/
+/*define for src_slice_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_offset 5
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_SRC_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_offset 6
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_offset 7
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_8 word*/
+/*define for dst_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_offset 8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_DST_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_shift)
+
+/*define for dst_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_offset 8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_DST_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_shift)
+
+/*define for DW_9 word*/
+/*define for dst_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_offset 9
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_mask   0x00001FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_DST_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_shift)
+
+/*define for dst_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_offset 9
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_mask   0x0007FFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_shift  13
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_DST_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_shift)
+
+/*define for DW_10 word*/
+/*define for dst_slice_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_offset 10
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_DST_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_shift)
+
+/*define for DW_11 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_offset 11
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_RECT_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_offset 11
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_RECT_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_shift)
+
+/*define for DW_12 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_mask   0x00001FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_RECT_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_DST_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_shift  24
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_SRC_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_LINEAR_SUBWIN_BC packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_op_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_op_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_sub_op_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_sub_op_shift)
+
+/*define for elementsize field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_elementsize_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_elementsize_mask   0x00000007
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_elementsize_shift  29
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_ELEMENTSIZE(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_elementsize_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_elementsize_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for src_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_src_x_offset 3
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_src_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_src_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_SRC_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_src_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_src_x_shift)
+
+/*define for src_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_src_y_offset 3
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_src_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_src_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_SRC_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_src_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_src_y_shift)
+
+/*define for DW_4 word*/
+/*define for src_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_src_z_offset 4
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_src_z_mask   0x000007FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_src_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_SRC_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_src_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_src_z_shift)
+
+/*define for src_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_src_pitch_offset 4
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_src_pitch_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_src_pitch_shift  13
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_SRC_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_src_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_src_pitch_shift)
+
+/*define for DW_5 word*/
+/*define for src_slice_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_5_src_slice_pitch_offset 5
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_5_src_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_5_src_slice_pitch_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_5_SRC_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_5_src_slice_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_5_src_slice_pitch_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_LO_dst_addr_31_0_offset 6
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_HI_dst_addr_63_32_offset 7
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_8 word*/
+/*define for dst_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_dst_x_offset 8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_dst_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_dst_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_DST_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_dst_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_dst_x_shift)
+
+/*define for dst_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_dst_y_offset 8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_dst_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_dst_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_DST_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_dst_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_dst_y_shift)
+
+/*define for DW_9 word*/
+/*define for dst_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_dst_z_offset 9
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_dst_z_mask   0x000007FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_dst_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_DST_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_dst_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_dst_z_shift)
+
+/*define for dst_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_dst_pitch_offset 9
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_dst_pitch_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_dst_pitch_shift  13
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_DST_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_dst_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_dst_pitch_shift)
+
+/*define for DW_10 word*/
+/*define for dst_slice_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_10_dst_slice_pitch_offset 10
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_10_dst_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_10_dst_slice_pitch_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_10_DST_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_10_dst_slice_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_10_dst_slice_pitch_shift)
+
+/*define for DW_11 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_rect_x_offset 11
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_rect_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_RECT_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_rect_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_rect_y_offset 11
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_rect_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_RECT_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_rect_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_rect_y_shift)
+
+/*define for DW_12 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_rect_z_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_rect_z_mask   0x000007FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_rect_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_RECT_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_rect_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_rect_z_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_dst_sw_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_dst_sw_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_DST_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_dst_sw_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_dst_sw_shift)
+
+/*define for dst_ha field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_dst_ha_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_dst_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_dst_ha_shift  22
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_DST_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_dst_ha_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_dst_ha_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_src_sw_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_src_sw_shift  24
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_SRC_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_src_sw_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_src_sw_shift)
+
+/*define for src_ha field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_src_ha_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_src_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_src_ha_shift  30
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_SRC_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_src_ha_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_src_ha_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_TILED packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_TILED_HEADER_op_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_HEADER_op_shift  0
+#define SDMA_PKT_COPY_TILED_HEADER_OP(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_op_mask) << SDMA_PKT_COPY_TILED_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_TILED_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_TILED_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_sub_op_mask) << SDMA_PKT_COPY_TILED_HEADER_sub_op_shift)
+
+/*define for encrypt field*/
+#define SDMA_PKT_COPY_TILED_HEADER_encrypt_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_encrypt_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_HEADER_encrypt_shift  16
+#define SDMA_PKT_COPY_TILED_HEADER_ENCRYPT(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_encrypt_mask) << SDMA_PKT_COPY_TILED_HEADER_encrypt_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_TILED_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_TILED_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_tmz_mask) << SDMA_PKT_COPY_TILED_HEADER_tmz_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_TILED_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_TILED_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_detile_mask) << SDMA_PKT_COPY_TILED_HEADER_detile_shift)
+
+/*define for TILED_ADDR_LO word*/
+/*define for tiled_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_offset 1
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_TILED_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_mask) << SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_shift)
+
+/*define for TILED_ADDR_HI word*/
+/*define for tiled_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_offset 2
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_TILED_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_mask) << SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for width field*/
+#define SDMA_PKT_COPY_TILED_DW_3_width_offset 3
+#define SDMA_PKT_COPY_TILED_DW_3_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_3_width_shift  0
+#define SDMA_PKT_COPY_TILED_DW_3_WIDTH(x) (((x) & SDMA_PKT_COPY_TILED_DW_3_width_mask) << SDMA_PKT_COPY_TILED_DW_3_width_shift)
+
+/*define for DW_4 word*/
+/*define for height field*/
+#define SDMA_PKT_COPY_TILED_DW_4_height_offset 4
+#define SDMA_PKT_COPY_TILED_DW_4_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_4_height_shift  0
+#define SDMA_PKT_COPY_TILED_DW_4_HEIGHT(x) (((x) & SDMA_PKT_COPY_TILED_DW_4_height_mask) << SDMA_PKT_COPY_TILED_DW_4_height_shift)
+
+/*define for depth field*/
+#define SDMA_PKT_COPY_TILED_DW_4_depth_offset 4
+#define SDMA_PKT_COPY_TILED_DW_4_depth_mask   0x00001FFF
+#define SDMA_PKT_COPY_TILED_DW_4_depth_shift  16
+#define SDMA_PKT_COPY_TILED_DW_4_DEPTH(x) (((x) & SDMA_PKT_COPY_TILED_DW_4_depth_mask) << SDMA_PKT_COPY_TILED_DW_4_depth_shift)
+
+/*define for DW_5 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_TILED_DW_5_element_size_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_DW_5_element_size_shift  0
+#define SDMA_PKT_COPY_TILED_DW_5_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_element_size_mask) << SDMA_PKT_COPY_TILED_DW_5_element_size_shift)
+
+/*define for swizzle_mode field*/
+#define SDMA_PKT_COPY_TILED_DW_5_swizzle_mode_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_swizzle_mode_mask   0x0000001F
+#define SDMA_PKT_COPY_TILED_DW_5_swizzle_mode_shift  3
+#define SDMA_PKT_COPY_TILED_DW_5_SWIZZLE_MODE(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_swizzle_mode_mask) << SDMA_PKT_COPY_TILED_DW_5_swizzle_mode_shift)
+
+/*define for dimension field*/
+#define SDMA_PKT_COPY_TILED_DW_5_dimension_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_dimension_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_5_dimension_shift  9
+#define SDMA_PKT_COPY_TILED_DW_5_DIMENSION(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_dimension_mask) << SDMA_PKT_COPY_TILED_DW_5_dimension_shift)
+
+/*define for mip_max field*/
+#define SDMA_PKT_COPY_TILED_DW_5_mip_max_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_mip_max_mask   0x0000000F
+#define SDMA_PKT_COPY_TILED_DW_5_mip_max_shift  16
+#define SDMA_PKT_COPY_TILED_DW_5_MIP_MAX(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_mip_max_mask) << SDMA_PKT_COPY_TILED_DW_5_mip_max_shift)
+
+/*define for DW_6 word*/
+/*define for x field*/
+#define SDMA_PKT_COPY_TILED_DW_6_x_offset 6
+#define SDMA_PKT_COPY_TILED_DW_6_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_6_x_shift  0
+#define SDMA_PKT_COPY_TILED_DW_6_X(x) (((x) & SDMA_PKT_COPY_TILED_DW_6_x_mask) << SDMA_PKT_COPY_TILED_DW_6_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_COPY_TILED_DW_6_y_offset 6
+#define SDMA_PKT_COPY_TILED_DW_6_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_6_y_shift  16
+#define SDMA_PKT_COPY_TILED_DW_6_Y(x) (((x) & SDMA_PKT_COPY_TILED_DW_6_y_mask) << SDMA_PKT_COPY_TILED_DW_6_y_shift)
+
+/*define for DW_7 word*/
+/*define for z field*/
+#define SDMA_PKT_COPY_TILED_DW_7_z_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_z_mask   0x00001FFF
+#define SDMA_PKT_COPY_TILED_DW_7_z_shift  0
+#define SDMA_PKT_COPY_TILED_DW_7_Z(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_z_mask) << SDMA_PKT_COPY_TILED_DW_7_z_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_TILED_DW_7_linear_sw_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_7_linear_sw_shift  16
+#define SDMA_PKT_COPY_TILED_DW_7_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_linear_sw_mask) << SDMA_PKT_COPY_TILED_DW_7_linear_sw_shift)
+
+/*define for linear_cc field*/
+#define SDMA_PKT_COPY_TILED_DW_7_linear_cc_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_linear_cc_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_DW_7_linear_cc_shift  20
+#define SDMA_PKT_COPY_TILED_DW_7_LINEAR_CC(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_linear_cc_mask) << SDMA_PKT_COPY_TILED_DW_7_linear_cc_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_TILED_DW_7_tile_sw_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_7_tile_sw_shift  24
+#define SDMA_PKT_COPY_TILED_DW_7_TILE_SW(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_tile_sw_mask) << SDMA_PKT_COPY_TILED_DW_7_tile_sw_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_offset 8
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_offset 9
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for LINEAR_PITCH word*/
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_offset 10
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_mask   0x0007FFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_mask) << SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_shift)
+
+/*define for LINEAR_SLICE_PITCH word*/
+/*define for linear_slice_pitch field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_SLICE_PITCH_linear_slice_pitch_offset 11
+#define SDMA_PKT_COPY_TILED_LINEAR_SLICE_PITCH_linear_slice_pitch_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_SLICE_PITCH_linear_slice_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_SLICE_PITCH_LINEAR_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_SLICE_PITCH_linear_slice_pitch_mask) << SDMA_PKT_COPY_TILED_LINEAR_SLICE_PITCH_linear_slice_pitch_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_TILED_COUNT_count_offset 12
+#define SDMA_PKT_COPY_TILED_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_COPY_TILED_COUNT_count_shift  0
+#define SDMA_PKT_COPY_TILED_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_TILED_COUNT_count_mask) << SDMA_PKT_COPY_TILED_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_TILED_BC packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_TILED_BC_HEADER_op_offset 0
+#define SDMA_PKT_COPY_TILED_BC_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_BC_HEADER_op_shift  0
+#define SDMA_PKT_COPY_TILED_BC_HEADER_OP(x) (((x) & SDMA_PKT_COPY_TILED_BC_HEADER_op_mask) << SDMA_PKT_COPY_TILED_BC_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_TILED_BC_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_TILED_BC_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_BC_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_TILED_BC_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_TILED_BC_HEADER_sub_op_mask) << SDMA_PKT_COPY_TILED_BC_HEADER_sub_op_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_TILED_BC_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_TILED_BC_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_BC_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_TILED_BC_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_TILED_BC_HEADER_detile_mask) << SDMA_PKT_COPY_TILED_BC_HEADER_detile_shift)
+
+/*define for TILED_ADDR_LO word*/
+/*define for tiled_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_BC_TILED_ADDR_LO_tiled_addr_31_0_offset 1
+#define SDMA_PKT_COPY_TILED_BC_TILED_ADDR_LO_tiled_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_BC_TILED_ADDR_LO_tiled_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_BC_TILED_ADDR_LO_TILED_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_BC_TILED_ADDR_LO_tiled_addr_31_0_mask) << SDMA_PKT_COPY_TILED_BC_TILED_ADDR_LO_tiled_addr_31_0_shift)
+
+/*define for TILED_ADDR_HI word*/
+/*define for tiled_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_BC_TILED_ADDR_HI_tiled_addr_63_32_offset 2
+#define SDMA_PKT_COPY_TILED_BC_TILED_ADDR_HI_tiled_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_BC_TILED_ADDR_HI_tiled_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_BC_TILED_ADDR_HI_TILED_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_BC_TILED_ADDR_HI_tiled_addr_63_32_mask) << SDMA_PKT_COPY_TILED_BC_TILED_ADDR_HI_tiled_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for width field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_3_width_offset 3
+#define SDMA_PKT_COPY_TILED_BC_DW_3_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_BC_DW_3_width_shift  0
+#define SDMA_PKT_COPY_TILED_BC_DW_3_WIDTH(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_3_width_mask) << SDMA_PKT_COPY_TILED_BC_DW_3_width_shift)
+
+/*define for DW_4 word*/
+/*define for height field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_4_height_offset 4
+#define SDMA_PKT_COPY_TILED_BC_DW_4_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_BC_DW_4_height_shift  0
+#define SDMA_PKT_COPY_TILED_BC_DW_4_HEIGHT(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_4_height_mask) << SDMA_PKT_COPY_TILED_BC_DW_4_height_shift)
+
+/*define for depth field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_4_depth_offset 4
+#define SDMA_PKT_COPY_TILED_BC_DW_4_depth_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_BC_DW_4_depth_shift  16
+#define SDMA_PKT_COPY_TILED_BC_DW_4_DEPTH(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_4_depth_mask) << SDMA_PKT_COPY_TILED_BC_DW_4_depth_shift)
+
+/*define for DW_5 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_5_element_size_offset 5
+#define SDMA_PKT_COPY_TILED_BC_DW_5_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_BC_DW_5_element_size_shift  0
+#define SDMA_PKT_COPY_TILED_BC_DW_5_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_5_element_size_mask) << SDMA_PKT_COPY_TILED_BC_DW_5_element_size_shift)
+
+/*define for array_mode field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_5_array_mode_offset 5
+#define SDMA_PKT_COPY_TILED_BC_DW_5_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_TILED_BC_DW_5_array_mode_shift  3
+#define SDMA_PKT_COPY_TILED_BC_DW_5_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_5_array_mode_mask) << SDMA_PKT_COPY_TILED_BC_DW_5_array_mode_shift)
+
+/*define for mit_mode field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_5_mit_mode_offset 5
+#define SDMA_PKT_COPY_TILED_BC_DW_5_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_BC_DW_5_mit_mode_shift  8
+#define SDMA_PKT_COPY_TILED_BC_DW_5_MIT_MODE(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_5_mit_mode_mask) << SDMA_PKT_COPY_TILED_BC_DW_5_mit_mode_shift)
+
+/*define for tilesplit_size field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_5_tilesplit_size_offset 5
+#define SDMA_PKT_COPY_TILED_BC_DW_5_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_BC_DW_5_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_TILED_BC_DW_5_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_5_tilesplit_size_mask) << SDMA_PKT_COPY_TILED_BC_DW_5_tilesplit_size_shift)
+
+/*define for bank_w field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_5_bank_w_offset 5
+#define SDMA_PKT_COPY_TILED_BC_DW_5_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_BC_DW_5_bank_w_shift  15
+#define SDMA_PKT_COPY_TILED_BC_DW_5_BANK_W(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_5_bank_w_mask) << SDMA_PKT_COPY_TILED_BC_DW_5_bank_w_shift)
+
+/*define for bank_h field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_5_bank_h_offset 5
+#define SDMA_PKT_COPY_TILED_BC_DW_5_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_BC_DW_5_bank_h_shift  18
+#define SDMA_PKT_COPY_TILED_BC_DW_5_BANK_H(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_5_bank_h_mask) << SDMA_PKT_COPY_TILED_BC_DW_5_bank_h_shift)
+
+/*define for num_bank field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_5_num_bank_offset 5
+#define SDMA_PKT_COPY_TILED_BC_DW_5_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_BC_DW_5_num_bank_shift  21
+#define SDMA_PKT_COPY_TILED_BC_DW_5_NUM_BANK(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_5_num_bank_mask) << SDMA_PKT_COPY_TILED_BC_DW_5_num_bank_shift)
+
+/*define for mat_aspt field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_5_mat_aspt_offset 5
+#define SDMA_PKT_COPY_TILED_BC_DW_5_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_BC_DW_5_mat_aspt_shift  24
+#define SDMA_PKT_COPY_TILED_BC_DW_5_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_5_mat_aspt_mask) << SDMA_PKT_COPY_TILED_BC_DW_5_mat_aspt_shift)
+
+/*define for pipe_config field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_5_pipe_config_offset 5
+#define SDMA_PKT_COPY_TILED_BC_DW_5_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_TILED_BC_DW_5_pipe_config_shift  26
+#define SDMA_PKT_COPY_TILED_BC_DW_5_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_5_pipe_config_mask) << SDMA_PKT_COPY_TILED_BC_DW_5_pipe_config_shift)
+
+/*define for DW_6 word*/
+/*define for x field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_6_x_offset 6
+#define SDMA_PKT_COPY_TILED_BC_DW_6_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_BC_DW_6_x_shift  0
+#define SDMA_PKT_COPY_TILED_BC_DW_6_X(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_6_x_mask) << SDMA_PKT_COPY_TILED_BC_DW_6_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_6_y_offset 6
+#define SDMA_PKT_COPY_TILED_BC_DW_6_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_BC_DW_6_y_shift  16
+#define SDMA_PKT_COPY_TILED_BC_DW_6_Y(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_6_y_mask) << SDMA_PKT_COPY_TILED_BC_DW_6_y_shift)
+
+/*define for DW_7 word*/
+/*define for z field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_7_z_offset 7
+#define SDMA_PKT_COPY_TILED_BC_DW_7_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_BC_DW_7_z_shift  0
+#define SDMA_PKT_COPY_TILED_BC_DW_7_Z(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_7_z_mask) << SDMA_PKT_COPY_TILED_BC_DW_7_z_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_7_linear_sw_offset 7
+#define SDMA_PKT_COPY_TILED_BC_DW_7_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_BC_DW_7_linear_sw_shift  16
+#define SDMA_PKT_COPY_TILED_BC_DW_7_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_7_linear_sw_mask) << SDMA_PKT_COPY_TILED_BC_DW_7_linear_sw_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_7_tile_sw_offset 7
+#define SDMA_PKT_COPY_TILED_BC_DW_7_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_BC_DW_7_tile_sw_shift  24
+#define SDMA_PKT_COPY_TILED_BC_DW_7_TILE_SW(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_7_tile_sw_mask) << SDMA_PKT_COPY_TILED_BC_DW_7_tile_sw_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_LO_linear_addr_31_0_offset 8
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_HI_linear_addr_63_32_offset 9
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for LINEAR_PITCH word*/
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_PITCH_linear_pitch_offset 10
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_PITCH_linear_pitch_mask   0x0007FFFF
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_PITCH_linear_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_PITCH_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_BC_LINEAR_PITCH_linear_pitch_mask) << SDMA_PKT_COPY_TILED_BC_LINEAR_PITCH_linear_pitch_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_TILED_BC_COUNT_count_offset 11
+#define SDMA_PKT_COPY_TILED_BC_COUNT_count_mask   0x000FFFFF
+#define SDMA_PKT_COPY_TILED_BC_COUNT_count_shift  2
+#define SDMA_PKT_COPY_TILED_BC_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_TILED_BC_COUNT_count_mask) << SDMA_PKT_COPY_TILED_BC_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_L2T_BROADCAST packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_OP(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_shift)
+
+/*define for encrypt field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_encrypt_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_encrypt_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_encrypt_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_ENCRYPT(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_encrypt_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_encrypt_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_tmz_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_tmz_shift)
+
+/*define for videocopy field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_shift  26
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_VIDEOCOPY(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_shift)
+
+/*define for broadcast field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_shift  27
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_BROADCAST(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_shift)
+
+/*define for TILED_ADDR_LO_0 word*/
+/*define for tiled_addr0_31_0 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_offset 1
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_TILED_ADDR0_31_0(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_shift)
+
+/*define for TILED_ADDR_HI_0 word*/
+/*define for tiled_addr0_63_32 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_offset 2
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_TILED_ADDR0_63_32(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_shift)
+
+/*define for TILED_ADDR_LO_1 word*/
+/*define for tiled_addr1_31_0 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_offset 3
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_TILED_ADDR1_31_0(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_shift)
+
+/*define for TILED_ADDR_HI_1 word*/
+/*define for tiled_addr1_63_32 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_offset 4
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_TILED_ADDR1_63_32(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_shift)
+
+/*define for DW_5 word*/
+/*define for width field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_width_offset 5
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_width_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_WIDTH(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_5_width_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_5_width_shift)
+
+/*define for DW_6 word*/
+/*define for height field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_height_offset 6
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_height_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_HEIGHT(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_6_height_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_6_height_shift)
+
+/*define for depth field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_depth_offset 6
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_depth_mask   0x00001FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_depth_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_DEPTH(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_6_depth_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_6_depth_shift)
+
+/*define for DW_7 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_shift)
+
+/*define for swizzle_mode field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_swizzle_mode_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_swizzle_mode_mask   0x0000001F
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_swizzle_mode_shift  3
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_SWIZZLE_MODE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_swizzle_mode_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_swizzle_mode_shift)
+
+/*define for dimension field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_dimension_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_dimension_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_dimension_shift  9
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_DIMENSION(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_dimension_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_dimension_shift)
+
+/*define for mip_max field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mip_max_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mip_max_mask   0x0000000F
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mip_max_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_MIP_MAX(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mip_max_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mip_max_shift)
+
+/*define for DW_8 word*/
+/*define for x field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_offset 8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_X(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_offset 8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_Y(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_shift)
+
+/*define for DW_9 word*/
+/*define for z field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_offset 9
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_mask   0x00001FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_Z(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_shift)
+
+/*define for DW_10 word*/
+/*define for dst2_sw field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_shift  8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_DST2_SW(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_shift  24
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_TILE_SW(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_offset 11
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_offset 12
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for LINEAR_PITCH word*/
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_offset 13
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_mask   0x0007FFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_shift)
+
+/*define for LINEAR_SLICE_PITCH word*/
+/*define for linear_slice_pitch field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_SLICE_PITCH_linear_slice_pitch_offset 14
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_SLICE_PITCH_linear_slice_pitch_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_SLICE_PITCH_linear_slice_pitch_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_SLICE_PITCH_LINEAR_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_SLICE_PITCH_linear_slice_pitch_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_SLICE_PITCH_linear_slice_pitch_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_offset 15
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_mask) << SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_T2T packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_T2T_HEADER_op_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_T2T_HEADER_op_shift  0
+#define SDMA_PKT_COPY_T2T_HEADER_OP(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_op_mask) << SDMA_PKT_COPY_T2T_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_T2T_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_T2T_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_T2T_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_sub_op_mask) << SDMA_PKT_COPY_T2T_HEADER_sub_op_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_T2T_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_T2T_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_T2T_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_tmz_mask) << SDMA_PKT_COPY_T2T_HEADER_tmz_shift)
+
+/*define for dcc field*/
+#define SDMA_PKT_COPY_T2T_HEADER_dcc_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_dcc_mask   0x00000001
+#define SDMA_PKT_COPY_T2T_HEADER_dcc_shift  19
+#define SDMA_PKT_COPY_T2T_HEADER_DCC(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_dcc_mask) << SDMA_PKT_COPY_T2T_HEADER_dcc_shift)
+
+/*define for dcc_dir field*/
+#define SDMA_PKT_COPY_T2T_HEADER_dcc_dir_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_dcc_dir_mask   0x00000001
+#define SDMA_PKT_COPY_T2T_HEADER_dcc_dir_shift  31
+#define SDMA_PKT_COPY_T2T_HEADER_DCC_DIR(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_dcc_dir_mask) << SDMA_PKT_COPY_T2T_HEADER_dcc_dir_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for src_x field*/
+#define SDMA_PKT_COPY_T2T_DW_3_src_x_offset 3
+#define SDMA_PKT_COPY_T2T_DW_3_src_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_3_src_x_shift  0
+#define SDMA_PKT_COPY_T2T_DW_3_SRC_X(x) (((x) & SDMA_PKT_COPY_T2T_DW_3_src_x_mask) << SDMA_PKT_COPY_T2T_DW_3_src_x_shift)
+
+/*define for src_y field*/
+#define SDMA_PKT_COPY_T2T_DW_3_src_y_offset 3
+#define SDMA_PKT_COPY_T2T_DW_3_src_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_3_src_y_shift  16
+#define SDMA_PKT_COPY_T2T_DW_3_SRC_Y(x) (((x) & SDMA_PKT_COPY_T2T_DW_3_src_y_mask) << SDMA_PKT_COPY_T2T_DW_3_src_y_shift)
+
+/*define for DW_4 word*/
+/*define for src_z field*/
+#define SDMA_PKT_COPY_T2T_DW_4_src_z_offset 4
+#define SDMA_PKT_COPY_T2T_DW_4_src_z_mask   0x00001FFF
+#define SDMA_PKT_COPY_T2T_DW_4_src_z_shift  0
+#define SDMA_PKT_COPY_T2T_DW_4_SRC_Z(x) (((x) & SDMA_PKT_COPY_T2T_DW_4_src_z_mask) << SDMA_PKT_COPY_T2T_DW_4_src_z_shift)
+
+/*define for src_width field*/
+#define SDMA_PKT_COPY_T2T_DW_4_src_width_offset 4
+#define SDMA_PKT_COPY_T2T_DW_4_src_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_4_src_width_shift  16
+#define SDMA_PKT_COPY_T2T_DW_4_SRC_WIDTH(x) (((x) & SDMA_PKT_COPY_T2T_DW_4_src_width_mask) << SDMA_PKT_COPY_T2T_DW_4_src_width_shift)
+
+/*define for DW_5 word*/
+/*define for src_height field*/
+#define SDMA_PKT_COPY_T2T_DW_5_src_height_offset 5
+#define SDMA_PKT_COPY_T2T_DW_5_src_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_5_src_height_shift  0
+#define SDMA_PKT_COPY_T2T_DW_5_SRC_HEIGHT(x) (((x) & SDMA_PKT_COPY_T2T_DW_5_src_height_mask) << SDMA_PKT_COPY_T2T_DW_5_src_height_shift)
+
+/*define for src_depth field*/
+#define SDMA_PKT_COPY_T2T_DW_5_src_depth_offset 5
+#define SDMA_PKT_COPY_T2T_DW_5_src_depth_mask   0x00001FFF
+#define SDMA_PKT_COPY_T2T_DW_5_src_depth_shift  16
+#define SDMA_PKT_COPY_T2T_DW_5_SRC_DEPTH(x) (((x) & SDMA_PKT_COPY_T2T_DW_5_src_depth_mask) << SDMA_PKT_COPY_T2T_DW_5_src_depth_shift)
+
+/*define for DW_6 word*/
+/*define for src_element_size field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_element_size_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_6_src_element_size_shift  0
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_element_size_mask) << SDMA_PKT_COPY_T2T_DW_6_src_element_size_shift)
+
+/*define for src_swizzle_mode field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_swizzle_mode_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_swizzle_mode_mask   0x0000001F
+#define SDMA_PKT_COPY_T2T_DW_6_src_swizzle_mode_shift  3
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_SWIZZLE_MODE(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_swizzle_mode_mask) << SDMA_PKT_COPY_T2T_DW_6_src_swizzle_mode_shift)
+
+/*define for src_dimension field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_dimension_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_dimension_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_6_src_dimension_shift  9
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_DIMENSION(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_dimension_mask) << SDMA_PKT_COPY_T2T_DW_6_src_dimension_shift)
+
+/*define for src_mip_max field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_mip_max_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_mip_max_mask   0x0000000F
+#define SDMA_PKT_COPY_T2T_DW_6_src_mip_max_shift  16
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_MIP_MAX(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_mip_max_mask) << SDMA_PKT_COPY_T2T_DW_6_src_mip_max_shift)
+
+/*define for src_mip_id field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_mip_id_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_mip_id_mask   0x0000000F
+#define SDMA_PKT_COPY_T2T_DW_6_src_mip_id_shift  20
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_MIP_ID(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_mip_id_mask) << SDMA_PKT_COPY_T2T_DW_6_src_mip_id_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_offset 7
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_offset 8
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_9 word*/
+/*define for dst_x field*/
+#define SDMA_PKT_COPY_T2T_DW_9_dst_x_offset 9
+#define SDMA_PKT_COPY_T2T_DW_9_dst_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_9_dst_x_shift  0
+#define SDMA_PKT_COPY_T2T_DW_9_DST_X(x) (((x) & SDMA_PKT_COPY_T2T_DW_9_dst_x_mask) << SDMA_PKT_COPY_T2T_DW_9_dst_x_shift)
+
+/*define for dst_y field*/
+#define SDMA_PKT_COPY_T2T_DW_9_dst_y_offset 9
+#define SDMA_PKT_COPY_T2T_DW_9_dst_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_9_dst_y_shift  16
+#define SDMA_PKT_COPY_T2T_DW_9_DST_Y(x) (((x) & SDMA_PKT_COPY_T2T_DW_9_dst_y_mask) << SDMA_PKT_COPY_T2T_DW_9_dst_y_shift)
+
+/*define for DW_10 word*/
+/*define for dst_z field*/
+#define SDMA_PKT_COPY_T2T_DW_10_dst_z_offset 10
+#define SDMA_PKT_COPY_T2T_DW_10_dst_z_mask   0x00001FFF
+#define SDMA_PKT_COPY_T2T_DW_10_dst_z_shift  0
+#define SDMA_PKT_COPY_T2T_DW_10_DST_Z(x) (((x) & SDMA_PKT_COPY_T2T_DW_10_dst_z_mask) << SDMA_PKT_COPY_T2T_DW_10_dst_z_shift)
+
+/*define for dst_width field*/
+#define SDMA_PKT_COPY_T2T_DW_10_dst_width_offset 10
+#define SDMA_PKT_COPY_T2T_DW_10_dst_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_10_dst_width_shift  16
+#define SDMA_PKT_COPY_T2T_DW_10_DST_WIDTH(x) (((x) & SDMA_PKT_COPY_T2T_DW_10_dst_width_mask) << SDMA_PKT_COPY_T2T_DW_10_dst_width_shift)
+
+/*define for DW_11 word*/
+/*define for dst_height field*/
+#define SDMA_PKT_COPY_T2T_DW_11_dst_height_offset 11
+#define SDMA_PKT_COPY_T2T_DW_11_dst_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_11_dst_height_shift  0
+#define SDMA_PKT_COPY_T2T_DW_11_DST_HEIGHT(x) (((x) & SDMA_PKT_COPY_T2T_DW_11_dst_height_mask) << SDMA_PKT_COPY_T2T_DW_11_dst_height_shift)
+
+/*define for dst_depth field*/
+#define SDMA_PKT_COPY_T2T_DW_11_dst_depth_offset 11
+#define SDMA_PKT_COPY_T2T_DW_11_dst_depth_mask   0x00001FFF
+#define SDMA_PKT_COPY_T2T_DW_11_dst_depth_shift  16
+#define SDMA_PKT_COPY_T2T_DW_11_DST_DEPTH(x) (((x) & SDMA_PKT_COPY_T2T_DW_11_dst_depth_mask) << SDMA_PKT_COPY_T2T_DW_11_dst_depth_shift)
+
+/*define for DW_12 word*/
+/*define for dst_element_size field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_element_size_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_12_dst_element_size_shift  0
+#define SDMA_PKT_COPY_T2T_DW_12_DST_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_element_size_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_element_size_shift)
+
+/*define for dst_swizzle_mode field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_swizzle_mode_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_swizzle_mode_mask   0x0000001F
+#define SDMA_PKT_COPY_T2T_DW_12_dst_swizzle_mode_shift  3
+#define SDMA_PKT_COPY_T2T_DW_12_DST_SWIZZLE_MODE(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_swizzle_mode_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_swizzle_mode_shift)
+
+/*define for dst_dimension field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_dimension_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_dimension_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_12_dst_dimension_shift  9
+#define SDMA_PKT_COPY_T2T_DW_12_DST_DIMENSION(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_dimension_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_dimension_shift)
+
+/*define for dst_mip_max field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mip_max_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mip_max_mask   0x0000000F
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mip_max_shift  16
+#define SDMA_PKT_COPY_T2T_DW_12_DST_MIP_MAX(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_mip_max_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_mip_max_shift)
+
+/*define for dst_mip_id field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mip_id_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mip_id_mask   0x0000000F
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mip_id_shift  20
+#define SDMA_PKT_COPY_T2T_DW_12_DST_MIP_ID(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_mip_id_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_mip_id_shift)
+
+/*define for DW_13 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_T2T_DW_13_rect_x_offset 13
+#define SDMA_PKT_COPY_T2T_DW_13_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_13_rect_x_shift  0
+#define SDMA_PKT_COPY_T2T_DW_13_RECT_X(x) (((x) & SDMA_PKT_COPY_T2T_DW_13_rect_x_mask) << SDMA_PKT_COPY_T2T_DW_13_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_T2T_DW_13_rect_y_offset 13
+#define SDMA_PKT_COPY_T2T_DW_13_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_13_rect_y_shift  16
+#define SDMA_PKT_COPY_T2T_DW_13_RECT_Y(x) (((x) & SDMA_PKT_COPY_T2T_DW_13_rect_y_mask) << SDMA_PKT_COPY_T2T_DW_13_rect_y_shift)
+
+/*define for DW_14 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_T2T_DW_14_rect_z_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_rect_z_mask   0x00001FFF
+#define SDMA_PKT_COPY_T2T_DW_14_rect_z_shift  0
+#define SDMA_PKT_COPY_T2T_DW_14_RECT_Z(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_rect_z_mask) << SDMA_PKT_COPY_T2T_DW_14_rect_z_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_T2T_DW_14_dst_sw_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_14_dst_sw_shift  16
+#define SDMA_PKT_COPY_T2T_DW_14_DST_SW(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_dst_sw_mask) << SDMA_PKT_COPY_T2T_DW_14_dst_sw_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_T2T_DW_14_src_sw_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_14_src_sw_shift  24
+#define SDMA_PKT_COPY_T2T_DW_14_SRC_SW(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_src_sw_mask) << SDMA_PKT_COPY_T2T_DW_14_src_sw_shift)
+
+/*define for META_ADDR_LO word*/
+/*define for meta_addr_31_0 field*/
+#define SDMA_PKT_COPY_T2T_META_ADDR_LO_meta_addr_31_0_offset 15
+#define SDMA_PKT_COPY_T2T_META_ADDR_LO_meta_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_META_ADDR_LO_meta_addr_31_0_shift  0
+#define SDMA_PKT_COPY_T2T_META_ADDR_LO_META_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_T2T_META_ADDR_LO_meta_addr_31_0_mask) << SDMA_PKT_COPY_T2T_META_ADDR_LO_meta_addr_31_0_shift)
+
+/*define for META_ADDR_HI word*/
+/*define for meta_addr_63_32 field*/
+#define SDMA_PKT_COPY_T2T_META_ADDR_HI_meta_addr_63_32_offset 16
+#define SDMA_PKT_COPY_T2T_META_ADDR_HI_meta_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_META_ADDR_HI_meta_addr_63_32_shift  0
+#define SDMA_PKT_COPY_T2T_META_ADDR_HI_META_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_T2T_META_ADDR_HI_meta_addr_63_32_mask) << SDMA_PKT_COPY_T2T_META_ADDR_HI_meta_addr_63_32_shift)
+
+/*define for META_CONFIG word*/
+/*define for data_format field*/
+#define SDMA_PKT_COPY_T2T_META_CONFIG_data_format_offset 17
+#define SDMA_PKT_COPY_T2T_META_CONFIG_data_format_mask   0x0000007F
+#define SDMA_PKT_COPY_T2T_META_CONFIG_data_format_shift  0
+#define SDMA_PKT_COPY_T2T_META_CONFIG_DATA_FORMAT(x) (((x) & SDMA_PKT_COPY_T2T_META_CONFIG_data_format_mask) << SDMA_PKT_COPY_T2T_META_CONFIG_data_format_shift)
+
+/*define for color_transform_disable field*/
+#define SDMA_PKT_COPY_T2T_META_CONFIG_color_transform_disable_offset 17
+#define SDMA_PKT_COPY_T2T_META_CONFIG_color_transform_disable_mask   0x00000001
+#define SDMA_PKT_COPY_T2T_META_CONFIG_color_transform_disable_shift  7
+#define SDMA_PKT_COPY_T2T_META_CONFIG_COLOR_TRANSFORM_DISABLE(x) (((x) & SDMA_PKT_COPY_T2T_META_CONFIG_color_transform_disable_mask) << SDMA_PKT_COPY_T2T_META_CONFIG_color_transform_disable_shift)
+
+/*define for alpha_is_on_msb field*/
+#define SDMA_PKT_COPY_T2T_META_CONFIG_alpha_is_on_msb_offset 17
+#define SDMA_PKT_COPY_T2T_META_CONFIG_alpha_is_on_msb_mask   0x00000001
+#define SDMA_PKT_COPY_T2T_META_CONFIG_alpha_is_on_msb_shift  8
+#define SDMA_PKT_COPY_T2T_META_CONFIG_ALPHA_IS_ON_MSB(x) (((x) & SDMA_PKT_COPY_T2T_META_CONFIG_alpha_is_on_msb_mask) << SDMA_PKT_COPY_T2T_META_CONFIG_alpha_is_on_msb_shift)
+
+/*define for number_type field*/
+#define SDMA_PKT_COPY_T2T_META_CONFIG_number_type_offset 17
+#define SDMA_PKT_COPY_T2T_META_CONFIG_number_type_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_META_CONFIG_number_type_shift  9
+#define SDMA_PKT_COPY_T2T_META_CONFIG_NUMBER_TYPE(x) (((x) & SDMA_PKT_COPY_T2T_META_CONFIG_number_type_mask) << SDMA_PKT_COPY_T2T_META_CONFIG_number_type_shift)
+
+/*define for surface_type field*/
+#define SDMA_PKT_COPY_T2T_META_CONFIG_surface_type_offset 17
+#define SDMA_PKT_COPY_T2T_META_CONFIG_surface_type_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_META_CONFIG_surface_type_shift  12
+#define SDMA_PKT_COPY_T2T_META_CONFIG_SURFACE_TYPE(x) (((x) & SDMA_PKT_COPY_T2T_META_CONFIG_surface_type_mask) << SDMA_PKT_COPY_T2T_META_CONFIG_surface_type_shift)
+
+/*define for max_comp_block_size field*/
+#define SDMA_PKT_COPY_T2T_META_CONFIG_max_comp_block_size_offset 17
+#define SDMA_PKT_COPY_T2T_META_CONFIG_max_comp_block_size_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_META_CONFIG_max_comp_block_size_shift  24
+#define SDMA_PKT_COPY_T2T_META_CONFIG_MAX_COMP_BLOCK_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_META_CONFIG_max_comp_block_size_mask) << SDMA_PKT_COPY_T2T_META_CONFIG_max_comp_block_size_shift)
+
+/*define for max_uncomp_block_size field*/
+#define SDMA_PKT_COPY_T2T_META_CONFIG_max_uncomp_block_size_offset 17
+#define SDMA_PKT_COPY_T2T_META_CONFIG_max_uncomp_block_size_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_META_CONFIG_max_uncomp_block_size_shift  26
+#define SDMA_PKT_COPY_T2T_META_CONFIG_MAX_UNCOMP_BLOCK_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_META_CONFIG_max_uncomp_block_size_mask) << SDMA_PKT_COPY_T2T_META_CONFIG_max_uncomp_block_size_shift)
+
+/*define for write_compress_enable field*/
+#define SDMA_PKT_COPY_T2T_META_CONFIG_write_compress_enable_offset 17
+#define SDMA_PKT_COPY_T2T_META_CONFIG_write_compress_enable_mask   0x00000001
+#define SDMA_PKT_COPY_T2T_META_CONFIG_write_compress_enable_shift  28
+#define SDMA_PKT_COPY_T2T_META_CONFIG_WRITE_COMPRESS_ENABLE(x) (((x) & SDMA_PKT_COPY_T2T_META_CONFIG_write_compress_enable_mask) << SDMA_PKT_COPY_T2T_META_CONFIG_write_compress_enable_shift)
+
+/*define for meta_tmz field*/
+#define SDMA_PKT_COPY_T2T_META_CONFIG_meta_tmz_offset 17
+#define SDMA_PKT_COPY_T2T_META_CONFIG_meta_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_T2T_META_CONFIG_meta_tmz_shift  29
+#define SDMA_PKT_COPY_T2T_META_CONFIG_META_TMZ(x) (((x) & SDMA_PKT_COPY_T2T_META_CONFIG_meta_tmz_mask) << SDMA_PKT_COPY_T2T_META_CONFIG_meta_tmz_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_T2T_BC packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_T2T_BC_HEADER_op_offset 0
+#define SDMA_PKT_COPY_T2T_BC_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_T2T_BC_HEADER_op_shift  0
+#define SDMA_PKT_COPY_T2T_BC_HEADER_OP(x) (((x) & SDMA_PKT_COPY_T2T_BC_HEADER_op_mask) << SDMA_PKT_COPY_T2T_BC_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_T2T_BC_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_T2T_BC_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_T2T_BC_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_T2T_BC_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_T2T_BC_HEADER_sub_op_mask) << SDMA_PKT_COPY_T2T_BC_HEADER_sub_op_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_T2T_BC_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_COPY_T2T_BC_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_BC_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_T2T_BC_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_T2T_BC_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_T2T_BC_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_T2T_BC_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_COPY_T2T_BC_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_BC_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_T2T_BC_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_T2T_BC_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_T2T_BC_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for src_x field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_3_src_x_offset 3
+#define SDMA_PKT_COPY_T2T_BC_DW_3_src_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_BC_DW_3_src_x_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DW_3_SRC_X(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_3_src_x_mask) << SDMA_PKT_COPY_T2T_BC_DW_3_src_x_shift)
+
+/*define for src_y field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_3_src_y_offset 3
+#define SDMA_PKT_COPY_T2T_BC_DW_3_src_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_BC_DW_3_src_y_shift  16
+#define SDMA_PKT_COPY_T2T_BC_DW_3_SRC_Y(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_3_src_y_mask) << SDMA_PKT_COPY_T2T_BC_DW_3_src_y_shift)
+
+/*define for DW_4 word*/
+/*define for src_z field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_4_src_z_offset 4
+#define SDMA_PKT_COPY_T2T_BC_DW_4_src_z_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_BC_DW_4_src_z_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DW_4_SRC_Z(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_4_src_z_mask) << SDMA_PKT_COPY_T2T_BC_DW_4_src_z_shift)
+
+/*define for src_width field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_4_src_width_offset 4
+#define SDMA_PKT_COPY_T2T_BC_DW_4_src_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_BC_DW_4_src_width_shift  16
+#define SDMA_PKT_COPY_T2T_BC_DW_4_SRC_WIDTH(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_4_src_width_mask) << SDMA_PKT_COPY_T2T_BC_DW_4_src_width_shift)
+
+/*define for DW_5 word*/
+/*define for src_height field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_5_src_height_offset 5
+#define SDMA_PKT_COPY_T2T_BC_DW_5_src_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_BC_DW_5_src_height_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DW_5_SRC_HEIGHT(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_5_src_height_mask) << SDMA_PKT_COPY_T2T_BC_DW_5_src_height_shift)
+
+/*define for src_depth field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_5_src_depth_offset 5
+#define SDMA_PKT_COPY_T2T_BC_DW_5_src_depth_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_BC_DW_5_src_depth_shift  16
+#define SDMA_PKT_COPY_T2T_BC_DW_5_SRC_DEPTH(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_5_src_depth_mask) << SDMA_PKT_COPY_T2T_BC_DW_5_src_depth_shift)
+
+/*define for DW_6 word*/
+/*define for src_element_size field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_element_size_offset 6
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_element_size_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DW_6_SRC_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_6_src_element_size_mask) << SDMA_PKT_COPY_T2T_BC_DW_6_src_element_size_shift)
+
+/*define for src_array_mode field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_array_mode_offset 6
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_array_mode_shift  3
+#define SDMA_PKT_COPY_T2T_BC_DW_6_SRC_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_6_src_array_mode_mask) << SDMA_PKT_COPY_T2T_BC_DW_6_src_array_mode_shift)
+
+/*define for src_mit_mode field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_mit_mode_offset 6
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_mit_mode_shift  8
+#define SDMA_PKT_COPY_T2T_BC_DW_6_SRC_MIT_MODE(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_6_src_mit_mode_mask) << SDMA_PKT_COPY_T2T_BC_DW_6_src_mit_mode_shift)
+
+/*define for src_tilesplit_size field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_tilesplit_size_offset 6
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_T2T_BC_DW_6_SRC_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_6_src_tilesplit_size_mask) << SDMA_PKT_COPY_T2T_BC_DW_6_src_tilesplit_size_shift)
+
+/*define for src_bank_w field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_bank_w_offset 6
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_bank_w_shift  15
+#define SDMA_PKT_COPY_T2T_BC_DW_6_SRC_BANK_W(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_6_src_bank_w_mask) << SDMA_PKT_COPY_T2T_BC_DW_6_src_bank_w_shift)
+
+/*define for src_bank_h field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_bank_h_offset 6
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_bank_h_shift  18
+#define SDMA_PKT_COPY_T2T_BC_DW_6_SRC_BANK_H(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_6_src_bank_h_mask) << SDMA_PKT_COPY_T2T_BC_DW_6_src_bank_h_shift)
+
+/*define for src_num_bank field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_num_bank_offset 6
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_num_bank_shift  21
+#define SDMA_PKT_COPY_T2T_BC_DW_6_SRC_NUM_BANK(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_6_src_num_bank_mask) << SDMA_PKT_COPY_T2T_BC_DW_6_src_num_bank_shift)
+
+/*define for src_mat_aspt field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_mat_aspt_offset 6
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_mat_aspt_shift  24
+#define SDMA_PKT_COPY_T2T_BC_DW_6_SRC_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_6_src_mat_aspt_mask) << SDMA_PKT_COPY_T2T_BC_DW_6_src_mat_aspt_shift)
+
+/*define for src_pipe_config field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_pipe_config_offset 6
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_pipe_config_shift  26
+#define SDMA_PKT_COPY_T2T_BC_DW_6_SRC_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_6_src_pipe_config_mask) << SDMA_PKT_COPY_T2T_BC_DW_6_src_pipe_config_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_T2T_BC_DST_ADDR_LO_dst_addr_31_0_offset 7
+#define SDMA_PKT_COPY_T2T_BC_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_BC_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_T2T_BC_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_T2T_BC_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_T2T_BC_DST_ADDR_HI_dst_addr_63_32_offset 8
+#define SDMA_PKT_COPY_T2T_BC_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_BC_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_T2T_BC_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_T2T_BC_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_9 word*/
+/*define for dst_x field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_9_dst_x_offset 9
+#define SDMA_PKT_COPY_T2T_BC_DW_9_dst_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_BC_DW_9_dst_x_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DW_9_DST_X(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_9_dst_x_mask) << SDMA_PKT_COPY_T2T_BC_DW_9_dst_x_shift)
+
+/*define for dst_y field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_9_dst_y_offset 9
+#define SDMA_PKT_COPY_T2T_BC_DW_9_dst_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_BC_DW_9_dst_y_shift  16
+#define SDMA_PKT_COPY_T2T_BC_DW_9_DST_Y(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_9_dst_y_mask) << SDMA_PKT_COPY_T2T_BC_DW_9_dst_y_shift)
+
+/*define for DW_10 word*/
+/*define for dst_z field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_10_dst_z_offset 10
+#define SDMA_PKT_COPY_T2T_BC_DW_10_dst_z_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_BC_DW_10_dst_z_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DW_10_DST_Z(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_10_dst_z_mask) << SDMA_PKT_COPY_T2T_BC_DW_10_dst_z_shift)
+
+/*define for dst_width field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_10_dst_width_offset 10
+#define SDMA_PKT_COPY_T2T_BC_DW_10_dst_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_BC_DW_10_dst_width_shift  16
+#define SDMA_PKT_COPY_T2T_BC_DW_10_DST_WIDTH(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_10_dst_width_mask) << SDMA_PKT_COPY_T2T_BC_DW_10_dst_width_shift)
+
+/*define for DW_11 word*/
+/*define for dst_height field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_11_dst_height_offset 11
+#define SDMA_PKT_COPY_T2T_BC_DW_11_dst_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_BC_DW_11_dst_height_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DW_11_DST_HEIGHT(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_11_dst_height_mask) << SDMA_PKT_COPY_T2T_BC_DW_11_dst_height_shift)
+
+/*define for dst_depth field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_11_dst_depth_offset 11
+#define SDMA_PKT_COPY_T2T_BC_DW_11_dst_depth_mask   0x00000FFF
+#define SDMA_PKT_COPY_T2T_BC_DW_11_dst_depth_shift  16
+#define SDMA_PKT_COPY_T2T_BC_DW_11_DST_DEPTH(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_11_dst_depth_mask) << SDMA_PKT_COPY_T2T_BC_DW_11_dst_depth_shift)
+
+/*define for DW_12 word*/
+/*define for dst_element_size field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_element_size_offset 12
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_element_size_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DW_12_DST_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_12_dst_element_size_mask) << SDMA_PKT_COPY_T2T_BC_DW_12_dst_element_size_shift)
+
+/*define for dst_array_mode field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_array_mode_offset 12
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_array_mode_shift  3
+#define SDMA_PKT_COPY_T2T_BC_DW_12_DST_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_12_dst_array_mode_mask) << SDMA_PKT_COPY_T2T_BC_DW_12_dst_array_mode_shift)
+
+/*define for dst_mit_mode field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_mit_mode_offset 12
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_mit_mode_shift  8
+#define SDMA_PKT_COPY_T2T_BC_DW_12_DST_MIT_MODE(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_12_dst_mit_mode_mask) << SDMA_PKT_COPY_T2T_BC_DW_12_dst_mit_mode_shift)
+
+/*define for dst_tilesplit_size field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_tilesplit_size_offset 12
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_T2T_BC_DW_12_DST_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_12_dst_tilesplit_size_mask) << SDMA_PKT_COPY_T2T_BC_DW_12_dst_tilesplit_size_shift)
+
+/*define for dst_bank_w field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_bank_w_offset 12
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_bank_w_shift  15
+#define SDMA_PKT_COPY_T2T_BC_DW_12_DST_BANK_W(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_12_dst_bank_w_mask) << SDMA_PKT_COPY_T2T_BC_DW_12_dst_bank_w_shift)
+
+/*define for dst_bank_h field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_bank_h_offset 12
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_bank_h_shift  18
+#define SDMA_PKT_COPY_T2T_BC_DW_12_DST_BANK_H(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_12_dst_bank_h_mask) << SDMA_PKT_COPY_T2T_BC_DW_12_dst_bank_h_shift)
+
+/*define for dst_num_bank field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_num_bank_offset 12
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_num_bank_shift  21
+#define SDMA_PKT_COPY_T2T_BC_DW_12_DST_NUM_BANK(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_12_dst_num_bank_mask) << SDMA_PKT_COPY_T2T_BC_DW_12_dst_num_bank_shift)
+
+/*define for dst_mat_aspt field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_mat_aspt_offset 12
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_mat_aspt_shift  24
+#define SDMA_PKT_COPY_T2T_BC_DW_12_DST_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_12_dst_mat_aspt_mask) << SDMA_PKT_COPY_T2T_BC_DW_12_dst_mat_aspt_shift)
+
+/*define for dst_pipe_config field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_pipe_config_offset 12
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_pipe_config_shift  26
+#define SDMA_PKT_COPY_T2T_BC_DW_12_DST_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_12_dst_pipe_config_mask) << SDMA_PKT_COPY_T2T_BC_DW_12_dst_pipe_config_shift)
+
+/*define for DW_13 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_13_rect_x_offset 13
+#define SDMA_PKT_COPY_T2T_BC_DW_13_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_BC_DW_13_rect_x_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DW_13_RECT_X(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_13_rect_x_mask) << SDMA_PKT_COPY_T2T_BC_DW_13_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_13_rect_y_offset 13
+#define SDMA_PKT_COPY_T2T_BC_DW_13_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_BC_DW_13_rect_y_shift  16
+#define SDMA_PKT_COPY_T2T_BC_DW_13_RECT_Y(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_13_rect_y_mask) << SDMA_PKT_COPY_T2T_BC_DW_13_rect_y_shift)
+
+/*define for DW_14 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_14_rect_z_offset 14
+#define SDMA_PKT_COPY_T2T_BC_DW_14_rect_z_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_BC_DW_14_rect_z_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DW_14_RECT_Z(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_14_rect_z_mask) << SDMA_PKT_COPY_T2T_BC_DW_14_rect_z_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_14_dst_sw_offset 14
+#define SDMA_PKT_COPY_T2T_BC_DW_14_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_BC_DW_14_dst_sw_shift  16
+#define SDMA_PKT_COPY_T2T_BC_DW_14_DST_SW(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_14_dst_sw_mask) << SDMA_PKT_COPY_T2T_BC_DW_14_dst_sw_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_14_src_sw_offset 14
+#define SDMA_PKT_COPY_T2T_BC_DW_14_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_BC_DW_14_src_sw_shift  24
+#define SDMA_PKT_COPY_T2T_BC_DW_14_SRC_SW(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_14_src_sw_mask) << SDMA_PKT_COPY_T2T_BC_DW_14_src_sw_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_TILED_SUBWIN packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_OP(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_tmz_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_tmz_shift)
+
+/*define for dcc field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_dcc_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_dcc_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_dcc_shift  19
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_DCC(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_dcc_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_dcc_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_shift)
+
+/*define for TILED_ADDR_LO word*/
+/*define for tiled_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_offset 1
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_TILED_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_mask) << SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_shift)
+
+/*define for TILED_ADDR_HI word*/
+/*define for tiled_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_offset 2
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_TILED_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_mask) << SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for tiled_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_offset 3
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_TILED_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_shift)
+
+/*define for tiled_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_offset 3
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_TILED_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_shift)
+
+/*define for DW_4 word*/
+/*define for tiled_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_offset 4
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_mask   0x00001FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_TILED_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_shift)
+
+/*define for width field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_width_offset 4
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_width_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_WIDTH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_4_width_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_4_width_shift)
+
+/*define for DW_5 word*/
+/*define for height field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_height_offset 5
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_height_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_HEIGHT(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_5_height_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_5_height_shift)
+
+/*define for depth field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_depth_offset 5
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_depth_mask   0x00001FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_depth_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_DEPTH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_5_depth_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_5_depth_shift)
+
+/*define for DW_6 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_shift)
+
+/*define for swizzle_mode field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_swizzle_mode_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_swizzle_mode_mask   0x0000001F
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_swizzle_mode_shift  3
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_SWIZZLE_MODE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_swizzle_mode_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_swizzle_mode_shift)
+
+/*define for dimension field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_dimension_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_dimension_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_dimension_shift  9
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_DIMENSION(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_dimension_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_dimension_shift)
+
+/*define for mip_max field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mip_max_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mip_max_mask   0x0000000F
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mip_max_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_MIP_MAX(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mip_max_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mip_max_shift)
+
+/*define for mip_id field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mip_id_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mip_id_mask   0x0000000F
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mip_id_shift  20
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_MIP_ID(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mip_id_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mip_id_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_offset 7
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_offset 8
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for DW_9 word*/
+/*define for linear_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_offset 9
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_LINEAR_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_shift)
+
+/*define for linear_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_offset 9
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_LINEAR_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_shift)
+
+/*define for DW_10 word*/
+/*define for linear_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_offset 10
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_mask   0x00001FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_LINEAR_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_shift)
+
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_offset 10
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_shift)
+
+/*define for DW_11 word*/
+/*define for linear_slice_pitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_offset 11
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_LINEAR_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_shift)
+
+/*define for DW_12 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_offset 12
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_RECT_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_offset 12
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_RECT_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_shift)
+
+/*define for DW_13 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_mask   0x00001FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_RECT_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_shift  24
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_TILE_SW(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_shift)
+
+/*define for META_ADDR_LO word*/
+/*define for meta_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_LO_meta_addr_31_0_offset 14
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_LO_meta_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_LO_meta_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_LO_META_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_LO_meta_addr_31_0_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_LO_meta_addr_31_0_shift)
+
+/*define for META_ADDR_HI word*/
+/*define for meta_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_HI_meta_addr_63_32_offset 15
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_HI_meta_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_HI_meta_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_HI_META_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_HI_meta_addr_63_32_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_HI_meta_addr_63_32_shift)
+
+/*define for META_CONFIG word*/
+/*define for data_format field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_data_format_offset 16
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_data_format_mask   0x0000007F
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_data_format_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_DATA_FORMAT(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_data_format_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_data_format_shift)
+
+/*define for color_transform_disable field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_color_transform_disable_offset 16
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_color_transform_disable_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_color_transform_disable_shift  7
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_COLOR_TRANSFORM_DISABLE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_color_transform_disable_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_color_transform_disable_shift)
+
+/*define for alpha_is_on_msb field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_alpha_is_on_msb_offset 16
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_alpha_is_on_msb_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_alpha_is_on_msb_shift  8
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_ALPHA_IS_ON_MSB(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_alpha_is_on_msb_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_alpha_is_on_msb_shift)
+
+/*define for number_type field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_number_type_offset 16
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_number_type_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_number_type_shift  9
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_NUMBER_TYPE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_number_type_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_number_type_shift)
+
+/*define for surface_type field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_surface_type_offset 16
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_surface_type_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_surface_type_shift  12
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_SURFACE_TYPE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_surface_type_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_surface_type_shift)
+
+/*define for max_comp_block_size field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_max_comp_block_size_offset 16
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_max_comp_block_size_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_max_comp_block_size_shift  24
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_MAX_COMP_BLOCK_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_max_comp_block_size_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_max_comp_block_size_shift)
+
+/*define for max_uncomp_block_size field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_max_uncomp_block_size_offset 16
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_max_uncomp_block_size_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_max_uncomp_block_size_shift  26
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_MAX_UNCOMP_BLOCK_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_max_uncomp_block_size_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_max_uncomp_block_size_shift)
+
+/*define for write_compress_enable field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_write_compress_enable_offset 16
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_write_compress_enable_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_write_compress_enable_shift  28
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_WRITE_COMPRESS_ENABLE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_write_compress_enable_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_write_compress_enable_shift)
+
+/*define for meta_tmz field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_meta_tmz_offset 16
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_meta_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_meta_tmz_shift  29
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_META_TMZ(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_meta_tmz_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_meta_tmz_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_TILED_SUBWIN_BC packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_op_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_op_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_OP(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_op_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_sub_op_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_sub_op_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_detile_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_detile_shift)
+
+/*define for TILED_ADDR_LO word*/
+/*define for tiled_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_LO_tiled_addr_31_0_offset 1
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_LO_tiled_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_LO_tiled_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_LO_TILED_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_LO_tiled_addr_31_0_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_LO_tiled_addr_31_0_shift)
+
+/*define for TILED_ADDR_HI word*/
+/*define for tiled_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_HI_tiled_addr_63_32_offset 2
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_HI_tiled_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_HI_tiled_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_HI_TILED_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_HI_tiled_addr_63_32_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_HI_tiled_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for tiled_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_tiled_x_offset 3
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_tiled_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_tiled_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_TILED_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_tiled_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_tiled_x_shift)
+
+/*define for tiled_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_tiled_y_offset 3
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_tiled_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_tiled_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_TILED_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_tiled_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_tiled_y_shift)
+
+/*define for DW_4 word*/
+/*define for tiled_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_tiled_z_offset 4
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_tiled_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_tiled_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_TILED_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_tiled_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_tiled_z_shift)
+
+/*define for width field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_width_offset 4
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_width_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_WIDTH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_width_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_width_shift)
+
+/*define for DW_5 word*/
+/*define for height field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_height_offset 5
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_height_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_HEIGHT(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_height_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_height_shift)
+
+/*define for depth field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_depth_offset 5
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_depth_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_depth_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_DEPTH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_depth_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_depth_shift)
+
+/*define for DW_6 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_element_size_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_element_size_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_element_size_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_element_size_shift)
+
+/*define for array_mode field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_array_mode_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_array_mode_shift  3
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_array_mode_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_array_mode_shift)
+
+/*define for mit_mode field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_mit_mode_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_mit_mode_shift  8
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_MIT_MODE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_mit_mode_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_mit_mode_shift)
+
+/*define for tilesplit_size field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_tilesplit_size_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_tilesplit_size_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_tilesplit_size_shift)
+
+/*define for bank_w field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_bank_w_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_bank_w_shift  15
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_BANK_W(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_bank_w_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_bank_w_shift)
+
+/*define for bank_h field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_bank_h_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_bank_h_shift  18
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_BANK_H(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_bank_h_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_bank_h_shift)
+
+/*define for num_bank field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_num_bank_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_num_bank_shift  21
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_NUM_BANK(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_num_bank_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_num_bank_shift)
+
+/*define for mat_aspt field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_mat_aspt_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_mat_aspt_shift  24
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_MAT_ASPT(x) ((x & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_mat_aspt_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_mat_aspt_shift)
+
+/*define for pipe_config field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_pipe_config_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_pipe_config_shift  26
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_pipe_config_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_pipe_config_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_LO_linear_addr_31_0_offset 7
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_HI_linear_addr_63_32_offset 8
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for DW_9 word*/
+/*define for linear_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_linear_x_offset 9
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_linear_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_linear_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_LINEAR_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_linear_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_linear_x_shift)
+
+/*define for linear_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_linear_y_offset 9
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_linear_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_linear_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_LINEAR_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_linear_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_linear_y_shift)
+
+/*define for DW_10 word*/
+/*define for linear_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_linear_z_offset 10
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_linear_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_linear_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_LINEAR_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_linear_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_linear_z_shift)
+
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_linear_pitch_offset 10
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_linear_pitch_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_linear_pitch_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_linear_pitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_linear_pitch_shift)
+
+/*define for DW_11 word*/
+/*define for linear_slice_pitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_11_linear_slice_pitch_offset 11
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_11_linear_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_11_linear_slice_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_11_LINEAR_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_11_linear_slice_pitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_11_linear_slice_pitch_shift)
+
+/*define for DW_12 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_rect_x_offset 12
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_rect_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_RECT_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_rect_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_rect_y_offset 12
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_rect_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_RECT_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_rect_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_rect_y_shift)
+
+/*define for DW_13 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_rect_z_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_rect_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_rect_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_RECT_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_rect_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_rect_z_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_linear_sw_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_linear_sw_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_linear_sw_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_linear_sw_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_tile_sw_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_tile_sw_shift  24
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_TILE_SW(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_tile_sw_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_tile_sw_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_STRUCT packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_op_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_STRUCT_HEADER_op_shift  0
+#define SDMA_PKT_COPY_STRUCT_HEADER_OP(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_op_mask) << SDMA_PKT_COPY_STRUCT_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_STRUCT_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_STRUCT_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_sub_op_mask) << SDMA_PKT_COPY_STRUCT_HEADER_sub_op_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_STRUCT_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_STRUCT_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_tmz_mask) << SDMA_PKT_COPY_STRUCT_HEADER_tmz_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_STRUCT_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_STRUCT_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_detile_mask) << SDMA_PKT_COPY_STRUCT_HEADER_detile_shift)
+
+/*define for SB_ADDR_LO word*/
+/*define for sb_addr_31_0 field*/
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_offset 1
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_shift  0
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_SB_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_mask) << SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_shift)
+
+/*define for SB_ADDR_HI word*/
+/*define for sb_addr_63_32 field*/
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_offset 2
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_shift  0
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_SB_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_mask) << SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_shift)
+
+/*define for START_INDEX word*/
+/*define for start_index field*/
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_offset 3
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_shift  0
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_START_INDEX(x) (((x) & SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_mask) << SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_STRUCT_COUNT_count_offset 4
+#define SDMA_PKT_COPY_STRUCT_COUNT_count_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_COUNT_count_shift  0
+#define SDMA_PKT_COPY_STRUCT_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_STRUCT_COUNT_count_mask) << SDMA_PKT_COPY_STRUCT_COUNT_count_shift)
+
+/*define for DW_5 word*/
+/*define for stride field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_stride_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_stride_mask   0x000007FF
+#define SDMA_PKT_COPY_STRUCT_DW_5_stride_shift  0
+#define SDMA_PKT_COPY_STRUCT_DW_5_STRIDE(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_stride_mask) << SDMA_PKT_COPY_STRUCT_DW_5_stride_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_shift  16
+#define SDMA_PKT_COPY_STRUCT_DW_5_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_mask) << SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_shift)
+
+/*define for struct_sw field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_mask   0x00000003
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_shift  24
+#define SDMA_PKT_COPY_STRUCT_DW_5_STRUCT_SW(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_mask) << SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_offset 6
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_offset 7
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_UNTILED packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_UNTILED_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_UNTILED_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_UNTILED_HEADER_op_mask) << SDMA_PKT_WRITE_UNTILED_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_UNTILED_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_mask) << SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_shift)
+
+/*define for encrypt field*/
+#define SDMA_PKT_WRITE_UNTILED_HEADER_encrypt_offset 0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_encrypt_mask   0x00000001
+#define SDMA_PKT_WRITE_UNTILED_HEADER_encrypt_shift  16
+#define SDMA_PKT_WRITE_UNTILED_HEADER_ENCRYPT(x) (((x) & SDMA_PKT_WRITE_UNTILED_HEADER_encrypt_mask) << SDMA_PKT_WRITE_UNTILED_HEADER_encrypt_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_WRITE_UNTILED_HEADER_tmz_offset 0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_WRITE_UNTILED_HEADER_tmz_shift  18
+#define SDMA_PKT_WRITE_UNTILED_HEADER_TMZ(x) (((x) & SDMA_PKT_WRITE_UNTILED_HEADER_tmz_mask) << SDMA_PKT_WRITE_UNTILED_HEADER_tmz_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_UNTILED_DW_3_count_offset 3
+#define SDMA_PKT_WRITE_UNTILED_DW_3_count_mask   0x000FFFFF
+#define SDMA_PKT_WRITE_UNTILED_DW_3_count_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(x) (((x) & SDMA_PKT_WRITE_UNTILED_DW_3_count_mask) << SDMA_PKT_WRITE_UNTILED_DW_3_count_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_WRITE_UNTILED_DW_3_sw_offset 3
+#define SDMA_PKT_WRITE_UNTILED_DW_3_sw_mask   0x00000003
+#define SDMA_PKT_WRITE_UNTILED_DW_3_sw_shift  24
+#define SDMA_PKT_WRITE_UNTILED_DW_3_SW(x) (((x) & SDMA_PKT_WRITE_UNTILED_DW_3_sw_mask) << SDMA_PKT_WRITE_UNTILED_DW_3_sw_shift)
+
+/*define for DATA0 word*/
+/*define for data0 field*/
+#define SDMA_PKT_WRITE_UNTILED_DATA0_data0_offset 4
+#define SDMA_PKT_WRITE_UNTILED_DATA0_data0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_UNTILED_DATA0_data0_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DATA0_DATA0(x) (((x) & SDMA_PKT_WRITE_UNTILED_DATA0_data0_mask) << SDMA_PKT_WRITE_UNTILED_DATA0_data0_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_TILED packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_TILED_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_TILED_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_TILED_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_TILED_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_TILED_HEADER_op_mask) << SDMA_PKT_WRITE_TILED_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_TILED_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_TILED_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_TILED_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_TILED_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_TILED_HEADER_sub_op_mask) << SDMA_PKT_WRITE_TILED_HEADER_sub_op_shift)
+
+/*define for encrypt field*/
+#define SDMA_PKT_WRITE_TILED_HEADER_encrypt_offset 0
+#define SDMA_PKT_WRITE_TILED_HEADER_encrypt_mask   0x00000001
+#define SDMA_PKT_WRITE_TILED_HEADER_encrypt_shift  16
+#define SDMA_PKT_WRITE_TILED_HEADER_ENCRYPT(x) (((x) & SDMA_PKT_WRITE_TILED_HEADER_encrypt_mask) << SDMA_PKT_WRITE_TILED_HEADER_encrypt_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_WRITE_TILED_HEADER_tmz_offset 0
+#define SDMA_PKT_WRITE_TILED_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_WRITE_TILED_HEADER_tmz_shift  18
+#define SDMA_PKT_WRITE_TILED_HEADER_TMZ(x) (((x) & SDMA_PKT_WRITE_TILED_HEADER_tmz_mask) << SDMA_PKT_WRITE_TILED_HEADER_tmz_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for width field*/
+#define SDMA_PKT_WRITE_TILED_DW_3_width_offset 3
+#define SDMA_PKT_WRITE_TILED_DW_3_width_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_3_width_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_3_WIDTH(x) (((x) & SDMA_PKT_WRITE_TILED_DW_3_width_mask) << SDMA_PKT_WRITE_TILED_DW_3_width_shift)
+
+/*define for DW_4 word*/
+/*define for height field*/
+#define SDMA_PKT_WRITE_TILED_DW_4_height_offset 4
+#define SDMA_PKT_WRITE_TILED_DW_4_height_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_4_height_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_4_HEIGHT(x) (((x) & SDMA_PKT_WRITE_TILED_DW_4_height_mask) << SDMA_PKT_WRITE_TILED_DW_4_height_shift)
+
+/*define for depth field*/
+#define SDMA_PKT_WRITE_TILED_DW_4_depth_offset 4
+#define SDMA_PKT_WRITE_TILED_DW_4_depth_mask   0x00001FFF
+#define SDMA_PKT_WRITE_TILED_DW_4_depth_shift  16
+#define SDMA_PKT_WRITE_TILED_DW_4_DEPTH(x) (((x) & SDMA_PKT_WRITE_TILED_DW_4_depth_mask) << SDMA_PKT_WRITE_TILED_DW_4_depth_shift)
+
+/*define for DW_5 word*/
+/*define for element_size field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_element_size_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_element_size_mask   0x00000007
+#define SDMA_PKT_WRITE_TILED_DW_5_element_size_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_5_ELEMENT_SIZE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_element_size_mask) << SDMA_PKT_WRITE_TILED_DW_5_element_size_shift)
+
+/*define for swizzle_mode field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_swizzle_mode_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_swizzle_mode_mask   0x0000001F
+#define SDMA_PKT_WRITE_TILED_DW_5_swizzle_mode_shift  3
+#define SDMA_PKT_WRITE_TILED_DW_5_SWIZZLE_MODE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_swizzle_mode_mask) << SDMA_PKT_WRITE_TILED_DW_5_swizzle_mode_shift)
+
+/*define for dimension field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_dimension_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_dimension_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_5_dimension_shift  9
+#define SDMA_PKT_WRITE_TILED_DW_5_DIMENSION(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_dimension_mask) << SDMA_PKT_WRITE_TILED_DW_5_dimension_shift)
+
+/*define for mip_max field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_mip_max_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_mip_max_mask   0x0000000F
+#define SDMA_PKT_WRITE_TILED_DW_5_mip_max_shift  16
+#define SDMA_PKT_WRITE_TILED_DW_5_MIP_MAX(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_mip_max_mask) << SDMA_PKT_WRITE_TILED_DW_5_mip_max_shift)
+
+/*define for DW_6 word*/
+/*define for x field*/
+#define SDMA_PKT_WRITE_TILED_DW_6_x_offset 6
+#define SDMA_PKT_WRITE_TILED_DW_6_x_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_6_x_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_6_X(x) (((x) & SDMA_PKT_WRITE_TILED_DW_6_x_mask) << SDMA_PKT_WRITE_TILED_DW_6_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_WRITE_TILED_DW_6_y_offset 6
+#define SDMA_PKT_WRITE_TILED_DW_6_y_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_6_y_shift  16
+#define SDMA_PKT_WRITE_TILED_DW_6_Y(x) (((x) & SDMA_PKT_WRITE_TILED_DW_6_y_mask) << SDMA_PKT_WRITE_TILED_DW_6_y_shift)
+
+/*define for DW_7 word*/
+/*define for z field*/
+#define SDMA_PKT_WRITE_TILED_DW_7_z_offset 7
+#define SDMA_PKT_WRITE_TILED_DW_7_z_mask   0x00001FFF
+#define SDMA_PKT_WRITE_TILED_DW_7_z_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_7_Z(x) (((x) & SDMA_PKT_WRITE_TILED_DW_7_z_mask) << SDMA_PKT_WRITE_TILED_DW_7_z_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_WRITE_TILED_DW_7_sw_offset 7
+#define SDMA_PKT_WRITE_TILED_DW_7_sw_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_7_sw_shift  24
+#define SDMA_PKT_WRITE_TILED_DW_7_SW(x) (((x) & SDMA_PKT_WRITE_TILED_DW_7_sw_mask) << SDMA_PKT_WRITE_TILED_DW_7_sw_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_TILED_COUNT_count_offset 8
+#define SDMA_PKT_WRITE_TILED_COUNT_count_mask   0x000FFFFF
+#define SDMA_PKT_WRITE_TILED_COUNT_count_shift  0
+#define SDMA_PKT_WRITE_TILED_COUNT_COUNT(x) (((x) & SDMA_PKT_WRITE_TILED_COUNT_count_mask) << SDMA_PKT_WRITE_TILED_COUNT_count_shift)
+
+/*define for DATA0 word*/
+/*define for data0 field*/
+#define SDMA_PKT_WRITE_TILED_DATA0_data0_offset 9
+#define SDMA_PKT_WRITE_TILED_DATA0_data0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_DATA0_data0_shift  0
+#define SDMA_PKT_WRITE_TILED_DATA0_DATA0(x) (((x) & SDMA_PKT_WRITE_TILED_DATA0_data0_mask) << SDMA_PKT_WRITE_TILED_DATA0_data0_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_TILED_BC packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_TILED_BC_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_TILED_BC_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_TILED_BC_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_TILED_BC_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_TILED_BC_HEADER_op_mask) << SDMA_PKT_WRITE_TILED_BC_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_TILED_BC_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_TILED_BC_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_TILED_BC_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_TILED_BC_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_TILED_BC_HEADER_sub_op_mask) << SDMA_PKT_WRITE_TILED_BC_HEADER_sub_op_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_TILED_BC_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_TILED_BC_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_BC_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_TILED_BC_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_TILED_BC_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_TILED_BC_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_TILED_BC_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_BC_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_TILED_BC_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_TILED_BC_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for width field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_3_width_offset 3
+#define SDMA_PKT_WRITE_TILED_BC_DW_3_width_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_BC_DW_3_width_shift  0
+#define SDMA_PKT_WRITE_TILED_BC_DW_3_WIDTH(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_3_width_mask) << SDMA_PKT_WRITE_TILED_BC_DW_3_width_shift)
+
+/*define for DW_4 word*/
+/*define for height field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_4_height_offset 4
+#define SDMA_PKT_WRITE_TILED_BC_DW_4_height_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_BC_DW_4_height_shift  0
+#define SDMA_PKT_WRITE_TILED_BC_DW_4_HEIGHT(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_4_height_mask) << SDMA_PKT_WRITE_TILED_BC_DW_4_height_shift)
+
+/*define for depth field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_4_depth_offset 4
+#define SDMA_PKT_WRITE_TILED_BC_DW_4_depth_mask   0x000007FF
+#define SDMA_PKT_WRITE_TILED_BC_DW_4_depth_shift  16
+#define SDMA_PKT_WRITE_TILED_BC_DW_4_DEPTH(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_4_depth_mask) << SDMA_PKT_WRITE_TILED_BC_DW_4_depth_shift)
+
+/*define for DW_5 word*/
+/*define for element_size field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_element_size_offset 5
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_element_size_mask   0x00000007
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_element_size_shift  0
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_ELEMENT_SIZE(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_5_element_size_mask) << SDMA_PKT_WRITE_TILED_BC_DW_5_element_size_shift)
+
+/*define for array_mode field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_array_mode_offset 5
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_array_mode_mask   0x0000000F
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_array_mode_shift  3
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_ARRAY_MODE(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_5_array_mode_mask) << SDMA_PKT_WRITE_TILED_BC_DW_5_array_mode_shift)
+
+/*define for mit_mode field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_mit_mode_offset 5
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_mit_mode_mask   0x00000007
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_mit_mode_shift  8
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_MIT_MODE(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_5_mit_mode_mask) << SDMA_PKT_WRITE_TILED_BC_DW_5_mit_mode_shift)
+
+/*define for tilesplit_size field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_tilesplit_size_offset 5
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_tilesplit_size_shift  11
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_5_tilesplit_size_mask) << SDMA_PKT_WRITE_TILED_BC_DW_5_tilesplit_size_shift)
+
+/*define for bank_w field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_bank_w_offset 5
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_bank_w_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_bank_w_shift  15
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_BANK_W(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_5_bank_w_mask) << SDMA_PKT_WRITE_TILED_BC_DW_5_bank_w_shift)
+
+/*define for bank_h field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_bank_h_offset 5
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_bank_h_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_bank_h_shift  18
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_BANK_H(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_5_bank_h_mask) << SDMA_PKT_WRITE_TILED_BC_DW_5_bank_h_shift)
+
+/*define for num_bank field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_num_bank_offset 5
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_num_bank_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_num_bank_shift  21
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_NUM_BANK(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_5_num_bank_mask) << SDMA_PKT_WRITE_TILED_BC_DW_5_num_bank_shift)
+
+/*define for mat_aspt field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_mat_aspt_offset 5
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_mat_aspt_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_mat_aspt_shift  24
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_MAT_ASPT(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_5_mat_aspt_mask) << SDMA_PKT_WRITE_TILED_BC_DW_5_mat_aspt_shift)
+
+/*define for pipe_config field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_pipe_config_offset 5
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_pipe_config_mask   0x0000001F
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_pipe_config_shift  26
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_PIPE_CONFIG(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_5_pipe_config_mask) << SDMA_PKT_WRITE_TILED_BC_DW_5_pipe_config_shift)
+
+/*define for DW_6 word*/
+/*define for x field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_6_x_offset 6
+#define SDMA_PKT_WRITE_TILED_BC_DW_6_x_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_BC_DW_6_x_shift  0
+#define SDMA_PKT_WRITE_TILED_BC_DW_6_X(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_6_x_mask) << SDMA_PKT_WRITE_TILED_BC_DW_6_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_6_y_offset 6
+#define SDMA_PKT_WRITE_TILED_BC_DW_6_y_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_BC_DW_6_y_shift  16
+#define SDMA_PKT_WRITE_TILED_BC_DW_6_Y(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_6_y_mask) << SDMA_PKT_WRITE_TILED_BC_DW_6_y_shift)
+
+/*define for DW_7 word*/
+/*define for z field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_7_z_offset 7
+#define SDMA_PKT_WRITE_TILED_BC_DW_7_z_mask   0x000007FF
+#define SDMA_PKT_WRITE_TILED_BC_DW_7_z_shift  0
+#define SDMA_PKT_WRITE_TILED_BC_DW_7_Z(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_7_z_mask) << SDMA_PKT_WRITE_TILED_BC_DW_7_z_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_7_sw_offset 7
+#define SDMA_PKT_WRITE_TILED_BC_DW_7_sw_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_BC_DW_7_sw_shift  24
+#define SDMA_PKT_WRITE_TILED_BC_DW_7_SW(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_7_sw_mask) << SDMA_PKT_WRITE_TILED_BC_DW_7_sw_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_TILED_BC_COUNT_count_offset 8
+#define SDMA_PKT_WRITE_TILED_BC_COUNT_count_mask   0x000FFFFF
+#define SDMA_PKT_WRITE_TILED_BC_COUNT_count_shift  2
+#define SDMA_PKT_WRITE_TILED_BC_COUNT_COUNT(x) (((x) & SDMA_PKT_WRITE_TILED_BC_COUNT_count_mask) << SDMA_PKT_WRITE_TILED_BC_COUNT_count_shift)
+
+/*define for DATA0 word*/
+/*define for data0 field*/
+#define SDMA_PKT_WRITE_TILED_BC_DATA0_data0_offset 9
+#define SDMA_PKT_WRITE_TILED_BC_DATA0_data0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_BC_DATA0_data0_shift  0
+#define SDMA_PKT_WRITE_TILED_BC_DATA0_DATA0(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DATA0_data0_mask) << SDMA_PKT_WRITE_TILED_BC_DATA0_data0_shift)
+
+
+/*
+** Definitions for SDMA_PKT_PTEPDE_COPY packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_PTEPDE_COPY_HEADER_op_offset 0
+#define SDMA_PKT_PTEPDE_COPY_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_COPY_HEADER_op_shift  0
+#define SDMA_PKT_PTEPDE_COPY_HEADER_OP(x) (((x) & SDMA_PKT_PTEPDE_COPY_HEADER_op_mask) << SDMA_PKT_PTEPDE_COPY_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_PTEPDE_COPY_HEADER_sub_op_offset 0
+#define SDMA_PKT_PTEPDE_COPY_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_COPY_HEADER_sub_op_shift  8
+#define SDMA_PKT_PTEPDE_COPY_HEADER_SUB_OP(x) (((x) & SDMA_PKT_PTEPDE_COPY_HEADER_sub_op_mask) << SDMA_PKT_PTEPDE_COPY_HEADER_sub_op_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_PTEPDE_COPY_HEADER_tmz_offset 0
+#define SDMA_PKT_PTEPDE_COPY_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_PTEPDE_COPY_HEADER_tmz_shift  18
+#define SDMA_PKT_PTEPDE_COPY_HEADER_TMZ(x) (((x) & SDMA_PKT_PTEPDE_COPY_HEADER_tmz_mask) << SDMA_PKT_PTEPDE_COPY_HEADER_tmz_shift)
+
+/*define for ptepde_op field*/
+#define SDMA_PKT_PTEPDE_COPY_HEADER_ptepde_op_offset 0
+#define SDMA_PKT_PTEPDE_COPY_HEADER_ptepde_op_mask   0x00000001
+#define SDMA_PKT_PTEPDE_COPY_HEADER_ptepde_op_shift  31
+#define SDMA_PKT_PTEPDE_COPY_HEADER_PTEPDE_OP(x) (((x) & SDMA_PKT_PTEPDE_COPY_HEADER_ptepde_op_mask) << SDMA_PKT_PTEPDE_COPY_HEADER_ptepde_op_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_PTEPDE_COPY_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_PTEPDE_COPY_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_PTEPDE_COPY_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_PTEPDE_COPY_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_LO_dst_addr_31_0_offset 3
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_PTEPDE_COPY_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_PTEPDE_COPY_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_HI_dst_addr_63_32_offset 4
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_PTEPDE_COPY_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_PTEPDE_COPY_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for MASK_DW0 word*/
+/*define for mask_dw0 field*/
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW0_mask_dw0_offset 5
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW0_mask_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW0_mask_dw0_shift  0
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW0_MASK_DW0(x) (((x) & SDMA_PKT_PTEPDE_COPY_MASK_DW0_mask_dw0_mask) << SDMA_PKT_PTEPDE_COPY_MASK_DW0_mask_dw0_shift)
+
+/*define for MASK_DW1 word*/
+/*define for mask_dw1 field*/
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW1_mask_dw1_offset 6
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW1_mask_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW1_mask_dw1_shift  0
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW1_MASK_DW1(x) (((x) & SDMA_PKT_PTEPDE_COPY_MASK_DW1_mask_dw1_mask) << SDMA_PKT_PTEPDE_COPY_MASK_DW1_mask_dw1_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_PTEPDE_COPY_COUNT_count_offset 7
+#define SDMA_PKT_PTEPDE_COPY_COUNT_count_mask   0x0007FFFF
+#define SDMA_PKT_PTEPDE_COPY_COUNT_count_shift  0
+#define SDMA_PKT_PTEPDE_COPY_COUNT_COUNT(x) (((x) & SDMA_PKT_PTEPDE_COPY_COUNT_count_mask) << SDMA_PKT_PTEPDE_COPY_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_PTEPDE_COPY_BACKWARDS packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_op_offset 0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_op_shift  0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_OP(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_op_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_sub_op_offset 0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_sub_op_shift  8
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_SUB_OP(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_sub_op_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_sub_op_shift)
+
+/*define for pte_size field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_pte_size_offset 0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_pte_size_mask   0x00000003
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_pte_size_shift  28
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_PTE_SIZE(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_pte_size_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_pte_size_shift)
+
+/*define for direction field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_direction_offset 0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_direction_mask   0x00000001
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_direction_shift  30
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_DIRECTION(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_direction_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_direction_shift)
+
+/*define for ptepde_op field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_ptepde_op_offset 0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_ptepde_op_mask   0x00000001
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_ptepde_op_shift  31
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_PTEPDE_OP(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_ptepde_op_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_ptepde_op_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_LO_dst_addr_31_0_offset 3
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_HI_dst_addr_63_32_offset 4
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for MASK_BIT_FOR_DW word*/
+/*define for mask_first_xfer field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_first_xfer_offset 5
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_first_xfer_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_first_xfer_shift  0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_MASK_FIRST_XFER(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_first_xfer_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_first_xfer_shift)
+
+/*define for mask_last_xfer field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_last_xfer_offset 5
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_last_xfer_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_last_xfer_shift  8
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_MASK_LAST_XFER(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_last_xfer_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_last_xfer_shift)
+
+/*define for COUNT_IN_32B_XFER word*/
+/*define for count field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_COUNT_IN_32B_XFER_count_offset 6
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_COUNT_IN_32B_XFER_count_mask   0x0001FFFF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_COUNT_IN_32B_XFER_count_shift  0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_COUNT_IN_32B_XFER_COUNT(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_COUNT_IN_32B_XFER_count_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_COUNT_IN_32B_XFER_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_PTEPDE_RMW packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_op_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_RMW_HEADER_op_shift  0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_OP(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_op_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_sub_op_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_RMW_HEADER_sub_op_shift  8
+#define SDMA_PKT_PTEPDE_RMW_HEADER_SUB_OP(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_sub_op_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_sub_op_shift)
+
+/*define for mtype field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_mtype_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_mtype_mask   0x00000007
+#define SDMA_PKT_PTEPDE_RMW_HEADER_mtype_shift  16
+#define SDMA_PKT_PTEPDE_RMW_HEADER_MTYPE(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_mtype_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_mtype_shift)
+
+/*define for gcc field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_gcc_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_gcc_mask   0x00000001
+#define SDMA_PKT_PTEPDE_RMW_HEADER_gcc_shift  19
+#define SDMA_PKT_PTEPDE_RMW_HEADER_GCC(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_gcc_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_gcc_shift)
+
+/*define for sys field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_sys_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_sys_mask   0x00000001
+#define SDMA_PKT_PTEPDE_RMW_HEADER_sys_shift  20
+#define SDMA_PKT_PTEPDE_RMW_HEADER_SYS(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_sys_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_sys_shift)
+
+/*define for snp field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_snp_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_snp_mask   0x00000001
+#define SDMA_PKT_PTEPDE_RMW_HEADER_snp_shift  22
+#define SDMA_PKT_PTEPDE_RMW_HEADER_SNP(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_snp_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_snp_shift)
+
+/*define for gpa field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_gpa_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_gpa_mask   0x00000001
+#define SDMA_PKT_PTEPDE_RMW_HEADER_gpa_shift  23
+#define SDMA_PKT_PTEPDE_RMW_HEADER_GPA(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_gpa_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_gpa_shift)
+
+/*define for l2_policy field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_l2_policy_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_l2_policy_mask   0x00000003
+#define SDMA_PKT_PTEPDE_RMW_HEADER_l2_policy_shift  24
+#define SDMA_PKT_PTEPDE_RMW_HEADER_L2_POLICY(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_l2_policy_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_l2_policy_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_PTEPDE_RMW_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_PTEPDE_RMW_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_RMW_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_PTEPDE_RMW_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_PTEPDE_RMW_ADDR_LO_addr_31_0_mask) << SDMA_PKT_PTEPDE_RMW_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_PTEPDE_RMW_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_PTEPDE_RMW_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_RMW_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_PTEPDE_RMW_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_PTEPDE_RMW_ADDR_HI_addr_63_32_mask) << SDMA_PKT_PTEPDE_RMW_ADDR_HI_addr_63_32_shift)
+
+/*define for MASK_LO word*/
+/*define for mask_31_0 field*/
+#define SDMA_PKT_PTEPDE_RMW_MASK_LO_mask_31_0_offset 3
+#define SDMA_PKT_PTEPDE_RMW_MASK_LO_mask_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_RMW_MASK_LO_mask_31_0_shift  0
+#define SDMA_PKT_PTEPDE_RMW_MASK_LO_MASK_31_0(x) (((x) & SDMA_PKT_PTEPDE_RMW_MASK_LO_mask_31_0_mask) << SDMA_PKT_PTEPDE_RMW_MASK_LO_mask_31_0_shift)
+
+/*define for MASK_HI word*/
+/*define for mask_63_32 field*/
+#define SDMA_PKT_PTEPDE_RMW_MASK_HI_mask_63_32_offset 4
+#define SDMA_PKT_PTEPDE_RMW_MASK_HI_mask_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_RMW_MASK_HI_mask_63_32_shift  0
+#define SDMA_PKT_PTEPDE_RMW_MASK_HI_MASK_63_32(x) (((x) & SDMA_PKT_PTEPDE_RMW_MASK_HI_mask_63_32_mask) << SDMA_PKT_PTEPDE_RMW_MASK_HI_mask_63_32_shift)
+
+/*define for VALUE_LO word*/
+/*define for value_31_0 field*/
+#define SDMA_PKT_PTEPDE_RMW_VALUE_LO_value_31_0_offset 5
+#define SDMA_PKT_PTEPDE_RMW_VALUE_LO_value_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_RMW_VALUE_LO_value_31_0_shift  0
+#define SDMA_PKT_PTEPDE_RMW_VALUE_LO_VALUE_31_0(x) (((x) & SDMA_PKT_PTEPDE_RMW_VALUE_LO_value_31_0_mask) << SDMA_PKT_PTEPDE_RMW_VALUE_LO_value_31_0_shift)
+
+/*define for VALUE_HI word*/
+/*define for value_63_32 field*/
+#define SDMA_PKT_PTEPDE_RMW_VALUE_HI_value_63_32_offset 6
+#define SDMA_PKT_PTEPDE_RMW_VALUE_HI_value_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_RMW_VALUE_HI_value_63_32_shift  0
+#define SDMA_PKT_PTEPDE_RMW_VALUE_HI_VALUE_63_32(x) (((x) & SDMA_PKT_PTEPDE_RMW_VALUE_HI_value_63_32_mask) << SDMA_PKT_PTEPDE_RMW_VALUE_HI_value_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_INCR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_INCR_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_INCR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_INCR_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_INCR_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_INCR_HEADER_op_mask) << SDMA_PKT_WRITE_INCR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_INCR_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_INCR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_INCR_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_INCR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_INCR_HEADER_sub_op_mask) << SDMA_PKT_WRITE_INCR_HEADER_sub_op_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for MASK_DW0 word*/
+/*define for mask_dw0 field*/
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_offset 3
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_shift  0
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_MASK_DW0(x) (((x) & SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_mask) << SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_shift)
+
+/*define for MASK_DW1 word*/
+/*define for mask_dw1 field*/
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_offset 4
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_shift  0
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_MASK_DW1(x) (((x) & SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_mask) << SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_shift)
+
+/*define for INIT_DW0 word*/
+/*define for init_dw0 field*/
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_offset 5
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_shift  0
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_INIT_DW0(x) (((x) & SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_mask) << SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_shift)
+
+/*define for INIT_DW1 word*/
+/*define for init_dw1 field*/
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_offset 6
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_shift  0
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_INIT_DW1(x) (((x) & SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_mask) << SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_shift)
+
+/*define for INCR_DW0 word*/
+/*define for incr_dw0 field*/
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_offset 7
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_shift  0
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_INCR_DW0(x) (((x) & SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_mask) << SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_shift)
+
+/*define for INCR_DW1 word*/
+/*define for incr_dw1 field*/
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_offset 8
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_shift  0
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_INCR_DW1(x) (((x) & SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_mask) << SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_INCR_COUNT_count_offset 9
+#define SDMA_PKT_WRITE_INCR_COUNT_count_mask   0x0007FFFF
+#define SDMA_PKT_WRITE_INCR_COUNT_count_shift  0
+#define SDMA_PKT_WRITE_INCR_COUNT_COUNT(x) (((x) & SDMA_PKT_WRITE_INCR_COUNT_count_mask) << SDMA_PKT_WRITE_INCR_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_INDIRECT packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_INDIRECT_HEADER_op_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_INDIRECT_HEADER_op_shift  0
+#define SDMA_PKT_INDIRECT_HEADER_OP(x) (((x) & SDMA_PKT_INDIRECT_HEADER_op_mask) << SDMA_PKT_INDIRECT_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_INDIRECT_HEADER_sub_op_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_INDIRECT_HEADER_sub_op_shift  8
+#define SDMA_PKT_INDIRECT_HEADER_SUB_OP(x) (((x) & SDMA_PKT_INDIRECT_HEADER_sub_op_mask) << SDMA_PKT_INDIRECT_HEADER_sub_op_shift)
+
+/*define for vmid field*/
+#define SDMA_PKT_INDIRECT_HEADER_vmid_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_vmid_mask   0x0000000F
+#define SDMA_PKT_INDIRECT_HEADER_vmid_shift  16
+#define SDMA_PKT_INDIRECT_HEADER_VMID(x) (((x) & SDMA_PKT_INDIRECT_HEADER_vmid_mask) << SDMA_PKT_INDIRECT_HEADER_vmid_shift)
+
+/*define for priv field*/
+#define SDMA_PKT_INDIRECT_HEADER_priv_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_priv_mask   0x00000001
+#define SDMA_PKT_INDIRECT_HEADER_priv_shift  31
+#define SDMA_PKT_INDIRECT_HEADER_PRIV(x) (((x) & SDMA_PKT_INDIRECT_HEADER_priv_mask) << SDMA_PKT_INDIRECT_HEADER_priv_shift)
+
+/*define for BASE_LO word*/
+/*define for ib_base_31_0 field*/
+#define SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_offset 1
+#define SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_shift  0
+#define SDMA_PKT_INDIRECT_BASE_LO_IB_BASE_31_0(x) (((x) & SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_mask) << SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_shift)
+
+/*define for BASE_HI word*/
+/*define for ib_base_63_32 field*/
+#define SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_offset 2
+#define SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_shift  0
+#define SDMA_PKT_INDIRECT_BASE_HI_IB_BASE_63_32(x) (((x) & SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_mask) << SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_shift)
+
+/*define for IB_SIZE word*/
+/*define for ib_size field*/
+#define SDMA_PKT_INDIRECT_IB_SIZE_ib_size_offset 3
+#define SDMA_PKT_INDIRECT_IB_SIZE_ib_size_mask   0x000FFFFF
+#define SDMA_PKT_INDIRECT_IB_SIZE_ib_size_shift  0
+#define SDMA_PKT_INDIRECT_IB_SIZE_IB_SIZE(x) (((x) & SDMA_PKT_INDIRECT_IB_SIZE_ib_size_mask) << SDMA_PKT_INDIRECT_IB_SIZE_ib_size_shift)
+
+/*define for CSA_ADDR_LO word*/
+/*define for csa_addr_31_0 field*/
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_offset 4
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_shift  0
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_CSA_ADDR_31_0(x) (((x) & SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_mask) << SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_shift)
+
+/*define for CSA_ADDR_HI word*/
+/*define for csa_addr_63_32 field*/
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_offset 5
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_shift  0
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_CSA_ADDR_63_32(x) (((x) & SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_mask) << SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_SEMAPHORE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_op_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_SEMAPHORE_HEADER_op_shift  0
+#define SDMA_PKT_SEMAPHORE_HEADER_OP(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_op_mask) << SDMA_PKT_SEMAPHORE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_sub_op_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_SEMAPHORE_HEADER_sub_op_shift  8
+#define SDMA_PKT_SEMAPHORE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_sub_op_mask) << SDMA_PKT_SEMAPHORE_HEADER_sub_op_shift)
+
+/*define for write_one field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_write_one_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_write_one_mask   0x00000001
+#define SDMA_PKT_SEMAPHORE_HEADER_write_one_shift  29
+#define SDMA_PKT_SEMAPHORE_HEADER_WRITE_ONE(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_write_one_mask) << SDMA_PKT_SEMAPHORE_HEADER_write_one_shift)
+
+/*define for signal field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_signal_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_signal_mask   0x00000001
+#define SDMA_PKT_SEMAPHORE_HEADER_signal_shift  30
+#define SDMA_PKT_SEMAPHORE_HEADER_SIGNAL(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_signal_mask) << SDMA_PKT_SEMAPHORE_HEADER_signal_shift)
+
+/*define for mailbox field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_mailbox_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_mailbox_mask   0x00000001
+#define SDMA_PKT_SEMAPHORE_HEADER_mailbox_shift  31
+#define SDMA_PKT_SEMAPHORE_HEADER_MAILBOX(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_mailbox_mask) << SDMA_PKT_SEMAPHORE_HEADER_mailbox_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_mask) << SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_mask) << SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_FENCE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_FENCE_HEADER_op_offset 0
+#define SDMA_PKT_FENCE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_FENCE_HEADER_op_shift  0
+#define SDMA_PKT_FENCE_HEADER_OP(x) (((x) & SDMA_PKT_FENCE_HEADER_op_mask) << SDMA_PKT_FENCE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_FENCE_HEADER_sub_op_offset 0
+#define SDMA_PKT_FENCE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_FENCE_HEADER_sub_op_shift  8
+#define SDMA_PKT_FENCE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_FENCE_HEADER_sub_op_mask) << SDMA_PKT_FENCE_HEADER_sub_op_shift)
+
+/*define for mtype field*/
+#define SDMA_PKT_FENCE_HEADER_mtype_offset 0
+#define SDMA_PKT_FENCE_HEADER_mtype_mask   0x00000007
+#define SDMA_PKT_FENCE_HEADER_mtype_shift  16
+#define SDMA_PKT_FENCE_HEADER_MTYPE(x) (((x) & SDMA_PKT_FENCE_HEADER_mtype_mask) << SDMA_PKT_FENCE_HEADER_mtype_shift)
+
+/*define for gcc field*/
+#define SDMA_PKT_FENCE_HEADER_gcc_offset 0
+#define SDMA_PKT_FENCE_HEADER_gcc_mask   0x00000001
+#define SDMA_PKT_FENCE_HEADER_gcc_shift  19
+#define SDMA_PKT_FENCE_HEADER_GCC(x) (((x) & SDMA_PKT_FENCE_HEADER_gcc_mask) << SDMA_PKT_FENCE_HEADER_gcc_shift)
+
+/*define for sys field*/
+#define SDMA_PKT_FENCE_HEADER_sys_offset 0
+#define SDMA_PKT_FENCE_HEADER_sys_mask   0x00000001
+#define SDMA_PKT_FENCE_HEADER_sys_shift  20
+#define SDMA_PKT_FENCE_HEADER_SYS(x) (((x) & SDMA_PKT_FENCE_HEADER_sys_mask) << SDMA_PKT_FENCE_HEADER_sys_shift)
+
+/*define for snp field*/
+#define SDMA_PKT_FENCE_HEADER_snp_offset 0
+#define SDMA_PKT_FENCE_HEADER_snp_mask   0x00000001
+#define SDMA_PKT_FENCE_HEADER_snp_shift  22
+#define SDMA_PKT_FENCE_HEADER_SNP(x) (((x) & SDMA_PKT_FENCE_HEADER_snp_mask) << SDMA_PKT_FENCE_HEADER_snp_shift)
+
+/*define for gpa field*/
+#define SDMA_PKT_FENCE_HEADER_gpa_offset 0
+#define SDMA_PKT_FENCE_HEADER_gpa_mask   0x00000001
+#define SDMA_PKT_FENCE_HEADER_gpa_shift  23
+#define SDMA_PKT_FENCE_HEADER_GPA(x) (((x) & SDMA_PKT_FENCE_HEADER_gpa_mask) << SDMA_PKT_FENCE_HEADER_gpa_shift)
+
+/*define for l2_policy field*/
+#define SDMA_PKT_FENCE_HEADER_l2_policy_offset 0
+#define SDMA_PKT_FENCE_HEADER_l2_policy_mask   0x00000003
+#define SDMA_PKT_FENCE_HEADER_l2_policy_shift  24
+#define SDMA_PKT_FENCE_HEADER_L2_POLICY(x) (((x) & SDMA_PKT_FENCE_HEADER_l2_policy_mask) << SDMA_PKT_FENCE_HEADER_l2_policy_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_FENCE_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_FENCE_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_FENCE_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_FENCE_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_FENCE_ADDR_LO_addr_31_0_mask) << SDMA_PKT_FENCE_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_FENCE_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_FENCE_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_FENCE_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_FENCE_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_FENCE_ADDR_HI_addr_63_32_mask) << SDMA_PKT_FENCE_ADDR_HI_addr_63_32_shift)
+
+/*define for DATA word*/
+/*define for data field*/
+#define SDMA_PKT_FENCE_DATA_data_offset 3
+#define SDMA_PKT_FENCE_DATA_data_mask   0xFFFFFFFF
+#define SDMA_PKT_FENCE_DATA_data_shift  0
+#define SDMA_PKT_FENCE_DATA_DATA(x) (((x) & SDMA_PKT_FENCE_DATA_data_mask) << SDMA_PKT_FENCE_DATA_data_shift)
+
+
+/*
+** Definitions for SDMA_PKT_SRBM_WRITE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_SRBM_WRITE_HEADER_op_offset 0
+#define SDMA_PKT_SRBM_WRITE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_SRBM_WRITE_HEADER_op_shift  0
+#define SDMA_PKT_SRBM_WRITE_HEADER_OP(x) (((x) & SDMA_PKT_SRBM_WRITE_HEADER_op_mask) << SDMA_PKT_SRBM_WRITE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_SRBM_WRITE_HEADER_sub_op_offset 0
+#define SDMA_PKT_SRBM_WRITE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_SRBM_WRITE_HEADER_sub_op_shift  8
+#define SDMA_PKT_SRBM_WRITE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_SRBM_WRITE_HEADER_sub_op_mask) << SDMA_PKT_SRBM_WRITE_HEADER_sub_op_shift)
+
+/*define for byte_en field*/
+#define SDMA_PKT_SRBM_WRITE_HEADER_byte_en_offset 0
+#define SDMA_PKT_SRBM_WRITE_HEADER_byte_en_mask   0x0000000F
+#define SDMA_PKT_SRBM_WRITE_HEADER_byte_en_shift  28
+#define SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(x) (((x) & SDMA_PKT_SRBM_WRITE_HEADER_byte_en_mask) << SDMA_PKT_SRBM_WRITE_HEADER_byte_en_shift)
+
+/*define for ADDR word*/
+/*define for addr field*/
+#define SDMA_PKT_SRBM_WRITE_ADDR_addr_offset 1
+#define SDMA_PKT_SRBM_WRITE_ADDR_addr_mask   0x0003FFFF
+#define SDMA_PKT_SRBM_WRITE_ADDR_addr_shift  0
+#define SDMA_PKT_SRBM_WRITE_ADDR_ADDR(x) (((x) & SDMA_PKT_SRBM_WRITE_ADDR_addr_mask) << SDMA_PKT_SRBM_WRITE_ADDR_addr_shift)
+
+/*define for apertureid field*/
+#define SDMA_PKT_SRBM_WRITE_ADDR_apertureid_offset 1
+#define SDMA_PKT_SRBM_WRITE_ADDR_apertureid_mask   0x00000FFF
+#define SDMA_PKT_SRBM_WRITE_ADDR_apertureid_shift  20
+#define SDMA_PKT_SRBM_WRITE_ADDR_APERTUREID(x) (((x) & SDMA_PKT_SRBM_WRITE_ADDR_apertureid_mask) << SDMA_PKT_SRBM_WRITE_ADDR_apertureid_shift)
+
+/*define for DATA word*/
+/*define for data field*/
+#define SDMA_PKT_SRBM_WRITE_DATA_data_offset 2
+#define SDMA_PKT_SRBM_WRITE_DATA_data_mask   0xFFFFFFFF
+#define SDMA_PKT_SRBM_WRITE_DATA_data_shift  0
+#define SDMA_PKT_SRBM_WRITE_DATA_DATA(x) (((x) & SDMA_PKT_SRBM_WRITE_DATA_data_mask) << SDMA_PKT_SRBM_WRITE_DATA_data_shift)
+
+
+/*
+** Definitions for SDMA_PKT_PRE_EXE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_PRE_EXE_HEADER_op_offset 0
+#define SDMA_PKT_PRE_EXE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_PRE_EXE_HEADER_op_shift  0
+#define SDMA_PKT_PRE_EXE_HEADER_OP(x) (((x) & SDMA_PKT_PRE_EXE_HEADER_op_mask) << SDMA_PKT_PRE_EXE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_PRE_EXE_HEADER_sub_op_offset 0
+#define SDMA_PKT_PRE_EXE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_PRE_EXE_HEADER_sub_op_shift  8
+#define SDMA_PKT_PRE_EXE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_PRE_EXE_HEADER_sub_op_mask) << SDMA_PKT_PRE_EXE_HEADER_sub_op_shift)
+
+/*define for dev_sel field*/
+#define SDMA_PKT_PRE_EXE_HEADER_dev_sel_offset 0
+#define SDMA_PKT_PRE_EXE_HEADER_dev_sel_mask   0x000000FF
+#define SDMA_PKT_PRE_EXE_HEADER_dev_sel_shift  16
+#define SDMA_PKT_PRE_EXE_HEADER_DEV_SEL(x) (((x) & SDMA_PKT_PRE_EXE_HEADER_dev_sel_mask) << SDMA_PKT_PRE_EXE_HEADER_dev_sel_shift)
+
+/*define for EXEC_COUNT word*/
+/*define for exec_count field*/
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_offset 1
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_mask   0x00003FFF
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_shift  0
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_EXEC_COUNT(x) (((x) & SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_mask) << SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COND_EXE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COND_EXE_HEADER_op_offset 0
+#define SDMA_PKT_COND_EXE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COND_EXE_HEADER_op_shift  0
+#define SDMA_PKT_COND_EXE_HEADER_OP(x) (((x) & SDMA_PKT_COND_EXE_HEADER_op_mask) << SDMA_PKT_COND_EXE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COND_EXE_HEADER_sub_op_offset 0
+#define SDMA_PKT_COND_EXE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COND_EXE_HEADER_sub_op_shift  8
+#define SDMA_PKT_COND_EXE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COND_EXE_HEADER_sub_op_mask) << SDMA_PKT_COND_EXE_HEADER_sub_op_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_COND_EXE_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_mask) << SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_COND_EXE_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_mask) << SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_shift)
+
+/*define for REFERENCE word*/
+/*define for reference field*/
+#define SDMA_PKT_COND_EXE_REFERENCE_reference_offset 3
+#define SDMA_PKT_COND_EXE_REFERENCE_reference_mask   0xFFFFFFFF
+#define SDMA_PKT_COND_EXE_REFERENCE_reference_shift  0
+#define SDMA_PKT_COND_EXE_REFERENCE_REFERENCE(x) (((x) & SDMA_PKT_COND_EXE_REFERENCE_reference_mask) << SDMA_PKT_COND_EXE_REFERENCE_reference_shift)
+
+/*define for EXEC_COUNT word*/
+/*define for exec_count field*/
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_offset 4
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_mask   0x00003FFF
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_shift  0
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_EXEC_COUNT(x) (((x) & SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_mask) << SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_CONSTANT_FILL packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_op_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_CONSTANT_FILL_HEADER_op_shift  0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_OP(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_op_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_shift  8
+#define SDMA_PKT_CONSTANT_FILL_HEADER_SUB_OP(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sw_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sw_mask   0x00000003
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sw_shift  16
+#define SDMA_PKT_CONSTANT_FILL_HEADER_SW(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_sw_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_sw_shift)
+
+/*define for fillsize field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_mask   0x00000003
+#define SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_shift  30
+#define SDMA_PKT_CONSTANT_FILL_HEADER_FILLSIZE(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DATA word*/
+/*define for src_data_31_0 field*/
+#define SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_offset 3
+#define SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_shift  0
+#define SDMA_PKT_CONSTANT_FILL_DATA_SRC_DATA_31_0(x) (((x) & SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_mask) << SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_CONSTANT_FILL_COUNT_count_offset 4
+#define SDMA_PKT_CONSTANT_FILL_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_CONSTANT_FILL_COUNT_count_shift  0
+#define SDMA_PKT_CONSTANT_FILL_COUNT_COUNT(x) (((x) & SDMA_PKT_CONSTANT_FILL_COUNT_count_mask) << SDMA_PKT_CONSTANT_FILL_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_DATA_FILL_MULTI packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_op_offset 0
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_op_shift  0
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_OP(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_HEADER_op_mask) << SDMA_PKT_DATA_FILL_MULTI_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_sub_op_offset 0
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_sub_op_shift  8
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_SUB_OP(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_HEADER_sub_op_mask) << SDMA_PKT_DATA_FILL_MULTI_HEADER_sub_op_shift)
+
+/*define for memlog_clr field*/
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_memlog_clr_offset 0
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_memlog_clr_mask   0x00000001
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_memlog_clr_shift  31
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_MEMLOG_CLR(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_HEADER_memlog_clr_mask) << SDMA_PKT_DATA_FILL_MULTI_HEADER_memlog_clr_shift)
+
+/*define for BYTE_STRIDE word*/
+/*define for byte_stride field*/
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_STRIDE_byte_stride_offset 1
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_STRIDE_byte_stride_mask   0xFFFFFFFF
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_STRIDE_byte_stride_shift  0
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_STRIDE_BYTE_STRIDE(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_BYTE_STRIDE_byte_stride_mask) << SDMA_PKT_DATA_FILL_MULTI_BYTE_STRIDE_byte_stride_shift)
+
+/*define for DMA_COUNT word*/
+/*define for dma_count field*/
+#define SDMA_PKT_DATA_FILL_MULTI_DMA_COUNT_dma_count_offset 2
+#define SDMA_PKT_DATA_FILL_MULTI_DMA_COUNT_dma_count_mask   0xFFFFFFFF
+#define SDMA_PKT_DATA_FILL_MULTI_DMA_COUNT_dma_count_shift  0
+#define SDMA_PKT_DATA_FILL_MULTI_DMA_COUNT_DMA_COUNT(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_DMA_COUNT_dma_count_mask) << SDMA_PKT_DATA_FILL_MULTI_DMA_COUNT_dma_count_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_LO_dst_addr_31_0_offset 3
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_HI_dst_addr_63_32_offset 4
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for BYTE_COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_COUNT_count_offset 5
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_COUNT_count_mask   0x03FFFFFF
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_COUNT_count_shift  0
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_COUNT_COUNT(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_BYTE_COUNT_count_mask) << SDMA_PKT_DATA_FILL_MULTI_BYTE_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_POLL_REGMEM packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_op_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_POLL_REGMEM_HEADER_op_shift  0
+#define SDMA_PKT_POLL_REGMEM_HEADER_OP(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_op_mask) << SDMA_PKT_POLL_REGMEM_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_sub_op_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_POLL_REGMEM_HEADER_sub_op_shift  8
+#define SDMA_PKT_POLL_REGMEM_HEADER_SUB_OP(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_sub_op_mask) << SDMA_PKT_POLL_REGMEM_HEADER_sub_op_shift)
+
+/*define for hdp_flush field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_mask   0x00000001
+#define SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_shift  26
+#define SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_mask) << SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_shift)
+
+/*define for func field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_func_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_func_mask   0x00000007
+#define SDMA_PKT_POLL_REGMEM_HEADER_func_shift  28
+#define SDMA_PKT_POLL_REGMEM_HEADER_FUNC(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_func_mask) << SDMA_PKT_POLL_REGMEM_HEADER_func_shift)
+
+/*define for mem_poll field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_mask   0x00000001
+#define SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_shift  31
+#define SDMA_PKT_POLL_REGMEM_HEADER_MEM_POLL(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_mask) << SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_mask) << SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_mask) << SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_shift)
+
+/*define for VALUE word*/
+/*define for value field*/
+#define SDMA_PKT_POLL_REGMEM_VALUE_value_offset 3
+#define SDMA_PKT_POLL_REGMEM_VALUE_value_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_VALUE_value_shift  0
+#define SDMA_PKT_POLL_REGMEM_VALUE_VALUE(x) (((x) & SDMA_PKT_POLL_REGMEM_VALUE_value_mask) << SDMA_PKT_POLL_REGMEM_VALUE_value_shift)
+
+/*define for MASK word*/
+/*define for mask field*/
+#define SDMA_PKT_POLL_REGMEM_MASK_mask_offset 4
+#define SDMA_PKT_POLL_REGMEM_MASK_mask_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_MASK_mask_shift  0
+#define SDMA_PKT_POLL_REGMEM_MASK_MASK(x) (((x) & SDMA_PKT_POLL_REGMEM_MASK_mask_mask) << SDMA_PKT_POLL_REGMEM_MASK_mask_shift)
+
+/*define for DW5 word*/
+/*define for interval field*/
+#define SDMA_PKT_POLL_REGMEM_DW5_interval_offset 5
+#define SDMA_PKT_POLL_REGMEM_DW5_interval_mask   0x0000FFFF
+#define SDMA_PKT_POLL_REGMEM_DW5_interval_shift  0
+#define SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(x) (((x) & SDMA_PKT_POLL_REGMEM_DW5_interval_mask) << SDMA_PKT_POLL_REGMEM_DW5_interval_shift)
+
+/*define for retry_count field*/
+#define SDMA_PKT_POLL_REGMEM_DW5_retry_count_offset 5
+#define SDMA_PKT_POLL_REGMEM_DW5_retry_count_mask   0x00000FFF
+#define SDMA_PKT_POLL_REGMEM_DW5_retry_count_shift  16
+#define SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(x) (((x) & SDMA_PKT_POLL_REGMEM_DW5_retry_count_mask) << SDMA_PKT_POLL_REGMEM_DW5_retry_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_POLL_REG_WRITE_MEM packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_op_offset 0
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_op_shift  0
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_OP(x) (((x) & SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_op_mask) << SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_sub_op_offset 0
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_sub_op_shift  8
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_SUB_OP(x) (((x) & SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_sub_op_mask) << SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_sub_op_shift)
+
+/*define for SRC_ADDR word*/
+/*define for addr_31_2 field*/
+#define SDMA_PKT_POLL_REG_WRITE_MEM_SRC_ADDR_addr_31_2_offset 1
+#define SDMA_PKT_POLL_REG_WRITE_MEM_SRC_ADDR_addr_31_2_mask   0x3FFFFFFF
+#define SDMA_PKT_POLL_REG_WRITE_MEM_SRC_ADDR_addr_31_2_shift  2
+#define SDMA_PKT_POLL_REG_WRITE_MEM_SRC_ADDR_ADDR_31_2(x) (((x) & SDMA_PKT_POLL_REG_WRITE_MEM_SRC_ADDR_addr_31_2_mask) << SDMA_PKT_POLL_REG_WRITE_MEM_SRC_ADDR_addr_31_2_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_LO_addr_31_0_offset 2
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_LO_addr_31_0_mask) << SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_LO_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_HI_addr_63_32_offset 3
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_HI_addr_63_32_mask) << SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_HI_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_POLL_DBIT_WRITE_MEM packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_op_offset 0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_op_shift  0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_OP(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_op_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_sub_op_offset 0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_sub_op_shift  8
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_SUB_OP(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_sub_op_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_sub_op_shift)
+
+/*define for ea field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_ea_offset 0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_ea_mask   0x00000003
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_ea_shift  16
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_EA(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_ea_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_ea_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_LO_addr_31_0_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_LO_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_HI_addr_63_32_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_HI_addr_63_32_shift)
+
+/*define for START_PAGE word*/
+/*define for addr_31_4 field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_START_PAGE_addr_31_4_offset 3
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_START_PAGE_addr_31_4_mask   0x0FFFFFFF
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_START_PAGE_addr_31_4_shift  4
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_START_PAGE_ADDR_31_4(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_START_PAGE_addr_31_4_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_START_PAGE_addr_31_4_shift)
+
+/*define for PAGE_NUM word*/
+/*define for page_num_31_0 field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_PAGE_NUM_page_num_31_0_offset 4
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_PAGE_NUM_page_num_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_PAGE_NUM_page_num_31_0_shift  0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_PAGE_NUM_PAGE_NUM_31_0(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_PAGE_NUM_page_num_31_0_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_PAGE_NUM_page_num_31_0_shift)
+
+
+/*
+** Definitions for SDMA_PKT_POLL_MEM_VERIFY packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_op_offset 0
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_op_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_OP(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_HEADER_op_mask) << SDMA_PKT_POLL_MEM_VERIFY_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_sub_op_offset 0
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_sub_op_shift  8
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_SUB_OP(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_HEADER_sub_op_mask) << SDMA_PKT_POLL_MEM_VERIFY_HEADER_sub_op_shift)
+
+/*define for mode field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_mode_offset 0
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_mode_mask   0x00000001
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_mode_shift  31
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_MODE(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_HEADER_mode_mask) << SDMA_PKT_POLL_MEM_VERIFY_HEADER_mode_shift)
+
+/*define for PATTERN word*/
+/*define for pattern field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_PATTERN_pattern_offset 1
+#define SDMA_PKT_POLL_MEM_VERIFY_PATTERN_pattern_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_PATTERN_pattern_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_PATTERN_PATTERN(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_PATTERN_pattern_mask) << SDMA_PKT_POLL_MEM_VERIFY_PATTERN_pattern_shift)
+
+/*define for CMP0_ADDR_START_LO word*/
+/*define for cmp0_start_31_0 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_LO_cmp0_start_31_0_offset 2
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_LO_cmp0_start_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_LO_cmp0_start_31_0_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_LO_CMP0_START_31_0(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_LO_cmp0_start_31_0_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_LO_cmp0_start_31_0_shift)
+
+/*define for CMP0_ADDR_START_HI word*/
+/*define for cmp0_start_63_32 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_HI_cmp0_start_63_32_offset 3
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_HI_cmp0_start_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_HI_cmp0_start_63_32_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_HI_CMP0_START_63_32(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_HI_cmp0_start_63_32_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_HI_cmp0_start_63_32_shift)
+
+/*define for CMP0_ADDR_END_LO word*/
+/*define for cmp1_end_31_0 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_LO_cmp1_end_31_0_offset 4
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_LO_cmp1_end_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_LO_cmp1_end_31_0_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_LO_CMP1_END_31_0(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_LO_cmp1_end_31_0_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_LO_cmp1_end_31_0_shift)
+
+/*define for CMP0_ADDR_END_HI word*/
+/*define for cmp1_end_63_32 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_HI_cmp1_end_63_32_offset 5
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_HI_cmp1_end_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_HI_cmp1_end_63_32_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_HI_CMP1_END_63_32(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_HI_cmp1_end_63_32_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_HI_cmp1_end_63_32_shift)
+
+/*define for CMP1_ADDR_START_LO word*/
+/*define for cmp1_start_31_0 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_LO_cmp1_start_31_0_offset 6
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_LO_cmp1_start_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_LO_cmp1_start_31_0_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_LO_CMP1_START_31_0(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_LO_cmp1_start_31_0_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_LO_cmp1_start_31_0_shift)
+
+/*define for CMP1_ADDR_START_HI word*/
+/*define for cmp1_start_63_32 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_HI_cmp1_start_63_32_offset 7
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_HI_cmp1_start_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_HI_cmp1_start_63_32_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_HI_CMP1_START_63_32(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_HI_cmp1_start_63_32_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_HI_cmp1_start_63_32_shift)
+
+/*define for CMP1_ADDR_END_LO word*/
+/*define for cmp1_end_31_0 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_LO_cmp1_end_31_0_offset 8
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_LO_cmp1_end_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_LO_cmp1_end_31_0_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_LO_CMP1_END_31_0(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_LO_cmp1_end_31_0_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_LO_cmp1_end_31_0_shift)
+
+/*define for CMP1_ADDR_END_HI word*/
+/*define for cmp1_end_63_32 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_HI_cmp1_end_63_32_offset 9
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_HI_cmp1_end_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_HI_cmp1_end_63_32_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_HI_CMP1_END_63_32(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_HI_cmp1_end_63_32_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_HI_cmp1_end_63_32_shift)
+
+/*define for REC_ADDR_LO word*/
+/*define for rec_31_0 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_LO_rec_31_0_offset 10
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_LO_rec_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_LO_rec_31_0_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_LO_REC_31_0(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_LO_rec_31_0_mask) << SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_LO_rec_31_0_shift)
+
+/*define for REC_ADDR_HI word*/
+/*define for rec_63_32 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_HI_rec_63_32_offset 11
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_HI_rec_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_HI_rec_63_32_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_HI_REC_63_32(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_HI_rec_63_32_mask) << SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_HI_rec_63_32_shift)
+
+/*define for RESERVED word*/
+/*define for reserved field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_RESERVED_reserved_offset 12
+#define SDMA_PKT_POLL_MEM_VERIFY_RESERVED_reserved_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_RESERVED_reserved_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_RESERVED_RESERVED(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_RESERVED_reserved_mask) << SDMA_PKT_POLL_MEM_VERIFY_RESERVED_reserved_shift)
+
+
+/*
+** Definitions for SDMA_PKT_VM_INVALIDATION packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_VM_INVALIDATION_HEADER_op_offset 0
+#define SDMA_PKT_VM_INVALIDATION_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_VM_INVALIDATION_HEADER_op_shift  0
+#define SDMA_PKT_VM_INVALIDATION_HEADER_OP(x) (((x) & SDMA_PKT_VM_INVALIDATION_HEADER_op_mask) << SDMA_PKT_VM_INVALIDATION_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_VM_INVALIDATION_HEADER_sub_op_offset 0
+#define SDMA_PKT_VM_INVALIDATION_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_VM_INVALIDATION_HEADER_sub_op_shift  8
+#define SDMA_PKT_VM_INVALIDATION_HEADER_SUB_OP(x) (((x) & SDMA_PKT_VM_INVALIDATION_HEADER_sub_op_mask) << SDMA_PKT_VM_INVALIDATION_HEADER_sub_op_shift)
+
+/*define for gfx_eng_id field*/
+#define SDMA_PKT_VM_INVALIDATION_HEADER_gfx_eng_id_offset 0
+#define SDMA_PKT_VM_INVALIDATION_HEADER_gfx_eng_id_mask   0x0000001F
+#define SDMA_PKT_VM_INVALIDATION_HEADER_gfx_eng_id_shift  16
+#define SDMA_PKT_VM_INVALIDATION_HEADER_GFX_ENG_ID(x) (((x) & SDMA_PKT_VM_INVALIDATION_HEADER_gfx_eng_id_mask) << SDMA_PKT_VM_INVALIDATION_HEADER_gfx_eng_id_shift)
+
+/*define for mm_eng_id field*/
+#define SDMA_PKT_VM_INVALIDATION_HEADER_mm_eng_id_offset 0
+#define SDMA_PKT_VM_INVALIDATION_HEADER_mm_eng_id_mask   0x0000001F
+#define SDMA_PKT_VM_INVALIDATION_HEADER_mm_eng_id_shift  24
+#define SDMA_PKT_VM_INVALIDATION_HEADER_MM_ENG_ID(x) (((x) & SDMA_PKT_VM_INVALIDATION_HEADER_mm_eng_id_mask) << SDMA_PKT_VM_INVALIDATION_HEADER_mm_eng_id_shift)
+
+/*define for INVALIDATEREQ word*/
+/*define for invalidatereq field*/
+#define SDMA_PKT_VM_INVALIDATION_INVALIDATEREQ_invalidatereq_offset 1
+#define SDMA_PKT_VM_INVALIDATION_INVALIDATEREQ_invalidatereq_mask   0xFFFFFFFF
+#define SDMA_PKT_VM_INVALIDATION_INVALIDATEREQ_invalidatereq_shift  0
+#define SDMA_PKT_VM_INVALIDATION_INVALIDATEREQ_INVALIDATEREQ(x) (((x) & SDMA_PKT_VM_INVALIDATION_INVALIDATEREQ_invalidatereq_mask) << SDMA_PKT_VM_INVALIDATION_INVALIDATEREQ_invalidatereq_shift)
+
+/*define for ADDRESSRANGELO word*/
+/*define for addressrangelo field*/
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGELO_addressrangelo_offset 2
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGELO_addressrangelo_mask   0xFFFFFFFF
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGELO_addressrangelo_shift  0
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGELO_ADDRESSRANGELO(x) (((x) & SDMA_PKT_VM_INVALIDATION_ADDRESSRANGELO_addressrangelo_mask) << SDMA_PKT_VM_INVALIDATION_ADDRESSRANGELO_addressrangelo_shift)
+
+/*define for ADDRESSRANGEHI word*/
+/*define for invalidateack field*/
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_invalidateack_offset 3
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_invalidateack_mask   0x0000FFFF
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_invalidateack_shift  0
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_INVALIDATEACK(x) (((x) & SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_invalidateack_mask) << SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_invalidateack_shift)
+
+/*define for addressrangehi field*/
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_addressrangehi_offset 3
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_addressrangehi_mask   0x0000001F
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_addressrangehi_shift  16
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_ADDRESSRANGEHI(x) (((x) & SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_addressrangehi_mask) << SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_addressrangehi_shift)
+
+/*define for reserved field*/
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_reserved_offset 3
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_reserved_mask   0x000001FF
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_reserved_shift  23
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_RESERVED(x) (((x) & SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_reserved_mask) << SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_reserved_shift)
+
+
+/*
+** Definitions for SDMA_PKT_ATOMIC packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_ATOMIC_HEADER_op_offset 0
+#define SDMA_PKT_ATOMIC_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_ATOMIC_HEADER_op_shift  0
+#define SDMA_PKT_ATOMIC_HEADER_OP(x) (((x) & SDMA_PKT_ATOMIC_HEADER_op_mask) << SDMA_PKT_ATOMIC_HEADER_op_shift)
+
+/*define for loop field*/
+#define SDMA_PKT_ATOMIC_HEADER_loop_offset 0
+#define SDMA_PKT_ATOMIC_HEADER_loop_mask   0x00000001
+#define SDMA_PKT_ATOMIC_HEADER_loop_shift  16
+#define SDMA_PKT_ATOMIC_HEADER_LOOP(x) (((x) & SDMA_PKT_ATOMIC_HEADER_loop_mask) << SDMA_PKT_ATOMIC_HEADER_loop_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_ATOMIC_HEADER_tmz_offset 0
+#define SDMA_PKT_ATOMIC_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_ATOMIC_HEADER_tmz_shift  18
+#define SDMA_PKT_ATOMIC_HEADER_TMZ(x) (((x) & SDMA_PKT_ATOMIC_HEADER_tmz_mask) << SDMA_PKT_ATOMIC_HEADER_tmz_shift)
+
+/*define for atomic_op field*/
+#define SDMA_PKT_ATOMIC_HEADER_atomic_op_offset 0
+#define SDMA_PKT_ATOMIC_HEADER_atomic_op_mask   0x0000007F
+#define SDMA_PKT_ATOMIC_HEADER_atomic_op_shift  25
+#define SDMA_PKT_ATOMIC_HEADER_ATOMIC_OP(x) (((x) & SDMA_PKT_ATOMIC_HEADER_atomic_op_mask) << SDMA_PKT_ATOMIC_HEADER_atomic_op_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_ATOMIC_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_mask) << SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_ATOMIC_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_mask) << SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_shift)
+
+/*define for SRC_DATA_LO word*/
+/*define for src_data_31_0 field*/
+#define SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_offset 3
+#define SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_shift  0
+#define SDMA_PKT_ATOMIC_SRC_DATA_LO_SRC_DATA_31_0(x) (((x) & SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_mask) << SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_shift)
+
+/*define for SRC_DATA_HI word*/
+/*define for src_data_63_32 field*/
+#define SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_offset 4
+#define SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_shift  0
+#define SDMA_PKT_ATOMIC_SRC_DATA_HI_SRC_DATA_63_32(x) (((x) & SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_mask) << SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_shift)
+
+/*define for CMP_DATA_LO word*/
+/*define for cmp_data_31_0 field*/
+#define SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_offset 5
+#define SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_shift  0
+#define SDMA_PKT_ATOMIC_CMP_DATA_LO_CMP_DATA_31_0(x) (((x) & SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_mask) << SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_shift)
+
+/*define for CMP_DATA_HI word*/
+/*define for cmp_data_63_32 field*/
+#define SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_offset 6
+#define SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_shift  0
+#define SDMA_PKT_ATOMIC_CMP_DATA_HI_CMP_DATA_63_32(x) (((x) & SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_mask) << SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_shift)
+
+/*define for LOOP_INTERVAL word*/
+/*define for loop_interval field*/
+#define SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_offset 7
+#define SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_mask   0x00001FFF
+#define SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_shift  0
+#define SDMA_PKT_ATOMIC_LOOP_INTERVAL_LOOP_INTERVAL(x) (((x) & SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_mask) << SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TIMESTAMP_SET packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_op_offset 0
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_op_shift  0
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_OP(x) (((x) & SDMA_PKT_TIMESTAMP_SET_HEADER_op_mask) << SDMA_PKT_TIMESTAMP_SET_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_offset 0
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_shift  8
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_mask) << SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_shift)
+
+/*define for INIT_DATA_LO word*/
+/*define for init_data_31_0 field*/
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_offset 1
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_shift  0
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_INIT_DATA_31_0(x) (((x) & SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_mask) << SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_shift)
+
+/*define for INIT_DATA_HI word*/
+/*define for init_data_63_32 field*/
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_offset 2
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_shift  0
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_INIT_DATA_63_32(x) (((x) & SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_mask) << SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TIMESTAMP_GET packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_op_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_HEADER_op_mask) << SDMA_PKT_TIMESTAMP_GET_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_shift  8
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_mask) << SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_shift)
+
+/*define for WRITE_ADDR_LO word*/
+/*define for write_addr_31_3 field*/
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_offset 1
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_mask   0x1FFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_shift  3
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_WRITE_ADDR_31_3(x) (((x) & SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_mask) << SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_shift)
+
+/*define for WRITE_ADDR_HI word*/
+/*define for write_addr_63_32 field*/
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_offset 2
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_WRITE_ADDR_63_32(x) (((x) & SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_mask) << SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TIMESTAMP_GET_GLOBAL packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_shift  8
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_shift)
+
+/*define for WRITE_ADDR_LO word*/
+/*define for write_addr_31_3 field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_offset 1
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_mask   0x1FFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_shift  3
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_WRITE_ADDR_31_3(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_shift)
+
+/*define for WRITE_ADDR_HI word*/
+/*define for write_addr_63_32 field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_offset 2
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_WRITE_ADDR_63_32(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TRAP packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TRAP_HEADER_op_offset 0
+#define SDMA_PKT_TRAP_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TRAP_HEADER_op_shift  0
+#define SDMA_PKT_TRAP_HEADER_OP(x) (((x) & SDMA_PKT_TRAP_HEADER_op_mask) << SDMA_PKT_TRAP_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TRAP_HEADER_sub_op_offset 0
+#define SDMA_PKT_TRAP_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TRAP_HEADER_sub_op_shift  8
+#define SDMA_PKT_TRAP_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TRAP_HEADER_sub_op_mask) << SDMA_PKT_TRAP_HEADER_sub_op_shift)
+
+/*define for INT_CONTEXT word*/
+/*define for int_context field*/
+#define SDMA_PKT_TRAP_INT_CONTEXT_int_context_offset 1
+#define SDMA_PKT_TRAP_INT_CONTEXT_int_context_mask   0x0FFFFFFF
+#define SDMA_PKT_TRAP_INT_CONTEXT_int_context_shift  0
+#define SDMA_PKT_TRAP_INT_CONTEXT_INT_CONTEXT(x) (((x) & SDMA_PKT_TRAP_INT_CONTEXT_int_context_mask) << SDMA_PKT_TRAP_INT_CONTEXT_int_context_shift)
+
+
+/*
+** Definitions for SDMA_PKT_DUMMY_TRAP packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_DUMMY_TRAP_HEADER_op_offset 0
+#define SDMA_PKT_DUMMY_TRAP_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_DUMMY_TRAP_HEADER_op_shift  0
+#define SDMA_PKT_DUMMY_TRAP_HEADER_OP(x) (((x) & SDMA_PKT_DUMMY_TRAP_HEADER_op_mask) << SDMA_PKT_DUMMY_TRAP_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_DUMMY_TRAP_HEADER_sub_op_offset 0
+#define SDMA_PKT_DUMMY_TRAP_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_DUMMY_TRAP_HEADER_sub_op_shift  8
+#define SDMA_PKT_DUMMY_TRAP_HEADER_SUB_OP(x) (((x) & SDMA_PKT_DUMMY_TRAP_HEADER_sub_op_mask) << SDMA_PKT_DUMMY_TRAP_HEADER_sub_op_shift)
+
+/*define for INT_CONTEXT word*/
+/*define for int_context field*/
+#define SDMA_PKT_DUMMY_TRAP_INT_CONTEXT_int_context_offset 1
+#define SDMA_PKT_DUMMY_TRAP_INT_CONTEXT_int_context_mask   0x0FFFFFFF
+#define SDMA_PKT_DUMMY_TRAP_INT_CONTEXT_int_context_shift  0
+#define SDMA_PKT_DUMMY_TRAP_INT_CONTEXT_INT_CONTEXT(x) (((x) & SDMA_PKT_DUMMY_TRAP_INT_CONTEXT_int_context_mask) << SDMA_PKT_DUMMY_TRAP_INT_CONTEXT_int_context_shift)
+
+
+/*
+** Definitions for SDMA_PKT_GPUVM_INV packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_GPUVM_INV_HEADER_op_offset 0
+#define SDMA_PKT_GPUVM_INV_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_GPUVM_INV_HEADER_op_shift  0
+#define SDMA_PKT_GPUVM_INV_HEADER_OP(x) (((x) & SDMA_PKT_GPUVM_INV_HEADER_op_mask) << SDMA_PKT_GPUVM_INV_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_GPUVM_INV_HEADER_sub_op_offset 0
+#define SDMA_PKT_GPUVM_INV_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_GPUVM_INV_HEADER_sub_op_shift  8
+#define SDMA_PKT_GPUVM_INV_HEADER_SUB_OP(x) (((x) & SDMA_PKT_GPUVM_INV_HEADER_sub_op_mask) << SDMA_PKT_GPUVM_INV_HEADER_sub_op_shift)
+
+/*define for PAYLOAD1 word*/
+/*define for per_vmid_inv_req field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_per_vmid_inv_req_offset 1
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_per_vmid_inv_req_mask   0x0000FFFF
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_per_vmid_inv_req_shift  0
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_PER_VMID_INV_REQ(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD1_per_vmid_inv_req_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD1_per_vmid_inv_req_shift)
+
+/*define for flush_type field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_flush_type_offset 1
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_flush_type_mask   0x00000007
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_flush_type_shift  16
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_FLUSH_TYPE(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD1_flush_type_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD1_flush_type_shift)
+
+/*define for l2_ptes field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_ptes_offset 1
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_ptes_mask   0x00000001
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_ptes_shift  19
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_L2_PTES(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_ptes_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_ptes_shift)
+
+/*define for l2_pde0 field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde0_offset 1
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde0_mask   0x00000001
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde0_shift  20
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_L2_PDE0(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde0_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde0_shift)
+
+/*define for l2_pde1 field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde1_offset 1
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde1_mask   0x00000001
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde1_shift  21
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_L2_PDE1(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde1_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde1_shift)
+
+/*define for l2_pde2 field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde2_offset 1
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde2_mask   0x00000001
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde2_shift  22
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_L2_PDE2(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde2_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde2_shift)
+
+/*define for l1_ptes field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l1_ptes_offset 1
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l1_ptes_mask   0x00000001
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l1_ptes_shift  23
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_L1_PTES(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD1_l1_ptes_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD1_l1_ptes_shift)
+
+/*define for clr_protection_fault_status_addr field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_clr_protection_fault_status_addr_offset 1
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_clr_protection_fault_status_addr_mask   0x00000001
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_clr_protection_fault_status_addr_shift  24
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_CLR_PROTECTION_FAULT_STATUS_ADDR(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD1_clr_protection_fault_status_addr_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD1_clr_protection_fault_status_addr_shift)
+
+/*define for log_request field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_log_request_offset 1
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_log_request_mask   0x00000001
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_log_request_shift  25
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_LOG_REQUEST(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD1_log_request_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD1_log_request_shift)
+
+/*define for four_kilobytes field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_four_kilobytes_offset 1
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_four_kilobytes_mask   0x00000001
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_four_kilobytes_shift  26
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_FOUR_KILOBYTES(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD1_four_kilobytes_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD1_four_kilobytes_shift)
+
+/*define for PAYLOAD2 word*/
+/*define for s field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD2_s_offset 2
+#define SDMA_PKT_GPUVM_INV_PAYLOAD2_s_mask   0x00000001
+#define SDMA_PKT_GPUVM_INV_PAYLOAD2_s_shift  0
+#define SDMA_PKT_GPUVM_INV_PAYLOAD2_S(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD2_s_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD2_s_shift)
+
+/*define for page_va_42_12 field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD2_page_va_42_12_offset 2
+#define SDMA_PKT_GPUVM_INV_PAYLOAD2_page_va_42_12_mask   0x7FFFFFFF
+#define SDMA_PKT_GPUVM_INV_PAYLOAD2_page_va_42_12_shift  1
+#define SDMA_PKT_GPUVM_INV_PAYLOAD2_PAGE_VA_42_12(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD2_page_va_42_12_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD2_page_va_42_12_shift)
+
+/*define for PAYLOAD3 word*/
+/*define for page_va_47_43 field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD3_page_va_47_43_offset 3
+#define SDMA_PKT_GPUVM_INV_PAYLOAD3_page_va_47_43_mask   0x0000003F
+#define SDMA_PKT_GPUVM_INV_PAYLOAD3_page_va_47_43_shift  0
+#define SDMA_PKT_GPUVM_INV_PAYLOAD3_PAGE_VA_47_43(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD3_page_va_47_43_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD3_page_va_47_43_shift)
+
+
+/*
+** Definitions for SDMA_PKT_GCR_REQ packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_GCR_REQ_HEADER_op_offset 0
+#define SDMA_PKT_GCR_REQ_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_GCR_REQ_HEADER_op_shift  0
+#define SDMA_PKT_GCR_REQ_HEADER_OP(x) (((x) & SDMA_PKT_GCR_REQ_HEADER_op_mask) << SDMA_PKT_GCR_REQ_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_GCR_REQ_HEADER_sub_op_offset 0
+#define SDMA_PKT_GCR_REQ_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_GCR_REQ_HEADER_sub_op_shift  8
+#define SDMA_PKT_GCR_REQ_HEADER_SUB_OP(x) (((x) & SDMA_PKT_GCR_REQ_HEADER_sub_op_mask) << SDMA_PKT_GCR_REQ_HEADER_sub_op_shift)
+
+/*define for PAYLOAD1 word*/
+/*define for base_va_31_7 field*/
+#define SDMA_PKT_GCR_REQ_PAYLOAD1_base_va_31_7_offset 1
+#define SDMA_PKT_GCR_REQ_PAYLOAD1_base_va_31_7_mask   0x01FFFFFF
+#define SDMA_PKT_GCR_REQ_PAYLOAD1_base_va_31_7_shift  7
+#define SDMA_PKT_GCR_REQ_PAYLOAD1_BASE_VA_31_7(x) (((x) & SDMA_PKT_GCR_REQ_PAYLOAD1_base_va_31_7_mask) << SDMA_PKT_GCR_REQ_PAYLOAD1_base_va_31_7_shift)
+
+/*define for PAYLOAD2 word*/
+/*define for base_va_47_32 field*/
+#define SDMA_PKT_GCR_REQ_PAYLOAD2_base_va_47_32_offset 2
+#define SDMA_PKT_GCR_REQ_PAYLOAD2_base_va_47_32_mask   0x0000FFFF
+#define SDMA_PKT_GCR_REQ_PAYLOAD2_base_va_47_32_shift  0
+#define SDMA_PKT_GCR_REQ_PAYLOAD2_BASE_VA_47_32(x) (((x) & SDMA_PKT_GCR_REQ_PAYLOAD2_base_va_47_32_mask) << SDMA_PKT_GCR_REQ_PAYLOAD2_base_va_47_32_shift)
+
+/*define for gcr_control_15_0 field*/
+#define SDMA_PKT_GCR_REQ_PAYLOAD2_gcr_control_15_0_offset 2
+#define SDMA_PKT_GCR_REQ_PAYLOAD2_gcr_control_15_0_mask   0x0000FFFF
+#define SDMA_PKT_GCR_REQ_PAYLOAD2_gcr_control_15_0_shift  16
+#define SDMA_PKT_GCR_REQ_PAYLOAD2_GCR_CONTROL_15_0(x) (((x) & SDMA_PKT_GCR_REQ_PAYLOAD2_gcr_control_15_0_mask) << SDMA_PKT_GCR_REQ_PAYLOAD2_gcr_control_15_0_shift)
+
+/*define for PAYLOAD3 word*/
+/*define for gcr_control_18_16 field*/
+#define SDMA_PKT_GCR_REQ_PAYLOAD3_gcr_control_18_16_offset 3
+#define SDMA_PKT_GCR_REQ_PAYLOAD3_gcr_control_18_16_mask   0x00000007
+#define SDMA_PKT_GCR_REQ_PAYLOAD3_gcr_control_18_16_shift  0
+#define SDMA_PKT_GCR_REQ_PAYLOAD3_GCR_CONTROL_18_16(x) (((x) & SDMA_PKT_GCR_REQ_PAYLOAD3_gcr_control_18_16_mask) << SDMA_PKT_GCR_REQ_PAYLOAD3_gcr_control_18_16_shift)
+
+/*define for limit_va_31_7 field*/
+#define SDMA_PKT_GCR_REQ_PAYLOAD3_limit_va_31_7_offset 3
+#define SDMA_PKT_GCR_REQ_PAYLOAD3_limit_va_31_7_mask   0x01FFFFFF
+#define SDMA_PKT_GCR_REQ_PAYLOAD3_limit_va_31_7_shift  7
+#define SDMA_PKT_GCR_REQ_PAYLOAD3_LIMIT_VA_31_7(x) (((x) & SDMA_PKT_GCR_REQ_PAYLOAD3_limit_va_31_7_mask) << SDMA_PKT_GCR_REQ_PAYLOAD3_limit_va_31_7_shift)
+
+/*define for PAYLOAD4 word*/
+/*define for limit_va_47_32 field*/
+#define SDMA_PKT_GCR_REQ_PAYLOAD4_limit_va_47_32_offset 4
+#define SDMA_PKT_GCR_REQ_PAYLOAD4_limit_va_47_32_mask   0x0000FFFF
+#define SDMA_PKT_GCR_REQ_PAYLOAD4_limit_va_47_32_shift  0
+#define SDMA_PKT_GCR_REQ_PAYLOAD4_LIMIT_VA_47_32(x) (((x) & SDMA_PKT_GCR_REQ_PAYLOAD4_limit_va_47_32_mask) << SDMA_PKT_GCR_REQ_PAYLOAD4_limit_va_47_32_shift)
+
+/*define for vmid field*/
+#define SDMA_PKT_GCR_REQ_PAYLOAD4_vmid_offset 4
+#define SDMA_PKT_GCR_REQ_PAYLOAD4_vmid_mask   0x0000000F
+#define SDMA_PKT_GCR_REQ_PAYLOAD4_vmid_shift  24
+#define SDMA_PKT_GCR_REQ_PAYLOAD4_VMID(x) (((x) & SDMA_PKT_GCR_REQ_PAYLOAD4_vmid_mask) << SDMA_PKT_GCR_REQ_PAYLOAD4_vmid_shift)
+
+
+/*
+** Definitions for SDMA_PKT_NOP packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_NOP_HEADER_op_offset 0
+#define SDMA_PKT_NOP_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_NOP_HEADER_op_shift  0
+#define SDMA_PKT_NOP_HEADER_OP(x) (((x) & SDMA_PKT_NOP_HEADER_op_mask) << SDMA_PKT_NOP_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_NOP_HEADER_sub_op_offset 0
+#define SDMA_PKT_NOP_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_NOP_HEADER_sub_op_shift  8
+#define SDMA_PKT_NOP_HEADER_SUB_OP(x) (((x) & SDMA_PKT_NOP_HEADER_sub_op_mask) << SDMA_PKT_NOP_HEADER_sub_op_shift)
+
+/*define for count field*/
+#define SDMA_PKT_NOP_HEADER_count_offset 0
+#define SDMA_PKT_NOP_HEADER_count_mask   0x00003FFF
+#define SDMA_PKT_NOP_HEADER_count_shift  16
+#define SDMA_PKT_NOP_HEADER_COUNT(x) (((x) & SDMA_PKT_NOP_HEADER_count_mask) << SDMA_PKT_NOP_HEADER_count_shift)
+
+/*define for DATA0 word*/
+/*define for data0 field*/
+#define SDMA_PKT_NOP_DATA0_data0_offset 1
+#define SDMA_PKT_NOP_DATA0_data0_mask   0xFFFFFFFF
+#define SDMA_PKT_NOP_DATA0_data0_shift  0
+#define SDMA_PKT_NOP_DATA0_DATA0(x) (((x) & SDMA_PKT_NOP_DATA0_data0_mask) << SDMA_PKT_NOP_DATA0_data0_shift)
+
+
+/*
+** Definitions for SDMA_AQL_PKT_HEADER packet
+*/
+
+/*define for HEADER word*/
+/*define for format field*/
+#define SDMA_AQL_PKT_HEADER_HEADER_format_offset 0
+#define SDMA_AQL_PKT_HEADER_HEADER_format_mask   0x000000FF
+#define SDMA_AQL_PKT_HEADER_HEADER_format_shift  0
+#define SDMA_AQL_PKT_HEADER_HEADER_FORMAT(x) (((x) & SDMA_AQL_PKT_HEADER_HEADER_format_mask) << SDMA_AQL_PKT_HEADER_HEADER_format_shift)
+
+/*define for barrier field*/
+#define SDMA_AQL_PKT_HEADER_HEADER_barrier_offset 0
+#define SDMA_AQL_PKT_HEADER_HEADER_barrier_mask   0x00000001
+#define SDMA_AQL_PKT_HEADER_HEADER_barrier_shift  8
+#define SDMA_AQL_PKT_HEADER_HEADER_BARRIER(x) (((x) & SDMA_AQL_PKT_HEADER_HEADER_barrier_mask) << SDMA_AQL_PKT_HEADER_HEADER_barrier_shift)
+
+/*define for acquire_fence_scope field*/
+#define SDMA_AQL_PKT_HEADER_HEADER_acquire_fence_scope_offset 0
+#define SDMA_AQL_PKT_HEADER_HEADER_acquire_fence_scope_mask   0x00000003
+#define SDMA_AQL_PKT_HEADER_HEADER_acquire_fence_scope_shift  9
+#define SDMA_AQL_PKT_HEADER_HEADER_ACQUIRE_FENCE_SCOPE(x) (((x) & SDMA_AQL_PKT_HEADER_HEADER_acquire_fence_scope_mask) << SDMA_AQL_PKT_HEADER_HEADER_acquire_fence_scope_shift)
+
+/*define for release_fence_scope field*/
+#define SDMA_AQL_PKT_HEADER_HEADER_release_fence_scope_offset 0
+#define SDMA_AQL_PKT_HEADER_HEADER_release_fence_scope_mask   0x00000003
+#define SDMA_AQL_PKT_HEADER_HEADER_release_fence_scope_shift  11
+#define SDMA_AQL_PKT_HEADER_HEADER_RELEASE_FENCE_SCOPE(x) (((x) & SDMA_AQL_PKT_HEADER_HEADER_release_fence_scope_mask) << SDMA_AQL_PKT_HEADER_HEADER_release_fence_scope_shift)
+
+/*define for reserved field*/
+#define SDMA_AQL_PKT_HEADER_HEADER_reserved_offset 0
+#define SDMA_AQL_PKT_HEADER_HEADER_reserved_mask   0x00000007
+#define SDMA_AQL_PKT_HEADER_HEADER_reserved_shift  13
+#define SDMA_AQL_PKT_HEADER_HEADER_RESERVED(x) (((x) & SDMA_AQL_PKT_HEADER_HEADER_reserved_mask) << SDMA_AQL_PKT_HEADER_HEADER_reserved_shift)
+
+/*define for op field*/
+#define SDMA_AQL_PKT_HEADER_HEADER_op_offset 0
+#define SDMA_AQL_PKT_HEADER_HEADER_op_mask   0x0000000F
+#define SDMA_AQL_PKT_HEADER_HEADER_op_shift  16
+#define SDMA_AQL_PKT_HEADER_HEADER_OP(x) (((x) & SDMA_AQL_PKT_HEADER_HEADER_op_mask) << SDMA_AQL_PKT_HEADER_HEADER_op_shift)
+
+/*define for subop field*/
+#define SDMA_AQL_PKT_HEADER_HEADER_subop_offset 0
+#define SDMA_AQL_PKT_HEADER_HEADER_subop_mask   0x00000007
+#define SDMA_AQL_PKT_HEADER_HEADER_subop_shift  20
+#define SDMA_AQL_PKT_HEADER_HEADER_SUBOP(x) (((x) & SDMA_AQL_PKT_HEADER_HEADER_subop_mask) << SDMA_AQL_PKT_HEADER_HEADER_subop_shift)
+
+
+/*
+** Definitions for SDMA_AQL_PKT_COPY_LINEAR packet
+*/
+
+/*define for HEADER word*/
+/*define for format field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_format_offset 0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_format_mask   0x000000FF
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_format_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_FORMAT(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_HEADER_format_mask) << SDMA_AQL_PKT_COPY_LINEAR_HEADER_format_shift)
+
+/*define for barrier field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_barrier_offset 0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_barrier_mask   0x00000001
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_barrier_shift  8
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_BARRIER(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_HEADER_barrier_mask) << SDMA_AQL_PKT_COPY_LINEAR_HEADER_barrier_shift)
+
+/*define for acquire_fence_scope field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_acquire_fence_scope_offset 0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_acquire_fence_scope_mask   0x00000003
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_acquire_fence_scope_shift  9
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_ACQUIRE_FENCE_SCOPE(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_HEADER_acquire_fence_scope_mask) << SDMA_AQL_PKT_COPY_LINEAR_HEADER_acquire_fence_scope_shift)
+
+/*define for release_fence_scope field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_release_fence_scope_offset 0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_release_fence_scope_mask   0x00000003
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_release_fence_scope_shift  11
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_RELEASE_FENCE_SCOPE(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_HEADER_release_fence_scope_mask) << SDMA_AQL_PKT_COPY_LINEAR_HEADER_release_fence_scope_shift)
+
+/*define for reserved field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_reserved_offset 0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_reserved_mask   0x00000007
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_reserved_shift  13
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_RESERVED(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_HEADER_reserved_mask) << SDMA_AQL_PKT_COPY_LINEAR_HEADER_reserved_shift)
+
+/*define for op field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_op_offset 0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_op_mask   0x0000000F
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_op_shift  16
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_OP(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_HEADER_op_mask) << SDMA_AQL_PKT_COPY_LINEAR_HEADER_op_shift)
+
+/*define for subop field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_subop_offset 0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_subop_mask   0x00000007
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_subop_shift  20
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_SUBOP(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_HEADER_subop_mask) << SDMA_AQL_PKT_COPY_LINEAR_HEADER_subop_shift)
+
+/*define for RESERVED_DW1 word*/
+/*define for reserved_dw1 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW1_reserved_dw1_offset 1
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW1_reserved_dw1_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW1_reserved_dw1_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW1_RESERVED_DW1(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW1_reserved_dw1_mask) << SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW1_reserved_dw1_shift)
+
+/*define for RETURN_ADDR_LO word*/
+/*define for return_addr_31_0 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_LO_return_addr_31_0_offset 2
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_LO_return_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_LO_return_addr_31_0_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_LO_RETURN_ADDR_31_0(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_LO_return_addr_31_0_mask) << SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_LO_return_addr_31_0_shift)
+
+/*define for RETURN_ADDR_HI word*/
+/*define for return_addr_63_32 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_HI_return_addr_63_32_offset 3
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_HI_return_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_HI_return_addr_63_32_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_HI_RETURN_ADDR_63_32(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_HI_return_addr_63_32_mask) << SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_HI_return_addr_63_32_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_COUNT_count_offset 4
+#define SDMA_AQL_PKT_COPY_LINEAR_COUNT_count_mask   0x003FFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_COUNT_count_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_COUNT_COUNT(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_COUNT_count_mask) << SDMA_AQL_PKT_COPY_LINEAR_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst_sw field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_dst_sw_offset 5
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_dst_sw_mask   0x00000003
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_dst_sw_shift  16
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_DST_SW(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_dst_sw_mask) << SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_dst_sw_shift)
+
+/*define for src_sw field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_src_sw_offset 5
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_src_sw_shift  24
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_SRC_SW(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_src_sw_mask) << SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_src_sw_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_offset 6
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_offset 7
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_offset 8
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_offset 9
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for RESERVED_DW10 word*/
+/*define for reserved_dw10 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW10_reserved_dw10_offset 10
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW10_reserved_dw10_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW10_reserved_dw10_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW10_RESERVED_DW10(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW10_reserved_dw10_mask) << SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW10_reserved_dw10_shift)
+
+/*define for RESERVED_DW11 word*/
+/*define for reserved_dw11 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW11_reserved_dw11_offset 11
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW11_reserved_dw11_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW11_reserved_dw11_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW11_RESERVED_DW11(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW11_reserved_dw11_mask) << SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW11_reserved_dw11_shift)
+
+/*define for RESERVED_DW12 word*/
+/*define for reserved_dw12 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW12_reserved_dw12_offset 12
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW12_reserved_dw12_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW12_reserved_dw12_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW12_RESERVED_DW12(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW12_reserved_dw12_mask) << SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW12_reserved_dw12_shift)
+
+/*define for RESERVED_DW13 word*/
+/*define for reserved_dw13 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW13_reserved_dw13_offset 13
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW13_reserved_dw13_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW13_reserved_dw13_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW13_RESERVED_DW13(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW13_reserved_dw13_mask) << SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW13_reserved_dw13_shift)
+
+/*define for COMPLETION_SIGNAL_LO word*/
+/*define for completion_signal_31_0 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_LO_completion_signal_31_0_offset 14
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_LO_completion_signal_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_LO_completion_signal_31_0_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_LO_COMPLETION_SIGNAL_31_0(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_LO_completion_signal_31_0_mask) << SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_LO_completion_signal_31_0_shift)
+
+/*define for COMPLETION_SIGNAL_HI word*/
+/*define for completion_signal_63_32 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_HI_completion_signal_63_32_offset 15
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_HI_completion_signal_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_HI_completion_signal_63_32_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_HI_COMPLETION_SIGNAL_63_32(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_HI_completion_signal_63_32_mask) << SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_HI_completion_signal_63_32_shift)
+
+
+/*
+** Definitions for SDMA_AQL_PKT_BARRIER_OR packet
+*/
+
+/*define for HEADER word*/
+/*define for format field*/
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_format_offset 0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_format_mask   0x000000FF
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_format_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_FORMAT(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_HEADER_format_mask) << SDMA_AQL_PKT_BARRIER_OR_HEADER_format_shift)
+
+/*define for barrier field*/
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_barrier_offset 0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_barrier_mask   0x00000001
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_barrier_shift  8
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_BARRIER(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_HEADER_barrier_mask) << SDMA_AQL_PKT_BARRIER_OR_HEADER_barrier_shift)
+
+/*define for acquire_fence_scope field*/
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_acquire_fence_scope_offset 0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_acquire_fence_scope_mask   0x00000003
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_acquire_fence_scope_shift  9
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_ACQUIRE_FENCE_SCOPE(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_HEADER_acquire_fence_scope_mask) << SDMA_AQL_PKT_BARRIER_OR_HEADER_acquire_fence_scope_shift)
+
+/*define for release_fence_scope field*/
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_release_fence_scope_offset 0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_release_fence_scope_mask   0x00000003
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_release_fence_scope_shift  11
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_RELEASE_FENCE_SCOPE(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_HEADER_release_fence_scope_mask) << SDMA_AQL_PKT_BARRIER_OR_HEADER_release_fence_scope_shift)
+
+/*define for reserved field*/
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_reserved_offset 0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_reserved_mask   0x00000007
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_reserved_shift  13
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_RESERVED(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_HEADER_reserved_mask) << SDMA_AQL_PKT_BARRIER_OR_HEADER_reserved_shift)
+
+/*define for op field*/
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_op_offset 0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_op_mask   0x0000000F
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_op_shift  16
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_OP(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_HEADER_op_mask) << SDMA_AQL_PKT_BARRIER_OR_HEADER_op_shift)
+
+/*define for subop field*/
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_subop_offset 0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_subop_mask   0x00000007
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_subop_shift  20
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_SUBOP(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_HEADER_subop_mask) << SDMA_AQL_PKT_BARRIER_OR_HEADER_subop_shift)
+
+/*define for RESERVED_DW1 word*/
+/*define for reserved_dw1 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW1_reserved_dw1_offset 1
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW1_reserved_dw1_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW1_reserved_dw1_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW1_RESERVED_DW1(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW1_reserved_dw1_mask) << SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW1_reserved_dw1_shift)
+
+/*define for DEPENDENT_ADDR_0_LO word*/
+/*define for dependent_addr_0_31_0 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_LO_dependent_addr_0_31_0_offset 2
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_LO_dependent_addr_0_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_LO_dependent_addr_0_31_0_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_LO_DEPENDENT_ADDR_0_31_0(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_LO_dependent_addr_0_31_0_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_LO_dependent_addr_0_31_0_shift)
+
+/*define for DEPENDENT_ADDR_0_HI word*/
+/*define for dependent_addr_0_63_32 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_HI_dependent_addr_0_63_32_offset 3
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_HI_dependent_addr_0_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_HI_dependent_addr_0_63_32_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_HI_DEPENDENT_ADDR_0_63_32(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_HI_dependent_addr_0_63_32_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_HI_dependent_addr_0_63_32_shift)
+
+/*define for DEPENDENT_ADDR_1_LO word*/
+/*define for dependent_addr_1_31_0 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_LO_dependent_addr_1_31_0_offset 4
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_LO_dependent_addr_1_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_LO_dependent_addr_1_31_0_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_LO_DEPENDENT_ADDR_1_31_0(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_LO_dependent_addr_1_31_0_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_LO_dependent_addr_1_31_0_shift)
+
+/*define for DEPENDENT_ADDR_1_HI word*/
+/*define for dependent_addr_1_63_32 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_HI_dependent_addr_1_63_32_offset 5
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_HI_dependent_addr_1_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_HI_dependent_addr_1_63_32_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_HI_DEPENDENT_ADDR_1_63_32(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_HI_dependent_addr_1_63_32_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_HI_dependent_addr_1_63_32_shift)
+
+/*define for DEPENDENT_ADDR_2_LO word*/
+/*define for dependent_addr_2_31_0 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_LO_dependent_addr_2_31_0_offset 6
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_LO_dependent_addr_2_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_LO_dependent_addr_2_31_0_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_LO_DEPENDENT_ADDR_2_31_0(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_LO_dependent_addr_2_31_0_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_LO_dependent_addr_2_31_0_shift)
+
+/*define for DEPENDENT_ADDR_2_HI word*/
+/*define for dependent_addr_2_63_32 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_HI_dependent_addr_2_63_32_offset 7
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_HI_dependent_addr_2_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_HI_dependent_addr_2_63_32_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_HI_DEPENDENT_ADDR_2_63_32(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_HI_dependent_addr_2_63_32_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_HI_dependent_addr_2_63_32_shift)
+
+/*define for DEPENDENT_ADDR_3_LO word*/
+/*define for dependent_addr_3_31_0 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_LO_dependent_addr_3_31_0_offset 8
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_LO_dependent_addr_3_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_LO_dependent_addr_3_31_0_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_LO_DEPENDENT_ADDR_3_31_0(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_LO_dependent_addr_3_31_0_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_LO_dependent_addr_3_31_0_shift)
+
+/*define for DEPENDENT_ADDR_3_HI word*/
+/*define for dependent_addr_3_63_32 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_HI_dependent_addr_3_63_32_offset 9
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_HI_dependent_addr_3_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_HI_dependent_addr_3_63_32_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_HI_DEPENDENT_ADDR_3_63_32(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_HI_dependent_addr_3_63_32_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_HI_dependent_addr_3_63_32_shift)
+
+/*define for DEPENDENT_ADDR_4_LO word*/
+/*define for dependent_addr_4_31_0 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_LO_dependent_addr_4_31_0_offset 10
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_LO_dependent_addr_4_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_LO_dependent_addr_4_31_0_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_LO_DEPENDENT_ADDR_4_31_0(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_LO_dependent_addr_4_31_0_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_LO_dependent_addr_4_31_0_shift)
+
+/*define for DEPENDENT_ADDR_4_HI word*/
+/*define for dependent_addr_4_63_32 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_HI_dependent_addr_4_63_32_offset 11
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_HI_dependent_addr_4_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_HI_dependent_addr_4_63_32_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_HI_DEPENDENT_ADDR_4_63_32(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_HI_dependent_addr_4_63_32_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_HI_dependent_addr_4_63_32_shift)
+
+/*define for RESERVED_DW12 word*/
+/*define for reserved_dw12 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW12_reserved_dw12_offset 12
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW12_reserved_dw12_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW12_reserved_dw12_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW12_RESERVED_DW12(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW12_reserved_dw12_mask) << SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW12_reserved_dw12_shift)
+
+/*define for RESERVED_DW13 word*/
+/*define for reserved_dw13 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW13_reserved_dw13_offset 13
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW13_reserved_dw13_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW13_reserved_dw13_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW13_RESERVED_DW13(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW13_reserved_dw13_mask) << SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW13_reserved_dw13_shift)
+
+/*define for COMPLETION_SIGNAL_LO word*/
+/*define for completion_signal_31_0 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_LO_completion_signal_31_0_offset 14
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_LO_completion_signal_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_LO_completion_signal_31_0_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_LO_COMPLETION_SIGNAL_31_0(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_LO_completion_signal_31_0_mask) << SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_LO_completion_signal_31_0_shift)
+
+/*define for COMPLETION_SIGNAL_HI word*/
+/*define for completion_signal_63_32 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_HI_completion_signal_63_32_offset 15
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_HI_completion_signal_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_HI_completion_signal_63_32_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_HI_COMPLETION_SIGNAL_63_32(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_HI_completion_signal_63_32_mask) << SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_HI_completion_signal_63_32_shift)
+
+
+#endif /* __NAVI10_SDMA_PKT_OPEN_H_ */
diff --git a/drivers/gpu/drm/amd/amdgpu/nbif_v6_3_1.c b/drivers/gpu/drm/amd/amdgpu/nbif_v6_3_1.c
new file mode 100644
index 000000000000..9b4025c39e44
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/nbif_v6_3_1.c
@@ -0,0 +1,554 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "nbif_v6_3_1.h"
+
+#include "nbif/nbif_6_3_1_offset.h"
+#include "nbif/nbif_6_3_1_sh_mask.h"
+#include "pcie/pcie_6_1_0_offset.h"
+#include "pcie/pcie_6_1_0_sh_mask.h"
+#include "ivsrcid/nbio/irqsrcs_nbif_7_4.h"
+#include <uapi/linux/kfd_ioctl.h>
+
+static void nbif_v6_3_1_remap_hdp_registers(struct amdgpu_device *adev)
+{
+	WREG32_SOC15(NBIO, 0, regBIF_BX0_REMAP_HDP_MEM_FLUSH_CNTL,
+		adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL);
+	WREG32_SOC15(NBIO, 0, regBIF_BX0_REMAP_HDP_REG_FLUSH_CNTL,
+		adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_REG_FLUSH_CNTL);
+}
+
+static u32 nbif_v6_3_1_get_rev_id(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_DEV0_EPF0_STRAP0);
+
+	tmp &= RCC_STRAP0_RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0_MASK;
+	tmp >>= RCC_STRAP0_RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0__SHIFT;
+
+	return tmp;
+}
+
+static void nbif_v6_3_1_mc_access_enable(struct amdgpu_device *adev, bool enable)
+{
+	if (enable)
+		WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_FB_EN,
+			     BIF_BX0_BIF_FB_EN__FB_READ_EN_MASK |
+			     BIF_BX0_BIF_FB_EN__FB_WRITE_EN_MASK);
+	else
+		WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_FB_EN, 0);
+}
+
+static u32 nbif_v6_3_1_get_memsize(struct amdgpu_device *adev)
+{
+	return RREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF0_RCC_CONFIG_MEMSIZE);
+}
+
+static void nbif_v6_3_1_sdma_doorbell_range(struct amdgpu_device *adev,
+					    int instance, bool use_doorbell,
+					    int doorbell_index,
+					    int doorbell_size)
+{
+	if (instance == 0) {
+		u32 doorbell_range = RREG32_SOC15(NBIO, 0, regGDC_S2A0_S2A_DOORBELL_ENTRY_2_CTRL);
+
+		if (use_doorbell) {
+			doorbell_range = REG_SET_FIELD(doorbell_range,
+						       GDC_S2A0_S2A_DOORBELL_ENTRY_2_CTRL,
+						       S2A_DOORBELL_PORT2_ENABLE,
+						       0x1);
+			doorbell_range = REG_SET_FIELD(doorbell_range,
+						       GDC_S2A0_S2A_DOORBELL_ENTRY_2_CTRL,
+						       S2A_DOORBELL_PORT2_AWID,
+						       0xe);
+			doorbell_range = REG_SET_FIELD(doorbell_range,
+						       GDC_S2A0_S2A_DOORBELL_ENTRY_2_CTRL,
+						       S2A_DOORBELL_PORT2_RANGE_OFFSET,
+						       doorbell_index);
+			doorbell_range = REG_SET_FIELD(doorbell_range,
+						       GDC_S2A0_S2A_DOORBELL_ENTRY_2_CTRL,
+						       S2A_DOORBELL_PORT2_RANGE_SIZE,
+						       doorbell_size);
+			doorbell_range = REG_SET_FIELD(doorbell_range,
+						       GDC_S2A0_S2A_DOORBELL_ENTRY_2_CTRL,
+						       S2A_DOORBELL_PORT2_AWADDR_31_28_VALUE,
+						       0x3);
+		} else
+			doorbell_range = REG_SET_FIELD(doorbell_range,
+						       GDC_S2A0_S2A_DOORBELL_ENTRY_2_CTRL,
+						       S2A_DOORBELL_PORT2_RANGE_SIZE,
+						       0);
+
+		WREG32_SOC15(NBIO, 0, regGDC_S2A0_S2A_DOORBELL_ENTRY_2_CTRL, doorbell_range);
+	}
+}
+
+static void nbif_v6_3_1_vcn_doorbell_range(struct amdgpu_device *adev,
+					   bool use_doorbell, int doorbell_index,
+					   int instance)
+{
+	u32 doorbell_range;
+
+	if (instance)
+		doorbell_range = RREG32_SOC15(NBIO, 0, regGDC_S2A0_S2A_DOORBELL_ENTRY_5_CTRL);
+	else
+		doorbell_range = RREG32_SOC15(NBIO, 0, regGDC_S2A0_S2A_DOORBELL_ENTRY_4_CTRL);
+
+	if (use_doorbell) {
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC_S2A0_S2A_DOORBELL_ENTRY_4_CTRL,
+					       S2A_DOORBELL_PORT4_ENABLE,
+					       0x1);
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC_S2A0_S2A_DOORBELL_ENTRY_4_CTRL,
+					       S2A_DOORBELL_PORT4_AWID,
+					       instance ? 0x7 : 0x4);
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC_S2A0_S2A_DOORBELL_ENTRY_4_CTRL,
+					       S2A_DOORBELL_PORT4_RANGE_OFFSET,
+					       doorbell_index);
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC_S2A0_S2A_DOORBELL_ENTRY_4_CTRL,
+					       S2A_DOORBELL_PORT4_RANGE_SIZE,
+					       8);
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC_S2A0_S2A_DOORBELL_ENTRY_4_CTRL,
+					       S2A_DOORBELL_PORT4_AWADDR_31_28_VALUE,
+					       instance ? 0x7 : 0x4);
+	} else
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC_S2A0_S2A_DOORBELL_ENTRY_4_CTRL,
+					       S2A_DOORBELL_PORT4_RANGE_SIZE,
+					       0);
+
+	if (instance)
+		WREG32_SOC15(NBIO, 0, regGDC_S2A0_S2A_DOORBELL_ENTRY_5_CTRL, doorbell_range);
+	else
+		WREG32_SOC15(NBIO, 0, regGDC_S2A0_S2A_DOORBELL_ENTRY_4_CTRL, doorbell_range);
+}
+
+static void nbif_v6_3_1_gc_doorbell_init(struct amdgpu_device *adev)
+{
+	WREG32_SOC15(NBIO, 0, regGDC_S2A0_S2A_DOORBELL_ENTRY_0_CTRL, 0x30000007);
+	WREG32_SOC15(NBIO, 0, regGDC_S2A0_S2A_DOORBELL_ENTRY_3_CTRL, 0x3000000d);
+}
+
+static void nbif_v6_3_1_enable_doorbell_aperture(struct amdgpu_device *adev,
+						 bool enable)
+{
+	WREG32_FIELD15_PREREG(NBIO, 0, RCC_DEV0_EPF0_RCC_DOORBELL_APER_EN,
+			BIF_DOORBELL_APER_EN, enable ? 1 : 0);
+}
+
+static void
+nbif_v6_3_1_enable_doorbell_selfring_aperture(struct amdgpu_device *adev,
+					      bool enable)
+{
+	u32 tmp = 0;
+
+	if (enable) {
+		tmp = REG_SET_FIELD(tmp, BIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL,
+				    DOORBELL_SELFRING_GPA_APER_EN, 1) |
+		      REG_SET_FIELD(tmp, BIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL,
+				    DOORBELL_SELFRING_GPA_APER_MODE, 1) |
+		      REG_SET_FIELD(tmp, BIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL,
+				    DOORBELL_SELFRING_GPA_APER_SIZE, 0);
+
+		WREG32_SOC15(NBIO, 0, regBIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_BASE_LOW,
+			     lower_32_bits(adev->doorbell.base));
+		WREG32_SOC15(NBIO, 0, regBIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_BASE_HIGH,
+			     upper_32_bits(adev->doorbell.base));
+	}
+
+	WREG32_SOC15(NBIO, 0, regBIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL, tmp);
+}
+
+static void nbif_v6_3_1_ih_doorbell_range(struct amdgpu_device *adev,
+					  bool use_doorbell, int doorbell_index)
+{
+	u32 ih_doorbell_range = RREG32_SOC15(NBIO, 0, regGDC_S2A0_S2A_DOORBELL_ENTRY_1_CTRL);
+
+	if (use_doorbell) {
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  GDC_S2A0_S2A_DOORBELL_ENTRY_1_CTRL,
+						  S2A_DOORBELL_PORT1_ENABLE,
+						  0x1);
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  GDC_S2A0_S2A_DOORBELL_ENTRY_1_CTRL,
+						  S2A_DOORBELL_PORT1_AWID,
+						  0x0);
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  GDC_S2A0_S2A_DOORBELL_ENTRY_1_CTRL,
+						  S2A_DOORBELL_PORT1_RANGE_OFFSET,
+						  doorbell_index);
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  GDC_S2A0_S2A_DOORBELL_ENTRY_1_CTRL,
+						  S2A_DOORBELL_PORT1_RANGE_SIZE,
+						  2);
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  GDC_S2A0_S2A_DOORBELL_ENTRY_1_CTRL,
+						  S2A_DOORBELL_PORT1_AWADDR_31_28_VALUE,
+						  0x0);
+	} else
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  GDC_S2A0_S2A_DOORBELL_ENTRY_1_CTRL,
+						  S2A_DOORBELL_PORT1_RANGE_SIZE,
+						  0);
+
+	WREG32_SOC15(NBIO, 0, regGDC_S2A0_S2A_DOORBELL_ENTRY_1_CTRL, ih_doorbell_range);
+}
+
+static void nbif_v6_3_1_ih_control(struct amdgpu_device *adev)
+{
+	u32 interrupt_cntl;
+
+	/* setup interrupt control */
+	WREG32_SOC15(NBIO, 0, regBIF_BX0_INTERRUPT_CNTL2, adev->dummy_page_addr >> 8);
+
+	interrupt_cntl = RREG32_SOC15(NBIO, 0, regBIF_BX0_INTERRUPT_CNTL);
+	/*
+	 * BIF_BX0_INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
+	 * BIF_BX0_INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
+	 */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, BIF_BX0_INTERRUPT_CNTL,
+				       IH_DUMMY_RD_OVERRIDE, 0);
+
+	/* BIF_BX0_INTERRUPT_CNTL__IH_REQ_NONSNOOP_EN_MASK=1 if ring is in non-cacheable memory, e.g., vram */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, BIF_BX0_INTERRUPT_CNTL,
+				       IH_REQ_NONSNOOP_EN, 0);
+
+	WREG32_SOC15(NBIO, 0, regBIF_BX0_INTERRUPT_CNTL, interrupt_cntl);
+}
+
+static void
+nbif_v6_3_1_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+					     bool enable)
+{
+}
+
+static void
+nbif_v6_3_1_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+					    bool enable)
+{
+}
+
+static void
+nbif_v6_3_1_get_clockgating_state(struct amdgpu_device *adev,
+				  u64 *flags)
+{
+}
+
+static u32 nbif_v6_3_1_get_hdp_flush_req_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF0_GPU_HDP_FLUSH_REQ);
+}
+
+static u32 nbif_v6_3_1_get_hdp_flush_done_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF0_GPU_HDP_FLUSH_DONE);
+}
+
+static u32 nbif_v6_3_1_get_pcie_index_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF0_RSMU_INDEX);
+}
+
+static u32 nbif_v6_3_1_get_pcie_data_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF0_RSMU_DATA);
+}
+
+const struct nbio_hdp_flush_reg nbif_v6_3_1_hdp_flush_reg = {
+	.ref_and_mask_cp0 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP0_MASK,
+	.ref_and_mask_cp1 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP1_MASK,
+	.ref_and_mask_cp2 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP2_MASK,
+	.ref_and_mask_cp3 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP3_MASK,
+	.ref_and_mask_cp4 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP4_MASK,
+	.ref_and_mask_cp5 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP5_MASK,
+	.ref_and_mask_cp6 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP6_MASK,
+	.ref_and_mask_cp7 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP7_MASK,
+	.ref_and_mask_cp8 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP8_MASK,
+	.ref_and_mask_cp9 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP9_MASK,
+	.ref_and_mask_sdma0 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__SDMA0_MASK,
+	.ref_and_mask_sdma1 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__SDMA1_MASK,
+};
+
+static void nbif_v6_3_1_init_registers(struct amdgpu_device *adev)
+{
+	uint32_t data;
+
+	data = RREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF2_STRAP2);
+	data &= ~RCC_DEV0_EPF2_STRAP2__STRAP_NO_SOFT_RESET_DEV0_F2_MASK;
+	WREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF2_STRAP2, data);
+}
+
+static u32 nbif_v6_3_1_get_rom_offset(struct amdgpu_device *adev)
+{
+	u32 data, rom_offset;
+
+	data = RREG32_SOC15(NBIO, 0, regREGS_ROM_OFFSET_CTRL);
+	rom_offset = REG_GET_FIELD(data, REGS_ROM_OFFSET_CTRL, ROM_OFFSET);
+
+	return rom_offset;
+}
+
+#ifdef CONFIG_PCIEASPM
+static void nbif_v6_3_1_program_ltr(struct amdgpu_device *adev)
+{
+	uint32_t def, data;
+	u16 devctl2;
+
+	def = RREG32_SOC15(NBIO, 0, regRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL);
+	data = 0x35EB;
+	data &= ~RCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL__LTR_PRIV_MSG_DIS_IN_PM_NON_D0_MASK;
+	data &= ~RCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL__LTR_PRIV_RST_LTR_IN_DL_DOWN_MASK;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL, data);
+
+	def = data = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_BIF_STRAP2);
+	data &= ~RCC_STRAP0_RCC_BIF_STRAP2__STRAP_LTR_IN_ASPML1_DIS_MASK;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_BIF_STRAP2, data);
+
+	pcie_capability_read_word(adev->pdev, PCI_EXP_DEVCTL2, &devctl2);
+
+	if (adev->pdev->ltr_path == (devctl2 & PCI_EXP_DEVCTL2_LTR_EN))
+		return;
+
+	if (adev->pdev->ltr_path)
+		pcie_capability_set_word(adev->pdev, PCI_EXP_DEVCTL2, PCI_EXP_DEVCTL2_LTR_EN);
+	else
+		pcie_capability_clear_word(adev->pdev, PCI_EXP_DEVCTL2, PCI_EXP_DEVCTL2_LTR_EN);
+}
+#endif
+
+static void nbif_v6_3_1_program_aspm(struct amdgpu_device *adev)
+{
+#ifdef CONFIG_PCIEASPM
+	uint32_t def, data;
+	u16 devctl2, ltr;
+
+	def = data = RREG32_SOC15(PCIE, 0, regPCIE_LC_CNTL);
+	data &= ~PCIE_LC_CNTL__LC_L1_INACTIVITY_MASK;
+	data &= ~PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK;
+	data |= PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK;
+	if (def != data)
+		WREG32_SOC15(PCIE, 0, regPCIE_LC_CNTL, data);
+
+	def = data = RREG32_SOC15(PCIE, 0, regPCIE_LC_CNTL7);
+	data |= PCIE_LC_CNTL7__LC_NBIF_ASPM_INPUT_EN_MASK;
+	if (def != data)
+		WREG32_SOC15(PCIE, 0, regPCIE_LC_CNTL7, data);
+
+	def = data = RREG32_SOC15(PCIE, 0, regPCIE_LC_CNTL3);
+	data |= PCIE_LC_CNTL3__LC_DSC_DONT_ENTER_L23_AFTER_PME_ACK_MASK;
+	if (def != data)
+		WREG32_SOC15(PCIE, 0, regPCIE_LC_CNTL3, data);
+
+	def = data = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_BIF_STRAP3);
+	data &= ~RCC_STRAP0_RCC_BIF_STRAP3__STRAP_VLINK_ASPM_IDLE_TIMER_MASK;
+	data &= ~RCC_STRAP0_RCC_BIF_STRAP3__STRAP_VLINK_PM_L1_ENTRY_TIMER_MASK;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_BIF_STRAP3, data);
+
+	def = data = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_BIF_STRAP5);
+	data &= ~RCC_STRAP0_RCC_BIF_STRAP5__STRAP_VLINK_LDN_ENTRY_TIMER_MASK;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_BIF_STRAP5, data);
+
+	pcie_capability_read_word(adev->pdev, PCI_EXP_DEVCTL2, &devctl2);
+	data = def = devctl2;
+	data &= ~PCI_EXP_DEVCTL2_LTR_EN;
+	if (def != data)
+		pcie_capability_set_word(adev->pdev, PCI_EXP_DEVCTL2, (u16)data);
+
+	ltr = pci_find_ext_capability(adev->pdev, PCI_EXT_CAP_ID_LTR);
+
+	if (ltr) {
+		pci_write_config_dword(adev->pdev, ltr + PCI_LTR_MAX_SNOOP_LAT, 0x10011001);
+	}
+
+#if 0
+	/* regPSWUSP0_PCIE_LC_CNTL2 should be replace by PCIE_LC_CNTL2 or someone else ? */
+	def = data = RREG32_SOC15(NBIO, 0, regPSWUSP0_PCIE_LC_CNTL2);
+	data |= PSWUSP0_PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L1_MASK |
+		PSWUSP0_PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L23_MASK;
+	data &= ~PSWUSP0_PCIE_LC_CNTL2__LC_RCV_L0_TO_RCV_L0S_DIS_MASK;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regPSWUSP0_PCIE_LC_CNTL2, data);
+#endif
+	def = data = RREG32_SOC15(PCIE, 0, regPCIE_LC_CNTL4);
+	data |= PCIE_LC_CNTL4__LC_L1_POWERDOWN_MASK;
+	if (def != data)
+		WREG32_SOC15(PCIE, 0, regPCIE_LC_CNTL4, data);
+
+	def = data = RREG32_SOC15(PCIE, 0, regPCIE_LC_RXRECOVER_RXSTANDBY_CNTL);
+	data |= PCIE_LC_RXRECOVER_RXSTANDBY_CNTL__LC_RX_L0S_STANDBY_EN_MASK;
+	if (def != data)
+		WREG32_SOC15(PCIE, 0, regPCIE_LC_RXRECOVER_RXSTANDBY_CNTL, data);
+
+	nbif_v6_3_1_program_ltr(adev);
+
+	def = data = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_BIF_STRAP3);
+	data |= 0x5DE0 << RCC_STRAP0_RCC_BIF_STRAP3__STRAP_VLINK_ASPM_IDLE_TIMER__SHIFT;
+	data |= 0x0010 << RCC_STRAP0_RCC_BIF_STRAP3__STRAP_VLINK_PM_L1_ENTRY_TIMER__SHIFT;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_BIF_STRAP3, data);
+
+	def = data = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_BIF_STRAP5);
+	data |= 0x0010 << RCC_STRAP0_RCC_BIF_STRAP5__STRAP_VLINK_LDN_ENTRY_TIMER__SHIFT;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_BIF_STRAP5, data);
+
+	def = data = RREG32_SOC15(PCIE, 0, regPCIE_LC_CNTL);
+	data |= 0x0 << PCIE_LC_CNTL__LC_L0S_INACTIVITY__SHIFT;
+	data |= 0x9 << PCIE_LC_CNTL__LC_L1_INACTIVITY__SHIFT;
+	data &= ~PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK;
+	if (def != data)
+		WREG32_SOC15(PCIE, 0, regPCIE_LC_CNTL, data);
+
+	def = data = RREG32_SOC15(PCIE, 0, regPCIE_LC_CNTL3);
+	data &= ~PCIE_LC_CNTL3__LC_DSC_DONT_ENTER_L23_AFTER_PME_ACK_MASK;
+	if (def != data)
+		WREG32_SOC15(PCIE, 0, regPCIE_LC_CNTL3, data);
+#endif
+}
+
+#define MMIO_REG_HOLE_OFFSET (0x80000 - PAGE_SIZE)
+
+static void nbif_v6_3_1_set_reg_remap(struct amdgpu_device *adev)
+{
+	if (!amdgpu_sriov_vf(adev) && (PAGE_SIZE <= 4096)) {
+		adev->rmmio_remap.reg_offset = MMIO_REG_HOLE_OFFSET;
+		adev->rmmio_remap.bus_addr = adev->rmmio_base + MMIO_REG_HOLE_OFFSET;
+	} else {
+		adev->rmmio_remap.reg_offset = SOC15_REG_OFFSET(NBIO, 0,
+			regBIF_BX_PF0_HDP_MEM_COHERENCY_FLUSH_CNTL) << 2;
+		adev->rmmio_remap.bus_addr = 0;
+	}
+}
+
+const struct amdgpu_nbio_funcs nbif_v6_3_1_funcs = {
+	.get_hdp_flush_req_offset = nbif_v6_3_1_get_hdp_flush_req_offset,
+	.get_hdp_flush_done_offset = nbif_v6_3_1_get_hdp_flush_done_offset,
+	.get_pcie_index_offset = nbif_v6_3_1_get_pcie_index_offset,
+	.get_pcie_data_offset = nbif_v6_3_1_get_pcie_data_offset,
+	.get_rev_id = nbif_v6_3_1_get_rev_id,
+	.mc_access_enable = nbif_v6_3_1_mc_access_enable,
+	.get_memsize = nbif_v6_3_1_get_memsize,
+	.sdma_doorbell_range = nbif_v6_3_1_sdma_doorbell_range,
+	.vcn_doorbell_range = nbif_v6_3_1_vcn_doorbell_range,
+	.gc_doorbell_init = nbif_v6_3_1_gc_doorbell_init,
+	.enable_doorbell_aperture = nbif_v6_3_1_enable_doorbell_aperture,
+	.enable_doorbell_selfring_aperture = nbif_v6_3_1_enable_doorbell_selfring_aperture,
+	.ih_doorbell_range = nbif_v6_3_1_ih_doorbell_range,
+	.update_medium_grain_clock_gating = nbif_v6_3_1_update_medium_grain_clock_gating,
+	.update_medium_grain_light_sleep = nbif_v6_3_1_update_medium_grain_light_sleep,
+	.get_clockgating_state = nbif_v6_3_1_get_clockgating_state,
+	.ih_control = nbif_v6_3_1_ih_control,
+	.init_registers = nbif_v6_3_1_init_registers,
+	.remap_hdp_registers = nbif_v6_3_1_remap_hdp_registers,
+	.get_rom_offset = nbif_v6_3_1_get_rom_offset,
+	.program_aspm = nbif_v6_3_1_program_aspm,
+	.set_reg_remap = nbif_v6_3_1_set_reg_remap,
+};
+
+
+static int nbif_v6_3_1_set_ras_err_event_athub_irq_state(struct amdgpu_device *adev,
+						       struct amdgpu_irq_src *src,
+						       unsigned type,
+						       enum amdgpu_interrupt_state state)
+{
+	/* The ras_controller_irq enablement should be done in psp bl when it
+	 * tries to enable ras feature. Driver only need to set the correct interrupt
+	 * vector for bare-metal and sriov use case respectively
+	 */
+	uint32_t bif_doorbell_int_cntl;
+
+	bif_doorbell_int_cntl = RREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_DOORBELL_INT_CNTL);
+	bif_doorbell_int_cntl = REG_SET_FIELD(bif_doorbell_int_cntl,
+					      BIF_BX0_BIF_DOORBELL_INT_CNTL,
+					      RAS_ATHUB_ERR_EVENT_INTERRUPT_DISABLE,
+					      (state == AMDGPU_IRQ_STATE_ENABLE) ? 0 : 1);
+	WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_DOORBELL_INT_CNTL, bif_doorbell_int_cntl);
+
+	return 0;
+}
+
+static int nbif_v6_3_1_process_err_event_athub_irq(struct amdgpu_device *adev,
+						 struct amdgpu_irq_src *source,
+						 struct amdgpu_iv_entry *entry)
+{
+	/* By design, the ih cookie for err_event_athub_irq should be written
+	 * to bif ring. since bif ring is not enabled, just leave process callback
+	 * as a dummy one.
+	 */
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs nbif_v6_3_1_ras_err_event_athub_irq_funcs = {
+	.set = nbif_v6_3_1_set_ras_err_event_athub_irq_state,
+	.process = nbif_v6_3_1_process_err_event_athub_irq,
+};
+
+static void nbif_v6_3_1_handle_ras_err_event_athub_intr_no_bifring(struct amdgpu_device *adev)
+{
+	uint32_t bif_doorbell_int_cntl;
+
+	bif_doorbell_int_cntl = RREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_DOORBELL_INT_CNTL);
+	if (REG_GET_FIELD(bif_doorbell_int_cntl,
+			  BIF_BX0_BIF_DOORBELL_INT_CNTL,
+			  RAS_ATHUB_ERR_EVENT_INTERRUPT_STATUS)) {
+		/* driver has to clear the interrupt status when bif ring is disabled */
+		bif_doorbell_int_cntl = REG_SET_FIELD(bif_doorbell_int_cntl,
+						BIF_BX0_BIF_DOORBELL_INT_CNTL,
+						RAS_ATHUB_ERR_EVENT_INTERRUPT_CLEAR, 1);
+		WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_DOORBELL_INT_CNTL, bif_doorbell_int_cntl);
+		amdgpu_ras_global_ras_isr(adev);
+	}
+}
+
+static int nbif_v6_3_1_init_ras_err_event_athub_interrupt(struct amdgpu_device *adev)
+{
+	int r;
+
+	/* init the irq funcs */
+	adev->nbio.ras_err_event_athub_irq.funcs =
+		&nbif_v6_3_1_ras_err_event_athub_irq_funcs;
+	adev->nbio.ras_err_event_athub_irq.num_types = 1;
+
+	/* register ras err event athub interrupt
+	 * nbif v6_3_1 uses the same irq source as nbio v7_4
+	 */
+	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_BIF,
+			      NBIF_7_4__SRCID__ERREVENT_ATHUB_INTERRUPT,
+			      &adev->nbio.ras_err_event_athub_irq);
+
+	return r;
+}
+
+struct amdgpu_nbio_ras nbif_v6_3_1_ras = {
+	.handle_ras_err_event_athub_intr_no_bifring =
+		nbif_v6_3_1_handle_ras_err_event_athub_intr_no_bifring,
+	.init_ras_err_event_athub_interrupt =
+		nbif_v6_3_1_init_ras_err_event_athub_interrupt,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/nbif_v6_3_1.h b/drivers/gpu/drm/amd/amdgpu/nbif_v6_3_1.h
new file mode 100644
index 000000000000..3afec715a9fe
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/nbif_v6_3_1.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __NBIO_V6_3_1_H__
+#define __NBIO_V6_3_1_H__
+
+#include "soc15_common.h"
+
+extern const struct nbio_hdp_flush_reg nbif_v6_3_1_hdp_flush_reg;
+extern const struct amdgpu_nbio_funcs nbif_v6_3_1_funcs;
+extern struct amdgpu_nbio_ras nbif_v6_3_1_ras;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v2_3.c b/drivers/gpu/drm/amd/amdgpu/nbio_v2_3.c
new file mode 100644
index 000000000000..04041b398781
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v2_3.c
@@ -0,0 +1,590 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "nbio_v2_3.h"
+
+#include "nbio/nbio_2_3_default.h"
+#include "nbio/nbio_2_3_offset.h"
+#include "nbio/nbio_2_3_sh_mask.h"
+#include <uapi/linux/kfd_ioctl.h>
+#include <linux/device.h>
+#include <linux/pci.h>
+
+#define smnPCIE_CONFIG_CNTL	0x11180044
+#define smnCPM_CONTROL		0x11180460
+#define smnPCIE_CNTL2		0x11180070
+#define smnPCIE_LC_CNTL		0x11140280
+#define smnPCIE_LC_CNTL3	0x111402d4
+#define smnPCIE_LC_CNTL6	0x111402ec
+#define smnPCIE_LC_CNTL7	0x111402f0
+#define smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2	0x1014008c
+#define smnRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL	0x10123538
+#define smnBIF_CFG_DEV0_EPF0_PCIE_LTR_CAP	0x10140324
+#define smnPSWUSP0_PCIE_LC_CNTL2		0x111402c4
+#define smnNBIF_MGCG_CTRL_LCLK			0x1013a21c
+
+#define mmBIF_SDMA2_DOORBELL_RANGE		0x01d6
+#define mmBIF_SDMA2_DOORBELL_RANGE_BASE_IDX	2
+#define mmBIF_SDMA3_DOORBELL_RANGE		0x01d7
+#define mmBIF_SDMA3_DOORBELL_RANGE_BASE_IDX	2
+
+#define mmBIF_MMSCH1_DOORBELL_RANGE		0x01d8
+#define mmBIF_MMSCH1_DOORBELL_RANGE_BASE_IDX	2
+
+#define smnPCIE_LC_LINK_WIDTH_CNTL		0x11140288
+
+#define GPU_HDP_FLUSH_DONE__RSVD_ENG0_MASK	0x00001000L /* Don't use.  Firmware uses this bit internally */
+#define GPU_HDP_FLUSH_DONE__RSVD_ENG1_MASK	0x00002000L
+#define GPU_HDP_FLUSH_DONE__RSVD_ENG2_MASK	0x00004000L
+#define GPU_HDP_FLUSH_DONE__RSVD_ENG3_MASK	0x00008000L
+#define GPU_HDP_FLUSH_DONE__RSVD_ENG4_MASK	0x00010000L
+#define GPU_HDP_FLUSH_DONE__RSVD_ENG5_MASK	0x00020000L
+#define GPU_HDP_FLUSH_DONE__RSVD_ENG6_MASK	0x00040000L
+#define GPU_HDP_FLUSH_DONE__RSVD_ENG7_MASK	0x00080000L
+#define GPU_HDP_FLUSH_DONE__RSVD_ENG8_MASK	0x00100000L
+
+static void nbio_v2_3_remap_hdp_registers(struct amdgpu_device *adev)
+{
+	WREG32_SOC15(NBIO, 0, mmREMAP_HDP_MEM_FLUSH_CNTL,
+		adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL);
+	WREG32_SOC15(NBIO, 0, mmREMAP_HDP_REG_FLUSH_CNTL,
+		adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_REG_FLUSH_CNTL);
+}
+
+static u32 nbio_v2_3_get_rev_id(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	/*
+	 * guest vm gets 0xffffffff when reading RCC_DEV0_EPF0_STRAP0,
+	 * therefore we force rev_id to 0 (which is the default value)
+	 */
+	if (amdgpu_sriov_vf(adev)) {
+		return 0;
+	}
+
+	tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0);
+	tmp &= RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0_MASK;
+	tmp >>= RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0__SHIFT;
+
+	return tmp;
+}
+
+static void nbio_v2_3_mc_access_enable(struct amdgpu_device *adev, bool enable)
+{
+	if (enable)
+		WREG32_SOC15(NBIO, 0, mmBIF_FB_EN,
+			     BIF_FB_EN__FB_READ_EN_MASK |
+			     BIF_FB_EN__FB_WRITE_EN_MASK);
+	else
+		WREG32_SOC15(NBIO, 0, mmBIF_FB_EN, 0);
+}
+
+static u32 nbio_v2_3_get_memsize(struct amdgpu_device *adev)
+{
+	return RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_RCC_CONFIG_MEMSIZE);
+}
+
+static void nbio_v2_3_sdma_doorbell_range(struct amdgpu_device *adev, int instance,
+					  bool use_doorbell, int doorbell_index,
+					  int doorbell_size)
+{
+	u32 reg = instance == 0 ? SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA0_DOORBELL_RANGE) :
+			instance == 1 ? SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA1_DOORBELL_RANGE) :
+			instance == 2 ? SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA2_DOORBELL_RANGE) :
+			SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA3_DOORBELL_RANGE);
+
+	u32 doorbell_range = RREG32(reg);
+
+	if (use_doorbell) {
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       BIF_SDMA0_DOORBELL_RANGE, OFFSET,
+					       doorbell_index);
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       BIF_SDMA0_DOORBELL_RANGE, SIZE,
+					       doorbell_size);
+	} else
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       BIF_SDMA0_DOORBELL_RANGE, SIZE,
+					       0);
+
+	WREG32(reg, doorbell_range);
+}
+
+static void nbio_v2_3_vcn_doorbell_range(struct amdgpu_device *adev, bool use_doorbell,
+					 int doorbell_index, int instance)
+{
+	u32 reg = instance ? SOC15_REG_OFFSET(NBIO, 0, mmBIF_MMSCH1_DOORBELL_RANGE) :
+		SOC15_REG_OFFSET(NBIO, 0, mmBIF_MMSCH0_DOORBELL_RANGE);
+
+	u32 doorbell_range = RREG32(reg);
+
+	if (use_doorbell) {
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       BIF_MMSCH0_DOORBELL_RANGE, OFFSET,
+					       doorbell_index);
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       BIF_MMSCH0_DOORBELL_RANGE, SIZE, 8);
+	} else
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       BIF_MMSCH0_DOORBELL_RANGE, SIZE, 0);
+
+	WREG32(reg, doorbell_range);
+}
+
+static void nbio_v2_3_enable_doorbell_aperture(struct amdgpu_device *adev,
+					       bool enable)
+{
+	WREG32_FIELD15(NBIO, 0, RCC_DEV0_EPF0_RCC_DOORBELL_APER_EN, BIF_DOORBELL_APER_EN,
+		       enable ? 1 : 0);
+}
+
+static void nbio_v2_3_enable_doorbell_selfring_aperture(struct amdgpu_device *adev,
+							bool enable)
+{
+	u32 tmp = 0;
+
+	if (enable) {
+		tmp = REG_SET_FIELD(tmp, BIF_BX_PF_DOORBELL_SELFRING_GPA_APER_CNTL,
+				    DOORBELL_SELFRING_GPA_APER_EN, 1) |
+		      REG_SET_FIELD(tmp, BIF_BX_PF_DOORBELL_SELFRING_GPA_APER_CNTL,
+				    DOORBELL_SELFRING_GPA_APER_MODE, 1) |
+		      REG_SET_FIELD(tmp, BIF_BX_PF_DOORBELL_SELFRING_GPA_APER_CNTL,
+				    DOORBELL_SELFRING_GPA_APER_SIZE, 0);
+
+		WREG32_SOC15(NBIO, 0, mmBIF_BX_PF_DOORBELL_SELFRING_GPA_APER_BASE_LOW,
+			     lower_32_bits(adev->doorbell.base));
+		WREG32_SOC15(NBIO, 0, mmBIF_BX_PF_DOORBELL_SELFRING_GPA_APER_BASE_HIGH,
+			     upper_32_bits(adev->doorbell.base));
+	}
+
+	WREG32_SOC15(NBIO, 0, mmBIF_BX_PF_DOORBELL_SELFRING_GPA_APER_CNTL,
+		     tmp);
+}
+
+
+static void nbio_v2_3_ih_doorbell_range(struct amdgpu_device *adev,
+					bool use_doorbell, int doorbell_index)
+{
+	u32 ih_doorbell_range = RREG32_SOC15(NBIO, 0, mmBIF_IH_DOORBELL_RANGE);
+
+	if (use_doorbell) {
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  BIF_IH_DOORBELL_RANGE, OFFSET,
+						  doorbell_index);
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  BIF_IH_DOORBELL_RANGE, SIZE,
+						  2);
+	} else
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  BIF_IH_DOORBELL_RANGE, SIZE,
+						  0);
+
+	WREG32_SOC15(NBIO, 0, mmBIF_IH_DOORBELL_RANGE, ih_doorbell_range);
+}
+
+static void nbio_v2_3_ih_control(struct amdgpu_device *adev)
+{
+	u32 interrupt_cntl;
+
+	/* setup interrupt control */
+	WREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL2, adev->dummy_page_addr >> 8);
+
+	interrupt_cntl = RREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL);
+	/*
+	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
+	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
+	 */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, INTERRUPT_CNTL,
+				       IH_DUMMY_RD_OVERRIDE, 0);
+
+	/* INTERRUPT_CNTL__IH_REQ_NONSNOOP_EN_MASK=1 if ring is in non-cacheable memory, e.g., vram */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, INTERRUPT_CNTL,
+				       IH_REQ_NONSNOOP_EN, 0);
+
+	WREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL, interrupt_cntl);
+}
+
+static void nbio_v2_3_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						       bool enable)
+{
+	uint32_t def, data;
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_BIF_MGCG))
+		return;
+
+	def = data = RREG32_PCIE(smnCPM_CONTROL);
+	if (enable) {
+		data |= (CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK |
+			 CPM_CONTROL__TXCLK_DYN_GATE_ENABLE_MASK |
+			 CPM_CONTROL__TXCLK_LCNT_GATE_ENABLE_MASK |
+			 CPM_CONTROL__TXCLK_REGS_GATE_ENABLE_MASK |
+			 CPM_CONTROL__TXCLK_PRBS_GATE_ENABLE_MASK |
+			 CPM_CONTROL__REFCLK_REGS_GATE_ENABLE_MASK);
+	} else {
+		data &= ~(CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK |
+			  CPM_CONTROL__TXCLK_DYN_GATE_ENABLE_MASK |
+			  CPM_CONTROL__TXCLK_LCNT_GATE_ENABLE_MASK |
+			  CPM_CONTROL__TXCLK_REGS_GATE_ENABLE_MASK |
+			  CPM_CONTROL__TXCLK_PRBS_GATE_ENABLE_MASK |
+			  CPM_CONTROL__REFCLK_REGS_GATE_ENABLE_MASK);
+	}
+
+	if (def != data)
+		WREG32_PCIE(smnCPM_CONTROL, data);
+}
+
+static void nbio_v2_3_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+						      bool enable)
+{
+	uint32_t def, data;
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_BIF_LS))
+		return;
+
+	def = data = RREG32_PCIE(smnPCIE_CNTL2);
+	if (enable) {
+		data |= (PCIE_CNTL2__SLV_MEM_LS_EN_MASK |
+			 PCIE_CNTL2__MST_MEM_LS_EN_MASK |
+			 PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK);
+	} else {
+		data &= ~(PCIE_CNTL2__SLV_MEM_LS_EN_MASK |
+			  PCIE_CNTL2__MST_MEM_LS_EN_MASK |
+			  PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK);
+	}
+
+	if (def != data)
+		WREG32_PCIE(smnPCIE_CNTL2, data);
+}
+
+static void nbio_v2_3_get_clockgating_state(struct amdgpu_device *adev,
+					    u64 *flags)
+{
+	int data;
+
+	/* AMD_CG_SUPPORT_BIF_MGCG */
+	data = RREG32_PCIE(smnCPM_CONTROL);
+	if (data & CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_BIF_MGCG;
+
+	/* AMD_CG_SUPPORT_BIF_LS */
+	data = RREG32_PCIE(smnPCIE_CNTL2);
+	if (data & PCIE_CNTL2__SLV_MEM_LS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_BIF_LS;
+}
+
+static u32 nbio_v2_3_get_hdp_flush_req_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF_GPU_HDP_FLUSH_REQ);
+}
+
+static u32 nbio_v2_3_get_hdp_flush_done_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF_GPU_HDP_FLUSH_DONE);
+}
+
+static u32 nbio_v2_3_get_pcie_index_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, mmPCIE_INDEX2);
+}
+
+static u32 nbio_v2_3_get_pcie_data_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, mmPCIE_DATA2);
+}
+
+const struct nbio_hdp_flush_reg nbio_v2_3_hdp_flush_reg = {
+	.ref_and_mask_cp0 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP0_MASK,
+	.ref_and_mask_cp1 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP1_MASK,
+	.ref_and_mask_cp2 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP2_MASK,
+	.ref_and_mask_cp3 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP3_MASK,
+	.ref_and_mask_cp4 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP4_MASK,
+	.ref_and_mask_cp5 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP5_MASK,
+	.ref_and_mask_cp6 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP6_MASK,
+	.ref_and_mask_cp7 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP7_MASK,
+	.ref_and_mask_cp8 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP8_MASK,
+	.ref_and_mask_cp9 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP9_MASK,
+	.ref_and_mask_sdma0 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__SDMA0_MASK,
+	.ref_and_mask_sdma1 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__SDMA1_MASK,
+};
+
+static void nbio_v2_3_init_registers(struct amdgpu_device *adev)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_PCIE(smnPCIE_CONFIG_CNTL);
+	data = REG_SET_FIELD(data, PCIE_CONFIG_CNTL, CI_SWUS_MAX_READ_REQUEST_SIZE_MODE, 1);
+	data = REG_SET_FIELD(data, PCIE_CONFIG_CNTL, CI_SWUS_MAX_READ_REQUEST_SIZE_PRIV, 1);
+
+	if (def != data)
+		WREG32_PCIE(smnPCIE_CONFIG_CNTL, data);
+}
+
+#define NAVI10_PCIE__LC_L0S_INACTIVITY_DEFAULT		0x00000000 // off by default, no gains over L1
+#define NAVI10_PCIE__LC_L1_INACTIVITY_DEFAULT		0x0000000A // 1=1us, 9=1ms, 10=4ms
+#define NAVI10_PCIE__LC_L1_INACTIVITY_TBT_DEFAULT	0x0000000E // 400ms
+
+static void nbio_v2_3_enable_aspm(struct amdgpu_device *adev,
+				  bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_PCIE(smnPCIE_LC_CNTL);
+
+	if (enable) {
+		/* Disable ASPM L0s/L1 first */
+		data &= ~(PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK | PCIE_LC_CNTL__LC_L1_INACTIVITY_MASK);
+
+		data |= NAVI10_PCIE__LC_L0S_INACTIVITY_DEFAULT << PCIE_LC_CNTL__LC_L0S_INACTIVITY__SHIFT;
+
+		if (dev_is_removable(&adev->pdev->dev))
+			data |= NAVI10_PCIE__LC_L1_INACTIVITY_TBT_DEFAULT  << PCIE_LC_CNTL__LC_L1_INACTIVITY__SHIFT;
+		else
+			data |= NAVI10_PCIE__LC_L1_INACTIVITY_DEFAULT << PCIE_LC_CNTL__LC_L1_INACTIVITY__SHIFT;
+
+		data &= ~PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK;
+	} else {
+		/* Disable ASPM L1 */
+		data &= ~PCIE_LC_CNTL__LC_L1_INACTIVITY_MASK;
+		/* Disable ASPM TxL0s */
+		data &= ~PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK;
+		/* Disable ACPI L1 */
+		data |= PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK;
+	}
+
+	if (def != data)
+		WREG32_PCIE(smnPCIE_LC_CNTL, data);
+}
+
+#ifdef CONFIG_PCIEASPM
+static void nbio_v2_3_program_ltr(struct amdgpu_device *adev)
+{
+	uint32_t def, data;
+
+	WREG32_PCIE(smnRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL, 0x75EB);
+
+	def = data = RREG32_SOC15(NBIO, 0, mmRCC_BIF_STRAP2);
+	data &= ~RCC_BIF_STRAP2__STRAP_LTR_IN_ASPML1_DIS_MASK;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, mmRCC_BIF_STRAP2, data);
+
+	def = data = RREG32_PCIE(smnRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL);
+	data &= ~EP_PCIE_TX_LTR_CNTL__LTR_PRIV_MSG_DIS_IN_PM_NON_D0_MASK;
+	if (def != data)
+		WREG32_PCIE(smnRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL, data);
+
+	def = data = RREG32_PCIE(smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2);
+	data |= BIF_CFG_DEV0_EPF0_DEVICE_CNTL2__LTR_EN_MASK;
+	if (def != data)
+		WREG32_PCIE(smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2, data);
+}
+#endif
+
+static void nbio_v2_3_program_aspm(struct amdgpu_device *adev)
+{
+#ifdef CONFIG_PCIEASPM
+	uint32_t def, data;
+
+	def = data = RREG32_PCIE(smnPCIE_LC_CNTL);
+	data &= ~PCIE_LC_CNTL__LC_L1_INACTIVITY_MASK;
+	data &= ~PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK;
+	data |= PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK;
+	if (def != data)
+		WREG32_PCIE(smnPCIE_LC_CNTL, data);
+
+	def = data = RREG32_PCIE(smnPCIE_LC_CNTL7);
+	data |= PCIE_LC_CNTL7__LC_NBIF_ASPM_INPUT_EN_MASK;
+	if (def != data)
+		WREG32_PCIE(smnPCIE_LC_CNTL7, data);
+
+	def = data = RREG32_PCIE(smnNBIF_MGCG_CTRL_LCLK);
+	data |= NBIF_MGCG_CTRL_LCLK__NBIF_MGCG_REG_DIS_LCLK_MASK;
+	if (def != data)
+		WREG32_PCIE(smnNBIF_MGCG_CTRL_LCLK, data);
+
+	def = data = RREG32_PCIE(smnPCIE_LC_CNTL3);
+	data |= PCIE_LC_CNTL3__LC_DSC_DONT_ENTER_L23_AFTER_PME_ACK_MASK;
+	if (def != data)
+		WREG32_PCIE(smnPCIE_LC_CNTL3, data);
+
+	def = data = RREG32_SOC15(NBIO, 0, mmRCC_BIF_STRAP3);
+	data &= ~RCC_BIF_STRAP3__STRAP_VLINK_ASPM_IDLE_TIMER_MASK;
+	data &= ~RCC_BIF_STRAP3__STRAP_VLINK_PM_L1_ENTRY_TIMER_MASK;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, mmRCC_BIF_STRAP3, data);
+
+	def = data = RREG32_SOC15(NBIO, 0, mmRCC_BIF_STRAP5);
+	data &= ~RCC_BIF_STRAP5__STRAP_VLINK_LDN_ENTRY_TIMER_MASK;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, mmRCC_BIF_STRAP5, data);
+
+	def = data = RREG32_PCIE(smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2);
+	data &= ~BIF_CFG_DEV0_EPF0_DEVICE_CNTL2__LTR_EN_MASK;
+	if (def != data)
+		WREG32_PCIE(smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2, data);
+
+	WREG32_PCIE(smnBIF_CFG_DEV0_EPF0_PCIE_LTR_CAP, 0x10011001);
+
+	def = data = RREG32_PCIE(smnPSWUSP0_PCIE_LC_CNTL2);
+	data |= PSWUSP0_PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L1_MASK |
+		PSWUSP0_PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L23_MASK;
+	data &= ~PSWUSP0_PCIE_LC_CNTL2__LC_RCV_L0_TO_RCV_L0S_DIS_MASK;
+	if (def != data)
+		WREG32_PCIE(smnPSWUSP0_PCIE_LC_CNTL2, data);
+
+	def = data = RREG32_PCIE(smnPCIE_LC_CNTL6);
+	data |= PCIE_LC_CNTL6__LC_L1_POWERDOWN_MASK |
+		PCIE_LC_CNTL6__LC_RX_L0S_STANDBY_EN_MASK;
+	if (def != data)
+		WREG32_PCIE(smnPCIE_LC_CNTL6, data);
+
+	/* Don't bother about LTR if LTR is not enabled
+	 * in the path */
+	if (adev->pdev->ltr_path)
+		nbio_v2_3_program_ltr(adev);
+
+	def = data = RREG32_SOC15(NBIO, 0, mmRCC_BIF_STRAP3);
+	data |= 0x5DE0 << RCC_BIF_STRAP3__STRAP_VLINK_ASPM_IDLE_TIMER__SHIFT;
+	data |= 0x0010 << RCC_BIF_STRAP3__STRAP_VLINK_PM_L1_ENTRY_TIMER__SHIFT;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, mmRCC_BIF_STRAP3, data);
+
+	def = data = RREG32_SOC15(NBIO, 0, mmRCC_BIF_STRAP5);
+	data |= 0x0010 << RCC_BIF_STRAP5__STRAP_VLINK_LDN_ENTRY_TIMER__SHIFT;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, mmRCC_BIF_STRAP5, data);
+
+	def = data = RREG32_PCIE(smnPCIE_LC_CNTL);
+	data |= NAVI10_PCIE__LC_L0S_INACTIVITY_DEFAULT << PCIE_LC_CNTL__LC_L0S_INACTIVITY__SHIFT;
+	if (dev_is_removable(&adev->pdev->dev))
+		data |= NAVI10_PCIE__LC_L1_INACTIVITY_TBT_DEFAULT  << PCIE_LC_CNTL__LC_L1_INACTIVITY__SHIFT;
+	else
+		data |= NAVI10_PCIE__LC_L1_INACTIVITY_DEFAULT << PCIE_LC_CNTL__LC_L1_INACTIVITY__SHIFT;
+	data &= ~PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK;
+	if (def != data)
+		WREG32_PCIE(smnPCIE_LC_CNTL, data);
+
+	def = data = RREG32_PCIE(smnPCIE_LC_CNTL3);
+	data &= ~PCIE_LC_CNTL3__LC_DSC_DONT_ENTER_L23_AFTER_PME_ACK_MASK;
+	if (def != data)
+		WREG32_PCIE(smnPCIE_LC_CNTL3, data);
+#endif
+}
+
+static void nbio_v2_3_apply_lc_spc_mode_wa(struct amdgpu_device *adev)
+{
+	uint32_t reg_data = 0;
+	uint32_t link_width = 0;
+
+	if (!((adev->asic_type >= CHIP_NAVI10) &&
+	     (adev->asic_type <= CHIP_NAVI12)))
+		return;
+
+	reg_data = RREG32_PCIE(smnPCIE_LC_LINK_WIDTH_CNTL);
+	link_width = (reg_data & PCIE_LC_LINK_WIDTH_CNTL__LC_LINK_WIDTH_RD_MASK)
+		>> PCIE_LC_LINK_WIDTH_CNTL__LC_LINK_WIDTH_RD__SHIFT;
+
+	/*
+	 * Program PCIE_LC_CNTL6.LC_SPC_MODE_8GT to 0x2 (4 symbols per clock data)
+	 * if link_width is 0x3 (x4)
+	 */
+	if (0x3 == link_width) {
+		reg_data = RREG32_PCIE(smnPCIE_LC_CNTL6);
+		reg_data &= ~PCIE_LC_CNTL6__LC_SPC_MODE_8GT_MASK;
+		reg_data |= (0x2 << PCIE_LC_CNTL6__LC_SPC_MODE_8GT__SHIFT);
+		WREG32_PCIE(smnPCIE_LC_CNTL6, reg_data);
+	}
+}
+
+static void nbio_v2_3_apply_l1_link_width_reconfig_wa(struct amdgpu_device *adev)
+{
+	uint32_t reg_data = 0;
+
+	if (adev->asic_type != CHIP_NAVI10)
+		return;
+
+	reg_data = RREG32_PCIE(smnPCIE_LC_LINK_WIDTH_CNTL);
+	reg_data |= PCIE_LC_LINK_WIDTH_CNTL__LC_L1_RECONFIG_EN_MASK;
+	WREG32_PCIE(smnPCIE_LC_LINK_WIDTH_CNTL, reg_data);
+}
+
+static void nbio_v2_3_clear_doorbell_interrupt(struct amdgpu_device *adev)
+{
+	uint32_t reg, reg_data;
+
+	if (amdgpu_ip_version(adev, NBIO_HWIP, 0) != IP_VERSION(3, 3, 0))
+		return;
+
+	reg = RREG32_SOC15(NBIO, 0, mmBIF_RB_CNTL);
+
+	/* Clear Interrupt Status
+	 */
+	if ((reg & BIF_RB_CNTL__RB_ENABLE_MASK) == 0) {
+		reg = RREG32_SOC15(NBIO, 0, mmBIF_DOORBELL_INT_CNTL);
+		if (reg & BIF_DOORBELL_INT_CNTL__DOORBELL_INTERRUPT_STATUS_MASK) {
+			reg_data = 1 << BIF_DOORBELL_INT_CNTL__DOORBELL_INTERRUPT_CLEAR__SHIFT;
+			WREG32_SOC15(NBIO, 0, mmBIF_DOORBELL_INT_CNTL, reg_data);
+		}
+	}
+}
+
+#define MMIO_REG_HOLE_OFFSET (0x80000 - PAGE_SIZE)
+
+static void nbio_v2_3_set_reg_remap(struct amdgpu_device *adev)
+{
+	if (!amdgpu_sriov_vf(adev) && (PAGE_SIZE <= 4096)) {
+		adev->rmmio_remap.reg_offset = MMIO_REG_HOLE_OFFSET;
+		adev->rmmio_remap.bus_addr = adev->rmmio_base + MMIO_REG_HOLE_OFFSET;
+	} else {
+		adev->rmmio_remap.reg_offset = SOC15_REG_OFFSET(NBIO, 0,
+			mmBIF_BX_DEV0_EPF0_VF0_HDP_MEM_COHERENCY_FLUSH_CNTL) << 2;
+		adev->rmmio_remap.bus_addr = 0;
+	}
+}
+
+const struct amdgpu_nbio_funcs nbio_v2_3_funcs = {
+	.get_hdp_flush_req_offset = nbio_v2_3_get_hdp_flush_req_offset,
+	.get_hdp_flush_done_offset = nbio_v2_3_get_hdp_flush_done_offset,
+	.get_pcie_index_offset = nbio_v2_3_get_pcie_index_offset,
+	.get_pcie_data_offset = nbio_v2_3_get_pcie_data_offset,
+	.get_rev_id = nbio_v2_3_get_rev_id,
+	.mc_access_enable = nbio_v2_3_mc_access_enable,
+	.get_memsize = nbio_v2_3_get_memsize,
+	.sdma_doorbell_range = nbio_v2_3_sdma_doorbell_range,
+	.vcn_doorbell_range = nbio_v2_3_vcn_doorbell_range,
+	.enable_doorbell_aperture = nbio_v2_3_enable_doorbell_aperture,
+	.enable_doorbell_selfring_aperture = nbio_v2_3_enable_doorbell_selfring_aperture,
+	.ih_doorbell_range = nbio_v2_3_ih_doorbell_range,
+	.update_medium_grain_clock_gating = nbio_v2_3_update_medium_grain_clock_gating,
+	.update_medium_grain_light_sleep = nbio_v2_3_update_medium_grain_light_sleep,
+	.get_clockgating_state = nbio_v2_3_get_clockgating_state,
+	.ih_control = nbio_v2_3_ih_control,
+	.init_registers = nbio_v2_3_init_registers,
+	.remap_hdp_registers = nbio_v2_3_remap_hdp_registers,
+	.enable_aspm = nbio_v2_3_enable_aspm,
+	.program_aspm =  nbio_v2_3_program_aspm,
+	.apply_lc_spc_mode_wa = nbio_v2_3_apply_lc_spc_mode_wa,
+	.apply_l1_link_width_reconfig_wa = nbio_v2_3_apply_l1_link_width_reconfig_wa,
+	.clear_doorbell_interrupt = nbio_v2_3_clear_doorbell_interrupt,
+	.set_reg_remap = nbio_v2_3_set_reg_remap,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v2_3.h b/drivers/gpu/drm/amd/amdgpu/nbio_v2_3.h
new file mode 100644
index 000000000000..a43b60acf7f6
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v2_3.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __NBIO_V2_3_H__
+#define __NBIO_V2_3_H__
+
+#include "soc15_common.h"
+
+extern const struct nbio_hdp_flush_reg nbio_v2_3_hdp_flush_reg;
+extern const struct amdgpu_nbio_funcs nbio_v2_3_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v4_3.c b/drivers/gpu/drm/amd/amdgpu/nbio_v4_3.c
new file mode 100644
index 000000000000..f89e5f40e1a5
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v4_3.c
@@ -0,0 +1,634 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "nbio_v4_3.h"
+
+#include "nbio/nbio_4_3_0_offset.h"
+#include "nbio/nbio_4_3_0_sh_mask.h"
+#include "ivsrcid/nbio/irqsrcs_nbif_7_4.h"
+#include <uapi/linux/kfd_ioctl.h>
+
+static void nbio_v4_3_remap_hdp_registers(struct amdgpu_device *adev)
+{
+	WREG32_SOC15(NBIO, 0, regBIF_BX0_REMAP_HDP_MEM_FLUSH_CNTL,
+		adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL);
+	WREG32_SOC15(NBIO, 0, regBIF_BX0_REMAP_HDP_REG_FLUSH_CNTL,
+		adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_REG_FLUSH_CNTL);
+}
+
+static u32 nbio_v4_3_get_rev_id(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_DEV0_EPF0_STRAP0);
+
+	tmp &= RCC_STRAP0_RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0_MASK;
+	tmp >>= RCC_STRAP0_RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0__SHIFT;
+
+	return tmp;
+}
+
+static void nbio_v4_3_mc_access_enable(struct amdgpu_device *adev, bool enable)
+{
+	if (enable)
+		WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_FB_EN,
+			     BIF_BX0_BIF_FB_EN__FB_READ_EN_MASK |
+			     BIF_BX0_BIF_FB_EN__FB_WRITE_EN_MASK);
+	else
+		WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_FB_EN, 0);
+}
+
+static u32 nbio_v4_3_get_memsize(struct amdgpu_device *adev)
+{
+	return RREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF0_RCC_CONFIG_MEMSIZE);
+}
+
+static void nbio_v4_3_sdma_doorbell_range(struct amdgpu_device *adev, int instance,
+					  bool use_doorbell, int doorbell_index,
+					  int doorbell_size)
+{
+	if (instance == 0) {
+		u32 doorbell_range = RREG32_SOC15(NBIO, 0, regS2A_DOORBELL_ENTRY_2_CTRL);
+
+		if (use_doorbell) {
+			doorbell_range = REG_SET_FIELD(doorbell_range,
+						       S2A_DOORBELL_ENTRY_2_CTRL,
+						       S2A_DOORBELL_PORT2_ENABLE,
+						       0x1);
+			doorbell_range = REG_SET_FIELD(doorbell_range,
+						       S2A_DOORBELL_ENTRY_2_CTRL,
+						       S2A_DOORBELL_PORT2_AWID,
+						       0xe);
+			doorbell_range = REG_SET_FIELD(doorbell_range,
+						       S2A_DOORBELL_ENTRY_2_CTRL,
+						       S2A_DOORBELL_PORT2_RANGE_OFFSET,
+						       doorbell_index);
+			doorbell_range = REG_SET_FIELD(doorbell_range,
+						       S2A_DOORBELL_ENTRY_2_CTRL,
+						       S2A_DOORBELL_PORT2_RANGE_SIZE,
+						       doorbell_size);
+			doorbell_range = REG_SET_FIELD(doorbell_range,
+						       S2A_DOORBELL_ENTRY_2_CTRL,
+						       S2A_DOORBELL_PORT2_AWADDR_31_28_VALUE,
+						       0x3);
+		} else
+			doorbell_range = REG_SET_FIELD(doorbell_range,
+						       S2A_DOORBELL_ENTRY_2_CTRL,
+						       S2A_DOORBELL_PORT2_RANGE_SIZE,
+						       0);
+
+		WREG32_SOC15(NBIO, 0, regS2A_DOORBELL_ENTRY_2_CTRL, doorbell_range);
+	}
+}
+
+static void nbio_v4_3_vcn_doorbell_range(struct amdgpu_device *adev, bool use_doorbell,
+					 int doorbell_index, int instance)
+{
+	u32 doorbell_range;
+
+	if (instance)
+		doorbell_range = RREG32_SOC15(NBIO, 0, regS2A_DOORBELL_ENTRY_5_CTRL);
+	else
+		doorbell_range = RREG32_SOC15(NBIO, 0, regS2A_DOORBELL_ENTRY_4_CTRL);
+
+	if (use_doorbell) {
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       S2A_DOORBELL_ENTRY_4_CTRL,
+					       S2A_DOORBELL_PORT4_ENABLE,
+					       0x1);
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       S2A_DOORBELL_ENTRY_4_CTRL,
+					       S2A_DOORBELL_PORT4_AWID,
+					       instance ? 0x7 : 0x4);
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       S2A_DOORBELL_ENTRY_4_CTRL,
+					       S2A_DOORBELL_PORT4_RANGE_OFFSET,
+					       doorbell_index);
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       S2A_DOORBELL_ENTRY_4_CTRL,
+					       S2A_DOORBELL_PORT4_RANGE_SIZE,
+					       8);
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       S2A_DOORBELL_ENTRY_4_CTRL,
+					       S2A_DOORBELL_PORT4_AWADDR_31_28_VALUE,
+					       instance ? 0x7 : 0x4);
+	} else
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       S2A_DOORBELL_ENTRY_4_CTRL,
+					       S2A_DOORBELL_PORT4_RANGE_SIZE,
+					       0);
+
+	if (instance)
+		WREG32_SOC15(NBIO, 0, regS2A_DOORBELL_ENTRY_5_CTRL, doorbell_range);
+	else
+		WREG32_SOC15(NBIO, 0, regS2A_DOORBELL_ENTRY_4_CTRL, doorbell_range);
+}
+
+static void nbio_v4_3_gc_doorbell_init(struct amdgpu_device *adev)
+{
+	WREG32_SOC15(NBIO, 0, regS2A_DOORBELL_ENTRY_0_CTRL, 0x30000007);
+	WREG32_SOC15(NBIO, 0, regS2A_DOORBELL_ENTRY_3_CTRL, 0x3000000d);
+}
+
+static void nbio_v4_3_enable_doorbell_aperture(struct amdgpu_device *adev,
+					       bool enable)
+{
+	WREG32_FIELD15_PREREG(NBIO, 0, RCC_DEV0_EPF0_RCC_DOORBELL_APER_EN,
+			BIF_DOORBELL_APER_EN, enable ? 1 : 0);
+}
+
+static void nbio_v4_3_enable_doorbell_selfring_aperture(struct amdgpu_device *adev,
+							bool enable)
+{
+	u32 tmp = 0;
+
+	if (enable) {
+		tmp = REG_SET_FIELD(tmp, BIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL,
+				    DOORBELL_SELFRING_GPA_APER_EN, 1) |
+		      REG_SET_FIELD(tmp, BIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL,
+				    DOORBELL_SELFRING_GPA_APER_MODE, 1) |
+		      REG_SET_FIELD(tmp, BIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL,
+				    DOORBELL_SELFRING_GPA_APER_SIZE, 0);
+
+		WREG32_SOC15(NBIO, 0, regBIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_BASE_LOW,
+			     lower_32_bits(adev->doorbell.base));
+		WREG32_SOC15(NBIO, 0, regBIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_BASE_HIGH,
+			     upper_32_bits(adev->doorbell.base));
+	}
+
+	WREG32_SOC15(NBIO, 0, regBIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL,
+		     tmp);
+}
+
+static void nbio_v4_3_ih_doorbell_range(struct amdgpu_device *adev,
+					bool use_doorbell, int doorbell_index)
+{
+	u32 ih_doorbell_range = RREG32_SOC15(NBIO, 0, regS2A_DOORBELL_ENTRY_1_CTRL);
+
+	if (use_doorbell) {
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  S2A_DOORBELL_ENTRY_1_CTRL,
+						  S2A_DOORBELL_PORT1_ENABLE,
+						  0x1);
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  S2A_DOORBELL_ENTRY_1_CTRL,
+						  S2A_DOORBELL_PORT1_AWID,
+						  0x0);
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  S2A_DOORBELL_ENTRY_1_CTRL,
+						  S2A_DOORBELL_PORT1_RANGE_OFFSET,
+						  doorbell_index);
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  S2A_DOORBELL_ENTRY_1_CTRL,
+						  S2A_DOORBELL_PORT1_RANGE_SIZE,
+						  2);
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  S2A_DOORBELL_ENTRY_1_CTRL,
+						  S2A_DOORBELL_PORT1_AWADDR_31_28_VALUE,
+						  0x0);
+	} else
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  S2A_DOORBELL_ENTRY_1_CTRL,
+						  S2A_DOORBELL_PORT1_RANGE_SIZE,
+						  0);
+
+	WREG32_SOC15(NBIO, 0, regS2A_DOORBELL_ENTRY_1_CTRL, ih_doorbell_range);
+}
+
+static void nbio_v4_3_ih_control(struct amdgpu_device *adev)
+{
+	u32 interrupt_cntl;
+
+	/* setup interrupt control */
+	WREG32_SOC15(NBIO, 0, regBIF_BX0_INTERRUPT_CNTL2, adev->dummy_page_addr >> 8);
+
+	interrupt_cntl = RREG32_SOC15(NBIO, 0, regBIF_BX0_INTERRUPT_CNTL);
+	/*
+	 * BIF_BX0_INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
+	 * BIF_BX0_INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
+	 */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, BIF_BX0_INTERRUPT_CNTL,
+				       IH_DUMMY_RD_OVERRIDE, 0);
+
+	/* BIF_BX0_INTERRUPT_CNTL__IH_REQ_NONSNOOP_EN_MASK=1 if ring is in non-cacheable memory, e.g., vram */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, BIF_BX0_INTERRUPT_CNTL,
+				       IH_REQ_NONSNOOP_EN, 0);
+
+	WREG32_SOC15(NBIO, 0, regBIF_BX0_INTERRUPT_CNTL, interrupt_cntl);
+}
+
+static void nbio_v4_3_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						       bool enable)
+{
+	uint32_t def, data;
+
+	if (enable && !(adev->cg_flags & AMD_CG_SUPPORT_BIF_MGCG))
+		return;
+
+	def = data = RREG32_SOC15(NBIO, 0, regCPM_CONTROL);
+	if (enable) {
+		data |= (CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK |
+			 CPM_CONTROL__TXCLK_DYN_GATE_ENABLE_MASK |
+			 CPM_CONTROL__TXCLK_LCNT_GATE_ENABLE_MASK |
+			 CPM_CONTROL__TXCLK_REGS_GATE_ENABLE_MASK |
+			 CPM_CONTROL__TXCLK_PRBS_GATE_ENABLE_MASK |
+			 CPM_CONTROL__REFCLK_REGS_GATE_ENABLE_MASK);
+	} else {
+		data &= ~(CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK |
+			  CPM_CONTROL__TXCLK_DYN_GATE_ENABLE_MASK |
+			  CPM_CONTROL__TXCLK_LCNT_GATE_ENABLE_MASK |
+			  CPM_CONTROL__TXCLK_REGS_GATE_ENABLE_MASK |
+			  CPM_CONTROL__TXCLK_PRBS_GATE_ENABLE_MASK |
+			  CPM_CONTROL__REFCLK_REGS_GATE_ENABLE_MASK);
+	}
+
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regCPM_CONTROL, data);
+}
+
+static void nbio_v4_3_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+						      bool enable)
+{
+	uint32_t def, data;
+
+	if (enable && !(adev->cg_flags & AMD_CG_SUPPORT_BIF_LS))
+		return;
+
+	/* TODO: need update in future */
+	def = data = RREG32_SOC15(NBIO, 0, regPCIE_CNTL2);
+	if (enable) {
+		data |= PCIE_CNTL2__SLV_MEM_LS_EN_MASK;
+	} else {
+		data &= ~PCIE_CNTL2__SLV_MEM_LS_EN_MASK;
+	}
+
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regPCIE_CNTL2, data);
+}
+
+static void nbio_v4_3_get_clockgating_state(struct amdgpu_device *adev,
+					    u64 *flags)
+{
+	int data;
+
+	/* AMD_CG_SUPPORT_BIF_MGCG */
+	data = RREG32_SOC15(NBIO, 0, regCPM_CONTROL);
+	if (data & CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_BIF_MGCG;
+
+	/* AMD_CG_SUPPORT_BIF_LS */
+	data = RREG32_SOC15(NBIO, 0, regPCIE_CNTL2);
+	if (data & PCIE_CNTL2__SLV_MEM_LS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_BIF_LS;
+}
+
+static u32 nbio_v4_3_get_hdp_flush_req_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF0_GPU_HDP_FLUSH_REQ);
+}
+
+static u32 nbio_v4_3_get_hdp_flush_done_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF0_GPU_HDP_FLUSH_DONE);
+}
+
+static u32 nbio_v4_3_get_pcie_index_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF0_RSMU_INDEX);
+}
+
+static u32 nbio_v4_3_get_pcie_data_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF0_RSMU_DATA);
+}
+
+const struct nbio_hdp_flush_reg nbio_v4_3_hdp_flush_reg = {
+	.ref_and_mask_cp0 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP0_MASK,
+	.ref_and_mask_cp1 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP1_MASK,
+	.ref_and_mask_cp2 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP2_MASK,
+	.ref_and_mask_cp3 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP3_MASK,
+	.ref_and_mask_cp4 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP4_MASK,
+	.ref_and_mask_cp5 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP5_MASK,
+	.ref_and_mask_cp6 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP6_MASK,
+	.ref_and_mask_cp7 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP7_MASK,
+	.ref_and_mask_cp8 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP8_MASK,
+	.ref_and_mask_cp9 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP9_MASK,
+	.ref_and_mask_sdma0 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__SDMA0_MASK,
+	.ref_and_mask_sdma1 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__SDMA1_MASK,
+};
+
+static void nbio_v4_3_init_registers(struct amdgpu_device *adev)
+{
+	if (amdgpu_ip_version(adev, NBIO_HWIP, 0) == IP_VERSION(4, 3, 0)) {
+		uint32_t data;
+
+		data = RREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF2_STRAP2);
+		data &= ~RCC_DEV0_EPF2_STRAP2__STRAP_NO_SOFT_RESET_DEV0_F2_MASK;
+		WREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF2_STRAP2, data);
+	}
+}
+
+static u32 nbio_v4_3_get_rom_offset(struct amdgpu_device *adev)
+{
+	u32 data, rom_offset;
+
+	data = RREG32_SOC15(NBIO, 0, regREGS_ROM_OFFSET_CTRL);
+	rom_offset = REG_GET_FIELD(data, REGS_ROM_OFFSET_CTRL, ROM_OFFSET);
+
+	return rom_offset;
+}
+
+#ifdef CONFIG_PCIEASPM
+static void nbio_v4_3_program_ltr(struct amdgpu_device *adev)
+{
+	uint32_t def, data;
+
+	def = RREG32_SOC15(NBIO, 0, regRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL);
+	data = 0x35EB;
+	data &= ~EP_PCIE_TX_LTR_CNTL__LTR_PRIV_MSG_DIS_IN_PM_NON_D0_MASK;
+	data &= ~EP_PCIE_TX_LTR_CNTL__LTR_PRIV_RST_LTR_IN_DL_DOWN_MASK;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL, data);
+
+	def = data = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_BIF_STRAP2);
+	data &= ~RCC_BIF_STRAP2__STRAP_LTR_IN_ASPML1_DIS_MASK;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_BIF_STRAP2, data);
+
+	def = data = RREG32_SOC15(NBIO, 0, regBIF_CFG_DEV0_EPF0_DEVICE_CNTL2);
+	if (adev->pdev->ltr_path)
+		data |= BIF_CFG_DEV0_EPF0_DEVICE_CNTL2__LTR_EN_MASK;
+	else
+		data &= ~BIF_CFG_DEV0_EPF0_DEVICE_CNTL2__LTR_EN_MASK;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regBIF_CFG_DEV0_EPF0_DEVICE_CNTL2, data);
+}
+#endif
+
+static void nbio_v4_3_program_aspm(struct amdgpu_device *adev)
+{
+#ifdef CONFIG_PCIEASPM
+	uint32_t def, data;
+
+	if (!(amdgpu_ip_version(adev, PCIE_HWIP, 0) == IP_VERSION(7, 4, 0)) &&
+	    !(amdgpu_ip_version(adev, PCIE_HWIP, 0) == IP_VERSION(7, 6, 0)))
+		return;
+
+	def = data = RREG32_SOC15(NBIO, 0, regPCIE_LC_CNTL);
+	data &= ~PCIE_LC_CNTL__LC_L1_INACTIVITY_MASK;
+	data &= ~PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK;
+	data |= PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regPCIE_LC_CNTL, data);
+
+	def = data = RREG32_SOC15(NBIO, 0, regPCIE_LC_CNTL7);
+	data |= PCIE_LC_CNTL7__LC_NBIF_ASPM_INPUT_EN_MASK;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regPCIE_LC_CNTL7, data);
+
+	def = data = RREG32_SOC15(NBIO, 0, regPCIE_LC_CNTL3);
+	data |= PCIE_LC_CNTL3__LC_DSC_DONT_ENTER_L23_AFTER_PME_ACK_MASK;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regPCIE_LC_CNTL3, data);
+
+	def = data = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_BIF_STRAP3);
+	data &= ~RCC_BIF_STRAP3__STRAP_VLINK_ASPM_IDLE_TIMER_MASK;
+	data &= ~RCC_BIF_STRAP3__STRAP_VLINK_PM_L1_ENTRY_TIMER_MASK;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_BIF_STRAP3, data);
+
+	def = data = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_BIF_STRAP5);
+	data &= ~RCC_BIF_STRAP5__STRAP_VLINK_LDN_ENTRY_TIMER_MASK;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_BIF_STRAP5, data);
+
+	def = data = RREG32_SOC15(NBIO, 0, regBIF_CFG_DEV0_EPF0_DEVICE_CNTL2);
+	data &= ~BIF_CFG_DEV0_EPF0_DEVICE_CNTL2__LTR_EN_MASK;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regBIF_CFG_DEV0_EPF0_DEVICE_CNTL2, data);
+
+	WREG32_SOC15(NBIO, 0, regBIF_CFG_DEV0_EPF0_PCIE_LTR_CAP, 0x10011001);
+
+	def = data = RREG32_SOC15(NBIO, 0, regPSWUSP0_PCIE_LC_CNTL2);
+	data |= PSWUSP0_PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L1_MASK |
+		PSWUSP0_PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L23_MASK;
+	data &= ~PSWUSP0_PCIE_LC_CNTL2__LC_RCV_L0_TO_RCV_L0S_DIS_MASK;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regPSWUSP0_PCIE_LC_CNTL2, data);
+
+	def = data = RREG32_SOC15(NBIO, 0, regPCIE_LC_CNTL4);
+	data |= PCIE_LC_CNTL4__LC_L1_POWERDOWN_MASK;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regPCIE_LC_CNTL4, data);
+
+	def = data = RREG32_SOC15(NBIO, 0, regPCIE_LC_RXRECOVER_RXSTANDBY_CNTL);
+	data |= PCIE_LC_RXRECOVER_RXSTANDBY_CNTL__LC_RX_L0S_STANDBY_EN_MASK;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regPCIE_LC_RXRECOVER_RXSTANDBY_CNTL, data);
+
+	nbio_v4_3_program_ltr(adev);
+
+	def = data = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_BIF_STRAP3);
+	data |= 0x5DE0 << RCC_BIF_STRAP3__STRAP_VLINK_ASPM_IDLE_TIMER__SHIFT;
+	data |= 0x0010 << RCC_BIF_STRAP3__STRAP_VLINK_PM_L1_ENTRY_TIMER__SHIFT;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_BIF_STRAP3, data);
+
+	def = data = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_BIF_STRAP5);
+	data |= 0x0010 << RCC_BIF_STRAP5__STRAP_VLINK_LDN_ENTRY_TIMER__SHIFT;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_BIF_STRAP5, data);
+
+	def = data = RREG32_SOC15(NBIO, 0, regPCIE_LC_CNTL);
+	data |= 0x0 << PCIE_LC_CNTL__LC_L0S_INACTIVITY__SHIFT;
+	data |= 0x9 << PCIE_LC_CNTL__LC_L1_INACTIVITY__SHIFT;
+	data &= ~PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regPCIE_LC_CNTL, data);
+
+	def = data = RREG32_SOC15(NBIO, 0, regPCIE_LC_CNTL3);
+	data &= ~PCIE_LC_CNTL3__LC_DSC_DONT_ENTER_L23_AFTER_PME_ACK_MASK;
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regPCIE_LC_CNTL3, data);
+#endif
+}
+
+#define MMIO_REG_HOLE_OFFSET (0x80000 - PAGE_SIZE)
+
+static void nbio_v4_3_set_reg_remap(struct amdgpu_device *adev)
+{
+	if (!amdgpu_sriov_vf(adev) && (PAGE_SIZE <= 4096)) {
+		adev->rmmio_remap.reg_offset = MMIO_REG_HOLE_OFFSET;
+		adev->rmmio_remap.bus_addr = adev->rmmio_base + MMIO_REG_HOLE_OFFSET;
+	} else {
+		adev->rmmio_remap.reg_offset = SOC15_REG_OFFSET(NBIO, 0,
+			regBIF_BX_DEV0_EPF0_VF0_HDP_MEM_COHERENCY_FLUSH_CNTL) << 2;
+		adev->rmmio_remap.bus_addr = 0;
+	}
+}
+
+const struct amdgpu_nbio_funcs nbio_v4_3_funcs = {
+	.get_hdp_flush_req_offset = nbio_v4_3_get_hdp_flush_req_offset,
+	.get_hdp_flush_done_offset = nbio_v4_3_get_hdp_flush_done_offset,
+	.get_pcie_index_offset = nbio_v4_3_get_pcie_index_offset,
+	.get_pcie_data_offset = nbio_v4_3_get_pcie_data_offset,
+	.get_rev_id = nbio_v4_3_get_rev_id,
+	.mc_access_enable = nbio_v4_3_mc_access_enable,
+	.get_memsize = nbio_v4_3_get_memsize,
+	.sdma_doorbell_range = nbio_v4_3_sdma_doorbell_range,
+	.vcn_doorbell_range = nbio_v4_3_vcn_doorbell_range,
+	.gc_doorbell_init = nbio_v4_3_gc_doorbell_init,
+	.enable_doorbell_aperture = nbio_v4_3_enable_doorbell_aperture,
+	.enable_doorbell_selfring_aperture = nbio_v4_3_enable_doorbell_selfring_aperture,
+	.ih_doorbell_range = nbio_v4_3_ih_doorbell_range,
+	.update_medium_grain_clock_gating = nbio_v4_3_update_medium_grain_clock_gating,
+	.update_medium_grain_light_sleep = nbio_v4_3_update_medium_grain_light_sleep,
+	.get_clockgating_state = nbio_v4_3_get_clockgating_state,
+	.ih_control = nbio_v4_3_ih_control,
+	.init_registers = nbio_v4_3_init_registers,
+	.remap_hdp_registers = nbio_v4_3_remap_hdp_registers,
+	.get_rom_offset = nbio_v4_3_get_rom_offset,
+	.program_aspm = nbio_v4_3_program_aspm,
+	.set_reg_remap = nbio_v4_3_set_reg_remap,
+};
+
+
+static void nbio_v4_3_sriov_ih_doorbell_range(struct amdgpu_device *adev,
+					bool use_doorbell, int doorbell_index)
+{
+}
+
+static void nbio_v4_3_sriov_sdma_doorbell_range(struct amdgpu_device *adev, int instance,
+					  bool use_doorbell, int doorbell_index,
+					  int doorbell_size)
+{
+}
+
+static void nbio_v4_3_sriov_vcn_doorbell_range(struct amdgpu_device *adev, bool use_doorbell,
+					 int doorbell_index, int instance)
+{
+}
+
+static void nbio_v4_3_sriov_gc_doorbell_init(struct amdgpu_device *adev)
+{
+}
+
+const struct amdgpu_nbio_funcs nbio_v4_3_sriov_funcs = {
+	.get_hdp_flush_req_offset = nbio_v4_3_get_hdp_flush_req_offset,
+	.get_hdp_flush_done_offset = nbio_v4_3_get_hdp_flush_done_offset,
+	.get_pcie_index_offset = nbio_v4_3_get_pcie_index_offset,
+	.get_pcie_data_offset = nbio_v4_3_get_pcie_data_offset,
+	.get_rev_id = nbio_v4_3_get_rev_id,
+	.mc_access_enable = nbio_v4_3_mc_access_enable,
+	.get_memsize = nbio_v4_3_get_memsize,
+	.sdma_doorbell_range = nbio_v4_3_sriov_sdma_doorbell_range,
+	.vcn_doorbell_range = nbio_v4_3_sriov_vcn_doorbell_range,
+	.gc_doorbell_init = nbio_v4_3_sriov_gc_doorbell_init,
+	.enable_doorbell_aperture = nbio_v4_3_enable_doorbell_aperture,
+	.enable_doorbell_selfring_aperture = nbio_v4_3_enable_doorbell_selfring_aperture,
+	.ih_doorbell_range = nbio_v4_3_sriov_ih_doorbell_range,
+	.update_medium_grain_clock_gating = nbio_v4_3_update_medium_grain_clock_gating,
+	.update_medium_grain_light_sleep = nbio_v4_3_update_medium_grain_light_sleep,
+	.get_clockgating_state = nbio_v4_3_get_clockgating_state,
+	.ih_control = nbio_v4_3_ih_control,
+	.init_registers = nbio_v4_3_init_registers,
+	.remap_hdp_registers = nbio_v4_3_remap_hdp_registers,
+	.get_rom_offset = nbio_v4_3_get_rom_offset,
+	.set_reg_remap = nbio_v4_3_set_reg_remap,
+};
+
+static int nbio_v4_3_set_ras_err_event_athub_irq_state(struct amdgpu_device *adev,
+						       struct amdgpu_irq_src *src,
+						       unsigned type,
+						       enum amdgpu_interrupt_state state)
+{
+	/* The ras_controller_irq enablement should be done in psp bl when it
+	 * tries to enable ras feature. Driver only need to set the correct interrupt
+	 * vector for bare-metal and sriov use case respectively
+	 */
+	uint32_t bif_doorbell_int_cntl;
+
+	bif_doorbell_int_cntl = RREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_DOORBELL_INT_CNTL);
+	bif_doorbell_int_cntl = REG_SET_FIELD(bif_doorbell_int_cntl,
+					      BIF_BX0_BIF_DOORBELL_INT_CNTL,
+					      RAS_ATHUB_ERR_EVENT_INTERRUPT_DISABLE,
+					      (state == AMDGPU_IRQ_STATE_ENABLE) ? 0 : 1);
+	WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_DOORBELL_INT_CNTL, bif_doorbell_int_cntl);
+
+	return 0;
+}
+
+static int nbio_v4_3_process_err_event_athub_irq(struct amdgpu_device *adev,
+						 struct amdgpu_irq_src *source,
+						 struct amdgpu_iv_entry *entry)
+{
+	/* By design, the ih cookie for err_event_athub_irq should be written
+	 * to bif ring. since bif ring is not enabled, just leave process callback
+	 * as a dummy one.
+	 */
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs nbio_v4_3_ras_err_event_athub_irq_funcs = {
+	.set = nbio_v4_3_set_ras_err_event_athub_irq_state,
+	.process = nbio_v4_3_process_err_event_athub_irq,
+};
+
+static void nbio_v4_3_handle_ras_err_event_athub_intr_no_bifring(struct amdgpu_device *adev)
+{
+	uint32_t bif_doorbell_int_cntl;
+
+	bif_doorbell_int_cntl = RREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_DOORBELL_INT_CNTL);
+	if (REG_GET_FIELD(bif_doorbell_int_cntl,
+			  BIF_DOORBELL_INT_CNTL,
+			  RAS_ATHUB_ERR_EVENT_INTERRUPT_STATUS)) {
+		/* driver has to clear the interrupt status when bif ring is disabled */
+		bif_doorbell_int_cntl = REG_SET_FIELD(bif_doorbell_int_cntl,
+						BIF_DOORBELL_INT_CNTL,
+						RAS_ATHUB_ERR_EVENT_INTERRUPT_CLEAR, 1);
+		WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_DOORBELL_INT_CNTL, bif_doorbell_int_cntl);
+		amdgpu_ras_global_ras_isr(adev);
+	}
+}
+
+static int nbio_v4_3_init_ras_err_event_athub_interrupt(struct amdgpu_device *adev)
+{
+
+	int r;
+
+	/* init the irq funcs */
+	adev->nbio.ras_err_event_athub_irq.funcs =
+		&nbio_v4_3_ras_err_event_athub_irq_funcs;
+	adev->nbio.ras_err_event_athub_irq.num_types = 1;
+
+	/* register ras err event athub interrupt
+	 * nbio v4_3 uses the same irq source as nbio v7_4 */
+	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_BIF,
+			      NBIF_7_4__SRCID__ERREVENT_ATHUB_INTERRUPT,
+			      &adev->nbio.ras_err_event_athub_irq);
+
+	return r;
+}
+
+struct amdgpu_nbio_ras nbio_v4_3_ras = {
+	.handle_ras_err_event_athub_intr_no_bifring = nbio_v4_3_handle_ras_err_event_athub_intr_no_bifring,
+	.init_ras_err_event_athub_interrupt = nbio_v4_3_init_ras_err_event_athub_interrupt,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v4_3.h b/drivers/gpu/drm/amd/amdgpu/nbio_v4_3.h
new file mode 100644
index 000000000000..399037cdf4fb
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v4_3.h
@@ -0,0 +1,34 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __NBIO_V4_3_H__
+#define __NBIO_V4_3_H__
+
+#include "soc15_common.h"
+
+extern const struct nbio_hdp_flush_reg nbio_v4_3_hdp_flush_reg;
+extern const struct amdgpu_nbio_funcs nbio_v4_3_funcs;
+extern const struct amdgpu_nbio_funcs nbio_v4_3_sriov_funcs;
+extern struct amdgpu_nbio_ras nbio_v4_3_ras;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v6_1.c b/drivers/gpu/drm/amd/amdgpu/nbio_v6_1.c
index 045988b18bc3..e911368c1aeb 100644
--- a/drivers/gpu/drm/amd/amdgpu/nbio_v6_1.c
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v6_1.c
@@ -21,22 +21,48 @@
  *
  */
 #include "amdgpu.h"
-#include "amdgpu_atombios.h"
 #include "nbio_v6_1.h"
 
-#include "vega10/soc15ip.h"
-#include "vega10/NBIO/nbio_6_1_default.h"
-#include "vega10/NBIO/nbio_6_1_offset.h"
-#include "vega10/NBIO/nbio_6_1_sh_mask.h"
-#include "vega10/vega10_enum.h"
-
-#define smnCPM_CONTROL                                                                                  0x11180460
-#define smnPCIE_CNTL2                                                                                   0x11180070
-#define smnPCIE_CONFIG_CNTL                                                                             0x11180044
+#include "nbio/nbio_6_1_default.h"
+#include "nbio/nbio_6_1_offset.h"
+#include "nbio/nbio_6_1_sh_mask.h"
+#include "nbio/nbio_6_1_smn.h"
+#include "vega10_enum.h"
+#include <uapi/linux/kfd_ioctl.h>
+
+#define smnPCIE_LC_CNTL		0x11140280
+#define smnPCIE_LC_CNTL3	0x111402d4
+#define smnPCIE_LC_CNTL6	0x111402ec
+#define smnPCIE_LC_CNTL7	0x111402f0
+#define smnNBIF_MGCG_CTRL_LCLK	0x1013a05c
+#define NBIF_MGCG_CTRL_LCLK__NBIF_MGCG_REG_DIS_LCLK_MASK	0x00001000L
+#define RCC_BIF_STRAP3__STRAP_VLINK_ASPM_IDLE_TIMER_MASK	0x0000FFFFL
+#define RCC_BIF_STRAP3__STRAP_VLINK_PM_L1_ENTRY_TIMER_MASK	0xFFFF0000L
+#define smnRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL	0x10123530
+#define smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2	0x1014008c
+#define smnBIF_CFG_DEV0_EPF0_PCIE_LTR_CAP	0x10140324
+#define smnPSWUSP0_PCIE_LC_CNTL2		0x111402c4
+#define smnRCC_BIF_STRAP2	0x10123488
+#define smnRCC_BIF_STRAP3	0x1012348c
+#define smnRCC_BIF_STRAP5	0x10123494
+#define BIF_CFG_DEV0_EPF0_DEVICE_CNTL2__LTR_EN_MASK			0x0400L
+#define RCC_BIF_STRAP5__STRAP_VLINK_LDN_ENTRY_TIMER_MASK	0x0000FFFFL
+#define RCC_BIF_STRAP2__STRAP_LTR_IN_ASPML1_DIS_MASK	0x00004000L
+#define RCC_BIF_STRAP3__STRAP_VLINK_ASPM_IDLE_TIMER__SHIFT	0x0
+#define RCC_BIF_STRAP3__STRAP_VLINK_PM_L1_ENTRY_TIMER__SHIFT	0x10
+#define RCC_BIF_STRAP5__STRAP_VLINK_LDN_ENTRY_TIMER__SHIFT	0x0
+
+static void nbio_v6_1_remap_hdp_registers(struct amdgpu_device *adev)
+{
+	WREG32_SOC15(NBIO, 0, mmREMAP_HDP_MEM_FLUSH_CNTL,
+		adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL);
+	WREG32_SOC15(NBIO, 0, mmREMAP_HDP_REG_FLUSH_CNTL,
+		adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_REG_FLUSH_CNTL);
+}
 
-u32 nbio_v6_1_get_rev_id(struct amdgpu_device *adev)
+static u32 nbio_v6_1_get_rev_id(struct amdgpu_device *adev)
 {
-        u32 tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0);
+	u32 tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0);
 
 	tmp &= RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0_MASK;
 	tmp >>= RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0__SHIFT;
@@ -44,19 +70,7 @@ u32 nbio_v6_1_get_rev_id(struct amdgpu_device *adev)
 	return tmp;
 }
 
-u32 nbio_v6_1_get_atombios_scratch_regs(struct amdgpu_device *adev,
-					uint32_t idx)
-{
-	return RREG32_SOC15_OFFSET(NBIO, 0, mmBIOS_SCRATCH_0, idx);
-}
-
-void nbio_v6_1_set_atombios_scratch_regs(struct amdgpu_device *adev,
-					 uint32_t idx, uint32_t val)
-{
-	WREG32_SOC15_OFFSET(NBIO, 0, mmBIOS_SCRATCH_0, idx, val);
-}
-
-void nbio_v6_1_mc_access_enable(struct amdgpu_device *adev, bool enable)
+static void nbio_v6_1_mc_access_enable(struct amdgpu_device *adev, bool enable)
 {
 	if (enable)
 		WREG32_SOC15(NBIO, 0, mmBIF_FB_EN,
@@ -66,44 +80,37 @@ void nbio_v6_1_mc_access_enable(struct amdgpu_device *adev, bool enable)
 		WREG32_SOC15(NBIO, 0, mmBIF_FB_EN, 0);
 }
 
-void nbio_v6_1_hdp_flush(struct amdgpu_device *adev)
-{
-	WREG32_SOC15_NO_KIQ(NBIO, 0, mmBIF_BX_PF0_HDP_MEM_COHERENCY_FLUSH_CNTL, 0);
-}
-
-u32 nbio_v6_1_get_memsize(struct amdgpu_device *adev)
+static u32 nbio_v6_1_get_memsize(struct amdgpu_device *adev)
 {
 	return RREG32_SOC15(NBIO, 0, mmRCC_PF_0_0_RCC_CONFIG_MEMSIZE);
 }
 
-static const u32 nbio_sdma_doorbell_range_reg[] =
+static void nbio_v6_1_sdma_doorbell_range(struct amdgpu_device *adev, int instance,
+			bool use_doorbell, int doorbell_index, int doorbell_size)
 {
-	SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA0_DOORBELL_RANGE),
-	SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA1_DOORBELL_RANGE)
-};
+	u32 reg = instance == 0 ? SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA0_DOORBELL_RANGE) :
+			SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA1_DOORBELL_RANGE);
 
-void nbio_v6_1_sdma_doorbell_range(struct amdgpu_device *adev, int instance,
-				  bool use_doorbell, int doorbell_index)
-{
-	u32 doorbell_range = RREG32(nbio_sdma_doorbell_range_reg[instance]);
+	u32 doorbell_range = RREG32(reg);
 
 	if (use_doorbell) {
 		doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, OFFSET, doorbell_index);
-		doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, 2);
+		doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, doorbell_size);
 	} else
 		doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, 0);
 
-	WREG32(nbio_sdma_doorbell_range_reg[instance], doorbell_range);
+	WREG32(reg, doorbell_range);
+
 }
 
-void nbio_v6_1_enable_doorbell_aperture(struct amdgpu_device *adev,
-					bool enable)
+static void nbio_v6_1_enable_doorbell_aperture(struct amdgpu_device *adev,
+					       bool enable)
 {
 	WREG32_FIELD15(NBIO, 0, RCC_PF_0_0_RCC_DOORBELL_APER_EN, BIF_DOORBELL_APER_EN, enable ? 1 : 0);
 }
 
-void nbio_v6_1_enable_doorbell_selfring_aperture(struct amdgpu_device *adev,
-					bool enable)
+static void nbio_v6_1_enable_doorbell_selfring_aperture(struct amdgpu_device *adev,
+							bool enable)
 {
 	u32 tmp = 0;
 
@@ -122,26 +129,27 @@ void nbio_v6_1_enable_doorbell_selfring_aperture(struct amdgpu_device *adev,
 }
 
 
-void nbio_v6_1_ih_doorbell_range(struct amdgpu_device *adev,
-				bool use_doorbell, int doorbell_index)
+static void nbio_v6_1_ih_doorbell_range(struct amdgpu_device *adev,
+					bool use_doorbell, int doorbell_index)
 {
-	u32 ih_doorbell_range = RREG32_SOC15(NBIO, 0 , mmBIF_IH_DOORBELL_RANGE);
+	u32 ih_doorbell_range = RREG32_SOC15(NBIO, 0, mmBIF_IH_DOORBELL_RANGE);
 
 	if (use_doorbell) {
 		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, OFFSET, doorbell_index);
-		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, SIZE, 2);
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  BIF_IH_DOORBELL_RANGE, SIZE, 6);
 	} else
 		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, SIZE, 0);
 
 	WREG32_SOC15(NBIO, 0, mmBIF_IH_DOORBELL_RANGE, ih_doorbell_range);
 }
 
-void nbio_v6_1_ih_control(struct amdgpu_device *adev)
+static void nbio_v6_1_ih_control(struct amdgpu_device *adev)
 {
 	u32 interrupt_cntl;
 
 	/* setup interrupt control */
-	WREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL2, adev->dummy_page.addr >> 8);
+	WREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL2, adev->dummy_page_addr >> 8);
 	interrupt_cntl = RREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL);
 	/* INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
 	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
@@ -152,8 +160,8 @@ void nbio_v6_1_ih_control(struct amdgpu_device *adev)
 	WREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL, interrupt_cntl);
 }
 
-void nbio_v6_1_update_medium_grain_clock_gating(struct amdgpu_device *adev,
-						bool enable)
+static void nbio_v6_1_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						       bool enable)
 {
 	uint32_t def, data;
 
@@ -180,8 +188,8 @@ void nbio_v6_1_update_medium_grain_clock_gating(struct amdgpu_device *adev,
 		WREG32_PCIE(smnCPM_CONTROL, data);
 }
 
-void nbio_v6_1_update_medium_grain_light_sleep(struct amdgpu_device *adev,
-					       bool enable)
+static void nbio_v6_1_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+						      bool enable)
 {
 	uint32_t def, data;
 
@@ -200,7 +208,8 @@ void nbio_v6_1_update_medium_grain_light_sleep(struct amdgpu_device *adev,
 		WREG32_PCIE(smnPCIE_CNTL2, data);
 }
 
-void nbio_v6_1_get_clockgating_state(struct amdgpu_device *adev, u32 *flags)
+static void nbio_v6_1_get_clockgating_state(struct amdgpu_device *adev,
+					    u64 *flags)
 {
 	int data;
 
@@ -215,50 +224,42 @@ void nbio_v6_1_get_clockgating_state(struct amdgpu_device *adev, u32 *flags)
 		*flags |= AMD_CG_SUPPORT_BIF_LS;
 }
 
-struct nbio_hdp_flush_reg nbio_v6_1_hdp_flush_reg;
-struct nbio_pcie_index_data nbio_v6_1_pcie_index_data;
-
-int nbio_v6_1_init(struct amdgpu_device *adev)
+static u32 nbio_v6_1_get_hdp_flush_req_offset(struct amdgpu_device *adev)
 {
-	nbio_v6_1_hdp_flush_reg.hdp_flush_req_offset = SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_GPU_HDP_FLUSH_REQ);
-	nbio_v6_1_hdp_flush_reg.hdp_flush_done_offset = SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_GPU_HDP_FLUSH_DONE);
-	nbio_v6_1_hdp_flush_reg.ref_and_mask_cp0 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP0_MASK;
-	nbio_v6_1_hdp_flush_reg.ref_and_mask_cp1 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP1_MASK;
-	nbio_v6_1_hdp_flush_reg.ref_and_mask_cp2 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP2_MASK;
-	nbio_v6_1_hdp_flush_reg.ref_and_mask_cp3 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP3_MASK;
-	nbio_v6_1_hdp_flush_reg.ref_and_mask_cp4 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP4_MASK;
-	nbio_v6_1_hdp_flush_reg.ref_and_mask_cp5 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP5_MASK;
-	nbio_v6_1_hdp_flush_reg.ref_and_mask_cp6 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP6_MASK;
-	nbio_v6_1_hdp_flush_reg.ref_and_mask_cp7 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP7_MASK;
-	nbio_v6_1_hdp_flush_reg.ref_and_mask_cp8 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP8_MASK;
-	nbio_v6_1_hdp_flush_reg.ref_and_mask_cp9 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP9_MASK;
-	nbio_v6_1_hdp_flush_reg.ref_and_mask_sdma0 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__SDMA0_MASK;
-	nbio_v6_1_hdp_flush_reg.ref_and_mask_sdma1 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__SDMA1_MASK;
-
-	nbio_v6_1_pcie_index_data.index_offset = SOC15_REG_OFFSET(NBIO, 0, mmPCIE_INDEX);
-	nbio_v6_1_pcie_index_data.data_offset = SOC15_REG_OFFSET(NBIO, 0, mmPCIE_DATA);
-
-	return 0;
+	return SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_GPU_HDP_FLUSH_REQ);
 }
 
-void nbio_v6_1_detect_hw_virt(struct amdgpu_device *adev)
+static u32 nbio_v6_1_get_hdp_flush_done_offset(struct amdgpu_device *adev)
 {
-	uint32_t reg;
-
-	reg = RREG32_SOC15(NBIO, 0, mmRCC_PF_0_0_RCC_IOV_FUNC_IDENTIFIER);
-	if (reg & 1)
-		adev->virt.caps |= AMDGPU_SRIOV_CAPS_IS_VF;
+	return SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_GPU_HDP_FLUSH_DONE);
+}
 
-	if (reg & 0x80000000)
-		adev->virt.caps |= AMDGPU_SRIOV_CAPS_ENABLE_IOV;
+static u32 nbio_v6_1_get_pcie_index_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, mmPCIE_INDEX2);
+}
 
-	if (!reg) {
-		if (is_virtual_machine())	/* passthrough mode exclus sriov mod */
-			adev->virt.caps |= AMDGPU_PASSTHROUGH_MODE;
-	}
+static u32 nbio_v6_1_get_pcie_data_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, mmPCIE_DATA2);
 }
 
-void nbio_v6_1_init_registers(struct amdgpu_device *adev)
+const struct nbio_hdp_flush_reg nbio_v6_1_hdp_flush_reg = {
+	.ref_and_mask_cp0 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP0_MASK,
+	.ref_and_mask_cp1 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP1_MASK,
+	.ref_and_mask_cp2 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP2_MASK,
+	.ref_and_mask_cp3 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP3_MASK,
+	.ref_and_mask_cp4 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP4_MASK,
+	.ref_and_mask_cp5 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP5_MASK,
+	.ref_and_mask_cp6 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP6_MASK,
+	.ref_and_mask_cp7 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP7_MASK,
+	.ref_and_mask_cp8 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP8_MASK,
+	.ref_and_mask_cp9 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP9_MASK,
+	.ref_and_mask_sdma0 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__SDMA0_MASK,
+	.ref_and_mask_sdma1 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__SDMA1_MASK
+};
+
+static void nbio_v6_1_init_registers(struct amdgpu_device *adev)
 {
 	uint32_t def, data;
 
@@ -268,4 +269,159 @@ void nbio_v6_1_init_registers(struct amdgpu_device *adev)
 
 	if (def != data)
 		WREG32_PCIE(smnPCIE_CONFIG_CNTL, data);
+
+	def = data = RREG32_PCIE(smnPCIE_CI_CNTL);
+	data = REG_SET_FIELD(data, PCIE_CI_CNTL, CI_SLV_ORDERING_DIS, 1);
+
+	if (def != data)
+		WREG32_PCIE(smnPCIE_CI_CNTL, data);
+}
+
+#ifdef CONFIG_PCIEASPM
+static void nbio_v6_1_program_ltr(struct amdgpu_device *adev)
+{
+	uint32_t def, data;
+
+	WREG32_PCIE(smnRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL, 0x75EB);
+
+	def = data = RREG32_PCIE(smnRCC_BIF_STRAP2);
+	data &= ~RCC_BIF_STRAP2__STRAP_LTR_IN_ASPML1_DIS_MASK;
+	if (def != data)
+		WREG32_PCIE(smnRCC_BIF_STRAP2, data);
+
+	def = data = RREG32_PCIE(smnRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL);
+	data &= ~EP_PCIE_TX_LTR_CNTL__LTR_PRIV_MSG_DIS_IN_PM_NON_D0_MASK;
+	if (def != data)
+		WREG32_PCIE(smnRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL, data);
+
+	def = data = RREG32_PCIE(smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2);
+	data |= BIF_CFG_DEV0_EPF0_DEVICE_CNTL2__LTR_EN_MASK;
+	if (def != data)
+		WREG32_PCIE(smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2, data);
 }
+#endif
+
+static void nbio_v6_1_program_aspm(struct amdgpu_device *adev)
+{
+#ifdef CONFIG_PCIEASPM
+	uint32_t def, data;
+
+	def = data = RREG32_PCIE(smnPCIE_LC_CNTL);
+	data &= ~PCIE_LC_CNTL__LC_L1_INACTIVITY_MASK;
+	data &= ~PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK;
+	data |= PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK;
+	if (def != data)
+		WREG32_PCIE(smnPCIE_LC_CNTL, data);
+
+	def = data = RREG32_PCIE(smnPCIE_LC_CNTL7);
+	data |= PCIE_LC_CNTL7__LC_NBIF_ASPM_INPUT_EN_MASK;
+	if (def != data)
+		WREG32_PCIE(smnPCIE_LC_CNTL7, data);
+
+	def = data = RREG32_PCIE(smnNBIF_MGCG_CTRL_LCLK);
+	data |= NBIF_MGCG_CTRL_LCLK__NBIF_MGCG_REG_DIS_LCLK_MASK;
+	if (def != data)
+		WREG32_PCIE(smnNBIF_MGCG_CTRL_LCLK, data);
+
+	def = data = RREG32_PCIE(smnPCIE_LC_CNTL3);
+	data |= PCIE_LC_CNTL3__LC_DSC_DONT_ENTER_L23_AFTER_PME_ACK_MASK;
+	if (def != data)
+		WREG32_PCIE(smnPCIE_LC_CNTL3, data);
+
+	def = data = RREG32_PCIE(smnRCC_BIF_STRAP3);
+	data &= ~RCC_BIF_STRAP3__STRAP_VLINK_ASPM_IDLE_TIMER_MASK;
+	data &= ~RCC_BIF_STRAP3__STRAP_VLINK_PM_L1_ENTRY_TIMER_MASK;
+	if (def != data)
+		WREG32_PCIE(smnRCC_BIF_STRAP3, data);
+
+	def = data = RREG32_PCIE(smnRCC_BIF_STRAP5);
+	data &= ~RCC_BIF_STRAP5__STRAP_VLINK_LDN_ENTRY_TIMER_MASK;
+	if (def != data)
+		WREG32_PCIE(smnRCC_BIF_STRAP5, data);
+
+	def = data = RREG32_PCIE(smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2);
+	data &= ~BIF_CFG_DEV0_EPF0_DEVICE_CNTL2__LTR_EN_MASK;
+	if (def != data)
+		WREG32_PCIE(smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2, data);
+
+	WREG32_PCIE(smnBIF_CFG_DEV0_EPF0_PCIE_LTR_CAP, 0x10011001);
+
+	def = data = RREG32_PCIE(smnPSWUSP0_PCIE_LC_CNTL2);
+	data |= PSWUSP0_PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L1_MASK |
+		PSWUSP0_PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L23_MASK;
+	data &= ~PSWUSP0_PCIE_LC_CNTL2__LC_RCV_L0_TO_RCV_L0S_DIS_MASK;
+	if (def != data)
+		WREG32_PCIE(smnPSWUSP0_PCIE_LC_CNTL2, data);
+
+	def = data = RREG32_PCIE(smnPCIE_LC_CNTL6);
+	data |= PCIE_LC_CNTL6__LC_L1_POWERDOWN_MASK |
+		PCIE_LC_CNTL6__LC_RX_L0S_STANDBY_EN_MASK;
+	if (def != data)
+		WREG32_PCIE(smnPCIE_LC_CNTL6, data);
+
+	/* Don't bother about LTR if LTR is not enabled
+	 * in the path */
+	if (adev->pdev->ltr_path)
+		nbio_v6_1_program_ltr(adev);
+
+	def = data = RREG32_PCIE(smnRCC_BIF_STRAP3);
+	data |= 0x5DE0 << RCC_BIF_STRAP3__STRAP_VLINK_ASPM_IDLE_TIMER__SHIFT;
+	data |= 0x0010 << RCC_BIF_STRAP3__STRAP_VLINK_PM_L1_ENTRY_TIMER__SHIFT;
+	if (def != data)
+		WREG32_PCIE(smnRCC_BIF_STRAP3, data);
+
+	def = data = RREG32_PCIE(smnRCC_BIF_STRAP5);
+	data |= 0x0010 << RCC_BIF_STRAP5__STRAP_VLINK_LDN_ENTRY_TIMER__SHIFT;
+	if (def != data)
+		WREG32_PCIE(smnRCC_BIF_STRAP5, data);
+
+	def = data = RREG32_PCIE(smnPCIE_LC_CNTL);
+	data &= ~PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK;
+	data |= 0x9 << PCIE_LC_CNTL__LC_L1_INACTIVITY__SHIFT;
+	data |= 0x1 << PCIE_LC_CNTL__LC_PMI_TO_L1_DIS__SHIFT;
+	if (def != data)
+		WREG32_PCIE(smnPCIE_LC_CNTL, data);
+
+	def = data = RREG32_PCIE(smnPCIE_LC_CNTL3);
+	data &= ~PCIE_LC_CNTL3__LC_DSC_DONT_ENTER_L23_AFTER_PME_ACK_MASK;
+	if (def != data)
+		WREG32_PCIE(smnPCIE_LC_CNTL3, data);
+#endif
+}
+
+#define MMIO_REG_HOLE_OFFSET (0x80000 - PAGE_SIZE)
+
+static void nbio_v6_1_set_reg_remap(struct amdgpu_device *adev)
+{
+	if (!amdgpu_sriov_vf(adev) && (PAGE_SIZE <= 4096)) {
+		adev->rmmio_remap.reg_offset = MMIO_REG_HOLE_OFFSET;
+		adev->rmmio_remap.bus_addr = adev->rmmio_base + MMIO_REG_HOLE_OFFSET;
+	} else {
+		adev->rmmio_remap.reg_offset =
+			SOC15_REG_OFFSET(NBIO, 0,
+					 mmBIF_BX_DEV0_EPF0_VF0_HDP_MEM_COHERENCY_FLUSH_CNTL) << 2;
+		adev->rmmio_remap.bus_addr = 0;
+	}
+}
+
+const struct amdgpu_nbio_funcs nbio_v6_1_funcs = {
+	.get_hdp_flush_req_offset = nbio_v6_1_get_hdp_flush_req_offset,
+	.get_hdp_flush_done_offset = nbio_v6_1_get_hdp_flush_done_offset,
+	.get_pcie_index_offset = nbio_v6_1_get_pcie_index_offset,
+	.get_pcie_data_offset = nbio_v6_1_get_pcie_data_offset,
+	.get_rev_id = nbio_v6_1_get_rev_id,
+	.mc_access_enable = nbio_v6_1_mc_access_enable,
+	.get_memsize = nbio_v6_1_get_memsize,
+	.sdma_doorbell_range = nbio_v6_1_sdma_doorbell_range,
+	.enable_doorbell_aperture = nbio_v6_1_enable_doorbell_aperture,
+	.enable_doorbell_selfring_aperture = nbio_v6_1_enable_doorbell_selfring_aperture,
+	.ih_doorbell_range = nbio_v6_1_ih_doorbell_range,
+	.update_medium_grain_clock_gating = nbio_v6_1_update_medium_grain_clock_gating,
+	.update_medium_grain_light_sleep = nbio_v6_1_update_medium_grain_light_sleep,
+	.get_clockgating_state = nbio_v6_1_get_clockgating_state,
+	.ih_control = nbio_v6_1_ih_control,
+	.init_registers = nbio_v6_1_init_registers,
+	.remap_hdp_registers = nbio_v6_1_remap_hdp_registers,
+	.program_aspm = nbio_v6_1_program_aspm,
+	.set_reg_remap = nbio_v6_1_set_reg_remap,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v6_1.h b/drivers/gpu/drm/amd/amdgpu/nbio_v6_1.h
index 686e4b4d296a..6dc743b73218 100644
--- a/drivers/gpu/drm/amd/amdgpu/nbio_v6_1.h
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v6_1.h
@@ -26,30 +26,7 @@
 
 #include "soc15_common.h"
 
-extern struct nbio_hdp_flush_reg nbio_v6_1_hdp_flush_reg;
-extern struct nbio_pcie_index_data nbio_v6_1_pcie_index_data;
-int nbio_v6_1_init(struct amdgpu_device *adev);
-u32 nbio_v6_1_get_atombios_scratch_regs(struct amdgpu_device *adev,
-                                        uint32_t idx);
-void nbio_v6_1_set_atombios_scratch_regs(struct amdgpu_device *adev,
-                                         uint32_t idx, uint32_t val);
-void nbio_v6_1_mc_access_enable(struct amdgpu_device *adev, bool enable);
-void nbio_v6_1_hdp_flush(struct amdgpu_device *adev);
-u32 nbio_v6_1_get_memsize(struct amdgpu_device *adev);
-void nbio_v6_1_sdma_doorbell_range(struct amdgpu_device *adev, int instance,
-				  bool use_doorbell, int doorbell_index);
-void nbio_v6_1_enable_doorbell_aperture(struct amdgpu_device *adev,
-					bool enable);
-void nbio_v6_1_enable_doorbell_selfring_aperture(struct amdgpu_device *adev,
-					bool enable);
-void nbio_v6_1_ih_doorbell_range(struct amdgpu_device *adev,
-				bool use_doorbell, int doorbell_index);
-void nbio_v6_1_ih_control(struct amdgpu_device *adev);
-u32 nbio_v6_1_get_rev_id(struct amdgpu_device *adev);
-void nbio_v6_1_update_medium_grain_clock_gating(struct amdgpu_device *adev, bool enable);
-void nbio_v6_1_update_medium_grain_light_sleep(struct amdgpu_device *adev, bool enable);
-void nbio_v6_1_get_clockgating_state(struct amdgpu_device *adev, u32 *flags);
-void nbio_v6_1_detect_hw_virt(struct amdgpu_device *adev);
-void nbio_v6_1_init_registers(struct amdgpu_device *adev);
+extern const struct nbio_hdp_flush_reg nbio_v6_1_hdp_flush_reg;
+extern const struct amdgpu_nbio_funcs nbio_v6_1_funcs;
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_0.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_0.c
index 11b70d601922..1569a1e934ec 100644
--- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_0.c
@@ -21,20 +21,28 @@
  *
  */
 #include "amdgpu.h"
-#include "amdgpu_atombios.h"
 #include "nbio_v7_0.h"
 
-#include "vega10/soc15ip.h"
-#include "raven1/NBIO/nbio_7_0_default.h"
-#include "raven1/NBIO/nbio_7_0_offset.h"
-#include "raven1/NBIO/nbio_7_0_sh_mask.h"
-#include "vega10/vega10_enum.h"
+#include "nbio/nbio_7_0_default.h"
+#include "nbio/nbio_7_0_offset.h"
+#include "nbio/nbio_7_0_sh_mask.h"
+#include "nbio/nbio_7_0_smn.h"
+#include "vega10_enum.h"
+#include <uapi/linux/kfd_ioctl.h>
 
 #define smnNBIF_MGCG_CTRL_LCLK	0x1013a05c
 
-u32 nbio_v7_0_get_rev_id(struct amdgpu_device *adev)
+static void nbio_v7_0_remap_hdp_registers(struct amdgpu_device *adev)
 {
-        u32 tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0);
+	WREG32_SOC15(NBIO, 0, mmREMAP_HDP_MEM_FLUSH_CNTL,
+		adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL);
+	WREG32_SOC15(NBIO, 0, mmREMAP_HDP_REG_FLUSH_CNTL,
+		adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_REG_FLUSH_CNTL);
+}
+
+static u32 nbio_v7_0_get_rev_id(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0);
 
 	tmp &= RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0_MASK;
 	tmp >>= RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0__SHIFT;
@@ -42,19 +50,7 @@ u32 nbio_v7_0_get_rev_id(struct amdgpu_device *adev)
 	return tmp;
 }
 
-u32 nbio_v7_0_get_atombios_scratch_regs(struct amdgpu_device *adev,
-					uint32_t idx)
-{
-	return RREG32_SOC15_OFFSET(NBIO, 0, mmBIOS_SCRATCH_0, idx);
-}
-
-void nbio_v7_0_set_atombios_scratch_regs(struct amdgpu_device *adev,
-					 uint32_t idx, uint32_t val)
-{
-	WREG32_SOC15_OFFSET(NBIO, 0, mmBIOS_SCRATCH_0, idx, val);
-}
-
-void nbio_v7_0_mc_access_enable(struct amdgpu_device *adev, bool enable)
+static void nbio_v7_0_mc_access_enable(struct amdgpu_device *adev, bool enable)
 {
 	if (enable)
 		WREG32_SOC15(NBIO, 0, mmBIF_FB_EN,
@@ -63,46 +59,64 @@ void nbio_v7_0_mc_access_enable(struct amdgpu_device *adev, bool enable)
 		WREG32_SOC15(NBIO, 0, mmBIF_FB_EN, 0);
 }
 
-void nbio_v7_0_hdp_flush(struct amdgpu_device *adev)
-{
-	WREG32_SOC15_NO_KIQ(NBIO, 0, mmHDP_MEM_COHERENCY_FLUSH_CNTL, 0);
-}
-
-u32 nbio_v7_0_get_memsize(struct amdgpu_device *adev)
+static u32 nbio_v7_0_get_memsize(struct amdgpu_device *adev)
 {
 	return RREG32_SOC15(NBIO, 0, mmRCC_CONFIG_MEMSIZE);
 }
 
-static const u32 nbio_sdma_doorbell_range_reg[] =
+static void nbio_v7_0_sdma_doorbell_range(struct amdgpu_device *adev, int instance,
+			bool use_doorbell, int doorbell_index, int doorbell_size)
 {
-	SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA0_DOORBELL_RANGE),
-	SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA1_DOORBELL_RANGE)
-};
+	u32 reg = instance == 0 ? SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA0_DOORBELL_RANGE) :
+			SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA1_DOORBELL_RANGE);
 
-void nbio_v7_0_sdma_doorbell_range(struct amdgpu_device *adev, int instance,
-				  bool use_doorbell, int doorbell_index)
-{
-	u32 doorbell_range = RREG32(nbio_sdma_doorbell_range_reg[instance]);
+	u32 doorbell_range = RREG32(reg);
 
 	if (use_doorbell) {
 		doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, OFFSET, doorbell_index);
-		doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, 2);
+		doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, doorbell_size);
 	} else
 		doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, 0);
 
-	WREG32(nbio_sdma_doorbell_range_reg[instance], doorbell_range);
+	WREG32(reg, doorbell_range);
 }
 
-void nbio_v7_0_enable_doorbell_aperture(struct amdgpu_device *adev,
-					bool enable)
+static void nbio_v7_0_vcn_doorbell_range(struct amdgpu_device *adev, bool use_doorbell,
+					 int doorbell_index, int instance)
+{
+	u32 reg = SOC15_REG_OFFSET(NBIO, 0, mmBIF_MMSCH0_DOORBELL_RANGE);
+
+	u32 doorbell_range = RREG32(reg);
+
+	if (use_doorbell) {
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       BIF_MMSCH0_DOORBELL_RANGE, OFFSET,
+					       doorbell_index);
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       BIF_MMSCH0_DOORBELL_RANGE, SIZE, 8);
+	} else
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       BIF_MMSCH0_DOORBELL_RANGE, SIZE, 0);
+
+	WREG32(reg, doorbell_range);
+}
+
+static void nbio_v7_0_enable_doorbell_aperture(struct amdgpu_device *adev,
+					       bool enable)
 {
 	WREG32_FIELD15(NBIO, 0, RCC_DOORBELL_APER_EN, BIF_DOORBELL_APER_EN, enable ? 1 : 0);
 }
 
-void nbio_v7_0_ih_doorbell_range(struct amdgpu_device *adev,
-				bool use_doorbell, int doorbell_index)
+static void nbio_v7_0_enable_doorbell_selfring_aperture(struct amdgpu_device *adev,
+							bool enable)
 {
-	u32 ih_doorbell_range = RREG32_SOC15(NBIO, 0 , mmBIF_IH_DOORBELL_RANGE);
+
+}
+
+static void nbio_v7_0_ih_doorbell_range(struct amdgpu_device *adev,
+					bool use_doorbell, int doorbell_index)
+{
+	u32 ih_doorbell_range = RREG32_SOC15(NBIO, 0, mmBIF_IH_DOORBELL_RANGE);
 
 	if (use_doorbell) {
 		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, OFFSET, doorbell_index);
@@ -130,8 +144,8 @@ static void nbio_7_0_write_syshub_ind_mmr(struct amdgpu_device *adev, uint32_t o
 	WREG32_SOC15(NBIO, 0, mmSYSHUB_DATA, data);
 }
 
-void nbio_v7_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
-						bool enable)
+static void nbio_v7_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						       bool enable)
 {
 	uint32_t def, data;
 
@@ -169,12 +183,48 @@ void nbio_v7_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
 		nbio_7_0_write_syshub_ind_mmr(adev, ixSYSHUB_MMREG_IND_SYSHUB_MGCG_CTRL_SHUBCLK, data);
 }
 
-void nbio_v7_0_ih_control(struct amdgpu_device *adev)
+static void nbio_v7_0_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+						      bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_PCIE(smnPCIE_CNTL2);
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_BIF_LS)) {
+		data |= (PCIE_CNTL2__SLV_MEM_LS_EN_MASK |
+			 PCIE_CNTL2__MST_MEM_LS_EN_MASK |
+			 PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK);
+	} else {
+		data &= ~(PCIE_CNTL2__SLV_MEM_LS_EN_MASK |
+			  PCIE_CNTL2__MST_MEM_LS_EN_MASK |
+			  PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK);
+	}
+
+	if (def != data)
+		WREG32_PCIE(smnPCIE_CNTL2, data);
+}
+
+static void nbio_v7_0_get_clockgating_state(struct amdgpu_device *adev,
+					    u64 *flags)
+{
+	int data;
+
+	/* AMD_CG_SUPPORT_BIF_MGCG */
+	data = RREG32_PCIE(smnCPM_CONTROL);
+	if (data & CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_BIF_MGCG;
+
+	/* AMD_CG_SUPPORT_BIF_LS */
+	data = RREG32_PCIE(smnPCIE_CNTL2);
+	if (data & PCIE_CNTL2__SLV_MEM_LS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_BIF_LS;
+}
+
+static void nbio_v7_0_ih_control(struct amdgpu_device *adev)
 {
 	u32 interrupt_cntl;
 
 	/* setup interrupt control */
-	WREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL2, adev->dummy_page.addr >> 8);
+	WREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL2, adev->dummy_page_addr >> 8);
 	interrupt_cntl = RREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL);
 	/* INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
 	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
@@ -185,28 +235,88 @@ void nbio_v7_0_ih_control(struct amdgpu_device *adev)
 	WREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL, interrupt_cntl);
 }
 
-struct nbio_hdp_flush_reg nbio_v7_0_hdp_flush_reg;
-struct nbio_pcie_index_data nbio_v7_0_pcie_index_data;
+static u32 nbio_v7_0_get_hdp_flush_req_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, mmGPU_HDP_FLUSH_REQ);
+}
 
-int nbio_v7_0_init(struct amdgpu_device *adev)
+static u32 nbio_v7_0_get_hdp_flush_done_offset(struct amdgpu_device *adev)
 {
-	nbio_v7_0_hdp_flush_reg.hdp_flush_req_offset = SOC15_REG_OFFSET(NBIO, 0, mmGPU_HDP_FLUSH_REQ);
-	nbio_v7_0_hdp_flush_reg.hdp_flush_done_offset = SOC15_REG_OFFSET(NBIO, 0, mmGPU_HDP_FLUSH_DONE);
-	nbio_v7_0_hdp_flush_reg.ref_and_mask_cp0 = GPU_HDP_FLUSH_DONE__CP0_MASK;
-	nbio_v7_0_hdp_flush_reg.ref_and_mask_cp1 = GPU_HDP_FLUSH_DONE__CP1_MASK;
-	nbio_v7_0_hdp_flush_reg.ref_and_mask_cp2 = GPU_HDP_FLUSH_DONE__CP2_MASK;
-	nbio_v7_0_hdp_flush_reg.ref_and_mask_cp3 = GPU_HDP_FLUSH_DONE__CP3_MASK;
-	nbio_v7_0_hdp_flush_reg.ref_and_mask_cp4 = GPU_HDP_FLUSH_DONE__CP4_MASK;
-	nbio_v7_0_hdp_flush_reg.ref_and_mask_cp5 = GPU_HDP_FLUSH_DONE__CP5_MASK;
-	nbio_v7_0_hdp_flush_reg.ref_and_mask_cp6 = GPU_HDP_FLUSH_DONE__CP6_MASK;
-	nbio_v7_0_hdp_flush_reg.ref_and_mask_cp7 = GPU_HDP_FLUSH_DONE__CP7_MASK;
-	nbio_v7_0_hdp_flush_reg.ref_and_mask_cp8 = GPU_HDP_FLUSH_DONE__CP8_MASK;
-	nbio_v7_0_hdp_flush_reg.ref_and_mask_cp9 = GPU_HDP_FLUSH_DONE__CP9_MASK;
-	nbio_v7_0_hdp_flush_reg.ref_and_mask_sdma0 = GPU_HDP_FLUSH_DONE__SDMA0_MASK;
-	nbio_v7_0_hdp_flush_reg.ref_and_mask_sdma1 = GPU_HDP_FLUSH_DONE__SDMA1_MASK;
+	return SOC15_REG_OFFSET(NBIO, 0, mmGPU_HDP_FLUSH_DONE);
+}
 
-	nbio_v7_0_pcie_index_data.index_offset = SOC15_REG_OFFSET(NBIO, 0, mmPCIE_INDEX2);
-	nbio_v7_0_pcie_index_data.data_offset = SOC15_REG_OFFSET(NBIO, 0, mmPCIE_DATA2);
+static u32 nbio_v7_0_get_pcie_index_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, mmPCIE_INDEX2);
+}
 
-	return 0;
+static u32 nbio_v7_0_get_pcie_data_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, mmPCIE_DATA2);
 }
+
+const struct nbio_hdp_flush_reg nbio_v7_0_hdp_flush_reg = {
+	.ref_and_mask_cp0 = GPU_HDP_FLUSH_DONE__CP0_MASK,
+	.ref_and_mask_cp1 = GPU_HDP_FLUSH_DONE__CP1_MASK,
+	.ref_and_mask_cp2 = GPU_HDP_FLUSH_DONE__CP2_MASK,
+	.ref_and_mask_cp3 = GPU_HDP_FLUSH_DONE__CP3_MASK,
+	.ref_and_mask_cp4 = GPU_HDP_FLUSH_DONE__CP4_MASK,
+	.ref_and_mask_cp5 = GPU_HDP_FLUSH_DONE__CP5_MASK,
+	.ref_and_mask_cp6 = GPU_HDP_FLUSH_DONE__CP6_MASK,
+	.ref_and_mask_cp7 = GPU_HDP_FLUSH_DONE__CP7_MASK,
+	.ref_and_mask_cp8 = GPU_HDP_FLUSH_DONE__CP8_MASK,
+	.ref_and_mask_cp9 = GPU_HDP_FLUSH_DONE__CP9_MASK,
+	.ref_and_mask_sdma0 = GPU_HDP_FLUSH_DONE__SDMA0_MASK,
+	.ref_and_mask_sdma1 = GPU_HDP_FLUSH_DONE__SDMA1_MASK,
+};
+
+#define regRCC_DEV0_EPF6_STRAP4                                                                         0xd304
+#define regRCC_DEV0_EPF6_STRAP4_BASE_IDX                                                                5
+
+static void nbio_v7_0_init_registers(struct amdgpu_device *adev)
+{
+	uint32_t data;
+
+	switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) {
+	case IP_VERSION(2, 5, 0):
+		data = RREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF6_STRAP4) & ~BIT(23);
+		WREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF6_STRAP4, data);
+		break;
+	}
+}
+
+#define MMIO_REG_HOLE_OFFSET (0x80000 - PAGE_SIZE)
+
+static void nbio_v7_0_set_reg_remap(struct amdgpu_device *adev)
+{
+	if (!amdgpu_sriov_vf(adev) && (PAGE_SIZE <= 4096)) {
+		adev->rmmio_remap.reg_offset = MMIO_REG_HOLE_OFFSET;
+		adev->rmmio_remap.bus_addr = adev->rmmio_base + MMIO_REG_HOLE_OFFSET;
+	} else {
+		adev->rmmio_remap.reg_offset =
+			SOC15_REG_OFFSET(NBIO, 0, mmHDP_MEM_COHERENCY_FLUSH_CNTL) << 2;
+		adev->rmmio_remap.bus_addr = 0;
+	}
+}
+
+const struct amdgpu_nbio_funcs nbio_v7_0_funcs = {
+	.get_hdp_flush_req_offset = nbio_v7_0_get_hdp_flush_req_offset,
+	.get_hdp_flush_done_offset = nbio_v7_0_get_hdp_flush_done_offset,
+	.get_pcie_index_offset = nbio_v7_0_get_pcie_index_offset,
+	.get_pcie_data_offset = nbio_v7_0_get_pcie_data_offset,
+	.get_rev_id = nbio_v7_0_get_rev_id,
+	.mc_access_enable = nbio_v7_0_mc_access_enable,
+	.get_memsize = nbio_v7_0_get_memsize,
+	.sdma_doorbell_range = nbio_v7_0_sdma_doorbell_range,
+	.vcn_doorbell_range = nbio_v7_0_vcn_doorbell_range,
+	.enable_doorbell_aperture = nbio_v7_0_enable_doorbell_aperture,
+	.enable_doorbell_selfring_aperture = nbio_v7_0_enable_doorbell_selfring_aperture,
+	.ih_doorbell_range = nbio_v7_0_ih_doorbell_range,
+	.update_medium_grain_clock_gating = nbio_v7_0_update_medium_grain_clock_gating,
+	.update_medium_grain_light_sleep = nbio_v7_0_update_medium_grain_light_sleep,
+	.get_clockgating_state = nbio_v7_0_get_clockgating_state,
+	.ih_control = nbio_v7_0_ih_control,
+	.init_registers = nbio_v7_0_init_registers,
+	.remap_hdp_registers = nbio_v7_0_remap_hdp_registers,
+	.set_reg_remap = nbio_v7_0_set_reg_remap,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_0.h b/drivers/gpu/drm/amd/amdgpu/nbio_v7_0.h
index 054ff49427e6..e7aefb252550 100644
--- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_0.h
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_0.h
@@ -26,24 +26,7 @@
 
 #include "soc15_common.h"
 
-extern struct nbio_hdp_flush_reg nbio_v7_0_hdp_flush_reg;
-extern struct nbio_pcie_index_data nbio_v7_0_pcie_index_data;
-int nbio_v7_0_init(struct amdgpu_device *adev);
-u32 nbio_v7_0_get_atombios_scratch_regs(struct amdgpu_device *adev,
-                                        uint32_t idx);
-void nbio_v7_0_set_atombios_scratch_regs(struct amdgpu_device *adev,
-                                         uint32_t idx, uint32_t val);
-void nbio_v7_0_mc_access_enable(struct amdgpu_device *adev, bool enable);
-void nbio_v7_0_hdp_flush(struct amdgpu_device *adev);
-u32 nbio_v7_0_get_memsize(struct amdgpu_device *adev);
-void nbio_v7_0_sdma_doorbell_range(struct amdgpu_device *adev, int instance,
-				  bool use_doorbell, int doorbell_index);
-void nbio_v7_0_enable_doorbell_aperture(struct amdgpu_device *adev,
-					bool enable);
-void nbio_v7_0_ih_doorbell_range(struct amdgpu_device *adev,
-				bool use_doorbell, int doorbell_index);
-void nbio_v7_0_ih_control(struct amdgpu_device *adev);
-u32 nbio_v7_0_get_rev_id(struct amdgpu_device *adev);
-void nbio_v7_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
-						bool enable);
+extern const struct nbio_hdp_flush_reg nbio_v7_0_hdp_flush_reg;
+extern const struct amdgpu_nbio_funcs nbio_v7_0_funcs;
+
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_11.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_11.c
new file mode 100644
index 000000000000..bed5ef4d8788
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_11.c
@@ -0,0 +1,400 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "nbio_v7_11.h"
+
+#include "nbio/nbio_7_11_0_offset.h"
+#include "nbio/nbio_7_11_0_sh_mask.h"
+#include <uapi/linux/kfd_ioctl.h>
+
+static void nbio_v7_11_remap_hdp_registers(struct amdgpu_device *adev)
+{
+	WREG32_SOC15(NBIO, 0, regBIF_BX0_REMAP_HDP_MEM_FLUSH_CNTL,
+		     adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL);
+	WREG32_SOC15(NBIO, 0, regBIF_BX0_REMAP_HDP_REG_FLUSH_CNTL,
+		     adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_REG_FLUSH_CNTL);
+}
+
+static u32 nbio_v7_11_get_rev_id(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	tmp = RREG32_SOC15(NBIO, 0, regRCC_STRAP1_RCC_DEV0_EPF0_STRAP0);
+	tmp &= RCC_STRAP0_RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0_MASK;
+	tmp >>= RCC_STRAP0_RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0__SHIFT;
+
+	return tmp;
+}
+
+static void nbio_v7_11_mc_access_enable(struct amdgpu_device *adev, bool enable)
+{
+	if (enable)
+		WREG32_SOC15(NBIO, 0, regBIF_BX1_BIF_FB_EN,
+			BIF_BX1_BIF_FB_EN__FB_READ_EN_MASK |
+			BIF_BX1_BIF_FB_EN__FB_WRITE_EN_MASK);
+	else
+		WREG32_SOC15(NBIO, 0, regBIF_BX1_BIF_FB_EN, 0);
+}
+
+static u32 nbio_v7_11_get_memsize(struct amdgpu_device *adev)
+{
+	return RREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF0_0_RCC_CONFIG_MEMSIZE);
+}
+
+static void nbio_v7_11_sdma_doorbell_range(struct amdgpu_device *adev, int instance,
+					  bool use_doorbell, int doorbell_index,
+					  int doorbell_size)
+{
+	u32 reg = SOC15_REG_OFFSET(NBIO, 0, regGDC0_BIF_CSDMA_DOORBELL_RANGE);
+	u32 doorbell_range = RREG32_PCIE_PORT(reg);
+
+	if (use_doorbell) {
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC0_BIF_CSDMA_DOORBELL_RANGE,
+					       OFFSET, doorbell_index);
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC0_BIF_CSDMA_DOORBELL_RANGE,
+					       SIZE, doorbell_size);
+	} else {
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC0_BIF_CSDMA_DOORBELL_RANGE,
+					       SIZE, 0);
+	}
+
+	WREG32_PCIE_PORT(reg, doorbell_range);
+}
+
+static void nbio_v7_11_vpe_doorbell_range(struct amdgpu_device *adev, int instance,
+					  bool use_doorbell, int doorbell_index,
+					  int doorbell_size)
+{
+	u32 reg = instance == 0 ?
+		  SOC15_REG_OFFSET(NBIO, 0, regGDC0_BIF_VPE_DOORBELL_RANGE) :
+		  SOC15_REG_OFFSET(NBIO, 0, regGDC0_BIF_VPE1_DOORBELL_RANGE);
+	u32 doorbell_range = RREG32_PCIE_PORT(reg);
+
+	if (use_doorbell) {
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC0_BIF_VPE_DOORBELL_RANGE,
+					       OFFSET, doorbell_index);
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC0_BIF_VPE_DOORBELL_RANGE,
+					       SIZE, doorbell_size);
+	} else {
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC0_BIF_VPE_DOORBELL_RANGE,
+					       SIZE, 0);
+	}
+
+	WREG32_PCIE_PORT(reg, doorbell_range);
+}
+
+static void nbio_v7_11_vcn_doorbell_range(struct amdgpu_device *adev,
+					  bool use_doorbell,
+					  int doorbell_index, int instance)
+{
+	u32 reg = instance == 0 ?
+		SOC15_REG_OFFSET(NBIO, 0, regGDC0_BIF_VCN0_DOORBELL_RANGE):
+		SOC15_REG_OFFSET(NBIO, 0, regGDC0_BIF_VCN1_DOORBELL_RANGE);
+
+	u32 doorbell_range = RREG32_PCIE_PORT(reg);
+
+	if (use_doorbell) {
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC0_BIF_VCN0_DOORBELL_RANGE, OFFSET,
+					       doorbell_index);
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC0_BIF_VCN0_DOORBELL_RANGE, SIZE, 8);
+	} else {
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC0_BIF_VCN0_DOORBELL_RANGE, SIZE, 0);
+	}
+
+	WREG32_PCIE_PORT(reg, doorbell_range);
+}
+
+static void nbio_v7_11_enable_doorbell_aperture(struct amdgpu_device *adev,
+					       bool enable)
+{
+	u32 reg;
+
+
+	reg = RREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF0_0_RCC_DOORBELL_APER_EN);
+	reg = REG_SET_FIELD(reg, RCC_DEV0_EPF0_0_RCC_DOORBELL_APER_EN,
+			    BIF_DOORBELL_APER_EN, enable ? 1 : 0);
+
+	WREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF0_0_RCC_DOORBELL_APER_EN, reg);
+}
+
+static void nbio_v7_11_enable_doorbell_selfring_aperture(struct amdgpu_device *adev,
+							bool enable)
+{
+	u32 tmp = 0;
+
+	if (enable) {
+		tmp = REG_SET_FIELD(tmp, BIF_BX_PF1_DOORBELL_SELFRING_GPA_APER_CNTL,
+				DOORBELL_SELFRING_GPA_APER_EN, 1) |
+		      REG_SET_FIELD(tmp, BIF_BX_PF1_DOORBELL_SELFRING_GPA_APER_CNTL,
+				DOORBELL_SELFRING_GPA_APER_MODE, 1) |
+		      REG_SET_FIELD(tmp, BIF_BX_PF1_DOORBELL_SELFRING_GPA_APER_CNTL,
+				DOORBELL_SELFRING_GPA_APER_SIZE, 0);
+
+		WREG32_SOC15(NBIO, 0,
+			regBIF_BX_PF1_DOORBELL_SELFRING_GPA_APER_BASE_LOW,
+			lower_32_bits(adev->doorbell.base));
+		WREG32_SOC15(NBIO, 0,
+			regBIF_BX_PF1_DOORBELL_SELFRING_GPA_APER_BASE_HIGH,
+			upper_32_bits(adev->doorbell.base));
+	}
+
+	WREG32_SOC15(NBIO, 0, regBIF_BX_PF1_DOORBELL_SELFRING_GPA_APER_CNTL, tmp);
+}
+
+
+static void nbio_v7_11_ih_doorbell_range(struct amdgpu_device *adev,
+					bool use_doorbell, int doorbell_index)
+{    
+	u32 ih_doorbell_range = RREG32_SOC15(NBIO, 0,regGDC0_BIF_IH_DOORBELL_RANGE);
+
+	if (use_doorbell) {
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  GDC0_BIF_IH_DOORBELL_RANGE, OFFSET,
+						  doorbell_index);
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  GDC0_BIF_IH_DOORBELL_RANGE, SIZE,
+						  2);
+	} else {
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  GDC0_BIF_IH_DOORBELL_RANGE, SIZE,
+						  0);
+	}
+
+	WREG32_SOC15(NBIO, 0, regGDC0_BIF_IH_DOORBELL_RANGE,
+			 ih_doorbell_range);
+}
+
+static void nbio_v7_11_ih_control(struct amdgpu_device *adev)
+{
+	u32 interrupt_cntl;
+
+	/* setup interrupt control */
+	WREG32_SOC15(NBIO, 0, regBIF_BX1_INTERRUPT_CNTL2,
+		     adev->dummy_page_addr >> 8);
+
+	interrupt_cntl = RREG32_SOC15(NBIO, 0, regBIF_BX1_INTERRUPT_CNTL);
+	/*
+	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
+	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
+	 */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, BIF_BX1_INTERRUPT_CNTL,
+				       IH_DUMMY_RD_OVERRIDE, 0);
+
+	/* INTERRUPT_CNTL__IH_REQ_NONSNOOP_EN_MASK=1 if ring is in non-cacheable memory, e.g., vram */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, BIF_BX1_INTERRUPT_CNTL,
+				       IH_REQ_NONSNOOP_EN, 0);
+
+	WREG32_SOC15(NBIO, 0, regBIF_BX1_INTERRUPT_CNTL, interrupt_cntl);
+}
+
+static u32 nbio_v7_11_get_hdp_flush_req_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF1_GPU_HDP_FLUSH_REQ);
+}
+
+static u32 nbio_v7_11_get_hdp_flush_done_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF1_GPU_HDP_FLUSH_DONE);
+}
+
+static u32 nbio_v7_11_get_pcie_index_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX1_PCIE_INDEX2);
+}
+
+static u32 nbio_v7_11_get_pcie_data_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX1_PCIE_DATA2);
+}
+
+static u32 nbio_v7_11_get_pcie_port_index_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF1_RSMU_INDEX);
+}
+
+static u32 nbio_v7_11_get_pcie_port_data_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF1_RSMU_DATA);
+}
+
+const struct nbio_hdp_flush_reg nbio_v7_11_hdp_flush_reg = {
+	.ref_and_mask_cp0 = BIF_BX_PF1_GPU_HDP_FLUSH_DONE__CP0_MASK,
+	.ref_and_mask_cp1 = BIF_BX_PF1_GPU_HDP_FLUSH_DONE__CP1_MASK,
+	.ref_and_mask_cp2 = BIF_BX_PF1_GPU_HDP_FLUSH_DONE__CP2_MASK,
+	.ref_and_mask_cp3 = BIF_BX_PF1_GPU_HDP_FLUSH_DONE__CP3_MASK,
+	.ref_and_mask_cp4 = BIF_BX_PF1_GPU_HDP_FLUSH_DONE__CP4_MASK,
+	.ref_and_mask_cp5 = BIF_BX_PF1_GPU_HDP_FLUSH_DONE__CP5_MASK,
+	.ref_and_mask_cp6 = BIF_BX_PF1_GPU_HDP_FLUSH_DONE__CP6_MASK,
+	.ref_and_mask_cp7 = BIF_BX_PF1_GPU_HDP_FLUSH_DONE__CP7_MASK,
+	.ref_and_mask_cp8 = BIF_BX_PF1_GPU_HDP_FLUSH_DONE__CP8_MASK,
+	.ref_and_mask_cp9 = BIF_BX_PF1_GPU_HDP_FLUSH_DONE__CP9_MASK,
+	.ref_and_mask_sdma0 = BIF_BX_PF1_GPU_HDP_FLUSH_DONE__SDMA0_MASK,
+	.ref_and_mask_sdma1 = BIF_BX_PF1_GPU_HDP_FLUSH_DONE__SDMA1_MASK,
+};
+
+static void nbio_v7_11_init_registers(struct amdgpu_device *adev)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_SOC15(NBIO, 0, regBIF_BIF256_CI256_RC3X4_USB4_PCIE_MST_CTRL_3);
+	data = REG_SET_FIELD(data, BIF_BIF256_CI256_RC3X4_USB4_PCIE_MST_CTRL_3,
+				CI_SWUS_MAX_READ_REQUEST_SIZE_MODE, 1);
+	data = REG_SET_FIELD(data, BIF_BIF256_CI256_RC3X4_USB4_PCIE_MST_CTRL_3,
+				CI_SWUS_MAX_READ_REQUEST_SIZE_PRIV, 1);
+
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regBIF_BIF256_CI256_RC3X4_USB4_PCIE_MST_CTRL_3, data);
+
+	switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) {
+	case IP_VERSION(7, 11, 0):
+	case IP_VERSION(7, 11, 1):
+	case IP_VERSION(7, 11, 2):
+	case IP_VERSION(7, 11, 3):
+		data = RREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF5_STRAP4) & ~BIT(23);
+		WREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF5_STRAP4, data);
+		break;
+	}
+}
+
+static void nbio_v7_11_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						       bool enable)
+{
+	uint32_t def, data;
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_BIF_MGCG))
+		return;
+
+	def = data = RREG32_SOC15(NBIO, 0, regBIF_BIF256_CI256_RC3X4_USB4_CPM_CONTROL);
+	if (enable) {
+		data |= (BIF_BIF256_CI256_RC3X4_USB4_CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK |
+			 BIF_BIF256_CI256_RC3X4_USB4_CPM_CONTROL__TXCLK_DYN_GATE_ENABLE_MASK |
+			 BIF_BIF256_CI256_RC3X4_USB4_CPM_CONTROL__TXCLK_LCNT_GATE_ENABLE_MASK |
+			 BIF_BIF256_CI256_RC3X4_USB4_CPM_CONTROL__TXCLK_REGS_GATE_ENABLE_MASK |
+			 BIF_BIF256_CI256_RC3X4_USB4_CPM_CONTROL__TXCLK_PRBS_GATE_ENABLE_MASK |
+			 BIF_BIF256_CI256_RC3X4_USB4_CPM_CONTROL__REFCLK_REGS_GATE_ENABLE_MASK);
+	} else {
+		data &= ~(BIF_BIF256_CI256_RC3X4_USB4_CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK |
+			  BIF_BIF256_CI256_RC3X4_USB4_CPM_CONTROL__TXCLK_DYN_GATE_ENABLE_MASK |
+			  BIF_BIF256_CI256_RC3X4_USB4_CPM_CONTROL__TXCLK_LCNT_GATE_ENABLE_MASK |
+			  BIF_BIF256_CI256_RC3X4_USB4_CPM_CONTROL__TXCLK_REGS_GATE_ENABLE_MASK |
+			  BIF_BIF256_CI256_RC3X4_USB4_CPM_CONTROL__TXCLK_PRBS_GATE_ENABLE_MASK |
+			  BIF_BIF256_CI256_RC3X4_USB4_CPM_CONTROL__REFCLK_REGS_GATE_ENABLE_MASK);
+	}
+
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regBIF_BIF256_CI256_RC3X4_USB4_CPM_CONTROL, data);
+}
+
+static void nbio_v7_11_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+						      bool enable)
+{
+	uint32_t def, data;
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_BIF_LS))
+		return;
+
+	def = data = RREG32_SOC15(NBIO, 0, regBIF_BIF256_CI256_RC3X4_USB4_PCIE_CNTL2);
+	if (enable)
+		data |= BIF_BIF256_CI256_RC3X4_USB4_PCIE_CNTL2__SLV_MEM_LS_EN_MASK;
+	else
+		data &= ~BIF_BIF256_CI256_RC3X4_USB4_PCIE_CNTL2__SLV_MEM_LS_EN_MASK;
+
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regBIF_BIF256_CI256_RC3X4_USB4_PCIE_CNTL2, data);
+
+	def = data = RREG32_SOC15(NBIO, 0, regBIF_BIF256_CI256_RC3X4_USB4_PCIE_TX_POWER_CTRL_1);
+	if (enable) {
+		data |= (BIF_BIF256_CI256_RC3X4_USB4_PCIE_TX_POWER_CTRL_1__MST_MEM_LS_EN_MASK |
+			BIF_BIF256_CI256_RC3X4_USB4_PCIE_TX_POWER_CTRL_1__REPLAY_MEM_LS_EN_MASK);
+	} else {
+		data &= ~(BIF_BIF256_CI256_RC3X4_USB4_PCIE_TX_POWER_CTRL_1__MST_MEM_LS_EN_MASK |
+			BIF_BIF256_CI256_RC3X4_USB4_PCIE_TX_POWER_CTRL_1__REPLAY_MEM_LS_EN_MASK);
+	}
+
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regBIF_BIF256_CI256_RC3X4_USB4_PCIE_TX_POWER_CTRL_1, data);
+}
+
+static void nbio_v7_11_get_clockgating_state(struct amdgpu_device *adev,
+					    u64 *flags)
+{
+	uint32_t data;
+
+	/* AMD_CG_SUPPORT_BIF_MGCG */
+	data = RREG32_SOC15(NBIO, 0, regBIF_BIF256_CI256_RC3X4_USB4_CPM_CONTROL);
+	if (data & BIF_BIF256_CI256_RC3X4_USB4_CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_BIF_MGCG;
+
+	/* AMD_CG_SUPPORT_BIF_LS */
+	data = RREG32_SOC15(NBIO, 0, regBIF_BIF256_CI256_RC3X4_USB4_PCIE_CNTL2);
+	if (data & BIF_BIF256_CI256_RC3X4_USB4_PCIE_CNTL2__SLV_MEM_LS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_BIF_LS;
+}
+
+#define MMIO_REG_HOLE_OFFSET 0x44000
+
+static void nbio_v7_11_set_reg_remap(struct amdgpu_device *adev)
+{
+	if (!amdgpu_sriov_vf(adev) && (PAGE_SIZE <= 4096)) {
+		adev->rmmio_remap.reg_offset = MMIO_REG_HOLE_OFFSET;
+		adev->rmmio_remap.bus_addr = adev->rmmio_base + MMIO_REG_HOLE_OFFSET;
+	} else {
+		adev->rmmio_remap.reg_offset =
+			SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF1_HDP_MEM_COHERENCY_FLUSH_CNTL) << 2;
+		adev->rmmio_remap.bus_addr = 0;
+	}
+}
+
+const struct amdgpu_nbio_funcs nbio_v7_11_funcs = {
+	.get_hdp_flush_req_offset = nbio_v7_11_get_hdp_flush_req_offset,
+	.get_hdp_flush_done_offset = nbio_v7_11_get_hdp_flush_done_offset,
+	.get_pcie_index_offset = nbio_v7_11_get_pcie_index_offset,
+	.get_pcie_data_offset = nbio_v7_11_get_pcie_data_offset,
+	.get_pcie_port_index_offset = nbio_v7_11_get_pcie_port_index_offset,
+	.get_pcie_port_data_offset = nbio_v7_11_get_pcie_port_data_offset,
+	.get_rev_id = nbio_v7_11_get_rev_id,
+	.mc_access_enable = nbio_v7_11_mc_access_enable,
+	.get_memsize = nbio_v7_11_get_memsize,
+	.sdma_doorbell_range = nbio_v7_11_sdma_doorbell_range,
+	.vcn_doorbell_range = nbio_v7_11_vcn_doorbell_range,
+	.vpe_doorbell_range = nbio_v7_11_vpe_doorbell_range,
+	.enable_doorbell_aperture = nbio_v7_11_enable_doorbell_aperture,
+	.enable_doorbell_selfring_aperture = nbio_v7_11_enable_doorbell_selfring_aperture,
+	.ih_doorbell_range = nbio_v7_11_ih_doorbell_range,
+	.update_medium_grain_clock_gating = nbio_v7_11_update_medium_grain_clock_gating,
+	.update_medium_grain_light_sleep = nbio_v7_11_update_medium_grain_light_sleep,
+	.get_clockgating_state = nbio_v7_11_get_clockgating_state,
+	.ih_control = nbio_v7_11_ih_control,
+	.init_registers = nbio_v7_11_init_registers,
+	.remap_hdp_registers = nbio_v7_11_remap_hdp_registers,
+	.set_reg_remap = nbio_v7_11_set_reg_remap,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_11.h b/drivers/gpu/drm/amd/amdgpu/nbio_v7_11.h
new file mode 100644
index 000000000000..9d8258ed3f0a
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_11.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __NBIO_V7_11_H__
+#define __NBIO_V7_11_H__
+
+#include "soc15_common.h"
+
+extern const struct nbio_hdp_flush_reg nbio_v7_11_hdp_flush_reg;
+extern const struct amdgpu_nbio_funcs nbio_v7_11_funcs;
+extern const struct amdgpu_nbio_ras_funcs nbio_v7_11_ras_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_2.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_2.c
new file mode 100644
index 000000000000..acc5f363684a
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_2.c
@@ -0,0 +1,443 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "nbio_v7_2.h"
+
+#include "nbio/nbio_7_2_0_offset.h"
+#include "nbio/nbio_7_2_0_sh_mask.h"
+#include <uapi/linux/kfd_ioctl.h>
+
+#define regRCC_STRAP0_RCC_DEV0_EPF0_STRAP0_YC				0x0015
+#define regRCC_STRAP0_RCC_DEV0_EPF0_STRAP0_YC_BASE_IDX		2
+#define regBIF_BX0_BIF_FB_EN_YC								0x0100
+#define regBIF_BX0_BIF_FB_EN_YC_BASE_IDX					2
+#define regBIF1_PCIE_MST_CTRL_3								0x4601c6
+#define regBIF1_PCIE_MST_CTRL_3_BASE_IDX					5
+#define BIF1_PCIE_MST_CTRL_3__CI_SWUS_MAX_READ_REQUEST_SIZE_MODE__SHIFT \
+			0x1b
+#define BIF1_PCIE_MST_CTRL_3__CI_SWUS_MAX_READ_REQUEST_SIZE_PRIV__SHIFT \
+			0x1c
+#define BIF1_PCIE_MST_CTRL_3__CI_SWUS_MAX_READ_REQUEST_SIZE_MODE_MASK \
+			0x08000000L
+#define BIF1_PCIE_MST_CTRL_3__CI_SWUS_MAX_READ_REQUEST_SIZE_PRIV_MASK \
+			0x30000000L
+#define regBIF1_PCIE_TX_POWER_CTRL_1						0x460187
+#define regBIF1_PCIE_TX_POWER_CTRL_1_BASE_IDX				5
+#define BIF1_PCIE_TX_POWER_CTRL_1__MST_MEM_LS_EN_MASK		0x00000001L
+#define BIF1_PCIE_TX_POWER_CTRL_1__REPLAY_MEM_LS_EN_MASK	0x00000008L
+
+static void nbio_v7_2_remap_hdp_registers(struct amdgpu_device *adev)
+{
+	WREG32_SOC15(NBIO, 0, regBIF_BX0_REMAP_HDP_MEM_FLUSH_CNTL,
+		adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL);
+	WREG32_SOC15(NBIO, 0, regBIF_BX0_REMAP_HDP_REG_FLUSH_CNTL,
+		adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_REG_FLUSH_CNTL);
+}
+
+static u32 nbio_v7_2_get_rev_id(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) {
+	case IP_VERSION(7, 2, 1):
+	case IP_VERSION(7, 3, 0):
+	case IP_VERSION(7, 5, 0):
+		tmp = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_DEV0_EPF0_STRAP0_YC);
+		break;
+	default:
+		tmp = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_DEV0_EPF0_STRAP0);
+		break;
+	}
+
+	tmp &= RCC_STRAP0_RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0_MASK;
+	tmp >>= RCC_STRAP0_RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0__SHIFT;
+
+	return tmp;
+}
+
+static void nbio_v7_2_mc_access_enable(struct amdgpu_device *adev, bool enable)
+{
+	switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) {
+	case IP_VERSION(7, 2, 1):
+	case IP_VERSION(7, 3, 0):
+	case IP_VERSION(7, 5, 0):
+		if (enable)
+			WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_FB_EN_YC,
+				BIF_BX0_BIF_FB_EN__FB_READ_EN_MASK |
+				BIF_BX0_BIF_FB_EN__FB_WRITE_EN_MASK);
+		else
+			WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_FB_EN_YC, 0);
+	break;
+	default:
+		if (enable)
+			WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_FB_EN,
+				BIF_BX0_BIF_FB_EN__FB_READ_EN_MASK |
+				BIF_BX0_BIF_FB_EN__FB_WRITE_EN_MASK);
+		else
+			WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_FB_EN, 0);
+		break;
+	}
+}
+
+static u32 nbio_v7_2_get_memsize(struct amdgpu_device *adev)
+{
+	return RREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF0_0_RCC_CONFIG_MEMSIZE);
+}
+
+static void nbio_v7_2_sdma_doorbell_range(struct amdgpu_device *adev, int instance,
+					  bool use_doorbell, int doorbell_index,
+					  int doorbell_size)
+{
+	u32 reg = SOC15_REG_OFFSET(NBIO, 0, regGDC0_BIF_SDMA0_DOORBELL_RANGE);
+	u32 doorbell_range = RREG32_PCIE_PORT(reg);
+
+	if (use_doorbell) {
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC0_BIF_SDMA0_DOORBELL_RANGE,
+					       OFFSET, doorbell_index);
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC0_BIF_SDMA0_DOORBELL_RANGE,
+					       SIZE, doorbell_size);
+	} else {
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC0_BIF_SDMA0_DOORBELL_RANGE,
+					       SIZE, 0);
+	}
+
+	WREG32_PCIE_PORT(reg, doorbell_range);
+}
+
+static void nbio_v7_2_vcn_doorbell_range(struct amdgpu_device *adev, bool use_doorbell,
+					 int doorbell_index, int instance)
+{
+	u32 reg = SOC15_REG_OFFSET(NBIO, 0, regGDC0_BIF_VCN0_DOORBELL_RANGE);
+	u32 doorbell_range = RREG32_PCIE_PORT(reg);
+
+	if (use_doorbell) {
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+							GDC0_BIF_VCN0_DOORBELL_RANGE, OFFSET,
+							doorbell_index);
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+							GDC0_BIF_VCN0_DOORBELL_RANGE, SIZE, 8);
+	} else {
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+							GDC0_BIF_VCN0_DOORBELL_RANGE, SIZE, 0);
+	}
+
+	WREG32_PCIE_PORT(reg, doorbell_range);
+}
+
+static void nbio_v7_2_enable_doorbell_aperture(struct amdgpu_device *adev,
+					       bool enable)
+{
+	u32 reg;
+
+	reg = RREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF0_0_RCC_DOORBELL_APER_EN);
+	reg = REG_SET_FIELD(reg, RCC_DEV0_EPF0_0_RCC_DOORBELL_APER_EN,
+			    BIF_DOORBELL_APER_EN, enable ? 1 : 0);
+
+	WREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF0_0_RCC_DOORBELL_APER_EN, reg);
+}
+
+static void nbio_v7_2_enable_doorbell_selfring_aperture(struct amdgpu_device *adev,
+							bool enable)
+{
+	u32 tmp = 0;
+
+	if (enable) {
+		tmp = REG_SET_FIELD(tmp, BIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL,
+				DOORBELL_SELFRING_GPA_APER_EN, 1) |
+			REG_SET_FIELD(tmp, BIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL,
+				DOORBELL_SELFRING_GPA_APER_MODE, 1) |
+			REG_SET_FIELD(tmp, BIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL,
+				DOORBELL_SELFRING_GPA_APER_SIZE, 0);
+
+		WREG32_SOC15(NBIO, 0,
+			regBIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_BASE_LOW,
+			lower_32_bits(adev->doorbell.base));
+		WREG32_SOC15(NBIO, 0,
+			regBIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_BASE_HIGH,
+			upper_32_bits(adev->doorbell.base));
+	}
+
+	WREG32_SOC15(NBIO, 0, regBIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL,
+		tmp);
+}
+
+
+static void nbio_v7_2_ih_doorbell_range(struct amdgpu_device *adev,
+					bool use_doorbell, int doorbell_index)
+{
+	u32 ih_doorbell_range = RREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regGDC0_BIF_IH_DOORBELL_RANGE));
+
+	if (use_doorbell) {
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  GDC0_BIF_IH_DOORBELL_RANGE, OFFSET,
+						  doorbell_index);
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  GDC0_BIF_IH_DOORBELL_RANGE, SIZE,
+						  2);
+	} else {
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  GDC0_BIF_IH_DOORBELL_RANGE, SIZE,
+						  0);
+	}
+
+	WREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regGDC0_BIF_IH_DOORBELL_RANGE),
+			 ih_doorbell_range);
+}
+
+static void nbio_v7_2_ih_control(struct amdgpu_device *adev)
+{
+	u32 interrupt_cntl;
+
+	/* setup interrupt control */
+	WREG32_SOC15(NBIO, 0, regBIF_BX0_INTERRUPT_CNTL2,
+		     adev->dummy_page_addr >> 8);
+
+	interrupt_cntl = RREG32_SOC15(NBIO, 0, regBIF_BX0_INTERRUPT_CNTL);
+	/*
+	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
+	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
+	 */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, BIF_BX0_INTERRUPT_CNTL,
+				       IH_DUMMY_RD_OVERRIDE, 0);
+
+	/* INTERRUPT_CNTL__IH_REQ_NONSNOOP_EN_MASK=1 if ring is in non-cacheable memory, e.g., vram */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, BIF_BX0_INTERRUPT_CNTL,
+				       IH_REQ_NONSNOOP_EN, 0);
+
+	WREG32_SOC15(NBIO, 0, regBIF_BX0_INTERRUPT_CNTL, interrupt_cntl);
+}
+
+static void nbio_v7_2_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						       bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regCPM_CONTROL));
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_BIF_MGCG)) {
+		data |= (CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK |
+			 CPM_CONTROL__TXCLK_DYN_GATE_ENABLE_MASK |
+			 CPM_CONTROL__TXCLK_LCNT_GATE_ENABLE_MASK |
+			 CPM_CONTROL__TXCLK_REGS_GATE_ENABLE_MASK |
+			 CPM_CONTROL__TXCLK_PRBS_GATE_ENABLE_MASK |
+			 CPM_CONTROL__REFCLK_REGS_GATE_ENABLE_MASK);
+	} else {
+		data &= ~(CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK |
+			  CPM_CONTROL__TXCLK_DYN_GATE_ENABLE_MASK |
+			  CPM_CONTROL__TXCLK_LCNT_GATE_ENABLE_MASK |
+			  CPM_CONTROL__TXCLK_REGS_GATE_ENABLE_MASK |
+			  CPM_CONTROL__TXCLK_PRBS_GATE_ENABLE_MASK |
+			  CPM_CONTROL__REFCLK_REGS_GATE_ENABLE_MASK);
+	}
+
+	if (def != data)
+		WREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regCPM_CONTROL), data);
+}
+
+static void nbio_v7_2_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+						      bool enable)
+{
+	uint32_t def, data;
+
+	switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) {
+	case IP_VERSION(7, 2, 1):
+	case IP_VERSION(7, 3, 0):
+	case IP_VERSION(7, 5, 0):
+		def = data = RREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regPCIE_CNTL2));
+		if (enable && (adev->cg_flags & AMD_CG_SUPPORT_BIF_LS))
+			data |= PCIE_CNTL2__SLV_MEM_LS_EN_MASK;
+		else
+			data &= ~PCIE_CNTL2__SLV_MEM_LS_EN_MASK;
+
+		if (def != data)
+			WREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regPCIE_CNTL2), data);
+
+		def = data = RREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0,
+			regBIF1_PCIE_TX_POWER_CTRL_1));
+		if (enable && (adev->cg_flags & AMD_CG_SUPPORT_BIF_LS))
+			data |= (BIF1_PCIE_TX_POWER_CTRL_1__MST_MEM_LS_EN_MASK |
+				BIF1_PCIE_TX_POWER_CTRL_1__REPLAY_MEM_LS_EN_MASK);
+		else
+			data &= ~(BIF1_PCIE_TX_POWER_CTRL_1__MST_MEM_LS_EN_MASK |
+				BIF1_PCIE_TX_POWER_CTRL_1__REPLAY_MEM_LS_EN_MASK);
+
+		if (def != data)
+			WREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regBIF1_PCIE_TX_POWER_CTRL_1),
+				data);
+		break;
+	default:
+		def = data = RREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regPCIE_CNTL2));
+		if (enable && (adev->cg_flags & AMD_CG_SUPPORT_BIF_LS))
+			data |= (PCIE_CNTL2__SLV_MEM_LS_EN_MASK |
+				 PCIE_CNTL2__MST_MEM_LS_EN_MASK |
+				 PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK);
+		else
+			data &= ~(PCIE_CNTL2__SLV_MEM_LS_EN_MASK |
+				  PCIE_CNTL2__MST_MEM_LS_EN_MASK |
+				  PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK);
+
+		if (def != data)
+			WREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regPCIE_CNTL2), data);
+		break;
+	}
+}
+
+static void nbio_v7_2_get_clockgating_state(struct amdgpu_device *adev,
+					    u64 *flags)
+{
+	int data;
+
+	/* AMD_CG_SUPPORT_BIF_MGCG */
+	data = RREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regCPM_CONTROL));
+	if (data & CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_BIF_MGCG;
+
+	/* AMD_CG_SUPPORT_BIF_LS */
+	data = RREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regPCIE_CNTL2));
+	if (data & PCIE_CNTL2__SLV_MEM_LS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_BIF_LS;
+}
+
+static u32 nbio_v7_2_get_hdp_flush_req_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF0_GPU_HDP_FLUSH_REQ);
+}
+
+static u32 nbio_v7_2_get_hdp_flush_done_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF0_GPU_HDP_FLUSH_DONE);
+}
+
+static u32 nbio_v7_2_get_pcie_index_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX0_PCIE_INDEX2);
+}
+
+static u32 nbio_v7_2_get_pcie_data_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX0_PCIE_DATA2);
+}
+
+static u32 nbio_v7_2_get_pcie_port_index_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF0_RSMU_INDEX);
+}
+
+static u32 nbio_v7_2_get_pcie_port_data_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF0_RSMU_DATA);
+}
+
+const struct nbio_hdp_flush_reg nbio_v7_2_hdp_flush_reg = {
+	.ref_and_mask_cp0 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP0_MASK,
+	.ref_and_mask_cp1 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP1_MASK,
+	.ref_and_mask_cp2 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP2_MASK,
+	.ref_and_mask_cp3 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP3_MASK,
+	.ref_and_mask_cp4 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP4_MASK,
+	.ref_and_mask_cp5 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP5_MASK,
+	.ref_and_mask_cp6 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP6_MASK,
+	.ref_and_mask_cp7 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP7_MASK,
+	.ref_and_mask_cp8 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP8_MASK,
+	.ref_and_mask_cp9 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP9_MASK,
+	.ref_and_mask_sdma0 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__SDMA0_MASK,
+	.ref_and_mask_sdma1 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__SDMA1_MASK,
+};
+
+static void nbio_v7_2_init_registers(struct amdgpu_device *adev)
+{
+	uint32_t def, data;
+	switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) {
+	case IP_VERSION(7, 2, 1):
+	case IP_VERSION(7, 3, 0):
+	case IP_VERSION(7, 5, 0):
+		def = data = RREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regBIF1_PCIE_MST_CTRL_3));
+		data = REG_SET_FIELD(data, BIF1_PCIE_MST_CTRL_3,
+			CI_SWUS_MAX_READ_REQUEST_SIZE_MODE, 1);
+		data = REG_SET_FIELD(data, BIF1_PCIE_MST_CTRL_3,
+			CI_SWUS_MAX_READ_REQUEST_SIZE_PRIV, 1);
+
+		if (def != data)
+			WREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regBIF1_PCIE_MST_CTRL_3), data);
+		break;
+	default:
+		def = data = RREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regPCIE_CONFIG_CNTL));
+		data = REG_SET_FIELD(data, PCIE_CONFIG_CNTL,
+			CI_SWUS_MAX_READ_REQUEST_SIZE_MODE, 1);
+		data = REG_SET_FIELD(data, PCIE_CONFIG_CNTL,
+			CI_SWUS_MAX_READ_REQUEST_SIZE_PRIV, 1);
+
+		if (def != data)
+			WREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regPCIE_CONFIG_CNTL), data);
+		break;
+	}
+
+	switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) {
+	case IP_VERSION(7, 3, 0):
+	case IP_VERSION(7, 5, 1):
+		data = RREG32_SOC15(NBIO, 0, regRCC_DEV2_EPF0_STRAP2);
+		data &= ~RCC_DEV2_EPF0_STRAP2__STRAP_NO_SOFT_RESET_DEV2_F0_MASK;
+		WREG32_SOC15(NBIO, 0, regRCC_DEV2_EPF0_STRAP2, data);
+		break;
+	}
+}
+
+#define MMIO_REG_HOLE_OFFSET (0x80000 - PAGE_SIZE)
+
+static void nbio_v7_2_set_reg_remap(struct amdgpu_device *adev)
+{
+	if (!amdgpu_sriov_vf(adev) && (PAGE_SIZE <= 4096)) {
+		adev->rmmio_remap.reg_offset = MMIO_REG_HOLE_OFFSET;
+		adev->rmmio_remap.bus_addr = adev->rmmio_base + MMIO_REG_HOLE_OFFSET;
+	} else {
+		adev->rmmio_remap.reg_offset =
+			SOC15_REG_OFFSET(NBIO, 0,
+					 regBIF_BX_PF0_HDP_MEM_COHERENCY_FLUSH_CNTL) << 2;
+		adev->rmmio_remap.bus_addr = 0;
+	}
+}
+
+const struct amdgpu_nbio_funcs nbio_v7_2_funcs = {
+	.get_hdp_flush_req_offset = nbio_v7_2_get_hdp_flush_req_offset,
+	.get_hdp_flush_done_offset = nbio_v7_2_get_hdp_flush_done_offset,
+	.get_pcie_index_offset = nbio_v7_2_get_pcie_index_offset,
+	.get_pcie_data_offset = nbio_v7_2_get_pcie_data_offset,
+	.get_pcie_port_index_offset = nbio_v7_2_get_pcie_port_index_offset,
+	.get_pcie_port_data_offset = nbio_v7_2_get_pcie_port_data_offset,
+	.get_rev_id = nbio_v7_2_get_rev_id,
+	.mc_access_enable = nbio_v7_2_mc_access_enable,
+	.get_memsize = nbio_v7_2_get_memsize,
+	.sdma_doorbell_range = nbio_v7_2_sdma_doorbell_range,
+	.vcn_doorbell_range = nbio_v7_2_vcn_doorbell_range,
+	.enable_doorbell_aperture = nbio_v7_2_enable_doorbell_aperture,
+	.enable_doorbell_selfring_aperture = nbio_v7_2_enable_doorbell_selfring_aperture,
+	.ih_doorbell_range = nbio_v7_2_ih_doorbell_range,
+	.update_medium_grain_clock_gating = nbio_v7_2_update_medium_grain_clock_gating,
+	.update_medium_grain_light_sleep = nbio_v7_2_update_medium_grain_light_sleep,
+	.get_clockgating_state = nbio_v7_2_get_clockgating_state,
+	.ih_control = nbio_v7_2_ih_control,
+	.init_registers = nbio_v7_2_init_registers,
+	.remap_hdp_registers = nbio_v7_2_remap_hdp_registers,
+	.set_reg_remap = nbio_v7_2_set_reg_remap,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_2.h b/drivers/gpu/drm/amd/amdgpu/nbio_v7_2.h
new file mode 100644
index 000000000000..a8e8e65648a0
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_2.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __NBIO_V7_2_H__
+#define __NBIO_V7_2_H__
+
+#include "soc15_common.h"
+
+extern const struct nbio_hdp_flush_reg nbio_v7_2_hdp_flush_reg;
+extern const struct amdgpu_nbio_funcs nbio_v7_2_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
new file mode 100644
index 000000000000..860bc5cb03c8
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
@@ -0,0 +1,827 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "nbio_v7_4.h"
+#include "amdgpu_ras.h"
+
+#include "nbio/nbio_7_4_offset.h"
+#include "nbio/nbio_7_4_sh_mask.h"
+#include "nbio/nbio_7_4_0_smn.h"
+#include "ivsrcid/nbio/irqsrcs_nbif_7_4.h"
+#include <uapi/linux/kfd_ioctl.h>
+
+#define smnPCIE_LC_CNTL		0x11140280
+#define smnPCIE_LC_CNTL3	0x111402d4
+#define smnPCIE_LC_CNTL6	0x111402ec
+#define smnPCIE_LC_CNTL7	0x111402f0
+#define smnNBIF_MGCG_CTRL_LCLK	0x1013a21c
+#define smnRCC_BIF_STRAP3	0x1012348c
+#define RCC_BIF_STRAP3__STRAP_VLINK_ASPM_IDLE_TIMER_MASK	0x0000FFFFL
+#define RCC_BIF_STRAP3__STRAP_VLINK_PM_L1_ENTRY_TIMER_MASK	0xFFFF0000L
+#define smnRCC_BIF_STRAP5	0x10123494
+#define RCC_BIF_STRAP5__STRAP_VLINK_LDN_ENTRY_TIMER_MASK	0x0000FFFFL
+#define smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2	0x1014008c
+#define BIF_CFG_DEV0_EPF0_DEVICE_CNTL2__LTR_EN_MASK			0x0400L
+#define smnBIF_CFG_DEV0_EPF0_PCIE_LTR_CAP	0x10140324
+#define smnPSWUSP0_PCIE_LC_CNTL2		0x111402c4
+#define smnRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL	0x10123538
+#define smnRCC_BIF_STRAP2	0x10123488
+#define RCC_BIF_STRAP2__STRAP_LTR_IN_ASPML1_DIS_MASK	0x00004000L
+#define RCC_BIF_STRAP3__STRAP_VLINK_ASPM_IDLE_TIMER__SHIFT	0x0
+#define RCC_BIF_STRAP3__STRAP_VLINK_PM_L1_ENTRY_TIMER__SHIFT	0x10
+#define RCC_BIF_STRAP5__STRAP_VLINK_LDN_ENTRY_TIMER__SHIFT	0x0
+
+/*
+ * These are nbio v7_4_1 registers mask. Temporarily define these here since
+ * nbio v7_4_1 header is incomplete.
+ */
+#define GPU_HDP_FLUSH_DONE__RSVD_ENG0_MASK	0x00001000L /* Don't use.  Firmware uses this bit internally */
+#define GPU_HDP_FLUSH_DONE__RSVD_ENG1_MASK	0x00002000L
+#define GPU_HDP_FLUSH_DONE__RSVD_ENG2_MASK	0x00004000L
+#define GPU_HDP_FLUSH_DONE__RSVD_ENG3_MASK	0x00008000L
+#define GPU_HDP_FLUSH_DONE__RSVD_ENG4_MASK	0x00010000L
+#define GPU_HDP_FLUSH_DONE__RSVD_ENG5_MASK	0x00020000L
+#define GPU_HDP_FLUSH_DONE__RSVD_ENG6_MASK	0x00040000L
+#define GPU_HDP_FLUSH_DONE__RSVD_ENG7_MASK	0x00080000L
+#define GPU_HDP_FLUSH_DONE__RSVD_ENG8_MASK	0x00100000L
+
+#define mmBIF_MMSCH1_DOORBELL_RANGE                     0x01dc
+#define mmBIF_MMSCH1_DOORBELL_RANGE_BASE_IDX            2
+//BIF_MMSCH1_DOORBELL_RANGE
+#define BIF_MMSCH1_DOORBELL_RANGE__OFFSET__SHIFT        0x2
+#define BIF_MMSCH1_DOORBELL_RANGE__SIZE__SHIFT          0x10
+#define BIF_MMSCH1_DOORBELL_RANGE__OFFSET_MASK          0x00000FFCL
+#define BIF_MMSCH1_DOORBELL_RANGE__SIZE_MASK            0x001F0000L
+
+#define BIF_MMSCH1_DOORBELL_RANGE__OFFSET_MASK          0x00000FFCL
+#define BIF_MMSCH1_DOORBELL_RANGE__SIZE_MASK            0x001F0000L
+
+#define mmBIF_MMSCH1_DOORBELL_RANGE_ALDE                0x01d8
+#define mmBIF_MMSCH1_DOORBELL_RANGE_ALDE_BASE_IDX       2
+//BIF_MMSCH1_DOORBELL_ALDE_RANGE
+#define BIF_MMSCH1_DOORBELL_RANGE_ALDE__OFFSET__SHIFT   0x2
+#define BIF_MMSCH1_DOORBELL_RANGE_ALDE__SIZE__SHIFT     0x10
+#define BIF_MMSCH1_DOORBELL_RANGE_ALDE__OFFSET_MASK     0x00000FFCL
+#define BIF_MMSCH1_DOORBELL_RANGE_ALDE__SIZE_MASK       0x001F0000L
+
+#define mmRCC_DEV0_EPF0_STRAP0_ALDE			0x0015
+#define mmRCC_DEV0_EPF0_STRAP0_ALDE_BASE_IDX		2
+
+#define mmBIF_DOORBELL_INT_CNTL_ALDE 			0x00fe
+#define mmBIF_DOORBELL_INT_CNTL_ALDE_BASE_IDX 		2
+#define BIF_DOORBELL_INT_CNTL_ALDE__DOORBELL_INTERRUPT_DISABLE__SHIFT	0x18
+#define BIF_DOORBELL_INT_CNTL_ALDE__DOORBELL_INTERRUPT_DISABLE_MASK	0x01000000L
+
+#define mmBIF_INTR_CNTL_ALDE 				0x0101
+#define mmBIF_INTR_CNTL_ALDE_BASE_IDX 			2
+
+static void nbio_v7_4_query_ras_error_count(struct amdgpu_device *adev,
+					void *ras_error_status);
+
+static void nbio_v7_4_remap_hdp_registers(struct amdgpu_device *adev)
+{
+	WREG32_SOC15(NBIO, 0, mmREMAP_HDP_MEM_FLUSH_CNTL,
+		adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL);
+	WREG32_SOC15(NBIO, 0, mmREMAP_HDP_REG_FLUSH_CNTL,
+		adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_REG_FLUSH_CNTL);
+}
+
+static u32 nbio_v7_4_get_rev_id(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	if (adev->asic_type == CHIP_ALDEBARAN)
+		tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0_ALDE);
+	else
+		tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0);
+
+	tmp &= RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0_MASK;
+	tmp >>= RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0__SHIFT;
+
+	return tmp;
+}
+
+static void nbio_v7_4_mc_access_enable(struct amdgpu_device *adev, bool enable)
+{
+	if (enable)
+		WREG32_SOC15(NBIO, 0, mmBIF_FB_EN,
+			BIF_FB_EN__FB_READ_EN_MASK | BIF_FB_EN__FB_WRITE_EN_MASK);
+	else
+		WREG32_SOC15(NBIO, 0, mmBIF_FB_EN, 0);
+}
+
+static u32 nbio_v7_4_get_memsize(struct amdgpu_device *adev)
+{
+	return RREG32_SOC15(NBIO, 0, mmRCC_CONFIG_MEMSIZE);
+}
+
+static void nbio_v7_4_sdma_doorbell_range(struct amdgpu_device *adev, int instance,
+			bool use_doorbell, int doorbell_index, int doorbell_size)
+{
+	u32 reg, doorbell_range;
+
+	if (instance < 2) {
+		reg = instance +
+			SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA0_DOORBELL_RANGE);
+	} else {
+		/*
+		 * These registers address of SDMA2~7 is not consecutive
+		 * from SDMA0~1. Need plus 4 dwords offset.
+		 *
+		 *   BIF_SDMA0_DOORBELL_RANGE:  0x3bc0
+		 *   BIF_SDMA1_DOORBELL_RANGE:  0x3bc4
+		 *   BIF_SDMA2_DOORBELL_RANGE:  0x3bd8
+		 *   BIF_SDMA4_DOORBELL_RANGE:
+		 *     ARCTURUS:  0x3be0
+		 *     ALDEBARAN: 0x3be4
+		 */
+		if (adev->asic_type == CHIP_ALDEBARAN && instance == 4)
+			reg = instance + 0x4 + 0x1 +
+				SOC15_REG_OFFSET(NBIO, 0,
+						 mmBIF_SDMA0_DOORBELL_RANGE);
+		else
+			reg = instance + 0x4 +
+				SOC15_REG_OFFSET(NBIO, 0,
+						 mmBIF_SDMA0_DOORBELL_RANGE);
+	}
+
+	doorbell_range = RREG32(reg);
+
+	if (use_doorbell) {
+		doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, OFFSET, doorbell_index);
+		doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, doorbell_size);
+	} else
+		doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, 0);
+
+	WREG32(reg, doorbell_range);
+}
+
+static void nbio_v7_4_vcn_doorbell_range(struct amdgpu_device *adev, bool use_doorbell,
+					 int doorbell_index, int instance)
+{
+	u32 reg;
+	u32 doorbell_range;
+
+	if (instance) {
+		if (adev->asic_type == CHIP_ALDEBARAN)
+			reg = SOC15_REG_OFFSET(NBIO, 0, mmBIF_MMSCH1_DOORBELL_RANGE_ALDE);
+		else
+			reg = SOC15_REG_OFFSET(NBIO, 0, mmBIF_MMSCH1_DOORBELL_RANGE);
+	} else
+		reg = SOC15_REG_OFFSET(NBIO, 0, mmBIF_MMSCH0_DOORBELL_RANGE);
+
+	doorbell_range = RREG32(reg);
+
+	if (use_doorbell) {
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       BIF_MMSCH0_DOORBELL_RANGE, OFFSET,
+					       doorbell_index);
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       BIF_MMSCH0_DOORBELL_RANGE, SIZE, 8);
+	} else
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       BIF_MMSCH0_DOORBELL_RANGE, SIZE, 0);
+
+	WREG32(reg, doorbell_range);
+}
+
+static void nbio_v7_4_enable_doorbell_aperture(struct amdgpu_device *adev,
+					       bool enable)
+{
+	WREG32_FIELD15(NBIO, 0, RCC_DOORBELL_APER_EN, BIF_DOORBELL_APER_EN, enable ? 1 : 0);
+}
+
+static void nbio_v7_4_enable_doorbell_selfring_aperture(struct amdgpu_device *adev,
+							bool enable)
+{
+	u32 tmp = 0;
+
+	if (enable) {
+		tmp = REG_SET_FIELD(tmp, DOORBELL_SELFRING_GPA_APER_CNTL, DOORBELL_SELFRING_GPA_APER_EN, 1) |
+		      REG_SET_FIELD(tmp, DOORBELL_SELFRING_GPA_APER_CNTL, DOORBELL_SELFRING_GPA_APER_MODE, 1) |
+		      REG_SET_FIELD(tmp, DOORBELL_SELFRING_GPA_APER_CNTL, DOORBELL_SELFRING_GPA_APER_SIZE, 0);
+
+		WREG32_SOC15(NBIO, 0, mmDOORBELL_SELFRING_GPA_APER_BASE_LOW,
+			     lower_32_bits(adev->doorbell.base));
+		WREG32_SOC15(NBIO, 0, mmDOORBELL_SELFRING_GPA_APER_BASE_HIGH,
+			     upper_32_bits(adev->doorbell.base));
+	}
+
+	WREG32_SOC15(NBIO, 0, mmDOORBELL_SELFRING_GPA_APER_CNTL, tmp);
+}
+
+static void nbio_v7_4_ih_doorbell_range(struct amdgpu_device *adev,
+					bool use_doorbell, int doorbell_index)
+{
+	u32 ih_doorbell_range = RREG32_SOC15(NBIO, 0 , mmBIF_IH_DOORBELL_RANGE);
+
+	if (use_doorbell) {
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, OFFSET, doorbell_index);
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, SIZE, 8);
+	} else
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, SIZE, 0);
+
+	WREG32_SOC15(NBIO, 0, mmBIF_IH_DOORBELL_RANGE, ih_doorbell_range);
+}
+
+
+static void nbio_v7_4_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						       bool enable)
+{
+	//TODO: Add support for v7.4
+}
+
+static void nbio_v7_4_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+						      bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_PCIE(smnPCIE_CNTL2);
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_BIF_LS)) {
+		data |= (PCIE_CNTL2__SLV_MEM_LS_EN_MASK |
+			 PCIE_CNTL2__MST_MEM_LS_EN_MASK |
+			 PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK);
+	} else {
+		data &= ~(PCIE_CNTL2__SLV_MEM_LS_EN_MASK |
+			  PCIE_CNTL2__MST_MEM_LS_EN_MASK |
+			  PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK);
+	}
+
+	if (def != data)
+		WREG32_PCIE(smnPCIE_CNTL2, data);
+}
+
+static void nbio_v7_4_get_clockgating_state(struct amdgpu_device *adev,
+					    u64 *flags)
+{
+	int data;
+
+	/* AMD_CG_SUPPORT_BIF_MGCG */
+	data = RREG32_PCIE(smnCPM_CONTROL);
+	if (data & CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_BIF_MGCG;
+
+	/* AMD_CG_SUPPORT_BIF_LS */
+	data = RREG32_PCIE(smnPCIE_CNTL2);
+	if (data & PCIE_CNTL2__SLV_MEM_LS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_BIF_LS;
+}
+
+static void nbio_v7_4_ih_control(struct amdgpu_device *adev)
+{
+	u32 interrupt_cntl;
+
+	/* setup interrupt control */
+	WREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL2, adev->dummy_page_addr >> 8);
+	interrupt_cntl = RREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL);
+	/* INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
+	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
+	 */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, INTERRUPT_CNTL, IH_DUMMY_RD_OVERRIDE, 0);
+	/* INTERRUPT_CNTL__IH_REQ_NONSNOOP_EN_MASK=1 if ring is in non-cacheable memory, e.g., vram */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, INTERRUPT_CNTL, IH_REQ_NONSNOOP_EN, 0);
+	WREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL, interrupt_cntl);
+}
+
+static u32 nbio_v7_4_get_hdp_flush_req_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, mmGPU_HDP_FLUSH_REQ);
+}
+
+static u32 nbio_v7_4_get_hdp_flush_done_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, mmGPU_HDP_FLUSH_DONE);
+}
+
+static u32 nbio_v7_4_get_pcie_index_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, mmPCIE_INDEX2);
+}
+
+static u32 nbio_v7_4_get_pcie_data_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, mmPCIE_DATA2);
+}
+
+const struct nbio_hdp_flush_reg nbio_v7_4_hdp_flush_reg = {
+	.ref_and_mask_cp0 = GPU_HDP_FLUSH_DONE__CP0_MASK,
+	.ref_and_mask_cp1 = GPU_HDP_FLUSH_DONE__CP1_MASK,
+	.ref_and_mask_cp2 = GPU_HDP_FLUSH_DONE__CP2_MASK,
+	.ref_and_mask_cp3 = GPU_HDP_FLUSH_DONE__CP3_MASK,
+	.ref_and_mask_cp4 = GPU_HDP_FLUSH_DONE__CP4_MASK,
+	.ref_and_mask_cp5 = GPU_HDP_FLUSH_DONE__CP5_MASK,
+	.ref_and_mask_cp6 = GPU_HDP_FLUSH_DONE__CP6_MASK,
+	.ref_and_mask_cp7 = GPU_HDP_FLUSH_DONE__CP7_MASK,
+	.ref_and_mask_cp8 = GPU_HDP_FLUSH_DONE__CP8_MASK,
+	.ref_and_mask_cp9 = GPU_HDP_FLUSH_DONE__CP9_MASK,
+	.ref_and_mask_sdma0 = GPU_HDP_FLUSH_DONE__SDMA0_MASK,
+	.ref_and_mask_sdma1 = GPU_HDP_FLUSH_DONE__SDMA1_MASK,
+};
+
+static void nbio_v7_4_init_registers(struct amdgpu_device *adev)
+{
+	uint32_t baco_cntl;
+
+	if (amdgpu_ip_version(adev, NBIO_HWIP, 0) == IP_VERSION(7, 4, 4) &&
+	    !amdgpu_sriov_vf(adev)) {
+		baco_cntl = RREG32_SOC15(NBIO, 0, mmBACO_CNTL);
+		if (baco_cntl &
+		    (BACO_CNTL__BACO_DUMMY_EN_MASK | BACO_CNTL__BACO_EN_MASK)) {
+			baco_cntl &= ~(BACO_CNTL__BACO_DUMMY_EN_MASK |
+				       BACO_CNTL__BACO_EN_MASK);
+			dev_dbg(adev->dev, "Unsetting baco dummy mode %x",
+				baco_cntl);
+			WREG32_SOC15(NBIO, 0, mmBACO_CNTL, baco_cntl);
+		}
+	}
+}
+
+static void nbio_v7_4_handle_ras_controller_intr_no_bifring(struct amdgpu_device *adev)
+{
+	uint32_t bif_doorbell_intr_cntl;
+	struct ras_manager *obj = amdgpu_ras_find_obj(adev, adev->nbio.ras_if);
+	struct ras_err_data err_data;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+
+	if (amdgpu_ras_error_data_init(&err_data))
+		return;
+
+	if (adev->asic_type == CHIP_ALDEBARAN)
+		bif_doorbell_intr_cntl = RREG32_SOC15(NBIO, 0, mmBIF_DOORBELL_INT_CNTL_ALDE);
+	else
+		bif_doorbell_intr_cntl = RREG32_SOC15(NBIO, 0, mmBIF_DOORBELL_INT_CNTL);
+
+	if (REG_GET_FIELD(bif_doorbell_intr_cntl,
+		BIF_DOORBELL_INT_CNTL, RAS_CNTLR_INTERRUPT_STATUS)) {
+		/* driver has to clear the interrupt status when bif ring is disabled */
+		bif_doorbell_intr_cntl = REG_SET_FIELD(bif_doorbell_intr_cntl,
+						BIF_DOORBELL_INT_CNTL,
+						RAS_CNTLR_INTERRUPT_CLEAR, 1);
+		if (adev->asic_type == CHIP_ALDEBARAN)
+			WREG32_SOC15(NBIO, 0, mmBIF_DOORBELL_INT_CNTL_ALDE, bif_doorbell_intr_cntl);
+		else
+			WREG32_SOC15(NBIO, 0, mmBIF_DOORBELL_INT_CNTL, bif_doorbell_intr_cntl);
+
+		if (ras && !ras->disable_ras_err_cnt_harvest && obj) {
+			/*
+			 * clear error status after ras_controller_intr
+			 * according to hw team and count ue number
+			 * for query
+			 */
+			nbio_v7_4_query_ras_error_count(adev, &err_data);
+
+			/* logging on error cnt and printing for awareness */
+			obj->err_data.ue_count += err_data.ue_count;
+			obj->err_data.ce_count += err_data.ce_count;
+
+			if (err_data.ce_count)
+				dev_info(adev->dev, "%ld correctable hardware "
+						"errors detected in %s block\n",
+						obj->err_data.ce_count,
+						get_ras_block_str(adev->nbio.ras_if));
+
+			if (err_data.ue_count)
+				dev_info(adev->dev, "%ld uncorrectable hardware "
+						"errors detected in %s block\n",
+						obj->err_data.ue_count,
+						get_ras_block_str(adev->nbio.ras_if));
+		}
+
+		dev_info(adev->dev, "RAS controller interrupt triggered "
+					"by NBIF error\n");
+
+		/* ras_controller_int is dedicated for nbif ras error,
+		 * not the global interrupt for sync flood
+		 */
+		amdgpu_ras_global_ras_isr(adev);
+	}
+
+	amdgpu_ras_error_data_fini(&err_data);
+}
+
+static void nbio_v7_4_handle_ras_err_event_athub_intr_no_bifring(struct amdgpu_device *adev)
+{
+	uint32_t bif_doorbell_intr_cntl;
+
+	if (adev->asic_type == CHIP_ALDEBARAN)
+		bif_doorbell_intr_cntl = RREG32_SOC15(NBIO, 0, mmBIF_DOORBELL_INT_CNTL_ALDE);
+	else
+		bif_doorbell_intr_cntl = RREG32_SOC15(NBIO, 0, mmBIF_DOORBELL_INT_CNTL);
+
+	if (REG_GET_FIELD(bif_doorbell_intr_cntl,
+		BIF_DOORBELL_INT_CNTL, RAS_ATHUB_ERR_EVENT_INTERRUPT_STATUS)) {
+		/* driver has to clear the interrupt status when bif ring is disabled */
+		bif_doorbell_intr_cntl = REG_SET_FIELD(bif_doorbell_intr_cntl,
+						BIF_DOORBELL_INT_CNTL,
+						RAS_ATHUB_ERR_EVENT_INTERRUPT_CLEAR, 1);
+
+		if (adev->asic_type == CHIP_ALDEBARAN)
+			WREG32_SOC15(NBIO, 0, mmBIF_DOORBELL_INT_CNTL_ALDE, bif_doorbell_intr_cntl);
+		else
+			WREG32_SOC15(NBIO, 0, mmBIF_DOORBELL_INT_CNTL, bif_doorbell_intr_cntl);
+
+		amdgpu_ras_global_ras_isr(adev);
+	}
+}
+
+
+static int nbio_v7_4_set_ras_controller_irq_state(struct amdgpu_device *adev,
+						  struct amdgpu_irq_src *src,
+						  unsigned type,
+						  enum amdgpu_interrupt_state state)
+{
+	/* The ras_controller_irq enablement should be done in psp bl when it
+	 * tries to enable ras feature. Driver only need to set the correct interrupt
+	 * vector for bare-metal and sriov use case respectively
+	 */
+	uint32_t bif_intr_cntl;
+
+	if (adev->asic_type == CHIP_ALDEBARAN)
+		bif_intr_cntl = RREG32_SOC15(NBIO, 0, mmBIF_INTR_CNTL_ALDE);
+	else
+		bif_intr_cntl = RREG32_SOC15(NBIO, 0, mmBIF_INTR_CNTL);
+
+	if (state == AMDGPU_IRQ_STATE_ENABLE) {
+		/* set interrupt vector select bit to 0 to select
+		 * vetcor 1 for bare metal case */
+		bif_intr_cntl = REG_SET_FIELD(bif_intr_cntl,
+					      BIF_INTR_CNTL,
+					      RAS_INTR_VEC_SEL, 0);
+
+		if (adev->asic_type == CHIP_ALDEBARAN)
+			WREG32_SOC15(NBIO, 0, mmBIF_INTR_CNTL_ALDE, bif_intr_cntl);
+		else
+			WREG32_SOC15(NBIO, 0, mmBIF_INTR_CNTL, bif_intr_cntl);
+
+	}
+
+	return 0;
+}
+
+static int nbio_v7_4_process_ras_controller_irq(struct amdgpu_device *adev,
+						struct amdgpu_irq_src *source,
+						struct amdgpu_iv_entry *entry)
+{
+	/* By design, the ih cookie for ras_controller_irq should be written
+	 * to BIFring instead of general iv ring. However, due to known bif ring
+	 * hw bug, it has to be disabled. There is no chance the process function
+	 * will be involked. Just left it as a dummy one.
+	 */
+	return 0;
+}
+
+static int nbio_v7_4_set_ras_err_event_athub_irq_state(struct amdgpu_device *adev,
+						       struct amdgpu_irq_src *src,
+						       unsigned type,
+						       enum amdgpu_interrupt_state state)
+{
+	/* The ras_controller_irq enablement should be done in psp bl when it
+	 * tries to enable ras feature. Driver only need to set the correct interrupt
+	 * vector for bare-metal and sriov use case respectively
+	 */
+	uint32_t bif_intr_cntl;
+
+	if (adev->asic_type == CHIP_ALDEBARAN)
+		bif_intr_cntl = RREG32_SOC15(NBIO, 0, mmBIF_INTR_CNTL_ALDE);
+	else
+		bif_intr_cntl = RREG32_SOC15(NBIO, 0, mmBIF_INTR_CNTL);
+
+	if (state == AMDGPU_IRQ_STATE_ENABLE) {
+		/* set interrupt vector select bit to 0 to select
+		 * vetcor 1 for bare metal case */
+		bif_intr_cntl = REG_SET_FIELD(bif_intr_cntl,
+					      BIF_INTR_CNTL,
+					      RAS_INTR_VEC_SEL, 0);
+
+		if (adev->asic_type == CHIP_ALDEBARAN)
+			WREG32_SOC15(NBIO, 0, mmBIF_INTR_CNTL_ALDE, bif_intr_cntl);
+		else
+			WREG32_SOC15(NBIO, 0, mmBIF_INTR_CNTL, bif_intr_cntl);
+	}
+
+	return 0;
+}
+
+static int nbio_v7_4_process_err_event_athub_irq(struct amdgpu_device *adev,
+						 struct amdgpu_irq_src *source,
+						 struct amdgpu_iv_entry *entry)
+{
+	/* By design, the ih cookie for err_event_athub_irq should be written
+	 * to BIFring instead of general iv ring. However, due to known bif ring
+	 * hw bug, it has to be disabled. There is no chance the process function
+	 * will be involked. Just left it as a dummy one.
+	 */
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs nbio_v7_4_ras_controller_irq_funcs = {
+	.set = nbio_v7_4_set_ras_controller_irq_state,
+	.process = nbio_v7_4_process_ras_controller_irq,
+};
+
+static const struct amdgpu_irq_src_funcs nbio_v7_4_ras_err_event_athub_irq_funcs = {
+	.set = nbio_v7_4_set_ras_err_event_athub_irq_state,
+	.process = nbio_v7_4_process_err_event_athub_irq,
+};
+
+static int nbio_v7_4_init_ras_controller_interrupt (struct amdgpu_device *adev)
+{
+	int r;
+
+	/* init the irq funcs */
+	adev->nbio.ras_controller_irq.funcs =
+		&nbio_v7_4_ras_controller_irq_funcs;
+	adev->nbio.ras_controller_irq.num_types = 1;
+
+	/* register ras controller interrupt */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_BIF,
+			      NBIF_7_4__SRCID__RAS_CONTROLLER_INTERRUPT,
+			      &adev->nbio.ras_controller_irq);
+
+	return r;
+}
+
+static int nbio_v7_4_init_ras_err_event_athub_interrupt (struct amdgpu_device *adev)
+{
+
+	int r;
+
+	/* init the irq funcs */
+	adev->nbio.ras_err_event_athub_irq.funcs =
+		&nbio_v7_4_ras_err_event_athub_irq_funcs;
+	adev->nbio.ras_err_event_athub_irq.num_types = 1;
+
+	/* register ras err event athub interrupt */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_BIF,
+			      NBIF_7_4__SRCID__ERREVENT_ATHUB_INTERRUPT,
+			      &adev->nbio.ras_err_event_athub_irq);
+
+	return r;
+}
+
+#define smnPARITY_ERROR_STATUS_UNCORR_GRP2	    0x13a20030
+#define smnPARITY_ERROR_STATUS_UNCORR_GRP2_ALDE	0x13b20030
+#define smnRAS_GLOBAL_STATUS_LO_ALDE            0x13b20020
+
+static void nbio_v7_4_query_ras_error_count(struct amdgpu_device *adev,
+					void *ras_error_status)
+{
+	uint32_t global_sts, central_sts, int_eoi, parity_sts;
+	uint32_t corr, fatal, non_fatal;
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+
+	if (adev->asic_type == CHIP_ALDEBARAN)
+		global_sts = RREG32_PCIE(smnRAS_GLOBAL_STATUS_LO_ALDE);
+	else
+		global_sts = RREG32_PCIE(smnRAS_GLOBAL_STATUS_LO);
+
+	corr = REG_GET_FIELD(global_sts, RAS_GLOBAL_STATUS_LO, ParityErrCorr);
+	fatal = REG_GET_FIELD(global_sts, RAS_GLOBAL_STATUS_LO, ParityErrFatal);
+	non_fatal = REG_GET_FIELD(global_sts, RAS_GLOBAL_STATUS_LO,
+				ParityErrNonFatal);
+
+	if (adev->asic_type == CHIP_ALDEBARAN)
+		parity_sts = RREG32_PCIE(smnPARITY_ERROR_STATUS_UNCORR_GRP2_ALDE);
+	else
+		parity_sts = RREG32_PCIE(smnPARITY_ERROR_STATUS_UNCORR_GRP2);
+
+	if (corr)
+		err_data->ce_count++;
+	if (fatal)
+		err_data->ue_count++;
+
+	if (corr || fatal || non_fatal) {
+		central_sts = RREG32_PCIE(smnBIFL_RAS_CENTRAL_STATUS);
+
+		/* clear error status register */
+		if (adev->asic_type == CHIP_ALDEBARAN)
+			WREG32_PCIE(smnRAS_GLOBAL_STATUS_LO_ALDE, global_sts);
+		else
+			WREG32_PCIE(smnRAS_GLOBAL_STATUS_LO, global_sts);
+
+		if (fatal)
+		{
+			/* clear parity fatal error indication field */
+			if (adev->asic_type == CHIP_ALDEBARAN)
+				WREG32_PCIE(smnPARITY_ERROR_STATUS_UNCORR_GRP2_ALDE, parity_sts);
+			else
+				WREG32_PCIE(smnPARITY_ERROR_STATUS_UNCORR_GRP2, parity_sts);
+		}
+
+		if (REG_GET_FIELD(central_sts, BIFL_RAS_CENTRAL_STATUS,
+				BIFL_RasContller_Intr_Recv)) {
+			/* clear interrupt status register */
+			WREG32_PCIE(smnBIFL_RAS_CENTRAL_STATUS, central_sts);
+			int_eoi = RREG32_PCIE(smnIOHC_INTERRUPT_EOI);
+			int_eoi = REG_SET_FIELD(int_eoi,
+					IOHC_INTERRUPT_EOI, SMI_EOI, 1);
+			WREG32_PCIE(smnIOHC_INTERRUPT_EOI, int_eoi);
+		}
+	}
+}
+
+static void nbio_v7_4_enable_doorbell_interrupt(struct amdgpu_device *adev,
+						bool enable)
+{
+	if (adev->asic_type == CHIP_ALDEBARAN)
+		WREG32_FIELD15(NBIO, 0, BIF_DOORBELL_INT_CNTL_ALDE,
+		       DOORBELL_INTERRUPT_DISABLE, enable ? 0 : 1);
+	else
+		WREG32_FIELD15(NBIO, 0, BIF_DOORBELL_INT_CNTL,
+		       DOORBELL_INTERRUPT_DISABLE, enable ? 0 : 1);
+}
+
+const struct amdgpu_ras_block_hw_ops nbio_v7_4_ras_hw_ops = {
+	.query_ras_error_count = nbio_v7_4_query_ras_error_count,
+};
+
+struct amdgpu_nbio_ras nbio_v7_4_ras = {
+	.ras_block = {
+		.ras_comm = {
+			.name = "pcie_bif",
+			.block = AMDGPU_RAS_BLOCK__PCIE_BIF,
+			.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE,
+		},
+		.hw_ops = &nbio_v7_4_ras_hw_ops,
+		.ras_late_init = amdgpu_nbio_ras_late_init,
+	},
+	.handle_ras_controller_intr_no_bifring = nbio_v7_4_handle_ras_controller_intr_no_bifring,
+	.handle_ras_err_event_athub_intr_no_bifring = nbio_v7_4_handle_ras_err_event_athub_intr_no_bifring,
+	.init_ras_controller_interrupt = nbio_v7_4_init_ras_controller_interrupt,
+	.init_ras_err_event_athub_interrupt = nbio_v7_4_init_ras_err_event_athub_interrupt,
+};
+
+
+#ifdef CONFIG_PCIEASPM
+static void nbio_v7_4_program_ltr(struct amdgpu_device *adev)
+{
+	uint32_t def, data;
+
+	WREG32_PCIE(smnRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL, 0x75EB);
+
+	def = data = RREG32_PCIE(smnRCC_BIF_STRAP2);
+	data &= ~RCC_BIF_STRAP2__STRAP_LTR_IN_ASPML1_DIS_MASK;
+	if (def != data)
+		WREG32_PCIE(smnRCC_BIF_STRAP2, data);
+
+	def = data = RREG32_PCIE(smnRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL);
+	data &= ~EP_PCIE_TX_LTR_CNTL__LTR_PRIV_MSG_DIS_IN_PM_NON_D0_MASK;
+	if (def != data)
+		WREG32_PCIE(smnRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL, data);
+
+	def = data = RREG32_PCIE(smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2);
+	data |= BIF_CFG_DEV0_EPF0_DEVICE_CNTL2__LTR_EN_MASK;
+	if (def != data)
+		WREG32_PCIE(smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2, data);
+}
+#endif
+
+static void nbio_v7_4_program_aspm(struct amdgpu_device *adev)
+{
+#ifdef CONFIG_PCIEASPM
+	uint32_t def, data;
+
+	if (amdgpu_ip_version(adev, NBIO_HWIP, 0) == IP_VERSION(7, 4, 4))
+		return;
+
+	def = data = RREG32_PCIE(smnPCIE_LC_CNTL);
+	data &= ~PCIE_LC_CNTL__LC_L1_INACTIVITY_MASK;
+	data &= ~PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK;
+	data |= PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK;
+	if (def != data)
+		WREG32_PCIE(smnPCIE_LC_CNTL, data);
+
+	def = data = RREG32_PCIE(smnPCIE_LC_CNTL7);
+	data |= PCIE_LC_CNTL7__LC_NBIF_ASPM_INPUT_EN_MASK;
+	if (def != data)
+		WREG32_PCIE(smnPCIE_LC_CNTL7, data);
+
+	def = data = RREG32_PCIE(smnNBIF_MGCG_CTRL_LCLK);
+	data |= NBIF_MGCG_CTRL_LCLK__NBIF_MGCG_REG_DIS_LCLK_MASK;
+	if (def != data)
+		WREG32_PCIE(smnNBIF_MGCG_CTRL_LCLK, data);
+
+	def = data = RREG32_PCIE(smnPCIE_LC_CNTL3);
+	data |= PCIE_LC_CNTL3__LC_DSC_DONT_ENTER_L23_AFTER_PME_ACK_MASK;
+	if (def != data)
+		WREG32_PCIE(smnPCIE_LC_CNTL3, data);
+
+	def = data = RREG32_PCIE(smnRCC_BIF_STRAP3);
+	data &= ~RCC_BIF_STRAP3__STRAP_VLINK_ASPM_IDLE_TIMER_MASK;
+	data &= ~RCC_BIF_STRAP3__STRAP_VLINK_PM_L1_ENTRY_TIMER_MASK;
+	if (def != data)
+		WREG32_PCIE(smnRCC_BIF_STRAP3, data);
+
+	def = data = RREG32_PCIE(smnRCC_BIF_STRAP5);
+	data &= ~RCC_BIF_STRAP5__STRAP_VLINK_LDN_ENTRY_TIMER_MASK;
+	if (def != data)
+		WREG32_PCIE(smnRCC_BIF_STRAP5, data);
+
+	def = data = RREG32_PCIE(smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2);
+	data &= ~BIF_CFG_DEV0_EPF0_DEVICE_CNTL2__LTR_EN_MASK;
+	if (def != data)
+		WREG32_PCIE(smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2, data);
+
+	WREG32_PCIE(smnBIF_CFG_DEV0_EPF0_PCIE_LTR_CAP, 0x10011001);
+
+	def = data = RREG32_PCIE(smnPSWUSP0_PCIE_LC_CNTL2);
+	data |= PSWUSP0_PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L1_MASK |
+		PSWUSP0_PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L23_MASK;
+	data &= ~PSWUSP0_PCIE_LC_CNTL2__LC_RCV_L0_TO_RCV_L0S_DIS_MASK;
+	if (def != data)
+		WREG32_PCIE(smnPSWUSP0_PCIE_LC_CNTL2, data);
+
+	def = data = RREG32_PCIE(smnPCIE_LC_CNTL6);
+	data |= PCIE_LC_CNTL6__LC_L1_POWERDOWN_MASK |
+		PCIE_LC_CNTL6__LC_RX_L0S_STANDBY_EN_MASK;
+	if (def != data)
+		WREG32_PCIE(smnPCIE_LC_CNTL6, data);
+
+	/* Don't bother about LTR if LTR is not enabled
+	 * in the path */
+	if (adev->pdev->ltr_path)
+		nbio_v7_4_program_ltr(adev);
+
+	def = data = RREG32_PCIE(smnRCC_BIF_STRAP3);
+	data |= 0x5DE0 << RCC_BIF_STRAP3__STRAP_VLINK_ASPM_IDLE_TIMER__SHIFT;
+	data |= 0x0010 << RCC_BIF_STRAP3__STRAP_VLINK_PM_L1_ENTRY_TIMER__SHIFT;
+	if (def != data)
+		WREG32_PCIE(smnRCC_BIF_STRAP3, data);
+
+	def = data = RREG32_PCIE(smnRCC_BIF_STRAP5);
+	data |= 0x0010 << RCC_BIF_STRAP5__STRAP_VLINK_LDN_ENTRY_TIMER__SHIFT;
+	if (def != data)
+		WREG32_PCIE(smnRCC_BIF_STRAP5, data);
+
+	def = data = RREG32_PCIE(smnPCIE_LC_CNTL);
+	data &= ~PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK;
+	data |= 0x9 << PCIE_LC_CNTL__LC_L1_INACTIVITY__SHIFT;
+	data |= 0x1 << PCIE_LC_CNTL__LC_PMI_TO_L1_DIS__SHIFT;
+	if (def != data)
+		WREG32_PCIE(smnPCIE_LC_CNTL, data);
+
+	def = data = RREG32_PCIE(smnPCIE_LC_CNTL3);
+	data &= ~PCIE_LC_CNTL3__LC_DSC_DONT_ENTER_L23_AFTER_PME_ACK_MASK;
+	if (def != data)
+		WREG32_PCIE(smnPCIE_LC_CNTL3, data);
+#endif
+}
+
+#define MMIO_REG_HOLE_OFFSET (0x80000 - PAGE_SIZE)
+
+static void nbio_v7_4_set_reg_remap(struct amdgpu_device *adev)
+{
+	if (!amdgpu_sriov_vf(adev) && (PAGE_SIZE <= 4096)) {
+		adev->rmmio_remap.reg_offset = MMIO_REG_HOLE_OFFSET;
+		adev->rmmio_remap.bus_addr = adev->rmmio_base + MMIO_REG_HOLE_OFFSET;
+	} else {
+		adev->rmmio_remap.reg_offset =
+			SOC15_REG_OFFSET(NBIO, 0,
+					 mmBIF_BX_DEV0_EPF0_VF0_HDP_MEM_COHERENCY_FLUSH_CNTL) << 2;
+		adev->rmmio_remap.bus_addr = 0;
+	}
+}
+
+const struct amdgpu_nbio_funcs nbio_v7_4_funcs = {
+	.get_hdp_flush_req_offset = nbio_v7_4_get_hdp_flush_req_offset,
+	.get_hdp_flush_done_offset = nbio_v7_4_get_hdp_flush_done_offset,
+	.get_pcie_index_offset = nbio_v7_4_get_pcie_index_offset,
+	.get_pcie_data_offset = nbio_v7_4_get_pcie_data_offset,
+	.get_rev_id = nbio_v7_4_get_rev_id,
+	.mc_access_enable = nbio_v7_4_mc_access_enable,
+	.get_memsize = nbio_v7_4_get_memsize,
+	.sdma_doorbell_range = nbio_v7_4_sdma_doorbell_range,
+	.vcn_doorbell_range = nbio_v7_4_vcn_doorbell_range,
+	.enable_doorbell_aperture = nbio_v7_4_enable_doorbell_aperture,
+	.enable_doorbell_selfring_aperture = nbio_v7_4_enable_doorbell_selfring_aperture,
+	.ih_doorbell_range = nbio_v7_4_ih_doorbell_range,
+	.enable_doorbell_interrupt = nbio_v7_4_enable_doorbell_interrupt,
+	.update_medium_grain_clock_gating = nbio_v7_4_update_medium_grain_clock_gating,
+	.update_medium_grain_light_sleep = nbio_v7_4_update_medium_grain_light_sleep,
+	.get_clockgating_state = nbio_v7_4_get_clockgating_state,
+	.ih_control = nbio_v7_4_ih_control,
+	.init_registers = nbio_v7_4_init_registers,
+	.remap_hdp_registers = nbio_v7_4_remap_hdp_registers,
+	.program_aspm =  nbio_v7_4_program_aspm,
+	.set_reg_remap = nbio_v7_4_set_reg_remap,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.h b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.h
new file mode 100644
index 000000000000..f27c41728822
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __NBIO_V7_4_H__
+#define __NBIO_V7_4_H__
+
+#include "soc15_common.h"
+
+extern const struct nbio_hdp_flush_reg nbio_v7_4_hdp_flush_reg;
+extern const struct amdgpu_nbio_funcs nbio_v7_4_funcs;
+extern struct amdgpu_nbio_ras nbio_v7_4_ras;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_7.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_7.c
new file mode 100644
index 000000000000..2ee60b8746a6
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_7.c
@@ -0,0 +1,369 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "nbio_v7_7.h"
+
+#include "nbio/nbio_7_7_0_offset.h"
+#include "nbio/nbio_7_7_0_sh_mask.h"
+#include <uapi/linux/kfd_ioctl.h>
+
+static void nbio_v7_7_remap_hdp_registers(struct amdgpu_device *adev)
+{
+	WREG32_SOC15(NBIO, 0, regBIF_BX0_REMAP_HDP_MEM_FLUSH_CNTL,
+		     adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL);
+	WREG32_SOC15(NBIO, 0, regBIF_BX0_REMAP_HDP_REG_FLUSH_CNTL,
+		     adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_REG_FLUSH_CNTL);
+}
+
+static u32 nbio_v7_7_get_rev_id(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	tmp = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_DEV0_EPF0_STRAP0);
+	tmp &= RCC_STRAP0_RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0_MASK;
+	tmp >>= RCC_STRAP0_RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0__SHIFT;
+
+	return tmp;
+}
+
+static void nbio_v7_7_mc_access_enable(struct amdgpu_device *adev, bool enable)
+{
+	if (enable)
+		WREG32_SOC15(NBIO, 0, regBIF_BX1_BIF_FB_EN,
+			BIF_BX1_BIF_FB_EN__FB_READ_EN_MASK |
+			BIF_BX1_BIF_FB_EN__FB_WRITE_EN_MASK);
+	else
+		WREG32_SOC15(NBIO, 0, regBIF_BX1_BIF_FB_EN, 0);
+}
+
+static u32 nbio_v7_7_get_memsize(struct amdgpu_device *adev)
+{
+	return RREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF0_0_RCC_CONFIG_MEMSIZE);
+}
+
+static void nbio_v7_7_sdma_doorbell_range(struct amdgpu_device *adev, int instance,
+					  bool use_doorbell, int doorbell_index,
+					  int doorbell_size)
+{
+	u32 reg = SOC15_REG_OFFSET(NBIO, 0, regGDC0_BIF_CSDMA_DOORBELL_RANGE);
+	u32 doorbell_range = RREG32_PCIE_PORT(reg);
+
+	if (use_doorbell) {
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC0_BIF_CSDMA_DOORBELL_RANGE,
+					       OFFSET, doorbell_index);
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC0_BIF_CSDMA_DOORBELL_RANGE,
+					       SIZE, doorbell_size);
+	} else {
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC0_BIF_SDMA0_DOORBELL_RANGE,
+					       SIZE, 0);
+	}
+
+	WREG32_PCIE_PORT(reg, doorbell_range);
+}
+
+static void nbio_v7_7_vcn_doorbell_range(struct amdgpu_device *adev, bool use_doorbell,
+					int doorbell_index, int instance)
+{
+	u32 reg = SOC15_REG_OFFSET(NBIO, 0, regGDC0_BIF_VCN0_DOORBELL_RANGE);
+	u32 doorbell_range = RREG32_PCIE_PORT(reg);
+
+	if (use_doorbell) {
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC0_BIF_VCN0_DOORBELL_RANGE, OFFSET,
+					       doorbell_index);
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC0_BIF_VCN0_DOORBELL_RANGE, SIZE, 8);
+	} else {
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+					       GDC0_BIF_VCN0_DOORBELL_RANGE, SIZE, 0);
+	}
+
+	WREG32_PCIE_PORT(reg, doorbell_range);
+}
+
+static void nbio_v7_7_enable_doorbell_aperture(struct amdgpu_device *adev,
+					       bool enable)
+{
+	u32 reg;
+
+	reg = RREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF0_0_RCC_DOORBELL_APER_EN);
+	reg = REG_SET_FIELD(reg, RCC_DEV0_EPF0_0_RCC_DOORBELL_APER_EN,
+			    BIF_DOORBELL_APER_EN, enable ? 1 : 0);
+
+	WREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF0_0_RCC_DOORBELL_APER_EN, reg);
+}
+
+static void nbio_v7_7_enable_doorbell_selfring_aperture(struct amdgpu_device *adev,
+							bool enable)
+{
+	u32 tmp = 0;
+
+	if (enable) {
+		tmp = REG_SET_FIELD(tmp, BIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL,
+				DOORBELL_SELFRING_GPA_APER_EN, 1) |
+			REG_SET_FIELD(tmp, BIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL,
+				DOORBELL_SELFRING_GPA_APER_MODE, 1) |
+			REG_SET_FIELD(tmp, BIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL,
+				DOORBELL_SELFRING_GPA_APER_SIZE, 0);
+
+		WREG32_SOC15(NBIO, 0,
+			regBIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_BASE_LOW,
+			lower_32_bits(adev->doorbell.base));
+		WREG32_SOC15(NBIO, 0,
+			regBIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_BASE_HIGH,
+			upper_32_bits(adev->doorbell.base));
+	}
+
+	WREG32_SOC15(NBIO, 0, regBIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL,
+		tmp);
+}
+
+
+static void nbio_v7_7_ih_doorbell_range(struct amdgpu_device *adev,
+					bool use_doorbell, int doorbell_index)
+{
+	u32 ih_doorbell_range = RREG32_SOC15(NBIO, 0,
+								regGDC0_BIF_IH_DOORBELL_RANGE);
+
+	if (use_doorbell) {
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  GDC0_BIF_IH_DOORBELL_RANGE, OFFSET,
+						  doorbell_index);
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  GDC0_BIF_IH_DOORBELL_RANGE, SIZE,
+						  2);
+	} else {
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+						  GDC0_BIF_IH_DOORBELL_RANGE, SIZE,
+						  0);
+	}
+
+	WREG32_SOC15(NBIO, 0, regGDC0_BIF_IH_DOORBELL_RANGE,
+			 ih_doorbell_range);
+}
+
+static void nbio_v7_7_ih_control(struct amdgpu_device *adev)
+{
+	u32 interrupt_cntl;
+
+	/* setup interrupt control */
+	WREG32_SOC15(NBIO, 0, regBIF_BX1_INTERRUPT_CNTL2,
+		     adev->dummy_page_addr >> 8);
+
+	interrupt_cntl = RREG32_SOC15(NBIO, 0, regBIF_BX1_INTERRUPT_CNTL);
+	/*
+	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
+	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
+	 */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, BIF_BX1_INTERRUPT_CNTL,
+				       IH_DUMMY_RD_OVERRIDE, 0);
+
+	/* INTERRUPT_CNTL__IH_REQ_NONSNOOP_EN_MASK=1 if ring is in non-cacheable memory, e.g., vram */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, BIF_BX1_INTERRUPT_CNTL,
+				       IH_REQ_NONSNOOP_EN, 0);
+
+	WREG32_SOC15(NBIO, 0, regBIF_BX1_INTERRUPT_CNTL, interrupt_cntl);
+}
+
+static u32 nbio_v7_7_get_hdp_flush_req_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF0_GPU_HDP_FLUSH_REQ);
+}
+
+static u32 nbio_v7_7_get_hdp_flush_done_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF0_GPU_HDP_FLUSH_DONE);
+}
+
+static u32 nbio_v7_7_get_pcie_index_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX0_PCIE_INDEX2);
+}
+
+static u32 nbio_v7_7_get_pcie_data_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX0_PCIE_DATA2);
+}
+
+static u32 nbio_v7_7_get_pcie_port_index_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF0_RSMU_INDEX);
+}
+
+static u32 nbio_v7_7_get_pcie_port_data_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF0_RSMU_DATA);
+}
+
+const struct nbio_hdp_flush_reg nbio_v7_7_hdp_flush_reg = {
+	.ref_and_mask_cp0 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP0_MASK,
+	.ref_and_mask_cp1 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP1_MASK,
+	.ref_and_mask_cp2 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP2_MASK,
+	.ref_and_mask_cp3 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP3_MASK,
+	.ref_and_mask_cp4 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP4_MASK,
+	.ref_and_mask_cp5 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP5_MASK,
+	.ref_and_mask_cp6 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP6_MASK,
+	.ref_and_mask_cp7 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP7_MASK,
+	.ref_and_mask_cp8 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP8_MASK,
+	.ref_and_mask_cp9 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP9_MASK,
+	.ref_and_mask_sdma0 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__SDMA0_MASK,
+	.ref_and_mask_sdma1 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__SDMA1_MASK,
+};
+
+static void nbio_v7_7_init_registers(struct amdgpu_device *adev)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_SOC15(NBIO, 0, regBIF0_PCIE_MST_CTRL_3);
+	data = REG_SET_FIELD(data, BIF0_PCIE_MST_CTRL_3,
+			     CI_SWUS_MAX_READ_REQUEST_SIZE_MODE, 1);
+	data = REG_SET_FIELD(data, BIF0_PCIE_MST_CTRL_3,
+			     CI_SWUS_MAX_READ_REQUEST_SIZE_PRIV, 1);
+
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regBIF0_PCIE_MST_CTRL_3, data);
+
+	switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) {
+	case IP_VERSION(7, 7, 0):
+		data = RREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF5_STRAP4) & ~BIT(23);
+		WREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF5_STRAP4, data);
+		break;
+	}
+}
+
+static void nbio_v7_7_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						       bool enable)
+{
+	uint32_t def, data;
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_BIF_MGCG))
+		return;
+
+	def = data = RREG32_SOC15(NBIO, 0, regBIF0_CPM_CONTROL);
+	if (enable) {
+		data |= (BIF0_CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK |
+			 BIF0_CPM_CONTROL__TXCLK_DYN_GATE_ENABLE_MASK |
+			 BIF0_CPM_CONTROL__TXCLK_LCNT_GATE_ENABLE_MASK |
+			 BIF0_CPM_CONTROL__TXCLK_REGS_GATE_ENABLE_MASK |
+			 BIF0_CPM_CONTROL__TXCLK_PRBS_GATE_ENABLE_MASK |
+			 BIF0_CPM_CONTROL__REFCLK_REGS_GATE_ENABLE_MASK);
+	} else {
+		data &= ~(BIF0_CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK |
+			  BIF0_CPM_CONTROL__TXCLK_DYN_GATE_ENABLE_MASK |
+			  BIF0_CPM_CONTROL__TXCLK_LCNT_GATE_ENABLE_MASK |
+			  BIF0_CPM_CONTROL__TXCLK_REGS_GATE_ENABLE_MASK |
+			  BIF0_CPM_CONTROL__TXCLK_PRBS_GATE_ENABLE_MASK |
+			  BIF0_CPM_CONTROL__REFCLK_REGS_GATE_ENABLE_MASK);
+	}
+
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regBIF0_CPM_CONTROL, data);
+}
+
+static void nbio_v7_7_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+						      bool enable)
+{
+	uint32_t def, data;
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_BIF_LS))
+		return;
+
+	def = data = RREG32_SOC15(NBIO, 0, regBIF0_PCIE_CNTL2);
+	if (enable)
+		data |= BIF0_PCIE_CNTL2__SLV_MEM_LS_EN_MASK;
+	else
+		data &= ~BIF0_PCIE_CNTL2__SLV_MEM_LS_EN_MASK;
+
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regBIF0_PCIE_CNTL2, data);
+
+	def = data = RREG32_SOC15(NBIO, 0, regBIF0_PCIE_TX_POWER_CTRL_1);
+	if (enable) {
+		data |= (BIF0_PCIE_TX_POWER_CTRL_1__MST_MEM_LS_EN_MASK |
+			BIF0_PCIE_TX_POWER_CTRL_1__REPLAY_MEM_LS_EN_MASK);
+	} else {
+		data &= ~(BIF0_PCIE_TX_POWER_CTRL_1__MST_MEM_LS_EN_MASK |
+			BIF0_PCIE_TX_POWER_CTRL_1__REPLAY_MEM_LS_EN_MASK);
+	}
+
+	if (def != data)
+		WREG32_SOC15(NBIO, 0, regBIF0_PCIE_TX_POWER_CTRL_1, data);
+}
+
+static void nbio_v7_7_get_clockgating_state(struct amdgpu_device *adev,
+					    u64 *flags)
+{
+	uint32_t data;
+
+	/* AMD_CG_SUPPORT_BIF_MGCG */
+	data = RREG32_SOC15(NBIO, 0, regBIF0_CPM_CONTROL);
+	if (data & BIF0_CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_BIF_MGCG;
+
+	/* AMD_CG_SUPPORT_BIF_LS */
+	data = RREG32_SOC15(NBIO, 0, regBIF0_PCIE_CNTL2);
+	if (data & BIF0_PCIE_CNTL2__SLV_MEM_LS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_BIF_LS;
+}
+
+#define MMIO_REG_HOLE_OFFSET (0x80000 - PAGE_SIZE)
+
+static void nbio_v7_7_set_reg_remap(struct amdgpu_device *adev)
+{
+	if (!amdgpu_sriov_vf(adev) && (PAGE_SIZE <= 4096)) {
+		adev->rmmio_remap.reg_offset = MMIO_REG_HOLE_OFFSET;
+		adev->rmmio_remap.bus_addr = adev->rmmio_base + MMIO_REG_HOLE_OFFSET;
+	} else {
+		adev->rmmio_remap.reg_offset =
+			SOC15_REG_OFFSET(NBIO, 0,
+					 regBIF_BX_PF0_HDP_MEM_COHERENCY_FLUSH_CNTL) << 2;
+		adev->rmmio_remap.bus_addr = 0;
+	}
+}
+
+const struct amdgpu_nbio_funcs nbio_v7_7_funcs = {
+	.get_hdp_flush_req_offset = nbio_v7_7_get_hdp_flush_req_offset,
+	.get_hdp_flush_done_offset = nbio_v7_7_get_hdp_flush_done_offset,
+	.get_pcie_index_offset = nbio_v7_7_get_pcie_index_offset,
+	.get_pcie_data_offset = nbio_v7_7_get_pcie_data_offset,
+	.get_pcie_port_index_offset = nbio_v7_7_get_pcie_port_index_offset,
+	.get_pcie_port_data_offset = nbio_v7_7_get_pcie_port_data_offset,
+	.get_rev_id = nbio_v7_7_get_rev_id,
+	.mc_access_enable = nbio_v7_7_mc_access_enable,
+	.get_memsize = nbio_v7_7_get_memsize,
+	.sdma_doorbell_range = nbio_v7_7_sdma_doorbell_range,
+	.vcn_doorbell_range = nbio_v7_7_vcn_doorbell_range,
+	.enable_doorbell_aperture = nbio_v7_7_enable_doorbell_aperture,
+	.enable_doorbell_selfring_aperture = nbio_v7_7_enable_doorbell_selfring_aperture,
+	.ih_doorbell_range = nbio_v7_7_ih_doorbell_range,
+	.update_medium_grain_clock_gating = nbio_v7_7_update_medium_grain_clock_gating,
+	.update_medium_grain_light_sleep = nbio_v7_7_update_medium_grain_light_sleep,
+	.get_clockgating_state = nbio_v7_7_get_clockgating_state,
+	.ih_control = nbio_v7_7_ih_control,
+	.init_registers = nbio_v7_7_init_registers,
+	.remap_hdp_registers = nbio_v7_7_remap_hdp_registers,
+	.set_reg_remap = nbio_v7_7_set_reg_remap,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_7.h b/drivers/gpu/drm/amd/amdgpu/nbio_v7_7.h
new file mode 100644
index 000000000000..2a33b256ba81
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_7.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __NBIO_V7_7_H__
+#define __NBIO_V7_7_H__
+
+#include "soc15_common.h"
+
+extern const struct nbio_hdp_flush_reg nbio_v7_7_hdp_flush_reg;
+extern const struct amdgpu_nbio_funcs nbio_v7_7_funcs;
+extern const struct amdgpu_nbio_ras_funcs nbio_v7_7_ras_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c
new file mode 100644
index 000000000000..1c22bc11c1f8
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c
@@ -0,0 +1,694 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "nbio_v7_9.h"
+#include "amdgpu_ras.h"
+
+#include "nbio/nbio_7_9_0_offset.h"
+#include "nbio/nbio_7_9_0_sh_mask.h"
+#include "ivsrcid/nbio/irqsrcs_nbif_7_4.h"
+#include <uapi/linux/kfd_ioctl.h>
+
+#define NPS_MODE_MASK 0x000000FFL
+
+static void nbio_v7_9_remap_hdp_registers(struct amdgpu_device *adev)
+{
+	WREG32_SOC15(NBIO, 0, regBIF_BX0_REMAP_HDP_MEM_FLUSH_CNTL,
+		adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL);
+	WREG32_SOC15(NBIO, 0, regBIF_BX0_REMAP_HDP_REG_FLUSH_CNTL,
+		adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_REG_FLUSH_CNTL);
+}
+
+static u32 nbio_v7_9_get_rev_id(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	tmp = IP_VERSION_SUBREV(amdgpu_ip_version_full(adev, NBIO_HWIP, 0));
+	/* If it is VF or subrevision holds a non-zero value, that should be used */
+	if (tmp || amdgpu_sriov_vf(adev))
+		return tmp;
+
+	/* If discovery subrev is not updated, use register version */
+	tmp = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_DEV0_EPF0_STRAP0);
+	tmp = REG_GET_FIELD(tmp, RCC_STRAP0_RCC_DEV0_EPF0_STRAP0,
+			    STRAP_ATI_REV_ID_DEV0_F0);
+
+	return tmp;
+}
+
+static void nbio_v7_9_mc_access_enable(struct amdgpu_device *adev, bool enable)
+{
+	if (enable)
+		WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_FB_EN,
+			BIF_BX0_BIF_FB_EN__FB_READ_EN_MASK | BIF_BX0_BIF_FB_EN__FB_WRITE_EN_MASK);
+	else
+		WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_FB_EN, 0);
+}
+
+static u32 nbio_v7_9_get_memsize(struct amdgpu_device *adev)
+{
+	return RREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF0_RCC_CONFIG_MEMSIZE);
+}
+
+static void nbio_v7_9_sdma_doorbell_range(struct amdgpu_device *adev, int instance,
+			bool use_doorbell, int doorbell_index, int doorbell_size)
+{
+	u32 doorbell_range = 0, doorbell_ctrl = 0;
+	int aid_id, dev_inst;
+
+	dev_inst = GET_INST(SDMA0, instance);
+	aid_id = adev->sdma.instance[instance].aid_id;
+
+	if (use_doorbell == false)
+		return;
+
+	doorbell_range =
+		REG_SET_FIELD(doorbell_range, DOORBELL0_CTRL_ENTRY_0,
+			BIF_DOORBELL0_RANGE_OFFSET_ENTRY, doorbell_index);
+	doorbell_range =
+		REG_SET_FIELD(doorbell_range, DOORBELL0_CTRL_ENTRY_0,
+			BIF_DOORBELL0_RANGE_SIZE_ENTRY, doorbell_size);
+	doorbell_ctrl =
+		REG_SET_FIELD(doorbell_ctrl, S2A_DOORBELL_ENTRY_1_CTRL,
+			S2A_DOORBELL_PORT1_ENABLE, 1);
+	doorbell_ctrl =
+		REG_SET_FIELD(doorbell_ctrl, S2A_DOORBELL_ENTRY_1_CTRL,
+			S2A_DOORBELL_PORT1_RANGE_SIZE, doorbell_size);
+
+	switch (dev_inst % adev->sdma.num_inst_per_aid) {
+	case 0:
+		WREG32_SOC15_OFFSET(NBIO, 0, regDOORBELL0_CTRL_ENTRY_1,
+			4 * aid_id, doorbell_range);
+
+		doorbell_ctrl = REG_SET_FIELD(doorbell_ctrl,
+					S2A_DOORBELL_ENTRY_1_CTRL,
+					S2A_DOORBELL_PORT1_AWID, 0xe);
+		doorbell_ctrl = REG_SET_FIELD(doorbell_ctrl,
+					S2A_DOORBELL_ENTRY_1_CTRL,
+					S2A_DOORBELL_PORT1_RANGE_OFFSET, 0xe);
+		doorbell_ctrl = REG_SET_FIELD(doorbell_ctrl,
+					S2A_DOORBELL_ENTRY_1_CTRL,
+					S2A_DOORBELL_PORT1_AWADDR_31_28_VALUE,
+					0x1);
+		WREG32_SOC15_EXT(NBIO, aid_id, regS2A_DOORBELL_ENTRY_1_CTRL,
+			aid_id, doorbell_ctrl);
+		break;
+	case 1:
+		WREG32_SOC15_OFFSET(NBIO, 0, regDOORBELL0_CTRL_ENTRY_2,
+			4 * aid_id, doorbell_range);
+
+		doorbell_ctrl = REG_SET_FIELD(doorbell_ctrl,
+					S2A_DOORBELL_ENTRY_1_CTRL,
+					S2A_DOORBELL_PORT1_AWID, 0x8);
+		doorbell_ctrl = REG_SET_FIELD(doorbell_ctrl,
+					S2A_DOORBELL_ENTRY_1_CTRL,
+					S2A_DOORBELL_PORT1_RANGE_OFFSET, 0x8);
+		doorbell_ctrl = REG_SET_FIELD(doorbell_ctrl,
+					S2A_DOORBELL_ENTRY_1_CTRL,
+					S2A_DOORBELL_PORT1_AWADDR_31_28_VALUE,
+					0x2);
+		WREG32_SOC15_EXT(NBIO, aid_id, regS2A_DOORBELL_ENTRY_2_CTRL,
+			aid_id, doorbell_ctrl);
+		break;
+	case 2:
+		WREG32_SOC15_OFFSET(NBIO, 0, regDOORBELL0_CTRL_ENTRY_3,
+			4 * aid_id, doorbell_range);
+
+		doorbell_ctrl = REG_SET_FIELD(doorbell_ctrl,
+					S2A_DOORBELL_ENTRY_1_CTRL,
+					S2A_DOORBELL_PORT1_AWID, 0x9);
+		doorbell_ctrl = REG_SET_FIELD(doorbell_ctrl,
+					S2A_DOORBELL_ENTRY_1_CTRL,
+					S2A_DOORBELL_PORT1_RANGE_OFFSET, 0x9);
+		doorbell_ctrl = REG_SET_FIELD(doorbell_ctrl,
+					S2A_DOORBELL_ENTRY_1_CTRL,
+					S2A_DOORBELL_PORT1_AWADDR_31_28_VALUE,
+					0x8);
+		WREG32_SOC15_EXT(NBIO, aid_id, regS2A_DOORBELL_ENTRY_5_CTRL,
+			aid_id, doorbell_ctrl);
+		break;
+	case 3:
+		WREG32_SOC15_OFFSET(NBIO, 0, regDOORBELL0_CTRL_ENTRY_4,
+			4 * aid_id, doorbell_range);
+
+		doorbell_ctrl = REG_SET_FIELD(doorbell_ctrl,
+					S2A_DOORBELL_ENTRY_1_CTRL,
+					S2A_DOORBELL_PORT1_AWID, 0xa);
+		doorbell_ctrl = REG_SET_FIELD(doorbell_ctrl,
+					S2A_DOORBELL_ENTRY_1_CTRL,
+					S2A_DOORBELL_PORT1_RANGE_OFFSET, 0xa);
+		doorbell_ctrl = REG_SET_FIELD(doorbell_ctrl,
+					S2A_DOORBELL_ENTRY_1_CTRL,
+					S2A_DOORBELL_PORT1_AWADDR_31_28_VALUE,
+					0x9);
+		WREG32_SOC15_EXT(NBIO, aid_id, regS2A_DOORBELL_ENTRY_6_CTRL,
+			aid_id, doorbell_ctrl);
+		break;
+	default:
+		break;
+	}
+}
+
+static void nbio_v7_9_vcn_doorbell_range(struct amdgpu_device *adev, bool use_doorbell,
+					 int doorbell_index, int instance)
+{
+	u32 doorbell_range = 0, doorbell_ctrl = 0;
+	u32 aid_id = instance;
+	u32 range_size;
+
+	if (use_doorbell) {
+		range_size = (amdgpu_ip_version(adev, GC_HWIP, 0) ==
+						IP_VERSION(9, 5, 0)) ?
+						0xb : 0x9;
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+				DOORBELL0_CTRL_ENTRY_0,
+				BIF_DOORBELL0_RANGE_OFFSET_ENTRY,
+				doorbell_index);
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+				DOORBELL0_CTRL_ENTRY_0,
+				BIF_DOORBELL0_RANGE_SIZE_ENTRY,
+				range_size);
+		if (aid_id)
+			doorbell_range = REG_SET_FIELD(doorbell_range,
+					DOORBELL0_CTRL_ENTRY_0,
+					DOORBELL0_FENCE_ENABLE_ENTRY,
+					0x4);
+
+		doorbell_ctrl = REG_SET_FIELD(doorbell_ctrl,
+				S2A_DOORBELL_ENTRY_1_CTRL,
+				S2A_DOORBELL_PORT1_ENABLE, 1);
+		doorbell_ctrl = REG_SET_FIELD(doorbell_ctrl,
+				S2A_DOORBELL_ENTRY_1_CTRL,
+				S2A_DOORBELL_PORT1_AWID, 0x4);
+		doorbell_ctrl = REG_SET_FIELD(doorbell_ctrl,
+				S2A_DOORBELL_ENTRY_1_CTRL,
+				S2A_DOORBELL_PORT1_RANGE_OFFSET, 0x4);
+		doorbell_ctrl = REG_SET_FIELD(doorbell_ctrl,
+				S2A_DOORBELL_ENTRY_1_CTRL,
+				S2A_DOORBELL_PORT1_RANGE_SIZE, range_size);
+		doorbell_ctrl = REG_SET_FIELD(doorbell_ctrl,
+				S2A_DOORBELL_ENTRY_1_CTRL,
+				S2A_DOORBELL_PORT1_AWADDR_31_28_VALUE, 0x4);
+
+		WREG32_SOC15_OFFSET(NBIO, 0, regDOORBELL0_CTRL_ENTRY_17,
+					aid_id, doorbell_range);
+		WREG32_SOC15_EXT(NBIO, aid_id, regS2A_DOORBELL_ENTRY_4_CTRL,
+				aid_id, doorbell_ctrl);
+	} else {
+		doorbell_range = REG_SET_FIELD(doorbell_range,
+				DOORBELL0_CTRL_ENTRY_0,
+				BIF_DOORBELL0_RANGE_SIZE_ENTRY, 0);
+		doorbell_ctrl = REG_SET_FIELD(doorbell_ctrl,
+				S2A_DOORBELL_ENTRY_1_CTRL,
+				S2A_DOORBELL_PORT1_RANGE_SIZE, 0);
+
+		WREG32_SOC15_OFFSET(NBIO, 0, regDOORBELL0_CTRL_ENTRY_17,
+					aid_id, doorbell_range);
+		WREG32_SOC15_EXT(NBIO, aid_id, regS2A_DOORBELL_ENTRY_4_CTRL,
+				aid_id, doorbell_ctrl);
+	}
+}
+
+static void nbio_v7_9_enable_doorbell_aperture(struct amdgpu_device *adev,
+					       bool enable)
+{
+	/* Enable to allow doorbell pass thru on pre-silicon bare-metal */
+	WREG32_SOC15(NBIO, 0, regBIFC_DOORBELL_ACCESS_EN_PF, 0xfffff);
+	WREG32_FIELD15_PREREG(NBIO, 0, RCC_DEV0_EPF0_RCC_DOORBELL_APER_EN,
+			BIF_DOORBELL_APER_EN, enable ? 1 : 0);
+}
+
+static void nbio_v7_9_enable_doorbell_selfring_aperture(struct amdgpu_device *adev,
+							bool enable)
+{
+	u32 tmp = 0;
+
+	if (enable) {
+		tmp = REG_SET_FIELD(tmp, BIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL,
+				    DOORBELL_SELFRING_GPA_APER_EN, 1) |
+		      REG_SET_FIELD(tmp, BIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL,
+				    DOORBELL_SELFRING_GPA_APER_MODE, 1) |
+		      REG_SET_FIELD(tmp, BIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL,
+				    DOORBELL_SELFRING_GPA_APER_SIZE, 0);
+
+		WREG32_SOC15(NBIO, 0, regBIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_BASE_LOW,
+			     lower_32_bits(adev->doorbell.base));
+		WREG32_SOC15(NBIO, 0, regBIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_BASE_HIGH,
+			     upper_32_bits(adev->doorbell.base));
+	}
+
+	WREG32_SOC15(NBIO, 0, regBIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL, tmp);
+}
+
+static void nbio_v7_9_ih_doorbell_range(struct amdgpu_device *adev,
+					bool use_doorbell, int doorbell_index)
+{
+	u32 ih_doorbell_range = 0, ih_doorbell_ctrl = 0;
+
+	if (use_doorbell) {
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+				DOORBELL0_CTRL_ENTRY_0,
+				BIF_DOORBELL0_RANGE_OFFSET_ENTRY,
+				doorbell_index);
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+				DOORBELL0_CTRL_ENTRY_0,
+				BIF_DOORBELL0_RANGE_SIZE_ENTRY,
+				0x8);
+
+		ih_doorbell_ctrl = REG_SET_FIELD(ih_doorbell_ctrl,
+				S2A_DOORBELL_ENTRY_1_CTRL,
+				S2A_DOORBELL_PORT1_ENABLE, 1);
+		ih_doorbell_ctrl = REG_SET_FIELD(ih_doorbell_ctrl,
+				S2A_DOORBELL_ENTRY_1_CTRL,
+				S2A_DOORBELL_PORT1_AWID, 0);
+		ih_doorbell_ctrl = REG_SET_FIELD(ih_doorbell_ctrl,
+				S2A_DOORBELL_ENTRY_1_CTRL,
+				S2A_DOORBELL_PORT1_RANGE_OFFSET, 0);
+		ih_doorbell_ctrl = REG_SET_FIELD(ih_doorbell_ctrl,
+				S2A_DOORBELL_ENTRY_1_CTRL,
+				S2A_DOORBELL_PORT1_RANGE_SIZE, 0x8);
+		ih_doorbell_ctrl = REG_SET_FIELD(ih_doorbell_ctrl,
+				S2A_DOORBELL_ENTRY_1_CTRL,
+				S2A_DOORBELL_PORT1_AWADDR_31_28_VALUE, 0);
+	} else {
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
+				DOORBELL0_CTRL_ENTRY_0,
+				BIF_DOORBELL0_RANGE_SIZE_ENTRY, 0);
+		ih_doorbell_ctrl = REG_SET_FIELD(ih_doorbell_ctrl,
+				S2A_DOORBELL_ENTRY_1_CTRL,
+				S2A_DOORBELL_PORT1_RANGE_SIZE, 0);
+	}
+
+	WREG32_SOC15(NBIO, 0, regDOORBELL0_CTRL_ENTRY_0, ih_doorbell_range);
+	WREG32_SOC15(NBIO, 0, regS2A_DOORBELL_ENTRY_3_CTRL, ih_doorbell_ctrl);
+}
+
+
+static void nbio_v7_9_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						       bool enable)
+{
+}
+
+static void nbio_v7_9_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+						      bool enable)
+{
+}
+
+static void nbio_v7_9_get_clockgating_state(struct amdgpu_device *adev,
+					    u64 *flags)
+{
+}
+
+static void nbio_v7_9_ih_control(struct amdgpu_device *adev)
+{
+	u32 interrupt_cntl;
+
+	/* setup interrupt control */
+	WREG32_SOC15(NBIO, 0, regBIF_BX0_INTERRUPT_CNTL2, adev->dummy_page_addr >> 8);
+	interrupt_cntl = RREG32_SOC15(NBIO, 0, regBIF_BX0_INTERRUPT_CNTL);
+	/* INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
+	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
+	 */
+	interrupt_cntl =
+		REG_SET_FIELD(interrupt_cntl, BIF_BX0_INTERRUPT_CNTL, IH_DUMMY_RD_OVERRIDE, 0);
+	/* INTERRUPT_CNTL__IH_REQ_NONSNOOP_EN_MASK=1 if ring is in non-cacheable memory, e.g., vram */
+	interrupt_cntl =
+		REG_SET_FIELD(interrupt_cntl, BIF_BX0_INTERRUPT_CNTL, IH_REQ_NONSNOOP_EN, 0);
+	WREG32_SOC15(NBIO, 0, regBIF_BX0_INTERRUPT_CNTL, interrupt_cntl);
+}
+
+static u32 nbio_v7_9_get_hdp_flush_req_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF0_GPU_HDP_FLUSH_REQ);
+}
+
+static u32 nbio_v7_9_get_hdp_flush_done_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF0_GPU_HDP_FLUSH_DONE);
+}
+
+static u32 nbio_v7_9_get_pcie_index_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX0_PCIE_INDEX2);
+}
+
+static u32 nbio_v7_9_get_pcie_data_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX0_PCIE_DATA2);
+}
+
+static u32 nbio_v7_9_get_pcie_index_hi_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX0_PCIE_INDEX2_HI);
+}
+
+const struct nbio_hdp_flush_reg nbio_v7_9_hdp_flush_reg = {
+	.ref_and_mask_cp0 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP0_MASK,
+	.ref_and_mask_cp1 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP1_MASK,
+	.ref_and_mask_cp2 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP2_MASK,
+	.ref_and_mask_cp3 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP3_MASK,
+	.ref_and_mask_cp4 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP4_MASK,
+	.ref_and_mask_cp5 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP5_MASK,
+	.ref_and_mask_cp6 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP6_MASK,
+	.ref_and_mask_cp7 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP7_MASK,
+	.ref_and_mask_cp8 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP8_MASK,
+	.ref_and_mask_cp9 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP9_MASK,
+	.ref_and_mask_sdma0 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__SDMA0_MASK,
+	.ref_and_mask_sdma1 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__SDMA1_MASK,
+	.ref_and_mask_sdma2 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__RSVD_ENG0_MASK,
+	.ref_and_mask_sdma3 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__RSVD_ENG1_MASK,
+	.ref_and_mask_sdma4 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__RSVD_ENG2_MASK,
+	.ref_and_mask_sdma5 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__RSVD_ENG3_MASK,
+	.ref_and_mask_sdma6 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__RSVD_ENG4_MASK,
+	.ref_and_mask_sdma7 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__RSVD_ENG5_MASK,
+};
+
+static void nbio_v7_9_enable_doorbell_interrupt(struct amdgpu_device *adev,
+						bool enable)
+{
+	WREG32_FIELD15_PREREG(NBIO, 0, BIF_BX0_BIF_DOORBELL_INT_CNTL,
+			      DOORBELL_INTERRUPT_DISABLE, enable ? 0 : 1);
+}
+
+static int nbio_v7_9_get_compute_partition_mode(struct amdgpu_device *adev)
+{
+	u32 tmp, px;
+
+	tmp = RREG32_SOC15(NBIO, 0, regBIF_BX_PF0_PARTITION_COMPUTE_STATUS);
+	px = REG_GET_FIELD(tmp, BIF_BX_PF0_PARTITION_COMPUTE_STATUS,
+			   PARTITION_MODE);
+
+	return px;
+}
+
+static bool nbio_v7_9_is_nps_switch_requested(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	tmp = RREG32_SOC15(NBIO, 0, regBIF_BX_PF0_PARTITION_MEM_STATUS);
+	tmp = REG_GET_FIELD(tmp, BIF_BX_PF0_PARTITION_MEM_STATUS,
+			    CHANGE_STATUE);
+
+	/* 0x8 - NPS switch requested */
+	return (tmp == 0x8);
+}
+static u32 nbio_v7_9_get_memory_partition_mode(struct amdgpu_device *adev,
+					       u32 *supp_modes)
+{
+	u32 tmp;
+
+	tmp = RREG32_SOC15(NBIO, 0, regBIF_BX_PF0_PARTITION_MEM_STATUS);
+	tmp = REG_GET_FIELD(tmp, BIF_BX_PF0_PARTITION_MEM_STATUS, NPS_MODE);
+
+	if (supp_modes) {
+		*supp_modes =
+			RREG32_SOC15(NBIO, 0, regBIF_BX_PF0_PARTITION_MEM_CAP);
+	}
+
+	return ffs(tmp);
+}
+
+static void nbio_v7_9_init_registers(struct amdgpu_device *adev)
+{
+	u32 inst_mask;
+	int i;
+
+	WREG32_SOC15(NBIO, 0, regXCC_DOORBELL_FENCE,
+		0xff & ~(adev->gfx.xcc_mask));
+
+	WREG32_SOC15(NBIO, 0, regBIFC_GFX_INT_MONITOR_MASK, 0x7ff);
+
+	inst_mask = adev->aid_mask & ~1U;
+	for_each_inst(i, inst_mask) {
+		WREG32_SOC15_EXT(NBIO, i, regXCC_DOORBELL_FENCE, i,
+			XCC_DOORBELL_FENCE__SHUB_SLV_MODE_MASK);
+
+	}
+
+	if (!amdgpu_sriov_vf(adev)) {
+		u32 baco_cntl;
+		for_each_inst(i, adev->aid_mask) {
+			baco_cntl = RREG32_SOC15(NBIO, i, regBIF_BX0_BACO_CNTL);
+			if (baco_cntl & (BIF_BX0_BACO_CNTL__BACO_DUMMY_EN_MASK |
+					 BIF_BX0_BACO_CNTL__BACO_EN_MASK)) {
+				baco_cntl &= ~(
+					BIF_BX0_BACO_CNTL__BACO_DUMMY_EN_MASK |
+					BIF_BX0_BACO_CNTL__BACO_EN_MASK);
+				dev_dbg(adev->dev,
+					"Unsetting baco dummy mode %x",
+					baco_cntl);
+				WREG32_SOC15(NBIO, i, regBIF_BX0_BACO_CNTL,
+					     baco_cntl);
+			}
+		}
+	}
+}
+
+#define MMIO_REG_HOLE_OFFSET 0x1A000
+
+static void nbio_v7_9_set_reg_remap(struct amdgpu_device *adev)
+{
+	if (!amdgpu_sriov_vf(adev) && (PAGE_SIZE <= 4096)) {
+		adev->rmmio_remap.reg_offset = MMIO_REG_HOLE_OFFSET;
+		adev->rmmio_remap.bus_addr = adev->rmmio_base + MMIO_REG_HOLE_OFFSET;
+	} else {
+		adev->rmmio_remap.reg_offset =
+			SOC15_REG_OFFSET(
+				NBIO, 0,
+				regBIF_BX_DEV0_EPF0_VF0_HDP_MEM_COHERENCY_FLUSH_CNTL)
+			<< 2;
+		adev->rmmio_remap.bus_addr = 0;
+	}
+}
+
+const struct amdgpu_nbio_funcs nbio_v7_9_funcs = {
+	.get_hdp_flush_req_offset = nbio_v7_9_get_hdp_flush_req_offset,
+	.get_hdp_flush_done_offset = nbio_v7_9_get_hdp_flush_done_offset,
+	.get_pcie_index_offset = nbio_v7_9_get_pcie_index_offset,
+	.get_pcie_data_offset = nbio_v7_9_get_pcie_data_offset,
+	.get_pcie_index_hi_offset = nbio_v7_9_get_pcie_index_hi_offset,
+	.get_rev_id = nbio_v7_9_get_rev_id,
+	.mc_access_enable = nbio_v7_9_mc_access_enable,
+	.get_memsize = nbio_v7_9_get_memsize,
+	.sdma_doorbell_range = nbio_v7_9_sdma_doorbell_range,
+	.vcn_doorbell_range = nbio_v7_9_vcn_doorbell_range,
+	.enable_doorbell_aperture = nbio_v7_9_enable_doorbell_aperture,
+	.enable_doorbell_selfring_aperture = nbio_v7_9_enable_doorbell_selfring_aperture,
+	.ih_doorbell_range = nbio_v7_9_ih_doorbell_range,
+	.enable_doorbell_interrupt = nbio_v7_9_enable_doorbell_interrupt,
+	.update_medium_grain_clock_gating = nbio_v7_9_update_medium_grain_clock_gating,
+	.update_medium_grain_light_sleep = nbio_v7_9_update_medium_grain_light_sleep,
+	.get_clockgating_state = nbio_v7_9_get_clockgating_state,
+	.ih_control = nbio_v7_9_ih_control,
+	.remap_hdp_registers = nbio_v7_9_remap_hdp_registers,
+	.get_compute_partition_mode = nbio_v7_9_get_compute_partition_mode,
+	.get_memory_partition_mode = nbio_v7_9_get_memory_partition_mode,
+	.is_nps_switch_requested = nbio_v7_9_is_nps_switch_requested,
+	.init_registers = nbio_v7_9_init_registers,
+	.set_reg_remap = nbio_v7_9_set_reg_remap,
+};
+
+static void nbio_v7_9_query_ras_error_count(struct amdgpu_device *adev,
+					void *ras_error_status)
+{
+}
+
+static void nbio_v7_9_handle_ras_controller_intr_no_bifring(struct amdgpu_device *adev)
+{
+	uint32_t bif_doorbell_intr_cntl;
+	struct ras_manager *obj = amdgpu_ras_find_obj(adev, adev->nbio.ras_if);
+	struct ras_err_data err_data;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+
+	if (amdgpu_ras_error_data_init(&err_data))
+		return;
+
+	bif_doorbell_intr_cntl = RREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_DOORBELL_INT_CNTL);
+
+	if (REG_GET_FIELD(bif_doorbell_intr_cntl,
+		BIF_BX0_BIF_DOORBELL_INT_CNTL, RAS_CNTLR_INTERRUPT_STATUS)) {
+		/* driver has to clear the interrupt status when bif ring is disabled */
+		bif_doorbell_intr_cntl = REG_SET_FIELD(bif_doorbell_intr_cntl,
+						BIF_BX0_BIF_DOORBELL_INT_CNTL,
+						RAS_CNTLR_INTERRUPT_CLEAR, 1);
+		WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_DOORBELL_INT_CNTL, bif_doorbell_intr_cntl);
+
+		if (!ras->disable_ras_err_cnt_harvest) {
+			/*
+			 * clear error status after ras_controller_intr
+			 * according to hw team and count ue number
+			 * for query
+			 */
+			nbio_v7_9_query_ras_error_count(adev, &err_data);
+
+			/* logging on error cnt and printing for awareness */
+			obj->err_data.ue_count += err_data.ue_count;
+			obj->err_data.ce_count += err_data.ce_count;
+
+			if (err_data.ce_count)
+				dev_info(adev->dev, "%ld correctable hardware "
+						"errors detected in %s block\n",
+						obj->err_data.ce_count,
+						get_ras_block_str(adev->nbio.ras_if));
+
+			if (err_data.ue_count)
+				dev_info(adev->dev, "%ld uncorrectable hardware "
+						"errors detected in %s block\n",
+						obj->err_data.ue_count,
+						get_ras_block_str(adev->nbio.ras_if));
+		}
+
+		dev_info(adev->dev, "RAS controller interrupt triggered "
+					"by NBIF error\n");
+	}
+
+	amdgpu_ras_error_data_fini(&err_data);
+}
+
+static void nbio_v7_9_handle_ras_err_event_athub_intr_no_bifring(struct amdgpu_device *adev)
+{
+	uint32_t bif_doorbell_intr_cntl;
+
+	bif_doorbell_intr_cntl = RREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_DOORBELL_INT_CNTL);
+
+	if (REG_GET_FIELD(bif_doorbell_intr_cntl,
+		BIF_BX0_BIF_DOORBELL_INT_CNTL, RAS_ATHUB_ERR_EVENT_INTERRUPT_STATUS)) {
+		/* driver has to clear the interrupt status when bif ring is disabled */
+		bif_doorbell_intr_cntl = REG_SET_FIELD(bif_doorbell_intr_cntl,
+						BIF_BX0_BIF_DOORBELL_INT_CNTL,
+						RAS_ATHUB_ERR_EVENT_INTERRUPT_CLEAR, 1);
+
+		WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_DOORBELL_INT_CNTL, bif_doorbell_intr_cntl);
+
+		amdgpu_ras_global_ras_isr(adev);
+	}
+}
+
+static int nbio_v7_9_set_ras_controller_irq_state(struct amdgpu_device *adev,
+						  struct amdgpu_irq_src *src,
+						  unsigned type,
+						  enum amdgpu_interrupt_state state)
+{
+	/* Dummy function, there is no initialization operation in driver */
+
+	return 0;
+}
+
+static int nbio_v7_9_process_ras_controller_irq(struct amdgpu_device *adev,
+						struct amdgpu_irq_src *source,
+						struct amdgpu_iv_entry *entry)
+{
+	/* By design, the ih cookie for ras_controller_irq should be written
+	 * to BIFring instead of general iv ring. However, due to known bif ring
+	 * hw bug, it has to be disabled. There is no chance the process function
+	 * will be involked. Just left it as a dummy one.
+	 */
+	return 0;
+}
+
+static int nbio_v7_9_set_ras_err_event_athub_irq_state(struct amdgpu_device *adev,
+						       struct amdgpu_irq_src *src,
+						       unsigned type,
+						       enum amdgpu_interrupt_state state)
+{
+	/* Dummy function, there is no initialization operation in driver */
+
+	return 0;
+}
+
+static int nbio_v7_9_process_err_event_athub_irq(struct amdgpu_device *adev,
+						 struct amdgpu_irq_src *source,
+						 struct amdgpu_iv_entry *entry)
+{
+	/* By design, the ih cookie for err_event_athub_irq should be written
+	 * to BIFring instead of general iv ring. However, due to known bif ring
+	 * hw bug, it has to be disabled. There is no chance the process function
+	 * will be involked. Just left it as a dummy one.
+	 */
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs nbio_v7_9_ras_controller_irq_funcs = {
+	.set = nbio_v7_9_set_ras_controller_irq_state,
+	.process = nbio_v7_9_process_ras_controller_irq,
+};
+
+static const struct amdgpu_irq_src_funcs nbio_v7_9_ras_err_event_athub_irq_funcs = {
+	.set = nbio_v7_9_set_ras_err_event_athub_irq_state,
+	.process = nbio_v7_9_process_err_event_athub_irq,
+};
+
+static int nbio_v7_9_init_ras_controller_interrupt (struct amdgpu_device *adev)
+{
+	int r;
+
+	/* init the irq funcs */
+	adev->nbio.ras_controller_irq.funcs =
+		&nbio_v7_9_ras_controller_irq_funcs;
+	adev->nbio.ras_controller_irq.num_types = 1;
+
+	/* register ras controller interrupt */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_BIF,
+			      NBIF_7_4__SRCID__RAS_CONTROLLER_INTERRUPT,
+			      &adev->nbio.ras_controller_irq);
+
+	return r;
+}
+
+static int nbio_v7_9_init_ras_err_event_athub_interrupt (struct amdgpu_device *adev)
+{
+
+	int r;
+
+	/* init the irq funcs */
+	adev->nbio.ras_err_event_athub_irq.funcs =
+		&nbio_v7_9_ras_err_event_athub_irq_funcs;
+	adev->nbio.ras_err_event_athub_irq.num_types = 1;
+
+	/* register ras err event athub interrupt */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_BIF,
+			      NBIF_7_4__SRCID__ERREVENT_ATHUB_INTERRUPT,
+			      &adev->nbio.ras_err_event_athub_irq);
+
+	return r;
+}
+
+const struct amdgpu_ras_block_hw_ops nbio_v7_9_ras_hw_ops = {
+	.query_ras_error_count = nbio_v7_9_query_ras_error_count,
+};
+
+struct amdgpu_nbio_ras nbio_v7_9_ras = {
+	.ras_block = {
+		.ras_comm = {
+			.name = "pcie_bif",
+			.block = AMDGPU_RAS_BLOCK__PCIE_BIF,
+			.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE,
+		},
+		.hw_ops = &nbio_v7_9_ras_hw_ops,
+		.ras_late_init = amdgpu_nbio_ras_late_init,
+	},
+	.handle_ras_controller_intr_no_bifring = nbio_v7_9_handle_ras_controller_intr_no_bifring,
+	.handle_ras_err_event_athub_intr_no_bifring = nbio_v7_9_handle_ras_err_event_athub_intr_no_bifring,
+	.init_ras_controller_interrupt = nbio_v7_9_init_ras_controller_interrupt,
+	.init_ras_err_event_athub_interrupt = nbio_v7_9_init_ras_err_event_athub_interrupt,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.h b/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.h
new file mode 100644
index 000000000000..73709771950d
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __NBIO_V7_9_H__
+#define __NBIO_V7_9_H__
+
+#include "soc15_common.h"
+
+extern const struct nbio_hdp_flush_reg nbio_v7_9_hdp_flush_reg;
+extern const struct amdgpu_nbio_funcs nbio_v7_9_funcs;
+extern struct amdgpu_nbio_ras nbio_v7_9_ras;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/nv.c b/drivers/gpu/drm/amd/amdgpu/nv.c
new file mode 100644
index 000000000000..50e77d9b30af
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/nv.c
@@ -0,0 +1,1108 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/firmware.h>
+#include <linux/slab.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+
+#include <drm/amdgpu_drm.h>
+
+#include "amdgpu.h"
+#include "amdgpu_atombios.h"
+#include "amdgpu_ih.h"
+#include "amdgpu_uvd.h"
+#include "amdgpu_vce.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_psp.h"
+#include "atom.h"
+#include "amd_pcie.h"
+
+#include "gc/gc_10_1_0_offset.h"
+#include "gc/gc_10_1_0_sh_mask.h"
+#include "mp/mp_11_0_offset.h"
+
+#include "soc15.h"
+#include "soc15_common.h"
+#include "gmc_v10_0.h"
+#include "gfxhub_v2_0.h"
+#include "mmhub_v2_0.h"
+#include "nbio_v2_3.h"
+#include "nbio_v7_2.h"
+#include "hdp_v5_0.h"
+#include "nv.h"
+#include "navi10_ih.h"
+#include "gfx_v10_0.h"
+#include "sdma_v5_0.h"
+#include "sdma_v5_2.h"
+#include "vcn_v2_0.h"
+#include "jpeg_v2_0.h"
+#include "vcn_v3_0.h"
+#include "jpeg_v3_0.h"
+#include "amdgpu_vkms.h"
+#include "mxgpu_nv.h"
+#include "smuio_v11_0.h"
+#include "smuio_v11_0_6.h"
+
+static const struct amd_ip_funcs nv_common_ip_funcs;
+
+/* Navi */
+static const struct amdgpu_video_codec_info nv_video_codecs_encode_array[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 4096, 4096, 0)},
+};
+
+static const struct amdgpu_video_codecs nv_video_codecs_encode = {
+	.codec_count = ARRAY_SIZE(nv_video_codecs_encode_array),
+	.codec_array = nv_video_codecs_encode_array,
+};
+
+/* Navi1x */
+static const struct amdgpu_video_codec_info nv_video_codecs_decode_array[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, 1920, 1088, 3)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, 1920, 1088, 5)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, 1920, 1088, 4)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 8192, 8192, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)},
+};
+
+static const struct amdgpu_video_codecs nv_video_codecs_decode = {
+	.codec_count = ARRAY_SIZE(nv_video_codecs_decode_array),
+	.codec_array = nv_video_codecs_decode_array,
+};
+
+/* Sienna Cichlid */
+static const struct amdgpu_video_codec_info sc_video_codecs_encode_array[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 0)},
+};
+
+static const struct amdgpu_video_codecs sc_video_codecs_encode = {
+	.codec_count = ARRAY_SIZE(sc_video_codecs_encode_array),
+	.codec_array = sc_video_codecs_encode_array,
+};
+
+static const struct amdgpu_video_codec_info sc_video_codecs_decode_array_vcn0[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, 1920, 1088, 3)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, 1920, 1088, 5)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, 1920, 1088, 4)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1, 8192, 4352, 0)},
+};
+
+static const struct amdgpu_video_codec_info sc_video_codecs_decode_array_vcn1[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, 1920, 1088, 3)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, 1920, 1088, 5)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, 1920, 1088, 4)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)},
+};
+
+static const struct amdgpu_video_codecs sc_video_codecs_decode_vcn0 = {
+	.codec_count = ARRAY_SIZE(sc_video_codecs_decode_array_vcn0),
+	.codec_array = sc_video_codecs_decode_array_vcn0,
+};
+
+static const struct amdgpu_video_codecs sc_video_codecs_decode_vcn1 = {
+	.codec_count = ARRAY_SIZE(sc_video_codecs_decode_array_vcn1),
+	.codec_array = sc_video_codecs_decode_array_vcn1,
+};
+
+/* SRIOV Sienna Cichlid, not const since data is controlled by host */
+static struct amdgpu_video_codec_info sriov_sc_video_codecs_encode_array[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 0)},
+};
+
+static struct amdgpu_video_codec_info sriov_sc_video_codecs_decode_array_vcn0[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, 1920, 1088, 3)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, 1920, 1088, 5)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, 1920, 1088, 4)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1, 8192, 4352, 0)},
+};
+
+static struct amdgpu_video_codec_info sriov_sc_video_codecs_decode_array_vcn1[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, 1920, 1088, 3)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, 1920, 1088, 5)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, 1920, 1088, 4)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)},
+};
+
+static struct amdgpu_video_codecs sriov_sc_video_codecs_encode = {
+	.codec_count = ARRAY_SIZE(sriov_sc_video_codecs_encode_array),
+	.codec_array = sriov_sc_video_codecs_encode_array,
+};
+
+static struct amdgpu_video_codecs sriov_sc_video_codecs_decode_vcn0 = {
+	.codec_count = ARRAY_SIZE(sriov_sc_video_codecs_decode_array_vcn0),
+	.codec_array = sriov_sc_video_codecs_decode_array_vcn0,
+};
+
+static struct amdgpu_video_codecs sriov_sc_video_codecs_decode_vcn1 = {
+	.codec_count = ARRAY_SIZE(sriov_sc_video_codecs_decode_array_vcn1),
+	.codec_array = sriov_sc_video_codecs_decode_array_vcn1,
+};
+
+/* Beige Goby*/
+static const struct amdgpu_video_codec_info bg_video_codecs_decode_array[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)},
+};
+
+static const struct amdgpu_video_codecs bg_video_codecs_decode = {
+	.codec_count = ARRAY_SIZE(bg_video_codecs_decode_array),
+	.codec_array = bg_video_codecs_decode_array,
+};
+
+static const struct amdgpu_video_codecs bg_video_codecs_encode = {
+	.codec_count = 0,
+	.codec_array = NULL,
+};
+
+/* Yellow Carp*/
+static const struct amdgpu_video_codec_info yc_video_codecs_decode_array[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1, 8192, 4352, 0)},
+};
+
+static const struct amdgpu_video_codecs yc_video_codecs_decode = {
+	.codec_count = ARRAY_SIZE(yc_video_codecs_decode_array),
+	.codec_array = yc_video_codecs_decode_array,
+};
+
+static int nv_query_video_codecs(struct amdgpu_device *adev, bool encode,
+				 const struct amdgpu_video_codecs **codecs)
+{
+	if (adev->vcn.num_vcn_inst == hweight8(adev->vcn.harvest_config))
+		return -EINVAL;
+
+	switch (amdgpu_ip_version(adev, UVD_HWIP, 0)) {
+	case IP_VERSION(3, 0, 0):
+	case IP_VERSION(3, 0, 64):
+	case IP_VERSION(3, 0, 192):
+		if (amdgpu_sriov_vf(adev)) {
+			if (adev->vcn.harvest_config & AMDGPU_VCN_HARVEST_VCN0) {
+				if (encode)
+					*codecs = &sriov_sc_video_codecs_encode;
+				else
+					*codecs = &sriov_sc_video_codecs_decode_vcn1;
+			} else {
+				if (encode)
+					*codecs = &sriov_sc_video_codecs_encode;
+				else
+					*codecs = &sriov_sc_video_codecs_decode_vcn0;
+			}
+		} else {
+			if (adev->vcn.harvest_config & AMDGPU_VCN_HARVEST_VCN0) {
+				if (encode)
+					*codecs = &sc_video_codecs_encode;
+				else
+					*codecs = &sc_video_codecs_decode_vcn1;
+			} else {
+				if (encode)
+					*codecs = &sc_video_codecs_encode;
+				else
+					*codecs = &sc_video_codecs_decode_vcn0;
+			}
+		}
+		return 0;
+	case IP_VERSION(3, 0, 16):
+	case IP_VERSION(3, 0, 2):
+		if (encode)
+			*codecs = &sc_video_codecs_encode;
+		else
+			*codecs = &sc_video_codecs_decode_vcn0;
+		return 0;
+	case IP_VERSION(3, 1, 1):
+	case IP_VERSION(3, 1, 2):
+		if (encode)
+			*codecs = &sc_video_codecs_encode;
+		else
+			*codecs = &yc_video_codecs_decode;
+		return 0;
+	case IP_VERSION(3, 0, 33):
+		if (encode)
+			*codecs = &bg_video_codecs_encode;
+		else
+			*codecs = &bg_video_codecs_decode;
+		return 0;
+	case IP_VERSION(2, 0, 0):
+	case IP_VERSION(2, 0, 2):
+		if (encode)
+			*codecs = &nv_video_codecs_encode;
+		else
+			*codecs = &nv_video_codecs_decode;
+		return 0;
+	default:
+		return -EINVAL;
+	}
+}
+
+static u32 nv_didt_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags, address, data;
+	u32 r;
+
+	address = SOC15_REG_OFFSET(GC, 0, mmDIDT_IND_INDEX);
+	data = SOC15_REG_OFFSET(GC, 0, mmDIDT_IND_DATA);
+
+	spin_lock_irqsave(&adev->didt_idx_lock, flags);
+	WREG32(address, (reg));
+	r = RREG32(data);
+	spin_unlock_irqrestore(&adev->didt_idx_lock, flags);
+	return r;
+}
+
+static void nv_didt_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags, address, data;
+
+	address = SOC15_REG_OFFSET(GC, 0, mmDIDT_IND_INDEX);
+	data = SOC15_REG_OFFSET(GC, 0, mmDIDT_IND_DATA);
+
+	spin_lock_irqsave(&adev->didt_idx_lock, flags);
+	WREG32(address, (reg));
+	WREG32(data, (v));
+	spin_unlock_irqrestore(&adev->didt_idx_lock, flags);
+}
+
+static u32 nv_get_config_memsize(struct amdgpu_device *adev)
+{
+	return adev->nbio.funcs->get_memsize(adev);
+}
+
+static u32 nv_get_xclk(struct amdgpu_device *adev)
+{
+	return adev->clock.spll.reference_freq;
+}
+
+
+void nv_grbm_select(struct amdgpu_device *adev,
+		     u32 me, u32 pipe, u32 queue, u32 vmid)
+{
+	u32 grbm_gfx_cntl = 0;
+	grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, PIPEID, pipe);
+	grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, MEID, me);
+	grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, VMID, vmid);
+	grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, QUEUEID, queue);
+
+	WREG32_SOC15(GC, 0, mmGRBM_GFX_CNTL, grbm_gfx_cntl);
+}
+
+static bool nv_read_disabled_bios(struct amdgpu_device *adev)
+{
+	/* todo */
+	return false;
+}
+
+static struct soc15_allowed_register_entry nv_allowed_read_registers[] = {
+	{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS)},
+	{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS2)},
+	{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS_SE0)},
+	{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS_SE1)},
+	{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS_SE2)},
+	{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS_SE3)},
+	{ SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_STATUS_REG)},
+	{ SOC15_REG_ENTRY(SDMA1, 0, mmSDMA1_STATUS_REG)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_STAT)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_STALLED_STAT1)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_STALLED_STAT2)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_STALLED_STAT3)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_CPF_BUSY_STAT)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_CPF_STALLED_STAT1)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_CPF_STATUS)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_CPC_BUSY_STAT)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_CPC_STALLED_STAT1)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_CPC_STATUS)},
+	{ SOC15_REG_ENTRY(GC, 0, mmGB_ADDR_CONFIG)},
+};
+
+static uint32_t nv_read_indexed_register(struct amdgpu_device *adev, u32 se_num,
+					 u32 sh_num, u32 reg_offset)
+{
+	uint32_t val;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	if (se_num != 0xffffffff || sh_num != 0xffffffff)
+		amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff, 0);
+
+	val = RREG32(reg_offset);
+
+	if (se_num != 0xffffffff || sh_num != 0xffffffff)
+		amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
+	mutex_unlock(&adev->grbm_idx_mutex);
+	return val;
+}
+
+static uint32_t nv_get_register_value(struct amdgpu_device *adev,
+				      bool indexed, u32 se_num,
+				      u32 sh_num, u32 reg_offset)
+{
+	if (indexed) {
+		return nv_read_indexed_register(adev, se_num, sh_num, reg_offset);
+	} else {
+		if (reg_offset == SOC15_REG_OFFSET(GC, 0, mmGB_ADDR_CONFIG))
+			return adev->gfx.config.gb_addr_config;
+		return RREG32(reg_offset);
+	}
+}
+
+static int nv_read_register(struct amdgpu_device *adev, u32 se_num,
+			    u32 sh_num, u32 reg_offset, u32 *value)
+{
+	uint32_t i;
+	struct soc15_allowed_register_entry  *en;
+
+	*value = 0;
+	for (i = 0; i < ARRAY_SIZE(nv_allowed_read_registers); i++) {
+		en = &nv_allowed_read_registers[i];
+		if (!adev->reg_offset[en->hwip][en->inst])
+			continue;
+		else if (reg_offset != (adev->reg_offset[en->hwip][en->inst][en->seg]
+					+ en->reg_offset))
+			continue;
+
+		*value = nv_get_register_value(adev,
+					       nv_allowed_read_registers[i].grbm_indexed,
+					       se_num, sh_num, reg_offset);
+		return 0;
+	}
+	return -EINVAL;
+}
+
+static int nv_asic_mode2_reset(struct amdgpu_device *adev)
+{
+	u32 i;
+	int ret = 0;
+
+	amdgpu_atombios_scratch_regs_engine_hung(adev, true);
+
+	/* disable BM */
+	pci_clear_master(adev->pdev);
+
+	amdgpu_device_cache_pci_state(adev->pdev);
+
+	ret = amdgpu_dpm_mode2_reset(adev);
+	if (ret)
+		dev_err(adev->dev, "GPU mode2 reset failed\n");
+
+	amdgpu_device_load_pci_state(adev->pdev);
+
+	/* wait for asic to come out of reset */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		u32 memsize = adev->nbio.funcs->get_memsize(adev);
+
+		if (memsize != 0xffffffff)
+			break;
+		udelay(1);
+	}
+
+	amdgpu_atombios_scratch_regs_engine_hung(adev, false);
+
+	return ret;
+}
+
+static enum amd_reset_method
+nv_asic_reset_method(struct amdgpu_device *adev)
+{
+	if (amdgpu_reset_method == AMD_RESET_METHOD_MODE1 ||
+	    amdgpu_reset_method == AMD_RESET_METHOD_MODE2 ||
+	    amdgpu_reset_method == AMD_RESET_METHOD_BACO ||
+	    amdgpu_reset_method == AMD_RESET_METHOD_PCI)
+		return amdgpu_reset_method;
+
+	if (amdgpu_reset_method != -1)
+		dev_warn(adev->dev, "Specified reset method:%d isn't supported, using AUTO instead.\n",
+				  amdgpu_reset_method);
+
+	switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) {
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(13, 0, 1):
+	case IP_VERSION(13, 0, 3):
+	case IP_VERSION(13, 0, 5):
+	case IP_VERSION(13, 0, 8):
+		return AMD_RESET_METHOD_MODE2;
+	case IP_VERSION(11, 0, 7):
+	case IP_VERSION(11, 0, 11):
+	case IP_VERSION(11, 0, 12):
+	case IP_VERSION(11, 0, 13):
+		return AMD_RESET_METHOD_MODE1;
+	default:
+		if (amdgpu_dpm_is_baco_supported(adev))
+			return AMD_RESET_METHOD_BACO;
+		else
+			return AMD_RESET_METHOD_MODE1;
+	}
+}
+
+static int nv_asic_reset(struct amdgpu_device *adev)
+{
+	int ret = 0;
+
+	switch (nv_asic_reset_method(adev)) {
+	case AMD_RESET_METHOD_PCI:
+		dev_info(adev->dev, "PCI reset\n");
+		ret = amdgpu_device_pci_reset(adev);
+		break;
+	case AMD_RESET_METHOD_BACO:
+		dev_info(adev->dev, "BACO reset\n");
+		ret = amdgpu_dpm_baco_reset(adev);
+		break;
+	case AMD_RESET_METHOD_MODE2:
+		dev_info(adev->dev, "MODE2 reset\n");
+		ret = nv_asic_mode2_reset(adev);
+		break;
+	default:
+		dev_info(adev->dev, "MODE1 reset\n");
+		ret = amdgpu_device_mode1_reset(adev);
+		break;
+	}
+
+	return ret;
+}
+
+static int nv_set_uvd_clocks(struct amdgpu_device *adev, u32 vclk, u32 dclk)
+{
+	/* todo */
+	return 0;
+}
+
+static int nv_set_vce_clocks(struct amdgpu_device *adev, u32 evclk, u32 ecclk)
+{
+	/* todo */
+	return 0;
+}
+
+static void nv_program_aspm(struct amdgpu_device *adev)
+{
+	if (!amdgpu_device_should_use_aspm(adev))
+		return;
+
+	if (adev->nbio.funcs->program_aspm)
+		adev->nbio.funcs->program_aspm(adev);
+
+}
+
+const struct amdgpu_ip_block_version nv_common_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_COMMON,
+	.major = 1,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &nv_common_ip_funcs,
+};
+
+void nv_set_virt_ops(struct amdgpu_device *adev)
+{
+	adev->virt.ops = &xgpu_nv_virt_ops;
+}
+
+static bool nv_need_full_reset(struct amdgpu_device *adev)
+{
+	return true;
+}
+
+static bool nv_need_reset_on_init(struct amdgpu_device *adev)
+{
+	u32 sol_reg;
+
+	if (adev->flags & AMD_IS_APU)
+		return false;
+
+	/* Check sOS sign of life register to confirm sys driver and sOS
+	 * are already been loaded.
+	 */
+	sol_reg = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_81);
+	if (sol_reg)
+		return true;
+
+	return false;
+}
+
+static void nv_init_doorbell_index(struct amdgpu_device *adev)
+{
+	adev->doorbell_index.kiq = AMDGPU_NAVI10_DOORBELL_KIQ;
+	adev->doorbell_index.mec_ring0 = AMDGPU_NAVI10_DOORBELL_MEC_RING0;
+	adev->doorbell_index.mec_ring1 = AMDGPU_NAVI10_DOORBELL_MEC_RING1;
+	adev->doorbell_index.mec_ring2 = AMDGPU_NAVI10_DOORBELL_MEC_RING2;
+	adev->doorbell_index.mec_ring3 = AMDGPU_NAVI10_DOORBELL_MEC_RING3;
+	adev->doorbell_index.mec_ring4 = AMDGPU_NAVI10_DOORBELL_MEC_RING4;
+	adev->doorbell_index.mec_ring5 = AMDGPU_NAVI10_DOORBELL_MEC_RING5;
+	adev->doorbell_index.mec_ring6 = AMDGPU_NAVI10_DOORBELL_MEC_RING6;
+	adev->doorbell_index.mec_ring7 = AMDGPU_NAVI10_DOORBELL_MEC_RING7;
+	adev->doorbell_index.userqueue_start = AMDGPU_NAVI10_DOORBELL_USERQUEUE_START;
+	adev->doorbell_index.userqueue_end = AMDGPU_NAVI10_DOORBELL_USERQUEUE_END;
+	adev->doorbell_index.gfx_ring0 = AMDGPU_NAVI10_DOORBELL_GFX_RING0;
+	adev->doorbell_index.gfx_ring1 = AMDGPU_NAVI10_DOORBELL_GFX_RING1;
+	adev->doorbell_index.gfx_userqueue_start =
+		AMDGPU_NAVI10_DOORBELL_GFX_USERQUEUE_START;
+	adev->doorbell_index.gfx_userqueue_end =
+		AMDGPU_NAVI10_DOORBELL_GFX_USERQUEUE_END;
+	adev->doorbell_index.mes_ring0 = AMDGPU_NAVI10_DOORBELL_MES_RING0;
+	adev->doorbell_index.mes_ring1 = AMDGPU_NAVI10_DOORBELL_MES_RING1;
+	adev->doorbell_index.sdma_engine[0] = AMDGPU_NAVI10_DOORBELL_sDMA_ENGINE0;
+	adev->doorbell_index.sdma_engine[1] = AMDGPU_NAVI10_DOORBELL_sDMA_ENGINE1;
+	adev->doorbell_index.sdma_engine[2] = AMDGPU_NAVI10_DOORBELL_sDMA_ENGINE2;
+	adev->doorbell_index.sdma_engine[3] = AMDGPU_NAVI10_DOORBELL_sDMA_ENGINE3;
+	adev->doorbell_index.ih = AMDGPU_NAVI10_DOORBELL_IH;
+	adev->doorbell_index.vcn.vcn_ring0_1 = AMDGPU_NAVI10_DOORBELL64_VCN0_1;
+	adev->doorbell_index.vcn.vcn_ring2_3 = AMDGPU_NAVI10_DOORBELL64_VCN2_3;
+	adev->doorbell_index.vcn.vcn_ring4_5 = AMDGPU_NAVI10_DOORBELL64_VCN4_5;
+	adev->doorbell_index.vcn.vcn_ring6_7 = AMDGPU_NAVI10_DOORBELL64_VCN6_7;
+	adev->doorbell_index.first_non_cp = AMDGPU_NAVI10_DOORBELL64_FIRST_NON_CP;
+	adev->doorbell_index.last_non_cp = AMDGPU_NAVI10_DOORBELL64_LAST_NON_CP;
+
+	adev->doorbell_index.max_assignment = AMDGPU_NAVI10_DOORBELL_MAX_ASSIGNMENT << 1;
+	adev->doorbell_index.sdma_doorbell_range = 20;
+}
+
+static void nv_pre_asic_init(struct amdgpu_device *adev)
+{
+}
+
+static int nv_update_umd_stable_pstate(struct amdgpu_device *adev,
+				       bool enter)
+{
+	if (enter)
+		amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
+	else
+		amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
+
+	if (adev->gfx.funcs->update_perfmon_mgcg)
+		adev->gfx.funcs->update_perfmon_mgcg(adev, !enter);
+
+	if (adev->nbio.funcs->enable_aspm &&
+	    amdgpu_device_should_use_aspm(adev))
+		adev->nbio.funcs->enable_aspm(adev, !enter);
+
+	return 0;
+}
+
+static const struct amdgpu_asic_funcs nv_asic_funcs = {
+	.read_disabled_bios = &nv_read_disabled_bios,
+	.read_bios_from_rom = &amdgpu_soc15_read_bios_from_rom,
+	.read_register = &nv_read_register,
+	.reset = &nv_asic_reset,
+	.reset_method = &nv_asic_reset_method,
+	.get_xclk = &nv_get_xclk,
+	.set_uvd_clocks = &nv_set_uvd_clocks,
+	.set_vce_clocks = &nv_set_vce_clocks,
+	.get_config_memsize = &nv_get_config_memsize,
+	.init_doorbell_index = &nv_init_doorbell_index,
+	.need_full_reset = &nv_need_full_reset,
+	.need_reset_on_init = &nv_need_reset_on_init,
+	.get_pcie_replay_count = &amdgpu_nbio_get_pcie_replay_count,
+	.supports_baco = &amdgpu_dpm_is_baco_supported,
+	.pre_asic_init = &nv_pre_asic_init,
+	.update_umd_stable_pstate = &nv_update_umd_stable_pstate,
+	.query_video_codecs = &nv_query_video_codecs,
+};
+
+static int nv_common_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	adev->nbio.funcs->set_reg_remap(adev);
+	adev->smc_rreg = NULL;
+	adev->smc_wreg = NULL;
+	adev->pcie_rreg = &amdgpu_device_indirect_rreg;
+	adev->pcie_wreg = &amdgpu_device_indirect_wreg;
+	adev->pcie_rreg64 = &amdgpu_device_indirect_rreg64;
+	adev->pcie_wreg64 = &amdgpu_device_indirect_wreg64;
+	adev->pciep_rreg = amdgpu_device_pcie_port_rreg;
+	adev->pciep_wreg = amdgpu_device_pcie_port_wreg;
+
+	/* TODO: will add them during VCN v2 implementation */
+	adev->uvd_ctx_rreg = NULL;
+	adev->uvd_ctx_wreg = NULL;
+
+	adev->didt_rreg = &nv_didt_rreg;
+	adev->didt_wreg = &nv_didt_wreg;
+
+	adev->asic_funcs = &nv_asic_funcs;
+
+	adev->rev_id = amdgpu_device_get_rev_id(adev);
+	adev->external_rev_id = 0xff;
+	/* TODO: split the GC and PG flags based on the relevant IP version for which
+	 * they are relevant.
+	 */
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 1, 10):
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_IH_CG |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_ATHUB_MGCG |
+			AMD_CG_SUPPORT_ATHUB_LS |
+			AMD_CG_SUPPORT_VCN_MGCG |
+			AMD_CG_SUPPORT_JPEG_MGCG |
+			AMD_CG_SUPPORT_BIF_MGCG |
+			AMD_CG_SUPPORT_BIF_LS;
+		adev->pg_flags = AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_VCN_DPG |
+			AMD_PG_SUPPORT_JPEG |
+			AMD_PG_SUPPORT_ATHUB;
+		adev->external_rev_id = adev->rev_id + 0x1;
+		break;
+	case IP_VERSION(10, 1, 1):
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_IH_CG |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_ATHUB_MGCG |
+			AMD_CG_SUPPORT_ATHUB_LS |
+			AMD_CG_SUPPORT_VCN_MGCG |
+			AMD_CG_SUPPORT_JPEG_MGCG |
+			AMD_CG_SUPPORT_BIF_MGCG |
+			AMD_CG_SUPPORT_BIF_LS;
+		adev->pg_flags = AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_JPEG |
+			AMD_PG_SUPPORT_VCN_DPG;
+		adev->external_rev_id = adev->rev_id + 20;
+		break;
+	case IP_VERSION(10, 1, 2):
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_GFX_RLC_LS |
+			AMD_CG_SUPPORT_IH_CG |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_ATHUB_MGCG |
+			AMD_CG_SUPPORT_ATHUB_LS |
+			AMD_CG_SUPPORT_VCN_MGCG |
+			AMD_CG_SUPPORT_JPEG_MGCG;
+		adev->pg_flags = AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_VCN_DPG |
+			AMD_PG_SUPPORT_JPEG |
+			AMD_PG_SUPPORT_ATHUB;
+		/* guest vm gets 0xffffffff when reading RCC_DEV0_EPF0_STRAP0,
+		 * as a consequence, the rev_id and external_rev_id are wrong.
+		 * workaround it by hardcoding rev_id to 0 (default value).
+		 */
+		if (amdgpu_sriov_vf(adev))
+			adev->rev_id = 0;
+		adev->external_rev_id = adev->rev_id + 0xa;
+		break;
+	case IP_VERSION(10, 3, 0):
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_VCN_MGCG |
+			AMD_CG_SUPPORT_JPEG_MGCG |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_IH_CG |
+			AMD_CG_SUPPORT_MC_LS;
+		adev->pg_flags = AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_VCN_DPG |
+			AMD_PG_SUPPORT_JPEG |
+			AMD_PG_SUPPORT_ATHUB |
+			AMD_PG_SUPPORT_MMHUB;
+		if (amdgpu_sriov_vf(adev)) {
+			/* hypervisor control CG and PG enablement */
+			adev->cg_flags = 0;
+			adev->pg_flags = 0;
+		}
+		adev->external_rev_id = adev->rev_id + 0x28;
+		break;
+	case IP_VERSION(10, 3, 2):
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_VCN_MGCG |
+			AMD_CG_SUPPORT_JPEG_MGCG |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_IH_CG;
+		adev->pg_flags = AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_VCN_DPG |
+			AMD_PG_SUPPORT_JPEG |
+			AMD_PG_SUPPORT_ATHUB |
+			AMD_PG_SUPPORT_MMHUB;
+		adev->external_rev_id = adev->rev_id + 0x32;
+		break;
+	case IP_VERSION(10, 3, 1):
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_GFX_RLC_LS |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGLS |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_GFX_FGCG |
+			AMD_CG_SUPPORT_VCN_MGCG |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_JPEG_MGCG;
+		adev->pg_flags = AMD_PG_SUPPORT_GFX_PG |
+			AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_VCN_DPG |
+			AMD_PG_SUPPORT_JPEG;
+		if (adev->apu_flags & AMD_APU_IS_VANGOGH)
+			adev->external_rev_id = adev->rev_id + 0x01;
+		break;
+	case IP_VERSION(10, 3, 4):
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_VCN_MGCG |
+			AMD_CG_SUPPORT_JPEG_MGCG |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_IH_CG;
+		adev->pg_flags = AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_VCN_DPG |
+			AMD_PG_SUPPORT_JPEG |
+			AMD_PG_SUPPORT_ATHUB |
+			AMD_PG_SUPPORT_MMHUB;
+		adev->external_rev_id = adev->rev_id + 0x3c;
+		break;
+	case IP_VERSION(10, 3, 5):
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_IH_CG |
+			AMD_CG_SUPPORT_VCN_MGCG;
+		adev->pg_flags = AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_VCN_DPG |
+			AMD_PG_SUPPORT_ATHUB |
+			AMD_PG_SUPPORT_MMHUB;
+		adev->external_rev_id = adev->rev_id + 0x46;
+		break;
+	case IP_VERSION(10, 3, 3):
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGLS |
+			AMD_CG_SUPPORT_GFX_RLC_LS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_GFX_FGCG |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_ATHUB_MGCG |
+			AMD_CG_SUPPORT_ATHUB_LS |
+			AMD_CG_SUPPORT_IH_CG |
+			AMD_CG_SUPPORT_VCN_MGCG |
+			AMD_CG_SUPPORT_JPEG_MGCG |
+			AMD_CG_SUPPORT_SDMA_MGCG;
+		adev->pg_flags = AMD_PG_SUPPORT_GFX_PG |
+			AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_VCN_DPG |
+			AMD_PG_SUPPORT_JPEG;
+		if (adev->pdev->device == 0x1681)
+			adev->external_rev_id = 0x20;
+		else
+			adev->external_rev_id = adev->rev_id + 0x01;
+		break;
+	case IP_VERSION(10, 1, 3):
+	case IP_VERSION(10, 1, 4):
+		adev->cg_flags = 0;
+		adev->pg_flags = 0;
+		adev->external_rev_id = adev->rev_id + 0x82;
+		break;
+	case IP_VERSION(10, 3, 6):
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGLS |
+			AMD_CG_SUPPORT_GFX_RLC_LS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_GFX_FGCG |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_ATHUB_MGCG |
+			AMD_CG_SUPPORT_ATHUB_LS |
+			AMD_CG_SUPPORT_IH_CG |
+			AMD_CG_SUPPORT_VCN_MGCG |
+			AMD_CG_SUPPORT_JPEG_MGCG;
+		adev->pg_flags = AMD_PG_SUPPORT_GFX_PG |
+			AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_VCN_DPG |
+			AMD_PG_SUPPORT_JPEG;
+		adev->external_rev_id = adev->rev_id + 0x01;
+		break;
+	case IP_VERSION(10, 3, 7):
+		adev->cg_flags =  AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGLS |
+			AMD_CG_SUPPORT_GFX_RLC_LS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_GFX_FGCG |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_ATHUB_MGCG |
+			AMD_CG_SUPPORT_ATHUB_LS |
+			AMD_CG_SUPPORT_IH_CG |
+			AMD_CG_SUPPORT_VCN_MGCG |
+			AMD_CG_SUPPORT_JPEG_MGCG |
+			AMD_CG_SUPPORT_SDMA_MGCG;
+		adev->pg_flags = AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_VCN_DPG |
+			AMD_PG_SUPPORT_JPEG |
+			AMD_PG_SUPPORT_GFX_PG;
+		adev->external_rev_id = adev->rev_id + 0x01;
+		break;
+	default:
+		/* FIXME: not supported yet */
+		return -EINVAL;
+	}
+
+	if (adev->harvest_ip_mask & AMD_HARVEST_IP_VCN_MASK)
+		adev->pg_flags &= ~(AMD_PG_SUPPORT_VCN |
+				    AMD_PG_SUPPORT_VCN_DPG |
+				    AMD_PG_SUPPORT_JPEG);
+
+	if (amdgpu_sriov_vf(adev)) {
+		amdgpu_virt_init_setting(adev);
+		xgpu_nv_mailbox_set_irq_funcs(adev);
+	}
+
+	return 0;
+}
+
+static int nv_common_late_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_sriov_vf(adev)) {
+		xgpu_nv_mailbox_get_irq(adev);
+		if (adev->vcn.harvest_config & AMDGPU_VCN_HARVEST_VCN0) {
+			amdgpu_virt_update_sriov_video_codec(adev,
+							     sriov_sc_video_codecs_encode_array,
+							     ARRAY_SIZE(sriov_sc_video_codecs_encode_array),
+							     sriov_sc_video_codecs_decode_array_vcn1,
+							     ARRAY_SIZE(sriov_sc_video_codecs_decode_array_vcn1));
+		} else {
+			amdgpu_virt_update_sriov_video_codec(adev,
+							     sriov_sc_video_codecs_encode_array,
+							     ARRAY_SIZE(sriov_sc_video_codecs_encode_array),
+							     sriov_sc_video_codecs_decode_array_vcn0,
+							     ARRAY_SIZE(sriov_sc_video_codecs_decode_array_vcn0));
+		}
+	}
+
+	/* Enable selfring doorbell aperture late because doorbell BAR
+	 * aperture will change if resize BAR successfully in gmc sw_init.
+	 */
+	adev->nbio.funcs->enable_doorbell_selfring_aperture(adev, true);
+
+	return 0;
+}
+
+static int nv_common_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_sriov_vf(adev))
+		xgpu_nv_mailbox_add_irq_id(adev);
+
+	return 0;
+}
+
+static int nv_common_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (adev->nbio.funcs->apply_lc_spc_mode_wa)
+		adev->nbio.funcs->apply_lc_spc_mode_wa(adev);
+
+	if (adev->nbio.funcs->apply_l1_link_width_reconfig_wa)
+		adev->nbio.funcs->apply_l1_link_width_reconfig_wa(adev);
+
+	/* enable aspm */
+	nv_program_aspm(adev);
+	/* setup nbio registers */
+	adev->nbio.funcs->init_registers(adev);
+	/* remap HDP registers to a hole in mmio space,
+	 * for the purpose of expose those registers
+	 * to process space
+	 */
+	if (adev->nbio.funcs->remap_hdp_registers && !amdgpu_sriov_vf(adev))
+		adev->nbio.funcs->remap_hdp_registers(adev);
+	/* enable the doorbell aperture */
+	adev->nbio.funcs->enable_doorbell_aperture(adev, true);
+
+	return 0;
+}
+
+static int nv_common_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	/* Disable the doorbell aperture and selfring doorbell aperture
+	 * separately in hw_fini because nv_enable_doorbell_aperture
+	 * has been removed and there is no need to delay disabling
+	 * selfring doorbell.
+	 */
+	adev->nbio.funcs->enable_doorbell_aperture(adev, false);
+	adev->nbio.funcs->enable_doorbell_selfring_aperture(adev, false);
+
+	return 0;
+}
+
+static int nv_common_suspend(struct amdgpu_ip_block *ip_block)
+{
+	return nv_common_hw_fini(ip_block);
+}
+
+static int nv_common_resume(struct amdgpu_ip_block *ip_block)
+{
+	return nv_common_hw_init(ip_block);
+}
+
+static bool nv_common_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	return true;
+}
+
+static int nv_common_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					   enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) {
+	case IP_VERSION(2, 3, 0):
+	case IP_VERSION(2, 3, 1):
+	case IP_VERSION(2, 3, 2):
+	case IP_VERSION(3, 3, 0):
+	case IP_VERSION(3, 3, 1):
+	case IP_VERSION(3, 3, 2):
+	case IP_VERSION(3, 3, 3):
+		adev->nbio.funcs->update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		adev->nbio.funcs->update_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+		adev->hdp.funcs->update_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		adev->smuio.funcs->update_rom_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int nv_common_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					   enum amd_powergating_state state)
+{
+	/* TODO */
+	return 0;
+}
+
+static void nv_common_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_sriov_vf(adev))
+		*flags = 0;
+
+	adev->nbio.funcs->get_clockgating_state(adev, flags);
+
+	adev->hdp.funcs->get_clock_gating_state(adev, flags);
+
+	adev->smuio.funcs->get_clock_gating_state(adev, flags);
+}
+
+static const struct amd_ip_funcs nv_common_ip_funcs = {
+	.name = "nv_common",
+	.early_init = nv_common_early_init,
+	.late_init = nv_common_late_init,
+	.sw_init = nv_common_sw_init,
+	.hw_init = nv_common_hw_init,
+	.hw_fini = nv_common_hw_fini,
+	.suspend = nv_common_suspend,
+	.resume = nv_common_resume,
+	.is_idle = nv_common_is_idle,
+	.set_clockgating_state = nv_common_set_clockgating_state,
+	.set_powergating_state = nv_common_set_powergating_state,
+	.get_clockgating_state = nv_common_get_clockgating_state,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/nv.h b/drivers/gpu/drm/amd/amdgpu/nv.h
new file mode 100644
index 000000000000..8f4817404f10
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/nv.h
@@ -0,0 +1,36 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __NV_H__
+#define __NV_H__
+
+#include "nbio_v2_3.h"
+
+extern const struct amdgpu_ip_block_version nv_common_ip_block;
+
+void nv_grbm_select(struct amdgpu_device *adev,
+		    u32 me, u32 pipe, u32 queue, u32 vmid);
+void nv_set_virt_ops(struct amdgpu_device *adev);
+int cyan_skillfish_reg_base_init(struct amdgpu_device *adev);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/nvd.h b/drivers/gpu/drm/amd/amdgpu/nvd.h
new file mode 100644
index 000000000000..56f1bfac0b20
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/nvd.h
@@ -0,0 +1,678 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef NVD_H
+#define NVD_H
+
+/**
+ * Navi's PM4 definitions
+ */
+#define	PACKET_TYPE0	0
+#define	PACKET_TYPE1	1
+#define	PACKET_TYPE2	2
+#define	PACKET_TYPE3	3
+
+#define CP_PACKET_GET_TYPE(h) (((h) >> 30) & 3)
+#define CP_PACKET_GET_COUNT(h) (((h) >> 16) & 0x3FFF)
+#define CP_PACKET0_GET_REG(h) ((h) & 0xFFFF)
+#define CP_PACKET3_GET_OPCODE(h) (((h) >> 8) & 0xFF)
+#define PACKET0(reg, n)	((PACKET_TYPE0 << 30) |				\
+			 ((reg) & 0xFFFF) |			\
+			 ((n) & 0x3FFF) << 16)
+#define CP_PACKET2			0x80000000
+#define		PACKET2_PAD_SHIFT		0
+#define		PACKET2_PAD_MASK		(0x3fffffff << 0)
+
+#define PACKET2(v)	(CP_PACKET2 | REG_SET(PACKET2_PAD, (v)))
+
+#define PACKET3(op, n)	((PACKET_TYPE3 << 30) |				\
+			 (((op) & 0xFF) << 8) |				\
+			 ((n) & 0x3FFF) << 16)
+
+#define PACKET3_COMPUTE(op, n) (PACKET3(op, n) | 1 << 1)
+
+/* Packet 3 types */
+#define	PACKET3_NOP					0x10
+#define	PACKET3_SET_BASE				0x11
+#define		PACKET3_BASE_INDEX(x)                  ((x) << 0)
+#define			CE_PARTITION_BASE		3
+#define	PACKET3_CLEAR_STATE				0x12
+#define	PACKET3_INDEX_BUFFER_SIZE			0x13
+#define	PACKET3_DISPATCH_DIRECT				0x15
+#define	PACKET3_DISPATCH_INDIRECT			0x16
+#define	PACKET3_INDIRECT_BUFFER_END			0x17
+#define	PACKET3_INDIRECT_BUFFER_CNST_END		0x19
+#define	PACKET3_ATOMIC_GDS				0x1D
+#define	PACKET3_ATOMIC_MEM				0x1E
+#define 	PACKET3_ATOMIC_MEM__ATOMIC(x) ((((unsigned)(x)) & 0x7F) << 0)
+#define 	PACKET3_ATOMIC_MEM__COMMAND(x) ((((unsigned)(x)) & 0xF) << 8)
+#define 	PACKET3_ATOMIC_MEM__CACHE_POLICY(x) ((((unsigned)(x)) & 0x3) << 25)
+#define 	PACKET3_ATOMIC_MEM__ADDR_LO(x) (((unsigned)(x)))
+#define 	PACKET3_ATOMIC_MEM__ADDR_HI(x) (((unsigned)(x)))
+#define 	PACKET3_ATOMIC_MEM__SRC_DATA_LO(x) (((unsigned)(x)))
+#define 	PACKET3_ATOMIC_MEM__SRC_DATA_HI(x) (((unsigned)(x)))
+#define 	PACKET3_ATOMIC_MEM__CMP_DATA_LO(x) (((unsigned)(x)))
+#define 	PACKET3_ATOMIC_MEM__CMP_DATA_HI(x) (((unsigned)(x)))
+#define 	PACKET3_ATOMIC_MEM__LOOP_INTERVAL(x) ((((unsigned)(x)) & 0x1FFF) << 0)
+#define 	PACKET3_ATOMIC_MEM__COMMAND__SINGLE_PASS_ATOMIC 0
+#define 	PACKET3_ATOMIC_MEM__COMMAND__LOOP_UNTIL_COMPARE_SATISFIED 1
+#define 	PACKET3_ATOMIC_MEM__COMMAND__WAIT_FOR_WRITE_CONFIRMATION 2
+#define 	PACKET3_ATOMIC_MEM__COMMAND__SEND_AND_CONTINUE 3
+#define 	PACKET3_ATOMIC_MEM__CACHE_POLICY__LRU 0
+#define 	PACKET3_ATOMIC_MEM__CACHE_POLICY__STREAM 1
+#define 	PACKET3_ATOMIC_MEM__CACHE_POLICY__NOA 2
+#define 	PACKET3_ATOMIC_MEM__CACHE_POLICY__BYPASS 3
+#define	PACKET3_OCCLUSION_QUERY				0x1F
+#define	PACKET3_SET_PREDICATION				0x20
+#define	PACKET3_REG_RMW					0x21
+#define	PACKET3_COND_EXEC				0x22
+#define	PACKET3_PRED_EXEC				0x23
+#define	PACKET3_DRAW_INDIRECT				0x24
+#define	PACKET3_DRAW_INDEX_INDIRECT			0x25
+#define	PACKET3_INDEX_BASE				0x26
+#define	PACKET3_DRAW_INDEX_2				0x27
+#define	PACKET3_CONTEXT_CONTROL				0x28
+#define	PACKET3_INDEX_TYPE				0x2A
+#define	PACKET3_DRAW_INDIRECT_MULTI			0x2C
+#define	PACKET3_DRAW_INDEX_AUTO				0x2D
+#define	PACKET3_NUM_INSTANCES				0x2F
+#define	PACKET3_DRAW_INDEX_MULTI_AUTO			0x30
+#define	PACKET3_INDIRECT_BUFFER_PRIV			0x32
+#define	PACKET3_INDIRECT_BUFFER_CNST			0x33
+#define	PACKET3_COND_INDIRECT_BUFFER_CNST		0x33
+#define	PACKET3_STRMOUT_BUFFER_UPDATE			0x34
+#define	PACKET3_DRAW_INDEX_OFFSET_2			0x35
+#define	PACKET3_DRAW_PREAMBLE				0x36
+#define	PACKET3_WRITE_DATA				0x37
+#define		WRITE_DATA_DST_SEL(x)                   ((x) << 8)
+		/* 0 - register
+		 * 1 - memory (sync - via GRBM)
+		 * 2 - gl2
+		 * 3 - gds
+		 * 4 - reserved
+		 * 5 - memory (async - direct)
+		 */
+#define		WR_ONE_ADDR                             (1 << 16)
+#define		WR_CONFIRM                              (1 << 20)
+#define		WRITE_DATA_CACHE_POLICY(x)              ((x) << 25)
+		/* 0 - LRU
+		 * 1 - Stream
+		 */
+#define		WRITE_DATA_ENGINE_SEL(x)                ((x) << 30)
+		/* 0 - me
+		 * 1 - pfp
+		 * 2 - ce
+		 */
+#define 	PACKET3_WRITE_DATA__DST_SEL(x) ((((unsigned)(x)) & 0xF) << 8)
+#define 	PACKET3_WRITE_DATA__ADDR_INCR(x) ((((unsigned)(x)) & 0x1) << 16)
+#define 	PACKET3_WRITE_DATA__WR_CONFIRM(x) ((((unsigned)(x)) & 0x1) << 20)
+#define 	PACKET3_WRITE_DATA__CACHE_POLICY(x) ((((unsigned)(x)) & 0x3) << 25)
+#define 	PACKET3_WRITE_DATA__DST_MMREG_ADDR(x) ((((unsigned)(x)) & 0x3FFFF) << 0)
+#define 	PACKET3_WRITE_DATA__DST_GDS_ADDR(x) ((((unsigned)(x)) & 0xFFFF) << 0)
+#define 	PACKET3_WRITE_DATA__DST_MEM_ADDR_LO(x) ((((unsigned)(x)) & 0x3FFFFFFF) << 2)
+#define 	PACKET3_WRITE_DATA__DST_MEM_ADDR_HI(x) ((unsigned)(x))
+#define 	PACKET3_WRITE_DATA__MODE(x) ((((unsigned)(x)) & 0x1) << 21)
+#define 	PACKET3_WRITE_DATA__AID_ID(x) ((((unsigned)(x)) & 0x3) << 22)
+#define 	PACKET3_WRITE_DATA__TEMPORAL(x) ((((unsigned)(x)) & 0x3) << 24)
+#define 	PACKET3_WRITE_DATA__DST_MMREG_ADDR_LO(x) ((unsigned)(x))
+#define 	PACKET3_WRITE_DATA__DST_MMREG_ADDR_HI(x) ((((unsigned)(x)) & 0xFF) << 0)
+#define 	PACKET3_WRITE_DATA__DST_SEL__MEM_MAPPED_REGISTER 0
+#define 	PACKET3_WRITE_DATA__DST_SEL__TC_L2 2
+#define 	PACKET3_WRITE_DATA__DST_SEL__GDS 3
+#define 	PACKET3_WRITE_DATA__DST_SEL__MEMORY 5
+#define 	PACKET3_WRITE_DATA__DST_SEL__MEMORY_MAPPED_ADC_PERSISTENT_STATE 6
+#define 	PACKET3_WRITE_DATA__ADDR_INCR__INCREMENT_ADDRESS 0
+#define 	PACKET3_WRITE_DATA__ADDR_INCR__DO_NOT_INCREMENT_ADDRESS 1
+#define 	PACKET3_WRITE_DATA__WR_CONFIRM__DO_NOT_WAIT_FOR_WRITE_CONFIRMATION 0
+#define 	PACKET3_WRITE_DATA__WR_CONFIRM__WAIT_FOR_WRITE_CONFIRMATION 1
+#define 	PACKET3_WRITE_DATA__MODE__PF_VF_DISABLED 0
+#define 	PACKET3_WRITE_DATA__MODE__PF_VF_ENABLED 1
+#define 	PACKET3_WRITE_DATA__TEMPORAL__RT 0
+#define 	PACKET3_WRITE_DATA__TEMPORAL__NT 1
+#define 	PACKET3_WRITE_DATA__TEMPORAL__HT 2
+#define 	PACKET3_WRITE_DATA__TEMPORAL__LU 3
+#define 	PACKET3_WRITE_DATA__CACHE_POLICY__LRU 0
+#define 	PACKET3_WRITE_DATA__CACHE_POLICY__STREAM 1
+#define 	PACKET3_WRITE_DATA__CACHE_POLICY__NOA 2
+#define 	PACKET3_WRITE_DATA__CACHE_POLICY__BYPASS 3
+#define	PACKET3_DRAW_INDEX_INDIRECT_MULTI		0x38
+#define	PACKET3_MEM_SEMAPHORE				0x39
+#              define PACKET3_SEM_USE_MAILBOX       (0x1 << 16)
+#              define PACKET3_SEM_SEL_SIGNAL_TYPE   (0x1 << 20) /* 0 = increment, 1 = write 1 */
+#              define PACKET3_SEM_SEL_SIGNAL	    (0x6 << 29)
+#              define PACKET3_SEM_SEL_WAIT	    (0x7 << 29)
+#define	PACKET3_DRAW_INDEX_MULTI_INST			0x3A
+#define	PACKET3_COPY_DW					0x3B
+#define	PACKET3_WAIT_REG_MEM				0x3C
+#define		WAIT_REG_MEM_FUNCTION(x)                ((x) << 0)
+		/* 0 - always
+		 * 1 - <
+		 * 2 - <=
+		 * 3 - ==
+		 * 4 - !=
+		 * 5 - >=
+		 * 6 - >
+		 */
+#define		WAIT_REG_MEM_MEM_SPACE(x)               ((x) << 4)
+		/* 0 - reg
+		 * 1 - mem
+		 */
+#define		WAIT_REG_MEM_OPERATION(x)               ((x) << 6)
+		/* 0 - wait_reg_mem
+		 * 1 - wr_wait_wr_reg
+		 */
+#define		WAIT_REG_MEM_ENGINE(x)                  ((x) << 8)
+		/* 0 - me
+		 * 1 - pfp
+		 */
+#define		PACKET3_WAIT_REG_MEM__FUNCTION(x) ((((unsigned)(x)) & 0x7) << 0)
+#define		PACKET3_WAIT_REG_MEM__MEM_SPACE(x) ((((unsigned)(x)) & 0x3) << 4)
+#define		PACKET3_WAIT_REG_MEM__OPERATION(x) ((((unsigned)(x)) & 0x3) << 6)
+#define		PACKET3_WAIT_REG_MEM__MES_INTR_PIPE(x) ((((unsigned)(x)) & 0x3) << 22)
+#define		PACKET3_WAIT_REG_MEM__MES_ACTION(x) ((((unsigned)(x)) & 0x1) << 24)
+#define		PACKET3_WAIT_REG_MEM__CACHE_POLICY(x) ((((unsigned)(x)) & 0x3) << 25)
+#define		PACKET3_WAIT_REG_MEM__TEMPORAL(x) ((((unsigned)(x)) & 0x3) << 25)
+#define		PACKET3_WAIT_REG_MEM__MEM_POLL_ADDR_LO(x) ((((unsigned)(x)) & 0x3FFFFFFF) << 2)
+#define		PACKET3_WAIT_REG_MEM__REG_POLL_ADDR(x) ((((unsigned)(x)) & 0X3FFFF) << 0)
+#define		PACKET3_WAIT_REG_MEM__REG_WRITE_ADDR1(x) ((((unsigned)(x)) & 0X3FFFF) << 0)
+#define		PACKET3_WAIT_REG_MEM__MEM_POLL_ADDR_HI(x) ((unsigned)(x))
+#define		PACKET3_WAIT_REG_MEM__REG_WRITE_ADDR2(x) ((((unsigned)(x)) & 0x3FFFF) << 0)
+#define		PACKET3_WAIT_REG_MEM__REFERENCE(x) ((unsigned)(x))
+#define		PACKET3_WAIT_REG_MEM__MASK(x) ((unsigned)(x))
+#define		PACKET3_WAIT_REG_MEM__POLL_INTERVAL(x) ((((unsigned)(x)) & 0xFFFF) << 0)
+#define		PACKET3_WAIT_REG_MEM__OPTIMIZE_ACE_OFFLOAD_MODE(x) ((((unsigned)(x)) & 0x1) << 31)
+#define 	PACKET3_WAIT_REG_MEM__FUNCTION__ALWAYS_PASS 0
+#define 	PACKET3_WAIT_REG_MEM__FUNCTION__LESS_THAN_REF_VALUE 1
+#define 	PACKET3_WAIT_REG_MEM__FUNCTION__LESS_THAN_EQUAL_TO_THE_REF_VALUE 2
+#define 	PACKET3_WAIT_REG_MEM__FUNCTION__EQUAL_TO_THE_REFERENCE_VALUE 3
+#define 	PACKET3_WAIT_REG_MEM__FUNCTION__NOT_EQUAL_REFERENCE_VALUE 4
+#define 	PACKET3_WAIT_REG_MEM__FUNCTION__GREATER_THAN_OR_EQUAL_REFERENCE_VALUE 5
+#define 	PACKET3_WAIT_REG_MEM__FUNCTION__GREATER_THAN_REFERENCE_VALUE 6
+#define 	PACKET3_WAIT_REG_MEM__MEM_SPACE__REGISTER_SPACE 0
+#define 	PACKET3_WAIT_REG_MEM__MEM_SPACE__MEMORY_SPACE 1
+#define 	PACKET3_WAIT_REG_MEM__OPERATION__WAIT_REG_MEM 0
+#define 	PACKET3_WAIT_REG_MEM__OPERATION__WR_WAIT_WR_REG 1
+#define 	PACKET3_WAIT_REG_MEM__OPERATION__WAIT_MEM_PREEMPTABLE 3
+#define 	PACKET3_WAIT_REG_MEM__CACHE_POLICY__LRU 0
+#define 	PACKET3_WAIT_REG_MEM__CACHE_POLICY__STREAM 1
+#define 	PACKET3_WAIT_REG_MEM__CACHE_POLICY__NOA 2
+#define 	PACKET3_WAIT_REG_MEM__CACHE_POLICY__BYPASS 3
+#define 	PACKET3_WAIT_REG_MEM__TEMPORAL__RT 0
+#define 	PACKET3_WAIT_REG_MEM__TEMPORAL__NT 1
+#define 	PACKET3_WAIT_REG_MEM__TEMPORAL__HT 2
+#define 	PACKET3_WAIT_REG_MEM__TEMPORAL__LU 3
+#define	PACKET3_INDIRECT_BUFFER				0x3F
+#define		INDIRECT_BUFFER_VALID                   (1 << 23)
+#define		INDIRECT_BUFFER_CACHE_POLICY(x)         ((x) << 28)
+		/* 0 - LRU
+		 * 1 - Stream
+		 * 2 - Bypass
+		 */
+#define		INDIRECT_BUFFER_PRE_ENB(x)		((x) << 21)
+#define		INDIRECT_BUFFER_PRE_RESUME(x)           ((x) << 30)
+#define 	PACKET3_INDIRECT_BUFFER__IB_BASE_LO(x) ((((unsigned)(x)) & 0x3FFFFFFF) << 2)
+#define 	PACKET3_INDIRECT_BUFFER__IB_BASE_HI(x) ((unsigned)(x))
+#define 	PACKET3_INDIRECT_BUFFER__IB_SIZE(x) ((((unsigned)(x)) & 0xFFFFF) << 0)
+#define 	PACKET3_INDIRECT_BUFFER__CHAIN(x) ((((unsigned)(x)) & 0x1) << 20)
+#define 	PACKET3_INDIRECT_BUFFER__OFFLOAD_POLLING(x) ((((unsigned)(x)) & 0x1) << 21)
+#define 	PACKET3_INDIRECT_BUFFER__VALID(x) ((((unsigned)(x)) & 0x1) << 23)
+#define 	PACKET3_INDIRECT_BUFFER__VMID(x) ((((unsigned)(x)) & 0xF) << 24)
+#define 	PACKET3_INDIRECT_BUFFER__CACHE_POLICY(x) ((((unsigned)(x)) & 0x3) << 28)
+#define 	PACKET3_INDIRECT_BUFFER__TEMPORAL(x) ((((unsigned)(x)) & 0x3) << 28)
+#define 	PACKET3_INDIRECT_BUFFER__PRIV(x) ((((unsigned)(x)) & 0x1) << 31)
+#define 	PACKET3_INDIRECT_BUFFER__TEMPORAL__RT 0
+#define 	PACKET3_INDIRECT_BUFFER__TEMPORAL__NT 1
+#define 	PACKET3_INDIRECT_BUFFER__TEMPORAL__HT 2
+#define 	PACKET3_INDIRECT_BUFFER__TEMPORAL__LU 3
+#define 	PACKET3_INDIRECT_BUFFER__CACHE_POLICY__LRU 0
+#define 	PACKET3_INDIRECT_BUFFER__CACHE_POLICY__STREAM 1
+#define 	PACKET3_INDIRECT_BUFFER__CACHE_POLICY__NOA 2
+#define 	PACKET3_INDIRECT_BUFFER__CACHE_POLICY__BYPASS 3
+#define	PACKET3_COND_INDIRECT_BUFFER			0x3F
+#define	PACKET3_COPY_DATA				0x40
+#define 	PACKET3_COPY_DATA__SRC_SEL(x) ((((unsigned)(x)) & 0xF) << 0)
+#define 	PACKET3_COPY_DATA__DST_SEL(x) ((((unsigned)(x)) & 0xF) << 8)
+#define 	PACKET3_COPY_DATA__SRC_CACHE_POLICY(x) ((((unsigned)(x)) & 0x3) << 13)
+#define 	PACKET3_COPY_DATA__SRC_TEMPORAL(x) ((((unsigned)(x)) & 0x3) << 13)
+#define 	PACKET3_COPY_DATA__COUNT_SEL(x) ((((unsigned)(x)) & 0x1) << 16)
+#define 	PACKET3_COPY_DATA__WR_CONFIRM(x) ((((unsigned)(x)) & 0x1) << 20)
+#define 	PACKET3_COPY_DATA__DST_CACHE_POLICY(x) ((((unsigned)(x)) & 0x3) << 25)
+#define 	PACKET3_COPY_DATA__PQ_EXE_STATUS(x) ((((unsigned)(x)) & 0x1) << 29)
+#define 	PACKET3_COPY_DATA__SRC_REG_OFFSET(x) ((((unsigned)(x)) & 0x3FFFF) << 0)
+#define 	PACKET3_COPY_DATA__SRC_32B_ADDR_LO(x) ((((unsigned)(x)) & 0x3FFFFFFF) << 2)
+#define 	PACKET3_COPY_DATA__SRC_64B_ADDR_LO(x) ((((unsigned)(x)) & 0x1FFFFFFF) << 3)
+#define 	PACKET3_COPY_DATA__SRC_GDS_ADDR_LO(x) ((((unsigned)(x)) & 0xFFFF) << 0)
+#define 	PACKET3_COPY_DATA__IMM_DATA(x) ((unsigned)(x))
+#define 	PACKET3_COPY_DATA__SRC_MEMTC_ADDR_HI(x) ((unsigned)(x))
+#define 	PACKET3_COPY_DATA__SRC_IMM_DATA(x) ((unsigned)(x))
+#define 	PACKET3_COPY_DATA__DST_REG_OFFSET(x) ((((unsigned)(x)) & 0x3FFFF) << 0)
+#define 	PACKET3_COPY_DATA__DST_32B_ADDR_LO(x) ((((unsigned)(x)) & 0x3FFFFFFF) << 2)
+#define 	PACKET3_COPY_DATA__DST_64B_ADDR_LO(x) ((((unsigned)(x)) & 0x1FFFFFFF) << 3)
+#define 	PACKET3_COPY_DATA__DST_GDS_ADDR_LO(x) ((((unsigned)(x)) & 0xFFFF) << 0)
+#define 	PACKET3_COPY_DATA__DST_ADDR_HI(x) ((unsigned)(x))
+#define 	PACKET3_COPY_DATA__MODE(x) ((((unsigned)(x)) & 0x1) << 21)
+#define 	PACKET3_COPY_DATA__AID_ID(x) ((((unsigned)(x)) & 0x3) << 23)
+#define 	PACKET3_COPY_DATA__DST_TEMPORAL(x) ((((unsigned)(x)) & 0x3) << 25)
+#define 	PACKET3_COPY_DATA__SRC_REG_OFFSET_LO(x) ((unsigned)(x))
+#define 	PACKET3_COPY_DATA__SRC_REG_OFFSET_HI(x) ((((unsigned)(x)) & 0xFF) << 0)
+#define 	PACKET3_COPY_DATA__DST_REG_OFFSET_LO(x) ((unsigned)(x))
+#define 	PACKET3_COPY_DATA__DST_REG_OFFSET_HI(x) ((((unsigned)(x)) & 0xFF) << 0)
+#define 	PACKET3_COPY_DATA__SRC_SEL__MEM_MAPPED_REGISTER 0
+#define 	PACKET3_COPY_DATA__SRC_SEL__TC_L2_OBSOLETE 1
+#define 	PACKET3_COPY_DATA__SRC_SEL__TC_L2 2
+#define 	PACKET3_COPY_DATA__SRC_SEL__GDS 3
+#define 	PACKET3_COPY_DATA__SRC_SEL__PERFCOUNTERS 4
+#define 	PACKET3_COPY_DATA__SRC_SEL__IMMEDIATE_DATA 5
+#define 	PACKET3_COPY_DATA__SRC_SEL__ATOMIC_RETURN_DATA 6
+#define 	PACKET3_COPY_DATA__SRC_SEL__GDS_ATOMIC_RETURN_DATA0 7
+#define 	PACKET3_COPY_DATA__SRC_SEL__GDS_ATOMIC_RETURN_DATA1 8
+#define 	PACKET3_COPY_DATA__SRC_SEL__GPU_CLOCK_COUNT 9
+#define 	PACKET3_COPY_DATA__SRC_SEL__SYSTEM_CLOCK_COUNT 10
+#define 	PACKET3_COPY_DATA__DST_SEL__MEM_MAPPED_REGISTER 0
+#define 	PACKET3_COPY_DATA__DST_SEL__TC_L2 2
+#define 	PACKET3_COPY_DATA__DST_SEL__GDS 3
+#define 	PACKET3_COPY_DATA__DST_SEL__PERFCOUNTERS 4
+#define 	PACKET3_COPY_DATA__DST_SEL__TC_L2_OBSOLETE 5
+#define 	PACKET3_COPY_DATA__DST_SEL__MEM_MAPPED_REG_DC 6
+#define 	PACKET3_COPY_DATA__SRC_TEMPORAL__RT 0
+#define 	PACKET3_COPY_DATA__SRC_TEMPORAL__NT 1
+#define 	PACKET3_COPY_DATA__SRC_TEMPORAL__HT 2
+#define 	PACKET3_COPY_DATA__SRC_TEMPORAL__LU 3
+#define 	PACKET3_COPY_DATA__SRC_CACHE_POLICY__LRU 0
+#define 	PACKET3_COPY_DATA__SRC_CACHE_POLICY__STREAM 1
+#define 	PACKET3_COPY_DATA__SRC_CACHE_POLICY__NOA 2
+#define 	PACKET3_COPY_DATA__SRC_CACHE_POLICY__BYPASS 3
+#define 	PACKET3_COPY_DATA__COUNT_SEL__32_BITS_OF_DATA 0
+#define 	PACKET3_COPY_DATA__COUNT_SEL__64_BITS_OF_DATA 1
+#define 	PACKET3_COPY_DATA__WR_CONFIRM__DO_NOT_WAIT_FOR_CONFIRMATION 0
+#define 	PACKET3_COPY_DATA__WR_CONFIRM__WAIT_FOR_CONFIRMATION 1
+#define 	PACKET3_COPY_DATA__MODE__PF_VF_DISABLED 0
+#define 	PACKET3_COPY_DATA__MODE__PF_VF_ENABLED 1
+#define 	PACKET3_COPY_DATA__DST_TEMPORAL__RT 0
+#define 	PACKET3_COPY_DATA__DST_TEMPORAL__NT 1
+#define 	PACKET3_COPY_DATA__DST_TEMPORAL__HT 2
+#define 	PACKET3_COPY_DATA__DST_TEMPORAL__LU 3
+#define 	PACKET3_COPY_DATA__DST_CACHE_POLICY__LRU 0
+#define 	PACKET3_COPY_DATA__DST_CACHE_POLICY__STREAM 1
+#define 	PACKET3_COPY_DATA__DST_CACHE_POLICY__NOA 2
+#define 	PACKET3_COPY_DATA__DST_CACHE_POLICY__BYPASS 3
+#define 	PACKET3_COPY_DATA__PQ_EXE_STATUS__DEFAULT 0
+#define 	PACKET3_COPY_DATA__PQ_EXE_STATUS__PHASE_UPDATE 1
+#define	PACKET3_CP_DMA					0x41
+#define	PACKET3_PFP_SYNC_ME				0x42
+#define	PACKET3_SURFACE_SYNC				0x43
+#define	PACKET3_ME_INITIALIZE				0x44
+#define	PACKET3_COND_WRITE				0x45
+#define	PACKET3_EVENT_WRITE				0x46
+#define		EVENT_TYPE(x)                           ((x) << 0)
+#define		EVENT_INDEX(x)                          ((x) << 8)
+		/* 0 - any non-TS event
+		 * 1 - ZPASS_DONE, PIXEL_PIPE_STAT_*
+		 * 2 - SAMPLE_PIPELINESTAT
+		 * 3 - SAMPLE_STREAMOUTSTAT*
+		 * 4 - *S_PARTIAL_FLUSH
+		 */
+#define		PACKET3_EVENT_WRITE__EVENT_TYPE(x) ((((unsigned)(x)) & 0x3F) << 0)
+#define		PACKET3_EVENT_WRITE__EVENT_INDEX(x) ((((unsigned)(x)) & 0xF) << 8)
+#define		PACKET3_EVENT_WRITE__SAMP_PLST_CNTR_MODE(x) ((((unsigned)(x)) & 0x3) << 29)
+#define		PACKET3_EVENT_WRITE__OFFLOAD_ENABLE(x) ((((unsigned)(x)) & 0x1) << 0)
+#define 	PACKET3_EVENT_WRITE__ADDRESS_LO(x) ((((unsigned)(x)) & 0x1FFFFFFF) << 3)
+#define 	PACKET3_EVENT_WRITE__ADDRESS_HI(x) ((unsigned)(x))
+#define 	PACKET3_EVENT_WRITE__EVENT_INDEX__OTHER 0
+#define 	PACKET3_EVENT_WRITE__EVENT_INDEX__SAMPLE_PIPELINESTAT 2
+#define 	PACKET3_EVENT_WRITE__EVENT_INDEX__CS_PARTIAL_FLUSH 4
+#define 	PACKET3_EVENT_WRITE__EVENT_INDEX__SAMPLE_STREAMOUTSTATS 8
+#define 	PACKET3_EVENT_WRITE__EVENT_INDEX__SAMPLE_STREAMOUTSTATS1 9
+#define 	PACKET3_EVENT_WRITE__EVENT_INDEX__SAMPLE_STREAMOUTSTATS2 10
+#define 	PACKET3_EVENT_WRITE__EVENT_INDEX__SAMPLE_STREAMOUTSTATS3 11
+#define 	PACKET3_EVENT_WRITE__SAMP_PLST_CNTR_MODE__LEGACY_MODE 0
+#define 	PACKET3_EVENT_WRITE__SAMP_PLST_CNTR_MODE__MIXED_MODE1 1
+#define 	PACKET3_EVENT_WRITE__SAMP_PLST_CNTR_MODE__NEW_MODE 2
+#define 	PACKET3_EVENT_WRITE__SAMP_PLST_CNTR_MODE__MIXED_MODE3 3
+#define	PACKET3_EVENT_WRITE_EOP				0x47
+#define	PACKET3_EVENT_WRITE_EOS				0x48
+#define	PACKET3_RELEASE_MEM				0x49
+#define		PACKET3_RELEASE_MEM_EVENT_TYPE(x)	((x) << 0)
+#define		PACKET3_RELEASE_MEM_EVENT_INDEX(x)	((x) << 8)
+#define		PACKET3_RELEASE_MEM_GCR_GLM_WB		(1 << 12)
+#define		PACKET3_RELEASE_MEM_GCR_GLM_INV		(1 << 13)
+#define		PACKET3_RELEASE_MEM_GCR_GLV_INV		(1 << 14)
+#define		PACKET3_RELEASE_MEM_GCR_GL1_INV		(1 << 15)
+#define		PACKET3_RELEASE_MEM_GCR_GL2_US		(1 << 16)
+#define		PACKET3_RELEASE_MEM_GCR_GL2_RANGE	(1 << 17)
+#define		PACKET3_RELEASE_MEM_GCR_GL2_DISCARD	(1 << 19)
+#define		PACKET3_RELEASE_MEM_GCR_GL2_INV		(1 << 20)
+#define		PACKET3_RELEASE_MEM_GCR_GL2_WB		(1 << 21)
+#define		PACKET3_RELEASE_MEM_GCR_SEQ		(1 << 22)
+#define		PACKET3_RELEASE_MEM_CACHE_POLICY(x)	((x) << 25)
+		/* 0 - cache_policy__me_release_mem__lru
+		 * 1 - cache_policy__me_release_mem__stream
+		 * 2 - cache_policy__me_release_mem__noa
+		 * 3 - cache_policy__me_release_mem__bypass
+		 */
+#define		PACKET3_RELEASE_MEM_EXECUTE		(1 << 28)
+
+#define		PACKET3_RELEASE_MEM_DATA_SEL(x)		((x) << 29)
+		/* 0 - discard
+		 * 1 - send low 32bit data
+		 * 2 - send 64bit data
+		 * 3 - send 64bit GPU counter value
+		 * 4 - send 64bit sys counter value
+		 */
+#define		PACKET3_RELEASE_MEM_INT_SEL(x)		((x) << 24)
+		/* 0 - none
+		 * 1 - interrupt only (DATA_SEL = 0)
+		 * 2 - interrupt when data write is confirmed
+		 */
+#define		PACKET3_RELEASE_MEM_DST_SEL(x)		((x) << 16)
+		/* 0 - MC
+		 * 1 - TC/L2
+		 */
+
+
+
+#define	PACKET3_PREAMBLE_CNTL				0x4A
+#              define PACKET3_PREAMBLE_BEGIN_CLEAR_STATE     (2 << 28)
+#              define PACKET3_PREAMBLE_END_CLEAR_STATE       (3 << 28)
+#define	PACKET3_DMA_DATA				0x50
+/* 1. header
+ * 2. CONTROL
+ * 3. SRC_ADDR_LO or DATA [31:0]
+ * 4. SRC_ADDR_HI [31:0]
+ * 5. DST_ADDR_LO [31:0]
+ * 6. DST_ADDR_HI [7:0]
+ * 7. COMMAND [31:26] | BYTE_COUNT [25:0]
+ */
+/* CONTROL */
+#              define PACKET3_DMA_DATA_ENGINE(x)     ((x) << 0)
+		/* 0 - ME
+		 * 1 - PFP
+		 */
+#              define PACKET3_DMA_DATA_SRC_CACHE_POLICY(x) ((x) << 13)
+		/* 0 - LRU
+		 * 1 - Stream
+		 */
+#              define PACKET3_DMA_DATA_DST_SEL(x)  ((x) << 20)
+		/* 0 - DST_ADDR using DAS
+		 * 1 - GDS
+		 * 3 - DST_ADDR using L2
+		 */
+#              define PACKET3_DMA_DATA_DST_CACHE_POLICY(x) ((x) << 25)
+		/* 0 - LRU
+		 * 1 - Stream
+		 */
+#              define PACKET3_DMA_DATA_SRC_SEL(x)  ((x) << 29)
+		/* 0 - SRC_ADDR using SAS
+		 * 1 - GDS
+		 * 2 - DATA
+		 * 3 - SRC_ADDR using L2
+		 */
+#              define PACKET3_DMA_DATA_CP_SYNC     (1 << 31)
+/* COMMAND */
+#              define PACKET3_DMA_DATA_CMD_SAS     (1 << 26)
+		/* 0 - memory
+		 * 1 - register
+		 */
+#              define PACKET3_DMA_DATA_CMD_DAS     (1 << 27)
+		/* 0 - memory
+		 * 1 - register
+		 */
+#              define PACKET3_DMA_DATA_CMD_SAIC    (1 << 28)
+#              define PACKET3_DMA_DATA_CMD_DAIC    (1 << 29)
+#              define PACKET3_DMA_DATA_CMD_RAW_WAIT  (1 << 30)
+#define	PACKET3_CONTEXT_REG_RMW				0x51
+#define	PACKET3_GFX_CNTX_UPDATE				0x52
+#define	PACKET3_BLK_CNTX_UPDATE				0x53
+#define	PACKET3_INCR_UPDT_STATE				0x55
+#define	PACKET3_ACQUIRE_MEM				0x58
+/* 1.  HEADER
+ * 2.  COHER_CNTL [30:0]
+ * 2.1 ENGINE_SEL [31:31]
+ * 2.  COHER_SIZE [31:0]
+ * 3.  COHER_SIZE_HI [7:0]
+ * 4.  COHER_BASE_LO [31:0]
+ * 5.  COHER_BASE_HI [23:0]
+ * 7.  POLL_INTERVAL [15:0]
+ * 8.  GCR_CNTL [18:0]
+ */
+#define 	PACKET3_ACQUIRE_MEM_GCR_CNTL_GLI_INV(x) ((x) << 0)
+		/*
+		 * 0:NOP
+		 * 1:ALL
+		 * 2:RANGE
+		 * 3:FIRST_LAST
+		 */
+#define 	PACKET3_ACQUIRE_MEM_GCR_CNTL_GL1_RANGE(x) ((x) << 2)
+		/*
+		 * 0:ALL
+		 * 1:reserved
+		 * 2:RANGE
+		 * 3:FIRST_LAST
+		 */
+#define 	PACKET3_ACQUIRE_MEM_GCR_CNTL_GLM_WB(x) ((x) << 4)
+#define 	PACKET3_ACQUIRE_MEM_GCR_CNTL_GLM_INV(x) ((x) << 5)
+#define 	PACKET3_ACQUIRE_MEM_GCR_CNTL_GLK_WB(x) ((x) << 6)
+#define 	PACKET3_ACQUIRE_MEM_GCR_CNTL_GLK_INV(x) ((x) << 7)
+#define 	PACKET3_ACQUIRE_MEM_GCR_CNTL_GLV_INV(x) ((x) << 8)
+#define 	PACKET3_ACQUIRE_MEM_GCR_CNTL_GL1_INV(x) ((x) << 9)
+#define 	PACKET3_ACQUIRE_MEM_GCR_CNTL_GL2_US(x) ((x) << 10)
+#define 	PACKET3_ACQUIRE_MEM_GCR_CNTL_GL2_RANGE(x) ((x) << 11)
+		/*
+		 * 0:ALL
+		 * 1:VOL
+		 * 2:RANGE
+		 * 3:FIRST_LAST
+		 */
+#define 	PACKET3_ACQUIRE_MEM_GCR_CNTL_GL2_DISCARD(x)  ((x) << 13)
+#define 	PACKET3_ACQUIRE_MEM_GCR_CNTL_GL2_INV(x) ((x) << 14)
+#define 	PACKET3_ACQUIRE_MEM_GCR_CNTL_GL2_WB(x) ((x) << 15)
+#define 	PACKET3_ACQUIRE_MEM_GCR_CNTL_SEQ(x) ((x) << 16)
+		/*
+		 * 0: PARALLEL
+		 * 1: FORWARD
+		 * 2: REVERSE
+		 */
+#define 	PACKET3_ACQUIRE_MEM_GCR_RANGE_IS_PA  (1 << 18)
+#define 	PACKET3_ACQUIRE_MEM__COHER_SIZE(x) ((unsigned)(x))
+#define 	PACKET3_ACQUIRE_MEM__COHER_SIZE_HI(x) ((((unsigned)(x)) & 0xFF) << 0)
+#define 	PACKET3_ACQUIRE_MEM__COHER_BASE_LO(x) ((unsigned)(x))
+#define 	PACKET3_ACQUIRE_MEM__COHER_BASE_HI(x) ((((unsigned)(x)) & 0xFFFFFF) << 0)
+#define 	PACKET3_ACQUIRE_MEM__POLL_INTERVAL(x) ((((unsigned)(x)) & 0xFFFF) << 0)
+#define 	PACKET3_ACQUIRE_MEM__GCR_CNTL(x) ((((unsigned)(x)) & 0x7FFFF) << 0)
+#define	PACKET3_REWIND					0x59
+#define	PACKET3_INTERRUPT				0x5A
+#define	PACKET3_GEN_PDEPTE				0x5B
+#define	PACKET3_INDIRECT_BUFFER_PASID			0x5C
+#define	PACKET3_PRIME_UTCL2				0x5D
+#define	PACKET3_LOAD_UCONFIG_REG			0x5E
+#define	PACKET3_LOAD_SH_REG				0x5F
+#define	PACKET3_LOAD_CONFIG_REG				0x60
+#define	PACKET3_LOAD_CONTEXT_REG			0x61
+#define	PACKET3_LOAD_COMPUTE_STATE			0x62
+#define	PACKET3_LOAD_SH_REG_INDEX			0x63
+#define	PACKET3_SET_CONFIG_REG				0x68
+#define		PACKET3_SET_CONFIG_REG_START			0x00002000
+#define		PACKET3_SET_CONFIG_REG_END			0x00002c00
+#define	PACKET3_SET_CONTEXT_REG				0x69
+#define		PACKET3_SET_CONTEXT_REG_START			0x0000a000
+#define		PACKET3_SET_CONTEXT_REG_END			0x0000a400
+#define	PACKET3_SET_CONTEXT_REG_INDEX			0x6A
+#define	PACKET3_SET_VGPR_REG_DI_MULTI			0x71
+#define	PACKET3_SET_SH_REG_DI				0x72
+#define	PACKET3_SET_CONTEXT_REG_INDIRECT		0x73
+#define	PACKET3_SET_SH_REG_DI_MULTI			0x74
+#define	PACKET3_GFX_PIPE_LOCK				0x75
+#define	PACKET3_SET_SH_REG				0x76
+#define		PACKET3_SET_SH_REG_START			0x00002c00
+#define		PACKET3_SET_SH_REG_END				0x00003000
+#define 	PACKET3_SET_SH_REG__REG_OFFSET(x) ((((unsigned)(x)) & 0xFFFF) << 0)
+#define 	PACKET3_SET_SH_REG__VMID_SHIFT(x) ((((unsigned)(x)) & 0x1F) << 23)
+#define 	PACKET3_SET_SH_REG__INDEX(x) ((((unsigned)(x)) & 0xF) << 28)
+#define 	PACKET3_SET_SH_REG__INDEX__DEFAULT 0
+#define 	PACKET3_SET_SH_REG__INDEX__INSERT_VMID 1
+#define	PACKET3_SET_SH_REG_OFFSET			0x77
+#define	PACKET3_SET_QUEUE_REG				0x78
+#define	PACKET3_SET_UCONFIG_REG				0x79
+#define		PACKET3_SET_UCONFIG_REG_START			0x0000c000
+#define		PACKET3_SET_UCONFIG_REG_END			0x0000c400
+#define 	PACKET3_SET_UCONFIG_REG__REG_OFFSET(x) ((((unsigned)(x)) & 0xFFFF) << 0)
+#define	PACKET3_SET_UCONFIG_REG_INDEX			0x7A
+#define	PACKET3_FORWARD_HEADER				0x7C
+#define	PACKET3_SCRATCH_RAM_WRITE			0x7D
+#define	PACKET3_SCRATCH_RAM_READ			0x7E
+#define	PACKET3_LOAD_CONST_RAM				0x80
+#define	PACKET3_WRITE_CONST_RAM				0x81
+#define	PACKET3_DUMP_CONST_RAM				0x83
+#define	PACKET3_INCREMENT_CE_COUNTER			0x84
+#define	PACKET3_INCREMENT_DE_COUNTER			0x85
+#define	PACKET3_WAIT_ON_CE_COUNTER			0x86
+#define	PACKET3_WAIT_ON_DE_COUNTER_DIFF			0x88
+#define	PACKET3_SWITCH_BUFFER				0x8B
+#define	PACKET3_DISPATCH_DRAW_PREAMBLE			0x8C
+#define	PACKET3_DISPATCH_DRAW_PREAMBLE_ACE		0x8C
+#define	PACKET3_DISPATCH_DRAW				0x8D
+#define	PACKET3_DISPATCH_DRAW_ACE			0x8D
+#define	PACKET3_GET_LOD_STATS				0x8E
+#define	PACKET3_DRAW_MULTI_PREAMBLE			0x8F
+#define	PACKET3_FRAME_CONTROL				0x90
+#			define FRAME_TMZ	(1 << 0)
+#			define FRAME_CMD(x) ((x) << 28)
+			/*
+			 * x=0: tmz_begin
+			 * x=1: tmz_end
+			 */
+#define	PACKET3_INDEX_ATTRIBUTES_INDIRECT		0x91
+#define	PACKET3_WAIT_REG_MEM64				0x93
+#define	PACKET3_COND_PREEMPT				0x94
+#define	PACKET3_HDP_FLUSH				0x95
+#define	PACKET3_COPY_DATA_RB				0x96
+#define	PACKET3_INVALIDATE_TLBS				0x98
+#              define PACKET3_INVALIDATE_TLBS_DST_SEL(x)     ((x) << 0)
+#              define PACKET3_INVALIDATE_TLBS_ALL_HUB(x)     ((x) << 4)
+#              define PACKET3_INVALIDATE_TLBS_PASID(x)       ((x) << 5)
+#              define PACKET3_INVALIDATE_TLBS_FLUSH_TYPE(x)  ((x) << 29)
+#define	PACKET3_AQL_PACKET				0x99
+#define	PACKET3_DMA_DATA_FILL_MULTI			0x9A
+#define	PACKET3_SET_SH_REG_INDEX			0x9B
+#define	PACKET3_DRAW_INDIRECT_COUNT_MULTI		0x9C
+#define	PACKET3_DRAW_INDEX_INDIRECT_COUNT_MULTI		0x9D
+#define	PACKET3_DUMP_CONST_RAM_OFFSET			0x9E
+#define	PACKET3_LOAD_CONTEXT_REG_INDEX			0x9F
+#define	PACKET3_SET_RESOURCES				0xA0
+/* 1. header
+ * 2. CONTROL
+ * 3. QUEUE_MASK_LO [31:0]
+ * 4. QUEUE_MASK_HI [31:0]
+ * 5. GWS_MASK_LO [31:0]
+ * 6. GWS_MASK_HI [31:0]
+ * 7. OAC_MASK [15:0]
+ * 8. GDS_HEAP_SIZE [16:11] | GDS_HEAP_BASE [5:0]
+ */
+#              define PACKET3_SET_RESOURCES_VMID_MASK(x)     ((x) << 0)
+#              define PACKET3_SET_RESOURCES_UNMAP_LATENTY(x) ((x) << 16)
+#              define PACKET3_SET_RESOURCES_QUEUE_TYPE(x)    ((x) << 29)
+#define PACKET3_MAP_PROCESS				0xA1
+#define PACKET3_MAP_QUEUES				0xA2
+/* 1. header
+ * 2. CONTROL
+ * 3. CONTROL2
+ * 4. MQD_ADDR_LO [31:0]
+ * 5. MQD_ADDR_HI [31:0]
+ * 6. WPTR_ADDR_LO [31:0]
+ * 7. WPTR_ADDR_HI [31:0]
+ */
+/* CONTROL */
+#              define PACKET3_MAP_QUEUES_QUEUE_SEL(x)       ((x) << 4)
+#              define PACKET3_MAP_QUEUES_VMID(x)            ((x) << 8)
+#              define PACKET3_MAP_QUEUES_QUEUE(x)           ((x) << 13)
+#              define PACKET3_MAP_QUEUES_PIPE(x)            ((x) << 16)
+#              define PACKET3_MAP_QUEUES_ME(x)              ((x) << 18)
+#              define PACKET3_MAP_QUEUES_QUEUE_TYPE(x)      ((x) << 21)
+#              define PACKET3_MAP_QUEUES_ALLOC_FORMAT(x)    ((x) << 24)
+#              define PACKET3_MAP_QUEUES_ENGINE_SEL(x)      ((x) << 26)
+#              define PACKET3_MAP_QUEUES_NUM_QUEUES(x)      ((x) << 29)
+/* CONTROL2 */
+#              define PACKET3_MAP_QUEUES_CHECK_DISABLE(x)   ((x) << 1)
+#              define PACKET3_MAP_QUEUES_DOORBELL_OFFSET(x) ((x) << 2)
+#define	PACKET3_UNMAP_QUEUES				0xA3
+/* 1. header
+ * 2. CONTROL
+ * 3. CONTROL2
+ * 4. CONTROL3
+ * 5. CONTROL4
+ * 6. CONTROL5
+ */
+/* CONTROL */
+#              define PACKET3_UNMAP_QUEUES_ACTION(x)           ((x) << 0)
+		/* 0 - PREEMPT_QUEUES
+		 * 1 - RESET_QUEUES
+		 * 2 - DISABLE_PROCESS_QUEUES
+		 * 3 - PREEMPT_QUEUES_NO_UNMAP
+		 */
+#              define PACKET3_UNMAP_QUEUES_QUEUE_SEL(x)        ((x) << 4)
+#              define PACKET3_UNMAP_QUEUES_ENGINE_SEL(x)       ((x) << 26)
+#              define PACKET3_UNMAP_QUEUES_NUM_QUEUES(x)       ((x) << 29)
+/* CONTROL2a */
+#              define PACKET3_UNMAP_QUEUES_PASID(x)            ((x) << 0)
+/* CONTROL2b */
+#              define PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET0(x) ((x) << 2)
+/* CONTROL3a */
+#              define PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET1(x) ((x) << 2)
+/* CONTROL3b */
+#              define PACKET3_UNMAP_QUEUES_RB_WPTR(x)          ((x) << 0)
+/* CONTROL4 */
+#              define PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET2(x) ((x) << 2)
+/* CONTROL5 */
+#              define PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET3(x) ((x) << 2)
+#define	PACKET3_QUERY_STATUS				0xA4
+/* 1. header
+ * 2. CONTROL
+ * 3. CONTROL2
+ * 4. ADDR_LO [31:0]
+ * 5. ADDR_HI [31:0]
+ * 6. DATA_LO [31:0]
+ * 7. DATA_HI [31:0]
+ */
+/* CONTROL */
+#              define PACKET3_QUERY_STATUS_CONTEXT_ID(x)       ((x) << 0)
+#              define PACKET3_QUERY_STATUS_INTERRUPT_SEL(x)    ((x) << 28)
+#              define PACKET3_QUERY_STATUS_COMMAND(x)          ((x) << 30)
+/* CONTROL2a */
+#              define PACKET3_QUERY_STATUS_PASID(x)            ((x) << 0)
+/* CONTROL2b */
+#              define PACKET3_QUERY_STATUS_DOORBELL_OFFSET(x)  ((x) << 2)
+#              define PACKET3_QUERY_STATUS_ENG_SEL(x)          ((x) << 25)
+#define	PACKET3_RUN_LIST				0xA5
+#define	PACKET3_MAP_PROCESS_VM				0xA6
+
+#define PACKET3_RUN_CLEANER_SHADER                      0xD2
+/* 1. header
+ * 2. RESERVED [31:0]
+ */
+
+/* GFX11 */
+#define	PACKET3_SET_Q_PREEMPTION_MODE			0xF0
+#              define PACKET3_SET_Q_PREEMPTION_MODE_IB_VMID(x)  ((x) << 0)
+#              define PACKET3_SET_Q_PREEMPTION_MODE_INIT_SHADOW_MEM    (1 << 0)
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h b/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h
index 8da6da90b1c9..73f87131a7e9 100644
--- a/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h
+++ b/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h
@@ -31,6 +31,9 @@
 #define GFX_CMD_RESERVED_MASK       0x7FF00000
 #define GFX_CMD_RESPONSE_MASK       0x80000000
 
+/* USBC PD FW version retrieval command */
+#define C2PMSG_CMD_GFX_USB_PD_FW_VER 0x2000000
+
 /* TEE Gfx Command IDs for the register interface.
 *  Command ID must be between 0x00010000 and 0x000F0000.
 */
@@ -40,11 +43,23 @@ enum psp_gfx_crtl_cmd_id
     GFX_CTRL_CMD_ID_INIT_GPCOM_RING = 0x00020000,   /* initialize GPCOM ring */
     GFX_CTRL_CMD_ID_DESTROY_RINGS   = 0x00030000,   /* destroy rings */
     GFX_CTRL_CMD_ID_CAN_INIT_RINGS  = 0x00040000,   /* is it allowed to initialized the rings */
+    GFX_CTRL_CMD_ID_ENABLE_INT      = 0x00050000,   /* enable PSP-to-Gfx interrupt */
+    GFX_CTRL_CMD_ID_DISABLE_INT     = 0x00060000,   /* disable PSP-to-Gfx interrupt */
+    GFX_CTRL_CMD_ID_MODE1_RST       = 0x00070000,   /* trigger the Mode 1 reset */
+    GFX_CTRL_CMD_ID_GBR_IH_SET      = 0x00080000,   /* set Gbr IH_RB_CNTL registers */
+    GFX_CTRL_CMD_ID_CONSUME_CMD     = 0x00090000,   /* send interrupt to psp for updating write pointer of vf */
+    GFX_CTRL_CMD_ID_DESTROY_GPCOM_RING = 0x000C0000, /* destroy GPCOM ring */
 
     GFX_CTRL_CMD_ID_MAX             = 0x000F0000,   /* max command ID */
 };
 
 
+/*-----------------------------------------------------------------------------
+    NOTE:   All physical addresses used in this interface are actually
+            GPU Virtual Addresses.
+*/
+
+
 /* Control registers of the TEE Gfx interface. These are located in
 *  SRBM-to-PSP mailbox registers (total 8 registers).
 */
@@ -55,8 +70,8 @@ struct psp_gfx_ctrl
     volatile uint32_t   rbi_rptr;         /* +8   Read pointer (index) of RBI ring */
     volatile uint32_t   gpcom_wptr;       /* +12  Write pointer (index) of GPCOM ring */
     volatile uint32_t   gpcom_rptr;       /* +16  Read pointer (index) of GPCOM ring */
-    volatile uint32_t   ring_addr_lo;     /* +20  bits [31:0] of physical address of ring buffer */
-    volatile uint32_t   ring_addr_hi;     /* +24  bits [63:32] of physical address of ring buffer */
+    volatile uint32_t   ring_addr_lo;     /* +20  bits [31:0] of GPU Virtual of ring buffer (VMID=0)*/
+    volatile uint32_t   ring_addr_hi;     /* +24  bits [63:32] of GPU Virtual of ring buffer (VMID=0) */
     volatile uint32_t   ring_buf_size;    /* +28  Ring buffer size (in bytes) */
 
 };
@@ -68,28 +83,56 @@ struct psp_gfx_ctrl
 */
 #define GFX_FLAG_RESPONSE               0x80000000
 
-
 /* TEE Gfx Command IDs for the ring buffer interface. */
 enum psp_gfx_cmd_id
 {
-    GFX_CMD_ID_LOAD_TA      = 0x00000001,   /* load TA */
-    GFX_CMD_ID_UNLOAD_TA    = 0x00000002,   /* unload TA */
-    GFX_CMD_ID_INVOKE_CMD   = 0x00000003,   /* send command to TA */
-    GFX_CMD_ID_LOAD_ASD     = 0x00000004,   /* load ASD Driver */
-    GFX_CMD_ID_SETUP_TMR    = 0x00000005,   /* setup TMR region */
-    GFX_CMD_ID_LOAD_IP_FW   = 0x00000006,   /* load HW IP FW */
+    GFX_CMD_ID_LOAD_TA            = 0x00000001,   /* load TA */
+    GFX_CMD_ID_UNLOAD_TA          = 0x00000002,   /* unload TA */
+    GFX_CMD_ID_INVOKE_CMD         = 0x00000003,   /* send command to TA */
+    GFX_CMD_ID_LOAD_ASD           = 0x00000004,   /* load ASD Driver */
+    GFX_CMD_ID_SETUP_TMR          = 0x00000005,   /* setup TMR region */
+    GFX_CMD_ID_LOAD_IP_FW         = 0x00000006,   /* load HW IP FW */
+    GFX_CMD_ID_DESTROY_TMR        = 0x00000007,   /* destroy TMR region */
+    GFX_CMD_ID_SAVE_RESTORE       = 0x00000008,   /* save/restore HW IP FW */
+    GFX_CMD_ID_SETUP_VMR          = 0x00000009,   /* setup VMR region */
+    GFX_CMD_ID_DESTROY_VMR        = 0x0000000A,   /* destroy VMR region */
+    GFX_CMD_ID_PROG_REG           = 0x0000000B,   /* program regs */
+    GFX_CMD_ID_GET_FW_ATTESTATION = 0x0000000F,   /* Query GPUVA of the Fw Attestation DB */
+    /* IDs upto 0x1F are reserved for older programs (Raven, Vega 10/12/20) */
+    GFX_CMD_ID_LOAD_TOC           = 0x00000020,   /* Load TOC and obtain TMR size */
+    GFX_CMD_ID_AUTOLOAD_RLC       = 0x00000021,   /* Indicates all graphics fw loaded, start RLC autoload */
+    GFX_CMD_ID_BOOT_CFG           = 0x00000022,   /* Boot Config */
+    GFX_CMD_ID_SRIOV_SPATIAL_PART = 0x00000027,   /* Configure spatial partitioning mode */
+    /*IDs of performance monitoring/profiling*/
+    GFX_CMD_ID_CONFIG_SQ_PERFMON  = 0x00000046,   /* Config CGTT_SQ_CLK_CTRL */
+    /* Dynamic memory partitioninig (NPS mode change)*/
+    GFX_CMD_ID_FB_NPS_MODE        = 0x00000048,  /* Configure memory partitioning mode */
+    GFX_CMD_ID_FB_FW_RESERV_ADDR  = 0x00000050,  /* Query FW reservation addr */
+    GFX_CMD_ID_FB_FW_RESERV_EXT_ADDR = 0x00000051,  /* Query FW reservation extended addr */
+};
 
+/* PSP boot config sub-commands */
+enum psp_gfx_boot_config_cmd
+{
+    BOOTCFG_CMD_SET         = 1, /* Set boot configuration settings */
+    BOOTCFG_CMD_GET         = 2, /* Get boot configuration settings */
+    BOOTCFG_CMD_INVALIDATE  = 3  /* Reset current boot configuration settings to VBIOS defaults */
 };
 
+/* PSP boot config bitmask values */
+enum psp_gfx_boot_config
+{
+    BOOT_CONFIG_GECC = 0x1,
+};
 
 /* Command to load Trusted Application binary into PSP OS. */
 struct psp_gfx_cmd_load_ta
 {
-    uint32_t        app_phy_addr_lo;        /* bits [31:0] of the physical address of the TA binary (must be 4 KB aligned) */
-    uint32_t        app_phy_addr_hi;        /* bits [63:32] of the physical address of the TA binary */
+    uint32_t        app_phy_addr_lo;        /* bits [31:0] of the GPU Virtual address of the TA binary (must be 4 KB aligned) */
+    uint32_t        app_phy_addr_hi;        /* bits [63:32] of the GPU Virtual address of the TA binary */
     uint32_t        app_len;                /* length of the TA binary in bytes */
-    uint32_t        cmd_buf_phy_addr_lo;    /* bits [31:0] of the physical address of CMD buffer (must be 4 KB aligned) */
-    uint32_t        cmd_buf_phy_addr_hi;    /* bits [63:32] of the physical address of CMD buffer */
+    uint32_t        cmd_buf_phy_addr_lo;    /* bits [31:0] of the GPU Virtual address of CMD buffer (must be 4 KB aligned) */
+    uint32_t        cmd_buf_phy_addr_hi;    /* bits [63:32] of the GPU Virtual address of CMD buffer */
     uint32_t        cmd_buf_len;            /* length of the CMD buffer in bytes; must be multiple of 4 KB */
 
     /* Note: CmdBufLen can be set to 0. In this case no persistent CMD buffer is provided
@@ -111,8 +154,8 @@ struct psp_gfx_cmd_unload_ta
 */
 struct psp_gfx_buf_desc
 {
-    uint32_t        buf_phy_addr_lo;       /* bits [31:0] of physical address of the buffer (must be 4 KB aligned) */
-    uint32_t        buf_phy_addr_hi;       /* bits [63:32] of physical address of the buffer */
+    uint32_t        buf_phy_addr_lo;       /* bits [31:0] of GPU Virtual address of the buffer (must be 4 KB aligned) */
+    uint32_t        buf_phy_addr_hi;       /* bits [63:32] of GPU Virtual address of the buffer */
     uint32_t        buf_size;              /* buffer size in bytes (must be multiple of 4 KB and no bigger than 64 MB) */
 
 };
@@ -145,47 +188,188 @@ struct psp_gfx_cmd_invoke_cmd
 /* Command to setup TMR region. */
 struct psp_gfx_cmd_setup_tmr
 {
-    uint32_t        buf_phy_addr_lo;       /* bits [31:0] of physical address of TMR buffer (must be 4 KB aligned) */
-    uint32_t        buf_phy_addr_hi;       /* bits [63:32] of physical address of TMR buffer */
+    uint32_t        buf_phy_addr_lo;       /* bits [31:0] of GPU Virtual address of TMR buffer (must be 4 KB aligned) */
+    uint32_t        buf_phy_addr_hi;       /* bits [63:32] of GPU Virtual address of TMR buffer */
     uint32_t        buf_size;              /* buffer size in bytes (must be multiple of 4 KB) */
+    union {
+	struct {
+		uint32_t	sriov_enabled:1; /* whether the device runs under SR-IOV*/
+		uint32_t	virt_phy_addr:1; /* driver passes both virtual and physical address to PSP*/
+		uint32_t	reserved:30;
+	} bitfield;
+	uint32_t        tmr_flags;
+    };
+    uint32_t        system_phy_addr_lo;        /* bits [31:0] of system physical address of TMR buffer (must be 4 KB aligned) */
+    uint32_t        system_phy_addr_hi;        /* bits [63:32] of system physical address of TMR buffer */
 
 };
 
-
 /* FW types for GFX_CMD_ID_LOAD_IP_FW command. Limit 31. */
-enum psp_gfx_fw_type
-{
-    GFX_FW_TYPE_NONE        = 0,
-    GFX_FW_TYPE_CP_ME       = 1,
-    GFX_FW_TYPE_CP_PFP      = 2,
-    GFX_FW_TYPE_CP_CE       = 3,
-    GFX_FW_TYPE_CP_MEC      = 4,
-    GFX_FW_TYPE_CP_MEC_ME1  = 5,
-    GFX_FW_TYPE_CP_MEC_ME2  = 6,
-    GFX_FW_TYPE_RLC_V       = 7,
-    GFX_FW_TYPE_RLC_G       = 8,
-    GFX_FW_TYPE_SDMA0       = 9,
-    GFX_FW_TYPE_SDMA1       = 10,
-    GFX_FW_TYPE_DMCU_ERAM   = 11,
-    GFX_FW_TYPE_DMCU_ISR    = 12,
-    GFX_FW_TYPE_VCN         = 13,
-    GFX_FW_TYPE_UVD         = 14,
-    GFX_FW_TYPE_VCE         = 15,
-    GFX_FW_TYPE_ISP         = 16,
-    GFX_FW_TYPE_ACP         = 17,
-    GFX_FW_TYPE_SMU         = 18,
+enum psp_gfx_fw_type {
+	GFX_FW_TYPE_NONE        = 0,    /* */
+	GFX_FW_TYPE_CP_ME       = 1,    /* CP-ME                    VG + RV */
+	GFX_FW_TYPE_CP_PFP      = 2,    /* CP-PFP                   VG + RV */
+	GFX_FW_TYPE_CP_CE       = 3,    /* CP-CE                    VG + RV */
+	GFX_FW_TYPE_CP_MEC      = 4,    /* CP-MEC FW                VG + RV */
+	GFX_FW_TYPE_CP_MEC_ME1  = 5,    /* CP-MEC Jump Table 1      VG + RV */
+	GFX_FW_TYPE_CP_MEC_ME2  = 6,    /* CP-MEC Jump Table 2      VG      */
+	GFX_FW_TYPE_RLC_V       = 7,    /* RLC-V                    VG      */
+	GFX_FW_TYPE_RLC_G       = 8,    /* RLC-G                    VG + RV */
+	GFX_FW_TYPE_SDMA0       = 9,    /* SDMA0                    VG + RV */
+	GFX_FW_TYPE_SDMA1       = 10,   /* SDMA1                    VG      */
+	GFX_FW_TYPE_DMCU_ERAM   = 11,   /* DMCU-ERAM                VG + RV */
+	GFX_FW_TYPE_DMCU_ISR    = 12,   /* DMCU-ISR                 VG + RV */
+	GFX_FW_TYPE_VCN         = 13,   /* VCN                           RV */
+	GFX_FW_TYPE_UVD         = 14,   /* UVD                      VG      */
+	GFX_FW_TYPE_VCE         = 15,   /* VCE                      VG      */
+	GFX_FW_TYPE_ISP         = 16,   /* ISP                           RV */
+	GFX_FW_TYPE_ACP         = 17,   /* ACP                           RV */
+	GFX_FW_TYPE_SMU         = 18,   /* SMU                      VG      */
+	GFX_FW_TYPE_MMSCH       = 19,   /* MMSCH                    VG      */
+	GFX_FW_TYPE_RLC_RESTORE_LIST_GPM_MEM        = 20,   /* RLC GPM                  VG + RV */
+	GFX_FW_TYPE_RLC_RESTORE_LIST_SRM_MEM        = 21,   /* RLC SRM                  VG + RV */
+	GFX_FW_TYPE_RLC_RESTORE_LIST_SRM_CNTL       = 22,   /* RLC CNTL                 VG + RV */
+	GFX_FW_TYPE_UVD1        = 23,   /* UVD1                     VG-20   */
+	GFX_FW_TYPE_TOC         = 24,   /* TOC                      NV-10   */
+	GFX_FW_TYPE_RLC_P                           = 25,   /* RLC P                    NV      */
+	GFX_FW_TYPE_RLC_IRAM                        = 26,   /* RLC_IRAM                 NV      */
+	GFX_FW_TYPE_GLOBAL_TAP_DELAYS               = 27,   /* GLOBAL TAP DELAYS        NV      */
+	GFX_FW_TYPE_SE0_TAP_DELAYS                  = 28,   /* SE0 TAP DELAYS           NV      */
+	GFX_FW_TYPE_SE1_TAP_DELAYS                  = 29,   /* SE1 TAP DELAYS           NV      */
+	GFX_FW_TYPE_GLOBAL_SE0_SE1_SKEW_DELAYS      = 30,   /* GLOBAL SE0/1 SKEW DELAYS NV      */
+	GFX_FW_TYPE_SDMA0_JT                        = 31,   /* SDMA0 JT                 NV      */
+	GFX_FW_TYPE_SDMA1_JT                        = 32,   /* SDNA1 JT                 NV      */
+	GFX_FW_TYPE_CP_MES                          = 33,   /* CP MES                   NV      */
+	GFX_FW_TYPE_MES_STACK                       = 34,   /* MES STACK                NV      */
+	GFX_FW_TYPE_RLC_SRM_DRAM_SR                 = 35,   /* RLC SRM DRAM             NV      */
+	GFX_FW_TYPE_RLCG_SCRATCH_SR                 = 36,   /* RLCG SCRATCH             NV      */
+	GFX_FW_TYPE_RLCP_SCRATCH_SR                 = 37,   /* RLCP SCRATCH             NV      */
+	GFX_FW_TYPE_RLCV_SCRATCH_SR                 = 38,   /* RLCV SCRATCH             NV      */
+	GFX_FW_TYPE_RLX6_DRAM_SR                    = 39,   /* RLX6 DRAM                NV      */
+	GFX_FW_TYPE_SDMA0_PG_CONTEXT                = 40,   /* SDMA0 PG CONTEXT         NV      */
+	GFX_FW_TYPE_SDMA1_PG_CONTEXT                = 41,   /* SDMA1 PG CONTEXT         NV      */
+	GFX_FW_TYPE_GLOBAL_MUX_SELECT_RAM           = 42,   /* GLOBAL MUX SEL RAM       NV      */
+	GFX_FW_TYPE_SE0_MUX_SELECT_RAM              = 43,   /* SE0 MUX SEL RAM          NV      */
+	GFX_FW_TYPE_SE1_MUX_SELECT_RAM              = 44,   /* SE1 MUX SEL RAM          NV      */
+	GFX_FW_TYPE_ACCUM_CTRL_RAM                  = 45,   /* ACCUM CTRL RAM           NV      */
+	GFX_FW_TYPE_RLCP_CAM                        = 46,   /* RLCP CAM                 NV      */
+	GFX_FW_TYPE_RLC_SPP_CAM_EXT                 = 47,   /* RLC SPP CAM EXT          NV      */
+	GFX_FW_TYPE_RLC_DRAM_BOOT                   = 48,   /* RLC DRAM BOOT            NV      */
+	GFX_FW_TYPE_VCN0_RAM                        = 49,   /* VCN_RAM                  NV + RN */
+	GFX_FW_TYPE_VCN1_RAM                        = 50,   /* VCN_RAM                  NV + RN */
+	GFX_FW_TYPE_DMUB                            = 51,   /* DMUB                          RN */
+	GFX_FW_TYPE_SDMA2                           = 52,   /* SDMA2                    MI      */
+	GFX_FW_TYPE_SDMA3                           = 53,   /* SDMA3                    MI      */
+	GFX_FW_TYPE_SDMA4                           = 54,   /* SDMA4                    MI      */
+	GFX_FW_TYPE_SDMA5                           = 55,   /* SDMA5                    MI      */
+	GFX_FW_TYPE_SDMA6                           = 56,   /* SDMA6                    MI      */
+	GFX_FW_TYPE_SDMA7                           = 57,   /* SDMA7                    MI      */
+	GFX_FW_TYPE_VCN1                            = 58,   /* VCN1                     MI      */
+	GFX_FW_TYPE_CAP                             = 62,   /* CAP_FW                           */
+	GFX_FW_TYPE_SE2_TAP_DELAYS                  = 65,   /* SE2 TAP DELAYS           NV      */
+	GFX_FW_TYPE_SE3_TAP_DELAYS                  = 66,   /* SE3 TAP DELAYS           NV      */
+	GFX_FW_TYPE_REG_LIST                        = 67,   /* REG_LIST                 MI      */
+	GFX_FW_TYPE_IMU_I                           = 68,   /* IMU Instruction FW       SOC21   */
+	GFX_FW_TYPE_IMU_D                           = 69,   /* IMU Data FW              SOC21   */
+	GFX_FW_TYPE_LSDMA                           = 70,   /* LSDMA FW                 SOC21   */
+	GFX_FW_TYPE_SDMA_UCODE_TH0                  = 71,   /* SDMA Thread 0/CTX        SOC21   */
+	GFX_FW_TYPE_SDMA_UCODE_TH1                  = 72,   /* SDMA Thread 1/CTL        SOC21   */
+	GFX_FW_TYPE_PPTABLE                         = 73,   /* PPTABLE                  SOC21   */
+	GFX_FW_TYPE_DISCRETE_USB4                   = 74,   /* dUSB4 FW                 SOC21   */
+	GFX_FW_TYPE_TA                              = 75,   /* SRIOV TA FW UUID         SOC21   */
+	GFX_FW_TYPE_RS64_MES                        = 76,   /* RS64 MES ucode           SOC21   */
+	GFX_FW_TYPE_RS64_MES_STACK                  = 77,   /* RS64 MES stack ucode     SOC21   */
+	GFX_FW_TYPE_RS64_KIQ                        = 78,   /* RS64 KIQ ucode           SOC21   */
+	GFX_FW_TYPE_RS64_KIQ_STACK                  = 79,   /* RS64 KIQ Heap stack      SOC21   */
+	GFX_FW_TYPE_ISP_DATA                        = 80,   /* ISP DATA                 SOC21   */
+	GFX_FW_TYPE_CP_MES_KIQ                      = 81,   /* MES KIQ ucode            SOC21   */
+	GFX_FW_TYPE_MES_KIQ_STACK                   = 82,   /* MES KIQ stack            SOC21   */
+	GFX_FW_TYPE_UMSCH_DATA                      = 83,   /* User Mode Scheduler Data SOC21   */
+	GFX_FW_TYPE_UMSCH_UCODE                     = 84,   /* User Mode Scheduler Ucode SOC21  */
+	GFX_FW_TYPE_UMSCH_CMD_BUFFER                = 85,   /* User Mode Scheduler Command Buffer SOC21 */
+	GFX_FW_TYPE_USB_DP_COMBO_PHY                = 86,   /* USB-Display port Combo   SOC21   */
+	GFX_FW_TYPE_RS64_PFP                        = 87,   /* RS64 PFP                 SOC21   */
+	GFX_FW_TYPE_RS64_ME                         = 88,   /* RS64 ME                  SOC21   */
+	GFX_FW_TYPE_RS64_MEC                        = 89,   /* RS64 MEC                 SOC21   */
+	GFX_FW_TYPE_RS64_PFP_P0_STACK               = 90,   /* RS64 PFP stack P0        SOC21   */
+	GFX_FW_TYPE_RS64_PFP_P1_STACK               = 91,   /* RS64 PFP stack P1        SOC21   */
+	GFX_FW_TYPE_RS64_ME_P0_STACK                = 92,   /* RS64 ME stack P0         SOC21   */
+	GFX_FW_TYPE_RS64_ME_P1_STACK                = 93,   /* RS64 ME stack P1         SOC21   */
+	GFX_FW_TYPE_RS64_MEC_P0_STACK               = 94,   /* RS64 MEC stack P0        SOC21   */
+	GFX_FW_TYPE_RS64_MEC_P1_STACK               = 95,   /* RS64 MEC stack P1        SOC21   */
+	GFX_FW_TYPE_RS64_MEC_P2_STACK               = 96,   /* RS64 MEC stack P2        SOC21   */
+	GFX_FW_TYPE_RS64_MEC_P3_STACK               = 97,   /* RS64 MEC stack P3        SOC21   */
+	GFX_FW_TYPE_VPEC_FW1                        = 100,  /* VPEC FW1 To Save         VPE     */
+	GFX_FW_TYPE_VPEC_FW2                        = 101,  /* VPEC FW2 To Save         VPE     */
+	GFX_FW_TYPE_VPE                             = 102,
+	GFX_FW_TYPE_JPEG_RAM                        = 128,  /**< JPEG Command buffer */
+	GFX_FW_TYPE_P2S_TABLE                       = 129,
+	GFX_FW_TYPE_MAX
 };
 
 /* Command to load HW IP FW. */
 struct psp_gfx_cmd_load_ip_fw
 {
-    uint32_t                fw_phy_addr_lo;    /* bits [31:0] of physical address of FW location (must be 4 KB aligned) */
-    uint32_t                fw_phy_addr_hi;    /* bits [63:32] of physical address of FW location */
+    uint32_t                fw_phy_addr_lo;    /* bits [31:0] of GPU Virtual address of FW location (must be 4 KB aligned) */
+    uint32_t                fw_phy_addr_hi;    /* bits [63:32] of GPU Virtual address of FW location */
     uint32_t                fw_size;           /* FW buffer size in bytes */
     enum psp_gfx_fw_type    fw_type;           /* FW type */
 
 };
 
+/* Command to save/restore HW IP FW. */
+struct psp_gfx_cmd_save_restore_ip_fw
+{
+    uint32_t                save_fw;              /* if set, command is used for saving fw otherwise for resetoring*/
+    uint32_t                save_restore_addr_lo; /* bits [31:0] of FB address of GART memory used as save/restore buffer (must be 4 KB aligned) */
+    uint32_t                save_restore_addr_hi; /* bits [63:32] of FB address of GART memory used as save/restore buffer */
+    uint32_t                buf_size;             /* Size of the save/restore buffer in bytes */
+    enum psp_gfx_fw_type    fw_type;              /* FW type */
+};
+
+/* Command to setup register program */
+struct psp_gfx_cmd_reg_prog {
+	uint32_t	reg_value;
+	uint32_t	reg_id;
+};
+
+/* Command to load TOC */
+struct psp_gfx_cmd_load_toc
+{
+    uint32_t        toc_phy_addr_lo;        /* bits [31:0] of GPU Virtual address of FW location (must be 4 KB aligned) */
+    uint32_t        toc_phy_addr_hi;        /* bits [63:32] of GPU Virtual address of FW location */
+    uint32_t        toc_size;               /* FW buffer size in bytes */
+};
+
+/* Dynamic boot configuration */
+struct psp_gfx_cmd_boot_cfg
+{
+    uint32_t                        timestamp;            /* calendar time as number of seconds */
+    enum psp_gfx_boot_config_cmd    sub_cmd;              /* sub-command indicating how to process command data */
+    uint32_t                        boot_config;          /* dynamic boot configuration bitmask */
+    uint32_t                        boot_config_valid;    /* dynamic boot configuration valid bits bitmask */
+};
+
+struct psp_gfx_cmd_sriov_spatial_part {
+	uint32_t mode;
+	uint32_t override_ips;
+	uint32_t override_xcds_avail;
+	uint32_t override_this_aid;
+};
+
+/*Structure for sq performance monitoring/profiling enable/disable*/
+struct psp_gfx_cmd_config_sq_perfmon {
+	uint32_t        gfx_xcp_mask;
+	uint8_t         core_override;
+	uint8_t         reg_override;
+	uint8_t         perfmon_override;
+	uint8_t         reserved[5];
+};
+
+struct psp_gfx_cmd_fb_memory_part {
+	uint32_t mode; /* requested NPS mode */
+	uint32_t resvd;
+};
 
 /* All GFX ring buffer commands. */
 union psp_gfx_commands
@@ -195,9 +379,47 @@ union psp_gfx_commands
     struct psp_gfx_cmd_invoke_cmd       cmd_invoke_cmd;
     struct psp_gfx_cmd_setup_tmr        cmd_setup_tmr;
     struct psp_gfx_cmd_load_ip_fw       cmd_load_ip_fw;
+    struct psp_gfx_cmd_save_restore_ip_fw cmd_save_restore_ip_fw;
+    struct psp_gfx_cmd_reg_prog       cmd_setup_reg_prog;
+    struct psp_gfx_cmd_setup_tmr        cmd_setup_vmr;
+    struct psp_gfx_cmd_load_toc         cmd_load_toc;
+    struct psp_gfx_cmd_boot_cfg         boot_cfg;
+    struct psp_gfx_cmd_sriov_spatial_part cmd_spatial_part;
+    struct psp_gfx_cmd_config_sq_perfmon config_sq_perfmon;
+    struct psp_gfx_cmd_fb_memory_part cmd_memory_part;
+};
+
+struct psp_gfx_uresp_reserved
+{
+    uint32_t reserved[8];
+};
 
+/* Command-specific response for Fw Attestation Db */
+struct psp_gfx_uresp_fwar_db_info
+{
+    uint32_t fwar_db_addr_lo;
+    uint32_t fwar_db_addr_hi;
+};
+
+/* Command-specific response for boot config. */
+struct psp_gfx_uresp_bootcfg {
+	uint32_t boot_cfg;	/* boot config data */
+};
+
+/* Command-specific response for fw reserve info */
+struct psp_gfx_uresp_fw_reserve_info {
+    uint32_t reserve_base_address_hi;
+    uint32_t reserve_base_address_lo;
+    uint32_t reserve_size;
 };
 
+/* Union of command-specific responses for GPCOM ring. */
+union psp_gfx_uresp {
+	struct psp_gfx_uresp_reserved		reserved;
+	struct psp_gfx_uresp_bootcfg		boot_cfg;
+	struct psp_gfx_uresp_fwar_db_info	fwar_db_info;
+	struct psp_gfx_uresp_fw_reserve_info	fw_reserve_info;
+};
 
 /* Structure of GFX Response buffer.
 * For GPCOM I/F it is part of GFX_CMD_RESP buffer, for RBI
@@ -205,14 +427,17 @@ union psp_gfx_commands
 */
 struct psp_gfx_resp
 {
-    uint32_t    status;             /* +0  status of command execution */
-    uint32_t    session_id;         /* +4  session ID in response to LoadTa command */
-    uint32_t    fw_addr_lo;         /* +8  bits [31:0] of FW address within TMR (in response to cmd_load_ip_fw command) */
-    uint32_t    fw_addr_hi;         /* +12 bits [63:32] of FW address within TMR (in response to cmd_load_ip_fw command) */
+    uint32_t	status;		/* +0  status of command execution */
+    uint32_t	session_id;	/* +4  session ID in response to LoadTa command */
+    uint32_t	fw_addr_lo;	/* +8  bits [31:0] of FW address within TMR (in response to cmd_load_ip_fw command) */
+    uint32_t	fw_addr_hi;	/* +12 bits [63:32] of FW address within TMR (in response to cmd_load_ip_fw command) */
+    uint32_t	tmr_size;	/* +16 size of the TMR to be reserved including MM fw and Gfx fw in response to cmd_load_toc command */
+
+    uint32_t	reserved[11];
 
-    uint32_t    reserved[4];
+    union psp_gfx_uresp uresp;      /* +64 response union containing command-specific responses */
 
-    /* total 32 bytes */
+    /* total 96 bytes */
 };
 
 /* Structure of Command buffer pointed by psp_gfx_rb_frame.cmd_buf_addr_hi
@@ -226,8 +451,8 @@ struct psp_gfx_cmd_resp
 
     /* These fields are used for RBI only. They are all 0 in GPCOM commands
     */
-    uint32_t        resp_buf_addr_lo;   /* +12 bits [31:0] of physical address of response buffer (must be 4 KB aligned) */
-    uint32_t        resp_buf_addr_hi;   /* +16 bits [63:32] of physical address of response buffer */
+    uint32_t        resp_buf_addr_lo;   /* +12 bits [31:0] of GPU Virtual address of response buffer (must be 4 KB aligned) */
+    uint32_t        resp_buf_addr_hi;   /* +16 bits [63:32] of GPU Virtual address of response buffer */
     uint32_t        resp_offset;        /* +20 offset within response buffer */
     uint32_t        resp_buf_size;      /* +24 total size of the response buffer in bytes */
 
@@ -251,19 +476,27 @@ struct psp_gfx_cmd_resp
 /* Structure of the Ring Buffer Frame */
 struct psp_gfx_rb_frame
 {
-    uint32_t    cmd_buf_addr_lo;    /* +0  bits [31:0] of physical address of command buffer (must be 4 KB aligned) */
-    uint32_t    cmd_buf_addr_hi;    /* +4  bits [63:32] of physical address of command buffer */
+    uint32_t    cmd_buf_addr_lo;    /* +0  bits [31:0] of GPU Virtual address of command buffer (must be 4 KB aligned) */
+    uint32_t    cmd_buf_addr_hi;    /* +4  bits [63:32] of GPU Virtual address of command buffer */
     uint32_t    cmd_buf_size;       /* +8  command buffer size in bytes */
-    uint32_t    fence_addr_lo;      /* +12 bits [31:0] of physical address of Fence for this frame */
-    uint32_t    fence_addr_hi;      /* +16 bits [63:32] of physical address of Fence for this frame */
+    uint32_t    fence_addr_lo;      /* +12 bits [31:0] of GPU Virtual address of Fence for this frame */
+    uint32_t    fence_addr_hi;      /* +16 bits [63:32] of GPU Virtual address of Fence for this frame */
     uint32_t    fence_value;        /* +20 Fence value */
     uint32_t    sid_lo;             /* +24 bits [31:0] of SID value (used only for RBI frames) */
     uint32_t    sid_hi;             /* +28 bits [63:32] of SID value (used only for RBI frames) */
     uint8_t     vmid;               /* +32 VMID value used for mapping of all addresses for this frame */
     uint8_t     frame_type;         /* +33 1: destory context frame, 0: all other frames; used only for RBI frames */
     uint8_t     reserved1[2];       /* +34 reserved, must be 0 */
-    uint32_t    reserved2[7];       /* +40 reserved, must be 0 */
-    /* total 64 bytes */
+    uint32_t    reserved2[7];       /* +36 reserved, must be 0 */
+                /* total 64 bytes */
+};
+
+#define PSP_ERR_UNKNOWN_COMMAND 0x00000100
+
+enum tee_error_code {
+	TEE_SUCCESS			= 0x00000000,
+	TEE_ERROR_CANCEL		= 0xFFFF0002,
+	TEE_ERROR_NOT_SUPPORTED		= 0xFFFF000A,
 };
 
 #endif /* _PSP_TEE_GFX_IF_H_ */
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v10_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v10_0.c
index f7cf994b1da2..3584b8c18fd9 100644
--- a/drivers/gpu/drm/amd/amdgpu/psp_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v10_0.c
@@ -24,161 +24,51 @@
  */
 
 #include <linux/firmware.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+
 #include "amdgpu.h"
 #include "amdgpu_psp.h"
 #include "amdgpu_ucode.h"
 #include "soc15_common.h"
 #include "psp_v10_0.h"
 
-#include "vega10/soc15ip.h"
-#include "raven1/MP/mp_10_0_offset.h"
-#include "raven1/GC/gc_9_1_offset.h"
-#include "raven1/SDMA0/sdma0_4_1_offset.h"
+#include "mp/mp_10_0_offset.h"
+#include "gc/gc_9_1_offset.h"
+#include "sdma0/sdma0_4_1_offset.h"
 
-static int
-psp_v10_0_get_fw_type(struct amdgpu_firmware_info *ucode, enum psp_gfx_fw_type *type)
-{
-	switch(ucode->ucode_id) {
-	case AMDGPU_UCODE_ID_SDMA0:
-		*type = GFX_FW_TYPE_SDMA0;
-		break;
-	case AMDGPU_UCODE_ID_SDMA1:
-		*type = GFX_FW_TYPE_SDMA1;
-		break;
-	case AMDGPU_UCODE_ID_CP_CE:
-		*type = GFX_FW_TYPE_CP_CE;
-		break;
-	case AMDGPU_UCODE_ID_CP_PFP:
-		*type = GFX_FW_TYPE_CP_PFP;
-		break;
-	case AMDGPU_UCODE_ID_CP_ME:
-		*type = GFX_FW_TYPE_CP_ME;
-		break;
-	case AMDGPU_UCODE_ID_CP_MEC1:
-		*type = GFX_FW_TYPE_CP_MEC;
-		break;
-	case AMDGPU_UCODE_ID_CP_MEC1_JT:
-		*type = GFX_FW_TYPE_CP_MEC_ME1;
-		break;
-	case AMDGPU_UCODE_ID_CP_MEC2:
-		*type = GFX_FW_TYPE_CP_MEC;
-		break;
-	case AMDGPU_UCODE_ID_CP_MEC2_JT:
-		*type = GFX_FW_TYPE_CP_MEC_ME2;
-		break;
-	case AMDGPU_UCODE_ID_RLC_G:
-		*type = GFX_FW_TYPE_RLC_G;
-		break;
-	case AMDGPU_UCODE_ID_SMC:
-		*type = GFX_FW_TYPE_SMU;
-		break;
-	case AMDGPU_UCODE_ID_UVD:
-		*type = GFX_FW_TYPE_UVD;
-		break;
-	case AMDGPU_UCODE_ID_VCE:
-		*type = GFX_FW_TYPE_VCE;
-		break;
-	case AMDGPU_UCODE_ID_MAXIMUM:
-	default:
-		return -EINVAL;
-	}
-
-	return 0;
-}
+MODULE_FIRMWARE("amdgpu/raven_asd.bin");
+MODULE_FIRMWARE("amdgpu/picasso_asd.bin");
+MODULE_FIRMWARE("amdgpu/raven2_asd.bin");
+MODULE_FIRMWARE("amdgpu/picasso_ta.bin");
+MODULE_FIRMWARE("amdgpu/raven2_ta.bin");
+MODULE_FIRMWARE("amdgpu/raven_ta.bin");
 
-int psp_v10_0_init_microcode(struct psp_context *psp)
+static int psp_v10_0_init_microcode(struct psp_context *psp)
 {
 	struct amdgpu_device *adev = psp->adev;
-	const char *chip_name;
-	char fw_name[30];
+	char ucode_prefix[30];
 	int err = 0;
-	const struct psp_firmware_header_v1_0 *hdr;
-
 	DRM_DEBUG("\n");
 
-	switch (adev->asic_type) {
-	case CHIP_RAVEN:
-		chip_name = "raven";
-		break;
-	default: BUG();
-	}
-
-	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_asd.bin", chip_name);
-	err = request_firmware(&adev->psp.asd_fw, fw_name, adev->dev);
-	if (err)
-		goto out;
+	amdgpu_ucode_ip_version_decode(adev, MP0_HWIP, ucode_prefix, sizeof(ucode_prefix));
 
-	err = amdgpu_ucode_validate(adev->psp.asd_fw);
+	err = psp_init_asd_microcode(psp, ucode_prefix);
 	if (err)
-		goto out;
-
-	hdr = (const struct psp_firmware_header_v1_0 *)adev->psp.asd_fw->data;
-	adev->psp.asd_fw_version = le32_to_cpu(hdr->header.ucode_version);
-	adev->psp.asd_feature_version = le32_to_cpu(hdr->ucode_feature_version);
-	adev->psp.asd_ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes);
-	adev->psp.asd_start_addr = (uint8_t *)hdr +
-				le32_to_cpu(hdr->header.ucode_array_offset_bytes);
-
-	return 0;
-out:
-	if (err) {
-		dev_err(adev->dev,
-			"psp v10.0: Failed to load firmware \"%s\"\n",
-			fw_name);
-		release_firmware(adev->psp.asd_fw);
-		adev->psp.asd_fw = NULL;
+		return err;
+
+	err = psp_init_ta_microcode(psp, ucode_prefix);
+	if ((amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 1, 0)) &&
+	    (adev->pdev->revision == 0xa1) &&
+	    (psp->securedisplay_context.context.bin_desc.fw_version >=
+	     0x27000008)) {
+		adev->psp.securedisplay_context.context.bin_desc.size_bytes = 0;
 	}
-
 	return err;
 }
 
-int psp_v10_0_prep_cmd_buf(struct amdgpu_firmware_info *ucode, struct psp_gfx_cmd_resp *cmd)
-{
-	int ret;
-	uint64_t fw_mem_mc_addr = ucode->mc_addr;
-	struct  common_firmware_header *header;
-
-	memset(cmd, 0, sizeof(struct psp_gfx_cmd_resp));
-	header = (struct common_firmware_header *)ucode->fw;
-
-	cmd->cmd_id = GFX_CMD_ID_LOAD_IP_FW;
-	cmd->cmd.cmd_load_ip_fw.fw_phy_addr_lo = lower_32_bits(fw_mem_mc_addr);
-	cmd->cmd.cmd_load_ip_fw.fw_phy_addr_hi = upper_32_bits(fw_mem_mc_addr);
-	cmd->cmd.cmd_load_ip_fw.fw_size = le32_to_cpu(header->ucode_size_bytes);
-
-	ret = psp_v10_0_get_fw_type(ucode, &cmd->cmd.cmd_load_ip_fw.fw_type);
-	if (ret)
-		DRM_ERROR("Unknown firmware type\n");
-
-	return ret;
-}
-
-int psp_v10_0_ring_init(struct psp_context *psp, enum psp_ring_type ring_type)
-{
-	int ret = 0;
-	struct psp_ring *ring;
-	struct amdgpu_device *adev = psp->adev;
-
-	ring = &psp->km_ring;
-
-	ring->ring_type = ring_type;
-
-	/* allocate 4k Page of Local Frame Buffer memory for ring */
-	ring->ring_size = 0x1000;
-	ret = amdgpu_bo_create_kernel(adev, ring->ring_size, PAGE_SIZE,
-				      AMDGPU_GEM_DOMAIN_VRAM,
-				      &adev->firmware.rbuf,
-				      &ring->ring_mem_mc_addr,
-				      (void **)&ring->ring_mem);
-	if (ret) {
-		ring->ring_size = 0;
-		return ret;
-	}
-
-	return 0;
-}
-
-int psp_v10_0_ring_create(struct psp_context *psp, enum psp_ring_type ring_type)
+static int psp_v10_0_ring_create(struct psp_context *psp,
+				 enum psp_ring_type ring_type)
 {
 	int ret = 0;
 	unsigned int psp_ring_reg = 0;
@@ -204,20 +94,18 @@ int psp_v10_0_ring_create(struct psp_context *psp, enum psp_ring_type ring_type)
 
 	/* Wait for response flag (bit 31) in C2PMSG_64 */
 	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
-			   0x80000000, 0x8000FFFF, false);
+			   MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
 
 	return ret;
 }
 
-int psp_v10_0_ring_destroy(struct psp_context *psp, enum psp_ring_type ring_type)
+static int psp_v10_0_ring_stop(struct psp_context *psp,
+			       enum psp_ring_type ring_type)
 {
 	int ret = 0;
-	struct psp_ring *ring;
 	unsigned int psp_ring_reg = 0;
 	struct amdgpu_device *adev = psp->adev;
 
-	ring = &psp->km_ring;
-
 	/* Write the ring destroy command to C2PMSG_64 */
 	psp_ring_reg = 3 << 16;
 	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64, psp_ring_reg);
@@ -227,166 +115,60 @@ int psp_v10_0_ring_destroy(struct psp_context *psp, enum psp_ring_type ring_type
 
 	/* Wait for response flag (bit 31) in C2PMSG_64 */
 	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
-			   0x80000000, 0x80000000, false);
-
-	amdgpu_bo_free_kernel(&adev->firmware.rbuf,
-			      &ring->ring_mem_mc_addr,
-			      (void **)&ring->ring_mem);
+			   MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
 
 	return ret;
 }
 
-int psp_v10_0_cmd_submit(struct psp_context *psp,
-		        struct amdgpu_firmware_info *ucode,
-		        uint64_t cmd_buf_mc_addr, uint64_t fence_mc_addr,
-		        int index)
+static int psp_v10_0_ring_destroy(struct psp_context *psp,
+				  enum psp_ring_type ring_type)
 {
-	unsigned int psp_write_ptr_reg = 0;
-	struct psp_gfx_rb_frame * write_frame = psp->km_ring.ring_mem;
+	int ret = 0;
 	struct psp_ring *ring = &psp->km_ring;
 	struct amdgpu_device *adev = psp->adev;
 
-	/* KM (GPCOM) prepare write pointer */
-	psp_write_ptr_reg = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67);
-
-	/* Update KM RB frame pointer to new frame */
-	if ((psp_write_ptr_reg % ring->ring_size) == 0)
-		write_frame = ring->ring_mem;
-	else
-		write_frame = ring->ring_mem + (psp_write_ptr_reg / (sizeof(struct psp_gfx_rb_frame) / 4));
-
-	/* Update KM RB frame */
-	write_frame->cmd_buf_addr_hi = upper_32_bits(cmd_buf_mc_addr);
-	write_frame->cmd_buf_addr_lo = lower_32_bits(cmd_buf_mc_addr);
-	write_frame->fence_addr_hi = upper_32_bits(fence_mc_addr);
-	write_frame->fence_addr_lo = lower_32_bits(fence_mc_addr);
-	write_frame->fence_value = index;
+	ret = psp_v10_0_ring_stop(psp, ring_type);
+	if (ret)
+		DRM_ERROR("Fail to stop psp ring\n");
 
-	/* Update the write Pointer in DWORDs */
-	psp_write_ptr_reg += sizeof(struct psp_gfx_rb_frame) / 4;
-	psp_write_ptr_reg = (psp_write_ptr_reg >= ring->ring_size) ? 0 : psp_write_ptr_reg;
-	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67, psp_write_ptr_reg);
+	amdgpu_bo_free_kernel(&adev->firmware.rbuf,
+			      &ring->ring_mem_mc_addr,
+			      (void **)&ring->ring_mem);
 
-	return 0;
+	return ret;
 }
 
-static int
-psp_v10_0_sram_map(unsigned int *sram_offset, unsigned int *sram_addr_reg_offset,
-		  unsigned int *sram_data_reg_offset,
-		  enum AMDGPU_UCODE_ID ucode_id)
+static int psp_v10_0_mode1_reset(struct psp_context *psp)
 {
-	int ret = 0;
-
-	switch(ucode_id) {
-/* TODO: needs to confirm */
-#if 0
-	case AMDGPU_UCODE_ID_SMC:
-		*sram_offset = 0;
-		*sram_addr_reg_offset = 0;
-		*sram_data_reg_offset = 0;
-		break;
-#endif
-
-	case AMDGPU_UCODE_ID_CP_CE:
-		*sram_offset = 0x0;
-		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_CE_UCODE_ADDR);
-		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_CE_UCODE_DATA);
-		break;
-
-	case AMDGPU_UCODE_ID_CP_PFP:
-		*sram_offset = 0x0;
-		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_PFP_UCODE_ADDR);
-		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_PFP_UCODE_DATA);
-		break;
-
-	case AMDGPU_UCODE_ID_CP_ME:
-		*sram_offset = 0x0;
-		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_HYP_ME_UCODE_ADDR);
-		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_HYP_ME_UCODE_DATA);
-		break;
-
-	case AMDGPU_UCODE_ID_CP_MEC1:
-		*sram_offset = 0x10000;
-		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_MEC_ME1_UCODE_ADDR);
-		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_MEC_ME1_UCODE_DATA);
-		break;
-
-	case AMDGPU_UCODE_ID_CP_MEC2:
-		*sram_offset = 0x10000;
-		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_HYP_MEC2_UCODE_ADDR);
-		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_HYP_MEC2_UCODE_DATA);
-		break;
-
-	case AMDGPU_UCODE_ID_RLC_G:
-		*sram_offset = 0x2000;
-		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_UCODE_ADDR);
-		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_UCODE_DATA);
-		break;
-
-	case AMDGPU_UCODE_ID_SDMA0:
-		*sram_offset = 0x0;
-		*sram_addr_reg_offset = SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_UCODE_ADDR);
-		*sram_data_reg_offset = SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_UCODE_DATA);
-		break;
-
-/* TODO: needs to confirm */
-#if 0
-	case AMDGPU_UCODE_ID_SDMA1:
-		*sram_offset = ;
-		*sram_addr_reg_offset = ;
-		break;
-
-	case AMDGPU_UCODE_ID_UVD:
-		*sram_offset = ;
-		*sram_addr_reg_offset = ;
-		break;
-
-	case AMDGPU_UCODE_ID_VCE:
-		*sram_offset = ;
-		*sram_addr_reg_offset = ;
-		break;
-#endif
-
-	case AMDGPU_UCODE_ID_MAXIMUM:
-	default:
-		ret = -EINVAL;
-		break;
-	}
-
-	return ret;
+	DRM_INFO("psp mode 1 reset not supported now! \n");
+	return -EINVAL;
 }
 
-bool psp_v10_0_compare_sram_data(struct psp_context *psp,
-				struct amdgpu_firmware_info *ucode,
-				enum AMDGPU_UCODE_ID ucode_type)
+static uint32_t psp_v10_0_ring_get_wptr(struct psp_context *psp)
 {
-	int err = 0;
-	unsigned int fw_sram_reg_val = 0;
-	unsigned int fw_sram_addr_reg_offset = 0;
-	unsigned int fw_sram_data_reg_offset = 0;
-	unsigned int ucode_size;
-	uint32_t *ucode_mem = NULL;
 	struct amdgpu_device *adev = psp->adev;
 
-	err = psp_v10_0_sram_map(&fw_sram_reg_val, &fw_sram_addr_reg_offset,
-				&fw_sram_data_reg_offset, ucode_type);
-	if (err)
-		return false;
-
-	WREG32(fw_sram_addr_reg_offset, fw_sram_reg_val);
-
-	ucode_size = ucode->ucode_size;
-	ucode_mem = (uint32_t *)ucode->kaddr;
-	while (!ucode_size) {
-		fw_sram_reg_val = RREG32(fw_sram_data_reg_offset);
+	return RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67);
+}
 
-		if (*ucode_mem != fw_sram_reg_val)
-			return false;
+static void psp_v10_0_ring_set_wptr(struct psp_context *psp, uint32_t value)
+{
+	struct amdgpu_device *adev = psp->adev;
 
-		ucode_mem++;
-		/* 4 bytes */
-		ucode_size -= 4;
-	}
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67, value);
+}
 
-	return true;
+static const struct psp_funcs psp_v10_0_funcs = {
+	.init_microcode = psp_v10_0_init_microcode,
+	.ring_create = psp_v10_0_ring_create,
+	.ring_stop = psp_v10_0_ring_stop,
+	.ring_destroy = psp_v10_0_ring_destroy,
+	.mode1_reset = psp_v10_0_mode1_reset,
+	.ring_get_wptr = psp_v10_0_ring_get_wptr,
+	.ring_set_wptr = psp_v10_0_ring_set_wptr,
+};
+
+void psp_v10_0_set_psp_funcs(struct psp_context *psp)
+{
+	psp->funcs = &psp_v10_0_funcs;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v10_0.h b/drivers/gpu/drm/amd/amdgpu/psp_v10_0.h
index e76cde2f01f9..20c2a94859d8 100644
--- a/drivers/gpu/drm/amd/amdgpu/psp_v10_0.h
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v10_0.h
@@ -27,20 +27,6 @@
 
 #include "amdgpu_psp.h"
 
-extern int psp_v10_0_init_microcode(struct psp_context *psp);
-extern int psp_v10_0_prep_cmd_buf(struct amdgpu_firmware_info *ucode,
-				 struct psp_gfx_cmd_resp *cmd);
-extern int psp_v10_0_ring_init(struct psp_context *psp,
-			      enum psp_ring_type ring_type);
-extern int psp_v10_0_ring_create(struct psp_context *psp,
-				 enum psp_ring_type ring_type);
-extern int psp_v10_0_ring_destroy(struct psp_context *psp,
-				  enum psp_ring_type ring_type);
-extern int psp_v10_0_cmd_submit(struct psp_context *psp,
-			       struct amdgpu_firmware_info *ucode,
-			       uint64_t cmd_buf_mc_addr, uint64_t fence_mc_addr,
-			       int index);
-extern bool psp_v10_0_compare_sram_data(struct psp_context *psp,
-				       struct amdgpu_firmware_info *ucode,
-				       enum AMDGPU_UCODE_ID ucode_type);
+void psp_v10_0_set_psp_funcs(struct psp_context *psp);
+
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c
new file mode 100644
index 000000000000..64b240b51f1a
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c
@@ -0,0 +1,663 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include <linux/vmalloc.h>
+#include <drm/drm_drv.h>
+
+#include "amdgpu.h"
+#include "amdgpu_psp.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_ucode.h"
+#include "soc15_common.h"
+#include "psp_v11_0.h"
+
+#include "mp/mp_11_0_offset.h"
+#include "mp/mp_11_0_sh_mask.h"
+#include "gc/gc_9_0_offset.h"
+#include "sdma0/sdma0_4_0_offset.h"
+#include "nbio/nbio_7_4_offset.h"
+
+#include "oss/osssys_4_0_offset.h"
+#include "oss/osssys_4_0_sh_mask.h"
+
+MODULE_FIRMWARE("amdgpu/vega20_sos.bin");
+MODULE_FIRMWARE("amdgpu/vega20_asd.bin");
+MODULE_FIRMWARE("amdgpu/vega20_ta.bin");
+MODULE_FIRMWARE("amdgpu/navi10_sos.bin");
+MODULE_FIRMWARE("amdgpu/navi10_asd.bin");
+MODULE_FIRMWARE("amdgpu/navi10_ta.bin");
+MODULE_FIRMWARE("amdgpu/navi14_sos.bin");
+MODULE_FIRMWARE("amdgpu/navi14_asd.bin");
+MODULE_FIRMWARE("amdgpu/navi14_ta.bin");
+MODULE_FIRMWARE("amdgpu/navi12_sos.bin");
+MODULE_FIRMWARE("amdgpu/navi12_asd.bin");
+MODULE_FIRMWARE("amdgpu/navi12_ta.bin");
+MODULE_FIRMWARE("amdgpu/navi12_cap.bin");
+MODULE_FIRMWARE("amdgpu/arcturus_sos.bin");
+MODULE_FIRMWARE("amdgpu/arcturus_asd.bin");
+MODULE_FIRMWARE("amdgpu/arcturus_ta.bin");
+MODULE_FIRMWARE("amdgpu/sienna_cichlid_sos.bin");
+MODULE_FIRMWARE("amdgpu/sienna_cichlid_ta.bin");
+MODULE_FIRMWARE("amdgpu/sienna_cichlid_cap.bin");
+MODULE_FIRMWARE("amdgpu/navy_flounder_sos.bin");
+MODULE_FIRMWARE("amdgpu/navy_flounder_ta.bin");
+MODULE_FIRMWARE("amdgpu/vangogh_asd.bin");
+MODULE_FIRMWARE("amdgpu/vangogh_toc.bin");
+MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_sos.bin");
+MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_ta.bin");
+MODULE_FIRMWARE("amdgpu/beige_goby_sos.bin");
+MODULE_FIRMWARE("amdgpu/beige_goby_ta.bin");
+
+/* address block */
+#define smnMP1_FIRMWARE_FLAGS		0x3010024
+/* navi10 reg offset define */
+#define mmRLC_GPM_UCODE_ADDR_NV10	0x5b61
+#define mmRLC_GPM_UCODE_DATA_NV10	0x5b62
+#define mmSDMA0_UCODE_ADDR_NV10		0x5880
+#define mmSDMA0_UCODE_DATA_NV10		0x5881
+/* memory training timeout define */
+#define MEM_TRAIN_SEND_MSG_TIMEOUT_US	3000000
+
+/* For large FW files the time to complete can be very long */
+#define USBC_PD_POLLING_LIMIT_S 240
+
+/* Read USB-PD from LFB */
+#define GFX_CMD_USB_PD_USE_LFB 0x480
+
+static int psp_v11_0_init_microcode(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+	char ucode_prefix[30];
+	int err = 0;
+
+	DRM_DEBUG("\n");
+
+	amdgpu_ucode_ip_version_decode(adev, MP0_HWIP, ucode_prefix, sizeof(ucode_prefix));
+
+	switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) {
+	case IP_VERSION(11, 0, 2):
+	case IP_VERSION(11, 0, 4):
+		err = psp_init_sos_microcode(psp, ucode_prefix);
+		if (err)
+			return err;
+		err = psp_init_asd_microcode(psp, ucode_prefix);
+		if (err)
+			return err;
+		err = psp_init_ta_microcode(psp, ucode_prefix);
+		adev->psp.securedisplay_context.context.bin_desc.size_bytes = 0;
+		break;
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 5):
+	case IP_VERSION(11, 0, 9):
+		err = psp_init_sos_microcode(psp, ucode_prefix);
+		if (err)
+			return err;
+		err = psp_init_asd_microcode(psp, ucode_prefix);
+		if (err)
+			return err;
+		err = psp_init_ta_microcode(psp, ucode_prefix);
+		adev->psp.securedisplay_context.context.bin_desc.size_bytes = 0;
+		break;
+	case IP_VERSION(11, 0, 7):
+	case IP_VERSION(11, 0, 11):
+	case IP_VERSION(11, 0, 12):
+	case IP_VERSION(11, 0, 13):
+		err = psp_init_sos_microcode(psp, ucode_prefix);
+		if (err)
+			return err;
+		err = psp_init_ta_microcode(psp, ucode_prefix);
+		break;
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 2):
+		err = psp_init_asd_microcode(psp, ucode_prefix);
+		if (err)
+			return err;
+		err = psp_init_toc_microcode(psp, ucode_prefix);
+		break;
+	default:
+		BUG();
+	}
+
+	return err;
+}
+
+static int psp_v11_0_wait_for_bootloader(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+
+	int ret;
+	int retry_loop;
+
+	for (retry_loop = 0; retry_loop < 20; retry_loop++) {
+		/* Wait for bootloader to signify that is
+		    ready having bit 31 of C2PMSG_35 set to 1 */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
+			0x80000000, 0x8000FFFF, PSP_WAITREG_NOVERBOSE);
+
+		if (ret == 0)
+			return 0;
+	}
+
+	return ret;
+}
+
+static bool psp_v11_0_is_sos_alive(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+	uint32_t sol_reg;
+
+	sol_reg = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_81);
+
+	return sol_reg != 0x0;
+}
+
+static int psp_v11_0_bootloader_load_component(struct psp_context  	*psp,
+					       struct psp_bin_desc 	*bin_desc,
+					       enum psp_bootloader_cmd  bl_cmd)
+{
+	int ret;
+	uint32_t psp_gfxdrv_command_reg = 0;
+	struct amdgpu_device *adev = psp->adev;
+
+	/* Check sOS sign of life register to confirm sys driver and sOS
+	 * are already been loaded.
+	 */
+	if (psp_v11_0_is_sos_alive(psp))
+		return 0;
+
+	ret = psp_v11_0_wait_for_bootloader(psp);
+	if (ret)
+		return ret;
+
+	/* Copy PSP System Driver binary to memory */
+	psp_copy_fw(psp, bin_desc->start_addr, bin_desc->size_bytes);
+
+	/* Provide the sys driver to bootloader */
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36,
+	       (uint32_t)(psp->fw_pri_mc_addr >> 20));
+	psp_gfxdrv_command_reg = bl_cmd;
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_35,
+	       psp_gfxdrv_command_reg);
+
+	ret = psp_v11_0_wait_for_bootloader(psp);
+
+	return ret;
+}
+
+static int psp_v11_0_bootloader_load_kdb(struct psp_context *psp)
+{
+	return psp_v11_0_bootloader_load_component(psp, &psp->kdb, PSP_BL__LOAD_KEY_DATABASE);
+}
+
+static int psp_v11_0_bootloader_load_spl(struct psp_context *psp)
+{
+	return psp_v11_0_bootloader_load_component(psp, &psp->spl, PSP_BL__LOAD_TOS_SPL_TABLE);
+}
+
+static int psp_v11_0_bootloader_load_sysdrv(struct psp_context *psp)
+{
+	return psp_v11_0_bootloader_load_component(psp, &psp->sys, PSP_BL__LOAD_SYSDRV);
+}
+
+static int psp_v11_0_bootloader_load_sos(struct psp_context *psp)
+{
+	int ret;
+	unsigned int psp_gfxdrv_command_reg = 0;
+	struct amdgpu_device *adev = psp->adev;
+
+	/* Check sOS sign of life register to confirm sys driver and sOS
+	 * are already been loaded.
+	 */
+	if (psp_v11_0_is_sos_alive(psp))
+		return 0;
+
+	ret = psp_v11_0_wait_for_bootloader(psp);
+	if (ret)
+		return ret;
+
+	/* Copy Secure OS binary to PSP memory */
+	psp_copy_fw(psp, psp->sos.start_addr, psp->sos.size_bytes);
+
+	/* Provide the PSP secure OS to bootloader */
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36,
+	       (uint32_t)(psp->fw_pri_mc_addr >> 20));
+	psp_gfxdrv_command_reg = PSP_BL__LOAD_SOSDRV;
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_35,
+	       psp_gfxdrv_command_reg);
+
+	/* there might be handshake issue with hardware which needs delay */
+	mdelay(20);
+	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_81),
+			   RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_81), 0,
+			   PSP_WAITREG_CHANGED);
+
+	return ret;
+}
+
+static int psp_v11_0_ring_stop(struct psp_context *psp,
+			      enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	struct amdgpu_device *adev = psp->adev;
+
+	/* Write the ring destroy command*/
+	if (amdgpu_sriov_vf(adev))
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_101,
+				     GFX_CTRL_CMD_ID_DESTROY_GPCOM_RING);
+	else
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64,
+				     GFX_CTRL_CMD_ID_DESTROY_RINGS);
+
+	/* there might be handshake issue with hardware which needs delay */
+	mdelay(20);
+
+	/* Wait for response flag (bit 31) */
+	if (amdgpu_sriov_vf(adev))
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_101),
+			MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+	else
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
+			MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+
+	return ret;
+}
+
+static int psp_v11_0_ring_create(struct psp_context *psp,
+				enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	unsigned int psp_ring_reg = 0;
+	struct psp_ring *ring = &psp->km_ring;
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_sriov_vf(adev)) {
+		ring->ring_wptr = 0;
+		ret = psp_v11_0_ring_stop(psp, ring_type);
+		if (ret) {
+			DRM_ERROR("psp_v11_0_ring_stop_sriov failed!\n");
+			return ret;
+		}
+
+		/* Write low address of the ring to C2PMSG_102 */
+		psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_102, psp_ring_reg);
+		/* Write high address of the ring to C2PMSG_103 */
+		psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_103, psp_ring_reg);
+
+		/* Write the ring initialization command to C2PMSG_101 */
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_101,
+					     GFX_CTRL_CMD_ID_INIT_GPCOM_RING);
+
+		/* there might be handshake issue with hardware which needs delay */
+		mdelay(20);
+
+		/* Wait for response flag (bit 31) in C2PMSG_101 */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_101),
+			MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+
+	} else {
+		/* Wait for sOS ready for ring creation */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
+			MBOX_TOS_READY_FLAG, MBOX_TOS_READY_MASK, 0);
+		if (ret) {
+			DRM_ERROR("Failed to wait for sOS ready for ring creation\n");
+			return ret;
+		}
+
+		/* Write low address of the ring to C2PMSG_69 */
+		psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_69, psp_ring_reg);
+		/* Write high address of the ring to C2PMSG_70 */
+		psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_70, psp_ring_reg);
+		/* Write size of ring to C2PMSG_71 */
+		psp_ring_reg = ring->ring_size;
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_71, psp_ring_reg);
+		/* Write the ring initialization command to C2PMSG_64 */
+		psp_ring_reg = ring_type;
+		psp_ring_reg = psp_ring_reg << 16;
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64, psp_ring_reg);
+
+		/* there might be handshake issue with hardware which needs delay */
+		mdelay(20);
+
+		/* Wait for response flag (bit 31) in C2PMSG_64 */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
+			MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+	}
+
+	return ret;
+}
+
+
+static int psp_v11_0_ring_destroy(struct psp_context *psp,
+				 enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	struct psp_ring *ring = &psp->km_ring;
+	struct amdgpu_device *adev = psp->adev;
+
+	ret = psp_v11_0_ring_stop(psp, ring_type);
+	if (ret)
+		DRM_ERROR("Fail to stop psp ring\n");
+
+	amdgpu_bo_free_kernel(&adev->firmware.rbuf,
+			      &ring->ring_mem_mc_addr,
+			      (void **)&ring->ring_mem);
+
+	return ret;
+}
+
+static int psp_v11_0_mode1_reset(struct psp_context *psp)
+{
+	int ret;
+	uint32_t offset;
+	struct amdgpu_device *adev = psp->adev;
+
+	offset = SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64);
+
+	ret = psp_wait_for(psp, offset, MBOX_TOS_READY_FLAG,
+			   MBOX_TOS_READY_MASK, 0);
+
+	if (ret) {
+		DRM_INFO("psp is not working correctly before mode1 reset!\n");
+		return -EINVAL;
+	}
+
+	/*send the mode 1 reset command*/
+	WREG32(offset, GFX_CTRL_CMD_ID_MODE1_RST);
+
+	msleep(500);
+
+	return 0;
+}
+
+static int psp_v11_0_memory_training_send_msg(struct psp_context *psp, int msg)
+{
+	int ret;
+	int i;
+	uint32_t data_32;
+	int max_wait;
+	struct amdgpu_device *adev = psp->adev;
+
+	data_32 = (psp->mem_train_ctx.c2p_train_data_offset >> 20);
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36, data_32);
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_35, msg);
+
+	max_wait = MEM_TRAIN_SEND_MSG_TIMEOUT_US / adev->usec_timeout;
+	for (i = 0; i < max_wait; i++) {
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
+			0x80000000, 0x80000000, PSP_WAITREG_NOVERBOSE);
+		if (ret == 0)
+			break;
+	}
+	if (i < max_wait)
+		ret = 0;
+	else
+		ret = -ETIME;
+
+	DRM_DEBUG("training %s %s, cost %d @ %d ms\n",
+		  (msg == PSP_BL__DRAM_SHORT_TRAIN) ? "short" : "long",
+		  (ret == 0) ? "succeed" : "failed",
+		  i, adev->usec_timeout/1000);
+	return ret;
+}
+
+/*
+ * save and restore process
+ */
+static int psp_v11_0_memory_training(struct psp_context *psp, uint32_t ops)
+{
+	struct psp_memory_training_context *ctx = &psp->mem_train_ctx;
+	uint32_t *pcache = (uint32_t *)ctx->sys_cache;
+	struct amdgpu_device *adev = psp->adev;
+	uint32_t p2c_header[4];
+	uint32_t sz;
+	void *buf;
+	int ret, idx;
+
+	if (ctx->init == PSP_MEM_TRAIN_NOT_SUPPORT) {
+		DRM_DEBUG("Memory training is not supported.\n");
+		return 0;
+	} else if (ctx->init != PSP_MEM_TRAIN_INIT_SUCCESS) {
+		DRM_ERROR("Memory training initialization failure.\n");
+		return -EINVAL;
+	}
+
+	if (psp_v11_0_is_sos_alive(psp)) {
+		DRM_DEBUG("SOS is alive, skip memory training.\n");
+		return 0;
+	}
+
+	amdgpu_device_vram_access(adev, ctx->p2c_train_data_offset, p2c_header, sizeof(p2c_header), false);
+	DRM_DEBUG("sys_cache[%08x,%08x,%08x,%08x] p2c_header[%08x,%08x,%08x,%08x]\n",
+		  pcache[0], pcache[1], pcache[2], pcache[3],
+		  p2c_header[0], p2c_header[1], p2c_header[2], p2c_header[3]);
+
+	if (ops & PSP_MEM_TRAIN_SEND_SHORT_MSG) {
+		DRM_DEBUG("Short training depends on restore.\n");
+		ops |= PSP_MEM_TRAIN_RESTORE;
+	}
+
+	if ((ops & PSP_MEM_TRAIN_RESTORE) &&
+	    pcache[0] != MEM_TRAIN_SYSTEM_SIGNATURE) {
+		DRM_DEBUG("sys_cache[0] is invalid, restore depends on save.\n");
+		ops |= PSP_MEM_TRAIN_SAVE;
+	}
+
+	if (p2c_header[0] == MEM_TRAIN_SYSTEM_SIGNATURE &&
+	    !(pcache[0] == MEM_TRAIN_SYSTEM_SIGNATURE &&
+	      pcache[3] == p2c_header[3])) {
+		DRM_DEBUG("sys_cache is invalid or out-of-date, need save training data to sys_cache.\n");
+		ops |= PSP_MEM_TRAIN_SAVE;
+	}
+
+	if ((ops & PSP_MEM_TRAIN_SAVE) &&
+	    p2c_header[0] != MEM_TRAIN_SYSTEM_SIGNATURE) {
+		DRM_DEBUG("p2c_header[0] is invalid, save depends on long training.\n");
+		ops |= PSP_MEM_TRAIN_SEND_LONG_MSG;
+	}
+
+	if (ops & PSP_MEM_TRAIN_SEND_LONG_MSG) {
+		ops &= ~PSP_MEM_TRAIN_SEND_SHORT_MSG;
+		ops |= PSP_MEM_TRAIN_SAVE;
+	}
+
+	DRM_DEBUG("Memory training ops:%x.\n", ops);
+
+	if (ops & PSP_MEM_TRAIN_SEND_LONG_MSG) {
+		/*
+		 * Long training will encroach a certain amount on the bottom of VRAM;
+		 * save the content from the bottom of VRAM to system memory
+		 * before training, and restore it after training to avoid
+		 * VRAM corruption.
+		 */
+		sz = BIST_MEM_TRAINING_ENCROACHED_SIZE;
+
+		if (adev->gmc.visible_vram_size < sz || !adev->mman.aper_base_kaddr) {
+			DRM_ERROR("visible_vram_size %llx or aper_base_kaddr %p is not initialized.\n",
+				  adev->gmc.visible_vram_size,
+				  adev->mman.aper_base_kaddr);
+			return -EINVAL;
+		}
+
+		buf = vmalloc(sz);
+		if (!buf) {
+			DRM_ERROR("failed to allocate system memory.\n");
+			return -ENOMEM;
+		}
+
+		if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+			memcpy_fromio(buf, adev->mman.aper_base_kaddr, sz);
+			ret = psp_v11_0_memory_training_send_msg(psp, PSP_BL__DRAM_LONG_TRAIN);
+			if (ret) {
+				DRM_ERROR("Send long training msg failed.\n");
+				vfree(buf);
+				drm_dev_exit(idx);
+				return ret;
+			}
+
+			memcpy_toio(adev->mman.aper_base_kaddr, buf, sz);
+			amdgpu_device_flush_hdp(adev, NULL);
+			vfree(buf);
+			drm_dev_exit(idx);
+		} else {
+			vfree(buf);
+			return -ENODEV;
+		}
+	}
+
+	if (ops & PSP_MEM_TRAIN_SAVE) {
+		amdgpu_device_vram_access(psp->adev, ctx->p2c_train_data_offset, ctx->sys_cache, ctx->train_data_size, false);
+	}
+
+	if (ops & PSP_MEM_TRAIN_RESTORE) {
+		amdgpu_device_vram_access(psp->adev, ctx->c2p_train_data_offset, ctx->sys_cache, ctx->train_data_size, true);
+	}
+
+	if (ops & PSP_MEM_TRAIN_SEND_SHORT_MSG) {
+		ret = psp_v11_0_memory_training_send_msg(psp, (amdgpu_force_long_training > 0) ?
+							 PSP_BL__DRAM_LONG_TRAIN : PSP_BL__DRAM_SHORT_TRAIN);
+		if (ret) {
+			DRM_ERROR("send training msg failed.\n");
+			return ret;
+		}
+	}
+	ctx->training_cnt++;
+	return 0;
+}
+
+static uint32_t psp_v11_0_ring_get_wptr(struct psp_context *psp)
+{
+	uint32_t data;
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_sriov_vf(adev))
+		data = psp->km_ring.ring_wptr;
+	else
+		data = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67);
+
+	return data;
+}
+
+static void psp_v11_0_ring_set_wptr(struct psp_context *psp, uint32_t value)
+{
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_sriov_vf(adev)) {
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_102, value);
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_101, GFX_CTRL_CMD_ID_CONSUME_CMD);
+		psp->km_ring.ring_wptr = value;
+	} else
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67, value);
+}
+
+static int psp_v11_0_load_usbc_pd_fw(struct psp_context *psp, uint64_t fw_pri_mc_addr)
+{
+	struct amdgpu_device *adev = psp->adev;
+	uint32_t reg_status;
+	int ret, i = 0;
+
+	/*
+	 * LFB address which is aligned to 1MB address and has to be
+	 * right-shifted by 20 so that LFB address can be passed on a 32-bit C2P
+	 * register
+	 */
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36, (fw_pri_mc_addr >> 20));
+
+	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
+			   0x80000000, 0x80000000, 0);
+	if (ret)
+		return ret;
+
+	/* Fireup interrupt so PSP can pick up the address */
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_35, (GFX_CMD_USB_PD_USE_LFB << 16));
+
+	/* FW load takes very long time */
+	do {
+		msleep(1000);
+		reg_status = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_35);
+
+		if (reg_status & 0x80000000)
+			goto done;
+
+	} while (++i < USBC_PD_POLLING_LIMIT_S);
+
+	return -ETIME;
+done:
+
+	if ((reg_status & 0xFFFF) != 0) {
+		DRM_ERROR("Address load failed - MP0_SMN_C2PMSG_35.Bits [15:0] = 0x%04x\n",
+				reg_status & 0xFFFF);
+		return -EIO;
+	}
+
+	return 0;
+}
+
+static int psp_v11_0_read_usbc_pd_fw(struct psp_context *psp, uint32_t *fw_ver)
+{
+	struct amdgpu_device *adev = psp->adev;
+	int ret;
+
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_35, C2PMSG_CMD_GFX_USB_PD_FW_VER);
+
+	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
+			   0x80000000, 0x80000000, 0);
+	if (!ret)
+		*fw_ver = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36);
+
+	return ret;
+}
+
+static const struct psp_funcs psp_v11_0_funcs = {
+	.init_microcode = psp_v11_0_init_microcode,
+	.bootloader_load_kdb = psp_v11_0_bootloader_load_kdb,
+	.bootloader_load_spl = psp_v11_0_bootloader_load_spl,
+	.bootloader_load_sysdrv = psp_v11_0_bootloader_load_sysdrv,
+	.bootloader_load_sos = psp_v11_0_bootloader_load_sos,
+	.ring_create = psp_v11_0_ring_create,
+	.ring_stop = psp_v11_0_ring_stop,
+	.ring_destroy = psp_v11_0_ring_destroy,
+	.mode1_reset = psp_v11_0_mode1_reset,
+	.mem_training = psp_v11_0_memory_training,
+	.ring_get_wptr = psp_v11_0_ring_get_wptr,
+	.ring_set_wptr = psp_v11_0_ring_set_wptr,
+	.load_usbc_pd_fw = psp_v11_0_load_usbc_pd_fw,
+	.read_usbc_pd_fw = psp_v11_0_read_usbc_pd_fw,
+	.wait_for_bootloader = psp_v11_0_wait_for_bootloader
+};
+
+void psp_v11_0_set_psp_funcs(struct psp_context *psp)
+{
+	psp->funcs = &psp_v11_0_funcs;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.h b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.h
new file mode 100644
index 000000000000..082c16c887bf
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __PSP_V11_0_H__
+#define __PSP_V11_0_H__
+
+#include "amdgpu_psp.h"
+
+void psp_v11_0_set_psp_funcs(struct psp_context *psp);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v11_0_8.c b/drivers/gpu/drm/amd/amdgpu/psp_v11_0_8.c
new file mode 100644
index 000000000000..93787a90d598
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v11_0_8.c
@@ -0,0 +1,186 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "amdgpu_psp.h"
+#include "amdgpu_ucode.h"
+#include "soc15_common.h"
+#include "psp_v11_0_8.h"
+
+#include "mp/mp_11_0_8_offset.h"
+
+static int psp_v11_0_8_ring_stop(struct psp_context *psp,
+			       enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_sriov_vf(adev)) {
+		/* Write the ring destroy command*/
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_101,
+			     GFX_CTRL_CMD_ID_DESTROY_GPCOM_RING);
+		/* there might be handshake issue with hardware which needs delay */
+		mdelay(20);
+		/* Wait for response flag (bit 31) */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_101),
+			MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+	} else {
+		/* Write the ring destroy command*/
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64,
+			     GFX_CTRL_CMD_ID_DESTROY_RINGS);
+		/* there might be handshake issue with hardware which needs delay */
+		mdelay(20);
+		/* Wait for response flag (bit 31) */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
+			MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+	}
+
+	return ret;
+}
+
+static int psp_v11_0_8_ring_create(struct psp_context *psp,
+				 enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	unsigned int psp_ring_reg = 0;
+	struct psp_ring *ring = &psp->km_ring;
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_sriov_vf(adev)) {
+		ret = psp_v11_0_8_ring_stop(psp, ring_type);
+		if (ret) {
+			DRM_ERROR("psp_v11_0_8_ring_stop_sriov failed!\n");
+			return ret;
+		}
+
+		/* Write low address of the ring to C2PMSG_102 */
+		psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_102, psp_ring_reg);
+		/* Write high address of the ring to C2PMSG_103 */
+		psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_103, psp_ring_reg);
+
+		/* Write the ring initialization command to C2PMSG_101 */
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_101,
+			     GFX_CTRL_CMD_ID_INIT_GPCOM_RING);
+
+		/* there might be handshake issue with hardware which needs delay */
+		mdelay(20);
+
+		/* Wait for response flag (bit 31) in C2PMSG_101 */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_101),
+			MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+
+	} else {
+		/* Wait for sOS ready for ring creation */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
+			MBOX_TOS_READY_FLAG, MBOX_TOS_READY_MASK, 0);
+		if (ret) {
+			DRM_ERROR("Failed to wait for trust OS ready for ring creation\n");
+			return ret;
+		}
+
+		/* Write low address of the ring to C2PMSG_69 */
+		psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_69, psp_ring_reg);
+		/* Write high address of the ring to C2PMSG_70 */
+		psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_70, psp_ring_reg);
+		/* Write size of ring to C2PMSG_71 */
+		psp_ring_reg = ring->ring_size;
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_71, psp_ring_reg);
+		/* Write the ring initialization command to C2PMSG_64 */
+		psp_ring_reg = ring_type;
+		psp_ring_reg = psp_ring_reg << 16;
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64, psp_ring_reg);
+
+		/* there might be handshake issue with hardware which needs delay */
+		mdelay(20);
+
+		/* Wait for response flag (bit 31) in C2PMSG_64 */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
+			MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+	}
+
+	return ret;
+}
+
+static int psp_v11_0_8_ring_destroy(struct psp_context *psp,
+				  enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	struct psp_ring *ring = &psp->km_ring;
+	struct amdgpu_device *adev = psp->adev;
+
+	ret = psp_v11_0_8_ring_stop(psp, ring_type);
+	if (ret)
+		DRM_ERROR("Fail to stop psp ring\n");
+
+	amdgpu_bo_free_kernel(&adev->firmware.rbuf,
+			      &ring->ring_mem_mc_addr,
+			      (void **)&ring->ring_mem);
+
+	return ret;
+}
+
+static uint32_t psp_v11_0_8_ring_get_wptr(struct psp_context *psp)
+{
+	uint32_t data;
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_sriov_vf(adev))
+		data = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_102);
+	else
+		data = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67);
+
+	return data;
+}
+
+static void psp_v11_0_8_ring_set_wptr(struct psp_context *psp, uint32_t value)
+{
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_sriov_vf(adev)) {
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_102, value);
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_101,
+			     GFX_CTRL_CMD_ID_CONSUME_CMD);
+	} else
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67, value);
+}
+
+static const struct psp_funcs psp_v11_0_8_funcs = {
+	.ring_create = psp_v11_0_8_ring_create,
+	.ring_stop = psp_v11_0_8_ring_stop,
+	.ring_destroy = psp_v11_0_8_ring_destroy,
+	.ring_get_wptr = psp_v11_0_8_ring_get_wptr,
+	.ring_set_wptr = psp_v11_0_8_ring_set_wptr,
+};
+
+void psp_v11_0_8_set_psp_funcs(struct psp_context *psp)
+{
+	psp->funcs = &psp_v11_0_8_funcs;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v11_0_8.h b/drivers/gpu/drm/amd/amdgpu/psp_v11_0_8.h
new file mode 100644
index 000000000000..890377a5afe0
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v11_0_8.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __PSP_V11_0_8_H__
+#define __PSP_V11_0_8_H__
+
+#include "amdgpu_psp.h"
+
+void psp_v11_0_8_set_psp_funcs(struct psp_context *psp);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v12_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v12_0.c
new file mode 100644
index 000000000000..4c6450d62299
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v12_0.c
@@ -0,0 +1,291 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include "amdgpu.h"
+#include "amdgpu_psp.h"
+#include "amdgpu_ucode.h"
+#include "soc15_common.h"
+#include "psp_v12_0.h"
+
+#include "mp/mp_12_0_0_offset.h"
+#include "mp/mp_12_0_0_sh_mask.h"
+#include "gc/gc_9_0_offset.h"
+#include "sdma0/sdma0_4_0_offset.h"
+#include "nbio/nbio_7_4_offset.h"
+
+MODULE_FIRMWARE("amdgpu/renoir_asd.bin");
+MODULE_FIRMWARE("amdgpu/renoir_ta.bin");
+MODULE_FIRMWARE("amdgpu/green_sardine_asd.bin");
+MODULE_FIRMWARE("amdgpu/green_sardine_ta.bin");
+
+/* address block */
+#define smnMP1_FIRMWARE_FLAGS		0x3010024
+
+static int psp_v12_0_init_microcode(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+	char ucode_prefix[30];
+	int err = 0;
+	DRM_DEBUG("\n");
+
+	amdgpu_ucode_ip_version_decode(adev, MP0_HWIP, ucode_prefix, sizeof(ucode_prefix));
+
+	err = psp_init_asd_microcode(psp, ucode_prefix);
+	if (err)
+		return err;
+
+	err = psp_init_ta_microcode(psp, ucode_prefix);
+	if (err)
+		return err;
+
+	/* only supported on renoir */
+	if (!(adev->apu_flags & AMD_APU_IS_RENOIR))
+		adev->psp.securedisplay_context.context.bin_desc.size_bytes = 0;
+
+	return 0;
+}
+
+static int psp_v12_0_bootloader_load_sysdrv(struct psp_context *psp)
+{
+	int ret;
+	uint32_t psp_gfxdrv_command_reg = 0;
+	struct amdgpu_device *adev = psp->adev;
+	uint32_t sol_reg;
+
+	/* Check sOS sign of life register to confirm sys driver and sOS
+	 * are already been loaded.
+	 */
+	sol_reg = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_81);
+	if (sol_reg)
+		return 0;
+
+	/* Wait for bootloader to signify that is ready having bit 31 of C2PMSG_35 set to 1 */
+	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
+			   0x80000000, 0x80000000, 0);
+	if (ret)
+		return ret;
+
+	/* Copy PSP System Driver binary to memory */
+	psp_copy_fw(psp, psp->sys.start_addr, psp->sys.size_bytes);
+
+	/* Provide the sys driver to bootloader */
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36,
+	       (uint32_t)(psp->fw_pri_mc_addr >> 20));
+	psp_gfxdrv_command_reg = 1 << 16;
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_35,
+	       psp_gfxdrv_command_reg);
+
+	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
+			   0x80000000, 0x80000000, 0);
+
+	return ret;
+}
+
+static int psp_v12_0_bootloader_load_sos(struct psp_context *psp)
+{
+	int ret;
+	unsigned int psp_gfxdrv_command_reg = 0;
+	struct amdgpu_device *adev = psp->adev;
+	uint32_t sol_reg;
+
+	/* Check sOS sign of life register to confirm sys driver and sOS
+	 * are already been loaded.
+	 */
+	sol_reg = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_81);
+	if (sol_reg)
+		return 0;
+
+	/* Wait for bootloader to signify that is ready having bit 31 of C2PMSG_35 set to 1 */
+	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
+			   0x80000000, 0x80000000, 0);
+	if (ret)
+		return ret;
+
+	/* Copy Secure OS binary to PSP memory */
+	psp_copy_fw(psp, psp->sos.start_addr, psp->sos.size_bytes);
+
+	/* Provide the PSP secure OS to bootloader */
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36,
+	       (uint32_t)(psp->fw_pri_mc_addr >> 20));
+	psp_gfxdrv_command_reg = 2 << 16;
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_35,
+	       psp_gfxdrv_command_reg);
+
+	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_81),
+			   RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_81), 0,
+			   PSP_WAITREG_CHANGED);
+
+	return ret;
+}
+
+static int psp_v12_0_ring_create(struct psp_context *psp,
+				enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	unsigned int psp_ring_reg = 0;
+	struct psp_ring *ring = &psp->km_ring;
+	struct amdgpu_device *adev = psp->adev;
+
+	/* Write low address of the ring to C2PMSG_69 */
+	psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr);
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_69, psp_ring_reg);
+	/* Write high address of the ring to C2PMSG_70 */
+	psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr);
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_70, psp_ring_reg);
+	/* Write size of ring to C2PMSG_71 */
+	psp_ring_reg = ring->ring_size;
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_71, psp_ring_reg);
+	/* Write the ring initialization command to C2PMSG_64 */
+	psp_ring_reg = ring_type;
+	psp_ring_reg = psp_ring_reg << 16;
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64, psp_ring_reg);
+
+	/* Wait for response flag (bit 31) in C2PMSG_64 */
+	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
+			   MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+
+	return ret;
+}
+
+static int psp_v12_0_ring_stop(struct psp_context *psp,
+			      enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	struct amdgpu_device *adev = psp->adev;
+
+	/* Write the ring destroy command*/
+	if (amdgpu_sriov_vf(adev))
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_101,
+				     GFX_CTRL_CMD_ID_DESTROY_GPCOM_RING);
+	else
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64,
+				     GFX_CTRL_CMD_ID_DESTROY_RINGS);
+
+	/* Wait for response flag (bit 31) */
+	if (amdgpu_sriov_vf(adev))
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_101),
+			MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+	else
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
+			MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+
+	return ret;
+}
+
+static int psp_v12_0_ring_destroy(struct psp_context *psp,
+				 enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	struct psp_ring *ring = &psp->km_ring;
+	struct amdgpu_device *adev = psp->adev;
+
+	ret = psp_v12_0_ring_stop(psp, ring_type);
+	if (ret)
+		DRM_ERROR("Fail to stop psp ring\n");
+
+	amdgpu_bo_free_kernel(&adev->firmware.rbuf,
+			      &ring->ring_mem_mc_addr,
+			      (void **)&ring->ring_mem);
+
+	return ret;
+}
+
+static int psp_v12_0_mode1_reset(struct psp_context *psp)
+{
+	int ret;
+	uint32_t offset;
+	struct amdgpu_device *adev = psp->adev;
+
+	offset = SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64);
+
+	ret = psp_wait_for(psp, offset, MBOX_TOS_READY_FLAG,
+			   MBOX_TOS_READY_MASK, 0);
+
+	if (ret) {
+		DRM_INFO("psp is not working correctly before mode1 reset!\n");
+		return -EINVAL;
+	}
+
+	/*send the mode 1 reset command*/
+	WREG32(offset, GFX_CTRL_CMD_ID_MODE1_RST);
+
+	msleep(500);
+
+	offset = SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_33);
+
+	ret = psp_wait_for(psp, offset, MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK,
+			   0);
+
+	if (ret) {
+		DRM_INFO("psp mode 1 reset failed!\n");
+		return -EINVAL;
+	}
+
+	DRM_INFO("psp mode1 reset succeed \n");
+
+	return 0;
+}
+
+static uint32_t psp_v12_0_ring_get_wptr(struct psp_context *psp)
+{
+	uint32_t data;
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_sriov_vf(adev))
+		data = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_102);
+	else
+		data = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67);
+
+	return data;
+}
+
+static void psp_v12_0_ring_set_wptr(struct psp_context *psp, uint32_t value)
+{
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_sriov_vf(adev)) {
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_102, value);
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_101, GFX_CTRL_CMD_ID_CONSUME_CMD);
+	} else
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67, value);
+}
+
+static const struct psp_funcs psp_v12_0_funcs = {
+	.init_microcode = psp_v12_0_init_microcode,
+	.bootloader_load_sysdrv = psp_v12_0_bootloader_load_sysdrv,
+	.bootloader_load_sos = psp_v12_0_bootloader_load_sos,
+	.ring_create = psp_v12_0_ring_create,
+	.ring_stop = psp_v12_0_ring_stop,
+	.ring_destroy = psp_v12_0_ring_destroy,
+	.mode1_reset = psp_v12_0_mode1_reset,
+	.ring_get_wptr = psp_v12_0_ring_get_wptr,
+	.ring_set_wptr = psp_v12_0_ring_set_wptr,
+};
+
+void psp_v12_0_set_psp_funcs(struct psp_context *psp)
+{
+	psp->funcs = &psp_v12_0_funcs;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v12_0.h b/drivers/gpu/drm/amd/amdgpu/psp_v12_0.h
new file mode 100644
index 000000000000..241693ab1990
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v12_0.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __PSP_V12_0_H__
+#define __PSP_V12_0_H__
+
+#include "amdgpu_psp.h"
+
+void psp_v12_0_set_psp_funcs(struct psp_context *psp);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c
new file mode 100644
index 000000000000..af4a7d7c4abd
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c
@@ -0,0 +1,977 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <drm/drm_drv.h>
+#include <linux/vmalloc.h>
+#include "amdgpu.h"
+#include "amdgpu_psp.h"
+#include "amdgpu_ucode.h"
+#include "soc15_common.h"
+#include "psp_v13_0.h"
+#include "amdgpu_ras.h"
+
+#include "mp/mp_13_0_2_offset.h"
+#include "mp/mp_13_0_2_sh_mask.h"
+
+MODULE_FIRMWARE("amdgpu/aldebaran_sos.bin");
+MODULE_FIRMWARE("amdgpu/aldebaran_ta.bin");
+MODULE_FIRMWARE("amdgpu/aldebaran_cap.bin");
+MODULE_FIRMWARE("amdgpu/yellow_carp_toc.bin");
+MODULE_FIRMWARE("amdgpu/yellow_carp_ta.bin");
+MODULE_FIRMWARE("amdgpu/psp_13_0_5_toc.bin");
+MODULE_FIRMWARE("amdgpu/psp_13_0_5_ta.bin");
+MODULE_FIRMWARE("amdgpu/psp_13_0_8_toc.bin");
+MODULE_FIRMWARE("amdgpu/psp_13_0_8_ta.bin");
+MODULE_FIRMWARE("amdgpu/psp_13_0_0_sos.bin");
+MODULE_FIRMWARE("amdgpu/psp_13_0_0_sos_kicker.bin");
+MODULE_FIRMWARE("amdgpu/psp_13_0_0_ta.bin");
+MODULE_FIRMWARE("amdgpu/psp_13_0_0_ta_kicker.bin");
+MODULE_FIRMWARE("amdgpu/psp_13_0_7_sos.bin");
+MODULE_FIRMWARE("amdgpu/psp_13_0_7_ta.bin");
+MODULE_FIRMWARE("amdgpu/psp_13_0_10_sos.bin");
+MODULE_FIRMWARE("amdgpu/psp_13_0_10_ta.bin");
+MODULE_FIRMWARE("amdgpu/psp_13_0_11_toc.bin");
+MODULE_FIRMWARE("amdgpu/psp_13_0_11_ta.bin");
+MODULE_FIRMWARE("amdgpu/psp_13_0_6_sos.bin");
+MODULE_FIRMWARE("amdgpu/psp_13_0_6_ta.bin");
+MODULE_FIRMWARE("amdgpu/psp_13_0_12_sos.bin");
+MODULE_FIRMWARE("amdgpu/psp_13_0_12_ta.bin");
+MODULE_FIRMWARE("amdgpu/psp_13_0_14_sos.bin");
+MODULE_FIRMWARE("amdgpu/psp_13_0_14_ta.bin");
+MODULE_FIRMWARE("amdgpu/psp_14_0_0_toc.bin");
+MODULE_FIRMWARE("amdgpu/psp_14_0_0_ta.bin");
+MODULE_FIRMWARE("amdgpu/psp_14_0_1_toc.bin");
+MODULE_FIRMWARE("amdgpu/psp_14_0_1_ta.bin");
+MODULE_FIRMWARE("amdgpu/psp_14_0_4_toc.bin");
+MODULE_FIRMWARE("amdgpu/psp_14_0_4_ta.bin");
+
+/* For large FW files the time to complete can be very long */
+#define USBC_PD_POLLING_LIMIT_S 240
+
+/* Read USB-PD from LFB */
+#define GFX_CMD_USB_PD_USE_LFB 0x480
+
+/* Retry times for vmbx ready wait */
+#define PSP_VMBX_POLLING_LIMIT 3000
+
+/* memory training timeout define */
+#define MEM_TRAIN_SEND_MSG_TIMEOUT_US	3000000
+
+#define regMP1_PUB_SCRATCH0	0x3b10090
+
+#define PSP13_BL_STATUS_SIZE 100
+
+static int psp_v13_0_init_microcode(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+	char ucode_prefix[30];
+	int err = 0;
+
+	amdgpu_ucode_ip_version_decode(adev, MP0_HWIP, ucode_prefix, sizeof(ucode_prefix));
+
+	switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) {
+	case IP_VERSION(13, 0, 2):
+		err = psp_init_sos_microcode(psp, ucode_prefix);
+		if (err)
+			return err;
+		/* It's not necessary to load ras ta on Guest side */
+		if (!amdgpu_sriov_vf(adev)) {
+			err = psp_init_ta_microcode(psp, ucode_prefix);
+			if (err)
+				return err;
+		}
+		break;
+	case IP_VERSION(13, 0, 1):
+	case IP_VERSION(13, 0, 3):
+	case IP_VERSION(13, 0, 5):
+	case IP_VERSION(13, 0, 8):
+	case IP_VERSION(13, 0, 11):
+	case IP_VERSION(14, 0, 0):
+	case IP_VERSION(14, 0, 1):
+	case IP_VERSION(14, 0, 4):
+		err = psp_init_toc_microcode(psp, ucode_prefix);
+		if (err)
+			return err;
+		err = psp_init_ta_microcode(psp, ucode_prefix);
+		if (err)
+			return err;
+		break;
+	case IP_VERSION(13, 0, 0):
+	case IP_VERSION(13, 0, 6):
+	case IP_VERSION(13, 0, 7):
+	case IP_VERSION(13, 0, 10):
+	case IP_VERSION(13, 0, 12):
+	case IP_VERSION(13, 0, 14):
+		err = psp_init_sos_microcode(psp, ucode_prefix);
+		if (err)
+			return err;
+		/* It's not necessary to load ras ta on Guest side */
+		err = psp_init_ta_microcode(psp, ucode_prefix);
+		if (err)
+			return err;
+		break;
+	default:
+		BUG();
+	}
+
+	return 0;
+}
+
+static bool psp_v13_0_is_sos_alive(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+	uint32_t sol_reg;
+
+	sol_reg = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_81);
+
+	return sol_reg != 0x0;
+}
+
+static void psp_v13_0_bootloader_print_status(struct psp_context *psp,
+					      const char *msg)
+{
+	struct amdgpu_device *adev = psp->adev;
+	u32 bl_status_reg;
+	char bl_status_msg[PSP13_BL_STATUS_SIZE];
+	int i, at;
+
+	if (amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 6) ||
+	    amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 12) ||
+	    amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 14)) {
+		at = 0;
+		for_each_inst(i, adev->aid_mask) {
+			bl_status_reg =
+				(SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_92)
+				 << 2) +
+				adev->asic_funcs->encode_ext_smn_addressing(i);
+			at += snprintf(bl_status_msg + at,
+				       PSP13_BL_STATUS_SIZE - at,
+				       " status(%02i): 0x%08x", i,
+				       RREG32_PCIE_EXT(bl_status_reg));
+		}
+		dev_info(adev->dev, "%s - %s", msg, bl_status_msg);
+	}
+}
+
+static int psp_v13_0_wait_for_vmbx_ready(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+	int retry_loop, ret;
+
+	for (retry_loop = 0; retry_loop < PSP_VMBX_POLLING_LIMIT; retry_loop++) {
+		/* Wait for bootloader to signify that is
+		   ready having bit 31 of C2PMSG_33 set to 1 */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_33),
+			0x80000000, 0xffffffff, PSP_WAITREG_NOVERBOSE);
+
+		if (ret == 0)
+			break;
+	}
+
+	if (ret)
+		dev_warn(adev->dev, "Bootloader wait timed out");
+
+	return ret;
+}
+
+static int psp_v13_0_wait_for_bootloader(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+	int retry_loop, retry_cnt, ret;
+
+	retry_cnt =
+		((amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 6) ||
+		  amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 12) ||
+		  amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 14))) ?
+			PSP_VMBX_POLLING_LIMIT :
+			10;
+	/* Wait for bootloader to signify that it is ready having bit 31 of
+	 * C2PMSG_35 set to 1. All other bits are expected to be cleared.
+	 * If there is an error in processing command, bits[7:0] will be set.
+	 * This is applicable for PSP v13.0.6 and newer.
+	 */
+	for (retry_loop = 0; retry_loop < retry_cnt; retry_loop++) {
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_35),
+			0x80000000, 0xffffffff, PSP_WAITREG_NOVERBOSE);
+
+		if (ret == 0)
+			return 0;
+		if (retry_loop && !(retry_loop % 10))
+			psp_v13_0_bootloader_print_status(
+				psp, "Waiting for bootloader completion");
+	}
+
+	return ret;
+}
+
+static int psp_v13_0_wait_for_bootloader_steady_state(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+	int ret;
+
+	if (amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 6) ||
+	    amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 12) ||
+	    amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 14)) {
+		ret = psp_v13_0_wait_for_vmbx_ready(psp);
+		if (ret)
+			amdgpu_ras_query_boot_status(adev, 4);
+
+		ret = psp_v13_0_wait_for_bootloader(psp);
+		if (ret)
+			amdgpu_ras_query_boot_status(adev, 4);
+
+		return ret;
+	}
+
+	return 0;
+}
+
+static int psp_v13_0_bootloader_load_component(struct psp_context  	*psp,
+					       struct psp_bin_desc 	*bin_desc,
+					       enum psp_bootloader_cmd  bl_cmd)
+{
+	int ret;
+	uint32_t psp_gfxdrv_command_reg = 0;
+	struct amdgpu_device *adev = psp->adev;
+
+	/* Check tOS sign of life register to confirm sys driver and sOS
+	 * are already been loaded.
+	 */
+	if (psp_v13_0_is_sos_alive(psp))
+		return 0;
+
+	ret = psp_v13_0_wait_for_bootloader(psp);
+	if (ret)
+		return ret;
+
+	memset(psp->fw_pri_buf, 0, PSP_1_MEG);
+
+	/* Copy PSP KDB binary to memory */
+	memcpy(psp->fw_pri_buf, bin_desc->start_addr, bin_desc->size_bytes);
+
+	/* Provide the PSP KDB to bootloader */
+	WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_36,
+	       (uint32_t)(psp->fw_pri_mc_addr >> 20));
+	psp_gfxdrv_command_reg = bl_cmd;
+	WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_35,
+	       psp_gfxdrv_command_reg);
+
+	ret = psp_v13_0_wait_for_bootloader(psp);
+
+	return ret;
+}
+
+static int psp_v13_0_bootloader_load_kdb(struct psp_context *psp)
+{
+	return psp_v13_0_bootloader_load_component(psp, &psp->kdb, PSP_BL__LOAD_KEY_DATABASE);
+}
+
+static int psp_v13_0_bootloader_load_spl(struct psp_context *psp)
+{
+	return psp_v13_0_bootloader_load_component(psp, &psp->kdb, PSP_BL__LOAD_TOS_SPL_TABLE);
+}
+
+static int psp_v13_0_bootloader_load_sysdrv(struct psp_context *psp)
+{
+	return psp_v13_0_bootloader_load_component(psp, &psp->sys, PSP_BL__LOAD_SYSDRV);
+}
+
+static int psp_v13_0_bootloader_load_soc_drv(struct psp_context *psp)
+{
+	return psp_v13_0_bootloader_load_component(psp, &psp->soc_drv, PSP_BL__LOAD_SOCDRV);
+}
+
+static int psp_v13_0_bootloader_load_intf_drv(struct psp_context *psp)
+{
+	return psp_v13_0_bootloader_load_component(psp, &psp->intf_drv, PSP_BL__LOAD_INTFDRV);
+}
+
+static int psp_v13_0_bootloader_load_dbg_drv(struct psp_context *psp)
+{
+	return psp_v13_0_bootloader_load_component(psp, &psp->dbg_drv, PSP_BL__LOAD_DBGDRV);
+}
+
+static int psp_v13_0_bootloader_load_ras_drv(struct psp_context *psp)
+{
+	return psp_v13_0_bootloader_load_component(psp, &psp->ras_drv, PSP_BL__LOAD_RASDRV);
+}
+
+static int psp_v13_0_bootloader_load_spdm_drv(struct psp_context *psp)
+{
+	return psp_v13_0_bootloader_load_component(psp, &psp->spdm_drv, PSP_BL__LOAD_SPDMDRV);
+}
+
+static inline void psp_v13_0_init_sos_version(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+
+	psp->sos.fw_version = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_58);
+}
+
+static int psp_v13_0_bootloader_load_sos(struct psp_context *psp)
+{
+	int ret;
+	unsigned int psp_gfxdrv_command_reg = 0;
+	struct amdgpu_device *adev = psp->adev;
+
+	/* Check sOS sign of life register to confirm sys driver and sOS
+	 * are already been loaded.
+	 */
+	if (psp_v13_0_is_sos_alive(psp)) {
+		psp_v13_0_init_sos_version(psp);
+		return 0;
+	}
+
+	ret = psp_v13_0_wait_for_bootloader(psp);
+	if (ret)
+		return ret;
+
+	memset(psp->fw_pri_buf, 0, PSP_1_MEG);
+
+	/* Copy Secure OS binary to PSP memory */
+	memcpy(psp->fw_pri_buf, psp->sos.start_addr, psp->sos.size_bytes);
+
+	/* Provide the PSP secure OS to bootloader */
+	WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_36,
+	       (uint32_t)(psp->fw_pri_mc_addr >> 20));
+	psp_gfxdrv_command_reg = PSP_BL__LOAD_SOSDRV;
+	WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_35,
+	       psp_gfxdrv_command_reg);
+
+	/* there might be handshake issue with hardware which needs delay */
+	mdelay(20);
+	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_81),
+			   RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_81), 0,
+			   PSP_WAITREG_CHANGED);
+
+	if (!ret)
+		psp_v13_0_init_sos_version(psp);
+
+	return ret;
+}
+
+static int psp_v13_0_ring_stop(struct psp_context *psp,
+			       enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_sriov_vf(adev)) {
+		/* Write the ring destroy command*/
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_101,
+			     GFX_CTRL_CMD_ID_DESTROY_GPCOM_RING);
+		/* there might be handshake issue with hardware which needs delay */
+		mdelay(20);
+		/* Wait for response flag (bit 31) */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_101),
+			MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+	} else {
+		/* Write the ring destroy command*/
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_64,
+			     GFX_CTRL_CMD_ID_DESTROY_RINGS);
+		/* there might be handshake issue with hardware which needs delay */
+		mdelay(20);
+		/* Wait for response flag (bit 31) */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_64),
+			MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+	}
+
+	return ret;
+}
+
+static int psp_v13_0_ring_create(struct psp_context *psp,
+				 enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	unsigned int psp_ring_reg = 0;
+	struct psp_ring *ring = &psp->km_ring;
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_sriov_vf(adev)) {
+		ret = psp_v13_0_ring_stop(psp, ring_type);
+		if (ret) {
+			DRM_ERROR("psp_v13_0_ring_stop_sriov failed!\n");
+			return ret;
+		}
+
+		/* Write low address of the ring to C2PMSG_102 */
+		psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_102, psp_ring_reg);
+		/* Write high address of the ring to C2PMSG_103 */
+		psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_103, psp_ring_reg);
+
+		/* Write the ring initialization command to C2PMSG_101 */
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_101,
+			     GFX_CTRL_CMD_ID_INIT_GPCOM_RING);
+
+		/* there might be handshake issue with hardware which needs delay */
+		mdelay(20);
+
+		/* Wait for response flag (bit 31) in C2PMSG_101 */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_101),
+			MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+
+	} else {
+		/* Wait for sOS ready for ring creation */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_64),
+			MBOX_TOS_READY_FLAG, MBOX_TOS_READY_MASK, 0);
+		if (ret) {
+			DRM_ERROR("Failed to wait for trust OS ready for ring creation\n");
+			return ret;
+		}
+
+		/* Write low address of the ring to C2PMSG_69 */
+		psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_69, psp_ring_reg);
+		/* Write high address of the ring to C2PMSG_70 */
+		psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_70, psp_ring_reg);
+		/* Write size of ring to C2PMSG_71 */
+		psp_ring_reg = ring->ring_size;
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_71, psp_ring_reg);
+		/* Write the ring initialization command to C2PMSG_64 */
+		psp_ring_reg = ring_type;
+		psp_ring_reg = psp_ring_reg << 16;
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_64, psp_ring_reg);
+
+		/* there might be handshake issue with hardware which needs delay */
+		mdelay(20);
+
+		/* Wait for response flag (bit 31) in C2PMSG_64 */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_64),
+			MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+	}
+
+	return ret;
+}
+
+static int psp_v13_0_ring_destroy(struct psp_context *psp,
+				  enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	struct psp_ring *ring = &psp->km_ring;
+	struct amdgpu_device *adev = psp->adev;
+
+	ret = psp_v13_0_ring_stop(psp, ring_type);
+	if (ret)
+		DRM_ERROR("Fail to stop psp ring\n");
+
+	amdgpu_bo_free_kernel(&adev->firmware.rbuf,
+			      &ring->ring_mem_mc_addr,
+			      (void **)&ring->ring_mem);
+
+	return ret;
+}
+
+static uint32_t psp_v13_0_ring_get_wptr(struct psp_context *psp)
+{
+	uint32_t data;
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_sriov_vf(adev))
+		data = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_102);
+	else
+		data = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_67);
+
+	return data;
+}
+
+static void psp_v13_0_ring_set_wptr(struct psp_context *psp, uint32_t value)
+{
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_sriov_vf(adev)) {
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_102, value);
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_101,
+			     GFX_CTRL_CMD_ID_CONSUME_CMD);
+	} else
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_67, value);
+}
+
+static int psp_v13_0_memory_training_send_msg(struct psp_context *psp, int msg)
+{
+	int ret;
+	int i;
+	uint32_t data_32;
+	int max_wait;
+	struct amdgpu_device *adev = psp->adev;
+
+	data_32 = (psp->mem_train_ctx.c2p_train_data_offset >> 20);
+	WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_36, data_32);
+	WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_35, msg);
+
+	max_wait = MEM_TRAIN_SEND_MSG_TIMEOUT_US / adev->usec_timeout;
+	for (i = 0; i < max_wait; i++) {
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_35),
+			0x80000000, 0x80000000, PSP_WAITREG_NOVERBOSE);
+		if (ret == 0)
+			break;
+	}
+	if (i < max_wait)
+		ret = 0;
+	else
+		ret = -ETIME;
+
+	dev_dbg(adev->dev, "training %s %s, cost %d @ %d ms\n",
+		  (msg == PSP_BL__DRAM_SHORT_TRAIN) ? "short" : "long",
+		  (ret == 0) ? "succeed" : "failed",
+		  i, adev->usec_timeout/1000);
+	return ret;
+}
+
+
+static int psp_v13_0_memory_training(struct psp_context *psp, uint32_t ops)
+{
+	struct psp_memory_training_context *ctx = &psp->mem_train_ctx;
+	uint32_t *pcache = (uint32_t *)ctx->sys_cache;
+	struct amdgpu_device *adev = psp->adev;
+	uint32_t p2c_header[4];
+	uint32_t sz;
+	void *buf;
+	int ret, idx;
+
+	if (ctx->init == PSP_MEM_TRAIN_NOT_SUPPORT) {
+		dev_dbg(adev->dev, "Memory training is not supported.\n");
+		return 0;
+	} else if (ctx->init != PSP_MEM_TRAIN_INIT_SUCCESS) {
+		dev_err(adev->dev, "Memory training initialization failure.\n");
+		return -EINVAL;
+	}
+
+	if (psp_v13_0_is_sos_alive(psp)) {
+		dev_dbg(adev->dev, "SOS is alive, skip memory training.\n");
+		return 0;
+	}
+
+	amdgpu_device_vram_access(adev, ctx->p2c_train_data_offset, p2c_header, sizeof(p2c_header), false);
+	dev_dbg(adev->dev, "sys_cache[%08x,%08x,%08x,%08x] p2c_header[%08x,%08x,%08x,%08x]\n",
+		  pcache[0], pcache[1], pcache[2], pcache[3],
+		  p2c_header[0], p2c_header[1], p2c_header[2], p2c_header[3]);
+
+	if (ops & PSP_MEM_TRAIN_SEND_SHORT_MSG) {
+		dev_dbg(adev->dev, "Short training depends on restore.\n");
+		ops |= PSP_MEM_TRAIN_RESTORE;
+	}
+
+	if ((ops & PSP_MEM_TRAIN_RESTORE) &&
+	    pcache[0] != MEM_TRAIN_SYSTEM_SIGNATURE) {
+		dev_dbg(adev->dev, "sys_cache[0] is invalid, restore depends on save.\n");
+		ops |= PSP_MEM_TRAIN_SAVE;
+	}
+
+	if (p2c_header[0] == MEM_TRAIN_SYSTEM_SIGNATURE &&
+	    !(pcache[0] == MEM_TRAIN_SYSTEM_SIGNATURE &&
+	      pcache[3] == p2c_header[3])) {
+		dev_dbg(adev->dev, "sys_cache is invalid or out-of-date, need save training data to sys_cache.\n");
+		ops |= PSP_MEM_TRAIN_SAVE;
+	}
+
+	if ((ops & PSP_MEM_TRAIN_SAVE) &&
+	    p2c_header[0] != MEM_TRAIN_SYSTEM_SIGNATURE) {
+		dev_dbg(adev->dev, "p2c_header[0] is invalid, save depends on long training.\n");
+		ops |= PSP_MEM_TRAIN_SEND_LONG_MSG;
+	}
+
+	if (ops & PSP_MEM_TRAIN_SEND_LONG_MSG) {
+		ops &= ~PSP_MEM_TRAIN_SEND_SHORT_MSG;
+		ops |= PSP_MEM_TRAIN_SAVE;
+	}
+
+	dev_dbg(adev->dev, "Memory training ops:%x.\n", ops);
+
+	if (ops & PSP_MEM_TRAIN_SEND_LONG_MSG) {
+		/*
+		 * Long training will encroach a certain amount on the bottom of VRAM;
+		 * save the content from the bottom of VRAM to system memory
+		 * before training, and restore it after training to avoid
+		 * VRAM corruption.
+		 */
+		sz = BIST_MEM_TRAINING_ENCROACHED_SIZE;
+
+		if (adev->gmc.visible_vram_size < sz || !adev->mman.aper_base_kaddr) {
+			dev_err(adev->dev, "visible_vram_size %llx or aper_base_kaddr %p is not initialized.\n",
+				  adev->gmc.visible_vram_size,
+				  adev->mman.aper_base_kaddr);
+			return -EINVAL;
+		}
+
+		buf = vmalloc(sz);
+		if (!buf) {
+			dev_err(adev->dev, "failed to allocate system memory.\n");
+			return -ENOMEM;
+		}
+
+		if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+			memcpy_fromio(buf, adev->mman.aper_base_kaddr, sz);
+			ret = psp_v13_0_memory_training_send_msg(psp, PSP_BL__DRAM_LONG_TRAIN);
+			if (ret) {
+				DRM_ERROR("Send long training msg failed.\n");
+				vfree(buf);
+				drm_dev_exit(idx);
+				return ret;
+			}
+
+			memcpy_toio(adev->mman.aper_base_kaddr, buf, sz);
+			amdgpu_device_flush_hdp(adev, NULL);
+			vfree(buf);
+			drm_dev_exit(idx);
+		} else {
+			vfree(buf);
+			return -ENODEV;
+		}
+	}
+
+	if (ops & PSP_MEM_TRAIN_SAVE) {
+		amdgpu_device_vram_access(psp->adev, ctx->p2c_train_data_offset, ctx->sys_cache, ctx->train_data_size, false);
+	}
+
+	if (ops & PSP_MEM_TRAIN_RESTORE) {
+		amdgpu_device_vram_access(psp->adev, ctx->c2p_train_data_offset, ctx->sys_cache, ctx->train_data_size, true);
+	}
+
+	if (ops & PSP_MEM_TRAIN_SEND_SHORT_MSG) {
+		ret = psp_v13_0_memory_training_send_msg(psp, (amdgpu_force_long_training > 0) ?
+							 PSP_BL__DRAM_LONG_TRAIN : PSP_BL__DRAM_SHORT_TRAIN);
+		if (ret) {
+			dev_err(adev->dev, "send training msg failed.\n");
+			return ret;
+		}
+	}
+	ctx->training_cnt++;
+	return 0;
+}
+
+static int psp_v13_0_load_usbc_pd_fw(struct psp_context *psp, uint64_t fw_pri_mc_addr)
+{
+	struct amdgpu_device *adev = psp->adev;
+	uint32_t reg_status;
+	int ret, i = 0;
+
+	/*
+	 * LFB address which is aligned to 1MB address and has to be
+	 * right-shifted by 20 so that LFB address can be passed on a 32-bit C2P
+	 * register
+	 */
+	WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_36, (fw_pri_mc_addr >> 20));
+
+	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_35),
+			   0x80000000, 0x80000000, 0);
+	if (ret)
+		return ret;
+
+	/* Fireup interrupt so PSP can pick up the address */
+	WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_35, (GFX_CMD_USB_PD_USE_LFB << 16));
+
+	/* FW load takes very long time */
+	do {
+		msleep(1000);
+		reg_status = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_35);
+
+		if (reg_status & 0x80000000)
+			goto done;
+
+	} while (++i < USBC_PD_POLLING_LIMIT_S);
+
+	return -ETIME;
+done:
+
+	if ((reg_status & 0xFFFF) != 0) {
+		DRM_ERROR("Address load failed - MP0_SMN_C2PMSG_35.Bits [15:0] = %04x\n",
+				reg_status & 0xFFFF);
+		return -EIO;
+	}
+
+	return 0;
+}
+
+static int psp_v13_0_read_usbc_pd_fw(struct psp_context *psp, uint32_t *fw_ver)
+{
+	struct amdgpu_device *adev = psp->adev;
+	int ret;
+
+	WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_35, C2PMSG_CMD_GFX_USB_PD_FW_VER);
+
+	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_35),
+			   0x80000000, 0x80000000, 0);
+	if (!ret)
+		*fw_ver = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_36);
+
+	return ret;
+}
+
+static int psp_v13_0_exec_spi_cmd(struct psp_context *psp, int cmd)
+{
+	uint32_t reg_status = 0, reg_val = 0;
+	struct amdgpu_device *adev = psp->adev;
+	int ret;
+
+	/* clear MBX ready (MBOX_READY_MASK bit is 0) and set update command */
+	reg_val |= (cmd << 16);
+	WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_115,  reg_val);
+
+	/* Ring the doorbell */
+	WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_73, 1);
+
+	if (cmd == C2PMSG_CMD_SPI_UPDATE_FLASH_IMAGE ||
+	    cmd == C2PMSG_CMD_SPI_GET_FLASH_IMAGE)
+		ret = psp_wait_for_spirom_update(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_115),
+						 MBOX_READY_FLAG, MBOX_READY_MASK, PSP_SPIROM_UPDATE_TIMEOUT);
+	else
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_115),
+			MBOX_READY_FLAG, MBOX_READY_MASK, 0);
+	if (ret) {
+		dev_err(adev->dev, "SPI cmd %x timed out, ret = %d", cmd, ret);
+		return ret;
+	}
+
+	reg_status = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_115);
+	if ((reg_status & 0xFFFF) != 0) {
+		dev_err(adev->dev, "SPI cmd %x failed, fail status = %04x\n",
+				cmd, reg_status & 0xFFFF);
+		return -EIO;
+	}
+
+	return 0;
+}
+
+static int psp_v13_0_update_spirom(struct psp_context *psp,
+				   uint64_t fw_pri_mc_addr)
+{
+	struct amdgpu_device *adev = psp->adev;
+	int ret;
+
+	/* Confirm PSP is ready to start */
+	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_115),
+			   MBOX_READY_FLAG, MBOX_READY_MASK, 0);
+	if (ret) {
+		dev_err(adev->dev, "PSP Not ready to start processing, ret = %d", ret);
+		return ret;
+	}
+
+	WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_116, lower_32_bits(fw_pri_mc_addr));
+
+	ret = psp_v13_0_exec_spi_cmd(psp, C2PMSG_CMD_SPI_UPDATE_ROM_IMAGE_ADDR_LO);
+	if (ret)
+		return ret;
+
+	WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_116, upper_32_bits(fw_pri_mc_addr));
+
+	ret = psp_v13_0_exec_spi_cmd(psp, C2PMSG_CMD_SPI_UPDATE_ROM_IMAGE_ADDR_HI);
+	if (ret)
+		return ret;
+
+	psp->vbflash_done = true;
+
+	ret = psp_v13_0_exec_spi_cmd(psp, C2PMSG_CMD_SPI_UPDATE_FLASH_IMAGE);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static int psp_v13_0_dump_spirom(struct psp_context *psp,
+				 uint64_t fw_pri_mc_addr)
+{
+	struct amdgpu_device *adev = psp->adev;
+	int ret;
+
+	/* Confirm PSP is ready to start */
+	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_115),
+			   MBOX_READY_FLAG, MBOX_READY_MASK, 0);
+	if (ret) {
+		dev_err(adev->dev, "PSP Not ready to start processing, ret = %d", ret);
+		return ret;
+	}
+
+	WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_116, lower_32_bits(fw_pri_mc_addr));
+
+	ret = psp_v13_0_exec_spi_cmd(psp, C2PMSG_CMD_SPI_GET_ROM_IMAGE_ADDR_LO);
+	if (ret)
+		return ret;
+
+	WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_116, upper_32_bits(fw_pri_mc_addr));
+
+	ret = psp_v13_0_exec_spi_cmd(psp, C2PMSG_CMD_SPI_GET_ROM_IMAGE_ADDR_HI);
+	if (ret)
+		return ret;
+
+	ret = psp_v13_0_exec_spi_cmd(psp, C2PMSG_CMD_SPI_GET_FLASH_IMAGE);
+
+	return ret;
+}
+
+static int psp_v13_0_vbflash_status(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+
+	return RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_115);
+}
+
+static int psp_v13_0_fatal_error_recovery_quirk(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 10)) {
+		uint32_t  reg_data;
+		/* MP1 fatal error: trigger PSP dram read to unhalt PSP
+		 * during MP1 triggered sync flood.
+		 */
+		reg_data = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_67);
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_67, reg_data + 0x10);
+
+		/* delay 1000ms for the mode1 reset for fatal error
+		 * to be recovered back.
+		 */
+		msleep(1000);
+	}
+
+	return 0;
+}
+
+static bool psp_v13_0_get_ras_capability(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	u32 reg_data;
+
+	/* query ras cap should be done from host side */
+	if (amdgpu_sriov_vf(adev))
+		return false;
+
+	if (!con)
+		return false;
+
+	if ((amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 6) ||
+	     amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 12) ||
+	     amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 14)) &&
+	    (!(adev->flags & AMD_IS_APU))) {
+		reg_data = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_127);
+		adev->ras_hw_enabled = (reg_data & GENMASK_ULL(23, 0));
+		con->poison_supported = ((reg_data & GENMASK_ULL(24, 24)) >> 24) ? true : false;
+		return true;
+	} else {
+		return false;
+	}
+}
+
+static bool psp_v13_0_is_aux_sos_load_required(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+	u32 pmfw_ver;
+
+	if (amdgpu_ip_version(adev, MP0_HWIP, 0) != IP_VERSION(13, 0, 6))
+		return false;
+
+	/* load 4e version of sos if pmfw version less than 85.115.0 */
+	pmfw_ver = RREG32(regMP1_PUB_SCRATCH0 / 4);
+
+	return (pmfw_ver < 0x557300);
+}
+
+static bool psp_v13_0_is_reload_needed(struct psp_context *psp)
+{
+	uint32_t ucode_ver;
+
+	if (!psp_v13_0_is_sos_alive(psp))
+		return false;
+
+	/* Restrict reload support only to specific IP versions */
+	switch (amdgpu_ip_version(psp->adev, MP0_HWIP, 0)) {
+	case IP_VERSION(13, 0, 2):
+	case IP_VERSION(13, 0, 6):
+	case IP_VERSION(13, 0, 14):
+		/* TOS version read from microcode header */
+		ucode_ver = psp->sos.fw_version;
+		/* Read TOS version from hardware */
+		psp_v13_0_init_sos_version(psp);
+		return (ucode_ver != psp->sos.fw_version);
+	default:
+		return false;
+	}
+
+	return false;
+}
+
+static int psp_v13_0_reg_program_no_ring(struct psp_context *psp, uint32_t val,
+					 enum psp_reg_prog_id id)
+{
+	struct amdgpu_device *adev = psp->adev;
+	int ret = -EOPNOTSUPP;
+
+	/* PSP will broadcast the value to all instances */
+	if (amdgpu_sriov_vf(adev)) {
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_101, GFX_CTRL_CMD_ID_GBR_IH_SET);
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_102, id);
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_103, val);
+
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_101),
+			0x80000000, 0x80000000, 0);
+	}
+
+	return ret;
+}
+
+static const struct psp_funcs psp_v13_0_funcs = {
+	.init_microcode = psp_v13_0_init_microcode,
+	.wait_for_bootloader = psp_v13_0_wait_for_bootloader_steady_state,
+	.bootloader_load_kdb = psp_v13_0_bootloader_load_kdb,
+	.bootloader_load_spl = psp_v13_0_bootloader_load_spl,
+	.bootloader_load_sysdrv = psp_v13_0_bootloader_load_sysdrv,
+	.bootloader_load_soc_drv = psp_v13_0_bootloader_load_soc_drv,
+	.bootloader_load_intf_drv = psp_v13_0_bootloader_load_intf_drv,
+	.bootloader_load_dbg_drv = psp_v13_0_bootloader_load_dbg_drv,
+	.bootloader_load_ras_drv = psp_v13_0_bootloader_load_ras_drv,
+	.bootloader_load_spdm_drv = psp_v13_0_bootloader_load_spdm_drv,
+	.bootloader_load_sos = psp_v13_0_bootloader_load_sos,
+	.ring_create = psp_v13_0_ring_create,
+	.ring_stop = psp_v13_0_ring_stop,
+	.ring_destroy = psp_v13_0_ring_destroy,
+	.ring_get_wptr = psp_v13_0_ring_get_wptr,
+	.ring_set_wptr = psp_v13_0_ring_set_wptr,
+	.mem_training = psp_v13_0_memory_training,
+	.load_usbc_pd_fw = psp_v13_0_load_usbc_pd_fw,
+	.read_usbc_pd_fw = psp_v13_0_read_usbc_pd_fw,
+	.update_spirom = psp_v13_0_update_spirom,
+	.dump_spirom = psp_v13_0_dump_spirom,
+	.vbflash_stat = psp_v13_0_vbflash_status,
+	.fatal_error_recovery_quirk = psp_v13_0_fatal_error_recovery_quirk,
+	.get_ras_capability = psp_v13_0_get_ras_capability,
+	.is_aux_sos_load_required = psp_v13_0_is_aux_sos_load_required,
+	.is_reload_needed = psp_v13_0_is_reload_needed,
+	.reg_program_no_ring = psp_v13_0_reg_program_no_ring,
+};
+
+void psp_v13_0_set_psp_funcs(struct psp_context *psp)
+{
+	psp->funcs = &psp_v13_0_funcs;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.h b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.h
new file mode 100644
index 000000000000..de5677ce4330
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __PSP_V13_0_H__
+#define __PSP_V13_0_H__
+
+#include "amdgpu_psp.h"
+
+#define PSP_SPIROM_UPDATE_TIMEOUT   60000   /* 60s */
+
+void psp_v13_0_set_psp_funcs(struct psp_context *psp);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v13_0_4.c b/drivers/gpu/drm/amd/amdgpu/psp_v13_0_4.c
new file mode 100644
index 000000000000..5f39a2edcc95
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v13_0_4.c
@@ -0,0 +1,355 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "amdgpu_psp.h"
+#include "amdgpu_ucode.h"
+#include "soc15_common.h"
+#include "psp_v13_0_4.h"
+
+#include "mp/mp_13_0_4_offset.h"
+#include "mp/mp_13_0_4_sh_mask.h"
+
+MODULE_FIRMWARE("amdgpu/psp_13_0_4_toc.bin");
+MODULE_FIRMWARE("amdgpu/psp_13_0_4_ta.bin");
+
+static int psp_v13_0_4_init_microcode(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+	char ucode_prefix[30];
+	int err = 0;
+
+	amdgpu_ucode_ip_version_decode(adev, MP0_HWIP, ucode_prefix, sizeof(ucode_prefix));
+
+	switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) {
+	case IP_VERSION(13, 0, 4):
+		err = psp_init_toc_microcode(psp, ucode_prefix);
+		if (err)
+			return err;
+		err = psp_init_ta_microcode(psp, ucode_prefix);
+		if (err)
+			return err;
+		break;
+	default:
+		BUG();
+	}
+
+	return 0;
+}
+
+static bool psp_v13_0_4_is_sos_alive(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+	uint32_t sol_reg;
+
+	sol_reg = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_81);
+
+	return sol_reg != 0x0;
+}
+
+static int psp_v13_0_4_wait_for_bootloader(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+
+	int ret;
+	int retry_loop;
+
+	for (retry_loop = 0; retry_loop < 10; retry_loop++) {
+		/* Wait for bootloader to signify that is
+		    ready having bit 31 of C2PMSG_35 set to 1 */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_35),
+			0x80000000, 0x80000000, PSP_WAITREG_NOVERBOSE);
+
+		if (ret == 0)
+			return 0;
+	}
+
+	return ret;
+}
+
+static int psp_v13_0_4_bootloader_load_component(struct psp_context  	*psp,
+					       struct psp_bin_desc 	*bin_desc,
+					       enum psp_bootloader_cmd  bl_cmd)
+{
+	int ret;
+	uint32_t psp_gfxdrv_command_reg = 0;
+	struct amdgpu_device *adev = psp->adev;
+
+	/* Check tOS sign of life register to confirm sys driver and sOS
+	 * are already been loaded.
+	 */
+	if (psp_v13_0_4_is_sos_alive(psp))
+		return 0;
+
+	ret = psp_v13_0_4_wait_for_bootloader(psp);
+	if (ret)
+		return ret;
+
+	memset(psp->fw_pri_buf, 0, PSP_1_MEG);
+
+	/* Copy PSP KDB binary to memory */
+	memcpy(psp->fw_pri_buf, bin_desc->start_addr, bin_desc->size_bytes);
+
+	/* Provide the PSP KDB to bootloader */
+	WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_36,
+	       (uint32_t)(psp->fw_pri_mc_addr >> 20));
+	psp_gfxdrv_command_reg = bl_cmd;
+	WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_35,
+	       psp_gfxdrv_command_reg);
+
+	ret = psp_v13_0_4_wait_for_bootloader(psp);
+
+	return ret;
+}
+
+static int psp_v13_0_4_bootloader_load_kdb(struct psp_context *psp)
+{
+	return psp_v13_0_4_bootloader_load_component(psp, &psp->kdb, PSP_BL__LOAD_KEY_DATABASE);
+}
+
+static int psp_v13_0_4_bootloader_load_spl(struct psp_context *psp)
+{
+	return psp_v13_0_4_bootloader_load_component(psp, &psp->kdb, PSP_BL__LOAD_TOS_SPL_TABLE);
+}
+
+static int psp_v13_0_4_bootloader_load_sysdrv(struct psp_context *psp)
+{
+	return psp_v13_0_4_bootloader_load_component(psp, &psp->sys, PSP_BL__LOAD_SYSDRV);
+}
+
+static int psp_v13_0_4_bootloader_load_soc_drv(struct psp_context *psp)
+{
+	return psp_v13_0_4_bootloader_load_component(psp, &psp->soc_drv, PSP_BL__LOAD_SOCDRV);
+}
+
+static int psp_v13_0_4_bootloader_load_intf_drv(struct psp_context *psp)
+{
+	return psp_v13_0_4_bootloader_load_component(psp, &psp->intf_drv, PSP_BL__LOAD_INTFDRV);
+}
+
+static int psp_v13_0_4_bootloader_load_dbg_drv(struct psp_context *psp)
+{
+	return psp_v13_0_4_bootloader_load_component(psp, &psp->dbg_drv, PSP_BL__LOAD_DBGDRV);
+}
+
+static int psp_v13_0_4_bootloader_load_sos(struct psp_context *psp)
+{
+	int ret;
+	unsigned int psp_gfxdrv_command_reg = 0;
+	struct amdgpu_device *adev = psp->adev;
+
+	/* Check sOS sign of life register to confirm sys driver and sOS
+	 * are already been loaded.
+	 */
+	if (psp_v13_0_4_is_sos_alive(psp))
+		return 0;
+
+	ret = psp_v13_0_4_wait_for_bootloader(psp);
+	if (ret)
+		return ret;
+
+	memset(psp->fw_pri_buf, 0, PSP_1_MEG);
+
+	/* Copy Secure OS binary to PSP memory */
+	memcpy(psp->fw_pri_buf, psp->sos.start_addr, psp->sos.size_bytes);
+
+	/* Provide the PSP secure OS to bootloader */
+	WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_36,
+	       (uint32_t)(psp->fw_pri_mc_addr >> 20));
+	psp_gfxdrv_command_reg = PSP_BL__LOAD_SOSDRV;
+	WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_35,
+	       psp_gfxdrv_command_reg);
+
+	/* there might be handshake issue with hardware which needs delay */
+	mdelay(20);
+	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_81),
+			   RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_81), 0,
+			   PSP_WAITREG_CHANGED);
+
+	return ret;
+}
+
+static int psp_v13_0_4_ring_stop(struct psp_context *psp,
+			       enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_sriov_vf(adev)) {
+		/* Write the ring destroy command*/
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_101,
+			     GFX_CTRL_CMD_ID_DESTROY_GPCOM_RING);
+		/* there might be handshake issue with hardware which needs delay */
+		mdelay(20);
+		/* Wait for response flag (bit 31) */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_101),
+			MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+	} else {
+		/* Write the ring destroy command*/
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_64,
+			     GFX_CTRL_CMD_ID_DESTROY_RINGS);
+		/* there might be handshake issue with hardware which needs delay */
+		mdelay(20);
+		/* Wait for response flag (bit 31) */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_64),
+			MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+	}
+
+	return ret;
+}
+
+static int psp_v13_0_4_ring_create(struct psp_context *psp,
+				 enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	unsigned int psp_ring_reg = 0;
+	struct psp_ring *ring = &psp->km_ring;
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_sriov_vf(adev)) {
+		ret = psp_v13_0_4_ring_stop(psp, ring_type);
+		if (ret) {
+			DRM_ERROR("psp_v13_0_ring_stop_sriov failed!\n");
+			return ret;
+		}
+
+		/* Write low address of the ring to C2PMSG_102 */
+		psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_102, psp_ring_reg);
+		/* Write high address of the ring to C2PMSG_103 */
+		psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_103, psp_ring_reg);
+
+		/* Write the ring initialization command to C2PMSG_101 */
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_101,
+			     GFX_CTRL_CMD_ID_INIT_GPCOM_RING);
+
+		/* there might be handshake issue with hardware which needs delay */
+		mdelay(20);
+
+		/* Wait for response flag (bit 31) in C2PMSG_101 */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_101),
+			MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+
+	} else {
+		/* Wait for sOS ready for ring creation */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_64),
+			MBOX_TOS_READY_FLAG, MBOX_TOS_READY_MASK, 0);
+		if (ret) {
+			DRM_ERROR("Failed to wait for trust OS ready for ring creation\n");
+			return ret;
+		}
+
+		/* Write low address of the ring to C2PMSG_69 */
+		psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_69, psp_ring_reg);
+		/* Write high address of the ring to C2PMSG_70 */
+		psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_70, psp_ring_reg);
+		/* Write size of ring to C2PMSG_71 */
+		psp_ring_reg = ring->ring_size;
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_71, psp_ring_reg);
+		/* Write the ring initialization command to C2PMSG_64 */
+		psp_ring_reg = ring_type;
+		psp_ring_reg = psp_ring_reg << 16;
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_64, psp_ring_reg);
+
+		/* there might be handshake issue with hardware which needs delay */
+		mdelay(20);
+
+		/* Wait for response flag (bit 31) in C2PMSG_64 */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_64),
+			MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+	}
+
+	return ret;
+}
+
+static int psp_v13_0_4_ring_destroy(struct psp_context *psp,
+				  enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	struct psp_ring *ring = &psp->km_ring;
+	struct amdgpu_device *adev = psp->adev;
+
+	ret = psp_v13_0_4_ring_stop(psp, ring_type);
+	if (ret)
+		DRM_ERROR("Fail to stop psp ring\n");
+
+	amdgpu_bo_free_kernel(&adev->firmware.rbuf,
+			      &ring->ring_mem_mc_addr,
+			      (void **)&ring->ring_mem);
+
+	return ret;
+}
+
+static uint32_t psp_v13_0_4_ring_get_wptr(struct psp_context *psp)
+{
+	uint32_t data;
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_sriov_vf(adev))
+		data = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_102);
+	else
+		data = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_67);
+
+	return data;
+}
+
+static void psp_v13_0_4_ring_set_wptr(struct psp_context *psp, uint32_t value)
+{
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_sriov_vf(adev)) {
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_102, value);
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_101,
+			     GFX_CTRL_CMD_ID_CONSUME_CMD);
+	} else
+		WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_67, value);
+}
+
+static const struct psp_funcs psp_v13_0_4_funcs = {
+	.init_microcode = psp_v13_0_4_init_microcode,
+	.bootloader_load_kdb = psp_v13_0_4_bootloader_load_kdb,
+	.bootloader_load_spl = psp_v13_0_4_bootloader_load_spl,
+	.bootloader_load_sysdrv = psp_v13_0_4_bootloader_load_sysdrv,
+	.bootloader_load_soc_drv = psp_v13_0_4_bootloader_load_soc_drv,
+	.bootloader_load_intf_drv = psp_v13_0_4_bootloader_load_intf_drv,
+	.bootloader_load_dbg_drv = psp_v13_0_4_bootloader_load_dbg_drv,
+	.bootloader_load_sos = psp_v13_0_4_bootloader_load_sos,
+	.ring_create = psp_v13_0_4_ring_create,
+	.ring_stop = psp_v13_0_4_ring_stop,
+	.ring_destroy = psp_v13_0_4_ring_destroy,
+	.ring_get_wptr = psp_v13_0_4_ring_get_wptr,
+	.ring_set_wptr = psp_v13_0_4_ring_set_wptr,
+};
+
+void psp_v13_0_4_set_psp_funcs(struct psp_context *psp)
+{
+	psp->funcs = &psp_v13_0_4_funcs;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v13_0_4.h b/drivers/gpu/drm/amd/amdgpu/psp_v13_0_4.h
new file mode 100644
index 000000000000..8547b8d514d5
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v13_0_4.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __PSP_V13_0_4_H__
+#define __PSP_V13_0_4_H__
+
+#include "amdgpu_psp.h"
+
+void psp_v13_0_4_set_psp_funcs(struct psp_context *psp);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v14_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v14_0.c
new file mode 100644
index 000000000000..38dfc5c19f2a
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v14_0.c
@@ -0,0 +1,705 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <drm/drm_drv.h>
+#include <linux/vmalloc.h>
+#include "amdgpu.h"
+#include "amdgpu_psp.h"
+#include "amdgpu_ucode.h"
+#include "soc15_common.h"
+#include "psp_v14_0.h"
+
+#include "mp/mp_14_0_2_offset.h"
+#include "mp/mp_14_0_2_sh_mask.h"
+
+MODULE_FIRMWARE("amdgpu/psp_14_0_2_sos.bin");
+MODULE_FIRMWARE("amdgpu/psp_14_0_2_ta.bin");
+MODULE_FIRMWARE("amdgpu/psp_14_0_3_sos.bin");
+MODULE_FIRMWARE("amdgpu/psp_14_0_3_sos_kicker.bin");
+MODULE_FIRMWARE("amdgpu/psp_14_0_3_ta.bin");
+MODULE_FIRMWARE("amdgpu/psp_14_0_3_ta_kicker.bin");
+MODULE_FIRMWARE("amdgpu/psp_14_0_5_toc.bin");
+MODULE_FIRMWARE("amdgpu/psp_14_0_5_ta.bin");
+
+/* For large FW files the time to complete can be very long */
+#define USBC_PD_POLLING_LIMIT_S 240
+
+/* Read USB-PD from LFB */
+#define GFX_CMD_USB_PD_USE_LFB 0x480
+
+/* VBIOS gfl defines */
+#define MBOX_READY_MASK 0x80000000
+#define MBOX_STATUS_MASK 0x0000FFFF
+#define MBOX_COMMAND_MASK 0x00FF0000
+#define MBOX_READY_FLAG 0x80000000
+#define C2PMSG_CMD_SPI_UPDATE_ROM_IMAGE_ADDR_LO 0x2
+#define C2PMSG_CMD_SPI_UPDATE_ROM_IMAGE_ADDR_HI 0x3
+#define C2PMSG_CMD_SPI_UPDATE_FLASH_IMAGE 0x4
+
+/* memory training timeout define */
+#define MEM_TRAIN_SEND_MSG_TIMEOUT_US	3000000
+
+static int psp_v14_0_init_microcode(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+	char ucode_prefix[30];
+	int err = 0;
+
+	amdgpu_ucode_ip_version_decode(adev, MP0_HWIP, ucode_prefix, sizeof(ucode_prefix));
+
+	switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) {
+	case IP_VERSION(14, 0, 2):
+	case IP_VERSION(14, 0, 3):
+		err = psp_init_sos_microcode(psp, ucode_prefix);
+		if (err)
+			return err;
+		err = psp_init_ta_microcode(psp, ucode_prefix);
+		if (err)
+			return err;
+		break;
+	case IP_VERSION(14, 0, 5):
+		err = psp_init_toc_microcode(psp, ucode_prefix);
+		if (err)
+			return err;
+		err = psp_init_ta_microcode(psp, ucode_prefix);
+		if (err)
+			return err;
+		break;
+	default:
+		BUG();
+	}
+
+	return 0;
+}
+
+static bool psp_v14_0_is_sos_alive(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+	uint32_t sol_reg;
+
+	sol_reg = RREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_81);
+
+	return sol_reg != 0x0;
+}
+
+static int psp_v14_0_wait_for_bootloader(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+
+	int ret;
+	int retry_loop;
+
+	for (retry_loop = 0; retry_loop < 10; retry_loop++) {
+		/* Wait for bootloader to signify that is
+		    ready having bit 31 of C2PMSG_35 set to 1 */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_35),
+			0x80000000, 0x80000000, PSP_WAITREG_NOVERBOSE);
+
+		if (ret == 0)
+			return 0;
+	}
+
+	return ret;
+}
+
+static int psp_v14_0_bootloader_load_component(struct psp_context  	*psp,
+					       struct psp_bin_desc 	*bin_desc,
+					       enum psp_bootloader_cmd  bl_cmd)
+{
+	int ret;
+	uint32_t psp_gfxdrv_command_reg = 0;
+	struct amdgpu_device *adev = psp->adev;
+
+	/* Check tOS sign of life register to confirm sys driver and sOS
+	 * are already been loaded.
+	 */
+	if (psp_v14_0_is_sos_alive(psp))
+		return 0;
+
+	ret = psp_v14_0_wait_for_bootloader(psp);
+	if (ret)
+		return ret;
+
+	memset(psp->fw_pri_buf, 0, PSP_1_MEG);
+
+	/* Copy PSP KDB binary to memory */
+	memcpy(psp->fw_pri_buf, bin_desc->start_addr, bin_desc->size_bytes);
+
+	/* Provide the PSP KDB to bootloader */
+	WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_36,
+	       (uint32_t)(psp->fw_pri_mc_addr >> 20));
+	psp_gfxdrv_command_reg = bl_cmd;
+	WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_35,
+	       psp_gfxdrv_command_reg);
+
+	ret = psp_v14_0_wait_for_bootloader(psp);
+
+	return ret;
+}
+
+static int psp_v14_0_bootloader_load_kdb(struct psp_context *psp)
+{
+	return psp_v14_0_bootloader_load_component(psp, &psp->kdb, PSP_BL__LOAD_KEY_DATABASE);
+}
+
+static int psp_v14_0_bootloader_load_spl(struct psp_context *psp)
+{
+	return psp_v14_0_bootloader_load_component(psp, &psp->spl, PSP_BL__LOAD_TOS_SPL_TABLE);
+}
+
+static int psp_v14_0_bootloader_load_sysdrv(struct psp_context *psp)
+{
+	return psp_v14_0_bootloader_load_component(psp, &psp->sys, PSP_BL__LOAD_SYSDRV);
+}
+
+static int psp_v14_0_bootloader_load_soc_drv(struct psp_context *psp)
+{
+	return psp_v14_0_bootloader_load_component(psp, &psp->soc_drv, PSP_BL__LOAD_SOCDRV);
+}
+
+static int psp_v14_0_bootloader_load_intf_drv(struct psp_context *psp)
+{
+	return psp_v14_0_bootloader_load_component(psp, &psp->intf_drv, PSP_BL__LOAD_INTFDRV);
+}
+
+static int psp_v14_0_bootloader_load_dbg_drv(struct psp_context *psp)
+{
+	/* dbg_drv was renamed to had_drv in psp v14 */
+	return psp_v14_0_bootloader_load_component(psp, &psp->dbg_drv, PSP_BL__LOAD_HADDRV);
+}
+
+static int psp_v14_0_bootloader_load_ras_drv(struct psp_context *psp)
+{
+	return psp_v14_0_bootloader_load_component(psp, &psp->ras_drv, PSP_BL__LOAD_RASDRV);
+}
+
+static int psp_v14_0_bootloader_load_ipkeymgr_drv(struct psp_context *psp)
+{
+	return psp_v14_0_bootloader_load_component(psp, &psp->ipkeymgr_drv, PSP_BL__LOAD_IPKEYMGRDRV);
+}
+
+static int psp_v14_0_bootloader_load_sos(struct psp_context *psp)
+{
+	int ret;
+	unsigned int psp_gfxdrv_command_reg = 0;
+	struct amdgpu_device *adev = psp->adev;
+
+	/* Check sOS sign of life register to confirm sys driver and sOS
+	 * are already been loaded.
+	 */
+	if (psp_v14_0_is_sos_alive(psp))
+		return 0;
+
+	ret = psp_v14_0_wait_for_bootloader(psp);
+	if (ret)
+		return ret;
+
+	memset(psp->fw_pri_buf, 0, PSP_1_MEG);
+
+	/* Copy Secure OS binary to PSP memory */
+	memcpy(psp->fw_pri_buf, psp->sos.start_addr, psp->sos.size_bytes);
+
+	/* Provide the PSP secure OS to bootloader */
+	WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_36,
+	       (uint32_t)(psp->fw_pri_mc_addr >> 20));
+	psp_gfxdrv_command_reg = PSP_BL__LOAD_SOSDRV;
+	WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_35,
+	       psp_gfxdrv_command_reg);
+
+	/* there might be handshake issue with hardware which needs delay */
+	mdelay(20);
+	ret = psp_wait_for(psp,
+			   SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_81),
+			   RREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_81), 0,
+			   PSP_WAITREG_CHANGED);
+
+	return ret;
+}
+
+static int psp_v14_0_ring_stop(struct psp_context *psp,
+			       enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_sriov_vf(adev)) {
+		/* Write the ring destroy command*/
+		WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_101,
+			     GFX_CTRL_CMD_ID_DESTROY_GPCOM_RING);
+		/* there might be handshake issue with hardware which needs delay */
+		mdelay(20);
+		/* Wait for response flag (bit 31) */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_101),
+			MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+	} else {
+		/* Write the ring destroy command*/
+		WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_64,
+			     GFX_CTRL_CMD_ID_DESTROY_RINGS);
+		/* there might be handshake issue with hardware which needs delay */
+		mdelay(20);
+		/* Wait for response flag (bit 31) */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_64),
+			MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+	}
+
+	return ret;
+}
+
+static int psp_v14_0_ring_create(struct psp_context *psp,
+				 enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	unsigned int psp_ring_reg = 0;
+	struct psp_ring *ring = &psp->km_ring;
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_sriov_vf(adev)) {
+		ret = psp_v14_0_ring_stop(psp, ring_type);
+		if (ret) {
+			DRM_ERROR("psp_v14_0_ring_stop_sriov failed!\n");
+			return ret;
+		}
+
+		/* Write low address of the ring to C2PMSG_102 */
+		psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_102, psp_ring_reg);
+		/* Write high address of the ring to C2PMSG_103 */
+		psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_103, psp_ring_reg);
+
+		/* Write the ring initialization command to C2PMSG_101 */
+		WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_101,
+			     GFX_CTRL_CMD_ID_INIT_GPCOM_RING);
+
+		/* there might be handshake issue with hardware which needs delay */
+		mdelay(20);
+
+		/* Wait for response flag (bit 31) in C2PMSG_101 */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_101),
+			MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+
+	} else {
+		/* Wait for sOS ready for ring creation */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_64),
+			MBOX_TOS_READY_FLAG, MBOX_TOS_READY_MASK, 0);
+		if (ret) {
+			DRM_ERROR("Failed to wait for trust OS ready for ring creation\n");
+			return ret;
+		}
+
+		/* Write low address of the ring to C2PMSG_69 */
+		psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_69, psp_ring_reg);
+		/* Write high address of the ring to C2PMSG_70 */
+		psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_70, psp_ring_reg);
+		/* Write size of ring to C2PMSG_71 */
+		psp_ring_reg = ring->ring_size;
+		WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_71, psp_ring_reg);
+		/* Write the ring initialization command to C2PMSG_64 */
+		psp_ring_reg = ring_type;
+		psp_ring_reg = psp_ring_reg << 16;
+		WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_64, psp_ring_reg);
+
+		/* there might be handshake issue with hardware which needs delay */
+		mdelay(20);
+
+		/* Wait for response flag (bit 31) in C2PMSG_64 */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_64),
+			MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, 0);
+	}
+
+	return ret;
+}
+
+static int psp_v14_0_ring_destroy(struct psp_context *psp,
+				  enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	struct psp_ring *ring = &psp->km_ring;
+	struct amdgpu_device *adev = psp->adev;
+
+	ret = psp_v14_0_ring_stop(psp, ring_type);
+	if (ret)
+		DRM_ERROR("Fail to stop psp ring\n");
+
+	amdgpu_bo_free_kernel(&adev->firmware.rbuf,
+			      &ring->ring_mem_mc_addr,
+			      (void **)&ring->ring_mem);
+
+	return ret;
+}
+
+static uint32_t psp_v14_0_ring_get_wptr(struct psp_context *psp)
+{
+	uint32_t data;
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_sriov_vf(adev))
+		data = RREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_102);
+	else
+		data = RREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_67);
+
+	return data;
+}
+
+static void psp_v14_0_ring_set_wptr(struct psp_context *psp, uint32_t value)
+{
+	struct amdgpu_device *adev = psp->adev;
+
+	if (amdgpu_sriov_vf(adev)) {
+		WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_102, value);
+		WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_101,
+			     GFX_CTRL_CMD_ID_CONSUME_CMD);
+	} else
+		WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_67, value);
+}
+
+static int psp_v14_0_memory_training_send_msg(struct psp_context *psp, int msg)
+{
+	int ret;
+	int i;
+	uint32_t data_32;
+	int max_wait;
+	struct amdgpu_device *adev = psp->adev;
+
+	data_32 = (psp->mem_train_ctx.c2p_train_data_offset >> 20);
+	WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_36, data_32);
+	WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_35, msg);
+
+	max_wait = MEM_TRAIN_SEND_MSG_TIMEOUT_US / adev->usec_timeout;
+	for (i = 0; i < max_wait; i++) {
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_35),
+			0x80000000, 0x80000000, PSP_WAITREG_NOVERBOSE);
+		if (ret == 0)
+			break;
+	}
+	if (i < max_wait)
+		ret = 0;
+	else
+		ret = -ETIME;
+
+	dev_dbg(adev->dev, "training %s %s, cost %d @ %d ms\n",
+		  (msg == PSP_BL__DRAM_SHORT_TRAIN) ? "short" : "long",
+		  (ret == 0) ? "succeed" : "failed",
+		  i, adev->usec_timeout/1000);
+	return ret;
+}
+
+
+static int psp_v14_0_memory_training(struct psp_context *psp, uint32_t ops)
+{
+	struct psp_memory_training_context *ctx = &psp->mem_train_ctx;
+	uint32_t *pcache = (uint32_t *)ctx->sys_cache;
+	struct amdgpu_device *adev = psp->adev;
+	uint32_t p2c_header[4];
+	uint32_t sz;
+	void *buf;
+	int ret, idx;
+
+	if (ctx->init == PSP_MEM_TRAIN_NOT_SUPPORT) {
+		dev_dbg(adev->dev, "Memory training is not supported.\n");
+		return 0;
+	} else if (ctx->init != PSP_MEM_TRAIN_INIT_SUCCESS) {
+		dev_err(adev->dev, "Memory training initialization failure.\n");
+		return -EINVAL;
+	}
+
+	if (psp_v14_0_is_sos_alive(psp)) {
+		dev_dbg(adev->dev, "SOS is alive, skip memory training.\n");
+		return 0;
+	}
+
+	amdgpu_device_vram_access(adev, ctx->p2c_train_data_offset, p2c_header, sizeof(p2c_header), false);
+	dev_dbg(adev->dev, "sys_cache[%08x,%08x,%08x,%08x] p2c_header[%08x,%08x,%08x,%08x]\n",
+		  pcache[0], pcache[1], pcache[2], pcache[3],
+		  p2c_header[0], p2c_header[1], p2c_header[2], p2c_header[3]);
+
+	if (ops & PSP_MEM_TRAIN_SEND_SHORT_MSG) {
+		dev_dbg(adev->dev, "Short training depends on restore.\n");
+		ops |= PSP_MEM_TRAIN_RESTORE;
+	}
+
+	if ((ops & PSP_MEM_TRAIN_RESTORE) &&
+	    pcache[0] != MEM_TRAIN_SYSTEM_SIGNATURE) {
+		dev_dbg(adev->dev, "sys_cache[0] is invalid, restore depends on save.\n");
+		ops |= PSP_MEM_TRAIN_SAVE;
+	}
+
+	if (p2c_header[0] == MEM_TRAIN_SYSTEM_SIGNATURE &&
+	    !(pcache[0] == MEM_TRAIN_SYSTEM_SIGNATURE &&
+	      pcache[3] == p2c_header[3])) {
+		dev_dbg(adev->dev, "sys_cache is invalid or out-of-date, need save training data to sys_cache.\n");
+		ops |= PSP_MEM_TRAIN_SAVE;
+	}
+
+	if ((ops & PSP_MEM_TRAIN_SAVE) &&
+	    p2c_header[0] != MEM_TRAIN_SYSTEM_SIGNATURE) {
+		dev_dbg(adev->dev, "p2c_header[0] is invalid, save depends on long training.\n");
+		ops |= PSP_MEM_TRAIN_SEND_LONG_MSG;
+	}
+
+	if (ops & PSP_MEM_TRAIN_SEND_LONG_MSG) {
+		ops &= ~PSP_MEM_TRAIN_SEND_SHORT_MSG;
+		ops |= PSP_MEM_TRAIN_SAVE;
+	}
+
+	dev_dbg(adev->dev, "Memory training ops:%x.\n", ops);
+
+	if (ops & PSP_MEM_TRAIN_SEND_LONG_MSG) {
+		/*
+		 * Long training will encroach a certain amount on the bottom of VRAM;
+		 * save the content from the bottom of VRAM to system memory
+		 * before training, and restore it after training to avoid
+		 * VRAM corruption.
+		 */
+		sz = BIST_MEM_TRAINING_ENCROACHED_SIZE;
+
+		if (adev->gmc.visible_vram_size < sz || !adev->mman.aper_base_kaddr) {
+			dev_err(adev->dev, "visible_vram_size %llx or aper_base_kaddr %p is not initialized.\n",
+				  adev->gmc.visible_vram_size,
+				  adev->mman.aper_base_kaddr);
+			return -EINVAL;
+		}
+
+		buf = vmalloc(sz);
+		if (!buf) {
+			dev_err(adev->dev, "failed to allocate system memory.\n");
+			return -ENOMEM;
+		}
+
+		if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+			memcpy_fromio(buf, adev->mman.aper_base_kaddr, sz);
+			ret = psp_v14_0_memory_training_send_msg(psp, PSP_BL__DRAM_LONG_TRAIN);
+			if (ret) {
+				DRM_ERROR("Send long training msg failed.\n");
+				vfree(buf);
+				drm_dev_exit(idx);
+				return ret;
+			}
+
+			memcpy_toio(adev->mman.aper_base_kaddr, buf, sz);
+			amdgpu_device_flush_hdp(adev, NULL);
+			vfree(buf);
+			drm_dev_exit(idx);
+		} else {
+			vfree(buf);
+			return -ENODEV;
+		}
+	}
+
+	if (ops & PSP_MEM_TRAIN_SAVE) {
+		amdgpu_device_vram_access(psp->adev, ctx->p2c_train_data_offset, ctx->sys_cache, ctx->train_data_size, false);
+	}
+
+	if (ops & PSP_MEM_TRAIN_RESTORE) {
+		amdgpu_device_vram_access(psp->adev, ctx->c2p_train_data_offset, ctx->sys_cache, ctx->train_data_size, true);
+	}
+
+	if (ops & PSP_MEM_TRAIN_SEND_SHORT_MSG) {
+		ret = psp_v14_0_memory_training_send_msg(psp, (amdgpu_force_long_training > 0) ?
+							 PSP_BL__DRAM_LONG_TRAIN : PSP_BL__DRAM_SHORT_TRAIN);
+		if (ret) {
+			dev_err(adev->dev, "send training msg failed.\n");
+			return ret;
+		}
+	}
+	ctx->training_cnt++;
+	return 0;
+}
+
+static int psp_v14_0_load_usbc_pd_fw(struct psp_context *psp, uint64_t fw_pri_mc_addr)
+{
+	struct amdgpu_device *adev = psp->adev;
+	uint32_t reg_status;
+	int ret, i = 0;
+
+	/*
+	 * LFB address which is aligned to 1MB address and has to be
+	 * right-shifted by 20 so that LFB address can be passed on a 32-bit C2P
+	 * register
+	 */
+	WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_36, (fw_pri_mc_addr >> 20));
+
+	ret = psp_wait_for(psp,
+			   SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_35),
+			   0x80000000, 0x80000000, 0);
+	if (ret)
+		return ret;
+
+	/* Fireup interrupt so PSP can pick up the address */
+	WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_35, (GFX_CMD_USB_PD_USE_LFB << 16));
+
+	/* FW load takes very long time */
+	do {
+		msleep(1000);
+		reg_status = RREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_35);
+
+		if (reg_status & 0x80000000)
+			goto done;
+
+	} while (++i < USBC_PD_POLLING_LIMIT_S);
+
+	return -ETIME;
+done:
+
+	if ((reg_status & 0xFFFF) != 0) {
+		DRM_ERROR("Address load failed - MP0_SMN_C2PMSG_35.Bits [15:0] = %04x\n",
+				reg_status & 0xFFFF);
+		return -EIO;
+	}
+
+	return 0;
+}
+
+static int psp_v14_0_read_usbc_pd_fw(struct psp_context *psp, uint32_t *fw_ver)
+{
+	struct amdgpu_device *adev = psp->adev;
+	int ret;
+
+	WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_35, C2PMSG_CMD_GFX_USB_PD_FW_VER);
+
+	ret = psp_wait_for(psp,
+			   SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_35),
+			   0x80000000, 0x80000000, 0);
+	if (!ret)
+		*fw_ver = RREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_36);
+
+	return ret;
+}
+
+static int psp_v14_0_exec_spi_cmd(struct psp_context *psp, int cmd)
+{
+	uint32_t reg_status = 0, reg_val = 0;
+	struct amdgpu_device *adev = psp->adev;
+	int ret;
+
+	/* clear MBX ready (MBOX_READY_MASK bit is 0) and set update command */
+	reg_val |= (cmd << 16);
+	WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_115,  reg_val);
+
+	/* Ring the doorbell */
+	WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_73, 1);
+
+	if (cmd == C2PMSG_CMD_SPI_UPDATE_FLASH_IMAGE)
+		ret = psp_wait_for_spirom_update(psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_115),
+						 MBOX_READY_FLAG, MBOX_READY_MASK, PSP_SPIROM_UPDATE_TIMEOUT);
+	else
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_115),
+			MBOX_READY_FLAG, MBOX_READY_MASK, 0);
+
+	ret = psp_wait_for(psp,
+			   SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_115),
+			   MBOX_READY_FLAG, MBOX_READY_MASK, 0);
+	if (ret) {
+		dev_err(adev->dev, "SPI cmd %x timed out, ret = %d", cmd, ret);
+		return ret;
+	}
+
+	reg_status = RREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_115);
+	if ((reg_status & 0xFFFF) != 0) {
+		dev_err(adev->dev, "SPI cmd %x failed, fail status = %04x\n",
+				cmd, reg_status & 0xFFFF);
+		return -EIO;
+	}
+
+	return 0;
+}
+
+static int psp_v14_0_update_spirom(struct psp_context *psp,
+				   uint64_t fw_pri_mc_addr)
+{
+	struct amdgpu_device *adev = psp->adev;
+	int ret;
+
+	/* Confirm PSP is ready to start */
+	ret = psp_wait_for(psp,
+			   SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_115),
+			   MBOX_READY_FLAG, MBOX_READY_MASK, 0);
+	if (ret) {
+		dev_err(adev->dev, "PSP Not ready to start processing, ret = %d", ret);
+		return ret;
+	}
+
+	WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_116, lower_32_bits(fw_pri_mc_addr));
+
+	ret = psp_v14_0_exec_spi_cmd(psp, C2PMSG_CMD_SPI_UPDATE_ROM_IMAGE_ADDR_LO);
+	if (ret)
+		return ret;
+
+	WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_116, upper_32_bits(fw_pri_mc_addr));
+
+	ret = psp_v14_0_exec_spi_cmd(psp, C2PMSG_CMD_SPI_UPDATE_ROM_IMAGE_ADDR_HI);
+	if (ret)
+		return ret;
+
+	psp->vbflash_done = true;
+
+	ret = psp_v14_0_exec_spi_cmd(psp, C2PMSG_CMD_SPI_UPDATE_FLASH_IMAGE);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static int psp_v14_0_vbflash_status(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+
+	return RREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_115);
+}
+
+static const struct psp_funcs psp_v14_0_funcs = {
+	.init_microcode = psp_v14_0_init_microcode,
+	.bootloader_load_kdb = psp_v14_0_bootloader_load_kdb,
+	.bootloader_load_spl = psp_v14_0_bootloader_load_spl,
+	.bootloader_load_sysdrv = psp_v14_0_bootloader_load_sysdrv,
+	.bootloader_load_soc_drv = psp_v14_0_bootloader_load_soc_drv,
+	.bootloader_load_intf_drv = psp_v14_0_bootloader_load_intf_drv,
+	.bootloader_load_dbg_drv = psp_v14_0_bootloader_load_dbg_drv,
+	.bootloader_load_ras_drv = psp_v14_0_bootloader_load_ras_drv,
+	.bootloader_load_ipkeymgr_drv = psp_v14_0_bootloader_load_ipkeymgr_drv,
+	.bootloader_load_sos = psp_v14_0_bootloader_load_sos,
+	.ring_create = psp_v14_0_ring_create,
+	.ring_stop = psp_v14_0_ring_stop,
+	.ring_destroy = psp_v14_0_ring_destroy,
+	.ring_get_wptr = psp_v14_0_ring_get_wptr,
+	.ring_set_wptr = psp_v14_0_ring_set_wptr,
+	.mem_training = psp_v14_0_memory_training,
+	.load_usbc_pd_fw = psp_v14_0_load_usbc_pd_fw,
+	.read_usbc_pd_fw = psp_v14_0_read_usbc_pd_fw,
+	.update_spirom = psp_v14_0_update_spirom,
+	.vbflash_stat = psp_v14_0_vbflash_status
+};
+
+void psp_v14_0_set_psp_funcs(struct psp_context *psp)
+{
+	psp->funcs = &psp_v14_0_funcs;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v14_0.h b/drivers/gpu/drm/amd/amdgpu/psp_v14_0.h
new file mode 100644
index 000000000000..dd18ba2cfad5
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v14_0.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __PSP_V14_0_H__
+#define __PSP_V14_0_H__
+
+#include "amdgpu_psp.h"
+
+#define PSP_SPIROM_UPDATE_TIMEOUT   60000   /* 60s */
+
+void psp_v14_0_set_psp_funcs(struct psp_context *psp);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c b/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c
index 2a535a4b8d5b..833830bc3e2e 100644
--- a/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c
@@ -24,145 +24,58 @@
  */
 
 #include <linux/firmware.h>
-#include <drm/drmP.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+
 #include "amdgpu.h"
 #include "amdgpu_psp.h"
 #include "amdgpu_ucode.h"
 #include "soc15_common.h"
 #include "psp_v3_1.h"
 
-#include "vega10/soc15ip.h"
-#include "vega10/MP/mp_9_0_offset.h"
-#include "vega10/MP/mp_9_0_sh_mask.h"
-#include "vega10/GC/gc_9_0_offset.h"
-#include "vega10/SDMA0/sdma0_4_0_offset.h"
-#include "vega10/NBIO/nbio_6_1_offset.h"
+#include "mp/mp_9_0_offset.h"
+#include "mp/mp_9_0_sh_mask.h"
+#include "gc/gc_9_0_offset.h"
+#include "sdma0/sdma0_4_0_offset.h"
+#include "nbio/nbio_6_1_offset.h"
+
+#include "oss/osssys_4_0_offset.h"
+#include "oss/osssys_4_0_sh_mask.h"
 
 MODULE_FIRMWARE("amdgpu/vega10_sos.bin");
 MODULE_FIRMWARE("amdgpu/vega10_asd.bin");
+MODULE_FIRMWARE("amdgpu/vega10_cap.bin");
+MODULE_FIRMWARE("amdgpu/vega12_sos.bin");
+MODULE_FIRMWARE("amdgpu/vega12_asd.bin");
 
-#define smnMP1_FIRMWARE_FLAGS 0x3010028
 
-static int
-psp_v3_1_get_fw_type(struct amdgpu_firmware_info *ucode, enum psp_gfx_fw_type *type)
-{
-	switch(ucode->ucode_id) {
-	case AMDGPU_UCODE_ID_SDMA0:
-		*type = GFX_FW_TYPE_SDMA0;
-		break;
-	case AMDGPU_UCODE_ID_SDMA1:
-		*type = GFX_FW_TYPE_SDMA1;
-		break;
-	case AMDGPU_UCODE_ID_CP_CE:
-		*type = GFX_FW_TYPE_CP_CE;
-		break;
-	case AMDGPU_UCODE_ID_CP_PFP:
-		*type = GFX_FW_TYPE_CP_PFP;
-		break;
-	case AMDGPU_UCODE_ID_CP_ME:
-		*type = GFX_FW_TYPE_CP_ME;
-		break;
-	case AMDGPU_UCODE_ID_CP_MEC1:
-		*type = GFX_FW_TYPE_CP_MEC;
-		break;
-	case AMDGPU_UCODE_ID_CP_MEC1_JT:
-		*type = GFX_FW_TYPE_CP_MEC_ME1;
-		break;
-	case AMDGPU_UCODE_ID_CP_MEC2:
-		*type = GFX_FW_TYPE_CP_MEC;
-		break;
-	case AMDGPU_UCODE_ID_CP_MEC2_JT:
-		*type = GFX_FW_TYPE_CP_MEC_ME2;
-		break;
-	case AMDGPU_UCODE_ID_RLC_G:
-		*type = GFX_FW_TYPE_RLC_G;
-		break;
-	case AMDGPU_UCODE_ID_SMC:
-		*type = GFX_FW_TYPE_SMU;
-		break;
-	case AMDGPU_UCODE_ID_UVD:
-		*type = GFX_FW_TYPE_UVD;
-		break;
-	case AMDGPU_UCODE_ID_VCE:
-		*type = GFX_FW_TYPE_VCE;
-		break;
-	case AMDGPU_UCODE_ID_MAXIMUM:
-	default:
-		return -EINVAL;
-	}
+#define smnMP1_FIRMWARE_FLAGS 0x3010028
 
-	return 0;
-}
+static int psp_v3_1_ring_stop(struct psp_context *psp,
+			      enum psp_ring_type ring_type);
 
-int psp_v3_1_init_microcode(struct psp_context *psp)
+static int psp_v3_1_init_microcode(struct psp_context *psp)
 {
 	struct amdgpu_device *adev = psp->adev;
-	const char *chip_name;
-	char fw_name[30];
+	char ucode_prefix[30];
 	int err = 0;
-	const struct psp_firmware_header_v1_0 *hdr;
 
 	DRM_DEBUG("\n");
 
-	switch (adev->asic_type) {
-	case CHIP_VEGA10:
-		chip_name = "vega10";
-		break;
-	default: BUG();
-	}
+	amdgpu_ucode_ip_version_decode(adev, MP0_HWIP, ucode_prefix, sizeof(ucode_prefix));
 
-	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_sos.bin", chip_name);
-	err = request_firmware(&adev->psp.sos_fw, fw_name, adev->dev);
+	err = psp_init_sos_microcode(psp, ucode_prefix);
 	if (err)
-		goto out;
+		return err;
 
-	err = amdgpu_ucode_validate(adev->psp.sos_fw);
-	if (err)
-		goto out;
-
-	hdr = (const struct psp_firmware_header_v1_0 *)adev->psp.sos_fw->data;
-	adev->psp.sos_fw_version = le32_to_cpu(hdr->header.ucode_version);
-	adev->psp.sos_feature_version = le32_to_cpu(hdr->ucode_feature_version);
-	adev->psp.sos_bin_size = le32_to_cpu(hdr->sos_size_bytes);
-	adev->psp.sys_bin_size = le32_to_cpu(hdr->header.ucode_size_bytes) -
-					le32_to_cpu(hdr->sos_size_bytes);
-	adev->psp.sys_start_addr = (uint8_t *)hdr +
-				le32_to_cpu(hdr->header.ucode_array_offset_bytes);
-	adev->psp.sos_start_addr = (uint8_t *)adev->psp.sys_start_addr +
-				le32_to_cpu(hdr->sos_offset_bytes);
-
-	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_asd.bin", chip_name);
-	err = request_firmware(&adev->psp.asd_fw, fw_name, adev->dev);
+	err = psp_init_asd_microcode(psp, ucode_prefix);
 	if (err)
-		goto out;
-
-	err = amdgpu_ucode_validate(adev->psp.asd_fw);
-	if (err)
-		goto out;
-
-	hdr = (const struct psp_firmware_header_v1_0 *)adev->psp.asd_fw->data;
-	adev->psp.asd_fw_version = le32_to_cpu(hdr->header.ucode_version);
-	adev->psp.asd_feature_version = le32_to_cpu(hdr->ucode_feature_version);
-	adev->psp.asd_ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes);
-	adev->psp.asd_start_addr = (uint8_t *)hdr +
-				le32_to_cpu(hdr->header.ucode_array_offset_bytes);
+		return err;
 
 	return 0;
-out:
-	if (err) {
-		dev_err(adev->dev,
-			"psp v3.1: Failed to load firmware \"%s\"\n",
-			fw_name);
-		release_firmware(adev->psp.sos_fw);
-		adev->psp.sos_fw = NULL;
-		release_firmware(adev->psp.asd_fw);
-		adev->psp.asd_fw = NULL;
-	}
-
-	return err;
 }
 
-int psp_v3_1_bootloader_load_sysdrv(struct psp_context *psp)
+static int psp_v3_1_bootloader_load_sysdrv(struct psp_context *psp)
 {
 	int ret;
 	uint32_t psp_gfxdrv_command_reg = 0;
@@ -178,19 +91,17 @@ int psp_v3_1_bootloader_load_sysdrv(struct psp_context *psp)
 
 	/* Wait for bootloader to signify that is ready having bit 31 of C2PMSG_35 set to 1 */
 	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
-			   0x80000000, 0x80000000, false);
+			   0x80000000, 0x80000000, 0);
 	if (ret)
 		return ret;
 
-	memset(psp->fw_pri_buf, 0, PSP_1_MEG);
-
 	/* Copy PSP System Driver binary to memory */
-	memcpy(psp->fw_pri_buf, psp->sys_start_addr, psp->sys_bin_size);
+	psp_copy_fw(psp, psp->sys.start_addr, psp->sys.size_bytes);
 
-	/* Provide the sys driver to bootrom */
+	/* Provide the sys driver to bootloader */
 	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36,
 	       (uint32_t)(psp->fw_pri_mc_addr >> 20));
-	psp_gfxdrv_command_reg = 1 << 16;
+	psp_gfxdrv_command_reg = PSP_BL__LOAD_SYSDRV;
 	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_35,
 	       psp_gfxdrv_command_reg);
 
@@ -198,12 +109,12 @@ int psp_v3_1_bootloader_load_sysdrv(struct psp_context *psp)
 	mdelay(20);
 
 	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
-			   0x80000000, 0x80000000, false);
+			   0x80000000, 0x80000000, 0);
 
 	return ret;
 }
 
-int psp_v3_1_bootloader_load_sos(struct psp_context *psp)
+static int psp_v3_1_bootloader_load_sos(struct psp_context *psp)
 {
 	int ret;
 	unsigned int psp_gfxdrv_command_reg = 0;
@@ -219,125 +130,163 @@ int psp_v3_1_bootloader_load_sos(struct psp_context *psp)
 
 	/* Wait for bootloader to signify that is ready having bit 31 of C2PMSG_35 set to 1 */
 	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
-			   0x80000000, 0x80000000, false);
+			   0x80000000, 0x80000000, 0);
 	if (ret)
 		return ret;
 
-	memset(psp->fw_pri_buf, 0, PSP_1_MEG);
-
 	/* Copy Secure OS binary to PSP memory */
-	memcpy(psp->fw_pri_buf, psp->sos_start_addr, psp->sos_bin_size);
+	psp_copy_fw(psp, psp->sos.start_addr, psp->sos.size_bytes);
 
-	/* Provide the PSP secure OS to bootrom */
+	/* Provide the PSP secure OS to bootloader */
 	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36,
 	       (uint32_t)(psp->fw_pri_mc_addr >> 20));
-	psp_gfxdrv_command_reg = 2 << 16;
+	psp_gfxdrv_command_reg = PSP_BL__LOAD_SOSDRV;
 	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_35,
 	       psp_gfxdrv_command_reg);
 
 	/* there might be handshake issue with hardware which needs delay */
 	mdelay(20);
 	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_81),
-			   RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_81),
-			   0, true);
-
+			   RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_81), 0,
+			   PSP_WAITREG_CHANGED);
 	return ret;
 }
 
-int psp_v3_1_prep_cmd_buf(struct amdgpu_firmware_info *ucode, struct psp_gfx_cmd_resp *cmd)
+static void psp_v3_1_reroute_ih(struct psp_context *psp)
 {
-	int ret;
-	uint64_t fw_mem_mc_addr = ucode->mc_addr;
+	struct amdgpu_device *adev = psp->adev;
+	uint32_t tmp;
 
-	memset(cmd, 0, sizeof(struct psp_gfx_cmd_resp));
+	/* Change IH ring for VMC */
+	tmp = REG_SET_FIELD(0, IH_CLIENT_CFG_DATA, CREDIT_RETURN_ADDR, 0x1244b);
+	tmp = REG_SET_FIELD(tmp, IH_CLIENT_CFG_DATA, CLIENT_TYPE, 1);
+	tmp = REG_SET_FIELD(tmp, IH_CLIENT_CFG_DATA, RING_ID, 1);
 
-	cmd->cmd_id = GFX_CMD_ID_LOAD_IP_FW;
-	cmd->cmd.cmd_load_ip_fw.fw_phy_addr_lo = lower_32_bits(fw_mem_mc_addr);
-	cmd->cmd.cmd_load_ip_fw.fw_phy_addr_hi = upper_32_bits(fw_mem_mc_addr);
-	cmd->cmd.cmd_load_ip_fw.fw_size = ucode->ucode_size;
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_69, 3);
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_70, tmp);
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64, GFX_CTRL_CMD_ID_GBR_IH_SET);
 
-	ret = psp_v3_1_get_fw_type(ucode, &cmd->cmd.cmd_load_ip_fw.fw_type);
-	if (ret)
-		DRM_ERROR("Unknown firmware type\n");
+	mdelay(20);
+	psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
+		     0x80000000, 0x8000FFFF, 0);
 
-	return ret;
+	/* Change IH ring for UMC */
+	tmp = REG_SET_FIELD(0, IH_CLIENT_CFG_DATA, CREDIT_RETURN_ADDR, 0x1216b);
+	tmp = REG_SET_FIELD(tmp, IH_CLIENT_CFG_DATA, RING_ID, 1);
+
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_69, 4);
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_70, tmp);
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64, GFX_CTRL_CMD_ID_GBR_IH_SET);
+
+	mdelay(20);
+	psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
+		     0x80000000, 0x8000FFFF, 0);
 }
 
-int psp_v3_1_ring_init(struct psp_context *psp, enum psp_ring_type ring_type)
+static int psp_v3_1_ring_create(struct psp_context *psp,
+				enum psp_ring_type ring_type)
 {
 	int ret = 0;
-	struct psp_ring *ring;
+	unsigned int psp_ring_reg = 0;
+	struct psp_ring *ring = &psp->km_ring;
 	struct amdgpu_device *adev = psp->adev;
 
-	ring = &psp->km_ring;
-
-	ring->ring_type = ring_type;
-
-	/* allocate 4k Page of Local Frame Buffer memory for ring */
-	ring->ring_size = 0x1000;
-	ret = amdgpu_bo_create_kernel(adev, ring->ring_size, PAGE_SIZE,
-				      AMDGPU_GEM_DOMAIN_VRAM,
-				      &adev->firmware.rbuf,
-				      &ring->ring_mem_mc_addr,
-				      (void **)&ring->ring_mem);
-	if (ret) {
-		ring->ring_size = 0;
-		return ret;
+	psp_v3_1_reroute_ih(psp);
+
+	if (amdgpu_sriov_vf(adev)) {
+		ring->ring_wptr = 0;
+		ret = psp_v3_1_ring_stop(psp, ring_type);
+		if (ret) {
+			DRM_ERROR("psp_v3_1_ring_stop_sriov failed!\n");
+			return ret;
+		}
+
+		/* Write low address of the ring to C2PMSG_102 */
+		psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_102, psp_ring_reg);
+		/* Write high address of the ring to C2PMSG_103 */
+		psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_103, psp_ring_reg);
+		/* No size initialization for sriov  */
+		/* Write the ring initialization command to C2PMSG_101 */
+		psp_ring_reg = ring_type;
+		psp_ring_reg = psp_ring_reg << 16;
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_101, psp_ring_reg);
+
+		/* there might be hardware handshake issue which needs delay */
+		mdelay(20);
+
+		/* Wait for response flag (bit 31) in C2PMSG_101 */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_101),
+			0x80000000, 0x8000FFFF, 0);
+	} else {
+
+		/* Write low address of the ring to C2PMSG_69 */
+		psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_69, psp_ring_reg);
+		/* Write high address of the ring to C2PMSG_70 */
+		psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr);
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_70, psp_ring_reg);
+		/* Write size of ring to C2PMSG_71 */
+		psp_ring_reg = ring->ring_size;
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_71, psp_ring_reg);
+		/* Write the ring initialization command to C2PMSG_64 */
+		psp_ring_reg = ring_type;
+		psp_ring_reg = psp_ring_reg << 16;
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64, psp_ring_reg);
+
+		/* there might be hardware handshake issue which needs delay */
+		mdelay(20);
+
+		/* Wait for response flag (bit 31) in C2PMSG_64 */
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
+			0x80000000, 0x8000FFFF, 0);
 	}
-
-	return 0;
+	return ret;
 }
 
-int psp_v3_1_ring_create(struct psp_context *psp, enum psp_ring_type ring_type)
+static int psp_v3_1_ring_stop(struct psp_context *psp,
+			      enum psp_ring_type ring_type)
 {
 	int ret = 0;
-	unsigned int psp_ring_reg = 0;
-	struct psp_ring *ring = &psp->km_ring;
 	struct amdgpu_device *adev = psp->adev;
 
-	/* Write low address of the ring to C2PMSG_69 */
-	psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr);
-	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_69, psp_ring_reg);
-	/* Write high address of the ring to C2PMSG_70 */
-	psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr);
-	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_70, psp_ring_reg);
-	/* Write size of ring to C2PMSG_71 */
-	psp_ring_reg = ring->ring_size;
-	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_71, psp_ring_reg);
-	/* Write the ring initialization command to C2PMSG_64 */
-	psp_ring_reg = ring_type;
-	psp_ring_reg = psp_ring_reg << 16;
-	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64, psp_ring_reg);
+	/* Write the ring destroy command*/
+	if (amdgpu_sriov_vf(adev))
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_101,
+				     GFX_CTRL_CMD_ID_DESTROY_GPCOM_RING);
+	else
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64,
+				     GFX_CTRL_CMD_ID_DESTROY_RINGS);
 
 	/* there might be handshake issue with hardware which needs delay */
 	mdelay(20);
 
-	/* Wait for response flag (bit 31) in C2PMSG_64 */
-	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
-			   0x80000000, 0x8000FFFF, false);
+	/* Wait for response flag (bit 31) */
+	if (amdgpu_sriov_vf(adev))
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_101),
+			0x80000000, 0x80000000, 0);
+	else
+		ret = psp_wait_for(
+			psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
+			0x80000000, 0x80000000, 0);
 
 	return ret;
 }
 
-int psp_v3_1_ring_destroy(struct psp_context *psp, enum psp_ring_type ring_type)
+static int psp_v3_1_ring_destroy(struct psp_context *psp,
+				 enum psp_ring_type ring_type)
 {
 	int ret = 0;
-	struct psp_ring *ring;
-	unsigned int psp_ring_reg = 0;
+	struct psp_ring *ring = &psp->km_ring;
 	struct amdgpu_device *adev = psp->adev;
 
-	ring = &psp->km_ring;
-
-	/* Write the ring destroy command to C2PMSG_64 */
-	psp_ring_reg = 3 << 16;
-	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64, psp_ring_reg);
-
-	/* there might be handshake issue with hardware which needs delay */
-	mdelay(20);
-
-	/* Wait for response flag (bit 31) in C2PMSG_64 */
-	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
-			   0x80000000, 0x80000000, false);
+	ret = psp_v3_1_ring_stop(psp, ring_type);
+	if (ret)
+		DRM_ERROR("Fail to stop psp ring\n");
 
 	amdgpu_bo_free_kernel(&adev->firmware.rbuf,
 			      &ring->ring_mem_mc_addr,
@@ -346,174 +295,89 @@ int psp_v3_1_ring_destroy(struct psp_context *psp, enum psp_ring_type ring_type)
 	return ret;
 }
 
-int psp_v3_1_cmd_submit(struct psp_context *psp,
-		        struct amdgpu_firmware_info *ucode,
-		        uint64_t cmd_buf_mc_addr, uint64_t fence_mc_addr,
-		        int index)
+static bool psp_v3_1_smu_reload_quirk(struct psp_context *psp)
 {
-	unsigned int psp_write_ptr_reg = 0;
-	struct psp_gfx_rb_frame * write_frame = psp->km_ring.ring_mem;
-	struct psp_ring *ring = &psp->km_ring;
 	struct amdgpu_device *adev = psp->adev;
-	uint32_t ring_size_dw = ring->ring_size / 4;
-	uint32_t rb_frame_size_dw = sizeof(struct psp_gfx_rb_frame) / 4;
+	uint32_t reg;
 
-	/* KM (GPCOM) prepare write pointer */
-	psp_write_ptr_reg = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67);
+	reg = RREG32_PCIE(smnMP1_FIRMWARE_FLAGS | 0x03b00000);
+	return (reg & MP1_FIRMWARE_FLAGS__INTERRUPTS_ENABLED_MASK) ? true : false;
+}
 
-	/* Update KM RB frame pointer to new frame */
-	/* write_frame ptr increments by size of rb_frame in bytes */
-	/* psp_write_ptr_reg increments by size of rb_frame in DWORDs */
-	if ((psp_write_ptr_reg % ring_size_dw) == 0)
-		write_frame = ring->ring_mem;
-	else
-		write_frame = ring->ring_mem + (psp_write_ptr_reg / rb_frame_size_dw);
+static int psp_v3_1_mode1_reset(struct psp_context *psp)
+{
+	int ret;
+	uint32_t offset;
+	struct amdgpu_device *adev = psp->adev;
 
-	/* Initialize KM RB frame */
-	memset(write_frame, 0, sizeof(struct psp_gfx_rb_frame));
+	offset = SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64);
 
-	/* Update KM RB frame */
-	write_frame->cmd_buf_addr_hi = upper_32_bits(cmd_buf_mc_addr);
-	write_frame->cmd_buf_addr_lo = lower_32_bits(cmd_buf_mc_addr);
-	write_frame->fence_addr_hi = upper_32_bits(fence_mc_addr);
-	write_frame->fence_addr_lo = lower_32_bits(fence_mc_addr);
-	write_frame->fence_value = index;
+	ret = psp_wait_for(psp, offset, 0x80000000, 0x8000FFFF, 0);
 
-	/* Update the write Pointer in DWORDs */
-	psp_write_ptr_reg = (psp_write_ptr_reg + rb_frame_size_dw) % ring_size_dw;
-	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67, psp_write_ptr_reg);
+	if (ret) {
+		DRM_INFO("psp is not working correctly before mode1 reset!\n");
+		return -EINVAL;
+	}
 
-	return 0;
-}
+	/*send the mode 1 reset command*/
+	WREG32(offset, GFX_CTRL_CMD_ID_MODE1_RST);
 
-static int
-psp_v3_1_sram_map(unsigned int *sram_offset, unsigned int *sram_addr_reg_offset,
-		  unsigned int *sram_data_reg_offset,
-		  enum AMDGPU_UCODE_ID ucode_id)
-{
-	int ret = 0;
+	msleep(500);
 
-	switch(ucode_id) {
-/* TODO: needs to confirm */
-#if 0
-	case AMDGPU_UCODE_ID_SMC:
-		*sram_offset = 0;
-		*sram_addr_reg_offset = 0;
-		*sram_data_reg_offset = 0;
-		break;
-#endif
-
-	case AMDGPU_UCODE_ID_CP_CE:
-		*sram_offset = 0x0;
-		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_CE_UCODE_ADDR);
-		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_CE_UCODE_DATA);
-		break;
-
-	case AMDGPU_UCODE_ID_CP_PFP:
-		*sram_offset = 0x0;
-		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_PFP_UCODE_ADDR);
-		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_PFP_UCODE_DATA);
-		break;
-
-	case AMDGPU_UCODE_ID_CP_ME:
-		*sram_offset = 0x0;
-		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_HYP_ME_UCODE_ADDR);
-		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_HYP_ME_UCODE_DATA);
-		break;
-
-	case AMDGPU_UCODE_ID_CP_MEC1:
-		*sram_offset = 0x10000;
-		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_MEC_ME1_UCODE_ADDR);
-		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_MEC_ME1_UCODE_DATA);
-		break;
-
-	case AMDGPU_UCODE_ID_CP_MEC2:
-		*sram_offset = 0x10000;
-		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_HYP_MEC2_UCODE_ADDR);
-		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_HYP_MEC2_UCODE_DATA);
-		break;
-
-	case AMDGPU_UCODE_ID_RLC_G:
-		*sram_offset = 0x2000;
-		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_UCODE_ADDR);
-		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_UCODE_DATA);
-		break;
-
-	case AMDGPU_UCODE_ID_SDMA0:
-		*sram_offset = 0x0;
-		*sram_addr_reg_offset = SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_UCODE_ADDR);
-		*sram_data_reg_offset = SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_UCODE_DATA);
-		break;
-
-/* TODO: needs to confirm */
-#if 0
-	case AMDGPU_UCODE_ID_SDMA1:
-		*sram_offset = ;
-		*sram_addr_reg_offset = ;
-		break;
-
-	case AMDGPU_UCODE_ID_UVD:
-		*sram_offset = ;
-		*sram_addr_reg_offset = ;
-		break;
-
-	case AMDGPU_UCODE_ID_VCE:
-		*sram_offset = ;
-		*sram_addr_reg_offset = ;
-		break;
-#endif
-
-	case AMDGPU_UCODE_ID_MAXIMUM:
-	default:
-		ret = -EINVAL;
-		break;
+	offset = SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_33);
+
+	ret = psp_wait_for(psp, offset, 0x80000000, 0x80000000, 0);
+
+	if (ret) {
+		DRM_INFO("psp mode 1 reset failed!\n");
+		return -EINVAL;
 	}
 
-	return ret;
+	DRM_INFO("psp mode1 reset succeed \n");
+
+	return 0;
 }
 
-bool psp_v3_1_compare_sram_data(struct psp_context *psp,
-				struct amdgpu_firmware_info *ucode,
-				enum AMDGPU_UCODE_ID ucode_type)
+static uint32_t psp_v3_1_ring_get_wptr(struct psp_context *psp)
 {
-	int err = 0;
-	unsigned int fw_sram_reg_val = 0;
-	unsigned int fw_sram_addr_reg_offset = 0;
-	unsigned int fw_sram_data_reg_offset = 0;
-	unsigned int ucode_size;
-	uint32_t *ucode_mem = NULL;
+	uint32_t data;
 	struct amdgpu_device *adev = psp->adev;
 
-	err = psp_v3_1_sram_map(&fw_sram_reg_val, &fw_sram_addr_reg_offset,
-				&fw_sram_data_reg_offset, ucode_type);
-	if (err)
-		return false;
-
-	WREG32(fw_sram_addr_reg_offset, fw_sram_reg_val);
-
-	ucode_size = ucode->ucode_size;
-	ucode_mem = (uint32_t *)ucode->kaddr;
-	while (ucode_size) {
-		fw_sram_reg_val = RREG32(fw_sram_data_reg_offset);
-
-		if (*ucode_mem != fw_sram_reg_val)
-			return false;
-
-		ucode_mem++;
-		/* 4 bytes */
-		ucode_size -= 4;
-	}
-
-	return true;
+	if (amdgpu_sriov_vf(adev))
+		data = psp->km_ring.ring_wptr;
+	else
+		data = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67);
+	return data;
 }
 
-bool psp_v3_1_smu_reload_quirk(struct psp_context *psp)
+static void psp_v3_1_ring_set_wptr(struct psp_context *psp, uint32_t value)
 {
 	struct amdgpu_device *adev = psp->adev;
-	uint32_t reg;
 
-	reg = smnMP1_FIRMWARE_FLAGS | 0x03b00000;
-	WREG32_SOC15(NBIO, 0, mmPCIE_INDEX2, reg);
-	reg = RREG32_SOC15(NBIO, 0, mmPCIE_DATA2);
-	return (reg & MP1_FIRMWARE_FLAGS__INTERRUPTS_ENABLED_MASK) ? true : false;
+	if (amdgpu_sriov_vf(adev)) {
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_102, value);
+		/* send interrupt to PSP for SRIOV ring write pointer update */
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_101,
+			GFX_CTRL_CMD_ID_CONSUME_CMD);
+		psp->km_ring.ring_wptr = value;
+	} else
+		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67, value);
+}
+
+static const struct psp_funcs psp_v3_1_funcs = {
+	.init_microcode = psp_v3_1_init_microcode,
+	.bootloader_load_sysdrv = psp_v3_1_bootloader_load_sysdrv,
+	.bootloader_load_sos = psp_v3_1_bootloader_load_sos,
+	.ring_create = psp_v3_1_ring_create,
+	.ring_stop = psp_v3_1_ring_stop,
+	.ring_destroy = psp_v3_1_ring_destroy,
+	.smu_reload_quirk = psp_v3_1_smu_reload_quirk,
+	.mode1_reset = psp_v3_1_mode1_reset,
+	.ring_get_wptr = psp_v3_1_ring_get_wptr,
+	.ring_set_wptr = psp_v3_1_ring_set_wptr,
+};
+
+void psp_v3_1_set_psp_funcs(struct psp_context *psp)
+{
+	psp->funcs = &psp_v3_1_funcs;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v3_1.h b/drivers/gpu/drm/amd/amdgpu/psp_v3_1.h
index 9dcd0b25c4c6..e411e31ba452 100644
--- a/drivers/gpu/drm/amd/amdgpu/psp_v3_1.h
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v3_1.h
@@ -32,23 +32,6 @@ enum { PSP_BINARY_ALIGNMENT = 64 };
 enum { PSP_BOOTLOADER_1_MEG_ALIGNMENT = 0x100000 };
 enum { PSP_BOOTLOADER_8_MEM_ALIGNMENT = 0x800000 };
 
-extern int psp_v3_1_init_microcode(struct psp_context *psp);
-extern int psp_v3_1_bootloader_load_sysdrv(struct psp_context *psp);
-extern int psp_v3_1_bootloader_load_sos(struct psp_context *psp);
-extern int psp_v3_1_prep_cmd_buf(struct amdgpu_firmware_info *ucode,
-				 struct psp_gfx_cmd_resp *cmd);
-extern int psp_v3_1_ring_init(struct psp_context *psp,
-			      enum psp_ring_type ring_type);
-extern int psp_v3_1_ring_create(struct psp_context *psp,
-				enum psp_ring_type ring_type);
-extern int psp_v3_1_ring_destroy(struct psp_context *psp,
-				enum psp_ring_type ring_type);
-extern int psp_v3_1_cmd_submit(struct psp_context *psp,
-			       struct amdgpu_firmware_info *ucode,
-			       uint64_t cmd_buf_mc_addr, uint64_t fence_mc_addr,
-			       int index);
-extern bool psp_v3_1_compare_sram_data(struct psp_context *psp,
-				       struct amdgpu_firmware_info *ucode,
-				       enum AMDGPU_UCODE_ID ucode_type);
-extern bool psp_v3_1_smu_reload_quirk(struct psp_context *psp);
+void psp_v3_1_set_psp_funcs(struct psp_context *psp);
+
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_common.h b/drivers/gpu/drm/amd/amdgpu/sdma_common.h
new file mode 100644
index 000000000000..8629ef7e8ad9
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_common.h
@@ -0,0 +1,42 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __SDMA_COMMON_H__
+#define __SDMA_COMMON_H__
+
+enum sdma_utcl2_cache_read_policy {
+	CACHE_READ_POLICY_L2__LRU    = 0x00000000,
+	CACHE_READ_POLICY_L2__STREAM = 0x00000001,
+	CACHE_READ_POLICY_L2__NOA    = 0x00000002,
+	CACHE_READ_POLICY_L2__DEFAULT = CACHE_READ_POLICY_L2__NOA,
+};
+
+enum sdma_utcl2_cache_write_policy {
+	CACHE_WRITE_POLICY_L2__LRU    = 0x00000000,
+	CACHE_WRITE_POLICY_L2__STREAM = 0x00000001,
+	CACHE_WRITE_POLICY_L2__NOA    = 0x00000002,
+	CACHE_WRITE_POLICY_L2__BYPASS = 0x00000003,
+	CACHE_WRITE_POLICY_L2__DEFAULT = CACHE_WRITE_POLICY_L2__BYPASS,
+};
+
+#endif /* __SDMA_COMMON_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
index f2d0710258cb..92ce580647cd 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
@@ -21,8 +21,11 @@
  *
  * Authors: Alex Deucher
  */
+
+#include <linux/delay.h>
 #include <linux/firmware.h>
-#include <drm/drmP.h>
+#include <linux/module.h>
+
 #include "amdgpu.h"
 #include "amdgpu_ucode.h"
 #include "amdgpu_trace.h"
@@ -44,6 +47,8 @@
 
 #include "iceland_sdma_pkt_open.h"
 
+#include "ivsrcid/ivsrcid_vislands30.h"
+
 static void sdma_v2_4_set_ring_funcs(struct amdgpu_device *adev);
 static void sdma_v2_4_set_buffer_funcs(struct amdgpu_device *adev);
 static void sdma_v2_4_set_vm_pte_funcs(struct amdgpu_device *adev);
@@ -52,22 +57,19 @@ static void sdma_v2_4_set_irq_funcs(struct amdgpu_device *adev);
 MODULE_FIRMWARE("amdgpu/topaz_sdma.bin");
 MODULE_FIRMWARE("amdgpu/topaz_sdma1.bin");
 
-static const u32 sdma_offsets[SDMA_MAX_INSTANCE] =
-{
+static const u32 sdma_offsets[SDMA_MAX_INSTANCE] = {
 	SDMA0_REGISTER_OFFSET,
 	SDMA1_REGISTER_OFFSET
 };
 
-static const u32 golden_settings_iceland_a11[] =
-{
+static const u32 golden_settings_iceland_a11[] = {
 	mmSDMA0_CHICKEN_BITS, 0xfc910007, 0x00810007,
 	mmSDMA0_CLK_CTRL, 0xff000fff, 0x00000000,
 	mmSDMA1_CHICKEN_BITS, 0xfc910007, 0x00810007,
 	mmSDMA1_CLK_CTRL, 0xff000fff, 0x00000000,
 };
 
-static const u32 iceland_mgcg_cgcg_init[] =
-{
+static const u32 iceland_mgcg_cgcg_init[] = {
 	mmSDMA0_CLK_CTRL, 0xff000ff0, 0x00000100,
 	mmSDMA1_CLK_CTRL, 0xff000ff0, 0x00000100
 };
@@ -93,12 +95,12 @@ static void sdma_v2_4_init_golden_registers(struct amdgpu_device *adev)
 {
 	switch (adev->asic_type) {
 	case CHIP_TOPAZ:
-		amdgpu_program_register_sequence(adev,
-						 iceland_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(iceland_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_iceland_a11,
-						 (const u32)ARRAY_SIZE(golden_settings_iceland_a11));
+		amdgpu_device_program_register_sequence(adev,
+							iceland_mgcg_cgcg_init,
+							ARRAY_SIZE(iceland_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_iceland_a11,
+							ARRAY_SIZE(golden_settings_iceland_a11));
 		break;
 	default:
 		break;
@@ -108,10 +110,9 @@ static void sdma_v2_4_init_golden_registers(struct amdgpu_device *adev)
 static void sdma_v2_4_free_microcode(struct amdgpu_device *adev)
 {
 	int i;
-	for (i = 0; i < adev->sdma.num_instances; i++) {
-		release_firmware(adev->sdma.instance[i].fw);
-		adev->sdma.instance[i].fw = NULL;
-	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		amdgpu_ucode_release(&adev->sdma.instance[i].fw);
 }
 
 /**
@@ -126,7 +127,6 @@ static void sdma_v2_4_free_microcode(struct amdgpu_device *adev)
 static int sdma_v2_4_init_microcode(struct amdgpu_device *adev)
 {
 	const char *chip_name;
-	char fw_name[30];
 	int err = 0, i;
 	struct amdgpu_firmware_info *info = NULL;
 	const struct common_firmware_header *header = NULL;
@@ -138,18 +138,19 @@ static int sdma_v2_4_init_microcode(struct amdgpu_device *adev)
 	case CHIP_TOPAZ:
 		chip_name = "topaz";
 		break;
-	default: BUG();
+	default:
+		BUG();
 	}
 
 	for (i = 0; i < adev->sdma.num_instances; i++) {
 		if (i == 0)
-			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_sdma.bin", chip_name);
+			err = amdgpu_ucode_request(adev, &adev->sdma.instance[i].fw,
+						   AMDGPU_UCODE_REQUIRED,
+						   "amdgpu/%s_sdma.bin", chip_name);
 		else
-			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_sdma1.bin", chip_name);
-		err = request_firmware(&adev->sdma.instance[i].fw, fw_name, adev->dev);
-		if (err)
-			goto out;
-		err = amdgpu_ucode_validate(adev->sdma.instance[i].fw);
+			err = amdgpu_ucode_request(adev, &adev->sdma.instance[i].fw,
+						   AMDGPU_UCODE_REQUIRED,
+						   "amdgpu/%s_sdma1.bin", chip_name);
 		if (err)
 			goto out;
 		hdr = (const struct sdma_firmware_header_v1_0 *)adev->sdma.instance[i].fw->data;
@@ -170,11 +171,10 @@ static int sdma_v2_4_init_microcode(struct amdgpu_device *adev)
 
 out:
 	if (err) {
-		pr_err("sdma_v2_4: Failed to load firmware \"%s\"\n", fw_name);
-		for (i = 0; i < adev->sdma.num_instances; i++) {
-			release_firmware(adev->sdma.instance[i].fw);
-			adev->sdma.instance[i].fw = NULL;
-		}
+		pr_err("sdma_v2_4: Failed to load firmware \"%s_sdma%s.bin\"\n",
+		       chip_name, i == 0 ? "" : "1");
+		for (i = 0; i < adev->sdma.num_instances; i++)
+			amdgpu_ucode_release(&adev->sdma.instance[i].fw);
 	}
 	return err;
 }
@@ -189,7 +189,7 @@ out:
 static uint64_t sdma_v2_4_ring_get_rptr(struct amdgpu_ring *ring)
 {
 	/* XXX check if swapping is necessary on BE */
-	return ring->adev->wb.wb[ring->rptr_offs] >> 2;
+	return *ring->rptr_cpu_addr >> 2;
 }
 
 /**
@@ -202,8 +202,7 @@ static uint64_t sdma_v2_4_ring_get_rptr(struct amdgpu_ring *ring)
 static uint64_t sdma_v2_4_ring_get_wptr(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
-	int me = (ring == &ring->adev->sdma.instance[0].ring) ? 0 : 1;
-	u32 wptr = RREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) >> 2;
+	u32 wptr = RREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) >> 2;
 
 	return wptr;
 }
@@ -218,14 +217,13 @@ static uint64_t sdma_v2_4_ring_get_wptr(struct amdgpu_ring *ring)
 static void sdma_v2_4_ring_set_wptr(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
-	int me = (ring == &ring->adev->sdma.instance[0].ring) ? 0 : 1;
 
-	WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[me], lower_32_bits(ring->wptr) << 2);
+	WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me], ring->wptr << 2);
 }
 
 static void sdma_v2_4_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
 {
-	struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring);
+	struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring);
 	int i;
 
 	for (i = 0; i < count; i++)
@@ -240,21 +238,24 @@ static void sdma_v2_4_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
  * sdma_v2_4_ring_emit_ib - Schedule an IB on the DMA engine
  *
  * @ring: amdgpu ring pointer
+ * @job: job to retrieve vmid from
  * @ib: IB object to schedule
+ * @flags: unused
  *
  * Schedule an IB in the DMA ring (VI).
  */
 static void sdma_v2_4_ring_emit_ib(struct amdgpu_ring *ring,
+				   struct amdgpu_job *job,
 				   struct amdgpu_ib *ib,
-				   unsigned vm_id, bool ctx_switch)
+				   uint32_t flags)
 {
-	u32 vmid = vm_id & 0xf;
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
 
 	/* IB packet must end on a 8 DW boundary */
-	sdma_v2_4_ring_insert_nop(ring, (10 - (lower_32_bits(ring->wptr) & 7)) % 8);
+	sdma_v2_4_ring_insert_nop(ring, (2 - lower_32_bits(ring->wptr)) & 7);
 
 	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_INDIRECT) |
-			  SDMA_PKT_INDIRECT_HEADER_VMID(vmid));
+			  SDMA_PKT_INDIRECT_HEADER_VMID(vmid & 0xf));
 	/* base must be 32 byte aligned */
 	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr) & 0xffffffe0);
 	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
@@ -265,7 +266,7 @@ static void sdma_v2_4_ring_emit_ib(struct amdgpu_ring *ring,
 }
 
 /**
- * sdma_v2_4_hdp_flush_ring_emit - emit an hdp flush on the DMA ring
+ * sdma_v2_4_ring_emit_hdp_flush - emit an hdp flush on the DMA ring
  *
  * @ring: amdgpu ring pointer
  *
@@ -275,7 +276,7 @@ static void sdma_v2_4_ring_emit_hdp_flush(struct amdgpu_ring *ring)
 {
 	u32 ref_and_mask = 0;
 
-	if (ring == &ring->adev->sdma.instance[0].ring)
+	if (ring->me == 0)
 		ref_and_mask = REG_SET_FIELD(ref_and_mask, GPU_HDP_FLUSH_DONE, SDMA0, 1);
 	else
 		ref_and_mask = REG_SET_FIELD(ref_and_mask, GPU_HDP_FLUSH_DONE, SDMA1, 1);
@@ -291,18 +292,13 @@ static void sdma_v2_4_ring_emit_hdp_flush(struct amdgpu_ring *ring)
 			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
 }
 
-static void sdma_v2_4_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
-{
-	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
-			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
-	amdgpu_ring_write(ring, mmHDP_DEBUG0);
-	amdgpu_ring_write(ring, 1);
-}
 /**
  * sdma_v2_4_ring_emit_fence - emit a fence on the DMA ring
  *
  * @ring: amdgpu ring pointer
- * @fence: amdgpu fence object
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
  *
  * Add a DMA fence packet to the ring to write
  * the fence seq number and DMA trap packet to generate
@@ -341,15 +337,9 @@ static void sdma_v2_4_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 se
  */
 static void sdma_v2_4_gfx_stop(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *sdma0 = &adev->sdma.instance[0].ring;
-	struct amdgpu_ring *sdma1 = &adev->sdma.instance[1].ring;
 	u32 rb_cntl, ib_cntl;
 	int i;
 
-	if ((adev->mman.buffer_funcs_ring == sdma0) ||
-	    (adev->mman.buffer_funcs_ring == sdma1))
-		amdgpu_ttm_set_active_vram_size(adev, adev->mc.visible_vram_size);
-
 	for (i = 0; i < adev->sdma.num_instances; i++) {
 		rb_cntl = RREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i]);
 		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 0);
@@ -358,8 +348,6 @@ static void sdma_v2_4_gfx_stop(struct amdgpu_device *adev)
 		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 0);
 		WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl);
 	}
-	sdma0->ready = false;
-	sdma1->ready = false;
 }
 
 /**
@@ -415,12 +403,10 @@ static int sdma_v2_4_gfx_resume(struct amdgpu_device *adev)
 	struct amdgpu_ring *ring;
 	u32 rb_cntl, ib_cntl;
 	u32 rb_bufsz;
-	u32 wb_offset;
 	int i, j, r;
 
 	for (i = 0; i < adev->sdma.num_instances; i++) {
 		ring = &adev->sdma.instance[i].ring;
-		wb_offset = (ring->rptr_offs * 4);
 
 		mutex_lock(&adev->srbm_mutex);
 		for (j = 0; j < 16; j++) {
@@ -456,9 +442,9 @@ static int sdma_v2_4_gfx_resume(struct amdgpu_device *adev)
 
 		/* set the wb address whether it's enabled or not */
 		WREG32(mmSDMA0_GFX_RB_RPTR_ADDR_HI + sdma_offsets[i],
-		       upper_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFF);
+		       upper_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFF);
 		WREG32(mmSDMA0_GFX_RB_RPTR_ADDR_LO + sdma_offsets[i],
-		       lower_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFC);
+		       lower_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFC);
 
 		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RPTR_WRITEBACK_ENABLE, 1);
 
@@ -466,7 +452,7 @@ static int sdma_v2_4_gfx_resume(struct amdgpu_device *adev)
 		WREG32(mmSDMA0_GFX_RB_BASE_HI + sdma_offsets[i], ring->gpu_addr >> 40);
 
 		ring->wptr = 0;
-		WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[i], lower_32_bits(ring->wptr) << 2);
+		WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[i], ring->wptr << 2);
 
 		/* enable DMA RB */
 		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 1);
@@ -479,21 +465,14 @@ static int sdma_v2_4_gfx_resume(struct amdgpu_device *adev)
 #endif
 		/* enable DMA IBs */
 		WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl);
-
-		ring->ready = true;
 	}
 
 	sdma_v2_4_enable(adev, true);
 	for (i = 0; i < adev->sdma.num_instances; i++) {
 		ring = &adev->sdma.instance[i].ring;
-		r = amdgpu_ring_test_ring(ring);
-		if (r) {
-			ring->ready = false;
+		r = amdgpu_ring_test_helper(ring);
+		if (r)
 			return r;
-		}
-
-		if (adev->mman.buffer_funcs_ring == ring)
-			amdgpu_ttm_set_active_vram_size(adev, adev->mc.real_vram_size);
 	}
 
 	return 0;
@@ -513,41 +492,6 @@ static int sdma_v2_4_rlc_resume(struct amdgpu_device *adev)
 	return 0;
 }
 
-/**
- * sdma_v2_4_load_microcode - load the sDMA ME ucode
- *
- * @adev: amdgpu_device pointer
- *
- * Loads the sDMA0/1 ucode.
- * Returns 0 for success, -EINVAL if the ucode is not available.
- */
-static int sdma_v2_4_load_microcode(struct amdgpu_device *adev)
-{
-	const struct sdma_firmware_header_v1_0 *hdr;
-	const __le32 *fw_data;
-	u32 fw_size;
-	int i, j;
-
-	/* halt the MEs */
-	sdma_v2_4_enable(adev, false);
-
-	for (i = 0; i < adev->sdma.num_instances; i++) {
-		if (!adev->sdma.instance[i].fw)
-			return -EINVAL;
-		hdr = (const struct sdma_firmware_header_v1_0 *)adev->sdma.instance[i].fw->data;
-		amdgpu_ucode_print_sdma_hdr(&hdr->header);
-		fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
-		fw_data = (const __le32 *)
-			(adev->sdma.instance[i].fw->data +
-			 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
-		WREG32(mmSDMA0_UCODE_ADDR + sdma_offsets[i], 0);
-		for (j = 0; j < fw_size; j++)
-			WREG32(mmSDMA0_UCODE_DATA + sdma_offsets[i], le32_to_cpup(fw_data++));
-		WREG32(mmSDMA0_UCODE_ADDR + sdma_offsets[i], adev->sdma.instance[i].fw_version);
-	}
-
-	return 0;
-}
 
 /**
  * sdma_v2_4_start - setup and start the async dma engines
@@ -561,23 +505,6 @@ static int sdma_v2_4_start(struct amdgpu_device *adev)
 {
 	int r;
 
-	if (!adev->pp_enabled) {
-		if (adev->firmware.load_type != AMDGPU_FW_LOAD_SMU) {
-			r = sdma_v2_4_load_microcode(adev);
-			if (r)
-				return r;
-		} else {
-			r = adev->smu.smumgr_funcs->check_fw_load_finish(adev,
-							AMDGPU_UCODE_ID_SDMA0);
-			if (r)
-				return -EINVAL;
-			r = adev->smu.smumgr_funcs->check_fw_load_finish(adev,
-							AMDGPU_UCODE_ID_SDMA1);
-			if (r)
-				return -EINVAL;
-		}
-	}
-
 	/* halt the engine before programing */
 	sdma_v2_4_enable(adev, false);
 
@@ -610,22 +537,17 @@ static int sdma_v2_4_ring_test_ring(struct amdgpu_ring *ring)
 	u32 tmp;
 	u64 gpu_addr;
 
-	r = amdgpu_wb_get(adev, &index);
-	if (r) {
-		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r)
 		return r;
-	}
 
 	gpu_addr = adev->wb.gpu_addr + (index * 4);
 	tmp = 0xCAFEDEAD;
 	adev->wb.wb[index] = cpu_to_le32(tmp);
 
 	r = amdgpu_ring_alloc(ring, 5);
-	if (r) {
-		DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r);
-		amdgpu_wb_free(adev, index);
-		return r;
-	}
+	if (r)
+		goto error_free_wb;
 
 	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
 			  SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR));
@@ -639,18 +561,14 @@ static int sdma_v2_4_ring_test_ring(struct amdgpu_ring *ring)
 		tmp = le32_to_cpu(adev->wb.wb[index]);
 		if (tmp == 0xDEADBEEF)
 			break;
-		DRM_UDELAY(1);
+		udelay(1);
 	}
 
-	if (i < adev->usec_timeout) {
-		DRM_INFO("ring test on %d succeeded in %d usecs\n", ring->idx, i);
-	} else {
-		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
-			  ring->idx, tmp);
-		r = -EINVAL;
-	}
-	amdgpu_wb_free(adev, index);
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
 
+error_free_wb:
+	amdgpu_device_wb_free(adev, index);
 	return r;
 }
 
@@ -658,6 +576,7 @@ static int sdma_v2_4_ring_test_ring(struct amdgpu_ring *ring)
  * sdma_v2_4_ring_test_ib - test an IB on the DMA engine
  *
  * @ring: amdgpu_ring structure holding ring information
+ * @timeout: timeout value in jiffies, or MAX_SCHEDULE_TIMEOUT
  *
  * Test a simple IB in the DMA ring (VI).
  * Returns 0 on success, error on failure.
@@ -672,21 +591,18 @@ static int sdma_v2_4_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 	u64 gpu_addr;
 	long r;
 
-	r = amdgpu_wb_get(adev, &index);
-	if (r) {
-		dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r)
 		return r;
-	}
 
 	gpu_addr = adev->wb.gpu_addr + (index * 4);
 	tmp = 0xCAFEDEAD;
 	adev->wb.wb[index] = cpu_to_le32(tmp);
 	memset(&ib, 0, sizeof(ib));
-	r = amdgpu_ib_get(adev, NULL, 256, &ib);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
+	r = amdgpu_ib_get(adev, NULL, 256,
+					AMDGPU_IB_POOL_DIRECT, &ib);
+	if (r)
 		goto err0;
-	}
 
 	ib.ptr[0] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
 		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR);
@@ -705,27 +621,22 @@ static int sdma_v2_4_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 
 	r = dma_fence_wait_timeout(f, false, timeout);
 	if (r == 0) {
-		DRM_ERROR("amdgpu: IB test timed out\n");
 		r = -ETIMEDOUT;
 		goto err1;
 	} else if (r < 0) {
-		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
 		goto err1;
 	}
 	tmp = le32_to_cpu(adev->wb.wb[index]);
-	if (tmp == 0xDEADBEEF) {
-		DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
+	if (tmp == 0xDEADBEEF)
 		r = 0;
-	} else {
-		DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp);
+	else
 		r = -EINVAL;
-	}
 
 err1:
-	amdgpu_ib_free(adev, &ib, NULL);
+	amdgpu_ib_free(&ib, NULL);
 	dma_fence_put(f);
 err0:
-	amdgpu_wb_free(adev, index);
+	amdgpu_device_wb_free(adev, index);
 	return r;
 }
 
@@ -816,16 +727,17 @@ static void sdma_v2_4_vm_set_pte_pde(struct amdgpu_ib *ib, uint64_t pe,
 /**
  * sdma_v2_4_ring_pad_ib - pad the IB to the required number of dw
  *
+ * @ring: amdgpu_ring structure holding ring information
  * @ib: indirect buffer to fill with padding
  *
  */
 static void sdma_v2_4_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib)
 {
-	struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring);
+	struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring);
 	u32 pad_count;
 	int i;
 
-	pad_count = (8 - (ib->length_dw & 0x7)) % 8;
+	pad_count = (-ib->length_dw) & 7;
 	for (i = 0; i < pad_count; i++)
 		if (sdma && sdma->burst_nop && (i == 0))
 			ib->ptr[ib->length_dw++] =
@@ -856,7 +768,7 @@ static void sdma_v2_4_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
 	amdgpu_ring_write(ring, addr & 0xfffffffc);
 	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xffffffff);
 	amdgpu_ring_write(ring, seq); /* reference */
-	amdgpu_ring_write(ring, 0xfffffff); /* mask */
+	amdgpu_ring_write(ring, 0xffffffff); /* mask */
 	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
 			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(4)); /* retry count, poll interval */
 }
@@ -865,28 +777,16 @@ static void sdma_v2_4_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
  * sdma_v2_4_ring_emit_vm_flush - cik vm flush using sDMA
  *
  * @ring: amdgpu_ring pointer
- * @vm: amdgpu_vm pointer
+ * @vmid: vmid number to use
+ * @pd_addr: address
  *
  * Update the page table base and flush the VM TLB
  * using sDMA (VI).
  */
 static void sdma_v2_4_ring_emit_vm_flush(struct amdgpu_ring *ring,
-					 unsigned vm_id, uint64_t pd_addr)
+					 unsigned vmid, uint64_t pd_addr)
 {
-	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
-			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
-	if (vm_id < 8) {
-		amdgpu_ring_write(ring, (mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + vm_id));
-	} else {
-		amdgpu_ring_write(ring, (mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vm_id - 8));
-	}
-	amdgpu_ring_write(ring, pd_addr >> 12);
-
-	/* flush TLB */
-	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
-			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
-	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST);
-	amdgpu_ring_write(ring, 1 << vm_id);
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
 
 	/* wait for flush */
 	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
@@ -900,12 +800,26 @@ static void sdma_v2_4_ring_emit_vm_flush(struct amdgpu_ring *ring,
 			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
 }
 
-static int sdma_v2_4_early_init(void *handle)
+static void sdma_v2_4_ring_emit_wreg(struct amdgpu_ring *ring,
+				     uint32_t reg, uint32_t val)
+{
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
+			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, val);
+}
+
+static int sdma_v2_4_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
 
 	adev->sdma.num_instances = SDMA_MAX_INSTANCE;
 
+	r = sdma_v2_4_init_microcode(adev);
+	if (r)
+		return r;
+
 	sdma_v2_4_set_ring_funcs(adev);
 	sdma_v2_4_set_buffer_funcs(adev);
 	sdma_v2_4_set_vm_pte_funcs(adev);
@@ -914,46 +828,39 @@ static int sdma_v2_4_early_init(void *handle)
 	return 0;
 }
 
-static int sdma_v2_4_sw_init(void *handle)
+static int sdma_v2_4_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_ring *ring;
 	int r, i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	/* SDMA trap event */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 224,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SDMA_TRAP,
 			      &adev->sdma.trap_irq);
 	if (r)
 		return r;
 
 	/* SDMA Privileged inst */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 241,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 241,
 			      &adev->sdma.illegal_inst_irq);
 	if (r)
 		return r;
 
 	/* SDMA Privileged inst */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 247,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SDMA_SRBM_WRITE,
 			      &adev->sdma.illegal_inst_irq);
 	if (r)
 		return r;
 
-	r = sdma_v2_4_init_microcode(adev);
-	if (r) {
-		DRM_ERROR("Failed to load sdma firmware!\n");
-		return r;
-	}
-
 	for (i = 0; i < adev->sdma.num_instances; i++) {
 		ring = &adev->sdma.instance[i].ring;
 		ring->ring_obj = NULL;
 		ring->use_doorbell = false;
 		sprintf(ring->name, "sdma%d", i);
-		r = amdgpu_ring_init(adev, ring, 1024,
-				     &adev->sdma.trap_irq,
-				     (i == 0) ?
-				     AMDGPU_SDMA_IRQ_TRAP0 :
-				     AMDGPU_SDMA_IRQ_TRAP1);
+		r = amdgpu_ring_init(adev, ring, 1024, &adev->sdma.trap_irq,
+				     (i == 0) ? AMDGPU_SDMA_IRQ_INSTANCE0 :
+				     AMDGPU_SDMA_IRQ_INSTANCE1,
+				     AMDGPU_RING_PRIO_DEFAULT, NULL);
 		if (r)
 			return r;
 	}
@@ -961,9 +868,9 @@ static int sdma_v2_4_sw_init(void *handle)
 	return r;
 }
 
-static int sdma_v2_4_sw_fini(void *handle)
+static int sdma_v2_4_sw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int i;
 
 	for (i = 0; i < adev->sdma.num_instances; i++)
@@ -973,10 +880,10 @@ static int sdma_v2_4_sw_fini(void *handle)
 	return 0;
 }
 
-static int sdma_v2_4_hw_init(void *handle)
+static int sdma_v2_4_hw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	sdma_v2_4_init_golden_registers(adev);
 
@@ -987,32 +894,26 @@ static int sdma_v2_4_hw_init(void *handle)
 	return r;
 }
 
-static int sdma_v2_4_hw_fini(void *handle)
+static int sdma_v2_4_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	sdma_v2_4_enable(adev, false);
+	sdma_v2_4_enable(ip_block->adev, false);
 
 	return 0;
 }
 
-static int sdma_v2_4_suspend(void *handle)
+static int sdma_v2_4_suspend(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return sdma_v2_4_hw_fini(adev);
+	return sdma_v2_4_hw_fini(ip_block);
 }
 
-static int sdma_v2_4_resume(void *handle)
+static int sdma_v2_4_resume(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return sdma_v2_4_hw_init(adev);
+	return sdma_v2_4_hw_init(ip_block);
 }
 
-static bool sdma_v2_4_is_idle(void *handle)
+static bool sdma_v2_4_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 tmp = RREG32(mmSRBM_STATUS2);
 
 	if (tmp & (SRBM_STATUS2__SDMA_BUSY_MASK |
@@ -1022,11 +923,11 @@ static bool sdma_v2_4_is_idle(void *handle)
 	return true;
 }
 
-static int sdma_v2_4_wait_for_idle(void *handle)
+static int sdma_v2_4_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
 	unsigned i;
 	u32 tmp;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
 		tmp = RREG32(mmSRBM_STATUS2) & (SRBM_STATUS2__SDMA_BUSY_MASK |
@@ -1039,10 +940,10 @@ static int sdma_v2_4_wait_for_idle(void *handle)
 	return -ETIMEDOUT;
 }
 
-static int sdma_v2_4_soft_reset(void *handle)
+static int sdma_v2_4_soft_reset(struct amdgpu_ip_block *ip_block)
 {
 	u32 srbm_soft_reset = 0;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 tmp = RREG32(mmSRBM_STATUS2);
 
 	if (tmp & SRBM_STATUS2__SDMA_BUSY_MASK) {
@@ -1088,7 +989,7 @@ static int sdma_v2_4_set_trap_irq_state(struct amdgpu_device *adev,
 	u32 sdma_cntl;
 
 	switch (type) {
-	case AMDGPU_SDMA_IRQ_TRAP0:
+	case AMDGPU_SDMA_IRQ_INSTANCE0:
 		switch (state) {
 		case AMDGPU_IRQ_STATE_DISABLE:
 			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET);
@@ -1104,7 +1005,7 @@ static int sdma_v2_4_set_trap_irq_state(struct amdgpu_device *adev,
 			break;
 		}
 		break;
-	case AMDGPU_SDMA_IRQ_TRAP1:
+	case AMDGPU_SDMA_IRQ_INSTANCE1:
 		switch (state) {
 		case AMDGPU_IRQ_STATE_DISABLE:
 			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET);
@@ -1170,19 +1071,25 @@ static int sdma_v2_4_process_illegal_inst_irq(struct amdgpu_device *adev,
 					      struct amdgpu_irq_src *source,
 					      struct amdgpu_iv_entry *entry)
 {
+	u8 instance_id, queue_id;
+
 	DRM_ERROR("Illegal instruction in SDMA command stream\n");
-	schedule_work(&adev->reset_work);
+	instance_id = (entry->ring_id & 0x3) >> 0;
+	queue_id = (entry->ring_id & 0xc) >> 2;
+
+	if (instance_id <= 1 && queue_id == 0)
+		drm_sched_fault(&adev->sdma.instance[instance_id].ring.sched);
 	return 0;
 }
 
-static int sdma_v2_4_set_clockgating_state(void *handle,
+static int sdma_v2_4_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
 	/* XXX handled via the smc on VI */
 	return 0;
 }
 
-static int sdma_v2_4_set_powergating_state(void *handle,
+static int sdma_v2_4_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	return 0;
@@ -1191,7 +1098,6 @@ static int sdma_v2_4_set_powergating_state(void *handle,
 static const struct amd_ip_funcs sdma_v2_4_ip_funcs = {
 	.name = "sdma_v2_4",
 	.early_init = sdma_v2_4_early_init,
-	.late_init = NULL,
 	.sw_init = sdma_v2_4_sw_init,
 	.sw_fini = sdma_v2_4_sw_fini,
 	.hw_init = sdma_v2_4_hw_init,
@@ -1210,14 +1116,15 @@ static const struct amdgpu_ring_funcs sdma_v2_4_ring_funcs = {
 	.align_mask = 0xf,
 	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
 	.support_64bit_ptrs = false,
+	.secure_submission_supported = true,
 	.get_rptr = sdma_v2_4_ring_get_rptr,
 	.get_wptr = sdma_v2_4_ring_get_wptr,
 	.set_wptr = sdma_v2_4_ring_set_wptr,
 	.emit_frame_size =
 		6 + /* sdma_v2_4_ring_emit_hdp_flush */
-		3 + /* sdma_v2_4_ring_emit_hdp_invalidate */
+		3 + /* hdp invalidate */
 		6 + /* sdma_v2_4_ring_emit_pipeline_sync */
-		12 + /* sdma_v2_4_ring_emit_vm_flush */
+		VI_FLUSH_GPU_TLB_NUM_WREG * 3 + 6 + /* sdma_v2_4_ring_emit_vm_flush */
 		10 + 10 + 10, /* sdma_v2_4_ring_emit_fence x3 for user fence, vm fence */
 	.emit_ib_size = 7 + 6, /* sdma_v2_4_ring_emit_ib */
 	.emit_ib = sdma_v2_4_ring_emit_ib,
@@ -1225,19 +1132,21 @@ static const struct amdgpu_ring_funcs sdma_v2_4_ring_funcs = {
 	.emit_pipeline_sync = sdma_v2_4_ring_emit_pipeline_sync,
 	.emit_vm_flush = sdma_v2_4_ring_emit_vm_flush,
 	.emit_hdp_flush = sdma_v2_4_ring_emit_hdp_flush,
-	.emit_hdp_invalidate = sdma_v2_4_ring_emit_hdp_invalidate,
 	.test_ring = sdma_v2_4_ring_test_ring,
 	.test_ib = sdma_v2_4_ring_test_ib,
 	.insert_nop = sdma_v2_4_ring_insert_nop,
 	.pad_ib = sdma_v2_4_ring_pad_ib,
+	.emit_wreg = sdma_v2_4_ring_emit_wreg,
 };
 
 static void sdma_v2_4_set_ring_funcs(struct amdgpu_device *adev)
 {
 	int i;
 
-	for (i = 0; i < adev->sdma.num_instances; i++)
+	for (i = 0; i < adev->sdma.num_instances; i++) {
 		adev->sdma.instance[i].ring.funcs = &sdma_v2_4_ring_funcs;
+		adev->sdma.instance[i].ring.me = i;
+	}
 }
 
 static const struct amdgpu_irq_src_funcs sdma_v2_4_trap_irq_funcs = {
@@ -1259,10 +1168,11 @@ static void sdma_v2_4_set_irq_funcs(struct amdgpu_device *adev)
 /**
  * sdma_v2_4_emit_copy_buffer - copy buffer using the sDMA engine
  *
- * @ring: amdgpu_ring structure holding ring information
+ * @ib: indirect buffer to copy to
  * @src_offset: src GPU address
  * @dst_offset: dst GPU address
  * @byte_count: number of bytes to xfer
+ * @copy_flags: unused
  *
  * Copy GPU buffers using the DMA engine (VI).
  * Used by the amdgpu ttm implementation to move pages if
@@ -1271,7 +1181,8 @@ static void sdma_v2_4_set_irq_funcs(struct amdgpu_device *adev)
 static void sdma_v2_4_emit_copy_buffer(struct amdgpu_ib *ib,
 				       uint64_t src_offset,
 				       uint64_t dst_offset,
-				       uint32_t byte_count)
+				       uint32_t byte_count,
+				       uint32_t copy_flags)
 {
 	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
 		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
@@ -1286,7 +1197,7 @@ static void sdma_v2_4_emit_copy_buffer(struct amdgpu_ib *ib,
 /**
  * sdma_v2_4_emit_fill_buffer - fill buffer using the sDMA engine
  *
- * @ring: amdgpu_ring structure holding ring information
+ * @ib: indirect buffer to copy to
  * @src_data: value to write to buffer
  * @dst_offset: dst GPU address
  * @byte_count: number of bytes to xfer
@@ -1317,14 +1228,14 @@ static const struct amdgpu_buffer_funcs sdma_v2_4_buffer_funcs = {
 
 static void sdma_v2_4_set_buffer_funcs(struct amdgpu_device *adev)
 {
-	if (adev->mman.buffer_funcs == NULL) {
-		adev->mman.buffer_funcs = &sdma_v2_4_buffer_funcs;
-		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
-	}
+	adev->mman.buffer_funcs = &sdma_v2_4_buffer_funcs;
+	adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
 }
 
 static const struct amdgpu_vm_pte_funcs sdma_v2_4_vm_pte_funcs = {
+	.copy_pte_num_dw = 7,
 	.copy_pte = sdma_v2_4_vm_copy_pte,
+
 	.write_pte = sdma_v2_4_vm_write_pte,
 	.set_pte_pde = sdma_v2_4_vm_set_pte_pde,
 };
@@ -1333,18 +1244,15 @@ static void sdma_v2_4_set_vm_pte_funcs(struct amdgpu_device *adev)
 {
 	unsigned i;
 
-	if (adev->vm_manager.vm_pte_funcs == NULL) {
-		adev->vm_manager.vm_pte_funcs = &sdma_v2_4_vm_pte_funcs;
-		for (i = 0; i < adev->sdma.num_instances; i++)
-			adev->vm_manager.vm_pte_rings[i] =
-				&adev->sdma.instance[i].ring;
-
-		adev->vm_manager.vm_pte_num_rings = adev->sdma.num_instances;
+	adev->vm_manager.vm_pte_funcs = &sdma_v2_4_vm_pte_funcs;
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		adev->vm_manager.vm_pte_scheds[i] =
+			&adev->sdma.instance[i].ring.sched;
 	}
+	adev->vm_manager.vm_pte_num_scheds = adev->sdma.num_instances;
 }
 
-const struct amdgpu_ip_block_version sdma_v2_4_ip_block =
-{
+const struct amdgpu_ip_block_version sdma_v2_4_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_SDMA,
 	.major = 2,
 	.minor = 4,
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
index b1de44f22824..1c076bd1cf73 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
@@ -21,8 +21,11 @@
  *
  * Authors: Alex Deucher
  */
+
+#include <linux/delay.h>
 #include <linux/firmware.h>
-#include <drm/drmP.h>
+#include <linux/module.h>
+
 #include "amdgpu.h"
 #include "amdgpu_ucode.h"
 #include "amdgpu_trace.h"
@@ -44,6 +47,8 @@
 
 #include "tonga_sdma_pkt_open.h"
 
+#include "ivsrcid/ivsrcid_vislands30.h"
+
 static void sdma_v3_0_set_ring_funcs(struct amdgpu_device *adev);
 static void sdma_v3_0_set_buffer_funcs(struct amdgpu_device *adev);
 static void sdma_v3_0_set_vm_pte_funcs(struct amdgpu_device *adev);
@@ -62,6 +67,8 @@ MODULE_FIRMWARE("amdgpu/polaris11_sdma.bin");
 MODULE_FIRMWARE("amdgpu/polaris11_sdma1.bin");
 MODULE_FIRMWARE("amdgpu/polaris12_sdma.bin");
 MODULE_FIRMWARE("amdgpu/polaris12_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/vegam_sdma.bin");
+MODULE_FIRMWARE("amdgpu/vegam_sdma1.bin");
 
 
 static const u32 sdma_offsets[SDMA_MAX_INSTANCE] =
@@ -192,47 +199,48 @@ static void sdma_v3_0_init_golden_registers(struct amdgpu_device *adev)
 {
 	switch (adev->asic_type) {
 	case CHIP_FIJI:
-		amdgpu_program_register_sequence(adev,
-						 fiji_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(fiji_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_fiji_a10,
-						 (const u32)ARRAY_SIZE(golden_settings_fiji_a10));
+		amdgpu_device_program_register_sequence(adev,
+							fiji_mgcg_cgcg_init,
+							ARRAY_SIZE(fiji_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_fiji_a10,
+							ARRAY_SIZE(golden_settings_fiji_a10));
 		break;
 	case CHIP_TONGA:
-		amdgpu_program_register_sequence(adev,
-						 tonga_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(tonga_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_tonga_a11,
-						 (const u32)ARRAY_SIZE(golden_settings_tonga_a11));
+		amdgpu_device_program_register_sequence(adev,
+							tonga_mgcg_cgcg_init,
+							ARRAY_SIZE(tonga_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_tonga_a11,
+							ARRAY_SIZE(golden_settings_tonga_a11));
 		break;
 	case CHIP_POLARIS11:
 	case CHIP_POLARIS12:
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_polaris11_a11,
-						 (const u32)ARRAY_SIZE(golden_settings_polaris11_a11));
+	case CHIP_VEGAM:
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_polaris11_a11,
+							ARRAY_SIZE(golden_settings_polaris11_a11));
 		break;
 	case CHIP_POLARIS10:
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_polaris10_a11,
-						 (const u32)ARRAY_SIZE(golden_settings_polaris10_a11));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_polaris10_a11,
+							ARRAY_SIZE(golden_settings_polaris10_a11));
 		break;
 	case CHIP_CARRIZO:
-		amdgpu_program_register_sequence(adev,
-						 cz_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(cz_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 cz_golden_settings_a11,
-						 (const u32)ARRAY_SIZE(cz_golden_settings_a11));
+		amdgpu_device_program_register_sequence(adev,
+							cz_mgcg_cgcg_init,
+							ARRAY_SIZE(cz_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							cz_golden_settings_a11,
+							ARRAY_SIZE(cz_golden_settings_a11));
 		break;
 	case CHIP_STONEY:
-		amdgpu_program_register_sequence(adev,
-						 stoney_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(stoney_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 stoney_golden_settings_a11,
-						 (const u32)ARRAY_SIZE(stoney_golden_settings_a11));
+		amdgpu_device_program_register_sequence(adev,
+							stoney_mgcg_cgcg_init,
+							ARRAY_SIZE(stoney_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							stoney_golden_settings_a11,
+							ARRAY_SIZE(stoney_golden_settings_a11));
 		break;
 	default:
 		break;
@@ -242,10 +250,9 @@ static void sdma_v3_0_init_golden_registers(struct amdgpu_device *adev)
 static void sdma_v3_0_free_microcode(struct amdgpu_device *adev)
 {
 	int i;
-	for (i = 0; i < adev->sdma.num_instances; i++) {
-		release_firmware(adev->sdma.instance[i].fw);
-		adev->sdma.instance[i].fw = NULL;
-	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		amdgpu_ucode_release(&adev->sdma.instance[i].fw);
 }
 
 /**
@@ -260,7 +267,6 @@ static void sdma_v3_0_free_microcode(struct amdgpu_device *adev)
 static int sdma_v3_0_init_microcode(struct amdgpu_device *adev)
 {
 	const char *chip_name;
-	char fw_name[30];
 	int err = 0, i;
 	struct amdgpu_firmware_info *info = NULL;
 	const struct common_firmware_header *header = NULL;
@@ -275,15 +281,18 @@ static int sdma_v3_0_init_microcode(struct amdgpu_device *adev)
 	case CHIP_FIJI:
 		chip_name = "fiji";
 		break;
-	case CHIP_POLARIS11:
-		chip_name = "polaris11";
-		break;
 	case CHIP_POLARIS10:
 		chip_name = "polaris10";
 		break;
+	case CHIP_POLARIS11:
+		chip_name = "polaris11";
+		break;
 	case CHIP_POLARIS12:
 		chip_name = "polaris12";
 		break;
+	case CHIP_VEGAM:
+		chip_name = "vegam";
+		break;
 	case CHIP_CARRIZO:
 		chip_name = "carrizo";
 		break;
@@ -295,13 +304,13 @@ static int sdma_v3_0_init_microcode(struct amdgpu_device *adev)
 
 	for (i = 0; i < adev->sdma.num_instances; i++) {
 		if (i == 0)
-			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_sdma.bin", chip_name);
+			err = amdgpu_ucode_request(adev, &adev->sdma.instance[i].fw,
+						   AMDGPU_UCODE_REQUIRED,
+						   "amdgpu/%s_sdma.bin", chip_name);
 		else
-			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_sdma1.bin", chip_name);
-		err = request_firmware(&adev->sdma.instance[i].fw, fw_name, adev->dev);
-		if (err)
-			goto out;
-		err = amdgpu_ucode_validate(adev->sdma.instance[i].fw);
+			err = amdgpu_ucode_request(adev, &adev->sdma.instance[i].fw,
+						   AMDGPU_UCODE_REQUIRED,
+						   "amdgpu/%s_sdma1.bin", chip_name);
 		if (err)
 			goto out;
 		hdr = (const struct sdma_firmware_header_v1_0 *)adev->sdma.instance[i].fw->data;
@@ -310,22 +319,20 @@ static int sdma_v3_0_init_microcode(struct amdgpu_device *adev)
 		if (adev->sdma.instance[i].feature_version >= 20)
 			adev->sdma.instance[i].burst_nop = true;
 
-		if (adev->firmware.load_type == AMDGPU_FW_LOAD_SMU) {
-			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_SDMA0 + i];
-			info->ucode_id = AMDGPU_UCODE_ID_SDMA0 + i;
-			info->fw = adev->sdma.instance[i].fw;
-			header = (const struct common_firmware_header *)info->fw->data;
-			adev->firmware.fw_size +=
-				ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
-		}
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_SDMA0 + i];
+		info->ucode_id = AMDGPU_UCODE_ID_SDMA0 + i;
+		info->fw = adev->sdma.instance[i].fw;
+		header = (const struct common_firmware_header *)info->fw->data;
+		adev->firmware.fw_size +=
+			ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+
 	}
 out:
 	if (err) {
-		pr_err("sdma_v3_0: Failed to load firmware \"%s\"\n", fw_name);
-		for (i = 0; i < adev->sdma.num_instances; i++) {
-			release_firmware(adev->sdma.instance[i].fw);
-			adev->sdma.instance[i].fw = NULL;
-		}
+		pr_err("sdma_v3_0: Failed to load firmware \"%s_sdma%s.bin\"\n",
+		       chip_name, i == 0 ? "" : "1");
+		for (i = 0; i < adev->sdma.num_instances; i++)
+			amdgpu_ucode_release(&adev->sdma.instance[i].fw);
 	}
 	return err;
 }
@@ -340,7 +347,7 @@ out:
 static uint64_t sdma_v3_0_ring_get_rptr(struct amdgpu_ring *ring)
 {
 	/* XXX check if swapping is necessary on BE */
-	return ring->adev->wb.wb[ring->rptr_offs] >> 2;
+	return *ring->rptr_cpu_addr >> 2;
 }
 
 /**
@@ -355,13 +362,11 @@ static uint64_t sdma_v3_0_ring_get_wptr(struct amdgpu_ring *ring)
 	struct amdgpu_device *adev = ring->adev;
 	u32 wptr;
 
-	if (ring->use_doorbell) {
+	if (ring->use_doorbell || ring->use_pollmem) {
 		/* XXX check if swapping is necessary on BE */
-		wptr = ring->adev->wb.wb[ring->wptr_offs] >> 2;
+		wptr = *ring->wptr_cpu_addr >> 2;
 	} else {
-		int me = (ring == &ring->adev->sdma.instance[0].ring) ? 0 : 1;
-
-		wptr = RREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) >> 2;
+		wptr = RREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) >> 2;
 	}
 
 	return wptr;
@@ -379,19 +384,22 @@ static void sdma_v3_0_ring_set_wptr(struct amdgpu_ring *ring)
 	struct amdgpu_device *adev = ring->adev;
 
 	if (ring->use_doorbell) {
+		u32 *wb = (u32 *)ring->wptr_cpu_addr;
 		/* XXX check if swapping is necessary on BE */
-		adev->wb.wb[ring->wptr_offs] = lower_32_bits(ring->wptr) << 2;
-		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr) << 2);
-	} else {
-		int me = (ring == &ring->adev->sdma.instance[0].ring) ? 0 : 1;
+		WRITE_ONCE(*wb, ring->wptr << 2);
+		WDOORBELL32(ring->doorbell_index, ring->wptr << 2);
+	} else if (ring->use_pollmem) {
+		u32 *wb = (u32 *)ring->wptr_cpu_addr;
 
-		WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[me], lower_32_bits(ring->wptr) << 2);
+		WRITE_ONCE(*wb, ring->wptr << 2);
+	} else {
+		WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me], ring->wptr << 2);
 	}
 }
 
 static void sdma_v3_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
 {
-	struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring);
+	struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring);
 	int i;
 
 	for (i = 0; i < count; i++)
@@ -406,21 +414,24 @@ static void sdma_v3_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
  * sdma_v3_0_ring_emit_ib - Schedule an IB on the DMA engine
  *
  * @ring: amdgpu ring pointer
+ * @job: job to retrieve vmid from
  * @ib: IB object to schedule
+ * @flags: unused
  *
  * Schedule an IB in the DMA ring (VI).
  */
 static void sdma_v3_0_ring_emit_ib(struct amdgpu_ring *ring,
+				   struct amdgpu_job *job,
 				   struct amdgpu_ib *ib,
-				   unsigned vm_id, bool ctx_switch)
+				   uint32_t flags)
 {
-	u32 vmid = vm_id & 0xf;
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
 
 	/* IB packet must end on a 8 DW boundary */
-	sdma_v3_0_ring_insert_nop(ring, (10 - (lower_32_bits(ring->wptr) & 7)) % 8);
+	sdma_v3_0_ring_insert_nop(ring, (2 - lower_32_bits(ring->wptr)) & 7);
 
 	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_INDIRECT) |
-			  SDMA_PKT_INDIRECT_HEADER_VMID(vmid));
+			  SDMA_PKT_INDIRECT_HEADER_VMID(vmid & 0xf));
 	/* base must be 32 byte aligned */
 	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr) & 0xffffffe0);
 	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
@@ -441,7 +452,7 @@ static void sdma_v3_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
 {
 	u32 ref_and_mask = 0;
 
-	if (ring == &ring->adev->sdma.instance[0].ring)
+	if (ring->me == 0)
 		ref_and_mask = REG_SET_FIELD(ref_and_mask, GPU_HDP_FLUSH_DONE, SDMA0, 1);
 	else
 		ref_and_mask = REG_SET_FIELD(ref_and_mask, GPU_HDP_FLUSH_DONE, SDMA1, 1);
@@ -457,19 +468,13 @@ static void sdma_v3_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
 			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
 }
 
-static void sdma_v3_0_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
-{
-	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
-			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
-	amdgpu_ring_write(ring, mmHDP_DEBUG0);
-	amdgpu_ring_write(ring, 1);
-}
-
 /**
  * sdma_v3_0_ring_emit_fence - emit a fence on the DMA ring
  *
  * @ring: amdgpu ring pointer
- * @fence: amdgpu fence object
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
  *
  * Add a DMA fence packet to the ring to write
  * the fence seq number and DMA trap packet to generate
@@ -508,15 +513,9 @@ static void sdma_v3_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 se
  */
 static void sdma_v3_0_gfx_stop(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *sdma0 = &adev->sdma.instance[0].ring;
-	struct amdgpu_ring *sdma1 = &adev->sdma.instance[1].ring;
 	u32 rb_cntl, ib_cntl;
 	int i;
 
-	if ((adev->mman.buffer_funcs_ring == sdma0) ||
-	    (adev->mman.buffer_funcs_ring == sdma1))
-		amdgpu_ttm_set_active_vram_size(adev, adev->mc.visible_vram_size);
-
 	for (i = 0; i < adev->sdma.num_instances; i++) {
 		rb_cntl = RREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i]);
 		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 0);
@@ -525,8 +524,6 @@ static void sdma_v3_0_gfx_stop(struct amdgpu_device *adev)
 		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 0);
 		WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl);
 	}
-	sdma0->ready = false;
-	sdma1->ready = false;
 }
 
 /**
@@ -641,16 +638,15 @@ static void sdma_v3_0_enable(struct amdgpu_device *adev, bool enable)
 static int sdma_v3_0_gfx_resume(struct amdgpu_device *adev)
 {
 	struct amdgpu_ring *ring;
-	u32 rb_cntl, ib_cntl;
+	u32 rb_cntl, ib_cntl, wptr_poll_cntl;
 	u32 rb_bufsz;
-	u32 wb_offset;
 	u32 doorbell;
+	u64 wptr_gpu_addr;
 	int i, j, r;
 
 	for (i = 0; i < adev->sdma.num_instances; i++) {
 		ring = &adev->sdma.instance[i].ring;
 		amdgpu_ring_clear_ring(ring);
-		wb_offset = (ring->rptr_offs * 4);
 
 		mutex_lock(&adev->srbm_mutex);
 		for (j = 0; j < 16; j++) {
@@ -687,9 +683,9 @@ static int sdma_v3_0_gfx_resume(struct amdgpu_device *adev)
 
 		/* set the wb address whether it's enabled or not */
 		WREG32(mmSDMA0_GFX_RB_RPTR_ADDR_HI + sdma_offsets[i],
-		       upper_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFF);
+		       upper_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFF);
 		WREG32(mmSDMA0_GFX_RB_RPTR_ADDR_LO + sdma_offsets[i],
-		       lower_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFC);
+		       lower_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFC);
 
 		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RPTR_WRITEBACK_ENABLE, 1);
 
@@ -707,6 +703,27 @@ static int sdma_v3_0_gfx_resume(struct amdgpu_device *adev)
 		}
 		WREG32(mmSDMA0_GFX_DOORBELL + sdma_offsets[i], doorbell);
 
+		/* setup the wptr shadow polling */
+		wptr_gpu_addr = ring->wptr_gpu_addr;
+
+		WREG32(mmSDMA0_GFX_RB_WPTR_POLL_ADDR_LO + sdma_offsets[i],
+		       lower_32_bits(wptr_gpu_addr));
+		WREG32(mmSDMA0_GFX_RB_WPTR_POLL_ADDR_HI + sdma_offsets[i],
+		       upper_32_bits(wptr_gpu_addr));
+		wptr_poll_cntl = RREG32(mmSDMA0_GFX_RB_WPTR_POLL_CNTL + sdma_offsets[i]);
+		if (ring->use_pollmem) {
+			/*wptr polling is not enough fast, directly clean the wptr register */
+			WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[i], 0);
+			wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl,
+						       SDMA0_GFX_RB_WPTR_POLL_CNTL,
+						       ENABLE, 1);
+		} else {
+			wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl,
+						       SDMA0_GFX_RB_WPTR_POLL_CNTL,
+						       ENABLE, 0);
+		}
+		WREG32(mmSDMA0_GFX_RB_WPTR_POLL_CNTL + sdma_offsets[i], wptr_poll_cntl);
+
 		/* enable DMA RB */
 		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 1);
 		WREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i], rb_cntl);
@@ -718,8 +735,6 @@ static int sdma_v3_0_gfx_resume(struct amdgpu_device *adev)
 #endif
 		/* enable DMA IBs */
 		WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl);
-
-		ring->ready = true;
 	}
 
 	/* unhalt the MEs */
@@ -729,14 +744,9 @@ static int sdma_v3_0_gfx_resume(struct amdgpu_device *adev)
 
 	for (i = 0; i < adev->sdma.num_instances; i++) {
 		ring = &adev->sdma.instance[i].ring;
-		r = amdgpu_ring_test_ring(ring);
-		if (r) {
-			ring->ready = false;
+		r = amdgpu_ring_test_helper(ring);
+		if (r)
 			return r;
-		}
-
-		if (adev->mman.buffer_funcs_ring == ring)
-			amdgpu_ttm_set_active_vram_size(adev, adev->mc.real_vram_size);
 	}
 
 	return 0;
@@ -757,42 +767,6 @@ static int sdma_v3_0_rlc_resume(struct amdgpu_device *adev)
 }
 
 /**
- * sdma_v3_0_load_microcode - load the sDMA ME ucode
- *
- * @adev: amdgpu_device pointer
- *
- * Loads the sDMA0/1 ucode.
- * Returns 0 for success, -EINVAL if the ucode is not available.
- */
-static int sdma_v3_0_load_microcode(struct amdgpu_device *adev)
-{
-	const struct sdma_firmware_header_v1_0 *hdr;
-	const __le32 *fw_data;
-	u32 fw_size;
-	int i, j;
-
-	/* halt the MEs */
-	sdma_v3_0_enable(adev, false);
-
-	for (i = 0; i < adev->sdma.num_instances; i++) {
-		if (!adev->sdma.instance[i].fw)
-			return -EINVAL;
-		hdr = (const struct sdma_firmware_header_v1_0 *)adev->sdma.instance[i].fw->data;
-		amdgpu_ucode_print_sdma_hdr(&hdr->header);
-		fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
-		fw_data = (const __le32 *)
-			(adev->sdma.instance[i].fw->data +
-				le32_to_cpu(hdr->header.ucode_array_offset_bytes));
-		WREG32(mmSDMA0_UCODE_ADDR + sdma_offsets[i], 0);
-		for (j = 0; j < fw_size; j++)
-			WREG32(mmSDMA0_UCODE_DATA + sdma_offsets[i], le32_to_cpup(fw_data++));
-		WREG32(mmSDMA0_UCODE_ADDR + sdma_offsets[i], adev->sdma.instance[i].fw_version);
-	}
-
-	return 0;
-}
-
-/**
  * sdma_v3_0_start - setup and start the async dma engines
  *
  * @adev: amdgpu_device pointer
@@ -802,24 +776,7 @@ static int sdma_v3_0_load_microcode(struct amdgpu_device *adev)
  */
 static int sdma_v3_0_start(struct amdgpu_device *adev)
 {
-	int r, i;
-
-	if (!adev->pp_enabled) {
-		if (adev->firmware.load_type != AMDGPU_FW_LOAD_SMU) {
-			r = sdma_v3_0_load_microcode(adev);
-			if (r)
-				return r;
-		} else {
-			for (i = 0; i < adev->sdma.num_instances; i++) {
-				r = adev->smu.smumgr_funcs->check_fw_load_finish(adev,
-										 (i == 0) ?
-										 AMDGPU_UCODE_ID_SDMA0 :
-										 AMDGPU_UCODE_ID_SDMA1);
-				if (r)
-					return -EINVAL;
-			}
-		}
-	}
+	int r;
 
 	/* disable sdma engine before programing it */
 	sdma_v3_0_ctx_switch_enable(adev, false);
@@ -854,22 +811,17 @@ static int sdma_v3_0_ring_test_ring(struct amdgpu_ring *ring)
 	u32 tmp;
 	u64 gpu_addr;
 
-	r = amdgpu_wb_get(adev, &index);
-	if (r) {
-		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r)
 		return r;
-	}
 
 	gpu_addr = adev->wb.gpu_addr + (index * 4);
 	tmp = 0xCAFEDEAD;
 	adev->wb.wb[index] = cpu_to_le32(tmp);
 
 	r = amdgpu_ring_alloc(ring, 5);
-	if (r) {
-		DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r);
-		amdgpu_wb_free(adev, index);
-		return r;
-	}
+	if (r)
+		goto error_free_wb;
 
 	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
 			  SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR));
@@ -883,18 +835,14 @@ static int sdma_v3_0_ring_test_ring(struct amdgpu_ring *ring)
 		tmp = le32_to_cpu(adev->wb.wb[index]);
 		if (tmp == 0xDEADBEEF)
 			break;
-		DRM_UDELAY(1);
+		udelay(1);
 	}
 
-	if (i < adev->usec_timeout) {
-		DRM_INFO("ring test on %d succeeded in %d usecs\n", ring->idx, i);
-	} else {
-		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
-			  ring->idx, tmp);
-		r = -EINVAL;
-	}
-	amdgpu_wb_free(adev, index);
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
 
+error_free_wb:
+	amdgpu_device_wb_free(adev, index);
 	return r;
 }
 
@@ -902,6 +850,7 @@ static int sdma_v3_0_ring_test_ring(struct amdgpu_ring *ring)
  * sdma_v3_0_ring_test_ib - test an IB on the DMA engine
  *
  * @ring: amdgpu_ring structure holding ring information
+ * @timeout: timeout value in jiffies, or MAX_SCHEDULE_TIMEOUT
  *
  * Test a simple IB in the DMA ring (VI).
  * Returns 0 on success, error on failure.
@@ -916,21 +865,18 @@ static int sdma_v3_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 	u64 gpu_addr;
 	long r;
 
-	r = amdgpu_wb_get(adev, &index);
-	if (r) {
-		dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r)
 		return r;
-	}
 
 	gpu_addr = adev->wb.gpu_addr + (index * 4);
 	tmp = 0xCAFEDEAD;
 	adev->wb.wb[index] = cpu_to_le32(tmp);
 	memset(&ib, 0, sizeof(ib));
-	r = amdgpu_ib_get(adev, NULL, 256, &ib);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
+	r = amdgpu_ib_get(adev, NULL, 256,
+					AMDGPU_IB_POOL_DIRECT, &ib);
+	if (r)
 		goto err0;
-	}
 
 	ib.ptr[0] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
 		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR);
@@ -949,26 +895,21 @@ static int sdma_v3_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 
 	r = dma_fence_wait_timeout(f, false, timeout);
 	if (r == 0) {
-		DRM_ERROR("amdgpu: IB test timed out\n");
 		r = -ETIMEDOUT;
 		goto err1;
 	} else if (r < 0) {
-		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
 		goto err1;
 	}
 	tmp = le32_to_cpu(adev->wb.wb[index]);
-	if (tmp == 0xDEADBEEF) {
-		DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
+	if (tmp == 0xDEADBEEF)
 		r = 0;
-	} else {
-		DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp);
+	else
 		r = -EINVAL;
-	}
 err1:
-	amdgpu_ib_free(adev, &ib, NULL);
+	amdgpu_ib_free(&ib, NULL);
 	dma_fence_put(f);
 err0:
-	amdgpu_wb_free(adev, index);
+	amdgpu_device_wb_free(adev, index);
 	return r;
 }
 
@@ -1059,16 +1000,17 @@ static void sdma_v3_0_vm_set_pte_pde(struct amdgpu_ib *ib, uint64_t pe,
 /**
  * sdma_v3_0_ring_pad_ib - pad the IB to the required number of dw
  *
+ * @ring: amdgpu_ring structure holding ring information
  * @ib: indirect buffer to fill with padding
  *
  */
 static void sdma_v3_0_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib)
 {
-	struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring);
+	struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring);
 	u32 pad_count;
 	int i;
 
-	pad_count = (8 - (ib->length_dw & 0x7)) % 8;
+	pad_count = (-ib->length_dw) & 7;
 	for (i = 0; i < pad_count; i++)
 		if (sdma && sdma->burst_nop && (i == 0))
 			ib->ptr[ib->length_dw++] =
@@ -1099,7 +1041,7 @@ static void sdma_v3_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
 	amdgpu_ring_write(ring, addr & 0xfffffffc);
 	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xffffffff);
 	amdgpu_ring_write(ring, seq); /* reference */
-	amdgpu_ring_write(ring, 0xfffffff); /* mask */
+	amdgpu_ring_write(ring, 0xffffffff); /* mask */
 	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
 			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(4)); /* retry count, poll interval */
 }
@@ -1108,28 +1050,16 @@ static void sdma_v3_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
  * sdma_v3_0_ring_emit_vm_flush - cik vm flush using sDMA
  *
  * @ring: amdgpu_ring pointer
- * @vm: amdgpu_vm pointer
+ * @vmid: vmid number to use
+ * @pd_addr: address
  *
  * Update the page table base and flush the VM TLB
  * using sDMA (VI).
  */
 static void sdma_v3_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
-					 unsigned vm_id, uint64_t pd_addr)
+					 unsigned vmid, uint64_t pd_addr)
 {
-	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
-			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
-	if (vm_id < 8) {
-		amdgpu_ring_write(ring, (mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + vm_id));
-	} else {
-		amdgpu_ring_write(ring, (mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vm_id - 8));
-	}
-	amdgpu_ring_write(ring, pd_addr >> 12);
-
-	/* flush TLB */
-	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
-			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
-	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST);
-	amdgpu_ring_write(ring, 1 << vm_id);
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
 
 	/* wait for flush */
 	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
@@ -1143,9 +1073,19 @@ static void sdma_v3_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
 			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
 }
 
-static int sdma_v3_0_early_init(void *handle)
+static void sdma_v3_0_ring_emit_wreg(struct amdgpu_ring *ring,
+				     uint32_t reg, uint32_t val)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
+			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, val);
+}
+
+static int sdma_v3_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
 
 	switch (adev->asic_type) {
 	case CHIP_STONEY:
@@ -1156,6 +1096,10 @@ static int sdma_v3_0_early_init(void *handle)
 		break;
 	}
 
+	r = sdma_v3_0_init_microcode(adev);
+	if (r)
+		return r;
+
 	sdma_v3_0_set_ring_funcs(adev);
 	sdma_v3_0_set_buffer_funcs(adev);
 	sdma_v3_0_set_vm_pte_funcs(adev);
@@ -1164,49 +1108,45 @@ static int sdma_v3_0_early_init(void *handle)
 	return 0;
 }
 
-static int sdma_v3_0_sw_init(void *handle)
+static int sdma_v3_0_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_ring *ring;
 	int r, i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	/* SDMA trap event */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 224,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SDMA_TRAP,
 			      &adev->sdma.trap_irq);
 	if (r)
 		return r;
 
 	/* SDMA Privileged inst */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 241,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 241,
 			      &adev->sdma.illegal_inst_irq);
 	if (r)
 		return r;
 
 	/* SDMA Privileged inst */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 247,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SDMA_SRBM_WRITE,
 			      &adev->sdma.illegal_inst_irq);
 	if (r)
 		return r;
 
-	r = sdma_v3_0_init_microcode(adev);
-	if (r) {
-		DRM_ERROR("Failed to load sdma firmware!\n");
-		return r;
-	}
-
 	for (i = 0; i < adev->sdma.num_instances; i++) {
 		ring = &adev->sdma.instance[i].ring;
 		ring->ring_obj = NULL;
-		ring->use_doorbell = true;
-		ring->doorbell_index = (i == 0) ?
-			AMDGPU_DOORBELL_sDMA_ENGINE0 : AMDGPU_DOORBELL_sDMA_ENGINE1;
+		if (!amdgpu_sriov_vf(adev)) {
+			ring->use_doorbell = true;
+			ring->doorbell_index = adev->doorbell_index.sdma_engine[i];
+		} else {
+			ring->use_pollmem = true;
+		}
 
 		sprintf(ring->name, "sdma%d", i);
-		r = amdgpu_ring_init(adev, ring, 1024,
-				     &adev->sdma.trap_irq,
-				     (i == 0) ?
-				     AMDGPU_SDMA_IRQ_TRAP0 :
-				     AMDGPU_SDMA_IRQ_TRAP1);
+		r = amdgpu_ring_init(adev, ring, 1024, &adev->sdma.trap_irq,
+				     (i == 0) ? AMDGPU_SDMA_IRQ_INSTANCE0 :
+				     AMDGPU_SDMA_IRQ_INSTANCE1,
+				     AMDGPU_RING_PRIO_DEFAULT, NULL);
 		if (r)
 			return r;
 	}
@@ -1214,9 +1154,9 @@ static int sdma_v3_0_sw_init(void *handle)
 	return r;
 }
 
-static int sdma_v3_0_sw_fini(void *handle)
+static int sdma_v3_0_sw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int i;
 
 	for (i = 0; i < adev->sdma.num_instances; i++)
@@ -1226,10 +1166,10 @@ static int sdma_v3_0_sw_fini(void *handle)
 	return 0;
 }
 
-static int sdma_v3_0_hw_init(void *handle)
+static int sdma_v3_0_hw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	sdma_v3_0_init_golden_registers(adev);
 
@@ -1240,9 +1180,9 @@ static int sdma_v3_0_hw_init(void *handle)
 	return r;
 }
 
-static int sdma_v3_0_hw_fini(void *handle)
+static int sdma_v3_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	sdma_v3_0_ctx_switch_enable(adev, false);
 	sdma_v3_0_enable(adev, false);
@@ -1250,23 +1190,19 @@ static int sdma_v3_0_hw_fini(void *handle)
 	return 0;
 }
 
-static int sdma_v3_0_suspend(void *handle)
+static int sdma_v3_0_suspend(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return sdma_v3_0_hw_fini(adev);
+	return sdma_v3_0_hw_fini(ip_block);
 }
 
-static int sdma_v3_0_resume(void *handle)
+static int sdma_v3_0_resume(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return sdma_v3_0_hw_init(adev);
+	return sdma_v3_0_hw_init(ip_block);
 }
 
-static bool sdma_v3_0_is_idle(void *handle)
+static bool sdma_v3_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 tmp = RREG32(mmSRBM_STATUS2);
 
 	if (tmp & (SRBM_STATUS2__SDMA_BUSY_MASK |
@@ -1276,11 +1212,11 @@ static bool sdma_v3_0_is_idle(void *handle)
 	return true;
 }
 
-static int sdma_v3_0_wait_for_idle(void *handle)
+static int sdma_v3_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
 	unsigned i;
 	u32 tmp;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
 		tmp = RREG32(mmSRBM_STATUS2) & (SRBM_STATUS2__SDMA_BUSY_MASK |
@@ -1293,9 +1229,9 @@ static int sdma_v3_0_wait_for_idle(void *handle)
 	return -ETIMEDOUT;
 }
 
-static bool sdma_v3_0_check_soft_reset(void *handle)
+static bool sdma_v3_0_check_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 srbm_soft_reset = 0;
 	u32 tmp = RREG32(mmSRBM_STATUS2);
 
@@ -1314,9 +1250,9 @@ static bool sdma_v3_0_check_soft_reset(void *handle)
 	}
 }
 
-static int sdma_v3_0_pre_soft_reset(void *handle)
+static int sdma_v3_0_pre_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 srbm_soft_reset = 0;
 
 	if (!adev->sdma.srbm_soft_reset)
@@ -1333,9 +1269,9 @@ static int sdma_v3_0_pre_soft_reset(void *handle)
 	return 0;
 }
 
-static int sdma_v3_0_post_soft_reset(void *handle)
+static int sdma_v3_0_post_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 srbm_soft_reset = 0;
 
 	if (!adev->sdma.srbm_soft_reset)
@@ -1352,9 +1288,9 @@ static int sdma_v3_0_post_soft_reset(void *handle)
 	return 0;
 }
 
-static int sdma_v3_0_soft_reset(void *handle)
+static int sdma_v3_0_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 srbm_soft_reset = 0;
 	u32 tmp;
 
@@ -1391,7 +1327,7 @@ static int sdma_v3_0_set_trap_irq_state(struct amdgpu_device *adev,
 	u32 sdma_cntl;
 
 	switch (type) {
-	case AMDGPU_SDMA_IRQ_TRAP0:
+	case AMDGPU_SDMA_IRQ_INSTANCE0:
 		switch (state) {
 		case AMDGPU_IRQ_STATE_DISABLE:
 			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET);
@@ -1407,7 +1343,7 @@ static int sdma_v3_0_set_trap_irq_state(struct amdgpu_device *adev,
 			break;
 		}
 		break;
-	case AMDGPU_SDMA_IRQ_TRAP1:
+	case AMDGPU_SDMA_IRQ_INSTANCE1:
 		switch (state) {
 		case AMDGPU_IRQ_STATE_DISABLE:
 			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET);
@@ -1473,8 +1409,14 @@ static int sdma_v3_0_process_illegal_inst_irq(struct amdgpu_device *adev,
 					      struct amdgpu_irq_src *source,
 					      struct amdgpu_iv_entry *entry)
 {
+	u8 instance_id, queue_id;
+
 	DRM_ERROR("Illegal instruction in SDMA command stream\n");
-	schedule_work(&adev->reset_work);
+	instance_id = (entry->ring_id & 0x3) >> 0;
+	queue_id = (entry->ring_id & 0xc) >> 2;
+
+	if (instance_id <= 1 && queue_id == 0)
+		drm_sched_fault(&adev->sdma.instance[instance_id].ring.sched);
 	return 0;
 }
 
@@ -1543,10 +1485,10 @@ static void sdma_v3_0_update_sdma_medium_grain_light_sleep(
 	}
 }
 
-static int sdma_v3_0_set_clockgating_state(void *handle,
+static int sdma_v3_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (amdgpu_sriov_vf(adev))
 		return 0;
@@ -1566,15 +1508,15 @@ static int sdma_v3_0_set_clockgating_state(void *handle,
 	return 0;
 }
 
-static int sdma_v3_0_set_powergating_state(void *handle,
+static int sdma_v3_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	return 0;
 }
 
-static void sdma_v3_0_get_clockgating_state(void *handle, u32 *flags)
+static void sdma_v3_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int data;
 
 	if (amdgpu_sriov_vf(adev))
@@ -1594,7 +1536,6 @@ static void sdma_v3_0_get_clockgating_state(void *handle, u32 *flags)
 static const struct amd_ip_funcs sdma_v3_0_ip_funcs = {
 	.name = "sdma_v3_0",
 	.early_init = sdma_v3_0_early_init,
-	.late_init = NULL,
 	.sw_init = sdma_v3_0_sw_init,
 	.sw_fini = sdma_v3_0_sw_fini,
 	.hw_init = sdma_v3_0_hw_init,
@@ -1617,14 +1558,15 @@ static const struct amdgpu_ring_funcs sdma_v3_0_ring_funcs = {
 	.align_mask = 0xf,
 	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
 	.support_64bit_ptrs = false,
+	.secure_submission_supported = true,
 	.get_rptr = sdma_v3_0_ring_get_rptr,
 	.get_wptr = sdma_v3_0_ring_get_wptr,
 	.set_wptr = sdma_v3_0_ring_set_wptr,
 	.emit_frame_size =
 		6 + /* sdma_v3_0_ring_emit_hdp_flush */
-		3 + /* sdma_v3_0_ring_emit_hdp_invalidate */
+		3 + /* hdp invalidate */
 		6 + /* sdma_v3_0_ring_emit_pipeline_sync */
-		12 + /* sdma_v3_0_ring_emit_vm_flush */
+		VI_FLUSH_GPU_TLB_NUM_WREG * 3 + 6 + /* sdma_v3_0_ring_emit_vm_flush */
 		10 + 10 + 10, /* sdma_v3_0_ring_emit_fence x3 for user fence, vm fence */
 	.emit_ib_size = 7 + 6, /* sdma_v3_0_ring_emit_ib */
 	.emit_ib = sdma_v3_0_ring_emit_ib,
@@ -1632,19 +1574,21 @@ static const struct amdgpu_ring_funcs sdma_v3_0_ring_funcs = {
 	.emit_pipeline_sync = sdma_v3_0_ring_emit_pipeline_sync,
 	.emit_vm_flush = sdma_v3_0_ring_emit_vm_flush,
 	.emit_hdp_flush = sdma_v3_0_ring_emit_hdp_flush,
-	.emit_hdp_invalidate = sdma_v3_0_ring_emit_hdp_invalidate,
 	.test_ring = sdma_v3_0_ring_test_ring,
 	.test_ib = sdma_v3_0_ring_test_ib,
 	.insert_nop = sdma_v3_0_ring_insert_nop,
 	.pad_ib = sdma_v3_0_ring_pad_ib,
+	.emit_wreg = sdma_v3_0_ring_emit_wreg,
 };
 
 static void sdma_v3_0_set_ring_funcs(struct amdgpu_device *adev)
 {
 	int i;
 
-	for (i = 0; i < adev->sdma.num_instances; i++)
+	for (i = 0; i < adev->sdma.num_instances; i++) {
 		adev->sdma.instance[i].ring.funcs = &sdma_v3_0_ring_funcs;
+		adev->sdma.instance[i].ring.me = i;
+	}
 }
 
 static const struct amdgpu_irq_src_funcs sdma_v3_0_trap_irq_funcs = {
@@ -1666,10 +1610,11 @@ static void sdma_v3_0_set_irq_funcs(struct amdgpu_device *adev)
 /**
  * sdma_v3_0_emit_copy_buffer - copy buffer using the sDMA engine
  *
- * @ring: amdgpu_ring structure holding ring information
+ * @ib: indirect buffer to copy to
  * @src_offset: src GPU address
  * @dst_offset: dst GPU address
  * @byte_count: number of bytes to xfer
+ * @copy_flags: unused
  *
  * Copy GPU buffers using the DMA engine (VI).
  * Used by the amdgpu ttm implementation to move pages if
@@ -1678,7 +1623,8 @@ static void sdma_v3_0_set_irq_funcs(struct amdgpu_device *adev)
 static void sdma_v3_0_emit_copy_buffer(struct amdgpu_ib *ib,
 				       uint64_t src_offset,
 				       uint64_t dst_offset,
-				       uint32_t byte_count)
+				       uint32_t byte_count,
+				       uint32_t copy_flags)
 {
 	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
 		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
@@ -1693,7 +1639,7 @@ static void sdma_v3_0_emit_copy_buffer(struct amdgpu_ib *ib,
 /**
  * sdma_v3_0_emit_fill_buffer - fill buffer using the sDMA engine
  *
- * @ring: amdgpu_ring structure holding ring information
+ * @ib: indirect buffer to copy to
  * @src_data: value to write to buffer
  * @dst_offset: dst GPU address
  * @byte_count: number of bytes to xfer
@@ -1713,25 +1659,25 @@ static void sdma_v3_0_emit_fill_buffer(struct amdgpu_ib *ib,
 }
 
 static const struct amdgpu_buffer_funcs sdma_v3_0_buffer_funcs = {
-	.copy_max_bytes = 0x1fffff,
+	.copy_max_bytes = 0x3fffe0, /* not 0x3fffff due to HW limitation */
 	.copy_num_dw = 7,
 	.emit_copy_buffer = sdma_v3_0_emit_copy_buffer,
 
-	.fill_max_bytes = 0x1fffff,
+	.fill_max_bytes = 0x3fffe0, /* not 0x3fffff due to HW limitation */
 	.fill_num_dw = 5,
 	.emit_fill_buffer = sdma_v3_0_emit_fill_buffer,
 };
 
 static void sdma_v3_0_set_buffer_funcs(struct amdgpu_device *adev)
 {
-	if (adev->mman.buffer_funcs == NULL) {
-		adev->mman.buffer_funcs = &sdma_v3_0_buffer_funcs;
-		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
-	}
+	adev->mman.buffer_funcs = &sdma_v3_0_buffer_funcs;
+	adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
 }
 
 static const struct amdgpu_vm_pte_funcs sdma_v3_0_vm_pte_funcs = {
+	.copy_pte_num_dw = 7,
 	.copy_pte = sdma_v3_0_vm_copy_pte,
+
 	.write_pte = sdma_v3_0_vm_write_pte,
 	.set_pte_pde = sdma_v3_0_vm_set_pte_pde,
 };
@@ -1740,14 +1686,12 @@ static void sdma_v3_0_set_vm_pte_funcs(struct amdgpu_device *adev)
 {
 	unsigned i;
 
-	if (adev->vm_manager.vm_pte_funcs == NULL) {
-		adev->vm_manager.vm_pte_funcs = &sdma_v3_0_vm_pte_funcs;
-		for (i = 0; i < adev->sdma.num_instances; i++)
-			adev->vm_manager.vm_pte_rings[i] =
-				&adev->sdma.instance[i].ring;
-
-		adev->vm_manager.vm_pte_num_rings = adev->sdma.num_instances;
+	adev->vm_manager.vm_pte_funcs = &sdma_v3_0_vm_pte_funcs;
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		adev->vm_manager.vm_pte_scheds[i] =
+			 &adev->sdma.instance[i].ring.sched;
 	}
+	adev->vm_manager.vm_pte_num_scheds = adev->sdma.num_instances;
 }
 
 const struct amdgpu_ip_block_version sdma_v3_0_ip_block =
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
index fd7c72aaafa6..f38004e6064e 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
@@ -21,135 +21,593 @@
  *
  */
 
+#include <linux/delay.h>
 #include <linux/firmware.h>
-#include <drm/drmP.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+
 #include "amdgpu.h"
 #include "amdgpu_ucode.h"
 #include "amdgpu_trace.h"
 
-#include "vega10/soc15ip.h"
-#include "vega10/SDMA0/sdma0_4_0_offset.h"
-#include "vega10/SDMA0/sdma0_4_0_sh_mask.h"
-#include "vega10/SDMA1/sdma1_4_0_offset.h"
-#include "vega10/SDMA1/sdma1_4_0_sh_mask.h"
-#include "vega10/MMHUB/mmhub_1_0_offset.h"
-#include "vega10/MMHUB/mmhub_1_0_sh_mask.h"
-#include "vega10/HDP/hdp_4_0_offset.h"
-#include "raven1/SDMA0/sdma0_4_1_default.h"
+#include "sdma0/sdma0_4_2_offset.h"
+#include "sdma0/sdma0_4_2_sh_mask.h"
+#include "sdma1/sdma1_4_2_offset.h"
+#include "sdma1/sdma1_4_2_sh_mask.h"
+#include "sdma2/sdma2_4_2_2_offset.h"
+#include "sdma2/sdma2_4_2_2_sh_mask.h"
+#include "sdma3/sdma3_4_2_2_offset.h"
+#include "sdma3/sdma3_4_2_2_sh_mask.h"
+#include "sdma4/sdma4_4_2_2_offset.h"
+#include "sdma4/sdma4_4_2_2_sh_mask.h"
+#include "sdma5/sdma5_4_2_2_offset.h"
+#include "sdma5/sdma5_4_2_2_sh_mask.h"
+#include "sdma6/sdma6_4_2_2_offset.h"
+#include "sdma6/sdma6_4_2_2_sh_mask.h"
+#include "sdma7/sdma7_4_2_2_offset.h"
+#include "sdma7/sdma7_4_2_2_sh_mask.h"
+#include "sdma0/sdma0_4_1_default.h"
 
 #include "soc15_common.h"
 #include "soc15.h"
 #include "vega10_sdma_pkt_open.h"
 
+#include "ivsrcid/sdma0/irqsrcs_sdma0_4_0.h"
+#include "ivsrcid/sdma1/irqsrcs_sdma1_4_0.h"
+
+#include "amdgpu_ras.h"
+#include "sdma_v4_4.h"
+
 MODULE_FIRMWARE("amdgpu/vega10_sdma.bin");
 MODULE_FIRMWARE("amdgpu/vega10_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/vega12_sdma.bin");
+MODULE_FIRMWARE("amdgpu/vega12_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/vega20_sdma.bin");
+MODULE_FIRMWARE("amdgpu/vega20_sdma1.bin");
 MODULE_FIRMWARE("amdgpu/raven_sdma.bin");
+MODULE_FIRMWARE("amdgpu/picasso_sdma.bin");
+MODULE_FIRMWARE("amdgpu/raven2_sdma.bin");
+MODULE_FIRMWARE("amdgpu/arcturus_sdma.bin");
+MODULE_FIRMWARE("amdgpu/renoir_sdma.bin");
+MODULE_FIRMWARE("amdgpu/green_sardine_sdma.bin");
+MODULE_FIRMWARE("amdgpu/aldebaran_sdma.bin");
+
+static const struct amdgpu_hwip_reg_entry sdma_reg_list_4_0[] = {
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_STATUS_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_STATUS1_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_STATUS2_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_STATUS3_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_UCODE_CHECKSUM),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RB_RPTR_FETCH_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RB_RPTR_FETCH),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_UTCL1_RD_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_UTCL1_WR_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_UTCL1_RD_XNACK0),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_UTCL1_RD_XNACK1),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_UTCL1_WR_XNACK0),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_UTCL1_WR_XNACK1),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_RB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_RB_RPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_RB_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_RB_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_IB_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_IB_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_IB_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_IB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_IB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_IB_SUB_REMAIN),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_DUMMY_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_RB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_RB_RPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_RB_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_RB_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_IB_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_IB_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_IB_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_DUMMY_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_RB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_RB_RPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_RB_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_RB_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_IB_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_IB_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_IB_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_DUMMY_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_VM_CNTL)
+};
 
 #define SDMA0_POWER_CNTL__ON_OFF_CONDITION_HOLD_TIME_MASK  0x000000F8L
 #define SDMA0_POWER_CNTL__ON_OFF_STATUS_DURATION_TIME_MASK 0xFC000000L
 
+#define WREG32_SDMA(instance, offset, value) \
+	WREG32(sdma_v4_0_get_reg_offset(adev, (instance), (offset)), value)
+#define RREG32_SDMA(instance, offset) \
+	RREG32(sdma_v4_0_get_reg_offset(adev, (instance), (offset)))
+
 static void sdma_v4_0_set_ring_funcs(struct amdgpu_device *adev);
 static void sdma_v4_0_set_buffer_funcs(struct amdgpu_device *adev);
 static void sdma_v4_0_set_vm_pte_funcs(struct amdgpu_device *adev);
 static void sdma_v4_0_set_irq_funcs(struct amdgpu_device *adev);
+static void sdma_v4_0_set_ras_funcs(struct amdgpu_device *adev);
+
+static const struct soc15_reg_golden golden_settings_sdma_4[] = {
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_CHICKEN_BITS, 0xfe931f07, 0x02831d07),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_CLK_CTRL, 0xff000ff0, 0x3f000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GFX_IB_CNTL, 0x800f0100, 0x00000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GFX_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_PAGE_IB_CNTL, 0x800f0100, 0x00000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_PAGE_RB_WPTR_POLL_CNTL, 0x0000fff0, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_POWER_CNTL, 0x003ff006, 0x0003c000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC0_IB_CNTL, 0x800f0100, 0x00000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC0_RB_WPTR_POLL_CNTL, 0x0000fff0, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC1_IB_CNTL, 0x800f0100, 0x00000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC1_RB_WPTR_POLL_CNTL, 0x0000fff0, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_UTCL1_PAGE, 0x000003ff, 0x000003c0),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_UTCL1_WATERMK, 0xfc000000, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_CLK_CTRL, 0xffffffff, 0x3f000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GFX_IB_CNTL, 0x800f0100, 0x00000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GFX_RB_WPTR_POLL_CNTL, 0x0000fff0, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_PAGE_IB_CNTL, 0x800f0100, 0x00000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_PAGE_RB_WPTR_POLL_CNTL, 0x0000fff0, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_POWER_CNTL, 0x003ff000, 0x0003c000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC0_IB_CNTL, 0x800f0100, 0x00000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC0_RB_WPTR_POLL_CNTL, 0x0000fff0, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC1_IB_CNTL, 0x800f0100, 0x00000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC1_RB_WPTR_POLL_CNTL, 0x0000fff0, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_UTCL1_PAGE, 0x000003ff, 0x000003c0),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_UTCL1_WATERMK, 0xfc000000, 0x00000000)
+};
+
+static const struct soc15_reg_golden golden_settings_sdma_vg10[] = {
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG, 0x0018773f, 0x00104002),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00104002),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_UTCL1_TIMEOUT, 0xffffffff, 0x00010001),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_CHICKEN_BITS, 0xfe931f07, 0x02831d07),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG, 0x0018773f, 0x00104002),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00104002),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_UTCL1_TIMEOUT, 0xffffffff, 0x00010001),
+};
 
-static const u32 golden_settings_sdma_4[] = {
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_CHICKEN_BITS), 0xfe931f07, 0x02831f07,
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_CLK_CTRL), 0xff000ff0, 0x3f000100,
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_GFX_IB_CNTL), 0x800f0100, 0x00000100,
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_GFX_RB_WPTR_POLL_CNTL), 0xfffffff7, 0x00403000,
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_PAGE_IB_CNTL), 0x800f0100, 0x00000100,
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_PAGE_RB_WPTR_POLL_CNTL), 0x0000fff0, 0x00403000,
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_POWER_CNTL), 0x003ff006, 0x0003c000,
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_RLC0_IB_CNTL), 0x800f0100, 0x00000100,
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_RLC0_RB_WPTR_POLL_CNTL), 0x0000fff0, 0x00403000,
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_RLC1_IB_CNTL), 0x800f0100, 0x00000100,
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_RLC1_RB_WPTR_POLL_CNTL), 0x0000fff0, 0x00403000,
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_UTCL1_PAGE), 0x000003ff, 0x000003c0,
-	SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_CHICKEN_BITS), 0xfe931f07, 0x02831f07,
-	SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_CLK_CTRL), 0xffffffff, 0x3f000100,
-	SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_GFX_IB_CNTL), 0x800f0100, 0x00000100,
-	SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_GFX_RB_WPTR_POLL_CNTL), 0x0000fff0, 0x00403000,
-	SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_PAGE_IB_CNTL), 0x800f0100, 0x00000100,
-	SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_PAGE_RB_WPTR_POLL_CNTL), 0x0000fff0, 0x00403000,
-	SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_POWER_CNTL), 0x003ff000, 0x0003c000,
-	SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_RLC0_IB_CNTL), 0x800f0100, 0x00000100,
-	SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_RLC0_RB_WPTR_POLL_CNTL), 0x0000fff0, 0x00403000,
-	SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_RLC1_IB_CNTL), 0x800f0100, 0x00000100,
-	SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_RLC1_RB_WPTR_POLL_CNTL), 0x0000fff0, 0x00403000,
-	SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_UTCL1_PAGE), 0x000003ff, 0x000003c0
+static const struct soc15_reg_golden golden_settings_sdma_vg12[] = {
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG, 0x0018773f, 0x00104001),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00104001),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_UTCL1_TIMEOUT, 0xffffffff, 0x00010001),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_CHICKEN_BITS, 0xfe931f07, 0x02831d07),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG, 0x0018773f, 0x00104001),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00104001),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_UTCL1_TIMEOUT, 0xffffffff, 0x00010001),
 };
 
-static const u32 golden_settings_sdma_vg10[] = {
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG), 0x0018773f, 0x00104002,
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG_READ), 0x0018773f, 0x00104002,
-	SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG), 0x0018773f, 0x00104002,
-	SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG_READ), 0x0018773f, 0x00104002
+static const struct soc15_reg_golden golden_settings_sdma_4_1[] = {
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_CHICKEN_BITS, 0xfe931f07, 0x02831d07),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_CLK_CTRL, 0xffffffff, 0x3f000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GFX_IB_CNTL, 0x800f0111, 0x00000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GFX_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_POWER_CNTL, 0xfc3fffff, 0x40000051),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC0_IB_CNTL, 0x800f0111, 0x00000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC0_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC1_IB_CNTL, 0x800f0111, 0x00000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC1_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_UTCL1_PAGE, 0x000003ff, 0x000003e0),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_UTCL1_WATERMK, 0xfc000000, 0x00000000)
 };
 
-static const u32 golden_settings_sdma_4_1[] =
-{
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_CHICKEN_BITS), 0xfe931f07, 0x02831f07,
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_CLK_CTRL), 0xffffffff, 0x3f000100,
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_GFX_IB_CNTL), 0x800f0111, 0x00000100,
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_GFX_RB_WPTR_POLL_CNTL), 0xfffffff7, 0x00403000,
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_POWER_CNTL), 0xfc3fffff, 0x40000051,
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_RLC0_IB_CNTL), 0x800f0111, 0x00000100,
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_RLC0_RB_WPTR_POLL_CNTL), 0xfffffff7, 0x00403000,
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_RLC1_IB_CNTL), 0x800f0111, 0x00000100,
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_RLC1_RB_WPTR_POLL_CNTL), 0xfffffff7, 0x00403000,
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_UTCL1_PAGE), 0x000003ff, 0x000003c0
+static const struct soc15_reg_golden golden_settings_sdma0_4_2_init[] = {
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC0_RB_WPTR_POLL_CNTL, 0xfffffff0, 0x00403000),
 };
 
-static const u32 golden_settings_sdma_rv1[] =
+static const struct soc15_reg_golden golden_settings_sdma0_4_2[] =
 {
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG), 0x0018773f, 0x00000002,
-	SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG_READ), 0x0018773f, 0x00000002
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_CHICKEN_BITS, 0xfe931f07, 0x02831f07),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_CLK_CTRL, 0xffffffff, 0x3f000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG_READ, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GFX_RB_RPTR_ADDR_LO, 0xfffffffd, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GFX_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_PAGE_RB_RPTR_ADDR_LO, 0xfffffffd, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_PAGE_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RD_BURST_CNTL, 0x0000000f, 0x00000003),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC0_RB_RPTR_ADDR_LO, 0xfffffffd, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC0_RB_WPTR_POLL_CNTL, 0xfffffff0, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC1_RB_RPTR_ADDR_LO, 0xfffffffd, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC1_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC2_RB_RPTR_ADDR_LO, 0xfffffffd, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC2_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC3_RB_RPTR_ADDR_LO, 0xfffffffd, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC3_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC4_RB_RPTR_ADDR_LO, 0xfffffffd, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC4_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC5_RB_RPTR_ADDR_LO, 0xfffffffd, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC5_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC6_RB_RPTR_ADDR_LO, 0xfffffffd, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC6_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC7_RB_RPTR_ADDR_LO, 0xfffffffd, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC7_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_UTCL1_PAGE, 0x000003ff, 0x000003c0),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_UTCL1_TIMEOUT, 0xffffffff, 0x00010001),
 };
 
-static u32 sdma_v4_0_get_reg_offset(u32 instance, u32 internal_offset)
+static const struct soc15_reg_golden golden_settings_sdma1_4_2[] = {
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_CHICKEN_BITS, 0xfe931f07, 0x02831f07),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_CLK_CTRL, 0xffffffff, 0x3f000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG_READ, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GFX_RB_RPTR_ADDR_LO, 0xfffffffd, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GFX_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_PAGE_RB_RPTR_ADDR_LO, 0xfffffffd, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_PAGE_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RD_BURST_CNTL, 0x0000000f, 0x00000003),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC0_RB_RPTR_ADDR_LO, 0xfffffffd, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC0_RB_WPTR_POLL_CNTL, 0xfffffff0, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC1_RB_RPTR_ADDR_LO, 0xfffffffd, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC1_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC2_RB_RPTR_ADDR_LO, 0xfffffffd, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC2_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC3_RB_RPTR_ADDR_LO, 0xfffffffd, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC3_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC4_RB_RPTR_ADDR_LO, 0xfffffffd, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC4_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC5_RB_RPTR_ADDR_LO, 0xfffffffd, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC5_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC6_RB_RPTR_ADDR_LO, 0xfffffffd, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC6_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC7_RB_RPTR_ADDR_LO, 0xfffffffd, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC7_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_UTCL1_PAGE, 0x000003ff, 0x000003c0),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_UTCL1_TIMEOUT, 0xffffffff, 0x00010001),
+};
+
+static const struct soc15_reg_golden golden_settings_sdma_rv1[] =
+{
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG, 0x0018773f, 0x00000002),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00000002)
+};
+
+static const struct soc15_reg_golden golden_settings_sdma_rv2[] =
 {
-	u32 base = 0;
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG, 0x0018773f, 0x00003001),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00003001)
+};
+
+static const struct soc15_reg_golden golden_settings_sdma_arct[] =
+{
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_CHICKEN_BITS, 0xfe931f07, 0x02831f07),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG_READ, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_UTCL1_TIMEOUT, 0xffffffff, 0x00010001),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_CHICKEN_BITS, 0xfe931f07, 0x02831f07),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG_READ, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_UTCL1_TIMEOUT, 0xffffffff, 0x00010001),
+	SOC15_REG_GOLDEN_VALUE(SDMA2, 0, mmSDMA2_CHICKEN_BITS, 0xfe931f07, 0x02831f07),
+	SOC15_REG_GOLDEN_VALUE(SDMA2, 0, mmSDMA2_GB_ADDR_CONFIG, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA2, 0, mmSDMA2_GB_ADDR_CONFIG_READ, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA2, 0, mmSDMA2_UTCL1_TIMEOUT, 0xffffffff, 0x00010001),
+	SOC15_REG_GOLDEN_VALUE(SDMA3, 0, mmSDMA3_CHICKEN_BITS, 0xfe931f07, 0x02831f07),
+	SOC15_REG_GOLDEN_VALUE(SDMA3, 0, mmSDMA3_GB_ADDR_CONFIG, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA3, 0, mmSDMA3_GB_ADDR_CONFIG_READ, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA3, 0, mmSDMA3_UTCL1_TIMEOUT, 0xffffffff, 0x00010001),
+	SOC15_REG_GOLDEN_VALUE(SDMA4, 0, mmSDMA4_CHICKEN_BITS, 0xfe931f07, 0x02831f07),
+	SOC15_REG_GOLDEN_VALUE(SDMA4, 0, mmSDMA4_GB_ADDR_CONFIG, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA4, 0, mmSDMA4_GB_ADDR_CONFIG_READ, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA4, 0, mmSDMA4_UTCL1_TIMEOUT, 0xffffffff, 0x00010001),
+	SOC15_REG_GOLDEN_VALUE(SDMA5, 0, mmSDMA5_CHICKEN_BITS, 0xfe931f07, 0x02831f07),
+	SOC15_REG_GOLDEN_VALUE(SDMA5, 0, mmSDMA5_GB_ADDR_CONFIG, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA5, 0, mmSDMA5_GB_ADDR_CONFIG_READ, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA5, 0, mmSDMA5_UTCL1_TIMEOUT, 0xffffffff, 0x00010001),
+	SOC15_REG_GOLDEN_VALUE(SDMA6, 0, mmSDMA6_CHICKEN_BITS, 0xfe931f07, 0x02831f07),
+	SOC15_REG_GOLDEN_VALUE(SDMA6, 0, mmSDMA6_GB_ADDR_CONFIG, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA6, 0, mmSDMA6_GB_ADDR_CONFIG_READ, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA6, 0, mmSDMA6_UTCL1_TIMEOUT, 0xffffffff, 0x00010001),
+	SOC15_REG_GOLDEN_VALUE(SDMA7, 0, mmSDMA7_CHICKEN_BITS, 0xfe931f07, 0x02831f07),
+	SOC15_REG_GOLDEN_VALUE(SDMA7, 0, mmSDMA7_GB_ADDR_CONFIG, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA7, 0, mmSDMA7_GB_ADDR_CONFIG_READ, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA7, 0, mmSDMA7_UTCL1_TIMEOUT, 0xffffffff, 0x00010001)
+};
 
+static const struct soc15_reg_golden golden_settings_sdma_aldebaran[] = {
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG, 0x0018773f, 0x00104002),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00104002),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_UTCL1_TIMEOUT, 0xffffffff, 0x00010001),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG, 0x0018773f, 0x00104002),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00104002),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_UTCL1_TIMEOUT, 0xffffffff, 0x00010001),
+	SOC15_REG_GOLDEN_VALUE(SDMA2, 0, mmSDMA2_GB_ADDR_CONFIG, 0x0018773f, 0x00104002),
+	SOC15_REG_GOLDEN_VALUE(SDMA2, 0, mmSDMA2_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00104002),
+	SOC15_REG_GOLDEN_VALUE(SDMA3, 0, mmSDMA2_UTCL1_TIMEOUT, 0xffffffff, 0x00010001),
+	SOC15_REG_GOLDEN_VALUE(SDMA3, 0, mmSDMA3_GB_ADDR_CONFIG, 0x0018773f, 0x00104002),
+	SOC15_REG_GOLDEN_VALUE(SDMA3, 0, mmSDMA3_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00104002),
+	SOC15_REG_GOLDEN_VALUE(SDMA3, 0, mmSDMA3_UTCL1_TIMEOUT, 0xffffffff, 0x00010001),
+	SOC15_REG_GOLDEN_VALUE(SDMA4, 0, mmSDMA4_GB_ADDR_CONFIG, 0x0018773f, 0x00104002),
+	SOC15_REG_GOLDEN_VALUE(SDMA4, 0, mmSDMA4_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00104002),
+	SOC15_REG_GOLDEN_VALUE(SDMA4, 0, mmSDMA4_UTCL1_TIMEOUT, 0xffffffff, 0x00010001),
+};
+
+static const struct soc15_reg_golden golden_settings_sdma_4_3[] = {
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_CHICKEN_BITS, 0xfe931f07, 0x02831f07),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_CLK_CTRL, 0xffffffff, 0x3f000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG, 0x0018773f, 0x00000002),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00000002),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GFX_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_POWER_CNTL, 0x003fff07, 0x40000051),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC0_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC1_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_UTCL1_PAGE, 0x000003ff, 0x000003e0),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_UTCL1_WATERMK, 0xfc000000, 0x03fbe1fe)
+};
+
+static const struct soc15_ras_field_entry sdma_v4_0_ras_fields[] = {
+	{ "SDMA_UCODE_BUF_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_UCODE_BUF_SED),
+	0, 0,
+	},
+	{ "SDMA_RB_CMD_BUF_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_RB_CMD_BUF_SED),
+	0, 0,
+	},
+	{ "SDMA_IB_CMD_BUF_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_IB_CMD_BUF_SED),
+	0, 0,
+	},
+	{ "SDMA_UTCL1_RD_FIFO_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_UTCL1_RD_FIFO_SED),
+	0, 0,
+	},
+	{ "SDMA_UTCL1_RDBST_FIFO_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_UTCL1_RDBST_FIFO_SED),
+	0, 0,
+	},
+	{ "SDMA_DATA_LUT_FIFO_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_DATA_LUT_FIFO_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF0_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF0_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF1_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF1_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF2_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF2_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF3_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF3_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF4_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF4_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF5_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF5_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF6_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF6_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF7_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF7_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF8_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF8_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF9_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF9_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF10_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF10_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF11_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF11_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF12_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF12_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF13_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF13_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF14_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF14_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF15_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF15_SED),
+	0, 0,
+	},
+	{ "SDMA_SPLIT_DAT_BUF_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_SPLIT_DAT_BUF_SED),
+	0, 0,
+	},
+	{ "SDMA_MC_WR_ADDR_FIFO_SED", SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MC_WR_ADDR_FIFO_SED),
+	0, 0,
+	},
+};
+
+static u32 sdma_v4_0_get_reg_offset(struct amdgpu_device *adev,
+		u32 instance, u32 offset)
+{
 	switch (instance) {
 	case 0:
-		base = SDMA0_BASE.instance[0].segment[0];
-		break;
+		return (adev->reg_offset[SDMA0_HWIP][0][0] + offset);
 	case 1:
-		base = SDMA1_BASE.instance[0].segment[0];
+		return (adev->reg_offset[SDMA1_HWIP][0][0] + offset);
+	case 2:
+		return (adev->reg_offset[SDMA2_HWIP][0][1] + offset);
+	case 3:
+		return (adev->reg_offset[SDMA3_HWIP][0][1] + offset);
+	case 4:
+		return (adev->reg_offset[SDMA4_HWIP][0][1] + offset);
+	case 5:
+		return (adev->reg_offset[SDMA5_HWIP][0][1] + offset);
+	case 6:
+		return (adev->reg_offset[SDMA6_HWIP][0][1] + offset);
+	case 7:
+		return (adev->reg_offset[SDMA7_HWIP][0][1] + offset);
+	default:
 		break;
+	}
+	return 0;
+}
+
+static unsigned sdma_v4_0_seq_to_irq_id(int seq_num)
+{
+	switch (seq_num) {
+	case 0:
+		return SOC15_IH_CLIENTID_SDMA0;
+	case 1:
+		return SOC15_IH_CLIENTID_SDMA1;
+	case 2:
+		return SOC15_IH_CLIENTID_SDMA2;
+	case 3:
+		return SOC15_IH_CLIENTID_SDMA3;
+	case 4:
+		return SOC15_IH_CLIENTID_SDMA4;
+	case 5:
+		return SOC15_IH_CLIENTID_SDMA5;
+	case 6:
+		return SOC15_IH_CLIENTID_SDMA6;
+	case 7:
+		return SOC15_IH_CLIENTID_SDMA7;
 	default:
-		BUG();
 		break;
 	}
+	return -EINVAL;
+}
 
-	return base + internal_offset;
+static int sdma_v4_0_irq_id_to_seq(unsigned client_id)
+{
+	switch (client_id) {
+	case SOC15_IH_CLIENTID_SDMA0:
+		return 0;
+	case SOC15_IH_CLIENTID_SDMA1:
+		return 1;
+	case SOC15_IH_CLIENTID_SDMA2:
+		return 2;
+	case SOC15_IH_CLIENTID_SDMA3:
+		return 3;
+	case SOC15_IH_CLIENTID_SDMA4:
+		return 4;
+	case SOC15_IH_CLIENTID_SDMA5:
+		return 5;
+	case SOC15_IH_CLIENTID_SDMA6:
+		return 6;
+	case SOC15_IH_CLIENTID_SDMA7:
+		return 7;
+	default:
+		break;
+	}
+	return -EINVAL;
 }
 
 static void sdma_v4_0_init_golden_registers(struct amdgpu_device *adev)
 {
-	switch (adev->asic_type) {
-	case CHIP_VEGA10:
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_sdma_4,
-						 (const u32)ARRAY_SIZE(golden_settings_sdma_4));
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_sdma_vg10,
-						 (const u32)ARRAY_SIZE(golden_settings_sdma_vg10));
+	switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) {
+	case IP_VERSION(4, 0, 0):
+		soc15_program_register_sequence(adev,
+						golden_settings_sdma_4,
+						ARRAY_SIZE(golden_settings_sdma_4));
+		soc15_program_register_sequence(adev,
+						golden_settings_sdma_vg10,
+						ARRAY_SIZE(golden_settings_sdma_vg10));
+		break;
+	case IP_VERSION(4, 0, 1):
+		soc15_program_register_sequence(adev,
+						golden_settings_sdma_4,
+						ARRAY_SIZE(golden_settings_sdma_4));
+		soc15_program_register_sequence(adev,
+						golden_settings_sdma_vg12,
+						ARRAY_SIZE(golden_settings_sdma_vg12));
+		break;
+	case IP_VERSION(4, 2, 0):
+		soc15_program_register_sequence(adev,
+						golden_settings_sdma0_4_2_init,
+						ARRAY_SIZE(golden_settings_sdma0_4_2_init));
+		soc15_program_register_sequence(adev,
+						golden_settings_sdma0_4_2,
+						ARRAY_SIZE(golden_settings_sdma0_4_2));
+		soc15_program_register_sequence(adev,
+						golden_settings_sdma1_4_2,
+						ARRAY_SIZE(golden_settings_sdma1_4_2));
+		break;
+	case IP_VERSION(4, 2, 2):
+		soc15_program_register_sequence(adev,
+						golden_settings_sdma_arct,
+						ARRAY_SIZE(golden_settings_sdma_arct));
 		break;
-	case CHIP_RAVEN:
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_sdma_4_1,
-						 (const u32)ARRAY_SIZE(golden_settings_sdma_4_1));
-		amdgpu_program_register_sequence(adev,
-						 golden_settings_sdma_rv1,
-						 (const u32)ARRAY_SIZE(golden_settings_sdma_rv1));
+	case IP_VERSION(4, 4, 0):
+		soc15_program_register_sequence(adev,
+						golden_settings_sdma_aldebaran,
+						ARRAY_SIZE(golden_settings_sdma_aldebaran));
+		break;
+	case IP_VERSION(4, 1, 0):
+	case IP_VERSION(4, 1, 1):
+		soc15_program_register_sequence(adev,
+						golden_settings_sdma_4_1,
+						ARRAY_SIZE(golden_settings_sdma_4_1));
+		if (adev->apu_flags & AMD_APU_IS_RAVEN2)
+			soc15_program_register_sequence(adev,
+							golden_settings_sdma_rv2,
+							ARRAY_SIZE(golden_settings_sdma_rv2));
+		else
+			soc15_program_register_sequence(adev,
+							golden_settings_sdma_rv1,
+							ARRAY_SIZE(golden_settings_sdma_rv1));
+		break;
+	case IP_VERSION(4, 1, 2):
+		soc15_program_register_sequence(adev,
+						golden_settings_sdma_4_3,
+						ARRAY_SIZE(golden_settings_sdma_4_3));
 		break;
 	default:
 		break;
 	}
 }
 
+static void sdma_v4_0_setup_ulv(struct amdgpu_device *adev)
+{
+	int i;
+
+	/*
+	 * The only chips with SDMAv4 and ULV are VG10 and VG20.
+	 * Server SKUs take a different hysteresis setting from other SKUs.
+	 */
+	switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) {
+	case IP_VERSION(4, 0, 0):
+		if (adev->pdev->device == 0x6860)
+			break;
+		return;
+	case IP_VERSION(4, 2, 0):
+		if (adev->pdev->device == 0x66a1)
+			break;
+		return;
+	default:
+		return;
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		uint32_t temp;
+
+		temp = RREG32_SDMA(i, mmSDMA0_ULV_CNTL);
+		temp = REG_SET_FIELD(temp, SDMA0_ULV_CNTL, HYSTERESIS, 0x0);
+		WREG32_SDMA(i, mmSDMA0_ULV_CNTL, temp);
+	}
+}
+
 /**
  * sdma_v4_0_init_microcode - load ucode images from disk
  *
@@ -164,63 +622,25 @@ static void sdma_v4_0_init_golden_registers(struct amdgpu_device *adev)
 // vega10 real chip need to use PSP to load firmware
 static int sdma_v4_0_init_microcode(struct amdgpu_device *adev)
 {
-	const char *chip_name;
-	char fw_name[30];
-	int err = 0, i;
-	struct amdgpu_firmware_info *info = NULL;
-	const struct common_firmware_header *header = NULL;
-	const struct sdma_firmware_header_v1_0 *hdr;
-
-	DRM_DEBUG("\n");
-
-	switch (adev->asic_type) {
-	case CHIP_VEGA10:
-		chip_name = "vega10";
-		break;
-	case CHIP_RAVEN:
-		chip_name = "raven";
-		break;
-	default:
-		BUG();
-	}
+	int ret, i;
 
 	for (i = 0; i < adev->sdma.num_instances; i++) {
-		if (i == 0)
-			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_sdma.bin", chip_name);
-		else
-			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_sdma1.bin", chip_name);
-		err = request_firmware(&adev->sdma.instance[i].fw, fw_name, adev->dev);
-		if (err)
-			goto out;
-		err = amdgpu_ucode_validate(adev->sdma.instance[i].fw);
-		if (err)
-			goto out;
-		hdr = (const struct sdma_firmware_header_v1_0 *)adev->sdma.instance[i].fw->data;
-		adev->sdma.instance[i].fw_version = le32_to_cpu(hdr->header.ucode_version);
-		adev->sdma.instance[i].feature_version = le32_to_cpu(hdr->ucode_feature_version);
-		if (adev->sdma.instance[i].feature_version >= 20)
-			adev->sdma.instance[i].burst_nop = true;
-		DRM_DEBUG("psp_load == '%s'\n",
-				adev->firmware.load_type == AMDGPU_FW_LOAD_PSP ? "true" : "false");
-
-		if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
-			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_SDMA0 + i];
-			info->ucode_id = AMDGPU_UCODE_ID_SDMA0 + i;
-			info->fw = adev->sdma.instance[i].fw;
-			header = (const struct common_firmware_header *)info->fw->data;
-			adev->firmware.fw_size +=
-				ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
-		}
-	}
-out:
-	if (err) {
-		DRM_ERROR("sdma_v4_0: Failed to load firmware \"%s\"\n", fw_name);
-		for (i = 0; i < adev->sdma.num_instances; i++) {
-			release_firmware(adev->sdma.instance[i].fw);
-			adev->sdma.instance[i].fw = NULL;
+		if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) ==
+			    IP_VERSION(4, 2, 2) ||
+		    amdgpu_ip_version(adev, SDMA0_HWIP, 0) ==
+			    IP_VERSION(4, 4, 0)) {
+			/* Acturus & Aldebaran will leverage the same FW memory
+			   for every SDMA instance */
+			ret = amdgpu_sdma_init_microcode(adev, 0, true);
+			break;
+		} else {
+			ret = amdgpu_sdma_init_microcode(adev, i, false);
+			if (ret)
+				return ret;
 		}
 	}
-	return err;
+
+	return ret;
 }
 
 /**
@@ -235,7 +655,7 @@ static uint64_t sdma_v4_0_ring_get_rptr(struct amdgpu_ring *ring)
 	u64 *rptr;
 
 	/* XXX check if swapping is necessary on BE */
-	rptr = ((u64 *)&ring->adev->wb.wb[ring->rptr_offs]);
+	rptr = ((u64 *)ring->rptr_cpu_addr);
 
 	DRM_DEBUG("rptr before shift == 0x%016llx\n", *rptr);
 	return ((*rptr) >> 2);
@@ -251,31 +671,21 @@ static uint64_t sdma_v4_0_ring_get_rptr(struct amdgpu_ring *ring)
 static uint64_t sdma_v4_0_ring_get_wptr(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
-	u64 *wptr = NULL;
-	uint64_t local_wptr = 0;
+	u64 wptr;
 
 	if (ring->use_doorbell) {
 		/* XXX check if swapping is necessary on BE */
-		wptr = ((u64 *)&adev->wb.wb[ring->wptr_offs]);
-		DRM_DEBUG("wptr/doorbell before shift == 0x%016llx\n", *wptr);
-		*wptr = (*wptr) >> 2;
-		DRM_DEBUG("wptr/doorbell after shift == 0x%016llx\n", *wptr);
+		wptr = READ_ONCE(*((u64 *)ring->wptr_cpu_addr));
+		DRM_DEBUG("wptr/doorbell before shift == 0x%016llx\n", wptr);
 	} else {
-		u32 lowbit, highbit;
-		int me = (ring == &adev->sdma.instance[0].ring) ? 0 : 1;
-
-		wptr = &local_wptr;
-		lowbit = RREG32(sdma_v4_0_get_reg_offset(me, mmSDMA0_GFX_RB_WPTR)) >> 2;
-		highbit = RREG32(sdma_v4_0_get_reg_offset(me, mmSDMA0_GFX_RB_WPTR_HI)) >> 2;
-
-		DRM_DEBUG("wptr [%i]high== 0x%08x low==0x%08x\n",
-				me, highbit, lowbit);
-		*wptr = highbit;
-		*wptr = (*wptr) << 32;
-		*wptr |= lowbit;
+		wptr = RREG32_SDMA(ring->me, mmSDMA0_GFX_RB_WPTR_HI);
+		wptr = wptr << 32;
+		wptr |= RREG32_SDMA(ring->me, mmSDMA0_GFX_RB_WPTR);
+		DRM_DEBUG("wptr before shift [%i] wptr == 0x%016llx\n",
+				ring->me, wptr);
 	}
 
-	return *wptr;
+	return wptr >> 2;
 }
 
 /**
@@ -291,12 +701,12 @@ static void sdma_v4_0_ring_set_wptr(struct amdgpu_ring *ring)
 
 	DRM_DEBUG("Setting write pointer\n");
 	if (ring->use_doorbell) {
-		u64 *wb = (u64 *)&adev->wb.wb[ring->wptr_offs];
+		u64 *wb = (u64 *)ring->wptr_cpu_addr;
 
 		DRM_DEBUG("Using doorbell -- "
 				"wptr_offs == 0x%08x "
-				"lower_32_bits(ring->wptr) << 2 == 0x%08x "
-				"upper_32_bits(ring->wptr) << 2 == 0x%08x\n",
+				"lower_32_bits(ring->wptr << 2) == 0x%08x "
+				"upper_32_bits(ring->wptr << 2) == 0x%08x\n",
 				ring->wptr_offs,
 				lower_32_bits(ring->wptr << 2),
 				upper_32_bits(ring->wptr << 2));
@@ -306,23 +716,74 @@ static void sdma_v4_0_ring_set_wptr(struct amdgpu_ring *ring)
 				ring->doorbell_index, ring->wptr << 2);
 		WDOORBELL64(ring->doorbell_index, ring->wptr << 2);
 	} else {
-		int me = (ring == &ring->adev->sdma.instance[0].ring) ? 0 : 1;
-
 		DRM_DEBUG("Not using doorbell -- "
 				"mmSDMA%i_GFX_RB_WPTR == 0x%08x "
 				"mmSDMA%i_GFX_RB_WPTR_HI == 0x%08x\n",
-				me,
+				ring->me,
 				lower_32_bits(ring->wptr << 2),
-				me,
+				ring->me,
 				upper_32_bits(ring->wptr << 2));
-		WREG32(sdma_v4_0_get_reg_offset(me, mmSDMA0_GFX_RB_WPTR), lower_32_bits(ring->wptr << 2));
-		WREG32(sdma_v4_0_get_reg_offset(me, mmSDMA0_GFX_RB_WPTR_HI), upper_32_bits(ring->wptr << 2));
+		WREG32_SDMA(ring->me, mmSDMA0_GFX_RB_WPTR,
+			    lower_32_bits(ring->wptr << 2));
+		WREG32_SDMA(ring->me, mmSDMA0_GFX_RB_WPTR_HI,
+			    upper_32_bits(ring->wptr << 2));
+	}
+}
+
+/**
+ * sdma_v4_0_page_ring_get_wptr - get the current write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current wptr from the hardware (VEGA10+).
+ */
+static uint64_t sdma_v4_0_page_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u64 wptr;
+
+	if (ring->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		wptr = READ_ONCE(*((u64 *)ring->wptr_cpu_addr));
+	} else {
+		wptr = RREG32_SDMA(ring->me, mmSDMA0_PAGE_RB_WPTR_HI);
+		wptr = wptr << 32;
+		wptr |= RREG32_SDMA(ring->me, mmSDMA0_PAGE_RB_WPTR);
+	}
+
+	return wptr >> 2;
+}
+
+/**
+ * sdma_v4_0_page_ring_set_wptr - commit the write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Write the wptr back to the hardware (VEGA10+).
+ */
+static void sdma_v4_0_page_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		u64 *wb = (u64 *)ring->wptr_cpu_addr;
+
+		/* XXX check if swapping is necessary on BE */
+		WRITE_ONCE(*wb, (ring->wptr << 2));
+		WDOORBELL64(ring->doorbell_index, ring->wptr << 2);
+	} else {
+		uint64_t wptr = ring->wptr << 2;
+
+		WREG32_SDMA(ring->me, mmSDMA0_PAGE_RB_WPTR,
+			    lower_32_bits(wptr));
+		WREG32_SDMA(ring->me, mmSDMA0_PAGE_RB_WPTR_HI,
+			    upper_32_bits(wptr));
 	}
 }
 
 static void sdma_v4_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
 {
-	struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring);
+	struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring);
 	int i;
 
 	for (i = 0; i < count; i++)
@@ -337,21 +798,24 @@ static void sdma_v4_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
  * sdma_v4_0_ring_emit_ib - Schedule an IB on the DMA engine
  *
  * @ring: amdgpu ring pointer
+ * @job: job to retrieve vmid from
  * @ib: IB object to schedule
+ * @flags: unused
  *
  * Schedule an IB in the DMA ring (VEGA10).
  */
 static void sdma_v4_0_ring_emit_ib(struct amdgpu_ring *ring,
-					struct amdgpu_ib *ib,
-					unsigned vm_id, bool ctx_switch)
+				   struct amdgpu_job *job,
+				   struct amdgpu_ib *ib,
+				   uint32_t flags)
 {
-	u32 vmid = vm_id & 0xf;
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
 
 	/* IB packet must end on a 8 DW boundary */
-	sdma_v4_0_ring_insert_nop(ring, (10 - (lower_32_bits(ring->wptr) & 7)) % 8);
+	sdma_v4_0_ring_insert_nop(ring, (2 - lower_32_bits(ring->wptr)) & 7);
 
 	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_INDIRECT) |
-			  SDMA_PKT_INDIRECT_HEADER_VMID(vmid));
+			  SDMA_PKT_INDIRECT_HEADER_VMID(vmid & 0xf));
 	/* base must be 32 byte aligned */
 	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr) & 0xffffffe0);
 	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
@@ -361,6 +825,31 @@ static void sdma_v4_0_ring_emit_ib(struct amdgpu_ring *ring,
 
 }
 
+static void sdma_v4_0_wait_reg_mem(struct amdgpu_ring *ring,
+				   int mem_space, int hdp,
+				   uint32_t addr0, uint32_t addr1,
+				   uint32_t ref, uint32_t mask,
+				   uint32_t inv)
+{
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(hdp) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_MEM_POLL(mem_space) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3)); /* == */
+	if (mem_space) {
+		/* memory */
+		amdgpu_ring_write(ring, addr0);
+		amdgpu_ring_write(ring, addr1);
+	} else {
+		/* registers */
+		amdgpu_ring_write(ring, addr0 << 2);
+		amdgpu_ring_write(ring, addr1 << 2);
+	}
+	amdgpu_ring_write(ring, ref); /* reference */
+	amdgpu_ring_write(ring, mask); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(inv)); /* retry count, poll interval */
+}
+
 /**
  * sdma_v4_0_ring_emit_hdp_flush - emit an hdp flush on the DMA ring
  *
@@ -370,43 +859,25 @@ static void sdma_v4_0_ring_emit_ib(struct amdgpu_ring *ring,
  */
 static void sdma_v4_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
 {
+	struct amdgpu_device *adev = ring->adev;
 	u32 ref_and_mask = 0;
-	struct nbio_hdp_flush_reg *nbio_hf_reg;
-
-	if (ring->adev->flags & AMD_IS_APU)
-		nbio_hf_reg = &nbio_v7_0_hdp_flush_reg;
-	else
-		nbio_hf_reg = &nbio_v6_1_hdp_flush_reg;
+	const struct nbio_hdp_flush_reg *nbio_hf_reg = adev->nbio.hdp_flush_reg;
 
-	if (ring == &ring->adev->sdma.instance[0].ring)
-		ref_and_mask = nbio_hf_reg->ref_and_mask_sdma0;
-	else
-		ref_and_mask = nbio_hf_reg->ref_and_mask_sdma1;
+	ref_and_mask = nbio_hf_reg->ref_and_mask_sdma0 << ring->me;
 
-	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
-			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(1) |
-			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3)); /* == */
-	amdgpu_ring_write(ring, nbio_hf_reg->hdp_flush_done_offset << 2);
-	amdgpu_ring_write(ring, nbio_hf_reg->hdp_flush_req_offset << 2);
-	amdgpu_ring_write(ring, ref_and_mask); /* reference */
-	amdgpu_ring_write(ring, ref_and_mask); /* mask */
-	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
-			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
-}
-
-static void sdma_v4_0_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
-{
-	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
-			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
-	amdgpu_ring_write(ring, SOC15_REG_OFFSET(HDP, 0, mmHDP_DEBUG0));
-	amdgpu_ring_write(ring, 1);
+	sdma_v4_0_wait_reg_mem(ring, 0, 1,
+			       adev->nbio.funcs->get_hdp_flush_done_offset(adev),
+			       adev->nbio.funcs->get_hdp_flush_req_offset(adev),
+			       ref_and_mask, ref_and_mask, 10);
 }
 
 /**
  * sdma_v4_0_ring_emit_fence - emit a fence on the DMA ring
  *
  * @ring: amdgpu ring pointer
- * @fence: amdgpu fence object
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
  *
  * Add a DMA fence packet to the ring to write
  * the fence seq number and DMA trap packet to generate
@@ -442,34 +913,25 @@ static void sdma_v4_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 se
 
 
 /**
- * sdma_v4_0_gfx_stop - stop the gfx async dma engines
+ * sdma_v4_0_gfx_enable - enable the gfx async dma engines
  *
  * @adev: amdgpu_device pointer
- *
- * Stop the gfx async dma ring buffers (VEGA10).
+ * @enable: enable SDMA RB/IB
+ * control the gfx async dma ring buffers (VEGA10).
  */
-static void sdma_v4_0_gfx_stop(struct amdgpu_device *adev)
+static void sdma_v4_0_gfx_enable(struct amdgpu_device *adev, bool enable)
 {
-	struct amdgpu_ring *sdma0 = &adev->sdma.instance[0].ring;
-	struct amdgpu_ring *sdma1 = &adev->sdma.instance[1].ring;
 	u32 rb_cntl, ib_cntl;
 	int i;
 
-	if ((adev->mman.buffer_funcs_ring == sdma0) ||
-	    (adev->mman.buffer_funcs_ring == sdma1))
-		amdgpu_ttm_set_active_vram_size(adev, adev->mc.visible_vram_size);
-
 	for (i = 0; i < adev->sdma.num_instances; i++) {
-		rb_cntl = RREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_CNTL));
-		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 0);
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
-		ib_cntl = RREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_IB_CNTL));
-		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 0);
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_IB_CNTL), ib_cntl);
+		rb_cntl = RREG32_SDMA(i, mmSDMA0_GFX_RB_CNTL);
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, enable ? 1 : 0);
+		WREG32_SDMA(i, mmSDMA0_GFX_RB_CNTL, rb_cntl);
+		ib_cntl = RREG32_SDMA(i, mmSDMA0_GFX_IB_CNTL);
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, enable ? 1 : 0);
+		WREG32_SDMA(i, mmSDMA0_GFX_IB_CNTL, ib_cntl);
 	}
-
-	sdma0->ready = false;
-	sdma1->ready = false;
 }
 
 /**
@@ -485,7 +947,31 @@ static void sdma_v4_0_rlc_stop(struct amdgpu_device *adev)
 }
 
 /**
- * sdma_v_0_ctx_switch_enable - stop the async dma engines context switch
+ * sdma_v4_0_page_stop - stop the page async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the page async dma ring buffers (VEGA10).
+ */
+static void sdma_v4_0_page_stop(struct amdgpu_device *adev)
+{
+	u32 rb_cntl, ib_cntl;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		rb_cntl = RREG32_SDMA(i, mmSDMA0_PAGE_RB_CNTL);
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_PAGE_RB_CNTL,
+					RB_ENABLE, 0);
+		WREG32_SDMA(i, mmSDMA0_PAGE_RB_CNTL, rb_cntl);
+		ib_cntl = RREG32_SDMA(i, mmSDMA0_PAGE_IB_CNTL);
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_PAGE_IB_CNTL,
+					IB_ENABLE, 0);
+		WREG32_SDMA(i, mmSDMA0_PAGE_IB_CNTL, ib_cntl);
+	}
+}
+
+/**
+ * sdma_v4_0_ctx_switch_enable - stop the async dma engines context switch
  *
  * @adev: amdgpu_device pointer
  * @enable: enable/disable the DMA MEs context switch.
@@ -522,18 +1008,27 @@ static void sdma_v4_0_ctx_switch_enable(struct amdgpu_device *adev, bool enable)
 	}
 
 	for (i = 0; i < adev->sdma.num_instances; i++) {
-		f32_cntl = RREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_CNTL));
+		f32_cntl = RREG32_SDMA(i, mmSDMA0_CNTL);
 		f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL,
 				AUTO_CTXSW_ENABLE, enable ? 1 : 0);
 		if (enable && amdgpu_sdma_phase_quantum) {
-			WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_PHASE0_QUANTUM),
-			       phase_quantum);
-			WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_PHASE1_QUANTUM),
-			       phase_quantum);
-			WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_PHASE2_QUANTUM),
-			       phase_quantum);
+			WREG32_SDMA(i, mmSDMA0_PHASE0_QUANTUM, phase_quantum);
+			WREG32_SDMA(i, mmSDMA0_PHASE1_QUANTUM, phase_quantum);
+			WREG32_SDMA(i, mmSDMA0_PHASE2_QUANTUM, phase_quantum);
 		}
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_CNTL), f32_cntl);
+		WREG32_SDMA(i, mmSDMA0_CNTL, f32_cntl);
+
+		/*
+		 * Enable SDMA utilization. Its only supported on
+		 * Arcturus for the moment and firmware version 14
+		 * and above.
+		 */
+		if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) ==
+			    IP_VERSION(4, 2, 2) &&
+		    adev->sdma.instance[i].fw_version >= 14)
+			WREG32_SDMA(i, mmSDMA0_PUB_DUMMY_REG2, enable);
+		/* Extend page fault timeout to avoid interrupt storm */
+		WREG32_SDMA(i, mmSDMA0_UTCL1_TIMEOUT, 0x00800080);
 	}
 
 }
@@ -551,161 +1046,206 @@ static void sdma_v4_0_enable(struct amdgpu_device *adev, bool enable)
 	u32 f32_cntl;
 	int i;
 
-	if (enable == false) {
-		sdma_v4_0_gfx_stop(adev);
+	if (!enable) {
+		sdma_v4_0_gfx_enable(adev, enable);
 		sdma_v4_0_rlc_stop(adev);
+		if (adev->sdma.has_page_queue)
+			sdma_v4_0_page_stop(adev);
 	}
 
 	for (i = 0; i < adev->sdma.num_instances; i++) {
-		f32_cntl = RREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_F32_CNTL));
+		f32_cntl = RREG32_SDMA(i, mmSDMA0_F32_CNTL);
 		f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_F32_CNTL, HALT, enable ? 0 : 1);
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_F32_CNTL), f32_cntl);
+		WREG32_SDMA(i, mmSDMA0_F32_CNTL, f32_cntl);
 	}
 }
 
+/*
+ * sdma_v4_0_rb_cntl - get parameters for rb_cntl
+ */
+static uint32_t sdma_v4_0_rb_cntl(struct amdgpu_ring *ring, uint32_t rb_cntl)
+{
+	/* Set ring buffer size in dwords */
+	uint32_t rb_bufsz = order_base_2(ring->ring_size / 4);
+
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SIZE, rb_bufsz);
+#ifdef __BIG_ENDIAN
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SWAP_ENABLE, 1);
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL,
+				RPTR_WRITEBACK_SWAP_ENABLE, 1);
+#endif
+	return rb_cntl;
+}
+
 /**
  * sdma_v4_0_gfx_resume - setup and start the async dma engines
  *
  * @adev: amdgpu_device pointer
+ * @i: instance to resume
  *
  * Set up the gfx DMA ring buffers and enable them (VEGA10).
  * Returns 0 for success, error for failure.
  */
-static int sdma_v4_0_gfx_resume(struct amdgpu_device *adev)
+static void sdma_v4_0_gfx_resume(struct amdgpu_device *adev, unsigned int i)
 {
-	struct amdgpu_ring *ring;
+	struct amdgpu_ring *ring = &adev->sdma.instance[i].ring;
 	u32 rb_cntl, ib_cntl, wptr_poll_cntl;
-	u32 rb_bufsz;
-	u32 wb_offset;
 	u32 doorbell;
 	u32 doorbell_offset;
-	u32 temp;
 	u64 wptr_gpu_addr;
-	int i, r;
 
-	for (i = 0; i < adev->sdma.num_instances; i++) {
-		ring = &adev->sdma.instance[i].ring;
-		wb_offset = (ring->rptr_offs * 4);
+	rb_cntl = RREG32_SDMA(i, mmSDMA0_GFX_RB_CNTL);
+	rb_cntl = sdma_v4_0_rb_cntl(ring, rb_cntl);
+	WREG32_SDMA(i, mmSDMA0_GFX_RB_CNTL, rb_cntl);
 
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_SEM_WAIT_FAIL_TIMER_CNTL), 0);
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32_SDMA(i, mmSDMA0_GFX_RB_RPTR, 0);
+	WREG32_SDMA(i, mmSDMA0_GFX_RB_RPTR_HI, 0);
+	WREG32_SDMA(i, mmSDMA0_GFX_RB_WPTR, 0);
+	WREG32_SDMA(i, mmSDMA0_GFX_RB_WPTR_HI, 0);
 
-		/* Set ring buffer size in dwords */
-		rb_bufsz = order_base_2(ring->ring_size / 4);
-		rb_cntl = RREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_CNTL));
-		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SIZE, rb_bufsz);
-#ifdef __BIG_ENDIAN
-		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SWAP_ENABLE, 1);
-		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL,
-					RPTR_WRITEBACK_SWAP_ENABLE, 1);
-#endif
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
+	/* set the wb address whether it's enabled or not */
+	WREG32_SDMA(i, mmSDMA0_GFX_RB_RPTR_ADDR_HI,
+	       upper_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFF);
+	WREG32_SDMA(i, mmSDMA0_GFX_RB_RPTR_ADDR_LO,
+	       lower_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFC);
 
-		/* Initialize the ring buffer's read and write pointers */
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_RPTR), 0);
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_RPTR_HI), 0);
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_WPTR), 0);
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_WPTR_HI), 0);
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL,
+				RPTR_WRITEBACK_ENABLE, 1);
 
-		/* set the wb address whether it's enabled or not */
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_RPTR_ADDR_HI),
-		       upper_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFF);
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_RPTR_ADDR_LO),
-		       lower_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFC);
+	WREG32_SDMA(i, mmSDMA0_GFX_RB_BASE, ring->gpu_addr >> 8);
+	WREG32_SDMA(i, mmSDMA0_GFX_RB_BASE_HI, ring->gpu_addr >> 40);
 
-		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RPTR_WRITEBACK_ENABLE, 1);
+	ring->wptr = 0;
 
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_BASE), ring->gpu_addr >> 8);
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_BASE_HI), ring->gpu_addr >> 40);
+	/* before programing wptr to a less value, need set minor_ptr_update first */
+	WREG32_SDMA(i, mmSDMA0_GFX_MINOR_PTR_UPDATE, 1);
 
-		ring->wptr = 0;
+	doorbell = RREG32_SDMA(i, mmSDMA0_GFX_DOORBELL);
+	doorbell_offset = RREG32_SDMA(i, mmSDMA0_GFX_DOORBELL_OFFSET);
 
-		/* before programing wptr to a less value, need set minor_ptr_update first */
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_MINOR_PTR_UPDATE), 1);
+	doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE,
+				 ring->use_doorbell);
+	doorbell_offset = REG_SET_FIELD(doorbell_offset,
+					SDMA0_GFX_DOORBELL_OFFSET,
+					OFFSET, ring->doorbell_index);
+	WREG32_SDMA(i, mmSDMA0_GFX_DOORBELL, doorbell);
+	WREG32_SDMA(i, mmSDMA0_GFX_DOORBELL_OFFSET, doorbell_offset);
+
+	sdma_v4_0_ring_set_wptr(ring);
+
+	/* set minor_ptr_update to 0 after wptr programed */
+	WREG32_SDMA(i, mmSDMA0_GFX_MINOR_PTR_UPDATE, 0);
+
+	/* setup the wptr shadow polling */
+	wptr_gpu_addr = ring->wptr_gpu_addr;
+	WREG32_SDMA(i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_LO,
+		    lower_32_bits(wptr_gpu_addr));
+	WREG32_SDMA(i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_HI,
+		    upper_32_bits(wptr_gpu_addr));
+	wptr_poll_cntl = RREG32_SDMA(i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL);
+	wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl,
+				       SDMA0_GFX_RB_WPTR_POLL_CNTL,
+				       F32_POLL_ENABLE, amdgpu_sriov_vf(adev)? 1 : 0);
+	WREG32_SDMA(i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL, wptr_poll_cntl);
+
+	/* enable DMA RB */
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 1);
+	WREG32_SDMA(i, mmSDMA0_GFX_RB_CNTL, rb_cntl);
+
+	ib_cntl = RREG32_SDMA(i, mmSDMA0_GFX_IB_CNTL);
+	ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 1);
+#ifdef __BIG_ENDIAN
+	ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_SWAP_ENABLE, 1);
+#endif
+	/* enable DMA IBs */
+	WREG32_SDMA(i, mmSDMA0_GFX_IB_CNTL, ib_cntl);
+}
 
-		if (!amdgpu_sriov_vf(adev)) { /* only bare-metal use register write for wptr */
-			WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_WPTR), lower_32_bits(ring->wptr) << 2);
-			WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_WPTR_HI), upper_32_bits(ring->wptr) << 2);
-		}
+/**
+ * sdma_v4_0_page_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ * @i: instance to resume
+ *
+ * Set up the page DMA ring buffers and enable them (VEGA10).
+ * Returns 0 for success, error for failure.
+ */
+static void sdma_v4_0_page_resume(struct amdgpu_device *adev, unsigned int i)
+{
+	struct amdgpu_ring *ring = &adev->sdma.instance[i].page;
+	u32 rb_cntl, ib_cntl, wptr_poll_cntl;
+	u32 doorbell;
+	u32 doorbell_offset;
+	u64 wptr_gpu_addr;
 
-		doorbell = RREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_DOORBELL));
-		doorbell_offset = RREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_DOORBELL_OFFSET));
+	rb_cntl = RREG32_SDMA(i, mmSDMA0_PAGE_RB_CNTL);
+	rb_cntl = sdma_v4_0_rb_cntl(ring, rb_cntl);
+	WREG32_SDMA(i, mmSDMA0_PAGE_RB_CNTL, rb_cntl);
 
-		if (ring->use_doorbell) {
-			doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, 1);
-			doorbell_offset = REG_SET_FIELD(doorbell_offset, SDMA0_GFX_DOORBELL_OFFSET,
-					OFFSET, ring->doorbell_index);
-		} else {
-			doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, 0);
-		}
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_DOORBELL), doorbell);
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_DOORBELL_OFFSET), doorbell_offset);
-		if (adev->flags & AMD_IS_APU)
-			nbio_v7_0_sdma_doorbell_range(adev, i, ring->use_doorbell, ring->doorbell_index);
-		else
-			nbio_v6_1_sdma_doorbell_range(adev, i, ring->use_doorbell, ring->doorbell_index);
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32_SDMA(i, mmSDMA0_PAGE_RB_RPTR, 0);
+	WREG32_SDMA(i, mmSDMA0_PAGE_RB_RPTR_HI, 0);
+	WREG32_SDMA(i, mmSDMA0_PAGE_RB_WPTR, 0);
+	WREG32_SDMA(i, mmSDMA0_PAGE_RB_WPTR_HI, 0);
 
-		if (amdgpu_sriov_vf(adev))
-			sdma_v4_0_ring_set_wptr(ring);
+	/* set the wb address whether it's enabled or not */
+	WREG32_SDMA(i, mmSDMA0_PAGE_RB_RPTR_ADDR_HI,
+	       upper_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFF);
+	WREG32_SDMA(i, mmSDMA0_PAGE_RB_RPTR_ADDR_LO,
+	       lower_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFC);
 
-		/* set minor_ptr_update to 0 after wptr programed */
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_MINOR_PTR_UPDATE), 0);
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_PAGE_RB_CNTL,
+				RPTR_WRITEBACK_ENABLE, 1);
 
-		/* set utc l1 enable flag always to 1 */
-		temp = RREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_CNTL));
-		temp = REG_SET_FIELD(temp, SDMA0_CNTL, UTC_L1_ENABLE, 1);
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_CNTL), temp);
+	WREG32_SDMA(i, mmSDMA0_PAGE_RB_BASE, ring->gpu_addr >> 8);
+	WREG32_SDMA(i, mmSDMA0_PAGE_RB_BASE_HI, ring->gpu_addr >> 40);
 
-		if (!amdgpu_sriov_vf(adev)) {
-			/* unhalt engine */
-			temp = RREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_F32_CNTL));
-			temp = REG_SET_FIELD(temp, SDMA0_F32_CNTL, HALT, 0);
-			WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_F32_CNTL), temp);
-		}
+	ring->wptr = 0;
 
-		/* setup the wptr shadow polling */
-		wptr_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_LO),
-		       lower_32_bits(wptr_gpu_addr));
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_HI),
-		       upper_32_bits(wptr_gpu_addr));
-		wptr_poll_cntl = RREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL));
-		if (amdgpu_sriov_vf(adev))
-			wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl, SDMA0_GFX_RB_WPTR_POLL_CNTL, F32_POLL_ENABLE, 1);
-		else
-			wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl, SDMA0_GFX_RB_WPTR_POLL_CNTL, F32_POLL_ENABLE, 0);
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL), wptr_poll_cntl);
+	/* before programing wptr to a less value, need set minor_ptr_update first */
+	WREG32_SDMA(i, mmSDMA0_PAGE_MINOR_PTR_UPDATE, 1);
 
-		/* enable DMA RB */
-		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 1);
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
+	doorbell = RREG32_SDMA(i, mmSDMA0_PAGE_DOORBELL);
+	doorbell_offset = RREG32_SDMA(i, mmSDMA0_PAGE_DOORBELL_OFFSET);
 
-		ib_cntl = RREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_IB_CNTL));
-		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 1);
+	doorbell = REG_SET_FIELD(doorbell, SDMA0_PAGE_DOORBELL, ENABLE,
+				 ring->use_doorbell);
+	doorbell_offset = REG_SET_FIELD(doorbell_offset,
+					SDMA0_PAGE_DOORBELL_OFFSET,
+					OFFSET, ring->doorbell_index);
+	WREG32_SDMA(i, mmSDMA0_PAGE_DOORBELL, doorbell);
+	WREG32_SDMA(i, mmSDMA0_PAGE_DOORBELL_OFFSET, doorbell_offset);
+
+	/* paging queue doorbell range is setup at sdma_v4_0_gfx_resume */
+	sdma_v4_0_page_ring_set_wptr(ring);
+
+	/* set minor_ptr_update to 0 after wptr programed */
+	WREG32_SDMA(i, mmSDMA0_PAGE_MINOR_PTR_UPDATE, 0);
+
+	/* setup the wptr shadow polling */
+	wptr_gpu_addr = ring->wptr_gpu_addr;
+	WREG32_SDMA(i, mmSDMA0_PAGE_RB_WPTR_POLL_ADDR_LO,
+		    lower_32_bits(wptr_gpu_addr));
+	WREG32_SDMA(i, mmSDMA0_PAGE_RB_WPTR_POLL_ADDR_HI,
+		    upper_32_bits(wptr_gpu_addr));
+	wptr_poll_cntl = RREG32_SDMA(i, mmSDMA0_PAGE_RB_WPTR_POLL_CNTL);
+	wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl,
+				       SDMA0_PAGE_RB_WPTR_POLL_CNTL,
+				       F32_POLL_ENABLE, amdgpu_sriov_vf(adev)? 1 : 0);
+	WREG32_SDMA(i, mmSDMA0_PAGE_RB_WPTR_POLL_CNTL, wptr_poll_cntl);
+
+	/* enable DMA RB */
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_PAGE_RB_CNTL, RB_ENABLE, 1);
+	WREG32_SDMA(i, mmSDMA0_PAGE_RB_CNTL, rb_cntl);
+
+	ib_cntl = RREG32_SDMA(i, mmSDMA0_PAGE_IB_CNTL);
+	ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_PAGE_IB_CNTL, IB_ENABLE, 1);
 #ifdef __BIG_ENDIAN
-		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_SWAP_ENABLE, 1);
+	ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_PAGE_IB_CNTL, IB_SWAP_ENABLE, 1);
 #endif
-		/* enable DMA IBs */
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_IB_CNTL), ib_cntl);
-
-		ring->ready = true;
-
-		if (amdgpu_sriov_vf(adev)) { /* bare-metal sequence doesn't need below to lines */
-			sdma_v4_0_ctx_switch_enable(adev, true);
-			sdma_v4_0_enable(adev, true);
-		}
-
-		r = amdgpu_ring_test_ring(ring);
-		if (r) {
-			ring->ready = false;
-			return r;
-		}
-
-		if (adev->mman.buffer_funcs_ring == ring)
-			amdgpu_ttm_set_active_vram_size(adev, adev->mc.real_vram_size);
-
-	}
-
-	return 0;
+	/* enable DMA IBs */
+	WREG32_SDMA(i, mmSDMA0_PAGE_IB_CNTL, ib_cntl);
 }
 
 static void
@@ -714,7 +1254,7 @@ sdma_v4_1_update_power_gating(struct amdgpu_device *adev, bool enable)
 	uint32_t def, data;
 
 	if (enable && (adev->pg_flags & AMD_PG_SUPPORT_SDMA)) {
-		/* disable idle interrupt */
+		/* enable idle interrupt */
 		def = data = RREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_CNTL));
 		data |= SDMA0_CNTL__CTXEMPTY_INT_ENABLE_MASK;
 
@@ -761,8 +1301,10 @@ static void sdma_v4_0_init_pg(struct amdgpu_device *adev)
 	if (!(adev->pg_flags & AMD_PG_SUPPORT_SDMA))
 		return;
 
-	switch (adev->asic_type) {
-	case CHIP_RAVEN:
+	switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) {
+	case IP_VERSION(4, 1, 0):
+        case IP_VERSION(4, 1, 1):
+	case IP_VERSION(4, 1, 2):
 		sdma_v4_1_init_power_gating(adev);
 		sdma_v4_1_update_power_gating(adev, true);
 		break;
@@ -816,12 +1358,14 @@ static int sdma_v4_0_load_microcode(struct amdgpu_device *adev)
 			(adev->sdma.instance[i].fw->data +
 				le32_to_cpu(hdr->header.ucode_array_offset_bytes));
 
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_UCODE_ADDR), 0);
+		WREG32_SDMA(i, mmSDMA0_UCODE_ADDR, 0);
 
 		for (j = 0; j < fw_size; j++)
-			WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_UCODE_DATA), le32_to_cpup(fw_data++));
+			WREG32_SDMA(i, mmSDMA0_UCODE_DATA,
+				    le32_to_cpup(fw_data++));
 
-		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_UCODE_ADDR), adev->sdma.instance[i].fw_version);
+		WREG32_SDMA(i, mmSDMA0_UCODE_ADDR,
+			    adev->sdma.instance[i].fw_version);
 	}
 
 	return 0;
@@ -837,33 +1381,72 @@ static int sdma_v4_0_load_microcode(struct amdgpu_device *adev)
  */
 static int sdma_v4_0_start(struct amdgpu_device *adev)
 {
-	int r = 0;
+	struct amdgpu_ring *ring;
+	int i, r = 0;
 
 	if (amdgpu_sriov_vf(adev)) {
 		sdma_v4_0_ctx_switch_enable(adev, false);
 		sdma_v4_0_enable(adev, false);
+	} else {
 
-		/* set RB registers */
-		r = sdma_v4_0_gfx_resume(adev);
-		return r;
+		if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
+			r = sdma_v4_0_load_microcode(adev);
+			if (r)
+				return r;
+		}
+
+		/* unhalt the MEs */
+		sdma_v4_0_enable(adev, true);
+		/* enable sdma ring preemption */
+		sdma_v4_0_ctx_switch_enable(adev, true);
 	}
 
-	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
-		r = sdma_v4_0_load_microcode(adev);
+	/* start the gfx rings and rlc compute queues */
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		uint32_t temp;
+
+		WREG32_SDMA(i, mmSDMA0_SEM_WAIT_FAIL_TIMER_CNTL, 0);
+		sdma_v4_0_gfx_resume(adev, i);
+		if (adev->sdma.has_page_queue)
+			sdma_v4_0_page_resume(adev, i);
+
+		/* set utc l1 enable flag always to 1 */
+		temp = RREG32_SDMA(i, mmSDMA0_CNTL);
+		temp = REG_SET_FIELD(temp, SDMA0_CNTL, UTC_L1_ENABLE, 1);
+		WREG32_SDMA(i, mmSDMA0_CNTL, temp);
+
+		if (!amdgpu_sriov_vf(adev)) {
+			/* unhalt engine */
+			temp = RREG32_SDMA(i, mmSDMA0_F32_CNTL);
+			temp = REG_SET_FIELD(temp, SDMA0_F32_CNTL, HALT, 0);
+			WREG32_SDMA(i, mmSDMA0_F32_CNTL, temp);
+		}
+	}
+
+	if (amdgpu_sriov_vf(adev)) {
+		sdma_v4_0_ctx_switch_enable(adev, true);
+		sdma_v4_0_enable(adev, true);
+	} else {
+		r = sdma_v4_0_rlc_resume(adev);
 		if (r)
 			return r;
 	}
 
-	/* unhalt the MEs */
-	sdma_v4_0_enable(adev, true);
-	/* enable sdma ring preemption */
-	sdma_v4_0_ctx_switch_enable(adev, true);
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+
+		r = amdgpu_ring_test_helper(ring);
+		if (r)
+			return r;
 
-	/* start the gfx rings and rlc compute queues */
-	r = sdma_v4_0_gfx_resume(adev);
-	if (r)
-		return r;
-	r = sdma_v4_0_rlc_resume(adev);
+		if (adev->sdma.has_page_queue) {
+			struct amdgpu_ring *page = &adev->sdma.instance[i].page;
+
+			r = amdgpu_ring_test_helper(page);
+			if (r)
+				return r;
+		}
+	}
 
 	return r;
 }
@@ -886,22 +1469,17 @@ static int sdma_v4_0_ring_test_ring(struct amdgpu_ring *ring)
 	u32 tmp;
 	u64 gpu_addr;
 
-	r = amdgpu_wb_get(adev, &index);
-	if (r) {
-		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r)
 		return r;
-	}
 
 	gpu_addr = adev->wb.gpu_addr + (index * 4);
 	tmp = 0xCAFEDEAD;
 	adev->wb.wb[index] = cpu_to_le32(tmp);
 
 	r = amdgpu_ring_alloc(ring, 5);
-	if (r) {
-		DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r);
-		amdgpu_wb_free(adev, index);
-		return r;
-	}
+	if (r)
+		goto error_free_wb;
 
 	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
 			  SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR));
@@ -915,18 +1493,14 @@ static int sdma_v4_0_ring_test_ring(struct amdgpu_ring *ring)
 		tmp = le32_to_cpu(adev->wb.wb[index]);
 		if (tmp == 0xDEADBEEF)
 			break;
-		DRM_UDELAY(1);
+		udelay(1);
 	}
 
-	if (i < adev->usec_timeout) {
-		DRM_INFO("ring test on %d succeeded in %d usecs\n", ring->idx, i);
-	} else {
-		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
-			  ring->idx, tmp);
-		r = -EINVAL;
-	}
-	amdgpu_wb_free(adev, index);
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
 
+error_free_wb:
+	amdgpu_device_wb_free(adev, index);
 	return r;
 }
 
@@ -934,6 +1508,7 @@ static int sdma_v4_0_ring_test_ring(struct amdgpu_ring *ring)
  * sdma_v4_0_ring_test_ib - test an IB on the DMA engine
  *
  * @ring: amdgpu_ring structure holding ring information
+ * @timeout: timeout value in jiffies, or MAX_SCHEDULE_TIMEOUT
  *
  * Test a simple IB in the DMA ring (VEGA10).
  * Returns 0 on success, error on failure.
@@ -948,21 +1523,18 @@ static int sdma_v4_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 	u32 tmp = 0;
 	u64 gpu_addr;
 
-	r = amdgpu_wb_get(adev, &index);
-	if (r) {
-		dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r)
 		return r;
-	}
 
 	gpu_addr = adev->wb.gpu_addr + (index * 4);
 	tmp = 0xCAFEDEAD;
 	adev->wb.wb[index] = cpu_to_le32(tmp);
 	memset(&ib, 0, sizeof(ib));
-	r = amdgpu_ib_get(adev, NULL, 256, &ib);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
+	r = amdgpu_ib_get(adev, NULL, 256,
+					AMDGPU_IB_POOL_DIRECT, &ib);
+	if (r)
 		goto err0;
-	}
 
 	ib.ptr[0] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
 		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR);
@@ -981,26 +1553,22 @@ static int sdma_v4_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 
 	r = dma_fence_wait_timeout(f, false, timeout);
 	if (r == 0) {
-		DRM_ERROR("amdgpu: IB test timed out\n");
 		r = -ETIMEDOUT;
 		goto err1;
 	} else if (r < 0) {
-		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
 		goto err1;
 	}
 	tmp = le32_to_cpu(adev->wb.wb[index]);
-	if (tmp == 0xDEADBEEF) {
-		DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
+	if (tmp == 0xDEADBEEF)
 		r = 0;
-	} else {
-		DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp);
+	else
 		r = -EINVAL;
-	}
+
 err1:
-	amdgpu_ib_free(adev, &ib, NULL);
+	amdgpu_ib_free(&ib, NULL);
 	dma_fence_put(f);
 err0:
-	amdgpu_wb_free(adev, index);
+	amdgpu_device_wb_free(adev, index);
 	return r;
 }
 
@@ -1037,10 +1605,9 @@ static void sdma_v4_0_vm_copy_pte(struct amdgpu_ib *ib,
  *
  * @ib: indirect buffer to fill with commands
  * @pe: addr of the page entry
- * @addr: dst addr to write into pe
+ * @value: dst addr to write into pe
  * @count: number of page entries to update
  * @incr: increase next addr by incr bytes
- * @flags: access flags
  *
  * Update PTEs by writing them manually using sDMA (VEGA10).
  */
@@ -1095,16 +1662,16 @@ static void sdma_v4_0_vm_set_pte_pde(struct amdgpu_ib *ib,
 /**
  * sdma_v4_0_ring_pad_ib - pad the IB to the required number of dw
  *
+ * @ring: amdgpu_ring structure holding ring information
  * @ib: indirect buffer to fill with padding
- *
  */
 static void sdma_v4_0_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib)
 {
-	struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring);
+	struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring);
 	u32 pad_count;
 	int i;
 
-	pad_count = (8 - (ib->length_dw & 0x7)) % 8;
+	pad_count = (-ib->length_dw) & 7;
 	for (i = 0; i < pad_count; i++)
 		if (sdma && sdma->burst_nop && (i == 0))
 			ib->ptr[ib->length_dw++] =
@@ -1129,16 +1696,10 @@ static void sdma_v4_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
 	uint64_t addr = ring->fence_drv.gpu_addr;
 
 	/* wait for idle */
-	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
-			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(0) |
-			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3) | /* equal */
-			  SDMA_PKT_POLL_REGMEM_HEADER_MEM_POLL(1));
-	amdgpu_ring_write(ring, addr & 0xfffffffc);
-	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xffffffff);
-	amdgpu_ring_write(ring, seq); /* reference */
-	amdgpu_ring_write(ring, 0xfffffff); /* mask */
-	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
-			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(4)); /* retry count, poll interval */
+	sdma_v4_0_wait_reg_mem(ring, 1, 0,
+			       addr & 0xfffffffc,
+			       upper_32_bits(addr) & 0xffffffff,
+			       seq, 0xffffffff, 4);
 }
 
 
@@ -1146,89 +1707,142 @@ static void sdma_v4_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
  * sdma_v4_0_ring_emit_vm_flush - vm flush using sDMA
  *
  * @ring: amdgpu_ring pointer
- * @vm: amdgpu_vm pointer
+ * @vmid: vmid number to use
+ * @pd_addr: address
  *
  * Update the page table base and flush the VM TLB
  * using sDMA (VEGA10).
  */
 static void sdma_v4_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
-					 unsigned vm_id, uint64_t pd_addr)
+					 unsigned vmid, uint64_t pd_addr)
 {
-	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
-	uint32_t req = ring->adev->gart.gart_funcs->get_invalidate_req(vm_id);
-	unsigned eng = ring->vm_inv_eng;
-
-	pd_addr = amdgpu_gart_get_vm_pde(ring->adev, pd_addr);
-	pd_addr |= AMDGPU_PTE_VALID;
-
-	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
-			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
-	amdgpu_ring_write(ring, hub->ctx0_ptb_addr_lo32 + vm_id * 2);
-	amdgpu_ring_write(ring, lower_32_bits(pd_addr));
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+}
 
+static void sdma_v4_0_ring_emit_wreg(struct amdgpu_ring *ring,
+				     uint32_t reg, uint32_t val)
+{
 	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
 			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
-	amdgpu_ring_write(ring, hub->ctx0_ptb_addr_hi32 + vm_id * 2);
-	amdgpu_ring_write(ring, upper_32_bits(pd_addr));
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, val);
+}
 
-	/* flush TLB */
-	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
-			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
-	amdgpu_ring_write(ring, hub->vm_inv_eng0_req + eng);
-	amdgpu_ring_write(ring, req);
+static void sdma_v4_0_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+					 uint32_t val, uint32_t mask)
+{
+	sdma_v4_0_wait_reg_mem(ring, 0, 0, reg, 0, val, mask, 10);
+}
 
-	/* wait for flush */
-	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
-			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(0) |
-			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3)); /* equal */
-	amdgpu_ring_write(ring, (hub->vm_inv_eng0_ack + eng) << 2);
-	amdgpu_ring_write(ring, 0);
-	amdgpu_ring_write(ring, 1 << vm_id); /* reference */
-	amdgpu_ring_write(ring, 1 << vm_id); /* mask */
-	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
-			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10));
+static bool sdma_v4_0_fw_support_paging_queue(struct amdgpu_device *adev)
+{
+	uint fw_version = adev->sdma.instance[0].fw_version;
+
+	switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) {
+	case IP_VERSION(4, 0, 0):
+		return fw_version >= 430;
+	case IP_VERSION(4, 0, 1):
+		/*return fw_version >= 31;*/
+		return false;
+	case IP_VERSION(4, 2, 0):
+		return fw_version >= 123;
+	default:
+		return false;
+	}
 }
 
-static int sdma_v4_0_early_init(void *handle)
+static int sdma_v4_0_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
 
-	if (adev->asic_type == CHIP_RAVEN)
-		adev->sdma.num_instances = 1;
-	else
-		adev->sdma.num_instances = 2;
+	r = sdma_v4_0_init_microcode(adev);
+	if (r)
+		return r;
+
+	/* TODO: Page queue breaks driver reload under SRIOV */
+	if ((amdgpu_ip_version(adev, SDMA0_HWIP, 0) == IP_VERSION(4, 0, 0)) &&
+	    amdgpu_sriov_vf((adev)))
+		adev->sdma.has_page_queue = false;
+	else if (sdma_v4_0_fw_support_paging_queue(adev))
+		adev->sdma.has_page_queue = true;
 
 	sdma_v4_0_set_ring_funcs(adev);
 	sdma_v4_0_set_buffer_funcs(adev);
 	sdma_v4_0_set_vm_pte_funcs(adev);
 	sdma_v4_0_set_irq_funcs(adev);
+	sdma_v4_0_set_ras_funcs(adev);
 
 	return 0;
 }
 
+static int sdma_v4_0_process_ras_data_cb(struct amdgpu_device *adev,
+		void *err_data,
+		struct amdgpu_iv_entry *entry);
 
-static int sdma_v4_0_sw_init(void *handle)
+static int sdma_v4_0_late_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	sdma_v4_0_setup_ulv(adev);
+
+	if (!amdgpu_persistent_edc_harvesting_supported(adev))
+		amdgpu_ras_reset_error_count(adev, AMDGPU_RAS_BLOCK__SDMA);
+
+	return 0;
+}
+
+static int sdma_v4_0_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_ring *ring;
 	int r, i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+	uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_4_0);
+	uint32_t *ptr;
 
 	/* SDMA trap event */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_SDMA0, 224,
-			      &adev->sdma.trap_irq);
-	if (r)
-		return r;
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		r = amdgpu_irq_add_id(adev, sdma_v4_0_seq_to_irq_id(i),
+				      SDMA0_4_0__SRCID__SDMA_TRAP,
+				      &adev->sdma.trap_irq);
+		if (r)
+			return r;
+	}
 
-	/* SDMA trap event */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_SDMA1, 224,
-			      &adev->sdma.trap_irq);
-	if (r)
-		return r;
+	/* SDMA SRAM ECC event */
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		r = amdgpu_irq_add_id(adev, sdma_v4_0_seq_to_irq_id(i),
+				      SDMA0_4_0__SRCID__SDMA_SRAM_ECC,
+				      &adev->sdma.ecc_irq);
+		if (r)
+			return r;
+	}
 
-	r = sdma_v4_0_init_microcode(adev);
-	if (r) {
-		DRM_ERROR("Failed to load sdma firmware!\n");
-		return r;
+	/* SDMA VM_HOLE/DOORBELL_INV/POLL_TIMEOUT/SRBM_WRITE_PROTECTION event*/
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		r = amdgpu_irq_add_id(adev, sdma_v4_0_seq_to_irq_id(i),
+				      SDMA0_4_0__SRCID__SDMA_VM_HOLE,
+				      &adev->sdma.vm_hole_irq);
+		if (r)
+			return r;
+
+		r = amdgpu_irq_add_id(adev, sdma_v4_0_seq_to_irq_id(i),
+				      SDMA0_4_0__SRCID__SDMA_DOORBELL_INVALID,
+				      &adev->sdma.doorbell_invalid_irq);
+		if (r)
+			return r;
+
+		r = amdgpu_irq_add_id(adev, sdma_v4_0_seq_to_irq_id(i),
+				      SDMA0_4_0__SRCID__SDMA_POLL_TIMEOUT,
+				      &adev->sdma.pool_timeout_irq);
+		if (r)
+			return r;
+
+		r = amdgpu_irq_add_id(adev, sdma_v4_0_seq_to_irq_id(i),
+				      SDMA0_4_0__SRCID__SDMA_SRBMWRITE,
+				      &adev->sdma.srbm_write_irq);
+		if (r)
+			return r;
 	}
 
 	for (i = 0; i < adev->sdma.num_instances; i++) {
@@ -1236,83 +1850,178 @@ static int sdma_v4_0_sw_init(void *handle)
 		ring->ring_obj = NULL;
 		ring->use_doorbell = true;
 
-		DRM_INFO("use_doorbell being set to: [%s]\n",
+		DRM_DEBUG("SDMA %d use_doorbell being set to: [%s]\n", i,
 				ring->use_doorbell?"true":"false");
 
-		ring->doorbell_index = (i == 0) ?
-			(AMDGPU_DOORBELL64_sDMA_ENGINE0 << 1) //get DWORD offset
-			: (AMDGPU_DOORBELL64_sDMA_ENGINE1 << 1); // get DWORD offset
+		/* doorbell size is 2 dwords, get DWORD offset */
+		ring->doorbell_index = adev->doorbell_index.sdma_engine[i] << 1;
+
+		/*
+		 * On Arcturus, SDMA instance 5~7 has a different vmhub
+		 * type(AMDGPU_MMHUB1).
+		 */
+		if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) ==
+			    IP_VERSION(4, 2, 2) &&
+		    i >= 5)
+			ring->vm_hub = AMDGPU_MMHUB1(0);
+		else
+			ring->vm_hub = AMDGPU_MMHUB0(0);
 
 		sprintf(ring->name, "sdma%d", i);
-		r = amdgpu_ring_init(adev, ring, 1024,
-				     &adev->sdma.trap_irq,
-				     (i == 0) ?
-				     AMDGPU_SDMA_IRQ_TRAP0 :
-				     AMDGPU_SDMA_IRQ_TRAP1);
+		r = amdgpu_ring_init(adev, ring, 1024, &adev->sdma.trap_irq,
+				     AMDGPU_SDMA_IRQ_INSTANCE0 + i,
+				     AMDGPU_RING_PRIO_DEFAULT, NULL);
 		if (r)
 			return r;
+
+		if (adev->sdma.has_page_queue) {
+			ring = &adev->sdma.instance[i].page;
+			ring->ring_obj = NULL;
+			ring->use_doorbell = true;
+
+			/* paging queue use same doorbell index/routing as gfx queue
+			 * with 0x400 (4096 dwords) offset on second doorbell page
+			 */
+			if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) >=
+				    IP_VERSION(4, 0, 0) &&
+			    amdgpu_ip_version(adev, SDMA0_HWIP, 0) <
+				    IP_VERSION(4, 2, 0)) {
+				ring->doorbell_index =
+					adev->doorbell_index.sdma_engine[i] << 1;
+				ring->doorbell_index += 0x400;
+			} else {
+				/* From vega20, the sdma_doorbell_range in 1st
+				 * doorbell page is reserved for page queue.
+				 */
+				ring->doorbell_index =
+					(adev->doorbell_index.sdma_engine[i] + 1) << 1;
+			}
+
+			if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) ==
+				    IP_VERSION(4, 2, 2) &&
+			    i >= 5)
+				ring->vm_hub = AMDGPU_MMHUB1(0);
+			else
+				ring->vm_hub = AMDGPU_MMHUB0(0);
+
+			sprintf(ring->name, "page%d", i);
+			r = amdgpu_ring_init(adev, ring, 1024,
+					     &adev->sdma.trap_irq,
+					     AMDGPU_SDMA_IRQ_INSTANCE0 + i,
+					     AMDGPU_RING_PRIO_DEFAULT, NULL);
+			if (r)
+				return r;
+		}
 	}
 
+	if (amdgpu_sdma_ras_sw_init(adev)) {
+		dev_err(adev->dev, "Failed to initialize sdma ras block!\n");
+		return -EINVAL;
+	}
+
+	/* Allocate memory for SDMA IP Dump buffer */
+	ptr = kcalloc(adev->sdma.num_instances * reg_count, sizeof(uint32_t), GFP_KERNEL);
+	if (ptr)
+		adev->sdma.ip_dump = ptr;
+	else
+		DRM_ERROR("Failed to allocated memory for SDMA IP Dump\n");
+
 	return r;
 }
 
-static int sdma_v4_0_sw_fini(void *handle)
+static int sdma_v4_0_sw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int i;
 
-	for (i = 0; i < adev->sdma.num_instances; i++)
+	for (i = 0; i < adev->sdma.num_instances; i++) {
 		amdgpu_ring_fini(&adev->sdma.instance[i].ring);
+		if (adev->sdma.has_page_queue)
+			amdgpu_ring_fini(&adev->sdma.instance[i].page);
+	}
+
+	if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) == IP_VERSION(4, 2, 2) ||
+	    amdgpu_ip_version(adev, SDMA0_HWIP, 0) == IP_VERSION(4, 4, 0))
+		amdgpu_sdma_destroy_inst_ctx(adev, true);
+	else
+		amdgpu_sdma_destroy_inst_ctx(adev, false);
+
+	kfree(adev->sdma.ip_dump);
 
 	return 0;
 }
 
-static int sdma_v4_0_hw_init(void *handle)
+static int sdma_v4_0_hw_init(struct amdgpu_ip_block *ip_block)
 {
-	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	sdma_v4_0_init_golden_registers(adev);
+	if (adev->flags & AMD_IS_APU)
+		amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_SDMA, false, 0);
 
-	r = sdma_v4_0_start(adev);
+	if (!amdgpu_sriov_vf(adev))
+		sdma_v4_0_init_golden_registers(adev);
 
-	return r;
+	return sdma_v4_0_start(adev);
 }
 
-static int sdma_v4_0_hw_fini(void *handle)
+static int sdma_v4_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+	int i;
 
 	if (amdgpu_sriov_vf(adev))
 		return 0;
 
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__SDMA)) {
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			amdgpu_irq_put(adev, &adev->sdma.ecc_irq,
+				       AMDGPU_SDMA_IRQ_INSTANCE0 + i);
+		}
+	}
+
 	sdma_v4_0_ctx_switch_enable(adev, false);
 	sdma_v4_0_enable(adev, false);
 
+	if (adev->flags & AMD_IS_APU)
+		amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_SDMA, true, 0);
+
 	return 0;
 }
 
-static int sdma_v4_0_suspend(void *handle)
+static int sdma_v4_0_suspend(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	return sdma_v4_0_hw_fini(adev);
+	/* SMU saves SDMA state for us */
+	if (adev->in_s0ix) {
+		sdma_v4_0_gfx_enable(adev, false);
+		return 0;
+	}
+
+	return sdma_v4_0_hw_fini(ip_block);
 }
 
-static int sdma_v4_0_resume(void *handle)
+static int sdma_v4_0_resume(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	return sdma_v4_0_hw_init(adev);
+	/* SMU restores SDMA state for us */
+	if (adev->in_s0ix) {
+		sdma_v4_0_enable(adev, true);
+		sdma_v4_0_gfx_enable(adev, true);
+		return 0;
+	}
+
+	return sdma_v4_0_hw_init(ip_block);
 }
 
-static bool sdma_v4_0_is_idle(void *handle)
+static bool sdma_v4_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 i;
 
 	for (i = 0; i < adev->sdma.num_instances; i++) {
-		u32 tmp = RREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_STATUS_REG));
+		u32 tmp = RREG32_SDMA(i, mmSDMA0_STATUS_REG);
 
 		if (!(tmp & SDMA0_STATUS_REG__IDLE_MASK))
 			return false;
@@ -1321,24 +2030,26 @@ static bool sdma_v4_0_is_idle(void *handle)
 	return true;
 }
 
-static int sdma_v4_0_wait_for_idle(void *handle)
+static int sdma_v4_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
-	unsigned i;
-	u32 sdma0, sdma1;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	unsigned i, j;
+	u32 sdma[AMDGPU_MAX_SDMA_INSTANCES];
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		sdma0 = RREG32(sdma_v4_0_get_reg_offset(0, mmSDMA0_STATUS_REG));
-		sdma1 = RREG32(sdma_v4_0_get_reg_offset(1, mmSDMA0_STATUS_REG));
-
-		if (sdma0 & sdma1 & SDMA0_STATUS_REG__IDLE_MASK)
+		for (j = 0; j < adev->sdma.num_instances; j++) {
+			sdma[j] = RREG32_SDMA(j, mmSDMA0_STATUS_REG);
+			if (!(sdma[j] & SDMA0_STATUS_REG__IDLE_MASK))
+				break;
+		}
+		if (j == adev->sdma.num_instances)
 			return 0;
 		udelay(1);
 	}
 	return -ETIMEDOUT;
 }
 
-static int sdma_v4_0_soft_reset(void *handle)
+static int sdma_v4_0_soft_reset(struct amdgpu_ip_block *ip_block)
 {
 	/* todo */
 
@@ -1352,14 +2063,10 @@ static int sdma_v4_0_set_trap_irq_state(struct amdgpu_device *adev,
 {
 	u32 sdma_cntl;
 
-	u32 reg_offset = (type == AMDGPU_SDMA_IRQ_TRAP0) ?
-		sdma_v4_0_get_reg_offset(0, mmSDMA0_CNTL) :
-		sdma_v4_0_get_reg_offset(1, mmSDMA0_CNTL);
-
-	sdma_cntl = RREG32(reg_offset);
+	sdma_cntl = RREG32_SDMA(type, mmSDMA0_CNTL);
 	sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE,
 		       state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
-	WREG32(reg_offset, sdma_cntl);
+	WREG32_SDMA(type, mmSDMA0_CNTL, sdma_cntl);
 
 	return 0;
 }
@@ -1368,114 +2075,197 @@ static int sdma_v4_0_process_trap_irq(struct amdgpu_device *adev,
 				      struct amdgpu_irq_src *source,
 				      struct amdgpu_iv_entry *entry)
 {
+	int instance;
+
 	DRM_DEBUG("IH: SDMA trap\n");
-	switch (entry->client_id) {
-	case AMDGPU_IH_CLIENTID_SDMA0:
-		switch (entry->ring_id) {
-		case 0:
-			amdgpu_fence_process(&adev->sdma.instance[0].ring);
-			break;
-		case 1:
-			/* XXX compute */
-			break;
-		case 2:
-			/* XXX compute */
-			break;
-		case 3:
-			/* XXX page queue*/
-			break;
-		}
+	instance = sdma_v4_0_irq_id_to_seq(entry->client_id);
+	if (instance < 0)
+		return instance;
+
+	switch (entry->ring_id) {
+	case 0:
+		amdgpu_fence_process(&adev->sdma.instance[instance].ring);
 		break;
-	case AMDGPU_IH_CLIENTID_SDMA1:
-		switch (entry->ring_id) {
-		case 0:
-			amdgpu_fence_process(&adev->sdma.instance[1].ring);
-			break;
-		case 1:
-			/* XXX compute */
-			break;
-		case 2:
-			/* XXX compute */
-			break;
-		case 3:
-			/* XXX page queue*/
-			break;
-		}
+	case 1:
+		if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) ==
+		    IP_VERSION(4, 2, 0))
+			amdgpu_fence_process(&adev->sdma.instance[instance].page);
+		break;
+	case 2:
+		/* XXX compute */
+		break;
+	case 3:
+		if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) !=
+		    IP_VERSION(4, 2, 0))
+			amdgpu_fence_process(&adev->sdma.instance[instance].page);
 		break;
 	}
 	return 0;
 }
 
+static int sdma_v4_0_process_ras_data_cb(struct amdgpu_device *adev,
+		void *err_data,
+		struct amdgpu_iv_entry *entry)
+{
+	int instance;
+
+	/* When “Full RAS” is enabled, the per-IP interrupt sources should
+	 * be disabled and the driver should only look for the aggregated
+	 * interrupt via sync flood
+	 */
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
+		goto out;
+
+	instance = sdma_v4_0_irq_id_to_seq(entry->client_id);
+	if (instance < 0)
+		goto out;
+
+	amdgpu_sdma_process_ras_data_cb(adev, err_data, entry);
+
+out:
+	return AMDGPU_RAS_SUCCESS;
+}
+
 static int sdma_v4_0_process_illegal_inst_irq(struct amdgpu_device *adev,
 					      struct amdgpu_irq_src *source,
 					      struct amdgpu_iv_entry *entry)
 {
+	int instance;
+
 	DRM_ERROR("Illegal instruction in SDMA command stream\n");
-	schedule_work(&adev->reset_work);
+
+	instance = sdma_v4_0_irq_id_to_seq(entry->client_id);
+	if (instance < 0)
+		return 0;
+
+	switch (entry->ring_id) {
+	case 0:
+		drm_sched_fault(&adev->sdma.instance[instance].ring.sched);
+		break;
+	}
 	return 0;
 }
 
+static int sdma_v4_0_set_ecc_irq_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	u32 sdma_edc_config;
+
+	sdma_edc_config = RREG32_SDMA(type, mmSDMA0_EDC_CONFIG);
+	sdma_edc_config = REG_SET_FIELD(sdma_edc_config, SDMA0_EDC_CONFIG, ECC_INT_ENABLE,
+		       state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+	WREG32_SDMA(type, mmSDMA0_EDC_CONFIG, sdma_edc_config);
+
+	return 0;
+}
+
+static int sdma_v4_0_print_iv_entry(struct amdgpu_device *adev,
+					      struct amdgpu_iv_entry *entry)
+{
+	int instance;
+	struct amdgpu_task_info *task_info;
+	u64 addr;
+
+	instance = sdma_v4_0_irq_id_to_seq(entry->client_id);
+	if (instance < 0 || instance >= adev->sdma.num_instances) {
+		dev_err(adev->dev, "sdma instance invalid %d\n", instance);
+		return -EINVAL;
+	}
+
+	addr = (u64)entry->src_data[0] << 12;
+	addr |= ((u64)entry->src_data[1] & 0xf) << 44;
+
+	dev_dbg_ratelimited(adev->dev,
+			   "[sdma%d] address:0x%016llx src_id:%u ring:%u vmid:%u pasid:%u\n",
+			   instance, addr, entry->src_id, entry->ring_id, entry->vmid,
+			   entry->pasid);
+
+	task_info = amdgpu_vm_get_task_info_pasid(adev, entry->pasid);
+	if (task_info) {
+		dev_dbg_ratelimited(adev->dev,
+				    " for process %s pid %d thread %s pid %d\n",
+				    task_info->process_name, task_info->tgid,
+				    task_info->task.comm, task_info->task.pid);
+		amdgpu_vm_put_task_info(task_info);
+	}
+
+	return 0;
+}
+
+static int sdma_v4_0_process_vm_hole_irq(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      struct amdgpu_iv_entry *entry)
+{
+	dev_dbg_ratelimited(adev->dev, "MC or SEM address in VM hole\n");
+	sdma_v4_0_print_iv_entry(adev, entry);
+	return 0;
+}
+
+static int sdma_v4_0_process_doorbell_invalid_irq(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      struct amdgpu_iv_entry *entry)
+{
+	dev_dbg_ratelimited(adev->dev, "SDMA received a doorbell from BIF with byte_enable !=0xff\n");
+	sdma_v4_0_print_iv_entry(adev, entry);
+	return 0;
+}
+
+static int sdma_v4_0_process_pool_timeout_irq(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      struct amdgpu_iv_entry *entry)
+{
+	dev_dbg_ratelimited(adev->dev,
+		"Polling register/memory timeout executing POLL_REG/MEM with finite timer\n");
+	sdma_v4_0_print_iv_entry(adev, entry);
+	return 0;
+}
+
+static int sdma_v4_0_process_srbm_write_irq(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      struct amdgpu_iv_entry *entry)
+{
+	dev_dbg_ratelimited(adev->dev,
+		"SDMA gets an Register Write SRBM_WRITE command in non-privilege command buffer\n");
+	sdma_v4_0_print_iv_entry(adev, entry);
+	return 0;
+}
 
 static void sdma_v4_0_update_medium_grain_clock_gating(
 		struct amdgpu_device *adev,
 		bool enable)
 {
 	uint32_t data, def;
+	int i;
 
 	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_SDMA_MGCG)) {
-		/* enable sdma0 clock gating */
-		def = data = RREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_CLK_CTRL));
-		data &= ~(SDMA0_CLK_CTRL__SOFT_OVERRIDE7_MASK |
-			  SDMA0_CLK_CTRL__SOFT_OVERRIDE6_MASK |
-			  SDMA0_CLK_CTRL__SOFT_OVERRIDE5_MASK |
-			  SDMA0_CLK_CTRL__SOFT_OVERRIDE4_MASK |
-			  SDMA0_CLK_CTRL__SOFT_OVERRIDE3_MASK |
-			  SDMA0_CLK_CTRL__SOFT_OVERRIDE2_MASK |
-			  SDMA0_CLK_CTRL__SOFT_OVERRIDE1_MASK |
-			  SDMA0_CLK_CTRL__SOFT_OVERRIDE0_MASK);
-		if (def != data)
-			WREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_CLK_CTRL), data);
-
-		if (adev->asic_type == CHIP_VEGA10) {
-			def = data = RREG32(SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_CLK_CTRL));
-			data &= ~(SDMA1_CLK_CTRL__SOFT_OVERRIDE7_MASK |
-				  SDMA1_CLK_CTRL__SOFT_OVERRIDE6_MASK |
-				  SDMA1_CLK_CTRL__SOFT_OVERRIDE5_MASK |
-				  SDMA1_CLK_CTRL__SOFT_OVERRIDE4_MASK |
-				  SDMA1_CLK_CTRL__SOFT_OVERRIDE3_MASK |
-				  SDMA1_CLK_CTRL__SOFT_OVERRIDE2_MASK |
-				  SDMA1_CLK_CTRL__SOFT_OVERRIDE1_MASK |
-				  SDMA1_CLK_CTRL__SOFT_OVERRIDE0_MASK);
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			def = data = RREG32_SDMA(i, mmSDMA0_CLK_CTRL);
+			data &= ~(SDMA0_CLK_CTRL__SOFT_OVERRIDE7_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE6_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE5_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE4_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE3_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE2_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE1_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE0_MASK);
 			if (def != data)
-				WREG32(SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_CLK_CTRL), data);
+				WREG32_SDMA(i, mmSDMA0_CLK_CTRL, data);
 		}
 	} else {
-		/* disable sdma0 clock gating */
-		def = data = RREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_CLK_CTRL));
-		data |= (SDMA0_CLK_CTRL__SOFT_OVERRIDE7_MASK |
-			 SDMA0_CLK_CTRL__SOFT_OVERRIDE6_MASK |
-			 SDMA0_CLK_CTRL__SOFT_OVERRIDE5_MASK |
-			 SDMA0_CLK_CTRL__SOFT_OVERRIDE4_MASK |
-			 SDMA0_CLK_CTRL__SOFT_OVERRIDE3_MASK |
-			 SDMA0_CLK_CTRL__SOFT_OVERRIDE2_MASK |
-			 SDMA0_CLK_CTRL__SOFT_OVERRIDE1_MASK |
-			 SDMA0_CLK_CTRL__SOFT_OVERRIDE0_MASK);
-
-		if (def != data)
-			WREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_CLK_CTRL), data);
-
-		if (adev->asic_type == CHIP_VEGA10) {
-			def = data = RREG32(SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_CLK_CTRL));
-			data |= (SDMA1_CLK_CTRL__SOFT_OVERRIDE7_MASK |
-				 SDMA1_CLK_CTRL__SOFT_OVERRIDE6_MASK |
-				 SDMA1_CLK_CTRL__SOFT_OVERRIDE5_MASK |
-				 SDMA1_CLK_CTRL__SOFT_OVERRIDE4_MASK |
-				 SDMA1_CLK_CTRL__SOFT_OVERRIDE3_MASK |
-				 SDMA1_CLK_CTRL__SOFT_OVERRIDE2_MASK |
-				 SDMA1_CLK_CTRL__SOFT_OVERRIDE1_MASK |
-				 SDMA1_CLK_CTRL__SOFT_OVERRIDE0_MASK);
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			def = data = RREG32_SDMA(i, mmSDMA0_CLK_CTRL);
+			data |= (SDMA0_CLK_CTRL__SOFT_OVERRIDE7_MASK |
+				 SDMA0_CLK_CTRL__SOFT_OVERRIDE6_MASK |
+				 SDMA0_CLK_CTRL__SOFT_OVERRIDE5_MASK |
+				 SDMA0_CLK_CTRL__SOFT_OVERRIDE4_MASK |
+				 SDMA0_CLK_CTRL__SOFT_OVERRIDE3_MASK |
+				 SDMA0_CLK_CTRL__SOFT_OVERRIDE2_MASK |
+				 SDMA0_CLK_CTRL__SOFT_OVERRIDE1_MASK |
+				 SDMA0_CLK_CTRL__SOFT_OVERRIDE0_MASK);
 			if (def != data)
-				WREG32(SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_CLK_CTRL), data);
+				WREG32_SDMA(i, mmSDMA0_CLK_CTRL, data);
 		}
 	}
 }
@@ -1486,69 +2276,53 @@ static void sdma_v4_0_update_medium_grain_light_sleep(
 		bool enable)
 {
 	uint32_t data, def;
+	int i;
 
 	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_SDMA_LS)) {
-		/* 1-not override: enable sdma0 mem light sleep */
-		def = data = RREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_POWER_CNTL));
-		data |= SDMA0_POWER_CNTL__MEM_POWER_OVERRIDE_MASK;
-		if (def != data)
-			WREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_POWER_CNTL), data);
-
-		/* 1-not override: enable sdma1 mem light sleep */
-		if (adev->asic_type == CHIP_VEGA10) {
-			def = data = RREG32(SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_POWER_CNTL));
-			data |= SDMA1_POWER_CNTL__MEM_POWER_OVERRIDE_MASK;
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			/* 1-not override: enable sdma mem light sleep */
+			def = data = RREG32_SDMA(0, mmSDMA0_POWER_CNTL);
+			data |= SDMA0_POWER_CNTL__MEM_POWER_OVERRIDE_MASK;
 			if (def != data)
-				WREG32(SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_POWER_CNTL), data);
+				WREG32_SDMA(0, mmSDMA0_POWER_CNTL, data);
 		}
 	} else {
-		/* 0-override:disable sdma0 mem light sleep */
-		def = data = RREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_POWER_CNTL));
-		data &= ~SDMA0_POWER_CNTL__MEM_POWER_OVERRIDE_MASK;
-		if (def != data)
-			WREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_POWER_CNTL), data);
-
-		/* 0-override:disable sdma1 mem light sleep */
-		if (adev->asic_type == CHIP_VEGA10) {
-			def = data = RREG32(SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_POWER_CNTL));
-			data &= ~SDMA1_POWER_CNTL__MEM_POWER_OVERRIDE_MASK;
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+		/* 0-override:disable sdma mem light sleep */
+			def = data = RREG32_SDMA(0, mmSDMA0_POWER_CNTL);
+			data &= ~SDMA0_POWER_CNTL__MEM_POWER_OVERRIDE_MASK;
 			if (def != data)
-				WREG32(SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_POWER_CNTL), data);
+				WREG32_SDMA(0, mmSDMA0_POWER_CNTL, data);
 		}
 	}
 }
 
-static int sdma_v4_0_set_clockgating_state(void *handle,
+static int sdma_v4_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (amdgpu_sriov_vf(adev))
 		return 0;
 
-	switch (adev->asic_type) {
-	case CHIP_VEGA10:
-	case CHIP_RAVEN:
-		sdma_v4_0_update_medium_grain_clock_gating(adev,
-				state == AMD_CG_STATE_GATE ? true : false);
-		sdma_v4_0_update_medium_grain_light_sleep(adev,
-				state == AMD_CG_STATE_GATE ? true : false);
-		break;
-	default:
-		break;
-	}
+	sdma_v4_0_update_medium_grain_clock_gating(adev,
+			state == AMD_CG_STATE_GATE);
+	sdma_v4_0_update_medium_grain_light_sleep(adev,
+			state == AMD_CG_STATE_GATE);
 	return 0;
 }
 
-static int sdma_v4_0_set_powergating_state(void *handle,
+static int sdma_v4_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	switch (adev->asic_type) {
-	case CHIP_RAVEN:
+	switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) {
+	case IP_VERSION(4, 1, 0):
+	case IP_VERSION(4, 1, 1):
+	case IP_VERSION(4, 1, 2):
 		sdma_v4_1_update_power_gating(adev,
-				state == AMD_PG_STATE_GATE ? true : false);
+				state == AMD_PG_STATE_GATE);
 		break;
 	default:
 		break;
@@ -1557,9 +2331,9 @@ static int sdma_v4_0_set_powergating_state(void *handle,
 	return 0;
 }
 
-static void sdma_v4_0_get_clockgating_state(void *handle, u32 *flags)
+static void sdma_v4_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int data;
 
 	if (amdgpu_sriov_vf(adev))
@@ -1576,10 +2350,50 @@ static void sdma_v4_0_get_clockgating_state(void *handle, u32 *flags)
 		*flags |= AMD_CG_SUPPORT_SDMA_LS;
 }
 
+static void sdma_v4_0_print_ip_state(struct amdgpu_ip_block *ip_block, struct drm_printer *p)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, j;
+	uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_4_0);
+	uint32_t instance_offset;
+
+	if (!adev->sdma.ip_dump)
+		return;
+
+	drm_printf(p, "num_instances:%d\n", adev->sdma.num_instances);
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		instance_offset = i * reg_count;
+		drm_printf(p, "\nInstance:%d\n", i);
+
+		for (j = 0; j < reg_count; j++)
+			drm_printf(p, "%-50s \t 0x%08x\n", sdma_reg_list_4_0[j].reg_name,
+				   adev->sdma.ip_dump[instance_offset + j]);
+	}
+}
+
+static void sdma_v4_0_dump_ip_state(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, j;
+	uint32_t instance_offset;
+	uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_4_0);
+
+	if (!adev->sdma.ip_dump)
+		return;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		instance_offset = i * reg_count;
+		for (j = 0; j < reg_count; j++)
+			adev->sdma.ip_dump[instance_offset + j] =
+				RREG32(sdma_v4_0_get_reg_offset(adev, i,
+				       sdma_reg_list_4_0[j].reg_offset));
+	}
+}
+
 const struct amd_ip_funcs sdma_v4_0_ip_funcs = {
 	.name = "sdma_v4_0",
 	.early_init = sdma_v4_0_early_init,
-	.late_init = NULL,
+	.late_init = sdma_v4_0_late_init,
 	.sw_init = sdma_v4_0_sw_init,
 	.sw_fini = sdma_v4_0_sw_fini,
 	.hw_init = sdma_v4_0_hw_init,
@@ -1592,22 +2406,26 @@ const struct amd_ip_funcs sdma_v4_0_ip_funcs = {
 	.set_clockgating_state = sdma_v4_0_set_clockgating_state,
 	.set_powergating_state = sdma_v4_0_set_powergating_state,
 	.get_clockgating_state = sdma_v4_0_get_clockgating_state,
+	.dump_ip_state = sdma_v4_0_dump_ip_state,
+	.print_ip_state = sdma_v4_0_print_ip_state,
 };
 
 static const struct amdgpu_ring_funcs sdma_v4_0_ring_funcs = {
 	.type = AMDGPU_RING_TYPE_SDMA,
-	.align_mask = 0xf,
+	.align_mask = 0xff,
 	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
 	.support_64bit_ptrs = true,
-	.vmhub = AMDGPU_MMHUB,
+	.secure_submission_supported = true,
 	.get_rptr = sdma_v4_0_ring_get_rptr,
 	.get_wptr = sdma_v4_0_ring_get_wptr,
 	.set_wptr = sdma_v4_0_ring_set_wptr,
 	.emit_frame_size =
 		6 + /* sdma_v4_0_ring_emit_hdp_flush */
-		3 + /* sdma_v4_0_ring_emit_hdp_invalidate */
+		3 + /* hdp invalidate */
 		6 + /* sdma_v4_0_ring_emit_pipeline_sync */
-		18 + /* sdma_v4_0_ring_emit_vm_flush */
+		/* sdma_v4_0_ring_emit_vm_flush */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 6 +
 		10 + 10 + 10, /* sdma_v4_0_ring_emit_fence x3 for user fence, vm fence */
 	.emit_ib_size = 7 + 6, /* sdma_v4_0_ring_emit_ib */
 	.emit_ib = sdma_v4_0_ring_emit_ib,
@@ -1615,19 +2433,60 @@ static const struct amdgpu_ring_funcs sdma_v4_0_ring_funcs = {
 	.emit_pipeline_sync = sdma_v4_0_ring_emit_pipeline_sync,
 	.emit_vm_flush = sdma_v4_0_ring_emit_vm_flush,
 	.emit_hdp_flush = sdma_v4_0_ring_emit_hdp_flush,
-	.emit_hdp_invalidate = sdma_v4_0_ring_emit_hdp_invalidate,
 	.test_ring = sdma_v4_0_ring_test_ring,
 	.test_ib = sdma_v4_0_ring_test_ib,
 	.insert_nop = sdma_v4_0_ring_insert_nop,
 	.pad_ib = sdma_v4_0_ring_pad_ib,
+	.emit_wreg = sdma_v4_0_ring_emit_wreg,
+	.emit_reg_wait = sdma_v4_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+};
+
+static const struct amdgpu_ring_funcs sdma_v4_0_page_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_SDMA,
+	.align_mask = 0xff,
+	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
+	.support_64bit_ptrs = true,
+	.secure_submission_supported = true,
+	.get_rptr = sdma_v4_0_ring_get_rptr,
+	.get_wptr = sdma_v4_0_page_ring_get_wptr,
+	.set_wptr = sdma_v4_0_page_ring_set_wptr,
+	.emit_frame_size =
+		6 + /* sdma_v4_0_ring_emit_hdp_flush */
+		3 + /* hdp invalidate */
+		6 + /* sdma_v4_0_ring_emit_pipeline_sync */
+		/* sdma_v4_0_ring_emit_vm_flush */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 6 +
+		10 + 10 + 10, /* sdma_v4_0_ring_emit_fence x3 for user fence, vm fence */
+	.emit_ib_size = 7 + 6, /* sdma_v4_0_ring_emit_ib */
+	.emit_ib = sdma_v4_0_ring_emit_ib,
+	.emit_fence = sdma_v4_0_ring_emit_fence,
+	.emit_pipeline_sync = sdma_v4_0_ring_emit_pipeline_sync,
+	.emit_vm_flush = sdma_v4_0_ring_emit_vm_flush,
+	.emit_hdp_flush = sdma_v4_0_ring_emit_hdp_flush,
+	.test_ring = sdma_v4_0_ring_test_ring,
+	.test_ib = sdma_v4_0_ring_test_ib,
+	.insert_nop = sdma_v4_0_ring_insert_nop,
+	.pad_ib = sdma_v4_0_ring_pad_ib,
+	.emit_wreg = sdma_v4_0_ring_emit_wreg,
+	.emit_reg_wait = sdma_v4_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
 };
 
 static void sdma_v4_0_set_ring_funcs(struct amdgpu_device *adev)
 {
 	int i;
 
-	for (i = 0; i < adev->sdma.num_instances; i++)
+	for (i = 0; i < adev->sdma.num_instances; i++) {
 		adev->sdma.instance[i].ring.funcs = &sdma_v4_0_ring_funcs;
+		adev->sdma.instance[i].ring.me = i;
+		if (adev->sdma.has_page_queue) {
+			adev->sdma.instance[i].page.funcs =
+					&sdma_v4_0_page_ring_funcs;
+			adev->sdma.instance[i].page.me = i;
+		}
+	}
 }
 
 static const struct amdgpu_irq_src_funcs sdma_v4_0_trap_irq_funcs = {
@@ -1639,32 +2498,74 @@ static const struct amdgpu_irq_src_funcs sdma_v4_0_illegal_inst_irq_funcs = {
 	.process = sdma_v4_0_process_illegal_inst_irq,
 };
 
+static const struct amdgpu_irq_src_funcs sdma_v4_0_ecc_irq_funcs = {
+	.set = sdma_v4_0_set_ecc_irq_state,
+	.process = amdgpu_sdma_process_ecc_irq,
+};
+
+static const struct amdgpu_irq_src_funcs sdma_v4_0_vm_hole_irq_funcs = {
+	.process = sdma_v4_0_process_vm_hole_irq,
+};
+
+static const struct amdgpu_irq_src_funcs sdma_v4_0_doorbell_invalid_irq_funcs = {
+	.process = sdma_v4_0_process_doorbell_invalid_irq,
+};
+
+static const struct amdgpu_irq_src_funcs sdma_v4_0_pool_timeout_irq_funcs = {
+	.process = sdma_v4_0_process_pool_timeout_irq,
+};
+
+static const struct amdgpu_irq_src_funcs sdma_v4_0_srbm_write_irq_funcs = {
+	.process = sdma_v4_0_process_srbm_write_irq,
+};
+
 static void sdma_v4_0_set_irq_funcs(struct amdgpu_device *adev)
 {
-	adev->sdma.trap_irq.num_types = AMDGPU_SDMA_IRQ_LAST;
+	adev->sdma.trap_irq.num_types = adev->sdma.num_instances;
+	adev->sdma.ecc_irq.num_types = adev->sdma.num_instances;
+	/*For Arcturus and Aldebaran, add another 4 irq handler*/
+	switch (adev->sdma.num_instances) {
+	case 5:
+	case 8:
+		adev->sdma.vm_hole_irq.num_types = adev->sdma.num_instances;
+		adev->sdma.doorbell_invalid_irq.num_types = adev->sdma.num_instances;
+		adev->sdma.pool_timeout_irq.num_types = adev->sdma.num_instances;
+		adev->sdma.srbm_write_irq.num_types = adev->sdma.num_instances;
+		break;
+	default:
+		break;
+	}
 	adev->sdma.trap_irq.funcs = &sdma_v4_0_trap_irq_funcs;
 	adev->sdma.illegal_inst_irq.funcs = &sdma_v4_0_illegal_inst_irq_funcs;
+	adev->sdma.ecc_irq.funcs = &sdma_v4_0_ecc_irq_funcs;
+	adev->sdma.vm_hole_irq.funcs = &sdma_v4_0_vm_hole_irq_funcs;
+	adev->sdma.doorbell_invalid_irq.funcs = &sdma_v4_0_doorbell_invalid_irq_funcs;
+	adev->sdma.pool_timeout_irq.funcs = &sdma_v4_0_pool_timeout_irq_funcs;
+	adev->sdma.srbm_write_irq.funcs = &sdma_v4_0_srbm_write_irq_funcs;
 }
 
 /**
  * sdma_v4_0_emit_copy_buffer - copy buffer using the sDMA engine
  *
- * @ring: amdgpu_ring structure holding ring information
+ * @ib: indirect buffer to copy to
  * @src_offset: src GPU address
  * @dst_offset: dst GPU address
  * @byte_count: number of bytes to xfer
+ * @copy_flags: copy flags for the buffers
  *
- * Copy GPU buffers using the DMA engine (VEGA10).
+ * Copy GPU buffers using the DMA engine (VEGA10/12).
  * Used by the amdgpu ttm implementation to move pages if
  * registered as the asic copy callback.
  */
 static void sdma_v4_0_emit_copy_buffer(struct amdgpu_ib *ib,
 				       uint64_t src_offset,
 				       uint64_t dst_offset,
-				       uint32_t byte_count)
+				       uint32_t byte_count,
+				       uint32_t copy_flags)
 {
 	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
-		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR) |
+		SDMA_PKT_COPY_LINEAR_HEADER_TMZ((copy_flags & AMDGPU_COPY_FLAGS_TMZ) ? 1 : 0);
 	ib->ptr[ib->length_dw++] = byte_count - 1;
 	ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
 	ib->ptr[ib->length_dw++] = lower_32_bits(src_offset);
@@ -1676,12 +2577,12 @@ static void sdma_v4_0_emit_copy_buffer(struct amdgpu_ib *ib,
 /**
  * sdma_v4_0_emit_fill_buffer - fill buffer using the sDMA engine
  *
- * @ring: amdgpu_ring structure holding ring information
+ * @ib: indirect buffer to copy to
  * @src_data: value to write to buffer
  * @dst_offset: dst GPU address
  * @byte_count: number of bytes to xfer
  *
- * Fill GPU buffers using the DMA engine (VEGA10).
+ * Fill GPU buffers using the DMA engine (VEGA10/12).
  */
 static void sdma_v4_0_emit_fill_buffer(struct amdgpu_ib *ib,
 				       uint32_t src_data,
@@ -1707,29 +2608,130 @@ static const struct amdgpu_buffer_funcs sdma_v4_0_buffer_funcs = {
 
 static void sdma_v4_0_set_buffer_funcs(struct amdgpu_device *adev)
 {
-	if (adev->mman.buffer_funcs == NULL) {
-		adev->mman.buffer_funcs = &sdma_v4_0_buffer_funcs;
+	adev->mman.buffer_funcs = &sdma_v4_0_buffer_funcs;
+	if (adev->sdma.has_page_queue)
+		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].page;
+	else
 		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
-	}
 }
 
 static const struct amdgpu_vm_pte_funcs sdma_v4_0_vm_pte_funcs = {
+	.copy_pte_num_dw = 7,
 	.copy_pte = sdma_v4_0_vm_copy_pte,
+
 	.write_pte = sdma_v4_0_vm_write_pte,
 	.set_pte_pde = sdma_v4_0_vm_set_pte_pde,
 };
 
 static void sdma_v4_0_set_vm_pte_funcs(struct amdgpu_device *adev)
 {
+	struct drm_gpu_scheduler *sched;
 	unsigned i;
 
-	if (adev->vm_manager.vm_pte_funcs == NULL) {
-		adev->vm_manager.vm_pte_funcs = &sdma_v4_0_vm_pte_funcs;
+	adev->vm_manager.vm_pte_funcs = &sdma_v4_0_vm_pte_funcs;
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (adev->sdma.has_page_queue)
+			sched = &adev->sdma.instance[i].page.sched;
+		else
+			sched = &adev->sdma.instance[i].ring.sched;
+		adev->vm_manager.vm_pte_scheds[i] = sched;
+	}
+	adev->vm_manager.vm_pte_num_scheds = adev->sdma.num_instances;
+}
+
+static void sdma_v4_0_get_ras_error_count(uint32_t value,
+					uint32_t instance,
+					uint32_t *sec_count)
+{
+	uint32_t i;
+	uint32_t sec_cnt;
+
+	/* double bits error (multiple bits) error detection is not supported */
+	for (i = 0; i < ARRAY_SIZE(sdma_v4_0_ras_fields); i++) {
+		/* the SDMA_EDC_COUNTER register in each sdma instance
+		 * shares the same sed shift_mask
+		 * */
+		sec_cnt = (value &
+			sdma_v4_0_ras_fields[i].sec_count_mask) >>
+			sdma_v4_0_ras_fields[i].sec_count_shift;
+		if (sec_cnt) {
+			DRM_INFO("Detected %s in SDMA%d, SED %d\n",
+				sdma_v4_0_ras_fields[i].name,
+				instance, sec_cnt);
+			*sec_count += sec_cnt;
+		}
+	}
+}
+
+static int sdma_v4_0_query_ras_error_count_by_instance(struct amdgpu_device *adev,
+			uint32_t instance, void *ras_error_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+	uint32_t sec_count = 0;
+	uint32_t reg_value = 0;
+
+	reg_value = RREG32_SDMA(instance, mmSDMA0_EDC_COUNTER);
+	/* double bit error is not supported */
+	if (reg_value)
+		sdma_v4_0_get_ras_error_count(reg_value,
+				instance, &sec_count);
+	/* err_data->ce_count should be initialized to 0
+	 * before calling into this function */
+	err_data->ce_count += sec_count;
+	/* double bit error is not supported
+	 * set ue count to 0 */
+	err_data->ue_count = 0;
+
+	return 0;
+};
+
+static void sdma_v4_0_query_ras_error_count(struct amdgpu_device *adev,  void *ras_error_status)
+{
+	int i = 0;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (sdma_v4_0_query_ras_error_count_by_instance(adev, i, ras_error_status)) {
+			dev_err(adev->dev, "Query ras error count failed in SDMA%d\n", i);
+			return;
+		}
+	}
+}
+
+static void sdma_v4_0_reset_ras_error_count(struct amdgpu_device *adev)
+{
+	int i;
+
+	/* read back edc counter registers to clear the counters */
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__SDMA)) {
 		for (i = 0; i < adev->sdma.num_instances; i++)
-			adev->vm_manager.vm_pte_rings[i] =
-				&adev->sdma.instance[i].ring;
+			RREG32_SDMA(i, mmSDMA0_EDC_COUNTER);
+	}
+}
+
+const struct amdgpu_ras_block_hw_ops sdma_v4_0_ras_hw_ops = {
+	.query_ras_error_count = sdma_v4_0_query_ras_error_count,
+	.reset_ras_error_count = sdma_v4_0_reset_ras_error_count,
+};
+
+static struct amdgpu_sdma_ras sdma_v4_0_ras = {
+	.ras_block = {
+		.hw_ops = &sdma_v4_0_ras_hw_ops,
+		.ras_cb = sdma_v4_0_process_ras_data_cb,
+	},
+};
 
-		adev->vm_manager.vm_pte_num_rings = adev->sdma.num_instances;
+static void sdma_v4_0_set_ras_funcs(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) {
+	case IP_VERSION(4, 2, 0):
+	case IP_VERSION(4, 2, 2):
+		adev->sdma.ras = &sdma_v4_0_ras;
+		break;
+	case IP_VERSION(4, 4, 0):
+		adev->sdma.ras = &sdma_v4_4_ras;
+		break;
+	default:
+		break;
 	}
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4.c
new file mode 100644
index 000000000000..0ddb6955a6d3
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4.c
@@ -0,0 +1,274 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "sdma/sdma_4_4_0_offset.h"
+#include "sdma/sdma_4_4_0_sh_mask.h"
+#include "soc15.h"
+#include "amdgpu_ras.h"
+
+#define SDMA1_REG_OFFSET 0x600
+#define SDMA2_REG_OFFSET 0x1cda0
+#define SDMA3_REG_OFFSET 0x1d1a0
+#define SDMA4_REG_OFFSET 0x1d5a0
+
+/* helper function that allow only use sdma0 register offset
+ * to calculate register offset for all the sdma instances */
+static uint32_t sdma_v4_4_get_reg_offset(struct amdgpu_device *adev,
+					 uint32_t instance,
+					 uint32_t offset)
+{
+	uint32_t sdma_base = adev->reg_offset[SDMA0_HWIP][0][0];
+
+	switch (instance) {
+	case 0:
+		return (sdma_base + offset);
+	case 1:
+		return (sdma_base + SDMA1_REG_OFFSET + offset);
+	case 2:
+		return (sdma_base + SDMA2_REG_OFFSET + offset);
+	case 3:
+		return (sdma_base + SDMA3_REG_OFFSET + offset);
+	case 4:
+		return (sdma_base + SDMA4_REG_OFFSET + offset);
+	default:
+		break;
+	}
+	return 0;
+}
+
+static const struct soc15_ras_field_entry sdma_v4_4_ras_fields[] = {
+	{ "SDMA_MBANK_DATA_BUF0_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF0_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF1_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF1_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF2_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF2_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF3_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF3_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF4_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF4_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF5_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF5_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF6_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF6_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF7_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF7_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF8_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF8_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF9_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF9_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF10_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF10_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF11_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF11_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF12_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF12_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF13_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF13_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF14_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF14_SED),
+	0, 0,
+	},
+	{ "SDMA_MBANK_DATA_BUF15_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF15_SED),
+	0, 0,
+	},
+	{ "SDMA_UCODE_BUF_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER2),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER2, SDMA_UCODE_BUF_SED),
+	0, 0,
+	},
+	{ "SDMA_RB_CMD_BUF_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER2),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER2, SDMA_RB_CMD_BUF_SED),
+	0, 0,
+	},
+	{ "SDMA_IB_CMD_BUF_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER2),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER2, SDMA_IB_CMD_BUF_SED),
+	0, 0,
+	},
+	{ "SDMA_UTCL1_RD_FIFO_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER2),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER2, SDMA_UTCL1_RD_FIFO_SED),
+	0, 0,
+	},
+	{ "SDMA_UTCL1_RDBST_FIFO_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER2),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER2, SDMA_UTCL1_RDBST_FIFO_SED),
+	0, 0,
+	},
+	{ "SDMA_UTCL1_WR_FIFO_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER2),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER2, SDMA_UTCL1_WR_FIFO_SED),
+	0, 0,
+	},
+	{ "SDMA_DATA_LUT_FIFO_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER2),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER2, SDMA_DATA_LUT_FIFO_SED),
+	0, 0,
+	},
+	{ "SDMA_SPLIT_DATA_BUF_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER2),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER2, SDMA_SPLIT_DATA_BUF_SED),
+	0, 0,
+	},
+	{ "SDMA_MC_WR_ADDR_FIFO_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER2),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER2, SDMA_MC_WR_ADDR_FIFO_SED),
+	0, 0,
+	},
+	{ "SDMA_MC_RDRET_BUF_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER2),
+	SOC15_REG_FIELD(SDMA0_EDC_COUNTER2, SDMA_MC_WR_ADDR_FIFO_SED),
+	0, 0,
+	},
+};
+
+static void sdma_v4_4_get_ras_error_count(struct amdgpu_device *adev,
+					  uint32_t reg_offset,
+					  uint32_t value,
+					  uint32_t instance,
+					  uint32_t *sec_count)
+{
+	uint32_t i;
+	uint32_t sec_cnt;
+
+	/* double bits error (multiple bits) error detection is not supported */
+	for (i = 0; i < ARRAY_SIZE(sdma_v4_4_ras_fields); i++) {
+		if (sdma_v4_4_ras_fields[i].reg_offset != reg_offset)
+			continue;
+
+		/* the SDMA_EDC_COUNTER register in each sdma instance
+		 * shares the same sed shift_mask
+		 * */
+		sec_cnt = (value &
+			sdma_v4_4_ras_fields[i].sec_count_mask) >>
+			sdma_v4_4_ras_fields[i].sec_count_shift;
+		if (sec_cnt) {
+			dev_info(adev->dev, "Detected %s in SDMA%d, SED %d\n",
+				 sdma_v4_4_ras_fields[i].name,
+				 instance, sec_cnt);
+			*sec_count += sec_cnt;
+		}
+	}
+}
+
+static int sdma_v4_4_query_ras_error_count_by_instance(struct amdgpu_device *adev,
+					   uint32_t instance,
+					   void *ras_error_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+	uint32_t sec_count = 0;
+	uint32_t reg_value = 0;
+	uint32_t reg_offset = 0;
+
+	reg_offset = sdma_v4_4_get_reg_offset(adev, instance, regSDMA0_EDC_COUNTER);
+	reg_value = RREG32(reg_offset);
+	/* double bit error is not supported */
+	if (reg_value)
+		sdma_v4_4_get_ras_error_count(adev, regSDMA0_EDC_COUNTER, reg_value,
+					      instance, &sec_count);
+
+	reg_offset = sdma_v4_4_get_reg_offset(adev, instance, regSDMA0_EDC_COUNTER2);
+	reg_value = RREG32(reg_offset);
+	/* double bit error is not supported */
+	if (reg_value)
+		sdma_v4_4_get_ras_error_count(adev, regSDMA0_EDC_COUNTER2, reg_value,
+					      instance, &sec_count);
+
+	/*
+	 * err_data->ue_count should be initialized to 0
+	 * before calling into this function
+	 *
+	 * SDMA RAS supports single bit uncorrectable error detection.
+	 * So, increment uncorrectable error count.
+	 */
+	err_data->ue_count += sec_count;
+
+	/*
+	 * SDMA RAS does not support correctable errors.
+	 * Set ce count to 0.
+	 */
+	err_data->ce_count = 0;
+
+	return 0;
+};
+
+static void sdma_v4_4_reset_ras_error_count(struct amdgpu_device *adev)
+{
+	int i;
+	uint32_t reg_offset;
+
+	/* write 0 to EDC_COUNTER reg to clear sdma edc counters */
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__SDMA)) {
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			reg_offset = sdma_v4_4_get_reg_offset(adev, i, regSDMA0_EDC_COUNTER);
+			WREG32(reg_offset, 0);
+			reg_offset = sdma_v4_4_get_reg_offset(adev, i, regSDMA0_EDC_COUNTER2);
+			WREG32(reg_offset, 0);
+		}
+	}
+}
+
+static void sdma_v4_4_query_ras_error_count(struct amdgpu_device *adev,  void *ras_error_status)
+{
+	int i = 0;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (sdma_v4_4_query_ras_error_count_by_instance(adev, i, ras_error_status)) {
+			dev_err(adev->dev, "Query ras error count failed in SDMA%d\n", i);
+			return;
+		}
+	}
+
+}
+
+const struct amdgpu_ras_block_hw_ops sdma_v4_4_ras_hw_ops = {
+	.query_ras_error_count = sdma_v4_4_query_ras_error_count,
+	.reset_ras_error_count = sdma_v4_4_reset_ras_error_count,
+};
+
+struct amdgpu_sdma_ras sdma_v4_4_ras = {
+	.ras_block = {
+		.hw_ops = &sdma_v4_4_ras_hw_ops,
+	},
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4.h b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4.h
new file mode 100644
index 000000000000..a9f0c68359e0
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4.h
@@ -0,0 +1,28 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __SDMA_V4_4_H__
+#define __SDMA_V4_4_H__
+
+extern struct amdgpu_sdma_ras sdma_v4_4_ras;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c
new file mode 100644
index 000000000000..36b1ca73c2ed
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c
@@ -0,0 +1,2609 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/delay.h>
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+
+#include "amdgpu.h"
+#include "amdgpu_xcp.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_trace.h"
+#include "amdgpu_reset.h"
+
+#include "sdma/sdma_4_4_2_offset.h"
+#include "sdma/sdma_4_4_2_sh_mask.h"
+
+#include "soc15_common.h"
+#include "soc15.h"
+#include "vega10_sdma_pkt_open.h"
+
+#include "ivsrcid/sdma0/irqsrcs_sdma0_4_0.h"
+#include "ivsrcid/sdma1/irqsrcs_sdma1_4_0.h"
+
+#include "amdgpu_ras.h"
+
+MODULE_FIRMWARE("amdgpu/sdma_4_4_2.bin");
+MODULE_FIRMWARE("amdgpu/sdma_4_4_4.bin");
+MODULE_FIRMWARE("amdgpu/sdma_4_4_5.bin");
+
+static const struct amdgpu_hwip_reg_entry sdma_reg_list_4_4_2[] = {
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_STATUS_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_STATUS1_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_STATUS2_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_STATUS3_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_UCODE_CHECKSUM),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_RB_RPTR_FETCH_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_RB_RPTR_FETCH),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_UTCL1_RD_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_UTCL1_WR_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_UTCL1_RD_XNACK0),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_UTCL1_RD_XNACK1),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_UTCL1_WR_XNACK0),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_UTCL1_WR_XNACK1),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_GFX_RB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_GFX_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_GFX_RB_RPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_GFX_RB_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_GFX_RB_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_GFX_IB_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_GFX_IB_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_GFX_IB_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_GFX_IB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_GFX_IB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_GFX_IB_SUB_REMAIN),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_GFX_DUMMY_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_PAGE_RB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_PAGE_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_PAGE_RB_RPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_PAGE_RB_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_PAGE_RB_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_PAGE_IB_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_PAGE_IB_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_PAGE_IB_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_PAGE_DUMMY_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_RLC0_RB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_RLC0_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_RLC0_RB_RPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_RLC0_RB_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_RLC0_RB_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_RLC0_IB_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_RLC0_IB_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_RLC0_IB_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_RLC0_DUMMY_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA_VM_CNTL)
+};
+
+#define mmSMNAID_AID0_MCA_SMU 0x03b30400
+
+#define WREG32_SDMA(instance, offset, value) \
+	WREG32(sdma_v4_4_2_get_reg_offset(adev, (instance), (offset)), value)
+#define RREG32_SDMA(instance, offset) \
+	RREG32(sdma_v4_4_2_get_reg_offset(adev, (instance), (offset)))
+
+static void sdma_v4_4_2_set_ring_funcs(struct amdgpu_device *adev);
+static void sdma_v4_4_2_set_buffer_funcs(struct amdgpu_device *adev);
+static void sdma_v4_4_2_set_vm_pte_funcs(struct amdgpu_device *adev);
+static void sdma_v4_4_2_set_irq_funcs(struct amdgpu_device *adev);
+static void sdma_v4_4_2_set_ras_funcs(struct amdgpu_device *adev);
+static void sdma_v4_4_2_update_reset_mask(struct amdgpu_device *adev);
+static int sdma_v4_4_2_stop_queue(struct amdgpu_ring *ring);
+static int sdma_v4_4_2_restore_queue(struct amdgpu_ring *ring);
+static int sdma_v4_4_2_soft_reset_engine(struct amdgpu_device *adev,
+					 u32 instance_id);
+
+static u32 sdma_v4_4_2_get_reg_offset(struct amdgpu_device *adev,
+		u32 instance, u32 offset)
+{
+	u32 dev_inst = GET_INST(SDMA0, instance);
+
+	return (adev->reg_offset[SDMA0_HWIP][dev_inst][0] + offset);
+}
+
+static unsigned sdma_v4_4_2_seq_to_irq_id(int seq_num)
+{
+	switch (seq_num) {
+	case 0:
+		return SOC15_IH_CLIENTID_SDMA0;
+	case 1:
+		return SOC15_IH_CLIENTID_SDMA1;
+	case 2:
+		return SOC15_IH_CLIENTID_SDMA2;
+	case 3:
+		return SOC15_IH_CLIENTID_SDMA3;
+	default:
+		return -EINVAL;
+	}
+}
+
+static int sdma_v4_4_2_irq_id_to_seq(struct amdgpu_device *adev, unsigned client_id)
+{
+	switch (client_id) {
+	case SOC15_IH_CLIENTID_SDMA0:
+		return 0;
+	case SOC15_IH_CLIENTID_SDMA1:
+		return 1;
+	case SOC15_IH_CLIENTID_SDMA2:
+		if (amdgpu_sriov_vf(adev) && (adev->gfx.xcc_mask == 0x1))
+			return 0;
+		else
+			return 2;
+	case SOC15_IH_CLIENTID_SDMA3:
+		if (amdgpu_sriov_vf(adev) && (adev->gfx.xcc_mask == 0x1))
+			return 1;
+		else
+			return 3;
+	default:
+		return -EINVAL;
+	}
+}
+
+static void sdma_v4_4_2_inst_init_golden_registers(struct amdgpu_device *adev,
+						   uint32_t inst_mask)
+{
+	u32 val;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		val = RREG32_SDMA(i, regSDMA_GB_ADDR_CONFIG);
+		val = REG_SET_FIELD(val, SDMA_GB_ADDR_CONFIG, NUM_BANKS, 4);
+		val = REG_SET_FIELD(val, SDMA_GB_ADDR_CONFIG,
+				    PIPE_INTERLEAVE_SIZE, 0);
+		WREG32_SDMA(i, regSDMA_GB_ADDR_CONFIG, val);
+
+		val = RREG32_SDMA(i, regSDMA_GB_ADDR_CONFIG_READ);
+		val = REG_SET_FIELD(val, SDMA_GB_ADDR_CONFIG_READ, NUM_BANKS,
+				    4);
+		val = REG_SET_FIELD(val, SDMA_GB_ADDR_CONFIG_READ,
+				    PIPE_INTERLEAVE_SIZE, 0);
+		WREG32_SDMA(i, regSDMA_GB_ADDR_CONFIG_READ, val);
+	}
+}
+
+/**
+ * sdma_v4_4_2_init_microcode - load ucode images from disk
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Use the firmware interface to load the ucode images into
+ * the driver (not loaded into hw).
+ * Returns 0 on success, error on failure.
+ */
+static int sdma_v4_4_2_init_microcode(struct amdgpu_device *adev)
+{
+	int ret, i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) == IP_VERSION(4, 4, 2) ||
+		    amdgpu_ip_version(adev, SDMA0_HWIP, 0) == IP_VERSION(4, 4, 4) ||
+		    amdgpu_ip_version(adev, SDMA0_HWIP, 0) == IP_VERSION(4, 4, 5)) {
+			ret = amdgpu_sdma_init_microcode(adev, 0, true);
+			break;
+		} else {
+			ret = amdgpu_sdma_init_microcode(adev, i, false);
+			if (ret)
+				return ret;
+		}
+	}
+
+	return ret;
+}
+
+/**
+ * sdma_v4_4_2_ring_get_rptr - get the current read pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current rptr from the hardware.
+ */
+static uint64_t sdma_v4_4_2_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	u64 rptr;
+
+	/* XXX check if swapping is necessary on BE */
+	rptr = READ_ONCE(*((u64 *)&ring->adev->wb.wb[ring->rptr_offs]));
+
+	DRM_DEBUG("rptr before shift == 0x%016llx\n", rptr);
+	return rptr >> 2;
+}
+
+/**
+ * sdma_v4_4_2_ring_get_wptr - get the current write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current wptr from the hardware.
+ */
+static uint64_t sdma_v4_4_2_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u64 wptr;
+
+	if (ring->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		wptr = READ_ONCE(*((u64 *)&adev->wb.wb[ring->wptr_offs]));
+		DRM_DEBUG("wptr/doorbell before shift == 0x%016llx\n", wptr);
+	} else {
+		wptr = RREG32_SDMA(ring->me, regSDMA_GFX_RB_WPTR_HI);
+		wptr = wptr << 32;
+		wptr |= RREG32_SDMA(ring->me, regSDMA_GFX_RB_WPTR);
+		DRM_DEBUG("wptr before shift [%i] wptr == 0x%016llx\n",
+				ring->me, wptr);
+	}
+
+	return wptr >> 2;
+}
+
+/**
+ * sdma_v4_4_2_ring_set_wptr - commit the write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Write the wptr back to the hardware.
+ */
+static void sdma_v4_4_2_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	DRM_DEBUG("Setting write pointer\n");
+	if (ring->use_doorbell) {
+		u64 *wb = (u64 *)&adev->wb.wb[ring->wptr_offs];
+
+		DRM_DEBUG("Using doorbell -- "
+				"wptr_offs == 0x%08x "
+				"lower_32_bits(ring->wptr) << 2 == 0x%08x "
+				"upper_32_bits(ring->wptr) << 2 == 0x%08x\n",
+				ring->wptr_offs,
+				lower_32_bits(ring->wptr << 2),
+				upper_32_bits(ring->wptr << 2));
+		/* XXX check if swapping is necessary on BE */
+		WRITE_ONCE(*wb, (ring->wptr << 2));
+		DRM_DEBUG("calling WDOORBELL64(0x%08x, 0x%016llx)\n",
+				ring->doorbell_index, ring->wptr << 2);
+		WDOORBELL64(ring->doorbell_index, ring->wptr << 2);
+	} else {
+		DRM_DEBUG("Not using doorbell -- "
+				"regSDMA%i_GFX_RB_WPTR == 0x%08x "
+				"regSDMA%i_GFX_RB_WPTR_HI == 0x%08x\n",
+				ring->me,
+				lower_32_bits(ring->wptr << 2),
+				ring->me,
+				upper_32_bits(ring->wptr << 2));
+		WREG32_SDMA(ring->me, regSDMA_GFX_RB_WPTR,
+			    lower_32_bits(ring->wptr << 2));
+		WREG32_SDMA(ring->me, regSDMA_GFX_RB_WPTR_HI,
+			    upper_32_bits(ring->wptr << 2));
+	}
+}
+
+/**
+ * sdma_v4_4_2_page_ring_get_wptr - get the current write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current wptr from the hardware.
+ */
+static uint64_t sdma_v4_4_2_page_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u64 wptr;
+
+	if (ring->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		wptr = READ_ONCE(*((u64 *)&adev->wb.wb[ring->wptr_offs]));
+	} else {
+		wptr = RREG32_SDMA(ring->me, regSDMA_PAGE_RB_WPTR_HI);
+		wptr = wptr << 32;
+		wptr |= RREG32_SDMA(ring->me, regSDMA_PAGE_RB_WPTR);
+	}
+
+	return wptr >> 2;
+}
+
+/**
+ * sdma_v4_4_2_page_ring_set_wptr - commit the write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Write the wptr back to the hardware.
+ */
+static void sdma_v4_4_2_page_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		u64 *wb = (u64 *)&adev->wb.wb[ring->wptr_offs];
+
+		/* XXX check if swapping is necessary on BE */
+		WRITE_ONCE(*wb, (ring->wptr << 2));
+		WDOORBELL64(ring->doorbell_index, ring->wptr << 2);
+	} else {
+		uint64_t wptr = ring->wptr << 2;
+
+		WREG32_SDMA(ring->me, regSDMA_PAGE_RB_WPTR,
+			    lower_32_bits(wptr));
+		WREG32_SDMA(ring->me, regSDMA_PAGE_RB_WPTR_HI,
+			    upper_32_bits(wptr));
+	}
+}
+
+static void sdma_v4_4_2_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring);
+	int i;
+
+	for (i = 0; i < count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			amdgpu_ring_write(ring, ring->funcs->nop |
+				SDMA_PKT_NOP_HEADER_COUNT(count - 1));
+		else
+			amdgpu_ring_write(ring, ring->funcs->nop);
+}
+
+/**
+ * sdma_v4_4_2_ring_emit_ib - Schedule an IB on the DMA engine
+ *
+ * @ring: amdgpu ring pointer
+ * @job: job to retrieve vmid from
+ * @ib: IB object to schedule
+ * @flags: unused
+ *
+ * Schedule an IB in the DMA ring.
+ */
+static void sdma_v4_4_2_ring_emit_ib(struct amdgpu_ring *ring,
+				   struct amdgpu_job *job,
+				   struct amdgpu_ib *ib,
+				   uint32_t flags)
+{
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+
+	/* IB packet must end on a 8 DW boundary */
+	sdma_v4_4_2_ring_insert_nop(ring, (2 - lower_32_bits(ring->wptr)) & 7);
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_INDIRECT) |
+			  SDMA_PKT_INDIRECT_HEADER_VMID(vmid & 0xf));
+	/* base must be 32 byte aligned */
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr) & 0xffffffe0);
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, ib->length_dw);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0);
+
+}
+
+static void sdma_v4_4_2_wait_reg_mem(struct amdgpu_ring *ring,
+				   int mem_space, int hdp,
+				   uint32_t addr0, uint32_t addr1,
+				   uint32_t ref, uint32_t mask,
+				   uint32_t inv)
+{
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(hdp) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_MEM_POLL(mem_space) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3)); /* == */
+	if (mem_space) {
+		/* memory */
+		amdgpu_ring_write(ring, addr0);
+		amdgpu_ring_write(ring, addr1);
+	} else {
+		/* registers */
+		amdgpu_ring_write(ring, addr0 << 2);
+		amdgpu_ring_write(ring, addr1 << 2);
+	}
+	amdgpu_ring_write(ring, ref); /* reference */
+	amdgpu_ring_write(ring, mask); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(inv)); /* retry count, poll interval */
+}
+
+/**
+ * sdma_v4_4_2_ring_emit_hdp_flush - emit an hdp flush on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Emit an hdp flush packet on the requested DMA ring.
+ */
+static void sdma_v4_4_2_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 ref_and_mask = 0;
+	const struct nbio_hdp_flush_reg *nbio_hf_reg = adev->nbio.hdp_flush_reg;
+
+	ref_and_mask = nbio_hf_reg->ref_and_mask_sdma0
+		       << (ring->me % adev->sdma.num_inst_per_aid);
+
+	sdma_v4_4_2_wait_reg_mem(ring, 0, 1,
+			       adev->nbio.funcs->get_hdp_flush_done_offset(adev),
+			       adev->nbio.funcs->get_hdp_flush_req_offset(adev),
+			       ref_and_mask, ref_and_mask, 10);
+}
+
+/**
+ * sdma_v4_4_2_ring_emit_fence - emit a fence on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
+ *
+ * Add a DMA fence packet to the ring to write
+ * the fence seq number and DMA trap packet to generate
+ * an interrupt if needed.
+ */
+static void sdma_v4_4_2_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				      unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	/* write the fence */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_FENCE));
+	/* zero in first two bits */
+	BUG_ON(addr & 0x3);
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+
+	/* optionally write high bits as well */
+	if (write64bit) {
+		addr += 4;
+		amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_FENCE));
+		/* zero in first two bits */
+		BUG_ON(addr & 0x3);
+		amdgpu_ring_write(ring, lower_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(seq));
+	}
+
+	/* generate an interrupt */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_TRAP));
+	amdgpu_ring_write(ring, SDMA_PKT_TRAP_INT_CONTEXT_INT_CONTEXT(0));
+}
+
+
+/**
+ * sdma_v4_4_2_inst_gfx_stop - stop the gfx async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ * @inst_mask: mask of dma engine instances to be disabled
+ *
+ * Stop the gfx async dma ring buffers.
+ */
+static void sdma_v4_4_2_inst_gfx_stop(struct amdgpu_device *adev,
+				      uint32_t inst_mask)
+{
+	struct amdgpu_ring *sdma[AMDGPU_MAX_SDMA_INSTANCES];
+	u32 doorbell_offset, doorbell;
+	u32 rb_cntl, ib_cntl, sdma_cntl;
+	int i;
+
+	for_each_inst(i, inst_mask) {
+		sdma[i] = &adev->sdma.instance[i].ring;
+
+		rb_cntl = RREG32_SDMA(i, regSDMA_GFX_RB_CNTL);
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA_GFX_RB_CNTL, RB_ENABLE, 0);
+		WREG32_SDMA(i, regSDMA_GFX_RB_CNTL, rb_cntl);
+		ib_cntl = RREG32_SDMA(i, regSDMA_GFX_IB_CNTL);
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA_GFX_IB_CNTL, IB_ENABLE, 0);
+		WREG32_SDMA(i, regSDMA_GFX_IB_CNTL, ib_cntl);
+		sdma_cntl = RREG32_SDMA(i, regSDMA_CNTL);
+		sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA_CNTL, UTC_L1_ENABLE, 0);
+		WREG32_SDMA(i, regSDMA_CNTL, sdma_cntl);
+
+		if (sdma[i]->use_doorbell) {
+			doorbell = RREG32_SDMA(i, regSDMA_GFX_DOORBELL);
+			doorbell_offset = RREG32_SDMA(i, regSDMA_GFX_DOORBELL_OFFSET);
+
+			doorbell = REG_SET_FIELD(doorbell, SDMA_GFX_DOORBELL, ENABLE, 0);
+			doorbell_offset = REG_SET_FIELD(doorbell_offset,
+					SDMA_GFX_DOORBELL_OFFSET,
+					OFFSET, 0);
+			WREG32_SDMA(i, regSDMA_GFX_DOORBELL, doorbell);
+			WREG32_SDMA(i, regSDMA_GFX_DOORBELL_OFFSET, doorbell_offset);
+		}
+	}
+}
+
+/**
+ * sdma_v4_4_2_inst_rlc_stop - stop the compute async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ * @inst_mask: mask of dma engine instances to be disabled
+ *
+ * Stop the compute async dma queues.
+ */
+static void sdma_v4_4_2_inst_rlc_stop(struct amdgpu_device *adev,
+				      uint32_t inst_mask)
+{
+	/* XXX todo */
+}
+
+/**
+ * sdma_v4_4_2_inst_page_stop - stop the page async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ * @inst_mask: mask of dma engine instances to be disabled
+ *
+ * Stop the page async dma ring buffers.
+ */
+static void sdma_v4_4_2_inst_page_stop(struct amdgpu_device *adev,
+				       uint32_t inst_mask)
+{
+	u32 rb_cntl, ib_cntl;
+	int i;
+
+	for_each_inst(i, inst_mask) {
+		rb_cntl = RREG32_SDMA(i, regSDMA_PAGE_RB_CNTL);
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA_PAGE_RB_CNTL,
+					RB_ENABLE, 0);
+		WREG32_SDMA(i, regSDMA_PAGE_RB_CNTL, rb_cntl);
+		ib_cntl = RREG32_SDMA(i, regSDMA_PAGE_IB_CNTL);
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA_PAGE_IB_CNTL,
+					IB_ENABLE, 0);
+		WREG32_SDMA(i, regSDMA_PAGE_IB_CNTL, ib_cntl);
+	}
+}
+
+/**
+ * sdma_v4_4_2_inst_ctx_switch_enable - stop the async dma engines context switch
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable the DMA MEs context switch.
+ * @inst_mask: mask of dma engine instances to be enabled
+ *
+ * Halt or unhalt the async dma engines context switch.
+ */
+static void sdma_v4_4_2_inst_ctx_switch_enable(struct amdgpu_device *adev,
+					       bool enable, uint32_t inst_mask)
+{
+	u32 f32_cntl, phase_quantum = 0;
+	int i;
+
+	if (amdgpu_sdma_phase_quantum) {
+		unsigned value = amdgpu_sdma_phase_quantum;
+		unsigned unit = 0;
+
+		while (value > (SDMA_PHASE0_QUANTUM__VALUE_MASK >>
+				SDMA_PHASE0_QUANTUM__VALUE__SHIFT)) {
+			value = (value + 1) >> 1;
+			unit++;
+		}
+		if (unit > (SDMA_PHASE0_QUANTUM__UNIT_MASK >>
+			    SDMA_PHASE0_QUANTUM__UNIT__SHIFT)) {
+			value = (SDMA_PHASE0_QUANTUM__VALUE_MASK >>
+				 SDMA_PHASE0_QUANTUM__VALUE__SHIFT);
+			unit = (SDMA_PHASE0_QUANTUM__UNIT_MASK >>
+				SDMA_PHASE0_QUANTUM__UNIT__SHIFT);
+			WARN_ONCE(1,
+			"clamping sdma_phase_quantum to %uK clock cycles\n",
+				  value << unit);
+		}
+		phase_quantum =
+			value << SDMA_PHASE0_QUANTUM__VALUE__SHIFT |
+			unit  << SDMA_PHASE0_QUANTUM__UNIT__SHIFT;
+	}
+
+	for_each_inst(i, inst_mask) {
+		f32_cntl = RREG32_SDMA(i, regSDMA_CNTL);
+		f32_cntl = REG_SET_FIELD(f32_cntl, SDMA_CNTL,
+				AUTO_CTXSW_ENABLE, enable ? 1 : 0);
+		if (enable && amdgpu_sdma_phase_quantum) {
+			WREG32_SDMA(i, regSDMA_PHASE0_QUANTUM, phase_quantum);
+			WREG32_SDMA(i, regSDMA_PHASE1_QUANTUM, phase_quantum);
+			WREG32_SDMA(i, regSDMA_PHASE2_QUANTUM, phase_quantum);
+		}
+		WREG32_SDMA(i, regSDMA_CNTL, f32_cntl);
+
+		/* Extend page fault timeout to avoid interrupt storm */
+		WREG32_SDMA(i, regSDMA_UTCL1_TIMEOUT, 0x00800080);
+	}
+}
+
+/**
+ * sdma_v4_4_2_inst_enable - stop the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable the DMA MEs.
+ * @inst_mask: mask of dma engine instances to be enabled
+ *
+ * Halt or unhalt the async dma engines.
+ */
+static void sdma_v4_4_2_inst_enable(struct amdgpu_device *adev, bool enable,
+				    uint32_t inst_mask)
+{
+	u32 f32_cntl;
+	int i;
+
+	if (!enable) {
+		sdma_v4_4_2_inst_gfx_stop(adev, inst_mask);
+		sdma_v4_4_2_inst_rlc_stop(adev, inst_mask);
+		if (adev->sdma.has_page_queue)
+			sdma_v4_4_2_inst_page_stop(adev, inst_mask);
+
+		/* SDMA FW needs to respond to FREEZE requests during reset.
+		 * Keep it running during reset */
+		if (!amdgpu_sriov_vf(adev) && amdgpu_in_reset(adev))
+			return;
+	}
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP)
+		return;
+
+	for_each_inst(i, inst_mask) {
+		f32_cntl = RREG32_SDMA(i, regSDMA_F32_CNTL);
+		f32_cntl = REG_SET_FIELD(f32_cntl, SDMA_F32_CNTL, HALT, enable ? 0 : 1);
+		WREG32_SDMA(i, regSDMA_F32_CNTL, f32_cntl);
+	}
+}
+
+/*
+ * sdma_v4_4_2_rb_cntl - get parameters for rb_cntl
+ */
+static uint32_t sdma_v4_4_2_rb_cntl(struct amdgpu_ring *ring, uint32_t rb_cntl)
+{
+	/* Set ring buffer size in dwords */
+	uint32_t rb_bufsz = order_base_2(ring->ring_size / 4);
+
+	barrier(); /* work around https://llvm.org/pr42576 */
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA_GFX_RB_CNTL, RB_SIZE, rb_bufsz);
+#ifdef __BIG_ENDIAN
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA_GFX_RB_CNTL, RB_SWAP_ENABLE, 1);
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA_GFX_RB_CNTL,
+				RPTR_WRITEBACK_SWAP_ENABLE, 1);
+#endif
+	return rb_cntl;
+}
+
+/**
+ * sdma_v4_4_2_gfx_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ * @i: instance to resume
+ * @restore: used to restore wptr when restart
+ *
+ * Set up the gfx DMA ring buffers and enable them.
+ * Returns 0 for success, error for failure.
+ */
+static void sdma_v4_4_2_gfx_resume(struct amdgpu_device *adev, unsigned int i, bool restore)
+{
+	struct amdgpu_ring *ring = &adev->sdma.instance[i].ring;
+	u32 rb_cntl, ib_cntl, wptr_poll_cntl;
+	u32 wb_offset;
+	u32 doorbell;
+	u32 doorbell_offset;
+	u64 wptr_gpu_addr;
+	u64 rwptr;
+
+	wb_offset = (ring->rptr_offs * 4);
+
+	rb_cntl = RREG32_SDMA(i, regSDMA_GFX_RB_CNTL);
+	rb_cntl = sdma_v4_4_2_rb_cntl(ring, rb_cntl);
+	WREG32_SDMA(i, regSDMA_GFX_RB_CNTL, rb_cntl);
+
+	/* set the wb address whether it's enabled or not */
+	WREG32_SDMA(i, regSDMA_GFX_RB_RPTR_ADDR_HI,
+	       upper_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFF);
+	WREG32_SDMA(i, regSDMA_GFX_RB_RPTR_ADDR_LO,
+	       lower_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFC);
+
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA_GFX_RB_CNTL,
+				RPTR_WRITEBACK_ENABLE, 1);
+
+	WREG32_SDMA(i, regSDMA_GFX_RB_BASE, ring->gpu_addr >> 8);
+	WREG32_SDMA(i, regSDMA_GFX_RB_BASE_HI, ring->gpu_addr >> 40);
+
+	if (!restore)
+		ring->wptr = 0;
+
+	/* before programing wptr to a less value, need set minor_ptr_update first */
+	WREG32_SDMA(i, regSDMA_GFX_MINOR_PTR_UPDATE, 1);
+
+	/* For the guilty queue, set RPTR to the current wptr to skip bad commands,
+	 * It is not a guilty queue, restore cache_rptr and continue execution.
+	 */
+	if (adev->sdma.instance[i].gfx_guilty)
+		rwptr = ring->wptr;
+	else
+		rwptr = ring->cached_rptr;
+
+	/* Initialize the ring buffer's read and write pointers */
+	if (restore) {
+		WREG32_SDMA(i, regSDMA_GFX_RB_RPTR, lower_32_bits(rwptr << 2));
+		WREG32_SDMA(i, regSDMA_GFX_RB_RPTR_HI, upper_32_bits(rwptr << 2));
+		WREG32_SDMA(i, regSDMA_GFX_RB_WPTR, lower_32_bits(rwptr << 2));
+		WREG32_SDMA(i, regSDMA_GFX_RB_WPTR_HI, upper_32_bits(rwptr << 2));
+	} else {
+		WREG32_SDMA(i, regSDMA_GFX_RB_RPTR, 0);
+		WREG32_SDMA(i, regSDMA_GFX_RB_RPTR_HI, 0);
+		WREG32_SDMA(i, regSDMA_GFX_RB_WPTR, 0);
+		WREG32_SDMA(i, regSDMA_GFX_RB_WPTR_HI, 0);
+	}
+
+	doorbell = RREG32_SDMA(i, regSDMA_GFX_DOORBELL);
+	doorbell_offset = RREG32_SDMA(i, regSDMA_GFX_DOORBELL_OFFSET);
+
+	doorbell = REG_SET_FIELD(doorbell, SDMA_GFX_DOORBELL, ENABLE,
+				 ring->use_doorbell);
+	doorbell_offset = REG_SET_FIELD(doorbell_offset,
+					SDMA_GFX_DOORBELL_OFFSET,
+					OFFSET, ring->doorbell_index);
+	WREG32_SDMA(i, regSDMA_GFX_DOORBELL, doorbell);
+	WREG32_SDMA(i, regSDMA_GFX_DOORBELL_OFFSET, doorbell_offset);
+
+	sdma_v4_4_2_ring_set_wptr(ring);
+
+	/* set minor_ptr_update to 0 after wptr programed */
+	WREG32_SDMA(i, regSDMA_GFX_MINOR_PTR_UPDATE, 0);
+
+	/* setup the wptr shadow polling */
+	wptr_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
+	WREG32_SDMA(i, regSDMA_GFX_RB_WPTR_POLL_ADDR_LO,
+		    lower_32_bits(wptr_gpu_addr));
+	WREG32_SDMA(i, regSDMA_GFX_RB_WPTR_POLL_ADDR_HI,
+		    upper_32_bits(wptr_gpu_addr));
+	wptr_poll_cntl = RREG32_SDMA(i, regSDMA_GFX_RB_WPTR_POLL_CNTL);
+	wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl,
+				       SDMA_GFX_RB_WPTR_POLL_CNTL,
+				       F32_POLL_ENABLE, amdgpu_sriov_vf(adev)? 1 : 0);
+	WREG32_SDMA(i, regSDMA_GFX_RB_WPTR_POLL_CNTL, wptr_poll_cntl);
+
+	/* enable DMA RB */
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA_GFX_RB_CNTL, RB_ENABLE, 1);
+	WREG32_SDMA(i, regSDMA_GFX_RB_CNTL, rb_cntl);
+
+	ib_cntl = RREG32_SDMA(i, regSDMA_GFX_IB_CNTL);
+	ib_cntl = REG_SET_FIELD(ib_cntl, SDMA_GFX_IB_CNTL, IB_ENABLE, 1);
+#ifdef __BIG_ENDIAN
+	ib_cntl = REG_SET_FIELD(ib_cntl, SDMA_GFX_IB_CNTL, IB_SWAP_ENABLE, 1);
+#endif
+	/* enable DMA IBs */
+	WREG32_SDMA(i, regSDMA_GFX_IB_CNTL, ib_cntl);
+}
+
+/**
+ * sdma_v4_4_2_page_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ * @i: instance to resume
+ * @restore: boolean to say restore needed or not
+ *
+ * Set up the page DMA ring buffers and enable them.
+ * Returns 0 for success, error for failure.
+ */
+static void sdma_v4_4_2_page_resume(struct amdgpu_device *adev, unsigned int i, bool restore)
+{
+	struct amdgpu_ring *ring = &adev->sdma.instance[i].page;
+	u32 rb_cntl, ib_cntl, wptr_poll_cntl;
+	u32 wb_offset;
+	u32 doorbell;
+	u32 doorbell_offset;
+	u64 wptr_gpu_addr;
+	u64 rwptr;
+
+	wb_offset = (ring->rptr_offs * 4);
+
+	rb_cntl = RREG32_SDMA(i, regSDMA_PAGE_RB_CNTL);
+	rb_cntl = sdma_v4_4_2_rb_cntl(ring, rb_cntl);
+	WREG32_SDMA(i, regSDMA_PAGE_RB_CNTL, rb_cntl);
+
+	/* For the guilty queue, set RPTR to the current wptr to skip bad commands,
+	 * It is not a guilty queue, restore cache_rptr and continue execution.
+	 */
+	if (adev->sdma.instance[i].page_guilty)
+		rwptr = ring->wptr;
+	else
+		rwptr = ring->cached_rptr;
+
+	/* Initialize the ring buffer's read and write pointers */
+	if (restore) {
+		WREG32_SDMA(i, regSDMA_PAGE_RB_RPTR, lower_32_bits(rwptr << 2));
+		WREG32_SDMA(i, regSDMA_PAGE_RB_RPTR_HI, upper_32_bits(rwptr << 2));
+		WREG32_SDMA(i, regSDMA_PAGE_RB_WPTR, lower_32_bits(rwptr << 2));
+		WREG32_SDMA(i, regSDMA_PAGE_RB_WPTR_HI, upper_32_bits(rwptr << 2));
+	} else {
+		WREG32_SDMA(i, regSDMA_PAGE_RB_RPTR, 0);
+		WREG32_SDMA(i, regSDMA_PAGE_RB_RPTR_HI, 0);
+		WREG32_SDMA(i, regSDMA_PAGE_RB_WPTR, 0);
+		WREG32_SDMA(i, regSDMA_PAGE_RB_WPTR_HI, 0);
+	}
+
+	/* set the wb address whether it's enabled or not */
+	WREG32_SDMA(i, regSDMA_PAGE_RB_RPTR_ADDR_HI,
+	       upper_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFF);
+	WREG32_SDMA(i, regSDMA_PAGE_RB_RPTR_ADDR_LO,
+	       lower_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFC);
+
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA_PAGE_RB_CNTL,
+				RPTR_WRITEBACK_ENABLE, 1);
+
+	WREG32_SDMA(i, regSDMA_PAGE_RB_BASE, ring->gpu_addr >> 8);
+	WREG32_SDMA(i, regSDMA_PAGE_RB_BASE_HI, ring->gpu_addr >> 40);
+
+	if (!restore)
+		ring->wptr = 0;
+
+	/* before programing wptr to a less value, need set minor_ptr_update first */
+	WREG32_SDMA(i, regSDMA_PAGE_MINOR_PTR_UPDATE, 1);
+
+	doorbell = RREG32_SDMA(i, regSDMA_PAGE_DOORBELL);
+	doorbell_offset = RREG32_SDMA(i, regSDMA_PAGE_DOORBELL_OFFSET);
+
+	doorbell = REG_SET_FIELD(doorbell, SDMA_PAGE_DOORBELL, ENABLE,
+				 ring->use_doorbell);
+	doorbell_offset = REG_SET_FIELD(doorbell_offset,
+					SDMA_PAGE_DOORBELL_OFFSET,
+					OFFSET, ring->doorbell_index);
+	WREG32_SDMA(i, regSDMA_PAGE_DOORBELL, doorbell);
+	WREG32_SDMA(i, regSDMA_PAGE_DOORBELL_OFFSET, doorbell_offset);
+
+	/* paging queue doorbell range is setup at sdma_v4_4_2_gfx_resume */
+	sdma_v4_4_2_page_ring_set_wptr(ring);
+
+	/* set minor_ptr_update to 0 after wptr programed */
+	WREG32_SDMA(i, regSDMA_PAGE_MINOR_PTR_UPDATE, 0);
+
+	/* setup the wptr shadow polling */
+	wptr_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
+	WREG32_SDMA(i, regSDMA_PAGE_RB_WPTR_POLL_ADDR_LO,
+		    lower_32_bits(wptr_gpu_addr));
+	WREG32_SDMA(i, regSDMA_PAGE_RB_WPTR_POLL_ADDR_HI,
+		    upper_32_bits(wptr_gpu_addr));
+	wptr_poll_cntl = RREG32_SDMA(i, regSDMA_PAGE_RB_WPTR_POLL_CNTL);
+	wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl,
+				       SDMA_PAGE_RB_WPTR_POLL_CNTL,
+				       F32_POLL_ENABLE, amdgpu_sriov_vf(adev)? 1 : 0);
+	WREG32_SDMA(i, regSDMA_PAGE_RB_WPTR_POLL_CNTL, wptr_poll_cntl);
+
+	/* enable DMA RB */
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA_PAGE_RB_CNTL, RB_ENABLE, 1);
+	WREG32_SDMA(i, regSDMA_PAGE_RB_CNTL, rb_cntl);
+
+	ib_cntl = RREG32_SDMA(i, regSDMA_PAGE_IB_CNTL);
+	ib_cntl = REG_SET_FIELD(ib_cntl, SDMA_PAGE_IB_CNTL, IB_ENABLE, 1);
+#ifdef __BIG_ENDIAN
+	ib_cntl = REG_SET_FIELD(ib_cntl, SDMA_PAGE_IB_CNTL, IB_SWAP_ENABLE, 1);
+#endif
+	/* enable DMA IBs */
+	WREG32_SDMA(i, regSDMA_PAGE_IB_CNTL, ib_cntl);
+}
+
+static void sdma_v4_4_2_init_pg(struct amdgpu_device *adev)
+{
+
+}
+
+/**
+ * sdma_v4_4_2_inst_rlc_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ * @inst_mask: mask of dma engine instances to be enabled
+ *
+ * Set up the compute DMA queues and enable them.
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v4_4_2_inst_rlc_resume(struct amdgpu_device *adev,
+				       uint32_t inst_mask)
+{
+	sdma_v4_4_2_init_pg(adev);
+
+	return 0;
+}
+
+/**
+ * sdma_v4_4_2_inst_load_microcode - load the sDMA ME ucode
+ *
+ * @adev: amdgpu_device pointer
+ * @inst_mask: mask of dma engine instances to be enabled
+ *
+ * Loads the sDMA0/1 ucode.
+ * Returns 0 for success, -EINVAL if the ucode is not available.
+ */
+static int sdma_v4_4_2_inst_load_microcode(struct amdgpu_device *adev,
+					   uint32_t inst_mask)
+{
+	const struct sdma_firmware_header_v1_0 *hdr;
+	const __le32 *fw_data;
+	u32 fw_size;
+	int i, j;
+
+	/* halt the MEs */
+	sdma_v4_4_2_inst_enable(adev, false, inst_mask);
+
+	for_each_inst(i, inst_mask) {
+		if (!adev->sdma.instance[i].fw)
+			return -EINVAL;
+
+		hdr = (const struct sdma_firmware_header_v1_0 *)adev->sdma.instance[i].fw->data;
+		amdgpu_ucode_print_sdma_hdr(&hdr->header);
+		fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+
+		fw_data = (const __le32 *)
+			(adev->sdma.instance[i].fw->data +
+				le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+
+		WREG32_SDMA(i, regSDMA_UCODE_ADDR, 0);
+
+		for (j = 0; j < fw_size; j++)
+			WREG32_SDMA(i, regSDMA_UCODE_DATA,
+				    le32_to_cpup(fw_data++));
+
+		WREG32_SDMA(i, regSDMA_UCODE_ADDR,
+			    adev->sdma.instance[i].fw_version);
+	}
+
+	return 0;
+}
+
+/**
+ * sdma_v4_4_2_inst_start - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ * @inst_mask: mask of dma engine instances to be enabled
+ * @restore: boolean to say restore needed or not
+ *
+ * Set up the DMA engines and enable them.
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v4_4_2_inst_start(struct amdgpu_device *adev,
+				  uint32_t inst_mask, bool restore)
+{
+	struct amdgpu_ring *ring;
+	uint32_t tmp_mask;
+	int i, r = 0;
+
+	if (amdgpu_sriov_vf(adev)) {
+		sdma_v4_4_2_inst_ctx_switch_enable(adev, false, inst_mask);
+		sdma_v4_4_2_inst_enable(adev, false, inst_mask);
+	} else {
+		/* bypass sdma microcode loading on Gopher */
+		if (!restore && adev->firmware.load_type != AMDGPU_FW_LOAD_PSP &&
+		    adev->sdma.instance[0].fw) {
+			r = sdma_v4_4_2_inst_load_microcode(adev, inst_mask);
+			if (r)
+				return r;
+		}
+
+		/* unhalt the MEs */
+		sdma_v4_4_2_inst_enable(adev, true, inst_mask);
+		/* enable sdma ring preemption */
+		sdma_v4_4_2_inst_ctx_switch_enable(adev, true, inst_mask);
+	}
+
+	/* start the gfx rings and rlc compute queues */
+	tmp_mask = inst_mask;
+	for_each_inst(i, tmp_mask) {
+		uint32_t temp;
+
+		WREG32_SDMA(i, regSDMA_SEM_WAIT_FAIL_TIMER_CNTL, 0);
+		sdma_v4_4_2_gfx_resume(adev, i, restore);
+		if (adev->sdma.has_page_queue)
+			sdma_v4_4_2_page_resume(adev, i, restore);
+
+		/* set utc l1 enable flag always to 1 */
+		temp = RREG32_SDMA(i, regSDMA_CNTL);
+		temp = REG_SET_FIELD(temp, SDMA_CNTL, UTC_L1_ENABLE, 1);
+		WREG32_SDMA(i, regSDMA_CNTL, temp);
+
+		if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) < IP_VERSION(4, 4, 5)) {
+			/* enable context empty interrupt during initialization */
+			temp = REG_SET_FIELD(temp, SDMA_CNTL, CTXEMPTY_INT_ENABLE, 1);
+			WREG32_SDMA(i, regSDMA_CNTL, temp);
+		}
+		if (!amdgpu_sriov_vf(adev)) {
+			if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
+				/* unhalt engine */
+				temp = RREG32_SDMA(i, regSDMA_F32_CNTL);
+				temp = REG_SET_FIELD(temp, SDMA_F32_CNTL, HALT, 0);
+				WREG32_SDMA(i, regSDMA_F32_CNTL, temp);
+			}
+		}
+	}
+
+	if (amdgpu_sriov_vf(adev)) {
+		sdma_v4_4_2_inst_ctx_switch_enable(adev, true, inst_mask);
+		sdma_v4_4_2_inst_enable(adev, true, inst_mask);
+	} else {
+		r = sdma_v4_4_2_inst_rlc_resume(adev, inst_mask);
+		if (r)
+			return r;
+	}
+
+	tmp_mask = inst_mask;
+	for_each_inst(i, tmp_mask) {
+		ring = &adev->sdma.instance[i].ring;
+
+		r = amdgpu_ring_test_helper(ring);
+		if (r)
+			return r;
+
+		if (adev->sdma.has_page_queue) {
+			struct amdgpu_ring *page = &adev->sdma.instance[i].page;
+
+			r = amdgpu_ring_test_helper(page);
+			if (r)
+				return r;
+		}
+	}
+
+	return r;
+}
+
+/**
+ * sdma_v4_4_2_ring_test_ring - simple async dma engine test
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Test the DMA engine by writing using it to write an
+ * value to memory.
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v4_4_2_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned i;
+	unsigned index;
+	int r;
+	u32 tmp;
+	u64 gpu_addr;
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r)
+		return r;
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	tmp = 0xCAFEDEAD;
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+
+	r = amdgpu_ring_alloc(ring, 5);
+	if (r)
+		goto error_free_wb;
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+			  SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR));
+	amdgpu_ring_write(ring, lower_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(0));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = le32_to_cpu(adev->wb.wb[index]);
+		if (tmp == 0xDEADBEEF)
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+
+error_free_wb:
+	amdgpu_device_wb_free(adev, index);
+	return r;
+}
+
+/**
+ * sdma_v4_4_2_ring_test_ib - test an IB on the DMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @timeout: timeout value in jiffies, or MAX_SCHEDULE_TIMEOUT
+ *
+ * Test a simple IB in the DMA ring.
+ * Returns 0 on success, error on failure.
+ */
+static int sdma_v4_4_2_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct dma_fence *f = NULL;
+	unsigned index;
+	long r;
+	u32 tmp = 0;
+	u64 gpu_addr;
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r)
+		return r;
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	tmp = 0xCAFEDEAD;
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+	memset(&ib, 0, sizeof(ib));
+	r = amdgpu_ib_get(adev, NULL, 256,
+					AMDGPU_IB_POOL_DIRECT, &ib);
+	if (r)
+		goto err0;
+
+	ib.ptr[0] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR);
+	ib.ptr[1] = lower_32_bits(gpu_addr);
+	ib.ptr[2] = upper_32_bits(gpu_addr);
+	ib.ptr[3] = SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(0);
+	ib.ptr[4] = 0xDEADBEEF;
+	ib.ptr[5] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.ptr[6] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.ptr[7] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.length_dw = 8;
+
+	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
+	if (r)
+		goto err1;
+
+	r = dma_fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		r = -ETIMEDOUT;
+		goto err1;
+	} else if (r < 0) {
+		goto err1;
+	}
+	tmp = le32_to_cpu(adev->wb.wb[index]);
+	if (tmp == 0xDEADBEEF)
+		r = 0;
+	else
+		r = -EINVAL;
+
+err1:
+	amdgpu_ib_free(&ib, NULL);
+	dma_fence_put(f);
+err0:
+	amdgpu_device_wb_free(adev, index);
+	return r;
+}
+
+
+/**
+ * sdma_v4_4_2_vm_copy_pte - update PTEs by copying them from the GART
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @src: src addr to copy from
+ * @count: number of page entries to update
+ *
+ * Update PTEs by copying them from the GART using sDMA.
+ */
+static void sdma_v4_4_2_vm_copy_pte(struct amdgpu_ib *ib,
+				  uint64_t pe, uint64_t src,
+				  unsigned count)
+{
+	unsigned bytes = count * 8;
+
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
+	ib->ptr[ib->length_dw++] = bytes - 1;
+	ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+	ib->ptr[ib->length_dw++] = lower_32_bits(src);
+	ib->ptr[ib->length_dw++] = upper_32_bits(src);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+
+}
+
+/**
+ * sdma_v4_4_2_vm_write_pte - update PTEs by writing them manually
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @value: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ *
+ * Update PTEs by writing them manually using sDMA.
+ */
+static void sdma_v4_4_2_vm_write_pte(struct amdgpu_ib *ib, uint64_t pe,
+				   uint64_t value, unsigned count,
+				   uint32_t incr)
+{
+	unsigned ndw = count * 2;
+
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+	ib->ptr[ib->length_dw++] = ndw - 1;
+	for (; ndw > 0; ndw -= 2) {
+		ib->ptr[ib->length_dw++] = lower_32_bits(value);
+		ib->ptr[ib->length_dw++] = upper_32_bits(value);
+		value += incr;
+	}
+}
+
+/**
+ * sdma_v4_4_2_vm_set_pte_pde - update the page tables using sDMA
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: access flags
+ *
+ * Update the page tables using sDMA.
+ */
+static void sdma_v4_4_2_vm_set_pte_pde(struct amdgpu_ib *ib,
+				     uint64_t pe,
+				     uint64_t addr, unsigned count,
+				     uint32_t incr, uint64_t flags)
+{
+	/* for physically contiguous pages (vram) */
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_PTEPDE);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe); /* dst addr */
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+	ib->ptr[ib->length_dw++] = lower_32_bits(flags); /* mask */
+	ib->ptr[ib->length_dw++] = upper_32_bits(flags);
+	ib->ptr[ib->length_dw++] = lower_32_bits(addr); /* value */
+	ib->ptr[ib->length_dw++] = upper_32_bits(addr);
+	ib->ptr[ib->length_dw++] = incr; /* increment size */
+	ib->ptr[ib->length_dw++] = 0;
+	ib->ptr[ib->length_dw++] = count - 1; /* number of entries */
+}
+
+/**
+ * sdma_v4_4_2_ring_pad_ib - pad the IB to the required number of dw
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @ib: indirect buffer to fill with padding
+ */
+static void sdma_v4_4_2_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring);
+	u32 pad_count;
+	int i;
+
+	pad_count = (-ib->length_dw) & 7;
+	for (i = 0; i < pad_count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			ib->ptr[ib->length_dw++] =
+				SDMA_PKT_HEADER_OP(SDMA_OP_NOP) |
+				SDMA_PKT_NOP_HEADER_COUNT(pad_count - 1);
+		else
+			ib->ptr[ib->length_dw++] =
+				SDMA_PKT_HEADER_OP(SDMA_OP_NOP);
+}
+
+
+/**
+ * sdma_v4_4_2_ring_emit_pipeline_sync - sync the pipeline
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Make sure all previous operations are completed (CIK).
+ */
+static void sdma_v4_4_2_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	/* wait for idle */
+	sdma_v4_4_2_wait_reg_mem(ring, 1, 0,
+			       addr & 0xfffffffc,
+			       upper_32_bits(addr) & 0xffffffff,
+			       seq, 0xffffffff, 4);
+}
+
+
+/**
+ * sdma_v4_4_2_ring_emit_vm_flush - vm flush using sDMA
+ *
+ * @ring: amdgpu_ring pointer
+ * @vmid: vmid number to use
+ * @pd_addr: address
+ *
+ * Update the page table base and flush the VM TLB
+ * using sDMA.
+ */
+static void sdma_v4_4_2_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					 unsigned vmid, uint64_t pd_addr)
+{
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+}
+
+static void sdma_v4_4_2_ring_emit_wreg(struct amdgpu_ring *ring,
+				     uint32_t reg, uint32_t val)
+{
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
+			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, val);
+}
+
+static void sdma_v4_4_2_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+					 uint32_t val, uint32_t mask)
+{
+	sdma_v4_4_2_wait_reg_mem(ring, 0, 0, reg, 0, val, mask, 10);
+}
+
+static bool sdma_v4_4_2_fw_support_paging_queue(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) {
+	case IP_VERSION(4, 4, 2):
+	case IP_VERSION(4, 4, 5):
+		return false;
+	default:
+		return false;
+	}
+}
+
+static const struct amdgpu_sdma_funcs sdma_v4_4_2_sdma_funcs = {
+	.stop_kernel_queue = &sdma_v4_4_2_stop_queue,
+	.start_kernel_queue = &sdma_v4_4_2_restore_queue,
+	.soft_reset_kernel_queue = &sdma_v4_4_2_soft_reset_engine,
+};
+
+static int sdma_v4_4_2_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = sdma_v4_4_2_init_microcode(adev);
+	if (r)
+		return r;
+
+	/* TODO: Page queue breaks driver reload under SRIOV */
+	if (sdma_v4_4_2_fw_support_paging_queue(adev))
+		adev->sdma.has_page_queue = true;
+
+	sdma_v4_4_2_set_ring_funcs(adev);
+	sdma_v4_4_2_set_buffer_funcs(adev);
+	sdma_v4_4_2_set_vm_pte_funcs(adev);
+	sdma_v4_4_2_set_irq_funcs(adev);
+	sdma_v4_4_2_set_ras_funcs(adev);
+	return 0;
+}
+
+#if 0
+static int sdma_v4_4_2_process_ras_data_cb(struct amdgpu_device *adev,
+		void *err_data,
+		struct amdgpu_iv_entry *entry);
+#endif
+
+static int sdma_v4_4_2_late_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+#if 0
+	struct ras_ih_if ih_info = {
+		.cb = sdma_v4_4_2_process_ras_data_cb,
+	};
+#endif
+	if (!amdgpu_persistent_edc_harvesting_supported(adev))
+		amdgpu_ras_reset_error_count(adev, AMDGPU_RAS_BLOCK__SDMA);
+
+	/* The initialization is done in the late_init stage to ensure that the SMU
+	 * initialization and capability setup are completed before we check the SDMA
+	 * reset capability
+	 */
+	sdma_v4_4_2_update_reset_mask(adev);
+
+	return 0;
+}
+
+static int sdma_v4_4_2_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_ring *ring;
+	int r, i;
+	struct amdgpu_device *adev = ip_block->adev;
+	u32 aid_id;
+	uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_4_4_2);
+	uint32_t *ptr;
+
+	/* SDMA trap event */
+	for (i = 0; i < adev->sdma.num_inst_per_aid; i++) {
+		r = amdgpu_irq_add_id(adev, sdma_v4_4_2_seq_to_irq_id(i),
+				      SDMA0_4_0__SRCID__SDMA_TRAP,
+				      &adev->sdma.trap_irq);
+		if (r)
+			return r;
+	}
+
+	/* SDMA SRAM ECC event */
+	for (i = 0; i < adev->sdma.num_inst_per_aid; i++) {
+		r = amdgpu_irq_add_id(adev, sdma_v4_4_2_seq_to_irq_id(i),
+				      SDMA0_4_0__SRCID__SDMA_SRAM_ECC,
+				      &adev->sdma.ecc_irq);
+		if (r)
+			return r;
+	}
+
+	/* SDMA VM_HOLE/DOORBELL_INV/POLL_TIMEOUT/SRBM_WRITE_PROTECTION event*/
+	for (i = 0; i < adev->sdma.num_inst_per_aid; i++) {
+		r = amdgpu_irq_add_id(adev, sdma_v4_4_2_seq_to_irq_id(i),
+				      SDMA0_4_0__SRCID__SDMA_VM_HOLE,
+				      &adev->sdma.vm_hole_irq);
+		if (r)
+			return r;
+
+		r = amdgpu_irq_add_id(adev, sdma_v4_4_2_seq_to_irq_id(i),
+				      SDMA0_4_0__SRCID__SDMA_DOORBELL_INVALID,
+				      &adev->sdma.doorbell_invalid_irq);
+		if (r)
+			return r;
+
+		r = amdgpu_irq_add_id(adev, sdma_v4_4_2_seq_to_irq_id(i),
+				      SDMA0_4_0__SRCID__SDMA_POLL_TIMEOUT,
+				      &adev->sdma.pool_timeout_irq);
+		if (r)
+			return r;
+
+		r = amdgpu_irq_add_id(adev, sdma_v4_4_2_seq_to_irq_id(i),
+				      SDMA0_4_0__SRCID__SDMA_SRBMWRITE,
+				      &adev->sdma.srbm_write_irq);
+		if (r)
+			return r;
+
+		r = amdgpu_irq_add_id(adev, sdma_v4_4_2_seq_to_irq_id(i),
+				      SDMA0_4_0__SRCID__SDMA_CTXEMPTY,
+				      &adev->sdma.ctxt_empty_irq);
+		if (r)
+			return r;
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		mutex_init(&adev->sdma.instance[i].engine_reset_mutex);
+		/* Initialize guilty flags for GFX and PAGE queues */
+		adev->sdma.instance[i].gfx_guilty = false;
+		adev->sdma.instance[i].page_guilty = false;
+		adev->sdma.instance[i].funcs = &sdma_v4_4_2_sdma_funcs;
+
+		ring = &adev->sdma.instance[i].ring;
+		ring->ring_obj = NULL;
+		ring->use_doorbell = true;
+		aid_id = adev->sdma.instance[i].aid_id;
+
+		DRM_DEBUG("SDMA %d use_doorbell being set to: [%s]\n", i,
+				ring->use_doorbell?"true":"false");
+
+		/* doorbell size is 2 dwords, get DWORD offset */
+		ring->doorbell_index = adev->doorbell_index.sdma_engine[i] << 1;
+		ring->vm_hub = AMDGPU_MMHUB0(aid_id);
+
+		sprintf(ring->name, "sdma%d.%d", aid_id,
+				i % adev->sdma.num_inst_per_aid);
+		r = amdgpu_ring_init(adev, ring, 1024, &adev->sdma.trap_irq,
+				     AMDGPU_SDMA_IRQ_INSTANCE0 + i,
+				     AMDGPU_RING_PRIO_DEFAULT, NULL);
+		if (r)
+			return r;
+
+		if (adev->sdma.has_page_queue) {
+			ring = &adev->sdma.instance[i].page;
+			ring->ring_obj = NULL;
+			ring->use_doorbell = true;
+
+			/* doorbell index of page queue is assigned right after
+			 * gfx queue on the same instance
+			 */
+			ring->doorbell_index =
+				(adev->doorbell_index.sdma_engine[i] + 1) << 1;
+			ring->vm_hub = AMDGPU_MMHUB0(aid_id);
+
+			sprintf(ring->name, "page%d.%d", aid_id,
+					i % adev->sdma.num_inst_per_aid);
+			r = amdgpu_ring_init(adev, ring, 1024,
+					     &adev->sdma.trap_irq,
+					     AMDGPU_SDMA_IRQ_INSTANCE0 + i,
+					     AMDGPU_RING_PRIO_DEFAULT, NULL);
+			if (r)
+				return r;
+		}
+	}
+
+	adev->sdma.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->sdma.instance[0].ring);
+
+	if (amdgpu_sdma_ras_sw_init(adev)) {
+		dev_err(adev->dev, "fail to initialize sdma ras block\n");
+		return -EINVAL;
+	}
+
+	/* Allocate memory for SDMA IP Dump buffer */
+	ptr = kcalloc(adev->sdma.num_instances * reg_count, sizeof(uint32_t), GFP_KERNEL);
+	if (ptr)
+		adev->sdma.ip_dump = ptr;
+	else
+		DRM_ERROR("Failed to allocated memory for SDMA IP Dump\n");
+
+	r = amdgpu_sdma_sysfs_reset_mask_init(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int sdma_v4_4_2_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		amdgpu_ring_fini(&adev->sdma.instance[i].ring);
+		if (adev->sdma.has_page_queue)
+			amdgpu_ring_fini(&adev->sdma.instance[i].page);
+	}
+
+	amdgpu_sdma_sysfs_reset_mask_fini(adev);
+	if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) == IP_VERSION(4, 4, 2) ||
+	    amdgpu_ip_version(adev, SDMA0_HWIP, 0) == IP_VERSION(4, 4, 4) ||
+	    amdgpu_ip_version(adev, SDMA0_HWIP, 0) == IP_VERSION(4, 4, 5))
+		amdgpu_sdma_destroy_inst_ctx(adev, true);
+	else
+		amdgpu_sdma_destroy_inst_ctx(adev, false);
+
+	kfree(adev->sdma.ip_dump);
+
+	return 0;
+}
+
+static int sdma_v4_4_2_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+	uint32_t inst_mask;
+
+	inst_mask = GENMASK(adev->sdma.num_instances - 1, 0);
+	if (!amdgpu_sriov_vf(adev))
+		sdma_v4_4_2_inst_init_golden_registers(adev, inst_mask);
+
+	r = sdma_v4_4_2_inst_start(adev, inst_mask, false);
+
+	return r;
+}
+
+static int sdma_v4_4_2_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	uint32_t inst_mask;
+	int i;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	inst_mask = GENMASK(adev->sdma.num_instances - 1, 0);
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__SDMA)) {
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			amdgpu_irq_put(adev, &adev->sdma.ecc_irq,
+				       AMDGPU_SDMA_IRQ_INSTANCE0 + i);
+		}
+	}
+
+	sdma_v4_4_2_inst_ctx_switch_enable(adev, false, inst_mask);
+	sdma_v4_4_2_inst_enable(adev, false, inst_mask);
+
+	return 0;
+}
+
+static int sdma_v4_4_2_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					     enum amd_clockgating_state state);
+
+static int sdma_v4_4_2_suspend(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_in_reset(adev))
+		sdma_v4_4_2_set_clockgating_state(ip_block, AMD_CG_STATE_UNGATE);
+
+	return sdma_v4_4_2_hw_fini(ip_block);
+}
+
+static int sdma_v4_4_2_resume(struct amdgpu_ip_block *ip_block)
+{
+	return sdma_v4_4_2_hw_init(ip_block);
+}
+
+static bool sdma_v4_4_2_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	u32 i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		u32 tmp = RREG32_SDMA(i, regSDMA_STATUS_REG);
+
+		if (!(tmp & SDMA_STATUS_REG__IDLE_MASK))
+			return false;
+	}
+
+	return true;
+}
+
+static int sdma_v4_4_2_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	unsigned i, j;
+	u32 sdma[AMDGPU_MAX_SDMA_INSTANCES];
+	struct amdgpu_device *adev = ip_block->adev;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		for (j = 0; j < adev->sdma.num_instances; j++) {
+			sdma[j] = RREG32_SDMA(j, regSDMA_STATUS_REG);
+			if (!(sdma[j] & SDMA_STATUS_REG__IDLE_MASK))
+				break;
+		}
+		if (j == adev->sdma.num_instances)
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static int sdma_v4_4_2_soft_reset(struct amdgpu_ip_block *ip_block)
+{
+	/* todo */
+
+	return 0;
+}
+
+static bool sdma_v4_4_2_is_queue_selected(struct amdgpu_device *adev, uint32_t instance_id, bool is_page_queue)
+{
+	uint32_t reg_offset = is_page_queue ? regSDMA_PAGE_CONTEXT_STATUS : regSDMA_GFX_CONTEXT_STATUS;
+	uint32_t context_status = RREG32(sdma_v4_4_2_get_reg_offset(adev, instance_id, reg_offset));
+
+	/* Check if the SELECTED bit is set */
+	return (context_status & SDMA_GFX_CONTEXT_STATUS__SELECTED_MASK) != 0;
+}
+
+static int sdma_v4_4_2_reset_queue(struct amdgpu_ring *ring,
+				   unsigned int vmid,
+				   struct amdgpu_fence *timedout_fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 id = ring->me;
+	int r;
+
+	amdgpu_amdkfd_suspend(adev, true);
+	r = amdgpu_sdma_reset_engine(adev, id, false);
+	amdgpu_amdkfd_resume(adev, true);
+	return r;
+}
+
+static int sdma_v4_4_2_stop_queue(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 instance_id = ring->me;
+	u32 inst_mask;
+	uint64_t rptr;
+
+	if (amdgpu_sriov_vf(adev))
+		return -EINVAL;
+
+	/* Check if this queue is the guilty one */
+	adev->sdma.instance[instance_id].gfx_guilty =
+		sdma_v4_4_2_is_queue_selected(adev, instance_id, false);
+	if (adev->sdma.has_page_queue)
+		adev->sdma.instance[instance_id].page_guilty =
+			sdma_v4_4_2_is_queue_selected(adev, instance_id, true);
+
+	/* Cache the rptr before reset, after the reset,
+	* all of the registers will be reset to 0
+	*/
+	rptr = amdgpu_ring_get_rptr(ring);
+	ring->cached_rptr = rptr;
+	/* Cache the rptr for the page queue if it exists */
+	if (adev->sdma.has_page_queue) {
+		struct amdgpu_ring *page_ring = &adev->sdma.instance[instance_id].page;
+		rptr = amdgpu_ring_get_rptr(page_ring);
+		page_ring->cached_rptr = rptr;
+	}
+
+	/* stop queue */
+	inst_mask = 1 << ring->me;
+	sdma_v4_4_2_inst_gfx_stop(adev, inst_mask);
+	if (adev->sdma.has_page_queue)
+		sdma_v4_4_2_inst_page_stop(adev, inst_mask);
+
+	return 0;
+}
+
+static int sdma_v4_4_2_restore_queue(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 inst_mask;
+	int i, r;
+
+	inst_mask = 1 << ring->me;
+	udelay(50);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (!REG_GET_FIELD(RREG32_SDMA(ring->me, regSDMA_F32_CNTL), SDMA_F32_CNTL, HALT))
+			break;
+		udelay(1);
+	}
+
+	if (i == adev->usec_timeout) {
+		dev_err(adev->dev, "timed out waiting for SDMA%d unhalt after reset\n",
+			ring->me);
+		return -ETIMEDOUT;
+	}
+
+	r = sdma_v4_4_2_inst_start(adev, inst_mask, true);
+
+	return r;
+}
+
+static int sdma_v4_4_2_soft_reset_engine(struct amdgpu_device *adev,
+					 u32 instance_id)
+{
+	/* For SDMA 4.x, use the existing DPM interface for backward compatibility
+	 * we need to convert the logical instance ID to physical instance ID before reset.
+	 */
+	return amdgpu_dpm_reset_sdma(adev, 1 << GET_INST(SDMA0, instance_id));
+}
+
+static int sdma_v4_4_2_set_trap_irq_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	u32 sdma_cntl;
+
+	sdma_cntl = RREG32_SDMA(type, regSDMA_CNTL);
+	sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA_CNTL, TRAP_ENABLE,
+		       state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+	WREG32_SDMA(type, regSDMA_CNTL, sdma_cntl);
+
+	return 0;
+}
+
+static int sdma_v4_4_2_process_trap_irq(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	uint32_t instance, i;
+
+	DRM_DEBUG("IH: SDMA trap\n");
+	instance = sdma_v4_4_2_irq_id_to_seq(adev, entry->client_id);
+
+	/* Client id gives the SDMA instance in AID. To know the exact SDMA
+	 * instance, interrupt entry gives the node id which corresponds to the AID instance.
+	 * Match node id with the AID id associated with the SDMA instance. */
+	for (i = instance; i < adev->sdma.num_instances;
+	     i += adev->sdma.num_inst_per_aid) {
+		if (adev->sdma.instance[i].aid_id ==
+		    node_id_to_phys_map[entry->node_id])
+			break;
+	}
+
+	if (i >= adev->sdma.num_instances) {
+		dev_WARN_ONCE(
+			adev->dev, 1,
+			"Couldn't find the right sdma instance in trap handler");
+		return 0;
+	}
+
+	switch (entry->ring_id) {
+	case 0:
+		amdgpu_fence_process(&adev->sdma.instance[i].ring);
+		break;
+	case 1:
+		amdgpu_fence_process(&adev->sdma.instance[i].page);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+#if 0
+static int sdma_v4_4_2_process_ras_data_cb(struct amdgpu_device *adev,
+		void *err_data,
+		struct amdgpu_iv_entry *entry)
+{
+	int instance;
+
+	/* When “Full RAS” is enabled, the per-IP interrupt sources should
+	 * be disabled and the driver should only look for the aggregated
+	 * interrupt via sync flood
+	 */
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__SDMA))
+		goto out;
+
+	instance = sdma_v4_4_2_irq_id_to_seq(adev, entry->client_id);
+	if (instance < 0)
+		goto out;
+
+	amdgpu_sdma_process_ras_data_cb(adev, err_data, entry);
+
+out:
+	return AMDGPU_RAS_SUCCESS;
+}
+#endif
+
+static int sdma_v4_4_2_process_illegal_inst_irq(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      struct amdgpu_iv_entry *entry)
+{
+	int instance;
+
+	DRM_ERROR("Illegal instruction in SDMA command stream\n");
+
+	instance = sdma_v4_4_2_irq_id_to_seq(adev, entry->client_id);
+	if (instance < 0)
+		return 0;
+
+	switch (entry->ring_id) {
+	case 0:
+		drm_sched_fault(&adev->sdma.instance[instance].ring.sched);
+		break;
+	}
+	return 0;
+}
+
+static int sdma_v4_4_2_set_ecc_irq_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	u32 sdma_cntl;
+
+	sdma_cntl = RREG32_SDMA(type, regSDMA_CNTL);
+	sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA_CNTL, DRAM_ECC_INT_ENABLE,
+					state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+	WREG32_SDMA(type, regSDMA_CNTL, sdma_cntl);
+
+	return 0;
+}
+
+static int sdma_v4_4_2_print_iv_entry(struct amdgpu_device *adev,
+					      struct amdgpu_iv_entry *entry)
+{
+	int instance;
+	struct amdgpu_task_info *task_info;
+	u64 addr;
+
+	instance = sdma_v4_4_2_irq_id_to_seq(adev, entry->client_id);
+	if (instance < 0 || instance >= adev->sdma.num_instances) {
+		dev_err(adev->dev, "sdma instance invalid %d\n", instance);
+		return -EINVAL;
+	}
+
+	addr = (u64)entry->src_data[0] << 12;
+	addr |= ((u64)entry->src_data[1] & 0xf) << 44;
+
+	dev_dbg_ratelimited(adev->dev,
+			    "[sdma%d] address:0x%016llx src_id:%u ring:%u vmid:%u pasid:%u\n",
+			    instance, addr, entry->src_id, entry->ring_id, entry->vmid,
+			    entry->pasid);
+
+	task_info = amdgpu_vm_get_task_info_pasid(adev, entry->pasid);
+	if (task_info) {
+		dev_dbg_ratelimited(adev->dev, " for process %s pid %d thread %s pid %d\n",
+				    task_info->process_name, task_info->tgid,
+				    task_info->task.comm, task_info->task.pid);
+		amdgpu_vm_put_task_info(task_info);
+	}
+
+	return 0;
+}
+
+static int sdma_v4_4_2_process_vm_hole_irq(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      struct amdgpu_iv_entry *entry)
+{
+	dev_dbg_ratelimited(adev->dev, "MC or SEM address in VM hole\n");
+	sdma_v4_4_2_print_iv_entry(adev, entry);
+	return 0;
+}
+
+static int sdma_v4_4_2_process_doorbell_invalid_irq(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      struct amdgpu_iv_entry *entry)
+{
+
+	dev_dbg_ratelimited(adev->dev, "SDMA received a doorbell from BIF with byte_enable !=0xff\n");
+	sdma_v4_4_2_print_iv_entry(adev, entry);
+	return 0;
+}
+
+static int sdma_v4_4_2_process_pool_timeout_irq(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      struct amdgpu_iv_entry *entry)
+{
+	dev_dbg_ratelimited(adev->dev,
+		"Polling register/memory timeout executing POLL_REG/MEM with finite timer\n");
+	sdma_v4_4_2_print_iv_entry(adev, entry);
+	return 0;
+}
+
+static int sdma_v4_4_2_process_srbm_write_irq(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      struct amdgpu_iv_entry *entry)
+{
+	dev_dbg_ratelimited(adev->dev,
+		"SDMA gets an Register Write SRBM_WRITE command in non-privilege command buffer\n");
+	sdma_v4_4_2_print_iv_entry(adev, entry);
+	return 0;
+}
+
+static int sdma_v4_4_2_process_ctxt_empty_irq(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      struct amdgpu_iv_entry *entry)
+{
+	/* There is nothing useful to be done here, only kept for debug */
+	dev_dbg_ratelimited(adev->dev, "SDMA context empty interrupt");
+	sdma_v4_4_2_print_iv_entry(adev, entry);
+	return 0;
+}
+
+static void sdma_v4_4_2_inst_update_medium_grain_light_sleep(
+	struct amdgpu_device *adev, bool enable, uint32_t inst_mask)
+{
+	uint32_t data, def;
+	int i;
+
+	/* leave as default if it is not driver controlled */
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_SDMA_LS))
+		return;
+
+	if (enable) {
+		for_each_inst(i, inst_mask) {
+			/* 1-not override: enable sdma mem light sleep */
+			def = data = RREG32_SDMA(i, regSDMA_POWER_CNTL);
+			data |= SDMA_POWER_CNTL__MEM_POWER_OVERRIDE_MASK;
+			if (def != data)
+				WREG32_SDMA(i, regSDMA_POWER_CNTL, data);
+		}
+	} else {
+		for_each_inst(i, inst_mask) {
+			/* 0-override:disable sdma mem light sleep */
+			def = data = RREG32_SDMA(i, regSDMA_POWER_CNTL);
+			data &= ~SDMA_POWER_CNTL__MEM_POWER_OVERRIDE_MASK;
+			if (def != data)
+				WREG32_SDMA(i, regSDMA_POWER_CNTL, data);
+		}
+	}
+}
+
+static void sdma_v4_4_2_inst_update_medium_grain_clock_gating(
+	struct amdgpu_device *adev, bool enable, uint32_t inst_mask)
+{
+	uint32_t data, def;
+	int i;
+
+	/* leave as default if it is not driver controlled */
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_SDMA_MGCG))
+		return;
+
+	if (enable) {
+		for_each_inst(i, inst_mask) {
+			def = data = RREG32_SDMA(i, regSDMA_CLK_CTRL);
+			data &= ~(SDMA_CLK_CTRL__SOFT_OVERRIDE5_MASK |
+				  SDMA_CLK_CTRL__SOFT_OVERRIDE4_MASK |
+				  SDMA_CLK_CTRL__SOFT_OVERRIDE3_MASK |
+				  SDMA_CLK_CTRL__SOFT_OVERRIDE2_MASK |
+				  SDMA_CLK_CTRL__SOFT_OVERRIDE1_MASK |
+				  SDMA_CLK_CTRL__SOFT_OVERRIDE0_MASK);
+			if (def != data)
+				WREG32_SDMA(i, regSDMA_CLK_CTRL, data);
+		}
+	} else {
+		for_each_inst(i, inst_mask) {
+			def = data = RREG32_SDMA(i, regSDMA_CLK_CTRL);
+			data |= (SDMA_CLK_CTRL__SOFT_OVERRIDE5_MASK |
+				 SDMA_CLK_CTRL__SOFT_OVERRIDE4_MASK |
+				 SDMA_CLK_CTRL__SOFT_OVERRIDE3_MASK |
+				 SDMA_CLK_CTRL__SOFT_OVERRIDE2_MASK |
+				 SDMA_CLK_CTRL__SOFT_OVERRIDE1_MASK |
+				 SDMA_CLK_CTRL__SOFT_OVERRIDE0_MASK);
+			if (def != data)
+				WREG32_SDMA(i, regSDMA_CLK_CTRL, data);
+		}
+	}
+}
+
+static int sdma_v4_4_2_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	uint32_t inst_mask;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	inst_mask = GENMASK(adev->sdma.num_instances - 1, 0);
+
+	sdma_v4_4_2_inst_update_medium_grain_clock_gating(
+		adev, state == AMD_CG_STATE_GATE, inst_mask);
+	sdma_v4_4_2_inst_update_medium_grain_light_sleep(
+		adev, state == AMD_CG_STATE_GATE, inst_mask);
+	return 0;
+}
+
+static int sdma_v4_4_2_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static void sdma_v4_4_2_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int data;
+
+	if (amdgpu_sriov_vf(adev))
+		*flags = 0;
+
+	/* AMD_CG_SUPPORT_SDMA_MGCG */
+	data = RREG32(SOC15_REG_OFFSET(SDMA0, GET_INST(SDMA0, 0), regSDMA_CLK_CTRL));
+	if (!(data & SDMA_CLK_CTRL__SOFT_OVERRIDE5_MASK))
+		*flags |= AMD_CG_SUPPORT_SDMA_MGCG;
+
+	/* AMD_CG_SUPPORT_SDMA_LS */
+	data = RREG32(SOC15_REG_OFFSET(SDMA0, GET_INST(SDMA0, 0), regSDMA_POWER_CNTL));
+	if (data & SDMA_POWER_CNTL__MEM_POWER_OVERRIDE_MASK)
+		*flags |= AMD_CG_SUPPORT_SDMA_LS;
+}
+
+static void sdma_v4_4_2_print_ip_state(struct amdgpu_ip_block *ip_block, struct drm_printer *p)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, j;
+	uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_4_4_2);
+	uint32_t instance_offset;
+
+	if (!adev->sdma.ip_dump)
+		return;
+
+	drm_printf(p, "num_instances:%d\n", adev->sdma.num_instances);
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		instance_offset = i * reg_count;
+		drm_printf(p, "\nInstance:%d\n", i);
+
+		for (j = 0; j < reg_count; j++)
+			drm_printf(p, "%-50s \t 0x%08x\n", sdma_reg_list_4_4_2[j].reg_name,
+				   adev->sdma.ip_dump[instance_offset + j]);
+	}
+}
+
+static void sdma_v4_4_2_dump_ip_state(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, j;
+	uint32_t instance_offset;
+	uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_4_4_2);
+
+	if (!adev->sdma.ip_dump)
+		return;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		instance_offset = i * reg_count;
+		for (j = 0; j < reg_count; j++)
+			adev->sdma.ip_dump[instance_offset + j] =
+				RREG32(sdma_v4_4_2_get_reg_offset(adev, i,
+				       sdma_reg_list_4_4_2[j].reg_offset));
+	}
+}
+
+const struct amd_ip_funcs sdma_v4_4_2_ip_funcs = {
+	.name = "sdma_v4_4_2",
+	.early_init = sdma_v4_4_2_early_init,
+	.late_init = sdma_v4_4_2_late_init,
+	.sw_init = sdma_v4_4_2_sw_init,
+	.sw_fini = sdma_v4_4_2_sw_fini,
+	.hw_init = sdma_v4_4_2_hw_init,
+	.hw_fini = sdma_v4_4_2_hw_fini,
+	.suspend = sdma_v4_4_2_suspend,
+	.resume = sdma_v4_4_2_resume,
+	.is_idle = sdma_v4_4_2_is_idle,
+	.wait_for_idle = sdma_v4_4_2_wait_for_idle,
+	.soft_reset = sdma_v4_4_2_soft_reset,
+	.set_clockgating_state = sdma_v4_4_2_set_clockgating_state,
+	.set_powergating_state = sdma_v4_4_2_set_powergating_state,
+	.get_clockgating_state = sdma_v4_4_2_get_clockgating_state,
+	.dump_ip_state = sdma_v4_4_2_dump_ip_state,
+	.print_ip_state = sdma_v4_4_2_print_ip_state,
+};
+
+static const struct amdgpu_ring_funcs sdma_v4_4_2_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_SDMA,
+	.align_mask = 0xff,
+	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
+	.support_64bit_ptrs = true,
+	.get_rptr = sdma_v4_4_2_ring_get_rptr,
+	.get_wptr = sdma_v4_4_2_ring_get_wptr,
+	.set_wptr = sdma_v4_4_2_ring_set_wptr,
+	.emit_frame_size =
+		6 + /* sdma_v4_4_2_ring_emit_hdp_flush */
+		3 + /* hdp invalidate */
+		6 + /* sdma_v4_4_2_ring_emit_pipeline_sync */
+		/* sdma_v4_4_2_ring_emit_vm_flush */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 6 +
+		10 + 10 + 10, /* sdma_v4_4_2_ring_emit_fence x3 for user fence, vm fence */
+	.emit_ib_size = 7 + 6, /* sdma_v4_4_2_ring_emit_ib */
+	.emit_ib = sdma_v4_4_2_ring_emit_ib,
+	.emit_fence = sdma_v4_4_2_ring_emit_fence,
+	.emit_pipeline_sync = sdma_v4_4_2_ring_emit_pipeline_sync,
+	.emit_vm_flush = sdma_v4_4_2_ring_emit_vm_flush,
+	.emit_hdp_flush = sdma_v4_4_2_ring_emit_hdp_flush,
+	.test_ring = sdma_v4_4_2_ring_test_ring,
+	.test_ib = sdma_v4_4_2_ring_test_ib,
+	.insert_nop = sdma_v4_4_2_ring_insert_nop,
+	.pad_ib = sdma_v4_4_2_ring_pad_ib,
+	.emit_wreg = sdma_v4_4_2_ring_emit_wreg,
+	.emit_reg_wait = sdma_v4_4_2_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = sdma_v4_4_2_reset_queue,
+};
+
+static const struct amdgpu_ring_funcs sdma_v4_4_2_page_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_SDMA,
+	.align_mask = 0xff,
+	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
+	.support_64bit_ptrs = true,
+	.get_rptr = sdma_v4_4_2_ring_get_rptr,
+	.get_wptr = sdma_v4_4_2_page_ring_get_wptr,
+	.set_wptr = sdma_v4_4_2_page_ring_set_wptr,
+	.emit_frame_size =
+		6 + /* sdma_v4_4_2_ring_emit_hdp_flush */
+		3 + /* hdp invalidate */
+		6 + /* sdma_v4_4_2_ring_emit_pipeline_sync */
+		/* sdma_v4_4_2_ring_emit_vm_flush */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 6 +
+		10 + 10 + 10, /* sdma_v4_4_2_ring_emit_fence x3 for user fence, vm fence */
+	.emit_ib_size = 7 + 6, /* sdma_v4_4_2_ring_emit_ib */
+	.emit_ib = sdma_v4_4_2_ring_emit_ib,
+	.emit_fence = sdma_v4_4_2_ring_emit_fence,
+	.emit_pipeline_sync = sdma_v4_4_2_ring_emit_pipeline_sync,
+	.emit_vm_flush = sdma_v4_4_2_ring_emit_vm_flush,
+	.emit_hdp_flush = sdma_v4_4_2_ring_emit_hdp_flush,
+	.test_ring = sdma_v4_4_2_ring_test_ring,
+	.test_ib = sdma_v4_4_2_ring_test_ib,
+	.insert_nop = sdma_v4_4_2_ring_insert_nop,
+	.pad_ib = sdma_v4_4_2_ring_pad_ib,
+	.emit_wreg = sdma_v4_4_2_ring_emit_wreg,
+	.emit_reg_wait = sdma_v4_4_2_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = sdma_v4_4_2_reset_queue,
+};
+
+static void sdma_v4_4_2_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i, dev_inst;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		adev->sdma.instance[i].ring.funcs = &sdma_v4_4_2_ring_funcs;
+		adev->sdma.instance[i].ring.me = i;
+		if (adev->sdma.has_page_queue) {
+			adev->sdma.instance[i].page.funcs =
+				&sdma_v4_4_2_page_ring_funcs;
+			adev->sdma.instance[i].page.me = i;
+		}
+
+		dev_inst = GET_INST(SDMA0, i);
+		/* AID to which SDMA belongs depends on physical instance */
+		adev->sdma.instance[i].aid_id =
+			dev_inst / adev->sdma.num_inst_per_aid;
+	}
+}
+
+static const struct amdgpu_irq_src_funcs sdma_v4_4_2_trap_irq_funcs = {
+	.set = sdma_v4_4_2_set_trap_irq_state,
+	.process = sdma_v4_4_2_process_trap_irq,
+};
+
+static const struct amdgpu_irq_src_funcs sdma_v4_4_2_illegal_inst_irq_funcs = {
+	.process = sdma_v4_4_2_process_illegal_inst_irq,
+};
+
+static const struct amdgpu_irq_src_funcs sdma_v4_4_2_ecc_irq_funcs = {
+	.set = sdma_v4_4_2_set_ecc_irq_state,
+	.process = amdgpu_sdma_process_ecc_irq,
+};
+
+static const struct amdgpu_irq_src_funcs sdma_v4_4_2_vm_hole_irq_funcs = {
+	.process = sdma_v4_4_2_process_vm_hole_irq,
+};
+
+static const struct amdgpu_irq_src_funcs sdma_v4_4_2_doorbell_invalid_irq_funcs = {
+	.process = sdma_v4_4_2_process_doorbell_invalid_irq,
+};
+
+static const struct amdgpu_irq_src_funcs sdma_v4_4_2_pool_timeout_irq_funcs = {
+	.process = sdma_v4_4_2_process_pool_timeout_irq,
+};
+
+static const struct amdgpu_irq_src_funcs sdma_v4_4_2_srbm_write_irq_funcs = {
+	.process = sdma_v4_4_2_process_srbm_write_irq,
+};
+
+static const struct amdgpu_irq_src_funcs sdma_v4_4_2_ctxt_empty_irq_funcs = {
+	.process = sdma_v4_4_2_process_ctxt_empty_irq,
+};
+
+static void sdma_v4_4_2_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->sdma.trap_irq.num_types = adev->sdma.num_instances;
+	adev->sdma.ecc_irq.num_types = adev->sdma.num_instances;
+	adev->sdma.vm_hole_irq.num_types = adev->sdma.num_instances;
+	adev->sdma.doorbell_invalid_irq.num_types = adev->sdma.num_instances;
+	adev->sdma.pool_timeout_irq.num_types = adev->sdma.num_instances;
+	adev->sdma.srbm_write_irq.num_types = adev->sdma.num_instances;
+	adev->sdma.ctxt_empty_irq.num_types = adev->sdma.num_instances;
+
+	adev->sdma.trap_irq.funcs = &sdma_v4_4_2_trap_irq_funcs;
+	adev->sdma.illegal_inst_irq.funcs = &sdma_v4_4_2_illegal_inst_irq_funcs;
+	adev->sdma.ecc_irq.funcs = &sdma_v4_4_2_ecc_irq_funcs;
+	adev->sdma.vm_hole_irq.funcs = &sdma_v4_4_2_vm_hole_irq_funcs;
+	adev->sdma.doorbell_invalid_irq.funcs = &sdma_v4_4_2_doorbell_invalid_irq_funcs;
+	adev->sdma.pool_timeout_irq.funcs = &sdma_v4_4_2_pool_timeout_irq_funcs;
+	adev->sdma.srbm_write_irq.funcs = &sdma_v4_4_2_srbm_write_irq_funcs;
+	adev->sdma.ctxt_empty_irq.funcs = &sdma_v4_4_2_ctxt_empty_irq_funcs;
+}
+
+/**
+ * sdma_v4_4_2_emit_copy_buffer - copy buffer using the sDMA engine
+ *
+ * @ib: indirect buffer to copy to
+ * @src_offset: src GPU address
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ * @copy_flags: copy flags for the buffers
+ *
+ * Copy GPU buffers using the DMA engine.
+ * Used by the amdgpu ttm implementation to move pages if
+ * registered as the asic copy callback.
+ */
+static void sdma_v4_4_2_emit_copy_buffer(struct amdgpu_ib *ib,
+				       uint64_t src_offset,
+				       uint64_t dst_offset,
+				       uint32_t byte_count,
+				       uint32_t copy_flags)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR) |
+		SDMA_PKT_COPY_LINEAR_HEADER_TMZ((copy_flags & AMDGPU_COPY_FLAGS_TMZ) ? 1 : 0);
+	ib->ptr[ib->length_dw++] = byte_count - 1;
+	ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+	ib->ptr[ib->length_dw++] = lower_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+}
+
+/**
+ * sdma_v4_4_2_emit_fill_buffer - fill buffer using the sDMA engine
+ *
+ * @ib: indirect buffer to copy to
+ * @src_data: value to write to buffer
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ *
+ * Fill GPU buffers using the DMA engine.
+ */
+static void sdma_v4_4_2_emit_fill_buffer(struct amdgpu_ib *ib,
+				       uint32_t src_data,
+				       uint64_t dst_offset,
+				       uint32_t byte_count)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_CONST_FILL);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = src_data;
+	ib->ptr[ib->length_dw++] = byte_count - 1;
+}
+
+static const struct amdgpu_buffer_funcs sdma_v4_4_2_buffer_funcs = {
+	.copy_max_bytes = 0x400000,
+	.copy_num_dw = 7,
+	.emit_copy_buffer = sdma_v4_4_2_emit_copy_buffer,
+
+	.fill_max_bytes = 0x400000,
+	.fill_num_dw = 5,
+	.emit_fill_buffer = sdma_v4_4_2_emit_fill_buffer,
+};
+
+static void sdma_v4_4_2_set_buffer_funcs(struct amdgpu_device *adev)
+{
+	adev->mman.buffer_funcs = &sdma_v4_4_2_buffer_funcs;
+	if (adev->sdma.has_page_queue)
+		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].page;
+	else
+		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
+}
+
+static const struct amdgpu_vm_pte_funcs sdma_v4_4_2_vm_pte_funcs = {
+	.copy_pte_num_dw = 7,
+	.copy_pte = sdma_v4_4_2_vm_copy_pte,
+
+	.write_pte = sdma_v4_4_2_vm_write_pte,
+	.set_pte_pde = sdma_v4_4_2_vm_set_pte_pde,
+};
+
+static void sdma_v4_4_2_set_vm_pte_funcs(struct amdgpu_device *adev)
+{
+	struct drm_gpu_scheduler *sched;
+	unsigned i;
+
+	adev->vm_manager.vm_pte_funcs = &sdma_v4_4_2_vm_pte_funcs;
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (adev->sdma.has_page_queue)
+			sched = &adev->sdma.instance[i].page.sched;
+		else
+			sched = &adev->sdma.instance[i].ring.sched;
+		adev->vm_manager.vm_pte_scheds[i] = sched;
+	}
+	adev->vm_manager.vm_pte_num_scheds = adev->sdma.num_instances;
+}
+
+/**
+ * sdma_v4_4_2_update_reset_mask - update  reset mask for SDMA
+ * @adev: Pointer to the AMDGPU device structure
+ *
+ * This function update reset mask for SDMA and sets the supported
+ * reset types based on the IP version and firmware versions.
+ *
+ */
+static void sdma_v4_4_2_update_reset_mask(struct amdgpu_device *adev)
+{
+	/* per queue reset not supported for SRIOV */
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/*
+	 * the user queue relies on MEC fw and pmfw when the sdma queue do reset.
+	 * it needs to check both of them at here to skip old mec and pmfw.
+	 */
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 4, 3):
+	case IP_VERSION(9, 4, 4):
+		if ((adev->gfx.mec_fw_version >= 0xb0) && amdgpu_dpm_reset_sdma_is_supported(adev))
+			adev->sdma.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+		break;
+	case IP_VERSION(9, 5, 0):
+		if ((adev->gfx.mec_fw_version >= 0xf) && amdgpu_dpm_reset_sdma_is_supported(adev))
+			adev->sdma.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+		break;
+	default:
+		break;
+	}
+
+}
+
+const struct amdgpu_ip_block_version sdma_v4_4_2_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_SDMA,
+	.major = 4,
+	.minor = 4,
+	.rev = 2,
+	.funcs = &sdma_v4_4_2_ip_funcs,
+};
+
+static int sdma_v4_4_2_xcp_resume(void *handle, uint32_t inst_mask)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int r;
+
+	if (!amdgpu_sriov_vf(adev))
+		sdma_v4_4_2_inst_init_golden_registers(adev, inst_mask);
+
+	r = sdma_v4_4_2_inst_start(adev, inst_mask, false);
+
+	return r;
+}
+
+static int sdma_v4_4_2_xcp_suspend(void *handle, uint32_t inst_mask)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	uint32_t tmp_mask = inst_mask;
+	int i;
+
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__SDMA)) {
+		for_each_inst(i, tmp_mask) {
+			amdgpu_irq_put(adev, &adev->sdma.ecc_irq,
+				       AMDGPU_SDMA_IRQ_INSTANCE0 + i);
+		}
+	}
+
+	sdma_v4_4_2_inst_ctx_switch_enable(adev, false, inst_mask);
+	sdma_v4_4_2_inst_enable(adev, false, inst_mask);
+
+	return 0;
+}
+
+struct amdgpu_xcp_ip_funcs sdma_v4_4_2_xcp_funcs = {
+	.suspend = &sdma_v4_4_2_xcp_suspend,
+	.resume = &sdma_v4_4_2_xcp_resume
+};
+
+static const struct amdgpu_ras_err_status_reg_entry sdma_v4_2_2_ue_reg_list[] = {
+	{AMDGPU_RAS_REG_ENTRY(SDMA0, 0, regSDMA_UE_ERR_STATUS_LO, regSDMA_UE_ERR_STATUS_HI),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "SDMA"},
+};
+
+static const struct amdgpu_ras_memory_id_entry sdma_v4_4_2_ras_memory_list[] = {
+	{AMDGPU_SDMA_MBANK_DATA_BUF0, "SDMA_MBANK_DATA_BUF0"},
+	{AMDGPU_SDMA_MBANK_DATA_BUF1, "SDMA_MBANK_DATA_BUF1"},
+	{AMDGPU_SDMA_MBANK_DATA_BUF2, "SDMA_MBANK_DATA_BUF2"},
+	{AMDGPU_SDMA_MBANK_DATA_BUF3, "SDMA_MBANK_DATA_BUF3"},
+	{AMDGPU_SDMA_MBANK_DATA_BUF4, "SDMA_MBANK_DATA_BUF4"},
+	{AMDGPU_SDMA_MBANK_DATA_BUF5, "SDMA_MBANK_DATA_BUF5"},
+	{AMDGPU_SDMA_MBANK_DATA_BUF6, "SDMA_MBANK_DATA_BUF6"},
+	{AMDGPU_SDMA_MBANK_DATA_BUF7, "SDMA_MBANK_DATA_BUF7"},
+	{AMDGPU_SDMA_MBANK_DATA_BUF8, "SDMA_MBANK_DATA_BUF8"},
+	{AMDGPU_SDMA_MBANK_DATA_BUF9, "SDMA_MBANK_DATA_BUF9"},
+	{AMDGPU_SDMA_MBANK_DATA_BUF10, "SDMA_MBANK_DATA_BUF10"},
+	{AMDGPU_SDMA_MBANK_DATA_BUF11, "SDMA_MBANK_DATA_BUF11"},
+	{AMDGPU_SDMA_MBANK_DATA_BUF12, "SDMA_MBANK_DATA_BUF12"},
+	{AMDGPU_SDMA_MBANK_DATA_BUF13, "SDMA_MBANK_DATA_BUF13"},
+	{AMDGPU_SDMA_MBANK_DATA_BUF14, "SDMA_MBANK_DATA_BUF14"},
+	{AMDGPU_SDMA_MBANK_DATA_BUF15, "SDMA_MBANK_DATA_BUF15"},
+	{AMDGPU_SDMA_UCODE_BUF, "SDMA_UCODE_BUF"},
+	{AMDGPU_SDMA_RB_CMD_BUF, "SDMA_RB_CMD_BUF"},
+	{AMDGPU_SDMA_IB_CMD_BUF, "SDMA_IB_CMD_BUF"},
+	{AMDGPU_SDMA_UTCL1_RD_FIFO, "SDMA_UTCL1_RD_FIFO"},
+	{AMDGPU_SDMA_UTCL1_RDBST_FIFO, "SDMA_UTCL1_RDBST_FIFO"},
+	{AMDGPU_SDMA_UTCL1_WR_FIFO, "SDMA_UTCL1_WR_FIFO"},
+	{AMDGPU_SDMA_DATA_LUT_FIFO, "SDMA_DATA_LUT_FIFO"},
+	{AMDGPU_SDMA_SPLIT_DAT_BUF, "SDMA_SPLIT_DAT_BUF"},
+};
+
+static void sdma_v4_4_2_inst_query_ras_error_count(struct amdgpu_device *adev,
+						   uint32_t sdma_inst,
+						   void *ras_err_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_err_status;
+	uint32_t sdma_dev_inst = GET_INST(SDMA0, sdma_inst);
+	unsigned long ue_count = 0;
+	struct amdgpu_smuio_mcm_config_info mcm_info = {
+		.socket_id = adev->smuio.funcs->get_socket_id(adev),
+		.die_id = adev->sdma.instance[sdma_inst].aid_id,
+	};
+
+	/* sdma v4_4_2 doesn't support query ce counts */
+	amdgpu_ras_inst_query_ras_error_count(adev,
+					sdma_v4_2_2_ue_reg_list,
+					ARRAY_SIZE(sdma_v4_2_2_ue_reg_list),
+					sdma_v4_4_2_ras_memory_list,
+					ARRAY_SIZE(sdma_v4_4_2_ras_memory_list),
+					sdma_dev_inst,
+					AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE,
+					&ue_count);
+
+	amdgpu_ras_error_statistic_ue_count(err_data, &mcm_info, ue_count);
+}
+
+static void sdma_v4_4_2_query_ras_error_count(struct amdgpu_device *adev,
+					      void *ras_err_status)
+{
+	uint32_t inst_mask;
+	int i = 0;
+
+	inst_mask = GENMASK(adev->sdma.num_instances - 1, 0);
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__SDMA)) {
+		for_each_inst(i, inst_mask)
+			sdma_v4_4_2_inst_query_ras_error_count(adev, i, ras_err_status);
+	} else {
+		dev_warn(adev->dev, "SDMA RAS is not supported\n");
+	}
+}
+
+static void sdma_v4_4_2_inst_reset_ras_error_count(struct amdgpu_device *adev,
+						   uint32_t sdma_inst)
+{
+	uint32_t sdma_dev_inst = GET_INST(SDMA0, sdma_inst);
+
+	amdgpu_ras_inst_reset_ras_error_count(adev,
+					sdma_v4_2_2_ue_reg_list,
+					ARRAY_SIZE(sdma_v4_2_2_ue_reg_list),
+					sdma_dev_inst);
+}
+
+static void sdma_v4_4_2_reset_ras_error_count(struct amdgpu_device *adev)
+{
+	uint32_t inst_mask;
+	int i = 0;
+
+	inst_mask = GENMASK(adev->sdma.num_instances - 1, 0);
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__SDMA)) {
+		for_each_inst(i, inst_mask)
+			sdma_v4_4_2_inst_reset_ras_error_count(adev, i);
+	} else {
+		dev_warn(adev->dev, "SDMA RAS is not supported\n");
+	}
+}
+
+static const struct amdgpu_ras_block_hw_ops sdma_v4_4_2_ras_hw_ops = {
+	.query_ras_error_count = sdma_v4_4_2_query_ras_error_count,
+	.reset_ras_error_count = sdma_v4_4_2_reset_ras_error_count,
+};
+
+static int sdma_v4_4_2_aca_bank_parser(struct aca_handle *handle, struct aca_bank *bank,
+				       enum aca_smu_type type, void *data)
+{
+	struct aca_bank_info info;
+	u64 misc0;
+	int ret;
+
+	ret = aca_bank_info_decode(bank, &info);
+	if (ret)
+		return ret;
+
+	misc0 = bank->regs[ACA_REG_IDX_MISC0];
+	switch (type) {
+	case ACA_SMU_TYPE_UE:
+		bank->aca_err_type = ACA_ERROR_TYPE_UE;
+		ret = aca_error_cache_log_bank_error(handle, &info, ACA_ERROR_TYPE_UE,
+						     1ULL);
+		break;
+	case ACA_SMU_TYPE_CE:
+		bank->aca_err_type = ACA_ERROR_TYPE_CE;
+		ret = aca_error_cache_log_bank_error(handle, &info, bank->aca_err_type,
+						     ACA_REG__MISC0__ERRCNT(misc0));
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return ret;
+}
+
+/* CODE_SDMA0 - CODE_SDMA4, reference to smu driver if header file */
+static int sdma_v4_4_2_err_codes[] = { 33, 34, 35, 36 };
+
+static bool sdma_v4_4_2_aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank,
+					  enum aca_smu_type type, void *data)
+{
+	u32 instlo;
+
+	instlo = ACA_REG__IPID__INSTANCEIDLO(bank->regs[ACA_REG_IDX_IPID]);
+	instlo &= GENMASK(31, 1);
+
+	if (instlo != mmSMNAID_AID0_MCA_SMU)
+		return false;
+
+	if (aca_bank_check_error_codes(handle->adev, bank,
+				       sdma_v4_4_2_err_codes,
+				       ARRAY_SIZE(sdma_v4_4_2_err_codes)))
+		return false;
+
+	return true;
+}
+
+static const struct aca_bank_ops sdma_v4_4_2_aca_bank_ops = {
+	.aca_bank_parser = sdma_v4_4_2_aca_bank_parser,
+	.aca_bank_is_valid = sdma_v4_4_2_aca_bank_is_valid,
+};
+
+static const struct aca_info sdma_v4_4_2_aca_info = {
+	.hwip = ACA_HWIP_TYPE_SMU,
+	.mask = ACA_ERROR_UE_MASK,
+	.bank_ops = &sdma_v4_4_2_aca_bank_ops,
+};
+
+static int sdma_v4_4_2_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
+{
+	int r;
+
+	r = amdgpu_sdma_ras_late_init(adev, ras_block);
+	if (r)
+		return r;
+
+	return amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__SDMA,
+				   &sdma_v4_4_2_aca_info, NULL);
+}
+
+static struct amdgpu_sdma_ras sdma_v4_4_2_ras = {
+	.ras_block = {
+		.hw_ops = &sdma_v4_4_2_ras_hw_ops,
+		.ras_late_init = sdma_v4_4_2_ras_late_init,
+	},
+};
+
+static void sdma_v4_4_2_set_ras_funcs(struct amdgpu_device *adev)
+{
+	adev->sdma.ras = &sdma_v4_4_2_ras;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.h b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.h
new file mode 100644
index 000000000000..d516145529bb
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __SDMA_V4_4_2_H__
+#define __SDMA_V4_4_2_H__
+
+extern const struct amd_ip_funcs sdma_v4_4_2_ip_funcs;
+extern const struct amdgpu_ip_block_version sdma_v4_4_2_ip_block;
+
+extern struct amdgpu_xcp_ip_funcs sdma_v4_4_2_xcp_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c
new file mode 100644
index 000000000000..7dc67a22a7a0
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c
@@ -0,0 +1,2102 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/delay.h>
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+
+#include "amdgpu.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_trace.h"
+
+#include "gc/gc_10_1_0_offset.h"
+#include "gc/gc_10_1_0_sh_mask.h"
+#include "ivsrcid/sdma0/irqsrcs_sdma0_5_0.h"
+#include "ivsrcid/sdma1/irqsrcs_sdma1_5_0.h"
+
+#include "soc15_common.h"
+#include "soc15.h"
+#include "navi10_sdma_pkt_open.h"
+#include "nbio_v2_3.h"
+#include "sdma_common.h"
+#include "sdma_v5_0.h"
+
+MODULE_FIRMWARE("amdgpu/navi10_sdma.bin");
+MODULE_FIRMWARE("amdgpu/navi10_sdma1.bin");
+
+MODULE_FIRMWARE("amdgpu/navi14_sdma.bin");
+MODULE_FIRMWARE("amdgpu/navi14_sdma1.bin");
+
+MODULE_FIRMWARE("amdgpu/navi12_sdma.bin");
+MODULE_FIRMWARE("amdgpu/navi12_sdma1.bin");
+
+MODULE_FIRMWARE("amdgpu/cyan_skillfish2_sdma.bin");
+MODULE_FIRMWARE("amdgpu/cyan_skillfish2_sdma1.bin");
+
+#define SDMA1_REG_OFFSET 0x600
+#define SDMA0_HYP_DEC_REG_START 0x5880
+#define SDMA0_HYP_DEC_REG_END 0x5893
+#define SDMA1_HYP_DEC_REG_OFFSET 0x20
+
+static const struct amdgpu_hwip_reg_entry sdma_reg_list_5_0[] = {
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_STATUS_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_STATUS1_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_STATUS2_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_STATUS3_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_UCODE_CHECKSUM),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RB_RPTR_FETCH_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RB_RPTR_FETCH),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_UTCL1_RD_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_UTCL1_WR_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_UTCL1_RD_XNACK0),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_UTCL1_RD_XNACK1),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_UTCL1_WR_XNACK0),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_UTCL1_WR_XNACK1),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_RB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_RB_RPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_RB_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_RB_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_IB_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_IB_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_IB_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_IB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_IB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_IB_SUB_REMAIN),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_DUMMY_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_RB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_RB_RPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_RB_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_RB_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_IB_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_IB_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_IB_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_DUMMY_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_RB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_RB_RPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_RB_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_RB_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_IB_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_IB_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_IB_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_DUMMY_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_INT_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_VM_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS2)
+};
+
+static void sdma_v5_0_set_ring_funcs(struct amdgpu_device *adev);
+static void sdma_v5_0_set_buffer_funcs(struct amdgpu_device *adev);
+static void sdma_v5_0_set_vm_pte_funcs(struct amdgpu_device *adev);
+static void sdma_v5_0_set_irq_funcs(struct amdgpu_device *adev);
+static int sdma_v5_0_stop_queue(struct amdgpu_ring *ring);
+static int sdma_v5_0_restore_queue(struct amdgpu_ring *ring);
+
+static const struct soc15_reg_golden golden_settings_sdma_5[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_CHICKEN_BITS, 0xffbf1f0f, 0x03ab0107),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_GFX_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_PAGE_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC0_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC1_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC2_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC3_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC4_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC5_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC6_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC7_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_UTCL1_PAGE, 0x00ffffff, 0x000c5c00),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_CHICKEN_BITS, 0xffbf1f0f, 0x03ab0107),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_GFX_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_PAGE_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC0_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC1_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC2_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC3_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC4_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC5_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC6_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC7_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_UTCL1_PAGE, 0x00ffffff, 0x000c5c00)
+};
+
+static const struct soc15_reg_golden golden_settings_sdma_5_sriov[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_GFX_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_PAGE_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC0_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC1_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC2_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC3_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC4_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC5_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC6_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC7_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_GFX_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_PAGE_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC0_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC1_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC2_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC3_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC4_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC5_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC6_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC7_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+};
+
+static const struct soc15_reg_golden golden_settings_sdma_nv10[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC3_RB_WPTR_POLL_CNTL, 0x0000fff0, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC3_RB_WPTR_POLL_CNTL, 0x0000fff0, 0x00403000),
+};
+
+static const struct soc15_reg_golden golden_settings_sdma_nv14[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC3_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC3_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+};
+
+static const struct soc15_reg_golden golden_settings_sdma_nv12[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC3_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_GB_ADDR_CONFIG, 0x001877ff, 0x00000044),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_GB_ADDR_CONFIG_READ, 0x001877ff, 0x00000044),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_GB_ADDR_CONFIG, 0x001877ff, 0x00000044),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_GB_ADDR_CONFIG_READ, 0x001877ff, 0x00000044),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC3_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+};
+
+static const struct soc15_reg_golden golden_settings_sdma_cyan_skillfish[] = {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_CHICKEN_BITS, 0xffbf1f0f, 0x03ab0107),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_GB_ADDR_CONFIG, 0x001877ff, 0x00000044),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_GB_ADDR_CONFIG_READ, 0x001877ff, 0x00000044),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_GFX_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_PAGE_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC0_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC1_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC2_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC3_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC4_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC5_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC6_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_RLC7_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA0_UTCL1_PAGE, 0x007fffff, 0x004c5c00),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_CHICKEN_BITS, 0xffbf1f0f, 0x03ab0107),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_GB_ADDR_CONFIG, 0x001877ff, 0x00000044),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_GB_ADDR_CONFIG_READ, 0x001877ff, 0x00000044),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_GFX_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_PAGE_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC0_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC1_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC2_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC3_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC4_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC5_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC6_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_RLC7_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSDMA1_UTCL1_PAGE, 0x007fffff, 0x004c5c00)
+};
+
+static u32 sdma_v5_0_get_reg_offset(struct amdgpu_device *adev, u32 instance, u32 internal_offset)
+{
+	u32 base;
+
+	if (internal_offset >= SDMA0_HYP_DEC_REG_START &&
+	    internal_offset <= SDMA0_HYP_DEC_REG_END) {
+		base = adev->reg_offset[GC_HWIP][0][1];
+		if (instance == 1)
+			internal_offset += SDMA1_HYP_DEC_REG_OFFSET;
+	} else {
+		base = adev->reg_offset[GC_HWIP][0][0];
+		if (instance == 1)
+			internal_offset += SDMA1_REG_OFFSET;
+	}
+
+	return base + internal_offset;
+}
+
+static void sdma_v5_0_init_golden_registers(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) {
+	case IP_VERSION(5, 0, 0):
+		soc15_program_register_sequence(adev,
+						golden_settings_sdma_5,
+						(const u32)ARRAY_SIZE(golden_settings_sdma_5));
+		soc15_program_register_sequence(adev,
+						golden_settings_sdma_nv10,
+						(const u32)ARRAY_SIZE(golden_settings_sdma_nv10));
+		break;
+	case IP_VERSION(5, 0, 2):
+		soc15_program_register_sequence(adev,
+						golden_settings_sdma_5,
+						(const u32)ARRAY_SIZE(golden_settings_sdma_5));
+		soc15_program_register_sequence(adev,
+						golden_settings_sdma_nv14,
+						(const u32)ARRAY_SIZE(golden_settings_sdma_nv14));
+		break;
+	case IP_VERSION(5, 0, 5):
+		if (amdgpu_sriov_vf(adev))
+			soc15_program_register_sequence(adev,
+							golden_settings_sdma_5_sriov,
+							(const u32)ARRAY_SIZE(golden_settings_sdma_5_sriov));
+		else
+			soc15_program_register_sequence(adev,
+							golden_settings_sdma_5,
+							(const u32)ARRAY_SIZE(golden_settings_sdma_5));
+		soc15_program_register_sequence(adev,
+						golden_settings_sdma_nv12,
+						(const u32)ARRAY_SIZE(golden_settings_sdma_nv12));
+		break;
+	case IP_VERSION(5, 0, 1):
+		soc15_program_register_sequence(adev,
+						golden_settings_sdma_cyan_skillfish,
+						(const u32)ARRAY_SIZE(golden_settings_sdma_cyan_skillfish));
+		break;
+	default:
+		break;
+	}
+}
+
+/**
+ * sdma_v5_0_init_microcode - load ucode images from disk
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Use the firmware interface to load the ucode images into
+ * the driver (not loaded into hw).
+ * Returns 0 on success, error on failure.
+ */
+
+// emulation only, won't work on real chip
+// navi10 real chip need to use PSP to load firmware
+static int sdma_v5_0_init_microcode(struct amdgpu_device *adev)
+{
+	int ret, i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ret = amdgpu_sdma_init_microcode(adev, i, false);
+		if (ret)
+			return ret;
+	}
+
+	return ret;
+}
+
+static unsigned sdma_v5_0_ring_init_cond_exec(struct amdgpu_ring *ring,
+					      uint64_t addr)
+{
+	unsigned ret;
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_COND_EXE));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, 1);
+	/* this is the offset we need patch later */
+	ret = ring->wptr & ring->buf_mask;
+	/* insert dummy here and patch it later */
+	amdgpu_ring_write(ring, 0);
+
+	return ret;
+}
+
+/**
+ * sdma_v5_0_ring_get_rptr - get the current read pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current rptr from the hardware (NAVI10+).
+ */
+static uint64_t sdma_v5_0_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	u64 *rptr;
+
+	/* XXX check if swapping is necessary on BE */
+	rptr = (u64 *)ring->rptr_cpu_addr;
+
+	DRM_DEBUG("rptr before shift == 0x%016llx\n", *rptr);
+	return ((*rptr) >> 2);
+}
+
+/**
+ * sdma_v5_0_ring_get_wptr - get the current write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current wptr from the hardware (NAVI10+).
+ */
+static uint64_t sdma_v5_0_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u64 wptr;
+
+	if (ring->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		wptr = READ_ONCE(*((u64 *)ring->wptr_cpu_addr));
+		DRM_DEBUG("wptr/doorbell before shift == 0x%016llx\n", wptr);
+	} else {
+		wptr = RREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI));
+		wptr = wptr << 32;
+		wptr |= RREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR));
+		DRM_DEBUG("wptr before shift [%i] wptr == 0x%016llx\n", ring->me, wptr);
+	}
+
+	return wptr >> 2;
+}
+
+/**
+ * sdma_v5_0_ring_set_wptr - commit the write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Write the wptr back to the hardware (NAVI10+).
+ */
+static void sdma_v5_0_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	DRM_DEBUG("Setting write pointer\n");
+	if (ring->use_doorbell) {
+		DRM_DEBUG("Using doorbell -- "
+			  "wptr_offs == 0x%08x "
+			  "lower_32_bits(ring->wptr) << 2 == 0x%08x "
+			  "upper_32_bits(ring->wptr) << 2 == 0x%08x\n",
+			  ring->wptr_offs,
+			  lower_32_bits(ring->wptr << 2),
+			  upper_32_bits(ring->wptr << 2));
+		/* XXX check if swapping is necessary on BE */
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr,
+			     ring->wptr << 2);
+		DRM_DEBUG("calling WDOORBELL64(0x%08x, 0x%016llx)\n",
+			  ring->doorbell_index, ring->wptr << 2);
+		WDOORBELL64(ring->doorbell_index, ring->wptr << 2);
+	} else {
+		DRM_DEBUG("Not using doorbell -- "
+			  "mmSDMA%i_GFX_RB_WPTR == 0x%08x "
+			  "mmSDMA%i_GFX_RB_WPTR_HI == 0x%08x\n",
+			  ring->me,
+			  lower_32_bits(ring->wptr << 2),
+			  ring->me,
+			  upper_32_bits(ring->wptr << 2));
+		WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev,
+							     ring->me, mmSDMA0_GFX_RB_WPTR),
+				lower_32_bits(ring->wptr << 2));
+		WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev,
+							     ring->me, mmSDMA0_GFX_RB_WPTR_HI),
+				upper_32_bits(ring->wptr << 2));
+	}
+}
+
+static void sdma_v5_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring);
+	int i;
+
+	for (i = 0; i < count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			amdgpu_ring_write(ring, ring->funcs->nop |
+				SDMA_PKT_NOP_HEADER_COUNT(count - 1));
+		else
+			amdgpu_ring_write(ring, ring->funcs->nop);
+}
+
+/**
+ * sdma_v5_0_ring_emit_ib - Schedule an IB on the DMA engine
+ *
+ * @ring: amdgpu ring pointer
+ * @job: job to retrieve vmid from
+ * @ib: IB object to schedule
+ * @flags: unused
+ *
+ * Schedule an IB in the DMA ring (NAVI10).
+ */
+static void sdma_v5_0_ring_emit_ib(struct amdgpu_ring *ring,
+				   struct amdgpu_job *job,
+				   struct amdgpu_ib *ib,
+				   uint32_t flags)
+{
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+	uint64_t csa_mc_addr = amdgpu_sdma_get_csa_mc_addr(ring, vmid);
+
+	/* An IB packet must end on a 8 DW boundary--the next dword
+	 * must be on a 8-dword boundary. Our IB packet below is 6
+	 * dwords long, thus add x number of NOPs, such that, in
+	 * modular arithmetic,
+	 * wptr + 6 + x = 8k, k >= 0, which in C is,
+	 * (wptr + 6 + x) % 8 = 0.
+	 * The expression below, is a solution of x.
+	 */
+	sdma_v5_0_ring_insert_nop(ring, (2 - lower_32_bits(ring->wptr)) & 7);
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_INDIRECT) |
+			  SDMA_PKT_INDIRECT_HEADER_VMID(vmid & 0xf));
+	/* base must be 32 byte aligned */
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr) & 0xffffffe0);
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, ib->length_dw);
+	amdgpu_ring_write(ring, lower_32_bits(csa_mc_addr));
+	amdgpu_ring_write(ring, upper_32_bits(csa_mc_addr));
+}
+
+/**
+ * sdma_v5_0_ring_emit_mem_sync - flush the IB by graphics cache rinse
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * flush the IB by graphics cache rinse.
+ */
+static void sdma_v5_0_ring_emit_mem_sync(struct amdgpu_ring *ring)
+{
+	uint32_t gcr_cntl = SDMA_GCR_GL2_INV | SDMA_GCR_GL2_WB | SDMA_GCR_GLM_INV |
+			    SDMA_GCR_GL1_INV | SDMA_GCR_GLV_INV | SDMA_GCR_GLK_INV |
+			    SDMA_GCR_GLI_INV(1);
+
+	/* flush entire cache L0/L1/L2, this can be optimized by performance requirement */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_GCR_REQ));
+	amdgpu_ring_write(ring, SDMA_PKT_GCR_REQ_PAYLOAD1_BASE_VA_31_7(0));
+	amdgpu_ring_write(ring, SDMA_PKT_GCR_REQ_PAYLOAD2_GCR_CONTROL_15_0(gcr_cntl) |
+			  SDMA_PKT_GCR_REQ_PAYLOAD2_BASE_VA_47_32(0));
+	amdgpu_ring_write(ring, SDMA_PKT_GCR_REQ_PAYLOAD3_LIMIT_VA_31_7(0) |
+			  SDMA_PKT_GCR_REQ_PAYLOAD3_GCR_CONTROL_18_16(gcr_cntl >> 16));
+	amdgpu_ring_write(ring, SDMA_PKT_GCR_REQ_PAYLOAD4_LIMIT_VA_47_32(0) |
+			  SDMA_PKT_GCR_REQ_PAYLOAD4_VMID(0));
+}
+
+/**
+ * sdma_v5_0_ring_emit_hdp_flush - emit an hdp flush on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Emit an hdp flush packet on the requested DMA ring.
+ */
+static void sdma_v5_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 ref_and_mask = 0;
+	const struct nbio_hdp_flush_reg *nbio_hf_reg = adev->nbio.hdp_flush_reg;
+
+	if (ring->me == 0)
+		ref_and_mask = nbio_hf_reg->ref_and_mask_sdma0;
+	else
+		ref_and_mask = nbio_hf_reg->ref_and_mask_sdma1;
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(1) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3)); /* == */
+	amdgpu_ring_write(ring, (adev->nbio.funcs->get_hdp_flush_done_offset(adev)) << 2);
+	amdgpu_ring_write(ring, (adev->nbio.funcs->get_hdp_flush_req_offset(adev)) << 2);
+	amdgpu_ring_write(ring, ref_and_mask); /* reference */
+	amdgpu_ring_write(ring, ref_and_mask); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
+}
+
+/**
+ * sdma_v5_0_ring_emit_fence - emit a fence on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
+ *
+ * Add a DMA fence packet to the ring to write
+ * the fence seq number and DMA trap packet to generate
+ * an interrupt if needed (NAVI10).
+ */
+static void sdma_v5_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				      unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	/* write the fence */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_FENCE) |
+			  SDMA_PKT_FENCE_HEADER_MTYPE(0x3)); /* Ucached(UC) */
+	/* zero in first two bits */
+	BUG_ON(addr & 0x3);
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+
+	/* optionally write high bits as well */
+	if (write64bit) {
+		addr += 4;
+		amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_FENCE) |
+				  SDMA_PKT_FENCE_HEADER_MTYPE(0x3));
+		/* zero in first two bits */
+		BUG_ON(addr & 0x3);
+		amdgpu_ring_write(ring, lower_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(seq));
+	}
+
+	if (flags & AMDGPU_FENCE_FLAG_INT) {
+		/* generate an interrupt */
+		amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_TRAP));
+		amdgpu_ring_write(ring, SDMA_PKT_TRAP_INT_CONTEXT_INT_CONTEXT(0));
+	}
+}
+
+
+/**
+ * sdma_v5_0_gfx_stop - stop the gfx async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ * @inst_mask: mask of dma engine instances to be disabled
+ * Stop the gfx async dma ring buffers (NAVI10).
+ */
+static void sdma_v5_0_gfx_stop(struct amdgpu_device *adev, uint32_t inst_mask)
+{
+	u32 rb_cntl, ib_cntl;
+	int i;
+
+	for_each_inst(i, inst_mask) {
+		rb_cntl = RREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL));
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 0);
+		WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
+		ib_cntl = RREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL));
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 0);
+		WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL), ib_cntl);
+	}
+}
+
+/**
+ * sdma_v5_0_rlc_stop - stop the compute async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the compute async dma queues (NAVI10).
+ */
+static void sdma_v5_0_rlc_stop(struct amdgpu_device *adev)
+{
+	/* XXX todo */
+}
+
+/**
+ * sdma_v5_0_ctx_switch_enable - stop the async dma engines context switch
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable the DMA MEs context switch.
+ *
+ * Halt or unhalt the async dma engines context switch (NAVI10).
+ */
+static void sdma_v5_0_ctx_switch_enable(struct amdgpu_device *adev, bool enable)
+{
+	u32 f32_cntl = 0, phase_quantum = 0;
+	int i;
+
+	if (amdgpu_sdma_phase_quantum) {
+		unsigned value = amdgpu_sdma_phase_quantum;
+		unsigned unit = 0;
+
+		while (value > (SDMA0_PHASE0_QUANTUM__VALUE_MASK >>
+				SDMA0_PHASE0_QUANTUM__VALUE__SHIFT)) {
+			value = (value + 1) >> 1;
+			unit++;
+		}
+		if (unit > (SDMA0_PHASE0_QUANTUM__UNIT_MASK >>
+			    SDMA0_PHASE0_QUANTUM__UNIT__SHIFT)) {
+			value = (SDMA0_PHASE0_QUANTUM__VALUE_MASK >>
+				 SDMA0_PHASE0_QUANTUM__VALUE__SHIFT);
+			unit = (SDMA0_PHASE0_QUANTUM__UNIT_MASK >>
+				SDMA0_PHASE0_QUANTUM__UNIT__SHIFT);
+			WARN_ONCE(1,
+			"clamping sdma_phase_quantum to %uK clock cycles\n",
+				  value << unit);
+		}
+		phase_quantum =
+			value << SDMA0_PHASE0_QUANTUM__VALUE__SHIFT |
+			unit  << SDMA0_PHASE0_QUANTUM__UNIT__SHIFT;
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (!amdgpu_sriov_vf(adev)) {
+			f32_cntl = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_CNTL));
+			f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL,
+						 AUTO_CTXSW_ENABLE, enable ? 1 : 0);
+		}
+
+		if (enable && amdgpu_sdma_phase_quantum) {
+			WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_PHASE0_QUANTUM),
+			       phase_quantum);
+			WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_PHASE1_QUANTUM),
+			       phase_quantum);
+			WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_PHASE2_QUANTUM),
+			       phase_quantum);
+		}
+		if (!amdgpu_sriov_vf(adev))
+			WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_CNTL), f32_cntl);
+	}
+
+}
+
+/**
+ * sdma_v5_0_enable - stop the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable the DMA MEs.
+ *
+ * Halt or unhalt the async dma engines (NAVI10).
+ */
+static void sdma_v5_0_enable(struct amdgpu_device *adev, bool enable)
+{
+	u32 f32_cntl;
+	int i;
+	uint32_t inst_mask;
+
+	inst_mask = GENMASK(adev->sdma.num_instances - 1, 0);
+	if (!enable) {
+		sdma_v5_0_gfx_stop(adev, 1 << inst_mask);
+		sdma_v5_0_rlc_stop(adev);
+	}
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		f32_cntl = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_F32_CNTL));
+		f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_F32_CNTL, HALT, enable ? 0 : 1);
+		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_F32_CNTL), f32_cntl);
+	}
+}
+
+/**
+ * sdma_v5_0_gfx_resume_instance - start/restart a certain sdma engine
+ *
+ * @adev: amdgpu_device pointer
+ * @i: instance
+ * @restore: used to restore wptr when restart
+ *
+ * Set up the gfx DMA ring buffers and enable them. On restart, we will restore wptr and rptr.
+ * Return 0 for success.
+ */
+static int sdma_v5_0_gfx_resume_instance(struct amdgpu_device *adev, int i, bool restore)
+{
+	struct amdgpu_ring *ring;
+	u32 rb_cntl, ib_cntl;
+	u32 rb_bufsz;
+	u32 doorbell;
+	u32 doorbell_offset;
+	u32 temp;
+	u32 wptr_poll_cntl;
+	u64 wptr_gpu_addr;
+
+	ring = &adev->sdma.instance[i].ring;
+
+	if (!amdgpu_sriov_vf(adev))
+		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_SEM_WAIT_FAIL_TIMER_CNTL), 0);
+
+	/* Set ring buffer size in dwords */
+	rb_bufsz = order_base_2(ring->ring_size / 4);
+	rb_cntl = RREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL));
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SIZE, rb_bufsz);
+#ifdef __BIG_ENDIAN
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SWAP_ENABLE, 1);
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL,
+				RPTR_WRITEBACK_SWAP_ENABLE, 1);
+#endif
+	WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
+
+	/* Initialize the ring buffer's read and write pointers */
+	if (restore) {
+		WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR), lower_32_bits(ring->wptr << 2));
+		WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_HI), upper_32_bits(ring->wptr << 2));
+		WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR), lower_32_bits(ring->wptr << 2));
+		WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_HI), upper_32_bits(ring->wptr << 2));
+	} else {
+		WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR), 0);
+		WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_HI), 0);
+		WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR), 0);
+		WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_HI), 0);
+	}
+	/* setup the wptr shadow polling */
+	wptr_gpu_addr = ring->wptr_gpu_addr;
+	WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_LO),
+	       lower_32_bits(wptr_gpu_addr));
+	WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_HI),
+	       upper_32_bits(wptr_gpu_addr));
+	wptr_poll_cntl = RREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i,
+						 mmSDMA0_GFX_RB_WPTR_POLL_CNTL));
+	wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl,
+				       SDMA0_GFX_RB_WPTR_POLL_CNTL,
+				       F32_POLL_ENABLE, 1);
+	WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL),
+	       wptr_poll_cntl);
+
+	/* set the wb address whether it's enabled or not */
+	WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_ADDR_HI),
+	       upper_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFF);
+	WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_ADDR_LO),
+	       lower_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFC);
+
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RPTR_WRITEBACK_ENABLE, 1);
+
+	WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_BASE),
+	       ring->gpu_addr >> 8);
+	WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_BASE_HI),
+	       ring->gpu_addr >> 40);
+
+	if (!restore)
+		ring->wptr = 0;
+
+	/* before programing wptr to a less value, need set minor_ptr_update first */
+	WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_MINOR_PTR_UPDATE), 1);
+
+	if (!amdgpu_sriov_vf(adev)) { /* only bare-metal use register write for wptr */
+		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR),
+		       lower_32_bits(ring->wptr << 2));
+		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_HI),
+		       upper_32_bits(ring->wptr << 2));
+	}
+
+	doorbell = RREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL));
+	doorbell_offset = RREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i,
+					mmSDMA0_GFX_DOORBELL_OFFSET));
+
+	if (ring->use_doorbell) {
+		doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, 1);
+		doorbell_offset = REG_SET_FIELD(doorbell_offset, SDMA0_GFX_DOORBELL_OFFSET,
+				OFFSET, ring->doorbell_index);
+	} else {
+		doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, 0);
+	}
+	WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL), doorbell);
+	WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL_OFFSET),
+	       doorbell_offset);
+
+	adev->nbio.funcs->sdma_doorbell_range(adev, i, ring->use_doorbell,
+					      ring->doorbell_index, 20);
+
+	if (amdgpu_sriov_vf(adev))
+		sdma_v5_0_ring_set_wptr(ring);
+
+	/* set minor_ptr_update to 0 after wptr programed */
+	WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_MINOR_PTR_UPDATE), 0);
+
+	if (!amdgpu_sriov_vf(adev)) {
+		/* set utc l1 enable flag always to 1 */
+		temp = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_CNTL));
+		temp = REG_SET_FIELD(temp, SDMA0_CNTL, UTC_L1_ENABLE, 1);
+
+		/* enable MCBP */
+		temp = REG_SET_FIELD(temp, SDMA0_CNTL, MIDCMD_PREEMPT_ENABLE, 1);
+		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_CNTL), temp);
+
+		/* Set up RESP_MODE to non-copy addresses */
+		temp = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_UTCL1_CNTL));
+		temp = REG_SET_FIELD(temp, SDMA0_UTCL1_CNTL, RESP_MODE, 3);
+		temp = REG_SET_FIELD(temp, SDMA0_UTCL1_CNTL, REDO_DELAY, 9);
+		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_UTCL1_CNTL), temp);
+
+		/* program default cache read and write policy */
+		temp = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_UTCL1_PAGE));
+		/* clean read policy and write policy bits */
+		temp &= 0xFF0FFF;
+		temp |= ((CACHE_READ_POLICY_L2__DEFAULT << 12) | (CACHE_WRITE_POLICY_L2__DEFAULT << 14));
+		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_UTCL1_PAGE), temp);
+	}
+
+	if (!amdgpu_sriov_vf(adev)) {
+		/* unhalt engine */
+		temp = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_F32_CNTL));
+		temp = REG_SET_FIELD(temp, SDMA0_F32_CNTL, HALT, 0);
+		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_F32_CNTL), temp);
+	}
+
+	/* enable DMA RB */
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 1);
+	WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
+
+	ib_cntl = RREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL));
+	ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 1);
+#ifdef __BIG_ENDIAN
+	ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_SWAP_ENABLE, 1);
+#endif
+	/* enable DMA IBs */
+	WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL), ib_cntl);
+
+	if (amdgpu_sriov_vf(adev)) { /* bare-metal sequence doesn't need below to lines */
+		sdma_v5_0_ctx_switch_enable(adev, true);
+		sdma_v5_0_enable(adev, true);
+	}
+
+	return amdgpu_ring_test_helper(ring);
+}
+
+/**
+ * sdma_v5_0_gfx_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the gfx DMA ring buffers and enable them (NAVI10).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v5_0_gfx_resume(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		r = sdma_v5_0_gfx_resume_instance(adev, i, false);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+/**
+ * sdma_v5_0_rlc_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the compute DMA queues and enable them (NAVI10).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v5_0_rlc_resume(struct amdgpu_device *adev)
+{
+	return 0;
+}
+
+/**
+ * sdma_v5_0_load_microcode - load the sDMA ME ucode
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Loads the sDMA0/1 ucode.
+ * Returns 0 for success, -EINVAL if the ucode is not available.
+ */
+static int sdma_v5_0_load_microcode(struct amdgpu_device *adev)
+{
+	const struct sdma_firmware_header_v1_0 *hdr;
+	const __le32 *fw_data;
+	u32 fw_size;
+	int i, j;
+
+	/* halt the MEs */
+	sdma_v5_0_enable(adev, false);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (!adev->sdma.instance[i].fw)
+			return -EINVAL;
+
+		hdr = (const struct sdma_firmware_header_v1_0 *)adev->sdma.instance[i].fw->data;
+		amdgpu_ucode_print_sdma_hdr(&hdr->header);
+		fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+
+		fw_data = (const __le32 *)
+			(adev->sdma.instance[i].fw->data +
+				le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+
+		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_UCODE_ADDR), 0);
+
+		for (j = 0; j < fw_size; j++) {
+			if (amdgpu_emu_mode == 1 && j % 500 == 0)
+				msleep(1);
+			WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_UCODE_DATA), le32_to_cpup(fw_data++));
+		}
+
+		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_UCODE_ADDR), adev->sdma.instance[i].fw_version);
+	}
+
+	return 0;
+}
+
+/**
+ * sdma_v5_0_start - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the DMA engines and enable them (NAVI10).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v5_0_start(struct amdgpu_device *adev)
+{
+	int r = 0;
+
+	if (amdgpu_sriov_vf(adev)) {
+		sdma_v5_0_ctx_switch_enable(adev, false);
+		sdma_v5_0_enable(adev, false);
+
+		/* set RB registers */
+		r = sdma_v5_0_gfx_resume(adev);
+		return r;
+	}
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+		r = sdma_v5_0_load_microcode(adev);
+		if (r)
+			return r;
+	}
+
+	/* unhalt the MEs */
+	sdma_v5_0_enable(adev, true);
+	/* enable sdma ring preemption */
+	sdma_v5_0_ctx_switch_enable(adev, true);
+
+	/* start the gfx rings and rlc compute queues */
+	r = sdma_v5_0_gfx_resume(adev);
+	if (r)
+		return r;
+	r = sdma_v5_0_rlc_resume(adev);
+
+	return r;
+}
+
+static int sdma_v5_0_mqd_init(struct amdgpu_device *adev, void *mqd,
+			      struct amdgpu_mqd_prop *prop)
+{
+	struct v10_sdma_mqd *m = mqd;
+	uint64_t wb_gpu_addr;
+
+	m->sdmax_rlcx_rb_cntl =
+		order_base_2(prop->queue_size / 4) << SDMA0_RLC0_RB_CNTL__RB_SIZE__SHIFT |
+		1 << SDMA0_RLC0_RB_CNTL__RPTR_WRITEBACK_ENABLE__SHIFT |
+		6 << SDMA0_RLC0_RB_CNTL__RPTR_WRITEBACK_TIMER__SHIFT |
+		1 << SDMA0_RLC0_RB_CNTL__RB_PRIV__SHIFT;
+
+	m->sdmax_rlcx_rb_base = lower_32_bits(prop->hqd_base_gpu_addr >> 8);
+	m->sdmax_rlcx_rb_base_hi = upper_32_bits(prop->hqd_base_gpu_addr >> 8);
+
+	m->sdmax_rlcx_rb_wptr_poll_cntl = RREG32(sdma_v5_0_get_reg_offset(adev, 0,
+						  mmSDMA0_GFX_RB_WPTR_POLL_CNTL));
+
+	wb_gpu_addr = prop->wptr_gpu_addr;
+	m->sdmax_rlcx_rb_wptr_poll_addr_lo = lower_32_bits(wb_gpu_addr);
+	m->sdmax_rlcx_rb_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr);
+
+	wb_gpu_addr = prop->rptr_gpu_addr;
+	m->sdmax_rlcx_rb_rptr_addr_lo = lower_32_bits(wb_gpu_addr);
+	m->sdmax_rlcx_rb_rptr_addr_hi = upper_32_bits(wb_gpu_addr);
+
+	m->sdmax_rlcx_ib_cntl = RREG32(sdma_v5_0_get_reg_offset(adev, 0,
+							mmSDMA0_GFX_IB_CNTL));
+
+	m->sdmax_rlcx_doorbell_offset =
+		prop->doorbell_index << SDMA0_RLC0_DOORBELL_OFFSET__OFFSET__SHIFT;
+
+	m->sdmax_rlcx_doorbell = REG_SET_FIELD(0, SDMA0_RLC0_DOORBELL, ENABLE, 1);
+
+	return 0;
+}
+
+static void sdma_v5_0_set_mqd_funcs(struct amdgpu_device *adev)
+{
+	adev->mqds[AMDGPU_HW_IP_DMA].mqd_size = sizeof(struct v10_sdma_mqd);
+	adev->mqds[AMDGPU_HW_IP_DMA].init_mqd = sdma_v5_0_mqd_init;
+}
+
+/**
+ * sdma_v5_0_ring_test_ring - simple async dma engine test
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Test the DMA engine by writing using it to write an
+ * value to memory. (NAVI10).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v5_0_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned i;
+	unsigned index;
+	int r;
+	u32 tmp;
+	u64 gpu_addr;
+
+	tmp = 0xCAFEDEAD;
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+
+	r = amdgpu_ring_alloc(ring, 20);
+	if (r) {
+		DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r);
+		amdgpu_device_wb_free(adev, index);
+		return r;
+	}
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+			  SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR));
+	amdgpu_ring_write(ring, lower_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(0));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = le32_to_cpu(adev->wb.wb[index]);
+		if (tmp == 0xDEADBEEF)
+			break;
+		if (amdgpu_emu_mode == 1)
+			msleep(1);
+		else
+			udelay(1);
+	}
+
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+
+	amdgpu_device_wb_free(adev, index);
+
+	return r;
+}
+
+/**
+ * sdma_v5_0_ring_test_ib - test an IB on the DMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @timeout: timeout value in jiffies, or MAX_SCHEDULE_TIMEOUT
+ *
+ * Test a simple IB in the DMA ring (NAVI10).
+ * Returns 0 on success, error on failure.
+ */
+static int sdma_v5_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct dma_fence *f = NULL;
+	unsigned index;
+	long r;
+	u32 tmp = 0;
+	u64 gpu_addr;
+
+	tmp = 0xCAFEDEAD;
+	memset(&ib, 0, sizeof(ib));
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+
+	r = amdgpu_ib_get(adev, NULL, 256,
+			  AMDGPU_IB_POOL_DIRECT, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
+		goto err0;
+	}
+
+	ib.ptr[0] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR);
+	ib.ptr[1] = lower_32_bits(gpu_addr);
+	ib.ptr[2] = upper_32_bits(gpu_addr);
+	ib.ptr[3] = SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(0);
+	ib.ptr[4] = 0xDEADBEEF;
+	ib.ptr[5] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.ptr[6] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.ptr[7] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.length_dw = 8;
+
+	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
+	if (r)
+		goto err1;
+
+	r = dma_fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out\n");
+		r = -ETIMEDOUT;
+		goto err1;
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
+		goto err1;
+	}
+
+	tmp = le32_to_cpu(adev->wb.wb[index]);
+
+	if (tmp == 0xDEADBEEF)
+		r = 0;
+	else
+		r = -EINVAL;
+
+err1:
+	amdgpu_ib_free(&ib, NULL);
+	dma_fence_put(f);
+err0:
+	amdgpu_device_wb_free(adev, index);
+	return r;
+}
+
+
+/**
+ * sdma_v5_0_vm_copy_pte - update PTEs by copying them from the GART
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @src: src addr to copy from
+ * @count: number of page entries to update
+ *
+ * Update PTEs by copying them from the GART using sDMA (NAVI10).
+ */
+static void sdma_v5_0_vm_copy_pte(struct amdgpu_ib *ib,
+				  uint64_t pe, uint64_t src,
+				  unsigned count)
+{
+	unsigned bytes = count * 8;
+
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
+	ib->ptr[ib->length_dw++] = bytes - 1;
+	ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+	ib->ptr[ib->length_dw++] = lower_32_bits(src);
+	ib->ptr[ib->length_dw++] = upper_32_bits(src);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+
+}
+
+/**
+ * sdma_v5_0_vm_write_pte - update PTEs by writing them manually
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @value: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ *
+ * Update PTEs by writing them manually using sDMA (NAVI10).
+ */
+static void sdma_v5_0_vm_write_pte(struct amdgpu_ib *ib, uint64_t pe,
+				   uint64_t value, unsigned count,
+				   uint32_t incr)
+{
+	unsigned ndw = count * 2;
+
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+	ib->ptr[ib->length_dw++] = ndw - 1;
+	for (; ndw > 0; ndw -= 2) {
+		ib->ptr[ib->length_dw++] = lower_32_bits(value);
+		ib->ptr[ib->length_dw++] = upper_32_bits(value);
+		value += incr;
+	}
+}
+
+/**
+ * sdma_v5_0_vm_set_pte_pde - update the page tables using sDMA
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: access flags
+ *
+ * Update the page tables using sDMA (NAVI10).
+ */
+static void sdma_v5_0_vm_set_pte_pde(struct amdgpu_ib *ib,
+				     uint64_t pe,
+				     uint64_t addr, unsigned count,
+				     uint32_t incr, uint64_t flags)
+{
+	/* for physically contiguous pages (vram) */
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_PTEPDE);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe); /* dst addr */
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+	ib->ptr[ib->length_dw++] = lower_32_bits(flags); /* mask */
+	ib->ptr[ib->length_dw++] = upper_32_bits(flags);
+	ib->ptr[ib->length_dw++] = lower_32_bits(addr); /* value */
+	ib->ptr[ib->length_dw++] = upper_32_bits(addr);
+	ib->ptr[ib->length_dw++] = incr; /* increment size */
+	ib->ptr[ib->length_dw++] = 0;
+	ib->ptr[ib->length_dw++] = count - 1; /* number of entries */
+}
+
+/**
+ * sdma_v5_0_ring_pad_ib - pad the IB
+ * @ring: amdgpu_ring structure holding ring information
+ * @ib: indirect buffer to fill with padding
+ *
+ * Pad the IB with NOPs to a boundary multiple of 8.
+ */
+static void sdma_v5_0_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring);
+	u32 pad_count;
+	int i;
+
+	pad_count = (-ib->length_dw) & 0x7;
+	for (i = 0; i < pad_count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			ib->ptr[ib->length_dw++] =
+				SDMA_PKT_HEADER_OP(SDMA_OP_NOP) |
+				SDMA_PKT_NOP_HEADER_COUNT(pad_count - 1);
+		else
+			ib->ptr[ib->length_dw++] =
+				SDMA_PKT_HEADER_OP(SDMA_OP_NOP);
+}
+
+
+/**
+ * sdma_v5_0_ring_emit_pipeline_sync - sync the pipeline
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Make sure all previous operations are completed (CIK).
+ */
+static void sdma_v5_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	/* wait for idle */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(0) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3) | /* equal */
+			  SDMA_PKT_POLL_REGMEM_HEADER_MEM_POLL(1));
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xffffffff);
+	amdgpu_ring_write(ring, seq); /* reference */
+	amdgpu_ring_write(ring, 0xffffffff); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(4)); /* retry count, poll interval */
+}
+
+
+/**
+ * sdma_v5_0_ring_emit_vm_flush - vm flush using sDMA
+ *
+ * @ring: amdgpu_ring pointer
+ * @vmid: vmid number to use
+ * @pd_addr: address
+ *
+ * Update the page table base and flush the VM TLB
+ * using sDMA (NAVI10).
+ */
+static void sdma_v5_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					 unsigned vmid, uint64_t pd_addr)
+{
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+}
+
+static void sdma_v5_0_ring_emit_wreg(struct amdgpu_ring *ring,
+				     uint32_t reg, uint32_t val)
+{
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
+			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, val);
+}
+
+static void sdma_v5_0_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+					 uint32_t val, uint32_t mask)
+{
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(0) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3)); /* equal */
+	amdgpu_ring_write(ring, reg << 2);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, val); /* reference */
+	amdgpu_ring_write(ring, mask); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10));
+}
+
+static void sdma_v5_0_ring_emit_reg_write_reg_wait(struct amdgpu_ring *ring,
+						   uint32_t reg0, uint32_t reg1,
+						   uint32_t ref, uint32_t mask)
+{
+	amdgpu_ring_emit_wreg(ring, reg0, ref);
+	/* wait for a cycle to reset vm_inv_eng*_ack */
+	amdgpu_ring_emit_reg_wait(ring, reg0, 0, 0);
+	amdgpu_ring_emit_reg_wait(ring, reg1, mask, mask);
+}
+
+static int sdma_v5_0_soft_reset_engine(struct amdgpu_device *adev, u32 instance_id)
+{
+	u32 grbm_soft_reset;
+	u32 tmp;
+
+	grbm_soft_reset = REG_SET_FIELD(0,
+					GRBM_SOFT_RESET, SOFT_RESET_SDMA0,
+					1);
+	grbm_soft_reset <<= instance_id;
+
+	tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
+	tmp |= grbm_soft_reset;
+	DRM_DEBUG("GRBM_SOFT_RESET=0x%08X\n", tmp);
+	WREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET, tmp);
+	tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
+
+	udelay(50);
+
+	tmp &= ~grbm_soft_reset;
+	WREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET, tmp);
+	tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
+	return 0;
+}
+
+static const struct amdgpu_sdma_funcs sdma_v5_0_sdma_funcs = {
+	.stop_kernel_queue = &sdma_v5_0_stop_queue,
+	.start_kernel_queue = &sdma_v5_0_restore_queue,
+	.soft_reset_kernel_queue = &sdma_v5_0_soft_reset_engine,
+};
+
+static int sdma_v5_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = sdma_v5_0_init_microcode(adev);
+	if (r)
+		return r;
+
+	sdma_v5_0_set_ring_funcs(adev);
+	sdma_v5_0_set_buffer_funcs(adev);
+	sdma_v5_0_set_vm_pte_funcs(adev);
+	sdma_v5_0_set_irq_funcs(adev);
+	sdma_v5_0_set_mqd_funcs(adev);
+
+	return 0;
+}
+
+
+static int sdma_v5_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_ring *ring;
+	int r, i;
+	struct amdgpu_device *adev = ip_block->adev;
+	uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_5_0);
+	uint32_t *ptr;
+
+	/* SDMA trap event */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_SDMA0,
+			      SDMA0_5_0__SRCID__SDMA_TRAP,
+			      &adev->sdma.trap_irq);
+	if (r)
+		return r;
+
+	/* SDMA trap event */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_SDMA1,
+			      SDMA1_5_0__SRCID__SDMA_TRAP,
+			      &adev->sdma.trap_irq);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		mutex_init(&adev->sdma.instance[i].engine_reset_mutex);
+		adev->sdma.instance[i].funcs = &sdma_v5_0_sdma_funcs;
+		ring = &adev->sdma.instance[i].ring;
+		ring->ring_obj = NULL;
+		ring->use_doorbell = true;
+
+		DRM_DEBUG("SDMA %d use_doorbell being set to: [%s]\n", i,
+				ring->use_doorbell?"true":"false");
+
+		ring->doorbell_index = (i == 0) ?
+			(adev->doorbell_index.sdma_engine[0] << 1) //get DWORD offset
+			: (adev->doorbell_index.sdma_engine[1] << 1); // get DWORD offset
+
+		ring->vm_hub = AMDGPU_GFXHUB(0);
+		sprintf(ring->name, "sdma%d", i);
+		r = amdgpu_ring_init(adev, ring, 1024, &adev->sdma.trap_irq,
+				     (i == 0) ? AMDGPU_SDMA_IRQ_INSTANCE0 :
+				     AMDGPU_SDMA_IRQ_INSTANCE1,
+				     AMDGPU_RING_PRIO_DEFAULT, NULL);
+		if (r)
+			return r;
+	}
+
+	adev->sdma.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->sdma.instance[0].ring);
+	switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) {
+	case IP_VERSION(5, 0, 0):
+	case IP_VERSION(5, 0, 2):
+	case IP_VERSION(5, 0, 5):
+		if ((adev->sdma.instance[0].fw_version >= 35) &&
+		    !amdgpu_sriov_vf(adev))
+			adev->sdma.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+		break;
+	default:
+		break;
+	}
+
+	/* Allocate memory for SDMA IP Dump buffer */
+	ptr = kcalloc(adev->sdma.num_instances * reg_count, sizeof(uint32_t), GFP_KERNEL);
+	if (ptr)
+		adev->sdma.ip_dump = ptr;
+	else
+		DRM_ERROR("Failed to allocated memory for SDMA IP Dump\n");
+
+	r = amdgpu_sdma_sysfs_reset_mask_init(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int sdma_v5_0_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		amdgpu_ring_fini(&adev->sdma.instance[i].ring);
+
+	amdgpu_sdma_sysfs_reset_mask_fini(adev);
+	amdgpu_sdma_destroy_inst_ctx(adev, false);
+
+	kfree(adev->sdma.ip_dump);
+
+	return 0;
+}
+
+static int sdma_v5_0_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	sdma_v5_0_init_golden_registers(adev);
+
+	r = sdma_v5_0_start(adev);
+
+	return r;
+}
+
+static int sdma_v5_0_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	sdma_v5_0_ctx_switch_enable(adev, false);
+	sdma_v5_0_enable(adev, false);
+
+	return 0;
+}
+
+static int sdma_v5_0_suspend(struct amdgpu_ip_block *ip_block)
+{
+	return sdma_v5_0_hw_fini(ip_block);
+}
+
+static int sdma_v5_0_resume(struct amdgpu_ip_block *ip_block)
+{
+	return sdma_v5_0_hw_init(ip_block);
+}
+
+static bool sdma_v5_0_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	u32 i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		u32 tmp = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_STATUS_REG));
+
+		if (!(tmp & SDMA0_STATUS_REG__IDLE_MASK))
+			return false;
+	}
+
+	return true;
+}
+
+static int sdma_v5_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	unsigned i;
+	u32 sdma0, sdma1;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		sdma0 = RREG32(sdma_v5_0_get_reg_offset(adev, 0, mmSDMA0_STATUS_REG));
+		sdma1 = RREG32(sdma_v5_0_get_reg_offset(adev, 1, mmSDMA0_STATUS_REG));
+
+		if (sdma0 & sdma1 & SDMA0_STATUS_REG__IDLE_MASK)
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static int sdma_v5_0_soft_reset(struct amdgpu_ip_block *ip_block)
+{
+	/* todo */
+
+	return 0;
+}
+
+static int sdma_v5_0_reset_queue(struct amdgpu_ring *ring,
+				 unsigned int vmid,
+				 struct amdgpu_fence *timedout_fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int r;
+
+	if (ring->me >= adev->sdma.num_instances) {
+		dev_err(adev->dev, "sdma instance not found\n");
+		return -EINVAL;
+	}
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+
+	amdgpu_amdkfd_suspend(adev, true);
+	r = amdgpu_sdma_reset_engine(adev, ring->me, true);
+	amdgpu_amdkfd_resume(adev, true);
+	if (r)
+		return r;
+
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static int sdma_v5_0_stop_queue(struct amdgpu_ring *ring)
+{
+	u32 f32_cntl, freeze, cntl, stat1_reg;
+	struct amdgpu_device *adev = ring->adev;
+	int i, j, r = 0;
+
+	if (amdgpu_sriov_vf(adev))
+		return -EINVAL;
+
+	i = ring->me;
+	amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
+
+	/* stop queue */
+	sdma_v5_0_gfx_stop(adev, 1 << i);
+
+	/* engine stop SDMA1_F32_CNTL.HALT to 1 and SDMAx_FREEZE freeze bit to 1 */
+	freeze = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_FREEZE));
+	freeze = REG_SET_FIELD(freeze, SDMA0_FREEZE, FREEZE, 1);
+	WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_FREEZE), freeze);
+
+	for (j = 0; j < adev->usec_timeout; j++) {
+		freeze = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_FREEZE));
+		if (REG_GET_FIELD(freeze, SDMA0_FREEZE, FROZEN) & 1)
+			break;
+		udelay(1);
+	}
+
+	/* check sdma copy engine all idle if frozen not received*/
+	if (j == adev->usec_timeout) {
+		stat1_reg = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_STATUS1_REG));
+		if ((stat1_reg & 0x3FF) != 0x3FF) {
+			DRM_ERROR("cannot soft reset as sdma not idle\n");
+			r = -ETIMEDOUT;
+			goto err0;
+		}
+	}
+
+	f32_cntl = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_F32_CNTL));
+	f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_F32_CNTL, HALT, 1);
+	WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_F32_CNTL), f32_cntl);
+
+	cntl = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_CNTL));
+	cntl = REG_SET_FIELD(cntl, SDMA0_CNTL, UTC_L1_ENABLE, 0);
+	WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_CNTL), cntl);
+err0:
+	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
+	return r;
+}
+
+static int sdma_v5_0_restore_queue(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 inst_id = ring->me;
+	u32 freeze;
+	int r;
+
+	amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
+	/* unfreeze*/
+	freeze = RREG32(sdma_v5_0_get_reg_offset(adev, inst_id, mmSDMA0_FREEZE));
+	freeze = REG_SET_FIELD(freeze, SDMA0_FREEZE, FREEZE, 0);
+	WREG32(sdma_v5_0_get_reg_offset(adev, inst_id, mmSDMA0_FREEZE), freeze);
+
+	r = sdma_v5_0_gfx_resume_instance(adev, inst_id, true);
+	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
+
+	return r;
+}
+
+static int sdma_v5_0_ring_preempt_ib(struct amdgpu_ring *ring)
+{
+	int i, r = 0;
+	struct amdgpu_device *adev = ring->adev;
+	u32 index = 0;
+	u64 sdma_gfx_preempt;
+
+	amdgpu_sdma_get_index_from_ring(ring, &index);
+	if (index == 0)
+		sdma_gfx_preempt = mmSDMA0_GFX_PREEMPT;
+	else
+		sdma_gfx_preempt = mmSDMA1_GFX_PREEMPT;
+
+	/* assert preemption condition */
+	amdgpu_ring_set_preempt_cond_exec(ring, false);
+
+	/* emit the trailing fence */
+	ring->trail_seq += 1;
+	amdgpu_ring_alloc(ring, 10);
+	sdma_v5_0_ring_emit_fence(ring, ring->trail_fence_gpu_addr,
+				  ring->trail_seq, 0);
+	amdgpu_ring_commit(ring);
+
+	/* assert IB preemption */
+	WREG32(sdma_gfx_preempt, 1);
+
+	/* poll the trailing fence */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (ring->trail_seq ==
+		    le32_to_cpu(*(ring->trail_fence_cpu_addr)))
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout) {
+		r = -EINVAL;
+		DRM_ERROR("ring %d failed to be preempted\n", ring->idx);
+	}
+
+	/* deassert IB preemption */
+	WREG32(sdma_gfx_preempt, 0);
+
+	/* deassert the preemption condition */
+	amdgpu_ring_set_preempt_cond_exec(ring, true);
+	return r;
+}
+
+static int sdma_v5_0_set_trap_irq_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	u32 sdma_cntl;
+
+	if (!amdgpu_sriov_vf(adev)) {
+		u32 reg_offset = (type == AMDGPU_SDMA_IRQ_INSTANCE0) ?
+			sdma_v5_0_get_reg_offset(adev, 0, mmSDMA0_CNTL) :
+			sdma_v5_0_get_reg_offset(adev, 1, mmSDMA0_CNTL);
+
+		sdma_cntl = RREG32(reg_offset);
+		sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE,
+					  state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+		WREG32(reg_offset, sdma_cntl);
+	}
+
+	return 0;
+}
+
+static int sdma_v5_0_process_trap_irq(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	uint32_t mes_queue_id = entry->src_data[0];
+
+	DRM_DEBUG("IH: SDMA trap\n");
+
+	if (adev->enable_mes && (mes_queue_id & AMDGPU_FENCE_MES_QUEUE_FLAG)) {
+		struct amdgpu_mes_queue *queue;
+
+		mes_queue_id &= AMDGPU_FENCE_MES_QUEUE_ID_MASK;
+
+		spin_lock(&adev->mes.queue_id_lock);
+		queue = idr_find(&adev->mes.queue_id_idr, mes_queue_id);
+		if (queue) {
+			DRM_DEBUG("process smda queue id = %d\n", mes_queue_id);
+			amdgpu_fence_process(queue->ring);
+		}
+		spin_unlock(&adev->mes.queue_id_lock);
+		return 0;
+	}
+
+	switch (entry->client_id) {
+	case SOC15_IH_CLIENTID_SDMA0:
+		switch (entry->ring_id) {
+		case 0:
+			amdgpu_fence_process(&adev->sdma.instance[0].ring);
+			break;
+		case 1:
+			/* XXX compute */
+			break;
+		case 2:
+			/* XXX compute */
+			break;
+		case 3:
+			/* XXX page queue*/
+			break;
+		}
+		break;
+	case SOC15_IH_CLIENTID_SDMA1:
+		switch (entry->ring_id) {
+		case 0:
+			amdgpu_fence_process(&adev->sdma.instance[1].ring);
+			break;
+		case 1:
+			/* XXX compute */
+			break;
+		case 2:
+			/* XXX compute */
+			break;
+		case 3:
+			/* XXX page queue*/
+			break;
+		}
+		break;
+	}
+	return 0;
+}
+
+static int sdma_v5_0_process_illegal_inst_irq(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      struct amdgpu_iv_entry *entry)
+{
+	return 0;
+}
+
+static void sdma_v5_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						       bool enable)
+{
+	uint32_t data, def;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (enable && (adev->cg_flags & AMD_CG_SUPPORT_SDMA_MGCG)) {
+			/* Enable sdma clock gating */
+			def = data = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_CLK_CTRL));
+			data &= ~(SDMA0_CLK_CTRL__SOFT_OVERRIDE7_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE6_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE5_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE4_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE3_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE2_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE1_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE0_MASK);
+			if (def != data)
+				WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_CLK_CTRL), data);
+		} else {
+			/* Disable sdma clock gating */
+			def = data = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_CLK_CTRL));
+			data |= (SDMA0_CLK_CTRL__SOFT_OVERRIDE7_MASK |
+				 SDMA0_CLK_CTRL__SOFT_OVERRIDE6_MASK |
+				 SDMA0_CLK_CTRL__SOFT_OVERRIDE5_MASK |
+				 SDMA0_CLK_CTRL__SOFT_OVERRIDE4_MASK |
+				 SDMA0_CLK_CTRL__SOFT_OVERRIDE3_MASK |
+				 SDMA0_CLK_CTRL__SOFT_OVERRIDE2_MASK |
+				 SDMA0_CLK_CTRL__SOFT_OVERRIDE1_MASK |
+				 SDMA0_CLK_CTRL__SOFT_OVERRIDE0_MASK);
+			if (def != data)
+				WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_CLK_CTRL), data);
+		}
+	}
+}
+
+static void sdma_v5_0_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+						      bool enable)
+{
+	uint32_t data, def;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (enable && (adev->cg_flags & AMD_CG_SUPPORT_SDMA_LS)) {
+			/* Enable sdma mem light sleep */
+			def = data = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_POWER_CNTL));
+			data |= SDMA0_POWER_CNTL__MEM_POWER_OVERRIDE_MASK;
+			if (def != data)
+				WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_POWER_CNTL), data);
+
+		} else {
+			/* Disable sdma mem light sleep */
+			def = data = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_POWER_CNTL));
+			data &= ~SDMA0_POWER_CNTL__MEM_POWER_OVERRIDE_MASK;
+			if (def != data)
+				WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_POWER_CNTL), data);
+
+		}
+	}
+}
+
+static int sdma_v5_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					   enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) {
+	case IP_VERSION(5, 0, 0):
+	case IP_VERSION(5, 0, 2):
+	case IP_VERSION(5, 0, 5):
+		sdma_v5_0_update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		sdma_v5_0_update_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int sdma_v5_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static void sdma_v5_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int data;
+
+	if (amdgpu_sriov_vf(adev))
+		*flags = 0;
+
+	/* AMD_CG_SUPPORT_SDMA_MGCG */
+	data = RREG32(sdma_v5_0_get_reg_offset(adev, 0, mmSDMA0_CLK_CTRL));
+	if (!(data & SDMA0_CLK_CTRL__SOFT_OVERRIDE7_MASK))
+		*flags |= AMD_CG_SUPPORT_SDMA_MGCG;
+
+	/* AMD_CG_SUPPORT_SDMA_LS */
+	data = RREG32(sdma_v5_0_get_reg_offset(adev, 0, mmSDMA0_POWER_CNTL));
+	if (data & SDMA0_POWER_CNTL__MEM_POWER_OVERRIDE_MASK)
+		*flags |= AMD_CG_SUPPORT_SDMA_LS;
+}
+
+static void sdma_v5_0_print_ip_state(struct amdgpu_ip_block *ip_block, struct drm_printer *p)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, j;
+	uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_5_0);
+	uint32_t instance_offset;
+
+	if (!adev->sdma.ip_dump)
+		return;
+
+	drm_printf(p, "num_instances:%d\n", adev->sdma.num_instances);
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		instance_offset = i * reg_count;
+		drm_printf(p, "\nInstance:%d\n", i);
+
+		for (j = 0; j < reg_count; j++)
+			drm_printf(p, "%-50s \t 0x%08x\n", sdma_reg_list_5_0[j].reg_name,
+				   adev->sdma.ip_dump[instance_offset + j]);
+	}
+}
+
+static void sdma_v5_0_dump_ip_state(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, j;
+	uint32_t instance_offset;
+	uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_5_0);
+
+	if (!adev->sdma.ip_dump)
+		return;
+
+	amdgpu_gfx_off_ctrl(adev, false);
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		instance_offset = i * reg_count;
+		for (j = 0; j < reg_count; j++)
+			adev->sdma.ip_dump[instance_offset + j] =
+				RREG32(sdma_v5_0_get_reg_offset(adev, i,
+				       sdma_reg_list_5_0[j].reg_offset));
+	}
+	amdgpu_gfx_off_ctrl(adev, true);
+}
+
+static const struct amd_ip_funcs sdma_v5_0_ip_funcs = {
+	.name = "sdma_v5_0",
+	.early_init = sdma_v5_0_early_init,
+	.sw_init = sdma_v5_0_sw_init,
+	.sw_fini = sdma_v5_0_sw_fini,
+	.hw_init = sdma_v5_0_hw_init,
+	.hw_fini = sdma_v5_0_hw_fini,
+	.suspend = sdma_v5_0_suspend,
+	.resume = sdma_v5_0_resume,
+	.is_idle = sdma_v5_0_is_idle,
+	.wait_for_idle = sdma_v5_0_wait_for_idle,
+	.soft_reset = sdma_v5_0_soft_reset,
+	.set_clockgating_state = sdma_v5_0_set_clockgating_state,
+	.set_powergating_state = sdma_v5_0_set_powergating_state,
+	.get_clockgating_state = sdma_v5_0_get_clockgating_state,
+	.dump_ip_state = sdma_v5_0_dump_ip_state,
+	.print_ip_state = sdma_v5_0_print_ip_state,
+};
+
+static const struct amdgpu_ring_funcs sdma_v5_0_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_SDMA,
+	.align_mask = 0xf,
+	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
+	.support_64bit_ptrs = true,
+	.secure_submission_supported = true,
+	.get_rptr = sdma_v5_0_ring_get_rptr,
+	.get_wptr = sdma_v5_0_ring_get_wptr,
+	.set_wptr = sdma_v5_0_ring_set_wptr,
+	.emit_frame_size =
+		5 + /* sdma_v5_0_ring_init_cond_exec */
+		6 + /* sdma_v5_0_ring_emit_hdp_flush */
+		3 + /* hdp_invalidate */
+		6 + /* sdma_v5_0_ring_emit_pipeline_sync */
+		/* sdma_v5_0_ring_emit_vm_flush */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 6 * 2 +
+		10 + 10 + 10, /* sdma_v5_0_ring_emit_fence x3 for user fence, vm fence */
+	.emit_ib_size = 5 + 7 + 6, /* sdma_v5_0_ring_emit_ib */
+	.emit_ib = sdma_v5_0_ring_emit_ib,
+	.emit_mem_sync = sdma_v5_0_ring_emit_mem_sync,
+	.emit_fence = sdma_v5_0_ring_emit_fence,
+	.emit_pipeline_sync = sdma_v5_0_ring_emit_pipeline_sync,
+	.emit_vm_flush = sdma_v5_0_ring_emit_vm_flush,
+	.emit_hdp_flush = sdma_v5_0_ring_emit_hdp_flush,
+	.test_ring = sdma_v5_0_ring_test_ring,
+	.test_ib = sdma_v5_0_ring_test_ib,
+	.insert_nop = sdma_v5_0_ring_insert_nop,
+	.pad_ib = sdma_v5_0_ring_pad_ib,
+	.emit_wreg = sdma_v5_0_ring_emit_wreg,
+	.emit_reg_wait = sdma_v5_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = sdma_v5_0_ring_emit_reg_write_reg_wait,
+	.init_cond_exec = sdma_v5_0_ring_init_cond_exec,
+	.preempt_ib = sdma_v5_0_ring_preempt_ib,
+	.reset = sdma_v5_0_reset_queue,
+};
+
+static void sdma_v5_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		adev->sdma.instance[i].ring.funcs = &sdma_v5_0_ring_funcs;
+		adev->sdma.instance[i].ring.me = i;
+	}
+}
+
+static const struct amdgpu_irq_src_funcs sdma_v5_0_trap_irq_funcs = {
+	.set = sdma_v5_0_set_trap_irq_state,
+	.process = sdma_v5_0_process_trap_irq,
+};
+
+static const struct amdgpu_irq_src_funcs sdma_v5_0_illegal_inst_irq_funcs = {
+	.process = sdma_v5_0_process_illegal_inst_irq,
+};
+
+static void sdma_v5_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->sdma.trap_irq.num_types = AMDGPU_SDMA_IRQ_INSTANCE0 +
+					adev->sdma.num_instances;
+	adev->sdma.trap_irq.funcs = &sdma_v5_0_trap_irq_funcs;
+	adev->sdma.illegal_inst_irq.funcs = &sdma_v5_0_illegal_inst_irq_funcs;
+}
+
+/**
+ * sdma_v5_0_emit_copy_buffer - copy buffer using the sDMA engine
+ *
+ * @ib: indirect buffer to copy to
+ * @src_offset: src GPU address
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ * @copy_flags: copy flags for the buffers
+ *
+ * Copy GPU buffers using the DMA engine (NAVI10).
+ * Used by the amdgpu ttm implementation to move pages if
+ * registered as the asic copy callback.
+ */
+static void sdma_v5_0_emit_copy_buffer(struct amdgpu_ib *ib,
+				       uint64_t src_offset,
+				       uint64_t dst_offset,
+				       uint32_t byte_count,
+				       uint32_t copy_flags)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR) |
+		SDMA_PKT_COPY_LINEAR_HEADER_TMZ((copy_flags & AMDGPU_COPY_FLAGS_TMZ) ? 1 : 0);
+	ib->ptr[ib->length_dw++] = byte_count - 1;
+	ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+	ib->ptr[ib->length_dw++] = lower_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+}
+
+/**
+ * sdma_v5_0_emit_fill_buffer - fill buffer using the sDMA engine
+ *
+ * @ib: indirect buffer to fill
+ * @src_data: value to write to buffer
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ *
+ * Fill GPU buffers using the DMA engine (NAVI10).
+ */
+static void sdma_v5_0_emit_fill_buffer(struct amdgpu_ib *ib,
+				       uint32_t src_data,
+				       uint64_t dst_offset,
+				       uint32_t byte_count)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_CONST_FILL);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = src_data;
+	ib->ptr[ib->length_dw++] = byte_count - 1;
+}
+
+static const struct amdgpu_buffer_funcs sdma_v5_0_buffer_funcs = {
+	.copy_max_bytes = 0x400000,
+	.copy_num_dw = 7,
+	.emit_copy_buffer = sdma_v5_0_emit_copy_buffer,
+
+	.fill_max_bytes = 0x400000,
+	.fill_num_dw = 5,
+	.emit_fill_buffer = sdma_v5_0_emit_fill_buffer,
+};
+
+static void sdma_v5_0_set_buffer_funcs(struct amdgpu_device *adev)
+{
+	if (adev->mman.buffer_funcs == NULL) {
+		adev->mman.buffer_funcs = &sdma_v5_0_buffer_funcs;
+		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
+	}
+}
+
+static const struct amdgpu_vm_pte_funcs sdma_v5_0_vm_pte_funcs = {
+	.copy_pte_num_dw = 7,
+	.copy_pte = sdma_v5_0_vm_copy_pte,
+	.write_pte = sdma_v5_0_vm_write_pte,
+	.set_pte_pde = sdma_v5_0_vm_set_pte_pde,
+};
+
+static void sdma_v5_0_set_vm_pte_funcs(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	if (adev->vm_manager.vm_pte_funcs == NULL) {
+		adev->vm_manager.vm_pte_funcs = &sdma_v5_0_vm_pte_funcs;
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			adev->vm_manager.vm_pte_scheds[i] =
+				&adev->sdma.instance[i].ring.sched;
+		}
+		adev->vm_manager.vm_pte_num_scheds = adev->sdma.num_instances;
+	}
+}
+
+const struct amdgpu_ip_block_version sdma_v5_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_SDMA,
+	.major = 5,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &sdma_v5_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.h b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.h
new file mode 100644
index 000000000000..2ab71f21755a
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __SDMA_V5_0_H__
+#define __SDMA_V5_0_H__
+
+extern const struct amdgpu_ip_block_version sdma_v5_0_ip_block;
+
+#endif /* __SDMA_V5_0_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c
new file mode 100644
index 000000000000..3bd44c24f692
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c
@@ -0,0 +1,2112 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/delay.h>
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+
+#include "amdgpu.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_trace.h"
+
+#include "gc/gc_10_3_0_offset.h"
+#include "gc/gc_10_3_0_sh_mask.h"
+#include "ivsrcid/sdma0/irqsrcs_sdma0_5_0.h"
+#include "ivsrcid/sdma1/irqsrcs_sdma1_5_0.h"
+#include "ivsrcid/sdma2/irqsrcs_sdma2_5_0.h"
+#include "ivsrcid/sdma3/irqsrcs_sdma3_5_0.h"
+
+#include "soc15_common.h"
+#include "soc15.h"
+#include "navi10_sdma_pkt_open.h"
+#include "nbio_v2_3.h"
+#include "sdma_common.h"
+#include "sdma_v5_2.h"
+
+MODULE_FIRMWARE("amdgpu/sienna_cichlid_sdma.bin");
+MODULE_FIRMWARE("amdgpu/navy_flounder_sdma.bin");
+MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_sdma.bin");
+MODULE_FIRMWARE("amdgpu/beige_goby_sdma.bin");
+
+MODULE_FIRMWARE("amdgpu/vangogh_sdma.bin");
+MODULE_FIRMWARE("amdgpu/yellow_carp_sdma.bin");
+MODULE_FIRMWARE("amdgpu/sdma_5_2_6.bin");
+MODULE_FIRMWARE("amdgpu/sdma_5_2_7.bin");
+
+#define SDMA1_REG_OFFSET 0x600
+#define SDMA3_REG_OFFSET 0x400
+#define SDMA0_HYP_DEC_REG_START 0x5880
+#define SDMA0_HYP_DEC_REG_END 0x5893
+#define SDMA1_HYP_DEC_REG_OFFSET 0x20
+
+static const struct amdgpu_hwip_reg_entry sdma_reg_list_5_2[] = {
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_STATUS_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_STATUS1_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_STATUS2_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_STATUS3_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_UCODE_CHECKSUM),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RB_RPTR_FETCH_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RB_RPTR_FETCH),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_UTCL1_RD_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_UTCL1_WR_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_UTCL1_RD_XNACK0),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_UTCL1_RD_XNACK1),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_UTCL1_WR_XNACK0),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_UTCL1_WR_XNACK1),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_RB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_RB_RPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_RB_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_RB_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_IB_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_IB_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_IB_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_IB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_IB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_IB_SUB_REMAIN),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_GFX_DUMMY_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_RB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_RB_RPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_RB_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_RB_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_IB_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_IB_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_IB_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_PAGE_DUMMY_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_RB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_RB_RPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_RB_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_RB_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_IB_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_IB_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_IB_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_RLC0_DUMMY_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_INT_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmSDMA0_VM_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS2)
+};
+
+static void sdma_v5_2_set_ring_funcs(struct amdgpu_device *adev);
+static void sdma_v5_2_set_buffer_funcs(struct amdgpu_device *adev);
+static void sdma_v5_2_set_vm_pte_funcs(struct amdgpu_device *adev);
+static void sdma_v5_2_set_irq_funcs(struct amdgpu_device *adev);
+static int sdma_v5_2_stop_queue(struct amdgpu_ring *ring);
+static int sdma_v5_2_restore_queue(struct amdgpu_ring *ring);
+
+static u32 sdma_v5_2_get_reg_offset(struct amdgpu_device *adev, u32 instance, u32 internal_offset)
+{
+	u32 base;
+
+	if (internal_offset >= SDMA0_HYP_DEC_REG_START &&
+	    internal_offset <= SDMA0_HYP_DEC_REG_END) {
+		base = adev->reg_offset[GC_HWIP][0][1];
+		if (instance != 0)
+			internal_offset += SDMA1_HYP_DEC_REG_OFFSET * instance;
+	} else {
+		if (instance < 2) {
+			base = adev->reg_offset[GC_HWIP][0][0];
+			if (instance == 1)
+				internal_offset += SDMA1_REG_OFFSET;
+		} else {
+			base = adev->reg_offset[GC_HWIP][0][2];
+			if (instance == 3)
+				internal_offset += SDMA3_REG_OFFSET;
+		}
+	}
+
+	return base + internal_offset;
+}
+
+static unsigned sdma_v5_2_ring_init_cond_exec(struct amdgpu_ring *ring,
+					      uint64_t addr)
+{
+	unsigned ret;
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_COND_EXE));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, 1);
+	/* this is the offset we need patch later */
+	ret = ring->wptr & ring->buf_mask;
+	/* insert dummy here and patch it later */
+	amdgpu_ring_write(ring, 0);
+
+	return ret;
+}
+
+/**
+ * sdma_v5_2_ring_get_rptr - get the current read pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current rptr from the hardware (NAVI10+).
+ */
+static uint64_t sdma_v5_2_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	u64 *rptr;
+
+	/* XXX check if swapping is necessary on BE */
+	rptr = (u64 *)ring->rptr_cpu_addr;
+
+	DRM_DEBUG("rptr before shift == 0x%016llx\n", *rptr);
+	return ((*rptr) >> 2);
+}
+
+/**
+ * sdma_v5_2_ring_get_wptr - get the current write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current wptr from the hardware (NAVI10+).
+ */
+static uint64_t sdma_v5_2_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u64 wptr;
+
+	if (ring->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		wptr = READ_ONCE(*((u64 *)ring->wptr_cpu_addr));
+		DRM_DEBUG("wptr/doorbell before shift == 0x%016llx\n", wptr);
+	} else {
+		wptr = RREG32(sdma_v5_2_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI));
+		wptr = wptr << 32;
+		wptr |= RREG32(sdma_v5_2_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR));
+		DRM_DEBUG("wptr before shift [%i] wptr == 0x%016llx\n", ring->me, wptr);
+	}
+
+	return wptr >> 2;
+}
+
+/**
+ * sdma_v5_2_ring_set_wptr - commit the write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Write the wptr back to the hardware (NAVI10+).
+ */
+static void sdma_v5_2_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	DRM_DEBUG("Setting write pointer\n");
+	if (ring->use_doorbell) {
+		DRM_DEBUG("Using doorbell -- "
+				"wptr_offs == 0x%08x "
+				"lower_32_bits(ring->wptr << 2) == 0x%08x "
+				"upper_32_bits(ring->wptr << 2) == 0x%08x\n",
+				ring->wptr_offs,
+				lower_32_bits(ring->wptr << 2),
+				upper_32_bits(ring->wptr << 2));
+		/* XXX check if swapping is necessary on BE */
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr,
+			     ring->wptr << 2);
+		DRM_DEBUG("calling WDOORBELL64(0x%08x, 0x%016llx)\n",
+				ring->doorbell_index, ring->wptr << 2);
+		WDOORBELL64(ring->doorbell_index, ring->wptr << 2);
+		if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) == IP_VERSION(5, 2, 1)) {
+			/* SDMA seems to miss doorbells sometimes when powergating kicks in.
+			 * Updating the wptr directly will wake it. This is only safe because
+			 * we disallow gfxoff in begin_use() and then allow it again in end_use().
+			 */
+			WREG32(sdma_v5_2_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR),
+			       lower_32_bits(ring->wptr << 2));
+			WREG32(sdma_v5_2_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI),
+			       upper_32_bits(ring->wptr << 2));
+		}
+	} else {
+		DRM_DEBUG("Not using doorbell -- "
+				"mmSDMA%i_GFX_RB_WPTR == 0x%08x "
+				"mmSDMA%i_GFX_RB_WPTR_HI == 0x%08x\n",
+				ring->me,
+				lower_32_bits(ring->wptr << 2),
+				ring->me,
+				upper_32_bits(ring->wptr << 2));
+		WREG32(sdma_v5_2_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR),
+			lower_32_bits(ring->wptr << 2));
+		WREG32(sdma_v5_2_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI),
+			upper_32_bits(ring->wptr << 2));
+	}
+}
+
+static void sdma_v5_2_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring);
+	int i;
+
+	for (i = 0; i < count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			amdgpu_ring_write(ring, ring->funcs->nop |
+				SDMA_PKT_NOP_HEADER_COUNT(count - 1));
+		else
+			amdgpu_ring_write(ring, ring->funcs->nop);
+}
+
+/**
+ * sdma_v5_2_ring_emit_ib - Schedule an IB on the DMA engine
+ *
+ * @ring: amdgpu ring pointer
+ * @job: job to retrieve vmid from
+ * @ib: IB object to schedule
+ * @flags: unused
+ *
+ * Schedule an IB in the DMA ring.
+ */
+static void sdma_v5_2_ring_emit_ib(struct amdgpu_ring *ring,
+				   struct amdgpu_job *job,
+				   struct amdgpu_ib *ib,
+				   uint32_t flags)
+{
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+	uint64_t csa_mc_addr = amdgpu_sdma_get_csa_mc_addr(ring, vmid);
+
+	/* An IB packet must end on a 8 DW boundary--the next dword
+	 * must be on a 8-dword boundary. Our IB packet below is 6
+	 * dwords long, thus add x number of NOPs, such that, in
+	 * modular arithmetic,
+	 * wptr + 6 + x = 8k, k >= 0, which in C is,
+	 * (wptr + 6 + x) % 8 = 0.
+	 * The expression below, is a solution of x.
+	 */
+	sdma_v5_2_ring_insert_nop(ring, (2 - lower_32_bits(ring->wptr)) & 7);
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_INDIRECT) |
+			  SDMA_PKT_INDIRECT_HEADER_VMID(vmid & 0xf));
+	/* base must be 32 byte aligned */
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr) & 0xffffffe0);
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, ib->length_dw);
+	amdgpu_ring_write(ring, lower_32_bits(csa_mc_addr));
+	amdgpu_ring_write(ring, upper_32_bits(csa_mc_addr));
+}
+
+/**
+ * sdma_v5_2_ring_emit_mem_sync - flush the IB by graphics cache rinse
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * flush the IB by graphics cache rinse.
+ */
+static void sdma_v5_2_ring_emit_mem_sync(struct amdgpu_ring *ring)
+{
+	uint32_t gcr_cntl = SDMA_GCR_GL2_INV | SDMA_GCR_GL2_WB |
+			    SDMA_GCR_GLM_INV | SDMA_GCR_GL1_INV |
+			    SDMA_GCR_GLV_INV | SDMA_GCR_GLK_INV |
+			    SDMA_GCR_GLI_INV(1);
+
+	/* flush entire cache L0/L1/L2, this can be optimized by performance requirement */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_GCR_REQ));
+	amdgpu_ring_write(ring, SDMA_PKT_GCR_REQ_PAYLOAD1_BASE_VA_31_7(0));
+	amdgpu_ring_write(ring, SDMA_PKT_GCR_REQ_PAYLOAD2_GCR_CONTROL_15_0(gcr_cntl) |
+			SDMA_PKT_GCR_REQ_PAYLOAD2_BASE_VA_47_32(0));
+	amdgpu_ring_write(ring, SDMA_PKT_GCR_REQ_PAYLOAD3_LIMIT_VA_31_7(0) |
+			SDMA_PKT_GCR_REQ_PAYLOAD3_GCR_CONTROL_18_16(gcr_cntl >> 16));
+	amdgpu_ring_write(ring, SDMA_PKT_GCR_REQ_PAYLOAD4_LIMIT_VA_47_32(0) |
+			SDMA_PKT_GCR_REQ_PAYLOAD4_VMID(0));
+}
+
+/**
+ * sdma_v5_2_ring_emit_hdp_flush - emit an hdp flush on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Emit an hdp flush packet on the requested DMA ring.
+ */
+static void sdma_v5_2_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 ref_and_mask = 0;
+	const struct nbio_hdp_flush_reg *nbio_hf_reg = adev->nbio.hdp_flush_reg;
+
+	if (ring->me > 1) {
+		amdgpu_asic_flush_hdp(adev, ring);
+	} else {
+		ref_and_mask = nbio_hf_reg->ref_and_mask_sdma0 << ring->me;
+
+		amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+				  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(1) |
+				  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3)); /* == */
+		amdgpu_ring_write(ring, (adev->nbio.funcs->get_hdp_flush_done_offset(adev)) << 2);
+		amdgpu_ring_write(ring, (adev->nbio.funcs->get_hdp_flush_req_offset(adev)) << 2);
+		amdgpu_ring_write(ring, ref_and_mask); /* reference */
+		amdgpu_ring_write(ring, ref_and_mask); /* mask */
+		amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+				  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
+	}
+}
+
+/**
+ * sdma_v5_2_ring_emit_fence - emit a fence on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
+ *
+ * Add a DMA fence packet to the ring to write
+ * the fence seq number and DMA trap packet to generate
+ * an interrupt if needed.
+ */
+static void sdma_v5_2_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				      unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	/* write the fence */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_FENCE) |
+			  SDMA_PKT_FENCE_HEADER_MTYPE(0x3)); /* Ucached(UC) */
+	/* zero in first two bits */
+	BUG_ON(addr & 0x3);
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+
+	/* optionally write high bits as well */
+	if (write64bit) {
+		addr += 4;
+		amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_FENCE) |
+				  SDMA_PKT_FENCE_HEADER_MTYPE(0x3));
+		/* zero in first two bits */
+		BUG_ON(addr & 0x3);
+		amdgpu_ring_write(ring, lower_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(seq));
+	}
+
+	if ((flags & AMDGPU_FENCE_FLAG_INT)) {
+		/* generate an interrupt */
+		amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_TRAP));
+		amdgpu_ring_write(ring, SDMA_PKT_TRAP_INT_CONTEXT_INT_CONTEXT(0));
+	}
+}
+
+
+/**
+ * sdma_v5_2_gfx_stop - stop the gfx async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ * @inst_mask: mask of dma engine instances to be disabled
+ * Stop the gfx async dma ring buffers.
+ */
+static void sdma_v5_2_gfx_stop(struct amdgpu_device *adev,  uint32_t inst_mask)
+{
+	u32 rb_cntl, ib_cntl;
+	int i;
+
+	for_each_inst(i, inst_mask) {
+		rb_cntl = RREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL));
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 0);
+		WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
+		ib_cntl = RREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL));
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 0);
+		WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL), ib_cntl);
+	}
+}
+
+/**
+ * sdma_v5_2_rlc_stop - stop the compute async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the compute async dma queues.
+ */
+static void sdma_v5_2_rlc_stop(struct amdgpu_device *adev)
+{
+	/* XXX todo */
+}
+
+/**
+ * sdma_v5_2_ctx_switch_enable - stop the async dma engines context switch
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable the DMA MEs context switch.
+ *
+ * Halt or unhalt the async dma engines context switch.
+ */
+static void sdma_v5_2_ctx_switch_enable(struct amdgpu_device *adev, bool enable)
+{
+	u32 f32_cntl, phase_quantum = 0;
+	int i;
+
+	if (amdgpu_sdma_phase_quantum) {
+		unsigned value = amdgpu_sdma_phase_quantum;
+		unsigned unit = 0;
+
+		while (value > (SDMA0_PHASE0_QUANTUM__VALUE_MASK >>
+				SDMA0_PHASE0_QUANTUM__VALUE__SHIFT)) {
+			value = (value + 1) >> 1;
+			unit++;
+		}
+		if (unit > (SDMA0_PHASE0_QUANTUM__UNIT_MASK >>
+			    SDMA0_PHASE0_QUANTUM__UNIT__SHIFT)) {
+			value = (SDMA0_PHASE0_QUANTUM__VALUE_MASK >>
+				 SDMA0_PHASE0_QUANTUM__VALUE__SHIFT);
+			unit = (SDMA0_PHASE0_QUANTUM__UNIT_MASK >>
+				SDMA0_PHASE0_QUANTUM__UNIT__SHIFT);
+			WARN_ONCE(1,
+			"clamping sdma_phase_quantum to %uK clock cycles\n",
+				  value << unit);
+		}
+		phase_quantum =
+			value << SDMA0_PHASE0_QUANTUM__VALUE__SHIFT |
+			unit  << SDMA0_PHASE0_QUANTUM__UNIT__SHIFT;
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (enable && amdgpu_sdma_phase_quantum) {
+			WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_PHASE0_QUANTUM),
+			       phase_quantum);
+			WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_PHASE1_QUANTUM),
+			       phase_quantum);
+			WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_PHASE2_QUANTUM),
+			       phase_quantum);
+		}
+
+		if (!amdgpu_sriov_vf(adev)) {
+			f32_cntl = RREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_CNTL));
+			f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL,
+					AUTO_CTXSW_ENABLE, enable ? 1 : 0);
+			WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_CNTL), f32_cntl);
+		}
+	}
+
+}
+
+/**
+ * sdma_v5_2_enable - stop the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable the DMA MEs.
+ *
+ * Halt or unhalt the async dma engines.
+ */
+static void sdma_v5_2_enable(struct amdgpu_device *adev, bool enable)
+{
+	u32 f32_cntl;
+	int i;
+	uint32_t inst_mask;
+
+	inst_mask = GENMASK(adev->sdma.num_instances - 1, 0);
+	if (!enable) {
+		sdma_v5_2_gfx_stop(adev, inst_mask);
+		sdma_v5_2_rlc_stop(adev);
+	}
+
+	if (!amdgpu_sriov_vf(adev)) {
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			f32_cntl = RREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_F32_CNTL));
+			f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_F32_CNTL, HALT, enable ? 0 : 1);
+			WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_F32_CNTL), f32_cntl);
+		}
+	}
+}
+
+/**
+ * sdma_v5_2_gfx_resume_instance - start/restart a certain sdma engine
+ *
+ * @adev: amdgpu_device pointer
+ * @i: instance
+ * @restore: used to restore wptr when restart
+ *
+ * Set up the gfx DMA ring buffers and enable them. On restart, we will restore wptr and rptr.
+ * Return 0 for success.
+ */
+
+static int sdma_v5_2_gfx_resume_instance(struct amdgpu_device *adev, int i, bool restore)
+{
+	struct amdgpu_ring *ring;
+	u32 rb_cntl, ib_cntl;
+	u32 rb_bufsz;
+	u32 doorbell;
+	u32 doorbell_offset;
+	u32 temp;
+	u32 wptr_poll_cntl;
+	u64 wptr_gpu_addr;
+
+	ring = &adev->sdma.instance[i].ring;
+
+	if (!amdgpu_sriov_vf(adev))
+		WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_SEM_WAIT_FAIL_TIMER_CNTL), 0);
+
+	/* Set ring buffer size in dwords */
+	rb_bufsz = order_base_2(ring->ring_size / 4);
+	rb_cntl = RREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL));
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SIZE, rb_bufsz);
+#ifdef __BIG_ENDIAN
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SWAP_ENABLE, 1);
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL,
+				RPTR_WRITEBACK_SWAP_ENABLE, 1);
+#endif
+	WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
+
+	/* Initialize the ring buffer's read and write pointers */
+	if (restore) {
+		WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR), lower_32_bits(ring->wptr << 2));
+		WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_HI), upper_32_bits(ring->wptr << 2));
+		WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR), lower_32_bits(ring->wptr << 2));
+		WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_HI), upper_32_bits(ring->wptr << 2));
+	} else {
+		WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR), 0);
+		WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_HI), 0);
+		WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR), 0);
+		WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_HI), 0);
+	}
+
+	/* setup the wptr shadow polling */
+	wptr_gpu_addr = ring->wptr_gpu_addr;
+	WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_LO),
+	       lower_32_bits(wptr_gpu_addr));
+	WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_HI),
+	       upper_32_bits(wptr_gpu_addr));
+	wptr_poll_cntl = RREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i,
+						 mmSDMA0_GFX_RB_WPTR_POLL_CNTL));
+	wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl,
+				       SDMA0_GFX_RB_WPTR_POLL_CNTL,
+				       F32_POLL_ENABLE, 1);
+	WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL),
+	       wptr_poll_cntl);
+
+	/* set the wb address whether it's enabled or not */
+	WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_ADDR_HI),
+	       upper_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFF);
+	WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_ADDR_LO),
+	       lower_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFC);
+
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RPTR_WRITEBACK_ENABLE, 1);
+
+	WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_BASE), ring->gpu_addr >> 8);
+	WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_BASE_HI), ring->gpu_addr >> 40);
+
+	if (!restore)
+		ring->wptr = 0;
+
+	/* before programing wptr to a less value, need set minor_ptr_update first */
+	WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_MINOR_PTR_UPDATE), 1);
+
+	if (!amdgpu_sriov_vf(adev)) { /* only bare-metal use register write for wptr */
+		WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR), lower_32_bits(ring->wptr << 2));
+		WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_HI), upper_32_bits(ring->wptr << 2));
+	}
+
+	doorbell = RREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL));
+	doorbell_offset = RREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL_OFFSET));
+
+	if (ring->use_doorbell) {
+		doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, 1);
+		doorbell_offset = REG_SET_FIELD(doorbell_offset, SDMA0_GFX_DOORBELL_OFFSET,
+				OFFSET, ring->doorbell_index);
+	} else {
+		doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, 0);
+	}
+	WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL), doorbell);
+	WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL_OFFSET), doorbell_offset);
+
+	adev->nbio.funcs->sdma_doorbell_range(adev, i, ring->use_doorbell,
+					      ring->doorbell_index,
+					      adev->doorbell_index.sdma_doorbell_range);
+
+	if (amdgpu_sriov_vf(adev))
+		sdma_v5_2_ring_set_wptr(ring);
+
+	/* set minor_ptr_update to 0 after wptr programed */
+
+	WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_MINOR_PTR_UPDATE), 0);
+
+	/* SRIOV VF has no control of any of registers below */
+	if (!amdgpu_sriov_vf(adev)) {
+		/* set utc l1 enable flag always to 1 */
+		temp = RREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_CNTL));
+		temp = REG_SET_FIELD(temp, SDMA0_CNTL, UTC_L1_ENABLE, 1);
+
+		/* enable MCBP */
+		temp = REG_SET_FIELD(temp, SDMA0_CNTL, MIDCMD_PREEMPT_ENABLE, 1);
+		WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_CNTL), temp);
+
+		/* Set up RESP_MODE to non-copy addresses */
+		temp = RREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_UTCL1_CNTL));
+		temp = REG_SET_FIELD(temp, SDMA0_UTCL1_CNTL, RESP_MODE, 3);
+		temp = REG_SET_FIELD(temp, SDMA0_UTCL1_CNTL, REDO_DELAY, 9);
+		WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_UTCL1_CNTL), temp);
+
+		/* program default cache read and write policy */
+		temp = RREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_UTCL1_PAGE));
+		/* clean read policy and write policy bits */
+		temp &= 0xFF0FFF;
+		temp |= ((CACHE_READ_POLICY_L2__DEFAULT << 12) |
+			 (CACHE_WRITE_POLICY_L2__DEFAULT << 14) |
+			 SDMA0_UTCL1_PAGE__LLC_NOALLOC_MASK);
+		WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_UTCL1_PAGE), temp);
+
+		/* unhalt engine */
+		temp = RREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_F32_CNTL));
+		temp = REG_SET_FIELD(temp, SDMA0_F32_CNTL, HALT, 0);
+		WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_F32_CNTL), temp);
+	}
+
+	/* enable DMA RB */
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 1);
+	WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
+
+	ib_cntl = RREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL));
+	ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 1);
+#ifdef __BIG_ENDIAN
+	ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_SWAP_ENABLE, 1);
+#endif
+	/* enable DMA IBs */
+	WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL), ib_cntl);
+
+	if (amdgpu_sriov_vf(adev)) { /* bare-metal sequence doesn't need below to lines */
+		sdma_v5_2_ctx_switch_enable(adev, true);
+		sdma_v5_2_enable(adev, true);
+	}
+
+	return amdgpu_ring_test_helper(ring);
+}
+
+/**
+ * sdma_v5_2_gfx_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the gfx DMA ring buffers and enable them.
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v5_2_gfx_resume(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		r = sdma_v5_2_gfx_resume_instance(adev, i, false);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+/**
+ * sdma_v5_2_rlc_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the compute DMA queues and enable them.
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v5_2_rlc_resume(struct amdgpu_device *adev)
+{
+	return 0;
+}
+
+/**
+ * sdma_v5_2_load_microcode - load the sDMA ME ucode
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Loads the sDMA0/1/2/3 ucode.
+ * Returns 0 for success, -EINVAL if the ucode is not available.
+ */
+static int sdma_v5_2_load_microcode(struct amdgpu_device *adev)
+{
+	const struct sdma_firmware_header_v1_0 *hdr;
+	const __le32 *fw_data;
+	u32 fw_size;
+	int i, j;
+
+	/* halt the MEs */
+	sdma_v5_2_enable(adev, false);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (!adev->sdma.instance[i].fw)
+			return -EINVAL;
+
+		hdr = (const struct sdma_firmware_header_v1_0 *)adev->sdma.instance[i].fw->data;
+		amdgpu_ucode_print_sdma_hdr(&hdr->header);
+		fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+
+		fw_data = (const __le32 *)
+			(adev->sdma.instance[i].fw->data +
+				le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+
+		WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_UCODE_ADDR), 0);
+
+		for (j = 0; j < fw_size; j++) {
+			if (amdgpu_emu_mode == 1 && j % 500 == 0)
+				msleep(1);
+			WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_UCODE_DATA), le32_to_cpup(fw_data++));
+		}
+
+		WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_UCODE_ADDR), adev->sdma.instance[i].fw_version);
+	}
+
+	return 0;
+}
+
+static int sdma_v5_2_soft_reset_engine(struct amdgpu_device *adev, u32 instance_id)
+{
+	u32 grbm_soft_reset;
+	u32 tmp;
+
+	grbm_soft_reset = REG_SET_FIELD(0,
+					GRBM_SOFT_RESET, SOFT_RESET_SDMA0,
+					1);
+	grbm_soft_reset <<= instance_id;
+
+	tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
+	tmp |= grbm_soft_reset;
+	DRM_DEBUG("GRBM_SOFT_RESET=0x%08X\n", tmp);
+	WREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET, tmp);
+	tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
+
+	udelay(50);
+
+	tmp &= ~grbm_soft_reset;
+	WREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET, tmp);
+	tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
+	return 0;
+}
+
+static int sdma_v5_2_soft_reset(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		sdma_v5_2_soft_reset_engine(adev, i);
+		udelay(50);
+	}
+
+	return 0;
+}
+
+static const struct amdgpu_sdma_funcs sdma_v5_2_sdma_funcs = {
+	.stop_kernel_queue = &sdma_v5_2_stop_queue,
+	.start_kernel_queue = &sdma_v5_2_restore_queue,
+	.soft_reset_kernel_queue = &sdma_v5_2_soft_reset_engine,
+};
+
+/**
+ * sdma_v5_2_start - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the DMA engines and enable them.
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v5_2_start(struct amdgpu_device *adev)
+{
+	int r = 0;
+	struct amdgpu_ip_block *ip_block;
+
+	if (amdgpu_sriov_vf(adev)) {
+		sdma_v5_2_ctx_switch_enable(adev, false);
+		sdma_v5_2_enable(adev, false);
+
+		/* set RB registers */
+		r = sdma_v5_2_gfx_resume(adev);
+		return r;
+	}
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+		r = sdma_v5_2_load_microcode(adev);
+		if (r)
+			return r;
+
+		/* The value of mmSDMA_F32_CNTL is invalid the moment after loading fw */
+		if (amdgpu_emu_mode == 1)
+			msleep(1000);
+	}
+
+	ip_block = amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_SDMA);
+	if (!ip_block)
+		return -EINVAL;
+
+	sdma_v5_2_soft_reset(ip_block);
+	/* unhalt the MEs */
+	sdma_v5_2_enable(adev, true);
+	/* enable sdma ring preemption */
+	sdma_v5_2_ctx_switch_enable(adev, true);
+
+	/* start the gfx rings and rlc compute queues */
+	r = sdma_v5_2_gfx_resume(adev);
+	if (r)
+		return r;
+	r = sdma_v5_2_rlc_resume(adev);
+
+	return r;
+}
+
+static int sdma_v5_2_mqd_init(struct amdgpu_device *adev, void *mqd,
+			      struct amdgpu_mqd_prop *prop)
+{
+	struct v10_sdma_mqd *m = mqd;
+	uint64_t wb_gpu_addr;
+
+	m->sdmax_rlcx_rb_cntl =
+		order_base_2(prop->queue_size / 4) << SDMA0_RLC0_RB_CNTL__RB_SIZE__SHIFT |
+		1 << SDMA0_RLC0_RB_CNTL__RPTR_WRITEBACK_ENABLE__SHIFT |
+		6 << SDMA0_RLC0_RB_CNTL__RPTR_WRITEBACK_TIMER__SHIFT |
+		1 << SDMA0_RLC0_RB_CNTL__RB_PRIV__SHIFT;
+
+	m->sdmax_rlcx_rb_base = lower_32_bits(prop->hqd_base_gpu_addr >> 8);
+	m->sdmax_rlcx_rb_base_hi = upper_32_bits(prop->hqd_base_gpu_addr >> 8);
+
+	m->sdmax_rlcx_rb_wptr_poll_cntl = RREG32(sdma_v5_2_get_reg_offset(adev, 0,
+						  mmSDMA0_GFX_RB_WPTR_POLL_CNTL));
+
+	wb_gpu_addr = prop->wptr_gpu_addr;
+	m->sdmax_rlcx_rb_wptr_poll_addr_lo = lower_32_bits(wb_gpu_addr);
+	m->sdmax_rlcx_rb_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr);
+
+	wb_gpu_addr = prop->rptr_gpu_addr;
+	m->sdmax_rlcx_rb_rptr_addr_lo = lower_32_bits(wb_gpu_addr);
+	m->sdmax_rlcx_rb_rptr_addr_hi = upper_32_bits(wb_gpu_addr);
+
+	m->sdmax_rlcx_ib_cntl = RREG32(sdma_v5_2_get_reg_offset(adev, 0,
+							mmSDMA0_GFX_IB_CNTL));
+
+	m->sdmax_rlcx_doorbell_offset =
+		prop->doorbell_index << SDMA0_RLC0_DOORBELL_OFFSET__OFFSET__SHIFT;
+
+	m->sdmax_rlcx_doorbell = REG_SET_FIELD(0, SDMA0_RLC0_DOORBELL, ENABLE, 1);
+
+	return 0;
+}
+
+static void sdma_v5_2_set_mqd_funcs(struct amdgpu_device *adev)
+{
+	adev->mqds[AMDGPU_HW_IP_DMA].mqd_size = sizeof(struct v10_sdma_mqd);
+	adev->mqds[AMDGPU_HW_IP_DMA].init_mqd = sdma_v5_2_mqd_init;
+}
+
+/**
+ * sdma_v5_2_ring_test_ring - simple async dma engine test
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Test the DMA engine by writing using it to write an
+ * value to memory.
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v5_2_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned i;
+	unsigned index;
+	int r;
+	u32 tmp;
+	u64 gpu_addr;
+
+	tmp = 0xCAFEDEAD;
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+
+	r = amdgpu_ring_alloc(ring, 20);
+	if (r) {
+		DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r);
+		amdgpu_device_wb_free(adev, index);
+		return r;
+	}
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+			  SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR));
+	amdgpu_ring_write(ring, lower_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(0));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = le32_to_cpu(adev->wb.wb[index]);
+		if (tmp == 0xDEADBEEF)
+			break;
+		if (amdgpu_emu_mode == 1)
+			msleep(1);
+		else
+			udelay(1);
+	}
+
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+
+	amdgpu_device_wb_free(adev, index);
+
+	return r;
+}
+
+/**
+ * sdma_v5_2_ring_test_ib - test an IB on the DMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @timeout: timeout value in jiffies, or MAX_SCHEDULE_TIMEOUT
+ *
+ * Test a simple IB in the DMA ring.
+ * Returns 0 on success, error on failure.
+ */
+static int sdma_v5_2_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct dma_fence *f = NULL;
+	unsigned index;
+	long r;
+	u32 tmp = 0;
+	u64 gpu_addr;
+
+	tmp = 0xCAFEDEAD;
+	memset(&ib, 0, sizeof(ib));
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+
+	r = amdgpu_ib_get(adev, NULL, 256, AMDGPU_IB_POOL_DIRECT, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
+		goto err0;
+	}
+
+	ib.ptr[0] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR);
+	ib.ptr[1] = lower_32_bits(gpu_addr);
+	ib.ptr[2] = upper_32_bits(gpu_addr);
+	ib.ptr[3] = SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(0);
+	ib.ptr[4] = 0xDEADBEEF;
+	ib.ptr[5] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.ptr[6] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.ptr[7] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.length_dw = 8;
+
+	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
+	if (r)
+		goto err1;
+
+	r = dma_fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out\n");
+		r = -ETIMEDOUT;
+		goto err1;
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
+		goto err1;
+	}
+
+	tmp = le32_to_cpu(adev->wb.wb[index]);
+
+	if (tmp == 0xDEADBEEF)
+		r = 0;
+	else
+		r = -EINVAL;
+
+err1:
+	amdgpu_ib_free(&ib, NULL);
+	dma_fence_put(f);
+err0:
+	amdgpu_device_wb_free(adev, index);
+	return r;
+}
+
+
+/**
+ * sdma_v5_2_vm_copy_pte - update PTEs by copying them from the GART
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @src: src addr to copy from
+ * @count: number of page entries to update
+ *
+ * Update PTEs by copying them from the GART using sDMA.
+ */
+static void sdma_v5_2_vm_copy_pte(struct amdgpu_ib *ib,
+				  uint64_t pe, uint64_t src,
+				  unsigned count)
+{
+	unsigned bytes = count * 8;
+
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
+	ib->ptr[ib->length_dw++] = bytes - 1;
+	ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+	ib->ptr[ib->length_dw++] = lower_32_bits(src);
+	ib->ptr[ib->length_dw++] = upper_32_bits(src);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+
+}
+
+/**
+ * sdma_v5_2_vm_write_pte - update PTEs by writing them manually
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @value: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ *
+ * Update PTEs by writing them manually using sDMA.
+ */
+static void sdma_v5_2_vm_write_pte(struct amdgpu_ib *ib, uint64_t pe,
+				   uint64_t value, unsigned count,
+				   uint32_t incr)
+{
+	unsigned ndw = count * 2;
+
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+	ib->ptr[ib->length_dw++] = ndw - 1;
+	for (; ndw > 0; ndw -= 2) {
+		ib->ptr[ib->length_dw++] = lower_32_bits(value);
+		ib->ptr[ib->length_dw++] = upper_32_bits(value);
+		value += incr;
+	}
+}
+
+/**
+ * sdma_v5_2_vm_set_pte_pde - update the page tables using sDMA
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: access flags
+ *
+ * Update the page tables using sDMA.
+ */
+static void sdma_v5_2_vm_set_pte_pde(struct amdgpu_ib *ib,
+				     uint64_t pe,
+				     uint64_t addr, unsigned count,
+				     uint32_t incr, uint64_t flags)
+{
+	/* for physically contiguous pages (vram) */
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_PTEPDE);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe); /* dst addr */
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+	ib->ptr[ib->length_dw++] = lower_32_bits(flags); /* mask */
+	ib->ptr[ib->length_dw++] = upper_32_bits(flags);
+	ib->ptr[ib->length_dw++] = lower_32_bits(addr); /* value */
+	ib->ptr[ib->length_dw++] = upper_32_bits(addr);
+	ib->ptr[ib->length_dw++] = incr; /* increment size */
+	ib->ptr[ib->length_dw++] = 0;
+	ib->ptr[ib->length_dw++] = count - 1; /* number of entries */
+}
+
+/**
+ * sdma_v5_2_ring_pad_ib - pad the IB
+ *
+ * @ib: indirect buffer to fill with padding
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Pad the IB with NOPs to a boundary multiple of 8.
+ */
+static void sdma_v5_2_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring);
+	u32 pad_count;
+	int i;
+
+	pad_count = (-ib->length_dw) & 0x7;
+	for (i = 0; i < pad_count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			ib->ptr[ib->length_dw++] =
+				SDMA_PKT_HEADER_OP(SDMA_OP_NOP) |
+				SDMA_PKT_NOP_HEADER_COUNT(pad_count - 1);
+		else
+			ib->ptr[ib->length_dw++] =
+				SDMA_PKT_HEADER_OP(SDMA_OP_NOP);
+}
+
+
+/**
+ * sdma_v5_2_ring_emit_pipeline_sync - sync the pipeline
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Make sure all previous operations are completed (CIK).
+ */
+static void sdma_v5_2_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	/* wait for idle */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(0) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3) | /* equal */
+			  SDMA_PKT_POLL_REGMEM_HEADER_MEM_POLL(1));
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xffffffff);
+	amdgpu_ring_write(ring, seq); /* reference */
+	amdgpu_ring_write(ring, 0xffffffff); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(4)); /* retry count, poll interval */
+}
+
+
+/**
+ * sdma_v5_2_ring_emit_vm_flush - vm flush using sDMA
+ *
+ * @ring: amdgpu_ring pointer
+ * @vmid: vmid number to use
+ * @pd_addr: address
+ *
+ * Update the page table base and flush the VM TLB
+ * using sDMA.
+ */
+static void sdma_v5_2_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					 unsigned vmid, uint64_t pd_addr)
+{
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
+	uint32_t req = hub->vmhub_funcs->get_invalidate_req(vmid, 0);
+
+	/* Update the PD address for this VMID. */
+	amdgpu_ring_emit_wreg(ring, hub->ctx0_ptb_addr_lo32 +
+			      (hub->ctx_addr_distance * vmid),
+			      lower_32_bits(pd_addr));
+	amdgpu_ring_emit_wreg(ring, hub->ctx0_ptb_addr_hi32 +
+			      (hub->ctx_addr_distance * vmid),
+			      upper_32_bits(pd_addr));
+
+	/* Trigger invalidation. */
+	amdgpu_ring_write(ring,
+			  SDMA_PKT_VM_INVALIDATION_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_VM_INVALIDATION_HEADER_SUB_OP(SDMA_SUBOP_VM_INVALIDATION) |
+			  SDMA_PKT_VM_INVALIDATION_HEADER_GFX_ENG_ID(ring->vm_inv_eng) |
+			  SDMA_PKT_VM_INVALIDATION_HEADER_MM_ENG_ID(0x1f));
+	amdgpu_ring_write(ring, req);
+	amdgpu_ring_write(ring, 0xFFFFFFFF);
+	amdgpu_ring_write(ring,
+			  SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_INVALIDATEACK(1 << vmid) |
+			  SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_ADDRESSRANGEHI(0x1F));
+}
+
+static void sdma_v5_2_ring_emit_wreg(struct amdgpu_ring *ring,
+				     uint32_t reg, uint32_t val)
+{
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
+			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, val);
+}
+
+static void sdma_v5_2_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+					 uint32_t val, uint32_t mask)
+{
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(0) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3)); /* equal */
+	amdgpu_ring_write(ring, reg << 2);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, val); /* reference */
+	amdgpu_ring_write(ring, mask); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10));
+}
+
+static void sdma_v5_2_ring_emit_reg_write_reg_wait(struct amdgpu_ring *ring,
+						   uint32_t reg0, uint32_t reg1,
+						   uint32_t ref, uint32_t mask)
+{
+	amdgpu_ring_emit_wreg(ring, reg0, ref);
+	/* wait for a cycle to reset vm_inv_eng*_ack */
+	amdgpu_ring_emit_reg_wait(ring, reg0, 0, 0);
+	amdgpu_ring_emit_reg_wait(ring, reg1, mask, mask);
+}
+
+static int sdma_v5_2_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = amdgpu_sdma_init_microcode(adev, 0, true);
+	if (r)
+		return r;
+
+	sdma_v5_2_set_ring_funcs(adev);
+	sdma_v5_2_set_buffer_funcs(adev);
+	sdma_v5_2_set_vm_pte_funcs(adev);
+	sdma_v5_2_set_irq_funcs(adev);
+	sdma_v5_2_set_mqd_funcs(adev);
+
+	return 0;
+}
+
+static unsigned sdma_v5_2_seq_to_irq_id(int seq_num)
+{
+	switch (seq_num) {
+	case 0:
+		return SOC15_IH_CLIENTID_SDMA0;
+	case 1:
+		return SOC15_IH_CLIENTID_SDMA1;
+	case 2:
+		return SOC15_IH_CLIENTID_SDMA2;
+	case 3:
+		return SOC15_IH_CLIENTID_SDMA3_Sienna_Cichlid;
+	default:
+		break;
+	}
+	return -EINVAL;
+}
+
+static unsigned sdma_v5_2_seq_to_trap_id(int seq_num)
+{
+	switch (seq_num) {
+	case 0:
+		return SDMA0_5_0__SRCID__SDMA_TRAP;
+	case 1:
+		return SDMA1_5_0__SRCID__SDMA_TRAP;
+	case 2:
+		return SDMA2_5_0__SRCID__SDMA_TRAP;
+	case 3:
+		return SDMA3_5_0__SRCID__SDMA_TRAP;
+	default:
+		break;
+	}
+	return -EINVAL;
+}
+
+static int sdma_v5_2_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_ring *ring;
+	int r, i;
+	struct amdgpu_device *adev = ip_block->adev;
+	uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_5_2);
+	uint32_t *ptr;
+
+	/* SDMA trap event */
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		r = amdgpu_irq_add_id(adev, sdma_v5_2_seq_to_irq_id(i),
+				      sdma_v5_2_seq_to_trap_id(i),
+				      &adev->sdma.trap_irq);
+		if (r)
+			return r;
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		mutex_init(&adev->sdma.instance[i].engine_reset_mutex);
+		adev->sdma.instance[i].funcs = &sdma_v5_2_sdma_funcs;
+		ring = &adev->sdma.instance[i].ring;
+		ring->ring_obj = NULL;
+		ring->use_doorbell = true;
+		ring->me = i;
+
+		DRM_INFO("use_doorbell being set to: [%s]\n",
+				ring->use_doorbell?"true":"false");
+
+		ring->doorbell_index =
+			(adev->doorbell_index.sdma_engine[i] << 1); //get DWORD offset
+
+		ring->vm_hub = AMDGPU_GFXHUB(0);
+		sprintf(ring->name, "sdma%d", i);
+		r = amdgpu_ring_init(adev, ring, 1024, &adev->sdma.trap_irq,
+				     AMDGPU_SDMA_IRQ_INSTANCE0 + i,
+				     AMDGPU_RING_PRIO_DEFAULT, NULL);
+		if (r)
+			return r;
+	}
+
+	adev->sdma.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->sdma.instance[0].ring);
+	switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) {
+	case IP_VERSION(5, 2, 0):
+	case IP_VERSION(5, 2, 2):
+	case IP_VERSION(5, 2, 3):
+	case IP_VERSION(5, 2, 4):
+		if ((adev->sdma.instance[0].fw_version >= 76) &&
+		    !amdgpu_sriov_vf(adev))
+			adev->sdma.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+		break;
+	case IP_VERSION(5, 2, 5):
+		if ((adev->sdma.instance[0].fw_version >= 34) &&
+		    !amdgpu_sriov_vf(adev))
+			adev->sdma.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+		break;
+	default:
+		break;
+	}
+
+	/* Allocate memory for SDMA IP Dump buffer */
+	ptr = kcalloc(adev->sdma.num_instances * reg_count, sizeof(uint32_t), GFP_KERNEL);
+	if (ptr)
+		adev->sdma.ip_dump = ptr;
+	else
+		DRM_ERROR("Failed to allocated memory for SDMA IP Dump\n");
+
+	r = amdgpu_sdma_sysfs_reset_mask_init(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int sdma_v5_2_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		amdgpu_ring_fini(&adev->sdma.instance[i].ring);
+
+	amdgpu_sdma_sysfs_reset_mask_fini(adev);
+	amdgpu_sdma_destroy_inst_ctx(adev, true);
+
+	kfree(adev->sdma.ip_dump);
+
+	return 0;
+}
+
+static int sdma_v5_2_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	return sdma_v5_2_start(adev);
+}
+
+static int sdma_v5_2_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	sdma_v5_2_ctx_switch_enable(adev, false);
+	sdma_v5_2_enable(adev, false);
+
+	return 0;
+}
+
+static int sdma_v5_2_suspend(struct amdgpu_ip_block *ip_block)
+{
+	return sdma_v5_2_hw_fini(ip_block);
+}
+
+static int sdma_v5_2_resume(struct amdgpu_ip_block *ip_block)
+{
+	return sdma_v5_2_hw_init(ip_block);
+}
+
+static bool sdma_v5_2_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	u32 i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		u32 tmp = RREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_STATUS_REG));
+
+		if (!(tmp & SDMA0_STATUS_REG__IDLE_MASK))
+			return false;
+	}
+
+	return true;
+}
+
+static int sdma_v5_2_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	unsigned i;
+	u32 sdma0, sdma1, sdma2, sdma3;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		sdma0 = RREG32(sdma_v5_2_get_reg_offset(adev, 0, mmSDMA0_STATUS_REG));
+		sdma1 = RREG32(sdma_v5_2_get_reg_offset(adev, 1, mmSDMA0_STATUS_REG));
+		sdma2 = RREG32(sdma_v5_2_get_reg_offset(adev, 2, mmSDMA0_STATUS_REG));
+		sdma3 = RREG32(sdma_v5_2_get_reg_offset(adev, 3, mmSDMA0_STATUS_REG));
+
+		if (sdma0 & sdma1 & sdma2 & sdma3 & SDMA0_STATUS_REG__IDLE_MASK)
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static int sdma_v5_2_reset_queue(struct amdgpu_ring *ring,
+				 unsigned int vmid,
+				 struct amdgpu_fence *timedout_fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int r;
+
+	if (ring->me >= adev->sdma.num_instances) {
+		dev_err(adev->dev, "sdma instance not found\n");
+		return -EINVAL;
+	}
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+
+	amdgpu_amdkfd_suspend(adev, true);
+	r = amdgpu_sdma_reset_engine(adev, ring->me, true);
+	amdgpu_amdkfd_resume(adev, true);
+	if (r)
+		return r;
+
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static int sdma_v5_2_stop_queue(struct amdgpu_ring *ring)
+{
+	u32 f32_cntl, freeze, cntl, stat1_reg;
+	struct amdgpu_device *adev = ring->adev;
+	int i, j, r = 0;
+
+	if (amdgpu_sriov_vf(adev))
+		return -EINVAL;
+
+	i = ring->me;
+	amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
+
+	/* stop queue */
+	sdma_v5_2_gfx_stop(adev, 1 << i);
+
+	/*engine stop SDMA1_F32_CNTL.HALT to 1 and SDMAx_FREEZE freeze bit to 1 */
+	freeze = RREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_FREEZE));
+	freeze = REG_SET_FIELD(freeze, SDMA0_FREEZE, FREEZE, 1);
+	WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_FREEZE), freeze);
+
+	for (j = 0; j < adev->usec_timeout; j++) {
+		freeze = RREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_FREEZE));
+
+		if (REG_GET_FIELD(freeze, SDMA0_FREEZE, FROZEN) & 1)
+			break;
+		udelay(1);
+	}
+
+
+	if (j == adev->usec_timeout) {
+		stat1_reg = RREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_STATUS1_REG));
+		if ((stat1_reg & 0x3FF) != 0x3FF) {
+			DRM_ERROR("cannot soft reset as sdma not idle\n");
+			r = -ETIMEDOUT;
+			goto err0;
+		}
+	}
+
+	f32_cntl = RREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_F32_CNTL));
+	f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_F32_CNTL, HALT, 1);
+	WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_F32_CNTL), f32_cntl);
+
+	cntl = RREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_CNTL));
+	cntl = REG_SET_FIELD(cntl, SDMA0_CNTL, UTC_L1_ENABLE, 0);
+	WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_CNTL), cntl);
+
+err0:
+	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
+	return r;
+}
+
+static int sdma_v5_2_restore_queue(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 inst_id = ring->me;
+	u32 freeze;
+	int r;
+
+	amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
+	/* unfreeze and unhalt */
+	freeze = RREG32(sdma_v5_2_get_reg_offset(adev, inst_id, mmSDMA0_FREEZE));
+	freeze = REG_SET_FIELD(freeze, SDMA0_FREEZE, FREEZE, 0);
+	WREG32(sdma_v5_2_get_reg_offset(adev, inst_id, mmSDMA0_FREEZE), freeze);
+
+	r = sdma_v5_2_gfx_resume_instance(adev, inst_id, true);
+
+	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
+
+	return r;
+}
+
+static int sdma_v5_2_ring_preempt_ib(struct amdgpu_ring *ring)
+{
+	int i, r = 0;
+	struct amdgpu_device *adev = ring->adev;
+	u32 index = 0;
+	u64 sdma_gfx_preempt;
+
+	amdgpu_sdma_get_index_from_ring(ring, &index);
+	sdma_gfx_preempt =
+		sdma_v5_2_get_reg_offset(adev, index, mmSDMA0_GFX_PREEMPT);
+
+	/* assert preemption condition */
+	amdgpu_ring_set_preempt_cond_exec(ring, false);
+
+	/* emit the trailing fence */
+	ring->trail_seq += 1;
+	amdgpu_ring_alloc(ring, 10);
+	sdma_v5_2_ring_emit_fence(ring, ring->trail_fence_gpu_addr,
+				  ring->trail_seq, 0);
+	amdgpu_ring_commit(ring);
+
+	/* assert IB preemption */
+	WREG32(sdma_gfx_preempt, 1);
+
+	/* poll the trailing fence */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (ring->trail_seq ==
+		    le32_to_cpu(*(ring->trail_fence_cpu_addr)))
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout) {
+		r = -EINVAL;
+		DRM_ERROR("ring %d failed to be preempted\n", ring->idx);
+	}
+
+	/* deassert IB preemption */
+	WREG32(sdma_gfx_preempt, 0);
+
+	/* deassert the preemption condition */
+	amdgpu_ring_set_preempt_cond_exec(ring, true);
+	return r;
+}
+
+static int sdma_v5_2_set_trap_irq_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	u32 sdma_cntl;
+	u32 reg_offset = sdma_v5_2_get_reg_offset(adev, type, mmSDMA0_CNTL);
+
+	if (!amdgpu_sriov_vf(adev)) {
+		sdma_cntl = RREG32(reg_offset);
+		sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE,
+			       state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+		WREG32(reg_offset, sdma_cntl);
+	}
+
+	return 0;
+}
+
+static int sdma_v5_2_process_trap_irq(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	uint32_t mes_queue_id = entry->src_data[0];
+
+	DRM_DEBUG("IH: SDMA trap\n");
+
+	if (adev->enable_mes && (mes_queue_id & AMDGPU_FENCE_MES_QUEUE_FLAG)) {
+		struct amdgpu_mes_queue *queue;
+
+		mes_queue_id &= AMDGPU_FENCE_MES_QUEUE_ID_MASK;
+
+		spin_lock(&adev->mes.queue_id_lock);
+		queue = idr_find(&adev->mes.queue_id_idr, mes_queue_id);
+		if (queue) {
+			DRM_DEBUG("process smda queue id = %d\n", mes_queue_id);
+			amdgpu_fence_process(queue->ring);
+		}
+		spin_unlock(&adev->mes.queue_id_lock);
+		return 0;
+	}
+
+	switch (entry->client_id) {
+	case SOC15_IH_CLIENTID_SDMA0:
+		switch (entry->ring_id) {
+		case 0:
+			amdgpu_fence_process(&adev->sdma.instance[0].ring);
+			break;
+		case 1:
+			/* XXX compute */
+			break;
+		case 2:
+			/* XXX compute */
+			break;
+		case 3:
+			/* XXX page queue*/
+			break;
+		}
+		break;
+	case SOC15_IH_CLIENTID_SDMA1:
+		switch (entry->ring_id) {
+		case 0:
+			amdgpu_fence_process(&adev->sdma.instance[1].ring);
+			break;
+		case 1:
+			/* XXX compute */
+			break;
+		case 2:
+			/* XXX compute */
+			break;
+		case 3:
+			/* XXX page queue*/
+			break;
+		}
+		break;
+	case SOC15_IH_CLIENTID_SDMA2:
+		switch (entry->ring_id) {
+		case 0:
+			amdgpu_fence_process(&adev->sdma.instance[2].ring);
+			break;
+		case 1:
+			/* XXX compute */
+			break;
+		case 2:
+			/* XXX compute */
+			break;
+		case 3:
+			/* XXX page queue*/
+			break;
+		}
+		break;
+	case SOC15_IH_CLIENTID_SDMA3_Sienna_Cichlid:
+		switch (entry->ring_id) {
+		case 0:
+			amdgpu_fence_process(&adev->sdma.instance[3].ring);
+			break;
+		case 1:
+			/* XXX compute */
+			break;
+		case 2:
+			/* XXX compute */
+			break;
+		case 3:
+			/* XXX page queue*/
+			break;
+		}
+		break;
+	}
+	return 0;
+}
+
+static int sdma_v5_2_process_illegal_inst_irq(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      struct amdgpu_iv_entry *entry)
+{
+	return 0;
+}
+
+static bool sdma_v5_2_firmware_mgcg_support(struct amdgpu_device *adev,
+						     int i)
+{
+	switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) {
+	case IP_VERSION(5, 2, 1):
+		if (adev->sdma.instance[i].fw_version < 70)
+			return false;
+		break;
+	case IP_VERSION(5, 2, 3):
+		if (adev->sdma.instance[i].fw_version < 47)
+			return false;
+		break;
+	case IP_VERSION(5, 2, 7):
+		if (adev->sdma.instance[i].fw_version < 9)
+			return false;
+		break;
+	default:
+		return true;
+	}
+
+	return true;
+
+}
+
+static void sdma_v5_2_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						       bool enable)
+{
+	uint32_t data, def;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+
+		if (!sdma_v5_2_firmware_mgcg_support(adev, i))
+			adev->cg_flags &= ~AMD_CG_SUPPORT_SDMA_MGCG;
+
+		if (enable && (adev->cg_flags & AMD_CG_SUPPORT_SDMA_MGCG)) {
+			/* Enable sdma clock gating */
+			def = data = RREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_CLK_CTRL));
+			data &= ~(SDMA0_CLK_CTRL__SOFT_OVERRIDE4_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE3_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE2_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE1_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE0_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDER_REG_MASK);
+			if (def != data)
+				WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_CLK_CTRL), data);
+		} else {
+			/* Disable sdma clock gating */
+			def = data = RREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_CLK_CTRL));
+			data |= (SDMA0_CLK_CTRL__SOFT_OVERRIDE4_MASK |
+				 SDMA0_CLK_CTRL__SOFT_OVERRIDE3_MASK |
+				 SDMA0_CLK_CTRL__SOFT_OVERRIDE2_MASK |
+				 SDMA0_CLK_CTRL__SOFT_OVERRIDE1_MASK |
+				 SDMA0_CLK_CTRL__SOFT_OVERRIDE0_MASK |
+				 SDMA0_CLK_CTRL__SOFT_OVERRIDER_REG_MASK);
+			if (def != data)
+				WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_CLK_CTRL), data);
+		}
+	}
+}
+
+static void sdma_v5_2_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+						      bool enable)
+{
+	uint32_t data, def;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (adev->sdma.instance[i].fw_version < 70 &&
+		    amdgpu_ip_version(adev, SDMA0_HWIP, 0) ==
+			    IP_VERSION(5, 2, 1))
+			adev->cg_flags &= ~AMD_CG_SUPPORT_SDMA_LS;
+
+		if (enable && (adev->cg_flags & AMD_CG_SUPPORT_SDMA_LS)) {
+			/* Enable sdma mem light sleep */
+			def = data = RREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_POWER_CNTL));
+			data |= SDMA0_POWER_CNTL__MEM_POWER_OVERRIDE_MASK;
+			if (def != data)
+				WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_POWER_CNTL), data);
+
+		} else {
+			/* Disable sdma mem light sleep */
+			def = data = RREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_POWER_CNTL));
+			data &= ~SDMA0_POWER_CNTL__MEM_POWER_OVERRIDE_MASK;
+			if (def != data)
+				WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_POWER_CNTL), data);
+
+		}
+	}
+}
+
+static int sdma_v5_2_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					   enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) {
+	case IP_VERSION(5, 2, 0):
+	case IP_VERSION(5, 2, 2):
+	case IP_VERSION(5, 2, 1):
+	case IP_VERSION(5, 2, 4):
+	case IP_VERSION(5, 2, 5):
+	case IP_VERSION(5, 2, 6):
+	case IP_VERSION(5, 2, 3):
+	case IP_VERSION(5, 2, 7):
+		sdma_v5_2_update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		sdma_v5_2_update_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int sdma_v5_2_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static void sdma_v5_2_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int data;
+
+	if (amdgpu_sriov_vf(adev))
+		*flags = 0;
+
+	/* AMD_CG_SUPPORT_SDMA_MGCG */
+	data = RREG32(sdma_v5_2_get_reg_offset(adev, 0, mmSDMA0_CLK_CTRL));
+	if (!(data & SDMA0_CLK_CTRL__CGCG_EN_OVERRIDE_MASK))
+		*flags |= AMD_CG_SUPPORT_SDMA_MGCG;
+
+	/* AMD_CG_SUPPORT_SDMA_LS */
+	data = RREG32_KIQ(sdma_v5_2_get_reg_offset(adev, 0, mmSDMA0_POWER_CNTL));
+	if (data & SDMA0_POWER_CNTL__MEM_POWER_OVERRIDE_MASK)
+		*flags |= AMD_CG_SUPPORT_SDMA_LS;
+}
+
+static void sdma_v5_2_ring_begin_use(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	/* SDMA 5.2.3 (RMB) FW doesn't seem to properly
+	 * disallow GFXOFF in some cases leading to
+	 * hangs in SDMA.  Disallow GFXOFF while SDMA is active.
+	 * We can probably just limit this to 5.2.3,
+	 * but it shouldn't hurt for other parts since
+	 * this GFXOFF will be disallowed anyway when SDMA is
+	 * active, this just makes it explicit.
+	 * sdma_v5_2_ring_set_wptr() takes advantage of this
+	 * to update the wptr because sometimes SDMA seems to miss
+	 * doorbells when entering PG.  If you remove this, update
+	 * sdma_v5_2_ring_set_wptr() as well!
+	 */
+	amdgpu_gfx_off_ctrl(adev, false);
+}
+
+static void sdma_v5_2_ring_end_use(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	/* SDMA 5.2.3 (RMB) FW doesn't seem to properly
+	 * disallow GFXOFF in some cases leading to
+	 * hangs in SDMA.  Allow GFXOFF when SDMA is complete.
+	 */
+	amdgpu_gfx_off_ctrl(adev, true);
+}
+
+static void sdma_v5_2_print_ip_state(struct amdgpu_ip_block *ip_block, struct drm_printer *p)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, j;
+	uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_5_2);
+	uint32_t instance_offset;
+
+	if (!adev->sdma.ip_dump)
+		return;
+
+	drm_printf(p, "num_instances:%d\n", adev->sdma.num_instances);
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		instance_offset = i * reg_count;
+		drm_printf(p, "\nInstance:%d\n", i);
+
+		for (j = 0; j < reg_count; j++)
+			drm_printf(p, "%-50s \t 0x%08x\n", sdma_reg_list_5_2[j].reg_name,
+				   adev->sdma.ip_dump[instance_offset + j]);
+	}
+}
+
+static void sdma_v5_2_dump_ip_state(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, j;
+	uint32_t instance_offset;
+	uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_5_2);
+
+	if (!adev->sdma.ip_dump)
+		return;
+
+	amdgpu_gfx_off_ctrl(adev, false);
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		instance_offset = i * reg_count;
+		for (j = 0; j < reg_count; j++)
+			adev->sdma.ip_dump[instance_offset + j] =
+				RREG32(sdma_v5_2_get_reg_offset(adev, i,
+				       sdma_reg_list_5_2[j].reg_offset));
+	}
+	amdgpu_gfx_off_ctrl(adev, true);
+}
+
+static const struct amd_ip_funcs sdma_v5_2_ip_funcs = {
+	.name = "sdma_v5_2",
+	.early_init = sdma_v5_2_early_init,
+	.sw_init = sdma_v5_2_sw_init,
+	.sw_fini = sdma_v5_2_sw_fini,
+	.hw_init = sdma_v5_2_hw_init,
+	.hw_fini = sdma_v5_2_hw_fini,
+	.suspend = sdma_v5_2_suspend,
+	.resume = sdma_v5_2_resume,
+	.is_idle = sdma_v5_2_is_idle,
+	.wait_for_idle = sdma_v5_2_wait_for_idle,
+	.soft_reset = sdma_v5_2_soft_reset,
+	.set_clockgating_state = sdma_v5_2_set_clockgating_state,
+	.set_powergating_state = sdma_v5_2_set_powergating_state,
+	.get_clockgating_state = sdma_v5_2_get_clockgating_state,
+	.dump_ip_state = sdma_v5_2_dump_ip_state,
+	.print_ip_state = sdma_v5_2_print_ip_state,
+};
+
+static const struct amdgpu_ring_funcs sdma_v5_2_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_SDMA,
+	.align_mask = 0xf,
+	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
+	.support_64bit_ptrs = true,
+	.secure_submission_supported = true,
+	.get_rptr = sdma_v5_2_ring_get_rptr,
+	.get_wptr = sdma_v5_2_ring_get_wptr,
+	.set_wptr = sdma_v5_2_ring_set_wptr,
+	.emit_frame_size =
+		5 + /* sdma_v5_2_ring_init_cond_exec */
+		6 + /* sdma_v5_2_ring_emit_hdp_flush */
+		3 + /* hdp_invalidate */
+		6 + /* sdma_v5_2_ring_emit_pipeline_sync */
+		/* sdma_v5_2_ring_emit_vm_flush */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 6 +
+		10 + 10 + 10, /* sdma_v5_2_ring_emit_fence x3 for user fence, vm fence */
+	.emit_ib_size = 7 + 6, /* sdma_v5_2_ring_emit_ib */
+	.emit_ib = sdma_v5_2_ring_emit_ib,
+	.emit_mem_sync = sdma_v5_2_ring_emit_mem_sync,
+	.emit_fence = sdma_v5_2_ring_emit_fence,
+	.emit_pipeline_sync = sdma_v5_2_ring_emit_pipeline_sync,
+	.emit_vm_flush = sdma_v5_2_ring_emit_vm_flush,
+	.emit_hdp_flush = sdma_v5_2_ring_emit_hdp_flush,
+	.test_ring = sdma_v5_2_ring_test_ring,
+	.test_ib = sdma_v5_2_ring_test_ib,
+	.insert_nop = sdma_v5_2_ring_insert_nop,
+	.pad_ib = sdma_v5_2_ring_pad_ib,
+	.begin_use = sdma_v5_2_ring_begin_use,
+	.end_use = sdma_v5_2_ring_end_use,
+	.emit_wreg = sdma_v5_2_ring_emit_wreg,
+	.emit_reg_wait = sdma_v5_2_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = sdma_v5_2_ring_emit_reg_write_reg_wait,
+	.init_cond_exec = sdma_v5_2_ring_init_cond_exec,
+	.preempt_ib = sdma_v5_2_ring_preempt_ib,
+	.reset = sdma_v5_2_reset_queue,
+};
+
+static void sdma_v5_2_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		adev->sdma.instance[i].ring.funcs = &sdma_v5_2_ring_funcs;
+		adev->sdma.instance[i].ring.me = i;
+	}
+}
+
+static const struct amdgpu_irq_src_funcs sdma_v5_2_trap_irq_funcs = {
+	.set = sdma_v5_2_set_trap_irq_state,
+	.process = sdma_v5_2_process_trap_irq,
+};
+
+static const struct amdgpu_irq_src_funcs sdma_v5_2_illegal_inst_irq_funcs = {
+	.process = sdma_v5_2_process_illegal_inst_irq,
+};
+
+static void sdma_v5_2_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->sdma.trap_irq.num_types = AMDGPU_SDMA_IRQ_INSTANCE0 +
+					adev->sdma.num_instances;
+	adev->sdma.trap_irq.funcs = &sdma_v5_2_trap_irq_funcs;
+	adev->sdma.illegal_inst_irq.funcs = &sdma_v5_2_illegal_inst_irq_funcs;
+}
+
+/**
+ * sdma_v5_2_emit_copy_buffer - copy buffer using the sDMA engine
+ *
+ * @ib: indirect buffer to copy to
+ * @src_offset: src GPU address
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ * @copy_flags: copy flags for the buffers
+ *
+ * Copy GPU buffers using the DMA engine.
+ * Used by the amdgpu ttm implementation to move pages if
+ * registered as the asic copy callback.
+ */
+static void sdma_v5_2_emit_copy_buffer(struct amdgpu_ib *ib,
+				       uint64_t src_offset,
+				       uint64_t dst_offset,
+				       uint32_t byte_count,
+				       uint32_t copy_flags)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR) |
+		SDMA_PKT_COPY_LINEAR_HEADER_TMZ((copy_flags & AMDGPU_COPY_FLAGS_TMZ) ? 1 : 0);
+	ib->ptr[ib->length_dw++] = byte_count - 1;
+	ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+	ib->ptr[ib->length_dw++] = lower_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+}
+
+/**
+ * sdma_v5_2_emit_fill_buffer - fill buffer using the sDMA engine
+ *
+ * @ib: indirect buffer to fill
+ * @src_data: value to write to buffer
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ *
+ * Fill GPU buffers using the DMA engine.
+ */
+static void sdma_v5_2_emit_fill_buffer(struct amdgpu_ib *ib,
+				       uint32_t src_data,
+				       uint64_t dst_offset,
+				       uint32_t byte_count)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_CONST_FILL);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = src_data;
+	ib->ptr[ib->length_dw++] = byte_count - 1;
+}
+
+static const struct amdgpu_buffer_funcs sdma_v5_2_buffer_funcs = {
+	.copy_max_bytes = 0x400000,
+	.copy_num_dw = 7,
+	.emit_copy_buffer = sdma_v5_2_emit_copy_buffer,
+
+	.fill_max_bytes = 0x400000,
+	.fill_num_dw = 5,
+	.emit_fill_buffer = sdma_v5_2_emit_fill_buffer,
+};
+
+static void sdma_v5_2_set_buffer_funcs(struct amdgpu_device *adev)
+{
+	if (adev->mman.buffer_funcs == NULL) {
+		adev->mman.buffer_funcs = &sdma_v5_2_buffer_funcs;
+		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
+	}
+}
+
+static const struct amdgpu_vm_pte_funcs sdma_v5_2_vm_pte_funcs = {
+	.copy_pte_num_dw = 7,
+	.copy_pte = sdma_v5_2_vm_copy_pte,
+	.write_pte = sdma_v5_2_vm_write_pte,
+	.set_pte_pde = sdma_v5_2_vm_set_pte_pde,
+};
+
+static void sdma_v5_2_set_vm_pte_funcs(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	if (adev->vm_manager.vm_pte_funcs == NULL) {
+		adev->vm_manager.vm_pte_funcs = &sdma_v5_2_vm_pte_funcs;
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			adev->vm_manager.vm_pte_scheds[i] =
+				&adev->sdma.instance[i].ring.sched;
+		}
+		adev->vm_manager.vm_pte_num_scheds = adev->sdma.num_instances;
+	}
+}
+
+const struct amdgpu_ip_block_version sdma_v5_2_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_SDMA,
+	.major = 5,
+	.minor = 2,
+	.rev = 0,
+	.funcs = &sdma_v5_2_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.h b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.h
new file mode 100644
index 000000000000..863145b3a77e
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __SDMA_V5_2_H__
+#define __SDMA_V5_2_H__
+
+extern const struct amdgpu_ip_block_version sdma_v5_2_ip_block;
+
+#endif /* __SDMA_V5_2_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c
new file mode 100644
index 000000000000..db6e41967f12
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c
@@ -0,0 +1,1916 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/delay.h>
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+
+#include "amdgpu.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_trace.h"
+
+#include "gc/gc_11_0_0_offset.h"
+#include "gc/gc_11_0_0_sh_mask.h"
+#include "gc/gc_11_0_0_default.h"
+#include "hdp/hdp_6_0_0_offset.h"
+#include "ivsrcid/gfx/irqsrcs_gfx_11_0_0.h"
+
+#include "soc15_common.h"
+#include "soc15.h"
+#include "sdma_v6_0_0_pkt_open.h"
+#include "nbio_v4_3.h"
+#include "sdma_common.h"
+#include "sdma_v6_0.h"
+#include "v11_structs.h"
+#include "mes_userqueue.h"
+#include "amdgpu_userq_fence.h"
+
+MODULE_FIRMWARE("amdgpu/sdma_6_0_0.bin");
+MODULE_FIRMWARE("amdgpu/sdma_6_0_1.bin");
+MODULE_FIRMWARE("amdgpu/sdma_6_0_2.bin");
+MODULE_FIRMWARE("amdgpu/sdma_6_0_3.bin");
+MODULE_FIRMWARE("amdgpu/sdma_6_1_0.bin");
+MODULE_FIRMWARE("amdgpu/sdma_6_1_1.bin");
+MODULE_FIRMWARE("amdgpu/sdma_6_1_2.bin");
+MODULE_FIRMWARE("amdgpu/sdma_6_1_3.bin");
+
+#define SDMA1_REG_OFFSET 0x600
+#define SDMA0_HYP_DEC_REG_START 0x5880
+#define SDMA0_HYP_DEC_REG_END 0x589a
+#define SDMA1_HYP_DEC_REG_OFFSET 0x20
+
+static const struct amdgpu_hwip_reg_entry sdma_reg_list_6_0[] = {
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_STATUS_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_STATUS1_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_STATUS2_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_STATUS3_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_STATUS4_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_STATUS5_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_STATUS6_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_UCODE_CHECKSUM),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_RB_RPTR_FETCH_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_RB_RPTR_FETCH),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_UTCL1_RD_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_UTCL1_WR_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_UTCL1_RD_XNACK0),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_UTCL1_RD_XNACK1),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_UTCL1_WR_XNACK0),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_UTCL1_WR_XNACK1),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_RB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_RB_RPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_RB_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_RB_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_IB_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_IB_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_IB_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_IB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_IB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_IB_SUB_REMAIN),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_DUMMY_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE_STATUS0),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_RB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_RB_RPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_RB_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_RB_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_IB_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_IB_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_IB_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_IB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_IB_SUB_REMAIN),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_DUMMY_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_RB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_RB_RPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_RB_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_RB_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_IB_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_IB_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_IB_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_IB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_IB_SUB_REMAIN),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_DUMMY_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_INT_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS2),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_CHICKEN_BITS),
+};
+
+static void sdma_v6_0_set_ring_funcs(struct amdgpu_device *adev);
+static void sdma_v6_0_set_buffer_funcs(struct amdgpu_device *adev);
+static void sdma_v6_0_set_vm_pte_funcs(struct amdgpu_device *adev);
+static void sdma_v6_0_set_irq_funcs(struct amdgpu_device *adev);
+static int sdma_v6_0_start(struct amdgpu_device *adev);
+
+static u32 sdma_v6_0_get_reg_offset(struct amdgpu_device *adev, u32 instance, u32 internal_offset)
+{
+	u32 base;
+
+	if (internal_offset >= SDMA0_HYP_DEC_REG_START &&
+	    internal_offset <= SDMA0_HYP_DEC_REG_END) {
+		base = adev->reg_offset[GC_HWIP][0][1];
+		if (instance != 0)
+			internal_offset += SDMA1_HYP_DEC_REG_OFFSET * instance;
+	} else {
+		base = adev->reg_offset[GC_HWIP][0][0];
+		if (instance == 1)
+			internal_offset += SDMA1_REG_OFFSET;
+	}
+
+	return base + internal_offset;
+}
+
+static unsigned sdma_v6_0_ring_init_cond_exec(struct amdgpu_ring *ring,
+					      uint64_t addr)
+{
+	unsigned ret;
+
+	amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_COND_EXE));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, 1);
+	/* this is the offset we need patch later */
+	ret = ring->wptr & ring->buf_mask;
+	/* insert dummy here and patch it later */
+	amdgpu_ring_write(ring, 0);
+
+	return ret;
+}
+
+/**
+ * sdma_v6_0_ring_get_rptr - get the current read pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current rptr from the hardware.
+ */
+static uint64_t sdma_v6_0_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	u64 *rptr;
+
+	/* XXX check if swapping is necessary on BE */
+	rptr = (u64 *)ring->rptr_cpu_addr;
+
+	DRM_DEBUG("rptr before shift == 0x%016llx\n", *rptr);
+	return ((*rptr) >> 2);
+}
+
+/**
+ * sdma_v6_0_ring_get_wptr - get the current write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current wptr from the hardware.
+ */
+static uint64_t sdma_v6_0_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	u64 wptr = 0;
+
+	if (ring->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		wptr = READ_ONCE(*((u64 *)ring->wptr_cpu_addr));
+		DRM_DEBUG("wptr/doorbell before shift == 0x%016llx\n", wptr);
+	}
+
+	return wptr >> 2;
+}
+
+/**
+ * sdma_v6_0_ring_set_wptr - commit the write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Write the wptr back to the hardware.
+ */
+static void sdma_v6_0_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		DRM_DEBUG("Using doorbell -- "
+			  "wptr_offs == 0x%08x "
+			  "lower_32_bits(ring->wptr) << 2 == 0x%08x "
+			  "upper_32_bits(ring->wptr) << 2 == 0x%08x\n",
+			  ring->wptr_offs,
+			  lower_32_bits(ring->wptr << 2),
+			  upper_32_bits(ring->wptr << 2));
+		/* XXX check if swapping is necessary on BE */
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr,
+			     ring->wptr << 2);
+		DRM_DEBUG("calling WDOORBELL64(0x%08x, 0x%016llx)\n",
+			  ring->doorbell_index, ring->wptr << 2);
+		WDOORBELL64(ring->doorbell_index, ring->wptr << 2);
+	} else {
+		DRM_DEBUG("Not using doorbell -- "
+			  "regSDMA%i_GFX_RB_WPTR == 0x%08x "
+			  "regSDMA%i_GFX_RB_WPTR_HI == 0x%08x\n",
+			  ring->me,
+			  lower_32_bits(ring->wptr << 2),
+			  ring->me,
+			  upper_32_bits(ring->wptr << 2));
+		WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev,
+							     ring->me, regSDMA0_QUEUE0_RB_WPTR),
+				lower_32_bits(ring->wptr << 2));
+		WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev,
+							     ring->me, regSDMA0_QUEUE0_RB_WPTR_HI),
+				upper_32_bits(ring->wptr << 2));
+	}
+}
+
+static void sdma_v6_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring);
+	int i;
+
+	for (i = 0; i < count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			amdgpu_ring_write(ring, ring->funcs->nop |
+				SDMA_PKT_NOP_HEADER_COUNT(count - 1));
+		else
+			amdgpu_ring_write(ring, ring->funcs->nop);
+}
+
+/*
+ * sdma_v6_0_ring_emit_ib - Schedule an IB on the DMA engine
+ *
+ * @ring: amdgpu ring pointer
+ * @ib: IB object to schedule
+ * @flags: unused
+ * @job: job to retrieve vmid from
+ *
+ * Schedule an IB in the DMA ring.
+ */
+static void sdma_v6_0_ring_emit_ib(struct amdgpu_ring *ring,
+				   struct amdgpu_job *job,
+				   struct amdgpu_ib *ib,
+				   uint32_t flags)
+{
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+	uint64_t csa_mc_addr = amdgpu_sdma_get_csa_mc_addr(ring, vmid);
+
+	/* An IB packet must end on a 8 DW boundary--the next dword
+	 * must be on a 8-dword boundary. Our IB packet below is 6
+	 * dwords long, thus add x number of NOPs, such that, in
+	 * modular arithmetic,
+	 * wptr + 6 + x = 8k, k >= 0, which in C is,
+	 * (wptr + 6 + x) % 8 = 0.
+	 * The expression below, is a solution of x.
+	 */
+	sdma_v6_0_ring_insert_nop(ring, (2 - lower_32_bits(ring->wptr)) & 7);
+
+	amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_INDIRECT) |
+			  SDMA_PKT_INDIRECT_HEADER_VMID(vmid & 0xf));
+	/* base must be 32 byte aligned */
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr) & 0xffffffe0);
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, ib->length_dw);
+	amdgpu_ring_write(ring, lower_32_bits(csa_mc_addr));
+	amdgpu_ring_write(ring, upper_32_bits(csa_mc_addr));
+}
+
+/**
+ * sdma_v6_0_ring_emit_mem_sync - flush the IB by graphics cache rinse
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * flush the IB by graphics cache rinse.
+ */
+static void sdma_v6_0_ring_emit_mem_sync(struct amdgpu_ring *ring)
+{
+        uint32_t gcr_cntl = SDMA_GCR_GL2_INV | SDMA_GCR_GL2_WB | SDMA_GCR_GLM_INV |
+                            SDMA_GCR_GL1_INV | SDMA_GCR_GLV_INV | SDMA_GCR_GLK_INV |
+                            SDMA_GCR_GLI_INV(1);
+
+        /* flush entire cache L0/L1/L2, this can be optimized by performance requirement */
+        amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_GCR_REQ));
+        amdgpu_ring_write(ring, SDMA_PKT_GCR_REQ_PAYLOAD1_BASE_VA_31_7(0));
+        amdgpu_ring_write(ring, SDMA_PKT_GCR_REQ_PAYLOAD2_GCR_CONTROL_15_0(gcr_cntl) |
+                          SDMA_PKT_GCR_REQ_PAYLOAD2_BASE_VA_47_32(0));
+        amdgpu_ring_write(ring, SDMA_PKT_GCR_REQ_PAYLOAD3_LIMIT_VA_31_7(0) |
+                          SDMA_PKT_GCR_REQ_PAYLOAD3_GCR_CONTROL_18_16(gcr_cntl >> 16));
+        amdgpu_ring_write(ring, SDMA_PKT_GCR_REQ_PAYLOAD4_LIMIT_VA_47_32(0) |
+                          SDMA_PKT_GCR_REQ_PAYLOAD4_VMID(0));
+}
+
+
+/**
+ * sdma_v6_0_ring_emit_hdp_flush - emit an hdp flush on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Emit an hdp flush packet on the requested DMA ring.
+ */
+static void sdma_v6_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 ref_and_mask = 0;
+	const struct nbio_hdp_flush_reg *nbio_hf_reg = adev->nbio.hdp_flush_reg;
+
+	ref_and_mask = nbio_hf_reg->ref_and_mask_sdma0 << ring->me;
+
+	amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(1) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3)); /* == */
+	amdgpu_ring_write(ring, (adev->nbio.funcs->get_hdp_flush_done_offset(adev)) << 2);
+	amdgpu_ring_write(ring, (adev->nbio.funcs->get_hdp_flush_req_offset(adev)) << 2);
+	amdgpu_ring_write(ring, ref_and_mask); /* reference */
+	amdgpu_ring_write(ring, ref_and_mask); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
+}
+
+/**
+ * sdma_v6_0_ring_emit_fence - emit a fence on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ * @addr: address
+ * @seq: fence seq number
+ * @flags: fence flags
+ *
+ * Add a DMA fence packet to the ring to write
+ * the fence seq number and DMA trap packet to generate
+ * an interrupt if needed.
+ */
+static void sdma_v6_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				      unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	/* write the fence */
+	amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_FENCE) |
+			  SDMA_PKT_FENCE_HEADER_MTYPE(0x3)); /* Ucached(UC) */
+	/* zero in first two bits */
+	BUG_ON(addr & 0x3);
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+
+	/* optionally write high bits as well */
+	if (write64bit) {
+		addr += 4;
+		amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_FENCE) |
+				  SDMA_PKT_FENCE_HEADER_MTYPE(0x3));
+		/* zero in first two bits */
+		BUG_ON(addr & 0x3);
+		amdgpu_ring_write(ring, lower_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(seq));
+	}
+
+	if (flags & AMDGPU_FENCE_FLAG_INT) {
+		/* generate an interrupt */
+		amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_TRAP));
+		amdgpu_ring_write(ring, SDMA_PKT_TRAP_INT_CONTEXT_INT_CONTEXT(0));
+	}
+}
+
+/**
+ * sdma_v6_0_gfx_stop - stop the gfx async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the gfx async dma ring buffers.
+ */
+static void sdma_v6_0_gfx_stop(struct amdgpu_device *adev)
+{
+	u32 rb_cntl, ib_cntl;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		rb_cntl = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_CNTL));
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RB_ENABLE, 0);
+		WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_CNTL), rb_cntl);
+		ib_cntl = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_IB_CNTL));
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_QUEUE0_IB_CNTL, IB_ENABLE, 0);
+		WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_IB_CNTL), ib_cntl);
+	}
+}
+
+/**
+ * sdma_v6_0_rlc_stop - stop the compute async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the compute async dma queues.
+ */
+static void sdma_v6_0_rlc_stop(struct amdgpu_device *adev)
+{
+	/* XXX todo */
+}
+
+/**
+ * sdma_v6_0_ctxempty_int_enable - enable or disable context empty interrupts
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable context switching due to queue empty conditions
+ *
+ * Enable or disable the async dma engines queue empty context switch.
+ */
+static void sdma_v6_0_ctxempty_int_enable(struct amdgpu_device *adev, bool enable)
+{
+	u32 f32_cntl;
+	int i;
+
+	if (!amdgpu_sriov_vf(adev)) {
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			f32_cntl = RREG32(sdma_v6_0_get_reg_offset(adev, i, regSDMA0_CNTL));
+			f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL,
+					CTXEMPTY_INT_ENABLE, enable ? 1 : 0);
+			WREG32(sdma_v6_0_get_reg_offset(adev, i, regSDMA0_CNTL), f32_cntl);
+		}
+	}
+}
+
+/**
+ * sdma_v6_0_enable - stop the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable the DMA MEs.
+ *
+ * Halt or unhalt the async dma engines.
+ */
+static void sdma_v6_0_enable(struct amdgpu_device *adev, bool enable)
+{
+	u32 f32_cntl;
+	int i;
+
+	if (!enable) {
+		sdma_v6_0_gfx_stop(adev);
+		sdma_v6_0_rlc_stop(adev);
+	}
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		f32_cntl = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_F32_CNTL));
+		f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_F32_CNTL, HALT, enable ? 0 : 1);
+		WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_F32_CNTL), f32_cntl);
+	}
+}
+
+/**
+ * sdma_v6_0_gfx_resume_instance - start/restart a certain sdma engine
+ *
+ * @adev: amdgpu_device pointer
+ * @i: instance
+ * @restore: used to restore wptr when restart
+ *
+ * Set up the gfx DMA ring buffers and enable them. On restart, we will restore wptr and rptr.
+ * Return 0 for success.
+ */
+static int sdma_v6_0_gfx_resume_instance(struct amdgpu_device *adev, int i, bool restore)
+{
+	struct amdgpu_ring *ring;
+	u32 rb_cntl, ib_cntl;
+	u32 rb_bufsz;
+	u32 doorbell;
+	u32 doorbell_offset;
+	u32 temp;
+	u64 wptr_gpu_addr;
+
+	ring = &adev->sdma.instance[i].ring;
+	if (!amdgpu_sriov_vf(adev))
+		WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_SEM_WAIT_FAIL_TIMER_CNTL), 0);
+
+	/* Set ring buffer size in dwords */
+	rb_bufsz = order_base_2(ring->ring_size / 4);
+	rb_cntl = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_CNTL));
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RB_SIZE, rb_bufsz);
+#ifdef __BIG_ENDIAN
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RB_SWAP_ENABLE, 1);
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL,
+				RPTR_WRITEBACK_SWAP_ENABLE, 1);
+#endif
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RB_PRIV, 1);
+	WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_CNTL), rb_cntl);
+
+	/* Initialize the ring buffer's read and write pointers */
+	if (restore) {
+		WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_RPTR), lower_32_bits(ring->wptr << 2));
+		WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_RPTR_HI), upper_32_bits(ring->wptr << 2));
+		WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR), lower_32_bits(ring->wptr << 2));
+		WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR_HI), upper_32_bits(ring->wptr << 2));
+	} else {
+		WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_RPTR), 0);
+		WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_RPTR_HI), 0);
+		WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR), 0);
+		WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR_HI), 0);
+	}
+	/* setup the wptr shadow polling */
+	wptr_gpu_addr = ring->wptr_gpu_addr;
+	WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR_POLL_ADDR_LO),
+	       lower_32_bits(wptr_gpu_addr));
+	WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR_POLL_ADDR_HI),
+	       upper_32_bits(wptr_gpu_addr));
+
+	/* set the wb address whether it's enabled or not */
+	WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_RPTR_ADDR_HI),
+	       upper_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFF);
+	WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_RPTR_ADDR_LO),
+	       lower_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFC);
+
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RPTR_WRITEBACK_ENABLE, 1);
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, WPTR_POLL_ENABLE, 0);
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, F32_WPTR_POLL_ENABLE, 1);
+
+	WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_BASE), ring->gpu_addr >> 8);
+	WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_BASE_HI), ring->gpu_addr >> 40);
+
+	if (!restore)
+		ring->wptr = 0;
+
+	/* before programing wptr to a less value, need set minor_ptr_update first */
+	WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_MINOR_PTR_UPDATE), 1);
+
+	if (!amdgpu_sriov_vf(adev)) { /* only bare-metal use register write for wptr */
+		WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR), lower_32_bits(ring->wptr) << 2);
+		WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR_HI), upper_32_bits(ring->wptr) << 2);
+	}
+
+	doorbell = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_DOORBELL));
+	doorbell_offset = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_DOORBELL_OFFSET));
+
+	if (ring->use_doorbell) {
+		doorbell = REG_SET_FIELD(doorbell, SDMA0_QUEUE0_DOORBELL, ENABLE, 1);
+		doorbell_offset = REG_SET_FIELD(doorbell_offset, SDMA0_QUEUE0_DOORBELL_OFFSET,
+				OFFSET, ring->doorbell_index);
+	} else {
+		doorbell = REG_SET_FIELD(doorbell, SDMA0_QUEUE0_DOORBELL, ENABLE, 0);
+	}
+	WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_DOORBELL), doorbell);
+	WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_DOORBELL_OFFSET), doorbell_offset);
+
+	if (i == 0)
+		adev->nbio.funcs->sdma_doorbell_range(adev, i, ring->use_doorbell,
+					      ring->doorbell_index,
+					      adev->doorbell_index.sdma_doorbell_range * adev->sdma.num_instances);
+
+	if (amdgpu_sriov_vf(adev))
+		sdma_v6_0_ring_set_wptr(ring);
+
+	/* set minor_ptr_update to 0 after wptr programed */
+	WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_MINOR_PTR_UPDATE), 0);
+
+	/* Set up sdma hang watchdog */
+	temp = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_WATCHDOG_CNTL));
+	/* 100ms per unit */
+	temp = REG_SET_FIELD(temp, SDMA0_WATCHDOG_CNTL, QUEUE_HANG_COUNT,
+			     max(adev->usec_timeout/100000, 1));
+	WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_WATCHDOG_CNTL), temp);
+
+	/* Set up RESP_MODE to non-copy addresses */
+	temp = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_UTCL1_CNTL));
+	temp = REG_SET_FIELD(temp, SDMA0_UTCL1_CNTL, RESP_MODE, 3);
+	temp = REG_SET_FIELD(temp, SDMA0_UTCL1_CNTL, REDO_DELAY, 9);
+	WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_UTCL1_CNTL), temp);
+
+	/* program default cache read and write policy */
+	temp = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_UTCL1_PAGE));
+	/* clean read policy and write policy bits */
+	temp &= 0xFF0FFF;
+	temp |= ((CACHE_READ_POLICY_L2__DEFAULT << 12) |
+		 (CACHE_WRITE_POLICY_L2__DEFAULT << 14) |
+		 SDMA0_UTCL1_PAGE__LLC_NOALLOC_MASK);
+	WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_UTCL1_PAGE), temp);
+
+	if (!amdgpu_sriov_vf(adev)) {
+		/* unhalt engine */
+		temp = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_F32_CNTL));
+		temp = REG_SET_FIELD(temp, SDMA0_F32_CNTL, HALT, 0);
+		temp = REG_SET_FIELD(temp, SDMA0_F32_CNTL, TH1_RESET, 0);
+		WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_F32_CNTL), temp);
+	}
+
+	/* enable DMA RB */
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RB_ENABLE, 1);
+	WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_CNTL), rb_cntl);
+
+	ib_cntl = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_IB_CNTL));
+	ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_QUEUE0_IB_CNTL, IB_ENABLE, 1);
+#ifdef __BIG_ENDIAN
+	ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_QUEUE0_IB_CNTL, IB_SWAP_ENABLE, 1);
+#endif
+	/* enable DMA IBs */
+	WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_IB_CNTL), ib_cntl);
+
+	if (amdgpu_sriov_vf(adev))
+		sdma_v6_0_enable(adev, true);
+
+	return amdgpu_ring_test_helper(ring);
+}
+
+/**
+ * sdma_v6_0_gfx_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the gfx DMA ring buffers and enable them.
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v6_0_gfx_resume(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		r = sdma_v6_0_gfx_resume_instance(adev, i, false);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+/**
+ * sdma_v6_0_rlc_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the compute DMA queues and enable them.
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v6_0_rlc_resume(struct amdgpu_device *adev)
+{
+	return 0;
+}
+
+/**
+ * sdma_v6_0_load_microcode - load the sDMA ME ucode
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Loads the sDMA0/1 ucode.
+ * Returns 0 for success, -EINVAL if the ucode is not available.
+ */
+static int sdma_v6_0_load_microcode(struct amdgpu_device *adev)
+{
+	const struct sdma_firmware_header_v2_0 *hdr;
+	const __le32 *fw_data;
+	u32 fw_size;
+	int i, j;
+	bool use_broadcast;
+
+	/* halt the MEs */
+	sdma_v6_0_enable(adev, false);
+
+	if (!adev->sdma.instance[0].fw)
+		return -EINVAL;
+
+	/* use broadcast mode to load SDMA microcode by default */
+	use_broadcast = true;
+
+	if (use_broadcast) {
+		dev_info(adev->dev, "Use broadcast method to load SDMA firmware\n");
+		/* load Control Thread microcode */
+		hdr = (const struct sdma_firmware_header_v2_0 *)adev->sdma.instance[0].fw->data;
+		amdgpu_ucode_print_sdma_hdr(&hdr->header);
+		fw_size = le32_to_cpu(hdr->ctx_jt_offset + hdr->ctx_jt_size) / 4;
+
+		fw_data = (const __le32 *)
+			(adev->sdma.instance[0].fw->data +
+				le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+
+		WREG32(sdma_v6_0_get_reg_offset(adev, 0, regSDMA0_BROADCAST_UCODE_ADDR), 0);
+
+		for (j = 0; j < fw_size; j++) {
+			if (amdgpu_emu_mode == 1 && j % 500 == 0)
+				msleep(1);
+			WREG32(sdma_v6_0_get_reg_offset(adev, 0, regSDMA0_BROADCAST_UCODE_DATA), le32_to_cpup(fw_data++));
+		}
+
+		/* load Context Switch microcode */
+		fw_size = le32_to_cpu(hdr->ctl_jt_offset + hdr->ctl_jt_size) / 4;
+
+		fw_data = (const __le32 *)
+			(adev->sdma.instance[0].fw->data +
+				le32_to_cpu(hdr->ctl_ucode_offset));
+
+		WREG32(sdma_v6_0_get_reg_offset(adev, 0, regSDMA0_BROADCAST_UCODE_ADDR), 0x8000);
+
+		for (j = 0; j < fw_size; j++) {
+			if (amdgpu_emu_mode == 1 && j % 500 == 0)
+				msleep(1);
+			WREG32(sdma_v6_0_get_reg_offset(adev, 0, regSDMA0_BROADCAST_UCODE_DATA), le32_to_cpup(fw_data++));
+		}
+	} else {
+		dev_info(adev->dev, "Use legacy method to load SDMA firmware\n");
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			/* load Control Thread microcode */
+			hdr = (const struct sdma_firmware_header_v2_0 *)adev->sdma.instance[0].fw->data;
+			amdgpu_ucode_print_sdma_hdr(&hdr->header);
+			fw_size = le32_to_cpu(hdr->ctx_jt_offset + hdr->ctx_jt_size) / 4;
+
+			fw_data = (const __le32 *)
+				(adev->sdma.instance[0].fw->data +
+					le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+
+			WREG32(sdma_v6_0_get_reg_offset(adev, i, regSDMA0_UCODE_ADDR), 0);
+
+			for (j = 0; j < fw_size; j++) {
+				if (amdgpu_emu_mode == 1 && j % 500 == 0)
+					msleep(1);
+				WREG32(sdma_v6_0_get_reg_offset(adev, i, regSDMA0_UCODE_DATA), le32_to_cpup(fw_data++));
+			}
+
+			WREG32(sdma_v6_0_get_reg_offset(adev, i, regSDMA0_UCODE_ADDR), adev->sdma.instance[0].fw_version);
+
+			/* load Context Switch microcode */
+			fw_size = le32_to_cpu(hdr->ctl_jt_offset + hdr->ctl_jt_size) / 4;
+
+			fw_data = (const __le32 *)
+				(adev->sdma.instance[0].fw->data +
+					le32_to_cpu(hdr->ctl_ucode_offset));
+
+			WREG32(sdma_v6_0_get_reg_offset(adev, i, regSDMA0_UCODE_ADDR), 0x8000);
+
+			for (j = 0; j < fw_size; j++) {
+				if (amdgpu_emu_mode == 1 && j % 500 == 0)
+					msleep(1);
+				WREG32(sdma_v6_0_get_reg_offset(adev, i, regSDMA0_UCODE_DATA), le32_to_cpup(fw_data++));
+			}
+
+			WREG32(sdma_v6_0_get_reg_offset(adev, i, regSDMA0_UCODE_ADDR), adev->sdma.instance[0].fw_version);
+		}
+	}
+
+	return 0;
+}
+
+static int sdma_v6_0_soft_reset(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	u32 tmp;
+	int i;
+
+	sdma_v6_0_gfx_stop(adev);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		tmp = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_FREEZE));
+		tmp |= SDMA0_FREEZE__FREEZE_MASK;
+		WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_FREEZE), tmp);
+		tmp = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_F32_CNTL));
+		tmp |= SDMA0_F32_CNTL__HALT_MASK;
+		tmp |= SDMA0_F32_CNTL__TH1_RESET_MASK;
+		WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_F32_CNTL), tmp);
+
+		WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_PREEMPT), 0);
+
+		udelay(100);
+
+		tmp = GRBM_SOFT_RESET__SOFT_RESET_SDMA0_MASK << i;
+		WREG32_SOC15(GC, 0, regGRBM_SOFT_RESET, tmp);
+		tmp = RREG32_SOC15(GC, 0, regGRBM_SOFT_RESET);
+
+		udelay(100);
+
+		WREG32_SOC15(GC, 0, regGRBM_SOFT_RESET, 0);
+		tmp = RREG32_SOC15(GC, 0, regGRBM_SOFT_RESET);
+
+		udelay(100);
+	}
+
+	return sdma_v6_0_start(adev);
+}
+
+static bool sdma_v6_0_check_soft_reset(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	int i, r;
+	long tmo = msecs_to_jiffies(1000);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		r = amdgpu_ring_test_ib(ring, tmo);
+		if (r)
+			return true;
+	}
+
+	return false;
+}
+
+/**
+ * sdma_v6_0_start - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the DMA engines and enable them.
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v6_0_start(struct amdgpu_device *adev)
+{
+	int r = 0;
+
+	if (amdgpu_sriov_vf(adev)) {
+		sdma_v6_0_enable(adev, false);
+
+		/* set RB registers */
+		r = sdma_v6_0_gfx_resume(adev);
+		return r;
+	}
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+		r = sdma_v6_0_load_microcode(adev);
+		if (r)
+			return r;
+
+		/* The value of regSDMA_F32_CNTL is invalid the moment after loading fw */
+		if (amdgpu_emu_mode == 1)
+			msleep(1000);
+	}
+
+	/* unhalt the MEs */
+	sdma_v6_0_enable(adev, true);
+	/* enable sdma ring preemption */
+	sdma_v6_0_ctxempty_int_enable(adev, true);
+
+	/* start the gfx rings and rlc compute queues */
+	r = sdma_v6_0_gfx_resume(adev);
+	if (r)
+		return r;
+	r = sdma_v6_0_rlc_resume(adev);
+
+	return r;
+}
+
+static int sdma_v6_0_mqd_init(struct amdgpu_device *adev, void *mqd,
+			      struct amdgpu_mqd_prop *prop)
+{
+	struct v11_sdma_mqd *m = mqd;
+	uint64_t wb_gpu_addr;
+
+	m->sdmax_rlcx_rb_cntl =
+		order_base_2(prop->queue_size / 4) << SDMA0_QUEUE0_RB_CNTL__RB_SIZE__SHIFT |
+		1 << SDMA0_QUEUE0_RB_CNTL__RPTR_WRITEBACK_ENABLE__SHIFT |
+		4 << SDMA0_QUEUE0_RB_CNTL__RPTR_WRITEBACK_TIMER__SHIFT |
+		1 << SDMA0_QUEUE0_RB_CNTL__F32_WPTR_POLL_ENABLE__SHIFT;
+
+	m->sdmax_rlcx_rb_base = lower_32_bits(prop->hqd_base_gpu_addr >> 8);
+	m->sdmax_rlcx_rb_base_hi = upper_32_bits(prop->hqd_base_gpu_addr >> 8);
+
+	wb_gpu_addr = prop->wptr_gpu_addr;
+	m->sdmax_rlcx_rb_wptr_poll_addr_lo = lower_32_bits(wb_gpu_addr);
+	m->sdmax_rlcx_rb_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr);
+
+	wb_gpu_addr = prop->rptr_gpu_addr;
+	m->sdmax_rlcx_rb_rptr_addr_lo = lower_32_bits(wb_gpu_addr);
+	m->sdmax_rlcx_rb_rptr_addr_hi = upper_32_bits(wb_gpu_addr);
+
+	m->sdmax_rlcx_ib_cntl = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, 0,
+							regSDMA0_QUEUE0_IB_CNTL));
+
+	m->sdmax_rlcx_doorbell_offset =
+		prop->doorbell_index << SDMA0_QUEUE0_DOORBELL_OFFSET__OFFSET__SHIFT;
+
+	m->sdmax_rlcx_doorbell = REG_SET_FIELD(0, SDMA0_QUEUE0_DOORBELL, ENABLE, 1);
+
+	m->sdmax_rlcx_skip_cntl = 0;
+	m->sdmax_rlcx_context_status = 0;
+	m->sdmax_rlcx_doorbell_log = 0;
+
+	m->sdmax_rlcx_rb_aql_cntl = regSDMA0_QUEUE0_RB_AQL_CNTL_DEFAULT;
+	m->sdmax_rlcx_dummy_reg = regSDMA0_QUEUE0_DUMMY_REG_DEFAULT;
+
+	m->sdmax_rlcx_csa_addr_lo = lower_32_bits(prop->csa_addr);
+	m->sdmax_rlcx_csa_addr_hi = upper_32_bits(prop->csa_addr);
+
+	m->sdmax_rlcx_f32_dbg0 = lower_32_bits(prop->fence_address);
+	m->sdmax_rlcx_f32_dbg1 = upper_32_bits(prop->fence_address);
+
+	return 0;
+}
+
+static void sdma_v6_0_set_mqd_funcs(struct amdgpu_device *adev)
+{
+	adev->mqds[AMDGPU_HW_IP_DMA].mqd_size = sizeof(struct v11_sdma_mqd);
+	adev->mqds[AMDGPU_HW_IP_DMA].init_mqd = sdma_v6_0_mqd_init;
+}
+
+/**
+ * sdma_v6_0_ring_test_ring - simple async dma engine test
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Test the DMA engine by writing using it to write an
+ * value to memory.
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v6_0_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned i;
+	unsigned index;
+	int r;
+	u32 tmp;
+	u64 gpu_addr;
+
+	tmp = 0xCAFEDEAD;
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+
+	r = amdgpu_ring_alloc(ring, 5);
+	if (r) {
+		DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r);
+		amdgpu_device_wb_free(adev, index);
+		return r;
+	}
+
+	amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_WRITE) |
+			  SDMA_PKT_COPY_LINEAR_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR));
+	amdgpu_ring_write(ring, lower_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(0));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = le32_to_cpu(adev->wb.wb[index]);
+		if (tmp == 0xDEADBEEF)
+			break;
+		if (amdgpu_emu_mode == 1)
+			msleep(1);
+		else
+			udelay(1);
+	}
+
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+
+	amdgpu_device_wb_free(adev, index);
+
+	return r;
+}
+
+/*
+ * sdma_v6_0_ring_test_ib - test an IB on the DMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @timeout: timeout value in jiffies, or MAX_SCHEDULE_TIMEOUT
+ *
+ * Test a simple IB in the DMA ring.
+ * Returns 0 on success, error on failure.
+ */
+static int sdma_v6_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct dma_fence *f = NULL;
+	unsigned index;
+	long r;
+	u32 tmp = 0;
+	u64 gpu_addr;
+
+	tmp = 0xCAFEDEAD;
+	memset(&ib, 0, sizeof(ib));
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+
+	r = amdgpu_ib_get(adev, NULL, 256, AMDGPU_IB_POOL_DIRECT, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
+		goto err0;
+	}
+
+	ib.ptr[0] = SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_WRITE) |
+		SDMA_PKT_COPY_LINEAR_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR);
+	ib.ptr[1] = lower_32_bits(gpu_addr);
+	ib.ptr[2] = upper_32_bits(gpu_addr);
+	ib.ptr[3] = SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(0);
+	ib.ptr[4] = 0xDEADBEEF;
+	ib.ptr[5] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.ptr[6] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.ptr[7] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.length_dw = 8;
+
+	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
+	if (r)
+		goto err1;
+
+	r = dma_fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out\n");
+		r = -ETIMEDOUT;
+		goto err1;
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
+		goto err1;
+	}
+
+	tmp = le32_to_cpu(adev->wb.wb[index]);
+
+	if (tmp == 0xDEADBEEF)
+		r = 0;
+	else
+		r = -EINVAL;
+
+err1:
+	amdgpu_ib_free(&ib, NULL);
+	dma_fence_put(f);
+err0:
+	amdgpu_device_wb_free(adev, index);
+	return r;
+}
+
+
+/**
+ * sdma_v6_0_vm_copy_pte - update PTEs by copying them from the GART
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @src: src addr to copy from
+ * @count: number of page entries to update
+ *
+ * Update PTEs by copying them from the GART using sDMA.
+ */
+static void sdma_v6_0_vm_copy_pte(struct amdgpu_ib *ib,
+				  uint64_t pe, uint64_t src,
+				  unsigned count)
+{
+	unsigned bytes = count * 8;
+
+	ib->ptr[ib->length_dw++] = SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_COPY) |
+		SDMA_PKT_COPY_LINEAR_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
+	ib->ptr[ib->length_dw++] = bytes - 1;
+	ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+	ib->ptr[ib->length_dw++] = lower_32_bits(src);
+	ib->ptr[ib->length_dw++] = upper_32_bits(src);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+
+}
+
+/**
+ * sdma_v6_0_vm_write_pte - update PTEs by writing them manually
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @value: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ *
+ * Update PTEs by writing them manually using sDMA.
+ */
+static void sdma_v6_0_vm_write_pte(struct amdgpu_ib *ib, uint64_t pe,
+				   uint64_t value, unsigned count,
+				   uint32_t incr)
+{
+	unsigned ndw = count * 2;
+
+	ib->ptr[ib->length_dw++] = SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_WRITE) |
+		SDMA_PKT_COPY_LINEAR_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+	ib->ptr[ib->length_dw++] = ndw - 1;
+	for (; ndw > 0; ndw -= 2) {
+		ib->ptr[ib->length_dw++] = lower_32_bits(value);
+		ib->ptr[ib->length_dw++] = upper_32_bits(value);
+		value += incr;
+	}
+}
+
+/**
+ * sdma_v6_0_vm_set_pte_pde - update the page tables using sDMA
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: access flags
+ *
+ * Update the page tables using sDMA.
+ */
+static void sdma_v6_0_vm_set_pte_pde(struct amdgpu_ib *ib,
+				     uint64_t pe,
+				     uint64_t addr, unsigned count,
+				     uint32_t incr, uint64_t flags)
+{
+	/* for physically contiguous pages (vram) */
+	ib->ptr[ib->length_dw++] = SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_PTEPDE);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe); /* dst addr */
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+	ib->ptr[ib->length_dw++] = lower_32_bits(flags); /* mask */
+	ib->ptr[ib->length_dw++] = upper_32_bits(flags);
+	ib->ptr[ib->length_dw++] = lower_32_bits(addr); /* value */
+	ib->ptr[ib->length_dw++] = upper_32_bits(addr);
+	ib->ptr[ib->length_dw++] = incr; /* increment size */
+	ib->ptr[ib->length_dw++] = 0;
+	ib->ptr[ib->length_dw++] = count - 1; /* number of entries */
+}
+
+/*
+ * sdma_v6_0_ring_pad_ib - pad the IB
+ * @ib: indirect buffer to fill with padding
+ * @ring: amdgpu ring pointer
+ *
+ * Pad the IB with NOPs to a boundary multiple of 8.
+ */
+static void sdma_v6_0_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring);
+	u32 pad_count;
+	int i;
+
+	pad_count = (-ib->length_dw) & 0x7;
+	for (i = 0; i < pad_count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			ib->ptr[ib->length_dw++] =
+				SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_NOP) |
+				SDMA_PKT_NOP_HEADER_COUNT(pad_count - 1);
+		else
+			ib->ptr[ib->length_dw++] =
+				SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_NOP);
+}
+
+/**
+ * sdma_v6_0_ring_emit_pipeline_sync - sync the pipeline
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Make sure all previous operations are completed (CIK).
+ */
+static void sdma_v6_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	/* wait for idle */
+	amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(0) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3) | /* equal */
+			  SDMA_PKT_POLL_REGMEM_HEADER_MEM_POLL(1));
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xffffffff);
+	amdgpu_ring_write(ring, seq); /* reference */
+	amdgpu_ring_write(ring, 0xffffffff); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(4)); /* retry count, poll interval */
+}
+
+/*
+ * sdma_v6_0_ring_emit_vm_flush - vm flush using sDMA
+ *
+ * @ring: amdgpu_ring pointer
+ * @vmid: vmid number to use
+ * @pd_addr: address
+ *
+ * Update the page table base and flush the VM TLB
+ * using sDMA.
+ */
+static void sdma_v6_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					 unsigned vmid, uint64_t pd_addr)
+{
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
+	uint32_t req = hub->vmhub_funcs->get_invalidate_req(vmid, 0);
+
+	/* Update the PD address for this VMID. */
+	amdgpu_ring_emit_wreg(ring, hub->ctx0_ptb_addr_lo32 +
+			      (hub->ctx_addr_distance * vmid),
+			      lower_32_bits(pd_addr));
+	amdgpu_ring_emit_wreg(ring, hub->ctx0_ptb_addr_hi32 +
+			      (hub->ctx_addr_distance * vmid),
+			      upper_32_bits(pd_addr));
+
+	/* Trigger invalidation. */
+	amdgpu_ring_write(ring,
+			  SDMA_PKT_VM_INVALIDATION_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_VM_INVALIDATION_HEADER_SUB_OP(SDMA_SUBOP_VM_INVALIDATION) |
+			  SDMA_PKT_VM_INVALIDATION_HEADER_GFX_ENG_ID(ring->vm_inv_eng) |
+			  SDMA_PKT_VM_INVALIDATION_HEADER_MM_ENG_ID(0x1f));
+	amdgpu_ring_write(ring, req);
+	amdgpu_ring_write(ring, 0xFFFFFFFF);
+	amdgpu_ring_write(ring,
+			  SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_INVALIDATEACK(1 << vmid) |
+			  SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_ADDRESSRANGEHI(0x1F));
+}
+
+static void sdma_v6_0_ring_emit_wreg(struct amdgpu_ring *ring,
+				     uint32_t reg, uint32_t val)
+{
+	amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_SRBM_WRITE) |
+			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, val);
+}
+
+static void sdma_v6_0_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+					 uint32_t val, uint32_t mask)
+{
+	amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(0) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3)); /* equal */
+	amdgpu_ring_write(ring, reg << 2);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, val); /* reference */
+	amdgpu_ring_write(ring, mask); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10));
+}
+
+static void sdma_v6_0_ring_emit_reg_write_reg_wait(struct amdgpu_ring *ring,
+						   uint32_t reg0, uint32_t reg1,
+						   uint32_t ref, uint32_t mask)
+{
+	amdgpu_ring_emit_wreg(ring, reg0, ref);
+	/* wait for a cycle to reset vm_inv_eng*_ack */
+	amdgpu_ring_emit_reg_wait(ring, reg0, 0, 0);
+	amdgpu_ring_emit_reg_wait(ring, reg1, mask, mask);
+}
+
+static struct amdgpu_sdma_ras sdma_v6_0_3_ras = {
+	.ras_block = {
+		.ras_late_init = amdgpu_ras_block_late_init,
+	},
+};
+
+static void sdma_v6_0_set_ras_funcs(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) {
+	case IP_VERSION(6, 0, 3):
+		adev->sdma.ras = &sdma_v6_0_3_ras;
+		break;
+	default:
+		break;
+	}
+}
+
+static int sdma_v6_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	switch (amdgpu_user_queue) {
+	case -1:
+	case 0:
+	default:
+		adev->sdma.no_user_submission = false;
+		adev->sdma.disable_uq = true;
+		break;
+	case 1:
+		adev->sdma.no_user_submission = false;
+		adev->sdma.disable_uq = false;
+		break;
+	case 2:
+		adev->sdma.no_user_submission = true;
+		adev->sdma.disable_uq = false;
+		break;
+	}
+
+	r = amdgpu_sdma_init_microcode(adev, 0, true);
+	if (r)
+		return r;
+
+	sdma_v6_0_set_ring_funcs(adev);
+	sdma_v6_0_set_buffer_funcs(adev);
+	sdma_v6_0_set_vm_pte_funcs(adev);
+	sdma_v6_0_set_irq_funcs(adev);
+	sdma_v6_0_set_mqd_funcs(adev);
+	sdma_v6_0_set_ras_funcs(adev);
+
+	return 0;
+}
+
+static int sdma_v6_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_ring *ring;
+	int r, i;
+	struct amdgpu_device *adev = ip_block->adev;
+	uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_6_0);
+	uint32_t *ptr;
+
+	/* SDMA trap event */
+	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GFX,
+			      GFX_11_0_0__SRCID__SDMA_TRAP,
+			      &adev->sdma.trap_irq);
+	if (r)
+		return r;
+
+	/* SDMA user fence event */
+	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GFX,
+			      GFX_11_0_0__SRCID__SDMA_FENCE,
+			      &adev->sdma.fence_irq);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		ring->ring_obj = NULL;
+		ring->use_doorbell = true;
+		ring->me = i;
+		ring->no_user_submission = adev->sdma.no_user_submission;
+
+		DRM_DEBUG("SDMA %d use_doorbell being set to: [%s]\n", i,
+				ring->use_doorbell?"true":"false");
+
+		ring->doorbell_index =
+			(adev->doorbell_index.sdma_engine[i] << 1); // get DWORD offset
+
+		ring->vm_hub = AMDGPU_GFXHUB(0);
+		sprintf(ring->name, "sdma%d", i);
+		r = amdgpu_ring_init(adev, ring, 1024,
+				     &adev->sdma.trap_irq,
+				     AMDGPU_SDMA_IRQ_INSTANCE0 + i,
+				     AMDGPU_RING_PRIO_DEFAULT, NULL);
+		if (r)
+			return r;
+	}
+
+	adev->sdma.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->sdma.instance[0].ring);
+	switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) {
+	case IP_VERSION(6, 0, 0):
+	case IP_VERSION(6, 0, 2):
+	case IP_VERSION(6, 0, 3):
+		if ((adev->sdma.instance[0].fw_version >= 21) &&
+		    !amdgpu_sriov_vf(adev))
+			adev->sdma.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+		break;
+	default:
+		break;
+	}
+
+	if (amdgpu_sdma_ras_sw_init(adev)) {
+		dev_err(adev->dev, "Failed to initialize sdma ras block!\n");
+		return -EINVAL;
+	}
+
+	/* Allocate memory for SDMA IP Dump buffer */
+	ptr = kcalloc(adev->sdma.num_instances * reg_count, sizeof(uint32_t), GFP_KERNEL);
+	if (ptr)
+		adev->sdma.ip_dump = ptr;
+	else
+		DRM_ERROR("Failed to allocated memory for SDMA IP Dump\n");
+
+	switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) {
+	case IP_VERSION(6, 0, 0):
+		if ((adev->sdma.instance[0].fw_version >= 27) && !adev->sdma.disable_uq)
+			adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs;
+		break;
+	case IP_VERSION(6, 0, 1):
+		if ((adev->sdma.instance[0].fw_version >= 18) && !adev->sdma.disable_uq)
+			adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs;
+		break;
+	case IP_VERSION(6, 0, 2):
+		if ((adev->sdma.instance[0].fw_version >= 23) && !adev->sdma.disable_uq)
+			adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs;
+		break;
+	case IP_VERSION(6, 0, 3):
+		if ((adev->sdma.instance[0].fw_version >= 27) && !adev->sdma.disable_uq)
+			adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs;
+		break;
+	case IP_VERSION(6, 1, 0):
+		if ((adev->sdma.instance[0].fw_version >= 14) && !adev->sdma.disable_uq)
+			adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs;
+		break;
+	case IP_VERSION(6, 1, 1):
+		if ((adev->sdma.instance[0].fw_version >= 17) && !adev->sdma.disable_uq)
+			adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs;
+		break;
+	case IP_VERSION(6, 1, 2):
+		if ((adev->sdma.instance[0].fw_version >= 15) && !adev->sdma.disable_uq)
+			adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs;
+		break;
+	case IP_VERSION(6, 1, 3):
+		if ((adev->sdma.instance[0].fw_version >= 10) && !adev->sdma.disable_uq)
+			adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs;
+		break;
+	default:
+		break;
+	}
+
+	r = amdgpu_sdma_sysfs_reset_mask_init(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int sdma_v6_0_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		amdgpu_ring_fini(&adev->sdma.instance[i].ring);
+
+	amdgpu_sdma_sysfs_reset_mask_fini(adev);
+	amdgpu_sdma_destroy_inst_ctx(adev, true);
+
+	kfree(adev->sdma.ip_dump);
+
+	return 0;
+}
+
+static int sdma_v6_0_set_userq_trap_interrupts(struct amdgpu_device *adev,
+					       bool enable)
+{
+	unsigned int irq_type;
+	int i, r;
+
+	if (adev->userq_funcs[AMDGPU_HW_IP_DMA]) {
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			irq_type = AMDGPU_SDMA_IRQ_INSTANCE0 + i;
+			if (enable)
+				r = amdgpu_irq_get(adev, &adev->sdma.trap_irq,
+						   irq_type);
+			else
+				r = amdgpu_irq_put(adev, &adev->sdma.trap_irq,
+						   irq_type);
+			if (r)
+				return r;
+		}
+	}
+
+	return 0;
+}
+
+static int sdma_v6_0_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = sdma_v6_0_start(adev);
+	if (r)
+		return r;
+
+	return sdma_v6_0_set_userq_trap_interrupts(adev, true);
+}
+
+static int sdma_v6_0_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	sdma_v6_0_ctxempty_int_enable(adev, false);
+	sdma_v6_0_enable(adev, false);
+	sdma_v6_0_set_userq_trap_interrupts(adev, false);
+
+	return 0;
+}
+
+static int sdma_v6_0_suspend(struct amdgpu_ip_block *ip_block)
+{
+	return sdma_v6_0_hw_fini(ip_block);
+}
+
+static int sdma_v6_0_resume(struct amdgpu_ip_block *ip_block)
+{
+	return sdma_v6_0_hw_init(ip_block);
+}
+
+static bool sdma_v6_0_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	u32 i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		u32 tmp = RREG32(sdma_v6_0_get_reg_offset(adev, i, regSDMA0_STATUS_REG));
+
+		if (!(tmp & SDMA0_STATUS_REG__IDLE_MASK))
+			return false;
+	}
+
+	return true;
+}
+
+static int sdma_v6_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	unsigned i;
+	u32 sdma0, sdma1;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		sdma0 = RREG32(sdma_v6_0_get_reg_offset(adev, 0, regSDMA0_STATUS_REG));
+		sdma1 = RREG32(sdma_v6_0_get_reg_offset(adev, 1, regSDMA0_STATUS_REG));
+
+		if (sdma0 & sdma1 & SDMA0_STATUS_REG__IDLE_MASK)
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static int sdma_v6_0_ring_preempt_ib(struct amdgpu_ring *ring)
+{
+	int i, r = 0;
+	struct amdgpu_device *adev = ring->adev;
+	u32 index = 0;
+	u64 sdma_gfx_preempt;
+
+	amdgpu_sdma_get_index_from_ring(ring, &index);
+	sdma_gfx_preempt =
+		sdma_v6_0_get_reg_offset(adev, index, regSDMA0_QUEUE0_PREEMPT);
+
+	/* assert preemption condition */
+	amdgpu_ring_set_preempt_cond_exec(ring, false);
+
+	/* emit the trailing fence */
+	ring->trail_seq += 1;
+	amdgpu_ring_alloc(ring, 10);
+	sdma_v6_0_ring_emit_fence(ring, ring->trail_fence_gpu_addr,
+				  ring->trail_seq, 0);
+	amdgpu_ring_commit(ring);
+
+	/* assert IB preemption */
+	WREG32(sdma_gfx_preempt, 1);
+
+	/* poll the trailing fence */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (ring->trail_seq ==
+		    le32_to_cpu(*(ring->trail_fence_cpu_addr)))
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout) {
+		r = -EINVAL;
+		DRM_ERROR("ring %d failed to be preempted\n", ring->idx);
+	}
+
+	/* deassert IB preemption */
+	WREG32(sdma_gfx_preempt, 0);
+
+	/* deassert the preemption condition */
+	amdgpu_ring_set_preempt_cond_exec(ring, true);
+	return r;
+}
+
+static int sdma_v6_0_reset_queue(struct amdgpu_ring *ring,
+				 unsigned int vmid,
+				 struct amdgpu_fence *timedout_fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int r;
+
+	if (ring->me >= adev->sdma.num_instances) {
+		dev_err(adev->dev, "sdma instance not found\n");
+		return -EINVAL;
+	}
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+
+	r = amdgpu_mes_reset_legacy_queue(adev, ring, vmid, true);
+	if (r)
+		return r;
+
+	r = sdma_v6_0_gfx_resume_instance(adev, ring->me, true);
+	if (r)
+		return r;
+
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static int sdma_v6_0_set_trap_irq_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	u32 sdma_cntl;
+
+	u32 reg_offset = sdma_v6_0_get_reg_offset(adev, type, regSDMA0_CNTL);
+
+	if (!amdgpu_sriov_vf(adev)) {
+		sdma_cntl = RREG32(reg_offset);
+		sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE,
+				state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+		WREG32(reg_offset, sdma_cntl);
+	}
+
+	return 0;
+}
+
+static int sdma_v6_0_process_trap_irq(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	int instances, queue;
+
+	DRM_DEBUG("IH: SDMA trap\n");
+
+	queue = entry->ring_id & 0xf;
+	instances = (entry->ring_id & 0xf0) >> 4;
+	if (instances > 1) {
+		DRM_ERROR("IH: wrong ring_ID detected, as wrong sdma instance\n");
+		return -EINVAL;
+	}
+
+	switch (entry->client_id) {
+	case SOC21_IH_CLIENTID_GFX:
+		switch (queue) {
+		case 0:
+			amdgpu_fence_process(&adev->sdma.instance[instances].ring);
+			break;
+		default:
+			break;
+		}
+		break;
+	}
+	return 0;
+}
+
+static int sdma_v6_0_process_fence_irq(struct amdgpu_device *adev,
+				       struct amdgpu_irq_src *source,
+				       struct amdgpu_iv_entry *entry)
+{
+	u32 doorbell_offset = entry->src_data[0];
+
+	if (adev->enable_mes && doorbell_offset) {
+		struct amdgpu_userq_fence_driver *fence_drv = NULL;
+		struct xarray *xa = &adev->userq_xa;
+		unsigned long flags;
+
+		doorbell_offset >>= SDMA0_QUEUE0_DOORBELL_OFFSET__OFFSET__SHIFT;
+
+		xa_lock_irqsave(xa, flags);
+		fence_drv = xa_load(xa, doorbell_offset);
+		if (fence_drv)
+			amdgpu_userq_fence_driver_process(fence_drv);
+		xa_unlock_irqrestore(xa, flags);
+	}
+
+	return 0;
+}
+
+static int sdma_v6_0_process_illegal_inst_irq(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      struct amdgpu_iv_entry *entry)
+{
+	return 0;
+}
+
+static int sdma_v6_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					   enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int sdma_v6_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static void sdma_v6_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
+{
+}
+
+static void sdma_v6_0_print_ip_state(struct amdgpu_ip_block *ip_block, struct drm_printer *p)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, j;
+	uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_6_0);
+	uint32_t instance_offset;
+
+	if (!adev->sdma.ip_dump)
+		return;
+
+	drm_printf(p, "num_instances:%d\n", adev->sdma.num_instances);
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		instance_offset = i * reg_count;
+		drm_printf(p, "\nInstance:%d\n", i);
+
+		for (j = 0; j < reg_count; j++)
+			drm_printf(p, "%-50s \t 0x%08x\n", sdma_reg_list_6_0[j].reg_name,
+				   adev->sdma.ip_dump[instance_offset + j]);
+	}
+}
+
+static void sdma_v6_0_dump_ip_state(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, j;
+	uint32_t instance_offset;
+	uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_6_0);
+
+	if (!adev->sdma.ip_dump)
+		return;
+
+	amdgpu_gfx_off_ctrl(adev, false);
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		instance_offset = i * reg_count;
+		for (j = 0; j < reg_count; j++)
+			adev->sdma.ip_dump[instance_offset + j] =
+				RREG32(sdma_v6_0_get_reg_offset(adev, i,
+				       sdma_reg_list_6_0[j].reg_offset));
+	}
+	amdgpu_gfx_off_ctrl(adev, true);
+}
+
+const struct amd_ip_funcs sdma_v6_0_ip_funcs = {
+	.name = "sdma_v6_0",
+	.early_init = sdma_v6_0_early_init,
+	.sw_init = sdma_v6_0_sw_init,
+	.sw_fini = sdma_v6_0_sw_fini,
+	.hw_init = sdma_v6_0_hw_init,
+	.hw_fini = sdma_v6_0_hw_fini,
+	.suspend = sdma_v6_0_suspend,
+	.resume = sdma_v6_0_resume,
+	.is_idle = sdma_v6_0_is_idle,
+	.wait_for_idle = sdma_v6_0_wait_for_idle,
+	.soft_reset = sdma_v6_0_soft_reset,
+	.check_soft_reset = sdma_v6_0_check_soft_reset,
+	.set_clockgating_state = sdma_v6_0_set_clockgating_state,
+	.set_powergating_state = sdma_v6_0_set_powergating_state,
+	.get_clockgating_state = sdma_v6_0_get_clockgating_state,
+	.dump_ip_state = sdma_v6_0_dump_ip_state,
+	.print_ip_state = sdma_v6_0_print_ip_state,
+};
+
+static const struct amdgpu_ring_funcs sdma_v6_0_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_SDMA,
+	.align_mask = 0xf,
+	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
+	.support_64bit_ptrs = true,
+	.secure_submission_supported = true,
+	.get_rptr = sdma_v6_0_ring_get_rptr,
+	.get_wptr = sdma_v6_0_ring_get_wptr,
+	.set_wptr = sdma_v6_0_ring_set_wptr,
+	.emit_frame_size =
+		5 + /* sdma_v6_0_ring_init_cond_exec */
+		6 + /* sdma_v6_0_ring_emit_hdp_flush */
+		6 + /* sdma_v6_0_ring_emit_pipeline_sync */
+		/* sdma_v6_0_ring_emit_vm_flush */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 6 +
+		10 + 10 + 10, /* sdma_v6_0_ring_emit_fence x3 for user fence, vm fence */
+	.emit_ib_size = 5 + 7 + 6, /* sdma_v6_0_ring_emit_ib */
+	.emit_ib = sdma_v6_0_ring_emit_ib,
+	.emit_mem_sync = sdma_v6_0_ring_emit_mem_sync,
+	.emit_fence = sdma_v6_0_ring_emit_fence,
+	.emit_pipeline_sync = sdma_v6_0_ring_emit_pipeline_sync,
+	.emit_vm_flush = sdma_v6_0_ring_emit_vm_flush,
+	.emit_hdp_flush = sdma_v6_0_ring_emit_hdp_flush,
+	.test_ring = sdma_v6_0_ring_test_ring,
+	.test_ib = sdma_v6_0_ring_test_ib,
+	.insert_nop = sdma_v6_0_ring_insert_nop,
+	.pad_ib = sdma_v6_0_ring_pad_ib,
+	.emit_wreg = sdma_v6_0_ring_emit_wreg,
+	.emit_reg_wait = sdma_v6_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = sdma_v6_0_ring_emit_reg_write_reg_wait,
+	.init_cond_exec = sdma_v6_0_ring_init_cond_exec,
+	.preempt_ib = sdma_v6_0_ring_preempt_ib,
+	.reset = sdma_v6_0_reset_queue,
+};
+
+static void sdma_v6_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		adev->sdma.instance[i].ring.funcs = &sdma_v6_0_ring_funcs;
+		adev->sdma.instance[i].ring.me = i;
+	}
+}
+
+static const struct amdgpu_irq_src_funcs sdma_v6_0_trap_irq_funcs = {
+	.set = sdma_v6_0_set_trap_irq_state,
+	.process = sdma_v6_0_process_trap_irq,
+};
+
+static const struct amdgpu_irq_src_funcs sdma_v6_0_fence_irq_funcs = {
+	.process = sdma_v6_0_process_fence_irq,
+};
+
+static const struct amdgpu_irq_src_funcs sdma_v6_0_illegal_inst_irq_funcs = {
+	.process = sdma_v6_0_process_illegal_inst_irq,
+};
+
+static void sdma_v6_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->sdma.trap_irq.num_types = AMDGPU_SDMA_IRQ_INSTANCE0 +
+					adev->sdma.num_instances;
+	adev->sdma.trap_irq.funcs = &sdma_v6_0_trap_irq_funcs;
+	adev->sdma.fence_irq.funcs = &sdma_v6_0_fence_irq_funcs;
+	adev->sdma.illegal_inst_irq.funcs = &sdma_v6_0_illegal_inst_irq_funcs;
+}
+
+/**
+ * sdma_v6_0_emit_copy_buffer - copy buffer using the sDMA engine
+ *
+ * @ib: indirect buffer to fill with commands
+ * @src_offset: src GPU address
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ * @copy_flags: copy flags for the buffers
+ *
+ * Copy GPU buffers using the DMA engine.
+ * Used by the amdgpu ttm implementation to move pages if
+ * registered as the asic copy callback.
+ */
+static void sdma_v6_0_emit_copy_buffer(struct amdgpu_ib *ib,
+				       uint64_t src_offset,
+				       uint64_t dst_offset,
+				       uint32_t byte_count,
+				       uint32_t copy_flags)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_COPY) |
+		SDMA_PKT_COPY_LINEAR_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR) |
+		SDMA_PKT_COPY_LINEAR_HEADER_TMZ((copy_flags & AMDGPU_COPY_FLAGS_TMZ) ? 1 : 0);
+	ib->ptr[ib->length_dw++] = byte_count - 1;
+	ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+	ib->ptr[ib->length_dw++] = lower_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+}
+
+/**
+ * sdma_v6_0_emit_fill_buffer - fill buffer using the sDMA engine
+ *
+ * @ib: indirect buffer to fill
+ * @src_data: value to write to buffer
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ *
+ * Fill GPU buffers using the DMA engine.
+ */
+static void sdma_v6_0_emit_fill_buffer(struct amdgpu_ib *ib,
+				       uint32_t src_data,
+				       uint64_t dst_offset,
+				       uint32_t byte_count)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PKT_CONSTANT_FILL_HEADER_OP(SDMA_OP_CONST_FILL);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = src_data;
+	ib->ptr[ib->length_dw++] = byte_count - 1;
+}
+
+static const struct amdgpu_buffer_funcs sdma_v6_0_buffer_funcs = {
+	.copy_max_bytes = 0x400000,
+	.copy_num_dw = 7,
+	.emit_copy_buffer = sdma_v6_0_emit_copy_buffer,
+
+	.fill_max_bytes = 0x400000,
+	.fill_num_dw = 5,
+	.emit_fill_buffer = sdma_v6_0_emit_fill_buffer,
+};
+
+static void sdma_v6_0_set_buffer_funcs(struct amdgpu_device *adev)
+{
+	adev->mman.buffer_funcs = &sdma_v6_0_buffer_funcs;
+	adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
+}
+
+static const struct amdgpu_vm_pte_funcs sdma_v6_0_vm_pte_funcs = {
+	.copy_pte_num_dw = 7,
+	.copy_pte = sdma_v6_0_vm_copy_pte,
+	.write_pte = sdma_v6_0_vm_write_pte,
+	.set_pte_pde = sdma_v6_0_vm_set_pte_pde,
+};
+
+static void sdma_v6_0_set_vm_pte_funcs(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	adev->vm_manager.vm_pte_funcs = &sdma_v6_0_vm_pte_funcs;
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		adev->vm_manager.vm_pte_scheds[i] =
+			&adev->sdma.instance[i].ring.sched;
+	}
+	adev->vm_manager.vm_pte_num_scheds = adev->sdma.num_instances;
+}
+
+const struct amdgpu_ip_block_version sdma_v6_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_SDMA,
+	.major = 6,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &sdma_v6_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.h b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.h
new file mode 100644
index 000000000000..e473ec7dfc8f
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __SDMA_V6_0_H__
+#define __SDMA_V6_0_H__
+
+extern const struct amd_ip_funcs sdma_v6_0_ip_funcs;
+extern const struct amdgpu_ip_block_version sdma_v6_0_ip_block;
+
+#endif /* __SDMA_V6_0_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0_0_pkt_open.h b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0_0_pkt_open.h
new file mode 100644
index 000000000000..d8cf830916b9
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0_0_pkt_open.h
@@ -0,0 +1,5672 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __SDMA_V6_0_0_PKT_OPEN_H_
+#define __SDMA_V6_0_0_PKT_OPEN_H_
+
+#define SDMA_OP_NOP  0
+#define SDMA_OP_COPY  1
+#define SDMA_OP_WRITE  2
+#define SDMA_OP_INDIRECT  4
+#define SDMA_OP_FENCE  5
+#define SDMA_OP_TRAP  6
+#define SDMA_OP_SEM  7
+#define SDMA_OP_POLL_REGMEM  8
+#define SDMA_OP_COND_EXE  9
+#define SDMA_OP_ATOMIC  10
+#define SDMA_OP_CONST_FILL  11
+#define SDMA_OP_PTEPDE  12
+#define SDMA_OP_TIMESTAMP  13
+#define SDMA_OP_SRBM_WRITE  14
+#define SDMA_OP_PRE_EXE  15
+#define SDMA_OP_GPUVM_INV  16
+#define SDMA_OP_GCR_REQ  17
+#define SDMA_OP_DUMMY_TRAP  32
+#define SDMA_SUBOP_TIMESTAMP_SET  0
+#define SDMA_SUBOP_TIMESTAMP_GET  1
+#define SDMA_SUBOP_TIMESTAMP_GET_GLOBAL  2
+#define SDMA_SUBOP_COPY_LINEAR  0
+#define SDMA_SUBOP_COPY_LINEAR_SUB_WIND  4
+#define SDMA_SUBOP_COPY_TILED  1
+#define SDMA_SUBOP_COPY_TILED_SUB_WIND  5
+#define SDMA_SUBOP_COPY_T2T_SUB_WIND  6
+#define SDMA_SUBOP_COPY_SOA  3
+#define SDMA_SUBOP_COPY_DIRTY_PAGE  7
+#define SDMA_SUBOP_COPY_LINEAR_PHY  8
+#define SDMA_SUBOP_COPY_LINEAR_SUB_WIND_LARGE  36
+#define SDMA_SUBOP_COPY_LINEAR_BC  16
+#define SDMA_SUBOP_COPY_TILED_BC  17
+#define SDMA_SUBOP_COPY_LINEAR_SUB_WIND_BC  20
+#define SDMA_SUBOP_COPY_TILED_SUB_WIND_BC  21
+#define SDMA_SUBOP_COPY_T2T_SUB_WIND_BC  22
+#define SDMA_SUBOP_WRITE_LINEAR  0
+#define SDMA_SUBOP_WRITE_TILED  1
+#define SDMA_SUBOP_WRITE_TILED_BC  17
+#define SDMA_SUBOP_PTEPDE_GEN  0
+#define SDMA_SUBOP_PTEPDE_COPY  1
+#define SDMA_SUBOP_PTEPDE_RMW  2
+#define SDMA_SUBOP_PTEPDE_COPY_BACKWARDS  3
+#define SDMA_SUBOP_MEM_INCR  1
+#define SDMA_SUBOP_DATA_FILL_MULTI  1
+#define SDMA_SUBOP_POLL_REG_WRITE_MEM  1
+#define SDMA_SUBOP_POLL_DBIT_WRITE_MEM  2
+#define SDMA_SUBOP_POLL_MEM_VERIFY  3
+#define SDMA_SUBOP_VM_INVALIDATION  4
+#define HEADER_AGENT_DISPATCH  4
+#define HEADER_BARRIER  5
+#define SDMA_OP_AQL_COPY  0
+#define SDMA_OP_AQL_BARRIER_OR  0
+
+#define SDMA_GCR_RANGE_IS_PA		(1 << 18)
+#define SDMA_GCR_SEQ(x)			(((x) & 0x3) << 16)
+#define SDMA_GCR_GL2_WB			(1 << 15)
+#define SDMA_GCR_GL2_INV		(1 << 14)
+#define SDMA_GCR_GL2_DISCARD		(1 << 13)
+#define SDMA_GCR_GL2_RANGE(x)		(((x) & 0x3) << 11)
+#define SDMA_GCR_GL2_US			(1 << 10)
+#define SDMA_GCR_GL1_INV		(1 << 9)
+#define SDMA_GCR_GLV_INV		(1 << 8)
+#define SDMA_GCR_GLK_INV		(1 << 7)
+#define SDMA_GCR_GLK_WB			(1 << 6)
+#define SDMA_GCR_GLM_INV		(1 << 5)
+#define SDMA_GCR_GLM_WB			(1 << 4)
+#define SDMA_GCR_GL1_RANGE(x)		(((x) & 0x3) << 2)
+#define SDMA_GCR_GLI_INV(x)		(((x) & 0x3) << 0)
+
+#define SDMA_DCC_DATA_FORMAT(x) ((x) & 0x3f)
+#define SDMA_DCC_NUM_TYPE(x) (((x) & 0x7) << 9)
+#define SDMA_DCC_READ_CM(x) (((x) & 0x3) << 16)
+#define SDMA_DCC_WRITE_CM(x) (((x) & 0x3) << 18)
+#define SDMA_DCC_MAX_COM(x) (((x) & 0x3) << 24)
+#define SDMA_DCC_MAX_UCOM(x) (((x) & 0x1) << 26)
+
+/*
+** Definitions for SDMA_PKT_COPY_LINEAR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_HEADER_op_shift  0
+#define SDMA_PKT_COPY_LINEAR_HEADER_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_op_mask) << SDMA_PKT_COPY_LINEAR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_LINEAR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_sub_op_mask) << SDMA_PKT_COPY_LINEAR_HEADER_sub_op_shift)
+
+/*define for encrypt field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_encrypt_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_encrypt_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_HEADER_encrypt_shift  16
+#define SDMA_PKT_COPY_LINEAR_HEADER_ENCRYPT(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_encrypt_mask) << SDMA_PKT_COPY_LINEAR_HEADER_encrypt_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_LINEAR_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_tmz_mask) << SDMA_PKT_COPY_LINEAR_HEADER_tmz_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_cpv_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_HEADER_cpv_shift  19
+#define SDMA_PKT_COPY_LINEAR_HEADER_CPV(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_cpv_mask) << SDMA_PKT_COPY_LINEAR_HEADER_cpv_shift)
+
+/*define for backwards field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_backwards_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_backwards_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_HEADER_backwards_shift  25
+#define SDMA_PKT_COPY_LINEAR_HEADER_BACKWARDS(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_backwards_mask) << SDMA_PKT_COPY_LINEAR_HEADER_backwards_shift)
+
+/*define for broadcast field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_broadcast_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_broadcast_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_HEADER_broadcast_shift  27
+#define SDMA_PKT_COPY_LINEAR_HEADER_BROADCAST(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_broadcast_mask) << SDMA_PKT_COPY_LINEAR_HEADER_broadcast_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_LINEAR_COUNT_count_offset 1
+#define SDMA_PKT_COPY_LINEAR_COUNT_count_mask   0x3FFFFFFF
+#define SDMA_PKT_COPY_LINEAR_COUNT_count_shift  0
+#define SDMA_PKT_COPY_LINEAR_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_LINEAR_COUNT_count_mask) << SDMA_PKT_COPY_LINEAR_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_shift  16
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_DST_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_shift)
+
+/*define for dst_cache_policy field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_cache_policy_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_cache_policy_mask   0x00000007
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_cache_policy_shift  18
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_DST_CACHE_POLICY(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_dst_cache_policy_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_dst_cache_policy_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_shift  24
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_SRC_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_shift)
+
+/*define for src_cache_policy field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_cache_policy_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_cache_policy_mask   0x00000007
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_cache_policy_shift  26
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_SRC_CACHE_POLICY(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_src_cache_policy_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_src_cache_policy_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_offset 3
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_offset 4
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_offset 5
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_offset 6
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_LINEAR_BC packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_LINEAR_BC_HEADER_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_BC_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_BC_HEADER_op_shift  0
+#define SDMA_PKT_COPY_LINEAR_BC_HEADER_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_BC_HEADER_op_mask) << SDMA_PKT_COPY_LINEAR_BC_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_LINEAR_BC_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_BC_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_BC_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_LINEAR_BC_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_BC_HEADER_sub_op_mask) << SDMA_PKT_COPY_LINEAR_BC_HEADER_sub_op_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_LINEAR_BC_COUNT_count_offset 1
+#define SDMA_PKT_COPY_LINEAR_BC_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_COPY_LINEAR_BC_COUNT_count_shift  0
+#define SDMA_PKT_COPY_LINEAR_BC_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_LINEAR_BC_COUNT_count_mask) << SDMA_PKT_COPY_LINEAR_BC_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_dst_sw_offset 2
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_dst_sw_shift  16
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_DST_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_BC_PARAMETER_dst_sw_mask) << SDMA_PKT_COPY_LINEAR_BC_PARAMETER_dst_sw_shift)
+
+/*define for dst_ha field*/
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_dst_ha_offset 2
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_dst_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_dst_ha_shift  19
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_DST_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_BC_PARAMETER_dst_ha_mask) << SDMA_PKT_COPY_LINEAR_BC_PARAMETER_dst_ha_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_src_sw_offset 2
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_src_sw_shift  24
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_SRC_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_BC_PARAMETER_src_sw_mask) << SDMA_PKT_COPY_LINEAR_BC_PARAMETER_src_sw_shift)
+
+/*define for src_ha field*/
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_src_ha_offset 2
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_src_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_src_ha_shift  27
+#define SDMA_PKT_COPY_LINEAR_BC_PARAMETER_SRC_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_BC_PARAMETER_src_ha_mask) << SDMA_PKT_COPY_LINEAR_BC_PARAMETER_src_ha_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_LO_src_addr_31_0_offset 3
+#define SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_HI_src_addr_63_32_offset 4
+#define SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_BC_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_LO_dst_addr_31_0_offset 5
+#define SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_HI_dst_addr_63_32_offset 6
+#define SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_BC_DST_ADDR_HI_dst_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_DIRTY_PAGE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_op_offset 0
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_op_shift  0
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_OP(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_HEADER_op_mask) << SDMA_PKT_COPY_DIRTY_PAGE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_HEADER_sub_op_mask) << SDMA_PKT_COPY_DIRTY_PAGE_HEADER_sub_op_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_HEADER_tmz_mask) << SDMA_PKT_COPY_DIRTY_PAGE_HEADER_tmz_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_cpv_offset 0
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_cpv_shift  19
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_CPV(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_HEADER_cpv_mask) << SDMA_PKT_COPY_DIRTY_PAGE_HEADER_cpv_shift)
+
+/*define for all field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_all_offset 0
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_all_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_all_shift  31
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_ALL(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_HEADER_all_mask) << SDMA_PKT_COPY_DIRTY_PAGE_HEADER_all_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_COUNT_count_offset 1
+#define SDMA_PKT_COPY_DIRTY_PAGE_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_COPY_DIRTY_PAGE_COUNT_count_shift  0
+#define SDMA_PKT_COPY_DIRTY_PAGE_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_COUNT_count_mask) << SDMA_PKT_COPY_DIRTY_PAGE_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst_mtype field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_mtype_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_mtype_mask   0x00000007
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_mtype_shift  3
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_DST_MTYPE(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_mtype_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_mtype_shift)
+
+/*define for dst_l2_policy field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_l2_policy_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_l2_policy_mask   0x00000003
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_l2_policy_shift  6
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_DST_L2_POLICY(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_l2_policy_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_l2_policy_shift)
+
+/*define for dst_llc field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_llc_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_llc_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_llc_shift  8
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_DST_LLC(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_llc_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_llc_shift)
+
+/*define for src_mtype field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_mtype_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_mtype_mask   0x00000007
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_mtype_shift  11
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_SRC_MTYPE(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_mtype_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_mtype_shift)
+
+/*define for src_l2_policy field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_l2_policy_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_l2_policy_mask   0x00000003
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_l2_policy_shift  14
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_SRC_L2_POLICY(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_l2_policy_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_l2_policy_shift)
+
+/*define for src_llc field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_llc_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_llc_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_llc_shift  16
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_SRC_LLC(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_llc_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_llc_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sw_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sw_shift  17
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_DST_SW(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sw_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sw_shift)
+
+/*define for dst_gcc field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gcc_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gcc_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gcc_shift  19
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_DST_GCC(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gcc_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gcc_shift)
+
+/*define for dst_sys field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sys_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sys_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sys_shift  20
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_DST_SYS(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sys_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sys_shift)
+
+/*define for dst_snoop field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_snoop_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_snoop_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_snoop_shift  22
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_DST_SNOOP(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_snoop_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_snoop_shift)
+
+/*define for dst_gpa field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gpa_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gpa_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gpa_shift  23
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_DST_GPA(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gpa_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gpa_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sw_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sw_shift  24
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_SRC_SW(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sw_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sw_shift)
+
+/*define for src_sys field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sys_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sys_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sys_shift  28
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_SRC_SYS(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sys_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sys_shift)
+
+/*define for src_snoop field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_snoop_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_snoop_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_snoop_shift  30
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_SRC_SNOOP(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_snoop_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_snoop_shift)
+
+/*define for src_gpa field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_gpa_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_gpa_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_gpa_shift  31
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_SRC_GPA(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_gpa_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_gpa_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_LO_src_addr_31_0_offset 3
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_HI_src_addr_63_32_offset 4
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_LO_dst_addr_31_0_offset 5
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_HI_dst_addr_63_32_offset 6
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_HI_dst_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_PHYSICAL_LINEAR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_op_offset 0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_op_shift  0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_OP(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_op_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_sub_op_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_sub_op_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_tmz_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_tmz_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_cpv_offset 0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_cpv_shift  19
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_CPV(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_cpv_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_cpv_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_count_offset 1
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_count_shift  0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_count_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_count_shift)
+
+/*define for addr_pair_num field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_addr_pair_num_offset 1
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_addr_pair_num_mask   0x000000FF
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_addr_pair_num_shift  24
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_ADDR_PAIR_NUM(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_addr_pair_num_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_addr_pair_num_shift)
+
+/*define for PARAMETER word*/
+/*define for dst_mtype field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_mtype_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_mtype_mask   0x00000007
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_mtype_shift  3
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_MTYPE(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_mtype_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_mtype_shift)
+
+/*define for dst_l2_policy field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_l2_policy_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_l2_policy_mask   0x00000003
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_l2_policy_shift  6
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_L2_POLICY(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_l2_policy_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_l2_policy_shift)
+
+/*define for dst_llc field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_llc_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_llc_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_llc_shift  8
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_LLC(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_llc_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_llc_shift)
+
+/*define for src_mtype field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_mtype_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_mtype_mask   0x00000007
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_mtype_shift  11
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_SRC_MTYPE(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_mtype_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_mtype_shift)
+
+/*define for src_l2_policy field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_l2_policy_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_l2_policy_mask   0x00000003
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_l2_policy_shift  14
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_SRC_L2_POLICY(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_l2_policy_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_l2_policy_shift)
+
+/*define for src_llc field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_llc_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_llc_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_llc_shift  16
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_SRC_LLC(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_llc_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_llc_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sw_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sw_shift  17
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_SW(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sw_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sw_shift)
+
+/*define for dst_gcc field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gcc_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gcc_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gcc_shift  19
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_GCC(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gcc_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gcc_shift)
+
+/*define for dst_sys field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sys_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sys_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sys_shift  20
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_SYS(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sys_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sys_shift)
+
+/*define for dst_log field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_log_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_log_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_log_shift  21
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_LOG(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_log_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_log_shift)
+
+/*define for dst_snoop field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_snoop_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_snoop_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_snoop_shift  22
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_SNOOP(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_snoop_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_snoop_shift)
+
+/*define for dst_gpa field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gpa_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gpa_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gpa_shift  23
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_GPA(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gpa_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gpa_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sw_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sw_shift  24
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_SRC_SW(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sw_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sw_shift)
+
+/*define for src_gcc field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gcc_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gcc_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gcc_shift  27
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_SRC_GCC(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gcc_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gcc_shift)
+
+/*define for src_sys field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sys_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sys_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sys_shift  28
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_SRC_SYS(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sys_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sys_shift)
+
+/*define for src_snoop field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_snoop_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_snoop_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_snoop_shift  30
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_SRC_SNOOP(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_snoop_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_snoop_shift)
+
+/*define for src_gpa field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gpa_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gpa_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gpa_shift  31
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_SRC_GPA(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gpa_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gpa_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_LO_src_addr_31_0_offset 3
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_HI_src_addr_63_32_offset 4
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_LO_dst_addr_31_0_offset 5
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_HI_dst_addr_63_32_offset 6
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_BROADCAST_LINEAR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_OP(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_shift)
+
+/*define for encrypt field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_encrypt_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_encrypt_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_encrypt_shift  16
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_ENCRYPT(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_encrypt_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_encrypt_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_tmz_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_tmz_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_cpv_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_cpv_shift  19
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_CPV(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_cpv_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_cpv_shift)
+
+/*define for broadcast field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_shift  27
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_BROADCAST(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_offset 1
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_mask   0x3FFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst2_sw field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_mask   0x00000003
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_shift  8
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST2_SW(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_shift)
+
+/*define for dst2_cache_policy field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_cache_policy_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_cache_policy_mask   0x00000007
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_cache_policy_shift  10
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST2_CACHE_POLICY(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_cache_policy_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_cache_policy_shift)
+
+/*define for dst1_sw field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_mask   0x00000003
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_shift  16
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST1_SW(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_shift)
+
+/*define for dst1_cache_policy field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_cache_policy_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_cache_policy_mask   0x00000007
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_cache_policy_shift  18
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST1_CACHE_POLICY(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_cache_policy_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_cache_policy_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_shift  24
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_SRC_SW(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_shift)
+
+/*define for src_cache_policy field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_cache_policy_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_cache_policy_mask   0x00000007
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_cache_policy_shift  26
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_SRC_CACHE_POLICY(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_cache_policy_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_cache_policy_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_offset 3
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_offset 4
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST1_ADDR_LO word*/
+/*define for dst1_addr_31_0 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_offset 5
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_DST1_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_shift)
+
+/*define for DST1_ADDR_HI word*/
+/*define for dst1_addr_63_32 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_offset 6
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_DST1_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_shift)
+
+/*define for DST2_ADDR_LO word*/
+/*define for dst2_addr_31_0 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_offset 7
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_DST2_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_shift)
+
+/*define for DST2_ADDR_HI word*/
+/*define for dst2_addr_63_32 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_offset 8
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_DST2_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_LINEAR_SUBWIN packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_tmz_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_tmz_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_cpv_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_cpv_shift  19
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_CPV(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_cpv_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_cpv_shift)
+
+/*define for elementsize field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_mask   0x00000007
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_shift  29
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_ELEMENTSIZE(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for src_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_offset 3
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_SRC_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_shift)
+
+/*define for src_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_offset 3
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_SRC_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_shift)
+
+/*define for DW_4 word*/
+/*define for src_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_offset 4
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_mask   0x00001FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_SRC_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_shift)
+
+/*define for src_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_offset 4
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_mask   0x0007FFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_shift  13
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_SRC_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_shift)
+
+/*define for DW_5 word*/
+/*define for src_slice_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_offset 5
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_SRC_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_offset 6
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_offset 7
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_8 word*/
+/*define for dst_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_offset 8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_DST_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_shift)
+
+/*define for dst_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_offset 8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_DST_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_shift)
+
+/*define for DW_9 word*/
+/*define for dst_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_offset 9
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_mask   0x00001FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_DST_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_shift)
+
+/*define for dst_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_offset 9
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_mask   0x0007FFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_shift  13
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_DST_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_shift)
+
+/*define for DW_10 word*/
+/*define for dst_slice_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_offset 10
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_DST_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_shift)
+
+/*define for DW_11 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_offset 11
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_RECT_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_offset 11
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_RECT_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_shift)
+
+/*define for DW_12 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_mask   0x00001FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_RECT_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_DST_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_shift)
+
+/*define for dst_cache_policy field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_cache_policy_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_cache_policy_mask   0x00000007
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_cache_policy_shift  18
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_DST_CACHE_POLICY(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_cache_policy_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_cache_policy_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_shift  24
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_SRC_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_shift)
+
+/*define for src_cache_policy field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_cache_policy_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_cache_policy_mask   0x00000007
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_cache_policy_shift  26
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_SRC_CACHE_POLICY(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_cache_policy_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_cache_policy_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_op_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_op_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_sub_op_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_sub_op_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_tmz_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_tmz_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_cpv_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_cpv_shift  19
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_CPV(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_cpv_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_HEADER_cpv_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for src_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_3_src_x_offset 3
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_3_src_x_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_3_src_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_3_SRC_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_3_src_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_3_src_x_shift)
+
+/*define for DW_4 word*/
+/*define for src_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_4_src_y_offset 4
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_4_src_y_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_4_src_y_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_4_SRC_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_4_src_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_4_src_y_shift)
+
+/*define for DW_5 word*/
+/*define for src_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_5_src_z_offset 5
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_5_src_z_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_5_src_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_5_SRC_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_5_src_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_5_src_z_shift)
+
+/*define for DW_6 word*/
+/*define for src_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_6_src_pitch_offset 6
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_6_src_pitch_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_6_src_pitch_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_6_SRC_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_6_src_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_6_src_pitch_shift)
+
+/*define for DW_7 word*/
+/*define for src_slice_pitch_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_7_src_slice_pitch_31_0_offset 7
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_7_src_slice_pitch_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_7_src_slice_pitch_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_7_SRC_SLICE_PITCH_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_7_src_slice_pitch_31_0_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_7_src_slice_pitch_31_0_shift)
+
+/*define for DW_8 word*/
+/*define for src_slice_pitch_47_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_8_src_slice_pitch_47_32_offset 8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_8_src_slice_pitch_47_32_mask   0x0000FFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_8_src_slice_pitch_47_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_8_SRC_SLICE_PITCH_47_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_8_src_slice_pitch_47_32_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_8_src_slice_pitch_47_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DST_ADDR_LO_dst_addr_31_0_offset 9
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DST_ADDR_HI_dst_addr_63_32_offset 10
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_11 word*/
+/*define for dst_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_11_dst_x_offset 11
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_11_dst_x_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_11_dst_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_11_DST_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_11_dst_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_11_dst_x_shift)
+
+/*define for DW_12 word*/
+/*define for dst_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_12_dst_y_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_12_dst_y_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_12_dst_y_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_12_DST_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_12_dst_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_12_dst_y_shift)
+
+/*define for DW_13 word*/
+/*define for dst_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_13_dst_z_offset 13
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_13_dst_z_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_13_dst_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_13_DST_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_13_dst_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_13_dst_z_shift)
+
+/*define for DW_14 word*/
+/*define for dst_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_14_dst_pitch_offset 14
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_14_dst_pitch_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_14_dst_pitch_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_14_DST_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_14_dst_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_14_dst_pitch_shift)
+
+/*define for DW_15 word*/
+/*define for dst_slice_pitch_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_15_dst_slice_pitch_31_0_offset 15
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_15_dst_slice_pitch_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_15_dst_slice_pitch_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_15_DST_SLICE_PITCH_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_15_dst_slice_pitch_31_0_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_15_dst_slice_pitch_31_0_shift)
+
+/*define for DW_16 word*/
+/*define for dst_slice_pitch_47_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_dst_slice_pitch_47_32_offset 16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_dst_slice_pitch_47_32_mask   0x0000FFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_dst_slice_pitch_47_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_DST_SLICE_PITCH_47_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_dst_slice_pitch_47_32_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_dst_slice_pitch_47_32_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_dst_sw_offset 16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_dst_sw_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_DST_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_dst_sw_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_dst_sw_shift)
+
+/*define for dst_policy field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_dst_policy_offset 16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_dst_policy_mask   0x00000007
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_dst_policy_shift  18
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_DST_POLICY(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_dst_policy_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_dst_policy_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_src_sw_offset 16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_src_sw_shift  24
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_SRC_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_src_sw_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_src_sw_shift)
+
+/*define for src_policy field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_src_policy_offset 16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_src_policy_mask   0x00000007
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_src_policy_shift  26
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_SRC_POLICY(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_src_policy_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_16_src_policy_shift)
+
+/*define for DW_17 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_17_rect_x_offset 17
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_17_rect_x_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_17_rect_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_17_RECT_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_17_rect_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_17_rect_x_shift)
+
+/*define for DW_18 word*/
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_18_rect_y_offset 18
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_18_rect_y_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_18_rect_y_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_18_RECT_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_18_rect_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_18_rect_y_shift)
+
+/*define for DW_19 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_19_rect_z_offset 19
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_19_rect_z_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_19_rect_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_19_RECT_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_19_rect_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_LARGE_DW_19_rect_z_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_LINEAR_SUBWIN_BC packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_op_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_op_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_sub_op_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_sub_op_shift)
+
+/*define for elementsize field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_elementsize_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_elementsize_mask   0x00000007
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_elementsize_shift  29
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_ELEMENTSIZE(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_elementsize_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_HEADER_elementsize_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for src_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_src_x_offset 3
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_src_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_src_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_SRC_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_src_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_src_x_shift)
+
+/*define for src_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_src_y_offset 3
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_src_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_src_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_SRC_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_src_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_3_src_y_shift)
+
+/*define for DW_4 word*/
+/*define for src_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_src_z_offset 4
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_src_z_mask   0x000007FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_src_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_SRC_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_src_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_src_z_shift)
+
+/*define for src_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_src_pitch_offset 4
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_src_pitch_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_src_pitch_shift  13
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_SRC_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_src_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_4_src_pitch_shift)
+
+/*define for DW_5 word*/
+/*define for src_slice_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_5_src_slice_pitch_offset 5
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_5_src_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_5_src_slice_pitch_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_5_SRC_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_5_src_slice_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_5_src_slice_pitch_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_LO_dst_addr_31_0_offset 6
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_HI_dst_addr_63_32_offset 7
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_8 word*/
+/*define for dst_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_dst_x_offset 8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_dst_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_dst_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_DST_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_dst_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_dst_x_shift)
+
+/*define for dst_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_dst_y_offset 8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_dst_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_dst_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_DST_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_dst_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_8_dst_y_shift)
+
+/*define for DW_9 word*/
+/*define for dst_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_dst_z_offset 9
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_dst_z_mask   0x000007FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_dst_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_DST_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_dst_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_dst_z_shift)
+
+/*define for dst_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_dst_pitch_offset 9
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_dst_pitch_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_dst_pitch_shift  13
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_DST_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_dst_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_9_dst_pitch_shift)
+
+/*define for DW_10 word*/
+/*define for dst_slice_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_10_dst_slice_pitch_offset 10
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_10_dst_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_10_dst_slice_pitch_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_10_DST_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_10_dst_slice_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_10_dst_slice_pitch_shift)
+
+/*define for DW_11 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_rect_x_offset 11
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_rect_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_RECT_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_rect_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_rect_y_offset 11
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_rect_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_RECT_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_rect_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_11_rect_y_shift)
+
+/*define for DW_12 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_rect_z_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_rect_z_mask   0x000007FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_rect_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_RECT_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_rect_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_rect_z_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_dst_sw_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_dst_sw_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_DST_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_dst_sw_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_dst_sw_shift)
+
+/*define for dst_ha field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_dst_ha_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_dst_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_dst_ha_shift  19
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_DST_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_dst_ha_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_dst_ha_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_src_sw_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_src_sw_shift  24
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_SRC_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_src_sw_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_src_sw_shift)
+
+/*define for src_ha field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_src_ha_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_src_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_src_ha_shift  27
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_SRC_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_src_ha_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_BC_DW_12_src_ha_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_TILED packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_TILED_HEADER_op_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_HEADER_op_shift  0
+#define SDMA_PKT_COPY_TILED_HEADER_OP(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_op_mask) << SDMA_PKT_COPY_TILED_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_TILED_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_TILED_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_sub_op_mask) << SDMA_PKT_COPY_TILED_HEADER_sub_op_shift)
+
+/*define for encrypt field*/
+#define SDMA_PKT_COPY_TILED_HEADER_encrypt_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_encrypt_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_HEADER_encrypt_shift  16
+#define SDMA_PKT_COPY_TILED_HEADER_ENCRYPT(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_encrypt_mask) << SDMA_PKT_COPY_TILED_HEADER_encrypt_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_TILED_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_TILED_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_tmz_mask) << SDMA_PKT_COPY_TILED_HEADER_tmz_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_COPY_TILED_HEADER_cpv_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_HEADER_cpv_shift  19
+#define SDMA_PKT_COPY_TILED_HEADER_CPV(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_cpv_mask) << SDMA_PKT_COPY_TILED_HEADER_cpv_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_TILED_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_TILED_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_detile_mask) << SDMA_PKT_COPY_TILED_HEADER_detile_shift)
+
+/*define for TILED_ADDR_LO word*/
+/*define for tiled_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_offset 1
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_TILED_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_mask) << SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_shift)
+
+/*define for TILED_ADDR_HI word*/
+/*define for tiled_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_offset 2
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_TILED_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_mask) << SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for width field*/
+#define SDMA_PKT_COPY_TILED_DW_3_width_offset 3
+#define SDMA_PKT_COPY_TILED_DW_3_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_3_width_shift  0
+#define SDMA_PKT_COPY_TILED_DW_3_WIDTH(x) (((x) & SDMA_PKT_COPY_TILED_DW_3_width_mask) << SDMA_PKT_COPY_TILED_DW_3_width_shift)
+
+/*define for DW_4 word*/
+/*define for height field*/
+#define SDMA_PKT_COPY_TILED_DW_4_height_offset 4
+#define SDMA_PKT_COPY_TILED_DW_4_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_4_height_shift  0
+#define SDMA_PKT_COPY_TILED_DW_4_HEIGHT(x) (((x) & SDMA_PKT_COPY_TILED_DW_4_height_mask) << SDMA_PKT_COPY_TILED_DW_4_height_shift)
+
+/*define for depth field*/
+#define SDMA_PKT_COPY_TILED_DW_4_depth_offset 4
+#define SDMA_PKT_COPY_TILED_DW_4_depth_mask   0x00001FFF
+#define SDMA_PKT_COPY_TILED_DW_4_depth_shift  16
+#define SDMA_PKT_COPY_TILED_DW_4_DEPTH(x) (((x) & SDMA_PKT_COPY_TILED_DW_4_depth_mask) << SDMA_PKT_COPY_TILED_DW_4_depth_shift)
+
+/*define for DW_5 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_TILED_DW_5_element_size_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_DW_5_element_size_shift  0
+#define SDMA_PKT_COPY_TILED_DW_5_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_element_size_mask) << SDMA_PKT_COPY_TILED_DW_5_element_size_shift)
+
+/*define for swizzle_mode field*/
+#define SDMA_PKT_COPY_TILED_DW_5_swizzle_mode_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_swizzle_mode_mask   0x0000001F
+#define SDMA_PKT_COPY_TILED_DW_5_swizzle_mode_shift  3
+#define SDMA_PKT_COPY_TILED_DW_5_SWIZZLE_MODE(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_swizzle_mode_mask) << SDMA_PKT_COPY_TILED_DW_5_swizzle_mode_shift)
+
+/*define for dimension field*/
+#define SDMA_PKT_COPY_TILED_DW_5_dimension_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_dimension_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_5_dimension_shift  9
+#define SDMA_PKT_COPY_TILED_DW_5_DIMENSION(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_dimension_mask) << SDMA_PKT_COPY_TILED_DW_5_dimension_shift)
+
+/*define for mip_max field*/
+#define SDMA_PKT_COPY_TILED_DW_5_mip_max_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_mip_max_mask   0x0000000F
+#define SDMA_PKT_COPY_TILED_DW_5_mip_max_shift  16
+#define SDMA_PKT_COPY_TILED_DW_5_MIP_MAX(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_mip_max_mask) << SDMA_PKT_COPY_TILED_DW_5_mip_max_shift)
+
+/*define for DW_6 word*/
+/*define for x field*/
+#define SDMA_PKT_COPY_TILED_DW_6_x_offset 6
+#define SDMA_PKT_COPY_TILED_DW_6_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_6_x_shift  0
+#define SDMA_PKT_COPY_TILED_DW_6_X(x) (((x) & SDMA_PKT_COPY_TILED_DW_6_x_mask) << SDMA_PKT_COPY_TILED_DW_6_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_COPY_TILED_DW_6_y_offset 6
+#define SDMA_PKT_COPY_TILED_DW_6_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_6_y_shift  16
+#define SDMA_PKT_COPY_TILED_DW_6_Y(x) (((x) & SDMA_PKT_COPY_TILED_DW_6_y_mask) << SDMA_PKT_COPY_TILED_DW_6_y_shift)
+
+/*define for DW_7 word*/
+/*define for z field*/
+#define SDMA_PKT_COPY_TILED_DW_7_z_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_z_mask   0x00001FFF
+#define SDMA_PKT_COPY_TILED_DW_7_z_shift  0
+#define SDMA_PKT_COPY_TILED_DW_7_Z(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_z_mask) << SDMA_PKT_COPY_TILED_DW_7_z_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_TILED_DW_7_linear_sw_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_7_linear_sw_shift  16
+#define SDMA_PKT_COPY_TILED_DW_7_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_linear_sw_mask) << SDMA_PKT_COPY_TILED_DW_7_linear_sw_shift)
+
+/*define for linear_cache_policy field*/
+#define SDMA_PKT_COPY_TILED_DW_7_linear_cache_policy_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_linear_cache_policy_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_DW_7_linear_cache_policy_shift  18
+#define SDMA_PKT_COPY_TILED_DW_7_LINEAR_CACHE_POLICY(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_linear_cache_policy_mask) << SDMA_PKT_COPY_TILED_DW_7_linear_cache_policy_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_TILED_DW_7_tile_sw_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_7_tile_sw_shift  24
+#define SDMA_PKT_COPY_TILED_DW_7_TILE_SW(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_tile_sw_mask) << SDMA_PKT_COPY_TILED_DW_7_tile_sw_shift)
+
+/*define for tile_cache_policy field*/
+#define SDMA_PKT_COPY_TILED_DW_7_tile_cache_policy_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_tile_cache_policy_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_DW_7_tile_cache_policy_shift  26
+#define SDMA_PKT_COPY_TILED_DW_7_TILE_CACHE_POLICY(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_tile_cache_policy_mask) << SDMA_PKT_COPY_TILED_DW_7_tile_cache_policy_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_offset 8
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_offset 9
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for LINEAR_PITCH word*/
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_offset 10
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_mask   0x0007FFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_mask) << SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_shift)
+
+/*define for LINEAR_SLICE_PITCH word*/
+/*define for linear_slice_pitch field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_SLICE_PITCH_linear_slice_pitch_offset 11
+#define SDMA_PKT_COPY_TILED_LINEAR_SLICE_PITCH_linear_slice_pitch_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_SLICE_PITCH_linear_slice_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_SLICE_PITCH_LINEAR_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_SLICE_PITCH_linear_slice_pitch_mask) << SDMA_PKT_COPY_TILED_LINEAR_SLICE_PITCH_linear_slice_pitch_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_TILED_COUNT_count_offset 12
+#define SDMA_PKT_COPY_TILED_COUNT_count_mask   0x3FFFFFFF
+#define SDMA_PKT_COPY_TILED_COUNT_count_shift  0
+#define SDMA_PKT_COPY_TILED_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_TILED_COUNT_count_mask) << SDMA_PKT_COPY_TILED_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_TILED_BC packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_TILED_BC_HEADER_op_offset 0
+#define SDMA_PKT_COPY_TILED_BC_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_BC_HEADER_op_shift  0
+#define SDMA_PKT_COPY_TILED_BC_HEADER_OP(x) (((x) & SDMA_PKT_COPY_TILED_BC_HEADER_op_mask) << SDMA_PKT_COPY_TILED_BC_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_TILED_BC_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_TILED_BC_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_BC_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_TILED_BC_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_TILED_BC_HEADER_sub_op_mask) << SDMA_PKT_COPY_TILED_BC_HEADER_sub_op_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_TILED_BC_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_TILED_BC_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_BC_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_TILED_BC_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_TILED_BC_HEADER_detile_mask) << SDMA_PKT_COPY_TILED_BC_HEADER_detile_shift)
+
+/*define for TILED_ADDR_LO word*/
+/*define for tiled_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_BC_TILED_ADDR_LO_tiled_addr_31_0_offset 1
+#define SDMA_PKT_COPY_TILED_BC_TILED_ADDR_LO_tiled_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_BC_TILED_ADDR_LO_tiled_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_BC_TILED_ADDR_LO_TILED_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_BC_TILED_ADDR_LO_tiled_addr_31_0_mask) << SDMA_PKT_COPY_TILED_BC_TILED_ADDR_LO_tiled_addr_31_0_shift)
+
+/*define for TILED_ADDR_HI word*/
+/*define for tiled_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_BC_TILED_ADDR_HI_tiled_addr_63_32_offset 2
+#define SDMA_PKT_COPY_TILED_BC_TILED_ADDR_HI_tiled_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_BC_TILED_ADDR_HI_tiled_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_BC_TILED_ADDR_HI_TILED_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_BC_TILED_ADDR_HI_tiled_addr_63_32_mask) << SDMA_PKT_COPY_TILED_BC_TILED_ADDR_HI_tiled_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for width field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_3_width_offset 3
+#define SDMA_PKT_COPY_TILED_BC_DW_3_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_BC_DW_3_width_shift  0
+#define SDMA_PKT_COPY_TILED_BC_DW_3_WIDTH(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_3_width_mask) << SDMA_PKT_COPY_TILED_BC_DW_3_width_shift)
+
+/*define for DW_4 word*/
+/*define for height field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_4_height_offset 4
+#define SDMA_PKT_COPY_TILED_BC_DW_4_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_BC_DW_4_height_shift  0
+#define SDMA_PKT_COPY_TILED_BC_DW_4_HEIGHT(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_4_height_mask) << SDMA_PKT_COPY_TILED_BC_DW_4_height_shift)
+
+/*define for depth field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_4_depth_offset 4
+#define SDMA_PKT_COPY_TILED_BC_DW_4_depth_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_BC_DW_4_depth_shift  16
+#define SDMA_PKT_COPY_TILED_BC_DW_4_DEPTH(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_4_depth_mask) << SDMA_PKT_COPY_TILED_BC_DW_4_depth_shift)
+
+/*define for DW_5 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_5_element_size_offset 5
+#define SDMA_PKT_COPY_TILED_BC_DW_5_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_BC_DW_5_element_size_shift  0
+#define SDMA_PKT_COPY_TILED_BC_DW_5_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_5_element_size_mask) << SDMA_PKT_COPY_TILED_BC_DW_5_element_size_shift)
+
+/*define for array_mode field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_5_array_mode_offset 5
+#define SDMA_PKT_COPY_TILED_BC_DW_5_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_TILED_BC_DW_5_array_mode_shift  3
+#define SDMA_PKT_COPY_TILED_BC_DW_5_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_5_array_mode_mask) << SDMA_PKT_COPY_TILED_BC_DW_5_array_mode_shift)
+
+/*define for mit_mode field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_5_mit_mode_offset 5
+#define SDMA_PKT_COPY_TILED_BC_DW_5_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_BC_DW_5_mit_mode_shift  8
+#define SDMA_PKT_COPY_TILED_BC_DW_5_MIT_MODE(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_5_mit_mode_mask) << SDMA_PKT_COPY_TILED_BC_DW_5_mit_mode_shift)
+
+/*define for tilesplit_size field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_5_tilesplit_size_offset 5
+#define SDMA_PKT_COPY_TILED_BC_DW_5_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_BC_DW_5_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_TILED_BC_DW_5_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_5_tilesplit_size_mask) << SDMA_PKT_COPY_TILED_BC_DW_5_tilesplit_size_shift)
+
+/*define for bank_w field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_5_bank_w_offset 5
+#define SDMA_PKT_COPY_TILED_BC_DW_5_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_BC_DW_5_bank_w_shift  15
+#define SDMA_PKT_COPY_TILED_BC_DW_5_BANK_W(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_5_bank_w_mask) << SDMA_PKT_COPY_TILED_BC_DW_5_bank_w_shift)
+
+/*define for bank_h field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_5_bank_h_offset 5
+#define SDMA_PKT_COPY_TILED_BC_DW_5_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_BC_DW_5_bank_h_shift  18
+#define SDMA_PKT_COPY_TILED_BC_DW_5_BANK_H(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_5_bank_h_mask) << SDMA_PKT_COPY_TILED_BC_DW_5_bank_h_shift)
+
+/*define for num_bank field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_5_num_bank_offset 5
+#define SDMA_PKT_COPY_TILED_BC_DW_5_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_BC_DW_5_num_bank_shift  21
+#define SDMA_PKT_COPY_TILED_BC_DW_5_NUM_BANK(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_5_num_bank_mask) << SDMA_PKT_COPY_TILED_BC_DW_5_num_bank_shift)
+
+/*define for mat_aspt field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_5_mat_aspt_offset 5
+#define SDMA_PKT_COPY_TILED_BC_DW_5_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_BC_DW_5_mat_aspt_shift  24
+#define SDMA_PKT_COPY_TILED_BC_DW_5_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_5_mat_aspt_mask) << SDMA_PKT_COPY_TILED_BC_DW_5_mat_aspt_shift)
+
+/*define for pipe_config field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_5_pipe_config_offset 5
+#define SDMA_PKT_COPY_TILED_BC_DW_5_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_TILED_BC_DW_5_pipe_config_shift  26
+#define SDMA_PKT_COPY_TILED_BC_DW_5_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_5_pipe_config_mask) << SDMA_PKT_COPY_TILED_BC_DW_5_pipe_config_shift)
+
+/*define for DW_6 word*/
+/*define for x field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_6_x_offset 6
+#define SDMA_PKT_COPY_TILED_BC_DW_6_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_BC_DW_6_x_shift  0
+#define SDMA_PKT_COPY_TILED_BC_DW_6_X(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_6_x_mask) << SDMA_PKT_COPY_TILED_BC_DW_6_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_6_y_offset 6
+#define SDMA_PKT_COPY_TILED_BC_DW_6_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_BC_DW_6_y_shift  16
+#define SDMA_PKT_COPY_TILED_BC_DW_6_Y(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_6_y_mask) << SDMA_PKT_COPY_TILED_BC_DW_6_y_shift)
+
+/*define for DW_7 word*/
+/*define for z field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_7_z_offset 7
+#define SDMA_PKT_COPY_TILED_BC_DW_7_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_BC_DW_7_z_shift  0
+#define SDMA_PKT_COPY_TILED_BC_DW_7_Z(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_7_z_mask) << SDMA_PKT_COPY_TILED_BC_DW_7_z_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_7_linear_sw_offset 7
+#define SDMA_PKT_COPY_TILED_BC_DW_7_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_BC_DW_7_linear_sw_shift  16
+#define SDMA_PKT_COPY_TILED_BC_DW_7_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_7_linear_sw_mask) << SDMA_PKT_COPY_TILED_BC_DW_7_linear_sw_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_TILED_BC_DW_7_tile_sw_offset 7
+#define SDMA_PKT_COPY_TILED_BC_DW_7_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_BC_DW_7_tile_sw_shift  24
+#define SDMA_PKT_COPY_TILED_BC_DW_7_TILE_SW(x) (((x) & SDMA_PKT_COPY_TILED_BC_DW_7_tile_sw_mask) << SDMA_PKT_COPY_TILED_BC_DW_7_tile_sw_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_LO_linear_addr_31_0_offset 8
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_HI_linear_addr_63_32_offset 9
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_TILED_BC_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for LINEAR_PITCH word*/
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_PITCH_linear_pitch_offset 10
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_PITCH_linear_pitch_mask   0x0007FFFF
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_PITCH_linear_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_PITCH_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_BC_LINEAR_PITCH_linear_pitch_mask) << SDMA_PKT_COPY_TILED_BC_LINEAR_PITCH_linear_pitch_shift)
+
+/*define for LINEAR_SLICE_PITCH word*/
+/*define for linear_slice_pitch field*/
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_SLICE_PITCH_linear_slice_pitch_offset 11
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_SLICE_PITCH_linear_slice_pitch_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_SLICE_PITCH_linear_slice_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_BC_LINEAR_SLICE_PITCH_LINEAR_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_BC_LINEAR_SLICE_PITCH_linear_slice_pitch_mask) << SDMA_PKT_COPY_TILED_BC_LINEAR_SLICE_PITCH_linear_slice_pitch_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_TILED_BC_COUNT_count_offset 12
+#define SDMA_PKT_COPY_TILED_BC_COUNT_count_mask   0x000FFFFF
+#define SDMA_PKT_COPY_TILED_BC_COUNT_count_shift  2
+#define SDMA_PKT_COPY_TILED_BC_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_TILED_BC_COUNT_count_mask) << SDMA_PKT_COPY_TILED_BC_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_L2T_BROADCAST packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_OP(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_shift)
+
+/*define for encrypt field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_encrypt_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_encrypt_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_encrypt_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_ENCRYPT(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_encrypt_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_encrypt_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_tmz_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_tmz_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_cpv_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_cpv_shift  19
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_CPV(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_cpv_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_cpv_shift)
+
+/*define for videocopy field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_shift  26
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_VIDEOCOPY(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_shift)
+
+/*define for broadcast field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_shift  27
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_BROADCAST(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_shift)
+
+/*define for TILED_ADDR_LO_0 word*/
+/*define for tiled_addr0_31_0 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_offset 1
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_TILED_ADDR0_31_0(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_shift)
+
+/*define for TILED_ADDR_HI_0 word*/
+/*define for tiled_addr0_63_32 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_offset 2
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_TILED_ADDR0_63_32(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_shift)
+
+/*define for TILED_ADDR_LO_1 word*/
+/*define for tiled_addr1_31_0 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_offset 3
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_TILED_ADDR1_31_0(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_shift)
+
+/*define for TILED_ADDR_HI_1 word*/
+/*define for tiled_addr1_63_32 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_offset 4
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_TILED_ADDR1_63_32(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_shift)
+
+/*define for DW_5 word*/
+/*define for width field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_width_offset 5
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_width_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_WIDTH(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_5_width_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_5_width_shift)
+
+/*define for DW_6 word*/
+/*define for height field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_height_offset 6
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_height_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_HEIGHT(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_6_height_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_6_height_shift)
+
+/*define for depth field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_depth_offset 6
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_depth_mask   0x00001FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_depth_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_DEPTH(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_6_depth_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_6_depth_shift)
+
+/*define for DW_7 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_shift)
+
+/*define for swizzle_mode field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_swizzle_mode_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_swizzle_mode_mask   0x0000001F
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_swizzle_mode_shift  3
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_SWIZZLE_MODE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_swizzle_mode_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_swizzle_mode_shift)
+
+/*define for dimension field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_dimension_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_dimension_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_dimension_shift  9
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_DIMENSION(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_dimension_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_dimension_shift)
+
+/*define for mip_max field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mip_max_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mip_max_mask   0x0000000F
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mip_max_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_MIP_MAX(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mip_max_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mip_max_shift)
+
+/*define for DW_8 word*/
+/*define for x field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_offset 8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_X(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_offset 8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_Y(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_shift)
+
+/*define for DW_9 word*/
+/*define for z field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_offset 9
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_mask   0x00001FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_Z(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_shift)
+
+/*define for DW_10 word*/
+/*define for dst2_sw field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_shift  8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_DST2_SW(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_shift)
+
+/*define for dst2_cache_policy field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_cache_policy_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_cache_policy_mask   0x00000007
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_cache_policy_shift  10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_DST2_CACHE_POLICY(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_cache_policy_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_cache_policy_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_shift)
+
+/*define for linear_cache_policy field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_cache_policy_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_cache_policy_mask   0x00000007
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_cache_policy_shift  18
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_LINEAR_CACHE_POLICY(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_cache_policy_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_cache_policy_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_shift  24
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_TILE_SW(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_shift)
+
+/*define for tile_cache_policy field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_cache_policy_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_cache_policy_mask   0x00000007
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_cache_policy_shift  26
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_TILE_CACHE_POLICY(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_cache_policy_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_cache_policy_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_offset 11
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_offset 12
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for LINEAR_PITCH word*/
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_offset 13
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_mask   0x0007FFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_shift)
+
+/*define for LINEAR_SLICE_PITCH word*/
+/*define for linear_slice_pitch field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_SLICE_PITCH_linear_slice_pitch_offset 14
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_SLICE_PITCH_linear_slice_pitch_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_SLICE_PITCH_linear_slice_pitch_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_SLICE_PITCH_LINEAR_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_SLICE_PITCH_linear_slice_pitch_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_SLICE_PITCH_linear_slice_pitch_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_offset 15
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_mask   0x3FFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_mask) << SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_T2T packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_T2T_HEADER_op_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_T2T_HEADER_op_shift  0
+#define SDMA_PKT_COPY_T2T_HEADER_OP(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_op_mask) << SDMA_PKT_COPY_T2T_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_T2T_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_T2T_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_T2T_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_sub_op_mask) << SDMA_PKT_COPY_T2T_HEADER_sub_op_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_T2T_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_T2T_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_T2T_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_tmz_mask) << SDMA_PKT_COPY_T2T_HEADER_tmz_shift)
+
+/*define for dcc field*/
+#define SDMA_PKT_COPY_T2T_HEADER_dcc_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_dcc_mask   0x00000001
+#define SDMA_PKT_COPY_T2T_HEADER_dcc_shift  19
+#define SDMA_PKT_COPY_T2T_HEADER_DCC(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_dcc_mask) << SDMA_PKT_COPY_T2T_HEADER_dcc_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_COPY_T2T_HEADER_cpv_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_COPY_T2T_HEADER_cpv_shift  28
+#define SDMA_PKT_COPY_T2T_HEADER_CPV(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_cpv_mask) << SDMA_PKT_COPY_T2T_HEADER_cpv_shift)
+
+/*define for dcc_dir field*/
+#define SDMA_PKT_COPY_T2T_HEADER_dcc_dir_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_dcc_dir_mask   0x00000001
+#define SDMA_PKT_COPY_T2T_HEADER_dcc_dir_shift  31
+#define SDMA_PKT_COPY_T2T_HEADER_DCC_DIR(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_dcc_dir_mask) << SDMA_PKT_COPY_T2T_HEADER_dcc_dir_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for src_x field*/
+#define SDMA_PKT_COPY_T2T_DW_3_src_x_offset 3
+#define SDMA_PKT_COPY_T2T_DW_3_src_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_3_src_x_shift  0
+#define SDMA_PKT_COPY_T2T_DW_3_SRC_X(x) (((x) & SDMA_PKT_COPY_T2T_DW_3_src_x_mask) << SDMA_PKT_COPY_T2T_DW_3_src_x_shift)
+
+/*define for src_y field*/
+#define SDMA_PKT_COPY_T2T_DW_3_src_y_offset 3
+#define SDMA_PKT_COPY_T2T_DW_3_src_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_3_src_y_shift  16
+#define SDMA_PKT_COPY_T2T_DW_3_SRC_Y(x) (((x) & SDMA_PKT_COPY_T2T_DW_3_src_y_mask) << SDMA_PKT_COPY_T2T_DW_3_src_y_shift)
+
+/*define for DW_4 word*/
+/*define for src_z field*/
+#define SDMA_PKT_COPY_T2T_DW_4_src_z_offset 4
+#define SDMA_PKT_COPY_T2T_DW_4_src_z_mask   0x00001FFF
+#define SDMA_PKT_COPY_T2T_DW_4_src_z_shift  0
+#define SDMA_PKT_COPY_T2T_DW_4_SRC_Z(x) (((x) & SDMA_PKT_COPY_T2T_DW_4_src_z_mask) << SDMA_PKT_COPY_T2T_DW_4_src_z_shift)
+
+/*define for src_width field*/
+#define SDMA_PKT_COPY_T2T_DW_4_src_width_offset 4
+#define SDMA_PKT_COPY_T2T_DW_4_src_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_4_src_width_shift  16
+#define SDMA_PKT_COPY_T2T_DW_4_SRC_WIDTH(x) (((x) & SDMA_PKT_COPY_T2T_DW_4_src_width_mask) << SDMA_PKT_COPY_T2T_DW_4_src_width_shift)
+
+/*define for DW_5 word*/
+/*define for src_height field*/
+#define SDMA_PKT_COPY_T2T_DW_5_src_height_offset 5
+#define SDMA_PKT_COPY_T2T_DW_5_src_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_5_src_height_shift  0
+#define SDMA_PKT_COPY_T2T_DW_5_SRC_HEIGHT(x) (((x) & SDMA_PKT_COPY_T2T_DW_5_src_height_mask) << SDMA_PKT_COPY_T2T_DW_5_src_height_shift)
+
+/*define for src_depth field*/
+#define SDMA_PKT_COPY_T2T_DW_5_src_depth_offset 5
+#define SDMA_PKT_COPY_T2T_DW_5_src_depth_mask   0x00001FFF
+#define SDMA_PKT_COPY_T2T_DW_5_src_depth_shift  16
+#define SDMA_PKT_COPY_T2T_DW_5_SRC_DEPTH(x) (((x) & SDMA_PKT_COPY_T2T_DW_5_src_depth_mask) << SDMA_PKT_COPY_T2T_DW_5_src_depth_shift)
+
+/*define for DW_6 word*/
+/*define for src_element_size field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_element_size_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_6_src_element_size_shift  0
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_element_size_mask) << SDMA_PKT_COPY_T2T_DW_6_src_element_size_shift)
+
+/*define for src_swizzle_mode field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_swizzle_mode_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_swizzle_mode_mask   0x0000001F
+#define SDMA_PKT_COPY_T2T_DW_6_src_swizzle_mode_shift  3
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_SWIZZLE_MODE(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_swizzle_mode_mask) << SDMA_PKT_COPY_T2T_DW_6_src_swizzle_mode_shift)
+
+/*define for src_dimension field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_dimension_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_dimension_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_6_src_dimension_shift  9
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_DIMENSION(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_dimension_mask) << SDMA_PKT_COPY_T2T_DW_6_src_dimension_shift)
+
+/*define for src_mip_max field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_mip_max_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_mip_max_mask   0x0000000F
+#define SDMA_PKT_COPY_T2T_DW_6_src_mip_max_shift  16
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_MIP_MAX(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_mip_max_mask) << SDMA_PKT_COPY_T2T_DW_6_src_mip_max_shift)
+
+/*define for src_mip_id field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_mip_id_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_mip_id_mask   0x0000000F
+#define SDMA_PKT_COPY_T2T_DW_6_src_mip_id_shift  20
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_MIP_ID(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_mip_id_mask) << SDMA_PKT_COPY_T2T_DW_6_src_mip_id_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_offset 7
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_offset 8
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_9 word*/
+/*define for dst_x field*/
+#define SDMA_PKT_COPY_T2T_DW_9_dst_x_offset 9
+#define SDMA_PKT_COPY_T2T_DW_9_dst_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_9_dst_x_shift  0
+#define SDMA_PKT_COPY_T2T_DW_9_DST_X(x) (((x) & SDMA_PKT_COPY_T2T_DW_9_dst_x_mask) << SDMA_PKT_COPY_T2T_DW_9_dst_x_shift)
+
+/*define for dst_y field*/
+#define SDMA_PKT_COPY_T2T_DW_9_dst_y_offset 9
+#define SDMA_PKT_COPY_T2T_DW_9_dst_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_9_dst_y_shift  16
+#define SDMA_PKT_COPY_T2T_DW_9_DST_Y(x) (((x) & SDMA_PKT_COPY_T2T_DW_9_dst_y_mask) << SDMA_PKT_COPY_T2T_DW_9_dst_y_shift)
+
+/*define for DW_10 word*/
+/*define for dst_z field*/
+#define SDMA_PKT_COPY_T2T_DW_10_dst_z_offset 10
+#define SDMA_PKT_COPY_T2T_DW_10_dst_z_mask   0x00001FFF
+#define SDMA_PKT_COPY_T2T_DW_10_dst_z_shift  0
+#define SDMA_PKT_COPY_T2T_DW_10_DST_Z(x) (((x) & SDMA_PKT_COPY_T2T_DW_10_dst_z_mask) << SDMA_PKT_COPY_T2T_DW_10_dst_z_shift)
+
+/*define for dst_width field*/
+#define SDMA_PKT_COPY_T2T_DW_10_dst_width_offset 10
+#define SDMA_PKT_COPY_T2T_DW_10_dst_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_10_dst_width_shift  16
+#define SDMA_PKT_COPY_T2T_DW_10_DST_WIDTH(x) (((x) & SDMA_PKT_COPY_T2T_DW_10_dst_width_mask) << SDMA_PKT_COPY_T2T_DW_10_dst_width_shift)
+
+/*define for DW_11 word*/
+/*define for dst_height field*/
+#define SDMA_PKT_COPY_T2T_DW_11_dst_height_offset 11
+#define SDMA_PKT_COPY_T2T_DW_11_dst_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_11_dst_height_shift  0
+#define SDMA_PKT_COPY_T2T_DW_11_DST_HEIGHT(x) (((x) & SDMA_PKT_COPY_T2T_DW_11_dst_height_mask) << SDMA_PKT_COPY_T2T_DW_11_dst_height_shift)
+
+/*define for dst_depth field*/
+#define SDMA_PKT_COPY_T2T_DW_11_dst_depth_offset 11
+#define SDMA_PKT_COPY_T2T_DW_11_dst_depth_mask   0x00001FFF
+#define SDMA_PKT_COPY_T2T_DW_11_dst_depth_shift  16
+#define SDMA_PKT_COPY_T2T_DW_11_DST_DEPTH(x) (((x) & SDMA_PKT_COPY_T2T_DW_11_dst_depth_mask) << SDMA_PKT_COPY_T2T_DW_11_dst_depth_shift)
+
+/*define for DW_12 word*/
+/*define for dst_element_size field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_element_size_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_12_dst_element_size_shift  0
+#define SDMA_PKT_COPY_T2T_DW_12_DST_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_element_size_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_element_size_shift)
+
+/*define for dst_swizzle_mode field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_swizzle_mode_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_swizzle_mode_mask   0x0000001F
+#define SDMA_PKT_COPY_T2T_DW_12_dst_swizzle_mode_shift  3
+#define SDMA_PKT_COPY_T2T_DW_12_DST_SWIZZLE_MODE(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_swizzle_mode_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_swizzle_mode_shift)
+
+/*define for dst_dimension field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_dimension_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_dimension_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_12_dst_dimension_shift  9
+#define SDMA_PKT_COPY_T2T_DW_12_DST_DIMENSION(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_dimension_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_dimension_shift)
+
+/*define for dst_mip_max field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mip_max_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mip_max_mask   0x0000000F
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mip_max_shift  16
+#define SDMA_PKT_COPY_T2T_DW_12_DST_MIP_MAX(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_mip_max_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_mip_max_shift)
+
+/*define for dst_mip_id field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mip_id_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mip_id_mask   0x0000000F
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mip_id_shift  20
+#define SDMA_PKT_COPY_T2T_DW_12_DST_MIP_ID(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_mip_id_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_mip_id_shift)
+
+/*define for DW_13 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_T2T_DW_13_rect_x_offset 13
+#define SDMA_PKT_COPY_T2T_DW_13_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_13_rect_x_shift  0
+#define SDMA_PKT_COPY_T2T_DW_13_RECT_X(x) (((x) & SDMA_PKT_COPY_T2T_DW_13_rect_x_mask) << SDMA_PKT_COPY_T2T_DW_13_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_T2T_DW_13_rect_y_offset 13
+#define SDMA_PKT_COPY_T2T_DW_13_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_13_rect_y_shift  16
+#define SDMA_PKT_COPY_T2T_DW_13_RECT_Y(x) (((x) & SDMA_PKT_COPY_T2T_DW_13_rect_y_mask) << SDMA_PKT_COPY_T2T_DW_13_rect_y_shift)
+
+/*define for DW_14 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_T2T_DW_14_rect_z_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_rect_z_mask   0x00001FFF
+#define SDMA_PKT_COPY_T2T_DW_14_rect_z_shift  0
+#define SDMA_PKT_COPY_T2T_DW_14_RECT_Z(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_rect_z_mask) << SDMA_PKT_COPY_T2T_DW_14_rect_z_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_T2T_DW_14_dst_sw_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_14_dst_sw_shift  16
+#define SDMA_PKT_COPY_T2T_DW_14_DST_SW(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_dst_sw_mask) << SDMA_PKT_COPY_T2T_DW_14_dst_sw_shift)
+
+/*define for dst_cache_policy field*/
+#define SDMA_PKT_COPY_T2T_DW_14_dst_cache_policy_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_dst_cache_policy_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_14_dst_cache_policy_shift  18
+#define SDMA_PKT_COPY_T2T_DW_14_DST_CACHE_POLICY(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_dst_cache_policy_mask) << SDMA_PKT_COPY_T2T_DW_14_dst_cache_policy_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_T2T_DW_14_src_sw_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_14_src_sw_shift  24
+#define SDMA_PKT_COPY_T2T_DW_14_SRC_SW(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_src_sw_mask) << SDMA_PKT_COPY_T2T_DW_14_src_sw_shift)
+
+/*define for src_cache_policy field*/
+#define SDMA_PKT_COPY_T2T_DW_14_src_cache_policy_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_src_cache_policy_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_14_src_cache_policy_shift  26
+#define SDMA_PKT_COPY_T2T_DW_14_SRC_CACHE_POLICY(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_src_cache_policy_mask) << SDMA_PKT_COPY_T2T_DW_14_src_cache_policy_shift)
+
+/*define for META_ADDR_LO word*/
+/*define for meta_addr_31_0 field*/
+#define SDMA_PKT_COPY_T2T_META_ADDR_LO_meta_addr_31_0_offset 15
+#define SDMA_PKT_COPY_T2T_META_ADDR_LO_meta_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_META_ADDR_LO_meta_addr_31_0_shift  0
+#define SDMA_PKT_COPY_T2T_META_ADDR_LO_META_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_T2T_META_ADDR_LO_meta_addr_31_0_mask) << SDMA_PKT_COPY_T2T_META_ADDR_LO_meta_addr_31_0_shift)
+
+/*define for META_ADDR_HI word*/
+/*define for meta_addr_63_32 field*/
+#define SDMA_PKT_COPY_T2T_META_ADDR_HI_meta_addr_63_32_offset 16
+#define SDMA_PKT_COPY_T2T_META_ADDR_HI_meta_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_META_ADDR_HI_meta_addr_63_32_shift  0
+#define SDMA_PKT_COPY_T2T_META_ADDR_HI_META_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_T2T_META_ADDR_HI_meta_addr_63_32_mask) << SDMA_PKT_COPY_T2T_META_ADDR_HI_meta_addr_63_32_shift)
+
+/*define for META_CONFIG word*/
+/*define for data_format field*/
+#define SDMA_PKT_COPY_T2T_META_CONFIG_data_format_offset 17
+#define SDMA_PKT_COPY_T2T_META_CONFIG_data_format_mask   0x0000007F
+#define SDMA_PKT_COPY_T2T_META_CONFIG_data_format_shift  0
+#define SDMA_PKT_COPY_T2T_META_CONFIG_DATA_FORMAT(x) (((x) & SDMA_PKT_COPY_T2T_META_CONFIG_data_format_mask) << SDMA_PKT_COPY_T2T_META_CONFIG_data_format_shift)
+
+/*define for color_transform_disable field*/
+#define SDMA_PKT_COPY_T2T_META_CONFIG_color_transform_disable_offset 17
+#define SDMA_PKT_COPY_T2T_META_CONFIG_color_transform_disable_mask   0x00000001
+#define SDMA_PKT_COPY_T2T_META_CONFIG_color_transform_disable_shift  7
+#define SDMA_PKT_COPY_T2T_META_CONFIG_COLOR_TRANSFORM_DISABLE(x) (((x) & SDMA_PKT_COPY_T2T_META_CONFIG_color_transform_disable_mask) << SDMA_PKT_COPY_T2T_META_CONFIG_color_transform_disable_shift)
+
+/*define for alpha_is_on_msb field*/
+#define SDMA_PKT_COPY_T2T_META_CONFIG_alpha_is_on_msb_offset 17
+#define SDMA_PKT_COPY_T2T_META_CONFIG_alpha_is_on_msb_mask   0x00000001
+#define SDMA_PKT_COPY_T2T_META_CONFIG_alpha_is_on_msb_shift  8
+#define SDMA_PKT_COPY_T2T_META_CONFIG_ALPHA_IS_ON_MSB(x) (((x) & SDMA_PKT_COPY_T2T_META_CONFIG_alpha_is_on_msb_mask) << SDMA_PKT_COPY_T2T_META_CONFIG_alpha_is_on_msb_shift)
+
+/*define for number_type field*/
+#define SDMA_PKT_COPY_T2T_META_CONFIG_number_type_offset 17
+#define SDMA_PKT_COPY_T2T_META_CONFIG_number_type_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_META_CONFIG_number_type_shift  9
+#define SDMA_PKT_COPY_T2T_META_CONFIG_NUMBER_TYPE(x) (((x) & SDMA_PKT_COPY_T2T_META_CONFIG_number_type_mask) << SDMA_PKT_COPY_T2T_META_CONFIG_number_type_shift)
+
+/*define for surface_type field*/
+#define SDMA_PKT_COPY_T2T_META_CONFIG_surface_type_offset 17
+#define SDMA_PKT_COPY_T2T_META_CONFIG_surface_type_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_META_CONFIG_surface_type_shift  12
+#define SDMA_PKT_COPY_T2T_META_CONFIG_SURFACE_TYPE(x) (((x) & SDMA_PKT_COPY_T2T_META_CONFIG_surface_type_mask) << SDMA_PKT_COPY_T2T_META_CONFIG_surface_type_shift)
+
+/*define for meta_llc field*/
+#define SDMA_PKT_COPY_T2T_META_CONFIG_meta_llc_offset 17
+#define SDMA_PKT_COPY_T2T_META_CONFIG_meta_llc_mask   0x00000001
+#define SDMA_PKT_COPY_T2T_META_CONFIG_meta_llc_shift  14
+#define SDMA_PKT_COPY_T2T_META_CONFIG_META_LLC(x) (((x) & SDMA_PKT_COPY_T2T_META_CONFIG_meta_llc_mask) << SDMA_PKT_COPY_T2T_META_CONFIG_meta_llc_shift)
+
+/*define for max_comp_block_size field*/
+#define SDMA_PKT_COPY_T2T_META_CONFIG_max_comp_block_size_offset 17
+#define SDMA_PKT_COPY_T2T_META_CONFIG_max_comp_block_size_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_META_CONFIG_max_comp_block_size_shift  24
+#define SDMA_PKT_COPY_T2T_META_CONFIG_MAX_COMP_BLOCK_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_META_CONFIG_max_comp_block_size_mask) << SDMA_PKT_COPY_T2T_META_CONFIG_max_comp_block_size_shift)
+
+/*define for max_uncomp_block_size field*/
+#define SDMA_PKT_COPY_T2T_META_CONFIG_max_uncomp_block_size_offset 17
+#define SDMA_PKT_COPY_T2T_META_CONFIG_max_uncomp_block_size_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_META_CONFIG_max_uncomp_block_size_shift  26
+#define SDMA_PKT_COPY_T2T_META_CONFIG_MAX_UNCOMP_BLOCK_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_META_CONFIG_max_uncomp_block_size_mask) << SDMA_PKT_COPY_T2T_META_CONFIG_max_uncomp_block_size_shift)
+
+/*define for write_compress_enable field*/
+#define SDMA_PKT_COPY_T2T_META_CONFIG_write_compress_enable_offset 17
+#define SDMA_PKT_COPY_T2T_META_CONFIG_write_compress_enable_mask   0x00000001
+#define SDMA_PKT_COPY_T2T_META_CONFIG_write_compress_enable_shift  28
+#define SDMA_PKT_COPY_T2T_META_CONFIG_WRITE_COMPRESS_ENABLE(x) (((x) & SDMA_PKT_COPY_T2T_META_CONFIG_write_compress_enable_mask) << SDMA_PKT_COPY_T2T_META_CONFIG_write_compress_enable_shift)
+
+/*define for meta_tmz field*/
+#define SDMA_PKT_COPY_T2T_META_CONFIG_meta_tmz_offset 17
+#define SDMA_PKT_COPY_T2T_META_CONFIG_meta_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_T2T_META_CONFIG_meta_tmz_shift  29
+#define SDMA_PKT_COPY_T2T_META_CONFIG_META_TMZ(x) (((x) & SDMA_PKT_COPY_T2T_META_CONFIG_meta_tmz_mask) << SDMA_PKT_COPY_T2T_META_CONFIG_meta_tmz_shift)
+
+/*define for pipe_aligned field*/
+#define SDMA_PKT_COPY_T2T_META_CONFIG_pipe_aligned_offset 17
+#define SDMA_PKT_COPY_T2T_META_CONFIG_pipe_aligned_mask   0x00000001
+#define SDMA_PKT_COPY_T2T_META_CONFIG_pipe_aligned_shift  31
+#define SDMA_PKT_COPY_T2T_META_CONFIG_PIPE_ALIGNED(x) (((x) & SDMA_PKT_COPY_T2T_META_CONFIG_pipe_aligned_mask) << SDMA_PKT_COPY_T2T_META_CONFIG_pipe_aligned_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_T2T_BC packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_T2T_BC_HEADER_op_offset 0
+#define SDMA_PKT_COPY_T2T_BC_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_T2T_BC_HEADER_op_shift  0
+#define SDMA_PKT_COPY_T2T_BC_HEADER_OP(x) (((x) & SDMA_PKT_COPY_T2T_BC_HEADER_op_mask) << SDMA_PKT_COPY_T2T_BC_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_T2T_BC_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_T2T_BC_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_T2T_BC_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_T2T_BC_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_T2T_BC_HEADER_sub_op_mask) << SDMA_PKT_COPY_T2T_BC_HEADER_sub_op_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_T2T_BC_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_COPY_T2T_BC_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_BC_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_T2T_BC_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_T2T_BC_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_T2T_BC_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_T2T_BC_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_COPY_T2T_BC_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_BC_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_T2T_BC_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_T2T_BC_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_T2T_BC_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for src_x field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_3_src_x_offset 3
+#define SDMA_PKT_COPY_T2T_BC_DW_3_src_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_BC_DW_3_src_x_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DW_3_SRC_X(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_3_src_x_mask) << SDMA_PKT_COPY_T2T_BC_DW_3_src_x_shift)
+
+/*define for src_y field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_3_src_y_offset 3
+#define SDMA_PKT_COPY_T2T_BC_DW_3_src_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_BC_DW_3_src_y_shift  16
+#define SDMA_PKT_COPY_T2T_BC_DW_3_SRC_Y(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_3_src_y_mask) << SDMA_PKT_COPY_T2T_BC_DW_3_src_y_shift)
+
+/*define for DW_4 word*/
+/*define for src_z field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_4_src_z_offset 4
+#define SDMA_PKT_COPY_T2T_BC_DW_4_src_z_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_BC_DW_4_src_z_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DW_4_SRC_Z(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_4_src_z_mask) << SDMA_PKT_COPY_T2T_BC_DW_4_src_z_shift)
+
+/*define for src_width field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_4_src_width_offset 4
+#define SDMA_PKT_COPY_T2T_BC_DW_4_src_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_BC_DW_4_src_width_shift  16
+#define SDMA_PKT_COPY_T2T_BC_DW_4_SRC_WIDTH(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_4_src_width_mask) << SDMA_PKT_COPY_T2T_BC_DW_4_src_width_shift)
+
+/*define for DW_5 word*/
+/*define for src_height field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_5_src_height_offset 5
+#define SDMA_PKT_COPY_T2T_BC_DW_5_src_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_BC_DW_5_src_height_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DW_5_SRC_HEIGHT(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_5_src_height_mask) << SDMA_PKT_COPY_T2T_BC_DW_5_src_height_shift)
+
+/*define for src_depth field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_5_src_depth_offset 5
+#define SDMA_PKT_COPY_T2T_BC_DW_5_src_depth_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_BC_DW_5_src_depth_shift  16
+#define SDMA_PKT_COPY_T2T_BC_DW_5_SRC_DEPTH(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_5_src_depth_mask) << SDMA_PKT_COPY_T2T_BC_DW_5_src_depth_shift)
+
+/*define for DW_6 word*/
+/*define for src_element_size field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_element_size_offset 6
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_element_size_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DW_6_SRC_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_6_src_element_size_mask) << SDMA_PKT_COPY_T2T_BC_DW_6_src_element_size_shift)
+
+/*define for src_array_mode field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_array_mode_offset 6
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_array_mode_shift  3
+#define SDMA_PKT_COPY_T2T_BC_DW_6_SRC_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_6_src_array_mode_mask) << SDMA_PKT_COPY_T2T_BC_DW_6_src_array_mode_shift)
+
+/*define for src_mit_mode field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_mit_mode_offset 6
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_mit_mode_shift  8
+#define SDMA_PKT_COPY_T2T_BC_DW_6_SRC_MIT_MODE(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_6_src_mit_mode_mask) << SDMA_PKT_COPY_T2T_BC_DW_6_src_mit_mode_shift)
+
+/*define for src_tilesplit_size field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_tilesplit_size_offset 6
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_T2T_BC_DW_6_SRC_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_6_src_tilesplit_size_mask) << SDMA_PKT_COPY_T2T_BC_DW_6_src_tilesplit_size_shift)
+
+/*define for src_bank_w field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_bank_w_offset 6
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_bank_w_shift  15
+#define SDMA_PKT_COPY_T2T_BC_DW_6_SRC_BANK_W(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_6_src_bank_w_mask) << SDMA_PKT_COPY_T2T_BC_DW_6_src_bank_w_shift)
+
+/*define for src_bank_h field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_bank_h_offset 6
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_bank_h_shift  18
+#define SDMA_PKT_COPY_T2T_BC_DW_6_SRC_BANK_H(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_6_src_bank_h_mask) << SDMA_PKT_COPY_T2T_BC_DW_6_src_bank_h_shift)
+
+/*define for src_num_bank field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_num_bank_offset 6
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_num_bank_shift  21
+#define SDMA_PKT_COPY_T2T_BC_DW_6_SRC_NUM_BANK(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_6_src_num_bank_mask) << SDMA_PKT_COPY_T2T_BC_DW_6_src_num_bank_shift)
+
+/*define for src_mat_aspt field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_mat_aspt_offset 6
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_mat_aspt_shift  24
+#define SDMA_PKT_COPY_T2T_BC_DW_6_SRC_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_6_src_mat_aspt_mask) << SDMA_PKT_COPY_T2T_BC_DW_6_src_mat_aspt_shift)
+
+/*define for src_pipe_config field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_pipe_config_offset 6
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_T2T_BC_DW_6_src_pipe_config_shift  26
+#define SDMA_PKT_COPY_T2T_BC_DW_6_SRC_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_6_src_pipe_config_mask) << SDMA_PKT_COPY_T2T_BC_DW_6_src_pipe_config_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_T2T_BC_DST_ADDR_LO_dst_addr_31_0_offset 7
+#define SDMA_PKT_COPY_T2T_BC_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_BC_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_T2T_BC_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_T2T_BC_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_T2T_BC_DST_ADDR_HI_dst_addr_63_32_offset 8
+#define SDMA_PKT_COPY_T2T_BC_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_BC_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_T2T_BC_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_T2T_BC_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_9 word*/
+/*define for dst_x field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_9_dst_x_offset 9
+#define SDMA_PKT_COPY_T2T_BC_DW_9_dst_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_BC_DW_9_dst_x_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DW_9_DST_X(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_9_dst_x_mask) << SDMA_PKT_COPY_T2T_BC_DW_9_dst_x_shift)
+
+/*define for dst_y field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_9_dst_y_offset 9
+#define SDMA_PKT_COPY_T2T_BC_DW_9_dst_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_BC_DW_9_dst_y_shift  16
+#define SDMA_PKT_COPY_T2T_BC_DW_9_DST_Y(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_9_dst_y_mask) << SDMA_PKT_COPY_T2T_BC_DW_9_dst_y_shift)
+
+/*define for DW_10 word*/
+/*define for dst_z field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_10_dst_z_offset 10
+#define SDMA_PKT_COPY_T2T_BC_DW_10_dst_z_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_BC_DW_10_dst_z_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DW_10_DST_Z(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_10_dst_z_mask) << SDMA_PKT_COPY_T2T_BC_DW_10_dst_z_shift)
+
+/*define for dst_width field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_10_dst_width_offset 10
+#define SDMA_PKT_COPY_T2T_BC_DW_10_dst_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_BC_DW_10_dst_width_shift  16
+#define SDMA_PKT_COPY_T2T_BC_DW_10_DST_WIDTH(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_10_dst_width_mask) << SDMA_PKT_COPY_T2T_BC_DW_10_dst_width_shift)
+
+/*define for DW_11 word*/
+/*define for dst_height field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_11_dst_height_offset 11
+#define SDMA_PKT_COPY_T2T_BC_DW_11_dst_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_BC_DW_11_dst_height_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DW_11_DST_HEIGHT(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_11_dst_height_mask) << SDMA_PKT_COPY_T2T_BC_DW_11_dst_height_shift)
+
+/*define for dst_depth field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_11_dst_depth_offset 11
+#define SDMA_PKT_COPY_T2T_BC_DW_11_dst_depth_mask   0x00000FFF
+#define SDMA_PKT_COPY_T2T_BC_DW_11_dst_depth_shift  16
+#define SDMA_PKT_COPY_T2T_BC_DW_11_DST_DEPTH(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_11_dst_depth_mask) << SDMA_PKT_COPY_T2T_BC_DW_11_dst_depth_shift)
+
+/*define for DW_12 word*/
+/*define for dst_element_size field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_element_size_offset 12
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_element_size_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DW_12_DST_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_12_dst_element_size_mask) << SDMA_PKT_COPY_T2T_BC_DW_12_dst_element_size_shift)
+
+/*define for dst_array_mode field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_array_mode_offset 12
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_array_mode_shift  3
+#define SDMA_PKT_COPY_T2T_BC_DW_12_DST_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_12_dst_array_mode_mask) << SDMA_PKT_COPY_T2T_BC_DW_12_dst_array_mode_shift)
+
+/*define for dst_mit_mode field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_mit_mode_offset 12
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_mit_mode_shift  8
+#define SDMA_PKT_COPY_T2T_BC_DW_12_DST_MIT_MODE(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_12_dst_mit_mode_mask) << SDMA_PKT_COPY_T2T_BC_DW_12_dst_mit_mode_shift)
+
+/*define for dst_tilesplit_size field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_tilesplit_size_offset 12
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_T2T_BC_DW_12_DST_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_12_dst_tilesplit_size_mask) << SDMA_PKT_COPY_T2T_BC_DW_12_dst_tilesplit_size_shift)
+
+/*define for dst_bank_w field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_bank_w_offset 12
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_bank_w_shift  15
+#define SDMA_PKT_COPY_T2T_BC_DW_12_DST_BANK_W(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_12_dst_bank_w_mask) << SDMA_PKT_COPY_T2T_BC_DW_12_dst_bank_w_shift)
+
+/*define for dst_bank_h field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_bank_h_offset 12
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_bank_h_shift  18
+#define SDMA_PKT_COPY_T2T_BC_DW_12_DST_BANK_H(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_12_dst_bank_h_mask) << SDMA_PKT_COPY_T2T_BC_DW_12_dst_bank_h_shift)
+
+/*define for dst_num_bank field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_num_bank_offset 12
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_num_bank_shift  21
+#define SDMA_PKT_COPY_T2T_BC_DW_12_DST_NUM_BANK(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_12_dst_num_bank_mask) << SDMA_PKT_COPY_T2T_BC_DW_12_dst_num_bank_shift)
+
+/*define for dst_mat_aspt field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_mat_aspt_offset 12
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_mat_aspt_shift  24
+#define SDMA_PKT_COPY_T2T_BC_DW_12_DST_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_12_dst_mat_aspt_mask) << SDMA_PKT_COPY_T2T_BC_DW_12_dst_mat_aspt_shift)
+
+/*define for dst_pipe_config field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_pipe_config_offset 12
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_T2T_BC_DW_12_dst_pipe_config_shift  26
+#define SDMA_PKT_COPY_T2T_BC_DW_12_DST_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_12_dst_pipe_config_mask) << SDMA_PKT_COPY_T2T_BC_DW_12_dst_pipe_config_shift)
+
+/*define for DW_13 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_13_rect_x_offset 13
+#define SDMA_PKT_COPY_T2T_BC_DW_13_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_BC_DW_13_rect_x_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DW_13_RECT_X(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_13_rect_x_mask) << SDMA_PKT_COPY_T2T_BC_DW_13_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_13_rect_y_offset 13
+#define SDMA_PKT_COPY_T2T_BC_DW_13_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_BC_DW_13_rect_y_shift  16
+#define SDMA_PKT_COPY_T2T_BC_DW_13_RECT_Y(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_13_rect_y_mask) << SDMA_PKT_COPY_T2T_BC_DW_13_rect_y_shift)
+
+/*define for DW_14 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_14_rect_z_offset 14
+#define SDMA_PKT_COPY_T2T_BC_DW_14_rect_z_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_BC_DW_14_rect_z_shift  0
+#define SDMA_PKT_COPY_T2T_BC_DW_14_RECT_Z(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_14_rect_z_mask) << SDMA_PKT_COPY_T2T_BC_DW_14_rect_z_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_14_dst_sw_offset 14
+#define SDMA_PKT_COPY_T2T_BC_DW_14_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_BC_DW_14_dst_sw_shift  16
+#define SDMA_PKT_COPY_T2T_BC_DW_14_DST_SW(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_14_dst_sw_mask) << SDMA_PKT_COPY_T2T_BC_DW_14_dst_sw_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_T2T_BC_DW_14_src_sw_offset 14
+#define SDMA_PKT_COPY_T2T_BC_DW_14_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_BC_DW_14_src_sw_shift  24
+#define SDMA_PKT_COPY_T2T_BC_DW_14_SRC_SW(x) (((x) & SDMA_PKT_COPY_T2T_BC_DW_14_src_sw_mask) << SDMA_PKT_COPY_T2T_BC_DW_14_src_sw_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_TILED_SUBWIN packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_OP(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_tmz_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_tmz_shift)
+
+/*define for dcc field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_dcc_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_dcc_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_dcc_shift  19
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_DCC(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_dcc_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_dcc_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_cpv_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_cpv_shift  28
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_CPV(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_cpv_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_cpv_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_shift)
+
+/*define for TILED_ADDR_LO word*/
+/*define for tiled_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_offset 1
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_TILED_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_mask) << SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_shift)
+
+/*define for TILED_ADDR_HI word*/
+/*define for tiled_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_offset 2
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_TILED_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_mask) << SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for tiled_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_offset 3
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_TILED_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_shift)
+
+/*define for tiled_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_offset 3
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_TILED_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_shift)
+
+/*define for DW_4 word*/
+/*define for tiled_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_offset 4
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_mask   0x00001FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_TILED_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_shift)
+
+/*define for width field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_width_offset 4
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_width_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_WIDTH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_4_width_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_4_width_shift)
+
+/*define for DW_5 word*/
+/*define for height field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_height_offset 5
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_height_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_HEIGHT(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_5_height_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_5_height_shift)
+
+/*define for depth field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_depth_offset 5
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_depth_mask   0x00001FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_depth_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_DEPTH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_5_depth_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_5_depth_shift)
+
+/*define for DW_6 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_shift)
+
+/*define for swizzle_mode field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_swizzle_mode_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_swizzle_mode_mask   0x0000001F
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_swizzle_mode_shift  3
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_SWIZZLE_MODE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_swizzle_mode_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_swizzle_mode_shift)
+
+/*define for dimension field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_dimension_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_dimension_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_dimension_shift  9
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_DIMENSION(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_dimension_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_dimension_shift)
+
+/*define for mip_max field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mip_max_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mip_max_mask   0x0000000F
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mip_max_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_MIP_MAX(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mip_max_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mip_max_shift)
+
+/*define for mip_id field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mip_id_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mip_id_mask   0x0000000F
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mip_id_shift  20
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_MIP_ID(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mip_id_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mip_id_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_offset 7
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_offset 8
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for DW_9 word*/
+/*define for linear_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_offset 9
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_LINEAR_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_shift)
+
+/*define for linear_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_offset 9
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_LINEAR_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_shift)
+
+/*define for DW_10 word*/
+/*define for linear_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_offset 10
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_mask   0x00001FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_LINEAR_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_shift)
+
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_offset 10
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_shift)
+
+/*define for DW_11 word*/
+/*define for linear_slice_pitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_offset 11
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_LINEAR_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_shift)
+
+/*define for DW_12 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_offset 12
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_RECT_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_offset 12
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_RECT_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_shift)
+
+/*define for DW_13 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_mask   0x00001FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_RECT_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_shift)
+
+/*define for linear_cache_policy field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_cache_policy_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_cache_policy_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_cache_policy_shift  18
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_LINEAR_CACHE_POLICY(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_cache_policy_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_cache_policy_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_shift  24
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_TILE_SW(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_shift)
+
+/*define for tile_cache_policy field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_cache_policy_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_cache_policy_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_cache_policy_shift  26
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_TILE_CACHE_POLICY(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_cache_policy_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_cache_policy_shift)
+
+/*define for META_ADDR_LO word*/
+/*define for meta_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_LO_meta_addr_31_0_offset 14
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_LO_meta_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_LO_meta_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_LO_META_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_LO_meta_addr_31_0_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_LO_meta_addr_31_0_shift)
+
+/*define for META_ADDR_HI word*/
+/*define for meta_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_HI_meta_addr_63_32_offset 15
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_HI_meta_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_HI_meta_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_HI_META_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_HI_meta_addr_63_32_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_ADDR_HI_meta_addr_63_32_shift)
+
+/*define for META_CONFIG word*/
+/*define for data_format field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_data_format_offset 16
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_data_format_mask   0x0000007F
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_data_format_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_DATA_FORMAT(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_data_format_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_data_format_shift)
+
+/*define for color_transform_disable field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_color_transform_disable_offset 16
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_color_transform_disable_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_color_transform_disable_shift  7
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_COLOR_TRANSFORM_DISABLE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_color_transform_disable_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_color_transform_disable_shift)
+
+/*define for alpha_is_on_msb field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_alpha_is_on_msb_offset 16
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_alpha_is_on_msb_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_alpha_is_on_msb_shift  8
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_ALPHA_IS_ON_MSB(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_alpha_is_on_msb_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_alpha_is_on_msb_shift)
+
+/*define for number_type field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_number_type_offset 16
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_number_type_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_number_type_shift  9
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_NUMBER_TYPE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_number_type_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_number_type_shift)
+
+/*define for surface_type field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_surface_type_offset 16
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_surface_type_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_surface_type_shift  12
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_SURFACE_TYPE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_surface_type_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_surface_type_shift)
+
+/*define for meta_llc field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_meta_llc_offset 16
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_meta_llc_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_meta_llc_shift  14
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_META_LLC(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_meta_llc_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_meta_llc_shift)
+
+/*define for max_comp_block_size field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_max_comp_block_size_offset 16
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_max_comp_block_size_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_max_comp_block_size_shift  24
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_MAX_COMP_BLOCK_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_max_comp_block_size_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_max_comp_block_size_shift)
+
+/*define for max_uncomp_block_size field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_max_uncomp_block_size_offset 16
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_max_uncomp_block_size_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_max_uncomp_block_size_shift  26
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_MAX_UNCOMP_BLOCK_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_max_uncomp_block_size_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_max_uncomp_block_size_shift)
+
+/*define for write_compress_enable field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_write_compress_enable_offset 16
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_write_compress_enable_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_write_compress_enable_shift  28
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_WRITE_COMPRESS_ENABLE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_write_compress_enable_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_write_compress_enable_shift)
+
+/*define for meta_tmz field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_meta_tmz_offset 16
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_meta_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_meta_tmz_shift  29
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_META_TMZ(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_meta_tmz_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_meta_tmz_shift)
+
+/*define for pipe_aligned field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_pipe_aligned_offset 16
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_pipe_aligned_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_pipe_aligned_shift  31
+#define SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_PIPE_ALIGNED(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_pipe_aligned_mask) << SDMA_PKT_COPY_TILED_SUBWIN_META_CONFIG_pipe_aligned_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_TILED_SUBWIN_BC packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_op_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_op_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_OP(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_op_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_sub_op_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_sub_op_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_detile_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_HEADER_detile_shift)
+
+/*define for TILED_ADDR_LO word*/
+/*define for tiled_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_LO_tiled_addr_31_0_offset 1
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_LO_tiled_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_LO_tiled_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_LO_TILED_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_LO_tiled_addr_31_0_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_LO_tiled_addr_31_0_shift)
+
+/*define for TILED_ADDR_HI word*/
+/*define for tiled_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_HI_tiled_addr_63_32_offset 2
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_HI_tiled_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_HI_tiled_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_HI_TILED_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_HI_tiled_addr_63_32_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_TILED_ADDR_HI_tiled_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for tiled_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_tiled_x_offset 3
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_tiled_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_tiled_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_TILED_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_tiled_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_tiled_x_shift)
+
+/*define for tiled_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_tiled_y_offset 3
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_tiled_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_tiled_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_TILED_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_tiled_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_3_tiled_y_shift)
+
+/*define for DW_4 word*/
+/*define for tiled_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_tiled_z_offset 4
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_tiled_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_tiled_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_TILED_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_tiled_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_tiled_z_shift)
+
+/*define for width field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_width_offset 4
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_width_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_WIDTH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_width_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_4_width_shift)
+
+/*define for DW_5 word*/
+/*define for height field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_height_offset 5
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_height_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_HEIGHT(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_height_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_height_shift)
+
+/*define for depth field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_depth_offset 5
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_depth_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_depth_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_DEPTH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_depth_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_5_depth_shift)
+
+/*define for DW_6 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_element_size_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_element_size_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_element_size_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_element_size_shift)
+
+/*define for array_mode field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_array_mode_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_array_mode_shift  3
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_array_mode_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_array_mode_shift)
+
+/*define for mit_mode field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_mit_mode_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_mit_mode_shift  8
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_MIT_MODE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_mit_mode_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_mit_mode_shift)
+
+/*define for tilesplit_size field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_tilesplit_size_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_tilesplit_size_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_tilesplit_size_shift)
+
+/*define for bank_w field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_bank_w_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_bank_w_shift  15
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_BANK_W(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_bank_w_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_bank_w_shift)
+
+/*define for bank_h field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_bank_h_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_bank_h_shift  18
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_BANK_H(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_bank_h_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_bank_h_shift)
+
+/*define for num_bank field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_num_bank_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_num_bank_shift  21
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_NUM_BANK(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_num_bank_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_num_bank_shift)
+
+/*define for mat_aspt field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_mat_aspt_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_mat_aspt_shift  24
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_mat_aspt_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_mat_aspt_shift)
+
+/*define for pipe_config field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_pipe_config_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_pipe_config_shift  26
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_pipe_config_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_6_pipe_config_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_LO_linear_addr_31_0_offset 7
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_HI_linear_addr_63_32_offset 8
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for DW_9 word*/
+/*define for linear_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_linear_x_offset 9
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_linear_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_linear_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_LINEAR_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_linear_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_linear_x_shift)
+
+/*define for linear_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_linear_y_offset 9
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_linear_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_linear_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_LINEAR_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_linear_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_9_linear_y_shift)
+
+/*define for DW_10 word*/
+/*define for linear_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_linear_z_offset 10
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_linear_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_linear_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_LINEAR_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_linear_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_linear_z_shift)
+
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_linear_pitch_offset 10
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_linear_pitch_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_linear_pitch_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_linear_pitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_10_linear_pitch_shift)
+
+/*define for DW_11 word*/
+/*define for linear_slice_pitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_11_linear_slice_pitch_offset 11
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_11_linear_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_11_linear_slice_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_11_LINEAR_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_11_linear_slice_pitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_11_linear_slice_pitch_shift)
+
+/*define for DW_12 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_rect_x_offset 12
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_rect_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_RECT_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_rect_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_rect_y_offset 12
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_rect_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_RECT_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_rect_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_12_rect_y_shift)
+
+/*define for DW_13 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_rect_z_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_rect_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_rect_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_RECT_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_rect_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_rect_z_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_linear_sw_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_linear_sw_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_linear_sw_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_linear_sw_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_tile_sw_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_tile_sw_shift  24
+#define SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_TILE_SW(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_tile_sw_mask) << SDMA_PKT_COPY_TILED_SUBWIN_BC_DW_13_tile_sw_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_STRUCT packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_op_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_STRUCT_HEADER_op_shift  0
+#define SDMA_PKT_COPY_STRUCT_HEADER_OP(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_op_mask) << SDMA_PKT_COPY_STRUCT_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_STRUCT_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_STRUCT_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_sub_op_mask) << SDMA_PKT_COPY_STRUCT_HEADER_sub_op_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_STRUCT_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_STRUCT_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_tmz_mask) << SDMA_PKT_COPY_STRUCT_HEADER_tmz_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_cpv_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_COPY_STRUCT_HEADER_cpv_shift  28
+#define SDMA_PKT_COPY_STRUCT_HEADER_CPV(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_cpv_mask) << SDMA_PKT_COPY_STRUCT_HEADER_cpv_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_STRUCT_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_STRUCT_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_detile_mask) << SDMA_PKT_COPY_STRUCT_HEADER_detile_shift)
+
+/*define for SB_ADDR_LO word*/
+/*define for sb_addr_31_0 field*/
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_offset 1
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_shift  0
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_SB_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_mask) << SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_shift)
+
+/*define for SB_ADDR_HI word*/
+/*define for sb_addr_63_32 field*/
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_offset 2
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_shift  0
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_SB_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_mask) << SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_shift)
+
+/*define for START_INDEX word*/
+/*define for start_index field*/
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_offset 3
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_shift  0
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_START_INDEX(x) (((x) & SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_mask) << SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_STRUCT_COUNT_count_offset 4
+#define SDMA_PKT_COPY_STRUCT_COUNT_count_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_COUNT_count_shift  0
+#define SDMA_PKT_COPY_STRUCT_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_STRUCT_COUNT_count_mask) << SDMA_PKT_COPY_STRUCT_COUNT_count_shift)
+
+/*define for DW_5 word*/
+/*define for stride field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_stride_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_stride_mask   0x000007FF
+#define SDMA_PKT_COPY_STRUCT_DW_5_stride_shift  0
+#define SDMA_PKT_COPY_STRUCT_DW_5_STRIDE(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_stride_mask) << SDMA_PKT_COPY_STRUCT_DW_5_stride_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_shift  16
+#define SDMA_PKT_COPY_STRUCT_DW_5_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_mask) << SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_shift)
+
+/*define for linear_cache_policy field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_cache_policy_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_cache_policy_mask   0x00000007
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_cache_policy_shift  18
+#define SDMA_PKT_COPY_STRUCT_DW_5_LINEAR_CACHE_POLICY(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_linear_cache_policy_mask) << SDMA_PKT_COPY_STRUCT_DW_5_linear_cache_policy_shift)
+
+/*define for struct_sw field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_mask   0x00000003
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_shift  24
+#define SDMA_PKT_COPY_STRUCT_DW_5_STRUCT_SW(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_mask) << SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_shift)
+
+/*define for struct_cache_policy field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_cache_policy_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_cache_policy_mask   0x00000007
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_cache_policy_shift  26
+#define SDMA_PKT_COPY_STRUCT_DW_5_STRUCT_CACHE_POLICY(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_struct_cache_policy_mask) << SDMA_PKT_COPY_STRUCT_DW_5_struct_cache_policy_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_offset 6
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_offset 7
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_UNTILED packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_UNTILED_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_UNTILED_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_UNTILED_HEADER_op_mask) << SDMA_PKT_WRITE_UNTILED_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_UNTILED_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_mask) << SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_shift)
+
+/*define for encrypt field*/
+#define SDMA_PKT_WRITE_UNTILED_HEADER_encrypt_offset 0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_encrypt_mask   0x00000001
+#define SDMA_PKT_WRITE_UNTILED_HEADER_encrypt_shift  16
+#define SDMA_PKT_WRITE_UNTILED_HEADER_ENCRYPT(x) (((x) & SDMA_PKT_WRITE_UNTILED_HEADER_encrypt_mask) << SDMA_PKT_WRITE_UNTILED_HEADER_encrypt_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_WRITE_UNTILED_HEADER_tmz_offset 0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_WRITE_UNTILED_HEADER_tmz_shift  18
+#define SDMA_PKT_WRITE_UNTILED_HEADER_TMZ(x) (((x) & SDMA_PKT_WRITE_UNTILED_HEADER_tmz_mask) << SDMA_PKT_WRITE_UNTILED_HEADER_tmz_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_WRITE_UNTILED_HEADER_cpv_offset 0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_WRITE_UNTILED_HEADER_cpv_shift  28
+#define SDMA_PKT_WRITE_UNTILED_HEADER_CPV(x) (((x) & SDMA_PKT_WRITE_UNTILED_HEADER_cpv_mask) << SDMA_PKT_WRITE_UNTILED_HEADER_cpv_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_UNTILED_DW_3_count_offset 3
+#define SDMA_PKT_WRITE_UNTILED_DW_3_count_mask   0x000FFFFF
+#define SDMA_PKT_WRITE_UNTILED_DW_3_count_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(x) (((x) & SDMA_PKT_WRITE_UNTILED_DW_3_count_mask) << SDMA_PKT_WRITE_UNTILED_DW_3_count_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_WRITE_UNTILED_DW_3_sw_offset 3
+#define SDMA_PKT_WRITE_UNTILED_DW_3_sw_mask   0x00000003
+#define SDMA_PKT_WRITE_UNTILED_DW_3_sw_shift  24
+#define SDMA_PKT_WRITE_UNTILED_DW_3_SW(x) (((x) & SDMA_PKT_WRITE_UNTILED_DW_3_sw_mask) << SDMA_PKT_WRITE_UNTILED_DW_3_sw_shift)
+
+/*define for cache_policy field*/
+#define SDMA_PKT_WRITE_UNTILED_DW_3_cache_policy_offset 3
+#define SDMA_PKT_WRITE_UNTILED_DW_3_cache_policy_mask   0x00000007
+#define SDMA_PKT_WRITE_UNTILED_DW_3_cache_policy_shift  26
+#define SDMA_PKT_WRITE_UNTILED_DW_3_CACHE_POLICY(x) (((x) & SDMA_PKT_WRITE_UNTILED_DW_3_cache_policy_mask) << SDMA_PKT_WRITE_UNTILED_DW_3_cache_policy_shift)
+
+/*define for DATA0 word*/
+/*define for data0 field*/
+#define SDMA_PKT_WRITE_UNTILED_DATA0_data0_offset 4
+#define SDMA_PKT_WRITE_UNTILED_DATA0_data0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_UNTILED_DATA0_data0_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DATA0_DATA0(x) (((x) & SDMA_PKT_WRITE_UNTILED_DATA0_data0_mask) << SDMA_PKT_WRITE_UNTILED_DATA0_data0_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_TILED packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_TILED_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_TILED_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_TILED_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_TILED_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_TILED_HEADER_op_mask) << SDMA_PKT_WRITE_TILED_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_TILED_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_TILED_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_TILED_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_TILED_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_TILED_HEADER_sub_op_mask) << SDMA_PKT_WRITE_TILED_HEADER_sub_op_shift)
+
+/*define for encrypt field*/
+#define SDMA_PKT_WRITE_TILED_HEADER_encrypt_offset 0
+#define SDMA_PKT_WRITE_TILED_HEADER_encrypt_mask   0x00000001
+#define SDMA_PKT_WRITE_TILED_HEADER_encrypt_shift  16
+#define SDMA_PKT_WRITE_TILED_HEADER_ENCRYPT(x) (((x) & SDMA_PKT_WRITE_TILED_HEADER_encrypt_mask) << SDMA_PKT_WRITE_TILED_HEADER_encrypt_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_WRITE_TILED_HEADER_tmz_offset 0
+#define SDMA_PKT_WRITE_TILED_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_WRITE_TILED_HEADER_tmz_shift  18
+#define SDMA_PKT_WRITE_TILED_HEADER_TMZ(x) (((x) & SDMA_PKT_WRITE_TILED_HEADER_tmz_mask) << SDMA_PKT_WRITE_TILED_HEADER_tmz_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_WRITE_TILED_HEADER_cpv_offset 0
+#define SDMA_PKT_WRITE_TILED_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_WRITE_TILED_HEADER_cpv_shift  28
+#define SDMA_PKT_WRITE_TILED_HEADER_CPV(x) (((x) & SDMA_PKT_WRITE_TILED_HEADER_cpv_mask) << SDMA_PKT_WRITE_TILED_HEADER_cpv_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for width field*/
+#define SDMA_PKT_WRITE_TILED_DW_3_width_offset 3
+#define SDMA_PKT_WRITE_TILED_DW_3_width_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_3_width_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_3_WIDTH(x) (((x) & SDMA_PKT_WRITE_TILED_DW_3_width_mask) << SDMA_PKT_WRITE_TILED_DW_3_width_shift)
+
+/*define for DW_4 word*/
+/*define for height field*/
+#define SDMA_PKT_WRITE_TILED_DW_4_height_offset 4
+#define SDMA_PKT_WRITE_TILED_DW_4_height_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_4_height_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_4_HEIGHT(x) (((x) & SDMA_PKT_WRITE_TILED_DW_4_height_mask) << SDMA_PKT_WRITE_TILED_DW_4_height_shift)
+
+/*define for depth field*/
+#define SDMA_PKT_WRITE_TILED_DW_4_depth_offset 4
+#define SDMA_PKT_WRITE_TILED_DW_4_depth_mask   0x00001FFF
+#define SDMA_PKT_WRITE_TILED_DW_4_depth_shift  16
+#define SDMA_PKT_WRITE_TILED_DW_4_DEPTH(x) (((x) & SDMA_PKT_WRITE_TILED_DW_4_depth_mask) << SDMA_PKT_WRITE_TILED_DW_4_depth_shift)
+
+/*define for DW_5 word*/
+/*define for element_size field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_element_size_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_element_size_mask   0x00000007
+#define SDMA_PKT_WRITE_TILED_DW_5_element_size_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_5_ELEMENT_SIZE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_element_size_mask) << SDMA_PKT_WRITE_TILED_DW_5_element_size_shift)
+
+/*define for swizzle_mode field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_swizzle_mode_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_swizzle_mode_mask   0x0000001F
+#define SDMA_PKT_WRITE_TILED_DW_5_swizzle_mode_shift  3
+#define SDMA_PKT_WRITE_TILED_DW_5_SWIZZLE_MODE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_swizzle_mode_mask) << SDMA_PKT_WRITE_TILED_DW_5_swizzle_mode_shift)
+
+/*define for dimension field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_dimension_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_dimension_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_5_dimension_shift  9
+#define SDMA_PKT_WRITE_TILED_DW_5_DIMENSION(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_dimension_mask) << SDMA_PKT_WRITE_TILED_DW_5_dimension_shift)
+
+/*define for mip_max field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_mip_max_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_mip_max_mask   0x0000000F
+#define SDMA_PKT_WRITE_TILED_DW_5_mip_max_shift  16
+#define SDMA_PKT_WRITE_TILED_DW_5_MIP_MAX(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_mip_max_mask) << SDMA_PKT_WRITE_TILED_DW_5_mip_max_shift)
+
+/*define for DW_6 word*/
+/*define for x field*/
+#define SDMA_PKT_WRITE_TILED_DW_6_x_offset 6
+#define SDMA_PKT_WRITE_TILED_DW_6_x_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_6_x_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_6_X(x) (((x) & SDMA_PKT_WRITE_TILED_DW_6_x_mask) << SDMA_PKT_WRITE_TILED_DW_6_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_WRITE_TILED_DW_6_y_offset 6
+#define SDMA_PKT_WRITE_TILED_DW_6_y_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_6_y_shift  16
+#define SDMA_PKT_WRITE_TILED_DW_6_Y(x) (((x) & SDMA_PKT_WRITE_TILED_DW_6_y_mask) << SDMA_PKT_WRITE_TILED_DW_6_y_shift)
+
+/*define for DW_7 word*/
+/*define for z field*/
+#define SDMA_PKT_WRITE_TILED_DW_7_z_offset 7
+#define SDMA_PKT_WRITE_TILED_DW_7_z_mask   0x00001FFF
+#define SDMA_PKT_WRITE_TILED_DW_7_z_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_7_Z(x) (((x) & SDMA_PKT_WRITE_TILED_DW_7_z_mask) << SDMA_PKT_WRITE_TILED_DW_7_z_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_WRITE_TILED_DW_7_sw_offset 7
+#define SDMA_PKT_WRITE_TILED_DW_7_sw_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_7_sw_shift  24
+#define SDMA_PKT_WRITE_TILED_DW_7_SW(x) (((x) & SDMA_PKT_WRITE_TILED_DW_7_sw_mask) << SDMA_PKT_WRITE_TILED_DW_7_sw_shift)
+
+/*define for cache_policy field*/
+#define SDMA_PKT_WRITE_TILED_DW_7_cache_policy_offset 7
+#define SDMA_PKT_WRITE_TILED_DW_7_cache_policy_mask   0x00000007
+#define SDMA_PKT_WRITE_TILED_DW_7_cache_policy_shift  26
+#define SDMA_PKT_WRITE_TILED_DW_7_CACHE_POLICY(x) (((x) & SDMA_PKT_WRITE_TILED_DW_7_cache_policy_mask) << SDMA_PKT_WRITE_TILED_DW_7_cache_policy_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_TILED_COUNT_count_offset 8
+#define SDMA_PKT_WRITE_TILED_COUNT_count_mask   0x000FFFFF
+#define SDMA_PKT_WRITE_TILED_COUNT_count_shift  0
+#define SDMA_PKT_WRITE_TILED_COUNT_COUNT(x) (((x) & SDMA_PKT_WRITE_TILED_COUNT_count_mask) << SDMA_PKT_WRITE_TILED_COUNT_count_shift)
+
+/*define for DATA0 word*/
+/*define for data0 field*/
+#define SDMA_PKT_WRITE_TILED_DATA0_data0_offset 9
+#define SDMA_PKT_WRITE_TILED_DATA0_data0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_DATA0_data0_shift  0
+#define SDMA_PKT_WRITE_TILED_DATA0_DATA0(x) (((x) & SDMA_PKT_WRITE_TILED_DATA0_data0_mask) << SDMA_PKT_WRITE_TILED_DATA0_data0_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_TILED_BC packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_TILED_BC_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_TILED_BC_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_TILED_BC_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_TILED_BC_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_TILED_BC_HEADER_op_mask) << SDMA_PKT_WRITE_TILED_BC_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_TILED_BC_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_TILED_BC_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_TILED_BC_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_TILED_BC_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_TILED_BC_HEADER_sub_op_mask) << SDMA_PKT_WRITE_TILED_BC_HEADER_sub_op_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_TILED_BC_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_TILED_BC_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_BC_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_TILED_BC_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_TILED_BC_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_TILED_BC_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_TILED_BC_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_BC_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_TILED_BC_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_TILED_BC_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for width field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_3_width_offset 3
+#define SDMA_PKT_WRITE_TILED_BC_DW_3_width_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_BC_DW_3_width_shift  0
+#define SDMA_PKT_WRITE_TILED_BC_DW_3_WIDTH(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_3_width_mask) << SDMA_PKT_WRITE_TILED_BC_DW_3_width_shift)
+
+/*define for DW_4 word*/
+/*define for height field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_4_height_offset 4
+#define SDMA_PKT_WRITE_TILED_BC_DW_4_height_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_BC_DW_4_height_shift  0
+#define SDMA_PKT_WRITE_TILED_BC_DW_4_HEIGHT(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_4_height_mask) << SDMA_PKT_WRITE_TILED_BC_DW_4_height_shift)
+
+/*define for depth field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_4_depth_offset 4
+#define SDMA_PKT_WRITE_TILED_BC_DW_4_depth_mask   0x000007FF
+#define SDMA_PKT_WRITE_TILED_BC_DW_4_depth_shift  16
+#define SDMA_PKT_WRITE_TILED_BC_DW_4_DEPTH(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_4_depth_mask) << SDMA_PKT_WRITE_TILED_BC_DW_4_depth_shift)
+
+/*define for DW_5 word*/
+/*define for element_size field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_element_size_offset 5
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_element_size_mask   0x00000007
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_element_size_shift  0
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_ELEMENT_SIZE(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_5_element_size_mask) << SDMA_PKT_WRITE_TILED_BC_DW_5_element_size_shift)
+
+/*define for array_mode field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_array_mode_offset 5
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_array_mode_mask   0x0000000F
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_array_mode_shift  3
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_ARRAY_MODE(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_5_array_mode_mask) << SDMA_PKT_WRITE_TILED_BC_DW_5_array_mode_shift)
+
+/*define for mit_mode field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_mit_mode_offset 5
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_mit_mode_mask   0x00000007
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_mit_mode_shift  8
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_MIT_MODE(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_5_mit_mode_mask) << SDMA_PKT_WRITE_TILED_BC_DW_5_mit_mode_shift)
+
+/*define for tilesplit_size field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_tilesplit_size_offset 5
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_tilesplit_size_shift  11
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_5_tilesplit_size_mask) << SDMA_PKT_WRITE_TILED_BC_DW_5_tilesplit_size_shift)
+
+/*define for bank_w field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_bank_w_offset 5
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_bank_w_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_bank_w_shift  15
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_BANK_W(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_5_bank_w_mask) << SDMA_PKT_WRITE_TILED_BC_DW_5_bank_w_shift)
+
+/*define for bank_h field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_bank_h_offset 5
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_bank_h_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_bank_h_shift  18
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_BANK_H(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_5_bank_h_mask) << SDMA_PKT_WRITE_TILED_BC_DW_5_bank_h_shift)
+
+/*define for num_bank field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_num_bank_offset 5
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_num_bank_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_num_bank_shift  21
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_NUM_BANK(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_5_num_bank_mask) << SDMA_PKT_WRITE_TILED_BC_DW_5_num_bank_shift)
+
+/*define for mat_aspt field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_mat_aspt_offset 5
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_mat_aspt_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_mat_aspt_shift  24
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_MAT_ASPT(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_5_mat_aspt_mask) << SDMA_PKT_WRITE_TILED_BC_DW_5_mat_aspt_shift)
+
+/*define for pipe_config field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_pipe_config_offset 5
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_pipe_config_mask   0x0000001F
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_pipe_config_shift  26
+#define SDMA_PKT_WRITE_TILED_BC_DW_5_PIPE_CONFIG(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_5_pipe_config_mask) << SDMA_PKT_WRITE_TILED_BC_DW_5_pipe_config_shift)
+
+/*define for DW_6 word*/
+/*define for x field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_6_x_offset 6
+#define SDMA_PKT_WRITE_TILED_BC_DW_6_x_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_BC_DW_6_x_shift  0
+#define SDMA_PKT_WRITE_TILED_BC_DW_6_X(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_6_x_mask) << SDMA_PKT_WRITE_TILED_BC_DW_6_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_6_y_offset 6
+#define SDMA_PKT_WRITE_TILED_BC_DW_6_y_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_BC_DW_6_y_shift  16
+#define SDMA_PKT_WRITE_TILED_BC_DW_6_Y(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_6_y_mask) << SDMA_PKT_WRITE_TILED_BC_DW_6_y_shift)
+
+/*define for DW_7 word*/
+/*define for z field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_7_z_offset 7
+#define SDMA_PKT_WRITE_TILED_BC_DW_7_z_mask   0x000007FF
+#define SDMA_PKT_WRITE_TILED_BC_DW_7_z_shift  0
+#define SDMA_PKT_WRITE_TILED_BC_DW_7_Z(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_7_z_mask) << SDMA_PKT_WRITE_TILED_BC_DW_7_z_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_WRITE_TILED_BC_DW_7_sw_offset 7
+#define SDMA_PKT_WRITE_TILED_BC_DW_7_sw_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_BC_DW_7_sw_shift  24
+#define SDMA_PKT_WRITE_TILED_BC_DW_7_SW(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DW_7_sw_mask) << SDMA_PKT_WRITE_TILED_BC_DW_7_sw_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_TILED_BC_COUNT_count_offset 8
+#define SDMA_PKT_WRITE_TILED_BC_COUNT_count_mask   0x000FFFFF
+#define SDMA_PKT_WRITE_TILED_BC_COUNT_count_shift  2
+#define SDMA_PKT_WRITE_TILED_BC_COUNT_COUNT(x) (((x) & SDMA_PKT_WRITE_TILED_BC_COUNT_count_mask) << SDMA_PKT_WRITE_TILED_BC_COUNT_count_shift)
+
+/*define for DATA0 word*/
+/*define for data0 field*/
+#define SDMA_PKT_WRITE_TILED_BC_DATA0_data0_offset 9
+#define SDMA_PKT_WRITE_TILED_BC_DATA0_data0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_BC_DATA0_data0_shift  0
+#define SDMA_PKT_WRITE_TILED_BC_DATA0_DATA0(x) (((x) & SDMA_PKT_WRITE_TILED_BC_DATA0_data0_mask) << SDMA_PKT_WRITE_TILED_BC_DATA0_data0_shift)
+
+
+/*
+** Definitions for SDMA_PKT_PTEPDE_COPY packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_PTEPDE_COPY_HEADER_op_offset 0
+#define SDMA_PKT_PTEPDE_COPY_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_COPY_HEADER_op_shift  0
+#define SDMA_PKT_PTEPDE_COPY_HEADER_OP(x) (((x) & SDMA_PKT_PTEPDE_COPY_HEADER_op_mask) << SDMA_PKT_PTEPDE_COPY_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_PTEPDE_COPY_HEADER_sub_op_offset 0
+#define SDMA_PKT_PTEPDE_COPY_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_COPY_HEADER_sub_op_shift  8
+#define SDMA_PKT_PTEPDE_COPY_HEADER_SUB_OP(x) (((x) & SDMA_PKT_PTEPDE_COPY_HEADER_sub_op_mask) << SDMA_PKT_PTEPDE_COPY_HEADER_sub_op_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_PTEPDE_COPY_HEADER_tmz_offset 0
+#define SDMA_PKT_PTEPDE_COPY_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_PTEPDE_COPY_HEADER_tmz_shift  18
+#define SDMA_PKT_PTEPDE_COPY_HEADER_TMZ(x) (((x) & SDMA_PKT_PTEPDE_COPY_HEADER_tmz_mask) << SDMA_PKT_PTEPDE_COPY_HEADER_tmz_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_PTEPDE_COPY_HEADER_cpv_offset 0
+#define SDMA_PKT_PTEPDE_COPY_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_PTEPDE_COPY_HEADER_cpv_shift  28
+#define SDMA_PKT_PTEPDE_COPY_HEADER_CPV(x) (((x) & SDMA_PKT_PTEPDE_COPY_HEADER_cpv_mask) << SDMA_PKT_PTEPDE_COPY_HEADER_cpv_shift)
+
+/*define for ptepde_op field*/
+#define SDMA_PKT_PTEPDE_COPY_HEADER_ptepde_op_offset 0
+#define SDMA_PKT_PTEPDE_COPY_HEADER_ptepde_op_mask   0x00000001
+#define SDMA_PKT_PTEPDE_COPY_HEADER_ptepde_op_shift  31
+#define SDMA_PKT_PTEPDE_COPY_HEADER_PTEPDE_OP(x) (((x) & SDMA_PKT_PTEPDE_COPY_HEADER_ptepde_op_mask) << SDMA_PKT_PTEPDE_COPY_HEADER_ptepde_op_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_PTEPDE_COPY_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_PTEPDE_COPY_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_PTEPDE_COPY_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_PTEPDE_COPY_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_LO_dst_addr_31_0_offset 3
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_PTEPDE_COPY_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_PTEPDE_COPY_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_HI_dst_addr_63_32_offset 4
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_PTEPDE_COPY_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_PTEPDE_COPY_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for MASK_DW0 word*/
+/*define for mask_dw0 field*/
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW0_mask_dw0_offset 5
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW0_mask_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW0_mask_dw0_shift  0
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW0_MASK_DW0(x) (((x) & SDMA_PKT_PTEPDE_COPY_MASK_DW0_mask_dw0_mask) << SDMA_PKT_PTEPDE_COPY_MASK_DW0_mask_dw0_shift)
+
+/*define for MASK_DW1 word*/
+/*define for mask_dw1 field*/
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW1_mask_dw1_offset 6
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW1_mask_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW1_mask_dw1_shift  0
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW1_MASK_DW1(x) (((x) & SDMA_PKT_PTEPDE_COPY_MASK_DW1_mask_dw1_mask) << SDMA_PKT_PTEPDE_COPY_MASK_DW1_mask_dw1_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_PTEPDE_COPY_COUNT_count_offset 7
+#define SDMA_PKT_PTEPDE_COPY_COUNT_count_mask   0x0007FFFF
+#define SDMA_PKT_PTEPDE_COPY_COUNT_count_shift  0
+#define SDMA_PKT_PTEPDE_COPY_COUNT_COUNT(x) (((x) & SDMA_PKT_PTEPDE_COPY_COUNT_count_mask) << SDMA_PKT_PTEPDE_COPY_COUNT_count_shift)
+
+/*define for dst_cache_policy field*/
+#define SDMA_PKT_PTEPDE_COPY_COUNT_dst_cache_policy_offset 7
+#define SDMA_PKT_PTEPDE_COPY_COUNT_dst_cache_policy_mask   0x00000007
+#define SDMA_PKT_PTEPDE_COPY_COUNT_dst_cache_policy_shift  22
+#define SDMA_PKT_PTEPDE_COPY_COUNT_DST_CACHE_POLICY(x) (((x) & SDMA_PKT_PTEPDE_COPY_COUNT_dst_cache_policy_mask) << SDMA_PKT_PTEPDE_COPY_COUNT_dst_cache_policy_shift)
+
+/*define for src_cache_policy field*/
+#define SDMA_PKT_PTEPDE_COPY_COUNT_src_cache_policy_offset 7
+#define SDMA_PKT_PTEPDE_COPY_COUNT_src_cache_policy_mask   0x00000007
+#define SDMA_PKT_PTEPDE_COPY_COUNT_src_cache_policy_shift  29
+#define SDMA_PKT_PTEPDE_COPY_COUNT_SRC_CACHE_POLICY(x) (((x) & SDMA_PKT_PTEPDE_COPY_COUNT_src_cache_policy_mask) << SDMA_PKT_PTEPDE_COPY_COUNT_src_cache_policy_shift)
+
+
+/*
+** Definitions for SDMA_PKT_PTEPDE_COPY_BACKWARDS packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_op_offset 0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_op_shift  0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_OP(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_op_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_sub_op_offset 0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_sub_op_shift  8
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_SUB_OP(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_sub_op_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_sub_op_shift)
+
+/*define for pte_size field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_pte_size_offset 0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_pte_size_mask   0x00000003
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_pte_size_shift  28
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_PTE_SIZE(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_pte_size_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_pte_size_shift)
+
+/*define for direction field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_direction_offset 0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_direction_mask   0x00000001
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_direction_shift  30
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_DIRECTION(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_direction_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_direction_shift)
+
+/*define for ptepde_op field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_ptepde_op_offset 0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_ptepde_op_mask   0x00000001
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_ptepde_op_shift  31
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_PTEPDE_OP(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_ptepde_op_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_ptepde_op_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_LO_dst_addr_31_0_offset 3
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_HI_dst_addr_63_32_offset 4
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for MASK_BIT_FOR_DW word*/
+/*define for mask_first_xfer field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_first_xfer_offset 5
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_first_xfer_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_first_xfer_shift  0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_MASK_FIRST_XFER(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_first_xfer_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_first_xfer_shift)
+
+/*define for mask_last_xfer field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_last_xfer_offset 5
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_last_xfer_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_last_xfer_shift  8
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_MASK_LAST_XFER(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_last_xfer_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_last_xfer_shift)
+
+/*define for COUNT_IN_32B_XFER word*/
+/*define for count field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_COUNT_IN_32B_XFER_count_offset 6
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_COUNT_IN_32B_XFER_count_mask   0x0001FFFF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_COUNT_IN_32B_XFER_count_shift  0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_COUNT_IN_32B_XFER_COUNT(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_COUNT_IN_32B_XFER_count_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_COUNT_IN_32B_XFER_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_PTEPDE_RMW packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_op_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_RMW_HEADER_op_shift  0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_OP(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_op_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_sub_op_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_RMW_HEADER_sub_op_shift  8
+#define SDMA_PKT_PTEPDE_RMW_HEADER_SUB_OP(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_sub_op_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_sub_op_shift)
+
+/*define for mtype field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_mtype_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_mtype_mask   0x00000007
+#define SDMA_PKT_PTEPDE_RMW_HEADER_mtype_shift  16
+#define SDMA_PKT_PTEPDE_RMW_HEADER_MTYPE(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_mtype_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_mtype_shift)
+
+/*define for gcc field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_gcc_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_gcc_mask   0x00000001
+#define SDMA_PKT_PTEPDE_RMW_HEADER_gcc_shift  19
+#define SDMA_PKT_PTEPDE_RMW_HEADER_GCC(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_gcc_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_gcc_shift)
+
+/*define for sys field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_sys_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_sys_mask   0x00000001
+#define SDMA_PKT_PTEPDE_RMW_HEADER_sys_shift  20
+#define SDMA_PKT_PTEPDE_RMW_HEADER_SYS(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_sys_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_sys_shift)
+
+/*define for snp field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_snp_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_snp_mask   0x00000001
+#define SDMA_PKT_PTEPDE_RMW_HEADER_snp_shift  22
+#define SDMA_PKT_PTEPDE_RMW_HEADER_SNP(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_snp_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_snp_shift)
+
+/*define for gpa field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_gpa_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_gpa_mask   0x00000001
+#define SDMA_PKT_PTEPDE_RMW_HEADER_gpa_shift  23
+#define SDMA_PKT_PTEPDE_RMW_HEADER_GPA(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_gpa_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_gpa_shift)
+
+/*define for l2_policy field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_l2_policy_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_l2_policy_mask   0x00000003
+#define SDMA_PKT_PTEPDE_RMW_HEADER_l2_policy_shift  24
+#define SDMA_PKT_PTEPDE_RMW_HEADER_L2_POLICY(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_l2_policy_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_l2_policy_shift)
+
+/*define for llc_policy field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_llc_policy_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_llc_policy_mask   0x00000001
+#define SDMA_PKT_PTEPDE_RMW_HEADER_llc_policy_shift  26
+#define SDMA_PKT_PTEPDE_RMW_HEADER_LLC_POLICY(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_llc_policy_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_llc_policy_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_cpv_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_PTEPDE_RMW_HEADER_cpv_shift  28
+#define SDMA_PKT_PTEPDE_RMW_HEADER_CPV(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_cpv_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_cpv_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_PTEPDE_RMW_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_PTEPDE_RMW_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_RMW_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_PTEPDE_RMW_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_PTEPDE_RMW_ADDR_LO_addr_31_0_mask) << SDMA_PKT_PTEPDE_RMW_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_PTEPDE_RMW_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_PTEPDE_RMW_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_RMW_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_PTEPDE_RMW_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_PTEPDE_RMW_ADDR_HI_addr_63_32_mask) << SDMA_PKT_PTEPDE_RMW_ADDR_HI_addr_63_32_shift)
+
+/*define for MASK_LO word*/
+/*define for mask_31_0 field*/
+#define SDMA_PKT_PTEPDE_RMW_MASK_LO_mask_31_0_offset 3
+#define SDMA_PKT_PTEPDE_RMW_MASK_LO_mask_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_RMW_MASK_LO_mask_31_0_shift  0
+#define SDMA_PKT_PTEPDE_RMW_MASK_LO_MASK_31_0(x) (((x) & SDMA_PKT_PTEPDE_RMW_MASK_LO_mask_31_0_mask) << SDMA_PKT_PTEPDE_RMW_MASK_LO_mask_31_0_shift)
+
+/*define for MASK_HI word*/
+/*define for mask_63_32 field*/
+#define SDMA_PKT_PTEPDE_RMW_MASK_HI_mask_63_32_offset 4
+#define SDMA_PKT_PTEPDE_RMW_MASK_HI_mask_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_RMW_MASK_HI_mask_63_32_shift  0
+#define SDMA_PKT_PTEPDE_RMW_MASK_HI_MASK_63_32(x) (((x) & SDMA_PKT_PTEPDE_RMW_MASK_HI_mask_63_32_mask) << SDMA_PKT_PTEPDE_RMW_MASK_HI_mask_63_32_shift)
+
+/*define for VALUE_LO word*/
+/*define for value_31_0 field*/
+#define SDMA_PKT_PTEPDE_RMW_VALUE_LO_value_31_0_offset 5
+#define SDMA_PKT_PTEPDE_RMW_VALUE_LO_value_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_RMW_VALUE_LO_value_31_0_shift  0
+#define SDMA_PKT_PTEPDE_RMW_VALUE_LO_VALUE_31_0(x) (((x) & SDMA_PKT_PTEPDE_RMW_VALUE_LO_value_31_0_mask) << SDMA_PKT_PTEPDE_RMW_VALUE_LO_value_31_0_shift)
+
+/*define for VALUE_HI word*/
+/*define for value_63_32 field*/
+#define SDMA_PKT_PTEPDE_RMW_VALUE_HI_value_63_32_offset 6
+#define SDMA_PKT_PTEPDE_RMW_VALUE_HI_value_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_RMW_VALUE_HI_value_63_32_shift  0
+#define SDMA_PKT_PTEPDE_RMW_VALUE_HI_VALUE_63_32(x) (((x) & SDMA_PKT_PTEPDE_RMW_VALUE_HI_value_63_32_mask) << SDMA_PKT_PTEPDE_RMW_VALUE_HI_value_63_32_shift)
+
+/*define for COUNT word*/
+/*define for num_of_pte field*/
+#define SDMA_PKT_PTEPDE_RMW_COUNT_num_of_pte_offset 7
+#define SDMA_PKT_PTEPDE_RMW_COUNT_num_of_pte_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_RMW_COUNT_num_of_pte_shift  0
+#define SDMA_PKT_PTEPDE_RMW_COUNT_NUM_OF_PTE(x) (((x) & SDMA_PKT_PTEPDE_RMW_COUNT_num_of_pte_mask) << SDMA_PKT_PTEPDE_RMW_COUNT_num_of_pte_shift)
+
+
+/*
+** Definitions for SDMA_PKT_REGISTER_RMW packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_REGISTER_RMW_HEADER_op_offset 0
+#define SDMA_PKT_REGISTER_RMW_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_REGISTER_RMW_HEADER_op_shift  0
+#define SDMA_PKT_REGISTER_RMW_HEADER_OP(x) (((x) & SDMA_PKT_REGISTER_RMW_HEADER_op_mask) << SDMA_PKT_REGISTER_RMW_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_REGISTER_RMW_HEADER_sub_op_offset 0
+#define SDMA_PKT_REGISTER_RMW_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_REGISTER_RMW_HEADER_sub_op_shift  8
+#define SDMA_PKT_REGISTER_RMW_HEADER_SUB_OP(x) (((x) & SDMA_PKT_REGISTER_RMW_HEADER_sub_op_mask) << SDMA_PKT_REGISTER_RMW_HEADER_sub_op_shift)
+
+/*define for ADDR word*/
+/*define for addr field*/
+#define SDMA_PKT_REGISTER_RMW_ADDR_addr_offset 1
+#define SDMA_PKT_REGISTER_RMW_ADDR_addr_mask   0x000FFFFF
+#define SDMA_PKT_REGISTER_RMW_ADDR_addr_shift  0
+#define SDMA_PKT_REGISTER_RMW_ADDR_ADDR(x) (((x) & SDMA_PKT_REGISTER_RMW_ADDR_addr_mask) << SDMA_PKT_REGISTER_RMW_ADDR_addr_shift)
+
+/*define for aperture_id field*/
+#define SDMA_PKT_REGISTER_RMW_ADDR_aperture_id_offset 1
+#define SDMA_PKT_REGISTER_RMW_ADDR_aperture_id_mask   0x00000FFF
+#define SDMA_PKT_REGISTER_RMW_ADDR_aperture_id_shift  20
+#define SDMA_PKT_REGISTER_RMW_ADDR_APERTURE_ID(x) (((x) & SDMA_PKT_REGISTER_RMW_ADDR_aperture_id_mask) << SDMA_PKT_REGISTER_RMW_ADDR_aperture_id_shift)
+
+/*define for MASK word*/
+/*define for mask field*/
+#define SDMA_PKT_REGISTER_RMW_MASK_mask_offset 2
+#define SDMA_PKT_REGISTER_RMW_MASK_mask_mask   0xFFFFFFFF
+#define SDMA_PKT_REGISTER_RMW_MASK_mask_shift  0
+#define SDMA_PKT_REGISTER_RMW_MASK_MASK(x) (((x) & SDMA_PKT_REGISTER_RMW_MASK_mask_mask) << SDMA_PKT_REGISTER_RMW_MASK_mask_shift)
+
+/*define for VALUE word*/
+/*define for value field*/
+#define SDMA_PKT_REGISTER_RMW_VALUE_value_offset 3
+#define SDMA_PKT_REGISTER_RMW_VALUE_value_mask   0xFFFFFFFF
+#define SDMA_PKT_REGISTER_RMW_VALUE_value_shift  0
+#define SDMA_PKT_REGISTER_RMW_VALUE_VALUE(x) (((x) & SDMA_PKT_REGISTER_RMW_VALUE_value_mask) << SDMA_PKT_REGISTER_RMW_VALUE_value_shift)
+
+/*define for MISC word*/
+/*define for stride field*/
+#define SDMA_PKT_REGISTER_RMW_MISC_stride_offset 4
+#define SDMA_PKT_REGISTER_RMW_MISC_stride_mask   0x000FFFFF
+#define SDMA_PKT_REGISTER_RMW_MISC_stride_shift  0
+#define SDMA_PKT_REGISTER_RMW_MISC_STRIDE(x) (((x) & SDMA_PKT_REGISTER_RMW_MISC_stride_mask) << SDMA_PKT_REGISTER_RMW_MISC_stride_shift)
+
+/*define for num_of_reg field*/
+#define SDMA_PKT_REGISTER_RMW_MISC_num_of_reg_offset 4
+#define SDMA_PKT_REGISTER_RMW_MISC_num_of_reg_mask   0x00000FFF
+#define SDMA_PKT_REGISTER_RMW_MISC_num_of_reg_shift  20
+#define SDMA_PKT_REGISTER_RMW_MISC_NUM_OF_REG(x) (((x) & SDMA_PKT_REGISTER_RMW_MISC_num_of_reg_mask) << SDMA_PKT_REGISTER_RMW_MISC_num_of_reg_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_INCR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_INCR_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_INCR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_INCR_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_INCR_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_INCR_HEADER_op_mask) << SDMA_PKT_WRITE_INCR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_INCR_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_INCR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_INCR_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_INCR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_INCR_HEADER_sub_op_mask) << SDMA_PKT_WRITE_INCR_HEADER_sub_op_shift)
+
+/*define for cache_policy field*/
+#define SDMA_PKT_WRITE_INCR_HEADER_cache_policy_offset 0
+#define SDMA_PKT_WRITE_INCR_HEADER_cache_policy_mask   0x00000007
+#define SDMA_PKT_WRITE_INCR_HEADER_cache_policy_shift  24
+#define SDMA_PKT_WRITE_INCR_HEADER_CACHE_POLICY(x) (((x) & SDMA_PKT_WRITE_INCR_HEADER_cache_policy_mask) << SDMA_PKT_WRITE_INCR_HEADER_cache_policy_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_WRITE_INCR_HEADER_cpv_offset 0
+#define SDMA_PKT_WRITE_INCR_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_WRITE_INCR_HEADER_cpv_shift  28
+#define SDMA_PKT_WRITE_INCR_HEADER_CPV(x) (((x) & SDMA_PKT_WRITE_INCR_HEADER_cpv_mask) << SDMA_PKT_WRITE_INCR_HEADER_cpv_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for MASK_DW0 word*/
+/*define for mask_dw0 field*/
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_offset 3
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_shift  0
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_MASK_DW0(x) (((x) & SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_mask) << SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_shift)
+
+/*define for MASK_DW1 word*/
+/*define for mask_dw1 field*/
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_offset 4
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_shift  0
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_MASK_DW1(x) (((x) & SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_mask) << SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_shift)
+
+/*define for INIT_DW0 word*/
+/*define for init_dw0 field*/
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_offset 5
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_shift  0
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_INIT_DW0(x) (((x) & SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_mask) << SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_shift)
+
+/*define for INIT_DW1 word*/
+/*define for init_dw1 field*/
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_offset 6
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_shift  0
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_INIT_DW1(x) (((x) & SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_mask) << SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_shift)
+
+/*define for INCR_DW0 word*/
+/*define for incr_dw0 field*/
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_offset 7
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_shift  0
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_INCR_DW0(x) (((x) & SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_mask) << SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_shift)
+
+/*define for INCR_DW1 word*/
+/*define for incr_dw1 field*/
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_offset 8
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_shift  0
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_INCR_DW1(x) (((x) & SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_mask) << SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_INCR_COUNT_count_offset 9
+#define SDMA_PKT_WRITE_INCR_COUNT_count_mask   0x0007FFFF
+#define SDMA_PKT_WRITE_INCR_COUNT_count_shift  0
+#define SDMA_PKT_WRITE_INCR_COUNT_COUNT(x) (((x) & SDMA_PKT_WRITE_INCR_COUNT_count_mask) << SDMA_PKT_WRITE_INCR_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_INDIRECT packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_INDIRECT_HEADER_op_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_INDIRECT_HEADER_op_shift  0
+#define SDMA_PKT_INDIRECT_HEADER_OP(x) (((x) & SDMA_PKT_INDIRECT_HEADER_op_mask) << SDMA_PKT_INDIRECT_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_INDIRECT_HEADER_sub_op_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_INDIRECT_HEADER_sub_op_shift  8
+#define SDMA_PKT_INDIRECT_HEADER_SUB_OP(x) (((x) & SDMA_PKT_INDIRECT_HEADER_sub_op_mask) << SDMA_PKT_INDIRECT_HEADER_sub_op_shift)
+
+/*define for vmid field*/
+#define SDMA_PKT_INDIRECT_HEADER_vmid_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_vmid_mask   0x0000000F
+#define SDMA_PKT_INDIRECT_HEADER_vmid_shift  16
+#define SDMA_PKT_INDIRECT_HEADER_VMID(x) (((x) & SDMA_PKT_INDIRECT_HEADER_vmid_mask) << SDMA_PKT_INDIRECT_HEADER_vmid_shift)
+
+/*define for priv field*/
+#define SDMA_PKT_INDIRECT_HEADER_priv_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_priv_mask   0x00000001
+#define SDMA_PKT_INDIRECT_HEADER_priv_shift  31
+#define SDMA_PKT_INDIRECT_HEADER_PRIV(x) (((x) & SDMA_PKT_INDIRECT_HEADER_priv_mask) << SDMA_PKT_INDIRECT_HEADER_priv_shift)
+
+/*define for BASE_LO word*/
+/*define for ib_base_31_0 field*/
+#define SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_offset 1
+#define SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_shift  0
+#define SDMA_PKT_INDIRECT_BASE_LO_IB_BASE_31_0(x) (((x) & SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_mask) << SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_shift)
+
+/*define for BASE_HI word*/
+/*define for ib_base_63_32 field*/
+#define SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_offset 2
+#define SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_shift  0
+#define SDMA_PKT_INDIRECT_BASE_HI_IB_BASE_63_32(x) (((x) & SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_mask) << SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_shift)
+
+/*define for IB_SIZE word*/
+/*define for ib_size field*/
+#define SDMA_PKT_INDIRECT_IB_SIZE_ib_size_offset 3
+#define SDMA_PKT_INDIRECT_IB_SIZE_ib_size_mask   0x000FFFFF
+#define SDMA_PKT_INDIRECT_IB_SIZE_ib_size_shift  0
+#define SDMA_PKT_INDIRECT_IB_SIZE_IB_SIZE(x) (((x) & SDMA_PKT_INDIRECT_IB_SIZE_ib_size_mask) << SDMA_PKT_INDIRECT_IB_SIZE_ib_size_shift)
+
+/*define for CSA_ADDR_LO word*/
+/*define for csa_addr_31_0 field*/
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_offset 4
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_shift  0
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_CSA_ADDR_31_0(x) (((x) & SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_mask) << SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_shift)
+
+/*define for CSA_ADDR_HI word*/
+/*define for csa_addr_63_32 field*/
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_offset 5
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_shift  0
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_CSA_ADDR_63_32(x) (((x) & SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_mask) << SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_SEMAPHORE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_op_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_SEMAPHORE_HEADER_op_shift  0
+#define SDMA_PKT_SEMAPHORE_HEADER_OP(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_op_mask) << SDMA_PKT_SEMAPHORE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_sub_op_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_SEMAPHORE_HEADER_sub_op_shift  8
+#define SDMA_PKT_SEMAPHORE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_sub_op_mask) << SDMA_PKT_SEMAPHORE_HEADER_sub_op_shift)
+
+/*define for write_one field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_write_one_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_write_one_mask   0x00000001
+#define SDMA_PKT_SEMAPHORE_HEADER_write_one_shift  29
+#define SDMA_PKT_SEMAPHORE_HEADER_WRITE_ONE(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_write_one_mask) << SDMA_PKT_SEMAPHORE_HEADER_write_one_shift)
+
+/*define for signal field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_signal_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_signal_mask   0x00000001
+#define SDMA_PKT_SEMAPHORE_HEADER_signal_shift  30
+#define SDMA_PKT_SEMAPHORE_HEADER_SIGNAL(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_signal_mask) << SDMA_PKT_SEMAPHORE_HEADER_signal_shift)
+
+/*define for mailbox field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_mailbox_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_mailbox_mask   0x00000001
+#define SDMA_PKT_SEMAPHORE_HEADER_mailbox_shift  31
+#define SDMA_PKT_SEMAPHORE_HEADER_MAILBOX(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_mailbox_mask) << SDMA_PKT_SEMAPHORE_HEADER_mailbox_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_mask) << SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_mask) << SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_MEM_INCR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_MEM_INCR_HEADER_op_offset 0
+#define SDMA_PKT_MEM_INCR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_MEM_INCR_HEADER_op_shift  0
+#define SDMA_PKT_MEM_INCR_HEADER_OP(x) (((x) & SDMA_PKT_MEM_INCR_HEADER_op_mask) << SDMA_PKT_MEM_INCR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_MEM_INCR_HEADER_sub_op_offset 0
+#define SDMA_PKT_MEM_INCR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_MEM_INCR_HEADER_sub_op_shift  8
+#define SDMA_PKT_MEM_INCR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_MEM_INCR_HEADER_sub_op_mask) << SDMA_PKT_MEM_INCR_HEADER_sub_op_shift)
+
+/*define for l2_policy field*/
+#define SDMA_PKT_MEM_INCR_HEADER_l2_policy_offset 0
+#define SDMA_PKT_MEM_INCR_HEADER_l2_policy_mask   0x00000003
+#define SDMA_PKT_MEM_INCR_HEADER_l2_policy_shift  24
+#define SDMA_PKT_MEM_INCR_HEADER_L2_POLICY(x) (((x) & SDMA_PKT_MEM_INCR_HEADER_l2_policy_mask) << SDMA_PKT_MEM_INCR_HEADER_l2_policy_shift)
+
+/*define for llc_policy field*/
+#define SDMA_PKT_MEM_INCR_HEADER_llc_policy_offset 0
+#define SDMA_PKT_MEM_INCR_HEADER_llc_policy_mask   0x00000001
+#define SDMA_PKT_MEM_INCR_HEADER_llc_policy_shift  26
+#define SDMA_PKT_MEM_INCR_HEADER_LLC_POLICY(x) (((x) & SDMA_PKT_MEM_INCR_HEADER_llc_policy_mask) << SDMA_PKT_MEM_INCR_HEADER_llc_policy_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_MEM_INCR_HEADER_cpv_offset 0
+#define SDMA_PKT_MEM_INCR_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_MEM_INCR_HEADER_cpv_shift  28
+#define SDMA_PKT_MEM_INCR_HEADER_CPV(x) (((x) & SDMA_PKT_MEM_INCR_HEADER_cpv_mask) << SDMA_PKT_MEM_INCR_HEADER_cpv_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_MEM_INCR_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_MEM_INCR_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_MEM_INCR_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_MEM_INCR_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_MEM_INCR_ADDR_LO_addr_31_0_mask) << SDMA_PKT_MEM_INCR_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_MEM_INCR_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_MEM_INCR_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_MEM_INCR_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_MEM_INCR_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_MEM_INCR_ADDR_HI_addr_63_32_mask) << SDMA_PKT_MEM_INCR_ADDR_HI_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_VM_INVALIDATION packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_VM_INVALIDATION_HEADER_op_offset 0
+#define SDMA_PKT_VM_INVALIDATION_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_VM_INVALIDATION_HEADER_op_shift  0
+#define SDMA_PKT_VM_INVALIDATION_HEADER_OP(x) (((x) & SDMA_PKT_VM_INVALIDATION_HEADER_op_mask) << SDMA_PKT_VM_INVALIDATION_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_VM_INVALIDATION_HEADER_sub_op_offset 0
+#define SDMA_PKT_VM_INVALIDATION_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_VM_INVALIDATION_HEADER_sub_op_shift  8
+#define SDMA_PKT_VM_INVALIDATION_HEADER_SUB_OP(x) (((x) & SDMA_PKT_VM_INVALIDATION_HEADER_sub_op_mask) << SDMA_PKT_VM_INVALIDATION_HEADER_sub_op_shift)
+
+/*define for gfx_eng_id field*/
+#define SDMA_PKT_VM_INVALIDATION_HEADER_gfx_eng_id_offset 0
+#define SDMA_PKT_VM_INVALIDATION_HEADER_gfx_eng_id_mask   0x0000001F
+#define SDMA_PKT_VM_INVALIDATION_HEADER_gfx_eng_id_shift  16
+#define SDMA_PKT_VM_INVALIDATION_HEADER_GFX_ENG_ID(x) (((x) & SDMA_PKT_VM_INVALIDATION_HEADER_gfx_eng_id_mask) << SDMA_PKT_VM_INVALIDATION_HEADER_gfx_eng_id_shift)
+
+/*define for mm_eng_id field*/
+#define SDMA_PKT_VM_INVALIDATION_HEADER_mm_eng_id_offset 0
+#define SDMA_PKT_VM_INVALIDATION_HEADER_mm_eng_id_mask   0x0000001F
+#define SDMA_PKT_VM_INVALIDATION_HEADER_mm_eng_id_shift  24
+#define SDMA_PKT_VM_INVALIDATION_HEADER_MM_ENG_ID(x) (((x) & SDMA_PKT_VM_INVALIDATION_HEADER_mm_eng_id_mask) << SDMA_PKT_VM_INVALIDATION_HEADER_mm_eng_id_shift)
+
+/*define for INVALIDATEREQ word*/
+/*define for invalidatereq field*/
+#define SDMA_PKT_VM_INVALIDATION_INVALIDATEREQ_invalidatereq_offset 1
+#define SDMA_PKT_VM_INVALIDATION_INVALIDATEREQ_invalidatereq_mask   0xFFFFFFFF
+#define SDMA_PKT_VM_INVALIDATION_INVALIDATEREQ_invalidatereq_shift  0
+#define SDMA_PKT_VM_INVALIDATION_INVALIDATEREQ_INVALIDATEREQ(x) (((x) & SDMA_PKT_VM_INVALIDATION_INVALIDATEREQ_invalidatereq_mask) << SDMA_PKT_VM_INVALIDATION_INVALIDATEREQ_invalidatereq_shift)
+
+/*define for ADDRESSRANGELO word*/
+/*define for addressrangelo field*/
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGELO_addressrangelo_offset 2
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGELO_addressrangelo_mask   0xFFFFFFFF
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGELO_addressrangelo_shift  0
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGELO_ADDRESSRANGELO(x) (((x) & SDMA_PKT_VM_INVALIDATION_ADDRESSRANGELO_addressrangelo_mask) << SDMA_PKT_VM_INVALIDATION_ADDRESSRANGELO_addressrangelo_shift)
+
+/*define for ADDRESSRANGEHI word*/
+/*define for invalidateack field*/
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_invalidateack_offset 3
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_invalidateack_mask   0x0000FFFF
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_invalidateack_shift  0
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_INVALIDATEACK(x) (((x) & SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_invalidateack_mask) << SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_invalidateack_shift)
+
+/*define for addressrangehi field*/
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_addressrangehi_offset 3
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_addressrangehi_mask   0x0000001F
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_addressrangehi_shift  16
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_ADDRESSRANGEHI(x) (((x) & SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_addressrangehi_mask) << SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_addressrangehi_shift)
+
+/*define for reserved field*/
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_reserved_offset 3
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_reserved_mask   0x000001FF
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_reserved_shift  23
+#define SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_RESERVED(x) (((x) & SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_reserved_mask) << SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_reserved_shift)
+
+
+/*
+** Definitions for SDMA_PKT_FENCE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_FENCE_HEADER_op_offset 0
+#define SDMA_PKT_FENCE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_FENCE_HEADER_op_shift  0
+#define SDMA_PKT_FENCE_HEADER_OP(x) (((x) & SDMA_PKT_FENCE_HEADER_op_mask) << SDMA_PKT_FENCE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_FENCE_HEADER_sub_op_offset 0
+#define SDMA_PKT_FENCE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_FENCE_HEADER_sub_op_shift  8
+#define SDMA_PKT_FENCE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_FENCE_HEADER_sub_op_mask) << SDMA_PKT_FENCE_HEADER_sub_op_shift)
+
+/*define for mtype field*/
+#define SDMA_PKT_FENCE_HEADER_mtype_offset 0
+#define SDMA_PKT_FENCE_HEADER_mtype_mask   0x00000007
+#define SDMA_PKT_FENCE_HEADER_mtype_shift  16
+#define SDMA_PKT_FENCE_HEADER_MTYPE(x) (((x) & SDMA_PKT_FENCE_HEADER_mtype_mask) << SDMA_PKT_FENCE_HEADER_mtype_shift)
+
+/*define for gcc field*/
+#define SDMA_PKT_FENCE_HEADER_gcc_offset 0
+#define SDMA_PKT_FENCE_HEADER_gcc_mask   0x00000001
+#define SDMA_PKT_FENCE_HEADER_gcc_shift  19
+#define SDMA_PKT_FENCE_HEADER_GCC(x) (((x) & SDMA_PKT_FENCE_HEADER_gcc_mask) << SDMA_PKT_FENCE_HEADER_gcc_shift)
+
+/*define for sys field*/
+#define SDMA_PKT_FENCE_HEADER_sys_offset 0
+#define SDMA_PKT_FENCE_HEADER_sys_mask   0x00000001
+#define SDMA_PKT_FENCE_HEADER_sys_shift  20
+#define SDMA_PKT_FENCE_HEADER_SYS(x) (((x) & SDMA_PKT_FENCE_HEADER_sys_mask) << SDMA_PKT_FENCE_HEADER_sys_shift)
+
+/*define for snp field*/
+#define SDMA_PKT_FENCE_HEADER_snp_offset 0
+#define SDMA_PKT_FENCE_HEADER_snp_mask   0x00000001
+#define SDMA_PKT_FENCE_HEADER_snp_shift  22
+#define SDMA_PKT_FENCE_HEADER_SNP(x) (((x) & SDMA_PKT_FENCE_HEADER_snp_mask) << SDMA_PKT_FENCE_HEADER_snp_shift)
+
+/*define for gpa field*/
+#define SDMA_PKT_FENCE_HEADER_gpa_offset 0
+#define SDMA_PKT_FENCE_HEADER_gpa_mask   0x00000001
+#define SDMA_PKT_FENCE_HEADER_gpa_shift  23
+#define SDMA_PKT_FENCE_HEADER_GPA(x) (((x) & SDMA_PKT_FENCE_HEADER_gpa_mask) << SDMA_PKT_FENCE_HEADER_gpa_shift)
+
+/*define for l2_policy field*/
+#define SDMA_PKT_FENCE_HEADER_l2_policy_offset 0
+#define SDMA_PKT_FENCE_HEADER_l2_policy_mask   0x00000003
+#define SDMA_PKT_FENCE_HEADER_l2_policy_shift  24
+#define SDMA_PKT_FENCE_HEADER_L2_POLICY(x) (((x) & SDMA_PKT_FENCE_HEADER_l2_policy_mask) << SDMA_PKT_FENCE_HEADER_l2_policy_shift)
+
+/*define for llc_policy field*/
+#define SDMA_PKT_FENCE_HEADER_llc_policy_offset 0
+#define SDMA_PKT_FENCE_HEADER_llc_policy_mask   0x00000001
+#define SDMA_PKT_FENCE_HEADER_llc_policy_shift  26
+#define SDMA_PKT_FENCE_HEADER_LLC_POLICY(x) (((x) & SDMA_PKT_FENCE_HEADER_llc_policy_mask) << SDMA_PKT_FENCE_HEADER_llc_policy_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_FENCE_HEADER_cpv_offset 0
+#define SDMA_PKT_FENCE_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_FENCE_HEADER_cpv_shift  28
+#define SDMA_PKT_FENCE_HEADER_CPV(x) (((x) & SDMA_PKT_FENCE_HEADER_cpv_mask) << SDMA_PKT_FENCE_HEADER_cpv_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_FENCE_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_FENCE_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_FENCE_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_FENCE_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_FENCE_ADDR_LO_addr_31_0_mask) << SDMA_PKT_FENCE_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_FENCE_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_FENCE_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_FENCE_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_FENCE_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_FENCE_ADDR_HI_addr_63_32_mask) << SDMA_PKT_FENCE_ADDR_HI_addr_63_32_shift)
+
+/*define for DATA word*/
+/*define for data field*/
+#define SDMA_PKT_FENCE_DATA_data_offset 3
+#define SDMA_PKT_FENCE_DATA_data_mask   0xFFFFFFFF
+#define SDMA_PKT_FENCE_DATA_data_shift  0
+#define SDMA_PKT_FENCE_DATA_DATA(x) (((x) & SDMA_PKT_FENCE_DATA_data_mask) << SDMA_PKT_FENCE_DATA_data_shift)
+
+
+/*
+** Definitions for SDMA_PKT_SRBM_WRITE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_SRBM_WRITE_HEADER_op_offset 0
+#define SDMA_PKT_SRBM_WRITE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_SRBM_WRITE_HEADER_op_shift  0
+#define SDMA_PKT_SRBM_WRITE_HEADER_OP(x) (((x) & SDMA_PKT_SRBM_WRITE_HEADER_op_mask) << SDMA_PKT_SRBM_WRITE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_SRBM_WRITE_HEADER_sub_op_offset 0
+#define SDMA_PKT_SRBM_WRITE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_SRBM_WRITE_HEADER_sub_op_shift  8
+#define SDMA_PKT_SRBM_WRITE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_SRBM_WRITE_HEADER_sub_op_mask) << SDMA_PKT_SRBM_WRITE_HEADER_sub_op_shift)
+
+/*define for byte_en field*/
+#define SDMA_PKT_SRBM_WRITE_HEADER_byte_en_offset 0
+#define SDMA_PKT_SRBM_WRITE_HEADER_byte_en_mask   0x0000000F
+#define SDMA_PKT_SRBM_WRITE_HEADER_byte_en_shift  28
+#define SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(x) (((x) & SDMA_PKT_SRBM_WRITE_HEADER_byte_en_mask) << SDMA_PKT_SRBM_WRITE_HEADER_byte_en_shift)
+
+/*define for ADDR word*/
+/*define for addr field*/
+#define SDMA_PKT_SRBM_WRITE_ADDR_addr_offset 1
+#define SDMA_PKT_SRBM_WRITE_ADDR_addr_mask   0x0003FFFF
+#define SDMA_PKT_SRBM_WRITE_ADDR_addr_shift  0
+#define SDMA_PKT_SRBM_WRITE_ADDR_ADDR(x) (((x) & SDMA_PKT_SRBM_WRITE_ADDR_addr_mask) << SDMA_PKT_SRBM_WRITE_ADDR_addr_shift)
+
+/*define for apertureid field*/
+#define SDMA_PKT_SRBM_WRITE_ADDR_apertureid_offset 1
+#define SDMA_PKT_SRBM_WRITE_ADDR_apertureid_mask   0x00000FFF
+#define SDMA_PKT_SRBM_WRITE_ADDR_apertureid_shift  20
+#define SDMA_PKT_SRBM_WRITE_ADDR_APERTUREID(x) (((x) & SDMA_PKT_SRBM_WRITE_ADDR_apertureid_mask) << SDMA_PKT_SRBM_WRITE_ADDR_apertureid_shift)
+
+/*define for DATA word*/
+/*define for data field*/
+#define SDMA_PKT_SRBM_WRITE_DATA_data_offset 2
+#define SDMA_PKT_SRBM_WRITE_DATA_data_mask   0xFFFFFFFF
+#define SDMA_PKT_SRBM_WRITE_DATA_data_shift  0
+#define SDMA_PKT_SRBM_WRITE_DATA_DATA(x) (((x) & SDMA_PKT_SRBM_WRITE_DATA_data_mask) << SDMA_PKT_SRBM_WRITE_DATA_data_shift)
+
+
+/*
+** Definitions for SDMA_PKT_PRE_EXE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_PRE_EXE_HEADER_op_offset 0
+#define SDMA_PKT_PRE_EXE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_PRE_EXE_HEADER_op_shift  0
+#define SDMA_PKT_PRE_EXE_HEADER_OP(x) (((x) & SDMA_PKT_PRE_EXE_HEADER_op_mask) << SDMA_PKT_PRE_EXE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_PRE_EXE_HEADER_sub_op_offset 0
+#define SDMA_PKT_PRE_EXE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_PRE_EXE_HEADER_sub_op_shift  8
+#define SDMA_PKT_PRE_EXE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_PRE_EXE_HEADER_sub_op_mask) << SDMA_PKT_PRE_EXE_HEADER_sub_op_shift)
+
+/*define for dev_sel field*/
+#define SDMA_PKT_PRE_EXE_HEADER_dev_sel_offset 0
+#define SDMA_PKT_PRE_EXE_HEADER_dev_sel_mask   0x000000FF
+#define SDMA_PKT_PRE_EXE_HEADER_dev_sel_shift  16
+#define SDMA_PKT_PRE_EXE_HEADER_DEV_SEL(x) (((x) & SDMA_PKT_PRE_EXE_HEADER_dev_sel_mask) << SDMA_PKT_PRE_EXE_HEADER_dev_sel_shift)
+
+/*define for EXEC_COUNT word*/
+/*define for exec_count field*/
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_offset 1
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_mask   0x00003FFF
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_shift  0
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_EXEC_COUNT(x) (((x) & SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_mask) << SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COND_EXE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COND_EXE_HEADER_op_offset 0
+#define SDMA_PKT_COND_EXE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COND_EXE_HEADER_op_shift  0
+#define SDMA_PKT_COND_EXE_HEADER_OP(x) (((x) & SDMA_PKT_COND_EXE_HEADER_op_mask) << SDMA_PKT_COND_EXE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COND_EXE_HEADER_sub_op_offset 0
+#define SDMA_PKT_COND_EXE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COND_EXE_HEADER_sub_op_shift  8
+#define SDMA_PKT_COND_EXE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COND_EXE_HEADER_sub_op_mask) << SDMA_PKT_COND_EXE_HEADER_sub_op_shift)
+
+/*define for cache_policy field*/
+#define SDMA_PKT_COND_EXE_HEADER_cache_policy_offset 0
+#define SDMA_PKT_COND_EXE_HEADER_cache_policy_mask   0x00000007
+#define SDMA_PKT_COND_EXE_HEADER_cache_policy_shift  24
+#define SDMA_PKT_COND_EXE_HEADER_CACHE_POLICY(x) (((x) & SDMA_PKT_COND_EXE_HEADER_cache_policy_mask) << SDMA_PKT_COND_EXE_HEADER_cache_policy_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_COND_EXE_HEADER_cpv_offset 0
+#define SDMA_PKT_COND_EXE_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_COND_EXE_HEADER_cpv_shift  28
+#define SDMA_PKT_COND_EXE_HEADER_CPV(x) (((x) & SDMA_PKT_COND_EXE_HEADER_cpv_mask) << SDMA_PKT_COND_EXE_HEADER_cpv_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_COND_EXE_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_mask) << SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_COND_EXE_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_mask) << SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_shift)
+
+/*define for REFERENCE word*/
+/*define for reference field*/
+#define SDMA_PKT_COND_EXE_REFERENCE_reference_offset 3
+#define SDMA_PKT_COND_EXE_REFERENCE_reference_mask   0xFFFFFFFF
+#define SDMA_PKT_COND_EXE_REFERENCE_reference_shift  0
+#define SDMA_PKT_COND_EXE_REFERENCE_REFERENCE(x) (((x) & SDMA_PKT_COND_EXE_REFERENCE_reference_mask) << SDMA_PKT_COND_EXE_REFERENCE_reference_shift)
+
+/*define for EXEC_COUNT word*/
+/*define for exec_count field*/
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_offset 4
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_mask   0x00003FFF
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_shift  0
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_EXEC_COUNT(x) (((x) & SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_mask) << SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_CONSTANT_FILL packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_op_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_CONSTANT_FILL_HEADER_op_shift  0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_OP(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_op_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_shift  8
+#define SDMA_PKT_CONSTANT_FILL_HEADER_SUB_OP(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sw_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sw_mask   0x00000003
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sw_shift  16
+#define SDMA_PKT_CONSTANT_FILL_HEADER_SW(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_sw_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_sw_shift)
+
+/*define for cache_policy field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_cache_policy_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_cache_policy_mask   0x00000007
+#define SDMA_PKT_CONSTANT_FILL_HEADER_cache_policy_shift  24
+#define SDMA_PKT_CONSTANT_FILL_HEADER_CACHE_POLICY(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_cache_policy_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_cache_policy_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_cpv_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_CONSTANT_FILL_HEADER_cpv_shift  28
+#define SDMA_PKT_CONSTANT_FILL_HEADER_CPV(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_cpv_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_cpv_shift)
+
+/*define for fillsize field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_mask   0x00000003
+#define SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_shift  30
+#define SDMA_PKT_CONSTANT_FILL_HEADER_FILLSIZE(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DATA word*/
+/*define for src_data_31_0 field*/
+#define SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_offset 3
+#define SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_shift  0
+#define SDMA_PKT_CONSTANT_FILL_DATA_SRC_DATA_31_0(x) (((x) & SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_mask) << SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_CONSTANT_FILL_COUNT_count_offset 4
+#define SDMA_PKT_CONSTANT_FILL_COUNT_count_mask   0x3FFFFFFF
+#define SDMA_PKT_CONSTANT_FILL_COUNT_count_shift  0
+#define SDMA_PKT_CONSTANT_FILL_COUNT_COUNT(x) (((x) & SDMA_PKT_CONSTANT_FILL_COUNT_count_mask) << SDMA_PKT_CONSTANT_FILL_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_DATA_FILL_MULTI packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_op_offset 0
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_op_shift  0
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_OP(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_HEADER_op_mask) << SDMA_PKT_DATA_FILL_MULTI_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_sub_op_offset 0
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_sub_op_shift  8
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_SUB_OP(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_HEADER_sub_op_mask) << SDMA_PKT_DATA_FILL_MULTI_HEADER_sub_op_shift)
+
+/*define for cache_policy field*/
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_cache_policy_offset 0
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_cache_policy_mask   0x00000007
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_cache_policy_shift  24
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_CACHE_POLICY(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_HEADER_cache_policy_mask) << SDMA_PKT_DATA_FILL_MULTI_HEADER_cache_policy_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_cpv_offset 0
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_cpv_shift  28
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_CPV(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_HEADER_cpv_mask) << SDMA_PKT_DATA_FILL_MULTI_HEADER_cpv_shift)
+
+/*define for memlog_clr field*/
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_memlog_clr_offset 0
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_memlog_clr_mask   0x00000001
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_memlog_clr_shift  31
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_MEMLOG_CLR(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_HEADER_memlog_clr_mask) << SDMA_PKT_DATA_FILL_MULTI_HEADER_memlog_clr_shift)
+
+/*define for BYTE_STRIDE word*/
+/*define for byte_stride field*/
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_STRIDE_byte_stride_offset 1
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_STRIDE_byte_stride_mask   0xFFFFFFFF
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_STRIDE_byte_stride_shift  0
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_STRIDE_BYTE_STRIDE(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_BYTE_STRIDE_byte_stride_mask) << SDMA_PKT_DATA_FILL_MULTI_BYTE_STRIDE_byte_stride_shift)
+
+/*define for DMA_COUNT word*/
+/*define for dma_count field*/
+#define SDMA_PKT_DATA_FILL_MULTI_DMA_COUNT_dma_count_offset 2
+#define SDMA_PKT_DATA_FILL_MULTI_DMA_COUNT_dma_count_mask   0xFFFFFFFF
+#define SDMA_PKT_DATA_FILL_MULTI_DMA_COUNT_dma_count_shift  0
+#define SDMA_PKT_DATA_FILL_MULTI_DMA_COUNT_DMA_COUNT(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_DMA_COUNT_dma_count_mask) << SDMA_PKT_DATA_FILL_MULTI_DMA_COUNT_dma_count_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_LO_dst_addr_31_0_offset 3
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_HI_dst_addr_63_32_offset 4
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for BYTE_COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_COUNT_count_offset 5
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_COUNT_count_mask   0x03FFFFFF
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_COUNT_count_shift  0
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_COUNT_COUNT(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_BYTE_COUNT_count_mask) << SDMA_PKT_DATA_FILL_MULTI_BYTE_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_POLL_REGMEM packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_op_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_POLL_REGMEM_HEADER_op_shift  0
+#define SDMA_PKT_POLL_REGMEM_HEADER_OP(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_op_mask) << SDMA_PKT_POLL_REGMEM_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_sub_op_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_POLL_REGMEM_HEADER_sub_op_shift  8
+#define SDMA_PKT_POLL_REGMEM_HEADER_SUB_OP(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_sub_op_mask) << SDMA_PKT_POLL_REGMEM_HEADER_sub_op_shift)
+
+/*define for cache_policy field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_cache_policy_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_cache_policy_mask   0x00000007
+#define SDMA_PKT_POLL_REGMEM_HEADER_cache_policy_shift  20
+#define SDMA_PKT_POLL_REGMEM_HEADER_CACHE_POLICY(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_cache_policy_mask) << SDMA_PKT_POLL_REGMEM_HEADER_cache_policy_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_cpv_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_POLL_REGMEM_HEADER_cpv_shift  24
+#define SDMA_PKT_POLL_REGMEM_HEADER_CPV(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_cpv_mask) << SDMA_PKT_POLL_REGMEM_HEADER_cpv_shift)
+
+/*define for hdp_flush field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_mask   0x00000001
+#define SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_shift  26
+#define SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_mask) << SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_shift)
+
+/*define for func field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_func_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_func_mask   0x00000007
+#define SDMA_PKT_POLL_REGMEM_HEADER_func_shift  28
+#define SDMA_PKT_POLL_REGMEM_HEADER_FUNC(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_func_mask) << SDMA_PKT_POLL_REGMEM_HEADER_func_shift)
+
+/*define for mem_poll field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_mask   0x00000001
+#define SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_shift  31
+#define SDMA_PKT_POLL_REGMEM_HEADER_MEM_POLL(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_mask) << SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_mask) << SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_mask) << SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_shift)
+
+/*define for VALUE word*/
+/*define for value field*/
+#define SDMA_PKT_POLL_REGMEM_VALUE_value_offset 3
+#define SDMA_PKT_POLL_REGMEM_VALUE_value_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_VALUE_value_shift  0
+#define SDMA_PKT_POLL_REGMEM_VALUE_VALUE(x) (((x) & SDMA_PKT_POLL_REGMEM_VALUE_value_mask) << SDMA_PKT_POLL_REGMEM_VALUE_value_shift)
+
+/*define for MASK word*/
+/*define for mask field*/
+#define SDMA_PKT_POLL_REGMEM_MASK_mask_offset 4
+#define SDMA_PKT_POLL_REGMEM_MASK_mask_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_MASK_mask_shift  0
+#define SDMA_PKT_POLL_REGMEM_MASK_MASK(x) (((x) & SDMA_PKT_POLL_REGMEM_MASK_mask_mask) << SDMA_PKT_POLL_REGMEM_MASK_mask_shift)
+
+/*define for DW5 word*/
+/*define for interval field*/
+#define SDMA_PKT_POLL_REGMEM_DW5_interval_offset 5
+#define SDMA_PKT_POLL_REGMEM_DW5_interval_mask   0x0000FFFF
+#define SDMA_PKT_POLL_REGMEM_DW5_interval_shift  0
+#define SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(x) (((x) & SDMA_PKT_POLL_REGMEM_DW5_interval_mask) << SDMA_PKT_POLL_REGMEM_DW5_interval_shift)
+
+/*define for retry_count field*/
+#define SDMA_PKT_POLL_REGMEM_DW5_retry_count_offset 5
+#define SDMA_PKT_POLL_REGMEM_DW5_retry_count_mask   0x00000FFF
+#define SDMA_PKT_POLL_REGMEM_DW5_retry_count_shift  16
+#define SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(x) (((x) & SDMA_PKT_POLL_REGMEM_DW5_retry_count_mask) << SDMA_PKT_POLL_REGMEM_DW5_retry_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_POLL_REG_WRITE_MEM packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_op_offset 0
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_op_shift  0
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_OP(x) (((x) & SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_op_mask) << SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_sub_op_offset 0
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_sub_op_shift  8
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_SUB_OP(x) (((x) & SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_sub_op_mask) << SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_sub_op_shift)
+
+/*define for cache_policy field*/
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_cache_policy_offset 0
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_cache_policy_mask   0x00000007
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_cache_policy_shift  24
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_CACHE_POLICY(x) (((x) & SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_cache_policy_mask) << SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_cache_policy_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_cpv_offset 0
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_cpv_shift  28
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_CPV(x) (((x) & SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_cpv_mask) << SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_cpv_shift)
+
+/*define for SRC_ADDR word*/
+/*define for addr_31_2 field*/
+#define SDMA_PKT_POLL_REG_WRITE_MEM_SRC_ADDR_addr_31_2_offset 1
+#define SDMA_PKT_POLL_REG_WRITE_MEM_SRC_ADDR_addr_31_2_mask   0x3FFFFFFF
+#define SDMA_PKT_POLL_REG_WRITE_MEM_SRC_ADDR_addr_31_2_shift  2
+#define SDMA_PKT_POLL_REG_WRITE_MEM_SRC_ADDR_ADDR_31_2(x) (((x) & SDMA_PKT_POLL_REG_WRITE_MEM_SRC_ADDR_addr_31_2_mask) << SDMA_PKT_POLL_REG_WRITE_MEM_SRC_ADDR_addr_31_2_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_LO_addr_31_0_offset 2
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_LO_addr_31_0_mask) << SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_LO_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_HI_addr_63_32_offset 3
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_HI_addr_63_32_mask) << SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_HI_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_POLL_DBIT_WRITE_MEM packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_op_offset 0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_op_shift  0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_OP(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_op_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_sub_op_offset 0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_sub_op_shift  8
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_SUB_OP(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_sub_op_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_sub_op_shift)
+
+/*define for ea field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_ea_offset 0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_ea_mask   0x00000003
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_ea_shift  16
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_EA(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_ea_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_ea_shift)
+
+/*define for cache_policy field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_cache_policy_offset 0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_cache_policy_mask   0x00000007
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_cache_policy_shift  24
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_CACHE_POLICY(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_cache_policy_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_cache_policy_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_cpv_offset 0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_cpv_shift  28
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_CPV(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_cpv_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_cpv_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_LO_addr_31_0_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_LO_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_HI_addr_63_32_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_HI_addr_63_32_shift)
+
+/*define for START_PAGE word*/
+/*define for addr_31_4 field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_START_PAGE_addr_31_4_offset 3
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_START_PAGE_addr_31_4_mask   0x0FFFFFFF
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_START_PAGE_addr_31_4_shift  4
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_START_PAGE_ADDR_31_4(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_START_PAGE_addr_31_4_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_START_PAGE_addr_31_4_shift)
+
+/*define for PAGE_NUM word*/
+/*define for page_num_31_0 field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_PAGE_NUM_page_num_31_0_offset 4
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_PAGE_NUM_page_num_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_PAGE_NUM_page_num_31_0_shift  0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_PAGE_NUM_PAGE_NUM_31_0(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_PAGE_NUM_page_num_31_0_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_PAGE_NUM_page_num_31_0_shift)
+
+
+/*
+** Definitions for SDMA_PKT_POLL_MEM_VERIFY packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_op_offset 0
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_op_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_OP(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_HEADER_op_mask) << SDMA_PKT_POLL_MEM_VERIFY_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_sub_op_offset 0
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_sub_op_shift  8
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_SUB_OP(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_HEADER_sub_op_mask) << SDMA_PKT_POLL_MEM_VERIFY_HEADER_sub_op_shift)
+
+/*define for cache_policy field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_cache_policy_offset 0
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_cache_policy_mask   0x00000007
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_cache_policy_shift  24
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_CACHE_POLICY(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_HEADER_cache_policy_mask) << SDMA_PKT_POLL_MEM_VERIFY_HEADER_cache_policy_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_cpv_offset 0
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_cpv_shift  28
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_CPV(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_HEADER_cpv_mask) << SDMA_PKT_POLL_MEM_VERIFY_HEADER_cpv_shift)
+
+/*define for mode field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_mode_offset 0
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_mode_mask   0x00000001
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_mode_shift  31
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_MODE(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_HEADER_mode_mask) << SDMA_PKT_POLL_MEM_VERIFY_HEADER_mode_shift)
+
+/*define for PATTERN word*/
+/*define for pattern field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_PATTERN_pattern_offset 1
+#define SDMA_PKT_POLL_MEM_VERIFY_PATTERN_pattern_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_PATTERN_pattern_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_PATTERN_PATTERN(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_PATTERN_pattern_mask) << SDMA_PKT_POLL_MEM_VERIFY_PATTERN_pattern_shift)
+
+/*define for CMP0_ADDR_START_LO word*/
+/*define for cmp0_start_31_0 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_LO_cmp0_start_31_0_offset 2
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_LO_cmp0_start_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_LO_cmp0_start_31_0_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_LO_CMP0_START_31_0(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_LO_cmp0_start_31_0_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_LO_cmp0_start_31_0_shift)
+
+/*define for CMP0_ADDR_START_HI word*/
+/*define for cmp0_start_63_32 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_HI_cmp0_start_63_32_offset 3
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_HI_cmp0_start_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_HI_cmp0_start_63_32_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_HI_CMP0_START_63_32(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_HI_cmp0_start_63_32_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_HI_cmp0_start_63_32_shift)
+
+/*define for CMP0_ADDR_END_LO word*/
+/*define for cmp0_end_31_0 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_LO_cmp0_end_31_0_offset 4
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_LO_cmp0_end_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_LO_cmp0_end_31_0_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_LO_CMP0_END_31_0(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_LO_cmp0_end_31_0_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_LO_cmp0_end_31_0_shift)
+
+/*define for CMP0_ADDR_END_HI word*/
+/*define for cmp0_end_63_32 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_HI_cmp0_end_63_32_offset 5
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_HI_cmp0_end_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_HI_cmp0_end_63_32_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_HI_CMP0_END_63_32(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_HI_cmp0_end_63_32_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_HI_cmp0_end_63_32_shift)
+
+/*define for CMP1_ADDR_START_LO word*/
+/*define for cmp1_start_31_0 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_LO_cmp1_start_31_0_offset 6
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_LO_cmp1_start_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_LO_cmp1_start_31_0_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_LO_CMP1_START_31_0(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_LO_cmp1_start_31_0_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_LO_cmp1_start_31_0_shift)
+
+/*define for CMP1_ADDR_START_HI word*/
+/*define for cmp1_start_63_32 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_HI_cmp1_start_63_32_offset 7
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_HI_cmp1_start_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_HI_cmp1_start_63_32_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_HI_CMP1_START_63_32(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_HI_cmp1_start_63_32_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_HI_cmp1_start_63_32_shift)
+
+/*define for CMP1_ADDR_END_LO word*/
+/*define for cmp1_end_31_0 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_LO_cmp1_end_31_0_offset 8
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_LO_cmp1_end_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_LO_cmp1_end_31_0_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_LO_CMP1_END_31_0(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_LO_cmp1_end_31_0_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_LO_cmp1_end_31_0_shift)
+
+/*define for CMP1_ADDR_END_HI word*/
+/*define for cmp1_end_63_32 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_HI_cmp1_end_63_32_offset 9
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_HI_cmp1_end_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_HI_cmp1_end_63_32_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_HI_CMP1_END_63_32(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_HI_cmp1_end_63_32_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_HI_cmp1_end_63_32_shift)
+
+/*define for REC_ADDR_LO word*/
+/*define for rec_31_0 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_LO_rec_31_0_offset 10
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_LO_rec_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_LO_rec_31_0_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_LO_REC_31_0(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_LO_rec_31_0_mask) << SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_LO_rec_31_0_shift)
+
+/*define for REC_ADDR_HI word*/
+/*define for rec_63_32 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_HI_rec_63_32_offset 11
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_HI_rec_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_HI_rec_63_32_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_HI_REC_63_32(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_HI_rec_63_32_mask) << SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_HI_rec_63_32_shift)
+
+/*define for RESERVED word*/
+/*define for reserved field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_RESERVED_reserved_offset 12
+#define SDMA_PKT_POLL_MEM_VERIFY_RESERVED_reserved_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_RESERVED_reserved_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_RESERVED_RESERVED(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_RESERVED_reserved_mask) << SDMA_PKT_POLL_MEM_VERIFY_RESERVED_reserved_shift)
+
+
+/*
+** Definitions for SDMA_PKT_ATOMIC packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_ATOMIC_HEADER_op_offset 0
+#define SDMA_PKT_ATOMIC_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_ATOMIC_HEADER_op_shift  0
+#define SDMA_PKT_ATOMIC_HEADER_OP(x) (((x) & SDMA_PKT_ATOMIC_HEADER_op_mask) << SDMA_PKT_ATOMIC_HEADER_op_shift)
+
+/*define for loop field*/
+#define SDMA_PKT_ATOMIC_HEADER_loop_offset 0
+#define SDMA_PKT_ATOMIC_HEADER_loop_mask   0x00000001
+#define SDMA_PKT_ATOMIC_HEADER_loop_shift  16
+#define SDMA_PKT_ATOMIC_HEADER_LOOP(x) (((x) & SDMA_PKT_ATOMIC_HEADER_loop_mask) << SDMA_PKT_ATOMIC_HEADER_loop_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_ATOMIC_HEADER_tmz_offset 0
+#define SDMA_PKT_ATOMIC_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_ATOMIC_HEADER_tmz_shift  18
+#define SDMA_PKT_ATOMIC_HEADER_TMZ(x) (((x) & SDMA_PKT_ATOMIC_HEADER_tmz_mask) << SDMA_PKT_ATOMIC_HEADER_tmz_shift)
+
+/*define for cache_policy field*/
+#define SDMA_PKT_ATOMIC_HEADER_cache_policy_offset 0
+#define SDMA_PKT_ATOMIC_HEADER_cache_policy_mask   0x00000007
+#define SDMA_PKT_ATOMIC_HEADER_cache_policy_shift  20
+#define SDMA_PKT_ATOMIC_HEADER_CACHE_POLICY(x) (((x) & SDMA_PKT_ATOMIC_HEADER_cache_policy_mask) << SDMA_PKT_ATOMIC_HEADER_cache_policy_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_ATOMIC_HEADER_cpv_offset 0
+#define SDMA_PKT_ATOMIC_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_ATOMIC_HEADER_cpv_shift  24
+#define SDMA_PKT_ATOMIC_HEADER_CPV(x) (((x) & SDMA_PKT_ATOMIC_HEADER_cpv_mask) << SDMA_PKT_ATOMIC_HEADER_cpv_shift)
+
+/*define for atomic_op field*/
+#define SDMA_PKT_ATOMIC_HEADER_atomic_op_offset 0
+#define SDMA_PKT_ATOMIC_HEADER_atomic_op_mask   0x0000007F
+#define SDMA_PKT_ATOMIC_HEADER_atomic_op_shift  25
+#define SDMA_PKT_ATOMIC_HEADER_ATOMIC_OP(x) (((x) & SDMA_PKT_ATOMIC_HEADER_atomic_op_mask) << SDMA_PKT_ATOMIC_HEADER_atomic_op_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_ATOMIC_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_mask) << SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_ATOMIC_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_mask) << SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_shift)
+
+/*define for SRC_DATA_LO word*/
+/*define for src_data_31_0 field*/
+#define SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_offset 3
+#define SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_shift  0
+#define SDMA_PKT_ATOMIC_SRC_DATA_LO_SRC_DATA_31_0(x) (((x) & SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_mask) << SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_shift)
+
+/*define for SRC_DATA_HI word*/
+/*define for src_data_63_32 field*/
+#define SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_offset 4
+#define SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_shift  0
+#define SDMA_PKT_ATOMIC_SRC_DATA_HI_SRC_DATA_63_32(x) (((x) & SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_mask) << SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_shift)
+
+/*define for CMP_DATA_LO word*/
+/*define for cmp_data_31_0 field*/
+#define SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_offset 5
+#define SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_shift  0
+#define SDMA_PKT_ATOMIC_CMP_DATA_LO_CMP_DATA_31_0(x) (((x) & SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_mask) << SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_shift)
+
+/*define for CMP_DATA_HI word*/
+/*define for cmp_data_63_32 field*/
+#define SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_offset 6
+#define SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_shift  0
+#define SDMA_PKT_ATOMIC_CMP_DATA_HI_CMP_DATA_63_32(x) (((x) & SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_mask) << SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_shift)
+
+/*define for LOOP_INTERVAL word*/
+/*define for loop_interval field*/
+#define SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_offset 7
+#define SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_mask   0x00001FFF
+#define SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_shift  0
+#define SDMA_PKT_ATOMIC_LOOP_INTERVAL_LOOP_INTERVAL(x) (((x) & SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_mask) << SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TIMESTAMP_SET packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_op_offset 0
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_op_shift  0
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_OP(x) (((x) & SDMA_PKT_TIMESTAMP_SET_HEADER_op_mask) << SDMA_PKT_TIMESTAMP_SET_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_offset 0
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_shift  8
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_mask) << SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_shift)
+
+/*define for INIT_DATA_LO word*/
+/*define for init_data_31_0 field*/
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_offset 1
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_shift  0
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_INIT_DATA_31_0(x) (((x) & SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_mask) << SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_shift)
+
+/*define for INIT_DATA_HI word*/
+/*define for init_data_63_32 field*/
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_offset 2
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_shift  0
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_INIT_DATA_63_32(x) (((x) & SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_mask) << SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TIMESTAMP_GET packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_op_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_HEADER_op_mask) << SDMA_PKT_TIMESTAMP_GET_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_shift  8
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_mask) << SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_shift)
+
+/*define for l2_policy field*/
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_l2_policy_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_l2_policy_mask   0x00000003
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_l2_policy_shift  24
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_L2_POLICY(x) (((x) & SDMA_PKT_TIMESTAMP_GET_HEADER_l2_policy_mask) << SDMA_PKT_TIMESTAMP_GET_HEADER_l2_policy_shift)
+
+/*define for llc_policy field*/
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_llc_policy_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_llc_policy_mask   0x00000001
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_llc_policy_shift  26
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_LLC_POLICY(x) (((x) & SDMA_PKT_TIMESTAMP_GET_HEADER_llc_policy_mask) << SDMA_PKT_TIMESTAMP_GET_HEADER_llc_policy_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_cpv_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_cpv_shift  28
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_CPV(x) (((x) & SDMA_PKT_TIMESTAMP_GET_HEADER_cpv_mask) << SDMA_PKT_TIMESTAMP_GET_HEADER_cpv_shift)
+
+/*define for WRITE_ADDR_LO word*/
+/*define for write_addr_31_3 field*/
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_offset 1
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_mask   0x1FFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_shift  3
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_WRITE_ADDR_31_3(x) (((x) & SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_mask) << SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_shift)
+
+/*define for WRITE_ADDR_HI word*/
+/*define for write_addr_63_32 field*/
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_offset 2
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_WRITE_ADDR_63_32(x) (((x) & SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_mask) << SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TIMESTAMP_GET_GLOBAL packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_shift  8
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_shift)
+
+/*define for l2_policy field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_l2_policy_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_l2_policy_mask   0x00000003
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_l2_policy_shift  24
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_L2_POLICY(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_l2_policy_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_l2_policy_shift)
+
+/*define for llc_policy field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_llc_policy_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_llc_policy_mask   0x00000001
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_llc_policy_shift  26
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_LLC_POLICY(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_llc_policy_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_llc_policy_shift)
+
+/*define for cpv field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_cpv_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_cpv_mask   0x00000001
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_cpv_shift  28
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_CPV(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_cpv_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_cpv_shift)
+
+/*define for WRITE_ADDR_LO word*/
+/*define for write_addr_31_3 field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_offset 1
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_mask   0x1FFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_shift  3
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_WRITE_ADDR_31_3(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_shift)
+
+/*define for WRITE_ADDR_HI word*/
+/*define for write_addr_63_32 field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_offset 2
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_WRITE_ADDR_63_32(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TRAP packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TRAP_HEADER_op_offset 0
+#define SDMA_PKT_TRAP_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TRAP_HEADER_op_shift  0
+#define SDMA_PKT_TRAP_HEADER_OP(x) (((x) & SDMA_PKT_TRAP_HEADER_op_mask) << SDMA_PKT_TRAP_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TRAP_HEADER_sub_op_offset 0
+#define SDMA_PKT_TRAP_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TRAP_HEADER_sub_op_shift  8
+#define SDMA_PKT_TRAP_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TRAP_HEADER_sub_op_mask) << SDMA_PKT_TRAP_HEADER_sub_op_shift)
+
+/*define for INT_CONTEXT word*/
+/*define for int_context field*/
+#define SDMA_PKT_TRAP_INT_CONTEXT_int_context_offset 1
+#define SDMA_PKT_TRAP_INT_CONTEXT_int_context_mask   0x0FFFFFFF
+#define SDMA_PKT_TRAP_INT_CONTEXT_int_context_shift  0
+#define SDMA_PKT_TRAP_INT_CONTEXT_INT_CONTEXT(x) (((x) & SDMA_PKT_TRAP_INT_CONTEXT_int_context_mask) << SDMA_PKT_TRAP_INT_CONTEXT_int_context_shift)
+
+
+/*
+** Definitions for SDMA_PKT_DUMMY_TRAP packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_DUMMY_TRAP_HEADER_op_offset 0
+#define SDMA_PKT_DUMMY_TRAP_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_DUMMY_TRAP_HEADER_op_shift  0
+#define SDMA_PKT_DUMMY_TRAP_HEADER_OP(x) (((x) & SDMA_PKT_DUMMY_TRAP_HEADER_op_mask) << SDMA_PKT_DUMMY_TRAP_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_DUMMY_TRAP_HEADER_sub_op_offset 0
+#define SDMA_PKT_DUMMY_TRAP_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_DUMMY_TRAP_HEADER_sub_op_shift  8
+#define SDMA_PKT_DUMMY_TRAP_HEADER_SUB_OP(x) (((x) & SDMA_PKT_DUMMY_TRAP_HEADER_sub_op_mask) << SDMA_PKT_DUMMY_TRAP_HEADER_sub_op_shift)
+
+/*define for INT_CONTEXT word*/
+/*define for int_context field*/
+#define SDMA_PKT_DUMMY_TRAP_INT_CONTEXT_int_context_offset 1
+#define SDMA_PKT_DUMMY_TRAP_INT_CONTEXT_int_context_mask   0x0FFFFFFF
+#define SDMA_PKT_DUMMY_TRAP_INT_CONTEXT_int_context_shift  0
+#define SDMA_PKT_DUMMY_TRAP_INT_CONTEXT_INT_CONTEXT(x) (((x) & SDMA_PKT_DUMMY_TRAP_INT_CONTEXT_int_context_mask) << SDMA_PKT_DUMMY_TRAP_INT_CONTEXT_int_context_shift)
+
+
+/*
+** Definitions for SDMA_PKT_GPUVM_INV packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_GPUVM_INV_HEADER_op_offset 0
+#define SDMA_PKT_GPUVM_INV_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_GPUVM_INV_HEADER_op_shift  0
+#define SDMA_PKT_GPUVM_INV_HEADER_OP(x) (((x) & SDMA_PKT_GPUVM_INV_HEADER_op_mask) << SDMA_PKT_GPUVM_INV_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_GPUVM_INV_HEADER_sub_op_offset 0
+#define SDMA_PKT_GPUVM_INV_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_GPUVM_INV_HEADER_sub_op_shift  8
+#define SDMA_PKT_GPUVM_INV_HEADER_SUB_OP(x) (((x) & SDMA_PKT_GPUVM_INV_HEADER_sub_op_mask) << SDMA_PKT_GPUVM_INV_HEADER_sub_op_shift)
+
+/*define for PAYLOAD1 word*/
+/*define for per_vmid_inv_req field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_per_vmid_inv_req_offset 1
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_per_vmid_inv_req_mask   0x0000FFFF
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_per_vmid_inv_req_shift  0
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_PER_VMID_INV_REQ(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD1_per_vmid_inv_req_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD1_per_vmid_inv_req_shift)
+
+/*define for flush_type field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_flush_type_offset 1
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_flush_type_mask   0x00000007
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_flush_type_shift  16
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_FLUSH_TYPE(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD1_flush_type_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD1_flush_type_shift)
+
+/*define for l2_ptes field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_ptes_offset 1
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_ptes_mask   0x00000001
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_ptes_shift  19
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_L2_PTES(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_ptes_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_ptes_shift)
+
+/*define for l2_pde0 field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde0_offset 1
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde0_mask   0x00000001
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde0_shift  20
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_L2_PDE0(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde0_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde0_shift)
+
+/*define for l2_pde1 field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde1_offset 1
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde1_mask   0x00000001
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde1_shift  21
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_L2_PDE1(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde1_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde1_shift)
+
+/*define for l2_pde2 field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde2_offset 1
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde2_mask   0x00000001
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde2_shift  22
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_L2_PDE2(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde2_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD1_l2_pde2_shift)
+
+/*define for l1_ptes field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l1_ptes_offset 1
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l1_ptes_mask   0x00000001
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_l1_ptes_shift  23
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_L1_PTES(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD1_l1_ptes_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD1_l1_ptes_shift)
+
+/*define for clr_protection_fault_status_addr field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_clr_protection_fault_status_addr_offset 1
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_clr_protection_fault_status_addr_mask   0x00000001
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_clr_protection_fault_status_addr_shift  24
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_CLR_PROTECTION_FAULT_STATUS_ADDR(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD1_clr_protection_fault_status_addr_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD1_clr_protection_fault_status_addr_shift)
+
+/*define for log_request field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_log_request_offset 1
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_log_request_mask   0x00000001
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_log_request_shift  25
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_LOG_REQUEST(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD1_log_request_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD1_log_request_shift)
+
+/*define for four_kilobytes field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_four_kilobytes_offset 1
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_four_kilobytes_mask   0x00000001
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_four_kilobytes_shift  26
+#define SDMA_PKT_GPUVM_INV_PAYLOAD1_FOUR_KILOBYTES(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD1_four_kilobytes_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD1_four_kilobytes_shift)
+
+/*define for PAYLOAD2 word*/
+/*define for s field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD2_s_offset 2
+#define SDMA_PKT_GPUVM_INV_PAYLOAD2_s_mask   0x00000001
+#define SDMA_PKT_GPUVM_INV_PAYLOAD2_s_shift  0
+#define SDMA_PKT_GPUVM_INV_PAYLOAD2_S(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD2_s_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD2_s_shift)
+
+/*define for page_va_42_12 field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD2_page_va_42_12_offset 2
+#define SDMA_PKT_GPUVM_INV_PAYLOAD2_page_va_42_12_mask   0x7FFFFFFF
+#define SDMA_PKT_GPUVM_INV_PAYLOAD2_page_va_42_12_shift  1
+#define SDMA_PKT_GPUVM_INV_PAYLOAD2_PAGE_VA_42_12(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD2_page_va_42_12_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD2_page_va_42_12_shift)
+
+/*define for PAYLOAD3 word*/
+/*define for page_va_47_43 field*/
+#define SDMA_PKT_GPUVM_INV_PAYLOAD3_page_va_47_43_offset 3
+#define SDMA_PKT_GPUVM_INV_PAYLOAD3_page_va_47_43_mask   0x0000003F
+#define SDMA_PKT_GPUVM_INV_PAYLOAD3_page_va_47_43_shift  0
+#define SDMA_PKT_GPUVM_INV_PAYLOAD3_PAGE_VA_47_43(x) (((x) & SDMA_PKT_GPUVM_INV_PAYLOAD3_page_va_47_43_mask) << SDMA_PKT_GPUVM_INV_PAYLOAD3_page_va_47_43_shift)
+
+
+/*
+** Definitions for SDMA_PKT_GCR_REQ packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_GCR_REQ_HEADER_op_offset 0
+#define SDMA_PKT_GCR_REQ_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_GCR_REQ_HEADER_op_shift  0
+#define SDMA_PKT_GCR_REQ_HEADER_OP(x) (((x) & SDMA_PKT_GCR_REQ_HEADER_op_mask) << SDMA_PKT_GCR_REQ_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_GCR_REQ_HEADER_sub_op_offset 0
+#define SDMA_PKT_GCR_REQ_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_GCR_REQ_HEADER_sub_op_shift  8
+#define SDMA_PKT_GCR_REQ_HEADER_SUB_OP(x) (((x) & SDMA_PKT_GCR_REQ_HEADER_sub_op_mask) << SDMA_PKT_GCR_REQ_HEADER_sub_op_shift)
+
+/*define for PAYLOAD1 word*/
+/*define for base_va_31_7 field*/
+#define SDMA_PKT_GCR_REQ_PAYLOAD1_base_va_31_7_offset 1
+#define SDMA_PKT_GCR_REQ_PAYLOAD1_base_va_31_7_mask   0x01FFFFFF
+#define SDMA_PKT_GCR_REQ_PAYLOAD1_base_va_31_7_shift  7
+#define SDMA_PKT_GCR_REQ_PAYLOAD1_BASE_VA_31_7(x) (((x) & SDMA_PKT_GCR_REQ_PAYLOAD1_base_va_31_7_mask) << SDMA_PKT_GCR_REQ_PAYLOAD1_base_va_31_7_shift)
+
+/*define for PAYLOAD2 word*/
+/*define for base_va_47_32 field*/
+#define SDMA_PKT_GCR_REQ_PAYLOAD2_base_va_47_32_offset 2
+#define SDMA_PKT_GCR_REQ_PAYLOAD2_base_va_47_32_mask   0x0000FFFF
+#define SDMA_PKT_GCR_REQ_PAYLOAD2_base_va_47_32_shift  0
+#define SDMA_PKT_GCR_REQ_PAYLOAD2_BASE_VA_47_32(x) (((x) & SDMA_PKT_GCR_REQ_PAYLOAD2_base_va_47_32_mask) << SDMA_PKT_GCR_REQ_PAYLOAD2_base_va_47_32_shift)
+
+/*define for gcr_control_15_0 field*/
+#define SDMA_PKT_GCR_REQ_PAYLOAD2_gcr_control_15_0_offset 2
+#define SDMA_PKT_GCR_REQ_PAYLOAD2_gcr_control_15_0_mask   0x0000FFFF
+#define SDMA_PKT_GCR_REQ_PAYLOAD2_gcr_control_15_0_shift  16
+#define SDMA_PKT_GCR_REQ_PAYLOAD2_GCR_CONTROL_15_0(x) (((x) & SDMA_PKT_GCR_REQ_PAYLOAD2_gcr_control_15_0_mask) << SDMA_PKT_GCR_REQ_PAYLOAD2_gcr_control_15_0_shift)
+
+/*define for PAYLOAD3 word*/
+/*define for gcr_control_18_16 field*/
+#define SDMA_PKT_GCR_REQ_PAYLOAD3_gcr_control_18_16_offset 3
+#define SDMA_PKT_GCR_REQ_PAYLOAD3_gcr_control_18_16_mask   0x00000007
+#define SDMA_PKT_GCR_REQ_PAYLOAD3_gcr_control_18_16_shift  0
+#define SDMA_PKT_GCR_REQ_PAYLOAD3_GCR_CONTROL_18_16(x) (((x) & SDMA_PKT_GCR_REQ_PAYLOAD3_gcr_control_18_16_mask) << SDMA_PKT_GCR_REQ_PAYLOAD3_gcr_control_18_16_shift)
+
+/*define for limit_va_31_7 field*/
+#define SDMA_PKT_GCR_REQ_PAYLOAD3_limit_va_31_7_offset 3
+#define SDMA_PKT_GCR_REQ_PAYLOAD3_limit_va_31_7_mask   0x01FFFFFF
+#define SDMA_PKT_GCR_REQ_PAYLOAD3_limit_va_31_7_shift  7
+#define SDMA_PKT_GCR_REQ_PAYLOAD3_LIMIT_VA_31_7(x) (((x) & SDMA_PKT_GCR_REQ_PAYLOAD3_limit_va_31_7_mask) << SDMA_PKT_GCR_REQ_PAYLOAD3_limit_va_31_7_shift)
+
+/*define for PAYLOAD4 word*/
+/*define for limit_va_47_32 field*/
+#define SDMA_PKT_GCR_REQ_PAYLOAD4_limit_va_47_32_offset 4
+#define SDMA_PKT_GCR_REQ_PAYLOAD4_limit_va_47_32_mask   0x0000FFFF
+#define SDMA_PKT_GCR_REQ_PAYLOAD4_limit_va_47_32_shift  0
+#define SDMA_PKT_GCR_REQ_PAYLOAD4_LIMIT_VA_47_32(x) (((x) & SDMA_PKT_GCR_REQ_PAYLOAD4_limit_va_47_32_mask) << SDMA_PKT_GCR_REQ_PAYLOAD4_limit_va_47_32_shift)
+
+/*define for vmid field*/
+#define SDMA_PKT_GCR_REQ_PAYLOAD4_vmid_offset 4
+#define SDMA_PKT_GCR_REQ_PAYLOAD4_vmid_mask   0x0000000F
+#define SDMA_PKT_GCR_REQ_PAYLOAD4_vmid_shift  24
+#define SDMA_PKT_GCR_REQ_PAYLOAD4_VMID(x) (((x) & SDMA_PKT_GCR_REQ_PAYLOAD4_vmid_mask) << SDMA_PKT_GCR_REQ_PAYLOAD4_vmid_shift)
+
+
+/*
+** Definitions for SDMA_PKT_NOP packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_NOP_HEADER_op_offset 0
+#define SDMA_PKT_NOP_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_NOP_HEADER_op_shift  0
+#define SDMA_PKT_NOP_HEADER_OP(x) (((x) & SDMA_PKT_NOP_HEADER_op_mask) << SDMA_PKT_NOP_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_NOP_HEADER_sub_op_offset 0
+#define SDMA_PKT_NOP_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_NOP_HEADER_sub_op_shift  8
+#define SDMA_PKT_NOP_HEADER_SUB_OP(x) (((x) & SDMA_PKT_NOP_HEADER_sub_op_mask) << SDMA_PKT_NOP_HEADER_sub_op_shift)
+
+/*define for count field*/
+#define SDMA_PKT_NOP_HEADER_count_offset 0
+#define SDMA_PKT_NOP_HEADER_count_mask   0x00003FFF
+#define SDMA_PKT_NOP_HEADER_count_shift  16
+#define SDMA_PKT_NOP_HEADER_COUNT(x) (((x) & SDMA_PKT_NOP_HEADER_count_mask) << SDMA_PKT_NOP_HEADER_count_shift)
+
+/*define for DATA0 word*/
+/*define for data0 field*/
+#define SDMA_PKT_NOP_DATA0_data0_offset 1
+#define SDMA_PKT_NOP_DATA0_data0_mask   0xFFFFFFFF
+#define SDMA_PKT_NOP_DATA0_data0_shift  0
+#define SDMA_PKT_NOP_DATA0_DATA0(x) (((x) & SDMA_PKT_NOP_DATA0_data0_mask) << SDMA_PKT_NOP_DATA0_data0_shift)
+
+
+/*
+** Definitions for SDMA_AQL_PKT_HEADER packet
+*/
+
+/*define for HEADER word*/
+/*define for format field*/
+#define SDMA_AQL_PKT_HEADER_HEADER_format_offset 0
+#define SDMA_AQL_PKT_HEADER_HEADER_format_mask   0x000000FF
+#define SDMA_AQL_PKT_HEADER_HEADER_format_shift  0
+#define SDMA_AQL_PKT_HEADER_HEADER_FORMAT(x) (((x) & SDMA_AQL_PKT_HEADER_HEADER_format_mask) << SDMA_AQL_PKT_HEADER_HEADER_format_shift)
+
+/*define for barrier field*/
+#define SDMA_AQL_PKT_HEADER_HEADER_barrier_offset 0
+#define SDMA_AQL_PKT_HEADER_HEADER_barrier_mask   0x00000001
+#define SDMA_AQL_PKT_HEADER_HEADER_barrier_shift  8
+#define SDMA_AQL_PKT_HEADER_HEADER_BARRIER(x) (((x) & SDMA_AQL_PKT_HEADER_HEADER_barrier_mask) << SDMA_AQL_PKT_HEADER_HEADER_barrier_shift)
+
+/*define for acquire_fence_scope field*/
+#define SDMA_AQL_PKT_HEADER_HEADER_acquire_fence_scope_offset 0
+#define SDMA_AQL_PKT_HEADER_HEADER_acquire_fence_scope_mask   0x00000003
+#define SDMA_AQL_PKT_HEADER_HEADER_acquire_fence_scope_shift  9
+#define SDMA_AQL_PKT_HEADER_HEADER_ACQUIRE_FENCE_SCOPE(x) (((x) & SDMA_AQL_PKT_HEADER_HEADER_acquire_fence_scope_mask) << SDMA_AQL_PKT_HEADER_HEADER_acquire_fence_scope_shift)
+
+/*define for release_fence_scope field*/
+#define SDMA_AQL_PKT_HEADER_HEADER_release_fence_scope_offset 0
+#define SDMA_AQL_PKT_HEADER_HEADER_release_fence_scope_mask   0x00000003
+#define SDMA_AQL_PKT_HEADER_HEADER_release_fence_scope_shift  11
+#define SDMA_AQL_PKT_HEADER_HEADER_RELEASE_FENCE_SCOPE(x) (((x) & SDMA_AQL_PKT_HEADER_HEADER_release_fence_scope_mask) << SDMA_AQL_PKT_HEADER_HEADER_release_fence_scope_shift)
+
+/*define for reserved field*/
+#define SDMA_AQL_PKT_HEADER_HEADER_reserved_offset 0
+#define SDMA_AQL_PKT_HEADER_HEADER_reserved_mask   0x00000007
+#define SDMA_AQL_PKT_HEADER_HEADER_reserved_shift  13
+#define SDMA_AQL_PKT_HEADER_HEADER_RESERVED(x) (((x) & SDMA_AQL_PKT_HEADER_HEADER_reserved_mask) << SDMA_AQL_PKT_HEADER_HEADER_reserved_shift)
+
+/*define for op field*/
+#define SDMA_AQL_PKT_HEADER_HEADER_op_offset 0
+#define SDMA_AQL_PKT_HEADER_HEADER_op_mask   0x0000000F
+#define SDMA_AQL_PKT_HEADER_HEADER_op_shift  16
+#define SDMA_AQL_PKT_HEADER_HEADER_OP(x) (((x) & SDMA_AQL_PKT_HEADER_HEADER_op_mask) << SDMA_AQL_PKT_HEADER_HEADER_op_shift)
+
+/*define for subop field*/
+#define SDMA_AQL_PKT_HEADER_HEADER_subop_offset 0
+#define SDMA_AQL_PKT_HEADER_HEADER_subop_mask   0x00000007
+#define SDMA_AQL_PKT_HEADER_HEADER_subop_shift  20
+#define SDMA_AQL_PKT_HEADER_HEADER_SUBOP(x) (((x) & SDMA_AQL_PKT_HEADER_HEADER_subop_mask) << SDMA_AQL_PKT_HEADER_HEADER_subop_shift)
+
+/*define for cpv field*/
+#define SDMA_AQL_PKT_HEADER_HEADER_cpv_offset 0
+#define SDMA_AQL_PKT_HEADER_HEADER_cpv_mask   0x00000001
+#define SDMA_AQL_PKT_HEADER_HEADER_cpv_shift  28
+#define SDMA_AQL_PKT_HEADER_HEADER_CPV(x) (((x) & SDMA_AQL_PKT_HEADER_HEADER_cpv_mask) << SDMA_AQL_PKT_HEADER_HEADER_cpv_shift)
+
+
+/*
+** Definitions for SDMA_AQL_PKT_COPY_LINEAR packet
+*/
+
+/*define for HEADER word*/
+/*define for format field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_format_offset 0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_format_mask   0x000000FF
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_format_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_FORMAT(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_HEADER_format_mask) << SDMA_AQL_PKT_COPY_LINEAR_HEADER_format_shift)
+
+/*define for barrier field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_barrier_offset 0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_barrier_mask   0x00000001
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_barrier_shift  8
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_BARRIER(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_HEADER_barrier_mask) << SDMA_AQL_PKT_COPY_LINEAR_HEADER_barrier_shift)
+
+/*define for acquire_fence_scope field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_acquire_fence_scope_offset 0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_acquire_fence_scope_mask   0x00000003
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_acquire_fence_scope_shift  9
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_ACQUIRE_FENCE_SCOPE(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_HEADER_acquire_fence_scope_mask) << SDMA_AQL_PKT_COPY_LINEAR_HEADER_acquire_fence_scope_shift)
+
+/*define for release_fence_scope field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_release_fence_scope_offset 0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_release_fence_scope_mask   0x00000003
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_release_fence_scope_shift  11
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_RELEASE_FENCE_SCOPE(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_HEADER_release_fence_scope_mask) << SDMA_AQL_PKT_COPY_LINEAR_HEADER_release_fence_scope_shift)
+
+/*define for reserved field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_reserved_offset 0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_reserved_mask   0x00000007
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_reserved_shift  13
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_RESERVED(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_HEADER_reserved_mask) << SDMA_AQL_PKT_COPY_LINEAR_HEADER_reserved_shift)
+
+/*define for op field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_op_offset 0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_op_mask   0x0000000F
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_op_shift  16
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_OP(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_HEADER_op_mask) << SDMA_AQL_PKT_COPY_LINEAR_HEADER_op_shift)
+
+/*define for subop field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_subop_offset 0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_subop_mask   0x00000007
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_subop_shift  20
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_SUBOP(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_HEADER_subop_mask) << SDMA_AQL_PKT_COPY_LINEAR_HEADER_subop_shift)
+
+/*define for cpv field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_cpv_offset 0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_cpv_mask   0x00000001
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_cpv_shift  28
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_CPV(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_HEADER_cpv_mask) << SDMA_AQL_PKT_COPY_LINEAR_HEADER_cpv_shift)
+
+/*define for RESERVED_DW1 word*/
+/*define for reserved_dw1 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW1_reserved_dw1_offset 1
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW1_reserved_dw1_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW1_reserved_dw1_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW1_RESERVED_DW1(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW1_reserved_dw1_mask) << SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW1_reserved_dw1_shift)
+
+/*define for RETURN_ADDR_LO word*/
+/*define for return_addr_31_0 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_LO_return_addr_31_0_offset 2
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_LO_return_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_LO_return_addr_31_0_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_LO_RETURN_ADDR_31_0(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_LO_return_addr_31_0_mask) << SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_LO_return_addr_31_0_shift)
+
+/*define for RETURN_ADDR_HI word*/
+/*define for return_addr_63_32 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_HI_return_addr_63_32_offset 3
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_HI_return_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_HI_return_addr_63_32_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_HI_RETURN_ADDR_63_32(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_HI_return_addr_63_32_mask) << SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_HI_return_addr_63_32_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_COUNT_count_offset 4
+#define SDMA_AQL_PKT_COPY_LINEAR_COUNT_count_mask   0x003FFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_COUNT_count_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_COUNT_COUNT(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_COUNT_count_mask) << SDMA_AQL_PKT_COPY_LINEAR_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst_sw field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_dst_sw_offset 5
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_dst_sw_mask   0x00000003
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_dst_sw_shift  16
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_DST_SW(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_dst_sw_mask) << SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_dst_sw_shift)
+
+/*define for dst_cache_policy field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_dst_cache_policy_offset 5
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_dst_cache_policy_mask   0x00000007
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_dst_cache_policy_shift  18
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_DST_CACHE_POLICY(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_dst_cache_policy_mask) << SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_dst_cache_policy_shift)
+
+/*define for src_sw field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_src_sw_offset 5
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_src_sw_shift  24
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_SRC_SW(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_src_sw_mask) << SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_src_sw_shift)
+
+/*define for src_cache_policy field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_src_cache_policy_offset 5
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_src_cache_policy_mask   0x00000007
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_src_cache_policy_shift  26
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_SRC_CACHE_POLICY(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_src_cache_policy_mask) << SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_src_cache_policy_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_offset 6
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_offset 7
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_offset 8
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_offset 9
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for RESERVED_DW10 word*/
+/*define for reserved_dw10 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW10_reserved_dw10_offset 10
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW10_reserved_dw10_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW10_reserved_dw10_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW10_RESERVED_DW10(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW10_reserved_dw10_mask) << SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW10_reserved_dw10_shift)
+
+/*define for RESERVED_DW11 word*/
+/*define for reserved_dw11 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW11_reserved_dw11_offset 11
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW11_reserved_dw11_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW11_reserved_dw11_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW11_RESERVED_DW11(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW11_reserved_dw11_mask) << SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW11_reserved_dw11_shift)
+
+/*define for RESERVED_DW12 word*/
+/*define for reserved_dw12 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW12_reserved_dw12_offset 12
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW12_reserved_dw12_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW12_reserved_dw12_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW12_RESERVED_DW12(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW12_reserved_dw12_mask) << SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW12_reserved_dw12_shift)
+
+/*define for RESERVED_DW13 word*/
+/*define for reserved_dw13 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW13_reserved_dw13_offset 13
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW13_reserved_dw13_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW13_reserved_dw13_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW13_RESERVED_DW13(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW13_reserved_dw13_mask) << SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW13_reserved_dw13_shift)
+
+/*define for COMPLETION_SIGNAL_LO word*/
+/*define for completion_signal_31_0 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_LO_completion_signal_31_0_offset 14
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_LO_completion_signal_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_LO_completion_signal_31_0_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_LO_COMPLETION_SIGNAL_31_0(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_LO_completion_signal_31_0_mask) << SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_LO_completion_signal_31_0_shift)
+
+/*define for COMPLETION_SIGNAL_HI word*/
+/*define for completion_signal_63_32 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_HI_completion_signal_63_32_offset 15
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_HI_completion_signal_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_HI_completion_signal_63_32_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_HI_COMPLETION_SIGNAL_63_32(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_HI_completion_signal_63_32_mask) << SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_HI_completion_signal_63_32_shift)
+
+
+/*
+** Definitions for SDMA_AQL_PKT_BARRIER_OR packet
+*/
+
+/*define for HEADER word*/
+/*define for format field*/
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_format_offset 0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_format_mask   0x000000FF
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_format_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_FORMAT(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_HEADER_format_mask) << SDMA_AQL_PKT_BARRIER_OR_HEADER_format_shift)
+
+/*define for barrier field*/
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_barrier_offset 0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_barrier_mask   0x00000001
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_barrier_shift  8
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_BARRIER(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_HEADER_barrier_mask) << SDMA_AQL_PKT_BARRIER_OR_HEADER_barrier_shift)
+
+/*define for acquire_fence_scope field*/
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_acquire_fence_scope_offset 0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_acquire_fence_scope_mask   0x00000003
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_acquire_fence_scope_shift  9
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_ACQUIRE_FENCE_SCOPE(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_HEADER_acquire_fence_scope_mask) << SDMA_AQL_PKT_BARRIER_OR_HEADER_acquire_fence_scope_shift)
+
+/*define for release_fence_scope field*/
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_release_fence_scope_offset 0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_release_fence_scope_mask   0x00000003
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_release_fence_scope_shift  11
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_RELEASE_FENCE_SCOPE(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_HEADER_release_fence_scope_mask) << SDMA_AQL_PKT_BARRIER_OR_HEADER_release_fence_scope_shift)
+
+/*define for reserved field*/
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_reserved_offset 0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_reserved_mask   0x00000007
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_reserved_shift  13
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_RESERVED(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_HEADER_reserved_mask) << SDMA_AQL_PKT_BARRIER_OR_HEADER_reserved_shift)
+
+/*define for op field*/
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_op_offset 0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_op_mask   0x0000000F
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_op_shift  16
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_OP(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_HEADER_op_mask) << SDMA_AQL_PKT_BARRIER_OR_HEADER_op_shift)
+
+/*define for subop field*/
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_subop_offset 0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_subop_mask   0x00000007
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_subop_shift  20
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_SUBOP(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_HEADER_subop_mask) << SDMA_AQL_PKT_BARRIER_OR_HEADER_subop_shift)
+
+/*define for cpv field*/
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_cpv_offset 0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_cpv_mask   0x00000001
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_cpv_shift  28
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_CPV(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_HEADER_cpv_mask) << SDMA_AQL_PKT_BARRIER_OR_HEADER_cpv_shift)
+
+/*define for RESERVED_DW1 word*/
+/*define for reserved_dw1 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW1_reserved_dw1_offset 1
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW1_reserved_dw1_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW1_reserved_dw1_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW1_RESERVED_DW1(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW1_reserved_dw1_mask) << SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW1_reserved_dw1_shift)
+
+/*define for DEPENDENT_ADDR_0_LO word*/
+/*define for dependent_addr_0_31_0 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_LO_dependent_addr_0_31_0_offset 2
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_LO_dependent_addr_0_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_LO_dependent_addr_0_31_0_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_LO_DEPENDENT_ADDR_0_31_0(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_LO_dependent_addr_0_31_0_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_LO_dependent_addr_0_31_0_shift)
+
+/*define for DEPENDENT_ADDR_0_HI word*/
+/*define for dependent_addr_0_63_32 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_HI_dependent_addr_0_63_32_offset 3
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_HI_dependent_addr_0_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_HI_dependent_addr_0_63_32_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_HI_DEPENDENT_ADDR_0_63_32(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_HI_dependent_addr_0_63_32_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_HI_dependent_addr_0_63_32_shift)
+
+/*define for DEPENDENT_ADDR_1_LO word*/
+/*define for dependent_addr_1_31_0 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_LO_dependent_addr_1_31_0_offset 4
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_LO_dependent_addr_1_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_LO_dependent_addr_1_31_0_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_LO_DEPENDENT_ADDR_1_31_0(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_LO_dependent_addr_1_31_0_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_LO_dependent_addr_1_31_0_shift)
+
+/*define for DEPENDENT_ADDR_1_HI word*/
+/*define for dependent_addr_1_63_32 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_HI_dependent_addr_1_63_32_offset 5
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_HI_dependent_addr_1_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_HI_dependent_addr_1_63_32_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_HI_DEPENDENT_ADDR_1_63_32(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_HI_dependent_addr_1_63_32_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_HI_dependent_addr_1_63_32_shift)
+
+/*define for DEPENDENT_ADDR_2_LO word*/
+/*define for dependent_addr_2_31_0 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_LO_dependent_addr_2_31_0_offset 6
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_LO_dependent_addr_2_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_LO_dependent_addr_2_31_0_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_LO_DEPENDENT_ADDR_2_31_0(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_LO_dependent_addr_2_31_0_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_LO_dependent_addr_2_31_0_shift)
+
+/*define for DEPENDENT_ADDR_2_HI word*/
+/*define for dependent_addr_2_63_32 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_HI_dependent_addr_2_63_32_offset 7
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_HI_dependent_addr_2_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_HI_dependent_addr_2_63_32_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_HI_DEPENDENT_ADDR_2_63_32(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_HI_dependent_addr_2_63_32_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_HI_dependent_addr_2_63_32_shift)
+
+/*define for DEPENDENT_ADDR_3_LO word*/
+/*define for dependent_addr_3_31_0 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_LO_dependent_addr_3_31_0_offset 8
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_LO_dependent_addr_3_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_LO_dependent_addr_3_31_0_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_LO_DEPENDENT_ADDR_3_31_0(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_LO_dependent_addr_3_31_0_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_LO_dependent_addr_3_31_0_shift)
+
+/*define for DEPENDENT_ADDR_3_HI word*/
+/*define for dependent_addr_3_63_32 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_HI_dependent_addr_3_63_32_offset 9
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_HI_dependent_addr_3_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_HI_dependent_addr_3_63_32_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_HI_DEPENDENT_ADDR_3_63_32(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_HI_dependent_addr_3_63_32_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_HI_dependent_addr_3_63_32_shift)
+
+/*define for DEPENDENT_ADDR_4_LO word*/
+/*define for dependent_addr_4_31_0 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_LO_dependent_addr_4_31_0_offset 10
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_LO_dependent_addr_4_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_LO_dependent_addr_4_31_0_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_LO_DEPENDENT_ADDR_4_31_0(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_LO_dependent_addr_4_31_0_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_LO_dependent_addr_4_31_0_shift)
+
+/*define for DEPENDENT_ADDR_4_HI word*/
+/*define for dependent_addr_4_63_32 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_HI_dependent_addr_4_63_32_offset 11
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_HI_dependent_addr_4_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_HI_dependent_addr_4_63_32_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_HI_DEPENDENT_ADDR_4_63_32(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_HI_dependent_addr_4_63_32_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_HI_dependent_addr_4_63_32_shift)
+
+/*define for CACHE_POLICY word*/
+/*define for cache_policy0 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy0_offset 12
+#define SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy0_mask   0x00000007
+#define SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy0_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_CACHE_POLICY0(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy0_mask) << SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy0_shift)
+
+/*define for cache_policy1 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy1_offset 12
+#define SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy1_mask   0x00000007
+#define SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy1_shift  5
+#define SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_CACHE_POLICY1(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy1_mask) << SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy1_shift)
+
+/*define for cache_policy2 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy2_offset 12
+#define SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy2_mask   0x00000007
+#define SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy2_shift  10
+#define SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_CACHE_POLICY2(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy2_mask) << SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy2_shift)
+
+/*define for cache_policy3 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy3_offset 12
+#define SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy3_mask   0x00000007
+#define SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy3_shift  15
+#define SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_CACHE_POLICY3(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy3_mask) << SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy3_shift)
+
+/*define for cache_policy4 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy4_offset 12
+#define SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy4_mask   0x00000007
+#define SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy4_shift  20
+#define SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_CACHE_POLICY4(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy4_mask) << SDMA_AQL_PKT_BARRIER_OR_CACHE_POLICY_cache_policy4_shift)
+
+/*define for RESERVED_DW13 word*/
+/*define for reserved_dw13 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW13_reserved_dw13_offset 13
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW13_reserved_dw13_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW13_reserved_dw13_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW13_RESERVED_DW13(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW13_reserved_dw13_mask) << SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW13_reserved_dw13_shift)
+
+/*define for COMPLETION_SIGNAL_LO word*/
+/*define for completion_signal_31_0 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_LO_completion_signal_31_0_offset 14
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_LO_completion_signal_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_LO_completion_signal_31_0_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_LO_COMPLETION_SIGNAL_31_0(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_LO_completion_signal_31_0_mask) << SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_LO_completion_signal_31_0_shift)
+
+/*define for COMPLETION_SIGNAL_HI word*/
+/*define for completion_signal_63_32 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_HI_completion_signal_63_32_offset 15
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_HI_completion_signal_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_HI_completion_signal_63_32_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_HI_COMPLETION_SIGNAL_63_32(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_HI_completion_signal_63_32_mask) << SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_HI_completion_signal_63_32_shift)
+
+
+#endif /* __SDMA_V6_0_0_PKT_OPEN_H_ */
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c
new file mode 100644
index 000000000000..326ecc8d37d2
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c
@@ -0,0 +1,1858 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/delay.h>
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+
+#include "amdgpu.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_trace.h"
+
+#include "gc/gc_12_0_0_offset.h"
+#include "gc/gc_12_0_0_sh_mask.h"
+#include "hdp/hdp_6_0_0_offset.h"
+#include "ivsrcid/gfx/irqsrcs_gfx_12_0_0.h"
+
+#include "soc15_common.h"
+#include "soc15.h"
+#include "sdma_v6_0_0_pkt_open.h"
+#include "nbio_v4_3.h"
+#include "sdma_common.h"
+#include "sdma_v7_0.h"
+#include "v12_structs.h"
+#include "mes_userqueue.h"
+#include "amdgpu_userq_fence.h"
+
+MODULE_FIRMWARE("amdgpu/sdma_7_0_0.bin");
+MODULE_FIRMWARE("amdgpu/sdma_7_0_1.bin");
+
+#define SDMA1_REG_OFFSET 0x600
+#define SDMA0_HYP_DEC_REG_START 0x5880
+#define SDMA0_HYP_DEC_REG_END 0x589a
+#define SDMA1_HYP_DEC_REG_OFFSET 0x20
+
+/*define for compression field for sdma7*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_compress_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_compress_mask   0x00000001
+#define SDMA_PKT_CONSTANT_FILL_HEADER_compress_shift  16
+#define SDMA_PKT_CONSTANT_FILL_HEADER_COMPRESS(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_compress_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_compress_shift)
+
+static const struct amdgpu_hwip_reg_entry sdma_reg_list_7_0[] = {
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_STATUS_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_STATUS1_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_STATUS2_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_STATUS3_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_STATUS4_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_STATUS5_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_STATUS6_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_UCODE_REV),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_RB_RPTR_FETCH_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_RB_RPTR_FETCH),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_UTCL1_RD_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_UTCL1_WR_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_UTCL1_RD_XNACK0),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_UTCL1_RD_XNACK1),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_UTCL1_WR_XNACK0),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_UTCL1_WR_XNACK1),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_RB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_RB_RPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_RB_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_RB_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_IB_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_IB_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_IB_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_IB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_IB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_IB_SUB_REMAIN),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_DUMMY_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE_STATUS0),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_RB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_RB_RPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_RB_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_RB_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_IB_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_IB_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_IB_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_IB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_IB_SUB_REMAIN),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_DUMMY_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_RB_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_RB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_RB_RPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_RB_WPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_RB_WPTR_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_IB_OFFSET),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_IB_BASE_LO),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_IB_BASE_HI),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_IB_RPTR),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_IB_SUB_REMAIN),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_DUMMY_REG),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_INT_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_VM_CNTL),
+	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS2),
+	SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_CHICKEN_BITS),
+};
+
+static void sdma_v7_0_set_ring_funcs(struct amdgpu_device *adev);
+static void sdma_v7_0_set_buffer_funcs(struct amdgpu_device *adev);
+static void sdma_v7_0_set_vm_pte_funcs(struct amdgpu_device *adev);
+static void sdma_v7_0_set_irq_funcs(struct amdgpu_device *adev);
+static int sdma_v7_0_start(struct amdgpu_device *adev);
+
+static u32 sdma_v7_0_get_reg_offset(struct amdgpu_device *adev, u32 instance, u32 internal_offset)
+{
+	u32 base;
+
+	if (internal_offset >= SDMA0_HYP_DEC_REG_START &&
+	    internal_offset <= SDMA0_HYP_DEC_REG_END) {
+		base = adev->reg_offset[GC_HWIP][0][1];
+		if (instance != 0)
+			internal_offset += SDMA1_HYP_DEC_REG_OFFSET * instance;
+	} else {
+		base = adev->reg_offset[GC_HWIP][0][0];
+		if (instance == 1)
+			internal_offset += SDMA1_REG_OFFSET;
+	}
+
+	return base + internal_offset;
+}
+
+static unsigned sdma_v7_0_ring_init_cond_exec(struct amdgpu_ring *ring,
+					      uint64_t addr)
+{
+	unsigned ret;
+
+	amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_COND_EXE));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, 1);
+	/* this is the offset we need patch later */
+	ret = ring->wptr & ring->buf_mask;
+	/* insert dummy here and patch it later */
+	amdgpu_ring_write(ring, 0);
+
+	return ret;
+}
+
+/**
+ * sdma_v7_0_ring_get_rptr - get the current read pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current rptr from the hardware.
+ */
+static uint64_t sdma_v7_0_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	u64 *rptr;
+
+	/* XXX check if swapping is necessary on BE */
+	rptr = (u64 *)ring->rptr_cpu_addr;
+
+	DRM_DEBUG("rptr before shift == 0x%016llx\n", *rptr);
+	return ((*rptr) >> 2);
+}
+
+/**
+ * sdma_v7_0_ring_get_wptr - get the current write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current wptr from the hardware.
+ */
+static uint64_t sdma_v7_0_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	u64 wptr = 0;
+
+	if (ring->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		wptr = READ_ONCE(*((u64 *)ring->wptr_cpu_addr));
+		DRM_DEBUG("wptr/doorbell before shift == 0x%016llx\n", wptr);
+	}
+
+	return wptr >> 2;
+}
+
+/**
+ * sdma_v7_0_ring_set_wptr - commit the write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Write the wptr back to the hardware.
+ */
+static void sdma_v7_0_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	DRM_DEBUG("Setting write pointer\n");
+
+	if (ring->use_doorbell) {
+		DRM_DEBUG("Using doorbell -- "
+			  "wptr_offs == 0x%08x "
+			  "lower_32_bits(ring->wptr) << 2 == 0x%08x "
+			  "upper_32_bits(ring->wptr) << 2 == 0x%08x\n",
+			  ring->wptr_offs,
+			  lower_32_bits(ring->wptr << 2),
+			  upper_32_bits(ring->wptr << 2));
+		/* XXX check if swapping is necessary on BE */
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr,
+			     ring->wptr << 2);
+		DRM_DEBUG("calling WDOORBELL64(0x%08x, 0x%016llx)\n",
+			  ring->doorbell_index, ring->wptr << 2);
+		WDOORBELL64(ring->doorbell_index, ring->wptr << 2);
+	} else {
+		DRM_DEBUG("Not using doorbell -- "
+			  "regSDMA%i_GFX_RB_WPTR == 0x%08x "
+			  "regSDMA%i_GFX_RB_WPTR_HI == 0x%08x\n",
+			  ring->me,
+			  lower_32_bits(ring->wptr << 2),
+			  ring->me,
+			  upper_32_bits(ring->wptr << 2));
+		WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev,
+							     ring->me,
+							     regSDMA0_QUEUE0_RB_WPTR),
+				lower_32_bits(ring->wptr << 2));
+		WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev,
+							     ring->me,
+							     regSDMA0_QUEUE0_RB_WPTR_HI),
+				upper_32_bits(ring->wptr << 2));
+	}
+}
+
+static void sdma_v7_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring);
+	int i;
+
+	for (i = 0; i < count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			amdgpu_ring_write(ring, ring->funcs->nop |
+				SDMA_PKT_NOP_HEADER_COUNT(count - 1));
+		else
+			amdgpu_ring_write(ring, ring->funcs->nop);
+}
+
+/**
+ * sdma_v7_0_ring_emit_ib - Schedule an IB on the DMA engine
+ *
+ * @ring: amdgpu ring pointer
+ * @job: job to retrieve vmid from
+ * @ib: IB object to schedule
+ * @flags: unused
+ *
+ * Schedule an IB in the DMA ring.
+ */
+static void sdma_v7_0_ring_emit_ib(struct amdgpu_ring *ring,
+				   struct amdgpu_job *job,
+				   struct amdgpu_ib *ib,
+				   uint32_t flags)
+{
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+	uint64_t csa_mc_addr = amdgpu_sdma_get_csa_mc_addr(ring, vmid);
+
+	/* An IB packet must end on a 8 DW boundary--the next dword
+	 * must be on a 8-dword boundary. Our IB packet below is 6
+	 * dwords long, thus add x number of NOPs, such that, in
+	 * modular arithmetic,
+	 * wptr + 6 + x = 8k, k >= 0, which in C is,
+	 * (wptr + 6 + x) % 8 = 0.
+	 * The expression below, is a solution of x.
+	 */
+	sdma_v7_0_ring_insert_nop(ring, (2 - lower_32_bits(ring->wptr)) & 7);
+
+	amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_INDIRECT) |
+			  SDMA_PKT_INDIRECT_HEADER_VMID(vmid & 0xf));
+	/* base must be 32 byte aligned */
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr) & 0xffffffe0);
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, ib->length_dw);
+	amdgpu_ring_write(ring, lower_32_bits(csa_mc_addr));
+	amdgpu_ring_write(ring, upper_32_bits(csa_mc_addr));
+}
+
+/**
+ * sdma_v7_0_ring_emit_mem_sync - flush the IB by graphics cache rinse
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * flush the IB by graphics cache rinse.
+ */
+static void sdma_v7_0_ring_emit_mem_sync(struct amdgpu_ring *ring)
+{
+	uint32_t gcr_cntl = SDMA_GCR_GL2_INV | SDMA_GCR_GL2_WB | SDMA_GCR_GLM_INV |
+		SDMA_GCR_GL1_INV | SDMA_GCR_GLV_INV | SDMA_GCR_GLK_INV |
+		SDMA_GCR_GLI_INV(1);
+
+	/* flush entire cache L0/L1/L2, this can be optimized by performance requirement */
+	amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_GCR_REQ));
+	amdgpu_ring_write(ring, SDMA_PKT_GCR_REQ_PAYLOAD1_BASE_VA_31_7(0));
+	amdgpu_ring_write(ring, SDMA_PKT_GCR_REQ_PAYLOAD2_GCR_CONTROL_15_0(gcr_cntl) |
+			  SDMA_PKT_GCR_REQ_PAYLOAD2_BASE_VA_47_32(0));
+	amdgpu_ring_write(ring, SDMA_PKT_GCR_REQ_PAYLOAD3_LIMIT_VA_31_7(0) |
+			  SDMA_PKT_GCR_REQ_PAYLOAD3_GCR_CONTROL_18_16(gcr_cntl >> 16));
+	amdgpu_ring_write(ring, SDMA_PKT_GCR_REQ_PAYLOAD4_LIMIT_VA_47_32(0) |
+			  SDMA_PKT_GCR_REQ_PAYLOAD4_VMID(0));
+}
+
+
+/**
+ * sdma_v7_0_ring_emit_hdp_flush - emit an hdp flush on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Emit an hdp flush packet on the requested DMA ring.
+ */
+static void sdma_v7_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 ref_and_mask = 0;
+	const struct nbio_hdp_flush_reg *nbio_hf_reg = adev->nbio.hdp_flush_reg;
+
+	ref_and_mask = nbio_hf_reg->ref_and_mask_sdma0 << ring->me;
+
+	amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(1) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3)); /* == */
+	amdgpu_ring_write(ring, (adev->nbio.funcs->get_hdp_flush_done_offset(adev)) << 2);
+	amdgpu_ring_write(ring, (adev->nbio.funcs->get_hdp_flush_req_offset(adev)) << 2);
+	amdgpu_ring_write(ring, ref_and_mask); /* reference */
+	amdgpu_ring_write(ring, ref_and_mask); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
+}
+
+/**
+ * sdma_v7_0_ring_emit_fence - emit a fence on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ * @addr: address
+ * @seq: fence seq number
+ * @flags: fence flags
+ *
+ * Add a DMA fence packet to the ring to write
+ * the fence seq number and DMA trap packet to generate
+ * an interrupt if needed.
+ */
+static void sdma_v7_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				      unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	/* write the fence */
+	amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_FENCE) |
+			  SDMA_PKT_FENCE_HEADER_MTYPE(0x3)); /* Ucached(UC) */
+	/* zero in first two bits */
+	BUG_ON(addr & 0x3);
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+
+	/* optionally write high bits as well */
+	if (write64bit) {
+		addr += 4;
+		amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_FENCE) |
+				  SDMA_PKT_FENCE_HEADER_MTYPE(0x3));
+		/* zero in first two bits */
+		BUG_ON(addr & 0x3);
+		amdgpu_ring_write(ring, lower_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(seq));
+	}
+
+	if (flags & AMDGPU_FENCE_FLAG_INT) {
+		/* generate an interrupt */
+		amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_TRAP));
+		amdgpu_ring_write(ring, SDMA_PKT_TRAP_INT_CONTEXT_INT_CONTEXT(0));
+	}
+}
+
+/**
+ * sdma_v7_0_gfx_stop - stop the gfx async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the gfx async dma ring buffers.
+ */
+static void sdma_v7_0_gfx_stop(struct amdgpu_device *adev)
+{
+	u32 rb_cntl, ib_cntl;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		rb_cntl = RREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_CNTL));
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RB_ENABLE, 0);
+		WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_CNTL), rb_cntl);
+		ib_cntl = RREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_IB_CNTL));
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_QUEUE0_IB_CNTL, IB_ENABLE, 0);
+		WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_IB_CNTL), ib_cntl);
+	}
+}
+
+/**
+ * sdma_v7_0_rlc_stop - stop the compute async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the compute async dma queues.
+ */
+static void sdma_v7_0_rlc_stop(struct amdgpu_device *adev)
+{
+	/* XXX todo */
+}
+
+/**
+ * sdma_v7_0_ctx_switch_enable - stop the async dma engines context switch
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable the DMA MEs context switch.
+ *
+ * Halt or unhalt the async dma engines context switch.
+ */
+static void sdma_v7_0_ctx_switch_enable(struct amdgpu_device *adev, bool enable)
+{
+}
+
+/**
+ * sdma_v7_0_enable - stop the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable the DMA MEs.
+ *
+ * Halt or unhalt the async dma engines.
+ */
+static void sdma_v7_0_enable(struct amdgpu_device *adev, bool enable)
+{
+	u32 mcu_cntl;
+	int i;
+
+	if (!enable) {
+		sdma_v7_0_gfx_stop(adev);
+		sdma_v7_0_rlc_stop(adev);
+	}
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		mcu_cntl = RREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_MCU_CNTL));
+		mcu_cntl = REG_SET_FIELD(mcu_cntl, SDMA0_MCU_CNTL, HALT, enable ? 0 : 1);
+		WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_MCU_CNTL), mcu_cntl);
+	}
+}
+
+/**
+ * sdma_v7_0_gfx_resume_instance - start/restart a certain sdma engine
+ *
+ * @adev: amdgpu_device pointer
+ * @i: instance
+ * @restore: used to restore wptr when restart
+ *
+ * Set up the gfx DMA ring buffers and enable them. On restart, we will restore wptr and rptr.
+ * Return 0 for success.
+ */
+static int sdma_v7_0_gfx_resume_instance(struct amdgpu_device *adev, int i, bool restore)
+{
+	struct amdgpu_ring *ring;
+	u32 rb_cntl, ib_cntl;
+	u32 rb_bufsz;
+	u32 doorbell;
+	u32 doorbell_offset;
+	u32 temp;
+	u64 wptr_gpu_addr;
+	int r;
+
+	ring = &adev->sdma.instance[i].ring;
+
+	/* Set ring buffer size in dwords */
+	rb_bufsz = order_base_2(ring->ring_size / 4);
+	rb_cntl = RREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_CNTL));
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RB_SIZE, rb_bufsz);
+#ifdef __BIG_ENDIAN
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RB_SWAP_ENABLE, 1);
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL,
+				RPTR_WRITEBACK_SWAP_ENABLE, 1);
+#endif
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RB_PRIV, 1);
+	WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_CNTL), rb_cntl);
+
+	/* Initialize the ring buffer's read and write pointers */
+	if (restore) {
+		WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_RPTR), lower_32_bits(ring->wptr << 2));
+		WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_RPTR_HI), upper_32_bits(ring->wptr << 2));
+		WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR), lower_32_bits(ring->wptr << 2));
+		WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR_HI), upper_32_bits(ring->wptr << 2));
+	} else {
+		WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_RPTR), 0);
+		WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_RPTR_HI), 0);
+		WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR), 0);
+		WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR_HI), 0);
+	}
+	/* setup the wptr shadow polling */
+	wptr_gpu_addr = ring->wptr_gpu_addr;
+	WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR_POLL_ADDR_LO),
+	       lower_32_bits(wptr_gpu_addr));
+	WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR_POLL_ADDR_HI),
+	       upper_32_bits(wptr_gpu_addr));
+
+	/* set the wb address whether it's enabled or not */
+	WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_RPTR_ADDR_HI),
+	       upper_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFF);
+	WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_RPTR_ADDR_LO),
+	       lower_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFC);
+
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RPTR_WRITEBACK_ENABLE, 1);
+	if (amdgpu_sriov_vf(adev))
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, WPTR_POLL_ENABLE, 1);
+	else
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, WPTR_POLL_ENABLE, 0);
+
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, MCU_WPTR_POLL_ENABLE, 1);
+
+	WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_BASE), ring->gpu_addr >> 8);
+	WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_BASE_HI), ring->gpu_addr >> 40);
+
+	if (!restore)
+		ring->wptr = 0;
+
+	/* before programing wptr to a less value, need set minor_ptr_update first */
+	WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_MINOR_PTR_UPDATE), 1);
+
+	if (!amdgpu_sriov_vf(adev)) { /* only bare-metal use register write for wptr */
+		WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR), lower_32_bits(ring->wptr) << 2);
+		WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR_HI), upper_32_bits(ring->wptr) << 2);
+	}
+
+	doorbell = RREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_DOORBELL));
+	doorbell_offset = RREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_DOORBELL_OFFSET));
+
+	if (ring->use_doorbell) {
+		doorbell = REG_SET_FIELD(doorbell, SDMA0_QUEUE0_DOORBELL, ENABLE, 1);
+		doorbell_offset = REG_SET_FIELD(doorbell_offset, SDMA0_QUEUE0_DOORBELL_OFFSET,
+				OFFSET, ring->doorbell_index);
+	} else {
+		doorbell = REG_SET_FIELD(doorbell, SDMA0_QUEUE0_DOORBELL, ENABLE, 0);
+	}
+	WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_DOORBELL), doorbell);
+	WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_DOORBELL_OFFSET), doorbell_offset);
+
+	if (i == 0)
+		adev->nbio.funcs->sdma_doorbell_range(adev, i, ring->use_doorbell,
+					      ring->doorbell_index,
+					      adev->doorbell_index.sdma_doorbell_range * adev->sdma.num_instances);
+
+	if (amdgpu_sriov_vf(adev))
+		sdma_v7_0_ring_set_wptr(ring);
+
+	/* set minor_ptr_update to 0 after wptr programed */
+	WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_MINOR_PTR_UPDATE), 0);
+
+	/* Set up sdma hang watchdog */
+	temp = RREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_WATCHDOG_CNTL));
+	/* 100ms per unit */
+	temp = REG_SET_FIELD(temp, SDMA0_WATCHDOG_CNTL, QUEUE_HANG_COUNT,
+			     max(adev->usec_timeout/100000, 1));
+	WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_WATCHDOG_CNTL), temp);
+
+	/* Set up RESP_MODE to non-copy addresses */
+	temp = RREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_UTCL1_CNTL));
+	temp = REG_SET_FIELD(temp, SDMA0_UTCL1_CNTL, RESP_MODE, 3);
+	temp = REG_SET_FIELD(temp, SDMA0_UTCL1_CNTL, REDO_DELAY, 9);
+	WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_UTCL1_CNTL), temp);
+
+	/* program default cache read and write policy */
+	temp = RREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_UTCL1_PAGE));
+	/* clean read policy and write policy bits */
+	temp &= 0xFF0FFF;
+	temp |= ((CACHE_READ_POLICY_L2__DEFAULT << 12) |
+		 (CACHE_WRITE_POLICY_L2__DEFAULT << 14));
+	WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_UTCL1_PAGE), temp);
+
+	if (!amdgpu_sriov_vf(adev)) {
+		/* unhalt engine */
+		temp = RREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_MCU_CNTL));
+		temp = REG_SET_FIELD(temp, SDMA0_MCU_CNTL, HALT, 0);
+		temp = REG_SET_FIELD(temp, SDMA0_MCU_CNTL, RESET, 0);
+		WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_MCU_CNTL), temp);
+	}
+
+	/* enable DMA RB */
+	rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RB_ENABLE, 1);
+	WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_CNTL), rb_cntl);
+
+	ib_cntl = RREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_IB_CNTL));
+	ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_QUEUE0_IB_CNTL, IB_ENABLE, 1);
+#ifdef __BIG_ENDIAN
+	ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_QUEUE0_IB_CNTL, IB_SWAP_ENABLE, 1);
+#endif
+	/* enable DMA IBs */
+	WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_IB_CNTL), ib_cntl);
+	ring->sched.ready = true;
+
+	if (amdgpu_sriov_vf(adev)) { /* bare-metal sequence doesn't need below to lines */
+		sdma_v7_0_ctx_switch_enable(adev, true);
+		sdma_v7_0_enable(adev, true);
+	}
+
+	r = amdgpu_ring_test_helper(ring);
+	if (r)
+		ring->sched.ready = false;
+
+	return r;
+}
+
+/**
+ * sdma_v7_0_gfx_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the gfx DMA ring buffers and enable them.
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v7_0_gfx_resume(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		r = sdma_v7_0_gfx_resume_instance(adev, i, false);
+		if (r)
+			return r;
+	}
+
+	return 0;
+
+}
+
+/**
+ * sdma_v7_0_rlc_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the compute DMA queues and enable them.
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v7_0_rlc_resume(struct amdgpu_device *adev)
+{
+	return 0;
+}
+
+static void sdma_v12_0_free_ucode_buffer(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		amdgpu_bo_free_kernel(&adev->sdma.instance[i].sdma_fw_obj,
+				      &adev->sdma.instance[i].sdma_fw_gpu_addr,
+				      (void **)&adev->sdma.instance[i].sdma_fw_ptr);
+	}
+}
+
+/**
+ * sdma_v7_0_load_microcode - load the sDMA ME ucode
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Loads the sDMA0/1 ucode.
+ * Returns 0 for success, -EINVAL if the ucode is not available.
+ */
+static int sdma_v7_0_load_microcode(struct amdgpu_device *adev)
+{
+	const struct sdma_firmware_header_v3_0 *hdr;
+	const __le32 *fw_data;
+	u32 fw_size;
+	uint32_t tmp, sdma_status, ic_op_cntl;
+	int i, r, j;
+
+	/* halt the MEs */
+	sdma_v7_0_enable(adev, false);
+
+	if (!adev->sdma.instance[0].fw)
+		return -EINVAL;
+
+	hdr = (const struct sdma_firmware_header_v3_0 *)
+		adev->sdma.instance[0].fw->data;
+	amdgpu_ucode_print_sdma_hdr(&hdr->header);
+
+	fw_data = (const __le32 *)(adev->sdma.instance[0].fw->data +
+			le32_to_cpu(hdr->ucode_offset_bytes));
+	fw_size = le32_to_cpu(hdr->ucode_size_bytes);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		r = amdgpu_bo_create_reserved(adev, fw_size,
+					      PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_VRAM,
+					      &adev->sdma.instance[i].sdma_fw_obj,
+					      &adev->sdma.instance[i].sdma_fw_gpu_addr,
+					      (void **)&adev->sdma.instance[i].sdma_fw_ptr);
+		if (r) {
+			dev_err(adev->dev, "(%d) failed to create sdma ucode bo\n", r);
+			return r;
+		}
+
+		memcpy(adev->sdma.instance[i].sdma_fw_ptr, fw_data, fw_size);
+
+		amdgpu_bo_kunmap(adev->sdma.instance[i].sdma_fw_obj);
+		amdgpu_bo_unreserve(adev->sdma.instance[i].sdma_fw_obj);
+
+		tmp = RREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_IC_CNTL));
+		tmp = REG_SET_FIELD(tmp, SDMA0_IC_CNTL, GPA, 0);
+		WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_IC_CNTL), tmp);
+
+		WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_IC_BASE_LO),
+			lower_32_bits(adev->sdma.instance[i].sdma_fw_gpu_addr));
+		WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_IC_BASE_HI),
+			upper_32_bits(adev->sdma.instance[i].sdma_fw_gpu_addr));
+
+		tmp = RREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_IC_OP_CNTL));
+		tmp = REG_SET_FIELD(tmp, SDMA0_IC_OP_CNTL, PRIME_ICACHE, 1);
+		WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_IC_OP_CNTL), tmp);
+
+		/* Wait for sdma ucode init complete */
+		for (j = 0; j < adev->usec_timeout; j++) {
+			ic_op_cntl = RREG32_SOC15_IP(GC,
+					sdma_v7_0_get_reg_offset(adev, i, regSDMA0_IC_OP_CNTL));
+			sdma_status = RREG32_SOC15_IP(GC,
+					sdma_v7_0_get_reg_offset(adev, i, regSDMA0_STATUS_REG));
+			if ((REG_GET_FIELD(ic_op_cntl, SDMA0_IC_OP_CNTL, ICACHE_PRIMED) == 1) &&
+			    (REG_GET_FIELD(sdma_status, SDMA0_STATUS_REG, UCODE_INIT_DONE) == 1))
+				break;
+			udelay(1);
+		}
+
+		if (j >= adev->usec_timeout) {
+			dev_err(adev->dev, "failed to init sdma ucode\n");
+			return -EINVAL;
+		}
+	}
+
+	return 0;
+}
+
+static int sdma_v7_0_soft_reset(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	u32 tmp;
+	int i;
+
+	sdma_v7_0_gfx_stop(adev);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		//tmp = RREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_FREEZE));
+		//tmp |= SDMA0_FREEZE__FREEZE_MASK;
+		//WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_FREEZE), tmp);
+		tmp = RREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_MCU_CNTL));
+		tmp |= SDMA0_MCU_CNTL__HALT_MASK;
+		tmp |= SDMA0_MCU_CNTL__RESET_MASK;
+		WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_MCU_CNTL), tmp);
+
+		WREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_PREEMPT), 0);
+
+		udelay(100);
+
+		tmp = GRBM_SOFT_RESET__SOFT_RESET_SDMA0_MASK << i;
+		WREG32_SOC15(GC, 0, regGRBM_SOFT_RESET, tmp);
+		tmp = RREG32_SOC15(GC, 0, regGRBM_SOFT_RESET);
+
+		udelay(100);
+
+		WREG32_SOC15(GC, 0, regGRBM_SOFT_RESET, 0);
+		tmp = RREG32_SOC15(GC, 0, regGRBM_SOFT_RESET);
+
+		udelay(100);
+	}
+
+	return sdma_v7_0_start(adev);
+}
+
+static bool sdma_v7_0_check_soft_reset(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	int i, r;
+	long tmo = msecs_to_jiffies(1000);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		r = amdgpu_ring_test_ib(ring, tmo);
+		if (r)
+			return true;
+	}
+
+	return false;
+}
+
+static int sdma_v7_0_reset_queue(struct amdgpu_ring *ring,
+				 unsigned int vmid,
+				 struct amdgpu_fence *timedout_fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int r;
+
+	if (ring->me >= adev->sdma.num_instances) {
+		dev_err(adev->dev, "sdma instance not found\n");
+		return -EINVAL;
+	}
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+
+	r = amdgpu_mes_reset_legacy_queue(adev, ring, vmid, true);
+	if (r)
+		return r;
+
+	r = sdma_v7_0_gfx_resume_instance(adev, ring->me, true);
+	if (r)
+		return r;
+
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+/**
+ * sdma_v7_0_start - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the DMA engines and enable them.
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v7_0_start(struct amdgpu_device *adev)
+{
+	int r = 0;
+
+	if (amdgpu_sriov_vf(adev)) {
+		sdma_v7_0_ctx_switch_enable(adev, false);
+		sdma_v7_0_enable(adev, false);
+
+		/* set RB registers */
+		r = sdma_v7_0_gfx_resume(adev);
+		return r;
+	}
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+		r = sdma_v7_0_load_microcode(adev);
+		if (r) {
+			sdma_v12_0_free_ucode_buffer(adev);
+			return r;
+		}
+
+		if (amdgpu_emu_mode == 1)
+			msleep(1000);
+	}
+
+	/* unhalt the MEs */
+	sdma_v7_0_enable(adev, true);
+	/* enable sdma ring preemption */
+	sdma_v7_0_ctx_switch_enable(adev, true);
+
+	/* start the gfx rings and rlc compute queues */
+	r = sdma_v7_0_gfx_resume(adev);
+	if (r)
+		return r;
+	r = sdma_v7_0_rlc_resume(adev);
+
+	return r;
+}
+
+static int sdma_v7_0_mqd_init(struct amdgpu_device *adev, void *mqd,
+			      struct amdgpu_mqd_prop *prop)
+{
+	struct v12_sdma_mqd *m = mqd;
+	uint64_t wb_gpu_addr;
+
+	m->sdmax_rlcx_rb_cntl =
+		order_base_2(prop->queue_size / 4) << SDMA0_QUEUE0_RB_CNTL__RB_SIZE__SHIFT |
+		1 << SDMA0_QUEUE0_RB_CNTL__RPTR_WRITEBACK_ENABLE__SHIFT |
+		4 << SDMA0_QUEUE0_RB_CNTL__RPTR_WRITEBACK_TIMER__SHIFT |
+		1 << SDMA0_QUEUE0_RB_CNTL__MCU_WPTR_POLL_ENABLE__SHIFT;
+
+	m->sdmax_rlcx_rb_base = lower_32_bits(prop->hqd_base_gpu_addr >> 8);
+	m->sdmax_rlcx_rb_base_hi = upper_32_bits(prop->hqd_base_gpu_addr >> 8);
+
+	wb_gpu_addr = prop->wptr_gpu_addr;
+	m->sdmax_rlcx_rb_wptr_poll_addr_lo = lower_32_bits(wb_gpu_addr);
+	m->sdmax_rlcx_rb_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr);
+
+	wb_gpu_addr = prop->rptr_gpu_addr;
+	m->sdmax_rlcx_rb_rptr_addr_lo = lower_32_bits(wb_gpu_addr);
+	m->sdmax_rlcx_rb_rptr_addr_hi = upper_32_bits(wb_gpu_addr);
+
+	m->sdmax_rlcx_ib_cntl = RREG32_SOC15_IP(GC, sdma_v7_0_get_reg_offset(adev, 0,
+							regSDMA0_QUEUE0_IB_CNTL));
+
+	m->sdmax_rlcx_doorbell_offset =
+		prop->doorbell_index << SDMA0_QUEUE0_DOORBELL_OFFSET__OFFSET__SHIFT;
+
+	m->sdmax_rlcx_doorbell = REG_SET_FIELD(0, SDMA0_QUEUE0_DOORBELL, ENABLE, 1);
+
+	m->sdmax_rlcx_doorbell_log = 0;
+	m->sdmax_rlcx_rb_aql_cntl = 0x4000;	//regSDMA0_QUEUE0_RB_AQL_CNTL_DEFAULT;
+	m->sdmax_rlcx_dummy_reg = 0xf;	//regSDMA0_QUEUE0_DUMMY_REG_DEFAULT;
+
+	m->sdmax_rlcx_csa_addr_lo = lower_32_bits(prop->csa_addr);
+	m->sdmax_rlcx_csa_addr_hi = upper_32_bits(prop->csa_addr);
+
+	m->sdmax_rlcx_mcu_dbg0 = lower_32_bits(prop->fence_address);
+	m->sdmax_rlcx_mcu_dbg1 = upper_32_bits(prop->fence_address);
+
+	return 0;
+}
+
+static void sdma_v7_0_set_mqd_funcs(struct amdgpu_device *adev)
+{
+	adev->mqds[AMDGPU_HW_IP_DMA].mqd_size = sizeof(struct v12_sdma_mqd);
+	adev->mqds[AMDGPU_HW_IP_DMA].init_mqd = sdma_v7_0_mqd_init;
+}
+
+/**
+ * sdma_v7_0_ring_test_ring - simple async dma engine test
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Test the DMA engine by writing using it to write an
+ * value to memory.
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v7_0_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned i;
+	unsigned index;
+	int r;
+	u32 tmp;
+	u64 gpu_addr;
+
+	tmp = 0xCAFEDEAD;
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+
+	r = amdgpu_ring_alloc(ring, 5);
+	if (r) {
+		DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r);
+		amdgpu_device_wb_free(adev, index);
+		return r;
+	}
+
+	amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_WRITE) |
+			  SDMA_PKT_COPY_LINEAR_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR));
+	amdgpu_ring_write(ring, lower_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(0));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = le32_to_cpu(adev->wb.wb[index]);
+		if (tmp == 0xDEADBEEF)
+			break;
+		if (amdgpu_emu_mode == 1)
+			msleep(1);
+		else
+			udelay(1);
+	}
+
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+
+	amdgpu_device_wb_free(adev, index);
+
+	return r;
+}
+
+/**
+ * sdma_v7_0_ring_test_ib - test an IB on the DMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @timeout: timeout value in jiffies, or MAX_SCHEDULE_TIMEOUT
+ *
+ * Test a simple IB in the DMA ring.
+ * Returns 0 on success, error on failure.
+ */
+static int sdma_v7_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct dma_fence *f = NULL;
+	unsigned index;
+	long r;
+	u32 tmp = 0;
+	u64 gpu_addr;
+
+	tmp = 0xCAFEDEAD;
+	memset(&ib, 0, sizeof(ib));
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+
+	r = amdgpu_ib_get(adev, NULL, 256, AMDGPU_IB_POOL_DIRECT, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
+		goto err0;
+	}
+
+	ib.ptr[0] = SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_WRITE) |
+		SDMA_PKT_COPY_LINEAR_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR);
+	ib.ptr[1] = lower_32_bits(gpu_addr);
+	ib.ptr[2] = upper_32_bits(gpu_addr);
+	ib.ptr[3] = SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(0);
+	ib.ptr[4] = 0xDEADBEEF;
+	ib.ptr[5] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.ptr[6] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.ptr[7] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.length_dw = 8;
+
+	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
+	if (r)
+		goto err1;
+
+	r = dma_fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out\n");
+		r = -ETIMEDOUT;
+		goto err1;
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
+		goto err1;
+	}
+
+	tmp = le32_to_cpu(adev->wb.wb[index]);
+
+	if (tmp == 0xDEADBEEF)
+		r = 0;
+	else
+		r = -EINVAL;
+
+err1:
+	amdgpu_ib_free(&ib, NULL);
+	dma_fence_put(f);
+err0:
+	amdgpu_device_wb_free(adev, index);
+	return r;
+}
+
+
+/**
+ * sdma_v7_0_vm_copy_pte - update PTEs by copying them from the GART
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @src: src addr to copy from
+ * @count: number of page entries to update
+ *
+ * Update PTEs by copying them from the GART using sDMA.
+ */
+static void sdma_v7_0_vm_copy_pte(struct amdgpu_ib *ib,
+				  uint64_t pe, uint64_t src,
+				  unsigned count)
+{
+	unsigned bytes = count * 8;
+
+	ib->ptr[ib->length_dw++] = SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_COPY) |
+		SDMA_PKT_COPY_LINEAR_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR) |
+		SDMA_PKT_COPY_LINEAR_HEADER_CPV(1);
+
+	ib->ptr[ib->length_dw++] = bytes - 1;
+	ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+	ib->ptr[ib->length_dw++] = lower_32_bits(src);
+	ib->ptr[ib->length_dw++] = upper_32_bits(src);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+	ib->ptr[ib->length_dw++] = 0;
+
+}
+
+/**
+ * sdma_v7_0_vm_write_pte - update PTEs by writing them manually
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @value: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ *
+ * Update PTEs by writing them manually using sDMA.
+ */
+static void sdma_v7_0_vm_write_pte(struct amdgpu_ib *ib, uint64_t pe,
+				   uint64_t value, unsigned count,
+				   uint32_t incr)
+{
+	unsigned ndw = count * 2;
+
+	ib->ptr[ib->length_dw++] = SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_WRITE) |
+		SDMA_PKT_COPY_LINEAR_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+	ib->ptr[ib->length_dw++] = ndw - 1;
+	for (; ndw > 0; ndw -= 2) {
+		ib->ptr[ib->length_dw++] = lower_32_bits(value);
+		ib->ptr[ib->length_dw++] = upper_32_bits(value);
+		value += incr;
+	}
+}
+
+/**
+ * sdma_v7_0_vm_set_pte_pde - update the page tables using sDMA
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: access flags
+ *
+ * Update the page tables using sDMA.
+ */
+static void sdma_v7_0_vm_set_pte_pde(struct amdgpu_ib *ib,
+				     uint64_t pe,
+				     uint64_t addr, unsigned count,
+				     uint32_t incr, uint64_t flags)
+{
+	/* for physically contiguous pages (vram) */
+	ib->ptr[ib->length_dw++] = SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_PTEPDE);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe); /* dst addr */
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+	ib->ptr[ib->length_dw++] = lower_32_bits(flags); /* mask */
+	ib->ptr[ib->length_dw++] = upper_32_bits(flags);
+	ib->ptr[ib->length_dw++] = lower_32_bits(addr); /* value */
+	ib->ptr[ib->length_dw++] = upper_32_bits(addr);
+	ib->ptr[ib->length_dw++] = incr; /* increment size */
+	ib->ptr[ib->length_dw++] = 0;
+	ib->ptr[ib->length_dw++] = count - 1; /* number of entries */
+}
+
+/**
+ * sdma_v7_0_ring_pad_ib - pad the IB
+ *
+ * @ring: amdgpu ring pointer
+ * @ib: indirect buffer to fill with padding
+ *
+ * Pad the IB with NOPs to a boundary multiple of 8.
+ */
+static void sdma_v7_0_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring);
+	u32 pad_count;
+	int i;
+
+	pad_count = (-ib->length_dw) & 0x7;
+	for (i = 0; i < pad_count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			ib->ptr[ib->length_dw++] =
+				SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_NOP) |
+				SDMA_PKT_NOP_HEADER_COUNT(pad_count - 1);
+		else
+			ib->ptr[ib->length_dw++] =
+				SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_NOP);
+}
+
+/**
+ * sdma_v7_0_ring_emit_pipeline_sync - sync the pipeline
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Make sure all previous operations are completed (CIK).
+ */
+static void sdma_v7_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	/* wait for idle */
+	amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(0) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3) | /* equal */
+			  SDMA_PKT_POLL_REGMEM_HEADER_MEM_POLL(1));
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xffffffff);
+	amdgpu_ring_write(ring, seq); /* reference */
+	amdgpu_ring_write(ring, 0xffffffff); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(4)); /* retry count, poll interval */
+}
+
+/**
+ * sdma_v7_0_ring_emit_vm_flush - vm flush using sDMA
+ *
+ * @ring: amdgpu_ring pointer
+ * @vmid: vmid number to use
+ * @pd_addr: address
+ *
+ * Update the page table base and flush the VM TLB
+ * using sDMA.
+ */
+static void sdma_v7_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					 unsigned vmid, uint64_t pd_addr)
+{
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+}
+
+static void sdma_v7_0_ring_emit_wreg(struct amdgpu_ring *ring,
+				     uint32_t reg, uint32_t val)
+{
+	/* SRBM WRITE command will not support on sdma v7.
+	 * Use Register WRITE command instead, which OPCODE is same as SRBM WRITE
+	 */
+	amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_SRBM_WRITE));
+	amdgpu_ring_write(ring, reg << 2);
+	amdgpu_ring_write(ring, val);
+}
+
+static void sdma_v7_0_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+					 uint32_t val, uint32_t mask)
+{
+	amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(0) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3)); /* equal */
+	amdgpu_ring_write(ring, reg << 2);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, val); /* reference */
+	amdgpu_ring_write(ring, mask); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10));
+}
+
+static void sdma_v7_0_ring_emit_reg_write_reg_wait(struct amdgpu_ring *ring,
+						   uint32_t reg0, uint32_t reg1,
+						   uint32_t ref, uint32_t mask)
+{
+	amdgpu_ring_emit_wreg(ring, reg0, ref);
+	/* wait for a cycle to reset vm_inv_eng*_ack */
+	amdgpu_ring_emit_reg_wait(ring, reg0, 0, 0);
+	amdgpu_ring_emit_reg_wait(ring, reg1, mask, mask);
+}
+
+static int sdma_v7_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	switch (amdgpu_user_queue) {
+	case -1:
+	case 0:
+	default:
+		adev->sdma.no_user_submission = false;
+		adev->sdma.disable_uq = true;
+		break;
+	case 1:
+		adev->sdma.no_user_submission = false;
+		adev->sdma.disable_uq = false;
+		break;
+	case 2:
+		adev->sdma.no_user_submission = true;
+		adev->sdma.disable_uq = false;
+		break;
+	}
+
+	r = amdgpu_sdma_init_microcode(adev, 0, true);
+	if (r) {
+		DRM_ERROR("Failed to init sdma firmware!\n");
+		return r;
+	}
+
+	sdma_v7_0_set_ring_funcs(adev);
+	sdma_v7_0_set_buffer_funcs(adev);
+	sdma_v7_0_set_vm_pte_funcs(adev);
+	sdma_v7_0_set_irq_funcs(adev);
+	sdma_v7_0_set_mqd_funcs(adev);
+
+	return 0;
+}
+
+static int sdma_v7_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_ring *ring;
+	int r, i;
+	struct amdgpu_device *adev = ip_block->adev;
+	uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_7_0);
+	uint32_t *ptr;
+
+	/* SDMA trap event */
+	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GFX,
+			      GFX_12_0_0__SRCID__SDMA_TRAP,
+			      &adev->sdma.trap_irq);
+	if (r)
+		return r;
+
+	/* SDMA user fence event */
+	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GFX,
+			      GFX_12_0_0__SRCID__SDMA_FENCE,
+			      &adev->sdma.fence_irq);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		ring->ring_obj = NULL;
+		ring->use_doorbell = true;
+		ring->me = i;
+		ring->no_user_submission = adev->sdma.no_user_submission;
+
+		DRM_DEBUG("SDMA %d use_doorbell being set to: [%s]\n", i,
+				ring->use_doorbell?"true":"false");
+
+		ring->doorbell_index =
+			(adev->doorbell_index.sdma_engine[i] << 1); // get DWORD offset
+
+		ring->vm_hub = AMDGPU_GFXHUB(0);
+		sprintf(ring->name, "sdma%d", i);
+		r = amdgpu_ring_init(adev, ring, 1024,
+				     &adev->sdma.trap_irq,
+				     AMDGPU_SDMA_IRQ_INSTANCE0 + i,
+				     AMDGPU_RING_PRIO_DEFAULT, NULL);
+		if (r)
+			return r;
+	}
+
+	adev->sdma.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->sdma.instance[0].ring);
+	if (!amdgpu_sriov_vf(adev))
+		adev->sdma.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+
+	r = amdgpu_sdma_sysfs_reset_mask_init(adev);
+	if (r)
+		return r;
+	/* Allocate memory for SDMA IP Dump buffer */
+	ptr = kcalloc(adev->sdma.num_instances * reg_count, sizeof(uint32_t), GFP_KERNEL);
+	if (ptr)
+		adev->sdma.ip_dump = ptr;
+	else
+		DRM_ERROR("Failed to allocated memory for SDMA IP Dump\n");
+
+	switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) {
+	case IP_VERSION(7, 0, 0):
+	case IP_VERSION(7, 0, 1):
+		if ((adev->sdma.instance[0].fw_version >= 7966358) && !adev->sdma.disable_uq)
+			adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs;
+		break;
+	default:
+		break;
+	}
+
+	return r;
+}
+
+static int sdma_v7_0_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		amdgpu_ring_fini(&adev->sdma.instance[i].ring);
+
+	amdgpu_sdma_sysfs_reset_mask_fini(adev);
+	amdgpu_sdma_destroy_inst_ctx(adev, true);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT)
+		sdma_v12_0_free_ucode_buffer(adev);
+
+	kfree(adev->sdma.ip_dump);
+
+	return 0;
+}
+
+static int sdma_v7_0_set_userq_trap_interrupts(struct amdgpu_device *adev,
+					       bool enable)
+{
+	unsigned int irq_type;
+	int i, r;
+
+	if (adev->userq_funcs[AMDGPU_HW_IP_DMA]) {
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			irq_type = AMDGPU_SDMA_IRQ_INSTANCE0 + i;
+			if (enable)
+				r = amdgpu_irq_get(adev, &adev->sdma.trap_irq,
+						   irq_type);
+			else
+				r = amdgpu_irq_put(adev, &adev->sdma.trap_irq,
+						   irq_type);
+			if (r)
+				return r;
+		}
+	}
+
+	return 0;
+}
+
+static int sdma_v7_0_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	r = sdma_v7_0_start(adev);
+	if (r)
+		return r;
+
+	return sdma_v7_0_set_userq_trap_interrupts(adev, true);
+}
+
+static int sdma_v7_0_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	sdma_v7_0_ctx_switch_enable(adev, false);
+	sdma_v7_0_enable(adev, false);
+	sdma_v7_0_set_userq_trap_interrupts(adev, false);
+
+	return 0;
+}
+
+static int sdma_v7_0_suspend(struct amdgpu_ip_block *ip_block)
+{
+	return sdma_v7_0_hw_fini(ip_block);
+}
+
+static int sdma_v7_0_resume(struct amdgpu_ip_block *ip_block)
+{
+	return sdma_v7_0_hw_init(ip_block);
+}
+
+static bool sdma_v7_0_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	u32 i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		u32 tmp = RREG32(sdma_v7_0_get_reg_offset(adev, i, regSDMA0_STATUS_REG));
+
+		if (!(tmp & SDMA0_STATUS_REG__IDLE_MASK))
+			return false;
+	}
+
+	return true;
+}
+
+static int sdma_v7_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	unsigned i;
+	u32 sdma0, sdma1;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		sdma0 = RREG32(sdma_v7_0_get_reg_offset(adev, 0, regSDMA0_STATUS_REG));
+		sdma1 = RREG32(sdma_v7_0_get_reg_offset(adev, 1, regSDMA0_STATUS_REG));
+
+		if (sdma0 & sdma1 & SDMA0_STATUS_REG__IDLE_MASK)
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static int sdma_v7_0_ring_preempt_ib(struct amdgpu_ring *ring)
+{
+	int i, r = 0;
+	struct amdgpu_device *adev = ring->adev;
+	u32 index = 0;
+	u64 sdma_gfx_preempt;
+
+	amdgpu_sdma_get_index_from_ring(ring, &index);
+	sdma_gfx_preempt =
+		sdma_v7_0_get_reg_offset(adev, index, regSDMA0_QUEUE0_PREEMPT);
+
+	/* assert preemption condition */
+	amdgpu_ring_set_preempt_cond_exec(ring, false);
+
+	/* emit the trailing fence */
+	ring->trail_seq += 1;
+	r = amdgpu_ring_alloc(ring, 10);
+	if (r) {
+		DRM_ERROR("ring %d failed to be allocated \n", ring->idx);
+		return r;
+	}
+	sdma_v7_0_ring_emit_fence(ring, ring->trail_fence_gpu_addr,
+				  ring->trail_seq, 0);
+	amdgpu_ring_commit(ring);
+
+	/* assert IB preemption */
+	WREG32(sdma_gfx_preempt, 1);
+
+	/* poll the trailing fence */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (ring->trail_seq ==
+		    le32_to_cpu(*(ring->trail_fence_cpu_addr)))
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout) {
+		r = -EINVAL;
+		DRM_ERROR("ring %d failed to be preempted\n", ring->idx);
+	}
+
+	/* deassert IB preemption */
+	WREG32(sdma_gfx_preempt, 0);
+
+	/* deassert the preemption condition */
+	amdgpu_ring_set_preempt_cond_exec(ring, true);
+	return r;
+}
+
+static int sdma_v7_0_set_trap_irq_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	u32 sdma_cntl;
+
+	u32 reg_offset = sdma_v7_0_get_reg_offset(adev, type, regSDMA0_CNTL);
+
+	sdma_cntl = RREG32(reg_offset);
+	sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE,
+		       state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+	WREG32(reg_offset, sdma_cntl);
+
+	return 0;
+}
+
+static int sdma_v7_0_process_trap_irq(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	int instances, queue;
+
+	DRM_DEBUG("IH: SDMA trap\n");
+
+	queue = entry->ring_id & 0xf;
+	instances = (entry->ring_id & 0xf0) >> 4;
+	if (instances > 1) {
+		DRM_ERROR("IH: wrong ring_ID detected, as wrong sdma instance\n");
+		return -EINVAL;
+	}
+
+	switch (entry->client_id) {
+	case SOC21_IH_CLIENTID_GFX:
+		switch (queue) {
+		case 0:
+			amdgpu_fence_process(&adev->sdma.instance[instances].ring);
+			break;
+		default:
+			break;
+		}
+		break;
+	}
+	return 0;
+}
+
+static int sdma_v7_0_process_fence_irq(struct amdgpu_device *adev,
+				       struct amdgpu_irq_src *source,
+				       struct amdgpu_iv_entry *entry)
+{
+	u32 doorbell_offset = entry->src_data[0];
+
+	if (adev->enable_mes && doorbell_offset) {
+		struct amdgpu_userq_fence_driver *fence_drv = NULL;
+		struct xarray *xa = &adev->userq_xa;
+		unsigned long flags;
+
+		doorbell_offset >>= SDMA0_QUEUE0_DOORBELL_OFFSET__OFFSET__SHIFT;
+
+		xa_lock_irqsave(xa, flags);
+		fence_drv = xa_load(xa, doorbell_offset);
+		if (fence_drv)
+			amdgpu_userq_fence_driver_process(fence_drv);
+		xa_unlock_irqrestore(xa, flags);
+	}
+
+	return 0;
+}
+
+static int sdma_v7_0_process_illegal_inst_irq(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      struct amdgpu_iv_entry *entry)
+{
+	return 0;
+}
+
+static int sdma_v7_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					   enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int sdma_v7_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static void sdma_v7_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
+{
+}
+
+static void sdma_v7_0_print_ip_state(struct amdgpu_ip_block *ip_block, struct drm_printer *p)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, j;
+	uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_7_0);
+	uint32_t instance_offset;
+
+	if (!adev->sdma.ip_dump)
+		return;
+
+	drm_printf(p, "num_instances:%d\n", adev->sdma.num_instances);
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		instance_offset = i * reg_count;
+		drm_printf(p, "\nInstance:%d\n", i);
+
+		for (j = 0; j < reg_count; j++)
+			drm_printf(p, "%-50s \t 0x%08x\n", sdma_reg_list_7_0[j].reg_name,
+				   adev->sdma.ip_dump[instance_offset + j]);
+	}
+}
+
+static void sdma_v7_0_dump_ip_state(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, j;
+	uint32_t instance_offset;
+	uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_7_0);
+
+	if (!adev->sdma.ip_dump)
+		return;
+
+	amdgpu_gfx_off_ctrl(adev, false);
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		instance_offset = i * reg_count;
+		for (j = 0; j < reg_count; j++)
+			adev->sdma.ip_dump[instance_offset + j] =
+				RREG32(sdma_v7_0_get_reg_offset(adev, i,
+				       sdma_reg_list_7_0[j].reg_offset));
+	}
+	amdgpu_gfx_off_ctrl(adev, true);
+}
+
+const struct amd_ip_funcs sdma_v7_0_ip_funcs = {
+	.name = "sdma_v7_0",
+	.early_init = sdma_v7_0_early_init,
+	.late_init = NULL,
+	.sw_init = sdma_v7_0_sw_init,
+	.sw_fini = sdma_v7_0_sw_fini,
+	.hw_init = sdma_v7_0_hw_init,
+	.hw_fini = sdma_v7_0_hw_fini,
+	.suspend = sdma_v7_0_suspend,
+	.resume = sdma_v7_0_resume,
+	.is_idle = sdma_v7_0_is_idle,
+	.wait_for_idle = sdma_v7_0_wait_for_idle,
+	.soft_reset = sdma_v7_0_soft_reset,
+	.check_soft_reset = sdma_v7_0_check_soft_reset,
+	.set_clockgating_state = sdma_v7_0_set_clockgating_state,
+	.set_powergating_state = sdma_v7_0_set_powergating_state,
+	.get_clockgating_state = sdma_v7_0_get_clockgating_state,
+	.dump_ip_state = sdma_v7_0_dump_ip_state,
+	.print_ip_state = sdma_v7_0_print_ip_state,
+};
+
+static const struct amdgpu_ring_funcs sdma_v7_0_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_SDMA,
+	.align_mask = 0xf,
+	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
+	.support_64bit_ptrs = true,
+	.secure_submission_supported = true,
+	.get_rptr = sdma_v7_0_ring_get_rptr,
+	.get_wptr = sdma_v7_0_ring_get_wptr,
+	.set_wptr = sdma_v7_0_ring_set_wptr,
+	.emit_frame_size =
+		5 + /* sdma_v7_0_ring_init_cond_exec */
+		6 + /* sdma_v7_0_ring_emit_hdp_flush */
+		6 + /* sdma_v7_0_ring_emit_pipeline_sync */
+		/* sdma_v7_0_ring_emit_vm_flush */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 6 +
+		10 + 10 + 10, /* sdma_v7_0_ring_emit_fence x3 for user fence, vm fence */
+	.emit_ib_size = 5 + 7 + 6, /* sdma_v7_0_ring_emit_ib */
+	.emit_ib = sdma_v7_0_ring_emit_ib,
+	.emit_mem_sync = sdma_v7_0_ring_emit_mem_sync,
+	.emit_fence = sdma_v7_0_ring_emit_fence,
+	.emit_pipeline_sync = sdma_v7_0_ring_emit_pipeline_sync,
+	.emit_vm_flush = sdma_v7_0_ring_emit_vm_flush,
+	.emit_hdp_flush = sdma_v7_0_ring_emit_hdp_flush,
+	.test_ring = sdma_v7_0_ring_test_ring,
+	.test_ib = sdma_v7_0_ring_test_ib,
+	.insert_nop = sdma_v7_0_ring_insert_nop,
+	.pad_ib = sdma_v7_0_ring_pad_ib,
+	.emit_wreg = sdma_v7_0_ring_emit_wreg,
+	.emit_reg_wait = sdma_v7_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = sdma_v7_0_ring_emit_reg_write_reg_wait,
+	.init_cond_exec = sdma_v7_0_ring_init_cond_exec,
+	.preempt_ib = sdma_v7_0_ring_preempt_ib,
+	.reset = sdma_v7_0_reset_queue,
+};
+
+static void sdma_v7_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		adev->sdma.instance[i].ring.funcs = &sdma_v7_0_ring_funcs;
+		adev->sdma.instance[i].ring.me = i;
+	}
+}
+
+static const struct amdgpu_irq_src_funcs sdma_v7_0_trap_irq_funcs = {
+	.set = sdma_v7_0_set_trap_irq_state,
+	.process = sdma_v7_0_process_trap_irq,
+};
+
+static const struct amdgpu_irq_src_funcs sdma_v7_0_fence_irq_funcs = {
+	.process = sdma_v7_0_process_fence_irq,
+};
+
+static const struct amdgpu_irq_src_funcs sdma_v7_0_illegal_inst_irq_funcs = {
+	.process = sdma_v7_0_process_illegal_inst_irq,
+};
+
+static void sdma_v7_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->sdma.trap_irq.num_types = AMDGPU_SDMA_IRQ_INSTANCE0 +
+					adev->sdma.num_instances;
+	adev->sdma.trap_irq.funcs = &sdma_v7_0_trap_irq_funcs;
+	adev->sdma.fence_irq.funcs = &sdma_v7_0_fence_irq_funcs;
+	adev->sdma.illegal_inst_irq.funcs = &sdma_v7_0_illegal_inst_irq_funcs;
+}
+
+/**
+ * sdma_v7_0_emit_copy_buffer - copy buffer using the sDMA engine
+ *
+ * @ib: indirect buffer to fill with commands
+ * @src_offset: src GPU address
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ * @copy_flags: copy flags for the buffers
+ *
+ * Copy GPU buffers using the DMA engine.
+ * Used by the amdgpu ttm implementation to move pages if
+ * registered as the asic copy callback.
+ */
+static void sdma_v7_0_emit_copy_buffer(struct amdgpu_ib *ib,
+				       uint64_t src_offset,
+				       uint64_t dst_offset,
+				       uint32_t byte_count,
+				       uint32_t copy_flags)
+{
+	uint32_t num_type, data_format, max_com, write_cm;
+
+	max_com = AMDGPU_COPY_FLAGS_GET(copy_flags, MAX_COMPRESSED);
+	data_format = AMDGPU_COPY_FLAGS_GET(copy_flags, DATA_FORMAT);
+	num_type = AMDGPU_COPY_FLAGS_GET(copy_flags, NUMBER_TYPE);
+	write_cm = AMDGPU_COPY_FLAGS_GET(copy_flags, WRITE_COMPRESS_DISABLE) ? 2 : 1;
+
+	ib->ptr[ib->length_dw++] = SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_COPY) |
+		SDMA_PKT_COPY_LINEAR_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR) |
+		SDMA_PKT_COPY_LINEAR_HEADER_TMZ((copy_flags & AMDGPU_COPY_FLAGS_TMZ) ? 1 : 0) |
+		SDMA_PKT_COPY_LINEAR_HEADER_CPV(1);
+
+	ib->ptr[ib->length_dw++] = byte_count - 1;
+	ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+	ib->ptr[ib->length_dw++] = lower_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+
+	if ((copy_flags & (AMDGPU_COPY_FLAGS_READ_DECOMPRESSED | AMDGPU_COPY_FLAGS_WRITE_COMPRESSED)))
+		ib->ptr[ib->length_dw++] = SDMA_DCC_DATA_FORMAT(data_format) | SDMA_DCC_NUM_TYPE(num_type) |
+			((copy_flags & AMDGPU_COPY_FLAGS_READ_DECOMPRESSED) ? SDMA_DCC_READ_CM(2) : 0) |
+			((copy_flags & AMDGPU_COPY_FLAGS_WRITE_COMPRESSED) ? SDMA_DCC_WRITE_CM(write_cm) : 0) |
+			SDMA_DCC_MAX_COM(max_com) | SDMA_DCC_MAX_UCOM(1);
+	else
+		ib->ptr[ib->length_dw++] = 0;
+}
+
+/**
+ * sdma_v7_0_emit_fill_buffer - fill buffer using the sDMA engine
+ *
+ * @ib: indirect buffer to fill
+ * @src_data: value to write to buffer
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ *
+ * Fill GPU buffers using the DMA engine.
+ */
+static void sdma_v7_0_emit_fill_buffer(struct amdgpu_ib *ib,
+				       uint32_t src_data,
+				       uint64_t dst_offset,
+				       uint32_t byte_count)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PKT_CONSTANT_FILL_HEADER_OP(SDMA_OP_CONST_FILL) |
+		SDMA_PKT_CONSTANT_FILL_HEADER_COMPRESS(1);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = src_data;
+	ib->ptr[ib->length_dw++] = byte_count - 1;
+}
+
+static const struct amdgpu_buffer_funcs sdma_v7_0_buffer_funcs = {
+	.copy_max_bytes = 0x400000,
+	.copy_num_dw = 8,
+	.emit_copy_buffer = sdma_v7_0_emit_copy_buffer,
+	.fill_max_bytes = 0x400000,
+	.fill_num_dw = 5,
+	.emit_fill_buffer = sdma_v7_0_emit_fill_buffer,
+};
+
+static void sdma_v7_0_set_buffer_funcs(struct amdgpu_device *adev)
+{
+	adev->mman.buffer_funcs = &sdma_v7_0_buffer_funcs;
+	adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
+}
+
+static const struct amdgpu_vm_pte_funcs sdma_v7_0_vm_pte_funcs = {
+	.copy_pte_num_dw = 8,
+	.copy_pte = sdma_v7_0_vm_copy_pte,
+	.write_pte = sdma_v7_0_vm_write_pte,
+	.set_pte_pde = sdma_v7_0_vm_set_pte_pde,
+};
+
+static void sdma_v7_0_set_vm_pte_funcs(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	adev->vm_manager.vm_pte_funcs = &sdma_v7_0_vm_pte_funcs;
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		adev->vm_manager.vm_pte_scheds[i] =
+			&adev->sdma.instance[i].ring.sched;
+	}
+	adev->vm_manager.vm_pte_num_scheds = adev->sdma.num_instances;
+}
+
+const struct amdgpu_ip_block_version sdma_v7_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_SDMA,
+	.major = 7,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &sdma_v7_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.h b/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.h
new file mode 100644
index 000000000000..5af863bb39c4
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __SDMA_V7_0_H__
+#define __SDMA_V7_0_H__
+
+extern const struct amd_ip_funcs sdma_v7_0_ip_funcs;
+extern const struct amdgpu_ip_block_version sdma_v7_0_ip_block;
+
+#endif /* __SDMA_V7_0_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/si.c b/drivers/gpu/drm/amd/amdgpu/si.c
index 8284d5dbfc30..e0f139de7991 100644
--- a/drivers/gpu/drm/amd/amdgpu/si.c
+++ b/drivers/gpu/drm/amd/amdgpu/si.c
@@ -24,14 +24,19 @@
 #include <linux/firmware.h>
 #include <linux/slab.h>
 #include <linux/module.h>
-#include <drm/drmP.h>
+#include <linux/pci.h>
+
+#include <drm/amdgpu_drm.h>
+
 #include "amdgpu.h"
 #include "amdgpu_atombios.h"
 #include "amdgpu_ih.h"
 #include "amdgpu_uvd.h"
 #include "amdgpu_vce.h"
 #include "atom.h"
-#include "amdgpu_powerplay.h"
+#include "amd_pcie.h"
+
+#include "si_dpm.h"
 #include "sid.h"
 #include "si_ih.h"
 #include "gfx_v6_0.h"
@@ -39,13 +44,31 @@
 #include "si_dma.h"
 #include "dce_v6_0.h"
 #include "si.h"
-#include "dce_virtual.h"
+#include "uvd_v3_1.h"
+
+#include "uvd/uvd_4_0_d.h"
+
+#include "smu/smu_6_0_d.h"
+#include "smu/smu_6_0_sh_mask.h"
+
 #include "gca/gfx_6_0_d.h"
+#include "gca/gfx_6_0_sh_mask.h"
+
 #include "oss/oss_1_0_d.h"
+#include "oss/oss_1_0_sh_mask.h"
+
 #include "gmc/gmc_6_0_d.h"
+#include"gmc/gmc_6_0_sh_mask.h"
+
 #include "dce/dce_6_0_d.h"
-#include "uvd/uvd_4_0_d.h"
+#include "dce/dce_6_0_sh_mask.h"
+
 #include "bif/bif_3_0_d.h"
+#include "bif/bif_3_0_sh_mask.h"
+#include "si_enums.h"
+
+#include "amdgpu_dm.h"
+#include "amdgpu_vkms.h"
 
 static const u32 tahiti_golden_registers[] =
 {
@@ -898,6 +921,114 @@ static const u32 hainan_mgcg_cgcg_init[] =
 	0x3630, 0xfffffff0, 0x00000100,
 };
 
+/* XXX: update when we support VCE */
+#if 0
+/* tahiti, pitcairn, verde */
+static const struct amdgpu_video_codec_info tahiti_video_codecs_encode_array[] =
+{
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC,
+		.max_width = 2048,
+		.max_height = 1152,
+		.max_pixels_per_frame = 2048 * 1152,
+		.max_level = 0,
+	},
+};
+
+static const struct amdgpu_video_codecs tahiti_video_codecs_encode =
+{
+	.codec_count = ARRAY_SIZE(tahiti_video_codecs_encode_array),
+	.codec_array = tahiti_video_codecs_encode_array,
+};
+#else
+static const struct amdgpu_video_codecs tahiti_video_codecs_encode =
+{
+	.codec_count = 0,
+	.codec_array = NULL,
+};
+#endif
+/* oland and hainan don't support encode */
+static const struct amdgpu_video_codecs hainan_video_codecs_encode =
+{
+	.codec_count = 0,
+	.codec_array = NULL,
+};
+
+/* tahiti, pitcairn, verde, oland */
+static const struct amdgpu_video_codec_info tahiti_video_codecs_decode_array[] =
+{
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2,
+		.max_width = 2048,
+		.max_height = 1152,
+		.max_pixels_per_frame = 2048 * 1152,
+		.max_level = 3,
+	},
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4,
+		.max_width = 2048,
+		.max_height = 1152,
+		.max_pixels_per_frame = 2048 * 1152,
+		.max_level = 5,
+	},
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC,
+		.max_width = 2048,
+		.max_height = 1152,
+		.max_pixels_per_frame = 2048 * 1152,
+		.max_level = 41,
+	},
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1,
+		.max_width = 2048,
+		.max_height = 1152,
+		.max_pixels_per_frame = 2048 * 1152,
+		.max_level = 4,
+	},
+};
+
+static const struct amdgpu_video_codecs tahiti_video_codecs_decode =
+{
+	.codec_count = ARRAY_SIZE(tahiti_video_codecs_decode_array),
+	.codec_array = tahiti_video_codecs_decode_array,
+};
+
+/* hainan doesn't support decode */
+static const struct amdgpu_video_codecs hainan_video_codecs_decode =
+{
+	.codec_count = 0,
+	.codec_array = NULL,
+};
+
+static int si_query_video_codecs(struct amdgpu_device *adev, bool encode,
+				 const struct amdgpu_video_codecs **codecs)
+{
+	switch (adev->asic_type) {
+	case CHIP_VERDE:
+	case CHIP_TAHITI:
+	case CHIP_PITCAIRN:
+		if (encode)
+			*codecs = &tahiti_video_codecs_encode;
+		else
+			*codecs = &tahiti_video_codecs_decode;
+		return 0;
+	case CHIP_OLAND:
+		if (encode)
+			*codecs = &hainan_video_codecs_encode;
+		else
+			*codecs = &tahiti_video_codecs_decode;
+		return 0;
+	case CHIP_HAINAN:
+		if (encode)
+			*codecs = &hainan_video_codecs_encode;
+		else
+			*codecs = &hainan_video_codecs_decode;
+		return 0;
+	default:
+		return -EINVAL;
+	}
+}
+
 static u32 si_pcie_rreg(struct amdgpu_device *adev, u32 reg)
 {
 	unsigned long flags;
@@ -954,8 +1085,8 @@ static u32 si_smc_rreg(struct amdgpu_device *adev, u32 reg)
 	u32 r;
 
 	spin_lock_irqsave(&adev->smc_idx_lock, flags);
-	WREG32(SMC_IND_INDEX_0, (reg));
-	r = RREG32(SMC_IND_DATA_0);
+	WREG32(mmSMC_IND_INDEX_0, (reg));
+	r = RREG32(mmSMC_IND_DATA_0);
 	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
 	return r;
 }
@@ -965,50 +1096,83 @@ static void si_smc_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
 	unsigned long flags;
 
 	spin_lock_irqsave(&adev->smc_idx_lock, flags);
-	WREG32(SMC_IND_INDEX_0, (reg));
-	WREG32(SMC_IND_DATA_0, (v));
+	WREG32(mmSMC_IND_INDEX_0, (reg));
+	WREG32(mmSMC_IND_DATA_0, (v));
 	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
 }
 
+static u32 si_uvd_ctx_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->uvd_ctx_idx_lock, flags);
+	WREG32(mmUVD_CTX_INDEX, ((reg) & 0x1ff));
+	r = RREG32(mmUVD_CTX_DATA);
+	spin_unlock_irqrestore(&adev->uvd_ctx_idx_lock, flags);
+	return r;
+}
+
+static void si_uvd_ctx_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->uvd_ctx_idx_lock, flags);
+	WREG32(mmUVD_CTX_INDEX, ((reg) & 0x1ff));
+	WREG32(mmUVD_CTX_DATA, (v));
+	spin_unlock_irqrestore(&adev->uvd_ctx_idx_lock, flags);
+}
+
 static struct amdgpu_allowed_register_entry si_allowed_read_registers[] = {
-	{GRBM_STATUS},
-	{GB_ADDR_CONFIG},
-	{MC_ARB_RAMCFG},
-	{GB_TILE_MODE0},
-	{GB_TILE_MODE1},
-	{GB_TILE_MODE2},
-	{GB_TILE_MODE3},
-	{GB_TILE_MODE4},
-	{GB_TILE_MODE5},
-	{GB_TILE_MODE6},
-	{GB_TILE_MODE7},
-	{GB_TILE_MODE8},
-	{GB_TILE_MODE9},
-	{GB_TILE_MODE10},
-	{GB_TILE_MODE11},
-	{GB_TILE_MODE12},
-	{GB_TILE_MODE13},
-	{GB_TILE_MODE14},
-	{GB_TILE_MODE15},
-	{GB_TILE_MODE16},
-	{GB_TILE_MODE17},
-	{GB_TILE_MODE18},
-	{GB_TILE_MODE19},
-	{GB_TILE_MODE20},
-	{GB_TILE_MODE21},
-	{GB_TILE_MODE22},
-	{GB_TILE_MODE23},
-	{GB_TILE_MODE24},
-	{GB_TILE_MODE25},
-	{GB_TILE_MODE26},
-	{GB_TILE_MODE27},
-	{GB_TILE_MODE28},
-	{GB_TILE_MODE29},
-	{GB_TILE_MODE30},
-	{GB_TILE_MODE31},
-	{CC_RB_BACKEND_DISABLE, true},
-	{GC_USER_RB_BACKEND_DISABLE, true},
-	{PA_SC_RASTER_CONFIG, true},
+	{mmGRBM_STATUS},
+	{mmGRBM_STATUS2},
+	{mmGRBM_STATUS_SE0},
+	{mmGRBM_STATUS_SE1},
+	{mmSRBM_STATUS},
+	{mmSRBM_STATUS2},
+	{mmDMA_STATUS_REG + DMA0_REGISTER_OFFSET},
+	{mmDMA_STATUS_REG + DMA1_REGISTER_OFFSET},
+	{mmCP_STAT},
+	{mmCP_STALLED_STAT1},
+	{mmCP_STALLED_STAT2},
+	{mmCP_STALLED_STAT3},
+	{mmGB_ADDR_CONFIG},
+	{mmMC_ARB_RAMCFG},
+	{mmGB_TILE_MODE0},
+	{mmGB_TILE_MODE1},
+	{mmGB_TILE_MODE2},
+	{mmGB_TILE_MODE3},
+	{mmGB_TILE_MODE4},
+	{mmGB_TILE_MODE5},
+	{mmGB_TILE_MODE6},
+	{mmGB_TILE_MODE7},
+	{mmGB_TILE_MODE8},
+	{mmGB_TILE_MODE9},
+	{mmGB_TILE_MODE10},
+	{mmGB_TILE_MODE11},
+	{mmGB_TILE_MODE12},
+	{mmGB_TILE_MODE13},
+	{mmGB_TILE_MODE14},
+	{mmGB_TILE_MODE15},
+	{mmGB_TILE_MODE16},
+	{mmGB_TILE_MODE17},
+	{mmGB_TILE_MODE18},
+	{mmGB_TILE_MODE19},
+	{mmGB_TILE_MODE20},
+	{mmGB_TILE_MODE21},
+	{mmGB_TILE_MODE22},
+	{mmGB_TILE_MODE23},
+	{mmGB_TILE_MODE24},
+	{mmGB_TILE_MODE25},
+	{mmGB_TILE_MODE26},
+	{mmGB_TILE_MODE27},
+	{mmGB_TILE_MODE28},
+	{mmGB_TILE_MODE29},
+	{mmGB_TILE_MODE30},
+	{mmGB_TILE_MODE31},
+	{mmCC_RB_BACKEND_DISABLE, true},
+	{mmGC_USER_RB_BACKEND_DISABLE, true},
+	{mmPA_SC_RASTER_CONFIG, true},
 };
 
 static uint32_t si_get_register_value(struct amdgpu_device *adev,
@@ -1031,12 +1195,12 @@ static uint32_t si_get_register_value(struct amdgpu_device *adev,
 
 		mutex_lock(&adev->grbm_idx_mutex);
 		if (se_num != 0xffffffff || sh_num != 0xffffffff)
-			amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff);
+			amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff, 0);
 
 		val = RREG32(reg_offset);
 
 		if (se_num != 0xffffffff || sh_num != 0xffffffff)
-			amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+			amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 		mutex_unlock(&adev->grbm_idx_mutex);
 		return val;
 	} else {
@@ -1114,37 +1278,37 @@ static bool si_read_disabled_bios(struct amdgpu_device *adev)
 	u32 rom_cntl;
 	bool r;
 
-	bus_cntl = RREG32(R600_BUS_CNTL);
+	bus_cntl = RREG32(mmBUS_CNTL);
 	if (adev->mode_info.num_crtc) {
-		d1vga_control = RREG32(AVIVO_D1VGA_CONTROL);
-		d2vga_control = RREG32(AVIVO_D2VGA_CONTROL);
-		vga_render_control = RREG32(VGA_RENDER_CONTROL);
+		d1vga_control = RREG32(mmD1VGA_CONTROL);
+		d2vga_control = RREG32(mmD2VGA_CONTROL);
+		vga_render_control = RREG32(mmVGA_RENDER_CONTROL);
 	}
 	rom_cntl = RREG32(R600_ROM_CNTL);
 
 	/* enable the rom */
-	WREG32(R600_BUS_CNTL, (bus_cntl & ~R600_BIOS_ROM_DIS));
+	WREG32(mmBUS_CNTL, (bus_cntl & ~BUS_CNTL__BIOS_ROM_DIS_MASK));
 	if (adev->mode_info.num_crtc) {
 		/* Disable VGA mode */
-		WREG32(AVIVO_D1VGA_CONTROL,
-		       (d1vga_control & ~(AVIVO_DVGA_CONTROL_MODE_ENABLE |
-					  AVIVO_DVGA_CONTROL_TIMING_SELECT)));
-		WREG32(AVIVO_D2VGA_CONTROL,
-		       (d2vga_control & ~(AVIVO_DVGA_CONTROL_MODE_ENABLE |
-					  AVIVO_DVGA_CONTROL_TIMING_SELECT)));
-		WREG32(VGA_RENDER_CONTROL,
-		       (vga_render_control & C_000300_VGA_VSTATUS_CNTL));
+		WREG32(mmD1VGA_CONTROL,
+		       (d1vga_control & ~(D1VGA_CONTROL__D1VGA_MODE_ENABLE_MASK |
+					  D1VGA_CONTROL__D1VGA_TIMING_SELECT_MASK)));
+		WREG32(mmD2VGA_CONTROL,
+		       (d2vga_control & ~(D1VGA_CONTROL__D1VGA_MODE_ENABLE_MASK |
+					  D1VGA_CONTROL__D1VGA_TIMING_SELECT_MASK)));
+		WREG32(mmVGA_RENDER_CONTROL,
+		       (vga_render_control & ~VGA_RENDER_CONTROL__VGA_VSTATUS_CNTL_MASK));
 	}
 	WREG32(R600_ROM_CNTL, rom_cntl | R600_SCK_OVERWRITE);
 
 	r = amdgpu_read_bios(adev);
 
 	/* restore regs */
-	WREG32(R600_BUS_CNTL, bus_cntl);
+	WREG32(mmBUS_CNTL, bus_cntl);
 	if (adev->mode_info.num_crtc) {
-		WREG32(AVIVO_D1VGA_CONTROL, d1vga_control);
-		WREG32(AVIVO_D2VGA_CONTROL, d2vga_control);
-		WREG32(VGA_RENDER_CONTROL, vga_render_control);
+		WREG32(mmD1VGA_CONTROL, d1vga_control);
+		WREG32(mmD2VGA_CONTROL, d2vga_control);
+		WREG32(mmVGA_RENDER_CONTROL, vga_render_control);
 	}
 	WREG32(R600_ROM_CNTL, rom_cntl);
 	return r;
@@ -1177,12 +1341,125 @@ static bool si_read_bios_from_rom(struct amdgpu_device *adev,
 	return true;
 }
 
-//xxx: not implemented
-static int si_asic_reset(struct amdgpu_device *adev)
+static void si_set_clk_bypass_mode(struct amdgpu_device *adev)
+{
+	u32 tmp, i;
+
+	tmp = RREG32(mmCG_SPLL_FUNC_CNTL);
+	tmp |= CG_SPLL_FUNC_CNTL__SPLL_BYPASS_EN_MASK;
+	WREG32(mmCG_SPLL_FUNC_CNTL, tmp);
+
+	tmp = RREG32(mmCG_SPLL_FUNC_CNTL_2);
+	tmp |= CG_SPLL_FUNC_CNTL_2__SPLL_CTLREQ_CHG_MASK;
+	WREG32(mmCG_SPLL_FUNC_CNTL_2, tmp);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (RREG32(mmCG_SPLL_STATUS) & CG_SPLL_STATUS__SPLL_CHG_STATUS_MASK)
+			break;
+		udelay(1);
+	}
+
+	tmp = RREG32(mmCG_SPLL_FUNC_CNTL_2);
+	tmp &= ~(CG_SPLL_FUNC_CNTL_2__SPLL_CTLREQ_CHG_MASK |
+		CG_SPLL_FUNC_CNTL_2__SCLK_MUX_UPDATE_MASK);
+	WREG32(mmCG_SPLL_FUNC_CNTL_2, tmp);
+
+	tmp = RREG32(MPLL_CNTL_MODE);
+	tmp &= ~MPLL_MCLK_SEL;
+	WREG32(MPLL_CNTL_MODE, tmp);
+}
+
+static void si_spll_powerdown(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	tmp = RREG32(mmSPLL_CNTL_MODE);
+	tmp |= SPLL_CNTL_MODE__SPLL_SW_DIR_CONTROL_MASK;
+	WREG32(mmSPLL_CNTL_MODE, tmp);
+
+	tmp = RREG32(mmCG_SPLL_FUNC_CNTL);
+	tmp |= CG_SPLL_FUNC_CNTL__SPLL_RESET_MASK;
+	WREG32(mmCG_SPLL_FUNC_CNTL, tmp);
+
+	tmp = RREG32(mmCG_SPLL_FUNC_CNTL);
+	tmp |= CG_SPLL_FUNC_CNTL__SPLL_SLEEP_MASK;
+	WREG32(mmCG_SPLL_FUNC_CNTL, tmp);
+
+	tmp = RREG32(mmSPLL_CNTL_MODE);
+	tmp &= ~SPLL_CNTL_MODE__SPLL_SW_DIR_CONTROL_MASK;
+	WREG32(mmSPLL_CNTL_MODE, tmp);
+}
+
+static int si_gpu_pci_config_reset(struct amdgpu_device *adev)
+{
+	u32 i;
+	int r = -EINVAL;
+
+	amdgpu_atombios_scratch_regs_engine_hung(adev, true);
+
+	/* set mclk/sclk to bypass */
+	si_set_clk_bypass_mode(adev);
+	/* powerdown spll */
+	si_spll_powerdown(adev);
+	/* disable BM */
+	pci_clear_master(adev->pdev);
+	/* reset */
+	amdgpu_device_pci_config_reset(adev);
+
+	udelay(100);
+
+	/* wait for asic to come out of reset */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (RREG32(mmCONFIG_MEMSIZE) != 0xffffffff) {
+			/* enable BM */
+			pci_set_master(adev->pdev);
+			adev->has_hw_reset = true;
+			r = 0;
+			break;
+		}
+		udelay(1);
+	}
+	amdgpu_atombios_scratch_regs_engine_hung(adev, false);
+
+	return r;
+}
+
+static int si_asic_supports_baco(struct amdgpu_device *adev)
 {
 	return 0;
 }
 
+static enum amd_reset_method
+si_asic_reset_method(struct amdgpu_device *adev)
+{
+	if (amdgpu_reset_method == AMD_RESET_METHOD_PCI)
+		return amdgpu_reset_method;
+	else if (amdgpu_reset_method != AMD_RESET_METHOD_LEGACY &&
+		 amdgpu_reset_method != -1)
+		dev_warn(adev->dev, "Specified reset method:%d isn't supported, using AUTO instead.\n",
+			 amdgpu_reset_method);
+
+	return AMD_RESET_METHOD_LEGACY;
+}
+
+static int si_asic_reset(struct amdgpu_device *adev)
+{
+	int r;
+
+	switch (si_asic_reset_method(adev)) {
+	case AMD_RESET_METHOD_PCI:
+		dev_info(adev->dev, "PCI reset\n");
+		r = amdgpu_device_pci_reset(adev);
+		break;
+	default:
+		dev_info(adev->dev, "PCI CONFIG reset\n");
+		r = si_gpu_pci_config_reset(adev);
+		break;
+	}
+
+	return r;
+}
+
 static u32 si_get_config_memsize(struct amdgpu_device *adev)
 {
 	return RREG32(mmCONFIG_MEMSIZE);
@@ -1192,42 +1469,541 @@ static void si_vga_set_state(struct amdgpu_device *adev, bool state)
 {
 	uint32_t temp;
 
-	temp = RREG32(CONFIG_CNTL);
-	if (state == false) {
+	temp = RREG32(mmCONFIG_CNTL);
+	if (!state) {
 		temp &= ~(1<<0);
 		temp |= (1<<1);
 	} else {
 		temp &= ~(1<<1);
 	}
-	WREG32(CONFIG_CNTL, temp);
+	WREG32(mmCONFIG_CNTL, temp);
 }
 
 static u32 si_get_xclk(struct amdgpu_device *adev)
 {
-        u32 reference_clock = adev->clock.spll.reference_freq;
+	u32 reference_clock = adev->clock.spll.reference_freq;
 	u32 tmp;
 
-	tmp = RREG32(CG_CLKPIN_CNTL_2);
-	if (tmp & MUX_TCLK_TO_XCLK)
+	tmp = RREG32(mmCG_CLKPIN_CNTL_2);
+	if (tmp & CG_CLKPIN_CNTL_2__MUX_TCLK_TO_XCLK_MASK)
 		return TCLK;
 
-	tmp = RREG32(CG_CLKPIN_CNTL);
-	if (tmp & XTALIN_DIVIDE)
+	tmp = RREG32(mmCG_CLKPIN_CNTL);
+	if (tmp & CG_CLKPIN_CNTL__XTALIN_DIVIDE_MASK)
 		return reference_clock / 4;
 
 	return reference_clock;
 }
 
-//xxx:not implemented
+static void si_flush_hdp(struct amdgpu_device *adev, struct amdgpu_ring *ring)
+{
+	if (!ring || !ring->funcs->emit_wreg) {
+		WREG32(mmHDP_MEM_COHERENCY_FLUSH_CNTL, 1);
+		RREG32(mmHDP_MEM_COHERENCY_FLUSH_CNTL);
+	} else {
+		amdgpu_ring_emit_wreg(ring, mmHDP_MEM_COHERENCY_FLUSH_CNTL, 1);
+	}
+}
+
+static void si_invalidate_hdp(struct amdgpu_device *adev,
+			      struct amdgpu_ring *ring)
+{
+	if (!ring || !ring->funcs->emit_wreg) {
+		WREG32(mmHDP_DEBUG0, 1);
+		RREG32(mmHDP_DEBUG0);
+	} else {
+		amdgpu_ring_emit_wreg(ring, mmHDP_DEBUG0, 1);
+	}
+}
+
+static bool si_need_full_reset(struct amdgpu_device *adev)
+{
+	/* change this when we support soft reset */
+	return true;
+}
+
+static bool si_need_reset_on_init(struct amdgpu_device *adev)
+{
+	return false;
+}
+
+static int si_get_pcie_lanes(struct amdgpu_device *adev)
+{
+	u32 link_width_cntl;
+
+	if (adev->flags & AMD_IS_APU)
+		return 0;
+
+	link_width_cntl = RREG32_PCIE_PORT(ixPCIE_LC_LINK_WIDTH_CNTL);
+
+	switch ((link_width_cntl & PCIE_LC_LINK_WIDTH_CNTL__LC_LINK_WIDTH_RD_MASK) >> PCIE_LC_LINK_WIDTH_CNTL__LC_LINK_WIDTH_RD__SHIFT) {
+	case LC_LINK_WIDTH_X1:
+		return 1;
+	case LC_LINK_WIDTH_X2:
+		return 2;
+	case LC_LINK_WIDTH_X4:
+		return 4;
+	case LC_LINK_WIDTH_X8:
+		return 8;
+	case LC_LINK_WIDTH_X0:
+	case LC_LINK_WIDTH_X16:
+	default:
+		return 16;
+	}
+}
+
+static void si_set_pcie_lanes(struct amdgpu_device *adev, int lanes)
+{
+	u32 link_width_cntl, mask;
+
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	switch (lanes) {
+	case 0:
+		mask = LC_LINK_WIDTH_X0;
+		break;
+	case 1:
+		mask = LC_LINK_WIDTH_X1;
+		break;
+	case 2:
+		mask = LC_LINK_WIDTH_X2;
+		break;
+	case 4:
+		mask = LC_LINK_WIDTH_X4;
+		break;
+	case 8:
+		mask = LC_LINK_WIDTH_X8;
+		break;
+	case 16:
+		mask = LC_LINK_WIDTH_X16;
+		break;
+	default:
+		DRM_ERROR("invalid pcie lane request: %d\n", lanes);
+		return;
+	}
+
+	link_width_cntl = RREG32_PCIE_PORT(ixPCIE_LC_LINK_WIDTH_CNTL);
+	link_width_cntl &= ~PCIE_LC_LINK_WIDTH_CNTL__LC_LINK_WIDTH_MASK;
+	link_width_cntl |= mask << PCIE_LC_LINK_WIDTH_CNTL__LC_LINK_WIDTH__SHIFT;
+	link_width_cntl |= (PCIE_LC_LINK_WIDTH_CNTL__LC_RECONFIG_NOW_MASK |
+		PCIE_LC_LINK_WIDTH_CNTL__LC_RECONFIG_ARC_MISSING_ESCAPE_MASK);
+
+	WREG32_PCIE_PORT(ixPCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
+}
+
+static void si_get_pcie_usage(struct amdgpu_device *adev, uint64_t *count0,
+			      uint64_t *count1)
+{
+	uint32_t perfctr = 0;
+	uint64_t cnt0_of, cnt1_of;
+	int tmp;
+
+	/* This reports 0 on APUs, so return to avoid writing/reading registers
+	 * that may or may not be different from their GPU counterparts
+	 */
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	/* Set the 2 events that we wish to watch, defined above */
+	/* Reg 40 is # received msgs, Reg 104 is # of posted requests sent */
+	perfctr = REG_SET_FIELD(perfctr, PCIE_PERF_CNTL_TXCLK, EVENT0_SEL, 40);
+	perfctr = REG_SET_FIELD(perfctr, PCIE_PERF_CNTL_TXCLK, EVENT1_SEL, 104);
+
+	/* Write to enable desired perf counters */
+	WREG32_PCIE(ixPCIE_PERF_CNTL_TXCLK, perfctr);
+	/* Zero out and enable the perf counters
+	 * Write 0x5:
+	 * Bit 0 = Start all counters(1)
+	 * Bit 2 = Global counter reset enable(1)
+	 */
+	WREG32_PCIE(ixPCIE_PERF_COUNT_CNTL, 0x00000005);
+
+	msleep(1000);
+
+	/* Load the shadow and disable the perf counters
+	 * Write 0x2:
+	 * Bit 0 = Stop counters(0)
+	 * Bit 1 = Load the shadow counters(1)
+	 */
+	WREG32_PCIE(ixPCIE_PERF_COUNT_CNTL, 0x00000002);
+
+	/* Read register values to get any >32bit overflow */
+	tmp = RREG32_PCIE(ixPCIE_PERF_CNTL_TXCLK);
+	cnt0_of = REG_GET_FIELD(tmp, PCIE_PERF_CNTL_TXCLK, COUNTER0_UPPER);
+	cnt1_of = REG_GET_FIELD(tmp, PCIE_PERF_CNTL_TXCLK, COUNTER1_UPPER);
+
+	/* Get the values and add the overflow */
+	*count0 = RREG32_PCIE(ixPCIE_PERF_COUNT0_TXCLK) | (cnt0_of << 32);
+	*count1 = RREG32_PCIE(ixPCIE_PERF_COUNT1_TXCLK) | (cnt1_of << 32);
+}
+
+static uint64_t si_get_pcie_replay_count(struct amdgpu_device *adev)
+{
+	uint64_t nak_r, nak_g;
+
+	/* Get the number of NAKs received and generated */
+	nak_r = RREG32_PCIE(ixPCIE_RX_NUM_NAK);
+	nak_g = RREG32_PCIE(ixPCIE_RX_NUM_NAK_GENERATED);
+
+	/* Add the total number of NAKs, i.e the number of replays */
+	return (nak_r + nak_g);
+}
+
+static int si_uvd_send_upll_ctlreq(struct amdgpu_device *adev,
+				   unsigned cg_upll_func_cntl)
+{
+	unsigned i;
+
+	/* Make sure UPLL_CTLREQ is deasserted */
+	WREG32_P(cg_upll_func_cntl, 0, ~UPLL_CTLREQ_MASK);
+
+	mdelay(10);
+
+	/* Assert UPLL_CTLREQ */
+	WREG32_P(cg_upll_func_cntl, UPLL_CTLREQ_MASK, ~UPLL_CTLREQ_MASK);
+
+	/* Wait for CTLACK and CTLACK2 to get asserted */
+	for (i = 0; i < SI_MAX_CTLACKS_ASSERTION_WAIT; ++i) {
+		uint32_t mask = UPLL_CTLACK_MASK | UPLL_CTLACK2_MASK;
+
+		if ((RREG32(cg_upll_func_cntl) & mask) == mask)
+			break;
+		mdelay(10);
+	}
+
+	/* Deassert UPLL_CTLREQ */
+	WREG32_P(cg_upll_func_cntl, 0, ~UPLL_CTLREQ_MASK);
+
+	if (i == SI_MAX_CTLACKS_ASSERTION_WAIT) {
+		DRM_ERROR("Timeout setting UVD clocks!\n");
+		return -ETIMEDOUT;
+	}
+
+	return 0;
+}
+
+static unsigned si_uvd_calc_upll_post_div(unsigned vco_freq,
+					  unsigned target_freq,
+					  unsigned pd_min,
+					  unsigned pd_even)
+{
+	unsigned post_div = vco_freq / target_freq;
+
+	/* Adjust to post divider minimum value */
+	if (post_div < pd_min)
+		post_div = pd_min;
+
+	/* We alway need a frequency less than or equal the target */
+	if ((vco_freq / post_div) > target_freq)
+		post_div += 1;
+
+	/* Post dividers above a certain value must be even */
+	if (post_div > pd_even && post_div % 2)
+		post_div += 1;
+
+	return post_div;
+}
+
+/**
+ * si_calc_upll_dividers - calc UPLL clock dividers
+ *
+ * @adev: amdgpu_device pointer
+ * @vclk: wanted VCLK
+ * @dclk: wanted DCLK
+ * @vco_min: minimum VCO frequency
+ * @vco_max: maximum VCO frequency
+ * @fb_factor: factor to multiply vco freq with
+ * @fb_mask: limit and bitmask for feedback divider
+ * @pd_min: post divider minimum
+ * @pd_max: post divider maximum
+ * @pd_even: post divider must be even above this value
+ * @optimal_fb_div: resulting feedback divider
+ * @optimal_vclk_div: resulting vclk post divider
+ * @optimal_dclk_div: resulting dclk post divider
+ *
+ * Calculate dividers for UVDs UPLL (except APUs).
+ * Returns zero on success; -EINVAL on error.
+ */
+static int si_calc_upll_dividers(struct amdgpu_device *adev,
+				 unsigned vclk, unsigned dclk,
+				 unsigned vco_min, unsigned vco_max,
+				 unsigned fb_factor, unsigned fb_mask,
+				 unsigned pd_min, unsigned pd_max,
+				 unsigned pd_even,
+				 unsigned *optimal_fb_div,
+				 unsigned *optimal_vclk_div,
+				 unsigned *optimal_dclk_div)
+{
+	unsigned vco_freq, ref_freq = adev->clock.spll.reference_freq;
+
+	/* Start off with something large */
+	unsigned optimal_score = ~0;
+
+	/* Loop through vco from low to high */
+	vco_min = max(max(vco_min, vclk), dclk);
+	for (vco_freq = vco_min; vco_freq <= vco_max; vco_freq += 100) {
+		uint64_t fb_div = (uint64_t)vco_freq * fb_factor;
+		unsigned vclk_div, dclk_div, score;
+
+		do_div(fb_div, ref_freq);
+
+		/* fb div out of range ? */
+		if (fb_div > fb_mask)
+			break; /* It can oly get worse */
+
+		fb_div &= fb_mask;
+
+		/* Calc vclk divider with current vco freq */
+		vclk_div = si_uvd_calc_upll_post_div(vco_freq, vclk,
+						     pd_min, pd_even);
+		if (vclk_div > pd_max)
+			break; /* vco is too big, it has to stop */
+
+		/* Calc dclk divider with current vco freq */
+		dclk_div = si_uvd_calc_upll_post_div(vco_freq, dclk,
+						     pd_min, pd_even);
+		if (dclk_div > pd_max)
+			break; /* vco is too big, it has to stop */
+
+		/* Calc score with current vco freq */
+		score = vclk - (vco_freq / vclk_div) + dclk - (vco_freq / dclk_div);
+
+		/* Determine if this vco setting is better than current optimal settings */
+		if (score < optimal_score) {
+			*optimal_fb_div = fb_div;
+			*optimal_vclk_div = vclk_div;
+			*optimal_dclk_div = dclk_div;
+			optimal_score = score;
+			if (optimal_score == 0)
+				break; /* It can't get better than this */
+		}
+	}
+
+	/* Did we found a valid setup ? */
+	if (optimal_score == ~0)
+		return -EINVAL;
+
+	return 0;
+}
+
 static int si_set_uvd_clocks(struct amdgpu_device *adev, u32 vclk, u32 dclk)
 {
+	unsigned fb_div = 0, vclk_div = 0, dclk_div = 0;
+	int r;
+
+	/* Bypass vclk and dclk with bclk */
+	WREG32_P(CG_UPLL_FUNC_CNTL_2,
+		 VCLK_SRC_SEL(1) | DCLK_SRC_SEL(1),
+		 ~(VCLK_SRC_SEL_MASK | DCLK_SRC_SEL_MASK));
+
+	/* Put PLL in bypass mode */
+	WREG32_P(CG_UPLL_FUNC_CNTL, UPLL_BYPASS_EN_MASK, ~UPLL_BYPASS_EN_MASK);
+
+	if (!vclk || !dclk) {
+		/* Keep the Bypass mode */
+		return 0;
+	}
+
+	r = si_calc_upll_dividers(adev, vclk, dclk, 125000, 250000,
+				  16384, 0x03FFFFFF, 0, 128, 5,
+				  &fb_div, &vclk_div, &dclk_div);
+	if (r)
+		return r;
+
+	/* Set RESET_ANTI_MUX to 0 */
+	WREG32_P(CG_UPLL_FUNC_CNTL_5, 0, ~RESET_ANTI_MUX_MASK);
+
+	/* Set VCO_MODE to 1 */
+	WREG32_P(CG_UPLL_FUNC_CNTL, UPLL_VCO_MODE_MASK, ~UPLL_VCO_MODE_MASK);
+
+	/* Disable sleep mode */
+	WREG32_P(CG_UPLL_FUNC_CNTL, 0, ~UPLL_SLEEP_MASK);
+
+	/* Deassert UPLL_RESET */
+	WREG32_P(CG_UPLL_FUNC_CNTL, 0, ~UPLL_RESET_MASK);
+
+	mdelay(1);
+
+	r = si_uvd_send_upll_ctlreq(adev, CG_UPLL_FUNC_CNTL);
+	if (r)
+		return r;
+
+	/* Assert UPLL_RESET again */
+	WREG32_P(CG_UPLL_FUNC_CNTL, UPLL_RESET_MASK, ~UPLL_RESET_MASK);
+
+	/* Disable spread spectrum. */
+	WREG32_P(CG_UPLL_SPREAD_SPECTRUM, 0, ~SSEN_MASK);
+
+	/* Set feedback divider */
+	WREG32_P(CG_UPLL_FUNC_CNTL_3, UPLL_FB_DIV(fb_div), ~UPLL_FB_DIV_MASK);
+
+	/* Set ref divider to 0 */
+	WREG32_P(CG_UPLL_FUNC_CNTL, 0, ~UPLL_REF_DIV_MASK);
+
+	if (fb_div < 307200)
+		WREG32_P(CG_UPLL_FUNC_CNTL_4, 0, ~UPLL_SPARE_ISPARE9);
+	else
+		WREG32_P(CG_UPLL_FUNC_CNTL_4,
+			 UPLL_SPARE_ISPARE9,
+			 ~UPLL_SPARE_ISPARE9);
+
+	/* Set PDIV_A and PDIV_B */
+	WREG32_P(CG_UPLL_FUNC_CNTL_2,
+		 UPLL_PDIV_A(vclk_div) | UPLL_PDIV_B(dclk_div),
+		 ~(UPLL_PDIV_A_MASK | UPLL_PDIV_B_MASK));
+
+	/* Give the PLL some time to settle */
+	mdelay(15);
+
+	/* Deassert PLL_RESET */
+	WREG32_P(CG_UPLL_FUNC_CNTL, 0, ~UPLL_RESET_MASK);
+
+	mdelay(15);
+
+	/* Switch from bypass mode to normal mode */
+	WREG32_P(CG_UPLL_FUNC_CNTL, 0, ~UPLL_BYPASS_EN_MASK);
+
+	r = si_uvd_send_upll_ctlreq(adev, CG_UPLL_FUNC_CNTL);
+	if (r)
+		return r;
+
+	/* Switch VCLK and DCLK selection */
+	WREG32_P(CG_UPLL_FUNC_CNTL_2,
+		 VCLK_SRC_SEL(2) | DCLK_SRC_SEL(2),
+		 ~(VCLK_SRC_SEL_MASK | DCLK_SRC_SEL_MASK));
+
+	mdelay(100);
+
+	return 0;
+}
+
+static int si_vce_send_vcepll_ctlreq(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	/* Make sure VCEPLL_CTLREQ is deasserted */
+	WREG32_SMC_P(CG_VCEPLL_FUNC_CNTL, 0, ~UPLL_CTLREQ_MASK);
+
+	mdelay(10);
+
+	/* Assert UPLL_CTLREQ */
+	WREG32_SMC_P(CG_VCEPLL_FUNC_CNTL, UPLL_CTLREQ_MASK, ~UPLL_CTLREQ_MASK);
+
+	/* Wait for CTLACK and CTLACK2 to get asserted */
+	for (i = 0; i < SI_MAX_CTLACKS_ASSERTION_WAIT; ++i) {
+		uint32_t mask = UPLL_CTLACK_MASK | UPLL_CTLACK2_MASK;
+
+		if ((RREG32_SMC(CG_VCEPLL_FUNC_CNTL) & mask) == mask)
+			break;
+		mdelay(10);
+	}
+
+	/* Deassert UPLL_CTLREQ */
+	WREG32_SMC_P(CG_VCEPLL_FUNC_CNTL, 0, ~UPLL_CTLREQ_MASK);
+
+	if (i == SI_MAX_CTLACKS_ASSERTION_WAIT) {
+		DRM_ERROR("Timeout setting VCE clocks!\n");
+		return -ETIMEDOUT;
+	}
+
 	return 0;
 }
 
-static void si_detect_hw_virtualization(struct amdgpu_device *adev)
+static int si_set_vce_clocks(struct amdgpu_device *adev, u32 evclk, u32 ecclk)
+{
+	unsigned fb_div = 0, evclk_div = 0, ecclk_div = 0;
+	int r;
+
+	/* Bypass evclk and ecclk with bclk */
+	WREG32_SMC_P(CG_VCEPLL_FUNC_CNTL_2,
+		     EVCLK_SRC_SEL(1) | ECCLK_SRC_SEL(1),
+		     ~(EVCLK_SRC_SEL_MASK | ECCLK_SRC_SEL_MASK));
+
+	/* Put PLL in bypass mode */
+	WREG32_SMC_P(CG_VCEPLL_FUNC_CNTL, VCEPLL_BYPASS_EN_MASK,
+		     ~VCEPLL_BYPASS_EN_MASK);
+
+	if (!evclk || !ecclk) {
+		/* Keep the Bypass mode, put PLL to sleep */
+		WREG32_SMC_P(CG_VCEPLL_FUNC_CNTL, VCEPLL_SLEEP_MASK,
+			     ~VCEPLL_SLEEP_MASK);
+		return 0;
+	}
+
+	r = si_calc_upll_dividers(adev, evclk, ecclk, 125000, 250000,
+				  16384, 0x03FFFFFF, 0, 128, 5,
+				  &fb_div, &evclk_div, &ecclk_div);
+	if (r)
+		return r;
+
+	/* Set RESET_ANTI_MUX to 0 */
+	WREG32_SMC_P(CG_VCEPLL_FUNC_CNTL_5, 0, ~RESET_ANTI_MUX_MASK);
+
+	/* Set VCO_MODE to 1 */
+	WREG32_SMC_P(CG_VCEPLL_FUNC_CNTL, VCEPLL_VCO_MODE_MASK,
+		     ~VCEPLL_VCO_MODE_MASK);
+
+	/* Toggle VCEPLL_SLEEP to 1 then back to 0 */
+	WREG32_SMC_P(CG_VCEPLL_FUNC_CNTL, VCEPLL_SLEEP_MASK,
+		     ~VCEPLL_SLEEP_MASK);
+	WREG32_SMC_P(CG_VCEPLL_FUNC_CNTL, 0, ~VCEPLL_SLEEP_MASK);
+
+	/* Deassert VCEPLL_RESET */
+	WREG32_SMC_P(CG_VCEPLL_FUNC_CNTL, 0, ~VCEPLL_RESET_MASK);
+
+	mdelay(1);
+
+	r = si_vce_send_vcepll_ctlreq(adev);
+	if (r)
+		return r;
+
+	/* Assert VCEPLL_RESET again */
+	WREG32_SMC_P(CG_VCEPLL_FUNC_CNTL, VCEPLL_RESET_MASK, ~VCEPLL_RESET_MASK);
+
+	/* Disable spread spectrum. */
+	WREG32_SMC_P(CG_VCEPLL_SPREAD_SPECTRUM, 0, ~SSEN_MASK);
+
+	/* Set feedback divider */
+	WREG32_SMC_P(CG_VCEPLL_FUNC_CNTL_3,
+		     VCEPLL_FB_DIV(fb_div),
+		     ~VCEPLL_FB_DIV_MASK);
+
+	/* Set ref divider to 0 */
+	WREG32_SMC_P(CG_VCEPLL_FUNC_CNTL, 0, ~VCEPLL_REF_DIV_MASK);
+
+	/* Set PDIV_A and PDIV_B */
+	WREG32_SMC_P(CG_VCEPLL_FUNC_CNTL_2,
+		     VCEPLL_PDIV_A(evclk_div) | VCEPLL_PDIV_B(ecclk_div),
+		     ~(VCEPLL_PDIV_A_MASK | VCEPLL_PDIV_B_MASK));
+
+	/* Give the PLL some time to settle */
+	mdelay(15);
+
+	/* Deassert PLL_RESET */
+	WREG32_SMC_P(CG_VCEPLL_FUNC_CNTL, 0, ~VCEPLL_RESET_MASK);
+
+	mdelay(15);
+
+	/* Switch from bypass mode to normal mode */
+	WREG32_SMC_P(CG_VCEPLL_FUNC_CNTL, 0, ~VCEPLL_BYPASS_EN_MASK);
+
+	r = si_vce_send_vcepll_ctlreq(adev);
+	if (r)
+		return r;
+
+	/* Switch VCLK and DCLK selection */
+	WREG32_SMC_P(CG_VCEPLL_FUNC_CNTL_2,
+		     EVCLK_SRC_SEL(16) | ECCLK_SRC_SEL(16),
+		     ~(EVCLK_SRC_SEL_MASK | ECCLK_SRC_SEL_MASK));
+
+	mdelay(100);
+
+	return 0;
+}
+
+static void si_pre_asic_init(struct amdgpu_device *adev)
 {
-	if (is_virtual_machine()) /* passthrough mode */
-		adev->virt.caps |= AMDGPU_PASSTHROUGH_MODE;
 }
 
 static const struct amdgpu_asic_funcs si_asic_funcs =
@@ -1236,22 +2012,34 @@ static const struct amdgpu_asic_funcs si_asic_funcs =
 	.read_bios_from_rom = &si_read_bios_from_rom,
 	.read_register = &si_read_register,
 	.reset = &si_asic_reset,
+	.reset_method = &si_asic_reset_method,
 	.set_vga_state = &si_vga_set_state,
 	.get_xclk = &si_get_xclk,
 	.set_uvd_clocks = &si_set_uvd_clocks,
-	.set_vce_clocks = NULL,
+	.set_vce_clocks = &si_set_vce_clocks,
+	.get_pcie_lanes = &si_get_pcie_lanes,
+	.set_pcie_lanes = &si_set_pcie_lanes,
 	.get_config_memsize = &si_get_config_memsize,
+	.flush_hdp = &si_flush_hdp,
+	.invalidate_hdp = &si_invalidate_hdp,
+	.need_full_reset = &si_need_full_reset,
+	.get_pcie_usage = &si_get_pcie_usage,
+	.need_reset_on_init = &si_need_reset_on_init,
+	.get_pcie_replay_count = &si_get_pcie_replay_count,
+	.supports_baco = &si_asic_supports_baco,
+	.pre_asic_init = &si_pre_asic_init,
+	.query_video_codecs = &si_query_video_codecs,
 };
 
 static uint32_t si_get_rev_id(struct amdgpu_device *adev)
 {
-	return (RREG32(CC_DRM_ID_STRAPS) & CC_DRM_ID_STRAPS__ATI_REV_ID_MASK)
+	return (RREG32(mmCC_DRM_ID_STRAPS) & CC_DRM_ID_STRAPS__ATI_REV_ID_MASK)
 		>> CC_DRM_ID_STRAPS__ATI_REV_ID__SHIFT;
 }
 
-static int si_common_early_init(void *handle)
+static int si_common_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	adev->smc_rreg = &si_smc_rreg;
 	adev->smc_wreg = &si_smc_wreg;
@@ -1259,8 +2047,8 @@ static int si_common_early_init(void *handle)
 	adev->pcie_wreg = &si_pcie_wreg;
 	adev->pciep_rreg = &si_pciep_rreg;
 	adev->pciep_wreg = &si_pciep_wreg;
-	adev->uvd_ctx_rreg = NULL;
-	adev->uvd_ctx_wreg = NULL;
+	adev->uvd_ctx_rreg = si_uvd_ctx_rreg;
+	adev->uvd_ctx_wreg = si_uvd_ctx_wreg;
 	adev->didt_rreg = NULL;
 	adev->didt_wreg = NULL;
 
@@ -1284,7 +2072,7 @@ static int si_common_early_init(void *handle)
 			AMD_CG_SUPPORT_UVD_MGCG |
 			AMD_CG_SUPPORT_HDP_LS |
 			AMD_CG_SUPPORT_HDP_MGCG;
-			adev->pg_flags = 0;
+		adev->pg_flags = 0;
 		adev->external_rev_id = (adev->rev_id == 0) ? 1 :
 					(adev->rev_id == 1) ? 5 : 6;
 		break;
@@ -1375,80 +2163,69 @@ static int si_common_early_init(void *handle)
 	return 0;
 }
 
-static int si_common_sw_init(void *handle)
-{
-	return 0;
-}
-
-static int si_common_sw_fini(void *handle)
-{
-	return 0;
-}
-
-
 static void si_init_golden_registers(struct amdgpu_device *adev)
 {
 	switch (adev->asic_type) {
 	case CHIP_TAHITI:
-		amdgpu_program_register_sequence(adev,
-						 tahiti_golden_registers,
-						 (const u32)ARRAY_SIZE(tahiti_golden_registers));
-		amdgpu_program_register_sequence(adev,
-						 tahiti_golden_rlc_registers,
-						 (const u32)ARRAY_SIZE(tahiti_golden_rlc_registers));
-		amdgpu_program_register_sequence(adev,
-						 tahiti_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(tahiti_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 tahiti_golden_registers2,
-						 (const u32)ARRAY_SIZE(tahiti_golden_registers2));
+		amdgpu_device_program_register_sequence(adev,
+							tahiti_golden_registers,
+							ARRAY_SIZE(tahiti_golden_registers));
+		amdgpu_device_program_register_sequence(adev,
+							tahiti_golden_rlc_registers,
+							ARRAY_SIZE(tahiti_golden_rlc_registers));
+		amdgpu_device_program_register_sequence(adev,
+							tahiti_mgcg_cgcg_init,
+							ARRAY_SIZE(tahiti_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							tahiti_golden_registers2,
+							ARRAY_SIZE(tahiti_golden_registers2));
 		break;
 	case CHIP_PITCAIRN:
-		amdgpu_program_register_sequence(adev,
-						 pitcairn_golden_registers,
-						 (const u32)ARRAY_SIZE(pitcairn_golden_registers));
-		amdgpu_program_register_sequence(adev,
-						 pitcairn_golden_rlc_registers,
-						 (const u32)ARRAY_SIZE(pitcairn_golden_rlc_registers));
-		amdgpu_program_register_sequence(adev,
-						 pitcairn_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(pitcairn_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							pitcairn_golden_registers,
+							ARRAY_SIZE(pitcairn_golden_registers));
+		amdgpu_device_program_register_sequence(adev,
+							pitcairn_golden_rlc_registers,
+							ARRAY_SIZE(pitcairn_golden_rlc_registers));
+		amdgpu_device_program_register_sequence(adev,
+							pitcairn_mgcg_cgcg_init,
+							ARRAY_SIZE(pitcairn_mgcg_cgcg_init));
 		break;
 	case CHIP_VERDE:
-		amdgpu_program_register_sequence(adev,
-						 verde_golden_registers,
-						 (const u32)ARRAY_SIZE(verde_golden_registers));
-		amdgpu_program_register_sequence(adev,
-						 verde_golden_rlc_registers,
-						 (const u32)ARRAY_SIZE(verde_golden_rlc_registers));
-		amdgpu_program_register_sequence(adev,
-						 verde_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(verde_mgcg_cgcg_init));
-		amdgpu_program_register_sequence(adev,
-						 verde_pg_init,
-						 (const u32)ARRAY_SIZE(verde_pg_init));
+		amdgpu_device_program_register_sequence(adev,
+							verde_golden_registers,
+							ARRAY_SIZE(verde_golden_registers));
+		amdgpu_device_program_register_sequence(adev,
+							verde_golden_rlc_registers,
+							ARRAY_SIZE(verde_golden_rlc_registers));
+		amdgpu_device_program_register_sequence(adev,
+							verde_mgcg_cgcg_init,
+							ARRAY_SIZE(verde_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							verde_pg_init,
+							ARRAY_SIZE(verde_pg_init));
 		break;
 	case CHIP_OLAND:
-		amdgpu_program_register_sequence(adev,
-						 oland_golden_registers,
-						 (const u32)ARRAY_SIZE(oland_golden_registers));
-		amdgpu_program_register_sequence(adev,
-						 oland_golden_rlc_registers,
-						 (const u32)ARRAY_SIZE(oland_golden_rlc_registers));
-		amdgpu_program_register_sequence(adev,
-						 oland_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(oland_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							oland_golden_registers,
+							ARRAY_SIZE(oland_golden_registers));
+		amdgpu_device_program_register_sequence(adev,
+							oland_golden_rlc_registers,
+							ARRAY_SIZE(oland_golden_rlc_registers));
+		amdgpu_device_program_register_sequence(adev,
+							oland_mgcg_cgcg_init,
+							ARRAY_SIZE(oland_mgcg_cgcg_init));
 		break;
 	case CHIP_HAINAN:
-		amdgpu_program_register_sequence(adev,
-						 hainan_golden_registers,
-						 (const u32)ARRAY_SIZE(hainan_golden_registers));
-		amdgpu_program_register_sequence(adev,
-						 hainan_golden_registers2,
-						 (const u32)ARRAY_SIZE(hainan_golden_registers2));
-		amdgpu_program_register_sequence(adev,
-						 hainan_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(hainan_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							hainan_golden_registers,
+							ARRAY_SIZE(hainan_golden_registers));
+		amdgpu_device_program_register_sequence(adev,
+							hainan_golden_registers2,
+							ARRAY_SIZE(hainan_golden_registers2));
+		amdgpu_device_program_register_sequence(adev,
+							hainan_mgcg_cgcg_init,
+							ARRAY_SIZE(hainan_mgcg_cgcg_init));
 		break;
 
 
@@ -1460,9 +2237,8 @@ static void si_init_golden_registers(struct amdgpu_device *adev)
 static void si_pcie_gen3_enable(struct amdgpu_device *adev)
 {
 	struct pci_dev *root = adev->pdev->bus->self;
-	int bridge_pos, gpu_pos;
-	u32 speed_cntl, mask, current_data_rate;
-	int ret, i;
+	u32 speed_cntl, current_data_rate;
+	int i;
 	u16 tmp16;
 
 	if (pci_is_root_bus(adev->pdev->bus))
@@ -1474,23 +2250,20 @@ static void si_pcie_gen3_enable(struct amdgpu_device *adev)
 	if (adev->flags & AMD_IS_APU)
 		return;
 
-	ret = drm_pcie_get_speed_cap_mask(adev->ddev, &mask);
-	if (ret != 0)
+	if (!(adev->pm.pcie_gen_mask & (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2 |
+					CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3)))
 		return;
 
-	if (!(mask & (DRM_PCIE_SPEED_50 | DRM_PCIE_SPEED_80)))
-		return;
-
-	speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL);
-	current_data_rate = (speed_cntl & LC_CURRENT_DATA_RATE_MASK) >>
-		LC_CURRENT_DATA_RATE_SHIFT;
-	if (mask & DRM_PCIE_SPEED_80) {
+	speed_cntl = RREG32_PCIE_PORT(ixPCIE_LC_SPEED_CNTL);
+	current_data_rate = (speed_cntl & PCIE_LC_SPEED_CNTL__LC_CURRENT_DATA_RATE_MASK) >>
+		PCIE_LC_SPEED_CNTL__LC_CURRENT_DATA_RATE__SHIFT;
+	if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3) {
 		if (current_data_rate == 2) {
 			DRM_INFO("PCIE gen 3 link speeds already enabled\n");
 			return;
 		}
 		DRM_INFO("enabling PCIE gen 3 link speeds, disable with amdgpu.pcie_gen2=0\n");
-	} else if (mask & DRM_PCIE_SPEED_50) {
+	} else if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2) {
 		if (current_data_rate == 1) {
 			DRM_INFO("PCIE gen 2 link speeds already enabled\n");
 			return;
@@ -1498,112 +2271,111 @@ static void si_pcie_gen3_enable(struct amdgpu_device *adev)
 		DRM_INFO("enabling PCIE gen 2 link speeds, disable with amdgpu.pcie_gen2=0\n");
 	}
 
-	bridge_pos = pci_pcie_cap(root);
-	if (!bridge_pos)
-		return;
-
-	gpu_pos = pci_pcie_cap(adev->pdev);
-	if (!gpu_pos)
+	if (!pci_is_pcie(root) || !pci_is_pcie(adev->pdev))
 		return;
 
-	if (mask & DRM_PCIE_SPEED_80) {
+	if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3) {
 		if (current_data_rate != 2) {
 			u16 bridge_cfg, gpu_cfg;
 			u16 bridge_cfg2, gpu_cfg2;
 			u32 max_lw, current_lw, tmp;
 
-			pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL, &bridge_cfg);
-			pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, &gpu_cfg);
+			pcie_capability_set_word(root, PCI_EXP_LNKCTL, PCI_EXP_LNKCTL_HAWD);
+			pcie_capability_set_word(adev->pdev, PCI_EXP_LNKCTL, PCI_EXP_LNKCTL_HAWD);
 
-			tmp16 = bridge_cfg | PCI_EXP_LNKCTL_HAWD;
-			pci_write_config_word(root, bridge_pos + PCI_EXP_LNKCTL, tmp16);
-
-			tmp16 = gpu_cfg | PCI_EXP_LNKCTL_HAWD;
-			pci_write_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, tmp16);
-
-			tmp = RREG32_PCIE(PCIE_LC_STATUS1);
-			max_lw = (tmp & LC_DETECTED_LINK_WIDTH_MASK) >> LC_DETECTED_LINK_WIDTH_SHIFT;
-			current_lw = (tmp & LC_OPERATING_LINK_WIDTH_MASK) >> LC_OPERATING_LINK_WIDTH_SHIFT;
+			tmp = RREG32_PCIE(ixPCIE_LC_STATUS1);
+			max_lw = (tmp & PCIE_LC_STATUS1__LC_DETECTED_LINK_WIDTH_MASK) >> PCIE_LC_STATUS1__LC_DETECTED_LINK_WIDTH__SHIFT;
+			current_lw = (tmp & PCIE_LC_STATUS1__LC_OPERATING_LINK_WIDTH_MASK) >> PCIE_LC_STATUS1__LC_OPERATING_LINK_WIDTH__SHIFT;
 
 			if (current_lw < max_lw) {
-				tmp = RREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL);
-				if (tmp & LC_RENEGOTIATION_SUPPORT) {
-					tmp &= ~(LC_LINK_WIDTH_MASK | LC_UPCONFIGURE_DIS);
-					tmp |= (max_lw << LC_LINK_WIDTH_SHIFT);
-					tmp |= LC_UPCONFIGURE_SUPPORT | LC_RENEGOTIATE_EN | LC_RECONFIG_NOW;
-					WREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL, tmp);
+				tmp = RREG32_PCIE_PORT(ixPCIE_LC_LINK_WIDTH_CNTL);
+				if (tmp & PCIE_LC_LINK_WIDTH_CNTL__LC_RENEGOTIATION_SUPPORT_MASK) {
+					tmp &= ~(PCIE_LC_LINK_WIDTH_CNTL__LC_LINK_WIDTH_MASK | PCIE_LC_LINK_WIDTH_CNTL__LC_UPCONFIGURE_DIS_MASK);
+					tmp |= (max_lw << PCIE_LC_LINK_WIDTH_CNTL__LC_LINK_WIDTH__SHIFT);
+					tmp |= PCIE_LC_LINK_WIDTH_CNTL__LC_UPCONFIGURE_SUPPORT_MASK | PCIE_LC_LINK_WIDTH_CNTL__LC_RENEGOTIATE_EN_MASK | PCIE_LC_LINK_WIDTH_CNTL__LC_RECONFIG_NOW_MASK;
+					WREG32_PCIE_PORT(ixPCIE_LC_LINK_WIDTH_CNTL, tmp);
 				}
 			}
 
 			for (i = 0; i < 10; i++) {
-				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_DEVSTA, &tmp16);
+				pcie_capability_read_word(adev->pdev,
+							  PCI_EXP_DEVSTA,
+							  &tmp16);
 				if (tmp16 & PCI_EXP_DEVSTA_TRPND)
 					break;
 
-				pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL, &bridge_cfg);
-				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, &gpu_cfg);
+				pcie_capability_read_word(root, PCI_EXP_LNKCTL,
+							  &bridge_cfg);
+				pcie_capability_read_word(adev->pdev,
+							  PCI_EXP_LNKCTL,
+							  &gpu_cfg);
 
-				pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL2, &bridge_cfg2);
-				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, &gpu_cfg2);
+				pcie_capability_read_word(root, PCI_EXP_LNKCTL2,
+							  &bridge_cfg2);
+				pcie_capability_read_word(adev->pdev,
+							  PCI_EXP_LNKCTL2,
+							  &gpu_cfg2);
 
-				tmp = RREG32_PCIE_PORT(PCIE_LC_CNTL4);
-				tmp |= LC_SET_QUIESCE;
-				WREG32_PCIE_PORT(PCIE_LC_CNTL4, tmp);
+				tmp = RREG32_PCIE_PORT(ixPCIE_LC_CNTL4);
+				tmp |= PCIE_LC_CNTL4__LC_SET_QUIESCE_MASK;
+				WREG32_PCIE_PORT(ixPCIE_LC_CNTL4, tmp);
 
-				tmp = RREG32_PCIE_PORT(PCIE_LC_CNTL4);
-				tmp |= LC_REDO_EQ;
-				WREG32_PCIE_PORT(PCIE_LC_CNTL4, tmp);
+				tmp = RREG32_PCIE_PORT(ixPCIE_LC_CNTL4);
+				tmp |= PCIE_LC_CNTL4__LC_REDO_EQ_MASK;
+				WREG32_PCIE_PORT(ixPCIE_LC_CNTL4, tmp);
 
 				mdelay(100);
 
-				pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL, &tmp16);
-				tmp16 &= ~PCI_EXP_LNKCTL_HAWD;
-				tmp16 |= (bridge_cfg & PCI_EXP_LNKCTL_HAWD);
-				pci_write_config_word(root, bridge_pos + PCI_EXP_LNKCTL, tmp16);
-
-				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, &tmp16);
-				tmp16 &= ~PCI_EXP_LNKCTL_HAWD;
-				tmp16 |= (gpu_cfg & PCI_EXP_LNKCTL_HAWD);
-				pci_write_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, tmp16);
-
-				pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL2, &tmp16);
-				tmp16 &= ~((1 << 4) | (7 << 9));
-				tmp16 |= (bridge_cfg2 & ((1 << 4) | (7 << 9)));
-				pci_write_config_word(root, bridge_pos + PCI_EXP_LNKCTL2, tmp16);
-
-				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, &tmp16);
-				tmp16 &= ~((1 << 4) | (7 << 9));
-				tmp16 |= (gpu_cfg2 & ((1 << 4) | (7 << 9)));
-				pci_write_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, tmp16);
-
-				tmp = RREG32_PCIE_PORT(PCIE_LC_CNTL4);
-				tmp &= ~LC_SET_QUIESCE;
-				WREG32_PCIE_PORT(PCIE_LC_CNTL4, tmp);
+				pcie_capability_clear_and_set_word(root, PCI_EXP_LNKCTL,
+								   PCI_EXP_LNKCTL_HAWD,
+								   bridge_cfg &
+								   PCI_EXP_LNKCTL_HAWD);
+				pcie_capability_clear_and_set_word(adev->pdev, PCI_EXP_LNKCTL,
+								   PCI_EXP_LNKCTL_HAWD,
+								   gpu_cfg &
+								   PCI_EXP_LNKCTL_HAWD);
+
+				pcie_capability_clear_and_set_word(root, PCI_EXP_LNKCTL2,
+								   PCI_EXP_LNKCTL2_ENTER_COMP |
+								   PCI_EXP_LNKCTL2_TX_MARGIN,
+								   bridge_cfg2 &
+								   (PCI_EXP_LNKCTL2_ENTER_COMP |
+								    PCI_EXP_LNKCTL2_TX_MARGIN));
+				pcie_capability_clear_and_set_word(adev->pdev, PCI_EXP_LNKCTL2,
+								   PCI_EXP_LNKCTL2_ENTER_COMP |
+								   PCI_EXP_LNKCTL2_TX_MARGIN,
+								   gpu_cfg2 &
+								   (PCI_EXP_LNKCTL2_ENTER_COMP |
+								    PCI_EXP_LNKCTL2_TX_MARGIN));
+
+				tmp = RREG32_PCIE_PORT(ixPCIE_LC_CNTL4);
+				tmp &= ~PCIE_LC_CNTL4__LC_SET_QUIESCE_MASK;
+				WREG32_PCIE_PORT(ixPCIE_LC_CNTL4, tmp);
 			}
 		}
 	}
 
-	speed_cntl |= LC_FORCE_EN_SW_SPEED_CHANGE | LC_FORCE_DIS_HW_SPEED_CHANGE;
-	speed_cntl &= ~LC_FORCE_DIS_SW_SPEED_CHANGE;
-	WREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL, speed_cntl);
+	speed_cntl |= PCIE_LC_SPEED_CNTL__LC_FORCE_EN_SW_SPEED_CHANGE_MASK | PCIE_LC_SPEED_CNTL__LC_FORCE_DIS_HW_SPEED_CHANGE_MASK;
+	speed_cntl &= ~PCIE_LC_SPEED_CNTL__LC_FORCE_DIS_SW_SPEED_CHANGE_MASK;
+	WREG32_PCIE_PORT(ixPCIE_LC_SPEED_CNTL, speed_cntl);
 
-	pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, &tmp16);
-	tmp16 &= ~0xf;
-	if (mask & DRM_PCIE_SPEED_80)
-		tmp16 |= 3;
-	else if (mask & DRM_PCIE_SPEED_50)
-		tmp16 |= 2;
+	tmp16 = 0;
+	if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3)
+		tmp16 |= PCI_EXP_LNKCTL2_TLS_8_0GT; /* gen3 */
+	else if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2)
+		tmp16 |= PCI_EXP_LNKCTL2_TLS_5_0GT; /* gen2 */
 	else
-		tmp16 |= 1;
-	pci_write_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, tmp16);
+		tmp16 |= PCI_EXP_LNKCTL2_TLS_2_5GT; /* gen1 */
+	pcie_capability_clear_and_set_word(adev->pdev, PCI_EXP_LNKCTL2,
+					   PCI_EXP_LNKCTL2_TLS, tmp16);
 
-	speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL);
-	speed_cntl |= LC_INITIATE_LINK_SPEED_CHANGE;
-	WREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL, speed_cntl);
+	speed_cntl = RREG32_PCIE_PORT(ixPCIE_LC_SPEED_CNTL);
+	speed_cntl |= PCIE_LC_SPEED_CNTL__LC_INITIATE_LINK_SPEED_CHANGE_MASK;
+	WREG32_PCIE_PORT(ixPCIE_LC_SPEED_CNTL, speed_cntl);
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL);
-		if ((speed_cntl & LC_INITIATE_LINK_SPEED_CHANGE) == 0)
+		speed_cntl = RREG32_PCIE_PORT(ixPCIE_LC_SPEED_CNTL);
+		if ((speed_cntl & PCIE_LC_SPEED_CNTL__LC_INITIATE_LINK_SPEED_CHANGE_MASK) == 0)
 			break;
 		udelay(1);
 	}
@@ -1658,126 +2430,124 @@ static void si_program_aspm(struct amdgpu_device *adev)
 	bool disable_l0s = false, disable_l1 = false, disable_plloff_in_l1 = false;
 	bool disable_clkreq = false;
 
-	if (amdgpu_aspm == 0)
+	if (!amdgpu_device_should_use_aspm(adev))
 		return;
 
-	if (adev->flags & AMD_IS_APU)
-		return;
-	orig = data = RREG32_PCIE_PORT(PCIE_LC_N_FTS_CNTL);
-	data &= ~LC_XMIT_N_FTS_MASK;
-	data |= LC_XMIT_N_FTS(0x24) | LC_XMIT_N_FTS_OVERRIDE_EN;
+	orig = data = RREG32_PCIE_PORT(ixPCIE_LC_N_FTS_CNTL);
+	data &= ~PCIE_LC_N_FTS_CNTL__LC_XMIT_N_FTS_MASK;
+	data |= (0x24 << PCIE_LC_N_FTS_CNTL__LC_XMIT_N_FTS__SHIFT) | PCIE_LC_N_FTS_CNTL__LC_XMIT_N_FTS_OVERRIDE_EN_MASK;
 	if (orig != data)
-		WREG32_PCIE_PORT(PCIE_LC_N_FTS_CNTL, data);
+		WREG32_PCIE_PORT(ixPCIE_LC_N_FTS_CNTL, data);
 
-	orig = data = RREG32_PCIE_PORT(PCIE_LC_CNTL3);
-	data |= LC_GO_TO_RECOVERY;
+	orig = data = RREG32_PCIE_PORT(ixPCIE_LC_CNTL3);
+	data |= PCIE_LC_CNTL3__LC_GO_TO_RECOVERY_MASK;
 	if (orig != data)
-		WREG32_PCIE_PORT(PCIE_LC_CNTL3, data);
+		WREG32_PCIE_PORT(ixPCIE_LC_CNTL3, data);
 
-	orig = data = RREG32_PCIE(PCIE_P_CNTL);
-	data |= P_IGNORE_EDB_ERR;
+	orig = data = RREG32_PCIE(ixPCIE_P_CNTL);
+	data |= PCIE_P_CNTL__P_IGNORE_EDB_ERR_MASK;
 	if (orig != data)
-		WREG32_PCIE(PCIE_P_CNTL, data);
+		WREG32_PCIE(ixPCIE_P_CNTL, data);
 
-	orig = data = RREG32_PCIE_PORT(PCIE_LC_CNTL);
-	data &= ~(LC_L0S_INACTIVITY_MASK | LC_L1_INACTIVITY_MASK);
-	data |= LC_PMI_TO_L1_DIS;
+	orig = data = RREG32_PCIE_PORT(ixPCIE_LC_CNTL);
+	data &= ~(PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK | PCIE_LC_CNTL__LC_L1_INACTIVITY_MASK);
+	data |= PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK;
 	if (!disable_l0s)
-		data |= LC_L0S_INACTIVITY(7);
+		data |= (7 << PCIE_LC_CNTL__LC_L0S_INACTIVITY__SHIFT);
 
 	if (!disable_l1) {
-		data |= LC_L1_INACTIVITY(7);
-		data &= ~LC_PMI_TO_L1_DIS;
+		data |= (7 << PCIE_LC_CNTL__LC_L1_INACTIVITY__SHIFT);
+		data &= ~PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK;
 		if (orig != data)
-			WREG32_PCIE_PORT(PCIE_LC_CNTL, data);
+			WREG32_PCIE_PORT(ixPCIE_LC_CNTL, data);
 
 		if (!disable_plloff_in_l1) {
 			bool clk_req_support;
 
-			orig = data = si_pif_phy0_rreg(adev,PB0_PIF_PWRDOWN_0);
-			data &= ~(PLL_POWER_STATE_IN_OFF_0_MASK | PLL_POWER_STATE_IN_TXS2_0_MASK);
-			data |= PLL_POWER_STATE_IN_OFF_0(7) | PLL_POWER_STATE_IN_TXS2_0(7);
+			orig = data = si_pif_phy0_rreg(adev,ixPB0_PIF_PWRDOWN_0);
+			data &= ~(PB0_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_OFF_0_MASK | PB0_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_TXS2_0_MASK);
+			data |= (7 << PB0_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_OFF_0__SHIFT) | (7 << PB0_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_TXS2_0__SHIFT);
 			if (orig != data)
-				si_pif_phy0_wreg(adev,PB0_PIF_PWRDOWN_0, data);
+				si_pif_phy0_wreg(adev,ixPB0_PIF_PWRDOWN_0, data);
 
-			orig = data = si_pif_phy0_rreg(adev,PB0_PIF_PWRDOWN_1);
-			data &= ~(PLL_POWER_STATE_IN_OFF_1_MASK | PLL_POWER_STATE_IN_TXS2_1_MASK);
-			data |= PLL_POWER_STATE_IN_OFF_1(7) | PLL_POWER_STATE_IN_TXS2_1(7);
+			orig = data = si_pif_phy0_rreg(adev,ixPB0_PIF_PWRDOWN_1);
+			data &= ~(PB0_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_OFF_1_MASK | PB0_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_TXS2_1_MASK);
+			data |= (7 << PB0_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_OFF_1__SHIFT) | (7 << PB0_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_TXS2_1__SHIFT);
 			if (orig != data)
-				si_pif_phy0_wreg(adev,PB0_PIF_PWRDOWN_1, data);
+				si_pif_phy0_wreg(adev,ixPB0_PIF_PWRDOWN_1, data);
 
-			orig = data = si_pif_phy1_rreg(adev,PB1_PIF_PWRDOWN_0);
-			data &= ~(PLL_POWER_STATE_IN_OFF_0_MASK | PLL_POWER_STATE_IN_TXS2_0_MASK);
-			data |= PLL_POWER_STATE_IN_OFF_0(7) | PLL_POWER_STATE_IN_TXS2_0(7);
+			orig = data = si_pif_phy1_rreg(adev,ixPB1_PIF_PWRDOWN_0);
+			data &= ~(PB1_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_OFF_0_MASK | PB1_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_TXS2_0_MASK);
+			data |= (7 << PB1_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_OFF_0__SHIFT) | (7 << PB1_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_TXS2_0__SHIFT);
 			if (orig != data)
-				si_pif_phy1_wreg(adev,PB1_PIF_PWRDOWN_0, data);
+				si_pif_phy1_wreg(adev,ixPB1_PIF_PWRDOWN_0, data);
 
-			orig = data = si_pif_phy1_rreg(adev,PB1_PIF_PWRDOWN_1);
-			data &= ~(PLL_POWER_STATE_IN_OFF_1_MASK | PLL_POWER_STATE_IN_TXS2_1_MASK);
-			data |= PLL_POWER_STATE_IN_OFF_1(7) | PLL_POWER_STATE_IN_TXS2_1(7);
+			orig = data = si_pif_phy1_rreg(adev,ixPB1_PIF_PWRDOWN_1);
+			data &= ~(PB1_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_OFF_1_MASK | PB1_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_TXS2_1_MASK);
+			data |= (7 << PB1_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_OFF_1__SHIFT) | (7 << PB1_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_TXS2_1__SHIFT);
 			if (orig != data)
-				si_pif_phy1_wreg(adev,PB1_PIF_PWRDOWN_1, data);
+				si_pif_phy1_wreg(adev,ixPB1_PIF_PWRDOWN_1, data);
 
-			if ((adev->family != CHIP_OLAND) && (adev->family != CHIP_HAINAN)) {
-				orig = data = si_pif_phy0_rreg(adev,PB0_PIF_PWRDOWN_0);
-				data &= ~PLL_RAMP_UP_TIME_0_MASK;
+			if ((adev->asic_type != CHIP_OLAND) && (adev->asic_type != CHIP_HAINAN)) {
+				orig = data = si_pif_phy0_rreg(adev,ixPB0_PIF_PWRDOWN_0);
+				data &= ~PB0_PIF_PWRDOWN_0__PLL_RAMP_UP_TIME_0_MASK;
 				if (orig != data)
-					si_pif_phy0_wreg(adev,PB0_PIF_PWRDOWN_0, data);
+					si_pif_phy0_wreg(adev,ixPB0_PIF_PWRDOWN_0, data);
 
-				orig = data = si_pif_phy0_rreg(adev,PB0_PIF_PWRDOWN_1);
-				data &= ~PLL_RAMP_UP_TIME_1_MASK;
+				orig = data = si_pif_phy0_rreg(adev,ixPB0_PIF_PWRDOWN_1);
+				data &= ~PB0_PIF_PWRDOWN_1__PLL_RAMP_UP_TIME_1_MASK;
 				if (orig != data)
-					si_pif_phy0_wreg(adev,PB0_PIF_PWRDOWN_1, data);
+					si_pif_phy0_wreg(adev,ixPB0_PIF_PWRDOWN_1, data);
 
-				orig = data = si_pif_phy0_rreg(adev,PB0_PIF_PWRDOWN_2);
-				data &= ~PLL_RAMP_UP_TIME_2_MASK;
+				orig = data = si_pif_phy0_rreg(adev,ixPB0_PIF_PWRDOWN_2);
+				data &= ~PB0_PIF_PWRDOWN_2__PLL_RAMP_UP_TIME_2_MASK;
 				if (orig != data)
-					si_pif_phy0_wreg(adev,PB0_PIF_PWRDOWN_2, data);
+					si_pif_phy0_wreg(adev,ixPB0_PIF_PWRDOWN_2, data);
 
-				orig = data = si_pif_phy0_rreg(adev,PB0_PIF_PWRDOWN_3);
-				data &= ~PLL_RAMP_UP_TIME_3_MASK;
+				orig = data = si_pif_phy0_rreg(adev,ixPB0_PIF_PWRDOWN_3);
+				data &= ~PB0_PIF_PWRDOWN_3__PLL_RAMP_UP_TIME_3_MASK;
 				if (orig != data)
-					si_pif_phy0_wreg(adev,PB0_PIF_PWRDOWN_3, data);
+					si_pif_phy0_wreg(adev,ixPB0_PIF_PWRDOWN_3, data);
 
-				orig = data = si_pif_phy1_rreg(adev,PB1_PIF_PWRDOWN_0);
-				data &= ~PLL_RAMP_UP_TIME_0_MASK;
+				orig = data = si_pif_phy1_rreg(adev,ixPB1_PIF_PWRDOWN_0);
+				data &= ~PB1_PIF_PWRDOWN_0__PLL_RAMP_UP_TIME_0_MASK;
 				if (orig != data)
-					si_pif_phy1_wreg(adev,PB1_PIF_PWRDOWN_0, data);
+					si_pif_phy1_wreg(adev,ixPB1_PIF_PWRDOWN_0, data);
 
-				orig = data = si_pif_phy1_rreg(adev,PB1_PIF_PWRDOWN_1);
-				data &= ~PLL_RAMP_UP_TIME_1_MASK;
+				orig = data = si_pif_phy1_rreg(adev,ixPB1_PIF_PWRDOWN_1);
+				data &= ~PB1_PIF_PWRDOWN_1__PLL_RAMP_UP_TIME_1_MASK;
 				if (orig != data)
-					si_pif_phy1_wreg(adev,PB1_PIF_PWRDOWN_1, data);
+					si_pif_phy1_wreg(adev,ixPB1_PIF_PWRDOWN_1, data);
 
-				orig = data = si_pif_phy1_rreg(adev,PB1_PIF_PWRDOWN_2);
-				data &= ~PLL_RAMP_UP_TIME_2_MASK;
+				orig = data = si_pif_phy1_rreg(adev,ixPB1_PIF_PWRDOWN_2);
+				data &= ~PB1_PIF_PWRDOWN_2__PLL_RAMP_UP_TIME_2_MASK;
 				if (orig != data)
-					si_pif_phy1_wreg(adev,PB1_PIF_PWRDOWN_2, data);
+					si_pif_phy1_wreg(adev,ixPB1_PIF_PWRDOWN_2, data);
 
-				orig = data = si_pif_phy1_rreg(adev,PB1_PIF_PWRDOWN_3);
-				data &= ~PLL_RAMP_UP_TIME_3_MASK;
+				orig = data = si_pif_phy1_rreg(adev,ixPB1_PIF_PWRDOWN_3);
+				data &= ~PB1_PIF_PWRDOWN_3__PLL_RAMP_UP_TIME_3_MASK;
 				if (orig != data)
-					si_pif_phy1_wreg(adev,PB1_PIF_PWRDOWN_3, data);
+					si_pif_phy1_wreg(adev,ixPB1_PIF_PWRDOWN_3, data);
 			}
-			orig = data = RREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL);
-			data &= ~LC_DYN_LANES_PWR_STATE_MASK;
-			data |= LC_DYN_LANES_PWR_STATE(3);
+			orig = data = RREG32_PCIE_PORT(ixPCIE_LC_LINK_WIDTH_CNTL);
+			data &= ~PCIE_LC_LINK_WIDTH_CNTL__LC_DYN_LANES_PWR_STATE_MASK;
+			data |= (3 << PCIE_LC_LINK_WIDTH_CNTL__LC_DYN_LANES_PWR_STATE__SHIFT);
 			if (orig != data)
-				WREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL, data);
+				WREG32_PCIE_PORT(ixPCIE_LC_LINK_WIDTH_CNTL, data);
 
-			orig = data = si_pif_phy0_rreg(adev,PB0_PIF_CNTL);
-			data &= ~LS2_EXIT_TIME_MASK;
-			if ((adev->family == CHIP_OLAND) || (adev->family == CHIP_HAINAN))
-				data |= LS2_EXIT_TIME(5);
+			orig = data = si_pif_phy0_rreg(adev,ixPB0_PIF_CNTL);
+			data &= ~PB0_PIF_CNTL__LS2_EXIT_TIME_MASK;
+			if ((adev->asic_type == CHIP_OLAND) || (adev->asic_type == CHIP_HAINAN))
+				data |= (5 << PB0_PIF_CNTL__LS2_EXIT_TIME__SHIFT);
 			if (orig != data)
-				si_pif_phy0_wreg(adev,PB0_PIF_CNTL, data);
+				si_pif_phy0_wreg(adev,ixPB0_PIF_CNTL, data);
 
-			orig = data = si_pif_phy1_rreg(adev,PB1_PIF_CNTL);
-			data &= ~LS2_EXIT_TIME_MASK;
-			if ((adev->family == CHIP_OLAND) || (adev->family == CHIP_HAINAN))
-				data |= LS2_EXIT_TIME(5);
+			orig = data = si_pif_phy1_rreg(adev,ixPB1_PIF_CNTL);
+			data &= ~PB1_PIF_CNTL__LS2_EXIT_TIME_MASK;
+			if ((adev->asic_type == CHIP_OLAND) || (adev->asic_type == CHIP_HAINAN))
+				data |= (5 << PB1_PIF_CNTL__LS2_EXIT_TIME__SHIFT);
 			if (orig != data)
-				si_pif_phy1_wreg(adev,PB1_PIF_CNTL, data);
+				si_pif_phy1_wreg(adev,ixPB1_PIF_CNTL, data);
 
 			if (!disable_clkreq &&
 			    !pci_is_root_bus(adev->pdev->bus)) {
@@ -1793,64 +2563,64 @@ static void si_program_aspm(struct amdgpu_device *adev)
 			}
 
 			if (clk_req_support) {
-				orig = data = RREG32_PCIE_PORT(PCIE_LC_CNTL2);
-				data |= LC_ALLOW_PDWN_IN_L1 | LC_ALLOW_PDWN_IN_L23;
+				orig = data = RREG32_PCIE_PORT(ixPCIE_LC_CNTL2);
+				data |= PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L1_MASK | PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L23_MASK;
 				if (orig != data)
-					WREG32_PCIE_PORT(PCIE_LC_CNTL2, data);
+					WREG32_PCIE_PORT(ixPCIE_LC_CNTL2, data);
 
-				orig = data = RREG32(THM_CLK_CNTL);
-				data &= ~(CMON_CLK_SEL_MASK | TMON_CLK_SEL_MASK);
-				data |= CMON_CLK_SEL(1) | TMON_CLK_SEL(1);
+				orig = data = RREG32(mmTHM_CLK_CNTL);
+				data &= ~(THM_CLK_CNTL__CMON_CLK_SEL_MASK | THM_CLK_CNTL__TMON_CLK_SEL_MASK);
+				data |= (1 << THM_CLK_CNTL__CMON_CLK_SEL__SHIFT) | (1 << THM_CLK_CNTL__TMON_CLK_SEL__SHIFT);
 				if (orig != data)
-					WREG32(THM_CLK_CNTL, data);
+					WREG32(mmTHM_CLK_CNTL, data);
 
-				orig = data = RREG32(MISC_CLK_CNTL);
-				data &= ~(DEEP_SLEEP_CLK_SEL_MASK | ZCLK_SEL_MASK);
-				data |= DEEP_SLEEP_CLK_SEL(1) | ZCLK_SEL(1);
+				orig = data = RREG32(mmMISC_CLK_CNTL);
+				data &= ~(MISC_CLK_CNTL__DEEP_SLEEP_CLK_SEL_MASK | MISC_CLK_CNTL__ZCLK_SEL_MASK);
+				data |= (1 << MISC_CLK_CNTL__DEEP_SLEEP_CLK_SEL__SHIFT) | (1 << MISC_CLK_CNTL__ZCLK_SEL__SHIFT);
 				if (orig != data)
-					WREG32(MISC_CLK_CNTL, data);
+					WREG32(mmMISC_CLK_CNTL, data);
 
-				orig = data = RREG32(CG_CLKPIN_CNTL);
-				data &= ~BCLK_AS_XCLK;
+				orig = data = RREG32(mmCG_CLKPIN_CNTL);
+				data &= ~CG_CLKPIN_CNTL__BCLK_AS_XCLK_MASK;
 				if (orig != data)
-					WREG32(CG_CLKPIN_CNTL, data);
+					WREG32(mmCG_CLKPIN_CNTL, data);
 
-				orig = data = RREG32(CG_CLKPIN_CNTL_2);
-				data &= ~FORCE_BIF_REFCLK_EN;
+				orig = data = RREG32(mmCG_CLKPIN_CNTL_2);
+				data &= ~CG_CLKPIN_CNTL_2__FORCE_BIF_REFCLK_EN_MASK;
 				if (orig != data)
-					WREG32(CG_CLKPIN_CNTL_2, data);
+					WREG32(mmCG_CLKPIN_CNTL_2, data);
 
-				orig = data = RREG32(MPLL_BYPASSCLK_SEL);
-				data &= ~MPLL_CLKOUT_SEL_MASK;
-				data |= MPLL_CLKOUT_SEL(4);
+				orig = data = RREG32(mmMPLL_BYPASSCLK_SEL);
+				data &= ~MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL_MASK;
+				data |= 4 << MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL__SHIFT;
 				if (orig != data)
-					WREG32(MPLL_BYPASSCLK_SEL, data);
+					WREG32(mmMPLL_BYPASSCLK_SEL, data);
 
-				orig = data = RREG32(SPLL_CNTL_MODE);
-				data &= ~SPLL_REFCLK_SEL_MASK;
+				orig = data = RREG32(mmSPLL_CNTL_MODE);
+				data &= ~SPLL_CNTL_MODE__SPLL_REFCLK_SEL_MASK;
 				if (orig != data)
-					WREG32(SPLL_CNTL_MODE, data);
+					WREG32(mmSPLL_CNTL_MODE, data);
 			}
 		}
 	} else {
 		if (orig != data)
-			WREG32_PCIE_PORT(PCIE_LC_CNTL, data);
+			WREG32_PCIE_PORT(ixPCIE_LC_CNTL, data);
 	}
 
-	orig = data = RREG32_PCIE(PCIE_CNTL2);
-	data |= SLV_MEM_LS_EN | MST_MEM_LS_EN | REPLAY_MEM_LS_EN;
+	orig = data = RREG32_PCIE(ixPCIE_CNTL2);
+	data |= PCIE_CNTL2__SLV_MEM_LS_EN_MASK | PCIE_CNTL2__MST_MEM_LS_EN_MASK | PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK;
 	if (orig != data)
-		WREG32_PCIE(PCIE_CNTL2, data);
+		WREG32_PCIE(ixPCIE_CNTL2, data);
 
 	if (!disable_l0s) {
-		data = RREG32_PCIE_PORT(PCIE_LC_N_FTS_CNTL);
-		if((data & LC_N_FTS_MASK) == LC_N_FTS_MASK) {
-			data = RREG32_PCIE(PCIE_LC_STATUS1);
-			if ((data & LC_REVERSE_XMIT) && (data & LC_REVERSE_RCVR)) {
-				orig = data = RREG32_PCIE_PORT(PCIE_LC_CNTL);
-				data &= ~LC_L0S_INACTIVITY_MASK;
+		data = RREG32_PCIE_PORT(ixPCIE_LC_N_FTS_CNTL);
+		if((data & PCIE_LC_N_FTS_CNTL__LC_N_FTS_MASK) == PCIE_LC_N_FTS_CNTL__LC_N_FTS_MASK) {
+			data = RREG32_PCIE(ixPCIE_LC_STATUS1);
+			if ((data & PCIE_LC_STATUS1__LC_REVERSE_XMIT_MASK) && (data & PCIE_LC_STATUS1__LC_REVERSE_RCVR_MASK)) {
+				orig = data = RREG32_PCIE_PORT(ixPCIE_LC_CNTL);
+				data &= ~PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK;
 				if (orig != data)
-					WREG32_PCIE_PORT(PCIE_LC_CNTL, data);
+					WREG32_PCIE_PORT(ixPCIE_LC_CNTL, data);
 			}
 		}
 	}
@@ -1867,9 +2637,9 @@ static void si_fix_pci_max_read_req_size(struct amdgpu_device *adev)
 		pcie_set_readrq(adev->pdev, 512);
 }
 
-static int si_common_hw_init(void *handle)
+static int si_common_hw_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	si_fix_pci_max_read_req_size(adev);
 	si_init_golden_registers(adev);
@@ -1879,47 +2649,28 @@ static int si_common_hw_init(void *handle)
 	return 0;
 }
 
-static int si_common_hw_fini(void *handle)
+static int si_common_hw_fini(struct amdgpu_ip_block *ip_block)
 {
 	return 0;
 }
 
-static int si_common_suspend(void *handle)
+static int si_common_resume(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return si_common_hw_fini(adev);
-}
-
-static int si_common_resume(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return si_common_hw_init(adev);
+	return si_common_hw_init(ip_block);
 }
 
-static bool si_common_is_idle(void *handle)
+static bool si_common_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	return true;
 }
 
-static int si_common_wait_for_idle(void *handle)
-{
-	return 0;
-}
-
-static int si_common_soft_reset(void *handle)
-{
-	return 0;
-}
-
-static int si_common_set_clockgating_state(void *handle,
+static int si_common_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					    enum amd_clockgating_state state)
 {
 	return 0;
 }
 
-static int si_common_set_powergating_state(void *handle,
+static int si_common_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					    enum amd_powergating_state state)
 {
 	return 0;
@@ -1928,16 +2679,10 @@ static int si_common_set_powergating_state(void *handle,
 static const struct amd_ip_funcs si_common_ip_funcs = {
 	.name = "si_common",
 	.early_init = si_common_early_init,
-	.late_init = NULL,
-	.sw_init = si_common_sw_init,
-	.sw_fini = si_common_sw_fini,
 	.hw_init = si_common_hw_init,
 	.hw_fini = si_common_hw_fini,
-	.suspend = si_common_suspend,
 	.resume = si_common_resume,
 	.is_idle = si_common_is_idle,
-	.wait_for_idle = si_common_wait_for_idle,
-	.soft_reset = si_common_soft_reset,
 	.set_clockgating_state = si_common_set_clockgating_state,
 	.set_powergating_state = si_common_set_powergating_state,
 };
@@ -1953,48 +2698,54 @@ static const struct amdgpu_ip_block_version si_common_ip_block =
 
 int si_set_ip_blocks(struct amdgpu_device *adev)
 {
-	si_detect_hw_virtualization(adev);
-
 	switch (adev->asic_type) {
 	case CHIP_VERDE:
 	case CHIP_TAHITI:
 	case CHIP_PITCAIRN:
-		amdgpu_ip_block_add(adev, &si_common_ip_block);
-		amdgpu_ip_block_add(adev, &gmc_v6_0_ip_block);
-		amdgpu_ip_block_add(adev, &si_ih_ip_block);
-		amdgpu_ip_block_add(adev, &amdgpu_pp_ip_block);
+		amdgpu_device_ip_block_add(adev, &si_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v6_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &si_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v6_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &si_dma_ip_block);
+		amdgpu_device_ip_block_add(adev, &si_smu_ip_block);
 		if (adev->enable_virtual_display)
-			amdgpu_ip_block_add(adev, &dce_virtual_ip_block);
+			amdgpu_device_ip_block_add(adev, &amdgpu_vkms_ip_block);
+#if defined(CONFIG_DRM_AMD_DC) && defined(CONFIG_DRM_AMD_DC_SI)
+		else if (amdgpu_device_has_dc_support(adev))
+			amdgpu_device_ip_block_add(adev, &dm_ip_block);
+#endif
 		else
-			amdgpu_ip_block_add(adev, &dce_v6_0_ip_block);
-		amdgpu_ip_block_add(adev, &gfx_v6_0_ip_block);
-		amdgpu_ip_block_add(adev, &si_dma_ip_block);
-		/* amdgpu_ip_block_add(adev, &uvd_v3_1_ip_block); */
-		/* amdgpu_ip_block_add(adev, &vce_v1_0_ip_block); */
+			amdgpu_device_ip_block_add(adev, &dce_v6_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &uvd_v3_1_ip_block);
+		/* amdgpu_device_ip_block_add(adev, &vce_v1_0_ip_block); */
 		break;
 	case CHIP_OLAND:
-		amdgpu_ip_block_add(adev, &si_common_ip_block);
-		amdgpu_ip_block_add(adev, &gmc_v6_0_ip_block);
-		amdgpu_ip_block_add(adev, &si_ih_ip_block);
-		amdgpu_ip_block_add(adev, &amdgpu_pp_ip_block);
+		amdgpu_device_ip_block_add(adev, &si_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v6_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &si_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v6_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &si_dma_ip_block);
+		amdgpu_device_ip_block_add(adev, &si_smu_ip_block);
 		if (adev->enable_virtual_display)
-			amdgpu_ip_block_add(adev, &dce_virtual_ip_block);
+			amdgpu_device_ip_block_add(adev, &amdgpu_vkms_ip_block);
+#if defined(CONFIG_DRM_AMD_DC) && defined(CONFIG_DRM_AMD_DC_SI)
+		else if (amdgpu_device_has_dc_support(adev))
+			amdgpu_device_ip_block_add(adev, &dm_ip_block);
+#endif
 		else
-			amdgpu_ip_block_add(adev, &dce_v6_4_ip_block);
-		amdgpu_ip_block_add(adev, &gfx_v6_0_ip_block);
-		amdgpu_ip_block_add(adev, &si_dma_ip_block);
-		/* amdgpu_ip_block_add(adev, &uvd_v3_1_ip_block); */
-		/* amdgpu_ip_block_add(adev, &vce_v1_0_ip_block); */
+			amdgpu_device_ip_block_add(adev, &dce_v6_4_ip_block);
+		amdgpu_device_ip_block_add(adev, &uvd_v3_1_ip_block);
+		/* amdgpu_device_ip_block_add(adev, &vce_v1_0_ip_block); */
 		break;
 	case CHIP_HAINAN:
-		amdgpu_ip_block_add(adev, &si_common_ip_block);
-		amdgpu_ip_block_add(adev, &gmc_v6_0_ip_block);
-		amdgpu_ip_block_add(adev, &si_ih_ip_block);
-		amdgpu_ip_block_add(adev, &amdgpu_pp_ip_block);
+		amdgpu_device_ip_block_add(adev, &si_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v6_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &si_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v6_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &si_dma_ip_block);
+		amdgpu_device_ip_block_add(adev, &si_smu_ip_block);
 		if (adev->enable_virtual_display)
-			amdgpu_ip_block_add(adev, &dce_virtual_ip_block);
-		amdgpu_ip_block_add(adev, &gfx_v6_0_ip_block);
-		amdgpu_ip_block_add(adev, &si_dma_ip_block);
+			amdgpu_device_ip_block_add(adev, &amdgpu_vkms_ip_block);
 		break;
 	default:
 		BUG();
diff --git a/drivers/gpu/drm/amd/amdgpu/si.h b/drivers/gpu/drm/amd/amdgpu/si.h
index 589225080c24..06ed7212a0d6 100644
--- a/drivers/gpu/drm/amd/amdgpu/si.h
+++ b/drivers/gpu/drm/amd/amdgpu/si.h
@@ -24,6 +24,8 @@
 #ifndef __SI_H__
 #define __SI_H__
 
+#define SI_FLUSH_GPU_TLB_NUM_WREG	2
+
 void si_srbm_select(struct amdgpu_device *adev,
 		     u32 me, u32 pipe, u32 queue, u32 vmid);
 int si_set_ip_blocks(struct amdgpu_device *adev);
diff --git a/drivers/gpu/drm/amd/amdgpu/si_dma.c b/drivers/gpu/drm/amd/amdgpu/si_dma.c
index 112969f3301a..7f18e4875287 100644
--- a/drivers/gpu/drm/amd/amdgpu/si_dma.c
+++ b/drivers/gpu/drm/amd/amdgpu/si_dma.c
@@ -21,11 +21,14 @@
  *
  * Authors: Alex Deucher
  */
-#include <drm/drmP.h>
+
 #include "amdgpu.h"
 #include "amdgpu_trace.h"
+#include "si.h"
 #include "sid.h"
 
+#include "oss/oss_1_0_d.h"
+#include "oss/oss_1_0_sh_mask.h"
 const u32 sdma_offsets[SDMA_MAX_INSTANCE] =
 {
 	DMA0_REGISTER_OFFSET,
@@ -37,17 +40,31 @@ static void si_dma_set_buffer_funcs(struct amdgpu_device *adev);
 static void si_dma_set_vm_pte_funcs(struct amdgpu_device *adev);
 static void si_dma_set_irq_funcs(struct amdgpu_device *adev);
 
+/**
+ * si_dma_ring_get_rptr - get the current read pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current rptr from the hardware (SI).
+ */
 static uint64_t si_dma_ring_get_rptr(struct amdgpu_ring *ring)
 {
-	return ring->adev->wb.wb[ring->rptr_offs>>2];
+	return *ring->rptr_cpu_addr;
 }
 
+/**
+ * si_dma_ring_get_wptr - get the current write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current wptr from the hardware (SI).
+ */
 static uint64_t si_dma_ring_get_wptr(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
 	u32 me = (ring == &adev->sdma.instance[0].ring) ? 0 : 1;
 
-	return (RREG32(DMA_RB_WPTR + sdma_offsets[me]) & 0x3fffc) >> 2;
+	return (RREG32(mmDMA_GFX_RB_WPTR + sdma_offsets[me]) & 0x3fffc) >> 2;
 }
 
 static void si_dma_ring_set_wptr(struct amdgpu_ring *ring)
@@ -55,44 +72,33 @@ static void si_dma_ring_set_wptr(struct amdgpu_ring *ring)
 	struct amdgpu_device *adev = ring->adev;
 	u32 me = (ring == &adev->sdma.instance[0].ring) ? 0 : 1;
 
-	WREG32(DMA_RB_WPTR + sdma_offsets[me],
-	       (lower_32_bits(ring->wptr) << 2) & 0x3fffc);
+	WREG32(mmDMA_GFX_RB_WPTR + sdma_offsets[me], (ring->wptr << 2) & 0x3fffc);
 }
 
 static void si_dma_ring_emit_ib(struct amdgpu_ring *ring,
+				struct amdgpu_job *job,
 				struct amdgpu_ib *ib,
-				unsigned vm_id, bool ctx_switch)
+				uint32_t flags)
 {
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
 	/* The indirect buffer packet must end on an 8 DW boundary in the DMA ring.
 	 * Pad as necessary with NOPs.
 	 */
 	while ((lower_32_bits(ring->wptr) & 7) != 5)
 		amdgpu_ring_write(ring, DMA_PACKET(DMA_PACKET_NOP, 0, 0, 0, 0));
-	amdgpu_ring_write(ring, DMA_IB_PACKET(DMA_PACKET_INDIRECT_BUFFER, vm_id, 0));
+	amdgpu_ring_write(ring, DMA_IB_PACKET(DMA_PACKET_INDIRECT_BUFFER, vmid, 0));
 	amdgpu_ring_write(ring, (ib->gpu_addr & 0xFFFFFFE0));
 	amdgpu_ring_write(ring, (ib->length_dw << 12) | (upper_32_bits(ib->gpu_addr) & 0xFF));
 
 }
 
-static void si_dma_ring_emit_hdp_flush(struct amdgpu_ring *ring)
-{
-	amdgpu_ring_write(ring, DMA_PACKET(DMA_PACKET_SRBM_WRITE, 0, 0, 0, 0));
-	amdgpu_ring_write(ring, (0xf << 16) | (HDP_MEM_COHERENCY_FLUSH_CNTL));
-	amdgpu_ring_write(ring, 1);
-}
-
-static void si_dma_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
-{
-	amdgpu_ring_write(ring, DMA_PACKET(DMA_PACKET_SRBM_WRITE, 0, 0, 0, 0));
-	amdgpu_ring_write(ring, (0xf << 16) | (HDP_DEBUG0));
-	amdgpu_ring_write(ring, 1);
-}
-
 /**
  * si_dma_ring_emit_fence - emit a fence on the DMA ring
  *
  * @ring: amdgpu ring pointer
- * @fence: amdgpu fence object
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
  *
  * Add a DMA fence packet to the ring to write
  * the fence seq number and DMA trap packet to generate
@@ -122,20 +128,14 @@ static void si_dma_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
 
 static void si_dma_stop(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *ring;
 	u32 rb_cntl;
 	unsigned i;
 
 	for (i = 0; i < adev->sdma.num_instances; i++) {
-		ring = &adev->sdma.instance[i].ring;
 		/* dma0 */
-		rb_cntl = RREG32(DMA_RB_CNTL + sdma_offsets[i]);
-		rb_cntl &= ~DMA_RB_ENABLE;
-		WREG32(DMA_RB_CNTL + sdma_offsets[i], rb_cntl);
-
-		if (adev->mman.buffer_funcs_ring == ring)
-			amdgpu_ttm_set_active_vram_size(adev, adev->mc.visible_vram_size);
-		ring->ready = false;
+		rb_cntl = RREG32(mmDMA_GFX_RB_CNTL + sdma_offsets[i]);
+		rb_cntl &= ~DMA_GFX_RB_CNTL__RB_ENABLE_MASK;
+		WREG32(mmDMA_GFX_RB_CNTL + sdma_offsets[i], rb_cntl);
 	}
 }
 
@@ -149,55 +149,48 @@ static int si_dma_start(struct amdgpu_device *adev)
 	for (i = 0; i < adev->sdma.num_instances; i++) {
 		ring = &adev->sdma.instance[i].ring;
 
-		WREG32(DMA_SEM_INCOMPLETE_TIMER_CNTL + sdma_offsets[i], 0);
-		WREG32(DMA_SEM_WAIT_FAIL_TIMER_CNTL + sdma_offsets[i], 0);
+		WREG32(mmDMA_SEM_INCOMPLETE_TIMER_CNTL + sdma_offsets[i], 0);
+		WREG32(mmDMA_SEM_WAIT_FAIL_TIMER_CNTL + sdma_offsets[i], 0);
 
 		/* Set ring buffer size in dwords */
 		rb_bufsz = order_base_2(ring->ring_size / 4);
 		rb_cntl = rb_bufsz << 1;
 #ifdef __BIG_ENDIAN
-		rb_cntl |= DMA_RB_SWAP_ENABLE | DMA_RPTR_WRITEBACK_SWAP_ENABLE;
+		rb_cntl |= DMA_GFX_RB_CNTL__RB_SWAP_ENABLE_MASK | DMA_GFX_RB_CNTL__RPTR_WRITEBACK_SWAP_ENABLE_MASK;
 #endif
-		WREG32(DMA_RB_CNTL + sdma_offsets[i], rb_cntl);
+		WREG32(mmDMA_GFX_RB_CNTL + sdma_offsets[i], rb_cntl);
 
 		/* Initialize the ring buffer's read and write pointers */
-		WREG32(DMA_RB_RPTR + sdma_offsets[i], 0);
-		WREG32(DMA_RB_WPTR + sdma_offsets[i], 0);
+		WREG32(mmDMA_GFX_RB_RPTR + sdma_offsets[i], 0);
+		WREG32(mmDMA_GFX_RB_WPTR + sdma_offsets[i], 0);
 
-		rptr_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+		rptr_addr = ring->rptr_gpu_addr;
 
-		WREG32(DMA_RB_RPTR_ADDR_LO + sdma_offsets[i], lower_32_bits(rptr_addr));
-		WREG32(DMA_RB_RPTR_ADDR_HI + sdma_offsets[i], upper_32_bits(rptr_addr) & 0xFF);
+		WREG32(mmDMA_GFX_RB_RPTR_ADDR_LO + sdma_offsets[i], lower_32_bits(rptr_addr));
+		WREG32(mmDMA_GFX_RB_RPTR_ADDR_HI + sdma_offsets[i], upper_32_bits(rptr_addr) & 0xFF);
 
-		rb_cntl |= DMA_RPTR_WRITEBACK_ENABLE;
+		rb_cntl |= DMA_GFX_RB_CNTL__RPTR_WRITEBACK_ENABLE_MASK;
 
-		WREG32(DMA_RB_BASE + sdma_offsets[i], ring->gpu_addr >> 8);
+		WREG32(mmDMA_GFX_RB_BASE + sdma_offsets[i], ring->gpu_addr >> 8);
 
 		/* enable DMA IBs */
-		ib_cntl = DMA_IB_ENABLE | CMD_VMID_FORCE;
+		ib_cntl = DMA_GFX_IB_CNTL__IB_ENABLE_MASK | DMA_GFX_IB_CNTL__CMD_VMID_FORCE_MASK;
 #ifdef __BIG_ENDIAN
-		ib_cntl |= DMA_IB_SWAP_ENABLE;
+		ib_cntl |= DMA_GFX_IB_CNTL__IB_SWAP_ENABLE_MASK;
 #endif
-		WREG32(DMA_IB_CNTL + sdma_offsets[i], ib_cntl);
+		WREG32(mmDMA_GFX_IB_CNTL + sdma_offsets[i], ib_cntl);
 
-		dma_cntl = RREG32(DMA_CNTL + sdma_offsets[i]);
-		dma_cntl &= ~CTXEMPTY_INT_ENABLE;
-		WREG32(DMA_CNTL + sdma_offsets[i], dma_cntl);
+		dma_cntl = RREG32(mmDMA_CNTL + sdma_offsets[i]);
+		dma_cntl &= ~DMA_CNTL__CTXEMPTY_INT_ENABLE_MASK;
+		WREG32(mmDMA_CNTL + sdma_offsets[i], dma_cntl);
 
 		ring->wptr = 0;
-		WREG32(DMA_RB_WPTR + sdma_offsets[i], lower_32_bits(ring->wptr) << 2);
-		WREG32(DMA_RB_CNTL + sdma_offsets[i], rb_cntl | DMA_RB_ENABLE);
-
-		ring->ready = true;
+		WREG32(mmDMA_GFX_RB_WPTR + sdma_offsets[i], ring->wptr << 2);
+		WREG32(mmDMA_GFX_RB_CNTL + sdma_offsets[i], rb_cntl | DMA_GFX_RB_CNTL__RB_ENABLE_MASK);
 
-		r = amdgpu_ring_test_ring(ring);
-		if (r) {
-			ring->ready = false;
+		r = amdgpu_ring_test_helper(ring);
+		if (r)
 			return r;
-		}
-
-		if (adev->mman.buffer_funcs_ring == ring)
-			amdgpu_ttm_set_active_vram_size(adev, adev->mc.real_vram_size);
 	}
 
 	return 0;
@@ -221,22 +214,17 @@ static int si_dma_ring_test_ring(struct amdgpu_ring *ring)
 	u32 tmp;
 	u64 gpu_addr;
 
-	r = amdgpu_wb_get(adev, &index);
-	if (r) {
-		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r)
 		return r;
-	}
 
 	gpu_addr = adev->wb.gpu_addr + (index * 4);
 	tmp = 0xCAFEDEAD;
 	adev->wb.wb[index] = cpu_to_le32(tmp);
 
 	r = amdgpu_ring_alloc(ring, 4);
-	if (r) {
-		DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r);
-		amdgpu_wb_free(adev, index);
-		return r;
-	}
+	if (r)
+		goto error_free_wb;
 
 	amdgpu_ring_write(ring, DMA_PACKET(DMA_PACKET_WRITE, 0, 0, 0, 1));
 	amdgpu_ring_write(ring, lower_32_bits(gpu_addr));
@@ -248,18 +236,14 @@ static int si_dma_ring_test_ring(struct amdgpu_ring *ring)
 		tmp = le32_to_cpu(adev->wb.wb[index]);
 		if (tmp == 0xDEADBEEF)
 			break;
-		DRM_UDELAY(1);
+		udelay(1);
 	}
 
-	if (i < adev->usec_timeout) {
-		DRM_INFO("ring test on %d succeeded in %d usecs\n", ring->idx, i);
-	} else {
-		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
-			  ring->idx, tmp);
-		r = -EINVAL;
-	}
-	amdgpu_wb_free(adev, index);
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
 
+error_free_wb:
+	amdgpu_device_wb_free(adev, index);
 	return r;
 }
 
@@ -267,6 +251,7 @@ static int si_dma_ring_test_ring(struct amdgpu_ring *ring)
  * si_dma_ring_test_ib - test an IB on the DMA engine
  *
  * @ring: amdgpu_ring structure holding ring information
+ * @timeout: timeout value in jiffies, or MAX_SCHEDULE_TIMEOUT
  *
  * Test a simple IB in the DMA ring (VI).
  * Returns 0 on success, error on failure.
@@ -281,21 +266,18 @@ static int si_dma_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 	u64 gpu_addr;
 	long r;
 
-	r = amdgpu_wb_get(adev, &index);
-	if (r) {
-		dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r)
 		return r;
-	}
 
 	gpu_addr = adev->wb.gpu_addr + (index * 4);
 	tmp = 0xCAFEDEAD;
 	adev->wb.wb[index] = cpu_to_le32(tmp);
 	memset(&ib, 0, sizeof(ib));
-	r = amdgpu_ib_get(adev, NULL, 256, &ib);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
+	r = amdgpu_ib_get(adev, NULL, 256,
+					AMDGPU_IB_POOL_DIRECT, &ib);
+	if (r)
 		goto err0;
-	}
 
 	ib.ptr[0] = DMA_PACKET(DMA_PACKET_WRITE, 0, 0, 0, 1);
 	ib.ptr[1] = lower_32_bits(gpu_addr);
@@ -308,32 +290,27 @@ static int si_dma_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 
 	r = dma_fence_wait_timeout(f, false, timeout);
 	if (r == 0) {
-		DRM_ERROR("amdgpu: IB test timed out\n");
 		r = -ETIMEDOUT;
 		goto err1;
 	} else if (r < 0) {
-		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
 		goto err1;
 	}
 	tmp = le32_to_cpu(adev->wb.wb[index]);
-	if (tmp == 0xDEADBEEF) {
-		DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
+	if (tmp == 0xDEADBEEF)
 		r = 0;
-	} else {
-		DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp);
+	else
 		r = -EINVAL;
-	}
 
 err1:
-	amdgpu_ib_free(adev, &ib, NULL);
+	amdgpu_ib_free(&ib, NULL);
 	dma_fence_put(f);
 err0:
-	amdgpu_wb_free(adev, index);
+	amdgpu_device_wb_free(adev, index);
 	return r;
 }
 
 /**
- * cik_dma_vm_copy_pte - update PTEs by copying them from the GART
+ * si_dma_vm_copy_pte - update PTEs by copying them from the GART
  *
  * @ib: indirect buffer to fill with commands
  * @pe: addr of the page entry
@@ -430,8 +407,9 @@ static void si_dma_vm_set_pte_pde(struct amdgpu_ib *ib,
 }
 
 /**
- * si_dma_pad_ib - pad the IB to the required number of dw
+ * si_dma_ring_pad_ib - pad the IB to the required number of dw
  *
+ * @ring: amdgpu_ring pointer
  * @ib: indirect buffer to fill with padding
  *
  */
@@ -442,7 +420,7 @@ static void si_dma_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib)
 }
 
 /**
- * cik_sdma_ring_emit_pipeline_sync - sync the pipeline
+ * si_dma_ring_emit_pipeline_sync - sync the pipeline
  *
  * @ring: amdgpu_ring pointer
  *
@@ -467,40 +445,39 @@ static void si_dma_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
  * si_dma_ring_emit_vm_flush - cik vm flush using sDMA
  *
  * @ring: amdgpu_ring pointer
- * @vm: amdgpu_vm pointer
+ * @vmid: vmid number to use
+ * @pd_addr: address
  *
  * Update the page table base and flush the VM TLB
  * using sDMA (VI).
  */
 static void si_dma_ring_emit_vm_flush(struct amdgpu_ring *ring,
-				      unsigned vm_id, uint64_t pd_addr)
+				      unsigned vmid, uint64_t pd_addr)
 {
-	amdgpu_ring_write(ring, DMA_PACKET(DMA_PACKET_SRBM_WRITE, 0, 0, 0, 0));
-	if (vm_id < 8)
-		amdgpu_ring_write(ring, (0xf << 16) | (VM_CONTEXT0_PAGE_TABLE_BASE_ADDR + vm_id));
-	else
-		amdgpu_ring_write(ring, (0xf << 16) | (VM_CONTEXT8_PAGE_TABLE_BASE_ADDR + (vm_id - 8)));
-	amdgpu_ring_write(ring, pd_addr >> 12);
-
-	/* bits 0-7 are the VM contexts0-7 */
-	amdgpu_ring_write(ring, DMA_PACKET(DMA_PACKET_SRBM_WRITE, 0, 0, 0, 0));
-	amdgpu_ring_write(ring, (0xf << 16) | (VM_INVALIDATE_REQUEST));
-	amdgpu_ring_write(ring, 1 << vm_id);
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
 
 	/* wait for invalidate to complete */
 	amdgpu_ring_write(ring, DMA_PACKET(DMA_PACKET_POLL_REG_MEM, 0, 0, 0, 0));
 	amdgpu_ring_write(ring, VM_INVALIDATE_REQUEST);
 	amdgpu_ring_write(ring, 0xff << 16); /* retry */
-	amdgpu_ring_write(ring, 1 << vm_id); /* mask */
+	amdgpu_ring_write(ring, 1 << vmid); /* mask */
 	amdgpu_ring_write(ring, 0); /* value */
 	amdgpu_ring_write(ring, (0 << 28) | 0x20); /* func(always) | poll interval */
 }
 
-static int si_dma_early_init(void *handle)
+static void si_dma_ring_emit_wreg(struct amdgpu_ring *ring,
+				  uint32_t reg, uint32_t val)
+{
+	amdgpu_ring_write(ring, DMA_PACKET(DMA_PACKET_SRBM_WRITE, 0, 0, 0, 0));
+	amdgpu_ring_write(ring, (0xf << 16) | reg);
+	amdgpu_ring_write(ring, val);
+}
+
+static int si_dma_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	adev->sdma.num_instances = 2;
+	adev->sdma.num_instances = SDMA_MAX_INSTANCE;
 
 	si_dma_set_ring_funcs(adev);
 	si_dma_set_buffer_funcs(adev);
@@ -510,19 +487,21 @@ static int si_dma_early_init(void *handle)
 	return 0;
 }
 
-static int si_dma_sw_init(void *handle)
+static int si_dma_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_ring *ring;
 	int r, i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	/* DMA0 trap event */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 224, &adev->sdma.trap_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 224,
+			      &adev->sdma.trap_irq);
 	if (r)
 		return r;
 
 	/* DMA1 trap event */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 244, &adev->sdma.trap_irq_1);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 244,
+			      &adev->sdma.trap_irq);
 	if (r)
 		return r;
 
@@ -533,9 +512,9 @@ static int si_dma_sw_init(void *handle)
 		sprintf(ring->name, "sdma%d", i);
 		r = amdgpu_ring_init(adev, ring, 1024,
 				     &adev->sdma.trap_irq,
-				     (i == 0) ?
-				     AMDGPU_SDMA_IRQ_TRAP0 :
-				     AMDGPU_SDMA_IRQ_TRAP1);
+				     (i == 0) ? AMDGPU_SDMA_IRQ_INSTANCE0 :
+				     AMDGPU_SDMA_IRQ_INSTANCE1,
+				     AMDGPU_RING_PRIO_DEFAULT, NULL);
 		if (r)
 			return r;
 	}
@@ -543,9 +522,9 @@ static int si_dma_sw_init(void *handle)
 	return r;
 }
 
-static int si_dma_sw_fini(void *handle)
+static int si_dma_sw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int i;
 
 	for (i = 0; i < adev->sdma.num_instances; i++)
@@ -554,61 +533,56 @@ static int si_dma_sw_fini(void *handle)
 	return 0;
 }
 
-static int si_dma_hw_init(void *handle)
+static int si_dma_hw_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	return si_dma_start(adev);
 }
 
-static int si_dma_hw_fini(void *handle)
+static int si_dma_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	si_dma_stop(adev);
+	si_dma_stop(ip_block->adev);
 
 	return 0;
 }
 
-static int si_dma_suspend(void *handle)
+static int si_dma_suspend(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return si_dma_hw_fini(adev);
+	return si_dma_hw_fini(ip_block);
 }
 
-static int si_dma_resume(void *handle)
+static int si_dma_resume(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return si_dma_hw_init(adev);
+	return si_dma_hw_init(ip_block);
 }
 
-static bool si_dma_is_idle(void *handle)
+static bool si_dma_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	u32 tmp = RREG32(SRBM_STATUS2);
+	struct amdgpu_device *adev = ip_block->adev;
+
+	u32 tmp = RREG32(mmSRBM_STATUS2);
 
-	if (tmp & (DMA_BUSY_MASK | DMA1_BUSY_MASK))
+	if (tmp & (SRBM_STATUS2__DMA_BUSY_MASK | SRBM_STATUS2__DMA1_BUSY_MASK))
 	    return false;
 
 	return true;
 }
 
-static int si_dma_wait_for_idle(void *handle)
+static int si_dma_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
 	unsigned i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		if (si_dma_is_idle(handle))
+		if (si_dma_is_idle(ip_block))
 			return 0;
 		udelay(1);
 	}
 	return -ETIMEDOUT;
 }
 
-static int si_dma_soft_reset(void *handle)
+static int si_dma_soft_reset(struct amdgpu_ip_block *ip_block)
 {
 	DRM_INFO("si_dma_soft_reset --- not implemented !!!!!!!\n");
 	return 0;
@@ -622,33 +596,33 @@ static int si_dma_set_trap_irq_state(struct amdgpu_device *adev,
 	u32 sdma_cntl;
 
 	switch (type) {
-	case AMDGPU_SDMA_IRQ_TRAP0:
+	case AMDGPU_SDMA_IRQ_INSTANCE0:
 		switch (state) {
 		case AMDGPU_IRQ_STATE_DISABLE:
-			sdma_cntl = RREG32(DMA_CNTL + DMA0_REGISTER_OFFSET);
-			sdma_cntl &= ~TRAP_ENABLE;
-			WREG32(DMA_CNTL + DMA0_REGISTER_OFFSET, sdma_cntl);
+			sdma_cntl = RREG32(mmDMA_CNTL + DMA0_REGISTER_OFFSET);
+			sdma_cntl &= ~DMA_CNTL__TRAP_ENABLE_MASK;
+			WREG32(mmDMA_CNTL + DMA0_REGISTER_OFFSET, sdma_cntl);
 			break;
 		case AMDGPU_IRQ_STATE_ENABLE:
-			sdma_cntl = RREG32(DMA_CNTL + DMA0_REGISTER_OFFSET);
-			sdma_cntl |= TRAP_ENABLE;
-			WREG32(DMA_CNTL + DMA0_REGISTER_OFFSET, sdma_cntl);
+			sdma_cntl = RREG32(mmDMA_CNTL + DMA0_REGISTER_OFFSET);
+			sdma_cntl |= DMA_CNTL__TRAP_ENABLE_MASK;
+			WREG32(mmDMA_CNTL + DMA0_REGISTER_OFFSET, sdma_cntl);
 			break;
 		default:
 			break;
 		}
 		break;
-	case AMDGPU_SDMA_IRQ_TRAP1:
+	case AMDGPU_SDMA_IRQ_INSTANCE1:
 		switch (state) {
 		case AMDGPU_IRQ_STATE_DISABLE:
-			sdma_cntl = RREG32(DMA_CNTL + DMA1_REGISTER_OFFSET);
-			sdma_cntl &= ~TRAP_ENABLE;
-			WREG32(DMA_CNTL + DMA1_REGISTER_OFFSET, sdma_cntl);
+			sdma_cntl = RREG32(mmDMA_CNTL + DMA1_REGISTER_OFFSET);
+			sdma_cntl &= ~DMA_CNTL__TRAP_ENABLE_MASK;
+			WREG32(mmDMA_CNTL + DMA1_REGISTER_OFFSET, sdma_cntl);
 			break;
 		case AMDGPU_IRQ_STATE_ENABLE:
-			sdma_cntl = RREG32(DMA_CNTL + DMA1_REGISTER_OFFSET);
-			sdma_cntl |= TRAP_ENABLE;
-			WREG32(DMA_CNTL + DMA1_REGISTER_OFFSET, sdma_cntl);
+			sdma_cntl = RREG32(mmDMA_CNTL + DMA1_REGISTER_OFFSET);
+			sdma_cntl |= DMA_CNTL__TRAP_ENABLE_MASK;
+			WREG32(mmDMA_CNTL + DMA1_REGISTER_OFFSET, sdma_cntl);
 			break;
 		default:
 			break;
@@ -664,38 +638,22 @@ static int si_dma_process_trap_irq(struct amdgpu_device *adev,
 				      struct amdgpu_irq_src *source,
 				      struct amdgpu_iv_entry *entry)
 {
-	amdgpu_fence_process(&adev->sdma.instance[0].ring);
-
-	return 0;
-}
-
-static int si_dma_process_trap_irq_1(struct amdgpu_device *adev,
-				      struct amdgpu_irq_src *source,
-				      struct amdgpu_iv_entry *entry)
-{
-	amdgpu_fence_process(&adev->sdma.instance[1].ring);
-
-	return 0;
-}
-
-static int si_dma_process_illegal_inst_irq(struct amdgpu_device *adev,
-					      struct amdgpu_irq_src *source,
-					      struct amdgpu_iv_entry *entry)
-{
-	DRM_ERROR("Illegal instruction in SDMA command stream\n");
-	schedule_work(&adev->reset_work);
+	if (entry->src_id == 224)
+		amdgpu_fence_process(&adev->sdma.instance[0].ring);
+	else
+		amdgpu_fence_process(&adev->sdma.instance[1].ring);
 	return 0;
 }
 
-static int si_dma_set_clockgating_state(void *handle,
+static int si_dma_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
 	u32 orig, data, offset;
 	int i;
 	bool enable;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	enable = (state == AMD_CG_STATE_GATE) ? true : false;
+	enable = (state == AMD_CG_STATE_GATE);
 
 	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_SDMA_MGCG)) {
 		for (i = 0; i < adev->sdma.num_instances; i++) {
@@ -703,11 +661,11 @@ static int si_dma_set_clockgating_state(void *handle,
 				offset = DMA0_REGISTER_OFFSET;
 			else
 				offset = DMA1_REGISTER_OFFSET;
-			orig = data = RREG32(DMA_POWER_CNTL + offset);
-			data &= ~MEM_POWER_OVERRIDE;
+			orig = data = RREG32(mmDMA_POWER_CNTL + offset);
+			data &= ~DMA_POWER_CNTL__MEM_POWER_OVERRIDE_MASK;
 			if (data != orig)
-				WREG32(DMA_POWER_CNTL + offset, data);
-			WREG32(DMA_CLK_CTRL + offset, 0x00000100);
+				WREG32(mmDMA_POWER_CNTL + offset, data);
+			WREG32(mmDMA_CLK_CTRL + offset, 0x00000100);
 		}
 	} else {
 		for (i = 0; i < adev->sdma.num_instances; i++) {
@@ -715,33 +673,33 @@ static int si_dma_set_clockgating_state(void *handle,
 				offset = DMA0_REGISTER_OFFSET;
 			else
 				offset = DMA1_REGISTER_OFFSET;
-			orig = data = RREG32(DMA_POWER_CNTL + offset);
-			data |= MEM_POWER_OVERRIDE;
+			orig = data = RREG32(mmDMA_POWER_CNTL + offset);
+			data |= DMA_POWER_CNTL__MEM_POWER_OVERRIDE_MASK;
 			if (data != orig)
-				WREG32(DMA_POWER_CNTL + offset, data);
+				WREG32(mmDMA_POWER_CNTL + offset, data);
 
-			orig = data = RREG32(DMA_CLK_CTRL + offset);
+			orig = data = RREG32(mmDMA_CLK_CTRL + offset);
 			data = 0xff000000;
 			if (data != orig)
-				WREG32(DMA_CLK_CTRL + offset, data);
+				WREG32(mmDMA_CLK_CTRL + offset, data);
 		}
 	}
 
 	return 0;
 }
 
-static int si_dma_set_powergating_state(void *handle,
+static int si_dma_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	u32 tmp;
 
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	WREG32(DMA_PGFSM_WRITE,  0x00002000);
-	WREG32(DMA_PGFSM_CONFIG, 0x100010ff);
+	WREG32(mmDMA_PGFSM_WRITE,  0x00002000);
+	WREG32(mmDMA_PGFSM_CONFIG, 0x100010ff);
 
 	for (tmp = 0; tmp < 5; tmp++)
-		WREG32(DMA_PGFSM_WRITE, 0);
+		WREG32(mmDMA_PGFSM_WRITE, 0);
 
 	return 0;
 }
@@ -749,7 +707,6 @@ static int si_dma_set_powergating_state(void *handle,
 static const struct amd_ip_funcs si_dma_ip_funcs = {
 	.name = "si_dma",
 	.early_init = si_dma_early_init,
-	.late_init = NULL,
 	.sw_init = si_dma_sw_init,
 	.sw_fini = si_dma_sw_fini,
 	.hw_init = si_dma_hw_init,
@@ -772,22 +729,20 @@ static const struct amdgpu_ring_funcs si_dma_ring_funcs = {
 	.get_wptr = si_dma_ring_get_wptr,
 	.set_wptr = si_dma_ring_set_wptr,
 	.emit_frame_size =
-		3 + /* si_dma_ring_emit_hdp_flush */
-		3 + /* si_dma_ring_emit_hdp_invalidate */
+		3 + 3 + /* hdp flush / invalidate */
 		6 + /* si_dma_ring_emit_pipeline_sync */
-		12 + /* si_dma_ring_emit_vm_flush */
+		SI_FLUSH_GPU_TLB_NUM_WREG * 3 + 6 + /* si_dma_ring_emit_vm_flush */
 		9 + 9 + 9, /* si_dma_ring_emit_fence x3 for user fence, vm fence */
 	.emit_ib_size = 7 + 3, /* si_dma_ring_emit_ib */
 	.emit_ib = si_dma_ring_emit_ib,
 	.emit_fence = si_dma_ring_emit_fence,
 	.emit_pipeline_sync = si_dma_ring_emit_pipeline_sync,
 	.emit_vm_flush = si_dma_ring_emit_vm_flush,
-	.emit_hdp_flush = si_dma_ring_emit_hdp_flush,
-	.emit_hdp_invalidate = si_dma_ring_emit_hdp_invalidate,
 	.test_ring = si_dma_ring_test_ring,
 	.test_ib = si_dma_ring_test_ib,
 	.insert_nop = amdgpu_ring_insert_nop,
 	.pad_ib = si_dma_ring_pad_ib,
+	.emit_wreg = si_dma_ring_emit_wreg,
 };
 
 static void si_dma_set_ring_funcs(struct amdgpu_device *adev)
@@ -803,30 +758,20 @@ static const struct amdgpu_irq_src_funcs si_dma_trap_irq_funcs = {
 	.process = si_dma_process_trap_irq,
 };
 
-static const struct amdgpu_irq_src_funcs si_dma_trap_irq_funcs_1 = {
-	.set = si_dma_set_trap_irq_state,
-	.process = si_dma_process_trap_irq_1,
-};
-
-static const struct amdgpu_irq_src_funcs si_dma_illegal_inst_irq_funcs = {
-	.process = si_dma_process_illegal_inst_irq,
-};
-
 static void si_dma_set_irq_funcs(struct amdgpu_device *adev)
 {
 	adev->sdma.trap_irq.num_types = AMDGPU_SDMA_IRQ_LAST;
 	adev->sdma.trap_irq.funcs = &si_dma_trap_irq_funcs;
-	adev->sdma.trap_irq_1.funcs = &si_dma_trap_irq_funcs_1;
-	adev->sdma.illegal_inst_irq.funcs = &si_dma_illegal_inst_irq_funcs;
 }
 
 /**
  * si_dma_emit_copy_buffer - copy buffer using the sDMA engine
  *
- * @ring: amdgpu_ring structure holding ring information
+ * @ib: indirect buffer to copy to
  * @src_offset: src GPU address
  * @dst_offset: dst GPU address
  * @byte_count: number of bytes to xfer
+ * @copy_flags: unused
  *
  * Copy GPU buffers using the DMA engine (VI).
  * Used by the amdgpu ttm implementation to move pages if
@@ -835,7 +780,8 @@ static void si_dma_set_irq_funcs(struct amdgpu_device *adev)
 static void si_dma_emit_copy_buffer(struct amdgpu_ib *ib,
 				       uint64_t src_offset,
 				       uint64_t dst_offset,
-				       uint32_t byte_count)
+				       uint32_t byte_count,
+				       uint32_t copy_flags)
 {
 	ib->ptr[ib->length_dw++] = DMA_PACKET(DMA_PACKET_COPY,
 					      1, 0, 0, byte_count);
@@ -848,7 +794,7 @@ static void si_dma_emit_copy_buffer(struct amdgpu_ib *ib,
 /**
  * si_dma_emit_fill_buffer - fill buffer using the sDMA engine
  *
- * @ring: amdgpu_ring structure holding ring information
+ * @ib: indirect buffer to copy to
  * @src_data: value to write to buffer
  * @dst_offset: dst GPU address
  * @byte_count: number of bytes to xfer
@@ -880,14 +826,14 @@ static const struct amdgpu_buffer_funcs si_dma_buffer_funcs = {
 
 static void si_dma_set_buffer_funcs(struct amdgpu_device *adev)
 {
-	if (adev->mman.buffer_funcs == NULL) {
-		adev->mman.buffer_funcs = &si_dma_buffer_funcs;
-		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
-	}
+	adev->mman.buffer_funcs = &si_dma_buffer_funcs;
+	adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
 }
 
 static const struct amdgpu_vm_pte_funcs si_dma_vm_pte_funcs = {
+	.copy_pte_num_dw = 5,
 	.copy_pte = si_dma_vm_copy_pte,
+
 	.write_pte = si_dma_vm_write_pte,
 	.set_pte_pde = si_dma_vm_set_pte_pde,
 };
@@ -896,14 +842,12 @@ static void si_dma_set_vm_pte_funcs(struct amdgpu_device *adev)
 {
 	unsigned i;
 
-	if (adev->vm_manager.vm_pte_funcs == NULL) {
-		adev->vm_manager.vm_pte_funcs = &si_dma_vm_pte_funcs;
-		for (i = 0; i < adev->sdma.num_instances; i++)
-			adev->vm_manager.vm_pte_rings[i] =
-				&adev->sdma.instance[i].ring;
-
-		adev->vm_manager.vm_pte_num_rings = adev->sdma.num_instances;
+	adev->vm_manager.vm_pte_funcs = &si_dma_vm_pte_funcs;
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		adev->vm_manager.vm_pte_scheds[i] =
+			&adev->sdma.instance[i].ring.sched;
 	}
+	adev->vm_manager.vm_pte_num_scheds = adev->sdma.num_instances;
 }
 
 const struct amdgpu_ip_block_version si_dma_ip_block =
diff --git a/drivers/gpu/drm/amd/amdgpu/si_enums.h b/drivers/gpu/drm/amd/amdgpu/si_enums.h
index dc9e0e6b4558..6da65778292b 100644
--- a/drivers/gpu/drm/amd/amdgpu/si_enums.h
+++ b/drivers/gpu/drm/amd/amdgpu/si_enums.h
@@ -23,104 +23,15 @@
 #ifndef SI_ENUMS_H
 #define SI_ENUMS_H
 
-#define VBLANK_INT_MASK                (1 << 0)
-#define DC_HPDx_INT_EN                 (1 << 16)
-#define VBLANK_ACK                     (1 << 4)
-#define VLINE_ACK                      (1 << 4)
-
-#define CURSOR_WIDTH 64
-#define CURSOR_HEIGHT 64
-
-#define VGA_VSTATUS_CNTL               0xFFFCFFFF
 #define PRIORITY_MARK_MASK             0x7fff
 #define PRIORITY_OFF                   (1 << 16)
 #define PRIORITY_ALWAYS_ON             (1 << 20)
-#define INTERLEAVE_EN                  (1 << 0)
-
-#define LATENCY_WATERMARK_MASK(x)      ((x) << 16)
-#define DC_LB_MEMORY_CONFIG(x)         ((x) << 20)
-#define ICON_DEGAMMA_MODE(x)           (((x) & 0x3) << 8)
-
-#define GRPH_ENDIAN_SWAP(x)            (((x) & 0x3) << 0)
-#define GRPH_ENDIAN_NONE               0
-#define GRPH_ENDIAN_8IN16              1
-#define GRPH_ENDIAN_8IN32              2
-#define GRPH_ENDIAN_8IN64              3
-
-#define GRPH_DEPTH(x)                  (((x) & 0x3) << 0)
-#define GRPH_DEPTH_8BPP                0
-#define GRPH_DEPTH_16BPP               1
-#define GRPH_DEPTH_32BPP               2
-
-#define GRPH_FORMAT(x)                 (((x) & 0x7) << 8)
-#define GRPH_FORMAT_INDEXED            0
-#define GRPH_FORMAT_ARGB1555           0
-#define GRPH_FORMAT_ARGB565            1
-#define GRPH_FORMAT_ARGB4444           2
-#define GRPH_FORMAT_AI88               3
-#define GRPH_FORMAT_MONO16             4
-#define GRPH_FORMAT_BGRA5551           5
-#define GRPH_FORMAT_ARGB8888           0
-#define GRPH_FORMAT_ARGB2101010        1
-#define GRPH_FORMAT_32BPP_DIG          2
-#define GRPH_FORMAT_8B_ARGB2101010     3
-#define GRPH_FORMAT_BGRA1010102        4
-#define GRPH_FORMAT_8B_BGRA1010102     5
-#define GRPH_FORMAT_RGB111110          6
-#define GRPH_FORMAT_BGR101111          7
-
-#define GRPH_NUM_BANKS(x)              (((x) & 0x3) << 2)
-#define GRPH_ARRAY_MODE(x)             (((x) & 0x7) << 20)
-#define GRPH_ARRAY_LINEAR_GENERAL      0
-#define GRPH_ARRAY_LINEAR_ALIGNED      1
-#define GRPH_ARRAY_1D_TILED_THIN1      2
-#define GRPH_ARRAY_2D_TILED_THIN1      4
-#define GRPH_TILE_SPLIT(x)             (((x) & 0x7) << 13)
-#define GRPH_BANK_WIDTH(x)             (((x) & 0x3) << 6)
-#define GRPH_BANK_HEIGHT(x)            (((x) & 0x3) << 11)
-#define GRPH_MACRO_TILE_ASPECT(x)      (((x) & 0x3) << 18)
-#define GRPH_ARRAY_MODE(x)             (((x) & 0x7) << 20)
-#define GRPH_PIPE_CONFIG(x)                   (((x) & 0x1f) << 24)
-
-#define CURSOR_EN                      (1 << 0)
-#define CURSOR_MODE(x)                 (((x) & 0x3) << 8)
-#define CURSOR_MONO                    0
-#define CURSOR_24_1                    1
-#define CURSOR_24_8_PRE_MULT           2
-#define CURSOR_24_8_UNPRE_MULT         3
-#define CURSOR_2X_MAGNIFY              (1 << 16)
-#define CURSOR_FORCE_MC_ON             (1 << 20)
-#define CURSOR_URGENT_CONTROL(x)       (((x) & 0x7) << 24)
-#define CURSOR_URGENT_ALWAYS           0
-#define CURSOR_URGENT_1_8              1
-#define CURSOR_URGENT_1_4              2
-#define CURSOR_URGENT_3_8              3
-#define CURSOR_URGENT_1_2              4
-#define CURSOR_UPDATE_PENDING          (1 << 0)
-#define CURSOR_UPDATE_TAKEN            (1 << 1)
-#define CURSOR_UPDATE_LOCK             (1 << 16)
-#define CURSOR_DISABLE_MULTIPLE_UPDATE (1 << 24)
-
-#define AMDGPU_NUM_OF_VMIDS                     8
-#define SI_CRTC0_REGISTER_OFFSET                0
-#define SI_CRTC1_REGISTER_OFFSET                0x300
-#define SI_CRTC2_REGISTER_OFFSET                0x2600
-#define SI_CRTC3_REGISTER_OFFSET                0x2900
-#define SI_CRTC4_REGISTER_OFFSET                0x2c00
-#define SI_CRTC5_REGISTER_OFFSET                0x2f00
-
-#define DMA0_REGISTER_OFFSET 0x000
-#define DMA1_REGISTER_OFFSET 0x200
-#define ES_AND_GS_AUTO       3
-#define RADEON_PACKET_TYPE3  3
-#define CE_PARTITION_BASE    3
-#define BUF_SWAP_32BIT       (2 << 16)
 
 #define GFX_POWER_STATUS                           (1 << 1)
 #define GFX_CLOCK_STATUS                           (1 << 2)
 #define GFX_LS_STATUS                              (1 << 3)
-#define RLC_BUSY_STATUS                            (1 << 0)
 
+#define RLC_BUSY_STATUS                            (1 << 0)
 #define RLC_PUD(x)                               ((x) << 0)
 #define RLC_PUD_MASK                             (0xff << 0)
 #define RLC_PDD(x)                               ((x) << 8)
@@ -129,144 +40,8 @@
 #define RLC_TTPD_MASK                            (0xff << 16)
 #define RLC_MSD(x)                               ((x) << 24)
 #define RLC_MSD_MASK                             (0xff << 24)
-#define WRITE_DATA_ENGINE_SEL(x) ((x) << 30)
-#define WRITE_DATA_DST_SEL(x) ((x) << 8)
-#define EVENT_TYPE(x) ((x) << 0)
-#define EVENT_INDEX(x) ((x) << 8)
-#define WAIT_REG_MEM_MEM_SPACE(x)               ((x) << 4)
-#define WAIT_REG_MEM_FUNCTION(x)                ((x) << 0)
-#define WAIT_REG_MEM_ENGINE(x)                  ((x) << 8)
 
-#define GFX6_NUM_GFX_RINGS     1
-#define GFX6_NUM_COMPUTE_RINGS 2
 #define RLC_SAVE_AND_RESTORE_STARTING_OFFSET 0x90
 #define RLC_CLEAR_STATE_DESCRIPTOR_OFFSET    0x3D
 
-#define TAHITI_GB_ADDR_CONFIG_GOLDEN        0x12011003
-#define VERDE_GB_ADDR_CONFIG_GOLDEN         0x02010002
-#define HAINAN_GB_ADDR_CONFIG_GOLDEN        0x02011003
-
-#define PACKET3(op, n)  ((RADEON_PACKET_TYPE3 << 30) |                  \
-                         (((op) & 0xFF) << 8) |                         \
-                         ((n) & 0x3FFF) << 16)
-#define PACKET3_COMPUTE(op, n) (PACKET3(op, n) | 1 << 1)
-#define	PACKET3_NOP					0x10
-#define	PACKET3_SET_BASE				0x11
-#define		PACKET3_BASE_INDEX(x)                  ((x) << 0)
-#define	PACKET3_CLEAR_STATE				0x12
-#define	PACKET3_INDEX_BUFFER_SIZE			0x13
-#define	PACKET3_DISPATCH_DIRECT				0x15
-#define	PACKET3_DISPATCH_INDIRECT			0x16
-#define	PACKET3_ALLOC_GDS				0x1B
-#define	PACKET3_WRITE_GDS_RAM				0x1C
-#define	PACKET3_ATOMIC_GDS				0x1D
-#define	PACKET3_ATOMIC					0x1E
-#define	PACKET3_OCCLUSION_QUERY				0x1F
-#define	PACKET3_SET_PREDICATION				0x20
-#define	PACKET3_REG_RMW					0x21
-#define	PACKET3_COND_EXEC				0x22
-#define	PACKET3_PRED_EXEC				0x23
-#define	PACKET3_DRAW_INDIRECT				0x24
-#define	PACKET3_DRAW_INDEX_INDIRECT			0x25
-#define	PACKET3_INDEX_BASE				0x26
-#define	PACKET3_DRAW_INDEX_2				0x27
-#define	PACKET3_CONTEXT_CONTROL				0x28
-#define	PACKET3_INDEX_TYPE				0x2A
-#define	PACKET3_DRAW_INDIRECT_MULTI			0x2C
-#define	PACKET3_DRAW_INDEX_AUTO				0x2D
-#define	PACKET3_DRAW_INDEX_IMMD				0x2E
-#define	PACKET3_NUM_INSTANCES				0x2F
-#define	PACKET3_DRAW_INDEX_MULTI_AUTO			0x30
-#define	PACKET3_INDIRECT_BUFFER_CONST			0x31
-#define	PACKET3_INDIRECT_BUFFER				0x3F
-#define	PACKET3_STRMOUT_BUFFER_UPDATE			0x34
-#define	PACKET3_DRAW_INDEX_OFFSET_2			0x35
-#define	PACKET3_DRAW_INDEX_MULTI_ELEMENT		0x36
-#define	PACKET3_WRITE_DATA				0x37
-#define	PACKET3_DRAW_INDEX_INDIRECT_MULTI		0x38
-#define	PACKET3_MEM_SEMAPHORE				0x39
-#define	PACKET3_MPEG_INDEX				0x3A
-#define	PACKET3_COPY_DW					0x3B
-#define	PACKET3_WAIT_REG_MEM				0x3C
-#define	PACKET3_MEM_WRITE				0x3D
-#define	PACKET3_COPY_DATA				0x40
-#define	PACKET3_CP_DMA					0x41
-#              define PACKET3_CP_DMA_DST_SEL(x)    ((x) << 20)
-#              define PACKET3_CP_DMA_ENGINE(x)     ((x) << 27)
-#              define PACKET3_CP_DMA_SRC_SEL(x)    ((x) << 29)
-#              define PACKET3_CP_DMA_CP_SYNC       (1 << 31)
-#              define PACKET3_CP_DMA_DIS_WC        (1 << 21)
-#              define PACKET3_CP_DMA_CMD_SRC_SWAP(x) ((x) << 22)
-#              define PACKET3_CP_DMA_CMD_DST_SWAP(x) ((x) << 24)
-#              define PACKET3_CP_DMA_CMD_SAS       (1 << 26)
-#              define PACKET3_CP_DMA_CMD_DAS       (1 << 27)
-#              define PACKET3_CP_DMA_CMD_SAIC      (1 << 28)
-#              define PACKET3_CP_DMA_CMD_DAIC      (1 << 29)
-#              define PACKET3_CP_DMA_CMD_RAW_WAIT  (1 << 30)
-#define	PACKET3_PFP_SYNC_ME				0x42
-#define	PACKET3_SURFACE_SYNC				0x43
-#              define PACKET3_DEST_BASE_0_ENA      (1 << 0)
-#              define PACKET3_DEST_BASE_1_ENA      (1 << 1)
-#              define PACKET3_CB0_DEST_BASE_ENA    (1 << 6)
-#              define PACKET3_CB1_DEST_BASE_ENA    (1 << 7)
-#              define PACKET3_CB2_DEST_BASE_ENA    (1 << 8)
-#              define PACKET3_CB3_DEST_BASE_ENA    (1 << 9)
-#              define PACKET3_CB4_DEST_BASE_ENA    (1 << 10)
-#              define PACKET3_CB5_DEST_BASE_ENA    (1 << 11)
-#              define PACKET3_CB6_DEST_BASE_ENA    (1 << 12)
-#              define PACKET3_CB7_DEST_BASE_ENA    (1 << 13)
-#              define PACKET3_DB_DEST_BASE_ENA     (1 << 14)
-#              define PACKET3_DEST_BASE_2_ENA      (1 << 19)
-#              define PACKET3_DEST_BASE_3_ENA      (1 << 21)
-#              define PACKET3_TCL1_ACTION_ENA      (1 << 22)
-#              define PACKET3_TC_ACTION_ENA        (1 << 23)
-#              define PACKET3_CB_ACTION_ENA        (1 << 25)
-#              define PACKET3_DB_ACTION_ENA        (1 << 26)
-#              define PACKET3_SH_KCACHE_ACTION_ENA (1 << 27)
-#              define PACKET3_SH_ICACHE_ACTION_ENA (1 << 29)
-#define	PACKET3_ME_INITIALIZE				0x44
-#define		PACKET3_ME_INITIALIZE_DEVICE_ID(x) ((x) << 16)
-#define	PACKET3_COND_WRITE				0x45
-#define	PACKET3_EVENT_WRITE				0x46
-#define	PACKET3_EVENT_WRITE_EOP				0x47
-#define	PACKET3_EVENT_WRITE_EOS				0x48
-#define	PACKET3_PREAMBLE_CNTL				0x4A
-#              define PACKET3_PREAMBLE_BEGIN_CLEAR_STATE     (2 << 28)
-#              define PACKET3_PREAMBLE_END_CLEAR_STATE       (3 << 28)
-#define	PACKET3_ONE_REG_WRITE				0x57
-#define	PACKET3_LOAD_CONFIG_REG				0x5F
-#define	PACKET3_LOAD_CONTEXT_REG			0x60
-#define	PACKET3_LOAD_SH_REG				0x61
-#define	PACKET3_SET_CONFIG_REG				0x68
-#define		PACKET3_SET_CONFIG_REG_START			0x00002000
-#define		PACKET3_SET_CONFIG_REG_END			0x00002c00
-#define	PACKET3_SET_CONTEXT_REG				0x69
-#define		PACKET3_SET_CONTEXT_REG_START			0x000a000
-#define		PACKET3_SET_CONTEXT_REG_END			0x000a400
-#define	PACKET3_SET_CONTEXT_REG_INDIRECT		0x73
-#define	PACKET3_SET_RESOURCE_INDIRECT			0x74
-#define	PACKET3_SET_SH_REG				0x76
-#define		PACKET3_SET_SH_REG_START			0x00002c00
-#define		PACKET3_SET_SH_REG_END				0x00003000
-#define	PACKET3_SET_SH_REG_OFFSET			0x77
-#define	PACKET3_ME_WRITE				0x7A
-#define	PACKET3_SCRATCH_RAM_WRITE			0x7D
-#define	PACKET3_SCRATCH_RAM_READ			0x7E
-#define	PACKET3_CE_WRITE				0x7F
-#define	PACKET3_LOAD_CONST_RAM				0x80
-#define	PACKET3_WRITE_CONST_RAM				0x81
-#define	PACKET3_WRITE_CONST_RAM_OFFSET			0x82
-#define	PACKET3_DUMP_CONST_RAM				0x83
-#define	PACKET3_INCREMENT_CE_COUNTER			0x84
-#define	PACKET3_INCREMENT_DE_COUNTER			0x85
-#define	PACKET3_WAIT_ON_CE_COUNTER			0x86
-#define	PACKET3_WAIT_ON_DE_COUNTER			0x87
-#define	PACKET3_WAIT_ON_DE_COUNTER_DIFF			0x88
-#define	PACKET3_SET_CE_DE_COUNTERS			0x89
-#define	PACKET3_WAIT_ON_AVAIL_BUFFER			0x8A
-#define	PACKET3_SWITCH_BUFFER				0x8B
-#define PACKET3_SEM_WAIT_ON_SIGNAL    (0x1 << 12)
-#define PACKET3_SEM_SEL_SIGNAL	    (0x6 << 29)
-#define PACKET3_SEM_SEL_WAIT	    (0x7 << 29)
-
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/si_ih.c b/drivers/gpu/drm/amd/amdgpu/si_ih.c
index ce25e03a077d..1df00f8a2406 100644
--- a/drivers/gpu/drm/amd/amdgpu/si_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/si_ih.c
@@ -20,12 +20,17 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  *
  */
-#include <drm/drmP.h>
+
+#include <linux/pci.h>
+
 #include "amdgpu.h"
 #include "amdgpu_ih.h"
 #include "sid.h"
 #include "si_ih.h"
 
+#include "oss/oss_1_0_d.h"
+#include "oss/oss_1_0_sh_mask.h"
+
 static void si_ih_set_interrupt_funcs(struct amdgpu_device *adev);
 
 static void si_ih_enable_interrupts(struct amdgpu_device *adev)
@@ -39,7 +44,7 @@ static void si_ih_enable_interrupts(struct amdgpu_device *adev)
 	WREG32(IH_RB_CNTL, ih_rb_cntl);
 	adev->irq.ih.enabled = true;
 }
-  
+
 static void si_ih_disable_interrupts(struct amdgpu_device *adev)
 {
 	u32 ih_rb_cntl = RREG32(IH_RB_CNTL);
@@ -57,12 +62,13 @@ static void si_ih_disable_interrupts(struct amdgpu_device *adev)
 
 static int si_ih_irq_init(struct amdgpu_device *adev)
 {
+	struct amdgpu_ih_ring *ih = &adev->irq.ih;
 	int rb_bufsz;
 	u32 interrupt_cntl, ih_cntl, ih_rb_cntl;
-	u64 wptr_off;
 
 	si_ih_disable_interrupts(adev);
-	WREG32(INTERRUPT_CNTL2, adev->irq.ih.gpu_addr >> 8);
+	/* set dummy read address to dummy page address */
+	WREG32(INTERRUPT_CNTL2, adev->dummy_page_addr >> 8);
 	interrupt_cntl = RREG32(INTERRUPT_CNTL);
 	interrupt_cntl &= ~IH_DUMMY_RD_OVERRIDE;
 	interrupt_cntl &= ~IH_REQ_NONSNOOP_EN;
@@ -76,9 +82,8 @@ static int si_ih_irq_init(struct amdgpu_device *adev)
 		     (rb_bufsz << 1) |
 		     IH_WPTR_WRITEBACK_ENABLE;
 
-	wptr_off = adev->wb.gpu_addr + (adev->irq.ih.wptr_offs * 4);
-	WREG32(IH_RB_WPTR_ADDR_LO, lower_32_bits(wptr_off));
-	WREG32(IH_RB_WPTR_ADDR_HI, upper_32_bits(wptr_off) & 0xFF);
+	WREG32(IH_RB_WPTR_ADDR_LO, lower_32_bits(ih->wptr_addr));
+	WREG32(IH_RB_WPTR_ADDR_HI, upper_32_bits(ih->wptr_addr) & 0xFF);
 	WREG32(IH_RB_CNTL, ih_rb_cntl);
 	WREG32(IH_RB_RPTR, 0);
 	WREG32(IH_RB_WPTR, 0);
@@ -100,114 +105,116 @@ static void si_ih_irq_disable(struct amdgpu_device *adev)
 	mdelay(1);
 }
 
-static u32 si_ih_get_wptr(struct amdgpu_device *adev)
+static u32 si_ih_get_wptr(struct amdgpu_device *adev,
+			  struct amdgpu_ih_ring *ih)
 {
 	u32 wptr, tmp;
 
-	wptr = le32_to_cpu(adev->wb.wb[adev->irq.ih.wptr_offs]);
+	wptr = le32_to_cpu(*ih->wptr_cpu);
 
 	if (wptr & IH_RB_WPTR__RB_OVERFLOW_MASK) {
 		wptr &= ~IH_RB_WPTR__RB_OVERFLOW_MASK;
 		dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
-			wptr, adev->irq.ih.rptr, (wptr + 16) & adev->irq.ih.ptr_mask);
-		adev->irq.ih.rptr = (wptr + 16) & adev->irq.ih.ptr_mask;
+			wptr, ih->rptr, (wptr + 16) & ih->ptr_mask);
+		ih->rptr = (wptr + 16) & ih->ptr_mask;
 		tmp = RREG32(IH_RB_CNTL);
 		tmp |= IH_RB_CNTL__WPTR_OVERFLOW_CLEAR_MASK;
 		WREG32(IH_RB_CNTL, tmp);
+
+		/* Unset the CLEAR_OVERFLOW bit immediately so new overflows
+		 * can be detected.
+		 */
+		tmp &= ~IH_RB_CNTL__WPTR_OVERFLOW_CLEAR_MASK;
+		WREG32(IH_RB_CNTL, tmp);
 	}
-	return (wptr & adev->irq.ih.ptr_mask);
+	return (wptr & ih->ptr_mask);
 }
 
 static void si_ih_decode_iv(struct amdgpu_device *adev,
-			     struct amdgpu_iv_entry *entry)
+			    struct amdgpu_ih_ring *ih,
+			    struct amdgpu_iv_entry *entry)
 {
-	u32 ring_index = adev->irq.ih.rptr >> 2;
+	u32 ring_index = ih->rptr >> 2;
 	uint32_t dw[4];
 
-	dw[0] = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
-	dw[1] = le32_to_cpu(adev->irq.ih.ring[ring_index + 1]);
-	dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
-	dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
+	dw[0] = le32_to_cpu(ih->ring[ring_index + 0]);
+	dw[1] = le32_to_cpu(ih->ring[ring_index + 1]);
+	dw[2] = le32_to_cpu(ih->ring[ring_index + 2]);
+	dw[3] = le32_to_cpu(ih->ring[ring_index + 3]);
 
-	entry->client_id = AMDGPU_IH_CLIENTID_LEGACY;
+	entry->client_id = AMDGPU_IRQ_CLIENTID_LEGACY;
 	entry->src_id = dw[0] & 0xff;
 	entry->src_data[0] = dw[1] & 0xfffffff;
 	entry->ring_id = dw[2] & 0xff;
-	entry->vm_id = (dw[2] >> 8) & 0xff;
+	entry->vmid = (dw[2] >> 8) & 0xff;
 
-	adev->irq.ih.rptr += 16;
+	ih->rptr += 16;
 }
 
-static void si_ih_set_rptr(struct amdgpu_device *adev)
+static void si_ih_set_rptr(struct amdgpu_device *adev,
+			   struct amdgpu_ih_ring *ih)
 {
-	WREG32(IH_RB_RPTR, adev->irq.ih.rptr);
+	WREG32(IH_RB_RPTR, ih->rptr);
 }
 
-static int si_ih_early_init(void *handle)
+static int si_ih_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	si_ih_set_interrupt_funcs(adev);
 
 	return 0;
 }
 
-static int si_ih_sw_init(void *handle)
+static int si_ih_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	r = amdgpu_ih_ring_init(adev, 64 * 1024, false);
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih, 64 * 1024, false);
 	if (r)
 		return r;
 
 	return amdgpu_irq_init(adev);
 }
 
-static int si_ih_sw_fini(void *handle)
+static int si_ih_sw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	amdgpu_irq_fini(adev);
-	amdgpu_ih_ring_fini(adev);
+	amdgpu_irq_fini_sw(adev);
 
 	return 0;
 }
 
-static int si_ih_hw_init(void *handle)
+static int si_ih_hw_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	return si_ih_irq_init(adev);
 }
 
-static int si_ih_hw_fini(void *handle)
+static int si_ih_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	si_ih_irq_disable(adev);
+	si_ih_irq_disable(ip_block->adev);
 
 	return 0;
 }
 
-static int si_ih_suspend(void *handle)
+static int si_ih_suspend(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return si_ih_hw_fini(adev);
+	return si_ih_hw_fini(ip_block);
 }
 
-static int si_ih_resume(void *handle)
+static int si_ih_resume(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return si_ih_hw_init(adev);
+	return si_ih_hw_init(ip_block);
 }
 
-static bool si_ih_is_idle(void *handle)
+static bool si_ih_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	u32 tmp = RREG32(SRBM_STATUS);
+	struct amdgpu_device *adev = ip_block->adev;
+	u32 tmp = RREG32(mmSRBM_STATUS);
 
 	if (tmp & SRBM_STATUS__IH_BUSY_MASK)
 		return false;
@@ -215,41 +222,41 @@ static bool si_ih_is_idle(void *handle)
 	return true;
 }
 
-static int si_ih_wait_for_idle(void *handle)
+static int si_ih_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
 	unsigned i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		if (si_ih_is_idle(handle))
+		if (si_ih_is_idle(ip_block))
 			return 0;
 		udelay(1);
 	}
 	return -ETIMEDOUT;
 }
 
-static int si_ih_soft_reset(void *handle)
+static int si_ih_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	u32 srbm_soft_reset = 0;
-	u32 tmp = RREG32(SRBM_STATUS);
+	u32 tmp = RREG32(mmSRBM_STATUS);
 
 	if (tmp & SRBM_STATUS__IH_BUSY_MASK)
 		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_IH_MASK;
 
 	if (srbm_soft_reset) {
-		tmp = RREG32(SRBM_SOFT_RESET);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
 		tmp |= srbm_soft_reset;
-		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
-		WREG32(SRBM_SOFT_RESET, tmp);
-		tmp = RREG32(SRBM_SOFT_RESET);
+		dev_info(adev->dev, "mmSRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
 
 		udelay(50);
 
 		tmp &= ~srbm_soft_reset;
-		WREG32(SRBM_SOFT_RESET, tmp);
-		tmp = RREG32(SRBM_SOFT_RESET);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
 
 		udelay(50);
 	}
@@ -257,13 +264,13 @@ static int si_ih_soft_reset(void *handle)
 	return 0;
 }
 
-static int si_ih_set_clockgating_state(void *handle,
+static int si_ih_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
 	return 0;
 }
 
-static int si_ih_set_powergating_state(void *handle,
+static int si_ih_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	return 0;
@@ -272,7 +279,6 @@ static int si_ih_set_powergating_state(void *handle,
 static const struct amd_ip_funcs si_ih_ip_funcs = {
 	.name = "si_ih",
 	.early_init = si_ih_early_init,
-	.late_init = NULL,
 	.sw_init = si_ih_sw_init,
 	.sw_fini = si_ih_sw_fini,
 	.hw_init = si_ih_hw_init,
@@ -294,8 +300,7 @@ static const struct amdgpu_ih_funcs si_ih_funcs = {
 
 static void si_ih_set_interrupt_funcs(struct amdgpu_device *adev)
 {
-	if (adev->irq.ih_funcs == NULL)
-		adev->irq.ih_funcs = &si_ih_funcs;
+	adev->irq.ih_funcs = &si_ih_funcs;
 }
 
 const struct amdgpu_ip_block_version si_ih_ip_block =
diff --git a/drivers/gpu/drm/amd/amdgpu/sid.h b/drivers/gpu/drm/amd/amdgpu/sid.h
index c57eff159374..cbd4f8951cfa 100644
--- a/drivers/gpu/drm/amd/amdgpu/sid.h
+++ b/drivers/gpu/drm/amd/amdgpu/sid.h
@@ -24,48 +24,12 @@
 #ifndef SI_H
 #define SI_H
 
-#define TAHITI_RB_BITMAP_WIDTH_PER_SH  2
-
-#define TAHITI_GB_ADDR_CONFIG_GOLDEN        0x12011003
-#define VERDE_GB_ADDR_CONFIG_GOLDEN         0x12010002
-#define HAINAN_GB_ADDR_CONFIG_GOLDEN        0x02010001
-
-#define SI_MAX_SH_GPRS		 	256
-#define SI_MAX_TEMP_GPRS         	16
-#define SI_MAX_SH_THREADS        	256
-#define SI_MAX_SH_STACK_ENTRIES  	4096
-#define SI_MAX_FRC_EOV_CNT       	16384
-#define SI_MAX_BACKENDS          	8
-#define SI_MAX_BACKENDS_MASK     	0xFF
-#define SI_MAX_BACKENDS_PER_SE_MASK     0x0F
-#define SI_MAX_SIMDS             	12
-#define SI_MAX_SIMDS_MASK        	0x0FFF
-#define SI_MAX_SIMDS_PER_SE_MASK        0x00FF
-#define SI_MAX_PIPES            	8
-#define SI_MAX_PIPES_MASK        	0xFF
-#define SI_MAX_PIPES_PER_SIMD_MASK      0x3F
-#define SI_MAX_LDS_NUM           	0xFFFF
-#define SI_MAX_TCC               	16
-#define SI_MAX_TCC_MASK          	0xFFFF
-
-#define AMDGPU_NUM_OF_VMIDS 		8
-
-/* SMC IND accessor regs */
-#define SMC_IND_INDEX_0                              0x80
-#define SMC_IND_DATA_0                               0x81
-
-#define SMC_IND_ACCESS_CNTL                          0x8A
-#       define AUTO_INCREMENT_IND_0                  (1 << 0)
-#define SMC_MESSAGE_0                                0x8B
-#define SMC_RESP_0                                   0x8C
+#define SI_MAX_CTLACKS_ASSERTION_WAIT   100
 
 /* CG IND registers are accessed via SMC indirect space + SMC_CG_IND_START */
 #define SMC_CG_IND_START                    0xc0030000
 #define SMC_CG_IND_END                      0xc0040000
 
-#define	CG_CGTT_LOCAL_0				0x400
-#define	CG_CGTT_LOCAL_1				0x401
-
 /* SMC IND registers */
 #define	SMC_SYSCON_RESET_CNTL				0x80000000
 #       define RST_REG                                  (1 << 0)
@@ -73,9 +37,6 @@
 #       define CK_DISABLE                               (1 << 0)
 #       define CKEN                                     (1 << 24)
 
-#define VGA_HDP_CONTROL  				0xCA
-#define		VGA_MEMORY_DISABLE				(1 << 4)
-
 #define DCCG_DISP_SLOW_SELECT_REG                       0x13F
 #define		DCCG_DISP1_SLOW_SELECT(x)		((x) << 0)
 #define		DCCG_DISP1_SLOW_SELECT_MASK		(7 << 0)
@@ -84,47 +45,6 @@
 #define		DCCG_DISP2_SLOW_SELECT_MASK		(7 << 4)
 #define		DCCG_DISP2_SLOW_SELECT_SHIFT		4
 
-#define	CG_SPLL_FUNC_CNTL				0x180
-#define		SPLL_RESET				(1 << 0)
-#define		SPLL_SLEEP				(1 << 1)
-#define		SPLL_BYPASS_EN				(1 << 3)
-#define		SPLL_REF_DIV(x)				((x) << 4)
-#define		SPLL_REF_DIV_MASK			(0x3f << 4)
-#define		SPLL_PDIV_A(x)				((x) << 20)
-#define		SPLL_PDIV_A_MASK			(0x7f << 20)
-#define		SPLL_PDIV_A_SHIFT			20
-#define	CG_SPLL_FUNC_CNTL_2				0x181
-#define		SCLK_MUX_SEL(x)				((x) << 0)
-#define		SCLK_MUX_SEL_MASK			(0x1ff << 0)
-#define		SPLL_CTLREQ_CHG				(1 << 23)
-#define		SCLK_MUX_UPDATE				(1 << 26)
-#define	CG_SPLL_FUNC_CNTL_3				0x182
-#define		SPLL_FB_DIV(x)				((x) << 0)
-#define		SPLL_FB_DIV_MASK			(0x3ffffff << 0)
-#define		SPLL_FB_DIV_SHIFT			0
-#define		SPLL_DITHEN				(1 << 28)
-#define	CG_SPLL_FUNC_CNTL_4				0x183
-
-#define	SPLL_STATUS					0x185
-#define		SPLL_CHG_STATUS				(1 << 1)
-#define	SPLL_CNTL_MODE					0x186
-#define		SPLL_SW_DIR_CONTROL			(1 << 0)
-#	define SPLL_REFCLK_SEL(x)			((x) << 26)
-#	define SPLL_REFCLK_SEL_MASK			(3 << 26)
-
-#define	CG_SPLL_SPREAD_SPECTRUM				0x188
-#define		SSEN					(1 << 0)
-#define		CLK_S(x)				((x) << 4)
-#define		CLK_S_MASK				(0xfff << 4)
-#define		CLK_S_SHIFT				4
-#define	CG_SPLL_SPREAD_SPECTRUM_2			0x189
-#define		CLK_V(x)				((x) << 0)
-#define		CLK_V_MASK				(0x3ffffff << 0)
-#define		CLK_V_SHIFT				0
-
-#define	CG_SPLL_AUTOSCALE_CNTL				0x18b
-#       define AUTOSCALE_ON_SS_CLEAR                    (1 << 9)
-
 /* discrete uvd clocks */
 #define	CG_UPLL_FUNC_CNTL				0x18d
 #	define UPLL_RESET_MASK				0x00000001
@@ -154,317 +74,13 @@
 #define	CG_UPLL_SPREAD_SPECTRUM				0x194
 #	define SSEN_MASK				0x00000001
 
-#define	MPLL_BYPASSCLK_SEL				0x197
-#	define MPLL_CLKOUT_SEL(x)			((x) << 8)
-#	define MPLL_CLKOUT_SEL_MASK			0xFF00
-
-#define CG_CLKPIN_CNTL                                    0x198
-#       define XTALIN_DIVIDE                              (1 << 1)
-#       define BCLK_AS_XCLK                               (1 << 2)
-#define CG_CLKPIN_CNTL_2                                  0x199
-#       define FORCE_BIF_REFCLK_EN                        (1 << 3)
-#       define MUX_TCLK_TO_XCLK                           (1 << 8)
-
-#define	THM_CLK_CNTL					0x19b
-#	define CMON_CLK_SEL(x)				((x) << 0)
-#	define CMON_CLK_SEL_MASK			0xFF
-#	define TMON_CLK_SEL(x)				((x) << 8)
-#	define TMON_CLK_SEL_MASK			0xFF00
-#define	MISC_CLK_CNTL					0x19c
-#	define DEEP_SLEEP_CLK_SEL(x)			((x) << 0)
-#	define DEEP_SLEEP_CLK_SEL_MASK			0xFF
-#	define ZCLK_SEL(x)				((x) << 8)
-#	define ZCLK_SEL_MASK				0xFF00
-
-#define	CG_THERMAL_CTRL					0x1c0
-#define 	DPM_EVENT_SRC(x)			((x) << 0)
-#define 	DPM_EVENT_SRC_MASK			(7 << 0)
-#define		DIG_THERM_DPM(x)			((x) << 14)
-#define		DIG_THERM_DPM_MASK			0x003FC000
-#define		DIG_THERM_DPM_SHIFT			14
-#define	CG_THERMAL_STATUS				0x1c1
-#define		FDO_PWM_DUTY(x)				((x) << 9)
-#define		FDO_PWM_DUTY_MASK			(0xff << 9)
-#define		FDO_PWM_DUTY_SHIFT			9
-#define	CG_THERMAL_INT					0x1c2
-#define		DIG_THERM_INTH(x)			((x) << 8)
-#define		DIG_THERM_INTH_MASK			0x0000FF00
-#define		DIG_THERM_INTH_SHIFT			8
-#define		DIG_THERM_INTL(x)			((x) << 16)
-#define		DIG_THERM_INTL_MASK			0x00FF0000
-#define		DIG_THERM_INTL_SHIFT			16
-#define 	THERM_INT_MASK_HIGH			(1 << 24)
-#define 	THERM_INT_MASK_LOW			(1 << 25)
-
-#define	CG_MULT_THERMAL_CTRL					0x1c4
-#define		TEMP_SEL(x)					((x) << 20)
-#define		TEMP_SEL_MASK					(0xff << 20)
-#define		TEMP_SEL_SHIFT					20
-#define	CG_MULT_THERMAL_STATUS					0x1c5
-#define		ASIC_MAX_TEMP(x)				((x) << 0)
-#define		ASIC_MAX_TEMP_MASK				0x000001ff
-#define		ASIC_MAX_TEMP_SHIFT				0
-#define		CTF_TEMP(x)					((x) << 9)
-#define		CTF_TEMP_MASK					0x0003fe00
-#define		CTF_TEMP_SHIFT					9
-
-#define	CG_FDO_CTRL0					0x1d5
-#define		FDO_STATIC_DUTY(x)			((x) << 0)
-#define		FDO_STATIC_DUTY_MASK			0x000000FF
-#define		FDO_STATIC_DUTY_SHIFT			0
-#define	CG_FDO_CTRL1					0x1d6
-#define		FMAX_DUTY100(x)				((x) << 0)
-#define		FMAX_DUTY100_MASK			0x000000FF
-#define		FMAX_DUTY100_SHIFT			0
-#define	CG_FDO_CTRL2					0x1d7
-#define		TMIN(x)					((x) << 0)
-#define		TMIN_MASK				0x000000FF
-#define		TMIN_SHIFT				0
-#define		FDO_PWM_MODE(x)				((x) << 11)
-#define		FDO_PWM_MODE_MASK			(7 << 11)
-#define		FDO_PWM_MODE_SHIFT			11
-#define		TACH_PWM_RESP_RATE(x)			((x) << 25)
-#define		TACH_PWM_RESP_RATE_MASK			(0x7f << 25)
-#define		TACH_PWM_RESP_RATE_SHIFT		25
-
-#define CG_TACH_CTRL                                    0x1dc
-#       define EDGE_PER_REV(x)                          ((x) << 0)
-#       define EDGE_PER_REV_MASK                        (0x7 << 0)
-#       define EDGE_PER_REV_SHIFT                       0
-#       define TARGET_PERIOD(x)                         ((x) << 3)
-#       define TARGET_PERIOD_MASK                       0xfffffff8
-#       define TARGET_PERIOD_SHIFT                      3
-#define CG_TACH_STATUS                                  0x1dd
-#       define TACH_PERIOD(x)                           ((x) << 0)
-#       define TACH_PERIOD_MASK                         0xffffffff
-#       define TACH_PERIOD_SHIFT                        0
-
-#define GENERAL_PWRMGT                                  0x1e0
-#       define GLOBAL_PWRMGT_EN                         (1 << 0)
-#       define STATIC_PM_EN                             (1 << 1)
-#       define THERMAL_PROTECTION_DIS                   (1 << 2)
-#       define THERMAL_PROTECTION_TYPE                  (1 << 3)
-#       define SW_SMIO_INDEX(x)                         ((x) << 6)
-#       define SW_SMIO_INDEX_MASK                       (1 << 6)
-#       define SW_SMIO_INDEX_SHIFT                      6
-#       define VOLT_PWRMGT_EN                           (1 << 10)
-#       define DYN_SPREAD_SPECTRUM_EN                   (1 << 23)
-#define CG_TPC                                            0x1e1
-#define SCLK_PWRMGT_CNTL                                  0x1e2
-#       define SCLK_PWRMGT_OFF                            (1 << 0)
-#       define SCLK_LOW_D1                                (1 << 1)
-#       define FIR_RESET                                  (1 << 4)
-#       define FIR_FORCE_TREND_SEL                        (1 << 5)
-#       define FIR_TREND_MODE                             (1 << 6)
-#       define DYN_GFX_CLK_OFF_EN                         (1 << 7)
-#       define GFX_CLK_FORCE_ON                           (1 << 8)
-#       define GFX_CLK_REQUEST_OFF                        (1 << 9)
-#       define GFX_CLK_FORCE_OFF                          (1 << 10)
-#       define GFX_CLK_OFF_ACPI_D1                        (1 << 11)
-#       define GFX_CLK_OFF_ACPI_D2                        (1 << 12)
-#       define GFX_CLK_OFF_ACPI_D3                        (1 << 13)
-#       define DYN_LIGHT_SLEEP_EN                         (1 << 14)
-
-#define TARGET_AND_CURRENT_PROFILE_INDEX                  0x1e6
-#       define CURRENT_STATE_INDEX_MASK                   (0xf << 4)
-#       define CURRENT_STATE_INDEX_SHIFT                  4
-
-#define CG_FTV                                            0x1ef
-
-#define CG_FFCT_0                                         0x1f0
-#       define UTC_0(x)                                   ((x) << 0)
-#       define UTC_0_MASK                                 (0x3ff << 0)
-#       define DTC_0(x)                                   ((x) << 10)
-#       define DTC_0_MASK                                 (0x3ff << 10)
-
-#define CG_BSP                                          0x1ff
-#       define BSP(x)					((x) << 0)
-#       define BSP_MASK					(0xffff << 0)
-#       define BSU(x)					((x) << 16)
-#       define BSU_MASK					(0xf << 16)
-#define CG_AT                                           0x200
-#       define CG_R(x)					((x) << 0)
-#       define CG_R_MASK				(0xffff << 0)
-#       define CG_L(x)					((x) << 16)
-#       define CG_L_MASK				(0xffff << 16)
-
-#define CG_GIT                                          0x201
-#       define CG_GICST(x)                              ((x) << 0)
-#       define CG_GICST_MASK                            (0xffff << 0)
-#       define CG_GIPOT(x)                              ((x) << 16)
-#       define CG_GIPOT_MASK                            (0xffff << 16)
-
-#define CG_SSP                                            0x203
-#       define SST(x)                                     ((x) << 0)
-#       define SST_MASK                                   (0xffff << 0)
-#       define SSTU(x)                                    ((x) << 16)
-#       define SSTU_MASK                                  (0xf << 16)
-
-#define CG_DISPLAY_GAP_CNTL                               0x20a
-#       define DISP1_GAP(x)                               ((x) << 0)
-#       define DISP1_GAP_MASK                             (3 << 0)
-#       define DISP2_GAP(x)                               ((x) << 2)
-#       define DISP2_GAP_MASK                             (3 << 2)
-#       define VBI_TIMER_COUNT(x)                         ((x) << 4)
-#       define VBI_TIMER_COUNT_MASK                       (0x3fff << 4)
-#       define VBI_TIMER_UNIT(x)                          ((x) << 20)
-#       define VBI_TIMER_UNIT_MASK                        (7 << 20)
-#       define DISP1_GAP_MCHG(x)                          ((x) << 24)
-#       define DISP1_GAP_MCHG_MASK                        (3 << 24)
-#       define DISP2_GAP_MCHG(x)                          ((x) << 26)
-#       define DISP2_GAP_MCHG_MASK                        (3 << 26)
-
-#define	CG_ULV_CONTROL					0x21e
-#define	CG_ULV_PARAMETER				0x21f
-
-#define	SMC_SCRATCH0					0x221
-
-#define	CG_CAC_CTRL					0x22e
-#	define CAC_WINDOW(x)				((x) << 0)
-#	define CAC_WINDOW_MASK				0x00ffffff
-
-#define DMIF_ADDR_CONFIG  				0x2F5
-
-#define DMIF_ADDR_CALC  				0x300
-
-#define	PIPE0_DMIF_BUFFER_CONTROL			  0x0328
-#       define DMIF_BUFFERS_ALLOCATED(x)                  ((x) << 0)
-#       define DMIF_BUFFERS_ALLOCATED_COMPLETED           (1 << 4)
-
-#define	SRBM_STATUS				        0x394
-#define		GRBM_RQ_PENDING 			(1 << 5)
-#define		VMC_BUSY 				(1 << 8)
-#define		MCB_BUSY 				(1 << 9)
-#define		MCB_NON_DISPLAY_BUSY 			(1 << 10)
-#define		MCC_BUSY 				(1 << 11)
-#define		MCD_BUSY 				(1 << 12)
-#define		SEM_BUSY 				(1 << 14)
-#define		IH_BUSY 				(1 << 17)
-
-#define	SRBM_SOFT_RESET				        0x398
-#define		SOFT_RESET_BIF				(1 << 1)
-#define		SOFT_RESET_DC				(1 << 5)
-#define		SOFT_RESET_DMA1				(1 << 6)
-#define		SOFT_RESET_GRBM				(1 << 8)
-#define		SOFT_RESET_HDP				(1 << 9)
-#define		SOFT_RESET_IH				(1 << 10)
-#define		SOFT_RESET_MC				(1 << 11)
-#define		SOFT_RESET_ROM				(1 << 14)
-#define		SOFT_RESET_SEM				(1 << 15)
-#define		SOFT_RESET_VMC				(1 << 17)
-#define		SOFT_RESET_DMA				(1 << 20)
-#define		SOFT_RESET_TST				(1 << 21)
-#define		SOFT_RESET_REGBB			(1 << 22)
-#define		SOFT_RESET_ORB				(1 << 23)
-
-#define	CC_SYS_RB_BACKEND_DISABLE			0x3A0
-#define	GC_USER_SYS_RB_BACKEND_DISABLE			0x3A1
-
-#define SRBM_READ_ERROR					0x3A6
-#define SRBM_INT_CNTL					0x3A8
-#define SRBM_INT_ACK					0x3AA
-
-#define	SRBM_STATUS2				        0x3B1
-#define		DMA_BUSY 				(1 << 5)
-#define		DMA1_BUSY 				(1 << 6)
-
-#define VM_L2_CNTL					0x500
-#define		ENABLE_L2_CACHE					(1 << 0)
-#define		ENABLE_L2_FRAGMENT_PROCESSING			(1 << 1)
-#define		L2_CACHE_PTE_ENDIAN_SWAP_MODE(x)		((x) << 2)
-#define		L2_CACHE_PDE_ENDIAN_SWAP_MODE(x)		((x) << 4)
-#define		ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE		(1 << 9)
-#define		ENABLE_L2_PDE0_CACHE_LRU_UPDATE_BY_WRITE	(1 << 10)
-#define		EFFECTIVE_L2_QUEUE_SIZE(x)			(((x) & 7) << 15)
-#define		CONTEXT1_IDENTITY_ACCESS_MODE(x)		(((x) & 3) << 19)
-#define VM_L2_CNTL2					0x501
-#define		INVALIDATE_ALL_L1_TLBS				(1 << 0)
-#define		INVALIDATE_L2_CACHE				(1 << 1)
-#define		INVALIDATE_CACHE_MODE(x)			((x) << 26)
-#define			INVALIDATE_PTE_AND_PDE_CACHES		0
-#define			INVALIDATE_ONLY_PTE_CACHES		1
-#define			INVALIDATE_ONLY_PDE_CACHES		2
-#define VM_L2_CNTL3					0x502
-#define		BANK_SELECT(x)					((x) << 0)
-#define		L2_CACHE_UPDATE_MODE(x)				((x) << 6)
-#define		L2_CACHE_BIGK_FRAGMENT_SIZE(x)			((x) << 15)
-#define		L2_CACHE_BIGK_ASSOCIATIVITY			(1 << 20)
-#define	VM_L2_STATUS					0x503
-#define		L2_BUSY						(1 << 0)
-#define VM_CONTEXT0_CNTL				0x504
-#define		ENABLE_CONTEXT					(1 << 0)
-#define		PAGE_TABLE_DEPTH(x)				(((x) & 3) << 1)
-#define		RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT		(1 << 3)
-#define		RANGE_PROTECTION_FAULT_ENABLE_DEFAULT		(1 << 4)
-#define		DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT	(1 << 6)
-#define		DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT	(1 << 7)
-#define		PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT		(1 << 9)
-#define		PDE0_PROTECTION_FAULT_ENABLE_DEFAULT		(1 << 10)
-#define		VALID_PROTECTION_FAULT_ENABLE_INTERRUPT		(1 << 12)
-#define		VALID_PROTECTION_FAULT_ENABLE_DEFAULT		(1 << 13)
-#define		READ_PROTECTION_FAULT_ENABLE_INTERRUPT		(1 << 15)
-#define		READ_PROTECTION_FAULT_ENABLE_DEFAULT		(1 << 16)
-#define		WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT		(1 << 18)
-#define		WRITE_PROTECTION_FAULT_ENABLE_DEFAULT		(1 << 19)
-#define		PAGE_TABLE_BLOCK_SIZE(x)			(((x) & 0xF) << 24)
-#define VM_CONTEXT1_CNTL				0x505
-#define VM_CONTEXT0_CNTL2				0x50C
-#define VM_CONTEXT1_CNTL2				0x50D
-#define	VM_CONTEXT8_PAGE_TABLE_BASE_ADDR		0x50E
-#define	VM_CONTEXT9_PAGE_TABLE_BASE_ADDR		0x50F
-#define	VM_CONTEXT10_PAGE_TABLE_BASE_ADDR		0x510
-#define	VM_CONTEXT11_PAGE_TABLE_BASE_ADDR		0x511
-#define	VM_CONTEXT12_PAGE_TABLE_BASE_ADDR		0x512
-#define	VM_CONTEXT13_PAGE_TABLE_BASE_ADDR		0x513
-#define	VM_CONTEXT14_PAGE_TABLE_BASE_ADDR		0x514
-#define	VM_CONTEXT15_PAGE_TABLE_BASE_ADDR		0x515
-
-#define	VM_CONTEXT1_PROTECTION_FAULT_ADDR		0x53f
-#define	VM_CONTEXT1_PROTECTION_FAULT_STATUS		0x537
-#define		PROTECTIONS_MASK			(0xf << 0)
-#define		PROTECTIONS_SHIFT			0
-		/* bit 0: range
-		 * bit 1: pde0
-		 * bit 2: valid
-		 * bit 3: read
-		 * bit 4: write
-		 */
-#define		MEMORY_CLIENT_ID_MASK			(0xff << 12)
-#define		MEMORY_CLIENT_ID_SHIFT			12
-#define		MEMORY_CLIENT_RW_MASK			(1 << 24)
-#define		MEMORY_CLIENT_RW_SHIFT			24
-#define		FAULT_VMID_MASK				(0xf << 25)
-#define		FAULT_VMID_SHIFT			25
-
 #define VM_INVALIDATE_REQUEST				0x51E
 #define VM_INVALIDATE_RESPONSE				0x51F
 
-#define VM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR	0x546
-#define VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR	0x547
-
-#define	VM_CONTEXT0_PAGE_TABLE_BASE_ADDR		0x54F
-#define	VM_CONTEXT1_PAGE_TABLE_BASE_ADDR		0x550
-#define	VM_CONTEXT2_PAGE_TABLE_BASE_ADDR		0x551
-#define	VM_CONTEXT3_PAGE_TABLE_BASE_ADDR		0x552
-#define	VM_CONTEXT4_PAGE_TABLE_BASE_ADDR		0x553
-#define	VM_CONTEXT5_PAGE_TABLE_BASE_ADDR		0x554
-#define	VM_CONTEXT6_PAGE_TABLE_BASE_ADDR		0x555
-#define	VM_CONTEXT7_PAGE_TABLE_BASE_ADDR		0x556
-#define	VM_CONTEXT0_PAGE_TABLE_START_ADDR		0x557
-#define	VM_CONTEXT1_PAGE_TABLE_START_ADDR		0x558
-
-#define	VM_CONTEXT0_PAGE_TABLE_END_ADDR			0x55F
-#define	VM_CONTEXT1_PAGE_TABLE_END_ADDR			0x560
-
 #define VM_L2_CG           				0x570
 #define		MC_CG_ENABLE				(1 << 18)
 #define		MC_LS_ENABLE				(1 << 19)
 
-#define MC_SHARED_CHMAP						0x801
-#define		NOOFCHAN_SHIFT					12
-#define		NOOFCHAN_MASK					0x0000f000
-#define MC_SHARED_CHREMAP					0x802
-
 #define	MC_VM_FB_LOCATION				0x809
 #define	MC_VM_AGP_TOP					0x80A
 #define	MC_VM_AGP_BOT					0x80B
@@ -496,21 +112,6 @@
 #define MC_CITF_MISC_WR_CG           			0x993
 #define MC_CITF_MISC_VM_CG           			0x994
 
-#define	MC_ARB_RAMCFG					0x9D8
-#define		NOOFBANK_SHIFT					0
-#define		NOOFBANK_MASK					0x00000003
-#define		NOOFRANK_SHIFT					2
-#define		NOOFRANK_MASK					0x00000004
-#define		NOOFROWS_SHIFT					3
-#define		NOOFROWS_MASK					0x00000038
-#define		NOOFCOLS_SHIFT					6
-#define		NOOFCOLS_MASK					0x000000C0
-#define		CHANSIZE_SHIFT					8
-#define		CHANSIZE_MASK					0x00000100
-#define		CHANSIZE_OVERRIDE				(1 << 11)
-#define		NOOFGROUPS_SHIFT				12
-#define		NOOFGROUPS_MASK					0x00001000
-
 #define	MC_ARB_DRAM_TIMING				0x9DD
 #define	MC_ARB_DRAM_TIMING2				0x9DE
 
@@ -636,20 +237,6 @@
 #define		CLKS(x)					((x) << 0)
 #define		CLKS_MASK				(0xfff << 0)
 
-#define	HDP_HOST_PATH_CNTL				0xB00
-#define 	CLOCK_GATING_DIS			(1 << 23)
-#define	HDP_NONSURFACE_BASE				0xB01
-#define	HDP_NONSURFACE_INFO				0xB02
-#define	HDP_NONSURFACE_SIZE				0xB03
-
-#define HDP_DEBUG0  					0xBCC
-
-#define HDP_ADDR_CONFIG  				0xBD2
-#define HDP_MISC_CNTL					0xBD3
-#define 	HDP_FLUSH_INVALIDATE_CACHE			(1 << 0)
-#define HDP_MEM_POWER_LS				0xBD4
-#define 	HDP_LS_ENABLE				(1 << 0)
-
 #define ATC_MISC_CG           				0xCD4
 
 #define IH_RB_CNTL                                        0xF80
@@ -679,8 +266,6 @@
 #       define MC_WR_CLEAN_CNT(x)                         ((x) << 20)
 #       define MC_VMID(x)                                 ((x) << 25)
 
-#define	CONFIG_MEMSIZE					0x150A
-
 #define INTERRUPT_CNTL                                    0x151A
 #       define IH_DUMMY_RD_OVERRIDE                       (1 << 0)
 #       define IH_DUMMY_RD_EN                             (1 << 1)
@@ -688,734 +273,28 @@
 #       define GEN_IH_INT_EN                              (1 << 8)
 #define INTERRUPT_CNTL2                                   0x151B
 
-#define HDP_MEM_COHERENCY_FLUSH_CNTL			0x1520
-
-#define	BIF_FB_EN						0x1524
-#define		FB_READ_EN					(1 << 0)
-#define		FB_WRITE_EN					(1 << 1)
-
-#define HDP_REG_COHERENCY_FLUSH_CNTL			0x1528
-
-/* DCE6 ELD audio interface */
-#define AZ_F0_CODEC_ENDPOINT_INDEX                       0x1780
-#       define AZ_ENDPOINT_REG_INDEX(x)                  (((x) & 0xff) << 0)
-#       define AZ_ENDPOINT_REG_WRITE_EN                  (1 << 8)
-#define AZ_F0_CODEC_ENDPOINT_DATA                        0x1781
-
-#define AZ_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER          0x25
-#define		SPEAKER_ALLOCATION(x)			(((x) & 0x7f) << 0)
-#define		SPEAKER_ALLOCATION_MASK			(0x7f << 0)
-#define		SPEAKER_ALLOCATION_SHIFT		0
-#define		HDMI_CONNECTION				(1 << 16)
-#define		DP_CONNECTION				(1 << 17)
-
-#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0        0x28 /* LPCM */
-#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR1        0x29 /* AC3 */
-#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR2        0x2A /* MPEG1 */
-#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR3        0x2B /* MP3 */
-#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR4        0x2C /* MPEG2 */
-#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR5        0x2D /* AAC */
-#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR6        0x2E /* DTS */
-#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR7        0x2F /* ATRAC */
-#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR8        0x30 /* one bit audio - leave at 0 (default) */
-#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR9        0x31 /* Dolby Digital */
-#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR10       0x32 /* DTS-HD */
-#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR11       0x33 /* MAT-MLP */
-#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR12       0x34 /* DTS */
-#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR13       0x35 /* WMA Pro */
-#       define MAX_CHANNELS(x)                            (((x) & 0x7) << 0)
-/* max channels minus one.  7 = 8 channels */
-#       define SUPPORTED_FREQUENCIES(x)                   (((x) & 0xff) << 8)
-#       define DESCRIPTOR_BYTE_2(x)                       (((x) & 0xff) << 16)
-#       define SUPPORTED_FREQUENCIES_STEREO(x)            (((x) & 0xff) << 24) /* LPCM only */
-/* SUPPORTED_FREQUENCIES, SUPPORTED_FREQUENCIES_STEREO
- * bit0 = 32 kHz
- * bit1 = 44.1 kHz
- * bit2 = 48 kHz
- * bit3 = 88.2 kHz
- * bit4 = 96 kHz
- * bit5 = 176.4 kHz
- * bit6 = 192 kHz
- */
-
-#define AZ_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC         0x37
-#       define VIDEO_LIPSYNC(x)                           (((x) & 0xff) << 0)
-#       define AUDIO_LIPSYNC(x)                           (((x) & 0xff) << 8)
-/* VIDEO_LIPSYNC, AUDIO_LIPSYNC
- * 0   = invalid
- * x   = legal delay value
- * 255 = sync not supported
- */
-#define AZ_F0_CODEC_PIN_CONTROL_RESPONSE_HBR             0x38
-#       define HBR_CAPABLE                                (1 << 0) /* enabled by default */
-
-#define AZ_F0_CODEC_PIN_CONTROL_SINK_INFO0               0x3a
-#       define MANUFACTURER_ID(x)                        (((x) & 0xffff) << 0)
-#       define PRODUCT_ID(x)                             (((x) & 0xffff) << 16)
-#define AZ_F0_CODEC_PIN_CONTROL_SINK_INFO1               0x3b
-#       define SINK_DESCRIPTION_LEN(x)                   (((x) & 0xff) << 0)
-#define AZ_F0_CODEC_PIN_CONTROL_SINK_INFO2               0x3c
-#       define PORT_ID0(x)                               (((x) & 0xffffffff) << 0)
-#define AZ_F0_CODEC_PIN_CONTROL_SINK_INFO3               0x3d
-#       define PORT_ID1(x)                               (((x) & 0xffffffff) << 0)
-#define AZ_F0_CODEC_PIN_CONTROL_SINK_INFO4               0x3e
-#       define DESCRIPTION0(x)                           (((x) & 0xff) << 0)
-#       define DESCRIPTION1(x)                           (((x) & 0xff) << 8)
-#       define DESCRIPTION2(x)                           (((x) & 0xff) << 16)
-#       define DESCRIPTION3(x)                           (((x) & 0xff) << 24)
-#define AZ_F0_CODEC_PIN_CONTROL_SINK_INFO5               0x3f
-#       define DESCRIPTION4(x)                           (((x) & 0xff) << 0)
-#       define DESCRIPTION5(x)                           (((x) & 0xff) << 8)
-#       define DESCRIPTION6(x)                           (((x) & 0xff) << 16)
-#       define DESCRIPTION7(x)                           (((x) & 0xff) << 24)
-#define AZ_F0_CODEC_PIN_CONTROL_SINK_INFO6               0x40
-#       define DESCRIPTION8(x)                           (((x) & 0xff) << 0)
-#       define DESCRIPTION9(x)                           (((x) & 0xff) << 8)
-#       define DESCRIPTION10(x)                          (((x) & 0xff) << 16)
-#       define DESCRIPTION11(x)                          (((x) & 0xff) << 24)
-#define AZ_F0_CODEC_PIN_CONTROL_SINK_INFO7               0x41
-#       define DESCRIPTION12(x)                          (((x) & 0xff) << 0)
-#       define DESCRIPTION13(x)                          (((x) & 0xff) << 8)
-#       define DESCRIPTION14(x)                          (((x) & 0xff) << 16)
-#       define DESCRIPTION15(x)                          (((x) & 0xff) << 24)
-#define AZ_F0_CODEC_PIN_CONTROL_SINK_INFO8               0x42
-#       define DESCRIPTION16(x)                          (((x) & 0xff) << 0)
-#       define DESCRIPTION17(x)                          (((x) & 0xff) << 8)
-
-#define AZ_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL         0x54
-#       define AUDIO_ENABLED                             (1 << 31)
-
-#define AZ_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT  0x56
-#define		PORT_CONNECTIVITY_MASK				(3 << 30)
-#define		PORT_CONNECTIVITY_SHIFT				30
-
-#define	DC_LB_MEMORY_SPLIT					0x1AC3
-#define		DC_LB_MEMORY_CONFIG(x)				((x) << 20)
-
-#define	PRIORITY_A_CNT						0x1AC6
-#define		PRIORITY_MARK_MASK				0x7fff
-#define		PRIORITY_OFF					(1 << 16)
-#define		PRIORITY_ALWAYS_ON				(1 << 20)
-#define	PRIORITY_B_CNT						0x1AC7
-
-#define	DPG_PIPE_ARBITRATION_CONTROL3				0x1B32
-#       define LATENCY_WATERMARK_MASK(x)			((x) << 16)
-#define	DPG_PIPE_LATENCY_CONTROL				0x1B33
-#       define LATENCY_LOW_WATERMARK(x)				((x) << 0)
-#       define LATENCY_HIGH_WATERMARK(x)			((x) << 16)
-
-/* 0x6bb8, 0x77b8, 0x103b8, 0x10fb8, 0x11bb8, 0x127b8 */
-#define VLINE_STATUS                                    0x1AEE
-#       define VLINE_OCCURRED                           (1 << 0)
-#       define VLINE_ACK                                (1 << 4)
-#       define VLINE_STAT                               (1 << 12)
-#       define VLINE_INTERRUPT                          (1 << 16)
-#       define VLINE_INTERRUPT_TYPE                     (1 << 17)
-/* 0x6bbc, 0x77bc, 0x103bc, 0x10fbc, 0x11bbc, 0x127bc */
-#define VBLANK_STATUS                                   0x1AEF
-#       define VBLANK_OCCURRED                          (1 << 0)
-#       define VBLANK_ACK                               (1 << 4)
-#       define VBLANK_STAT                              (1 << 12)
-#       define VBLANK_INTERRUPT                         (1 << 16)
-#       define VBLANK_INTERRUPT_TYPE                    (1 << 17)
-
-/* 0x6b40, 0x7740, 0x10340, 0x10f40, 0x11b40, 0x12740 */
-#define INT_MASK                                        0x1AD0
-#       define VBLANK_INT_MASK                          (1 << 0)
-#       define VLINE_INT_MASK                           (1 << 4)
-
-#define DISP_INTERRUPT_STATUS                           0x183D
-#       define LB_D1_VLINE_INTERRUPT                    (1 << 2)
-#       define LB_D1_VBLANK_INTERRUPT                   (1 << 3)
-#       define DC_HPD1_INTERRUPT                        (1 << 17)
-#       define DC_HPD1_RX_INTERRUPT                     (1 << 18)
-#       define DACA_AUTODETECT_INTERRUPT                (1 << 22)
-#       define DACB_AUTODETECT_INTERRUPT                (1 << 23)
-#       define DC_I2C_SW_DONE_INTERRUPT                 (1 << 24)
-#       define DC_I2C_HW_DONE_INTERRUPT                 (1 << 25)
-#define DISP_INTERRUPT_STATUS_CONTINUE                  0x183E
-#       define LB_D2_VLINE_INTERRUPT                    (1 << 2)
-#       define LB_D2_VBLANK_INTERRUPT                   (1 << 3)
-#       define DC_HPD2_INTERRUPT                        (1 << 17)
-#       define DC_HPD2_RX_INTERRUPT                     (1 << 18)
-#       define DISP_TIMER_INTERRUPT                     (1 << 24)
-#define DISP_INTERRUPT_STATUS_CONTINUE2                 0x183F
-#       define LB_D3_VLINE_INTERRUPT                    (1 << 2)
-#       define LB_D3_VBLANK_INTERRUPT                   (1 << 3)
-#       define DC_HPD3_INTERRUPT                        (1 << 17)
-#       define DC_HPD3_RX_INTERRUPT                     (1 << 18)
-#define DISP_INTERRUPT_STATUS_CONTINUE3                 0x1840
-#       define LB_D4_VLINE_INTERRUPT                    (1 << 2)
-#       define LB_D4_VBLANK_INTERRUPT                   (1 << 3)
-#       define DC_HPD4_INTERRUPT                        (1 << 17)
-#       define DC_HPD4_RX_INTERRUPT                     (1 << 18)
-#define DISP_INTERRUPT_STATUS_CONTINUE4                 0x1853
-#       define LB_D5_VLINE_INTERRUPT                    (1 << 2)
-#       define LB_D5_VBLANK_INTERRUPT                   (1 << 3)
-#       define DC_HPD5_INTERRUPT                        (1 << 17)
-#       define DC_HPD5_RX_INTERRUPT                     (1 << 18)
-#define DISP_INTERRUPT_STATUS_CONTINUE5                 0x1854
-#       define LB_D6_VLINE_INTERRUPT                    (1 << 2)
-#       define LB_D6_VBLANK_INTERRUPT                   (1 << 3)
-#       define DC_HPD6_INTERRUPT                        (1 << 17)
-#       define DC_HPD6_RX_INTERRUPT                     (1 << 18)
-
-/* 0x6858, 0x7458, 0x10058, 0x10c58, 0x11858, 0x12458 */
-#define GRPH_INT_STATUS                                 0x1A16
-#       define GRPH_PFLIP_INT_OCCURRED                  (1 << 0)
-#       define GRPH_PFLIP_INT_CLEAR                     (1 << 8)
-/* 0x685c, 0x745c, 0x1005c, 0x10c5c, 0x1185c, 0x1245c */
-#define	GRPH_INT_CONTROL			        0x1A17
-#       define GRPH_PFLIP_INT_MASK                      (1 << 0)
-#       define GRPH_PFLIP_INT_TYPE                      (1 << 8)
-
-#define	DAC_AUTODETECT_INT_CONTROL			0x19F2
-
-#define DC_HPD1_INT_STATUS                              0x1807
-#define DC_HPD2_INT_STATUS                              0x180A
-#define DC_HPD3_INT_STATUS                              0x180D
-#define DC_HPD4_INT_STATUS                              0x1810
-#define DC_HPD5_INT_STATUS                              0x1813
-#define DC_HPD6_INT_STATUS                              0x1816
-#       define DC_HPDx_INT_STATUS                       (1 << 0)
-#       define DC_HPDx_SENSE                            (1 << 1)
-#       define DC_HPDx_RX_INT_STATUS                    (1 << 8)
-
-#define DC_HPD1_INT_CONTROL                             0x1808
-#define DC_HPD2_INT_CONTROL                             0x180B
-#define DC_HPD3_INT_CONTROL                             0x180E
-#define DC_HPD4_INT_CONTROL                             0x1811
-#define DC_HPD5_INT_CONTROL                             0x1814
-#define DC_HPD6_INT_CONTROL                             0x1817
-#       define DC_HPDx_INT_ACK                          (1 << 0)
-#       define DC_HPDx_INT_POLARITY                     (1 << 8)
-#       define DC_HPDx_INT_EN                           (1 << 16)
-#       define DC_HPDx_RX_INT_ACK                       (1 << 20)
-#       define DC_HPDx_RX_INT_EN                        (1 << 24)
-
-#define DC_HPD1_CONTROL                                   0x1809
-#define DC_HPD2_CONTROL                                   0x180C
-#define DC_HPD3_CONTROL                                   0x180F
-#define DC_HPD4_CONTROL                                   0x1812
-#define DC_HPD5_CONTROL                                   0x1815
-#define DC_HPD6_CONTROL                                   0x1818
-#       define DC_HPDx_CONNECTION_TIMER(x)                ((x) << 0)
-#       define DC_HPDx_RX_INT_TIMER(x)                    ((x) << 16)
-#       define DC_HPDx_EN                                 (1 << 28)
-
-#define DPG_PIPE_STUTTER_CONTROL                          0x1B35
-#       define STUTTER_ENABLE                             (1 << 0)
-
-/* 0x6e98, 0x7a98, 0x10698, 0x11298, 0x11e98, 0x12a98 */
-#define CRTC_STATUS_FRAME_COUNT                         0x1BA6
-
-/* Audio clocks */
-#define DCCG_AUDIO_DTO_SOURCE                           0x05ac
-#       define DCCG_AUDIO_DTO0_SOURCE_SEL(x) ((x) << 0) /* crtc0 - crtc5 */
-#       define DCCG_AUDIO_DTO_SEL            (1 << 4)   /* 0=dto0 1=dto1 */
-
-#define DCCG_AUDIO_DTO0_PHASE                           0x05b0
-#define DCCG_AUDIO_DTO0_MODULE                          0x05b4
-#define DCCG_AUDIO_DTO1_PHASE                           0x05c0
-#define DCCG_AUDIO_DTO1_MODULE                          0x05c4
-
-#define AFMT_AUDIO_SRC_CONTROL                          0x1c4f
-#define		AFMT_AUDIO_SRC_SELECT(x)		(((x) & 7) << 0)
-/* AFMT_AUDIO_SRC_SELECT
- * 0 = stream0
- * 1 = stream1
- * 2 = stream2
- * 3 = stream3
- * 4 = stream4
- * 5 = stream5
- */
-
-#define	GRBM_CNTL					0x2000
-#define		GRBM_READ_TIMEOUT(x)				((x) << 0)
-
-#define	GRBM_STATUS2					0x2002
-#define		RLC_RQ_PENDING 					(1 << 0)
-#define		RLC_BUSY 					(1 << 8)
-#define		TC_BUSY 					(1 << 9)
-
-#define	GRBM_STATUS					0x2004
-#define		CMDFIFO_AVAIL_MASK				0x0000000F
-#define		RING2_RQ_PENDING				(1 << 4)
-#define		SRBM_RQ_PENDING					(1 << 5)
-#define		RING1_RQ_PENDING				(1 << 6)
-#define		CF_RQ_PENDING					(1 << 7)
-#define		PF_RQ_PENDING					(1 << 8)
-#define		GDS_DMA_RQ_PENDING				(1 << 9)
-#define		GRBM_EE_BUSY					(1 << 10)
-#define		DB_CLEAN					(1 << 12)
-#define		CB_CLEAN					(1 << 13)
-#define		TA_BUSY 					(1 << 14)
-#define		GDS_BUSY 					(1 << 15)
-#define		VGT_BUSY					(1 << 17)
-#define		IA_BUSY_NO_DMA					(1 << 18)
-#define		IA_BUSY						(1 << 19)
-#define		SX_BUSY 					(1 << 20)
-#define		SPI_BUSY					(1 << 22)
-#define		BCI_BUSY					(1 << 23)
-#define		SC_BUSY 					(1 << 24)
-#define		PA_BUSY 					(1 << 25)
-#define		DB_BUSY 					(1 << 26)
-#define		CP_COHERENCY_BUSY      				(1 << 28)
-#define		CP_BUSY 					(1 << 29)
-#define		CB_BUSY 					(1 << 30)
-#define		GUI_ACTIVE					(1 << 31)
-#define	GRBM_STATUS_SE0					0x2005
-#define	GRBM_STATUS_SE1					0x2006
-#define		SE_DB_CLEAN					(1 << 1)
-#define		SE_CB_CLEAN					(1 << 2)
-#define		SE_BCI_BUSY					(1 << 22)
-#define		SE_VGT_BUSY					(1 << 23)
-#define		SE_PA_BUSY					(1 << 24)
-#define		SE_TA_BUSY					(1 << 25)
-#define		SE_SX_BUSY					(1 << 26)
-#define		SE_SPI_BUSY					(1 << 27)
-#define		SE_SC_BUSY					(1 << 29)
-#define		SE_DB_BUSY					(1 << 30)
-#define		SE_CB_BUSY					(1 << 31)
-
-#define	GRBM_SOFT_RESET					0x2008
-#define		SOFT_RESET_CP					(1 << 0)
-#define		SOFT_RESET_CB					(1 << 1)
-#define		SOFT_RESET_RLC					(1 << 2)
-#define		SOFT_RESET_DB					(1 << 3)
-#define		SOFT_RESET_GDS					(1 << 4)
-#define		SOFT_RESET_PA					(1 << 5)
-#define		SOFT_RESET_SC					(1 << 6)
-#define		SOFT_RESET_BCI					(1 << 7)
-#define		SOFT_RESET_SPI					(1 << 8)
-#define		SOFT_RESET_SX					(1 << 10)
-#define		SOFT_RESET_TC					(1 << 11)
-#define		SOFT_RESET_TA					(1 << 12)
-#define		SOFT_RESET_VGT					(1 << 14)
-#define		SOFT_RESET_IA					(1 << 15)
-
-#define GRBM_GFX_INDEX          			0x200B
-#define		INSTANCE_INDEX(x)			((x) << 0)
-#define		SH_INDEX(x)     			((x) << 8)
-#define		SE_INDEX(x)     			((x) << 16)
-#define		SH_BROADCAST_WRITES      		(1 << 29)
-#define		INSTANCE_BROADCAST_WRITES      		(1 << 30)
-#define		SE_BROADCAST_WRITES      		(1 << 31)
-
-#define GRBM_INT_CNTL                                   0x2018
-#       define RDERR_INT_ENABLE                         (1 << 0)
-#       define GUI_IDLE_INT_ENABLE                      (1 << 19)
-
-#define	CP_STRMOUT_CNTL					0x213F
-#define	SCRATCH_REG0					0x2140
-#define	SCRATCH_REG1					0x2141
-#define	SCRATCH_REG2					0x2142
-#define	SCRATCH_REG3					0x2143
-#define	SCRATCH_REG4					0x2144
-#define	SCRATCH_REG5					0x2145
-#define	SCRATCH_REG6					0x2146
-#define	SCRATCH_REG7					0x2147
-
-#define	SCRATCH_UMSK					0x2150
-#define	SCRATCH_ADDR					0x2151
-
-#define	CP_SEM_WAIT_TIMER				0x216F
-
-#define	CP_SEM_INCOMPLETE_TIMER_CNTL			0x2172
-
-#define CP_ME_CNTL					0x21B6
-#define		CP_CE_HALT					(1 << 24)
-#define		CP_PFP_HALT					(1 << 26)
-#define		CP_ME_HALT					(1 << 28)
-
-#define	CP_COHER_CNTL2					0x217A
-
-#define	CP_RB2_RPTR					0x21BE
-#define	CP_RB1_RPTR					0x21BF
-#define	CP_RB0_RPTR					0x21C0
-#define	CP_RB_WPTR_DELAY				0x21C1
-
-#define	CP_QUEUE_THRESHOLDS				0x21D8
-#define		ROQ_IB1_START(x)				((x) << 0)
-#define		ROQ_IB2_START(x)				((x) << 8)
-#define CP_MEQ_THRESHOLDS				0x21D9
-#define		MEQ1_START(x)				((x) << 0)
-#define		MEQ2_START(x)				((x) << 8)
-
-#define	CP_PERFMON_CNTL					0x21FF
-
 #define	VGT_VTX_VECT_EJECT_REG				0x222C
-
-#define	VGT_CACHE_INVALIDATION				0x2231
-#define		CACHE_INVALIDATION(x)				((x) << 0)
-#define			VC_ONLY						0
-#define			TC_ONLY						1
-#define			VC_AND_TC					2
-#define		AUTO_INVLD_EN(x)				((x) << 6)
-#define			NO_AUTO						0
-#define			ES_AUTO						1
-#define			GS_AUTO						2
-#define			ES_AND_GS_AUTO					3
 #define	VGT_ESGS_RING_SIZE				0x2232
 #define	VGT_GSVS_RING_SIZE				0x2233
-
 #define	VGT_GS_VERTEX_REUSE				0x2235
-
 #define	VGT_PRIMITIVE_TYPE				0x2256
 #define	VGT_INDEX_TYPE					0x2257
-
 #define	VGT_NUM_INDICES					0x225C
 #define	VGT_NUM_INSTANCES				0x225D
-
 #define	VGT_TF_RING_SIZE				0x2262
-
 #define	VGT_HS_OFFCHIP_PARAM				0x226C
-
 #define	VGT_TF_MEMORY_BASE				0x226E
 
-#define CC_GC_SHADER_ARRAY_CONFIG			0x226F
-#define		INACTIVE_CUS_MASK			0xFFFF0000
-#define		INACTIVE_CUS_SHIFT			16
-#define GC_USER_SHADER_ARRAY_CONFIG			0x2270
-
-#define	PA_CL_ENHANCE					0x2285
-#define		CLIP_VTX_REORDER_ENA				(1 << 0)
-#define		NUM_CLIP_SEQ(x)					((x) << 1)
-
-#define	PA_SU_LINE_STIPPLE_VALUE			0x2298
-
-#define	PA_SC_LINE_STIPPLE_STATE			0x22C4
-
-#define	PA_SC_FORCE_EOV_MAX_CNTS			0x22C9
-#define		FORCE_EOV_MAX_CLK_CNT(x)			((x) << 0)
-#define		FORCE_EOV_MAX_REZ_CNT(x)			((x) << 16)
-
-#define	PA_SC_FIFO_SIZE					0x22F3
-#define		SC_FRONTEND_PRIM_FIFO_SIZE(x)			((x) << 0)
-#define		SC_BACKEND_PRIM_FIFO_SIZE(x)			((x) << 6)
-#define		SC_HIZ_TILE_FIFO_SIZE(x)			((x) << 15)
-#define		SC_EARLYZ_TILE_FIFO_SIZE(x)			((x) << 23)
-
 #define	PA_SC_ENHANCE					0x22FC
 
-#define	SQ_CONFIG					0x2300
-
-#define	SQC_CACHES					0x2302
-
-#define SQ_POWER_THROTTLE                               0x2396
-#define		MIN_POWER(x)				((x) << 0)
-#define		MIN_POWER_MASK				(0x3fff << 0)
-#define		MIN_POWER_SHIFT				0
-#define		MAX_POWER(x)				((x) << 16)
-#define		MAX_POWER_MASK				(0x3fff << 16)
-#define		MAX_POWER_SHIFT				0
-#define SQ_POWER_THROTTLE2                              0x2397
-#define		MAX_POWER_DELTA(x)			((x) << 0)
-#define		MAX_POWER_DELTA_MASK			(0x3fff << 0)
-#define		MAX_POWER_DELTA_SHIFT			0
-#define		STI_SIZE(x)				((x) << 16)
-#define		STI_SIZE_MASK				(0x3ff << 16)
-#define		STI_SIZE_SHIFT				16
-#define		LTI_RATIO(x)				((x) << 27)
-#define		LTI_RATIO_MASK				(0xf << 27)
-#define		LTI_RATIO_SHIFT				27
-
-#define	SX_DEBUG_1					0x2418
-
-#define	SPI_STATIC_THREAD_MGMT_1			0x2438
-#define	SPI_STATIC_THREAD_MGMT_2			0x2439
-#define	SPI_STATIC_THREAD_MGMT_3			0x243A
-#define	SPI_PS_MAX_WAVE_ID				0x243B
-
-#define	SPI_CONFIG_CNTL					0x2440
-
-#define	SPI_CONFIG_CNTL_1				0x244F
-#define		VTX_DONE_DELAY(x)				((x) << 0)
-#define		INTERP_ONE_PRIM_PER_ROW				(1 << 4)
-
-#define	CGTS_TCC_DISABLE				0x2452
-#define	CGTS_USER_TCC_DISABLE				0x2453
-#define		TCC_DISABLE_MASK				0xFFFF0000
-#define		TCC_DISABLE_SHIFT				16
-#define	CGTS_SM_CTRL_REG				0x2454
-#define		OVERRIDE				(1 << 21)
-#define		LS_OVERRIDE				(1 << 22)
-
-#define	SPI_LB_CU_MASK					0x24D5
-
 #define	TA_CNTL_AUX					0x2542
 
-#define CC_RB_BACKEND_DISABLE				0x263D
-#define		BACKEND_DISABLE(x)     			((x) << 16)
-#define GB_ADDR_CONFIG  				0x263E
-#define		NUM_PIPES(x)				((x) << 0)
-#define		NUM_PIPES_MASK				0x00000007
-#define		NUM_PIPES_SHIFT				0
-#define		PIPE_INTERLEAVE_SIZE(x)			((x) << 4)
-#define		PIPE_INTERLEAVE_SIZE_MASK		0x00000070
-#define		PIPE_INTERLEAVE_SIZE_SHIFT		4
-#define		NUM_SHADER_ENGINES(x)			((x) << 12)
-#define		NUM_SHADER_ENGINES_MASK			0x00003000
-#define		NUM_SHADER_ENGINES_SHIFT		12
-#define		SHADER_ENGINE_TILE_SIZE(x)     		((x) << 16)
-#define		SHADER_ENGINE_TILE_SIZE_MASK		0x00070000
-#define		SHADER_ENGINE_TILE_SIZE_SHIFT		16
-#define		NUM_GPUS(x)     			((x) << 20)
-#define		NUM_GPUS_MASK				0x00700000
-#define		NUM_GPUS_SHIFT				20
-#define		MULTI_GPU_TILE_SIZE(x)     		((x) << 24)
-#define		MULTI_GPU_TILE_SIZE_MASK		0x03000000
-#define		MULTI_GPU_TILE_SIZE_SHIFT		24
-#define		ROW_SIZE(x)             		((x) << 28)
-#define		ROW_SIZE_MASK				0x30000000
-#define		ROW_SIZE_SHIFT				28
-
-#define	GB_TILE_MODE0					0x2644
-#       define MICRO_TILE_MODE(x)				((x) << 0)
-#              define	ADDR_SURF_DISPLAY_MICRO_TILING		0
-#              define	ADDR_SURF_THIN_MICRO_TILING		1
-#              define	ADDR_SURF_DEPTH_MICRO_TILING		2
-#       define ARRAY_MODE(x)					((x) << 2)
-#              define	ARRAY_LINEAR_GENERAL			0
-#              define	ARRAY_LINEAR_ALIGNED			1
-#              define	ARRAY_1D_TILED_THIN1			2
-#              define	ARRAY_2D_TILED_THIN1			4
-#       define PIPE_CONFIG(x)					((x) << 6)
-#              define	ADDR_SURF_P2				0
-#              define	ADDR_SURF_P4_8x16			4
-#              define	ADDR_SURF_P4_16x16			5
-#              define	ADDR_SURF_P4_16x32			6
-#              define	ADDR_SURF_P4_32x32			7
-#              define	ADDR_SURF_P8_16x16_8x16			8
-#              define	ADDR_SURF_P8_16x32_8x16			9
-#              define	ADDR_SURF_P8_32x32_8x16			10
-#              define	ADDR_SURF_P8_16x32_16x16		11
-#              define	ADDR_SURF_P8_32x32_16x16		12
-#              define	ADDR_SURF_P8_32x32_16x32		13
-#              define	ADDR_SURF_P8_32x64_32x32		14
-#       define TILE_SPLIT(x)					((x) << 11)
-#              define	ADDR_SURF_TILE_SPLIT_64B		0
-#              define	ADDR_SURF_TILE_SPLIT_128B		1
-#              define	ADDR_SURF_TILE_SPLIT_256B		2
-#              define	ADDR_SURF_TILE_SPLIT_512B		3
-#              define	ADDR_SURF_TILE_SPLIT_1KB		4
-#              define	ADDR_SURF_TILE_SPLIT_2KB		5
-#              define	ADDR_SURF_TILE_SPLIT_4KB		6
-#       define BANK_WIDTH(x)					((x) << 14)
-#              define	ADDR_SURF_BANK_WIDTH_1			0
-#              define	ADDR_SURF_BANK_WIDTH_2			1
-#              define	ADDR_SURF_BANK_WIDTH_4			2
-#              define	ADDR_SURF_BANK_WIDTH_8			3
-#       define BANK_HEIGHT(x)					((x) << 16)
-#              define	ADDR_SURF_BANK_HEIGHT_1			0
-#              define	ADDR_SURF_BANK_HEIGHT_2			1
-#              define	ADDR_SURF_BANK_HEIGHT_4			2
-#              define	ADDR_SURF_BANK_HEIGHT_8			3
-#       define MACRO_TILE_ASPECT(x)				((x) << 18)
-#              define	ADDR_SURF_MACRO_ASPECT_1		0
-#              define	ADDR_SURF_MACRO_ASPECT_2		1
-#              define	ADDR_SURF_MACRO_ASPECT_4		2
-#              define	ADDR_SURF_MACRO_ASPECT_8		3
-#       define NUM_BANKS(x)					((x) << 20)
-#              define	ADDR_SURF_2_BANK			0
-#              define	ADDR_SURF_4_BANK			1
-#              define	ADDR_SURF_8_BANK			2
-#              define	ADDR_SURF_16_BANK			3
-#define	GB_TILE_MODE1					0x2645
-#define	GB_TILE_MODE2					0x2646
-#define	GB_TILE_MODE3					0x2647
-#define	GB_TILE_MODE4					0x2648
-#define	GB_TILE_MODE5					0x2649
-#define	GB_TILE_MODE6					0x264a
-#define	GB_TILE_MODE7					0x264b
-#define	GB_TILE_MODE8					0x264c
-#define	GB_TILE_MODE9					0x264d
-#define	GB_TILE_MODE10					0x264e
-#define	GB_TILE_MODE11					0x264f
-#define	GB_TILE_MODE12					0x2650
-#define	GB_TILE_MODE13					0x2651
-#define	GB_TILE_MODE14					0x2652
-#define	GB_TILE_MODE15					0x2653
-#define	GB_TILE_MODE16					0x2654
-#define	GB_TILE_MODE17					0x2655
-#define	GB_TILE_MODE18					0x2656
-#define	GB_TILE_MODE19					0x2657
-#define	GB_TILE_MODE20					0x2658
-#define	GB_TILE_MODE21					0x2659
-#define	GB_TILE_MODE22					0x265a
-#define	GB_TILE_MODE23					0x265b
-#define	GB_TILE_MODE24					0x265c
-#define	GB_TILE_MODE25					0x265d
-#define	GB_TILE_MODE26					0x265e
-#define	GB_TILE_MODE27					0x265f
-#define	GB_TILE_MODE28					0x2660
-#define	GB_TILE_MODE29					0x2661
-#define	GB_TILE_MODE30					0x2662
-#define	GB_TILE_MODE31					0x2663
-
-#define	CB_PERFCOUNTER0_SELECT0				0x2688
-#define	CB_PERFCOUNTER0_SELECT1				0x2689
-#define	CB_PERFCOUNTER1_SELECT0				0x268A
-#define	CB_PERFCOUNTER1_SELECT1				0x268B
-#define	CB_PERFCOUNTER2_SELECT0				0x268C
-#define	CB_PERFCOUNTER2_SELECT1				0x268D
-#define	CB_PERFCOUNTER3_SELECT0				0x268E
-#define	CB_PERFCOUNTER3_SELECT1				0x268F
-
-#define	CB_CGTT_SCLK_CTRL				0x2698
-
-#define	GC_USER_RB_BACKEND_DISABLE			0x26DF
-#define		BACKEND_DISABLE_MASK			0x00FF0000
-#define		BACKEND_DISABLE_SHIFT			16
-
-#define	TCP_CHAN_STEER_LO				0x2B03
-#define	TCP_CHAN_STEER_HI				0x2B94
-
-#define	CP_RB0_BASE					0x3040
-#define	CP_RB0_CNTL					0x3041
-#define		RB_BUFSZ(x)					((x) << 0)
-#define		RB_BLKSZ(x)					((x) << 8)
-#define		BUF_SWAP_32BIT					(2 << 16)
-#define		RB_NO_UPDATE					(1 << 27)
-#define		RB_RPTR_WR_ENA					(1 << 31)
-
-#define	CP_RB0_RPTR_ADDR				0x3043
-#define	CP_RB0_RPTR_ADDR_HI				0x3044
-#define	CP_RB0_WPTR					0x3045
-
-#define	CP_PFP_UCODE_ADDR				0x3054
-#define	CP_PFP_UCODE_DATA				0x3055
-#define	CP_ME_RAM_RADDR					0x3056
-#define	CP_ME_RAM_WADDR					0x3057
-#define	CP_ME_RAM_DATA					0x3058
-
-#define	CP_CE_UCODE_ADDR				0x305A
-#define	CP_CE_UCODE_DATA				0x305B
-
-#define	CP_RB1_BASE					0x3060
-#define	CP_RB1_CNTL					0x3061
-#define	CP_RB1_RPTR_ADDR				0x3062
-#define	CP_RB1_RPTR_ADDR_HI				0x3063
-#define	CP_RB1_WPTR					0x3064
-#define	CP_RB2_BASE					0x3065
-#define	CP_RB2_CNTL					0x3066
-#define	CP_RB2_RPTR_ADDR				0x3067
-#define	CP_RB2_RPTR_ADDR_HI				0x3068
-#define	CP_RB2_WPTR					0x3069
-#define CP_INT_CNTL_RING0                               0x306A
-#define CP_INT_CNTL_RING1                               0x306B
-#define CP_INT_CNTL_RING2                               0x306C
-#       define CNTX_BUSY_INT_ENABLE                     (1 << 19)
-#       define CNTX_EMPTY_INT_ENABLE                    (1 << 20)
-#       define WAIT_MEM_SEM_INT_ENABLE                  (1 << 21)
-#       define TIME_STAMP_INT_ENABLE                    (1 << 26)
-#       define CP_RINGID2_INT_ENABLE                    (1 << 29)
-#       define CP_RINGID1_INT_ENABLE                    (1 << 30)
-#       define CP_RINGID0_INT_ENABLE                    (1 << 31)
-#define CP_INT_STATUS_RING0                             0x306D
-#define CP_INT_STATUS_RING1                             0x306E
-#define CP_INT_STATUS_RING2                             0x306F
-#       define WAIT_MEM_SEM_INT_STAT                    (1 << 21)
-#       define TIME_STAMP_INT_STAT                      (1 << 26)
-#       define CP_RINGID2_INT_STAT                      (1 << 29)
-#       define CP_RINGID1_INT_STAT                      (1 << 30)
-#       define CP_RINGID0_INT_STAT                      (1 << 31)
-
-#define	CP_MEM_SLP_CNTL					0x3079
-#       define CP_MEM_LS_EN                             (1 << 0)
-
-#define	CP_DEBUG					0x307F
-
-#define RLC_CNTL                                          0x30C0
-#       define RLC_ENABLE                                 (1 << 0)
-#define RLC_RL_BASE                                       0x30C1
-#define RLC_RL_SIZE                                       0x30C2
-#define RLC_LB_CNTL                                       0x30C3
-#       define LOAD_BALANCE_ENABLE                        (1 << 0)
-#define RLC_SAVE_AND_RESTORE_BASE                         0x30C4
-#define RLC_LB_CNTR_MAX                                   0x30C5
-#define RLC_LB_CNTR_INIT                                  0x30C6
-
-#define RLC_CLEAR_STATE_RESTORE_BASE                      0x30C8
-
-#define RLC_UCODE_ADDR                                    0x30CB
-#define RLC_UCODE_DATA                                    0x30CC
-
-#define RLC_GPU_CLOCK_COUNT_LSB                           0x30CE
-#define RLC_GPU_CLOCK_COUNT_MSB                           0x30CF
-#define RLC_CAPTURE_GPU_CLOCK_COUNT                       0x30D0
-#define RLC_MC_CNTL                                       0x30D1
-#define RLC_UCODE_CNTL                                    0x30D2
-#define RLC_STAT                                          0x30D3
-#       define RLC_BUSY_STATUS                            (1 << 0)
-#       define GFX_POWER_STATUS                           (1 << 1)
-#       define GFX_CLOCK_STATUS                           (1 << 2)
-#       define GFX_LS_STATUS                              (1 << 3)
-
-#define	RLC_PG_CNTL					0x30D7
-#	define GFX_PG_ENABLE				(1 << 0)
-#	define GFX_PG_SRC				(1 << 1)
-
-#define	RLC_CGTT_MGCG_OVERRIDE				0x3100
-#define	RLC_CGCG_CGLS_CTRL				0x3101
-#	define CGCG_EN					(1 << 0)
-#	define CGLS_EN					(1 << 1)
-
-#define	RLC_TTOP_D					0x3105
-#	define RLC_PUD(x)				((x) << 0)
-#	define RLC_PUD_MASK				(0xff << 0)
-#	define RLC_PDD(x)				((x) << 8)
-#	define RLC_PDD_MASK				(0xff << 8)
-#	define RLC_TTPD(x)				((x) << 16)
-#	define RLC_TTPD_MASK				(0xff << 16)
-#	define RLC_MSD(x)				((x) << 24)
-#	define RLC_MSD_MASK				(0xff << 24)
-
-#define RLC_LB_INIT_CU_MASK                               0x3107
-
-#define	RLC_PG_AO_CU_MASK				0x310B
-#define	RLC_MAX_PG_CU					0x310C
-#	define MAX_PU_CU(x)				((x) << 0)
-#	define MAX_PU_CU_MASK				(0xff << 0)
-#define	RLC_AUTO_PG_CTRL				0x310C
-#	define AUTO_PG_EN				(1 << 0)
-#	define GRBM_REG_SGIT(x)				((x) << 3)
-#	define GRBM_REG_SGIT_MASK			(0xffff << 3)
-#	define PG_AFTER_GRBM_REG_ST(x)			((x) << 19)
-#	define PG_AFTER_GRBM_REG_ST_MASK		(0x1fff << 19)
-
-#define RLC_SERDES_WR_MASTER_MASK_0                       0x3115
-#define RLC_SERDES_WR_MASTER_MASK_1                       0x3116
-#define RLC_SERDES_WR_CTRL                                0x3117
-
-#define RLC_SERDES_MASTER_BUSY_0                          0x3119
-#define RLC_SERDES_MASTER_BUSY_1                          0x311A
-
-#define RLC_GCPM_GENERAL_3                                0x311E
-
-#define	DB_RENDER_CONTROL				0xA000
-
-#define DB_DEPTH_INFO                                   0xA00F
-
-#define PA_SC_RASTER_CONFIG                             0xA0D4
-#	define RB_MAP_PKR0(x)				((x) << 0)
-#	define RB_MAP_PKR0_MASK				(0x3 << 0)
-#	define RB_MAP_PKR1(x)				((x) << 2)
-#	define RB_MAP_PKR1_MASK				(0x3 << 2)
-#       define RASTER_CONFIG_RB_MAP_0                   0
-#       define RASTER_CONFIG_RB_MAP_1                   1
-#       define RASTER_CONFIG_RB_MAP_2                   2
-#       define RASTER_CONFIG_RB_MAP_3                   3
+// #define PA_SC_RASTER_CONFIG                             0xA0D4
 #	define RB_XSEL2(x)				((x) << 4)
 #	define RB_XSEL2_MASK				(0x3 << 4)
 #	define RB_XSEL					(1 << 6)
 #	define RB_YSEL					(1 << 7)
 #	define PKR_MAP(x)				((x) << 8)
-#	define PKR_MAP_MASK				(0x3 << 8)
-#       define RASTER_CONFIG_PKR_MAP_0			0
-#       define RASTER_CONFIG_PKR_MAP_1			1
-#       define RASTER_CONFIG_PKR_MAP_2			2
-#       define RASTER_CONFIG_PKR_MAP_3			3
 #	define PKR_XSEL(x)				((x) << 10)
 #	define PKR_XSEL_MASK				(0x3 << 10)
 #	define PKR_YSEL(x)				((x) << 12)
@@ -1427,228 +306,27 @@
 #	define SC_YSEL(x)				((x) << 20)
 #	define SC_YSEL_MASK				(0x3 << 20)
 #	define SE_MAP(x)				((x) << 24)
-#	define SE_MAP_MASK				(0x3 << 24)
-#       define RASTER_CONFIG_SE_MAP_0			0
-#       define RASTER_CONFIG_SE_MAP_1			1
-#       define RASTER_CONFIG_SE_MAP_2			2
-#       define RASTER_CONFIG_SE_MAP_3			3
 #	define SE_XSEL(x)				((x) << 26)
 #	define SE_XSEL_MASK				(0x3 << 26)
 #	define SE_YSEL(x)				((x) << 28)
 #	define SE_YSEL_MASK				(0x3 << 28)
 
-
-#define VGT_EVENT_INITIATOR                             0xA2A4
-#       define SAMPLE_STREAMOUTSTATS1                   (1 << 0)
-#       define SAMPLE_STREAMOUTSTATS2                   (2 << 0)
-#       define SAMPLE_STREAMOUTSTATS3                   (3 << 0)
-#       define CACHE_FLUSH_TS                           (4 << 0)
-#       define CACHE_FLUSH                              (6 << 0)
-#       define CS_PARTIAL_FLUSH                         (7 << 0)
-#       define VGT_STREAMOUT_RESET                      (10 << 0)
-#       define END_OF_PIPE_INCR_DE                      (11 << 0)
-#       define END_OF_PIPE_IB_END                       (12 << 0)
-#       define RST_PIX_CNT                              (13 << 0)
-#       define VS_PARTIAL_FLUSH                         (15 << 0)
-#       define PS_PARTIAL_FLUSH                         (16 << 0)
-#       define CACHE_FLUSH_AND_INV_TS_EVENT             (20 << 0)
-#       define ZPASS_DONE                               (21 << 0)
-#       define CACHE_FLUSH_AND_INV_EVENT                (22 << 0)
-#       define PERFCOUNTER_START                        (23 << 0)
-#       define PERFCOUNTER_STOP                         (24 << 0)
-#       define PIPELINESTAT_START                       (25 << 0)
-#       define PIPELINESTAT_STOP                        (26 << 0)
-#       define PERFCOUNTER_SAMPLE                       (27 << 0)
-#       define SAMPLE_PIPELINESTAT                      (30 << 0)
-#       define SAMPLE_STREAMOUTSTATS                    (32 << 0)
-#       define RESET_VTX_CNT                            (33 << 0)
-#       define VGT_FLUSH                                (36 << 0)
-#       define BOTTOM_OF_PIPE_TS                        (40 << 0)
-#       define DB_CACHE_FLUSH_AND_INV                   (42 << 0)
-#       define FLUSH_AND_INV_DB_DATA_TS                 (43 << 0)
-#       define FLUSH_AND_INV_DB_META                    (44 << 0)
-#       define FLUSH_AND_INV_CB_DATA_TS                 (45 << 0)
-#       define FLUSH_AND_INV_CB_META                    (46 << 0)
-#       define CS_DONE                                  (47 << 0)
-#       define PS_DONE                                  (48 << 0)
-#       define FLUSH_AND_INV_CB_PIXEL_DATA              (49 << 0)
-#       define THREAD_TRACE_START                       (51 << 0)
-#       define THREAD_TRACE_STOP                        (52 << 0)
-#       define THREAD_TRACE_FLUSH                       (54 << 0)
-#       define THREAD_TRACE_FINISH                      (55 << 0)
-
-/* PIF PHY0 registers idx/data 0x8/0xc */
-#define PB0_PIF_CNTL                                      0x10
-#       define LS2_EXIT_TIME(x)                           ((x) << 17)
-#       define LS2_EXIT_TIME_MASK                         (0x7 << 17)
-#       define LS2_EXIT_TIME_SHIFT                        17
-#define PB0_PIF_PAIRING                                   0x11
-#       define MULTI_PIF                                  (1 << 25)
-#define PB0_PIF_PWRDOWN_0                                 0x12
-#       define PLL_POWER_STATE_IN_TXS2_0(x)               ((x) << 7)
-#       define PLL_POWER_STATE_IN_TXS2_0_MASK             (0x7 << 7)
-#       define PLL_POWER_STATE_IN_TXS2_0_SHIFT            7
-#       define PLL_POWER_STATE_IN_OFF_0(x)                ((x) << 10)
-#       define PLL_POWER_STATE_IN_OFF_0_MASK              (0x7 << 10)
-#       define PLL_POWER_STATE_IN_OFF_0_SHIFT             10
-#       define PLL_RAMP_UP_TIME_0(x)                      ((x) << 24)
-#       define PLL_RAMP_UP_TIME_0_MASK                    (0x7 << 24)
-#       define PLL_RAMP_UP_TIME_0_SHIFT                   24
-#define PB0_PIF_PWRDOWN_1                                 0x13
-#       define PLL_POWER_STATE_IN_TXS2_1(x)               ((x) << 7)
-#       define PLL_POWER_STATE_IN_TXS2_1_MASK             (0x7 << 7)
-#       define PLL_POWER_STATE_IN_TXS2_1_SHIFT            7
-#       define PLL_POWER_STATE_IN_OFF_1(x)                ((x) << 10)
-#       define PLL_POWER_STATE_IN_OFF_1_MASK              (0x7 << 10)
-#       define PLL_POWER_STATE_IN_OFF_1_SHIFT             10
-#       define PLL_RAMP_UP_TIME_1(x)                      ((x) << 24)
-#       define PLL_RAMP_UP_TIME_1_MASK                    (0x7 << 24)
-#       define PLL_RAMP_UP_TIME_1_SHIFT                   24
-
-#define PB0_PIF_PWRDOWN_2                                 0x17
-#       define PLL_POWER_STATE_IN_TXS2_2(x)               ((x) << 7)
-#       define PLL_POWER_STATE_IN_TXS2_2_MASK             (0x7 << 7)
-#       define PLL_POWER_STATE_IN_TXS2_2_SHIFT            7
-#       define PLL_POWER_STATE_IN_OFF_2(x)                ((x) << 10)
-#       define PLL_POWER_STATE_IN_OFF_2_MASK              (0x7 << 10)
-#       define PLL_POWER_STATE_IN_OFF_2_SHIFT             10
-#       define PLL_RAMP_UP_TIME_2(x)                      ((x) << 24)
-#       define PLL_RAMP_UP_TIME_2_MASK                    (0x7 << 24)
-#       define PLL_RAMP_UP_TIME_2_SHIFT                   24
-#define PB0_PIF_PWRDOWN_3                                 0x18
-#       define PLL_POWER_STATE_IN_TXS2_3(x)               ((x) << 7)
-#       define PLL_POWER_STATE_IN_TXS2_3_MASK             (0x7 << 7)
-#       define PLL_POWER_STATE_IN_TXS2_3_SHIFT            7
-#       define PLL_POWER_STATE_IN_OFF_3(x)                ((x) << 10)
-#       define PLL_POWER_STATE_IN_OFF_3_MASK              (0x7 << 10)
-#       define PLL_POWER_STATE_IN_OFF_3_SHIFT             10
-#       define PLL_RAMP_UP_TIME_3(x)                      ((x) << 24)
-#       define PLL_RAMP_UP_TIME_3_MASK                    (0x7 << 24)
-#       define PLL_RAMP_UP_TIME_3_SHIFT                   24
-/* PIF PHY1 registers idx/data 0x10/0x14 */
-#define PB1_PIF_CNTL                                      0x10
-#define PB1_PIF_PAIRING                                   0x11
-#define PB1_PIF_PWRDOWN_0                                 0x12
-#define PB1_PIF_PWRDOWN_1                                 0x13
-
-#define PB1_PIF_PWRDOWN_2                                 0x17
-#define PB1_PIF_PWRDOWN_3                                 0x18
-/* PCIE registers idx/data 0x30/0x34 */
-#define PCIE_CNTL2                                        0x1c /* PCIE */
-#       define SLV_MEM_LS_EN                              (1 << 16)
-#       define SLV_MEM_AGGRESSIVE_LS_EN                   (1 << 17)
-#       define MST_MEM_LS_EN                              (1 << 18)
-#       define REPLAY_MEM_LS_EN                           (1 << 19)
-#define PCIE_LC_STATUS1                                   0x28 /* PCIE */
-#       define LC_REVERSE_RCVR                            (1 << 0)
-#       define LC_REVERSE_XMIT                            (1 << 1)
-#       define LC_OPERATING_LINK_WIDTH_MASK               (0x7 << 2)
-#       define LC_OPERATING_LINK_WIDTH_SHIFT              2
-#       define LC_DETECTED_LINK_WIDTH_MASK                (0x7 << 5)
-#       define LC_DETECTED_LINK_WIDTH_SHIFT               5
-
-#define PCIE_P_CNTL                                       0x40 /* PCIE */
-#       define P_IGNORE_EDB_ERR                           (1 << 6)
-
 /* PCIE PORT registers idx/data 0x38/0x3c */
-#define PCIE_LC_CNTL                                      0xa0
-#       define LC_L0S_INACTIVITY(x)                       ((x) << 8)
-#       define LC_L0S_INACTIVITY_MASK                     (0xf << 8)
-#       define LC_L0S_INACTIVITY_SHIFT                    8
-#       define LC_L1_INACTIVITY(x)                        ((x) << 12)
-#       define LC_L1_INACTIVITY_MASK                      (0xf << 12)
-#       define LC_L1_INACTIVITY_SHIFT                     12
-#       define LC_PMI_TO_L1_DIS                           (1 << 16)
-#       define LC_ASPM_TO_L1_DIS                          (1 << 24)
-#define PCIE_LC_LINK_WIDTH_CNTL                           0xa2 /* PCIE_P */
-#       define LC_LINK_WIDTH_SHIFT                        0
-#       define LC_LINK_WIDTH_MASK                         0x7
+// #define PCIE_LC_LINK_WIDTH_CNTL                           0xa2 /* PCIE_P */
 #       define LC_LINK_WIDTH_X0                           0
 #       define LC_LINK_WIDTH_X1                           1
 #       define LC_LINK_WIDTH_X2                           2
 #       define LC_LINK_WIDTH_X4                           3
 #       define LC_LINK_WIDTH_X8                           4
 #       define LC_LINK_WIDTH_X16                          6
-#       define LC_LINK_WIDTH_RD_SHIFT                     4
-#       define LC_LINK_WIDTH_RD_MASK                      0x70
-#       define LC_RECONFIG_ARC_MISSING_ESCAPE             (1 << 7)
-#       define LC_RECONFIG_NOW                            (1 << 8)
-#       define LC_RENEGOTIATION_SUPPORT                   (1 << 9)
-#       define LC_RENEGOTIATE_EN                          (1 << 10)
-#       define LC_SHORT_RECONFIG_EN                       (1 << 11)
-#       define LC_UPCONFIGURE_SUPPORT                     (1 << 12)
-#       define LC_UPCONFIGURE_DIS                         (1 << 13)
-#       define LC_DYN_LANES_PWR_STATE(x)                  ((x) << 21)
-#       define LC_DYN_LANES_PWR_STATE_MASK                (0x3 << 21)
-#       define LC_DYN_LANES_PWR_STATE_SHIFT               21
-#define PCIE_LC_N_FTS_CNTL                                0xa3 /* PCIE_P */
-#       define LC_XMIT_N_FTS(x)                           ((x) << 0)
-#       define LC_XMIT_N_FTS_MASK                         (0xff << 0)
-#       define LC_XMIT_N_FTS_SHIFT                        0
-#       define LC_XMIT_N_FTS_OVERRIDE_EN                  (1 << 8)
-#       define LC_N_FTS_MASK                              (0xff << 24)
-#define PCIE_LC_SPEED_CNTL                                0xa4 /* PCIE_P */
-#       define LC_GEN2_EN_STRAP                           (1 << 0)
-#       define LC_GEN3_EN_STRAP                           (1 << 1)
-#       define LC_TARGET_LINK_SPEED_OVERRIDE_EN           (1 << 2)
-#       define LC_TARGET_LINK_SPEED_OVERRIDE_MASK         (0x3 << 3)
-#       define LC_TARGET_LINK_SPEED_OVERRIDE_SHIFT        3
-#       define LC_FORCE_EN_SW_SPEED_CHANGE                (1 << 5)
-#       define LC_FORCE_DIS_SW_SPEED_CHANGE               (1 << 6)
-#       define LC_FORCE_EN_HW_SPEED_CHANGE                (1 << 7)
-#       define LC_FORCE_DIS_HW_SPEED_CHANGE               (1 << 8)
-#       define LC_INITIATE_LINK_SPEED_CHANGE              (1 << 9)
-#       define LC_SPEED_CHANGE_ATTEMPTS_ALLOWED_MASK      (0x3 << 10)
-#       define LC_SPEED_CHANGE_ATTEMPTS_ALLOWED_SHIFT     10
-#       define LC_CURRENT_DATA_RATE_MASK                  (0x3 << 13) /* 0/1/2 = gen1/2/3 */
-#       define LC_CURRENT_DATA_RATE_SHIFT                 13
-#       define LC_CLR_FAILED_SPD_CHANGE_CNT               (1 << 16)
-#       define LC_OTHER_SIDE_EVER_SENT_GEN2               (1 << 18)
-#       define LC_OTHER_SIDE_SUPPORTS_GEN2                (1 << 19)
-#       define LC_OTHER_SIDE_EVER_SENT_GEN3               (1 << 20)
-#       define LC_OTHER_SIDE_SUPPORTS_GEN3                (1 << 21)
-
-#define PCIE_LC_CNTL2                                     0xb1
-#       define LC_ALLOW_PDWN_IN_L1                        (1 << 17)
-#       define LC_ALLOW_PDWN_IN_L23                       (1 << 18)
-
-#define PCIE_LC_CNTL3                                     0xb5 /* PCIE_P */
-#       define LC_GO_TO_RECOVERY                          (1 << 30)
-#define PCIE_LC_CNTL4                                     0xb6 /* PCIE_P */
-#       define LC_REDO_EQ                                 (1 << 5)
-#       define LC_SET_QUIESCE                             (1 << 13)
-
-/*
- * UVD
- */
-#define UVD_UDEC_ADDR_CONFIG				0x3bd3
-#define UVD_UDEC_DB_ADDR_CONFIG				0x3bd4
-#define UVD_UDEC_DBW_ADDR_CONFIG			0x3bd5
-#define UVD_RBC_RB_RPTR					0x3da4
-#define UVD_RBC_RB_WPTR					0x3da5
-#define UVD_STATUS					0x3daf
-
-#define	UVD_CGC_CTRL					0x3dc2
-#	define DCM					(1 << 0)
-#	define CG_DT(x)					((x) << 2)
-#	define CG_DT_MASK				(0xf << 2)
-#	define CLK_OD(x)				((x) << 6)
-#	define CLK_OD_MASK				(0x1f << 6)
-
- /* UVD CTX indirect */
-#define	UVD_CGC_MEM_CTRL				0xC0
-#define	UVD_CGC_CTRL2					0xC1
-#	define DYN_OR_EN				(1 << 0)
-#	define DYN_RR_EN				(1 << 1)
-#	define G_DIV_ID(x)				((x) << 2)
-#	define G_DIV_ID_MASK				(0x7 << 2)
 
 /*
  * PM4
  */
-#define PACKET0(reg, n)	((RADEON_PACKET_TYPE0 << 30) |			\
-			 (((reg) >> 2) & 0xFFFF) |			\
-			 ((n) & 0x3FFF) << 16)
+#define PACKET_TYPE0    0
+#define PACKET0(reg, n) ((PACKET_TYPE0 << 30) |				\
+                         ((reg) & 0xFFFF) |				\
+                         ((n) & 0x3FFF) << 16)
 #define CP_PACKET2			0x80000000
 #define		PACKET2_PAD_SHIFT		0
 #define		PACKET2_PAD_MASK		(0x3fffffff << 0)
@@ -1874,45 +552,7 @@
 /* ASYNC DMA - first instance at 0xd000, second at 0xd800 */
 #define DMA0_REGISTER_OFFSET                              0x0 /* not a register */
 #define DMA1_REGISTER_OFFSET                              0x200 /* not a register */
-
-#define DMA_RB_CNTL                                       0x3400
-#       define DMA_RB_ENABLE                              (1 << 0)
-#       define DMA_RB_SIZE(x)                             ((x) << 1) /* log2 */
-#       define DMA_RB_SWAP_ENABLE                         (1 << 9) /* 8IN32 */
-#       define DMA_RPTR_WRITEBACK_ENABLE                  (1 << 12)
-#       define DMA_RPTR_WRITEBACK_SWAP_ENABLE             (1 << 13)  /* 8IN32 */
-#       define DMA_RPTR_WRITEBACK_TIMER(x)                ((x) << 16) /* log2 */
-#define DMA_RB_BASE                                       0x3401
-#define DMA_RB_RPTR                                       0x3402
-#define DMA_RB_WPTR                                       0x3403
-
-#define DMA_RB_RPTR_ADDR_HI                               0x3407
-#define DMA_RB_RPTR_ADDR_LO                               0x3408
-
-#define DMA_IB_CNTL                                       0x3409
-#       define DMA_IB_ENABLE                              (1 << 0)
-#       define DMA_IB_SWAP_ENABLE                         (1 << 4)
-#       define CMD_VMID_FORCE                             (1 << 31)
-#define DMA_IB_RPTR                                       0x340a
-#define DMA_CNTL                                          0x340b
-#       define TRAP_ENABLE                                (1 << 0)
-#       define SEM_INCOMPLETE_INT_ENABLE                  (1 << 1)
-#       define SEM_WAIT_INT_ENABLE                        (1 << 2)
-#       define DATA_SWAP_ENABLE                           (1 << 3)
-#       define FENCE_SWAP_ENABLE                          (1 << 4)
-#       define CTXEMPTY_INT_ENABLE                        (1 << 28)
-#define DMA_STATUS_REG                                    0x340d
-#       define DMA_IDLE                                   (1 << 0)
-#define DMA_TILING_CONFIG  				  0x342e
-
-#define	DMA_POWER_CNTL					0x342f
-#       define MEM_POWER_OVERRIDE                       (1 << 8)
-#define	DMA_CLK_CTRL					0x3430
-
-#define	DMA_PG						0x3435
-#	define PG_CNTL_ENABLE				(1 << 0)
-#define	DMA_PGFSM_CONFIG				0x3436
-#define	DMA_PGFSM_WRITE					0x3437
+#define SDMA_MAX_INSTANCE 2
 
 #define DMA_PACKET(cmd, b, t, s, n)	((((cmd) & 0xF) << 28) |	\
 					 (((b) & 0x1) << 26) |		\
@@ -1941,6 +581,7 @@
 #define	DMA_PACKET_POLL_REG_MEM				  0xe
 #define	DMA_PACKET_NOP					  0xf
 
+/* VCE */
 #define VCE_STATUS					0x20004
 #define VCE_VCPU_CNTL					0x20014
 #define		VCE_CLK_EN				(1 << 0)
@@ -1991,416 +632,144 @@
 
 //#dce stupp
 /* display controller offsets used for crtc/cur/lut/grph/viewport/etc. */
-#define SI_CRTC0_REGISTER_OFFSET                0 //(0x6df0 - 0x6df0)/4
-#define SI_CRTC1_REGISTER_OFFSET                0x300 //(0x79f0 - 0x6df0)/4
-#define SI_CRTC2_REGISTER_OFFSET                0x2600 //(0x105f0 - 0x6df0)/4
-#define SI_CRTC3_REGISTER_OFFSET                0x2900 //(0x111f0 - 0x6df0)/4
-#define SI_CRTC4_REGISTER_OFFSET                0x2c00 //(0x11df0 - 0x6df0)/4
-#define SI_CRTC5_REGISTER_OFFSET                0x2f00 //(0x129f0 - 0x6df0)/4
+#define CRTC0_REGISTER_OFFSET                 (0x1b7c - 0x1b7c) //(0x6df0 - 0x6df0)/4
+#define CRTC1_REGISTER_OFFSET                 (0x1e7c - 0x1b7c) //(0x79f0 - 0x6df0)/4
+#define CRTC2_REGISTER_OFFSET                 (0x417c - 0x1b7c) //(0x105f0 - 0x6df0)/4
+#define CRTC3_REGISTER_OFFSET                 (0x447c - 0x1b7c) //(0x111f0 - 0x6df0)/4
+#define CRTC4_REGISTER_OFFSET                 (0x477c - 0x1b7c) //(0x11df0 - 0x6df0)/4
+#define CRTC5_REGISTER_OFFSET                 (0x4a7c - 0x1b7c) //(0x129f0 - 0x6df0)/4
+
+/* hpd instance offsets */
+#define HPD0_REGISTER_OFFSET                 (0x1807 - 0x1807)
+#define HPD1_REGISTER_OFFSET                 (0x180a - 0x1807)
+#define HPD2_REGISTER_OFFSET                 (0x180d - 0x1807)
+#define HPD3_REGISTER_OFFSET                 (0x1810 - 0x1807)
+#define HPD4_REGISTER_OFFSET                 (0x1813 - 0x1807)
+#define HPD5_REGISTER_OFFSET                 (0x1816 - 0x1807)
+
+/* audio endpt instance offsets */
+#define AUD0_REGISTER_OFFSET                 (0x1780 - 0x1780)
+#define AUD1_REGISTER_OFFSET                 (0x1786 - 0x1780)
+#define AUD2_REGISTER_OFFSET                 (0x178c - 0x1780)
+#define AUD3_REGISTER_OFFSET                 (0x1792 - 0x1780)
+#define AUD4_REGISTER_OFFSET                 (0x1798 - 0x1780)
+#define AUD5_REGISTER_OFFSET                 (0x179d - 0x1780)
+#define AUD6_REGISTER_OFFSET                 (0x17a4 - 0x1780)
 
 #define CURSOR_WIDTH 64
 #define CURSOR_HEIGHT 64
-#define AMDGPU_MM_INDEX		        0x0000
-#define AMDGPU_MM_DATA		        0x0001
-
-#define VERDE_NUM_CRTC 6
-#define	BLACKOUT_MODE_MASK			0x00000007
-#define	VGA_RENDER_CONTROL			0xC0
-#define R_000300_VGA_RENDER_CONTROL             0xC0
-#define C_000300_VGA_VSTATUS_CNTL               0xFFFCFFFF
-#define EVERGREEN_CRTC_STATUS                   0x1BA3
-#define EVERGREEN_CRTC_V_BLANK                  (1 << 0)
-#define EVERGREEN_CRTC_STATUS_POSITION          0x1BA4
-/* CRTC blocks at 0x6df0, 0x79f0, 0x105f0, 0x111f0, 0x11df0, 0x129f0 */
-#define EVERGREEN_CRTC_V_BLANK_START_END                0x1b8d
-#define EVERGREEN_CRTC_CONTROL                          0x1b9c
-#define EVERGREEN_CRTC_MASTER_EN                 (1 << 0)
-#define EVERGREEN_CRTC_DISP_READ_REQUEST_DISABLE (1 << 24)
-#define EVERGREEN_CRTC_BLANK_CONTROL                    0x1b9d
-#define EVERGREEN_CRTC_BLANK_DATA_EN             (1 << 8)
-#define EVERGREEN_CRTC_V_BLANK                   (1 << 0)
-#define EVERGREEN_CRTC_STATUS_HV_COUNT                  0x1ba8
-#define EVERGREEN_CRTC_UPDATE_LOCK                      0x1bb5
-#define EVERGREEN_MASTER_UPDATE_LOCK                    0x1bbd
-#define EVERGREEN_MASTER_UPDATE_MODE                    0x1bbe
-#define EVERGREEN_GRPH_UPDATE_LOCK               (1 << 16)
-#define EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH     0x1a07
-#define EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS_HIGH   0x1a08
-#define EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS          0x1a04
-#define EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS        0x1a05
-#define EVERGREEN_GRPH_UPDATE                           0x1a11
-#define EVERGREEN_VGA_MEMORY_BASE_ADDRESS               0xc4
-#define EVERGREEN_VGA_MEMORY_BASE_ADDRESS_HIGH          0xc9
-#define EVERGREEN_GRPH_SURFACE_UPDATE_PENDING    (1 << 2)
-
-#define EVERGREEN_DATA_FORMAT                           0x1ac0
-#       define EVERGREEN_INTERLEAVE_EN                  (1 << 0)
-
-#define MC_SHARED_CHMAP__NOOFCHAN_MASK 0xf000
-#define MC_SHARED_CHMAP__NOOFCHAN__SHIFT 0xc
-
-#define R600_D1GRPH_ARRAY_MODE_LINEAR_GENERAL            (0 << 20)
-#define R600_D1GRPH_ARRAY_MODE_LINEAR_ALIGNED            (1 << 20)
-#define R600_D1GRPH_ARRAY_MODE_1D_TILED_THIN1            (2 << 20)
-#define R600_D1GRPH_ARRAY_MODE_2D_TILED_THIN1            (4 << 20)
-
-#define R700_D1GRPH_PRIMARY_SURFACE_ADDRESS_HIGH                0x1a45
-#define R700_D2GRPH_PRIMARY_SURFACE_ADDRESS_HIGH                0x1845
-
-#define R700_D2GRPH_SECONDARY_SURFACE_ADDRESS_HIGH              0x1847
-#define R700_D1GRPH_SECONDARY_SURFACE_ADDRESS_HIGH              0x1a47
-
-#define DISP_INTERRUPT_STATUS__LB_D1_VBLANK_INTERRUPT_MASK 0x8
-#define DISP_INTERRUPT_STATUS_CONTINUE__LB_D2_VBLANK_INTERRUPT_MASK 0x8
-#define DISP_INTERRUPT_STATUS_CONTINUE2__LB_D3_VBLANK_INTERRUPT_MASK 0x8
-#define DISP_INTERRUPT_STATUS_CONTINUE3__LB_D4_VBLANK_INTERRUPT_MASK 0x8
-#define DISP_INTERRUPT_STATUS_CONTINUE4__LB_D5_VBLANK_INTERRUPT_MASK 0x8
-#define DISP_INTERRUPT_STATUS_CONTINUE5__LB_D6_VBLANK_INTERRUPT_MASK 0x8
-
-#define DISP_INTERRUPT_STATUS__LB_D1_VLINE_INTERRUPT_MASK 0x4
-#define DISP_INTERRUPT_STATUS_CONTINUE__LB_D2_VLINE_INTERRUPT_MASK 0x4
-#define DISP_INTERRUPT_STATUS_CONTINUE2__LB_D3_VLINE_INTERRUPT_MASK 0x4
-#define DISP_INTERRUPT_STATUS_CONTINUE3__LB_D4_VLINE_INTERRUPT_MASK 0x4
-#define DISP_INTERRUPT_STATUS_CONTINUE4__LB_D5_VLINE_INTERRUPT_MASK 0x4
-#define DISP_INTERRUPT_STATUS_CONTINUE5__LB_D6_VLINE_INTERRUPT_MASK 0x4
-
-#define DISP_INTERRUPT_STATUS__DC_HPD1_INTERRUPT_MASK 0x20000
-#define DISP_INTERRUPT_STATUS_CONTINUE__DC_HPD2_INTERRUPT_MASK 0x20000
-#define DISP_INTERRUPT_STATUS_CONTINUE2__DC_HPD3_INTERRUPT_MASK 0x20000
-#define DISP_INTERRUPT_STATUS_CONTINUE3__DC_HPD4_INTERRUPT_MASK 0x20000
-#define DISP_INTERRUPT_STATUS_CONTINUE4__DC_HPD5_INTERRUPT_MASK 0x20000
-#define DISP_INTERRUPT_STATUS_CONTINUE5__DC_HPD6_INTERRUPT_MASK 0x20000
-
-#define GRPH_INTERRUPT_STATUS__GRPH_PFLIP_INT_OCCURRED_MASK 0x1
-#define GRPH_INTERRUPT_STATUS__GRPH_PFLIP_INT_CLEAR_MASK 0x100
-
-#define DC_HPD1_INT_CONTROL__DC_HPD1_INT_ACK_MASK 0x1
-
-#define R600_D1GRPH_SWAP_CONTROL                               0x1843
-#define R600_D1GRPH_SWAP_ENDIAN_NONE                    (0 << 0)
-#define R600_D1GRPH_SWAP_ENDIAN_16BIT                   (1 << 0)
-#define R600_D1GRPH_SWAP_ENDIAN_32BIT                   (2 << 0)
-#define R600_D1GRPH_SWAP_ENDIAN_64BIT                   (3 << 0)
-
-#define AVIVO_D1VGA_CONTROL					0x00cc
-#       define AVIVO_DVGA_CONTROL_MODE_ENABLE            (1 << 0)
-#       define AVIVO_DVGA_CONTROL_TIMING_SELECT          (1 << 8)
-#       define AVIVO_DVGA_CONTROL_SYNC_POLARITY_SELECT   (1 << 9)
-#       define AVIVO_DVGA_CONTROL_OVERSCAN_TIMING_SELECT (1 << 10)
-#       define AVIVO_DVGA_CONTROL_OVERSCAN_COLOR_EN      (1 << 16)
-#       define AVIVO_DVGA_CONTROL_ROTATE                 (1 << 24)
-#define AVIVO_D2VGA_CONTROL					0x00ce
-
-#define R600_BUS_CNTL                                           0x1508
-#       define R600_BIOS_ROM_DIS                                (1 << 1)
+
 
 #define R600_ROM_CNTL                              0x580
 #       define R600_SCK_OVERWRITE                  (1 << 1)
 #       define R600_SCK_PRESCALE_CRYSTAL_CLK_SHIFT 28
 #       define R600_SCK_PRESCALE_CRYSTAL_CLK_MASK  (0xf << 28)
 
-#define GRPH_INTERRUPT_CONTROL__GRPH_PFLIP_INT_MASK_MASK 0x1
-
-#define FMT_BIT_DEPTH_CONTROL                0x1bf2
-#define FMT_TRUNCATE_EN               (1 << 0)
-#define FMT_TRUNCATE_DEPTH            (1 << 4)
-#define FMT_SPATIAL_DITHER_EN         (1 << 8)
-#define FMT_SPATIAL_DITHER_MODE(x)    ((x) << 9)
-#define FMT_SPATIAL_DITHER_DEPTH      (1 << 12)
-#define FMT_FRAME_RANDOM_ENABLE       (1 << 13)
-#define FMT_RGB_RANDOM_ENABLE         (1 << 14)
-#define FMT_HIGHPASS_RANDOM_ENABLE    (1 << 15)
-#define FMT_TEMPORAL_DITHER_EN        (1 << 16)
-#define FMT_TEMPORAL_DITHER_DEPTH     (1 << 20)
-#define FMT_TEMPORAL_DITHER_OFFSET(x) ((x) << 21)
-#define FMT_TEMPORAL_LEVEL            (1 << 24)
-#define FMT_TEMPORAL_DITHER_RESET     (1 << 25)
-#define FMT_25FRC_SEL(x)              ((x) << 26)
-#define FMT_50FRC_SEL(x)              ((x) << 28)
-#define FMT_75FRC_SEL(x)              ((x) << 30)
-
-#define EVERGREEN_DC_LUT_CONTROL                        0x1a80
-#define EVERGREEN_DC_LUT_BLACK_OFFSET_BLUE              0x1a81
-#define EVERGREEN_DC_LUT_BLACK_OFFSET_GREEN             0x1a82
-#define EVERGREEN_DC_LUT_BLACK_OFFSET_RED               0x1a83
-#define EVERGREEN_DC_LUT_WHITE_OFFSET_BLUE              0x1a84
-#define EVERGREEN_DC_LUT_WHITE_OFFSET_GREEN             0x1a85
-#define EVERGREEN_DC_LUT_WHITE_OFFSET_RED               0x1a86
-#define EVERGREEN_DC_LUT_30_COLOR                       0x1a7c
-#define EVERGREEN_DC_LUT_RW_INDEX                       0x1a79
-#define EVERGREEN_DC_LUT_WRITE_EN_MASK                  0x1a7e
-#define EVERGREEN_DC_LUT_RW_MODE                        0x1a78
-
-#define EVERGREEN_GRPH_ENABLE                           0x1a00
-#define EVERGREEN_GRPH_CONTROL                          0x1a01
-#define EVERGREEN_GRPH_DEPTH(x)                  (((x) & 0x3) << 0)
-#define EVERGREEN_GRPH_DEPTH_8BPP                0
-#define EVERGREEN_GRPH_DEPTH_16BPP               1
-#define EVERGREEN_GRPH_DEPTH_32BPP               2
-#define EVERGREEN_GRPH_NUM_BANKS(x)              (((x) & 0x3) << 2)
-#define EVERGREEN_ADDR_SURF_2_BANK               0
-#define EVERGREEN_ADDR_SURF_4_BANK               1
-#define EVERGREEN_ADDR_SURF_8_BANK               2
-#define EVERGREEN_ADDR_SURF_16_BANK              3
-#define EVERGREEN_GRPH_Z(x)                      (((x) & 0x3) << 4)
-#define EVERGREEN_GRPH_BANK_WIDTH(x)             (((x) & 0x3) << 6)
-#define EVERGREEN_ADDR_SURF_BANK_WIDTH_1         0
-#define EVERGREEN_ADDR_SURF_BANK_WIDTH_2         1
-#define EVERGREEN_ADDR_SURF_BANK_WIDTH_4         2
-#define EVERGREEN_ADDR_SURF_BANK_WIDTH_8         3
-#define EVERGREEN_GRPH_FORMAT(x)                 (((x) & 0x7) << 8)
-
-#define EVERGREEN_GRPH_FORMAT_INDEXED            0
-#define EVERGREEN_GRPH_FORMAT_ARGB1555           0
-#define EVERGREEN_GRPH_FORMAT_ARGB565            1
-#define EVERGREEN_GRPH_FORMAT_ARGB4444           2
-#define EVERGREEN_GRPH_FORMAT_AI88               3
-#define EVERGREEN_GRPH_FORMAT_MONO16             4
-#define EVERGREEN_GRPH_FORMAT_BGRA5551           5
+#define GRPH_ARRAY_LINEAR_GENERAL      0
+#define GRPH_ARRAY_LINEAR_ALIGNED      1
+#define GRPH_ARRAY_1D_TILED_THIN1      2
+#define GRPH_ARRAY_2D_TILED_THIN1      4
+
+#define ES_AND_GS_AUTO       3
+#define BUF_SWAP_32BIT       (2 << 16)
+
+#define GRPH_DEPTH_8BPP                0
+#define GRPH_DEPTH_16BPP               1
+#define GRPH_DEPTH_32BPP               2
+
+/* 8 BPP */
+#define GRPH_FORMAT_INDEXED            0
+
+/* 16 BPP */
+#define GRPH_FORMAT_ARGB1555           0
+#define GRPH_FORMAT_ARGB565            1
+#define GRPH_FORMAT_ARGB4444           2
+#define GRPH_FORMAT_AI88               3
+#define GRPH_FORMAT_MONO16             4
+#define GRPH_FORMAT_BGRA5551           5
 
 /* 32 BPP */
-#define EVERGREEN_GRPH_FORMAT_ARGB8888           0
-#define EVERGREEN_GRPH_FORMAT_ARGB2101010        1
-#define EVERGREEN_GRPH_FORMAT_32BPP_DIG          2
-#define EVERGREEN_GRPH_FORMAT_8B_ARGB2101010     3
-#define EVERGREEN_GRPH_FORMAT_BGRA1010102        4
-#define EVERGREEN_GRPH_FORMAT_8B_BGRA1010102     5
-#define EVERGREEN_GRPH_FORMAT_RGB111110          6
-#define EVERGREEN_GRPH_FORMAT_BGR101111          7
-#define EVERGREEN_GRPH_BANK_HEIGHT(x)            (((x) & 0x3) << 11)
-#define EVERGREEN_ADDR_SURF_BANK_HEIGHT_1        0
-#define EVERGREEN_ADDR_SURF_BANK_HEIGHT_2        1
-#define EVERGREEN_ADDR_SURF_BANK_HEIGHT_4        2
-#define EVERGREEN_ADDR_SURF_BANK_HEIGHT_8        3
-#define EVERGREEN_GRPH_TILE_SPLIT(x)             (((x) & 0x7) << 13)
-#define EVERGREEN_ADDR_SURF_TILE_SPLIT_64B       0
-#define EVERGREEN_ADDR_SURF_TILE_SPLIT_128B      1
-#define EVERGREEN_ADDR_SURF_TILE_SPLIT_256B      2
-#define EVERGREEN_ADDR_SURF_TILE_SPLIT_512B      3
-#define EVERGREEN_ADDR_SURF_TILE_SPLIT_1KB       4
-#define EVERGREEN_ADDR_SURF_TILE_SPLIT_2KB       5
-#define EVERGREEN_ADDR_SURF_TILE_SPLIT_4KB       6
-#define EVERGREEN_GRPH_MACRO_TILE_ASPECT(x)      (((x) & 0x3) << 18)
-#define EVERGREEN_ADDR_SURF_MACRO_TILE_ASPECT_1  0
-#define EVERGREEN_ADDR_SURF_MACRO_TILE_ASPECT_2  1
-#define EVERGREEN_ADDR_SURF_MACRO_TILE_ASPECT_4  2
-#define EVERGREEN_ADDR_SURF_MACRO_TILE_ASPECT_8  3
-#define EVERGREEN_GRPH_ARRAY_MODE(x)             (((x) & 0x7) << 20)
-#define EVERGREEN_GRPH_ARRAY_LINEAR_GENERAL      0
-#define EVERGREEN_GRPH_ARRAY_LINEAR_ALIGNED      1
-#define EVERGREEN_GRPH_ARRAY_1D_TILED_THIN1      2
-#define EVERGREEN_GRPH_ARRAY_2D_TILED_THIN1      4
-#define EVERGREEN_ADDR_SURF_MACRO_TILE_ASPECT_1  0
-#define EVERGREEN_ADDR_SURF_MACRO_TILE_ASPECT_2  1
-#define EVERGREEN_ADDR_SURF_MACRO_TILE_ASPECT_4  2
-#define EVERGREEN_ADDR_SURF_MACRO_TILE_ASPECT_8  3
-
-#define EVERGREEN_GRPH_SWAP_CONTROL                     0x1a03
-#define EVERGREEN_GRPH_ENDIAN_SWAP(x)            (((x) & 0x3) << 0)
-#       define EVERGREEN_GRPH_ENDIAN_NONE               0
-#       define EVERGREEN_GRPH_ENDIAN_8IN16              1
-#       define EVERGREEN_GRPH_ENDIAN_8IN32              2
-#       define EVERGREEN_GRPH_ENDIAN_8IN64              3
-
-#define EVERGREEN_D3VGA_CONTROL                         0xf8
-#define EVERGREEN_D4VGA_CONTROL                         0xf9
-#define EVERGREEN_D5VGA_CONTROL                         0xfa
-#define EVERGREEN_D6VGA_CONTROL                         0xfb
-
-#define EVERGREEN_GRPH_SURFACE_ADDRESS_MASK      0xffffff00
-
-#define EVERGREEN_GRPH_LUT_10BIT_BYPASS_CONTROL         0x1a02
-#define EVERGREEN_LUT_10BIT_BYPASS_EN            (1 << 8)
-
-#define EVERGREEN_GRPH_PITCH                            0x1a06
-#define EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH     0x1a07
-#define EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS_HIGH   0x1a08
-#define EVERGREEN_GRPH_SURFACE_OFFSET_X                 0x1a09
-#define EVERGREEN_GRPH_SURFACE_OFFSET_Y                 0x1a0a
-#define EVERGREEN_GRPH_X_START                          0x1a0b
-#define EVERGREEN_GRPH_Y_START                          0x1a0c
-#define EVERGREEN_GRPH_X_END                            0x1a0d
-#define EVERGREEN_GRPH_Y_END                            0x1a0e
-#define EVERGREEN_GRPH_UPDATE                           0x1a11
-#define EVERGREEN_GRPH_SURFACE_UPDATE_PENDING    (1 << 2)
-#define EVERGREEN_GRPH_UPDATE_LOCK               (1 << 16)
-#define EVERGREEN_GRPH_FLIP_CONTROL                     0x1a12
-#define EVERGREEN_GRPH_SURFACE_UPDATE_H_RETRACE_EN (1 << 0)
-
-#define EVERGREEN_VIEWPORT_START                        0x1b5c
-#define EVERGREEN_VIEWPORT_SIZE                         0x1b5d
-#define EVERGREEN_DESKTOP_HEIGHT                        0x1ac1
-
-/* CUR blocks at 0x6998, 0x7598, 0x10198, 0x10d98, 0x11998, 0x12598 */
-#define EVERGREEN_CUR_CONTROL                           0x1a66
-#       define EVERGREEN_CURSOR_EN                      (1 << 0)
-#       define EVERGREEN_CURSOR_MODE(x)                 (((x) & 0x3) << 8)
-#       define EVERGREEN_CURSOR_MONO                    0
-#       define EVERGREEN_CURSOR_24_1                    1
-#       define EVERGREEN_CURSOR_24_8_PRE_MULT           2
-#       define EVERGREEN_CURSOR_24_8_UNPRE_MULT         3
-#       define EVERGREEN_CURSOR_2X_MAGNIFY              (1 << 16)
-#       define EVERGREEN_CURSOR_FORCE_MC_ON             (1 << 20)
-#       define EVERGREEN_CURSOR_URGENT_CONTROL(x)       (((x) & 0x7) << 24)
-#       define EVERGREEN_CURSOR_URGENT_ALWAYS           0
-#       define EVERGREEN_CURSOR_URGENT_1_8              1
-#       define EVERGREEN_CURSOR_URGENT_1_4              2
-#       define EVERGREEN_CURSOR_URGENT_3_8              3
-#       define EVERGREEN_CURSOR_URGENT_1_2              4
-#define EVERGREEN_CUR_SURFACE_ADDRESS                   0x1a67
-#       define EVERGREEN_CUR_SURFACE_ADDRESS_MASK       0xfffff000
-#define EVERGREEN_CUR_SIZE                              0x1a68
-#define EVERGREEN_CUR_SURFACE_ADDRESS_HIGH              0x1a69
-#define EVERGREEN_CUR_POSITION                          0x1a6a
-#define EVERGREEN_CUR_HOT_SPOT                          0x1a6b
-#define EVERGREEN_CUR_COLOR1                            0x1a6c
-#define EVERGREEN_CUR_COLOR2                            0x1a6d
-#define EVERGREEN_CUR_UPDATE                            0x1a6e
-#       define EVERGREEN_CURSOR_UPDATE_PENDING          (1 << 0)
-#       define EVERGREEN_CURSOR_UPDATE_TAKEN            (1 << 1)
-#       define EVERGREEN_CURSOR_UPDATE_LOCK             (1 << 16)
-#       define EVERGREEN_CURSOR_DISABLE_MULTIPLE_UPDATE (1 << 24)
-
-
-#define NI_INPUT_CSC_CONTROL                           0x1a35
-#       define NI_INPUT_CSC_GRPH_MODE(x)               (((x) & 0x3) << 0)
-#       define NI_INPUT_CSC_BYPASS                     0
-#       define NI_INPUT_CSC_PROG_COEFF                 1
-#       define NI_INPUT_CSC_PROG_SHARED_MATRIXA        2
-#       define NI_INPUT_CSC_OVL_MODE(x)                (((x) & 0x3) << 4)
-
-#define NI_OUTPUT_CSC_CONTROL                          0x1a3c
-#       define NI_OUTPUT_CSC_GRPH_MODE(x)              (((x) & 0x7) << 0)
-#       define NI_OUTPUT_CSC_BYPASS                    0
-#       define NI_OUTPUT_CSC_TV_RGB                    1
-#       define NI_OUTPUT_CSC_YCBCR_601                 2
-#       define NI_OUTPUT_CSC_YCBCR_709                 3
-#       define NI_OUTPUT_CSC_PROG_COEFF                4
-#       define NI_OUTPUT_CSC_PROG_SHARED_MATRIXB       5
-#       define NI_OUTPUT_CSC_OVL_MODE(x)               (((x) & 0x7) << 4)
-
-#define NI_DEGAMMA_CONTROL                             0x1a58
-#       define NI_GRPH_DEGAMMA_MODE(x)                 (((x) & 0x3) << 0)
-#       define NI_DEGAMMA_BYPASS                       0
-#       define NI_DEGAMMA_SRGB_24                      1
-#       define NI_DEGAMMA_XVYCC_222                    2
-#       define NI_OVL_DEGAMMA_MODE(x)                  (((x) & 0x3) << 4)
-#       define NI_ICON_DEGAMMA_MODE(x)                 (((x) & 0x3) << 8)
-#       define NI_CURSOR_DEGAMMA_MODE(x)               (((x) & 0x3) << 12)
-
-#define NI_GAMUT_REMAP_CONTROL                         0x1a59
-#       define NI_GRPH_GAMUT_REMAP_MODE(x)             (((x) & 0x3) << 0)
-#       define NI_GAMUT_REMAP_BYPASS                   0
-#       define NI_GAMUT_REMAP_PROG_COEFF               1
-#       define NI_GAMUT_REMAP_PROG_SHARED_MATRIXA      2
-#       define NI_GAMUT_REMAP_PROG_SHARED_MATRIXB      3
-#       define NI_OVL_GAMUT_REMAP_MODE(x)              (((x) & 0x3) << 4)
-
-#define NI_REGAMMA_CONTROL                             0x1aa0
-#       define NI_GRPH_REGAMMA_MODE(x)                 (((x) & 0x7) << 0)
-#       define NI_REGAMMA_BYPASS                       0
-#       define NI_REGAMMA_SRGB_24                      1
-#       define NI_REGAMMA_XVYCC_222                    2
-#       define NI_REGAMMA_PROG_A                       3
-#       define NI_REGAMMA_PROG_B                       4
-#       define NI_OVL_REGAMMA_MODE(x)                  (((x) & 0x7) << 4)
-
-
-#define NI_PRESCALE_GRPH_CONTROL                       0x1a2d
-#       define NI_GRPH_PRESCALE_BYPASS                 (1 << 4)
-
-#define NI_PRESCALE_OVL_CONTROL                        0x1a31
-#       define NI_OVL_PRESCALE_BYPASS                  (1 << 4)
-
-#define NI_INPUT_GAMMA_CONTROL                         0x1a10
-#       define NI_GRPH_INPUT_GAMMA_MODE(x)             (((x) & 0x3) << 0)
-#       define NI_INPUT_GAMMA_USE_LUT                  0
-#       define NI_INPUT_GAMMA_BYPASS                   1
-#       define NI_INPUT_GAMMA_SRGB_24                  2
-#       define NI_INPUT_GAMMA_XVYCC_222                3
-#       define NI_OVL_INPUT_GAMMA_MODE(x)              (((x) & 0x3) << 4)
-
-#define IH_RB_WPTR__RB_OVERFLOW_MASK	0x1
-#define IH_RB_CNTL__WPTR_OVERFLOW_CLEAR_MASK 0x80000000
-#define SRBM_STATUS__IH_BUSY_MASK	0x20000
-#define SRBM_SOFT_RESET__SOFT_RESET_IH_MASK	0x400
-
-#define	BLACKOUT_MODE_MASK			0x00000007
-#define	VGA_RENDER_CONTROL			0xC0
-#define R_000300_VGA_RENDER_CONTROL             0xC0
-#define C_000300_VGA_VSTATUS_CNTL               0xFFFCFFFF
-#define EVERGREEN_CRTC_STATUS                   0x1BA3
-#define EVERGREEN_CRTC_V_BLANK                  (1 << 0)
-#define EVERGREEN_CRTC_STATUS_POSITION          0x1BA4
-/* CRTC blocks at 0x6df0, 0x79f0, 0x105f0, 0x111f0, 0x11df0, 0x129f0 */
-#define EVERGREEN_CRTC_V_BLANK_START_END                0x1b8d
-#define EVERGREEN_CRTC_CONTROL                          0x1b9c
-#       define EVERGREEN_CRTC_MASTER_EN                 (1 << 0)
-#       define EVERGREEN_CRTC_DISP_READ_REQUEST_DISABLE (1 << 24)
-#define EVERGREEN_CRTC_BLANK_CONTROL                    0x1b9d
-#       define EVERGREEN_CRTC_BLANK_DATA_EN             (1 << 8)
-#       define EVERGREEN_CRTC_V_BLANK                   (1 << 0)
-#define EVERGREEN_CRTC_STATUS_HV_COUNT                  0x1ba8
-#define EVERGREEN_CRTC_UPDATE_LOCK                      0x1bb5
-#define EVERGREEN_MASTER_UPDATE_LOCK                    0x1bbd
-#define EVERGREEN_MASTER_UPDATE_MODE                    0x1bbe
-#define EVERGREEN_GRPH_UPDATE_LOCK               (1 << 16)
-#define EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH     0x1a07
-#define EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS_HIGH   0x1a08
-#define EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS          0x1a04
-#define EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS        0x1a05
-#define EVERGREEN_GRPH_UPDATE                           0x1a11
-#define EVERGREEN_VGA_MEMORY_BASE_ADDRESS               0xc4
-#define EVERGREEN_VGA_MEMORY_BASE_ADDRESS_HIGH          0xc9
-#define EVERGREEN_GRPH_SURFACE_UPDATE_PENDING    (1 << 2)
-
-#define mmVM_CONTEXT1_CNTL__xxRANGE_PROTECTION_FAULT_ENABLE_DEFAULT_MASK 0x10
-#define mmVM_CONTEXT1_CNTL__xxRANGE_PROTECTION_FAULT_ENABLE_DEFAULT__SHIFT 0x4
-#define mmVM_CONTEXT1_CNTL__xxDUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT_MASK 0x80
-#define mmVM_CONTEXT1_CNTL__xxDUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT__SHIFT 0x7
-#define mmVM_CONTEXT1_CNTL__xxPDE0_PROTECTION_FAULT_ENABLE_DEFAULT_MASK 0x400
-#define mmVM_CONTEXT1_CNTL__xxPDE0_PROTECTION_FAULT_ENABLE_DEFAULT__SHIFT 0xa
-#define mmVM_CONTEXT1_CNTL__xxVALID_PROTECTION_FAULT_ENABLE_DEFAULT_MASK 0x2000
-#define mmVM_CONTEXT1_CNTL__xxVALID_PROTECTION_FAULT_ENABLE_DEFAULT__SHIFT 0xd
-#define mmVM_CONTEXT1_CNTL__xxREAD_PROTECTION_FAULT_ENABLE_DEFAULT_MASK 0x10000
-#define mmVM_CONTEXT1_CNTL__xxREAD_PROTECTION_FAULT_ENABLE_DEFAULT__SHIFT 0x10
-#define mmVM_CONTEXT1_CNTL__xxWRITE_PROTECTION_FAULT_ENABLE_DEFAULT_MASK 0x80000
-#define mmVM_CONTEXT1_CNTL__xxWRITE_PROTECTION_FAULT_ENABLE_DEFAULT__SHIFT 0x13
-
-#define mmVM_CONTEXT1_PROTECTION_FAULT_STATUS__xxVMID_MASK 0x1e000000
-#define mmVM_CONTEXT1_PROTECTION_FAULT_STATUS__xxVMID__SHIFT 0x19
-#define mmVM_CONTEXT1_PROTECTION_FAULT_STATUS__xxPROTECTIONS_MASK 0xff
-#define mmVM_CONTEXT1_PROTECTION_FAULT_STATUS__xxPROTECTIONS__SHIFT 0x0
-#define mmVM_CONTEXT1_PROTECTION_FAULT_STATUS__xxMEMORY_CLIENT_ID_MASK 0xff000
-#define mmVM_CONTEXT1_PROTECTION_FAULT_STATUS__xxMEMORY_CLIENT_ID__SHIFT 0xc
-#define mmVM_CONTEXT1_PROTECTION_FAULT_STATUS__xxMEMORY_CLIENT_RW_MASK 0x1000000
-#define mmVM_CONTEXT1_PROTECTION_FAULT_STATUS__xxMEMORY_CLIENT_RW__SHIFT 0x18
-
-#define mmMC_SHARED_BLACKOUT_CNTL__xxBLACKOUT_MODE_MASK 0x7
-#define mmMC_SHARED_BLACKOUT_CNTL__xxBLACKOUT_MODE__SHIFT 0x0
-
-#define mmBIF_FB_EN__xxFB_READ_EN_MASK 0x1
-#define mmBIF_FB_EN__xxFB_READ_EN__SHIFT 0x0
-#define mmBIF_FB_EN__xxFB_WRITE_EN_MASK 0x2
-#define mmBIF_FB_EN__xxFB_WRITE_EN__SHIFT 0x1
-
-#define mmSRBM_SOFT_RESET__xxSOFT_RESET_VMC_MASK 0x20000
-#define mmSRBM_SOFT_RESET__xxSOFT_RESET_VMC__SHIFT 0x11
-#define mmSRBM_SOFT_RESET__xxSOFT_RESET_MC_MASK 0x800
-#define mmSRBM_SOFT_RESET__xxSOFT_RESET_MC__SHIFT 0xb
-
-#define VM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK 0x8
-#define VM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT__SHIFT 0x3
-#define VM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK 0x40
-#define VM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT__SHIFT 0x6
-#define VM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK 0x200
-#define VM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT__SHIFT 0x9
-#define VM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK 0x1000
-#define VM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT__SHIFT 0xc
-#define VM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK 0x8000
-#define VM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT__SHIFT 0xf
-#define VM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK 0x40000
-#define VM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT__SHIFT 0x12
+#define GRPH_FORMAT_ARGB8888           0
+#define GRPH_FORMAT_ARGB2101010        1
+#define GRPH_FORMAT_32BPP_DIG          2
+#define GRPH_FORMAT_8B_ARGB2101010     3
+#define GRPH_FORMAT_BGRA1010102        4
+#define GRPH_FORMAT_8B_BGRA1010102     5
+#define GRPH_FORMAT_RGB111110          6
+#define GRPH_FORMAT_BGR101111          7
+
+#define GRPH_ENDIAN_NONE               0
+#define GRPH_ENDIAN_8IN16              1
+#define GRPH_ENDIAN_8IN32              2
+#define GRPH_ENDIAN_8IN64              3
+#define GRPH_RED_SEL_R                 0
+#define GRPH_RED_SEL_G                 1
+#define GRPH_RED_SEL_B                 2
+#define GRPH_RED_SEL_A                 3
+
+#define GRPH_GREEN_SEL_G               0
+#define GRPH_GREEN_SEL_B               1
+#define GRPH_GREEN_SEL_A               2
+#define GRPH_GREEN_SEL_R               3
+
+#define GRPH_BLUE_SEL_B                0
+#define GRPH_BLUE_SEL_A                1
+#define GRPH_BLUE_SEL_R                2
+#define GRPH_BLUE_SEL_G                3
+
+#define GRPH_ALPHA_SEL_A               0
+#define GRPH_ALPHA_SEL_R               1
+#define GRPH_ALPHA_SEL_G               2
+#define GRPH_ALPHA_SEL_B               3
+
+/* CUR_CONTROL */
+	#define CURSOR_MONO                    0
+	#define CURSOR_24_1                    1
+	#define CURSOR_24_8_PRE_MULT           2
+	#define CURSOR_24_8_UNPRE_MULT         3
+	#define CURSOR_URGENT_ALWAYS           0
+	#define CURSOR_URGENT_1_8              1
+	#define CURSOR_URGENT_1_4              2
+	#define CURSOR_URGENT_3_8              3
+	#define CURSOR_URGENT_1_2              4
+
+/* INPUT_CSC_CONTROL */
+#       define INPUT_CSC_BYPASS                     0
+#       define INPUT_CSC_PROG_COEFF                 1
+#       define INPUT_CSC_PROG_SHARED_MATRIXA        2
+
+/* OUTPUT_CSC_CONTROL */
+#       define OUTPUT_CSC_BYPASS                    0
+#       define OUTPUT_CSC_TV_RGB                    1
+#       define OUTPUT_CSC_YCBCR_601                 2
+#       define OUTPUT_CSC_YCBCR_709                 3
+#       define OUTPUT_CSC_PROG_COEFF                4
+#       define OUTPUT_CSC_PROG_SHARED_MATRIXB       5
+
+/* DEGAMMA_CONTROL */
+#       define DEGAMMA_BYPASS                       0
+#       define DEGAMMA_SRGB_24                      1
+#       define DEGAMMA_XVYCC_222                    2
+
+/* GAMUT_REMAP_CONTROL */
+#       define GAMUT_REMAP_BYPASS                   0
+#       define GAMUT_REMAP_PROG_COEFF               1
+#       define GAMUT_REMAP_PROG_SHARED_MATRIXA      2
+#       define GAMUT_REMAP_PROG_SHARED_MATRIXB      3
+
+/* REGAMMA_CONTROL */
+#       define REGAMMA_BYPASS                       0
+#       define REGAMMA_SRGB_24                      1
+#       define REGAMMA_XVYCC_222                    2
+#       define REGAMMA_PROG_A                       3
+#       define REGAMMA_PROG_B                       4
+
+
+/* INPUT_GAMMA_CONTROL */
+#       define INPUT_GAMMA_USE_LUT                  0
+#       define INPUT_GAMMA_BYPASS                   1
+#       define INPUT_GAMMA_SRGB_24                  2
+#       define INPUT_GAMMA_XVYCC_222                3
 
 #define MC_SEQ_MISC0__MT__MASK	0xf0000000
 #define MC_SEQ_MISC0__MT__GDDR1  0x10000000
@@ -2411,44 +780,16 @@
 #define MC_SEQ_MISC0__MT__HBM    0x60000000
 #define MC_SEQ_MISC0__MT__DDR3   0xB0000000
 
-#define SRBM_STATUS__MCB_BUSY_MASK 0x200
-#define SRBM_STATUS__MCB_BUSY__SHIFT 0x9
-#define SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK 0x400
-#define SRBM_STATUS__MCB_NON_DISPLAY_BUSY__SHIFT 0xa
-#define SRBM_STATUS__MCC_BUSY_MASK 0x800
-#define SRBM_STATUS__MCC_BUSY__SHIFT 0xb
-#define SRBM_STATUS__MCD_BUSY_MASK 0x1000
-#define SRBM_STATUS__MCD_BUSY__SHIFT 0xc
-#define SRBM_STATUS__VMC_BUSY_MASK 0x100
-#define SRBM_STATUS__VMC_BUSY__SHIFT 0x8
-
-
-#define GRBM_STATUS__GUI_ACTIVE_MASK 0x80000000
 #define CP_INT_CNTL_RING__TIME_STAMP_INT_ENABLE_MASK 0x4000000
-#define CP_INT_CNTL_RING0__PRIV_REG_INT_ENABLE_MASK 0x800000
-#define CP_INT_CNTL_RING0__PRIV_INSTR_INT_ENABLE_MASK 0x400000
 #define PACKET3_SEM_WAIT_ON_SIGNAL    (0x1 << 12)
 #define PACKET3_SEM_SEL_SIGNAL	    (0x6 << 29)
 #define PACKET3_SEM_SEL_WAIT	    (0x7 << 29)
 
-#define CONFIG_CNTL	0x1509
-#define CC_DRM_ID_STRAPS	0X1559
 #define AMDGPU_PCIE_INDEX	0xc
 #define AMDGPU_PCIE_DATA	0xd
 
-#define DMA_SEM_INCOMPLETE_TIMER_CNTL                     0x3411
-#define DMA_SEM_WAIT_FAIL_TIMER_CNTL                      0x3412
-#define DMA_MODE                                          0x342f
-#define DMA_RB_RPTR_ADDR_HI                               0x3407
-#define DMA_RB_RPTR_ADDR_LO                               0x3408
-#define DMA_BUSY_MASK 0x20
-#define DMA1_BUSY_MASK 0X40
-#define SDMA_MAX_INSTANCE 2
-
 #define PCIE_BUS_CLK    10000
 #define TCLK            (PCIE_BUS_CLK / 10)
-#define CC_DRM_ID_STRAPS__ATI_REV_ID_MASK		0xf0000000
-#define CC_DRM_ID_STRAPS__ATI_REV_ID__SHIFT 0x1c
 #define	PCIE_PORT_INDEX					0xe
 #define	PCIE_PORT_DATA					0xf
 #define EVERGREEN_PIF_PHY0_INDEX                        0x8
@@ -2456,6 +797,36 @@
 #define EVERGREEN_PIF_PHY1_INDEX                        0x10
 #define EVERGREEN_PIF_PHY1_DATA				0x14
 
-#define	MC_VM_FB_OFFSET					0x81a
+/* Discrete VCE clocks */
+#define CG_VCEPLL_FUNC_CNTL                             0xc0030600
+#define    VCEPLL_RESET_MASK                            0x00000001
+#define    VCEPLL_SLEEP_MASK                            0x00000002
+#define    VCEPLL_BYPASS_EN_MASK                        0x00000004
+#define    VCEPLL_CTLREQ_MASK                           0x00000008
+#define    VCEPLL_VCO_MODE_MASK                         0x00000600
+#define    VCEPLL_REF_DIV_MASK                          0x003F0000
+#define    VCEPLL_CTLACK_MASK                           0x40000000
+#define    VCEPLL_CTLACK2_MASK                          0x80000000
+
+#define CG_VCEPLL_FUNC_CNTL_2                           0xc0030601
+#define    VCEPLL_PDIV_A(x)                             ((x) << 0)
+#define    VCEPLL_PDIV_A_MASK                           0x0000007F
+#define    VCEPLL_PDIV_B(x)                             ((x) << 8)
+#define    VCEPLL_PDIV_B_MASK                           0x00007F00
+#define    EVCLK_SRC_SEL(x)                             ((x) << 20)
+#define    EVCLK_SRC_SEL_MASK                           0x01F00000
+#define    ECCLK_SRC_SEL(x)                             ((x) << 25)
+#define    ECCLK_SRC_SEL_MASK                           0x3E000000
+
+#define CG_VCEPLL_FUNC_CNTL_3                           0xc0030602
+#define    VCEPLL_FB_DIV(x)                             ((x) << 0)
+#define    VCEPLL_FB_DIV_MASK                           0x01FFFFFF
+
+#define CG_VCEPLL_FUNC_CNTL_4                           0xc0030603
+
+#define CG_VCEPLL_FUNC_CNTL_5                           0xc0030604
+#define CG_VCEPLL_SPREAD_SPECTRUM                       0xc0030606
+#define    VCEPLL_SSEN_MASK                             0x00000001
+
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/sienna_cichlid.c b/drivers/gpu/drm/amd/amdgpu/sienna_cichlid.c
new file mode 100644
index 000000000000..2594467bdd87
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/sienna_cichlid.c
@@ -0,0 +1,298 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "sienna_cichlid.h"
+#include "amdgpu_reset.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_dpm.h"
+#include "amdgpu_job.h"
+#include "amdgpu_ring.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_psp.h"
+#include "amdgpu_xgmi.h"
+
+static bool sienna_cichlid_is_mode2_default(struct amdgpu_reset_control *reset_ctl)
+{
+#if 0
+	struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle;
+
+	if (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(11, 0, 7) &&
+	    adev->pm.fw_version >= 0x3a5500 && !amdgpu_sriov_vf(adev))
+		return true;
+#endif
+	return amdgpu_reset_method == AMD_RESET_METHOD_MODE2;
+}
+
+static struct amdgpu_reset_handler *
+sienna_cichlid_get_reset_handler(struct amdgpu_reset_control *reset_ctl,
+			    struct amdgpu_reset_context *reset_context)
+{
+	struct amdgpu_reset_handler *handler;
+	int i;
+
+	if (reset_context->method != AMD_RESET_METHOD_NONE) {
+		for_each_handler(i, handler, reset_ctl)	{
+			if (handler->reset_method == reset_context->method)
+				return handler;
+		}
+	}
+
+	if (sienna_cichlid_is_mode2_default(reset_ctl)) {
+		for_each_handler(i, handler, reset_ctl)	{
+			if (handler->reset_method == AMD_RESET_METHOD_MODE2)
+				return handler;
+		}
+	}
+
+	return NULL;
+}
+
+static int sienna_cichlid_mode2_suspend_ip(struct amdgpu_device *adev)
+{
+	int r, i;
+
+	amdgpu_device_set_pg_state(adev, AMD_PG_STATE_UNGATE);
+	amdgpu_device_set_cg_state(adev, AMD_CG_STATE_UNGATE);
+
+	for (i = adev->num_ip_blocks - 1; i >= 0; i--) {
+		if (!(adev->ip_blocks[i].version->type ==
+			      AMD_IP_BLOCK_TYPE_GFX ||
+		      adev->ip_blocks[i].version->type ==
+			      AMD_IP_BLOCK_TYPE_SDMA))
+			continue;
+
+		r = amdgpu_ip_block_suspend(&adev->ip_blocks[i]);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+static int
+sienna_cichlid_mode2_prepare_hwcontext(struct amdgpu_reset_control *reset_ctl,
+				  struct amdgpu_reset_context *reset_context)
+{
+	int r = 0;
+	struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle;
+
+	if (!amdgpu_sriov_vf(adev)) {
+		if (adev->gfxhub.funcs->mode2_save_regs)
+			adev->gfxhub.funcs->mode2_save_regs(adev);
+		if (adev->gfxhub.funcs->halt)
+			adev->gfxhub.funcs->halt(adev);
+		r = sienna_cichlid_mode2_suspend_ip(adev);
+	}
+
+	return r;
+}
+
+static void sienna_cichlid_async_reset(struct work_struct *work)
+{
+	struct amdgpu_reset_handler *handler;
+	struct amdgpu_reset_control *reset_ctl =
+		container_of(work, struct amdgpu_reset_control, reset_work);
+	struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle;
+	int i;
+
+	for_each_handler(i, handler, reset_ctl)	{
+		if (handler->reset_method == reset_ctl->active_reset) {
+			dev_dbg(adev->dev, "Resetting device\n");
+			handler->do_reset(adev);
+			break;
+		}
+	}
+}
+
+static int sienna_cichlid_mode2_reset(struct amdgpu_device *adev)
+{
+	/* disable BM */
+	pci_clear_master(adev->pdev);
+	return amdgpu_dpm_mode2_reset(adev);
+}
+
+static int
+sienna_cichlid_mode2_perform_reset(struct amdgpu_reset_control *reset_ctl,
+			      struct amdgpu_reset_context *reset_context)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle;
+	int r;
+
+	r = sienna_cichlid_mode2_reset(adev);
+	if (r) {
+		dev_err(adev->dev,
+			"ASIC reset failed with error, %d ", r);
+	}
+	return r;
+}
+
+static int sienna_cichlid_mode2_restore_ip(struct amdgpu_device *adev)
+{
+	int i, r;
+	struct psp_context *psp = &adev->psp;
+
+	r = psp_rlc_autoload_start(psp);
+	if (r) {
+		dev_err(adev->dev, "Failed to start rlc autoload\n");
+		return r;
+	}
+
+	/* Reinit GFXHUB */
+	if (adev->gfxhub.funcs->mode2_restore_regs)
+		adev->gfxhub.funcs->mode2_restore_regs(adev);
+	adev->gfxhub.funcs->init(adev);
+	r = adev->gfxhub.funcs->gart_enable(adev);
+	if (r) {
+		dev_err(adev->dev, "GFXHUB gart reenable failed after reset\n");
+		return r;
+	}
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_IH) {
+			r = amdgpu_ip_block_resume(&adev->ip_blocks[i]);
+			if (r)
+				return r;
+		}
+	}
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!(adev->ip_blocks[i].version->type ==
+			      AMD_IP_BLOCK_TYPE_GFX ||
+		      adev->ip_blocks[i].version->type ==
+			      AMD_IP_BLOCK_TYPE_SDMA))
+			continue;
+		r = amdgpu_ip_block_resume(&adev->ip_blocks[i]);
+		if (r)
+			return r;
+	}
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!(adev->ip_blocks[i].version->type ==
+			      AMD_IP_BLOCK_TYPE_GFX ||
+		      adev->ip_blocks[i].version->type ==
+			      AMD_IP_BLOCK_TYPE_SDMA))
+			continue;
+
+		if (adev->ip_blocks[i].version->funcs->late_init) {
+			r = adev->ip_blocks[i].version->funcs->late_init(
+				&adev->ip_blocks[i]);
+			if (r) {
+				dev_err(adev->dev,
+					"late_init of IP block <%s> failed %d after reset\n",
+					adev->ip_blocks[i].version->funcs->name,
+					r);
+				return r;
+			}
+		}
+		adev->ip_blocks[i].status.late_initialized = true;
+	}
+
+	amdgpu_device_set_cg_state(adev, AMD_CG_STATE_GATE);
+	amdgpu_device_set_pg_state(adev, AMD_PG_STATE_GATE);
+
+	return r;
+}
+
+static int
+sienna_cichlid_mode2_restore_hwcontext(struct amdgpu_reset_control *reset_ctl,
+				  struct amdgpu_reset_context *reset_context)
+{
+	int r;
+	struct amdgpu_device *tmp_adev = (struct amdgpu_device *)reset_ctl->handle;
+
+	amdgpu_set_init_level(tmp_adev, AMDGPU_INIT_LEVEL_RESET_RECOVERY);
+	dev_info(tmp_adev->dev,
+			"GPU reset succeeded, trying to resume\n");
+	r = sienna_cichlid_mode2_restore_ip(tmp_adev);
+	if (r)
+		goto end;
+
+	/*
+	* Add this ASIC as tracked as reset was already
+	* complete successfully.
+	*/
+	amdgpu_register_gpu_instance(tmp_adev);
+
+	/* Resume RAS */
+	amdgpu_ras_resume(tmp_adev);
+
+	amdgpu_irq_gpu_reset_resume_helper(tmp_adev);
+
+	amdgpu_set_init_level(tmp_adev, AMDGPU_INIT_LEVEL_DEFAULT);
+	r = amdgpu_ib_ring_tests(tmp_adev);
+	if (r) {
+		dev_err(tmp_adev->dev,
+			"ib ring test failed (%d).\n", r);
+		r = -EAGAIN;
+		goto end;
+	}
+
+end:
+	if (r)
+		return -EAGAIN;
+	else
+		return r;
+}
+
+static struct amdgpu_reset_handler sienna_cichlid_mode2_handler = {
+	.reset_method		= AMD_RESET_METHOD_MODE2,
+	.prepare_env		= NULL,
+	.prepare_hwcontext	= sienna_cichlid_mode2_prepare_hwcontext,
+	.perform_reset		= sienna_cichlid_mode2_perform_reset,
+	.restore_hwcontext	= sienna_cichlid_mode2_restore_hwcontext,
+	.restore_env		= NULL,
+	.do_reset		= sienna_cichlid_mode2_reset,
+};
+
+static struct amdgpu_reset_handler
+	*sienna_cichlid_rst_handlers[AMDGPU_RESET_MAX_HANDLERS] = {
+		&sienna_cichlid_mode2_handler,
+	};
+
+int sienna_cichlid_reset_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_reset_control *reset_ctl;
+
+	reset_ctl = kzalloc(sizeof(*reset_ctl), GFP_KERNEL);
+	if (!reset_ctl)
+		return -ENOMEM;
+
+	reset_ctl->handle = adev;
+	reset_ctl->async_reset = sienna_cichlid_async_reset;
+	reset_ctl->active_reset = AMD_RESET_METHOD_NONE;
+	reset_ctl->get_reset_handler = sienna_cichlid_get_reset_handler;
+
+	INIT_WORK(&reset_ctl->reset_work, reset_ctl->async_reset);
+	/* Only mode2 is handled through reset control now */
+	reset_ctl->reset_handlers = &sienna_cichlid_rst_handlers;
+	adev->reset_cntl = reset_ctl;
+
+	return 0;
+}
+
+int sienna_cichlid_reset_fini(struct amdgpu_device *adev)
+{
+	kfree(adev->reset_cntl);
+	adev->reset_cntl = NULL;
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/sienna_cichlid.h b/drivers/gpu/drm/amd/amdgpu/sienna_cichlid.h
new file mode 100644
index 000000000000..5213b162dacd
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/sienna_cichlid.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __SIENNA_CICHLID_H__
+#define __SIENNA_CICHLID_H__
+
+#include "amdgpu.h"
+
+int sienna_cichlid_reset_init(struct amdgpu_device *adev);
+int sienna_cichlid_reset_fini(struct amdgpu_device *adev);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/smu_v11_0_i2c.c b/drivers/gpu/drm/amd/amdgpu/smu_v11_0_i2c.c
new file mode 100644
index 000000000000..68aef47254a9
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/smu_v11_0_i2c.c
@@ -0,0 +1,794 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "smuio/smuio_11_0_0_offset.h"
+#include "smuio/smuio_11_0_0_sh_mask.h"
+
+#include "smu_v11_0_i2c.h"
+#include "amdgpu.h"
+#include "amdgpu_dpm.h"
+#include "soc15_common.h"
+#include <drm/drm_fixed.h>
+#include <drm/drm_drv.h>
+#include "amdgpu_amdkfd.h"
+#include <linux/i2c.h>
+#include <linux/pci.h>
+
+/* error codes */
+#define I2C_OK                0
+#define I2C_NAK_7B_ADDR_NOACK 1
+#define I2C_NAK_TXDATA_NOACK  2
+#define I2C_TIMEOUT           4
+#define I2C_SW_TIMEOUT        8
+#define I2C_ABORT             0x10
+
+#define I2C_X_RESTART         BIT(31)
+
+static void smu_v11_0_i2c_set_clock_gating(struct i2c_adapter *control, bool en)
+{
+	struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control);
+	struct amdgpu_device *adev = smu_i2c->adev;
+	uint32_t reg = RREG32_SOC15(SMUIO, 0, mmSMUIO_PWRMGT);
+
+	reg = REG_SET_FIELD(reg, SMUIO_PWRMGT, i2c_clk_gate_en, en ? 1 : 0);
+	WREG32_SOC15(SMUIO, 0, mmSMUIO_PWRMGT, reg);
+}
+
+/* The T_I2C_POLL_US is defined as follows:
+ *
+ * "Define a timer interval (t_i2c_poll) equal to 10 times the
+ *  signalling period for the highest I2C transfer speed used in the
+ *  system and supported by DW_apb_i2c. For instance, if the highest
+ *  I2C data transfer mode is 400 kb/s, then t_i2c_poll is 25 us."  --
+ * DesignWare DW_apb_i2c Databook, Version 1.21a, section 3.8.3.1,
+ * page 56, with grammar and syntax corrections.
+ *
+ * Vcc for our device is at 1.8V which puts it at 400 kHz,
+ * see Atmel AT24CM02 datasheet, section 8.3 DC Characteristics table, page 14.
+ *
+ * The procedure to disable the IP block is described in section
+ * 3.8.3 Disabling DW_apb_i2c on page 56.
+ */
+#define I2C_SPEED_MODE_FAST     2
+#define T_I2C_POLL_US           25
+#define I2C_MAX_T_POLL_COUNT    1000
+
+static int smu_v11_0_i2c_enable(struct i2c_adapter *control, bool enable)
+{
+	struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control);
+	struct amdgpu_device *adev = smu_i2c->adev;
+
+	WREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_ENABLE, enable ? 1 : 0);
+
+	if (!enable) {
+		int ii;
+
+		for (ii = I2C_MAX_T_POLL_COUNT; ii > 0; ii--) {
+			u32 en_stat = RREG32_SOC15(SMUIO,
+						   0,
+						   mmCKSVII2C_IC_ENABLE_STATUS);
+			if (REG_GET_FIELD(en_stat, CKSVII2C_IC_ENABLE_STATUS, IC_EN))
+				udelay(T_I2C_POLL_US);
+			else
+				return I2C_OK;
+		}
+
+		return I2C_ABORT;
+	}
+
+	return I2C_OK;
+}
+
+static void smu_v11_0_i2c_clear_status(struct i2c_adapter *control)
+{
+	struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control);
+	struct amdgpu_device *adev = smu_i2c->adev;
+	/* do */
+	{
+		RREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_CLR_INTR);
+
+	} /* while (reg_CKSVII2C_ic_clr_intr == 0) */
+}
+
+static void smu_v11_0_i2c_configure(struct i2c_adapter *control)
+{
+	struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control);
+	struct amdgpu_device *adev = smu_i2c->adev;
+	uint32_t reg = 0;
+
+	reg = REG_SET_FIELD(reg, CKSVII2C_IC_CON, IC_SLAVE_DISABLE, 1);
+	reg = REG_SET_FIELD(reg, CKSVII2C_IC_CON, IC_RESTART_EN, 1);
+	reg = REG_SET_FIELD(reg, CKSVII2C_IC_CON, IC_10BITADDR_MASTER, 0);
+	reg = REG_SET_FIELD(reg, CKSVII2C_IC_CON, IC_10BITADDR_SLAVE, 0);
+	/* The values of IC_MAX_SPEED_MODE are,
+	 * 1: standard mode, 0 - 100 Kb/s,
+	 * 2: fast mode, <= 400 Kb/s, or fast mode plus, <= 1000 Kb/s,
+	 * 3: high speed mode, <= 3.4 Mb/s.
+	 */
+	reg = REG_SET_FIELD(reg, CKSVII2C_IC_CON, IC_MAX_SPEED_MODE,
+			    I2C_SPEED_MODE_FAST);
+	reg = REG_SET_FIELD(reg, CKSVII2C_IC_CON, IC_MASTER_MODE, 1);
+
+	WREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_CON, reg);
+}
+
+static void smu_v11_0_i2c_set_clock(struct i2c_adapter *control)
+{
+	struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control);
+	struct amdgpu_device *adev = smu_i2c->adev;
+
+	/*
+	 * Standard mode speed, These values are taken from SMUIO MAS,
+	 * but are different from what is given is
+	 * Synopsys spec. The values here are based on assumption
+	 * that refclock is 100MHz
+	 *
+	 * Configuration for standard mode; Speed = 100kbps
+	 * Scale linearly, for now only support standard speed clock
+	 * This will work only with 100M ref clock
+	 *
+	 * TBD:Change the calculation to take into account ref clock values also.
+	 */
+
+	WREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_FS_SPKLEN, 2);
+	WREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_SS_SCL_HCNT, 120);
+	WREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_SS_SCL_LCNT, 130);
+	WREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_SDA_HOLD, 20);
+}
+
+static void smu_v11_0_i2c_set_address(struct i2c_adapter *control, u16 address)
+{
+	struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control);
+	struct amdgpu_device *adev = smu_i2c->adev;
+
+	/* The IC_TAR::IC_TAR field is 10-bits wide.
+	 * It takes a 7-bit or 10-bit addresses as an address,
+	 * i.e. no read/write bit--no wire format, just the address.
+	 */
+	WREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_TAR, address & 0x3FF);
+}
+
+static uint32_t smu_v11_0_i2c_poll_tx_status(struct i2c_adapter *control)
+{
+	struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control);
+	struct amdgpu_device *adev = smu_i2c->adev;
+	uint32_t ret = I2C_OK;
+	uint32_t reg, reg_c_tx_abrt_source;
+
+	/*Check if transmission is completed */
+	unsigned long  timeout_counter = jiffies + msecs_to_jiffies(20);
+
+	do {
+		if (time_after(jiffies, timeout_counter)) {
+			ret |= I2C_SW_TIMEOUT;
+			break;
+		}
+
+		reg = RREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_STATUS);
+
+	} while (REG_GET_FIELD(reg, CKSVII2C_IC_STATUS, TFE) == 0);
+
+	if (ret != I2C_OK)
+		return ret;
+
+	/* This only checks if NAK is received and transaction got aborted */
+	reg = RREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_INTR_STAT);
+
+	if (REG_GET_FIELD(reg, CKSVII2C_IC_INTR_STAT, R_TX_ABRT) == 1) {
+		reg_c_tx_abrt_source = RREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_TX_ABRT_SOURCE);
+		DRM_INFO("TX was terminated, IC_TX_ABRT_SOURCE val is:%x", reg_c_tx_abrt_source);
+
+		/* Check for stop due to NACK */
+		if (REG_GET_FIELD(reg_c_tx_abrt_source,
+				  CKSVII2C_IC_TX_ABRT_SOURCE,
+				  ABRT_TXDATA_NOACK) == 1) {
+
+			ret |= I2C_NAK_TXDATA_NOACK;
+
+		} else if (REG_GET_FIELD(reg_c_tx_abrt_source,
+					 CKSVII2C_IC_TX_ABRT_SOURCE,
+					 ABRT_7B_ADDR_NOACK) == 1) {
+
+			ret |= I2C_NAK_7B_ADDR_NOACK;
+		} else {
+			ret |= I2C_ABORT;
+		}
+
+		smu_v11_0_i2c_clear_status(control);
+	}
+
+	return ret;
+}
+
+static uint32_t smu_v11_0_i2c_poll_rx_status(struct i2c_adapter *control)
+{
+	struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control);
+	struct amdgpu_device *adev = smu_i2c->adev;
+	uint32_t ret = I2C_OK;
+	uint32_t reg_ic_status, reg_c_tx_abrt_source;
+
+	reg_c_tx_abrt_source = RREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_TX_ABRT_SOURCE);
+
+	/* If slave is not present */
+	if (REG_GET_FIELD(reg_c_tx_abrt_source,
+			  CKSVII2C_IC_TX_ABRT_SOURCE,
+			  ABRT_7B_ADDR_NOACK) == 1) {
+		ret |= I2C_NAK_7B_ADDR_NOACK;
+
+		smu_v11_0_i2c_clear_status(control);
+	} else {  /* wait till some data is there in RXFIFO */
+		/* Poll for some byte in RXFIFO */
+		unsigned long  timeout_counter = jiffies + msecs_to_jiffies(20);
+
+		do {
+			if (time_after(jiffies, timeout_counter)) {
+				ret |= I2C_SW_TIMEOUT;
+				break;
+			}
+
+			reg_ic_status = RREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_STATUS);
+
+		} while (REG_GET_FIELD(reg_ic_status, CKSVII2C_IC_STATUS, RFNE) == 0);
+	}
+
+	return ret;
+}
+
+/**
+ * smu_v11_0_i2c_transmit - Send a block of data over the I2C bus to a slave device.
+ *
+ * @control: I2C adapter reference
+ * @address: The I2C address of the slave device.
+ * @data: The data to transmit over the bus.
+ * @numbytes: The amount of data to transmit.
+ * @i2c_flag: Flags for transmission
+ *
+ * Returns 0 on success or error.
+ */
+static uint32_t smu_v11_0_i2c_transmit(struct i2c_adapter *control,
+				       u16 address, u8 *data,
+				       u32 numbytes, u32 i2c_flag)
+{
+	struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control);
+	struct amdgpu_device *adev = smu_i2c->adev;
+	u32 bytes_sent, reg, ret = I2C_OK;
+	unsigned long  timeout_counter;
+
+	bytes_sent = 0;
+
+	DRM_DEBUG_DRIVER("I2C_Transmit(), address = %x, bytes = %d , data: ",
+			 address, numbytes);
+
+	if (drm_debug_enabled(DRM_UT_DRIVER)) {
+		print_hex_dump(KERN_INFO, "data: ", DUMP_PREFIX_NONE,
+			       16, 1, data, numbytes, false);
+	}
+
+	/* Set the I2C slave address */
+	smu_v11_0_i2c_set_address(control, address);
+	/* Enable I2C */
+	smu_v11_0_i2c_enable(control, true);
+
+	/* Clear status bits */
+	smu_v11_0_i2c_clear_status(control);
+
+	timeout_counter = jiffies + msecs_to_jiffies(20);
+
+	while (numbytes > 0) {
+		reg = RREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_STATUS);
+		if (!REG_GET_FIELD(reg, CKSVII2C_IC_STATUS, TFNF)) {
+			/*
+			 * We waited for too long for the transmission
+			 * FIFO to become not-full.  Exit the loop
+			 * with error.
+			 */
+			if (time_after(jiffies, timeout_counter)) {
+				ret |= I2C_SW_TIMEOUT;
+				goto Err;
+			}
+		} else {
+			reg = REG_SET_FIELD(reg, CKSVII2C_IC_DATA_CMD, DAT,
+					    data[bytes_sent]);
+
+			/* Final message, final byte, must generate a
+			 * STOP to release the bus, i.e. don't hold
+			 * SCL low.
+			 */
+			if (numbytes == 1 && i2c_flag & I2C_M_STOP)
+				reg = REG_SET_FIELD(reg,
+						    CKSVII2C_IC_DATA_CMD,
+						    STOP, 1);
+
+			if (bytes_sent == 0 && i2c_flag & I2C_X_RESTART)
+				reg = REG_SET_FIELD(reg,
+						    CKSVII2C_IC_DATA_CMD,
+						    RESTART, 1);
+
+			/* Write */
+			reg = REG_SET_FIELD(reg, CKSVII2C_IC_DATA_CMD, CMD, 0);
+			WREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_DATA_CMD, reg);
+
+			/* Record that the bytes were transmitted */
+			bytes_sent++;
+			numbytes--;
+		}
+	}
+
+	ret = smu_v11_0_i2c_poll_tx_status(control);
+Err:
+	/* Any error, no point in proceeding */
+	if (ret != I2C_OK) {
+		if (ret & I2C_SW_TIMEOUT)
+			DRM_ERROR("TIMEOUT ERROR !!!");
+
+		if (ret & I2C_NAK_7B_ADDR_NOACK)
+			DRM_ERROR("Received I2C_NAK_7B_ADDR_NOACK !!!");
+
+
+		if (ret & I2C_NAK_TXDATA_NOACK)
+			DRM_ERROR("Received I2C_NAK_TXDATA_NOACK !!!");
+	}
+
+	return ret;
+}
+
+
+/**
+ * smu_v11_0_i2c_receive - Receive a block of data over the I2C bus from a slave device.
+ *
+ * @control: I2C adapter reference
+ * @address: The I2C address of the slave device.
+ * @data: Placeholder to store received data.
+ * @numbytes: The amount of data to transmit.
+ * @i2c_flag: Flags for transmission
+ *
+ * Returns 0 on success or error.
+ */
+static uint32_t smu_v11_0_i2c_receive(struct i2c_adapter *control,
+				      u16 address, u8 *data,
+				      u32 numbytes, u32 i2c_flag)
+{
+	struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control);
+	struct amdgpu_device *adev = smu_i2c->adev;
+	uint32_t bytes_received, ret = I2C_OK;
+
+	bytes_received = 0;
+
+	/* Set the I2C slave address */
+	smu_v11_0_i2c_set_address(control, address);
+
+	/* Enable I2C */
+	smu_v11_0_i2c_enable(control, true);
+
+	while (numbytes > 0) {
+		uint32_t reg = 0;
+
+		smu_v11_0_i2c_clear_status(control);
+
+		/* Prepare transaction */
+		reg = REG_SET_FIELD(reg, CKSVII2C_IC_DATA_CMD, DAT, 0);
+		/* Read */
+		reg = REG_SET_FIELD(reg, CKSVII2C_IC_DATA_CMD, CMD, 1);
+
+		/* Final message, final byte, must generate a STOP
+		 * to release the bus, i.e. don't hold SCL low.
+		 */
+		if (numbytes == 1 && i2c_flag & I2C_M_STOP)
+			reg = REG_SET_FIELD(reg, CKSVII2C_IC_DATA_CMD,
+					    STOP, 1);
+
+		if (bytes_received == 0 && i2c_flag & I2C_X_RESTART)
+			reg = REG_SET_FIELD(reg, CKSVII2C_IC_DATA_CMD,
+					    RESTART, 1);
+
+		WREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_DATA_CMD, reg);
+
+		ret = smu_v11_0_i2c_poll_rx_status(control);
+
+		/* Any error, no point in proceeding */
+		if (ret != I2C_OK) {
+			if (ret & I2C_SW_TIMEOUT)
+				DRM_ERROR("TIMEOUT ERROR !!!");
+
+			if (ret & I2C_NAK_7B_ADDR_NOACK)
+				DRM_ERROR("Received I2C_NAK_7B_ADDR_NOACK !!!");
+
+			if (ret & I2C_NAK_TXDATA_NOACK)
+				DRM_ERROR("Received I2C_NAK_TXDATA_NOACK !!!");
+
+			break;
+		}
+
+		reg = RREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_DATA_CMD);
+		data[bytes_received] = REG_GET_FIELD(reg, CKSVII2C_IC_DATA_CMD, DAT);
+
+		/* Record that the bytes were received */
+		bytes_received++;
+		numbytes--;
+	}
+
+	DRM_DEBUG_DRIVER("I2C_Receive(), address = %x, bytes = %d, data :",
+		  (uint16_t)address, bytes_received);
+
+	if (drm_debug_enabled(DRM_UT_DRIVER)) {
+		print_hex_dump(KERN_INFO, "data: ", DUMP_PREFIX_NONE,
+			       16, 1, data, bytes_received, false);
+	}
+
+	return ret;
+}
+
+static void smu_v11_0_i2c_abort(struct i2c_adapter *control)
+{
+	struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control);
+	struct amdgpu_device *adev = smu_i2c->adev;
+	uint32_t reg = 0;
+
+	/* Enable I2C engine; */
+	reg = REG_SET_FIELD(reg, CKSVII2C_IC_ENABLE, ENABLE, 1);
+	WREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_ENABLE, reg);
+
+	/* Abort previous transaction */
+	reg = REG_SET_FIELD(reg, CKSVII2C_IC_ENABLE, ABORT, 1);
+	WREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_ENABLE, reg);
+
+	DRM_DEBUG_DRIVER("I2C_Abort() Done.");
+}
+
+static bool smu_v11_0_i2c_activity_done(struct i2c_adapter *control)
+{
+	struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control);
+	struct amdgpu_device *adev = smu_i2c->adev;
+
+	const uint32_t IDLE_TIMEOUT = 1024;
+	uint32_t timeout_count = 0;
+	uint32_t reg_ic_enable, reg_ic_enable_status, reg_ic_clr_activity;
+
+	reg_ic_enable_status = RREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_ENABLE_STATUS);
+	reg_ic_enable = RREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_ENABLE);
+
+	if ((REG_GET_FIELD(reg_ic_enable, CKSVII2C_IC_ENABLE, ENABLE) == 0) &&
+	    (REG_GET_FIELD(reg_ic_enable_status, CKSVII2C_IC_ENABLE_STATUS, IC_EN) == 1)) {
+		/*
+		 * Nobody is using I2C engine, but engine remains active because
+		 * someone missed to send STOP
+		 */
+		smu_v11_0_i2c_abort(control);
+	} else if (REG_GET_FIELD(reg_ic_enable, CKSVII2C_IC_ENABLE, ENABLE) == 0) {
+		/* Nobody is using I2C engine */
+		return true;
+	}
+
+	/* Keep reading activity bit until it's cleared */
+	do {
+		reg_ic_clr_activity = RREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_CLR_ACTIVITY);
+
+		if (REG_GET_FIELD(reg_ic_clr_activity,
+		    CKSVII2C_IC_CLR_ACTIVITY, CLR_ACTIVITY) == 0)
+			return true;
+
+		++timeout_count;
+
+	} while (timeout_count < IDLE_TIMEOUT);
+
+	return false;
+}
+
+static void smu_v11_0_i2c_init(struct i2c_adapter *control)
+{
+	int res;
+
+	/* Disable clock gating */
+	smu_v11_0_i2c_set_clock_gating(control, false);
+
+	if (!smu_v11_0_i2c_activity_done(control))
+		DRM_WARN("I2C busy !");
+
+	/* Disable I2C */
+	res = smu_v11_0_i2c_enable(control, false);
+	if (res != I2C_OK)
+		smu_v11_0_i2c_abort(control);
+
+	/* Configure I2C to operate as master and in standard mode */
+	smu_v11_0_i2c_configure(control);
+
+	/* Initialize the clock to 50 kHz default */
+	smu_v11_0_i2c_set_clock(control);
+
+}
+
+static void smu_v11_0_i2c_fini(struct i2c_adapter *control)
+{
+	struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control);
+	struct amdgpu_device *adev = smu_i2c->adev;
+	u32 status, enable, en_stat;
+	int res;
+
+	res = smu_v11_0_i2c_enable(control, false);
+	if (res != I2C_OK) {
+		status  = RREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_STATUS);
+		enable  = RREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_ENABLE);
+		en_stat = RREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_ENABLE_STATUS);
+
+		/* Nobody is using the I2C engine, yet it remains
+		 * active, possibly because someone missed to send
+		 * STOP.
+		 */
+		DRM_DEBUG_DRIVER("Aborting from fini: status:0x%08x "
+				 "enable:0x%08x enable_stat:0x%08x",
+				 status, enable, en_stat);
+		smu_v11_0_i2c_abort(control);
+	}
+
+	/* Restore clock gating */
+
+	/*
+	 * TODO Reenabling clock gating seems to break subsequent SMU operation
+	 *      on the I2C bus. My guess is that SMU doesn't disable clock gating like
+	 *      we do here before working with the bus. So for now just don't restore
+	 *      it but later work with SMU to see if they have this issue and can
+	 *      update their code appropriately
+	 */
+	/* smu_v11_0_i2c_set_clock_gating(control, true); */
+
+}
+
+static bool smu_v11_0_i2c_bus_lock(struct i2c_adapter *control)
+{
+	struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control);
+	struct amdgpu_device *adev = smu_i2c->adev;
+
+	/* Send  PPSMC_MSG_RequestI2CBus */
+	if (!amdgpu_dpm_smu_i2c_bus_access(adev, true))
+		return true;
+
+	return false;
+}
+
+static bool smu_v11_0_i2c_bus_unlock(struct i2c_adapter *control)
+{
+	struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control);
+	struct amdgpu_device *adev = smu_i2c->adev;
+
+	/* Send  PPSMC_MSG_ReleaseI2CBus */
+	if (!amdgpu_dpm_smu_i2c_bus_access(adev, false))
+		return true;
+
+	return false;
+}
+
+/***************************** I2C GLUE ****************************/
+
+static uint32_t smu_v11_0_i2c_read_data(struct i2c_adapter *control,
+					struct i2c_msg *msg, uint32_t i2c_flag)
+{
+	uint32_t  ret;
+
+	ret = smu_v11_0_i2c_receive(control, msg->addr, msg->buf, msg->len, i2c_flag);
+
+	if (ret != I2C_OK)
+		DRM_ERROR("ReadData() - I2C error occurred :%x", ret);
+
+	return ret;
+}
+
+static uint32_t smu_v11_0_i2c_write_data(struct i2c_adapter *control,
+					struct i2c_msg *msg, uint32_t i2c_flag)
+{
+	uint32_t  ret;
+
+	ret = smu_v11_0_i2c_transmit(control, msg->addr, msg->buf, msg->len, i2c_flag);
+
+	if (ret != I2C_OK)
+		DRM_ERROR("WriteI2CData() - I2C error occurred :%x", ret);
+
+	return ret;
+
+}
+
+static void lock_bus(struct i2c_adapter *i2c, unsigned int flags)
+{
+	struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(i2c);
+	struct amdgpu_device *adev = smu_i2c->adev;
+
+	mutex_lock(&smu_i2c->mutex);
+	if (!smu_v11_0_i2c_bus_lock(i2c))
+		DRM_ERROR("Failed to lock the bus from SMU");
+	else
+		adev->pm.bus_locked = true;
+}
+
+static int trylock_bus(struct i2c_adapter *i2c, unsigned int flags)
+{
+	WARN_ONCE(1, "This operation not supposed to run in atomic context!");
+	return false;
+}
+
+static void unlock_bus(struct i2c_adapter *i2c, unsigned int flags)
+{
+	struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(i2c);
+	struct amdgpu_device *adev = smu_i2c->adev;
+
+	if (!smu_v11_0_i2c_bus_unlock(i2c))
+		DRM_ERROR("Failed to unlock the bus from SMU");
+	else
+		adev->pm.bus_locked = false;
+	mutex_unlock(&smu_i2c->mutex);
+}
+
+static const struct i2c_lock_operations smu_v11_0_i2c_i2c_lock_ops = {
+	.lock_bus = lock_bus,
+	.trylock_bus = trylock_bus,
+	.unlock_bus = unlock_bus,
+};
+
+static int smu_v11_0_i2c_xfer(struct i2c_adapter *i2c_adap,
+			      struct i2c_msg *msg, int num)
+{
+	int i, ret;
+	u16 addr, dir;
+
+	smu_v11_0_i2c_init(i2c_adap);
+
+	/* From the client's point of view, this sequence of
+	 * messages-- the array i2c_msg *msg, is a single transaction
+	 * on the bus, starting with START and ending with STOP.
+	 *
+	 * The client is welcome to send any sequence of messages in
+	 * this array, as processing under this function here is
+	 * striving to be agnostic.
+	 *
+	 * Record the first address and direction we see. If either
+	 * changes for a subsequent message, generate ReSTART. The
+	 * DW_apb_i2c databook, v1.21a, specifies that ReSTART is
+	 * generated when the direction changes, with the default IP
+	 * block parameter settings, but it doesn't specify if ReSTART
+	 * is generated when the address changes (possibly...). We
+	 * don't rely on the default IP block parameter settings as
+	 * the block is shared and they may change.
+	 */
+	if (num > 0) {
+		addr = msg[0].addr;
+		dir  = msg[0].flags & I2C_M_RD;
+	}
+
+	for (i = 0; i < num; i++) {
+		u32 i2c_flag = 0;
+
+		if (msg[i].addr != addr || (msg[i].flags ^ dir) & I2C_M_RD) {
+			addr = msg[i].addr;
+			dir  = msg[i].flags & I2C_M_RD;
+			i2c_flag |= I2C_X_RESTART;
+		}
+
+		if (i == num - 1) {
+			/* Set the STOP bit on the last message, so
+			 * that the IP block generates a STOP after
+			 * the last byte of the message.
+			 */
+			i2c_flag |= I2C_M_STOP;
+		}
+
+		if (msg[i].flags & I2C_M_RD)
+			ret = smu_v11_0_i2c_read_data(i2c_adap,
+						      msg + i,
+						      i2c_flag);
+		else
+			ret = smu_v11_0_i2c_write_data(i2c_adap,
+						       msg + i,
+						       i2c_flag);
+
+		if (ret != I2C_OK) {
+			num = -EIO;
+			break;
+		}
+	}
+
+	smu_v11_0_i2c_fini(i2c_adap);
+	return num;
+}
+
+static u32 smu_v11_0_i2c_func(struct i2c_adapter *adap)
+{
+	return I2C_FUNC_I2C | I2C_FUNC_SMBUS_EMUL;
+}
+
+static const struct i2c_algorithm smu_v11_0_i2c_algo = {
+	.master_xfer = smu_v11_0_i2c_xfer,
+	.functionality = smu_v11_0_i2c_func,
+};
+
+static const struct i2c_adapter_quirks smu_v11_0_i2c_control_quirks = {
+	.flags = I2C_AQ_NO_ZERO_LEN,
+};
+
+int smu_v11_0_i2c_control_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_smu_i2c_bus *smu_i2c = &adev->pm.smu_i2c[0];
+	struct i2c_adapter *control = &smu_i2c->adapter;
+	int res;
+
+	smu_i2c->adev = adev;
+	smu_i2c->port = 0;
+	mutex_init(&smu_i2c->mutex);
+	control->owner = THIS_MODULE;
+	control->class = I2C_CLASS_HWMON;
+	control->dev.parent = &adev->pdev->dev;
+	control->algo = &smu_v11_0_i2c_algo;
+	snprintf(control->name, sizeof(control->name), "AMDGPU SMU 0");
+	control->lock_ops = &smu_v11_0_i2c_i2c_lock_ops;
+	control->quirks = &smu_v11_0_i2c_control_quirks;
+	i2c_set_adapdata(control, smu_i2c);
+
+	adev->pm.ras_eeprom_i2c_bus = &adev->pm.smu_i2c[0].adapter;
+	adev->pm.fru_eeprom_i2c_bus = &adev->pm.smu_i2c[0].adapter;
+
+	res = devm_i2c_add_adapter(adev->dev, control);
+	if (res)
+		DRM_ERROR("Failed to register hw i2c, err: %d\n", res);
+
+	return res;
+}
+
+void smu_v11_0_i2c_control_fini(struct amdgpu_device *adev)
+{
+	adev->pm.ras_eeprom_i2c_bus = NULL;
+	adev->pm.fru_eeprom_i2c_bus = NULL;
+}
+
+/*
+ * Keep this for future unit test if bugs arise
+ */
+#if 0
+#define I2C_TARGET_ADDR 0xA0
+
+bool smu_v11_0_i2c_test_bus(struct i2c_adapter *control)
+{
+
+	uint32_t ret = I2C_OK;
+	uint8_t data[6] = {0xf, 0, 0xde, 0xad, 0xbe, 0xef};
+
+
+	DRM_INFO("Begin");
+
+	if (!smu_v11_0_i2c_bus_lock(control)) {
+		DRM_ERROR("Failed to lock the bus!.");
+		return false;
+	}
+
+	smu_v11_0_i2c_init(control);
+
+	/* Write 0xde to address 0x0000 on the EEPROM */
+	ret = smu_v11_0_i2c_write_data(control, I2C_TARGET_ADDR, data, 6);
+
+	ret = smu_v11_0_i2c_read_data(control, I2C_TARGET_ADDR, data, 6);
+
+	smu_v11_0_i2c_fini(control);
+
+	smu_v11_0_i2c_bus_unlock(control);
+
+
+	DRM_INFO("End");
+	return true;
+}
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/smu_v11_0_i2c.h b/drivers/gpu/drm/amd/amdgpu/smu_v11_0_i2c.h
new file mode 100644
index 000000000000..96ad14288a0c
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/smu_v11_0_i2c.h
@@ -0,0 +1,34 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef SMU_V11_I2C_CONTROL_H
+#define SMU_V11_I2C_CONTROL_H
+
+#include <linux/types.h>
+
+struct amdgpu_device;
+
+int smu_v11_0_i2c_control_init(struct amdgpu_device *adev);
+void smu_v11_0_i2c_control_fini(struct amdgpu_device *adev);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/smu_v13_0_10.c b/drivers/gpu/drm/amd/amdgpu/smu_v13_0_10.c
new file mode 100644
index 000000000000..70569ea906bc
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/smu_v13_0_10.c
@@ -0,0 +1,296 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "smu_v13_0_10.h"
+#include "amdgpu_reset.h"
+#include "amdgpu_dpm.h"
+#include "amdgpu_job.h"
+#include "amdgpu_ring.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_psp.h"
+
+static bool smu_v13_0_10_is_mode2_default(struct amdgpu_reset_control *reset_ctl)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle;
+	if (adev->pm.fw_version >= 0x00502005 && !amdgpu_sriov_vf(adev))
+		return true;
+
+	return false;
+}
+
+static struct amdgpu_reset_handler *
+smu_v13_0_10_get_reset_handler(struct amdgpu_reset_control *reset_ctl,
+			    struct amdgpu_reset_context *reset_context)
+{
+	struct amdgpu_reset_handler *handler;
+	struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle;
+	int i;
+
+	if (reset_context->method != AMD_RESET_METHOD_NONE) {
+		for_each_handler(i, handler, reset_ctl) {
+			if (handler->reset_method == reset_context->method)
+				return handler;
+		}
+	}
+
+	if (smu_v13_0_10_is_mode2_default(reset_ctl) &&
+		amdgpu_asic_reset_method(adev) == AMD_RESET_METHOD_MODE2) {
+		for_each_handler(i, handler, reset_ctl)	{
+			if (handler->reset_method == AMD_RESET_METHOD_MODE2)
+				return handler;
+		}
+	}
+
+	return NULL;
+}
+
+static int smu_v13_0_10_mode2_suspend_ip(struct amdgpu_device *adev)
+{
+	int r, i;
+
+	amdgpu_device_set_pg_state(adev, AMD_PG_STATE_UNGATE);
+	amdgpu_device_set_cg_state(adev, AMD_CG_STATE_UNGATE);
+
+	for (i = adev->num_ip_blocks - 1; i >= 0; i--) {
+		if (!(adev->ip_blocks[i].version->type ==
+			      AMD_IP_BLOCK_TYPE_GFX ||
+		      adev->ip_blocks[i].version->type ==
+			      AMD_IP_BLOCK_TYPE_SDMA ||
+		      adev->ip_blocks[i].version->type ==
+			      AMD_IP_BLOCK_TYPE_MES))
+			continue;
+
+		r = amdgpu_ip_block_suspend(&adev->ip_blocks[i]);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+static int
+smu_v13_0_10_mode2_prepare_hwcontext(struct amdgpu_reset_control *reset_ctl,
+				  struct amdgpu_reset_context *reset_context)
+{
+	int r = 0;
+	struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle;
+
+	if (!amdgpu_sriov_vf(adev))
+		r = smu_v13_0_10_mode2_suspend_ip(adev);
+
+	return r;
+}
+
+static int smu_v13_0_10_mode2_reset(struct amdgpu_device *adev)
+{
+	return amdgpu_dpm_mode2_reset(adev);
+}
+
+static void smu_v13_0_10_async_reset(struct work_struct *work)
+{
+	struct amdgpu_reset_handler *handler;
+	struct amdgpu_reset_control *reset_ctl =
+		container_of(work, struct amdgpu_reset_control, reset_work);
+	struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle;
+	int i;
+
+	for_each_handler(i, handler, reset_ctl)	{
+		if (handler->reset_method == reset_ctl->active_reset) {
+			dev_dbg(adev->dev, "Resetting device\n");
+			handler->do_reset(adev);
+			break;
+		}
+	}
+}
+static int
+smu_v13_0_10_mode2_perform_reset(struct amdgpu_reset_control *reset_ctl,
+			      struct amdgpu_reset_context *reset_context)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle;
+	int r;
+
+	r = smu_v13_0_10_mode2_reset(adev);
+	if (r) {
+		dev_err(adev->dev,
+			"ASIC reset failed with error, %d ", r);
+	}
+	return r;
+}
+
+static int smu_v13_0_10_mode2_restore_ip(struct amdgpu_device *adev)
+{
+	int i, r;
+	struct psp_context *psp = &adev->psp;
+	struct amdgpu_firmware_info *ucode;
+	struct amdgpu_firmware_info *ucode_list[2];
+	int ucode_count = 0;
+
+	for (i = 0; i < adev->firmware.max_ucodes; i++) {
+		ucode = &adev->firmware.ucode[i];
+
+		switch (ucode->ucode_id) {
+		case AMDGPU_UCODE_ID_IMU_I:
+		case AMDGPU_UCODE_ID_IMU_D:
+			ucode_list[ucode_count++] = ucode;
+			break;
+		default:
+			break;
+		}
+	}
+
+	r = psp_load_fw_list(psp, ucode_list, ucode_count);
+	if (r) {
+		dev_err(adev->dev, "IMU ucode load failed after mode2 reset\n");
+		return r;
+	}
+
+	r = psp_rlc_autoload_start(psp);
+	if (r) {
+		DRM_ERROR("Failed to start rlc autoload after mode2 reset\n");
+		return r;
+	}
+
+	amdgpu_dpm_enable_gfx_features(adev);
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!(adev->ip_blocks[i].version->type ==
+			      AMD_IP_BLOCK_TYPE_GFX ||
+		      adev->ip_blocks[i].version->type ==
+			      AMD_IP_BLOCK_TYPE_MES ||
+		      adev->ip_blocks[i].version->type ==
+			      AMD_IP_BLOCK_TYPE_SDMA))
+			continue;
+		r = amdgpu_ip_block_resume(&adev->ip_blocks[i]);
+		if (r)
+			return r;
+	}
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!(adev->ip_blocks[i].version->type ==
+			      AMD_IP_BLOCK_TYPE_GFX ||
+		      adev->ip_blocks[i].version->type ==
+			      AMD_IP_BLOCK_TYPE_MES ||
+		      adev->ip_blocks[i].version->type ==
+			      AMD_IP_BLOCK_TYPE_SDMA))
+			continue;
+
+		if (adev->ip_blocks[i].version->funcs->late_init) {
+			r = adev->ip_blocks[i].version->funcs->late_init(
+				&adev->ip_blocks[i]);
+			if (r) {
+				dev_err(adev->dev,
+					"late_init of IP block <%s> failed %d after reset\n",
+					adev->ip_blocks[i].version->funcs->name,
+					r);
+				return r;
+			}
+		}
+		adev->ip_blocks[i].status.late_initialized = true;
+	}
+
+	amdgpu_device_set_cg_state(adev, AMD_CG_STATE_GATE);
+	amdgpu_device_set_pg_state(adev, AMD_PG_STATE_GATE);
+
+	return r;
+}
+
+static int
+smu_v13_0_10_mode2_restore_hwcontext(struct amdgpu_reset_control *reset_ctl,
+				  struct amdgpu_reset_context *reset_context)
+{
+	int r;
+	struct amdgpu_device *tmp_adev = (struct amdgpu_device *)reset_ctl->handle;
+
+	amdgpu_set_init_level(tmp_adev, AMDGPU_INIT_LEVEL_RESET_RECOVERY);
+	dev_info(tmp_adev->dev,
+			"GPU reset succeeded, trying to resume\n");
+	r = smu_v13_0_10_mode2_restore_ip(tmp_adev);
+	if (r)
+		goto end;
+
+	amdgpu_register_gpu_instance(tmp_adev);
+
+	/* Resume RAS */
+	amdgpu_ras_resume(tmp_adev);
+
+	amdgpu_irq_gpu_reset_resume_helper(tmp_adev);
+
+	amdgpu_set_init_level(tmp_adev, AMDGPU_INIT_LEVEL_DEFAULT);
+	r = amdgpu_ib_ring_tests(tmp_adev);
+	if (r) {
+		dev_err(tmp_adev->dev,
+			"ib ring test failed (%d).\n", r);
+		r = -EAGAIN;
+		goto end;
+	}
+
+end:
+	if (r)
+		return -EAGAIN;
+	else
+		return r;
+}
+
+static struct amdgpu_reset_handler smu_v13_0_10_mode2_handler = {
+	.reset_method		= AMD_RESET_METHOD_MODE2,
+	.prepare_env		= NULL,
+	.prepare_hwcontext	= smu_v13_0_10_mode2_prepare_hwcontext,
+	.perform_reset		= smu_v13_0_10_mode2_perform_reset,
+	.restore_hwcontext	= smu_v13_0_10_mode2_restore_hwcontext,
+	.restore_env		= NULL,
+	.do_reset		= smu_v13_0_10_mode2_reset,
+};
+
+static struct amdgpu_reset_handler
+	*smu_v13_0_10_rst_handlers[AMDGPU_RESET_MAX_HANDLERS] = {
+		&smu_v13_0_10_mode2_handler,
+	};
+
+int smu_v13_0_10_reset_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_reset_control *reset_ctl;
+
+	reset_ctl = kzalloc(sizeof(*reset_ctl), GFP_KERNEL);
+	if (!reset_ctl)
+		return -ENOMEM;
+
+	reset_ctl->handle = adev;
+	reset_ctl->async_reset = smu_v13_0_10_async_reset;
+	reset_ctl->active_reset = AMD_RESET_METHOD_NONE;
+	reset_ctl->get_reset_handler = smu_v13_0_10_get_reset_handler;
+
+	INIT_WORK(&reset_ctl->reset_work, reset_ctl->async_reset);
+	/* Only mode2 is handled through reset control now */
+	reset_ctl->reset_handlers = &smu_v13_0_10_rst_handlers;
+
+	adev->reset_cntl = reset_ctl;
+
+	return 0;
+}
+
+int smu_v13_0_10_reset_fini(struct amdgpu_device *adev)
+{
+	kfree(adev->reset_cntl);
+	adev->reset_cntl = NULL;
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/smu_v13_0_10.h b/drivers/gpu/drm/amd/amdgpu/smu_v13_0_10.h
new file mode 100644
index 000000000000..e0cb72a0eec6
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/smu_v13_0_10.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __SMU_V13_0_10_H__
+#define __SMU_V13_0_10_H__
+
+#include "amdgpu.h"
+
+int smu_v13_0_10_reset_init(struct amdgpu_device *adev);
+int smu_v13_0_10_reset_fini(struct amdgpu_device *adev);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/smuio_v11_0.c b/drivers/gpu/drm/amd/amdgpu/smuio_v11_0.c
new file mode 100644
index 000000000000..acdc40f99ab3
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/smuio_v11_0.c
@@ -0,0 +1,80 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "smuio_v11_0.h"
+#include "smuio/smuio_11_0_0_offset.h"
+#include "smuio/smuio_11_0_0_sh_mask.h"
+
+static u32 smuio_v11_0_get_rom_index_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(SMUIO, 0, mmROM_INDEX);
+}
+
+static u32 smuio_v11_0_get_rom_data_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(SMUIO, 0, mmROM_DATA);
+}
+
+static void smuio_v11_0_update_rom_clock_gating(struct amdgpu_device *adev, bool enable)
+{
+	u32 def, data;
+
+	/* enable/disable ROM CG is not supported on APU */
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_ROM_MGCG))
+		return;
+
+	def = data = RREG32_SOC15(SMUIO, 0, mmCGTT_ROM_CLK_CTRL0);
+
+	if (enable)
+		data &= ~(CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK |
+			CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE1_MASK);
+	else
+		data |= CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK |
+			CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE1_MASK;
+
+	if (def != data)
+		WREG32_SOC15(SMUIO, 0, mmCGTT_ROM_CLK_CTRL0, data);
+}
+
+static void smuio_v11_0_get_clock_gating_state(struct amdgpu_device *adev, u64 *flags)
+{
+	u32 data;
+
+	/* CGTT_ROM_CLK_CTRL0 is not available for APU */
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	data = RREG32_SOC15(SMUIO, 0, mmCGTT_ROM_CLK_CTRL0);
+	if (!(data & CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK))
+		*flags |= AMD_CG_SUPPORT_ROM_MGCG;
+}
+
+const struct amdgpu_smuio_funcs smuio_v11_0_funcs = {
+	.get_rom_index_offset = smuio_v11_0_get_rom_index_offset,
+	.get_rom_data_offset = smuio_v11_0_get_rom_data_offset,
+	.update_rom_clock_gating = smuio_v11_0_update_rom_clock_gating,
+	.get_clock_gating_state = smuio_v11_0_get_clock_gating_state,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/smuio_v11_0.h b/drivers/gpu/drm/amd/amdgpu/smuio_v11_0.h
new file mode 100644
index 000000000000..43c4262f2b8b
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/smuio_v11_0.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __SMUIO_V11_0_H__
+#define __SMUIO_V11_0_H__
+
+#include "soc15_common.h"
+
+extern const struct amdgpu_smuio_funcs smuio_v11_0_funcs;
+
+#endif /* __SMUIO_V11_0_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/smuio_v11_0_6.c b/drivers/gpu/drm/amd/amdgpu/smuio_v11_0_6.c
new file mode 100644
index 000000000000..2afeb8b37f62
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/smuio_v11_0_6.c
@@ -0,0 +1,77 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "smuio_v11_0_6.h"
+#include "smuio/smuio_11_0_6_offset.h"
+#include "smuio/smuio_11_0_6_sh_mask.h"
+
+static u32 smuio_v11_0_6_get_rom_index_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(SMUIO, 0, mmROM_INDEX);
+}
+
+static u32 smuio_v11_0_6_get_rom_data_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(SMUIO, 0, mmROM_DATA);
+}
+
+static void smuio_v11_0_6_update_rom_clock_gating(struct amdgpu_device *adev, bool enable)
+{
+	u32 def, data;
+
+	/* enable/disable ROM CG is not supported on APU */
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	def = data = RREG32_SOC15(SMUIO, 0, mmCGTT_ROM_CLK_CTRL0);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_ROM_MGCG))
+		data &= ~(CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK |
+			CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE1_MASK);
+	else
+		data |= CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK |
+			CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE1_MASK;
+
+	if (def != data)
+		WREG32_SOC15(SMUIO, 0, mmCGTT_ROM_CLK_CTRL0, data);
+}
+
+static void smuio_v11_0_6_get_clock_gating_state(struct amdgpu_device *adev, u64 *flags)
+{
+	u32 data;
+
+	/* CGTT_ROM_CLK_CTRL0 is not available for APU */
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	data = RREG32_SOC15(SMUIO, 0, mmCGTT_ROM_CLK_CTRL0);
+	if (!(data & CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK))
+		*flags |= AMD_CG_SUPPORT_ROM_MGCG;
+}
+
+const struct amdgpu_smuio_funcs smuio_v11_0_6_funcs = {
+	.get_rom_index_offset = smuio_v11_0_6_get_rom_index_offset,
+	.get_rom_data_offset = smuio_v11_0_6_get_rom_data_offset,
+	.update_rom_clock_gating = smuio_v11_0_6_update_rom_clock_gating,
+	.get_clock_gating_state = smuio_v11_0_6_get_clock_gating_state,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/smuio_v11_0_6.h b/drivers/gpu/drm/amd/amdgpu/smuio_v11_0_6.h
new file mode 100644
index 000000000000..3c3f4ab0bc9b
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/smuio_v11_0_6.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __SMUIO_V11_0_6_H__
+#define __SMUIO_V11_0_6_H__
+
+#include "soc15_common.h"
+
+extern const struct amdgpu_smuio_funcs smuio_v11_0_6_funcs;
+
+#endif /* __SMUIO_V11_0_6_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/smuio_v13_0.c b/drivers/gpu/drm/amd/amdgpu/smuio_v13_0.c
new file mode 100644
index 000000000000..bf8b8e5ddf5d
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/smuio_v13_0.c
@@ -0,0 +1,161 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "smuio_v13_0.h"
+#include "smuio/smuio_13_0_2_offset.h"
+#include "smuio/smuio_13_0_2_sh_mask.h"
+
+#define SMUIO_MCM_CONFIG__HOST_GPU_XGMI_MASK	0x00000001L
+
+static u32 smuio_v13_0_get_rom_index_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(SMUIO, 0, regROM_INDEX);
+}
+
+static u32 smuio_v13_0_get_rom_data_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(SMUIO, 0, regROM_DATA);
+}
+
+static void smuio_v13_0_update_rom_clock_gating(struct amdgpu_device *adev, bool enable)
+{
+	u32 def, data;
+
+	/* enable/disable ROM CG is not supported on APU */
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	def = data = RREG32_SOC15(SMUIO, 0, regCGTT_ROM_CLK_CTRL0);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_ROM_MGCG))
+		data &= ~(CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK |
+			CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE1_MASK);
+	else
+		data |= CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK |
+			CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE1_MASK;
+
+	if (def != data)
+		WREG32_SOC15(SMUIO, 0, regCGTT_ROM_CLK_CTRL0, data);
+}
+
+static void smuio_v13_0_get_clock_gating_state(struct amdgpu_device *adev, u64 *flags)
+{
+	u32 data;
+
+	/* CGTT_ROM_CLK_CTRL0 is not available for APU */
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	data = RREG32_SOC15(SMUIO, 0, regCGTT_ROM_CLK_CTRL0);
+	if (!(data & CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK))
+		*flags |= AMD_CG_SUPPORT_ROM_MGCG;
+}
+
+/**
+ * smuio_v13_0_get_die_id - query die id from FCH.
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Returns die id
+ */
+static u32 smuio_v13_0_get_die_id(struct amdgpu_device *adev)
+{
+	u32 data, die_id;
+
+	data = RREG32_SOC15(SMUIO, 0, regSMUIO_MCM_CONFIG);
+	die_id = REG_GET_FIELD(data, SMUIO_MCM_CONFIG, DIE_ID);
+
+	return die_id;
+}
+
+/**
+ * smuio_v13_0_get_socket_id - query socket id from FCH
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Returns socket id
+ */
+static u32 smuio_v13_0_get_socket_id(struct amdgpu_device *adev)
+{
+	u32 data, socket_id;
+
+	data = RREG32_SOC15(SMUIO, 0, regSMUIO_MCM_CONFIG);
+	socket_id = REG_GET_FIELD(data, SMUIO_MCM_CONFIG, SOCKET_ID);
+
+	return socket_id;
+}
+
+/**
+ * smuio_v13_0_is_host_gpu_xgmi_supported - detect xgmi interface between cpu and gpu/s.
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Returns true on success or false otherwise.
+ */
+static bool smuio_v13_0_is_host_gpu_xgmi_supported(struct amdgpu_device *adev)
+{
+	u32 data;
+
+	data = RREG32_SOC15(SMUIO, 0, regSMUIO_MCM_CONFIG);
+	data = REG_GET_FIELD(data, SMUIO_MCM_CONFIG, TOPOLOGY_ID);
+	/* data[4:0]
+	 * bit 0 == 0 host-gpu interface is PCIE
+	 * bit 0 == 1 host-gpu interface is Alternate Protocal
+	 * for AMD, this is XGMI
+	 */
+	data &= SMUIO_MCM_CONFIG__HOST_GPU_XGMI_MASK;
+
+	return data ? true : false;
+}
+
+static enum amdgpu_pkg_type smuio_v13_0_get_pkg_type(struct amdgpu_device *adev)
+{
+	enum amdgpu_pkg_type pkg_type;
+	u32 data;
+
+	data = RREG32_SOC15(SMUIO, 0, regSMUIO_MCM_CONFIG);
+	data = REG_GET_FIELD(data, SMUIO_MCM_CONFIG, TOPOLOGY_ID);
+
+	switch (data) {
+	case 0x4:
+	case 0xC:
+		pkg_type = AMDGPU_PKG_TYPE_CEM;
+		break;
+	default:
+		pkg_type = AMDGPU_PKG_TYPE_OAM;
+		break;
+	}
+
+	return pkg_type;
+}
+
+const struct amdgpu_smuio_funcs smuio_v13_0_funcs = {
+	.get_rom_index_offset = smuio_v13_0_get_rom_index_offset,
+	.get_rom_data_offset = smuio_v13_0_get_rom_data_offset,
+	.get_die_id = smuio_v13_0_get_die_id,
+	.get_socket_id = smuio_v13_0_get_socket_id,
+	.is_host_gpu_xgmi_supported = smuio_v13_0_is_host_gpu_xgmi_supported,
+	.update_rom_clock_gating = smuio_v13_0_update_rom_clock_gating,
+	.get_clock_gating_state = smuio_v13_0_get_clock_gating_state,
+	.get_pkg_type = smuio_v13_0_get_pkg_type,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/smuio_v13_0.h b/drivers/gpu/drm/amd/amdgpu/smuio_v13_0.h
new file mode 100644
index 000000000000..a3bfe3e4fb46
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/smuio_v13_0.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __SMUIO_V13_0_H__
+#define __SMUIO_V13_0_H__
+
+#include "soc15_common.h"
+
+extern const struct amdgpu_smuio_funcs smuio_v13_0_funcs;
+
+#endif /* __SMUIO_V13_0_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/smuio_v13_0_3.c b/drivers/gpu/drm/amd/amdgpu/smuio_v13_0_3.c
new file mode 100644
index 000000000000..5461b5289793
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/smuio_v13_0_3.c
@@ -0,0 +1,109 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "smuio_v13_0_3.h"
+#include "soc15_common.h"
+#include "smuio/smuio_13_0_3_offset.h"
+#include "smuio/smuio_13_0_3_sh_mask.h"
+
+#define PKG_TYPE_MASK		0x00000003L
+
+/**
+ * smuio_v13_0_3_get_die_id - query die id from FCH.
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Returns die id
+ */
+static u32 smuio_v13_0_3_get_die_id(struct amdgpu_device *adev)
+{
+	u32 data, die_id;
+
+	data = RREG32_SOC15(SMUIO, 0, regSMUIO_MCM_CONFIG);
+	die_id = REG_GET_FIELD(data, SMUIO_MCM_CONFIG, DIE_ID);
+
+	return die_id;
+}
+
+/**
+ * smuio_v13_0_3_get_socket_id - query socket id from FCH
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Returns socket id
+ */
+static u32 smuio_v13_0_3_get_socket_id(struct amdgpu_device *adev)
+{
+	u32 data, socket_id;
+
+	data = RREG32_SOC15(SMUIO, 0, regSMUIO_MCM_CONFIG);
+	socket_id = REG_GET_FIELD(data, SMUIO_MCM_CONFIG, SOCKET_ID);
+
+	return socket_id;
+}
+
+/**
+ * smuio_v13_0_3_get_pkg_type - query package type set by MP1/bootcode
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Returns package type
+ */
+
+static enum amdgpu_pkg_type smuio_v13_0_3_get_pkg_type(struct amdgpu_device *adev)
+{
+	enum amdgpu_pkg_type pkg_type;
+	u32 data;
+
+	data = RREG32_SOC15(SMUIO, 0, regSMUIO_MCM_CONFIG);
+	data = REG_GET_FIELD(data, SMUIO_MCM_CONFIG, PKG_TYPE);
+	/* pkg_type[4:0]
+	 *
+	 * bit 1 == 1 APU form factor
+	 *
+	 * b0100 - b1111 - Reserved
+	 */
+	switch (data & PKG_TYPE_MASK) {
+	case 0x0:
+		pkg_type = AMDGPU_PKG_TYPE_CEM;
+		break;
+	case 0x1:
+		pkg_type = AMDGPU_PKG_TYPE_OAM;
+		break;
+	case 0x2:
+		pkg_type = AMDGPU_PKG_TYPE_APU;
+		break;
+	default:
+		pkg_type = AMDGPU_PKG_TYPE_UNKNOWN;
+		break;
+	}
+
+	return pkg_type;
+}
+
+
+const struct amdgpu_smuio_funcs smuio_v13_0_3_funcs = {
+	.get_die_id = smuio_v13_0_3_get_die_id,
+	.get_socket_id = smuio_v13_0_3_get_socket_id,
+	.get_pkg_type = smuio_v13_0_3_get_pkg_type,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/smuio_v13_0_3.h b/drivers/gpu/drm/amd/amdgpu/smuio_v13_0_3.h
new file mode 100644
index 000000000000..795f66c5a58b
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/smuio_v13_0_3.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __SMUIO_V13_0_3_H__
+#define __SMUIO_V13_0_3_H__
+
+#include "soc15_common.h"
+
+extern const struct amdgpu_smuio_funcs smuio_v13_0_3_funcs;
+
+#endif /* __SMUIO_V13_0_3_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/smuio_v13_0_6.c b/drivers/gpu/drm/amd/amdgpu/smuio_v13_0_6.c
new file mode 100644
index 000000000000..de998e328b08
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/smuio_v13_0_6.c
@@ -0,0 +1,41 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "smuio_v13_0_6.h"
+#include "smuio/smuio_13_0_6_offset.h"
+#include "smuio/smuio_13_0_6_sh_mask.h"
+
+static u32 smuio_v13_0_6_get_rom_index_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(SMUIO, 0, regROM_INDEX);
+}
+
+static u32 smuio_v13_0_6_get_rom_data_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(SMUIO, 0, regROM_DATA);
+}
+
+const struct amdgpu_smuio_funcs smuio_v13_0_6_funcs = {
+	.get_rom_index_offset = smuio_v13_0_6_get_rom_index_offset,
+	.get_rom_data_offset = smuio_v13_0_6_get_rom_data_offset,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/smuio_v13_0_6.h b/drivers/gpu/drm/amd/amdgpu/smuio_v13_0_6.h
new file mode 100644
index 000000000000..c75621de5ab5
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/smuio_v13_0_6.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __SMUIO_V13_0_6_H__
+#define __SMUIO_V13_0_6_H__
+
+#include "soc15_common.h"
+
+extern const struct amdgpu_smuio_funcs smuio_v13_0_6_funcs;
+
+#endif /* __SMUIO_V13_0_6_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/smuio_v14_0_2.c b/drivers/gpu/drm/amd/amdgpu/smuio_v14_0_2.c
new file mode 100644
index 000000000000..2a51a70d4846
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/smuio_v14_0_2.c
@@ -0,0 +1,62 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "smuio_v14_0_2.h"
+#include "smuio/smuio_14_0_2_offset.h"
+#include "smuio/smuio_14_0_2_sh_mask.h"
+#include <linux/preempt.h>
+
+static u32 smuio_v14_0_2_get_rom_index_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(SMUIO, 0, regROM_INDEX);
+}
+
+static u32 smuio_v14_0_2_get_rom_data_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(SMUIO, 0, regROM_DATA);
+}
+
+static u64 smuio_v14_0_2_get_gpu_clock_counter(struct amdgpu_device *adev)
+{
+	u64 clock;
+	u64 clock_counter_lo, clock_counter_hi_pre, clock_counter_hi_after;
+
+	preempt_disable();
+	clock_counter_hi_pre = (u64)RREG32_SOC15(SMUIO, 0, regGOLDEN_TSC_COUNT_UPPER);
+	clock_counter_lo = (u64)RREG32_SOC15(SMUIO, 0, regGOLDEN_TSC_COUNT_LOWER);
+	/* the clock counter may be udpated during polling the counters */
+	clock_counter_hi_after = (u64)RREG32_SOC15(SMUIO, 0, regGOLDEN_TSC_COUNT_UPPER);
+	if (clock_counter_hi_pre != clock_counter_hi_after)
+		clock_counter_lo = (u64)RREG32_SOC15(SMUIO, 0, regGOLDEN_TSC_COUNT_LOWER);
+	preempt_enable();
+
+	clock = clock_counter_lo | (clock_counter_hi_after << 32ULL);
+
+	return clock;
+}
+
+const struct amdgpu_smuio_funcs smuio_v14_0_2_funcs = {
+	.get_rom_index_offset = smuio_v14_0_2_get_rom_index_offset,
+	.get_rom_data_offset = smuio_v14_0_2_get_rom_data_offset,
+	.get_gpu_clock_counter = smuio_v14_0_2_get_gpu_clock_counter,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/smuio_v14_0_2.h b/drivers/gpu/drm/amd/amdgpu/smuio_v14_0_2.h
new file mode 100644
index 000000000000..6e617f832d90
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/smuio_v14_0_2.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __SMUIO_V14_0_2_H__
+#define __SMUIO_V14_0_2_H__
+
+#include "soc15_common.h"
+
+extern const struct amdgpu_smuio_funcs smuio_v14_0_2_funcs;
+
+#endif /* __SMUIO_V14_0_2_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/smuio_v9_0.c b/drivers/gpu/drm/amd/amdgpu/smuio_v9_0.c
new file mode 100644
index 000000000000..c04fdd2d5b38
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/smuio_v9_0.c
@@ -0,0 +1,77 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "smuio_v9_0.h"
+#include "smuio/smuio_9_0_offset.h"
+#include "smuio/smuio_9_0_sh_mask.h"
+
+static u32 smuio_v9_0_get_rom_index_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(SMUIO, 0, mmROM_INDEX);
+}
+
+static u32 smuio_v9_0_get_rom_data_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(SMUIO, 0, mmROM_DATA);
+}
+
+static void smuio_v9_0_update_rom_clock_gating(struct amdgpu_device *adev, bool enable)
+{
+	u32 def, data;
+
+	/* enable/disable ROM CG is not supported on APU */
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	def = data = RREG32_SOC15(SMUIO, 0, mmCGTT_ROM_CLK_CTRL0);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_ROM_MGCG))
+		data &= ~(CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK |
+			CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE1_MASK);
+	else
+		data |= CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK |
+			CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE1_MASK;
+
+	if (def != data)
+		WREG32_SOC15(SMUIO, 0, mmCGTT_ROM_CLK_CTRL0, data);
+}
+
+static void smuio_v9_0_get_clock_gating_state(struct amdgpu_device *adev, u64 *flags)
+{
+	u32 data;
+
+	/* CGTT_ROM_CLK_CTRL0 is not available for APUs */
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	data = RREG32_SOC15(SMUIO, 0, mmCGTT_ROM_CLK_CTRL0);
+	if (!(data & CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK))
+		*flags |= AMD_CG_SUPPORT_ROM_MGCG;
+}
+
+const struct amdgpu_smuio_funcs smuio_v9_0_funcs = {
+	.get_rom_index_offset = smuio_v9_0_get_rom_index_offset,
+	.get_rom_data_offset = smuio_v9_0_get_rom_data_offset,
+	.update_rom_clock_gating = smuio_v9_0_update_rom_clock_gating,
+	.get_clock_gating_state = smuio_v9_0_get_clock_gating_state,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/smuio_v9_0.h b/drivers/gpu/drm/amd/amdgpu/smuio_v9_0.h
new file mode 100644
index 000000000000..fc265ce9837d
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/smuio_v9_0.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __SMUIO_V9_0_H__
+#define __SMUIO_V9_0_H__
+
+#include "soc15_common.h"
+
+extern const struct amdgpu_smuio_funcs smuio_v9_0_funcs;
+
+#endif /* __SMUIO_V9_0_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.c b/drivers/gpu/drm/amd/amdgpu/soc15.c
index f2c3a49f73a0..9785fada4fa7 100644
--- a/drivers/gpu/drm/amd/amdgpu/soc15.c
+++ b/drivers/gpu/drm/amd/amdgpu/soc15.c
@@ -23,9 +23,11 @@
 #include <linux/firmware.h>
 #include <linux/slab.h>
 #include <linux/module.h>
-#include <drm/drmP.h>
+#include <linux/pci.h>
+
+#include <drm/amdgpu_drm.h>
+
 #include "amdgpu.h"
-#include "amdgpu_atombios.h"
 #include "amdgpu_ih.h"
 #include "amdgpu_uvd.h"
 #include "amdgpu_vce.h"
@@ -34,18 +36,16 @@
 #include "atom.h"
 #include "amd_pcie.h"
 
-#include "vega10/soc15ip.h"
-#include "vega10/UVD/uvd_7_0_offset.h"
-#include "vega10/GC/gc_9_0_offset.h"
-#include "vega10/GC/gc_9_0_sh_mask.h"
-#include "vega10/SDMA0/sdma0_4_0_offset.h"
-#include "vega10/SDMA1/sdma1_4_0_offset.h"
-#include "vega10/HDP/hdp_4_0_offset.h"
-#include "vega10/HDP/hdp_4_0_sh_mask.h"
-#include "vega10/MP/mp_9_0_offset.h"
-#include "vega10/MP/mp_9_0_sh_mask.h"
-#include "vega10/SMUIO/smuio_9_0_offset.h"
-#include "vega10/SMUIO/smuio_9_0_sh_mask.h"
+#include "uvd/uvd_7_0_offset.h"
+#include "gc/gc_9_0_offset.h"
+#include "gc/gc_9_0_sh_mask.h"
+#include "sdma0/sdma0_4_0_offset.h"
+#include "sdma1/sdma1_4_0_offset.h"
+#include "nbio/nbio_7_0_default.h"
+#include "nbio/nbio_7_0_offset.h"
+#include "nbio/nbio_7_0_sh_mask.h"
+#include "nbio/nbio_7_0_smn.h"
+#include "mp/mp_9_0_offset.h"
 
 #include "soc15.h"
 #include "soc15_common.h"
@@ -53,91 +53,188 @@
 #include "gmc_v9_0.h"
 #include "gfxhub_v1_0.h"
 #include "mmhub_v1_0.h"
+#include "df_v1_7.h"
+#include "df_v3_6.h"
+#include "nbio_v6_1.h"
+#include "nbio_v7_0.h"
+#include "nbio_v7_4.h"
+#include "hdp_v4_0.h"
 #include "vega10_ih.h"
+#include "vega20_ih.h"
+#include "navi10_ih.h"
 #include "sdma_v4_0.h"
 #include "uvd_v7_0.h"
 #include "vce_v4_0.h"
 #include "vcn_v1_0.h"
-#include "amdgpu_powerplay.h"
-#include "dce_virtual.h"
+#include "vcn_v2_0.h"
+#include "jpeg_v2_0.h"
+#include "vcn_v2_5.h"
+#include "jpeg_v2_5.h"
+#include "smuio_v9_0.h"
+#include "smuio_v11_0.h"
+#include "smuio_v13_0.h"
+#include "amdgpu_vkms.h"
 #include "mxgpu_ai.h"
-
-#define mmFabricConfigAccessControl                                                                    0x0410
-#define mmFabricConfigAccessControl_BASE_IDX                                                           0
-#define mmFabricConfigAccessControl_DEFAULT                                      0x00000000
-//FabricConfigAccessControl
-#define FabricConfigAccessControl__CfgRegInstAccEn__SHIFT                                                     0x0
-#define FabricConfigAccessControl__CfgRegInstAccRegLock__SHIFT                                                0x1
-#define FabricConfigAccessControl__CfgRegInstID__SHIFT                                                        0x10
-#define FabricConfigAccessControl__CfgRegInstAccEn_MASK                                                       0x00000001L
-#define FabricConfigAccessControl__CfgRegInstAccRegLock_MASK                                                  0x00000002L
-#define FabricConfigAccessControl__CfgRegInstID_MASK                                                          0x00FF0000L
-
-
-#define mmDF_PIE_AON0_DfGlobalClkGater                                                                 0x00fc
-#define mmDF_PIE_AON0_DfGlobalClkGater_BASE_IDX                                                        0
-//DF_PIE_AON0_DfGlobalClkGater
-#define DF_PIE_AON0_DfGlobalClkGater__MGCGMode__SHIFT                                                         0x0
-#define DF_PIE_AON0_DfGlobalClkGater__MGCGMode_MASK                                                           0x0000000FL
-
-enum {
-	DF_MGCG_DISABLE = 0,
-	DF_MGCG_ENABLE_00_CYCLE_DELAY =1,
-	DF_MGCG_ENABLE_01_CYCLE_DELAY =2,
-	DF_MGCG_ENABLE_15_CYCLE_DELAY =13,
-	DF_MGCG_ENABLE_31_CYCLE_DELAY =14,
-	DF_MGCG_ENABLE_63_CYCLE_DELAY =15
-};
+#include "amdgpu_ras.h"
+#include "amdgpu_xgmi.h"
+#include <uapi/linux/kfd_ioctl.h>
 
 #define mmMP0_MISC_CGTT_CTRL0                                                                   0x01b9
 #define mmMP0_MISC_CGTT_CTRL0_BASE_IDX                                                          0
 #define mmMP0_MISC_LIGHT_SLEEP_CTRL                                                             0x01ba
 #define mmMP0_MISC_LIGHT_SLEEP_CTRL_BASE_IDX                                                    0
 
-/*
- * Indirect registers accessor
- */
-static u32 soc15_pcie_rreg(struct amdgpu_device *adev, u32 reg)
+static const struct amd_ip_funcs soc15_common_ip_funcs;
+
+/* Vega, Raven, Arcturus */
+static const struct amdgpu_video_codec_info vega_video_codecs_encode_array[] =
 {
-	unsigned long flags, address, data;
-	u32 r;
-	struct nbio_pcie_index_data *nbio_pcie_id;
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 4096, 4096, 0)},
+};
 
-	if (adev->flags & AMD_IS_APU)
-		nbio_pcie_id = &nbio_v7_0_pcie_index_data;
-	else
-		nbio_pcie_id = &nbio_v6_1_pcie_index_data;
+static const struct amdgpu_video_codecs vega_video_codecs_encode =
+{
+	.codec_count = ARRAY_SIZE(vega_video_codecs_encode_array),
+	.codec_array = vega_video_codecs_encode_array,
+};
 
-	address = nbio_pcie_id->index_offset;
-	data = nbio_pcie_id->data_offset;
+/* Vega */
+static const struct amdgpu_video_codec_info vega_video_codecs_decode_array[] =
+{
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, 1920, 1088, 3)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, 1920, 1088, 5)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, 1920, 1088, 4)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 4096, 4096, 186)},
+};
 
-	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
-	WREG32(address, reg);
-	(void)RREG32(address);
-	r = RREG32(data);
-	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
-	return r;
-}
+static const struct amdgpu_video_codecs vega_video_codecs_decode =
+{
+	.codec_count = ARRAY_SIZE(vega_video_codecs_decode_array),
+	.codec_array = vega_video_codecs_decode_array,
+};
 
-static void soc15_pcie_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+/* Raven */
+static const struct amdgpu_video_codec_info rv_video_codecs_decode_array[] =
 {
-	unsigned long flags, address, data;
-	struct nbio_pcie_index_data *nbio_pcie_id;
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, 1920, 1088, 3)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, 1920, 1088, 5)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, 1920, 1088, 4)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 4096, 4096, 186)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 8192, 8192, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 4096, 4096, 0)},
+};
 
-	if (adev->flags & AMD_IS_APU)
-		nbio_pcie_id = &nbio_v7_0_pcie_index_data;
-	else
-		nbio_pcie_id = &nbio_v6_1_pcie_index_data;
+static const struct amdgpu_video_codecs rv_video_codecs_decode =
+{
+	.codec_count = ARRAY_SIZE(rv_video_codecs_decode_array),
+	.codec_array = rv_video_codecs_decode_array,
+};
+
+/* Renoir, Arcturus */
+static const struct amdgpu_video_codec_info rn_video_codecs_decode_array[] =
+{
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, 1920, 1088, 3)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, 1920, 1088, 5)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, 1920, 1088, 4)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)},
+};
 
-	address = nbio_pcie_id->index_offset;
-	data = nbio_pcie_id->data_offset;
+static const struct amdgpu_video_codecs rn_video_codecs_decode =
+{
+	.codec_count = ARRAY_SIZE(rn_video_codecs_decode_array),
+	.codec_array = rn_video_codecs_decode_array,
+};
+
+static const struct amdgpu_video_codec_info vcn_4_0_3_video_codecs_decode_array[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1, 8192, 4352, 0)},
+};
+
+static const struct amdgpu_video_codecs vcn_4_0_3_video_codecs_decode = {
+	.codec_count = ARRAY_SIZE(vcn_4_0_3_video_codecs_decode_array),
+	.codec_array = vcn_4_0_3_video_codecs_decode_array,
+};
 
-	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
-	WREG32(address, reg);
-	(void)RREG32(address);
-	WREG32(data, v);
-	(void)RREG32(data);
-	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+static const struct amdgpu_video_codecs vcn_4_0_3_video_codecs_encode = {
+	.codec_count = 0,
+	.codec_array = NULL,
+};
+
+static const struct amdgpu_video_codecs vcn_5_0_1_video_codecs_encode_vcn0 = {
+	.codec_count = 0,
+	.codec_array = NULL,
+};
+
+static const struct amdgpu_video_codec_info vcn_5_0_1_video_codecs_decode_array_vcn0[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1, 8192, 4352, 0)},
+};
+
+static const struct amdgpu_video_codecs vcn_5_0_1_video_codecs_decode_vcn0 = {
+	.codec_count = ARRAY_SIZE(vcn_5_0_1_video_codecs_decode_array_vcn0),
+	.codec_array = vcn_5_0_1_video_codecs_decode_array_vcn0,
+};
+
+static int soc15_query_video_codecs(struct amdgpu_device *adev, bool encode,
+				    const struct amdgpu_video_codecs **codecs)
+{
+	if (amdgpu_ip_version(adev, VCE_HWIP, 0)) {
+		switch (amdgpu_ip_version(adev, VCE_HWIP, 0)) {
+		case IP_VERSION(4, 0, 0):
+		case IP_VERSION(4, 1, 0):
+			if (encode)
+				*codecs = &vega_video_codecs_encode;
+			else
+				*codecs = &vega_video_codecs_decode;
+			return 0;
+		default:
+			return -EINVAL;
+		}
+	} else {
+		switch (amdgpu_ip_version(adev, UVD_HWIP, 0)) {
+		case IP_VERSION(1, 0, 0):
+		case IP_VERSION(1, 0, 1):
+			if (encode)
+				*codecs = &vega_video_codecs_encode;
+			else
+				*codecs = &rv_video_codecs_decode;
+			return 0;
+		case IP_VERSION(2, 5, 0):
+		case IP_VERSION(2, 6, 0):
+		case IP_VERSION(2, 2, 0):
+			if (encode)
+				*codecs = &vega_video_codecs_encode;
+			else
+				*codecs = &rn_video_codecs_decode;
+			return 0;
+		case IP_VERSION(4, 0, 3):
+			if (encode)
+				*codecs = &vcn_4_0_3_video_codecs_encode;
+			else
+				*codecs = &vcn_4_0_3_video_codecs_decode;
+			return 0;
+		case IP_VERSION(5, 0, 1):
+			if (encode)
+				*codecs = &vcn_5_0_1_video_codecs_encode_vcn0;
+			else
+				*codecs = &vcn_5_0_1_video_codecs_decode_vcn0;
+			return 0;
+		default:
+			return -EINVAL;
+		}
+	}
 }
 
 static u32 soc15_uvd_ctx_rreg(struct amdgpu_device *adev, u32 reg)
@@ -242,52 +339,29 @@ static void soc15_se_cac_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
 
 static u32 soc15_get_config_memsize(struct amdgpu_device *adev)
 {
-	if (adev->flags & AMD_IS_APU)
-		return nbio_v7_0_get_memsize(adev);
-	else
-		return nbio_v6_1_get_memsize(adev);
+	return adev->nbio.funcs->get_memsize(adev);
 }
 
-static const u32 vega10_golden_init[] =
-{
-};
-
-static const u32 raven_golden_init[] =
-{
-};
-
-static void soc15_init_golden_registers(struct amdgpu_device *adev)
-{
-	/* Some of the registers might be dependent on GRBM_GFX_INDEX */
-	mutex_lock(&adev->grbm_idx_mutex);
-
-	switch (adev->asic_type) {
-	case CHIP_VEGA10:
-		amdgpu_program_register_sequence(adev,
-						 vega10_golden_init,
-						 (const u32)ARRAY_SIZE(vega10_golden_init));
-		break;
-	case CHIP_RAVEN:
-		amdgpu_program_register_sequence(adev,
-						 raven_golden_init,
-						 (const u32)ARRAY_SIZE(raven_golden_init));
-		break;
-	default:
-		break;
-	}
-	mutex_unlock(&adev->grbm_idx_mutex);
-}
 static u32 soc15_get_xclk(struct amdgpu_device *adev)
 {
-	if (adev->asic_type == CHIP_VEGA10)
-		return adev->clock.spll.reference_freq/4;
-	else
-		return adev->clock.spll.reference_freq;
+	u32 reference_clock = adev->clock.spll.reference_freq;
+
+	if (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(12, 0, 0) ||
+	    amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(12, 0, 1) ||
+	    amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 6) ||
+	    amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 12) ||
+	    amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 14))
+		return 10000;
+	if (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(10, 0, 0) ||
+	    amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(10, 0, 1))
+		return reference_clock / 4;
+
+	return reference_clock;
 }
 
 
 void soc15_grbm_select(struct amdgpu_device *adev,
-		     u32 me, u32 pipe, u32 queue, u32 vmid)
+		     u32 me, u32 pipe, u32 queue, u32 vmid, int xcc_id)
 {
 	u32 grbm_gfx_cntl = 0;
 	grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, PIPEID, pipe);
@@ -295,12 +369,7 @@ void soc15_grbm_select(struct amdgpu_device *adev,
 	grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, VMID, vmid);
 	grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, QUEUEID, queue);
 
-	WREG32(SOC15_REG_OFFSET(GC, 0, mmGRBM_GFX_CNTL), grbm_gfx_cntl);
-}
-
-static void soc15_vga_set_state(struct amdgpu_device *adev, bool state)
-{
-	/* todo */
+	WREG32_SOC15_RLC_SHADOW(GC, xcc_id, mmGRBM_GFX_CNTL, grbm_gfx_cntl);
 }
 
 static bool soc15_read_disabled_bios(struct amdgpu_device *adev)
@@ -309,51 +378,27 @@ static bool soc15_read_disabled_bios(struct amdgpu_device *adev)
 	return false;
 }
 
-static bool soc15_read_bios_from_rom(struct amdgpu_device *adev,
-				     u8 *bios, u32 length_bytes)
-{
-	u32 *dw_ptr;
-	u32 i, length_dw;
-
-	if (bios == NULL)
-		return false;
-	if (length_bytes == 0)
-		return false;
-	/* APU vbios image is part of sbios image */
-	if (adev->flags & AMD_IS_APU)
-		return false;
-
-	dw_ptr = (u32 *)bios;
-	length_dw = ALIGN(length_bytes, 4) / 4;
-
-	/* set rom index to 0 */
-	WREG32(SOC15_REG_OFFSET(SMUIO, 0, mmROM_INDEX), 0);
-	/* read out the rom data */
-	for (i = 0; i < length_dw; i++)
-		dw_ptr[i] = RREG32(SOC15_REG_OFFSET(SMUIO, 0, mmROM_DATA));
-
-	return true;
-}
-
-static struct amdgpu_allowed_register_entry soc15_allowed_read_registers[] = {
-	{ SOC15_REG_OFFSET(GC, 0, mmGRBM_STATUS)},
-	{ SOC15_REG_OFFSET(GC, 0, mmGRBM_STATUS2)},
-	{ SOC15_REG_OFFSET(GC, 0, mmGRBM_STATUS_SE0)},
-	{ SOC15_REG_OFFSET(GC, 0, mmGRBM_STATUS_SE1)},
-	{ SOC15_REG_OFFSET(GC, 0, mmGRBM_STATUS_SE2)},
-	{ SOC15_REG_OFFSET(GC, 0, mmGRBM_STATUS_SE3)},
-	{ SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_STATUS_REG)},
-	{ SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_STATUS_REG)},
-	{ SOC15_REG_OFFSET(GC, 0, mmCP_STAT)},
-	{ SOC15_REG_OFFSET(GC, 0, mmCP_STALLED_STAT1)},
-	{ SOC15_REG_OFFSET(GC, 0, mmCP_STALLED_STAT2)},
-	{ SOC15_REG_OFFSET(GC, 0, mmCP_STALLED_STAT3)},
-	{ SOC15_REG_OFFSET(GC, 0, mmCP_CPF_BUSY_STAT)},
-	{ SOC15_REG_OFFSET(GC, 0, mmCP_CPF_STALLED_STAT1)},
-	{ SOC15_REG_OFFSET(GC, 0, mmCP_CPF_STATUS)},
-	{ SOC15_REG_OFFSET(GC, 0, mmCP_CPC_STALLED_STAT1)},
-	{ SOC15_REG_OFFSET(GC, 0, mmCP_CPC_STATUS)},
-	{ SOC15_REG_OFFSET(GC, 0, mmGB_ADDR_CONFIG)},
+static struct soc15_allowed_register_entry soc15_allowed_read_registers[] = {
+	{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS)},
+	{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS2)},
+	{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS_SE0)},
+	{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS_SE1)},
+	{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS_SE2)},
+	{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS_SE3)},
+	{ SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_STATUS_REG)},
+	{ SOC15_REG_ENTRY(SDMA1, 0, mmSDMA1_STATUS_REG)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_STAT)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_STALLED_STAT1)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_STALLED_STAT2)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_STALLED_STAT3)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_CPF_BUSY_STAT)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_CPF_STALLED_STAT1)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_CPF_STATUS)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_CPC_BUSY_STAT)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_CPC_STALLED_STAT1)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_CPC_STATUS)},
+	{ SOC15_REG_ENTRY(GC, 0, mmGB_ADDR_CONFIG)},
+	{ SOC15_REG_ENTRY(GC, 0, mmDB_DEBUG2)},
 };
 
 static uint32_t soc15_read_indexed_register(struct amdgpu_device *adev, u32 se_num,
@@ -363,12 +408,12 @@ static uint32_t soc15_read_indexed_register(struct amdgpu_device *adev, u32 se_n
 
 	mutex_lock(&adev->grbm_idx_mutex);
 	if (se_num != 0xffffffff || sh_num != 0xffffffff)
-		amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff);
+		amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff, 0);
 
 	val = RREG32(reg_offset);
 
 	if (se_num != 0xffffffff || sh_num != 0xffffffff)
-		amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+		amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 	mutex_unlock(&adev->grbm_idx_mutex);
 	return val;
 }
@@ -380,12 +425,11 @@ static uint32_t soc15_get_register_value(struct amdgpu_device *adev,
 	if (indexed) {
 		return soc15_read_indexed_register(adev, se_num, sh_num, reg_offset);
 	} else {
-		switch (reg_offset) {
-		case SOC15_REG_OFFSET(GC, 0, mmGB_ADDR_CONFIG):
+		if (reg_offset == SOC15_REG_OFFSET(GC, 0, mmGB_ADDR_CONFIG))
 			return adev->gfx.config.gb_addr_config;
-		default:
-			return RREG32(reg_offset);
-		}
+		else if (reg_offset == SOC15_REG_OFFSET(GC, 0, mmDB_DEBUG2))
+			return adev->gfx.config.db_debug2;
+		return RREG32(reg_offset);
 	}
 }
 
@@ -393,10 +437,15 @@ static int soc15_read_register(struct amdgpu_device *adev, u32 se_num,
 			    u32 sh_num, u32 reg_offset, u32 *value)
 {
 	uint32_t i;
+	struct soc15_allowed_register_entry  *en;
 
 	*value = 0;
 	for (i = 0; i < ARRAY_SIZE(soc15_allowed_read_registers); i++) {
-		if (reg_offset != soc15_allowed_read_registers[i].reg_offset)
+		en = &soc15_allowed_read_registers[i];
+		if (!adev->reg_offset[en->hwip][en->inst])
+			continue;
+		else if (reg_offset != (adev->reg_offset[en->hwip][en->inst][en->seg]
+					+ en->reg_offset))
 			continue;
 
 		*value = soc15_get_register_value(adev,
@@ -407,40 +456,217 @@ static int soc15_read_register(struct amdgpu_device *adev, u32 se_num,
 	return -EINVAL;
 }
 
-static void soc15_gpu_pci_config_reset(struct amdgpu_device *adev)
+
+/**
+ * soc15_program_register_sequence - program an array of registers.
+ *
+ * @adev: amdgpu_device pointer
+ * @regs: pointer to the register array
+ * @array_size: size of the register array
+ *
+ * Programs an array or registers with and and or masks.
+ * This is a helper for setting golden registers.
+ */
+
+void soc15_program_register_sequence(struct amdgpu_device *adev,
+					     const struct soc15_reg_golden *regs,
+					     const u32 array_size)
+{
+	const struct soc15_reg_golden *entry;
+	u32 tmp, reg;
+	int i;
+
+	for (i = 0; i < array_size; ++i) {
+		entry = &regs[i];
+		reg =  adev->reg_offset[entry->hwip][entry->instance][entry->segment] + entry->reg;
+
+		if (entry->and_mask == 0xffffffff) {
+			tmp = entry->or_mask;
+		} else {
+			tmp = (entry->hwip == GC_HWIP) ?
+				RREG32_SOC15_IP(GC, reg) : RREG32(reg);
+
+			tmp &= ~(entry->and_mask);
+			tmp |= (entry->or_mask & entry->and_mask);
+		}
+
+		if (reg == SOC15_REG_OFFSET(GC, 0, mmPA_SC_BINNER_EVENT_CNTL_3) ||
+			reg == SOC15_REG_OFFSET(GC, 0, mmPA_SC_ENHANCE) ||
+			reg == SOC15_REG_OFFSET(GC, 0, mmPA_SC_ENHANCE_1) ||
+			reg == SOC15_REG_OFFSET(GC, 0, mmSH_MEM_CONFIG))
+			WREG32_RLC(reg, tmp);
+		else
+			(entry->hwip == GC_HWIP) ?
+				WREG32_SOC15_IP(GC, reg, tmp) : WREG32(reg, tmp);
+
+	}
+
+}
+
+static int soc15_asic_baco_reset(struct amdgpu_device *adev)
 {
-	u32 i;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+	int ret = 0;
 
-	dev_info(adev->dev, "GPU pci config reset\n");
+	/* avoid NBIF got stuck when do RAS recovery in BACO reset */
+	if (ras && adev->ras_enabled)
+		adev->nbio.funcs->enable_doorbell_interrupt(adev, false);
 
-	/* disable BM */
-	pci_clear_master(adev->pdev);
-	/* reset */
-	amdgpu_pci_config_reset(adev);
+	ret = amdgpu_dpm_baco_reset(adev);
+	if (ret)
+		return ret;
 
-	udelay(100);
+	/* re-enable doorbell interrupt after BACO exit */
+	if (ras && adev->ras_enabled)
+		adev->nbio.funcs->enable_doorbell_interrupt(adev, true);
 
-	/* wait for asic to come out of reset */
-	for (i = 0; i < adev->usec_timeout; i++) {
-		u32 memsize = (adev->flags & AMD_IS_APU) ?
-			nbio_v7_0_get_memsize(adev) :
-			nbio_v6_1_get_memsize(adev);
-		if (memsize != 0xffffffff)
-			break;
-		udelay(1);
+	return 0;
+}
+
+static enum amd_reset_method
+soc15_asic_reset_method(struct amdgpu_device *adev)
+{
+	int baco_reset = 0;
+	bool connected_to_cpu = false;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+
+        if (adev->gmc.xgmi.supported && adev->gmc.xgmi.connected_to_cpu)
+                connected_to_cpu = true;
+
+	if (amdgpu_reset_method == AMD_RESET_METHOD_MODE1 ||
+	    amdgpu_reset_method == AMD_RESET_METHOD_MODE2 ||
+	    amdgpu_reset_method == AMD_RESET_METHOD_BACO ||
+	    amdgpu_reset_method == AMD_RESET_METHOD_PCI) {
+		/* If connected to cpu, driver only support mode2 */
+                if (connected_to_cpu)
+                        return AMD_RESET_METHOD_MODE2;
+                return amdgpu_reset_method;
+        }
+
+	if (amdgpu_reset_method != -1)
+		dev_warn(adev->dev, "Specified reset method:%d isn't supported, using AUTO instead.\n",
+				  amdgpu_reset_method);
+
+	switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) {
+	case IP_VERSION(10, 0, 0):
+	case IP_VERSION(10, 0, 1):
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 0, 1):
+		return AMD_RESET_METHOD_MODE2;
+	case IP_VERSION(9, 0, 0):
+	case IP_VERSION(11, 0, 2):
+		if (adev->asic_type == CHIP_VEGA20) {
+			if (adev->psp.sos.fw_version >= 0x80067)
+				baco_reset = amdgpu_dpm_is_baco_supported(adev);
+			/*
+			 * 1. PMFW version > 0x284300: all cases use baco
+			 * 2. PMFW version <= 0x284300: only sGPU w/o RAS use baco
+			 */
+			if (ras && adev->ras_enabled &&
+			    adev->pm.fw_version <= 0x283400)
+				baco_reset = 0;
+		} else {
+			baco_reset = amdgpu_dpm_is_baco_supported(adev);
+		}
+		break;
+	case IP_VERSION(13, 0, 2):
+		 /*
+		 * 1.connected to cpu: driver issue mode2 reset
+		 * 2.discret gpu: driver issue mode1 reset
+		 */
+		if (connected_to_cpu)
+			return AMD_RESET_METHOD_MODE2;
+		break;
+	case IP_VERSION(13, 0, 6):
+	case IP_VERSION(13, 0, 14):
+	case IP_VERSION(13, 0, 12):
+		/* Use gpu_recovery param to target a reset method.
+		 * Enable triggering of GPU reset only if specified
+		 * by module parameter.
+		 */
+		if (adev->pcie_reset_ctx.in_link_reset)
+			return AMD_RESET_METHOD_LINK;
+		if (amdgpu_gpu_recovery == 4 || amdgpu_gpu_recovery == 5)
+			return AMD_RESET_METHOD_MODE2;
+		else if (!(adev->flags & AMD_IS_APU))
+			return AMD_RESET_METHOD_MODE1;
+		else
+			return AMD_RESET_METHOD_MODE2;
+	default:
+		break;
 	}
 
+	if (baco_reset)
+		return AMD_RESET_METHOD_BACO;
+	else
+		return AMD_RESET_METHOD_MODE1;
+}
+
+static bool soc15_need_reset_on_resume(struct amdgpu_device *adev)
+{
+	/* Will reset for the following suspend abort cases.
+	 * 1) S3 suspend aborted in the normal S3 suspend
+	 * 2) S3 suspend aborted in performing pm core test.
+	 */
+	if (adev->in_s3 && !pm_resume_via_firmware())
+		return true;
+	else
+		return false;
 }
 
 static int soc15_asic_reset(struct amdgpu_device *adev)
 {
-	amdgpu_atombios_scratch_regs_engine_hung(adev, true);
+	/* original raven doesn't have full asic reset */
+	/* On the latest Raven, the GPU reset can be performed
+	 * successfully. So now, temporarily enable it for the
+	 * S3 suspend abort case.
+	 */
 
-	soc15_gpu_pci_config_reset(adev);
+	if ((adev->apu_flags & AMD_APU_IS_PICASSO ||
+			!(adev->apu_flags & AMD_APU_IS_RAVEN)) &&
+			soc15_need_reset_on_resume(adev))
+		goto asic_reset;
 
-	amdgpu_atombios_scratch_regs_engine_hung(adev, false);
+	if ((adev->apu_flags & AMD_APU_IS_RAVEN) ||
+			(adev->apu_flags & AMD_APU_IS_RAVEN2))
+		return 0;
 
-	return 0;
+asic_reset:
+	switch (soc15_asic_reset_method(adev)) {
+	case AMD_RESET_METHOD_PCI:
+		dev_info(adev->dev, "PCI reset\n");
+		return amdgpu_device_pci_reset(adev);
+	case AMD_RESET_METHOD_BACO:
+		dev_info(adev->dev, "BACO reset\n");
+		return soc15_asic_baco_reset(adev);
+	case AMD_RESET_METHOD_MODE2:
+		dev_info(adev->dev, "MODE2 reset\n");
+		return amdgpu_dpm_mode2_reset(adev);
+	case AMD_RESET_METHOD_LINK:
+		dev_info(adev->dev, "Link reset\n");
+		return amdgpu_device_link_reset(adev);
+	default:
+		dev_info(adev->dev, "MODE1 reset\n");
+		return amdgpu_device_mode1_reset(adev);
+	}
+}
+
+static int soc15_supports_baco(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) {
+	case IP_VERSION(9, 0, 0):
+	case IP_VERSION(11, 0, 2):
+		if (adev->asic_type == CHIP_VEGA20) {
+			if (adev->psp.sos.fw_version >= 0x80067)
+				return amdgpu_dpm_is_baco_supported(adev);
+			return 0;
+		} else {
+			return amdgpu_dpm_is_baco_supported(adev);
+		}
+		break;
+	default:
+		return 0;
+	}
 }
 
 /*static int soc15_set_uvd_clock(struct amdgpu_device *adev, u32 clock,
@@ -469,45 +695,16 @@ static int soc15_set_vce_clocks(struct amdgpu_device *adev, u32 evclk, u32 ecclk
 	return 0;
 }
 
-static void soc15_pcie_gen3_enable(struct amdgpu_device *adev)
-{
-	if (pci_is_root_bus(adev->pdev->bus))
-		return;
-
-	if (amdgpu_pcie_gen2 == 0)
-		return;
-
-	if (adev->flags & AMD_IS_APU)
-		return;
-
-	if (!(adev->pm.pcie_gen_mask & (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2 |
-					CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3)))
-		return;
-
-	/* todo */
-}
-
 static void soc15_program_aspm(struct amdgpu_device *adev)
 {
-
-	if (amdgpu_aspm == 0)
+	if (!amdgpu_device_should_use_aspm(adev))
 		return;
 
-	/* todo */
+	if (adev->nbio.funcs->program_aspm)
+		adev->nbio.funcs->program_aspm(adev);
 }
 
-static void soc15_enable_doorbell_aperture(struct amdgpu_device *adev,
-					bool enable)
-{
-	if (adev->flags & AMD_IS_APU) {
-		nbio_v7_0_enable_doorbell_aperture(adev, enable);
-	} else {
-		nbio_v6_1_enable_doorbell_aperture(adev, enable);
-		nbio_v6_1_enable_doorbell_selfring_aperture(adev, enable);
-	}
-}
-
-static const struct amdgpu_ip_block_version vega10_common_ip_block =
+const struct amdgpu_ip_block_version vega10_common_ip_block =
 {
 	.type = AMD_IP_BLOCK_TYPE_COMMON,
 	.major = 2,
@@ -516,78 +713,269 @@ static const struct amdgpu_ip_block_version vega10_common_ip_block =
 	.funcs = &soc15_common_ip_funcs,
 };
 
-int soc15_set_ip_blocks(struct amdgpu_device *adev)
+static void soc15_reg_base_init(struct amdgpu_device *adev)
 {
-	nbio_v6_1_detect_hw_virt(adev);
-
-	if (amdgpu_sriov_vf(adev))
-		adev->virt.ops = &xgpu_ai_virt_ops;
-
+	/* Set IP register base before any HW register access */
 	switch (adev->asic_type) {
 	case CHIP_VEGA10:
-		amdgpu_ip_block_add(adev, &vega10_common_ip_block);
-		amdgpu_ip_block_add(adev, &gmc_v9_0_ip_block);
-		amdgpu_ip_block_add(adev, &vega10_ih_ip_block);
-		if (amdgpu_fw_load_type == 2 || amdgpu_fw_load_type == -1)
-			amdgpu_ip_block_add(adev, &psp_v3_1_ip_block);
-		if (!amdgpu_sriov_vf(adev))
-			amdgpu_ip_block_add(adev, &amdgpu_pp_ip_block);
-		if (adev->enable_virtual_display || amdgpu_sriov_vf(adev))
-			amdgpu_ip_block_add(adev, &dce_virtual_ip_block);
-		amdgpu_ip_block_add(adev, &gfx_v9_0_ip_block);
-		amdgpu_ip_block_add(adev, &sdma_v4_0_ip_block);
-		amdgpu_ip_block_add(adev, &uvd_v7_0_ip_block);
-		amdgpu_ip_block_add(adev, &vce_v4_0_ip_block);
-		break;
+	case CHIP_VEGA12:
 	case CHIP_RAVEN:
-		amdgpu_ip_block_add(adev, &vega10_common_ip_block);
-		amdgpu_ip_block_add(adev, &gmc_v9_0_ip_block);
-		amdgpu_ip_block_add(adev, &vega10_ih_ip_block);
-		amdgpu_ip_block_add(adev, &psp_v10_0_ip_block);
-		amdgpu_ip_block_add(adev, &amdgpu_pp_ip_block);
-		if (adev->enable_virtual_display || amdgpu_sriov_vf(adev))
-			amdgpu_ip_block_add(adev, &dce_virtual_ip_block);
-		amdgpu_ip_block_add(adev, &gfx_v9_0_ip_block);
-		amdgpu_ip_block_add(adev, &sdma_v4_0_ip_block);
-		amdgpu_ip_block_add(adev, &vcn_v1_0_ip_block);
+	case CHIP_RENOIR:
+		vega10_reg_base_init(adev);
+		break;
+	case CHIP_VEGA20:
+		vega20_reg_base_init(adev);
+		break;
+	case CHIP_ARCTURUS:
+		arct_reg_base_init(adev);
+		break;
+	case CHIP_ALDEBARAN:
+		aldebaran_reg_base_init(adev);
 		break;
 	default:
-		return -EINVAL;
+		DRM_ERROR("Unsupported asic type: %d!\n", adev->asic_type);
+		break;
 	}
+}
 
-	return 0;
+void soc15_set_virt_ops(struct amdgpu_device *adev)
+{
+	adev->virt.ops = &xgpu_ai_virt_ops;
+	/* init soc15 reg base early enough so we can
+	 * request request full access for sriov before
+	 * set_ip_blocks. */
+	soc15_reg_base_init(adev);
+}
+
+static bool soc15_need_full_reset(struct amdgpu_device *adev)
+{
+	/* change this when we implement soft reset */
+	return true;
 }
 
-static uint32_t soc15_get_rev_id(struct amdgpu_device *adev)
+static void soc15_get_pcie_usage(struct amdgpu_device *adev, uint64_t *count0,
+				 uint64_t *count1)
 {
+	uint32_t perfctr = 0;
+	uint64_t cnt0_of, cnt1_of;
+	int tmp;
+
+	/* This reports 0 on APUs, so return to avoid writing/reading registers
+	 * that may or may not be different from their GPU counterparts
+	 */
 	if (adev->flags & AMD_IS_APU)
-		return nbio_v7_0_get_rev_id(adev);
-	else
-		return nbio_v6_1_get_rev_id(adev);
+		return;
+
+	/* Set the 2 events that we wish to watch, defined above */
+	/* Reg 40 is # received msgs */
+	/* Reg 104 is # of posted requests sent */
+	perfctr = REG_SET_FIELD(perfctr, PCIE_PERF_CNTL_TXCLK, EVENT0_SEL, 40);
+	perfctr = REG_SET_FIELD(perfctr, PCIE_PERF_CNTL_TXCLK, EVENT1_SEL, 104);
+
+	/* Write to enable desired perf counters */
+	WREG32_PCIE(smnPCIE_PERF_CNTL_TXCLK, perfctr);
+	/* Zero out and enable the perf counters
+	 * Write 0x5:
+	 * Bit 0 = Start all counters(1)
+	 * Bit 2 = Global counter reset enable(1)
+	 */
+	WREG32_PCIE(smnPCIE_PERF_COUNT_CNTL, 0x00000005);
+
+	msleep(1000);
+
+	/* Load the shadow and disable the perf counters
+	 * Write 0x2:
+	 * Bit 0 = Stop counters(0)
+	 * Bit 1 = Load the shadow counters(1)
+	 */
+	WREG32_PCIE(smnPCIE_PERF_COUNT_CNTL, 0x00000002);
+
+	/* Read register values to get any >32bit overflow */
+	tmp = RREG32_PCIE(smnPCIE_PERF_CNTL_TXCLK);
+	cnt0_of = REG_GET_FIELD(tmp, PCIE_PERF_CNTL_TXCLK, COUNTER0_UPPER);
+	cnt1_of = REG_GET_FIELD(tmp, PCIE_PERF_CNTL_TXCLK, COUNTER1_UPPER);
+
+	/* Get the values and add the overflow */
+	*count0 = RREG32_PCIE(smnPCIE_PERF_COUNT0_TXCLK) | (cnt0_of << 32);
+	*count1 = RREG32_PCIE(smnPCIE_PERF_COUNT1_TXCLK) | (cnt1_of << 32);
+}
+
+static void vega20_get_pcie_usage(struct amdgpu_device *adev, uint64_t *count0,
+				 uint64_t *count1)
+{
+	uint32_t perfctr = 0;
+	uint64_t cnt0_of, cnt1_of;
+	int tmp;
+
+	/* This reports 0 on APUs, so return to avoid writing/reading registers
+	 * that may or may not be different from their GPU counterparts
+	 */
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	/* Set the 2 events that we wish to watch, defined above */
+	/* Reg 40 is # received msgs */
+	/* Reg 108 is # of posted requests sent on VG20 */
+	perfctr = REG_SET_FIELD(perfctr, PCIE_PERF_CNTL_TXCLK3,
+				EVENT0_SEL, 40);
+	perfctr = REG_SET_FIELD(perfctr, PCIE_PERF_CNTL_TXCLK3,
+				EVENT1_SEL, 108);
+
+	/* Write to enable desired perf counters */
+	WREG32_PCIE(smnPCIE_PERF_CNTL_TXCLK3, perfctr);
+	/* Zero out and enable the perf counters
+	 * Write 0x5:
+	 * Bit 0 = Start all counters(1)
+	 * Bit 2 = Global counter reset enable(1)
+	 */
+	WREG32_PCIE(smnPCIE_PERF_COUNT_CNTL, 0x00000005);
+
+	msleep(1000);
+
+	/* Load the shadow and disable the perf counters
+	 * Write 0x2:
+	 * Bit 0 = Stop counters(0)
+	 * Bit 1 = Load the shadow counters(1)
+	 */
+	WREG32_PCIE(smnPCIE_PERF_COUNT_CNTL, 0x00000002);
+
+	/* Read register values to get any >32bit overflow */
+	tmp = RREG32_PCIE(smnPCIE_PERF_CNTL_TXCLK3);
+	cnt0_of = REG_GET_FIELD(tmp, PCIE_PERF_CNTL_TXCLK3, COUNTER0_UPPER);
+	cnt1_of = REG_GET_FIELD(tmp, PCIE_PERF_CNTL_TXCLK3, COUNTER1_UPPER);
+
+	/* Get the values and add the overflow */
+	*count0 = RREG32_PCIE(smnPCIE_PERF_COUNT0_TXCLK3) | (cnt0_of << 32);
+	*count1 = RREG32_PCIE(smnPCIE_PERF_COUNT1_TXCLK3) | (cnt1_of << 32);
+}
+
+static bool soc15_need_reset_on_init(struct amdgpu_device *adev)
+{
+	u32 sol_reg;
+
+	/* CP hangs in IGT reloading test on RN, reset to WA */
+	if (adev->asic_type == CHIP_RENOIR)
+		return true;
+
+	if (amdgpu_gmc_need_reset_on_init(adev))
+		return true;
+	if (amdgpu_psp_tos_reload_needed(adev))
+		return true;
+	/* Just return false for soc15 GPUs.  Reset does not seem to
+	 * be necessary.
+	 */
+	if (!amdgpu_passthrough(adev))
+		return false;
+
+	if (adev->flags & AMD_IS_APU)
+		return false;
+
+	/* Check sOS sign of life register to confirm sys driver and sOS
+	 * are already been loaded.
+	 */
+	sol_reg = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_81);
+	if (sol_reg)
+		return true;
+
+	return false;
+}
+
+static uint64_t soc15_get_pcie_replay_count(struct amdgpu_device *adev)
+{
+	uint64_t nak_r, nak_g;
+
+	/* Get the number of NAKs received and generated */
+	nak_r = RREG32_PCIE(smnPCIE_RX_NUM_NAK);
+	nak_g = RREG32_PCIE(smnPCIE_RX_NUM_NAK_GENERATED);
+
+	/* Add the total number of NAKs, i.e the number of replays */
+	return (nak_r + nak_g);
+}
+
+static void soc15_pre_asic_init(struct amdgpu_device *adev)
+{
+	gmc_v9_0_restore_registers(adev);
 }
 
 static const struct amdgpu_asic_funcs soc15_asic_funcs =
 {
 	.read_disabled_bios = &soc15_read_disabled_bios,
-	.read_bios_from_rom = &soc15_read_bios_from_rom,
+	.read_bios_from_rom = &amdgpu_soc15_read_bios_from_rom,
+	.read_register = &soc15_read_register,
+	.reset = &soc15_asic_reset,
+	.reset_method = &soc15_asic_reset_method,
+	.get_xclk = &soc15_get_xclk,
+	.set_uvd_clocks = &soc15_set_uvd_clocks,
+	.set_vce_clocks = &soc15_set_vce_clocks,
+	.get_config_memsize = &soc15_get_config_memsize,
+	.need_full_reset = &soc15_need_full_reset,
+	.init_doorbell_index = &vega10_doorbell_index_init,
+	.get_pcie_usage = &soc15_get_pcie_usage,
+	.need_reset_on_init = &soc15_need_reset_on_init,
+	.get_pcie_replay_count = &soc15_get_pcie_replay_count,
+	.supports_baco = &soc15_supports_baco,
+	.pre_asic_init = &soc15_pre_asic_init,
+	.query_video_codecs = &soc15_query_video_codecs,
+};
+
+static const struct amdgpu_asic_funcs vega20_asic_funcs =
+{
+	.read_disabled_bios = &soc15_read_disabled_bios,
+	.read_bios_from_rom = &amdgpu_soc15_read_bios_from_rom,
+	.read_register = &soc15_read_register,
+	.reset = &soc15_asic_reset,
+	.reset_method = &soc15_asic_reset_method,
+	.get_xclk = &soc15_get_xclk,
+	.set_uvd_clocks = &soc15_set_uvd_clocks,
+	.set_vce_clocks = &soc15_set_vce_clocks,
+	.get_config_memsize = &soc15_get_config_memsize,
+	.need_full_reset = &soc15_need_full_reset,
+	.init_doorbell_index = &vega20_doorbell_index_init,
+	.get_pcie_usage = &vega20_get_pcie_usage,
+	.need_reset_on_init = &soc15_need_reset_on_init,
+	.get_pcie_replay_count = &soc15_get_pcie_replay_count,
+	.supports_baco = &soc15_supports_baco,
+	.pre_asic_init = &soc15_pre_asic_init,
+	.query_video_codecs = &soc15_query_video_codecs,
+};
+
+static const struct amdgpu_asic_funcs aqua_vanjaram_asic_funcs =
+{
+	.read_disabled_bios = &soc15_read_disabled_bios,
+	.read_bios_from_rom = &amdgpu_soc15_read_bios_from_rom,
 	.read_register = &soc15_read_register,
 	.reset = &soc15_asic_reset,
-	.set_vga_state = &soc15_vga_set_state,
+	.reset_method = &soc15_asic_reset_method,
 	.get_xclk = &soc15_get_xclk,
 	.set_uvd_clocks = &soc15_set_uvd_clocks,
 	.set_vce_clocks = &soc15_set_vce_clocks,
 	.get_config_memsize = &soc15_get_config_memsize,
+	.need_full_reset = &soc15_need_full_reset,
+	.init_doorbell_index = &aqua_vanjaram_doorbell_index_init,
+	.need_reset_on_init = &soc15_need_reset_on_init,
+	.get_pcie_replay_count = &amdgpu_nbio_get_pcie_replay_count,
+	.supports_baco = &soc15_supports_baco,
+	.pre_asic_init = &soc15_pre_asic_init,
+	.query_video_codecs = &soc15_query_video_codecs,
+	.encode_ext_smn_addressing = &aqua_vanjaram_encode_ext_smn_addressing,
+	.get_reg_state = &aqua_vanjaram_get_reg_state,
 };
 
-static int soc15_common_early_init(void *handle)
+static int soc15_common_early_init(struct amdgpu_ip_block *ip_block)
 {
-	bool psp_enabled = false;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
+	adev->nbio.funcs->set_reg_remap(adev);
 	adev->smc_rreg = NULL;
 	adev->smc_wreg = NULL;
-	adev->pcie_rreg = &soc15_pcie_rreg;
-	adev->pcie_wreg = &soc15_pcie_wreg;
+	adev->pcie_rreg = &amdgpu_device_indirect_rreg;
+	adev->pcie_wreg = &amdgpu_device_indirect_wreg;
+	adev->pcie_rreg_ext = &amdgpu_device_indirect_rreg_ext;
+	adev->pcie_wreg_ext = &amdgpu_device_indirect_wreg_ext;
+	adev->pcie_rreg64 = &amdgpu_device_indirect_rreg64;
+	adev->pcie_wreg64 = &amdgpu_device_indirect_wreg64;
+	adev->pcie_rreg64_ext = &amdgpu_device_indirect_rreg64_ext;
+	adev->pcie_wreg64_ext = &amdgpu_device_indirect_wreg64_ext;
 	adev->uvd_ctx_rreg = &soc15_uvd_ctx_rreg;
 	adev->uvd_ctx_wreg = &soc15_uvd_ctx_wreg;
 	adev->didt_rreg = &soc15_didt_rreg;
@@ -597,31 +985,14 @@ static int soc15_common_early_init(void *handle)
 	adev->se_cac_rreg = &soc15_se_cac_rreg;
 	adev->se_cac_wreg = &soc15_se_cac_wreg;
 
-	adev->asic_funcs = &soc15_asic_funcs;
-
-	if (amdgpu_get_ip_block(adev, AMD_IP_BLOCK_TYPE_PSP) &&
-		(amdgpu_ip_block_mask & (1 << AMD_IP_BLOCK_TYPE_PSP)))
-		psp_enabled = true;
-
-	/*
-	 * nbio need be used for both sdma and gfx9, but only
-	 * initializes once
-	 */
-	switch(adev->asic_type) {
-	case CHIP_VEGA10:
-		nbio_v6_1_init(adev);
-		break;
-	case CHIP_RAVEN:
-		nbio_v7_0_init(adev);
-		break;
-	default:
-		return -EINVAL;
-	}
-
-	adev->rev_id = soc15_get_rev_id(adev);
+	adev->rev_id = amdgpu_device_get_rev_id(adev);
 	adev->external_rev_id = 0xFF;
-	switch (adev->asic_type) {
-	case CHIP_VEGA10:
+	/* TODO: split the GC and PG flags based on the relevant IP version for which
+	 * they are relevant.
+	 */
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 0, 1):
+		adev->asic_funcs = &soc15_asic_funcs;
 		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
 			AMD_CG_SUPPORT_GFX_MGLS |
 			AMD_CG_SUPPORT_GFX_RLC_LS |
@@ -644,29 +1015,210 @@ static int soc15_common_early_init(void *handle)
 		adev->pg_flags = 0;
 		adev->external_rev_id = 0x1;
 		break;
-	case CHIP_RAVEN:
+	case IP_VERSION(9, 2, 1):
+		adev->asic_funcs = &soc15_asic_funcs;
 		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
 			AMD_CG_SUPPORT_GFX_MGLS |
-			AMD_CG_SUPPORT_GFX_RLC_LS |
-			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
 			AMD_CG_SUPPORT_GFX_3D_CGCG |
 			AMD_CG_SUPPORT_GFX_3D_CGLS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_BIF_MGCG |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_ROM_MGCG |
+			AMD_CG_SUPPORT_VCE_MGCG |
+			AMD_CG_SUPPORT_UVD_MGCG;
+		adev->pg_flags = 0;
+		adev->external_rev_id = adev->rev_id + 0x14;
+		break;
+	case IP_VERSION(9, 4, 0):
+		adev->asic_funcs = &vega20_asic_funcs;
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
 			AMD_CG_SUPPORT_GFX_CGCG |
 			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGLS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
 			AMD_CG_SUPPORT_BIF_MGCG |
 			AMD_CG_SUPPORT_BIF_LS |
 			AMD_CG_SUPPORT_HDP_MGCG |
 			AMD_CG_SUPPORT_HDP_LS |
-			AMD_CG_SUPPORT_DRM_MGCG |
-			AMD_CG_SUPPORT_DRM_LS |
 			AMD_CG_SUPPORT_ROM_MGCG |
+			AMD_CG_SUPPORT_VCE_MGCG |
+			AMD_CG_SUPPORT_UVD_MGCG;
+		adev->pg_flags = 0;
+		adev->external_rev_id = adev->rev_id + 0x28;
+		break;
+	case IP_VERSION(9, 1, 0):
+	case IP_VERSION(9, 2, 2):
+		adev->asic_funcs = &soc15_asic_funcs;
+
+		if (adev->rev_id >= 0x8)
+			adev->apu_flags |= AMD_APU_IS_RAVEN2;
+
+		if (adev->apu_flags & AMD_APU_IS_RAVEN2)
+			adev->external_rev_id = adev->rev_id + 0x79;
+		else if (adev->apu_flags & AMD_APU_IS_PICASSO)
+			adev->external_rev_id = adev->rev_id + 0x41;
+		else if (adev->rev_id == 1)
+			adev->external_rev_id = adev->rev_id + 0x20;
+		else
+			adev->external_rev_id = adev->rev_id + 0x01;
+
+		if (adev->apu_flags & AMD_APU_IS_RAVEN2) {
+			adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+				AMD_CG_SUPPORT_GFX_MGLS |
+				AMD_CG_SUPPORT_GFX_CP_LS |
+				AMD_CG_SUPPORT_GFX_3D_CGCG |
+				AMD_CG_SUPPORT_GFX_3D_CGLS |
+				AMD_CG_SUPPORT_GFX_CGCG |
+				AMD_CG_SUPPORT_GFX_CGLS |
+				AMD_CG_SUPPORT_BIF_LS |
+				AMD_CG_SUPPORT_HDP_LS |
+				AMD_CG_SUPPORT_MC_MGCG |
+				AMD_CG_SUPPORT_MC_LS |
+				AMD_CG_SUPPORT_SDMA_MGCG |
+				AMD_CG_SUPPORT_SDMA_LS |
+				AMD_CG_SUPPORT_VCN_MGCG;
+
+			adev->pg_flags = AMD_PG_SUPPORT_SDMA | AMD_PG_SUPPORT_VCN;
+		} else if (adev->apu_flags & AMD_APU_IS_PICASSO) {
+			adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+				AMD_CG_SUPPORT_GFX_MGLS |
+				AMD_CG_SUPPORT_GFX_CP_LS |
+				AMD_CG_SUPPORT_GFX_3D_CGLS |
+				AMD_CG_SUPPORT_GFX_CGCG |
+				AMD_CG_SUPPORT_GFX_CGLS |
+				AMD_CG_SUPPORT_BIF_LS |
+				AMD_CG_SUPPORT_HDP_LS |
+				AMD_CG_SUPPORT_MC_MGCG |
+				AMD_CG_SUPPORT_MC_LS |
+				AMD_CG_SUPPORT_SDMA_MGCG |
+				AMD_CG_SUPPORT_SDMA_LS |
+				AMD_CG_SUPPORT_VCN_MGCG;
+
+			/*
+			 * MMHUB PG needs to be disabled for Picasso for
+			 * stability reasons.
+			 */
+			adev->pg_flags = AMD_PG_SUPPORT_SDMA |
+				AMD_PG_SUPPORT_VCN;
+		} else {
+			adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+				AMD_CG_SUPPORT_GFX_MGLS |
+				AMD_CG_SUPPORT_GFX_RLC_LS |
+				AMD_CG_SUPPORT_GFX_CP_LS |
+				AMD_CG_SUPPORT_GFX_3D_CGLS |
+				AMD_CG_SUPPORT_GFX_CGCG |
+				AMD_CG_SUPPORT_GFX_CGLS |
+				AMD_CG_SUPPORT_BIF_MGCG |
+				AMD_CG_SUPPORT_BIF_LS |
+				AMD_CG_SUPPORT_HDP_MGCG |
+				AMD_CG_SUPPORT_HDP_LS |
+				AMD_CG_SUPPORT_DRM_MGCG |
+				AMD_CG_SUPPORT_DRM_LS |
+				AMD_CG_SUPPORT_MC_MGCG |
+				AMD_CG_SUPPORT_MC_LS |
+				AMD_CG_SUPPORT_SDMA_MGCG |
+				AMD_CG_SUPPORT_SDMA_LS |
+				AMD_CG_SUPPORT_VCN_MGCG;
+
+			adev->pg_flags = AMD_PG_SUPPORT_SDMA | AMD_PG_SUPPORT_VCN;
+		}
+		break;
+	case IP_VERSION(9, 4, 1):
+		adev->asic_funcs = &vega20_asic_funcs;
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
 			AMD_CG_SUPPORT_MC_MGCG |
 			AMD_CG_SUPPORT_MC_LS |
-			AMD_CG_SUPPORT_SDMA_MGCG |
-			AMD_CG_SUPPORT_SDMA_LS;
+			AMD_CG_SUPPORT_IH_CG |
+			AMD_CG_SUPPORT_VCN_MGCG |
+			AMD_CG_SUPPORT_JPEG_MGCG;
+		adev->pg_flags = AMD_PG_SUPPORT_VCN | AMD_PG_SUPPORT_VCN_DPG;
+		adev->external_rev_id = adev->rev_id + 0x32;
+		break;
+	case IP_VERSION(9, 3, 0):
+		adev->asic_funcs = &soc15_asic_funcs;
+
+		if (adev->apu_flags & AMD_APU_IS_RENOIR)
+			adev->external_rev_id = adev->rev_id + 0x91;
+		else
+			adev->external_rev_id = adev->rev_id + 0xa1;
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+				 AMD_CG_SUPPORT_GFX_MGLS |
+				 AMD_CG_SUPPORT_GFX_3D_CGCG |
+				 AMD_CG_SUPPORT_GFX_3D_CGLS |
+				 AMD_CG_SUPPORT_GFX_CGCG |
+				 AMD_CG_SUPPORT_GFX_CGLS |
+				 AMD_CG_SUPPORT_GFX_CP_LS |
+				 AMD_CG_SUPPORT_MC_MGCG |
+				 AMD_CG_SUPPORT_MC_LS |
+				 AMD_CG_SUPPORT_SDMA_MGCG |
+				 AMD_CG_SUPPORT_SDMA_LS |
+				 AMD_CG_SUPPORT_BIF_LS |
+				 AMD_CG_SUPPORT_HDP_LS |
+				 AMD_CG_SUPPORT_VCN_MGCG |
+				 AMD_CG_SUPPORT_JPEG_MGCG |
+				 AMD_CG_SUPPORT_IH_CG |
+				 AMD_CG_SUPPORT_ATHUB_LS |
+				 AMD_CG_SUPPORT_ATHUB_MGCG |
+				 AMD_CG_SUPPORT_DF_MGCG;
 		adev->pg_flags = AMD_PG_SUPPORT_SDMA |
-				 AMD_PG_SUPPORT_MMHUB;
-		adev->external_rev_id = 0x1;
+				 AMD_PG_SUPPORT_VCN |
+				 AMD_PG_SUPPORT_JPEG |
+				 AMD_PG_SUPPORT_VCN_DPG;
+		break;
+	case IP_VERSION(9, 4, 2):
+		adev->asic_funcs = &vega20_asic_funcs;
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_IH_CG |
+			AMD_CG_SUPPORT_VCN_MGCG | AMD_CG_SUPPORT_JPEG_MGCG;
+		adev->pg_flags = AMD_PG_SUPPORT_VCN_DPG;
+		adev->external_rev_id = adev->rev_id + 0x3c;
+		break;
+	case IP_VERSION(9, 4, 3):
+	case IP_VERSION(9, 4, 4):
+	case IP_VERSION(9, 5, 0):
+		adev->asic_funcs = &aqua_vanjaram_asic_funcs;
+		adev->cg_flags =
+			AMD_CG_SUPPORT_GFX_MGCG | AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS | AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_GFX_FGCG | AMD_CG_SUPPORT_REPEATER_FGCG |
+			AMD_CG_SUPPORT_VCN_MGCG | AMD_CG_SUPPORT_JPEG_MGCG |
+			AMD_CG_SUPPORT_IH_CG;
+		adev->pg_flags =
+			AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_VCN_DPG |
+			AMD_PG_SUPPORT_JPEG;
+		/*TODO: need a new external_rev_id for GC 9.4.4? */
+		adev->external_rev_id = adev->rev_id + 0x46;
+		if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0))
+			adev->external_rev_id = adev->rev_id + 0x50;
 		break;
 	default:
 		/* FIXME: not supported yet */
@@ -678,113 +1230,146 @@ static int soc15_common_early_init(void *handle)
 		xgpu_ai_mailbox_set_irq_funcs(adev);
 	}
 
-	adev->firmware.load_type = amdgpu_ucode_get_load_type(adev, amdgpu_fw_load_type);
-
-	amdgpu_get_pcie_info(adev);
-
 	return 0;
 }
 
-static int soc15_common_late_init(void *handle)
+static int soc15_common_late_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (amdgpu_sriov_vf(adev))
 		xgpu_ai_mailbox_get_irq(adev);
 
+	/* Enable selfring doorbell aperture late because doorbell BAR
+	 * aperture will change if resize BAR successfully in gmc sw_init.
+	 */
+	adev->nbio.funcs->enable_doorbell_selfring_aperture(adev, true);
+
 	return 0;
 }
 
-static int soc15_common_sw_init(void *handle)
+static int soc15_common_sw_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (amdgpu_sriov_vf(adev))
 		xgpu_ai_mailbox_add_irq_id(adev);
 
+	if (adev->df.funcs &&
+	    adev->df.funcs->sw_init)
+		adev->df.funcs->sw_init(adev);
+
 	return 0;
 }
 
-static int soc15_common_sw_fini(void *handle)
+static int soc15_common_sw_fini(struct amdgpu_ip_block *ip_block)
 {
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (adev->df.funcs &&
+	    adev->df.funcs->sw_fini)
+		adev->df.funcs->sw_fini(adev);
 	return 0;
 }
 
-static int soc15_common_hw_init(void *handle)
+static void soc15_sdma_doorbell_range_init(struct amdgpu_device *adev)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int i;
+
+	/* sdma doorbell range is programed by hypervisor */
+	if (!amdgpu_sriov_vf(adev)) {
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			adev->nbio.funcs->sdma_doorbell_range(adev, i,
+				true, adev->doorbell_index.sdma_engine[i] << 1,
+				adev->doorbell_index.sdma_doorbell_range);
+		}
+	}
+}
+
+static int soc15_common_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
 
-	/* move the golden regs per IP block */
-	soc15_init_golden_registers(adev);
-	/* enable pcie gen2/3 link */
-	soc15_pcie_gen3_enable(adev);
 	/* enable aspm */
 	soc15_program_aspm(adev);
 	/* setup nbio registers */
-	if (!(adev->flags & AMD_IS_APU))
-		nbio_v6_1_init_registers(adev);
+	adev->nbio.funcs->init_registers(adev);
+	/* remap HDP registers to a hole in mmio space,
+	 * for the purpose of expose those registers
+	 * to process space
+	 */
+	if (adev->nbio.funcs->remap_hdp_registers && !amdgpu_sriov_vf(adev))
+		adev->nbio.funcs->remap_hdp_registers(adev);
+
 	/* enable the doorbell aperture */
-	soc15_enable_doorbell_aperture(adev, true);
+	adev->nbio.funcs->enable_doorbell_aperture(adev, true);
+
+	/* HW doorbell routing policy: doorbell writing not
+	 * in SDMA/IH/MM/ACV range will be routed to CP. So
+	 * we need to init SDMA doorbell range prior
+	 * to CP ip block init and ring test.  IH already
+	 * happens before CP.
+	 */
+	soc15_sdma_doorbell_range_init(adev);
 
 	return 0;
 }
 
-static int soc15_common_hw_fini(void *handle)
+static int soc15_common_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	/* Disable the doorbell aperture and selfring doorbell aperture
+	 * separately in hw_fini because soc15_enable_doorbell_aperture
+	 * has been removed and there is no need to delay disabling
+	 * selfring doorbell.
+	 */
+	adev->nbio.funcs->enable_doorbell_aperture(adev, false);
+	adev->nbio.funcs->enable_doorbell_selfring_aperture(adev, false);
 
-	/* disable the doorbell aperture */
-	soc15_enable_doorbell_aperture(adev, false);
 	if (amdgpu_sriov_vf(adev))
 		xgpu_ai_mailbox_put_irq(adev);
 
+	/*
+	 * For minimal init, late_init is not called, hence RAS irqs are not
+	 * enabled.
+	 */
+	if ((!amdgpu_sriov_vf(adev)) &&
+	    (adev->init_lvl->level != AMDGPU_INIT_LEVEL_MINIMAL_XGMI) &&
+	    adev->nbio.ras_if &&
+	    amdgpu_ras_is_supported(adev, adev->nbio.ras_if->block)) {
+		if (adev->nbio.ras &&
+		    adev->nbio.ras->init_ras_controller_interrupt)
+			amdgpu_irq_put(adev, &adev->nbio.ras_controller_irq, 0);
+		if (adev->nbio.ras &&
+		    adev->nbio.ras->init_ras_err_event_athub_interrupt)
+			amdgpu_irq_put(adev, &adev->nbio.ras_err_event_athub_irq, 0);
+	}
+
 	return 0;
 }
 
-static int soc15_common_suspend(void *handle)
+static int soc15_common_suspend(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return soc15_common_hw_fini(adev);
+	return soc15_common_hw_fini(ip_block);
 }
 
-static int soc15_common_resume(void *handle)
+static int soc15_common_resume(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	return soc15_common_hw_init(adev);
+	if (soc15_need_reset_on_resume(adev)) {
+		dev_info(adev->dev, "S3 suspend abort case, let's reset ASIC.\n");
+		soc15_asic_reset(adev);
+	}
+	return soc15_common_hw_init(ip_block);
 }
 
-static bool soc15_common_is_idle(void *handle)
+static bool soc15_common_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	return true;
 }
 
-static int soc15_common_wait_for_idle(void *handle)
-{
-	return 0;
-}
-
-static int soc15_common_soft_reset(void *handle)
-{
-	return 0;
-}
-
-static void soc15_update_hdp_light_sleep(struct amdgpu_device *adev, bool enable)
-{
-	uint32_t def, data;
-
-	def = data = RREG32(SOC15_REG_OFFSET(HDP, 0, mmHDP_MEM_POWER_LS));
-
-	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS))
-		data |= HDP_MEM_POWER_LS__LS_ENABLE_MASK;
-	else
-		data &= ~HDP_MEM_POWER_LS__LS_ENABLE_MASK;
-
-	if (def != data)
-		WREG32(SOC15_REG_OFFSET(HDP, 0, mmHDP_MEM_POWER_LS), data);
-}
-
 static void soc15_update_drm_clock_gating(struct amdgpu_device *adev, bool enable)
 {
 	uint32_t def, data;
@@ -829,88 +1414,51 @@ static void soc15_update_drm_light_sleep(struct amdgpu_device *adev, bool enable
 		WREG32(SOC15_REG_OFFSET(MP0, 0, mmMP0_MISC_LIGHT_SLEEP_CTRL), data);
 }
 
-static void soc15_update_rom_medium_grain_clock_gating(struct amdgpu_device *adev,
-						       bool enable)
-{
-	uint32_t def, data;
-
-	def = data = RREG32(SOC15_REG_OFFSET(SMUIO, 0, mmCGTT_ROM_CLK_CTRL0));
-
-	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_ROM_MGCG))
-		data &= ~(CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK |
-			CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE1_MASK);
-	else
-		data |= CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK |
-			CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE1_MASK;
-
-	if (def != data)
-		WREG32(SOC15_REG_OFFSET(SMUIO, 0, mmCGTT_ROM_CLK_CTRL0), data);
-}
-
-static void soc15_update_df_medium_grain_clock_gating(struct amdgpu_device *adev,
-						       bool enable)
-{
-	uint32_t data;
-
-	/* Put DF on broadcast mode */
-	data = RREG32(SOC15_REG_OFFSET(DF, 0, mmFabricConfigAccessControl));
-	data &= ~FabricConfigAccessControl__CfgRegInstAccEn_MASK;
-	WREG32(SOC15_REG_OFFSET(DF, 0, mmFabricConfigAccessControl), data);
-
-	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_DF_MGCG)) {
-		data = RREG32(SOC15_REG_OFFSET(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater));
-		data &= ~DF_PIE_AON0_DfGlobalClkGater__MGCGMode_MASK;
-		data |= DF_MGCG_ENABLE_15_CYCLE_DELAY;
-		WREG32(SOC15_REG_OFFSET(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater), data);
-	} else {
-		data = RREG32(SOC15_REG_OFFSET(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater));
-		data &= ~DF_PIE_AON0_DfGlobalClkGater__MGCGMode_MASK;
-		data |= DF_MGCG_DISABLE;
-		WREG32(SOC15_REG_OFFSET(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater), data);
-	}
-
-	WREG32(SOC15_REG_OFFSET(DF, 0, mmFabricConfigAccessControl),
-	       mmFabricConfigAccessControl_DEFAULT);
-}
-
-static int soc15_common_set_clockgating_state(void *handle,
+static int soc15_common_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					    enum amd_clockgating_state state)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (amdgpu_sriov_vf(adev))
 		return 0;
 
-	switch (adev->asic_type) {
-	case CHIP_VEGA10:
-		nbio_v6_1_update_medium_grain_clock_gating(adev,
-				state == AMD_CG_STATE_GATE ? true : false);
-		nbio_v6_1_update_medium_grain_light_sleep(adev,
-				state == AMD_CG_STATE_GATE ? true : false);
-		soc15_update_hdp_light_sleep(adev,
-				state == AMD_CG_STATE_GATE ? true : false);
+	switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) {
+	case IP_VERSION(6, 1, 0):
+	case IP_VERSION(6, 2, 0):
+	case IP_VERSION(7, 4, 0):
+		adev->nbio.funcs->update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		adev->nbio.funcs->update_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+		adev->hdp.funcs->update_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
 		soc15_update_drm_clock_gating(adev,
-				state == AMD_CG_STATE_GATE ? true : false);
+				state == AMD_CG_STATE_GATE);
 		soc15_update_drm_light_sleep(adev,
-				state == AMD_CG_STATE_GATE ? true : false);
-		soc15_update_rom_medium_grain_clock_gating(adev,
-				state == AMD_CG_STATE_GATE ? true : false);
-		soc15_update_df_medium_grain_clock_gating(adev,
-				state == AMD_CG_STATE_GATE ? true : false);
+				state == AMD_CG_STATE_GATE);
+		adev->smuio.funcs->update_rom_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		adev->df.funcs->update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
 		break;
-	case CHIP_RAVEN:
-		nbio_v7_0_update_medium_grain_clock_gating(adev,
-				state == AMD_CG_STATE_GATE ? true : false);
-		nbio_v6_1_update_medium_grain_light_sleep(adev,
-				state == AMD_CG_STATE_GATE ? true : false);
-		soc15_update_hdp_light_sleep(adev,
-				state == AMD_CG_STATE_GATE ? true : false);
+	case IP_VERSION(7, 0, 0):
+	case IP_VERSION(7, 0, 1):
+	case IP_VERSION(2, 5, 0):
+		adev->nbio.funcs->update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		adev->nbio.funcs->update_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+		adev->hdp.funcs->update_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
 		soc15_update_drm_clock_gating(adev,
-				state == AMD_CG_STATE_GATE ? true : false);
+				state == AMD_CG_STATE_GATE);
 		soc15_update_drm_light_sleep(adev,
-				state == AMD_CG_STATE_GATE ? true : false);
-		soc15_update_rom_medium_grain_clock_gating(adev,
-				state == AMD_CG_STATE_GATE ? true : false);
+				state == AMD_CG_STATE_GATE);
+		break;
+	case IP_VERSION(7, 4, 1):
+	case IP_VERSION(7, 4, 4):
+		adev->hdp.funcs->update_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
 		break;
 	default:
 		break;
@@ -918,50 +1466,51 @@ static int soc15_common_set_clockgating_state(void *handle,
 	return 0;
 }
 
-static void soc15_common_get_clockgating_state(void *handle, u32 *flags)
+static void soc15_common_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int data;
 
 	if (amdgpu_sriov_vf(adev))
 		*flags = 0;
 
-	nbio_v6_1_get_clockgating_state(adev, flags);
-
-	/* AMD_CG_SUPPORT_HDP_LS */
-	data = RREG32(SOC15_REG_OFFSET(HDP, 0, mmHDP_MEM_POWER_LS));
-	if (data & HDP_MEM_POWER_LS__LS_ENABLE_MASK)
-		*flags |= AMD_CG_SUPPORT_HDP_LS;
-
-	/* AMD_CG_SUPPORT_DRM_MGCG */
-	data = RREG32(SOC15_REG_OFFSET(MP0, 0, mmMP0_MISC_CGTT_CTRL0));
-	if (!(data & 0x01000000))
-		*flags |= AMD_CG_SUPPORT_DRM_MGCG;
-
-	/* AMD_CG_SUPPORT_DRM_LS */
-	data = RREG32(SOC15_REG_OFFSET(MP0, 0, mmMP0_MISC_LIGHT_SLEEP_CTRL));
-	if (data & 0x1)
-		*flags |= AMD_CG_SUPPORT_DRM_LS;
+	if (adev->nbio.funcs && adev->nbio.funcs->get_clockgating_state)
+		adev->nbio.funcs->get_clockgating_state(adev, flags);
+
+	if (adev->hdp.funcs && adev->hdp.funcs->get_clock_gating_state)
+		adev->hdp.funcs->get_clock_gating_state(adev, flags);
+
+	if ((amdgpu_ip_version(adev, MP0_HWIP, 0) != IP_VERSION(13, 0, 2)) &&
+	    (amdgpu_ip_version(adev, MP0_HWIP, 0) != IP_VERSION(13, 0, 6)) &&
+	    (amdgpu_ip_version(adev, MP0_HWIP, 0) != IP_VERSION(13, 0, 12)) &&
+	    (amdgpu_ip_version(adev, MP0_HWIP, 0) != IP_VERSION(13, 0, 14))) {
+		/* AMD_CG_SUPPORT_DRM_MGCG */
+		data = RREG32(SOC15_REG_OFFSET(MP0, 0, mmMP0_MISC_CGTT_CTRL0));
+		if (!(data & 0x01000000))
+			*flags |= AMD_CG_SUPPORT_DRM_MGCG;
+
+		/* AMD_CG_SUPPORT_DRM_LS */
+		data = RREG32(SOC15_REG_OFFSET(MP0, 0, mmMP0_MISC_LIGHT_SLEEP_CTRL));
+		if (data & 0x1)
+			*flags |= AMD_CG_SUPPORT_DRM_LS;
+	}
 
 	/* AMD_CG_SUPPORT_ROM_MGCG */
-	data = RREG32(SOC15_REG_OFFSET(SMUIO, 0, mmCGTT_ROM_CLK_CTRL0));
-	if (!(data & CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK))
-		*flags |= AMD_CG_SUPPORT_ROM_MGCG;
+	if (adev->smuio.funcs && adev->smuio.funcs->get_clock_gating_state)
+		adev->smuio.funcs->get_clock_gating_state(adev, flags);
 
-	/* AMD_CG_SUPPORT_DF_MGCG */
-	data = RREG32(SOC15_REG_OFFSET(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater));
-	if (data & DF_MGCG_ENABLE_15_CYCLE_DELAY)
-		*flags |= AMD_CG_SUPPORT_DF_MGCG;
+	if (adev->df.funcs && adev->df.funcs->get_clockgating_state)
+		adev->df.funcs->get_clockgating_state(adev, flags);
 }
 
-static int soc15_common_set_powergating_state(void *handle,
+static int soc15_common_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					    enum amd_powergating_state state)
 {
 	/* todo */
 	return 0;
 }
 
-const struct amd_ip_funcs soc15_common_ip_funcs = {
+static const struct amd_ip_funcs soc15_common_ip_funcs = {
 	.name = "soc15_common",
 	.early_init = soc15_common_early_init,
 	.late_init = soc15_common_late_init,
@@ -972,8 +1521,6 @@ const struct amd_ip_funcs soc15_common_ip_funcs = {
 	.suspend = soc15_common_suspend,
 	.resume = soc15_common_resume,
 	.is_idle = soc15_common_is_idle,
-	.wait_for_idle = soc15_common_wait_for_idle,
-	.soft_reset = soc15_common_soft_reset,
 	.set_clockgating_state = soc15_common_set_clockgating_state,
 	.set_powergating_state = soc15_common_set_powergating_state,
 	.get_clockgating_state= soc15_common_get_clockgating_state,
diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.h b/drivers/gpu/drm/amd/amdgpu/soc15.h
index acb3cdb119f2..c8ac11a9cdef 100644
--- a/drivers/gpu/drm/amd/amdgpu/soc15.h
+++ b/drivers/gpu/drm/amd/amdgpu/soc15.h
@@ -26,11 +26,105 @@
 
 #include "nbio_v6_1.h"
 #include "nbio_v7_0.h"
+#include "nbio_v7_4.h"
+#include "amdgpu_reg_state.h"
 
-extern const struct amd_ip_funcs soc15_common_ip_funcs;
+extern const struct amdgpu_ip_block_version vega10_common_ip_block;
+
+#define SOC15_FLUSH_GPU_TLB_NUM_WREG		6
+#define SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT	3
+
+struct soc15_reg_golden {
+	u32	hwip;
+	u32	instance;
+	u32	segment;
+	u32	reg;
+	u32	and_mask;
+	u32	or_mask;
+};
+
+struct soc15_reg_rlcg {
+	u32	hwip;
+	u32	instance;
+	u32	segment;
+	u32	reg;
+};
+
+struct soc15_reg {
+	uint32_t hwip;
+	uint32_t inst;
+	uint32_t seg;
+	uint32_t reg_offset;
+};
+
+struct soc15_reg_entry {
+	uint32_t hwip;
+	uint32_t inst;
+	uint32_t seg;
+	uint32_t reg_offset;
+	uint32_t reg_value;
+	uint32_t se_num;
+	uint32_t instance;
+};
+
+struct soc15_allowed_register_entry {
+	uint32_t hwip;
+	uint32_t inst;
+	uint32_t seg;
+	uint32_t reg_offset;
+	bool grbm_indexed;
+};
+
+struct soc15_ras_field_entry {
+	const char *name;
+	uint32_t hwip;
+	uint32_t inst;
+	uint32_t seg;
+	uint32_t reg_offset;
+	uint32_t sec_count_mask;
+	uint32_t sec_count_shift;
+	uint32_t ded_count_mask;
+	uint32_t ded_count_shift;
+};
+
+#define SOC15_REG_ENTRY(ip, inst, reg)	ip##_HWIP, inst, reg##_BASE_IDX, reg
+#define SOC15_REG_ENTRY_STR(ip, inst, reg) \
+	{ ip##_HWIP, inst, reg##_BASE_IDX, reg, #reg }
+
+#define SOC15_REG_ENTRY_OFFSET(entry)	(adev->reg_offset[entry.hwip][entry.inst][entry.seg] + entry.reg_offset)
+
+/* Over ride the instance id */
+#define SOC15_REG_ENTRY_OFFSET_INST(entry, inst) \
+	(adev->reg_offset[entry.hwip][inst][entry.seg] + entry.reg_offset)
+
+#define SOC15_REG_GOLDEN_VALUE(ip, inst, reg, and_mask, or_mask) \
+	{ ip##_HWIP, inst, reg##_BASE_IDX, reg, and_mask, or_mask }
+
+#define SOC15_REG_FIELD(reg, field) reg##__##field##_MASK, reg##__##field##__SHIFT
+
+#define SOC15_REG_FIELD_VAL(val, mask, shift)	(((val) & mask) >> shift)
+
+#define SOC15_RAS_REG_FIELD_VAL(val, entry, field) SOC15_REG_FIELD_VAL((val), (entry).field##_count_mask, (entry).field##_count_shift)
 
 void soc15_grbm_select(struct amdgpu_device *adev,
-		    u32 me, u32 pipe, u32 queue, u32 vmid);
-int soc15_set_ip_blocks(struct amdgpu_device *adev);
+		    u32 me, u32 pipe, u32 queue, u32 vmid, int xcc_id);
+void soc15_set_virt_ops(struct amdgpu_device *adev);
+
+void soc15_program_register_sequence(struct amdgpu_device *adev,
+					     const struct soc15_reg_golden *registers,
+					     const u32 array_size);
+
+int vega10_reg_base_init(struct amdgpu_device *adev);
+int vega20_reg_base_init(struct amdgpu_device *adev);
+int arct_reg_base_init(struct amdgpu_device *adev);
+int aldebaran_reg_base_init(struct amdgpu_device *adev);
+u64 aqua_vanjaram_encode_ext_smn_addressing(int ext_id);
+int aqua_vanjaram_init_soc_config(struct amdgpu_device *adev);
+ssize_t aqua_vanjaram_get_reg_state(struct amdgpu_device *adev,
+				    enum amdgpu_reg_state reg_state, void *buf,
+				    size_t max_size);
 
+void vega10_doorbell_index_init(struct amdgpu_device *adev);
+void vega20_doorbell_index_init(struct amdgpu_device *adev);
+void aqua_vanjaram_doorbell_index_init(struct amdgpu_device *adev);
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/soc15_common.h b/drivers/gpu/drm/amd/amdgpu/soc15_common.h
index 7a8e4e28abb2..242b24f73c17 100644
--- a/drivers/gpu/drm/amd/amdgpu/soc15_common.h
+++ b/drivers/gpu/drm/amd/amdgpu/soc15_common.h
@@ -24,73 +24,190 @@
 #ifndef __SOC15_COMMON_H__
 #define __SOC15_COMMON_H__
 
-struct nbio_hdp_flush_reg {
-	u32 hdp_flush_req_offset;
-	u32 hdp_flush_done_offset;
-	u32 ref_and_mask_cp0;
-	u32 ref_and_mask_cp1;
-	u32 ref_and_mask_cp2;
-	u32 ref_and_mask_cp3;
-	u32 ref_and_mask_cp4;
-	u32 ref_and_mask_cp5;
-	u32 ref_and_mask_cp6;
-	u32 ref_and_mask_cp7;
-	u32 ref_and_mask_cp8;
-	u32 ref_and_mask_cp9;
-	u32 ref_and_mask_sdma0;
-	u32 ref_and_mask_sdma1;
-};
-
-struct nbio_pcie_index_data {
-	u32 index_offset;
-	u32 data_offset;
-};
+/* GET_INST returns the physical instance corresponding to a logical instance */
+#define GET_INST(ip, inst) \
+	(adev->ip_map.logical_to_dev_inst ? \
+	adev->ip_map.logical_to_dev_inst(adev, ip##_HWIP, inst) : inst)
+#define GET_MASK(ip, mask) \
+	(adev->ip_map.logical_to_dev_mask ? \
+	adev->ip_map.logical_to_dev_mask(adev, ip##_HWIP, mask) : mask)
 
 /* Register Access Macros */
-#define SOC15_REG_OFFSET(ip, inst, reg)       (0 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG0 + reg : \
-                                                (1 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG1 + reg : \
-                                                    (2 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG2 + reg : \
-                                                        (3 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG3 + reg : \
-                                                            (ip##_BASE__INST##inst##_SEG4 + reg)))))
+#define SOC15_REG_OFFSET(ip, inst, reg)	(adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg)
+#define SOC15_REG_OFFSET1(ip, inst, reg, offset) \
+	(adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + (reg)+(offset))
+
+#define __WREG32_SOC15_RLC__(reg, value, flag, hwip, inst) \
+	((amdgpu_sriov_vf(adev) && adev->gfx.rlc.funcs && adev->gfx.rlc.rlcg_reg_access_supported) ? \
+	 amdgpu_sriov_wreg(adev, reg, value, flag, hwip, inst) : \
+	 WREG32(reg, value))
+
+#define __RREG32_SOC15_RLC__(reg, flag, hwip, inst) \
+	((amdgpu_sriov_vf(adev) && adev->gfx.rlc.funcs && adev->gfx.rlc.rlcg_reg_access_supported) ? \
+	 amdgpu_sriov_rreg(adev, reg, flag, hwip, inst) : \
+	 RREG32(reg))
 
 #define WREG32_FIELD15(ip, idx, reg, field, val)	\
-	WREG32(SOC15_REG_OFFSET(ip, idx, mm##reg), (RREG32(SOC15_REG_OFFSET(ip, idx, mm##reg)) & ~REG_FIELD_MASK(reg, field)) | (val) << REG_FIELD_SHIFT(reg, field))
+	 __WREG32_SOC15_RLC__(adev->reg_offset[ip##_HWIP][idx][mm##reg##_BASE_IDX] + mm##reg,	\
+				(__RREG32_SOC15_RLC__( \
+					adev->reg_offset[ip##_HWIP][idx][mm##reg##_BASE_IDX] + mm##reg, \
+					0, ip##_HWIP, idx) & \
+				~REG_FIELD_MASK(reg, field)) | (val) << REG_FIELD_SHIFT(reg, field), \
+			      0, ip##_HWIP, idx)
+
+#define WREG32_FIELD15_PREREG(ip, idx, reg_name, field, val)        \
+	__WREG32_SOC15_RLC__(adev->reg_offset[ip##_HWIP][idx][reg##reg_name##_BASE_IDX] + reg##reg_name,   \
+			(__RREG32_SOC15_RLC__( \
+					adev->reg_offset[ip##_HWIP][idx][reg##reg_name##_BASE_IDX] + reg##reg_name, \
+					0, ip##_HWIP, idx) & \
+					~REG_FIELD_MASK(reg_name, field)) | (val) << REG_FIELD_SHIFT(reg_name, field), \
+			0, ip##_HWIP, idx)
 
 #define RREG32_SOC15(ip, inst, reg) \
-	RREG32( (0 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG0 + reg : \
-		(1 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG1 + reg : \
-		(2 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG2 + reg : \
-		(3 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG3 + reg : \
-		(ip##_BASE__INST##inst##_SEG4 + reg))))))
+	__RREG32_SOC15_RLC__(adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg, \
+			 0, ip##_HWIP, inst)
+
+#define RREG32_SOC15_IP(ip, reg) __RREG32_SOC15_RLC__(reg, 0, ip##_HWIP, 0)
+
+#define RREG32_SOC15_IP_NO_KIQ(ip, reg, inst) __RREG32_SOC15_RLC__(reg, AMDGPU_REGS_NO_KIQ, ip##_HWIP, inst)
+
+#define RREG32_SOC15_NO_KIQ(ip, inst, reg) \
+	__RREG32_SOC15_RLC__(adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg, \
+			 AMDGPU_REGS_NO_KIQ, ip##_HWIP, inst)
 
 #define RREG32_SOC15_OFFSET(ip, inst, reg, offset) \
-	RREG32( (0 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG0 + reg : \
-		(1 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG1 + reg : \
-		(2 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG2 + reg : \
-		(3 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG3 + reg : \
-		(ip##_BASE__INST##inst##_SEG4 + reg))))) + offset)
+	 __RREG32_SOC15_RLC__((adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + (reg)) + \
+			 (offset), 0, ip##_HWIP, inst)
 
 #define WREG32_SOC15(ip, inst, reg, value) \
-	WREG32( (0 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG0 + reg : \
-		(1 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG1 + reg : \
-		(2 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG2 + reg : \
-		(3 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG3 + reg : \
-		(ip##_BASE__INST##inst##_SEG4 + reg))))), value)
+	 __WREG32_SOC15_RLC__((adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg), \
+			  value, 0, ip##_HWIP, inst)
+
+#define WREG32_SOC15_IP(ip, reg, value) \
+	 __WREG32_SOC15_RLC__(reg, value, 0, ip##_HWIP, 0)
+
+#define WREG32_SOC15_IP_NO_KIQ(ip, reg, value, inst) \
+	 __WREG32_SOC15_RLC__(reg, value, AMDGPU_REGS_NO_KIQ, ip##_HWIP, inst)
 
 #define WREG32_SOC15_NO_KIQ(ip, inst, reg, value) \
-	WREG32_NO_KIQ( (0 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG0 + reg : \
-		(1 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG1 + reg : \
-		(2 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG2 + reg : \
-		(3 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG3 + reg : \
-		(ip##_BASE__INST##inst##_SEG4 + reg))))), value)
+	__WREG32_SOC15_RLC__(adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg, \
+			     value, AMDGPU_REGS_NO_KIQ, ip##_HWIP, inst)
 
 #define WREG32_SOC15_OFFSET(ip, inst, reg, offset, value) \
-	WREG32( (0 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG0 + reg : \
-		(1 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG1 + reg : \
-		(2 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG2 + reg : \
-		(3 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG3 + reg : \
-		(ip##_BASE__INST##inst##_SEG4 + reg))))) + offset, value)
+	 __WREG32_SOC15_RLC__((adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg) + offset, \
+			  value, 0, ip##_HWIP, inst)
 
-#endif
+#define SOC15_WAIT_ON_RREG(ip, inst, reg, expected_value, mask)      \
+	amdgpu_device_wait_on_rreg(adev, inst,                       \
+	(adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + (reg)), \
+	#reg, expected_value, mask)
+
+#define SOC15_WAIT_ON_RREG_OFFSET(ip, inst, reg, offset, expected_value, mask)  \
+	amdgpu_device_wait_on_rreg(adev, inst,                                  \
+	(adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + (reg) + (offset)), \
+	#reg, expected_value, mask)
+
+#define WREG32_RLC(reg, value) \
+	__WREG32_SOC15_RLC__(reg, value, AMDGPU_REGS_RLC, GC_HWIP, 0)
+
+#define WREG32_RLC_EX(prefix, reg, value, inst) \
+	do {							\
+		if (amdgpu_sriov_fullaccess(adev)) {    \
+			uint32_t i = 0;	\
+			uint32_t retries = 50000;	\
+			uint32_t r0 = adev->reg_offset[GC_HWIP][inst][prefix##SCRATCH_REG0_BASE_IDX] + prefix##SCRATCH_REG0;	\
+			uint32_t r1 = adev->reg_offset[GC_HWIP][inst][prefix##SCRATCH_REG1_BASE_IDX] + prefix##SCRATCH_REG1;	\
+			uint32_t spare_int = adev->reg_offset[GC_HWIP][inst][prefix##RLC_SPARE_INT_BASE_IDX] + prefix##RLC_SPARE_INT;	\
+			WREG32(r0, value);	\
+			WREG32(r1, (reg | 0x80000000));	\
+			WREG32(spare_int, 0x1);	\
+			for (i = 0; i < retries; i++) {	\
+				u32 tmp = RREG32(r1);	\
+				if (!(tmp & 0x80000000))	\
+					break;	\
+				udelay(10);	\
+			}	\
+			if (i >= retries)	\
+				pr_err("timeout: rlcg program reg:0x%05x failed !\n", reg);	\
+		} else {	\
+			WREG32(reg, value); \
+		}	\
+	} while (0)
+
+/* shadow the registers in the callback function */
+#define WREG32_SOC15_RLC_SHADOW(ip, inst, reg, value) \
+	__WREG32_SOC15_RLC__((adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg), value, AMDGPU_REGS_RLC, GC_HWIP, inst)
+
+/* for GC only */
+#define RREG32_RLC(reg) \
+	__RREG32_SOC15_RLC__(reg, AMDGPU_REGS_RLC, GC_HWIP, 0)
 
+#define WREG32_RLC_NO_KIQ(reg, value, hwip) \
+	__WREG32_SOC15_RLC__(reg, value, AMDGPU_REGS_NO_KIQ | AMDGPU_REGS_RLC, hwip, 0)
 
+#define RREG32_RLC_NO_KIQ(reg, hwip) \
+	__RREG32_SOC15_RLC__(reg, AMDGPU_REGS_NO_KIQ | AMDGPU_REGS_RLC, hwip, 0)
+
+#define WREG32_SOC15_RLC_SHADOW_EX(prefix, ip, inst, reg, value) \
+	do {							\
+		uint32_t target_reg = adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg;\
+		if (amdgpu_sriov_fullaccess(adev)) {    \
+			uint32_t r2 = adev->reg_offset[GC_HWIP][inst][prefix##SCRATCH_REG1_BASE_IDX] + prefix##SCRATCH_REG2;	\
+			uint32_t r3 = adev->reg_offset[GC_HWIP][inst][prefix##SCRATCH_REG1_BASE_IDX] + prefix##SCRATCH_REG3;	\
+			uint32_t grbm_cntl = adev->reg_offset[GC_HWIP][inst][prefix##GRBM_GFX_CNTL_BASE_IDX] + prefix##GRBM_GFX_CNTL;   \
+			uint32_t grbm_idx = adev->reg_offset[GC_HWIP][inst][prefix##GRBM_GFX_INDEX_BASE_IDX] + prefix##GRBM_GFX_INDEX;   \
+			if (target_reg == grbm_cntl) \
+				WREG32(r2, value);	\
+			else if (target_reg == grbm_idx) \
+				WREG32(r3, value);	\
+			WREG32(target_reg, value);	\
+		} else {	\
+			WREG32(target_reg, value); \
+		}	\
+	} while (0)
+
+#define RREG32_SOC15_RLC(ip, inst, reg) \
+	__RREG32_SOC15_RLC__(adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg, AMDGPU_REGS_RLC, ip##_HWIP, inst)
+
+#define WREG32_SOC15_RLC(ip, inst, reg, value) \
+	do {							\
+		uint32_t target_reg = adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg;\
+		__WREG32_SOC15_RLC__(target_reg, value, AMDGPU_REGS_RLC, ip##_HWIP, inst); \
+	} while (0)
+
+#define WREG32_SOC15_RLC_EX(prefix, ip, inst, reg, value) \
+	do {							\
+			uint32_t target_reg = adev->reg_offset[GC_HWIP][inst][reg##_BASE_IDX] + reg;\
+			WREG32_RLC_EX(prefix, target_reg, value, inst); \
+	} while (0)
+
+#define WREG32_FIELD15_RLC(ip, idx, reg, field, val)   \
+	__WREG32_SOC15_RLC__((adev->reg_offset[ip##_HWIP][idx][mm##reg##_BASE_IDX] + mm##reg), \
+			     (__RREG32_SOC15_RLC__(adev->reg_offset[ip##_HWIP][idx][mm##reg##_BASE_IDX] + mm##reg, \
+						   AMDGPU_REGS_RLC, ip##_HWIP, idx) & \
+			      ~REG_FIELD_MASK(reg, field)) | (val) << REG_FIELD_SHIFT(reg, field), \
+			     AMDGPU_REGS_RLC, ip##_HWIP, idx)
+
+#define WREG32_SOC15_OFFSET_RLC(ip, inst, reg, offset, value) \
+	__WREG32_SOC15_RLC__((adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg) + offset, value, AMDGPU_REGS_RLC, ip##_HWIP, inst)
+
+#define RREG32_SOC15_OFFSET_RLC(ip, inst, reg, offset) \
+	__RREG32_SOC15_RLC__((adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg) + offset, AMDGPU_REGS_RLC, ip##_HWIP, inst)
+
+/* inst equals to ext for some IPs */
+#define RREG32_SOC15_EXT(ip, inst, reg, ext) \
+	RREG32_PCIE_EXT((adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg) * 4 \
+			+ adev->asic_funcs->encode_ext_smn_addressing(ext)) \
+
+#define WREG32_SOC15_EXT(ip, inst, reg, ext, value) \
+	WREG32_PCIE_EXT((adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg) * 4 \
+			+ adev->asic_funcs->encode_ext_smn_addressing(ext), \
+			value) \
+
+#define RREG64_MCA(ext, mca_base, idx) \
+	RREG64_PCIE_EXT(adev->asic_funcs->encode_ext_smn_addressing(ext) + mca_base + (idx * 8))
+
+#define WREG64_MCA(ext, mca_base, idx, val) \
+	WREG64_PCIE_EXT(adev->asic_funcs->encode_ext_smn_addressing(ext) + mca_base + (idx * 8), val)
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/soc15d.h b/drivers/gpu/drm/amd/amdgpu/soc15d.h
index 7f408f85fdb6..cf93fa477674 100644
--- a/drivers/gpu/drm/amd/amdgpu/soc15d.h
+++ b/drivers/gpu/drm/amd/amdgpu/soc15d.h
@@ -53,6 +53,35 @@
 
 #define PACKET3_COMPUTE(op, n) (PACKET3(op, n) | 1 << 1)
 
+#define	PACKETJ_CONDITION_CHECK0	0
+#define	PACKETJ_CONDITION_CHECK1	1
+#define	PACKETJ_CONDITION_CHECK2	2
+#define	PACKETJ_CONDITION_CHECK3	3
+#define	PACKETJ_CONDITION_CHECK4	4
+#define	PACKETJ_CONDITION_CHECK5	5
+#define	PACKETJ_CONDITION_CHECK6	6
+#define	PACKETJ_CONDITION_CHECK7	7
+
+#define	PACKETJ_TYPE0	0
+#define	PACKETJ_TYPE1	1
+#define	PACKETJ_TYPE2	2
+#define	PACKETJ_TYPE3	3
+#define	PACKETJ_TYPE4	4
+#define	PACKETJ_TYPE5	5
+#define	PACKETJ_TYPE6	6
+#define	PACKETJ_TYPE7	7
+
+#define PACKETJ(reg, r, cond, type)	((reg & 0x3FFFF) |			\
+			 ((r & 0x3F) << 18) |			\
+			 ((cond & 0xF) << 24) |				\
+			 ((type & 0xF) << 28))
+
+#define CP_PACKETJ_NOP		0x60000000
+#define CP_PACKETJ_GET_REG(x)  ((x) & 0x3FFFF)
+#define CP_PACKETJ_GET_RES(x)  (((x) >> 18) & 0x3F)
+#define CP_PACKETJ_GET_COND(x) (((x) >> 24) & 0xF)
+#define CP_PACKETJ_GET_TYPE(x) (((x) >> 28) & 0xF)
+
 /* Packet 3 types */
 #define	PACKET3_NOP					0x10
 #define	PACKET3_SET_BASE				0x11
@@ -64,11 +93,25 @@
 #define	PACKET3_DISPATCH_INDIRECT			0x16
 #define	PACKET3_ATOMIC_GDS				0x1D
 #define	PACKET3_ATOMIC_MEM				0x1E
+#define		PACKET3_ATOMIC_MEM__ATOMIC(x) ((((unsigned)(x)) & 0x3F) << 0)
+#define		PACKET3_ATOMIC_MEM__COMMAND(x) ((((unsigned)(x)) & 0xF) << 8)
+#define		PACKET3_ATOMIC_MEM__CACHE_POLICY(x) ((((unsigned)(x)) & 0x3) << 25)
+#define		PACKET3_ATOMIC_MEM__ADDR_LO(x) (((unsigned)(x)) << 0)
+#define		PACKET3_ATOMIC_MEM__ADDR_HI(x) (((unsigned)(x)) << 0)
+#define		PACKET3_ATOMIC_MEM__SRC_DATA_LO(x) (((unsigned)(x)) << 0)
+#define		PACKET3_ATOMIC_MEM__SRC_DATA_HI(x) (((unsigned)(x)) << 0)
+#define		PACKET3_ATOMIC_MEM__CMP_DATA_LO(x) (((unsigned)(x)) << 0)
+#define		PACKET3_ATOMIC_MEM__CMP_DATA_HI(x) (((unsigned)(x)) << 0)
+#define		PACKET3_ATOMIC_MEM__LOOP_INTERVAL(x) ((((unsigned)(x)) & 0x1FFF) << 0)
+#define 	PACKET3_ATOMIC_MEM__COMMAND__SINGLE_PASS_ATOMIC	0
+#define 	PACKET3_ATOMIC_MEM__COMMAND__LOOP_UNTIL_COMPARE_SATISFIED 1
 #define	PACKET3_OCCLUSION_QUERY				0x1F
 #define	PACKET3_SET_PREDICATION				0x20
 #define	PACKET3_REG_RMW					0x21
 #define	PACKET3_COND_EXEC				0x22
 #define	PACKET3_PRED_EXEC				0x23
+#define		PACKET3_PRED_EXEC__EXEC_COUNT(x)	((((unsigned)(x)) & 0x3FFF) << 0)
+#define		PACKET3_PRED_EXEC__VIRTUAL_XCC_ID_SELECT(x)	((((unsigned)(x)) & 0xFF) << 24)
 #define	PACKET3_DRAW_INDIRECT				0x24
 #define	PACKET3_DRAW_INDEX_INDIRECT			0x25
 #define	PACKET3_INDEX_BASE				0x26
@@ -103,6 +146,28 @@
 		 * 1 - pfp
 		 * 2 - ce
 		 */
+#define		PACKET3_WRITE_DATA__DST_SEL(x) ((((unsigned)(x)) & 0xF) << 8)
+#define		PACKET3_WRITE_DATA__ADDR_INCR(x) ((((unsigned)(x)) & 0x1) << 16)
+#define		PACKET3_WRITE_DATA__RESUME_VF_MI300(x) ((((unsigned)(x)) & 0x1) << 19)
+#define		PACKET3_WRITE_DATA__WR_CONFIRM(x) ((((unsigned)(x)) & 0x1) << 20)
+#define		PACKET3_WRITE_DATA__CACHE_POLICY(x) ((((unsigned)(x)) & 0x3) << 25)
+#define		PACKET3_WRITE_DATA__DST_MMREG_ADDR(x) ((((unsigned)(x)) & 0x3FFFF) << 0)
+#define		PACKET3_WRITE_DATA__DST_GDS_ADDR(x) ((((unsigned)(x)) & 0xFFFF) << 0)
+#define		PACKET3_WRITE_DATA__DST_MEM_ADDR_LO(x) ((((unsigned)(x)) & 0x3FFFFFFF) << 2)
+#define		PACKET3_WRITE_DATA__DST_MEM_ADDR_HI(x) ((unsigned)(x))
+#define		PACKET3_WRITE_DATA__DST_SEL__MEM_MAPPED_REGISTER	0
+#define		PACKET3_WRITE_DATA__DST_SEL__TC_L2	2
+#define		PACKET3_WRITE_DATA__DST_SEL__GDS	3
+#define		PACKET3_WRITE_DATA__DST_SEL__MEMORY	5
+#define		PACKET3_WRITE_DATA__DST_SEL__MEMORY_MAPPED_ADC_PERSISTENT_STATE	6
+#define		PACKET3_WRITE_DATA__ADDR_INCR__INCREMENT_ADDRESS	0
+#define		PACKET3_WRITE_DATA__ADDR_INCR__DO_NOT_INCREMENT_ADDRESS	1
+#define		PACKET3_WRITE_DATA__WR_CONFIRM__DO_NOT_WAIT_FOR_WRITE_CONFIRMATION	0
+#define		PACKET3_WRITE_DATA__WR_CONFIRM__WAIT_FOR_WRITE_CONFIRMATION	1
+#define		PACKET3_WRITE_DATA__CACHE_POLICY__LRU	0
+#define		PACKET3_WRITE_DATA__CACHE_POLICY__STREAM	1
+#define		PACKET3_WRITE_DATA__CACHE_POLICY__NOA	2
+#define		PACKET3_WRITE_DATA__CACHE_POLICY__BYPASS	3
 #define	PACKET3_DRAW_INDEX_INDIRECT_MULTI		0x38
 #define	PACKET3_MEM_SEMAPHORE				0x39
 #              define PACKET3_SEM_USE_MAILBOX       (0x1 << 16)
@@ -131,6 +196,33 @@
 		/* 0 - me
 		 * 1 - pfp
 		 */
+#define		PACKET3_WAIT_REG_MEM__FUNCTION(x) ((((unsigned)(x)) & 0x7) << 0)
+#define		PACKET3_WAIT_REG_MEM__MEM_SPACE(x) ((((unsigned)(x)) & 0x3) << 4)
+#define		PACKET3_WAIT_REG_MEM__OPERATION(x) ((((unsigned)(x)) & 0x3) << 6)
+#define		PACKET3_WAIT_REG_MEM__MES_INTR_PIPE(x) ((((unsigned)(x)) & 0x3) << 22)
+#define		PACKET3_WAIT_REG_MEM__MES_ACTION(x) ((((unsigned)(x)) & 0x1) << 24)
+#define		PACKET3_WAIT_REG_MEM__CACHE_POLICY(x) ((((unsigned)(x)) & 0x3) << 25)
+#define		PACKET3_WAIT_REG_MEM__MEM_POLL_ADDR_LO(x) ((((unsigned)(x)) & 0x3FFFFFFF) << 2)
+#define		PACKET3_WAIT_REG_MEM__REG_POLL_ADDR(x) ((((unsigned)(x)) & 0x3FFFF) << 0)
+#define		PACKET3_WAIT_REG_MEM__REG_WRITE_ADDR1(x) ((((unsigned)(x)) & 0x3FFFF) << 0)
+#define		PACKET3_WAIT_REG_MEM__MEM_POLL_ADDR_HI(x) ((unsigned)(x))
+#define		PACKET3_WAIT_REG_MEM__REG_WRITE_ADDR2(x) ((((unsigned)(x)) & 0x3FFFF) << 0)
+#define		PACKET3_WAIT_REG_MEM__REFERENCE(x) ((unsigned)(x))
+#define		PACKET3_WAIT_REG_MEM__MASK(x) ((unsigned)(x))
+#define		PACKET3_WAIT_REG_MEM__POLL_INTERVAL(x) ((((unsigned)(x)) & 0xFFFF) << 0)
+#define		PACKET3_WAIT_REG_MEM__OPTIMIZE_ACE_OFFLOAD_MODE(x) ((((unsigned)(x)) & 0x1) << 31)
+#define		PACKET3_WAIT_REG_MEM__FUNCTION__ALWAYS_PASS	0
+#define		PACKET3_WAIT_REG_MEM__FUNCTION__LESS_THAN_REF_VALUE	1
+#define		PACKET3_WAIT_REG_MEM__FUNCTION__LESS_THAN_EQUAL_TO_THE_REF_VALUE	2
+#define		PACKET3_WAIT_REG_MEM__FUNCTION__EQUAL_TO_THE_REFERENCE_VALUE	3
+#define		PACKET3_WAIT_REG_MEM__FUNCTION__NOT_EQUAL_REFERENCE_VALUE	4
+#define		PACKET3_WAIT_REG_MEM__FUNCTION__GREATER_THAN_OR_EQUAL_REFERENCE_VALUE	5
+#define		PACKET3_WAIT_REG_MEM__FUNCTION__GREATER_THAN_REFERENCE_VALUE	6
+#define		PACKET3_WAIT_REG_MEM__MEM_SPACE__REGISTER_SPACE	0
+#define		PACKET3_WAIT_REG_MEM__MEM_SPACE__MEMORY_SPACE	1
+#define		PACKET3_WAIT_REG_MEM__OPERATION__WAIT_REG_MEM	0
+#define		PACKET3_WAIT_REG_MEM__OPERATION__WR_WAIT_WR_REG	1
+#define		PACKET3_WAIT_REG_MEM__OPERATION__WAIT_MEM_PREEMPTABLE	3
 #define	PACKET3_INDIRECT_BUFFER				0x3F
 #define		INDIRECT_BUFFER_VALID                   (1 << 23)
 #define		INDIRECT_BUFFER_CACHE_POLICY(x)         ((x) << 28)
@@ -139,7 +231,64 @@
 		 * 2 - Bypass
 		 */
 #define     INDIRECT_BUFFER_PRE_ENB(x)		 ((x) << 21)
+#define     INDIRECT_BUFFER_PRE_RESUME(x)               ((x) << 30)
+#define		PACKET3_INDIRECT_BUFFER__IB_BASE_LO(x) ((((unsigned)(x)) & 0x3FFFFFFF) << 2)
+#define		PACKET3_INDIRECT_BUFFER__IB_BASE_HI(x) ((unsigned)(x))
+#define		PACKET3_INDIRECT_BUFFER__IB_SIZE(x) ((((unsigned)(x)) & 0xFFFFF) << 0)
+#define		PACKET3_INDIRECT_BUFFER__CHAIN(x) ((((unsigned)(x)) & 0x1) << 20)
+#define		PACKET3_INDIRECT_BUFFER__OFFLOAD_POLLING(x) ((((unsigned)(x)) & 0x1) << 21)
+#define		PACKET3_INDIRECT_BUFFER__VALID(x) ((((unsigned)(x)) & 0x1) << 23)
+#define		PACKET3_INDIRECT_BUFFER__VMID(x) ((((unsigned)(x)) & 0xF) << 24)
+#define		PACKET3_INDIRECT_BUFFER__CACHE_POLICY(x) ((((unsigned)(x)) & 0x3) << 28)
+#define		PACKET3_INDIRECT_BUFFER__PRIV(x) ((((unsigned)(x)) & 0x1) << 31)
+#define		PACKET3_INDIRECT_BUFFER__CACHE_POLICY__LRU	0
+#define		PACKET3_INDIRECT_BUFFER__CACHE_POLICY__STREAM	1
 #define	PACKET3_COPY_DATA				0x40
+#define		PACKET3_COPY_DATA__SRC_SEL(x) ((((unsigned)(x)) & 0xF) << 0)
+#define		PACKET3_COPY_DATA__DST_SEL(x) ((((unsigned)(x)) & 0xF) << 8)
+#define		PACKET3_COPY_DATA__SRC_CACHE_POLICY(x) ((((unsigned)(x)) & 0x3) << 13)
+#define		PACKET3_COPY_DATA__COUNT_SEL(x) ((((unsigned)(x)) & 0x1) << 16)
+#define		PACKET3_COPY_DATA__WR_CONFIRM(x) ((((unsigned)(x)) & 0x1) << 20)
+#define		PACKET3_COPY_DATA__DST_CACHE_POLICY(x) ((((unsigned)(x)) & 0x3) << 25)
+#define		PACKET3_COPY_DATA__PQ_EXE_STATUS(x) ((((unsigned)(x)) & 0x1) << 29)
+#define		PACKET3_COPY_DATA__SRC_REG_OFFSET(x) ((((unsigned)(x)) & 0x3FFFF) << 0)
+#define		PACKET3_COPY_DATA__SRC_32B_ADDR_LO(x) ((((unsigned)(x)) & 0x3FFFFFFF) << 2)
+#define		PACKET3_COPY_DATA__SRC_64B_ADDR_LO(x) ((((unsigned)(x)) & 0x1FFFFFFF) << 3)
+#define		PACKET3_COPY_DATA__SRC_GDS_ADDR_LO(x) ((((unsigned)(x)) & 0xFFFF) << 0)
+#define		PACKET3_COPY_DATA__IMM_DATA(x) ((unsigned)(x))
+#define		PACKET3_COPY_DATA__SRC_MEMTC_ADDR_HI(x) ((unsigned)(x))
+#define		PACKET3_COPY_DATA__SRC_IMM_DATA(x) ((unsigned)(x))
+#define		PACKET3_COPY_DATA__DST_REG_OFFSET(x) ((((unsigned)(x)) & 0x3FFFF) << 0)
+#define		PACKET3_COPY_DATA__DST_32B_ADDR_LO(x) ((((unsigned)(x)) & 0x3FFFFFFF) << 2)
+#define		PACKET3_COPY_DATA__DST_64B_ADDR_LO(x) ((((unsigned)(x)) & 0x1FFFFFFF) << 3)
+#define		PACKET3_COPY_DATA__DST_GDS_ADDR_LO(x) ((((unsigned)(x)) & 0xFFFF) << 0)
+#define		PACKET3_COPY_DATA__DST_ADDR_HI(x) ((unsigned)(x))
+#define		PACKET3_COPY_DATA__SRC_SEL__MEM_MAPPED_REGISTER	0
+#define		PACKET3_COPY_DATA__SRC_SEL__MEMORY	1
+#define		PACKET3_COPY_DATA__SRC_SEL__TC_L2	2
+#define		PACKET3_COPY_DATA__SRC_SEL__GDS	3
+#define		PACKET3_COPY_DATA__SRC_SEL__PERFCOUNTERS	4
+#define		PACKET3_COPY_DATA__SRC_SEL__IMMEDIATE_DATA	5
+#define		PACKET3_COPY_DATA__SRC_SEL__ATOMIC_RETURN_DATA	6
+#define		PACKET3_COPY_DATA__SRC_SEL__GDS_ATOMIC_RETURN_DATA0	7
+#define		PACKET3_COPY_DATA__SRC_SEL__GDS_ATOMIC_RETURN_DATA1	8
+#define		PACKET3_COPY_DATA__SRC_SEL__GPU_CLOCK_COUNT	9
+#define		PACKET3_COPY_DATA__DST_SEL__MEM_MAPPED_REGISTER	0
+#define		PACKET3_COPY_DATA__DST_SEL__TC_L2	2
+#define		PACKET3_COPY_DATA__DST_SEL__GDS	3
+#define		PACKET3_COPY_DATA__DST_SEL__PERFCOUNTERS	4
+#define		PACKET3_COPY_DATA__DST_SEL__MEMORY	5
+#define		PACKET3_COPY_DATA__DST_SEL__MEM_MAPPED_REG_DC	6
+#define		PACKET3_COPY_DATA__SRC_CACHE_POLICY__LRU	0
+#define		PACKET3_COPY_DATA__SRC_CACHE_POLICY__STREAM 1
+#define		PACKET3_COPY_DATA__COUNT_SEL__32_BITS_OF_DATA	0
+#define		PACKET3_COPY_DATA__COUNT_SEL__64_BITS_OF_DATA	1
+#define		PACKET3_COPY_DATA__WR_CONFIRM__DO_NOT_WAIT_FOR_CONFIRMATION	0
+#define		PACKET3_COPY_DATA__WR_CONFIRM__WAIT_FOR_CONFIRMATION	1
+#define		PACKET3_COPY_DATA__DST_CACHE_POLICY__LRU	0
+#define		PACKET3_COPY_DATA__DST_CACHE_POLICY__STREAM	1
+#define		PACKET3_COPY_DATA__PQ_EXE_STATUS__DEFAULT	0
+#define		PACKET3_COPY_DATA__PQ_EXE_STATUS__PHASE_UPDATE	1
 #define	PACKET3_PFP_SYNC_ME				0x42
 #define	PACKET3_COND_WRITE				0x45
 #define	PACKET3_EVENT_WRITE				0x46
@@ -151,6 +300,15 @@
 		 * 3 - SAMPLE_STREAMOUTSTAT*
 		 * 4 - *S_PARTIAL_FLUSH
 		 */
+#define		PACKET3_EVENT_WRITE__EVENT_TYPE(x) ((((unsigned)(x)) & 0x3F) << 0)
+#define		PACKET3_EVENT_WRITE__EVENT_INDEX(x) ((((unsigned)(x)) & 0xF) << 8)
+#define		PACKET3_EVENT_WRITE__OFFLOAD_ENABLE(x) ((((unsigned)(x)) & 0x1) << 31)
+#define		PACKET3_EVENT_WRITE__SAMP_PLST_CNTR_MODE(x) ((((unsigned)(x)) & 0x3) << 29)
+#define		PACKET3_EVENT_WRITE__ADDRESS_LO(x) ((((unsigned)(x)) & 0x1FFFFFFF) << 3)
+#define		PACKET3_EVENT_WRITE__ADDRESS_HI(x) (((unsigned)(x)) << 0)
+#define		PACKET3_EVENT_WRITE__EVENT_INDEX__OTHER	0
+#define		PACKET3_EVENT_WRITE__EVENT_INDEX__SAMPLE_PIPELINESTATS	2
+#define		PACKET3_EVENT_WRITE__EVENT_INDEX__CS_PARTIAL_FLUSH	4
 #define	PACKET3_RELEASE_MEM				0x49
 #define		EVENT_TYPE(x)                           ((x) << 0)
 #define		EVENT_INDEX(x)                          ((x) << 8)
@@ -159,7 +317,9 @@
 #define		EOP_TC_WB_ACTION_EN                     (1 << 15) /* L2 */
 #define		EOP_TCL1_ACTION_EN                      (1 << 16)
 #define		EOP_TC_ACTION_EN                        (1 << 17) /* L2 */
+#define		EOP_TC_NC_ACTION_EN			(1 << 19)
 #define		EOP_TC_MD_ACTION_EN			(1 << 21) /* L2 metadata */
+#define		EOP_EXEC				(1 << 28) /* For Trailing Fence */
 
 #define		DATA_SEL(x)                             ((x) << 29)
 		/* 0 - discard
@@ -229,8 +389,38 @@
 #              define PACKET3_DMA_DATA_CMD_SAIC    (1 << 28)
 #              define PACKET3_DMA_DATA_CMD_DAIC    (1 << 29)
 #              define PACKET3_DMA_DATA_CMD_RAW_WAIT  (1 << 30)
-#define	PACKET3_AQUIRE_MEM				0x58
+#define	PACKET3_ACQUIRE_MEM				0x58
+/* 1.  HEADER
+ * 2.  COHER_CNTL [30:0]
+ * 2.1 ENGINE_SEL [31:31]
+ * 3.  COHER_SIZE [31:0]
+ * 4.  COHER_SIZE_HI [7:0]
+ * 5.  COHER_BASE_LO [31:0]
+ * 6.  COHER_BASE_HI [23:0]
+ * 7.  POLL_INTERVAL [15:0]
+ */
+/* COHER_CNTL fields for CP_COHER_CNTL */
+#define 	PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_TC_NC_ACTION_ENA(x) ((x) << 3)
+#define 	PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_TC_WC_ACTION_ENA(x) ((x) << 4)
+#define 	PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_TC_INV_METADATA_ACTION_ENA(x) ((x) << 5)
+#define 	PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_TCL1_VOL_ACTION_ENA(x) ((x) << 15)
+#define 	PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_TC_WB_ACTION_ENA(x) ((x) << 18)
+#define 	PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_TCL1_ACTION_ENA(x) ((x) << 22)
+#define 	PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_TC_ACTION_ENA(x) ((x) << 23)
+#define 	PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_CB_ACTION_ENA(x) ((x) << 25)
+#define 	PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_DB_ACTION_ENA(x) ((x) << 26)
+#define 	PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_SH_KCACHE_ACTION_ENA(x) ((x) << 27)
+#define 	PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_SH_KCACHE_VOL_ACTION_ENA(x) ((x) << 28)
+#define 	PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_SH_ICACHE_ACTION_ENA(x) ((x) << 29)
+#define 	PACKET3_ACQUIRE_MEM_CP_COHER_CNTL_SH_KCACHE_WB_ACTION_ENA(x) ((x) << 30)
 #define	PACKET3_REWIND					0x59
+#define		PACKET3_ACQUIRE_MEM__COHER_SIZE(x) ((unsigned)(x))
+#define 	PACKET3_ACQUIRE_MEM__COHER_SIZE_HI(x) ((((unsigned)(x)) & 0xFF) << 0)
+#define 	PACKET3_ACQUIRE_MEM__COHER_SIZE_HI_VG10(x) ((((unsigned)(x)) & 0xFFFFFF) << 0)
+#define 	PACKET3_ACQUIRE_MEM__COHER_BASE_LO(x) ((unsigned)(x))
+#define 	PACKET3_ACQUIRE_MEM__COHER_BASE_HI(x) ((((unsigned)(x)) & 0xFFFFFF) << 0)
+#define 	PACKET3_ACQUIRE_MEM__POLL_INTERVAL(x) ((((unsigned)(x)) & 0xFFFF) << 0)
+#define 	PACKET3_ACQUIRE_MEM__GCR_CNTL(x) ((((unsigned)(x)) & 0x7FF) << 0)
 #define	PACKET3_LOAD_UCONFIG_REG			0x5E
 #define	PACKET3_LOAD_SH_REG				0x5F
 #define	PACKET3_LOAD_CONFIG_REG				0x60
@@ -245,12 +435,16 @@
 #define	PACKET3_SET_SH_REG				0x76
 #define		PACKET3_SET_SH_REG_START			0x00002c00
 #define		PACKET3_SET_SH_REG_END				0x00003000
+#define 	PACKET3_SET_SH_REG__REG_OFFSET(x) ((((unsigned)(x)) & 0xFFFF) << 0)
+#define 	PACKET3_SET_SH_REG__VMID_SHIFT(x) ((((unsigned)(x)) & 0x1F) << 23)
+#define 	PACKET3_SET_SH_REG__INDEX(x) ((((unsigned)(x)) & 0xF) << 28)
 #define	PACKET3_SET_SH_REG_OFFSET			0x77
 #define	PACKET3_SET_QUEUE_REG				0x78
 #define	PACKET3_SET_UCONFIG_REG				0x79
 #define		PACKET3_SET_UCONFIG_REG_START			0x0000c000
 #define		PACKET3_SET_UCONFIG_REG_END			0x0000c400
 #define		PACKET3_SET_UCONFIG_REG_INDEX_TYPE		(2 << 28)
+#define 	PACKET3_SET_UCONFIG_REG__REG_OFFSET(x) ((((unsigned)(x)) & 0xFFFF) << 0)
 #define	PACKET3_SCRATCH_RAM_WRITE			0x7D
 #define	PACKET3_SCRATCH_RAM_READ			0x7E
 #define	PACKET3_LOAD_CONST_RAM				0x80
@@ -262,12 +456,18 @@
 #define	PACKET3_WAIT_ON_DE_COUNTER_DIFF			0x88
 #define	PACKET3_SWITCH_BUFFER				0x8B
 #define PACKET3_FRAME_CONTROL				0x90
+#			define FRAME_TMZ	(1 << 0)
 #			define FRAME_CMD(x) ((x) << 28)
 			/*
 			 * x=0: tmz_begin
 			 * x=1: tmz_end
 			 */
 
+#define	PACKET3_INVALIDATE_TLBS				0x98
+#              define PACKET3_INVALIDATE_TLBS_DST_SEL(x)     ((x) << 0)
+#              define PACKET3_INVALIDATE_TLBS_ALL_HUB(x)     ((x) << 4)
+#              define PACKET3_INVALIDATE_TLBS_PASID(x)       ((x) << 5)
+#              define PACKET3_INVALIDATE_TLBS_FLUSH_TYPE(x)  ((x) << 29)
 #define PACKET3_SET_RESOURCES				0xA0
 /* 1. header
  * 2. CONTROL
@@ -352,6 +552,15 @@
 #              define PACKET3_QUERY_STATUS_DOORBELL_OFFSET(x)  ((x) << 2)
 #              define PACKET3_QUERY_STATUS_ENG_SEL(x)          ((x) << 25)
 
+#define PACKET3_RUN_CLEANER_SHADER_9_0                      0xD7
+/* 1. header
+ * 2. RESERVED [31:0]
+ */
+
+#define PACKET3_RUN_CLEANER_SHADER                      0xD2
+/* 1. header
+ * 2. RESERVED [31:0]
+ */
 
 #define VCE_CMD_NO_OP		0x00000000
 #define VCE_CMD_END		0x00000001
diff --git a/drivers/gpu/drm/amd/amdgpu/soc21.c b/drivers/gpu/drm/amd/amdgpu/soc21.c
new file mode 100644
index 000000000000..ad36c96478a8
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/soc21.c
@@ -0,0 +1,1021 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/firmware.h>
+#include <linux/slab.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+
+#include "amdgpu.h"
+#include "amdgpu_atombios.h"
+#include "amdgpu_ih.h"
+#include "amdgpu_uvd.h"
+#include "amdgpu_vce.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_psp.h"
+#include "amdgpu_smu.h"
+#include "atom.h"
+#include "amd_pcie.h"
+
+#include "gc/gc_11_0_0_offset.h"
+#include "gc/gc_11_0_0_sh_mask.h"
+#include "mp/mp_13_0_0_offset.h"
+
+#include "soc15.h"
+#include "soc15_common.h"
+#include "soc21.h"
+#include "mxgpu_nv.h"
+
+static const struct amd_ip_funcs soc21_common_ip_funcs;
+
+/* SOC21 */
+static const struct amdgpu_video_codec_info vcn_4_0_0_video_codecs_encode_array_vcn0[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1, 8192, 4352, 0)},
+};
+
+static const struct amdgpu_video_codec_info vcn_4_0_0_video_codecs_encode_array_vcn1[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 0)},
+};
+
+static const struct amdgpu_video_codecs vcn_4_0_0_video_codecs_encode_vcn0 = {
+	.codec_count = ARRAY_SIZE(vcn_4_0_0_video_codecs_encode_array_vcn0),
+	.codec_array = vcn_4_0_0_video_codecs_encode_array_vcn0,
+};
+
+static const struct amdgpu_video_codecs vcn_4_0_0_video_codecs_encode_vcn1 = {
+	.codec_count = ARRAY_SIZE(vcn_4_0_0_video_codecs_encode_array_vcn1),
+	.codec_array = vcn_4_0_0_video_codecs_encode_array_vcn1,
+};
+
+static const struct amdgpu_video_codec_info vcn_4_0_0_video_codecs_decode_array_vcn0[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1, 8192, 4352, 0)},
+};
+
+static const struct amdgpu_video_codec_info vcn_4_0_0_video_codecs_decode_array_vcn1[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)},
+};
+
+static const struct amdgpu_video_codecs vcn_4_0_0_video_codecs_decode_vcn0 = {
+	.codec_count = ARRAY_SIZE(vcn_4_0_0_video_codecs_decode_array_vcn0),
+	.codec_array = vcn_4_0_0_video_codecs_decode_array_vcn0,
+};
+
+static const struct amdgpu_video_codecs vcn_4_0_0_video_codecs_decode_vcn1 = {
+	.codec_count = ARRAY_SIZE(vcn_4_0_0_video_codecs_decode_array_vcn1),
+	.codec_array = vcn_4_0_0_video_codecs_decode_array_vcn1,
+};
+
+/* SRIOV SOC21, not const since data is controlled by host */
+static struct amdgpu_video_codec_info sriov_vcn_4_0_0_video_codecs_encode_array_vcn0[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1, 8192, 4352, 0)},
+};
+
+static struct amdgpu_video_codec_info sriov_vcn_4_0_0_video_codecs_encode_array_vcn1[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 0)},
+};
+
+static struct amdgpu_video_codecs sriov_vcn_4_0_0_video_codecs_encode_vcn0 = {
+	.codec_count = ARRAY_SIZE(sriov_vcn_4_0_0_video_codecs_encode_array_vcn0),
+	.codec_array = sriov_vcn_4_0_0_video_codecs_encode_array_vcn0,
+};
+
+static struct amdgpu_video_codecs sriov_vcn_4_0_0_video_codecs_encode_vcn1 = {
+	.codec_count = ARRAY_SIZE(sriov_vcn_4_0_0_video_codecs_encode_array_vcn1),
+	.codec_array = sriov_vcn_4_0_0_video_codecs_encode_array_vcn1,
+};
+
+static struct amdgpu_video_codec_info sriov_vcn_4_0_0_video_codecs_decode_array_vcn0[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1, 8192, 4352, 0)},
+};
+
+static struct amdgpu_video_codec_info sriov_vcn_4_0_0_video_codecs_decode_array_vcn1[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)},
+};
+
+static struct amdgpu_video_codecs sriov_vcn_4_0_0_video_codecs_decode_vcn0 = {
+	.codec_count = ARRAY_SIZE(sriov_vcn_4_0_0_video_codecs_decode_array_vcn0),
+	.codec_array = sriov_vcn_4_0_0_video_codecs_decode_array_vcn0,
+};
+
+static struct amdgpu_video_codecs sriov_vcn_4_0_0_video_codecs_decode_vcn1 = {
+	.codec_count = ARRAY_SIZE(sriov_vcn_4_0_0_video_codecs_decode_array_vcn1),
+	.codec_array = sriov_vcn_4_0_0_video_codecs_decode_array_vcn1,
+};
+
+static int soc21_query_video_codecs(struct amdgpu_device *adev, bool encode,
+				 const struct amdgpu_video_codecs **codecs)
+{
+	if (adev->vcn.num_vcn_inst == hweight8(adev->vcn.harvest_config))
+		return -EINVAL;
+
+	switch (amdgpu_ip_version(adev, UVD_HWIP, 0)) {
+	case IP_VERSION(4, 0, 0):
+	case IP_VERSION(4, 0, 2):
+	case IP_VERSION(4, 0, 4):
+	case IP_VERSION(4, 0, 5):
+		if (amdgpu_sriov_vf(adev)) {
+			if ((adev->vcn.harvest_config & AMDGPU_VCN_HARVEST_VCN0) ||
+			!amdgpu_sriov_is_av1_support(adev)) {
+				if (encode)
+					*codecs = &sriov_vcn_4_0_0_video_codecs_encode_vcn1;
+				else
+					*codecs = &sriov_vcn_4_0_0_video_codecs_decode_vcn1;
+			} else {
+				if (encode)
+					*codecs = &sriov_vcn_4_0_0_video_codecs_encode_vcn0;
+				else
+					*codecs = &sriov_vcn_4_0_0_video_codecs_decode_vcn0;
+			}
+		} else {
+			if ((adev->vcn.harvest_config & AMDGPU_VCN_HARVEST_VCN0)) {
+				if (encode)
+					*codecs = &vcn_4_0_0_video_codecs_encode_vcn1;
+				else
+					*codecs = &vcn_4_0_0_video_codecs_decode_vcn1;
+			} else {
+				if (encode)
+					*codecs = &vcn_4_0_0_video_codecs_encode_vcn0;
+				else
+					*codecs = &vcn_4_0_0_video_codecs_decode_vcn0;
+			}
+		}
+		return 0;
+	case IP_VERSION(4, 0, 6):
+		if (encode)
+			*codecs = &vcn_4_0_0_video_codecs_encode_vcn0;
+		else
+			*codecs = &vcn_4_0_0_video_codecs_decode_vcn0;
+		return 0;
+	default:
+		return -EINVAL;
+	}
+}
+
+static u32 soc21_didt_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags, address, data;
+	u32 r;
+
+	address = SOC15_REG_OFFSET(GC, 0, regDIDT_IND_INDEX);
+	data = SOC15_REG_OFFSET(GC, 0, regDIDT_IND_DATA);
+
+	spin_lock_irqsave(&adev->didt_idx_lock, flags);
+	WREG32(address, (reg));
+	r = RREG32(data);
+	spin_unlock_irqrestore(&adev->didt_idx_lock, flags);
+	return r;
+}
+
+static void soc21_didt_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags, address, data;
+
+	address = SOC15_REG_OFFSET(GC, 0, regDIDT_IND_INDEX);
+	data = SOC15_REG_OFFSET(GC, 0, regDIDT_IND_DATA);
+
+	spin_lock_irqsave(&adev->didt_idx_lock, flags);
+	WREG32(address, (reg));
+	WREG32(data, (v));
+	spin_unlock_irqrestore(&adev->didt_idx_lock, flags);
+}
+
+static u32 soc21_get_config_memsize(struct amdgpu_device *adev)
+{
+	return adev->nbio.funcs->get_memsize(adev);
+}
+
+static u32 soc21_get_xclk(struct amdgpu_device *adev)
+{
+	return adev->clock.spll.reference_freq;
+}
+
+
+void soc21_grbm_select(struct amdgpu_device *adev,
+		     u32 me, u32 pipe, u32 queue, u32 vmid)
+{
+	u32 grbm_gfx_cntl = 0;
+	grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, PIPEID, pipe);
+	grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, MEID, me);
+	grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, VMID, vmid);
+	grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, QUEUEID, queue);
+
+	WREG32_SOC15(GC, 0, regGRBM_GFX_CNTL, grbm_gfx_cntl);
+}
+
+static bool soc21_read_disabled_bios(struct amdgpu_device *adev)
+{
+	/* todo */
+	return false;
+}
+
+static struct soc15_allowed_register_entry soc21_allowed_read_registers[] = {
+	{ SOC15_REG_ENTRY(GC, 0, regGRBM_STATUS)},
+	{ SOC15_REG_ENTRY(GC, 0, regGRBM_STATUS2)},
+	{ SOC15_REG_ENTRY(GC, 0, regGRBM_STATUS_SE0)},
+	{ SOC15_REG_ENTRY(GC, 0, regGRBM_STATUS_SE1)},
+	{ SOC15_REG_ENTRY(GC, 0, regGRBM_STATUS_SE2)},
+	{ SOC15_REG_ENTRY(GC, 0, regGRBM_STATUS_SE3)},
+	{ SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_STATUS_REG)},
+	{ SOC15_REG_ENTRY(SDMA1, 0, regSDMA1_STATUS_REG)},
+	{ SOC15_REG_ENTRY(GC, 0, regCP_STAT)},
+	{ SOC15_REG_ENTRY(GC, 0, regCP_STALLED_STAT1)},
+	{ SOC15_REG_ENTRY(GC, 0, regCP_STALLED_STAT2)},
+	{ SOC15_REG_ENTRY(GC, 0, regCP_STALLED_STAT3)},
+	{ SOC15_REG_ENTRY(GC, 0, regCP_CPF_BUSY_STAT)},
+	{ SOC15_REG_ENTRY(GC, 0, regCP_CPF_STALLED_STAT1)},
+	{ SOC15_REG_ENTRY(GC, 0, regCP_CPF_STATUS)},
+	{ SOC15_REG_ENTRY(GC, 0, regCP_CPC_BUSY_STAT)},
+	{ SOC15_REG_ENTRY(GC, 0, regCP_CPC_STALLED_STAT1)},
+	{ SOC15_REG_ENTRY(GC, 0, regCP_CPC_STATUS)},
+	{ SOC15_REG_ENTRY(GC, 0, regGB_ADDR_CONFIG)},
+};
+
+static uint32_t soc21_read_indexed_register(struct amdgpu_device *adev, u32 se_num,
+					 u32 sh_num, u32 reg_offset)
+{
+	uint32_t val;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	if (se_num != 0xffffffff || sh_num != 0xffffffff)
+		amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff, 0);
+
+	val = RREG32(reg_offset);
+
+	if (se_num != 0xffffffff || sh_num != 0xffffffff)
+		amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
+	mutex_unlock(&adev->grbm_idx_mutex);
+	return val;
+}
+
+static uint32_t soc21_get_register_value(struct amdgpu_device *adev,
+				      bool indexed, u32 se_num,
+				      u32 sh_num, u32 reg_offset)
+{
+	if (indexed) {
+		return soc21_read_indexed_register(adev, se_num, sh_num, reg_offset);
+	} else {
+		if (reg_offset == SOC15_REG_OFFSET(GC, 0, regGB_ADDR_CONFIG) && adev->gfx.config.gb_addr_config)
+			return adev->gfx.config.gb_addr_config;
+		return RREG32(reg_offset);
+	}
+}
+
+static int soc21_read_register(struct amdgpu_device *adev, u32 se_num,
+			    u32 sh_num, u32 reg_offset, u32 *value)
+{
+	uint32_t i;
+	struct soc15_allowed_register_entry  *en;
+
+	*value = 0;
+	for (i = 0; i < ARRAY_SIZE(soc21_allowed_read_registers); i++) {
+		en = &soc21_allowed_read_registers[i];
+		if (!adev->reg_offset[en->hwip][en->inst])
+			continue;
+		else if (reg_offset != (adev->reg_offset[en->hwip][en->inst][en->seg]
+					+ en->reg_offset))
+			continue;
+
+		*value = soc21_get_register_value(adev,
+					       soc21_allowed_read_registers[i].grbm_indexed,
+					       se_num, sh_num, reg_offset);
+		return 0;
+	}
+	return -EINVAL;
+}
+
+#if 0
+static int soc21_asic_mode1_reset(struct amdgpu_device *adev)
+{
+	u32 i;
+	int ret = 0;
+
+	amdgpu_atombios_scratch_regs_engine_hung(adev, true);
+
+	/* disable BM */
+	pci_clear_master(adev->pdev);
+
+	amdgpu_device_cache_pci_state(adev->pdev);
+
+	if (amdgpu_dpm_is_mode1_reset_supported(adev)) {
+		dev_info(adev->dev, "GPU smu mode1 reset\n");
+		ret = amdgpu_dpm_mode1_reset(adev);
+	} else {
+		dev_info(adev->dev, "GPU psp mode1 reset\n");
+		ret = psp_gpu_reset(adev);
+	}
+
+	if (ret)
+		dev_err(adev->dev, "GPU mode1 reset failed\n");
+	amdgpu_device_load_pci_state(adev->pdev);
+
+	/* wait for asic to come out of reset */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		u32 memsize = adev->nbio.funcs->get_memsize(adev);
+
+		if (memsize != 0xffffffff)
+			break;
+		udelay(1);
+	}
+
+	amdgpu_atombios_scratch_regs_engine_hung(adev, false);
+
+	return ret;
+}
+#endif
+
+static enum amd_reset_method
+soc21_asic_reset_method(struct amdgpu_device *adev)
+{
+	if (amdgpu_reset_method == AMD_RESET_METHOD_MODE1 ||
+	    amdgpu_reset_method == AMD_RESET_METHOD_MODE2 ||
+	    amdgpu_reset_method == AMD_RESET_METHOD_BACO)
+		return amdgpu_reset_method;
+
+	if (amdgpu_reset_method != -1)
+		dev_warn(adev->dev, "Specified reset method:%d isn't supported, using AUTO instead.\n",
+				  amdgpu_reset_method);
+
+	switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) {
+	case IP_VERSION(13, 0, 0):
+	case IP_VERSION(13, 0, 7):
+	case IP_VERSION(13, 0, 10):
+		return AMD_RESET_METHOD_MODE1;
+	case IP_VERSION(13, 0, 4):
+	case IP_VERSION(13, 0, 11):
+	case IP_VERSION(14, 0, 0):
+	case IP_VERSION(14, 0, 1):
+	case IP_VERSION(14, 0, 4):
+	case IP_VERSION(14, 0, 5):
+		return AMD_RESET_METHOD_MODE2;
+	default:
+		if (amdgpu_dpm_is_baco_supported(adev))
+			return AMD_RESET_METHOD_BACO;
+		else
+			return AMD_RESET_METHOD_MODE1;
+	}
+}
+
+static int soc21_asic_reset(struct amdgpu_device *adev)
+{
+	int ret = 0;
+
+	switch (soc21_asic_reset_method(adev)) {
+	case AMD_RESET_METHOD_PCI:
+		dev_info(adev->dev, "PCI reset\n");
+		ret = amdgpu_device_pci_reset(adev);
+		break;
+	case AMD_RESET_METHOD_BACO:
+		dev_info(adev->dev, "BACO reset\n");
+		ret = amdgpu_dpm_baco_reset(adev);
+		break;
+	case AMD_RESET_METHOD_MODE2:
+		dev_info(adev->dev, "MODE2 reset\n");
+		ret = amdgpu_dpm_mode2_reset(adev);
+		break;
+	default:
+		dev_info(adev->dev, "MODE1 reset\n");
+		ret = amdgpu_device_mode1_reset(adev);
+		break;
+	}
+
+	return ret;
+}
+
+static int soc21_set_uvd_clocks(struct amdgpu_device *adev, u32 vclk, u32 dclk)
+{
+	/* todo */
+	return 0;
+}
+
+static int soc21_set_vce_clocks(struct amdgpu_device *adev, u32 evclk, u32 ecclk)
+{
+	/* todo */
+	return 0;
+}
+
+static void soc21_program_aspm(struct amdgpu_device *adev)
+{
+	if (!amdgpu_device_should_use_aspm(adev))
+		return;
+
+	if (adev->nbio.funcs->program_aspm)
+		adev->nbio.funcs->program_aspm(adev);
+}
+
+const struct amdgpu_ip_block_version soc21_common_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_COMMON,
+	.major = 1,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &soc21_common_ip_funcs,
+};
+
+static bool soc21_need_full_reset(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 2):
+	case IP_VERSION(11, 0, 3):
+	default:
+		return true;
+	}
+}
+
+static bool soc21_need_reset_on_init(struct amdgpu_device *adev)
+{
+	u32 sol_reg;
+
+	if (adev->flags & AMD_IS_APU)
+		return false;
+
+	/* Check sOS sign of life register to confirm sys driver and sOS
+	 * are already been loaded.
+	 */
+	sol_reg = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_81);
+	if (sol_reg)
+		return true;
+
+	return false;
+}
+
+static void soc21_init_doorbell_index(struct amdgpu_device *adev)
+{
+	adev->doorbell_index.kiq = AMDGPU_NAVI10_DOORBELL_KIQ;
+	adev->doorbell_index.mec_ring0 = AMDGPU_NAVI10_DOORBELL_MEC_RING0;
+	adev->doorbell_index.mec_ring1 = AMDGPU_NAVI10_DOORBELL_MEC_RING1;
+	adev->doorbell_index.mec_ring2 = AMDGPU_NAVI10_DOORBELL_MEC_RING2;
+	adev->doorbell_index.mec_ring3 = AMDGPU_NAVI10_DOORBELL_MEC_RING3;
+	adev->doorbell_index.mec_ring4 = AMDGPU_NAVI10_DOORBELL_MEC_RING4;
+	adev->doorbell_index.mec_ring5 = AMDGPU_NAVI10_DOORBELL_MEC_RING5;
+	adev->doorbell_index.mec_ring6 = AMDGPU_NAVI10_DOORBELL_MEC_RING6;
+	adev->doorbell_index.mec_ring7 = AMDGPU_NAVI10_DOORBELL_MEC_RING7;
+	adev->doorbell_index.userqueue_start = AMDGPU_NAVI10_DOORBELL_USERQUEUE_START;
+	adev->doorbell_index.userqueue_end = AMDGPU_NAVI10_DOORBELL_USERQUEUE_END;
+	adev->doorbell_index.gfx_ring0 = AMDGPU_NAVI10_DOORBELL_GFX_RING0;
+	adev->doorbell_index.gfx_ring1 = AMDGPU_NAVI10_DOORBELL_GFX_RING1;
+	adev->doorbell_index.gfx_userqueue_start =
+		AMDGPU_NAVI10_DOORBELL_GFX_USERQUEUE_START;
+	adev->doorbell_index.gfx_userqueue_end =
+		AMDGPU_NAVI10_DOORBELL_GFX_USERQUEUE_END;
+	adev->doorbell_index.mes_ring0 = AMDGPU_NAVI10_DOORBELL_MES_RING0;
+	adev->doorbell_index.mes_ring1 = AMDGPU_NAVI10_DOORBELL_MES_RING1;
+	adev->doorbell_index.sdma_engine[0] = AMDGPU_NAVI10_DOORBELL_sDMA_ENGINE0;
+	adev->doorbell_index.sdma_engine[1] = AMDGPU_NAVI10_DOORBELL_sDMA_ENGINE1;
+	adev->doorbell_index.ih = AMDGPU_NAVI10_DOORBELL_IH;
+	adev->doorbell_index.vcn.vcn_ring0_1 = AMDGPU_NAVI10_DOORBELL64_VCN0_1;
+	adev->doorbell_index.vcn.vcn_ring2_3 = AMDGPU_NAVI10_DOORBELL64_VCN2_3;
+	adev->doorbell_index.vcn.vcn_ring4_5 = AMDGPU_NAVI10_DOORBELL64_VCN4_5;
+	adev->doorbell_index.vcn.vcn_ring6_7 = AMDGPU_NAVI10_DOORBELL64_VCN6_7;
+	adev->doorbell_index.vpe_ring = AMDGPU_NAVI10_DOORBELL64_VPE;
+	adev->doorbell_index.first_non_cp = AMDGPU_NAVI10_DOORBELL64_FIRST_NON_CP;
+	adev->doorbell_index.last_non_cp = AMDGPU_NAVI10_DOORBELL64_LAST_NON_CP;
+
+	adev->doorbell_index.max_assignment = AMDGPU_NAVI10_DOORBELL_MAX_ASSIGNMENT << 1;
+	adev->doorbell_index.sdma_doorbell_range = 20;
+}
+
+static void soc21_pre_asic_init(struct amdgpu_device *adev)
+{
+}
+
+static int soc21_update_umd_stable_pstate(struct amdgpu_device *adev,
+					  bool enter)
+{
+	if (enter)
+		amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
+	else
+		amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
+
+	if (adev->gfx.funcs->update_perfmon_mgcg)
+		adev->gfx.funcs->update_perfmon_mgcg(adev, !enter);
+
+	return 0;
+}
+
+static const struct amdgpu_asic_funcs soc21_asic_funcs = {
+	.read_disabled_bios = &soc21_read_disabled_bios,
+	.read_bios_from_rom = &amdgpu_soc15_read_bios_from_rom,
+	.read_register = &soc21_read_register,
+	.reset = &soc21_asic_reset,
+	.reset_method = &soc21_asic_reset_method,
+	.get_xclk = &soc21_get_xclk,
+	.set_uvd_clocks = &soc21_set_uvd_clocks,
+	.set_vce_clocks = &soc21_set_vce_clocks,
+	.get_config_memsize = &soc21_get_config_memsize,
+	.init_doorbell_index = &soc21_init_doorbell_index,
+	.need_full_reset = &soc21_need_full_reset,
+	.need_reset_on_init = &soc21_need_reset_on_init,
+	.get_pcie_replay_count = &amdgpu_nbio_get_pcie_replay_count,
+	.supports_baco = &amdgpu_dpm_is_baco_supported,
+	.pre_asic_init = &soc21_pre_asic_init,
+	.query_video_codecs = &soc21_query_video_codecs,
+	.update_umd_stable_pstate = &soc21_update_umd_stable_pstate,
+};
+
+static int soc21_common_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	adev->nbio.funcs->set_reg_remap(adev);
+	adev->smc_rreg = NULL;
+	adev->smc_wreg = NULL;
+	adev->pcie_rreg = &amdgpu_device_indirect_rreg;
+	adev->pcie_wreg = &amdgpu_device_indirect_wreg;
+	adev->pcie_rreg64 = &amdgpu_device_indirect_rreg64;
+	adev->pcie_wreg64 = &amdgpu_device_indirect_wreg64;
+	adev->pciep_rreg = amdgpu_device_pcie_port_rreg;
+	adev->pciep_wreg = amdgpu_device_pcie_port_wreg;
+
+	/* TODO: will add them during VCN v2 implementation */
+	adev->uvd_ctx_rreg = NULL;
+	adev->uvd_ctx_wreg = NULL;
+
+	adev->didt_rreg = &soc21_didt_rreg;
+	adev->didt_wreg = &soc21_didt_wreg;
+
+	adev->asic_funcs = &soc21_asic_funcs;
+
+	adev->rev_id = amdgpu_device_get_rev_id(adev);
+	adev->external_rev_id = 0xff;
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(11, 0, 0):
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+#if 0
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGLS |
+#endif
+			AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_REPEATER_FGCG |
+			AMD_CG_SUPPORT_GFX_FGCG |
+			AMD_CG_SUPPORT_GFX_PERF_CLK |
+			AMD_CG_SUPPORT_VCN_MGCG |
+			AMD_CG_SUPPORT_JPEG_MGCG |
+			AMD_CG_SUPPORT_ATHUB_MGCG |
+			AMD_CG_SUPPORT_ATHUB_LS |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_IH_CG |
+			AMD_CG_SUPPORT_HDP_SD;
+		adev->pg_flags = AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_VCN_DPG |
+			AMD_PG_SUPPORT_JPEG |
+			AMD_PG_SUPPORT_ATHUB |
+			AMD_PG_SUPPORT_MMHUB;
+		adev->external_rev_id = adev->rev_id + 0x1; // TODO: need update
+		break;
+	case IP_VERSION(11, 0, 2):
+		adev->cg_flags =
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_REPEATER_FGCG |
+			AMD_CG_SUPPORT_VCN_MGCG |
+			AMD_CG_SUPPORT_JPEG_MGCG |
+			AMD_CG_SUPPORT_ATHUB_MGCG |
+			AMD_CG_SUPPORT_ATHUB_LS |
+			AMD_CG_SUPPORT_IH_CG |
+			AMD_CG_SUPPORT_HDP_SD;
+		adev->pg_flags =
+			AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_VCN_DPG |
+			AMD_PG_SUPPORT_JPEG |
+			AMD_PG_SUPPORT_ATHUB |
+			AMD_PG_SUPPORT_MMHUB;
+		adev->external_rev_id = adev->rev_id + 0x10;
+		break;
+	case IP_VERSION(11, 0, 1):
+		adev->cg_flags =
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_FGCG |
+			AMD_CG_SUPPORT_REPEATER_FGCG |
+			AMD_CG_SUPPORT_GFX_PERF_CLK |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_ATHUB_MGCG |
+			AMD_CG_SUPPORT_ATHUB_LS |
+			AMD_CG_SUPPORT_IH_CG |
+			AMD_CG_SUPPORT_BIF_MGCG |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_VCN_MGCG |
+			AMD_CG_SUPPORT_JPEG_MGCG;
+		adev->pg_flags =
+			AMD_PG_SUPPORT_GFX_PG |
+			AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_VCN_DPG |
+			AMD_PG_SUPPORT_JPEG;
+		adev->external_rev_id = adev->rev_id + 0x1;
+		break;
+	case IP_VERSION(11, 0, 3):
+		adev->cg_flags = AMD_CG_SUPPORT_VCN_MGCG |
+			AMD_CG_SUPPORT_JPEG_MGCG |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_REPEATER_FGCG |
+			AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_HDP_SD |
+			AMD_CG_SUPPORT_ATHUB_MGCG |
+			AMD_CG_SUPPORT_ATHUB_LS;
+		adev->pg_flags = AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_VCN_DPG |
+			AMD_PG_SUPPORT_JPEG;
+		adev->external_rev_id = adev->rev_id + 0x20;
+		break;
+	case IP_VERSION(11, 0, 4):
+		adev->cg_flags =
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_FGCG |
+			AMD_CG_SUPPORT_REPEATER_FGCG |
+			AMD_CG_SUPPORT_GFX_PERF_CLK |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_ATHUB_MGCG |
+			AMD_CG_SUPPORT_ATHUB_LS |
+			AMD_CG_SUPPORT_IH_CG |
+			AMD_CG_SUPPORT_BIF_MGCG |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_VCN_MGCG |
+			AMD_CG_SUPPORT_JPEG_MGCG;
+		adev->pg_flags = AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_VCN_DPG |
+			AMD_PG_SUPPORT_GFX_PG |
+			AMD_PG_SUPPORT_JPEG;
+		adev->external_rev_id = adev->rev_id + 0x80;
+		break;
+	case IP_VERSION(11, 5, 0):
+		adev->cg_flags = AMD_CG_SUPPORT_VCN_MGCG |
+			AMD_CG_SUPPORT_JPEG_MGCG |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_FGCG |
+			AMD_CG_SUPPORT_REPEATER_FGCG |
+			AMD_CG_SUPPORT_GFX_PERF_CLK	|
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGLS	|
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_HDP_DS |
+			AMD_CG_SUPPORT_HDP_SD |
+			AMD_CG_SUPPORT_ATHUB_MGCG |
+			AMD_CG_SUPPORT_ATHUB_LS |
+			AMD_CG_SUPPORT_IH_CG |
+			AMD_CG_SUPPORT_BIF_MGCG |
+			AMD_CG_SUPPORT_BIF_LS;
+		adev->pg_flags = AMD_PG_SUPPORT_VCN_DPG |
+			AMD_PG_SUPPORT_JPEG_DPG |
+			AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_JPEG |
+			AMD_PG_SUPPORT_GFX_PG;
+		if (adev->rev_id == 0)
+			adev->external_rev_id = 0x1;
+		else
+			adev->external_rev_id = adev->rev_id + 0x10;
+		break;
+	case IP_VERSION(11, 5, 1):
+		adev->cg_flags =
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_FGCG |
+			AMD_CG_SUPPORT_REPEATER_FGCG |
+			AMD_CG_SUPPORT_GFX_PERF_CLK	|
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGLS	|
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_HDP_DS |
+			AMD_CG_SUPPORT_HDP_SD |
+			AMD_CG_SUPPORT_ATHUB_MGCG |
+			AMD_CG_SUPPORT_ATHUB_LS |
+			AMD_CG_SUPPORT_IH_CG |
+			AMD_CG_SUPPORT_BIF_MGCG |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_VCN_MGCG |
+			AMD_CG_SUPPORT_JPEG_MGCG;
+		adev->pg_flags =
+			AMD_PG_SUPPORT_GFX_PG |
+			AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_VCN_DPG |
+			AMD_PG_SUPPORT_JPEG;
+		adev->external_rev_id = adev->rev_id + 0xc1;
+		break;
+	case IP_VERSION(11, 5, 2):
+		adev->cg_flags = AMD_CG_SUPPORT_VCN_MGCG |
+			AMD_CG_SUPPORT_JPEG_MGCG |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_FGCG |
+			AMD_CG_SUPPORT_REPEATER_FGCG |
+			AMD_CG_SUPPORT_GFX_PERF_CLK	|
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGLS	|
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_HDP_DS |
+			AMD_CG_SUPPORT_HDP_SD |
+			AMD_CG_SUPPORT_ATHUB_MGCG |
+			AMD_CG_SUPPORT_ATHUB_LS |
+			AMD_CG_SUPPORT_IH_CG |
+			AMD_CG_SUPPORT_BIF_MGCG |
+			AMD_CG_SUPPORT_BIF_LS;
+		adev->pg_flags = AMD_PG_SUPPORT_VCN_DPG |
+			AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_JPEG_DPG |
+			AMD_PG_SUPPORT_JPEG |
+			AMD_PG_SUPPORT_GFX_PG;
+		adev->external_rev_id = adev->rev_id + 0x40;
+		break;
+	case IP_VERSION(11, 5, 3):
+		adev->cg_flags = AMD_CG_SUPPORT_VCN_MGCG |
+			AMD_CG_SUPPORT_JPEG_MGCG |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_FGCG |
+			AMD_CG_SUPPORT_REPEATER_FGCG |
+			AMD_CG_SUPPORT_GFX_PERF_CLK |
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGLS |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_HDP_DS |
+			AMD_CG_SUPPORT_HDP_SD |
+			AMD_CG_SUPPORT_ATHUB_MGCG |
+			AMD_CG_SUPPORT_ATHUB_LS |
+			AMD_CG_SUPPORT_IH_CG |
+			AMD_CG_SUPPORT_BIF_MGCG |
+			AMD_CG_SUPPORT_BIF_LS;
+		adev->pg_flags = AMD_PG_SUPPORT_VCN_DPG |
+			AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_JPEG_DPG |
+			AMD_PG_SUPPORT_JPEG |
+			AMD_PG_SUPPORT_GFX_PG;
+		adev->external_rev_id = adev->rev_id + 0x50;
+		break;
+	default:
+		/* FIXME: not supported yet */
+		return -EINVAL;
+	}
+
+	if (amdgpu_sriov_vf(adev)) {
+		amdgpu_virt_init_setting(adev);
+		xgpu_nv_mailbox_set_irq_funcs(adev);
+	}
+
+	return 0;
+}
+
+static int soc21_common_late_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_sriov_vf(adev)) {
+		xgpu_nv_mailbox_get_irq(adev);
+		if ((adev->vcn.harvest_config & AMDGPU_VCN_HARVEST_VCN0) ||
+		!amdgpu_sriov_is_av1_support(adev)) {
+			amdgpu_virt_update_sriov_video_codec(adev,
+							     sriov_vcn_4_0_0_video_codecs_encode_array_vcn1,
+							     ARRAY_SIZE(sriov_vcn_4_0_0_video_codecs_encode_array_vcn1),
+							     sriov_vcn_4_0_0_video_codecs_decode_array_vcn1,
+							     ARRAY_SIZE(sriov_vcn_4_0_0_video_codecs_decode_array_vcn1));
+		} else {
+			amdgpu_virt_update_sriov_video_codec(adev,
+							     sriov_vcn_4_0_0_video_codecs_encode_array_vcn0,
+							     ARRAY_SIZE(sriov_vcn_4_0_0_video_codecs_encode_array_vcn0),
+							     sriov_vcn_4_0_0_video_codecs_decode_array_vcn0,
+							     ARRAY_SIZE(sriov_vcn_4_0_0_video_codecs_decode_array_vcn0));
+		}
+	} else {
+		if (adev->nbio.ras &&
+		    adev->nbio.ras_err_event_athub_irq.funcs)
+			/* don't need to fail gpu late init
+			 * if enabling athub_err_event interrupt failed
+			 * nbio v4_3 only support fatal error hanlding
+			 * just enable the interrupt directly */
+			amdgpu_irq_get(adev, &adev->nbio.ras_err_event_athub_irq, 0);
+	}
+
+	/* Enable selfring doorbell aperture late because doorbell BAR
+	 * aperture will change if resize BAR successfully in gmc sw_init.
+	 */
+	adev->nbio.funcs->enable_doorbell_selfring_aperture(adev, true);
+
+	return 0;
+}
+
+static int soc21_common_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_sriov_vf(adev))
+		xgpu_nv_mailbox_add_irq_id(adev);
+
+	return 0;
+}
+
+static int soc21_common_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	/* enable aspm */
+	soc21_program_aspm(adev);
+	/* setup nbio registers */
+	adev->nbio.funcs->init_registers(adev);
+	/* remap HDP registers to a hole in mmio space,
+	 * for the purpose of expose those registers
+	 * to process space
+	 */
+	if (adev->nbio.funcs->remap_hdp_registers && !amdgpu_sriov_vf(adev))
+		adev->nbio.funcs->remap_hdp_registers(adev);
+	/* enable the doorbell aperture */
+	adev->nbio.funcs->enable_doorbell_aperture(adev, true);
+
+	return 0;
+}
+
+static int soc21_common_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	/* Disable the doorbell aperture and selfring doorbell aperture
+	 * separately in hw_fini because soc21_enable_doorbell_aperture
+	 * has been removed and there is no need to delay disabling
+	 * selfring doorbell.
+	 */
+	adev->nbio.funcs->enable_doorbell_aperture(adev, false);
+	adev->nbio.funcs->enable_doorbell_selfring_aperture(adev, false);
+
+	if (amdgpu_sriov_vf(adev)) {
+		xgpu_nv_mailbox_put_irq(adev);
+	} else {
+		if (adev->nbio.ras &&
+		    adev->nbio.ras_err_event_athub_irq.funcs)
+			amdgpu_irq_put(adev, &adev->nbio.ras_err_event_athub_irq, 0);
+	}
+
+	return 0;
+}
+
+static int soc21_common_suspend(struct amdgpu_ip_block *ip_block)
+{
+	return soc21_common_hw_fini(ip_block);
+}
+
+static bool soc21_need_reset_on_resume(struct amdgpu_device *adev)
+{
+	u32 sol_reg1, sol_reg2;
+
+	/* Will reset for the following suspend abort cases.
+	 * 1) Only reset dGPU side.
+	 * 2) S3 suspend got aborted and TOS is active.
+	 *    As for dGPU suspend abort cases the SOL value
+	 *    will be kept as zero at this resume point.
+	 */
+	if (!(adev->flags & AMD_IS_APU) && adev->in_s3) {
+		sol_reg1 = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_81);
+		msleep(100);
+		sol_reg2 = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_81);
+
+		return (sol_reg1 != sol_reg2);
+	}
+
+	return false;
+}
+
+static int soc21_common_resume(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (soc21_need_reset_on_resume(adev)) {
+		dev_info(adev->dev, "S3 suspend aborted, resetting...");
+		soc21_asic_reset(adev);
+	}
+
+	return soc21_common_hw_init(ip_block);
+}
+
+static bool soc21_common_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	return true;
+}
+
+static int soc21_common_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					   enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) {
+	case IP_VERSION(4, 3, 0):
+	case IP_VERSION(4, 3, 1):
+	case IP_VERSION(7, 7, 0):
+	case IP_VERSION(7, 7, 1):
+	case IP_VERSION(7, 11, 0):
+	case IP_VERSION(7, 11, 1):
+	case IP_VERSION(7, 11, 2):
+	case IP_VERSION(7, 11, 3):
+		adev->nbio.funcs->update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		adev->nbio.funcs->update_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+		adev->hdp.funcs->update_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int soc21_common_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					   enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	switch (amdgpu_ip_version(adev, LSDMA_HWIP, 0)) {
+	case IP_VERSION(6, 0, 0):
+	case IP_VERSION(6, 0, 2):
+		adev->lsdma.funcs->update_memory_power_gating(adev,
+				state == AMD_PG_STATE_GATE);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static void soc21_common_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	adev->nbio.funcs->get_clockgating_state(adev, flags);
+
+	adev->hdp.funcs->get_clock_gating_state(adev, flags);
+}
+
+static const struct amd_ip_funcs soc21_common_ip_funcs = {
+	.name = "soc21_common",
+	.early_init = soc21_common_early_init,
+	.late_init = soc21_common_late_init,
+	.sw_init = soc21_common_sw_init,
+	.hw_init = soc21_common_hw_init,
+	.hw_fini = soc21_common_hw_fini,
+	.suspend = soc21_common_suspend,
+	.resume = soc21_common_resume,
+	.is_idle = soc21_common_is_idle,
+	.set_clockgating_state = soc21_common_set_clockgating_state,
+	.set_powergating_state = soc21_common_set_powergating_state,
+	.get_clockgating_state = soc21_common_get_clockgating_state,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/soc21.h b/drivers/gpu/drm/amd/amdgpu/soc21.h
new file mode 100644
index 000000000000..4c8067af1b65
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/soc21.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __SOC21_H__
+#define __SOC21_H__
+
+extern const struct amdgpu_ip_block_version soc21_common_ip_block;
+
+void soc21_grbm_select(struct amdgpu_device *adev,
+		    u32 me, u32 pipe, u32 queue, u32 vmid);
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/soc24.c b/drivers/gpu/drm/amd/amdgpu/soc24.c
new file mode 100644
index 000000000000..972b449ab89f
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/soc24.c
@@ -0,0 +1,601 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/firmware.h>
+#include <linux/slab.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+#include "amdgpu_uvd.h"
+#include "amdgpu_vce.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_psp.h"
+#include "amdgpu_smu.h"
+#include "atom.h"
+#include "amd_pcie.h"
+
+#include "gc/gc_12_0_0_offset.h"
+#include "gc/gc_12_0_0_sh_mask.h"
+#include "mp/mp_14_0_2_offset.h"
+
+#include "soc15.h"
+#include "soc15_common.h"
+#include "soc24.h"
+#include "mxgpu_nv.h"
+
+static const struct amd_ip_funcs soc24_common_ip_funcs;
+
+static const struct amdgpu_video_codec_info vcn_5_0_0_video_codecs_encode_array_vcn0[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1, 8192, 4352, 0)},
+};
+
+static const struct amdgpu_video_codecs vcn_5_0_0_video_codecs_encode_vcn0 = {
+	.codec_count = ARRAY_SIZE(vcn_5_0_0_video_codecs_encode_array_vcn0),
+	.codec_array = vcn_5_0_0_video_codecs_encode_array_vcn0,
+};
+
+static const struct amdgpu_video_codec_info vcn_5_0_0_video_codecs_decode_array_vcn0[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1, 8192, 4352, 0)},
+};
+
+static const struct amdgpu_video_codecs vcn_5_0_0_video_codecs_decode_vcn0 = {
+	.codec_count = ARRAY_SIZE(vcn_5_0_0_video_codecs_decode_array_vcn0),
+	.codec_array = vcn_5_0_0_video_codecs_decode_array_vcn0,
+};
+
+static int soc24_query_video_codecs(struct amdgpu_device *adev, bool encode,
+				 const struct amdgpu_video_codecs **codecs)
+{
+	if (adev->vcn.num_vcn_inst == hweight8(adev->vcn.harvest_config))
+		return -EINVAL;
+
+	switch (amdgpu_ip_version(adev, UVD_HWIP, 0)) {
+	case IP_VERSION(5, 0, 0):
+		if (encode)
+			*codecs = &vcn_5_0_0_video_codecs_encode_vcn0;
+		else
+			*codecs = &vcn_5_0_0_video_codecs_decode_vcn0;
+		return 0;
+	default:
+		return -EINVAL;
+	}
+}
+
+static u32 soc24_get_config_memsize(struct amdgpu_device *adev)
+{
+	return adev->nbio.funcs->get_memsize(adev);
+}
+
+static u32 soc24_get_xclk(struct amdgpu_device *adev)
+{
+	return adev->clock.spll.reference_freq;
+}
+
+void soc24_grbm_select(struct amdgpu_device *adev,
+		       u32 me, u32 pipe, u32 queue, u32 vmid)
+{
+	u32 grbm_gfx_cntl = 0;
+	grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, PIPEID, pipe);
+	grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, MEID, me);
+	grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, VMID, vmid);
+	grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, QUEUEID, queue);
+
+	WREG32_SOC15(GC, 0, regGRBM_GFX_CNTL, grbm_gfx_cntl);
+}
+
+static struct soc15_allowed_register_entry soc24_allowed_read_registers[] = {
+	{ SOC15_REG_ENTRY(GC, 0, regGRBM_STATUS)},
+	{ SOC15_REG_ENTRY(GC, 0, regGRBM_STATUS2)},
+	{ SOC15_REG_ENTRY(GC, 0, regGRBM_STATUS_SE0)},
+	{ SOC15_REG_ENTRY(GC, 0, regGRBM_STATUS_SE1)},
+	{ SOC15_REG_ENTRY(GC, 0, regGRBM_STATUS_SE2)},
+	{ SOC15_REG_ENTRY(GC, 0, regGRBM_STATUS_SE3)},
+	{ SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_STATUS_REG)},
+	{ SOC15_REG_ENTRY(SDMA1, 0, regSDMA1_STATUS_REG)},
+	{ SOC15_REG_ENTRY(GC, 0, regCP_STAT)},
+	{ SOC15_REG_ENTRY(GC, 0, regCP_STALLED_STAT1)},
+	{ SOC15_REG_ENTRY(GC, 0, regCP_STALLED_STAT2)},
+	{ SOC15_REG_ENTRY(GC, 0, regCP_STALLED_STAT3)},
+	{ SOC15_REG_ENTRY(GC, 0, regCP_CPF_BUSY_STAT)},
+	{ SOC15_REG_ENTRY(GC, 0, regCP_CPF_STALLED_STAT1)},
+	{ SOC15_REG_ENTRY(GC, 0, regCP_CPF_STATUS)},
+	{ SOC15_REG_ENTRY(GC, 0, regCP_CPC_BUSY_STAT)},
+	{ SOC15_REG_ENTRY(GC, 0, regCP_CPC_STALLED_STAT1)},
+	{ SOC15_REG_ENTRY(GC, 0, regCP_CPC_STATUS)},
+	{ SOC15_REG_ENTRY(GC, 0, regGB_ADDR_CONFIG)},
+};
+
+static uint32_t soc24_read_indexed_register(struct amdgpu_device *adev,
+					    u32 se_num,
+					    u32 sh_num,
+					    u32 reg_offset)
+{
+	uint32_t val;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	if (se_num != 0xffffffff || sh_num != 0xffffffff)
+		amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff, 0);
+
+	val = RREG32(reg_offset);
+
+	if (se_num != 0xffffffff || sh_num != 0xffffffff)
+		amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
+	mutex_unlock(&adev->grbm_idx_mutex);
+	return val;
+}
+
+static uint32_t soc24_get_register_value(struct amdgpu_device *adev,
+					 bool indexed, u32 se_num,
+					 u32 sh_num, u32 reg_offset)
+{
+	if (indexed) {
+		return soc24_read_indexed_register(adev, se_num, sh_num, reg_offset);
+	} else {
+		if (reg_offset == SOC15_REG_OFFSET(GC, 0, regGB_ADDR_CONFIG) &&
+		    adev->gfx.config.gb_addr_config)
+			return adev->gfx.config.gb_addr_config;
+		return RREG32(reg_offset);
+	}
+}
+
+static int soc24_read_register(struct amdgpu_device *adev, u32 se_num,
+			       u32 sh_num, u32 reg_offset, u32 *value)
+{
+	uint32_t i;
+	struct soc15_allowed_register_entry  *en;
+
+	*value = 0;
+	for (i = 0; i < ARRAY_SIZE(soc24_allowed_read_registers); i++) {
+		en = &soc24_allowed_read_registers[i];
+		if (!adev->reg_offset[en->hwip][en->inst])
+			continue;
+		else if (reg_offset != (adev->reg_offset[en->hwip][en->inst][en->seg]
+					+ en->reg_offset))
+			continue;
+
+		*value = soc24_get_register_value(adev,
+				soc24_allowed_read_registers[i].grbm_indexed,
+				se_num, sh_num, reg_offset);
+		return 0;
+	}
+	return -EINVAL;
+}
+
+static enum amd_reset_method
+soc24_asic_reset_method(struct amdgpu_device *adev)
+{
+	if (amdgpu_reset_method == AMD_RESET_METHOD_MODE1 ||
+	    amdgpu_reset_method == AMD_RESET_METHOD_MODE2 ||
+	    amdgpu_reset_method == AMD_RESET_METHOD_BACO)
+		return amdgpu_reset_method;
+
+	if (amdgpu_reset_method != -1)
+		dev_warn(adev->dev,
+			 "Specified reset method:%d isn't supported, using AUTO instead.\n",
+			 amdgpu_reset_method);
+
+	switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) {
+	case IP_VERSION(14, 0, 2):
+	case IP_VERSION(14, 0, 3):
+		return AMD_RESET_METHOD_MODE1;
+	default:
+		if (amdgpu_dpm_is_baco_supported(adev))
+			return AMD_RESET_METHOD_BACO;
+		else
+			return AMD_RESET_METHOD_MODE1;
+	}
+}
+
+static int soc24_asic_reset(struct amdgpu_device *adev)
+{
+	int ret = 0;
+
+	switch (soc24_asic_reset_method(adev)) {
+	case AMD_RESET_METHOD_PCI:
+		dev_info(adev->dev, "PCI reset\n");
+		ret = amdgpu_device_pci_reset(adev);
+		break;
+	case AMD_RESET_METHOD_BACO:
+		dev_info(adev->dev, "BACO reset\n");
+		ret = amdgpu_dpm_baco_reset(adev);
+		break;
+	case AMD_RESET_METHOD_MODE2:
+		dev_info(adev->dev, "MODE2 reset\n");
+		ret = amdgpu_dpm_mode2_reset(adev);
+		break;
+	default:
+		dev_info(adev->dev, "MODE1 reset\n");
+		ret = amdgpu_device_mode1_reset(adev);
+		break;
+	}
+
+	return ret;
+}
+
+static void soc24_program_aspm(struct amdgpu_device *adev)
+{
+	if (!amdgpu_device_should_use_aspm(adev))
+		return;
+
+	if (!(adev->flags & AMD_IS_APU) &&
+	    (adev->nbio.funcs->program_aspm))
+		adev->nbio.funcs->program_aspm(adev);
+}
+
+const struct amdgpu_ip_block_version soc24_common_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_COMMON,
+	.major = 1,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &soc24_common_ip_funcs,
+};
+
+static bool soc24_need_full_reset(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 0, 1):
+	default:
+		return true;
+	}
+}
+
+static bool soc24_need_reset_on_init(struct amdgpu_device *adev)
+{
+	u32 sol_reg;
+
+	if (adev->flags & AMD_IS_APU)
+		return false;
+
+	/* Check sOS sign of life register to confirm sys driver and sOS
+	 * are already been loaded.
+	 */
+	sol_reg = RREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_81);
+	if (sol_reg)
+		return true;
+
+	return false;
+}
+
+static uint64_t soc24_get_pcie_replay_count(struct amdgpu_device *adev)
+{
+	/* TODO
+	 * dummy implement for pcie_replay_count sysfs interface
+	 * */
+	return 0;
+}
+
+static void soc24_init_doorbell_index(struct amdgpu_device *adev)
+{
+	adev->doorbell_index.kiq = AMDGPU_NAVI10_DOORBELL_KIQ;
+	adev->doorbell_index.mec_ring0 = AMDGPU_NAVI10_DOORBELL_MEC_RING0;
+	adev->doorbell_index.mec_ring1 = AMDGPU_NAVI10_DOORBELL_MEC_RING1;
+	adev->doorbell_index.mec_ring2 = AMDGPU_NAVI10_DOORBELL_MEC_RING2;
+	adev->doorbell_index.mec_ring3 = AMDGPU_NAVI10_DOORBELL_MEC_RING3;
+	adev->doorbell_index.mec_ring4 = AMDGPU_NAVI10_DOORBELL_MEC_RING4;
+	adev->doorbell_index.mec_ring5 = AMDGPU_NAVI10_DOORBELL_MEC_RING5;
+	adev->doorbell_index.mec_ring6 = AMDGPU_NAVI10_DOORBELL_MEC_RING6;
+	adev->doorbell_index.mec_ring7 = AMDGPU_NAVI10_DOORBELL_MEC_RING7;
+	adev->doorbell_index.userqueue_start = AMDGPU_NAVI10_DOORBELL_USERQUEUE_START;
+	adev->doorbell_index.userqueue_end = AMDGPU_NAVI10_DOORBELL_USERQUEUE_END;
+	adev->doorbell_index.gfx_ring0 = AMDGPU_NAVI10_DOORBELL_GFX_RING0;
+	adev->doorbell_index.gfx_ring1 = AMDGPU_NAVI10_DOORBELL_GFX_RING1;
+	adev->doorbell_index.gfx_userqueue_start =
+		AMDGPU_NAVI10_DOORBELL_GFX_USERQUEUE_START;
+	adev->doorbell_index.gfx_userqueue_end =
+		AMDGPU_NAVI10_DOORBELL_GFX_USERQUEUE_END;
+	adev->doorbell_index.mes_ring0 = AMDGPU_NAVI10_DOORBELL_MES_RING0;
+	adev->doorbell_index.mes_ring1 = AMDGPU_NAVI10_DOORBELL_MES_RING1;
+	adev->doorbell_index.sdma_engine[0] = AMDGPU_NAVI10_DOORBELL_sDMA_ENGINE0;
+	adev->doorbell_index.sdma_engine[1] = AMDGPU_NAVI10_DOORBELL_sDMA_ENGINE1;
+	adev->doorbell_index.ih = AMDGPU_NAVI10_DOORBELL_IH;
+	adev->doorbell_index.vcn.vcn_ring0_1 = AMDGPU_NAVI10_DOORBELL64_VCN0_1;
+	adev->doorbell_index.vcn.vcn_ring2_3 = AMDGPU_NAVI10_DOORBELL64_VCN2_3;
+	adev->doorbell_index.vcn.vcn_ring4_5 = AMDGPU_NAVI10_DOORBELL64_VCN4_5;
+	adev->doorbell_index.vcn.vcn_ring6_7 = AMDGPU_NAVI10_DOORBELL64_VCN6_7;
+	adev->doorbell_index.first_non_cp = AMDGPU_NAVI10_DOORBELL64_FIRST_NON_CP;
+	adev->doorbell_index.last_non_cp = AMDGPU_NAVI10_DOORBELL64_LAST_NON_CP;
+
+	adev->doorbell_index.max_assignment = AMDGPU_NAVI10_DOORBELL_MAX_ASSIGNMENT << 1;
+	adev->doorbell_index.sdma_doorbell_range = 20;
+}
+
+static void soc24_pre_asic_init(struct amdgpu_device *adev)
+{
+}
+
+static int soc24_update_umd_stable_pstate(struct amdgpu_device *adev,
+					  bool enter)
+{
+	if (enter)
+		amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
+	else
+		amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
+
+	if (adev->gfx.funcs->update_perfmon_mgcg)
+		adev->gfx.funcs->update_perfmon_mgcg(adev, !enter);
+
+	return 0;
+}
+
+static const struct amdgpu_asic_funcs soc24_asic_funcs = {
+	.read_bios_from_rom = &amdgpu_soc15_read_bios_from_rom,
+	.read_register = &soc24_read_register,
+	.reset = &soc24_asic_reset,
+	.reset_method = &soc24_asic_reset_method,
+	.get_xclk = &soc24_get_xclk,
+	.get_config_memsize = &soc24_get_config_memsize,
+	.init_doorbell_index = &soc24_init_doorbell_index,
+	.need_full_reset = &soc24_need_full_reset,
+	.need_reset_on_init = &soc24_need_reset_on_init,
+	.get_pcie_replay_count = &soc24_get_pcie_replay_count,
+	.supports_baco = &amdgpu_dpm_is_baco_supported,
+	.pre_asic_init = &soc24_pre_asic_init,
+	.query_video_codecs = &soc24_query_video_codecs,
+	.update_umd_stable_pstate = &soc24_update_umd_stable_pstate,
+};
+
+static int soc24_common_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	adev->nbio.funcs->set_reg_remap(adev);
+	adev->smc_rreg = NULL;
+	adev->smc_wreg = NULL;
+	adev->pcie_rreg = &amdgpu_device_indirect_rreg;
+	adev->pcie_wreg = &amdgpu_device_indirect_wreg;
+	adev->pcie_rreg64 = &amdgpu_device_indirect_rreg64;
+	adev->pcie_wreg64 = &amdgpu_device_indirect_wreg64;
+	adev->pciep_rreg = amdgpu_device_pcie_port_rreg;
+	adev->pciep_wreg = amdgpu_device_pcie_port_wreg;
+	adev->uvd_ctx_rreg = NULL;
+	adev->uvd_ctx_wreg = NULL;
+	adev->didt_rreg = NULL;
+	adev->didt_wreg = NULL;
+
+	adev->asic_funcs = &soc24_asic_funcs;
+
+	adev->rev_id = amdgpu_device_get_rev_id(adev);
+	adev->external_rev_id = 0xff;
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(12, 0, 0):
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGLS |
+			AMD_CG_SUPPORT_REPEATER_FGCG |
+			AMD_CG_SUPPORT_GFX_FGCG |
+			AMD_CG_SUPPORT_GFX_PERF_CLK |
+			AMD_CG_SUPPORT_ATHUB_MGCG |
+			AMD_CG_SUPPORT_ATHUB_LS |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_HDP_SD |
+			AMD_CG_SUPPORT_MC_LS;
+		adev->pg_flags = AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_JPEG |
+			AMD_PG_SUPPORT_VCN_DPG;
+		adev->external_rev_id = adev->rev_id + 0x40;
+		break;
+	case IP_VERSION(12, 0, 1):
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGLS |
+			AMD_CG_SUPPORT_REPEATER_FGCG |
+			AMD_CG_SUPPORT_GFX_FGCG |
+			AMD_CG_SUPPORT_GFX_PERF_CLK |
+			AMD_CG_SUPPORT_ATHUB_MGCG |
+			AMD_CG_SUPPORT_ATHUB_LS |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_HDP_SD |
+			AMD_CG_SUPPORT_MC_LS;
+
+		adev->pg_flags = AMD_PG_SUPPORT_VCN |
+			AMD_PG_SUPPORT_JPEG |
+			AMD_PG_SUPPORT_JPEG_DPG |
+			AMD_PG_SUPPORT_VCN_DPG;
+		adev->external_rev_id = adev->rev_id + 0x50;
+		break;
+	default:
+		/* FIXME: not supported yet */
+		return -EINVAL;
+	}
+
+	if (amdgpu_sriov_vf(adev)) {
+		amdgpu_virt_init_setting(adev);
+		xgpu_nv_mailbox_set_irq_funcs(adev);
+	}
+
+	return 0;
+}
+
+static int soc24_common_late_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_sriov_vf(adev)) {
+		xgpu_nv_mailbox_get_irq(adev);
+	} else {
+		if (adev->nbio.ras &&
+		    adev->nbio.ras_err_event_athub_irq.funcs)
+			/* don't need to fail gpu late init
+			 * if enabling athub_err_event interrupt failed
+			 * nbif v6_3_1 only support fatal error hanlding
+			 * just enable the interrupt directly
+			 */
+			amdgpu_irq_get(adev, &adev->nbio.ras_err_event_athub_irq, 0);
+	}
+
+	/* Enable selfring doorbell aperture late because doorbell BAR
+	 * aperture will change if resize BAR successfully in gmc sw_init.
+	 */
+	adev->nbio.funcs->enable_doorbell_selfring_aperture(adev, true);
+
+	return 0;
+}
+
+static int soc24_common_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_sriov_vf(adev))
+		xgpu_nv_mailbox_add_irq_id(adev);
+
+	return 0;
+}
+
+static int soc24_common_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	/* enable aspm */
+	soc24_program_aspm(adev);
+	/* setup nbio registers */
+	adev->nbio.funcs->init_registers(adev);
+	/* remap HDP registers to a hole in mmio space,
+	 * for the purpose of expose those registers
+	 * to process space
+	 */
+	if (adev->nbio.funcs->remap_hdp_registers)
+		adev->nbio.funcs->remap_hdp_registers(adev);
+
+	if (adev->df.funcs->hw_init)
+		adev->df.funcs->hw_init(adev);
+
+	/* enable the doorbell aperture */
+	adev->nbio.funcs->enable_doorbell_aperture(adev, true);
+
+	return 0;
+}
+
+static int soc24_common_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	/* Disable the doorbell aperture and selfring doorbell aperture
+	 * separately in hw_fini because soc21_enable_doorbell_aperture
+	 * has been removed and there is no need to delay disabling
+	 * selfring doorbell.
+	 */
+	adev->nbio.funcs->enable_doorbell_aperture(adev, false);
+	adev->nbio.funcs->enable_doorbell_selfring_aperture(adev, false);
+
+	if (amdgpu_sriov_vf(adev)) {
+		xgpu_nv_mailbox_put_irq(adev);
+	} else {
+		if (adev->nbio.ras &&
+		    adev->nbio.ras_err_event_athub_irq.funcs)
+			amdgpu_irq_put(adev, &adev->nbio.ras_err_event_athub_irq, 0);
+	}
+
+	return 0;
+}
+
+static int soc24_common_suspend(struct amdgpu_ip_block *ip_block)
+{
+	return soc24_common_hw_fini(ip_block);
+}
+
+static int soc24_common_resume(struct amdgpu_ip_block *ip_block)
+{
+	return soc24_common_hw_init(ip_block);
+}
+
+static bool soc24_common_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	return true;
+}
+
+static int soc24_common_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					      enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) {
+	case IP_VERSION(6, 3, 1):
+		adev->nbio.funcs->update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		adev->nbio.funcs->update_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+		adev->hdp.funcs->update_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int soc24_common_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					      enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	switch (amdgpu_ip_version(adev, LSDMA_HWIP, 0)) {
+	case IP_VERSION(7, 0, 0):
+	case IP_VERSION(7, 0, 1):
+		adev->lsdma.funcs->update_memory_power_gating(adev,
+				state == AMD_PG_STATE_GATE);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static void soc24_common_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	adev->nbio.funcs->get_clockgating_state(adev, flags);
+
+	adev->hdp.funcs->get_clock_gating_state(adev, flags);
+
+	return;
+}
+
+static const struct amd_ip_funcs soc24_common_ip_funcs = {
+	.name = "soc24_common",
+	.early_init = soc24_common_early_init,
+	.late_init = soc24_common_late_init,
+	.sw_init = soc24_common_sw_init,
+	.hw_init = soc24_common_hw_init,
+	.hw_fini = soc24_common_hw_fini,
+	.suspend = soc24_common_suspend,
+	.resume = soc24_common_resume,
+	.is_idle = soc24_common_is_idle,
+	.set_clockgating_state = soc24_common_set_clockgating_state,
+	.set_powergating_state = soc24_common_set_powergating_state,
+	.get_clockgating_state = soc24_common_get_clockgating_state,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/soc24.h b/drivers/gpu/drm/amd/amdgpu/soc24.h
new file mode 100644
index 000000000000..fa7e442e0b62
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/soc24.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __SOC24_H__
+#define __SOC24_H__
+
+extern const struct amdgpu_ip_block_version soc24_common_ip_block;
+
+void soc24_grbm_select(struct amdgpu_device *adev,
+		       u32 me, u32 pipe, u32 queue, u32 vmid);
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/ta_rap_if.h b/drivers/gpu/drm/amd/amdgpu/ta_rap_if.h
new file mode 100644
index 000000000000..f14833fae07c
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/ta_rap_if.h
@@ -0,0 +1,84 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef _TA_RAP_IF_H
+#define _TA_RAP_IF_H
+
+/* Responses have bit 31 set */
+#define RSP_ID_MASK (1U << 31)
+#define RSP_ID(cmdId) (((uint32_t)(cmdId)) | RSP_ID_MASK)
+
+enum ta_rap_status {
+	TA_RAP_STATUS__SUCCESS                              = 1,
+	TA_RAP_STATUS__ERROR_GENERIC_FAILURE                = 2,
+	TA_RAP_STATUS__ERROR_CMD_NOT_SUPPORTED              = 3,
+	TA_RAP_STATUS__ERROR_INVALID_VALIDATION_METHOD      = 4,
+	TA_RAP_STATUS__ERROR_NULL_POINTER                   = 5,
+	TA_RAP_STATUS__ERROR_NOT_INITIALIZED                = 6,
+	TA_RAP_STATUS__ERROR_VALIDATION_FAILED              = 7,
+	TA_RAP_STATUS__ERROR_ASIC_NOT_SUPPORTED             = 8,
+	TA_RAP_STATUS__ERROR_OPERATION_NOT_PERMISSABLE      = 9,
+	TA_RAP_STATUS__ERROR_ALREADY_INIT                   = 10,
+};
+
+enum ta_rap_cmd {
+	TA_CMD_RAP__INITIALIZE              = 1,
+	TA_CMD_RAP__VALIDATE_L0             = 2,
+};
+
+enum ta_rap_validation_method {
+	METHOD_A           = 1,
+};
+
+struct ta_rap_cmd_input_data {
+	uint8_t reserved[8];
+};
+
+struct ta_rap_cmd_output_data {
+	uint32_t    last_subsection;
+	uint32_t    num_total_validate;
+	uint32_t    num_valid;
+	uint32_t    last_validate_addr;
+	uint32_t    last_validate_val;
+	uint32_t    last_validate_val_exptd;
+};
+
+union ta_rap_cmd_input {
+	struct ta_rap_cmd_input_data input;
+};
+
+union ta_rap_cmd_output {
+	struct ta_rap_cmd_output_data output;
+};
+
+struct ta_rap_shared_memory {
+	uint32_t                    cmd_id;
+	uint32_t                    validation_method_id;
+	uint32_t                    resp_id;
+	enum ta_rap_status          rap_status;
+	union ta_rap_cmd_input      rap_in_message;
+	union ta_rap_cmd_output     rap_out_message;
+	uint8_t                     reserved[64];
+};
+
+#endif  // #define _TA_RAP_IF_H
diff --git a/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h b/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h
new file mode 100644
index 000000000000..8a3f326474e5
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h
@@ -0,0 +1,223 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef _TA_RAS_IF_H
+#define _TA_RAS_IF_H
+
+#define RAS_TA_HOST_IF_VER	0
+
+/* Responses have bit 31 set */
+#define RSP_ID_MASK (1U << 31)
+#define RSP_ID(cmdId) (((uint32_t)(cmdId)) | RSP_ID_MASK)
+
+/* invalid node instance value */
+#define TA_RAS_INV_NODE 0xffff
+
+/* RAS related enumerations */
+/**********************************************************/
+enum ras_command {
+	TA_RAS_COMMAND__ENABLE_FEATURES = 0,
+	TA_RAS_COMMAND__DISABLE_FEATURES,
+	TA_RAS_COMMAND__TRIGGER_ERROR,
+	TA_RAS_COMMAND__QUERY_BLOCK_INFO,
+	TA_RAS_COMMAND__QUERY_SUB_BLOCK_INFO,
+	TA_RAS_COMMAND__QUERY_ADDRESS,
+};
+
+enum ta_ras_status {
+	TA_RAS_STATUS__SUCCESS                          = 0x0000,
+	TA_RAS_STATUS__RESET_NEEDED                     = 0xA001,
+	TA_RAS_STATUS__ERROR_INVALID_PARAMETER          = 0xA002,
+	TA_RAS_STATUS__ERROR_RAS_NOT_AVAILABLE          = 0xA003,
+	TA_RAS_STATUS__ERROR_RAS_DUPLICATE_CMD          = 0xA004,
+	TA_RAS_STATUS__ERROR_INJECTION_FAILED           = 0xA005,
+	TA_RAS_STATUS__ERROR_ASD_READ_WRITE             = 0xA006,
+	TA_RAS_STATUS__ERROR_TOGGLE_DF_CSTATE           = 0xA007,
+	TA_RAS_STATUS__ERROR_TIMEOUT                    = 0xA008,
+	TA_RAS_STATUS__ERROR_BLOCK_DISABLED             = 0XA009,
+	TA_RAS_STATUS__ERROR_GENERIC                    = 0xA00A,
+	TA_RAS_STATUS__ERROR_RAS_MMHUB_INIT             = 0xA00B,
+	TA_RAS_STATUS__ERROR_GET_DEV_INFO               = 0xA00C,
+	TA_RAS_STATUS__ERROR_UNSUPPORTED_DEV            = 0xA00D,
+	TA_RAS_STATUS__ERROR_NOT_INITIALIZED            = 0xA00E,
+	TA_RAS_STATUS__ERROR_TEE_INTERNAL               = 0xA00F,
+	TA_RAS_STATUS__ERROR_UNSUPPORTED_FUNCTION       = 0xA010,
+	TA_RAS_STATUS__ERROR_SYS_DRV_REG_ACCESS         = 0xA011,
+	TA_RAS_STATUS__ERROR_RAS_READ_WRITE             = 0xA012,
+	TA_RAS_STATUS__ERROR_NULL_PTR                   = 0xA013,
+	TA_RAS_STATUS__ERROR_UNSUPPORTED_IP             = 0xA014,
+	TA_RAS_STATUS__ERROR_PCS_STATE_QUIET            = 0xA015,
+	TA_RAS_STATUS__ERROR_PCS_STATE_ERROR            = 0xA016,
+	TA_RAS_STATUS__ERROR_PCS_STATE_HANG             = 0xA017,
+	TA_RAS_STATUS__ERROR_PCS_STATE_UNKNOWN          = 0xA018,
+	TA_RAS_STATUS__ERROR_UNSUPPORTED_ERROR_INJ      = 0xA019,
+	TA_RAS_STATUS__TEE_ERROR_ACCESS_DENIED          = 0xA01A
+};
+
+enum ta_ras_block {
+	TA_RAS_BLOCK__UMC = 0,
+	TA_RAS_BLOCK__SDMA,
+	TA_RAS_BLOCK__GFX,
+	TA_RAS_BLOCK__MMHUB,
+	TA_RAS_BLOCK__ATHUB,
+	TA_RAS_BLOCK__PCIE_BIF,
+	TA_RAS_BLOCK__HDP,
+	TA_RAS_BLOCK__XGMI_WAFL,
+	TA_RAS_BLOCK__DF,
+	TA_RAS_BLOCK__SMN,
+	TA_RAS_BLOCK__SEM,
+	TA_RAS_BLOCK__MP0,
+	TA_RAS_BLOCK__MP1,
+	TA_RAS_BLOCK__FUSE,
+	TA_RAS_BLOCK__MCA,
+	TA_RAS_BLOCK__VCN,
+	TA_RAS_BLOCK__JPEG,
+	TA_RAS_BLOCK__IH,
+	TA_RAS_BLOCK__MPIO,
+	TA_RAS_BLOCK__MMSCH,
+	TA_NUM_BLOCK_MAX
+};
+
+enum ta_ras_mca_block {
+	TA_RAS_MCA_BLOCK__MP0   = 0,
+	TA_RAS_MCA_BLOCK__MP1   = 1,
+	TA_RAS_MCA_BLOCK__MPIO  = 2,
+	TA_RAS_MCA_BLOCK__IOHC  = 3,
+	TA_MCA_NUM_BLOCK_MAX
+};
+
+enum ta_ras_error_type {
+	TA_RAS_ERROR__NONE			= 0,
+	TA_RAS_ERROR__PARITY			= 1,
+	TA_RAS_ERROR__SINGLE_CORRECTABLE	= 2,
+	TA_RAS_ERROR__MULTI_UNCORRECTABLE	= 4,
+	TA_RAS_ERROR__POISON			= 8,
+};
+
+enum ta_ras_address_type {
+	TA_RAS_MCA_TO_PA,
+	TA_RAS_PA_TO_MCA,
+};
+
+enum ta_ras_nps_mode {
+	TA_RAS_UNKNOWN_MODE = 0,
+	TA_RAS_NPS1_MODE = 1,
+	TA_RAS_NPS2_MODE = 2,
+	TA_RAS_NPS4_MODE = 4,
+	TA_RAS_NPS8_MODE = 8,
+};
+
+/* Input/output structures for RAS commands */
+/**********************************************************/
+
+struct ta_ras_enable_features_input {
+	enum ta_ras_block	block_id;
+	enum ta_ras_error_type	error_type;
+};
+
+struct ta_ras_disable_features_input {
+	enum ta_ras_block	block_id;
+	enum ta_ras_error_type	error_type;
+};
+
+struct ta_ras_trigger_error_input {
+	enum ta_ras_block	block_id;		// ras-block. i.e. umc, gfx
+	enum ta_ras_error_type	inject_error_type;	// type of error. i.e. single_correctable
+	uint32_t		sub_block_index;	// mem block. i.e. hbm, sram etc.
+	uint64_t		address;		// explicit address of error
+	uint64_t		value;			// method if error injection. i.e persistent, coherent etc.
+};
+
+struct ta_ras_init_flags {
+	uint8_t poison_mode_en;
+	uint8_t dgpu_mode;
+	uint16_t xcc_mask;
+	uint8_t channel_dis_num;
+	uint8_t nps_mode;
+	uint32_t active_umc_mask;
+};
+
+struct ta_ras_mca_addr {
+	uint64_t err_addr;
+	uint32_t ch_inst;
+	uint32_t umc_inst;
+	uint32_t node_inst;
+	uint32_t socket_id;
+};
+
+struct ta_ras_phy_addr {
+	uint64_t pa;
+	uint32_t bank;
+	uint32_t channel_idx;
+};
+
+struct ta_ras_query_address_input {
+	enum ta_ras_address_type addr_type;
+	struct ta_ras_mca_addr ma;
+	struct ta_ras_phy_addr pa;
+};
+
+struct ta_ras_output_flags {
+	uint8_t ras_init_success_flag;
+	uint8_t err_inject_switch_disable_flag;
+	uint8_t reg_access_failure_flag;
+};
+
+struct ta_ras_query_address_output {
+	/* don't use the flags here */
+	struct ta_ras_output_flags flags;
+	struct ta_ras_mca_addr ma;
+	struct ta_ras_phy_addr pa;
+};
+
+/* Common input structure for RAS callbacks */
+/**********************************************************/
+union ta_ras_cmd_input {
+	struct ta_ras_init_flags		init_flags;
+	struct ta_ras_enable_features_input	enable_features;
+	struct ta_ras_disable_features_input	disable_features;
+	struct ta_ras_trigger_error_input	trigger_error;
+	struct ta_ras_query_address_input	address;
+
+	uint32_t reserve_pad[256];
+};
+
+union ta_ras_cmd_output {
+	struct ta_ras_output_flags flags;
+	struct ta_ras_query_address_output address;
+
+	uint32_t reserve_pad[256];
+};
+
+/* Shared Memory structures */
+/**********************************************************/
+struct ta_ras_shared_memory {
+	uint32_t		    cmd_id;
+	uint32_t		    resp_id;
+	uint32_t	    	    ras_status;
+	uint32_t		    if_version;
+	union ta_ras_cmd_input	    ras_in_message;
+	union ta_ras_cmd_output     ras_out_message;
+};
+
+#endif // TL_RAS_IF_H_
diff --git a/drivers/gpu/drm/amd/amdgpu/ta_secureDisplay_if.h b/drivers/gpu/drm/amd/amdgpu/ta_secureDisplay_if.h
new file mode 100644
index 000000000000..9ec2e03d41c7
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/ta_secureDisplay_if.h
@@ -0,0 +1,175 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef _TA_SECUREDISPLAY_IF_H
+#define _TA_SECUREDISPLAY_IF_H
+
+/** Secure Display related enumerations */
+/**********************************************************/
+
+/** @enum ta_securedisplay_command
+ *    Secure Display Command ID
+ */
+enum ta_securedisplay_command {
+	/* Query whether TA is responding. It is used only for validation purpose */
+	TA_SECUREDISPLAY_COMMAND__QUERY_TA              = 1,
+	/* Send region of Interest and CRC value to I2C */
+	TA_SECUREDISPLAY_COMMAND__SEND_ROI_CRC          = 2,
+	/* V2 to send multiple regions of Interest and CRC value to I2C */
+	TA_SECUREDISPLAY_COMMAND__SEND_ROI_CRC_V2       = 3,
+	/* Maximum Command ID */
+	TA_SECUREDISPLAY_COMMAND__MAX_ID                = 0x7FFFFFFF,
+};
+
+/** @enum ta_securedisplay_status
+ *    Secure Display status returns in shared buffer status
+ */
+enum ta_securedisplay_status {
+	TA_SECUREDISPLAY_STATUS__SUCCESS                 = 0x00,         /* Success */
+	TA_SECUREDISPLAY_STATUS__GENERIC_FAILURE         = 0x01,         /* Generic Failure */
+	TA_SECUREDISPLAY_STATUS__INVALID_PARAMETER       = 0x02,         /* Invalid Parameter */
+	TA_SECUREDISPLAY_STATUS__NULL_POINTER            = 0x03,         /* Null Pointer*/
+	TA_SECUREDISPLAY_STATUS__I2C_WRITE_ERROR         = 0x04,         /* Fail to Write to I2C */
+	TA_SECUREDISPLAY_STATUS__READ_DIO_SCRATCH_ERROR  = 0x05, /*Fail Read DIO Scratch Register*/
+	TA_SECUREDISPLAY_STATUS__READ_CRC_ERROR          = 0x06,         /* Fail to Read CRC*/
+	TA_SECUREDISPLAY_STATUS__I2C_INIT_ERROR          = 0x07,     /* Failed to initialize I2C */
+
+	TA_SECUREDISPLAY_STATUS__MAX                     = 0x7FFFFFFF,/* Maximum Value for status*/
+};
+
+/** @enum ta_securedisplay_phy_ID
+ *    Physical ID number to use for reading corresponding DIO Scratch register for ROI
+ */
+enum  ta_securedisplay_phy_ID {
+	TA_SECUREDISPLAY_PHY0                           = 0,
+	TA_SECUREDISPLAY_PHY1                           = 1,
+	TA_SECUREDISPLAY_PHY2                           = 2,
+	TA_SECUREDISPLAY_PHY3                           = 3,
+	TA_SECUREDISPLAY_MAX_PHY                        = 4,
+};
+
+/** @enum ta_securedisplay_ta_query_cmd_ret
+ *    A predefined specific reteurn value which is 0xAB only used to validate
+ *    communication to Secure Display TA is functional.
+ *    This value is used to validate whether TA is responding successfully
+ */
+enum ta_securedisplay_ta_query_cmd_ret {
+	/* This is a value to validate if TA is loaded successfully */
+	TA_SECUREDISPLAY_QUERY_CMD_RET                 = 0xAB,
+};
+
+/** @enum ta_securedisplay_buffer_size
+ *    I2C Buffer size which contains 8 bytes of ROI  (X start, X end, Y start, Y end)
+ *    and 6 bytes of CRC( R,G,B) and 1  byte for physical ID
+ */
+enum ta_securedisplay_buffer_size {
+	/* 15 bytes = 8 byte (ROI) + 6 byte(CRC) + 1 byte(phy_id) */
+	TA_SECUREDISPLAY_I2C_BUFFER_SIZE                = 15,
+	/* 16 bytes = 8 byte (ROI) + 6 byte(CRC) + 1 byte(phy_id) + 1 byte(roi_idx) */
+	TA_SECUREDISPLAY_V2_I2C_BUFFER_SIZE             = 16,
+};
+
+/** Input/output structures for Secure Display commands */
+/**********************************************************/
+/**
+ * Input structures
+ */
+
+/** @struct ta_securedisplay_send_roi_crc_input
+ *    Physical ID to determine which DIO scratch register should be used to get ROI
+ */
+struct ta_securedisplay_send_roi_crc_input {
+	/* Physical ID */
+	uint32_t  phy_id;
+};
+
+struct ta_securedisplay_send_roi_crc_v2_input {
+	/* Physical ID */
+	uint32_t phy_id;
+	/* Region of interest index */
+	uint8_t  roi_idx;
+};
+
+/** @union ta_securedisplay_cmd_input
+ *    Input buffer
+ */
+union ta_securedisplay_cmd_input {
+	/* send ROI and CRC input buffer format */
+	struct ta_securedisplay_send_roi_crc_input        send_roi_crc;
+	/* send ROI and CRC input buffer format, v2 adds a ROI index */
+	struct ta_securedisplay_send_roi_crc_v2_input     send_roi_crc_v2;
+	uint32_t                                          reserved[4];
+};
+
+/**
+ * Output structures
+ */
+
+/** @struct ta_securedisplay_query_ta_output
+ *  Output buffer format for query TA whether TA is responding used only for validation purpose
+ */
+struct ta_securedisplay_query_ta_output {
+	/* return value from TA when it is queried for validation purpose only */
+	uint32_t  query_cmd_ret;
+};
+
+/** @struct ta_securedisplay_send_roi_crc_output
+ *  Output buffer format for send ROI CRC command which will pass I2c buffer created inside TA
+ *  and used to write to I2C used only for validation purpose
+ */
+struct ta_securedisplay_send_roi_crc_output {
+	uint8_t  i2c_buf[TA_SECUREDISPLAY_I2C_BUFFER_SIZE];  /* I2C buffer */
+	uint8_t  reserved;
+};
+
+struct ta_securedisplay_send_roi_crc_v2_output {
+	uint8_t  i2c_buf[TA_SECUREDISPLAY_V2_I2C_BUFFER_SIZE];  /* I2C buffer */
+};
+
+/** @union ta_securedisplay_cmd_output
+ *    Output buffer
+ */
+union ta_securedisplay_cmd_output {
+	/* Query TA output buffer format used only for validation purpose*/
+	struct ta_securedisplay_query_ta_output            query_ta;
+	/* Send ROI CRC output buffer format used only for validation purpose */
+	struct ta_securedisplay_send_roi_crc_output        send_roi_crc;
+	/* Send ROI CRC output buffer format used only for validation purpose */
+	struct ta_securedisplay_send_roi_crc_v2_output     send_roi_crc_v2;
+	uint32_t                                           reserved[4];
+};
+
+/** @struct ta_securedisplay_cmd
+*    Secure display command which is shared buffer memory
+*/
+struct ta_securedisplay_cmd {
+    uint32_t                                           cmd_id;                         /**< +0  Bytes Command ID */
+    enum ta_securedisplay_status                       status;                         /**< +4  Bytes Status code returned by the secure display TA */
+    uint32_t                                           reserved[2];                    /**< +8  Bytes Reserved */
+    union ta_securedisplay_cmd_input                   securedisplay_in_message;       /**< +16 Bytes Command input buffer */
+    union ta_securedisplay_cmd_output                  securedisplay_out_message;      /**< +32 Bytes Command output buffer */
+    /**@note Total 48 Bytes */
+};
+
+#endif   //_TA_SECUREDISPLAY_IF_H
+
diff --git a/drivers/gpu/drm/amd/amdgpu/ta_xgmi_if.h b/drivers/gpu/drm/amd/amdgpu/ta_xgmi_if.h
new file mode 100644
index 000000000000..d5748032674e
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/ta_xgmi_if.h
@@ -0,0 +1,177 @@
+/*
+ * Copyright 2018-2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef _TA_XGMI_IF_H
+#define _TA_XGMI_IF_H
+
+/* Responses have bit 31 set */
+#define RSP_ID_MASK (1U << 31)
+#define RSP_ID(cmdId) (((uint32_t)(cmdId)) | RSP_ID_MASK)
+
+#define EXTEND_PEER_LINK_INFO_CMD_FLAG 1
+
+enum ta_command_xgmi {
+	/* Initialize the Context and Session Topology */
+	TA_COMMAND_XGMI__INITIALIZE			= 0x00,
+	/* Gets the current GPU's node ID */
+	TA_COMMAND_XGMI__GET_NODE_ID			= 0x01,
+	/* Gets the current GPU's hive ID */
+	TA_COMMAND_XGMI__GET_HIVE_ID			= 0x02,
+	/* Gets the Peer's topology Information */
+	TA_COMMAND_XGMI__GET_TOPOLOGY_INFO		= 0x03,
+	/* Sets the Peer's topology Information */
+	TA_COMMAND_XGMI__SET_TOPOLOGY_INFO		= 0x04,
+	/* Gets the total links between adjacent peer dies in hive */
+	TA_COMMAND_XGMI__GET_PEER_LINKS			= 0x0B,
+	/* Gets the total links and connected port numbers between adjacent peer dies in hive */
+	TA_COMMAND_XGMI__GET_EXTEND_PEER_LINKS		= 0x0C
+};
+
+/* XGMI related enumerations */
+/**********************************************************/;
+enum { TA_XGMI__MAX_CONNECTED_NODES = 64 };
+enum { TA_XGMI__MAX_INTERNAL_STATE = 32 };
+enum { TA_XGMI__MAX_INTERNAL_STATE_BUFFER = 128 };
+enum { TA_XGMI__MAX_PORT_NUM = 8 };
+
+enum ta_xgmi_status {
+	TA_XGMI_STATUS__SUCCESS				= 0x00,
+	TA_XGMI_STATUS__GENERIC_FAILURE			= 0x01,
+	TA_XGMI_STATUS__NULL_POINTER			= 0x02,
+	TA_XGMI_STATUS__INVALID_PARAMETER		= 0x03,
+	TA_XGMI_STATUS__NOT_INITIALIZED			= 0x04,
+	TA_XGMI_STATUS__INVALID_NODE_NUM		= 0x05,
+	TA_XGMI_STATUS__INVALID_NODE_ID			= 0x06,
+	TA_XGMI_STATUS__INVALID_TOPOLOGY		= 0x07,
+	TA_XGMI_STATUS__FAILED_ID_GEN			= 0x08,
+	TA_XGMI_STATUS__FAILED_TOPOLOGY_INIT		= 0x09,
+	TA_XGMI_STATUS__SET_SHARING_ERROR		= 0x0A
+};
+
+enum ta_xgmi_assigned_sdma_engine {
+	TA_XGMI_ASSIGNED_SDMA_ENGINE__NOT_ASSIGNED	= -1,
+	TA_XGMI_ASSIGNED_SDMA_ENGINE__SDMA0		= 0,
+	TA_XGMI_ASSIGNED_SDMA_ENGINE__SDMA1		= 1,
+	TA_XGMI_ASSIGNED_SDMA_ENGINE__SDMA2		= 2,
+	TA_XGMI_ASSIGNED_SDMA_ENGINE__SDMA3		= 3,
+	TA_XGMI_ASSIGNED_SDMA_ENGINE__SDMA4		= 4,
+	TA_XGMI_ASSIGNED_SDMA_ENGINE__SDMA5		= 5
+};
+
+/* input/output structures for XGMI commands */
+/**********************************************************/
+struct ta_xgmi_node_info {
+	uint64_t				node_id;
+	uint8_t					num_hops;
+	uint8_t					is_sharing_enabled;
+	enum ta_xgmi_assigned_sdma_engine	sdma_engine;
+};
+
+struct ta_xgmi_peer_link_info {
+	uint64_t				node_id;
+	uint8_t					num_links;
+};
+
+struct xgmi_connected_port_num {
+	uint8_t		dst_xgmi_port_num;
+	uint8_t		src_xgmi_port_num;
+};
+
+/* support both the port num and num_links */
+struct ta_xgmi_extend_peer_link_info {
+	uint64_t				node_id;
+	uint8_t					num_links;
+	struct xgmi_connected_port_num		port_num[TA_XGMI__MAX_PORT_NUM];
+};
+
+struct ta_xgmi_cmd_initialize_output {
+	uint32_t	status;
+};
+
+struct ta_xgmi_cmd_get_node_id_output {
+	uint64_t	node_id;
+};
+
+struct ta_xgmi_cmd_get_hive_id_output {
+	uint64_t	hive_id;
+};
+
+struct ta_xgmi_cmd_get_topology_info_input {
+	uint32_t			num_nodes;
+	struct ta_xgmi_node_info	nodes[TA_XGMI__MAX_CONNECTED_NODES];
+};
+
+struct ta_xgmi_cmd_get_topology_info_output {
+	uint32_t			num_nodes;
+	struct ta_xgmi_node_info	nodes[TA_XGMI__MAX_CONNECTED_NODES];
+};
+
+struct ta_xgmi_cmd_set_topology_info_input {
+	uint32_t			num_nodes;
+	struct ta_xgmi_node_info	nodes[TA_XGMI__MAX_CONNECTED_NODES];
+};
+
+/* support XGMI TA w/ and w/o port_num both so two similar structs defined */
+struct ta_xgmi_cmd_get_peer_link_info {
+	uint32_t			num_nodes;
+	struct ta_xgmi_peer_link_info	nodes[TA_XGMI__MAX_CONNECTED_NODES];
+};
+
+struct ta_xgmi_cmd_get_extend_peer_link_info {
+	uint32_t				num_nodes;
+	struct ta_xgmi_extend_peer_link_info nodes[TA_XGMI__MAX_CONNECTED_NODES];
+};
+/**********************************************************/
+/* Common input structure for XGMI callbacks */
+union ta_xgmi_cmd_input {
+	struct ta_xgmi_cmd_get_topology_info_input	get_topology_info;
+	struct ta_xgmi_cmd_set_topology_info_input	set_topology_info;
+};
+
+/* Common output structure for XGMI callbacks */
+union ta_xgmi_cmd_output {
+	struct ta_xgmi_cmd_initialize_output		initialize;
+	struct ta_xgmi_cmd_get_node_id_output		get_node_id;
+	struct ta_xgmi_cmd_get_hive_id_output		get_hive_id;
+	struct ta_xgmi_cmd_get_topology_info_output	get_topology_info;
+	struct ta_xgmi_cmd_get_peer_link_info		get_link_info;
+	struct ta_xgmi_cmd_get_extend_peer_link_info	get_extend_link_info;
+};
+
+struct ta_xgmi_shared_memory {
+	uint32_t			cmd_id;
+	uint32_t			resp_id;
+	enum ta_xgmi_status		xgmi_status;
+
+	/* if the number of xgmi link record is more than 128, driver will set the
+	 * flag 0 to get the first 128 of the link records and will set to 1, to get
+	 * the second set
+	 */
+	uint8_t				flag_extend_link_record;
+	/* bit0: port_num info support flag for GET_EXTEND_PEER_LINKS commmand */
+	uint8_t				caps_flag;
+	uint8_t				reserved[2];
+	union ta_xgmi_cmd_input		xgmi_in_message;
+	union ta_xgmi_cmd_output	xgmi_out_message;
+};
+
+#endif   //_TA_XGMI_IF_H
diff --git a/drivers/gpu/drm/amd/amdgpu/tonga_ih.c b/drivers/gpu/drm/amd/amdgpu/tonga_ih.c
index 923df2c0e535..7d17ae56f901 100644
--- a/drivers/gpu/drm/amd/amdgpu/tonga_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/tonga_ih.c
@@ -20,7 +20,9 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  *
  */
-#include <drm/drmP.h>
+
+#include <linux/pci.h>
+
 #include "amdgpu.h"
 #include "amdgpu_ih.h"
 #include "vid.h"
@@ -99,15 +101,15 @@ static void tonga_ih_disable_interrupts(struct amdgpu_device *adev)
  */
 static int tonga_ih_irq_init(struct amdgpu_device *adev)
 {
-	int rb_bufsz;
 	u32 interrupt_cntl, ih_rb_cntl, ih_doorbell_rtpr;
-	u64 wptr_off;
+	struct amdgpu_ih_ring *ih = &adev->irq.ih;
+	int rb_bufsz;
 
 	/* disable irqs */
 	tonga_ih_disable_interrupts(adev);
 
 	/* setup interrupt control */
-	WREG32(mmINTERRUPT_CNTL2, adev->dummy_page.addr >> 8);
+	WREG32(mmINTERRUPT_CNTL2, adev->dummy_page_addr >> 8);
 	interrupt_cntl = RREG32(mmINTERRUPT_CNTL);
 	/* INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
 	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
@@ -118,10 +120,7 @@ static int tonga_ih_irq_init(struct amdgpu_device *adev)
 	WREG32(mmINTERRUPT_CNTL, interrupt_cntl);
 
 	/* Ring Buffer base. [39:8] of 40-bit address of the beginning of the ring buffer*/
-	if (adev->irq.ih.use_bus_addr)
-		WREG32(mmIH_RB_BASE, adev->irq.ih.rb_dma_addr >> 8);
-	else
-		WREG32(mmIH_RB_BASE, adev->irq.ih.gpu_addr >> 8);
+	WREG32(mmIH_RB_BASE, ih->gpu_addr >> 8);
 
 	rb_bufsz = order_base_2(adev->irq.ih.ring_size / 4);
 	ih_rb_cntl = REG_SET_FIELD(0, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
@@ -136,12 +135,8 @@ static int tonga_ih_irq_init(struct amdgpu_device *adev)
 	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
 
 	/* set the writeback address whether it's enabled or not */
-	if (adev->irq.ih.use_bus_addr)
-		wptr_off = adev->irq.ih.rb_dma_addr + (adev->irq.ih.wptr_offs * 4);
-	else
-		wptr_off = adev->wb.gpu_addr + (adev->irq.ih.wptr_offs * 4);
-	WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(wptr_off));
-	WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(wptr_off) & 0xFF);
+	WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(ih->wptr_addr));
+	WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(ih->wptr_addr) & 0xFF);
 
 	/* set rptr, wptr to 0 */
 	WREG32(mmIH_RB_RPTR, 0);
@@ -186,6 +181,7 @@ static void tonga_ih_irq_disable(struct amdgpu_device *adev)
  * tonga_ih_get_wptr - get the IH ring buffer wptr
  *
  * @adev: amdgpu_device pointer
+ * @ih: IH ring buffer to fetch wptr
  *
  * Get the IH ring buffer wptr from either the register
  * or the writeback memory buffer (VI).  Also check for
@@ -193,86 +189,103 @@ static void tonga_ih_irq_disable(struct amdgpu_device *adev)
  * Used by cz_irq_process(VI).
  * Returns the value of the wptr.
  */
-static u32 tonga_ih_get_wptr(struct amdgpu_device *adev)
+static u32 tonga_ih_get_wptr(struct amdgpu_device *adev,
+			     struct amdgpu_ih_ring *ih)
 {
 	u32 wptr, tmp;
 
-	if (adev->irq.ih.use_bus_addr)
-		wptr = le32_to_cpu(adev->irq.ih.ring[adev->irq.ih.wptr_offs]);
-	else
-		wptr = le32_to_cpu(adev->wb.wb[adev->irq.ih.wptr_offs]);
-
-	if (REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW)) {
-		wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
-		/* When a ring buffer overflow happen start parsing interrupt
-		 * from the last not overwritten vector (wptr + 16). Hopefully
-		 * this should allow us to catchup.
-		 */
-		dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
-			wptr, adev->irq.ih.rptr, (wptr + 16) & adev->irq.ih.ptr_mask);
-		adev->irq.ih.rptr = (wptr + 16) & adev->irq.ih.ptr_mask;
-		tmp = RREG32(mmIH_RB_CNTL);
-		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
-		WREG32(mmIH_RB_CNTL, tmp);
-	}
-	return (wptr & adev->irq.ih.ptr_mask);
+	wptr = le32_to_cpu(*ih->wptr_cpu);
+
+	if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
+		goto out;
+
+	/* Double check that the overflow wasn't already cleared. */
+	wptr = RREG32(mmIH_RB_WPTR);
+
+	if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
+		goto out;
+
+	wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
+
+	/* When a ring buffer overflow happen start parsing interrupt
+	 * from the last not overwritten vector (wptr + 16). Hopefully
+	 * this should allow us to catchup.
+	 */
+
+	dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
+		wptr, ih->rptr, (wptr + 16) & ih->ptr_mask);
+	ih->rptr = (wptr + 16) & ih->ptr_mask;
+	tmp = RREG32(mmIH_RB_CNTL);
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+	WREG32(mmIH_RB_CNTL, tmp);
+
+	/* Unset the CLEAR_OVERFLOW bit immediately so new overflows
+	 * can be detected.
+	 */
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 0);
+	WREG32(mmIH_RB_CNTL, tmp);
+
+out:
+	return (wptr & ih->ptr_mask);
 }
 
 /**
  * tonga_ih_decode_iv - decode an interrupt vector
  *
  * @adev: amdgpu_device pointer
+ * @ih: IH ring buffer to decode
+ * @entry: IV entry to place decoded information into
  *
  * Decodes the interrupt vector at the current rptr
  * position and also advance the position.
  */
 static void tonga_ih_decode_iv(struct amdgpu_device *adev,
-				 struct amdgpu_iv_entry *entry)
+			       struct amdgpu_ih_ring *ih,
+			       struct amdgpu_iv_entry *entry)
 {
 	/* wptr/rptr are in bytes! */
-	u32 ring_index = adev->irq.ih.rptr >> 2;
+	u32 ring_index = ih->rptr >> 2;
 	uint32_t dw[4];
 
-	dw[0] = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
-	dw[1] = le32_to_cpu(adev->irq.ih.ring[ring_index + 1]);
-	dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
-	dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
+	dw[0] = le32_to_cpu(ih->ring[ring_index + 0]);
+	dw[1] = le32_to_cpu(ih->ring[ring_index + 1]);
+	dw[2] = le32_to_cpu(ih->ring[ring_index + 2]);
+	dw[3] = le32_to_cpu(ih->ring[ring_index + 3]);
 
-	entry->client_id = AMDGPU_IH_CLIENTID_LEGACY;
+	entry->client_id = AMDGPU_IRQ_CLIENTID_LEGACY;
 	entry->src_id = dw[0] & 0xff;
 	entry->src_data[0] = dw[1] & 0xfffffff;
 	entry->ring_id = dw[2] & 0xff;
-	entry->vm_id = (dw[2] >> 8) & 0xff;
-	entry->pas_id = (dw[2] >> 16) & 0xffff;
+	entry->vmid = (dw[2] >> 8) & 0xff;
+	entry->pasid = (dw[2] >> 16) & 0xffff;
 
 	/* wptr/rptr are in bytes! */
-	adev->irq.ih.rptr += 16;
+	ih->rptr += 16;
 }
 
 /**
  * tonga_ih_set_rptr - set the IH ring buffer rptr
  *
  * @adev: amdgpu_device pointer
+ * @ih: IH ring buffer to set rptr
  *
  * Set the IH ring buffer rptr.
  */
-static void tonga_ih_set_rptr(struct amdgpu_device *adev)
+static void tonga_ih_set_rptr(struct amdgpu_device *adev,
+			      struct amdgpu_ih_ring *ih)
 {
-	if (adev->irq.ih.use_doorbell) {
+	if (ih->use_doorbell) {
 		/* XXX check if swapping is necessary on BE */
-		if (adev->irq.ih.use_bus_addr)
-			adev->irq.ih.ring[adev->irq.ih.rptr_offs] = adev->irq.ih.rptr;
-		else
-			adev->wb.wb[adev->irq.ih.rptr_offs] = adev->irq.ih.rptr;
-		WDOORBELL32(adev->irq.ih.doorbell_index, adev->irq.ih.rptr);
+		*ih->rptr_cpu = ih->rptr;
+		WDOORBELL32(ih->doorbell_index, ih->rptr);
 	} else {
-		WREG32(mmIH_RB_RPTR, adev->irq.ih.rptr);
+		WREG32(mmIH_RB_RPTR, ih->rptr);
 	}
 }
 
-static int tonga_ih_early_init(void *handle)
+static int tonga_ih_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int ret;
 
 	ret = amdgpu_irq_add_domain(adev);
@@ -284,38 +297,37 @@ static int tonga_ih_early_init(void *handle)
 	return 0;
 }
 
-static int tonga_ih_sw_init(void *handle)
+static int tonga_ih_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	r = amdgpu_ih_ring_init(adev, 64 * 1024, true);
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih, 64 * 1024, true);
 	if (r)
 		return r;
 
 	adev->irq.ih.use_doorbell = true;
-	adev->irq.ih.doorbell_index = AMDGPU_DOORBELL_IH;
+	adev->irq.ih.doorbell_index = adev->doorbell_index.ih;
 
 	r = amdgpu_irq_init(adev);
 
 	return r;
 }
 
-static int tonga_ih_sw_fini(void *handle)
+static int tonga_ih_sw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	amdgpu_irq_fini(adev);
-	amdgpu_ih_ring_fini(adev);
+	amdgpu_irq_fini_sw(adev);
 	amdgpu_irq_remove_domain(adev);
 
 	return 0;
 }
 
-static int tonga_ih_hw_init(void *handle)
+static int tonga_ih_hw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	r = tonga_ih_irq_init(adev);
 	if (r)
@@ -324,32 +336,26 @@ static int tonga_ih_hw_init(void *handle)
 	return 0;
 }
 
-static int tonga_ih_hw_fini(void *handle)
+static int tonga_ih_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	tonga_ih_irq_disable(adev);
+	tonga_ih_irq_disable(ip_block->adev);
 
 	return 0;
 }
 
-static int tonga_ih_suspend(void *handle)
+static int tonga_ih_suspend(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return tonga_ih_hw_fini(adev);
+	return tonga_ih_hw_fini(ip_block);
 }
 
-static int tonga_ih_resume(void *handle)
+static int tonga_ih_resume(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return tonga_ih_hw_init(adev);
+	return tonga_ih_hw_init(ip_block);
 }
 
-static bool tonga_ih_is_idle(void *handle)
+static bool tonga_ih_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 tmp = RREG32(mmSRBM_STATUS);
 
 	if (REG_GET_FIELD(tmp, SRBM_STATUS, IH_BUSY))
@@ -358,11 +364,11 @@ static bool tonga_ih_is_idle(void *handle)
 	return true;
 }
 
-static int tonga_ih_wait_for_idle(void *handle)
+static int tonga_ih_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
 	unsigned i;
 	u32 tmp;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
 		/* read MC_STATUS */
@@ -374,9 +380,9 @@ static int tonga_ih_wait_for_idle(void *handle)
 	return -ETIMEDOUT;
 }
 
-static bool tonga_ih_check_soft_reset(void *handle)
+static bool tonga_ih_check_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 srbm_soft_reset = 0;
 	u32 tmp = RREG32(mmSRBM_STATUS);
 
@@ -393,29 +399,27 @@ static bool tonga_ih_check_soft_reset(void *handle)
 	}
 }
 
-static int tonga_ih_pre_soft_reset(void *handle)
+static int tonga_ih_pre_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (!adev->irq.srbm_soft_reset)
+	if (!ip_block->adev->irq.srbm_soft_reset)
 		return 0;
 
-	return tonga_ih_hw_fini(adev);
+	return tonga_ih_hw_fini(ip_block);
 }
 
-static int tonga_ih_post_soft_reset(void *handle)
+static int tonga_ih_post_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (!adev->irq.srbm_soft_reset)
 		return 0;
 
-	return tonga_ih_hw_init(adev);
+	return tonga_ih_hw_init(ip_block);
 }
 
-static int tonga_ih_soft_reset(void *handle)
+static int tonga_ih_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 srbm_soft_reset;
 
 	if (!adev->irq.srbm_soft_reset)
@@ -444,13 +448,13 @@ static int tonga_ih_soft_reset(void *handle)
 	return 0;
 }
 
-static int tonga_ih_set_clockgating_state(void *handle,
+static int tonga_ih_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
 	return 0;
 }
 
-static int tonga_ih_set_powergating_state(void *handle,
+static int tonga_ih_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	return 0;
@@ -459,7 +463,6 @@ static int tonga_ih_set_powergating_state(void *handle,
 static const struct amd_ip_funcs tonga_ih_ip_funcs = {
 	.name = "tonga_ih",
 	.early_init = tonga_ih_early_init,
-	.late_init = NULL,
 	.sw_init = tonga_ih_sw_init,
 	.sw_fini = tonga_ih_sw_fini,
 	.hw_init = tonga_ih_hw_init,
@@ -484,12 +487,10 @@ static const struct amdgpu_ih_funcs tonga_ih_funcs = {
 
 static void tonga_ih_set_interrupt_funcs(struct amdgpu_device *adev)
 {
-	if (adev->irq.ih_funcs == NULL)
-		adev->irq.ih_funcs = &tonga_ih_funcs;
+	adev->irq.ih_funcs = &tonga_ih_funcs;
 }
 
-const struct amdgpu_ip_block_version tonga_ih_ip_block =
-{
+const struct amdgpu_ip_block_version tonga_ih_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_IH,
 	.major = 3,
 	.minor = 0,
diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c
new file mode 100644
index 000000000000..8dc32787d625
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c
@@ -0,0 +1,728 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "umc_v12_0.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_umc.h"
+#include "amdgpu.h"
+#include "umc/umc_12_0_0_offset.h"
+#include "umc/umc_12_0_0_sh_mask.h"
+#include "mp/mp_13_0_6_sh_mask.h"
+
+#define MAX_ECC_NUM_PER_RETIREMENT  32
+#define DELAYED_TIME_FOR_GPU_RESET  1000  //ms
+
+static inline uint64_t get_umc_v12_0_reg_offset(struct amdgpu_device *adev,
+					    uint32_t node_inst,
+					    uint32_t umc_inst,
+					    uint32_t ch_inst)
+{
+	uint32_t index = umc_inst * adev->umc.channel_inst_num + ch_inst;
+	uint64_t cross_node_offset = (node_inst == 0) ? 0 : UMC_V12_0_CROSS_NODE_OFFSET;
+
+	umc_inst = index / 4;
+	ch_inst = index % 4;
+
+	return adev->umc.channel_offs * ch_inst + UMC_V12_0_INST_DIST * umc_inst +
+		UMC_V12_0_NODE_DIST * node_inst + cross_node_offset;
+}
+
+static int umc_v12_0_reset_error_count_per_channel(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
+{
+	uint64_t odecc_err_cnt_addr;
+	uint64_t umc_reg_offset =
+		get_umc_v12_0_reg_offset(adev, node_inst, umc_inst, ch_inst);
+
+	odecc_err_cnt_addr =
+		SOC15_REG_OFFSET(UMC, 0, regUMCCH0_OdEccErrCnt);
+
+	/* clear error count */
+	WREG32_PCIE_EXT((odecc_err_cnt_addr + umc_reg_offset) * 4,
+			UMC_V12_0_CE_CNT_INIT);
+
+	return 0;
+}
+
+static void umc_v12_0_reset_error_count(struct amdgpu_device *adev)
+{
+	amdgpu_umc_loop_channels(adev,
+		umc_v12_0_reset_error_count_per_channel, NULL);
+}
+
+bool umc_v12_0_is_deferred_error(struct amdgpu_device *adev, uint64_t mc_umc_status)
+{
+	dev_dbg(adev->dev,
+		"MCA_UMC_STATUS(0x%llx): Val:%llu, Poison:%llu, Deferred:%llu, PCC:%llu, UC:%llu, TCC:%llu\n",
+		mc_umc_status,
+		REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val),
+		REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Poison),
+		REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred),
+		REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, PCC),
+		REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UC),
+		REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, TCC)
+	);
+
+	return (amdgpu_ras_is_poison_mode_supported(adev) &&
+		(REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1) &&
+		((REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1) ||
+		(REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Poison) == 1)));
+}
+
+bool umc_v12_0_is_uncorrectable_error(struct amdgpu_device *adev, uint64_t mc_umc_status)
+{
+	if (umc_v12_0_is_deferred_error(adev, mc_umc_status))
+		return false;
+
+	return ((REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1) &&
+		(REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, PCC) == 1 ||
+		REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UC) == 1 ||
+		REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, TCC) == 1));
+}
+
+bool umc_v12_0_is_correctable_error(struct amdgpu_device *adev, uint64_t mc_umc_status)
+{
+	if (umc_v12_0_is_deferred_error(adev, mc_umc_status))
+		return false;
+
+	return (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
+		(REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, CECC) == 1 ||
+		(REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1 &&
+		REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UC) == 0) ||
+		/* Identify data parity error in replay mode */
+		((REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, ErrorCodeExt) == 0x5 ||
+		REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, ErrorCodeExt) == 0xb) &&
+		!(umc_v12_0_is_uncorrectable_error(adev, mc_umc_status)))));
+}
+
+static void umc_v12_0_query_error_count_per_type(struct amdgpu_device *adev,
+						   uint64_t umc_reg_offset,
+						   unsigned long *error_count,
+						   check_error_type_func error_type_func)
+{
+	uint64_t mc_umc_status;
+	uint64_t mc_umc_status_addr;
+
+	mc_umc_status_addr =
+		SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_STATUST0);
+
+	/* Check MCUMC_STATUS */
+	mc_umc_status =
+		RREG64_PCIE_EXT((mc_umc_status_addr + umc_reg_offset) * 4);
+
+	if (error_type_func(adev, mc_umc_status))
+		*error_count += 1;
+}
+
+static int umc_v12_0_query_error_count(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)data;
+	unsigned long ue_count = 0, ce_count = 0, de_count = 0;
+
+	/* NOTE: node_inst is converted by adev->umc.active_mask and the range is [0-3],
+	 * which can be used as die ID directly */
+	struct amdgpu_smuio_mcm_config_info mcm_info = {
+		.socket_id = adev->smuio.funcs->get_socket_id(adev),
+		.die_id = node_inst,
+	};
+
+	uint64_t umc_reg_offset =
+		get_umc_v12_0_reg_offset(adev, node_inst, umc_inst, ch_inst);
+
+	umc_v12_0_query_error_count_per_type(adev, umc_reg_offset,
+					    &ce_count, umc_v12_0_is_correctable_error);
+	umc_v12_0_query_error_count_per_type(adev, umc_reg_offset,
+					    &ue_count, umc_v12_0_is_uncorrectable_error);
+	umc_v12_0_query_error_count_per_type(adev, umc_reg_offset,
+					    &de_count, umc_v12_0_is_deferred_error);
+
+	amdgpu_ras_error_statistic_ue_count(err_data, &mcm_info, ue_count);
+	amdgpu_ras_error_statistic_ce_count(err_data, &mcm_info, ce_count);
+	amdgpu_ras_error_statistic_de_count(err_data, &mcm_info, de_count);
+
+	return 0;
+}
+
+static void umc_v12_0_query_ras_error_count(struct amdgpu_device *adev,
+					   void *ras_error_status)
+{
+	amdgpu_umc_loop_channels(adev,
+		umc_v12_0_query_error_count, ras_error_status);
+
+	umc_v12_0_reset_error_count(adev);
+}
+
+static void umc_v12_0_get_retire_flip_bits(struct amdgpu_device *adev)
+{
+	enum amdgpu_memory_partition nps = AMDGPU_NPS1_PARTITION_MODE;
+	uint32_t vram_type = adev->gmc.vram_type;
+	struct amdgpu_umc_flip_bits *flip_bits = &(adev->umc.flip_bits);
+
+	if (adev->gmc.gmc_funcs->query_mem_partition_mode)
+		nps = adev->gmc.gmc_funcs->query_mem_partition_mode(adev);
+
+	/* default setting */
+	flip_bits->flip_bits_in_pa[0] = UMC_V12_0_PA_C2_BIT;
+	flip_bits->flip_bits_in_pa[1] = UMC_V12_0_PA_C3_BIT;
+	flip_bits->flip_bits_in_pa[2] = UMC_V12_0_PA_C4_BIT;
+	flip_bits->flip_bits_in_pa[3] = UMC_V12_0_PA_R13_BIT;
+	flip_bits->flip_row_bit = 13;
+	flip_bits->bit_num = 4;
+	flip_bits->r13_in_pa = UMC_V12_0_PA_R13_BIT;
+
+	if (nps == AMDGPU_NPS2_PARTITION_MODE) {
+		flip_bits->flip_bits_in_pa[0] = UMC_V12_0_PA_CH5_BIT;
+		flip_bits->flip_bits_in_pa[1] = UMC_V12_0_PA_C2_BIT;
+		flip_bits->flip_bits_in_pa[2] = UMC_V12_0_PA_B1_BIT;
+		flip_bits->r13_in_pa = UMC_V12_0_PA_R12_BIT;
+	} else if (nps == AMDGPU_NPS4_PARTITION_MODE) {
+		flip_bits->flip_bits_in_pa[0] = UMC_V12_0_PA_CH4_BIT;
+		flip_bits->flip_bits_in_pa[1] = UMC_V12_0_PA_CH5_BIT;
+		flip_bits->flip_bits_in_pa[2] = UMC_V12_0_PA_B0_BIT;
+		flip_bits->r13_in_pa = UMC_V12_0_PA_R11_BIT;
+	}
+
+	switch (vram_type) {
+	case AMDGPU_VRAM_TYPE_HBM:
+		/* other nps modes are taken as nps1 */
+		if (nps == AMDGPU_NPS2_PARTITION_MODE)
+			flip_bits->flip_bits_in_pa[3] = UMC_V12_0_PA_R12_BIT;
+		else if (nps == AMDGPU_NPS4_PARTITION_MODE)
+			flip_bits->flip_bits_in_pa[3] = UMC_V12_0_PA_R11_BIT;
+
+		break;
+	case AMDGPU_VRAM_TYPE_HBM3E:
+		flip_bits->flip_bits_in_pa[3] = UMC_V12_0_PA_R12_BIT;
+		flip_bits->flip_row_bit = 12;
+
+		if (nps == AMDGPU_NPS2_PARTITION_MODE)
+			flip_bits->flip_bits_in_pa[3] = UMC_V12_0_PA_R11_BIT;
+		else if (nps == AMDGPU_NPS4_PARTITION_MODE)
+			flip_bits->flip_bits_in_pa[3] = UMC_V12_0_PA_R10_BIT;
+
+		break;
+	default:
+		dev_warn(adev->dev,
+			"Unknown HBM type, set RAS retire flip bits to the value in NPS1 mode.\n");
+		break;
+	}
+
+	adev->umc.retire_unit = 0x1 << flip_bits->bit_num;
+}
+
+static int umc_v12_0_convert_error_address(struct amdgpu_device *adev,
+					struct ras_err_data *err_data,
+					struct ta_ras_query_address_input *addr_in,
+					struct ta_ras_query_address_output *addr_out,
+					bool dump_addr)
+{
+	uint32_t col, col_lower, row, row_lower, row_high, bank;
+	uint32_t channel_index = 0, umc_inst = 0;
+	uint32_t i, bit_num, retire_unit, *flip_bits;
+	uint64_t soc_pa, column, err_addr;
+	struct ta_ras_query_address_output addr_out_tmp;
+	struct ta_ras_query_address_output *paddr_out;
+	int ret = 0;
+
+	if (!addr_out)
+		paddr_out = &addr_out_tmp;
+	else
+		paddr_out = addr_out;
+
+	err_addr = bank = 0;
+	if (addr_in) {
+		err_addr = addr_in->ma.err_addr;
+		addr_in->addr_type = TA_RAS_MCA_TO_PA;
+		ret = psp_ras_query_address(&adev->psp, addr_in, paddr_out);
+		if (ret) {
+			dev_warn(adev->dev, "Failed to query RAS physical address for 0x%llx",
+				err_addr);
+
+			goto out;
+		}
+
+		bank = paddr_out->pa.bank;
+		/* no need to care about umc inst if addr_in is NULL */
+		umc_inst = addr_in->ma.umc_inst;
+	}
+
+	flip_bits = adev->umc.flip_bits.flip_bits_in_pa;
+	bit_num = adev->umc.flip_bits.bit_num;
+	retire_unit = adev->umc.retire_unit;
+
+	soc_pa = paddr_out->pa.pa;
+	channel_index = paddr_out->pa.channel_idx;
+	/* clear loop bits in soc physical address */
+	for (i = 0; i < bit_num; i++)
+		soc_pa &= ~BIT_ULL(flip_bits[i]);
+
+	paddr_out->pa.pa = soc_pa;
+	/* get column bit 0 and 1 in mca address */
+	col_lower = (err_addr >> 1) & 0x3ULL;
+	/* extra row bit will be handled later */
+	row_lower = (err_addr >> UMC_V12_0_MA_R0_BIT) & 0x1fffULL;
+	row_lower &= ~BIT_ULL(adev->umc.flip_bits.flip_row_bit);
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(9, 5, 0)) {
+		row_high = (soc_pa >> adev->umc.flip_bits.r13_in_pa) & 0x3ULL;
+		/* it's 2.25GB in each channel, from MCA address to PA
+		 * [R14 R13] is converted if the two bits value are 0x3,
+		 * get them from PA instead of MCA address.
+		 */
+		row_lower |= (row_high << 13);
+	}
+
+	if (!err_data && !dump_addr)
+		goto out;
+
+	/* loop for all possibilities of retired bits */
+	for (column = 0; column < retire_unit; column++) {
+		soc_pa = paddr_out->pa.pa;
+		for (i = 0; i < bit_num; i++)
+			soc_pa |= (((column >> i) & 0x1ULL) << flip_bits[i]);
+
+		col = ((column & 0x7) << 2) | col_lower;
+		/* handle extra row bit */
+		if (bit_num == RETIRE_FLIP_BITS_NUM)
+			row = ((column >> 3) << adev->umc.flip_bits.flip_row_bit) |
+					row_lower;
+
+		if (dump_addr)
+			dev_info(adev->dev,
+				"Error Address(PA):0x%-10llx Row:0x%-4x Col:0x%-2x Bank:0x%x Channel:0x%x\n",
+				soc_pa, row, col, bank, channel_index);
+
+		if (err_data)
+			amdgpu_umc_fill_error_record(err_data, err_addr,
+				soc_pa, channel_index, umc_inst);
+	}
+
+out:
+	return ret;
+}
+
+static int umc_v12_0_query_error_address(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)data;
+	struct ta_ras_query_address_input addr_in;
+	uint64_t mc_umc_status_addr;
+	uint64_t mc_umc_status, err_addr;
+	uint64_t mc_umc_addrt0;
+	uint64_t umc_reg_offset =
+		get_umc_v12_0_reg_offset(adev, node_inst, umc_inst, ch_inst);
+
+	mc_umc_status_addr =
+		SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_STATUST0);
+
+	mc_umc_status = RREG64_PCIE_EXT((mc_umc_status_addr + umc_reg_offset) * 4);
+
+	if (mc_umc_status == 0)
+		return 0;
+
+	if (!err_data->err_addr) {
+		/* clear umc status */
+		WREG64_PCIE_EXT((mc_umc_status_addr + umc_reg_offset) * 4, 0x0ULL);
+
+		return 0;
+	}
+
+	/* calculate error address if ue error is detected */
+	if (umc_v12_0_is_uncorrectable_error(adev, mc_umc_status) ||
+	    umc_v12_0_is_deferred_error(adev, mc_umc_status)) {
+		mc_umc_addrt0 =
+			SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_ADDRT0);
+
+		err_addr = RREG64_PCIE_EXT((mc_umc_addrt0 + umc_reg_offset) * 4);
+
+		err_addr = REG_GET_FIELD(err_addr, MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr);
+
+		if (!adev->aid_mask &&
+		    adev->smuio.funcs &&
+		    adev->smuio.funcs->get_socket_id)
+			addr_in.ma.socket_id = adev->smuio.funcs->get_socket_id(adev);
+		else
+			addr_in.ma.socket_id = 0;
+
+		addr_in.ma.err_addr = err_addr;
+		addr_in.ma.ch_inst = ch_inst;
+		addr_in.ma.umc_inst = umc_inst;
+		addr_in.ma.node_inst = node_inst;
+
+		umc_v12_0_convert_error_address(adev, err_data, &addr_in, NULL, true);
+	}
+
+	/* clear umc status */
+	WREG64_PCIE_EXT((mc_umc_status_addr + umc_reg_offset) * 4, 0x0ULL);
+
+	return 0;
+}
+
+static void umc_v12_0_query_ras_error_address(struct amdgpu_device *adev,
+					     void *ras_error_status)
+{
+	amdgpu_umc_loop_channels(adev,
+		umc_v12_0_query_error_address, ras_error_status);
+}
+
+static int umc_v12_0_err_cnt_init_per_channel(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
+{
+	uint32_t odecc_cnt_sel;
+	uint64_t odecc_cnt_sel_addr, odecc_err_cnt_addr;
+	uint64_t umc_reg_offset =
+		get_umc_v12_0_reg_offset(adev, node_inst, umc_inst, ch_inst);
+
+	odecc_cnt_sel_addr =
+		SOC15_REG_OFFSET(UMC, 0, regUMCCH0_OdEccCntSel);
+	odecc_err_cnt_addr =
+		SOC15_REG_OFFSET(UMC, 0, regUMCCH0_OdEccErrCnt);
+
+	odecc_cnt_sel = RREG32_PCIE_EXT((odecc_cnt_sel_addr + umc_reg_offset) * 4);
+
+	/* set ce error interrupt type to APIC based interrupt */
+	odecc_cnt_sel = REG_SET_FIELD(odecc_cnt_sel, UMCCH0_OdEccCntSel,
+					OdEccErrInt, 0x1);
+	WREG32_PCIE_EXT((odecc_cnt_sel_addr + umc_reg_offset) * 4, odecc_cnt_sel);
+
+	/* set error count to initial value */
+	WREG32_PCIE_EXT((odecc_err_cnt_addr + umc_reg_offset) * 4, UMC_V12_0_CE_CNT_INIT);
+
+	return 0;
+}
+
+static bool umc_v12_0_check_ecc_err_status(struct amdgpu_device *adev,
+			enum amdgpu_mca_error_type type, void *ras_error_status)
+{
+	uint64_t mc_umc_status = *(uint64_t *)ras_error_status;
+
+	switch (type) {
+	case AMDGPU_MCA_ERROR_TYPE_UE:
+		return umc_v12_0_is_uncorrectable_error(adev, mc_umc_status);
+	case AMDGPU_MCA_ERROR_TYPE_CE:
+		return umc_v12_0_is_correctable_error(adev, mc_umc_status);
+	case AMDGPU_MCA_ERROR_TYPE_DE:
+		return umc_v12_0_is_deferred_error(adev, mc_umc_status);
+	default:
+		return false;
+	}
+
+	return false;
+}
+
+static void umc_v12_0_err_cnt_init(struct amdgpu_device *adev)
+{
+	amdgpu_umc_loop_channels(adev,
+		umc_v12_0_err_cnt_init_per_channel, NULL);
+}
+
+static bool umc_v12_0_query_ras_poison_mode(struct amdgpu_device *adev)
+{
+	/*
+	 * Force return true, because regUMCCH0_EccCtrl
+	 * is not accessible from host side
+	 */
+	return true;
+}
+
+const struct amdgpu_ras_block_hw_ops umc_v12_0_ras_hw_ops = {
+	.query_ras_error_count = umc_v12_0_query_ras_error_count,
+	.query_ras_error_address = umc_v12_0_query_ras_error_address,
+};
+
+static int umc_v12_0_aca_bank_parser(struct aca_handle *handle, struct aca_bank *bank,
+				     enum aca_smu_type type, void *data)
+{
+	struct amdgpu_device *adev = handle->adev;
+	struct aca_bank_info info;
+	enum aca_error_type err_type;
+	u64 status, count;
+	u32 ext_error_code;
+	int ret;
+
+	status = bank->regs[ACA_REG_IDX_STATUS];
+	if (umc_v12_0_is_deferred_error(adev, status))
+		err_type = ACA_ERROR_TYPE_DEFERRED;
+	else if (umc_v12_0_is_uncorrectable_error(adev, status))
+		err_type = ACA_ERROR_TYPE_UE;
+	else if (umc_v12_0_is_correctable_error(adev, status))
+		err_type = ACA_ERROR_TYPE_CE;
+	else
+		return 0;
+	bank->aca_err_type = err_type;
+
+	ret = aca_bank_info_decode(bank, &info);
+	if (ret)
+		return ret;
+
+	amdgpu_umc_update_ecc_status(adev,
+		bank->regs[ACA_REG_IDX_STATUS],
+		bank->regs[ACA_REG_IDX_IPID],
+		bank->regs[ACA_REG_IDX_ADDR]);
+
+	ext_error_code = ACA_REG__STATUS__ERRORCODEEXT(status);
+	if (umc_v12_0_is_deferred_error(adev, status))
+		count = ext_error_code == 0 ?
+			adev->umc.err_addr_cnt / adev->umc.retire_unit : 1ULL;
+	else
+		count = ext_error_code == 0 ?
+			ACA_REG__MISC0__ERRCNT(bank->regs[ACA_REG_IDX_MISC0]) : 1ULL;
+
+	return aca_error_cache_log_bank_error(handle, &info, err_type, count);
+}
+
+static const struct aca_bank_ops umc_v12_0_aca_bank_ops = {
+	.aca_bank_parser = umc_v12_0_aca_bank_parser,
+};
+
+const struct aca_info umc_v12_0_aca_info = {
+	.hwip = ACA_HWIP_TYPE_UMC,
+	.mask = ACA_ERROR_UE_MASK | ACA_ERROR_CE_MASK | ACA_ERROR_DEFERRED_MASK,
+	.bank_ops = &umc_v12_0_aca_bank_ops,
+};
+
+static int umc_v12_0_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
+{
+	int ret;
+
+	ret = amdgpu_umc_ras_late_init(adev, ras_block);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__UMC,
+				  &umc_v12_0_aca_info, NULL);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static int umc_v12_0_update_ecc_status(struct amdgpu_device *adev,
+			uint64_t status, uint64_t ipid, uint64_t addr)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	uint16_t hwid, mcatype;
+	uint64_t page_pfn[UMC_V12_0_BAD_PAGE_NUM_PER_CHANNEL];
+	uint64_t err_addr, pa_addr = 0;
+	struct ras_ecc_err *ecc_err;
+	struct ta_ras_query_address_output addr_out;
+	uint32_t shift_bit = adev->umc.flip_bits.flip_bits_in_pa[2];
+	int count, ret, i;
+
+	hwid = REG_GET_FIELD(ipid, MCMP1_IPIDT0, HardwareID);
+	mcatype = REG_GET_FIELD(ipid, MCMP1_IPIDT0, McaType);
+
+	/* The IP block decode of consumption is SMU */
+	if (hwid != MCA_UMC_HWID_V12_0 || mcatype != MCA_UMC_MCATYPE_V12_0) {
+		con->umc_ecc_log.consumption_q_count++;
+		return 0;
+	}
+
+	if (!status)
+		return 0;
+
+	if (!umc_v12_0_is_deferred_error(adev, status))
+		return 0;
+
+	err_addr = REG_GET_FIELD(addr,
+				MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr);
+
+	dev_dbg(adev->dev,
+		"UMC:IPID:0x%llx, socket:%llu, aid:%llu, inst:%llu, ch:%llu, err_addr:0x%llx\n",
+		ipid,
+		MCA_IPID_2_SOCKET_ID(ipid),
+		MCA_IPID_2_DIE_ID(ipid),
+		MCA_IPID_2_UMC_INST(ipid),
+		MCA_IPID_2_UMC_CH(ipid),
+		err_addr);
+
+	ret = amdgpu_umc_mca_to_addr(adev,
+			err_addr, MCA_IPID_2_UMC_CH(ipid),
+			MCA_IPID_2_UMC_INST(ipid), MCA_IPID_2_DIE_ID(ipid),
+			MCA_IPID_2_SOCKET_ID(ipid), &addr_out, true);
+	if (ret)
+		return ret;
+
+	ecc_err = kzalloc(sizeof(*ecc_err), GFP_KERNEL);
+	if (!ecc_err)
+		return -ENOMEM;
+
+	pa_addr = addr_out.pa.pa;
+	ecc_err->status = status;
+	ecc_err->ipid = ipid;
+	ecc_err->addr = addr;
+	ecc_err->pa_pfn = pa_addr >> AMDGPU_GPU_PAGE_SHIFT;
+	ecc_err->channel_idx = addr_out.pa.channel_idx;
+
+	/* If converted pa_pfn is 0, use pa C4 pfn. */
+	if (!ecc_err->pa_pfn)
+		ecc_err->pa_pfn = BIT_ULL(shift_bit) >> AMDGPU_GPU_PAGE_SHIFT;
+
+	ret = amdgpu_umc_logs_ecc_err(adev, &con->umc_ecc_log.de_page_tree, ecc_err);
+	if (ret) {
+		if (ret == -EEXIST)
+			con->umc_ecc_log.de_queried_count++;
+		else
+			dev_err(adev->dev, "Fail to log ecc error! ret:%d\n", ret);
+
+		kfree(ecc_err);
+		return ret;
+	}
+
+	con->umc_ecc_log.de_queried_count++;
+
+	memset(page_pfn, 0, sizeof(page_pfn));
+	count = amdgpu_umc_lookup_bad_pages_in_a_row(adev,
+				pa_addr,
+				page_pfn, ARRAY_SIZE(page_pfn));
+	if (count <= 0) {
+		dev_warn(adev->dev, "Fail to convert error address! count:%d\n", count);
+		return 0;
+	}
+
+	/* Reserve memory */
+	for (i = 0; i < count; i++)
+		amdgpu_ras_reserve_page(adev, page_pfn[i]);
+
+	/* The problem case is as follows:
+	 * 1. GPU A triggers a gpu ras reset, and GPU A drives
+	 *    GPU B to also perform a gpu ras reset.
+	 * 2. After gpu B ras reset started, gpu B queried a DE
+	 *    data. Since the DE data was queried in the ras reset
+	 *    thread instead of the page retirement thread, bad
+	 *    page retirement work would not be triggered. Then
+	 *    even if all gpu resets are completed, the bad pages
+	 *    will be cached in RAM until GPU B's bad page retirement
+	 *    work is triggered again and then saved to eeprom.
+	 * Trigger delayed work to save the bad pages to eeprom in time
+	 * after gpu ras reset is completed.
+	 */
+	if (amdgpu_ras_in_recovery(adev))
+		schedule_delayed_work(&con->page_retirement_dwork,
+			msecs_to_jiffies(DELAYED_TIME_FOR_GPU_RESET));
+
+	return 0;
+}
+
+static int umc_v12_0_fill_error_record(struct amdgpu_device *adev,
+				struct ras_ecc_err *ecc_err, void *ras_error_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+	uint64_t page_pfn[UMC_V12_0_BAD_PAGE_NUM_PER_CHANNEL];
+	int ret, i, count;
+
+	if (!err_data || !ecc_err)
+		return -EINVAL;
+
+	memset(page_pfn, 0, sizeof(page_pfn));
+	count = amdgpu_umc_lookup_bad_pages_in_a_row(adev,
+				ecc_err->pa_pfn << AMDGPU_GPU_PAGE_SHIFT,
+				page_pfn, ARRAY_SIZE(page_pfn));
+
+	for (i = 0; i < count; i++) {
+		ret = amdgpu_umc_fill_error_record(err_data,
+				ecc_err->addr,
+				page_pfn[i] << AMDGPU_GPU_PAGE_SHIFT,
+				ecc_err->channel_idx,
+				MCA_IPID_2_UMC_INST(ecc_err->ipid));
+		if (ret)
+			break;
+	}
+
+	err_data->de_count++;
+
+	return ret;
+}
+
+static void umc_v12_0_query_ras_ecc_err_addr(struct amdgpu_device *adev,
+					void *ras_error_status)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_ecc_err *entries[MAX_ECC_NUM_PER_RETIREMENT];
+	struct radix_tree_root *ecc_tree;
+	int new_detected, ret, i;
+
+	ecc_tree = &con->umc_ecc_log.de_page_tree;
+
+	mutex_lock(&con->umc_ecc_log.lock);
+	new_detected = radix_tree_gang_lookup_tag(ecc_tree, (void **)entries,
+			0, ARRAY_SIZE(entries), UMC_ECC_NEW_DETECTED_TAG);
+	for (i = 0; i < new_detected; i++) {
+		if (!entries[i])
+			continue;
+
+		ret = umc_v12_0_fill_error_record(adev, entries[i], ras_error_status);
+		if (ret) {
+			dev_err(adev->dev, "Fail to fill umc error record, ret:%d\n", ret);
+			break;
+		}
+		radix_tree_tag_clear(ecc_tree,
+				entries[i]->pa_pfn, UMC_ECC_NEW_DETECTED_TAG);
+	}
+	mutex_unlock(&con->umc_ecc_log.lock);
+}
+
+static uint32_t umc_v12_0_get_die_id(struct amdgpu_device *adev,
+		uint64_t mca_addr, uint64_t retired_page)
+{
+	uint32_t die = 0;
+
+	/* we only calculate die id for nps1 mode right now */
+	die += ((((retired_page >> 12) & 0x1ULL)^
+	    ((retired_page >> 20) & 0x1ULL) ^
+	    ((retired_page >> 27) & 0x1ULL) ^
+	    ((retired_page >> 34) & 0x1ULL) ^
+	    ((retired_page >> 41) & 0x1ULL)) << 0);
+
+	/* the original PA_C4 and PA_R13 may be cleared in retired_page, so
+	 * get them from mca_addr.
+	 */
+	die += ((((retired_page >> 13) & 0x1ULL) ^
+	    ((mca_addr >> 5) & 0x1ULL) ^
+	    ((retired_page >> 28) & 0x1ULL) ^
+	    ((mca_addr >> 23) & 0x1ULL) ^
+	    ((retired_page >> 42) & 0x1ULL)) << 1);
+	die &= 3;
+
+	return die;
+}
+
+struct amdgpu_umc_ras umc_v12_0_ras = {
+	.ras_block = {
+		.hw_ops = &umc_v12_0_ras_hw_ops,
+		.ras_late_init = umc_v12_0_ras_late_init,
+	},
+	.err_cnt_init = umc_v12_0_err_cnt_init,
+	.query_ras_poison_mode = umc_v12_0_query_ras_poison_mode,
+	.ecc_info_query_ras_error_address = umc_v12_0_query_ras_ecc_err_addr,
+	.check_ecc_err_status = umc_v12_0_check_ecc_err_status,
+	.update_ecc_status = umc_v12_0_update_ecc_status,
+	.convert_ras_err_addr = umc_v12_0_convert_error_address,
+	.get_die_id_from_pa = umc_v12_0_get_die_id,
+	.get_retire_flip_bits = umc_v12_0_get_retire_flip_bits,
+};
+
diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.h b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.h
new file mode 100644
index 000000000000..63b7e7254526
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.h
@@ -0,0 +1,105 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __UMC_V12_0_H__
+#define __UMC_V12_0_H__
+
+#include "soc15_common.h"
+#include "amdgpu.h"
+
+#define UMC_V12_0_NODE_DIST		0x40000000
+#define UMC_V12_0_INST_DIST		0x40000
+
+/* UMC register per channel offset */
+#define UMC_V12_0_PER_CHANNEL_OFFSET	0x400
+
+/* UMC cross node offset */
+#define UMC_V12_0_CROSS_NODE_OFFSET		0x100000000
+
+/* OdEccErrCnt max value */
+#define UMC_V12_0_CE_CNT_MAX		0xffff
+/* umc ce interrupt threshold */
+#define UMC_V12_0_CE_INT_THRESHOLD	0xffff
+/* umc ce count initial value */
+#define UMC_V12_0_CE_CNT_INIT	(UMC_V12_0_CE_CNT_MAX - UMC_V12_0_CE_INT_THRESHOLD)
+
+/* number of umc channel instance with memory map register access */
+#define UMC_V12_0_CHANNEL_INSTANCE_NUM		8
+/* number of umc instance with memory map register access */
+#define UMC_V12_0_UMC_INSTANCE_NUM		4
+
+/* Total channel instances for all available umc nodes */
+#define UMC_V12_0_TOTAL_CHANNEL_NUM(adev) \
+	(UMC_V12_0_CHANNEL_INSTANCE_NUM * (adev)->gmc.num_umc)
+
+/* one piece of normalized address is mapped to 8 pieces of physical address */
+#define UMC_V12_0_NA_MAP_PA_NUM        8
+/* R13 bit shift should be considered, double the number */
+#define UMC_V12_0_BAD_PAGE_NUM_PER_CHANNEL (UMC_V12_0_NA_MAP_PA_NUM * 2)
+
+/* column bits in SOC physical address */
+#define UMC_V12_0_PA_C2_BIT 15
+#define UMC_V12_0_PA_C3_BIT 16
+#define UMC_V12_0_PA_C4_BIT 21
+/* row bits in SOC physical address */
+#define UMC_V12_0_PA_R0_BIT 22
+#define UMC_V12_0_PA_R10_BIT 32
+#define UMC_V12_0_PA_R11_BIT 33
+#define UMC_V12_0_PA_R12_BIT 34
+#define UMC_V12_0_PA_R13_BIT 35
+/* channel bit in SOC physical address */
+#define UMC_V12_0_PA_CH4_BIT 12
+#define UMC_V12_0_PA_CH5_BIT 13
+/* bank bit in SOC physical address */
+#define UMC_V12_0_PA_B0_BIT 19
+#define UMC_V12_0_PA_B1_BIT 20
+/* row bits in MCA address */
+#define UMC_V12_0_MA_R0_BIT 10
+
+#define MCA_UMC_HWID_V12_0     0x96
+#define MCA_UMC_MCATYPE_V12_0  0x0
+
+#define MCA_IPID_LO_2_UMC_CH(_ipid_lo) (((((_ipid_lo) >> 20) & 0x1) * 4) + \
+			(((_ipid_lo) >> 12) & 0xF))
+#define MCA_IPID_LO_2_UMC_INST(_ipid_lo) (((_ipid_lo) >> 21) & 0x7)
+
+#define MCA_IPID_2_DIE_ID(ipid)  ((REG_GET_FIELD(ipid, MCMP1_IPIDT0, InstanceIdHi) >> 2) & 0x03)
+
+#define MCA_IPID_2_UMC_CH(ipid) \
+	(MCA_IPID_LO_2_UMC_CH(REG_GET_FIELD(ipid, MCMP1_IPIDT0, InstanceIdLo)))
+
+#define MCA_IPID_2_UMC_INST(ipid) \
+	(MCA_IPID_LO_2_UMC_INST(REG_GET_FIELD(ipid, MCMP1_IPIDT0, InstanceIdLo)))
+
+#define MCA_IPID_2_SOCKET_ID(ipid) \
+	(((REG_GET_FIELD(ipid, MCMP1_IPIDT0, InstanceIdLo) & 0x1) << 2) | \
+	 (REG_GET_FIELD(ipid, MCMP1_IPIDT0, InstanceIdHi) & 0x03))
+
+bool umc_v12_0_is_deferred_error(struct amdgpu_device *adev, uint64_t mc_umc_status);
+bool umc_v12_0_is_uncorrectable_error(struct amdgpu_device *adev, uint64_t mc_umc_status);
+bool umc_v12_0_is_correctable_error(struct amdgpu_device *adev, uint64_t mc_umc_status);
+
+typedef bool (*check_error_type_func)(struct amdgpu_device *adev, uint64_t mc_umc_status);
+
+extern struct amdgpu_umc_ras umc_v12_0_ras;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_0.c b/drivers/gpu/drm/amd/amdgpu/umc_v6_0.c
new file mode 100644
index 000000000000..97fa88ed770c
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_0.c
@@ -0,0 +1,37 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "umc_v6_0.h"
+#include "amdgpu.h"
+
+static void umc_v6_0_init_registers(struct amdgpu_device *adev)
+{
+	unsigned i, j;
+
+	for (i = 0; i < 4; i++)
+		for (j = 0; j < 4; j++)
+			WREG32((i*0x100000 + 0x5010c + j*0x2000)/4, 0x1002);
+}
+
+const struct amdgpu_umc_funcs umc_v6_0_funcs = {
+	.init_registers = umc_v6_0_init_registers,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_0.h b/drivers/gpu/drm/amd/amdgpu/umc_v6_0.h
new file mode 100644
index 000000000000..109f1a57a46e
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_0.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __UMC_V6_0_H__
+#define __UMC_V6_0_H__
+
+#include "soc15_common.h"
+#include "amdgpu.h"
+
+extern const struct amdgpu_umc_funcs umc_v6_0_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c b/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c
new file mode 100644
index 000000000000..f17d297b594b
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c
@@ -0,0 +1,462 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "umc_v6_1.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_umc.h"
+#include "amdgpu.h"
+
+#include "rsmu/rsmu_0_0_2_offset.h"
+#include "rsmu/rsmu_0_0_2_sh_mask.h"
+#include "umc/umc_6_1_1_offset.h"
+#include "umc/umc_6_1_1_sh_mask.h"
+#include "umc/umc_6_1_2_offset.h"
+
+#define UMC_6_INST_DIST			0x40000
+
+const uint32_t
+	umc_v6_1_channel_idx_tbl[UMC_V6_1_UMC_INSTANCE_NUM][UMC_V6_1_CHANNEL_INSTANCE_NUM] = {
+		{2, 18, 11, 27},	{4, 20, 13, 29},
+		{1, 17, 8, 24},		{7, 23, 14, 30},
+		{10, 26, 3, 19},	{12, 28, 5, 21},
+		{9, 25, 0, 16},		{15, 31, 6, 22}
+};
+
+static void umc_v6_1_enable_umc_index_mode(struct amdgpu_device *adev)
+{
+	uint32_t rsmu_umc_addr, rsmu_umc_val;
+
+	rsmu_umc_addr = SOC15_REG_OFFSET(RSMU, 0,
+			mmRSMU_UMC_INDEX_REGISTER_NBIF_VG20_GPU);
+	rsmu_umc_val = RREG32_PCIE(rsmu_umc_addr * 4);
+
+	rsmu_umc_val = REG_SET_FIELD(rsmu_umc_val,
+			RSMU_UMC_INDEX_REGISTER_NBIF_VG20_GPU,
+			RSMU_UMC_INDEX_MODE_EN, 1);
+
+	WREG32_PCIE(rsmu_umc_addr * 4, rsmu_umc_val);
+}
+
+static void umc_v6_1_disable_umc_index_mode(struct amdgpu_device *adev)
+{
+	uint32_t rsmu_umc_addr, rsmu_umc_val;
+
+	rsmu_umc_addr = SOC15_REG_OFFSET(RSMU, 0,
+			mmRSMU_UMC_INDEX_REGISTER_NBIF_VG20_GPU);
+	rsmu_umc_val = RREG32_PCIE(rsmu_umc_addr * 4);
+
+	rsmu_umc_val = REG_SET_FIELD(rsmu_umc_val,
+			RSMU_UMC_INDEX_REGISTER_NBIF_VG20_GPU,
+			RSMU_UMC_INDEX_MODE_EN, 0);
+
+	WREG32_PCIE(rsmu_umc_addr * 4, rsmu_umc_val);
+}
+
+static uint32_t umc_v6_1_get_umc_index_mode_state(struct amdgpu_device *adev)
+{
+	uint32_t rsmu_umc_addr, rsmu_umc_val;
+
+	rsmu_umc_addr = SOC15_REG_OFFSET(RSMU, 0,
+			mmRSMU_UMC_INDEX_REGISTER_NBIF_VG20_GPU);
+	rsmu_umc_val = RREG32_PCIE(rsmu_umc_addr * 4);
+
+	return REG_GET_FIELD(rsmu_umc_val,
+			RSMU_UMC_INDEX_REGISTER_NBIF_VG20_GPU,
+			RSMU_UMC_INDEX_MODE_EN);
+}
+
+static inline uint32_t get_umc_6_reg_offset(struct amdgpu_device *adev,
+					    uint32_t umc_inst,
+					    uint32_t ch_inst)
+{
+	return adev->umc.channel_offs*ch_inst + UMC_6_INST_DIST*umc_inst;
+}
+
+static void umc_v6_1_clear_error_count_per_channel(struct amdgpu_device *adev,
+					uint32_t umc_reg_offset)
+{
+	uint32_t ecc_err_cnt_addr;
+	uint32_t ecc_err_cnt_sel, ecc_err_cnt_sel_addr;
+
+	if (adev->asic_type == CHIP_ARCTURUS) {
+		/* UMC 6_1_2 registers */
+		ecc_err_cnt_sel_addr =
+			SOC15_REG_OFFSET(UMC, 0,
+					mmUMCCH0_0_EccErrCntSel_ARCT);
+		ecc_err_cnt_addr =
+			SOC15_REG_OFFSET(UMC, 0,
+					mmUMCCH0_0_EccErrCnt_ARCT);
+	} else {
+		/* UMC 6_1_1 registers */
+		ecc_err_cnt_sel_addr =
+			SOC15_REG_OFFSET(UMC, 0,
+					mmUMCCH0_0_EccErrCntSel);
+		ecc_err_cnt_addr =
+			SOC15_REG_OFFSET(UMC, 0,
+					mmUMCCH0_0_EccErrCnt);
+	}
+
+	/* select the lower chip */
+	ecc_err_cnt_sel = RREG32_PCIE((ecc_err_cnt_sel_addr +
+					umc_reg_offset) * 4);
+	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel,
+					UMCCH0_0_EccErrCntSel,
+					EccErrCntCsSel, 0);
+	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4,
+			ecc_err_cnt_sel);
+
+	/* clear lower chip error count */
+	WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4,
+			UMC_V6_1_CE_CNT_INIT);
+
+	/* select the higher chip */
+	ecc_err_cnt_sel = RREG32_PCIE((ecc_err_cnt_sel_addr +
+					umc_reg_offset) * 4);
+	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel,
+					UMCCH0_0_EccErrCntSel,
+					EccErrCntCsSel, 1);
+	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4,
+			ecc_err_cnt_sel);
+
+	/* clear higher chip error count */
+	WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4,
+			UMC_V6_1_CE_CNT_INIT);
+}
+
+static void umc_v6_1_clear_error_count(struct amdgpu_device *adev)
+{
+	uint32_t umc_inst        = 0;
+	uint32_t ch_inst         = 0;
+	uint32_t umc_reg_offset  = 0;
+	uint32_t rsmu_umc_index_state =
+				umc_v6_1_get_umc_index_mode_state(adev);
+
+	if (rsmu_umc_index_state)
+		umc_v6_1_disable_umc_index_mode(adev);
+
+	LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) {
+		umc_reg_offset = get_umc_6_reg_offset(adev,
+						umc_inst,
+						ch_inst);
+
+		umc_v6_1_clear_error_count_per_channel(adev,
+						umc_reg_offset);
+	}
+
+	if (rsmu_umc_index_state)
+		umc_v6_1_enable_umc_index_mode(adev);
+}
+
+static void umc_v6_1_query_correctable_error_count(struct amdgpu_device *adev,
+						   uint32_t umc_reg_offset,
+						   unsigned long *error_count)
+{
+	uint32_t ecc_err_cnt_sel, ecc_err_cnt_sel_addr;
+	uint32_t ecc_err_cnt, ecc_err_cnt_addr;
+	uint64_t mc_umc_status;
+	uint32_t mc_umc_status_addr;
+
+	if (adev->asic_type == CHIP_ARCTURUS) {
+		/* UMC 6_1_2 registers */
+		ecc_err_cnt_sel_addr =
+			SOC15_REG_OFFSET(UMC, 0, mmUMCCH0_0_EccErrCntSel_ARCT);
+		ecc_err_cnt_addr =
+			SOC15_REG_OFFSET(UMC, 0, mmUMCCH0_0_EccErrCnt_ARCT);
+		mc_umc_status_addr =
+			SOC15_REG_OFFSET(UMC, 0, mmMCA_UMC_UMC0_MCUMC_STATUST0_ARCT);
+	} else {
+		/* UMC 6_1_1 registers */
+		ecc_err_cnt_sel_addr =
+			SOC15_REG_OFFSET(UMC, 0, mmUMCCH0_0_EccErrCntSel);
+		ecc_err_cnt_addr =
+			SOC15_REG_OFFSET(UMC, 0, mmUMCCH0_0_EccErrCnt);
+		mc_umc_status_addr =
+			SOC15_REG_OFFSET(UMC, 0, mmMCA_UMC_UMC0_MCUMC_STATUST0);
+	}
+
+	/* select the lower chip and check the error count */
+	ecc_err_cnt_sel = RREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4);
+	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, UMCCH0_0_EccErrCntSel,
+					EccErrCntCsSel, 0);
+	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4, ecc_err_cnt_sel);
+
+	ecc_err_cnt = RREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4);
+	*error_count +=
+		(REG_GET_FIELD(ecc_err_cnt, UMCCH0_0_EccErrCnt, EccErrCnt) -
+		 UMC_V6_1_CE_CNT_INIT);
+
+	/* select the higher chip and check the err counter */
+	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, UMCCH0_0_EccErrCntSel,
+					EccErrCntCsSel, 1);
+	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4, ecc_err_cnt_sel);
+
+	ecc_err_cnt = RREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4);
+	*error_count +=
+		(REG_GET_FIELD(ecc_err_cnt, UMCCH0_0_EccErrCnt, EccErrCnt) -
+		 UMC_V6_1_CE_CNT_INIT);
+
+	/* check for SRAM correctable error
+	  MCUMC_STATUS is a 64 bit register */
+	mc_umc_status = RREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4);
+	if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, ErrorCodeExt) == 6 &&
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, CECC) == 1)
+		*error_count += 1;
+}
+
+static void umc_v6_1_querry_uncorrectable_error_count(struct amdgpu_device *adev,
+						      uint32_t umc_reg_offset,
+						      unsigned long *error_count)
+{
+	uint64_t mc_umc_status;
+	uint32_t mc_umc_status_addr;
+
+	if (adev->asic_type == CHIP_ARCTURUS) {
+		/* UMC 6_1_2 registers */
+		mc_umc_status_addr =
+			SOC15_REG_OFFSET(UMC, 0, mmMCA_UMC_UMC0_MCUMC_STATUST0_ARCT);
+	} else {
+		/* UMC 6_1_1 registers */
+		mc_umc_status_addr =
+			SOC15_REG_OFFSET(UMC, 0, mmMCA_UMC_UMC0_MCUMC_STATUST0);
+	}
+
+	/* check the MCUMC_STATUS */
+	mc_umc_status = RREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4);
+	if ((REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1) &&
+	    (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, PCC) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UC) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, TCC) == 1))
+		*error_count += 1;
+}
+
+static void umc_v6_1_query_ras_error_count(struct amdgpu_device *adev,
+					   void *ras_error_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+
+	uint32_t umc_inst        = 0;
+	uint32_t ch_inst         = 0;
+	uint32_t umc_reg_offset  = 0;
+
+	uint32_t rsmu_umc_index_state = umc_v6_1_get_umc_index_mode_state(adev);
+
+	if (rsmu_umc_index_state)
+		umc_v6_1_disable_umc_index_mode(adev);
+
+	if ((adev->asic_type == CHIP_ARCTURUS) &&
+		amdgpu_dpm_set_df_cstate(adev, DF_CSTATE_DISALLOW))
+		DRM_WARN("Fail to disable DF-Cstate.\n");
+
+	LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) {
+		umc_reg_offset = get_umc_6_reg_offset(adev,
+						      umc_inst,
+						      ch_inst);
+
+		umc_v6_1_query_correctable_error_count(adev,
+						       umc_reg_offset,
+						       &(err_data->ce_count));
+		umc_v6_1_querry_uncorrectable_error_count(adev,
+							  umc_reg_offset,
+							  &(err_data->ue_count));
+	}
+
+	if ((adev->asic_type == CHIP_ARCTURUS) &&
+		amdgpu_dpm_set_df_cstate(adev, DF_CSTATE_ALLOW))
+		DRM_WARN("Fail to enable DF-Cstate\n");
+
+	if (rsmu_umc_index_state)
+		umc_v6_1_enable_umc_index_mode(adev);
+
+	umc_v6_1_clear_error_count(adev);
+}
+
+static void umc_v6_1_query_error_address(struct amdgpu_device *adev,
+					 struct ras_err_data *err_data,
+					 uint32_t umc_reg_offset,
+					 uint32_t ch_inst,
+					 uint32_t umc_inst)
+{
+	uint32_t lsb, mc_umc_status_addr;
+	uint64_t mc_umc_status, err_addr, retired_page, mc_umc_addrt0;
+	uint32_t channel_index = adev->umc.channel_idx_tbl[umc_inst * adev->umc.channel_inst_num + ch_inst];
+
+	if (adev->asic_type == CHIP_ARCTURUS) {
+		/* UMC 6_1_2 registers */
+		mc_umc_status_addr =
+			SOC15_REG_OFFSET(UMC, 0, mmMCA_UMC_UMC0_MCUMC_STATUST0_ARCT);
+		mc_umc_addrt0 =
+			SOC15_REG_OFFSET(UMC, 0, mmMCA_UMC_UMC0_MCUMC_ADDRT0_ARCT);
+	} else {
+		/* UMC 6_1_1 registers */
+		mc_umc_status_addr =
+			SOC15_REG_OFFSET(UMC, 0, mmMCA_UMC_UMC0_MCUMC_STATUST0);
+		mc_umc_addrt0 =
+			SOC15_REG_OFFSET(UMC, 0, mmMCA_UMC_UMC0_MCUMC_ADDRT0);
+	}
+
+	mc_umc_status = RREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4);
+
+	if (mc_umc_status == 0)
+		return;
+
+	if (!err_data->err_addr) {
+		/* clear umc status */
+		WREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4, 0x0ULL);
+		return;
+	}
+
+	/* calculate error address if ue error is detected */
+	if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1) {
+
+		err_addr = RREG64_PCIE((mc_umc_addrt0 + umc_reg_offset) * 4);
+		/* the lowest lsb bits should be ignored */
+		lsb = REG_GET_FIELD(err_addr, MCA_UMC_UMC0_MCUMC_ADDRT0, LSB);
+		err_addr = REG_GET_FIELD(err_addr, MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr);
+		err_addr &= ~((0x1ULL << lsb) - 1);
+
+		/* translate umc channel address to soc pa, 3 parts are included */
+		retired_page = ADDR_OF_8KB_BLOCK(err_addr) |
+				ADDR_OF_256B_BLOCK(channel_index) |
+				OFFSET_IN_256B_BLOCK(err_addr);
+
+		amdgpu_umc_fill_error_record(err_data, err_addr,
+					retired_page, channel_index, umc_inst);
+	}
+
+	/* clear umc status */
+	WREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4, 0x0ULL);
+}
+
+static void umc_v6_1_query_ras_error_address(struct amdgpu_device *adev,
+					     void *ras_error_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+
+	uint32_t umc_inst        = 0;
+	uint32_t ch_inst         = 0;
+	uint32_t umc_reg_offset  = 0;
+
+	uint32_t rsmu_umc_index_state = umc_v6_1_get_umc_index_mode_state(adev);
+
+	if (rsmu_umc_index_state)
+		umc_v6_1_disable_umc_index_mode(adev);
+
+	if ((adev->asic_type == CHIP_ARCTURUS) &&
+		amdgpu_dpm_set_df_cstate(adev, DF_CSTATE_DISALLOW))
+		DRM_WARN("Fail to disable DF-Cstate.\n");
+
+	LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) {
+		umc_reg_offset = get_umc_6_reg_offset(adev,
+						      umc_inst,
+						      ch_inst);
+
+		umc_v6_1_query_error_address(adev,
+					     err_data,
+					     umc_reg_offset,
+					     ch_inst,
+					     umc_inst);
+	}
+
+	if ((adev->asic_type == CHIP_ARCTURUS) &&
+		amdgpu_dpm_set_df_cstate(adev, DF_CSTATE_ALLOW))
+		DRM_WARN("Fail to enable DF-Cstate\n");
+
+	if (rsmu_umc_index_state)
+		umc_v6_1_enable_umc_index_mode(adev);
+}
+
+static void umc_v6_1_err_cnt_init_per_channel(struct amdgpu_device *adev,
+					      uint32_t umc_reg_offset)
+{
+	uint32_t ecc_err_cnt_sel, ecc_err_cnt_sel_addr;
+	uint32_t ecc_err_cnt_addr;
+
+	if (adev->asic_type == CHIP_ARCTURUS) {
+		/* UMC 6_1_2 registers */
+		ecc_err_cnt_sel_addr =
+			SOC15_REG_OFFSET(UMC, 0, mmUMCCH0_0_EccErrCntSel_ARCT);
+		ecc_err_cnt_addr =
+			SOC15_REG_OFFSET(UMC, 0, mmUMCCH0_0_EccErrCnt_ARCT);
+	} else {
+		/* UMC 6_1_1 registers */
+		ecc_err_cnt_sel_addr =
+			SOC15_REG_OFFSET(UMC, 0, mmUMCCH0_0_EccErrCntSel);
+		ecc_err_cnt_addr =
+			SOC15_REG_OFFSET(UMC, 0, mmUMCCH0_0_EccErrCnt);
+	}
+
+	/* select the lower chip and check the error count */
+	ecc_err_cnt_sel = RREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4);
+	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, UMCCH0_0_EccErrCntSel,
+					EccErrCntCsSel, 0);
+	/* set ce error interrupt type to APIC based interrupt */
+	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, UMCCH0_0_EccErrCntSel,
+					EccErrInt, 0x1);
+	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4, ecc_err_cnt_sel);
+	/* set error count to initial value */
+	WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4, UMC_V6_1_CE_CNT_INIT);
+
+	/* select the higher chip and check the err counter */
+	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, UMCCH0_0_EccErrCntSel,
+					EccErrCntCsSel, 1);
+	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4, ecc_err_cnt_sel);
+	WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4, UMC_V6_1_CE_CNT_INIT);
+}
+
+static void umc_v6_1_err_cnt_init(struct amdgpu_device *adev)
+{
+	uint32_t umc_inst        = 0;
+	uint32_t ch_inst         = 0;
+	uint32_t umc_reg_offset  = 0;
+
+	uint32_t rsmu_umc_index_state = umc_v6_1_get_umc_index_mode_state(adev);
+
+	if (rsmu_umc_index_state)
+		umc_v6_1_disable_umc_index_mode(adev);
+
+	LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) {
+		umc_reg_offset = get_umc_6_reg_offset(adev,
+						      umc_inst,
+						      ch_inst);
+
+		umc_v6_1_err_cnt_init_per_channel(adev, umc_reg_offset);
+	}
+
+	if (rsmu_umc_index_state)
+		umc_v6_1_enable_umc_index_mode(adev);
+}
+
+const struct amdgpu_ras_block_hw_ops umc_v6_1_ras_hw_ops = {
+	.query_ras_error_count = umc_v6_1_query_ras_error_count,
+	.query_ras_error_address = umc_v6_1_query_ras_error_address,
+};
+
+struct amdgpu_umc_ras umc_v6_1_ras = {
+	.ras_block = {
+		.hw_ops = &umc_v6_1_ras_hw_ops,
+	},
+	.err_cnt_init = umc_v6_1_err_cnt_init,
+};
+\ No newline at end of file
diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_1.h b/drivers/gpu/drm/amd/amdgpu/umc_v6_1.h
new file mode 100644
index 000000000000..50c632eb4cc6
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_1.h
@@ -0,0 +1,52 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __UMC_V6_1_H__
+#define __UMC_V6_1_H__
+
+#include "soc15_common.h"
+#include "amdgpu.h"
+
+/* HBM  Memory Channel Width */
+#define UMC_V6_1_HBM_MEMORY_CHANNEL_WIDTH	128
+/* number of umc channel instance with memory map register access */
+#define UMC_V6_1_CHANNEL_INSTANCE_NUM		4
+/* number of umc instance with memory map register access */
+#define UMC_V6_1_UMC_INSTANCE_NUM		8
+/* total channel instances in one umc block */
+#define UMC_V6_1_TOTAL_CHANNEL_NUM	(UMC_V6_1_CHANNEL_INSTANCE_NUM * UMC_V6_1_UMC_INSTANCE_NUM)
+/* UMC regiser per channel offset */
+#define UMC_V6_1_PER_CHANNEL_OFFSET_VG20	0x800
+#define UMC_V6_1_PER_CHANNEL_OFFSET_ARCT	0x400
+
+/* EccErrCnt max value */
+#define UMC_V6_1_CE_CNT_MAX		0xffff
+/* umc ce interrupt threshold */
+#define UMC_V6_1_CE_INT_THRESHOLD	0xffff
+/* umc ce count initial value */
+#define UMC_V6_1_CE_CNT_INIT	(UMC_V6_1_CE_CNT_MAX - UMC_V6_1_CE_INT_THRESHOLD)
+
+extern struct amdgpu_umc_ras umc_v6_1_ras;
+extern const uint32_t
+	umc_v6_1_channel_idx_tbl[UMC_V6_1_UMC_INSTANCE_NUM][UMC_V6_1_CHANNEL_INSTANCE_NUM];
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c
new file mode 100644
index 000000000000..a3ee3c4c650f
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c
@@ -0,0 +1,528 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "umc_v6_7.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_umc.h"
+#include "amdgpu.h"
+
+#include "umc/umc_6_7_0_offset.h"
+#include "umc/umc_6_7_0_sh_mask.h"
+
+const uint32_t
+	umc_v6_7_channel_idx_tbl_second[UMC_V6_7_UMC_INSTANCE_NUM][UMC_V6_7_CHANNEL_INSTANCE_NUM] = {
+		{28, 20, 24, 16, 12, 4, 8, 0},
+		{6, 30, 2, 26, 22, 14, 18, 10},
+		{19, 11, 15, 7, 3, 27, 31, 23},
+		{9, 1, 5, 29, 25, 17, 21, 13}
+};
+const uint32_t
+	umc_v6_7_channel_idx_tbl_first[UMC_V6_7_UMC_INSTANCE_NUM][UMC_V6_7_CHANNEL_INSTANCE_NUM] = {
+		{19, 11, 15, 7,	3, 27, 31, 23},
+		{9, 1, 5, 29, 25, 17, 21, 13},
+		{28, 20, 24, 16, 12, 4, 8, 0},
+		{6, 30, 2, 26, 22, 14, 18, 10},
+};
+
+static inline uint32_t get_umc_v6_7_reg_offset(struct amdgpu_device *adev,
+					      uint32_t umc_inst,
+					      uint32_t ch_inst)
+{
+	uint32_t index = umc_inst * adev->umc.channel_inst_num + ch_inst;
+
+	/* adjust umc and channel index offset,
+	 * the register address is not linear on each umc instace */
+	umc_inst = index / 4;
+	ch_inst = index % 4;
+
+	return adev->umc.channel_offs * ch_inst + UMC_V6_7_INST_DIST * umc_inst;
+}
+
+static void umc_v6_7_query_error_status_helper(struct amdgpu_device *adev,
+						  uint64_t mc_umc_status, uint32_t umc_reg_offset)
+{
+	uint32_t mc_umc_addr;
+	uint64_t reg_value;
+
+	if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1)
+		dev_info(adev->dev, "Deferred error\n");
+
+	if (mc_umc_status)
+		dev_info(adev->dev, "MCA STATUS 0x%llx, umc_reg_offset 0x%x\n", mc_umc_status, umc_reg_offset);
+
+	/* print IPID registers value */
+	mc_umc_addr =
+		SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_IPIDT0);
+	reg_value = RREG64_PCIE((mc_umc_addr + umc_reg_offset) * 4);
+	if (reg_value)
+		dev_info(adev->dev, "MCA IPID 0x%llx, umc_reg_offset 0x%x\n", reg_value, umc_reg_offset);
+
+	/* print SYND registers value */
+	mc_umc_addr =
+		SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_SYNDT0);
+	reg_value = RREG64_PCIE((mc_umc_addr + umc_reg_offset) * 4);
+	if (reg_value)
+		dev_info(adev->dev, "MCA SYND 0x%llx, umc_reg_offset 0x%x\n", reg_value, umc_reg_offset);
+
+	/* print MISC0 registers value */
+	mc_umc_addr =
+		SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_MISC0T0);
+	reg_value = RREG64_PCIE((mc_umc_addr + umc_reg_offset) * 4);
+	if (reg_value)
+		dev_info(adev->dev, "MCA MISC0 0x%llx, umc_reg_offset 0x%x\n", reg_value, umc_reg_offset);
+}
+
+static void umc_v6_7_ecc_info_query_correctable_error_count(struct amdgpu_device *adev,
+						   uint32_t umc_inst, uint32_t ch_inst,
+						   unsigned long *error_count)
+{
+	uint64_t mc_umc_status;
+	uint32_t eccinfo_table_idx;
+	uint32_t umc_reg_offset;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+
+	umc_reg_offset = get_umc_v6_7_reg_offset(adev,
+						umc_inst, ch_inst);
+
+	eccinfo_table_idx = umc_inst * adev->umc.channel_inst_num + ch_inst;
+	/* check for SRAM correctable error
+	  MCUMC_STATUS is a 64 bit register */
+	mc_umc_status = ras->umc_ecc.ecc[eccinfo_table_idx].mca_umc_status;
+	if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, CECC) == 1) {
+		*error_count += 1;
+
+		umc_v6_7_query_error_status_helper(adev, mc_umc_status, umc_reg_offset);
+
+		if (ras->umc_ecc.record_ce_addr_supported)	{
+			uint64_t err_addr, soc_pa;
+			uint32_t channel_index =
+				adev->umc.channel_idx_tbl[umc_inst * adev->umc.channel_inst_num + ch_inst];
+
+			err_addr = ras->umc_ecc.ecc[eccinfo_table_idx].mca_ceumc_addr;
+			err_addr = REG_GET_FIELD(err_addr, MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr);
+			/* translate umc channel address to soc pa, 3 parts are included */
+			soc_pa = ADDR_OF_8KB_BLOCK(err_addr) |
+					ADDR_OF_256B_BLOCK(channel_index) |
+					OFFSET_IN_256B_BLOCK(err_addr);
+
+			/* The umc channel bits are not original values, they are hashed */
+			SET_CHANNEL_HASH(channel_index, soc_pa);
+
+			dev_info(adev->dev, "Error Address(PA): 0x%llx\n", soc_pa);
+		}
+	}
+}
+
+static void umc_v6_7_ecc_info_querry_uncorrectable_error_count(struct amdgpu_device *adev,
+							  uint32_t umc_inst, uint32_t ch_inst,
+						      unsigned long *error_count)
+{
+	uint64_t mc_umc_status;
+	uint32_t eccinfo_table_idx;
+	uint32_t umc_reg_offset;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+
+	umc_reg_offset = get_umc_v6_7_reg_offset(adev,
+						umc_inst, ch_inst);
+
+	eccinfo_table_idx = umc_inst * adev->umc.channel_inst_num + ch_inst;
+	/* check the MCUMC_STATUS */
+	mc_umc_status = ras->umc_ecc.ecc[eccinfo_table_idx].mca_umc_status;
+	if ((REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1) &&
+	    (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, PCC) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UC) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, TCC) == 1)) {
+		*error_count += 1;
+
+		umc_v6_7_query_error_status_helper(adev, mc_umc_status, umc_reg_offset);
+	}
+}
+
+static int umc_v6_7_ecc_info_querry_ecc_error_count(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)data;
+
+	umc_v6_7_ecc_info_query_correctable_error_count(adev,
+		umc_inst, ch_inst,
+		&(err_data->ce_count));
+
+	umc_v6_7_ecc_info_querry_uncorrectable_error_count(adev,
+		umc_inst, ch_inst,
+		&(err_data->ue_count));
+
+	return 0;
+}
+
+static void umc_v6_7_ecc_info_query_ras_error_count(struct amdgpu_device *adev,
+					   void *ras_error_status)
+{
+	amdgpu_umc_loop_channels(adev,
+		umc_v6_7_ecc_info_querry_ecc_error_count, ras_error_status);
+}
+
+void umc_v6_7_convert_error_address(struct amdgpu_device *adev,
+				    struct ras_err_data *err_data, uint64_t err_addr,
+				    uint32_t ch_inst, uint32_t umc_inst)
+{
+	uint32_t channel_index;
+	uint64_t soc_pa, retired_page, column;
+
+	channel_index =
+		adev->umc.channel_idx_tbl[umc_inst * adev->umc.channel_inst_num + ch_inst];
+	/* translate umc channel address to soc pa, 3 parts are included */
+	soc_pa = ADDR_OF_8KB_BLOCK(err_addr) |
+			ADDR_OF_256B_BLOCK(channel_index) |
+			OFFSET_IN_256B_BLOCK(err_addr);
+
+	/* The umc channel bits are not original values, they are hashed */
+	SET_CHANNEL_HASH(channel_index, soc_pa);
+
+	/* clear [C4 C3 C2] in soc physical address */
+	soc_pa &= ~(0x7ULL << UMC_V6_7_PA_C2_BIT);
+
+	/* loop for all possibilities of [C4 C3 C2] */
+	for (column = 0; column < UMC_V6_7_NA_MAP_PA_NUM; column++) {
+		retired_page = soc_pa | (column << UMC_V6_7_PA_C2_BIT);
+		dev_info(adev->dev, "Error Address(PA): 0x%llx\n", retired_page);
+		amdgpu_umc_fill_error_record(err_data, err_addr,
+			retired_page, channel_index, umc_inst);
+
+		/* shift R14 bit */
+		retired_page ^= (0x1ULL << UMC_V6_7_PA_R14_BIT);
+		dev_info(adev->dev, "Error Address(PA): 0x%llx\n", retired_page);
+		amdgpu_umc_fill_error_record(err_data, err_addr,
+			retired_page, channel_index, umc_inst);
+	}
+}
+
+static int umc_v6_7_ecc_info_query_error_address(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
+{
+	uint64_t mc_umc_status, err_addr;
+	uint32_t eccinfo_table_idx;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+	struct ras_err_data *err_data = (struct ras_err_data *)data;
+
+	eccinfo_table_idx = umc_inst * adev->umc.channel_inst_num + ch_inst;
+	mc_umc_status = ras->umc_ecc.ecc[eccinfo_table_idx].mca_umc_status;
+
+	if (mc_umc_status == 0)
+		return 0;
+
+	if (!err_data->err_addr)
+		return 0;
+
+	/* calculate error address if ue error is detected */
+	if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1) {
+
+		err_addr = ras->umc_ecc.ecc[eccinfo_table_idx].mca_umc_addr;
+		err_addr = REG_GET_FIELD(err_addr, MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr);
+
+		umc_v6_7_convert_error_address(adev, err_data, err_addr,
+					ch_inst, umc_inst);
+	}
+
+	return 0;
+}
+
+static void umc_v6_7_ecc_info_query_ras_error_address(struct amdgpu_device *adev,
+					     void *ras_error_status)
+{
+	amdgpu_umc_loop_channels(adev,
+	    umc_v6_7_ecc_info_query_error_address, ras_error_status);
+}
+
+static void umc_v6_7_query_correctable_error_count(struct amdgpu_device *adev,
+						   uint32_t umc_reg_offset,
+						   unsigned long *error_count,
+						   uint32_t ch_inst,
+						   uint32_t umc_inst)
+{
+	uint32_t ecc_err_cnt_sel, ecc_err_cnt_sel_addr;
+	uint32_t ecc_err_cnt, ecc_err_cnt_addr;
+	uint64_t mc_umc_status;
+	uint32_t mc_umc_status_addr;
+
+	/* UMC 6_1_1 registers */
+	ecc_err_cnt_sel_addr =
+		SOC15_REG_OFFSET(UMC, 0, regUMCCH0_0_EccErrCntSel);
+	ecc_err_cnt_addr =
+		SOC15_REG_OFFSET(UMC, 0, regUMCCH0_0_EccErrCnt);
+	mc_umc_status_addr =
+		SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_STATUST0);
+
+	/* select the lower chip and check the error count */
+	ecc_err_cnt_sel = RREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4);
+	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, UMCCH0_0_EccErrCntSel,
+					EccErrCntCsSel, 0);
+	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4, ecc_err_cnt_sel);
+
+	ecc_err_cnt = RREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4);
+	*error_count +=
+		(REG_GET_FIELD(ecc_err_cnt, UMCCH0_0_EccErrCnt, EccErrCnt) -
+		 UMC_V6_7_CE_CNT_INIT);
+
+	/* select the higher chip and check the err counter */
+	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, UMCCH0_0_EccErrCntSel,
+					EccErrCntCsSel, 1);
+	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4, ecc_err_cnt_sel);
+
+	ecc_err_cnt = RREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4);
+	*error_count +=
+		(REG_GET_FIELD(ecc_err_cnt, UMCCH0_0_EccErrCnt, EccErrCnt) -
+		 UMC_V6_7_CE_CNT_INIT);
+
+	/* check for SRAM correctable error
+	  MCUMC_STATUS is a 64 bit register */
+	mc_umc_status = RREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4);
+	if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, CECC) == 1) {
+		*error_count += 1;
+
+		umc_v6_7_query_error_status_helper(adev, mc_umc_status, umc_reg_offset);
+
+		{
+			uint64_t err_addr, soc_pa;
+			uint32_t mc_umc_addrt0;
+			uint32_t channel_index;
+
+			mc_umc_addrt0 =
+				SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_ADDRT0);
+
+			channel_index =
+				adev->umc.channel_idx_tbl[umc_inst * adev->umc.channel_inst_num + ch_inst];
+
+			err_addr = RREG64_PCIE((mc_umc_addrt0 + umc_reg_offset) * 4);
+			err_addr = REG_GET_FIELD(err_addr, MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr);
+
+			/* translate umc channel address to soc pa, 3 parts are included */
+			soc_pa = ADDR_OF_8KB_BLOCK(err_addr) |
+					ADDR_OF_256B_BLOCK(channel_index) |
+					OFFSET_IN_256B_BLOCK(err_addr);
+
+			/* The umc channel bits are not original values, they are hashed */
+			SET_CHANNEL_HASH(channel_index, soc_pa);
+
+			dev_info(adev->dev, "Error Address(PA): 0x%llx\n", soc_pa);
+		}
+	}
+}
+
+static void umc_v6_7_querry_uncorrectable_error_count(struct amdgpu_device *adev,
+						      uint32_t umc_reg_offset,
+						      unsigned long *error_count)
+{
+	uint64_t mc_umc_status;
+	uint32_t mc_umc_status_addr;
+
+	mc_umc_status_addr =
+		SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_STATUST0);
+
+	/* check the MCUMC_STATUS */
+	mc_umc_status = RREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4);
+	if ((REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1) &&
+	    (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, PCC) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UC) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, TCC) == 1)) {
+		*error_count += 1;
+
+		umc_v6_7_query_error_status_helper(adev, mc_umc_status, umc_reg_offset);
+	}
+}
+
+static int umc_v6_7_reset_error_count_per_channel(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
+{
+	uint32_t ecc_err_cnt_addr;
+	uint32_t ecc_err_cnt_sel, ecc_err_cnt_sel_addr;
+	uint32_t umc_reg_offset =
+		get_umc_v6_7_reg_offset(adev, umc_inst, ch_inst);
+
+	ecc_err_cnt_sel_addr =
+		SOC15_REG_OFFSET(UMC, 0,
+				regUMCCH0_0_EccErrCntSel);
+	ecc_err_cnt_addr =
+		SOC15_REG_OFFSET(UMC, 0,
+				regUMCCH0_0_EccErrCnt);
+
+	/* select the lower chip */
+	ecc_err_cnt_sel = RREG32_PCIE((ecc_err_cnt_sel_addr +
+				       umc_reg_offset) * 4);
+	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel,
+					UMCCH0_0_EccErrCntSel,
+					EccErrCntCsSel, 0);
+	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4,
+			ecc_err_cnt_sel);
+
+	/* clear lower chip error count */
+	WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4,
+			UMC_V6_7_CE_CNT_INIT);
+
+	/* select the higher chip */
+	ecc_err_cnt_sel = RREG32_PCIE((ecc_err_cnt_sel_addr +
+					umc_reg_offset) * 4);
+	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel,
+					UMCCH0_0_EccErrCntSel,
+					EccErrCntCsSel, 1);
+	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4,
+			ecc_err_cnt_sel);
+
+	/* clear higher chip error count */
+	WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4,
+			UMC_V6_7_CE_CNT_INIT);
+
+	return 0;
+}
+
+static void umc_v6_7_reset_error_count(struct amdgpu_device *adev)
+{
+	amdgpu_umc_loop_channels(adev,
+		umc_v6_7_reset_error_count_per_channel, NULL);
+}
+
+static int umc_v6_7_query_ecc_error_count(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)data;
+	uint32_t umc_reg_offset =
+		get_umc_v6_7_reg_offset(adev, umc_inst, ch_inst);
+
+	umc_v6_7_query_correctable_error_count(adev,
+					umc_reg_offset,
+					&(err_data->ce_count),
+					ch_inst, umc_inst);
+
+	umc_v6_7_querry_uncorrectable_error_count(adev,
+					umc_reg_offset,
+					&(err_data->ue_count));
+
+	return 0;
+}
+
+static void umc_v6_7_query_ras_error_count(struct amdgpu_device *adev,
+					   void *ras_error_status)
+{
+	amdgpu_umc_loop_channels(adev,
+		umc_v6_7_query_ecc_error_count, ras_error_status);
+
+	umc_v6_7_reset_error_count(adev);
+}
+
+static int umc_v6_7_query_error_address(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
+{
+	uint32_t mc_umc_status_addr;
+	uint64_t mc_umc_status = 0, mc_umc_addrt0, err_addr;
+	struct ras_err_data *err_data = (struct ras_err_data *)data;
+	uint32_t umc_reg_offset =
+		get_umc_v6_7_reg_offset(adev, umc_inst, ch_inst);
+
+	mc_umc_status_addr =
+		SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_STATUST0);
+	mc_umc_addrt0 =
+		SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_ADDRT0);
+
+	mc_umc_status = RREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4);
+
+	if (mc_umc_status == 0)
+		return 0;
+
+	if (!err_data->err_addr) {
+		/* clear umc status */
+		WREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4, 0x0ULL);
+		return 0;
+	}
+
+	/* calculate error address if ue error is detected */
+	if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1) {
+		err_addr = RREG64_PCIE((mc_umc_addrt0 + umc_reg_offset) * 4);
+		err_addr =
+			REG_GET_FIELD(err_addr, MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr);
+
+		umc_v6_7_convert_error_address(adev, err_data, err_addr,
+					ch_inst, umc_inst);
+	}
+
+	/* clear umc status */
+	WREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4, 0x0ULL);
+
+	return 0;
+}
+
+static void umc_v6_7_query_ras_error_address(struct amdgpu_device *adev,
+					     void *ras_error_status)
+{
+	amdgpu_umc_loop_channels(adev,
+		umc_v6_7_query_error_address, ras_error_status);
+}
+
+static uint32_t umc_v6_7_query_ras_poison_mode_per_channel(
+						struct amdgpu_device *adev,
+						uint32_t umc_reg_offset)
+{
+	uint32_t ecc_ctrl_addr, ecc_ctrl;
+
+	ecc_ctrl_addr =
+		SOC15_REG_OFFSET(UMC, 0, regUMCCH0_0_EccCtrl);
+	ecc_ctrl = RREG32_PCIE((ecc_ctrl_addr +
+					umc_reg_offset) * 4);
+
+	return REG_GET_FIELD(ecc_ctrl, UMCCH0_0_EccCtrl, UCFatalEn);
+}
+
+static bool umc_v6_7_query_ras_poison_mode(struct amdgpu_device *adev)
+{
+	uint32_t umc_reg_offset  = 0;
+
+	/* Enabling fatal error in umc instance0 channel0 will be
+	 * considered as fatal error mode
+	 */
+	umc_reg_offset = get_umc_v6_7_reg_offset(adev, 0, 0);
+	return !umc_v6_7_query_ras_poison_mode_per_channel(adev, umc_reg_offset);
+}
+
+const struct amdgpu_ras_block_hw_ops umc_v6_7_ras_hw_ops = {
+	.query_ras_error_count = umc_v6_7_query_ras_error_count,
+	.query_ras_error_address = umc_v6_7_query_ras_error_address,
+};
+
+struct amdgpu_umc_ras umc_v6_7_ras = {
+	.ras_block = {
+		.hw_ops = &umc_v6_7_ras_hw_ops,
+	},
+	.query_ras_poison_mode = umc_v6_7_query_ras_poison_mode,
+	.ecc_info_query_ras_error_count = umc_v6_7_ecc_info_query_ras_error_count,
+	.ecc_info_query_ras_error_address = umc_v6_7_ecc_info_query_ras_error_address,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_7.h b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.h
new file mode 100644
index 000000000000..105245d5b6e5
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.h
@@ -0,0 +1,77 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __UMC_V6_7_H__
+#define __UMC_V6_7_H__
+
+#include "soc15_common.h"
+#include "amdgpu.h"
+
+/* EccErrCnt max value */
+#define UMC_V6_7_CE_CNT_MAX		0xffff
+/* umc ce interrupt threshold */
+#define UMC_V6_7_CE_INT_THRESHOLD	0xffff
+/* umc ce count initial value */
+#define UMC_V6_7_CE_CNT_INIT	(UMC_V6_7_CE_CNT_MAX - UMC_V6_7_CE_INT_THRESHOLD)
+
+#define UMC_V6_7_INST_DIST	0x40000
+
+/* number of umc channel instance with memory map register access */
+#define UMC_V6_7_UMC_INSTANCE_NUM		4
+/* number of umc instance with memory map register access */
+#define UMC_V6_7_CHANNEL_INSTANCE_NUM		8
+/* total channel instances in one umc block */
+#define UMC_V6_7_TOTAL_CHANNEL_NUM	(UMC_V6_7_CHANNEL_INSTANCE_NUM * UMC_V6_7_UMC_INSTANCE_NUM)
+/* one piece of normalizing address is mapped to 8 pieces of physical address */
+#define UMC_V6_7_NA_MAP_PA_NUM	8
+/* R14 bit shift should be considered, double the number */
+#define UMC_V6_7_BAD_PAGE_NUM_PER_CHANNEL	(UMC_V6_7_NA_MAP_PA_NUM * 2)
+/* The CH4 bit in SOC physical address */
+#define UMC_V6_7_PA_CH4_BIT	12
+/* The C2 bit in SOC physical address */
+#define UMC_V6_7_PA_C2_BIT	17
+/* The R14 bit in SOC physical address */
+#define UMC_V6_7_PA_R14_BIT	34
+/* UMC regiser per channel offset */
+#define UMC_V6_7_PER_CHANNEL_OFFSET		0x400
+
+/* XOR bit 20, 25, 34 of PA into CH4 bit (bit 12 of PA),
+ * hash bit is only effective when related setting is enabled
+ */
+#define CHANNEL_HASH(channel_idx, pa) (((channel_idx) >> 4) ^ \
+			(((pa)  >> 20) & 0x1ULL & adev->df.hash_status.hash_64k) ^ \
+			(((pa)  >> 25) & 0x1ULL & adev->df.hash_status.hash_2m) ^ \
+			(((pa)  >> 34) & 0x1ULL & adev->df.hash_status.hash_1g))
+#define SET_CHANNEL_HASH(channel_idx, pa) do { \
+		(pa) &= ~(0x1ULL << UMC_V6_7_PA_CH4_BIT); \
+		(pa) |= (CHANNEL_HASH(channel_idx, pa) << UMC_V6_7_PA_CH4_BIT); \
+	} while (0)
+
+extern struct amdgpu_umc_ras umc_v6_7_ras;
+extern const uint32_t
+	umc_v6_7_channel_idx_tbl_second[UMC_V6_7_UMC_INSTANCE_NUM][UMC_V6_7_CHANNEL_INSTANCE_NUM];
+extern const uint32_t
+	umc_v6_7_channel_idx_tbl_first[UMC_V6_7_UMC_INSTANCE_NUM][UMC_V6_7_CHANNEL_INSTANCE_NUM];
+void umc_v6_7_convert_error_address(struct amdgpu_device *adev,
+                                    struct ras_err_data *err_data, uint64_t err_addr,
+                                    uint32_t ch_inst, uint32_t umc_inst);
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v8_10.c b/drivers/gpu/drm/amd/amdgpu/umc_v8_10.c
new file mode 100644
index 000000000000..a32f87992f20
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/umc_v8_10.c
@@ -0,0 +1,458 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "umc_v8_10.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_umc.h"
+#include "amdgpu.h"
+#include "umc/umc_8_10_0_offset.h"
+#include "umc/umc_8_10_0_sh_mask.h"
+
+#define UMC_8_NODE_DIST   0x800000
+#define UMC_8_INST_DIST   0x4000
+
+struct channelnum_map_colbit {
+	uint32_t channel_num;
+	uint32_t col_bit;
+};
+
+const struct channelnum_map_colbit umc_v8_10_channelnum_map_colbit_table[] = {
+	{24, 13},
+	{20, 13},
+	{16, 12},
+	{14, 12},
+	{12, 12},
+	{10, 12},
+	{6,  11},
+};
+
+const uint32_t
+	umc_v8_10_channel_idx_tbl_ext0[]
+				[UMC_V8_10_UMC_INSTANCE_NUM]
+				[UMC_V8_10_CHANNEL_INSTANCE_NUM] = {
+	   {{1,   5}, {7,  3}},
+	   {{14, 15}, {13, 12}},
+	   {{10, 11}, {9,  8}},
+	   {{6,   2}, {0,  4}}
+	};
+
+const uint32_t
+	umc_v8_10_channel_idx_tbl[]
+				[UMC_V8_10_UMC_INSTANCE_NUM]
+				[UMC_V8_10_CHANNEL_INSTANCE_NUM] = {
+	   {{16, 18}, {17, 19}},
+	   {{15, 11}, {3,   7}},
+	   {{1,   5}, {13,  9}},
+	   {{23, 21}, {22, 20}},
+	   {{0,   4}, {12,  8}},
+	   {{14, 10}, {2,   6}}
+	};
+
+static inline uint32_t get_umc_v8_10_reg_offset(struct amdgpu_device *adev,
+					    uint32_t node_inst,
+					    uint32_t umc_inst,
+					    uint32_t ch_inst)
+{
+	return adev->umc.channel_offs * ch_inst + UMC_8_INST_DIST * umc_inst +
+		UMC_8_NODE_DIST * node_inst;
+}
+
+static int umc_v8_10_clear_error_count_per_channel(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
+{
+	uint32_t ecc_err_cnt_addr;
+	uint32_t umc_reg_offset =
+		get_umc_v8_10_reg_offset(adev, node_inst, umc_inst, ch_inst);
+
+	ecc_err_cnt_addr =
+		SOC15_REG_OFFSET(UMC, 0, regUMCCH0_0_GeccErrCnt);
+
+	/* clear error count */
+	WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4,
+			UMC_V8_10_CE_CNT_INIT);
+
+	return 0;
+}
+
+static void umc_v8_10_clear_error_count(struct amdgpu_device *adev)
+{
+	amdgpu_umc_loop_channels(adev,
+		umc_v8_10_clear_error_count_per_channel, NULL);
+}
+
+static void umc_v8_10_query_correctable_error_count(struct amdgpu_device *adev,
+						   uint32_t umc_reg_offset,
+						   unsigned long *error_count)
+{
+	uint64_t mc_umc_status;
+	uint32_t mc_umc_status_addr;
+
+	/* UMC 8_10 registers */
+	mc_umc_status_addr =
+		SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_STATUST0);
+
+	/* Rely on MCUMC_STATUS for correctable error counter
+	 * MCUMC_STATUS is a 64 bit register
+	 */
+	mc_umc_status = RREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4);
+	if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, CECC) == 1)
+		*error_count += 1;
+}
+
+static void umc_v8_10_query_uncorrectable_error_count(struct amdgpu_device *adev,
+						      uint32_t umc_reg_offset,
+						      unsigned long *error_count)
+{
+	uint64_t mc_umc_status;
+	uint32_t mc_umc_status_addr;
+
+	mc_umc_status_addr = SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_STATUST0);
+
+	/* Check the MCUMC_STATUS. */
+	mc_umc_status = RREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4);
+	if ((REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1) &&
+	    (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, PCC) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UC) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, TCC) == 1))
+		*error_count += 1;
+}
+
+static int umc_v8_10_query_ecc_error_count(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)data;
+	uint32_t umc_reg_offset =
+		get_umc_v8_10_reg_offset(adev, node_inst, umc_inst, ch_inst);
+
+	umc_v8_10_query_correctable_error_count(adev,
+					umc_reg_offset,
+					&(err_data->ce_count));
+	umc_v8_10_query_uncorrectable_error_count(adev,
+					umc_reg_offset,
+					&(err_data->ue_count));
+
+	return 0;
+}
+
+static void umc_v8_10_query_ras_error_count(struct amdgpu_device *adev,
+					   void *ras_error_status)
+{
+	amdgpu_umc_loop_channels(adev,
+		umc_v8_10_query_ecc_error_count, ras_error_status);
+
+	umc_v8_10_clear_error_count(adev);
+}
+
+static uint32_t umc_v8_10_get_col_bit(uint32_t channel_num)
+{
+	uint32_t t = 0;
+
+	for (t = 0; t < ARRAY_SIZE(umc_v8_10_channelnum_map_colbit_table); t++)
+		if (channel_num == umc_v8_10_channelnum_map_colbit_table[t].channel_num)
+			return umc_v8_10_channelnum_map_colbit_table[t].col_bit;
+
+	/* Failed to get col_bit. */
+	return U32_MAX;
+}
+
+/*
+ * Mapping normal address to soc physical address in swizzle mode.
+ */
+static int umc_v8_10_swizzle_mode_na_to_pa(struct amdgpu_device *adev,
+					uint32_t channel_idx,
+					uint64_t na, uint64_t *soc_pa)
+{
+	uint32_t channel_num = UMC_V8_10_TOTAL_CHANNEL_NUM(adev);
+	uint32_t col_bit = umc_v8_10_get_col_bit(channel_num);
+	uint64_t tmp_addr;
+
+	if (col_bit == U32_MAX)
+		return -1;
+
+	tmp_addr = SWIZZLE_MODE_TMP_ADDR(na, channel_num, channel_idx);
+	*soc_pa = SWIZZLE_MODE_ADDR_HI(tmp_addr, col_bit) |
+		SWIZZLE_MODE_ADDR_MID(na, col_bit) |
+		SWIZZLE_MODE_ADDR_LOW(tmp_addr, col_bit) |
+		SWIZZLE_MODE_ADDR_LSB(na);
+
+	return 0;
+}
+
+static void umc_v8_10_convert_error_address(struct amdgpu_device *adev,
+					    struct ras_err_data *err_data, uint64_t err_addr,
+					    uint32_t ch_inst, uint32_t umc_inst,
+					    uint32_t node_inst, uint64_t mc_umc_status)
+{
+	uint64_t na_err_addr_base;
+	uint64_t na_err_addr, retired_page_addr;
+	uint32_t channel_index, addr_lsb, col = 0;
+	int ret = 0;
+
+	channel_index =
+		adev->umc.channel_idx_tbl[node_inst * adev->umc.umc_inst_num *
+					adev->umc.channel_inst_num +
+					umc_inst * adev->umc.channel_inst_num +
+					ch_inst];
+
+	/* the lowest lsb bits should be ignored */
+	addr_lsb = REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, AddrLsb);
+	err_addr &= ~((0x1ULL << addr_lsb) - 1);
+	na_err_addr_base = err_addr & ~(0x3ULL << UMC_V8_10_NA_C5_BIT);
+
+	/* loop for all possibilities of [C6 C5] in normal address. */
+	for (col = 0; col < UMC_V8_10_NA_COL_2BITS_POWER_OF_2_NUM; col++) {
+		na_err_addr = na_err_addr_base | (col << UMC_V8_10_NA_C5_BIT);
+
+		/* Mapping normal error address to retired soc physical address. */
+		ret = umc_v8_10_swizzle_mode_na_to_pa(adev, channel_index,
+						na_err_addr, &retired_page_addr);
+		if (ret) {
+			dev_err(adev->dev, "Failed to map pa from umc na.\n");
+			break;
+		}
+		dev_info(adev->dev, "Error Address(PA): 0x%llx\n",
+			retired_page_addr);
+		amdgpu_umc_fill_error_record(err_data, na_err_addr,
+				retired_page_addr, channel_index, umc_inst);
+	}
+}
+
+static int umc_v8_10_query_error_address(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
+{
+	uint64_t mc_umc_status_addr;
+	uint64_t mc_umc_status, err_addr;
+	uint64_t mc_umc_addrt0;
+	struct ras_err_data *err_data = (struct ras_err_data *)data;
+	uint32_t umc_reg_offset =
+		get_umc_v8_10_reg_offset(adev, node_inst, umc_inst, ch_inst);
+
+	mc_umc_status_addr =
+		SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_STATUST0);
+	mc_umc_status = RREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4);
+
+	if (mc_umc_status == 0)
+		return 0;
+
+	if (!err_data->err_addr) {
+		/* clear umc status */
+		WREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4, 0x0ULL);
+		return 0;
+	}
+
+	/* calculate error address if ue error is detected */
+	if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, AddrV) == 1 &&
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1) {
+
+		mc_umc_addrt0 = SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_ADDRT0);
+		err_addr = RREG64_PCIE((mc_umc_addrt0 + umc_reg_offset) * 4);
+		err_addr = REG_GET_FIELD(err_addr, MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr);
+
+		umc_v8_10_convert_error_address(adev, err_data, err_addr,
+					ch_inst, umc_inst, node_inst, mc_umc_status);
+	}
+
+	/* clear umc status */
+	WREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4, 0x0ULL);
+
+	return 0;
+}
+
+static void umc_v8_10_query_ras_error_address(struct amdgpu_device *adev,
+					     void *ras_error_status)
+{
+	amdgpu_umc_loop_channels(adev,
+		umc_v8_10_query_error_address, ras_error_status);
+}
+
+static int umc_v8_10_err_cnt_init_per_channel(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
+{
+	uint32_t ecc_err_cnt_sel, ecc_err_cnt_sel_addr;
+	uint32_t ecc_err_cnt_addr;
+	uint32_t umc_reg_offset =
+		get_umc_v8_10_reg_offset(adev, node_inst, umc_inst, ch_inst);
+
+	ecc_err_cnt_sel_addr =
+		SOC15_REG_OFFSET(UMC, 0, regUMCCH0_0_GeccErrCntSel);
+	ecc_err_cnt_addr =
+		SOC15_REG_OFFSET(UMC, 0, regUMCCH0_0_GeccErrCnt);
+
+	ecc_err_cnt_sel = RREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4);
+
+	/* set ce error interrupt type to APIC based interrupt */
+	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, UMCCH0_0_GeccErrCntSel,
+					GeccErrInt, 0x1);
+	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4, ecc_err_cnt_sel);
+	/* set error count to initial value */
+	WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4, UMC_V8_10_CE_CNT_INIT);
+
+	return 0;
+}
+
+static void umc_v8_10_err_cnt_init(struct amdgpu_device *adev)
+{
+	amdgpu_umc_loop_channels(adev,
+		umc_v8_10_err_cnt_init_per_channel, NULL);
+}
+
+static bool umc_v8_10_query_ras_poison_mode(struct amdgpu_device *adev)
+{
+	/*
+	 * Force return true, because UMCCH0_0_GeccCtrl
+	 * is not accessible from host side
+	 */
+	return true;
+}
+
+static void umc_v8_10_ecc_info_query_correctable_error_count(struct amdgpu_device *adev,
+				      uint32_t node_inst, uint32_t umc_inst, uint32_t ch_inst,
+				      unsigned long *error_count)
+{
+	uint16_t ecc_ce_cnt;
+	uint32_t eccinfo_table_idx;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+
+	eccinfo_table_idx = node_inst * adev->umc.umc_inst_num *
+				  adev->umc.channel_inst_num +
+				  umc_inst * adev->umc.channel_inst_num +
+				  ch_inst;
+
+	/* Retrieve CE count */
+	ecc_ce_cnt = ras->umc_ecc.ecc[eccinfo_table_idx].ce_count_lo_chip;
+	if (ecc_ce_cnt)
+		*error_count += ecc_ce_cnt;
+}
+
+static void umc_v8_10_ecc_info_query_uncorrectable_error_count(struct amdgpu_device *adev,
+				      uint32_t node_inst, uint32_t umc_inst, uint32_t ch_inst,
+				      unsigned long *error_count)
+{
+	uint64_t mc_umc_status;
+	uint32_t eccinfo_table_idx;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+
+	eccinfo_table_idx = node_inst * adev->umc.umc_inst_num *
+				  adev->umc.channel_inst_num +
+				  umc_inst * adev->umc.channel_inst_num +
+				  ch_inst;
+
+	/* check the MCUMC_STATUS */
+	mc_umc_status = ras->umc_ecc.ecc[eccinfo_table_idx].mca_umc_status;
+	if ((REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1) &&
+	    (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, PCC) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UC) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, TCC) == 1)) {
+		*error_count += 1;
+	}
+}
+
+static int umc_v8_10_ecc_info_query_ecc_error_count(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)data;
+
+	umc_v8_10_ecc_info_query_correctable_error_count(adev,
+					node_inst, umc_inst, ch_inst,
+					&(err_data->ce_count));
+	umc_v8_10_ecc_info_query_uncorrectable_error_count(adev,
+					node_inst, umc_inst, ch_inst,
+					&(err_data->ue_count));
+	return 0;
+}
+
+static void umc_v8_10_ecc_info_query_ras_error_count(struct amdgpu_device *adev,
+					void *ras_error_status)
+{
+	amdgpu_umc_loop_channels(adev,
+		umc_v8_10_ecc_info_query_ecc_error_count, ras_error_status);
+}
+
+static int umc_v8_10_ecc_info_query_error_address(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
+{
+	uint32_t eccinfo_table_idx;
+	uint64_t mc_umc_status, err_addr;
+	struct ras_err_data *err_data = (struct ras_err_data *)data;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+
+	eccinfo_table_idx = node_inst * adev->umc.umc_inst_num *
+				  adev->umc.channel_inst_num +
+				  umc_inst * adev->umc.channel_inst_num +
+				  ch_inst;
+
+	mc_umc_status = ras->umc_ecc.ecc[eccinfo_table_idx].mca_umc_status;
+
+	if (mc_umc_status == 0)
+		return 0;
+
+	if (!err_data->err_addr)
+		return 0;
+
+	/* calculate error address if ue error is detected */
+	if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, AddrV) == 1 &&
+	    (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1)) {
+
+		err_addr = ras->umc_ecc.ecc[eccinfo_table_idx].mca_umc_addr;
+		err_addr = REG_GET_FIELD(err_addr, MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr);
+
+		umc_v8_10_convert_error_address(adev, err_data, err_addr,
+					ch_inst, umc_inst, node_inst, mc_umc_status);
+	}
+
+	return 0;
+}
+
+static void umc_v8_10_ecc_info_query_ras_error_address(struct amdgpu_device *adev,
+					void *ras_error_status)
+{
+	amdgpu_umc_loop_channels(adev,
+		umc_v8_10_ecc_info_query_error_address, ras_error_status);
+}
+
+const struct amdgpu_ras_block_hw_ops umc_v8_10_ras_hw_ops = {
+	.query_ras_error_count = umc_v8_10_query_ras_error_count,
+	.query_ras_error_address = umc_v8_10_query_ras_error_address,
+};
+
+struct amdgpu_umc_ras umc_v8_10_ras = {
+	.ras_block = {
+		.hw_ops = &umc_v8_10_ras_hw_ops,
+	},
+	.err_cnt_init = umc_v8_10_err_cnt_init,
+	.query_ras_poison_mode = umc_v8_10_query_ras_poison_mode,
+	.ecc_info_query_ras_error_count = umc_v8_10_ecc_info_query_ras_error_count,
+	.ecc_info_query_ras_error_address = umc_v8_10_ecc_info_query_ras_error_address,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v8_10.h b/drivers/gpu/drm/amd/amdgpu/umc_v8_10.h
new file mode 100644
index 000000000000..dc12e0af5451
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/umc_v8_10.h
@@ -0,0 +1,75 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __UMC_V8_10_H__
+#define __UMC_V8_10_H__
+
+#include "soc15_common.h"
+#include "amdgpu.h"
+
+/* number of umc channel instance with memory map register access */
+#define UMC_V8_10_CHANNEL_INSTANCE_NUM		2
+/* number of umc instance with memory map register access */
+#define UMC_V8_10_UMC_INSTANCE_NUM		2
+
+/* Total channel instances for all available umc nodes */
+#define UMC_V8_10_TOTAL_CHANNEL_NUM(adev) \
+	(UMC_V8_10_CHANNEL_INSTANCE_NUM * UMC_V8_10_UMC_INSTANCE_NUM * \
+	(adev)->gmc.num_umc - hweight32((adev)->gmc.m_half_use) * 2)
+
+/* UMC regiser per channel offset */
+#define UMC_V8_10_PER_CHANNEL_OFFSET	0x400
+
+/* EccErrCnt max value */
+#define UMC_V8_10_CE_CNT_MAX		0xffff
+/* umc ce interrupt threshold */
+#define UUMC_V8_10_CE_INT_THRESHOLD	0xffff
+/* umc ce count initial value */
+#define UMC_V8_10_CE_CNT_INIT	(UMC_V8_10_CE_CNT_MAX - UUMC_V8_10_CE_INT_THRESHOLD)
+
+#define UMC_V8_10_NA_COL_2BITS_POWER_OF_2_NUM	 4
+
+/* The C5 bit in NA  address */
+#define UMC_V8_10_NA_C5_BIT	14
+
+/* Map to swizzle mode address */
+#define SWIZZLE_MODE_TMP_ADDR(na, ch_num, ch_idx) \
+		((((na) >> 10) * (ch_num) + (ch_idx)) << 10)
+#define SWIZZLE_MODE_ADDR_HI(addr, col_bit)  \
+		(((addr) >> ((col_bit) + 2)) << ((col_bit) + 2))
+#define SWIZZLE_MODE_ADDR_MID(na, col_bit) ((((na) >> 8) & 0x3) << (col_bit))
+#define SWIZZLE_MODE_ADDR_LOW(addr, col_bit) \
+		((((addr) >> 10) & ((0x1ULL << (col_bit - 8)) - 1)) << 8)
+#define SWIZZLE_MODE_ADDR_LSB(na) ((na) & 0xFF)
+
+extern struct amdgpu_umc_ras umc_v8_10_ras;
+extern const uint32_t
+	umc_v8_10_channel_idx_tbl[]
+				[UMC_V8_10_UMC_INSTANCE_NUM]
+				[UMC_V8_10_CHANNEL_INSTANCE_NUM];
+
+extern const uint32_t
+	umc_v8_10_channel_idx_tbl_ext0[]
+				[UMC_V8_10_UMC_INSTANCE_NUM]
+				[UMC_V8_10_CHANNEL_INSTANCE_NUM];
+#endif
+
diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v8_14.c b/drivers/gpu/drm/amd/amdgpu/umc_v8_14.c
new file mode 100644
index 000000000000..eaca10a3c4a9
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/umc_v8_14.c
@@ -0,0 +1,160 @@
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "umc_v8_14.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_umc.h"
+#include "amdgpu.h"
+#include "umc/umc_8_14_0_offset.h"
+#include "umc/umc_8_14_0_sh_mask.h"
+
+static inline uint32_t get_umc_v8_14_reg_offset(struct amdgpu_device *adev,
+					    uint32_t umc_inst,
+					    uint32_t ch_inst)
+{
+	return adev->umc.channel_offs * ch_inst + UMC_V8_14_INST_DIST * umc_inst;
+}
+
+static int umc_v8_14_clear_error_count_per_channel(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
+{
+	uint32_t ecc_err_cnt_addr;
+	uint32_t umc_reg_offset =
+		get_umc_v8_14_reg_offset(adev, umc_inst, ch_inst);
+
+	ecc_err_cnt_addr =
+		SOC15_REG_OFFSET(UMC, 0, regUMCCH0_GeccErrCnt);
+
+	/* clear error count */
+	WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4,
+			UMC_V8_14_CE_CNT_INIT);
+
+	return 0;
+}
+
+static void umc_v8_14_clear_error_count(struct amdgpu_device *adev)
+{
+	amdgpu_umc_loop_channels(adev,
+		umc_v8_14_clear_error_count_per_channel, NULL);
+}
+
+static void umc_v8_14_query_correctable_error_count(struct amdgpu_device *adev,
+						   uint32_t umc_reg_offset,
+						   unsigned long *error_count)
+{
+	uint32_t ecc_err_cnt, ecc_err_cnt_addr;
+
+	/* UMC 8_14 registers */
+	ecc_err_cnt_addr =
+		SOC15_REG_OFFSET(UMC, 0, regUMCCH0_GeccErrCnt);
+
+	ecc_err_cnt = RREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4);
+	*error_count +=
+		(REG_GET_FIELD(ecc_err_cnt, UMCCH0_GeccErrCnt, GeccErrCnt) -
+		 UMC_V8_14_CE_CNT_INIT);
+}
+
+static void umc_v8_14_query_uncorrectable_error_count(struct amdgpu_device *adev,
+						      uint32_t umc_reg_offset,
+						      unsigned long *error_count)
+{
+	uint32_t ecc_err_cnt, ecc_err_cnt_addr;
+	/* UMC 8_14 registers */
+	ecc_err_cnt_addr =
+		SOC15_REG_OFFSET(UMC, 0, regUMCCH0_GeccErrCnt);
+
+	ecc_err_cnt = RREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4);
+	*error_count +=
+		(REG_GET_FIELD(ecc_err_cnt, UMCCH0_GeccErrCnt, GeccUnCorrErrCnt) -
+		 UMC_V8_14_CE_CNT_INIT);
+}
+
+static int umc_v8_14_query_error_count_per_channel(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)data;
+	uint32_t umc_reg_offset =
+		get_umc_v8_14_reg_offset(adev, umc_inst, ch_inst);
+
+	umc_v8_14_query_correctable_error_count(adev,
+					umc_reg_offset,
+					&(err_data->ce_count));
+	umc_v8_14_query_uncorrectable_error_count(adev,
+					umc_reg_offset,
+					&(err_data->ue_count));
+
+	return 0;
+}
+
+static void umc_v8_14_query_ras_error_count(struct amdgpu_device *adev,
+					   void *ras_error_status)
+{
+	amdgpu_umc_loop_channels(adev,
+		umc_v8_14_query_error_count_per_channel, ras_error_status);
+
+	umc_v8_14_clear_error_count(adev);
+}
+
+static int umc_v8_14_err_cnt_init_per_channel(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
+{
+	uint32_t ecc_err_cnt_sel, ecc_err_cnt_sel_addr;
+	uint32_t ecc_err_cnt_addr;
+	uint32_t umc_reg_offset =
+		get_umc_v8_14_reg_offset(adev, umc_inst, ch_inst);
+
+	ecc_err_cnt_sel_addr =
+		SOC15_REG_OFFSET(UMC, 0, regUMCCH0_GeccErrCntSel);
+	ecc_err_cnt_addr =
+		SOC15_REG_OFFSET(UMC, 0, regUMCCH0_GeccErrCnt);
+
+	ecc_err_cnt_sel = RREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4);
+
+	/* set ce error interrupt type to APIC based interrupt */
+	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, UMCCH0_GeccErrCntSel,
+					GeccErrInt, 0x1);
+	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4, ecc_err_cnt_sel);
+	/* set error count to initial value */
+	WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4, UMC_V8_14_CE_CNT_INIT);
+
+	return 0;
+}
+
+static void umc_v8_14_err_cnt_init(struct amdgpu_device *adev)
+{
+	amdgpu_umc_loop_channels(adev,
+		umc_v8_14_err_cnt_init_per_channel, NULL);
+}
+
+const struct amdgpu_ras_block_hw_ops umc_v8_14_ras_hw_ops = {
+	.query_ras_error_count = umc_v8_14_query_ras_error_count,
+};
+
+struct amdgpu_umc_ras umc_v8_14_ras = {
+	.ras_block = {
+		.hw_ops = &umc_v8_14_ras_hw_ops,
+	},
+	.err_cnt_init = umc_v8_14_err_cnt_init,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v8_14.h b/drivers/gpu/drm/amd/amdgpu/umc_v8_14.h
new file mode 100644
index 000000000000..20a258f0017a
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/umc_v8_14.h
@@ -0,0 +1,51 @@
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __UMC_V8_14_H__
+#define __UMC_V8_14_H__
+
+#include "soc15_common.h"
+#include "amdgpu.h"
+
+/* number of umc channel instance with memory map register access */
+#define UMC_V8_14_CHANNEL_INSTANCE_NUM		2
+/* number of umc instance with memory map register access */
+#define UMC_V8_14_UMC_INSTANCE_NUM(adev)		((adev)->umc.node_inst_num)
+
+/* Total channel instances for all available umc nodes */
+#define UMC_V8_14_TOTAL_CHANNEL_NUM(adev) \
+	(UMC_V8_14_CHANNEL_INSTANCE_NUM * (adev)->gmc.num_umc)
+
+/* UMC register per channel offset */
+#define UMC_V8_14_PER_CHANNEL_OFFSET		0x400
+
+#define UMC_V8_14_INST_DIST		0x40000
+
+/* EccErrCnt max value */
+#define UMC_V8_14_CE_CNT_MAX		0xffff
+/* umc ce interrupt threshold */
+#define UMC_V8_14_CE_INT_THRESHOLD		0xffff
+/* umc ce count initial value */
+#define UMC_V8_14_CE_CNT_INIT	(UMC_V8_14_CE_CNT_MAX - UMC_V8_14_CE_INT_THRESHOLD)
+
+extern struct amdgpu_umc_ras umc_v8_14_ras;
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v8_7.c b/drivers/gpu/drm/amd/amdgpu/umc_v8_7.c
new file mode 100644
index 000000000000..b717fdaa46e4
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/umc_v8_7.c
@@ -0,0 +1,446 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "umc_v8_7.h"
+#include "amdgpu_ras.h"
+#include "amdgpu_umc.h"
+#include "amdgpu.h"
+
+#include "rsmu/rsmu_0_0_2_offset.h"
+#include "rsmu/rsmu_0_0_2_sh_mask.h"
+#include "umc/umc_8_7_0_offset.h"
+#include "umc/umc_8_7_0_sh_mask.h"
+
+#define UMC_8_INST_DIST			0x40000
+
+const uint32_t
+	umc_v8_7_channel_idx_tbl[UMC_V8_7_UMC_INSTANCE_NUM][UMC_V8_7_CHANNEL_INSTANCE_NUM] = {
+		{2, 11},  {4, 13},
+		{1, 8},   {7, 14},
+		{10, 3},  {12, 5},
+		{9, 0},   {15, 6}
+};
+
+static inline uint32_t get_umc_v8_7_reg_offset(struct amdgpu_device *adev,
+					    uint32_t umc_inst,
+					    uint32_t ch_inst)
+{
+	return adev->umc.channel_offs*ch_inst + UMC_8_INST_DIST*umc_inst;
+}
+
+static void umc_v8_7_ecc_info_query_correctable_error_count(struct amdgpu_device *adev,
+						uint32_t umc_inst, uint32_t ch_inst,
+						unsigned long *error_count)
+{
+	uint64_t mc_umc_status;
+	uint32_t eccinfo_table_idx;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+
+	eccinfo_table_idx = umc_inst * adev->umc.channel_inst_num + ch_inst;
+
+	/* check for SRAM correctable error
+	 * MCUMC_STATUS is a 64 bit register
+	 */
+	mc_umc_status = ras->umc_ecc.ecc[eccinfo_table_idx].mca_umc_status;
+	if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, CECC) == 1)
+		*error_count += 1;
+}
+
+static void umc_v8_7_ecc_info_querry_uncorrectable_error_count(struct amdgpu_device *adev,
+							uint32_t umc_inst, uint32_t ch_inst,
+							unsigned long *error_count)
+{
+	uint64_t mc_umc_status;
+	uint32_t eccinfo_table_idx;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+
+	eccinfo_table_idx = umc_inst * adev->umc.channel_inst_num + ch_inst;
+
+	/* check the MCUMC_STATUS */
+	mc_umc_status = ras->umc_ecc.ecc[eccinfo_table_idx].mca_umc_status;
+	if ((REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1) &&
+	    (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, PCC) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UC) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, TCC) == 1))
+		*error_count += 1;
+}
+
+static void umc_v8_7_ecc_info_query_ras_error_count(struct amdgpu_device *adev,
+					void *ras_error_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+
+	uint32_t umc_inst        = 0;
+	uint32_t ch_inst         = 0;
+
+	/* TODO: driver needs to toggle DF Cstate to ensure
+	 * safe access of UMC registers. Will add the protection
+	 */
+	LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) {
+		umc_v8_7_ecc_info_query_correctable_error_count(adev,
+							umc_inst, ch_inst,
+							&(err_data->ce_count));
+		umc_v8_7_ecc_info_querry_uncorrectable_error_count(adev,
+							umc_inst, ch_inst,
+							&(err_data->ue_count));
+	}
+}
+
+static void umc_v8_7_convert_error_address(struct amdgpu_device *adev,
+					struct ras_err_data *err_data, uint64_t err_addr,
+					uint32_t ch_inst, uint32_t umc_inst)
+{
+	uint64_t retired_page;
+	uint32_t channel_index;
+
+	channel_index =
+		adev->umc.channel_idx_tbl[umc_inst * adev->umc.channel_inst_num + ch_inst];
+
+	/* translate umc channel address to soc pa, 3 parts are included */
+	retired_page = ADDR_OF_4KB_BLOCK(err_addr) |
+			ADDR_OF_256B_BLOCK(channel_index) |
+			OFFSET_IN_256B_BLOCK(err_addr);
+
+	amdgpu_umc_fill_error_record(err_data, err_addr,
+				retired_page, channel_index, umc_inst);
+}
+
+static void umc_v8_7_ecc_info_query_error_address(struct amdgpu_device *adev,
+					struct ras_err_data *err_data,
+					uint32_t ch_inst,
+					uint32_t umc_inst)
+{
+	uint64_t mc_umc_status, err_addr;
+	uint32_t eccinfo_table_idx;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+
+	eccinfo_table_idx = umc_inst * adev->umc.channel_inst_num + ch_inst;
+	mc_umc_status = ras->umc_ecc.ecc[eccinfo_table_idx].mca_umc_status;
+
+	if (mc_umc_status == 0)
+		return;
+
+	if (!err_data->err_addr)
+		return;
+
+	/* calculate error address if ue error is detected */
+	if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1) {
+
+		err_addr = ras->umc_ecc.ecc[eccinfo_table_idx].mca_umc_addr;
+		err_addr = REG_GET_FIELD(err_addr, MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr);
+
+		umc_v8_7_convert_error_address(adev, err_data, err_addr,
+						ch_inst, umc_inst);
+	}
+}
+
+static void umc_v8_7_ecc_info_query_ras_error_address(struct amdgpu_device *adev,
+					void *ras_error_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+
+	uint32_t umc_inst        = 0;
+	uint32_t ch_inst         = 0;
+
+	/* TODO: driver needs to toggle DF Cstate to ensure
+	 * safe access of UMC resgisters. Will add the protection
+	 * when firmware interface is ready
+	 */
+	LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) {
+		umc_v8_7_ecc_info_query_error_address(adev,
+						err_data,
+						ch_inst,
+						umc_inst);
+	}
+}
+
+static void umc_v8_7_clear_error_count_per_channel(struct amdgpu_device *adev,
+					uint32_t umc_reg_offset)
+{
+	uint32_t ecc_err_cnt_addr;
+	uint32_t ecc_err_cnt_sel, ecc_err_cnt_sel_addr;
+
+	ecc_err_cnt_sel_addr =
+		SOC15_REG_OFFSET(UMC, 0, mmUMCCH0_0_GeccErrCntSel);
+	ecc_err_cnt_addr =
+		SOC15_REG_OFFSET(UMC, 0, mmUMCCH0_0_GeccErrCnt);
+
+	/* select the lower chip */
+	ecc_err_cnt_sel = RREG32_PCIE((ecc_err_cnt_sel_addr +
+					umc_reg_offset) * 4);
+	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel,
+					UMCCH0_0_GeccErrCntSel,
+					GeccErrCntCsSel, 0);
+	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4,
+			ecc_err_cnt_sel);
+
+	/* clear lower chip error count */
+	WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4,
+			UMC_V8_7_CE_CNT_INIT);
+
+	/* select the higher chip */
+	ecc_err_cnt_sel = RREG32_PCIE((ecc_err_cnt_sel_addr +
+					umc_reg_offset) * 4);
+	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel,
+					UMCCH0_0_GeccErrCntSel,
+					GeccErrCntCsSel, 1);
+	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4,
+			ecc_err_cnt_sel);
+
+	/* clear higher chip error count */
+	WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4,
+			UMC_V8_7_CE_CNT_INIT);
+}
+
+static void umc_v8_7_clear_error_count(struct amdgpu_device *adev)
+{
+	uint32_t umc_inst        = 0;
+	uint32_t ch_inst         = 0;
+	uint32_t umc_reg_offset  = 0;
+
+	LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) {
+		umc_reg_offset = get_umc_v8_7_reg_offset(adev,
+						umc_inst,
+						ch_inst);
+
+		umc_v8_7_clear_error_count_per_channel(adev,
+						umc_reg_offset);
+	}
+}
+
+static void umc_v8_7_query_correctable_error_count(struct amdgpu_device *adev,
+						   uint32_t umc_reg_offset,
+						   unsigned long *error_count)
+{
+	uint32_t ecc_err_cnt_sel, ecc_err_cnt_sel_addr;
+	uint32_t ecc_err_cnt, ecc_err_cnt_addr;
+	uint64_t mc_umc_status;
+	uint32_t mc_umc_status_addr;
+
+	/* UMC 8_7_2 registers */
+	ecc_err_cnt_sel_addr =
+		SOC15_REG_OFFSET(UMC, 0, mmUMCCH0_0_GeccErrCntSel);
+	ecc_err_cnt_addr =
+		SOC15_REG_OFFSET(UMC, 0, mmUMCCH0_0_GeccErrCnt);
+	mc_umc_status_addr =
+		SOC15_REG_OFFSET(UMC, 0, mmMCA_UMC_UMC0_MCUMC_STATUST0);
+
+	/* select the lower chip and check the error count */
+	ecc_err_cnt_sel = RREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4);
+	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, UMCCH0_0_GeccErrCntSel,
+					GeccErrCntCsSel, 0);
+	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4, ecc_err_cnt_sel);
+
+	ecc_err_cnt = RREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4);
+	*error_count +=
+		(REG_GET_FIELD(ecc_err_cnt, UMCCH0_0_GeccErrCnt, GeccErrCnt) -
+		 UMC_V8_7_CE_CNT_INIT);
+
+	/* select the higher chip and check the err counter */
+	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, UMCCH0_0_GeccErrCntSel,
+					GeccErrCntCsSel, 1);
+	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4, ecc_err_cnt_sel);
+
+	ecc_err_cnt = RREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4);
+	*error_count +=
+		(REG_GET_FIELD(ecc_err_cnt, UMCCH0_0_GeccErrCnt, GeccErrCnt) -
+		 UMC_V8_7_CE_CNT_INIT);
+
+	/* check for SRAM correctable error
+	  MCUMC_STATUS is a 64 bit register */
+	mc_umc_status = RREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4);
+	if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, ErrorCodeExt) == 6 &&
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, CECC) == 1)
+		*error_count += 1;
+}
+
+static void umc_v8_7_querry_uncorrectable_error_count(struct amdgpu_device *adev,
+						      uint32_t umc_reg_offset,
+						      unsigned long *error_count)
+{
+	uint64_t mc_umc_status;
+	uint32_t mc_umc_status_addr;
+
+	mc_umc_status_addr = SOC15_REG_OFFSET(UMC, 0, mmMCA_UMC_UMC0_MCUMC_STATUST0);
+
+	/* check the MCUMC_STATUS */
+	mc_umc_status = RREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4);
+	if ((REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1) &&
+	    (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, PCC) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UC) == 1 ||
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, TCC) == 1))
+		*error_count += 1;
+}
+
+static void umc_v8_7_query_ras_error_count(struct amdgpu_device *adev,
+					   void *ras_error_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+
+	uint32_t umc_inst        = 0;
+	uint32_t ch_inst         = 0;
+	uint32_t umc_reg_offset  = 0;
+
+	LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) {
+		umc_reg_offset = get_umc_v8_7_reg_offset(adev,
+						      umc_inst,
+						      ch_inst);
+
+		umc_v8_7_query_correctable_error_count(adev,
+						       umc_reg_offset,
+						       &(err_data->ce_count));
+		umc_v8_7_querry_uncorrectable_error_count(adev,
+							  umc_reg_offset,
+							  &(err_data->ue_count));
+	}
+
+	umc_v8_7_clear_error_count(adev);
+}
+
+static void umc_v8_7_query_error_address(struct amdgpu_device *adev,
+					 struct ras_err_data *err_data,
+					 uint32_t umc_reg_offset,
+					 uint32_t ch_inst,
+					 uint32_t umc_inst)
+{
+	uint32_t lsb, mc_umc_status_addr;
+	uint64_t mc_umc_status, err_addr, mc_umc_addrt0;
+
+	mc_umc_status_addr =
+		SOC15_REG_OFFSET(UMC, 0, mmMCA_UMC_UMC0_MCUMC_STATUST0);
+	mc_umc_addrt0 =
+		SOC15_REG_OFFSET(UMC, 0, mmMCA_UMC_UMC0_MCUMC_ADDRT0);
+	mc_umc_status = RREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4);
+
+	if (mc_umc_status == 0)
+		return;
+
+	if (!err_data->err_addr) {
+		/* clear umc status */
+		WREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4, 0x0ULL);
+		return;
+	}
+
+	/* calculate error address if ue error is detected */
+	if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
+	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1) {
+
+		err_addr = RREG64_PCIE((mc_umc_addrt0 + umc_reg_offset) * 4);
+		/* the lowest lsb bits should be ignored */
+		lsb = REG_GET_FIELD(err_addr, MCA_UMC_UMC0_MCUMC_ADDRT0, LSB);
+		err_addr = REG_GET_FIELD(err_addr, MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr);
+		err_addr &= ~((0x1ULL << lsb) - 1);
+
+		umc_v8_7_convert_error_address(adev, err_data, err_addr,
+								ch_inst, umc_inst);
+	}
+
+	/* clear umc status */
+	WREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4, 0x0ULL);
+}
+
+static void umc_v8_7_query_ras_error_address(struct amdgpu_device *adev,
+					     void *ras_error_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+
+	uint32_t umc_inst        = 0;
+	uint32_t ch_inst         = 0;
+	uint32_t umc_reg_offset  = 0;
+
+	LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) {
+		umc_reg_offset = get_umc_v8_7_reg_offset(adev,
+						      umc_inst,
+						      ch_inst);
+
+		umc_v8_7_query_error_address(adev,
+					     err_data,
+					     umc_reg_offset,
+					     ch_inst,
+					     umc_inst);
+	}
+}
+
+static void umc_v8_7_err_cnt_init_per_channel(struct amdgpu_device *adev,
+					      uint32_t umc_reg_offset)
+{
+	uint32_t ecc_err_cnt_sel, ecc_err_cnt_sel_addr;
+	uint32_t ecc_err_cnt_addr;
+
+	ecc_err_cnt_sel_addr =
+		SOC15_REG_OFFSET(UMC, 0, mmUMCCH0_0_GeccErrCntSel);
+	ecc_err_cnt_addr =
+		SOC15_REG_OFFSET(UMC, 0, mmUMCCH0_0_GeccErrCnt);
+
+	/* select the lower chip and check the error count */
+	ecc_err_cnt_sel = RREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4);
+	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, UMCCH0_0_GeccErrCntSel,
+					GeccErrCntCsSel, 0);
+	/* set ce error interrupt type to APIC based interrupt */
+	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, UMCCH0_0_GeccErrCntSel,
+					GeccErrInt, 0x1);
+	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4, ecc_err_cnt_sel);
+	/* set error count to initial value */
+	WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4, UMC_V8_7_CE_CNT_INIT);
+
+	/* select the higher chip and check the err counter */
+	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, UMCCH0_0_GeccErrCntSel,
+					GeccErrCntCsSel, 1);
+	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4, ecc_err_cnt_sel);
+	WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4, UMC_V8_7_CE_CNT_INIT);
+}
+
+static void umc_v8_7_err_cnt_init(struct amdgpu_device *adev)
+{
+	uint32_t umc_inst        = 0;
+	uint32_t ch_inst         = 0;
+	uint32_t umc_reg_offset  = 0;
+
+	LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) {
+		umc_reg_offset = get_umc_v8_7_reg_offset(adev,
+						      umc_inst,
+						      ch_inst);
+
+		umc_v8_7_err_cnt_init_per_channel(adev, umc_reg_offset);
+	}
+}
+
+const struct amdgpu_ras_block_hw_ops umc_v8_7_ras_hw_ops = {
+	.query_ras_error_count = umc_v8_7_query_ras_error_count,
+	.query_ras_error_address = umc_v8_7_query_ras_error_address,
+};
+
+struct amdgpu_umc_ras umc_v8_7_ras = {
+	.ras_block = {
+		.hw_ops = &umc_v8_7_ras_hw_ops,
+	},
+	.err_cnt_init = umc_v8_7_err_cnt_init,
+	.ecc_info_query_ras_error_count = umc_v8_7_ecc_info_query_ras_error_count,
+	.ecc_info_query_ras_error_address = umc_v8_7_ecc_info_query_ras_error_address,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v8_7.h b/drivers/gpu/drm/amd/amdgpu/umc_v8_7.h
new file mode 100644
index 000000000000..dd4993f5f78f
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/umc_v8_7.h
@@ -0,0 +1,51 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __UMC_V8_7_H__
+#define __UMC_V8_7_H__
+
+#include "soc15_common.h"
+#include "amdgpu.h"
+
+/* HBM  Memory Channel Width */
+#define UMC_V8_7_HBM_MEMORY_CHANNEL_WIDTH	128
+/* number of umc channel instance with memory map register access */
+#define UMC_V8_7_CHANNEL_INSTANCE_NUM		2
+/* number of umc instance with memory map register access */
+#define UMC_V8_7_UMC_INSTANCE_NUM		8
+/* total channel instances in one umc block */
+#define UMC_V8_7_TOTAL_CHANNEL_NUM	(UMC_V8_7_CHANNEL_INSTANCE_NUM * UMC_V8_7_UMC_INSTANCE_NUM)
+/* UMC regiser per channel offset */
+#define UMC_V8_7_PER_CHANNEL_OFFSET_SIENNA	0x400
+
+/* EccErrCnt max value */
+#define UMC_V8_7_CE_CNT_MAX		0xffff
+/* umc ce interrupt threshold */
+#define UMC_V8_7_CE_INT_THRESHOLD	0xffff
+/* umc ce count initial value */
+#define UMC_V8_7_CE_CNT_INIT	(UMC_V8_7_CE_CNT_MAX - UMC_V8_7_CE_INT_THRESHOLD)
+
+extern struct amdgpu_umc_ras umc_v8_7_ras;
+extern const uint32_t
+	umc_v8_7_channel_idx_tbl[UMC_V8_7_UMC_INSTANCE_NUM][UMC_V8_7_CHANNEL_INSTANCE_NUM];
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/umsch_mm_v4_0.c b/drivers/gpu/drm/amd/amdgpu/umsch_mm_v4_0.c
new file mode 100644
index 000000000000..ce3bb12e3572
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/umsch_mm_v4_0.c
@@ -0,0 +1,434 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include <linux/debugfs.h>
+#include "amdgpu.h"
+#include "soc15_common.h"
+#include "soc21.h"
+#include "vcn/vcn_4_0_0_offset.h"
+#include "vcn/vcn_4_0_0_sh_mask.h"
+
+#include "amdgpu_umsch_mm.h"
+#include "umsch_mm_4_0_api_def.h"
+#include "umsch_mm_v4_0.h"
+
+#define regUVD_IPX_DLDO_CONFIG                             0x0064
+#define regUVD_IPX_DLDO_CONFIG_BASE_IDX                    1
+#define regUVD_IPX_DLDO_STATUS                             0x0065
+#define regUVD_IPX_DLDO_STATUS_BASE_IDX                    1
+
+#define UVD_IPX_DLDO_CONFIG__ONO0_PWR_CONFIG__SHIFT        0x00000002
+#define UVD_IPX_DLDO_CONFIG__ONO0_PWR_CONFIG_MASK          0x0000000cUL
+#define UVD_IPX_DLDO_STATUS__ONO0_PWR_STATUS__SHIFT        0x00000001
+#define UVD_IPX_DLDO_STATUS__ONO0_PWR_STATUS_MASK          0x00000002UL
+
+static int umsch_mm_v4_0_load_microcode(struct amdgpu_umsch_mm *umsch)
+{
+	struct amdgpu_device *adev = umsch->ring.adev;
+	uint64_t data;
+	int r;
+
+	r = amdgpu_umsch_mm_allocate_ucode_buffer(umsch);
+	if (r)
+		return r;
+
+	r = amdgpu_umsch_mm_allocate_ucode_data_buffer(umsch);
+	if (r)
+		goto err_free_ucode_bo;
+
+	umsch->cmd_buf_curr_ptr = umsch->cmd_buf_ptr;
+
+	if (amdgpu_ip_version(adev, VCN_HWIP, 0) >= IP_VERSION(4, 0, 5)) {
+		WREG32_SOC15(VCN, 0, regUVD_IPX_DLDO_CONFIG,
+			1 << UVD_IPX_DLDO_CONFIG__ONO0_PWR_CONFIG__SHIFT);
+		SOC15_WAIT_ON_RREG(VCN, 0, regUVD_IPX_DLDO_STATUS,
+			0 << UVD_IPX_DLDO_STATUS__ONO0_PWR_STATUS__SHIFT,
+			UVD_IPX_DLDO_STATUS__ONO0_PWR_STATUS_MASK);
+	}
+
+	data = RREG32_SOC15(VCN, 0, regUMSCH_MES_RESET_CTRL);
+	data = REG_SET_FIELD(data, UMSCH_MES_RESET_CTRL, MES_CORE_SOFT_RESET, 0);
+	WREG32_SOC15_UMSCH(regUMSCH_MES_RESET_CTRL, data);
+
+	data = RREG32_SOC15(VCN, 0, regVCN_MES_CNTL);
+	data = REG_SET_FIELD(data, VCN_MES_CNTL, MES_INVALIDATE_ICACHE, 1);
+	data = REG_SET_FIELD(data, VCN_MES_CNTL, MES_PIPE0_RESET, 1);
+	data = REG_SET_FIELD(data, VCN_MES_CNTL, MES_PIPE0_ACTIVE, 0);
+	data = REG_SET_FIELD(data, VCN_MES_CNTL, MES_HALT, 1);
+	WREG32_SOC15_UMSCH(regVCN_MES_CNTL, data);
+
+	data = RREG32_SOC15(VCN, 0, regVCN_MES_IC_BASE_CNTL);
+	data = REG_SET_FIELD(data, VCN_MES_IC_BASE_CNTL, VMID, 0);
+	data = REG_SET_FIELD(data, VCN_MES_IC_BASE_CNTL, EXE_DISABLE, 0);
+	data = REG_SET_FIELD(data, VCN_MES_IC_BASE_CNTL, CACHE_POLICY, 0);
+	WREG32_SOC15_UMSCH(regVCN_MES_IC_BASE_CNTL, data);
+
+	WREG32_SOC15_UMSCH(regVCN_MES_INTR_ROUTINE_START,
+		lower_32_bits(adev->umsch_mm.irq_start_addr >> 2));
+	WREG32_SOC15_UMSCH(regVCN_MES_INTR_ROUTINE_START_HI,
+		upper_32_bits(adev->umsch_mm.irq_start_addr >> 2));
+
+	WREG32_SOC15_UMSCH(regVCN_MES_PRGRM_CNTR_START,
+		lower_32_bits(adev->umsch_mm.uc_start_addr >> 2));
+	WREG32_SOC15_UMSCH(regVCN_MES_PRGRM_CNTR_START_HI,
+		upper_32_bits(adev->umsch_mm.uc_start_addr >> 2));
+
+	WREG32_SOC15_UMSCH(regVCN_MES_LOCAL_INSTR_BASE_LO, 0);
+	WREG32_SOC15_UMSCH(regVCN_MES_LOCAL_INSTR_BASE_HI, 0);
+
+	data = adev->umsch_mm.uc_start_addr + adev->umsch_mm.ucode_size - 1;
+	WREG32_SOC15_UMSCH(regVCN_MES_LOCAL_INSTR_MASK_LO, lower_32_bits(data));
+	WREG32_SOC15_UMSCH(regVCN_MES_LOCAL_INSTR_MASK_HI, upper_32_bits(data));
+
+	data = adev->firmware.load_type == AMDGPU_FW_LOAD_PSP ?
+	       0 : adev->umsch_mm.ucode_fw_gpu_addr;
+	WREG32_SOC15_UMSCH(regVCN_MES_IC_BASE_LO, lower_32_bits(data));
+	WREG32_SOC15_UMSCH(regVCN_MES_IC_BASE_HI, upper_32_bits(data));
+
+	WREG32_SOC15_UMSCH(regVCN_MES_MIBOUND_LO, 0x1FFFFF);
+
+	WREG32_SOC15_UMSCH(regVCN_MES_LOCAL_BASE0_LO,
+		lower_32_bits(adev->umsch_mm.data_start_addr));
+	WREG32_SOC15_UMSCH(regVCN_MES_LOCAL_BASE0_HI,
+		upper_32_bits(adev->umsch_mm.data_start_addr));
+
+	WREG32_SOC15_UMSCH(regVCN_MES_LOCAL_MASK0_LO,
+		adev->umsch_mm.data_size - 1);
+	WREG32_SOC15_UMSCH(regVCN_MES_LOCAL_MASK0_HI, 0);
+
+	data = adev->firmware.load_type == AMDGPU_FW_LOAD_PSP ?
+	       0 : adev->umsch_mm.data_fw_gpu_addr;
+	WREG32_SOC15_UMSCH(regVCN_MES_DC_BASE_LO, lower_32_bits(data));
+	WREG32_SOC15_UMSCH(regVCN_MES_DC_BASE_HI, upper_32_bits(data));
+
+	WREG32_SOC15_UMSCH(regVCN_MES_MDBOUND_LO, 0x3FFFF);
+
+	data = RREG32_SOC15(VCN, 0, regUVD_UMSCH_FORCE);
+	data = REG_SET_FIELD(data, UVD_UMSCH_FORCE, IC_FORCE_GPUVM, 1);
+	data = REG_SET_FIELD(data, UVD_UMSCH_FORCE, DC_FORCE_GPUVM, 1);
+	WREG32_SOC15_UMSCH(regUVD_UMSCH_FORCE, data);
+
+	data = RREG32_SOC15(VCN, 0, regVCN_MES_IC_OP_CNTL);
+	data = REG_SET_FIELD(data, VCN_MES_IC_OP_CNTL, PRIME_ICACHE, 0);
+	data = REG_SET_FIELD(data, VCN_MES_IC_OP_CNTL, INVALIDATE_CACHE, 1);
+	WREG32_SOC15_UMSCH(regVCN_MES_IC_OP_CNTL, data);
+
+	data = RREG32_SOC15(VCN, 0, regVCN_MES_IC_OP_CNTL);
+	data = REG_SET_FIELD(data, VCN_MES_IC_OP_CNTL, PRIME_ICACHE, 1);
+	WREG32_SOC15_UMSCH(regVCN_MES_IC_OP_CNTL, data);
+
+	WREG32_SOC15_UMSCH(regVCN_MES_GP0_LO, 0);
+	WREG32_SOC15_UMSCH(regVCN_MES_GP0_HI, 0);
+
+#if defined(CONFIG_DEBUG_FS)
+	WREG32_SOC15_UMSCH(regVCN_MES_GP0_LO, lower_32_bits(umsch->log_gpu_addr));
+	WREG32_SOC15_UMSCH(regVCN_MES_GP0_HI, upper_32_bits(umsch->log_gpu_addr));
+#endif
+
+	WREG32_SOC15_UMSCH(regVCN_MES_GP1_LO, 0);
+	WREG32_SOC15_UMSCH(regVCN_MES_GP1_HI, 0);
+
+	data = RREG32_SOC15(VCN, 0, regVCN_MES_CNTL);
+	data = REG_SET_FIELD(data, VCN_MES_CNTL, MES_INVALIDATE_ICACHE, 0);
+	data = REG_SET_FIELD(data, VCN_MES_CNTL, MES_PIPE0_RESET, 0);
+	data = REG_SET_FIELD(data, VCN_MES_CNTL, MES_HALT, 0);
+	data = REG_SET_FIELD(data, VCN_MES_CNTL, MES_PIPE0_ACTIVE, 1);
+	WREG32_SOC15_UMSCH(regVCN_MES_CNTL, data);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP)
+		amdgpu_umsch_mm_psp_execute_cmd_buf(umsch);
+
+	r = SOC15_WAIT_ON_RREG(VCN, 0, regVCN_MES_MSTATUS_LO, 0xAAAAAAAA, 0xFFFFFFFF);
+	if (r) {
+		dev_err(adev->dev, "UMSCH FW Load: Failed, regVCN_MES_MSTATUS_LO: 0x%08x\n",
+			RREG32_SOC15(VCN, 0, regVCN_MES_MSTATUS_LO));
+		goto err_free_data_bo;
+	}
+
+	return 0;
+
+err_free_data_bo:
+	amdgpu_bo_free_kernel(&adev->umsch_mm.data_fw_obj,
+			      &adev->umsch_mm.data_fw_gpu_addr,
+			      (void **)&adev->umsch_mm.data_fw_ptr);
+err_free_ucode_bo:
+	amdgpu_bo_free_kernel(&adev->umsch_mm.ucode_fw_obj,
+			      &adev->umsch_mm.ucode_fw_gpu_addr,
+			      (void **)&adev->umsch_mm.ucode_fw_ptr);
+	return r;
+}
+
+static void umsch_mm_v4_0_aggregated_doorbell_init(struct amdgpu_umsch_mm *umsch)
+{
+	struct amdgpu_device *adev = umsch->ring.adev;
+	uint32_t data;
+
+	data = RREG32_SOC15(VCN, 0, regVCN_AGDB_CTRL0);
+	data = REG_SET_FIELD(data, VCN_AGDB_CTRL0, OFFSET,
+	       umsch->agdb_index[CONTEXT_PRIORITY_LEVEL_REALTIME]);
+	data = REG_SET_FIELD(data, VCN_AGDB_CTRL0, EN, 1);
+	WREG32_SOC15(VCN, 0, regVCN_AGDB_CTRL0, data);
+
+	data = RREG32_SOC15(VCN, 0, regVCN_AGDB_CTRL1);
+	data = REG_SET_FIELD(data, VCN_AGDB_CTRL1, OFFSET,
+	       umsch->agdb_index[CONTEXT_PRIORITY_LEVEL_FOCUS]);
+	data = REG_SET_FIELD(data, VCN_AGDB_CTRL1, EN, 1);
+	WREG32_SOC15(VCN, 0, regVCN_AGDB_CTRL1, data);
+
+	data = RREG32_SOC15(VCN, 0, regVCN_AGDB_CTRL2);
+	data = REG_SET_FIELD(data, VCN_AGDB_CTRL2, OFFSET,
+	       umsch->agdb_index[CONTEXT_PRIORITY_LEVEL_NORMAL]);
+	data = REG_SET_FIELD(data, VCN_AGDB_CTRL2, EN, 1);
+	WREG32_SOC15(VCN, 0, regVCN_AGDB_CTRL2, data);
+
+	data = RREG32_SOC15(VCN, 0, regVCN_AGDB_CTRL3);
+	data = REG_SET_FIELD(data, VCN_AGDB_CTRL3, OFFSET,
+	       umsch->agdb_index[CONTEXT_PRIORITY_LEVEL_IDLE]);
+	data = REG_SET_FIELD(data, VCN_AGDB_CTRL3, EN, 1);
+	WREG32_SOC15(VCN, 0, regVCN_AGDB_CTRL3, data);
+}
+
+static int umsch_mm_v4_0_ring_start(struct amdgpu_umsch_mm *umsch)
+{
+	struct amdgpu_ring *ring = &umsch->ring;
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t data;
+
+	data = RREG32_SOC15(VCN, 0, regVCN_UMSCH_RB_DB_CTRL);
+	data = REG_SET_FIELD(data, VCN_UMSCH_RB_DB_CTRL, OFFSET, ring->doorbell_index);
+	data = REG_SET_FIELD(data, VCN_UMSCH_RB_DB_CTRL, EN, 1);
+	WREG32_SOC15(VCN, 0, regVCN_UMSCH_RB_DB_CTRL, data);
+
+	adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
+		(adev->doorbell_index.vcn.vcn_ring0_1 << 1), 0);
+
+	WREG32_SOC15(VCN, 0, regVCN_UMSCH_RB_BASE_LO, lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15(VCN, 0, regVCN_UMSCH_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+
+	WREG32_SOC15(VCN, 0, regVCN_UMSCH_RB_SIZE, ring->ring_size);
+
+	ring->wptr = 0;
+
+	data = RREG32_SOC15(VCN, 0, regVCN_RB_ENABLE);
+	data &= ~(VCN_RB_ENABLE__AUDIO_RB_EN_MASK);
+	WREG32_SOC15(VCN, 0, regVCN_RB_ENABLE, data);
+
+	umsch_mm_v4_0_aggregated_doorbell_init(umsch);
+
+	return 0;
+}
+
+static int umsch_mm_v4_0_ring_stop(struct amdgpu_umsch_mm *umsch)
+{
+	struct amdgpu_ring *ring = &umsch->ring;
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t data;
+
+	data = RREG32_SOC15(VCN, 0, regVCN_RB_ENABLE);
+	data = REG_SET_FIELD(data, VCN_RB_ENABLE, UMSCH_RB_EN, 0);
+	WREG32_SOC15(VCN, 0, regVCN_RB_ENABLE, data);
+
+	data = RREG32_SOC15(VCN, 0, regVCN_UMSCH_RB_DB_CTRL);
+	data = REG_SET_FIELD(data, VCN_UMSCH_RB_DB_CTRL, EN, 0);
+	WREG32_SOC15(VCN, 0, regVCN_UMSCH_RB_DB_CTRL, data);
+
+	if (amdgpu_ip_version(adev, VCN_HWIP, 0) >= IP_VERSION(4, 0, 5)) {
+		WREG32_SOC15(VCN, 0, regUVD_IPX_DLDO_CONFIG,
+			2 << UVD_IPX_DLDO_CONFIG__ONO0_PWR_CONFIG__SHIFT);
+		SOC15_WAIT_ON_RREG(VCN, 0, regUVD_IPX_DLDO_STATUS,
+			1 << UVD_IPX_DLDO_STATUS__ONO0_PWR_STATUS__SHIFT,
+			UVD_IPX_DLDO_STATUS__ONO0_PWR_STATUS_MASK);
+	}
+
+	return 0;
+}
+
+static int umsch_mm_v4_0_set_hw_resources(struct amdgpu_umsch_mm *umsch)
+{
+	union UMSCHAPI__SET_HW_RESOURCES set_hw_resources = {};
+	struct amdgpu_device *adev = umsch->ring.adev;
+	int r;
+
+	set_hw_resources.header.type = UMSCH_API_TYPE_SCHEDULER;
+	set_hw_resources.header.opcode = UMSCH_API_SET_HW_RSRC;
+	set_hw_resources.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	set_hw_resources.vmid_mask_mm_vcn = umsch->vmid_mask_mm_vcn;
+	set_hw_resources.vmid_mask_mm_vpe = umsch->vmid_mask_mm_vpe;
+	set_hw_resources.collaboration_mask_vpe =
+		adev->vpe.collaborate_mode ? 0x3 : 0x0;
+	set_hw_resources.engine_mask = umsch->engine_mask;
+
+	set_hw_resources.vcn0_hqd_mask[0] = umsch->vcn0_hqd_mask;
+	set_hw_resources.vcn1_hqd_mask[0] = umsch->vcn1_hqd_mask;
+	set_hw_resources.vcn_hqd_mask[0] = umsch->vcn_hqd_mask[0];
+	set_hw_resources.vcn_hqd_mask[1] = umsch->vcn_hqd_mask[1];
+	set_hw_resources.vpe_hqd_mask[0] = umsch->vpe_hqd_mask;
+
+	set_hw_resources.g_sch_ctx_gpu_mc_ptr = umsch->sch_ctx_gpu_addr;
+
+	set_hw_resources.enable_level_process_quantum_check = 1;
+
+	memcpy(set_hw_resources.mmhub_base, adev->reg_offset[MMHUB_HWIP][0],
+	       sizeof(uint32_t) * 5);
+	set_hw_resources.mmhub_version =
+		IP_VERSION_MAJ_MIN_REV(amdgpu_ip_version(adev, MMHUB_HWIP, 0));
+
+	memcpy(set_hw_resources.osssys_base, adev->reg_offset[OSSSYS_HWIP][0],
+	       sizeof(uint32_t) * 5);
+	set_hw_resources.osssys_version =
+		IP_VERSION_MAJ_MIN_REV(amdgpu_ip_version(adev, OSSSYS_HWIP, 0));
+
+	set_hw_resources.vcn_version =
+		IP_VERSION_MAJ_MIN_REV(amdgpu_ip_version(adev, VCN_HWIP, 0));
+	set_hw_resources.vpe_version =
+		IP_VERSION_MAJ_MIN_REV(amdgpu_ip_version(adev, VPE_HWIP, 0));
+
+	set_hw_resources.api_status.api_completion_fence_addr = umsch->ring.fence_drv.gpu_addr;
+	set_hw_resources.api_status.api_completion_fence_value = ++umsch->ring.fence_drv.sync_seq;
+
+	r = amdgpu_umsch_mm_submit_pkt(umsch, &set_hw_resources.max_dwords_in_api,
+				       API_FRAME_SIZE_IN_DWORDS);
+	if (r)
+		return r;
+
+	r = amdgpu_umsch_mm_query_fence(umsch);
+	if (r) {
+		dev_err(adev->dev, "UMSCH SET_HW_RESOURCES: Failed\n");
+		return r;
+	}
+
+	return 0;
+}
+
+static int umsch_mm_v4_0_add_queue(struct amdgpu_umsch_mm *umsch,
+				   struct umsch_mm_add_queue_input *input_ptr)
+{
+	struct amdgpu_device *adev = umsch->ring.adev;
+	union UMSCHAPI__ADD_QUEUE add_queue = {};
+	int r;
+
+	add_queue.header.type = UMSCH_API_TYPE_SCHEDULER;
+	add_queue.header.opcode = UMSCH_API_ADD_QUEUE;
+	add_queue.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	add_queue.process_id = input_ptr->process_id;
+	add_queue.page_table_base_addr = input_ptr->page_table_base_addr;
+	add_queue.process_va_start = input_ptr->process_va_start;
+	add_queue.process_va_end = input_ptr->process_va_end;
+	add_queue.process_quantum = input_ptr->process_quantum;
+	add_queue.process_csa_addr = input_ptr->process_csa_addr;
+	add_queue.context_quantum = input_ptr->context_quantum;
+	add_queue.context_csa_addr = input_ptr->context_csa_addr;
+	add_queue.inprocess_context_priority = input_ptr->inprocess_context_priority;
+	add_queue.context_global_priority_level =
+		(enum UMSCH_AMD_PRIORITY_LEVEL)input_ptr->context_global_priority_level;
+	add_queue.doorbell_offset_0 = input_ptr->doorbell_offset_0;
+	add_queue.doorbell_offset_1 = input_ptr->doorbell_offset_1;
+	add_queue.affinity.u32All = input_ptr->affinity;
+	add_queue.mqd_addr = input_ptr->mqd_addr;
+	add_queue.engine_type = (enum UMSCH_ENGINE_TYPE)input_ptr->engine_type;
+	add_queue.h_context = input_ptr->h_context;
+	add_queue.h_queue = input_ptr->h_queue;
+	add_queue.vm_context_cntl = input_ptr->vm_context_cntl;
+	add_queue.is_context_suspended = input_ptr->is_context_suspended;
+	add_queue.collaboration_mode = adev->vpe.collaborate_mode ? 1 : 0;
+
+	add_queue.api_status.api_completion_fence_addr = umsch->ring.fence_drv.gpu_addr;
+	add_queue.api_status.api_completion_fence_value = ++umsch->ring.fence_drv.sync_seq;
+
+	r = amdgpu_umsch_mm_submit_pkt(umsch, &add_queue.max_dwords_in_api,
+				       API_FRAME_SIZE_IN_DWORDS);
+	if (r)
+		return r;
+
+	r = amdgpu_umsch_mm_query_fence(umsch);
+	if (r) {
+		dev_err(adev->dev, "UMSCH ADD_QUEUE: Failed\n");
+		return r;
+	}
+
+	return 0;
+}
+
+static int umsch_mm_v4_0_remove_queue(struct amdgpu_umsch_mm *umsch,
+				      struct umsch_mm_remove_queue_input *input_ptr)
+{
+	union UMSCHAPI__REMOVE_QUEUE remove_queue = {};
+	struct amdgpu_device *adev = umsch->ring.adev;
+	int r;
+
+	remove_queue.header.type = UMSCH_API_TYPE_SCHEDULER;
+	remove_queue.header.opcode = UMSCH_API_REMOVE_QUEUE;
+	remove_queue.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	remove_queue.doorbell_offset_0 = input_ptr->doorbell_offset_0;
+	remove_queue.doorbell_offset_1 = input_ptr->doorbell_offset_1;
+	remove_queue.context_csa_addr = input_ptr->context_csa_addr;
+
+	remove_queue.api_status.api_completion_fence_addr = umsch->ring.fence_drv.gpu_addr;
+	remove_queue.api_status.api_completion_fence_value = ++umsch->ring.fence_drv.sync_seq;
+
+	r = amdgpu_umsch_mm_submit_pkt(umsch, &remove_queue.max_dwords_in_api,
+				       API_FRAME_SIZE_IN_DWORDS);
+	if (r)
+		return r;
+
+	r = amdgpu_umsch_mm_query_fence(umsch);
+	if (r) {
+		dev_err(adev->dev, "UMSCH REMOVE_QUEUE: Failed\n");
+		return r;
+	}
+
+	return 0;
+}
+
+static int umsch_mm_v4_0_set_regs(struct amdgpu_umsch_mm *umsch)
+{
+	struct amdgpu_device *adev = container_of(umsch, struct amdgpu_device, umsch_mm);
+
+	umsch->rb_wptr = SOC15_REG_OFFSET(VCN, 0, regVCN_UMSCH_RB_WPTR);
+	umsch->rb_rptr = SOC15_REG_OFFSET(VCN, 0, regVCN_UMSCH_RB_RPTR);
+
+	return 0;
+}
+
+static const struct umsch_mm_funcs umsch_mm_v4_0_funcs = {
+	.set_hw_resources = umsch_mm_v4_0_set_hw_resources,
+	.add_queue = umsch_mm_v4_0_add_queue,
+	.remove_queue = umsch_mm_v4_0_remove_queue,
+	.set_regs = umsch_mm_v4_0_set_regs,
+	.init_microcode = amdgpu_umsch_mm_init_microcode,
+	.load_microcode = umsch_mm_v4_0_load_microcode,
+	.ring_init = amdgpu_umsch_mm_ring_init,
+	.ring_start = umsch_mm_v4_0_ring_start,
+	.ring_stop = umsch_mm_v4_0_ring_stop,
+};
+
+void umsch_mm_v4_0_set_funcs(struct amdgpu_umsch_mm *umsch)
+{
+	umsch->funcs = &umsch_mm_v4_0_funcs;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/umsch_mm_v4_0.h b/drivers/gpu/drm/amd/amdgpu/umsch_mm_v4_0.h
new file mode 100644
index 000000000000..06bc0fa74996
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/umsch_mm_v4_0.h
@@ -0,0 +1,30 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __UMSCH_MM_V4_0_H__
+#define __UMSCH_MM_V4_0_H__
+
+void umsch_mm_v4_0_set_funcs(struct amdgpu_umsch_mm *umsch);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v3_1.c b/drivers/gpu/drm/amd/amdgpu/uvd_v3_1.c
new file mode 100644
index 000000000000..2e79a3afc774
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/uvd_v3_1.c
@@ -0,0 +1,849 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Sonny Jiang <sonny.jiang@amd.com>
+ */
+
+#include <linux/firmware.h>
+
+#include "amdgpu.h"
+#include "amdgpu_uvd.h"
+#include "sid.h"
+
+#include "uvd/uvd_3_1_d.h"
+#include "uvd/uvd_3_1_sh_mask.h"
+
+#include "oss/oss_1_0_d.h"
+#include "oss/oss_1_0_sh_mask.h"
+
+/**
+ * uvd_v3_1_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t uvd_v3_1_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32(mmUVD_RBC_RB_RPTR);
+}
+
+/**
+ * uvd_v3_1_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t uvd_v3_1_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32(mmUVD_RBC_RB_WPTR);
+}
+
+/**
+ * uvd_v3_1_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void uvd_v3_1_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	WREG32(mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr));
+}
+
+/**
+ * uvd_v3_1_ring_emit_ib - execute indirect buffer
+ *
+ * @ring: amdgpu_ring pointer
+ * @job: iob associated with the indirect buffer
+ * @ib: indirect buffer to execute
+ * @flags: flags associated with the indirect buffer
+ *
+ * Write ring commands to execute the indirect buffer
+ */
+static void uvd_v3_1_ring_emit_ib(struct amdgpu_ring *ring,
+				  struct amdgpu_job *job,
+				  struct amdgpu_ib *ib,
+				  uint32_t flags)
+{
+	amdgpu_ring_write(ring, PACKET0(mmUVD_RBC_IB_BASE, 0));
+	amdgpu_ring_write(ring, ib->gpu_addr);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_RBC_IB_SIZE, 0));
+	amdgpu_ring_write(ring, ib->length_dw);
+}
+
+/**
+ * uvd_v3_1_ring_emit_fence - emit a fence & trap command
+ *
+ * @ring: amdgpu_ring pointer
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
+ *
+ * Write a fence and a trap command to the ring.
+ */
+static void uvd_v3_1_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				 unsigned flags)
+{
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_CONTEXT_ID, 0));
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
+	amdgpu_ring_write(ring, addr & 0xffffffff);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xff);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
+	amdgpu_ring_write(ring, 2);
+}
+
+/**
+ * uvd_v3_1_ring_test_ring - register write test
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Test if we can successfully write to the context register
+ */
+static int uvd_v3_1_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	WREG32(mmUVD_CONTEXT_ID, 0xCAFEDEAD);
+	r = amdgpu_ring_alloc(ring, 3);
+	if (r)
+		return r;
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_CONTEXT_ID, 0));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(mmUVD_CONTEXT_ID);
+		if (tmp == 0xDEADBEEF)
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+
+	return r;
+}
+
+static void uvd_v3_1_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	int i;
+
+	WARN_ON(ring->wptr % 2 || count % 2);
+
+	for (i = 0; i < count / 2; i++) {
+		amdgpu_ring_write(ring, PACKET0(mmUVD_NO_OP, 0));
+		amdgpu_ring_write(ring, 0);
+	}
+}
+
+static const struct amdgpu_ring_funcs uvd_v3_1_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_UVD,
+	.align_mask = 0xf,
+	.support_64bit_ptrs = false,
+	.no_user_fence = true,
+	.get_rptr = uvd_v3_1_ring_get_rptr,
+	.get_wptr = uvd_v3_1_ring_get_wptr,
+	.set_wptr = uvd_v3_1_ring_set_wptr,
+	.parse_cs = amdgpu_uvd_ring_parse_cs,
+	.emit_frame_size =
+		14, /* uvd_v3_1_ring_emit_fence  x1 no user fence */
+	.emit_ib_size = 4, /* uvd_v3_1_ring_emit_ib */
+	.emit_ib = uvd_v3_1_ring_emit_ib,
+	.emit_fence = uvd_v3_1_ring_emit_fence,
+	.test_ring = uvd_v3_1_ring_test_ring,
+	.test_ib = amdgpu_uvd_ring_test_ib,
+	.insert_nop = uvd_v3_1_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_uvd_ring_begin_use,
+	.end_use = amdgpu_uvd_ring_end_use,
+};
+
+static void uvd_v3_1_set_ring_funcs(struct amdgpu_device *adev)
+{
+	adev->uvd.inst->ring.funcs = &uvd_v3_1_ring_funcs;
+}
+
+static void uvd_v3_1_set_dcm(struct amdgpu_device *adev,
+							 bool sw_mode)
+{
+	u32 tmp, tmp2;
+
+	WREG32_FIELD(UVD_CGC_GATE, REGS, 0);
+
+	tmp = RREG32(mmUVD_CGC_CTRL);
+	tmp &= ~(UVD_CGC_CTRL__CLK_OFF_DELAY_MASK | UVD_CGC_CTRL__CLK_GATE_DLY_TIMER_MASK);
+	tmp |= UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK |
+		(1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT) |
+		(4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT);
+
+	if (sw_mode) {
+		tmp &= ~0x7ffff800;
+		tmp2 = UVD_CGC_CTRL2__DYN_OCLK_RAMP_EN_MASK |
+			UVD_CGC_CTRL2__DYN_RCLK_RAMP_EN_MASK |
+			(7 << UVD_CGC_CTRL2__GATER_DIV_ID__SHIFT);
+	} else {
+		tmp |= 0x7ffff800;
+		tmp2 = 0;
+	}
+
+	WREG32(mmUVD_CGC_CTRL, tmp);
+	WREG32_UVD_CTX(ixUVD_CGC_CTRL2, tmp2);
+}
+
+/**
+ * uvd_v3_1_mc_resume - memory controller programming
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Let the UVD memory controller know it's offsets
+ */
+static void uvd_v3_1_mc_resume(struct amdgpu_device *adev)
+{
+	uint64_t addr;
+	uint32_t size;
+
+	/* program the VCPU memory controller bits 0-27 */
+	addr = (adev->uvd.inst->gpu_addr + AMDGPU_UVD_FIRMWARE_OFFSET) >> 3;
+	size = AMDGPU_UVD_FIRMWARE_SIZE(adev) >> 3;
+	WREG32(mmUVD_VCPU_CACHE_OFFSET0, addr);
+	WREG32(mmUVD_VCPU_CACHE_SIZE0, size);
+
+	addr += size;
+	size = AMDGPU_UVD_HEAP_SIZE >> 3;
+	WREG32(mmUVD_VCPU_CACHE_OFFSET1, addr);
+	WREG32(mmUVD_VCPU_CACHE_SIZE1, size);
+
+	addr += size;
+	size = (AMDGPU_UVD_STACK_SIZE +
+		(AMDGPU_UVD_SESSION_SIZE * adev->uvd.max_handles)) >> 3;
+	WREG32(mmUVD_VCPU_CACHE_OFFSET2, addr);
+	WREG32(mmUVD_VCPU_CACHE_SIZE2, size);
+
+	/* bits 28-31 */
+	addr = (adev->uvd.inst->gpu_addr >> 28) & 0xF;
+	WREG32(mmUVD_LMI_ADDR_EXT, (addr << 12) | (addr << 0));
+
+	/* bits 32-39 */
+	addr = (adev->uvd.inst->gpu_addr >> 32) & 0xFF;
+	WREG32(mmUVD_LMI_EXT40_ADDR, addr | (0x9 << 16) | (0x1 << 31));
+
+	WREG32(mmUVD_UDEC_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+	WREG32(mmUVD_UDEC_DB_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+	WREG32(mmUVD_UDEC_DBW_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+}
+
+/**
+ * uvd_v3_1_fw_validate - FW validation operation
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialate and check UVD validation.
+ */
+static int uvd_v3_1_fw_validate(struct amdgpu_device *adev)
+{
+	int i;
+	uint32_t keysel = adev->uvd.keyselect;
+
+	WREG32(mmUVD_FW_START, keysel);
+
+	for (i = 0; i < 10; ++i) {
+		mdelay(10);
+		if (RREG32(mmUVD_FW_STATUS) & UVD_FW_STATUS__DONE_MASK)
+			break;
+	}
+
+	if (i == 10)
+		return -ETIMEDOUT;
+
+	if (!(RREG32(mmUVD_FW_STATUS) & UVD_FW_STATUS__PASS_MASK))
+		return -EINVAL;
+
+	for (i = 0; i < 10; ++i) {
+		mdelay(10);
+		if (!(RREG32(mmUVD_FW_STATUS) & UVD_FW_STATUS__BUSY_MASK))
+			break;
+	}
+
+	if (i == 10)
+		return -ETIMEDOUT;
+
+	return 0;
+}
+
+/**
+ * uvd_v3_1_start - start UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the UVD block
+ */
+static int uvd_v3_1_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = &adev->uvd.inst->ring;
+	uint32_t rb_bufsz;
+	int i, j, r;
+	u32 tmp;
+	/* disable byte swapping */
+	u32 lmi_swap_cntl = 0;
+	u32 mp_swap_cntl = 0;
+
+	/* set uvd busy */
+	WREG32_P(mmUVD_STATUS, 1<<2, ~(1<<2));
+
+	uvd_v3_1_set_dcm(adev, true);
+	WREG32(mmUVD_CGC_GATE, 0);
+
+	/* take UVD block out of reset */
+	WREG32_P(mmSRBM_SOFT_RESET, 0, ~SRBM_SOFT_RESET__SOFT_RESET_UVD_MASK);
+	mdelay(5);
+
+	/* enable VCPU clock */
+	WREG32(mmUVD_VCPU_CNTL,  1 << 9);
+
+	/* disable interrupt */
+	WREG32_P(mmUVD_MASTINT_EN, 0, ~(1 << 1));
+
+#ifdef __BIG_ENDIAN
+	/* swap (8 in 32) RB and IB */
+	lmi_swap_cntl = 0xa;
+	mp_swap_cntl = 0;
+#endif
+	WREG32(mmUVD_LMI_SWAP_CNTL, lmi_swap_cntl);
+	WREG32(mmUVD_MP_SWAP_CNTL, mp_swap_cntl);
+
+	/* initialize UVD memory controller */
+	WREG32(mmUVD_LMI_CTRL, 0x40 | (1 << 8) | (1 << 13) |
+		(1 << 21) | (1 << 9) | (1 << 20));
+
+	tmp = RREG32(mmUVD_MPC_CNTL);
+	WREG32(mmUVD_MPC_CNTL, tmp | 0x10);
+
+	WREG32(mmUVD_MPC_SET_MUXA0, 0x40c2040);
+	WREG32(mmUVD_MPC_SET_MUXA1, 0x0);
+	WREG32(mmUVD_MPC_SET_MUXB0, 0x40c2040);
+	WREG32(mmUVD_MPC_SET_MUXB1, 0x0);
+	WREG32(mmUVD_MPC_SET_ALU, 0);
+	WREG32(mmUVD_MPC_SET_MUX, 0x88);
+
+	tmp = RREG32_UVD_CTX(ixUVD_LMI_CACHE_CTRL);
+	WREG32_UVD_CTX(ixUVD_LMI_CACHE_CTRL, tmp & (~0x10));
+
+	/* enable UMC */
+	WREG32_P(mmUVD_LMI_CTRL2, 0, ~(1 << 8));
+
+	WREG32_P(mmUVD_SOFT_RESET, 0, ~UVD_SOFT_RESET__LMI_SOFT_RESET_MASK);
+
+	WREG32_P(mmUVD_SOFT_RESET, 0, ~UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK);
+
+	WREG32_P(mmUVD_SOFT_RESET, 0, ~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+
+	mdelay(10);
+
+	for (i = 0; i < 10; ++i) {
+		uint32_t status;
+		for (j = 0; j < 100; ++j) {
+			status = RREG32(mmUVD_STATUS);
+			if (status & 2)
+				break;
+			mdelay(10);
+		}
+		r = 0;
+		if (status & 2)
+			break;
+
+		DRM_ERROR("UVD not responding, trying to reset the VCPU!!!\n");
+		WREG32_P(mmUVD_SOFT_RESET, UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK,
+				 ~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+		mdelay(10);
+		WREG32_P(mmUVD_SOFT_RESET, 0, ~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+		mdelay(10);
+		r = -1;
+	}
+
+	if (r) {
+		DRM_ERROR("UVD not responding, giving up!!!\n");
+		return r;
+	}
+
+	/* enable interrupt */
+	WREG32_P(mmUVD_MASTINT_EN, 3<<1, ~(3 << 1));
+
+	WREG32_P(mmUVD_STATUS, 0, ~(1<<2));
+
+	/* force RBC into idle state */
+	WREG32(mmUVD_RBC_RB_CNTL, 0x11010101);
+
+	/* Set the write pointer delay */
+	WREG32(mmUVD_RBC_RB_WPTR_CNTL, 0);
+
+	/* Program the 4GB memory segment for rptr and ring buffer */
+	WREG32(mmUVD_LMI_EXT40_ADDR, upper_32_bits(ring->gpu_addr) |
+		   (0x7 << 16) | (0x1 << 31));
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32(mmUVD_RBC_RB_RPTR, 0x0);
+
+	ring->wptr = RREG32(mmUVD_RBC_RB_RPTR);
+	WREG32(mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr));
+
+	/* set the ring address */
+	WREG32(mmUVD_RBC_RB_BASE, ring->gpu_addr);
+
+	/* Set ring buffer size */
+	rb_bufsz = order_base_2(ring->ring_size);
+	rb_bufsz = (0x1 << 8) | rb_bufsz;
+	WREG32_P(mmUVD_RBC_RB_CNTL, rb_bufsz, ~0x11f1f);
+
+	return 0;
+}
+
+/**
+ * uvd_v3_1_stop - stop UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * stop the UVD block
+ */
+static void uvd_v3_1_stop(struct amdgpu_device *adev)
+{
+	uint32_t i, j;
+	uint32_t status;
+
+	WREG32(mmUVD_RBC_RB_CNTL, 0x11010101);
+
+	for (i = 0; i < 10; ++i) {
+		for (j = 0; j < 100; ++j) {
+			status = RREG32(mmUVD_STATUS);
+			if (status & 2)
+				break;
+			mdelay(1);
+		}
+		if (status & 2)
+			break;
+	}
+
+	for (i = 0; i < 10; ++i) {
+		for (j = 0; j < 100; ++j) {
+			status = RREG32(mmUVD_LMI_STATUS);
+			if (status & 0xf)
+				break;
+			mdelay(1);
+		}
+		if (status & 0xf)
+			break;
+	}
+
+	/* Stall UMC and register bus before resetting VCPU */
+	WREG32_P(mmUVD_LMI_CTRL2, 1 << 8, ~(1 << 8));
+
+	for (i = 0; i < 10; ++i) {
+		for (j = 0; j < 100; ++j) {
+			status = RREG32(mmUVD_LMI_STATUS);
+			if (status & 0x240)
+				break;
+			mdelay(1);
+		}
+		if (status & 0x240)
+			break;
+	}
+
+	WREG32_P(0x3D49, 0, ~(1 << 2));
+
+	WREG32_P(mmUVD_VCPU_CNTL, 0, ~(1 << 9));
+
+	/* put LMI, VCPU, RBC etc... into reset */
+	WREG32(mmUVD_SOFT_RESET, UVD_SOFT_RESET__LMI_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK);
+
+	WREG32(mmUVD_STATUS, 0);
+
+	uvd_v3_1_set_dcm(adev, false);
+}
+
+static int uvd_v3_1_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+static int uvd_v3_1_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	DRM_DEBUG("IH: UVD TRAP\n");
+	amdgpu_fence_process(&adev->uvd.inst->ring);
+	return 0;
+}
+
+
+static const struct amdgpu_irq_src_funcs uvd_v3_1_irq_funcs = {
+	.set = uvd_v3_1_set_interrupt_state,
+	.process = uvd_v3_1_process_interrupt,
+};
+
+static void uvd_v3_1_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->uvd.inst->irq.num_types = 1;
+	adev->uvd.inst->irq.funcs = &uvd_v3_1_irq_funcs;
+}
+
+
+static int uvd_v3_1_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	adev->uvd.num_uvd_inst = 1;
+
+	uvd_v3_1_set_ring_funcs(adev);
+	uvd_v3_1_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int uvd_v3_1_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_ring *ring;
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+	void *ptr;
+	uint32_t ucode_len;
+
+	/* UVD TRAP */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 124, &adev->uvd.inst->irq);
+	if (r)
+		return r;
+
+	r = amdgpu_uvd_sw_init(adev);
+	if (r)
+		return r;
+
+	ring = &adev->uvd.inst->ring;
+	sprintf(ring->name, "uvd");
+	r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst->irq, 0,
+			 AMDGPU_RING_PRIO_DEFAULT, NULL);
+	if (r)
+		return r;
+
+	r = amdgpu_uvd_resume(adev);
+	if (r)
+		return r;
+
+	/* Retrieval firmware validate key */
+	ptr = adev->uvd.inst[0].cpu_addr;
+	ptr += 192 + 16;
+	memcpy(&ucode_len, ptr, 4);
+	ptr += ucode_len;
+	memcpy(&adev->uvd.keyselect, ptr, 4);
+
+	return r;
+}
+
+static int uvd_v3_1_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	r = amdgpu_uvd_suspend(adev);
+	if (r)
+		return r;
+
+	return amdgpu_uvd_sw_fini(adev);
+}
+
+static void uvd_v3_1_enable_mgcg(struct amdgpu_device *adev,
+				 bool enable)
+{
+	u32 orig, data;
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_UVD_MGCG)) {
+		data = RREG32_UVD_CTX(ixUVD_CGC_MEM_CTRL);
+		data |= 0x3fff;
+		WREG32_UVD_CTX(ixUVD_CGC_MEM_CTRL, data);
+
+		orig = data = RREG32(mmUVD_CGC_CTRL);
+		data |= UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK;
+		if (orig != data)
+			WREG32(mmUVD_CGC_CTRL, data);
+	} else {
+		data = RREG32_UVD_CTX(ixUVD_CGC_MEM_CTRL);
+		data &= ~0x3fff;
+		WREG32_UVD_CTX(ixUVD_CGC_MEM_CTRL, data);
+
+		orig = data = RREG32(mmUVD_CGC_CTRL);
+		data &= ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK;
+		if (orig != data)
+			WREG32(mmUVD_CGC_CTRL, data);
+	}
+}
+
+/**
+ * uvd_v3_1_hw_init - start and test UVD block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Initialize the hardware, boot up the VCPU and do some testing.
+ *
+ * On SI, the UVD is meant to be used in a specific power state,
+ * or alternatively the driver can manually enable its clock.
+ * In amdgpu we use the dedicated UVD power state when DPM is enabled.
+ * Calling amdgpu_dpm_enable_uvd makes DPM select the UVD power state
+ * for the SMU and afterwards enables the UVD clock.
+ * This is automatically done by amdgpu_uvd_ring_begin_use when work
+ * is submitted to the UVD ring. Here, we have to call it manually
+ * in order to power up UVD before firmware validation.
+ *
+ * Note that we must not disable the UVD clock here, as that would
+ * cause the ring test to fail. However, UVD is powered off
+ * automatically after the ring test: amdgpu_uvd_ring_end_use calls
+ * the UVD idle work handler which will disable the UVD clock when
+ * all fences are signalled.
+ */
+static int uvd_v3_1_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring = &adev->uvd.inst->ring;
+	uint32_t tmp;
+	int r;
+
+	uvd_v3_1_mc_resume(adev);
+	uvd_v3_1_enable_mgcg(adev, true);
+
+	/* Make sure UVD is powered during FW validation.
+	 * It's going to be automatically powered off after the ring test.
+	 */
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_uvd(adev, true);
+	else
+		amdgpu_asic_set_uvd_clocks(adev, 53300, 40000);
+
+	r = uvd_v3_1_fw_validate(adev);
+	if (r) {
+		DRM_ERROR("amdgpu: UVD Firmware validate fail (%d).\n", r);
+		return r;
+	}
+
+	uvd_v3_1_start(adev);
+
+	r = amdgpu_ring_test_helper(ring);
+	if (r) {
+		DRM_ERROR("amdgpu: UVD ring test fail (%d).\n", r);
+		goto done;
+	}
+
+	r = amdgpu_ring_alloc(ring, 10);
+	if (r) {
+		DRM_ERROR("amdgpu: ring failed to lock UVD ring (%d).\n", r);
+		goto done;
+	}
+
+	tmp = PACKET0(mmUVD_SEMA_WAIT_FAULT_TIMEOUT_CNTL, 0);
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0xFFFFF);
+
+	tmp = PACKET0(mmUVD_SEMA_WAIT_INCOMPLETE_TIMEOUT_CNTL, 0);
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0xFFFFF);
+
+	tmp = PACKET0(mmUVD_SEMA_SIGNAL_INCOMPLETE_TIMEOUT_CNTL, 0);
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0xFFFFF);
+
+	/* Clear timeout status bits */
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_TIMEOUT_STATUS, 0));
+	amdgpu_ring_write(ring, 0x8);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_CNTL, 0));
+	amdgpu_ring_write(ring, 3);
+
+	amdgpu_ring_commit(ring);
+
+done:
+	if (!r)
+		DRM_INFO("UVD initialized successfully.\n");
+
+	return r;
+}
+
+/**
+ * uvd_v3_1_hw_fini - stop the hardware block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Stop the UVD block, mark ring as not ready any more
+ */
+static int uvd_v3_1_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	cancel_delayed_work_sync(&adev->uvd.idle_work);
+
+	if (RREG32(mmUVD_STATUS) != 0)
+		uvd_v3_1_stop(adev);
+
+	return 0;
+}
+
+static int uvd_v3_1_prepare_suspend(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	return amdgpu_uvd_prepare_suspend(adev);
+}
+
+static int uvd_v3_1_suspend(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	/*
+	 * Proper cleanups before halting the HW engine:
+	 *   - cancel the delayed idle work
+	 *   - enable powergating
+	 *   - enable clockgating
+	 *   - disable dpm
+	 *
+	 * TODO: to align with the VCN implementation, move the
+	 * jobs for clockgating/powergating/dpm setting to
+	 * ->set_powergating_state().
+	 */
+	cancel_delayed_work_sync(&adev->uvd.idle_work);
+
+	if (adev->pm.dpm_enabled) {
+		amdgpu_dpm_enable_uvd(adev, false);
+	} else {
+		amdgpu_asic_set_uvd_clocks(adev, 0, 0);
+		/* shutdown the UVD block */
+		amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+						       AMD_PG_STATE_GATE);
+		amdgpu_device_ip_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+						       AMD_CG_STATE_GATE);
+	}
+
+	r = uvd_v3_1_hw_fini(ip_block);
+	if (r)
+		return r;
+
+	return amdgpu_uvd_suspend(adev);
+}
+
+static int uvd_v3_1_resume(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	r = amdgpu_uvd_resume(ip_block->adev);
+	if (r)
+		return r;
+
+	return uvd_v3_1_hw_init(ip_block);
+}
+
+static bool uvd_v3_1_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	return !(RREG32(mmSRBM_STATUS) & SRBM_STATUS__UVD_BUSY_MASK);
+}
+
+static int uvd_v3_1_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	unsigned i;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (!(RREG32(mmSRBM_STATUS) & SRBM_STATUS__UVD_BUSY_MASK))
+			return 0;
+	}
+	return -ETIMEDOUT;
+}
+
+static int uvd_v3_1_soft_reset(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	uvd_v3_1_stop(adev);
+
+	WREG32_P(mmSRBM_SOFT_RESET, SRBM_SOFT_RESET__SOFT_RESET_UVD_MASK,
+			 ~SRBM_SOFT_RESET__SOFT_RESET_UVD_MASK);
+	mdelay(5);
+
+	return uvd_v3_1_start(adev);
+}
+
+static int uvd_v3_1_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int uvd_v3_1_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static const struct amd_ip_funcs uvd_v3_1_ip_funcs = {
+	.name = "uvd_v3_1",
+	.early_init = uvd_v3_1_early_init,
+	.sw_init = uvd_v3_1_sw_init,
+	.sw_fini = uvd_v3_1_sw_fini,
+	.hw_init = uvd_v3_1_hw_init,
+	.hw_fini = uvd_v3_1_hw_fini,
+	.prepare_suspend = uvd_v3_1_prepare_suspend,
+	.suspend = uvd_v3_1_suspend,
+	.resume = uvd_v3_1_resume,
+	.is_idle = uvd_v3_1_is_idle,
+	.wait_for_idle = uvd_v3_1_wait_for_idle,
+	.soft_reset = uvd_v3_1_soft_reset,
+	.set_clockgating_state = uvd_v3_1_set_clockgating_state,
+	.set_powergating_state = uvd_v3_1_set_powergating_state,
+};
+
+const struct amdgpu_ip_block_version uvd_v3_1_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_UVD,
+	.major = 3,
+	.minor = 1,
+	.rev = 0,
+	.funcs = &uvd_v3_1_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v3_1.h b/drivers/gpu/drm/amd/amdgpu/uvd_v3_1.h
new file mode 100644
index 000000000000..8c2f9b207574
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/uvd_v3_1.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __UVD_V3_1_H__
+#define __UVD_V3_1_H__
+
+extern const struct amdgpu_ip_block_version uvd_v3_1_ip_block;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c b/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c
index 8ab0f78794a5..4b96fd583772 100644
--- a/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c
+++ b/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c
@@ -23,7 +23,7 @@
  */
 
 #include <linux/firmware.h>
-#include <drm/drmP.h>
+
 #include "amdgpu.h"
 #include "amdgpu_uvd.h"
 #include "cikd.h"
@@ -44,7 +44,7 @@ static void uvd_v4_2_set_ring_funcs(struct amdgpu_device *adev);
 static void uvd_v4_2_set_irq_funcs(struct amdgpu_device *adev);
 static int uvd_v4_2_start(struct amdgpu_device *adev);
 static void uvd_v4_2_stop(struct amdgpu_device *adev);
-static int uvd_v4_2_set_clockgating_state(void *handle,
+static int uvd_v4_2_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 				enum amd_clockgating_state state);
 static void uvd_v4_2_set_dcm(struct amdgpu_device *adev,
 			     bool sw_mode);
@@ -90,9 +90,10 @@ static void uvd_v4_2_ring_set_wptr(struct amdgpu_ring *ring)
 	WREG32(mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr));
 }
 
-static int uvd_v4_2_early_init(void *handle)
+static int uvd_v4_2_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+	adev->uvd.num_uvd_inst = 1;
 
 	uvd_v4_2_set_ring_funcs(adev);
 	uvd_v4_2_set_irq_funcs(adev);
@@ -100,14 +101,14 @@ static int uvd_v4_2_early_init(void *handle)
 	return 0;
 }
 
-static int uvd_v4_2_sw_init(void *handle)
+static int uvd_v4_2_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_ring *ring;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int r;
 
 	/* UVD TRAP */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 124, &adev->uvd.irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 124, &adev->uvd.inst->irq);
 	if (r)
 		return r;
 
@@ -115,21 +116,24 @@ static int uvd_v4_2_sw_init(void *handle)
 	if (r)
 		return r;
 
-	r = amdgpu_uvd_resume(adev);
+	ring = &adev->uvd.inst->ring;
+	sprintf(ring->name, "uvd");
+	r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst->irq, 0,
+			     AMDGPU_RING_PRIO_DEFAULT, NULL);
 	if (r)
 		return r;
 
-	ring = &adev->uvd.ring;
-	sprintf(ring->name, "uvd");
-	r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.irq, 0);
+	r = amdgpu_uvd_resume(adev);
+	if (r)
+		return r;
 
 	return r;
 }
 
-static int uvd_v4_2_sw_fini(void *handle)
+static int uvd_v4_2_sw_fini(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	r = amdgpu_uvd_suspend(adev);
 	if (r)
@@ -143,26 +147,23 @@ static void uvd_v4_2_enable_mgcg(struct amdgpu_device *adev,
 /**
  * uvd_v4_2_hw_init - start and test UVD block
  *
- * @adev: amdgpu_device pointer
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
  *
  * Initialize the hardware, boot up the VCPU and do some testing
  */
-static int uvd_v4_2_hw_init(void *handle)
+static int uvd_v4_2_hw_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	struct amdgpu_ring *ring = &adev->uvd.ring;
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring = &adev->uvd.inst->ring;
 	uint32_t tmp;
 	int r;
 
 	uvd_v4_2_enable_mgcg(adev, true);
 	amdgpu_asic_set_uvd_clocks(adev, 10000, 10000);
 
-	ring->ready = true;
-	r = amdgpu_ring_test_ring(ring);
-	if (r) {
-		ring->ready = false;
+	r = amdgpu_ring_test_helper(ring);
+	if (r)
 		goto done;
-	}
 
 	r = amdgpu_ring_alloc(ring, 10);
 	if (r) {
@@ -201,45 +202,74 @@ done:
 /**
  * uvd_v4_2_hw_fini - stop the hardware block
  *
- * @adev: amdgpu_device pointer
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
  *
  * Stop the UVD block, mark ring as not ready any more
  */
-static int uvd_v4_2_hw_fini(void *handle)
+static int uvd_v4_2_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	struct amdgpu_ring *ring = &adev->uvd.ring;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	cancel_delayed_work_sync(&adev->uvd.idle_work);
 
 	if (RREG32(mmUVD_STATUS) != 0)
 		uvd_v4_2_stop(adev);
 
-	ring->ready = false;
-
 	return 0;
 }
 
-static int uvd_v4_2_suspend(void *handle)
+static int uvd_v4_2_prepare_suspend(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	return amdgpu_uvd_prepare_suspend(adev);
+}
+
+static int uvd_v4_2_suspend(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	/*
+	 * Proper cleanups before halting the HW engine:
+	 *   - cancel the delayed idle work
+	 *   - enable powergating
+	 *   - enable clockgating
+	 *   - disable dpm
+	 *
+	 * TODO: to align with the VCN implementation, move the
+	 * jobs for clockgating/powergating/dpm setting to
+	 * ->set_powergating_state().
+	 */
+	cancel_delayed_work_sync(&adev->uvd.idle_work);
 
-	r = uvd_v4_2_hw_fini(adev);
+	if (adev->pm.dpm_enabled) {
+		amdgpu_dpm_enable_uvd(adev, false);
+	} else {
+		amdgpu_asic_set_uvd_clocks(adev, 0, 0);
+		/* shutdown the UVD block */
+		amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+						       AMD_PG_STATE_GATE);
+		amdgpu_device_ip_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+						       AMD_CG_STATE_GATE);
+	}
+
+	r = uvd_v4_2_hw_fini(ip_block);
 	if (r)
 		return r;
 
 	return amdgpu_uvd_suspend(adev);
 }
 
-static int uvd_v4_2_resume(void *handle)
+static int uvd_v4_2_resume(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	r = amdgpu_uvd_resume(adev);
+	r = amdgpu_uvd_resume(ip_block->adev);
 	if (r)
 		return r;
 
-	return uvd_v4_2_hw_init(adev);
+	return uvd_v4_2_hw_init(ip_block);
 }
 
 /**
@@ -251,7 +281,7 @@ static int uvd_v4_2_resume(void *handle)
  */
 static int uvd_v4_2_start(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *ring = &adev->uvd.ring;
+	struct amdgpu_ring *ring = &adev->uvd.inst->ring;
 	uint32_t rb_bufsz;
 	int i, j, r;
 	u32 tmp;
@@ -272,7 +302,7 @@ static int uvd_v4_2_start(struct amdgpu_device *adev)
 	/* enable VCPU clock */
 	WREG32(mmUVD_VCPU_CNTL,  1 << 9);
 
-	/* disable interupt */
+	/* disable interrupt */
 	WREG32_P(mmUVD_MASTINT_EN, 0, ~(1 << 1));
 
 #ifdef __BIG_ENDIAN
@@ -282,6 +312,7 @@ static int uvd_v4_2_start(struct amdgpu_device *adev)
 #endif
 	WREG32(mmUVD_LMI_SWAP_CNTL, lmi_swap_cntl);
 	WREG32(mmUVD_MP_SWAP_CNTL, mp_swap_cntl);
+
 	/* initialize UVD memory controller */
 	WREG32(mmUVD_LMI_CTRL, 0x203108);
 
@@ -348,7 +379,7 @@ static int uvd_v4_2_start(struct amdgpu_device *adev)
 	/* Set the write pointer delay */
 	WREG32(mmUVD_RBC_RB_WPTR_CNTL, 0);
 
-	/* programm the 4GB memory segment for rptr and ring buffer */
+	/* program the 4GB memory segment for rptr and ring buffer */
 	WREG32(mmUVD_LMI_EXT40_ADDR, upper_32_bits(ring->gpu_addr) |
 				   (0x7 << 16) | (0x1 << 31));
 
@@ -437,7 +468,9 @@ static void uvd_v4_2_stop(struct amdgpu_device *adev)
  * uvd_v4_2_ring_emit_fence - emit an fence & trap command
  *
  * @ring: amdgpu_ring pointer
- * @fence: fence to emit
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
  *
  * Write a fence and a trap command to the ring.
  */
@@ -464,32 +497,6 @@ static void uvd_v4_2_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq
 }
 
 /**
- * uvd_v4_2_ring_emit_hdp_flush - emit an hdp flush
- *
- * @ring: amdgpu_ring pointer
- *
- * Emits an hdp flush.
- */
-static void uvd_v4_2_ring_emit_hdp_flush(struct amdgpu_ring *ring)
-{
-	amdgpu_ring_write(ring, PACKET0(mmHDP_MEM_COHERENCY_FLUSH_CNTL, 0));
-	amdgpu_ring_write(ring, 0);
-}
-
-/**
- * uvd_v4_2_ring_hdp_invalidate - emit an hdp invalidate
- *
- * @ring: amdgpu_ring pointer
- *
- * Emits an hdp invalidate.
- */
-static void uvd_v4_2_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
-{
-	amdgpu_ring_write(ring, PACKET0(mmHDP_DEBUG0, 0));
-	amdgpu_ring_write(ring, 1);
-}
-
-/**
  * uvd_v4_2_ring_test_ring - register write test
  *
  * @ring: amdgpu_ring pointer
@@ -505,11 +512,9 @@ static int uvd_v4_2_ring_test_ring(struct amdgpu_ring *ring)
 
 	WREG32(mmUVD_CONTEXT_ID, 0xCAFEDEAD);
 	r = amdgpu_ring_alloc(ring, 3);
-	if (r) {
-		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n",
-			  ring->idx, r);
+	if (r)
 		return r;
-	}
+
 	amdgpu_ring_write(ring, PACKET0(mmUVD_CONTEXT_ID, 0));
 	amdgpu_ring_write(ring, 0xDEADBEEF);
 	amdgpu_ring_commit(ring);
@@ -517,17 +522,12 @@ static int uvd_v4_2_ring_test_ring(struct amdgpu_ring *ring)
 		tmp = RREG32(mmUVD_CONTEXT_ID);
 		if (tmp == 0xDEADBEEF)
 			break;
-		DRM_UDELAY(1);
+		udelay(1);
 	}
 
-	if (i < adev->usec_timeout) {
-		DRM_INFO("ring test on %d succeeded in %d usecs\n",
-			 ring->idx, i);
-	} else {
-		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
-			  ring->idx, tmp);
-		r = -EINVAL;
-	}
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+
 	return r;
 }
 
@@ -535,13 +535,16 @@ static int uvd_v4_2_ring_test_ring(struct amdgpu_ring *ring)
  * uvd_v4_2_ring_emit_ib - execute indirect buffer
  *
  * @ring: amdgpu_ring pointer
+ * @job: iob associated with the indirect buffer
  * @ib: indirect buffer to execute
+ * @flags: flags associated with the indirect buffer
  *
  * Write ring commands to execute the indirect buffer
  */
 static void uvd_v4_2_ring_emit_ib(struct amdgpu_ring *ring,
+				  struct amdgpu_job *job,
 				  struct amdgpu_ib *ib,
-				  unsigned vm_id, bool ctx_switch)
+				  uint32_t flags)
 {
 	amdgpu_ring_write(ring, PACKET0(mmUVD_RBC_IB_BASE, 0));
 	amdgpu_ring_write(ring, ib->gpu_addr);
@@ -549,6 +552,18 @@ static void uvd_v4_2_ring_emit_ib(struct amdgpu_ring *ring,
 	amdgpu_ring_write(ring, ib->length_dw);
 }
 
+static void uvd_v4_2_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	int i;
+
+	WARN_ON(ring->wptr % 2 || count % 2);
+
+	for (i = 0; i < count / 2; i++) {
+		amdgpu_ring_write(ring, PACKET0(mmUVD_NO_OP, 0));
+		amdgpu_ring_write(ring, 0);
+	}
+}
+
 /**
  * uvd_v4_2_mc_resume - memory controller programming
  *
@@ -561,9 +576,9 @@ static void uvd_v4_2_mc_resume(struct amdgpu_device *adev)
 	uint64_t addr;
 	uint32_t size;
 
-	/* programm the VCPU memory controller bits 0-27 */
-	addr = (adev->uvd.gpu_addr + AMDGPU_UVD_FIRMWARE_OFFSET) >> 3;
-	size = AMDGPU_GPU_PAGE_ALIGN(adev->uvd.fw->size + 4) >> 3;
+	/* program the VCPU memory controller bits 0-27 */
+	addr = (adev->uvd.inst->gpu_addr + AMDGPU_UVD_FIRMWARE_OFFSET) >> 3;
+	size = AMDGPU_UVD_FIRMWARE_SIZE(adev) >> 3;
 	WREG32(mmUVD_VCPU_CACHE_OFFSET0, addr);
 	WREG32(mmUVD_VCPU_CACHE_SIZE0, size);
 
@@ -579,11 +594,11 @@ static void uvd_v4_2_mc_resume(struct amdgpu_device *adev)
 	WREG32(mmUVD_VCPU_CACHE_SIZE2, size);
 
 	/* bits 28-31 */
-	addr = (adev->uvd.gpu_addr >> 28) & 0xF;
+	addr = (adev->uvd.inst->gpu_addr >> 28) & 0xF;
 	WREG32(mmUVD_LMI_ADDR_EXT, (addr << 12) | (addr << 0));
 
 	/* bits 32-39 */
-	addr = (adev->uvd.gpu_addr >> 32) & 0xFF;
+	addr = (adev->uvd.inst->gpu_addr >> 32) & 0xFF;
 	WREG32(mmUVD_LMI_EXT40_ADDR, addr | (0x9 << 16) | (0x1 << 31));
 
 	WREG32(mmUVD_UDEC_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
@@ -644,17 +659,17 @@ static void uvd_v4_2_set_dcm(struct amdgpu_device *adev,
 	WREG32_UVD_CTX(ixUVD_CGC_CTRL2, tmp2);
 }
 
-static bool uvd_v4_2_is_idle(void *handle)
+static bool uvd_v4_2_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	return !(RREG32(mmSRBM_STATUS) & SRBM_STATUS__UVD_BUSY_MASK);
 }
 
-static int uvd_v4_2_wait_for_idle(void *handle)
+static int uvd_v4_2_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
 	unsigned i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
 		if (!(RREG32(mmSRBM_STATUS) & SRBM_STATUS__UVD_BUSY_MASK))
@@ -663,9 +678,9 @@ static int uvd_v4_2_wait_for_idle(void *handle)
 	return -ETIMEDOUT;
 }
 
-static int uvd_v4_2_soft_reset(void *handle)
+static int uvd_v4_2_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	uvd_v4_2_stop(adev);
 
@@ -690,17 +705,17 @@ static int uvd_v4_2_process_interrupt(struct amdgpu_device *adev,
 				      struct amdgpu_iv_entry *entry)
 {
 	DRM_DEBUG("IH: UVD TRAP\n");
-	amdgpu_fence_process(&adev->uvd.ring);
+	amdgpu_fence_process(&adev->uvd.inst->ring);
 	return 0;
 }
 
-static int uvd_v4_2_set_clockgating_state(void *handle,
+static int uvd_v4_2_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
 	return 0;
 }
 
-static int uvd_v4_2_set_powergating_state(void *handle,
+static int uvd_v4_2_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	/* This doesn't actually powergate the UVD block.
@@ -710,11 +725,11 @@ static int uvd_v4_2_set_powergating_state(void *handle,
 	 * revisit this when there is a cleaner line between
 	 * the smc and the hw blocks
 	 */
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (state == AMD_PG_STATE_GATE) {
 		uvd_v4_2_stop(adev);
-		if (adev->pg_flags & AMD_PG_SUPPORT_UVD && amdgpu_dpm == 0) {
+		if (adev->pg_flags & AMD_PG_SUPPORT_UVD && !adev->pm.dpm_enabled) {
 			if (!(RREG32_SMC(ixCURRENT_PG_STATUS) &
 				CURRENT_PG_STATUS__UVD_PG_STATUS_MASK)) {
 				WREG32(mmUVD_PGFSM_CONFIG, (UVD_PGFSM_CONFIG__UVD_PGFSM_FSM_ADDR_MASK   |
@@ -725,7 +740,7 @@ static int uvd_v4_2_set_powergating_state(void *handle,
 		}
 		return 0;
 	} else {
-		if (adev->pg_flags & AMD_PG_SUPPORT_UVD && amdgpu_dpm == 0) {
+		if (adev->pg_flags & AMD_PG_SUPPORT_UVD && !adev->pm.dpm_enabled) {
 			if (RREG32_SMC(ixCURRENT_PG_STATUS) &
 				CURRENT_PG_STATUS__UVD_PG_STATUS_MASK) {
 				WREG32(mmUVD_PGFSM_CONFIG, (UVD_PGFSM_CONFIG__UVD_PGFSM_FSM_ADDR_MASK   |
@@ -741,11 +756,11 @@ static int uvd_v4_2_set_powergating_state(void *handle,
 static const struct amd_ip_funcs uvd_v4_2_ip_funcs = {
 	.name = "uvd_v4_2",
 	.early_init = uvd_v4_2_early_init,
-	.late_init = NULL,
 	.sw_init = uvd_v4_2_sw_init,
 	.sw_fini = uvd_v4_2_sw_fini,
 	.hw_init = uvd_v4_2_hw_init,
 	.hw_fini = uvd_v4_2_hw_fini,
+	.prepare_suspend = uvd_v4_2_prepare_suspend,
 	.suspend = uvd_v4_2_suspend,
 	.resume = uvd_v4_2_resume,
 	.is_idle = uvd_v4_2_is_idle,
@@ -758,24 +773,20 @@ static const struct amd_ip_funcs uvd_v4_2_ip_funcs = {
 static const struct amdgpu_ring_funcs uvd_v4_2_ring_funcs = {
 	.type = AMDGPU_RING_TYPE_UVD,
 	.align_mask = 0xf,
-	.nop = PACKET0(mmUVD_NO_OP, 0),
 	.support_64bit_ptrs = false,
+	.no_user_fence = true,
 	.get_rptr = uvd_v4_2_ring_get_rptr,
 	.get_wptr = uvd_v4_2_ring_get_wptr,
 	.set_wptr = uvd_v4_2_ring_set_wptr,
 	.parse_cs = amdgpu_uvd_ring_parse_cs,
 	.emit_frame_size =
-		2 + /* uvd_v4_2_ring_emit_hdp_flush */
-		2 + /* uvd_v4_2_ring_emit_hdp_invalidate */
 		14, /* uvd_v4_2_ring_emit_fence  x1 no user fence */
 	.emit_ib_size = 4, /* uvd_v4_2_ring_emit_ib */
 	.emit_ib = uvd_v4_2_ring_emit_ib,
 	.emit_fence = uvd_v4_2_ring_emit_fence,
-	.emit_hdp_flush = uvd_v4_2_ring_emit_hdp_flush,
-	.emit_hdp_invalidate = uvd_v4_2_ring_emit_hdp_invalidate,
 	.test_ring = uvd_v4_2_ring_test_ring,
 	.test_ib = amdgpu_uvd_ring_test_ib,
-	.insert_nop = amdgpu_ring_insert_nop,
+	.insert_nop = uvd_v4_2_ring_insert_nop,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
 	.begin_use = amdgpu_uvd_ring_begin_use,
 	.end_use = amdgpu_uvd_ring_end_use,
@@ -783,7 +794,7 @@ static const struct amdgpu_ring_funcs uvd_v4_2_ring_funcs = {
 
 static void uvd_v4_2_set_ring_funcs(struct amdgpu_device *adev)
 {
-	adev->uvd.ring.funcs = &uvd_v4_2_ring_funcs;
+	adev->uvd.inst->ring.funcs = &uvd_v4_2_ring_funcs;
 }
 
 static const struct amdgpu_irq_src_funcs uvd_v4_2_irq_funcs = {
@@ -793,8 +804,8 @@ static const struct amdgpu_irq_src_funcs uvd_v4_2_irq_funcs = {
 
 static void uvd_v4_2_set_irq_funcs(struct amdgpu_device *adev)
 {
-	adev->uvd.irq.num_types = 1;
-	adev->uvd.irq.funcs = &uvd_v4_2_irq_funcs;
+	adev->uvd.inst->irq.num_types = 1;
+	adev->uvd.inst->irq.funcs = &uvd_v4_2_irq_funcs;
 }
 
 const struct amdgpu_ip_block_version uvd_v4_2_ip_block =
diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c
index bb6d46e168a3..71409ad8b7ed 100644
--- a/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c
@@ -22,8 +22,9 @@
  * Authors: Christian König <christian.koenig@amd.com>
  */
 
+#include <linux/delay.h>
 #include <linux/firmware.h>
-#include <drm/drmP.h>
+
 #include "amdgpu.h"
 #include "amdgpu_uvd.h"
 #include "vid.h"
@@ -35,12 +36,13 @@
 #include "vi.h"
 #include "smu/smu_7_1_2_d.h"
 #include "smu/smu_7_1_2_sh_mask.h"
+#include "ivsrcid/ivsrcid_vislands30.h"
 
 static void uvd_v5_0_set_ring_funcs(struct amdgpu_device *adev);
 static void uvd_v5_0_set_irq_funcs(struct amdgpu_device *adev);
 static int uvd_v5_0_start(struct amdgpu_device *adev);
 static void uvd_v5_0_stop(struct amdgpu_device *adev);
-static int uvd_v5_0_set_clockgating_state(void *handle,
+static int uvd_v5_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state);
 static void uvd_v5_0_enable_mgcg(struct amdgpu_device *adev,
 				 bool enable);
@@ -86,9 +88,10 @@ static void uvd_v5_0_ring_set_wptr(struct amdgpu_ring *ring)
 	WREG32(mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr));
 }
 
-static int uvd_v5_0_early_init(void *handle)
+static int uvd_v5_0_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+	adev->uvd.num_uvd_inst = 1;
 
 	uvd_v5_0_set_ring_funcs(adev);
 	uvd_v5_0_set_irq_funcs(adev);
@@ -96,14 +99,14 @@ static int uvd_v5_0_early_init(void *handle)
 	return 0;
 }
 
-static int uvd_v5_0_sw_init(void *handle)
+static int uvd_v5_0_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_ring *ring;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int r;
 
 	/* UVD TRAP */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 124, &adev->uvd.irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_UVD_SYSTEM_MESSAGE, &adev->uvd.inst->irq);
 	if (r)
 		return r;
 
@@ -111,21 +114,24 @@ static int uvd_v5_0_sw_init(void *handle)
 	if (r)
 		return r;
 
-	r = amdgpu_uvd_resume(adev);
+	ring = &adev->uvd.inst->ring;
+	sprintf(ring->name, "uvd");
+	r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst->irq, 0,
+			     AMDGPU_RING_PRIO_DEFAULT, NULL);
 	if (r)
 		return r;
 
-	ring = &adev->uvd.ring;
-	sprintf(ring->name, "uvd");
-	r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.irq, 0);
+	r = amdgpu_uvd_resume(adev);
+	if (r)
+		return r;
 
 	return r;
 }
 
-static int uvd_v5_0_sw_fini(void *handle)
+static int uvd_v5_0_sw_fini(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	r = amdgpu_uvd_suspend(adev);
 	if (r)
@@ -137,27 +143,24 @@ static int uvd_v5_0_sw_fini(void *handle)
 /**
  * uvd_v5_0_hw_init - start and test UVD block
  *
- * @adev: amdgpu_device pointer
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
  *
  * Initialize the hardware, boot up the VCPU and do some testing
  */
-static int uvd_v5_0_hw_init(void *handle)
+static int uvd_v5_0_hw_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	struct amdgpu_ring *ring = &adev->uvd.ring;
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring = &adev->uvd.inst->ring;
 	uint32_t tmp;
 	int r;
 
 	amdgpu_asic_set_uvd_clocks(adev, 10000, 10000);
-	uvd_v5_0_set_clockgating_state(adev, AMD_CG_STATE_UNGATE);
+	uvd_v5_0_set_clockgating_state(ip_block, AMD_CG_STATE_UNGATE);
 	uvd_v5_0_enable_mgcg(adev, true);
 
-	ring->ready = true;
-	r = amdgpu_ring_test_ring(ring);
-	if (r) {
-		ring->ready = false;
+	r = amdgpu_ring_test_helper(ring);
+	if (r)
 		goto done;
-	}
 
 	r = amdgpu_ring_alloc(ring, 10);
 	if (r) {
@@ -197,46 +200,74 @@ done:
 /**
  * uvd_v5_0_hw_fini - stop the hardware block
  *
- * @adev: amdgpu_device pointer
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
  *
  * Stop the UVD block, mark ring as not ready any more
  */
-static int uvd_v5_0_hw_fini(void *handle)
+static int uvd_v5_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	struct amdgpu_ring *ring = &adev->uvd.ring;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	cancel_delayed_work_sync(&adev->uvd.idle_work);
 
 	if (RREG32(mmUVD_STATUS) != 0)
 		uvd_v5_0_stop(adev);
 
-	ring->ready = false;
-
 	return 0;
 }
 
-static int uvd_v5_0_suspend(void *handle)
+static int uvd_v5_0_prepare_suspend(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	return amdgpu_uvd_prepare_suspend(adev);
+}
+
+static int uvd_v5_0_suspend(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	/*
+	 * Proper cleanups before halting the HW engine:
+	 *   - cancel the delayed idle work
+	 *   - enable powergating
+	 *   - enable clockgating
+	 *   - disable dpm
+	 *
+	 * TODO: to align with the VCN implementation, move the
+	 * jobs for clockgating/powergating/dpm setting to
+	 * ->set_powergating_state().
+	 */
+	cancel_delayed_work_sync(&adev->uvd.idle_work);
 
-	r = uvd_v5_0_hw_fini(adev);
+	if (adev->pm.dpm_enabled) {
+		amdgpu_dpm_enable_uvd(adev, false);
+	} else {
+		amdgpu_asic_set_uvd_clocks(adev, 0, 0);
+		/* shutdown the UVD block */
+		amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+						       AMD_PG_STATE_GATE);
+		amdgpu_device_ip_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+						       AMD_CG_STATE_GATE);
+	}
+
+	r = uvd_v5_0_hw_fini(ip_block);
 	if (r)
 		return r;
-	uvd_v5_0_set_clockgating_state(adev, AMD_CG_STATE_GATE);
 
 	return amdgpu_uvd_suspend(adev);
 }
 
-static int uvd_v5_0_resume(void *handle)
+static int uvd_v5_0_resume(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	r = amdgpu_uvd_resume(adev);
+	r = amdgpu_uvd_resume(ip_block->adev);
 	if (r)
 		return r;
 
-	return uvd_v5_0_hw_init(adev);
+	return uvd_v5_0_hw_init(ip_block);
 }
 
 /**
@@ -251,14 +282,14 @@ static void uvd_v5_0_mc_resume(struct amdgpu_device *adev)
 	uint64_t offset;
 	uint32_t size;
 
-	/* programm memory controller bits 0-27 */
+	/* program memory controller bits 0-27 */
 	WREG32(mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
-			lower_32_bits(adev->uvd.gpu_addr));
+			lower_32_bits(adev->uvd.inst->gpu_addr));
 	WREG32(mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
-			upper_32_bits(adev->uvd.gpu_addr));
+			upper_32_bits(adev->uvd.inst->gpu_addr));
 
 	offset = AMDGPU_UVD_FIRMWARE_OFFSET;
-	size = AMDGPU_GPU_PAGE_ALIGN(adev->uvd.fw->size + 4);
+	size = AMDGPU_UVD_FIRMWARE_SIZE(adev);
 	WREG32(mmUVD_VCPU_CACHE_OFFSET0, offset >> 3);
 	WREG32(mmUVD_VCPU_CACHE_SIZE0, size);
 
@@ -287,7 +318,7 @@ static void uvd_v5_0_mc_resume(struct amdgpu_device *adev)
  */
 static int uvd_v5_0_start(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *ring = &adev->uvd.ring;
+	struct amdgpu_ring *ring = &adev->uvd.inst->ring;
 	uint32_t rb_bufsz, tmp;
 	uint32_t lmi_swap_cntl;
 	uint32_t mp_swap_cntl;
@@ -402,7 +433,7 @@ static int uvd_v5_0_start(struct amdgpu_device *adev)
 	/* set the wb address */
 	WREG32(mmUVD_RBC_RB_RPTR_ADDR, (upper_32_bits(ring->gpu_addr) >> 2));
 
-	/* programm the RB_BASE for ring buffer */
+	/* program the RB_BASE for ring buffer */
 	WREG32(mmUVD_LMI_RBC_RB_64BIT_BAR_LOW,
 			lower_32_bits(ring->gpu_addr));
 	WREG32(mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH,
@@ -452,7 +483,9 @@ static void uvd_v5_0_stop(struct amdgpu_device *adev)
  * uvd_v5_0_ring_emit_fence - emit an fence & trap command
  *
  * @ring: amdgpu_ring pointer
- * @fence: fence to emit
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
  *
  * Write a fence and a trap command to the ring.
  */
@@ -479,32 +512,6 @@ static void uvd_v5_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq
 }
 
 /**
- * uvd_v5_0_ring_emit_hdp_flush - emit an hdp flush
- *
- * @ring: amdgpu_ring pointer
- *
- * Emits an hdp flush.
- */
-static void uvd_v5_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
-{
-	amdgpu_ring_write(ring, PACKET0(mmHDP_MEM_COHERENCY_FLUSH_CNTL, 0));
-	amdgpu_ring_write(ring, 0);
-}
-
-/**
- * uvd_v5_0_ring_hdp_invalidate - emit an hdp invalidate
- *
- * @ring: amdgpu_ring pointer
- *
- * Emits an hdp invalidate.
- */
-static void uvd_v5_0_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
-{
-	amdgpu_ring_write(ring, PACKET0(mmHDP_DEBUG0, 0));
-	amdgpu_ring_write(ring, 1);
-}
-
-/**
  * uvd_v5_0_ring_test_ring - register write test
  *
  * @ring: amdgpu_ring pointer
@@ -520,11 +527,8 @@ static int uvd_v5_0_ring_test_ring(struct amdgpu_ring *ring)
 
 	WREG32(mmUVD_CONTEXT_ID, 0xCAFEDEAD);
 	r = amdgpu_ring_alloc(ring, 3);
-	if (r) {
-		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n",
-			  ring->idx, r);
+	if (r)
 		return r;
-	}
 	amdgpu_ring_write(ring, PACKET0(mmUVD_CONTEXT_ID, 0));
 	amdgpu_ring_write(ring, 0xDEADBEEF);
 	amdgpu_ring_commit(ring);
@@ -532,17 +536,12 @@ static int uvd_v5_0_ring_test_ring(struct amdgpu_ring *ring)
 		tmp = RREG32(mmUVD_CONTEXT_ID);
 		if (tmp == 0xDEADBEEF)
 			break;
-		DRM_UDELAY(1);
+		udelay(1);
 	}
 
-	if (i < adev->usec_timeout) {
-		DRM_INFO("ring test on %d succeeded in %d usecs\n",
-			 ring->idx, i);
-	} else {
-		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
-			  ring->idx, tmp);
-		r = -EINVAL;
-	}
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+
 	return r;
 }
 
@@ -550,13 +549,16 @@ static int uvd_v5_0_ring_test_ring(struct amdgpu_ring *ring)
  * uvd_v5_0_ring_emit_ib - execute indirect buffer
  *
  * @ring: amdgpu_ring pointer
+ * @job: job to retrieve vmid from
  * @ib: indirect buffer to execute
+ * @flags: unused
  *
  * Write ring commands to execute the indirect buffer
  */
 static void uvd_v5_0_ring_emit_ib(struct amdgpu_ring *ring,
+				  struct amdgpu_job *job,
 				  struct amdgpu_ib *ib,
-				  unsigned vm_id, bool ctx_switch)
+				  uint32_t flags)
 {
 	amdgpu_ring_write(ring, PACKET0(mmUVD_LMI_RBC_IB_64BIT_BAR_LOW, 0));
 	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
@@ -566,17 +568,29 @@ static void uvd_v5_0_ring_emit_ib(struct amdgpu_ring *ring,
 	amdgpu_ring_write(ring, ib->length_dw);
 }
 
-static bool uvd_v5_0_is_idle(void *handle)
+static void uvd_v5_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	int i;
+
+	WARN_ON(ring->wptr % 2 || count % 2);
+
+	for (i = 0; i < count / 2; i++) {
+		amdgpu_ring_write(ring, PACKET0(mmUVD_NO_OP, 0));
+		amdgpu_ring_write(ring, 0);
+	}
+}
+
+static bool uvd_v5_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	return !(RREG32(mmSRBM_STATUS) & SRBM_STATUS__UVD_BUSY_MASK);
 }
 
-static int uvd_v5_0_wait_for_idle(void *handle)
+static int uvd_v5_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
 	unsigned i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
 		if (!(RREG32(mmSRBM_STATUS) & SRBM_STATUS__UVD_BUSY_MASK))
@@ -585,9 +599,9 @@ static int uvd_v5_0_wait_for_idle(void *handle)
 	return -ETIMEDOUT;
 }
 
-static int uvd_v5_0_soft_reset(void *handle)
+static int uvd_v5_0_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	uvd_v5_0_stop(adev);
 
@@ -612,7 +626,7 @@ static int uvd_v5_0_process_interrupt(struct amdgpu_device *adev,
 				      struct amdgpu_iv_entry *entry)
 {
 	DRM_DEBUG("IH: UVD TRAP\n");
-	amdgpu_fence_process(&adev->uvd.ring);
+	amdgpu_fence_process(&adev->uvd.inst->ring);
 	return 0;
 }
 
@@ -776,15 +790,15 @@ static void uvd_v5_0_enable_mgcg(struct amdgpu_device *adev,
 	}
 }
 
-static int uvd_v5_0_set_clockgating_state(void *handle,
+static int uvd_v5_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	bool enable = (state == AMD_CG_STATE_GATE) ? true : false;
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = (state == AMD_CG_STATE_GATE);
 
 	if (enable) {
 		/* wait for STATUS to clear */
-		if (uvd_v5_0_wait_for_idle(handle))
+		if (uvd_v5_0_wait_for_idle(ip_block))
 			return -EBUSY;
 		uvd_v5_0_enable_clock_gating(adev, true);
 
@@ -798,7 +812,7 @@ static int uvd_v5_0_set_clockgating_state(void *handle,
 	return 0;
 }
 
-static int uvd_v5_0_set_powergating_state(void *handle,
+static int uvd_v5_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	/* This doesn't actually powergate the UVD block.
@@ -808,7 +822,7 @@ static int uvd_v5_0_set_powergating_state(void *handle,
 	 * revisit this when there is a cleaner line between
 	 * the smc and the hw blocks
 	 */
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int ret = 0;
 
 	if (state == AMD_PG_STATE_GATE) {
@@ -823,9 +837,9 @@ out:
 	return ret;
 }
 
-static void uvd_v5_0_get_clockgating_state(void *handle, u32 *flags)
+static void uvd_v5_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int data;
 
 	mutex_lock(&adev->pm.mutex);
@@ -848,11 +862,11 @@ out:
 static const struct amd_ip_funcs uvd_v5_0_ip_funcs = {
 	.name = "uvd_v5_0",
 	.early_init = uvd_v5_0_early_init,
-	.late_init = NULL,
 	.sw_init = uvd_v5_0_sw_init,
 	.sw_fini = uvd_v5_0_sw_fini,
 	.hw_init = uvd_v5_0_hw_init,
 	.hw_fini = uvd_v5_0_hw_fini,
+	.prepare_suspend = uvd_v5_0_prepare_suspend,
 	.suspend = uvd_v5_0_suspend,
 	.resume = uvd_v5_0_resume,
 	.is_idle = uvd_v5_0_is_idle,
@@ -866,24 +880,20 @@ static const struct amd_ip_funcs uvd_v5_0_ip_funcs = {
 static const struct amdgpu_ring_funcs uvd_v5_0_ring_funcs = {
 	.type = AMDGPU_RING_TYPE_UVD,
 	.align_mask = 0xf,
-	.nop = PACKET0(mmUVD_NO_OP, 0),
 	.support_64bit_ptrs = false,
+	.no_user_fence = true,
 	.get_rptr = uvd_v5_0_ring_get_rptr,
 	.get_wptr = uvd_v5_0_ring_get_wptr,
 	.set_wptr = uvd_v5_0_ring_set_wptr,
 	.parse_cs = amdgpu_uvd_ring_parse_cs,
 	.emit_frame_size =
-		2 + /* uvd_v5_0_ring_emit_hdp_flush */
-		2 + /* uvd_v5_0_ring_emit_hdp_invalidate */
 		14, /* uvd_v5_0_ring_emit_fence  x1 no user fence */
 	.emit_ib_size = 6, /* uvd_v5_0_ring_emit_ib */
 	.emit_ib = uvd_v5_0_ring_emit_ib,
 	.emit_fence = uvd_v5_0_ring_emit_fence,
-	.emit_hdp_flush = uvd_v5_0_ring_emit_hdp_flush,
-	.emit_hdp_invalidate = uvd_v5_0_ring_emit_hdp_invalidate,
 	.test_ring = uvd_v5_0_ring_test_ring,
 	.test_ib = amdgpu_uvd_ring_test_ib,
-	.insert_nop = amdgpu_ring_insert_nop,
+	.insert_nop = uvd_v5_0_ring_insert_nop,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
 	.begin_use = amdgpu_uvd_ring_begin_use,
 	.end_use = amdgpu_uvd_ring_end_use,
@@ -891,7 +901,7 @@ static const struct amdgpu_ring_funcs uvd_v5_0_ring_funcs = {
 
 static void uvd_v5_0_set_ring_funcs(struct amdgpu_device *adev)
 {
-	adev->uvd.ring.funcs = &uvd_v5_0_ring_funcs;
+	adev->uvd.inst->ring.funcs = &uvd_v5_0_ring_funcs;
 }
 
 static const struct amdgpu_irq_src_funcs uvd_v5_0_irq_funcs = {
@@ -901,8 +911,8 @@ static const struct amdgpu_irq_src_funcs uvd_v5_0_irq_funcs = {
 
 static void uvd_v5_0_set_irq_funcs(struct amdgpu_device *adev)
 {
-	adev->uvd.irq.num_types = 1;
-	adev->uvd.irq.funcs = &uvd_v5_0_irq_funcs;
+	adev->uvd.inst->irq.num_types = 1;
+	adev->uvd.inst->irq.funcs = &uvd_v5_0_irq_funcs;
 }
 
 const struct amdgpu_ip_block_version uvd_v5_0_ip_block =
diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c
index 31db356476f8..ceb94bbb03a4 100644
--- a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c
@@ -23,7 +23,7 @@
  */
 
 #include <linux/firmware.h>
-#include <drm/drmP.h>
+
 #include "amdgpu.h"
 #include "amdgpu_uvd.h"
 #include "vid.h"
@@ -36,18 +36,38 @@
 #include "bif/bif_5_1_d.h"
 #include "gmc/gmc_8_1_d.h"
 #include "vi.h"
+#include "ivsrcid/ivsrcid_vislands30.h"
+
+/* Polaris10/11/12 firmware version */
+#define FW_1_130_16 ((1 << 24) | (130 << 16) | (16 << 8))
 
 static void uvd_v6_0_set_ring_funcs(struct amdgpu_device *adev);
+static void uvd_v6_0_set_enc_ring_funcs(struct amdgpu_device *adev);
+
 static void uvd_v6_0_set_irq_funcs(struct amdgpu_device *adev);
 static int uvd_v6_0_start(struct amdgpu_device *adev);
 static void uvd_v6_0_stop(struct amdgpu_device *adev);
 static void uvd_v6_0_set_sw_clock_gating(struct amdgpu_device *adev);
-static int uvd_v6_0_set_clockgating_state(void *handle,
+static int uvd_v6_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state);
 static void uvd_v6_0_enable_mgcg(struct amdgpu_device *adev,
 				 bool enable);
 
 /**
+* uvd_v6_0_enc_support - get encode support status
+*
+* @adev: amdgpu_device pointer
+*
+* Returns the current hardware encode support status
+*/
+static inline bool uvd_v6_0_enc_support(struct amdgpu_device *adev)
+{
+	return ((adev->asic_type >= CHIP_POLARIS10) &&
+			(adev->asic_type <= CHIP_VEGAM) &&
+			(!adev->uvd.fw_version || adev->uvd.fw_version >= FW_1_130_16));
+}
+
+/**
  * uvd_v6_0_ring_get_rptr - get read pointer
  *
  * @ring: amdgpu_ring pointer
@@ -62,6 +82,22 @@ static uint64_t uvd_v6_0_ring_get_rptr(struct amdgpu_ring *ring)
 }
 
 /**
+ * uvd_v6_0_enc_ring_get_rptr - get enc read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware enc read pointer
+ */
+static uint64_t uvd_v6_0_enc_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->uvd.inst->ring_enc[0])
+		return RREG32(mmUVD_RB_RPTR);
+	else
+		return RREG32(mmUVD_RB_RPTR2);
+}
+/**
  * uvd_v6_0_ring_get_wptr - get write pointer
  *
  * @ring: amdgpu_ring pointer
@@ -76,6 +112,23 @@ static uint64_t uvd_v6_0_ring_get_wptr(struct amdgpu_ring *ring)
 }
 
 /**
+ * uvd_v6_0_enc_ring_get_wptr - get enc write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware enc write pointer
+ */
+static uint64_t uvd_v6_0_enc_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->uvd.inst->ring_enc[0])
+		return RREG32(mmUVD_RB_WPTR);
+	else
+		return RREG32(mmUVD_RB_WPTR2);
+}
+
+/**
  * uvd_v6_0_ring_set_wptr - set write pointer
  *
  * @ring: amdgpu_ring pointer
@@ -89,78 +142,339 @@ static void uvd_v6_0_ring_set_wptr(struct amdgpu_ring *ring)
 	WREG32(mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr));
 }
 
-static int uvd_v6_0_early_init(void *handle)
+/**
+ * uvd_v6_0_enc_ring_set_wptr - set enc write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the enc write pointer to the hardware
+ */
+static void uvd_v6_0_enc_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->uvd.inst->ring_enc[0])
+		WREG32(mmUVD_RB_WPTR,
+			lower_32_bits(ring->wptr));
+	else
+		WREG32(mmUVD_RB_WPTR2,
+			lower_32_bits(ring->wptr));
+}
+
+/**
+ * uvd_v6_0_enc_ring_test_ring - test if UVD ENC ring is working
+ *
+ * @ring: the engine to test on
+ *
+ */
+static int uvd_v6_0_enc_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t rptr;
+	unsigned i;
+	int r;
+
+	r = amdgpu_ring_alloc(ring, 16);
+	if (r)
+		return r;
+
+	rptr = amdgpu_ring_get_rptr(ring);
+
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_END);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (amdgpu_ring_get_rptr(ring) != rptr)
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+
+	return r;
+}
+
+/**
+ * uvd_v6_0_enc_get_create_msg - generate a UVD ENC create msg
+ *
+ * @ring: ring we should submit the msg to
+ * @handle: session handle to use
+ * @bo: amdgpu object for which we query the offset
+ * @fence: optional fence to return
+ *
+ * Open up a stream for HW test
+ */
+static int uvd_v6_0_enc_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
+				       struct amdgpu_bo *bo,
+				       struct dma_fence **fence)
+{
+	const unsigned ib_size_dw = 16;
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	struct dma_fence *f = NULL;
+	uint64_t addr;
+	int i, r;
+
+	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL, ib_size_dw * 4,
+				     AMDGPU_IB_POOL_DIRECT, &job,
+				     AMDGPU_KERNEL_JOB_ID_VCN_RING_TEST);
+	if (r)
+		return r;
+
+	ib = &job->ibs[0];
+	addr = amdgpu_bo_gpu_offset(bo);
+
+	ib->length_dw = 0;
+	ib->ptr[ib->length_dw++] = 0x00000018;
+	ib->ptr[ib->length_dw++] = 0x00000001; /* session info */
+	ib->ptr[ib->length_dw++] = handle;
+	ib->ptr[ib->length_dw++] = 0x00010000;
+	ib->ptr[ib->length_dw++] = upper_32_bits(addr);
+	ib->ptr[ib->length_dw++] = addr;
+
+	ib->ptr[ib->length_dw++] = 0x00000014;
+	ib->ptr[ib->length_dw++] = 0x00000002; /* task info */
+	ib->ptr[ib->length_dw++] = 0x0000001c;
+	ib->ptr[ib->length_dw++] = 0x00000001;
+	ib->ptr[ib->length_dw++] = 0x00000000;
+
+	ib->ptr[ib->length_dw++] = 0x00000008;
+	ib->ptr[ib->length_dw++] = 0x08000001; /* op initialize */
+
+	for (i = ib->length_dw; i < ib_size_dw; ++i)
+		ib->ptr[i] = 0x0;
+
+	r = amdgpu_job_submit_direct(job, ring, &f);
+	if (r)
+		goto err;
+
+	if (fence)
+		*fence = dma_fence_get(f);
+	dma_fence_put(f);
+	return 0;
+
+err:
+	amdgpu_job_free(job);
+	return r;
+}
+
+/**
+ * uvd_v6_0_enc_get_destroy_msg - generate a UVD ENC destroy msg
+ *
+ * @ring: ring we should submit the msg to
+ * @handle: session handle to use
+ * @bo: amdgpu object for which we query the offset
+ * @fence: optional fence to return
+ *
+ * Close up a stream for HW test or if userspace failed to do so
+ */
+static int uvd_v6_0_enc_get_destroy_msg(struct amdgpu_ring *ring,
+					uint32_t handle,
+					struct amdgpu_bo *bo,
+					struct dma_fence **fence)
+{
+	const unsigned ib_size_dw = 16;
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	struct dma_fence *f = NULL;
+	uint64_t addr;
+	int i, r;
+
+	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL, ib_size_dw * 4,
+				     AMDGPU_IB_POOL_DIRECT, &job,
+				     AMDGPU_KERNEL_JOB_ID_VCN_RING_TEST);
+	if (r)
+		return r;
+
+	ib = &job->ibs[0];
+	addr = amdgpu_bo_gpu_offset(bo);
+
+	ib->length_dw = 0;
+	ib->ptr[ib->length_dw++] = 0x00000018;
+	ib->ptr[ib->length_dw++] = 0x00000001; /* session info */
+	ib->ptr[ib->length_dw++] = handle;
+	ib->ptr[ib->length_dw++] = 0x00010000;
+	ib->ptr[ib->length_dw++] = upper_32_bits(addr);
+	ib->ptr[ib->length_dw++] = addr;
+
+	ib->ptr[ib->length_dw++] = 0x00000014;
+	ib->ptr[ib->length_dw++] = 0x00000002; /* task info */
+	ib->ptr[ib->length_dw++] = 0x0000001c;
+	ib->ptr[ib->length_dw++] = 0x00000001;
+	ib->ptr[ib->length_dw++] = 0x00000000;
+
+	ib->ptr[ib->length_dw++] = 0x00000008;
+	ib->ptr[ib->length_dw++] = 0x08000002; /* op close session */
+
+	for (i = ib->length_dw; i < ib_size_dw; ++i)
+		ib->ptr[i] = 0x0;
+
+	r = amdgpu_job_submit_direct(job, ring, &f);
+	if (r)
+		goto err;
+
+	if (fence)
+		*fence = dma_fence_get(f);
+	dma_fence_put(f);
+	return 0;
+
+err:
+	amdgpu_job_free(job);
+	return r;
+}
+
+/**
+ * uvd_v6_0_enc_ring_test_ib - test if UVD ENC IBs are working
+ *
+ * @ring: the engine to test on
+ * @timeout: timeout value in jiffies, or MAX_SCHEDULE_TIMEOUT
+ *
+ */
+static int uvd_v6_0_enc_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct dma_fence *fence = NULL;
+	struct amdgpu_bo *bo = ring->adev->uvd.ib_bo;
+	long r;
+
+	r = uvd_v6_0_enc_get_create_msg(ring, 1, bo, NULL);
+	if (r)
+		goto error;
+
+	r = uvd_v6_0_enc_get_destroy_msg(ring, 1, bo, &fence);
+	if (r)
+		goto error;
+
+	r = dma_fence_wait_timeout(fence, false, timeout);
+	if (r == 0)
+		r = -ETIMEDOUT;
+	else if (r > 0)
+		r = 0;
+
+error:
+	dma_fence_put(fence);
+	return r;
+}
+
+static int uvd_v6_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	adev->uvd.num_uvd_inst = 1;
+
+	if (!(adev->flags & AMD_IS_APU) &&
+	    (RREG32_SMC(ixCC_HARVEST_FUSES) & CC_HARVEST_FUSES__UVD_DISABLE_MASK))
+		return -ENOENT;
 
 	uvd_v6_0_set_ring_funcs(adev);
+
+	if (uvd_v6_0_enc_support(adev)) {
+		adev->uvd.num_enc_rings = 2;
+		uvd_v6_0_set_enc_ring_funcs(adev);
+	}
+
 	uvd_v6_0_set_irq_funcs(adev);
 
 	return 0;
 }
 
-static int uvd_v6_0_sw_init(void *handle)
+static int uvd_v6_0_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_ring *ring;
-	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int i, r;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	/* UVD TRAP */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 124, &adev->uvd.irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_UVD_SYSTEM_MESSAGE, &adev->uvd.inst->irq);
 	if (r)
 		return r;
 
+	/* UVD ENC TRAP */
+	if (uvd_v6_0_enc_support(adev)) {
+		for (i = 0; i < adev->uvd.num_enc_rings; ++i) {
+			r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, i + VISLANDS30_IV_SRCID_UVD_ENC_GEN_PURP, &adev->uvd.inst->irq);
+			if (r)
+				return r;
+		}
+	}
+
 	r = amdgpu_uvd_sw_init(adev);
 	if (r)
 		return r;
 
+	if (!uvd_v6_0_enc_support(adev)) {
+		for (i = 0; i < adev->uvd.num_enc_rings; ++i)
+			adev->uvd.inst->ring_enc[i].funcs = NULL;
+
+		adev->uvd.inst->irq.num_types = 1;
+		adev->uvd.num_enc_rings = 0;
+
+		DRM_INFO("UVD ENC is disabled\n");
+	}
+
+	ring = &adev->uvd.inst->ring;
+	sprintf(ring->name, "uvd");
+	r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst->irq, 0,
+			     AMDGPU_RING_PRIO_DEFAULT, NULL);
+	if (r)
+		return r;
+
 	r = amdgpu_uvd_resume(adev);
 	if (r)
 		return r;
 
-	ring = &adev->uvd.ring;
-	sprintf(ring->name, "uvd");
-	r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.irq, 0);
+	if (uvd_v6_0_enc_support(adev)) {
+		for (i = 0; i < adev->uvd.num_enc_rings; ++i) {
+			ring = &adev->uvd.inst->ring_enc[i];
+			sprintf(ring->name, "uvd_enc%d", i);
+			r = amdgpu_ring_init(adev, ring, 512,
+					     &adev->uvd.inst->irq, 0,
+					     AMDGPU_RING_PRIO_DEFAULT, NULL);
+			if (r)
+				return r;
+		}
+	}
 
 	return r;
 }
 
-static int uvd_v6_0_sw_fini(void *handle)
+static int uvd_v6_0_sw_fini(struct amdgpu_ip_block *ip_block)
 {
-	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int i, r;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	r = amdgpu_uvd_suspend(adev);
 	if (r)
 		return r;
 
+	if (uvd_v6_0_enc_support(adev)) {
+		for (i = 0; i < adev->uvd.num_enc_rings; ++i)
+			amdgpu_ring_fini(&adev->uvd.inst->ring_enc[i]);
+	}
+
 	return amdgpu_uvd_sw_fini(adev);
 }
 
 /**
  * uvd_v6_0_hw_init - start and test UVD block
  *
- * @adev: amdgpu_device pointer
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
  *
  * Initialize the hardware, boot up the VCPU and do some testing
  */
-static int uvd_v6_0_hw_init(void *handle)
+static int uvd_v6_0_hw_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	struct amdgpu_ring *ring = &adev->uvd.ring;
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring = &adev->uvd.inst->ring;
 	uint32_t tmp;
-	int r;
+	int i, r;
 
 	amdgpu_asic_set_uvd_clocks(adev, 10000, 10000);
-	uvd_v6_0_set_clockgating_state(adev, AMD_CG_STATE_UNGATE);
+	uvd_v6_0_set_clockgating_state(ip_block, AMD_CG_STATE_UNGATE);
 	uvd_v6_0_enable_mgcg(adev, true);
 
-	ring->ready = true;
-	r = amdgpu_ring_test_ring(ring);
-	if (r) {
-		ring->ready = false;
+	r = amdgpu_ring_test_helper(ring);
+	if (r)
 		goto done;
-	}
 
 	r = amdgpu_ring_alloc(ring, 10);
 	if (r) {
@@ -189,9 +503,22 @@ static int uvd_v6_0_hw_init(void *handle)
 
 	amdgpu_ring_commit(ring);
 
+	if (uvd_v6_0_enc_support(adev)) {
+		for (i = 0; i < adev->uvd.num_enc_rings; ++i) {
+			ring = &adev->uvd.inst->ring_enc[i];
+			r = amdgpu_ring_test_helper(ring);
+			if (r)
+				goto done;
+		}
+	}
+
 done:
-	if (!r)
-		DRM_INFO("UVD initialized successfully.\n");
+	if (!r) {
+		if (uvd_v6_0_enc_support(adev))
+			DRM_INFO("UVD and UVD ENC initialized successfully.\n");
+		else
+			DRM_INFO("UVD initialized successfully.\n");
+	}
 
 	return r;
 }
@@ -199,51 +526,74 @@ done:
 /**
  * uvd_v6_0_hw_fini - stop the hardware block
  *
- * @adev: amdgpu_device pointer
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
  *
  * Stop the UVD block, mark ring as not ready any more
  */
-static int uvd_v6_0_hw_fini(void *handle)
+static int uvd_v6_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	struct amdgpu_ring *ring = &adev->uvd.ring;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	cancel_delayed_work_sync(&adev->uvd.idle_work);
 
 	if (RREG32(mmUVD_STATUS) != 0)
 		uvd_v6_0_stop(adev);
 
-	ring->ready = false;
-
 	return 0;
 }
 
-static int uvd_v6_0_suspend(void *handle)
+static int uvd_v6_0_prepare_suspend(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	return amdgpu_uvd_prepare_suspend(adev);
+}
+
+static int uvd_v6_0_suspend(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	/*
+	 * Proper cleanups before halting the HW engine:
+	 *   - cancel the delayed idle work
+	 *   - enable powergating
+	 *   - enable clockgating
+	 *   - disable dpm
+	 *
+	 * TODO: to align with the VCN implementation, move the
+	 * jobs for clockgating/powergating/dpm setting to
+	 * ->set_powergating_state().
+	 */
+	cancel_delayed_work_sync(&adev->uvd.idle_work);
+
+	if (adev->pm.dpm_enabled) {
+		amdgpu_dpm_enable_uvd(adev, false);
+	} else {
+		amdgpu_asic_set_uvd_clocks(adev, 0, 0);
+		/* shutdown the UVD block */
+		amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+						       AMD_PG_STATE_GATE);
+		amdgpu_device_ip_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+						       AMD_CG_STATE_GATE);
+	}
 
-	r = uvd_v6_0_hw_fini(adev);
+	r = uvd_v6_0_hw_fini(ip_block);
 	if (r)
 		return r;
 
-	/* Skip this for APU for now */
-	if (!(adev->flags & AMD_IS_APU))
-		r = amdgpu_uvd_suspend(adev);
-
-	return r;
+	return amdgpu_uvd_suspend(adev);
 }
 
-static int uvd_v6_0_resume(void *handle)
+static int uvd_v6_0_resume(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	/* Skip this for APU for now */
-	if (!(adev->flags & AMD_IS_APU)) {
-		r = amdgpu_uvd_resume(adev);
-		if (r)
-			return r;
-	}
-	return uvd_v6_0_hw_init(adev);
+	r = amdgpu_uvd_resume(ip_block->adev);
+	if (r)
+		return r;
+
+	return uvd_v6_0_hw_init(ip_block);
 }
 
 /**
@@ -258,14 +608,14 @@ static void uvd_v6_0_mc_resume(struct amdgpu_device *adev)
 	uint64_t offset;
 	uint32_t size;
 
-	/* programm memory controller bits 0-27 */
+	/* program memory controller bits 0-27 */
 	WREG32(mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
-			lower_32_bits(adev->uvd.gpu_addr));
+			lower_32_bits(adev->uvd.inst->gpu_addr));
 	WREG32(mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
-			upper_32_bits(adev->uvd.gpu_addr));
+			upper_32_bits(adev->uvd.inst->gpu_addr));
 
 	offset = AMDGPU_UVD_FIRMWARE_OFFSET;
-	size = AMDGPU_GPU_PAGE_ALIGN(adev->uvd.fw->size + 4);
+	size = AMDGPU_UVD_FIRMWARE_SIZE(adev);
 	WREG32(mmUVD_VCPU_CACHE_OFFSET0, offset >> 3);
 	WREG32(mmUVD_VCPU_CACHE_SIZE0, size);
 
@@ -375,7 +725,7 @@ static void cz_set_uvd_clock_gating_branches(struct amdgpu_device *adev,
  */
 static int uvd_v6_0_start(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *ring = &adev->uvd.ring;
+	struct amdgpu_ring *ring = &adev->uvd.inst->ring;
 	uint32_t rb_bufsz, tmp;
 	uint32_t lmi_swap_cntl;
 	uint32_t mp_swap_cntl;
@@ -500,7 +850,7 @@ static int uvd_v6_0_start(struct amdgpu_device *adev)
 	/* set the wb address */
 	WREG32(mmUVD_RBC_RB_RPTR_ADDR, (upper_32_bits(ring->gpu_addr) >> 2));
 
-	/* programm the RB_BASE for ring buffer */
+	/* program the RB_BASE for ring buffer */
 	WREG32(mmUVD_LMI_RBC_RB_64BIT_BAR_LOW,
 			lower_32_bits(ring->gpu_addr));
 	WREG32(mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH,
@@ -514,6 +864,22 @@ static int uvd_v6_0_start(struct amdgpu_device *adev)
 
 	WREG32_FIELD(UVD_RBC_RB_CNTL, RB_NO_FETCH, 0);
 
+	if (uvd_v6_0_enc_support(adev)) {
+		ring = &adev->uvd.inst->ring_enc[0];
+		WREG32(mmUVD_RB_RPTR, lower_32_bits(ring->wptr));
+		WREG32(mmUVD_RB_WPTR, lower_32_bits(ring->wptr));
+		WREG32(mmUVD_RB_BASE_LO, ring->gpu_addr);
+		WREG32(mmUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+		WREG32(mmUVD_RB_SIZE, ring->ring_size / 4);
+
+		ring = &adev->uvd.inst->ring_enc[1];
+		WREG32(mmUVD_RB_RPTR2, lower_32_bits(ring->wptr));
+		WREG32(mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
+		WREG32(mmUVD_RB_BASE_LO2, ring->gpu_addr);
+		WREG32(mmUVD_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
+		WREG32(mmUVD_RB_SIZE2, ring->ring_size / 4);
+	}
+
 	return 0;
 }
 
@@ -550,7 +916,9 @@ static void uvd_v6_0_stop(struct amdgpu_device *adev)
  * uvd_v6_0_ring_emit_fence - emit an fence & trap command
  *
  * @ring: amdgpu_ring pointer
- * @fence: fence to emit
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
  *
  * Write a fence and a trap command to the ring.
  */
@@ -577,29 +945,35 @@ static void uvd_v6_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq
 }
 
 /**
- * uvd_v6_0_ring_emit_hdp_flush - emit an hdp flush
+ * uvd_v6_0_enc_ring_emit_fence - emit an enc fence & trap command
  *
  * @ring: amdgpu_ring pointer
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
  *
- * Emits an hdp flush.
+ * Write enc a fence and a trap command to the ring.
  */
-static void uvd_v6_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+static void uvd_v6_0_enc_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
+			u64 seq, unsigned flags)
 {
-	amdgpu_ring_write(ring, PACKET0(mmHDP_MEM_COHERENCY_FLUSH_CNTL, 0));
-	amdgpu_ring_write(ring, 0);
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_FENCE);
+	amdgpu_ring_write(ring, addr);
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_TRAP);
 }
 
 /**
- * uvd_v6_0_ring_hdp_invalidate - emit an hdp invalidate
+ * uvd_v6_0_ring_emit_hdp_flush - skip HDP flushing
  *
  * @ring: amdgpu_ring pointer
- *
- * Emits an hdp invalidate.
  */
-static void uvd_v6_0_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
+static void uvd_v6_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
 {
-	amdgpu_ring_write(ring, PACKET0(mmHDP_DEBUG0, 0));
-	amdgpu_ring_write(ring, 1);
+	/* The firmware doesn't seem to like touching registers at this point. */
 }
 
 /**
@@ -618,11 +992,9 @@ static int uvd_v6_0_ring_test_ring(struct amdgpu_ring *ring)
 
 	WREG32(mmUVD_CONTEXT_ID, 0xCAFEDEAD);
 	r = amdgpu_ring_alloc(ring, 3);
-	if (r) {
-		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n",
-			  ring->idx, r);
+	if (r)
 		return r;
-	}
+
 	amdgpu_ring_write(ring, PACKET0(mmUVD_CONTEXT_ID, 0));
 	amdgpu_ring_write(ring, 0xDEADBEEF);
 	amdgpu_ring_commit(ring);
@@ -630,17 +1002,12 @@ static int uvd_v6_0_ring_test_ring(struct amdgpu_ring *ring)
 		tmp = RREG32(mmUVD_CONTEXT_ID);
 		if (tmp == 0xDEADBEEF)
 			break;
-		DRM_UDELAY(1);
+		udelay(1);
 	}
 
-	if (i < adev->usec_timeout) {
-		DRM_INFO("ring test on %d succeeded in %d usecs\n",
-			 ring->idx, i);
-	} else {
-		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
-			  ring->idx, tmp);
-		r = -EINVAL;
-	}
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+
 	return r;
 }
 
@@ -648,16 +1015,21 @@ static int uvd_v6_0_ring_test_ring(struct amdgpu_ring *ring)
  * uvd_v6_0_ring_emit_ib - execute indirect buffer
  *
  * @ring: amdgpu_ring pointer
+ * @job: job to retrieve vmid from
  * @ib: indirect buffer to execute
+ * @flags: unused
  *
  * Write ring commands to execute the indirect buffer
  */
 static void uvd_v6_0_ring_emit_ib(struct amdgpu_ring *ring,
+				  struct amdgpu_job *job,
 				  struct amdgpu_ib *ib,
-				  unsigned vm_id, bool ctx_switch)
+				  uint32_t flags)
 {
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+
 	amdgpu_ring_write(ring, PACKET0(mmUVD_LMI_RBC_IB_VMID, 0));
-	amdgpu_ring_write(ring, vm_id);
+	amdgpu_ring_write(ring, vmid);
 
 	amdgpu_ring_write(ring, PACKET0(mmUVD_LMI_RBC_IB_64BIT_BAR_LOW, 0));
 	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
@@ -667,36 +1039,52 @@ static void uvd_v6_0_ring_emit_ib(struct amdgpu_ring *ring,
 	amdgpu_ring_write(ring, ib->length_dw);
 }
 
-static void uvd_v6_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
-					 unsigned vm_id, uint64_t pd_addr)
+/**
+ * uvd_v6_0_enc_ring_emit_ib - enc execute indirect buffer
+ *
+ * @ring: amdgpu_ring pointer
+ * @job: job to retrive vmid from
+ * @ib: indirect buffer to execute
+ * @flags: unused
+ *
+ * Write enc ring commands to execute the indirect buffer
+ */
+static void uvd_v6_0_enc_ring_emit_ib(struct amdgpu_ring *ring,
+					struct amdgpu_job *job,
+					struct amdgpu_ib *ib,
+					uint32_t flags)
 {
-	uint32_t reg;
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
 
-	if (vm_id < 8)
-		reg = mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + vm_id;
-	else
-		reg = mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vm_id - 8;
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_IB_VM);
+	amdgpu_ring_write(ring, vmid);
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, ib->length_dw);
+}
 
+static void uvd_v6_0_ring_emit_wreg(struct amdgpu_ring *ring,
+				    uint32_t reg, uint32_t val)
+{
 	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
 	amdgpu_ring_write(ring, reg << 2);
 	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
-	amdgpu_ring_write(ring, pd_addr >> 12);
+	amdgpu_ring_write(ring, val);
 	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
 	amdgpu_ring_write(ring, 0x8);
+}
 
-	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
-	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST << 2);
-	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
-	amdgpu_ring_write(ring, 1 << vm_id);
-	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
-	amdgpu_ring_write(ring, 0x8);
+static void uvd_v6_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					unsigned vmid, uint64_t pd_addr)
+{
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
 
 	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
 	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST << 2);
 	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
 	amdgpu_ring_write(ring, 0);
 	amdgpu_ring_write(ring, PACKET0(mmUVD_GP_SCRATCH8, 0));
-	amdgpu_ring_write(ring, 1 << vm_id); /* mask */
+	amdgpu_ring_write(ring, 1 << vmid); /* mask */
 	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
 	amdgpu_ring_write(ring, 0xC);
 }
@@ -718,29 +1106,68 @@ static void uvd_v6_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
 	amdgpu_ring_write(ring, 0xE);
 }
 
-static bool uvd_v6_0_is_idle(void *handle)
+static void uvd_v6_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int i;
+
+	WARN_ON(ring->wptr % 2 || count % 2);
+
+	for (i = 0; i < count / 2; i++) {
+		amdgpu_ring_write(ring, PACKET0(mmUVD_NO_OP, 0));
+		amdgpu_ring_write(ring, 0);
+	}
+}
+
+static void uvd_v6_0_enc_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_WAIT_GE);
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, seq);
+}
+
+static void uvd_v6_0_enc_ring_insert_end(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_END);
+}
+
+static void uvd_v6_0_enc_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					    unsigned int vmid, uint64_t pd_addr)
+{
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_UPDATE_PTB);
+	amdgpu_ring_write(ring, vmid);
+	amdgpu_ring_write(ring, pd_addr >> 12);
+
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_FLUSH_TLB);
+	amdgpu_ring_write(ring, vmid);
+}
+
+static bool uvd_v6_0_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
 
 	return !(RREG32(mmSRBM_STATUS) & SRBM_STATUS__UVD_BUSY_MASK);
 }
 
-static int uvd_v6_0_wait_for_idle(void *handle)
+static int uvd_v6_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
 	unsigned i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		if (uvd_v6_0_is_idle(handle))
+		if (uvd_v6_0_is_idle(ip_block))
 			return 0;
 	}
 	return -ETIMEDOUT;
 }
 
 #define AMDGPU_UVD_STATUS_BUSY_MASK    0xfd
-static bool uvd_v6_0_check_soft_reset(void *handle)
+static bool uvd_v6_0_check_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 srbm_soft_reset = 0;
 	u32 tmp = RREG32(mmSRBM_STATUS);
 
@@ -750,33 +1177,33 @@ static bool uvd_v6_0_check_soft_reset(void *handle)
 		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_UVD, 1);
 
 	if (srbm_soft_reset) {
-		adev->uvd.srbm_soft_reset = srbm_soft_reset;
+		adev->uvd.inst->srbm_soft_reset = srbm_soft_reset;
 		return true;
 	} else {
-		adev->uvd.srbm_soft_reset = 0;
+		adev->uvd.inst->srbm_soft_reset = 0;
 		return false;
 	}
 }
 
-static int uvd_v6_0_pre_soft_reset(void *handle)
+static int uvd_v6_0_pre_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	if (!adev->uvd.srbm_soft_reset)
+	if (!adev->uvd.inst->srbm_soft_reset)
 		return 0;
 
 	uvd_v6_0_stop(adev);
 	return 0;
 }
 
-static int uvd_v6_0_soft_reset(void *handle)
+static int uvd_v6_0_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 srbm_soft_reset;
 
-	if (!adev->uvd.srbm_soft_reset)
+	if (!adev->uvd.inst->srbm_soft_reset)
 		return 0;
-	srbm_soft_reset = adev->uvd.srbm_soft_reset;
+	srbm_soft_reset = adev->uvd.inst->srbm_soft_reset;
 
 	if (srbm_soft_reset) {
 		u32 tmp;
@@ -800,11 +1227,11 @@ static int uvd_v6_0_soft_reset(void *handle)
 	return 0;
 }
 
-static int uvd_v6_0_post_soft_reset(void *handle)
+static int uvd_v6_0_post_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	if (!adev->uvd.srbm_soft_reset)
+	if (!adev->uvd.inst->srbm_soft_reset)
 		return 0;
 
 	mdelay(5);
@@ -825,8 +1252,31 @@ static int uvd_v6_0_process_interrupt(struct amdgpu_device *adev,
 				      struct amdgpu_irq_src *source,
 				      struct amdgpu_iv_entry *entry)
 {
+	bool int_handled = true;
 	DRM_DEBUG("IH: UVD TRAP\n");
-	amdgpu_fence_process(&adev->uvd.ring);
+
+	switch (entry->src_id) {
+	case 124:
+		amdgpu_fence_process(&adev->uvd.inst->ring);
+		break;
+	case 119:
+		if (likely(uvd_v6_0_enc_support(adev)))
+			amdgpu_fence_process(&adev->uvd.inst->ring_enc[0]);
+		else
+			int_handled = false;
+		break;
+	case 120:
+		if (likely(uvd_v6_0_enc_support(adev)))
+			amdgpu_fence_process(&adev->uvd.inst->ring_enc[1]);
+		else
+			int_handled = false;
+		break;
+	}
+
+	if (!int_handled)
+		DRM_ERROR("Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+
 	return 0;
 }
 
@@ -1002,15 +1452,15 @@ static void uvd_v6_0_enable_mgcg(struct amdgpu_device *adev,
 	}
 }
 
-static int uvd_v6_0_set_clockgating_state(void *handle,
+static int uvd_v6_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	bool enable = (state == AMD_CG_STATE_GATE) ? true : false;
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = (state == AMD_CG_STATE_GATE);
 
 	if (enable) {
 		/* wait for STATUS to clear */
-		if (uvd_v6_0_wait_for_idle(handle))
+		if (uvd_v6_0_wait_for_idle(ip_block))
 			return -EBUSY;
 		uvd_v6_0_enable_clock_gating(adev, true);
 		/* enable HW gates because UVD is idle */
@@ -1023,7 +1473,7 @@ static int uvd_v6_0_set_clockgating_state(void *handle,
 	return 0;
 }
 
-static int uvd_v6_0_set_powergating_state(void *handle,
+static int uvd_v6_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	/* This doesn't actually powergate the UVD block.
@@ -1033,7 +1483,7 @@ static int uvd_v6_0_set_powergating_state(void *handle,
 	 * revisit this when there is a cleaner line between
 	 * the smc and the hw blocks
 	 */
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int ret = 0;
 
 	WREG32(mmUVD_POWER_STATUS, UVD_POWER_STATUS__UVD_PG_EN_MASK);
@@ -1050,9 +1500,9 @@ out:
 	return ret;
 }
 
-static void uvd_v6_0_get_clockgating_state(void *handle, u32 *flags)
+static void uvd_v6_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int data;
 
 	mutex_lock(&adev->pm.mutex);
@@ -1079,11 +1529,11 @@ out:
 static const struct amd_ip_funcs uvd_v6_0_ip_funcs = {
 	.name = "uvd_v6_0",
 	.early_init = uvd_v6_0_early_init,
-	.late_init = NULL,
 	.sw_init = uvd_v6_0_sw_init,
 	.sw_fini = uvd_v6_0_sw_fini,
 	.hw_init = uvd_v6_0_hw_init,
 	.hw_fini = uvd_v6_0_hw_fini,
+	.prepare_suspend = uvd_v6_0_prepare_suspend,
 	.suspend = uvd_v6_0_suspend,
 	.resume = uvd_v6_0_resume,
 	.is_idle = uvd_v6_0_is_idle,
@@ -1100,43 +1550,41 @@ static const struct amd_ip_funcs uvd_v6_0_ip_funcs = {
 static const struct amdgpu_ring_funcs uvd_v6_0_ring_phys_funcs = {
 	.type = AMDGPU_RING_TYPE_UVD,
 	.align_mask = 0xf,
-	.nop = PACKET0(mmUVD_NO_OP, 0),
 	.support_64bit_ptrs = false,
+	.no_user_fence = true,
 	.get_rptr = uvd_v6_0_ring_get_rptr,
 	.get_wptr = uvd_v6_0_ring_get_wptr,
 	.set_wptr = uvd_v6_0_ring_set_wptr,
 	.parse_cs = amdgpu_uvd_ring_parse_cs,
 	.emit_frame_size =
-		2 + /* uvd_v6_0_ring_emit_hdp_flush */
-		2 + /* uvd_v6_0_ring_emit_hdp_invalidate */
+		6 + /* hdp invalidate */
 		10 + /* uvd_v6_0_ring_emit_pipeline_sync */
 		14, /* uvd_v6_0_ring_emit_fence x1 no user fence */
 	.emit_ib_size = 8, /* uvd_v6_0_ring_emit_ib */
 	.emit_ib = uvd_v6_0_ring_emit_ib,
 	.emit_fence = uvd_v6_0_ring_emit_fence,
 	.emit_hdp_flush = uvd_v6_0_ring_emit_hdp_flush,
-	.emit_hdp_invalidate = uvd_v6_0_ring_emit_hdp_invalidate,
 	.test_ring = uvd_v6_0_ring_test_ring,
 	.test_ib = amdgpu_uvd_ring_test_ib,
-	.insert_nop = amdgpu_ring_insert_nop,
+	.insert_nop = uvd_v6_0_ring_insert_nop,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
 	.begin_use = amdgpu_uvd_ring_begin_use,
 	.end_use = amdgpu_uvd_ring_end_use,
+	.emit_wreg = uvd_v6_0_ring_emit_wreg,
 };
 
 static const struct amdgpu_ring_funcs uvd_v6_0_ring_vm_funcs = {
 	.type = AMDGPU_RING_TYPE_UVD,
 	.align_mask = 0xf,
-	.nop = PACKET0(mmUVD_NO_OP, 0),
 	.support_64bit_ptrs = false,
+	.no_user_fence = true,
 	.get_rptr = uvd_v6_0_ring_get_rptr,
 	.get_wptr = uvd_v6_0_ring_get_wptr,
 	.set_wptr = uvd_v6_0_ring_set_wptr,
 	.emit_frame_size =
-		2 + /* uvd_v6_0_ring_emit_hdp_flush */
-		2 + /* uvd_v6_0_ring_emit_hdp_invalidate */
+		6 + /* hdp invalidate */
 		10 + /* uvd_v6_0_ring_emit_pipeline_sync */
-		20 + /* uvd_v6_0_ring_emit_vm_flush */
+		VI_FLUSH_GPU_TLB_NUM_WREG * 6 + 8 + /* uvd_v6_0_ring_emit_vm_flush */
 		14 + 14, /* uvd_v6_0_ring_emit_fence x2 vm fence */
 	.emit_ib_size = 8, /* uvd_v6_0_ring_emit_ib */
 	.emit_ib = uvd_v6_0_ring_emit_ib,
@@ -1144,10 +1592,38 @@ static const struct amdgpu_ring_funcs uvd_v6_0_ring_vm_funcs = {
 	.emit_vm_flush = uvd_v6_0_ring_emit_vm_flush,
 	.emit_pipeline_sync = uvd_v6_0_ring_emit_pipeline_sync,
 	.emit_hdp_flush = uvd_v6_0_ring_emit_hdp_flush,
-	.emit_hdp_invalidate = uvd_v6_0_ring_emit_hdp_invalidate,
 	.test_ring = uvd_v6_0_ring_test_ring,
 	.test_ib = amdgpu_uvd_ring_test_ib,
+	.insert_nop = uvd_v6_0_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_uvd_ring_begin_use,
+	.end_use = amdgpu_uvd_ring_end_use,
+	.emit_wreg = uvd_v6_0_ring_emit_wreg,
+};
+
+static const struct amdgpu_ring_funcs uvd_v6_0_enc_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_UVD_ENC,
+	.align_mask = 0x3f,
+	.nop = HEVC_ENC_CMD_NO_OP,
+	.support_64bit_ptrs = false,
+	.no_user_fence = true,
+	.get_rptr = uvd_v6_0_enc_ring_get_rptr,
+	.get_wptr = uvd_v6_0_enc_ring_get_wptr,
+	.set_wptr = uvd_v6_0_enc_ring_set_wptr,
+	.emit_frame_size =
+		4 + /* uvd_v6_0_enc_ring_emit_pipeline_sync */
+		5 + /* uvd_v6_0_enc_ring_emit_vm_flush */
+		5 + 5 + /* uvd_v6_0_enc_ring_emit_fence x2 vm fence */
+		1, /* uvd_v6_0_enc_ring_insert_end */
+	.emit_ib_size = 5, /* uvd_v6_0_enc_ring_emit_ib */
+	.emit_ib = uvd_v6_0_enc_ring_emit_ib,
+	.emit_fence = uvd_v6_0_enc_ring_emit_fence,
+	.emit_vm_flush = uvd_v6_0_enc_ring_emit_vm_flush,
+	.emit_pipeline_sync = uvd_v6_0_enc_ring_emit_pipeline_sync,
+	.test_ring = uvd_v6_0_enc_ring_test_ring,
+	.test_ib = uvd_v6_0_enc_ring_test_ib,
 	.insert_nop = amdgpu_ring_insert_nop,
+	.insert_end = uvd_v6_0_enc_ring_insert_end,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
 	.begin_use = amdgpu_uvd_ring_begin_use,
 	.end_use = amdgpu_uvd_ring_end_use,
@@ -1156,14 +1632,24 @@ static const struct amdgpu_ring_funcs uvd_v6_0_ring_vm_funcs = {
 static void uvd_v6_0_set_ring_funcs(struct amdgpu_device *adev)
 {
 	if (adev->asic_type >= CHIP_POLARIS10) {
-		adev->uvd.ring.funcs = &uvd_v6_0_ring_vm_funcs;
+		adev->uvd.inst->ring.funcs = &uvd_v6_0_ring_vm_funcs;
 		DRM_INFO("UVD is enabled in VM mode\n");
 	} else {
-		adev->uvd.ring.funcs = &uvd_v6_0_ring_phys_funcs;
+		adev->uvd.inst->ring.funcs = &uvd_v6_0_ring_phys_funcs;
 		DRM_INFO("UVD is enabled in physical mode\n");
 	}
 }
 
+static void uvd_v6_0_set_enc_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->uvd.num_enc_rings; ++i)
+		adev->uvd.inst->ring_enc[i].funcs = &uvd_v6_0_enc_ring_vm_funcs;
+
+	DRM_INFO("UVD ENC is enabled in VM mode\n");
+}
+
 static const struct amdgpu_irq_src_funcs uvd_v6_0_irq_funcs = {
 	.set = uvd_v6_0_set_interrupt_state,
 	.process = uvd_v6_0_process_interrupt,
@@ -1171,8 +1657,12 @@ static const struct amdgpu_irq_src_funcs uvd_v6_0_irq_funcs = {
 
 static void uvd_v6_0_set_irq_funcs(struct amdgpu_device *adev)
 {
-	adev->uvd.irq.num_types = 1;
-	adev->uvd.irq.funcs = &uvd_v6_0_irq_funcs;
+	if (uvd_v6_0_enc_support(adev))
+		adev->uvd.inst->irq.num_types = adev->uvd.num_enc_rings + 1;
+	else
+		adev->uvd.inst->irq.num_types = 1;
+
+	adev->uvd.inst->irq.funcs = &uvd_v6_0_irq_funcs;
 }
 
 const struct amdgpu_ip_block_version uvd_v6_0_ip_block =
diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c
index 23a85750edd6..1f8866f3f63c 100644
--- a/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c
@@ -22,23 +22,32 @@
  */
 
 #include <linux/firmware.h>
-#include <drm/drmP.h>
+
 #include "amdgpu.h"
 #include "amdgpu_uvd.h"
+#include "amdgpu_cs.h"
+#include "soc15.h"
 #include "soc15d.h"
 #include "soc15_common.h"
 #include "mmsch_v1_0.h"
 
-#include "vega10/soc15ip.h"
-#include "vega10/UVD/uvd_7_0_offset.h"
-#include "vega10/UVD/uvd_7_0_sh_mask.h"
-#include "vega10/VCE/vce_4_0_offset.h"
-#include "vega10/VCE/vce_4_0_default.h"
-#include "vega10/VCE/vce_4_0_sh_mask.h"
-#include "vega10/NBIF/nbif_6_1_offset.h"
-#include "vega10/HDP/hdp_4_0_offset.h"
-#include "vega10/MMHUB/mmhub_1_0_offset.h"
-#include "vega10/MMHUB/mmhub_1_0_sh_mask.h"
+#include "uvd/uvd_7_0_offset.h"
+#include "uvd/uvd_7_0_sh_mask.h"
+#include "vce/vce_4_0_offset.h"
+#include "vce/vce_4_0_default.h"
+#include "vce/vce_4_0_sh_mask.h"
+#include "nbif/nbif_6_1_offset.h"
+#include "mmhub/mmhub_1_0_offset.h"
+#include "mmhub/mmhub_1_0_sh_mask.h"
+#include "ivsrcid/uvd/irqsrcs_uvd_7_0.h"
+
+#define mmUVD_PG0_CC_UVD_HARVESTING                                                                    0x00c7
+#define mmUVD_PG0_CC_UVD_HARVESTING_BASE_IDX                                                           1
+//UVD_PG0_CC_UVD_HARVESTING
+#define UVD_PG0_CC_UVD_HARVESTING__UVD_DISABLE__SHIFT                                                         0x1
+#define UVD_PG0_CC_UVD_HARVESTING__UVD_DISABLE_MASK                                                           0x00000002L
+
+#define UVD7_MAX_HW_INSTANCES_VEGA20			2
 
 static void uvd_v7_0_set_ring_funcs(struct amdgpu_device *adev);
 static void uvd_v7_0_set_enc_ring_funcs(struct amdgpu_device *adev);
@@ -47,6 +56,11 @@ static int uvd_v7_0_start(struct amdgpu_device *adev);
 static void uvd_v7_0_stop(struct amdgpu_device *adev);
 static int uvd_v7_0_sriov_start(struct amdgpu_device *adev);
 
+static int amdgpu_ih_clientid_uvds[] = {
+	SOC15_IH_CLIENTID_UVD,
+	SOC15_IH_CLIENTID_UVD1
+};
+
 /**
  * uvd_v7_0_ring_get_rptr - get read pointer
  *
@@ -58,7 +72,7 @@ static uint64_t uvd_v7_0_ring_get_rptr(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
 
-	return RREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR);
+	return RREG32_SOC15(UVD, ring->me, mmUVD_RBC_RB_RPTR);
 }
 
 /**
@@ -72,10 +86,10 @@ static uint64_t uvd_v7_0_enc_ring_get_rptr(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
 
-	if (ring == &adev->uvd.ring_enc[0])
-		return RREG32_SOC15(UVD, 0, mmUVD_RB_RPTR);
+	if (ring == &adev->uvd.inst[ring->me].ring_enc[0])
+		return RREG32_SOC15(UVD, ring->me, mmUVD_RB_RPTR);
 	else
-		return RREG32_SOC15(UVD, 0, mmUVD_RB_RPTR2);
+		return RREG32_SOC15(UVD, ring->me, mmUVD_RB_RPTR2);
 }
 
 /**
@@ -89,7 +103,7 @@ static uint64_t uvd_v7_0_ring_get_wptr(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
 
-	return RREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR);
+	return RREG32_SOC15(UVD, ring->me, mmUVD_RBC_RB_WPTR);
 }
 
 /**
@@ -104,12 +118,12 @@ static uint64_t uvd_v7_0_enc_ring_get_wptr(struct amdgpu_ring *ring)
 	struct amdgpu_device *adev = ring->adev;
 
 	if (ring->use_doorbell)
-		return adev->wb.wb[ring->wptr_offs];
+		return *ring->wptr_cpu_addr;
 
-	if (ring == &adev->uvd.ring_enc[0])
-		return RREG32_SOC15(UVD, 0, mmUVD_RB_WPTR);
+	if (ring == &adev->uvd.inst[ring->me].ring_enc[0])
+		return RREG32_SOC15(UVD, ring->me, mmUVD_RB_WPTR);
 	else
-		return RREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2);
+		return RREG32_SOC15(UVD, ring->me, mmUVD_RB_WPTR2);
 }
 
 /**
@@ -123,7 +137,7 @@ static void uvd_v7_0_ring_set_wptr(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
 
-	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr));
+	WREG32_SOC15(UVD, ring->me, mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr));
 }
 
 /**
@@ -139,16 +153,16 @@ static void uvd_v7_0_enc_ring_set_wptr(struct amdgpu_ring *ring)
 
 	if (ring->use_doorbell) {
 		/* XXX check if swapping is necessary on BE */
-		adev->wb.wb[ring->wptr_offs] = lower_32_bits(ring->wptr);
+		*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
 		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
 		return;
 	}
 
-	if (ring == &adev->uvd.ring_enc[0])
-		WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR,
+	if (ring == &adev->uvd.inst[ring->me].ring_enc[0])
+		WREG32_SOC15(UVD, ring->me, mmUVD_RB_WPTR,
 			lower_32_bits(ring->wptr));
 	else
-		WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2,
+		WREG32_SOC15(UVD, ring->me, mmUVD_RB_WPTR2,
 			lower_32_bits(ring->wptr));
 }
 
@@ -161,7 +175,7 @@ static void uvd_v7_0_enc_ring_set_wptr(struct amdgpu_ring *ring)
 static int uvd_v7_0_enc_ring_test_ring(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
-	uint32_t rptr = amdgpu_ring_get_rptr(ring);
+	uint32_t rptr;
 	unsigned i;
 	int r;
 
@@ -169,28 +183,22 @@ static int uvd_v7_0_enc_ring_test_ring(struct amdgpu_ring *ring)
 		return 0;
 
 	r = amdgpu_ring_alloc(ring, 16);
-	if (r) {
-		DRM_ERROR("amdgpu: uvd enc failed to lock ring %d (%d).\n",
-			  ring->idx, r);
+	if (r)
 		return r;
-	}
+
+	rptr = amdgpu_ring_get_rptr(ring);
+
 	amdgpu_ring_write(ring, HEVC_ENC_CMD_END);
 	amdgpu_ring_commit(ring);
 
 	for (i = 0; i < adev->usec_timeout; i++) {
 		if (amdgpu_ring_get_rptr(ring) != rptr)
 			break;
-		DRM_UDELAY(1);
+		udelay(1);
 	}
 
-	if (i < adev->usec_timeout) {
-		DRM_INFO("ring test on %d succeeded in %d usecs\n",
-			 ring->idx, i);
-	} else {
-		DRM_ERROR("amdgpu: ring %d test failed\n",
-			  ring->idx);
+	if (i >= adev->usec_timeout)
 		r = -ETIMEDOUT;
-	}
 
 	return r;
 }
@@ -198,37 +206,40 @@ static int uvd_v7_0_enc_ring_test_ring(struct amdgpu_ring *ring)
 /**
  * uvd_v7_0_enc_get_create_msg - generate a UVD ENC create msg
  *
- * @adev: amdgpu_device pointer
  * @ring: ring we should submit the msg to
  * @handle: session handle to use
+ * @bo: amdgpu object for which we query the offset
  * @fence: optional fence to return
  *
  * Open up a stream for HW test
  */
-static int uvd_v7_0_enc_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
+static int uvd_v7_0_enc_get_create_msg(struct amdgpu_ring *ring, u32 handle,
+				       struct amdgpu_bo *bo,
 				       struct dma_fence **fence)
 {
 	const unsigned ib_size_dw = 16;
 	struct amdgpu_job *job;
 	struct amdgpu_ib *ib;
 	struct dma_fence *f = NULL;
-	uint64_t dummy;
+	uint64_t addr;
 	int i, r;
 
-	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4, &job);
+	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL, ib_size_dw * 4,
+				     AMDGPU_IB_POOL_DIRECT, &job,
+				     AMDGPU_KERNEL_JOB_ID_VCN_RING_TEST);
 	if (r)
 		return r;
 
 	ib = &job->ibs[0];
-	dummy = ib->gpu_addr + 1024;
+	addr = amdgpu_bo_gpu_offset(bo);
 
 	ib->length_dw = 0;
 	ib->ptr[ib->length_dw++] = 0x00000018;
 	ib->ptr[ib->length_dw++] = 0x00000001; /* session info */
 	ib->ptr[ib->length_dw++] = handle;
 	ib->ptr[ib->length_dw++] = 0x00000000;
-	ib->ptr[ib->length_dw++] = upper_32_bits(dummy);
-	ib->ptr[ib->length_dw++] = dummy;
+	ib->ptr[ib->length_dw++] = upper_32_bits(addr);
+	ib->ptr[ib->length_dw++] = addr;
 
 	ib->ptr[ib->length_dw++] = 0x00000014;
 	ib->ptr[ib->length_dw++] = 0x00000002; /* task info */
@@ -242,12 +253,10 @@ static int uvd_v7_0_enc_get_create_msg(struct amdgpu_ring *ring, uint32_t handle
 	for (i = ib->length_dw; i < ib_size_dw; ++i)
 		ib->ptr[i] = 0x0;
 
-	r = amdgpu_ib_schedule(ring, 1, ib, NULL, &f);
-	job->fence = dma_fence_get(f);
+	r = amdgpu_job_submit_direct(job, ring, &f);
 	if (r)
 		goto err;
 
-	amdgpu_job_free(job);
 	if (fence)
 		*fence = dma_fence_get(f);
 	dma_fence_put(f);
@@ -261,37 +270,40 @@ err:
 /**
  * uvd_v7_0_enc_get_destroy_msg - generate a UVD ENC destroy msg
  *
- * @adev: amdgpu_device pointer
  * @ring: ring we should submit the msg to
  * @handle: session handle to use
+ * @bo: amdgpu object for which we query the offset
  * @fence: optional fence to return
  *
  * Close up a stream for HW test or if userspace failed to do so
  */
-int uvd_v7_0_enc_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
-				 bool direct, struct dma_fence **fence)
+static int uvd_v7_0_enc_get_destroy_msg(struct amdgpu_ring *ring, u32 handle,
+					struct amdgpu_bo *bo,
+					struct dma_fence **fence)
 {
 	const unsigned ib_size_dw = 16;
 	struct amdgpu_job *job;
 	struct amdgpu_ib *ib;
 	struct dma_fence *f = NULL;
-	uint64_t dummy;
+	uint64_t addr;
 	int i, r;
 
-	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4, &job);
+	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL, ib_size_dw * 4,
+				     AMDGPU_IB_POOL_DIRECT, &job,
+				     AMDGPU_KERNEL_JOB_ID_VCN_RING_TEST);
 	if (r)
 		return r;
 
 	ib = &job->ibs[0];
-	dummy = ib->gpu_addr + 1024;
+	addr = amdgpu_bo_gpu_offset(bo);
 
 	ib->length_dw = 0;
 	ib->ptr[ib->length_dw++] = 0x00000018;
 	ib->ptr[ib->length_dw++] = 0x00000001;
 	ib->ptr[ib->length_dw++] = handle;
 	ib->ptr[ib->length_dw++] = 0x00000000;
-	ib->ptr[ib->length_dw++] = upper_32_bits(dummy);
-	ib->ptr[ib->length_dw++] = dummy;
+	ib->ptr[ib->length_dw++] = upper_32_bits(addr);
+	ib->ptr[ib->length_dw++] = addr;
 
 	ib->ptr[ib->length_dw++] = 0x00000014;
 	ib->ptr[ib->length_dw++] = 0x00000002;
@@ -305,19 +317,9 @@ int uvd_v7_0_enc_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
 	for (i = ib->length_dw; i < ib_size_dw; ++i)
 		ib->ptr[i] = 0x0;
 
-	if (direct) {
-		r = amdgpu_ib_schedule(ring, 1, ib, NULL, &f);
-		job->fence = dma_fence_get(f);
-		if (r)
-			goto err;
-
-		amdgpu_job_free(job);
-	} else {
-		r = amdgpu_job_submit(job, ring, &ring->adev->vce.entity,
-				      AMDGPU_FENCE_OWNER_UNDEFINED, &f);
-		if (r)
-			goto err;
-	}
+	r = amdgpu_job_submit_direct(job, ring, &f);
+	if (r)
+		goto err;
 
 	if (fence)
 		*fence = dma_fence_get(f);
@@ -333,43 +335,56 @@ err:
  * uvd_v7_0_enc_ring_test_ib - test if UVD ENC IBs are working
  *
  * @ring: the engine to test on
+ * @timeout: timeout value in jiffies, or MAX_SCHEDULE_TIMEOUT
  *
  */
 static int uvd_v7_0_enc_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 {
 	struct dma_fence *fence = NULL;
+	struct amdgpu_bo *bo = ring->adev->uvd.ib_bo;
 	long r;
 
-	r = uvd_v7_0_enc_get_create_msg(ring, 1, NULL);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get create msg (%ld).\n", r);
+	r = uvd_v7_0_enc_get_create_msg(ring, 1, bo, NULL);
+	if (r)
 		goto error;
-	}
 
-	r = uvd_v7_0_enc_get_destroy_msg(ring, 1, true, &fence);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get destroy ib (%ld).\n", r);
+	r = uvd_v7_0_enc_get_destroy_msg(ring, 1, bo, &fence);
+	if (r)
 		goto error;
-	}
 
 	r = dma_fence_wait_timeout(fence, false, timeout);
-	if (r == 0) {
-		DRM_ERROR("amdgpu: IB test timed out.\n");
+	if (r == 0)
 		r = -ETIMEDOUT;
-	} else if (r < 0) {
-		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
-	} else {
-		DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
+	else if (r > 0)
 		r = 0;
-	}
+
 error:
 	dma_fence_put(fence);
 	return r;
 }
 
-static int uvd_v7_0_early_init(void *handle)
+static int uvd_v7_0_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (adev->asic_type == CHIP_VEGA20) {
+		u32 harvest;
+		int i;
+
+		adev->uvd.num_uvd_inst = UVD7_MAX_HW_INSTANCES_VEGA20;
+		for (i = 0; i < adev->uvd.num_uvd_inst; i++) {
+			harvest = RREG32_SOC15(UVD, i, mmUVD_PG0_CC_UVD_HARVESTING);
+			if (harvest & UVD_PG0_CC_UVD_HARVESTING__UVD_DISABLE_MASK) {
+				adev->uvd.harvest_config |= 1 << i;
+			}
+		}
+		if (adev->uvd.harvest_config == (AMDGPU_UVD_HARVEST_UVD0 |
+						 AMDGPU_UVD_HARVEST_UVD1))
+			/* both instances are harvested, disable the block */
+			return -ENOENT;
+	} else {
+		adev->uvd.num_uvd_inst = 1;
+	}
 
 	if (amdgpu_sriov_vf(adev))
 		adev->uvd.num_enc_rings = 1;
@@ -382,23 +397,27 @@ static int uvd_v7_0_early_init(void *handle)
 	return 0;
 }
 
-static int uvd_v7_0_sw_init(void *handle)
+static int uvd_v7_0_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_ring *ring;
-	struct amd_sched_rq *rq;
-	int i, r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	/* UVD TRAP */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_UVD, 124, &adev->uvd.irq);
-	if (r)
-		return r;
+	int i, j, r;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	/* UVD ENC TRAP */
-	for (i = 0; i < adev->uvd.num_enc_rings; ++i) {
-		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_UVD, i + 119, &adev->uvd.irq);
+	for (j = 0; j < adev->uvd.num_uvd_inst; j++) {
+		if (adev->uvd.harvest_config & (1 << j))
+			continue;
+		/* UVD TRAP */
+		r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_uvds[j], UVD_7_0__SRCID__UVD_SYSTEM_MESSAGE_INTERRUPT, &adev->uvd.inst[j].irq);
 		if (r)
 			return r;
+
+		/* UVD ENC TRAP */
+		for (i = 0; i < adev->uvd.num_enc_rings; ++i) {
+			r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_uvds[j], i + UVD_7_0__SRCID__UVD_ENC_GEN_PURP, &adev->uvd.inst[j].irq);
+			if (r)
+				return r;
+		}
 	}
 
 	r = amdgpu_uvd_sw_init(adev);
@@ -412,47 +431,56 @@ static int uvd_v7_0_sw_init(void *handle)
 		adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].fw = adev->uvd.fw;
 		adev->firmware.fw_size +=
 			ALIGN(le32_to_cpu(hdr->ucode_size_bytes), PAGE_SIZE);
+
+		if (adev->uvd.num_uvd_inst == UVD7_MAX_HW_INSTANCES_VEGA20) {
+			adev->firmware.ucode[AMDGPU_UCODE_ID_UVD1].ucode_id = AMDGPU_UCODE_ID_UVD1;
+			adev->firmware.ucode[AMDGPU_UCODE_ID_UVD1].fw = adev->uvd.fw;
+			adev->firmware.fw_size +=
+				ALIGN(le32_to_cpu(hdr->ucode_size_bytes), PAGE_SIZE);
+		}
 		DRM_INFO("PSP loading UVD firmware\n");
 	}
 
-	ring = &adev->uvd.ring_enc[0];
-	rq = &ring->sched.sched_rq[AMD_SCHED_PRIORITY_NORMAL];
-	r = amd_sched_entity_init(&ring->sched, &adev->uvd.entity_enc,
-				  rq, amdgpu_sched_jobs);
-	if (r) {
-		DRM_ERROR("Failed setting up UVD ENC run queue.\n");
-		return r;
+	for (j = 0; j < adev->uvd.num_uvd_inst; j++) {
+		if (adev->uvd.harvest_config & (1 << j))
+			continue;
+		if (!amdgpu_sriov_vf(adev)) {
+			ring = &adev->uvd.inst[j].ring;
+			ring->vm_hub = AMDGPU_MMHUB0(0);
+			sprintf(ring->name, "uvd_%d", ring->me);
+			r = amdgpu_ring_init(adev, ring, 512,
+					     &adev->uvd.inst[j].irq, 0,
+					     AMDGPU_RING_PRIO_DEFAULT, NULL);
+			if (r)
+				return r;
+		}
+
+		for (i = 0; i < adev->uvd.num_enc_rings; ++i) {
+			ring = &adev->uvd.inst[j].ring_enc[i];
+			ring->vm_hub = AMDGPU_MMHUB0(0);
+			sprintf(ring->name, "uvd_enc_%d.%d", ring->me, i);
+			if (amdgpu_sriov_vf(adev)) {
+				ring->use_doorbell = true;
+
+				/* currently only use the first enconding ring for
+				 * sriov, so set unused location for other unused rings.
+				 */
+				if (i == 0)
+					ring->doorbell_index = adev->doorbell_index.uvd_vce.uvd_ring0_1 * 2;
+				else
+					ring->doorbell_index = adev->doorbell_index.uvd_vce.uvd_ring2_3 * 2 + 1;
+			}
+			r = amdgpu_ring_init(adev, ring, 512,
+					     &adev->uvd.inst[j].irq, 0,
+					     AMDGPU_RING_PRIO_DEFAULT, NULL);
+			if (r)
+				return r;
+		}
 	}
 
 	r = amdgpu_uvd_resume(adev);
 	if (r)
 		return r;
-	if (!amdgpu_sriov_vf(adev)) {
-		ring = &adev->uvd.ring;
-		sprintf(ring->name, "uvd");
-		r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.irq, 0);
-		if (r)
-			return r;
-	}
-
-	for (i = 0; i < adev->uvd.num_enc_rings; ++i) {
-		ring = &adev->uvd.ring_enc[i];
-		sprintf(ring->name, "uvd_enc%d", i);
-		if (amdgpu_sriov_vf(adev)) {
-			ring->use_doorbell = true;
-
-			/* currently only use the first enconding ring for
-			 * sriov, so set unused location for other unused rings.
-			 */
-			if (i == 0)
-				ring->doorbell_index = AMDGPU_DOORBELL64_UVD_RING0_1 * 2;
-			else
-				ring->doorbell_index = AMDGPU_DOORBELL64_UVD_RING2_3 * 2 + 1;
-		}
-		r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.irq, 0);
-		if (r)
-			return r;
-	}
 
 	r = amdgpu_virt_alloc_mm_table(adev);
 	if (r)
@@ -461,10 +489,10 @@ static int uvd_v7_0_sw_init(void *handle)
 	return r;
 }
 
-static int uvd_v7_0_sw_fini(void *handle)
+static int uvd_v7_0_sw_fini(struct amdgpu_ip_block *ip_block)
 {
-	int i, r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int i, j, r;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	amdgpu_virt_free_mm_table(adev);
 
@@ -472,27 +500,28 @@ static int uvd_v7_0_sw_fini(void *handle)
 	if (r)
 		return r;
 
-	amd_sched_entity_fini(&adev->uvd.ring_enc[0].sched, &adev->uvd.entity_enc);
-
-	for (i = 0; i < adev->uvd.num_enc_rings; ++i)
-		amdgpu_ring_fini(&adev->uvd.ring_enc[i]);
-
+	for (j = 0; j < adev->uvd.num_uvd_inst; ++j) {
+		if (adev->uvd.harvest_config & (1 << j))
+			continue;
+		for (i = 0; i < adev->uvd.num_enc_rings; ++i)
+			amdgpu_ring_fini(&adev->uvd.inst[j].ring_enc[i]);
+	}
 	return amdgpu_uvd_sw_fini(adev);
 }
 
 /**
  * uvd_v7_0_hw_init - start and test UVD block
  *
- * @adev: amdgpu_device pointer
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
  *
  * Initialize the hardware, boot up the VCPU and do some testing
  */
-static int uvd_v7_0_hw_init(void *handle)
+static int uvd_v7_0_hw_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	struct amdgpu_ring *ring = &adev->uvd.ring;
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
 	uint32_t tmp;
-	int i, r;
+	int i, j, r;
 
 	if (amdgpu_sriov_vf(adev))
 		r = uvd_v7_0_sriov_start(adev);
@@ -501,57 +530,56 @@ static int uvd_v7_0_hw_init(void *handle)
 	if (r)
 		goto done;
 
-	if (!amdgpu_sriov_vf(adev)) {
-		ring->ready = true;
-		r = amdgpu_ring_test_ring(ring);
-		if (r) {
-			ring->ready = false;
-			goto done;
+	for (j = 0; j < adev->uvd.num_uvd_inst; ++j) {
+		if (adev->uvd.harvest_config & (1 << j))
+			continue;
+		ring = &adev->uvd.inst[j].ring;
+
+		if (!amdgpu_sriov_vf(adev)) {
+			r = amdgpu_ring_test_helper(ring);
+			if (r)
+				goto done;
+
+			r = amdgpu_ring_alloc(ring, 10);
+			if (r) {
+				DRM_ERROR("amdgpu: (%d)ring failed to lock UVD ring (%d).\n", j, r);
+				goto done;
+			}
+
+			tmp = PACKET0(SOC15_REG_OFFSET(UVD, j,
+				mmUVD_SEMA_WAIT_FAULT_TIMEOUT_CNTL), 0);
+			amdgpu_ring_write(ring, tmp);
+			amdgpu_ring_write(ring, 0xFFFFF);
+
+			tmp = PACKET0(SOC15_REG_OFFSET(UVD, j,
+				mmUVD_SEMA_WAIT_INCOMPLETE_TIMEOUT_CNTL), 0);
+			amdgpu_ring_write(ring, tmp);
+			amdgpu_ring_write(ring, 0xFFFFF);
+
+			tmp = PACKET0(SOC15_REG_OFFSET(UVD, j,
+				mmUVD_SEMA_SIGNAL_INCOMPLETE_TIMEOUT_CNTL), 0);
+			amdgpu_ring_write(ring, tmp);
+			amdgpu_ring_write(ring, 0xFFFFF);
+
+			/* Clear timeout status bits */
+			amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(UVD, j,
+				mmUVD_SEMA_TIMEOUT_STATUS), 0));
+			amdgpu_ring_write(ring, 0x8);
+
+			amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(UVD, j,
+				mmUVD_SEMA_CNTL), 0));
+			amdgpu_ring_write(ring, 3);
+
+			amdgpu_ring_commit(ring);
 		}
 
-		r = amdgpu_ring_alloc(ring, 10);
-		if (r) {
-			DRM_ERROR("amdgpu: ring failed to lock UVD ring (%d).\n", r);
-			goto done;
-		}
-
-		tmp = PACKET0(SOC15_REG_OFFSET(UVD, 0,
-			mmUVD_SEMA_WAIT_FAULT_TIMEOUT_CNTL), 0);
-		amdgpu_ring_write(ring, tmp);
-		amdgpu_ring_write(ring, 0xFFFFF);
-
-		tmp = PACKET0(SOC15_REG_OFFSET(UVD, 0,
-			mmUVD_SEMA_WAIT_INCOMPLETE_TIMEOUT_CNTL), 0);
-		amdgpu_ring_write(ring, tmp);
-		amdgpu_ring_write(ring, 0xFFFFF);
-
-		tmp = PACKET0(SOC15_REG_OFFSET(UVD, 0,
-			mmUVD_SEMA_SIGNAL_INCOMPLETE_TIMEOUT_CNTL), 0);
-		amdgpu_ring_write(ring, tmp);
-		amdgpu_ring_write(ring, 0xFFFFF);
-
-		/* Clear timeout status bits */
-		amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(UVD, 0,
-			mmUVD_SEMA_TIMEOUT_STATUS), 0));
-		amdgpu_ring_write(ring, 0x8);
-
-		amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(UVD, 0,
-			mmUVD_SEMA_CNTL), 0));
-		amdgpu_ring_write(ring, 3);
-
-		amdgpu_ring_commit(ring);
-	}
-
-	for (i = 0; i < adev->uvd.num_enc_rings; ++i) {
-		ring = &adev->uvd.ring_enc[i];
-		ring->ready = true;
-		r = amdgpu_ring_test_ring(ring);
-		if (r) {
-			ring->ready = false;
-			goto done;
+		for (i = 0; i < adev->uvd.num_enc_rings; ++i) {
+			ring = &adev->uvd.inst[j].ring_enc[i];
+			r = amdgpu_ring_test_helper(ring);
+			if (r)
+				goto done;
 		}
 	}
-
 done:
 	if (!r)
 		DRM_INFO("UVD and UVD ENC initialized successfully.\n");
@@ -562,14 +590,15 @@ done:
 /**
  * uvd_v7_0_hw_fini - stop the hardware block
  *
- * @adev: amdgpu_device pointer
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
  *
  * Stop the UVD block, mark ring as not ready any more
  */
-static int uvd_v7_0_hw_fini(void *handle)
+static int uvd_v7_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	struct amdgpu_ring *ring = &adev->uvd.ring;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	cancel_delayed_work_sync(&adev->uvd.idle_work);
 
 	if (!amdgpu_sriov_vf(adev))
 		uvd_v7_0_stop(adev);
@@ -578,39 +607,61 @@ static int uvd_v7_0_hw_fini(void *handle)
 		DRM_DEBUG("For SRIOV client, shouldn't do anything.\n");
 	}
 
-	ring->ready = false;
-
 	return 0;
 }
 
-static int uvd_v7_0_suspend(void *handle)
+static int uvd_v7_0_prepare_suspend(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	return amdgpu_uvd_prepare_suspend(adev);
+}
+
+static int uvd_v7_0_suspend(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	/*
+	 * Proper cleanups before halting the HW engine:
+	 *   - cancel the delayed idle work
+	 *   - enable powergating
+	 *   - enable clockgating
+	 *   - disable dpm
+	 *
+	 * TODO: to align with the VCN implementation, move the
+	 * jobs for clockgating/powergating/dpm setting to
+	 * ->set_powergating_state().
+	 */
+	cancel_delayed_work_sync(&adev->uvd.idle_work);
 
-	r = uvd_v7_0_hw_fini(adev);
+	if (adev->pm.dpm_enabled) {
+		amdgpu_dpm_enable_uvd(adev, false);
+	} else {
+		amdgpu_asic_set_uvd_clocks(adev, 0, 0);
+		/* shutdown the UVD block */
+		amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+						       AMD_PG_STATE_GATE);
+		amdgpu_device_ip_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+						       AMD_CG_STATE_GATE);
+	}
+
+	r = uvd_v7_0_hw_fini(ip_block);
 	if (r)
 		return r;
 
-	/* Skip this for APU for now */
-	if (!(adev->flags & AMD_IS_APU))
-		r = amdgpu_uvd_suspend(adev);
-
-	return r;
+	return amdgpu_uvd_suspend(adev);
 }
 
-static int uvd_v7_0_resume(void *handle)
+static int uvd_v7_0_resume(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	/* Skip this for APU for now */
-	if (!(adev->flags & AMD_IS_APU)) {
-		r = amdgpu_uvd_resume(adev);
-		if (r)
-			return r;
-	}
-	return uvd_v7_0_hw_init(adev);
+	r = amdgpu_uvd_resume(ip_block->adev);
+	if (r)
+		return r;
+
+	return uvd_v7_0_hw_init(ip_block);
 }
 
 /**
@@ -622,50 +673,60 @@ static int uvd_v7_0_resume(void *handle)
  */
 static void uvd_v7_0_mc_resume(struct amdgpu_device *adev)
 {
-	uint32_t size = AMDGPU_GPU_PAGE_ALIGN(adev->uvd.fw->size + 4);
+	uint32_t size = AMDGPU_UVD_FIRMWARE_SIZE(adev);
 	uint32_t offset;
+	int i;
 
-	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
-		WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
-			lower_32_bits(adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].mc_addr));
-		WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
-			upper_32_bits(adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].mc_addr));
-		offset = 0;
-	} else {
-		WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
-			lower_32_bits(adev->uvd.gpu_addr));
-		WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
-			upper_32_bits(adev->uvd.gpu_addr));
-		offset = size;
-	}
+	for (i = 0; i < adev->uvd.num_uvd_inst; ++i) {
+		if (adev->uvd.harvest_config & (1 << i))
+			continue;
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+			WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+				i == 0 ?
+				adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].tmr_mc_addr_lo :
+				adev->firmware.ucode[AMDGPU_UCODE_ID_UVD1].tmr_mc_addr_lo);
+			WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+				i == 0 ?
+				adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].tmr_mc_addr_hi :
+				adev->firmware.ucode[AMDGPU_UCODE_ID_UVD1].tmr_mc_addr_hi);
+			WREG32_SOC15(UVD, i, mmUVD_VCPU_CACHE_OFFSET0, 0);
+			offset = 0;
+		} else {
+			WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+				lower_32_bits(adev->uvd.inst[i].gpu_addr));
+			WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+				upper_32_bits(adev->uvd.inst[i].gpu_addr));
+			offset = size;
+			WREG32_SOC15(UVD, i, mmUVD_VCPU_CACHE_OFFSET0,
+					AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
+		}
 
-	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_OFFSET0,
-				AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
-	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_SIZE0, size);
-
-	WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW,
-			lower_32_bits(adev->uvd.gpu_addr + offset));
-	WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH,
-			upper_32_bits(adev->uvd.gpu_addr + offset));
-	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_OFFSET1, (1 << 21));
-	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_SIZE1, AMDGPU_UVD_HEAP_SIZE);
-
-	WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW,
-			lower_32_bits(adev->uvd.gpu_addr + offset + AMDGPU_UVD_HEAP_SIZE));
-	WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH,
-			upper_32_bits(adev->uvd.gpu_addr + offset + AMDGPU_UVD_HEAP_SIZE));
-	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_OFFSET2, (2 << 21));
-	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_SIZE2,
-			AMDGPU_UVD_STACK_SIZE + (AMDGPU_UVD_SESSION_SIZE * 40));
-
-	WREG32_SOC15(UVD, 0, mmUVD_UDEC_ADDR_CONFIG,
-			adev->gfx.config.gb_addr_config);
-	WREG32_SOC15(UVD, 0, mmUVD_UDEC_DB_ADDR_CONFIG,
-			adev->gfx.config.gb_addr_config);
-	WREG32_SOC15(UVD, 0, mmUVD_UDEC_DBW_ADDR_CONFIG,
-			adev->gfx.config.gb_addr_config);
-
-	WREG32_SOC15(UVD, 0, mmUVD_GP_SCRATCH4, adev->uvd.max_handles);
+		WREG32_SOC15(UVD, i, mmUVD_VCPU_CACHE_SIZE0, size);
+
+		WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW,
+				lower_32_bits(adev->uvd.inst[i].gpu_addr + offset));
+		WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH,
+				upper_32_bits(adev->uvd.inst[i].gpu_addr + offset));
+		WREG32_SOC15(UVD, i, mmUVD_VCPU_CACHE_OFFSET1, (1 << 21));
+		WREG32_SOC15(UVD, i, mmUVD_VCPU_CACHE_SIZE1, AMDGPU_UVD_HEAP_SIZE);
+
+		WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW,
+				lower_32_bits(adev->uvd.inst[i].gpu_addr + offset + AMDGPU_UVD_HEAP_SIZE));
+		WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH,
+				upper_32_bits(adev->uvd.inst[i].gpu_addr + offset + AMDGPU_UVD_HEAP_SIZE));
+		WREG32_SOC15(UVD, i, mmUVD_VCPU_CACHE_OFFSET2, (2 << 21));
+		WREG32_SOC15(UVD, i, mmUVD_VCPU_CACHE_SIZE2,
+				AMDGPU_UVD_STACK_SIZE + (AMDGPU_UVD_SESSION_SIZE * 40));
+
+		WREG32_SOC15(UVD, i, mmUVD_UDEC_ADDR_CONFIG,
+				adev->gfx.config.gb_addr_config);
+		WREG32_SOC15(UVD, i, mmUVD_UDEC_DB_ADDR_CONFIG,
+				adev->gfx.config.gb_addr_config);
+		WREG32_SOC15(UVD, i, mmUVD_UDEC_DBW_ADDR_CONFIG,
+				adev->gfx.config.gb_addr_config);
+
+		WREG32_SOC15(UVD, i, mmUVD_GP_SCRATCH4, adev->uvd.max_handles);
+	}
 }
 
 static int uvd_v7_0_mmsch_start(struct amdgpu_device *adev,
@@ -675,6 +736,7 @@ static int uvd_v7_0_mmsch_start(struct amdgpu_device *adev,
 	uint64_t addr = table->gpu_addr;
 	struct mmsch_v1_0_init_header *header = (struct mmsch_v1_0_init_header *)table->cpu_addr;
 	uint32_t size;
+	int i;
 
 	size = header->header_size + header->vce_table_size + header->uvd_table_size;
 
@@ -694,11 +756,14 @@ static int uvd_v7_0_mmsch_start(struct amdgpu_device *adev,
 	/* 4, set resp to zero */
 	WREG32_SOC15(VCE, 0, mmVCE_MMSCH_VF_MAILBOX_RESP, 0);
 
-	WDOORBELL32(adev->uvd.ring_enc[0].doorbell_index, 0);
-	adev->wb.wb[adev->uvd.ring_enc[0].wptr_offs] = 0;
-	adev->uvd.ring_enc[0].wptr = 0;
-	adev->uvd.ring_enc[0].wptr_old = 0;
-
+	for (i = 0; i < adev->uvd.num_uvd_inst; ++i) {
+		if (adev->uvd.harvest_config & (1 << i))
+			continue;
+		WDOORBELL32(adev->uvd.inst[i].ring_enc[0].doorbell_index, 0);
+		*adev->uvd.inst[i].ring_enc[0].wptr_cpu_addr = 0;
+		adev->uvd.inst[i].ring_enc[0].wptr = 0;
+		adev->uvd.inst[i].ring_enc[0].wptr_old = 0;
+	}
 	/* 5, kick off the initialization and wait until VCE_MMSCH_VF_MAILBOX_RESP becomes non-zero */
 	WREG32_SOC15(VCE, 0, mmVCE_MMSCH_VF_MAILBOX_HOST, 0x10000001);
 
@@ -731,6 +796,7 @@ static int uvd_v7_0_sriov_start(struct amdgpu_device *adev)
 	struct mmsch_v1_0_cmd_end end = { {0} };
 	uint32_t *init_table = adev->virt.mm_table.cpu_addr;
 	struct mmsch_v1_0_init_header *header = (struct mmsch_v1_0_init_header *)init_table;
+	uint8_t i = 0;
 
 	direct_wt.cmd_header.command_type = MMSCH_COMMAND__DIRECT_REG_WRITE;
 	direct_rd_mod_wt.cmd_header.command_type = MMSCH_COMMAND__DIRECT_REG_READ_MODIFY_WRITE;
@@ -748,120 +814,127 @@ static int uvd_v7_0_sriov_start(struct amdgpu_device *adev)
 
 		init_table += header->uvd_table_offset;
 
-		ring = &adev->uvd.ring;
-		ring->wptr = 0;
-		size = AMDGPU_GPU_PAGE_ALIGN(adev->uvd.fw->size + 4);
-
-		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_STATUS),
-						   0xFFFFFFFF, 0x00000004);
-		/* mc resume*/
-		if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
-			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
-						    lower_32_bits(adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].mc_addr));
-			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
-						    upper_32_bits(adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].mc_addr));
-			offset = 0;
-		} else {
-			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
-						    lower_32_bits(adev->uvd.gpu_addr));
-			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
-						    upper_32_bits(adev->uvd.gpu_addr));
-			offset = size;
+		for (i = 0; i < adev->uvd.num_uvd_inst; ++i) {
+			if (adev->uvd.harvest_config & (1 << i))
+				continue;
+			ring = &adev->uvd.inst[i].ring;
+			ring->wptr = 0;
+			size = AMDGPU_GPU_PAGE_ALIGN(adev->uvd.fw->size + 4);
+
+			MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_STATUS),
+							   0xFFFFFFFF, 0x00000004);
+			/* mc resume*/
+			if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+				MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i,
+							mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+							adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].tmr_mc_addr_lo);
+				MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i,
+							mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+							adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].tmr_mc_addr_hi);
+				MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CACHE_OFFSET0), 0);
+				offset = 0;
+			} else {
+				MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+							    lower_32_bits(adev->uvd.inst[i].gpu_addr));
+				MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+							    upper_32_bits(adev->uvd.inst[i].gpu_addr));
+				offset = size;
+				MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CACHE_OFFSET0),
+							AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
+
+			}
+
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_SIZE0), size);
+
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW),
+						    lower_32_bits(adev->uvd.inst[i].gpu_addr + offset));
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH),
+						    upper_32_bits(adev->uvd.inst[i].gpu_addr + offset));
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_OFFSET1), (1 << 21));
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_SIZE1), AMDGPU_UVD_HEAP_SIZE);
+
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW),
+						    lower_32_bits(adev->uvd.inst[i].gpu_addr + offset + AMDGPU_UVD_HEAP_SIZE));
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH),
+						    upper_32_bits(adev->uvd.inst[i].gpu_addr + offset + AMDGPU_UVD_HEAP_SIZE));
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_OFFSET2), (2 << 21));
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_SIZE2),
+						    AMDGPU_UVD_STACK_SIZE + (AMDGPU_UVD_SESSION_SIZE * 40));
+
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_GP_SCRATCH4), adev->uvd.max_handles);
+			/* mc resume end*/
+
+			/* disable clock gating */
+			MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_CGC_CTRL),
+							   ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK, 0);
+
+			/* disable interupt */
+			MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_MASTINT_EN),
+							   ~UVD_MASTINT_EN__VCPU_EN_MASK, 0);
+
+			/* stall UMC and register bus before resetting VCPU */
+			MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_CTRL2),
+							   ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK,
+							   UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+
+			/* put LMI, VCPU, RBC etc... into reset */
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_SOFT_RESET),
+						    (uint32_t)(UVD_SOFT_RESET__LMI_SOFT_RESET_MASK |
+							       UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK |
+							       UVD_SOFT_RESET__LBSI_SOFT_RESET_MASK |
+							       UVD_SOFT_RESET__RBC_SOFT_RESET_MASK |
+							       UVD_SOFT_RESET__CSM_SOFT_RESET_MASK |
+							       UVD_SOFT_RESET__CXW_SOFT_RESET_MASK |
+							       UVD_SOFT_RESET__TAP_SOFT_RESET_MASK |
+							       UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK));
+
+			/* initialize UVD memory controller */
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_CTRL),
+						    (uint32_t)((0x40 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
+							       UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+							       UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+							       UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
+							       UVD_LMI_CTRL__REQ_MODE_MASK |
+							       0x00100000L));
+
+			/* take all subblocks out of reset, except VCPU */
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_SOFT_RESET),
+						    UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+
+			/* enable VCPU clock */
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CNTL),
+						    UVD_VCPU_CNTL__CLK_EN_MASK);
+
+			/* enable master interrupt */
+			MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_MASTINT_EN),
+							   ~(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK),
+							   (UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK));
+
+			/* clear the bit 4 of UVD_STATUS */
+			MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_STATUS),
+							   ~(2 << UVD_STATUS__VCPU_REPORT__SHIFT), 0);
+
+			/* force RBC into idle state */
+			size = order_base_2(ring->ring_size);
+			tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, size);
+			tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_RBC_RB_CNTL), tmp);
+
+			ring = &adev->uvd.inst[i].ring_enc[0];
+			ring->wptr = 0;
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_RB_BASE_LO), ring->gpu_addr);
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_RB_BASE_HI), upper_32_bits(ring->gpu_addr));
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_RB_SIZE), ring->ring_size / 4);
+
+			/* boot up the VCPU */
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_SOFT_RESET), 0);
+
+			/* enable UMC */
+			MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_CTRL2),
+											   ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK, 0);
+
+			MMSCH_V1_0_INSERT_DIRECT_POLL(SOC15_REG_OFFSET(UVD, i, mmUVD_STATUS), 0x02, 0x02);
 		}
-
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CACHE_OFFSET0),
-					    AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CACHE_SIZE0), size);
-
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW),
-					    lower_32_bits(adev->uvd.gpu_addr + offset));
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH),
-					    upper_32_bits(adev->uvd.gpu_addr + offset));
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CACHE_OFFSET1), (1 << 21));
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CACHE_SIZE1), AMDGPU_UVD_HEAP_SIZE);
-
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW),
-					    lower_32_bits(adev->uvd.gpu_addr + offset + AMDGPU_UVD_HEAP_SIZE));
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH),
-					    upper_32_bits(adev->uvd.gpu_addr + offset + AMDGPU_UVD_HEAP_SIZE));
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CACHE_OFFSET2), (2 << 21));
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CACHE_SIZE2),
-					    AMDGPU_UVD_STACK_SIZE + (AMDGPU_UVD_SESSION_SIZE * 40));
-
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_GP_SCRATCH4), adev->uvd.max_handles);
-		/* mc resume end*/
-
-		/* disable clock gating */
-		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_CGC_CTRL),
-						   ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK, 0);
-
-		/* disable interupt */
-		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_MASTINT_EN),
-						   ~UVD_MASTINT_EN__VCPU_EN_MASK, 0);
-
-		/* stall UMC and register bus before resetting VCPU */
-		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2),
-						   ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK,
-						   UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
-
-		/* put LMI, VCPU, RBC etc... into reset */
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET),
-					    (uint32_t)(UVD_SOFT_RESET__LMI_SOFT_RESET_MASK |
-						       UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK |
-						       UVD_SOFT_RESET__LBSI_SOFT_RESET_MASK |
-						       UVD_SOFT_RESET__RBC_SOFT_RESET_MASK |
-						       UVD_SOFT_RESET__CSM_SOFT_RESET_MASK |
-						       UVD_SOFT_RESET__CXW_SOFT_RESET_MASK |
-						       UVD_SOFT_RESET__TAP_SOFT_RESET_MASK |
-						       UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK));
-
-		/* initialize UVD memory controller */
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL),
-					    (uint32_t)((0x40 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
-						       UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
-						       UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
-						       UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
-						       UVD_LMI_CTRL__REQ_MODE_MASK |
-						       0x00100000L));
-
-		/* take all subblocks out of reset, except VCPU */
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET),
-					    UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
-
-		/* enable VCPU clock */
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CNTL),
-					    UVD_VCPU_CNTL__CLK_EN_MASK);
-
-		/* enable master interrupt */
-		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_MASTINT_EN),
-						   ~(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK),
-						   (UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK));
-
-		/* clear the bit 4 of UVD_STATUS */
-		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_STATUS),
-						   ~(2 << UVD_STATUS__VCPU_REPORT__SHIFT), 0);
-
-		/* force RBC into idle state */
-		size = order_base_2(ring->ring_size);
-		tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, size);
-		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_RBC_RB_CNTL), tmp);
-
-		ring = &adev->uvd.ring_enc[0];
-		ring->wptr = 0;
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_RB_BASE_LO), ring->gpu_addr);
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_RB_BASE_HI), upper_32_bits(ring->gpu_addr));
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_RB_SIZE), ring->ring_size / 4);
-
-		/* boot up the VCPU */
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET), 0);
-
-		/* enable UMC */
-		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2),
-										   ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK, 0);
-
-		MMSCH_V1_0_INSERT_DIRECT_POLL(SOC15_REG_OFFSET(UVD, 0, mmUVD_STATUS), 0x02, 0x02);
-
 		/* add end packet */
 		memcpy((void *)init_table, &end, sizeof(struct mmsch_v1_0_cmd_end));
 		table_size += sizeof(struct mmsch_v1_0_cmd_end) / 4;
@@ -880,15 +953,19 @@ static int uvd_v7_0_sriov_start(struct amdgpu_device *adev)
  */
 static int uvd_v7_0_start(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *ring = &adev->uvd.ring;
+	struct amdgpu_ring *ring;
 	uint32_t rb_bufsz, tmp;
 	uint32_t lmi_swap_cntl;
 	uint32_t mp_swap_cntl;
-	int i, j, r;
-
-	/* disable DPG */
-	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_POWER_STATUS), 0,
-			~UVD_POWER_STATUS__UVD_PG_MODE_MASK);
+	int i, j, k, r;
+
+	for (k = 0; k < adev->uvd.num_uvd_inst; ++k) {
+		if (adev->uvd.harvest_config & (1 << k))
+			continue;
+		/* disable DPG */
+		WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_POWER_STATUS), 0,
+				~UVD_POWER_STATUS__UVD_PG_MODE_MASK);
+	}
 
 	/* disable byte swapping */
 	lmi_swap_cntl = 0;
@@ -896,157 +973,161 @@ static int uvd_v7_0_start(struct amdgpu_device *adev)
 
 	uvd_v7_0_mc_resume(adev);
 
-	/* disable clock gating */
-	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_CGC_CTRL), 0,
-			~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK);
-
-	/* disable interupt */
-	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_MASTINT_EN), 0,
-			~UVD_MASTINT_EN__VCPU_EN_MASK);
-
-	/* stall UMC and register bus before resetting VCPU */
-	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2),
-			UVD_LMI_CTRL2__STALL_ARB_UMC_MASK,
-			~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
-	mdelay(1);
-
-	/* put LMI, VCPU, RBC etc... into reset */
-	WREG32_SOC15(UVD, 0, mmUVD_SOFT_RESET,
-		UVD_SOFT_RESET__LMI_SOFT_RESET_MASK |
-		UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK |
-		UVD_SOFT_RESET__LBSI_SOFT_RESET_MASK |
-		UVD_SOFT_RESET__RBC_SOFT_RESET_MASK |
-		UVD_SOFT_RESET__CSM_SOFT_RESET_MASK |
-		UVD_SOFT_RESET__CXW_SOFT_RESET_MASK |
-		UVD_SOFT_RESET__TAP_SOFT_RESET_MASK |
-		UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK);
-	mdelay(5);
-
-	/* initialize UVD memory controller */
-	WREG32_SOC15(UVD, 0, mmUVD_LMI_CTRL,
-		(0x40 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
-		UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
-		UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
-		UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
-		UVD_LMI_CTRL__REQ_MODE_MASK |
-		0x00100000L);
+	for (k = 0; k < adev->uvd.num_uvd_inst; ++k) {
+		if (adev->uvd.harvest_config & (1 << k))
+			continue;
+		ring = &adev->uvd.inst[k].ring;
+		/* disable clock gating */
+		WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_CGC_CTRL), 0,
+				~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK);
+
+		/* disable interupt */
+		WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_MASTINT_EN), 0,
+				~UVD_MASTINT_EN__VCPU_EN_MASK);
+
+		/* stall UMC and register bus before resetting VCPU */
+		WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_LMI_CTRL2),
+				UVD_LMI_CTRL2__STALL_ARB_UMC_MASK,
+				~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+		mdelay(1);
+
+		/* put LMI, VCPU, RBC etc... into reset */
+		WREG32_SOC15(UVD, k, mmUVD_SOFT_RESET,
+			UVD_SOFT_RESET__LMI_SOFT_RESET_MASK |
+			UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK |
+			UVD_SOFT_RESET__LBSI_SOFT_RESET_MASK |
+			UVD_SOFT_RESET__RBC_SOFT_RESET_MASK |
+			UVD_SOFT_RESET__CSM_SOFT_RESET_MASK |
+			UVD_SOFT_RESET__CXW_SOFT_RESET_MASK |
+			UVD_SOFT_RESET__TAP_SOFT_RESET_MASK |
+			UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK);
+		mdelay(5);
+
+		/* initialize UVD memory controller */
+		WREG32_SOC15(UVD, k, mmUVD_LMI_CTRL,
+			(0x40 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
+			UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+			UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+			UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
+			UVD_LMI_CTRL__REQ_MODE_MASK |
+			0x00100000L);
 
 #ifdef __BIG_ENDIAN
-	/* swap (8 in 32) RB and IB */
-	lmi_swap_cntl = 0xa;
-	mp_swap_cntl = 0;
+		/* swap (8 in 32) RB and IB */
+		lmi_swap_cntl = 0xa;
+		mp_swap_cntl = 0;
 #endif
-	WREG32_SOC15(UVD, 0, mmUVD_LMI_SWAP_CNTL, lmi_swap_cntl);
-	WREG32_SOC15(UVD, 0, mmUVD_MP_SWAP_CNTL, mp_swap_cntl);
-
-	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUXA0, 0x40c2040);
-	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUXA1, 0x0);
-	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUXB0, 0x40c2040);
-	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUXB1, 0x0);
-	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_ALU, 0);
-	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUX, 0x88);
-
-	/* take all subblocks out of reset, except VCPU */
-	WREG32_SOC15(UVD, 0, mmUVD_SOFT_RESET,
-			UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
-	mdelay(5);
-
-	/* enable VCPU clock */
-	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CNTL,
-			UVD_VCPU_CNTL__CLK_EN_MASK);
-
-	/* enable UMC */
-	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2), 0,
-			~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
-
-	/* boot up the VCPU */
-	WREG32_SOC15(UVD, 0, mmUVD_SOFT_RESET, 0);
-	mdelay(10);
-
-	for (i = 0; i < 10; ++i) {
-		uint32_t status;
-
-		for (j = 0; j < 100; ++j) {
-			status = RREG32_SOC15(UVD, 0, mmUVD_STATUS);
-			if (status & 2)
-				break;
-			mdelay(10);
-		}
-		r = 0;
-		if (status & 2)
-			break;
+		WREG32_SOC15(UVD, k, mmUVD_LMI_SWAP_CNTL, lmi_swap_cntl);
+		WREG32_SOC15(UVD, k, mmUVD_MP_SWAP_CNTL, mp_swap_cntl);
 
-		DRM_ERROR("UVD not responding, trying to reset the VCPU!!!\n");
-		WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET),
-				UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK,
-				~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
-		mdelay(10);
-		WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET), 0,
-				~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+		WREG32_SOC15(UVD, k, mmUVD_MPC_SET_MUXA0, 0x40c2040);
+		WREG32_SOC15(UVD, k, mmUVD_MPC_SET_MUXA1, 0x0);
+		WREG32_SOC15(UVD, k, mmUVD_MPC_SET_MUXB0, 0x40c2040);
+		WREG32_SOC15(UVD, k, mmUVD_MPC_SET_MUXB1, 0x0);
+		WREG32_SOC15(UVD, k, mmUVD_MPC_SET_ALU, 0);
+		WREG32_SOC15(UVD, k, mmUVD_MPC_SET_MUX, 0x88);
+
+		/* take all subblocks out of reset, except VCPU */
+		WREG32_SOC15(UVD, k, mmUVD_SOFT_RESET,
+				UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+		mdelay(5);
+
+		/* enable VCPU clock */
+		WREG32_SOC15(UVD, k, mmUVD_VCPU_CNTL,
+				UVD_VCPU_CNTL__CLK_EN_MASK);
+
+		/* enable UMC */
+		WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_LMI_CTRL2), 0,
+				~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+
+		/* boot up the VCPU */
+		WREG32_SOC15(UVD, k, mmUVD_SOFT_RESET, 0);
 		mdelay(10);
-		r = -1;
-	}
 
-	if (r) {
-		DRM_ERROR("UVD not responding, giving up!!!\n");
-		return r;
-	}
-	/* enable master interrupt */
-	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_MASTINT_EN),
-		(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK),
-		~(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK));
-
-	/* clear the bit 4 of UVD_STATUS */
-	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_STATUS), 0,
-			~(2 << UVD_STATUS__VCPU_REPORT__SHIFT));
-
-	/* force RBC into idle state */
-	rb_bufsz = order_base_2(ring->ring_size);
-	tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz);
-	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
-	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
-	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_WPTR_POLL_EN, 0);
-	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
-	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
-	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_CNTL, tmp);
-
-	/* set the write pointer delay */
-	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR_CNTL, 0);
-
-	/* set the wb address */
-	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR_ADDR,
-			(upper_32_bits(ring->gpu_addr) >> 2));
-
-	/* programm the RB_BASE for ring buffer */
-	WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW,
-			lower_32_bits(ring->gpu_addr));
-	WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH,
-			upper_32_bits(ring->gpu_addr));
-
-	/* Initialize the ring buffer's read and write pointers */
-	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR, 0);
-
-	ring->wptr = RREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR);
-	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR,
-			lower_32_bits(ring->wptr));
+		for (i = 0; i < 10; ++i) {
+			uint32_t status;
+
+			for (j = 0; j < 100; ++j) {
+				status = RREG32_SOC15(UVD, k, mmUVD_STATUS);
+				if (status & 2)
+					break;
+				mdelay(10);
+			}
+			r = 0;
+			if (status & 2)
+				break;
 
-	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_RBC_RB_CNTL), 0,
-			~UVD_RBC_RB_CNTL__RB_NO_FETCH_MASK);
+			DRM_ERROR("UVD(%d) not responding, trying to reset the VCPU!!!\n", k);
+			WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_SOFT_RESET),
+					UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK,
+					~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+			mdelay(10);
+			WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_SOFT_RESET), 0,
+					~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+			mdelay(10);
+			r = -1;
+		}
 
-	ring = &adev->uvd.ring_enc[0];
-	WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR, lower_32_bits(ring->wptr));
-	WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR, lower_32_bits(ring->wptr));
-	WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO, ring->gpu_addr);
-	WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
-	WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE, ring->ring_size / 4);
+		if (r) {
+			DRM_ERROR("UVD(%d) not responding, giving up!!!\n", k);
+			return r;
+		}
+		/* enable master interrupt */
+		WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_MASTINT_EN),
+			(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK),
+			~(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK));
 
-	ring = &adev->uvd.ring_enc[1];
-	WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR2, lower_32_bits(ring->wptr));
-	WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
-	WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO2, ring->gpu_addr);
-	WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
-	WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE2, ring->ring_size / 4);
+		/* clear the bit 4 of UVD_STATUS */
+		WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_STATUS), 0,
+				~(2 << UVD_STATUS__VCPU_REPORT__SHIFT));
 
+		/* force RBC into idle state */
+		rb_bufsz = order_base_2(ring->ring_size);
+		tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_WPTR_POLL_EN, 0);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
+		WREG32_SOC15(UVD, k, mmUVD_RBC_RB_CNTL, tmp);
+
+		/* set the write pointer delay */
+		WREG32_SOC15(UVD, k, mmUVD_RBC_RB_WPTR_CNTL, 0);
+
+		/* set the wb address */
+		WREG32_SOC15(UVD, k, mmUVD_RBC_RB_RPTR_ADDR,
+				(upper_32_bits(ring->gpu_addr) >> 2));
+
+		/* program the RB_BASE for ring buffer */
+		WREG32_SOC15(UVD, k, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW,
+				lower_32_bits(ring->gpu_addr));
+		WREG32_SOC15(UVD, k, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH,
+				upper_32_bits(ring->gpu_addr));
+
+		/* Initialize the ring buffer's read and write pointers */
+		WREG32_SOC15(UVD, k, mmUVD_RBC_RB_RPTR, 0);
+
+		ring->wptr = RREG32_SOC15(UVD, k, mmUVD_RBC_RB_RPTR);
+		WREG32_SOC15(UVD, k, mmUVD_RBC_RB_WPTR,
+				lower_32_bits(ring->wptr));
+
+		WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_RBC_RB_CNTL), 0,
+				~UVD_RBC_RB_CNTL__RB_NO_FETCH_MASK);
+
+		ring = &adev->uvd.inst[k].ring_enc[0];
+		WREG32_SOC15(UVD, k, mmUVD_RB_RPTR, lower_32_bits(ring->wptr));
+		WREG32_SOC15(UVD, k, mmUVD_RB_WPTR, lower_32_bits(ring->wptr));
+		WREG32_SOC15(UVD, k, mmUVD_RB_BASE_LO, ring->gpu_addr);
+		WREG32_SOC15(UVD, k, mmUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+		WREG32_SOC15(UVD, k, mmUVD_RB_SIZE, ring->ring_size / 4);
+
+		ring = &adev->uvd.inst[k].ring_enc[1];
+		WREG32_SOC15(UVD, k, mmUVD_RB_RPTR2, lower_32_bits(ring->wptr));
+		WREG32_SOC15(UVD, k, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
+		WREG32_SOC15(UVD, k, mmUVD_RB_BASE_LO2, ring->gpu_addr);
+		WREG32_SOC15(UVD, k, mmUVD_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
+		WREG32_SOC15(UVD, k, mmUVD_RB_SIZE2, ring->ring_size / 4);
+	}
 	return 0;
 }
 
@@ -1059,62 +1140,72 @@ static int uvd_v7_0_start(struct amdgpu_device *adev)
  */
 static void uvd_v7_0_stop(struct amdgpu_device *adev)
 {
-	/* force RBC into idle state */
-	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_CNTL, 0x11010101);
-
-	/* Stall UMC and register bus before resetting VCPU */
-	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2),
-			UVD_LMI_CTRL2__STALL_ARB_UMC_MASK,
-			~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
-	mdelay(1);
-
-	/* put VCPU into reset */
-	WREG32_SOC15(UVD, 0, mmUVD_SOFT_RESET,
-			UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
-	mdelay(5);
-
-	/* disable VCPU clock */
-	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CNTL, 0x0);
-
-	/* Unstall UMC and register bus */
-	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2), 0,
-			~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+	uint8_t i = 0;
+
+	for (i = 0; i < adev->uvd.num_uvd_inst; ++i) {
+		if (adev->uvd.harvest_config & (1 << i))
+			continue;
+		/* force RBC into idle state */
+		WREG32_SOC15(UVD, i, mmUVD_RBC_RB_CNTL, 0x11010101);
+
+		/* Stall UMC and register bus before resetting VCPU */
+		WREG32_P(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_CTRL2),
+				UVD_LMI_CTRL2__STALL_ARB_UMC_MASK,
+				~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+		mdelay(1);
+
+		/* put VCPU into reset */
+		WREG32_SOC15(UVD, i, mmUVD_SOFT_RESET,
+				UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+		mdelay(5);
+
+		/* disable VCPU clock */
+		WREG32_SOC15(UVD, i, mmUVD_VCPU_CNTL, 0x0);
+
+		/* Unstall UMC and register bus */
+		WREG32_P(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_CTRL2), 0,
+				~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+	}
 }
 
 /**
  * uvd_v7_0_ring_emit_fence - emit an fence & trap command
  *
  * @ring: amdgpu_ring pointer
- * @fence: fence to emit
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
  *
  * Write a fence and a trap command to the ring.
  */
 static void uvd_v7_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
 				     unsigned flags)
 {
+	struct amdgpu_device *adev = ring->adev;
+
 	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
 
 	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_CONTEXT_ID), 0));
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_CONTEXT_ID), 0));
 	amdgpu_ring_write(ring, seq);
 	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0), 0));
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA0), 0));
 	amdgpu_ring_write(ring, addr & 0xffffffff);
 	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA1), 0));
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA1), 0));
 	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xff);
 	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD), 0));
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_CMD), 0));
 	amdgpu_ring_write(ring, 0);
 
 	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0), 0));
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA0), 0));
 	amdgpu_ring_write(ring, 0);
 	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA1), 0));
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA1), 0));
 	amdgpu_ring_write(ring, 0);
 	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD), 0));
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_CMD), 0));
 	amdgpu_ring_write(ring, 2);
 }
 
@@ -1122,13 +1213,16 @@ static void uvd_v7_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq
  * uvd_v7_0_enc_ring_emit_fence - emit an enc fence & trap command
  *
  * @ring: amdgpu_ring pointer
- * @fence: fence to emit
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
  *
  * Write enc a fence and a trap command to the ring.
  */
 static void uvd_v7_0_enc_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
 			u64 seq, unsigned flags)
 {
+
 	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
 
 	amdgpu_ring_write(ring, HEVC_ENC_CMD_FENCE);
@@ -1139,30 +1233,13 @@ static void uvd_v7_0_enc_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
 }
 
 /**
- * uvd_v7_0_ring_emit_hdp_flush - emit an hdp flush
+ * uvd_v7_0_ring_emit_hdp_flush - skip HDP flushing
  *
  * @ring: amdgpu_ring pointer
- *
- * Emits an hdp flush.
  */
 static void uvd_v7_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
 {
-	amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(NBIF, 0,
-		mmHDP_MEM_COHERENCY_FLUSH_CNTL), 0));
-	amdgpu_ring_write(ring, 0);
-}
-
-/**
- * uvd_v7_0_ring_hdp_invalidate - emit an hdp invalidate
- *
- * @ring: amdgpu_ring pointer
- *
- * Emits an hdp invalidate.
- */
-static void uvd_v7_0_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
-{
-	amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(HDP, 0, mmHDP_DEBUG0), 0));
-	amdgpu_ring_write(ring, 1);
+	/* The firmware doesn't seem to like touching registers at this point. */
 }
 
 /**
@@ -1179,59 +1256,88 @@ static int uvd_v7_0_ring_test_ring(struct amdgpu_ring *ring)
 	unsigned i;
 	int r;
 
-	WREG32_SOC15(UVD, 0, mmUVD_CONTEXT_ID, 0xCAFEDEAD);
+	WREG32_SOC15(UVD, ring->me, mmUVD_CONTEXT_ID, 0xCAFEDEAD);
 	r = amdgpu_ring_alloc(ring, 3);
-	if (r) {
-		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n",
-			  ring->idx, r);
+	if (r)
 		return r;
-	}
+
 	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_CONTEXT_ID), 0));
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_CONTEXT_ID), 0));
 	amdgpu_ring_write(ring, 0xDEADBEEF);
 	amdgpu_ring_commit(ring);
 	for (i = 0; i < adev->usec_timeout; i++) {
-		tmp = RREG32_SOC15(UVD, 0, mmUVD_CONTEXT_ID);
+		tmp = RREG32_SOC15(UVD, ring->me, mmUVD_CONTEXT_ID);
 		if (tmp == 0xDEADBEEF)
 			break;
-		DRM_UDELAY(1);
+		udelay(1);
 	}
 
-	if (i < adev->usec_timeout) {
-		DRM_INFO("ring test on %d succeeded in %d usecs\n",
-			 ring->idx, i);
-	} else {
-		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
-			  ring->idx, tmp);
-		r = -EINVAL;
-	}
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+
 	return r;
 }
 
 /**
+ * uvd_v7_0_ring_patch_cs_in_place - Patch the IB for command submission.
+ *
+ * @p: the CS parser with the IBs
+ * @job: which job this ib is in
+ * @ib: which IB to patch
+ *
+ */
+static int uvd_v7_0_ring_patch_cs_in_place(struct amdgpu_cs_parser *p,
+					   struct amdgpu_job *job,
+					   struct amdgpu_ib *ib)
+{
+	struct amdgpu_ring *ring = amdgpu_job_ring(job);
+	unsigned i;
+
+	/* No patching necessary for the first instance */
+	if (!ring->me)
+		return 0;
+
+	for (i = 0; i < ib->length_dw; i += 2) {
+		uint32_t reg = amdgpu_ib_get_value(ib, i);
+
+		reg -= p->adev->reg_offset[UVD_HWIP][0][1];
+		reg += p->adev->reg_offset[UVD_HWIP][1][1];
+
+		amdgpu_ib_set_value(ib, i, reg);
+	}
+	return 0;
+}
+
+/**
  * uvd_v7_0_ring_emit_ib - execute indirect buffer
  *
  * @ring: amdgpu_ring pointer
+ * @job: job to retrieve vmid from
  * @ib: indirect buffer to execute
+ * @flags: unused
  *
  * Write ring commands to execute the indirect buffer
  */
 static void uvd_v7_0_ring_emit_ib(struct amdgpu_ring *ring,
+				  struct amdgpu_job *job,
 				  struct amdgpu_ib *ib,
-				  unsigned vm_id, bool ctx_switch)
+				  uint32_t flags)
 {
+	struct amdgpu_device *adev = ring->adev;
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+
 	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_RBC_IB_VMID), 0));
-	amdgpu_ring_write(ring, vm_id);
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_LMI_RBC_IB_VMID), 0));
+	amdgpu_ring_write(ring, vmid);
 
 	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_RBC_IB_64BIT_BAR_LOW), 0));
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_LMI_RBC_IB_64BIT_BAR_LOW), 0));
 	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
 	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH), 0));
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH), 0));
 	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
 	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_RBC_IB_SIZE), 0));
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_RBC_IB_SIZE), 0));
 	amdgpu_ring_write(ring, ib->length_dw);
 }
 
@@ -1239,224 +1345,124 @@ static void uvd_v7_0_ring_emit_ib(struct amdgpu_ring *ring,
  * uvd_v7_0_enc_ring_emit_ib - enc execute indirect buffer
  *
  * @ring: amdgpu_ring pointer
+ * @job: job to retrive vmid from
  * @ib: indirect buffer to execute
+ * @flags: unused
  *
  * Write enc ring commands to execute the indirect buffer
  */
 static void uvd_v7_0_enc_ring_emit_ib(struct amdgpu_ring *ring,
-		struct amdgpu_ib *ib, unsigned int vm_id, bool ctx_switch)
+					struct amdgpu_job *job,
+					struct amdgpu_ib *ib,
+					uint32_t flags)
 {
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+
 	amdgpu_ring_write(ring, HEVC_ENC_CMD_IB_VM);
-	amdgpu_ring_write(ring, vm_id);
+	amdgpu_ring_write(ring, vmid);
 	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
 	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
 	amdgpu_ring_write(ring, ib->length_dw);
 }
 
-static void uvd_v7_0_vm_reg_write(struct amdgpu_ring *ring,
-				uint32_t data0, uint32_t data1)
+static void uvd_v7_0_ring_emit_wreg(struct amdgpu_ring *ring,
+				    uint32_t reg, uint32_t val)
 {
+	struct amdgpu_device *adev = ring->adev;
+
 	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0), 0));
-	amdgpu_ring_write(ring, data0);
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA0), 0));
+	amdgpu_ring_write(ring, reg << 2);
 	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA1), 0));
-	amdgpu_ring_write(ring, data1);
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA1), 0));
+	amdgpu_ring_write(ring, val);
 	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD), 0));
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_CMD), 0));
 	amdgpu_ring_write(ring, 8);
 }
 
-static void uvd_v7_0_vm_reg_wait(struct amdgpu_ring *ring,
-				uint32_t data0, uint32_t data1, uint32_t mask)
+static void uvd_v7_0_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+					uint32_t val, uint32_t mask)
 {
+	struct amdgpu_device *adev = ring->adev;
+
 	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0), 0));
-	amdgpu_ring_write(ring, data0);
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA0), 0));
+	amdgpu_ring_write(ring, reg << 2);
 	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA1), 0));
-	amdgpu_ring_write(ring, data1);
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA1), 0));
+	amdgpu_ring_write(ring, val);
 	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GP_SCRATCH8), 0));
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GP_SCRATCH8), 0));
 	amdgpu_ring_write(ring, mask);
 	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD), 0));
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_CMD), 0));
 	amdgpu_ring_write(ring, 12);
 }
 
 static void uvd_v7_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
-					unsigned vm_id, uint64_t pd_addr)
+					unsigned vmid, uint64_t pd_addr)
 {
-	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
-	uint32_t req = ring->adev->gart.gart_funcs->get_invalidate_req(vm_id);
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
 	uint32_t data0, data1, mask;
-	unsigned eng = ring->vm_inv_eng;
-
-	pd_addr = amdgpu_gart_get_vm_pde(ring->adev, pd_addr);
-	pd_addr |= AMDGPU_PTE_VALID;
 
-	data0 = (hub->ctx0_ptb_addr_hi32 + vm_id * 2) << 2;
-	data1 = upper_32_bits(pd_addr);
-	uvd_v7_0_vm_reg_write(ring, data0, data1);
+	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
 
-	data0 = (hub->ctx0_ptb_addr_lo32 + vm_id * 2) << 2;
-	data1 = lower_32_bits(pd_addr);
-	uvd_v7_0_vm_reg_write(ring, data0, data1);
-
-	data0 = (hub->ctx0_ptb_addr_lo32 + vm_id * 2) << 2;
+	/* wait for reg writes */
+	data0 = hub->ctx0_ptb_addr_lo32 + vmid * hub->ctx_addr_distance;
 	data1 = lower_32_bits(pd_addr);
 	mask = 0xffffffff;
-	uvd_v7_0_vm_reg_wait(ring, data0, data1, mask);
-
-	/* flush TLB */
-	data0 = (hub->vm_inv_eng0_req + eng) << 2;
-	data1 = req;
-	uvd_v7_0_vm_reg_write(ring, data0, data1);
-
-	/* wait for flush */
-	data0 = (hub->vm_inv_eng0_ack + eng) << 2;
-	data1 = 1 << vm_id;
-	mask =  1 << vm_id;
-	uvd_v7_0_vm_reg_wait(ring, data0, data1, mask);
-}
-
-static void uvd_v7_0_enc_ring_insert_end(struct amdgpu_ring *ring)
-{
-	amdgpu_ring_write(ring, HEVC_ENC_CMD_END);
-}
-
-static void uvd_v7_0_enc_ring_emit_vm_flush(struct amdgpu_ring *ring,
-			 unsigned int vm_id, uint64_t pd_addr)
-{
-	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
-	uint32_t req = ring->adev->gart.gart_funcs->get_invalidate_req(vm_id);
-	unsigned eng = ring->vm_inv_eng;
-
-	pd_addr = amdgpu_gart_get_vm_pde(ring->adev, pd_addr);
-	pd_addr |= AMDGPU_PTE_VALID;
-
-	amdgpu_ring_write(ring, HEVC_ENC_CMD_REG_WRITE);
-	amdgpu_ring_write(ring,	(hub->ctx0_ptb_addr_hi32 + vm_id * 2) << 2);
-	amdgpu_ring_write(ring, upper_32_bits(pd_addr));
-
-	amdgpu_ring_write(ring, HEVC_ENC_CMD_REG_WRITE);
-	amdgpu_ring_write(ring,	(hub->ctx0_ptb_addr_lo32 + vm_id * 2) << 2);
-	amdgpu_ring_write(ring, lower_32_bits(pd_addr));
-
-	amdgpu_ring_write(ring, HEVC_ENC_CMD_REG_WAIT);
-	amdgpu_ring_write(ring,	(hub->ctx0_ptb_addr_lo32 + vm_id * 2) << 2);
-	amdgpu_ring_write(ring, 0xffffffff);
-	amdgpu_ring_write(ring, lower_32_bits(pd_addr));
-
-	/* flush TLB */
-	amdgpu_ring_write(ring, HEVC_ENC_CMD_REG_WRITE);
-	amdgpu_ring_write(ring,	(hub->vm_inv_eng0_req + eng) << 2);
-	amdgpu_ring_write(ring, req);
-
-	/* wait for flush */
-	amdgpu_ring_write(ring, HEVC_ENC_CMD_REG_WAIT);
-	amdgpu_ring_write(ring, (hub->vm_inv_eng0_ack + eng) << 2);
-	amdgpu_ring_write(ring, 1 << vm_id);
-	amdgpu_ring_write(ring, 1 << vm_id);
+	uvd_v7_0_ring_emit_reg_wait(ring, data0, data1, mask);
 }
 
-#if 0
-static bool uvd_v7_0_is_idle(void *handle)
+static void uvd_v7_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return !(RREG32(mmSRBM_STATUS) & SRBM_STATUS__UVD_BUSY_MASK);
-}
+	struct amdgpu_device *adev = ring->adev;
+	int i;
 
-static int uvd_v7_0_wait_for_idle(void *handle)
-{
-	unsigned i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	WARN_ON(ring->wptr % 2 || count % 2);
 
-	for (i = 0; i < adev->usec_timeout; i++) {
-		if (uvd_v7_0_is_idle(handle))
-			return 0;
+	for (i = 0; i < count / 2; i++) {
+		amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_NO_OP), 0));
+		amdgpu_ring_write(ring, 0);
 	}
-	return -ETIMEDOUT;
 }
 
-#define AMDGPU_UVD_STATUS_BUSY_MASK    0xfd
-static bool uvd_v7_0_check_soft_reset(void *handle)
+static void uvd_v7_0_enc_ring_insert_end(struct amdgpu_ring *ring)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	u32 srbm_soft_reset = 0;
-	u32 tmp = RREG32(mmSRBM_STATUS);
-
-	if (REG_GET_FIELD(tmp, SRBM_STATUS, UVD_RQ_PENDING) ||
-	    REG_GET_FIELD(tmp, SRBM_STATUS, UVD_BUSY) ||
-	    (RREG32_SOC15(UVD, 0, mmUVD_STATUS) &
-		    AMDGPU_UVD_STATUS_BUSY_MASK))
-		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset,
-				SRBM_SOFT_RESET, SOFT_RESET_UVD, 1);
-
-	if (srbm_soft_reset) {
-		adev->uvd.srbm_soft_reset = srbm_soft_reset;
-		return true;
-	} else {
-		adev->uvd.srbm_soft_reset = 0;
-		return false;
-	}
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_END);
 }
 
-static int uvd_v7_0_pre_soft_reset(void *handle)
+static void uvd_v7_0_enc_ring_emit_reg_wait(struct amdgpu_ring *ring,
+					    uint32_t reg, uint32_t val,
+					    uint32_t mask)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (!adev->uvd.srbm_soft_reset)
-		return 0;
-
-	uvd_v7_0_stop(adev);
-	return 0;
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_REG_WAIT);
+	amdgpu_ring_write(ring,	reg << 2);
+	amdgpu_ring_write(ring, mask);
+	amdgpu_ring_write(ring, val);
 }
 
-static int uvd_v7_0_soft_reset(void *handle)
+static void uvd_v7_0_enc_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					    unsigned int vmid, uint64_t pd_addr)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	u32 srbm_soft_reset;
-
-	if (!adev->uvd.srbm_soft_reset)
-		return 0;
-	srbm_soft_reset = adev->uvd.srbm_soft_reset;
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
 
-	if (srbm_soft_reset) {
-		u32 tmp;
-
-		tmp = RREG32(mmSRBM_SOFT_RESET);
-		tmp |= srbm_soft_reset;
-		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
-		WREG32(mmSRBM_SOFT_RESET, tmp);
-		tmp = RREG32(mmSRBM_SOFT_RESET);
-
-		udelay(50);
-
-		tmp &= ~srbm_soft_reset;
-		WREG32(mmSRBM_SOFT_RESET, tmp);
-		tmp = RREG32(mmSRBM_SOFT_RESET);
-
-		/* Wait a little for things to settle down */
-		udelay(50);
-	}
+	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
 
-	return 0;
+	/* wait for reg writes */
+	uvd_v7_0_enc_ring_emit_reg_wait(ring, hub->ctx0_ptb_addr_lo32 +
+					vmid * hub->ctx_addr_distance,
+					lower_32_bits(pd_addr), 0xffffffff);
 }
 
-static int uvd_v7_0_post_soft_reset(void *handle)
+static void uvd_v7_0_enc_ring_emit_wreg(struct amdgpu_ring *ring,
+					uint32_t reg, uint32_t val)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (!adev->uvd.srbm_soft_reset)
-		return 0;
-
-	mdelay(5);
-
-	return uvd_v7_0_start(adev);
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_REG_WRITE);
+	amdgpu_ring_write(ring,	reg << 2);
+	amdgpu_ring_write(ring, val);
 }
-#endif
 
 static int uvd_v7_0_set_interrupt_state(struct amdgpu_device *adev,
 					struct amdgpu_irq_src *source,
@@ -1471,17 +1477,32 @@ static int uvd_v7_0_process_interrupt(struct amdgpu_device *adev,
 				      struct amdgpu_irq_src *source,
 				      struct amdgpu_iv_entry *entry)
 {
+	uint32_t ip_instance;
+
+	switch (entry->client_id) {
+	case SOC15_IH_CLIENTID_UVD:
+		ip_instance = 0;
+		break;
+	case SOC15_IH_CLIENTID_UVD1:
+		ip_instance = 1;
+		break;
+	default:
+		DRM_ERROR("Unhandled client id: %d\n", entry->client_id);
+		return 0;
+	}
+
 	DRM_DEBUG("IH: UVD TRAP\n");
+
 	switch (entry->src_id) {
 	case 124:
-		amdgpu_fence_process(&adev->uvd.ring);
+		amdgpu_fence_process(&adev->uvd.inst[ip_instance].ring);
 		break;
 	case 119:
-		amdgpu_fence_process(&adev->uvd.ring_enc[0]);
+		amdgpu_fence_process(&adev->uvd.inst[ip_instance].ring_enc[0]);
 		break;
 	case 120:
 		if (!amdgpu_sriov_vf(adev))
-			amdgpu_fence_process(&adev->uvd.ring_enc[1]);
+			amdgpu_fence_process(&adev->uvd.inst[ip_instance].ring_enc[1]);
 		break;
 	default:
 		DRM_ERROR("Unhandled interrupt: %d %d\n",
@@ -1492,172 +1513,7 @@ static int uvd_v7_0_process_interrupt(struct amdgpu_device *adev,
 	return 0;
 }
 
-#if 0
-static void uvd_v7_0_set_sw_clock_gating(struct amdgpu_device *adev)
-{
-	uint32_t data, data1, data2, suvd_flags;
-
-	data = RREG32_SOC15(UVD, 0, mmUVD_CGC_CTRL);
-	data1 = RREG32_SOC15(UVD, 0, mmUVD_SUVD_CGC_GATE);
-	data2 = RREG32_SOC15(UVD, 0, mmUVD_SUVD_CGC_CTRL);
-
-	data &= ~(UVD_CGC_CTRL__CLK_OFF_DELAY_MASK |
-		  UVD_CGC_CTRL__CLK_GATE_DLY_TIMER_MASK);
-
-	suvd_flags = UVD_SUVD_CGC_GATE__SRE_MASK |
-		     UVD_SUVD_CGC_GATE__SIT_MASK |
-		     UVD_SUVD_CGC_GATE__SMP_MASK |
-		     UVD_SUVD_CGC_GATE__SCM_MASK |
-		     UVD_SUVD_CGC_GATE__SDB_MASK;
-
-	data |= UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK |
-		(1 << REG_FIELD_SHIFT(UVD_CGC_CTRL, CLK_GATE_DLY_TIMER)) |
-		(4 << REG_FIELD_SHIFT(UVD_CGC_CTRL, CLK_OFF_DELAY));
-
-	data &= ~(UVD_CGC_CTRL__UDEC_RE_MODE_MASK |
-			UVD_CGC_CTRL__UDEC_CM_MODE_MASK |
-			UVD_CGC_CTRL__UDEC_IT_MODE_MASK |
-			UVD_CGC_CTRL__UDEC_DB_MODE_MASK |
-			UVD_CGC_CTRL__UDEC_MP_MODE_MASK |
-			UVD_CGC_CTRL__SYS_MODE_MASK |
-			UVD_CGC_CTRL__UDEC_MODE_MASK |
-			UVD_CGC_CTRL__MPEG2_MODE_MASK |
-			UVD_CGC_CTRL__REGS_MODE_MASK |
-			UVD_CGC_CTRL__RBC_MODE_MASK |
-			UVD_CGC_CTRL__LMI_MC_MODE_MASK |
-			UVD_CGC_CTRL__LMI_UMC_MODE_MASK |
-			UVD_CGC_CTRL__IDCT_MODE_MASK |
-			UVD_CGC_CTRL__MPRD_MODE_MASK |
-			UVD_CGC_CTRL__MPC_MODE_MASK |
-			UVD_CGC_CTRL__LBSI_MODE_MASK |
-			UVD_CGC_CTRL__LRBBM_MODE_MASK |
-			UVD_CGC_CTRL__WCB_MODE_MASK |
-			UVD_CGC_CTRL__VCPU_MODE_MASK |
-			UVD_CGC_CTRL__JPEG_MODE_MASK |
-			UVD_CGC_CTRL__JPEG2_MODE_MASK |
-			UVD_CGC_CTRL__SCPU_MODE_MASK);
-	data2 &= ~(UVD_SUVD_CGC_CTRL__SRE_MODE_MASK |
-			UVD_SUVD_CGC_CTRL__SIT_MODE_MASK |
-			UVD_SUVD_CGC_CTRL__SMP_MODE_MASK |
-			UVD_SUVD_CGC_CTRL__SCM_MODE_MASK |
-			UVD_SUVD_CGC_CTRL__SDB_MODE_MASK);
-	data1 |= suvd_flags;
-
-	WREG32_SOC15(UVD, 0, mmUVD_CGC_CTRL, data);
-	WREG32_SOC15(UVD, 0, mmUVD_CGC_GATE, 0);
-	WREG32_SOC15(UVD, 0, mmUVD_SUVD_CGC_GATE, data1);
-	WREG32_SOC15(UVD, 0, mmUVD_SUVD_CGC_CTRL, data2);
-}
-
-static void uvd_v7_0_set_hw_clock_gating(struct amdgpu_device *adev)
-{
-	uint32_t data, data1, cgc_flags, suvd_flags;
-
-	data = RREG32_SOC15(UVD, 0, mmUVD_CGC_GATE);
-	data1 = RREG32_SOC15(UVD, 0, mmUVD_SUVD_CGC_GATE);
-
-	cgc_flags = UVD_CGC_GATE__SYS_MASK |
-		UVD_CGC_GATE__UDEC_MASK |
-		UVD_CGC_GATE__MPEG2_MASK |
-		UVD_CGC_GATE__RBC_MASK |
-		UVD_CGC_GATE__LMI_MC_MASK |
-		UVD_CGC_GATE__IDCT_MASK |
-		UVD_CGC_GATE__MPRD_MASK |
-		UVD_CGC_GATE__MPC_MASK |
-		UVD_CGC_GATE__LBSI_MASK |
-		UVD_CGC_GATE__LRBBM_MASK |
-		UVD_CGC_GATE__UDEC_RE_MASK |
-		UVD_CGC_GATE__UDEC_CM_MASK |
-		UVD_CGC_GATE__UDEC_IT_MASK |
-		UVD_CGC_GATE__UDEC_DB_MASK |
-		UVD_CGC_GATE__UDEC_MP_MASK |
-		UVD_CGC_GATE__WCB_MASK |
-		UVD_CGC_GATE__VCPU_MASK |
-		UVD_CGC_GATE__SCPU_MASK |
-		UVD_CGC_GATE__JPEG_MASK |
-		UVD_CGC_GATE__JPEG2_MASK;
-
-	suvd_flags = UVD_SUVD_CGC_GATE__SRE_MASK |
-				UVD_SUVD_CGC_GATE__SIT_MASK |
-				UVD_SUVD_CGC_GATE__SMP_MASK |
-				UVD_SUVD_CGC_GATE__SCM_MASK |
-				UVD_SUVD_CGC_GATE__SDB_MASK;
-
-	data |= cgc_flags;
-	data1 |= suvd_flags;
-
-	WREG32_SOC15(UVD, 0, mmUVD_CGC_GATE, data);
-	WREG32_SOC15(UVD, 0, mmUVD_SUVD_CGC_GATE, data1);
-}
-
-static void uvd_v7_0_set_bypass_mode(struct amdgpu_device *adev, bool enable)
-{
-	u32 tmp = RREG32_SMC(ixGCK_DFS_BYPASS_CNTL);
-
-	if (enable)
-		tmp |= (GCK_DFS_BYPASS_CNTL__BYPASSDCLK_MASK |
-			GCK_DFS_BYPASS_CNTL__BYPASSVCLK_MASK);
-	else
-		tmp &= ~(GCK_DFS_BYPASS_CNTL__BYPASSDCLK_MASK |
-			 GCK_DFS_BYPASS_CNTL__BYPASSVCLK_MASK);
-
-	WREG32_SMC(ixGCK_DFS_BYPASS_CNTL, tmp);
-}
-
-
-static int uvd_v7_0_set_clockgating_state(void *handle,
-					  enum amd_clockgating_state state)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	bool enable = (state == AMD_CG_STATE_GATE) ? true : false;
-
-	uvd_v7_0_set_bypass_mode(adev, enable);
-
-	if (!(adev->cg_flags & AMD_CG_SUPPORT_UVD_MGCG))
-		return 0;
-
-	if (enable) {
-		/* disable HW gating and enable Sw gating */
-		uvd_v7_0_set_sw_clock_gating(adev);
-	} else {
-		/* wait for STATUS to clear */
-		if (uvd_v7_0_wait_for_idle(handle))
-			return -EBUSY;
-
-		/* enable HW gates because UVD is idle */
-		/* uvd_v7_0_set_hw_clock_gating(adev); */
-	}
-
-	return 0;
-}
-
-static int uvd_v7_0_set_powergating_state(void *handle,
-					  enum amd_powergating_state state)
-{
-	/* This doesn't actually powergate the UVD block.
-	 * That's done in the dpm code via the SMC.  This
-	 * just re-inits the block as necessary.  The actual
-	 * gating still happens in the dpm code.  We should
-	 * revisit this when there is a cleaner line between
-	 * the smc and the hw blocks
-	 */
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (!(adev->pg_flags & AMD_PG_SUPPORT_UVD))
-		return 0;
-
-	WREG32_SOC15(UVD, 0, mmUVD_POWER_STATUS, UVD_POWER_STATUS__UVD_PG_EN_MASK);
-
-	if (state == AMD_PG_STATE_GATE) {
-		uvd_v7_0_stop(adev);
-		return 0;
-	} else {
-		return uvd_v7_0_start(adev);
-	}
-}
-#endif
-
-static int uvd_v7_0_set_clockgating_state(void *handle,
+static int uvd_v7_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
 	/* needed for driver unload*/
@@ -1667,19 +1523,13 @@ static int uvd_v7_0_set_clockgating_state(void *handle,
 const struct amd_ip_funcs uvd_v7_0_ip_funcs = {
 	.name = "uvd_v7_0",
 	.early_init = uvd_v7_0_early_init,
-	.late_init = NULL,
 	.sw_init = uvd_v7_0_sw_init,
 	.sw_fini = uvd_v7_0_sw_fini,
 	.hw_init = uvd_v7_0_hw_init,
 	.hw_fini = uvd_v7_0_hw_fini,
+	.prepare_suspend = uvd_v7_0_prepare_suspend,
 	.suspend = uvd_v7_0_suspend,
 	.resume = uvd_v7_0_resume,
-	.is_idle = NULL /* uvd_v7_0_is_idle */,
-	.wait_for_idle = NULL /* uvd_v7_0_wait_for_idle */,
-	.check_soft_reset = NULL /* uvd_v7_0_check_soft_reset */,
-	.pre_soft_reset = NULL /* uvd_v7_0_pre_soft_reset */,
-	.soft_reset = NULL /* uvd_v7_0_soft_reset */,
-	.post_soft_reset = NULL /* uvd_v7_0_post_soft_reset */,
 	.set_clockgating_state = uvd_v7_0_set_clockgating_state,
 	.set_powergating_state = NULL /* uvd_v7_0_set_powergating_state */,
 };
@@ -1687,29 +1537,32 @@ const struct amd_ip_funcs uvd_v7_0_ip_funcs = {
 static const struct amdgpu_ring_funcs uvd_v7_0_ring_vm_funcs = {
 	.type = AMDGPU_RING_TYPE_UVD,
 	.align_mask = 0xf,
-	.nop = PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_NO_OP), 0),
 	.support_64bit_ptrs = false,
-	.vmhub = AMDGPU_MMHUB,
+	.no_user_fence = true,
 	.get_rptr = uvd_v7_0_ring_get_rptr,
 	.get_wptr = uvd_v7_0_ring_get_wptr,
 	.set_wptr = uvd_v7_0_ring_set_wptr,
+	.patch_cs_in_place = uvd_v7_0_ring_patch_cs_in_place,
 	.emit_frame_size =
-		2 + /* uvd_v7_0_ring_emit_hdp_flush */
-		2 + /* uvd_v7_0_ring_emit_hdp_invalidate */
-		34 + /* uvd_v7_0_ring_emit_vm_flush */
+		6 + /* hdp invalidate */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
+		8 + /* uvd_v7_0_ring_emit_vm_flush */
 		14 + 14, /* uvd_v7_0_ring_emit_fence x2 vm fence */
 	.emit_ib_size = 8, /* uvd_v7_0_ring_emit_ib */
 	.emit_ib = uvd_v7_0_ring_emit_ib,
 	.emit_fence = uvd_v7_0_ring_emit_fence,
 	.emit_vm_flush = uvd_v7_0_ring_emit_vm_flush,
 	.emit_hdp_flush = uvd_v7_0_ring_emit_hdp_flush,
-	.emit_hdp_invalidate = uvd_v7_0_ring_emit_hdp_invalidate,
 	.test_ring = uvd_v7_0_ring_test_ring,
 	.test_ib = amdgpu_uvd_ring_test_ib,
-	.insert_nop = amdgpu_ring_insert_nop,
+	.insert_nop = uvd_v7_0_ring_insert_nop,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
 	.begin_use = amdgpu_uvd_ring_begin_use,
 	.end_use = amdgpu_uvd_ring_end_use,
+	.emit_wreg = uvd_v7_0_ring_emit_wreg,
+	.emit_reg_wait = uvd_v7_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
 };
 
 static const struct amdgpu_ring_funcs uvd_v7_0_enc_ring_vm_funcs = {
@@ -1717,12 +1570,15 @@ static const struct amdgpu_ring_funcs uvd_v7_0_enc_ring_vm_funcs = {
 	.align_mask = 0x3f,
 	.nop = HEVC_ENC_CMD_NO_OP,
 	.support_64bit_ptrs = false,
-	.vmhub = AMDGPU_MMHUB,
+	.no_user_fence = true,
 	.get_rptr = uvd_v7_0_enc_ring_get_rptr,
 	.get_wptr = uvd_v7_0_enc_ring_get_wptr,
 	.set_wptr = uvd_v7_0_enc_ring_set_wptr,
 	.emit_frame_size =
-		17 + /* uvd_v7_0_enc_ring_emit_vm_flush */
+		3 + 3 + /* hdp flush / invalidate */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 4 +
+		4 + /* uvd_v7_0_enc_ring_emit_vm_flush */
 		5 + 5 + /* uvd_v7_0_enc_ring_emit_fence x2 vm fence */
 		1, /* uvd_v7_0_enc_ring_insert_end */
 	.emit_ib_size = 5, /* uvd_v7_0_enc_ring_emit_ib */
@@ -1736,22 +1592,38 @@ static const struct amdgpu_ring_funcs uvd_v7_0_enc_ring_vm_funcs = {
 	.pad_ib = amdgpu_ring_generic_pad_ib,
 	.begin_use = amdgpu_uvd_ring_begin_use,
 	.end_use = amdgpu_uvd_ring_end_use,
+	.emit_wreg = uvd_v7_0_enc_ring_emit_wreg,
+	.emit_reg_wait = uvd_v7_0_enc_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
 };
 
 static void uvd_v7_0_set_ring_funcs(struct amdgpu_device *adev)
 {
-	adev->uvd.ring.funcs = &uvd_v7_0_ring_vm_funcs;
-	DRM_INFO("UVD is enabled in VM mode\n");
+	int i;
+
+	for (i = 0; i < adev->uvd.num_uvd_inst; i++) {
+		if (adev->uvd.harvest_config & (1 << i))
+			continue;
+		adev->uvd.inst[i].ring.funcs = &uvd_v7_0_ring_vm_funcs;
+		adev->uvd.inst[i].ring.me = i;
+		DRM_INFO("UVD(%d) is enabled in VM mode\n", i);
+	}
 }
 
 static void uvd_v7_0_set_enc_ring_funcs(struct amdgpu_device *adev)
 {
-	int i;
-
-	for (i = 0; i < adev->uvd.num_enc_rings; ++i)
-		adev->uvd.ring_enc[i].funcs = &uvd_v7_0_enc_ring_vm_funcs;
+	int i, j;
+
+	for (j = 0; j < adev->uvd.num_uvd_inst; j++) {
+		if (adev->uvd.harvest_config & (1 << j))
+			continue;
+		for (i = 0; i < adev->uvd.num_enc_rings; ++i) {
+			adev->uvd.inst[j].ring_enc[i].funcs = &uvd_v7_0_enc_ring_vm_funcs;
+			adev->uvd.inst[j].ring_enc[i].me = j;
+		}
 
-	DRM_INFO("UVD ENC is enabled in VM mode\n");
+		DRM_INFO("UVD(%d) ENC is enabled in VM mode\n", j);
+	}
 }
 
 static const struct amdgpu_irq_src_funcs uvd_v7_0_irq_funcs = {
@@ -1761,12 +1633,17 @@ static const struct amdgpu_irq_src_funcs uvd_v7_0_irq_funcs = {
 
 static void uvd_v7_0_set_irq_funcs(struct amdgpu_device *adev)
 {
-	adev->uvd.irq.num_types = adev->uvd.num_enc_rings + 1;
-	adev->uvd.irq.funcs = &uvd_v7_0_irq_funcs;
+	int i;
+
+	for (i = 0; i < adev->uvd.num_uvd_inst; i++) {
+		if (adev->uvd.harvest_config & (1 << i))
+			continue;
+		adev->uvd.inst[i].irq.num_types = adev->uvd.num_enc_rings + 1;
+		adev->uvd.inst[i].irq.funcs = &uvd_v7_0_irq_funcs;
+	}
 }
 
-const struct amdgpu_ip_block_version uvd_v7_0_ip_block =
-{
+const struct amdgpu_ip_block_version uvd_v7_0_ip_block = {
 		.type = AMD_IP_BLOCK_TYPE_UVD,
 		.major = 7,
 		.minor = 0,
diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c
index 47f70827195b..bee3e904a6bc 100644
--- a/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c
@@ -26,7 +26,7 @@
  */
 
 #include <linux/firmware.h>
-#include <drm/drmP.h>
+
 #include "amdgpu.h"
 #include "amdgpu_vce.h"
 #include "cikd.h"
@@ -56,7 +56,7 @@ static uint64_t vce_v2_0_ring_get_rptr(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
 
-	if (ring == &adev->vce.ring[0])
+	if (ring->me == 0)
 		return RREG32(mmVCE_RB_RPTR);
 	else
 		return RREG32(mmVCE_RB_RPTR2);
@@ -73,7 +73,7 @@ static uint64_t vce_v2_0_ring_get_wptr(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
 
-	if (ring == &adev->vce.ring[0])
+	if (ring->me == 0)
 		return RREG32(mmVCE_RB_WPTR);
 	else
 		return RREG32(mmVCE_RB_WPTR2);
@@ -90,7 +90,7 @@ static void vce_v2_0_ring_set_wptr(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
 
-	if (ring == &adev->vce.ring[0])
+	if (ring->me == 0)
 		WREG32(mmVCE_RB_WPTR, lower_32_bits(ring->wptr));
 	else
 		WREG32(mmVCE_RB_WPTR2, lower_32_bits(ring->wptr));
@@ -201,20 +201,20 @@ static void vce_v2_0_mc_resume(struct amdgpu_device *adev)
 	WREG32_FIELD(VCE_SYS_INT_EN, VCE_SYS_INT_TRAP_INTERRUPT_EN, 1);
 }
 
-static bool vce_v2_0_is_idle(void *handle)
+static bool vce_v2_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	return !(RREG32(mmSRBM_STATUS2) & SRBM_STATUS2__VCE_BUSY_MASK);
 }
 
-static int vce_v2_0_wait_for_idle(void *handle)
+static int vce_v2_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	unsigned i;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		if (vce_v2_0_is_idle(handle))
+		if (vce_v2_0_is_idle(ip_block))
 			return 0;
 	}
 	return -ETIMEDOUT;
@@ -274,16 +274,22 @@ static int vce_v2_0_start(struct amdgpu_device *adev)
 
 static int vce_v2_0_stop(struct amdgpu_device *adev)
 {
+	struct amdgpu_ip_block *ip_block;
 	int i;
 	int status;
 
+
 	if (vce_v2_0_lmi_clean(adev)) {
-		DRM_INFO("vce is not idle \n");
+		DRM_INFO("VCE is not idle \n");
 		return 0;
 	}
 
-	if (vce_v2_0_wait_for_idle(adev)) {
-		DRM_INFO("VCE is busy, Can't set clock gateing");
+	ip_block = amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_VCE);
+	if (!ip_block)
+		return -EINVAL;
+
+	if (vce_v2_0_wait_for_idle(ip_block)) {
+		DRM_INFO("VCE is busy, Can't set clock gating");
 		return 0;
 	}
 
@@ -398,9 +404,9 @@ static void vce_v2_0_enable_mgcg(struct amdgpu_device *adev, bool enable,
 	}
 }
 
-static int vce_v2_0_early_init(void *handle)
+static int vce_v2_0_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	adev->vce.num_rings = 2;
 
@@ -410,14 +416,14 @@ static int vce_v2_0_early_init(void *handle)
 	return 0;
 }
 
-static int vce_v2_0_sw_init(void *handle)
+static int vce_v2_0_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_ring *ring;
 	int r, i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	/* VCE */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 167, &adev->vce.irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 167, &adev->vce.irq);
 	if (r)
 		return r;
 
@@ -431,10 +437,12 @@ static int vce_v2_0_sw_init(void *handle)
 		return r;
 
 	for (i = 0; i < adev->vce.num_rings; i++) {
+		enum amdgpu_ring_priority_level hw_prio = amdgpu_vce_get_ring_prio(i);
+
 		ring = &adev->vce.ring[i];
 		sprintf(ring->name, "vce%d", i);
-		r = amdgpu_ring_init(adev, ring, 512,
-				     &adev->vce.irq, 0);
+		r = amdgpu_ring_init(adev, ring, 512, &adev->vce.irq, 0,
+				     hw_prio, NULL);
 		if (r)
 			return r;
 	}
@@ -442,10 +450,10 @@ static int vce_v2_0_sw_init(void *handle)
 	return r;
 }
 
-static int vce_v2_0_sw_fini(void *handle)
+static int vce_v2_0_sw_fini(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	r = amdgpu_vce_suspend(adev);
 	if (r)
@@ -454,22 +462,18 @@ static int vce_v2_0_sw_fini(void *handle)
 	return amdgpu_vce_sw_fini(adev);
 }
 
-static int vce_v2_0_hw_init(void *handle)
+static int vce_v2_0_hw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r, i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	amdgpu_asic_set_vce_clocks(adev, 10000, 10000);
 	vce_v2_0_enable_mgcg(adev, true, false);
-	for (i = 0; i < adev->vce.num_rings; i++)
-		adev->vce.ring[i].ready = false;
 
 	for (i = 0; i < adev->vce.num_rings; i++) {
-		r = amdgpu_ring_test_ring(&adev->vce.ring[i]);
+		r = amdgpu_ring_test_helper(&adev->vce.ring[i]);
 		if (r)
 			return r;
-		else
-			adev->vce.ring[i].ready = true;
 	}
 
 	DRM_INFO("VCE initialized successfully.\n");
@@ -477,38 +481,63 @@ static int vce_v2_0_hw_init(void *handle)
 	return 0;
 }
 
-static int vce_v2_0_hw_fini(void *handle)
+static int vce_v2_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
+	cancel_delayed_work_sync(&ip_block->adev->vce.idle_work);
+
 	return 0;
 }
 
-static int vce_v2_0_suspend(void *handle)
+static int vce_v2_0_suspend(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+
+
+	/*
+	 * Proper cleanups before halting the HW engine:
+	 *   - cancel the delayed idle work
+	 *   - enable powergating
+	 *   - enable clockgating
+	 *   - disable dpm
+	 *
+	 * TODO: to align with the VCN implementation, move the
+	 * jobs for clockgating/powergating/dpm setting to
+	 * ->set_powergating_state().
+	 */
+	cancel_delayed_work_sync(&adev->vce.idle_work);
 
-	r = vce_v2_0_hw_fini(adev);
+	if (adev->pm.dpm_enabled) {
+		amdgpu_dpm_enable_vce(adev, false);
+	} else {
+		amdgpu_asic_set_vce_clocks(adev, 0, 0);
+		amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+						       AMD_PG_STATE_GATE);
+		amdgpu_device_ip_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+						       AMD_CG_STATE_GATE);
+	}
+
+	r = vce_v2_0_hw_fini(ip_block);
 	if (r)
 		return r;
 
 	return amdgpu_vce_suspend(adev);
 }
 
-static int vce_v2_0_resume(void *handle)
+static int vce_v2_0_resume(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	r = amdgpu_vce_resume(adev);
+	r = amdgpu_vce_resume(ip_block->adev);
 	if (r)
 		return r;
 
-	return vce_v2_0_hw_init(adev);
+	return vce_v2_0_hw_init(ip_block);
 }
 
-static int vce_v2_0_soft_reset(void *handle)
+static int vce_v2_0_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	WREG32_FIELD(SRBM_SOFT_RESET, SOFT_RESET_VCE, 1);
 	mdelay(5);
@@ -549,13 +578,13 @@ static int vce_v2_0_process_interrupt(struct amdgpu_device *adev,
 	return 0;
 }
 
-static int vce_v2_0_set_clockgating_state(void *handle,
+static int vce_v2_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
 	bool gate = false;
 	bool sw_cg = false;
 
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (state == AMD_CG_STATE_GATE) {
 		gate = true;
@@ -567,7 +596,7 @@ static int vce_v2_0_set_clockgating_state(void *handle,
 	return 0;
 }
 
-static int vce_v2_0_set_powergating_state(void *handle,
+static int vce_v2_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	/* This doesn't actually powergate the VCE block.
@@ -577,7 +606,7 @@ static int vce_v2_0_set_powergating_state(void *handle,
 	 * revisit this when there is a cleaner line between
 	 * the smc and the hw blocks
 	 */
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (state == AMD_PG_STATE_GATE)
 		return vce_v2_0_stop(adev);
@@ -588,7 +617,6 @@ static int vce_v2_0_set_powergating_state(void *handle,
 static const struct amd_ip_funcs vce_v2_0_ip_funcs = {
 	.name = "vce_v2_0",
 	.early_init = vce_v2_0_early_init,
-	.late_init = NULL,
 	.sw_init = vce_v2_0_sw_init,
 	.sw_fini = vce_v2_0_sw_fini,
 	.hw_init = vce_v2_0_hw_init,
@@ -607,6 +635,7 @@ static const struct amdgpu_ring_funcs vce_v2_0_ring_funcs = {
 	.align_mask = 0xf,
 	.nop = VCE_CMD_NO_OP,
 	.support_64bit_ptrs = false,
+	.no_user_fence = true,
 	.get_rptr = vce_v2_0_ring_get_rptr,
 	.get_wptr = vce_v2_0_ring_get_wptr,
 	.set_wptr = vce_v2_0_ring_set_wptr,
@@ -627,8 +656,10 @@ static void vce_v2_0_set_ring_funcs(struct amdgpu_device *adev)
 {
 	int i;
 
-	for (i = 0; i < adev->vce.num_rings; i++)
+	for (i = 0; i < adev->vce.num_rings; i++) {
 		adev->vce.ring[i].funcs = &vce_v2_0_ring_funcs;
+		adev->vce.ring[i].me = i;
+	}
 }
 
 static const struct amdgpu_irq_src_funcs vce_v2_0_irq_funcs = {
diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c
index 90332f55cfba..708123899c41 100644
--- a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c
@@ -26,7 +26,7 @@
  */
 
 #include <linux/firmware.h>
-#include <drm/drmP.h>
+
 #include "amdgpu.h"
 #include "amdgpu_vce.h"
 #include "vid.h"
@@ -37,8 +37,8 @@
 #include "gca/gfx_8_0_d.h"
 #include "smu/smu_7_1_2_d.h"
 #include "smu/smu_7_1_2_sh_mask.h"
-#include "gca/gfx_8_0_d.h"
 #include "gca/gfx_8_0_sh_mask.h"
+#include "ivsrcid/ivsrcid_vislands30.h"
 
 
 #define GRBM_GFX_INDEX__VCE_INSTANCE__SHIFT	0x04
@@ -64,8 +64,8 @@
 static void vce_v3_0_mc_resume(struct amdgpu_device *adev, int idx);
 static void vce_v3_0_set_ring_funcs(struct amdgpu_device *adev);
 static void vce_v3_0_set_irq_funcs(struct amdgpu_device *adev);
-static int vce_v3_0_wait_for_idle(void *handle);
-static int vce_v3_0_set_clockgating_state(void *handle,
+static int vce_v3_0_wait_for_idle(struct amdgpu_ip_block *ip_block);
+static int vce_v3_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state);
 /**
  * vce_v3_0_ring_get_rptr - get read pointer
@@ -86,9 +86,9 @@ static uint64_t vce_v3_0_ring_get_rptr(struct amdgpu_ring *ring)
 	else if (adev->vce.harvest_config == AMDGPU_VCE_HARVEST_VCE0)
 		WREG32(mmGRBM_GFX_INDEX, GET_VCE_INSTANCE(1));
 
-	if (ring == &adev->vce.ring[0])
+	if (ring->me == 0)
 		v = RREG32(mmVCE_RB_RPTR);
-	else if (ring == &adev->vce.ring[1])
+	else if (ring->me == 1)
 		v = RREG32(mmVCE_RB_RPTR2);
 	else
 		v = RREG32(mmVCE_RB_RPTR3);
@@ -118,9 +118,9 @@ static uint64_t vce_v3_0_ring_get_wptr(struct amdgpu_ring *ring)
 	else if (adev->vce.harvest_config == AMDGPU_VCE_HARVEST_VCE0)
 		WREG32(mmGRBM_GFX_INDEX, GET_VCE_INSTANCE(1));
 
-	if (ring == &adev->vce.ring[0])
+	if (ring->me == 0)
 		v = RREG32(mmVCE_RB_WPTR);
-	else if (ring == &adev->vce.ring[1])
+	else if (ring->me == 1)
 		v = RREG32(mmVCE_RB_WPTR2);
 	else
 		v = RREG32(mmVCE_RB_WPTR3);
@@ -149,9 +149,9 @@ static void vce_v3_0_ring_set_wptr(struct amdgpu_ring *ring)
 	else if (adev->vce.harvest_config == AMDGPU_VCE_HARVEST_VCE0)
 		WREG32(mmGRBM_GFX_INDEX, GET_VCE_INSTANCE(1));
 
-	if (ring == &adev->vce.ring[0])
+	if (ring->me == 0)
 		WREG32(mmVCE_RB_WPTR, lower_32_bits(ring->wptr));
-	else if (ring == &adev->vce.ring[1])
+	else if (ring->me == 1)
 		WREG32(mmVCE_RB_WPTR2, lower_32_bits(ring->wptr));
 	else
 		WREG32(mmVCE_RB_WPTR3, lower_32_bits(ring->wptr));
@@ -365,15 +365,10 @@ static unsigned vce_v3_0_get_harvest_config(struct amdgpu_device *adev)
 {
 	u32 tmp;
 
-	/* Fiji, Stoney, Polaris10, Polaris11, Polaris12 are single pipe */
 	if ((adev->asic_type == CHIP_FIJI) ||
-	    (adev->asic_type == CHIP_STONEY) ||
-	    (adev->asic_type == CHIP_POLARIS10) ||
-	    (adev->asic_type == CHIP_POLARIS11) ||
-	    (adev->asic_type == CHIP_POLARIS12))
+	    (adev->asic_type == CHIP_STONEY))
 		return AMDGPU_VCE_HARVEST_VCE1;
 
-	/* Tonga and CZ are dual or single pipe */
 	if (adev->flags & AMD_IS_APU)
 		tmp = (RREG32_SMC(ixVCE_HARVEST_FUSE_MACRO__ADDRESS) &
 		       VCE_HARVEST_FUSE_MACRO__MASK) >>
@@ -391,13 +386,19 @@ static unsigned vce_v3_0_get_harvest_config(struct amdgpu_device *adev)
 	case 3:
 		return AMDGPU_VCE_HARVEST_VCE0 | AMDGPU_VCE_HARVEST_VCE1;
 	default:
+		if ((adev->asic_type == CHIP_POLARIS10) ||
+		    (adev->asic_type == CHIP_POLARIS11) ||
+		    (adev->asic_type == CHIP_POLARIS12) ||
+		    (adev->asic_type == CHIP_VEGAM))
+			return AMDGPU_VCE_HARVEST_VCE1;
+
 		return 0;
 	}
 }
 
-static int vce_v3_0_early_init(void *handle)
+static int vce_v3_0_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	adev->vce.harvest_config = vce_v3_0_get_harvest_config(adev);
 
@@ -414,14 +415,14 @@ static int vce_v3_0_early_init(void *handle)
 	return 0;
 }
 
-static int vce_v3_0_sw_init(void *handle)
+static int vce_v3_0_sw_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	struct amdgpu_ring *ring;
 	int r, i;
 
 	/* VCE */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 167, &adev->vce.irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_VCE_TRAP, &adev->vce.irq);
 	if (r)
 		return r;
 
@@ -439,9 +440,12 @@ static int vce_v3_0_sw_init(void *handle)
 		return r;
 
 	for (i = 0; i < adev->vce.num_rings; i++) {
+		enum amdgpu_ring_priority_level hw_prio = amdgpu_vce_get_ring_prio(i);
+
 		ring = &adev->vce.ring[i];
 		sprintf(ring->name, "vce%d", i);
-		r = amdgpu_ring_init(adev, ring, 512, &adev->vce.irq, 0);
+		r = amdgpu_ring_init(adev, ring, 512, &adev->vce.irq, 0,
+				     hw_prio, NULL);
 		if (r)
 			return r;
 	}
@@ -449,10 +453,10 @@ static int vce_v3_0_sw_init(void *handle)
 	return r;
 }
 
-static int vce_v3_0_sw_fini(void *handle)
+static int vce_v3_0_sw_fini(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	r = amdgpu_vce_suspend(adev);
 	if (r)
@@ -461,24 +465,19 @@ static int vce_v3_0_sw_fini(void *handle)
 	return amdgpu_vce_sw_fini(adev);
 }
 
-static int vce_v3_0_hw_init(void *handle)
+static int vce_v3_0_hw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r, i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	vce_v3_0_override_vce_clock_gating(adev, true);
-	if (!(adev->flags & AMD_IS_APU))
-		amdgpu_asic_set_vce_clocks(adev, 10000, 10000);
 
-	for (i = 0; i < adev->vce.num_rings; i++)
-		adev->vce.ring[i].ready = false;
+	amdgpu_asic_set_vce_clocks(adev, 10000, 10000);
 
 	for (i = 0; i < adev->vce.num_rings; i++) {
-		r = amdgpu_ring_test_ring(&adev->vce.ring[i]);
+		r = amdgpu_ring_test_helper(&adev->vce.ring[i]);
 		if (r)
 			return r;
-		else
-			adev->vce.ring[i].ready = true;
 	}
 
 	DRM_INFO("VCE initialized successfully.\n");
@@ -486,41 +485,65 @@ static int vce_v3_0_hw_init(void *handle)
 	return 0;
 }
 
-static int vce_v3_0_hw_fini(void *handle)
+static int vce_v3_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	cancel_delayed_work_sync(&adev->vce.idle_work);
 
-	r = vce_v3_0_wait_for_idle(handle);
+	r = vce_v3_0_wait_for_idle(ip_block);
 	if (r)
 		return r;
 
 	vce_v3_0_stop(adev);
-	return vce_v3_0_set_clockgating_state(adev, AMD_CG_STATE_GATE);
+	return vce_v3_0_set_clockgating_state(ip_block, AMD_CG_STATE_GATE);
 }
 
-static int vce_v3_0_suspend(void *handle)
+static int vce_v3_0_suspend(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	/*
+	 * Proper cleanups before halting the HW engine:
+	 *   - cancel the delayed idle work
+	 *   - enable powergating
+	 *   - enable clockgating
+	 *   - disable dpm
+	 *
+	 * TODO: to align with the VCN implementation, move the
+	 * jobs for clockgating/powergating/dpm setting to
+	 * ->set_powergating_state().
+	 */
+	cancel_delayed_work_sync(&adev->vce.idle_work);
+
+	if (adev->pm.dpm_enabled) {
+		amdgpu_dpm_enable_vce(adev, false);
+	} else {
+		amdgpu_asic_set_vce_clocks(adev, 0, 0);
+		amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+						       AMD_PG_STATE_GATE);
+		amdgpu_device_ip_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+						       AMD_CG_STATE_GATE);
+	}
 
-	r = vce_v3_0_hw_fini(adev);
+	r = vce_v3_0_hw_fini(ip_block);
 	if (r)
 		return r;
 
 	return amdgpu_vce_suspend(adev);
 }
 
-static int vce_v3_0_resume(void *handle)
+static int vce_v3_0_resume(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	r = amdgpu_vce_resume(adev);
+	r = amdgpu_vce_resume(ip_block->adev);
 	if (r)
 		return r;
 
-	return vce_v3_0_hw_init(adev);
+	return vce_v3_0_hw_init(ip_block);
 }
 
 static void vce_v3_0_mc_resume(struct amdgpu_device *adev, int idx)
@@ -574,9 +597,9 @@ static void vce_v3_0_mc_resume(struct amdgpu_device *adev, int idx)
 	WREG32_FIELD(VCE_SYS_INT_EN, VCE_SYS_INT_TRAP_INTERRUPT_EN, 1);
 }
 
-static bool vce_v3_0_is_idle(void *handle)
+static bool vce_v3_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 mask = 0;
 
 	mask |= (adev->vce.harvest_config & AMDGPU_VCE_HARVEST_VCE0) ? 0 : SRBM_STATUS2__VCE0_BUSY_MASK;
@@ -585,13 +608,13 @@ static bool vce_v3_0_is_idle(void *handle)
 	return !(RREG32(mmSRBM_STATUS2) & mask);
 }
 
-static int vce_v3_0_wait_for_idle(void *handle)
+static int vce_v3_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
 	unsigned i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++)
-		if (vce_v3_0_is_idle(handle))
+		if (vce_v3_0_is_idle(ip_block))
 			return 0;
 
 	return -ETIMEDOUT;
@@ -603,9 +626,9 @@ static int vce_v3_0_wait_for_idle(void *handle)
 #define  AMDGPU_VCE_STATUS_BUSY_MASK (VCE_STATUS_VCPU_REPORT_AUTO_BUSY_MASK | \
 				      VCE_STATUS_VCPU_REPORT_RB0_BUSY_MASK)
 
-static bool vce_v3_0_check_soft_reset(void *handle)
+static bool vce_v3_0_check_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 srbm_soft_reset = 0;
 
 	/* According to VCE team , we should use VCE_STATUS instead
@@ -644,9 +667,9 @@ static bool vce_v3_0_check_soft_reset(void *handle)
 	}
 }
 
-static int vce_v3_0_soft_reset(void *handle)
+static int vce_v3_0_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 srbm_soft_reset;
 
 	if (!adev->vce.srbm_soft_reset)
@@ -675,29 +698,29 @@ static int vce_v3_0_soft_reset(void *handle)
 	return 0;
 }
 
-static int vce_v3_0_pre_soft_reset(void *handle)
+static int vce_v3_0_pre_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (!adev->vce.srbm_soft_reset)
 		return 0;
 
 	mdelay(5);
 
-	return vce_v3_0_suspend(adev);
+	return vce_v3_0_suspend(ip_block);
 }
 
 
-static int vce_v3_0_post_soft_reset(void *handle)
+static int vce_v3_0_post_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (!adev->vce.srbm_soft_reset)
 		return 0;
 
 	mdelay(5);
 
-	return vce_v3_0_resume(adev);
+	return vce_v3_0_resume(ip_block);
 }
 
 static int vce_v3_0_set_interrupt_state(struct amdgpu_device *adev,
@@ -737,11 +760,11 @@ static int vce_v3_0_process_interrupt(struct amdgpu_device *adev,
 	return 0;
 }
 
-static int vce_v3_0_set_clockgating_state(void *handle,
+static int vce_v3_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	bool enable = (state == AMD_CG_STATE_GATE) ? true : false;
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = (state == AMD_CG_STATE_GATE);
 	int i;
 
 	if (!(adev->cg_flags & AMD_CG_SUPPORT_VCE_MGCG))
@@ -778,7 +801,7 @@ static int vce_v3_0_set_clockgating_state(void *handle,
 	return 0;
 }
 
-static int vce_v3_0_set_powergating_state(void *handle,
+static int vce_v3_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	/* This doesn't actually powergate the VCE block.
@@ -788,7 +811,7 @@ static int vce_v3_0_set_powergating_state(void *handle,
 	 * revisit this when there is a cleaner line between
 	 * the smc and the hw blocks
 	 */
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int ret = 0;
 
 	if (state == AMD_PG_STATE_GATE) {
@@ -805,9 +828,9 @@ out:
 	return ret;
 }
 
-static void vce_v3_0_get_clockgating_state(void *handle, u32 *flags)
+static void vce_v3_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int data;
 
 	mutex_lock(&adev->pm.mutex);
@@ -834,24 +857,28 @@ out:
 }
 
 static void vce_v3_0_ring_emit_ib(struct amdgpu_ring *ring,
-		struct amdgpu_ib *ib, unsigned int vm_id, bool ctx_switch)
+				  struct amdgpu_job *job,
+				  struct amdgpu_ib *ib,
+				  uint32_t flags)
 {
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+
 	amdgpu_ring_write(ring, VCE_CMD_IB_VM);
-	amdgpu_ring_write(ring, vm_id);
+	amdgpu_ring_write(ring, vmid);
 	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
 	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
 	amdgpu_ring_write(ring, ib->length_dw);
 }
 
 static void vce_v3_0_emit_vm_flush(struct amdgpu_ring *ring,
-			 unsigned int vm_id, uint64_t pd_addr)
+				   unsigned int vmid, uint64_t pd_addr)
 {
 	amdgpu_ring_write(ring, VCE_CMD_UPDATE_PTB);
-	amdgpu_ring_write(ring, vm_id);
+	amdgpu_ring_write(ring, vmid);
 	amdgpu_ring_write(ring, pd_addr >> 12);
 
 	amdgpu_ring_write(ring, VCE_CMD_FLUSH_TLB);
-	amdgpu_ring_write(ring, vm_id);
+	amdgpu_ring_write(ring, vmid);
 	amdgpu_ring_write(ring, VCE_CMD_END);
 }
 
@@ -869,7 +896,6 @@ static void vce_v3_0_emit_pipeline_sync(struct amdgpu_ring *ring)
 static const struct amd_ip_funcs vce_v3_0_ip_funcs = {
 	.name = "vce_v3_0",
 	.early_init = vce_v3_0_early_init,
-	.late_init = NULL,
 	.sw_init = vce_v3_0_sw_init,
 	.sw_fini = vce_v3_0_sw_fini,
 	.hw_init = vce_v3_0_hw_init,
@@ -892,6 +918,7 @@ static const struct amdgpu_ring_funcs vce_v3_0_ring_phys_funcs = {
 	.align_mask = 0xf,
 	.nop = VCE_CMD_NO_OP,
 	.support_64bit_ptrs = false,
+	.no_user_fence = true,
 	.get_rptr = vce_v3_0_ring_get_rptr,
 	.get_wptr = vce_v3_0_ring_get_wptr,
 	.set_wptr = vce_v3_0_ring_set_wptr,
@@ -899,7 +926,7 @@ static const struct amdgpu_ring_funcs vce_v3_0_ring_phys_funcs = {
 	.emit_frame_size =
 		4 + /* vce_v3_0_emit_pipeline_sync */
 		6, /* amdgpu_vce_ring_emit_fence x1 no user fence */
-	.emit_ib_size = 5, /* vce_v3_0_ring_emit_ib */
+	.emit_ib_size = 4, /* amdgpu_vce_ring_emit_ib */
 	.emit_ib = amdgpu_vce_ring_emit_ib,
 	.emit_fence = amdgpu_vce_ring_emit_fence,
 	.test_ring = amdgpu_vce_ring_test_ring,
@@ -915,15 +942,16 @@ static const struct amdgpu_ring_funcs vce_v3_0_ring_vm_funcs = {
 	.align_mask = 0xf,
 	.nop = VCE_CMD_NO_OP,
 	.support_64bit_ptrs = false,
+	.no_user_fence = true,
 	.get_rptr = vce_v3_0_ring_get_rptr,
 	.get_wptr = vce_v3_0_ring_get_wptr,
 	.set_wptr = vce_v3_0_ring_set_wptr,
-	.parse_cs = amdgpu_vce_ring_parse_cs_vm,
+	.patch_cs_in_place = amdgpu_vce_ring_parse_cs_vm,
 	.emit_frame_size =
 		6 + /* vce_v3_0_emit_vm_flush */
 		4 + /* vce_v3_0_emit_pipeline_sync */
 		6 + 6, /* amdgpu_vce_ring_emit_fence x2 vm fence */
-	.emit_ib_size = 4, /* amdgpu_vce_ring_emit_ib */
+	.emit_ib_size = 5, /* vce_v3_0_ring_emit_ib */
 	.emit_ib = vce_v3_0_ring_emit_ib,
 	.emit_vm_flush = vce_v3_0_emit_vm_flush,
 	.emit_pipeline_sync = vce_v3_0_emit_pipeline_sync,
@@ -941,12 +969,16 @@ static void vce_v3_0_set_ring_funcs(struct amdgpu_device *adev)
 	int i;
 
 	if (adev->asic_type >= CHIP_STONEY) {
-		for (i = 0; i < adev->vce.num_rings; i++)
+		for (i = 0; i < adev->vce.num_rings; i++) {
 			adev->vce.ring[i].funcs = &vce_v3_0_ring_vm_funcs;
+			adev->vce.ring[i].me = i;
+		}
 		DRM_INFO("VCE enabled in VM mode\n");
 	} else {
-		for (i = 0; i < adev->vce.num_rings; i++)
+		for (i = 0; i < adev->vce.num_rings; i++) {
 			adev->vce.ring[i].funcs = &vce_v3_0_ring_phys_funcs;
+			adev->vce.ring[i].me = i;
+		}
 		DRM_INFO("VCE enabled in physical mode\n");
 	}
 }
@@ -962,8 +994,7 @@ static void vce_v3_0_set_irq_funcs(struct amdgpu_device *adev)
 	adev->vce.irq.funcs = &vce_v3_0_irq_funcs;
 };
 
-const struct amdgpu_ip_block_version vce_v3_0_ip_block =
-{
+const struct amdgpu_ip_block_version vce_v3_0_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_VCE,
 	.major = 3,
 	.minor = 0,
@@ -971,8 +1002,7 @@ const struct amdgpu_ip_block_version vce_v3_0_ip_block =
 	.funcs = &vce_v3_0_ip_funcs,
 };
 
-const struct amdgpu_ip_block_version vce_v3_1_ip_block =
-{
+const struct amdgpu_ip_block_version vce_v3_1_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_VCE,
 	.major = 3,
 	.minor = 1,
@@ -980,8 +1010,7 @@ const struct amdgpu_ip_block_version vce_v3_1_ip_block =
 	.funcs = &vce_v3_0_ip_funcs,
 };
 
-const struct amdgpu_ip_block_version vce_v3_4_ip_block =
-{
+const struct amdgpu_ip_block_version vce_v3_4_ip_block = {
 	.type = AMD_IP_BLOCK_TYPE_VCE,
 	.major = 3,
 	.minor = 4,
diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c
index 11134d5f7443..335bda64ff5b 100644
--- a/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c
@@ -25,19 +25,22 @@
  */
 
 #include <linux/firmware.h>
-#include <drm/drmP.h>
+#include <drm/drm_drv.h>
+
 #include "amdgpu.h"
 #include "amdgpu_vce.h"
+#include "soc15.h"
 #include "soc15d.h"
 #include "soc15_common.h"
 #include "mmsch_v1_0.h"
 
-#include "vega10/soc15ip.h"
-#include "vega10/VCE/vce_4_0_offset.h"
-#include "vega10/VCE/vce_4_0_default.h"
-#include "vega10/VCE/vce_4_0_sh_mask.h"
-#include "vega10/MMHUB/mmhub_1_0_offset.h"
-#include "vega10/MMHUB/mmhub_1_0_sh_mask.h"
+#include "vce/vce_4_0_offset.h"
+#include "vce/vce_4_0_default.h"
+#include "vce/vce_4_0_sh_mask.h"
+#include "mmhub/mmhub_1_0_offset.h"
+#include "mmhub/mmhub_1_0_sh_mask.h"
+
+#include "ivsrcid/vce/irqsrcs_vce_4_0.h"
 
 #define VCE_STATUS_VCPU_REPORT_FW_LOADED_MASK	0x02
 
@@ -60,9 +63,9 @@ static uint64_t vce_v4_0_ring_get_rptr(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
 
-	if (ring == &adev->vce.ring[0])
+	if (ring->me == 0)
 		return RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_RPTR));
-	else if (ring == &adev->vce.ring[1])
+	else if (ring->me == 1)
 		return RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_RPTR2));
 	else
 		return RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_RPTR3));
@@ -80,11 +83,11 @@ static uint64_t vce_v4_0_ring_get_wptr(struct amdgpu_ring *ring)
 	struct amdgpu_device *adev = ring->adev;
 
 	if (ring->use_doorbell)
-		return adev->wb.wb[ring->wptr_offs];
+		return *ring->wptr_cpu_addr;
 
-	if (ring == &adev->vce.ring[0])
+	if (ring->me == 0)
 		return RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR));
-	else if (ring == &adev->vce.ring[1])
+	else if (ring->me == 1)
 		return RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR2));
 	else
 		return RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR3));
@@ -103,15 +106,15 @@ static void vce_v4_0_ring_set_wptr(struct amdgpu_ring *ring)
 
 	if (ring->use_doorbell) {
 		/* XXX check if swapping is necessary on BE */
-		adev->wb.wb[ring->wptr_offs] = lower_32_bits(ring->wptr);
+		*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
 		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
 		return;
 	}
 
-	if (ring == &adev->vce.ring[0])
+	if (ring->me == 0)
 		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR),
 			lower_32_bits(ring->wptr));
-	else if (ring == &adev->vce.ring[1])
+	else if (ring->me == 1)
 		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR2),
 			lower_32_bits(ring->wptr));
 	else
@@ -174,7 +177,7 @@ static int vce_v4_0_mmsch_start(struct amdgpu_device *adev,
 	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_MMSCH_VF_MAILBOX_RESP), 0);
 
 	WDOORBELL32(adev->vce.ring[0].doorbell_index, 0);
-	adev->wb.wb[adev->vce.ring[0].wptr_offs] = 0;
+	*adev->vce.ring[0].wptr_cpu_addr = 0;
 	adev->vce.ring[0].wptr = 0;
 	adev->vce.ring[0].wptr_old = 0;
 
@@ -242,38 +245,55 @@ static int vce_v4_0_sriov_start(struct amdgpu_device *adev)
 		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_SWAP_CNTL1), 0);
 		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VM_CTRL), 0);
 
+		offset = AMDGPU_VCE_FIRMWARE_OFFSET;
 		if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
-		    MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_40BIT_BAR0),
-						adev->firmware.ucode[AMDGPU_UCODE_ID_VCE].mc_addr >> 8);
-		    MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_40BIT_BAR1),
-						adev->firmware.ucode[AMDGPU_UCODE_ID_VCE].mc_addr >> 8);
-		    MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_40BIT_BAR2),
-						adev->firmware.ucode[AMDGPU_UCODE_ID_VCE].mc_addr >> 8);
+			uint32_t low = adev->firmware.ucode[AMDGPU_UCODE_ID_VCE].tmr_mc_addr_lo;
+			uint32_t hi = adev->firmware.ucode[AMDGPU_UCODE_ID_VCE].tmr_mc_addr_hi;
+			uint64_t tmr_mc_addr = (uint64_t)(hi) << 32 | low;
+
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0,
+						mmVCE_LMI_VCPU_CACHE_40BIT_BAR0), tmr_mc_addr >> 8);
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0,
+						mmVCE_LMI_VCPU_CACHE_64BIT_BAR0),
+						(tmr_mc_addr >> 40) & 0xff);
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_OFFSET0), 0);
 		} else {
-		    MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_40BIT_BAR0),
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0,
+						mmVCE_LMI_VCPU_CACHE_40BIT_BAR0),
 						adev->vce.gpu_addr >> 8);
-		    MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_40BIT_BAR1),
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0,
+						mmVCE_LMI_VCPU_CACHE_64BIT_BAR0),
+						(adev->vce.gpu_addr >> 40) & 0xff);
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_OFFSET0),
+						offset & ~0x0f000000);
+
+		}
+		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0,
+						mmVCE_LMI_VCPU_CACHE_40BIT_BAR1),
 						adev->vce.gpu_addr >> 8);
-		    MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_40BIT_BAR2),
+		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0,
+						mmVCE_LMI_VCPU_CACHE_64BIT_BAR1),
+						(adev->vce.gpu_addr >> 40) & 0xff);
+		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0,
+						mmVCE_LMI_VCPU_CACHE_40BIT_BAR2),
 						adev->vce.gpu_addr >> 8);
-		}
+		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0,
+						mmVCE_LMI_VCPU_CACHE_64BIT_BAR2),
+						(adev->vce.gpu_addr >> 40) & 0xff);
 
-		offset = AMDGPU_VCE_FIRMWARE_OFFSET;
 		size = VCE_V4_0_FW_SIZE;
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_OFFSET0),
-					    offset & 0x7FFFFFFF);
 		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_SIZE0), size);
 
-		offset += size;
+		offset = (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) ? offset + size : 0;
 		size = VCE_V4_0_STACK_SIZE;
 		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_OFFSET1),
-					    offset & 0x7FFFFFFF);
+					(offset & ~0x0f000000) | (1 << 24));
 		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_SIZE1), size);
 
 		offset += size;
 		size = VCE_V4_0_DATA_SIZE;
 		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_OFFSET2),
-					    offset & 0x7FFFFFFF);
+					(offset & ~0x0f000000) | (2 << 24));
 		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_SIZE2), size);
 
 		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_CTRL2), ~0x100, 0);
@@ -368,6 +388,7 @@ static int vce_v4_0_start(struct amdgpu_device *adev)
 static int vce_v4_0_stop(struct amdgpu_device *adev)
 {
 
+	/* Disable VCPU */
 	WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CNTL), 0, ~0x200001);
 
 	/* hold on ECPU */
@@ -375,8 +396,8 @@ static int vce_v4_0_stop(struct amdgpu_device *adev)
 			VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK,
 			~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK);
 
-	/* clear BUSY flag */
-	WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_STATUS), 0, ~VCE_STATUS__JOB_BUSY_MASK);
+	/* clear VCE_STATUS */
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_STATUS), 0);
 
 	/* Set Clock-Gating off */
 	/* if (adev->cg_flags & AMD_CG_SUPPORT_VCE_MGCG)
@@ -386,9 +407,9 @@ static int vce_v4_0_stop(struct amdgpu_device *adev)
 	return 0;
 }
 
-static int vce_v4_0_early_init(void *handle)
+static int vce_v4_0_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (amdgpu_sriov_vf(adev)) /* currently only VCN0 support SRIOV */
 		adev->vce.num_rings = 1;
@@ -401,18 +422,19 @@ static int vce_v4_0_early_init(void *handle)
 	return 0;
 }
 
-static int vce_v4_0_sw_init(void *handle)
+static int vce_v4_0_sw_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	struct amdgpu_ring *ring;
+
 	unsigned size;
 	int r, i;
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_VCE0, 167, &adev->vce.irq);
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCE0, 167, &adev->vce.irq);
 	if (r)
 		return r;
 
-	size  = (VCE_V4_0_STACK_SIZE + VCE_V4_0_DATA_SIZE) * 2;
+	size  = VCE_V4_0_STACK_SIZE + VCE_V4_0_DATA_SIZE;
 	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
 		size += VCE_V4_0_FW_SIZE;
 
@@ -424,7 +446,7 @@ static int vce_v4_0_sw_init(void *handle)
 		const struct common_firmware_header *hdr;
 		unsigned size = amdgpu_bo_size(adev->vce.vcpu_bo);
 
-		adev->vce.saved_bo = kmalloc(size, GFP_KERNEL);
+		adev->vce.saved_bo = kvmalloc(size, GFP_KERNEL);
 		if (!adev->vce.saved_bo)
 			return -ENOMEM;
 
@@ -441,7 +463,10 @@ static int vce_v4_0_sw_init(void *handle)
 	}
 
 	for (i = 0; i < adev->vce.num_rings; i++) {
+		enum amdgpu_ring_priority_level hw_prio = amdgpu_vce_get_ring_prio(i);
+
 		ring = &adev->vce.ring[i];
+		ring->vm_hub = AMDGPU_MMHUB0(0);
 		sprintf(ring->name, "vce%d", i);
 		if (amdgpu_sriov_vf(adev)) {
 			/* DOORBELL only works under SRIOV */
@@ -451,11 +476,12 @@ static int vce_v4_0_sw_init(void *handle)
 			 * so set unused location for other unused rings.
 			 */
 			if (i == 0)
-				ring->doorbell_index = AMDGPU_DOORBELL64_VCE_RING0_1 * 2;
+				ring->doorbell_index = adev->doorbell_index.uvd_vce.vce_ring0_1 * 2;
 			else
-				ring->doorbell_index = AMDGPU_DOORBELL64_VCE_RING2_3 * 2 + 1;
+				ring->doorbell_index = adev->doorbell_index.uvd_vce.vce_ring2_3 * 2 + 1;
 		}
-		r = amdgpu_ring_init(adev, ring, 512, &adev->vce.irq, 0);
+		r = amdgpu_ring_init(adev, ring, 512, &adev->vce.irq, 0,
+				     hw_prio, NULL);
 		if (r)
 			return r;
 	}
@@ -467,16 +493,16 @@ static int vce_v4_0_sw_init(void *handle)
 	return r;
 }
 
-static int vce_v4_0_sw_fini(void *handle)
+static int vce_v4_0_sw_fini(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	/* free MM table */
 	amdgpu_virt_free_mm_table(adev);
 
 	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
-		kfree(adev->vce.saved_bo);
+		kvfree(adev->vce.saved_bo);
 		adev->vce.saved_bo = NULL;
 	}
 
@@ -487,10 +513,10 @@ static int vce_v4_0_sw_fini(void *handle)
 	return amdgpu_vce_sw_fini(adev);
 }
 
-static int vce_v4_0_hw_init(void *handle)
+static int vce_v4_0_hw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r, i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (amdgpu_sriov_vf(adev))
 		r = vce_v4_0_sriov_start(adev);
@@ -499,15 +525,10 @@ static int vce_v4_0_hw_init(void *handle)
 	if (r)
 		return r;
 
-	for (i = 0; i < adev->vce.num_rings; i++)
-		adev->vce.ring[i].ready = false;
-
 	for (i = 0; i < adev->vce.num_rings; i++) {
-		r = amdgpu_ring_test_ring(&adev->vce.ring[i]);
+		r = amdgpu_ring_test_helper(&adev->vce.ring[i]);
 		if (r)
 			return r;
-		else
-			adev->vce.ring[i].ready = true;
 	}
 
 	DRM_INFO("VCE initialized successfully.\n");
@@ -515,72 +536,101 @@ static int vce_v4_0_hw_init(void *handle)
 	return 0;
 }
 
-static int vce_v4_0_hw_fini(void *handle)
+static int vce_v4_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	int i;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	cancel_delayed_work_sync(&adev->vce.idle_work);
 
 	if (!amdgpu_sriov_vf(adev)) {
-		/* vce_v4_0_wait_for_idle(handle); */
+		/* vce_v4_0_wait_for_idle(ip_block); */
 		vce_v4_0_stop(adev);
 	} else {
 		/* full access mode, so don't touch any VCE register */
 		DRM_DEBUG("For SRIOV client, shouldn't do anything.\n");
 	}
 
-	for (i = 0; i < adev->vce.num_rings; i++)
-		adev->vce.ring[i].ready = false;
-
 	return 0;
 }
 
-static int vce_v4_0_suspend(void *handle)
+static int vce_v4_0_suspend(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+	int r, idx;
 
 	if (adev->vce.vcpu_bo == NULL)
 		return 0;
 
-	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
-		unsigned size = amdgpu_bo_size(adev->vce.vcpu_bo);
-		void *ptr = adev->vce.cpu_addr;
+	if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+			unsigned size = amdgpu_bo_size(adev->vce.vcpu_bo);
+			void *ptr = adev->vce.cpu_addr;
+
+			memcpy_fromio(adev->vce.saved_bo, ptr, size);
+		}
+		drm_dev_exit(idx);
+	}
+
+	/*
+	 * Proper cleanups before halting the HW engine:
+	 *   - cancel the delayed idle work
+	 *   - enable powergating
+	 *   - enable clockgating
+	 *   - disable dpm
+	 *
+	 * TODO: to align with the VCN implementation, move the
+	 * jobs for clockgating/powergating/dpm setting to
+	 * ->set_powergating_state().
+	 */
+	cancel_delayed_work_sync(&adev->vce.idle_work);
 
-		memcpy_fromio(adev->vce.saved_bo, ptr, size);
+	if (adev->pm.dpm_enabled) {
+		amdgpu_dpm_enable_vce(adev, false);
+	} else {
+		amdgpu_asic_set_vce_clocks(adev, 0, 0);
+		amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+						       AMD_PG_STATE_GATE);
+		amdgpu_device_ip_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+						       AMD_CG_STATE_GATE);
 	}
 
-	r = vce_v4_0_hw_fini(adev);
+	r = vce_v4_0_hw_fini(ip_block);
 	if (r)
 		return r;
 
 	return amdgpu_vce_suspend(adev);
 }
 
-static int vce_v4_0_resume(void *handle)
+static int vce_v4_0_resume(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+	int r, idx;
 
 	if (adev->vce.vcpu_bo == NULL)
 		return -EINVAL;
 
 	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
-		unsigned size = amdgpu_bo_size(adev->vce.vcpu_bo);
-		void *ptr = adev->vce.cpu_addr;
 
-		memcpy_toio(ptr, adev->vce.saved_bo, size);
+		if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+			unsigned size = amdgpu_bo_size(adev->vce.vcpu_bo);
+			void *ptr = adev->vce.cpu_addr;
+
+			memcpy_toio(ptr, adev->vce.saved_bo, size);
+			drm_dev_exit(idx);
+		}
 	} else {
 		r = amdgpu_vce_resume(adev);
 		if (r)
 			return r;
 	}
 
-	return vce_v4_0_hw_init(adev);
+	return vce_v4_0_hw_init(ip_block);
 }
 
 static void vce_v4_0_mc_resume(struct amdgpu_device *adev)
 {
 	uint32_t offset, size;
+	uint64_t tmr_mc_addr;
 
 	WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_A), 0, ~(1 << 16));
 	WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING), 0x1FF000, ~0xFF9FF000);
@@ -593,21 +643,25 @@ static void vce_v4_0_mc_resume(struct amdgpu_device *adev)
 	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_SWAP_CNTL1), 0);
 	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VM_CTRL), 0);
 
+	offset = AMDGPU_VCE_FIRMWARE_OFFSET;
+
 	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		tmr_mc_addr = (uint64_t)(adev->firmware.ucode[AMDGPU_UCODE_ID_VCE].tmr_mc_addr_hi) << 32 |
+										adev->firmware.ucode[AMDGPU_UCODE_ID_VCE].tmr_mc_addr_lo;
 		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_40BIT_BAR0),
-			(adev->firmware.ucode[AMDGPU_UCODE_ID_VCE].mc_addr >> 8));
+			(tmr_mc_addr >> 8));
 		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_64BIT_BAR0),
-			(adev->firmware.ucode[AMDGPU_UCODE_ID_VCE].mc_addr >> 40) & 0xff);
+			(tmr_mc_addr >> 40) & 0xff);
+		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_OFFSET0), 0);
 	} else {
 		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_40BIT_BAR0),
 			(adev->vce.gpu_addr >> 8));
 		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_64BIT_BAR0),
 			(adev->vce.gpu_addr >> 40) & 0xff);
+		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_OFFSET0), offset & ~0x0f000000);
 	}
 
-	offset = AMDGPU_VCE_FIRMWARE_OFFSET;
 	size = VCE_V4_0_FW_SIZE;
-	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_OFFSET0), offset & ~0x0f000000);
 	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_SIZE0), size);
 
 	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_40BIT_BAR1), (adev->vce.gpu_addr >> 8));
@@ -630,280 +684,14 @@ static void vce_v4_0_mc_resume(struct amdgpu_device *adev)
 			~VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK);
 }
 
-static int vce_v4_0_set_clockgating_state(void *handle,
+static int vce_v4_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
 	/* needed for driver unload*/
 	return 0;
 }
 
-#if 0
-static bool vce_v4_0_is_idle(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	u32 mask = 0;
-
-	mask |= (adev->vce.harvest_config & AMDGPU_VCE_HARVEST_VCE0) ? 0 : SRBM_STATUS2__VCE0_BUSY_MASK;
-	mask |= (adev->vce.harvest_config & AMDGPU_VCE_HARVEST_VCE1) ? 0 : SRBM_STATUS2__VCE1_BUSY_MASK;
-
-	return !(RREG32(mmSRBM_STATUS2) & mask);
-}
-
-static int vce_v4_0_wait_for_idle(void *handle)
-{
-	unsigned i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	for (i = 0; i < adev->usec_timeout; i++)
-		if (vce_v4_0_is_idle(handle))
-			return 0;
-
-	return -ETIMEDOUT;
-}
-
-#define  VCE_STATUS_VCPU_REPORT_AUTO_BUSY_MASK  0x00000008L   /* AUTO_BUSY */
-#define  VCE_STATUS_VCPU_REPORT_RB0_BUSY_MASK   0x00000010L   /* RB0_BUSY */
-#define  VCE_STATUS_VCPU_REPORT_RB1_BUSY_MASK   0x00000020L   /* RB1_BUSY */
-#define  AMDGPU_VCE_STATUS_BUSY_MASK (VCE_STATUS_VCPU_REPORT_AUTO_BUSY_MASK | \
-				      VCE_STATUS_VCPU_REPORT_RB0_BUSY_MASK)
-
-static bool vce_v4_0_check_soft_reset(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	u32 srbm_soft_reset = 0;
-
-	/* According to VCE team , we should use VCE_STATUS instead
-	 * SRBM_STATUS.VCE_BUSY bit for busy status checking.
-	 * GRBM_GFX_INDEX.INSTANCE_INDEX is used to specify which VCE
-	 * instance's registers are accessed
-	 * (0 for 1st instance, 10 for 2nd instance).
-	 *
-	 *VCE_STATUS
-	 *|UENC|ACPI|AUTO ACTIVE|RB1 |RB0 |RB2 |          |FW_LOADED|JOB |
-	 *|----+----+-----------+----+----+----+----------+---------+----|
-	 *|bit8|bit7|    bit6   |bit5|bit4|bit3|   bit2   |  bit1   |bit0|
-	 *
-	 * VCE team suggest use bit 3--bit 6 for busy status check
-	 */
-	mutex_lock(&adev->grbm_idx_mutex);
-	WREG32_FIELD(GRBM_GFX_INDEX, INSTANCE_INDEX, 0);
-	if (RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_STATUS) & AMDGPU_VCE_STATUS_BUSY_MASK) {
-		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VCE0, 1);
-		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VCE1, 1);
-	}
-	WREG32_FIELD(GRBM_GFX_INDEX, INSTANCE_INDEX, 0x10);
-	if (RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_STATUS) & AMDGPU_VCE_STATUS_BUSY_MASK) {
-		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VCE0, 1);
-		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VCE1, 1);
-	}
-	WREG32_FIELD(GRBM_GFX_INDEX, INSTANCE_INDEX, 0);
-	mutex_unlock(&adev->grbm_idx_mutex);
-
-	if (srbm_soft_reset) {
-		adev->vce.srbm_soft_reset = srbm_soft_reset;
-		return true;
-	} else {
-		adev->vce.srbm_soft_reset = 0;
-		return false;
-	}
-}
-
-static int vce_v4_0_soft_reset(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	u32 srbm_soft_reset;
-
-	if (!adev->vce.srbm_soft_reset)
-		return 0;
-	srbm_soft_reset = adev->vce.srbm_soft_reset;
-
-	if (srbm_soft_reset) {
-		u32 tmp;
-
-		tmp = RREG32(mmSRBM_SOFT_RESET);
-		tmp |= srbm_soft_reset;
-		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
-		WREG32(mmSRBM_SOFT_RESET, tmp);
-		tmp = RREG32(mmSRBM_SOFT_RESET);
-
-		udelay(50);
-
-		tmp &= ~srbm_soft_reset;
-		WREG32(mmSRBM_SOFT_RESET, tmp);
-		tmp = RREG32(mmSRBM_SOFT_RESET);
-
-		/* Wait a little for things to settle down */
-		udelay(50);
-	}
-
-	return 0;
-}
-
-static int vce_v4_0_pre_soft_reset(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (!adev->vce.srbm_soft_reset)
-		return 0;
-
-	mdelay(5);
-
-	return vce_v4_0_suspend(adev);
-}
-
-
-static int vce_v4_0_post_soft_reset(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (!adev->vce.srbm_soft_reset)
-		return 0;
-
-	mdelay(5);
-
-	return vce_v4_0_resume(adev);
-}
-
-static void vce_v4_0_override_vce_clock_gating(struct amdgpu_device *adev, bool override)
-{
-	u32 tmp, data;
-
-	tmp = data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_ARB_CTRL));
-	if (override)
-		data |= VCE_RB_ARB_CTRL__VCE_CGTT_OVERRIDE_MASK;
-	else
-		data &= ~VCE_RB_ARB_CTRL__VCE_CGTT_OVERRIDE_MASK;
-
-	if (tmp != data)
-		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_ARB_CTRL), data);
-}
-
-static void vce_v4_0_set_vce_sw_clock_gating(struct amdgpu_device *adev,
-					     bool gated)
-{
-	u32 data;
-
-	/* Set Override to disable Clock Gating */
-	vce_v4_0_override_vce_clock_gating(adev, true);
-
-	/* This function enables MGCG which is controlled by firmware.
-	   With the clocks in the gated state the core is still
-	   accessible but the firmware will throttle the clocks on the
-	   fly as necessary.
-	*/
-	if (gated) {
-		data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_B));
-		data |= 0x1ff;
-		data &= ~0xef0000;
-		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_B), data);
-
-		data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING));
-		data |= 0x3ff000;
-		data &= ~0xffc00000;
-		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING), data);
-
-		data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING_2));
-		data |= 0x2;
-		data &= ~0x00010000;
-		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING_2), data);
-
-		data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_REG_CLOCK_GATING));
-		data |= 0x37f;
-		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_REG_CLOCK_GATING), data);
-
-		data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_DMA_DCLK_CTRL));
-		data |= VCE_UENC_DMA_DCLK_CTRL__WRDMCLK_FORCEON_MASK |
-			VCE_UENC_DMA_DCLK_CTRL__RDDMCLK_FORCEON_MASK |
-			VCE_UENC_DMA_DCLK_CTRL__REGCLK_FORCEON_MASK  |
-			0x8;
-		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_DMA_DCLK_CTRL), data);
-	} else {
-		data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_B));
-		data &= ~0x80010;
-		data |= 0xe70008;
-		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_B), data);
-
-		data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING));
-		data |= 0xffc00000;
-		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING), data);
-
-		data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING_2));
-		data |= 0x10000;
-		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING_2), data);
-
-		data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_REG_CLOCK_GATING));
-		data &= ~0xffc00000;
-		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_REG_CLOCK_GATING), data);
-
-		data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_DMA_DCLK_CTRL));
-		data &= ~(VCE_UENC_DMA_DCLK_CTRL__WRDMCLK_FORCEON_MASK |
-			  VCE_UENC_DMA_DCLK_CTRL__RDDMCLK_FORCEON_MASK |
-			  VCE_UENC_DMA_DCLK_CTRL__REGCLK_FORCEON_MASK  |
-			  0x8);
-		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_DMA_DCLK_CTRL), data);
-	}
-	vce_v4_0_override_vce_clock_gating(adev, false);
-}
-
-static void vce_v4_0_set_bypass_mode(struct amdgpu_device *adev, bool enable)
-{
-	u32 tmp = RREG32_SMC(ixGCK_DFS_BYPASS_CNTL);
-
-	if (enable)
-		tmp |= GCK_DFS_BYPASS_CNTL__BYPASSECLK_MASK;
-	else
-		tmp &= ~GCK_DFS_BYPASS_CNTL__BYPASSECLK_MASK;
-
-	WREG32_SMC(ixGCK_DFS_BYPASS_CNTL, tmp);
-}
-
-static int vce_v4_0_set_clockgating_state(void *handle,
-					  enum amd_clockgating_state state)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	bool enable = (state == AMD_CG_STATE_GATE) ? true : false;
-	int i;
-
-	if ((adev->asic_type == CHIP_POLARIS10) ||
-		(adev->asic_type == CHIP_TONGA) ||
-		(adev->asic_type == CHIP_FIJI))
-		vce_v4_0_set_bypass_mode(adev, enable);
-
-	if (!(adev->cg_flags & AMD_CG_SUPPORT_VCE_MGCG))
-		return 0;
-
-	mutex_lock(&adev->grbm_idx_mutex);
-	for (i = 0; i < 2; i++) {
-		/* Program VCE Instance 0 or 1 if not harvested */
-		if (adev->vce.harvest_config & (1 << i))
-			continue;
-
-		WREG32_FIELD(GRBM_GFX_INDEX, VCE_INSTANCE, i);
-
-		if (enable) {
-			/* initialize VCE_CLOCK_GATING_A: Clock ON/OFF delay */
-			uint32_t data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_A);
-			data &= ~(0xf | 0xff0);
-			data |= ((0x0 << 0) | (0x04 << 4));
-			WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_A, data);
-
-			/* initialize VCE_UENC_CLOCK_GATING: Clock ON/OFF delay */
-			data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING);
-			data &= ~(0xf | 0xff0);
-			data |= ((0x0 << 0) | (0x04 << 4));
-			WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING, data);
-		}
-
-		vce_v4_0_set_vce_sw_clock_gating(adev, enable);
-	}
-
-	WREG32_FIELD(GRBM_GFX_INDEX, VCE_INSTANCE, 0);
-	mutex_unlock(&adev->grbm_idx_mutex);
-
-	return 0;
-}
-
-static int vce_v4_0_set_powergating_state(void *handle,
+static int vce_v4_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	/* This doesn't actually powergate the VCE block.
@@ -913,24 +701,21 @@ static int vce_v4_0_set_powergating_state(void *handle,
 	 * revisit this when there is a cleaner line between
 	 * the smc and the hw blocks
 	 */
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (!(adev->pg_flags & AMD_PG_SUPPORT_VCE))
-		return 0;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (state == AMD_PG_STATE_GATE)
-		/* XXX do we need a vce_v4_0_stop()? */
-		return 0;
+		return vce_v4_0_stop(adev);
 	else
 		return vce_v4_0_start(adev);
 }
-#endif
 
-static void vce_v4_0_ring_emit_ib(struct amdgpu_ring *ring,
-		struct amdgpu_ib *ib, unsigned int vm_id, bool ctx_switch)
+static void vce_v4_0_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_job *job,
+					struct amdgpu_ib *ib, uint32_t flags)
 {
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+
 	amdgpu_ring_write(ring, VCE_CMD_IB_VM);
-	amdgpu_ring_write(ring, vm_id);
+	amdgpu_ring_write(ring, vmid);
 	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
 	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
 	amdgpu_ring_write(ring, ib->length_dw);
@@ -953,39 +738,34 @@ static void vce_v4_0_ring_insert_end(struct amdgpu_ring *ring)
 	amdgpu_ring_write(ring, VCE_CMD_END);
 }
 
-static void vce_v4_0_emit_vm_flush(struct amdgpu_ring *ring,
-			 unsigned int vm_id, uint64_t pd_addr)
+static void vce_v4_0_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+				   uint32_t val, uint32_t mask)
 {
-	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
-	uint32_t req = ring->adev->gart.gart_funcs->get_invalidate_req(vm_id);
-	unsigned eng = ring->vm_inv_eng;
-
-	pd_addr = amdgpu_gart_get_vm_pde(ring->adev, pd_addr);
-	pd_addr |= AMDGPU_PTE_VALID;
+	amdgpu_ring_write(ring, VCE_CMD_REG_WAIT);
+	amdgpu_ring_write(ring,	reg << 2);
+	amdgpu_ring_write(ring, mask);
+	amdgpu_ring_write(ring, val);
+}
 
-	amdgpu_ring_write(ring, VCE_CMD_REG_WRITE);
-	amdgpu_ring_write(ring,	(hub->ctx0_ptb_addr_hi32 + vm_id * 2) << 2);
-	amdgpu_ring_write(ring, upper_32_bits(pd_addr));
+static void vce_v4_0_emit_vm_flush(struct amdgpu_ring *ring,
+				   unsigned int vmid, uint64_t pd_addr)
+{
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
 
-	amdgpu_ring_write(ring, VCE_CMD_REG_WRITE);
-	amdgpu_ring_write(ring,	(hub->ctx0_ptb_addr_lo32 + vm_id * 2) << 2);
-	amdgpu_ring_write(ring, lower_32_bits(pd_addr));
+	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
 
-	amdgpu_ring_write(ring, VCE_CMD_REG_WAIT);
-	amdgpu_ring_write(ring,	(hub->ctx0_ptb_addr_lo32 + vm_id * 2) << 2);
-	amdgpu_ring_write(ring, 0xffffffff);
-	amdgpu_ring_write(ring, lower_32_bits(pd_addr));
+	/* wait for reg writes */
+	vce_v4_0_emit_reg_wait(ring, hub->ctx0_ptb_addr_lo32 +
+			       vmid * hub->ctx_addr_distance,
+			       lower_32_bits(pd_addr), 0xffffffff);
+}
 
-	/* flush TLB */
+static void vce_v4_0_emit_wreg(struct amdgpu_ring *ring,
+			       uint32_t reg, uint32_t val)
+{
 	amdgpu_ring_write(ring, VCE_CMD_REG_WRITE);
-	amdgpu_ring_write(ring,	(hub->vm_inv_eng0_req + eng) << 2);
-	amdgpu_ring_write(ring, req);
-
-	/* wait for flush */
-	amdgpu_ring_write(ring, VCE_CMD_REG_WAIT);
-	amdgpu_ring_write(ring, (hub->vm_inv_eng0_ack + eng) << 2);
-	amdgpu_ring_write(ring, 1 << vm_id);
-	amdgpu_ring_write(ring, 1 << vm_id);
+	amdgpu_ring_write(ring,	reg << 2);
+	amdgpu_ring_write(ring, val);
 }
 
 static int vce_v4_0_set_interrupt_state(struct amdgpu_device *adev,
@@ -1011,10 +791,6 @@ static int vce_v4_0_process_interrupt(struct amdgpu_device *adev,
 {
 	DRM_DEBUG("IH: VCE\n");
 
-	WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_SYS_INT_STATUS),
-			VCE_SYS_INT_STATUS__VCE_SYS_INT_TRAP_INTERRUPT_INT_MASK,
-			~VCE_SYS_INT_STATUS__VCE_SYS_INT_TRAP_INTERRUPT_INT_MASK);
-
 	switch (entry->src_data[0]) {
 	case 0:
 	case 1:
@@ -1033,21 +809,14 @@ static int vce_v4_0_process_interrupt(struct amdgpu_device *adev,
 const struct amd_ip_funcs vce_v4_0_ip_funcs = {
 	.name = "vce_v4_0",
 	.early_init = vce_v4_0_early_init,
-	.late_init = NULL,
 	.sw_init = vce_v4_0_sw_init,
 	.sw_fini = vce_v4_0_sw_fini,
 	.hw_init = vce_v4_0_hw_init,
 	.hw_fini = vce_v4_0_hw_fini,
 	.suspend = vce_v4_0_suspend,
 	.resume = vce_v4_0_resume,
-	.is_idle = NULL /* vce_v4_0_is_idle */,
-	.wait_for_idle = NULL /* vce_v4_0_wait_for_idle */,
-	.check_soft_reset = NULL /* vce_v4_0_check_soft_reset */,
-	.pre_soft_reset = NULL /* vce_v4_0_pre_soft_reset */,
-	.soft_reset = NULL /* vce_v4_0_soft_reset */,
-	.post_soft_reset = NULL /* vce_v4_0_post_soft_reset */,
 	.set_clockgating_state = vce_v4_0_set_clockgating_state,
-	.set_powergating_state = NULL /* vce_v4_0_set_powergating_state */,
+	.set_powergating_state = vce_v4_0_set_powergating_state,
 };
 
 static const struct amdgpu_ring_funcs vce_v4_0_ring_vm_funcs = {
@@ -1055,13 +824,15 @@ static const struct amdgpu_ring_funcs vce_v4_0_ring_vm_funcs = {
 	.align_mask = 0x3f,
 	.nop = VCE_CMD_NO_OP,
 	.support_64bit_ptrs = false,
-	.vmhub = AMDGPU_MMHUB,
+	.no_user_fence = true,
 	.get_rptr = vce_v4_0_ring_get_rptr,
 	.get_wptr = vce_v4_0_ring_get_wptr,
 	.set_wptr = vce_v4_0_ring_set_wptr,
-	.parse_cs = amdgpu_vce_ring_parse_cs_vm,
+	.patch_cs_in_place = amdgpu_vce_ring_parse_cs_vm,
 	.emit_frame_size =
-		17 + /* vce_v4_0_emit_vm_flush */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 4 +
+		4 + /* vce_v4_0_emit_vm_flush */
 		5 + 5 + /* amdgpu_vce_ring_emit_fence x2 vm fence */
 		1, /* vce_v4_0_ring_insert_end */
 	.emit_ib_size = 5, /* vce_v4_0_ring_emit_ib */
@@ -1075,14 +846,19 @@ static const struct amdgpu_ring_funcs vce_v4_0_ring_vm_funcs = {
 	.pad_ib = amdgpu_ring_generic_pad_ib,
 	.begin_use = amdgpu_vce_ring_begin_use,
 	.end_use = amdgpu_vce_ring_end_use,
+	.emit_wreg = vce_v4_0_emit_wreg,
+	.emit_reg_wait = vce_v4_0_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
 };
 
 static void vce_v4_0_set_ring_funcs(struct amdgpu_device *adev)
 {
 	int i;
 
-	for (i = 0; i < adev->vce.num_rings; i++)
+	for (i = 0; i < adev->vce.num_rings; i++) {
 		adev->vce.ring[i].funcs = &vce_v4_0_ring_vm_funcs;
+		adev->vce.ring[i].me = i;
+	}
 	DRM_INFO("VCE enabled in VM mode\n");
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_sw_ring.c b/drivers/gpu/drm/amd/amdgpu/vcn_sw_ring.c
new file mode 100644
index 000000000000..2b9ddb3d2fe1
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_sw_ring.c
@@ -0,0 +1,86 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "vcn_sw_ring.h"
+
+void vcn_dec_sw_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
+	u64 seq, uint32_t flags)
+{
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	amdgpu_ring_write(ring, VCN_DEC_SW_CMD_FENCE);
+	amdgpu_ring_write(ring, addr);
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, VCN_DEC_SW_CMD_TRAP);
+}
+
+void vcn_dec_sw_ring_insert_end(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, VCN_DEC_SW_CMD_END);
+}
+
+void vcn_dec_sw_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_job *job,
+	struct amdgpu_ib *ib, uint32_t flags)
+{
+	uint32_t vmid = AMDGPU_JOB_GET_VMID(job);
+
+	amdgpu_ring_write(ring, VCN_DEC_SW_CMD_IB);
+	amdgpu_ring_write(ring, vmid);
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, ib->length_dw);
+}
+
+void vcn_dec_sw_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+	uint32_t val, uint32_t mask)
+{
+	amdgpu_ring_write(ring, VCN_DEC_SW_CMD_REG_WAIT);
+	amdgpu_ring_write(ring, reg << 2);
+	amdgpu_ring_write(ring, mask);
+	amdgpu_ring_write(ring, val);
+}
+
+void vcn_dec_sw_ring_emit_vm_flush(struct amdgpu_ring *ring,
+	uint32_t vmid, uint64_t pd_addr)
+{
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
+	uint32_t data0, data1, mask;
+
+	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* wait for register write */
+	data0 = hub->ctx0_ptb_addr_lo32 + vmid * hub->ctx_addr_distance;
+	data1 = lower_32_bits(pd_addr);
+	mask = 0xffffffff;
+	vcn_dec_sw_ring_emit_reg_wait(ring, data0, data1, mask);
+}
+
+void vcn_dec_sw_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg,
+	uint32_t val)
+{
+	amdgpu_ring_write(ring, VCN_DEC_SW_CMD_REG_WRITE);
+	amdgpu_ring_write(ring,	reg << 2);
+	amdgpu_ring_write(ring, val);
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_sw_ring.h b/drivers/gpu/drm/amd/amdgpu/vcn_sw_ring.h
new file mode 100644
index 000000000000..7e775725f120
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_sw_ring.h
@@ -0,0 +1,44 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VCN_SW_RING_H__
+#define __VCN_SW_RING_H__
+
+#define VCN_SW_RING_EMIT_FRAME_SIZE \
+		(4 + /* vcn_dec_sw_ring_emit_vm_flush */ \
+		5 + 5 + /* vcn_dec_sw_ring_emit_fence x2 vm fence */ \
+		1) /* vcn_dec_sw_ring_insert_end */
+
+void vcn_dec_sw_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
+	u64 seq, uint32_t flags);
+void vcn_dec_sw_ring_insert_end(struct amdgpu_ring *ring);
+void vcn_dec_sw_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_job *job,
+	struct amdgpu_ib *ib, uint32_t flags);
+void vcn_dec_sw_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+	uint32_t val, uint32_t mask);
+void vcn_dec_sw_ring_emit_vm_flush(struct amdgpu_ring *ring,
+	uint32_t vmid, uint64_t pd_addr);
+void vcn_dec_sw_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg,
+	uint32_t val);
+
+#endif /* __VCN_SW_RING_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c
index 21e7b88401e1..a316797875a8 100644
--- a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c
@@ -22,154 +22,247 @@
  */
 
 #include <linux/firmware.h>
-#include <drm/drmP.h>
+
 #include "amdgpu.h"
+#include "amdgpu_cs.h"
 #include "amdgpu_vcn.h"
+#include "amdgpu_pm.h"
+#include "soc15.h"
 #include "soc15d.h"
 #include "soc15_common.h"
 
-#include "vega10/soc15ip.h"
-#include "raven1/VCN/vcn_1_0_offset.h"
-#include "raven1/VCN/vcn_1_0_sh_mask.h"
-#include "vega10/HDP/hdp_4_0_offset.h"
-#include "raven1/MMHUB/mmhub_9_1_offset.h"
-#include "raven1/MMHUB/mmhub_9_1_sh_mask.h"
+#include "vcn/vcn_1_0_offset.h"
+#include "vcn/vcn_1_0_sh_mask.h"
+#include "mmhub/mmhub_9_1_offset.h"
+#include "mmhub/mmhub_9_1_sh_mask.h"
+
+#include "ivsrcid/vcn/irqsrcs_vcn_1_0.h"
+#include "jpeg_v1_0.h"
+#include "vcn_v1_0.h"
+
+#define mmUVD_RBC_XX_IB_REG_CHECK_1_0		0x05ab
+#define mmUVD_RBC_XX_IB_REG_CHECK_1_0_BASE_IDX	1
+#define mmUVD_REG_XX_MASK_1_0			0x05ac
+#define mmUVD_REG_XX_MASK_1_0_BASE_IDX		1
+
+static const struct amdgpu_hwip_reg_entry vcn_reg_list_1_0[] = {
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_CONTEXT_ID),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_CONTEXT_ID2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_GPCOM_VCPU_DATA0),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_GPCOM_VCPU_DATA1),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_GPCOM_VCPU_CMD),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_HI),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_LO),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_HI2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_LO2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_HI3),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_LO3),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_HI4),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_LO4),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_RPTR),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_WPTR),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_RPTR2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_WPTR2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_RPTR3),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_WPTR3),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_RPTR4),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_WPTR4),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_SIZE),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_SIZE2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_SIZE3),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_SIZE4),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_PGFSM_CONFIG),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_PGFSM_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_DPG_LMA_CTL),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_DPG_LMA_DATA),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_DPG_LMA_MASK),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_DPG_PAUSE)
+};
 
-static int vcn_v1_0_start(struct amdgpu_device *adev);
-static int vcn_v1_0_stop(struct amdgpu_device *adev);
+static int vcn_v1_0_stop(struct amdgpu_vcn_inst *vinst);
 static void vcn_v1_0_set_dec_ring_funcs(struct amdgpu_device *adev);
 static void vcn_v1_0_set_enc_ring_funcs(struct amdgpu_device *adev);
 static void vcn_v1_0_set_irq_funcs(struct amdgpu_device *adev);
+static int vcn_v1_0_set_pg_state(struct amdgpu_vcn_inst *vinst,
+				 enum amd_powergating_state state);
+static int vcn_v1_0_pause_dpg_mode(struct amdgpu_vcn_inst *vinst,
+				   struct dpg_pause_state *new_state);
+
+static void vcn_v1_0_idle_work_handler(struct work_struct *work);
+static void vcn_v1_0_ring_begin_use(struct amdgpu_ring *ring);
 
 /**
- * vcn_v1_0_early_init - set function pointers
+ * vcn_v1_0_early_init - set function pointers and load microcode
  *
- * @handle: amdgpu_device pointer
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
  *
  * Set ring and irq function pointers
+ * Load microcode from filesystem
  */
-static int vcn_v1_0_early_init(void *handle)
+static int vcn_v1_0_early_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	adev->vcn.num_enc_rings = 2;
+	adev->vcn.inst[0].num_enc_rings = 2;
+	adev->vcn.inst[0].set_pg_state = vcn_v1_0_set_pg_state;
 
 	vcn_v1_0_set_dec_ring_funcs(adev);
 	vcn_v1_0_set_enc_ring_funcs(adev);
 	vcn_v1_0_set_irq_funcs(adev);
 
-	return 0;
+	jpeg_v1_0_early_init(ip_block);
+
+	return amdgpu_vcn_early_init(adev, 0);
 }
 
 /**
  * vcn_v1_0_sw_init - sw init for VCN block
  *
- * @handle: amdgpu_device pointer
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
  *
  * Load firmware and sw initialization
  */
-static int vcn_v1_0_sw_init(void *handle)
+static int vcn_v1_0_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_ring *ring;
 	int i, r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	uint32_t reg_count = ARRAY_SIZE(vcn_reg_list_1_0);
+	uint32_t *ptr;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	/* VCN DEC TRAP */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_VCN, 124, &adev->vcn.irq);
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+			VCN_1_0__SRCID__UVD_SYSTEM_MESSAGE_INTERRUPT, &adev->vcn.inst->irq);
 	if (r)
 		return r;
 
 	/* VCN ENC TRAP */
-	for (i = 0; i < adev->vcn.num_enc_rings; ++i) {
-		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_VCN, i + 119,
-					&adev->vcn.irq);
+	for (i = 0; i < adev->vcn.inst[0].num_enc_rings; ++i) {
+		r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN, i + VCN_1_0__SRCID__UVD_ENC_GENERAL_PURPOSE,
+					&adev->vcn.inst->irq);
 		if (r)
 			return r;
 	}
 
-	r = amdgpu_vcn_sw_init(adev);
+	r = amdgpu_vcn_sw_init(adev, 0);
 	if (r)
 		return r;
 
-	r = amdgpu_vcn_resume(adev);
+	/* Override the work func */
+	adev->vcn.inst[0].idle_work.work.func = vcn_v1_0_idle_work_handler;
+
+	amdgpu_vcn_setup_ucode(adev, 0);
+
+	r = amdgpu_vcn_resume(adev, 0);
 	if (r)
 		return r;
 
-	ring = &adev->vcn.ring_dec;
+	ring = &adev->vcn.inst->ring_dec;
+	ring->vm_hub = AMDGPU_MMHUB0(0);
 	sprintf(ring->name, "vcn_dec");
-	r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.irq, 0);
+	r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst->irq, 0,
+			     AMDGPU_RING_PRIO_DEFAULT, NULL);
 	if (r)
 		return r;
 
-	for (i = 0; i < adev->vcn.num_enc_rings; ++i) {
-		ring = &adev->vcn.ring_enc[i];
+	adev->vcn.inst[0].internal.scratch9 = adev->vcn.inst->external.scratch9 =
+		SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9);
+	adev->vcn.inst[0].internal.data0 = adev->vcn.inst->external.data0 =
+		SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0);
+	adev->vcn.inst[0].internal.data1 = adev->vcn.inst->external.data1 =
+		SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA1);
+	adev->vcn.inst[0].internal.cmd = adev->vcn.inst->external.cmd =
+		SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD);
+	adev->vcn.inst[0].internal.nop = adev->vcn.inst->external.nop =
+		SOC15_REG_OFFSET(UVD, 0, mmUVD_NO_OP);
+
+	for (i = 0; i < adev->vcn.inst[0].num_enc_rings; ++i) {
+		enum amdgpu_ring_priority_level hw_prio = amdgpu_vcn_get_enc_ring_prio(i);
+
+		ring = &adev->vcn.inst->ring_enc[i];
+		ring->vm_hub = AMDGPU_MMHUB0(0);
 		sprintf(ring->name, "vcn_enc%d", i);
-		r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.irq, 0);
+		r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst->irq, 0,
+				     hw_prio, NULL);
 		if (r)
 			return r;
 	}
 
+	adev->vcn.inst[0].pause_dpg_mode = vcn_v1_0_pause_dpg_mode;
+
+	if (amdgpu_vcnfw_log) {
+		struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared.cpu_addr;
+
+		fw_shared->present_flag_0 = 0;
+		amdgpu_vcn_fwlog_init(adev->vcn.inst);
+	}
+
+	r = jpeg_v1_0_sw_init(ip_block);
+
+	/* Allocate memory for VCN IP Dump buffer */
+	ptr = kcalloc(adev->vcn.num_vcn_inst * reg_count, sizeof(uint32_t), GFP_KERNEL);
+	if (!ptr) {
+		DRM_ERROR("Failed to allocate memory for VCN IP Dump\n");
+		adev->vcn.ip_dump = NULL;
+	} else {
+		adev->vcn.ip_dump = ptr;
+	}
 	return r;
 }
 
 /**
  * vcn_v1_0_sw_fini - sw fini for VCN block
  *
- * @handle: amdgpu_device pointer
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
  *
  * VCN suspend and free up sw allocation
  */
-static int vcn_v1_0_sw_fini(void *handle)
+static int vcn_v1_0_sw_fini(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	r = amdgpu_vcn_suspend(adev);
+	r = amdgpu_vcn_suspend(adev, 0);
 	if (r)
 		return r;
 
-	r = amdgpu_vcn_sw_fini(adev);
+	jpeg_v1_0_sw_fini(ip_block);
 
-	return r;
+	amdgpu_vcn_sw_fini(adev, 0);
+
+	kfree(adev->vcn.ip_dump);
+
+	return 0;
 }
 
 /**
  * vcn_v1_0_hw_init - start and test VCN block
  *
- * @handle: amdgpu_device pointer
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
  *
  * Initialize the hardware, boot up the VCPU and do some testing
  */
-static int vcn_v1_0_hw_init(void *handle)
+static int vcn_v1_0_hw_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	struct amdgpu_ring *ring = &adev->vcn.ring_dec;
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring = &adev->vcn.inst->ring_dec;
 	int i, r;
 
-	r = vcn_v1_0_start(adev);
+	r = amdgpu_ring_test_helper(ring);
 	if (r)
-		goto done;
-
-	ring->ready = true;
-	r = amdgpu_ring_test_ring(ring);
-	if (r) {
-		ring->ready = false;
-		goto done;
-	}
+		return r;
 
-	for (i = 0; i < adev->vcn.num_enc_rings; ++i) {
-		ring = &adev->vcn.ring_enc[i];
-		ring->ready = true;
-		r = amdgpu_ring_test_ring(ring);
-		if (r) {
-			ring->ready = false;
-			goto done;
-		}
+	for (i = 0; i < adev->vcn.inst[0].num_enc_rings; ++i) {
+		ring = &adev->vcn.inst->ring_enc[i];
+		r = amdgpu_ring_test_helper(ring);
+		if (r)
+			return r;
 	}
 
-done:
-	if (!r)
-		DRM_INFO("VCN decode and encode initialized successfully.\n");
+	ring = adev->jpeg.inst->ring_dec;
+	r = amdgpu_ring_test_helper(ring);
 
 	return r;
 }
@@ -177,21 +270,22 @@ done:
 /**
  * vcn_v1_0_hw_fini - stop the hardware block
  *
- * @handle: amdgpu_device pointer
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
  *
  * Stop the VCN block, mark ring as not ready any more
  */
-static int vcn_v1_0_hw_fini(void *handle)
+static int vcn_v1_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	struct amdgpu_ring *ring = &adev->vcn.ring_dec;
-	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_vcn_inst *vinst = adev->vcn.inst;
 
-	r = vcn_v1_0_stop(adev);
-	if (r)
-		return r;
+	cancel_delayed_work_sync(&vinst->idle_work);
 
-	ring->ready = false;
+	if ((adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) ||
+	    (vinst->cur_state != AMD_PG_STATE_GATE &&
+	     RREG32_SOC15(VCN, 0, mmUVD_STATUS))) {
+		vinst->set_pg_state(vinst, AMD_PG_STATE_GATE);
+	}
 
 	return 0;
 }
@@ -199,20 +293,27 @@ static int vcn_v1_0_hw_fini(void *handle)
 /**
  * vcn_v1_0_suspend - suspend VCN block
  *
- * @handle: amdgpu_device pointer
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
  *
  * HW fini and suspend VCN block
  */
-static int vcn_v1_0_suspend(void *handle)
+static int vcn_v1_0_suspend(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
+	bool idle_work_unexecuted;
+
+	idle_work_unexecuted = cancel_delayed_work_sync(&adev->vcn.inst[0].idle_work);
+	if (idle_work_unexecuted) {
+		if (adev->pm.dpm_enabled)
+			amdgpu_dpm_enable_vcn(adev, false, 0);
+	}
 
-	r = vcn_v1_0_hw_fini(adev);
+	r = vcn_v1_0_hw_fini(ip_block);
 	if (r)
 		return r;
 
-	r = amdgpu_vcn_suspend(adev);
+	r = amdgpu_vcn_suspend(adev, 0);
 
 	return r;
 }
@@ -220,57 +321,71 @@ static int vcn_v1_0_suspend(void *handle)
 /**
  * vcn_v1_0_resume - resume VCN block
  *
- * @handle: amdgpu_device pointer
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
  *
  * Resume firmware and hw init VCN block
  */
-static int vcn_v1_0_resume(void *handle)
+static int vcn_v1_0_resume(struct amdgpu_ip_block *ip_block)
 {
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	r = amdgpu_vcn_resume(adev);
+	r = amdgpu_vcn_resume(ip_block->adev, 0);
 	if (r)
 		return r;
 
-	r = vcn_v1_0_hw_init(adev);
+	r = vcn_v1_0_hw_init(ip_block);
 
 	return r;
 }
 
 /**
- * vcn_v1_0_mc_resume - memory controller programming
+ * vcn_v1_0_mc_resume_spg_mode - memory controller programming
  *
- * @adev: amdgpu_device pointer
+ * @vinst: VCN instance
  *
  * Let the VCN memory controller know it's offsets
  */
-static void vcn_v1_0_mc_resume(struct amdgpu_device *adev)
+static void vcn_v1_0_mc_resume_spg_mode(struct amdgpu_vcn_inst *vinst)
 {
-	uint32_t size = AMDGPU_GPU_PAGE_ALIGN(adev->vcn.fw->size + 4);
-
-	WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
-			lower_32_bits(adev->vcn.gpu_addr));
-	WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
-			upper_32_bits(adev->vcn.gpu_addr));
-	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_OFFSET0,
-				AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
+	struct amdgpu_device *adev = vinst->adev;
+	uint32_t size = AMDGPU_GPU_PAGE_ALIGN(adev->vcn.inst[0].fw->size + 4);
+	uint32_t offset;
+
+	/* cache window 0: fw */
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			     (adev->firmware.ucode[AMDGPU_UCODE_ID_VCN].tmr_mc_addr_lo));
+		WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			     (adev->firmware.ucode[AMDGPU_UCODE_ID_VCN].tmr_mc_addr_hi));
+		WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_OFFSET0, 0);
+		offset = 0;
+	} else {
+		WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			lower_32_bits(adev->vcn.inst->gpu_addr));
+		WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			upper_32_bits(adev->vcn.inst->gpu_addr));
+		offset = size;
+		WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_OFFSET0,
+			     AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
+	}
+
 	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_SIZE0, size);
 
+	/* cache window 1: stack */
 	WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW,
-			lower_32_bits(adev->vcn.gpu_addr + size));
+		     lower_32_bits(adev->vcn.inst->gpu_addr + offset));
 	WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH,
-			upper_32_bits(adev->vcn.gpu_addr + size));
+		     upper_32_bits(adev->vcn.inst->gpu_addr + offset));
 	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_OFFSET1, 0);
-	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_SIZE1, AMDGPU_VCN_HEAP_SIZE);
+	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_SIZE1, AMDGPU_VCN_STACK_SIZE);
 
+	/* cache window 2: context */
 	WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW,
-			lower_32_bits(adev->vcn.gpu_addr + size + AMDGPU_VCN_HEAP_SIZE));
+		     lower_32_bits(adev->vcn.inst->gpu_addr + offset + AMDGPU_VCN_STACK_SIZE));
 	WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH,
-			upper_32_bits(adev->vcn.gpu_addr + size + AMDGPU_VCN_HEAP_SIZE));
+		     upper_32_bits(adev->vcn.inst->gpu_addr + offset + AMDGPU_VCN_STACK_SIZE));
 	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_OFFSET2, 0);
-	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_SIZE2,
-			AMDGPU_VCN_STACK_SIZE + (AMDGPU_VCN_SESSION_SIZE * 40));
+	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_SIZE2, AMDGPU_VCN_CONTEXT_SIZE);
 
 	WREG32_SOC15(UVD, 0, mmUVD_UDEC_ADDR_CONFIG,
 			adev->gfx.config.gb_addr_config);
@@ -278,24 +393,115 @@ static void vcn_v1_0_mc_resume(struct amdgpu_device *adev)
 			adev->gfx.config.gb_addr_config);
 	WREG32_SOC15(UVD, 0, mmUVD_UDEC_DBW_ADDR_CONFIG,
 			adev->gfx.config.gb_addr_config);
+	WREG32_SOC15(UVD, 0, mmUVD_UDEC_DBW_UV_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config);
+	WREG32_SOC15(UVD, 0, mmUVD_MIF_CURR_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config);
+	WREG32_SOC15(UVD, 0, mmUVD_MIF_CURR_UV_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config);
+	WREG32_SOC15(UVD, 0, mmUVD_MIF_RECON1_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config);
+	WREG32_SOC15(UVD, 0, mmUVD_MIF_RECON1_UV_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config);
+	WREG32_SOC15(UVD, 0, mmUVD_MIF_REF_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config);
+	WREG32_SOC15(UVD, 0, mmUVD_MIF_REF_UV_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config);
+	WREG32_SOC15(UVD, 0, mmUVD_JPEG_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config);
+	WREG32_SOC15(UVD, 0, mmUVD_JPEG_UV_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config);
+}
+
+static void vcn_v1_0_mc_resume_dpg_mode(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	uint32_t size = AMDGPU_GPU_PAGE_ALIGN(adev->vcn.inst[0].fw->size + 4);
+	uint32_t offset;
+
+	/* cache window 0: fw */
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			     (adev->firmware.ucode[AMDGPU_UCODE_ID_VCN].tmr_mc_addr_lo),
+			     0xFFFFFFFF, 0);
+		WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			     (adev->firmware.ucode[AMDGPU_UCODE_ID_VCN].tmr_mc_addr_hi),
+			     0xFFFFFFFF, 0);
+		WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_VCPU_CACHE_OFFSET0, 0,
+			     0xFFFFFFFF, 0);
+		offset = 0;
+	} else {
+		WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			lower_32_bits(adev->vcn.inst->gpu_addr), 0xFFFFFFFF, 0);
+		WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			upper_32_bits(adev->vcn.inst->gpu_addr), 0xFFFFFFFF, 0);
+		offset = size;
+		WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_VCPU_CACHE_OFFSET0,
+			     AMDGPU_UVD_FIRMWARE_OFFSET >> 3, 0xFFFFFFFF, 0);
+	}
+
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_VCPU_CACHE_SIZE0, size, 0xFFFFFFFF, 0);
+
+	/* cache window 1: stack */
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW,
+		     lower_32_bits(adev->vcn.inst->gpu_addr + offset), 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH,
+		     upper_32_bits(adev->vcn.inst->gpu_addr + offset), 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_VCPU_CACHE_OFFSET1, 0,
+			     0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_VCPU_CACHE_SIZE1, AMDGPU_VCN_STACK_SIZE,
+			     0xFFFFFFFF, 0);
+
+	/* cache window 2: context */
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW,
+		     lower_32_bits(adev->vcn.inst->gpu_addr + offset + AMDGPU_VCN_STACK_SIZE),
+			     0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH,
+		     upper_32_bits(adev->vcn.inst->gpu_addr + offset + AMDGPU_VCN_STACK_SIZE),
+			     0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_VCPU_CACHE_OFFSET2, 0, 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_VCPU_CACHE_SIZE2, AMDGPU_VCN_CONTEXT_SIZE,
+			     0xFFFFFFFF, 0);
+
+	/* VCN global tiling registers */
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_UDEC_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config, 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_UDEC_DB_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config, 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_UDEC_DBW_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config, 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_UDEC_DBW_UV_ADDR_CONFIG,
+		adev->gfx.config.gb_addr_config, 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_MIF_CURR_ADDR_CONFIG,
+		adev->gfx.config.gb_addr_config, 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_MIF_CURR_UV_ADDR_CONFIG,
+		adev->gfx.config.gb_addr_config, 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_MIF_RECON1_ADDR_CONFIG,
+		adev->gfx.config.gb_addr_config, 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_MIF_RECON1_UV_ADDR_CONFIG,
+		adev->gfx.config.gb_addr_config, 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_MIF_REF_ADDR_CONFIG,
+		adev->gfx.config.gb_addr_config, 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_MIF_REF_UV_ADDR_CONFIG,
+		adev->gfx.config.gb_addr_config, 0xFFFFFFFF, 0);
 }
 
 /**
  * vcn_v1_0_disable_clock_gating - disable VCN clock gating
  *
- * @adev: amdgpu_device pointer
- * @sw: enable SW clock gating
+ * @vinst: VCN instance
  *
  * Disable clock gating for VCN block
  */
-static void vcn_v1_0_disable_clock_gating(struct amdgpu_device *adev, bool sw)
+static void vcn_v1_0_disable_clock_gating(struct amdgpu_vcn_inst *vinst)
 {
+	struct amdgpu_device *adev = vinst->adev;
 	uint32_t data;
 
 	/* JPEG disable CGC */
 	data = RREG32_SOC15(VCN, 0, mmJPEG_CGC_CTRL);
 
-	if (sw)
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
 		data |= 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
 	else
 		data &= ~JPEG_CGC_CTRL__DYN_CLOCK_MODE_MASK;
@@ -310,10 +516,10 @@ static void vcn_v1_0_disable_clock_gating(struct amdgpu_device *adev, bool sw)
 
 	/* UVD disable CGC */
 	data = RREG32_SOC15(VCN, 0, mmUVD_CGC_CTRL);
-	if (sw)
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
 		data |= 1 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
 	else
-		data &= ~ UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK;
+		data &= ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK;
 
 	data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
 	data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
@@ -410,18 +616,18 @@ static void vcn_v1_0_disable_clock_gating(struct amdgpu_device *adev, bool sw)
 /**
  * vcn_v1_0_enable_clock_gating - enable VCN clock gating
  *
- * @adev: amdgpu_device pointer
- * @sw: enable SW clock gating
+ * @vinst: Pointer to the VCN instance structure
  *
  * Enable clock gating for VCN block
  */
-static void vcn_v1_0_enable_clock_gating(struct amdgpu_device *adev, bool sw)
+static void vcn_v1_0_enable_clock_gating(struct amdgpu_vcn_inst *vinst)
 {
+	struct amdgpu_device *adev = vinst->adev;
 	uint32_t data = 0;
 
 	/* enable JPEG CGC */
 	data = RREG32_SOC15(VCN, 0, mmJPEG_CGC_CTRL);
-	if (sw)
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
 		data |= 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
 	else
 		data |= 0 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
@@ -435,7 +641,7 @@ static void vcn_v1_0_enable_clock_gating(struct amdgpu_device *adev, bool sw)
 
 	/* enable UVD CGC */
 	data = RREG32_SOC15(VCN, 0, mmUVD_CGC_CTRL);
-	if (sw)
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
 		data |= 1 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
 	else
 		data |= 0 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
@@ -480,16 +686,161 @@ static void vcn_v1_0_enable_clock_gating(struct amdgpu_device *adev, bool sw)
 	WREG32_SOC15(VCN, 0, mmUVD_SUVD_CGC_CTRL, data);
 }
 
+static void vcn_v1_0_clock_gating_dpg_mode(struct amdgpu_vcn_inst *vinst,
+					   uint8_t sram_sel)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	uint32_t reg_data = 0;
+
+	/* disable JPEG CGC */
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		reg_data = 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		reg_data = 0 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	reg_data |= 1 << JPEG_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	reg_data |= 4 << JPEG_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmJPEG_CGC_CTRL, reg_data, 0xFFFFFFFF, sram_sel);
+
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmJPEG_CGC_GATE, 0, 0xFFFFFFFF, sram_sel);
+
+	/* enable sw clock gating control */
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		reg_data = 1 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		reg_data = 0 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	reg_data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	reg_data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	reg_data &= ~(UVD_CGC_CTRL__UDEC_RE_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_CM_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_IT_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_DB_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_MP_MODE_MASK |
+		 UVD_CGC_CTRL__SYS_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_MODE_MASK |
+		 UVD_CGC_CTRL__MPEG2_MODE_MASK |
+		 UVD_CGC_CTRL__REGS_MODE_MASK |
+		 UVD_CGC_CTRL__RBC_MODE_MASK |
+		 UVD_CGC_CTRL__LMI_MC_MODE_MASK |
+		 UVD_CGC_CTRL__LMI_UMC_MODE_MASK |
+		 UVD_CGC_CTRL__IDCT_MODE_MASK |
+		 UVD_CGC_CTRL__MPRD_MODE_MASK |
+		 UVD_CGC_CTRL__MPC_MODE_MASK |
+		 UVD_CGC_CTRL__LBSI_MODE_MASK |
+		 UVD_CGC_CTRL__LRBBM_MODE_MASK |
+		 UVD_CGC_CTRL__WCB_MODE_MASK |
+		 UVD_CGC_CTRL__VCPU_MODE_MASK |
+		 UVD_CGC_CTRL__SCPU_MODE_MASK);
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_CGC_CTRL, reg_data, 0xFFFFFFFF, sram_sel);
+
+	/* turn off clock gating */
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_CGC_GATE, 0, 0xFFFFFFFF, sram_sel);
+
+	/* turn on SUVD clock gating */
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_SUVD_CGC_GATE, 1, 0xFFFFFFFF, sram_sel);
+
+	/* turn on sw mode in UVD_SUVD_CGC_CTRL */
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_SUVD_CGC_CTRL, 0, 0xFFFFFFFF, sram_sel);
+}
+
+static void vcn_1_0_disable_static_power_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	uint32_t data = 0;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN) {
+		data = (1 << UVD_PGFSM_CONFIG__UVDM_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDU_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDF_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDC_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDB_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDIL_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDIR_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTD_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTE_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDE_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDW_PWR_CONFIG__SHIFT);
+
+		WREG32_SOC15(VCN, 0, mmUVD_PGFSM_CONFIG, data);
+		SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_PGFSM_STATUS, UVD_PGFSM_STATUS__UVDM_UVDU_PWR_ON, 0xFFFFFF);
+	} else {
+		data = (1 << UVD_PGFSM_CONFIG__UVDM_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDU_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDF_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDC_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDB_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDIL_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDIR_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDTD_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDTE_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDE_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDW_PWR_CONFIG__SHIFT);
+		WREG32_SOC15(VCN, 0, mmUVD_PGFSM_CONFIG, data);
+		SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_PGFSM_STATUS, 0,  0xFFFFFFFF);
+	}
+
+	/* polling UVD_PGFSM_STATUS to confirm UVDM_PWR_STATUS , UVDU_PWR_STATUS are 0 (power on) */
+
+	data = RREG32_SOC15(VCN, 0, mmUVD_POWER_STATUS);
+	data &= ~0x103;
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN)
+		data |= UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON | UVD_POWER_STATUS__UVD_PG_EN_MASK;
+
+	WREG32_SOC15(VCN, 0, mmUVD_POWER_STATUS, data);
+}
+
+static void vcn_1_0_enable_static_power_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	uint32_t data = 0;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN) {
+		/* Before power off, this indicator has to be turned on */
+		data = RREG32_SOC15(VCN, 0, mmUVD_POWER_STATUS);
+		data &= ~UVD_POWER_STATUS__UVD_POWER_STATUS_MASK;
+		data |= UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF;
+		WREG32_SOC15(VCN, 0, mmUVD_POWER_STATUS, data);
+
+
+		data = (2 << UVD_PGFSM_CONFIG__UVDM_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDU_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDF_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDC_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDB_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDIL_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDIR_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTD_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTE_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDE_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDW_PWR_CONFIG__SHIFT);
+
+		WREG32_SOC15(VCN, 0, mmUVD_PGFSM_CONFIG, data);
+
+		data = (2 << UVD_PGFSM_STATUS__UVDM_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDU_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDF_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDC_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDB_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDIL_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDIR_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDTD_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDTE_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDE_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDW_PWR_STATUS__SHIFT);
+		SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_PGFSM_STATUS, data, 0xFFFFFFFF);
+	}
+}
+
 /**
- * vcn_v1_0_start - start VCN block
+ * vcn_v1_0_start_spg_mode - start VCN block
  *
- * @adev: amdgpu_device pointer
+ * @vinst: VCN instance
  *
  * Setup and start the VCN block
  */
-static int vcn_v1_0_start(struct amdgpu_device *adev)
+static int vcn_v1_0_start_spg_mode(struct amdgpu_vcn_inst *vinst)
 {
-	struct amdgpu_ring *ring = &adev->vcn.ring_dec;
+	struct amdgpu_device *adev = vinst->adev;
+	struct amdgpu_ring *ring = &adev->vcn.inst->ring_dec;
 	uint32_t rb_bufsz, tmp;
 	uint32_t lmi_swap_cntl;
 	int i, j, r;
@@ -497,41 +848,25 @@ static int vcn_v1_0_start(struct amdgpu_device *adev)
 	/* disable byte swapping */
 	lmi_swap_cntl = 0;
 
-	vcn_v1_0_mc_resume(adev);
+	vcn_1_0_disable_static_power_gating(vinst);
+
+	tmp = RREG32_SOC15(UVD, 0, mmUVD_STATUS) | UVD_STATUS__UVD_BUSY;
+	WREG32_SOC15(UVD, 0, mmUVD_STATUS, tmp);
 
 	/* disable clock gating */
-	vcn_v1_0_disable_clock_gating(adev, true);
+	vcn_v1_0_disable_clock_gating(vinst);
 
 	/* disable interupt */
 	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_MASTINT_EN), 0,
 			~UVD_MASTINT_EN__VCPU_EN_MASK);
 
-	/* stall UMC and register bus before resetting VCPU */
-	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2),
-			UVD_LMI_CTRL2__STALL_ARB_UMC_MASK,
-			~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
-	mdelay(1);
-
-	/* put LMI, VCPU, RBC etc... into reset */
-	WREG32_SOC15(UVD, 0, mmUVD_SOFT_RESET,
-		UVD_SOFT_RESET__LMI_SOFT_RESET_MASK |
-		UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK |
-		UVD_SOFT_RESET__LBSI_SOFT_RESET_MASK |
-		UVD_SOFT_RESET__RBC_SOFT_RESET_MASK |
-		UVD_SOFT_RESET__CSM_SOFT_RESET_MASK |
-		UVD_SOFT_RESET__CXW_SOFT_RESET_MASK |
-		UVD_SOFT_RESET__TAP_SOFT_RESET_MASK |
-		UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK);
-	mdelay(5);
-
 	/* initialize VCN memory controller */
-	WREG32_SOC15(UVD, 0, mmUVD_LMI_CTRL,
-		(0x40 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
-		UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
-		UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
-		UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
-		UVD_LMI_CTRL__REQ_MODE_MASK |
-		0x00100000L);
+	tmp = RREG32_SOC15(UVD, 0, mmUVD_LMI_CTRL);
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_CTRL, tmp		|
+		UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK	|
+		UVD_LMI_CTRL__MASK_MC_URGENT_MASK			|
+		UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK		|
+		UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK);
 
 #ifdef __BIG_ENDIAN
 	/* swap (8 in 32) RB and IB */
@@ -539,41 +874,61 @@ static int vcn_v1_0_start(struct amdgpu_device *adev)
 #endif
 	WREG32_SOC15(UVD, 0, mmUVD_LMI_SWAP_CNTL, lmi_swap_cntl);
 
-	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUXA0, 0x40c2040);
-	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUXA1, 0x0);
-	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUXB0, 0x40c2040);
-	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUXB1, 0x0);
-	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_ALU, 0);
-	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUX, 0x88);
+	tmp = RREG32_SOC15(UVD, 0, mmUVD_MPC_CNTL);
+	tmp &= ~UVD_MPC_CNTL__REPLACEMENT_MODE_MASK;
+	tmp |= 0x2 << UVD_MPC_CNTL__REPLACEMENT_MODE__SHIFT;
+	WREG32_SOC15(UVD, 0, mmUVD_MPC_CNTL, tmp);
 
-	/* take all subblocks out of reset, except VCPU */
-	WREG32_SOC15(UVD, 0, mmUVD_SOFT_RESET,
-			UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
-	mdelay(5);
+	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUXA0,
+		((0x1 << UVD_MPC_SET_MUXA0__VARA_1__SHIFT) |
+		(0x2 << UVD_MPC_SET_MUXA0__VARA_2__SHIFT) |
+		(0x3 << UVD_MPC_SET_MUXA0__VARA_3__SHIFT) |
+		(0x4 << UVD_MPC_SET_MUXA0__VARA_4__SHIFT)));
+
+	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUXB0,
+		((0x1 << UVD_MPC_SET_MUXB0__VARB_1__SHIFT) |
+		(0x2 << UVD_MPC_SET_MUXB0__VARB_2__SHIFT) |
+		(0x3 << UVD_MPC_SET_MUXB0__VARB_3__SHIFT) |
+		(0x4 << UVD_MPC_SET_MUXB0__VARB_4__SHIFT)));
+
+	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUX,
+		((0x0 << UVD_MPC_SET_MUX__SET_0__SHIFT) |
+		(0x1 << UVD_MPC_SET_MUX__SET_1__SHIFT) |
+		(0x2 << UVD_MPC_SET_MUX__SET_2__SHIFT)));
+
+	vcn_v1_0_mc_resume_spg_mode(vinst);
+
+	WREG32_SOC15(UVD, 0, mmUVD_REG_XX_MASK_1_0, 0x10);
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_XX_IB_REG_CHECK_1_0,
+		RREG32_SOC15(UVD, 0, mmUVD_RBC_XX_IB_REG_CHECK_1_0) | 0x3);
 
 	/* enable VCPU clock */
-	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CNTL,
-			UVD_VCPU_CNTL__CLK_EN_MASK);
+	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CNTL, UVD_VCPU_CNTL__CLK_EN_MASK);
+
+	/* boot up the VCPU */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET), 0,
+			~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
 
 	/* enable UMC */
 	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2), 0,
 			~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
 
-	/* boot up the VCPU */
-	WREG32_SOC15(UVD, 0, mmUVD_SOFT_RESET, 0);
-	mdelay(10);
+	tmp = RREG32_SOC15(UVD, 0, mmUVD_SOFT_RESET);
+	tmp &= ~UVD_SOFT_RESET__LMI_SOFT_RESET_MASK;
+	tmp &= ~UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK;
+	WREG32_SOC15(UVD, 0, mmUVD_SOFT_RESET, tmp);
 
 	for (i = 0; i < 10; ++i) {
 		uint32_t status;
 
 		for (j = 0; j < 100; ++j) {
 			status = RREG32_SOC15(UVD, 0, mmUVD_STATUS);
-			if (status & 2)
+			if (status & UVD_STATUS__IDLE)
 				break;
 			mdelay(10);
 		}
 		r = 0;
-		if (status & 2)
+		if (status & UVD_STATUS__IDLE)
 			break;
 
 		DRM_ERROR("VCN decode not responding, trying to reset the VCPU!!!\n");
@@ -593,19 +948,22 @@ static int vcn_v1_0_start(struct amdgpu_device *adev)
 	}
 	/* enable master interrupt */
 	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_MASTINT_EN),
-		(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK),
-		~(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK));
+		UVD_MASTINT_EN__VCPU_EN_MASK, ~UVD_MASTINT_EN__VCPU_EN_MASK);
+
+	/* enable system interrupt for JRBC, TODO: move to set interrupt*/
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_SYS_INT_EN),
+		UVD_SYS_INT_EN__UVD_JRBC_EN_MASK,
+		~UVD_SYS_INT_EN__UVD_JRBC_EN_MASK);
 
-	/* clear the bit 4 of VCN_STATUS */
-	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_STATUS), 0,
-			~(2 << UVD_STATUS__VCPU_REPORT__SHIFT));
+	/* clear the busy bit of UVD_STATUS */
+	tmp = RREG32_SOC15(UVD, 0, mmUVD_STATUS) & ~UVD_STATUS__UVD_BUSY;
+	WREG32_SOC15(UVD, 0, mmUVD_STATUS, tmp);
 
 	/* force RBC into idle state */
 	rb_bufsz = order_base_2(ring->ring_size);
 	tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz);
 	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
 	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
-	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_WPTR_POLL_EN, 0);
 	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
 	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
 	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_CNTL, tmp);
@@ -617,7 +975,7 @@ static int vcn_v1_0_start(struct amdgpu_device *adev)
 	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR_ADDR,
 			(upper_32_bits(ring->gpu_addr) >> 2));
 
-	/* programm the RB_BASE for ring buffer */
+	/* program the RB_BASE for ring buffer */
 	WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW,
 			lower_32_bits(ring->gpu_addr));
 	WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH,
@@ -626,6 +984,8 @@ static int vcn_v1_0_start(struct amdgpu_device *adev)
 	/* Initialize the ring buffer's read and write pointers */
 	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR, 0);
 
+	WREG32_SOC15(UVD, 0, mmUVD_SCRATCH2, 0);
+
 	ring->wptr = RREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR);
 	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR,
 			lower_32_bits(ring->wptr));
@@ -633,63 +993,462 @@ static int vcn_v1_0_start(struct amdgpu_device *adev)
 	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_RBC_RB_CNTL), 0,
 			~UVD_RBC_RB_CNTL__RB_NO_FETCH_MASK);
 
-	ring = &adev->vcn.ring_enc[0];
+	ring = &adev->vcn.inst->ring_enc[0];
 	WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR, lower_32_bits(ring->wptr));
 	WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR, lower_32_bits(ring->wptr));
 	WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO, ring->gpu_addr);
 	WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
 	WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE, ring->ring_size / 4);
 
-	ring = &adev->vcn.ring_enc[1];
+	ring = &adev->vcn.inst->ring_enc[1];
 	WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR2, lower_32_bits(ring->wptr));
 	WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
 	WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO2, ring->gpu_addr);
 	WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
 	WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE2, ring->ring_size / 4);
 
+	jpeg_v1_0_start(adev, 0);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(UVD, 0, mmUVD_STATUS);
+
 	return 0;
 }
 
+static int vcn_v1_0_start_dpg_mode(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	struct amdgpu_ring *ring = &adev->vcn.inst->ring_dec;
+	uint32_t rb_bufsz, tmp;
+	uint32_t lmi_swap_cntl;
+
+	/* disable byte swapping */
+	lmi_swap_cntl = 0;
+
+	vcn_1_0_enable_static_power_gating(vinst);
+
+	/* enable dynamic power gating mode */
+	tmp = RREG32_SOC15(UVD, 0, mmUVD_POWER_STATUS);
+	tmp |= UVD_POWER_STATUS__UVD_PG_MODE_MASK;
+	tmp |= UVD_POWER_STATUS__UVD_PG_EN_MASK;
+	WREG32_SOC15(UVD, 0, mmUVD_POWER_STATUS, tmp);
+
+	/* enable clock gating */
+	vcn_v1_0_clock_gating_dpg_mode(vinst, 0);
+
+	/* enable VCPU clock */
+	tmp = (0xFF << UVD_VCPU_CNTL__PRB_TIMEOUT_VAL__SHIFT);
+	tmp |= UVD_VCPU_CNTL__CLK_EN_MASK;
+	tmp |= UVD_VCPU_CNTL__MIF_WR_LOW_THRESHOLD_BP_MASK;
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_VCPU_CNTL, tmp, 0xFFFFFFFF, 0);
+
+	/* disable interupt */
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_MASTINT_EN,
+			0, UVD_MASTINT_EN__VCPU_EN_MASK, 0);
+
+	/* initialize VCN memory controller */
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_LMI_CTRL,
+		(8 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
+		UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+		UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+		UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
+		UVD_LMI_CTRL__REQ_MODE_MASK |
+		UVD_LMI_CTRL__CRC_RESET_MASK |
+		UVD_LMI_CTRL__MASK_MC_URGENT_MASK |
+		0x00100000L, 0xFFFFFFFF, 0);
+
+#ifdef __BIG_ENDIAN
+	/* swap (8 in 32) RB and IB */
+	lmi_swap_cntl = 0xa;
+#endif
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_LMI_SWAP_CNTL, lmi_swap_cntl, 0xFFFFFFFF, 0);
+
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_MPC_CNTL,
+		0x2 << UVD_MPC_CNTL__REPLACEMENT_MODE__SHIFT, 0xFFFFFFFF, 0);
+
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_MPC_SET_MUXA0,
+		((0x1 << UVD_MPC_SET_MUXA0__VARA_1__SHIFT) |
+		 (0x2 << UVD_MPC_SET_MUXA0__VARA_2__SHIFT) |
+		 (0x3 << UVD_MPC_SET_MUXA0__VARA_3__SHIFT) |
+		 (0x4 << UVD_MPC_SET_MUXA0__VARA_4__SHIFT)), 0xFFFFFFFF, 0);
+
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_MPC_SET_MUXB0,
+		((0x1 << UVD_MPC_SET_MUXB0__VARB_1__SHIFT) |
+		 (0x2 << UVD_MPC_SET_MUXB0__VARB_2__SHIFT) |
+		 (0x3 << UVD_MPC_SET_MUXB0__VARB_3__SHIFT) |
+		 (0x4 << UVD_MPC_SET_MUXB0__VARB_4__SHIFT)), 0xFFFFFFFF, 0);
+
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_MPC_SET_MUX,
+		((0x0 << UVD_MPC_SET_MUX__SET_0__SHIFT) |
+		 (0x1 << UVD_MPC_SET_MUX__SET_1__SHIFT) |
+		 (0x2 << UVD_MPC_SET_MUX__SET_2__SHIFT)), 0xFFFFFFFF, 0);
+
+	vcn_v1_0_mc_resume_dpg_mode(vinst);
+
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_REG_XX_MASK, 0x10, 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_RBC_XX_IB_REG_CHECK, 0x3, 0xFFFFFFFF, 0);
+
+	/* boot up the VCPU */
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_SOFT_RESET, 0, 0xFFFFFFFF, 0);
+
+	/* enable UMC */
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_LMI_CTRL2,
+		0x1F << UVD_LMI_CTRL2__RE_OFLD_MIF_WR_REQ_NUM__SHIFT,
+		0xFFFFFFFF, 0);
+
+	/* enable master interrupt */
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_MASTINT_EN,
+			UVD_MASTINT_EN__VCPU_EN_MASK, UVD_MASTINT_EN__VCPU_EN_MASK, 0);
+
+	vcn_v1_0_clock_gating_dpg_mode(vinst, 1);
+	/* setup mmUVD_LMI_CTRL */
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_LMI_CTRL,
+		(8 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
+		UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+		UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+		UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
+		UVD_LMI_CTRL__REQ_MODE_MASK |
+		UVD_LMI_CTRL__CRC_RESET_MASK |
+		UVD_LMI_CTRL__MASK_MC_URGENT_MASK |
+		0x00100000L, 0xFFFFFFFF, 1);
+
+	tmp = adev->gfx.config.gb_addr_config;
+	/* setup VCN global tiling registers */
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_JPEG_ADDR_CONFIG, tmp, 0xFFFFFFFF, 1);
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_JPEG_UV_ADDR_CONFIG, tmp, 0xFFFFFFFF, 1);
+
+	/* enable System Interrupt for JRBC */
+	WREG32_SOC15_DPG_MODE_1_0(UVD, 0, mmUVD_SYS_INT_EN,
+									UVD_SYS_INT_EN__UVD_JRBC_EN_MASK, 0xFFFFFFFF, 1);
+
+	/* force RBC into idle state */
+	rb_bufsz = order_base_2(ring->ring_size);
+	tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_CNTL, tmp);
+
+	/* set the write pointer delay */
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR_CNTL, 0);
+
+	/* set the wb address */
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR_ADDR,
+								(upper_32_bits(ring->gpu_addr) >> 2));
+
+	/* program the RB_BASE for ring buffer */
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW,
+								lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH,
+								upper_32_bits(ring->gpu_addr));
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR, 0);
+
+	WREG32_SOC15(UVD, 0, mmUVD_SCRATCH2, 0);
+
+	ring->wptr = RREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR);
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR,
+								lower_32_bits(ring->wptr));
+
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_RBC_RB_CNTL), 0,
+			~UVD_RBC_RB_CNTL__RB_NO_FETCH_MASK);
+
+	jpeg_v1_0_start(adev, 1);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(UVD, 0, mmUVD_STATUS);
+
+	return 0;
+}
+
+static int vcn_v1_0_start(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+
+	return (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) ?
+		vcn_v1_0_start_dpg_mode(vinst) : vcn_v1_0_start_spg_mode(vinst);
+}
+
 /**
- * vcn_v1_0_stop - stop VCN block
+ * vcn_v1_0_stop_spg_mode - stop VCN block
  *
- * @adev: amdgpu_device pointer
+ * @vinst: VCN instance
  *
  * stop the VCN block
  */
-static int vcn_v1_0_stop(struct amdgpu_device *adev)
+static int vcn_v1_0_stop_spg_mode(struct amdgpu_vcn_inst *vinst)
 {
-	/* force RBC into idle state */
-	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_CNTL, 0x11010101);
+	struct amdgpu_device *adev = vinst->adev;
+	int tmp;
+
+	SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_STATUS, UVD_STATUS__IDLE, 0x7);
+
+	tmp = UVD_LMI_STATUS__VCPU_LMI_WRITE_CLEAN_MASK |
+		UVD_LMI_STATUS__READ_CLEAN_MASK |
+		UVD_LMI_STATUS__WRITE_CLEAN_MASK |
+		UVD_LMI_STATUS__WRITE_CLEAN_RAW_MASK;
+	SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_LMI_STATUS, tmp, tmp);
 
-	/* Stall UMC and register bus before resetting VCPU */
+	/* stall UMC channel */
 	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2),
-			UVD_LMI_CTRL2__STALL_ARB_UMC_MASK,
-			~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
-	mdelay(1);
+		UVD_LMI_CTRL2__STALL_ARB_UMC_MASK,
+		~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
 
-	/* put VCPU into reset */
-	WREG32_SOC15(UVD, 0, mmUVD_SOFT_RESET,
-			UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
-	mdelay(5);
+	tmp = UVD_LMI_STATUS__UMC_READ_CLEAN_RAW_MASK |
+		UVD_LMI_STATUS__UMC_WRITE_CLEAN_RAW_MASK;
+	SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_LMI_STATUS, tmp, tmp);
 
 	/* disable VCPU clock */
-	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CNTL, 0x0);
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CNTL), 0,
+		~UVD_VCPU_CNTL__CLK_EN_MASK);
 
-	/* Unstall UMC and register bus */
-	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2), 0,
-			~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+	/* reset LMI UMC/LMI */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET),
+		UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK,
+		~UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK);
 
-	/* enable clock gating */
-	vcn_v1_0_enable_clock_gating(adev, true);
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET),
+		UVD_SOFT_RESET__LMI_SOFT_RESET_MASK,
+		~UVD_SOFT_RESET__LMI_SOFT_RESET_MASK);
+
+	/* put VCPU into reset */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET),
+		UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK,
+		~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+
+	WREG32_SOC15(UVD, 0, mmUVD_STATUS, 0);
+
+	vcn_v1_0_enable_clock_gating(vinst);
+	vcn_1_0_enable_static_power_gating(vinst);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(UVD, 0, mmUVD_STATUS);
 
 	return 0;
 }
 
-static int vcn_v1_0_set_clockgating_state(void *handle,
+static int vcn_v1_0_stop_dpg_mode(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	uint32_t tmp;
+
+	/* Wait for power status to be UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF */
+	SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS,
+			UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF,
+			UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+	/* wait for read ptr to be equal to write ptr */
+	tmp = RREG32_SOC15(UVD, 0, mmUVD_RB_WPTR);
+	SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_RB_RPTR, tmp, 0xFFFFFFFF);
+
+	tmp = RREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2);
+	SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_RB_RPTR2, tmp, 0xFFFFFFFF);
+
+	tmp = RREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_WPTR);
+	SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_JRBC_RB_RPTR, tmp, 0xFFFFFFFF);
+
+	tmp = RREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR) & 0x7FFFFFFF;
+	SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_RBC_RB_RPTR, tmp, 0xFFFFFFFF);
+
+	SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS,
+		UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF,
+		UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+	/* disable dynamic power gating mode */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_POWER_STATUS), 0,
+			~UVD_POWER_STATUS__UVD_PG_MODE_MASK);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(UVD, 0, mmUVD_STATUS);
+
+	return 0;
+}
+
+static int vcn_v1_0_stop(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int r;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)
+		r = vcn_v1_0_stop_dpg_mode(vinst);
+	else
+		r = vcn_v1_0_stop_spg_mode(vinst);
+
+	return r;
+}
+
+static int vcn_v1_0_pause_dpg_mode(struct amdgpu_vcn_inst *vinst,
+				   struct dpg_pause_state *new_state)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	int ret_code;
+	uint32_t reg_data = 0;
+	uint32_t reg_data2 = 0;
+	struct amdgpu_ring *ring;
+
+	/* pause/unpause if state is changed */
+	if (adev->vcn.inst[inst_idx].pause_state.fw_based != new_state->fw_based) {
+		DRM_DEBUG("dpg pause state changed %d:%d -> %d:%d",
+			adev->vcn.inst[inst_idx].pause_state.fw_based,
+			adev->vcn.inst[inst_idx].pause_state.jpeg,
+			new_state->fw_based, new_state->jpeg);
+
+		reg_data = RREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE) &
+			(~UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK);
+
+		if (new_state->fw_based == VCN_DPG_STATE__PAUSE) {
+			ret_code = 0;
+
+			if (!(reg_data & UVD_DPG_PAUSE__JPEG_PAUSE_DPG_ACK_MASK))
+				ret_code = SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS,
+						   UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF,
+						   UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+			if (!ret_code) {
+				/* pause DPG non-jpeg */
+				reg_data |= UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
+				WREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE, reg_data);
+				SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_DPG_PAUSE,
+						   UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK,
+						   UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK);
+
+				/* Restore */
+				ring = &adev->vcn.inst->ring_enc[0];
+				WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO, ring->gpu_addr);
+				WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+				WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE, ring->ring_size / 4);
+				WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR, lower_32_bits(ring->wptr));
+				WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR, lower_32_bits(ring->wptr));
+
+				ring = &adev->vcn.inst->ring_enc[1];
+				WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO2, ring->gpu_addr);
+				WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
+				WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE2, ring->ring_size / 4);
+				WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR2, lower_32_bits(ring->wptr));
+				WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
+
+				WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR,
+						   RREG32_SOC15(UVD, 0, mmUVD_SCRATCH2) & 0x7FFFFFFF);
+				SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS,
+						   UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON,
+						   UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+			}
+		} else {
+			/* unpause dpg non-jpeg, no need to wait */
+			reg_data &= ~UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
+			WREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE, reg_data);
+		}
+		adev->vcn.inst[inst_idx].pause_state.fw_based = new_state->fw_based;
+	}
+
+	/* pause/unpause if state is changed */
+	if (adev->vcn.inst[inst_idx].pause_state.jpeg != new_state->jpeg) {
+		DRM_DEBUG("dpg pause state changed %d:%d -> %d:%d",
+			adev->vcn.inst[inst_idx].pause_state.fw_based,
+			adev->vcn.inst[inst_idx].pause_state.jpeg,
+			new_state->fw_based, new_state->jpeg);
+
+		reg_data = RREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE) &
+			(~UVD_DPG_PAUSE__JPEG_PAUSE_DPG_ACK_MASK);
+
+		if (new_state->jpeg == VCN_DPG_STATE__PAUSE) {
+			ret_code = 0;
+
+			if (!(reg_data & UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK))
+				ret_code = SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS,
+						   UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF,
+						   UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+			if (!ret_code) {
+				/* Make sure JPRG Snoop is disabled before sending the pause */
+				reg_data2 = RREG32_SOC15(UVD, 0, mmUVD_POWER_STATUS);
+				reg_data2 |= UVD_POWER_STATUS__JRBC_SNOOP_DIS_MASK;
+				WREG32_SOC15(UVD, 0, mmUVD_POWER_STATUS, reg_data2);
+
+				/* pause DPG jpeg */
+				reg_data |= UVD_DPG_PAUSE__JPEG_PAUSE_DPG_REQ_MASK;
+				WREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE, reg_data);
+				SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_DPG_PAUSE,
+							UVD_DPG_PAUSE__JPEG_PAUSE_DPG_ACK_MASK,
+							UVD_DPG_PAUSE__JPEG_PAUSE_DPG_ACK_MASK);
+
+				/* Restore */
+				ring = adev->jpeg.inst->ring_dec;
+				WREG32_SOC15(UVD, 0, mmUVD_LMI_JRBC_RB_VMID, 0);
+				WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_CNTL,
+							UVD_JRBC_RB_CNTL__RB_NO_FETCH_MASK |
+							UVD_JRBC_RB_CNTL__RB_RPTR_WR_EN_MASK);
+				WREG32_SOC15(UVD, 0, mmUVD_LMI_JRBC_RB_64BIT_BAR_LOW,
+							lower_32_bits(ring->gpu_addr));
+				WREG32_SOC15(UVD, 0, mmUVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
+							upper_32_bits(ring->gpu_addr));
+				WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_RPTR, ring->wptr);
+				WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_WPTR, ring->wptr);
+				WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_CNTL,
+							UVD_JRBC_RB_CNTL__RB_RPTR_WR_EN_MASK);
+
+				WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR,
+						   RREG32_SOC15(UVD, 0, mmUVD_SCRATCH2) & 0x7FFFFFFF);
+				SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS,
+						   UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON,
+						   UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+			}
+		} else {
+			/* unpause dpg jpeg, no need to wait */
+			reg_data &= ~UVD_DPG_PAUSE__JPEG_PAUSE_DPG_REQ_MASK;
+			WREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE, reg_data);
+		}
+		adev->vcn.inst[inst_idx].pause_state.jpeg = new_state->jpeg;
+	}
+
+	return 0;
+}
+
+static bool vcn_v1_0_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	return (RREG32_SOC15(VCN, 0, mmUVD_STATUS) == UVD_STATUS__IDLE);
+}
+
+static int vcn_v1_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int ret;
+
+	ret = SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_STATUS, UVD_STATUS__IDLE,
+		UVD_STATUS__IDLE);
+
+	return ret;
+}
+
+static int vcn_v1_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
-	/* needed for driver unload*/
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_vcn_inst *vinst = adev->vcn.inst;
+	bool enable = (state == AMD_CG_STATE_GATE);
+
+	if (enable) {
+		/* wait for STATUS to clear */
+		if (!vcn_v1_0_is_idle(ip_block))
+			return -EBUSY;
+		vcn_v1_0_enable_clock_gating(vinst);
+	} else {
+		/* disable HW gating and enable Sw gating */
+		vcn_v1_0_disable_clock_gating(vinst);
+	}
 	return 0;
 }
 
@@ -732,6 +1491,10 @@ static void vcn_v1_0_dec_ring_set_wptr(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
 
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)
+		WREG32_SOC15(UVD, 0, mmUVD_SCRATCH2,
+			lower_32_bits(ring->wptr) | 0x80000000);
+
 	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr));
 }
 
@@ -744,6 +1507,8 @@ static void vcn_v1_0_dec_ring_set_wptr(struct amdgpu_ring *ring)
  */
 static void vcn_v1_0_dec_ring_insert_start(struct amdgpu_ring *ring)
 {
+	struct amdgpu_device *adev = ring->adev;
+
 	amdgpu_ring_write(ring,
 		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0), 0));
 	amdgpu_ring_write(ring, 0);
@@ -761,6 +1526,8 @@ static void vcn_v1_0_dec_ring_insert_start(struct amdgpu_ring *ring)
  */
 static void vcn_v1_0_dec_ring_insert_end(struct amdgpu_ring *ring)
 {
+	struct amdgpu_device *adev = ring->adev;
+
 	amdgpu_ring_write(ring,
 		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD), 0));
 	amdgpu_ring_write(ring, VCN_DEC_CMD_PACKET_END << 1);
@@ -770,13 +1537,17 @@ static void vcn_v1_0_dec_ring_insert_end(struct amdgpu_ring *ring)
  * vcn_v1_0_dec_ring_emit_fence - emit an fence & trap command
  *
  * @ring: amdgpu_ring pointer
- * @fence: fence to emit
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
  *
  * Write a fence and a trap command to the ring.
  */
 static void vcn_v1_0_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
 				     unsigned flags)
 {
+	struct amdgpu_device *adev = ring->adev;
+
 	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
 
 	amdgpu_ring_write(ring,
@@ -804,33 +1575,26 @@ static void vcn_v1_0_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64
 }
 
 /**
- * vcn_v1_0_dec_ring_hdp_invalidate - emit an hdp invalidate
- *
- * @ring: amdgpu_ring pointer
- *
- * Emits an hdp invalidate.
- */
-static void vcn_v1_0_dec_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
-{
-	amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(HDP, 0, mmHDP_DEBUG0), 0));
-	amdgpu_ring_write(ring, 1);
-}
-
-/**
  * vcn_v1_0_dec_ring_emit_ib - execute indirect buffer
  *
  * @ring: amdgpu_ring pointer
+ * @job: job to retrieve vmid from
  * @ib: indirect buffer to execute
+ * @flags: unused
  *
  * Write ring commands to execute the indirect buffer
  */
 static void vcn_v1_0_dec_ring_emit_ib(struct amdgpu_ring *ring,
-				  struct amdgpu_ib *ib,
-				  unsigned vm_id, bool ctx_switch)
+					struct amdgpu_job *job,
+					struct amdgpu_ib *ib,
+					uint32_t flags)
 {
+	struct amdgpu_device *adev = ring->adev;
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+
 	amdgpu_ring_write(ring,
 		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_RBC_IB_VMID), 0));
-	amdgpu_ring_write(ring, vm_id);
+	amdgpu_ring_write(ring, vmid);
 
 	amdgpu_ring_write(ring,
 		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_RBC_IB_64BIT_BAR_LOW), 0));
@@ -843,29 +1607,18 @@ static void vcn_v1_0_dec_ring_emit_ib(struct amdgpu_ring *ring,
 	amdgpu_ring_write(ring, ib->length_dw);
 }
 
-static void vcn_v1_0_dec_vm_reg_write(struct amdgpu_ring *ring,
-				uint32_t data0, uint32_t data1)
+static void vcn_v1_0_dec_ring_emit_reg_wait(struct amdgpu_ring *ring,
+					    uint32_t reg, uint32_t val,
+					    uint32_t mask)
 {
-	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0), 0));
-	amdgpu_ring_write(ring, data0);
-	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA1), 0));
-	amdgpu_ring_write(ring, data1);
-	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD), 0));
-	amdgpu_ring_write(ring, VCN_DEC_CMD_WRITE_REG << 1);
-}
+	struct amdgpu_device *adev = ring->adev;
 
-static void vcn_v1_0_dec_vm_reg_wait(struct amdgpu_ring *ring,
-				uint32_t data0, uint32_t data1, uint32_t mask)
-{
 	amdgpu_ring_write(ring,
 		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0), 0));
-	amdgpu_ring_write(ring, data0);
+	amdgpu_ring_write(ring, reg << 2);
 	amdgpu_ring_write(ring,
 		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA1), 0));
-	amdgpu_ring_write(ring, data1);
+	amdgpu_ring_write(ring, val);
 	amdgpu_ring_write(ring,
 		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GP_SCRATCH8), 0));
 	amdgpu_ring_write(ring, mask);
@@ -875,39 +1628,34 @@ static void vcn_v1_0_dec_vm_reg_wait(struct amdgpu_ring *ring,
 }
 
 static void vcn_v1_0_dec_ring_emit_vm_flush(struct amdgpu_ring *ring,
-					unsigned vm_id, uint64_t pd_addr)
+					    unsigned vmid, uint64_t pd_addr)
 {
-	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
-	uint32_t req = ring->adev->gart.gart_funcs->get_invalidate_req(vm_id);
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
 	uint32_t data0, data1, mask;
-	unsigned eng = ring->vm_inv_eng;
-
-	pd_addr = amdgpu_gart_get_vm_pde(ring->adev, pd_addr);
-	pd_addr |= AMDGPU_PTE_VALID;
 
-	data0 = (hub->ctx0_ptb_addr_hi32 + vm_id * 2) << 2;
-	data1 = upper_32_bits(pd_addr);
-	vcn_v1_0_dec_vm_reg_write(ring, data0, data1);
+	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
 
-	data0 = (hub->ctx0_ptb_addr_lo32 + vm_id * 2) << 2;
-	data1 = lower_32_bits(pd_addr);
-	vcn_v1_0_dec_vm_reg_write(ring, data0, data1);
-
-	data0 = (hub->ctx0_ptb_addr_lo32 + vm_id * 2) << 2;
+	/* wait for register write */
+	data0 = hub->ctx0_ptb_addr_lo32 + vmid * hub->ctx_addr_distance;
 	data1 = lower_32_bits(pd_addr);
 	mask = 0xffffffff;
-	vcn_v1_0_dec_vm_reg_wait(ring, data0, data1, mask);
-
-	/* flush TLB */
-	data0 = (hub->vm_inv_eng0_req + eng) << 2;
-	data1 = req;
-	vcn_v1_0_dec_vm_reg_write(ring, data0, data1);
-
-	/* wait for flush */
-	data0 = (hub->vm_inv_eng0_ack + eng) << 2;
-	data1 = 1 << vm_id;
-	mask =  1 << vm_id;
-	vcn_v1_0_dec_vm_reg_wait(ring, data0, data1, mask);
+	vcn_v1_0_dec_ring_emit_reg_wait(ring, data0, data1, mask);
+}
+
+static void vcn_v1_0_dec_ring_emit_wreg(struct amdgpu_ring *ring,
+					uint32_t reg, uint32_t val)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0), 0));
+	amdgpu_ring_write(ring, reg << 2);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA1), 0));
+	amdgpu_ring_write(ring, val);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD), 0));
+	amdgpu_ring_write(ring, VCN_DEC_CMD_WRITE_REG << 1);
 }
 
 /**
@@ -921,7 +1669,7 @@ static uint64_t vcn_v1_0_enc_ring_get_rptr(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
 
-	if (ring == &adev->vcn.ring_enc[0])
+	if (ring == &adev->vcn.inst->ring_enc[0])
 		return RREG32_SOC15(UVD, 0, mmUVD_RB_RPTR);
 	else
 		return RREG32_SOC15(UVD, 0, mmUVD_RB_RPTR2);
@@ -938,7 +1686,7 @@ static uint64_t vcn_v1_0_enc_ring_get_wptr(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
 
-	if (ring == &adev->vcn.ring_enc[0])
+	if (ring == &adev->vcn.inst->ring_enc[0])
 		return RREG32_SOC15(UVD, 0, mmUVD_RB_WPTR);
 	else
 		return RREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2);
@@ -955,7 +1703,7 @@ static void vcn_v1_0_enc_ring_set_wptr(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
 
-	if (ring == &adev->vcn.ring_enc[0])
+	if (ring == &adev->vcn.inst->ring_enc[0])
 		WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR,
 			lower_32_bits(ring->wptr));
 	else
@@ -967,7 +1715,9 @@ static void vcn_v1_0_enc_ring_set_wptr(struct amdgpu_ring *ring)
  * vcn_v1_0_enc_ring_emit_fence - emit an enc fence & trap command
  *
  * @ring: amdgpu_ring pointer
- * @fence: fence to emit
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
  *
  * Write enc a fence and a trap command to the ring.
  */
@@ -992,56 +1742,55 @@ static void vcn_v1_0_enc_ring_insert_end(struct amdgpu_ring *ring)
  * vcn_v1_0_enc_ring_emit_ib - enc execute indirect buffer
  *
  * @ring: amdgpu_ring pointer
+ * @job: job to retrive vmid from
  * @ib: indirect buffer to execute
+ * @flags: unused
  *
  * Write enc ring commands to execute the indirect buffer
  */
 static void vcn_v1_0_enc_ring_emit_ib(struct amdgpu_ring *ring,
-		struct amdgpu_ib *ib, unsigned int vm_id, bool ctx_switch)
+					struct amdgpu_job *job,
+					struct amdgpu_ib *ib,
+					uint32_t flags)
 {
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+
 	amdgpu_ring_write(ring, VCN_ENC_CMD_IB);
-	amdgpu_ring_write(ring, vm_id);
+	amdgpu_ring_write(ring, vmid);
 	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
 	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
 	amdgpu_ring_write(ring, ib->length_dw);
 }
 
-static void vcn_v1_0_enc_ring_emit_vm_flush(struct amdgpu_ring *ring,
-			 unsigned int vm_id, uint64_t pd_addr)
+static void vcn_v1_0_enc_ring_emit_reg_wait(struct amdgpu_ring *ring,
+					    uint32_t reg, uint32_t val,
+					    uint32_t mask)
 {
-	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
-	uint32_t req = ring->adev->gart.gart_funcs->get_invalidate_req(vm_id);
-	unsigned eng = ring->vm_inv_eng;
-
-	pd_addr = amdgpu_gart_get_vm_pde(ring->adev, pd_addr);
-	pd_addr |= AMDGPU_PTE_VALID;
+	amdgpu_ring_write(ring, VCN_ENC_CMD_REG_WAIT);
+	amdgpu_ring_write(ring, reg << 2);
+	amdgpu_ring_write(ring, mask);
+	amdgpu_ring_write(ring, val);
+}
 
-	amdgpu_ring_write(ring, VCN_ENC_CMD_REG_WRITE);
-	amdgpu_ring_write(ring,
-			  (hub->ctx0_ptb_addr_hi32 + vm_id * 2) << 2);
-	amdgpu_ring_write(ring, upper_32_bits(pd_addr));
+static void vcn_v1_0_enc_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					    unsigned int vmid, uint64_t pd_addr)
+{
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
 
-	amdgpu_ring_write(ring, VCN_ENC_CMD_REG_WRITE);
-	amdgpu_ring_write(ring,
-			  (hub->ctx0_ptb_addr_lo32 + vm_id * 2) << 2);
-	amdgpu_ring_write(ring, lower_32_bits(pd_addr));
+	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
 
-	amdgpu_ring_write(ring, VCN_ENC_CMD_REG_WAIT);
-	amdgpu_ring_write(ring,
-			  (hub->ctx0_ptb_addr_lo32 + vm_id * 2) << 2);
-	amdgpu_ring_write(ring, 0xffffffff);
-	amdgpu_ring_write(ring, lower_32_bits(pd_addr));
+	/* wait for reg writes */
+	vcn_v1_0_enc_ring_emit_reg_wait(ring, hub->ctx0_ptb_addr_lo32 +
+					vmid * hub->ctx_addr_distance,
+					lower_32_bits(pd_addr), 0xffffffff);
+}
 
-	/* flush TLB */
+static void vcn_v1_0_enc_ring_emit_wreg(struct amdgpu_ring *ring,
+					uint32_t reg, uint32_t val)
+{
 	amdgpu_ring_write(ring, VCN_ENC_CMD_REG_WRITE);
-	amdgpu_ring_write(ring,	(hub->vm_inv_eng0_req + eng) << 2);
-	amdgpu_ring_write(ring, req);
-
-	/* wait for flush */
-	amdgpu_ring_write(ring, VCN_ENC_CMD_REG_WAIT);
-	amdgpu_ring_write(ring, (hub->vm_inv_eng0_ack + eng) << 2);
-	amdgpu_ring_write(ring, 1 << vm_id);
-	amdgpu_ring_write(ring, 1 << vm_id);
+	amdgpu_ring_write(ring,	reg << 2);
+	amdgpu_ring_write(ring, val);
 }
 
 static int vcn_v1_0_set_interrupt_state(struct amdgpu_device *adev,
@@ -1060,13 +1809,13 @@ static int vcn_v1_0_process_interrupt(struct amdgpu_device *adev,
 
 	switch (entry->src_id) {
 	case 124:
-		amdgpu_fence_process(&adev->vcn.ring_dec);
+		amdgpu_fence_process(&adev->vcn.inst->ring_dec);
 		break;
 	case 119:
-		amdgpu_fence_process(&adev->vcn.ring_enc[0]);
+		amdgpu_fence_process(&adev->vcn.inst->ring_enc[0]);
 		break;
 	case 120:
-		amdgpu_fence_process(&adev->vcn.ring_enc[1]);
+		amdgpu_fence_process(&adev->vcn.inst->ring_enc[1]);
 		break;
 	default:
 		DRM_ERROR("Unhandled interrupt: %d %d\n",
@@ -1077,53 +1826,323 @@ static int vcn_v1_0_process_interrupt(struct amdgpu_device *adev,
 	return 0;
 }
 
+static void vcn_v1_0_dec_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int i;
+
+	WARN_ON(ring->wptr % 2 || count % 2);
+
+	for (i = 0; i < count / 2; i++) {
+		amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_NO_OP), 0));
+		amdgpu_ring_write(ring, 0);
+	}
+}
+
+static int vcn_v1_0_set_pg_state(struct amdgpu_vcn_inst *vinst,
+				 enum amd_powergating_state state)
+{
+	/* This doesn't actually powergate the VCN block.
+	 * That's done in the dpm code via the SMC.  This
+	 * just re-inits the block as necessary.  The actual
+	 * gating still happens in the dpm code.  We should
+	 * revisit this when there is a cleaner line between
+	 * the smc and the hw blocks
+	 */
+	int ret;
+
+	if (state == vinst->cur_state)
+		return 0;
+
+	if (state == AMD_PG_STATE_GATE)
+		ret = vcn_v1_0_stop(vinst);
+	else
+		ret = vcn_v1_0_start(vinst);
+
+	if (!ret)
+		vinst->cur_state = state;
+
+	return ret;
+}
+
+static void vcn_v1_0_idle_work_handler(struct work_struct *work)
+{
+	struct amdgpu_vcn_inst *vcn_inst =
+		container_of(work, struct amdgpu_vcn_inst, idle_work.work);
+	struct amdgpu_device *adev = vcn_inst->adev;
+	unsigned int fences = 0, i;
+
+	for (i = 0; i < adev->vcn.inst[0].num_enc_rings; ++i)
+		fences += amdgpu_fence_count_emitted(&adev->vcn.inst->ring_enc[i]);
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) {
+		struct dpg_pause_state new_state;
+
+		if (fences)
+			new_state.fw_based = VCN_DPG_STATE__PAUSE;
+		else
+			new_state.fw_based = VCN_DPG_STATE__UNPAUSE;
+
+		if (amdgpu_fence_count_emitted(adev->jpeg.inst->ring_dec))
+			new_state.jpeg = VCN_DPG_STATE__PAUSE;
+		else
+			new_state.jpeg = VCN_DPG_STATE__UNPAUSE;
+
+		adev->vcn.inst->pause_dpg_mode(vcn_inst, &new_state);
+	}
+
+	fences += amdgpu_fence_count_emitted(adev->jpeg.inst->ring_dec);
+	fences += amdgpu_fence_count_emitted(&adev->vcn.inst->ring_dec);
+
+	if (fences == 0) {
+		amdgpu_gfx_off_ctrl(adev, true);
+		if (adev->pm.dpm_enabled)
+			amdgpu_dpm_enable_vcn(adev, false, 0);
+		else
+			amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN,
+			       AMD_PG_STATE_GATE);
+	} else {
+		schedule_delayed_work(&adev->vcn.inst[0].idle_work, VCN_IDLE_TIMEOUT);
+	}
+}
+
+static void vcn_v1_0_ring_begin_use(struct amdgpu_ring *ring)
+{
+	struct	amdgpu_device *adev = ring->adev;
+	bool set_clocks = !cancel_delayed_work_sync(&adev->vcn.inst[0].idle_work);
+
+	mutex_lock(&adev->vcn.inst[0].vcn1_jpeg1_workaround);
+
+	if (amdgpu_fence_wait_empty(ring->adev->jpeg.inst->ring_dec))
+		DRM_ERROR("VCN dec: jpeg dec ring may not be empty\n");
+
+	vcn_v1_0_set_pg_for_begin_use(ring, set_clocks);
+
+}
+
+void vcn_v1_0_set_pg_for_begin_use(struct amdgpu_ring *ring, bool set_clocks)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (set_clocks) {
+		amdgpu_gfx_off_ctrl(adev, false);
+		if (adev->pm.dpm_enabled)
+			amdgpu_dpm_enable_vcn(adev, true, 0);
+		else
+			amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN,
+			       AMD_PG_STATE_UNGATE);
+	}
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) {
+		struct dpg_pause_state new_state;
+		unsigned int fences = 0, i;
+
+		for (i = 0; i < adev->vcn.inst[0].num_enc_rings; ++i)
+			fences += amdgpu_fence_count_emitted(&adev->vcn.inst->ring_enc[i]);
+
+		if (fences)
+			new_state.fw_based = VCN_DPG_STATE__PAUSE;
+		else
+			new_state.fw_based = VCN_DPG_STATE__UNPAUSE;
+
+		if (amdgpu_fence_count_emitted(adev->jpeg.inst->ring_dec))
+			new_state.jpeg = VCN_DPG_STATE__PAUSE;
+		else
+			new_state.jpeg = VCN_DPG_STATE__UNPAUSE;
+
+		if (ring->funcs->type == AMDGPU_RING_TYPE_VCN_ENC)
+			new_state.fw_based = VCN_DPG_STATE__PAUSE;
+		else if (ring->funcs->type == AMDGPU_RING_TYPE_VCN_JPEG)
+			new_state.jpeg = VCN_DPG_STATE__PAUSE;
+
+		adev->vcn.inst->pause_dpg_mode(adev->vcn.inst, &new_state);
+	}
+}
+
+void vcn_v1_0_ring_end_use(struct amdgpu_ring *ring)
+{
+	schedule_delayed_work(&ring->adev->vcn.inst[0].idle_work, VCN_IDLE_TIMEOUT);
+	mutex_unlock(&ring->adev->vcn.inst[0].vcn1_jpeg1_workaround);
+}
+
+static void vcn_v1_0_print_ip_state(struct amdgpu_ip_block *ip_block, struct drm_printer *p)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, j;
+	uint32_t reg_count = ARRAY_SIZE(vcn_reg_list_1_0);
+	uint32_t inst_off, is_powered;
+
+	if (!adev->vcn.ip_dump)
+		return;
+
+	drm_printf(p, "num_instances:%d\n", adev->vcn.num_vcn_inst);
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		if (adev->vcn.harvest_config & (1 << i)) {
+			drm_printf(p, "\nHarvested Instance:VCN%d Skipping dump\n", i);
+			continue;
+		}
+
+		inst_off = i * reg_count;
+		is_powered = (adev->vcn.ip_dump[inst_off] &
+				UVD_POWER_STATUS__UVD_POWER_STATUS_MASK) != 1;
+
+		if (is_powered) {
+			drm_printf(p, "\nActive Instance:VCN%d\n", i);
+			for (j = 0; j < reg_count; j++)
+				drm_printf(p, "%-50s \t 0x%08x\n", vcn_reg_list_1_0[j].reg_name,
+					   adev->vcn.ip_dump[inst_off + j]);
+		} else {
+			drm_printf(p, "\nInactive Instance:VCN%d\n", i);
+		}
+	}
+}
+
+static void vcn_v1_0_dump_ip_state(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, j;
+	bool is_powered;
+	uint32_t inst_off;
+	uint32_t reg_count = ARRAY_SIZE(vcn_reg_list_1_0);
+
+	if (!adev->vcn.ip_dump)
+		return;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		inst_off = i * reg_count;
+		/* mmUVD_POWER_STATUS is always readable and is first element of the array */
+		adev->vcn.ip_dump[inst_off] = RREG32_SOC15(VCN, i, mmUVD_POWER_STATUS);
+		is_powered = (adev->vcn.ip_dump[inst_off] &
+				UVD_POWER_STATUS__UVD_POWER_STATUS_MASK) != 1;
+
+		if (is_powered)
+			for (j = 1; j < reg_count; j++)
+				adev->vcn.ip_dump[inst_off + j] =
+					RREG32(SOC15_REG_ENTRY_OFFSET_INST(vcn_reg_list_1_0[j], i));
+	}
+}
+
 static const struct amd_ip_funcs vcn_v1_0_ip_funcs = {
 	.name = "vcn_v1_0",
 	.early_init = vcn_v1_0_early_init,
-	.late_init = NULL,
 	.sw_init = vcn_v1_0_sw_init,
 	.sw_fini = vcn_v1_0_sw_fini,
 	.hw_init = vcn_v1_0_hw_init,
 	.hw_fini = vcn_v1_0_hw_fini,
 	.suspend = vcn_v1_0_suspend,
 	.resume = vcn_v1_0_resume,
-	.is_idle = NULL /* vcn_v1_0_is_idle */,
-	.wait_for_idle = NULL /* vcn_v1_0_wait_for_idle */,
-	.check_soft_reset = NULL /* vcn_v1_0_check_soft_reset */,
-	.pre_soft_reset = NULL /* vcn_v1_0_pre_soft_reset */,
-	.soft_reset = NULL /* vcn_v1_0_soft_reset */,
-	.post_soft_reset = NULL /* vcn_v1_0_post_soft_reset */,
+	.is_idle = vcn_v1_0_is_idle,
+	.wait_for_idle = vcn_v1_0_wait_for_idle,
 	.set_clockgating_state = vcn_v1_0_set_clockgating_state,
-	.set_powergating_state = NULL /* vcn_v1_0_set_powergating_state */,
+	.set_powergating_state = vcn_set_powergating_state,
+	.dump_ip_state = vcn_v1_0_dump_ip_state,
+	.print_ip_state = vcn_v1_0_print_ip_state,
 };
 
+/*
+ * It is a hardware issue that VCN can't handle a GTT TMZ buffer on
+ * CHIP_RAVEN series ASIC. Move such a GTT TMZ buffer to VRAM domain
+ * before command submission as a workaround.
+ */
+static int vcn_v1_0_validate_bo(struct amdgpu_cs_parser *parser,
+				struct amdgpu_job *job,
+				uint64_t addr)
+{
+	struct ttm_operation_ctx ctx = { false, false };
+	struct amdgpu_fpriv *fpriv = parser->filp->driver_priv;
+	struct amdgpu_vm *vm = &fpriv->vm;
+	struct amdgpu_bo_va_mapping *mapping;
+	struct amdgpu_bo *bo;
+	int r;
+
+	addr &= AMDGPU_GMC_HOLE_MASK;
+	if (addr & 0x7) {
+		DRM_ERROR("VCN messages must be 8 byte aligned!\n");
+		return -EINVAL;
+	}
+
+	mapping = amdgpu_vm_bo_lookup_mapping(vm, addr/AMDGPU_GPU_PAGE_SIZE);
+	if (!mapping || !mapping->bo_va || !mapping->bo_va->base.bo)
+		return -EINVAL;
+
+	bo = mapping->bo_va->base.bo;
+	if (!(bo->flags & AMDGPU_GEM_CREATE_ENCRYPTED))
+		return 0;
+
+	amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_VRAM);
+	r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+	if (r) {
+		DRM_ERROR("Failed to validate the VCN message BO (%d)!\n", r);
+		return r;
+	}
+
+	return r;
+}
+
+static int vcn_v1_0_ring_patch_cs_in_place(struct amdgpu_cs_parser *p,
+					   struct amdgpu_job *job,
+					   struct amdgpu_ib *ib)
+{
+	uint32_t msg_lo = 0, msg_hi = 0;
+	int i, r;
+
+	if (!(ib->flags & AMDGPU_IB_FLAGS_SECURE))
+		return 0;
+
+	for (i = 0; i < ib->length_dw; i += 2) {
+		uint32_t reg = amdgpu_ib_get_value(ib, i);
+		uint32_t val = amdgpu_ib_get_value(ib, i + 1);
+
+		if (reg == PACKET0(p->adev->vcn.inst[0].internal.data0, 0)) {
+			msg_lo = val;
+		} else if (reg == PACKET0(p->adev->vcn.inst[0].internal.data1, 0)) {
+			msg_hi = val;
+		} else if (reg == PACKET0(p->adev->vcn.inst[0].internal.cmd, 0)) {
+			r = vcn_v1_0_validate_bo(p, job,
+						 ((u64)msg_hi) << 32 | msg_lo);
+			if (r)
+				return r;
+		}
+	}
+
+	return 0;
+}
+
 static const struct amdgpu_ring_funcs vcn_v1_0_dec_ring_vm_funcs = {
 	.type = AMDGPU_RING_TYPE_VCN_DEC,
 	.align_mask = 0xf,
-	.nop = PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_NO_OP), 0),
 	.support_64bit_ptrs = false,
-	.vmhub = AMDGPU_MMHUB,
+	.no_user_fence = true,
+	.secure_submission_supported = true,
 	.get_rptr = vcn_v1_0_dec_ring_get_rptr,
 	.get_wptr = vcn_v1_0_dec_ring_get_wptr,
 	.set_wptr = vcn_v1_0_dec_ring_set_wptr,
+	.patch_cs_in_place = vcn_v1_0_ring_patch_cs_in_place,
 	.emit_frame_size =
-		2 + /* vcn_v1_0_dec_ring_emit_hdp_invalidate */
-		34 + /* vcn_v1_0_dec_ring_emit_vm_flush */
+		6 + 6 + /* hdp invalidate / flush */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
+		8 + /* vcn_v1_0_dec_ring_emit_vm_flush */
 		14 + 14 + /* vcn_v1_0_dec_ring_emit_fence x2 vm fence */
 		6,
 	.emit_ib_size = 8, /* vcn_v1_0_dec_ring_emit_ib */
 	.emit_ib = vcn_v1_0_dec_ring_emit_ib,
 	.emit_fence = vcn_v1_0_dec_ring_emit_fence,
 	.emit_vm_flush = vcn_v1_0_dec_ring_emit_vm_flush,
-	.emit_hdp_invalidate = vcn_v1_0_dec_ring_emit_hdp_invalidate,
 	.test_ring = amdgpu_vcn_dec_ring_test_ring,
 	.test_ib = amdgpu_vcn_dec_ring_test_ib,
-	.insert_nop = amdgpu_ring_insert_nop,
+	.insert_nop = vcn_v1_0_dec_ring_insert_nop,
 	.insert_start = vcn_v1_0_dec_ring_insert_start,
 	.insert_end = vcn_v1_0_dec_ring_insert_end,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
-	.begin_use = amdgpu_vcn_ring_begin_use,
-	.end_use = amdgpu_vcn_ring_end_use,
+	.begin_use = vcn_v1_0_ring_begin_use,
+	.end_use = vcn_v1_0_ring_end_use,
+	.emit_wreg = vcn_v1_0_dec_ring_emit_wreg,
+	.emit_reg_wait = vcn_v1_0_dec_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
 };
 
 static const struct amdgpu_ring_funcs vcn_v1_0_enc_ring_vm_funcs = {
@@ -1131,12 +2150,14 @@ static const struct amdgpu_ring_funcs vcn_v1_0_enc_ring_vm_funcs = {
 	.align_mask = 0x3f,
 	.nop = VCN_ENC_CMD_NO_OP,
 	.support_64bit_ptrs = false,
-	.vmhub = AMDGPU_MMHUB,
+	.no_user_fence = true,
 	.get_rptr = vcn_v1_0_enc_ring_get_rptr,
 	.get_wptr = vcn_v1_0_enc_ring_get_wptr,
 	.set_wptr = vcn_v1_0_enc_ring_set_wptr,
 	.emit_frame_size =
-		17 + /* vcn_v1_0_enc_ring_emit_vm_flush */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 4 +
+		4 + /* vcn_v1_0_enc_ring_emit_vm_flush */
 		5 + 5 + /* vcn_v1_0_enc_ring_emit_fence x2 vm fence */
 		1, /* vcn_v1_0_enc_ring_insert_end */
 	.emit_ib_size = 5, /* vcn_v1_0_enc_ring_emit_ib */
@@ -1148,24 +2169,24 @@ static const struct amdgpu_ring_funcs vcn_v1_0_enc_ring_vm_funcs = {
 	.insert_nop = amdgpu_ring_insert_nop,
 	.insert_end = vcn_v1_0_enc_ring_insert_end,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
-	.begin_use = amdgpu_vcn_ring_begin_use,
-	.end_use = amdgpu_vcn_ring_end_use,
+	.begin_use = vcn_v1_0_ring_begin_use,
+	.end_use = vcn_v1_0_ring_end_use,
+	.emit_wreg = vcn_v1_0_enc_ring_emit_wreg,
+	.emit_reg_wait = vcn_v1_0_enc_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
 };
 
 static void vcn_v1_0_set_dec_ring_funcs(struct amdgpu_device *adev)
 {
-	adev->vcn.ring_dec.funcs = &vcn_v1_0_dec_ring_vm_funcs;
-	DRM_INFO("VCN decode is enabled in VM mode\n");
+	adev->vcn.inst->ring_dec.funcs = &vcn_v1_0_dec_ring_vm_funcs;
 }
 
 static void vcn_v1_0_set_enc_ring_funcs(struct amdgpu_device *adev)
 {
 	int i;
 
-	for (i = 0; i < adev->vcn.num_enc_rings; ++i)
-		adev->vcn.ring_enc[i].funcs = &vcn_v1_0_enc_ring_vm_funcs;
-
-	DRM_INFO("VCN encode is enabled in VM mode\n");
+	for (i = 0; i < adev->vcn.inst[0].num_enc_rings; ++i)
+		adev->vcn.inst->ring_enc[i].funcs = &vcn_v1_0_enc_ring_vm_funcs;
 }
 
 static const struct amdgpu_irq_src_funcs vcn_v1_0_irq_funcs = {
@@ -1175,12 +2196,11 @@ static const struct amdgpu_irq_src_funcs vcn_v1_0_irq_funcs = {
 
 static void vcn_v1_0_set_irq_funcs(struct amdgpu_device *adev)
 {
-	adev->uvd.irq.num_types = adev->vcn.num_enc_rings + 1;
-	adev->vcn.irq.funcs = &vcn_v1_0_irq_funcs;
+	adev->vcn.inst->irq.num_types = adev->vcn.inst[0].num_enc_rings + 2;
+	adev->vcn.inst->irq.funcs = &vcn_v1_0_irq_funcs;
 }
 
-const struct amdgpu_ip_block_version vcn_v1_0_ip_block =
-{
+const struct amdgpu_ip_block_version vcn_v1_0_ip_block = {
 		.type = AMD_IP_BLOCK_TYPE_VCN,
 		.major = 1,
 		.minor = 0,
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.h b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.h
index 2a497a7a4840..1f1cc7f0ece7 100644
--- a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.h
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.h
@@ -24,6 +24,9 @@
 #ifndef __VCN_V1_0_H__
 #define __VCN_V1_0_H__
 
+void vcn_v1_0_ring_end_use(struct amdgpu_ring *ring);
+void vcn_v1_0_set_pg_for_begin_use(struct amdgpu_ring *ring, bool set_clocks);
+
 extern const struct amdgpu_ip_block_version vcn_v1_0_ip_block;
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c
new file mode 100644
index 000000000000..8897dcc9c1a0
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c
@@ -0,0 +1,2204 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include <drm/drm_drv.h>
+
+#include "amdgpu.h"
+#include "amdgpu_vcn.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_psp.h"
+#include "mmsch_v2_0.h"
+#include "vcn_v2_0.h"
+
+#include "vcn/vcn_2_0_0_offset.h"
+#include "vcn/vcn_2_0_0_sh_mask.h"
+#include "ivsrcid/vcn/irqsrcs_vcn_2_0.h"
+
+#define VCN_VID_SOC_ADDRESS_2_0					0x1fa00
+#define VCN1_VID_SOC_ADDRESS_3_0				0x48200
+#define VCN1_AON_SOC_ADDRESS_3_0				0x48000
+
+#define mmUVD_CONTEXT_ID_INTERNAL_OFFSET			0x1fd
+#define mmUVD_GPCOM_VCPU_CMD_INTERNAL_OFFSET			0x503
+#define mmUVD_GPCOM_VCPU_DATA0_INTERNAL_OFFSET			0x504
+#define mmUVD_GPCOM_VCPU_DATA1_INTERNAL_OFFSET			0x505
+#define mmUVD_NO_OP_INTERNAL_OFFSET				0x53f
+#define mmUVD_GP_SCRATCH8_INTERNAL_OFFSET			0x54a
+#define mmUVD_SCRATCH9_INTERNAL_OFFSET				0xc01d
+
+#define mmUVD_LMI_RBC_IB_VMID_INTERNAL_OFFSET			0x1e1
+#define mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH_INTERNAL_OFFSET		0x5a6
+#define mmUVD_LMI_RBC_IB_64BIT_BAR_LOW_INTERNAL_OFFSET		0x5a7
+#define mmUVD_RBC_IB_SIZE_INTERNAL_OFFSET			0x1e2
+
+static const struct amdgpu_hwip_reg_entry vcn_reg_list_2_0[] = {
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_CONTEXT_ID),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_CONTEXT_ID2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_GPCOM_VCPU_DATA0),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_GPCOM_VCPU_DATA1),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_GPCOM_VCPU_CMD),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_HI),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_LO),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_HI2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_LO2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_HI3),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_LO3),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_HI4),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_LO4),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_RPTR),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_WPTR),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_RPTR2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_WPTR2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_RPTR3),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_WPTR3),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_RPTR4),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_WPTR4),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_SIZE),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_SIZE2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_SIZE3),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_SIZE4),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_PGFSM_CONFIG),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_PGFSM_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_DPG_LMA_CTL),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_DPG_LMA_DATA),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_DPG_LMA_MASK),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_DPG_PAUSE)
+};
+
+static void vcn_v2_0_set_dec_ring_funcs(struct amdgpu_device *adev);
+static void vcn_v2_0_set_enc_ring_funcs(struct amdgpu_device *adev);
+static void vcn_v2_0_set_irq_funcs(struct amdgpu_device *adev);
+static int vcn_v2_0_set_pg_state(struct amdgpu_vcn_inst *vinst,
+				 enum amd_powergating_state state);
+static int vcn_v2_0_pause_dpg_mode(struct amdgpu_vcn_inst *vinst,
+				   struct dpg_pause_state *new_state);
+static int vcn_v2_0_start_sriov(struct amdgpu_device *adev);
+static int vcn_v2_0_reset(struct amdgpu_vcn_inst *vinst);
+
+/**
+ * vcn_v2_0_early_init - set function pointers and load microcode
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Set ring and irq function pointers
+ * Load microcode from filesystem
+ */
+static int vcn_v2_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (amdgpu_sriov_vf(adev))
+		adev->vcn.inst[0].num_enc_rings = 1;
+	else
+		adev->vcn.inst[0].num_enc_rings = 2;
+
+	adev->vcn.inst->set_pg_state = vcn_v2_0_set_pg_state;
+	vcn_v2_0_set_dec_ring_funcs(adev);
+	vcn_v2_0_set_enc_ring_funcs(adev);
+	vcn_v2_0_set_irq_funcs(adev);
+
+	return amdgpu_vcn_early_init(adev, 0);
+}
+
+/**
+ * vcn_v2_0_sw_init - sw init for VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Load firmware and sw initialization
+ */
+static int vcn_v2_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_ring *ring;
+	int i, r;
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_fw_shared *fw_shared;
+
+	/* VCN DEC TRAP */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+			      VCN_2_0__SRCID__UVD_SYSTEM_MESSAGE_INTERRUPT,
+			      &adev->vcn.inst->irq);
+	if (r)
+		return r;
+
+	/* VCN ENC TRAP */
+	for (i = 0; i < adev->vcn.inst[0].num_enc_rings; ++i) {
+		r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+				      i + VCN_2_0__SRCID__UVD_ENC_GENERAL_PURPOSE,
+				      &adev->vcn.inst->irq);
+		if (r)
+			return r;
+	}
+
+	r = amdgpu_vcn_sw_init(adev, 0);
+	if (r)
+		return r;
+
+	amdgpu_vcn_setup_ucode(adev, 0);
+
+	r = amdgpu_vcn_resume(adev, 0);
+	if (r)
+		return r;
+
+	ring = &adev->vcn.inst->ring_dec;
+
+	ring->use_doorbell = true;
+	ring->doorbell_index = adev->doorbell_index.vcn.vcn_ring0_1 << 1;
+	ring->vm_hub = AMDGPU_MMHUB0(0);
+
+	sprintf(ring->name, "vcn_dec");
+	r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst->irq, 0,
+			     AMDGPU_RING_PRIO_DEFAULT, NULL);
+	if (r)
+		return r;
+
+	adev->vcn.inst[0].internal.context_id = mmUVD_CONTEXT_ID_INTERNAL_OFFSET;
+	adev->vcn.inst[0].internal.ib_vmid = mmUVD_LMI_RBC_IB_VMID_INTERNAL_OFFSET;
+	adev->vcn.inst[0].internal.ib_bar_low = mmUVD_LMI_RBC_IB_64BIT_BAR_LOW_INTERNAL_OFFSET;
+	adev->vcn.inst[0].internal.ib_bar_high = mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH_INTERNAL_OFFSET;
+	adev->vcn.inst[0].internal.ib_size = mmUVD_RBC_IB_SIZE_INTERNAL_OFFSET;
+	adev->vcn.inst[0].internal.gp_scratch8 = mmUVD_GP_SCRATCH8_INTERNAL_OFFSET;
+
+	adev->vcn.inst[0].internal.scratch9 = mmUVD_SCRATCH9_INTERNAL_OFFSET;
+	adev->vcn.inst->external.scratch9 = SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9);
+	adev->vcn.inst[0].internal.data0 = mmUVD_GPCOM_VCPU_DATA0_INTERNAL_OFFSET;
+	adev->vcn.inst->external.data0 = SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0);
+	adev->vcn.inst[0].internal.data1 = mmUVD_GPCOM_VCPU_DATA1_INTERNAL_OFFSET;
+	adev->vcn.inst->external.data1 = SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA1);
+	adev->vcn.inst[0].internal.cmd = mmUVD_GPCOM_VCPU_CMD_INTERNAL_OFFSET;
+	adev->vcn.inst->external.cmd = SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD);
+	adev->vcn.inst[0].internal.nop = mmUVD_NO_OP_INTERNAL_OFFSET;
+	adev->vcn.inst->external.nop = SOC15_REG_OFFSET(UVD, 0, mmUVD_NO_OP);
+
+	for (i = 0; i < adev->vcn.inst[0].num_enc_rings; ++i) {
+		enum amdgpu_ring_priority_level hw_prio = amdgpu_vcn_get_enc_ring_prio(i);
+
+		ring = &adev->vcn.inst->ring_enc[i];
+		ring->use_doorbell = true;
+		ring->vm_hub = AMDGPU_MMHUB0(0);
+		if (!amdgpu_sriov_vf(adev))
+			ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 2 + i;
+		else
+			ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 1 + i;
+		sprintf(ring->name, "vcn_enc%d", i);
+		r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst->irq, 0,
+				     hw_prio, NULL);
+		if (r)
+			return r;
+	}
+
+	adev->vcn.inst[0].pause_dpg_mode = vcn_v2_0_pause_dpg_mode;
+	adev->vcn.inst[0].reset = vcn_v2_0_reset;
+
+	adev->vcn.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->vcn.inst[0].ring_enc[0]);
+	if (!amdgpu_sriov_vf(adev))
+		adev->vcn.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+
+	r = amdgpu_virt_alloc_mm_table(adev);
+	if (r)
+		return r;
+
+	fw_shared = adev->vcn.inst->fw_shared.cpu_addr;
+	fw_shared->present_flag_0 = cpu_to_le32(AMDGPU_VCN_MULTI_QUEUE_FLAG);
+
+	if (amdgpu_vcnfw_log)
+		amdgpu_vcn_fwlog_init(adev->vcn.inst);
+
+	r = amdgpu_vcn_reg_dump_init(adev, vcn_reg_list_2_0, ARRAY_SIZE(vcn_reg_list_2_0));
+	if (r)
+		return r;
+
+	r = amdgpu_vcn_sysfs_reset_mask_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+/**
+ * vcn_v2_0_sw_fini - sw fini for VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * VCN suspend and free up sw allocation
+ */
+static int vcn_v2_0_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	int r, idx;
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared.cpu_addr;
+
+	if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+		fw_shared->present_flag_0 = 0;
+		drm_dev_exit(idx);
+	}
+
+	amdgpu_virt_free_mm_table(adev);
+
+	r = amdgpu_vcn_suspend(adev, 0);
+	if (r)
+		return r;
+
+	amdgpu_vcn_sysfs_reset_mask_fini(adev);
+
+	amdgpu_vcn_sw_fini(adev, 0);
+
+	return 0;
+}
+
+/**
+ * vcn_v2_0_hw_init - start and test VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Initialize the hardware, boot up the VCPU and do some testing
+ */
+static int vcn_v2_0_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring = &adev->vcn.inst->ring_dec;
+	int i, r;
+
+	adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
+					     ring->doorbell_index, 0);
+
+	if (amdgpu_sriov_vf(adev))
+		vcn_v2_0_start_sriov(adev);
+
+	r = amdgpu_ring_test_helper(ring);
+	if (r)
+		return r;
+
+	//Disable vcn decode for sriov
+	if (amdgpu_sriov_vf(adev))
+		ring->sched.ready = false;
+
+	for (i = 0; i < adev->vcn.inst[0].num_enc_rings; ++i) {
+		ring = &adev->vcn.inst->ring_enc[i];
+		r = amdgpu_ring_test_helper(ring);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v2_0_hw_fini - stop the hardware block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Stop the VCN block, mark ring as not ready any more
+ */
+static int vcn_v2_0_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_vcn_inst *vinst = adev->vcn.inst;
+
+	cancel_delayed_work_sync(&vinst->idle_work);
+
+	if ((adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) ||
+	    (vinst->cur_state != AMD_PG_STATE_GATE &&
+	     RREG32_SOC15(VCN, 0, mmUVD_STATUS)))
+		vinst->set_pg_state(vinst, AMD_PG_STATE_GATE);
+
+	return 0;
+}
+
+/**
+ * vcn_v2_0_suspend - suspend VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * HW fini and suspend VCN block
+ */
+static int vcn_v2_0_suspend(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	r = vcn_v2_0_hw_fini(ip_block);
+	if (r)
+		return r;
+
+	r = amdgpu_vcn_suspend(ip_block->adev, 0);
+
+	return r;
+}
+
+/**
+ * vcn_v2_0_resume - resume VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Resume firmware and hw init VCN block
+ */
+static int vcn_v2_0_resume(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+
+	r = amdgpu_vcn_resume(ip_block->adev, 0);
+	if (r)
+		return r;
+
+	r = vcn_v2_0_hw_init(ip_block);
+
+	return r;
+}
+
+/**
+ * vcn_v2_0_mc_resume - memory controller programming
+ *
+ * @vinst: Pointer to the VCN instance structure
+ *
+ * Let the VCN memory controller know it's offsets
+ */
+static void vcn_v2_0_mc_resume(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	uint32_t size = AMDGPU_GPU_PAGE_ALIGN(adev->vcn.inst[0].fw->size + 4);
+	uint32_t offset;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/* cache window 0: fw */
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN].tmr_mc_addr_lo));
+		WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN].tmr_mc_addr_hi));
+		WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_OFFSET0, 0);
+		offset = 0;
+	} else {
+		WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			lower_32_bits(adev->vcn.inst->gpu_addr));
+		WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			upper_32_bits(adev->vcn.inst->gpu_addr));
+		offset = size;
+		WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_OFFSET0,
+			AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
+	}
+
+	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_SIZE0, size);
+
+	/* cache window 1: stack */
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW,
+		lower_32_bits(adev->vcn.inst->gpu_addr + offset));
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH,
+		upper_32_bits(adev->vcn.inst->gpu_addr + offset));
+	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_OFFSET1, 0);
+	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_SIZE1, AMDGPU_VCN_STACK_SIZE);
+
+	/* cache window 2: context */
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW,
+		lower_32_bits(adev->vcn.inst->gpu_addr + offset + AMDGPU_VCN_STACK_SIZE));
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH,
+		upper_32_bits(adev->vcn.inst->gpu_addr + offset + AMDGPU_VCN_STACK_SIZE));
+	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_OFFSET2, 0);
+	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_SIZE2, AMDGPU_VCN_CONTEXT_SIZE);
+
+	/* non-cache window */
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_LOW,
+		lower_32_bits(adev->vcn.inst->fw_shared.gpu_addr));
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH,
+		upper_32_bits(adev->vcn.inst->fw_shared.gpu_addr));
+	WREG32_SOC15(UVD, 0, mmUVD_VCPU_NONCACHE_OFFSET0, 0);
+	WREG32_SOC15(UVD, 0, mmUVD_VCPU_NONCACHE_SIZE0,
+		AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_fw_shared)));
+
+	WREG32_SOC15(UVD, 0, mmUVD_GFX10_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+}
+
+static void vcn_v2_0_mc_resume_dpg_mode(struct amdgpu_vcn_inst *vinst,
+					bool indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	uint32_t size = AMDGPU_GPU_PAGE_ALIGN(adev->vcn.inst[0].fw->size + 4);
+	uint32_t offset;
+
+	/* cache window 0: fw */
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		if (!indirect) {
+			WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+				UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+				(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN].tmr_mc_addr_lo), 0, indirect);
+			WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+				UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+				(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN].tmr_mc_addr_hi), 0, indirect);
+			WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+				UVD, 0, mmUVD_VCPU_CACHE_OFFSET0), 0, 0, indirect);
+		} else {
+			WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+				UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW), 0, 0, indirect);
+			WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+				UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH), 0, 0, indirect);
+			WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+				UVD, 0, mmUVD_VCPU_CACHE_OFFSET0), 0, 0, indirect);
+		}
+		offset = 0;
+	} else {
+		WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+			UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst->gpu_addr), 0, indirect);
+		WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+			UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst->gpu_addr), 0, indirect);
+		offset = size;
+		WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+			UVD, 0, mmUVD_VCPU_CACHE_OFFSET0),
+			AMDGPU_UVD_FIRMWARE_OFFSET >> 3, 0, indirect);
+	}
+
+	if (!indirect)
+		WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+			UVD, 0, mmUVD_VCPU_CACHE_SIZE0), size, 0, indirect);
+	else
+		WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+			UVD, 0, mmUVD_VCPU_CACHE_SIZE0), 0, 0, indirect);
+
+	/* cache window 1: stack */
+	if (!indirect) {
+		WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+			UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst->gpu_addr + offset), 0, indirect);
+		WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+			UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst->gpu_addr + offset), 0, indirect);
+		WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+			UVD, 0, mmUVD_VCPU_CACHE_OFFSET1), 0, 0, indirect);
+	} else {
+		WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+			UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW), 0, 0, indirect);
+		WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+			UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH), 0, 0, indirect);
+		WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+			UVD, 0, mmUVD_VCPU_CACHE_OFFSET1), 0, 0, indirect);
+	}
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_VCPU_CACHE_SIZE1), AMDGPU_VCN_STACK_SIZE, 0, indirect);
+
+	/* cache window 2: context */
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW),
+		lower_32_bits(adev->vcn.inst->gpu_addr + offset + AMDGPU_VCN_STACK_SIZE), 0, indirect);
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH),
+		upper_32_bits(adev->vcn.inst->gpu_addr + offset + AMDGPU_VCN_STACK_SIZE), 0, indirect);
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_VCPU_CACHE_OFFSET2), 0, 0, indirect);
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_VCPU_CACHE_SIZE2), AMDGPU_VCN_CONTEXT_SIZE, 0, indirect);
+
+	/* non-cache window */
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_LOW),
+		lower_32_bits(adev->vcn.inst->fw_shared.gpu_addr), 0, indirect);
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH),
+		upper_32_bits(adev->vcn.inst->fw_shared.gpu_addr), 0, indirect);
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_VCPU_NONCACHE_OFFSET0), 0, 0, indirect);
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_VCPU_NONCACHE_SIZE0),
+		AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_fw_shared)), 0, indirect);
+
+	/* VCN global tiling registers */
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_GFX10_ADDR_CONFIG), adev->gfx.config.gb_addr_config, 0, indirect);
+}
+
+/**
+ * vcn_v2_0_disable_clock_gating - disable VCN clock gating
+ *
+ * @vinst: VCN instance
+ *
+ * Disable clock gating for VCN block
+ */
+static void vcn_v2_0_disable_clock_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	uint32_t data;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/* UVD disable CGC */
+	data = RREG32_SOC15(VCN, 0, mmUVD_CGC_CTRL);
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		data |= 1 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		data &= ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK;
+	data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(VCN, 0, mmUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, 0, mmUVD_CGC_GATE);
+	data &= ~(UVD_CGC_GATE__SYS_MASK
+		| UVD_CGC_GATE__UDEC_MASK
+		| UVD_CGC_GATE__MPEG2_MASK
+		| UVD_CGC_GATE__REGS_MASK
+		| UVD_CGC_GATE__RBC_MASK
+		| UVD_CGC_GATE__LMI_MC_MASK
+		| UVD_CGC_GATE__LMI_UMC_MASK
+		| UVD_CGC_GATE__IDCT_MASK
+		| UVD_CGC_GATE__MPRD_MASK
+		| UVD_CGC_GATE__MPC_MASK
+		| UVD_CGC_GATE__LBSI_MASK
+		| UVD_CGC_GATE__LRBBM_MASK
+		| UVD_CGC_GATE__UDEC_RE_MASK
+		| UVD_CGC_GATE__UDEC_CM_MASK
+		| UVD_CGC_GATE__UDEC_IT_MASK
+		| UVD_CGC_GATE__UDEC_DB_MASK
+		| UVD_CGC_GATE__UDEC_MP_MASK
+		| UVD_CGC_GATE__WCB_MASK
+		| UVD_CGC_GATE__VCPU_MASK
+		| UVD_CGC_GATE__SCPU_MASK);
+	WREG32_SOC15(VCN, 0, mmUVD_CGC_GATE, data);
+
+	data = RREG32_SOC15(VCN, 0, mmUVD_CGC_CTRL);
+	data &= ~(UVD_CGC_CTRL__UDEC_RE_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_CM_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_IT_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_DB_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_MP_MODE_MASK
+		| UVD_CGC_CTRL__SYS_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_MODE_MASK
+		| UVD_CGC_CTRL__MPEG2_MODE_MASK
+		| UVD_CGC_CTRL__REGS_MODE_MASK
+		| UVD_CGC_CTRL__RBC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_MC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_UMC_MODE_MASK
+		| UVD_CGC_CTRL__IDCT_MODE_MASK
+		| UVD_CGC_CTRL__MPRD_MODE_MASK
+		| UVD_CGC_CTRL__MPC_MODE_MASK
+		| UVD_CGC_CTRL__LBSI_MODE_MASK
+		| UVD_CGC_CTRL__LRBBM_MODE_MASK
+		| UVD_CGC_CTRL__WCB_MODE_MASK
+		| UVD_CGC_CTRL__VCPU_MODE_MASK
+		| UVD_CGC_CTRL__SCPU_MODE_MASK);
+	WREG32_SOC15(VCN, 0, mmUVD_CGC_CTRL, data);
+
+	/* turn on */
+	data = RREG32_SOC15(VCN, 0, mmUVD_SUVD_CGC_GATE);
+	data |= (UVD_SUVD_CGC_GATE__SRE_MASK
+		| UVD_SUVD_CGC_GATE__SIT_MASK
+		| UVD_SUVD_CGC_GATE__SMP_MASK
+		| UVD_SUVD_CGC_GATE__SCM_MASK
+		| UVD_SUVD_CGC_GATE__SDB_MASK
+		| UVD_SUVD_CGC_GATE__SRE_H264_MASK
+		| UVD_SUVD_CGC_GATE__SRE_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SIT_H264_MASK
+		| UVD_SUVD_CGC_GATE__SIT_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SCM_H264_MASK
+		| UVD_SUVD_CGC_GATE__SCM_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SDB_H264_MASK
+		| UVD_SUVD_CGC_GATE__SDB_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SCLR_MASK
+		| UVD_SUVD_CGC_GATE__UVD_SC_MASK
+		| UVD_SUVD_CGC_GATE__ENT_MASK
+		| UVD_SUVD_CGC_GATE__SIT_HEVC_DEC_MASK
+		| UVD_SUVD_CGC_GATE__SIT_HEVC_ENC_MASK
+		| UVD_SUVD_CGC_GATE__SITE_MASK
+		| UVD_SUVD_CGC_GATE__SRE_VP9_MASK
+		| UVD_SUVD_CGC_GATE__SCM_VP9_MASK
+		| UVD_SUVD_CGC_GATE__SIT_VP9_DEC_MASK
+		| UVD_SUVD_CGC_GATE__SDB_VP9_MASK
+		| UVD_SUVD_CGC_GATE__IME_HEVC_MASK);
+	WREG32_SOC15(VCN, 0, mmUVD_SUVD_CGC_GATE, data);
+
+	data = RREG32_SOC15(VCN, 0, mmUVD_SUVD_CGC_CTRL);
+	data &= ~(UVD_SUVD_CGC_CTRL__SRE_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SIT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SMP_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SCM_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SDB_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SCLR_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__UVD_SC_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__ENT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__IME_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SITE_MODE_MASK);
+	WREG32_SOC15(VCN, 0, mmUVD_SUVD_CGC_CTRL, data);
+}
+
+static void vcn_v2_0_clock_gating_dpg_mode(struct amdgpu_vcn_inst *vinst,
+		uint8_t sram_sel, uint8_t indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	uint32_t reg_data = 0;
+
+	/* enable sw clock gating control */
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		reg_data = 1 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		reg_data = 0 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	reg_data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	reg_data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	reg_data &= ~(UVD_CGC_CTRL__UDEC_RE_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_CM_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_IT_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_DB_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_MP_MODE_MASK |
+		 UVD_CGC_CTRL__SYS_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_MODE_MASK |
+		 UVD_CGC_CTRL__MPEG2_MODE_MASK |
+		 UVD_CGC_CTRL__REGS_MODE_MASK |
+		 UVD_CGC_CTRL__RBC_MODE_MASK |
+		 UVD_CGC_CTRL__LMI_MC_MODE_MASK |
+		 UVD_CGC_CTRL__LMI_UMC_MODE_MASK |
+		 UVD_CGC_CTRL__IDCT_MODE_MASK |
+		 UVD_CGC_CTRL__MPRD_MODE_MASK |
+		 UVD_CGC_CTRL__MPC_MODE_MASK |
+		 UVD_CGC_CTRL__LBSI_MODE_MASK |
+		 UVD_CGC_CTRL__LRBBM_MODE_MASK |
+		 UVD_CGC_CTRL__WCB_MODE_MASK |
+		 UVD_CGC_CTRL__VCPU_MODE_MASK |
+		 UVD_CGC_CTRL__SCPU_MODE_MASK);
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_CGC_CTRL), reg_data, sram_sel, indirect);
+
+	/* turn off clock gating */
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_CGC_GATE), 0, sram_sel, indirect);
+
+	/* turn on SUVD clock gating */
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_SUVD_CGC_GATE), 1, sram_sel, indirect);
+
+	/* turn on sw mode in UVD_SUVD_CGC_CTRL */
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_SUVD_CGC_CTRL), 0, sram_sel, indirect);
+}
+
+/**
+ * vcn_v2_0_enable_clock_gating - enable VCN clock gating
+ *
+ * @vinst: VCN instance
+ *
+ * Enable clock gating for VCN block
+ */
+static void vcn_v2_0_enable_clock_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	uint32_t data = 0;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/* enable UVD CGC */
+	data = RREG32_SOC15(VCN, 0, mmUVD_CGC_CTRL);
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		data |= 1 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		data |= 0 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(VCN, 0, mmUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, 0, mmUVD_CGC_CTRL);
+	data |= (UVD_CGC_CTRL__UDEC_RE_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_CM_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_IT_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_DB_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_MP_MODE_MASK
+		| UVD_CGC_CTRL__SYS_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_MODE_MASK
+		| UVD_CGC_CTRL__MPEG2_MODE_MASK
+		| UVD_CGC_CTRL__REGS_MODE_MASK
+		| UVD_CGC_CTRL__RBC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_MC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_UMC_MODE_MASK
+		| UVD_CGC_CTRL__IDCT_MODE_MASK
+		| UVD_CGC_CTRL__MPRD_MODE_MASK
+		| UVD_CGC_CTRL__MPC_MODE_MASK
+		| UVD_CGC_CTRL__LBSI_MODE_MASK
+		| UVD_CGC_CTRL__LRBBM_MODE_MASK
+		| UVD_CGC_CTRL__WCB_MODE_MASK
+		| UVD_CGC_CTRL__VCPU_MODE_MASK
+		| UVD_CGC_CTRL__SCPU_MODE_MASK);
+	WREG32_SOC15(VCN, 0, mmUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, 0, mmUVD_SUVD_CGC_CTRL);
+	data |= (UVD_SUVD_CGC_CTRL__SRE_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SIT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SMP_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SCM_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SDB_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SCLR_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__UVD_SC_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__ENT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__IME_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SITE_MODE_MASK);
+	WREG32_SOC15(VCN, 0, mmUVD_SUVD_CGC_CTRL, data);
+}
+
+static void vcn_v2_0_disable_static_power_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	uint32_t data = 0;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN) {
+		data = (1 << UVD_PGFSM_CONFIG__UVDM_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDU_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDF_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDC_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDB_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDIL_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDIR_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTD_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTE_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDE_PWR_CONFIG__SHIFT);
+
+		WREG32_SOC15(VCN, 0, mmUVD_PGFSM_CONFIG, data);
+		SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_PGFSM_STATUS,
+			UVD_PGFSM_STATUS__UVDM_UVDU_PWR_ON_2_0, 0xFFFFF);
+	} else {
+		data = (1 << UVD_PGFSM_CONFIG__UVDM_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDU_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDF_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDC_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDB_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDIL_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDIR_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDTD_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDTE_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDE_PWR_CONFIG__SHIFT);
+		WREG32_SOC15(VCN, 0, mmUVD_PGFSM_CONFIG, data);
+		SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_PGFSM_STATUS, 0,  0xFFFFF);
+	}
+
+	/* polling UVD_PGFSM_STATUS to confirm UVDM_PWR_STATUS,
+	 * UVDU_PWR_STATUS are 0 (power on) */
+
+	data = RREG32_SOC15(VCN, 0, mmUVD_POWER_STATUS);
+	data &= ~0x103;
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN)
+		data |= UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON |
+			UVD_POWER_STATUS__UVD_PG_EN_MASK;
+
+	WREG32_SOC15(VCN, 0, mmUVD_POWER_STATUS, data);
+}
+
+static void vcn_v2_0_enable_static_power_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	uint32_t data = 0;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN) {
+		/* Before power off, this indicator has to be turned on */
+		data = RREG32_SOC15(VCN, 0, mmUVD_POWER_STATUS);
+		data &= ~UVD_POWER_STATUS__UVD_POWER_STATUS_MASK;
+		data |= UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF;
+		WREG32_SOC15(VCN, 0, mmUVD_POWER_STATUS, data);
+
+
+		data = (2 << UVD_PGFSM_CONFIG__UVDM_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDU_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDF_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDC_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDB_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDIL_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDIR_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTD_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTE_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDE_PWR_CONFIG__SHIFT);
+
+		WREG32_SOC15(VCN, 0, mmUVD_PGFSM_CONFIG, data);
+
+		data = (2 << UVD_PGFSM_STATUS__UVDM_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDU_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDF_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDC_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDB_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDIL_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDIR_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDTD_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDTE_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDE_PWR_STATUS__SHIFT);
+		SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_PGFSM_STATUS, data, 0xFFFFF);
+	}
+}
+
+static int vcn_v2_0_start_dpg_mode(struct amdgpu_vcn_inst *vinst, bool indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared.cpu_addr;
+	struct amdgpu_ring *ring = &adev->vcn.inst->ring_dec;
+	uint32_t rb_bufsz, tmp;
+	int ret;
+
+	vcn_v2_0_enable_static_power_gating(vinst);
+
+	/* enable dynamic power gating mode */
+	tmp = RREG32_SOC15(UVD, 0, mmUVD_POWER_STATUS);
+	tmp |= UVD_POWER_STATUS__UVD_PG_MODE_MASK;
+	tmp |= UVD_POWER_STATUS__UVD_PG_EN_MASK;
+	WREG32_SOC15(UVD, 0, mmUVD_POWER_STATUS, tmp);
+
+	if (indirect)
+		adev->vcn.inst->dpg_sram_curr_addr = (uint32_t *)adev->vcn.inst->dpg_sram_cpu_addr;
+
+	/* enable clock gating */
+	vcn_v2_0_clock_gating_dpg_mode(vinst, 0, indirect);
+
+	/* enable VCPU clock */
+	tmp = (0xFF << UVD_VCPU_CNTL__PRB_TIMEOUT_VAL__SHIFT);
+	tmp |= UVD_VCPU_CNTL__CLK_EN_MASK;
+	tmp |= UVD_VCPU_CNTL__MIF_WR_LOW_THRESHOLD_BP_MASK;
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_VCPU_CNTL), tmp, 0, indirect);
+
+	/* disable master interupt */
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_MASTINT_EN), 0, 0, indirect);
+
+	/* setup mmUVD_LMI_CTRL */
+	tmp = (UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+		UVD_LMI_CTRL__REQ_MODE_MASK |
+		UVD_LMI_CTRL__CRC_RESET_MASK |
+		UVD_LMI_CTRL__MASK_MC_URGENT_MASK |
+		UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+		UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
+		(8 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
+		0x00100000L);
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_LMI_CTRL), tmp, 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_MPC_CNTL),
+		0x2 << UVD_MPC_CNTL__REPLACEMENT_MODE__SHIFT, 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_MPC_SET_MUXA0),
+		((0x1 << UVD_MPC_SET_MUXA0__VARA_1__SHIFT) |
+		 (0x2 << UVD_MPC_SET_MUXA0__VARA_2__SHIFT) |
+		 (0x3 << UVD_MPC_SET_MUXA0__VARA_3__SHIFT) |
+		 (0x4 << UVD_MPC_SET_MUXA0__VARA_4__SHIFT)), 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_MPC_SET_MUXB0),
+		((0x1 << UVD_MPC_SET_MUXB0__VARB_1__SHIFT) |
+		 (0x2 << UVD_MPC_SET_MUXB0__VARB_2__SHIFT) |
+		 (0x3 << UVD_MPC_SET_MUXB0__VARB_3__SHIFT) |
+		 (0x4 << UVD_MPC_SET_MUXB0__VARB_4__SHIFT)), 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_MPC_SET_MUX),
+		((0x0 << UVD_MPC_SET_MUX__SET_0__SHIFT) |
+		 (0x1 << UVD_MPC_SET_MUX__SET_1__SHIFT) |
+		 (0x2 << UVD_MPC_SET_MUX__SET_2__SHIFT)), 0, indirect);
+
+	vcn_v2_0_mc_resume_dpg_mode(vinst, indirect);
+
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_REG_XX_MASK), 0x10, 0, indirect);
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_RBC_XX_IB_REG_CHECK), 0x3, 0, indirect);
+
+	/* release VCPU reset to boot */
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_SOFT_RESET), 0, 0, indirect);
+
+	/* enable LMI MC and UMC channels */
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_LMI_CTRL2),
+		0x1F << UVD_LMI_CTRL2__RE_OFLD_MIF_WR_REQ_NUM__SHIFT, 0, indirect);
+
+	/* enable master interrupt */
+	WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+		UVD, 0, mmUVD_MASTINT_EN),
+		UVD_MASTINT_EN__VCPU_EN_MASK, 0, indirect);
+
+	if (indirect) {
+		ret = amdgpu_vcn_psp_update_sram(adev, 0, 0);
+		if (ret) {
+			dev_err(adev->dev, "vcn sram load failed %d\n", ret);
+			return ret;
+		}
+	}
+
+	/* force RBC into idle state */
+	rb_bufsz = order_base_2(ring->ring_size);
+	tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_CNTL, tmp);
+
+	/* Stall DPG before WPTR/RPTR reset */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_POWER_STATUS),
+		UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK,
+		~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK);
+	fw_shared->multi_queue.decode_queue_mode |= FW_QUEUE_RING_RESET;
+
+	/* set the write pointer delay */
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR_CNTL, 0);
+
+	/* set the wb address */
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR_ADDR,
+		(upper_32_bits(ring->gpu_addr) >> 2));
+
+	/* program the RB_BASE for ring buffer */
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW,
+		lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH,
+		upper_32_bits(ring->gpu_addr));
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR, 0);
+
+	WREG32_SOC15(UVD, 0, mmUVD_SCRATCH2, 0);
+
+	ring->wptr = RREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR);
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR,
+		lower_32_bits(ring->wptr));
+
+	fw_shared->multi_queue.decode_queue_mode &= ~FW_QUEUE_RING_RESET;
+	/* Unstall DPG */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_POWER_STATUS),
+		0, ~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(UVD, 0, mmUVD_STATUS);
+
+	return 0;
+}
+
+static int vcn_v2_0_start(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared.cpu_addr;
+	struct amdgpu_ring *ring = &adev->vcn.inst->ring_dec;
+	uint32_t rb_bufsz, tmp;
+	uint32_t lmi_swap_cntl;
+	int i, j, r;
+
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_vcn(adev, true, 0);
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)
+		return vcn_v2_0_start_dpg_mode(vinst, adev->vcn.inst->indirect_sram);
+
+	vcn_v2_0_disable_static_power_gating(vinst);
+
+	/* set uvd status busy */
+	tmp = RREG32_SOC15(UVD, 0, mmUVD_STATUS) | UVD_STATUS__UVD_BUSY;
+	WREG32_SOC15(UVD, 0, mmUVD_STATUS, tmp);
+
+	/*SW clock gating */
+	vcn_v2_0_disable_clock_gating(vinst);
+
+	/* enable VCPU clock */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CNTL),
+		UVD_VCPU_CNTL__CLK_EN_MASK, ~UVD_VCPU_CNTL__CLK_EN_MASK);
+
+	/* disable master interrupt */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_MASTINT_EN), 0,
+		~UVD_MASTINT_EN__VCPU_EN_MASK);
+
+	/* setup mmUVD_LMI_CTRL */
+	tmp = RREG32_SOC15(UVD, 0, mmUVD_LMI_CTRL);
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_CTRL, tmp |
+		UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK	|
+		UVD_LMI_CTRL__MASK_MC_URGENT_MASK |
+		UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+		UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK);
+
+	/* setup mmUVD_MPC_CNTL */
+	tmp = RREG32_SOC15(UVD, 0, mmUVD_MPC_CNTL);
+	tmp &= ~UVD_MPC_CNTL__REPLACEMENT_MODE_MASK;
+	tmp |= 0x2 << UVD_MPC_CNTL__REPLACEMENT_MODE__SHIFT;
+	WREG32_SOC15(VCN, 0, mmUVD_MPC_CNTL, tmp);
+
+	/* setup UVD_MPC_SET_MUXA0 */
+	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUXA0,
+		((0x1 << UVD_MPC_SET_MUXA0__VARA_1__SHIFT) |
+		(0x2 << UVD_MPC_SET_MUXA0__VARA_2__SHIFT) |
+		(0x3 << UVD_MPC_SET_MUXA0__VARA_3__SHIFT) |
+		(0x4 << UVD_MPC_SET_MUXA0__VARA_4__SHIFT)));
+
+	/* setup UVD_MPC_SET_MUXB0 */
+	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUXB0,
+		((0x1 << UVD_MPC_SET_MUXB0__VARB_1__SHIFT) |
+		(0x2 << UVD_MPC_SET_MUXB0__VARB_2__SHIFT) |
+		(0x3 << UVD_MPC_SET_MUXB0__VARB_3__SHIFT) |
+		(0x4 << UVD_MPC_SET_MUXB0__VARB_4__SHIFT)));
+
+	/* setup mmUVD_MPC_SET_MUX */
+	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUX,
+		((0x0 << UVD_MPC_SET_MUX__SET_0__SHIFT) |
+		(0x1 << UVD_MPC_SET_MUX__SET_1__SHIFT) |
+		(0x2 << UVD_MPC_SET_MUX__SET_2__SHIFT)));
+
+	vcn_v2_0_mc_resume(vinst);
+
+	/* release VCPU reset to boot */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET), 0,
+		~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+
+	/* enable LMI MC and UMC channels */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2), 0,
+		~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+
+	tmp = RREG32_SOC15(VCN, 0, mmUVD_SOFT_RESET);
+	tmp &= ~UVD_SOFT_RESET__LMI_SOFT_RESET_MASK;
+	tmp &= ~UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK;
+	WREG32_SOC15(VCN, 0, mmUVD_SOFT_RESET, tmp);
+
+	/* disable byte swapping */
+	lmi_swap_cntl = 0;
+#ifdef __BIG_ENDIAN
+	/* swap (8 in 32) RB and IB */
+	lmi_swap_cntl = 0xa;
+#endif
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_SWAP_CNTL, lmi_swap_cntl);
+
+	for (i = 0; i < 10; ++i) {
+		uint32_t status;
+
+		for (j = 0; j < 100; ++j) {
+			status = RREG32_SOC15(UVD, 0, mmUVD_STATUS);
+			if (status & 2)
+				break;
+			mdelay(10);
+		}
+		r = 0;
+		if (status & 2)
+			break;
+
+		DRM_ERROR("VCN decode not responding, trying to reset the VCPU!!!\n");
+		WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET),
+			UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK,
+			~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+		mdelay(10);
+		WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET), 0,
+			~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+		mdelay(10);
+		r = -1;
+	}
+
+	if (r) {
+		DRM_ERROR("VCN decode not responding, giving up!!!\n");
+		return r;
+	}
+
+	/* enable master interrupt */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_MASTINT_EN),
+		UVD_MASTINT_EN__VCPU_EN_MASK,
+		~UVD_MASTINT_EN__VCPU_EN_MASK);
+
+	/* clear the busy bit of VCN_STATUS */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_STATUS), 0,
+		~(2 << UVD_STATUS__VCPU_REPORT__SHIFT));
+
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_VMID, 0);
+
+	/* force RBC into idle state */
+	rb_bufsz = order_base_2(ring->ring_size);
+	tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_CNTL, tmp);
+
+	fw_shared->multi_queue.decode_queue_mode |= FW_QUEUE_RING_RESET;
+	/* program the RB_BASE for ring buffer */
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW,
+		lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH,
+		upper_32_bits(ring->gpu_addr));
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR, 0);
+
+	ring->wptr = RREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR);
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR,
+			lower_32_bits(ring->wptr));
+	fw_shared->multi_queue.decode_queue_mode &= ~FW_QUEUE_RING_RESET;
+
+	fw_shared->multi_queue.encode_generalpurpose_queue_mode |= FW_QUEUE_RING_RESET;
+	ring = &adev->vcn.inst->ring_enc[0];
+	WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR, lower_32_bits(ring->wptr));
+	WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR, lower_32_bits(ring->wptr));
+	WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO, ring->gpu_addr);
+	WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE, ring->ring_size / 4);
+	fw_shared->multi_queue.encode_generalpurpose_queue_mode &= ~FW_QUEUE_RING_RESET;
+
+	fw_shared->multi_queue.encode_lowlatency_queue_mode |= FW_QUEUE_RING_RESET;
+	ring = &adev->vcn.inst->ring_enc[1];
+	WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR2, lower_32_bits(ring->wptr));
+	WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
+	WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO2, ring->gpu_addr);
+	WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE2, ring->ring_size / 4);
+	fw_shared->multi_queue.encode_lowlatency_queue_mode &= ~FW_QUEUE_RING_RESET;
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(UVD, 0, mmUVD_STATUS);
+
+	return 0;
+}
+
+static int vcn_v2_0_stop_dpg_mode(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	struct dpg_pause_state state = {.fw_based = VCN_DPG_STATE__UNPAUSE};
+	uint32_t tmp;
+
+	vcn_v2_0_pause_dpg_mode(vinst, &state);
+	/* Wait for power status to be 1 */
+	SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS, 1,
+		UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+	/* wait for read ptr to be equal to write ptr */
+	tmp = RREG32_SOC15(UVD, 0, mmUVD_RB_WPTR);
+	SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_RB_RPTR, tmp, 0xFFFFFFFF);
+
+	tmp = RREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2);
+	SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_RB_RPTR2, tmp, 0xFFFFFFFF);
+
+	tmp = RREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR) & 0x7FFFFFFF;
+	SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_RBC_RB_RPTR, tmp, 0xFFFFFFFF);
+
+	SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS, 1,
+		UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+	/* disable dynamic power gating mode */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_POWER_STATUS), 0,
+			~UVD_POWER_STATUS__UVD_PG_MODE_MASK);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(UVD, 0, mmUVD_STATUS);
+
+	return 0;
+}
+
+static int vcn_v2_0_stop(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	uint32_t tmp;
+	int r;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) {
+		r = vcn_v2_0_stop_dpg_mode(vinst);
+		if (r)
+			return r;
+		goto power_off;
+	}
+
+	/* wait for uvd idle */
+	r = SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_STATUS, UVD_STATUS__IDLE, 0x7);
+	if (r)
+		return r;
+
+	tmp = UVD_LMI_STATUS__VCPU_LMI_WRITE_CLEAN_MASK |
+		UVD_LMI_STATUS__READ_CLEAN_MASK |
+		UVD_LMI_STATUS__WRITE_CLEAN_MASK |
+		UVD_LMI_STATUS__WRITE_CLEAN_RAW_MASK;
+	r = SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_LMI_STATUS, tmp, tmp);
+	if (r)
+		return r;
+
+	/* stall UMC channel */
+	tmp = RREG32_SOC15(VCN, 0, mmUVD_LMI_CTRL2);
+	tmp |= UVD_LMI_CTRL2__STALL_ARB_UMC_MASK;
+	WREG32_SOC15(VCN, 0, mmUVD_LMI_CTRL2, tmp);
+
+	tmp = UVD_LMI_STATUS__UMC_READ_CLEAN_RAW_MASK|
+		UVD_LMI_STATUS__UMC_WRITE_CLEAN_RAW_MASK;
+	r = SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_LMI_STATUS, tmp, tmp);
+	if (r)
+		return r;
+
+	/* disable VCPU clock */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CNTL), 0,
+		~(UVD_VCPU_CNTL__CLK_EN_MASK));
+
+	/* reset LMI UMC */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET),
+		UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK,
+		~UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK);
+
+	/* reset LMI */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET),
+		UVD_SOFT_RESET__LMI_SOFT_RESET_MASK,
+		~UVD_SOFT_RESET__LMI_SOFT_RESET_MASK);
+
+	/* reset VCPU */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET),
+		UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK,
+		~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+
+	/* clear status */
+	WREG32_SOC15(VCN, 0, mmUVD_STATUS, 0);
+
+	vcn_v2_0_enable_clock_gating(vinst);
+	vcn_v2_0_enable_static_power_gating(vinst);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, 0, mmUVD_STATUS);
+
+power_off:
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_vcn(adev, false, 0);
+
+	return 0;
+}
+
+static int vcn_v2_0_pause_dpg_mode(struct amdgpu_vcn_inst *vinst,
+				   struct dpg_pause_state *new_state)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	struct amdgpu_ring *ring;
+	uint32_t reg_data = 0;
+	int ret_code;
+
+	/* pause/unpause if state is changed */
+	if (adev->vcn.inst[inst_idx].pause_state.fw_based != new_state->fw_based) {
+		DRM_DEBUG("dpg pause state changed %d -> %d",
+			adev->vcn.inst[inst_idx].pause_state.fw_based,	new_state->fw_based);
+		reg_data = RREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE) &
+			(~UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK);
+
+		if (new_state->fw_based == VCN_DPG_STATE__PAUSE) {
+			ret_code = SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS, 0x1,
+				UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+			if (!ret_code) {
+				struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared.cpu_addr;
+				/* pause DPG */
+				reg_data |= UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
+				WREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE, reg_data);
+
+				/* wait for ACK */
+				SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_DPG_PAUSE,
+					   UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK,
+					   UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK);
+
+				/* Stall DPG before WPTR/RPTR reset */
+				WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_POWER_STATUS),
+					   UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK,
+					   ~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK);
+				/* Restore */
+				fw_shared->multi_queue.encode_generalpurpose_queue_mode |= FW_QUEUE_RING_RESET;
+				ring = &adev->vcn.inst->ring_enc[0];
+				ring->wptr = 0;
+				WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO, ring->gpu_addr);
+				WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+				WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE, ring->ring_size / 4);
+				WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR, lower_32_bits(ring->wptr));
+				WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR, lower_32_bits(ring->wptr));
+				fw_shared->multi_queue.encode_generalpurpose_queue_mode &= ~FW_QUEUE_RING_RESET;
+
+				fw_shared->multi_queue.encode_lowlatency_queue_mode |= FW_QUEUE_RING_RESET;
+				ring = &adev->vcn.inst->ring_enc[1];
+				ring->wptr = 0;
+				WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO2, ring->gpu_addr);
+				WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
+				WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE2, ring->ring_size / 4);
+				WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR2, lower_32_bits(ring->wptr));
+				WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
+				fw_shared->multi_queue.encode_lowlatency_queue_mode &= ~FW_QUEUE_RING_RESET;
+
+				fw_shared->multi_queue.decode_queue_mode |= FW_QUEUE_RING_RESET;
+				WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR,
+					   RREG32_SOC15(UVD, 0, mmUVD_SCRATCH2) & 0x7FFFFFFF);
+				fw_shared->multi_queue.decode_queue_mode &= ~FW_QUEUE_RING_RESET;
+				/* Unstall DPG */
+				WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_POWER_STATUS),
+					   0, ~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK);
+
+				SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS,
+					   UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON,
+					   UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+			}
+		} else {
+			/* unpause dpg, no need to wait */
+			reg_data &= ~UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
+			WREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE, reg_data);
+		}
+		adev->vcn.inst[inst_idx].pause_state.fw_based = new_state->fw_based;
+	}
+
+	return 0;
+}
+
+static int vcn_v2_0_reset(struct amdgpu_vcn_inst *vinst)
+{
+	int r;
+
+	r = vcn_v2_0_stop(vinst);
+	if (r)
+		return r;
+	return vcn_v2_0_start(vinst);
+}
+
+static bool vcn_v2_0_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	return (RREG32_SOC15(VCN, 0, mmUVD_STATUS) == UVD_STATUS__IDLE);
+}
+
+static int vcn_v2_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int ret;
+
+	ret = SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_STATUS, UVD_STATUS__IDLE,
+		UVD_STATUS__IDLE);
+
+	return ret;
+}
+
+static int vcn_v2_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = (state == AMD_CG_STATE_GATE);
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	if (enable) {
+		/* wait for STATUS to clear */
+		if (!vcn_v2_0_is_idle(ip_block))
+			return -EBUSY;
+		vcn_v2_0_enable_clock_gating(&adev->vcn.inst[0]);
+	} else {
+		/* disable HW gating and enable Sw gating */
+		vcn_v2_0_disable_clock_gating(&adev->vcn.inst[0]);
+	}
+	return 0;
+}
+
+/**
+ * vcn_v2_0_dec_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t vcn_v2_0_dec_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR);
+}
+
+/**
+ * vcn_v2_0_dec_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t vcn_v2_0_dec_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell)
+		return *ring->wptr_cpu_addr;
+	else
+		return RREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR);
+}
+
+/**
+ * vcn_v2_0_dec_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void vcn_v2_0_dec_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)
+		WREG32_SOC15(UVD, 0, mmUVD_SCRATCH2,
+			lower_32_bits(ring->wptr) | 0x80000000);
+
+	if (ring->use_doorbell) {
+		*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
+		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+	} else {
+		WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr));
+	}
+}
+
+/**
+ * vcn_v2_0_dec_ring_insert_start - insert a start command
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Write a start command to the ring.
+ */
+void vcn_v2_0_dec_ring_insert_start(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[ring->me].internal.data0, 0));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[ring->me].internal.cmd, 0));
+	amdgpu_ring_write(ring, VCN_DEC_KMD_CMD | (VCN_DEC_CMD_PACKET_START << 1));
+}
+
+/**
+ * vcn_v2_0_dec_ring_insert_end - insert a end command
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Write a end command to the ring.
+ */
+void vcn_v2_0_dec_ring_insert_end(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[0].internal.cmd, 0));
+	amdgpu_ring_write(ring, VCN_DEC_KMD_CMD | (VCN_DEC_CMD_PACKET_END << 1));
+}
+
+/**
+ * vcn_v2_0_dec_ring_insert_nop - insert a nop command
+ *
+ * @ring: amdgpu_ring pointer
+ * @count: the number of NOP packets to insert
+ *
+ * Write a nop command to the ring.
+ */
+void vcn_v2_0_dec_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int i;
+
+	WARN_ON(ring->wptr % 2 || count % 2);
+
+	for (i = 0; i < count / 2; i++) {
+		amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[ring->me].internal.nop, 0));
+		amdgpu_ring_write(ring, 0);
+	}
+}
+
+/**
+ * vcn_v2_0_dec_ring_emit_fence - emit an fence & trap command
+ *
+ * @ring: amdgpu_ring pointer
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
+ *
+ * Write a fence and a trap command to the ring.
+ */
+void vcn_v2_0_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				unsigned flags)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+	amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[ring->me].internal.context_id, 0));
+	amdgpu_ring_write(ring, seq);
+
+	amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[ring->me].internal.data0, 0));
+	amdgpu_ring_write(ring, addr & 0xffffffff);
+
+	amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[ring->me].internal.data1, 0));
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xff);
+
+	amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[ring->me].internal.cmd, 0));
+	amdgpu_ring_write(ring, VCN_DEC_KMD_CMD | (VCN_DEC_CMD_FENCE << 1));
+
+	amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[ring->me].internal.data0, 0));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[ring->me].internal.data1, 0));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[ring->me].internal.cmd, 0));
+
+	amdgpu_ring_write(ring, VCN_DEC_KMD_CMD | (VCN_DEC_CMD_TRAP << 1));
+}
+
+/**
+ * vcn_v2_0_dec_ring_emit_ib - execute indirect buffer
+ *
+ * @ring: amdgpu_ring pointer
+ * @job: job to retrieve vmid from
+ * @ib: indirect buffer to execute
+ * @flags: unused
+ *
+ * Write ring commands to execute the indirect buffer
+ */
+void vcn_v2_0_dec_ring_emit_ib(struct amdgpu_ring *ring,
+			       struct amdgpu_job *job,
+			       struct amdgpu_ib *ib,
+			       uint32_t flags)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+
+	amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[ring->me].internal.ib_vmid, 0));
+	amdgpu_ring_write(ring, vmid);
+
+	amdgpu_ring_write(ring,	PACKET0(adev->vcn.inst[ring->me].internal.ib_bar_low, 0));
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring,	PACKET0(adev->vcn.inst[ring->me].internal.ib_bar_high, 0));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring,	PACKET0(adev->vcn.inst[ring->me].internal.ib_size, 0));
+	amdgpu_ring_write(ring, ib->length_dw);
+}
+
+void vcn_v2_0_dec_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+				uint32_t val, uint32_t mask)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[ring->me].internal.data0, 0));
+	amdgpu_ring_write(ring, reg << 2);
+
+	amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[ring->me].internal.data1, 0));
+	amdgpu_ring_write(ring, val);
+
+	amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[ring->me].internal.gp_scratch8, 0));
+	amdgpu_ring_write(ring, mask);
+
+	amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[ring->me].internal.cmd, 0));
+
+	amdgpu_ring_write(ring, VCN_DEC_KMD_CMD | (VCN_DEC_CMD_REG_READ_COND_WAIT << 1));
+}
+
+void vcn_v2_0_dec_ring_emit_vm_flush(struct amdgpu_ring *ring,
+				unsigned vmid, uint64_t pd_addr)
+{
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
+	uint32_t data0, data1, mask;
+
+	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* wait for register write */
+	data0 = hub->ctx0_ptb_addr_lo32 + vmid * hub->ctx_addr_distance;
+	data1 = lower_32_bits(pd_addr);
+	mask = 0xffffffff;
+	vcn_v2_0_dec_ring_emit_reg_wait(ring, data0, data1, mask);
+}
+
+void vcn_v2_0_dec_ring_emit_wreg(struct amdgpu_ring *ring,
+				uint32_t reg, uint32_t val)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[ring->me].internal.data0, 0));
+	amdgpu_ring_write(ring, reg << 2);
+
+	amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[ring->me].internal.data1, 0));
+	amdgpu_ring_write(ring, val);
+
+	amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[ring->me].internal.cmd, 0));
+
+	amdgpu_ring_write(ring, VCN_DEC_KMD_CMD | (VCN_DEC_CMD_WRITE_REG << 1));
+}
+
+/**
+ * vcn_v2_0_enc_ring_get_rptr - get enc read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware enc read pointer
+ */
+static uint64_t vcn_v2_0_enc_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->vcn.inst->ring_enc[0])
+		return RREG32_SOC15(UVD, 0, mmUVD_RB_RPTR);
+	else
+		return RREG32_SOC15(UVD, 0, mmUVD_RB_RPTR2);
+}
+
+ /**
+ * vcn_v2_0_enc_ring_get_wptr - get enc write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware enc write pointer
+ */
+static uint64_t vcn_v2_0_enc_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->vcn.inst->ring_enc[0]) {
+		if (ring->use_doorbell)
+			return *ring->wptr_cpu_addr;
+		else
+			return RREG32_SOC15(UVD, 0, mmUVD_RB_WPTR);
+	} else {
+		if (ring->use_doorbell)
+			return *ring->wptr_cpu_addr;
+		else
+			return RREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2);
+	}
+}
+
+ /**
+ * vcn_v2_0_enc_ring_set_wptr - set enc write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the enc write pointer to the hardware
+ */
+static void vcn_v2_0_enc_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->vcn.inst->ring_enc[0]) {
+		if (ring->use_doorbell) {
+			*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
+			WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+		} else {
+			WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR, lower_32_bits(ring->wptr));
+		}
+	} else {
+		if (ring->use_doorbell) {
+			*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
+			WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+		} else {
+			WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
+		}
+	}
+}
+
+/**
+ * vcn_v2_0_enc_ring_emit_fence - emit an enc fence & trap command
+ *
+ * @ring: amdgpu_ring pointer
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
+ *
+ * Write enc a fence and a trap command to the ring.
+ */
+void vcn_v2_0_enc_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
+				u64 seq, unsigned flags)
+{
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	amdgpu_ring_write(ring, VCN_ENC_CMD_FENCE);
+	amdgpu_ring_write(ring, addr);
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, VCN_ENC_CMD_TRAP);
+}
+
+void vcn_v2_0_enc_ring_insert_end(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, VCN_ENC_CMD_END);
+}
+
+/**
+ * vcn_v2_0_enc_ring_emit_ib - enc execute indirect buffer
+ *
+ * @ring: amdgpu_ring pointer
+ * @job: job to retrive vmid from
+ * @ib: indirect buffer to execute
+ * @flags: unused
+ *
+ * Write enc ring commands to execute the indirect buffer
+ */
+void vcn_v2_0_enc_ring_emit_ib(struct amdgpu_ring *ring,
+			       struct amdgpu_job *job,
+			       struct amdgpu_ib *ib,
+			       uint32_t flags)
+{
+	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
+
+	amdgpu_ring_write(ring, VCN_ENC_CMD_IB);
+	amdgpu_ring_write(ring, vmid);
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, ib->length_dw);
+}
+
+void vcn_v2_0_enc_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+				uint32_t val, uint32_t mask)
+{
+	amdgpu_ring_write(ring, VCN_ENC_CMD_REG_WAIT);
+	amdgpu_ring_write(ring, reg << 2);
+	amdgpu_ring_write(ring, mask);
+	amdgpu_ring_write(ring, val);
+}
+
+void vcn_v2_0_enc_ring_emit_vm_flush(struct amdgpu_ring *ring,
+				unsigned int vmid, uint64_t pd_addr)
+{
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
+
+	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* wait for reg writes */
+	vcn_v2_0_enc_ring_emit_reg_wait(ring, hub->ctx0_ptb_addr_lo32 +
+					vmid * hub->ctx_addr_distance,
+					lower_32_bits(pd_addr), 0xffffffff);
+}
+
+void vcn_v2_0_enc_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg, uint32_t val)
+{
+	amdgpu_ring_write(ring, VCN_ENC_CMD_REG_WRITE);
+	amdgpu_ring_write(ring,	reg << 2);
+	amdgpu_ring_write(ring, val);
+}
+
+static int vcn_v2_0_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+static int vcn_v2_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	DRM_DEBUG("IH: VCN TRAP\n");
+
+	switch (entry->src_id) {
+	case VCN_2_0__SRCID__UVD_SYSTEM_MESSAGE_INTERRUPT:
+		amdgpu_fence_process(&adev->vcn.inst->ring_dec);
+		break;
+	case VCN_2_0__SRCID__UVD_ENC_GENERAL_PURPOSE:
+		amdgpu_fence_process(&adev->vcn.inst->ring_enc[0]);
+		break;
+	case VCN_2_0__SRCID__UVD_ENC_LOW_LATENCY:
+		amdgpu_fence_process(&adev->vcn.inst->ring_enc[1]);
+		break;
+	default:
+		DRM_ERROR("Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+int vcn_v2_0_dec_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	WREG32(adev->vcn.inst[ring->me].external.scratch9, 0xCAFEDEAD);
+	r = amdgpu_ring_alloc(ring, 4);
+	if (r)
+		return r;
+	amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[ring->me].internal.cmd, 0));
+	amdgpu_ring_write(ring, VCN_DEC_KMD_CMD | (VCN_DEC_CMD_PACKET_START << 1));
+	amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[ring->me].internal.scratch9, 0));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(adev->vcn.inst[ring->me].external.scratch9);
+		if (tmp == 0xDEADBEEF)
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout)
+		r = -ETIMEDOUT;
+
+	return r;
+}
+
+
+static int vcn_v2_0_set_pg_state(struct amdgpu_vcn_inst *vinst,
+				 enum amd_powergating_state state)
+{
+	/* This doesn't actually powergate the VCN block.
+	 * That's done in the dpm code via the SMC.  This
+	 * just re-inits the block as necessary.  The actual
+	 * gating still happens in the dpm code.  We should
+	 * revisit this when there is a cleaner line between
+	 * the smc and the hw blocks
+	 */
+	int ret;
+	struct amdgpu_device *adev = vinst->adev;
+
+	if (amdgpu_sriov_vf(adev)) {
+		vinst->cur_state = AMD_PG_STATE_UNGATE;
+		return 0;
+	}
+
+	if (state == vinst->cur_state)
+		return 0;
+
+	if (state == AMD_PG_STATE_GATE)
+		ret = vcn_v2_0_stop(vinst);
+	else
+		ret = vcn_v2_0_start(vinst);
+
+	if (!ret)
+		vinst->cur_state = state;
+
+	return ret;
+}
+
+static int vcn_v2_0_start_mmsch(struct amdgpu_device *adev,
+				struct amdgpu_mm_table *table)
+{
+	uint32_t data = 0, loop;
+	uint64_t addr = table->gpu_addr;
+	struct mmsch_v2_0_init_header *header;
+	uint32_t size;
+	int i;
+
+	header = (struct mmsch_v2_0_init_header *)table->cpu_addr;
+	size = header->header_size + header->vcn_table_size;
+
+	/* 1, write to vce_mmsch_vf_ctx_addr_lo/hi register with GPU mc addr
+	 * of memory descriptor location
+	 */
+	WREG32_SOC15(UVD, 0, mmMMSCH_VF_CTX_ADDR_LO, lower_32_bits(addr));
+	WREG32_SOC15(UVD, 0, mmMMSCH_VF_CTX_ADDR_HI, upper_32_bits(addr));
+
+	/* 2, update vmid of descriptor */
+	data = RREG32_SOC15(UVD, 0, mmMMSCH_VF_VMID);
+	data &= ~MMSCH_VF_VMID__VF_CTX_VMID_MASK;
+	/* use domain0 for MM scheduler */
+	data |= (0 << MMSCH_VF_VMID__VF_CTX_VMID__SHIFT);
+	WREG32_SOC15(UVD, 0, mmMMSCH_VF_VMID, data);
+
+	/* 3, notify mmsch about the size of this descriptor */
+	WREG32_SOC15(UVD, 0, mmMMSCH_VF_CTX_SIZE, size);
+
+	/* 4, set resp to zero */
+	WREG32_SOC15(UVD, 0, mmMMSCH_VF_MAILBOX_RESP, 0);
+
+	adev->vcn.inst->ring_dec.wptr = 0;
+	adev->vcn.inst->ring_dec.wptr_old = 0;
+	vcn_v2_0_dec_ring_set_wptr(&adev->vcn.inst->ring_dec);
+
+	for (i = 0; i < adev->vcn.inst[0].num_enc_rings; ++i) {
+		adev->vcn.inst->ring_enc[i].wptr = 0;
+		adev->vcn.inst->ring_enc[i].wptr_old = 0;
+		vcn_v2_0_enc_ring_set_wptr(&adev->vcn.inst->ring_enc[i]);
+	}
+
+	/* 5, kick off the initialization and wait until
+	 * VCE_MMSCH_VF_MAILBOX_RESP becomes non-zero
+	 */
+	WREG32_SOC15(UVD, 0, mmMMSCH_VF_MAILBOX_HOST, 0x10000001);
+
+	data = RREG32_SOC15(UVD, 0, mmMMSCH_VF_MAILBOX_RESP);
+	loop = 1000;
+	while ((data & 0x10000002) != 0x10000002) {
+		udelay(10);
+		data = RREG32_SOC15(UVD, 0, mmMMSCH_VF_MAILBOX_RESP);
+		loop--;
+		if (!loop)
+			break;
+	}
+
+	if (!loop) {
+		DRM_ERROR("failed to init MMSCH, " \
+			"mmMMSCH_VF_MAILBOX_RESP = 0x%08x\n", data);
+		return -EBUSY;
+	}
+
+	return 0;
+}
+
+static int vcn_v2_0_start_sriov(struct amdgpu_device *adev)
+{
+	int r;
+	uint32_t tmp;
+	struct amdgpu_ring *ring;
+	uint32_t offset, size;
+	uint32_t table_size = 0;
+	struct mmsch_v2_0_cmd_direct_write direct_wt = { {0} };
+	struct mmsch_v2_0_cmd_direct_read_modify_write direct_rd_mod_wt = { {0} };
+	struct mmsch_v2_0_cmd_end end = { {0} };
+	struct mmsch_v2_0_init_header *header;
+	uint32_t *init_table = adev->virt.mm_table.cpu_addr;
+	uint8_t i = 0;
+
+	header = (struct mmsch_v2_0_init_header *)init_table;
+	direct_wt.cmd_header.command_type = MMSCH_COMMAND__DIRECT_REG_WRITE;
+	direct_rd_mod_wt.cmd_header.command_type =
+		MMSCH_COMMAND__DIRECT_REG_READ_MODIFY_WRITE;
+	end.cmd_header.command_type = MMSCH_COMMAND__END;
+
+	if (header->vcn_table_offset == 0 && header->vcn_table_size == 0) {
+		header->version = MMSCH_VERSION;
+		header->header_size = sizeof(struct mmsch_v2_0_init_header) >> 2;
+
+		header->vcn_table_offset = header->header_size;
+
+		init_table += header->vcn_table_offset;
+
+		size = AMDGPU_GPU_PAGE_ALIGN(adev->vcn.inst[0].fw->size + 4);
+
+		MMSCH_V2_0_INSERT_DIRECT_RD_MOD_WT(
+			SOC15_REG_OFFSET(UVD, i, mmUVD_STATUS),
+			0xFFFFFFFF, 0x00000004);
+
+		/* mc resume*/
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+			MMSCH_V2_0_INSERT_DIRECT_WT(
+				SOC15_REG_OFFSET(UVD, i,
+					mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+				adev->firmware.ucode[AMDGPU_UCODE_ID_VCN].tmr_mc_addr_lo);
+			MMSCH_V2_0_INSERT_DIRECT_WT(
+				SOC15_REG_OFFSET(UVD, i,
+					mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+				adev->firmware.ucode[AMDGPU_UCODE_ID_VCN].tmr_mc_addr_hi);
+			offset = 0;
+		} else {
+			MMSCH_V2_0_INSERT_DIRECT_WT(
+				SOC15_REG_OFFSET(UVD, i,
+					mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+				lower_32_bits(adev->vcn.inst->gpu_addr));
+			MMSCH_V2_0_INSERT_DIRECT_WT(
+				SOC15_REG_OFFSET(UVD, i,
+					mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+				upper_32_bits(adev->vcn.inst->gpu_addr));
+			offset = size;
+		}
+
+		MMSCH_V2_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_OFFSET0),
+			0);
+		MMSCH_V2_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_SIZE0),
+			size);
+
+		MMSCH_V2_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i,
+				mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst->gpu_addr + offset));
+		MMSCH_V2_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i,
+				mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst->gpu_addr + offset));
+		MMSCH_V2_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_OFFSET1),
+			0);
+		MMSCH_V2_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_SIZE1),
+			AMDGPU_VCN_STACK_SIZE);
+
+		MMSCH_V2_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i,
+				mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst->gpu_addr + offset +
+				AMDGPU_VCN_STACK_SIZE));
+		MMSCH_V2_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i,
+				mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst->gpu_addr + offset +
+				AMDGPU_VCN_STACK_SIZE));
+		MMSCH_V2_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_OFFSET2),
+			0);
+		MMSCH_V2_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_SIZE2),
+			AMDGPU_VCN_CONTEXT_SIZE);
+
+		for (r = 0; r < adev->vcn.inst[0].num_enc_rings; ++r) {
+			ring = &adev->vcn.inst->ring_enc[r];
+			ring->wptr = 0;
+			MMSCH_V2_0_INSERT_DIRECT_WT(
+				SOC15_REG_OFFSET(UVD, i, mmUVD_RB_BASE_LO),
+				lower_32_bits(ring->gpu_addr));
+			MMSCH_V2_0_INSERT_DIRECT_WT(
+				SOC15_REG_OFFSET(UVD, i, mmUVD_RB_BASE_HI),
+				upper_32_bits(ring->gpu_addr));
+			MMSCH_V2_0_INSERT_DIRECT_WT(
+				SOC15_REG_OFFSET(UVD, i, mmUVD_RB_SIZE),
+				ring->ring_size / 4);
+		}
+
+		ring = &adev->vcn.inst->ring_dec;
+		ring->wptr = 0;
+		MMSCH_V2_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i,
+				mmUVD_LMI_RBC_RB_64BIT_BAR_LOW),
+			lower_32_bits(ring->gpu_addr));
+		MMSCH_V2_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i,
+				mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH),
+			upper_32_bits(ring->gpu_addr));
+		/* force RBC into idle state */
+		tmp = order_base_2(ring->ring_size);
+		tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, tmp);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
+		MMSCH_V2_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i, mmUVD_RBC_RB_CNTL), tmp);
+
+		/* add end packet */
+		tmp = sizeof(struct mmsch_v2_0_cmd_end);
+		memcpy((void *)init_table, &end, tmp);
+		table_size += (tmp / 4);
+		header->vcn_table_size = table_size;
+
+	}
+	return vcn_v2_0_start_mmsch(adev, &adev->virt.mm_table);
+}
+
+static const struct amd_ip_funcs vcn_v2_0_ip_funcs = {
+	.name = "vcn_v2_0",
+	.early_init = vcn_v2_0_early_init,
+	.sw_init = vcn_v2_0_sw_init,
+	.sw_fini = vcn_v2_0_sw_fini,
+	.hw_init = vcn_v2_0_hw_init,
+	.hw_fini = vcn_v2_0_hw_fini,
+	.suspend = vcn_v2_0_suspend,
+	.resume = vcn_v2_0_resume,
+	.is_idle = vcn_v2_0_is_idle,
+	.wait_for_idle = vcn_v2_0_wait_for_idle,
+	.set_clockgating_state = vcn_v2_0_set_clockgating_state,
+	.set_powergating_state = vcn_set_powergating_state,
+	.dump_ip_state = amdgpu_vcn_dump_ip_state,
+	.print_ip_state = amdgpu_vcn_print_ip_state,
+};
+
+static const struct amdgpu_ring_funcs vcn_v2_0_dec_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_DEC,
+	.align_mask = 0xf,
+	.secure_submission_supported = true,
+	.get_rptr = vcn_v2_0_dec_ring_get_rptr,
+	.get_wptr = vcn_v2_0_dec_ring_get_wptr,
+	.set_wptr = vcn_v2_0_dec_ring_set_wptr,
+	.emit_frame_size =
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
+		8 + /* vcn_v2_0_dec_ring_emit_vm_flush */
+		14 + 14 + /* vcn_v2_0_dec_ring_emit_fence x2 vm fence */
+		6,
+	.emit_ib_size = 8, /* vcn_v2_0_dec_ring_emit_ib */
+	.emit_ib = vcn_v2_0_dec_ring_emit_ib,
+	.emit_fence = vcn_v2_0_dec_ring_emit_fence,
+	.emit_vm_flush = vcn_v2_0_dec_ring_emit_vm_flush,
+	.test_ring = vcn_v2_0_dec_ring_test_ring,
+	.test_ib = amdgpu_vcn_dec_ring_test_ib,
+	.insert_nop = vcn_v2_0_dec_ring_insert_nop,
+	.insert_start = vcn_v2_0_dec_ring_insert_start,
+	.insert_end = vcn_v2_0_dec_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_vcn_ring_begin_use,
+	.end_use = amdgpu_vcn_ring_end_use,
+	.emit_wreg = vcn_v2_0_dec_ring_emit_wreg,
+	.emit_reg_wait = vcn_v2_0_dec_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = amdgpu_vcn_ring_reset,
+};
+
+static const struct amdgpu_ring_funcs vcn_v2_0_enc_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_ENC,
+	.align_mask = 0x3f,
+	.nop = VCN_ENC_CMD_NO_OP,
+	.get_rptr = vcn_v2_0_enc_ring_get_rptr,
+	.get_wptr = vcn_v2_0_enc_ring_get_wptr,
+	.set_wptr = vcn_v2_0_enc_ring_set_wptr,
+	.emit_frame_size =
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 4 +
+		4 + /* vcn_v2_0_enc_ring_emit_vm_flush */
+		5 + 5 + /* vcn_v2_0_enc_ring_emit_fence x2 vm fence */
+		1, /* vcn_v2_0_enc_ring_insert_end */
+	.emit_ib_size = 5, /* vcn_v2_0_enc_ring_emit_ib */
+	.emit_ib = vcn_v2_0_enc_ring_emit_ib,
+	.emit_fence = vcn_v2_0_enc_ring_emit_fence,
+	.emit_vm_flush = vcn_v2_0_enc_ring_emit_vm_flush,
+	.test_ring = amdgpu_vcn_enc_ring_test_ring,
+	.test_ib = amdgpu_vcn_enc_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.insert_end = vcn_v2_0_enc_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_vcn_ring_begin_use,
+	.end_use = amdgpu_vcn_ring_end_use,
+	.emit_wreg = vcn_v2_0_enc_ring_emit_wreg,
+	.emit_reg_wait = vcn_v2_0_enc_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = amdgpu_vcn_ring_reset,
+};
+
+static void vcn_v2_0_set_dec_ring_funcs(struct amdgpu_device *adev)
+{
+	adev->vcn.inst->ring_dec.funcs = &vcn_v2_0_dec_ring_vm_funcs;
+}
+
+static void vcn_v2_0_set_enc_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->vcn.inst[0].num_enc_rings; ++i)
+		adev->vcn.inst->ring_enc[i].funcs = &vcn_v2_0_enc_ring_vm_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs vcn_v2_0_irq_funcs = {
+	.set = vcn_v2_0_set_interrupt_state,
+	.process = vcn_v2_0_process_interrupt,
+};
+
+static void vcn_v2_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->vcn.inst->irq.num_types = adev->vcn.inst[0].num_enc_rings + 1;
+	adev->vcn.inst->irq.funcs = &vcn_v2_0_irq_funcs;
+}
+
+const struct amdgpu_ip_block_version vcn_v2_0_ip_block =
+{
+		.type = AMD_IP_BLOCK_TYPE_VCN,
+		.major = 2,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &vcn_v2_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.h b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.h
new file mode 100644
index 000000000000..6c9de1882428
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.h
@@ -0,0 +1,55 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VCN_V2_0_H__
+#define __VCN_V2_0_H__
+
+extern void vcn_v2_0_dec_ring_insert_start(struct amdgpu_ring *ring);
+extern void vcn_v2_0_dec_ring_insert_end(struct amdgpu_ring *ring);
+extern void vcn_v2_0_dec_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count);
+extern void vcn_v2_0_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				unsigned flags);
+extern void vcn_v2_0_dec_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_job *job,
+				struct amdgpu_ib *ib, uint32_t flags);
+extern void vcn_v2_0_dec_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+				uint32_t val, uint32_t mask);
+extern void vcn_v2_0_dec_ring_emit_vm_flush(struct amdgpu_ring *ring,
+				unsigned vmid, uint64_t pd_addr);
+extern void vcn_v2_0_dec_ring_emit_wreg(struct amdgpu_ring *ring,
+				uint32_t reg, uint32_t val);
+extern int vcn_v2_0_dec_ring_test_ring(struct amdgpu_ring *ring);
+
+extern void vcn_v2_0_enc_ring_insert_end(struct amdgpu_ring *ring);
+extern void vcn_v2_0_enc_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
+				u64 seq, unsigned flags);
+extern void vcn_v2_0_enc_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_job *job,
+				struct amdgpu_ib *ib, uint32_t flags);
+extern void vcn_v2_0_enc_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+				uint32_t val, uint32_t mask);
+extern void vcn_v2_0_enc_ring_emit_vm_flush(struct amdgpu_ring *ring,
+				unsigned int vmid, uint64_t pd_addr);
+extern void vcn_v2_0_enc_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg, uint32_t val);
+
+extern const struct amdgpu_ip_block_version vcn_v2_0_ip_block;
+
+#endif /* __VCN_V2_0_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c
new file mode 100644
index 000000000000..cebee453871c
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c
@@ -0,0 +1,2208 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include <drm/drm_drv.h>
+
+#include "amdgpu.h"
+#include "amdgpu_vcn.h"
+#include "amdgpu_pm.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "vcn_v2_0.h"
+#include "mmsch_v1_0.h"
+#include "vcn_v2_5.h"
+
+#include "vcn/vcn_2_5_offset.h"
+#include "vcn/vcn_2_5_sh_mask.h"
+#include "ivsrcid/vcn/irqsrcs_vcn_2_0.h"
+
+#define VCN_VID_SOC_ADDRESS_2_0					0x1fa00
+#define VCN1_VID_SOC_ADDRESS_3_0				0x48200
+#define VCN1_AON_SOC_ADDRESS_3_0				0x48000
+
+#define mmUVD_CONTEXT_ID_INTERNAL_OFFSET			0x27
+#define mmUVD_GPCOM_VCPU_CMD_INTERNAL_OFFSET			0x0f
+#define mmUVD_GPCOM_VCPU_DATA0_INTERNAL_OFFSET			0x10
+#define mmUVD_GPCOM_VCPU_DATA1_INTERNAL_OFFSET			0x11
+#define mmUVD_NO_OP_INTERNAL_OFFSET				0x29
+#define mmUVD_GP_SCRATCH8_INTERNAL_OFFSET			0x66
+#define mmUVD_SCRATCH9_INTERNAL_OFFSET				0xc01d
+
+#define mmUVD_LMI_RBC_IB_VMID_INTERNAL_OFFSET			0x431
+#define mmUVD_LMI_RBC_IB_64BIT_BAR_LOW_INTERNAL_OFFSET		0x3b4
+#define mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH_INTERNAL_OFFSET		0x3b5
+#define mmUVD_RBC_IB_SIZE_INTERNAL_OFFSET			0x25c
+
+#define VCN25_MAX_HW_INSTANCES_ARCTURUS			2
+
+static const struct amdgpu_hwip_reg_entry vcn_reg_list_2_5[] = {
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_CONTEXT_ID),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_CONTEXT_ID2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_GPCOM_VCPU_DATA0),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_GPCOM_VCPU_DATA1),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_GPCOM_VCPU_CMD),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_HI),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_LO),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_HI2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_LO2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_HI3),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_LO3),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_HI4),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_LO4),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_RPTR),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_WPTR),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_RPTR2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_WPTR2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_RPTR3),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_WPTR3),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_RPTR4),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_WPTR4),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_SIZE),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_SIZE2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_SIZE3),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_SIZE4),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_PGFSM_CONFIG),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_PGFSM_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_DPG_LMA_CTL),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_DPG_LMA_DATA),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_DPG_LMA_MASK),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_DPG_PAUSE)
+};
+
+static void vcn_v2_5_set_dec_ring_funcs(struct amdgpu_device *adev);
+static void vcn_v2_5_set_enc_ring_funcs(struct amdgpu_device *adev);
+static void vcn_v2_5_set_irq_funcs(struct amdgpu_device *adev);
+static int vcn_v2_5_set_pg_state(struct amdgpu_vcn_inst *vinst,
+				 enum amd_powergating_state state);
+static int vcn_v2_5_pause_dpg_mode(struct amdgpu_vcn_inst *vinst,
+				   struct dpg_pause_state *new_state);
+static int vcn_v2_5_sriov_start(struct amdgpu_device *adev);
+static void vcn_v2_5_set_ras_funcs(struct amdgpu_device *adev);
+static int vcn_v2_5_reset(struct amdgpu_vcn_inst *vinst);
+
+static int amdgpu_ih_clientid_vcns[] = {
+	SOC15_IH_CLIENTID_VCN,
+	SOC15_IH_CLIENTID_VCN1
+};
+
+static void vcn_v2_5_idle_work_handler(struct work_struct *work)
+{
+	struct amdgpu_vcn_inst *vcn_inst =
+		container_of(work, struct amdgpu_vcn_inst, idle_work.work);
+	struct amdgpu_device *adev = vcn_inst->adev;
+	unsigned int fences = 0, fence[AMDGPU_MAX_VCN_INSTANCES] = {0};
+	unsigned int i, j;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		struct amdgpu_vcn_inst *v = &adev->vcn.inst[i];
+
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		for (j = 0; j < v->num_enc_rings; ++j)
+			fence[i] += amdgpu_fence_count_emitted(&v->ring_enc[j]);
+
+		/* Only set DPG pause for VCN3 or below, VCN4 and above will be handled by FW */
+		if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG &&
+		    !v->using_unified_queue) {
+			struct dpg_pause_state new_state;
+
+			if (fence[i] ||
+			    unlikely(atomic_read(&v->dpg_enc_submission_cnt)))
+				new_state.fw_based = VCN_DPG_STATE__PAUSE;
+			else
+				new_state.fw_based = VCN_DPG_STATE__UNPAUSE;
+
+			v->pause_dpg_mode(v, &new_state);
+		}
+
+		fence[i] += amdgpu_fence_count_emitted(&v->ring_dec);
+		fences += fence[i];
+
+	}
+
+	if (!fences && !atomic_read(&adev->vcn.inst[0].total_submission_cnt)) {
+		amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN,
+						       AMD_PG_STATE_GATE);
+		amdgpu_vcn_put_profile(adev);
+	} else {
+		schedule_delayed_work(&adev->vcn.inst[0].idle_work, VCN_IDLE_TIMEOUT);
+	}
+}
+
+static void vcn_v2_5_ring_begin_use(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_vcn_inst *v = &adev->vcn.inst[ring->me];
+
+	atomic_inc(&adev->vcn.inst[0].total_submission_cnt);
+
+	cancel_delayed_work_sync(&adev->vcn.inst[0].idle_work);
+
+	/* We can safely return early here because we've cancelled the
+	 * the delayed work so there is no one else to set it to false
+	 * and we don't care if someone else sets it to true.
+	 */
+	mutex_lock(&adev->vcn.inst[0].vcn_pg_lock);
+	amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN,
+					       AMD_PG_STATE_UNGATE);
+
+	/* Only set DPG pause for VCN3 or below, VCN4 and above will be handled by FW */
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG &&
+	    !v->using_unified_queue) {
+		struct dpg_pause_state new_state;
+
+		if (ring->funcs->type == AMDGPU_RING_TYPE_VCN_ENC) {
+			atomic_inc(&v->dpg_enc_submission_cnt);
+			new_state.fw_based = VCN_DPG_STATE__PAUSE;
+		} else {
+			unsigned int fences = 0;
+			unsigned int i;
+
+			for (i = 0; i < v->num_enc_rings; ++i)
+				fences += amdgpu_fence_count_emitted(&v->ring_enc[i]);
+
+			if (fences || atomic_read(&v->dpg_enc_submission_cnt))
+				new_state.fw_based = VCN_DPG_STATE__PAUSE;
+			else
+				new_state.fw_based = VCN_DPG_STATE__UNPAUSE;
+		}
+		v->pause_dpg_mode(v, &new_state);
+	}
+	mutex_unlock(&adev->vcn.inst[0].vcn_pg_lock);
+	amdgpu_vcn_get_profile(adev);
+}
+
+static void vcn_v2_5_ring_end_use(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	/* Only set DPG pause for VCN3 or below, VCN4 and above will be handled by FW */
+	if (ring->adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG &&
+	    ring->funcs->type == AMDGPU_RING_TYPE_VCN_ENC &&
+	    !adev->vcn.inst[ring->me].using_unified_queue)
+		atomic_dec(&adev->vcn.inst[ring->me].dpg_enc_submission_cnt);
+
+	atomic_dec(&adev->vcn.inst[0].total_submission_cnt);
+
+	schedule_delayed_work(&adev->vcn.inst[0].idle_work,
+			      VCN_IDLE_TIMEOUT);
+}
+
+/**
+ * vcn_v2_5_early_init - set function pointers and load microcode
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Set ring and irq function pointers
+ * Load microcode from filesystem
+ */
+static int vcn_v2_5_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, r;
+
+	if (amdgpu_sriov_vf(adev)) {
+		adev->vcn.num_vcn_inst = 2;
+		adev->vcn.harvest_config = 0;
+		for (i = 0; i < adev->vcn.num_vcn_inst; i++)
+			adev->vcn.inst[i].num_enc_rings = 1;
+	} else {
+		u32 harvest;
+		int i;
+
+		for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+			harvest = RREG32_SOC15(VCN, i, mmCC_UVD_HARVESTING);
+			if (harvest & CC_UVD_HARVESTING__UVD_DISABLE_MASK)
+				adev->vcn.harvest_config |= 1 << i;
+			adev->vcn.inst[i].num_enc_rings = 2;
+		}
+		if (adev->vcn.harvest_config == (AMDGPU_VCN_HARVEST_VCN0 |
+						 AMDGPU_VCN_HARVEST_VCN1))
+			/* both instances are harvested, disable the block */
+			return -ENOENT;
+	}
+
+	vcn_v2_5_set_dec_ring_funcs(adev);
+	vcn_v2_5_set_enc_ring_funcs(adev);
+	vcn_v2_5_set_irq_funcs(adev);
+	vcn_v2_5_set_ras_funcs(adev);
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		adev->vcn.inst[i].set_pg_state = vcn_v2_5_set_pg_state;
+
+		r = amdgpu_vcn_early_init(adev, i);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v2_5_sw_init - sw init for VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Load firmware and sw initialization
+ */
+static int vcn_v2_5_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_ring *ring;
+	int i, j, r;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	for (j = 0; j < adev->vcn.num_vcn_inst; j++) {
+		struct amdgpu_fw_shared *fw_shared;
+
+		if (adev->vcn.harvest_config & (1 << j))
+			continue;
+		/* VCN DEC TRAP */
+		r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_vcns[j],
+				VCN_2_0__SRCID__UVD_SYSTEM_MESSAGE_INTERRUPT, &adev->vcn.inst[j].irq);
+		if (r)
+			return r;
+
+		/* VCN ENC TRAP */
+		for (i = 0; i < adev->vcn.inst[j].num_enc_rings; ++i) {
+			r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_vcns[j],
+				i + VCN_2_0__SRCID__UVD_ENC_GENERAL_PURPOSE, &adev->vcn.inst[j].irq);
+			if (r)
+				return r;
+		}
+
+		/* VCN POISON TRAP */
+		r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_vcns[j],
+			VCN_2_6__SRCID_UVD_POISON, &adev->vcn.inst[j].ras_poison_irq);
+		if (r)
+			return r;
+
+		r = amdgpu_vcn_sw_init(adev, j);
+		if (r)
+			return r;
+
+		/* Override the work func */
+		adev->vcn.inst[j].idle_work.work.func = vcn_v2_5_idle_work_handler;
+
+		amdgpu_vcn_setup_ucode(adev, j);
+
+		r = amdgpu_vcn_resume(adev, j);
+		if (r)
+			return r;
+
+		adev->vcn.inst[j].internal.context_id = mmUVD_CONTEXT_ID_INTERNAL_OFFSET;
+		adev->vcn.inst[j].internal.ib_vmid = mmUVD_LMI_RBC_IB_VMID_INTERNAL_OFFSET;
+		adev->vcn.inst[j].internal.ib_bar_low = mmUVD_LMI_RBC_IB_64BIT_BAR_LOW_INTERNAL_OFFSET;
+		adev->vcn.inst[j].internal.ib_bar_high = mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH_INTERNAL_OFFSET;
+		adev->vcn.inst[j].internal.ib_size = mmUVD_RBC_IB_SIZE_INTERNAL_OFFSET;
+		adev->vcn.inst[j].internal.gp_scratch8 = mmUVD_GP_SCRATCH8_INTERNAL_OFFSET;
+
+		adev->vcn.inst[j].internal.scratch9 = mmUVD_SCRATCH9_INTERNAL_OFFSET;
+		adev->vcn.inst[j].external.scratch9 = SOC15_REG_OFFSET(VCN, j, mmUVD_SCRATCH9);
+		adev->vcn.inst[j].internal.data0 = mmUVD_GPCOM_VCPU_DATA0_INTERNAL_OFFSET;
+		adev->vcn.inst[j].external.data0 = SOC15_REG_OFFSET(VCN, j, mmUVD_GPCOM_VCPU_DATA0);
+		adev->vcn.inst[j].internal.data1 = mmUVD_GPCOM_VCPU_DATA1_INTERNAL_OFFSET;
+		adev->vcn.inst[j].external.data1 = SOC15_REG_OFFSET(VCN, j, mmUVD_GPCOM_VCPU_DATA1);
+		adev->vcn.inst[j].internal.cmd = mmUVD_GPCOM_VCPU_CMD_INTERNAL_OFFSET;
+		adev->vcn.inst[j].external.cmd = SOC15_REG_OFFSET(VCN, j, mmUVD_GPCOM_VCPU_CMD);
+		adev->vcn.inst[j].internal.nop = mmUVD_NO_OP_INTERNAL_OFFSET;
+		adev->vcn.inst[j].external.nop = SOC15_REG_OFFSET(VCN, j, mmUVD_NO_OP);
+
+		ring = &adev->vcn.inst[j].ring_dec;
+		ring->use_doorbell = true;
+
+		ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
+				(amdgpu_sriov_vf(adev) ? 2*j : 8*j);
+
+		if (amdgpu_ip_version(adev, UVD_HWIP, 0) == IP_VERSION(2, 5, 0))
+			ring->vm_hub = AMDGPU_MMHUB1(0);
+		else
+			ring->vm_hub = AMDGPU_MMHUB0(0);
+
+		sprintf(ring->name, "vcn_dec_%d", j);
+		r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst[j].irq,
+				     0, AMDGPU_RING_PRIO_DEFAULT, NULL);
+		if (r)
+			return r;
+
+		for (i = 0; i < adev->vcn.inst[j].num_enc_rings; ++i) {
+			enum amdgpu_ring_priority_level hw_prio = amdgpu_vcn_get_enc_ring_prio(i);
+
+			ring = &adev->vcn.inst[j].ring_enc[i];
+			ring->use_doorbell = true;
+
+			ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
+					(amdgpu_sriov_vf(adev) ? (1 + i + 2*j) : (2 + i + 8*j));
+
+			if (amdgpu_ip_version(adev, UVD_HWIP, 0) ==
+			    IP_VERSION(2, 5, 0))
+				ring->vm_hub = AMDGPU_MMHUB1(0);
+			else
+				ring->vm_hub = AMDGPU_MMHUB0(0);
+
+			sprintf(ring->name, "vcn_enc_%d.%d", j, i);
+			r = amdgpu_ring_init(adev, ring, 512,
+					     &adev->vcn.inst[j].irq, 0,
+					     hw_prio, NULL);
+			if (r)
+				return r;
+		}
+
+		fw_shared = adev->vcn.inst[j].fw_shared.cpu_addr;
+		fw_shared->present_flag_0 = cpu_to_le32(AMDGPU_VCN_MULTI_QUEUE_FLAG);
+
+		if (amdgpu_vcnfw_log)
+			amdgpu_vcn_fwlog_init(&adev->vcn.inst[i]);
+
+		if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)
+			adev->vcn.inst[j].pause_dpg_mode = vcn_v2_5_pause_dpg_mode;
+		adev->vcn.inst[j].reset = vcn_v2_5_reset;
+	}
+
+	adev->vcn.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->vcn.inst[0].ring_enc[0]);
+	if (!amdgpu_sriov_vf(adev))
+		adev->vcn.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+
+	if (amdgpu_sriov_vf(adev)) {
+		r = amdgpu_virt_alloc_mm_table(adev);
+		if (r)
+			return r;
+	}
+
+	r = amdgpu_vcn_ras_sw_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_vcn_reg_dump_init(adev, vcn_reg_list_2_5, ARRAY_SIZE(vcn_reg_list_2_5));
+	if (r)
+		return r;
+
+	r = amdgpu_vcn_sysfs_reset_mask_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+/**
+ * vcn_v2_5_sw_fini - sw fini for VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * VCN suspend and free up sw allocation
+ */
+static int vcn_v2_5_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	int i, r, idx;
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_fw_shared *fw_shared;
+
+	if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+		for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+			if (adev->vcn.harvest_config & (1 << i))
+				continue;
+			fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+			fw_shared->present_flag_0 = 0;
+		}
+		drm_dev_exit(idx);
+	}
+
+
+	if (amdgpu_sriov_vf(adev))
+		amdgpu_virt_free_mm_table(adev);
+
+	amdgpu_vcn_sysfs_reset_mask_fini(adev);
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		r = amdgpu_vcn_suspend(adev, i);
+		if (r)
+			return r;
+		amdgpu_vcn_sw_fini(adev, i);
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v2_5_hw_init - start and test VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Initialize the hardware, boot up the VCPU and do some testing
+ */
+static int vcn_v2_5_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	int i, j, r = 0;
+
+	if (amdgpu_sriov_vf(adev))
+		r = vcn_v2_5_sriov_start(adev);
+
+	for (j = 0; j < adev->vcn.num_vcn_inst; ++j) {
+		if (adev->vcn.harvest_config & (1 << j))
+			continue;
+
+		if (amdgpu_sriov_vf(adev)) {
+			adev->vcn.inst[j].ring_enc[0].sched.ready = true;
+			adev->vcn.inst[j].ring_enc[1].sched.ready = false;
+			adev->vcn.inst[j].ring_enc[2].sched.ready = false;
+			adev->vcn.inst[j].ring_dec.sched.ready = true;
+		} else {
+
+			ring = &adev->vcn.inst[j].ring_dec;
+
+			adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
+						     ring->doorbell_index, j);
+
+			r = amdgpu_ring_test_helper(ring);
+			if (r)
+				return r;
+
+			for (i = 0; i < adev->vcn.inst[j].num_enc_rings; ++i) {
+				ring = &adev->vcn.inst[j].ring_enc[i];
+				r = amdgpu_ring_test_helper(ring);
+				if (r)
+					return r;
+			}
+		}
+	}
+
+	return r;
+}
+
+/**
+ * vcn_v2_5_hw_fini - stop the hardware block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Stop the VCN block, mark ring as not ready any more
+ */
+static int vcn_v2_5_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[i];
+
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		cancel_delayed_work_sync(&vinst->idle_work);
+
+		if ((adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) ||
+		    (vinst->cur_state != AMD_PG_STATE_GATE &&
+		     RREG32_SOC15(VCN, i, mmUVD_STATUS)))
+			vinst->set_pg_state(vinst, AMD_PG_STATE_GATE);
+
+		if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__VCN))
+			amdgpu_irq_put(adev, &vinst->ras_poison_irq, 0);
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v2_5_suspend - suspend VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * HW fini and suspend VCN block
+ */
+static int vcn_v2_5_suspend(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r, i;
+
+	r = vcn_v2_5_hw_fini(ip_block);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		r = amdgpu_vcn_suspend(ip_block->adev, i);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v2_5_resume - resume VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Resume firmware and hw init VCN block
+ */
+static int vcn_v2_5_resume(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r, i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		r = amdgpu_vcn_resume(ip_block->adev, i);
+		if (r)
+			return r;
+	}
+
+	r = vcn_v2_5_hw_init(ip_block);
+
+	return r;
+}
+
+/**
+ * vcn_v2_5_mc_resume - memory controller programming
+ *
+ * @vinst: VCN instance
+ *
+ * Let the VCN memory controller know it's offsets
+ */
+static void vcn_v2_5_mc_resume(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int i = vinst->inst;
+	uint32_t size;
+	uint32_t offset;
+
+	if (adev->vcn.harvest_config & (1 << i))
+		return;
+
+	size = AMDGPU_GPU_PAGE_ALIGN(adev->vcn.inst[i].fw->size + 4);
+	/* cache window 0: fw */
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		WREG32_SOC15(VCN, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			     (adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + i].tmr_mc_addr_lo));
+		WREG32_SOC15(VCN, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			     (adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + i].tmr_mc_addr_hi));
+		WREG32_SOC15(VCN, i, mmUVD_VCPU_CACHE_OFFSET0, 0);
+		offset = 0;
+	} else {
+		WREG32_SOC15(VCN, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			     lower_32_bits(adev->vcn.inst[i].gpu_addr));
+		WREG32_SOC15(VCN, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			     upper_32_bits(adev->vcn.inst[i].gpu_addr));
+		offset = size;
+		WREG32_SOC15(VCN, i, mmUVD_VCPU_CACHE_OFFSET0,
+			     AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
+	}
+	WREG32_SOC15(VCN, i, mmUVD_VCPU_CACHE_SIZE0, size);
+
+	/* cache window 1: stack */
+	WREG32_SOC15(VCN, i, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW,
+		     lower_32_bits(adev->vcn.inst[i].gpu_addr + offset));
+	WREG32_SOC15(VCN, i, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH,
+		     upper_32_bits(adev->vcn.inst[i].gpu_addr + offset));
+	WREG32_SOC15(VCN, i, mmUVD_VCPU_CACHE_OFFSET1, 0);
+	WREG32_SOC15(VCN, i, mmUVD_VCPU_CACHE_SIZE1, AMDGPU_VCN_STACK_SIZE);
+
+	/* cache window 2: context */
+	WREG32_SOC15(VCN, i, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW,
+		     lower_32_bits(adev->vcn.inst[i].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE));
+	WREG32_SOC15(VCN, i, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH,
+		     upper_32_bits(adev->vcn.inst[i].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE));
+	WREG32_SOC15(VCN, i, mmUVD_VCPU_CACHE_OFFSET2, 0);
+	WREG32_SOC15(VCN, i, mmUVD_VCPU_CACHE_SIZE2, AMDGPU_VCN_CONTEXT_SIZE);
+
+	/* non-cache window */
+	WREG32_SOC15(VCN, i, mmUVD_LMI_VCPU_NC0_64BIT_BAR_LOW,
+		     lower_32_bits(adev->vcn.inst[i].fw_shared.gpu_addr));
+	WREG32_SOC15(VCN, i, mmUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH,
+		     upper_32_bits(adev->vcn.inst[i].fw_shared.gpu_addr));
+	WREG32_SOC15(VCN, i, mmUVD_VCPU_NONCACHE_OFFSET0, 0);
+	WREG32_SOC15(VCN, i, mmUVD_VCPU_NONCACHE_SIZE0,
+		     AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_fw_shared)));
+}
+
+static void vcn_v2_5_mc_resume_dpg_mode(struct amdgpu_vcn_inst *vinst,
+					bool indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t size = AMDGPU_GPU_PAGE_ALIGN(adev->vcn.inst[inst_idx].fw->size + 4);
+	uint32_t offset;
+
+	/* cache window 0: fw */
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		if (!indirect) {
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+				(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst_idx].tmr_mc_addr_lo), 0, indirect);
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+				(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst_idx].tmr_mc_addr_hi), 0, indirect);
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, 0, mmUVD_VCPU_CACHE_OFFSET0), 0, 0, indirect);
+		} else {
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW), 0, 0, indirect);
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH), 0, 0, indirect);
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, 0, mmUVD_VCPU_CACHE_OFFSET0), 0, 0, indirect);
+		}
+		offset = 0;
+	} else {
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr), 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr), 0, indirect);
+		offset = size;
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, mmUVD_VCPU_CACHE_OFFSET0),
+			AMDGPU_UVD_FIRMWARE_OFFSET >> 3, 0, indirect);
+	}
+
+	if (!indirect)
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, mmUVD_VCPU_CACHE_SIZE0), size, 0, indirect);
+	else
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, mmUVD_VCPU_CACHE_SIZE0), 0, 0, indirect);
+
+	/* cache window 1: stack */
+	if (!indirect) {
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset), 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset), 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, mmUVD_VCPU_CACHE_OFFSET1), 0, 0, indirect);
+	} else {
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW), 0, 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH), 0, 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, mmUVD_VCPU_CACHE_OFFSET1), 0, 0, indirect);
+	}
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_VCPU_CACHE_SIZE1), AMDGPU_VCN_STACK_SIZE, 0, indirect);
+
+	/* cache window 2: context */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW),
+		lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE), 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH),
+		upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE), 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_VCPU_CACHE_OFFSET2), 0, 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_VCPU_CACHE_SIZE2), AMDGPU_VCN_CONTEXT_SIZE, 0, indirect);
+
+	/* non-cache window */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_LOW),
+		lower_32_bits(adev->vcn.inst[inst_idx].fw_shared.gpu_addr), 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH),
+		upper_32_bits(adev->vcn.inst[inst_idx].fw_shared.gpu_addr), 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_VCPU_NONCACHE_OFFSET0), 0, 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_VCPU_NONCACHE_SIZE0),
+		AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_fw_shared)), 0, indirect);
+
+	/* VCN global tiling registers */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_GFX8_ADDR_CONFIG), adev->gfx.config.gb_addr_config, 0, indirect);
+}
+
+/**
+ * vcn_v2_5_disable_clock_gating - disable VCN clock gating
+ *
+ * @vinst: VCN instance
+ *
+ * Disable clock gating for VCN block
+ */
+static void vcn_v2_5_disable_clock_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int i = vinst->inst;
+	uint32_t data;
+
+	if (adev->vcn.harvest_config & (1 << i))
+		return;
+	/* UVD disable CGC */
+	data = RREG32_SOC15(VCN, i, mmUVD_CGC_CTRL);
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		data |= 1 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		data &= ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK;
+	data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(VCN, i, mmUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, i, mmUVD_CGC_GATE);
+	data &= ~(UVD_CGC_GATE__SYS_MASK
+		  | UVD_CGC_GATE__UDEC_MASK
+		  | UVD_CGC_GATE__MPEG2_MASK
+		  | UVD_CGC_GATE__REGS_MASK
+		  | UVD_CGC_GATE__RBC_MASK
+		  | UVD_CGC_GATE__LMI_MC_MASK
+		  | UVD_CGC_GATE__LMI_UMC_MASK
+		  | UVD_CGC_GATE__IDCT_MASK
+		  | UVD_CGC_GATE__MPRD_MASK
+		  | UVD_CGC_GATE__MPC_MASK
+		  | UVD_CGC_GATE__LBSI_MASK
+		  | UVD_CGC_GATE__LRBBM_MASK
+		  | UVD_CGC_GATE__UDEC_RE_MASK
+		  | UVD_CGC_GATE__UDEC_CM_MASK
+		  | UVD_CGC_GATE__UDEC_IT_MASK
+		  | UVD_CGC_GATE__UDEC_DB_MASK
+		  | UVD_CGC_GATE__UDEC_MP_MASK
+		  | UVD_CGC_GATE__WCB_MASK
+		  | UVD_CGC_GATE__VCPU_MASK
+		  | UVD_CGC_GATE__MMSCH_MASK);
+
+	WREG32_SOC15(VCN, i, mmUVD_CGC_GATE, data);
+
+	SOC15_WAIT_ON_RREG(VCN, i, mmUVD_CGC_GATE, 0,  0xFFFFFFFF);
+
+	data = RREG32_SOC15(VCN, i, mmUVD_CGC_CTRL);
+	data &= ~(UVD_CGC_CTRL__UDEC_RE_MODE_MASK
+		  | UVD_CGC_CTRL__UDEC_CM_MODE_MASK
+		  | UVD_CGC_CTRL__UDEC_IT_MODE_MASK
+		  | UVD_CGC_CTRL__UDEC_DB_MODE_MASK
+		  | UVD_CGC_CTRL__UDEC_MP_MODE_MASK
+		  | UVD_CGC_CTRL__SYS_MODE_MASK
+		  | UVD_CGC_CTRL__UDEC_MODE_MASK
+		  | UVD_CGC_CTRL__MPEG2_MODE_MASK
+		  | UVD_CGC_CTRL__REGS_MODE_MASK
+		  | UVD_CGC_CTRL__RBC_MODE_MASK
+		  | UVD_CGC_CTRL__LMI_MC_MODE_MASK
+		  | UVD_CGC_CTRL__LMI_UMC_MODE_MASK
+		  | UVD_CGC_CTRL__IDCT_MODE_MASK
+		  | UVD_CGC_CTRL__MPRD_MODE_MASK
+		  | UVD_CGC_CTRL__MPC_MODE_MASK
+		  | UVD_CGC_CTRL__LBSI_MODE_MASK
+		  | UVD_CGC_CTRL__LRBBM_MODE_MASK
+		  | UVD_CGC_CTRL__WCB_MODE_MASK
+		  | UVD_CGC_CTRL__VCPU_MODE_MASK
+		  | UVD_CGC_CTRL__MMSCH_MODE_MASK);
+	WREG32_SOC15(VCN, i, mmUVD_CGC_CTRL, data);
+
+	/* turn on */
+	data = RREG32_SOC15(VCN, i, mmUVD_SUVD_CGC_GATE);
+	data |= (UVD_SUVD_CGC_GATE__SRE_MASK
+		 | UVD_SUVD_CGC_GATE__SIT_MASK
+		 | UVD_SUVD_CGC_GATE__SMP_MASK
+		 | UVD_SUVD_CGC_GATE__SCM_MASK
+		 | UVD_SUVD_CGC_GATE__SDB_MASK
+		 | UVD_SUVD_CGC_GATE__SRE_H264_MASK
+		 | UVD_SUVD_CGC_GATE__SRE_HEVC_MASK
+		 | UVD_SUVD_CGC_GATE__SIT_H264_MASK
+		 | UVD_SUVD_CGC_GATE__SIT_HEVC_MASK
+		 | UVD_SUVD_CGC_GATE__SCM_H264_MASK
+		 | UVD_SUVD_CGC_GATE__SCM_HEVC_MASK
+		 | UVD_SUVD_CGC_GATE__SDB_H264_MASK
+		 | UVD_SUVD_CGC_GATE__SDB_HEVC_MASK
+		 | UVD_SUVD_CGC_GATE__SCLR_MASK
+		 | UVD_SUVD_CGC_GATE__UVD_SC_MASK
+		 | UVD_SUVD_CGC_GATE__ENT_MASK
+		 | UVD_SUVD_CGC_GATE__SIT_HEVC_DEC_MASK
+		 | UVD_SUVD_CGC_GATE__SIT_HEVC_ENC_MASK
+		 | UVD_SUVD_CGC_GATE__SITE_MASK
+		 | UVD_SUVD_CGC_GATE__SRE_VP9_MASK
+		 | UVD_SUVD_CGC_GATE__SCM_VP9_MASK
+		 | UVD_SUVD_CGC_GATE__SIT_VP9_DEC_MASK
+		 | UVD_SUVD_CGC_GATE__SDB_VP9_MASK
+		 | UVD_SUVD_CGC_GATE__IME_HEVC_MASK);
+	WREG32_SOC15(VCN, i, mmUVD_SUVD_CGC_GATE, data);
+
+	data = RREG32_SOC15(VCN, i, mmUVD_SUVD_CGC_CTRL);
+	data &= ~(UVD_SUVD_CGC_CTRL__SRE_MODE_MASK
+		  | UVD_SUVD_CGC_CTRL__SIT_MODE_MASK
+		  | UVD_SUVD_CGC_CTRL__SMP_MODE_MASK
+		  | UVD_SUVD_CGC_CTRL__SCM_MODE_MASK
+		  | UVD_SUVD_CGC_CTRL__SDB_MODE_MASK
+		  | UVD_SUVD_CGC_CTRL__SCLR_MODE_MASK
+		  | UVD_SUVD_CGC_CTRL__UVD_SC_MODE_MASK
+		  | UVD_SUVD_CGC_CTRL__ENT_MODE_MASK
+			| UVD_SUVD_CGC_CTRL__IME_MODE_MASK
+		  | UVD_SUVD_CGC_CTRL__SITE_MODE_MASK);
+	WREG32_SOC15(VCN, i, mmUVD_SUVD_CGC_CTRL, data);
+}
+
+static void vcn_v2_5_clock_gating_dpg_mode(struct amdgpu_vcn_inst *vinst,
+		uint8_t sram_sel, uint8_t indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t reg_data = 0;
+
+	/* enable sw clock gating control */
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		reg_data = 1 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		reg_data = 0 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	reg_data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	reg_data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	reg_data &= ~(UVD_CGC_CTRL__UDEC_RE_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_CM_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_IT_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_DB_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_MP_MODE_MASK |
+		 UVD_CGC_CTRL__SYS_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_MODE_MASK |
+		 UVD_CGC_CTRL__MPEG2_MODE_MASK |
+		 UVD_CGC_CTRL__REGS_MODE_MASK |
+		 UVD_CGC_CTRL__RBC_MODE_MASK |
+		 UVD_CGC_CTRL__LMI_MC_MODE_MASK |
+		 UVD_CGC_CTRL__LMI_UMC_MODE_MASK |
+		 UVD_CGC_CTRL__IDCT_MODE_MASK |
+		 UVD_CGC_CTRL__MPRD_MODE_MASK |
+		 UVD_CGC_CTRL__MPC_MODE_MASK |
+		 UVD_CGC_CTRL__LBSI_MODE_MASK |
+		 UVD_CGC_CTRL__LRBBM_MODE_MASK |
+		 UVD_CGC_CTRL__WCB_MODE_MASK |
+		 UVD_CGC_CTRL__VCPU_MODE_MASK |
+		 UVD_CGC_CTRL__MMSCH_MODE_MASK);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_CGC_CTRL), reg_data, sram_sel, indirect);
+
+	/* turn off clock gating */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_CGC_GATE), 0, sram_sel, indirect);
+
+	/* turn on SUVD clock gating */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_SUVD_CGC_GATE), 1, sram_sel, indirect);
+
+	/* turn on sw mode in UVD_SUVD_CGC_CTRL */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_SUVD_CGC_CTRL), 0, sram_sel, indirect);
+}
+
+/**
+ * vcn_v2_5_enable_clock_gating - enable VCN clock gating
+ *
+ * @vinst: VCN instance
+ *
+ * Enable clock gating for VCN block
+ */
+static void vcn_v2_5_enable_clock_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int i = vinst->inst;
+	uint32_t data = 0;
+
+	if (adev->vcn.harvest_config & (1 << i))
+		return;
+	/* enable UVD CGC */
+	data = RREG32_SOC15(VCN, i, mmUVD_CGC_CTRL);
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		data |= 1 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		data |= 0 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(VCN, i, mmUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, i, mmUVD_CGC_CTRL);
+	data |= (UVD_CGC_CTRL__UDEC_RE_MODE_MASK
+		 | UVD_CGC_CTRL__UDEC_CM_MODE_MASK
+		 | UVD_CGC_CTRL__UDEC_IT_MODE_MASK
+		 | UVD_CGC_CTRL__UDEC_DB_MODE_MASK
+		 | UVD_CGC_CTRL__UDEC_MP_MODE_MASK
+		 | UVD_CGC_CTRL__SYS_MODE_MASK
+		 | UVD_CGC_CTRL__UDEC_MODE_MASK
+		 | UVD_CGC_CTRL__MPEG2_MODE_MASK
+		 | UVD_CGC_CTRL__REGS_MODE_MASK
+		 | UVD_CGC_CTRL__RBC_MODE_MASK
+		 | UVD_CGC_CTRL__LMI_MC_MODE_MASK
+		 | UVD_CGC_CTRL__LMI_UMC_MODE_MASK
+		 | UVD_CGC_CTRL__IDCT_MODE_MASK
+		 | UVD_CGC_CTRL__MPRD_MODE_MASK
+		 | UVD_CGC_CTRL__MPC_MODE_MASK
+		 | UVD_CGC_CTRL__LBSI_MODE_MASK
+		 | UVD_CGC_CTRL__LRBBM_MODE_MASK
+		 | UVD_CGC_CTRL__WCB_MODE_MASK
+		 | UVD_CGC_CTRL__VCPU_MODE_MASK);
+	WREG32_SOC15(VCN, i, mmUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, i, mmUVD_SUVD_CGC_CTRL);
+	data |= (UVD_SUVD_CGC_CTRL__SRE_MODE_MASK
+		 | UVD_SUVD_CGC_CTRL__SIT_MODE_MASK
+		 | UVD_SUVD_CGC_CTRL__SMP_MODE_MASK
+		 | UVD_SUVD_CGC_CTRL__SCM_MODE_MASK
+		 | UVD_SUVD_CGC_CTRL__SDB_MODE_MASK
+		 | UVD_SUVD_CGC_CTRL__SCLR_MODE_MASK
+		 | UVD_SUVD_CGC_CTRL__UVD_SC_MODE_MASK
+		 | UVD_SUVD_CGC_CTRL__ENT_MODE_MASK
+		 | UVD_SUVD_CGC_CTRL__IME_MODE_MASK
+		 | UVD_SUVD_CGC_CTRL__SITE_MODE_MASK);
+	WREG32_SOC15(VCN, i, mmUVD_SUVD_CGC_CTRL, data);
+}
+
+static void vcn_v2_6_enable_ras(struct amdgpu_vcn_inst *vinst,
+				bool indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t tmp;
+
+	if (amdgpu_ip_version(adev, UVD_HWIP, 0) != IP_VERSION(2, 6, 0))
+		return;
+
+	tmp = VCN_RAS_CNTL__VCPU_VCODEC_REARM_MASK |
+	      VCN_RAS_CNTL__VCPU_VCODEC_IH_EN_MASK |
+	      VCN_RAS_CNTL__VCPU_VCODEC_PMI_EN_MASK |
+	      VCN_RAS_CNTL__VCPU_VCODEC_STALL_EN_MASK;
+	WREG32_SOC15_DPG_MODE(inst_idx,
+			      SOC15_DPG_MODE_OFFSET(VCN, 0, mmVCN_RAS_CNTL),
+			      tmp, 0, indirect);
+
+	tmp = UVD_VCPU_INT_EN__RASCNTL_VCPU_VCODEC_EN_MASK;
+	WREG32_SOC15_DPG_MODE(inst_idx,
+			      SOC15_DPG_MODE_OFFSET(VCN, 0, mmUVD_VCPU_INT_EN),
+			      tmp, 0, indirect);
+
+	tmp = UVD_SYS_INT_EN__RASCNTL_VCPU_VCODEC_EN_MASK;
+	WREG32_SOC15_DPG_MODE(inst_idx,
+			      SOC15_DPG_MODE_OFFSET(VCN, 0, mmUVD_SYS_INT_EN),
+			      tmp, 0, indirect);
+}
+
+static int vcn_v2_5_start_dpg_mode(struct amdgpu_vcn_inst *vinst, bool indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	struct amdgpu_fw_shared *fw_shared = adev->vcn.inst[inst_idx].fw_shared.cpu_addr;
+	struct amdgpu_ring *ring;
+	uint32_t rb_bufsz, tmp;
+	int ret;
+
+	/* disable register anti-hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, mmUVD_POWER_STATUS), 1,
+		~UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+	/* enable dynamic power gating mode */
+	tmp = RREG32_SOC15(VCN, inst_idx, mmUVD_POWER_STATUS);
+	tmp |= UVD_POWER_STATUS__UVD_PG_MODE_MASK;
+	tmp |= UVD_POWER_STATUS__UVD_PG_EN_MASK;
+	WREG32_SOC15(VCN, inst_idx, mmUVD_POWER_STATUS, tmp);
+
+	if (indirect)
+		adev->vcn.inst[inst_idx].dpg_sram_curr_addr = (uint32_t *)adev->vcn.inst[inst_idx].dpg_sram_cpu_addr;
+
+	/* enable clock gating */
+	vcn_v2_5_clock_gating_dpg_mode(vinst, 0, indirect);
+
+	/* enable VCPU clock */
+	tmp = (0xFF << UVD_VCPU_CNTL__PRB_TIMEOUT_VAL__SHIFT);
+	tmp |= UVD_VCPU_CNTL__CLK_EN_MASK;
+	tmp |= UVD_VCPU_CNTL__BLK_RST_MASK;
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_VCPU_CNTL), tmp, 0, indirect);
+
+	/* disable master interupt */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_MASTINT_EN), 0, 0, indirect);
+
+	/* setup mmUVD_LMI_CTRL */
+	tmp = (0x8 | UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+		UVD_LMI_CTRL__REQ_MODE_MASK |
+		UVD_LMI_CTRL__CRC_RESET_MASK |
+		UVD_LMI_CTRL__MASK_MC_URGENT_MASK |
+		UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+		UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
+		(8 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
+		0x00100000L);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_LMI_CTRL), tmp, 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_MPC_CNTL),
+		0x2 << UVD_MPC_CNTL__REPLACEMENT_MODE__SHIFT, 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_MPC_SET_MUXA0),
+		((0x1 << UVD_MPC_SET_MUXA0__VARA_1__SHIFT) |
+		 (0x2 << UVD_MPC_SET_MUXA0__VARA_2__SHIFT) |
+		 (0x3 << UVD_MPC_SET_MUXA0__VARA_3__SHIFT) |
+		 (0x4 << UVD_MPC_SET_MUXA0__VARA_4__SHIFT)), 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_MPC_SET_MUXB0),
+		((0x1 << UVD_MPC_SET_MUXB0__VARB_1__SHIFT) |
+		 (0x2 << UVD_MPC_SET_MUXB0__VARB_2__SHIFT) |
+		 (0x3 << UVD_MPC_SET_MUXB0__VARB_3__SHIFT) |
+		 (0x4 << UVD_MPC_SET_MUXB0__VARB_4__SHIFT)), 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_MPC_SET_MUX),
+		((0x0 << UVD_MPC_SET_MUX__SET_0__SHIFT) |
+		 (0x1 << UVD_MPC_SET_MUX__SET_1__SHIFT) |
+		 (0x2 << UVD_MPC_SET_MUX__SET_2__SHIFT)), 0, indirect);
+
+	vcn_v2_5_mc_resume_dpg_mode(vinst, indirect);
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_REG_XX_MASK), 0x10, 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_RBC_XX_IB_REG_CHECK), 0x3, 0, indirect);
+
+	/* enable LMI MC and UMC channels */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_LMI_CTRL2), 0, 0, indirect);
+
+	vcn_v2_6_enable_ras(vinst, indirect);
+
+	/* unblock VCPU register access */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_RB_ARB_CTRL), 0, 0, indirect);
+
+	tmp = (0xFF << UVD_VCPU_CNTL__PRB_TIMEOUT_VAL__SHIFT);
+	tmp |= UVD_VCPU_CNTL__CLK_EN_MASK;
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_VCPU_CNTL), tmp, 0, indirect);
+
+	/* enable master interrupt */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, mmUVD_MASTINT_EN),
+		UVD_MASTINT_EN__VCPU_EN_MASK, 0, indirect);
+
+	if (indirect) {
+		ret = amdgpu_vcn_psp_update_sram(adev, inst_idx, 0);
+		if (ret) {
+			dev_err(adev->dev, "vcn sram load failed %d\n", ret);
+			return ret;
+		}
+	}
+
+	ring = &adev->vcn.inst[inst_idx].ring_dec;
+	/* force RBC into idle state */
+	rb_bufsz = order_base_2(ring->ring_size);
+	tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
+	WREG32_SOC15(VCN, inst_idx, mmUVD_RBC_RB_CNTL, tmp);
+
+	/* Stall DPG before WPTR/RPTR reset */
+	WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, mmUVD_POWER_STATUS),
+		UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK,
+		~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK);
+	fw_shared->multi_queue.decode_queue_mode |= FW_QUEUE_RING_RESET;
+
+	/* set the write pointer delay */
+	WREG32_SOC15(VCN, inst_idx, mmUVD_RBC_RB_WPTR_CNTL, 0);
+
+	/* set the wb address */
+	WREG32_SOC15(VCN, inst_idx, mmUVD_RBC_RB_RPTR_ADDR,
+		(upper_32_bits(ring->gpu_addr) >> 2));
+
+	/* program the RB_BASE for ring buffer */
+	WREG32_SOC15(VCN, inst_idx, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW,
+		lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15(VCN, inst_idx, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH,
+		upper_32_bits(ring->gpu_addr));
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32_SOC15(VCN, inst_idx, mmUVD_RBC_RB_RPTR, 0);
+
+	WREG32_SOC15(VCN, inst_idx, mmUVD_SCRATCH2, 0);
+
+	ring->wptr = RREG32_SOC15(VCN, inst_idx, mmUVD_RBC_RB_RPTR);
+	WREG32_SOC15(VCN, inst_idx, mmUVD_RBC_RB_WPTR,
+		lower_32_bits(ring->wptr));
+
+	fw_shared->multi_queue.decode_queue_mode &= ~FW_QUEUE_RING_RESET;
+	/* Unstall DPG */
+	WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, mmUVD_POWER_STATUS),
+		0, ~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, inst_idx, mmUVD_STATUS);
+
+	return 0;
+}
+
+static int vcn_v2_5_start(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int i = vinst->inst;
+	struct amdgpu_fw_shared *fw_shared =
+		adev->vcn.inst[i].fw_shared.cpu_addr;
+	struct amdgpu_ring *ring;
+	uint32_t rb_bufsz, tmp;
+	int j, k, r;
+
+	if (adev->vcn.harvest_config & (1 << i))
+		return 0;
+
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_vcn(adev, true, i);
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)
+		return vcn_v2_5_start_dpg_mode(vinst, adev->vcn.inst[i].indirect_sram);
+
+	/* disable register anti-hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_POWER_STATUS), 0,
+		 ~UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+	/* set uvd status busy */
+	tmp = RREG32_SOC15(VCN, i, mmUVD_STATUS) | UVD_STATUS__UVD_BUSY;
+	WREG32_SOC15(VCN, i, mmUVD_STATUS, tmp);
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)
+		return 0;
+
+	/* SW clock gating */
+	vcn_v2_5_disable_clock_gating(vinst);
+
+	/* enable VCPU clock */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_VCPU_CNTL),
+		 UVD_VCPU_CNTL__CLK_EN_MASK, ~UVD_VCPU_CNTL__CLK_EN_MASK);
+
+	/* disable master interrupt */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_MASTINT_EN), 0,
+		 ~UVD_MASTINT_EN__VCPU_EN_MASK);
+
+	/* setup mmUVD_LMI_CTRL */
+	tmp = RREG32_SOC15(VCN, i, mmUVD_LMI_CTRL);
+	tmp &= ~0xff;
+	WREG32_SOC15(VCN, i, mmUVD_LMI_CTRL, tmp | 0x8|
+		     UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK	|
+		     UVD_LMI_CTRL__MASK_MC_URGENT_MASK |
+		     UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+		     UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK);
+
+	/* setup mmUVD_MPC_CNTL */
+	tmp = RREG32_SOC15(VCN, i, mmUVD_MPC_CNTL);
+	tmp &= ~UVD_MPC_CNTL__REPLACEMENT_MODE_MASK;
+	tmp |= 0x2 << UVD_MPC_CNTL__REPLACEMENT_MODE__SHIFT;
+	WREG32_SOC15(VCN, i, mmUVD_MPC_CNTL, tmp);
+
+	/* setup UVD_MPC_SET_MUXA0 */
+	WREG32_SOC15(VCN, i, mmUVD_MPC_SET_MUXA0,
+		     ((0x1 << UVD_MPC_SET_MUXA0__VARA_1__SHIFT) |
+		      (0x2 << UVD_MPC_SET_MUXA0__VARA_2__SHIFT) |
+		      (0x3 << UVD_MPC_SET_MUXA0__VARA_3__SHIFT) |
+		      (0x4 << UVD_MPC_SET_MUXA0__VARA_4__SHIFT)));
+
+	/* setup UVD_MPC_SET_MUXB0 */
+	WREG32_SOC15(VCN, i, mmUVD_MPC_SET_MUXB0,
+		     ((0x1 << UVD_MPC_SET_MUXB0__VARB_1__SHIFT) |
+		      (0x2 << UVD_MPC_SET_MUXB0__VARB_2__SHIFT) |
+		      (0x3 << UVD_MPC_SET_MUXB0__VARB_3__SHIFT) |
+		      (0x4 << UVD_MPC_SET_MUXB0__VARB_4__SHIFT)));
+
+	/* setup mmUVD_MPC_SET_MUX */
+	WREG32_SOC15(VCN, i, mmUVD_MPC_SET_MUX,
+		     ((0x0 << UVD_MPC_SET_MUX__SET_0__SHIFT) |
+		      (0x1 << UVD_MPC_SET_MUX__SET_1__SHIFT) |
+		      (0x2 << UVD_MPC_SET_MUX__SET_2__SHIFT)));
+
+	vcn_v2_5_mc_resume(vinst);
+
+	/* VCN global tiling registers */
+	WREG32_SOC15(VCN, i, mmUVD_GFX8_ADDR_CONFIG,
+		     adev->gfx.config.gb_addr_config);
+	WREG32_SOC15(VCN, i, mmUVD_GFX8_ADDR_CONFIG,
+		     adev->gfx.config.gb_addr_config);
+
+	/* enable LMI MC and UMC channels */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_LMI_CTRL2), 0,
+		 ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+
+	/* unblock VCPU register access */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_RB_ARB_CTRL), 0,
+		 ~UVD_RB_ARB_CTRL__VCPU_DIS_MASK);
+
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_VCPU_CNTL), 0,
+		 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+	for (k = 0; k < 10; ++k) {
+		uint32_t status;
+
+		for (j = 0; j < 100; ++j) {
+			status = RREG32_SOC15(VCN, i, mmUVD_STATUS);
+			if (status & 2)
+				break;
+			if (amdgpu_emu_mode == 1)
+				msleep(500);
+			else
+				mdelay(10);
+		}
+		r = 0;
+		if (status & 2)
+			break;
+
+		DRM_ERROR("VCN decode not responding, trying to reset the VCPU!!!\n");
+		WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_VCPU_CNTL),
+			 UVD_VCPU_CNTL__BLK_RST_MASK,
+			 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+		mdelay(10);
+		WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_VCPU_CNTL), 0,
+			 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+		mdelay(10);
+		r = -1;
+	}
+
+	if (r) {
+		DRM_ERROR("VCN decode not responding, giving up!!!\n");
+		return r;
+	}
+
+	/* enable master interrupt */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_MASTINT_EN),
+		 UVD_MASTINT_EN__VCPU_EN_MASK,
+		 ~UVD_MASTINT_EN__VCPU_EN_MASK);
+
+	/* clear the busy bit of VCN_STATUS */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_STATUS), 0,
+		 ~(2 << UVD_STATUS__VCPU_REPORT__SHIFT));
+
+	WREG32_SOC15(VCN, i, mmUVD_LMI_RBC_RB_VMID, 0);
+
+	ring = &adev->vcn.inst[i].ring_dec;
+	/* force RBC into idle state */
+	rb_bufsz = order_base_2(ring->ring_size);
+	tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
+	WREG32_SOC15(VCN, i, mmUVD_RBC_RB_CNTL, tmp);
+
+	fw_shared->multi_queue.decode_queue_mode |= FW_QUEUE_RING_RESET;
+	/* program the RB_BASE for ring buffer */
+	WREG32_SOC15(VCN, i, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW,
+		     lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15(VCN, i, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH,
+		     upper_32_bits(ring->gpu_addr));
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32_SOC15(VCN, i, mmUVD_RBC_RB_RPTR, 0);
+
+	ring->wptr = RREG32_SOC15(VCN, i, mmUVD_RBC_RB_RPTR);
+	WREG32_SOC15(VCN, i, mmUVD_RBC_RB_WPTR,
+		     lower_32_bits(ring->wptr));
+	fw_shared->multi_queue.decode_queue_mode &= ~FW_QUEUE_RING_RESET;
+
+	fw_shared->multi_queue.encode_generalpurpose_queue_mode |= FW_QUEUE_RING_RESET;
+	ring = &adev->vcn.inst[i].ring_enc[0];
+	WREG32_SOC15(VCN, i, mmUVD_RB_RPTR, lower_32_bits(ring->wptr));
+	WREG32_SOC15(VCN, i, mmUVD_RB_WPTR, lower_32_bits(ring->wptr));
+	WREG32_SOC15(VCN, i, mmUVD_RB_BASE_LO, ring->gpu_addr);
+	WREG32_SOC15(VCN, i, mmUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(VCN, i, mmUVD_RB_SIZE, ring->ring_size / 4);
+	fw_shared->multi_queue.encode_generalpurpose_queue_mode &= ~FW_QUEUE_RING_RESET;
+
+	fw_shared->multi_queue.encode_lowlatency_queue_mode |= FW_QUEUE_RING_RESET;
+	ring = &adev->vcn.inst[i].ring_enc[1];
+	WREG32_SOC15(VCN, i, mmUVD_RB_RPTR2, lower_32_bits(ring->wptr));
+	WREG32_SOC15(VCN, i, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
+	WREG32_SOC15(VCN, i, mmUVD_RB_BASE_LO2, ring->gpu_addr);
+	WREG32_SOC15(VCN, i, mmUVD_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(VCN, i, mmUVD_RB_SIZE2, ring->ring_size / 4);
+	fw_shared->multi_queue.encode_lowlatency_queue_mode &= ~FW_QUEUE_RING_RESET;
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, i, mmUVD_STATUS);
+
+	return 0;
+}
+
+static int vcn_v2_5_mmsch_start(struct amdgpu_device *adev,
+				struct amdgpu_mm_table *table)
+{
+	uint32_t data = 0, loop = 0, size = 0;
+	uint64_t addr = table->gpu_addr;
+	struct mmsch_v1_1_init_header *header = NULL;
+
+	header = (struct mmsch_v1_1_init_header *)table->cpu_addr;
+	size = header->total_size;
+
+	/*
+	 * 1, write to vce_mmsch_vf_ctx_addr_lo/hi register with GPU mc addr of
+	 *  memory descriptor location
+	 */
+	WREG32_SOC15(VCN, 0, mmMMSCH_VF_CTX_ADDR_LO, lower_32_bits(addr));
+	WREG32_SOC15(VCN, 0, mmMMSCH_VF_CTX_ADDR_HI, upper_32_bits(addr));
+
+	/* 2, update vmid of descriptor */
+	data = RREG32_SOC15(VCN, 0, mmMMSCH_VF_VMID);
+	data &= ~MMSCH_VF_VMID__VF_CTX_VMID_MASK;
+	/* use domain0 for MM scheduler */
+	data |= (0 << MMSCH_VF_VMID__VF_CTX_VMID__SHIFT);
+	WREG32_SOC15(VCN, 0, mmMMSCH_VF_VMID, data);
+
+	/* 3, notify mmsch about the size of this descriptor */
+	WREG32_SOC15(VCN, 0, mmMMSCH_VF_CTX_SIZE, size);
+
+	/* 4, set resp to zero */
+	WREG32_SOC15(VCN, 0, mmMMSCH_VF_MAILBOX_RESP, 0);
+
+	/*
+	 * 5, kick off the initialization and wait until
+	 * VCE_MMSCH_VF_MAILBOX_RESP becomes non-zero
+	 */
+	WREG32_SOC15(VCN, 0, mmMMSCH_VF_MAILBOX_HOST, 0x10000001);
+
+	data = RREG32_SOC15(VCN, 0, mmMMSCH_VF_MAILBOX_RESP);
+	loop = 10;
+	while ((data & 0x10000002) != 0x10000002) {
+		udelay(100);
+		data = RREG32_SOC15(VCN, 0, mmMMSCH_VF_MAILBOX_RESP);
+		loop--;
+		if (!loop)
+			break;
+	}
+
+	if (!loop) {
+		dev_err(adev->dev,
+			"failed to init MMSCH, mmMMSCH_VF_MAILBOX_RESP = %x\n",
+			data);
+		return -EBUSY;
+	}
+
+	return 0;
+}
+
+static int vcn_v2_5_sriov_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	uint32_t offset, size, tmp, i, rb_bufsz;
+	uint32_t table_size = 0;
+	struct mmsch_v1_0_cmd_direct_write direct_wt = { { 0 } };
+	struct mmsch_v1_0_cmd_direct_read_modify_write direct_rd_mod_wt = { { 0 } };
+	struct mmsch_v1_0_cmd_end end = { { 0 } };
+	uint32_t *init_table = adev->virt.mm_table.cpu_addr;
+	struct mmsch_v1_1_init_header *header = (struct mmsch_v1_1_init_header *)init_table;
+
+	direct_wt.cmd_header.command_type = MMSCH_COMMAND__DIRECT_REG_WRITE;
+	direct_rd_mod_wt.cmd_header.command_type = MMSCH_COMMAND__DIRECT_REG_READ_MODIFY_WRITE;
+	end.cmd_header.command_type = MMSCH_COMMAND__END;
+
+	header->version = MMSCH_VERSION;
+	header->total_size = sizeof(struct mmsch_v1_1_init_header) >> 2;
+	init_table += header->total_size;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		header->eng[i].table_offset = header->total_size;
+		header->eng[i].init_status = 0;
+		header->eng[i].table_size = 0;
+
+		table_size = 0;
+
+		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(
+			SOC15_REG_OFFSET(VCN, i, mmUVD_STATUS),
+			~UVD_STATUS__UVD_BUSY, UVD_STATUS__UVD_BUSY);
+
+		size = AMDGPU_GPU_PAGE_ALIGN(adev->vcn.inst[i].fw->size + 4);
+		/* mc resume*/
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+			MMSCH_V1_0_INSERT_DIRECT_WT(
+				SOC15_REG_OFFSET(VCN, i,
+					mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+				adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + i].tmr_mc_addr_lo);
+			MMSCH_V1_0_INSERT_DIRECT_WT(
+				SOC15_REG_OFFSET(VCN, i,
+					mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+				adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + i].tmr_mc_addr_hi);
+			offset = 0;
+			MMSCH_V1_0_INSERT_DIRECT_WT(
+				SOC15_REG_OFFSET(VCN, i, mmUVD_VCPU_CACHE_OFFSET0), 0);
+		} else {
+			MMSCH_V1_0_INSERT_DIRECT_WT(
+				SOC15_REG_OFFSET(VCN, i,
+					mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+				lower_32_bits(adev->vcn.inst[i].gpu_addr));
+			MMSCH_V1_0_INSERT_DIRECT_WT(
+				SOC15_REG_OFFSET(VCN, i,
+					mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+				upper_32_bits(adev->vcn.inst[i].gpu_addr));
+			offset = size;
+			MMSCH_V1_0_INSERT_DIRECT_WT(
+				SOC15_REG_OFFSET(VCN, i, mmUVD_VCPU_CACHE_OFFSET0),
+				AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
+		}
+
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(VCN, i, mmUVD_VCPU_CACHE_SIZE0),
+			size);
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(VCN, i,
+				mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[i].gpu_addr + offset));
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(VCN, i,
+				mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[i].gpu_addr + offset));
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(VCN, i, mmUVD_VCPU_CACHE_OFFSET1),
+			0);
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(VCN, i, mmUVD_VCPU_CACHE_SIZE1),
+			AMDGPU_VCN_STACK_SIZE);
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(VCN, i,
+				mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[i].gpu_addr + offset +
+				AMDGPU_VCN_STACK_SIZE));
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(VCN, i,
+				mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[i].gpu_addr + offset +
+				AMDGPU_VCN_STACK_SIZE));
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(VCN, i, mmUVD_VCPU_CACHE_OFFSET2),
+			0);
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(VCN, i, mmUVD_VCPU_CACHE_SIZE2),
+			AMDGPU_VCN_CONTEXT_SIZE);
+
+		ring = &adev->vcn.inst[i].ring_enc[0];
+		ring->wptr = 0;
+
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(VCN, i, mmUVD_RB_BASE_LO),
+			lower_32_bits(ring->gpu_addr));
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(VCN, i, mmUVD_RB_BASE_HI),
+			upper_32_bits(ring->gpu_addr));
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(VCN, i, mmUVD_RB_SIZE),
+			ring->ring_size / 4);
+
+		ring = &adev->vcn.inst[i].ring_dec;
+		ring->wptr = 0;
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(VCN, i,
+				mmUVD_LMI_RBC_RB_64BIT_BAR_LOW),
+			lower_32_bits(ring->gpu_addr));
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(VCN, i,
+				mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH),
+			upper_32_bits(ring->gpu_addr));
+
+		/* force RBC into idle state */
+		rb_bufsz = order_base_2(ring->ring_size);
+		tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(VCN, i, mmUVD_RBC_RB_CNTL), tmp);
+
+		/* add end packet */
+		memcpy((void *)init_table, &end, sizeof(struct mmsch_v1_0_cmd_end));
+		table_size += sizeof(struct mmsch_v1_0_cmd_end) / 4;
+		init_table += sizeof(struct mmsch_v1_0_cmd_end) / 4;
+
+		/* refine header */
+		header->eng[i].table_size = table_size;
+		header->total_size += table_size;
+	}
+
+	return vcn_v2_5_mmsch_start(adev, &adev->virt.mm_table);
+}
+
+static int vcn_v2_5_stop_dpg_mode(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t tmp;
+
+	/* Wait for power status to be 1 */
+	SOC15_WAIT_ON_RREG(VCN, inst_idx, mmUVD_POWER_STATUS, 1,
+		UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+	/* wait for read ptr to be equal to write ptr */
+	tmp = RREG32_SOC15(VCN, inst_idx, mmUVD_RB_WPTR);
+	SOC15_WAIT_ON_RREG(VCN, inst_idx, mmUVD_RB_RPTR, tmp, 0xFFFFFFFF);
+
+	tmp = RREG32_SOC15(VCN, inst_idx, mmUVD_RB_WPTR2);
+	SOC15_WAIT_ON_RREG(VCN, inst_idx, mmUVD_RB_RPTR2, tmp, 0xFFFFFFFF);
+
+	tmp = RREG32_SOC15(VCN, inst_idx, mmUVD_RBC_RB_WPTR) & 0x7FFFFFFF;
+	SOC15_WAIT_ON_RREG(VCN, inst_idx, mmUVD_RBC_RB_RPTR, tmp, 0xFFFFFFFF);
+
+	SOC15_WAIT_ON_RREG(VCN, inst_idx, mmUVD_POWER_STATUS, 1,
+		UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+	/* disable dynamic power gating mode */
+	WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, mmUVD_POWER_STATUS), 0,
+			~UVD_POWER_STATUS__UVD_PG_MODE_MASK);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, inst_idx, mmUVD_STATUS);
+
+	return 0;
+}
+
+static int vcn_v2_5_stop(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int i = vinst->inst;
+	uint32_t tmp;
+	int r;
+
+	if (adev->vcn.harvest_config & (1 << i))
+		return 0;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) {
+		r = vcn_v2_5_stop_dpg_mode(vinst);
+		goto done;
+	}
+
+	/* wait for vcn idle */
+	r = SOC15_WAIT_ON_RREG(VCN, i, mmUVD_STATUS, UVD_STATUS__IDLE, 0x7);
+	if (r)
+		goto done;
+
+	tmp = UVD_LMI_STATUS__VCPU_LMI_WRITE_CLEAN_MASK |
+		UVD_LMI_STATUS__READ_CLEAN_MASK |
+		UVD_LMI_STATUS__WRITE_CLEAN_MASK |
+		UVD_LMI_STATUS__WRITE_CLEAN_RAW_MASK;
+	r = SOC15_WAIT_ON_RREG(VCN, i, mmUVD_LMI_STATUS, tmp, tmp);
+	if (r)
+		goto done;
+
+	/* block LMI UMC channel */
+	tmp = RREG32_SOC15(VCN, i, mmUVD_LMI_CTRL2);
+	tmp |= UVD_LMI_CTRL2__STALL_ARB_UMC_MASK;
+	WREG32_SOC15(VCN, i, mmUVD_LMI_CTRL2, tmp);
+
+	tmp = UVD_LMI_STATUS__UMC_READ_CLEAN_RAW_MASK|
+		UVD_LMI_STATUS__UMC_WRITE_CLEAN_RAW_MASK;
+	r = SOC15_WAIT_ON_RREG(VCN, i, mmUVD_LMI_STATUS, tmp, tmp);
+	if (r)
+		goto done;
+
+	/* block VCPU register access */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_RB_ARB_CTRL),
+		 UVD_RB_ARB_CTRL__VCPU_DIS_MASK,
+		 ~UVD_RB_ARB_CTRL__VCPU_DIS_MASK);
+
+	/* reset VCPU */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_VCPU_CNTL),
+		 UVD_VCPU_CNTL__BLK_RST_MASK,
+		 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+	/* disable VCPU clock */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_VCPU_CNTL), 0,
+		 ~(UVD_VCPU_CNTL__CLK_EN_MASK));
+
+	/* clear status */
+	WREG32_SOC15(VCN, i, mmUVD_STATUS, 0);
+
+	vcn_v2_5_enable_clock_gating(vinst);
+
+	/* enable register anti-hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_POWER_STATUS),
+		 UVD_POWER_STATUS__UVD_POWER_STATUS_MASK,
+		 ~UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, i, mmUVD_STATUS);
+done:
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_vcn(adev, false, i);
+
+	return r;
+}
+
+static int vcn_v2_5_pause_dpg_mode(struct amdgpu_vcn_inst *vinst,
+				   struct dpg_pause_state *new_state)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	struct amdgpu_ring *ring;
+	uint32_t reg_data = 0;
+	int ret_code = 0;
+
+	/* pause/unpause if state is changed */
+	if (adev->vcn.inst[inst_idx].pause_state.fw_based != new_state->fw_based) {
+		DRM_DEBUG("dpg pause state changed %d -> %d",
+			adev->vcn.inst[inst_idx].pause_state.fw_based,	new_state->fw_based);
+		reg_data = RREG32_SOC15(VCN, inst_idx, mmUVD_DPG_PAUSE) &
+			(~UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK);
+
+		if (new_state->fw_based == VCN_DPG_STATE__PAUSE) {
+			ret_code = SOC15_WAIT_ON_RREG(VCN, inst_idx, mmUVD_POWER_STATUS, 0x1,
+				UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+			if (!ret_code) {
+				struct amdgpu_fw_shared *fw_shared = adev->vcn.inst[inst_idx].fw_shared.cpu_addr;
+
+				/* pause DPG */
+				reg_data |= UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
+				WREG32_SOC15(VCN, inst_idx, mmUVD_DPG_PAUSE, reg_data);
+
+				/* wait for ACK */
+				SOC15_WAIT_ON_RREG(VCN, inst_idx, mmUVD_DPG_PAUSE,
+					   UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK,
+					   UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK);
+
+				/* Stall DPG before WPTR/RPTR reset */
+				WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, mmUVD_POWER_STATUS),
+					   UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK,
+					   ~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK);
+
+				/* Restore */
+				fw_shared->multi_queue.encode_generalpurpose_queue_mode |= FW_QUEUE_RING_RESET;
+				ring = &adev->vcn.inst[inst_idx].ring_enc[0];
+				ring->wptr = 0;
+				WREG32_SOC15(VCN, inst_idx, mmUVD_RB_BASE_LO, ring->gpu_addr);
+				WREG32_SOC15(VCN, inst_idx, mmUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+				WREG32_SOC15(VCN, inst_idx, mmUVD_RB_SIZE, ring->ring_size / 4);
+				WREG32_SOC15(VCN, inst_idx, mmUVD_RB_RPTR, lower_32_bits(ring->wptr));
+				WREG32_SOC15(VCN, inst_idx, mmUVD_RB_WPTR, lower_32_bits(ring->wptr));
+				fw_shared->multi_queue.encode_generalpurpose_queue_mode &= ~FW_QUEUE_RING_RESET;
+
+				fw_shared->multi_queue.encode_lowlatency_queue_mode |= FW_QUEUE_RING_RESET;
+				ring = &adev->vcn.inst[inst_idx].ring_enc[1];
+				ring->wptr = 0;
+				WREG32_SOC15(VCN, inst_idx, mmUVD_RB_BASE_LO2, ring->gpu_addr);
+				WREG32_SOC15(VCN, inst_idx, mmUVD_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
+				WREG32_SOC15(VCN, inst_idx, mmUVD_RB_SIZE2, ring->ring_size / 4);
+				WREG32_SOC15(VCN, inst_idx, mmUVD_RB_RPTR2, lower_32_bits(ring->wptr));
+				WREG32_SOC15(VCN, inst_idx, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
+				fw_shared->multi_queue.encode_lowlatency_queue_mode &= ~FW_QUEUE_RING_RESET;
+
+				/* Unstall DPG */
+				WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, mmUVD_POWER_STATUS),
+					   0, ~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK);
+
+				SOC15_WAIT_ON_RREG(VCN, inst_idx, mmUVD_POWER_STATUS,
+					   UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON, UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+			}
+		} else {
+			reg_data &= ~UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
+			WREG32_SOC15(VCN, inst_idx, mmUVD_DPG_PAUSE, reg_data);
+			SOC15_WAIT_ON_RREG(VCN, inst_idx, mmUVD_POWER_STATUS, 0x1,
+				UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+		}
+		adev->vcn.inst[inst_idx].pause_state.fw_based = new_state->fw_based;
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v2_5_dec_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t vcn_v2_5_dec_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32_SOC15(VCN, ring->me, mmUVD_RBC_RB_RPTR);
+}
+
+/**
+ * vcn_v2_5_dec_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t vcn_v2_5_dec_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell)
+		return *ring->wptr_cpu_addr;
+	else
+		return RREG32_SOC15(VCN, ring->me, mmUVD_RBC_RB_WPTR);
+}
+
+/**
+ * vcn_v2_5_dec_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void vcn_v2_5_dec_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
+		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+	} else {
+		WREG32_SOC15(VCN, ring->me, mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr));
+	}
+}
+
+static const struct amdgpu_ring_funcs vcn_v2_5_dec_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_DEC,
+	.align_mask = 0xf,
+	.secure_submission_supported = true,
+	.get_rptr = vcn_v2_5_dec_ring_get_rptr,
+	.get_wptr = vcn_v2_5_dec_ring_get_wptr,
+	.set_wptr = vcn_v2_5_dec_ring_set_wptr,
+	.emit_frame_size =
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
+		8 + /* vcn_v2_0_dec_ring_emit_vm_flush */
+		14 + 14 + /* vcn_v2_0_dec_ring_emit_fence x2 vm fence */
+		6,
+	.emit_ib_size = 8, /* vcn_v2_0_dec_ring_emit_ib */
+	.emit_ib = vcn_v2_0_dec_ring_emit_ib,
+	.emit_fence = vcn_v2_0_dec_ring_emit_fence,
+	.emit_vm_flush = vcn_v2_0_dec_ring_emit_vm_flush,
+	.test_ring = vcn_v2_0_dec_ring_test_ring,
+	.test_ib = amdgpu_vcn_dec_ring_test_ib,
+	.insert_nop = vcn_v2_0_dec_ring_insert_nop,
+	.insert_start = vcn_v2_0_dec_ring_insert_start,
+	.insert_end = vcn_v2_0_dec_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = vcn_v2_5_ring_begin_use,
+	.end_use = vcn_v2_5_ring_end_use,
+	.emit_wreg = vcn_v2_0_dec_ring_emit_wreg,
+	.emit_reg_wait = vcn_v2_0_dec_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = amdgpu_vcn_ring_reset,
+};
+
+/**
+ * vcn_v2_5_enc_ring_get_rptr - get enc read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware enc read pointer
+ */
+static uint64_t vcn_v2_5_enc_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->vcn.inst[ring->me].ring_enc[0])
+		return RREG32_SOC15(VCN, ring->me, mmUVD_RB_RPTR);
+	else
+		return RREG32_SOC15(VCN, ring->me, mmUVD_RB_RPTR2);
+}
+
+/**
+ * vcn_v2_5_enc_ring_get_wptr - get enc write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware enc write pointer
+ */
+static uint64_t vcn_v2_5_enc_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->vcn.inst[ring->me].ring_enc[0]) {
+		if (ring->use_doorbell)
+			return *ring->wptr_cpu_addr;
+		else
+			return RREG32_SOC15(VCN, ring->me, mmUVD_RB_WPTR);
+	} else {
+		if (ring->use_doorbell)
+			return *ring->wptr_cpu_addr;
+		else
+			return RREG32_SOC15(VCN, ring->me, mmUVD_RB_WPTR2);
+	}
+}
+
+/**
+ * vcn_v2_5_enc_ring_set_wptr - set enc write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the enc write pointer to the hardware
+ */
+static void vcn_v2_5_enc_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->vcn.inst[ring->me].ring_enc[0]) {
+		if (ring->use_doorbell) {
+			*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
+			WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+		} else {
+			WREG32_SOC15(VCN, ring->me, mmUVD_RB_WPTR, lower_32_bits(ring->wptr));
+		}
+	} else {
+		if (ring->use_doorbell) {
+			*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
+			WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+		} else {
+			WREG32_SOC15(VCN, ring->me, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
+		}
+	}
+}
+
+static const struct amdgpu_ring_funcs vcn_v2_5_enc_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_ENC,
+	.align_mask = 0x3f,
+	.nop = VCN_ENC_CMD_NO_OP,
+	.get_rptr = vcn_v2_5_enc_ring_get_rptr,
+	.get_wptr = vcn_v2_5_enc_ring_get_wptr,
+	.set_wptr = vcn_v2_5_enc_ring_set_wptr,
+	.emit_frame_size =
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 4 +
+		4 + /* vcn_v2_0_enc_ring_emit_vm_flush */
+		5 + 5 + /* vcn_v2_0_enc_ring_emit_fence x2 vm fence */
+		1, /* vcn_v2_0_enc_ring_insert_end */
+	.emit_ib_size = 5, /* vcn_v2_0_enc_ring_emit_ib */
+	.emit_ib = vcn_v2_0_enc_ring_emit_ib,
+	.emit_fence = vcn_v2_0_enc_ring_emit_fence,
+	.emit_vm_flush = vcn_v2_0_enc_ring_emit_vm_flush,
+	.test_ring = amdgpu_vcn_enc_ring_test_ring,
+	.test_ib = amdgpu_vcn_enc_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.insert_end = vcn_v2_0_enc_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = vcn_v2_5_ring_begin_use,
+	.end_use = vcn_v2_5_ring_end_use,
+	.emit_wreg = vcn_v2_0_enc_ring_emit_wreg,
+	.emit_reg_wait = vcn_v2_0_enc_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = amdgpu_vcn_ring_reset,
+};
+
+static void vcn_v2_5_set_dec_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+		adev->vcn.inst[i].ring_dec.funcs = &vcn_v2_5_dec_ring_vm_funcs;
+		adev->vcn.inst[i].ring_dec.me = i;
+	}
+}
+
+static void vcn_v2_5_set_enc_ring_funcs(struct amdgpu_device *adev)
+{
+	int i, j;
+
+	for (j = 0; j < adev->vcn.num_vcn_inst; ++j) {
+		if (adev->vcn.harvest_config & (1 << j))
+			continue;
+		for (i = 0; i < adev->vcn.inst[j].num_enc_rings; ++i) {
+			adev->vcn.inst[j].ring_enc[i].funcs = &vcn_v2_5_enc_ring_vm_funcs;
+			adev->vcn.inst[j].ring_enc[i].me = j;
+		}
+	}
+}
+
+static int vcn_v2_5_reset(struct amdgpu_vcn_inst *vinst)
+{
+	int r;
+
+	r = vcn_v2_5_stop(vinst);
+	if (r)
+		return r;
+	return vcn_v2_5_start(vinst);
+}
+
+static bool vcn_v2_5_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, ret = 1;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		ret &= (RREG32_SOC15(VCN, i, mmUVD_STATUS) == UVD_STATUS__IDLE);
+	}
+
+	return ret;
+}
+
+static int vcn_v2_5_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, ret = 0;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+		ret = SOC15_WAIT_ON_RREG(VCN, i, mmUVD_STATUS, UVD_STATUS__IDLE,
+			UVD_STATUS__IDLE);
+		if (ret)
+			return ret;
+	}
+
+	return ret;
+}
+
+static int vcn_v2_5_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = (state == AMD_CG_STATE_GATE);
+	int i;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[i];
+
+		if (enable) {
+			if (!vcn_v2_5_is_idle(ip_block))
+				return -EBUSY;
+			vcn_v2_5_enable_clock_gating(vinst);
+		} else {
+			vcn_v2_5_disable_clock_gating(vinst);
+		}
+	}
+
+	return 0;
+}
+
+static int vcn_v2_5_set_pg_state(struct amdgpu_vcn_inst *vinst,
+				 enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int ret;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	if (state == vinst->cur_state)
+		return 0;
+
+	if (state == AMD_PG_STATE_GATE)
+		ret = vcn_v2_5_stop(vinst);
+	else
+		ret = vcn_v2_5_start(vinst);
+
+	if (!ret)
+		vinst->cur_state = state;
+
+	return ret;
+}
+
+static int vcn_v2_5_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+static int vcn_v2_6_set_ras_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned int type,
+					enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+static int vcn_v2_5_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	uint32_t ip_instance;
+
+	switch (entry->client_id) {
+	case SOC15_IH_CLIENTID_VCN:
+		ip_instance = 0;
+		break;
+	case SOC15_IH_CLIENTID_VCN1:
+		ip_instance = 1;
+		break;
+	default:
+		DRM_ERROR("Unhandled client id: %d\n", entry->client_id);
+		return 0;
+	}
+
+	DRM_DEBUG("IH: VCN TRAP\n");
+
+	switch (entry->src_id) {
+	case VCN_2_0__SRCID__UVD_SYSTEM_MESSAGE_INTERRUPT:
+		amdgpu_fence_process(&adev->vcn.inst[ip_instance].ring_dec);
+		break;
+	case VCN_2_0__SRCID__UVD_ENC_GENERAL_PURPOSE:
+		amdgpu_fence_process(&adev->vcn.inst[ip_instance].ring_enc[0]);
+		break;
+	case VCN_2_0__SRCID__UVD_ENC_LOW_LATENCY:
+		amdgpu_fence_process(&adev->vcn.inst[ip_instance].ring_enc[1]);
+		break;
+	default:
+		DRM_ERROR("Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs vcn_v2_5_irq_funcs = {
+	.set = vcn_v2_5_set_interrupt_state,
+	.process = vcn_v2_5_process_interrupt,
+};
+
+static const struct amdgpu_irq_src_funcs vcn_v2_6_ras_irq_funcs = {
+	.set = vcn_v2_6_set_ras_interrupt_state,
+	.process = amdgpu_vcn_process_poison_irq,
+};
+
+static void vcn_v2_5_set_irq_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+		adev->vcn.inst[i].irq.num_types = adev->vcn.inst[i].num_enc_rings + 1;
+		adev->vcn.inst[i].irq.funcs = &vcn_v2_5_irq_funcs;
+
+		adev->vcn.inst[i].ras_poison_irq.num_types = adev->vcn.inst[i].num_enc_rings + 1;
+		adev->vcn.inst[i].ras_poison_irq.funcs = &vcn_v2_6_ras_irq_funcs;
+	}
+}
+
+static const struct amd_ip_funcs vcn_v2_5_ip_funcs = {
+	.name = "vcn_v2_5",
+	.early_init = vcn_v2_5_early_init,
+	.sw_init = vcn_v2_5_sw_init,
+	.sw_fini = vcn_v2_5_sw_fini,
+	.hw_init = vcn_v2_5_hw_init,
+	.hw_fini = vcn_v2_5_hw_fini,
+	.suspend = vcn_v2_5_suspend,
+	.resume = vcn_v2_5_resume,
+	.is_idle = vcn_v2_5_is_idle,
+	.wait_for_idle = vcn_v2_5_wait_for_idle,
+	.set_clockgating_state = vcn_v2_5_set_clockgating_state,
+	.set_powergating_state = vcn_set_powergating_state,
+	.dump_ip_state = amdgpu_vcn_dump_ip_state,
+	.print_ip_state = amdgpu_vcn_print_ip_state,
+};
+
+static const struct amd_ip_funcs vcn_v2_6_ip_funcs = {
+        .name = "vcn_v2_6",
+        .early_init = vcn_v2_5_early_init,
+        .sw_init = vcn_v2_5_sw_init,
+        .sw_fini = vcn_v2_5_sw_fini,
+        .hw_init = vcn_v2_5_hw_init,
+        .hw_fini = vcn_v2_5_hw_fini,
+        .suspend = vcn_v2_5_suspend,
+        .resume = vcn_v2_5_resume,
+        .is_idle = vcn_v2_5_is_idle,
+        .wait_for_idle = vcn_v2_5_wait_for_idle,
+        .set_clockgating_state = vcn_v2_5_set_clockgating_state,
+	.set_powergating_state = vcn_set_powergating_state,
+	.dump_ip_state = amdgpu_vcn_dump_ip_state,
+	.print_ip_state = amdgpu_vcn_print_ip_state,
+};
+
+const struct amdgpu_ip_block_version vcn_v2_5_ip_block =
+{
+		.type = AMD_IP_BLOCK_TYPE_VCN,
+		.major = 2,
+		.minor = 5,
+		.rev = 0,
+		.funcs = &vcn_v2_5_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version vcn_v2_6_ip_block =
+{
+		.type = AMD_IP_BLOCK_TYPE_VCN,
+		.major = 2,
+		.minor = 6,
+		.rev = 0,
+		.funcs = &vcn_v2_6_ip_funcs,
+};
+
+static uint32_t vcn_v2_6_query_poison_by_instance(struct amdgpu_device *adev,
+			uint32_t instance, uint32_t sub_block)
+{
+	uint32_t poison_stat = 0, reg_value = 0;
+
+	switch (sub_block) {
+	case AMDGPU_VCN_V2_6_VCPU_VCODEC:
+		reg_value = RREG32_SOC15(VCN, instance, mmUVD_RAS_VCPU_VCODEC_STATUS);
+		poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_VCPU_VCODEC_STATUS, POISONED_PF);
+		break;
+	default:
+		break;
+	}
+
+	if (poison_stat)
+		dev_info(adev->dev, "Poison detected in VCN%d, sub_block%d\n",
+			instance, sub_block);
+
+	return poison_stat;
+}
+
+static bool vcn_v2_6_query_poison_status(struct amdgpu_device *adev)
+{
+	uint32_t inst, sub;
+	uint32_t poison_stat = 0;
+
+	for (inst = 0; inst < adev->vcn.num_vcn_inst; inst++)
+		for (sub = 0; sub < AMDGPU_VCN_V2_6_MAX_SUB_BLOCK; sub++)
+			poison_stat +=
+			vcn_v2_6_query_poison_by_instance(adev, inst, sub);
+
+	return !!poison_stat;
+}
+
+const struct amdgpu_ras_block_hw_ops vcn_v2_6_ras_hw_ops = {
+	.query_poison_status = vcn_v2_6_query_poison_status,
+};
+
+static struct amdgpu_vcn_ras vcn_v2_6_ras = {
+	.ras_block = {
+		.hw_ops = &vcn_v2_6_ras_hw_ops,
+		.ras_late_init = amdgpu_vcn_ras_late_init,
+	},
+};
+
+static void vcn_v2_5_set_ras_funcs(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, VCN_HWIP, 0)) {
+	case IP_VERSION(2, 6, 0):
+		adev->vcn.ras = &vcn_v2_6_ras;
+		break;
+	default:
+		break;
+	}
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.h b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.h
new file mode 100644
index 000000000000..1c19af74e4fd
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.h
@@ -0,0 +1,36 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VCN_V2_5_H__
+#define __VCN_V2_5_H__
+
+enum amdgpu_vcn_v2_6_sub_block {
+	AMDGPU_VCN_V2_6_VCPU_VCODEC = 0,
+
+	AMDGPU_VCN_V2_6_MAX_SUB_BLOCK,
+};
+
+extern const struct amdgpu_ip_block_version vcn_v2_5_ip_block;
+extern const struct amdgpu_ip_block_version vcn_v2_6_ip_block;
+
+#endif /* __VCN_V2_5_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c
new file mode 100644
index 000000000000..d9cf8f0feeb3
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c
@@ -0,0 +1,2368 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include "amdgpu.h"
+#include "amdgpu_vcn.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_cs.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "vcn_v2_0.h"
+#include "mmsch_v3_0.h"
+#include "vcn_sw_ring.h"
+
+#include "vcn/vcn_3_0_0_offset.h"
+#include "vcn/vcn_3_0_0_sh_mask.h"
+#include "ivsrcid/vcn/irqsrcs_vcn_2_0.h"
+
+#include <drm/drm_drv.h>
+
+#define VCN_VID_SOC_ADDRESS_2_0					0x1fa00
+#define VCN1_VID_SOC_ADDRESS_3_0				0x48200
+#define VCN1_AON_SOC_ADDRESS_3_0				0x48000
+
+#define mmUVD_CONTEXT_ID_INTERNAL_OFFSET			0x27
+#define mmUVD_GPCOM_VCPU_CMD_INTERNAL_OFFSET			0x0f
+#define mmUVD_GPCOM_VCPU_DATA0_INTERNAL_OFFSET			0x10
+#define mmUVD_GPCOM_VCPU_DATA1_INTERNAL_OFFSET			0x11
+#define mmUVD_NO_OP_INTERNAL_OFFSET				0x29
+#define mmUVD_GP_SCRATCH8_INTERNAL_OFFSET			0x66
+#define mmUVD_SCRATCH9_INTERNAL_OFFSET				0xc01d
+
+#define mmUVD_LMI_RBC_IB_VMID_INTERNAL_OFFSET			0x431
+#define mmUVD_LMI_RBC_IB_64BIT_BAR_LOW_INTERNAL_OFFSET		0x3b4
+#define mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH_INTERNAL_OFFSET		0x3b5
+#define mmUVD_RBC_IB_SIZE_INTERNAL_OFFSET			0x25c
+
+#define VCN_INSTANCES_SIENNA_CICHLID				2
+#define DEC_SW_RING_ENABLED					FALSE
+
+#define RDECODE_MSG_CREATE					0x00000000
+#define RDECODE_MESSAGE_CREATE					0x00000001
+
+static const struct amdgpu_hwip_reg_entry vcn_reg_list_3_0[] = {
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_CONTEXT_ID),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_CONTEXT_ID2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_GPCOM_VCPU_DATA0),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_GPCOM_VCPU_DATA1),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_GPCOM_VCPU_CMD),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_HI),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_LO),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_HI2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_LO2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_HI3),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_LO3),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_HI4),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_BASE_LO4),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_RPTR),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_WPTR),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_RPTR2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_WPTR2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_RPTR3),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_WPTR3),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_RPTR4),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_WPTR4),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_SIZE),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_SIZE2),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_SIZE3),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_RB_SIZE4),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_PGFSM_CONFIG),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_PGFSM_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_DPG_LMA_CTL),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_DPG_LMA_DATA),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_DPG_LMA_MASK),
+	SOC15_REG_ENTRY_STR(VCN, 0, mmUVD_DPG_PAUSE)
+};
+
+static int amdgpu_ih_clientid_vcns[] = {
+	SOC15_IH_CLIENTID_VCN,
+	SOC15_IH_CLIENTID_VCN1
+};
+
+static int vcn_v3_0_start_sriov(struct amdgpu_device *adev);
+static void vcn_v3_0_set_dec_ring_funcs(struct amdgpu_device *adev);
+static void vcn_v3_0_set_enc_ring_funcs(struct amdgpu_device *adev);
+static void vcn_v3_0_set_irq_funcs(struct amdgpu_device *adev);
+static int vcn_v3_0_set_pg_state(struct amdgpu_vcn_inst *vinst,
+				 enum amd_powergating_state state);
+static int vcn_v3_0_pause_dpg_mode(struct amdgpu_vcn_inst *vinst,
+				   struct dpg_pause_state *new_state);
+static int vcn_v3_0_reset(struct amdgpu_vcn_inst *vinst);
+
+static void vcn_v3_0_dec_ring_set_wptr(struct amdgpu_ring *ring);
+static void vcn_v3_0_enc_ring_set_wptr(struct amdgpu_ring *ring);
+
+/**
+ * vcn_v3_0_early_init - set function pointers and load microcode
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Set ring and irq function pointers
+ * Load microcode from filesystem
+ */
+static int vcn_v3_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, r;
+
+	if (amdgpu_sriov_vf(adev)) {
+		adev->vcn.num_vcn_inst = VCN_INSTANCES_SIENNA_CICHLID;
+		adev->vcn.harvest_config = 0;
+		for (i = 0; i < adev->vcn.num_vcn_inst; i++)
+			adev->vcn.inst[i].num_enc_rings = 1;
+
+	} else {
+		if (adev->vcn.harvest_config == (AMDGPU_VCN_HARVEST_VCN0 |
+						 AMDGPU_VCN_HARVEST_VCN1))
+			/* both instances are harvested, disable the block */
+			return -ENOENT;
+
+		for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+			if (amdgpu_ip_version(adev, UVD_HWIP, 0) ==
+			    IP_VERSION(3, 0, 33))
+				adev->vcn.inst[i].num_enc_rings = 0;
+			else
+				adev->vcn.inst[i].num_enc_rings = 2;
+		}
+	}
+
+	vcn_v3_0_set_dec_ring_funcs(adev);
+	vcn_v3_0_set_enc_ring_funcs(adev);
+	vcn_v3_0_set_irq_funcs(adev);
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		adev->vcn.inst[i].set_pg_state = vcn_v3_0_set_pg_state;
+
+		r = amdgpu_vcn_early_init(adev, i);
+		if (r)
+			return r;
+	}
+	return 0;
+}
+
+/**
+ * vcn_v3_0_sw_init - sw init for VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Load firmware and sw initialization
+ */
+static int vcn_v3_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_ring *ring;
+	int i, j, r;
+	int vcn_doorbell_index = 0;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	/*
+	 * Note: doorbell assignment is fixed for SRIOV multiple VCN engines
+	 * Formula:
+	 *   vcn_db_base  = adev->doorbell_index.vcn.vcn_ring0_1 << 1;
+	 *   dec_ring_i   = vcn_db_base + i * (adev->vcn.num_enc_rings + 1)
+	 *   enc_ring_i,j = vcn_db_base + i * (adev->vcn.num_enc_rings + 1) + 1 + j
+	 */
+	if (amdgpu_sriov_vf(adev)) {
+		vcn_doorbell_index = adev->doorbell_index.vcn.vcn_ring0_1;
+		/* get DWORD offset */
+		vcn_doorbell_index = vcn_doorbell_index << 1;
+	}
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		struct amdgpu_fw_shared *fw_shared;
+
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		r = amdgpu_vcn_sw_init(adev, i);
+		if (r)
+			return r;
+
+		amdgpu_vcn_setup_ucode(adev, i);
+
+		r = amdgpu_vcn_resume(adev, i);
+		if (r)
+			return r;
+
+		adev->vcn.inst[i].internal.context_id = mmUVD_CONTEXT_ID_INTERNAL_OFFSET;
+		adev->vcn.inst[i].internal.ib_vmid = mmUVD_LMI_RBC_IB_VMID_INTERNAL_OFFSET;
+		adev->vcn.inst[i].internal.ib_bar_low = mmUVD_LMI_RBC_IB_64BIT_BAR_LOW_INTERNAL_OFFSET;
+		adev->vcn.inst[i].internal.ib_bar_high = mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH_INTERNAL_OFFSET;
+		adev->vcn.inst[i].internal.ib_size = mmUVD_RBC_IB_SIZE_INTERNAL_OFFSET;
+		adev->vcn.inst[i].internal.gp_scratch8 = mmUVD_GP_SCRATCH8_INTERNAL_OFFSET;
+
+		adev->vcn.inst[i].internal.scratch9 = mmUVD_SCRATCH9_INTERNAL_OFFSET;
+		adev->vcn.inst[i].external.scratch9 = SOC15_REG_OFFSET(VCN, i, mmUVD_SCRATCH9);
+		adev->vcn.inst[i].internal.data0 = mmUVD_GPCOM_VCPU_DATA0_INTERNAL_OFFSET;
+		adev->vcn.inst[i].external.data0 = SOC15_REG_OFFSET(VCN, i, mmUVD_GPCOM_VCPU_DATA0);
+		adev->vcn.inst[i].internal.data1 = mmUVD_GPCOM_VCPU_DATA1_INTERNAL_OFFSET;
+		adev->vcn.inst[i].external.data1 = SOC15_REG_OFFSET(VCN, i, mmUVD_GPCOM_VCPU_DATA1);
+		adev->vcn.inst[i].internal.cmd = mmUVD_GPCOM_VCPU_CMD_INTERNAL_OFFSET;
+		adev->vcn.inst[i].external.cmd = SOC15_REG_OFFSET(VCN, i, mmUVD_GPCOM_VCPU_CMD);
+		adev->vcn.inst[i].internal.nop = mmUVD_NO_OP_INTERNAL_OFFSET;
+		adev->vcn.inst[i].external.nop = SOC15_REG_OFFSET(VCN, i, mmUVD_NO_OP);
+
+		/* VCN DEC TRAP */
+		r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_vcns[i],
+				VCN_2_0__SRCID__UVD_SYSTEM_MESSAGE_INTERRUPT, &adev->vcn.inst[i].irq);
+		if (r)
+			return r;
+
+		atomic_set(&adev->vcn.inst[i].sched_score, 0);
+
+		ring = &adev->vcn.inst[i].ring_dec;
+		ring->use_doorbell = true;
+		if (amdgpu_sriov_vf(adev)) {
+			ring->doorbell_index = vcn_doorbell_index + i * (adev->vcn.inst[i].num_enc_rings + 1);
+		} else {
+			ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 8 * i;
+		}
+		ring->vm_hub = AMDGPU_MMHUB0(0);
+		sprintf(ring->name, "vcn_dec_%d", i);
+		r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst[i].irq, 0,
+				     AMDGPU_RING_PRIO_DEFAULT,
+				     &adev->vcn.inst[i].sched_score);
+		if (r)
+			return r;
+
+		for (j = 0; j < adev->vcn.inst[i].num_enc_rings; ++j) {
+			enum amdgpu_ring_priority_level hw_prio = amdgpu_vcn_get_enc_ring_prio(j);
+
+			/* VCN ENC TRAP */
+			r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_vcns[i],
+				j + VCN_2_0__SRCID__UVD_ENC_GENERAL_PURPOSE, &adev->vcn.inst[i].irq);
+			if (r)
+				return r;
+
+			ring = &adev->vcn.inst[i].ring_enc[j];
+			ring->use_doorbell = true;
+			if (amdgpu_sriov_vf(adev)) {
+				ring->doorbell_index = vcn_doorbell_index + i * (adev->vcn.inst[i].num_enc_rings + 1) + 1 + j;
+			} else {
+				ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 2 + j + 8 * i;
+			}
+			ring->vm_hub = AMDGPU_MMHUB0(0);
+			sprintf(ring->name, "vcn_enc_%d.%d", i, j);
+			r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst[i].irq, 0,
+					     hw_prio, &adev->vcn.inst[i].sched_score);
+			if (r)
+				return r;
+		}
+
+		fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+		fw_shared->present_flag_0 |= cpu_to_le32(AMDGPU_VCN_SW_RING_FLAG) |
+					     cpu_to_le32(AMDGPU_VCN_MULTI_QUEUE_FLAG) |
+					     cpu_to_le32(AMDGPU_VCN_FW_SHARED_FLAG_0_RB);
+		fw_shared->sw_ring.is_enabled = cpu_to_le32(DEC_SW_RING_ENABLED);
+		fw_shared->present_flag_0 |= AMDGPU_VCN_SMU_VERSION_INFO_FLAG;
+		if (amdgpu_ip_version(adev, UVD_HWIP, 0) == IP_VERSION(3, 1, 2))
+			fw_shared->smu_interface_info.smu_interface_type = 2;
+		else if (amdgpu_ip_version(adev, UVD_HWIP, 0) ==
+			 IP_VERSION(3, 1, 1))
+			fw_shared->smu_interface_info.smu_interface_type = 1;
+
+		if (amdgpu_vcnfw_log)
+			amdgpu_vcn_fwlog_init(&adev->vcn.inst[i]);
+
+		if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)
+			adev->vcn.inst[i].pause_dpg_mode = vcn_v3_0_pause_dpg_mode;
+		adev->vcn.inst[i].reset = vcn_v3_0_reset;
+	}
+
+	adev->vcn.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->vcn.inst[0].ring_enc[0]);
+	if (!amdgpu_sriov_vf(adev))
+		adev->vcn.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+
+	if (amdgpu_sriov_vf(adev)) {
+		r = amdgpu_virt_alloc_mm_table(adev);
+		if (r)
+			return r;
+	}
+
+	r = amdgpu_vcn_reg_dump_init(adev, vcn_reg_list_3_0, ARRAY_SIZE(vcn_reg_list_3_0));
+	if (r)
+		return r;
+
+	r = amdgpu_vcn_sysfs_reset_mask_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+/**
+ * vcn_v3_0_sw_fini - sw fini for VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * VCN suspend and free up sw allocation
+ */
+static int vcn_v3_0_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, r, idx;
+
+	if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+		for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+			struct amdgpu_fw_shared *fw_shared;
+
+			if (adev->vcn.harvest_config & (1 << i))
+				continue;
+			fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+			fw_shared->present_flag_0 = 0;
+			fw_shared->sw_ring.is_enabled = false;
+		}
+
+		drm_dev_exit(idx);
+	}
+
+	if (amdgpu_sriov_vf(adev))
+		amdgpu_virt_free_mm_table(adev);
+
+	amdgpu_vcn_sysfs_reset_mask_fini(adev);
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		r = amdgpu_vcn_suspend(adev, i);
+		if (r)
+			return r;
+
+		amdgpu_vcn_sw_fini(adev, i);
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v3_0_hw_init - start and test VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Initialize the hardware, boot up the VCPU and do some testing
+ */
+static int vcn_v3_0_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	int i, j, r;
+
+	if (amdgpu_sriov_vf(adev)) {
+		r = vcn_v3_0_start_sriov(adev);
+		if (r)
+			return r;
+
+		/* initialize VCN dec and enc ring buffers */
+		for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+			if (adev->vcn.harvest_config & (1 << i))
+				continue;
+
+			ring = &adev->vcn.inst[i].ring_dec;
+			if (amdgpu_vcn_is_disabled_vcn(adev, VCN_DECODE_RING, i)) {
+				ring->sched.ready = false;
+				ring->no_scheduler = true;
+				dev_info(adev->dev, "ring %s is disabled by hypervisor\n", ring->name);
+			} else {
+				ring->wptr = 0;
+				ring->wptr_old = 0;
+				vcn_v3_0_dec_ring_set_wptr(ring);
+				ring->sched.ready = true;
+			}
+
+			for (j = 0; j < adev->vcn.inst[i].num_enc_rings; ++j) {
+				ring = &adev->vcn.inst[i].ring_enc[j];
+				if (amdgpu_vcn_is_disabled_vcn(adev, VCN_ENCODE_RING, i)) {
+					ring->sched.ready = false;
+					ring->no_scheduler = true;
+					dev_info(adev->dev, "ring %s is disabled by hypervisor\n", ring->name);
+				} else {
+					ring->wptr = 0;
+					ring->wptr_old = 0;
+					vcn_v3_0_enc_ring_set_wptr(ring);
+					ring->sched.ready = true;
+				}
+			}
+		}
+	} else {
+		for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+			if (adev->vcn.harvest_config & (1 << i))
+				continue;
+
+			ring = &adev->vcn.inst[i].ring_dec;
+
+			adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
+						     ring->doorbell_index, i);
+
+			r = amdgpu_ring_test_helper(ring);
+			if (r)
+				return r;
+
+			for (j = 0; j < adev->vcn.inst[i].num_enc_rings; ++j) {
+				ring = &adev->vcn.inst[i].ring_enc[j];
+				r = amdgpu_ring_test_helper(ring);
+				if (r)
+					return r;
+			}
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v3_0_hw_fini - stop the hardware block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Stop the VCN block, mark ring as not ready any more
+ */
+static int vcn_v3_0_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[i];
+
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		cancel_delayed_work_sync(&vinst->idle_work);
+
+		if (!amdgpu_sriov_vf(adev)) {
+			if ((adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) ||
+			    (vinst->cur_state != AMD_PG_STATE_GATE &&
+			     RREG32_SOC15(VCN, i, mmUVD_STATUS))) {
+				vinst->set_pg_state(vinst, AMD_PG_STATE_GATE);
+			}
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v3_0_suspend - suspend VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * HW fini and suspend VCN block
+ */
+static int vcn_v3_0_suspend(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r, i;
+
+	r = vcn_v3_0_hw_fini(ip_block);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		r = amdgpu_vcn_suspend(ip_block->adev, i);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v3_0_resume - resume VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Resume firmware and hw init VCN block
+ */
+static int vcn_v3_0_resume(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r, i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		r = amdgpu_vcn_resume(ip_block->adev, i);
+		if (r)
+			return r;
+	}
+
+	r = vcn_v3_0_hw_init(ip_block);
+
+	return r;
+}
+
+/**
+ * vcn_v3_0_mc_resume - memory controller programming
+ *
+ * @vinst: VCN instance
+ *
+ * Let the VCN memory controller know it's offsets
+ */
+static void vcn_v3_0_mc_resume(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst = vinst->inst;
+	uint32_t size = AMDGPU_GPU_PAGE_ALIGN(adev->vcn.inst[inst].fw->size + 4);
+	uint32_t offset;
+
+	/* cache window 0: fw */
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		WREG32_SOC15(VCN, inst, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst].tmr_mc_addr_lo));
+		WREG32_SOC15(VCN, inst, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst].tmr_mc_addr_hi));
+		WREG32_SOC15(VCN, inst, mmUVD_VCPU_CACHE_OFFSET0, 0);
+		offset = 0;
+	} else {
+		WREG32_SOC15(VCN, inst, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			lower_32_bits(adev->vcn.inst[inst].gpu_addr));
+		WREG32_SOC15(VCN, inst, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			upper_32_bits(adev->vcn.inst[inst].gpu_addr));
+		offset = size;
+		WREG32_SOC15(VCN, inst, mmUVD_VCPU_CACHE_OFFSET0,
+			AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
+	}
+	WREG32_SOC15(VCN, inst, mmUVD_VCPU_CACHE_SIZE0, size);
+
+	/* cache window 1: stack */
+	WREG32_SOC15(VCN, inst, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW,
+		lower_32_bits(adev->vcn.inst[inst].gpu_addr + offset));
+	WREG32_SOC15(VCN, inst, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH,
+		upper_32_bits(adev->vcn.inst[inst].gpu_addr + offset));
+	WREG32_SOC15(VCN, inst, mmUVD_VCPU_CACHE_OFFSET1, 0);
+	WREG32_SOC15(VCN, inst, mmUVD_VCPU_CACHE_SIZE1, AMDGPU_VCN_STACK_SIZE);
+
+	/* cache window 2: context */
+	WREG32_SOC15(VCN, inst, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW,
+		lower_32_bits(adev->vcn.inst[inst].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE));
+	WREG32_SOC15(VCN, inst, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH,
+		upper_32_bits(adev->vcn.inst[inst].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE));
+	WREG32_SOC15(VCN, inst, mmUVD_VCPU_CACHE_OFFSET2, 0);
+	WREG32_SOC15(VCN, inst, mmUVD_VCPU_CACHE_SIZE2, AMDGPU_VCN_CONTEXT_SIZE);
+
+	/* non-cache window */
+	WREG32_SOC15(VCN, inst, mmUVD_LMI_VCPU_NC0_64BIT_BAR_LOW,
+		lower_32_bits(adev->vcn.inst[inst].fw_shared.gpu_addr));
+	WREG32_SOC15(VCN, inst, mmUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH,
+		upper_32_bits(adev->vcn.inst[inst].fw_shared.gpu_addr));
+	WREG32_SOC15(VCN, inst, mmUVD_VCPU_NONCACHE_OFFSET0, 0);
+	WREG32_SOC15(VCN, inst, mmUVD_VCPU_NONCACHE_SIZE0,
+		AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_fw_shared)));
+}
+
+static void vcn_v3_0_mc_resume_dpg_mode(struct amdgpu_vcn_inst *vinst,
+					bool indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t size = AMDGPU_GPU_PAGE_ALIGN(adev->vcn.inst[inst_idx].fw->size + 4);
+	uint32_t offset;
+
+	/* cache window 0: fw */
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		if (!indirect) {
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, inst_idx, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+				(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst_idx].tmr_mc_addr_lo), 0, indirect);
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, inst_idx, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+				(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst_idx].tmr_mc_addr_hi), 0, indirect);
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, inst_idx, mmUVD_VCPU_CACHE_OFFSET0), 0, 0, indirect);
+		} else {
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, inst_idx, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW), 0, 0, indirect);
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, inst_idx, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH), 0, 0, indirect);
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, inst_idx, mmUVD_VCPU_CACHE_OFFSET0), 0, 0, indirect);
+		}
+		offset = 0;
+	} else {
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr), 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr), 0, indirect);
+		offset = size;
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, mmUVD_VCPU_CACHE_OFFSET0),
+			AMDGPU_UVD_FIRMWARE_OFFSET >> 3, 0, indirect);
+	}
+
+	if (!indirect)
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, mmUVD_VCPU_CACHE_SIZE0), size, 0, indirect);
+	else
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, mmUVD_VCPU_CACHE_SIZE0), 0, 0, indirect);
+
+	/* cache window 1: stack */
+	if (!indirect) {
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset), 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset), 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, mmUVD_VCPU_CACHE_OFFSET1), 0, 0, indirect);
+	} else {
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW), 0, 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH), 0, 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, mmUVD_VCPU_CACHE_OFFSET1), 0, 0, indirect);
+	}
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, mmUVD_VCPU_CACHE_SIZE1), AMDGPU_VCN_STACK_SIZE, 0, indirect);
+
+	/* cache window 2: context */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE), 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE), 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, mmUVD_VCPU_CACHE_OFFSET2), 0, 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, mmUVD_VCPU_CACHE_SIZE2), AMDGPU_VCN_CONTEXT_SIZE, 0, indirect);
+
+	/* non-cache window */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, mmUVD_LMI_VCPU_NC0_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[inst_idx].fw_shared.gpu_addr), 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, mmUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[inst_idx].fw_shared.gpu_addr), 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, mmUVD_VCPU_NONCACHE_OFFSET0), 0, 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, mmUVD_VCPU_NONCACHE_SIZE0),
+			AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_fw_shared)), 0, indirect);
+
+	/* VCN global tiling registers */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		UVD, inst_idx, mmUVD_GFX10_ADDR_CONFIG), adev->gfx.config.gb_addr_config, 0, indirect);
+}
+
+static void vcn_v3_0_disable_static_power_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst = vinst->inst;
+	uint32_t data = 0;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN) {
+		data = (1 << UVD_PGFSM_CONFIG__UVDM_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDU_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDF_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDC_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDB_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDIRL_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDLM_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTD_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTE_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDE_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDAB_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDATD_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDNA_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDNB_PWR_CONFIG__SHIFT);
+
+		WREG32_SOC15(VCN, inst, mmUVD_PGFSM_CONFIG, data);
+		SOC15_WAIT_ON_RREG(VCN, inst, mmUVD_PGFSM_STATUS,
+			UVD_PGFSM_STATUS__UVDM_UVDU_UVDLM_PWR_ON_3_0, 0x3F3FFFFF);
+	} else {
+		data = (1 << UVD_PGFSM_CONFIG__UVDM_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDU_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDF_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDC_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDB_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDIRL_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDLM_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDTD_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDTE_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDE_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDAB_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDATD_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDNA_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDNB_PWR_CONFIG__SHIFT);
+		WREG32_SOC15(VCN, inst, mmUVD_PGFSM_CONFIG, data);
+		SOC15_WAIT_ON_RREG(VCN, inst, mmUVD_PGFSM_STATUS, 0,  0x3F3FFFFF);
+	}
+
+	data = RREG32_SOC15(VCN, inst, mmUVD_POWER_STATUS);
+	data &= ~0x103;
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN)
+		data |= UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON |
+			UVD_POWER_STATUS__UVD_PG_EN_MASK;
+
+	WREG32_SOC15(VCN, inst, mmUVD_POWER_STATUS, data);
+}
+
+static void vcn_v3_0_enable_static_power_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst = vinst->inst;
+	uint32_t data;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN) {
+		/* Before power off, this indicator has to be turned on */
+		data = RREG32_SOC15(VCN, inst, mmUVD_POWER_STATUS);
+		data &= ~UVD_POWER_STATUS__UVD_POWER_STATUS_MASK;
+		data |= UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF;
+		WREG32_SOC15(VCN, inst, mmUVD_POWER_STATUS, data);
+
+		data = (2 << UVD_PGFSM_CONFIG__UVDM_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDU_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDF_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDC_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDB_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDIRL_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDLM_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTD_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTE_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDE_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDAB_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDATD_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDNA_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDNB_PWR_CONFIG__SHIFT);
+		WREG32_SOC15(VCN, inst, mmUVD_PGFSM_CONFIG, data);
+
+		data = (2 << UVD_PGFSM_STATUS__UVDM_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDU_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDF_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDC_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDB_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDIRL_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDLM_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDTD_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDTE_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDE_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDAB_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDATD_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDNA_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDNB_PWR_STATUS__SHIFT);
+		SOC15_WAIT_ON_RREG(VCN, inst, mmUVD_PGFSM_STATUS, data, 0x3F3FFFFF);
+	}
+}
+
+/**
+ * vcn_v3_0_disable_clock_gating - disable VCN clock gating
+ *
+ * @vinst: Pointer to the VCN instance structure
+ *
+ * Disable clock gating for VCN block
+ */
+static void vcn_v3_0_disable_clock_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst = vinst->inst;
+	uint32_t data;
+
+	/* VCN disable CGC */
+	data = RREG32_SOC15(VCN, inst, mmUVD_CGC_CTRL);
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		data |= 1 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		data &= ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK;
+	data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(VCN, inst, mmUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, inst, mmUVD_CGC_GATE);
+	data &= ~(UVD_CGC_GATE__SYS_MASK
+		| UVD_CGC_GATE__UDEC_MASK
+		| UVD_CGC_GATE__MPEG2_MASK
+		| UVD_CGC_GATE__REGS_MASK
+		| UVD_CGC_GATE__RBC_MASK
+		| UVD_CGC_GATE__LMI_MC_MASK
+		| UVD_CGC_GATE__LMI_UMC_MASK
+		| UVD_CGC_GATE__IDCT_MASK
+		| UVD_CGC_GATE__MPRD_MASK
+		| UVD_CGC_GATE__MPC_MASK
+		| UVD_CGC_GATE__LBSI_MASK
+		| UVD_CGC_GATE__LRBBM_MASK
+		| UVD_CGC_GATE__UDEC_RE_MASK
+		| UVD_CGC_GATE__UDEC_CM_MASK
+		| UVD_CGC_GATE__UDEC_IT_MASK
+		| UVD_CGC_GATE__UDEC_DB_MASK
+		| UVD_CGC_GATE__UDEC_MP_MASK
+		| UVD_CGC_GATE__WCB_MASK
+		| UVD_CGC_GATE__VCPU_MASK
+		| UVD_CGC_GATE__MMSCH_MASK);
+
+	WREG32_SOC15(VCN, inst, mmUVD_CGC_GATE, data);
+
+	SOC15_WAIT_ON_RREG(VCN, inst, mmUVD_CGC_GATE, 0,  0xFFFFFFFF);
+
+	data = RREG32_SOC15(VCN, inst, mmUVD_CGC_CTRL);
+	data &= ~(UVD_CGC_CTRL__UDEC_RE_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_CM_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_IT_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_DB_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_MP_MODE_MASK
+		| UVD_CGC_CTRL__SYS_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_MODE_MASK
+		| UVD_CGC_CTRL__MPEG2_MODE_MASK
+		| UVD_CGC_CTRL__REGS_MODE_MASK
+		| UVD_CGC_CTRL__RBC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_MC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_UMC_MODE_MASK
+		| UVD_CGC_CTRL__IDCT_MODE_MASK
+		| UVD_CGC_CTRL__MPRD_MODE_MASK
+		| UVD_CGC_CTRL__MPC_MODE_MASK
+		| UVD_CGC_CTRL__LBSI_MODE_MASK
+		| UVD_CGC_CTRL__LRBBM_MODE_MASK
+		| UVD_CGC_CTRL__WCB_MODE_MASK
+		| UVD_CGC_CTRL__VCPU_MODE_MASK
+		| UVD_CGC_CTRL__MMSCH_MODE_MASK);
+	WREG32_SOC15(VCN, inst, mmUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, inst, mmUVD_SUVD_CGC_GATE);
+	data |= (UVD_SUVD_CGC_GATE__SRE_MASK
+		| UVD_SUVD_CGC_GATE__SIT_MASK
+		| UVD_SUVD_CGC_GATE__SMP_MASK
+		| UVD_SUVD_CGC_GATE__SCM_MASK
+		| UVD_SUVD_CGC_GATE__SDB_MASK
+		| UVD_SUVD_CGC_GATE__SRE_H264_MASK
+		| UVD_SUVD_CGC_GATE__SRE_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SIT_H264_MASK
+		| UVD_SUVD_CGC_GATE__SIT_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SCM_H264_MASK
+		| UVD_SUVD_CGC_GATE__SCM_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SDB_H264_MASK
+		| UVD_SUVD_CGC_GATE__SDB_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SCLR_MASK
+		| UVD_SUVD_CGC_GATE__ENT_MASK
+		| UVD_SUVD_CGC_GATE__IME_MASK
+		| UVD_SUVD_CGC_GATE__SIT_HEVC_DEC_MASK
+		| UVD_SUVD_CGC_GATE__SIT_HEVC_ENC_MASK
+		| UVD_SUVD_CGC_GATE__SITE_MASK
+		| UVD_SUVD_CGC_GATE__SRE_VP9_MASK
+		| UVD_SUVD_CGC_GATE__SCM_VP9_MASK
+		| UVD_SUVD_CGC_GATE__SIT_VP9_DEC_MASK
+		| UVD_SUVD_CGC_GATE__SDB_VP9_MASK
+		| UVD_SUVD_CGC_GATE__IME_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__EFC_MASK
+		| UVD_SUVD_CGC_GATE__SAOE_MASK
+		| UVD_SUVD_CGC_GATE__SRE_AV1_MASK
+		| UVD_SUVD_CGC_GATE__FBC_PCLK_MASK
+		| UVD_SUVD_CGC_GATE__FBC_CCLK_MASK
+		| UVD_SUVD_CGC_GATE__SCM_AV1_MASK
+		| UVD_SUVD_CGC_GATE__SMPA_MASK);
+	WREG32_SOC15(VCN, inst, mmUVD_SUVD_CGC_GATE, data);
+
+	data = RREG32_SOC15(VCN, inst, mmUVD_SUVD_CGC_GATE2);
+	data |= (UVD_SUVD_CGC_GATE2__MPBE0_MASK
+		| UVD_SUVD_CGC_GATE2__MPBE1_MASK
+		| UVD_SUVD_CGC_GATE2__SIT_AV1_MASK
+		| UVD_SUVD_CGC_GATE2__SDB_AV1_MASK
+		| UVD_SUVD_CGC_GATE2__MPC1_MASK);
+	WREG32_SOC15(VCN, inst, mmUVD_SUVD_CGC_GATE2, data);
+
+	data = RREG32_SOC15(VCN, inst, mmUVD_SUVD_CGC_CTRL);
+	data &= ~(UVD_SUVD_CGC_CTRL__SRE_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SIT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SMP_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SCM_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SDB_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SCLR_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__ENT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__IME_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SITE_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__EFC_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SAOE_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SMPA_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__MPBE0_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__MPBE1_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SIT_AV1_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SDB_AV1_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__MPC1_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__FBC_PCLK_MASK
+		| UVD_SUVD_CGC_CTRL__FBC_CCLK_MASK);
+	WREG32_SOC15(VCN, inst, mmUVD_SUVD_CGC_CTRL, data);
+}
+
+static void vcn_v3_0_clock_gating_dpg_mode(struct amdgpu_vcn_inst *vinst,
+					   uint8_t sram_sel,
+					   uint8_t indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t reg_data = 0;
+
+	/* enable sw clock gating control */
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		reg_data = 1 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		reg_data = 0 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	reg_data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	reg_data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	reg_data &= ~(UVD_CGC_CTRL__UDEC_RE_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_CM_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_IT_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_DB_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_MP_MODE_MASK |
+		 UVD_CGC_CTRL__SYS_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_MODE_MASK |
+		 UVD_CGC_CTRL__MPEG2_MODE_MASK |
+		 UVD_CGC_CTRL__REGS_MODE_MASK |
+		 UVD_CGC_CTRL__RBC_MODE_MASK |
+		 UVD_CGC_CTRL__LMI_MC_MODE_MASK |
+		 UVD_CGC_CTRL__LMI_UMC_MODE_MASK |
+		 UVD_CGC_CTRL__IDCT_MODE_MASK |
+		 UVD_CGC_CTRL__MPRD_MODE_MASK |
+		 UVD_CGC_CTRL__MPC_MODE_MASK |
+		 UVD_CGC_CTRL__LBSI_MODE_MASK |
+		 UVD_CGC_CTRL__LRBBM_MODE_MASK |
+		 UVD_CGC_CTRL__WCB_MODE_MASK |
+		 UVD_CGC_CTRL__VCPU_MODE_MASK |
+		 UVD_CGC_CTRL__MMSCH_MODE_MASK);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, mmUVD_CGC_CTRL), reg_data, sram_sel, indirect);
+
+	/* turn off clock gating */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, mmUVD_CGC_GATE), 0, sram_sel, indirect);
+
+	/* turn on SUVD clock gating */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, mmUVD_SUVD_CGC_GATE), 1, sram_sel, indirect);
+
+	/* turn on sw mode in UVD_SUVD_CGC_CTRL */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, mmUVD_SUVD_CGC_CTRL), 0, sram_sel, indirect);
+}
+
+/**
+ * vcn_v3_0_enable_clock_gating - enable VCN clock gating
+ *
+ * @vinst: Pointer to the VCN instance structure
+ *
+ * Enable clock gating for VCN block
+ */
+static void vcn_v3_0_enable_clock_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst = vinst->inst;
+	uint32_t data;
+
+	/* enable VCN CGC */
+	data = RREG32_SOC15(VCN, inst, mmUVD_CGC_CTRL);
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		data |= 1 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		data |= 0 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(VCN, inst, mmUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, inst, mmUVD_CGC_CTRL);
+	data |= (UVD_CGC_CTRL__UDEC_RE_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_CM_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_IT_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_DB_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_MP_MODE_MASK
+		| UVD_CGC_CTRL__SYS_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_MODE_MASK
+		| UVD_CGC_CTRL__MPEG2_MODE_MASK
+		| UVD_CGC_CTRL__REGS_MODE_MASK
+		| UVD_CGC_CTRL__RBC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_MC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_UMC_MODE_MASK
+		| UVD_CGC_CTRL__IDCT_MODE_MASK
+		| UVD_CGC_CTRL__MPRD_MODE_MASK
+		| UVD_CGC_CTRL__MPC_MODE_MASK
+		| UVD_CGC_CTRL__LBSI_MODE_MASK
+		| UVD_CGC_CTRL__LRBBM_MODE_MASK
+		| UVD_CGC_CTRL__WCB_MODE_MASK
+		| UVD_CGC_CTRL__VCPU_MODE_MASK
+		| UVD_CGC_CTRL__MMSCH_MODE_MASK);
+	WREG32_SOC15(VCN, inst, mmUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, inst, mmUVD_SUVD_CGC_CTRL);
+	data |= (UVD_SUVD_CGC_CTRL__SRE_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SIT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SMP_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SCM_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SDB_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SCLR_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__ENT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__IME_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SITE_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__EFC_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SAOE_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SMPA_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__MPBE0_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__MPBE1_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SIT_AV1_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SDB_AV1_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__MPC1_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__FBC_PCLK_MASK
+		| UVD_SUVD_CGC_CTRL__FBC_CCLK_MASK);
+	WREG32_SOC15(VCN, inst, mmUVD_SUVD_CGC_CTRL, data);
+}
+
+static int vcn_v3_0_start_dpg_mode(struct amdgpu_vcn_inst *vinst, bool indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	struct amdgpu_fw_shared *fw_shared = adev->vcn.inst[inst_idx].fw_shared.cpu_addr;
+	struct amdgpu_ring *ring;
+	uint32_t rb_bufsz, tmp;
+	int ret;
+
+	/* disable register anti-hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, mmUVD_POWER_STATUS), 1,
+		~UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+	/* enable dynamic power gating mode */
+	tmp = RREG32_SOC15(VCN, inst_idx, mmUVD_POWER_STATUS);
+	tmp |= UVD_POWER_STATUS__UVD_PG_MODE_MASK;
+	tmp |= UVD_POWER_STATUS__UVD_PG_EN_MASK;
+	WREG32_SOC15(VCN, inst_idx, mmUVD_POWER_STATUS, tmp);
+
+	if (indirect)
+		adev->vcn.inst[inst_idx].dpg_sram_curr_addr = (uint32_t *)adev->vcn.inst[inst_idx].dpg_sram_cpu_addr;
+
+	/* enable clock gating */
+	vcn_v3_0_clock_gating_dpg_mode(vinst, 0, indirect);
+
+	/* enable VCPU clock */
+	tmp = (0xFF << UVD_VCPU_CNTL__PRB_TIMEOUT_VAL__SHIFT);
+	tmp |= UVD_VCPU_CNTL__CLK_EN_MASK;
+	tmp |= UVD_VCPU_CNTL__BLK_RST_MASK;
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, mmUVD_VCPU_CNTL), tmp, 0, indirect);
+
+	/* disable master interupt */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, mmUVD_MASTINT_EN), 0, 0, indirect);
+
+	/* setup mmUVD_LMI_CTRL */
+	tmp = (0x8 | UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+		UVD_LMI_CTRL__REQ_MODE_MASK |
+		UVD_LMI_CTRL__CRC_RESET_MASK |
+		UVD_LMI_CTRL__MASK_MC_URGENT_MASK |
+		UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+		UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
+		(8 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
+		0x00100000L);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, mmUVD_LMI_CTRL), tmp, 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, mmUVD_MPC_CNTL),
+		0x2 << UVD_MPC_CNTL__REPLACEMENT_MODE__SHIFT, 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, mmUVD_MPC_SET_MUXA0),
+		((0x1 << UVD_MPC_SET_MUXA0__VARA_1__SHIFT) |
+		 (0x2 << UVD_MPC_SET_MUXA0__VARA_2__SHIFT) |
+		 (0x3 << UVD_MPC_SET_MUXA0__VARA_3__SHIFT) |
+		 (0x4 << UVD_MPC_SET_MUXA0__VARA_4__SHIFT)), 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, mmUVD_MPC_SET_MUXB0),
+		 ((0x1 << UVD_MPC_SET_MUXB0__VARB_1__SHIFT) |
+		 (0x2 << UVD_MPC_SET_MUXB0__VARB_2__SHIFT) |
+		 (0x3 << UVD_MPC_SET_MUXB0__VARB_3__SHIFT) |
+		 (0x4 << UVD_MPC_SET_MUXB0__VARB_4__SHIFT)), 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, mmUVD_MPC_SET_MUX),
+		((0x0 << UVD_MPC_SET_MUX__SET_0__SHIFT) |
+		 (0x1 << UVD_MPC_SET_MUX__SET_1__SHIFT) |
+		 (0x2 << UVD_MPC_SET_MUX__SET_2__SHIFT)), 0, indirect);
+
+	vcn_v3_0_mc_resume_dpg_mode(vinst, indirect);
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, mmUVD_REG_XX_MASK), 0x10, 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, mmUVD_RBC_XX_IB_REG_CHECK), 0x3, 0, indirect);
+
+	/* enable LMI MC and UMC channels */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, mmUVD_LMI_CTRL2), 0, 0, indirect);
+
+	/* unblock VCPU register access */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, mmUVD_RB_ARB_CTRL), 0, 0, indirect);
+
+	tmp = (0xFF << UVD_VCPU_CNTL__PRB_TIMEOUT_VAL__SHIFT);
+	tmp |= UVD_VCPU_CNTL__CLK_EN_MASK;
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, mmUVD_VCPU_CNTL), tmp, 0, indirect);
+
+	/* enable master interrupt */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, mmUVD_MASTINT_EN),
+		UVD_MASTINT_EN__VCPU_EN_MASK, 0, indirect);
+
+	/* add nop to workaround PSP size check */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, mmUVD_VCPU_CNTL), tmp, 0, indirect);
+
+	if (indirect) {
+		ret = amdgpu_vcn_psp_update_sram(adev, inst_idx, 0);
+		if (ret) {
+			dev_err(adev->dev, "vcn sram load failed %d\n", ret);
+			return ret;
+		}
+	}
+
+	ring = &adev->vcn.inst[inst_idx].ring_dec;
+	/* force RBC into idle state */
+	rb_bufsz = order_base_2(ring->ring_size);
+	tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
+	WREG32_SOC15(VCN, inst_idx, mmUVD_RBC_RB_CNTL, tmp);
+
+	/* Stall DPG before WPTR/RPTR reset */
+	WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, mmUVD_POWER_STATUS),
+		UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK,
+		~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK);
+	fw_shared->multi_queue.decode_queue_mode |= cpu_to_le32(FW_QUEUE_RING_RESET);
+
+	/* set the write pointer delay */
+	WREG32_SOC15(VCN, inst_idx, mmUVD_RBC_RB_WPTR_CNTL, 0);
+
+	/* set the wb address */
+	WREG32_SOC15(VCN, inst_idx, mmUVD_RBC_RB_RPTR_ADDR,
+		(upper_32_bits(ring->gpu_addr) >> 2));
+
+	/* programm the RB_BASE for ring buffer */
+	WREG32_SOC15(VCN, inst_idx, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW,
+		lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15(VCN, inst_idx, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH,
+		upper_32_bits(ring->gpu_addr));
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32_SOC15(VCN, inst_idx, mmUVD_RBC_RB_RPTR, 0);
+
+	WREG32_SOC15(VCN, inst_idx, mmUVD_SCRATCH2, 0);
+
+	ring->wptr = RREG32_SOC15(VCN, inst_idx, mmUVD_RBC_RB_RPTR);
+	WREG32_SOC15(VCN, inst_idx, mmUVD_RBC_RB_WPTR,
+		lower_32_bits(ring->wptr));
+
+	/* Reset FW shared memory RBC WPTR/RPTR */
+	fw_shared->rb.rptr = 0;
+	fw_shared->rb.wptr = lower_32_bits(ring->wptr);
+
+	/*resetting done, fw can check RB ring */
+	fw_shared->multi_queue.decode_queue_mode &= cpu_to_le32(~FW_QUEUE_RING_RESET);
+
+	/* Unstall DPG */
+	WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, mmUVD_POWER_STATUS),
+		0, ~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, inst_idx, mmUVD_STATUS);
+
+	return 0;
+}
+
+static int vcn_v3_0_start(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int i = vinst->inst;
+	struct amdgpu_fw_shared *fw_shared;
+	struct amdgpu_ring *ring;
+	uint32_t rb_bufsz, tmp;
+	int j, k, r;
+
+	if (adev->vcn.harvest_config & (1 << i))
+		return 0;
+
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_vcn(adev, true, i);
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)
+		return vcn_v3_0_start_dpg_mode(vinst, vinst->indirect_sram);
+
+	/* disable VCN power gating */
+	vcn_v3_0_disable_static_power_gating(vinst);
+
+	/* set VCN status busy */
+	tmp = RREG32_SOC15(VCN, i, mmUVD_STATUS) | UVD_STATUS__UVD_BUSY;
+	WREG32_SOC15(VCN, i, mmUVD_STATUS, tmp);
+
+	/* SW clock gating */
+	vcn_v3_0_disable_clock_gating(vinst);
+
+	/* enable VCPU clock */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_VCPU_CNTL),
+		 UVD_VCPU_CNTL__CLK_EN_MASK, ~UVD_VCPU_CNTL__CLK_EN_MASK);
+
+	/* disable master interrupt */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_MASTINT_EN), 0,
+		 ~UVD_MASTINT_EN__VCPU_EN_MASK);
+
+	/* enable LMI MC and UMC channels */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_LMI_CTRL2), 0,
+		 ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+
+	tmp = RREG32_SOC15(VCN, i, mmUVD_SOFT_RESET);
+	tmp &= ~UVD_SOFT_RESET__LMI_SOFT_RESET_MASK;
+	tmp &= ~UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK;
+	WREG32_SOC15(VCN, i, mmUVD_SOFT_RESET, tmp);
+
+	/* setup mmUVD_LMI_CTRL */
+	tmp = RREG32_SOC15(VCN, i, mmUVD_LMI_CTRL);
+	WREG32_SOC15(VCN, i, mmUVD_LMI_CTRL, tmp |
+		     UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK	|
+		     UVD_LMI_CTRL__MASK_MC_URGENT_MASK |
+		     UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+		     UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK);
+
+	/* setup mmUVD_MPC_CNTL */
+	tmp = RREG32_SOC15(VCN, i, mmUVD_MPC_CNTL);
+	tmp &= ~UVD_MPC_CNTL__REPLACEMENT_MODE_MASK;
+	tmp |= 0x2 << UVD_MPC_CNTL__REPLACEMENT_MODE__SHIFT;
+	WREG32_SOC15(VCN, i, mmUVD_MPC_CNTL, tmp);
+
+	/* setup UVD_MPC_SET_MUXA0 */
+	WREG32_SOC15(VCN, i, mmUVD_MPC_SET_MUXA0,
+		     ((0x1 << UVD_MPC_SET_MUXA0__VARA_1__SHIFT) |
+		      (0x2 << UVD_MPC_SET_MUXA0__VARA_2__SHIFT) |
+		      (0x3 << UVD_MPC_SET_MUXA0__VARA_3__SHIFT) |
+		      (0x4 << UVD_MPC_SET_MUXA0__VARA_4__SHIFT)));
+
+	/* setup UVD_MPC_SET_MUXB0 */
+	WREG32_SOC15(VCN, i, mmUVD_MPC_SET_MUXB0,
+		     ((0x1 << UVD_MPC_SET_MUXB0__VARB_1__SHIFT) |
+		      (0x2 << UVD_MPC_SET_MUXB0__VARB_2__SHIFT) |
+		      (0x3 << UVD_MPC_SET_MUXB0__VARB_3__SHIFT) |
+		      (0x4 << UVD_MPC_SET_MUXB0__VARB_4__SHIFT)));
+
+	/* setup mmUVD_MPC_SET_MUX */
+	WREG32_SOC15(VCN, i, mmUVD_MPC_SET_MUX,
+		     ((0x0 << UVD_MPC_SET_MUX__SET_0__SHIFT) |
+		      (0x1 << UVD_MPC_SET_MUX__SET_1__SHIFT) |
+		      (0x2 << UVD_MPC_SET_MUX__SET_2__SHIFT)));
+
+	vcn_v3_0_mc_resume(vinst);
+
+	/* VCN global tiling registers */
+	WREG32_SOC15(VCN, i, mmUVD_GFX10_ADDR_CONFIG,
+		     adev->gfx.config.gb_addr_config);
+
+	/* unblock VCPU register access */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_RB_ARB_CTRL), 0,
+		 ~UVD_RB_ARB_CTRL__VCPU_DIS_MASK);
+
+	/* release VCPU reset to boot */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_VCPU_CNTL), 0,
+		 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+	for (j = 0; j < 10; ++j) {
+		uint32_t status;
+
+		for (k = 0; k < 100; ++k) {
+			status = RREG32_SOC15(VCN, i, mmUVD_STATUS);
+			if (status & 2)
+				break;
+			mdelay(10);
+		}
+		r = 0;
+		if (status & 2)
+			break;
+
+		DRM_ERROR("VCN[%d] decode not responding, trying to reset the VCPU!!!\n", i);
+		WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_VCPU_CNTL),
+			 UVD_VCPU_CNTL__BLK_RST_MASK,
+			 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+		mdelay(10);
+		WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_VCPU_CNTL), 0,
+			 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+		mdelay(10);
+		r = -1;
+	}
+
+	if (r) {
+		DRM_ERROR("VCN[%d] decode not responding, giving up!!!\n", i);
+		return r;
+	}
+
+	/* enable master interrupt */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_MASTINT_EN),
+		 UVD_MASTINT_EN__VCPU_EN_MASK,
+		 ~UVD_MASTINT_EN__VCPU_EN_MASK);
+
+	/* clear the busy bit of VCN_STATUS */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_STATUS), 0,
+		 ~(2 << UVD_STATUS__VCPU_REPORT__SHIFT));
+
+	WREG32_SOC15(VCN, i, mmUVD_LMI_RBC_RB_VMID, 0);
+
+	ring = &adev->vcn.inst[i].ring_dec;
+	/* force RBC into idle state */
+	rb_bufsz = order_base_2(ring->ring_size);
+	tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
+	WREG32_SOC15(VCN, i, mmUVD_RBC_RB_CNTL, tmp);
+
+	fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+	fw_shared->multi_queue.decode_queue_mode |= cpu_to_le32(FW_QUEUE_RING_RESET);
+
+	/* programm the RB_BASE for ring buffer */
+	WREG32_SOC15(VCN, i, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW,
+		     lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15(VCN, i, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH,
+		     upper_32_bits(ring->gpu_addr));
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32_SOC15(VCN, i, mmUVD_RBC_RB_RPTR, 0);
+
+	WREG32_SOC15(VCN, i, mmUVD_SCRATCH2, 0);
+	ring->wptr = RREG32_SOC15(VCN, i, mmUVD_RBC_RB_RPTR);
+	WREG32_SOC15(VCN, i, mmUVD_RBC_RB_WPTR,
+		     lower_32_bits(ring->wptr));
+	fw_shared->rb.wptr = lower_32_bits(ring->wptr);
+	fw_shared->multi_queue.decode_queue_mode &= cpu_to_le32(~FW_QUEUE_RING_RESET);
+
+	if (amdgpu_ip_version(adev, UVD_HWIP, 0) !=
+	    IP_VERSION(3, 0, 33)) {
+		fw_shared->multi_queue.encode_generalpurpose_queue_mode |= cpu_to_le32(FW_QUEUE_RING_RESET);
+		ring = &adev->vcn.inst[i].ring_enc[0];
+		WREG32_SOC15(VCN, i, mmUVD_RB_RPTR, lower_32_bits(ring->wptr));
+		WREG32_SOC15(VCN, i, mmUVD_RB_WPTR, lower_32_bits(ring->wptr));
+		WREG32_SOC15(VCN, i, mmUVD_RB_BASE_LO, ring->gpu_addr);
+		WREG32_SOC15(VCN, i, mmUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+		WREG32_SOC15(VCN, i, mmUVD_RB_SIZE, ring->ring_size / 4);
+		fw_shared->multi_queue.encode_generalpurpose_queue_mode &= cpu_to_le32(~FW_QUEUE_RING_RESET);
+
+		fw_shared->multi_queue.encode_lowlatency_queue_mode |= cpu_to_le32(FW_QUEUE_RING_RESET);
+		ring = &adev->vcn.inst[i].ring_enc[1];
+		WREG32_SOC15(VCN, i, mmUVD_RB_RPTR2, lower_32_bits(ring->wptr));
+		WREG32_SOC15(VCN, i, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
+		WREG32_SOC15(VCN, i, mmUVD_RB_BASE_LO2, ring->gpu_addr);
+		WREG32_SOC15(VCN, i, mmUVD_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
+		WREG32_SOC15(VCN, i, mmUVD_RB_SIZE2, ring->ring_size / 4);
+		fw_shared->multi_queue.encode_lowlatency_queue_mode &= cpu_to_le32(~FW_QUEUE_RING_RESET);
+	}
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, i, mmUVD_STATUS);
+
+	return 0;
+}
+
+static int vcn_v3_0_start_sriov(struct amdgpu_device *adev)
+{
+	int i, j;
+	struct amdgpu_ring *ring;
+	uint64_t cache_addr;
+	uint64_t rb_addr;
+	uint64_t ctx_addr;
+	uint32_t param, resp, expected;
+	uint32_t offset, cache_size;
+	uint32_t tmp, timeout;
+
+	struct amdgpu_mm_table *table = &adev->virt.mm_table;
+	uint32_t *table_loc;
+	uint32_t table_size;
+	uint32_t size, size_dw;
+
+	struct mmsch_v3_0_cmd_direct_write
+		direct_wt = { {0} };
+	struct mmsch_v3_0_cmd_direct_read_modify_write
+		direct_rd_mod_wt = { {0} };
+	struct mmsch_v3_0_cmd_end end = { {0} };
+	struct mmsch_v3_0_init_header header;
+
+	direct_wt.cmd_header.command_type =
+		MMSCH_COMMAND__DIRECT_REG_WRITE;
+	direct_rd_mod_wt.cmd_header.command_type =
+		MMSCH_COMMAND__DIRECT_REG_READ_MODIFY_WRITE;
+	end.cmd_header.command_type =
+		MMSCH_COMMAND__END;
+
+	header.version = MMSCH_VERSION;
+	header.total_size = sizeof(struct mmsch_v3_0_init_header) >> 2;
+	for (i = 0; i < MMSCH_V3_0_VCN_INSTANCES; i++) {
+		header.inst[i].init_status = 0;
+		header.inst[i].table_offset = 0;
+		header.inst[i].table_size = 0;
+	}
+
+	table_loc = (uint32_t *)table->cpu_addr;
+	table_loc += header.total_size;
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		table_size = 0;
+
+		MMSCH_V3_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(VCN, i,
+			mmUVD_STATUS),
+			~UVD_STATUS__UVD_BUSY, UVD_STATUS__UVD_BUSY);
+
+		cache_size = AMDGPU_GPU_PAGE_ALIGN(adev->vcn.inst[i].fw->size + 4);
+
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+			MMSCH_V3_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+				mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+				adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + i].tmr_mc_addr_lo);
+			MMSCH_V3_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+				mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+				adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + i].tmr_mc_addr_hi);
+			offset = 0;
+			MMSCH_V3_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+				mmUVD_VCPU_CACHE_OFFSET0),
+				0);
+		} else {
+			MMSCH_V3_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+				mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+				lower_32_bits(adev->vcn.inst[i].gpu_addr));
+			MMSCH_V3_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+				mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+				upper_32_bits(adev->vcn.inst[i].gpu_addr));
+			offset = cache_size;
+			MMSCH_V3_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+				mmUVD_VCPU_CACHE_OFFSET0),
+				AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
+		}
+
+		MMSCH_V3_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			mmUVD_VCPU_CACHE_SIZE0),
+			cache_size);
+
+		cache_addr = adev->vcn.inst[i].gpu_addr + offset;
+		MMSCH_V3_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW),
+			lower_32_bits(cache_addr));
+		MMSCH_V3_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH),
+			upper_32_bits(cache_addr));
+		MMSCH_V3_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			mmUVD_VCPU_CACHE_OFFSET1),
+			0);
+		MMSCH_V3_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			mmUVD_VCPU_CACHE_SIZE1),
+			AMDGPU_VCN_STACK_SIZE);
+
+		cache_addr = adev->vcn.inst[i].gpu_addr + offset +
+			AMDGPU_VCN_STACK_SIZE;
+		MMSCH_V3_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW),
+			lower_32_bits(cache_addr));
+		MMSCH_V3_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH),
+			upper_32_bits(cache_addr));
+		MMSCH_V3_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			mmUVD_VCPU_CACHE_OFFSET2),
+			0);
+		MMSCH_V3_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			mmUVD_VCPU_CACHE_SIZE2),
+			AMDGPU_VCN_CONTEXT_SIZE);
+
+		for (j = 0; j < adev->vcn.inst[i].num_enc_rings; ++j) {
+			ring = &adev->vcn.inst[i].ring_enc[j];
+			ring->wptr = 0;
+			rb_addr = ring->gpu_addr;
+			MMSCH_V3_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+				mmUVD_RB_BASE_LO),
+				lower_32_bits(rb_addr));
+			MMSCH_V3_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+				mmUVD_RB_BASE_HI),
+				upper_32_bits(rb_addr));
+			MMSCH_V3_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+				mmUVD_RB_SIZE),
+				ring->ring_size / 4);
+		}
+
+		ring = &adev->vcn.inst[i].ring_dec;
+		ring->wptr = 0;
+		rb_addr = ring->gpu_addr;
+		MMSCH_V3_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			mmUVD_LMI_RBC_RB_64BIT_BAR_LOW),
+			lower_32_bits(rb_addr));
+		MMSCH_V3_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH),
+			upper_32_bits(rb_addr));
+		/* force RBC into idle state */
+		tmp = order_base_2(ring->ring_size);
+		tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, tmp);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
+		MMSCH_V3_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			mmUVD_RBC_RB_CNTL),
+			tmp);
+
+		/* add end packet */
+		MMSCH_V3_0_INSERT_END();
+
+		/* refine header */
+		header.inst[i].init_status = 0;
+		header.inst[i].table_offset = header.total_size;
+		header.inst[i].table_size = table_size;
+		header.total_size += table_size;
+	}
+
+	/* Update init table header in memory */
+	size = sizeof(struct mmsch_v3_0_init_header);
+	table_loc = (uint32_t *)table->cpu_addr;
+	memcpy((void *)table_loc, &header, size);
+
+	/* message MMSCH (in VCN[0]) to initialize this client
+	 * 1, write to mmsch_vf_ctx_addr_lo/hi register with GPU mc addr
+	 * of memory descriptor location
+	 */
+	ctx_addr = table->gpu_addr;
+	WREG32_SOC15(VCN, 0, mmMMSCH_VF_CTX_ADDR_LO, lower_32_bits(ctx_addr));
+	WREG32_SOC15(VCN, 0, mmMMSCH_VF_CTX_ADDR_HI, upper_32_bits(ctx_addr));
+
+	/* 2, update vmid of descriptor */
+	tmp = RREG32_SOC15(VCN, 0, mmMMSCH_VF_VMID);
+	tmp &= ~MMSCH_VF_VMID__VF_CTX_VMID_MASK;
+	/* use domain0 for MM scheduler */
+	tmp |= (0 << MMSCH_VF_VMID__VF_CTX_VMID__SHIFT);
+	WREG32_SOC15(VCN, 0, mmMMSCH_VF_VMID, tmp);
+
+	/* 3, notify mmsch about the size of this descriptor */
+	size = header.total_size;
+	WREG32_SOC15(VCN, 0, mmMMSCH_VF_CTX_SIZE, size);
+
+	/* 4, set resp to zero */
+	WREG32_SOC15(VCN, 0, mmMMSCH_VF_MAILBOX_RESP, 0);
+
+	/* 5, kick off the initialization and wait until
+	 * MMSCH_VF_MAILBOX_RESP becomes non-zero
+	 */
+	param = 0x10000001;
+	WREG32_SOC15(VCN, 0, mmMMSCH_VF_MAILBOX_HOST, param);
+	tmp = 0;
+	timeout = 1000;
+	resp = 0;
+	expected = param + 1;
+	while (resp != expected) {
+		resp = RREG32_SOC15(VCN, 0, mmMMSCH_VF_MAILBOX_RESP);
+		if (resp == expected)
+			break;
+
+		udelay(10);
+		tmp = tmp + 10;
+		if (tmp >= timeout) {
+			DRM_ERROR("failed to init MMSCH. TIME-OUT after %d usec"\
+				" waiting for mmMMSCH_VF_MAILBOX_RESP "\
+				"(expected=0x%08x, readback=0x%08x)\n",
+				tmp, expected, resp);
+			return -EBUSY;
+		}
+	}
+
+	return 0;
+}
+
+static int vcn_v3_0_stop_dpg_mode(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	struct dpg_pause_state state = {.fw_based = VCN_DPG_STATE__UNPAUSE};
+	uint32_t tmp;
+
+	vcn_v3_0_pause_dpg_mode(vinst, &state);
+
+	/* Wait for power status to be 1 */
+	SOC15_WAIT_ON_RREG(VCN, inst_idx, mmUVD_POWER_STATUS, 1,
+		UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+	/* wait for read ptr to be equal to write ptr */
+	tmp = RREG32_SOC15(VCN, inst_idx, mmUVD_RB_WPTR);
+	SOC15_WAIT_ON_RREG(VCN, inst_idx, mmUVD_RB_RPTR, tmp, 0xFFFFFFFF);
+
+	tmp = RREG32_SOC15(VCN, inst_idx, mmUVD_RB_WPTR2);
+	SOC15_WAIT_ON_RREG(VCN, inst_idx, mmUVD_RB_RPTR2, tmp, 0xFFFFFFFF);
+
+	tmp = RREG32_SOC15(VCN, inst_idx, mmUVD_RBC_RB_WPTR) & 0x7FFFFFFF;
+	SOC15_WAIT_ON_RREG(VCN, inst_idx, mmUVD_RBC_RB_RPTR, tmp, 0xFFFFFFFF);
+
+	SOC15_WAIT_ON_RREG(VCN, inst_idx, mmUVD_POWER_STATUS, 1,
+		UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+	/* disable dynamic power gating mode */
+	WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, mmUVD_POWER_STATUS), 0,
+		~UVD_POWER_STATUS__UVD_PG_MODE_MASK);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, inst_idx, mmUVD_STATUS);
+
+	return 0;
+}
+
+static int vcn_v3_0_stop(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int i = vinst->inst;
+	uint32_t tmp;
+	int r = 0;
+
+	if (adev->vcn.harvest_config & (1 << i))
+		return 0;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) {
+		r = vcn_v3_0_stop_dpg_mode(vinst);
+		goto done;
+	}
+
+	/* wait for vcn idle */
+	r = SOC15_WAIT_ON_RREG(VCN, i, mmUVD_STATUS, UVD_STATUS__IDLE, 0x7);
+	if (r)
+		goto done;
+
+	tmp = UVD_LMI_STATUS__VCPU_LMI_WRITE_CLEAN_MASK |
+		UVD_LMI_STATUS__READ_CLEAN_MASK |
+		UVD_LMI_STATUS__WRITE_CLEAN_MASK |
+		UVD_LMI_STATUS__WRITE_CLEAN_RAW_MASK;
+	r = SOC15_WAIT_ON_RREG(VCN, i, mmUVD_LMI_STATUS, tmp, tmp);
+	if (r)
+		goto done;
+
+	/* disable LMI UMC channel */
+	tmp = RREG32_SOC15(VCN, i, mmUVD_LMI_CTRL2);
+	tmp |= UVD_LMI_CTRL2__STALL_ARB_UMC_MASK;
+	WREG32_SOC15(VCN, i, mmUVD_LMI_CTRL2, tmp);
+	tmp = UVD_LMI_STATUS__UMC_READ_CLEAN_RAW_MASK|
+		UVD_LMI_STATUS__UMC_WRITE_CLEAN_RAW_MASK;
+	r = SOC15_WAIT_ON_RREG(VCN, i, mmUVD_LMI_STATUS, tmp, tmp);
+	if (r)
+		goto done;
+
+	/* block VCPU register access */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_RB_ARB_CTRL),
+		 UVD_RB_ARB_CTRL__VCPU_DIS_MASK,
+		 ~UVD_RB_ARB_CTRL__VCPU_DIS_MASK);
+
+	/* reset VCPU */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_VCPU_CNTL),
+		 UVD_VCPU_CNTL__BLK_RST_MASK,
+		 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+	/* disable VCPU clock */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, mmUVD_VCPU_CNTL), 0,
+		 ~(UVD_VCPU_CNTL__CLK_EN_MASK));
+
+	/* apply soft reset */
+	tmp = RREG32_SOC15(VCN, i, mmUVD_SOFT_RESET);
+	tmp |= UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK;
+	WREG32_SOC15(VCN, i, mmUVD_SOFT_RESET, tmp);
+	tmp = RREG32_SOC15(VCN, i, mmUVD_SOFT_RESET);
+	tmp |= UVD_SOFT_RESET__LMI_SOFT_RESET_MASK;
+	WREG32_SOC15(VCN, i, mmUVD_SOFT_RESET, tmp);
+
+	/* clear status */
+	WREG32_SOC15(VCN, i, mmUVD_STATUS, 0);
+
+	/* apply HW clock gating */
+	vcn_v3_0_enable_clock_gating(vinst);
+
+	/* enable VCN power gating */
+	vcn_v3_0_enable_static_power_gating(vinst);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, i, mmUVD_STATUS);
+
+done:
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_vcn(adev, false, i);
+
+	return r;
+}
+
+static int vcn_v3_0_pause_dpg_mode(struct amdgpu_vcn_inst *vinst,
+				   struct dpg_pause_state *new_state)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	struct amdgpu_fw_shared *fw_shared;
+	struct amdgpu_ring *ring;
+	uint32_t reg_data = 0;
+	int ret_code;
+
+	/* pause/unpause if state is changed */
+	if (adev->vcn.inst[inst_idx].pause_state.fw_based != new_state->fw_based) {
+		DRM_DEBUG("dpg pause state changed %d -> %d",
+			adev->vcn.inst[inst_idx].pause_state.fw_based,	new_state->fw_based);
+		reg_data = RREG32_SOC15(VCN, inst_idx, mmUVD_DPG_PAUSE) &
+			(~UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK);
+
+		if (new_state->fw_based == VCN_DPG_STATE__PAUSE) {
+			ret_code = SOC15_WAIT_ON_RREG(VCN, inst_idx, mmUVD_POWER_STATUS, 0x1,
+				UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+			if (!ret_code) {
+				/* pause DPG */
+				reg_data |= UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
+				WREG32_SOC15(VCN, inst_idx, mmUVD_DPG_PAUSE, reg_data);
+
+				/* wait for ACK */
+				SOC15_WAIT_ON_RREG(VCN, inst_idx, mmUVD_DPG_PAUSE,
+					UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK,
+					UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK);
+
+				/* Stall DPG before WPTR/RPTR reset */
+				WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, mmUVD_POWER_STATUS),
+					UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK,
+					~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK);
+
+				if (amdgpu_ip_version(adev, UVD_HWIP, 0) !=
+				    IP_VERSION(3, 0, 33)) {
+					/* Restore */
+					fw_shared = adev->vcn.inst[inst_idx].fw_shared.cpu_addr;
+					fw_shared->multi_queue.encode_generalpurpose_queue_mode |= cpu_to_le32(FW_QUEUE_RING_RESET);
+					ring = &adev->vcn.inst[inst_idx].ring_enc[0];
+					ring->wptr = 0;
+					WREG32_SOC15(VCN, inst_idx, mmUVD_RB_BASE_LO, ring->gpu_addr);
+					WREG32_SOC15(VCN, inst_idx, mmUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+					WREG32_SOC15(VCN, inst_idx, mmUVD_RB_SIZE, ring->ring_size / 4);
+					WREG32_SOC15(VCN, inst_idx, mmUVD_RB_RPTR, lower_32_bits(ring->wptr));
+					WREG32_SOC15(VCN, inst_idx, mmUVD_RB_WPTR, lower_32_bits(ring->wptr));
+					fw_shared->multi_queue.encode_generalpurpose_queue_mode &= cpu_to_le32(~FW_QUEUE_RING_RESET);
+
+					fw_shared->multi_queue.encode_lowlatency_queue_mode |= cpu_to_le32(FW_QUEUE_RING_RESET);
+					ring = &adev->vcn.inst[inst_idx].ring_enc[1];
+					ring->wptr = 0;
+					WREG32_SOC15(VCN, inst_idx, mmUVD_RB_BASE_LO2, ring->gpu_addr);
+					WREG32_SOC15(VCN, inst_idx, mmUVD_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
+					WREG32_SOC15(VCN, inst_idx, mmUVD_RB_SIZE2, ring->ring_size / 4);
+					WREG32_SOC15(VCN, inst_idx, mmUVD_RB_RPTR2, lower_32_bits(ring->wptr));
+					WREG32_SOC15(VCN, inst_idx, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
+					fw_shared->multi_queue.encode_lowlatency_queue_mode &= cpu_to_le32(~FW_QUEUE_RING_RESET);
+
+					/* restore wptr/rptr with pointers saved in FW shared memory*/
+					WREG32_SOC15(VCN, inst_idx, mmUVD_RBC_RB_RPTR, fw_shared->rb.rptr);
+					WREG32_SOC15(VCN, inst_idx, mmUVD_RBC_RB_WPTR, fw_shared->rb.wptr);
+				}
+
+				/* Unstall DPG */
+				WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, mmUVD_POWER_STATUS),
+					0, ~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK);
+
+				SOC15_WAIT_ON_RREG(VCN, inst_idx, mmUVD_POWER_STATUS,
+					UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON, UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+			}
+		} else {
+			/* unpause dpg, no need to wait */
+			reg_data &= ~UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
+			WREG32_SOC15(VCN, inst_idx, mmUVD_DPG_PAUSE, reg_data);
+		}
+		adev->vcn.inst[inst_idx].pause_state.fw_based = new_state->fw_based;
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v3_0_dec_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t vcn_v3_0_dec_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32_SOC15(VCN, ring->me, mmUVD_RBC_RB_RPTR);
+}
+
+/**
+ * vcn_v3_0_dec_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t vcn_v3_0_dec_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell)
+		return *ring->wptr_cpu_addr;
+	else
+		return RREG32_SOC15(VCN, ring->me, mmUVD_RBC_RB_WPTR);
+}
+
+/**
+ * vcn_v3_0_dec_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void vcn_v3_0_dec_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_fw_shared *fw_shared;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) {
+		/*whenever update RBC_RB_WPTR, we save the wptr in shared rb.wptr and scratch2 */
+		fw_shared = adev->vcn.inst[ring->me].fw_shared.cpu_addr;
+		fw_shared->rb.wptr = lower_32_bits(ring->wptr);
+		WREG32_SOC15(VCN, ring->me, mmUVD_SCRATCH2,
+			lower_32_bits(ring->wptr));
+	}
+
+	if (ring->use_doorbell) {
+		*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
+		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+	} else {
+		WREG32_SOC15(VCN, ring->me, mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr));
+	}
+}
+
+static const struct amdgpu_ring_funcs vcn_v3_0_dec_sw_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_DEC,
+	.align_mask = 0x3f,
+	.nop = VCN_DEC_SW_CMD_NO_OP,
+	.secure_submission_supported = true,
+	.get_rptr = vcn_v3_0_dec_ring_get_rptr,
+	.get_wptr = vcn_v3_0_dec_ring_get_wptr,
+	.set_wptr = vcn_v3_0_dec_ring_set_wptr,
+	.emit_frame_size =
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 4 +
+		VCN_SW_RING_EMIT_FRAME_SIZE,
+	.emit_ib_size = 5, /* vcn_dec_sw_ring_emit_ib */
+	.emit_ib = vcn_dec_sw_ring_emit_ib,
+	.emit_fence = vcn_dec_sw_ring_emit_fence,
+	.emit_vm_flush = vcn_dec_sw_ring_emit_vm_flush,
+	.test_ring = amdgpu_vcn_dec_sw_ring_test_ring,
+	.test_ib = NULL,//amdgpu_vcn_dec_sw_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.insert_end = vcn_dec_sw_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_vcn_ring_begin_use,
+	.end_use = amdgpu_vcn_ring_end_use,
+	.emit_wreg = vcn_dec_sw_ring_emit_wreg,
+	.emit_reg_wait = vcn_dec_sw_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+};
+
+static int vcn_v3_0_limit_sched(struct amdgpu_cs_parser *p,
+				struct amdgpu_job *job)
+{
+	struct drm_gpu_scheduler **scheds;
+	struct dma_fence *fence;
+
+	/* if VCN0 is harvested, we can't support AV1 */
+	if (p->adev->vcn.harvest_config & AMDGPU_VCN_HARVEST_VCN0)
+		return -EINVAL;
+
+	/* wait for all jobs to finish before switching to instance 0 */
+	fence = amdgpu_ctx_get_fence(p->ctx, job->base.entity, ~0ull);
+	if (fence) {
+		dma_fence_wait(fence, false);
+		dma_fence_put(fence);
+	}
+
+	scheds = p->adev->gpu_sched[AMDGPU_HW_IP_VCN_DEC]
+		[AMDGPU_RING_PRIO_DEFAULT].sched;
+	drm_sched_entity_modify_sched(job->base.entity, scheds, 1);
+	return 0;
+}
+
+static int vcn_v3_0_dec_msg(struct amdgpu_cs_parser *p, struct amdgpu_job *job,
+			    uint64_t addr)
+{
+	struct ttm_operation_ctx ctx = { false, false };
+	struct amdgpu_bo_va_mapping *map;
+	uint32_t *msg, num_buffers;
+	struct amdgpu_bo *bo;
+	uint64_t start, end;
+	unsigned int i;
+	void *ptr;
+	int r;
+
+	addr &= AMDGPU_GMC_HOLE_MASK;
+	r = amdgpu_cs_find_mapping(p, addr, &bo, &map);
+	if (r) {
+		DRM_ERROR("Can't find BO for addr 0x%08Lx\n", addr);
+		return r;
+	}
+
+	start = map->start * AMDGPU_GPU_PAGE_SIZE;
+	end = (map->last + 1) * AMDGPU_GPU_PAGE_SIZE;
+	if (addr & 0x7) {
+		DRM_ERROR("VCN messages must be 8 byte aligned!\n");
+		return -EINVAL;
+	}
+
+	bo->flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
+	amdgpu_bo_placement_from_domain(bo, bo->allowed_domains);
+	r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+	if (r) {
+		DRM_ERROR("Failed validating the VCN message BO (%d)!\n", r);
+		return r;
+	}
+
+	r = amdgpu_bo_kmap(bo, &ptr);
+	if (r) {
+		DRM_ERROR("Failed mapping the VCN message (%d)!\n", r);
+		return r;
+	}
+
+	msg = ptr + addr - start;
+
+	/* Check length */
+	if (msg[1] > end - addr) {
+		r = -EINVAL;
+		goto out;
+	}
+
+	if (msg[3] != RDECODE_MSG_CREATE)
+		goto out;
+
+	num_buffers = msg[2];
+	for (i = 0, msg = &msg[6]; i < num_buffers; ++i, msg += 4) {
+		uint32_t offset, size, *create;
+
+		if (msg[0] != RDECODE_MESSAGE_CREATE)
+			continue;
+
+		offset = msg[1];
+		size = msg[2];
+
+		if (offset + size > end) {
+			r = -EINVAL;
+			goto out;
+		}
+
+		create = ptr + addr + offset - start;
+
+		/* H246, HEVC and VP9 can run on any instance */
+		if (create[0] == 0x7 || create[0] == 0x10 || create[0] == 0x11)
+			continue;
+
+		r = vcn_v3_0_limit_sched(p, job);
+		if (r)
+			goto out;
+	}
+
+out:
+	amdgpu_bo_kunmap(bo);
+	return r;
+}
+
+static int vcn_v3_0_ring_patch_cs_in_place(struct amdgpu_cs_parser *p,
+					   struct amdgpu_job *job,
+					   struct amdgpu_ib *ib)
+{
+	struct amdgpu_ring *ring = amdgpu_job_ring(job);
+	uint32_t msg_lo = 0, msg_hi = 0;
+	unsigned i;
+	int r;
+
+	/* The first instance can decode anything */
+	if (!ring->me)
+		return 0;
+
+	for (i = 0; i < ib->length_dw; i += 2) {
+		uint32_t reg = amdgpu_ib_get_value(ib, i);
+		uint32_t val = amdgpu_ib_get_value(ib, i + 1);
+
+		if (reg == PACKET0(p->adev->vcn.inst[ring->me].internal.data0, 0)) {
+			msg_lo = val;
+		} else if (reg == PACKET0(p->adev->vcn.inst[ring->me].internal.data1, 0)) {
+			msg_hi = val;
+		} else if (reg == PACKET0(p->adev->vcn.inst[ring->me].internal.cmd, 0) &&
+			   val == 0) {
+			r = vcn_v3_0_dec_msg(p, job,
+					     ((u64)msg_hi) << 32 | msg_lo);
+			if (r)
+				return r;
+		}
+	}
+	return 0;
+}
+
+static const struct amdgpu_ring_funcs vcn_v3_0_dec_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_DEC,
+	.align_mask = 0xf,
+	.secure_submission_supported = true,
+	.get_rptr = vcn_v3_0_dec_ring_get_rptr,
+	.get_wptr = vcn_v3_0_dec_ring_get_wptr,
+	.set_wptr = vcn_v3_0_dec_ring_set_wptr,
+	.patch_cs_in_place = vcn_v3_0_ring_patch_cs_in_place,
+	.emit_frame_size =
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
+		8 + /* vcn_v2_0_dec_ring_emit_vm_flush */
+		14 + 14 + /* vcn_v2_0_dec_ring_emit_fence x2 vm fence */
+		6,
+	.emit_ib_size = 8, /* vcn_v2_0_dec_ring_emit_ib */
+	.emit_ib = vcn_v2_0_dec_ring_emit_ib,
+	.emit_fence = vcn_v2_0_dec_ring_emit_fence,
+	.emit_vm_flush = vcn_v2_0_dec_ring_emit_vm_flush,
+	.test_ring = vcn_v2_0_dec_ring_test_ring,
+	.test_ib = amdgpu_vcn_dec_ring_test_ib,
+	.insert_nop = vcn_v2_0_dec_ring_insert_nop,
+	.insert_start = vcn_v2_0_dec_ring_insert_start,
+	.insert_end = vcn_v2_0_dec_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_vcn_ring_begin_use,
+	.end_use = amdgpu_vcn_ring_end_use,
+	.emit_wreg = vcn_v2_0_dec_ring_emit_wreg,
+	.emit_reg_wait = vcn_v2_0_dec_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = amdgpu_vcn_ring_reset,
+};
+
+/**
+ * vcn_v3_0_enc_ring_get_rptr - get enc read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware enc read pointer
+ */
+static uint64_t vcn_v3_0_enc_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->vcn.inst[ring->me].ring_enc[0])
+		return RREG32_SOC15(VCN, ring->me, mmUVD_RB_RPTR);
+	else
+		return RREG32_SOC15(VCN, ring->me, mmUVD_RB_RPTR2);
+}
+
+/**
+ * vcn_v3_0_enc_ring_get_wptr - get enc write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware enc write pointer
+ */
+static uint64_t vcn_v3_0_enc_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->vcn.inst[ring->me].ring_enc[0]) {
+		if (ring->use_doorbell)
+			return *ring->wptr_cpu_addr;
+		else
+			return RREG32_SOC15(VCN, ring->me, mmUVD_RB_WPTR);
+	} else {
+		if (ring->use_doorbell)
+			return *ring->wptr_cpu_addr;
+		else
+			return RREG32_SOC15(VCN, ring->me, mmUVD_RB_WPTR2);
+	}
+}
+
+/**
+ * vcn_v3_0_enc_ring_set_wptr - set enc write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the enc write pointer to the hardware
+ */
+static void vcn_v3_0_enc_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->vcn.inst[ring->me].ring_enc[0]) {
+		if (ring->use_doorbell) {
+			*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
+			WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+		} else {
+			WREG32_SOC15(VCN, ring->me, mmUVD_RB_WPTR, lower_32_bits(ring->wptr));
+		}
+	} else {
+		if (ring->use_doorbell) {
+			*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
+			WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+		} else {
+			WREG32_SOC15(VCN, ring->me, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
+		}
+	}
+}
+
+static const struct amdgpu_ring_funcs vcn_v3_0_enc_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_ENC,
+	.align_mask = 0x3f,
+	.nop = VCN_ENC_CMD_NO_OP,
+	.get_rptr = vcn_v3_0_enc_ring_get_rptr,
+	.get_wptr = vcn_v3_0_enc_ring_get_wptr,
+	.set_wptr = vcn_v3_0_enc_ring_set_wptr,
+	.emit_frame_size =
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 4 +
+		4 + /* vcn_v2_0_enc_ring_emit_vm_flush */
+		5 + 5 + /* vcn_v2_0_enc_ring_emit_fence x2 vm fence */
+		1, /* vcn_v2_0_enc_ring_insert_end */
+	.emit_ib_size = 5, /* vcn_v2_0_enc_ring_emit_ib */
+	.emit_ib = vcn_v2_0_enc_ring_emit_ib,
+	.emit_fence = vcn_v2_0_enc_ring_emit_fence,
+	.emit_vm_flush = vcn_v2_0_enc_ring_emit_vm_flush,
+	.test_ring = amdgpu_vcn_enc_ring_test_ring,
+	.test_ib = amdgpu_vcn_enc_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.insert_end = vcn_v2_0_enc_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_vcn_ring_begin_use,
+	.end_use = amdgpu_vcn_ring_end_use,
+	.emit_wreg = vcn_v2_0_enc_ring_emit_wreg,
+	.emit_reg_wait = vcn_v2_0_enc_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = amdgpu_vcn_ring_reset,
+};
+
+static void vcn_v3_0_set_dec_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		if (!DEC_SW_RING_ENABLED)
+			adev->vcn.inst[i].ring_dec.funcs = &vcn_v3_0_dec_ring_vm_funcs;
+		else
+			adev->vcn.inst[i].ring_dec.funcs = &vcn_v3_0_dec_sw_ring_vm_funcs;
+		adev->vcn.inst[i].ring_dec.me = i;
+	}
+}
+
+static void vcn_v3_0_set_enc_ring_funcs(struct amdgpu_device *adev)
+{
+	int i, j;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		for (j = 0; j < adev->vcn.inst[i].num_enc_rings; ++j) {
+			adev->vcn.inst[i].ring_enc[j].funcs = &vcn_v3_0_enc_ring_vm_funcs;
+			adev->vcn.inst[i].ring_enc[j].me = i;
+		}
+	}
+}
+
+static int vcn_v3_0_reset(struct amdgpu_vcn_inst *vinst)
+{
+	int r;
+
+	r = vcn_v3_0_stop(vinst);
+	if (r)
+		return r;
+	vcn_v3_0_enable_clock_gating(vinst);
+	vcn_v3_0_enable_static_power_gating(vinst);
+	return vcn_v3_0_start(vinst);
+}
+
+static bool vcn_v3_0_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, ret = 1;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		ret &= (RREG32_SOC15(VCN, i, mmUVD_STATUS) == UVD_STATUS__IDLE);
+	}
+
+	return ret;
+}
+
+static int vcn_v3_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, ret = 0;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		ret = SOC15_WAIT_ON_RREG(VCN, i, mmUVD_STATUS, UVD_STATUS__IDLE,
+			UVD_STATUS__IDLE);
+		if (ret)
+			return ret;
+	}
+
+	return ret;
+}
+
+static int vcn_v3_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = state == AMD_CG_STATE_GATE;
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[i];
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		if (enable) {
+			if (RREG32_SOC15(VCN, i, mmUVD_STATUS) != UVD_STATUS__IDLE)
+				return -EBUSY;
+			vcn_v3_0_enable_clock_gating(vinst);
+		} else {
+			vcn_v3_0_disable_clock_gating(vinst);
+		}
+	}
+
+	return 0;
+}
+
+static int vcn_v3_0_set_pg_state(struct amdgpu_vcn_inst *vinst,
+				 enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int ret = 0;
+
+	/* for SRIOV, guest should not control VCN Power-gating
+	 * MMSCH FW should control Power-gating and clock-gating
+	 * guest should avoid touching CGC and PG
+	 */
+	if (amdgpu_sriov_vf(adev)) {
+		vinst->cur_state = AMD_PG_STATE_UNGATE;
+		return 0;
+	}
+
+	if (state == vinst->cur_state)
+		return 0;
+
+	if (state == AMD_PG_STATE_GATE)
+		ret = vcn_v3_0_stop(vinst);
+	else
+		ret = vcn_v3_0_start(vinst);
+
+	if (!ret)
+		vinst->cur_state = state;
+
+	return ret;
+}
+
+static int vcn_v3_0_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+static int vcn_v3_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	uint32_t ip_instance;
+
+	switch (entry->client_id) {
+	case SOC15_IH_CLIENTID_VCN:
+		ip_instance = 0;
+		break;
+	case SOC15_IH_CLIENTID_VCN1:
+		ip_instance = 1;
+		break;
+	default:
+		DRM_ERROR("Unhandled client id: %d\n", entry->client_id);
+		return 0;
+	}
+
+	DRM_DEBUG("IH: VCN TRAP\n");
+
+	switch (entry->src_id) {
+	case VCN_2_0__SRCID__UVD_SYSTEM_MESSAGE_INTERRUPT:
+		amdgpu_fence_process(&adev->vcn.inst[ip_instance].ring_dec);
+		break;
+	case VCN_2_0__SRCID__UVD_ENC_GENERAL_PURPOSE:
+		amdgpu_fence_process(&adev->vcn.inst[ip_instance].ring_enc[0]);
+		break;
+	case VCN_2_0__SRCID__UVD_ENC_LOW_LATENCY:
+		amdgpu_fence_process(&adev->vcn.inst[ip_instance].ring_enc[1]);
+		break;
+	default:
+		DRM_ERROR("Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs vcn_v3_0_irq_funcs = {
+	.set = vcn_v3_0_set_interrupt_state,
+	.process = vcn_v3_0_process_interrupt,
+};
+
+static void vcn_v3_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		adev->vcn.inst[i].irq.num_types = adev->vcn.inst[i].num_enc_rings + 1;
+		adev->vcn.inst[i].irq.funcs = &vcn_v3_0_irq_funcs;
+	}
+}
+
+static const struct amd_ip_funcs vcn_v3_0_ip_funcs = {
+	.name = "vcn_v3_0",
+	.early_init = vcn_v3_0_early_init,
+	.sw_init = vcn_v3_0_sw_init,
+	.sw_fini = vcn_v3_0_sw_fini,
+	.hw_init = vcn_v3_0_hw_init,
+	.hw_fini = vcn_v3_0_hw_fini,
+	.suspend = vcn_v3_0_suspend,
+	.resume = vcn_v3_0_resume,
+	.is_idle = vcn_v3_0_is_idle,
+	.wait_for_idle = vcn_v3_0_wait_for_idle,
+	.set_clockgating_state = vcn_v3_0_set_clockgating_state,
+	.set_powergating_state = vcn_set_powergating_state,
+	.dump_ip_state = amdgpu_vcn_dump_ip_state,
+	.print_ip_state = amdgpu_vcn_print_ip_state,
+};
+
+const struct amdgpu_ip_block_version vcn_v3_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_VCN,
+	.major = 3,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &vcn_v3_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.h b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.h
new file mode 100644
index 000000000000..31683582d778
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VCN_V3_0_H__
+#define __VCN_V3_0_H__
+
+extern const struct amdgpu_ip_block_version vcn_v3_0_ip_block;
+
+#endif /* __VCN_V3_0_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c
new file mode 100644
index 000000000000..3ae666522d57
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c
@@ -0,0 +1,2315 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include "amdgpu.h"
+#include "amdgpu_vcn.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_cs.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "soc15_hw_ip.h"
+#include "vcn_v2_0.h"
+#include "mmsch_v4_0.h"
+#include "vcn_v4_0.h"
+
+#include "vcn/vcn_4_0_0_offset.h"
+#include "vcn/vcn_4_0_0_sh_mask.h"
+#include "ivsrcid/vcn/irqsrcs_vcn_4_0.h"
+
+#include <drm/drm_drv.h>
+
+#define mmUVD_DPG_LMA_CTL							regUVD_DPG_LMA_CTL
+#define mmUVD_DPG_LMA_CTL_BASE_IDX						regUVD_DPG_LMA_CTL_BASE_IDX
+#define mmUVD_DPG_LMA_DATA							regUVD_DPG_LMA_DATA
+#define mmUVD_DPG_LMA_DATA_BASE_IDX						regUVD_DPG_LMA_DATA_BASE_IDX
+
+#define VCN_VID_SOC_ADDRESS_2_0							0x1fb00
+#define VCN1_VID_SOC_ADDRESS_3_0						0x48300
+#define VCN1_AON_SOC_ADDRESS_3_0						0x48000
+
+#define VCN_HARVEST_MMSCH								0
+
+#define RDECODE_MSG_CREATE							0x00000000
+#define RDECODE_MESSAGE_CREATE							0x00000001
+
+static const struct amdgpu_hwip_reg_entry vcn_reg_list_4_0[] = {
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_CONTEXT_ID),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_CONTEXT_ID2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_GPCOM_VCPU_DATA0),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_GPCOM_VCPU_DATA1),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_GPCOM_VCPU_CMD),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_HI),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_LO),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_HI2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_LO2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_HI3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_LO3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_HI4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_LO4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_RPTR),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_WPTR),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_RPTR2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_WPTR2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_RPTR3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_WPTR3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_RPTR4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_WPTR4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_SIZE),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_SIZE2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_SIZE3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_SIZE4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_PGFSM_CONFIG),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_PGFSM_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_DPG_LMA_CTL),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_DPG_LMA_DATA),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_DPG_LMA_MASK),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_DPG_PAUSE)
+};
+
+static int amdgpu_ih_clientid_vcns[] = {
+	SOC15_IH_CLIENTID_VCN,
+	SOC15_IH_CLIENTID_VCN1
+};
+
+static int vcn_v4_0_start_sriov(struct amdgpu_device *adev);
+static void vcn_v4_0_set_unified_ring_funcs(struct amdgpu_device *adev);
+static void vcn_v4_0_set_irq_funcs(struct amdgpu_device *adev);
+static int vcn_v4_0_set_pg_state(struct amdgpu_vcn_inst *vinst,
+				 enum amd_powergating_state state);
+static int vcn_v4_0_pause_dpg_mode(struct amdgpu_vcn_inst *vinst,
+				   struct dpg_pause_state *new_state);
+static void vcn_v4_0_unified_ring_set_wptr(struct amdgpu_ring *ring);
+static void vcn_v4_0_set_ras_funcs(struct amdgpu_device *adev);
+
+/**
+ * vcn_v4_0_early_init - set function pointers and load microcode
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Set ring and irq function pointers
+ * Load microcode from filesystem
+ */
+static int vcn_v4_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, r;
+
+	if (amdgpu_sriov_vf(adev)) {
+		adev->vcn.harvest_config = VCN_HARVEST_MMSCH;
+		for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+			if (amdgpu_vcn_is_disabled_vcn(adev, VCN_ENCODE_RING, i)) {
+				adev->vcn.harvest_config |= 1 << i;
+				dev_info(adev->dev, "VCN%d is disabled by hypervisor\n", i);
+			}
+		}
+	}
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i)
+		/* re-use enc ring as unified ring */
+		adev->vcn.inst[i].num_enc_rings = 1;
+
+	vcn_v4_0_set_unified_ring_funcs(adev);
+	vcn_v4_0_set_irq_funcs(adev);
+	vcn_v4_0_set_ras_funcs(adev);
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		adev->vcn.inst[i].set_pg_state = vcn_v4_0_set_pg_state;
+
+		r = amdgpu_vcn_early_init(adev, i);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+static int vcn_v4_0_fw_shared_init(struct amdgpu_device *adev, int inst_idx)
+{
+	struct amdgpu_vcn4_fw_shared *fw_shared;
+
+	fw_shared = adev->vcn.inst[inst_idx].fw_shared.cpu_addr;
+	fw_shared->present_flag_0 = cpu_to_le32(AMDGPU_FW_SHARED_FLAG_0_UNIFIED_QUEUE);
+	fw_shared->sq.is_enabled = 1;
+
+	fw_shared->present_flag_0 |= cpu_to_le32(AMDGPU_VCN_SMU_DPM_INTERFACE_FLAG);
+	fw_shared->smu_dpm_interface.smu_interface_type = (adev->flags & AMD_IS_APU) ?
+		AMDGPU_VCN_SMU_DPM_INTERFACE_APU : AMDGPU_VCN_SMU_DPM_INTERFACE_DGPU;
+
+	if (amdgpu_ip_version(adev, VCN_HWIP, 0) ==
+	    IP_VERSION(4, 0, 2)) {
+		fw_shared->present_flag_0 |= AMDGPU_FW_SHARED_FLAG_0_DRM_KEY_INJECT;
+		fw_shared->drm_key_wa.method =
+			AMDGPU_DRM_KEY_INJECT_WORKAROUND_VCNFW_ASD_HANDSHAKING;
+	}
+
+	if (amdgpu_vcnfw_log)
+		amdgpu_vcn_fwlog_init(&adev->vcn.inst[inst_idx]);
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_sw_init - sw init for VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Load firmware and sw initialization
+ */
+static int vcn_v4_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_ring *ring;
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, r;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		r = amdgpu_vcn_sw_init(adev, i);
+		if (r)
+			return r;
+
+		amdgpu_vcn_setup_ucode(adev, i);
+
+		r = amdgpu_vcn_resume(adev, i);
+		if (r)
+			return r;
+
+		/* Init instance 0 sched_score to 1, so it's scheduled after other instances */
+		if (i == 0)
+			atomic_set(&adev->vcn.inst[i].sched_score, 1);
+		else
+			atomic_set(&adev->vcn.inst[i].sched_score, 0);
+
+		/* VCN UNIFIED TRAP */
+		r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_vcns[i],
+				VCN_4_0__SRCID__UVD_ENC_GENERAL_PURPOSE, &adev->vcn.inst[i].irq);
+		if (r)
+			return r;
+
+		/* VCN POISON TRAP */
+		r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_vcns[i],
+				VCN_4_0__SRCID_UVD_POISON, &adev->vcn.inst[i].ras_poison_irq);
+		if (r)
+			return r;
+
+		ring = &adev->vcn.inst[i].ring_enc[0];
+		ring->use_doorbell = true;
+		if (amdgpu_sriov_vf(adev))
+			ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + i *
+				(adev->vcn.inst[i].num_enc_rings + 1) + 1;
+		else
+			ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 2 + 8 * i;
+		ring->vm_hub = AMDGPU_MMHUB0(0);
+		sprintf(ring->name, "vcn_unified_%d", i);
+
+		r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst[i].irq, 0,
+						AMDGPU_RING_PRIO_0, &adev->vcn.inst[i].sched_score);
+		if (r)
+			return r;
+
+		vcn_v4_0_fw_shared_init(adev, i);
+
+		if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)
+			adev->vcn.inst[i].pause_dpg_mode = vcn_v4_0_pause_dpg_mode;
+	}
+
+	adev->vcn.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->vcn.inst[0].ring_enc[0]);
+	if (!amdgpu_sriov_vf(adev))
+		adev->vcn.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+
+	if (amdgpu_sriov_vf(adev)) {
+		r = amdgpu_virt_alloc_mm_table(adev);
+		if (r)
+			return r;
+	}
+
+
+	r = amdgpu_vcn_ras_sw_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_vcn_reg_dump_init(adev, vcn_reg_list_4_0, ARRAY_SIZE(vcn_reg_list_4_0));
+	if (r)
+		return r;
+
+	r = amdgpu_vcn_sysfs_reset_mask_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_sw_fini - sw fini for VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * VCN suspend and free up sw allocation
+ */
+static int vcn_v4_0_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, r, idx;
+
+	if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+		for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+			struct amdgpu_vcn4_fw_shared *fw_shared;
+
+			if (adev->vcn.harvest_config & (1 << i))
+				continue;
+
+			fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+			fw_shared->present_flag_0 = 0;
+			fw_shared->sq.is_enabled = 0;
+		}
+
+		drm_dev_exit(idx);
+	}
+
+	if (amdgpu_sriov_vf(adev))
+		amdgpu_virt_free_mm_table(adev);
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		r = amdgpu_vcn_suspend(adev, i);
+		if (r)
+			return r;
+	}
+
+	amdgpu_vcn_sysfs_reset_mask_fini(adev);
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++)
+		amdgpu_vcn_sw_fini(adev, i);
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_hw_init - start and test VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Initialize the hardware, boot up the VCPU and do some testing
+ */
+static int vcn_v4_0_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	int i, r;
+
+	if (amdgpu_sriov_vf(adev)) {
+		r = vcn_v4_0_start_sriov(adev);
+		if (r)
+			return r;
+
+		for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+			if (adev->vcn.harvest_config & (1 << i))
+				continue;
+
+			ring = &adev->vcn.inst[i].ring_enc[0];
+			ring->wptr = 0;
+			ring->wptr_old = 0;
+			vcn_v4_0_unified_ring_set_wptr(ring);
+			ring->sched.ready = true;
+		}
+	} else {
+		for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+			if (adev->vcn.harvest_config & (1 << i))
+				continue;
+
+			ring = &adev->vcn.inst[i].ring_enc[0];
+
+			adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
+					((adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 8 * i), i);
+
+			r = amdgpu_ring_test_helper(ring);
+			if (r)
+				return r;
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_hw_fini - stop the hardware block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Stop the VCN block, mark ring as not ready any more
+ */
+static int vcn_v4_0_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[i];
+
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		cancel_delayed_work_sync(&vinst->idle_work);
+
+		if (!amdgpu_sriov_vf(adev)) {
+			if ((adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) ||
+			    (vinst->cur_state != AMD_PG_STATE_GATE &&
+				 RREG32_SOC15(VCN, i, regUVD_STATUS))) {
+				vinst->set_pg_state(vinst, AMD_PG_STATE_GATE);
+			}
+		}
+		if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__VCN))
+			amdgpu_irq_put(adev, &vinst->ras_poison_irq, 0);
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_suspend - suspend VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * HW fini and suspend VCN block
+ */
+static int vcn_v4_0_suspend(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r, i;
+
+	r = vcn_v4_0_hw_fini(ip_block);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		r = amdgpu_vcn_suspend(ip_block->adev, i);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_resume - resume VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Resume firmware and hw init VCN block
+ */
+static int vcn_v4_0_resume(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r, i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		r = amdgpu_vcn_resume(ip_block->adev, i);
+		if (r)
+			return r;
+	}
+
+	r = vcn_v4_0_hw_init(ip_block);
+
+	return r;
+}
+
+/**
+ * vcn_v4_0_mc_resume - memory controller programming
+ *
+ * @vinst: VCN instance
+ *
+ * Let the VCN memory controller know it's offsets
+ */
+static void vcn_v4_0_mc_resume(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst = vinst->inst;
+	uint32_t offset, size;
+	const struct common_firmware_header *hdr;
+
+	hdr = (const struct common_firmware_header *)adev->vcn.inst[inst].fw->data;
+	size = AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8);
+
+	/* cache window 0: fw */
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst].tmr_mc_addr_lo));
+		WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst].tmr_mc_addr_hi));
+		WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_OFFSET0, 0);
+		offset = 0;
+	} else {
+		WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			lower_32_bits(adev->vcn.inst[inst].gpu_addr));
+		WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			upper_32_bits(adev->vcn.inst[inst].gpu_addr));
+		offset = size;
+                WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_OFFSET0, AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
+	}
+	WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_SIZE0, size);
+
+	/* cache window 1: stack */
+	WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW,
+		lower_32_bits(adev->vcn.inst[inst].gpu_addr + offset));
+	WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH,
+		upper_32_bits(adev->vcn.inst[inst].gpu_addr + offset));
+	WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_OFFSET1, 0);
+	WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_SIZE1, AMDGPU_VCN_STACK_SIZE);
+
+	/* cache window 2: context */
+	WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW,
+		lower_32_bits(adev->vcn.inst[inst].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE));
+	WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH,
+		upper_32_bits(adev->vcn.inst[inst].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE));
+	WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_OFFSET2, 0);
+	WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_SIZE2, AMDGPU_VCN_CONTEXT_SIZE);
+
+	/* non-cache window */
+	WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_NC0_64BIT_BAR_LOW,
+		lower_32_bits(adev->vcn.inst[inst].fw_shared.gpu_addr));
+	WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH,
+		upper_32_bits(adev->vcn.inst[inst].fw_shared.gpu_addr));
+	WREG32_SOC15(VCN, inst, regUVD_VCPU_NONCACHE_OFFSET0, 0);
+	WREG32_SOC15(VCN, inst, regUVD_VCPU_NONCACHE_SIZE0,
+		AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn4_fw_shared)));
+}
+
+/**
+ * vcn_v4_0_mc_resume_dpg_mode - memory controller programming for dpg mode
+ *
+ * @vinst: VCN instance
+ * @indirect: indirectly write sram
+ *
+ * Let the VCN memory controller know it's offsets with dpg mode
+ */
+static void vcn_v4_0_mc_resume_dpg_mode(struct amdgpu_vcn_inst *vinst,
+					bool indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t offset, size;
+	const struct common_firmware_header *hdr;
+	hdr = (const struct common_firmware_header *)adev->vcn.inst[inst_idx].fw->data;
+	size = AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8);
+
+	/* cache window 0: fw */
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		if (!indirect) {
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+				(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst_idx].tmr_mc_addr_lo), 0, indirect);
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+				(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst_idx].tmr_mc_addr_hi), 0, indirect);
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET0), 0, 0, indirect);
+		} else {
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW), 0, 0, indirect);
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH), 0, 0, indirect);
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET0), 0, 0, indirect);
+		}
+		offset = 0;
+	} else {
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr), 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr), 0, indirect);
+		offset = size;
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET0),
+			AMDGPU_UVD_FIRMWARE_OFFSET >> 3, 0, indirect);
+	}
+
+	if (!indirect)
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_VCPU_CACHE_SIZE0), size, 0, indirect);
+	else
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_VCPU_CACHE_SIZE0), 0, 0, indirect);
+
+	/* cache window 1: stack */
+	if (!indirect) {
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset), 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset), 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET1), 0, 0, indirect);
+	} else {
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW), 0, 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH), 0, 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET1), 0, 0, indirect);
+	}
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_VCPU_CACHE_SIZE1), AMDGPU_VCN_STACK_SIZE, 0, indirect);
+
+	/* cache window 2: context */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE), 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE), 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET2), 0, 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_VCPU_CACHE_SIZE2), AMDGPU_VCN_CONTEXT_SIZE, 0, indirect);
+
+	/* non-cache window */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_NC0_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[inst_idx].fw_shared.gpu_addr), 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[inst_idx].fw_shared.gpu_addr), 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_VCPU_NONCACHE_OFFSET0), 0, 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_VCPU_NONCACHE_SIZE0),
+			AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn4_fw_shared)), 0, indirect);
+
+	/* VCN global tiling registers */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_GFX10_ADDR_CONFIG),
+			adev->gfx.config.gb_addr_config, 0, indirect);
+}
+
+/**
+ * vcn_v4_0_disable_static_power_gating - disable VCN static power gating
+ *
+ * @vinst: VCN instance
+ *
+ * Disable static power gating for VCN block
+ */
+static void vcn_v4_0_disable_static_power_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst = vinst->inst;
+	uint32_t data = 0;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN) {
+		data = (1 << UVD_PGFSM_CONFIG__UVDM_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDS_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDLM_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDF_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTC_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDB_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTA_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTD_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTE_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDE_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDAB_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTB_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDNA_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDNB_PWR_CONFIG__SHIFT);
+
+		WREG32_SOC15(VCN, inst, regUVD_PGFSM_CONFIG, data);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_PGFSM_STATUS,
+			UVD_PGFSM_STATUS__UVDM_UVDU_UVDLM_PWR_ON_3_0, 0x3F3FFFFF);
+	} else {
+		uint32_t value;
+
+		value = (inst) ? 0x2200800 : 0;
+		data = (1 << UVD_PGFSM_CONFIG__UVDM_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDS_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDLM_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDF_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDTC_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDB_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDTA_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDTD_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDTE_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDE_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDAB_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDTB_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDNA_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDNB_PWR_CONFIG__SHIFT);
+
+                WREG32_SOC15(VCN, inst, regUVD_PGFSM_CONFIG, data);
+                SOC15_WAIT_ON_RREG(VCN, inst, regUVD_PGFSM_STATUS, value,  0x3F3FFFFF);
+        }
+
+        data = RREG32_SOC15(VCN, inst, regUVD_POWER_STATUS);
+        data &= ~0x103;
+        if (adev->pg_flags & AMD_PG_SUPPORT_VCN)
+                data |= UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON |
+                        UVD_POWER_STATUS__UVD_PG_EN_MASK;
+
+        WREG32_SOC15(VCN, inst, regUVD_POWER_STATUS, data);
+
+        return;
+}
+
+/**
+ * vcn_v4_0_enable_static_power_gating - enable VCN static power gating
+ *
+ * @vinst: VCN instance
+ *
+ * Enable static power gating for VCN block
+ */
+static void vcn_v4_0_enable_static_power_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst = vinst->inst;
+	uint32_t data;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN) {
+		/* Before power off, this indicator has to be turned on */
+		data = RREG32_SOC15(VCN, inst, regUVD_POWER_STATUS);
+		data &= ~UVD_POWER_STATUS__UVD_POWER_STATUS_MASK;
+		data |= UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF;
+		WREG32_SOC15(VCN, inst, regUVD_POWER_STATUS, data);
+
+		data = (2 << UVD_PGFSM_CONFIG__UVDM_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDS_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDF_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTC_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDB_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTA_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDLM_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTD_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTE_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDE_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDAB_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTB_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDNA_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDNB_PWR_CONFIG__SHIFT);
+		WREG32_SOC15(VCN, inst, regUVD_PGFSM_CONFIG, data);
+
+		data = (2 << UVD_PGFSM_STATUS__UVDM_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDS_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDF_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDTC_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDB_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDTA_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDLM_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDTD_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDTE_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDE_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDAB_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDTB_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDNA_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDNB_PWR_STATUS__SHIFT);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_PGFSM_STATUS, data, 0x3F3FFFFF);
+	}
+
+        return;
+}
+
+/**
+ * vcn_v4_0_disable_clock_gating - disable VCN clock gating
+ *
+ * @vinst: VCN instance
+ *
+ * Disable clock gating for VCN block
+ */
+static void vcn_v4_0_disable_clock_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst = vinst->inst;
+	uint32_t data;
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		return;
+
+	/* VCN disable CGC */
+	data = RREG32_SOC15(VCN, inst, regUVD_CGC_CTRL);
+	data &= ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK;
+	data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(VCN, inst, regUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, inst, regUVD_CGC_GATE);
+	data &= ~(UVD_CGC_GATE__SYS_MASK
+		| UVD_CGC_GATE__UDEC_MASK
+		| UVD_CGC_GATE__MPEG2_MASK
+		| UVD_CGC_GATE__REGS_MASK
+		| UVD_CGC_GATE__RBC_MASK
+		| UVD_CGC_GATE__LMI_MC_MASK
+		| UVD_CGC_GATE__LMI_UMC_MASK
+		| UVD_CGC_GATE__IDCT_MASK
+		| UVD_CGC_GATE__MPRD_MASK
+		| UVD_CGC_GATE__MPC_MASK
+		| UVD_CGC_GATE__LBSI_MASK
+		| UVD_CGC_GATE__LRBBM_MASK
+		| UVD_CGC_GATE__UDEC_RE_MASK
+		| UVD_CGC_GATE__UDEC_CM_MASK
+		| UVD_CGC_GATE__UDEC_IT_MASK
+		| UVD_CGC_GATE__UDEC_DB_MASK
+		| UVD_CGC_GATE__UDEC_MP_MASK
+		| UVD_CGC_GATE__WCB_MASK
+		| UVD_CGC_GATE__VCPU_MASK
+		| UVD_CGC_GATE__MMSCH_MASK);
+
+	WREG32_SOC15(VCN, inst, regUVD_CGC_GATE, data);
+	SOC15_WAIT_ON_RREG(VCN, inst, regUVD_CGC_GATE, 0,  0xFFFFFFFF);
+
+	data = RREG32_SOC15(VCN, inst, regUVD_CGC_CTRL);
+	data &= ~(UVD_CGC_CTRL__UDEC_RE_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_CM_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_IT_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_DB_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_MP_MODE_MASK
+		| UVD_CGC_CTRL__SYS_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_MODE_MASK
+		| UVD_CGC_CTRL__MPEG2_MODE_MASK
+		| UVD_CGC_CTRL__REGS_MODE_MASK
+		| UVD_CGC_CTRL__RBC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_MC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_UMC_MODE_MASK
+		| UVD_CGC_CTRL__IDCT_MODE_MASK
+		| UVD_CGC_CTRL__MPRD_MODE_MASK
+		| UVD_CGC_CTRL__MPC_MODE_MASK
+		| UVD_CGC_CTRL__LBSI_MODE_MASK
+		| UVD_CGC_CTRL__LRBBM_MODE_MASK
+		| UVD_CGC_CTRL__WCB_MODE_MASK
+		| UVD_CGC_CTRL__VCPU_MODE_MASK
+		| UVD_CGC_CTRL__MMSCH_MODE_MASK);
+	WREG32_SOC15(VCN, inst, regUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, inst, regUVD_SUVD_CGC_GATE);
+	data |= (UVD_SUVD_CGC_GATE__SRE_MASK
+		| UVD_SUVD_CGC_GATE__SIT_MASK
+		| UVD_SUVD_CGC_GATE__SMP_MASK
+		| UVD_SUVD_CGC_GATE__SCM_MASK
+		| UVD_SUVD_CGC_GATE__SDB_MASK
+		| UVD_SUVD_CGC_GATE__SRE_H264_MASK
+		| UVD_SUVD_CGC_GATE__SRE_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SIT_H264_MASK
+		| UVD_SUVD_CGC_GATE__SIT_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SCM_H264_MASK
+		| UVD_SUVD_CGC_GATE__SCM_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SDB_H264_MASK
+		| UVD_SUVD_CGC_GATE__SDB_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SCLR_MASK
+		| UVD_SUVD_CGC_GATE__UVD_SC_MASK
+		| UVD_SUVD_CGC_GATE__ENT_MASK
+		| UVD_SUVD_CGC_GATE__SIT_HEVC_DEC_MASK
+		| UVD_SUVD_CGC_GATE__SIT_HEVC_ENC_MASK
+		| UVD_SUVD_CGC_GATE__SITE_MASK
+		| UVD_SUVD_CGC_GATE__SRE_VP9_MASK
+		| UVD_SUVD_CGC_GATE__SCM_VP9_MASK
+		| UVD_SUVD_CGC_GATE__SIT_VP9_DEC_MASK
+		| UVD_SUVD_CGC_GATE__SDB_VP9_MASK
+		| UVD_SUVD_CGC_GATE__IME_HEVC_MASK);
+	WREG32_SOC15(VCN, inst, regUVD_SUVD_CGC_GATE, data);
+
+	data = RREG32_SOC15(VCN, inst, regUVD_SUVD_CGC_CTRL);
+	data &= ~(UVD_SUVD_CGC_CTRL__SRE_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SIT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SMP_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SCM_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SDB_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SCLR_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__UVD_SC_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__ENT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__IME_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SITE_MODE_MASK);
+	WREG32_SOC15(VCN, inst, regUVD_SUVD_CGC_CTRL, data);
+}
+
+/**
+ * vcn_v4_0_disable_clock_gating_dpg_mode - disable VCN clock gating dpg mode
+ *
+ * @vinst: VCN instance
+ * @sram_sel: sram select
+ * @indirect: indirectly write sram
+ *
+ * Disable clock gating for VCN block with dpg mode
+ */
+static void vcn_v4_0_disable_clock_gating_dpg_mode(struct amdgpu_vcn_inst *vinst,
+						   uint8_t sram_sel,
+						   uint8_t indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t reg_data = 0;
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		return;
+
+	/* enable sw clock gating control */
+	reg_data = 0 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	reg_data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	reg_data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	reg_data &= ~(UVD_CGC_CTRL__UDEC_RE_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_CM_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_IT_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_DB_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_MP_MODE_MASK |
+		 UVD_CGC_CTRL__SYS_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_MODE_MASK |
+		 UVD_CGC_CTRL__MPEG2_MODE_MASK |
+		 UVD_CGC_CTRL__REGS_MODE_MASK |
+		 UVD_CGC_CTRL__RBC_MODE_MASK |
+		 UVD_CGC_CTRL__LMI_MC_MODE_MASK |
+		 UVD_CGC_CTRL__LMI_UMC_MODE_MASK |
+		 UVD_CGC_CTRL__IDCT_MODE_MASK |
+		 UVD_CGC_CTRL__MPRD_MODE_MASK |
+		 UVD_CGC_CTRL__MPC_MODE_MASK |
+		 UVD_CGC_CTRL__LBSI_MODE_MASK |
+		 UVD_CGC_CTRL__LRBBM_MODE_MASK |
+		 UVD_CGC_CTRL__WCB_MODE_MASK |
+		 UVD_CGC_CTRL__VCPU_MODE_MASK);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_CGC_CTRL), reg_data, sram_sel, indirect);
+
+	/* turn off clock gating */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_CGC_GATE), 0, sram_sel, indirect);
+
+	/* turn on SUVD clock gating */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_SUVD_CGC_GATE), 1, sram_sel, indirect);
+
+	/* turn on sw mode in UVD_SUVD_CGC_CTRL */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_SUVD_CGC_CTRL), 0, sram_sel, indirect);
+}
+
+/**
+ * vcn_v4_0_enable_clock_gating - enable VCN clock gating
+ *
+ * @vinst: VCN instance
+ *
+ * Enable clock gating for VCN block
+ */
+static void vcn_v4_0_enable_clock_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst = vinst->inst;
+	uint32_t data;
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		return;
+
+	/* enable VCN CGC */
+	data = RREG32_SOC15(VCN, inst, regUVD_CGC_CTRL);
+	data |= 0 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(VCN, inst, regUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, inst, regUVD_CGC_CTRL);
+	data |= (UVD_CGC_CTRL__UDEC_RE_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_CM_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_IT_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_DB_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_MP_MODE_MASK
+		| UVD_CGC_CTRL__SYS_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_MODE_MASK
+		| UVD_CGC_CTRL__MPEG2_MODE_MASK
+		| UVD_CGC_CTRL__REGS_MODE_MASK
+		| UVD_CGC_CTRL__RBC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_MC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_UMC_MODE_MASK
+		| UVD_CGC_CTRL__IDCT_MODE_MASK
+		| UVD_CGC_CTRL__MPRD_MODE_MASK
+		| UVD_CGC_CTRL__MPC_MODE_MASK
+		| UVD_CGC_CTRL__LBSI_MODE_MASK
+		| UVD_CGC_CTRL__LRBBM_MODE_MASK
+		| UVD_CGC_CTRL__WCB_MODE_MASK
+		| UVD_CGC_CTRL__VCPU_MODE_MASK
+		| UVD_CGC_CTRL__MMSCH_MODE_MASK);
+	WREG32_SOC15(VCN, inst, regUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, inst, regUVD_SUVD_CGC_CTRL);
+	data |= (UVD_SUVD_CGC_CTRL__SRE_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SIT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SMP_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SCM_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SDB_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SCLR_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__UVD_SC_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__ENT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__IME_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SITE_MODE_MASK);
+	WREG32_SOC15(VCN, inst, regUVD_SUVD_CGC_CTRL, data);
+}
+
+static void vcn_v4_0_enable_ras(struct amdgpu_vcn_inst *vinst,
+				bool indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t tmp;
+
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__VCN))
+		return;
+
+	tmp = VCN_RAS_CNTL__VCPU_VCODEC_REARM_MASK |
+	      VCN_RAS_CNTL__VCPU_VCODEC_IH_EN_MASK |
+	      VCN_RAS_CNTL__VCPU_VCODEC_PMI_EN_MASK |
+	      VCN_RAS_CNTL__VCPU_VCODEC_STALL_EN_MASK;
+	WREG32_SOC15_DPG_MODE(inst_idx,
+			      SOC15_DPG_MODE_OFFSET(VCN, 0, regVCN_RAS_CNTL),
+			      tmp, 0, indirect);
+
+	tmp = UVD_SYS_INT_EN__RASCNTL_VCPU_VCODEC_EN_MASK;
+	WREG32_SOC15_DPG_MODE(inst_idx,
+			      SOC15_DPG_MODE_OFFSET(VCN, 0, regUVD_SYS_INT_EN),
+			      tmp, 0, indirect);
+}
+
+/**
+ * vcn_v4_0_start_dpg_mode - VCN start with dpg mode
+ *
+ * @vinst: VCN instance
+ * @indirect: indirectly write sram
+ *
+ * Start VCN block with dpg mode
+ */
+static int vcn_v4_0_start_dpg_mode(struct amdgpu_vcn_inst *vinst, bool indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	struct amdgpu_vcn4_fw_shared *fw_shared = adev->vcn.inst[inst_idx].fw_shared.cpu_addr;
+	struct amdgpu_ring *ring;
+	uint32_t tmp;
+	int ret;
+
+	/* disable register anti-hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, regUVD_POWER_STATUS), 1,
+		~UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+	/* enable dynamic power gating mode */
+	tmp = RREG32_SOC15(VCN, inst_idx, regUVD_POWER_STATUS);
+	tmp |= UVD_POWER_STATUS__UVD_PG_MODE_MASK;
+	tmp |= UVD_POWER_STATUS__UVD_PG_EN_MASK;
+	WREG32_SOC15(VCN, inst_idx, regUVD_POWER_STATUS, tmp);
+
+	if (indirect)
+		adev->vcn.inst[inst_idx].dpg_sram_curr_addr = (uint32_t *)adev->vcn.inst[inst_idx].dpg_sram_cpu_addr;
+
+	/* enable clock gating */
+	vcn_v4_0_disable_clock_gating_dpg_mode(vinst, 0, indirect);
+
+	/* enable VCPU clock */
+	tmp = (0xFF << UVD_VCPU_CNTL__PRB_TIMEOUT_VAL__SHIFT);
+	tmp |= UVD_VCPU_CNTL__CLK_EN_MASK | UVD_VCPU_CNTL__BLK_RST_MASK;
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_VCPU_CNTL), tmp, 0, indirect);
+
+	/* disable master interupt */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_MASTINT_EN), 0, 0, indirect);
+
+	/* setup regUVD_LMI_CTRL */
+	tmp = (UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+		UVD_LMI_CTRL__REQ_MODE_MASK |
+		UVD_LMI_CTRL__CRC_RESET_MASK |
+		UVD_LMI_CTRL__MASK_MC_URGENT_MASK |
+		UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+		UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
+		(8 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
+		0x00100000L);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_LMI_CTRL), tmp, 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_MPC_CNTL),
+		0x2 << UVD_MPC_CNTL__REPLACEMENT_MODE__SHIFT, 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_MPC_SET_MUXA0),
+		((0x1 << UVD_MPC_SET_MUXA0__VARA_1__SHIFT) |
+		 (0x2 << UVD_MPC_SET_MUXA0__VARA_2__SHIFT) |
+		 (0x3 << UVD_MPC_SET_MUXA0__VARA_3__SHIFT) |
+		 (0x4 << UVD_MPC_SET_MUXA0__VARA_4__SHIFT)), 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_MPC_SET_MUXB0),
+		 ((0x1 << UVD_MPC_SET_MUXB0__VARB_1__SHIFT) |
+		 (0x2 << UVD_MPC_SET_MUXB0__VARB_2__SHIFT) |
+		 (0x3 << UVD_MPC_SET_MUXB0__VARB_3__SHIFT) |
+		 (0x4 << UVD_MPC_SET_MUXB0__VARB_4__SHIFT)), 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_MPC_SET_MUX),
+		((0x0 << UVD_MPC_SET_MUX__SET_0__SHIFT) |
+		 (0x1 << UVD_MPC_SET_MUX__SET_1__SHIFT) |
+		 (0x2 << UVD_MPC_SET_MUX__SET_2__SHIFT)), 0, indirect);
+
+	vcn_v4_0_mc_resume_dpg_mode(vinst, indirect);
+
+	tmp = (0xFF << UVD_VCPU_CNTL__PRB_TIMEOUT_VAL__SHIFT);
+	tmp |= UVD_VCPU_CNTL__CLK_EN_MASK;
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_VCPU_CNTL), tmp, 0, indirect);
+
+	/* enable LMI MC and UMC channels */
+	tmp = 0x1f << UVD_LMI_CTRL2__RE_OFLD_MIF_WR_REQ_NUM__SHIFT;
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_LMI_CTRL2), tmp, 0, indirect);
+
+	vcn_v4_0_enable_ras(vinst, indirect);
+
+	/* enable master interrupt */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_MASTINT_EN),
+		UVD_MASTINT_EN__VCPU_EN_MASK, 0, indirect);
+
+
+	if (indirect) {
+		ret = amdgpu_vcn_psp_update_sram(adev, inst_idx, 0);
+		if (ret) {
+			dev_err(adev->dev, "vcn sram load failed %d\n", ret);
+			return ret;
+		}
+	}
+
+	ring = &adev->vcn.inst[inst_idx].ring_enc[0];
+
+	WREG32_SOC15(VCN, inst_idx, regUVD_RB_BASE_LO, ring->gpu_addr);
+	WREG32_SOC15(VCN, inst_idx, regUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(VCN, inst_idx, regUVD_RB_SIZE, ring->ring_size / 4);
+
+	tmp = RREG32_SOC15(VCN, inst_idx, regVCN_RB_ENABLE);
+	tmp &= ~(VCN_RB_ENABLE__RB1_EN_MASK);
+	WREG32_SOC15(VCN, inst_idx, regVCN_RB_ENABLE, tmp);
+	fw_shared->sq.queue_mode |= FW_QUEUE_RING_RESET;
+	WREG32_SOC15(VCN, inst_idx, regUVD_RB_RPTR, 0);
+	WREG32_SOC15(VCN, inst_idx, regUVD_RB_WPTR, 0);
+
+	tmp = RREG32_SOC15(VCN, inst_idx, regUVD_RB_RPTR);
+	WREG32_SOC15(VCN, inst_idx, regUVD_RB_WPTR, tmp);
+	ring->wptr = RREG32_SOC15(VCN, inst_idx, regUVD_RB_WPTR);
+
+	tmp = RREG32_SOC15(VCN, inst_idx, regVCN_RB_ENABLE);
+	tmp |= VCN_RB_ENABLE__RB1_EN_MASK;
+	WREG32_SOC15(VCN, inst_idx, regVCN_RB_ENABLE, tmp);
+	fw_shared->sq.queue_mode &= ~(FW_QUEUE_RING_RESET | FW_QUEUE_DPG_HOLD_OFF);
+
+	WREG32_SOC15(VCN, inst_idx, regVCN_RB1_DB_CTRL,
+			ring->doorbell_index << VCN_RB1_DB_CTRL__OFFSET__SHIFT |
+			VCN_RB1_DB_CTRL__EN_MASK);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, inst_idx, regUVD_STATUS);
+
+	return 0;
+}
+
+
+/**
+ * vcn_v4_0_start - VCN start
+ *
+ * @vinst: VCN instance
+ *
+ * Start VCN block
+ */
+static int vcn_v4_0_start(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int i = vinst->inst;
+	struct amdgpu_vcn4_fw_shared *fw_shared;
+	struct amdgpu_ring *ring;
+	uint32_t tmp;
+	int j, k, r;
+
+	if (adev->vcn.harvest_config & (1 << i))
+		return 0;
+
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_vcn(adev, true, i);
+
+	fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)
+		return vcn_v4_0_start_dpg_mode(vinst, adev->vcn.inst[i].indirect_sram);
+
+	/* disable VCN power gating */
+	vcn_v4_0_disable_static_power_gating(vinst);
+
+	/* set VCN status busy */
+	tmp = RREG32_SOC15(VCN, i, regUVD_STATUS) | UVD_STATUS__UVD_BUSY;
+	WREG32_SOC15(VCN, i, regUVD_STATUS, tmp);
+
+	/*SW clock gating */
+	vcn_v4_0_disable_clock_gating(vinst);
+
+	/* enable VCPU clock */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL),
+		 UVD_VCPU_CNTL__CLK_EN_MASK, ~UVD_VCPU_CNTL__CLK_EN_MASK);
+
+	/* disable master interrupt */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_MASTINT_EN), 0,
+		 ~UVD_MASTINT_EN__VCPU_EN_MASK);
+
+	/* enable LMI MC and UMC channels */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_LMI_CTRL2), 0,
+		 ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+
+	tmp = RREG32_SOC15(VCN, i, regUVD_SOFT_RESET);
+	tmp &= ~UVD_SOFT_RESET__LMI_SOFT_RESET_MASK;
+	tmp &= ~UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK;
+	WREG32_SOC15(VCN, i, regUVD_SOFT_RESET, tmp);
+
+	/* setup regUVD_LMI_CTRL */
+	tmp = RREG32_SOC15(VCN, i, regUVD_LMI_CTRL);
+	WREG32_SOC15(VCN, i, regUVD_LMI_CTRL, tmp |
+		     UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+		     UVD_LMI_CTRL__MASK_MC_URGENT_MASK |
+		     UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+		     UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK);
+
+	/* setup regUVD_MPC_CNTL */
+	tmp = RREG32_SOC15(VCN, i, regUVD_MPC_CNTL);
+	tmp &= ~UVD_MPC_CNTL__REPLACEMENT_MODE_MASK;
+	tmp |= 0x2 << UVD_MPC_CNTL__REPLACEMENT_MODE__SHIFT;
+	WREG32_SOC15(VCN, i, regUVD_MPC_CNTL, tmp);
+
+	/* setup UVD_MPC_SET_MUXA0 */
+	WREG32_SOC15(VCN, i, regUVD_MPC_SET_MUXA0,
+		     ((0x1 << UVD_MPC_SET_MUXA0__VARA_1__SHIFT) |
+		      (0x2 << UVD_MPC_SET_MUXA0__VARA_2__SHIFT) |
+		      (0x3 << UVD_MPC_SET_MUXA0__VARA_3__SHIFT) |
+		      (0x4 << UVD_MPC_SET_MUXA0__VARA_4__SHIFT)));
+
+	/* setup UVD_MPC_SET_MUXB0 */
+	WREG32_SOC15(VCN, i, regUVD_MPC_SET_MUXB0,
+		     ((0x1 << UVD_MPC_SET_MUXB0__VARB_1__SHIFT) |
+		      (0x2 << UVD_MPC_SET_MUXB0__VARB_2__SHIFT) |
+		      (0x3 << UVD_MPC_SET_MUXB0__VARB_3__SHIFT) |
+		      (0x4 << UVD_MPC_SET_MUXB0__VARB_4__SHIFT)));
+
+	/* setup UVD_MPC_SET_MUX */
+	WREG32_SOC15(VCN, i, regUVD_MPC_SET_MUX,
+		     ((0x0 << UVD_MPC_SET_MUX__SET_0__SHIFT) |
+		      (0x1 << UVD_MPC_SET_MUX__SET_1__SHIFT) |
+		      (0x2 << UVD_MPC_SET_MUX__SET_2__SHIFT)));
+
+	vcn_v4_0_mc_resume(vinst);
+
+	/* VCN global tiling registers */
+	WREG32_SOC15(VCN, i, regUVD_GFX10_ADDR_CONFIG,
+		     adev->gfx.config.gb_addr_config);
+
+	/* unblock VCPU register access */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_RB_ARB_CTRL), 0,
+		 ~UVD_RB_ARB_CTRL__VCPU_DIS_MASK);
+
+	/* release VCPU reset to boot */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), 0,
+		 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+	for (j = 0; j < 10; ++j) {
+		uint32_t status;
+
+		for (k = 0; k < 100; ++k) {
+			status = RREG32_SOC15(VCN, i, regUVD_STATUS);
+			if (status & 2)
+				break;
+			mdelay(10);
+			if (amdgpu_emu_mode == 1)
+				msleep(1);
+		}
+
+		if (amdgpu_emu_mode == 1) {
+			r = -1;
+			if (status & 2) {
+				r = 0;
+				break;
+			}
+		} else {
+			r = 0;
+			if (status & 2)
+				break;
+
+			dev_err(adev->dev, "VCN[%d] is not responding, trying to reset the VCPU!!!\n", i);
+			WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL),
+				 UVD_VCPU_CNTL__BLK_RST_MASK,
+				 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+			mdelay(10);
+			WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), 0,
+				 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+			mdelay(10);
+			r = -1;
+		}
+	}
+
+	if (r) {
+		dev_err(adev->dev, "VCN[%d] is not responding, giving up!!!\n", i);
+		return r;
+	}
+
+	/* enable master interrupt */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_MASTINT_EN),
+		 UVD_MASTINT_EN__VCPU_EN_MASK,
+		 ~UVD_MASTINT_EN__VCPU_EN_MASK);
+
+	/* clear the busy bit of VCN_STATUS */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_STATUS), 0,
+		 ~(2 << UVD_STATUS__VCPU_REPORT__SHIFT));
+
+	ring = &adev->vcn.inst[i].ring_enc[0];
+	WREG32_SOC15(VCN, i, regVCN_RB1_DB_CTRL,
+		     ring->doorbell_index << VCN_RB1_DB_CTRL__OFFSET__SHIFT |
+		     VCN_RB1_DB_CTRL__EN_MASK);
+
+	WREG32_SOC15(VCN, i, regUVD_RB_BASE_LO, ring->gpu_addr);
+	WREG32_SOC15(VCN, i, regUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(VCN, i, regUVD_RB_SIZE, ring->ring_size / 4);
+
+	tmp = RREG32_SOC15(VCN, i, regVCN_RB_ENABLE);
+	tmp &= ~(VCN_RB_ENABLE__RB1_EN_MASK);
+	WREG32_SOC15(VCN, i, regVCN_RB_ENABLE, tmp);
+	fw_shared->sq.queue_mode |= FW_QUEUE_RING_RESET;
+	WREG32_SOC15(VCN, i, regUVD_RB_RPTR, 0);
+	WREG32_SOC15(VCN, i, regUVD_RB_WPTR, 0);
+
+	tmp = RREG32_SOC15(VCN, i, regUVD_RB_RPTR);
+	WREG32_SOC15(VCN, i, regUVD_RB_WPTR, tmp);
+	ring->wptr = RREG32_SOC15(VCN, i, regUVD_RB_WPTR);
+
+	tmp = RREG32_SOC15(VCN, i, regVCN_RB_ENABLE);
+	tmp |= VCN_RB_ENABLE__RB1_EN_MASK;
+	WREG32_SOC15(VCN, i, regVCN_RB_ENABLE, tmp);
+	fw_shared->sq.queue_mode &= ~(FW_QUEUE_RING_RESET | FW_QUEUE_DPG_HOLD_OFF);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, i, regUVD_STATUS);
+
+	return 0;
+}
+
+static int vcn_v4_0_init_ring_metadata(struct amdgpu_device *adev, uint32_t vcn_inst, struct amdgpu_ring *ring_enc)
+{
+	struct amdgpu_vcn_rb_metadata *rb_metadata = NULL;
+	uint8_t *rb_ptr = (uint8_t *)ring_enc->ring;
+
+	rb_ptr += ring_enc->ring_size;
+	rb_metadata = (struct amdgpu_vcn_rb_metadata *)rb_ptr;
+
+	memset(rb_metadata, 0, sizeof(struct amdgpu_vcn_rb_metadata));
+	rb_metadata->size = sizeof(struct amdgpu_vcn_rb_metadata);
+	rb_metadata->present_flag_0 |= cpu_to_le32(AMDGPU_VCN_VF_RB_SETUP_FLAG);
+	rb_metadata->present_flag_0 |= cpu_to_le32(AMDGPU_VCN_VF_RB_DECOUPLE_FLAG);
+	rb_metadata->version = 1;
+	rb_metadata->ring_id = vcn_inst & 0xFF;
+
+	return 0;
+}
+
+static int vcn_v4_0_start_sriov(struct amdgpu_device *adev)
+{
+	int i;
+	struct amdgpu_ring *ring_enc;
+	uint64_t cache_addr;
+	uint64_t rb_enc_addr;
+	uint64_t ctx_addr;
+	uint32_t param, resp, expected;
+	uint32_t offset, cache_size;
+	uint32_t tmp, timeout;
+
+	struct amdgpu_mm_table *table = &adev->virt.mm_table;
+	uint32_t *table_loc;
+	uint32_t table_size;
+	uint32_t size, size_dw;
+	uint32_t init_status;
+	uint32_t enabled_vcn;
+
+	struct mmsch_v4_0_cmd_direct_write
+		direct_wt = { {0} };
+	struct mmsch_v4_0_cmd_direct_read_modify_write
+		direct_rd_mod_wt = { {0} };
+	struct mmsch_v4_0_cmd_end end = { {0} };
+	struct mmsch_v4_0_init_header header;
+
+	struct amdgpu_vcn4_fw_shared *fw_shared;
+	struct amdgpu_fw_shared_rb_setup *rb_setup;
+
+	direct_wt.cmd_header.command_type =
+		MMSCH_COMMAND__DIRECT_REG_WRITE;
+	direct_rd_mod_wt.cmd_header.command_type =
+		MMSCH_COMMAND__DIRECT_REG_READ_MODIFY_WRITE;
+	end.cmd_header.command_type =
+		MMSCH_COMMAND__END;
+
+	header.version = MMSCH_VERSION;
+	header.total_size = sizeof(struct mmsch_v4_0_init_header) >> 2;
+	for (i = 0; i < MMSCH_V4_0_VCN_INSTANCES; i++) {
+		header.inst[i].init_status = 0;
+		header.inst[i].table_offset = 0;
+		header.inst[i].table_size = 0;
+	}
+
+	table_loc = (uint32_t *)table->cpu_addr;
+	table_loc += header.total_size;
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		// Must re/init fw_shared at beginning
+		vcn_v4_0_fw_shared_init(adev, i);
+
+		table_size = 0;
+
+		MMSCH_V4_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(VCN, i,
+			regUVD_STATUS),
+			~UVD_STATUS__UVD_BUSY, UVD_STATUS__UVD_BUSY);
+
+		cache_size = AMDGPU_GPU_PAGE_ALIGN(adev->vcn.inst[i].fw->size + 4);
+
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+			MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+				regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+				adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + i].tmr_mc_addr_lo);
+			MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+				regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+				adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + i].tmr_mc_addr_hi);
+			offset = 0;
+			MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+				regUVD_VCPU_CACHE_OFFSET0),
+				0);
+		} else {
+			MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+				regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+				lower_32_bits(adev->vcn.inst[i].gpu_addr));
+			MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+				regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+				upper_32_bits(adev->vcn.inst[i].gpu_addr));
+			offset = cache_size;
+			MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+				regUVD_VCPU_CACHE_OFFSET0),
+				AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
+		}
+
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			regUVD_VCPU_CACHE_SIZE0),
+			cache_size);
+
+		cache_addr = adev->vcn.inst[i].gpu_addr + offset;
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			regUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW),
+			lower_32_bits(cache_addr));
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			regUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH),
+			upper_32_bits(cache_addr));
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			regUVD_VCPU_CACHE_OFFSET1),
+			0);
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			regUVD_VCPU_CACHE_SIZE1),
+			AMDGPU_VCN_STACK_SIZE);
+
+		cache_addr = adev->vcn.inst[i].gpu_addr + offset +
+			AMDGPU_VCN_STACK_SIZE;
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			regUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW),
+			lower_32_bits(cache_addr));
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			regUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH),
+			upper_32_bits(cache_addr));
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			regUVD_VCPU_CACHE_OFFSET2),
+			0);
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			regUVD_VCPU_CACHE_SIZE2),
+			AMDGPU_VCN_CONTEXT_SIZE);
+
+		fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+		rb_setup = &fw_shared->rb_setup;
+
+		ring_enc = &adev->vcn.inst[i].ring_enc[0];
+		ring_enc->wptr = 0;
+		rb_enc_addr = ring_enc->gpu_addr;
+
+		rb_setup->is_rb_enabled_flags |= RB_ENABLED;
+		fw_shared->present_flag_0 |= cpu_to_le32(AMDGPU_VCN_VF_RB_SETUP_FLAG);
+
+		if (amdgpu_sriov_is_vcn_rb_decouple(adev)) {
+			vcn_v4_0_init_ring_metadata(adev, i, ring_enc);
+
+			memset((void *)&rb_setup->rb_info, 0, sizeof(struct amdgpu_vcn_rb_setup_info) * MAX_NUM_VCN_RB_SETUP);
+			if (!(adev->vcn.harvest_config & (1 << 0))) {
+				rb_setup->rb_info[0].rb_addr_lo = lower_32_bits(adev->vcn.inst[0].ring_enc[0].gpu_addr);
+				rb_setup->rb_info[0].rb_addr_hi = upper_32_bits(adev->vcn.inst[0].ring_enc[0].gpu_addr);
+				rb_setup->rb_info[0].rb_size = adev->vcn.inst[0].ring_enc[0].ring_size / 4;
+			}
+			if (!(adev->vcn.harvest_config & (1 << 1))) {
+				rb_setup->rb_info[2].rb_addr_lo = lower_32_bits(adev->vcn.inst[1].ring_enc[0].gpu_addr);
+				rb_setup->rb_info[2].rb_addr_hi = upper_32_bits(adev->vcn.inst[1].ring_enc[0].gpu_addr);
+				rb_setup->rb_info[2].rb_size = adev->vcn.inst[1].ring_enc[0].ring_size / 4;
+			}
+			fw_shared->decouple.is_enabled = 1;
+			fw_shared->present_flag_0 |= cpu_to_le32(AMDGPU_VCN_VF_RB_DECOUPLE_FLAG);
+		} else {
+			rb_setup->rb_addr_lo = lower_32_bits(rb_enc_addr);
+			rb_setup->rb_addr_hi = upper_32_bits(rb_enc_addr);
+			rb_setup->rb_size = ring_enc->ring_size / 4;
+		}
+
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			regUVD_LMI_VCPU_NC0_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[i].fw_shared.gpu_addr));
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			regUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[i].fw_shared.gpu_addr));
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, i,
+			regUVD_VCPU_NONCACHE_SIZE0),
+			AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn4_fw_shared)));
+
+		/* add end packet */
+		MMSCH_V4_0_INSERT_END();
+
+		/* refine header */
+		header.inst[i].init_status = 0;
+		header.inst[i].table_offset = header.total_size;
+		header.inst[i].table_size = table_size;
+		header.total_size += table_size;
+	}
+
+	/* Update init table header in memory */
+	size = sizeof(struct mmsch_v4_0_init_header);
+	table_loc = (uint32_t *)table->cpu_addr;
+	memcpy((void *)table_loc, &header, size);
+
+	/* message MMSCH (in VCN[0]) to initialize this client
+	 * 1, write to mmsch_vf_ctx_addr_lo/hi register with GPU mc addr
+	 * of memory descriptor location
+	 */
+	ctx_addr = table->gpu_addr;
+	WREG32_SOC15(VCN, 0, regMMSCH_VF_CTX_ADDR_LO, lower_32_bits(ctx_addr));
+	WREG32_SOC15(VCN, 0, regMMSCH_VF_CTX_ADDR_HI, upper_32_bits(ctx_addr));
+
+	/* 2, update vmid of descriptor */
+	tmp = RREG32_SOC15(VCN, 0, regMMSCH_VF_VMID);
+	tmp &= ~MMSCH_VF_VMID__VF_CTX_VMID_MASK;
+	/* use domain0 for MM scheduler */
+	tmp |= (0 << MMSCH_VF_VMID__VF_CTX_VMID__SHIFT);
+	WREG32_SOC15(VCN, 0, regMMSCH_VF_VMID, tmp);
+
+	/* 3, notify mmsch about the size of this descriptor */
+	size = header.total_size;
+	WREG32_SOC15(VCN, 0, regMMSCH_VF_CTX_SIZE, size);
+
+	/* 4, set resp to zero */
+	WREG32_SOC15(VCN, 0, regMMSCH_VF_MAILBOX_RESP, 0);
+
+	/* 5, kick off the initialization and wait until
+	 * MMSCH_VF_MAILBOX_RESP becomes non-zero
+	 */
+	param = 0x00000001;
+	WREG32_SOC15(VCN, 0, regMMSCH_VF_MAILBOX_HOST, param);
+	tmp = 0;
+	timeout = 1000;
+	resp = 0;
+	expected = MMSCH_VF_MAILBOX_RESP__OK;
+	while (resp != expected) {
+		resp = RREG32_SOC15(VCN, 0, regMMSCH_VF_MAILBOX_RESP);
+		if (resp != 0)
+			break;
+
+		udelay(10);
+		tmp = tmp + 10;
+		if (tmp >= timeout) {
+			DRM_ERROR("failed to init MMSCH. TIME-OUT after %d usec"\
+				" waiting for regMMSCH_VF_MAILBOX_RESP "\
+				"(expected=0x%08x, readback=0x%08x)\n",
+				tmp, expected, resp);
+			return -EBUSY;
+		}
+	}
+	enabled_vcn = amdgpu_vcn_is_disabled_vcn(adev, VCN_DECODE_RING, 0) ? 1 : 0;
+	init_status = ((struct mmsch_v4_0_init_header *)(table_loc))->inst[enabled_vcn].init_status;
+	if (resp != expected && resp != MMSCH_VF_MAILBOX_RESP__INCOMPLETE
+	&& init_status != MMSCH_VF_ENGINE_STATUS__PASS)
+		DRM_ERROR("MMSCH init status is incorrect! readback=0x%08x, header init "\
+			"status for VCN%x: 0x%x\n", resp, enabled_vcn, init_status);
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_stop_dpg_mode - VCN stop with dpg mode
+ *
+ * @vinst: VCN instance
+ *
+ * Stop VCN block with dpg mode
+ */
+static void vcn_v4_0_stop_dpg_mode(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	struct dpg_pause_state state = {.fw_based = VCN_DPG_STATE__UNPAUSE};
+	uint32_t tmp;
+
+	vcn_v4_0_pause_dpg_mode(vinst, &state);
+	/* Wait for power status to be 1 */
+	SOC15_WAIT_ON_RREG(VCN, inst_idx, regUVD_POWER_STATUS, 1,
+		UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+	/* wait for read ptr to be equal to write ptr */
+	tmp = RREG32_SOC15(VCN, inst_idx, regUVD_RB_WPTR);
+	SOC15_WAIT_ON_RREG(VCN, inst_idx, regUVD_RB_RPTR, tmp, 0xFFFFFFFF);
+
+	SOC15_WAIT_ON_RREG(VCN, inst_idx, regUVD_POWER_STATUS, 1,
+		UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+	/* disable dynamic power gating mode */
+	WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, regUVD_POWER_STATUS), 0,
+		~UVD_POWER_STATUS__UVD_PG_MODE_MASK);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, inst_idx, regUVD_STATUS);
+}
+
+/**
+ * vcn_v4_0_stop - VCN stop
+ *
+ * @vinst: VCN instance
+ *
+ * Stop VCN block
+ */
+static int vcn_v4_0_stop(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int i = vinst->inst;
+	struct amdgpu_vcn4_fw_shared *fw_shared;
+	uint32_t tmp;
+	int r = 0;
+
+	if (adev->vcn.harvest_config & (1 << i))
+		return 0;
+
+	fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+	fw_shared->sq.queue_mode |= FW_QUEUE_DPG_HOLD_OFF;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) {
+		vcn_v4_0_stop_dpg_mode(vinst);
+		goto done;
+	}
+
+	/* wait for vcn idle */
+	r = SOC15_WAIT_ON_RREG(VCN, i, regUVD_STATUS, UVD_STATUS__IDLE, 0x7);
+	if (r)
+		goto done;
+
+	tmp = UVD_LMI_STATUS__VCPU_LMI_WRITE_CLEAN_MASK |
+		UVD_LMI_STATUS__READ_CLEAN_MASK |
+		UVD_LMI_STATUS__WRITE_CLEAN_MASK |
+		UVD_LMI_STATUS__WRITE_CLEAN_RAW_MASK;
+	r = SOC15_WAIT_ON_RREG(VCN, i, regUVD_LMI_STATUS, tmp, tmp);
+	if (r)
+		goto done;
+
+	/* disable LMI UMC channel */
+	tmp = RREG32_SOC15(VCN, i, regUVD_LMI_CTRL2);
+	tmp |= UVD_LMI_CTRL2__STALL_ARB_UMC_MASK;
+	WREG32_SOC15(VCN, i, regUVD_LMI_CTRL2, tmp);
+	tmp = UVD_LMI_STATUS__UMC_READ_CLEAN_RAW_MASK |
+		UVD_LMI_STATUS__UMC_WRITE_CLEAN_RAW_MASK;
+	r = SOC15_WAIT_ON_RREG(VCN, i, regUVD_LMI_STATUS, tmp, tmp);
+	if (r)
+		goto done;
+
+	/* block VCPU register access */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_RB_ARB_CTRL),
+		 UVD_RB_ARB_CTRL__VCPU_DIS_MASK,
+		 ~UVD_RB_ARB_CTRL__VCPU_DIS_MASK);
+
+	/* reset VCPU */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL),
+		 UVD_VCPU_CNTL__BLK_RST_MASK,
+		 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+	/* disable VCPU clock */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), 0,
+		 ~(UVD_VCPU_CNTL__CLK_EN_MASK));
+
+	/* apply soft reset */
+	tmp = RREG32_SOC15(VCN, i, regUVD_SOFT_RESET);
+	tmp |= UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK;
+	WREG32_SOC15(VCN, i, regUVD_SOFT_RESET, tmp);
+	tmp = RREG32_SOC15(VCN, i, regUVD_SOFT_RESET);
+	tmp |= UVD_SOFT_RESET__LMI_SOFT_RESET_MASK;
+	WREG32_SOC15(VCN, i, regUVD_SOFT_RESET, tmp);
+
+	/* clear status */
+	WREG32_SOC15(VCN, i, regUVD_STATUS, 0);
+
+	/* apply HW clock gating */
+	vcn_v4_0_enable_clock_gating(vinst);
+
+	/* enable VCN power gating */
+	vcn_v4_0_enable_static_power_gating(vinst);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, i, regUVD_STATUS);
+
+done:
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_vcn(adev, false, i);
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_pause_dpg_mode - VCN pause with dpg mode
+ *
+ * @vinst: VCN instance
+ * @new_state: pause state
+ *
+ * Pause dpg mode for VCN block
+ */
+static int vcn_v4_0_pause_dpg_mode(struct amdgpu_vcn_inst *vinst,
+				   struct dpg_pause_state *new_state)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t reg_data = 0;
+	int ret_code;
+
+	/* pause/unpause if state is changed */
+	if (adev->vcn.inst[inst_idx].pause_state.fw_based != new_state->fw_based) {
+		DRM_DEV_DEBUG(adev->dev, "dpg pause state changed %d -> %d",
+			adev->vcn.inst[inst_idx].pause_state.fw_based,	new_state->fw_based);
+		reg_data = RREG32_SOC15(VCN, inst_idx, regUVD_DPG_PAUSE) &
+			(~UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK);
+
+		if (new_state->fw_based == VCN_DPG_STATE__PAUSE) {
+			ret_code = SOC15_WAIT_ON_RREG(VCN, inst_idx, regUVD_POWER_STATUS, 0x1,
+				UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+			if (!ret_code) {
+				/* pause DPG */
+				reg_data |= UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
+				WREG32_SOC15(VCN, inst_idx, regUVD_DPG_PAUSE, reg_data);
+
+				/* wait for ACK */
+				SOC15_WAIT_ON_RREG(VCN, inst_idx, regUVD_DPG_PAUSE,
+					UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK,
+					UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK);
+
+				SOC15_WAIT_ON_RREG(VCN, inst_idx, regUVD_POWER_STATUS,
+					UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON, UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+			}
+		} else {
+			/* unpause dpg, no need to wait */
+			reg_data &= ~UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
+			WREG32_SOC15(VCN, inst_idx, regUVD_DPG_PAUSE, reg_data);
+		}
+		adev->vcn.inst[inst_idx].pause_state.fw_based = new_state->fw_based;
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_unified_ring_get_rptr - get unified read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware unified read pointer
+ */
+static uint64_t vcn_v4_0_unified_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring != &adev->vcn.inst[ring->me].ring_enc[0])
+		DRM_ERROR("wrong ring id is identified in %s", __func__);
+
+	return RREG32_SOC15(VCN, ring->me, regUVD_RB_RPTR);
+}
+
+/**
+ * vcn_v4_0_unified_ring_get_wptr - get unified write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware unified write pointer
+ */
+static uint64_t vcn_v4_0_unified_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring != &adev->vcn.inst[ring->me].ring_enc[0])
+		DRM_ERROR("wrong ring id is identified in %s", __func__);
+
+	if (ring->use_doorbell)
+		return *ring->wptr_cpu_addr;
+	else
+		return RREG32_SOC15(VCN, ring->me, regUVD_RB_WPTR);
+}
+
+/**
+ * vcn_v4_0_unified_ring_set_wptr - set enc write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the enc write pointer to the hardware
+ */
+static void vcn_v4_0_unified_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring != &adev->vcn.inst[ring->me].ring_enc[0])
+		DRM_ERROR("wrong ring id is identified in %s", __func__);
+
+	if (ring->use_doorbell) {
+		*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
+		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+	} else {
+		WREG32_SOC15(VCN, ring->me, regUVD_RB_WPTR, lower_32_bits(ring->wptr));
+	}
+}
+
+static int vcn_v4_0_limit_sched(struct amdgpu_cs_parser *p,
+				struct amdgpu_job *job)
+{
+	struct drm_gpu_scheduler **scheds;
+	struct dma_fence *fence;
+
+	/* if VCN0 is harvested, we can't support AV1 */
+	if (p->adev->vcn.harvest_config & AMDGPU_VCN_HARVEST_VCN0)
+		return -EINVAL;
+
+	/* wait for all jobs to finish before switching to instance 0 */
+	fence = amdgpu_ctx_get_fence(p->ctx, job->base.entity, ~0ull);
+	if (fence) {
+		dma_fence_wait(fence, false);
+		dma_fence_put(fence);
+	}
+
+	scheds = p->adev->gpu_sched[AMDGPU_HW_IP_VCN_ENC]
+		[AMDGPU_RING_PRIO_0].sched;
+	drm_sched_entity_modify_sched(job->base.entity, scheds, 1);
+	return 0;
+}
+
+static int vcn_v4_0_dec_msg(struct amdgpu_cs_parser *p, struct amdgpu_job *job,
+			    uint64_t addr)
+{
+	struct ttm_operation_ctx ctx = { false, false };
+	struct amdgpu_bo_va_mapping *map;
+	uint32_t *msg, num_buffers;
+	struct amdgpu_bo *bo;
+	uint64_t start, end;
+	unsigned int i;
+	void *ptr;
+	int r;
+
+	addr &= AMDGPU_GMC_HOLE_MASK;
+	r = amdgpu_cs_find_mapping(p, addr, &bo, &map);
+	if (r) {
+		DRM_ERROR("Can't find BO for addr 0x%08llx\n", addr);
+		return r;
+	}
+
+	start = map->start * AMDGPU_GPU_PAGE_SIZE;
+	end = (map->last + 1) * AMDGPU_GPU_PAGE_SIZE;
+	if (addr & 0x7) {
+		DRM_ERROR("VCN messages must be 8 byte aligned!\n");
+		return -EINVAL;
+	}
+
+	bo->flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
+	amdgpu_bo_placement_from_domain(bo, bo->allowed_domains);
+	r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+	if (r) {
+		DRM_ERROR("Failed validating the VCN message BO (%d)!\n", r);
+		return r;
+	}
+
+	r = amdgpu_bo_kmap(bo, &ptr);
+	if (r) {
+		DRM_ERROR("Failed mapping the VCN message (%d)!\n", r);
+		return r;
+	}
+
+	msg = ptr + addr - start;
+
+	/* Check length */
+	if (msg[1] > end - addr) {
+		r = -EINVAL;
+		goto out;
+	}
+
+	if (msg[3] != RDECODE_MSG_CREATE)
+		goto out;
+
+	num_buffers = msg[2];
+	for (i = 0, msg = &msg[6]; i < num_buffers; ++i, msg += 4) {
+		uint32_t offset, size, *create;
+
+		if (msg[0] != RDECODE_MESSAGE_CREATE)
+			continue;
+
+		offset = msg[1];
+		size = msg[2];
+
+		if (offset + size > end) {
+			r = -EINVAL;
+			goto out;
+		}
+
+		create = ptr + addr + offset - start;
+
+		/* H264, HEVC and VP9 can run on any instance */
+		if (create[0] == 0x7 || create[0] == 0x10 || create[0] == 0x11)
+			continue;
+
+		r = vcn_v4_0_limit_sched(p, job);
+		if (r)
+			goto out;
+	}
+
+out:
+	amdgpu_bo_kunmap(bo);
+	return r;
+}
+
+#define RADEON_VCN_ENGINE_TYPE_ENCODE			(0x00000002)
+#define RADEON_VCN_ENGINE_TYPE_DECODE			(0x00000003)
+#define RADEON_VCN_ENGINE_INFO				(0x30000001)
+#define RENCODE_ENCODE_STANDARD_AV1			2
+#define RENCODE_IB_PARAM_SESSION_INIT			0x00000003
+
+/* return the offset in ib if id is found, -1 otherwise */
+static int vcn_v4_0_enc_find_ib_param(struct amdgpu_ib *ib, uint32_t id, int start)
+{
+	int i;
+
+	for (i = start; i < ib->length_dw && ib->ptr[i] >= 8; i += ib->ptr[i] / 4) {
+		if (ib->ptr[i + 1] == id)
+			return i;
+	}
+	return -1;
+}
+
+static int vcn_v4_0_ring_patch_cs_in_place(struct amdgpu_cs_parser *p,
+					   struct amdgpu_job *job,
+					   struct amdgpu_ib *ib)
+{
+	struct amdgpu_ring *ring = amdgpu_job_ring(job);
+	struct amdgpu_vcn_decode_buffer *decode_buffer;
+	uint64_t addr;
+	uint32_t val;
+	int idx = 0, sidx;
+
+	/* The first instance can decode anything */
+	if (!ring->me)
+		return 0;
+
+	while ((idx = vcn_v4_0_enc_find_ib_param(ib, RADEON_VCN_ENGINE_INFO, idx)) >= 0) {
+		val = amdgpu_ib_get_value(ib, idx + 2); /* RADEON_VCN_ENGINE_TYPE */
+		if (val == RADEON_VCN_ENGINE_TYPE_DECODE) {
+			decode_buffer = (struct amdgpu_vcn_decode_buffer *)&ib->ptr[idx + 6];
+
+			if (!(decode_buffer->valid_buf_flag & 0x1))
+				return 0;
+
+			addr = ((u64)decode_buffer->msg_buffer_address_hi) << 32 |
+				decode_buffer->msg_buffer_address_lo;
+			return vcn_v4_0_dec_msg(p, job, addr);
+		} else if (val == RADEON_VCN_ENGINE_TYPE_ENCODE) {
+			sidx = vcn_v4_0_enc_find_ib_param(ib, RENCODE_IB_PARAM_SESSION_INIT, idx);
+			if (sidx >= 0 && ib->ptr[sidx + 2] == RENCODE_ENCODE_STANDARD_AV1)
+				return vcn_v4_0_limit_sched(p, job);
+		}
+		idx += ib->ptr[idx] / 4;
+	}
+	return 0;
+}
+
+static int vcn_v4_0_ring_reset(struct amdgpu_ring *ring,
+			       unsigned int vmid,
+			       struct amdgpu_fence *timedout_fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[ring->me];
+	int r;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+	r = vcn_v4_0_stop(vinst);
+	if (r)
+		return r;
+	r = vcn_v4_0_start(vinst);
+	if (r)
+		return r;
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static struct amdgpu_ring_funcs vcn_v4_0_unified_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_ENC,
+	.align_mask = 0x3f,
+	.nop = VCN_ENC_CMD_NO_OP,
+	.extra_bytes = sizeof(struct amdgpu_vcn_rb_metadata),
+	.get_rptr = vcn_v4_0_unified_ring_get_rptr,
+	.get_wptr = vcn_v4_0_unified_ring_get_wptr,
+	.set_wptr = vcn_v4_0_unified_ring_set_wptr,
+	.patch_cs_in_place = vcn_v4_0_ring_patch_cs_in_place,
+	.emit_frame_size =
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 4 +
+		4 + /* vcn_v2_0_enc_ring_emit_vm_flush */
+		5 + 5 + /* vcn_v2_0_enc_ring_emit_fence x2 vm fence */
+		1, /* vcn_v2_0_enc_ring_insert_end */
+	.emit_ib_size = 5, /* vcn_v2_0_enc_ring_emit_ib */
+	.emit_ib = vcn_v2_0_enc_ring_emit_ib,
+	.emit_fence = vcn_v2_0_enc_ring_emit_fence,
+	.emit_vm_flush = vcn_v2_0_enc_ring_emit_vm_flush,
+	.test_ring = amdgpu_vcn_enc_ring_test_ring,
+	.test_ib = amdgpu_vcn_unified_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.insert_end = vcn_v2_0_enc_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_vcn_ring_begin_use,
+	.end_use = amdgpu_vcn_ring_end_use,
+	.emit_wreg = vcn_v2_0_enc_ring_emit_wreg,
+	.emit_reg_wait = vcn_v2_0_enc_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = vcn_v4_0_ring_reset,
+};
+
+/**
+ * vcn_v4_0_set_unified_ring_funcs - set unified ring functions
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set unified ring functions
+ */
+static void vcn_v4_0_set_unified_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		if (amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 2))
+			vcn_v4_0_unified_ring_vm_funcs.secure_submission_supported = true;
+
+		adev->vcn.inst[i].ring_enc[0].funcs =
+		       (const struct amdgpu_ring_funcs *)&vcn_v4_0_unified_ring_vm_funcs;
+		adev->vcn.inst[i].ring_enc[0].me = i;
+	}
+}
+
+/**
+ * vcn_v4_0_is_idle - check VCN block is idle
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block structure
+ *
+ * Check whether VCN block is idle
+ */
+static bool vcn_v4_0_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, ret = 1;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		ret &= (RREG32_SOC15(VCN, i, regUVD_STATUS) == UVD_STATUS__IDLE);
+	}
+
+	return ret;
+}
+
+/**
+ * vcn_v4_0_wait_for_idle - wait for VCN block idle
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Wait for VCN block idle
+ */
+static int vcn_v4_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, ret = 0;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		ret = SOC15_WAIT_ON_RREG(VCN, i, regUVD_STATUS, UVD_STATUS__IDLE,
+			UVD_STATUS__IDLE);
+		if (ret)
+			return ret;
+	}
+
+	return ret;
+}
+
+/**
+ * vcn_v4_0_set_clockgating_state - set VCN block clockgating state
+ *
+ * @ip_block: amdgpu_ip_block pointer
+ * @state: clock gating state
+ *
+ * Set VCN block clockgating state
+ */
+static int vcn_v4_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = state == AMD_CG_STATE_GATE;
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[i];
+
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		if (enable) {
+			if (RREG32_SOC15(VCN, i, regUVD_STATUS) != UVD_STATUS__IDLE)
+				return -EBUSY;
+			vcn_v4_0_enable_clock_gating(vinst);
+		} else {
+			vcn_v4_0_disable_clock_gating(vinst);
+		}
+	}
+
+	return 0;
+}
+
+static int vcn_v4_0_set_pg_state(struct amdgpu_vcn_inst *vinst,
+				 enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int ret = 0;
+
+	/* for SRIOV, guest should not control VCN Power-gating
+	 * MMSCH FW should control Power-gating and clock-gating
+	 * guest should avoid touching CGC and PG
+	 */
+	if (amdgpu_sriov_vf(adev)) {
+		vinst->cur_state = AMD_PG_STATE_UNGATE;
+		return 0;
+	}
+
+	if (state == vinst->cur_state)
+		return 0;
+
+	if (state == AMD_PG_STATE_GATE)
+		ret = vcn_v4_0_stop(vinst);
+	else
+		ret = vcn_v4_0_start(vinst);
+
+	if (!ret)
+		vinst->cur_state = state;
+
+	return ret;
+}
+
+/**
+ * vcn_v4_0_set_ras_interrupt_state - set VCN block RAS interrupt state
+ *
+ * @adev: amdgpu_device pointer
+ * @source: interrupt sources
+ * @type: interrupt types
+ * @state: interrupt states
+ *
+ * Set VCN block RAS interrupt state
+ */
+static int vcn_v4_0_set_ras_interrupt_state(struct amdgpu_device *adev,
+	struct amdgpu_irq_src *source,
+	unsigned int type,
+	enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+/**
+ * vcn_v4_0_process_interrupt - process VCN block interrupt
+ *
+ * @adev: amdgpu_device pointer
+ * @source: interrupt sources
+ * @entry: interrupt entry from clients and sources
+ *
+ * Process VCN block interrupt
+ */
+static int vcn_v4_0_process_interrupt(struct amdgpu_device *adev, struct amdgpu_irq_src *source,
+      struct amdgpu_iv_entry *entry)
+{
+	uint32_t ip_instance;
+
+	if (amdgpu_sriov_is_vcn_rb_decouple(adev)) {
+		ip_instance = entry->ring_id;
+	} else {
+		switch (entry->client_id) {
+		case SOC15_IH_CLIENTID_VCN:
+			ip_instance = 0;
+			break;
+		case SOC15_IH_CLIENTID_VCN1:
+			ip_instance = 1;
+			break;
+		default:
+			DRM_ERROR("Unhandled client id: %d\n", entry->client_id);
+			return 0;
+		}
+	}
+
+	DRM_DEBUG("IH: VCN TRAP\n");
+
+	switch (entry->src_id) {
+	case VCN_4_0__SRCID__UVD_ENC_GENERAL_PURPOSE:
+		amdgpu_fence_process(&adev->vcn.inst[ip_instance].ring_enc[0]);
+		break;
+	default:
+		DRM_ERROR("Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs vcn_v4_0_irq_funcs = {
+	.process = vcn_v4_0_process_interrupt,
+};
+
+static const struct amdgpu_irq_src_funcs vcn_v4_0_ras_irq_funcs = {
+	.set = vcn_v4_0_set_ras_interrupt_state,
+	.process = amdgpu_vcn_process_poison_irq,
+};
+
+/**
+ * vcn_v4_0_set_irq_funcs - set VCN block interrupt irq functions
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set VCN block interrupt irq functions
+ */
+static void vcn_v4_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		adev->vcn.inst[i].irq.num_types = adev->vcn.inst[i].num_enc_rings + 1;
+		adev->vcn.inst[i].irq.funcs = &vcn_v4_0_irq_funcs;
+
+		adev->vcn.inst[i].ras_poison_irq.num_types = adev->vcn.inst[i].num_enc_rings + 1;
+		adev->vcn.inst[i].ras_poison_irq.funcs = &vcn_v4_0_ras_irq_funcs;
+	}
+}
+
+static const struct amd_ip_funcs vcn_v4_0_ip_funcs = {
+	.name = "vcn_v4_0",
+	.early_init = vcn_v4_0_early_init,
+	.sw_init = vcn_v4_0_sw_init,
+	.sw_fini = vcn_v4_0_sw_fini,
+	.hw_init = vcn_v4_0_hw_init,
+	.hw_fini = vcn_v4_0_hw_fini,
+	.suspend = vcn_v4_0_suspend,
+	.resume = vcn_v4_0_resume,
+	.is_idle = vcn_v4_0_is_idle,
+	.wait_for_idle = vcn_v4_0_wait_for_idle,
+	.set_clockgating_state = vcn_v4_0_set_clockgating_state,
+	.set_powergating_state = vcn_set_powergating_state,
+	.dump_ip_state = amdgpu_vcn_dump_ip_state,
+	.print_ip_state = amdgpu_vcn_print_ip_state,
+};
+
+const struct amdgpu_ip_block_version vcn_v4_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_VCN,
+	.major = 4,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &vcn_v4_0_ip_funcs,
+};
+
+static uint32_t vcn_v4_0_query_poison_by_instance(struct amdgpu_device *adev,
+			uint32_t instance, uint32_t sub_block)
+{
+	uint32_t poison_stat = 0, reg_value = 0;
+
+	switch (sub_block) {
+	case AMDGPU_VCN_V4_0_VCPU_VCODEC:
+		reg_value = RREG32_SOC15(VCN, instance, regUVD_RAS_VCPU_VCODEC_STATUS);
+		poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_VCPU_VCODEC_STATUS, POISONED_PF);
+		break;
+	default:
+		break;
+	}
+
+	if (poison_stat)
+		dev_info(adev->dev, "Poison detected in VCN%d, sub_block%d\n",
+			instance, sub_block);
+
+	return poison_stat;
+}
+
+static bool vcn_v4_0_query_ras_poison_status(struct amdgpu_device *adev)
+{
+	uint32_t inst, sub;
+	uint32_t poison_stat = 0;
+
+	for (inst = 0; inst < adev->vcn.num_vcn_inst; inst++)
+		for (sub = 0; sub < AMDGPU_VCN_V4_0_MAX_SUB_BLOCK; sub++)
+			poison_stat +=
+				vcn_v4_0_query_poison_by_instance(adev, inst, sub);
+
+	return !!poison_stat;
+}
+
+const struct amdgpu_ras_block_hw_ops vcn_v4_0_ras_hw_ops = {
+	.query_poison_status = vcn_v4_0_query_ras_poison_status,
+};
+
+static struct amdgpu_vcn_ras vcn_v4_0_ras = {
+	.ras_block = {
+		.hw_ops = &vcn_v4_0_ras_hw_ops,
+		.ras_late_init = amdgpu_vcn_ras_late_init,
+	},
+};
+
+static void vcn_v4_0_set_ras_funcs(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, VCN_HWIP, 0)) {
+	case IP_VERSION(4, 0, 0):
+		adev->vcn.ras = &vcn_v4_0_ras;
+		break;
+	default:
+		break;
+	}
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.h b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.h
new file mode 100644
index 000000000000..7d3d11f40f27
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.h
@@ -0,0 +1,35 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VCN_V4_0_H__
+#define __VCN_V4_0_H__
+
+enum amdgpu_vcn_v4_0_sub_block {
+	AMDGPU_VCN_V4_0_VCPU_VCODEC = 0,
+
+	AMDGPU_VCN_V4_0_MAX_SUB_BLOCK,
+};
+
+extern const struct amdgpu_ip_block_version vcn_v4_0_ip_block;
+
+#endif /* __VCN_V4_0_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c
new file mode 100644
index 000000000000..eacf4e93ba2f
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c
@@ -0,0 +1,2131 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include <drm/drm_drv.h>
+
+#include "amdgpu.h"
+#include "amdgpu_vcn.h"
+#include "amdgpu_pm.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "soc15_hw_ip.h"
+#include "vcn_v2_0.h"
+#include "vcn_v4_0_3.h"
+#include "mmsch_v4_0_3.h"
+
+#include "vcn/vcn_4_0_3_offset.h"
+#include "vcn/vcn_4_0_3_sh_mask.h"
+#include "ivsrcid/vcn/irqsrcs_vcn_4_0.h"
+
+#define mmUVD_DPG_LMA_CTL		regUVD_DPG_LMA_CTL
+#define mmUVD_DPG_LMA_CTL_BASE_IDX	regUVD_DPG_LMA_CTL_BASE_IDX
+#define mmUVD_DPG_LMA_DATA		regUVD_DPG_LMA_DATA
+#define mmUVD_DPG_LMA_DATA_BASE_IDX	regUVD_DPG_LMA_DATA_BASE_IDX
+
+#define VCN_VID_SOC_ADDRESS_2_0		0x1fb00
+#define VCN1_VID_SOC_ADDRESS_3_0	0x48300
+#define VCN1_AON_SOC_ADDRESS_3_0	0x48000
+
+static const struct amdgpu_hwip_reg_entry vcn_reg_list_4_0_3[] = {
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_CONTEXT_ID),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_CONTEXT_ID2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_GPCOM_VCPU_DATA0),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_GPCOM_VCPU_DATA1),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_GPCOM_VCPU_CMD),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_HI),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_LO),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_HI2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_LO2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_HI3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_LO3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_HI4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_LO4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_RPTR),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_WPTR),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_RPTR2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_WPTR2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_RPTR3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_WPTR3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_RPTR4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_WPTR4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_SIZE),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_SIZE2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_SIZE3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_SIZE4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_PGFSM_CONFIG),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_PGFSM_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_DPG_LMA_CTL),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_DPG_LMA_DATA),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_DPG_LMA_MASK),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_DPG_PAUSE)
+};
+
+#define NORMALIZE_VCN_REG_OFFSET(offset) \
+		(offset & 0x1FFFF)
+
+static int vcn_v4_0_3_start_sriov(struct amdgpu_device *adev);
+static void vcn_v4_0_3_set_unified_ring_funcs(struct amdgpu_device *adev);
+static void vcn_v4_0_3_set_irq_funcs(struct amdgpu_device *adev);
+static int vcn_v4_0_3_set_pg_state(struct amdgpu_vcn_inst *vinst,
+				   enum amd_powergating_state state);
+static int vcn_v4_0_3_pause_dpg_mode(struct amdgpu_vcn_inst *vinst,
+				     struct dpg_pause_state *new_state);
+static void vcn_v4_0_3_unified_ring_set_wptr(struct amdgpu_ring *ring);
+static void vcn_v4_0_3_set_ras_funcs(struct amdgpu_device *adev);
+static void vcn_v4_0_3_enable_ras(struct amdgpu_device *adev,
+				  int inst_idx, bool indirect);
+
+static inline bool vcn_v4_0_3_normalizn_reqd(struct amdgpu_device *adev)
+{
+	return (adev->vcn.caps & AMDGPU_VCN_CAPS(RRMT_ENABLED)) == 0;
+}
+
+/**
+ * vcn_v4_0_3_early_init - set function pointers
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Set ring and irq function pointers
+ */
+static int vcn_v4_0_3_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, r;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i)
+		/* re-use enc ring as unified ring */
+		adev->vcn.inst[i].num_enc_rings = 1;
+
+	vcn_v4_0_3_set_unified_ring_funcs(adev);
+	vcn_v4_0_3_set_irq_funcs(adev);
+	vcn_v4_0_3_set_ras_funcs(adev);
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		adev->vcn.inst[i].set_pg_state = vcn_v4_0_3_set_pg_state;
+
+		r = amdgpu_vcn_early_init(adev, i);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+static int vcn_v4_0_3_late_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	adev->vcn.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->vcn.inst[0].ring_enc[0]);
+
+	if (amdgpu_dpm_reset_vcn_is_supported(adev))
+		adev->vcn.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+
+	return 0;
+}
+
+static int vcn_v4_0_3_fw_shared_init(struct amdgpu_device *adev, int inst_idx)
+{
+	struct amdgpu_vcn4_fw_shared *fw_shared;
+
+	fw_shared = adev->vcn.inst[inst_idx].fw_shared.cpu_addr;
+	fw_shared->present_flag_0 = cpu_to_le32(AMDGPU_FW_SHARED_FLAG_0_UNIFIED_QUEUE);
+	fw_shared->sq.is_enabled = 1;
+
+	if (amdgpu_vcnfw_log)
+		amdgpu_vcn_fwlog_init(&adev->vcn.inst[inst_idx]);
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_3_sw_init - sw init for VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Load firmware and sw initialization
+ */
+static int vcn_v4_0_3_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	int i, r, vcn_inst;
+
+	/* VCN DEC TRAP */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+		VCN_4_0__SRCID__UVD_ENC_GENERAL_PURPOSE, &adev->vcn.inst->irq);
+	if (r)
+		return r;
+
+	/* VCN POISON TRAP */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+		VCN_4_0__SRCID_UVD_POISON, &adev->vcn.inst->ras_poison_irq);
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+
+		r = amdgpu_vcn_sw_init(adev, i);
+		if (r)
+			return r;
+
+		amdgpu_vcn_setup_ucode(adev, i);
+
+		r = amdgpu_vcn_resume(adev, i);
+		if (r)
+			return r;
+
+		vcn_inst = GET_INST(VCN, i);
+
+		ring = &adev->vcn.inst[i].ring_enc[0];
+		ring->use_doorbell = true;
+
+		if (!amdgpu_sriov_vf(adev))
+			ring->doorbell_index =
+				(adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
+				9 * vcn_inst;
+		else
+			ring->doorbell_index =
+				(adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
+				32 * vcn_inst;
+
+		ring->vm_hub = AMDGPU_MMHUB0(adev->vcn.inst[i].aid_id);
+		sprintf(ring->name, "vcn_unified_%d", adev->vcn.inst[i].aid_id);
+
+		/* There are no per-instance irq source IDs on 4.0.3, the IH
+		 * packets use a separate field to differentiate instances.
+		 */
+		r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst[0].irq, 0,
+				     AMDGPU_RING_PRIO_DEFAULT,
+				     &adev->vcn.inst[i].sched_score);
+		if (r)
+			return r;
+
+		vcn_v4_0_3_fw_shared_init(adev, i);
+
+		if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)
+			adev->vcn.inst[i].pause_dpg_mode = vcn_v4_0_3_pause_dpg_mode;
+	}
+
+	if (amdgpu_sriov_vf(adev)) {
+		r = amdgpu_virt_alloc_mm_table(adev);
+		if (r)
+			return r;
+	}
+
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__VCN)) {
+		r = amdgpu_vcn_ras_sw_init(adev);
+		if (r) {
+			dev_err(adev->dev, "Failed to initialize vcn ras block!\n");
+			return r;
+		}
+	}
+
+	r = amdgpu_vcn_reg_dump_init(adev, vcn_reg_list_4_0_3, ARRAY_SIZE(vcn_reg_list_4_0_3));
+	if (r)
+		return r;
+
+	return amdgpu_vcn_sysfs_reset_mask_init(adev);
+}
+
+/**
+ * vcn_v4_0_3_sw_fini - sw fini for VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * VCN suspend and free up sw allocation
+ */
+static int vcn_v4_0_3_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, r, idx;
+
+	if (drm_dev_enter(&adev->ddev, &idx)) {
+		for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+			struct amdgpu_vcn4_fw_shared *fw_shared;
+
+			fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+			fw_shared->present_flag_0 = 0;
+			fw_shared->sq.is_enabled = cpu_to_le32(false);
+		}
+		drm_dev_exit(idx);
+	}
+
+	if (amdgpu_sriov_vf(adev))
+		amdgpu_virt_free_mm_table(adev);
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		r = amdgpu_vcn_suspend(adev, i);
+		if (r)
+			return r;
+	}
+
+	amdgpu_vcn_sysfs_reset_mask_fini(adev);
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++)
+		amdgpu_vcn_sw_fini(adev, i);
+
+	return 0;
+}
+
+static int vcn_v4_0_3_hw_init_inst(struct amdgpu_vcn_inst *vinst)
+{
+	int vcn_inst;
+	struct amdgpu_device *adev = vinst->adev;
+	struct amdgpu_ring *ring;
+	int inst_idx = vinst->inst;
+
+	vcn_inst = GET_INST(VCN, inst_idx);
+	ring = &adev->vcn.inst[inst_idx].ring_enc[0];
+	if (ring->use_doorbell) {
+		adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
+			(adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 9 * vcn_inst,
+			adev->vcn.inst[inst_idx].aid_id);
+
+		WREG32_SOC15(VCN, vcn_inst, regVCN_RB1_DB_CTRL,
+			ring->doorbell_index << VCN_RB1_DB_CTRL__OFFSET__SHIFT |
+			VCN_RB1_DB_CTRL__EN_MASK);
+
+		/* Read DB_CTRL to flush the write DB_CTRL command. */
+		RREG32_SOC15(VCN, vcn_inst, regVCN_RB1_DB_CTRL);
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_3_hw_init - start and test VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Initialize the hardware, boot up the VCPU and do some testing
+ */
+static int vcn_v4_0_3_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	struct amdgpu_vcn_inst *vinst;
+	int i, r;
+
+	if (amdgpu_sriov_vf(adev)) {
+		r = vcn_v4_0_3_start_sriov(adev);
+		if (r)
+			return r;
+
+		for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+			ring = &adev->vcn.inst[i].ring_enc[0];
+			ring->wptr = 0;
+			ring->wptr_old = 0;
+			vcn_v4_0_3_unified_ring_set_wptr(ring);
+			ring->sched.ready = true;
+		}
+	} else {
+		/* This flag is not set for VF, assumed to be disabled always */
+		if (RREG32_SOC15(VCN, GET_INST(VCN, 0), regVCN_RRMT_CNTL) &
+		    0x100)
+			adev->vcn.caps |= AMDGPU_VCN_CAPS(RRMT_ENABLED);
+
+		for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+			struct amdgpu_vcn4_fw_shared *fw_shared;
+
+			ring = &adev->vcn.inst[i].ring_enc[0];
+			vinst = &adev->vcn.inst[i];
+			vcn_v4_0_3_hw_init_inst(vinst);
+
+			/* Re-init fw_shared when RAS fatal error occurred */
+			fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+			if (!fw_shared->sq.is_enabled)
+				vcn_v4_0_3_fw_shared_init(adev, i);
+
+			r = amdgpu_ring_test_helper(ring);
+			if (r)
+				return r;
+		}
+	}
+
+	return r;
+}
+
+/**
+ * vcn_v4_0_3_hw_fini - stop the hardware block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Stop the VCN block, mark ring as not ready any more
+ */
+static int vcn_v4_0_3_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[i];
+
+		cancel_delayed_work_sync(&vinst->idle_work);
+
+		if (vinst->cur_state != AMD_PG_STATE_GATE)
+			vinst->set_pg_state(vinst, AMD_PG_STATE_GATE);
+	}
+
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__VCN) && !amdgpu_sriov_vf(adev))
+		amdgpu_irq_put(adev, &adev->vcn.inst->ras_poison_irq, 0);
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_3_suspend - suspend VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * HW fini and suspend VCN block
+ */
+static int vcn_v4_0_3_suspend(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r, i;
+
+	r = vcn_v4_0_3_hw_fini(ip_block);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		r = amdgpu_vcn_suspend(adev, i);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_3_resume - resume VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Resume firmware and hw init VCN block
+ */
+static int vcn_v4_0_3_resume(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r, i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		r = amdgpu_vcn_resume(ip_block->adev, i);
+		if (r)
+			return r;
+	}
+
+	r = vcn_v4_0_3_hw_init(ip_block);
+
+	return r;
+}
+
+/**
+ * vcn_v4_0_3_mc_resume - memory controller programming
+ *
+ * @vinst: VCN instance
+ *
+ * Let the VCN memory controller know it's offsets
+ */
+static void vcn_v4_0_3_mc_resume(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t offset, size, vcn_inst;
+	const struct common_firmware_header *hdr;
+
+	hdr = (const struct common_firmware_header *)adev->vcn.inst[inst_idx].fw->data;
+	size = AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8);
+
+	vcn_inst = GET_INST(VCN, inst_idx);
+	/* cache window 0: fw */
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		WREG32_SOC15(
+			VCN, vcn_inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst_idx]
+				 .tmr_mc_addr_lo));
+		WREG32_SOC15(
+			VCN, vcn_inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst_idx]
+				 .tmr_mc_addr_hi));
+		WREG32_SOC15(VCN, vcn_inst, regUVD_VCPU_CACHE_OFFSET0, 0);
+		offset = 0;
+	} else {
+		WREG32_SOC15(VCN, vcn_inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			     lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr));
+		WREG32_SOC15(VCN, vcn_inst,
+			     regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			     upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr));
+		offset = size;
+		WREG32_SOC15(VCN, vcn_inst, regUVD_VCPU_CACHE_OFFSET0,
+			     AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
+	}
+	WREG32_SOC15(VCN, vcn_inst, regUVD_VCPU_CACHE_SIZE0, size);
+
+	/* cache window 1: stack */
+	WREG32_SOC15(VCN, vcn_inst, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW,
+		     lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset));
+	WREG32_SOC15(VCN, vcn_inst, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH,
+		     upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset));
+	WREG32_SOC15(VCN, vcn_inst, regUVD_VCPU_CACHE_OFFSET1, 0);
+	WREG32_SOC15(VCN, vcn_inst, regUVD_VCPU_CACHE_SIZE1,
+		     AMDGPU_VCN_STACK_SIZE);
+
+	/* cache window 2: context */
+	WREG32_SOC15(VCN, vcn_inst, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW,
+		     lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset +
+				   AMDGPU_VCN_STACK_SIZE));
+	WREG32_SOC15(VCN, vcn_inst, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH,
+		     upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset +
+				   AMDGPU_VCN_STACK_SIZE));
+	WREG32_SOC15(VCN, vcn_inst, regUVD_VCPU_CACHE_OFFSET2, 0);
+	WREG32_SOC15(VCN, vcn_inst, regUVD_VCPU_CACHE_SIZE2,
+		     AMDGPU_VCN_CONTEXT_SIZE);
+
+	/* non-cache window */
+	WREG32_SOC15(
+		VCN, vcn_inst, regUVD_LMI_VCPU_NC0_64BIT_BAR_LOW,
+		lower_32_bits(adev->vcn.inst[inst_idx].fw_shared.gpu_addr));
+	WREG32_SOC15(
+		VCN, vcn_inst, regUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH,
+		upper_32_bits(adev->vcn.inst[inst_idx].fw_shared.gpu_addr));
+	WREG32_SOC15(VCN, vcn_inst, regUVD_VCPU_NONCACHE_OFFSET0, 0);
+	WREG32_SOC15(
+		VCN, vcn_inst, regUVD_VCPU_NONCACHE_SIZE0,
+		AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn4_fw_shared)));
+}
+
+/**
+ * vcn_v4_0_3_mc_resume_dpg_mode - memory controller programming for dpg mode
+ *
+ * @vinst: VCN instance
+ * @indirect: indirectly write sram
+ *
+ * Let the VCN memory controller know it's offsets with dpg mode
+ */
+static void vcn_v4_0_3_mc_resume_dpg_mode(struct amdgpu_vcn_inst *vinst,
+					  bool indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t offset, size;
+	const struct common_firmware_header *hdr;
+
+	hdr = (const struct common_firmware_header *)adev->vcn.inst[inst_idx].fw->data;
+	size = AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8);
+
+	/* cache window 0: fw */
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		if (!indirect) {
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, 0, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+				(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN +
+					inst_idx].tmr_mc_addr_lo), 0, indirect);
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, 0, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+				(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN +
+					inst_idx].tmr_mc_addr_hi), 0, indirect);
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, 0, regUVD_VCPU_CACHE_OFFSET0), 0, 0, indirect);
+		} else {
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, 0, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW), 0, 0, indirect);
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, 0, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH), 0, 0, indirect);
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, 0, regUVD_VCPU_CACHE_OFFSET0), 0, 0, indirect);
+		}
+		offset = 0;
+	} else {
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr), 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr), 0, indirect);
+		offset = size;
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_VCPU_CACHE_OFFSET0),
+			AMDGPU_UVD_FIRMWARE_OFFSET >> 3, 0, indirect);
+	}
+
+	if (!indirect)
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_VCPU_CACHE_SIZE0), size, 0, indirect);
+	else
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_VCPU_CACHE_SIZE0), 0, 0, indirect);
+
+	/* cache window 1: stack */
+	if (!indirect) {
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset), 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset), 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_VCPU_CACHE_OFFSET1), 0, 0, indirect);
+	} else {
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW), 0, 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH), 0, 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_VCPU_CACHE_OFFSET1), 0, 0, indirect);
+	}
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_VCPU_CACHE_SIZE1), AMDGPU_VCN_STACK_SIZE, 0, indirect);
+
+	/* cache window 2: context */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset +
+				AMDGPU_VCN_STACK_SIZE), 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset +
+				AMDGPU_VCN_STACK_SIZE), 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_VCPU_CACHE_OFFSET2), 0, 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_VCPU_CACHE_SIZE2), AMDGPU_VCN_CONTEXT_SIZE, 0, indirect);
+
+	/* non-cache window */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_LMI_VCPU_NC0_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[inst_idx].fw_shared.gpu_addr), 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[inst_idx].fw_shared.gpu_addr), 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_VCPU_NONCACHE_OFFSET0), 0, 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_VCPU_NONCACHE_SIZE0),
+			AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn4_fw_shared)), 0, indirect);
+
+	/* VCN global tiling registers */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_GFX8_ADDR_CONFIG), adev->gfx.config.gb_addr_config, 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_GFX10_ADDR_CONFIG), adev->gfx.config.gb_addr_config, 0, indirect);
+}
+
+/**
+ * vcn_v4_0_3_disable_clock_gating - disable VCN clock gating
+ *
+ * @vinst: VCN instance
+ *
+ * Disable clock gating for VCN block
+ */
+static void vcn_v4_0_3_disable_clock_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t data;
+	int vcn_inst;
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		return;
+
+	vcn_inst = GET_INST(VCN, inst_idx);
+
+	/* VCN disable CGC */
+	data = RREG32_SOC15(VCN, vcn_inst, regUVD_CGC_CTRL);
+	data &= ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK;
+	data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(VCN, vcn_inst, regUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, vcn_inst, regUVD_CGC_GATE);
+	data &= ~(UVD_CGC_GATE__SYS_MASK
+		| UVD_CGC_GATE__MPEG2_MASK
+		| UVD_CGC_GATE__REGS_MASK
+		| UVD_CGC_GATE__RBC_MASK
+		| UVD_CGC_GATE__LMI_MC_MASK
+		| UVD_CGC_GATE__LMI_UMC_MASK
+		| UVD_CGC_GATE__MPC_MASK
+		| UVD_CGC_GATE__LBSI_MASK
+		| UVD_CGC_GATE__LRBBM_MASK
+		| UVD_CGC_GATE__WCB_MASK
+		| UVD_CGC_GATE__VCPU_MASK
+		| UVD_CGC_GATE__MMSCH_MASK);
+
+	WREG32_SOC15(VCN, vcn_inst, regUVD_CGC_GATE, data);
+	SOC15_WAIT_ON_RREG(VCN, vcn_inst, regUVD_CGC_GATE, 0, 0xFFFFFFFF);
+
+	data = RREG32_SOC15(VCN, vcn_inst, regUVD_CGC_CTRL);
+	data &= ~(UVD_CGC_CTRL__SYS_MODE_MASK
+		| UVD_CGC_CTRL__MPEG2_MODE_MASK
+		| UVD_CGC_CTRL__REGS_MODE_MASK
+		| UVD_CGC_CTRL__RBC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_MC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_UMC_MODE_MASK
+		| UVD_CGC_CTRL__MPC_MODE_MASK
+		| UVD_CGC_CTRL__LBSI_MODE_MASK
+		| UVD_CGC_CTRL__LRBBM_MODE_MASK
+		| UVD_CGC_CTRL__WCB_MODE_MASK
+		| UVD_CGC_CTRL__VCPU_MODE_MASK
+		| UVD_CGC_CTRL__MMSCH_MODE_MASK);
+	WREG32_SOC15(VCN, vcn_inst, regUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, vcn_inst, regUVD_SUVD_CGC_GATE);
+	data |= (UVD_SUVD_CGC_GATE__SRE_MASK
+		| UVD_SUVD_CGC_GATE__SIT_MASK
+		| UVD_SUVD_CGC_GATE__SMP_MASK
+		| UVD_SUVD_CGC_GATE__SCM_MASK
+		| UVD_SUVD_CGC_GATE__SDB_MASK
+		| UVD_SUVD_CGC_GATE__SRE_H264_MASK
+		| UVD_SUVD_CGC_GATE__SRE_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SIT_H264_MASK
+		| UVD_SUVD_CGC_GATE__SIT_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SCM_H264_MASK
+		| UVD_SUVD_CGC_GATE__SCM_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SDB_H264_MASK
+		| UVD_SUVD_CGC_GATE__SDB_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__ENT_MASK
+		| UVD_SUVD_CGC_GATE__SIT_HEVC_DEC_MASK
+		| UVD_SUVD_CGC_GATE__SITE_MASK
+		| UVD_SUVD_CGC_GATE__SRE_VP9_MASK
+		| UVD_SUVD_CGC_GATE__SCM_VP9_MASK
+		| UVD_SUVD_CGC_GATE__SIT_VP9_DEC_MASK
+		| UVD_SUVD_CGC_GATE__SDB_VP9_MASK
+		| UVD_SUVD_CGC_GATE__IME_HEVC_MASK);
+	WREG32_SOC15(VCN, vcn_inst, regUVD_SUVD_CGC_GATE, data);
+
+	data = RREG32_SOC15(VCN, vcn_inst, regUVD_SUVD_CGC_CTRL);
+	data &= ~(UVD_SUVD_CGC_CTRL__SRE_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SIT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SMP_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SCM_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SDB_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__ENT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__IME_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SITE_MODE_MASK);
+	WREG32_SOC15(VCN, vcn_inst, regUVD_SUVD_CGC_CTRL, data);
+}
+
+/**
+ * vcn_v4_0_3_disable_clock_gating_dpg_mode - disable VCN clock gating dpg mode
+ *
+ * @vinst: VCN instance
+ * @sram_sel: sram select
+ * @indirect: indirectly write sram
+ *
+ * Disable clock gating for VCN block with dpg mode
+ */
+static void vcn_v4_0_3_disable_clock_gating_dpg_mode(struct amdgpu_vcn_inst *vinst,
+						     uint8_t sram_sel,
+						     uint8_t indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t reg_data = 0;
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		return;
+
+	/* enable sw clock gating control */
+	reg_data = 0 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	reg_data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	reg_data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	reg_data &= ~(UVD_CGC_CTRL__SYS_MODE_MASK |
+		 UVD_CGC_CTRL__MPEG2_MODE_MASK |
+		 UVD_CGC_CTRL__REGS_MODE_MASK |
+		 UVD_CGC_CTRL__RBC_MODE_MASK |
+		 UVD_CGC_CTRL__LMI_MC_MODE_MASK |
+		 UVD_CGC_CTRL__LMI_UMC_MODE_MASK |
+		 UVD_CGC_CTRL__IDCT_MODE_MASK |
+		 UVD_CGC_CTRL__MPRD_MODE_MASK |
+		 UVD_CGC_CTRL__MPC_MODE_MASK |
+		 UVD_CGC_CTRL__LBSI_MODE_MASK |
+		 UVD_CGC_CTRL__LRBBM_MODE_MASK |
+		 UVD_CGC_CTRL__WCB_MODE_MASK |
+		 UVD_CGC_CTRL__VCPU_MODE_MASK);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_CGC_CTRL), reg_data, sram_sel, indirect);
+
+	/* turn off clock gating */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_CGC_GATE), 0, sram_sel, indirect);
+
+	/* turn on SUVD clock gating */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_SUVD_CGC_GATE), 1, sram_sel, indirect);
+
+	/* turn on sw mode in UVD_SUVD_CGC_CTRL */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_SUVD_CGC_CTRL), 0, sram_sel, indirect);
+}
+
+/**
+ * vcn_v4_0_3_enable_clock_gating - enable VCN clock gating
+ *
+ * @vinst: VCN instance
+ *
+ * Enable clock gating for VCN block
+ */
+static void vcn_v4_0_3_enable_clock_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t data;
+	int vcn_inst;
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		return;
+
+	vcn_inst = GET_INST(VCN, inst_idx);
+
+	/* enable VCN CGC */
+	data = RREG32_SOC15(VCN, vcn_inst, regUVD_CGC_CTRL);
+	data |= 0 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(VCN, vcn_inst, regUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, vcn_inst, regUVD_CGC_CTRL);
+	data |= (UVD_CGC_CTRL__SYS_MODE_MASK
+		| UVD_CGC_CTRL__MPEG2_MODE_MASK
+		| UVD_CGC_CTRL__REGS_MODE_MASK
+		| UVD_CGC_CTRL__RBC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_MC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_UMC_MODE_MASK
+		| UVD_CGC_CTRL__MPC_MODE_MASK
+		| UVD_CGC_CTRL__LBSI_MODE_MASK
+		| UVD_CGC_CTRL__LRBBM_MODE_MASK
+		| UVD_CGC_CTRL__WCB_MODE_MASK
+		| UVD_CGC_CTRL__VCPU_MODE_MASK);
+	WREG32_SOC15(VCN, vcn_inst, regUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, vcn_inst, regUVD_SUVD_CGC_CTRL);
+	data |= (UVD_SUVD_CGC_CTRL__SRE_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SIT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SMP_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SCM_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SDB_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__ENT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__IME_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SITE_MODE_MASK);
+	WREG32_SOC15(VCN, vcn_inst, regUVD_SUVD_CGC_CTRL, data);
+}
+
+/**
+ * vcn_v4_0_3_start_dpg_mode - VCN start with dpg mode
+ *
+ * @vinst: VCN instance
+ * @indirect: indirectly write sram
+ *
+ * Start VCN block with dpg mode
+ */
+static int vcn_v4_0_3_start_dpg_mode(struct amdgpu_vcn_inst *vinst,
+				     bool indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	struct amdgpu_vcn4_fw_shared *fw_shared =
+						adev->vcn.inst[inst_idx].fw_shared.cpu_addr;
+	struct amdgpu_ring *ring;
+	int vcn_inst, ret;
+	uint32_t tmp;
+
+	vcn_inst = GET_INST(VCN, inst_idx);
+	/* disable register anti-hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_POWER_STATUS), 1,
+		 ~UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+	/* enable dynamic power gating mode */
+	tmp = RREG32_SOC15(VCN, vcn_inst, regUVD_POWER_STATUS);
+	tmp |= UVD_POWER_STATUS__UVD_PG_MODE_MASK;
+	tmp |= UVD_POWER_STATUS__UVD_PG_EN_MASK;
+	WREG32_SOC15(VCN, vcn_inst, regUVD_POWER_STATUS, tmp);
+
+	if (indirect) {
+		DRM_DEV_DEBUG(adev->dev, "VCN %d start: on AID %d",
+			inst_idx, adev->vcn.inst[inst_idx].aid_id);
+		adev->vcn.inst[inst_idx].dpg_sram_curr_addr =
+				(uint32_t *)adev->vcn.inst[inst_idx].dpg_sram_cpu_addr;
+		/* Use dummy register 0xDEADBEEF passing AID selection to PSP FW */
+		WREG32_SOC15_DPG_MODE(inst_idx, 0xDEADBEEF,
+			adev->vcn.inst[inst_idx].aid_id, 0, true);
+	}
+
+	/* enable clock gating */
+	vcn_v4_0_3_disable_clock_gating_dpg_mode(vinst, 0, indirect);
+
+	/* enable VCPU clock */
+	tmp = (0xFF << UVD_VCPU_CNTL__PRB_TIMEOUT_VAL__SHIFT);
+	tmp |= UVD_VCPU_CNTL__CLK_EN_MASK;
+	tmp |= UVD_VCPU_CNTL__BLK_RST_MASK;
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_VCPU_CNTL), tmp, 0, indirect);
+
+	/* disable master interrupt */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_MASTINT_EN), 0, 0, indirect);
+
+	/* setup regUVD_LMI_CTRL */
+	tmp = (UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+		UVD_LMI_CTRL__REQ_MODE_MASK |
+		UVD_LMI_CTRL__CRC_RESET_MASK |
+		UVD_LMI_CTRL__MASK_MC_URGENT_MASK |
+		UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+		UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
+		(8 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
+		0x00100000L);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_LMI_CTRL), tmp, 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_MPC_CNTL),
+		0x2 << UVD_MPC_CNTL__REPLACEMENT_MODE__SHIFT, 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_MPC_SET_MUXA0),
+		((0x1 << UVD_MPC_SET_MUXA0__VARA_1__SHIFT) |
+		 (0x2 << UVD_MPC_SET_MUXA0__VARA_2__SHIFT) |
+		 (0x3 << UVD_MPC_SET_MUXA0__VARA_3__SHIFT) |
+		 (0x4 << UVD_MPC_SET_MUXA0__VARA_4__SHIFT)), 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_MPC_SET_MUXB0),
+		 ((0x1 << UVD_MPC_SET_MUXB0__VARB_1__SHIFT) |
+		 (0x2 << UVD_MPC_SET_MUXB0__VARB_2__SHIFT) |
+		 (0x3 << UVD_MPC_SET_MUXB0__VARB_3__SHIFT) |
+		 (0x4 << UVD_MPC_SET_MUXB0__VARB_4__SHIFT)), 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_MPC_SET_MUX),
+		((0x0 << UVD_MPC_SET_MUX__SET_0__SHIFT) |
+		 (0x1 << UVD_MPC_SET_MUX__SET_1__SHIFT) |
+		 (0x2 << UVD_MPC_SET_MUX__SET_2__SHIFT)), 0, indirect);
+
+	vcn_v4_0_3_mc_resume_dpg_mode(vinst, indirect);
+
+	tmp = (0xFF << UVD_VCPU_CNTL__PRB_TIMEOUT_VAL__SHIFT);
+	tmp |= UVD_VCPU_CNTL__CLK_EN_MASK;
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_VCPU_CNTL), tmp, 0, indirect);
+
+	/* enable LMI MC and UMC channels */
+	tmp = 0x1f << UVD_LMI_CTRL2__RE_OFLD_MIF_WR_REQ_NUM__SHIFT;
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_LMI_CTRL2), tmp, 0, indirect);
+
+	vcn_v4_0_3_enable_ras(adev, inst_idx, indirect);
+
+	/* enable master interrupt */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_MASTINT_EN),
+		UVD_MASTINT_EN__VCPU_EN_MASK, 0, indirect);
+
+	if (indirect) {
+		ret = amdgpu_vcn_psp_update_sram(adev, inst_idx, AMDGPU_UCODE_ID_VCN0_RAM);
+		if (ret) {
+			dev_err(adev->dev, "vcn sram load failed %d\n", ret);
+			return ret;
+		}
+	}
+
+	ring = &adev->vcn.inst[inst_idx].ring_enc[0];
+
+	/* program the RB_BASE for ring buffer */
+	WREG32_SOC15(VCN, vcn_inst, regUVD_RB_BASE_LO,
+		     lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15(VCN, vcn_inst, regUVD_RB_BASE_HI,
+		     upper_32_bits(ring->gpu_addr));
+
+	WREG32_SOC15(VCN, vcn_inst, regUVD_RB_SIZE,
+		     ring->ring_size / sizeof(uint32_t));
+
+	/* resetting ring, fw should not check RB ring */
+	tmp = RREG32_SOC15(VCN, vcn_inst, regVCN_RB_ENABLE);
+	tmp &= ~(VCN_RB_ENABLE__RB_EN_MASK);
+	WREG32_SOC15(VCN, vcn_inst, regVCN_RB_ENABLE, tmp);
+	fw_shared->sq.queue_mode |= FW_QUEUE_RING_RESET;
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32_SOC15(VCN, vcn_inst, regUVD_RB_RPTR, 0);
+	WREG32_SOC15(VCN, vcn_inst, regUVD_RB_WPTR, 0);
+	ring->wptr = RREG32_SOC15(VCN, vcn_inst, regUVD_RB_WPTR);
+
+	tmp = RREG32_SOC15(VCN, vcn_inst, regVCN_RB_ENABLE);
+	tmp |= VCN_RB_ENABLE__RB_EN_MASK;
+	WREG32_SOC15(VCN, vcn_inst, regVCN_RB_ENABLE, tmp);
+	fw_shared->sq.queue_mode &= ~(FW_QUEUE_RING_RESET | FW_QUEUE_DPG_HOLD_OFF);
+
+	/*resetting done, fw can check RB ring */
+	fw_shared->sq.queue_mode &= cpu_to_le32(~FW_QUEUE_RING_RESET);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, vcn_inst, regUVD_STATUS);
+
+	return 0;
+}
+
+static int vcn_v4_0_3_start_sriov(struct amdgpu_device *adev)
+{
+	int i, vcn_inst;
+	struct amdgpu_ring *ring_enc;
+	uint64_t cache_addr;
+	uint64_t rb_enc_addr;
+	uint64_t ctx_addr;
+	uint32_t param, resp, expected;
+	uint32_t offset, cache_size;
+	uint32_t tmp, timeout;
+
+	struct amdgpu_mm_table *table = &adev->virt.mm_table;
+	uint32_t *table_loc;
+	uint32_t table_size;
+	uint32_t size, size_dw;
+	uint32_t init_status;
+	uint32_t enabled_vcn;
+
+	struct mmsch_v4_0_cmd_direct_write
+		direct_wt = { {0} };
+	struct mmsch_v4_0_cmd_direct_read_modify_write
+		direct_rd_mod_wt = { {0} };
+	struct mmsch_v4_0_cmd_end end = { {0} };
+	struct mmsch_v4_0_3_init_header header;
+
+	struct amdgpu_vcn4_fw_shared *fw_shared;
+	struct amdgpu_fw_shared_rb_setup *rb_setup;
+
+	direct_wt.cmd_header.command_type =
+		MMSCH_COMMAND__DIRECT_REG_WRITE;
+	direct_rd_mod_wt.cmd_header.command_type =
+		MMSCH_COMMAND__DIRECT_REG_READ_MODIFY_WRITE;
+	end.cmd_header.command_type = MMSCH_COMMAND__END;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		vcn_inst = GET_INST(VCN, i);
+
+		vcn_v4_0_3_fw_shared_init(adev, vcn_inst);
+
+		memset(&header, 0, sizeof(struct mmsch_v4_0_3_init_header));
+		header.version = MMSCH_VERSION;
+		header.total_size = sizeof(struct mmsch_v4_0_3_init_header) >> 2;
+
+		table_loc = (uint32_t *)table->cpu_addr;
+		table_loc += header.total_size;
+
+		table_size = 0;
+
+		MMSCH_V4_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(VCN, 0, regUVD_STATUS),
+			~UVD_STATUS__UVD_BUSY, UVD_STATUS__UVD_BUSY);
+
+		cache_size = AMDGPU_GPU_PAGE_ALIGN(adev->vcn.inst[i].fw->size + 4);
+
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+			MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+				regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+				adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + i].tmr_mc_addr_lo);
+
+			MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+				regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+				adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + i].tmr_mc_addr_hi);
+
+			offset = 0;
+			MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+				regUVD_VCPU_CACHE_OFFSET0), 0);
+		} else {
+			MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+				regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+				lower_32_bits(adev->vcn.inst[i].gpu_addr));
+			MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+				regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+				upper_32_bits(adev->vcn.inst[i].gpu_addr));
+			offset = cache_size;
+			MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+				regUVD_VCPU_CACHE_OFFSET0),
+				AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
+		}
+
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_VCPU_CACHE_SIZE0),
+			cache_size);
+
+		cache_addr = adev->vcn.inst[vcn_inst].gpu_addr + offset;
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW), lower_32_bits(cache_addr));
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH), upper_32_bits(cache_addr));
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_VCPU_CACHE_OFFSET1), 0);
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_VCPU_CACHE_SIZE1), AMDGPU_VCN_STACK_SIZE);
+
+		cache_addr = adev->vcn.inst[vcn_inst].gpu_addr + offset +
+			AMDGPU_VCN_STACK_SIZE;
+
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW), lower_32_bits(cache_addr));
+
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH), upper_32_bits(cache_addr));
+
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_VCPU_CACHE_OFFSET2), 0);
+
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_VCPU_CACHE_SIZE2), AMDGPU_VCN_CONTEXT_SIZE);
+
+		fw_shared = adev->vcn.inst[vcn_inst].fw_shared.cpu_addr;
+		rb_setup = &fw_shared->rb_setup;
+
+		ring_enc = &adev->vcn.inst[vcn_inst].ring_enc[0];
+		ring_enc->wptr = 0;
+		rb_enc_addr = ring_enc->gpu_addr;
+
+		rb_setup->is_rb_enabled_flags |= RB_ENABLED;
+		rb_setup->rb_addr_lo = lower_32_bits(rb_enc_addr);
+		rb_setup->rb_addr_hi = upper_32_bits(rb_enc_addr);
+		rb_setup->rb_size = ring_enc->ring_size / 4;
+		fw_shared->present_flag_0 |= cpu_to_le32(AMDGPU_VCN_VF_RB_SETUP_FLAG);
+
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_LMI_VCPU_NC0_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[vcn_inst].fw_shared.gpu_addr));
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[vcn_inst].fw_shared.gpu_addr));
+		MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_VCPU_NONCACHE_SIZE0),
+			AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn4_fw_shared)));
+		MMSCH_V4_0_INSERT_END();
+
+		header.vcn0.init_status = 0;
+		header.vcn0.table_offset = header.total_size;
+		header.vcn0.table_size = table_size;
+		header.total_size += table_size;
+
+		/* Send init table to mmsch */
+		size = sizeof(struct mmsch_v4_0_3_init_header);
+		table_loc = (uint32_t *)table->cpu_addr;
+		memcpy((void *)table_loc, &header, size);
+
+		ctx_addr = table->gpu_addr;
+		WREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_CTX_ADDR_LO, lower_32_bits(ctx_addr));
+		WREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_CTX_ADDR_HI, upper_32_bits(ctx_addr));
+
+		tmp = RREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_VMID);
+		tmp &= ~MMSCH_VF_VMID__VF_CTX_VMID_MASK;
+		tmp |= (0 << MMSCH_VF_VMID__VF_CTX_VMID__SHIFT);
+		WREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_VMID, tmp);
+
+		size = header.total_size;
+		WREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_CTX_SIZE, size);
+
+		WREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_MAILBOX_RESP, 0);
+
+		param = 0x00000001;
+		WREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_MAILBOX_HOST, param);
+		tmp = 0;
+		timeout = 1000;
+		resp = 0;
+		expected = MMSCH_VF_MAILBOX_RESP__OK;
+		while (resp != expected) {
+			resp = RREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_MAILBOX_RESP);
+			if (resp != 0)
+				break;
+
+			udelay(10);
+			tmp = tmp + 10;
+			if (tmp >= timeout) {
+				DRM_ERROR("failed to init MMSCH. TIME-OUT after %d usec"\
+					" waiting for regMMSCH_VF_MAILBOX_RESP "\
+					"(expected=0x%08x, readback=0x%08x)\n",
+					tmp, expected, resp);
+				return -EBUSY;
+			}
+		}
+
+		enabled_vcn = amdgpu_vcn_is_disabled_vcn(adev, VCN_DECODE_RING, 0) ? 1 : 0;
+		init_status = ((struct mmsch_v4_0_3_init_header *)(table_loc))->vcn0.init_status;
+		if (resp != expected && resp != MMSCH_VF_MAILBOX_RESP__INCOMPLETE
+					&& init_status != MMSCH_VF_ENGINE_STATUS__PASS) {
+			DRM_ERROR("MMSCH init status is incorrect! readback=0x%08x, header init "\
+				"status for VCN%x: 0x%x\n", resp, enabled_vcn, init_status);
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_3_start - VCN start
+ *
+ * @vinst: VCN instance
+ *
+ * Start VCN block
+ */
+static int vcn_v4_0_3_start(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int i = vinst->inst;
+	struct amdgpu_vcn4_fw_shared *fw_shared;
+	struct amdgpu_ring *ring;
+	int j, k, r, vcn_inst;
+	uint32_t tmp;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)
+		return vcn_v4_0_3_start_dpg_mode(vinst, adev->vcn.inst[i].indirect_sram);
+
+	vcn_inst = GET_INST(VCN, i);
+	/* set VCN status busy */
+	tmp = RREG32_SOC15(VCN, vcn_inst, regUVD_STATUS) |
+		UVD_STATUS__UVD_BUSY;
+	WREG32_SOC15(VCN, vcn_inst, regUVD_STATUS, tmp);
+
+	/* SW clock gating */
+	vcn_v4_0_3_disable_clock_gating(vinst);
+
+	/* enable VCPU clock */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_VCPU_CNTL),
+		 UVD_VCPU_CNTL__CLK_EN_MASK,
+		 ~UVD_VCPU_CNTL__CLK_EN_MASK);
+
+	/* disable master interrupt */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_MASTINT_EN), 0,
+		 ~UVD_MASTINT_EN__VCPU_EN_MASK);
+
+	/* enable LMI MC and UMC channels */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_LMI_CTRL2), 0,
+		 ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+
+	tmp = RREG32_SOC15(VCN, vcn_inst, regUVD_SOFT_RESET);
+	tmp &= ~UVD_SOFT_RESET__LMI_SOFT_RESET_MASK;
+	tmp &= ~UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK;
+	WREG32_SOC15(VCN, vcn_inst, regUVD_SOFT_RESET, tmp);
+
+	/* setup regUVD_LMI_CTRL */
+	tmp = RREG32_SOC15(VCN, vcn_inst, regUVD_LMI_CTRL);
+	WREG32_SOC15(VCN, vcn_inst, regUVD_LMI_CTRL,
+		     tmp | UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+		     UVD_LMI_CTRL__MASK_MC_URGENT_MASK |
+		     UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+		     UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK);
+
+	/* setup regUVD_MPC_CNTL */
+	tmp = RREG32_SOC15(VCN, vcn_inst, regUVD_MPC_CNTL);
+	tmp &= ~UVD_MPC_CNTL__REPLACEMENT_MODE_MASK;
+	tmp |= 0x2 << UVD_MPC_CNTL__REPLACEMENT_MODE__SHIFT;
+	WREG32_SOC15(VCN, vcn_inst, regUVD_MPC_CNTL, tmp);
+
+	/* setup UVD_MPC_SET_MUXA0 */
+	WREG32_SOC15(VCN, vcn_inst, regUVD_MPC_SET_MUXA0,
+		     ((0x1 << UVD_MPC_SET_MUXA0__VARA_1__SHIFT) |
+		      (0x2 << UVD_MPC_SET_MUXA0__VARA_2__SHIFT) |
+		      (0x3 << UVD_MPC_SET_MUXA0__VARA_3__SHIFT) |
+		      (0x4 << UVD_MPC_SET_MUXA0__VARA_4__SHIFT)));
+
+	/* setup UVD_MPC_SET_MUXB0 */
+	WREG32_SOC15(VCN, vcn_inst, regUVD_MPC_SET_MUXB0,
+		     ((0x1 << UVD_MPC_SET_MUXB0__VARB_1__SHIFT) |
+		      (0x2 << UVD_MPC_SET_MUXB0__VARB_2__SHIFT) |
+		      (0x3 << UVD_MPC_SET_MUXB0__VARB_3__SHIFT) |
+		      (0x4 << UVD_MPC_SET_MUXB0__VARB_4__SHIFT)));
+
+	/* setup UVD_MPC_SET_MUX */
+	WREG32_SOC15(VCN, vcn_inst, regUVD_MPC_SET_MUX,
+		     ((0x0 << UVD_MPC_SET_MUX__SET_0__SHIFT) |
+		      (0x1 << UVD_MPC_SET_MUX__SET_1__SHIFT) |
+		      (0x2 << UVD_MPC_SET_MUX__SET_2__SHIFT)));
+
+	vcn_v4_0_3_mc_resume(vinst);
+
+	/* VCN global tiling registers */
+	WREG32_SOC15(VCN, vcn_inst, regUVD_GFX8_ADDR_CONFIG,
+		     adev->gfx.config.gb_addr_config);
+	WREG32_SOC15(VCN, vcn_inst, regUVD_GFX10_ADDR_CONFIG,
+		     adev->gfx.config.gb_addr_config);
+
+	/* unblock VCPU register access */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_RB_ARB_CTRL), 0,
+		 ~UVD_RB_ARB_CTRL__VCPU_DIS_MASK);
+
+	/* release VCPU reset to boot */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_VCPU_CNTL), 0,
+		 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+	for (j = 0; j < 10; ++j) {
+		uint32_t status;
+
+		for (k = 0; k < 100; ++k) {
+			status = RREG32_SOC15(VCN, vcn_inst,
+					      regUVD_STATUS);
+			if (status & 2)
+				break;
+			mdelay(10);
+		}
+		r = 0;
+		if (status & 2)
+			break;
+
+		DRM_DEV_ERROR(adev->dev,
+			      "VCN decode not responding, trying to reset the VCPU!!!\n");
+		WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst,
+					  regUVD_VCPU_CNTL),
+			 UVD_VCPU_CNTL__BLK_RST_MASK,
+			 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+		mdelay(10);
+		WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst,
+					  regUVD_VCPU_CNTL),
+			 0, ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+		mdelay(10);
+		r = -1;
+	}
+
+	if (r) {
+		DRM_DEV_ERROR(adev->dev, "VCN decode not responding, giving up!!!\n");
+		return r;
+	}
+
+	/* enable master interrupt */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_MASTINT_EN),
+		 UVD_MASTINT_EN__VCPU_EN_MASK,
+		 ~UVD_MASTINT_EN__VCPU_EN_MASK);
+
+	/* clear the busy bit of VCN_STATUS */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_STATUS), 0,
+		 ~(2 << UVD_STATUS__VCPU_REPORT__SHIFT));
+
+	ring = &adev->vcn.inst[i].ring_enc[0];
+	fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+
+	/* program the RB_BASE for ring buffer */
+	WREG32_SOC15(VCN, vcn_inst, regUVD_RB_BASE_LO,
+		     lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15(VCN, vcn_inst, regUVD_RB_BASE_HI,
+		     upper_32_bits(ring->gpu_addr));
+
+	WREG32_SOC15(VCN, vcn_inst, regUVD_RB_SIZE,
+		     ring->ring_size / sizeof(uint32_t));
+
+	/* resetting ring, fw should not check RB ring */
+	tmp = RREG32_SOC15(VCN, vcn_inst, regVCN_RB_ENABLE);
+	tmp &= ~(VCN_RB_ENABLE__RB_EN_MASK);
+	WREG32_SOC15(VCN, vcn_inst, regVCN_RB_ENABLE, tmp);
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32_SOC15(VCN, vcn_inst, regUVD_RB_RPTR, 0);
+	WREG32_SOC15(VCN, vcn_inst, regUVD_RB_WPTR, 0);
+
+	tmp = RREG32_SOC15(VCN, vcn_inst, regVCN_RB_ENABLE);
+	tmp |= VCN_RB_ENABLE__RB_EN_MASK;
+	WREG32_SOC15(VCN, vcn_inst, regVCN_RB_ENABLE, tmp);
+
+	ring->wptr = RREG32_SOC15(VCN, vcn_inst, regUVD_RB_WPTR);
+	fw_shared->sq.queue_mode &=
+		cpu_to_le32(~(FW_QUEUE_RING_RESET | FW_QUEUE_DPG_HOLD_OFF));
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_3_stop_dpg_mode - VCN stop with dpg mode
+ *
+ * @vinst: VCN instance
+ *
+ * Stop VCN block with dpg mode
+ */
+static int vcn_v4_0_3_stop_dpg_mode(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t tmp;
+	int vcn_inst;
+
+	vcn_inst = GET_INST(VCN, inst_idx);
+
+	/* Wait for power status to be 1 */
+	SOC15_WAIT_ON_RREG(VCN, vcn_inst, regUVD_POWER_STATUS, 1,
+			   UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+	/* wait for read ptr to be equal to write ptr */
+	tmp = RREG32_SOC15(VCN, vcn_inst, regUVD_RB_WPTR);
+	SOC15_WAIT_ON_RREG(VCN, vcn_inst, regUVD_RB_RPTR, tmp, 0xFFFFFFFF);
+
+	SOC15_WAIT_ON_RREG(VCN, vcn_inst, regUVD_POWER_STATUS, 1,
+			   UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+	/* disable dynamic power gating mode */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_POWER_STATUS), 0,
+		 ~UVD_POWER_STATUS__UVD_PG_MODE_MASK);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, vcn_inst, regUVD_STATUS);
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_3_stop - VCN stop
+ *
+ * @vinst: VCN instance
+ *
+ * Stop VCN block
+ */
+static int vcn_v4_0_3_stop(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int i = vinst->inst;
+	struct amdgpu_vcn4_fw_shared *fw_shared;
+	int r = 0, vcn_inst;
+	uint32_t tmp;
+
+	vcn_inst = GET_INST(VCN, i);
+
+	fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+	fw_shared->sq.queue_mode |= FW_QUEUE_DPG_HOLD_OFF;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) {
+		vcn_v4_0_3_stop_dpg_mode(vinst);
+		goto Done;
+	}
+
+	/* wait for vcn idle */
+	r = SOC15_WAIT_ON_RREG(VCN, vcn_inst, regUVD_STATUS,
+			       UVD_STATUS__IDLE, 0x7);
+	if (r)
+		goto Done;
+
+	tmp = UVD_LMI_STATUS__VCPU_LMI_WRITE_CLEAN_MASK |
+		UVD_LMI_STATUS__READ_CLEAN_MASK |
+		UVD_LMI_STATUS__WRITE_CLEAN_MASK |
+		UVD_LMI_STATUS__WRITE_CLEAN_RAW_MASK;
+	r = SOC15_WAIT_ON_RREG(VCN, vcn_inst, regUVD_LMI_STATUS, tmp,
+			       tmp);
+	if (r)
+		goto Done;
+
+	/* stall UMC channel */
+	tmp = RREG32_SOC15(VCN, vcn_inst, regUVD_LMI_CTRL2);
+	tmp |= UVD_LMI_CTRL2__STALL_ARB_UMC_MASK;
+	WREG32_SOC15(VCN, vcn_inst, regUVD_LMI_CTRL2, tmp);
+	tmp = UVD_LMI_STATUS__UMC_READ_CLEAN_RAW_MASK |
+		UVD_LMI_STATUS__UMC_WRITE_CLEAN_RAW_MASK;
+	r = SOC15_WAIT_ON_RREG(VCN, vcn_inst, regUVD_LMI_STATUS, tmp,
+			       tmp);
+	if (r)
+		goto Done;
+
+	/* Unblock VCPU Register access */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_RB_ARB_CTRL),
+		 UVD_RB_ARB_CTRL__VCPU_DIS_MASK,
+		 ~UVD_RB_ARB_CTRL__VCPU_DIS_MASK);
+
+	/* release VCPU reset to boot */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_VCPU_CNTL),
+		 UVD_VCPU_CNTL__BLK_RST_MASK,
+		 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+	/* disable VCPU clock */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_VCPU_CNTL), 0,
+		 ~(UVD_VCPU_CNTL__CLK_EN_MASK));
+
+	/* reset LMI UMC/LMI/VCPU */
+	tmp = RREG32_SOC15(VCN, vcn_inst, regUVD_SOFT_RESET);
+	tmp |= UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK;
+	WREG32_SOC15(VCN, vcn_inst, regUVD_SOFT_RESET, tmp);
+
+	tmp = RREG32_SOC15(VCN, vcn_inst, regUVD_SOFT_RESET);
+	tmp |= UVD_SOFT_RESET__LMI_SOFT_RESET_MASK;
+	WREG32_SOC15(VCN, vcn_inst, regUVD_SOFT_RESET, tmp);
+
+	/* clear VCN status */
+	WREG32_SOC15(VCN, vcn_inst, regUVD_STATUS, 0);
+
+	/* apply HW clock gating */
+	vcn_v4_0_3_enable_clock_gating(vinst);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, vcn_inst, regUVD_STATUS);
+
+Done:
+	return 0;
+}
+
+/**
+ * vcn_v4_0_3_pause_dpg_mode - VCN pause with dpg mode
+ *
+ * @vinst: VCN instance
+ * @new_state: pause state
+ *
+ * Pause dpg mode for VCN block
+ */
+static int vcn_v4_0_3_pause_dpg_mode(struct amdgpu_vcn_inst *vinst,
+				     struct dpg_pause_state *new_state)
+{
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_3_unified_ring_get_rptr - get unified read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware unified read pointer
+ */
+static uint64_t vcn_v4_0_3_unified_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring != &adev->vcn.inst[ring->me].ring_enc[0])
+		DRM_ERROR("wrong ring id is identified in %s", __func__);
+
+	return RREG32_SOC15(VCN, GET_INST(VCN, ring->me), regUVD_RB_RPTR);
+}
+
+/**
+ * vcn_v4_0_3_unified_ring_get_wptr - get unified write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware unified write pointer
+ */
+static uint64_t vcn_v4_0_3_unified_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring != &adev->vcn.inst[ring->me].ring_enc[0])
+		DRM_ERROR("wrong ring id is identified in %s", __func__);
+
+	if (ring->use_doorbell)
+		return *ring->wptr_cpu_addr;
+	else
+		return RREG32_SOC15(VCN, GET_INST(VCN, ring->me),
+				    regUVD_RB_WPTR);
+}
+
+void vcn_v4_0_3_enc_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+				       uint32_t val, uint32_t mask)
+{
+	/* Use normalized offsets when required */
+	if (vcn_v4_0_3_normalizn_reqd(ring->adev))
+		reg = NORMALIZE_VCN_REG_OFFSET(reg);
+
+	amdgpu_ring_write(ring, VCN_ENC_CMD_REG_WAIT);
+	amdgpu_ring_write(ring, reg << 2);
+	amdgpu_ring_write(ring, mask);
+	amdgpu_ring_write(ring, val);
+}
+
+void vcn_v4_0_3_enc_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg,
+				   uint32_t val)
+{
+	/* Use normalized offsets when required */
+	if (vcn_v4_0_3_normalizn_reqd(ring->adev))
+		reg = NORMALIZE_VCN_REG_OFFSET(reg);
+
+	amdgpu_ring_write(ring, VCN_ENC_CMD_REG_WRITE);
+	amdgpu_ring_write(ring,	reg << 2);
+	amdgpu_ring_write(ring, val);
+}
+
+void vcn_v4_0_3_enc_ring_emit_vm_flush(struct amdgpu_ring *ring,
+				       unsigned int vmid, uint64_t pd_addr)
+{
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
+
+	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* wait for reg writes */
+	vcn_v4_0_3_enc_ring_emit_reg_wait(ring, hub->ctx0_ptb_addr_lo32 +
+					vmid * hub->ctx_addr_distance,
+					lower_32_bits(pd_addr), 0xffffffff);
+}
+
+void vcn_v4_0_3_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	/* VCN engine access for HDP flush doesn't work when RRMT is enabled.
+	 * This is a workaround to avoid any HDP flush through VCN ring.
+	 */
+}
+
+/**
+ * vcn_v4_0_3_unified_ring_set_wptr - set enc write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the enc write pointer to the hardware
+ */
+static void vcn_v4_0_3_unified_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring != &adev->vcn.inst[ring->me].ring_enc[0])
+		DRM_ERROR("wrong ring id is identified in %s", __func__);
+
+	if (ring->use_doorbell) {
+		*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
+		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+	} else {
+		WREG32_SOC15(VCN, GET_INST(VCN, ring->me), regUVD_RB_WPTR,
+			     lower_32_bits(ring->wptr));
+	}
+}
+
+static int vcn_v4_0_3_ring_reset(struct amdgpu_ring *ring,
+				 unsigned int vmid,
+				 struct amdgpu_fence *timedout_fence)
+{
+	int r = 0;
+	int vcn_inst;
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[ring->me];
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+
+	vcn_inst = GET_INST(VCN, ring->me);
+	r = amdgpu_dpm_reset_vcn(adev, 1 << vcn_inst);
+
+	if (r) {
+		DRM_DEV_ERROR(adev->dev, "VCN reset fail : %d\n", r);
+		return r;
+	}
+
+	/* This flag is not set for VF, assumed to be disabled always */
+	if (RREG32_SOC15(VCN, GET_INST(VCN, 0), regVCN_RRMT_CNTL) & 0x100)
+		adev->vcn.caps |= AMDGPU_VCN_CAPS(RRMT_ENABLED);
+	vcn_v4_0_3_hw_init_inst(vinst);
+	vcn_v4_0_3_start_dpg_mode(vinst, adev->vcn.inst[ring->me].indirect_sram);
+
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static const struct amdgpu_ring_funcs vcn_v4_0_3_unified_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_ENC,
+	.align_mask = 0x3f,
+	.nop = VCN_ENC_CMD_NO_OP,
+	.get_rptr = vcn_v4_0_3_unified_ring_get_rptr,
+	.get_wptr = vcn_v4_0_3_unified_ring_get_wptr,
+	.set_wptr = vcn_v4_0_3_unified_ring_set_wptr,
+	.emit_frame_size =
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 4 +
+		4 + /* vcn_v2_0_enc_ring_emit_vm_flush */
+		5 + 5 + /* vcn_v2_0_enc_ring_emit_fence x2 vm fence */
+		1, /* vcn_v2_0_enc_ring_insert_end */
+	.emit_ib_size = 5, /* vcn_v2_0_enc_ring_emit_ib */
+	.emit_ib = vcn_v2_0_enc_ring_emit_ib,
+	.emit_fence = vcn_v2_0_enc_ring_emit_fence,
+	.emit_vm_flush = vcn_v4_0_3_enc_ring_emit_vm_flush,
+	.emit_hdp_flush = vcn_v4_0_3_ring_emit_hdp_flush,
+	.test_ring = amdgpu_vcn_enc_ring_test_ring,
+	.test_ib = amdgpu_vcn_unified_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.insert_end = vcn_v2_0_enc_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_vcn_ring_begin_use,
+	.end_use = amdgpu_vcn_ring_end_use,
+	.emit_wreg = vcn_v4_0_3_enc_ring_emit_wreg,
+	.emit_reg_wait = vcn_v4_0_3_enc_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = vcn_v4_0_3_ring_reset,
+};
+
+/**
+ * vcn_v4_0_3_set_unified_ring_funcs - set unified ring functions
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set unified ring functions
+ */
+static void vcn_v4_0_3_set_unified_ring_funcs(struct amdgpu_device *adev)
+{
+	int i, vcn_inst;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		adev->vcn.inst[i].ring_enc[0].funcs = &vcn_v4_0_3_unified_ring_vm_funcs;
+		adev->vcn.inst[i].ring_enc[0].me = i;
+		vcn_inst = GET_INST(VCN, i);
+		adev->vcn.inst[i].aid_id =
+			vcn_inst / adev->vcn.num_inst_per_aid;
+	}
+}
+
+/**
+ * vcn_v4_0_3_is_idle - check VCN block is idle
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block structure
+ *
+ * Check whether VCN block is idle
+ */
+static bool vcn_v4_0_3_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, ret = 1;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		ret &= (RREG32_SOC15(VCN, GET_INST(VCN, i), regUVD_STATUS) ==
+			UVD_STATUS__IDLE);
+	}
+
+	return ret;
+}
+
+/**
+ * vcn_v4_0_3_wait_for_idle - wait for VCN block idle
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Wait for VCN block idle
+ */
+static int vcn_v4_0_3_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, ret = 0;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		ret = SOC15_WAIT_ON_RREG(VCN, GET_INST(VCN, i), regUVD_STATUS,
+					 UVD_STATUS__IDLE, UVD_STATUS__IDLE);
+		if (ret)
+			return ret;
+	}
+
+	return ret;
+}
+
+/* vcn_v4_0_3_set_clockgating_state - set VCN block clockgating state
+ *
+ * @ip_block: amdgpu_ip_block pointer
+ * @state: clock gating state
+ *
+ * Set VCN block clockgating state
+ */
+static int vcn_v4_0_3_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = state == AMD_CG_STATE_GATE;
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[i];
+
+		if (enable) {
+			if (RREG32_SOC15(VCN, GET_INST(VCN, i),
+					 regUVD_STATUS) != UVD_STATUS__IDLE)
+				return -EBUSY;
+			vcn_v4_0_3_enable_clock_gating(vinst);
+		} else {
+			vcn_v4_0_3_disable_clock_gating(vinst);
+		}
+	}
+	return 0;
+}
+
+static int vcn_v4_0_3_set_pg_state(struct amdgpu_vcn_inst *vinst,
+				   enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int ret = 0;
+
+	/* for SRIOV, guest should not control VCN Power-gating
+	 * MMSCH FW should control Power-gating and clock-gating
+	 * guest should avoid touching CGC and PG
+	 */
+	if (amdgpu_sriov_vf(adev)) {
+		vinst->cur_state = AMD_PG_STATE_UNGATE;
+		return 0;
+	}
+
+	if (state == vinst->cur_state)
+		return 0;
+
+	if (state == AMD_PG_STATE_GATE)
+		ret = vcn_v4_0_3_stop(vinst);
+	else
+		ret = vcn_v4_0_3_start(vinst);
+
+	if (!ret)
+		vinst->cur_state = state;
+
+	return ret;
+}
+
+/**
+ * vcn_v4_0_3_set_interrupt_state - set VCN block interrupt state
+ *
+ * @adev: amdgpu_device pointer
+ * @source: interrupt sources
+ * @type: interrupt types
+ * @state: interrupt states
+ *
+ * Set VCN block interrupt state
+ */
+static int vcn_v4_0_3_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned int type,
+					enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+/**
+ * vcn_v4_0_3_process_interrupt - process VCN block interrupt
+ *
+ * @adev: amdgpu_device pointer
+ * @source: interrupt sources
+ * @entry: interrupt entry from clients and sources
+ *
+ * Process VCN block interrupt
+ */
+static int vcn_v4_0_3_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	uint32_t i, inst;
+
+	i = node_id_to_phys_map[entry->node_id];
+
+	DRM_DEV_DEBUG(adev->dev, "IH: VCN TRAP\n");
+
+	for (inst = 0; inst < adev->vcn.num_vcn_inst; ++inst)
+		if (adev->vcn.inst[inst].aid_id == i)
+			break;
+
+	if (inst >= adev->vcn.num_vcn_inst) {
+		dev_WARN_ONCE(adev->dev, 1,
+			      "Interrupt received for unknown VCN instance %d",
+			      entry->node_id);
+		return 0;
+	}
+
+	switch (entry->src_id) {
+	case VCN_4_0__SRCID__UVD_ENC_GENERAL_PURPOSE:
+		amdgpu_fence_process(&adev->vcn.inst[inst].ring_enc[0]);
+		break;
+	default:
+		DRM_DEV_ERROR(adev->dev, "Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+static int vcn_v4_0_3_set_ras_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned int type,
+					enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs vcn_v4_0_3_irq_funcs = {
+	.set = vcn_v4_0_3_set_interrupt_state,
+	.process = vcn_v4_0_3_process_interrupt,
+};
+
+static const struct amdgpu_irq_src_funcs vcn_v4_0_3_ras_irq_funcs = {
+	.set = vcn_v4_0_3_set_ras_interrupt_state,
+	.process = amdgpu_vcn_process_poison_irq,
+};
+
+/**
+ * vcn_v4_0_3_set_irq_funcs - set VCN block interrupt irq functions
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set VCN block interrupt irq functions
+ */
+static void vcn_v4_0_3_set_irq_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		adev->vcn.inst->irq.num_types++;
+	}
+	adev->vcn.inst->irq.funcs = &vcn_v4_0_3_irq_funcs;
+
+	adev->vcn.inst->ras_poison_irq.num_types = 1;
+	adev->vcn.inst->ras_poison_irq.funcs = &vcn_v4_0_3_ras_irq_funcs;
+}
+
+static const struct amd_ip_funcs vcn_v4_0_3_ip_funcs = {
+	.name = "vcn_v4_0_3",
+	.early_init = vcn_v4_0_3_early_init,
+	.late_init = vcn_v4_0_3_late_init,
+	.sw_init = vcn_v4_0_3_sw_init,
+	.sw_fini = vcn_v4_0_3_sw_fini,
+	.hw_init = vcn_v4_0_3_hw_init,
+	.hw_fini = vcn_v4_0_3_hw_fini,
+	.suspend = vcn_v4_0_3_suspend,
+	.resume = vcn_v4_0_3_resume,
+	.is_idle = vcn_v4_0_3_is_idle,
+	.wait_for_idle = vcn_v4_0_3_wait_for_idle,
+	.set_clockgating_state = vcn_v4_0_3_set_clockgating_state,
+	.set_powergating_state = vcn_set_powergating_state,
+	.dump_ip_state = amdgpu_vcn_dump_ip_state,
+	.print_ip_state = amdgpu_vcn_print_ip_state,
+};
+
+const struct amdgpu_ip_block_version vcn_v4_0_3_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_VCN,
+	.major = 4,
+	.minor = 0,
+	.rev = 3,
+	.funcs = &vcn_v4_0_3_ip_funcs,
+};
+
+static const struct amdgpu_ras_err_status_reg_entry vcn_v4_0_3_ue_reg_list[] = {
+	{AMDGPU_RAS_REG_ENTRY(VCN, 0, regVCN_UE_ERR_STATUS_LO_VIDD, regVCN_UE_ERR_STATUS_HI_VIDD),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "VIDD"},
+	{AMDGPU_RAS_REG_ENTRY(VCN, 0, regVCN_UE_ERR_STATUS_LO_VIDV, regVCN_UE_ERR_STATUS_HI_VIDV),
+	1, (AMDGPU_RAS_ERR_INFO_VALID | AMDGPU_RAS_ERR_STATUS_VALID), "VIDV"},
+};
+
+static void vcn_v4_0_3_inst_query_ras_error_count(struct amdgpu_device *adev,
+						  uint32_t vcn_inst,
+						  void *ras_err_status)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)ras_err_status;
+
+	/* vcn v4_0_3 only support query uncorrectable errors */
+	amdgpu_ras_inst_query_ras_error_count(adev,
+			vcn_v4_0_3_ue_reg_list,
+			ARRAY_SIZE(vcn_v4_0_3_ue_reg_list),
+			NULL, 0, GET_INST(VCN, vcn_inst),
+			AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE,
+			&err_data->ue_count);
+}
+
+static void vcn_v4_0_3_query_ras_error_count(struct amdgpu_device *adev,
+					     void *ras_err_status)
+{
+	uint32_t i;
+
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__VCN)) {
+		dev_warn(adev->dev, "VCN RAS is not supported\n");
+		return;
+	}
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++)
+		vcn_v4_0_3_inst_query_ras_error_count(adev, i, ras_err_status);
+}
+
+static void vcn_v4_0_3_inst_reset_ras_error_count(struct amdgpu_device *adev,
+						  uint32_t vcn_inst)
+{
+	amdgpu_ras_inst_reset_ras_error_count(adev,
+					vcn_v4_0_3_ue_reg_list,
+					ARRAY_SIZE(vcn_v4_0_3_ue_reg_list),
+					GET_INST(VCN, vcn_inst));
+}
+
+static void vcn_v4_0_3_reset_ras_error_count(struct amdgpu_device *adev)
+{
+	uint32_t i;
+
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__VCN)) {
+		dev_warn(adev->dev, "VCN RAS is not supported\n");
+		return;
+	}
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++)
+		vcn_v4_0_3_inst_reset_ras_error_count(adev, i);
+}
+
+static uint32_t vcn_v4_0_3_query_poison_by_instance(struct amdgpu_device *adev,
+			uint32_t instance, uint32_t sub_block)
+{
+	uint32_t poison_stat = 0, reg_value = 0;
+
+	switch (sub_block) {
+	case AMDGPU_VCN_V4_0_3_VCPU_VCODEC:
+		reg_value = RREG32_SOC15(VCN, instance, regUVD_RAS_VCPU_VCODEC_STATUS);
+		poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_VCPU_VCODEC_STATUS, POISONED_PF);
+		break;
+	default:
+		break;
+	}
+
+	if (poison_stat)
+		dev_info(adev->dev, "Poison detected in VCN%d, sub_block%d\n",
+			instance, sub_block);
+
+	return poison_stat;
+}
+
+static bool vcn_v4_0_3_query_poison_status(struct amdgpu_device *adev)
+{
+	uint32_t inst, sub;
+	uint32_t poison_stat = 0;
+
+	for (inst = 0; inst < adev->vcn.num_vcn_inst; inst++)
+		for (sub = 0; sub < AMDGPU_VCN_V4_0_3_MAX_SUB_BLOCK; sub++)
+			poison_stat +=
+			vcn_v4_0_3_query_poison_by_instance(adev, inst, sub);
+
+	return !!poison_stat;
+}
+
+static const struct amdgpu_ras_block_hw_ops vcn_v4_0_3_ras_hw_ops = {
+	.query_ras_error_count = vcn_v4_0_3_query_ras_error_count,
+	.reset_ras_error_count = vcn_v4_0_3_reset_ras_error_count,
+	.query_poison_status = vcn_v4_0_3_query_poison_status,
+};
+
+static int vcn_v4_0_3_aca_bank_parser(struct aca_handle *handle, struct aca_bank *bank,
+				      enum aca_smu_type type, void *data)
+{
+	struct aca_bank_info info;
+	u64 misc0;
+	int ret;
+
+	ret = aca_bank_info_decode(bank, &info);
+	if (ret)
+		return ret;
+
+	misc0 = bank->regs[ACA_REG_IDX_MISC0];
+	switch (type) {
+	case ACA_SMU_TYPE_UE:
+		bank->aca_err_type = ACA_ERROR_TYPE_UE;
+		ret = aca_error_cache_log_bank_error(handle, &info, ACA_ERROR_TYPE_UE,
+						     1ULL);
+		break;
+	case ACA_SMU_TYPE_CE:
+		bank->aca_err_type = ACA_ERROR_TYPE_CE;
+		ret = aca_error_cache_log_bank_error(handle, &info, bank->aca_err_type,
+						     ACA_REG__MISC0__ERRCNT(misc0));
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return ret;
+}
+
+/* reference to smu driver if header file */
+static int vcn_v4_0_3_err_codes[] = {
+	14, 15, /* VCN */
+};
+
+static bool vcn_v4_0_3_aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank,
+					 enum aca_smu_type type, void *data)
+{
+	u32 instlo;
+
+	instlo = ACA_REG__IPID__INSTANCEIDLO(bank->regs[ACA_REG_IDX_IPID]);
+	instlo &= GENMASK(31, 1);
+
+	if (instlo != mmSMNAID_AID0_MCA_SMU)
+		return false;
+
+	if (aca_bank_check_error_codes(handle->adev, bank,
+				       vcn_v4_0_3_err_codes,
+				       ARRAY_SIZE(vcn_v4_0_3_err_codes)))
+		return false;
+
+	return true;
+}
+
+static const struct aca_bank_ops vcn_v4_0_3_aca_bank_ops = {
+	.aca_bank_parser = vcn_v4_0_3_aca_bank_parser,
+	.aca_bank_is_valid = vcn_v4_0_3_aca_bank_is_valid,
+};
+
+static const struct aca_info vcn_v4_0_3_aca_info = {
+	.hwip = ACA_HWIP_TYPE_SMU,
+	.mask = ACA_ERROR_UE_MASK,
+	.bank_ops = &vcn_v4_0_3_aca_bank_ops,
+};
+
+static int vcn_v4_0_3_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
+{
+	int r;
+
+	r = amdgpu_ras_block_late_init(adev, ras_block);
+	if (r)
+		return r;
+
+	if (amdgpu_ras_is_supported(adev, ras_block->block) &&
+		adev->vcn.inst->ras_poison_irq.funcs) {
+		r = amdgpu_irq_get(adev, &adev->vcn.inst->ras_poison_irq, 0);
+		if (r)
+			goto late_fini;
+	}
+
+	r = amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__VCN,
+				&vcn_v4_0_3_aca_info, NULL);
+	if (r)
+		goto late_fini;
+
+	return 0;
+
+late_fini:
+	amdgpu_ras_block_late_fini(adev, ras_block);
+
+	return r;
+}
+
+static struct amdgpu_vcn_ras vcn_v4_0_3_ras = {
+	.ras_block = {
+		.hw_ops = &vcn_v4_0_3_ras_hw_ops,
+		.ras_late_init = vcn_v4_0_3_ras_late_init,
+	},
+};
+
+static void vcn_v4_0_3_set_ras_funcs(struct amdgpu_device *adev)
+{
+	adev->vcn.ras = &vcn_v4_0_3_ras;
+}
+
+static void vcn_v4_0_3_enable_ras(struct amdgpu_device *adev,
+				  int inst_idx, bool indirect)
+{
+	uint32_t tmp;
+
+	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__VCN))
+		return;
+
+	tmp = VCN_RAS_CNTL__VCPU_VCODEC_REARM_MASK |
+	      VCN_RAS_CNTL__VCPU_VCODEC_IH_EN_MASK |
+	      VCN_RAS_CNTL__VCPU_VCODEC_PMI_EN_MASK |
+	      VCN_RAS_CNTL__VCPU_VCODEC_STALL_EN_MASK;
+	WREG32_SOC15_DPG_MODE(inst_idx,
+			      SOC15_DPG_MODE_OFFSET(VCN, 0, regVCN_RAS_CNTL),
+			      tmp, 0, indirect);
+
+	tmp = UVD_VCPU_INT_EN2__RASCNTL_VCPU_VCODEC_EN_MASK;
+	WREG32_SOC15_DPG_MODE(inst_idx,
+			      SOC15_DPG_MODE_OFFSET(VCN, 0, regUVD_VCPU_INT_EN2),
+			      tmp, 0, indirect);
+
+	tmp = UVD_SYS_INT_EN__RASCNTL_VCPU_VCODEC_EN_MASK;
+	WREG32_SOC15_DPG_MODE(inst_idx,
+			      SOC15_DPG_MODE_OFFSET(VCN, 0, regUVD_SYS_INT_EN),
+			      tmp, 0, indirect);
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.h b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.h
new file mode 100644
index 000000000000..aeab89853a92
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.h
@@ -0,0 +1,44 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VCN_V4_0_3_H__
+#define __VCN_V4_0_3_H__
+
+enum amdgpu_vcn_v4_0_3_sub_block {
+	AMDGPU_VCN_V4_0_3_VCPU_VCODEC = 0,
+
+	AMDGPU_VCN_V4_0_3_MAX_SUB_BLOCK,
+};
+
+extern const struct amdgpu_ip_block_version vcn_v4_0_3_ip_block;
+
+void vcn_v4_0_3_enc_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+				       uint32_t val, uint32_t mask);
+
+void vcn_v4_0_3_enc_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg,
+				   uint32_t val);
+void vcn_v4_0_3_enc_ring_emit_vm_flush(struct amdgpu_ring *ring,
+				       unsigned int vmid, uint64_t pd_addr);
+void vcn_v4_0_3_ring_emit_hdp_flush(struct amdgpu_ring *ring);
+
+#endif /* __VCN_V4_0_3_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c
new file mode 100644
index 000000000000..b107ee80e472
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c
@@ -0,0 +1,1723 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include "amdgpu.h"
+#include "amdgpu_vcn.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_cs.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "soc15_hw_ip.h"
+#include "vcn_v2_0.h"
+#include "mmsch_v4_0.h"
+#include "vcn_v4_0_5.h"
+
+#include "vcn/vcn_4_0_5_offset.h"
+#include "vcn/vcn_4_0_5_sh_mask.h"
+#include "ivsrcid/vcn/irqsrcs_vcn_4_0.h"
+
+#include <drm/drm_drv.h>
+
+#define mmUVD_DPG_LMA_CTL							regUVD_DPG_LMA_CTL
+#define mmUVD_DPG_LMA_CTL_BASE_IDX					regUVD_DPG_LMA_CTL_BASE_IDX
+#define mmUVD_DPG_LMA_DATA							regUVD_DPG_LMA_DATA
+#define mmUVD_DPG_LMA_DATA_BASE_IDX					regUVD_DPG_LMA_DATA_BASE_IDX
+
+#define VCN_VID_SOC_ADDRESS_2_0						0x1fb00
+#define VCN1_VID_SOC_ADDRESS_3_0					(0x48300 + 0x38000)
+#define VCN1_AON_SOC_ADDRESS_3_0					(0x48000 + 0x38000)
+
+#define VCN_HARVEST_MMSCH							0
+
+#define RDECODE_MSG_CREATE							0x00000000
+#define RDECODE_MESSAGE_CREATE						0x00000001
+
+static const struct amdgpu_hwip_reg_entry vcn_reg_list_4_0_5[] = {
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_CONTEXT_ID),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_CONTEXT_ID2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_GPCOM_VCPU_DATA0),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_GPCOM_VCPU_DATA1),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_GPCOM_VCPU_CMD),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_HI),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_LO),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_HI2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_LO2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_HI3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_LO3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_HI4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_LO4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_RPTR),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_WPTR),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_RPTR2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_WPTR2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_RPTR3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_WPTR3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_RPTR4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_WPTR4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_SIZE),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_SIZE2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_SIZE3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_SIZE4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_PGFSM_CONFIG),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_PGFSM_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_DPG_LMA_CTL),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_DPG_LMA_DATA),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_DPG_LMA_MASK),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_DPG_PAUSE)
+};
+
+static int amdgpu_ih_clientid_vcns[] = {
+	SOC15_IH_CLIENTID_VCN,
+	SOC15_IH_CLIENTID_VCN1
+};
+
+static void vcn_v4_0_5_set_unified_ring_funcs(struct amdgpu_device *adev);
+static void vcn_v4_0_5_set_irq_funcs(struct amdgpu_device *adev);
+static int vcn_v4_0_5_set_pg_state(struct amdgpu_vcn_inst *vinst,
+				   enum amd_powergating_state state);
+static int vcn_v4_0_5_pause_dpg_mode(struct amdgpu_vcn_inst *vinst,
+				     struct dpg_pause_state *new_state);
+static void vcn_v4_0_5_unified_ring_set_wptr(struct amdgpu_ring *ring);
+
+/**
+ * vcn_v4_0_5_early_init - set function pointers and load microcode
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Set ring and irq function pointers
+ * Load microcode from filesystem
+ */
+static int vcn_v4_0_5_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, r;
+
+	if (amdgpu_ip_version(adev, UVD_HWIP, 0) == IP_VERSION(4, 0, 6))
+		adev->vcn.per_inst_fw = true;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i)
+		/* re-use enc ring as unified ring */
+		adev->vcn.inst[i].num_enc_rings = 1;
+	vcn_v4_0_5_set_unified_ring_funcs(adev);
+	vcn_v4_0_5_set_irq_funcs(adev);
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		adev->vcn.inst[i].set_pg_state = vcn_v4_0_5_set_pg_state;
+
+		r = amdgpu_vcn_early_init(adev, i);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_5_sw_init - sw init for VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Load firmware and sw initialization
+ */
+static int vcn_v4_0_5_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_ring *ring;
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, r;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		struct amdgpu_vcn4_fw_shared *fw_shared;
+
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		r = amdgpu_vcn_sw_init(adev, i);
+		if (r)
+			return r;
+
+		amdgpu_vcn_setup_ucode(adev, i);
+
+		r = amdgpu_vcn_resume(adev, i);
+		if (r)
+			return r;
+
+		atomic_set(&adev->vcn.inst[i].sched_score, 0);
+
+		/* VCN UNIFIED TRAP */
+		r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_vcns[i],
+				VCN_4_0__SRCID__UVD_ENC_GENERAL_PURPOSE, &adev->vcn.inst[i].irq);
+		if (r)
+			return r;
+
+		/* VCN POISON TRAP */
+		r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_vcns[i],
+				VCN_4_0__SRCID_UVD_POISON, &adev->vcn.inst[i].irq);
+		if (r)
+			return r;
+
+		ring = &adev->vcn.inst[i].ring_enc[0];
+		ring->use_doorbell = true;
+		if (amdgpu_sriov_vf(adev))
+			ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
+						i * (adev->vcn.inst[i].num_enc_rings + 1) + 1;
+		else
+			ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
+						2 + 8 * i;
+		ring->vm_hub = AMDGPU_MMHUB0(0);
+		sprintf(ring->name, "vcn_unified_%d", i);
+
+		r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst[i].irq, 0,
+				AMDGPU_RING_PRIO_0, &adev->vcn.inst[i].sched_score);
+		if (r)
+			return r;
+
+		fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+		fw_shared->present_flag_0 = cpu_to_le32(AMDGPU_FW_SHARED_FLAG_0_UNIFIED_QUEUE);
+		fw_shared->sq.is_enabled = 1;
+
+		fw_shared->present_flag_0 |= cpu_to_le32(AMDGPU_VCN_SMU_DPM_INTERFACE_FLAG);
+		fw_shared->smu_dpm_interface.smu_interface_type = (adev->flags & AMD_IS_APU) ?
+			AMDGPU_VCN_SMU_DPM_INTERFACE_APU : AMDGPU_VCN_SMU_DPM_INTERFACE_DGPU;
+
+		if (amdgpu_sriov_vf(adev))
+			fw_shared->present_flag_0 |= cpu_to_le32(AMDGPU_VCN_VF_RB_SETUP_FLAG);
+
+		fw_shared->present_flag_0 |= AMDGPU_FW_SHARED_FLAG_0_DRM_KEY_INJECT;
+		fw_shared->drm_key_wa.method =
+			AMDGPU_DRM_KEY_INJECT_WORKAROUND_VCNFW_ASD_HANDSHAKING;
+
+		if (amdgpu_vcnfw_log)
+			amdgpu_vcn_fwlog_init(&adev->vcn.inst[i]);
+
+		if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)
+			adev->vcn.inst[i].pause_dpg_mode = vcn_v4_0_5_pause_dpg_mode;
+	}
+
+	adev->vcn.supported_reset = amdgpu_get_soft_full_reset_mask(&adev->vcn.inst[0].ring_enc[0]);
+	if (!amdgpu_sriov_vf(adev))
+		adev->vcn.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+
+	r = amdgpu_vcn_sysfs_reset_mask_init(adev);
+	if (r)
+		return r;
+
+	if (amdgpu_sriov_vf(adev)) {
+		r = amdgpu_virt_alloc_mm_table(adev);
+		if (r)
+			return r;
+	}
+
+	r = amdgpu_vcn_reg_dump_init(adev, vcn_reg_list_4_0_5, ARRAY_SIZE(vcn_reg_list_4_0_5));
+
+	return r;
+}
+
+/**
+ * vcn_v4_0_5_sw_fini - sw fini for VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * VCN suspend and free up sw allocation
+ */
+static int vcn_v4_0_5_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, r, idx;
+
+	if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+		for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+			struct amdgpu_vcn4_fw_shared *fw_shared;
+
+			if (adev->vcn.harvest_config & (1 << i))
+				continue;
+
+			fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+			fw_shared->present_flag_0 = 0;
+			fw_shared->sq.is_enabled = 0;
+		}
+
+		drm_dev_exit(idx);
+	}
+
+	if (amdgpu_sriov_vf(adev))
+		amdgpu_virt_free_mm_table(adev);
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		r = amdgpu_vcn_suspend(adev, i);
+		if (r)
+			return r;
+
+		amdgpu_vcn_sw_fini(adev, i);
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_5_hw_init - start and test VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Initialize the hardware, boot up the VCPU and do some testing
+ */
+static int vcn_v4_0_5_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	int i, r;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		ring = &adev->vcn.inst[i].ring_enc[0];
+
+		adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
+				((adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 8 * i), i);
+
+		r = amdgpu_ring_test_helper(ring);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_5_hw_fini - stop the hardware block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Stop the VCN block, mark ring as not ready any more
+ */
+static int vcn_v4_0_5_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[i];
+
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		cancel_delayed_work_sync(&vinst->idle_work);
+
+		if (!amdgpu_sriov_vf(adev)) {
+			if ((adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) ||
+			    (vinst->cur_state != AMD_PG_STATE_GATE &&
+			     RREG32_SOC15(VCN, i, regUVD_STATUS))) {
+				vinst->set_pg_state(vinst, AMD_PG_STATE_GATE);
+			}
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_5_suspend - suspend VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * HW fini and suspend VCN block
+ */
+static int vcn_v4_0_5_suspend(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r, i;
+
+	r = vcn_v4_0_5_hw_fini(ip_block);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		r = amdgpu_vcn_suspend(ip_block->adev, i);
+		if (r)
+			return r;
+	}
+
+	return r;
+}
+
+/**
+ * vcn_v4_0_5_resume - resume VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Resume firmware and hw init VCN block
+ */
+static int vcn_v4_0_5_resume(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r, i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		r = amdgpu_vcn_resume(ip_block->adev, i);
+		if (r)
+			return r;
+	}
+
+	r = vcn_v4_0_5_hw_init(ip_block);
+
+	return r;
+}
+
+/**
+ * vcn_v4_0_5_mc_resume - memory controller programming
+ *
+ * @vinst: VCN instance
+ *
+ * Let the VCN memory controller know it's offsets
+ */
+static void vcn_v4_0_5_mc_resume(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst = vinst->inst;
+	uint32_t offset, size;
+	const struct common_firmware_header *hdr;
+
+	hdr = (const struct common_firmware_header *)adev->vcn.inst[inst].fw->data;
+	size = AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8);
+
+	/* cache window 0: fw */
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst].tmr_mc_addr_lo));
+		WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst].tmr_mc_addr_hi));
+		WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_OFFSET0, 0);
+		offset = 0;
+	} else {
+		WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			lower_32_bits(adev->vcn.inst[inst].gpu_addr));
+		WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			upper_32_bits(adev->vcn.inst[inst].gpu_addr));
+		offset = size;
+		WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_OFFSET0, AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
+	}
+	WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_SIZE0, size);
+
+	/* cache window 1: stack */
+	WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW,
+		lower_32_bits(adev->vcn.inst[inst].gpu_addr + offset));
+	WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH,
+		upper_32_bits(adev->vcn.inst[inst].gpu_addr + offset));
+	WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_OFFSET1, 0);
+	WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_SIZE1, AMDGPU_VCN_STACK_SIZE);
+
+	/* cache window 2: context */
+	WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW,
+		lower_32_bits(adev->vcn.inst[inst].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE));
+	WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH,
+		upper_32_bits(adev->vcn.inst[inst].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE));
+	WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_OFFSET2, 0);
+	WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_SIZE2, AMDGPU_VCN_CONTEXT_SIZE);
+
+	/* non-cache window */
+	WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_NC0_64BIT_BAR_LOW,
+		lower_32_bits(adev->vcn.inst[inst].fw_shared.gpu_addr));
+	WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH,
+		upper_32_bits(adev->vcn.inst[inst].fw_shared.gpu_addr));
+	WREG32_SOC15(VCN, inst, regUVD_VCPU_NONCACHE_OFFSET0, 0);
+	WREG32_SOC15(VCN, inst, regUVD_VCPU_NONCACHE_SIZE0,
+		AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn4_fw_shared)));
+}
+
+/**
+ * vcn_v4_0_5_mc_resume_dpg_mode - memory controller programming for dpg mode
+ *
+ * @vinst: VCN instance
+ * @indirect: indirectly write sram
+ *
+ * Let the VCN memory controller know it's offsets with dpg mode
+ */
+static void vcn_v4_0_5_mc_resume_dpg_mode(struct amdgpu_vcn_inst *vinst,
+					  bool indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t offset, size;
+	const struct common_firmware_header *hdr;
+
+	hdr = (const struct common_firmware_header *)adev->vcn.inst[inst_idx].fw->data;
+	size = AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8);
+
+	/* cache window 0: fw */
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		if (!indirect) {
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+			(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst_idx].tmr_mc_addr_lo),
+			0, indirect);
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+			(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst_idx].tmr_mc_addr_hi),
+			0, indirect);
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET0), 0, 0, indirect);
+		} else {
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW), 0, 0, indirect);
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH), 0, 0, indirect);
+			WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+				VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET0), 0, 0, indirect);
+		}
+		offset = 0;
+	} else {
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr), 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr), 0, indirect);
+		offset = size;
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET0),
+			AMDGPU_UVD_FIRMWARE_OFFSET >> 3, 0, indirect);
+	}
+
+	if (!indirect)
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_VCPU_CACHE_SIZE0), size, 0, indirect);
+	else
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_VCPU_CACHE_SIZE0), 0, 0, indirect);
+
+	/* cache window 1: stack */
+	if (!indirect) {
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset), 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset), 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET1), 0, 0, indirect);
+	} else {
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW), 0, 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH), 0, 0, indirect);
+		WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET1), 0, 0, indirect);
+	}
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_VCPU_CACHE_SIZE1), AMDGPU_VCN_STACK_SIZE, 0, indirect);
+
+	/* cache window 2: context */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW),
+		lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE),
+		0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH),
+		upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE),
+		0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET2), 0, 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_VCPU_CACHE_SIZE2), AMDGPU_VCN_CONTEXT_SIZE, 0, indirect);
+
+	/* non-cache window */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_LMI_VCPU_NC0_64BIT_BAR_LOW),
+		lower_32_bits(adev->vcn.inst[inst_idx].fw_shared.gpu_addr), 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH),
+		upper_32_bits(adev->vcn.inst[inst_idx].fw_shared.gpu_addr), 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_VCPU_NONCACHE_OFFSET0), 0, 0, indirect);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_VCPU_NONCACHE_SIZE0),
+		AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn4_fw_shared)), 0, indirect);
+
+	/* VCN global tiling registers */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_GFX10_ADDR_CONFIG),
+		adev->gfx.config.gb_addr_config, 0, indirect);
+}
+
+/**
+ * vcn_v4_0_5_disable_static_power_gating - disable VCN static power gating
+ *
+ * @vinst: VCN instance
+ *
+ * Disable static power gating for VCN block
+ */
+static void vcn_v4_0_5_disable_static_power_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst = vinst->inst;
+	uint32_t data = 0;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN) {
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG,
+					1 << UVD_IPX_DLDO_CONFIG__ONO2_PWR_CONFIG__SHIFT);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS, 0,
+					UVD_IPX_DLDO_STATUS__ONO2_PWR_STATUS_MASK);
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG,
+					2 << UVD_IPX_DLDO_CONFIG__ONO3_PWR_CONFIG__SHIFT);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+					1 << UVD_IPX_DLDO_STATUS__ONO3_PWR_STATUS__SHIFT,
+					UVD_IPX_DLDO_STATUS__ONO3_PWR_STATUS_MASK);
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG,
+					2 << UVD_IPX_DLDO_CONFIG__ONO4_PWR_CONFIG__SHIFT);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+					1 << UVD_IPX_DLDO_STATUS__ONO4_PWR_STATUS__SHIFT,
+					UVD_IPX_DLDO_STATUS__ONO4_PWR_STATUS_MASK);
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG,
+					2 << UVD_IPX_DLDO_CONFIG__ONO5_PWR_CONFIG__SHIFT);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+					1 << UVD_IPX_DLDO_STATUS__ONO5_PWR_STATUS__SHIFT,
+					UVD_IPX_DLDO_STATUS__ONO5_PWR_STATUS_MASK);
+	} else {
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG,
+			1 << UVD_IPX_DLDO_CONFIG__ONO2_PWR_CONFIG__SHIFT);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+			0, UVD_IPX_DLDO_STATUS__ONO2_PWR_STATUS_MASK);
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG,
+			1 << UVD_IPX_DLDO_CONFIG__ONO3_PWR_CONFIG__SHIFT);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+			0, UVD_IPX_DLDO_STATUS__ONO3_PWR_STATUS_MASK);
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG,
+			1 << UVD_IPX_DLDO_CONFIG__ONO4_PWR_CONFIG__SHIFT);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+			0, UVD_IPX_DLDO_STATUS__ONO4_PWR_STATUS_MASK);
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG,
+			1 << UVD_IPX_DLDO_CONFIG__ONO5_PWR_CONFIG__SHIFT);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+			0, UVD_IPX_DLDO_STATUS__ONO5_PWR_STATUS_MASK);
+	}
+
+	data = RREG32_SOC15(VCN, inst, regUVD_POWER_STATUS);
+	data &= ~0x103;
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN)
+		data |= UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON |
+			UVD_POWER_STATUS__UVD_PG_EN_MASK;
+	WREG32_SOC15(VCN, inst, regUVD_POWER_STATUS, data);
+}
+
+/**
+ * vcn_v4_0_5_enable_static_power_gating - enable VCN static power gating
+ *
+ * @vinst: VCN instance
+ *
+ * Enable static power gating for VCN block
+ */
+static void vcn_v4_0_5_enable_static_power_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst = vinst->inst;
+	uint32_t data;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN) {
+		/* Before power off, this indicator has to be turned on */
+		data = RREG32_SOC15(VCN, inst, regUVD_POWER_STATUS);
+		data &= ~UVD_POWER_STATUS__UVD_POWER_STATUS_MASK;
+		data |= UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF;
+		WREG32_SOC15(VCN, inst, regUVD_POWER_STATUS, data);
+
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG,
+			2 << UVD_IPX_DLDO_CONFIG__ONO5_PWR_CONFIG__SHIFT);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+			1 << UVD_IPX_DLDO_STATUS__ONO5_PWR_STATUS__SHIFT,
+			UVD_IPX_DLDO_STATUS__ONO5_PWR_STATUS_MASK);
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG,
+			2 << UVD_IPX_DLDO_CONFIG__ONO4_PWR_CONFIG__SHIFT);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+			1 << UVD_IPX_DLDO_STATUS__ONO4_PWR_STATUS__SHIFT,
+			UVD_IPX_DLDO_STATUS__ONO4_PWR_STATUS_MASK);
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG,
+			2 << UVD_IPX_DLDO_CONFIG__ONO3_PWR_CONFIG__SHIFT);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+			1 << UVD_IPX_DLDO_STATUS__ONO3_PWR_STATUS__SHIFT,
+			UVD_IPX_DLDO_STATUS__ONO3_PWR_STATUS_MASK);
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG,
+			2 << UVD_IPX_DLDO_CONFIG__ONO2_PWR_CONFIG__SHIFT);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+			1 << UVD_IPX_DLDO_STATUS__ONO2_PWR_STATUS__SHIFT,
+			UVD_IPX_DLDO_STATUS__ONO2_PWR_STATUS_MASK);
+	}
+}
+
+/**
+ * vcn_v4_0_5_disable_clock_gating - disable VCN clock gating
+ *
+ * @vinst: VCN instance
+ *
+ * Disable clock gating for VCN block
+ */
+static void vcn_v4_0_5_disable_clock_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst = vinst->inst;
+	uint32_t data;
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		return;
+
+	/* VCN disable CGC */
+	data = RREG32_SOC15(VCN, inst, regUVD_CGC_CTRL);
+	data &= ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK;
+	data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(VCN, inst, regUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, inst, regUVD_CGC_GATE);
+	data &= ~(UVD_CGC_GATE__SYS_MASK
+		| UVD_CGC_GATE__UDEC_MASK
+		| UVD_CGC_GATE__MPEG2_MASK
+		| UVD_CGC_GATE__REGS_MASK
+		| UVD_CGC_GATE__RBC_MASK
+		| UVD_CGC_GATE__LMI_MC_MASK
+		| UVD_CGC_GATE__LMI_UMC_MASK
+		| UVD_CGC_GATE__IDCT_MASK
+		| UVD_CGC_GATE__MPRD_MASK
+		| UVD_CGC_GATE__MPC_MASK
+		| UVD_CGC_GATE__LBSI_MASK
+		| UVD_CGC_GATE__LRBBM_MASK
+		| UVD_CGC_GATE__UDEC_RE_MASK
+		| UVD_CGC_GATE__UDEC_CM_MASK
+		| UVD_CGC_GATE__UDEC_IT_MASK
+		| UVD_CGC_GATE__UDEC_DB_MASK
+		| UVD_CGC_GATE__UDEC_MP_MASK
+		| UVD_CGC_GATE__WCB_MASK
+		| UVD_CGC_GATE__VCPU_MASK
+		| UVD_CGC_GATE__MMSCH_MASK);
+
+	WREG32_SOC15(VCN, inst, regUVD_CGC_GATE, data);
+	SOC15_WAIT_ON_RREG(VCN, inst, regUVD_CGC_GATE, 0,  0xFFFFFFFF);
+
+	data = RREG32_SOC15(VCN, inst, regUVD_CGC_CTRL);
+	data &= ~(UVD_CGC_CTRL__UDEC_RE_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_CM_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_IT_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_DB_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_MP_MODE_MASK
+		| UVD_CGC_CTRL__SYS_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_MODE_MASK
+		| UVD_CGC_CTRL__MPEG2_MODE_MASK
+		| UVD_CGC_CTRL__REGS_MODE_MASK
+		| UVD_CGC_CTRL__RBC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_MC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_UMC_MODE_MASK
+		| UVD_CGC_CTRL__IDCT_MODE_MASK
+		| UVD_CGC_CTRL__MPRD_MODE_MASK
+		| UVD_CGC_CTRL__MPC_MODE_MASK
+		| UVD_CGC_CTRL__LBSI_MODE_MASK
+		| UVD_CGC_CTRL__LRBBM_MODE_MASK
+		| UVD_CGC_CTRL__WCB_MODE_MASK
+		| UVD_CGC_CTRL__VCPU_MODE_MASK
+		| UVD_CGC_CTRL__MMSCH_MODE_MASK);
+	WREG32_SOC15(VCN, inst, regUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, inst, regUVD_SUVD_CGC_GATE);
+	data |= (UVD_SUVD_CGC_GATE__SRE_MASK
+		| UVD_SUVD_CGC_GATE__SIT_MASK
+		| UVD_SUVD_CGC_GATE__SMP_MASK
+		| UVD_SUVD_CGC_GATE__SCM_MASK
+		| UVD_SUVD_CGC_GATE__SDB_MASK
+		| UVD_SUVD_CGC_GATE__SRE_H264_MASK
+		| UVD_SUVD_CGC_GATE__SRE_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SIT_H264_MASK
+		| UVD_SUVD_CGC_GATE__SIT_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SCM_H264_MASK
+		| UVD_SUVD_CGC_GATE__SCM_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SDB_H264_MASK
+		| UVD_SUVD_CGC_GATE__SDB_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SCLR_MASK
+		| UVD_SUVD_CGC_GATE__UVD_SC_MASK
+		| UVD_SUVD_CGC_GATE__ENT_MASK
+		| UVD_SUVD_CGC_GATE__SIT_HEVC_DEC_MASK
+		| UVD_SUVD_CGC_GATE__SIT_HEVC_ENC_MASK
+		| UVD_SUVD_CGC_GATE__SITE_MASK
+		| UVD_SUVD_CGC_GATE__SRE_VP9_MASK
+		| UVD_SUVD_CGC_GATE__SCM_VP9_MASK
+		| UVD_SUVD_CGC_GATE__SIT_VP9_DEC_MASK
+		| UVD_SUVD_CGC_GATE__SDB_VP9_MASK
+		| UVD_SUVD_CGC_GATE__IME_HEVC_MASK);
+	WREG32_SOC15(VCN, inst, regUVD_SUVD_CGC_GATE, data);
+
+	data = RREG32_SOC15(VCN, inst, regUVD_SUVD_CGC_CTRL);
+	data &= ~(UVD_SUVD_CGC_CTRL__SRE_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SIT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SMP_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SCM_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SDB_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SCLR_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__UVD_SC_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__ENT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__IME_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SITE_MODE_MASK);
+	WREG32_SOC15(VCN, inst, regUVD_SUVD_CGC_CTRL, data);
+}
+
+/**
+ * vcn_v4_0_5_disable_clock_gating_dpg_mode - disable VCN clock gating dpg mode
+ *
+ * @vinst: VCN instance
+ * @sram_sel: sram select
+ * @indirect: indirectly write sram
+ *
+ * Disable clock gating for VCN block with dpg mode
+ */
+static void vcn_v4_0_5_disable_clock_gating_dpg_mode(struct amdgpu_vcn_inst *vinst,
+						     uint8_t sram_sel,
+						     uint8_t indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t reg_data = 0;
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		return;
+
+	/* enable sw clock gating control */
+	reg_data = 0 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	reg_data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	reg_data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	reg_data &= ~(UVD_CGC_CTRL__UDEC_RE_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_CM_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_IT_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_DB_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_MP_MODE_MASK |
+		 UVD_CGC_CTRL__SYS_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_MODE_MASK |
+		 UVD_CGC_CTRL__MPEG2_MODE_MASK |
+		 UVD_CGC_CTRL__REGS_MODE_MASK |
+		 UVD_CGC_CTRL__RBC_MODE_MASK |
+		 UVD_CGC_CTRL__LMI_MC_MODE_MASK |
+		 UVD_CGC_CTRL__LMI_UMC_MODE_MASK |
+		 UVD_CGC_CTRL__IDCT_MODE_MASK |
+		 UVD_CGC_CTRL__MPRD_MODE_MASK |
+		 UVD_CGC_CTRL__MPC_MODE_MASK |
+		 UVD_CGC_CTRL__LBSI_MODE_MASK |
+		 UVD_CGC_CTRL__LRBBM_MODE_MASK |
+		 UVD_CGC_CTRL__WCB_MODE_MASK |
+		 UVD_CGC_CTRL__VCPU_MODE_MASK);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_CGC_CTRL), reg_data, sram_sel, indirect);
+
+	/* turn off clock gating */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_CGC_GATE), 0, sram_sel, indirect);
+
+	/* turn on SUVD clock gating */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_SUVD_CGC_GATE), 1, sram_sel, indirect);
+
+	/* turn on sw mode in UVD_SUVD_CGC_CTRL */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_SUVD_CGC_CTRL), 0, sram_sel, indirect);
+}
+
+/**
+ * vcn_v4_0_5_enable_clock_gating - enable VCN clock gating
+ *
+ * @vinst: VCN instance
+ *
+ * Enable clock gating for VCN block
+ */
+static void vcn_v4_0_5_enable_clock_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst = vinst->inst;
+	uint32_t data;
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		return;
+
+	/* enable VCN CGC */
+	data = RREG32_SOC15(VCN, inst, regUVD_CGC_CTRL);
+	data |= 0 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(VCN, inst, regUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, inst, regUVD_CGC_CTRL);
+	data |= (UVD_CGC_CTRL__UDEC_RE_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_CM_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_IT_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_DB_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_MP_MODE_MASK
+		| UVD_CGC_CTRL__SYS_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_MODE_MASK
+		| UVD_CGC_CTRL__MPEG2_MODE_MASK
+		| UVD_CGC_CTRL__REGS_MODE_MASK
+		| UVD_CGC_CTRL__RBC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_MC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_UMC_MODE_MASK
+		| UVD_CGC_CTRL__IDCT_MODE_MASK
+		| UVD_CGC_CTRL__MPRD_MODE_MASK
+		| UVD_CGC_CTRL__MPC_MODE_MASK
+		| UVD_CGC_CTRL__LBSI_MODE_MASK
+		| UVD_CGC_CTRL__LRBBM_MODE_MASK
+		| UVD_CGC_CTRL__WCB_MODE_MASK
+		| UVD_CGC_CTRL__VCPU_MODE_MASK
+		| UVD_CGC_CTRL__MMSCH_MODE_MASK);
+	WREG32_SOC15(VCN, inst, regUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, inst, regUVD_SUVD_CGC_CTRL);
+	data |= (UVD_SUVD_CGC_CTRL__SRE_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SIT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SMP_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SCM_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SDB_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SCLR_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__UVD_SC_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__ENT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__IME_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SITE_MODE_MASK);
+	WREG32_SOC15(VCN, inst, regUVD_SUVD_CGC_CTRL, data);
+}
+
+/**
+ * vcn_v4_0_5_start_dpg_mode - VCN start with dpg mode
+ *
+ * @vinst: VCN instance
+ * @indirect: indirectly write sram
+ *
+ * Start VCN block with dpg mode
+ */
+static int vcn_v4_0_5_start_dpg_mode(struct amdgpu_vcn_inst *vinst,
+				     bool indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	struct amdgpu_vcn4_fw_shared *fw_shared = adev->vcn.inst[inst_idx].fw_shared.cpu_addr;
+	struct amdgpu_ring *ring;
+	uint32_t tmp;
+	int ret;
+
+	/* disable register anti-hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, regUVD_POWER_STATUS), 1,
+		~UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+	/* enable dynamic power gating mode */
+	tmp = RREG32_SOC15(VCN, inst_idx, regUVD_POWER_STATUS);
+	tmp |= UVD_POWER_STATUS__UVD_PG_MODE_MASK;
+	tmp |= UVD_POWER_STATUS__UVD_PG_EN_MASK;
+	WREG32_SOC15(VCN, inst_idx, regUVD_POWER_STATUS, tmp);
+
+	if (indirect)
+		adev->vcn.inst[inst_idx].dpg_sram_curr_addr =
+					(uint32_t *)adev->vcn.inst[inst_idx].dpg_sram_cpu_addr;
+
+	/* enable clock gating */
+	vcn_v4_0_5_disable_clock_gating_dpg_mode(vinst, 0, indirect);
+
+	/* enable VCPU clock */
+	tmp = (0xFF << UVD_VCPU_CNTL__PRB_TIMEOUT_VAL__SHIFT);
+	tmp |= UVD_VCPU_CNTL__CLK_EN_MASK | UVD_VCPU_CNTL__BLK_RST_MASK;
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_VCPU_CNTL), tmp, 0, indirect);
+
+	/* disable master interrupt */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_MASTINT_EN), 0, 0, indirect);
+
+	/* setup regUVD_LMI_CTRL */
+	tmp = (UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+		UVD_LMI_CTRL__REQ_MODE_MASK |
+		UVD_LMI_CTRL__CRC_RESET_MASK |
+		UVD_LMI_CTRL__MASK_MC_URGENT_MASK |
+		UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+		UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
+		(8 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
+		0x00100000L);
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_LMI_CTRL), tmp, 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_MPC_CNTL),
+		0x2 << UVD_MPC_CNTL__REPLACEMENT_MODE__SHIFT, 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_MPC_SET_MUXA0),
+		((0x1 << UVD_MPC_SET_MUXA0__VARA_1__SHIFT) |
+		 (0x2 << UVD_MPC_SET_MUXA0__VARA_2__SHIFT) |
+		 (0x3 << UVD_MPC_SET_MUXA0__VARA_3__SHIFT) |
+		 (0x4 << UVD_MPC_SET_MUXA0__VARA_4__SHIFT)), 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_MPC_SET_MUXB0),
+		 ((0x1 << UVD_MPC_SET_MUXB0__VARB_1__SHIFT) |
+		 (0x2 << UVD_MPC_SET_MUXB0__VARB_2__SHIFT) |
+		 (0x3 << UVD_MPC_SET_MUXB0__VARB_3__SHIFT) |
+		 (0x4 << UVD_MPC_SET_MUXB0__VARB_4__SHIFT)), 0, indirect);
+
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_MPC_SET_MUX),
+		((0x0 << UVD_MPC_SET_MUX__SET_0__SHIFT) |
+		 (0x1 << UVD_MPC_SET_MUX__SET_1__SHIFT) |
+		 (0x2 << UVD_MPC_SET_MUX__SET_2__SHIFT)), 0, indirect);
+
+	vcn_v4_0_5_mc_resume_dpg_mode(vinst, indirect);
+
+	tmp = (0xFF << UVD_VCPU_CNTL__PRB_TIMEOUT_VAL__SHIFT);
+	tmp |= UVD_VCPU_CNTL__CLK_EN_MASK;
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_VCPU_CNTL), tmp, 0, indirect);
+
+	/* enable LMI MC and UMC channels */
+	tmp = 0x1f << UVD_LMI_CTRL2__RE_OFLD_MIF_WR_REQ_NUM__SHIFT;
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_LMI_CTRL2), tmp, 0, indirect);
+
+	/* enable master interrupt */
+	WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_MASTINT_EN),
+		UVD_MASTINT_EN__VCPU_EN_MASK, 0, indirect);
+
+	if (indirect) {
+		ret = amdgpu_vcn_psp_update_sram(adev, inst_idx, 0);
+		if (ret) {
+			dev_err(adev->dev, "vcn sram load failed %d\n", ret);
+			return ret;
+		}
+	}
+
+	ring = &adev->vcn.inst[inst_idx].ring_enc[0];
+
+	WREG32_SOC15(VCN, inst_idx, regUVD_RB_BASE_LO, ring->gpu_addr);
+	WREG32_SOC15(VCN, inst_idx, regUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(VCN, inst_idx, regUVD_RB_SIZE, ring->ring_size / 4);
+
+	tmp = RREG32_SOC15(VCN, inst_idx, regVCN_RB_ENABLE);
+	tmp &= ~(VCN_RB_ENABLE__RB1_EN_MASK);
+	WREG32_SOC15(VCN, inst_idx, regVCN_RB_ENABLE, tmp);
+	fw_shared->sq.queue_mode |= FW_QUEUE_RING_RESET;
+	WREG32_SOC15(VCN, inst_idx, regUVD_RB_RPTR, 0);
+	WREG32_SOC15(VCN, inst_idx, regUVD_RB_WPTR, 0);
+
+	tmp = RREG32_SOC15(VCN, inst_idx, regUVD_RB_RPTR);
+	WREG32_SOC15(VCN, inst_idx, regUVD_RB_WPTR, tmp);
+	ring->wptr = RREG32_SOC15(VCN, inst_idx, regUVD_RB_WPTR);
+
+	tmp = RREG32_SOC15(VCN, inst_idx, regVCN_RB_ENABLE);
+	tmp |= VCN_RB_ENABLE__RB1_EN_MASK;
+	WREG32_SOC15(VCN, inst_idx, regVCN_RB_ENABLE, tmp);
+	fw_shared->sq.queue_mode &= ~(FW_QUEUE_RING_RESET | FW_QUEUE_DPG_HOLD_OFF);
+
+	WREG32_SOC15(VCN, inst_idx, regVCN_RB1_DB_CTRL,
+			ring->doorbell_index << VCN_RB1_DB_CTRL__OFFSET__SHIFT |
+			VCN_RB1_DB_CTRL__EN_MASK);
+
+	/* Keeping one read-back to ensure all register writes are done, otherwise
+	 * it may introduce race conditions */
+	RREG32_SOC15(VCN, inst_idx, regVCN_RB1_DB_CTRL);
+
+	return 0;
+}
+
+
+/**
+ * vcn_v4_0_5_start - VCN start
+ *
+ * @vinst: VCN instance
+ *
+ * Start VCN block
+ */
+static int vcn_v4_0_5_start(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int i = vinst->inst;
+	struct amdgpu_vcn4_fw_shared *fw_shared;
+	struct amdgpu_ring *ring;
+	uint32_t tmp;
+	int j, k, r;
+
+	if (adev->vcn.harvest_config & (1 << i))
+		return 0;
+
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_vcn(adev, true, i);
+
+	fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)
+		return vcn_v4_0_5_start_dpg_mode(vinst, adev->vcn.inst[i].indirect_sram);
+
+	/* disable VCN power gating */
+	vcn_v4_0_5_disable_static_power_gating(vinst);
+
+	/* set VCN status busy */
+	tmp = RREG32_SOC15(VCN, i, regUVD_STATUS) | UVD_STATUS__UVD_BUSY;
+	WREG32_SOC15(VCN, i, regUVD_STATUS, tmp);
+
+	/* SW clock gating */
+	vcn_v4_0_5_disable_clock_gating(vinst);
+
+	/* enable VCPU clock */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL),
+		 UVD_VCPU_CNTL__CLK_EN_MASK, ~UVD_VCPU_CNTL__CLK_EN_MASK);
+
+	/* disable master interrupt */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_MASTINT_EN), 0,
+		 ~UVD_MASTINT_EN__VCPU_EN_MASK);
+
+	/* enable LMI MC and UMC channels */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_LMI_CTRL2), 0,
+		 ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+
+	tmp = RREG32_SOC15(VCN, i, regUVD_SOFT_RESET);
+	tmp &= ~UVD_SOFT_RESET__LMI_SOFT_RESET_MASK;
+	tmp &= ~UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK;
+	WREG32_SOC15(VCN, i, regUVD_SOFT_RESET, tmp);
+
+	/* setup regUVD_LMI_CTRL */
+	tmp = RREG32_SOC15(VCN, i, regUVD_LMI_CTRL);
+	WREG32_SOC15(VCN, i, regUVD_LMI_CTRL, tmp |
+		     UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+		     UVD_LMI_CTRL__MASK_MC_URGENT_MASK |
+		     UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+		     UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK);
+
+	/* setup regUVD_MPC_CNTL */
+	tmp = RREG32_SOC15(VCN, i, regUVD_MPC_CNTL);
+	tmp &= ~UVD_MPC_CNTL__REPLACEMENT_MODE_MASK;
+	tmp |= 0x2 << UVD_MPC_CNTL__REPLACEMENT_MODE__SHIFT;
+	WREG32_SOC15(VCN, i, regUVD_MPC_CNTL, tmp);
+
+	/* setup UVD_MPC_SET_MUXA0 */
+	WREG32_SOC15(VCN, i, regUVD_MPC_SET_MUXA0,
+		     ((0x1 << UVD_MPC_SET_MUXA0__VARA_1__SHIFT) |
+		      (0x2 << UVD_MPC_SET_MUXA0__VARA_2__SHIFT) |
+		      (0x3 << UVD_MPC_SET_MUXA0__VARA_3__SHIFT) |
+		      (0x4 << UVD_MPC_SET_MUXA0__VARA_4__SHIFT)));
+
+	/* setup UVD_MPC_SET_MUXB0 */
+	WREG32_SOC15(VCN, i, regUVD_MPC_SET_MUXB0,
+		     ((0x1 << UVD_MPC_SET_MUXB0__VARB_1__SHIFT) |
+		      (0x2 << UVD_MPC_SET_MUXB0__VARB_2__SHIFT) |
+		      (0x3 << UVD_MPC_SET_MUXB0__VARB_3__SHIFT) |
+		      (0x4 << UVD_MPC_SET_MUXB0__VARB_4__SHIFT)));
+
+	/* setup UVD_MPC_SET_MUX */
+	WREG32_SOC15(VCN, i, regUVD_MPC_SET_MUX,
+		     ((0x0 << UVD_MPC_SET_MUX__SET_0__SHIFT) |
+		      (0x1 << UVD_MPC_SET_MUX__SET_1__SHIFT) |
+		      (0x2 << UVD_MPC_SET_MUX__SET_2__SHIFT)));
+
+	vcn_v4_0_5_mc_resume(vinst);
+
+	/* VCN global tiling registers */
+	WREG32_SOC15(VCN, i, regUVD_GFX10_ADDR_CONFIG,
+		     adev->gfx.config.gb_addr_config);
+
+	/* unblock VCPU register access */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_RB_ARB_CTRL), 0,
+		 ~UVD_RB_ARB_CTRL__VCPU_DIS_MASK);
+
+	/* release VCPU reset to boot */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), 0,
+		 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+	for (j = 0; j < 10; ++j) {
+		uint32_t status;
+
+		for (k = 0; k < 100; ++k) {
+			status = RREG32_SOC15(VCN, i, regUVD_STATUS);
+			if (status & 2)
+				break;
+			mdelay(10);
+			if (amdgpu_emu_mode == 1)
+				msleep(1);
+		}
+
+		if (amdgpu_emu_mode == 1) {
+			r = -1;
+			if (status & 2) {
+				r = 0;
+				break;
+			}
+		} else {
+			r = 0;
+			if (status & 2)
+				break;
+
+			dev_err(adev->dev,
+				"VCN[%d] is not responding, trying to reset VCPU!!!\n", i);
+			WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL),
+				 UVD_VCPU_CNTL__BLK_RST_MASK,
+				 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+			mdelay(10);
+			WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), 0,
+				 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+			mdelay(10);
+			r = -1;
+		}
+	}
+
+	if (r) {
+		dev_err(adev->dev, "VCN[%d] is not responding, giving up!!!\n", i);
+		return r;
+	}
+
+	/* enable master interrupt */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_MASTINT_EN),
+		 UVD_MASTINT_EN__VCPU_EN_MASK,
+		 ~UVD_MASTINT_EN__VCPU_EN_MASK);
+
+	/* clear the busy bit of VCN_STATUS */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_STATUS), 0,
+		 ~(2 << UVD_STATUS__VCPU_REPORT__SHIFT));
+
+	ring = &adev->vcn.inst[i].ring_enc[0];
+	WREG32_SOC15(VCN, i, regVCN_RB1_DB_CTRL,
+		     ring->doorbell_index << VCN_RB1_DB_CTRL__OFFSET__SHIFT |
+		     VCN_RB1_DB_CTRL__EN_MASK);
+
+	WREG32_SOC15(VCN, i, regUVD_RB_BASE_LO, ring->gpu_addr);
+	WREG32_SOC15(VCN, i, regUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(VCN, i, regUVD_RB_SIZE, ring->ring_size / 4);
+
+	tmp = RREG32_SOC15(VCN, i, regVCN_RB_ENABLE);
+	tmp &= ~(VCN_RB_ENABLE__RB1_EN_MASK);
+	WREG32_SOC15(VCN, i, regVCN_RB_ENABLE, tmp);
+	fw_shared->sq.queue_mode |= FW_QUEUE_RING_RESET;
+	WREG32_SOC15(VCN, i, regUVD_RB_RPTR, 0);
+	WREG32_SOC15(VCN, i, regUVD_RB_WPTR, 0);
+
+	tmp = RREG32_SOC15(VCN, i, regUVD_RB_RPTR);
+	WREG32_SOC15(VCN, i, regUVD_RB_WPTR, tmp);
+	ring->wptr = RREG32_SOC15(VCN, i, regUVD_RB_WPTR);
+
+	tmp = RREG32_SOC15(VCN, i, regVCN_RB_ENABLE);
+	tmp |= VCN_RB_ENABLE__RB1_EN_MASK;
+	WREG32_SOC15(VCN, i, regVCN_RB_ENABLE, tmp);
+	fw_shared->sq.queue_mode &= ~(FW_QUEUE_RING_RESET | FW_QUEUE_DPG_HOLD_OFF);
+
+	/* Keeping one read-back to ensure all register writes are done, otherwise
+	 * it may introduce race conditions */
+	RREG32_SOC15(VCN, i, regVCN_RB_ENABLE);
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_5_stop_dpg_mode - VCN stop with dpg mode
+ *
+ * @vinst: VCN instance
+ *
+ * Stop VCN block with dpg mode
+ */
+static void vcn_v4_0_5_stop_dpg_mode(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t tmp;
+
+	/* Wait for power status to be 1 */
+	SOC15_WAIT_ON_RREG(VCN, inst_idx, regUVD_POWER_STATUS, 1,
+		UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+	/* wait for read ptr to be equal to write ptr */
+	tmp = RREG32_SOC15(VCN, inst_idx, regUVD_RB_WPTR);
+	SOC15_WAIT_ON_RREG(VCN, inst_idx, regUVD_RB_RPTR, tmp, 0xFFFFFFFF);
+
+	SOC15_WAIT_ON_RREG(VCN, inst_idx, regUVD_POWER_STATUS, 1,
+		UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+	/* disable dynamic power gating mode */
+	WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, regUVD_POWER_STATUS), 0,
+		~UVD_POWER_STATUS__UVD_PG_MODE_MASK);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, inst_idx, regUVD_STATUS);
+}
+
+/**
+ * vcn_v4_0_5_stop - VCN stop
+ *
+ * @vinst: VCN instance
+ *
+ * Stop VCN block
+ */
+static int vcn_v4_0_5_stop(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int i = vinst->inst;
+	struct amdgpu_vcn4_fw_shared *fw_shared;
+	uint32_t tmp;
+	int r = 0;
+
+	if (adev->vcn.harvest_config & (1 << i))
+		return 0;
+
+	fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+	fw_shared->sq.queue_mode |= FW_QUEUE_DPG_HOLD_OFF;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) {
+		vcn_v4_0_5_stop_dpg_mode(vinst);
+		r = 0;
+		goto done;
+	}
+
+	/* wait for vcn idle */
+	r = SOC15_WAIT_ON_RREG(VCN, i, regUVD_STATUS, UVD_STATUS__IDLE, 0x7);
+	if (r)
+		goto done;
+
+	tmp = UVD_LMI_STATUS__VCPU_LMI_WRITE_CLEAN_MASK |
+		UVD_LMI_STATUS__READ_CLEAN_MASK |
+		UVD_LMI_STATUS__WRITE_CLEAN_MASK |
+		UVD_LMI_STATUS__WRITE_CLEAN_RAW_MASK;
+	r = SOC15_WAIT_ON_RREG(VCN, i, regUVD_LMI_STATUS, tmp, tmp);
+	if (r)
+		goto done;
+
+	/* disable LMI UMC channel */
+	tmp = RREG32_SOC15(VCN, i, regUVD_LMI_CTRL2);
+	tmp |= UVD_LMI_CTRL2__STALL_ARB_UMC_MASK;
+	WREG32_SOC15(VCN, i, regUVD_LMI_CTRL2, tmp);
+	tmp = UVD_LMI_STATUS__UMC_READ_CLEAN_RAW_MASK |
+		UVD_LMI_STATUS__UMC_WRITE_CLEAN_RAW_MASK;
+	r = SOC15_WAIT_ON_RREG(VCN, i, regUVD_LMI_STATUS, tmp, tmp);
+	if (r)
+		goto done;
+
+	/* block VCPU register access */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_RB_ARB_CTRL),
+		 UVD_RB_ARB_CTRL__VCPU_DIS_MASK,
+		 ~UVD_RB_ARB_CTRL__VCPU_DIS_MASK);
+
+	/* reset VCPU */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL),
+		 UVD_VCPU_CNTL__BLK_RST_MASK,
+		 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+	/* disable VCPU clock */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), 0,
+		 ~(UVD_VCPU_CNTL__CLK_EN_MASK));
+
+	/* apply soft reset */
+	tmp = RREG32_SOC15(VCN, i, regUVD_SOFT_RESET);
+	tmp |= UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK;
+	WREG32_SOC15(VCN, i, regUVD_SOFT_RESET, tmp);
+	tmp = RREG32_SOC15(VCN, i, regUVD_SOFT_RESET);
+	tmp |= UVD_SOFT_RESET__LMI_SOFT_RESET_MASK;
+	WREG32_SOC15(VCN, i, regUVD_SOFT_RESET, tmp);
+
+	/* clear status */
+	WREG32_SOC15(VCN, i, regUVD_STATUS, 0);
+
+	/* apply HW clock gating */
+	vcn_v4_0_5_enable_clock_gating(vinst);
+
+	/* enable VCN power gating */
+	vcn_v4_0_5_enable_static_power_gating(vinst);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, i, regUVD_STATUS);
+
+done:
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_vcn(adev, false, i);
+
+	return r;
+}
+
+/**
+ * vcn_v4_0_5_pause_dpg_mode - VCN pause with dpg mode
+ *
+ * @vinst: VCN instance
+ * @new_state: pause state
+ *
+ * Pause dpg mode for VCN block
+ */
+static int vcn_v4_0_5_pause_dpg_mode(struct amdgpu_vcn_inst *vinst,
+				     struct dpg_pause_state *new_state)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t reg_data = 0;
+	int ret_code;
+
+	/* pause/unpause if state is changed */
+	if (adev->vcn.inst[inst_idx].pause_state.fw_based != new_state->fw_based) {
+		DRM_DEV_DEBUG(adev->dev, "dpg pause state changed %d -> %d",
+			adev->vcn.inst[inst_idx].pause_state.fw_based,	new_state->fw_based);
+		reg_data = RREG32_SOC15(VCN, inst_idx, regUVD_DPG_PAUSE) &
+			(~UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK);
+
+		if (new_state->fw_based == VCN_DPG_STATE__PAUSE) {
+			ret_code = SOC15_WAIT_ON_RREG(VCN, inst_idx, regUVD_POWER_STATUS, 0x1,
+				UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+			if (!ret_code) {
+				/* pause DPG */
+				reg_data |= UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
+				WREG32_SOC15(VCN, inst_idx, regUVD_DPG_PAUSE, reg_data);
+
+				/* wait for ACK */
+				SOC15_WAIT_ON_RREG(VCN, inst_idx, regUVD_DPG_PAUSE,
+					UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK,
+					UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK);
+
+				SOC15_WAIT_ON_RREG(VCN, inst_idx, regUVD_POWER_STATUS,
+					UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON,
+					UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+			}
+		} else {
+			/* unpause dpg, no need to wait */
+			reg_data &= ~UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
+			WREG32_SOC15(VCN, inst_idx, regUVD_DPG_PAUSE, reg_data);
+		}
+		adev->vcn.inst[inst_idx].pause_state.fw_based = new_state->fw_based;
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v4_0_5_unified_ring_get_rptr - get unified read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware unified read pointer
+ */
+static uint64_t vcn_v4_0_5_unified_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring != &adev->vcn.inst[ring->me].ring_enc[0])
+		DRM_ERROR("wrong ring id is identified in %s", __func__);
+
+	return RREG32_SOC15(VCN, ring->me, regUVD_RB_RPTR);
+}
+
+/**
+ * vcn_v4_0_5_unified_ring_get_wptr - get unified write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware unified write pointer
+ */
+static uint64_t vcn_v4_0_5_unified_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring != &adev->vcn.inst[ring->me].ring_enc[0])
+		DRM_ERROR("wrong ring id is identified in %s", __func__);
+
+	if (ring->use_doorbell)
+		return *ring->wptr_cpu_addr;
+	else
+		return RREG32_SOC15(VCN, ring->me, regUVD_RB_WPTR);
+}
+
+/**
+ * vcn_v4_0_5_unified_ring_set_wptr - set enc write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the enc write pointer to the hardware
+ */
+static void vcn_v4_0_5_unified_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring != &adev->vcn.inst[ring->me].ring_enc[0])
+		DRM_ERROR("wrong ring id is identified in %s", __func__);
+
+	if (ring->use_doorbell) {
+		*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
+		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+	} else {
+		WREG32_SOC15(VCN, ring->me, regUVD_RB_WPTR, lower_32_bits(ring->wptr));
+	}
+}
+
+static int vcn_v4_0_5_ring_reset(struct amdgpu_ring *ring,
+				 unsigned int vmid,
+				 struct amdgpu_fence *timedout_fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[ring->me];
+	int r;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+	r = vcn_v4_0_5_stop(vinst);
+	if (r)
+		return r;
+	r = vcn_v4_0_5_start(vinst);
+	if (r)
+		return r;
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static struct amdgpu_ring_funcs vcn_v4_0_5_unified_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_ENC,
+	.align_mask = 0x3f,
+	.nop = VCN_ENC_CMD_NO_OP,
+	.get_rptr = vcn_v4_0_5_unified_ring_get_rptr,
+	.get_wptr = vcn_v4_0_5_unified_ring_get_wptr,
+	.set_wptr = vcn_v4_0_5_unified_ring_set_wptr,
+	.emit_frame_size =
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 4 +
+		4 + /* vcn_v2_0_enc_ring_emit_vm_flush */
+		5 + 5 + /* vcn_v2_0_enc_ring_emit_fence x2 vm fence */
+		1, /* vcn_v2_0_enc_ring_insert_end */
+	.emit_ib_size = 5, /* vcn_v2_0_enc_ring_emit_ib */
+	.emit_ib = vcn_v2_0_enc_ring_emit_ib,
+	.emit_fence = vcn_v2_0_enc_ring_emit_fence,
+	.emit_vm_flush = vcn_v2_0_enc_ring_emit_vm_flush,
+	.test_ring = amdgpu_vcn_enc_ring_test_ring,
+	.test_ib = amdgpu_vcn_unified_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.insert_end = vcn_v2_0_enc_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_vcn_ring_begin_use,
+	.end_use = amdgpu_vcn_ring_end_use,
+	.emit_wreg = vcn_v2_0_enc_ring_emit_wreg,
+	.emit_reg_wait = vcn_v2_0_enc_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = vcn_v4_0_5_ring_reset,
+};
+
+/**
+ * vcn_v4_0_5_set_unified_ring_funcs - set unified ring functions
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set unified ring functions
+ */
+static void vcn_v4_0_5_set_unified_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		if (amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 5))
+			vcn_v4_0_5_unified_ring_vm_funcs.secure_submission_supported = true;
+
+		adev->vcn.inst[i].ring_enc[0].funcs = &vcn_v4_0_5_unified_ring_vm_funcs;
+		adev->vcn.inst[i].ring_enc[0].me = i;
+	}
+}
+
+/**
+ * vcn_v4_0_5_is_idle - check VCN block is idle
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block structure
+ *
+ * Check whether VCN block is idle
+ */
+static bool vcn_v4_0_5_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, ret = 1;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		ret &= (RREG32_SOC15(VCN, i, regUVD_STATUS) == UVD_STATUS__IDLE);
+	}
+
+	return ret;
+}
+
+/**
+ * vcn_v4_0_5_wait_for_idle - wait for VCN block idle
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Wait for VCN block idle
+ */
+static int vcn_v4_0_5_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, ret = 0;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		ret = SOC15_WAIT_ON_RREG(VCN, i, regUVD_STATUS, UVD_STATUS__IDLE,
+			UVD_STATUS__IDLE);
+		if (ret)
+			return ret;
+	}
+
+	return ret;
+}
+
+/**
+ * vcn_v4_0_5_set_clockgating_state - set VCN block clockgating state
+ *
+ * @ip_block: amdgpu_ip_block pointer
+ * @state: clock gating state
+ *
+ * Set VCN block clockgating state
+ */
+static int vcn_v4_0_5_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = state == AMD_CG_STATE_GATE;
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[i];
+
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		if (enable) {
+			if (RREG32_SOC15(VCN, i, regUVD_STATUS) != UVD_STATUS__IDLE)
+				return -EBUSY;
+			vcn_v4_0_5_enable_clock_gating(vinst);
+		} else {
+			vcn_v4_0_5_disable_clock_gating(vinst);
+		}
+	}
+
+	return 0;
+}
+
+static int vcn_v4_0_5_set_pg_state(struct amdgpu_vcn_inst *vinst,
+				   enum amd_powergating_state state)
+{
+	int ret = 0;
+
+	if (state == vinst->cur_state)
+		return 0;
+
+	if (state == AMD_PG_STATE_GATE)
+		ret = vcn_v4_0_5_stop(vinst);
+	else
+		ret = vcn_v4_0_5_start(vinst);
+
+	if (!ret)
+		vinst->cur_state = state;
+
+	return ret;
+}
+
+/**
+ * vcn_v4_0_5_process_interrupt - process VCN block interrupt
+ *
+ * @adev: amdgpu_device pointer
+ * @source: interrupt sources
+ * @entry: interrupt entry from clients and sources
+ *
+ * Process VCN block interrupt
+ */
+static int vcn_v4_0_5_process_interrupt(struct amdgpu_device *adev, struct amdgpu_irq_src *source,
+		struct amdgpu_iv_entry *entry)
+{
+	uint32_t ip_instance;
+
+	switch (entry->client_id) {
+	case SOC15_IH_CLIENTID_VCN:
+		ip_instance = 0;
+		break;
+	case SOC15_IH_CLIENTID_VCN1:
+		ip_instance = 1;
+		break;
+	default:
+		DRM_ERROR("Unhandled client id: %d\n", entry->client_id);
+		return 0;
+	}
+
+	DRM_DEBUG("IH: VCN TRAP\n");
+
+	switch (entry->src_id) {
+	case VCN_4_0__SRCID__UVD_ENC_GENERAL_PURPOSE:
+		amdgpu_fence_process(&adev->vcn.inst[ip_instance].ring_enc[0]);
+		break;
+	case VCN_4_0__SRCID_UVD_POISON:
+		amdgpu_vcn_process_poison_irq(adev, source, entry);
+		break;
+	default:
+		DRM_ERROR("Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs vcn_v4_0_5_irq_funcs = {
+	.process = vcn_v4_0_5_process_interrupt,
+};
+
+/**
+ * vcn_v4_0_5_set_irq_funcs - set VCN block interrupt irq functions
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set VCN block interrupt irq functions
+ */
+static void vcn_v4_0_5_set_irq_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		adev->vcn.inst[i].irq.num_types = adev->vcn.inst[i].num_enc_rings + 1;
+		adev->vcn.inst[i].irq.funcs = &vcn_v4_0_5_irq_funcs;
+	}
+}
+
+static const struct amd_ip_funcs vcn_v4_0_5_ip_funcs = {
+	.name = "vcn_v4_0_5",
+	.early_init = vcn_v4_0_5_early_init,
+	.sw_init = vcn_v4_0_5_sw_init,
+	.sw_fini = vcn_v4_0_5_sw_fini,
+	.hw_init = vcn_v4_0_5_hw_init,
+	.hw_fini = vcn_v4_0_5_hw_fini,
+	.suspend = vcn_v4_0_5_suspend,
+	.resume = vcn_v4_0_5_resume,
+	.is_idle = vcn_v4_0_5_is_idle,
+	.wait_for_idle = vcn_v4_0_5_wait_for_idle,
+	.set_clockgating_state = vcn_v4_0_5_set_clockgating_state,
+	.set_powergating_state = vcn_set_powergating_state,
+	.dump_ip_state = amdgpu_vcn_dump_ip_state,
+	.print_ip_state = amdgpu_vcn_print_ip_state,
+};
+
+const struct amdgpu_ip_block_version vcn_v4_0_5_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_VCN,
+	.major = 4,
+	.minor = 0,
+	.rev = 5,
+	.funcs = &vcn_v4_0_5_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.h b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.h
new file mode 100644
index 000000000000..ff9b3d6f6a47
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.h
@@ -0,0 +1,35 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VCN_V4_0_5_H__
+#define __VCN_V4_0_5_H__
+
+enum amdgpu_vcn_v4_0_5_sub_block {
+	AMDGPU_VCN_V4_0_5_VCPU_VCODEC = 0,
+
+	AMDGPU_VCN_V4_0_5_MAX_SUB_BLOCK,
+};
+
+extern const struct amdgpu_ip_block_version vcn_v4_0_5_ip_block;
+
+#endif /* __VCN_V4_0_5_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c
new file mode 100644
index 000000000000..0202df5db1e1
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c
@@ -0,0 +1,1442 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include "amdgpu.h"
+#include "amdgpu_vcn.h"
+#include "amdgpu_pm.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "soc15_hw_ip.h"
+#include "vcn_v2_0.h"
+
+#include "vcn/vcn_5_0_0_offset.h"
+#include "vcn/vcn_5_0_0_sh_mask.h"
+#include "ivsrcid/vcn/irqsrcs_vcn_5_0.h"
+#include "vcn_v5_0_0.h"
+
+#include <drm/drm_drv.h>
+
+static const struct amdgpu_hwip_reg_entry vcn_reg_list_5_0[] = {
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_CONTEXT_ID),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_CONTEXT_ID2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_GPCOM_VCPU_DATA0),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_GPCOM_VCPU_DATA1),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_GPCOM_VCPU_CMD),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_HI),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_LO),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_HI2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_LO2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_HI3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_LO3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_HI4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_LO4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_RPTR),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_WPTR),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_RPTR2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_WPTR2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_RPTR3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_WPTR3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_RPTR4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_WPTR4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_SIZE),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_SIZE2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_SIZE3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_SIZE4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_DPG_LMA_CTL),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_DPG_LMA_DATA),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_DPG_LMA_MASK),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_DPG_PAUSE)
+};
+
+static int amdgpu_ih_clientid_vcns[] = {
+	SOC15_IH_CLIENTID_VCN,
+	SOC15_IH_CLIENTID_VCN1
+};
+
+static void vcn_v5_0_0_set_unified_ring_funcs(struct amdgpu_device *adev);
+static void vcn_v5_0_0_set_irq_funcs(struct amdgpu_device *adev);
+static int vcn_v5_0_0_set_pg_state(struct amdgpu_vcn_inst *vinst,
+				   enum amd_powergating_state state);
+static int vcn_v5_0_0_pause_dpg_mode(struct amdgpu_vcn_inst *vinst,
+				     struct dpg_pause_state *new_state);
+static void vcn_v5_0_0_unified_ring_set_wptr(struct amdgpu_ring *ring);
+
+/**
+ * vcn_v5_0_0_early_init - set function pointers and load microcode
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Set ring and irq function pointers
+ * Load microcode from filesystem
+ */
+static int vcn_v5_0_0_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, r;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i)
+		/* re-use enc ring as unified ring */
+		adev->vcn.inst[i].num_enc_rings = 1;
+
+	vcn_v5_0_0_set_unified_ring_funcs(adev);
+	vcn_v5_0_0_set_irq_funcs(adev);
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		adev->vcn.inst[i].set_pg_state = vcn_v5_0_0_set_pg_state;
+
+		r = amdgpu_vcn_early_init(adev, i);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v5_0_0_sw_init - sw init for VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Load firmware and sw initialization
+ */
+static int vcn_v5_0_0_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_ring *ring;
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, r;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		struct amdgpu_vcn5_fw_shared *fw_shared;
+
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		r = amdgpu_vcn_sw_init(adev, i);
+		if (r)
+			return r;
+
+		amdgpu_vcn_setup_ucode(adev, i);
+
+		r = amdgpu_vcn_resume(adev, i);
+		if (r)
+			return r;
+
+		atomic_set(&adev->vcn.inst[i].sched_score, 0);
+
+		/* VCN UNIFIED TRAP */
+		r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_vcns[i],
+				VCN_5_0__SRCID__UVD_ENC_GENERAL_PURPOSE, &adev->vcn.inst[i].irq);
+		if (r)
+			return r;
+
+		/* VCN POISON TRAP */
+		r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_vcns[i],
+				VCN_5_0__SRCID_UVD_POISON, &adev->vcn.inst[i].irq);
+		if (r)
+			return r;
+
+		ring = &adev->vcn.inst[i].ring_enc[0];
+		ring->use_doorbell = true;
+		ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 2 + 8 * i;
+
+		ring->vm_hub = AMDGPU_MMHUB0(0);
+		sprintf(ring->name, "vcn_unified_%d", i);
+
+		r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst[i].irq, 0,
+						AMDGPU_RING_PRIO_0, &adev->vcn.inst[i].sched_score);
+		if (r)
+			return r;
+
+		fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+		fw_shared->present_flag_0 = cpu_to_le32(AMDGPU_FW_SHARED_FLAG_0_UNIFIED_QUEUE);
+		fw_shared->sq.is_enabled = 1;
+
+		if (amdgpu_vcnfw_log)
+			amdgpu_vcn_fwlog_init(&adev->vcn.inst[i]);
+
+		if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)
+			adev->vcn.inst[i].pause_dpg_mode = vcn_v5_0_0_pause_dpg_mode;
+	}
+
+	adev->vcn.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->vcn.inst[0].ring_enc[0]);
+	if (!amdgpu_sriov_vf(adev))
+		adev->vcn.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+
+	r = amdgpu_vcn_reg_dump_init(adev, vcn_reg_list_5_0, ARRAY_SIZE(vcn_reg_list_5_0));
+	if (r)
+		return r;
+
+	r = amdgpu_vcn_sysfs_reset_mask_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+/**
+ * vcn_v5_0_0_sw_fini - sw fini for VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * VCN suspend and free up sw allocation
+ */
+static int vcn_v5_0_0_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, r, idx;
+
+	if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+		for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+			struct amdgpu_vcn5_fw_shared *fw_shared;
+
+			if (adev->vcn.harvest_config & (1 << i))
+				continue;
+
+			fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+			fw_shared->present_flag_0 = 0;
+			fw_shared->sq.is_enabled = 0;
+		}
+
+		drm_dev_exit(idx);
+	}
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		r = amdgpu_vcn_suspend(adev, i);
+		if (r)
+			return r;
+	}
+
+	amdgpu_vcn_sysfs_reset_mask_fini(adev);
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++)
+		amdgpu_vcn_sw_fini(adev, i);
+
+	return 0;
+}
+
+/**
+ * vcn_v5_0_0_hw_init - start and test VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Initialize the hardware, boot up the VCPU and do some testing
+ */
+static int vcn_v5_0_0_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	int i, r;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		ring = &adev->vcn.inst[i].ring_enc[0];
+
+		adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
+			((adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 8 * i), i);
+
+		r = amdgpu_ring_test_helper(ring);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v5_0_0_hw_fini - stop the hardware block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Stop the VCN block, mark ring as not ready any more
+ */
+static int vcn_v5_0_0_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[i];
+
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		cancel_delayed_work_sync(&vinst->idle_work);
+
+		if (!amdgpu_sriov_vf(adev)) {
+			if ((adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) ||
+			    (vinst->cur_state != AMD_PG_STATE_GATE &&
+			     RREG32_SOC15(VCN, i, regUVD_STATUS))) {
+				vinst->set_pg_state(vinst, AMD_PG_STATE_GATE);
+			}
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v5_0_0_suspend - suspend VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * HW fini and suspend VCN block
+ */
+static int vcn_v5_0_0_suspend(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r, i;
+
+	r = vcn_v5_0_0_hw_fini(ip_block);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		r = amdgpu_vcn_suspend(ip_block->adev, i);
+		if (r)
+			return r;
+	}
+
+	return r;
+}
+
+/**
+ * vcn_v5_0_0_resume - resume VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Resume firmware and hw init VCN block
+ */
+static int vcn_v5_0_0_resume(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r, i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		r = amdgpu_vcn_resume(ip_block->adev, i);
+		if (r)
+			return r;
+	}
+
+	r = vcn_v5_0_0_hw_init(ip_block);
+
+	return r;
+}
+
+/**
+ * vcn_v5_0_0_mc_resume - memory controller programming
+ *
+ * @vinst: VCN instance
+ *
+ * Let the VCN memory controller know it's offsets
+ */
+static void vcn_v5_0_0_mc_resume(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst = vinst->inst;
+	uint32_t offset, size;
+	const struct common_firmware_header *hdr;
+
+	hdr = (const struct common_firmware_header *)adev->vcn.inst[inst].fw->data;
+	size = AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8);
+
+	/* cache window 0: fw */
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst].tmr_mc_addr_lo));
+		WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst].tmr_mc_addr_hi));
+		WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_OFFSET0, 0);
+		offset = 0;
+	} else {
+		WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			lower_32_bits(adev->vcn.inst[inst].gpu_addr));
+		WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			upper_32_bits(adev->vcn.inst[inst].gpu_addr));
+		offset = size;
+		WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_OFFSET0, AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
+	}
+	WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_SIZE0, size);
+
+	/* cache window 1: stack */
+	WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW,
+		lower_32_bits(adev->vcn.inst[inst].gpu_addr + offset));
+	WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH,
+		upper_32_bits(adev->vcn.inst[inst].gpu_addr + offset));
+	WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_OFFSET1, 0);
+	WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_SIZE1, AMDGPU_VCN_STACK_SIZE);
+
+	/* cache window 2: context */
+	WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW,
+		lower_32_bits(adev->vcn.inst[inst].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE));
+	WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH,
+		upper_32_bits(adev->vcn.inst[inst].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE));
+	WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_OFFSET2, 0);
+	WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_SIZE2, AMDGPU_VCN_CONTEXT_SIZE);
+
+	/* non-cache window */
+	WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_NC0_64BIT_BAR_LOW,
+		lower_32_bits(adev->vcn.inst[inst].fw_shared.gpu_addr));
+	WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH,
+		upper_32_bits(adev->vcn.inst[inst].fw_shared.gpu_addr));
+	WREG32_SOC15(VCN, inst, regUVD_VCPU_NONCACHE_OFFSET0, 0);
+	WREG32_SOC15(VCN, inst, regUVD_VCPU_NONCACHE_SIZE0,
+		AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn5_fw_shared)));
+}
+
+/**
+ * vcn_v5_0_0_mc_resume_dpg_mode - memory controller programming for dpg mode
+ *
+ * @vinst: VCN instance
+ * @indirect: indirectly write sram
+ *
+ * Let the VCN memory controller know it's offsets with dpg mode
+ */
+static void vcn_v5_0_0_mc_resume_dpg_mode(struct amdgpu_vcn_inst *vinst,
+					  bool indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t offset, size;
+	const struct common_firmware_header *hdr;
+
+	hdr = (const struct common_firmware_header *)adev->vcn.inst[inst_idx].fw->data;
+	size = AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8);
+
+	/* cache window 0: fw */
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		if (!indirect) {
+			WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+				VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+				(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst_idx].tmr_mc_addr_lo), 0, indirect);
+			WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+				VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+				(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst_idx].tmr_mc_addr_hi), 0, indirect);
+			WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+				VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET0), 0, 0, indirect);
+		} else {
+			WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+				VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW), 0, 0, indirect);
+			WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+				VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH), 0, 0, indirect);
+			WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+				VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET0), 0, 0, indirect);
+		}
+		offset = 0;
+	} else {
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr), 0, indirect);
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr), 0, indirect);
+		offset = size;
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET0),
+			AMDGPU_UVD_FIRMWARE_OFFSET >> 3, 0, indirect);
+	}
+
+	if (!indirect)
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_VCPU_CACHE_SIZE0), size, 0, indirect);
+	else
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_VCPU_CACHE_SIZE0), 0, 0, indirect);
+
+	/* cache window 1: stack */
+	if (!indirect) {
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset), 0, indirect);
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset), 0, indirect);
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET1), 0, 0, indirect);
+	} else {
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW), 0, 0, indirect);
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH), 0, 0, indirect);
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET1), 0, 0, indirect);
+	}
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, inst_idx, regUVD_VCPU_CACHE_SIZE1), AMDGPU_VCN_STACK_SIZE, 0, indirect);
+
+	/* cache window 2: context */
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW),
+		lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE), 0, indirect);
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH),
+		upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE), 0, indirect);
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET2), 0, 0, indirect);
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_VCPU_CACHE_SIZE2), AMDGPU_VCN_CONTEXT_SIZE, 0, indirect);
+
+	/* non-cache window */
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_LMI_VCPU_NC0_64BIT_BAR_LOW),
+		lower_32_bits(adev->vcn.inst[inst_idx].fw_shared.gpu_addr), 0, indirect);
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH),
+		upper_32_bits(adev->vcn.inst[inst_idx].fw_shared.gpu_addr), 0, indirect);
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_VCPU_NONCACHE_OFFSET0), 0, 0, indirect);
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_VCPU_NONCACHE_SIZE0),
+		AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn5_fw_shared)), 0, indirect);
+
+	/* VCN global tiling registers */
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_GFX10_ADDR_CONFIG),
+		adev->gfx.config.gb_addr_config, 0, indirect);
+
+	return;
+}
+
+/**
+ * vcn_v5_0_0_disable_static_power_gating - disable VCN static power gating
+ *
+ * @vinst: VCN instance
+ *
+ * Disable static power gating for VCN block
+ */
+static void vcn_v5_0_0_disable_static_power_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst = vinst->inst;
+	uint32_t data = 0;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN) {
+		data = 1 << UVD_IPX_DLDO_CONFIG__ONO2_PWR_CONFIG__SHIFT;
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS, 0,
+				UVD_IPX_DLDO_STATUS__ONO2_PWR_STATUS_MASK);
+
+		data = 2 << UVD_IPX_DLDO_CONFIG__ONO3_PWR_CONFIG__SHIFT;
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+				1 << UVD_IPX_DLDO_STATUS__ONO3_PWR_STATUS__SHIFT,
+				UVD_IPX_DLDO_STATUS__ONO3_PWR_STATUS_MASK);
+
+		data = 2 << UVD_IPX_DLDO_CONFIG__ONO4_PWR_CONFIG__SHIFT;
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+				1 << UVD_IPX_DLDO_STATUS__ONO4_PWR_STATUS__SHIFT,
+				UVD_IPX_DLDO_STATUS__ONO4_PWR_STATUS_MASK);
+
+		data = 2 << UVD_IPX_DLDO_CONFIG__ONO5_PWR_CONFIG__SHIFT;
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+				1 << UVD_IPX_DLDO_STATUS__ONO5_PWR_STATUS__SHIFT,
+				UVD_IPX_DLDO_STATUS__ONO5_PWR_STATUS_MASK);
+	} else {
+		data = 1 << UVD_IPX_DLDO_CONFIG__ONO2_PWR_CONFIG__SHIFT;
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS, 0,
+				UVD_IPX_DLDO_STATUS__ONO2_PWR_STATUS_MASK);
+
+		data = 1 << UVD_IPX_DLDO_CONFIG__ONO3_PWR_CONFIG__SHIFT;
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS, 0,
+				UVD_IPX_DLDO_STATUS__ONO3_PWR_STATUS_MASK);
+
+		data = 1 << UVD_IPX_DLDO_CONFIG__ONO4_PWR_CONFIG__SHIFT;
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS, 0,
+				UVD_IPX_DLDO_STATUS__ONO4_PWR_STATUS_MASK);
+
+		data = 1 << UVD_IPX_DLDO_CONFIG__ONO5_PWR_CONFIG__SHIFT;
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS, 0,
+				UVD_IPX_DLDO_STATUS__ONO5_PWR_STATUS_MASK);
+	}
+
+	data = RREG32_SOC15(VCN, inst, regUVD_POWER_STATUS);
+	data &= ~0x103;
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN)
+		data |= UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON |
+			UVD_POWER_STATUS__UVD_PG_EN_MASK;
+
+	WREG32_SOC15(VCN, inst, regUVD_POWER_STATUS, data);
+	return;
+}
+
+/**
+ * vcn_v5_0_0_enable_static_power_gating - enable VCN static power gating
+ *
+ * @vinst: VCN instance
+ *
+ * Enable static power gating for VCN block
+ */
+static void vcn_v5_0_0_enable_static_power_gating(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst = vinst->inst;
+	uint32_t data;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN) {
+		/* Before power off, this indicator has to be turned on */
+		data = RREG32_SOC15(VCN, inst, regUVD_POWER_STATUS);
+		data &= ~UVD_POWER_STATUS__UVD_POWER_STATUS_MASK;
+		data |= UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF;
+		WREG32_SOC15(VCN, inst, regUVD_POWER_STATUS, data);
+
+		data = 2 << UVD_IPX_DLDO_CONFIG__ONO5_PWR_CONFIG__SHIFT;
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+				1 << UVD_IPX_DLDO_STATUS__ONO5_PWR_STATUS__SHIFT,
+				UVD_IPX_DLDO_STATUS__ONO5_PWR_STATUS_MASK);
+
+		data = 2 << UVD_IPX_DLDO_CONFIG__ONO4_PWR_CONFIG__SHIFT;
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+				1 << UVD_IPX_DLDO_STATUS__ONO4_PWR_STATUS__SHIFT,
+				UVD_IPX_DLDO_STATUS__ONO4_PWR_STATUS_MASK);
+
+		data = 2 << UVD_IPX_DLDO_CONFIG__ONO3_PWR_CONFIG__SHIFT;
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+				1 << UVD_IPX_DLDO_STATUS__ONO3_PWR_STATUS__SHIFT,
+				UVD_IPX_DLDO_STATUS__ONO3_PWR_STATUS_MASK);
+
+		data = 2 << UVD_IPX_DLDO_CONFIG__ONO2_PWR_CONFIG__SHIFT;
+		WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+		SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+				1 << UVD_IPX_DLDO_STATUS__ONO2_PWR_STATUS__SHIFT,
+				UVD_IPX_DLDO_STATUS__ONO2_PWR_STATUS_MASK);
+	}
+	return;
+}
+
+/**
+ * vcn_v5_0_0_disable_clock_gating - disable VCN clock gating
+ *
+ * @vinst: VCN instance
+ *
+ * Disable clock gating for VCN block
+ */
+static void vcn_v5_0_0_disable_clock_gating(struct amdgpu_vcn_inst *vinst)
+{
+	return;
+}
+
+#if 0
+/**
+ * vcn_v5_0_0_disable_clock_gating_dpg_mode - disable VCN clock gating dpg mode
+ *
+ * @vinst: VCN instance
+ * @sram_sel: sram select
+ * @indirect: indirectly write sram
+ *
+ * Disable clock gating for VCN block with dpg mode
+ */
+static void vcn_v5_0_0_disable_clock_gating_dpg_mode(struct amdgpu_vcn_inst *vinst,
+						     uint8_t sram_sel,
+						     uint8_t indirect)
+{
+	return;
+}
+#endif
+
+/**
+ * vcn_v5_0_0_enable_clock_gating - enable VCN clock gating
+ *
+ * @vinst: VCN instance
+ *
+ * Enable clock gating for VCN block
+ */
+static void vcn_v5_0_0_enable_clock_gating(struct amdgpu_vcn_inst *vinst)
+{
+	return;
+}
+
+/**
+ * vcn_v5_0_0_start_dpg_mode - VCN start with dpg mode
+ *
+ * @vinst: VCN instance
+ * @indirect: indirectly write sram
+ *
+ * Start VCN block with dpg mode
+ */
+static int vcn_v5_0_0_start_dpg_mode(struct amdgpu_vcn_inst *vinst,
+				     bool indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	struct amdgpu_vcn5_fw_shared *fw_shared = adev->vcn.inst[inst_idx].fw_shared.cpu_addr;
+	struct amdgpu_ring *ring;
+	uint32_t tmp;
+	int ret;
+
+	/* disable register anti-hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, regUVD_POWER_STATUS), 1,
+		~UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+	/* enable dynamic power gating mode */
+	tmp = RREG32_SOC15(VCN, inst_idx, regUVD_POWER_STATUS);
+	tmp |= UVD_POWER_STATUS__UVD_PG_MODE_MASK;
+	tmp |= UVD_POWER_STATUS__UVD_PG_EN_MASK;
+	WREG32_SOC15(VCN, inst_idx, regUVD_POWER_STATUS, tmp);
+
+	if (indirect)
+		adev->vcn.inst[inst_idx].dpg_sram_curr_addr = (uint32_t *)adev->vcn.inst[inst_idx].dpg_sram_cpu_addr;
+
+	/* enable VCPU clock */
+	tmp = (0xFF << UVD_VCPU_CNTL__PRB_TIMEOUT_VAL__SHIFT);
+	tmp |= UVD_VCPU_CNTL__CLK_EN_MASK | UVD_VCPU_CNTL__BLK_RST_MASK;
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_VCPU_CNTL), tmp, 0, indirect);
+
+	/* disable master interrupt */
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_MASTINT_EN), 0, 0, indirect);
+
+	/* setup regUVD_LMI_CTRL */
+	tmp = (UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+		UVD_LMI_CTRL__REQ_MODE_MASK |
+		UVD_LMI_CTRL__CRC_RESET_MASK |
+		UVD_LMI_CTRL__MASK_MC_URGENT_MASK |
+		UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+		UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
+		(8 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
+		0x00100000L);
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_LMI_CTRL), tmp, 0, indirect);
+
+	vcn_v5_0_0_mc_resume_dpg_mode(vinst, indirect);
+
+	tmp = (0xFF << UVD_VCPU_CNTL__PRB_TIMEOUT_VAL__SHIFT);
+	tmp |= UVD_VCPU_CNTL__CLK_EN_MASK;
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_VCPU_CNTL), tmp, 0, indirect);
+
+	/* enable LMI MC and UMC channels */
+	tmp = 0x1f << UVD_LMI_CTRL2__RE_OFLD_MIF_WR_REQ_NUM__SHIFT;
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_LMI_CTRL2), tmp, 0, indirect);
+
+	/* enable master interrupt */
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, inst_idx, regUVD_MASTINT_EN),
+		UVD_MASTINT_EN__VCPU_EN_MASK, 0, indirect);
+
+	if (indirect) {
+		ret = amdgpu_vcn_psp_update_sram(adev, inst_idx, 0);
+		if (ret) {
+			dev_err(adev->dev, "%s: vcn sram load failed %d\n", __func__, ret);
+			return ret;
+		}
+	}
+
+	ring = &adev->vcn.inst[inst_idx].ring_enc[0];
+
+	WREG32_SOC15(VCN, inst_idx, regUVD_RB_BASE_LO, ring->gpu_addr);
+	WREG32_SOC15(VCN, inst_idx, regUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(VCN, inst_idx, regUVD_RB_SIZE, ring->ring_size / 4);
+
+	tmp = RREG32_SOC15(VCN, inst_idx, regVCN_RB_ENABLE);
+	tmp &= ~(VCN_RB_ENABLE__RB1_EN_MASK);
+	WREG32_SOC15(VCN, inst_idx, regVCN_RB_ENABLE, tmp);
+	fw_shared->sq.queue_mode |= FW_QUEUE_RING_RESET;
+	WREG32_SOC15(VCN, inst_idx, regUVD_RB_RPTR, 0);
+	WREG32_SOC15(VCN, inst_idx, regUVD_RB_WPTR, 0);
+
+	tmp = RREG32_SOC15(VCN, inst_idx, regUVD_RB_RPTR);
+	WREG32_SOC15(VCN, inst_idx, regUVD_RB_WPTR, tmp);
+	ring->wptr = RREG32_SOC15(VCN, inst_idx, regUVD_RB_WPTR);
+
+	tmp = RREG32_SOC15(VCN, inst_idx, regVCN_RB_ENABLE);
+	tmp |= VCN_RB_ENABLE__RB1_EN_MASK;
+	WREG32_SOC15(VCN, inst_idx, regVCN_RB_ENABLE, tmp);
+	fw_shared->sq.queue_mode &= ~(FW_QUEUE_RING_RESET | FW_QUEUE_DPG_HOLD_OFF);
+
+	WREG32_SOC15(VCN, inst_idx, regVCN_RB1_DB_CTRL,
+		ring->doorbell_index << VCN_RB1_DB_CTRL__OFFSET__SHIFT |
+		VCN_RB1_DB_CTRL__EN_MASK);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, inst_idx, regUVD_STATUS);
+
+	return 0;
+}
+
+/**
+ * vcn_v5_0_0_start - VCN start
+ *
+ * @vinst: VCN instance
+ *
+ * Start VCN block
+ */
+static int vcn_v5_0_0_start(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int i = vinst->inst;
+	struct amdgpu_vcn5_fw_shared *fw_shared;
+	struct amdgpu_ring *ring;
+	uint32_t tmp;
+	int j, k, r;
+
+	if (adev->vcn.harvest_config & (1 << i))
+		return 0;
+
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_vcn(adev, true, i);
+
+	fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)
+		return vcn_v5_0_0_start_dpg_mode(vinst, adev->vcn.inst[i].indirect_sram);
+
+	/* disable VCN power gating */
+	vcn_v5_0_0_disable_static_power_gating(vinst);
+
+	/* set VCN status busy */
+	tmp = RREG32_SOC15(VCN, i, regUVD_STATUS) | UVD_STATUS__UVD_BUSY;
+	WREG32_SOC15(VCN, i, regUVD_STATUS, tmp);
+
+	/* enable VCPU clock */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL),
+		 UVD_VCPU_CNTL__CLK_EN_MASK, ~UVD_VCPU_CNTL__CLK_EN_MASK);
+
+	/* disable master interrupt */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_MASTINT_EN), 0,
+		 ~UVD_MASTINT_EN__VCPU_EN_MASK);
+
+	/* enable LMI MC and UMC channels */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_LMI_CTRL2), 0,
+		 ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+
+	tmp = RREG32_SOC15(VCN, i, regUVD_SOFT_RESET);
+	tmp &= ~UVD_SOFT_RESET__LMI_SOFT_RESET_MASK;
+	tmp &= ~UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK;
+	WREG32_SOC15(VCN, i, regUVD_SOFT_RESET, tmp);
+
+	/* setup regUVD_LMI_CTRL */
+	tmp = RREG32_SOC15(VCN, i, regUVD_LMI_CTRL);
+	WREG32_SOC15(VCN, i, regUVD_LMI_CTRL, tmp |
+		     UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+		     UVD_LMI_CTRL__MASK_MC_URGENT_MASK |
+		     UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+		     UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK);
+
+	vcn_v5_0_0_mc_resume(vinst);
+
+	/* VCN global tiling registers */
+	WREG32_SOC15(VCN, i, regUVD_GFX10_ADDR_CONFIG,
+		     adev->gfx.config.gb_addr_config);
+
+	/* unblock VCPU register access */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_RB_ARB_CTRL), 0,
+		 ~UVD_RB_ARB_CTRL__VCPU_DIS_MASK);
+
+	/* release VCPU reset to boot */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), 0,
+		 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+	for (j = 0; j < 10; ++j) {
+		uint32_t status;
+
+		for (k = 0; k < 100; ++k) {
+			status = RREG32_SOC15(VCN, i, regUVD_STATUS);
+			if (status & 2)
+				break;
+			mdelay(10);
+			if (amdgpu_emu_mode == 1)
+				msleep(1);
+		}
+
+		if (amdgpu_emu_mode == 1) {
+			r = -1;
+			if (status & 2) {
+				r = 0;
+				break;
+			}
+		} else {
+			r = 0;
+			if (status & 2)
+				break;
+
+			dev_err(adev->dev,
+				"VCN[%d] is not responding, trying to reset the VCPU!!!\n", i);
+			WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL),
+				 UVD_VCPU_CNTL__BLK_RST_MASK,
+				 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+			mdelay(10);
+			WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), 0,
+				 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+			mdelay(10);
+			r = -1;
+		}
+	}
+
+	if (r) {
+		dev_err(adev->dev, "VCN[%d] is not responding, giving up!!!\n", i);
+		return r;
+	}
+
+	/* enable master interrupt */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_MASTINT_EN),
+		 UVD_MASTINT_EN__VCPU_EN_MASK,
+		 ~UVD_MASTINT_EN__VCPU_EN_MASK);
+
+	/* clear the busy bit of VCN_STATUS */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_STATUS), 0,
+		 ~(2 << UVD_STATUS__VCPU_REPORT__SHIFT));
+
+	ring = &adev->vcn.inst[i].ring_enc[0];
+	WREG32_SOC15(VCN, i, regVCN_RB1_DB_CTRL,
+		     ring->doorbell_index << VCN_RB1_DB_CTRL__OFFSET__SHIFT |
+		     VCN_RB1_DB_CTRL__EN_MASK);
+
+	WREG32_SOC15(VCN, i, regUVD_RB_BASE_LO, ring->gpu_addr);
+	WREG32_SOC15(VCN, i, regUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(VCN, i, regUVD_RB_SIZE, ring->ring_size / 4);
+
+	tmp = RREG32_SOC15(VCN, i, regVCN_RB_ENABLE);
+	tmp &= ~(VCN_RB_ENABLE__RB1_EN_MASK);
+	WREG32_SOC15(VCN, i, regVCN_RB_ENABLE, tmp);
+	fw_shared->sq.queue_mode |= FW_QUEUE_RING_RESET;
+	WREG32_SOC15(VCN, i, regUVD_RB_RPTR, 0);
+	WREG32_SOC15(VCN, i, regUVD_RB_WPTR, 0);
+
+	tmp = RREG32_SOC15(VCN, i, regUVD_RB_RPTR);
+	WREG32_SOC15(VCN, i, regUVD_RB_WPTR, tmp);
+	ring->wptr = RREG32_SOC15(VCN, i, regUVD_RB_WPTR);
+
+	tmp = RREG32_SOC15(VCN, i, regVCN_RB_ENABLE);
+	tmp |= VCN_RB_ENABLE__RB1_EN_MASK;
+	WREG32_SOC15(VCN, i, regVCN_RB_ENABLE, tmp);
+	fw_shared->sq.queue_mode &= ~(FW_QUEUE_RING_RESET | FW_QUEUE_DPG_HOLD_OFF);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, i, regUVD_STATUS);
+
+	return 0;
+}
+
+/**
+ * vcn_v5_0_0_stop_dpg_mode - VCN stop with dpg mode
+ *
+ * @vinst: VCN instance
+ *
+ * Stop VCN block with dpg mode
+ */
+static void vcn_v5_0_0_stop_dpg_mode(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	struct dpg_pause_state state = {.fw_based = VCN_DPG_STATE__UNPAUSE};
+	uint32_t tmp;
+
+	vcn_v5_0_0_pause_dpg_mode(vinst, &state);
+
+	/* Wait for power status to be 1 */
+	SOC15_WAIT_ON_RREG(VCN, inst_idx, regUVD_POWER_STATUS, 1,
+		UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+	/* wait for read ptr to be equal to write ptr */
+	tmp = RREG32_SOC15(VCN, inst_idx, regUVD_RB_WPTR);
+	SOC15_WAIT_ON_RREG(VCN, inst_idx, regUVD_RB_RPTR, tmp, 0xFFFFFFFF);
+
+	/* disable dynamic power gating mode */
+	WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, regUVD_POWER_STATUS), 0,
+		~UVD_POWER_STATUS__UVD_PG_MODE_MASK);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, inst_idx, regUVD_STATUS);
+
+	return;
+}
+
+/**
+ * vcn_v5_0_0_stop - VCN stop
+ *
+ * @vinst: VCN instance
+ *
+ * Stop VCN block
+ */
+static int vcn_v5_0_0_stop(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int i = vinst->inst;
+	struct amdgpu_vcn5_fw_shared *fw_shared;
+	uint32_t tmp;
+	int r = 0;
+
+	if (adev->vcn.harvest_config & (1 << i))
+		return 0;
+
+	fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+	fw_shared->sq.queue_mode |= FW_QUEUE_DPG_HOLD_OFF;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) {
+		vcn_v5_0_0_stop_dpg_mode(vinst);
+		r = 0;
+		goto done;
+	}
+
+	/* wait for vcn idle */
+	r = SOC15_WAIT_ON_RREG(VCN, i, regUVD_STATUS, UVD_STATUS__IDLE, 0x7);
+	if (r)
+		goto done;
+
+	tmp = UVD_LMI_STATUS__VCPU_LMI_WRITE_CLEAN_MASK |
+		UVD_LMI_STATUS__READ_CLEAN_MASK |
+		UVD_LMI_STATUS__WRITE_CLEAN_MASK |
+		UVD_LMI_STATUS__WRITE_CLEAN_RAW_MASK;
+	r = SOC15_WAIT_ON_RREG(VCN, i, regUVD_LMI_STATUS, tmp, tmp);
+	if (r)
+		goto done;
+
+	/* disable LMI UMC channel */
+	tmp = RREG32_SOC15(VCN, i, regUVD_LMI_CTRL2);
+	tmp |= UVD_LMI_CTRL2__STALL_ARB_UMC_MASK;
+	WREG32_SOC15(VCN, i, regUVD_LMI_CTRL2, tmp);
+	tmp = UVD_LMI_STATUS__UMC_READ_CLEAN_RAW_MASK |
+		UVD_LMI_STATUS__UMC_WRITE_CLEAN_RAW_MASK;
+	r = SOC15_WAIT_ON_RREG(VCN, i, regUVD_LMI_STATUS, tmp, tmp);
+	if (r)
+		goto done;
+
+	/* block VCPU register access */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_RB_ARB_CTRL),
+		 UVD_RB_ARB_CTRL__VCPU_DIS_MASK,
+		 ~UVD_RB_ARB_CTRL__VCPU_DIS_MASK);
+
+	/* reset VCPU */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL),
+		 UVD_VCPU_CNTL__BLK_RST_MASK,
+		 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+	/* disable VCPU clock */
+	WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), 0,
+		 ~(UVD_VCPU_CNTL__CLK_EN_MASK));
+
+	/* apply soft reset */
+	tmp = RREG32_SOC15(VCN, i, regUVD_SOFT_RESET);
+	tmp |= UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK;
+	WREG32_SOC15(VCN, i, regUVD_SOFT_RESET, tmp);
+	tmp = RREG32_SOC15(VCN, i, regUVD_SOFT_RESET);
+	tmp |= UVD_SOFT_RESET__LMI_SOFT_RESET_MASK;
+	WREG32_SOC15(VCN, i, regUVD_SOFT_RESET, tmp);
+
+	/* clear status */
+	WREG32_SOC15(VCN, i, regUVD_STATUS, 0);
+
+	/* enable VCN power gating */
+	vcn_v5_0_0_enable_static_power_gating(vinst);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, i, regUVD_STATUS);
+
+done:
+	if (adev->pm.dpm_enabled)
+		amdgpu_dpm_enable_vcn(adev, false, i);
+
+	return r;
+}
+
+/**
+ * vcn_v5_0_0_pause_dpg_mode - VCN pause with dpg mode
+ *
+ * @vinst: VCN instance
+ * @new_state: pause state
+ *
+ * Pause dpg mode for VCN block
+ */
+static int vcn_v5_0_0_pause_dpg_mode(struct amdgpu_vcn_inst *vinst,
+				     struct dpg_pause_state *new_state)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t reg_data = 0;
+	int ret_code;
+
+	/* pause/unpause if state is changed */
+	if (adev->vcn.inst[inst_idx].pause_state.fw_based != new_state->fw_based) {
+		DRM_DEV_DEBUG(adev->dev, "dpg pause state changed %d -> %d",
+			adev->vcn.inst[inst_idx].pause_state.fw_based,  new_state->fw_based);
+		reg_data = RREG32_SOC15(VCN, inst_idx, regUVD_DPG_PAUSE) &
+			(~UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK);
+
+		if (new_state->fw_based == VCN_DPG_STATE__PAUSE) {
+			ret_code = SOC15_WAIT_ON_RREG(VCN, inst_idx, regUVD_POWER_STATUS, 0x1,
+					UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+			if (!ret_code) {
+				/* pause DPG */
+				reg_data |= UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
+				WREG32_SOC15(VCN, inst_idx, regUVD_DPG_PAUSE, reg_data);
+
+				/* wait for ACK */
+				SOC15_WAIT_ON_RREG(VCN, inst_idx, regUVD_DPG_PAUSE,
+					UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK,
+					UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK);
+			}
+		} else {
+			/* unpause dpg, no need to wait */
+			reg_data &= ~UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
+			WREG32_SOC15(VCN, inst_idx, regUVD_DPG_PAUSE, reg_data);
+		}
+		adev->vcn.inst[inst_idx].pause_state.fw_based = new_state->fw_based;
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v5_0_0_unified_ring_get_rptr - get unified read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware unified read pointer
+ */
+static uint64_t vcn_v5_0_0_unified_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring != &adev->vcn.inst[ring->me].ring_enc[0])
+		DRM_ERROR("wrong ring id is identified in %s", __func__);
+
+	return RREG32_SOC15(VCN, ring->me, regUVD_RB_RPTR);
+}
+
+/**
+ * vcn_v5_0_0_unified_ring_get_wptr - get unified write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware unified write pointer
+ */
+static uint64_t vcn_v5_0_0_unified_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring != &adev->vcn.inst[ring->me].ring_enc[0])
+		DRM_ERROR("wrong ring id is identified in %s", __func__);
+
+	if (ring->use_doorbell)
+		return *ring->wptr_cpu_addr;
+	else
+		return RREG32_SOC15(VCN, ring->me, regUVD_RB_WPTR);
+}
+
+/**
+ * vcn_v5_0_0_unified_ring_set_wptr - set enc write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the enc write pointer to the hardware
+ */
+static void vcn_v5_0_0_unified_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring != &adev->vcn.inst[ring->me].ring_enc[0])
+		DRM_ERROR("wrong ring id is identified in %s", __func__);
+
+	if (ring->use_doorbell) {
+		*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
+		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+	} else {
+		WREG32_SOC15(VCN, ring->me, regUVD_RB_WPTR, lower_32_bits(ring->wptr));
+	}
+}
+
+static int vcn_v5_0_0_ring_reset(struct amdgpu_ring *ring,
+				 unsigned int vmid,
+				 struct amdgpu_fence *timedout_fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[ring->me];
+	int r;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+	r = vcn_v5_0_0_stop(vinst);
+	if (r)
+		return r;
+	r = vcn_v5_0_0_start(vinst);
+	if (r)
+		return r;
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static const struct amdgpu_ring_funcs vcn_v5_0_0_unified_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_ENC,
+	.align_mask = 0x3f,
+	.nop = VCN_ENC_CMD_NO_OP,
+	.get_rptr = vcn_v5_0_0_unified_ring_get_rptr,
+	.get_wptr = vcn_v5_0_0_unified_ring_get_wptr,
+	.set_wptr = vcn_v5_0_0_unified_ring_set_wptr,
+	.emit_frame_size =
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 4 +
+		4 + /* vcn_v2_0_enc_ring_emit_vm_flush */
+		5 + 5 + /* vcn_v2_0_enc_ring_emit_fence x2 vm fence */
+		1, /* vcn_v2_0_enc_ring_insert_end */
+	.emit_ib_size = 5, /* vcn_v2_0_enc_ring_emit_ib */
+	.emit_ib = vcn_v2_0_enc_ring_emit_ib,
+	.emit_fence = vcn_v2_0_enc_ring_emit_fence,
+	.emit_vm_flush = vcn_v2_0_enc_ring_emit_vm_flush,
+	.test_ring = amdgpu_vcn_enc_ring_test_ring,
+	.test_ib = amdgpu_vcn_unified_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.insert_end = vcn_v2_0_enc_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_vcn_ring_begin_use,
+	.end_use = amdgpu_vcn_ring_end_use,
+	.emit_wreg = vcn_v2_0_enc_ring_emit_wreg,
+	.emit_reg_wait = vcn_v2_0_enc_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = vcn_v5_0_0_ring_reset,
+};
+
+/**
+ * vcn_v5_0_0_set_unified_ring_funcs - set unified ring functions
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set unified ring functions
+ */
+static void vcn_v5_0_0_set_unified_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		adev->vcn.inst[i].ring_enc[0].funcs = &vcn_v5_0_0_unified_ring_vm_funcs;
+		adev->vcn.inst[i].ring_enc[0].me = i;
+	}
+}
+
+/**
+ * vcn_v5_0_0_is_idle - check VCN block is idle
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block structure
+ *
+ * Check whether VCN block is idle
+ */
+static bool vcn_v5_0_0_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, ret = 1;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		ret &= (RREG32_SOC15(VCN, i, regUVD_STATUS) == UVD_STATUS__IDLE);
+	}
+
+	return ret;
+}
+
+/**
+ * vcn_v5_0_0_wait_for_idle - wait for VCN block idle
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Wait for VCN block idle
+ */
+static int vcn_v5_0_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, ret = 0;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		ret = SOC15_WAIT_ON_RREG(VCN, i, regUVD_STATUS, UVD_STATUS__IDLE,
+			UVD_STATUS__IDLE);
+		if (ret)
+			return ret;
+	}
+
+	return ret;
+}
+
+/**
+ * vcn_v5_0_0_set_clockgating_state - set VCN block clockgating state
+ *
+ * @ip_block: amdgpu_ip_block pointer
+ * @state: clock gating state
+ *
+ * Set VCN block clockgating state
+ */
+static int vcn_v5_0_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = state == AMD_CG_STATE_GATE;
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[i];
+
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		if (enable) {
+			if (RREG32_SOC15(VCN, i, regUVD_STATUS) != UVD_STATUS__IDLE)
+				return -EBUSY;
+			vcn_v5_0_0_enable_clock_gating(vinst);
+		} else {
+			vcn_v5_0_0_disable_clock_gating(vinst);
+		}
+	}
+
+	return 0;
+}
+
+static int vcn_v5_0_0_set_pg_state(struct amdgpu_vcn_inst *vinst,
+				   enum amd_powergating_state state)
+{
+	int ret = 0;
+
+	if (state == vinst->cur_state)
+		return 0;
+
+	if (state == AMD_PG_STATE_GATE)
+		ret = vcn_v5_0_0_stop(vinst);
+	else
+		ret = vcn_v5_0_0_start(vinst);
+
+	if (!ret)
+		vinst->cur_state = state;
+
+	return ret;
+}
+
+/**
+ * vcn_v5_0_0_process_interrupt - process VCN block interrupt
+ *
+ * @adev: amdgpu_device pointer
+ * @source: interrupt sources
+ * @entry: interrupt entry from clients and sources
+ *
+ * Process VCN block interrupt
+ */
+static int vcn_v5_0_0_process_interrupt(struct amdgpu_device *adev, struct amdgpu_irq_src *source,
+	struct amdgpu_iv_entry *entry)
+{
+	uint32_t ip_instance;
+
+	switch (entry->client_id) {
+	case SOC15_IH_CLIENTID_VCN:
+		ip_instance = 0;
+		break;
+	case SOC15_IH_CLIENTID_VCN1:
+		ip_instance = 1;
+		break;
+	default:
+		DRM_ERROR("Unhandled client id: %d\n", entry->client_id);
+		return 0;
+	}
+
+	DRM_DEBUG("IH: VCN TRAP\n");
+
+	switch (entry->src_id) {
+	case VCN_5_0__SRCID__UVD_ENC_GENERAL_PURPOSE:
+		amdgpu_fence_process(&adev->vcn.inst[ip_instance].ring_enc[0]);
+		break;
+	case VCN_5_0__SRCID_UVD_POISON:
+		amdgpu_vcn_process_poison_irq(adev, source, entry);
+		break;
+	default:
+		DRM_ERROR("Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs vcn_v5_0_0_irq_funcs = {
+	.process = vcn_v5_0_0_process_interrupt,
+};
+
+/**
+ * vcn_v5_0_0_set_irq_funcs - set VCN block interrupt irq functions
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set VCN block interrupt irq functions
+ */
+static void vcn_v5_0_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		adev->vcn.inst[i].irq.num_types = adev->vcn.inst[i].num_enc_rings + 1;
+		adev->vcn.inst[i].irq.funcs = &vcn_v5_0_0_irq_funcs;
+	}
+}
+
+static const struct amd_ip_funcs vcn_v5_0_0_ip_funcs = {
+	.name = "vcn_v5_0_0",
+	.early_init = vcn_v5_0_0_early_init,
+	.sw_init = vcn_v5_0_0_sw_init,
+	.sw_fini = vcn_v5_0_0_sw_fini,
+	.hw_init = vcn_v5_0_0_hw_init,
+	.hw_fini = vcn_v5_0_0_hw_fini,
+	.suspend = vcn_v5_0_0_suspend,
+	.resume = vcn_v5_0_0_resume,
+	.is_idle = vcn_v5_0_0_is_idle,
+	.wait_for_idle = vcn_v5_0_0_wait_for_idle,
+	.set_clockgating_state = vcn_v5_0_0_set_clockgating_state,
+	.set_powergating_state = vcn_set_powergating_state,
+	.dump_ip_state = amdgpu_vcn_dump_ip_state,
+	.print_ip_state = amdgpu_vcn_print_ip_state,
+};
+
+const struct amdgpu_ip_block_version vcn_v5_0_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_VCN,
+	.major = 5,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &vcn_v5_0_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.h b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.h
new file mode 100644
index 000000000000..51bbccd4360f
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.h
@@ -0,0 +1,37 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VCN_V5_0_0_H__
+#define __VCN_V5_0_0_H__
+
+#define VCN_VID_SOC_ADDRESS                        0x1FC00
+#define VCN_AON_SOC_ADDRESS                        0x1F800
+#define VCN1_VID_SOC_ADDRESS                       0x48300
+#define VCN1_AON_SOC_ADDRESS                       0x48000
+
+#define VCN_VID_IP_ADDRESS                         0x0
+#define VCN_AON_IP_ADDRESS                         0x30000
+
+extern const struct amdgpu_ip_block_version vcn_v5_0_0_ip_block;
+
+#endif /* __VCN_V5_0_0_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.c b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.c
new file mode 100644
index 000000000000..714350cabf2f
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.c
@@ -0,0 +1,1727 @@
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc. All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include "amdgpu.h"
+#include "amdgpu_vcn.h"
+#include "amdgpu_pm.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "soc15_hw_ip.h"
+#include "vcn_v2_0.h"
+#include "vcn_v4_0_3.h"
+#include "mmsch_v5_0.h"
+
+#include "vcn/vcn_5_0_0_offset.h"
+#include "vcn/vcn_5_0_0_sh_mask.h"
+#include "ivsrcid/vcn/irqsrcs_vcn_5_0.h"
+#include "vcn_v5_0_0.h"
+#include "vcn_v5_0_1.h"
+
+#include <drm/drm_drv.h>
+
+static const struct amdgpu_hwip_reg_entry vcn_reg_list_5_0_1[] = {
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_STATUS),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_CONTEXT_ID),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_CONTEXT_ID2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_GPCOM_VCPU_DATA0),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_GPCOM_VCPU_DATA1),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_GPCOM_VCPU_CMD),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_HI),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_LO),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_HI2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_LO2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_HI3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_LO3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_HI4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_BASE_LO4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_RPTR),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_WPTR),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_RPTR2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_WPTR2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_RPTR3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_WPTR3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_RPTR4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_WPTR4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_SIZE),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_SIZE2),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_SIZE3),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_RB_SIZE4),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_DPG_LMA_CTL),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_DPG_LMA_DATA),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_DPG_LMA_MASK),
+	SOC15_REG_ENTRY_STR(VCN, 0, regUVD_DPG_PAUSE)
+};
+
+static int vcn_v5_0_1_start_sriov(struct amdgpu_device *adev);
+static void vcn_v5_0_1_set_unified_ring_funcs(struct amdgpu_device *adev);
+static void vcn_v5_0_1_set_irq_funcs(struct amdgpu_device *adev);
+static int vcn_v5_0_1_set_pg_state(struct amdgpu_vcn_inst *vinst,
+				   enum amd_powergating_state state);
+static void vcn_v5_0_1_unified_ring_set_wptr(struct amdgpu_ring *ring);
+static void vcn_v5_0_1_set_ras_funcs(struct amdgpu_device *adev);
+/**
+ * vcn_v5_0_1_early_init - set function pointers and load microcode
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Set ring and irq function pointers
+ * Load microcode from filesystem
+ */
+static int vcn_v5_0_1_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, r;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i)
+		/* re-use enc ring as unified ring */
+		adev->vcn.inst[i].num_enc_rings = 1;
+
+	vcn_v5_0_1_set_unified_ring_funcs(adev);
+	vcn_v5_0_1_set_irq_funcs(adev);
+	vcn_v5_0_1_set_ras_funcs(adev);
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		adev->vcn.inst[i].set_pg_state = vcn_v5_0_1_set_pg_state;
+
+		r = amdgpu_vcn_early_init(adev, i);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+static int vcn_v5_0_1_late_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	adev->vcn.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->vcn.inst[0].ring_enc[0]);
+
+	switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) {
+	case IP_VERSION(13, 0, 12):
+		if ((adev->psp.sos.fw_version >= 0x00450025) && amdgpu_dpm_reset_vcn_is_supported(adev))
+			adev->vcn.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static void vcn_v5_0_1_fw_shared_init(struct amdgpu_device *adev, int inst_idx)
+{
+	struct amdgpu_vcn5_fw_shared *fw_shared;
+
+	fw_shared = adev->vcn.inst[inst_idx].fw_shared.cpu_addr;
+
+	if (fw_shared->sq.is_enabled)
+		return;
+	fw_shared->present_flag_0 =
+		cpu_to_le32(AMDGPU_FW_SHARED_FLAG_0_UNIFIED_QUEUE);
+	fw_shared->sq.is_enabled = 1;
+
+	if (amdgpu_vcnfw_log)
+		amdgpu_vcn_fwlog_init(&adev->vcn.inst[inst_idx]);
+}
+
+/**
+ * vcn_v5_0_1_sw_init - sw init for VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Load firmware and sw initialization
+ */
+static int vcn_v5_0_1_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	int i, r, vcn_inst;
+
+	/* VCN UNIFIED TRAP */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+		VCN_5_0__SRCID__UVD_ENC_GENERAL_PURPOSE, &adev->vcn.inst->irq);
+	if (r)
+		return r;
+
+	/* VCN POISON TRAP */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+		VCN_5_0__SRCID_UVD_POISON, &adev->vcn.inst->ras_poison_irq);
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		vcn_inst = GET_INST(VCN, i);
+
+		r = amdgpu_vcn_sw_init(adev, i);
+		if (r)
+			return r;
+
+		amdgpu_vcn_setup_ucode(adev, i);
+
+		r = amdgpu_vcn_resume(adev, i);
+		if (r)
+			return r;
+
+		ring = &adev->vcn.inst[i].ring_enc[0];
+		ring->use_doorbell = true;
+		if (!amdgpu_sriov_vf(adev))
+			ring->doorbell_index =
+				(adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
+				11 * vcn_inst;
+		else
+			ring->doorbell_index =
+				(adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
+				32 * vcn_inst;
+
+		ring->vm_hub = AMDGPU_MMHUB0(adev->vcn.inst[i].aid_id);
+		sprintf(ring->name, "vcn_unified_%d", adev->vcn.inst[i].aid_id);
+
+		r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst[i].irq, 0,
+					AMDGPU_RING_PRIO_DEFAULT, &adev->vcn.inst[i].sched_score);
+		if (r)
+			return r;
+
+		vcn_v5_0_1_fw_shared_init(adev, i);
+	}
+
+	if (amdgpu_sriov_vf(adev)) {
+		r = amdgpu_virt_alloc_mm_table(adev);
+		if (r)
+			return r;
+	}
+
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__VCN)) {
+		r = amdgpu_vcn_ras_sw_init(adev);
+		if (r) {
+			dev_err(adev->dev, "Failed to initialize vcn ras block!\n");
+			return r;
+		}
+	}
+
+	r = amdgpu_vcn_reg_dump_init(adev, vcn_reg_list_5_0_1, ARRAY_SIZE(vcn_reg_list_5_0_1));
+	if (r)
+		return r;
+
+	return amdgpu_vcn_sysfs_reset_mask_init(adev);
+}
+
+/**
+ * vcn_v5_0_1_sw_fini - sw fini for VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * VCN suspend and free up sw allocation
+ */
+static int vcn_v5_0_1_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, r, idx;
+
+	if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+		for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+			struct amdgpu_vcn5_fw_shared *fw_shared;
+
+			fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+			fw_shared->present_flag_0 = 0;
+			fw_shared->sq.is_enabled = 0;
+		}
+
+		drm_dev_exit(idx);
+	}
+
+	if (amdgpu_sriov_vf(adev))
+		amdgpu_virt_free_mm_table(adev);
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		r = amdgpu_vcn_suspend(adev, i);
+		if (r)
+			return r;
+	}
+
+	amdgpu_vcn_sysfs_reset_mask_fini(adev);
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++)
+		amdgpu_vcn_sw_fini(adev, i);
+
+	return 0;
+}
+
+static int vcn_v5_0_1_hw_init_inst(struct amdgpu_device *adev, int i)
+{
+	struct amdgpu_ring *ring;
+	int vcn_inst;
+
+	vcn_inst = GET_INST(VCN, i);
+	ring = &adev->vcn.inst[i].ring_enc[0];
+
+	if (ring->use_doorbell)
+		adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
+			((adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
+			11 * vcn_inst),
+			adev->vcn.inst[i].aid_id);
+
+	return 0;
+}
+
+/**
+ * vcn_v5_0_1_hw_init - start and test VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Initialize the hardware, boot up the VCPU and do some testing
+ */
+static int vcn_v5_0_1_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_ring *ring;
+	int i, r;
+
+	if (amdgpu_sriov_vf(adev)) {
+		r = vcn_v5_0_1_start_sriov(adev);
+		if (r)
+			return r;
+
+		for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+			ring = &adev->vcn.inst[i].ring_enc[0];
+			ring->wptr = 0;
+			ring->wptr_old = 0;
+			vcn_v5_0_1_unified_ring_set_wptr(ring);
+			ring->sched.ready = true;
+		}
+	} else {
+		if (RREG32_SOC15(VCN, GET_INST(VCN, 0), regVCN_RRMT_CNTL) & 0x100)
+			adev->vcn.caps |= AMDGPU_VCN_CAPS(RRMT_ENABLED);
+		for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+			ring = &adev->vcn.inst[i].ring_enc[0];
+			vcn_v5_0_1_hw_init_inst(adev, i);
+
+			/* Re-init fw_shared, if required */
+			vcn_v5_0_1_fw_shared_init(adev, i);
+
+			r = amdgpu_ring_test_helper(ring);
+			if (r)
+				return r;
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v5_0_1_hw_fini - stop the hardware block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Stop the VCN block, mark ring as not ready any more
+ */
+static int vcn_v5_0_1_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[i];
+
+		cancel_delayed_work_sync(&adev->vcn.inst[i].idle_work);
+		if (vinst->cur_state != AMD_PG_STATE_GATE)
+			vinst->set_pg_state(vinst, AMD_PG_STATE_GATE);
+	}
+
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__VCN) && !amdgpu_sriov_vf(adev))
+		amdgpu_irq_put(adev, &adev->vcn.inst->ras_poison_irq, 0);
+
+	return 0;
+}
+
+/**
+ * vcn_v5_0_1_suspend - suspend VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * HW fini and suspend VCN block
+ */
+static int vcn_v5_0_1_suspend(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r, i;
+
+	r = vcn_v5_0_1_hw_fini(ip_block);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		r = amdgpu_vcn_suspend(ip_block->adev, i);
+		if (r)
+			return r;
+	}
+
+	return r;
+}
+
+/**
+ * vcn_v5_0_1_resume - resume VCN block
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Resume firmware and hw init VCN block
+ */
+static int vcn_v5_0_1_resume(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r, i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[i];
+
+		if (amdgpu_in_reset(adev))
+			vinst->cur_state = AMD_PG_STATE_GATE;
+
+		r = amdgpu_vcn_resume(ip_block->adev, i);
+		if (r)
+			return r;
+	}
+
+	r = vcn_v5_0_1_hw_init(ip_block);
+
+	return r;
+}
+
+/**
+ * vcn_v5_0_1_mc_resume - memory controller programming
+ *
+ * @vinst: VCN instance
+ *
+ * Let the VCN memory controller know it's offsets
+ */
+static void vcn_v5_0_1_mc_resume(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst = vinst->inst;
+	uint32_t offset, size, vcn_inst;
+	const struct common_firmware_header *hdr;
+
+	hdr = (const struct common_firmware_header *)adev->vcn.inst[inst].fw->data;
+	size = AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8);
+
+	vcn_inst = GET_INST(VCN, inst);
+	/* cache window 0: fw */
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		WREG32_SOC15(VCN, vcn_inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst].tmr_mc_addr_lo));
+		WREG32_SOC15(VCN, vcn_inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst].tmr_mc_addr_hi));
+		WREG32_SOC15(VCN, vcn_inst, regUVD_VCPU_CACHE_OFFSET0, 0);
+		offset = 0;
+	} else {
+		WREG32_SOC15(VCN, vcn_inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			lower_32_bits(adev->vcn.inst[inst].gpu_addr));
+		WREG32_SOC15(VCN, vcn_inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			upper_32_bits(adev->vcn.inst[inst].gpu_addr));
+		offset = size;
+		WREG32_SOC15(VCN, vcn_inst, regUVD_VCPU_CACHE_OFFSET0,
+				AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
+	}
+	WREG32_SOC15(VCN, vcn_inst, regUVD_VCPU_CACHE_SIZE0, size);
+
+	/* cache window 1: stack */
+	WREG32_SOC15(VCN, vcn_inst, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW,
+		lower_32_bits(adev->vcn.inst[inst].gpu_addr + offset));
+	WREG32_SOC15(VCN, vcn_inst, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH,
+		upper_32_bits(adev->vcn.inst[inst].gpu_addr + offset));
+	WREG32_SOC15(VCN, vcn_inst, regUVD_VCPU_CACHE_OFFSET1, 0);
+	WREG32_SOC15(VCN, vcn_inst, regUVD_VCPU_CACHE_SIZE1, AMDGPU_VCN_STACK_SIZE);
+
+	/* cache window 2: context */
+	WREG32_SOC15(VCN, vcn_inst, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW,
+		lower_32_bits(adev->vcn.inst[inst].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE));
+	WREG32_SOC15(VCN, vcn_inst, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH,
+		upper_32_bits(adev->vcn.inst[inst].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE));
+	WREG32_SOC15(VCN, vcn_inst, regUVD_VCPU_CACHE_OFFSET2, 0);
+	WREG32_SOC15(VCN, vcn_inst, regUVD_VCPU_CACHE_SIZE2, AMDGPU_VCN_CONTEXT_SIZE);
+
+	/* non-cache window */
+	WREG32_SOC15(VCN, vcn_inst, regUVD_LMI_VCPU_NC0_64BIT_BAR_LOW,
+		lower_32_bits(adev->vcn.inst[inst].fw_shared.gpu_addr));
+	WREG32_SOC15(VCN, vcn_inst, regUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH,
+		upper_32_bits(adev->vcn.inst[inst].fw_shared.gpu_addr));
+	WREG32_SOC15(VCN, vcn_inst, regUVD_VCPU_NONCACHE_OFFSET0, 0);
+	WREG32_SOC15(VCN, vcn_inst, regUVD_VCPU_NONCACHE_SIZE0,
+		AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn5_fw_shared)));
+}
+
+/**
+ * vcn_v5_0_1_mc_resume_dpg_mode - memory controller programming for dpg mode
+ *
+ * @vinst: VCN instance
+ * @indirect: indirectly write sram
+ *
+ * Let the VCN memory controller know it's offsets with dpg mode
+ */
+static void vcn_v5_0_1_mc_resume_dpg_mode(struct amdgpu_vcn_inst *vinst,
+					  bool indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t offset, size;
+	const struct common_firmware_header *hdr;
+
+	hdr = (const struct common_firmware_header *)adev->vcn.inst[inst_idx].fw->data;
+	size = AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8);
+
+	/* cache window 0: fw */
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		if (!indirect) {
+			WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+				VCN, 0, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+				(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN +
+				 inst_idx].tmr_mc_addr_lo), 0, indirect);
+			WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+				VCN, 0, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+				(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN +
+				 inst_idx].tmr_mc_addr_hi), 0, indirect);
+			WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+				VCN, 0, regUVD_VCPU_CACHE_OFFSET0), 0, 0, indirect);
+		} else {
+			WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+				VCN, 0, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW), 0, 0, indirect);
+			WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+				VCN, 0, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH), 0, 0, indirect);
+			WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+				VCN, 0, regUVD_VCPU_CACHE_OFFSET0), 0, 0, indirect);
+		}
+		offset = 0;
+	} else {
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr), 0, indirect);
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr), 0, indirect);
+		offset = size;
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_VCPU_CACHE_OFFSET0),
+			AMDGPU_UVD_FIRMWARE_OFFSET >> 3, 0, indirect);
+	}
+
+	if (!indirect)
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_VCPU_CACHE_SIZE0), size, 0, indirect);
+	else
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_VCPU_CACHE_SIZE0), 0, 0, indirect);
+
+	/* cache window 1: stack */
+	if (!indirect) {
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset), 0, indirect);
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset), 0, indirect);
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_VCPU_CACHE_OFFSET1), 0, 0, indirect);
+	} else {
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW), 0, 0, indirect);
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH), 0, 0, indirect);
+		WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_VCPU_CACHE_OFFSET1), 0, 0, indirect);
+	}
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+			VCN, 0, regUVD_VCPU_CACHE_SIZE1), AMDGPU_VCN_STACK_SIZE, 0, indirect);
+
+	/* cache window 2: context */
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW),
+		lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset +
+			AMDGPU_VCN_STACK_SIZE), 0, indirect);
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH),
+		upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset +
+			AMDGPU_VCN_STACK_SIZE), 0, indirect);
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_VCPU_CACHE_OFFSET2), 0, 0, indirect);
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_VCPU_CACHE_SIZE2), AMDGPU_VCN_CONTEXT_SIZE, 0, indirect);
+
+	/* non-cache window */
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_LMI_VCPU_NC0_64BIT_BAR_LOW),
+		lower_32_bits(adev->vcn.inst[inst_idx].fw_shared.gpu_addr), 0, indirect);
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH),
+		upper_32_bits(adev->vcn.inst[inst_idx].fw_shared.gpu_addr), 0, indirect);
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_VCPU_NONCACHE_OFFSET0), 0, 0, indirect);
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_VCPU_NONCACHE_SIZE0),
+		AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn5_fw_shared)), 0, indirect);
+
+	/* VCN global tiling registers */
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_GFX10_ADDR_CONFIG), adev->gfx.config.gb_addr_config, 0, indirect);
+}
+
+/**
+ * vcn_v5_0_1_disable_clock_gating - disable VCN clock gating
+ *
+ * @vinst: VCN instance
+ *
+ * Disable clock gating for VCN block
+ */
+static void vcn_v5_0_1_disable_clock_gating(struct amdgpu_vcn_inst *vinst)
+{
+}
+
+/**
+ * vcn_v5_0_1_enable_clock_gating - enable VCN clock gating
+ *
+ * @vinst: VCN instance
+ *
+ * Enable clock gating for VCN block
+ */
+static void vcn_v5_0_1_enable_clock_gating(struct amdgpu_vcn_inst *vinst)
+{
+}
+
+/**
+ * vcn_v5_0_1_pause_dpg_mode - VCN pause with dpg mode
+ *
+ * @vinst: VCN instance
+ * @new_state: pause state
+ *
+ * Pause dpg mode for VCN block
+ */
+static int vcn_v5_0_1_pause_dpg_mode(struct amdgpu_vcn_inst *vinst,
+				     struct dpg_pause_state *new_state)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	uint32_t reg_data = 0;
+	int vcn_inst;
+
+	vcn_inst = GET_INST(VCN, vinst->inst);
+
+	/* pause/unpause if state is changed */
+	if (vinst->pause_state.fw_based != new_state->fw_based) {
+		DRM_DEV_DEBUG(adev->dev, "dpg pause state changed %d -> %d %s\n",
+			vinst->pause_state.fw_based, new_state->fw_based,
+			new_state->fw_based ? "VCN_DPG_STATE__PAUSE" : "VCN_DPG_STATE__UNPAUSE");
+		reg_data = RREG32_SOC15(VCN, vcn_inst, regUVD_DPG_PAUSE) &
+			(~UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK);
+
+		if (new_state->fw_based == VCN_DPG_STATE__PAUSE) {
+			/* pause DPG */
+			reg_data |= UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
+			WREG32_SOC15(VCN, vcn_inst, regUVD_DPG_PAUSE, reg_data);
+
+			/* wait for ACK */
+			SOC15_WAIT_ON_RREG(VCN, vcn_inst, regUVD_DPG_PAUSE,
+					UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK,
+					UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK);
+		} else {
+			/* unpause DPG, no need to wait */
+			reg_data &= ~UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
+			WREG32_SOC15(VCN, vcn_inst, regUVD_DPG_PAUSE, reg_data);
+		}
+		vinst->pause_state.fw_based = new_state->fw_based;
+	}
+
+	return 0;
+}
+
+
+/**
+ * vcn_v5_0_1_start_dpg_mode - VCN start with dpg mode
+ *
+ * @vinst: VCN instance
+ * @indirect: indirectly write sram
+ *
+ * Start VCN block with dpg mode
+ */
+static int vcn_v5_0_1_start_dpg_mode(struct amdgpu_vcn_inst *vinst,
+				     bool indirect)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	struct amdgpu_vcn5_fw_shared *fw_shared =
+		adev->vcn.inst[inst_idx].fw_shared.cpu_addr;
+	struct amdgpu_ring *ring;
+	struct dpg_pause_state state = {.fw_based = VCN_DPG_STATE__PAUSE};
+	int vcn_inst, ret;
+	uint32_t tmp;
+
+	vcn_inst = GET_INST(VCN, inst_idx);
+
+	/* disable register anti-hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_POWER_STATUS), 1,
+		~UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+	/* enable dynamic power gating mode */
+	tmp = RREG32_SOC15(VCN, vcn_inst, regUVD_POWER_STATUS);
+	tmp |= UVD_POWER_STATUS__UVD_PG_MODE_MASK;
+	WREG32_SOC15(VCN, vcn_inst, regUVD_POWER_STATUS, tmp);
+
+	if (indirect) {
+		adev->vcn.inst[inst_idx].dpg_sram_curr_addr =
+			(uint32_t *)adev->vcn.inst[inst_idx].dpg_sram_cpu_addr;
+		/* Use dummy register 0xDEADBEEF passing AID selection to PSP FW */
+		WREG32_SOC24_DPG_MODE(inst_idx, 0xDEADBEEF,
+				adev->vcn.inst[inst_idx].aid_id, 0, true);
+	}
+
+	/* enable VCPU clock */
+	tmp = (0xFF << UVD_VCPU_CNTL__PRB_TIMEOUT_VAL__SHIFT);
+	tmp |= UVD_VCPU_CNTL__CLK_EN_MASK | UVD_VCPU_CNTL__BLK_RST_MASK;
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_VCPU_CNTL), tmp, 0, indirect);
+
+	/* disable master interrupt */
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_MASTINT_EN), 0, 0, indirect);
+
+	/* setup regUVD_LMI_CTRL */
+	tmp = (UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+		UVD_LMI_CTRL__REQ_MODE_MASK |
+		UVD_LMI_CTRL__CRC_RESET_MASK |
+		UVD_LMI_CTRL__MASK_MC_URGENT_MASK |
+		UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+		UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
+		(8 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
+		0x00100000L);
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_LMI_CTRL), tmp, 0, indirect);
+
+	vcn_v5_0_1_mc_resume_dpg_mode(vinst, indirect);
+
+	tmp = (0xFF << UVD_VCPU_CNTL__PRB_TIMEOUT_VAL__SHIFT);
+	tmp |= UVD_VCPU_CNTL__CLK_EN_MASK;
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_VCPU_CNTL), tmp, 0, indirect);
+
+	/* enable LMI MC and UMC channels */
+	tmp = 0x1f << UVD_LMI_CTRL2__RE_OFLD_MIF_WR_REQ_NUM__SHIFT;
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_LMI_CTRL2), tmp, 0, indirect);
+
+	/* enable master interrupt */
+	WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+		VCN, 0, regUVD_MASTINT_EN),
+		UVD_MASTINT_EN__VCPU_EN_MASK, 0, indirect);
+
+	if (indirect) {
+		ret = amdgpu_vcn_psp_update_sram(adev, inst_idx, AMDGPU_UCODE_ID_VCN0_RAM);
+		if (ret) {
+			dev_err(adev->dev, "vcn sram load failed %d\n", ret);
+			return ret;
+		}
+	}
+
+	/* resetting ring, fw should not check RB ring */
+	fw_shared->sq.queue_mode |= FW_QUEUE_RING_RESET;
+
+	/* Pause dpg */
+	vcn_v5_0_1_pause_dpg_mode(vinst, &state);
+
+	ring = &adev->vcn.inst[inst_idx].ring_enc[0];
+
+	WREG32_SOC15(VCN, vcn_inst, regUVD_RB_BASE_LO, lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15(VCN, vcn_inst, regUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(VCN, vcn_inst, regUVD_RB_SIZE, ring->ring_size / sizeof(uint32_t));
+
+	tmp = RREG32_SOC15(VCN, vcn_inst, regVCN_RB_ENABLE);
+	tmp &= ~(VCN_RB_ENABLE__RB1_EN_MASK);
+	WREG32_SOC15(VCN, vcn_inst, regVCN_RB_ENABLE, tmp);
+
+	WREG32_SOC15(VCN, vcn_inst, regUVD_RB_RPTR, 0);
+	WREG32_SOC15(VCN, vcn_inst, regUVD_RB_WPTR, 0);
+
+	tmp = RREG32_SOC15(VCN, vcn_inst, regUVD_RB_RPTR);
+	WREG32_SOC15(VCN, vcn_inst, regUVD_RB_WPTR, tmp);
+	ring->wptr = RREG32_SOC15(VCN, vcn_inst, regUVD_RB_WPTR);
+
+	tmp = RREG32_SOC15(VCN, vcn_inst, regVCN_RB_ENABLE);
+	tmp |= VCN_RB_ENABLE__RB1_EN_MASK;
+	WREG32_SOC15(VCN, vcn_inst, regVCN_RB_ENABLE, tmp);
+	/* resetting done, fw can check RB ring */
+	fw_shared->sq.queue_mode &= ~(FW_QUEUE_RING_RESET | FW_QUEUE_DPG_HOLD_OFF);
+
+	WREG32_SOC15(VCN, vcn_inst, regVCN_RB1_DB_CTRL,
+		ring->doorbell_index << VCN_RB1_DB_CTRL__OFFSET__SHIFT |
+		VCN_RB1_DB_CTRL__EN_MASK);
+	/* Read DB_CTRL to flush the write DB_CTRL command. */
+	RREG32_SOC15(VCN, vcn_inst, regVCN_RB1_DB_CTRL);
+
+	return 0;
+}
+
+static int vcn_v5_0_1_start_sriov(struct amdgpu_device *adev)
+{
+	int i, vcn_inst;
+	struct amdgpu_ring *ring_enc;
+	uint64_t cache_addr;
+	uint64_t rb_enc_addr;
+	uint64_t ctx_addr;
+	uint32_t param, resp, expected;
+	uint32_t offset, cache_size;
+	uint32_t tmp, timeout;
+
+	struct amdgpu_mm_table *table = &adev->virt.mm_table;
+	uint32_t *table_loc;
+	uint32_t table_size;
+	uint32_t size, size_dw;
+	uint32_t init_status;
+	uint32_t enabled_vcn;
+
+	struct mmsch_v5_0_cmd_direct_write
+		direct_wt = { {0} };
+	struct mmsch_v5_0_cmd_direct_read_modify_write
+		direct_rd_mod_wt = { {0} };
+	struct mmsch_v5_0_cmd_end end = { {0} };
+	struct mmsch_v5_0_init_header header;
+
+	struct amdgpu_vcn5_fw_shared *fw_shared;
+	struct amdgpu_fw_shared_rb_setup *rb_setup;
+
+	direct_wt.cmd_header.command_type =
+		MMSCH_COMMAND__DIRECT_REG_WRITE;
+	direct_rd_mod_wt.cmd_header.command_type =
+		MMSCH_COMMAND__DIRECT_REG_READ_MODIFY_WRITE;
+	end.cmd_header.command_type = MMSCH_COMMAND__END;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		vcn_inst = GET_INST(VCN, i);
+
+		vcn_v5_0_1_fw_shared_init(adev, vcn_inst);
+
+		memset(&header, 0, sizeof(struct mmsch_v5_0_init_header));
+		header.version = MMSCH_VERSION;
+		header.total_size = sizeof(struct mmsch_v5_0_init_header) >> 2;
+
+		table_loc = (uint32_t *)table->cpu_addr;
+		table_loc += header.total_size;
+
+		table_size = 0;
+
+		MMSCH_V5_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(VCN, 0, regUVD_STATUS),
+			~UVD_STATUS__UVD_BUSY, UVD_STATUS__UVD_BUSY);
+
+		cache_size = AMDGPU_GPU_PAGE_ALIGN(adev->vcn.inst[i].fw->size + 4);
+
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+			MMSCH_V5_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+				regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+				adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + i].tmr_mc_addr_lo);
+
+			MMSCH_V5_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+				regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+				adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + i].tmr_mc_addr_hi);
+
+			offset = 0;
+			MMSCH_V5_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+				regUVD_VCPU_CACHE_OFFSET0), 0);
+		} else {
+			MMSCH_V5_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+				regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+				lower_32_bits(adev->vcn.inst[i].gpu_addr));
+			MMSCH_V5_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+				regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+				upper_32_bits(adev->vcn.inst[i].gpu_addr));
+			offset = cache_size;
+			MMSCH_V5_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+				regUVD_VCPU_CACHE_OFFSET0),
+				AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
+		}
+
+		MMSCH_V5_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_VCPU_CACHE_SIZE0),
+			cache_size);
+
+		cache_addr = adev->vcn.inst[vcn_inst].gpu_addr + offset;
+		MMSCH_V5_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW), lower_32_bits(cache_addr));
+		MMSCH_V5_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH), upper_32_bits(cache_addr));
+		MMSCH_V5_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_VCPU_CACHE_OFFSET1), 0);
+		MMSCH_V5_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_VCPU_CACHE_SIZE1), AMDGPU_VCN_STACK_SIZE);
+
+		cache_addr = adev->vcn.inst[vcn_inst].gpu_addr + offset +
+			AMDGPU_VCN_STACK_SIZE;
+
+		MMSCH_V5_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW), lower_32_bits(cache_addr));
+
+		MMSCH_V5_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH), upper_32_bits(cache_addr));
+
+		MMSCH_V5_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_VCPU_CACHE_OFFSET2), 0);
+
+		MMSCH_V5_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_VCPU_CACHE_SIZE2), AMDGPU_VCN_CONTEXT_SIZE);
+
+		fw_shared = adev->vcn.inst[vcn_inst].fw_shared.cpu_addr;
+		rb_setup = &fw_shared->rb_setup;
+
+		ring_enc = &adev->vcn.inst[vcn_inst].ring_enc[0];
+		ring_enc->wptr = 0;
+		rb_enc_addr = ring_enc->gpu_addr;
+
+		rb_setup->is_rb_enabled_flags |= RB_ENABLED;
+		rb_setup->rb_addr_lo = lower_32_bits(rb_enc_addr);
+		rb_setup->rb_addr_hi = upper_32_bits(rb_enc_addr);
+		rb_setup->rb_size = ring_enc->ring_size / 4;
+		fw_shared->present_flag_0 |= cpu_to_le32(AMDGPU_VCN_VF_RB_SETUP_FLAG);
+
+		MMSCH_V5_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_LMI_VCPU_NC0_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[vcn_inst].fw_shared.gpu_addr));
+		MMSCH_V5_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[vcn_inst].fw_shared.gpu_addr));
+		MMSCH_V5_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0,
+			regUVD_VCPU_NONCACHE_SIZE0),
+			AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn4_fw_shared)));
+		MMSCH_V5_0_INSERT_END();
+
+		header.vcn0.init_status = 0;
+		header.vcn0.table_offset = header.total_size;
+		header.vcn0.table_size = table_size;
+		header.total_size += table_size;
+
+		/* Send init table to mmsch */
+		size = sizeof(struct mmsch_v5_0_init_header);
+		table_loc = (uint32_t *)table->cpu_addr;
+		memcpy((void *)table_loc, &header, size);
+
+		ctx_addr = table->gpu_addr;
+		WREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_CTX_ADDR_LO, lower_32_bits(ctx_addr));
+		WREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_CTX_ADDR_HI, upper_32_bits(ctx_addr));
+
+		tmp = RREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_VMID);
+		tmp &= ~MMSCH_VF_VMID__VF_CTX_VMID_MASK;
+		tmp |= (0 << MMSCH_VF_VMID__VF_CTX_VMID__SHIFT);
+		WREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_VMID, tmp);
+
+		size = header.total_size;
+		WREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_CTX_SIZE, size);
+
+		WREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_MAILBOX_RESP, 0);
+
+		param = 0x00000001;
+		WREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_MAILBOX_HOST, param);
+		tmp = 0;
+		timeout = 1000;
+		resp = 0;
+		expected = MMSCH_VF_MAILBOX_RESP__OK;
+		while (resp != expected) {
+			resp = RREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_MAILBOX_RESP);
+			if (resp != 0)
+				break;
+
+			udelay(10);
+			tmp = tmp + 10;
+			if (tmp >= timeout) {
+				DRM_ERROR("failed to init MMSCH. TIME-OUT after %d usec"\
+					" waiting for regMMSCH_VF_MAILBOX_RESP "\
+					"(expected=0x%08x, readback=0x%08x)\n",
+					tmp, expected, resp);
+				return -EBUSY;
+			}
+		}
+
+		enabled_vcn = amdgpu_vcn_is_disabled_vcn(adev, VCN_DECODE_RING, 0) ? 1 : 0;
+		init_status = ((struct mmsch_v5_0_init_header *)(table_loc))->vcn0.init_status;
+		if (resp != expected && resp != MMSCH_VF_MAILBOX_RESP__INCOMPLETE
+					&& init_status != MMSCH_VF_ENGINE_STATUS__PASS) {
+			DRM_ERROR("MMSCH init status is incorrect! readback=0x%08x, header init "\
+				"status for VCN%x: 0x%x\n", resp, enabled_vcn, init_status);
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * vcn_v5_0_1_start - VCN start
+ *
+ * @vinst: VCN instance
+ *
+ * Start VCN block
+ */
+static int vcn_v5_0_1_start(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int i = vinst->inst;
+	struct amdgpu_vcn5_fw_shared *fw_shared;
+	struct amdgpu_ring *ring;
+	uint32_t tmp;
+	int j, k, r, vcn_inst;
+
+	fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)
+		return vcn_v5_0_1_start_dpg_mode(vinst, adev->vcn.inst[i].indirect_sram);
+
+	vcn_inst = GET_INST(VCN, i);
+
+	/* set VCN status busy */
+	tmp = RREG32_SOC15(VCN, vcn_inst, regUVD_STATUS) | UVD_STATUS__UVD_BUSY;
+	WREG32_SOC15(VCN, vcn_inst, regUVD_STATUS, tmp);
+
+	/* enable VCPU clock */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_VCPU_CNTL),
+		 UVD_VCPU_CNTL__CLK_EN_MASK, ~UVD_VCPU_CNTL__CLK_EN_MASK);
+
+	/* disable master interrupt */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_MASTINT_EN), 0,
+		 ~UVD_MASTINT_EN__VCPU_EN_MASK);
+
+	/* enable LMI MC and UMC channels */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_LMI_CTRL2), 0,
+		 ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+
+	tmp = RREG32_SOC15(VCN, vcn_inst, regUVD_SOFT_RESET);
+	tmp &= ~UVD_SOFT_RESET__LMI_SOFT_RESET_MASK;
+	tmp &= ~UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK;
+	WREG32_SOC15(VCN, vcn_inst, regUVD_SOFT_RESET, tmp);
+
+	/* setup regUVD_LMI_CTRL */
+	tmp = RREG32_SOC15(VCN, vcn_inst, regUVD_LMI_CTRL);
+	WREG32_SOC15(VCN, vcn_inst, regUVD_LMI_CTRL, tmp |
+		     UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+		     UVD_LMI_CTRL__MASK_MC_URGENT_MASK |
+		     UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+		     UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK);
+
+	vcn_v5_0_1_mc_resume(vinst);
+
+	/* VCN global tiling registers */
+	WREG32_SOC15(VCN, vcn_inst, regUVD_GFX10_ADDR_CONFIG,
+		     adev->gfx.config.gb_addr_config);
+
+	/* unblock VCPU register access */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_RB_ARB_CTRL), 0,
+		 ~UVD_RB_ARB_CTRL__VCPU_DIS_MASK);
+
+	/* release VCPU reset to boot */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_VCPU_CNTL), 0,
+		 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+	for (j = 0; j < 10; ++j) {
+		uint32_t status;
+
+		for (k = 0; k < 100; ++k) {
+			status = RREG32_SOC15(VCN, vcn_inst, regUVD_STATUS);
+			if (status & 2)
+				break;
+			mdelay(100);
+			if (amdgpu_emu_mode == 1)
+				msleep(20);
+		}
+
+		if (amdgpu_emu_mode == 1) {
+			r = -1;
+			if (status & 2) {
+				r = 0;
+				break;
+			}
+		} else {
+			r = 0;
+			if (status & 2)
+				break;
+
+			dev_err(adev->dev,
+				"VCN[%d] is not responding, trying to reset the VCPU!!!\n", i);
+			WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_VCPU_CNTL),
+				 UVD_VCPU_CNTL__BLK_RST_MASK,
+				 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+			mdelay(10);
+			WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_VCPU_CNTL), 0,
+				 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+			mdelay(10);
+			r = -1;
+		}
+	}
+
+	if (r) {
+		dev_err(adev->dev, "VCN[%d] is not responding, giving up!!!\n", i);
+		return r;
+	}
+
+	/* enable master interrupt */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_MASTINT_EN),
+		 UVD_MASTINT_EN__VCPU_EN_MASK,
+		 ~UVD_MASTINT_EN__VCPU_EN_MASK);
+
+	/* clear the busy bit of VCN_STATUS */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_STATUS), 0,
+		 ~(2 << UVD_STATUS__VCPU_REPORT__SHIFT));
+
+	ring = &adev->vcn.inst[i].ring_enc[0];
+
+	WREG32_SOC15(VCN, vcn_inst, regVCN_RB1_DB_CTRL,
+		     ring->doorbell_index << VCN_RB1_DB_CTRL__OFFSET__SHIFT |
+		     VCN_RB1_DB_CTRL__EN_MASK);
+
+	/* Read DB_CTRL to flush the write DB_CTRL command. */
+	RREG32_SOC15(VCN, vcn_inst, regVCN_RB1_DB_CTRL);
+
+	WREG32_SOC15(VCN, vcn_inst, regUVD_RB_BASE_LO, ring->gpu_addr);
+	WREG32_SOC15(VCN, vcn_inst, regUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(VCN, vcn_inst, regUVD_RB_SIZE, ring->ring_size / 4);
+
+	tmp = RREG32_SOC15(VCN, vcn_inst, regVCN_RB_ENABLE);
+	tmp &= ~(VCN_RB_ENABLE__RB1_EN_MASK);
+	WREG32_SOC15(VCN, vcn_inst, regVCN_RB_ENABLE, tmp);
+	fw_shared->sq.queue_mode |= FW_QUEUE_RING_RESET;
+	WREG32_SOC15(VCN, vcn_inst, regUVD_RB_RPTR, 0);
+	WREG32_SOC15(VCN, vcn_inst, regUVD_RB_WPTR, 0);
+
+	tmp = RREG32_SOC15(VCN, vcn_inst, regUVD_RB_RPTR);
+	WREG32_SOC15(VCN, vcn_inst, regUVD_RB_WPTR, tmp);
+	ring->wptr = RREG32_SOC15(VCN, vcn_inst, regUVD_RB_WPTR);
+
+	tmp = RREG32_SOC15(VCN, vcn_inst, regVCN_RB_ENABLE);
+	tmp |= VCN_RB_ENABLE__RB1_EN_MASK;
+	WREG32_SOC15(VCN, vcn_inst, regVCN_RB_ENABLE, tmp);
+	fw_shared->sq.queue_mode &= ~(FW_QUEUE_RING_RESET | FW_QUEUE_DPG_HOLD_OFF);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, vcn_inst, regUVD_STATUS);
+
+	return 0;
+}
+
+/**
+ * vcn_v5_0_1_stop_dpg_mode - VCN stop with dpg mode
+ *
+ * @vinst: VCN instance
+ *
+ * Stop VCN block with dpg mode
+ */
+static void vcn_v5_0_1_stop_dpg_mode(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int inst_idx = vinst->inst;
+	uint32_t tmp;
+	int vcn_inst;
+	struct dpg_pause_state state = {.fw_based = VCN_DPG_STATE__UNPAUSE};
+
+	vcn_inst = GET_INST(VCN, inst_idx);
+
+	/* Unpause dpg */
+	vcn_v5_0_1_pause_dpg_mode(vinst, &state);
+
+	/* Wait for power status to be 1 */
+	SOC15_WAIT_ON_RREG(VCN, vcn_inst, regUVD_POWER_STATUS, 1,
+		UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+	/* wait for read ptr to be equal to write ptr */
+	tmp = RREG32_SOC15(VCN, vcn_inst, regUVD_RB_WPTR);
+	SOC15_WAIT_ON_RREG(VCN, vcn_inst, regUVD_RB_RPTR, tmp, 0xFFFFFFFF);
+
+	/* disable dynamic power gating mode */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_POWER_STATUS), 0,
+		~UVD_POWER_STATUS__UVD_PG_MODE_MASK);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, vcn_inst, regUVD_STATUS);
+}
+
+/**
+ * vcn_v5_0_1_stop - VCN stop
+ *
+ * @vinst: VCN instance
+ *
+ * Stop VCN block
+ */
+static int vcn_v5_0_1_stop(struct amdgpu_vcn_inst *vinst)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int i = vinst->inst;
+	struct amdgpu_vcn5_fw_shared *fw_shared;
+	uint32_t tmp;
+	int r = 0, vcn_inst;
+
+	vcn_inst = GET_INST(VCN, i);
+
+	fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+	fw_shared->sq.queue_mode |= FW_QUEUE_DPG_HOLD_OFF;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) {
+		vcn_v5_0_1_stop_dpg_mode(vinst);
+		return 0;
+	}
+
+	/* wait for vcn idle */
+	r = SOC15_WAIT_ON_RREG(VCN, vcn_inst, regUVD_STATUS, UVD_STATUS__IDLE, 0x7);
+	if (r)
+		return r;
+
+	tmp = UVD_LMI_STATUS__VCPU_LMI_WRITE_CLEAN_MASK |
+		UVD_LMI_STATUS__READ_CLEAN_MASK |
+		UVD_LMI_STATUS__WRITE_CLEAN_MASK |
+		UVD_LMI_STATUS__WRITE_CLEAN_RAW_MASK;
+	r = SOC15_WAIT_ON_RREG(VCN, vcn_inst, regUVD_LMI_STATUS, tmp, tmp);
+	if (r)
+		return r;
+
+	/* disable LMI UMC channel */
+	tmp = RREG32_SOC15(VCN, vcn_inst, regUVD_LMI_CTRL2);
+	tmp |= UVD_LMI_CTRL2__STALL_ARB_UMC_MASK;
+	WREG32_SOC15(VCN, vcn_inst, regUVD_LMI_CTRL2, tmp);
+	tmp = UVD_LMI_STATUS__UMC_READ_CLEAN_RAW_MASK |
+		UVD_LMI_STATUS__UMC_WRITE_CLEAN_RAW_MASK;
+	r = SOC15_WAIT_ON_RREG(VCN, vcn_inst, regUVD_LMI_STATUS, tmp, tmp);
+	if (r)
+		return r;
+
+	/* block VCPU register access */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_RB_ARB_CTRL),
+		 UVD_RB_ARB_CTRL__VCPU_DIS_MASK,
+		 ~UVD_RB_ARB_CTRL__VCPU_DIS_MASK);
+
+	/* reset VCPU */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_VCPU_CNTL),
+		 UVD_VCPU_CNTL__BLK_RST_MASK,
+		 ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+	/* disable VCPU clock */
+	WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_VCPU_CNTL), 0,
+		 ~(UVD_VCPU_CNTL__CLK_EN_MASK));
+
+	/* apply soft reset */
+	tmp = RREG32_SOC15(VCN, vcn_inst, regUVD_SOFT_RESET);
+	tmp |= UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK;
+	WREG32_SOC15(VCN, vcn_inst, regUVD_SOFT_RESET, tmp);
+	tmp = RREG32_SOC15(VCN, vcn_inst, regUVD_SOFT_RESET);
+	tmp |= UVD_SOFT_RESET__LMI_SOFT_RESET_MASK;
+	WREG32_SOC15(VCN, vcn_inst, regUVD_SOFT_RESET, tmp);
+
+	/* clear status */
+	WREG32_SOC15(VCN, vcn_inst, regUVD_STATUS, 0);
+
+	/* Keeping one read-back to ensure all register writes are done,
+	 * otherwise it may introduce race conditions.
+	 */
+	RREG32_SOC15(VCN, vcn_inst, regUVD_STATUS);
+
+	return 0;
+}
+
+/**
+ * vcn_v5_0_1_unified_ring_get_rptr - get unified read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware unified read pointer
+ */
+static uint64_t vcn_v5_0_1_unified_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring != &adev->vcn.inst[ring->me].ring_enc[0])
+		DRM_ERROR("wrong ring id is identified in %s", __func__);
+
+	return RREG32_SOC15(VCN, GET_INST(VCN, ring->me), regUVD_RB_RPTR);
+}
+
+/**
+ * vcn_v5_0_1_unified_ring_get_wptr - get unified write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware unified write pointer
+ */
+static uint64_t vcn_v5_0_1_unified_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring != &adev->vcn.inst[ring->me].ring_enc[0])
+		DRM_ERROR("wrong ring id is identified in %s", __func__);
+
+	if (ring->use_doorbell)
+		return *ring->wptr_cpu_addr;
+	else
+		return RREG32_SOC15(VCN, GET_INST(VCN, ring->me), regUVD_RB_WPTR);
+}
+
+/**
+ * vcn_v5_0_1_unified_ring_set_wptr - set enc write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the enc write pointer to the hardware
+ */
+static void vcn_v5_0_1_unified_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring != &adev->vcn.inst[ring->me].ring_enc[0])
+		DRM_ERROR("wrong ring id is identified in %s", __func__);
+
+	if (ring->use_doorbell) {
+		*ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
+		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+	} else {
+		WREG32_SOC15(VCN, GET_INST(VCN, ring->me), regUVD_RB_WPTR,
+				lower_32_bits(ring->wptr));
+	}
+}
+
+static int vcn_v5_0_1_ring_reset(struct amdgpu_ring *ring,
+				 unsigned int vmid,
+				 struct amdgpu_fence *timedout_fence)
+{
+	int r = 0;
+	int vcn_inst;
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[ring->me];
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+
+	vcn_inst = GET_INST(VCN, ring->me);
+	r = amdgpu_dpm_reset_vcn(adev, 1 << vcn_inst);
+
+	if (r) {
+		DRM_DEV_ERROR(adev->dev, "VCN reset fail : %d\n", r);
+		return r;
+	}
+
+	vcn_v5_0_1_hw_init_inst(adev, ring->me);
+	vcn_v5_0_1_start_dpg_mode(vinst, vinst->indirect_sram);
+
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static const struct amdgpu_ring_funcs vcn_v5_0_1_unified_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_ENC,
+	.align_mask = 0x3f,
+	.nop = VCN_ENC_CMD_NO_OP,
+	.get_rptr = vcn_v5_0_1_unified_ring_get_rptr,
+	.get_wptr = vcn_v5_0_1_unified_ring_get_wptr,
+	.set_wptr = vcn_v5_0_1_unified_ring_set_wptr,
+	.emit_frame_size = SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+			   SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 4 +
+			   4 + /* vcn_v2_0_enc_ring_emit_vm_flush */
+			   5 +
+			   5 + /* vcn_v2_0_enc_ring_emit_fence x2 vm fence */
+			   1, /* vcn_v2_0_enc_ring_insert_end */
+	.emit_ib_size = 5, /* vcn_v2_0_enc_ring_emit_ib */
+	.emit_ib = vcn_v2_0_enc_ring_emit_ib,
+	.emit_fence = vcn_v2_0_enc_ring_emit_fence,
+	.emit_vm_flush = vcn_v4_0_3_enc_ring_emit_vm_flush,
+	.emit_hdp_flush = vcn_v4_0_3_ring_emit_hdp_flush,
+	.test_ring = amdgpu_vcn_enc_ring_test_ring,
+	.test_ib = amdgpu_vcn_unified_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.insert_end = vcn_v2_0_enc_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_vcn_ring_begin_use,
+	.end_use = amdgpu_vcn_ring_end_use,
+	.emit_wreg = vcn_v4_0_3_enc_ring_emit_wreg,
+	.emit_reg_wait = vcn_v4_0_3_enc_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = vcn_v5_0_1_ring_reset,
+};
+
+/**
+ * vcn_v5_0_1_set_unified_ring_funcs - set unified ring functions
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set unified ring functions
+ */
+static void vcn_v5_0_1_set_unified_ring_funcs(struct amdgpu_device *adev)
+{
+	int i, vcn_inst;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		adev->vcn.inst[i].ring_enc[0].funcs = &vcn_v5_0_1_unified_ring_vm_funcs;
+		adev->vcn.inst[i].ring_enc[0].me = i;
+		vcn_inst = GET_INST(VCN, i);
+		adev->vcn.inst[i].aid_id = vcn_inst / adev->vcn.num_inst_per_aid;
+	}
+}
+
+/**
+ * vcn_v5_0_1_is_idle - check VCN block is idle
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block structure
+ *
+ * Check whether VCN block is idle
+ */
+static bool vcn_v5_0_1_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, ret = 1;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i)
+		ret &= (RREG32_SOC15(VCN, GET_INST(VCN, i), regUVD_STATUS) == UVD_STATUS__IDLE);
+
+	return ret;
+}
+
+/**
+ * vcn_v5_0_1_wait_for_idle - wait for VCN block idle
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Wait for VCN block idle
+ */
+static int vcn_v5_0_1_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, ret = 0;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		ret = SOC15_WAIT_ON_RREG(VCN, GET_INST(VCN, i), regUVD_STATUS, UVD_STATUS__IDLE,
+			UVD_STATUS__IDLE);
+		if (ret)
+			return ret;
+	}
+
+	return ret;
+}
+
+/**
+ * vcn_v5_0_1_set_clockgating_state - set VCN block clockgating state
+ *
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ * @state: clock gating state
+ *
+ * Set VCN block clockgating state
+ */
+static int vcn_v5_0_1_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					    enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool enable = state == AMD_CG_STATE_GATE;
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[i];
+
+		if (enable) {
+			if (RREG32_SOC15(VCN, GET_INST(VCN, i), regUVD_STATUS) != UVD_STATUS__IDLE)
+				return -EBUSY;
+			vcn_v5_0_1_enable_clock_gating(vinst);
+		} else {
+			vcn_v5_0_1_disable_clock_gating(vinst);
+		}
+	}
+
+	return 0;
+}
+
+static int vcn_v5_0_1_set_pg_state(struct amdgpu_vcn_inst *vinst,
+				   enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = vinst->adev;
+	int ret = 0;
+
+	/* for SRIOV, guest should not control VCN Power-gating
+	 * MMSCH FW should control Power-gating and clock-gating
+	 * guest should avoid touching CGC and PG
+	 */
+	if (amdgpu_sriov_vf(adev)) {
+		vinst->cur_state = AMD_PG_STATE_UNGATE;
+		return 0;
+	}
+
+	if (state == vinst->cur_state)
+		return 0;
+
+	if (state == AMD_PG_STATE_GATE)
+		ret = vcn_v5_0_1_stop(vinst);
+	else
+		ret = vcn_v5_0_1_start(vinst);
+
+	if (!ret)
+		vinst->cur_state = state;
+
+	return ret;
+}
+
+/**
+ * vcn_v5_0_1_process_interrupt - process VCN block interrupt
+ *
+ * @adev: amdgpu_device pointer
+ * @source: interrupt sources
+ * @entry: interrupt entry from clients and sources
+ *
+ * Process VCN block interrupt
+ */
+static int vcn_v5_0_1_process_interrupt(struct amdgpu_device *adev, struct amdgpu_irq_src *source,
+	struct amdgpu_iv_entry *entry)
+{
+	uint32_t i, inst;
+
+	i = node_id_to_phys_map[entry->node_id];
+
+	DRM_DEV_DEBUG(adev->dev, "IH: VCN TRAP\n");
+
+	for (inst = 0; inst < adev->vcn.num_vcn_inst; ++inst)
+		if (adev->vcn.inst[inst].aid_id == i)
+			break;
+	if (inst >= adev->vcn.num_vcn_inst) {
+		dev_WARN_ONCE(adev->dev, 1,
+				"Interrupt received for unknown VCN instance %d",
+				entry->node_id);
+		return 0;
+	}
+
+	switch (entry->src_id) {
+	case VCN_5_0__SRCID__UVD_ENC_GENERAL_PURPOSE:
+		amdgpu_fence_process(&adev->vcn.inst[inst].ring_enc[0]);
+		break;
+	default:
+		DRM_DEV_ERROR(adev->dev, "Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+static int vcn_v5_0_1_set_ras_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned int type,
+					enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs vcn_v5_0_1_irq_funcs = {
+	.process = vcn_v5_0_1_process_interrupt,
+};
+
+static const struct amdgpu_irq_src_funcs vcn_v5_0_1_ras_irq_funcs = {
+	.set = vcn_v5_0_1_set_ras_interrupt_state,
+	.process = amdgpu_vcn_process_poison_irq,
+};
+
+
+/**
+ * vcn_v5_0_1_set_irq_funcs - set VCN block interrupt irq functions
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set VCN block interrupt irq functions
+ */
+static void vcn_v5_0_1_set_irq_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i)
+		adev->vcn.inst->irq.num_types++;
+
+	adev->vcn.inst->irq.funcs = &vcn_v5_0_1_irq_funcs;
+
+	adev->vcn.inst->ras_poison_irq.num_types = 1;
+	adev->vcn.inst->ras_poison_irq.funcs = &vcn_v5_0_1_ras_irq_funcs;
+
+}
+
+static const struct amd_ip_funcs vcn_v5_0_1_ip_funcs = {
+	.name = "vcn_v5_0_1",
+	.early_init = vcn_v5_0_1_early_init,
+	.late_init = vcn_v5_0_1_late_init,
+	.sw_init = vcn_v5_0_1_sw_init,
+	.sw_fini = vcn_v5_0_1_sw_fini,
+	.hw_init = vcn_v5_0_1_hw_init,
+	.hw_fini = vcn_v5_0_1_hw_fini,
+	.suspend = vcn_v5_0_1_suspend,
+	.resume = vcn_v5_0_1_resume,
+	.is_idle = vcn_v5_0_1_is_idle,
+	.wait_for_idle = vcn_v5_0_1_wait_for_idle,
+	.check_soft_reset = NULL,
+	.pre_soft_reset = NULL,
+	.soft_reset = NULL,
+	.post_soft_reset = NULL,
+	.set_clockgating_state = vcn_v5_0_1_set_clockgating_state,
+	.set_powergating_state = vcn_set_powergating_state,
+	.dump_ip_state = amdgpu_vcn_dump_ip_state,
+	.print_ip_state = amdgpu_vcn_print_ip_state,
+};
+
+const struct amdgpu_ip_block_version vcn_v5_0_1_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_VCN,
+	.major = 5,
+	.minor = 0,
+	.rev = 1,
+	.funcs = &vcn_v5_0_1_ip_funcs,
+};
+
+static uint32_t vcn_v5_0_1_query_poison_by_instance(struct amdgpu_device *adev,
+			uint32_t instance, uint32_t sub_block)
+{
+	uint32_t poison_stat = 0, reg_value = 0;
+
+	switch (sub_block) {
+	case AMDGPU_VCN_V5_0_1_VCPU_VCODEC:
+		reg_value = RREG32_SOC15(VCN, instance, regUVD_RAS_VCPU_VCODEC_STATUS);
+		poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_VCPU_VCODEC_STATUS, POISONED_PF);
+		break;
+	default:
+		break;
+	}
+
+	if (poison_stat)
+		dev_info(adev->dev, "Poison detected in VCN%d, sub_block%d\n",
+			instance, sub_block);
+
+	return poison_stat;
+}
+
+static bool vcn_v5_0_1_query_poison_status(struct amdgpu_device *adev)
+{
+	uint32_t inst, sub;
+	uint32_t poison_stat = 0;
+
+	for (inst = 0; inst < adev->vcn.num_vcn_inst; inst++)
+		for (sub = 0; sub < AMDGPU_VCN_V5_0_1_MAX_SUB_BLOCK; sub++)
+			poison_stat +=
+			vcn_v5_0_1_query_poison_by_instance(adev, inst, sub);
+
+	return !!poison_stat;
+}
+
+static const struct amdgpu_ras_block_hw_ops vcn_v5_0_1_ras_hw_ops = {
+	.query_poison_status = vcn_v5_0_1_query_poison_status,
+};
+
+static int vcn_v5_0_1_aca_bank_parser(struct aca_handle *handle, struct aca_bank *bank,
+				      enum aca_smu_type type, void *data)
+{
+	struct aca_bank_info info;
+	u64 misc0;
+	int ret;
+
+	ret = aca_bank_info_decode(bank, &info);
+	if (ret)
+		return ret;
+
+	misc0 = bank->regs[ACA_REG_IDX_MISC0];
+	switch (type) {
+	case ACA_SMU_TYPE_UE:
+		bank->aca_err_type = ACA_ERROR_TYPE_UE;
+		ret = aca_error_cache_log_bank_error(handle, &info, ACA_ERROR_TYPE_UE,
+						     1ULL);
+		break;
+	case ACA_SMU_TYPE_CE:
+		bank->aca_err_type = ACA_ERROR_TYPE_CE;
+		ret = aca_error_cache_log_bank_error(handle, &info, bank->aca_err_type,
+						     ACA_REG__MISC0__ERRCNT(misc0));
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return ret;
+}
+
+/* reference to smu driver if header file */
+static int vcn_v5_0_1_err_codes[] = {
+	14, 15, 47, /* VCN [D|V|S] */
+};
+
+static bool vcn_v5_0_1_aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank,
+					 enum aca_smu_type type, void *data)
+{
+	u32 instlo;
+
+	instlo = ACA_REG__IPID__INSTANCEIDLO(bank->regs[ACA_REG_IDX_IPID]);
+	instlo &= GENMASK(31, 1);
+
+	if (instlo != mmSMNAID_AID0_MCA_SMU)
+		return false;
+
+	if (aca_bank_check_error_codes(handle->adev, bank,
+				       vcn_v5_0_1_err_codes,
+				       ARRAY_SIZE(vcn_v5_0_1_err_codes)))
+		return false;
+
+	return true;
+}
+
+static const struct aca_bank_ops vcn_v5_0_1_aca_bank_ops = {
+	.aca_bank_parser = vcn_v5_0_1_aca_bank_parser,
+	.aca_bank_is_valid = vcn_v5_0_1_aca_bank_is_valid,
+};
+
+static const struct aca_info vcn_v5_0_1_aca_info = {
+	.hwip = ACA_HWIP_TYPE_SMU,
+	.mask = ACA_ERROR_UE_MASK,
+	.bank_ops = &vcn_v5_0_1_aca_bank_ops,
+};
+
+static int vcn_v5_0_1_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
+{
+	int r;
+
+	r = amdgpu_ras_block_late_init(adev, ras_block);
+	if (r)
+		return r;
+
+	r = amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__VCN,
+				&vcn_v5_0_1_aca_info, NULL);
+	if (r)
+		goto late_fini;
+
+	if (amdgpu_ras_is_supported(adev, ras_block->block) &&
+		adev->vcn.inst->ras_poison_irq.funcs) {
+		r = amdgpu_irq_get(adev, &adev->vcn.inst->ras_poison_irq, 0);
+		if (r)
+			goto late_fini;
+	}
+
+	return 0;
+
+late_fini:
+	amdgpu_ras_block_late_fini(adev, ras_block);
+
+	return r;
+}
+
+static struct amdgpu_vcn_ras vcn_v5_0_1_ras = {
+	.ras_block = {
+		.hw_ops = &vcn_v5_0_1_ras_hw_ops,
+		.ras_late_init = vcn_v5_0_1_ras_late_init,
+	},
+};
+
+static void vcn_v5_0_1_set_ras_funcs(struct amdgpu_device *adev)
+{
+	adev->vcn.ras = &vcn_v5_0_1_ras;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.h b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.h
new file mode 100644
index 000000000000..b72e4da68317
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.h
@@ -0,0 +1,39 @@
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc. All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VCN_v5_0_1_H__
+#define __VCN_v5_0_1_H__
+
+#define regVCN_RRMT_CNTL                          0x0940
+#define regVCN_RRMT_CNTL_BASE_IDX                 1
+
+
+enum amdgpu_vcn_v5_0_1_sub_block {
+	AMDGPU_VCN_V5_0_1_VCPU_VCODEC = 0,
+
+	AMDGPU_VCN_V5_0_1_MAX_SUB_BLOCK,
+};
+
+extern const struct amdgpu_ip_block_version vcn_v5_0_1_ip_block;
+
+#endif /* __VCN_v5_0_1_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/vega10_ih.c b/drivers/gpu/drm/amd/amdgpu/vega10_ih.c
index 56150e8d1ed2..eb16916c6473 100644
--- a/drivers/gpu/drm/amd/amdgpu/vega10_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/vega10_ih.c
@@ -20,157 +20,290 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  *
  */
-#include <drm/drmP.h>
+
+#include <linux/pci.h>
+
 #include "amdgpu.h"
 #include "amdgpu_ih.h"
 #include "soc15.h"
 
-
-#include "vega10/soc15ip.h"
-#include "vega10/OSSSYS/osssys_4_0_offset.h"
-#include "vega10/OSSSYS/osssys_4_0_sh_mask.h"
+#include "oss/osssys_4_0_offset.h"
+#include "oss/osssys_4_0_sh_mask.h"
 
 #include "soc15_common.h"
 #include "vega10_ih.h"
 
-
+#define MAX_REARM_RETRY 10
 
 static void vega10_ih_set_interrupt_funcs(struct amdgpu_device *adev);
 
 /**
- * vega10_ih_enable_interrupts - Enable the interrupt ring buffer
+ * vega10_ih_init_register_offset - Initialize register offset for ih rings
  *
  * @adev: amdgpu_device pointer
  *
- * Enable the interrupt ring buffer (VEGA10).
+ * Initialize register offset ih rings (VEGA10).
  */
-static void vega10_ih_enable_interrupts(struct amdgpu_device *adev)
+static void vega10_ih_init_register_offset(struct amdgpu_device *adev)
 {
-	u32 ih_rb_cntl = RREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL));
+	struct amdgpu_ih_regs *ih_regs;
+
+	if (adev->irq.ih.ring_size) {
+		ih_regs = &adev->irq.ih.ih_regs;
+		ih_regs->ih_rb_base = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_BASE);
+		ih_regs->ih_rb_base_hi = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_BASE_HI);
+		ih_regs->ih_rb_cntl = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL);
+		ih_regs->ih_rb_wptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR);
+		ih_regs->ih_rb_rptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_RPTR);
+		ih_regs->ih_doorbell_rptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_DOORBELL_RPTR);
+		ih_regs->ih_rb_wptr_addr_lo = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR_ADDR_LO);
+		ih_regs->ih_rb_wptr_addr_hi = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR_ADDR_HI);
+		ih_regs->psp_reg_id = PSP_REG_IH_RB_CNTL;
+	}
 
-	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_ENABLE, 1);
-	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, ENABLE_INTR, 1);
-	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL), ih_rb_cntl);
-	adev->irq.ih.enabled = true;
+	if (adev->irq.ih1.ring_size) {
+		ih_regs = &adev->irq.ih1.ih_regs;
+		ih_regs->ih_rb_base = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_BASE_RING1);
+		ih_regs->ih_rb_base_hi = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_BASE_HI_RING1);
+		ih_regs->ih_rb_cntl = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL_RING1);
+		ih_regs->ih_rb_wptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR_RING1);
+		ih_regs->ih_rb_rptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_RPTR_RING1);
+		ih_regs->ih_doorbell_rptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_DOORBELL_RPTR_RING1);
+		ih_regs->psp_reg_id = PSP_REG_IH_RB_CNTL_RING1;
+	}
+
+	if (adev->irq.ih2.ring_size) {
+		ih_regs = &adev->irq.ih2.ih_regs;
+		ih_regs->ih_rb_base = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_BASE_RING2);
+		ih_regs->ih_rb_base_hi = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_BASE_HI_RING2);
+		ih_regs->ih_rb_cntl = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL_RING2);
+		ih_regs->ih_rb_wptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR_RING2);
+		ih_regs->ih_rb_rptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_RPTR_RING2);
+		ih_regs->ih_doorbell_rptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_DOORBELL_RPTR_RING2);
+		ih_regs->psp_reg_id = PSP_REG_IH_RB_CNTL_RING2;
+	}
 }
 
 /**
- * vega10_ih_disable_interrupts - Disable the interrupt ring buffer
+ * vega10_ih_toggle_ring_interrupts - toggle the interrupt ring buffer
  *
  * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointet
+ * @enable: true - enable the interrupts, false - disable the interrupts
  *
- * Disable the interrupt ring buffer (VEGA10).
+ * Toggle the interrupt ring buffer (VEGA10)
  */
-static void vega10_ih_disable_interrupts(struct amdgpu_device *adev)
+static int vega10_ih_toggle_ring_interrupts(struct amdgpu_device *adev,
+					    struct amdgpu_ih_ring *ih,
+					    bool enable)
 {
-	u32 ih_rb_cntl = RREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL));
+	struct amdgpu_ih_regs *ih_regs;
+	uint32_t tmp;
+
+	ih_regs = &ih->ih_regs;
+
+	tmp = RREG32(ih_regs->ih_rb_cntl);
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RB_ENABLE, (enable ? 1 : 0));
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RB_GPU_TS_ENABLE, 1);
+	/* enable_intr field is only valid in ring0 */
+	if (ih == &adev->irq.ih)
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, ENABLE_INTR, (enable ? 1 : 0));
+	if (amdgpu_sriov_vf(adev)) {
+		if (psp_reg_program(&adev->psp, ih_regs->psp_reg_id, tmp)) {
+			dev_err(adev->dev, "PSP program IH_RB_CNTL failed!\n");
+			return -ETIMEDOUT;
+		}
+	} else {
+		WREG32(ih_regs->ih_rb_cntl, tmp);
+	}
 
-	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_ENABLE, 0);
-	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, ENABLE_INTR, 0);
-	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL), ih_rb_cntl);
-	/* set rptr, wptr to 0 */
-	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_RPTR), 0);
-	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR), 0);
-	adev->irq.ih.enabled = false;
-	adev->irq.ih.rptr = 0;
+	if (enable) {
+		ih->enabled = true;
+	} else {
+		/* set rptr, wptr to 0 */
+		WREG32(ih_regs->ih_rb_rptr, 0);
+		WREG32(ih_regs->ih_rb_wptr, 0);
+		ih->enabled = false;
+		ih->rptr = 0;
+	}
+
+	return 0;
 }
 
 /**
- * vega10_ih_irq_init - init and enable the interrupt ring
+ * vega10_ih_toggle_interrupts - Toggle all the available interrupt ring buffers
  *
  * @adev: amdgpu_device pointer
+ * @enable: enable or disable interrupt ring buffers
  *
- * Allocate a ring buffer for the interrupt controller,
- * enable the RLC, disable interrupts, enable the IH
- * ring buffer and enable it (VI).
- * Called at device load and reume.
- * Returns 0 for success, errors for failure.
+ * Toggle all the available interrupt ring buffers (VEGA10).
  */
-static int vega10_ih_irq_init(struct amdgpu_device *adev)
+static int vega10_ih_toggle_interrupts(struct amdgpu_device *adev, bool enable)
 {
-	int ret = 0;
-	int rb_bufsz;
-	u32 ih_rb_cntl, ih_doorbell_rtpr;
-	u32 tmp;
-	u64 wptr_off;
+	struct amdgpu_ih_ring *ih[] = {&adev->irq.ih, &adev->irq.ih1, &adev->irq.ih2};
+	int i;
+	int r;
 
-	/* disable irqs */
-	vega10_ih_disable_interrupts(adev);
+	for (i = 0; i < ARRAY_SIZE(ih); i++) {
+		if (ih[i]->ring_size) {
+			r = vega10_ih_toggle_ring_interrupts(adev, ih[i], enable);
+			if (r)
+				return r;
+		}
+	}
 
-	if (adev->flags & AMD_IS_APU)
-		nbio_v7_0_ih_control(adev);
-	else
-		nbio_v6_1_ih_control(adev);
+	return 0;
+}
 
-	ih_rb_cntl = RREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL));
-	/* Ring Buffer base. [39:8] of 40-bit address of the beginning of the ring buffer*/
-	if (adev->irq.ih.use_bus_addr) {
-		WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_BASE), adev->irq.ih.rb_dma_addr >> 8);
-		WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_BASE_HI), ((u64)adev->irq.ih.rb_dma_addr >> 40) & 0xff);
-		ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_SPACE, 1);
-	} else {
-		WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_BASE), adev->irq.ih.gpu_addr >> 8);
-		WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_BASE_HI), (adev->irq.ih.gpu_addr >> 40) & 0xff);
-		ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_SPACE, 4);
-	}
-	rb_bufsz = order_base_2(adev->irq.ih.ring_size / 4);
-	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
-	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_OVERFLOW_ENABLE, 1);
+static uint32_t vega10_ih_rb_cntl(struct amdgpu_ih_ring *ih, uint32_t ih_rb_cntl)
+{
+	int rb_bufsz = order_base_2(ih->ring_size / 4);
+
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   MC_SPACE, ih->use_bus_addr ? 1 : 4);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   WPTR_OVERFLOW_CLEAR, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   WPTR_OVERFLOW_ENABLE, 1);
 	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_SIZE, rb_bufsz);
-	/* Ring Buffer write pointer writeback. If enabled, IH_RB_WPTR register value is written to memory */
-	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_WRITEBACK_ENABLE, 1);
+	/* Ring Buffer write pointer writeback. If enabled, IH_RB_WPTR register
+	 * value is written to memory
+	 */
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   WPTR_WRITEBACK_ENABLE, 1);
 	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_SNOOP, 1);
 	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_RO, 0);
 	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_VMID, 0);
 
-	if (adev->irq.msi_enabled)
-		ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RPTR_REARM, 1);
-
-	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL), ih_rb_cntl);
-
-	/* set the writeback address whether it's enabled or not */
-	if (adev->irq.ih.use_bus_addr)
-		wptr_off = adev->irq.ih.rb_dma_addr + (adev->irq.ih.wptr_offs * 4);
-	else
-		wptr_off = adev->wb.gpu_addr + (adev->irq.ih.wptr_offs * 4);
-	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR_ADDR_LO), lower_32_bits(wptr_off));
-	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR_ADDR_HI), upper_32_bits(wptr_off) & 0xFF);
+	return ih_rb_cntl;
+}
 
-	/* set rptr, wptr to 0 */
-	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_RPTR), 0);
-	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR), 0);
-
-	ih_doorbell_rtpr = RREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_DOORBELL_RPTR));
-	if (adev->irq.ih.use_doorbell) {
-		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr, IH_DOORBELL_RPTR,
-						 OFFSET, adev->irq.ih.doorbell_index);
-		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr, IH_DOORBELL_RPTR,
+static uint32_t vega10_ih_doorbell_rptr(struct amdgpu_ih_ring *ih)
+{
+	u32 ih_doorbell_rtpr = 0;
+
+	if (ih->use_doorbell) {
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+						 IH_DOORBELL_RPTR, OFFSET,
+						 ih->doorbell_index);
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+						 IH_DOORBELL_RPTR,
 						 ENABLE, 1);
 	} else {
-		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr, IH_DOORBELL_RPTR,
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+						 IH_DOORBELL_RPTR,
 						 ENABLE, 0);
 	}
-	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_DOORBELL_RPTR), ih_doorbell_rtpr);
-	if (adev->flags & AMD_IS_APU)
-		nbio_v7_0_ih_doorbell_range(adev, adev->irq.ih.use_doorbell, adev->irq.ih.doorbell_index);
-	else
-		nbio_v6_1_ih_doorbell_range(adev, adev->irq.ih.use_doorbell, adev->irq.ih.doorbell_index);
+	return ih_doorbell_rtpr;
+}
+
+/**
+ * vega10_ih_enable_ring - enable an ih ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointer
+ *
+ * Enable an ih ring buffer (VEGA10)
+ */
+static int vega10_ih_enable_ring(struct amdgpu_device *adev,
+				 struct amdgpu_ih_ring *ih)
+{
+	struct amdgpu_ih_regs *ih_regs;
+	uint32_t tmp;
+
+	ih_regs = &ih->ih_regs;
+
+	/* Ring Buffer base. [39:8] of 40-bit address of the beginning of the ring buffer*/
+	WREG32(ih_regs->ih_rb_base, ih->gpu_addr >> 8);
+	WREG32(ih_regs->ih_rb_base_hi, (ih->gpu_addr >> 40) & 0xff);
+
+	tmp = RREG32(ih_regs->ih_rb_cntl);
+	tmp = vega10_ih_rb_cntl(ih, tmp);
+	if (ih == &adev->irq.ih)
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RPTR_REARM, !!adev->irq.msi_enabled);
+	if (ih == &adev->irq.ih1)
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RB_FULL_DRAIN_ENABLE, 1);
+	if (amdgpu_sriov_vf(adev)) {
+		if (psp_reg_program(&adev->psp, ih_regs->psp_reg_id, tmp)) {
+			dev_err(adev->dev, "PSP program IH_RB_CNTL failed!\n");
+			return -ETIMEDOUT;
+		}
+	} else {
+		WREG32(ih_regs->ih_rb_cntl, tmp);
+	}
+
+	if (ih == &adev->irq.ih) {
+		/* set the ih ring 0 writeback address whether it's enabled or not */
+		WREG32(ih_regs->ih_rb_wptr_addr_lo, lower_32_bits(ih->wptr_addr));
+		WREG32(ih_regs->ih_rb_wptr_addr_hi, upper_32_bits(ih->wptr_addr) & 0xFFFF);
+	}
+
+	/* set rptr, wptr to 0 */
+	WREG32(ih_regs->ih_rb_wptr, 0);
+	WREG32(ih_regs->ih_rb_rptr, 0);
+
+	WREG32(ih_regs->ih_doorbell_rptr, vega10_ih_doorbell_rptr(ih));
+
+	return 0;
+}
+
+/**
+ * vega10_ih_irq_init - init and enable the interrupt ring
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate a ring buffer for the interrupt controller,
+ * enable the RLC, disable interrupts, enable the IH
+ * ring buffer and enable it (VI).
+ * Called at device load and reume.
+ * Returns 0 for success, errors for failure.
+ */
+static int vega10_ih_irq_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_ih_ring *ih[] = {&adev->irq.ih, &adev->irq.ih1, &adev->irq.ih2};
+	u32 ih_chicken;
+	int ret;
+	int i;
+
+	/* disable irqs */
+	ret = vega10_ih_toggle_interrupts(adev, false);
+	if (ret)
+		return ret;
+
+	adev->nbio.funcs->ih_control(adev);
+
+	if (adev->asic_type == CHIP_RENOIR) {
+		ih_chicken = RREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN);
+		if (adev->irq.ih.use_bus_addr) {
+			ih_chicken = REG_SET_FIELD(ih_chicken, IH_CHICKEN,
+						   MC_SPACE_GPA_ENABLE, 1);
+		}
+		WREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN, ih_chicken);
+	}
 
-	tmp = RREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_STORM_CLIENT_LIST_CNTL));
-	tmp = REG_SET_FIELD(tmp, IH_STORM_CLIENT_LIST_CNTL,
-			    CLIENT18_IS_STORM_CLIENT, 1);
-	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_STORM_CLIENT_LIST_CNTL), tmp);
+	for (i = 0; i < ARRAY_SIZE(ih); i++) {
+		if (ih[i]->ring_size) {
+			ret = vega10_ih_enable_ring(adev, ih[i]);
+			if (ret)
+				return ret;
+		}
+	}
 
-	tmp = RREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_INT_FLOOD_CNTL));
-	tmp = REG_SET_FIELD(tmp, IH_INT_FLOOD_CNTL, FLOOD_CNTL_ENABLE, 1);
-	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_INT_FLOOD_CNTL), tmp);
+	if (!amdgpu_sriov_vf(adev))
+		adev->nbio.funcs->ih_doorbell_range(adev, adev->irq.ih.use_doorbell,
+						    adev->irq.ih.doorbell_index);
 
 	pci_set_master(adev->pdev);
 
 	/* enable interrupts */
-	vega10_ih_enable_interrupts(adev);
+	ret = vega10_ih_toggle_interrupts(adev, true);
+	if (ret)
+		return ret;
 
-	return ret;
+	if (adev->irq.ih_soft.ring_size)
+		adev->irq.ih_soft.enabled = true;
+
+	return 0;
 }
 
 /**
@@ -182,7 +315,7 @@ static int vega10_ih_irq_init(struct amdgpu_device *adev)
  */
 static void vega10_ih_irq_disable(struct amdgpu_device *adev)
 {
-	vega10_ih_disable_interrupts(adev);
+	vega10_ih_toggle_interrupts(adev, false);
 
 	/* Wait and acknowledge irq */
 	mdelay(1);
@@ -192,200 +325,297 @@ static void vega10_ih_irq_disable(struct amdgpu_device *adev)
  * vega10_ih_get_wptr - get the IH ring buffer wptr
  *
  * @adev: amdgpu_device pointer
+ * @ih: IH ring buffer to fetch wptr
  *
  * Get the IH ring buffer wptr from either the register
  * or the writeback memory buffer (VEGA10).  Also check for
  * ring buffer overflow and deal with it.
  * Returns the value of the wptr.
  */
-static u32 vega10_ih_get_wptr(struct amdgpu_device *adev)
+static u32 vega10_ih_get_wptr(struct amdgpu_device *adev,
+			      struct amdgpu_ih_ring *ih)
 {
 	u32 wptr, tmp;
+	struct amdgpu_ih_regs *ih_regs;
 
-	if (adev->irq.ih.use_bus_addr)
-		wptr = le32_to_cpu(adev->irq.ih.ring[adev->irq.ih.wptr_offs]);
-	else
-		wptr = le32_to_cpu(adev->wb.wb[adev->irq.ih.wptr_offs]);
-
-	if (REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW)) {
-		wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
-
-		/* When a ring buffer overflow happen start parsing interrupt
-		 * from the last not overwritten vector (wptr + 32). Hopefully
-		 * this should allow us to catchup.
+	if (ih == &adev->irq.ih || ih == &adev->irq.ih_soft) {
+		/* Only ring0 supports writeback. On other rings fall back
+		 * to register-based code with overflow checking below.
+		 * ih_soft ring doesn't have any backing hardware registers,
+		 * update wptr and return.
 		 */
-		tmp = (wptr + 32) & adev->irq.ih.ptr_mask;
-		dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
-			wptr, adev->irq.ih.rptr, tmp);
-		adev->irq.ih.rptr = tmp;
-
-		tmp = RREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL));
-		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
-		WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL), tmp);
+		wptr = le32_to_cpu(*ih->wptr_cpu);
+
+		if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
+			goto out;
 	}
-	return (wptr & adev->irq.ih.ptr_mask);
+
+	ih_regs = &ih->ih_regs;
+
+	/* Double check that the overflow wasn't already cleared. */
+	wptr = RREG32_NO_KIQ(ih_regs->ih_rb_wptr);
+	if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
+		goto out;
+
+	wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
+
+	/* When a ring buffer overflow happen start parsing interrupt
+	 * from the last not overwritten vector (wptr + 32). Hopefully
+	 * this should allow us to catchup.
+	 */
+	tmp = (wptr + 32) & ih->ptr_mask;
+	dev_warn_ratelimited(adev->dev, "%s ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
+			     amdgpu_ih_ring_name(adev, ih), wptr, ih->rptr, tmp);
+	ih->rptr = tmp;
+
+	tmp = RREG32_NO_KIQ(ih_regs->ih_rb_cntl);
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+	WREG32_NO_KIQ(ih_regs->ih_rb_cntl, tmp);
+
+	/* Unset the CLEAR_OVERFLOW bit immediately so new overflows
+	 * can be detected.
+	 */
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 0);
+	WREG32_NO_KIQ(ih_regs->ih_rb_cntl, tmp);
+
+out:
+	return (wptr & ih->ptr_mask);
 }
 
 /**
- * vega10_ih_decode_iv - decode an interrupt vector
+ * vega10_ih_irq_rearm - rearm IRQ if lost
  *
  * @adev: amdgpu_device pointer
+ * @ih: IH ring to match
  *
- * Decodes the interrupt vector at the current rptr
- * position and also advance the position.
  */
-static void vega10_ih_decode_iv(struct amdgpu_device *adev,
-				 struct amdgpu_iv_entry *entry)
+static void vega10_ih_irq_rearm(struct amdgpu_device *adev,
+			       struct amdgpu_ih_ring *ih)
 {
-	/* wptr/rptr are in bytes! */
-	u32 ring_index = adev->irq.ih.rptr >> 2;
-	uint32_t dw[8];
-
-	dw[0] = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
-	dw[1] = le32_to_cpu(adev->irq.ih.ring[ring_index + 1]);
-	dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
-	dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
-	dw[4] = le32_to_cpu(adev->irq.ih.ring[ring_index + 4]);
-	dw[5] = le32_to_cpu(adev->irq.ih.ring[ring_index + 5]);
-	dw[6] = le32_to_cpu(adev->irq.ih.ring[ring_index + 6]);
-	dw[7] = le32_to_cpu(adev->irq.ih.ring[ring_index + 7]);
-
-	entry->client_id = dw[0] & 0xff;
-	entry->src_id = (dw[0] >> 8) & 0xff;
-	entry->ring_id = (dw[0] >> 16) & 0xff;
-	entry->vm_id = (dw[0] >> 24) & 0xf;
-	entry->vm_id_src = (dw[0] >> 31);
-	entry->timestamp = dw[1] | ((u64)(dw[2] & 0xffff) << 32);
-	entry->timestamp_src = dw[2] >> 31;
-	entry->pas_id = dw[3] & 0xffff;
-	entry->pasid_src = dw[3] >> 31;
-	entry->src_data[0] = dw[4];
-	entry->src_data[1] = dw[5];
-	entry->src_data[2] = dw[6];
-	entry->src_data[3] = dw[7];
-
-
-	/* wptr/rptr are in bytes! */
-	adev->irq.ih.rptr += 32;
+	uint32_t v = 0;
+	uint32_t i = 0;
+	struct amdgpu_ih_regs *ih_regs;
+
+	ih_regs = &ih->ih_regs;
+	/* Rearm IRQ / re-wwrite doorbell if doorbell write is lost */
+	for (i = 0; i < MAX_REARM_RETRY; i++) {
+		v = RREG32_NO_KIQ(ih_regs->ih_rb_rptr);
+		if ((v < ih->ring_size) && (v != ih->rptr))
+			WDOORBELL32(ih->doorbell_index, ih->rptr);
+		else
+			break;
+	}
 }
 
 /**
  * vega10_ih_set_rptr - set the IH ring buffer rptr
  *
  * @adev: amdgpu_device pointer
+ * @ih: IH ring buffer to set rptr
  *
  * Set the IH ring buffer rptr.
  */
-static void vega10_ih_set_rptr(struct amdgpu_device *adev)
+static void vega10_ih_set_rptr(struct amdgpu_device *adev,
+			       struct amdgpu_ih_ring *ih)
 {
-	if (adev->irq.ih.use_doorbell) {
+	struct amdgpu_ih_regs *ih_regs;
+
+	if (ih == &adev->irq.ih_soft)
+		return;
+
+	if (ih->use_doorbell) {
 		/* XXX check if swapping is necessary on BE */
-		if (adev->irq.ih.use_bus_addr)
-			adev->irq.ih.ring[adev->irq.ih.rptr_offs] = adev->irq.ih.rptr;
-		else
-			adev->wb.wb[adev->irq.ih.rptr_offs] = adev->irq.ih.rptr;
-		WDOORBELL32(adev->irq.ih.doorbell_index, adev->irq.ih.rptr);
+		*ih->rptr_cpu = ih->rptr;
+		WDOORBELL32(ih->doorbell_index, ih->rptr);
+
+		if (amdgpu_sriov_vf(adev))
+			vega10_ih_irq_rearm(adev, ih);
 	} else {
-		WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_RPTR), adev->irq.ih.rptr);
+		ih_regs = &ih->ih_regs;
+		WREG32(ih_regs->ih_rb_rptr, ih->rptr);
+	}
+}
+
+/**
+ * vega10_ih_self_irq - dispatch work for ring 1 and 2
+ *
+ * @adev: amdgpu_device pointer
+ * @source: irq source
+ * @entry: IV with WPTR update
+ *
+ * Update the WPTR from the IV and schedule work to handle the entries.
+ */
+static int vega10_ih_self_irq(struct amdgpu_device *adev,
+			      struct amdgpu_irq_src *source,
+			      struct amdgpu_iv_entry *entry)
+{
+	switch (entry->ring_id) {
+	case 1:
+		schedule_work(&adev->irq.ih1_work);
+		break;
+	case 2:
+		schedule_work(&adev->irq.ih2_work);
+		break;
+	default: break;
 	}
+	return 0;
 }
 
-static int vega10_ih_early_init(void *handle)
+static const struct amdgpu_irq_src_funcs vega10_ih_self_irq_funcs = {
+	.process = vega10_ih_self_irq,
+};
+
+static void vega10_ih_set_self_irq_funcs(struct amdgpu_device *adev)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	adev->irq.self_irq.num_types = 0;
+	adev->irq.self_irq.funcs = &vega10_ih_self_irq_funcs;
+}
+
+static int vega10_ih_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
 
 	vega10_ih_set_interrupt_funcs(adev);
+	vega10_ih_set_self_irq_funcs(adev);
 	return 0;
 }
 
-static int vega10_ih_sw_init(void *handle)
+static int vega10_ih_sw_init(struct amdgpu_ip_block *ip_block)
 {
+	struct amdgpu_device *adev = ip_block->adev;
 	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	r = amdgpu_ih_ring_init(adev, 256 * 1024, true);
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_IH, 0,
+			      &adev->irq.self_irq);
+	if (r)
+		return r;
+
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih, IH_RING_SIZE, true);
 	if (r)
 		return r;
 
 	adev->irq.ih.use_doorbell = true;
-	adev->irq.ih.doorbell_index = AMDGPU_DOORBELL64_IH << 1;
+	adev->irq.ih.doorbell_index = adev->doorbell_index.ih << 1;
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		r = amdgpu_ih_ring_init(adev, &adev->irq.ih1, PAGE_SIZE, true);
+		if (r)
+			return r;
+
+		adev->irq.ih1.use_doorbell = true;
+		adev->irq.ih1.doorbell_index = (adev->doorbell_index.ih + 1) << 1;
+
+		r = amdgpu_ih_ring_init(adev, &adev->irq.ih2, PAGE_SIZE, true);
+		if (r)
+			return r;
+
+		adev->irq.ih2.use_doorbell = true;
+		adev->irq.ih2.doorbell_index = (adev->doorbell_index.ih + 2) << 1;
+	}
+	/* initialize ih control registers offset */
+	vega10_ih_init_register_offset(adev);
+
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih_soft, IH_SW_RING_SIZE, true);
+	if (r)
+		return r;
 
 	r = amdgpu_irq_init(adev);
 
 	return r;
 }
 
-static int vega10_ih_sw_fini(void *handle)
+static int vega10_ih_sw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
-	amdgpu_irq_fini(adev);
-	amdgpu_ih_ring_fini(adev);
+	amdgpu_irq_fini_sw(adev);
 
 	return 0;
 }
 
-static int vega10_ih_hw_init(void *handle)
+static int vega10_ih_hw_init(struct amdgpu_ip_block *ip_block)
 {
-	int r;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	r = vega10_ih_irq_init(adev);
-	if (r)
-		return r;
-
-	return 0;
+	return vega10_ih_irq_init(ip_block->adev);
 }
 
-static int vega10_ih_hw_fini(void *handle)
+static int vega10_ih_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	vega10_ih_irq_disable(adev);
+	vega10_ih_irq_disable(ip_block->adev);
 
 	return 0;
 }
 
-static int vega10_ih_suspend(void *handle)
+static int vega10_ih_suspend(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return vega10_ih_hw_fini(adev);
+	return vega10_ih_hw_fini(ip_block);
 }
 
-static int vega10_ih_resume(void *handle)
+static int vega10_ih_resume(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return vega10_ih_hw_init(adev);
+	return vega10_ih_hw_init(ip_block);
 }
 
-static bool vega10_ih_is_idle(void *handle)
+static bool vega10_ih_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	/* todo */
 	return true;
 }
 
-static int vega10_ih_wait_for_idle(void *handle)
+static int vega10_ih_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
 	/* todo */
 	return -ETIMEDOUT;
 }
 
-static int vega10_ih_soft_reset(void *handle)
+static int vega10_ih_soft_reset(struct amdgpu_ip_block *ip_block)
 {
 	/* todo */
 
 	return 0;
 }
 
-static int vega10_ih_set_clockgating_state(void *handle,
+static void vega10_ih_update_clockgating_state(struct amdgpu_device *adev,
+					       bool enable)
+{
+	uint32_t data, def, field_val;
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_IH_CG) {
+		def = data = RREG32_SOC15(OSSSYS, 0, mmIH_CLK_CTRL);
+		field_val = enable ? 0 : 1;
+		/**
+		 * Vega10/12 and RAVEN don't have IH_BUFFER_MEM_CLK_SOFT_OVERRIDE field.
+		 */
+		if (adev->asic_type == CHIP_RENOIR)
+			data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     IH_BUFFER_MEM_CLK_SOFT_OVERRIDE, field_val);
+
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     DBUS_MUX_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     OSSSYS_SHARE_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     LIMIT_SMN_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     DYN_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     REG_CLK_SOFT_OVERRIDE, field_val);
+		if (def != data)
+			WREG32_SOC15(OSSSYS, 0, mmIH_CLK_CTRL, data);
+	}
+}
+
+static int vega10_ih_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
+	struct amdgpu_device *adev = ip_block->adev;
+
+	vega10_ih_update_clockgating_state(adev,
+				state == AMD_CG_STATE_GATE);
 	return 0;
+
 }
 
-static int vega10_ih_set_powergating_state(void *handle,
+static int vega10_ih_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_powergating_state state)
 {
 	return 0;
@@ -394,7 +624,6 @@ static int vega10_ih_set_powergating_state(void *handle,
 const struct amd_ip_funcs vega10_ih_ip_funcs = {
 	.name = "vega10_ih",
 	.early_init = vega10_ih_early_init,
-	.late_init = NULL,
 	.sw_init = vega10_ih_sw_init,
 	.sw_fini = vega10_ih_sw_fini,
 	.hw_init = vega10_ih_hw_init,
@@ -410,14 +639,14 @@ const struct amd_ip_funcs vega10_ih_ip_funcs = {
 
 static const struct amdgpu_ih_funcs vega10_ih_funcs = {
 	.get_wptr = vega10_ih_get_wptr,
-	.decode_iv = vega10_ih_decode_iv,
+	.decode_iv = amdgpu_ih_decode_iv_helper,
+	.decode_iv_ts = amdgpu_ih_decode_iv_ts_helper,
 	.set_rptr = vega10_ih_set_rptr
 };
 
 static void vega10_ih_set_interrupt_funcs(struct amdgpu_device *adev)
 {
-	if (adev->irq.ih_funcs == NULL)
-		adev->irq.ih_funcs = &vega10_ih_funcs;
+	adev->irq.ih_funcs = &vega10_ih_funcs;
 }
 
 const struct amdgpu_ip_block_version vega10_ih_ip_block =
diff --git a/drivers/gpu/drm/amd/amdgpu/vega10_reg_init.c b/drivers/gpu/drm/amd/amdgpu/vega10_reg_init.c
new file mode 100644
index 000000000000..6b52a539d51b
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vega10_reg_init.c
@@ -0,0 +1,95 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "soc15.h"
+
+#include "soc15_common.h"
+#include "vega10_ip_offset.h"
+
+int vega10_reg_base_init(struct amdgpu_device *adev)
+{
+	/* HW has more IP blocks,  only initialized the blocke beend by our driver  */
+	uint32_t i;
+	for (i = 0 ; i < MAX_INSTANCE ; ++i) {
+		adev->reg_offset[GC_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i]));
+		adev->reg_offset[HDP_HWIP][i] = (uint32_t *)(&(HDP_BASE.instance[i]));
+		adev->reg_offset[MMHUB_HWIP][i] = (uint32_t *)(&(MMHUB_BASE.instance[i]));
+		adev->reg_offset[ATHUB_HWIP][i] = (uint32_t *)(&(ATHUB_BASE.instance[i]));
+		adev->reg_offset[NBIO_HWIP][i] = (uint32_t *)(&(NBIO_BASE.instance[i]));
+		adev->reg_offset[MP0_HWIP][i] = (uint32_t *)(&(MP0_BASE.instance[i]));
+		adev->reg_offset[MP1_HWIP][i] = (uint32_t *)(&(MP1_BASE.instance[i]));
+		adev->reg_offset[UVD_HWIP][i] = (uint32_t *)(&(UVD_BASE.instance[i]));
+		adev->reg_offset[VCE_HWIP][i] = (uint32_t *)(&(VCE_BASE.instance[i]));
+		adev->reg_offset[VCN_HWIP][i] = (uint32_t *)(&(VCN_BASE.instance[i]));
+		adev->reg_offset[DF_HWIP][i] = (uint32_t *)(&(DF_BASE.instance[i]));
+		adev->reg_offset[DCE_HWIP][i] = (uint32_t *)(&(DCE_BASE.instance[i]));
+		adev->reg_offset[OSSSYS_HWIP][i] = (uint32_t *)(&(OSSSYS_BASE.instance[i]));
+		adev->reg_offset[SDMA0_HWIP][i] = (uint32_t *)(&(SDMA0_BASE.instance[i]));
+		adev->reg_offset[SDMA1_HWIP][i] = (uint32_t *)(&(SDMA1_BASE.instance[i]));
+		adev->reg_offset[SMUIO_HWIP][i] = (uint32_t *)(&(SMUIO_BASE.instance[i]));
+		adev->reg_offset[PWR_HWIP][i] = (uint32_t *)(&(PWR_BASE.instance[i]));
+		adev->reg_offset[NBIF_HWIP][i] = (uint32_t *)(&(NBIF_BASE.instance[i]));
+		adev->reg_offset[THM_HWIP][i] = (uint32_t *)(&(THM_BASE.instance[i]));
+		adev->reg_offset[CLK_HWIP][i] = (uint32_t *)(&(CLK_BASE.instance[i]));
+	}
+	return 0;
+}
+
+void vega10_doorbell_index_init(struct amdgpu_device *adev)
+{
+	adev->doorbell_index.kiq = AMDGPU_DOORBELL64_KIQ;
+	adev->doorbell_index.mec_ring0 = AMDGPU_DOORBELL64_MEC_RING0;
+	adev->doorbell_index.mec_ring1 = AMDGPU_DOORBELL64_MEC_RING1;
+	adev->doorbell_index.mec_ring2 = AMDGPU_DOORBELL64_MEC_RING2;
+	adev->doorbell_index.mec_ring3 = AMDGPU_DOORBELL64_MEC_RING3;
+	adev->doorbell_index.mec_ring4 = AMDGPU_DOORBELL64_MEC_RING4;
+	adev->doorbell_index.mec_ring5 = AMDGPU_DOORBELL64_MEC_RING5;
+	adev->doorbell_index.mec_ring6 = AMDGPU_DOORBELL64_MEC_RING6;
+	adev->doorbell_index.mec_ring7 = AMDGPU_DOORBELL64_MEC_RING7;
+	adev->doorbell_index.userqueue_start = AMDGPU_DOORBELL64_USERQUEUE_START;
+	adev->doorbell_index.userqueue_end = AMDGPU_DOORBELL64_USERQUEUE_END;
+	adev->doorbell_index.gfx_ring0 = AMDGPU_DOORBELL64_GFX_RING0;
+	adev->doorbell_index.sdma_engine[0] = AMDGPU_DOORBELL64_sDMA_ENGINE0;
+	adev->doorbell_index.sdma_engine[1] = AMDGPU_DOORBELL64_sDMA_ENGINE1;
+	adev->doorbell_index.ih = AMDGPU_DOORBELL64_IH;
+	adev->doorbell_index.uvd_vce.uvd_ring0_1 = AMDGPU_DOORBELL64_UVD_RING0_1;
+	adev->doorbell_index.uvd_vce.uvd_ring2_3 = AMDGPU_DOORBELL64_UVD_RING2_3;
+	adev->doorbell_index.uvd_vce.uvd_ring4_5 = AMDGPU_DOORBELL64_UVD_RING4_5;
+	adev->doorbell_index.uvd_vce.uvd_ring6_7 = AMDGPU_DOORBELL64_UVD_RING6_7;
+	adev->doorbell_index.uvd_vce.vce_ring0_1 = AMDGPU_DOORBELL64_VCE_RING0_1;
+	adev->doorbell_index.uvd_vce.vce_ring2_3 = AMDGPU_DOORBELL64_VCE_RING2_3;
+	adev->doorbell_index.uvd_vce.vce_ring4_5 = AMDGPU_DOORBELL64_VCE_RING4_5;
+	adev->doorbell_index.uvd_vce.vce_ring6_7 = AMDGPU_DOORBELL64_VCE_RING6_7;
+	adev->doorbell_index.vcn.vcn_ring0_1 = AMDGPU_DOORBELL64_VCN0_1;
+	adev->doorbell_index.vcn.vcn_ring2_3 = AMDGPU_DOORBELL64_VCN2_3;
+	adev->doorbell_index.vcn.vcn_ring4_5 = AMDGPU_DOORBELL64_VCN4_5;
+	adev->doorbell_index.vcn.vcn_ring6_7 = AMDGPU_DOORBELL64_VCN6_7;
+
+	adev->doorbell_index.first_non_cp = AMDGPU_DOORBELL64_FIRST_NON_CP;
+	adev->doorbell_index.last_non_cp = AMDGPU_DOORBELL64_LAST_NON_CP;
+
+	/* In unit of dword doorbell */
+	adev->doorbell_index.max_assignment = AMDGPU_DOORBELL64_MAX_ASSIGNMENT << 1;
+	adev->doorbell_index.sdma_doorbell_range = 4;
+}
+
diff --git a/drivers/gpu/drm/amd/amdgpu/vega20_ih.c b/drivers/gpu/drm/amd/amdgpu/vega20_ih.c
new file mode 100644
index 000000000000..85846fd08ce4
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vega20_ih.c
@@ -0,0 +1,755 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/pci.h>
+
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+#include "soc15.h"
+
+#include "oss/osssys_4_2_0_offset.h"
+#include "oss/osssys_4_2_0_sh_mask.h"
+
+#include "soc15_common.h"
+#include "vega20_ih.h"
+
+#define MAX_REARM_RETRY 10
+
+#define mmIH_CHICKEN_ALDEBARAN			0x18d
+#define mmIH_CHICKEN_ALDEBARAN_BASE_IDX		0
+
+#define mmIH_RETRY_INT_CAM_CNTL_ALDEBARAN		0x00ea
+#define mmIH_RETRY_INT_CAM_CNTL_ALDEBARAN_BASE_IDX	0
+#define IH_RETRY_INT_CAM_CNTL_ALDEBARAN__ENABLE__SHIFT	0x10
+#define IH_RETRY_INT_CAM_CNTL_ALDEBARAN__ENABLE_MASK	0x00010000L
+
+static void vega20_ih_set_interrupt_funcs(struct amdgpu_device *adev);
+
+/**
+ * vega20_ih_init_register_offset - Initialize register offset for ih rings
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialize register offset ih rings (VEGA20).
+ */
+static void vega20_ih_init_register_offset(struct amdgpu_device *adev)
+{
+	struct amdgpu_ih_regs *ih_regs;
+
+	if (adev->irq.ih.ring_size) {
+		ih_regs = &adev->irq.ih.ih_regs;
+		ih_regs->ih_rb_base = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_BASE);
+		ih_regs->ih_rb_base_hi = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_BASE_HI);
+		ih_regs->ih_rb_cntl = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL);
+		ih_regs->ih_rb_wptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR);
+		ih_regs->ih_rb_rptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_RPTR);
+		ih_regs->ih_doorbell_rptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_DOORBELL_RPTR);
+		ih_regs->ih_rb_wptr_addr_lo = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR_ADDR_LO);
+		ih_regs->ih_rb_wptr_addr_hi = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR_ADDR_HI);
+		ih_regs->psp_reg_id = PSP_REG_IH_RB_CNTL;
+	}
+
+	if (adev->irq.ih1.ring_size) {
+		ih_regs = &adev->irq.ih1.ih_regs;
+		ih_regs->ih_rb_base = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_BASE_RING1);
+		ih_regs->ih_rb_base_hi = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_BASE_HI_RING1);
+		ih_regs->ih_rb_cntl = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL_RING1);
+		ih_regs->ih_rb_wptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR_RING1);
+		ih_regs->ih_rb_rptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_RPTR_RING1);
+		ih_regs->ih_doorbell_rptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_DOORBELL_RPTR_RING1);
+		ih_regs->psp_reg_id = PSP_REG_IH_RB_CNTL_RING1;
+	}
+
+	if (adev->irq.ih2.ring_size) {
+		ih_regs = &adev->irq.ih2.ih_regs;
+		ih_regs->ih_rb_base = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_BASE_RING2);
+		ih_regs->ih_rb_base_hi = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_BASE_HI_RING2);
+		ih_regs->ih_rb_cntl = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL_RING2);
+		ih_regs->ih_rb_wptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR_RING2);
+		ih_regs->ih_rb_rptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_RPTR_RING2);
+		ih_regs->ih_doorbell_rptr = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_DOORBELL_RPTR_RING2);
+		ih_regs->psp_reg_id = PSP_REG_IH_RB_CNTL_RING2;
+	}
+}
+
+/**
+ * vega20_ih_toggle_ring_interrupts - toggle the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointer
+ * @enable: true - enable the interrupts, false - disable the interrupts
+ *
+ * Toggle the interrupt ring buffer (VEGA20)
+ */
+static int vega20_ih_toggle_ring_interrupts(struct amdgpu_device *adev,
+					    struct amdgpu_ih_ring *ih,
+					    bool enable)
+{
+	struct amdgpu_ih_regs *ih_regs;
+	uint32_t tmp;
+
+	ih_regs = &ih->ih_regs;
+
+	tmp = RREG32(ih_regs->ih_rb_cntl);
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RB_ENABLE, (enable ? 1 : 0));
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RB_GPU_TS_ENABLE, 1);
+
+	if (enable) {
+		/* Unset the CLEAR_OVERFLOW bit to make sure the next step
+		 * is switching the bit from 0 to 1
+		 */
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 0);
+		if (amdgpu_sriov_vf(adev) && amdgpu_sriov_reg_indirect_ih(adev)) {
+			if (psp_reg_program(&adev->psp, ih_regs->psp_reg_id, tmp))
+				return -ETIMEDOUT;
+		} else {
+			WREG32_NO_KIQ(ih_regs->ih_rb_cntl, tmp);
+		}
+
+		/* Clear RB_OVERFLOW bit */
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+		if (amdgpu_sriov_vf(adev) && amdgpu_sriov_reg_indirect_ih(adev)) {
+			if (psp_reg_program(&adev->psp, ih_regs->psp_reg_id, tmp))
+				return -ETIMEDOUT;
+		} else {
+			WREG32_NO_KIQ(ih_regs->ih_rb_cntl, tmp);
+		}
+
+		/* Unset the CLEAR_OVERFLOW bit immediately so new overflows
+		 * can be detected.
+		 */
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 0);
+	}
+
+	/* enable_intr field is only valid in ring0 */
+	if (ih == &adev->irq.ih)
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, ENABLE_INTR, (enable ? 1 : 0));
+	if (amdgpu_sriov_vf(adev)) {
+		if (psp_reg_program(&adev->psp, ih_regs->psp_reg_id, tmp)) {
+			dev_err(adev->dev, "PSP program IH_RB_CNTL failed!\n");
+			return -ETIMEDOUT;
+		}
+	} else {
+		WREG32(ih_regs->ih_rb_cntl, tmp);
+	}
+
+	if (enable) {
+		ih->enabled = true;
+	} else {
+		/* set rptr, wptr to 0 */
+		WREG32(ih_regs->ih_rb_rptr, 0);
+		WREG32(ih_regs->ih_rb_wptr, 0);
+		ih->enabled = false;
+		ih->rptr = 0;
+	}
+
+	return 0;
+}
+
+/**
+ * vega20_ih_toggle_interrupts - Toggle all the available interrupt ring buffers
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable or disable interrupt ring buffers
+ *
+ * Toggle all the available interrupt ring buffers (VEGA20).
+ */
+static int vega20_ih_toggle_interrupts(struct amdgpu_device *adev, bool enable)
+{
+	struct amdgpu_ih_ring *ih[] = {&adev->irq.ih, &adev->irq.ih1, &adev->irq.ih2};
+	int i;
+	int r;
+
+	for (i = 0; i < ARRAY_SIZE(ih); i++) {
+		if (ih[i]->ring_size) {
+			r = vega20_ih_toggle_ring_interrupts(adev, ih[i], enable);
+			if (r)
+				return r;
+		}
+	}
+
+	return 0;
+}
+
+static uint32_t vega20_ih_rb_cntl(struct amdgpu_ih_ring *ih, uint32_t ih_rb_cntl)
+{
+	int rb_bufsz = order_base_2(ih->ring_size / 4);
+
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   MC_SPACE, ih->use_bus_addr ? 1 : 4);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   WPTR_OVERFLOW_CLEAR, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   WPTR_OVERFLOW_ENABLE, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_SIZE, rb_bufsz);
+	/* Ring Buffer write pointer writeback. If enabled, IH_RB_WPTR register
+	 * value is written to memory
+	 */
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+				   WPTR_WRITEBACK_ENABLE, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_SNOOP, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_RO, 0);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_VMID, 0);
+
+	return ih_rb_cntl;
+}
+
+static uint32_t vega20_ih_doorbell_rptr(struct amdgpu_ih_ring *ih)
+{
+	u32 ih_doorbell_rtpr = 0;
+
+	if (ih->use_doorbell) {
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+						 IH_DOORBELL_RPTR, OFFSET,
+						 ih->doorbell_index);
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+						 IH_DOORBELL_RPTR,
+						 ENABLE, 1);
+	} else {
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+						 IH_DOORBELL_RPTR,
+						 ENABLE, 0);
+	}
+	return ih_doorbell_rtpr;
+}
+
+/**
+ * vega20_ih_enable_ring - enable an ih ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointer
+ *
+ * Enable an ih ring buffer (VEGA20)
+ */
+static int vega20_ih_enable_ring(struct amdgpu_device *adev,
+				 struct amdgpu_ih_ring *ih)
+{
+	struct amdgpu_ih_regs *ih_regs;
+	uint32_t tmp;
+
+	ih_regs = &ih->ih_regs;
+
+	/* Ring Buffer base. [39:8] of 40-bit address of the beginning of the ring buffer*/
+	WREG32(ih_regs->ih_rb_base, ih->gpu_addr >> 8);
+	WREG32(ih_regs->ih_rb_base_hi, (ih->gpu_addr >> 40) & 0xff);
+
+	tmp = RREG32(ih_regs->ih_rb_cntl);
+	tmp = vega20_ih_rb_cntl(ih, tmp);
+	if (ih == &adev->irq.ih)
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RPTR_REARM, !!adev->irq.msi_enabled);
+	if (ih == &adev->irq.ih1)
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RB_FULL_DRAIN_ENABLE, 1);
+	if (amdgpu_sriov_vf(adev)) {
+		if (psp_reg_program(&adev->psp, ih_regs->psp_reg_id, tmp)) {
+			dev_err(adev->dev, "PSP program IH_RB_CNTL failed!\n");
+			return -ETIMEDOUT;
+		}
+	} else {
+		WREG32(ih_regs->ih_rb_cntl, tmp);
+	}
+
+	if (ih == &adev->irq.ih) {
+		/* set the ih ring 0 writeback address whether it's enabled or not */
+		WREG32(ih_regs->ih_rb_wptr_addr_lo, lower_32_bits(ih->wptr_addr));
+		WREG32(ih_regs->ih_rb_wptr_addr_hi, upper_32_bits(ih->wptr_addr) & 0xFFFF);
+	}
+
+	/* set rptr, wptr to 0 */
+	WREG32(ih_regs->ih_rb_wptr, 0);
+	WREG32(ih_regs->ih_rb_rptr, 0);
+
+	WREG32(ih_regs->ih_doorbell_rptr, vega20_ih_doorbell_rptr(ih));
+
+	return 0;
+}
+
+static uint32_t vega20_setup_retry_doorbell(u32 doorbell_index)
+{
+	u32 val = 0;
+
+	val = REG_SET_FIELD(val, IH_DOORBELL_RPTR, OFFSET, doorbell_index);
+	val = REG_SET_FIELD(val, IH_DOORBELL_RPTR, ENABLE, 1);
+
+	return val;
+}
+
+/**
+ * vega20_ih_irq_init - init and enable the interrupt ring
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate a ring buffer for the interrupt controller,
+ * enable the RLC, disable interrupts, enable the IH
+ * ring buffer and enable it (VI).
+ * Called at device load and reume.
+ * Returns 0 for success, errors for failure.
+ */
+static int vega20_ih_irq_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_ih_ring *ih[] = {&adev->irq.ih, &adev->irq.ih1, &adev->irq.ih2};
+	u32 ih_chicken;
+	int ret;
+	int i;
+
+	/* disable irqs */
+	ret = vega20_ih_toggle_interrupts(adev, false);
+	if (ret)
+		return ret;
+
+	adev->nbio.funcs->ih_control(adev);
+
+	if (!amdgpu_sriov_vf(adev)) {
+		if ((amdgpu_ip_version(adev, OSSSYS_HWIP, 0) == IP_VERSION(4, 2, 1)) &&
+		    adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+			ih_chicken = RREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN);
+			if (adev->irq.ih.use_bus_addr) {
+				ih_chicken = REG_SET_FIELD(ih_chicken, IH_CHICKEN,
+							   MC_SPACE_GPA_ENABLE, 1);
+			}
+			WREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN, ih_chicken);
+		}
+
+		/* psp firmware won't program IH_CHICKEN for aldebaran
+		 * driver needs to program it properly according to
+		 * MC_SPACE type in IH_RB_CNTL */
+		if ((amdgpu_ip_version(adev, OSSSYS_HWIP, 0) == IP_VERSION(4, 4, 0)) ||
+		    (amdgpu_ip_version(adev, OSSSYS_HWIP, 0) == IP_VERSION(4, 4, 2)) ||
+		    (amdgpu_ip_version(adev, OSSSYS_HWIP, 0) == IP_VERSION(4, 4, 5))) {
+			ih_chicken = RREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN_ALDEBARAN);
+			if (adev->irq.ih.use_bus_addr) {
+				ih_chicken = REG_SET_FIELD(ih_chicken, IH_CHICKEN,
+							   MC_SPACE_GPA_ENABLE, 1);
+			}
+			WREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN_ALDEBARAN, ih_chicken);
+		}
+	}
+
+	for (i = 0; i < ARRAY_SIZE(ih); i++) {
+		if (ih[i]->ring_size) {
+			ret = vega20_ih_enable_ring(adev, ih[i]);
+			if (ret)
+				return ret;
+		}
+		ih[i]->overflow = false;
+	}
+
+	if (!amdgpu_sriov_vf(adev))
+		adev->nbio.funcs->ih_doorbell_range(adev, adev->irq.ih.use_doorbell,
+						    adev->irq.ih.doorbell_index);
+
+	pci_set_master(adev->pdev);
+
+	/* Allocate the doorbell for IH Retry CAM */
+	adev->irq.retry_cam_doorbell_index = (adev->doorbell_index.ih + 3) << 1;
+	WREG32_SOC15(OSSSYS, 0, mmIH_DOORBELL_RETRY_CAM,
+		vega20_setup_retry_doorbell(adev->irq.retry_cam_doorbell_index));
+
+	/* Enable IH Retry CAM */
+	if (amdgpu_ip_version(adev, OSSSYS_HWIP, 0) == IP_VERSION(4, 4, 0) ||
+	    amdgpu_ip_version(adev, OSSSYS_HWIP, 0) == IP_VERSION(4, 4, 2) ||
+	    amdgpu_ip_version(adev, OSSSYS_HWIP, 0) == IP_VERSION(4, 4, 4) ||
+	    amdgpu_ip_version(adev, OSSSYS_HWIP, 0) == IP_VERSION(4, 4, 5))
+		WREG32_FIELD15(OSSSYS, 0, IH_RETRY_INT_CAM_CNTL_ALDEBARAN,
+			       ENABLE, 1);
+	else
+		WREG32_FIELD15(OSSSYS, 0, IH_RETRY_INT_CAM_CNTL, ENABLE, 1);
+
+	adev->irq.retry_cam_enabled = true;
+
+	/* enable interrupts */
+	ret = vega20_ih_toggle_interrupts(adev, true);
+	if (ret)
+		return ret;
+
+	if (adev->irq.ih_soft.ring_size)
+		adev->irq.ih_soft.enabled = true;
+
+	return 0;
+}
+
+/**
+ * vega20_ih_irq_disable - disable interrupts
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable interrupts on the hw (VEGA20).
+ */
+static void vega20_ih_irq_disable(struct amdgpu_device *adev)
+{
+	vega20_ih_toggle_interrupts(adev, false);
+
+	/* Wait and acknowledge irq */
+	mdelay(1);
+}
+
+/**
+ * vega20_ih_get_wptr - get the IH ring buffer wptr
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointer
+ *
+ * Get the IH ring buffer wptr from either the register
+ * or the writeback memory buffer (VEGA20).  Also check for
+ * ring buffer overflow and deal with it.
+ * Returns the value of the wptr.
+ */
+static u32 vega20_ih_get_wptr(struct amdgpu_device *adev,
+			      struct amdgpu_ih_ring *ih)
+{
+	u32 wptr, tmp;
+	struct amdgpu_ih_regs *ih_regs;
+
+	if (ih == &adev->irq.ih || ih == &adev->irq.ih_soft) {
+		/* Only ring0 supports writeback. On other rings fall back
+		 * to register-based code with overflow checking below.
+		 * ih_soft ring doesn't have any backing hardware registers,
+		 * update wptr and return.
+		 */
+		wptr = le32_to_cpu(*ih->wptr_cpu);
+
+		if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
+			goto out;
+	}
+
+	ih_regs = &ih->ih_regs;
+
+	/* Double check that the overflow wasn't already cleared. */
+	wptr = RREG32_NO_KIQ(ih_regs->ih_rb_wptr);
+	if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
+		goto out;
+
+	if (!amdgpu_sriov_vf(adev))
+		wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
+	else
+		ih->overflow = true;
+
+	/* When a ring buffer overflow happen start parsing interrupt
+	 * from the last not overwritten vector (wptr + 32). Hopefully
+	 * this should allow us to catchup.
+	 */
+	tmp = (wptr + 32) & ih->ptr_mask;
+	dev_warn_ratelimited(adev->dev, "%s ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
+			     amdgpu_ih_ring_name(adev, ih), wptr, ih->rptr, tmp);
+	ih->rptr = tmp;
+
+	tmp = RREG32_NO_KIQ(ih_regs->ih_rb_cntl);
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+	WREG32_NO_KIQ(ih_regs->ih_rb_cntl, tmp);
+
+	/* Unset the CLEAR_OVERFLOW bit immediately so new overflows
+	 * can be detected.
+	 */
+	tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 0);
+	WREG32_NO_KIQ(ih_regs->ih_rb_cntl, tmp);
+
+out:
+	return (wptr & ih->ptr_mask);
+}
+
+/**
+ * vega20_ih_irq_rearm - rearm IRQ if lost
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointer
+ *
+ */
+static void vega20_ih_irq_rearm(struct amdgpu_device *adev,
+			       struct amdgpu_ih_ring *ih)
+{
+	uint32_t v = 0;
+	uint32_t i = 0;
+	struct amdgpu_ih_regs *ih_regs;
+
+	ih_regs = &ih->ih_regs;
+
+	/* Rearm IRQ / re-wwrite doorbell if doorbell write is lost */
+	for (i = 0; i < MAX_REARM_RETRY; i++) {
+		v = RREG32_NO_KIQ(ih_regs->ih_rb_rptr);
+		if ((v < ih->ring_size) && (v != ih->rptr))
+			WDOORBELL32(ih->doorbell_index, ih->rptr);
+		else
+			break;
+	}
+}
+
+/**
+ * vega20_ih_set_rptr - set the IH ring buffer rptr
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointer
+ *
+ * Set the IH ring buffer rptr.
+ */
+static void vega20_ih_set_rptr(struct amdgpu_device *adev,
+			       struct amdgpu_ih_ring *ih)
+{
+	struct amdgpu_ih_regs *ih_regs;
+
+	if (ih == &adev->irq.ih_soft)
+		return;
+
+	if (ih->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		*ih->rptr_cpu = ih->rptr;
+		WDOORBELL32(ih->doorbell_index, ih->rptr);
+
+		if (amdgpu_sriov_vf(adev))
+			vega20_ih_irq_rearm(adev, ih);
+	} else {
+		ih_regs = &ih->ih_regs;
+		WREG32(ih_regs->ih_rb_rptr, ih->rptr);
+	}
+}
+
+/**
+ * vega20_ih_self_irq - dispatch work for ring 1 and 2
+ *
+ * @adev: amdgpu_device pointer
+ * @source: irq source
+ * @entry: IV with WPTR update
+ *
+ * Update the WPTR from the IV and schedule work to handle the entries.
+ */
+static int vega20_ih_self_irq(struct amdgpu_device *adev,
+			      struct amdgpu_irq_src *source,
+			      struct amdgpu_iv_entry *entry)
+{
+	switch (entry->ring_id) {
+	case 1:
+		schedule_work(&adev->irq.ih1_work);
+		break;
+	case 2:
+		schedule_work(&adev->irq.ih2_work);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs vega20_ih_self_irq_funcs = {
+	.process = vega20_ih_self_irq,
+};
+
+static void vega20_ih_set_self_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->irq.self_irq.num_types = 0;
+	adev->irq.self_irq.funcs = &vega20_ih_self_irq_funcs;
+}
+
+static int vega20_ih_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	vega20_ih_set_interrupt_funcs(adev);
+	vega20_ih_set_self_irq_funcs(adev);
+	return 0;
+}
+
+static int vega20_ih_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	bool use_bus_addr = true;
+	int r;
+
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_IH, 0,
+			      &adev->irq.self_irq);
+	if (r)
+		return r;
+
+	if ((adev->flags & AMD_IS_APU) &&
+	    (amdgpu_ip_version(adev, OSSSYS_HWIP, 0) == IP_VERSION(4, 4, 2)))
+		use_bus_addr = false;
+
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih, IH_RING_SIZE, use_bus_addr);
+	if (r)
+		return r;
+
+	adev->irq.ih.use_doorbell = true;
+	adev->irq.ih.doorbell_index = adev->doorbell_index.ih << 1;
+
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih1, PAGE_SIZE, use_bus_addr);
+	if (r)
+		return r;
+
+	adev->irq.ih1.use_doorbell = true;
+	adev->irq.ih1.doorbell_index = (adev->doorbell_index.ih + 1) << 1;
+
+	if (amdgpu_ip_version(adev, OSSSYS_HWIP, 0) != IP_VERSION(4, 4, 2) &&
+	    amdgpu_ip_version(adev, OSSSYS_HWIP, 0) != IP_VERSION(4, 4, 5)) {
+		r = amdgpu_ih_ring_init(adev, &adev->irq.ih2, PAGE_SIZE, true);
+		if (r)
+			return r;
+
+		adev->irq.ih2.use_doorbell = true;
+		adev->irq.ih2.doorbell_index = (adev->doorbell_index.ih + 2) << 1;
+	}
+
+	/* initialize ih control registers offset */
+	vega20_ih_init_register_offset(adev);
+
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih_soft, IH_SW_RING_SIZE, use_bus_addr);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_init(adev);
+
+	return r;
+}
+
+static int vega20_ih_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	amdgpu_irq_fini_sw(adev);
+
+	return 0;
+}
+
+static int vega20_ih_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	int r;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	r = vega20_ih_irq_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static int vega20_ih_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	vega20_ih_irq_disable(ip_block->adev);
+
+	return 0;
+}
+
+static int vega20_ih_suspend(struct amdgpu_ip_block *ip_block)
+{
+	return vega20_ih_hw_fini(ip_block);
+}
+
+static int vega20_ih_resume(struct amdgpu_ip_block *ip_block)
+{
+	return vega20_ih_hw_init(ip_block);
+}
+
+static bool vega20_ih_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	/* todo */
+	return true;
+}
+
+static int vega20_ih_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	/* todo */
+	return -ETIMEDOUT;
+}
+
+static int vega20_ih_soft_reset(struct amdgpu_ip_block *ip_block)
+{
+	/* todo */
+
+	return 0;
+}
+
+static void vega20_ih_update_clockgating_state(struct amdgpu_device *adev,
+					       bool enable)
+{
+	uint32_t data, def, field_val;
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_IH_CG) {
+		def = data = RREG32_SOC15(OSSSYS, 0, mmIH_CLK_CTRL);
+		field_val = enable ? 0 : 1;
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     IH_RETRY_INT_CAM_MEM_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     IH_BUFFER_MEM_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     DBUS_MUX_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     OSSSYS_SHARE_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     LIMIT_SMN_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     DYN_CLK_SOFT_OVERRIDE, field_val);
+		data = REG_SET_FIELD(data, IH_CLK_CTRL,
+				     REG_CLK_SOFT_OVERRIDE, field_val);
+		if (def != data)
+			WREG32_SOC15(OSSSYS, 0, mmIH_CLK_CTRL, data);
+	}
+}
+
+static int vega20_ih_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	vega20_ih_update_clockgating_state(adev,
+				state == AMD_CG_STATE_GATE);
+	return 0;
+
+}
+
+static int vega20_ih_set_powergating_state(struct amdgpu_ip_block *ip_block,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs vega20_ih_ip_funcs = {
+	.name = "vega20_ih",
+	.early_init = vega20_ih_early_init,
+	.sw_init = vega20_ih_sw_init,
+	.sw_fini = vega20_ih_sw_fini,
+	.hw_init = vega20_ih_hw_init,
+	.hw_fini = vega20_ih_hw_fini,
+	.suspend = vega20_ih_suspend,
+	.resume = vega20_ih_resume,
+	.is_idle = vega20_ih_is_idle,
+	.wait_for_idle = vega20_ih_wait_for_idle,
+	.soft_reset = vega20_ih_soft_reset,
+	.set_clockgating_state = vega20_ih_set_clockgating_state,
+	.set_powergating_state = vega20_ih_set_powergating_state,
+};
+
+static const struct amdgpu_ih_funcs vega20_ih_funcs = {
+	.get_wptr = vega20_ih_get_wptr,
+	.decode_iv = amdgpu_ih_decode_iv_helper,
+	.decode_iv_ts = amdgpu_ih_decode_iv_ts_helper,
+	.set_rptr = vega20_ih_set_rptr
+};
+
+static void vega20_ih_set_interrupt_funcs(struct amdgpu_device *adev)
+{
+	adev->irq.ih_funcs = &vega20_ih_funcs;
+}
+
+const struct amdgpu_ip_block_version vega20_ih_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_IH,
+	.major = 4,
+	.minor = 2,
+	.rev = 0,
+	.funcs = &vega20_ih_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/vega20_ih.h b/drivers/gpu/drm/amd/amdgpu/vega20_ih.h
new file mode 100644
index 000000000000..7af6d8758ee3
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vega20_ih.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VEGA20_IH_H__
+#define __VEGA20_IH_H__
+
+extern const struct amd_ip_funcs vega20_ih_ip_funcs;
+extern const struct amdgpu_ip_block_version vega20_ih_ip_block;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/vega20_reg_init.c b/drivers/gpu/drm/amd/amdgpu/vega20_reg_init.c
new file mode 100644
index 000000000000..556f854e3551
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vega20_reg_init.c
@@ -0,0 +1,100 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "soc15.h"
+
+#include "soc15_common.h"
+#include "vega20_ip_offset.h"
+
+int vega20_reg_base_init(struct amdgpu_device *adev)
+{
+	/* HW has more IP blocks,  only initialized the blocke beend by our driver  */
+	uint32_t i;
+	for (i = 0 ; i < MAX_INSTANCE ; ++i) {
+		adev->reg_offset[GC_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i]));
+		adev->reg_offset[HDP_HWIP][i] = (uint32_t *)(&(HDP_BASE.instance[i]));
+		adev->reg_offset[MMHUB_HWIP][i] = (uint32_t *)(&(MMHUB_BASE.instance[i]));
+		adev->reg_offset[ATHUB_HWIP][i] = (uint32_t *)(&(ATHUB_BASE.instance[i]));
+		adev->reg_offset[NBIO_HWIP][i] = (uint32_t *)(&(NBIO_BASE.instance[i]));
+		adev->reg_offset[MP0_HWIP][i] = (uint32_t *)(&(MP0_BASE.instance[i]));
+		adev->reg_offset[MP1_HWIP][i] = (uint32_t *)(&(MP1_BASE.instance[i]));
+		adev->reg_offset[UVD_HWIP][i] = (uint32_t *)(&(UVD_BASE.instance[i]));
+		adev->reg_offset[VCE_HWIP][i] = (uint32_t *)(&(VCE_BASE.instance[i]));
+		adev->reg_offset[DF_HWIP][i] = (uint32_t *)(&(DF_BASE.instance[i]));
+		adev->reg_offset[DCE_HWIP][i] = (uint32_t *)(&(DCE_BASE.instance[i]));
+		adev->reg_offset[OSSSYS_HWIP][i] = (uint32_t *)(&(OSSSYS_BASE.instance[i]));
+		adev->reg_offset[SDMA0_HWIP][i] = (uint32_t *)(&(SDMA0_BASE.instance[i]));
+		adev->reg_offset[SDMA1_HWIP][i] = (uint32_t *)(&(SDMA1_BASE.instance[i]));
+		adev->reg_offset[SMUIO_HWIP][i] = (uint32_t *)(&(SMUIO_BASE.instance[i]));
+		adev->reg_offset[NBIF_HWIP][i] = (uint32_t *)(&(NBIO_BASE.instance[i]));
+		adev->reg_offset[THM_HWIP][i] = (uint32_t *)(&(THM_BASE.instance[i]));
+		adev->reg_offset[CLK_HWIP][i] = (uint32_t *)(&(CLK_BASE.instance[i]));
+		adev->reg_offset[UMC_HWIP][i] = (uint32_t *)(&(UMC_BASE.instance[i]));
+		adev->reg_offset[RSMU_HWIP][i] = (uint32_t *)(&(RSMU_BASE.instance[i]));
+	}
+	return 0;
+}
+
+void vega20_doorbell_index_init(struct amdgpu_device *adev)
+{
+	adev->doorbell_index.kiq = AMDGPU_VEGA20_DOORBELL_KIQ;
+	adev->doorbell_index.mec_ring0 = AMDGPU_VEGA20_DOORBELL_MEC_RING0;
+	adev->doorbell_index.mec_ring1 = AMDGPU_VEGA20_DOORBELL_MEC_RING1;
+	adev->doorbell_index.mec_ring2 = AMDGPU_VEGA20_DOORBELL_MEC_RING2;
+	adev->doorbell_index.mec_ring3 = AMDGPU_VEGA20_DOORBELL_MEC_RING3;
+	adev->doorbell_index.mec_ring4 = AMDGPU_VEGA20_DOORBELL_MEC_RING4;
+	adev->doorbell_index.mec_ring5 = AMDGPU_VEGA20_DOORBELL_MEC_RING5;
+	adev->doorbell_index.mec_ring6 = AMDGPU_VEGA20_DOORBELL_MEC_RING6;
+	adev->doorbell_index.mec_ring7 = AMDGPU_VEGA20_DOORBELL_MEC_RING7;
+	adev->doorbell_index.userqueue_start = AMDGPU_VEGA20_DOORBELL_USERQUEUE_START;
+	adev->doorbell_index.userqueue_end = AMDGPU_VEGA20_DOORBELL_USERQUEUE_END;
+	adev->doorbell_index.gfx_ring0 = AMDGPU_VEGA20_DOORBELL_GFX_RING0;
+	adev->doorbell_index.sdma_engine[0] = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE0;
+	adev->doorbell_index.sdma_engine[1] = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE1;
+	adev->doorbell_index.sdma_engine[2] = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE2;
+	adev->doorbell_index.sdma_engine[3] = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE3;
+	adev->doorbell_index.sdma_engine[4] = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE4;
+	adev->doorbell_index.sdma_engine[5] = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE5;
+	adev->doorbell_index.sdma_engine[6] = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE6;
+	adev->doorbell_index.sdma_engine[7] = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE7;
+	adev->doorbell_index.ih = AMDGPU_VEGA20_DOORBELL_IH;
+	adev->doorbell_index.uvd_vce.uvd_ring0_1 = AMDGPU_VEGA20_DOORBELL64_UVD_RING0_1;
+	adev->doorbell_index.uvd_vce.uvd_ring2_3 = AMDGPU_VEGA20_DOORBELL64_UVD_RING2_3;
+	adev->doorbell_index.uvd_vce.uvd_ring4_5 = AMDGPU_VEGA20_DOORBELL64_UVD_RING4_5;
+	adev->doorbell_index.uvd_vce.uvd_ring6_7 = AMDGPU_VEGA20_DOORBELL64_UVD_RING6_7;
+	adev->doorbell_index.uvd_vce.vce_ring0_1 = AMDGPU_VEGA20_DOORBELL64_VCE_RING0_1;
+	adev->doorbell_index.uvd_vce.vce_ring2_3 = AMDGPU_VEGA20_DOORBELL64_VCE_RING2_3;
+	adev->doorbell_index.uvd_vce.vce_ring4_5 = AMDGPU_VEGA20_DOORBELL64_VCE_RING4_5;
+	adev->doorbell_index.uvd_vce.vce_ring6_7 = AMDGPU_VEGA20_DOORBELL64_VCE_RING6_7;
+	adev->doorbell_index.vcn.vcn_ring0_1 = AMDGPU_VEGA20_DOORBELL64_VCN0_1;
+	adev->doorbell_index.vcn.vcn_ring2_3 = AMDGPU_VEGA20_DOORBELL64_VCN2_3;
+	adev->doorbell_index.vcn.vcn_ring4_5 = AMDGPU_VEGA20_DOORBELL64_VCN4_5;
+	adev->doorbell_index.vcn.vcn_ring6_7 = AMDGPU_VEGA20_DOORBELL64_VCN6_7;
+
+	adev->doorbell_index.first_non_cp = AMDGPU_VEGA20_DOORBELL64_FIRST_NON_CP;
+	adev->doorbell_index.last_non_cp = AMDGPU_VEGA20_DOORBELL64_LAST_NON_CP;
+
+	adev->doorbell_index.max_assignment = AMDGPU_VEGA20_DOORBELL_MAX_ASSIGNMENT << 1;
+	adev->doorbell_index.sdma_doorbell_range = 20;
+}
+
diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c b/drivers/gpu/drm/amd/amdgpu/vi.c
index 9ff69b90df36..a611a7345125 100644
--- a/drivers/gpu/drm/amd/amdgpu/vi.c
+++ b/drivers/gpu/drm/amd/amdgpu/vi.c
@@ -20,8 +20,12 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  *
  */
+
+#include <linux/pci.h>
 #include <linux/slab.h>
-#include <drm/drmP.h>
+
+#include <drm/amdgpu_drm.h>
+
 #include "amdgpu.h"
 #include "amdgpu_atombios.h"
 #include "amdgpu_ih.h"
@@ -57,26 +61,235 @@
 
 #include "vid.h"
 #include "vi.h"
-#include "vi_dpm.h"
 #include "gmc_v8_0.h"
 #include "gmc_v7_0.h"
 #include "gfx_v8_0.h"
 #include "sdma_v2_4.h"
 #include "sdma_v3_0.h"
 #include "dce_v10_0.h"
-#include "dce_v11_0.h"
 #include "iceland_ih.h"
 #include "tonga_ih.h"
 #include "cz_ih.h"
 #include "uvd_v5_0.h"
 #include "uvd_v6_0.h"
 #include "vce_v3_0.h"
-#include "amdgpu_powerplay.h"
 #if defined(CONFIG_DRM_AMD_ACP)
 #include "amdgpu_acp.h"
 #endif
-#include "dce_virtual.h"
+#include "amdgpu_vkms.h"
 #include "mxgpu_vi.h"
+#include "amdgpu_dm.h"
+
+#define ixPCIE_LC_L1_PM_SUBSTATE	0x100100C6
+#define PCIE_LC_L1_PM_SUBSTATE__LC_L1_SUBSTATES_OVERRIDE_EN_MASK	0x00000001L
+#define PCIE_LC_L1_PM_SUBSTATE__LC_PCI_PM_L1_2_OVERRIDE_MASK	0x00000002L
+#define PCIE_LC_L1_PM_SUBSTATE__LC_PCI_PM_L1_1_OVERRIDE_MASK	0x00000004L
+#define PCIE_LC_L1_PM_SUBSTATE__LC_ASPM_L1_2_OVERRIDE_MASK		0x00000008L
+#define PCIE_LC_L1_PM_SUBSTATE__LC_ASPM_L1_1_OVERRIDE_MASK		0x00000010L
+#define ixPCIE_L1_PM_SUB_CNTL	0x378
+#define PCIE_L1_PM_SUB_CNTL__ASPM_L1_2_EN_MASK	0x00000004L
+#define PCIE_L1_PM_SUB_CNTL__ASPM_L1_1_EN_MASK	0x00000008L
+#define PCIE_L1_PM_SUB_CNTL__PCI_PM_L1_2_EN_MASK	0x00000001L
+#define PCIE_L1_PM_SUB_CNTL__PCI_PM_L1_1_EN_MASK	0x00000002L
+#define PCIE_LC_CNTL6__LC_L1_POWERDOWN_MASK		0x00200000L
+#define LINK_CAP	0x64
+#define PCIE_LINK_CAP__CLOCK_POWER_MANAGEMENT_MASK	0x00040000L
+#define ixCPM_CONTROL	0x1400118
+#define ixPCIE_LC_CNTL7	0x100100BC
+#define PCIE_LC_CNTL7__LC_L1_SIDEBAND_CLKREQ_PDWN_EN_MASK	0x00000400L
+#define PCIE_LC_CNTL__LC_L0S_INACTIVITY_DEFAULT	0x00000007
+#define PCIE_LC_CNTL__LC_L1_INACTIVITY_DEFAULT	0x00000009
+#define CPM_CONTROL__CLKREQb_UNGATE_TXCLK_ENABLE_MASK	0x01000000L
+#define PCIE_L1_PM_SUB_CNTL	0x378
+#define ASIC_IS_P22(asic_type, rid)	((asic_type >= CHIP_POLARIS10) && \
+									(asic_type <= CHIP_POLARIS12) && \
+									(rid >= 0x6E))
+/* Topaz */
+static const struct amdgpu_video_codecs topaz_video_codecs_encode =
+{
+	.codec_count = 0,
+	.codec_array = NULL,
+};
+
+/* Tonga, CZ, ST, Fiji */
+static const struct amdgpu_video_codec_info tonga_video_codecs_encode_array[] =
+{
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC,
+		.max_width = 4096,
+		.max_height = 2304,
+		.max_pixels_per_frame = 4096 * 2304,
+		.max_level = 0,
+	},
+};
+
+static const struct amdgpu_video_codecs tonga_video_codecs_encode =
+{
+	.codec_count = ARRAY_SIZE(tonga_video_codecs_encode_array),
+	.codec_array = tonga_video_codecs_encode_array,
+};
+
+/* Polaris */
+static const struct amdgpu_video_codec_info polaris_video_codecs_encode_array[] =
+{
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC,
+		.max_width = 4096,
+		.max_height = 4096,
+		.max_pixels_per_frame = 4096 * 4096,
+		.max_level = 0,
+	},
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC,
+		.max_width = 4096,
+		.max_height = 4096,
+		.max_pixels_per_frame = 4096 * 4096,
+		.max_level = 0,
+	},
+};
+
+static const struct amdgpu_video_codecs polaris_video_codecs_encode =
+{
+	.codec_count = ARRAY_SIZE(polaris_video_codecs_encode_array),
+	.codec_array = polaris_video_codecs_encode_array,
+};
+
+/* Topaz */
+static const struct amdgpu_video_codecs topaz_video_codecs_decode =
+{
+	.codec_count = 0,
+	.codec_array = NULL,
+};
+
+/* Tonga */
+static const struct amdgpu_video_codec_info tonga_video_codecs_decode_array[] =
+{
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2,
+		.max_width = 1920,
+		.max_height = 1088,
+		.max_pixels_per_frame = 1920 * 1088,
+		.max_level = 3,
+	},
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4,
+		.max_width = 1920,
+		.max_height = 1088,
+		.max_pixels_per_frame = 1920 * 1088,
+		.max_level = 5,
+	},
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC,
+		.max_width = 4096,
+		.max_height = 4096,
+		.max_pixels_per_frame = 4096 * 4096,
+		.max_level = 52,
+	},
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1,
+		.max_width = 1920,
+		.max_height = 1088,
+		.max_pixels_per_frame = 1920 * 1088,
+		.max_level = 4,
+	},
+};
+
+static const struct amdgpu_video_codecs tonga_video_codecs_decode =
+{
+	.codec_count = ARRAY_SIZE(tonga_video_codecs_decode_array),
+	.codec_array = tonga_video_codecs_decode_array,
+};
+
+/* CZ, ST, Fiji, Polaris */
+static const struct amdgpu_video_codec_info cz_video_codecs_decode_array[] =
+{
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2,
+		.max_width = 1920,
+		.max_height = 1088,
+		.max_pixels_per_frame = 1920 * 1088,
+		.max_level = 3,
+	},
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4,
+		.max_width = 1920,
+		.max_height = 1088,
+		.max_pixels_per_frame = 1920 * 1088,
+		.max_level = 5,
+	},
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC,
+		.max_width = 4096,
+		.max_height = 4096,
+		.max_pixels_per_frame = 4096 * 4096,
+		.max_level = 52,
+	},
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1,
+		.max_width = 1920,
+		.max_height = 1088,
+		.max_pixels_per_frame = 1920 * 1088,
+		.max_level = 4,
+	},
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC,
+		.max_width = 4096,
+		.max_height = 4096,
+		.max_pixels_per_frame = 4096 * 4096,
+		.max_level = 186,
+	},
+	{
+		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG,
+		.max_width = 4096,
+		.max_height = 4096,
+		.max_pixels_per_frame = 4096 * 4096,
+		.max_level = 0,
+	},
+};
+
+static const struct amdgpu_video_codecs cz_video_codecs_decode =
+{
+	.codec_count = ARRAY_SIZE(cz_video_codecs_decode_array),
+	.codec_array = cz_video_codecs_decode_array,
+};
+
+static int vi_query_video_codecs(struct amdgpu_device *adev, bool encode,
+				 const struct amdgpu_video_codecs **codecs)
+{
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+		if (encode)
+			*codecs = &topaz_video_codecs_encode;
+		else
+			*codecs = &topaz_video_codecs_decode;
+		return 0;
+	case CHIP_TONGA:
+		if (encode)
+			*codecs = &tonga_video_codecs_encode;
+		else
+			*codecs = &tonga_video_codecs_decode;
+		return 0;
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
+		if (encode)
+			*codecs = &polaris_video_codecs_encode;
+		else
+			*codecs = &cz_video_codecs_decode;
+		return 0;
+	case CHIP_FIJI:
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+		if (encode)
+			*codecs = &tonga_video_codecs_encode;
+		else
+			*codecs = &cz_video_codecs_decode;
+		return 0;
+	default:
+		return -EINVAL;
+	}
+}
 
 /*
  * Indirect registers accessor
@@ -87,9 +300,9 @@ static u32 vi_pcie_rreg(struct amdgpu_device *adev, u32 reg)
 	u32 r;
 
 	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
-	WREG32(mmPCIE_INDEX, reg);
-	(void)RREG32(mmPCIE_INDEX);
-	r = RREG32(mmPCIE_DATA);
+	WREG32_NO_KIQ(mmPCIE_INDEX, reg);
+	(void)RREG32_NO_KIQ(mmPCIE_INDEX);
+	r = RREG32_NO_KIQ(mmPCIE_DATA);
 	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
 	return r;
 }
@@ -99,10 +312,10 @@ static void vi_pcie_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
 	unsigned long flags;
 
 	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
-	WREG32(mmPCIE_INDEX, reg);
-	(void)RREG32(mmPCIE_INDEX);
-	WREG32(mmPCIE_DATA, v);
-	(void)RREG32(mmPCIE_DATA);
+	WREG32_NO_KIQ(mmPCIE_INDEX, reg);
+	(void)RREG32_NO_KIQ(mmPCIE_INDEX);
+	WREG32_NO_KIQ(mmPCIE_DATA, v);
+	(void)RREG32_NO_KIQ(mmPCIE_DATA);
 	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
 }
 
@@ -112,8 +325,8 @@ static u32 vi_smc_rreg(struct amdgpu_device *adev, u32 reg)
 	u32 r;
 
 	spin_lock_irqsave(&adev->smc_idx_lock, flags);
-	WREG32(mmSMC_IND_INDEX_11, (reg));
-	r = RREG32(mmSMC_IND_DATA_11);
+	WREG32_NO_KIQ(mmSMC_IND_INDEX_11, (reg));
+	r = RREG32_NO_KIQ(mmSMC_IND_DATA_11);
 	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
 	return r;
 }
@@ -123,8 +336,8 @@ static void vi_smc_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
 	unsigned long flags;
 
 	spin_lock_irqsave(&adev->smc_idx_lock, flags);
-	WREG32(mmSMC_IND_INDEX_11, (reg));
-	WREG32(mmSMC_IND_DATA_11, (v));
+	WREG32_NO_KIQ(mmSMC_IND_INDEX_11, (reg));
+	WREG32_NO_KIQ(mmSMC_IND_DATA_11, (v));
 	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
 }
 
@@ -281,33 +494,34 @@ static void vi_init_golden_registers(struct amdgpu_device *adev)
 
 	switch (adev->asic_type) {
 	case CHIP_TOPAZ:
-		amdgpu_program_register_sequence(adev,
-						 iceland_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(iceland_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							iceland_mgcg_cgcg_init,
+							ARRAY_SIZE(iceland_mgcg_cgcg_init));
 		break;
 	case CHIP_FIJI:
-		amdgpu_program_register_sequence(adev,
-						 fiji_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(fiji_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							fiji_mgcg_cgcg_init,
+							ARRAY_SIZE(fiji_mgcg_cgcg_init));
 		break;
 	case CHIP_TONGA:
-		amdgpu_program_register_sequence(adev,
-						 tonga_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(tonga_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							tonga_mgcg_cgcg_init,
+							ARRAY_SIZE(tonga_mgcg_cgcg_init));
 		break;
 	case CHIP_CARRIZO:
-		amdgpu_program_register_sequence(adev,
-						 cz_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(cz_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							cz_mgcg_cgcg_init,
+							ARRAY_SIZE(cz_mgcg_cgcg_init));
 		break;
 	case CHIP_STONEY:
-		amdgpu_program_register_sequence(adev,
-						 stoney_mgcg_cgcg_init,
-						 (const u32)ARRAY_SIZE(stoney_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							stoney_mgcg_cgcg_init,
+							ARRAY_SIZE(stoney_mgcg_cgcg_init));
 		break;
-	case CHIP_POLARIS11:
 	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
 	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
 	default:
 		break;
 	}
@@ -327,8 +541,15 @@ static u32 vi_get_xclk(struct amdgpu_device *adev)
 	u32 reference_clock = adev->clock.spll.reference_freq;
 	u32 tmp;
 
-	if (adev->flags & AMD_IS_APU)
-		return reference_clock;
+	if (adev->flags & AMD_IS_APU) {
+		switch (adev->asic_type) {
+		case CHIP_STONEY:
+			/* vbios says 48Mhz, but the actual freq is 100Mhz */
+			return 10000;
+		default:
+			return reference_clock;
+		}
+	}
 
 	tmp = RREG32_SMC(ixCG_CLKPIN_CNTL_2);
 	if (REG_GET_FIELD(tmp, CG_CLKPIN_CNTL_2, MUX_TCLK_TO_XCLK))
@@ -365,11 +586,6 @@ void vi_srbm_select(struct amdgpu_device *adev,
 	WREG32(mmSRBM_GFX_CNTL, srbm_gfx_cntl);
 }
 
-static void vi_vga_set_state(struct amdgpu_device *adev, bool state)
-{
-	/* todo */
-}
-
 static bool vi_read_disabled_bios(struct amdgpu_device *adev)
 {
 	u32 bus_cntl;
@@ -446,23 +662,6 @@ static bool vi_read_bios_from_rom(struct amdgpu_device *adev,
 	return true;
 }
 
-static void vi_detect_hw_virtualization(struct amdgpu_device *adev)
-{
-	uint32_t reg = RREG32(mmBIF_IOV_FUNC_IDENTIFIER);
-	/* bit0: 0 means pf and 1 means vf */
-	/* bit31: 0 means disable IOV and 1 means enable */
-	if (reg & 1)
-		adev->virt.caps |= AMDGPU_SRIOV_CAPS_IS_VF;
-
-	if (reg & 0x80000000)
-		adev->virt.caps |= AMDGPU_SRIOV_CAPS_ENABLE_IOV;
-
-	if (reg == 0) {
-		if (is_virtual_machine()) /* passthrough mode exclus sr-iov mode */
-			adev->virt.caps |= AMDGPU_PASSTHROUGH_MODE;
-	}
-}
-
 static const struct amdgpu_allowed_register_entry vi_allowed_read_registers[] = {
 	{mmGRBM_STATUS},
 	{mmGRBM_STATUS2},
@@ -564,12 +763,12 @@ static uint32_t vi_get_register_value(struct amdgpu_device *adev,
 
 		mutex_lock(&adev->grbm_idx_mutex);
 		if (se_num != 0xffffffff || sh_num != 0xffffffff)
-			amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff);
+			amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff, 0);
 
 		val = RREG32(reg_offset);
 
 		if (se_num != 0xffffffff || sh_num != 0xffffffff)
-			amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+			amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 		mutex_unlock(&adev->grbm_idx_mutex);
 		return val;
 	} else {
@@ -657,16 +856,26 @@ static int vi_read_register(struct amdgpu_device *adev, u32 se_num,
 	return -EINVAL;
 }
 
-static int vi_gpu_pci_config_reset(struct amdgpu_device *adev)
+/**
+ * vi_asic_pci_config_reset - soft reset GPU
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Use PCI Config method to reset the GPU.
+ *
+ * Returns 0 for success.
+ */
+static int vi_asic_pci_config_reset(struct amdgpu_device *adev)
 {
 	u32 i;
+	int r = -EINVAL;
 
-	dev_info(adev->dev, "GPU pci config reset\n");
+	amdgpu_atombios_scratch_regs_engine_hung(adev, true);
 
 	/* disable BM */
 	pci_clear_master(adev->pdev);
 	/* reset */
-	amdgpu_pci_config_reset(adev);
+	amdgpu_device_pci_config_reset(adev);
 
 	udelay(100);
 
@@ -676,11 +885,63 @@ static int vi_gpu_pci_config_reset(struct amdgpu_device *adev)
 			/* enable BM */
 			pci_set_master(adev->pdev);
 			adev->has_hw_reset = true;
-			return 0;
+			r = 0;
+			break;
 		}
 		udelay(1);
 	}
-	return -EINVAL;
+
+	amdgpu_atombios_scratch_regs_engine_hung(adev, false);
+
+	return r;
+}
+
+static int vi_asic_supports_baco(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_FIJI:
+	case CHIP_TONGA:
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_TOPAZ:
+		return amdgpu_dpm_is_baco_supported(adev);
+	default:
+		return 0;
+	}
+}
+
+static enum amd_reset_method
+vi_asic_reset_method(struct amdgpu_device *adev)
+{
+	int baco_reset;
+
+	if (amdgpu_reset_method == AMD_RESET_METHOD_LEGACY ||
+	    amdgpu_reset_method == AMD_RESET_METHOD_BACO)
+		return amdgpu_reset_method;
+
+	if (amdgpu_reset_method != -1)
+		dev_warn(adev->dev, "Specified reset method:%d isn't supported, using AUTO instead.\n",
+				  amdgpu_reset_method);
+
+	switch (adev->asic_type) {
+	case CHIP_FIJI:
+	case CHIP_TONGA:
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_TOPAZ:
+		baco_reset = amdgpu_dpm_is_baco_supported(adev);
+		break;
+	default:
+		baco_reset = 0;
+		break;
+	}
+
+	if (baco_reset)
+		return AMD_RESET_METHOD_BACO;
+	else
+		return AMD_RESET_METHOD_LEGACY;
 }
 
 /**
@@ -696,11 +957,17 @@ static int vi_asic_reset(struct amdgpu_device *adev)
 {
 	int r;
 
-	amdgpu_atombios_scratch_regs_engine_hung(adev, true);
-
-	r = vi_gpu_pci_config_reset(adev);
+	/* APUs don't have full asic reset */
+	if (adev->flags & AMD_IS_APU)
+		return 0;
 
-	amdgpu_atombios_scratch_regs_engine_hung(adev, false);
+	if (vi_asic_reset_method(adev) == AMD_RESET_METHOD_BACO) {
+		dev_info(adev->dev, "BACO reset\n");
+		r = amdgpu_dpm_baco_reset(adev);
+	} else {
+		dev_info(adev->dev, "PCI CONFIG reset\n");
+		r = vi_asic_pci_config_reset(adev);
+	}
 
 	return r;
 }
@@ -724,33 +991,59 @@ static int vi_set_uvd_clock(struct amdgpu_device *adev, u32 clock,
 		return r;
 
 	tmp = RREG32_SMC(cntl_reg);
-	tmp &= ~(CG_DCLK_CNTL__DCLK_DIR_CNTL_EN_MASK |
-		CG_DCLK_CNTL__DCLK_DIVIDER_MASK);
+
+	if (adev->flags & AMD_IS_APU)
+		tmp &= ~CG_DCLK_CNTL__DCLK_DIVIDER_MASK;
+	else
+		tmp &= ~(CG_DCLK_CNTL__DCLK_DIR_CNTL_EN_MASK |
+				CG_DCLK_CNTL__DCLK_DIVIDER_MASK);
 	tmp |= dividers.post_divider;
 	WREG32_SMC(cntl_reg, tmp);
 
 	for (i = 0; i < 100; i++) {
-		if (RREG32_SMC(status_reg) & CG_DCLK_STATUS__DCLK_STATUS_MASK)
-			break;
+		tmp = RREG32_SMC(status_reg);
+		if (adev->flags & AMD_IS_APU) {
+			if (tmp & 0x10000)
+				break;
+		} else {
+			if (tmp & CG_DCLK_STATUS__DCLK_STATUS_MASK)
+				break;
+		}
 		mdelay(10);
 	}
 	if (i == 100)
 		return -ETIMEDOUT;
-
 	return 0;
 }
 
+#define ixGNB_CLK1_DFS_CNTL 0xD82200F0
+#define ixGNB_CLK1_STATUS   0xD822010C
+#define ixGNB_CLK2_DFS_CNTL 0xD8220110
+#define ixGNB_CLK2_STATUS   0xD822012C
+#define ixGNB_CLK3_DFS_CNTL 0xD8220130
+#define ixGNB_CLK3_STATUS   0xD822014C
+
 static int vi_set_uvd_clocks(struct amdgpu_device *adev, u32 vclk, u32 dclk)
 {
 	int r;
 
-	r = vi_set_uvd_clock(adev, vclk, ixCG_VCLK_CNTL, ixCG_VCLK_STATUS);
-	if (r)
-		return r;
+	if (adev->flags & AMD_IS_APU) {
+		r = vi_set_uvd_clock(adev, vclk, ixGNB_CLK2_DFS_CNTL, ixGNB_CLK2_STATUS);
+		if (r)
+			return r;
 
-	r = vi_set_uvd_clock(adev, dclk, ixCG_DCLK_CNTL, ixCG_DCLK_STATUS);
-	if (r)
-		return r;
+		r = vi_set_uvd_clock(adev, dclk, ixGNB_CLK1_DFS_CNTL, ixGNB_CLK1_STATUS);
+		if (r)
+			return r;
+	} else {
+		r = vi_set_uvd_clock(adev, vclk, ixCG_VCLK_CNTL, ixCG_VCLK_STATUS);
+		if (r)
+			return r;
+
+		r = vi_set_uvd_clock(adev, dclk, ixCG_DCLK_CNTL, ixCG_DCLK_STATUS);
+		if (r)
+			return r;
+	}
 
 	return 0;
 }
@@ -760,6 +1053,22 @@ static int vi_set_vce_clocks(struct amdgpu_device *adev, u32 evclk, u32 ecclk)
 	int r, i;
 	struct atom_clock_dividers dividers;
 	u32 tmp;
+	u32 reg_ctrl;
+	u32 reg_status;
+	u32 status_mask;
+	u32 reg_mask;
+
+	if (adev->flags & AMD_IS_APU) {
+		reg_ctrl = ixGNB_CLK3_DFS_CNTL;
+		reg_status = ixGNB_CLK3_STATUS;
+		status_mask = 0x00010000;
+		reg_mask = CG_ECLK_CNTL__ECLK_DIVIDER_MASK;
+	} else {
+		reg_ctrl = ixCG_ECLK_CNTL;
+		reg_status = ixCG_ECLK_STATUS;
+		status_mask = CG_ECLK_STATUS__ECLK_STATUS_MASK;
+		reg_mask = CG_ECLK_CNTL__ECLK_DIR_CNTL_EN_MASK | CG_ECLK_CNTL__ECLK_DIVIDER_MASK;
+	}
 
 	r = amdgpu_atombios_get_clock_dividers(adev,
 					       COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
@@ -768,55 +1077,202 @@ static int vi_set_vce_clocks(struct amdgpu_device *adev, u32 evclk, u32 ecclk)
 		return r;
 
 	for (i = 0; i < 100; i++) {
-		if (RREG32_SMC(ixCG_ECLK_STATUS) & CG_ECLK_STATUS__ECLK_STATUS_MASK)
+		if (RREG32_SMC(reg_status) & status_mask)
 			break;
 		mdelay(10);
 	}
+
 	if (i == 100)
 		return -ETIMEDOUT;
 
-	tmp = RREG32_SMC(ixCG_ECLK_CNTL);
-	tmp &= ~(CG_ECLK_CNTL__ECLK_DIR_CNTL_EN_MASK |
-		CG_ECLK_CNTL__ECLK_DIVIDER_MASK);
+	tmp = RREG32_SMC(reg_ctrl);
+	tmp &= ~reg_mask;
 	tmp |= dividers.post_divider;
-	WREG32_SMC(ixCG_ECLK_CNTL, tmp);
+	WREG32_SMC(reg_ctrl, tmp);
 
 	for (i = 0; i < 100; i++) {
-		if (RREG32_SMC(ixCG_ECLK_STATUS) & CG_ECLK_STATUS__ECLK_STATUS_MASK)
+		if (RREG32_SMC(reg_status) & status_mask)
 			break;
 		mdelay(10);
 	}
+
 	if (i == 100)
 		return -ETIMEDOUT;
 
 	return 0;
 }
 
-static void vi_pcie_gen3_enable(struct amdgpu_device *adev)
+static void vi_enable_aspm(struct amdgpu_device *adev)
 {
-	if (pci_is_root_bus(adev->pdev->bus))
-		return;
+	u32 data, orig;
+
+	orig = data = RREG32_PCIE(ixPCIE_LC_CNTL);
+	data |= PCIE_LC_CNTL__LC_L0S_INACTIVITY_DEFAULT <<
+			PCIE_LC_CNTL__LC_L0S_INACTIVITY__SHIFT;
+	data |= PCIE_LC_CNTL__LC_L1_INACTIVITY_DEFAULT <<
+			PCIE_LC_CNTL__LC_L1_INACTIVITY__SHIFT;
+	data &= ~PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK;
+	data |= PCIE_LC_CNTL__LC_DELAY_L1_EXIT_MASK;
+	if (orig != data)
+		WREG32_PCIE(ixPCIE_LC_CNTL, data);
+}
 
-	if (amdgpu_pcie_gen2 == 0)
-		return;
+static void vi_program_aspm(struct amdgpu_device *adev)
+{
+	u32 data, data1, orig;
+	bool bL1SS = false;
+	bool bClkReqSupport = true;
 
-	if (adev->flags & AMD_IS_APU)
+	if (!amdgpu_device_should_use_aspm(adev))
 		return;
 
-	if (!(adev->pm.pcie_gen_mask & (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2 |
-					CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3)))
+	if (adev->asic_type < CHIP_POLARIS10)
 		return;
 
-	/* todo */
-}
+	orig = data = RREG32_PCIE(ixPCIE_LC_CNTL);
+	data &= ~PCIE_LC_CNTL__LC_L1_INACTIVITY_MASK;
+	data &= ~PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK;
+	data |= PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK;
+	if (orig != data)
+		WREG32_PCIE(ixPCIE_LC_CNTL, data);
+
+	orig = data = RREG32_PCIE(ixPCIE_LC_N_FTS_CNTL);
+	data &= ~PCIE_LC_N_FTS_CNTL__LC_XMIT_N_FTS_MASK;
+	data |= 0x0024 << PCIE_LC_N_FTS_CNTL__LC_XMIT_N_FTS__SHIFT;
+	data |= PCIE_LC_N_FTS_CNTL__LC_XMIT_N_FTS_OVERRIDE_EN_MASK;
+	if (orig != data)
+		WREG32_PCIE(ixPCIE_LC_N_FTS_CNTL, data);
+
+	orig = data = RREG32_PCIE(ixPCIE_LC_CNTL3);
+	data |= PCIE_LC_CNTL3__LC_GO_TO_RECOVERY_MASK;
+	if (orig != data)
+		WREG32_PCIE(ixPCIE_LC_CNTL3, data);
+
+	orig = data = RREG32_PCIE(ixPCIE_P_CNTL);
+	data |= PCIE_P_CNTL__P_IGNORE_EDB_ERR_MASK;
+	if (orig != data)
+		WREG32_PCIE(ixPCIE_P_CNTL, data);
+
+	data = RREG32_PCIE(ixPCIE_LC_L1_PM_SUBSTATE);
+	pci_read_config_dword(adev->pdev, PCIE_L1_PM_SUB_CNTL, &data1);
+	if (data & PCIE_LC_L1_PM_SUBSTATE__LC_L1_SUBSTATES_OVERRIDE_EN_MASK &&
+	    (data & (PCIE_LC_L1_PM_SUBSTATE__LC_PCI_PM_L1_2_OVERRIDE_MASK |
+		    PCIE_LC_L1_PM_SUBSTATE__LC_PCI_PM_L1_1_OVERRIDE_MASK |
+			PCIE_LC_L1_PM_SUBSTATE__LC_ASPM_L1_2_OVERRIDE_MASK |
+			PCIE_LC_L1_PM_SUBSTATE__LC_ASPM_L1_1_OVERRIDE_MASK))) {
+		bL1SS = true;
+	} else if (data1 & (PCIE_L1_PM_SUB_CNTL__ASPM_L1_2_EN_MASK |
+	    PCIE_L1_PM_SUB_CNTL__ASPM_L1_1_EN_MASK |
+	    PCIE_L1_PM_SUB_CNTL__PCI_PM_L1_2_EN_MASK |
+	    PCIE_L1_PM_SUB_CNTL__PCI_PM_L1_1_EN_MASK)) {
+		bL1SS = true;
+	}
 
-static void vi_program_aspm(struct amdgpu_device *adev)
-{
+	orig = data = RREG32_PCIE(ixPCIE_LC_CNTL6);
+	data |= PCIE_LC_CNTL6__LC_L1_POWERDOWN_MASK;
+	if (orig != data)
+		WREG32_PCIE(ixPCIE_LC_CNTL6, data);
+
+	orig = data = RREG32_PCIE(ixPCIE_LC_LINK_WIDTH_CNTL);
+	data |= PCIE_LC_LINK_WIDTH_CNTL__LC_DYN_LANES_PWR_STATE_MASK;
+	if (orig != data)
+		WREG32_PCIE(ixPCIE_LC_LINK_WIDTH_CNTL, data);
+
+	pci_read_config_dword(adev->pdev, LINK_CAP, &data);
+	if (!(data & PCIE_LINK_CAP__CLOCK_POWER_MANAGEMENT_MASK))
+		bClkReqSupport = false;
+
+	if (bClkReqSupport) {
+		orig = data = RREG32_SMC(ixTHM_CLK_CNTL);
+		data &= ~(THM_CLK_CNTL__CMON_CLK_SEL_MASK | THM_CLK_CNTL__TMON_CLK_SEL_MASK);
+		data |= (1 << THM_CLK_CNTL__CMON_CLK_SEL__SHIFT) |
+				(1 << THM_CLK_CNTL__TMON_CLK_SEL__SHIFT);
+		if (orig != data)
+			WREG32_SMC(ixTHM_CLK_CNTL, data);
+
+		orig = data = RREG32_SMC(ixMISC_CLK_CTRL);
+		data &= ~(MISC_CLK_CTRL__DEEP_SLEEP_CLK_SEL_MASK |
+			MISC_CLK_CTRL__ZCLK_SEL_MASK | MISC_CLK_CTRL__DFT_SMS_PG_CLK_SEL_MASK);
+		data |= (1 << MISC_CLK_CTRL__DEEP_SLEEP_CLK_SEL__SHIFT) |
+				(1 << MISC_CLK_CTRL__ZCLK_SEL__SHIFT);
+		data |= (0x20 << MISC_CLK_CTRL__DFT_SMS_PG_CLK_SEL__SHIFT);
+		if (orig != data)
+			WREG32_SMC(ixMISC_CLK_CTRL, data);
+
+		orig = data = RREG32_SMC(ixCG_CLKPIN_CNTL);
+		data |= CG_CLKPIN_CNTL__XTALIN_DIVIDE_MASK;
+		if (orig != data)
+			WREG32_SMC(ixCG_CLKPIN_CNTL, data);
+
+		orig = data = RREG32_SMC(ixCG_CLKPIN_CNTL_2);
+		data |= CG_CLKPIN_CNTL_2__ENABLE_XCLK_MASK;
+		if (orig != data)
+			WREG32_SMC(ixCG_CLKPIN_CNTL, data);
+
+		orig = data = RREG32_SMC(ixMPLL_BYPASSCLK_SEL);
+		data &= ~MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL_MASK;
+		data |= (4 << MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL__SHIFT);
+		if (orig != data)
+			WREG32_SMC(ixMPLL_BYPASSCLK_SEL, data);
+
+		orig = data = RREG32_PCIE(ixCPM_CONTROL);
+		data |= (CPM_CONTROL__REFCLK_XSTCLK_ENABLE_MASK |
+				CPM_CONTROL__CLKREQb_UNGATE_TXCLK_ENABLE_MASK);
+		if (orig != data)
+			WREG32_PCIE(ixCPM_CONTROL, data);
+
+		orig = data = RREG32_PCIE(ixPCIE_CONFIG_CNTL);
+		data &= ~PCIE_CONFIG_CNTL__DYN_CLK_LATENCY_MASK;
+		data |= (0xE << PCIE_CONFIG_CNTL__DYN_CLK_LATENCY__SHIFT);
+		if (orig != data)
+			WREG32_PCIE(ixPCIE_CONFIG_CNTL, data);
+
+		orig = data = RREG32(mmBIF_CLK_CTRL);
+		data |= BIF_CLK_CTRL__BIF_XSTCLK_READY_MASK;
+		if (orig != data)
+			WREG32(mmBIF_CLK_CTRL, data);
+
+		orig = data = RREG32_PCIE(ixPCIE_LC_CNTL7);
+		data |= PCIE_LC_CNTL7__LC_L1_SIDEBAND_CLKREQ_PDWN_EN_MASK;
+		if (orig != data)
+			WREG32_PCIE(ixPCIE_LC_CNTL7, data);
+
+		orig = data = RREG32_PCIE(ixPCIE_HW_DEBUG);
+		data |= PCIE_HW_DEBUG__HW_01_DEBUG_MASK;
+		if (orig != data)
+			WREG32_PCIE(ixPCIE_HW_DEBUG, data);
+
+		orig = data = RREG32_PCIE(ixPCIE_LC_CNTL2);
+		data |= PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L23_MASK;
+		data |= PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L1_MASK;
+		if (bL1SS)
+			data &= ~PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L1_MASK;
+		if (orig != data)
+			WREG32_PCIE(ixPCIE_LC_CNTL2, data);
 
-	if (amdgpu_aspm == 0)
-		return;
+	}
+
+	vi_enable_aspm(adev);
+
+	data = RREG32_PCIE(ixPCIE_LC_N_FTS_CNTL);
+	data1 = RREG32_PCIE(ixPCIE_LC_STATUS1);
+	if (((data & PCIE_LC_N_FTS_CNTL__LC_N_FTS_MASK) == PCIE_LC_N_FTS_CNTL__LC_N_FTS_MASK) &&
+	    data1 & PCIE_LC_STATUS1__LC_REVERSE_XMIT_MASK &&
+	    data1 & PCIE_LC_STATUS1__LC_REVERSE_RCVR_MASK) {
+		orig = data = RREG32_PCIE(ixPCIE_LC_CNTL);
+		data &= ~PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK;
+		if (orig != data)
+			WREG32_PCIE(ixPCIE_LC_CNTL, data);
+	}
 
-	/* todo */
+	if ((adev->asic_type == CHIP_POLARIS12 &&
+	    !(ASICID_IS_P23(adev->pdev->device, adev->pdev->revision))) ||
+	    ASIC_IS_P22(adev->asic_type, adev->external_rev_id)) {
+		orig = data = RREG32_PCIE(ixPCIE_LC_TRAINING_CNTL);
+		data &= ~PCIE_LC_TRAINING_CNTL__LC_DISABLE_TRAINING_BIT_ARCH_MASK;
+		if (orig != data)
+			WREG32_PCIE(ixPCIE_LC_TRAINING_CNTL, data);
+	}
 }
 
 static void vi_enable_doorbell_aperture(struct amdgpu_device *adev,
@@ -851,26 +1307,156 @@ static uint32_t vi_get_rev_id(struct amdgpu_device *adev)
 			>> PCIE_EFUSE4__STRAP_BIF_ATI_REV_ID__SHIFT;
 }
 
+static void vi_flush_hdp(struct amdgpu_device *adev, struct amdgpu_ring *ring)
+{
+	if (!ring || !ring->funcs->emit_wreg) {
+		WREG32(mmHDP_MEM_COHERENCY_FLUSH_CNTL, 1);
+		RREG32(mmHDP_MEM_COHERENCY_FLUSH_CNTL);
+	} else {
+		amdgpu_ring_emit_wreg(ring, mmHDP_MEM_COHERENCY_FLUSH_CNTL, 1);
+	}
+}
+
+static void vi_invalidate_hdp(struct amdgpu_device *adev,
+			      struct amdgpu_ring *ring)
+{
+	if (!ring || !ring->funcs->emit_wreg) {
+		WREG32(mmHDP_DEBUG0, 1);
+		RREG32(mmHDP_DEBUG0);
+	} else {
+		amdgpu_ring_emit_wreg(ring, mmHDP_DEBUG0, 1);
+	}
+}
+
+static bool vi_need_full_reset(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+		/* CZ has hang issues with full reset at the moment */
+		return false;
+	case CHIP_FIJI:
+	case CHIP_TONGA:
+		/* XXX: soft reset should work on fiji and tonga */
+		return true;
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_TOPAZ:
+	default:
+		/* change this when we support soft reset */
+		return true;
+	}
+}
+
+static void vi_get_pcie_usage(struct amdgpu_device *adev, uint64_t *count0,
+			      uint64_t *count1)
+{
+	uint32_t perfctr = 0;
+	uint64_t cnt0_of, cnt1_of;
+	int tmp;
+
+	/* This reports 0 on APUs, so return to avoid writing/reading registers
+	 * that may or may not be different from their GPU counterparts
+	 */
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	/* Set the 2 events that we wish to watch, defined above */
+	/* Reg 40 is # received msgs, Reg 104 is # of posted requests sent */
+	perfctr = REG_SET_FIELD(perfctr, PCIE_PERF_CNTL_TXCLK, EVENT0_SEL, 40);
+	perfctr = REG_SET_FIELD(perfctr, PCIE_PERF_CNTL_TXCLK, EVENT1_SEL, 104);
+
+	/* Write to enable desired perf counters */
+	WREG32_PCIE(ixPCIE_PERF_CNTL_TXCLK, perfctr);
+	/* Zero out and enable the perf counters
+	 * Write 0x5:
+	 * Bit 0 = Start all counters(1)
+	 * Bit 2 = Global counter reset enable(1)
+	 */
+	WREG32_PCIE(ixPCIE_PERF_COUNT_CNTL, 0x00000005);
+
+	msleep(1000);
+
+	/* Load the shadow and disable the perf counters
+	 * Write 0x2:
+	 * Bit 0 = Stop counters(0)
+	 * Bit 1 = Load the shadow counters(1)
+	 */
+	WREG32_PCIE(ixPCIE_PERF_COUNT_CNTL, 0x00000002);
+
+	/* Read register values to get any >32bit overflow */
+	tmp = RREG32_PCIE(ixPCIE_PERF_CNTL_TXCLK);
+	cnt0_of = REG_GET_FIELD(tmp, PCIE_PERF_CNTL_TXCLK, COUNTER0_UPPER);
+	cnt1_of = REG_GET_FIELD(tmp, PCIE_PERF_CNTL_TXCLK, COUNTER1_UPPER);
+
+	/* Get the values and add the overflow */
+	*count0 = RREG32_PCIE(ixPCIE_PERF_COUNT0_TXCLK) | (cnt0_of << 32);
+	*count1 = RREG32_PCIE(ixPCIE_PERF_COUNT1_TXCLK) | (cnt1_of << 32);
+}
+
+static uint64_t vi_get_pcie_replay_count(struct amdgpu_device *adev)
+{
+	uint64_t nak_r, nak_g;
+
+	/* Get the number of NAKs received and generated */
+	nak_r = RREG32_PCIE(ixPCIE_RX_NUM_NAK);
+	nak_g = RREG32_PCIE(ixPCIE_RX_NUM_NAK_GENERATED);
+
+	/* Add the total number of NAKs, i.e the number of replays */
+	return (nak_r + nak_g);
+}
+
+static bool vi_need_reset_on_init(struct amdgpu_device *adev)
+{
+	u32 clock_cntl, pc;
+
+	if (adev->flags & AMD_IS_APU)
+		return false;
+
+	/* check if the SMC is already running */
+	clock_cntl = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+	pc = RREG32_SMC(ixSMC_PC_C);
+	if ((0 == REG_GET_FIELD(clock_cntl, SMC_SYSCON_CLOCK_CNTL_0, ck_disable)) &&
+	    (0x20100 <= pc))
+		return true;
+
+	return false;
+}
+
+static void vi_pre_asic_init(struct amdgpu_device *adev)
+{
+}
+
 static const struct amdgpu_asic_funcs vi_asic_funcs =
 {
 	.read_disabled_bios = &vi_read_disabled_bios,
 	.read_bios_from_rom = &vi_read_bios_from_rom,
 	.read_register = &vi_read_register,
 	.reset = &vi_asic_reset,
-	.set_vga_state = &vi_vga_set_state,
+	.reset_method = &vi_asic_reset_method,
 	.get_xclk = &vi_get_xclk,
 	.set_uvd_clocks = &vi_set_uvd_clocks,
 	.set_vce_clocks = &vi_set_vce_clocks,
 	.get_config_memsize = &vi_get_config_memsize,
+	.flush_hdp = &vi_flush_hdp,
+	.invalidate_hdp = &vi_invalidate_hdp,
+	.need_full_reset = &vi_need_full_reset,
+	.init_doorbell_index = &legacy_doorbell_index_init,
+	.get_pcie_usage = &vi_get_pcie_usage,
+	.need_reset_on_init = &vi_need_reset_on_init,
+	.get_pcie_replay_count = &vi_get_pcie_replay_count,
+	.supports_baco = &vi_asic_supports_baco,
+	.pre_asic_init = &vi_pre_asic_init,
+	.query_video_codecs = &vi_query_video_codecs,
 };
 
 #define CZ_REV_BRISTOL(rev)	 \
 	((rev >= 0xC8 && rev <= 0xCE) || (rev >= 0xE1 && rev <= 0xE6))
 
-static int vi_common_early_init(void *handle)
+static int vi_common_early_init(struct amdgpu_ip_block *ip_block)
 {
-	bool smc_enabled = false;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (adev->flags & AMD_IS_APU) {
 		adev->smc_rreg = &cz_smc_rreg;
@@ -890,10 +1476,6 @@ static int vi_common_early_init(void *handle)
 
 	adev->asic_funcs = &vi_asic_funcs;
 
-	if (amdgpu_get_ip_block(adev, AMD_IP_BLOCK_TYPE_SMC) &&
-		(amdgpu_ip_block_mask & (1 << AMD_IP_BLOCK_TYPE_SMC)))
-		smc_enabled = true;
-
 	adev->rev_id = vi_get_rev_id(adev);
 	adev->external_rev_id = 0xFF;
 	switch (adev->asic_type) {
@@ -1009,6 +1591,30 @@ static int vi_common_early_init(void *handle)
 		adev->pg_flags = 0;
 		adev->external_rev_id = adev->rev_id + 0x64;
 		break;
+	case CHIP_VEGAM:
+		adev->cg_flags = 0;
+			/*AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_RLC_LS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGLS |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_BIF_MGCG |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_ROM_MGCG |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_DRM_LS |
+			AMD_CG_SUPPORT_UVD_MGCG |
+			AMD_CG_SUPPORT_VCE_MGCG;*/
+		adev->pg_flags = 0;
+		adev->external_rev_id = adev->rev_id + 0x6E;
+		break;
 	case CHIP_CARRIZO:
 		adev->cg_flags = AMD_CG_SUPPORT_UVD_MGCG |
 			AMD_CG_SUPPORT_GFX_MGCG |
@@ -1044,7 +1650,6 @@ static int vi_common_early_init(void *handle)
 			AMD_CG_SUPPORT_GFX_CP_LS |
 			AMD_CG_SUPPORT_GFX_CGTS |
 			AMD_CG_SUPPORT_GFX_CGTS_LS |
-			AMD_CG_SUPPORT_GFX_CGCG |
 			AMD_CG_SUPPORT_GFX_CGLS |
 			AMD_CG_SUPPORT_BIF_LS |
 			AMD_CG_SUPPORT_HDP_MGCG |
@@ -1070,17 +1675,12 @@ static int vi_common_early_init(void *handle)
 		xgpu_vi_mailbox_set_irq_funcs(adev);
 	}
 
-	/* vi use smc load by default */
-	adev->firmware.load_type = amdgpu_ucode_get_load_type(adev, amdgpu_fw_load_type);
-
-	amdgpu_get_pcie_info(adev);
-
 	return 0;
 }
 
-static int vi_common_late_init(void *handle)
+static int vi_common_late_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (amdgpu_sriov_vf(adev))
 		xgpu_vi_mailbox_get_irq(adev);
@@ -1088,9 +1688,9 @@ static int vi_common_late_init(void *handle)
 	return 0;
 }
 
-static int vi_common_sw_init(void *handle)
+static int vi_common_sw_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (amdgpu_sriov_vf(adev))
 		xgpu_vi_mailbox_add_irq_id(adev);
@@ -1098,19 +1698,12 @@ static int vi_common_sw_init(void *handle)
 	return 0;
 }
 
-static int vi_common_sw_fini(void *handle)
-{
-	return 0;
-}
-
-static int vi_common_hw_init(void *handle)
+static int vi_common_hw_init(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	/* move the golden regs per IP block */
 	vi_init_golden_registers(adev);
-	/* enable pcie gen2/3 link */
-	vi_pcie_gen3_enable(adev);
 	/* enable aspm */
 	vi_program_aspm(adev);
 	/* enable the doorbell aperture */
@@ -1119,9 +1712,9 @@ static int vi_common_hw_init(void *handle)
 	return 0;
 }
 
-static int vi_common_hw_fini(void *handle)
+static int vi_common_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	/* enable the doorbell aperture */
 	vi_enable_doorbell_aperture(adev, false);
@@ -1132,35 +1725,21 @@ static int vi_common_hw_fini(void *handle)
 	return 0;
 }
 
-static int vi_common_suspend(void *handle)
+static int vi_common_suspend(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return vi_common_hw_fini(adev);
+	return vi_common_hw_fini(ip_block);
 }
 
-static int vi_common_resume(void *handle)
+static int vi_common_resume(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	return vi_common_hw_init(adev);
+	return vi_common_hw_init(ip_block);
 }
 
-static bool vi_common_is_idle(void *handle)
+static bool vi_common_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	return true;
 }
 
-static int vi_common_wait_for_idle(void *handle)
-{
-	return 0;
-}
-
-static int vi_common_soft_reset(void *handle)
-{
-	return 0;
-}
-
 static void vi_update_bif_medium_grain_light_sleep(struct amdgpu_device *adev,
 						   bool enable)
 {
@@ -1254,15 +1833,14 @@ static int vi_common_set_clockgating_state_by_smu(void *handle,
 	uint32_t msg_id, pp_state = 0;
 	uint32_t pp_support_state = 0;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	void *pp_handle = adev->powerplay.pp_handle;
 
 	if (adev->cg_flags & (AMD_CG_SUPPORT_MC_LS | AMD_CG_SUPPORT_MC_MGCG)) {
 		if (adev->cg_flags & AMD_CG_SUPPORT_MC_LS) {
-			pp_support_state = AMD_CG_SUPPORT_MC_LS;
+			pp_support_state = PP_STATE_SUPPORT_LS;
 			pp_state = PP_STATE_LS;
 		}
 		if (adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG) {
-			pp_support_state |= AMD_CG_SUPPORT_MC_MGCG;
+			pp_support_state |= PP_STATE_SUPPORT_CG;
 			pp_state |= PP_STATE_CG;
 		}
 		if (state == AMD_CG_STATE_UNGATE)
@@ -1271,16 +1849,16 @@ static int vi_common_set_clockgating_state_by_smu(void *handle,
 			       PP_BLOCK_SYS_MC,
 			       pp_support_state,
 			       pp_state);
-		amd_set_clockgating_by_smu(pp_handle, msg_id);
+		amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
 	}
 
 	if (adev->cg_flags & (AMD_CG_SUPPORT_SDMA_LS | AMD_CG_SUPPORT_SDMA_MGCG)) {
 		if (adev->cg_flags & AMD_CG_SUPPORT_SDMA_LS) {
-			pp_support_state = AMD_CG_SUPPORT_SDMA_LS;
+			pp_support_state = PP_STATE_SUPPORT_LS;
 			pp_state = PP_STATE_LS;
 		}
 		if (adev->cg_flags & AMD_CG_SUPPORT_SDMA_MGCG) {
-			pp_support_state |= AMD_CG_SUPPORT_SDMA_MGCG;
+			pp_support_state |= PP_STATE_SUPPORT_CG;
 			pp_state |= PP_STATE_CG;
 		}
 		if (state == AMD_CG_STATE_UNGATE)
@@ -1289,16 +1867,16 @@ static int vi_common_set_clockgating_state_by_smu(void *handle,
 			       PP_BLOCK_SYS_SDMA,
 			       pp_support_state,
 			       pp_state);
-		amd_set_clockgating_by_smu(pp_handle, msg_id);
+		amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
 	}
 
 	if (adev->cg_flags & (AMD_CG_SUPPORT_HDP_LS | AMD_CG_SUPPORT_HDP_MGCG)) {
 		if (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS) {
-			pp_support_state = AMD_CG_SUPPORT_HDP_LS;
+			pp_support_state = PP_STATE_SUPPORT_LS;
 			pp_state = PP_STATE_LS;
 		}
 		if (adev->cg_flags & AMD_CG_SUPPORT_HDP_MGCG) {
-			pp_support_state |= AMD_CG_SUPPORT_HDP_MGCG;
+			pp_support_state |= PP_STATE_SUPPORT_CG;
 			pp_state |= PP_STATE_CG;
 		}
 		if (state == AMD_CG_STATE_UNGATE)
@@ -1307,7 +1885,7 @@ static int vi_common_set_clockgating_state_by_smu(void *handle,
 			       PP_BLOCK_SYS_HDP,
 			       pp_support_state,
 			       pp_state);
-		amd_set_clockgating_by_smu(pp_handle, msg_id);
+		amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
 	}
 
 
@@ -1321,7 +1899,7 @@ static int vi_common_set_clockgating_state_by_smu(void *handle,
 			       PP_BLOCK_SYS_BIF,
 			       PP_STATE_SUPPORT_LS,
 			        pp_state);
-		amd_set_clockgating_by_smu(pp_handle, msg_id);
+		amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
 	}
 	if (adev->cg_flags & AMD_CG_SUPPORT_BIF_MGCG) {
 		if (state == AMD_CG_STATE_UNGATE)
@@ -1333,7 +1911,7 @@ static int vi_common_set_clockgating_state_by_smu(void *handle,
 			       PP_BLOCK_SYS_BIF,
 			       PP_STATE_SUPPORT_CG,
 			       pp_state);
-		amd_set_clockgating_by_smu(pp_handle, msg_id);
+		amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
 	}
 
 	if (adev->cg_flags & AMD_CG_SUPPORT_DRM_LS) {
@@ -1347,7 +1925,7 @@ static int vi_common_set_clockgating_state_by_smu(void *handle,
 			       PP_BLOCK_SYS_DRM,
 			       PP_STATE_SUPPORT_LS,
 			       pp_state);
-		amd_set_clockgating_by_smu(pp_handle, msg_id);
+		amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
 	}
 
 	if (adev->cg_flags & AMD_CG_SUPPORT_ROM_MGCG) {
@@ -1361,15 +1939,15 @@ static int vi_common_set_clockgating_state_by_smu(void *handle,
 			       PP_BLOCK_SYS_ROM,
 			       PP_STATE_SUPPORT_CG,
 			       pp_state);
-		amd_set_clockgating_by_smu(pp_handle, msg_id);
+		amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
 	}
 	return 0;
 }
 
-static int vi_common_set_clockgating_state(void *handle,
+static int vi_common_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					   enum amd_clockgating_state state)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (amdgpu_sriov_vf(adev))
 		return 0;
@@ -1400,22 +1978,24 @@ static int vi_common_set_clockgating_state(void *handle,
 	case CHIP_POLARIS10:
 	case CHIP_POLARIS11:
 	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
 		vi_common_set_clockgating_state_by_smu(adev, state);
+		break;
 	default:
 		break;
 	}
 	return 0;
 }
 
-static int vi_common_set_powergating_state(void *handle,
+static int vi_common_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					    enum amd_powergating_state state)
 {
 	return 0;
 }
 
-static void vi_common_get_clockgating_state(void *handle, u32 *flags)
+static void vi_common_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int data;
 
 	if (amdgpu_sriov_vf(adev))
@@ -1447,14 +2027,11 @@ static const struct amd_ip_funcs vi_common_ip_funcs = {
 	.early_init = vi_common_early_init,
 	.late_init = vi_common_late_init,
 	.sw_init = vi_common_sw_init,
-	.sw_fini = vi_common_sw_fini,
 	.hw_init = vi_common_hw_init,
 	.hw_fini = vi_common_hw_fini,
 	.suspend = vi_common_suspend,
 	.resume = vi_common_resume,
 	.is_idle = vi_common_is_idle,
-	.wait_for_idle = vi_common_wait_for_idle,
-	.soft_reset = vi_common_soft_reset,
 	.set_clockgating_state = vi_common_set_clockgating_state,
 	.set_powergating_state = vi_common_set_powergating_state,
 	.get_clockgating_state = vi_common_get_clockgating_state,
@@ -1469,106 +2046,122 @@ static const struct amdgpu_ip_block_version vi_common_ip_block =
 	.funcs = &vi_common_ip_funcs,
 };
 
-int vi_set_ip_blocks(struct amdgpu_device *adev)
+void vi_set_virt_ops(struct amdgpu_device *adev)
 {
-	/* in early init stage, vbios code won't work */
-	vi_detect_hw_virtualization(adev);
+	adev->virt.ops = &xgpu_vi_virt_ops;
+}
 
-	if (amdgpu_sriov_vf(adev))
-		adev->virt.ops = &xgpu_vi_virt_ops;
+int vi_set_ip_blocks(struct amdgpu_device *adev)
+{
+	amdgpu_device_set_sriov_virtual_display(adev);
 
 	switch (adev->asic_type) {
 	case CHIP_TOPAZ:
 		/* topaz has no DCE, UVD, VCE */
-		amdgpu_ip_block_add(adev, &vi_common_ip_block);
-		amdgpu_ip_block_add(adev, &gmc_v7_4_ip_block);
-		amdgpu_ip_block_add(adev, &iceland_ih_ip_block);
-		amdgpu_ip_block_add(adev, &amdgpu_pp_ip_block);
+		amdgpu_device_ip_block_add(adev, &vi_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v7_4_ip_block);
+		amdgpu_device_ip_block_add(adev, &iceland_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v8_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &sdma_v2_4_ip_block);
+		amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
 		if (adev->enable_virtual_display)
-			amdgpu_ip_block_add(adev, &dce_virtual_ip_block);
-		amdgpu_ip_block_add(adev, &gfx_v8_0_ip_block);
-		amdgpu_ip_block_add(adev, &sdma_v2_4_ip_block);
+			amdgpu_device_ip_block_add(adev, &amdgpu_vkms_ip_block);
 		break;
 	case CHIP_FIJI:
-		amdgpu_ip_block_add(adev, &vi_common_ip_block);
-		amdgpu_ip_block_add(adev, &gmc_v8_5_ip_block);
-		amdgpu_ip_block_add(adev, &tonga_ih_ip_block);
-		amdgpu_ip_block_add(adev, &amdgpu_pp_ip_block);
-		if (adev->enable_virtual_display || amdgpu_sriov_vf(adev))
-			amdgpu_ip_block_add(adev, &dce_virtual_ip_block);
+		amdgpu_device_ip_block_add(adev, &vi_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v8_5_ip_block);
+		amdgpu_device_ip_block_add(adev, &tonga_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v8_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &sdma_v3_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
+		if (adev->enable_virtual_display)
+			amdgpu_device_ip_block_add(adev, &amdgpu_vkms_ip_block);
+#if defined(CONFIG_DRM_AMD_DC)
+		else if (amdgpu_device_has_dc_support(adev))
+			amdgpu_device_ip_block_add(adev, &dm_ip_block);
+#endif
 		else
-			amdgpu_ip_block_add(adev, &dce_v10_1_ip_block);
-		amdgpu_ip_block_add(adev, &gfx_v8_0_ip_block);
-		amdgpu_ip_block_add(adev, &sdma_v3_0_ip_block);
+			amdgpu_device_ip_block_add(adev, &dce_v10_1_ip_block);
 		if (!amdgpu_sriov_vf(adev)) {
-			amdgpu_ip_block_add(adev, &uvd_v6_0_ip_block);
-			amdgpu_ip_block_add(adev, &vce_v3_0_ip_block);
+			amdgpu_device_ip_block_add(adev, &uvd_v6_0_ip_block);
+			amdgpu_device_ip_block_add(adev, &vce_v3_0_ip_block);
 		}
 		break;
 	case CHIP_TONGA:
-		amdgpu_ip_block_add(adev, &vi_common_ip_block);
-		amdgpu_ip_block_add(adev, &gmc_v8_0_ip_block);
-		amdgpu_ip_block_add(adev, &tonga_ih_ip_block);
-		amdgpu_ip_block_add(adev, &amdgpu_pp_ip_block);
-		if (adev->enable_virtual_display || amdgpu_sriov_vf(adev))
-			amdgpu_ip_block_add(adev, &dce_virtual_ip_block);
+		amdgpu_device_ip_block_add(adev, &vi_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v8_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &tonga_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v8_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &sdma_v3_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
+		if (adev->enable_virtual_display)
+			amdgpu_device_ip_block_add(adev, &amdgpu_vkms_ip_block);
+#if defined(CONFIG_DRM_AMD_DC)
+		else if (amdgpu_device_has_dc_support(adev))
+			amdgpu_device_ip_block_add(adev, &dm_ip_block);
+#endif
 		else
-			amdgpu_ip_block_add(adev, &dce_v10_0_ip_block);
-		amdgpu_ip_block_add(adev, &gfx_v8_0_ip_block);
-		amdgpu_ip_block_add(adev, &sdma_v3_0_ip_block);
+			amdgpu_device_ip_block_add(adev, &dce_v10_0_ip_block);
 		if (!amdgpu_sriov_vf(adev)) {
-			amdgpu_ip_block_add(adev, &uvd_v5_0_ip_block);
-			amdgpu_ip_block_add(adev, &vce_v3_0_ip_block);
+			amdgpu_device_ip_block_add(adev, &uvd_v5_0_ip_block);
+			amdgpu_device_ip_block_add(adev, &vce_v3_0_ip_block);
 		}
 		break;
-	case CHIP_POLARIS11:
 	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
 	case CHIP_POLARIS12:
-		amdgpu_ip_block_add(adev, &vi_common_ip_block);
-		amdgpu_ip_block_add(adev, &gmc_v8_1_ip_block);
-		amdgpu_ip_block_add(adev, &tonga_ih_ip_block);
-		amdgpu_ip_block_add(adev, &amdgpu_pp_ip_block);
+	case CHIP_VEGAM:
+		amdgpu_device_ip_block_add(adev, &vi_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v8_1_ip_block);
+		amdgpu_device_ip_block_add(adev, &tonga_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v8_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &sdma_v3_1_ip_block);
+		amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
 		if (adev->enable_virtual_display)
-			amdgpu_ip_block_add(adev, &dce_virtual_ip_block);
-		else
-			amdgpu_ip_block_add(adev, &dce_v11_2_ip_block);
-		amdgpu_ip_block_add(adev, &gfx_v8_0_ip_block);
-		amdgpu_ip_block_add(adev, &sdma_v3_1_ip_block);
-		amdgpu_ip_block_add(adev, &uvd_v6_3_ip_block);
-		amdgpu_ip_block_add(adev, &vce_v3_4_ip_block);
+			amdgpu_device_ip_block_add(adev, &amdgpu_vkms_ip_block);
+#if defined(CONFIG_DRM_AMD_DC)
+		else if (amdgpu_device_has_dc_support(adev))
+			amdgpu_device_ip_block_add(adev, &dm_ip_block);
+#endif
+		amdgpu_device_ip_block_add(adev, &uvd_v6_3_ip_block);
+		amdgpu_device_ip_block_add(adev, &vce_v3_4_ip_block);
 		break;
 	case CHIP_CARRIZO:
-		amdgpu_ip_block_add(adev, &vi_common_ip_block);
-		amdgpu_ip_block_add(adev, &gmc_v8_0_ip_block);
-		amdgpu_ip_block_add(adev, &cz_ih_ip_block);
-		amdgpu_ip_block_add(adev, &amdgpu_pp_ip_block);
+		amdgpu_device_ip_block_add(adev, &vi_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v8_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &cz_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v8_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &sdma_v3_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
 		if (adev->enable_virtual_display)
-			amdgpu_ip_block_add(adev, &dce_virtual_ip_block);
-		else
-			amdgpu_ip_block_add(adev, &dce_v11_0_ip_block);
-		amdgpu_ip_block_add(adev, &gfx_v8_0_ip_block);
-		amdgpu_ip_block_add(adev, &sdma_v3_0_ip_block);
-		amdgpu_ip_block_add(adev, &uvd_v6_0_ip_block);
-		amdgpu_ip_block_add(adev, &vce_v3_1_ip_block);
+			amdgpu_device_ip_block_add(adev, &amdgpu_vkms_ip_block);
+#if defined(CONFIG_DRM_AMD_DC)
+		else if (amdgpu_device_has_dc_support(adev))
+			amdgpu_device_ip_block_add(adev, &dm_ip_block);
+#endif
+		amdgpu_device_ip_block_add(adev, &uvd_v6_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &vce_v3_1_ip_block);
 #if defined(CONFIG_DRM_AMD_ACP)
-		amdgpu_ip_block_add(adev, &acp_ip_block);
+		amdgpu_device_ip_block_add(adev, &acp_ip_block);
 #endif
 		break;
 	case CHIP_STONEY:
-		amdgpu_ip_block_add(adev, &vi_common_ip_block);
-		amdgpu_ip_block_add(adev, &gmc_v8_0_ip_block);
-		amdgpu_ip_block_add(adev, &cz_ih_ip_block);
-		amdgpu_ip_block_add(adev, &amdgpu_pp_ip_block);
+		amdgpu_device_ip_block_add(adev, &vi_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v8_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &cz_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v8_1_ip_block);
+		amdgpu_device_ip_block_add(adev, &sdma_v3_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
 		if (adev->enable_virtual_display)
-			amdgpu_ip_block_add(adev, &dce_virtual_ip_block);
-		else
-			amdgpu_ip_block_add(adev, &dce_v11_0_ip_block);
-		amdgpu_ip_block_add(adev, &gfx_v8_1_ip_block);
-		amdgpu_ip_block_add(adev, &sdma_v3_0_ip_block);
-		amdgpu_ip_block_add(adev, &uvd_v6_2_ip_block);
-		amdgpu_ip_block_add(adev, &vce_v3_4_ip_block);
+			amdgpu_device_ip_block_add(adev, &amdgpu_vkms_ip_block);
+#if defined(CONFIG_DRM_AMD_DC)
+		else if (amdgpu_device_has_dc_support(adev))
+			amdgpu_device_ip_block_add(adev, &dm_ip_block);
+#endif
+		amdgpu_device_ip_block_add(adev, &uvd_v6_2_ip_block);
+		amdgpu_device_ip_block_add(adev, &vce_v3_4_ip_block);
 #if defined(CONFIG_DRM_AMD_ACP)
-		amdgpu_ip_block_add(adev, &acp_ip_block);
+		amdgpu_device_ip_block_add(adev, &acp_ip_block);
 #endif
 		break;
 	default:
@@ -1578,3 +2171,21 @@ int vi_set_ip_blocks(struct amdgpu_device *adev)
 
 	return 0;
 }
+
+void legacy_doorbell_index_init(struct amdgpu_device *adev)
+{
+	adev->doorbell_index.kiq = AMDGPU_DOORBELL_KIQ;
+	adev->doorbell_index.mec_ring0 = AMDGPU_DOORBELL_MEC_RING0;
+	adev->doorbell_index.mec_ring1 = AMDGPU_DOORBELL_MEC_RING1;
+	adev->doorbell_index.mec_ring2 = AMDGPU_DOORBELL_MEC_RING2;
+	adev->doorbell_index.mec_ring3 = AMDGPU_DOORBELL_MEC_RING3;
+	adev->doorbell_index.mec_ring4 = AMDGPU_DOORBELL_MEC_RING4;
+	adev->doorbell_index.mec_ring5 = AMDGPU_DOORBELL_MEC_RING5;
+	adev->doorbell_index.mec_ring6 = AMDGPU_DOORBELL_MEC_RING6;
+	adev->doorbell_index.mec_ring7 = AMDGPU_DOORBELL_MEC_RING7;
+	adev->doorbell_index.gfx_ring0 = AMDGPU_DOORBELL_GFX_RING0;
+	adev->doorbell_index.sdma_engine[0] = AMDGPU_DOORBELL_sDMA_ENGINE0;
+	adev->doorbell_index.sdma_engine[1] = AMDGPU_DOORBELL_sDMA_ENGINE1;
+	adev->doorbell_index.ih = AMDGPU_DOORBELL_IH;
+	adev->doorbell_index.max_assignment = AMDGPU_DOORBELL_MAX_ASSIGNMENT;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/vi.h b/drivers/gpu/drm/amd/amdgpu/vi.h
index 575d7aed5d32..9718f98f8533 100644
--- a/drivers/gpu/drm/amd/amdgpu/vi.h
+++ b/drivers/gpu/drm/amd/amdgpu/vi.h
@@ -24,8 +24,13 @@
 #ifndef __VI_H__
 #define __VI_H__
 
+#define VI_FLUSH_GPU_TLB_NUM_WREG	3
+
 void vi_srbm_select(struct amdgpu_device *adev,
 		    u32 me, u32 pipe, u32 queue, u32 vmid);
+void vi_set_virt_ops(struct amdgpu_device *adev);
 int vi_set_ip_blocks(struct amdgpu_device *adev);
 
+void legacy_doorbell_index_init(struct amdgpu_device *adev);
+
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/vi_dpm.h b/drivers/gpu/drm/amd/amdgpu/vi_dpm.h
deleted file mode 100644
index c43e03fddfba..000000000000
--- a/drivers/gpu/drm/amd/amdgpu/vi_dpm.h
+++ /dev/null
@@ -1,32 +0,0 @@
-/*
- * Copyright 2014 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-
-#ifndef __VI_DPM_H__
-#define __VI_DPM_H__
-
-extern const struct amd_ip_funcs cz_dpm_ip_funcs;
-int cz_smu_init(struct amdgpu_device *adev);
-int cz_smu_start(struct amdgpu_device *adev);
-int cz_smu_fini(struct amdgpu_device *adev);
-
-#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/vid.h b/drivers/gpu/drm/amd/amdgpu/vid.h
index a6485254a169..b61f6b838ec2 100644
--- a/drivers/gpu/drm/amd/amdgpu/vid.h
+++ b/drivers/gpu/drm/amd/amdgpu/vid.h
@@ -27,6 +27,8 @@
 #define SDMA1_REGISTER_OFFSET                             0x200 /* not a register */
 #define SDMA_MAX_INSTANCE 2
 
+#define KFD_VI_SDMA_QUEUE_OFFSET                      0x80 /* not a register */
+
 /* crtc instance offsets */
 #define CRTC0_REGISTER_OFFSET                 (0x1b9c - 0x1b9c)
 #define CRTC1_REGISTER_OFFSET                 (0x1d9c - 0x1b9c)
@@ -65,8 +67,6 @@
 #define HPD4_REGISTER_OFFSET                 (0x18b8 - 0x1898)
 #define HPD5_REGISTER_OFFSET                 (0x18c0 - 0x1898)
 
-#define AMDGPU_NUM_OF_VMIDS			8
-
 #define		PIPEID(x)					((x) << 0)
 #define		MEID(x)						((x) << 2)
 #define		VMID(x)						((x) << 4)
@@ -246,6 +246,7 @@
 		 * 1 - Stream
 		 * 2 - Bypass
 		 */
+#define		EOP_EXEC				(1 << 28) /* For Trailing Fence */
 #define		DATA_SEL(x)                             ((x) << 29)
 		/* 0 - discard
 		 * 1 - send low 32bit data
@@ -330,7 +331,7 @@
 #              define PACKET3_DMA_DATA_CMD_SAIC    (1 << 28)
 #              define PACKET3_DMA_DATA_CMD_DAIC    (1 << 29)
 #              define PACKET3_DMA_DATA_CMD_RAW_WAIT  (1 << 30)
-#define	PACKET3_AQUIRE_MEM				0x58
+#define	PACKET3_ACQUIRE_MEM				0x58
 #define	PACKET3_REWIND					0x59
 #define	PACKET3_LOAD_UCONFIG_REG			0x5E
 #define	PACKET3_LOAD_SH_REG				0x5F
@@ -465,6 +466,16 @@
 #define VCE_CMD_UPDATE_PTB      0x00000107
 #define VCE_CMD_FLUSH_TLB       0x00000108
 
+/* HEVC ENC */
+#define HEVC_ENC_CMD_NO_OP         0x00000000
+#define HEVC_ENC_CMD_END           0x00000001
+#define HEVC_ENC_CMD_FENCE         0x00000003
+#define HEVC_ENC_CMD_TRAP          0x00000004
+#define HEVC_ENC_CMD_IB_VM         0x00000102
+#define HEVC_ENC_CMD_WAIT_GE       0x00000106
+#define HEVC_ENC_CMD_UPDATE_PTB    0x00000107
+#define HEVC_ENC_CMD_FLUSH_TLB     0x00000108
+
 /* mmPA_SC_RASTER_CONFIG mask */
 #define RB_MAP_PKR0(x)				((x) << 0)
 #define RB_MAP_PKR0_MASK			(0x3 << 0)
diff --git a/drivers/gpu/drm/amd/amdgpu/vpe_6_1_fw_if.h b/drivers/gpu/drm/amd/amdgpu/vpe_6_1_fw_if.h
new file mode 100644
index 000000000000..47534dbbd137
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vpe_6_1_fw_if.h
@@ -0,0 +1,218 @@
+/* Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __VPE_6_1_FW_IF_H_
+#define __VPE_6_1_FW_IF_H_
+
+/****************
+ * VPE OP Codes
+ ****************/
+enum VPE_CMD_OPCODE {
+    VPE_CMD_OPCODE_NOP          = 0x0,
+    VPE_CMD_OPCODE_VPE_DESC     = 0x1,
+    VPE_CMD_OPCODE_PLANE_CFG    = 0x2,
+    VPE_CMD_OPCODE_VPEP_CFG     = 0x3,
+    VPE_CMD_OPCODE_INDIRECT     = 0x4,
+    VPE_CMD_OPCODE_FENCE        = 0x5,
+    VPE_CMD_OPCODE_TRAP         = 0x6,
+    VPE_CMD_OPCODE_REG_WRITE    = 0x7,
+    VPE_CMD_OPCODE_POLL_REGMEM  = 0x8,
+    VPE_CMD_OPCODE_COND_EXE     = 0x9,
+    VPE_CMD_OPCODE_ATOMIC       = 0xA,
+    VPE_CMD_OPCODE_PRED_EXE     = 0xB,
+    VPE_CMD_OPCODE_COLLAB_SYNC  = 0xC,
+    VPE_CMD_OPCODE_TIMESTAMP    = 0xD
+};
+
+/** Generic Command Header
+ * Generic Commands include:
+ *  Noop, Fence, Trap,
+ *  RegisterWrite, PollRegisterWriteMemory,
+ *  SetLocalTimestamp, GetLocalTimestamp
+ *  GetGlobalGPUTimestamp */
+#define VPE_HEADER_SUB_OPCODE__SHIFT    8
+#define VPE_HEADER_SUB_OPCODE_MASK      0x0000FF00
+#define VPE_HEADER_OPCODE__SHIFT        0
+#define VPE_HEADER_OPCODE_MASK          0x000000FF
+
+#define VPE_CMD_HEADER(op, subop) \
+    (((subop << VPE_HEADER_SUB_OPCODE__SHIFT) & VPE_HEADER_SUB_OPCODE_MASK) | \
+     ((op << VPE_HEADER_OPCODE__SHIFT) & VPE_HEADER_OPCODE_MASK))
+
+
+ /***************************
+  * VPE NOP
+  ***************************/
+#define VPE_CMD_NOP_HEADER_COUNT__SHIFT    16
+#define VPE_CMD_NOP_HEADER_COUNT_MASK      0x00003FFF
+
+#define VPE_CMD_NOP_HEADER_COUNT(count) \
+     (((count) & VPE_CMD_NOP_HEADER_COUNT_MASK) << VPE_CMD_NOP_HEADER_COUNT__SHIFT)
+
+ /***************************
+  * VPE Descriptor
+  ***************************/
+#define VPE_DESC_CD__SHIFT          16
+#define VPE_DESC_CD_MASK            0x000F0000
+
+#define VPE_DESC_CMD_HEADER(cd) \
+    (VPE_CMD_HEADER(VPE_CMD_OPCODE_VPE_DESC, 0) | \
+     (((cd) << VPE_DESC_CD__SHIFT) & VPE_DESC_CD_MASK))
+
+ /***************************
+  * VPE Plane Config
+  ***************************/
+enum VPE_PLANE_CFG_SUBOP {
+    VPE_PLANE_CFG_SUBOP_1_TO_1 = 0x0,
+    VPE_PLANE_CFG_SUBOP_2_TO_1 = 0x1,
+    VPE_PLANE_CFG_SUBOP_2_TO_2 = 0x2
+};
+
+#define VPE_PLANE_CFG_ONE_PLANE     0
+#define VPE_PLANE_CFG_TWO_PLANES    1
+
+#define VPE_PLANE_CFG_NPS0__SHIFT   16
+#define VPE_PLANE_CFG_NPS0_MASK     0x00030000
+
+#define VPE_PLANE_CFG_NPD0__SHIFT   18
+#define VPE_PLANE_CFG_NPD0_MASK     0x000C0000
+
+#define VPE_PLANE_CFG_NPS1__SHIFT   20
+#define VPE_PLANE_CFG_NPS1_MASK     0x00300000
+
+#define VPE_PLANE_CFG_NPD1__SHIFT   22
+#define VPE_PLANE_CFG_NPD1_MASK     0x00C00000
+
+#define VPE_PLANE_CFG_TMZ__SHIFT    16
+#define VPE_PLANE_CFG_TMZ_MASK      0x00010000
+
+#define VPE_PLANE_CFG_SWIZZLE_MODE__SHIFT   3
+#define VPE_PLANE_CFG_SWIZZLE_MODE_MASK     0x000000F8
+
+#define VPE_PLANE_CFG_ROTATION__SHIFT       0
+#define VPE_PLANE_CFG_ROTATION_MASK         0x00000003
+
+#define VPE_PLANE_ADDR_LO__SHIFT        0
+#define VPE_PLANE_ADDR_LO_MASK          0xFFFFFF00
+
+#define VPE_PLANE_CFG_PITCH__SHIFT      0
+#define VPE_PLANE_CFG_PITCH_MASK        0x00003FFF
+
+#define VPE_PLANE_CFG_VIEWPORT_Y__SHIFT 16
+#define VPE_PLANE_CFG_VIEWPORT_Y_MASK   0x3FFF0000
+#define VPE_PLANE_CFG_VIEWPORT_X__SHIFT 0
+#define VPE_PLANE_CFG_VIEWPORT_X_MASK   0x00003FFF
+
+
+#define VPE_PLANE_CFG_VIEWPORT_HEIGHT__SHIFT 16
+#define VPE_PLANE_CFG_VIEWPORT_HEIGHT_MASK   0x1FFF0000
+#define VPE_PLANE_CFG_VIEWPORT_ELEMENT_SIZE__SHIFT  13
+#define VPE_PLANE_CFG_VIEWPORT_ELEMENT_SIZE_MASK    0x0000E000
+#define VPE_PLANE_CFG_VIEWPORT_WIDTH__SHIFT 0
+#define VPE_PLANE_CFG_VIEWPORT_WIDTH_MASK   0x00001FFF
+
+enum VPE_PLANE_CFG_ELEMENT_SIZE {
+    VPE_PLANE_CFG_ELEMENT_SIZE_8BPE     = 0,
+    VPE_PLANE_CFG_ELEMENT_SIZE_16BPE    = 1,
+    VPE_PLANE_CFG_ELEMENT_SIZE_32BPE    = 2,
+    VPE_PLANE_CFG_ELEMENT_SIZE_64BPE    = 3
+};
+
+#define VPE_PLANE_CFG_CMD_HEADER(subop, nps0, npd0, nps1, npd1) \
+    (VPE_CMD_HEADER(VPE_CMD_OPCODE_PLANE_CFG, subop) | \
+     (((nps0) << VPE_PLANE_CFG_NPS0__SHIFT) & VPE_PLANE_CFG_NPS0_MASK) | \
+     (((npd0) << VPE_PLANE_CFG_NPD0__SHIFT) & VPE_PLANE_CFG_NPD0_MASK) | \
+     (((nps1) << VPE_PLANE_CFG_NPS1__SHIFT) & VPE_PLANE_CFG_NPS1_MASK) | \
+     (((npd0) << VPE_PLANE_CFG_NPD1__SHIFT) & VPE_PLANE_CFG_NPD1_MASK))
+
+
+/************************
+ * VPEP Config
+ ************************/
+enum VPE_VPEP_CFG_SUBOP {
+    VPE_VPEP_CFG_SUBOP_DIR_CFG = 0x0,
+    VPE_VPEP_CFG_SUBOP_IND_CFG = 0x1
+};
+
+
+// Direct Config Command Header
+#define VPE_DIR_CFG_HEADER_ARRAY_SIZE__SHIFT    16
+#define VPE_DIR_CFG_HEADER_ARRAY_SIZE_MASK      0xFFFF0000
+
+#define VPE_DIR_CFG_CMD_HEADER(subop, arr_sz) \
+    (VPE_CMD_HEADER(VPE_CMD_OPCODE_VPEP_CFG, subop) | \
+     (((arr_sz) << VPE_DIR_CFG_HEADER_ARRAY_SIZE__SHIFT) & VPE_DIR_CFG_HEADER_ARRAY_SIZE_MASK))
+
+
+#define VPE_DIR_CFG_PKT_REGISTER_OFFSET__SHIFT  2
+#define VPE_DIR_CFG_PKT_REGISTER_OFFSET_MASK    0x000FFFFC
+
+#define VPE_DIR_CFG_PKT_DATA_SIZE__SHIFT        20
+#define VPE_DIR_CFG_PKT_DATA_SIZE_MASK          0xFFF00000
+
+
+// InDirect Config Command Header
+#define VPE_IND_CFG_HEADER_NUM_DST__SHIFT   28
+#define VPE_IND_CFG_HEADER_NUM_DST_MASK     0xF0000000
+
+#define VPE_IND_CFG_CMD_HEADER(subop, num_dst) \
+    (VPE_CMD_HEADER(VPE_CMD_OPCODE_VPEP_CFG, subop) | \
+     (((num_dst) << VPE_IND_CFG_HEADER_NUM_DST__SHIFT) & VPE_IND_CFG_HEADER_NUM_DST_MASK))
+
+// Indirect Buffer Command Header
+#define VPE_CMD_INDIRECT_HEADER_VMID__SHIFT   16
+#define VPE_CMD_INDIRECT_HEADER_VMID_MASK     0x0000000F
+#define VPE_CMD_INDIRECT_HEADER_VMID(vmid) \
+     (((vmid) & VPE_CMD_INDIRECT_HEADER_VMID_MASK) << VPE_CMD_INDIRECT_HEADER_VMID__SHIFT)
+
+
+/**************************
+ * Poll Reg/Mem Sub-OpCode
+ **************************/
+enum VPE_POLL_REGMEM_SUBOP {
+    VPE_POLL_REGMEM_SUBOP_REGMEM = 0x0,
+    VPE_POLL_REGMEM_SUBOP_REGMEM_WRITE = 0x1
+};
+
+#define VPE_CMD_POLL_REGMEM_HEADER_FUNC__SHIFT   28
+#define VPE_CMD_POLL_REGMEM_HEADER_FUNC_MASK     0x00000007
+#define VPE_CMD_POLL_REGMEM_HEADER_FUNC(func) \
+     (((func) & VPE_CMD_POLL_REGMEM_HEADER_FUNC_MASK) << VPE_CMD_POLL_REGMEM_HEADER_FUNC__SHIFT)
+
+#define VPE_CMD_POLL_REGMEM_HEADER_MEM__SHIFT   31
+#define VPE_CMD_POLL_REGMEM_HEADER_MEM_MASK     0x00000001
+#define VPE_CMD_POLL_REGMEM_HEADER_MEM(mem) \
+     (((mem) & VPE_CMD_POLL_REGMEM_HEADER_MEM_MASK) << VPE_CMD_POLL_REGMEM_HEADER_MEM__SHIFT)
+
+#define VPE_CMD_POLL_REGMEM_DW5_INTERVAL__SHIFT   0
+#define VPE_CMD_POLL_REGMEM_DW5_INTERVAL_MASK     0x0000FFFF
+#define VPE_CMD_POLL_REGMEM_DW5_INTERVAL(interval) \
+     (((interval) & VPE_CMD_POLL_REGMEM_DW5_INTERVAL_MASK) << VPE_CMD_POLL_REGMEM_DW5_INTERVAL__SHIFT)
+
+#define VPE_CMD_POLL_REGMEM_DW5_RETRY_COUNT__SHIFT   16
+#define VPE_CMD_POLL_REGMEM_DW5_RETRY_COUNT_MASK     0x00000FFF
+#define VPE_CMD_POLL_REGMEM_DW5_RETRY_COUNT(count) \
+     (((count) & VPE_CMD_POLL_REGMEM_DW5_RETRY_COUNT_MASK) << VPE_CMD_POLL_REGMEM_DW5_RETRY_COUNT__SHIFT)
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/vpe_v6_1.c b/drivers/gpu/drm/amd/amdgpu/vpe_v6_1.c
new file mode 100644
index 000000000000..45876883bbf3
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vpe_v6_1.c
@@ -0,0 +1,398 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include <linux/firmware.h>
+#include <drm/drm_drv.h>
+
+#include "amdgpu.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_vpe.h"
+#include "vpe_v6_1.h"
+#include "soc15_common.h"
+#include "ivsrcid/vpe/irqsrcs_vpe_6_1.h"
+#include "vpe/vpe_6_1_0_offset.h"
+#include "vpe/vpe_6_1_0_sh_mask.h"
+
+MODULE_FIRMWARE("amdgpu/vpe_6_1_0.bin");
+MODULE_FIRMWARE("amdgpu/vpe_6_1_1.bin");
+MODULE_FIRMWARE("amdgpu/vpe_6_1_3.bin");
+
+#define VPE_THREAD1_UCODE_OFFSET	0x8000
+
+#define regVPEC_COLLABORATE_CNTL                                                0x0013
+#define regVPEC_COLLABORATE_CNTL_BASE_IDX                                       0
+#define VPEC_COLLABORATE_CNTL__COLLABORATE_MODE_EN__SHIFT                       0x0
+#define VPEC_COLLABORATE_CNTL__COLLABORATE_MODE_EN_MASK                         0x00000001L
+
+#define regVPEC_COLLABORATE_CFG                                                 0x0014
+#define regVPEC_COLLABORATE_CFG_BASE_IDX                                        0
+#define VPEC_COLLABORATE_CFG__MASTER_ID__SHIFT                                  0x0
+#define VPEC_COLLABORATE_CFG__MASTER_EN__SHIFT                                  0x3
+#define VPEC_COLLABORATE_CFG__SLAVE0_ID__SHIFT                                  0x4
+#define VPEC_COLLABORATE_CFG__SLAVE0_EN__SHIFT                                  0x7
+#define VPEC_COLLABORATE_CFG__MASTER_ID_MASK                                    0x00000007L
+#define VPEC_COLLABORATE_CFG__MASTER_EN_MASK                                    0x00000008L
+#define VPEC_COLLABORATE_CFG__SLAVE0_ID_MASK                                    0x00000070L
+#define VPEC_COLLABORATE_CFG__SLAVE0_EN_MASK                                    0x00000080L
+
+#define regVPEC_CNTL_6_1_1                                                      0x0016
+#define regVPEC_CNTL_6_1_1_BASE_IDX                                             0
+#define regVPEC_QUEUE_RESET_REQ_6_1_1                                           0x002c
+#define regVPEC_QUEUE_RESET_REQ_6_1_1_BASE_IDX                                  0
+#define regVPEC_PUB_DUMMY2_6_1_1                                                0x004c
+#define regVPEC_PUB_DUMMY2_6_1_1_BASE_IDX                                       0
+
+static uint32_t vpe_v6_1_get_reg_offset(struct amdgpu_vpe *vpe, uint32_t inst, uint32_t offset)
+{
+	uint32_t base;
+
+	base = vpe->ring.adev->reg_offset[VPE_HWIP][inst][0];
+
+	return base + offset;
+}
+
+static void vpe_v6_1_halt(struct amdgpu_vpe *vpe, bool halt)
+{
+	struct amdgpu_device *adev = vpe->ring.adev;
+	uint32_t i, f32_cntl;
+
+	for (i = 0; i < vpe->num_instances; i++) {
+		f32_cntl = RREG32(vpe_get_reg_offset(vpe, i, regVPEC_F32_CNTL));
+		f32_cntl = REG_SET_FIELD(f32_cntl, VPEC_F32_CNTL, HALT, halt ? 1 : 0);
+		f32_cntl = REG_SET_FIELD(f32_cntl, VPEC_F32_CNTL, TH1_RESET, halt ? 1 : 0);
+		WREG32(vpe_get_reg_offset(vpe, i, regVPEC_F32_CNTL), f32_cntl);
+	}
+}
+
+static int vpe_v6_1_irq_init(struct amdgpu_vpe *vpe)
+{
+	struct amdgpu_device *adev = container_of(vpe, struct amdgpu_device, vpe);
+	int ret;
+
+	ret = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_VPE,
+				VPE_6_1_SRCID__VPE_TRAP,
+				&adev->vpe.trap_irq);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static void vpe_v6_1_set_collaborate_mode(struct amdgpu_vpe *vpe, bool enable)
+{
+	struct amdgpu_device *adev = vpe->ring.adev;
+	uint32_t vpe_colla_cntl, vpe_colla_cfg, i;
+
+	if (!vpe->collaborate_mode)
+		return;
+
+	for (i = 0; i < vpe->num_instances; i++) {
+		vpe_colla_cntl = RREG32(vpe_get_reg_offset(vpe, i, regVPEC_COLLABORATE_CNTL));
+		vpe_colla_cntl = REG_SET_FIELD(vpe_colla_cntl, VPEC_COLLABORATE_CNTL,
+					       COLLABORATE_MODE_EN, enable ? 1 : 0);
+		WREG32(vpe_get_reg_offset(vpe, i, regVPEC_COLLABORATE_CNTL), vpe_colla_cntl);
+
+		vpe_colla_cfg = RREG32(vpe_get_reg_offset(vpe, i, regVPEC_COLLABORATE_CFG));
+		vpe_colla_cfg = REG_SET_FIELD(vpe_colla_cfg, VPEC_COLLABORATE_CFG, MASTER_ID, 0);
+		vpe_colla_cfg = REG_SET_FIELD(vpe_colla_cfg, VPEC_COLLABORATE_CFG, MASTER_EN, enable ? 1 : 0);
+		vpe_colla_cfg = REG_SET_FIELD(vpe_colla_cfg, VPEC_COLLABORATE_CFG, SLAVE0_ID, 1);
+		vpe_colla_cfg = REG_SET_FIELD(vpe_colla_cfg, VPEC_COLLABORATE_CFG, SLAVE0_EN, enable ? 1 : 0);
+		WREG32(vpe_get_reg_offset(vpe, i, regVPEC_COLLABORATE_CFG), vpe_colla_cfg);
+	}
+}
+
+static int vpe_v6_1_load_microcode(struct amdgpu_vpe *vpe)
+{
+	struct amdgpu_device *adev = vpe->ring.adev;
+	const struct vpe_firmware_header_v1_0 *vpe_hdr;
+	const __le32 *data;
+	uint32_t ucode_offset[2], ucode_size[2];
+	uint32_t i, j, size_dw;
+	uint32_t ret;
+
+	/* disable UMSCH_INT_ENABLE */
+	for (j = 0; j < vpe->num_instances; j++) {
+
+		if (amdgpu_ip_version(adev, VPE_HWIP, 0) == IP_VERSION(6, 1, 1))
+			ret = RREG32(vpe_get_reg_offset(vpe, j, regVPEC_CNTL_6_1_1));
+		else
+			ret = RREG32(vpe_get_reg_offset(vpe, j, regVPEC_CNTL));
+
+		ret = REG_SET_FIELD(ret, VPEC_CNTL, UMSCH_INT_ENABLE, 0);
+
+		if (amdgpu_ip_version(adev, VPE_HWIP, 0) == IP_VERSION(6, 1, 1))
+			WREG32(vpe_get_reg_offset(vpe, j, regVPEC_CNTL_6_1_1), ret);
+		else
+			WREG32(vpe_get_reg_offset(vpe, j, regVPEC_CNTL), ret);
+	}
+
+	/* setup collaborate mode */
+	vpe_v6_1_set_collaborate_mode(vpe, true);
+	/* setup DPM */
+	if (amdgpu_vpe_configure_dpm(vpe))
+		dev_warn(adev->dev, "VPE failed to enable DPM\n");
+
+	/*
+	 * For VPE 6.1.1, still only need to add master's offset, and psp will apply it to slave as well.
+	 * Here use instance 0 as master.
+	 */
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		uint32_t f32_offset, f32_cntl;
+
+		f32_offset = vpe_get_reg_offset(vpe, 0, regVPEC_F32_CNTL);
+		f32_cntl = RREG32(f32_offset);
+		f32_cntl = REG_SET_FIELD(f32_cntl, VPEC_F32_CNTL, HALT, 0);
+		f32_cntl = REG_SET_FIELD(f32_cntl, VPEC_F32_CNTL, TH1_RESET, 0);
+
+		adev->vpe.cmdbuf_cpu_addr[0] = f32_offset;
+		adev->vpe.cmdbuf_cpu_addr[1] = f32_cntl;
+
+		return amdgpu_vpe_psp_update_sram(adev);
+	}
+
+	vpe_hdr = (const struct vpe_firmware_header_v1_0 *)adev->vpe.fw->data;
+
+	/* Thread 0(command thread) ucode offset/size */
+	ucode_offset[0] = le32_to_cpu(vpe_hdr->header.ucode_array_offset_bytes);
+	ucode_size[0] = le32_to_cpu(vpe_hdr->ctx_ucode_size_bytes);
+	/* Thread 1(control thread) ucode offset/size */
+	ucode_offset[1] = le32_to_cpu(vpe_hdr->ctl_ucode_offset);
+	ucode_size[1] = le32_to_cpu(vpe_hdr->ctl_ucode_size_bytes);
+
+	vpe_v6_1_halt(vpe, true);
+
+	for (j = 0; j < vpe->num_instances; j++) {
+		for (i = 0; i < 2; i++) {
+			if (i > 0)
+				WREG32(vpe_get_reg_offset(vpe, j, regVPEC_UCODE_ADDR), VPE_THREAD1_UCODE_OFFSET);
+			else
+				WREG32(vpe_get_reg_offset(vpe, j, regVPEC_UCODE_ADDR), 0);
+
+			data = (const __le32 *)(adev->vpe.fw->data + ucode_offset[i]);
+			size_dw = ucode_size[i] / sizeof(__le32);
+
+			while (size_dw--) {
+				if (amdgpu_emu_mode && size_dw % 500 == 0)
+					msleep(1);
+				WREG32(vpe_get_reg_offset(vpe, j, regVPEC_UCODE_DATA), le32_to_cpup(data++));
+			}
+		}
+	}
+
+	vpe_v6_1_halt(vpe, false);
+
+	return 0;
+}
+
+static int vpe_v6_1_ring_start(struct amdgpu_vpe *vpe)
+{
+	struct amdgpu_ring *ring = &vpe->ring;
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t doorbell, doorbell_offset;
+	uint32_t rb_bufsz, rb_cntl;
+	uint32_t ib_cntl, i;
+	int ret;
+
+	for (i = 0; i < vpe->num_instances; i++) {
+		/* Set ring buffer size in dwords */
+		rb_bufsz = order_base_2(ring->ring_size / 4);
+		rb_cntl = RREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE0_RB_CNTL));
+		rb_cntl = REG_SET_FIELD(rb_cntl, VPEC_QUEUE0_RB_CNTL, RB_SIZE, rb_bufsz);
+		rb_cntl = REG_SET_FIELD(rb_cntl, VPEC_QUEUE0_RB_CNTL, RB_PRIV, 1);
+		rb_cntl = REG_SET_FIELD(rb_cntl, VPEC_QUEUE0_RB_CNTL, RB_VMID, 0);
+		WREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE0_RB_CNTL), rb_cntl);
+
+		/* Initialize the ring buffer's read and write pointers */
+		WREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE0_RB_RPTR), 0);
+		WREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE0_RB_RPTR_HI), 0);
+		WREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE0_RB_WPTR), 0);
+		WREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE0_RB_WPTR_HI), 0);
+
+		/* set the wb address whether it's enabled or not */
+		WREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE0_RB_RPTR_ADDR_LO),
+			lower_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFC);
+		WREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE0_RB_RPTR_ADDR_HI),
+			upper_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFF);
+
+		rb_cntl = REG_SET_FIELD(rb_cntl, VPEC_QUEUE0_RB_CNTL, RPTR_WRITEBACK_ENABLE, 1);
+
+		WREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE0_RB_BASE), ring->gpu_addr >> 8);
+		WREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE0_RB_BASE_HI), ring->gpu_addr >> 40);
+
+		ring->wptr = 0;
+
+		/* before programing wptr to a less value, need set minor_ptr_update first */
+		WREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE0_MINOR_PTR_UPDATE), 1);
+		WREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE0_RB_WPTR), lower_32_bits(ring->wptr) << 2);
+		WREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE0_RB_WPTR_HI), upper_32_bits(ring->wptr) << 2);
+		/* set minor_ptr_update to 0 after wptr programed */
+		WREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE0_MINOR_PTR_UPDATE), 0);
+
+		doorbell_offset = RREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE0_DOORBELL_OFFSET));
+		doorbell_offset = REG_SET_FIELD(doorbell_offset, VPEC_QUEUE0_DOORBELL_OFFSET, OFFSET, ring->doorbell_index + i*4);
+		WREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE0_DOORBELL_OFFSET), doorbell_offset);
+
+		doorbell = RREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE0_DOORBELL));
+		doorbell = REG_SET_FIELD(doorbell, VPEC_QUEUE0_DOORBELL, ENABLE, ring->use_doorbell ? 1 : 0);
+		WREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE0_DOORBELL), doorbell);
+
+		adev->nbio.funcs->vpe_doorbell_range(adev, i, ring->use_doorbell, ring->doorbell_index + i*4, 4);
+
+		rb_cntl = REG_SET_FIELD(rb_cntl, VPEC_QUEUE0_RB_CNTL, RPTR_WRITEBACK_ENABLE, 1);
+		rb_cntl = REG_SET_FIELD(rb_cntl, VPEC_QUEUE0_RB_CNTL, RB_ENABLE, 1);
+		WREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE0_RB_CNTL), rb_cntl);
+
+		ib_cntl = RREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE0_IB_CNTL));
+		ib_cntl = REG_SET_FIELD(ib_cntl, VPEC_QUEUE0_IB_CNTL, IB_ENABLE, 1);
+		WREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE0_IB_CNTL), ib_cntl);
+	}
+
+	ret = amdgpu_ring_test_helper(ring);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static int vpe_v_6_1_ring_stop(struct amdgpu_vpe *vpe)
+{
+	struct amdgpu_device *adev = vpe->ring.adev;
+	uint32_t queue_reset, i;
+	int ret;
+
+	for (i = 0; i < vpe->num_instances; i++) {
+		if (amdgpu_ip_version(adev, VPE_HWIP, 0) == IP_VERSION(6, 1, 1))
+			queue_reset = RREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE_RESET_REQ_6_1_1));
+		else
+			queue_reset = RREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE_RESET_REQ));
+
+		queue_reset = REG_SET_FIELD(queue_reset, VPEC_QUEUE_RESET_REQ, QUEUE0_RESET, 1);
+
+		if (amdgpu_ip_version(adev, VPE_HWIP, 0) == IP_VERSION(6, 1, 1)) {
+			WREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE_RESET_REQ_6_1_1), queue_reset);
+			ret = SOC15_WAIT_ON_RREG(VPE, i, regVPEC_QUEUE_RESET_REQ_6_1_1, 0,
+						 VPEC_QUEUE_RESET_REQ__QUEUE0_RESET_MASK);
+		} else {
+			WREG32(vpe_get_reg_offset(vpe, i, regVPEC_QUEUE_RESET_REQ), queue_reset);
+			ret = SOC15_WAIT_ON_RREG(VPE, i, regVPEC_QUEUE_RESET_REQ, 0,
+						 VPEC_QUEUE_RESET_REQ__QUEUE0_RESET_MASK);
+		}
+
+		if (ret)
+			dev_err(adev->dev, "VPE queue reset failed\n");
+	}
+
+	vpe->ring.sched.ready = false;
+
+	return ret;
+}
+
+static int vpe_v6_1_set_trap_irq_state(struct amdgpu_device *adev,
+				       struct amdgpu_irq_src *source,
+				       unsigned int type,
+				       enum amdgpu_interrupt_state state)
+{
+	struct amdgpu_vpe *vpe = &adev->vpe;
+	uint32_t vpe_cntl;
+
+	if (amdgpu_ip_version(adev, VPE_HWIP, 0) == IP_VERSION(6, 1, 1))
+		vpe_cntl = RREG32(vpe_get_reg_offset(vpe, 0, regVPEC_CNTL_6_1_1));
+	else
+		vpe_cntl = RREG32(vpe_get_reg_offset(vpe, 0, regVPEC_CNTL));
+
+	vpe_cntl = REG_SET_FIELD(vpe_cntl, VPEC_CNTL, TRAP_ENABLE,
+				 state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+
+	if (amdgpu_ip_version(adev, VPE_HWIP, 0) == IP_VERSION(6, 1, 1))
+		WREG32(vpe_get_reg_offset(vpe, 0, regVPEC_CNTL_6_1_1), vpe_cntl);
+	else
+		WREG32(vpe_get_reg_offset(vpe, 0, regVPEC_CNTL), vpe_cntl);
+
+	return 0;
+}
+
+static int vpe_v6_1_process_trap_irq(struct amdgpu_device *adev,
+				     struct amdgpu_irq_src *source,
+				     struct amdgpu_iv_entry *entry)
+{
+
+	dev_dbg(adev->dev, "IH: VPE trap\n");
+
+	switch (entry->client_id) {
+	case SOC21_IH_CLIENTID_VPE:
+		amdgpu_fence_process(&adev->vpe.ring);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int vpe_v6_1_set_regs(struct amdgpu_vpe *vpe)
+{
+	struct amdgpu_device *adev = container_of(vpe, struct amdgpu_device, vpe);
+
+	vpe->regs.queue0_rb_rptr_lo = regVPEC_QUEUE0_RB_RPTR;
+	vpe->regs.queue0_rb_rptr_hi = regVPEC_QUEUE0_RB_RPTR_HI;
+	vpe->regs.queue0_rb_wptr_lo = regVPEC_QUEUE0_RB_WPTR;
+	vpe->regs.queue0_rb_wptr_hi = regVPEC_QUEUE0_RB_WPTR_HI;
+	vpe->regs.queue0_preempt = regVPEC_QUEUE0_PREEMPT;
+
+	if (amdgpu_ip_version(adev, VPE_HWIP, 0) == IP_VERSION(6, 1, 1))
+		vpe->regs.dpm_enable = regVPEC_PUB_DUMMY2_6_1_1;
+	else
+		vpe->regs.dpm_enable = regVPEC_PUB_DUMMY2;
+
+	vpe->regs.dpm_pratio = regVPEC_QUEUE6_DUMMY4;
+	vpe->regs.dpm_request_interval = regVPEC_QUEUE5_DUMMY3;
+	vpe->regs.dpm_decision_threshold = regVPEC_QUEUE5_DUMMY4;
+	vpe->regs.dpm_busy_clamp_threshold = regVPEC_QUEUE7_DUMMY2;
+	vpe->regs.dpm_idle_clamp_threshold = regVPEC_QUEUE7_DUMMY3;
+	vpe->regs.dpm_request_lv = regVPEC_QUEUE7_DUMMY1;
+	vpe->regs.context_indicator = regVPEC_QUEUE6_DUMMY3;
+
+	return 0;
+}
+
+static const struct vpe_funcs vpe_v6_1_funcs = {
+	.get_reg_offset = vpe_v6_1_get_reg_offset,
+	.set_regs = vpe_v6_1_set_regs,
+	.irq_init = vpe_v6_1_irq_init,
+	.init_microcode = amdgpu_vpe_init_microcode,
+	.load_microcode = vpe_v6_1_load_microcode,
+	.ring_init = amdgpu_vpe_ring_init,
+	.ring_start = vpe_v6_1_ring_start,
+	.ring_stop = vpe_v_6_1_ring_stop,
+	.ring_fini = amdgpu_vpe_ring_fini,
+};
+
+static const struct amdgpu_irq_src_funcs vpe_v6_1_trap_irq_funcs = {
+	.set = vpe_v6_1_set_trap_irq_state,
+	.process = vpe_v6_1_process_trap_irq,
+};
+
+void vpe_v6_1_set_funcs(struct amdgpu_vpe *vpe)
+{
+	vpe->funcs = &vpe_v6_1_funcs;
+	vpe->trap_irq.funcs = &vpe_v6_1_trap_irq_funcs;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/vpe_v6_1.h b/drivers/gpu/drm/amd/amdgpu/vpe_v6_1.h
new file mode 100644
index 000000000000..a9bea7905a77
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vpe_v6_1.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+#ifndef __VPE_V6_1_H__
+#define __VPE_V6_1_H__
+
+#include "amdgpu_vpe.h"
+
+void vpe_v6_1_set_funcs(struct amdgpu_vpe *vpe);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdkfd/Kconfig b/drivers/gpu/drm/amd/amdkfd/Kconfig
index e13c67c8d2c0..16e12c9913f9 100644
--- a/drivers/gpu/drm/amd/amdkfd/Kconfig
+++ b/drivers/gpu/drm/amd/amdkfd/Kconfig
@@ -1,9 +1,40 @@
+# SPDX-License-Identifier: MIT
 #
-# Heterogenous system architecture configuration
+# Heterogeneous system architecture configuration
 #
 
 config HSA_AMD
-	tristate "HSA kernel driver for AMD GPU devices"
-	depends on (DRM_RADEON || DRM_AMDGPU) && AMD_IOMMU_V2 && X86_64
+	bool "HSA kernel driver for AMD GPU devices"
+	depends on DRM_AMDGPU && (X86_64 || ARM64 || PPC64 || (RISCV && 64BIT) || (LOONGARCH && 64BIT))
+	select HMM_MIRROR
+	select MMU_NOTIFIER
+	select DRM_AMDGPU_USERPTR
 	help
 	  Enable this if you want to use HSA features on AMD GPU devices.
+
+config HSA_AMD_SVM
+	bool "Enable HMM-based shared virtual memory manager"
+	depends on HSA_AMD && DEVICE_PRIVATE
+	default y
+	select HMM_MIRROR
+	select MMU_NOTIFIER
+	help
+	  Enable this to use unified memory and managed memory in HIP. This
+	  memory manager supports two modes of operation. One based on
+	  preemptions and one based on page faults. To enable page fault
+	  based memory management on most GFXv9 GPUs, set the module
+	  parameter amdgpu.noretry=0.
+
+config HSA_AMD_P2P
+	bool "HSA kernel driver support for peer-to-peer for AMD GPU devices"
+	depends on HSA_AMD && PCI_P2PDMA && DMABUF_MOVE_NOTIFY
+	help
+	  Enable peer-to-peer (P2P) communication between AMD GPUs over
+	  the PCIe bus. This can improve performance of multi-GPU compute
+	  applications and libraries by enabling GPUs to access data directly
+	  in peer GPUs' memory without intermediate copies in system memory.
+
+	  This P2P feature is only enabled on compatible chipsets, and between
+	  GPUs with large memory BARs that expose the entire VRAM in PCIe bus
+	  address space within the physical address limits of the GPUs.
+
diff --git a/drivers/gpu/drm/amd/amdkfd/Makefile b/drivers/gpu/drm/amd/amdkfd/Makefile
index b400d5664252..0ce08113c9f0 100644
--- a/drivers/gpu/drm/amd/amdkfd/Makefile
+++ b/drivers/gpu/drm/amd/amdkfd/Makefile
@@ -1,19 +1,70 @@
 #
+# Copyright 2017 Advanced Micro Devices, Inc.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+#
 # Makefile for Heterogenous System Architecture support for AMD GPU devices
 #
 
-ccflags-y := -Idrivers/gpu/drm/amd/include/  \
-		-Idrivers/gpu/drm/amd/include/asic_reg
+AMDKFD_FILES	:= $(AMDKFD_PATH)/kfd_module.o \
+		$(AMDKFD_PATH)/kfd_device.o \
+		$(AMDKFD_PATH)/kfd_chardev.o \
+		$(AMDKFD_PATH)/kfd_topology.o \
+		$(AMDKFD_PATH)/kfd_doorbell.o \
+		$(AMDKFD_PATH)/kfd_flat_memory.o \
+		$(AMDKFD_PATH)/kfd_process.o \
+		$(AMDKFD_PATH)/kfd_queue.o \
+		$(AMDKFD_PATH)/kfd_mqd_manager.o \
+		$(AMDKFD_PATH)/kfd_mqd_manager_cik.o \
+		$(AMDKFD_PATH)/kfd_mqd_manager_vi.o \
+		$(AMDKFD_PATH)/kfd_mqd_manager_v9.o \
+		$(AMDKFD_PATH)/kfd_mqd_manager_v10.o \
+		$(AMDKFD_PATH)/kfd_mqd_manager_v11.o \
+		$(AMDKFD_PATH)/kfd_mqd_manager_v12.o \
+		$(AMDKFD_PATH)/kfd_kernel_queue.o \
+		$(AMDKFD_PATH)/kfd_packet_manager.o \
+		$(AMDKFD_PATH)/kfd_packet_manager_vi.o \
+		$(AMDKFD_PATH)/kfd_packet_manager_v9.o \
+		$(AMDKFD_PATH)/kfd_process_queue_manager.o \
+		$(AMDKFD_PATH)/kfd_device_queue_manager.o \
+		$(AMDKFD_PATH)/kfd_device_queue_manager_cik.o \
+		$(AMDKFD_PATH)/kfd_device_queue_manager_vi.o \
+		$(AMDKFD_PATH)/kfd_device_queue_manager_v9.o \
+		$(AMDKFD_PATH)/kfd_device_queue_manager_v10.o \
+		$(AMDKFD_PATH)/kfd_device_queue_manager_v11.o \
+		$(AMDKFD_PATH)/kfd_device_queue_manager_v12.o \
+		$(AMDKFD_PATH)/kfd_interrupt.o \
+		$(AMDKFD_PATH)/kfd_events.o \
+		$(AMDKFD_PATH)/cik_event_interrupt.o \
+		$(AMDKFD_PATH)/kfd_int_process_v9.o \
+		$(AMDKFD_PATH)/kfd_int_process_v10.o \
+		$(AMDKFD_PATH)/kfd_int_process_v11.o \
+		$(AMDKFD_PATH)/kfd_smi_events.o \
+		$(AMDKFD_PATH)/kfd_crat.o \
+		$(AMDKFD_PATH)/kfd_debug.o
 
-amdkfd-y	:= kfd_module.o kfd_device.o kfd_chardev.o kfd_topology.o \
-		kfd_pasid.o kfd_doorbell.o kfd_flat_memory.o \
-		kfd_process.o kfd_queue.o kfd_mqd_manager.o \
-		kfd_mqd_manager_cik.o kfd_mqd_manager_vi.o \
-		kfd_kernel_queue.o kfd_kernel_queue_cik.o \
-		kfd_kernel_queue_vi.o kfd_packet_manager.o \
-		kfd_process_queue_manager.o kfd_device_queue_manager.o \
-		kfd_device_queue_manager_cik.o kfd_device_queue_manager_vi.o \
-		kfd_interrupt.o kfd_events.o cik_event_interrupt.o \
-		kfd_dbgdev.o kfd_dbgmgr.o
+ifneq ($(CONFIG_DEBUG_FS),)
+AMDKFD_FILES += $(AMDKFD_PATH)/kfd_debugfs.o
+endif
 
-obj-$(CONFIG_HSA_AMD)	+= amdkfd.o
+ifneq ($(CONFIG_HSA_AMD_SVM),)
+AMDKFD_FILES += $(AMDKFD_PATH)/kfd_svm.o \
+		$(AMDKFD_PATH)/kfd_migrate.o
+endif
diff --git a/drivers/gpu/drm/amd/amdkfd/cik_event_interrupt.c b/drivers/gpu/drm/amd/amdkfd/cik_event_interrupt.c
index 211fc48697fa..73acbe0b7c21 100644
--- a/drivers/gpu/drm/amd/amdkfd/cik_event_interrupt.c
+++ b/drivers/gpu/drm/amd/amdkfd/cik_event_interrupt.c
@@ -23,41 +23,110 @@
 #include "kfd_priv.h"
 #include "kfd_events.h"
 #include "cik_int.h"
+#include "amdgpu_amdkfd.h"
+#include "kfd_smi_events.h"
 
-static bool cik_event_interrupt_isr(struct kfd_dev *dev,
-					const uint32_t *ih_ring_entry)
+static bool cik_event_interrupt_isr(struct kfd_node *dev,
+					const uint32_t *ih_ring_entry,
+					uint32_t *patched_ihre,
+					bool *patched_flag)
 {
-	unsigned int pasid;
 	const struct cik_ih_ring_entry *ihre =
 			(const struct cik_ih_ring_entry *)ih_ring_entry;
+	const struct kfd2kgd_calls *f2g = dev->kfd2kgd;
+	unsigned int vmid;
+	uint16_t pasid;
+	bool ret;
+
+	/* This workaround is due to HW/FW limitation on Hawaii that
+	 * VMID and PASID are not written into ih_ring_entry
+	 */
+	if ((ihre->source_id == CIK_INTSRC_GFX_PAGE_INV_FAULT ||
+		ihre->source_id == CIK_INTSRC_GFX_MEM_PROT_FAULT) &&
+		dev->adev->asic_type == CHIP_HAWAII) {
+		struct cik_ih_ring_entry *tmp_ihre =
+			(struct cik_ih_ring_entry *)patched_ihre;
+
+		*patched_flag = true;
+		*tmp_ihre = *ihre;
+
+		vmid = f2g->read_vmid_from_vmfault_reg(dev->adev);
+		ret = f2g->get_atc_vmid_pasid_mapping_info(dev->adev, vmid, &pasid);
+
+		tmp_ihre->ring_id &= 0x000000ff;
+		tmp_ihre->ring_id |= vmid << 8;
+		tmp_ihre->ring_id |= pasid << 16;
+
+		return ret && (pasid != 0) &&
+			vmid >= dev->vm_info.first_vmid_kfd &&
+			vmid <= dev->vm_info.last_vmid_kfd;
+	}
+
+	/* Only handle interrupts from KFD VMIDs */
+	vmid  = (ihre->ring_id & 0x0000ff00) >> 8;
+	if (vmid < dev->vm_info.first_vmid_kfd ||
+	    vmid > dev->vm_info.last_vmid_kfd)
+		return false;
 
+	/* If there is no valid PASID, it's likely a firmware bug */
 	pasid = (ihre->ring_id & 0xffff0000) >> 16;
+	if (WARN_ONCE(pasid == 0, "FW bug: No PASID in KFD interrupt"))
+		return false;
 
-	/* Do not process in ISR, just request it to be forwarded to WQ. */
-	return (pasid != 0) &&
-		(ihre->source_id == CIK_INTSRC_CP_END_OF_PIPE ||
+	/* Interrupt types we care about: various signals and faults.
+	 * They will be forwarded to a work queue (see below).
+	 */
+	return ihre->source_id == CIK_INTSRC_CP_END_OF_PIPE ||
+		ihre->source_id == CIK_INTSRC_SDMA_TRAP ||
 		ihre->source_id == CIK_INTSRC_SQ_INTERRUPT_MSG ||
-		ihre->source_id == CIK_INTSRC_CP_BAD_OPCODE);
+		ihre->source_id == CIK_INTSRC_CP_BAD_OPCODE ||
+		((ihre->source_id == CIK_INTSRC_GFX_PAGE_INV_FAULT ||
+		ihre->source_id == CIK_INTSRC_GFX_MEM_PROT_FAULT) &&
+		!amdgpu_no_queue_eviction_on_vm_fault);
 }
 
-static void cik_event_interrupt_wq(struct kfd_dev *dev,
+static void cik_event_interrupt_wq(struct kfd_node *dev,
 					const uint32_t *ih_ring_entry)
 {
-	unsigned int pasid;
 	const struct cik_ih_ring_entry *ihre =
 			(const struct cik_ih_ring_entry *)ih_ring_entry;
-
-	pasid = (ihre->ring_id & 0xffff0000) >> 16;
+	uint32_t context_id = ihre->data & 0xfffffff;
+	u32 pasid = (ihre->ring_id & 0xffff0000) >> 16;
 
 	if (pasid == 0)
 		return;
 
 	if (ihre->source_id == CIK_INTSRC_CP_END_OF_PIPE)
-		kfd_signal_event_interrupt(pasid, 0, 0);
+		kfd_signal_event_interrupt(pasid, context_id, 28);
+	else if (ihre->source_id == CIK_INTSRC_SDMA_TRAP)
+		kfd_signal_event_interrupt(pasid, context_id, 28);
 	else if (ihre->source_id == CIK_INTSRC_SQ_INTERRUPT_MSG)
-		kfd_signal_event_interrupt(pasid, ihre->data & 0xFF, 8);
+		kfd_signal_event_interrupt(pasid, context_id & 0xff, 8);
 	else if (ihre->source_id == CIK_INTSRC_CP_BAD_OPCODE)
 		kfd_signal_hw_exception_event(pasid);
+	else if (ihre->source_id == CIK_INTSRC_GFX_PAGE_INV_FAULT ||
+		ihre->source_id == CIK_INTSRC_GFX_MEM_PROT_FAULT) {
+		struct kfd_process_device *pdd = NULL;
+		struct kfd_vm_fault_info info;
+		struct kfd_process *p;
+
+		kfd_smi_event_update_vmfault(dev, pasid);
+		p = kfd_lookup_process_by_pasid(pasid, &pdd);
+		if (!pdd)
+			return;
+
+		kfd_evict_process_device(pdd);
+
+		memset(&info, 0, sizeof(info));
+		amdgpu_amdkfd_gpuvm_get_vm_fault_info(dev->adev, &info);
+		if (!info.page_addr && !info.status) {
+			kfd_unref_process(p);
+			return;
+		}
+
+		kfd_signal_vm_fault_event(pdd, &info, NULL);
+		kfd_unref_process(p);
+	}
 }
 
 const struct kfd_event_interrupt_class event_interrupt_class_cik = {
diff --git a/drivers/gpu/drm/amd/amdkfd/cik_int.h b/drivers/gpu/drm/amd/amdkfd/cik_int.h
index 79a16d24c1b8..76f8677a7926 100644
--- a/drivers/gpu/drm/amd/amdkfd/cik_int.h
+++ b/drivers/gpu/drm/amd/amdkfd/cik_int.h
@@ -20,8 +20,8 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  */
 
-#ifndef HSA_RADEON_CIK_INT_H_INCLUDED
-#define HSA_RADEON_CIK_INT_H_INCLUDED
+#ifndef CIK_INT_H_INCLUDED
+#define CIK_INT_H_INCLUDED
 
 #include <linux/types.h>
 
@@ -32,10 +32,12 @@ struct cik_ih_ring_entry {
 	uint32_t reserved;
 };
 
-#define CIK_INTSRC_DEQUEUE_COMPLETE	0xC6
 #define CIK_INTSRC_CP_END_OF_PIPE	0xB5
 #define CIK_INTSRC_CP_BAD_OPCODE	0xB7
+#define CIK_INTSRC_SDMA_TRAP		0xE0
 #define CIK_INTSRC_SQ_INTERRUPT_MSG	0xEF
+#define CIK_INTSRC_GFX_PAGE_INV_FAULT	0x92
+#define CIK_INTSRC_GFX_MEM_PROT_FAULT	0x93
 
 #endif
 
diff --git a/drivers/gpu/drm/amd/amdkfd/cik_regs.h b/drivers/gpu/drm/amd/amdkfd/cik_regs.h
index 48769d12dd7b..8e2a1663c4db 100644
--- a/drivers/gpu/drm/amd/amdkfd/cik_regs.h
+++ b/drivers/gpu/drm/amd/amdkfd/cik_regs.h
@@ -33,7 +33,8 @@
 #define	APE1_MTYPE(x)					((x) << 7)
 
 /* valid for both DEFAULT_MTYPE and APE1_MTYPE */
-#define	MTYPE_CACHED					0
+#define	MTYPE_CACHED_NV					0
+#define	MTYPE_CACHED					1
 #define	MTYPE_NONCACHED					3
 
 #define	DEFAULT_CP_HQD_PERSISTENT_STATE			(0x33U << 8)
@@ -67,6 +68,4 @@
 
 #define GRBM_GFX_INDEX					0x30800
 
-#define	ATC_VMID_PASID_MAPPING_VALID			(1U << 31)
-
 #endif
diff --git a/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler.h b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler.h
new file mode 100644
index 000000000000..0320163b6e74
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler.h
@@ -0,0 +1,4656 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+static const uint32_t cwsr_trap_gfx8_hex[] = {
+	0xbf820001, 0xbf820121,
+	0xb8f4f802, 0x89748674,
+	0xb8f5f803, 0x8675ff75,
+	0x00000400, 0xbf850017,
+	0xc00a1e37, 0x00000000,
+	0xbf8c007f, 0x87777978,
+	0xbf840005, 0x8f728374,
+	0xb972e0c2, 0xbf800002,
+	0xb9740002, 0xbe801d78,
+	0xb8f5f803, 0x8675ff75,
+	0x000001ff, 0xbf850002,
+	0x80708470, 0x82718071,
+	0x8671ff71, 0x0000ffff,
+	0x8f728374, 0xb972e0c2,
+	0xbf800002, 0xb9740002,
+	0xbe801f70, 0xbefa0080,
+	0xb97a0283, 0xbef60068,
+	0xbef70069, 0xb8fa1c07,
+	0x8e7a9c7a, 0x87717a71,
+	0xb8fa03c7, 0x8e7a9b7a,
+	0x87717a71, 0xb8faf807,
+	0x867aff7a, 0x00007fff,
+	0xb97af807, 0xbef2007e,
+	0xbef3007f, 0xbefe0180,
+	0xbf900004, 0x877a8474,
+	0xb97af802, 0xbf8e0002,
+	0xbf88fffe, 0xbef8007e,
+	0x8679ff7f, 0x0000ffff,
+	0x8779ff79, 0x00040000,
+	0xbefa0080, 0xbefb00ff,
+	0x00807fac, 0x867aff7f,
+	0x08000000, 0x8f7a837a,
+	0x877b7a7b, 0x867aff7f,
+	0x70000000, 0x8f7a817a,
+	0x877b7a7b, 0xbeef007c,
+	0xbeee0080, 0xb8ee2a05,
+	0x806e816e, 0x8e6e8a6e,
+	0xb8fa1605, 0x807a817a,
+	0x8e7a867a, 0x806e7a6e,
+	0xbefa0084, 0xbefa00ff,
+	0x01000000, 0xbefe007c,
+	0xbefc006e, 0xc0611bfc,
+	0x0000007c, 0x806e846e,
+	0xbefc007e, 0xbefe007c,
+	0xbefc006e, 0xc0611c3c,
+	0x0000007c, 0x806e846e,
+	0xbefc007e, 0xbefe007c,
+	0xbefc006e, 0xc0611c7c,
+	0x0000007c, 0x806e846e,
+	0xbefc007e, 0xbefe007c,
+	0xbefc006e, 0xc0611cbc,
+	0x0000007c, 0x806e846e,
+	0xbefc007e, 0xbefe007c,
+	0xbefc006e, 0xc0611cfc,
+	0x0000007c, 0x806e846e,
+	0xbefc007e, 0xbefe007c,
+	0xbefc006e, 0xc0611d3c,
+	0x0000007c, 0x806e846e,
+	0xbefc007e, 0xb8f5f803,
+	0xbefe007c, 0xbefc006e,
+	0xc0611d7c, 0x0000007c,
+	0x806e846e, 0xbefc007e,
+	0xbefe007c, 0xbefc006e,
+	0xc0611dbc, 0x0000007c,
+	0x806e846e, 0xbefc007e,
+	0xbefe007c, 0xbefc006e,
+	0xc0611dfc, 0x0000007c,
+	0x806e846e, 0xbefc007e,
+	0xb8eff801, 0xbefe007c,
+	0xbefc006e, 0xc0611bfc,
+	0x0000007c, 0x806e846e,
+	0xbefc007e, 0xbefe007c,
+	0xbefc006e, 0xc0611b3c,
+	0x0000007c, 0x806e846e,
+	0xbefc007e, 0xbefe007c,
+	0xbefc006e, 0xc0611b7c,
+	0x0000007c, 0x806e846e,
+	0xbefc007e, 0x867aff7f,
+	0x04000000, 0xbef30080,
+	0x8773737a, 0xb8ee2a05,
+	0x806e816e, 0x8e6e8a6e,
+	0xb8f51605, 0x80758175,
+	0x8e758475, 0x8e7a8275,
+	0xbefa00ff, 0x01000000,
+	0xbef60178, 0x80786e78,
+	0x82798079, 0xbefc0080,
+	0xbe802b00, 0xbe822b02,
+	0xbe842b04, 0xbe862b06,
+	0xbe882b08, 0xbe8a2b0a,
+	0xbe8c2b0c, 0xbe8e2b0e,
+	0xc06b003c, 0x00000000,
+	0xc06b013c, 0x00000010,
+	0xc06b023c, 0x00000020,
+	0xc06b033c, 0x00000030,
+	0x8078c078, 0x82798079,
+	0x807c907c, 0xbf0a757c,
+	0xbf85ffeb, 0xbef80176,
+	0xbeee0080, 0xbefe00c1,
+	0xbeff00c1, 0xbefa00ff,
+	0x01000000, 0xe0724000,
+	0x6e1e0000, 0xe0724100,
+	0x6e1e0100, 0xe0724200,
+	0x6e1e0200, 0xe0724300,
+	0x6e1e0300, 0xbefe00c1,
+	0xbeff00c1, 0xb8f54306,
+	0x8675c175, 0xbf84002c,
+	0xbf8a0000, 0x867aff73,
+	0x04000000, 0xbf840028,
+	0x8e758675, 0x8e758275,
+	0xbefa0075, 0xb8ee2a05,
+	0x806e816e, 0x8e6e8a6e,
+	0xb8fa1605, 0x807a817a,
+	0x8e7a867a, 0x806e7a6e,
+	0x806eff6e, 0x00000080,
+	0xbefa00ff, 0x01000000,
+	0xbefc0080, 0xd28c0002,
+	0x000100c1, 0xd28d0003,
+	0x000204c1, 0xd1060002,
+	0x00011103, 0x7e0602ff,
+	0x00000200, 0xbefc00ff,
+	0x00010000, 0xbe80007b,
+	0x867bff7b, 0xff7fffff,
+	0x877bff7b, 0x00058000,
+	0xd8ec0000, 0x00000002,
+	0xbf8c007f, 0xe0765000,
+	0x6e1e0002, 0x32040702,
+	0xd0c9006a, 0x0000eb02,
+	0xbf87fff7, 0xbefb0000,
+	0xbeee00ff, 0x00000400,
+	0xbefe00c1, 0xbeff00c1,
+	0xb8f52a05, 0x80758175,
+	0x8e758275, 0x8e7a8875,
+	0xbefa00ff, 0x01000000,
+	0xbefc0084, 0xbf0a757c,
+	0xbf840015, 0xbf11017c,
+	0x8075ff75, 0x00001000,
+	0x7e000300, 0x7e020301,
+	0x7e040302, 0x7e060303,
+	0xe0724000, 0x6e1e0000,
+	0xe0724100, 0x6e1e0100,
+	0xe0724200, 0x6e1e0200,
+	0xe0724300, 0x6e1e0300,
+	0x807c847c, 0x806eff6e,
+	0x00000400, 0xbf0a757c,
+	0xbf85ffef, 0xbf9c0000,
+	0xbf8200ce, 0xbef8007e,
+	0x8679ff7f, 0x0000ffff,
+	0x8779ff79, 0x00040000,
+	0xbefa0080, 0xbefb00ff,
+	0x00807fac, 0x8676ff7f,
+	0x08000000, 0x8f768376,
+	0x877b767b, 0x8676ff7f,
+	0x70000000, 0x8f768176,
+	0x877b767b, 0x8676ff7f,
+	0x04000000, 0xbf84001e,
+	0xbefe00c1, 0xbeff00c1,
+	0xb8f34306, 0x8673c173,
+	0xbf840019, 0x8e738673,
+	0x8e738273, 0xbefa0073,
+	0xb8f22a05, 0x80728172,
+	0x8e728a72, 0xb8f61605,
+	0x80768176, 0x8e768676,
+	0x80727672, 0x8072ff72,
+	0x00000080, 0xbefa00ff,
+	0x01000000, 0xbefc0080,
+	0xe0510000, 0x721e0000,
+	0xe0510100, 0x721e0000,
+	0x807cff7c, 0x00000200,
+	0x8072ff72, 0x00000200,
+	0xbf0a737c, 0xbf85fff6,
+	0xbef20080, 0xbefe00c1,
+	0xbeff00c1, 0xb8f32a05,
+	0x80738173, 0x8e738273,
+	0x8e7a8873, 0xbefa00ff,
+	0x01000000, 0xbef60072,
+	0x8072ff72, 0x00000400,
+	0xbefc0084, 0xbf11087c,
+	0x8073ff73, 0x00008000,
+	0xe0524000, 0x721e0000,
+	0xe0524100, 0x721e0100,
+	0xe0524200, 0x721e0200,
+	0xe0524300, 0x721e0300,
+	0xbf8c0f70, 0x7e000300,
+	0x7e020301, 0x7e040302,
+	0x7e060303, 0x807c847c,
+	0x8072ff72, 0x00000400,
+	0xbf0a737c, 0xbf85ffee,
+	0xbf9c0000, 0xe0524000,
+	0x761e0000, 0xe0524100,
+	0x761e0100, 0xe0524200,
+	0x761e0200, 0xe0524300,
+	0x761e0300, 0xbf8c0f70,
+	0xb8f22a05, 0x80728172,
+	0x8e728a72, 0xb8f61605,
+	0x80768176, 0x8e768676,
+	0x80727672, 0x80f2c072,
+	0xb8f31605, 0x80738173,
+	0x8e738473, 0x8e7a8273,
+	0xbefa00ff, 0x01000000,
+	0xbefc0073, 0xc031003c,
+	0x00000072, 0x80f2c072,
+	0xbf8c007f, 0x80fc907c,
+	0xbe802d00, 0xbe822d02,
+	0xbe842d04, 0xbe862d06,
+	0xbe882d08, 0xbe8a2d0a,
+	0xbe8c2d0c, 0xbe8e2d0e,
+	0xbf06807c, 0xbf84fff1,
+	0xb8f22a05, 0x80728172,
+	0x8e728a72, 0xb8f61605,
+	0x80768176, 0x8e768676,
+	0x80727672, 0xbefa0084,
+	0xbefa00ff, 0x01000000,
+	0xc0211cfc, 0x00000072,
+	0x80728472, 0xc0211c3c,
+	0x00000072, 0x80728472,
+	0xc0211c7c, 0x00000072,
+	0x80728472, 0xc0211bbc,
+	0x00000072, 0x80728472,
+	0xc0211bfc, 0x00000072,
+	0x80728472, 0xc0211d3c,
+	0x00000072, 0x80728472,
+	0xc0211d7c, 0x00000072,
+	0x80728472, 0xc0211a3c,
+	0x00000072, 0x80728472,
+	0xc0211a7c, 0x00000072,
+	0x80728472, 0xc0211dfc,
+	0x00000072, 0x80728472,
+	0xc0211b3c, 0x00000072,
+	0x80728472, 0xc0211b7c,
+	0x00000072, 0x80728472,
+	0xbf8c007f, 0xbefc0073,
+	0xbefe006e, 0xbeff006f,
+	0x867375ff, 0x000003ff,
+	0xb9734803, 0x867375ff,
+	0xfffff800, 0x8f738b73,
+	0xb973a2c3, 0xb977f801,
+	0x8673ff71, 0xf0000000,
+	0x8f739c73, 0x8e739073,
+	0xbef60080, 0x87767376,
+	0x8673ff71, 0x08000000,
+	0x8f739b73, 0x8e738f73,
+	0x87767376, 0x8673ff74,
+	0x00800000, 0x8f739773,
+	0xb976f807, 0x8671ff71,
+	0x0000ffff, 0x86fe7e7e,
+	0x86ea6a6a, 0x8f768374,
+	0xb976e0c2, 0xbf800002,
+	0xb9740002, 0xbf8a0000,
+	0x95807370, 0xbf810000,
+};
+
+
+static const uint32_t cwsr_trap_gfx9_hex[] = {
+	0xbf820001, 0xbf820259,
+	0xb8f8f802, 0x8978ff78,
+	0x00020006, 0xb8fbf803,
+	0x866eff78, 0x00002000,
+	0xbf840009, 0x866eff6d,
+	0x00ff0000, 0xbf85001e,
+	0x866eff7b, 0x00000400,
+	0xbf850055, 0xbf8e0010,
+	0xb8fbf803, 0xbf82fffa,
+	0x866eff7b, 0x03c00900,
+	0xbf850015, 0x866eff7b,
+	0x000071ff, 0xbf840008,
+	0x866fff7b, 0x00007080,
+	0xbf840001, 0xbeee1a87,
+	0xb8eff801, 0x8e6e8c6e,
+	0x866e6f6e, 0xbf85000a,
+	0x866eff6d, 0x00ff0000,
+	0xbf850007, 0xb8eef801,
+	0x866eff6e, 0x00000800,
+	0xbf850003, 0x866eff7b,
+	0x00000400, 0xbf85003a,
+	0xb8faf807, 0x867aff7a,
+	0x001f8000, 0x8e7a8b7a,
+	0x8977ff77, 0xfc000000,
+	0x87777a77, 0xba7ff807,
+	0x00000000, 0xb8faf812,
+	0xb8fbf813, 0x8efa887a,
+	0xbf0d8f7b, 0xbf840002,
+	0x877bff7b, 0xffff0000,
+	0xc0031bbd, 0x00000010,
+	0xbf8cc07f, 0x8e6e976e,
+	0x8977ff77, 0x00800000,
+	0x87776e77, 0xc0071bbd,
+	0x00000000, 0xbf8cc07f,
+	0xc0071ebd, 0x00000008,
+	0xbf8cc07f, 0x86ee6e6e,
+	0xbf840001, 0xbe801d6e,
+	0x866eff6d, 0x01ff0000,
+	0xbf850005, 0x8778ff78,
+	0x00002000, 0x80ec886c,
+	0x82ed806d, 0xbf820005,
+	0x866eff6d, 0x01000000,
+	0xbf850002, 0x806c846c,
+	0x826d806d, 0x866dff6d,
+	0x0000ffff, 0x8f7a8b77,
+	0x867aff7a, 0x001f8000,
+	0xb97af807, 0x86fe7e7e,
+	0x86ea6a6a, 0x8f6e8378,
+	0xb96ee0c2, 0xbf800002,
+	0xb9780002, 0xbe801f6c,
+	0x866dff6d, 0x0000ffff,
+	0xbefa0080, 0xb97a0283,
+	0xb8faf807, 0x867aff7a,
+	0x001f8000, 0x8e7a8b7a,
+	0x8977ff77, 0xfc000000,
+	0x87777a77, 0xba7ff807,
+	0x00000000, 0xbeee007e,
+	0xbeef007f, 0xbefe0180,
+	0xbf900004, 0x877a8478,
+	0xb97af802, 0xbf8e0002,
+	0xbf88fffe, 0xb8fa2a05,
+	0x807a817a, 0x8e7a8a7a,
+	0xb8fb1605, 0x807b817b,
+	0x8e7b867b, 0x807a7b7a,
+	0x807a7e7a, 0x827b807f,
+	0x867bff7b, 0x0000ffff,
+	0xc04b1c3d, 0x00000050,
+	0xbf8cc07f, 0xc04b1d3d,
+	0x00000060, 0xbf8cc07f,
+	0xc0431e7d, 0x00000074,
+	0xbf8cc07f, 0xbef4007e,
+	0x8675ff7f, 0x0000ffff,
+	0x8775ff75, 0x00040000,
+	0xbef60080, 0xbef700ff,
+	0x00807fac, 0xbef1007c,
+	0xbef00080, 0xb8f02a05,
+	0x80708170, 0x8e708a70,
+	0xb8fa1605, 0x807a817a,
+	0x8e7a867a, 0x80707a70,
+	0xbef60084, 0xbef600ff,
+	0x01000000, 0xbefe007c,
+	0xbefc0070, 0xc0611c7a,
+	0x0000007c, 0xbf8cc07f,
+	0x80708470, 0xbefc007e,
+	0xbefe007c, 0xbefc0070,
+	0xc0611b3a, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xbefe007c,
+	0xbefc0070, 0xc0611b7a,
+	0x0000007c, 0xbf8cc07f,
+	0x80708470, 0xbefc007e,
+	0xbefe007c, 0xbefc0070,
+	0xc0611bba, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xbefe007c,
+	0xbefc0070, 0xc0611bfa,
+	0x0000007c, 0xbf8cc07f,
+	0x80708470, 0xbefc007e,
+	0xbefe007c, 0xbefc0070,
+	0xc0611e3a, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xb8fbf803,
+	0xbefe007c, 0xbefc0070,
+	0xc0611efa, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xbefe007c,
+	0xbefc0070, 0xc0611a3a,
+	0x0000007c, 0xbf8cc07f,
+	0x80708470, 0xbefc007e,
+	0xbefe007c, 0xbefc0070,
+	0xc0611a7a, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xb8f1f801,
+	0xbefe007c, 0xbefc0070,
+	0xc0611c7a, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xbf108080,
+	0x867aff7f, 0x04000000,
+	0xbeef0080, 0x876f6f7a,
+	0xb8f02a05, 0x80708170,
+	0x8e708a70, 0xb8fb1605,
+	0x807b817b, 0x8e7b847b,
+	0x8e76827b, 0xbef600ff,
+	0x01000000, 0xbef20174,
+	0x80747074, 0x82758075,
+	0xbefc0080, 0xbf800000,
+	0xbe802b00, 0xbe822b02,
+	0xbe842b04, 0xbe862b06,
+	0xbe882b08, 0xbe8a2b0a,
+	0xbe8c2b0c, 0xbe8e2b0e,
+	0xc06b003a, 0x00000000,
+	0xbf8cc07f, 0xc06b013a,
+	0x00000010, 0xbf8cc07f,
+	0xc06b023a, 0x00000020,
+	0xbf8cc07f, 0xc06b033a,
+	0x00000030, 0xbf8cc07f,
+	0x8074c074, 0x82758075,
+	0x807c907c, 0xbf0a7b7c,
+	0xbf85ffe7, 0xbef40172,
+	0xbef00080, 0xbefe00c1,
+	0xbeff00c1, 0xbee80080,
+	0xbee90080, 0xbef600ff,
+	0x01000000, 0x867aff78,
+	0x00400000, 0xbf850003,
+	0xb8faf803, 0x897a7aff,
+	0x10000000, 0xbf85004d,
+	0xbe840080, 0xd2890000,
+	0x00000900, 0x80048104,
+	0xd2890001, 0x00000900,
+	0x80048104, 0xd2890002,
+	0x00000900, 0x80048104,
+	0xd2890003, 0x00000900,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000901,
+	0x80048104, 0xd2890001,
+	0x00000901, 0x80048104,
+	0xd2890002, 0x00000901,
+	0x80048104, 0xd2890003,
+	0x00000901, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000902, 0x80048104,
+	0xd2890001, 0x00000902,
+	0x80048104, 0xd2890002,
+	0x00000902, 0x80048104,
+	0xd2890003, 0x00000902,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000903,
+	0x80048104, 0xd2890001,
+	0x00000903, 0x80048104,
+	0xd2890002, 0x00000903,
+	0x80048104, 0xd2890003,
+	0x00000903, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbf820008, 0xe0724000,
+	0x701d0000, 0xe0724100,
+	0x701d0100, 0xe0724200,
+	0x701d0200, 0xe0724300,
+	0x701d0300, 0xbefe00c1,
+	0xbeff00c1, 0xb8fb4306,
+	0x867bc17b, 0xbf840063,
+	0xbf8a0000, 0x867aff6f,
+	0x04000000, 0xbf84005f,
+	0x8e7b867b, 0x8e7b827b,
+	0xbef6007b, 0xb8f02a05,
+	0x80708170, 0x8e708a70,
+	0xb8fa1605, 0x807a817a,
+	0x8e7a867a, 0x80707a70,
+	0x8070ff70, 0x00000080,
+	0xbef600ff, 0x01000000,
+	0xbefc0080, 0xd28c0002,
+	0x000100c1, 0xd28d0003,
+	0x000204c1, 0x867aff78,
+	0x00400000, 0xbf850003,
+	0xb8faf803, 0x897a7aff,
+	0x10000000, 0xbf850030,
+	0x24040682, 0xd86e4000,
+	0x00000002, 0xbf8cc07f,
+	0xbe840080, 0xd2890000,
+	0x00000900, 0x80048104,
+	0xd2890001, 0x00000900,
+	0x80048104, 0xd2890002,
+	0x00000900, 0x80048104,
+	0xd2890003, 0x00000900,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000901,
+	0x80048104, 0xd2890001,
+	0x00000901, 0x80048104,
+	0xd2890002, 0x00000901,
+	0x80048104, 0xd2890003,
+	0x00000901, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0x680404ff, 0x00000200,
+	0xd0c9006a, 0x0000f702,
+	0xbf87ffd2, 0xbf820015,
+	0xd1060002, 0x00011103,
+	0x7e0602ff, 0x00000200,
+	0xbefc00ff, 0x00010000,
+	0xbe800077, 0x8677ff77,
+	0xff7fffff, 0x8777ff77,
+	0x00058000, 0xd8ec0000,
+	0x00000002, 0xbf8cc07f,
+	0xe0765000, 0x701d0002,
+	0x68040702, 0xd0c9006a,
+	0x0000f702, 0xbf87fff7,
+	0xbef70000, 0xbef000ff,
+	0x00000400, 0xbefe00c1,
+	0xbeff00c1, 0xb8fb2a05,
+	0x807b817b, 0x8e7b827b,
+	0xbef600ff, 0x01000000,
+	0xbefc0084, 0xbf0a7b7c,
+	0xbf84006d, 0xbf11017c,
+	0x807bff7b, 0x00001000,
+	0x867aff78, 0x00400000,
+	0xbf850003, 0xb8faf803,
+	0x897a7aff, 0x10000000,
+	0xbf850051, 0xbe840080,
+	0xd2890000, 0x00000900,
+	0x80048104, 0xd2890001,
+	0x00000900, 0x80048104,
+	0xd2890002, 0x00000900,
+	0x80048104, 0xd2890003,
+	0x00000900, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000901, 0x80048104,
+	0xd2890001, 0x00000901,
+	0x80048104, 0xd2890002,
+	0x00000901, 0x80048104,
+	0xd2890003, 0x00000901,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000902,
+	0x80048104, 0xd2890001,
+	0x00000902, 0x80048104,
+	0xd2890002, 0x00000902,
+	0x80048104, 0xd2890003,
+	0x00000902, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000903, 0x80048104,
+	0xd2890001, 0x00000903,
+	0x80048104, 0xd2890002,
+	0x00000903, 0x80048104,
+	0xd2890003, 0x00000903,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0x807c847c,
+	0xbf0a7b7c, 0xbf85ffb1,
+	0xbf9c0000, 0xbf820012,
+	0x7e000300, 0x7e020301,
+	0x7e040302, 0x7e060303,
+	0xe0724000, 0x701d0000,
+	0xe0724100, 0x701d0100,
+	0xe0724200, 0x701d0200,
+	0xe0724300, 0x701d0300,
+	0x807c847c, 0x8070ff70,
+	0x00000400, 0xbf0a7b7c,
+	0xbf85ffef, 0xbf9c0000,
+	0xbf8200c7, 0xbef4007e,
+	0x8675ff7f, 0x0000ffff,
+	0x8775ff75, 0x00040000,
+	0xbef60080, 0xbef700ff,
+	0x00807fac, 0x866eff7f,
+	0x04000000, 0xbf84001e,
+	0xbefe00c1, 0xbeff00c1,
+	0xb8ef4306, 0x866fc16f,
+	0xbf840019, 0x8e6f866f,
+	0x8e6f826f, 0xbef6006f,
+	0xb8f82a05, 0x80788178,
+	0x8e788a78, 0xb8ee1605,
+	0x806e816e, 0x8e6e866e,
+	0x80786e78, 0x8078ff78,
+	0x00000080, 0xbef600ff,
+	0x01000000, 0xbefc0080,
+	0xe0510000, 0x781d0000,
+	0xe0510100, 0x781d0000,
+	0x807cff7c, 0x00000200,
+	0x8078ff78, 0x00000200,
+	0xbf0a6f7c, 0xbf85fff6,
+	0xbefe00c1, 0xbeff00c1,
+	0xbef600ff, 0x01000000,
+	0xb8ef2a05, 0x806f816f,
+	0x8e6f826f, 0x806fff6f,
+	0x00008000, 0xbef80080,
+	0xbeee0078, 0x8078ff78,
+	0x00000400, 0xbefc0084,
+	0xbf11087c, 0xe0524000,
+	0x781d0000, 0xe0524100,
+	0x781d0100, 0xe0524200,
+	0x781d0200, 0xe0524300,
+	0x781d0300, 0xbf8c0f70,
+	0x7e000300, 0x7e020301,
+	0x7e040302, 0x7e060303,
+	0x807c847c, 0x8078ff78,
+	0x00000400, 0xbf0a6f7c,
+	0xbf85ffee, 0xbf9c0000,
+	0xe0524000, 0x6e1d0000,
+	0xe0524100, 0x6e1d0100,
+	0xe0524200, 0x6e1d0200,
+	0xe0524300, 0x6e1d0300,
+	0xbf8c0f70, 0xb8f82a05,
+	0x80788178, 0x8e788a78,
+	0xb8ee1605, 0x806e816e,
+	0x8e6e866e, 0x80786e78,
+	0x80f8c078, 0xb8ef1605,
+	0x806f816f, 0x8e6f846f,
+	0x8e76826f, 0xbef600ff,
+	0x01000000, 0xbefc006f,
+	0xc031003a, 0x00000078,
+	0x80f8c078, 0xbf8cc07f,
+	0x80fc907c, 0xbf800000,
+	0xbe802d00, 0xbe822d02,
+	0xbe842d04, 0xbe862d06,
+	0xbe882d08, 0xbe8a2d0a,
+	0xbe8c2d0c, 0xbe8e2d0e,
+	0xbf06807c, 0xbf84fff0,
+	0xb8f82a05, 0x80788178,
+	0x8e788a78, 0xb8ee1605,
+	0x806e816e, 0x8e6e866e,
+	0x80786e78, 0xbef60084,
+	0xbef600ff, 0x01000000,
+	0xc0211bfa, 0x00000078,
+	0x80788478, 0xc0211b3a,
+	0x00000078, 0x80788478,
+	0xc0211b7a, 0x00000078,
+	0x80788478, 0xc0211c3a,
+	0x00000078, 0x80788478,
+	0xc0211c7a, 0x00000078,
+	0x80788478, 0xc0211eba,
+	0x00000078, 0x80788478,
+	0xc0211efa, 0x00000078,
+	0x80788478, 0xc0211a3a,
+	0x00000078, 0x80788478,
+	0xc0211a7a, 0x00000078,
+	0x80788478, 0xc0211cfa,
+	0x00000078, 0x80788478,
+	0xbf8cc07f, 0xbefc006f,
+	0xbefe0070, 0xbeff0071,
+	0x866f7bff, 0x000003ff,
+	0xb96f4803, 0x866f7bff,
+	0xfffff800, 0x8f6f8b6f,
+	0xb96fa2c3, 0xb973f801,
+	0xb8ee2a05, 0x806e816e,
+	0x8e6e8a6e, 0xb8ef1605,
+	0x806f816f, 0x8e6f866f,
+	0x806e6f6e, 0x806e746e,
+	0x826f8075, 0x866fff6f,
+	0x0000ffff, 0xc00b1c37,
+	0x00000050, 0xc00b1d37,
+	0x00000060, 0xc0031e77,
+	0x00000074, 0xbf8cc07f,
+	0x8f6e8b77, 0x866eff6e,
+	0x001f8000, 0xb96ef807,
+	0x866dff6d, 0x0000ffff,
+	0x86fe7e7e, 0x86ea6a6a,
+	0x8f6e837a, 0xb96ee0c2,
+	0xbf800002, 0xb97a0002,
+	0xbf8a0000, 0xbe801f6c,
+	0xbf9b0000, 0x00000000,
+};
+
+static const uint32_t cwsr_trap_nv1x_hex[] = {
+	0xbf820001, 0xbf820393,
+	0xb0804004, 0xb978f802,
+	0x8a78ff78, 0x00020006,
+	0xb97bf803, 0x876eff78,
+	0x00002000, 0xbf840009,
+	0x876eff6d, 0x00ff0000,
+	0xbf85001e, 0x876eff7b,
+	0x00000400, 0xbf85005b,
+	0xbf8e0010, 0xb97bf803,
+	0xbf82fffa, 0x876eff7b,
+	0x00000900, 0xbf850015,
+	0x876eff7b, 0x000071ff,
+	0xbf840008, 0x876fff7b,
+	0x00007080, 0xbf840001,
+	0xbeee1d87, 0xb96ff801,
+	0x8f6e8c6e, 0x876e6f6e,
+	0xbf85000a, 0x876eff6d,
+	0x00ff0000, 0xbf850007,
+	0xb96ef801, 0x876eff6e,
+	0x00000800, 0xbf850003,
+	0x876eff7b, 0x00000400,
+	0xbf850040, 0x8a77ff77,
+	0xff000000, 0xb97af807,
+	0x877bff7a, 0x02000000,
+	0x8f7b867b, 0x88777b77,
+	0x877bff7a, 0x003f8000,
+	0x8f7b897b, 0x88777b77,
+	0x8a7aff7a, 0x023f8000,
+	0xb9faf807, 0xb97af812,
+	0xb97bf813, 0x8ffa887a,
+	0xbf0d8f7b, 0xbf840002,
+	0x887bff7b, 0xffff0000,
+	0xf4011bbd, 0xfa000010,
+	0xbf8c0000, 0x8f6e976e,
+	0x8a77ff77, 0x00800000,
+	0x88776e77, 0xf4051bbd,
+	0xfa000000, 0xbf8c0000,
+	0xf4051ebd, 0xfa000008,
+	0xbf8c0000, 0x87ee6e6e,
+	0xbf840001, 0xbe80206e,
+	0x876eff6d, 0x00ff0000,
+	0xbf850008, 0x876eff6d,
+	0x01000000, 0xbf850007,
+	0x8878ff78, 0x00002000,
+	0x80ec886c, 0x82ed806d,
+	0xbf820002, 0x806c846c,
+	0x826d806d, 0x876dff6d,
+	0x0000ffff, 0x907a8977,
+	0x877bff7a, 0x003f8000,
+	0x907a8677, 0x877aff7a,
+	0x02000000, 0x887a7b7a,
+	0xb9faf807, 0x87fe7e7e,
+	0x87ea6a6a, 0xb9f8f802,
+	0xbe80226c, 0x876dff6d,
+	0x0000ffff, 0xbefa0380,
+	0xb9fa0283, 0x8a77ff77,
+	0xff000000, 0xb97af807,
+	0x877bff7a, 0x02000000,
+	0x8f7b867b, 0x88777b77,
+	0x877bff7a, 0x003f8000,
+	0x8f7b897b, 0x88777b77,
+	0x8a7aff7a, 0x023f8000,
+	0xb9faf807, 0xbeee037e,
+	0xbeef037f, 0xbefe0480,
+	0xbf900004, 0xbf8e0002,
+	0xbf88fffe, 0x877aff7f,
+	0x04000000, 0x8f7a857a,
+	0x886d7a6d, 0xb97b02dc,
+	0x8f7b997b, 0xb97a3a05,
+	0x807a817a, 0xbf0d997b,
+	0xbf850002, 0x8f7a897a,
+	0xbf820001, 0x8f7a8a7a,
+	0xb97b1e06, 0x8f7b8a7b,
+	0x807a7b7a, 0x877bff7f,
+	0x0000ffff, 0x807aff7a,
+	0x00000200, 0x807a7e7a,
+	0x827b807b, 0xf4491c3d,
+	0xfa000050, 0xf4491d3d,
+	0xfa000060, 0xf4411e7d,
+	0xfa000074, 0xbef4037e,
+	0x8775ff7f, 0x0000ffff,
+	0x8875ff75, 0x00040000,
+	0xbef60380, 0xbef703ff,
+	0x10807fac, 0xbef1037c,
+	0xbef00380, 0xb97302dc,
+	0x8f739973, 0xb97bf816,
+	0xba80f816, 0x00000000,
+	0xbefe03c1, 0x907c9973,
+	0x877c817c, 0xbf06817c,
+	0xbf850002, 0xbeff0380,
+	0xbf820002, 0xbeff03c1,
+	0xbf820058, 0xbef603ff,
+	0x01000000, 0xb97af803,
+	0x8a7a7aff, 0x10000000,
+	0xbf850049, 0xbe840380,
+	0xd7600000, 0x00000900,
+	0x80048104, 0xd7600001,
+	0x00000900, 0x80048104,
+	0xd7600002, 0x00000900,
+	0x80048104, 0xd7600003,
+	0x00000900, 0x80048104,
+	0xf469003a, 0xe0000000,
+	0x80709070, 0xbf06a004,
+	0xbf84ffef, 0xbe840380,
+	0xd7600000, 0x00000901,
+	0x80048104, 0xd7600001,
+	0x00000901, 0x80048104,
+	0xd7600002, 0x00000901,
+	0x80048104, 0xd7600003,
+	0x00000901, 0x80048104,
+	0xf469003a, 0xe0000000,
+	0x80709070, 0xbf06a004,
+	0xbf84ffef, 0xbe840380,
+	0xd7600000, 0x00000902,
+	0x80048104, 0xd7600001,
+	0x00000902, 0x80048104,
+	0xd7600002, 0x00000902,
+	0x80048104, 0xd7600003,
+	0x00000902, 0x80048104,
+	0xf469003a, 0xe0000000,
+	0x80709070, 0xbf06a004,
+	0xbf84ffef, 0xbe840380,
+	0xd7600000, 0x00000903,
+	0x80048104, 0xd7600001,
+	0x00000903, 0x80048104,
+	0xd7600002, 0x00000903,
+	0x80048104, 0xd7600003,
+	0x00000903, 0x80048104,
+	0xf469003a, 0xe0000000,
+	0x80709070, 0xbf06a004,
+	0xbf84ffef, 0xbf820060,
+	0xe0704000, 0x705d0000,
+	0xe0704080, 0x705d0100,
+	0xe0704100, 0x705d0200,
+	0xe0704180, 0x705d0300,
+	0xbf820057, 0xbef603ff,
+	0x01000000, 0xb97af803,
+	0x8a7a7aff, 0x10000000,
+	0xbf850049, 0xbe840380,
+	0xd7600000, 0x00000900,
+	0x80048104, 0xd7600001,
+	0x00000900, 0x80048104,
+	0xd7600002, 0x00000900,
+	0x80048104, 0xd7600003,
+	0x00000900, 0x80048104,
+	0xf469003a, 0xe0000000,
+	0x80709070, 0xbf06c004,
+	0xbf84ffef, 0xbe840380,
+	0xd7600000, 0x00000901,
+	0x80048104, 0xd7600001,
+	0x00000901, 0x80048104,
+	0xd7600002, 0x00000901,
+	0x80048104, 0xd7600003,
+	0x00000901, 0x80048104,
+	0xf469003a, 0xe0000000,
+	0x80709070, 0xbf06c004,
+	0xbf84ffef, 0xbe840380,
+	0xd7600000, 0x00000902,
+	0x80048104, 0xd7600001,
+	0x00000902, 0x80048104,
+	0xd7600002, 0x00000902,
+	0x80048104, 0xd7600003,
+	0x00000902, 0x80048104,
+	0xf469003a, 0xe0000000,
+	0x80709070, 0xbf06c004,
+	0xbf84ffef, 0xbe840380,
+	0xd7600000, 0x00000903,
+	0x80048104, 0xd7600001,
+	0x00000903, 0x80048104,
+	0xd7600002, 0x00000903,
+	0x80048104, 0xd7600003,
+	0x00000903, 0x80048104,
+	0xf469003a, 0xe0000000,
+	0x80709070, 0xbf06c004,
+	0xbf84ffef, 0xbf820008,
+	0xe0704000, 0x705d0000,
+	0xe0704100, 0x705d0100,
+	0xe0704200, 0x705d0200,
+	0xe0704300, 0x705d0300,
+	0xb9703a05, 0x80708170,
+	0xbf0d9973, 0xbf850002,
+	0x8f708970, 0xbf820001,
+	0x8f708a70, 0xb97a1e06,
+	0x8f7a8a7a, 0x80707a70,
+	0x8070ff70, 0x00000200,
+	0xbef603ff, 0x01000000,
+	0xbefe037c, 0xbefc0370,
+	0xf4611c7a, 0xf8000000,
+	0x80708470, 0xbefc037e,
+	0xbefe037c, 0xbefc0370,
+	0xf4611b3a, 0xf8000000,
+	0x80708470, 0xbefc037e,
+	0x8a7aff6d, 0x80000000,
+	0xbefe037c, 0xbefc0370,
+	0xf4611eba, 0xf8000000,
+	0x80708470, 0xbefc037e,
+	0xbefe037c, 0xbefc0370,
+	0xf4611bba, 0xf8000000,
+	0x80708470, 0xbefc037e,
+	0xbefe037c, 0xbefc0370,
+	0xf4611bfa, 0xf8000000,
+	0x80708470, 0xbefc037e,
+	0xbefe037c, 0xbefc0370,
+	0xf4611e3a, 0xf8000000,
+	0x80708470, 0xbefc037e,
+	0xb97af803, 0xbefe037c,
+	0xbefc0370, 0xf4611eba,
+	0xf8000000, 0x80708470,
+	0xbefc037e, 0xbefe037c,
+	0xbefc0370, 0xf4611efa,
+	0xf8000000, 0x80708470,
+	0xbefc037e, 0xb971f801,
+	0xbefe037c, 0xbefc0370,
+	0xf4611c7a, 0xf8000000,
+	0x80708470, 0xbefc037e,
+	0xb971f814, 0xbefe037c,
+	0xbefc0370, 0xf4611c7a,
+	0xf8000000, 0x80708470,
+	0xbefc037e, 0xb971f815,
+	0xbefe037c, 0xbefc0370,
+	0xf4611c7a, 0xf8000000,
+	0x80708470, 0xbefc037e,
+	0xb9703a05, 0x80708170,
+	0xbf0d9973, 0xbf850002,
+	0x8f708970, 0xbf820001,
+	0x8f708a70, 0xb97a1e06,
+	0x8f7a8a7a, 0x80707a70,
+	0xbef603ff, 0x01000000,
+	0xbefb0374, 0x80747074,
+	0x82758075, 0xbefc0380,
+	0xbf800000, 0xbe802f00,
+	0xbe822f02, 0xbe842f04,
+	0xbe862f06, 0xbe882f08,
+	0xbe8a2f0a, 0xbe8c2f0c,
+	0xbe8e2f0e, 0xf469003a,
+	0xfa000000, 0xf469013a,
+	0xfa000010, 0xf469023a,
+	0xfa000020, 0xf469033a,
+	0xfa000030, 0x8074c074,
+	0x82758075, 0x807c907c,
+	0xbf0aff7c, 0x00000060,
+	0xbf85ffea, 0xbe802f00,
+	0xbe822f02, 0xbe842f04,
+	0xbe862f06, 0xbe882f08,
+	0xbe8a2f0a, 0xf469003a,
+	0xfa000000, 0xf469013a,
+	0xfa000010, 0xf469023a,
+	0xfa000020, 0x8074b074,
+	0x82758075, 0xbef4037b,
+	0xbefe03c1, 0x907c9973,
+	0x877c817c, 0xbf06817c,
+	0xbf850002, 0xbeff0380,
+	0xbf820001, 0xbeff03c1,
+	0xb97b4306, 0x877bc17b,
+	0xbf840085, 0xbf8a0000,
+	0x877aff6d, 0x80000000,
+	0xbf840081, 0x8f7b887b,
+	0xbef6037b, 0xb9703a05,
+	0x80708170, 0xbf0d9973,
+	0xbf850002, 0x8f708970,
+	0xbf820001, 0x8f708a70,
+	0xb97a1e06, 0x8f7a8a7a,
+	0x80707a70, 0x8070ff70,
+	0x00000200, 0x8070ff70,
+	0x00000080, 0xbef603ff,
+	0x01000000, 0xd7650000,
+	0x000100c1, 0xd7660000,
+	0x000200c1, 0x16000084,
+	0x907c9973, 0x877c817c,
+	0xbf06817c, 0xbefc0380,
+	0xbf850033, 0xb97af803,
+	0x8a7a7aff, 0x10000000,
+	0xbf85001d, 0xd8d80000,
+	0x01000000, 0xbf8c0000,
+	0xbe840380, 0xd7600000,
+	0x00000901, 0x80048104,
+	0xd7600001, 0x00000901,
+	0x80048104, 0xd7600002,
+	0x00000901, 0x80048104,
+	0xd7600003, 0x00000901,
+	0x80048104, 0xf469003a,
+	0xe0000000, 0x80709070,
+	0xbf06a004, 0xbf84ffef,
+	0x807cff7c, 0x00000080,
+	0xd5250000, 0x0001ff00,
+	0x00000080, 0xbf0a7b7c,
+	0xbf85ffe4, 0xbf820044,
+	0xbe8303ff, 0x00000080,
+	0xbf800000, 0xbf800000,
+	0xbf800000, 0xd8d80000,
+	0x01000000, 0xbf8c0000,
+	0xe0704000, 0x705d0100,
+	0x807c037c, 0x80700370,
+	0xd5250000, 0x0001ff00,
+	0x00000080, 0xbf0a7b7c,
+	0xbf85fff4, 0xbf820032,
+	0xb97af803, 0x8a7a7aff,
+	0x10000000, 0xbf85001d,
+	0xd8d80000, 0x01000000,
+	0xbf8c0000, 0xbe840380,
+	0xd7600000, 0x00000901,
+	0x80048104, 0xd7600001,
+	0x00000901, 0x80048104,
+	0xd7600002, 0x00000901,
+	0x80048104, 0xd7600003,
+	0x00000901, 0x80048104,
+	0xf469003a, 0xe0000000,
+	0x80709070, 0xbf06c004,
+	0xbf84ffef, 0x807cff7c,
+	0x00000100, 0xd5250000,
+	0x0001ff00, 0x00000100,
+	0xbf0a7b7c, 0xbf85ffe4,
+	0xbf820011, 0xbe8303ff,
+	0x00000100, 0xbf800000,
+	0xbf800000, 0xbf800000,
+	0xd8d80000, 0x01000000,
+	0xbf8c0000, 0xe0704000,
+	0x705d0100, 0x807c037c,
+	0x80700370, 0xd5250000,
+	0x0001ff00, 0x00000100,
+	0xbf0a7b7c, 0xbf85fff4,
+	0xbefe03c1, 0x907c9973,
+	0x877c817c, 0xbf06817c,
+	0xbf850004, 0xbef003ff,
+	0x00000200, 0xbeff0380,
+	0xbf820003, 0xbef003ff,
+	0x00000400, 0xbeff03c1,
+	0xb97b3a05, 0x807b817b,
+	0x8f7b827b, 0x907c9973,
+	0x877c817c, 0xbf06817c,
+	0xbf85006b, 0xbef603ff,
+	0x01000000, 0xbefc0384,
+	0xbf0a7b7c, 0xbf8400fa,
+	0xb97af803, 0x8a7a7aff,
+	0x10000000, 0xbf850050,
+	0x7e008700, 0x7e028701,
+	0x7e048702, 0x7e068703,
+	0xbe840380, 0xd7600000,
+	0x00000900, 0x80048104,
+	0xd7600001, 0x00000900,
+	0x80048104, 0xd7600002,
+	0x00000900, 0x80048104,
+	0xd7600003, 0x00000900,
+	0x80048104, 0xf469003a,
+	0xe0000000, 0x80709070,
+	0xbf06a004, 0xbf84ffef,
+	0xbe840380, 0xd7600000,
+	0x00000901, 0x80048104,
+	0xd7600001, 0x00000901,
+	0x80048104, 0xd7600002,
+	0x00000901, 0x80048104,
+	0xd7600003, 0x00000901,
+	0x80048104, 0xf469003a,
+	0xe0000000, 0x80709070,
+	0xbf06a004, 0xbf84ffef,
+	0xbe840380, 0xd7600000,
+	0x00000902, 0x80048104,
+	0xd7600001, 0x00000902,
+	0x80048104, 0xd7600002,
+	0x00000902, 0x80048104,
+	0xd7600003, 0x00000902,
+	0x80048104, 0xf469003a,
+	0xe0000000, 0x80709070,
+	0xbf06a004, 0xbf84ffef,
+	0xbe840380, 0xd7600000,
+	0x00000903, 0x80048104,
+	0xd7600001, 0x00000903,
+	0x80048104, 0xd7600002,
+	0x00000903, 0x80048104,
+	0xd7600003, 0x00000903,
+	0x80048104, 0xf469003a,
+	0xe0000000, 0x80709070,
+	0xbf06a004, 0xbf84ffef,
+	0x807c847c, 0xbf0a7b7c,
+	0xbf85ffb1, 0xbf8200a6,
+	0x7e008700, 0x7e028701,
+	0x7e048702, 0x7e068703,
+	0xe0704000, 0x705d0000,
+	0xe0704080, 0x705d0100,
+	0xe0704100, 0x705d0200,
+	0xe0704180, 0x705d0300,
+	0x807c847c, 0x8070ff70,
+	0x00000200, 0xbf0a7b7c,
+	0xbf85ffef, 0xbf820094,
+	0xbef603ff, 0x01000000,
+	0xbefc0384, 0xbf0a7b7c,
+	0xbf840065, 0xb97af803,
+	0x8a7a7aff, 0x10000000,
+	0xbf850050, 0x7e008700,
+	0x7e028701, 0x7e048702,
+	0x7e068703, 0xbe840380,
+	0xd7600000, 0x00000900,
+	0x80048104, 0xd7600001,
+	0x00000900, 0x80048104,
+	0xd7600002, 0x00000900,
+	0x80048104, 0xd7600003,
+	0x00000900, 0x80048104,
+	0xf469003a, 0xe0000000,
+	0x80709070, 0xbf06c004,
+	0xbf84ffef, 0xbe840380,
+	0xd7600000, 0x00000901,
+	0x80048104, 0xd7600001,
+	0x00000901, 0x80048104,
+	0xd7600002, 0x00000901,
+	0x80048104, 0xd7600003,
+	0x00000901, 0x80048104,
+	0xf469003a, 0xe0000000,
+	0x80709070, 0xbf06c004,
+	0xbf84ffef, 0xbe840380,
+	0xd7600000, 0x00000902,
+	0x80048104, 0xd7600001,
+	0x00000902, 0x80048104,
+	0xd7600002, 0x00000902,
+	0x80048104, 0xd7600003,
+	0x00000902, 0x80048104,
+	0xf469003a, 0xe0000000,
+	0x80709070, 0xbf06c004,
+	0xbf84ffef, 0xbe840380,
+	0xd7600000, 0x00000903,
+	0x80048104, 0xd7600001,
+	0x00000903, 0x80048104,
+	0xd7600002, 0x00000903,
+	0x80048104, 0xd7600003,
+	0x00000903, 0x80048104,
+	0xf469003a, 0xe0000000,
+	0x80709070, 0xbf06c004,
+	0xbf84ffef, 0x807c847c,
+	0xbf0a7b7c, 0xbf85ffb1,
+	0xbf82003b, 0x7e008700,
+	0x7e028701, 0x7e048702,
+	0x7e068703, 0xe0704000,
+	0x705d0000, 0xe0704100,
+	0x705d0100, 0xe0704200,
+	0x705d0200, 0xe0704300,
+	0x705d0300, 0x807c847c,
+	0x8070ff70, 0x00000400,
+	0xbf0a7b7c, 0xbf85ffef,
+	0xb97b1e06, 0x877bc17b,
+	0xbf840027, 0x8f7b837b,
+	0x807b7c7b, 0xbefe03c1,
+	0xbeff0380, 0xb97af803,
+	0x8a7a7aff, 0x10000000,
+	0xbf850017, 0x7e008700,
+	0xbe840380, 0xd7600000,
+	0x00000900, 0x80048104,
+	0xd7600001, 0x00000900,
+	0x80048104, 0xd7600002,
+	0x00000900, 0x80048104,
+	0xd7600003, 0x00000900,
+	0x80048104, 0xf469003a,
+	0xe0000000, 0x80709070,
+	0xbf06c004, 0xbf84ffef,
+	0x807c817c, 0xbf0a7b7c,
+	0xbf85ffea, 0xbf820008,
+	0x7e008700, 0xe0704000,
+	0x705d0000, 0x807c817c,
+	0x8070ff70, 0x00000080,
+	0xbf0a7b7c, 0xbf85fff8,
+	0xbf82013f, 0xbef4037e,
+	0x8775ff7f, 0x0000ffff,
+	0x8875ff75, 0x00040000,
+	0xbef60380, 0xbef703ff,
+	0x10807fac, 0xb97202dc,
+	0x8f729972, 0x876eff7f,
+	0x04000000, 0xbf840033,
+	0xbefe03c1, 0x907c9972,
+	0x877c817c, 0xbf06817c,
+	0xbf850002, 0xbeff0380,
+	0xbf820001, 0xbeff03c1,
+	0xb96f4306, 0x876fc16f,
+	0xbf840028, 0x8f6f886f,
+	0xbef6036f, 0xb9783a05,
+	0x80788178, 0xbf0d9972,
+	0xbf850002, 0x8f788978,
+	0xbf820001, 0x8f788a78,
+	0xb96e1e06, 0x8f6e8a6e,
+	0x80786e78, 0x8078ff78,
+	0x00000200, 0x8078ff78,
+	0x00000080, 0xbef603ff,
+	0x01000000, 0x907c9972,
+	0x877c817c, 0xbf06817c,
+	0xbefc0380, 0xbf850009,
+	0xe0310000, 0x781d0000,
+	0x807cff7c, 0x00000080,
+	0x8078ff78, 0x00000080,
+	0xbf0a6f7c, 0xbf85fff8,
+	0xbf820008, 0xe0310000,
+	0x781d0000, 0x807cff7c,
+	0x00000100, 0x8078ff78,
+	0x00000100, 0xbf0a6f7c,
+	0xbf85fff8, 0xbef80380,
+	0xbefe03c1, 0x907c9972,
+	0x877c817c, 0xbf06817c,
+	0xbf850002, 0xbeff0380,
+	0xbf820001, 0xbeff03c1,
+	0xb96f3a05, 0x806f816f,
+	0x8f6f826f, 0x907c9972,
+	0x877c817c, 0xbf06817c,
+	0xbf850024, 0xbef603ff,
+	0x01000000, 0xbeee0378,
+	0x8078ff78, 0x00000200,
+	0xbefc0384, 0xbf0a6f7c,
+	0xbf840050, 0xe0304000,
+	0x785d0000, 0xe0304080,
+	0x785d0100, 0xe0304100,
+	0x785d0200, 0xe0304180,
+	0x785d0300, 0xbf8c0000,
+	0x7e008500, 0x7e028501,
+	0x7e048502, 0x7e068503,
+	0x807c847c, 0x8078ff78,
+	0x00000200, 0xbf0a6f7c,
+	0xbf85ffee, 0xe0304000,
+	0x6e5d0000, 0xe0304080,
+	0x6e5d0100, 0xe0304100,
+	0x6e5d0200, 0xe0304180,
+	0x6e5d0300, 0xbf8c0000,
+	0xbf820034, 0xbef603ff,
+	0x01000000, 0xbeee0378,
+	0x8078ff78, 0x00000400,
+	0xbefc0384, 0xbf0a6f7c,
+	0xbf840012, 0xe0304000,
+	0x785d0000, 0xe0304100,
+	0x785d0100, 0xe0304200,
+	0x785d0200, 0xe0304300,
+	0x785d0300, 0xbf8c0000,
+	0x7e008500, 0x7e028501,
+	0x7e048502, 0x7e068503,
+	0x807c847c, 0x8078ff78,
+	0x00000400, 0xbf0a6f7c,
+	0xbf85ffee, 0xb96f1e06,
+	0x876fc16f, 0xbf84000e,
+	0x8f6f836f, 0x806f7c6f,
+	0xbefe03c1, 0xbeff0380,
+	0xe0304000, 0x785d0000,
+	0xbf8c0000, 0x7e008500,
+	0x807c817c, 0x8078ff78,
+	0x00000080, 0xbf0a6f7c,
+	0xbf85fff7, 0xbeff03c1,
+	0xe0304000, 0x6e5d0000,
+	0xe0304100, 0x6e5d0100,
+	0xe0304200, 0x6e5d0200,
+	0xe0304300, 0x6e5d0300,
+	0xbf8c0000, 0xb9783a05,
+	0x80788178, 0xbf0d9972,
+	0xbf850002, 0x8f788978,
+	0xbf820001, 0x8f788a78,
+	0xb96e1e06, 0x8f6e8a6e,
+	0x80786e78, 0x8078ff78,
+	0x00000200, 0x80f8ff78,
+	0x00000050, 0xbef603ff,
+	0x01000000, 0xbefc03ff,
+	0x0000006c, 0x80f89078,
+	0xf429003a, 0xf0000000,
+	0xbf8c0000, 0x80fc847c,
+	0xbf800000, 0xbe803100,
+	0xbe823102, 0x80f8a078,
+	0xf42d003a, 0xf0000000,
+	0xbf8c0000, 0x80fc887c,
+	0xbf800000, 0xbe803100,
+	0xbe823102, 0xbe843104,
+	0xbe863106, 0x80f8c078,
+	0xf431003a, 0xf0000000,
+	0xbf8c0000, 0x80fc907c,
+	0xbf800000, 0xbe803100,
+	0xbe823102, 0xbe843104,
+	0xbe863106, 0xbe883108,
+	0xbe8a310a, 0xbe8c310c,
+	0xbe8e310e, 0xbf06807c,
+	0xbf84fff0, 0xba80f801,
+	0x00000000, 0xbf8a0000,
+	0xb9783a05, 0x80788178,
+	0xbf0d9972, 0xbf850002,
+	0x8f788978, 0xbf820001,
+	0x8f788a78, 0xb96e1e06,
+	0x8f6e8a6e, 0x80786e78,
+	0x8078ff78, 0x00000200,
+	0xbef603ff, 0x01000000,
+	0xf4211bfa, 0xf0000000,
+	0x80788478, 0xf4211b3a,
+	0xf0000000, 0x80788478,
+	0xf4211b7a, 0xf0000000,
+	0x80788478, 0xf4211c3a,
+	0xf0000000, 0x80788478,
+	0xf4211c7a, 0xf0000000,
+	0x80788478, 0xf4211eba,
+	0xf0000000, 0x80788478,
+	0xf4211efa, 0xf0000000,
+	0x80788478, 0xf4211e7a,
+	0xf0000000, 0x80788478,
+	0xf4211cfa, 0xf0000000,
+	0x80788478, 0xf4211bba,
+	0xf0000000, 0x80788478,
+	0xbf8c0000, 0xb9eef814,
+	0xf4211bba, 0xf0000000,
+	0x80788478, 0xbf8c0000,
+	0xb9eef815, 0xbefc036f,
+	0xbefe0370, 0xbeff0371,
+	0xb9f9f816, 0xb9fb4803,
+	0x907b8b7b, 0xb9fba2c3,
+	0xb9f3f801, 0xb96e3a05,
+	0x806e816e, 0xbf0d9972,
+	0xbf850002, 0x8f6e896e,
+	0xbf820001, 0x8f6e8a6e,
+	0xb96f1e06, 0x8f6f8a6f,
+	0x806e6f6e, 0x806eff6e,
+	0x00000200, 0x806e746e,
+	0x826f8075, 0x876fff6f,
+	0x0000ffff, 0xf4091c37,
+	0xfa000050, 0xf4091d37,
+	0xfa000060, 0xf4011e77,
+	0xfa000074, 0xbf8c0000,
+	0x906e8977, 0x876fff6e,
+	0x003f8000, 0x906e8677,
+	0x876eff6e, 0x02000000,
+	0x886e6f6e, 0xb9eef807,
+	0x876dff6d, 0x0000ffff,
+	0x87fe7e7e, 0x87ea6a6a,
+	0xb9faf802, 0xbe80226c,
+	0xbf9b0000, 0xbf9f0000,
+	0xbf9f0000, 0xbf9f0000,
+	0xbf9f0000, 0xbf9f0000,
+};
+
+static const uint32_t cwsr_trap_arcturus_hex[] = {
+	0xbf820001, 0xbf8202d5,
+	0xb8f8f802, 0x8978ff78,
+	0x00020006, 0xb8fbf803,
+	0x866eff78, 0x00002000,
+	0xbf840009, 0x866eff6d,
+	0x00ff0000, 0xbf85001e,
+	0x866eff7b, 0x00000400,
+	0xbf850055, 0xbf8e0010,
+	0xb8fbf803, 0xbf82fffa,
+	0x866eff7b, 0x03c00900,
+	0xbf850015, 0x866eff7b,
+	0x000071ff, 0xbf840008,
+	0x866fff7b, 0x00007080,
+	0xbf840001, 0xbeee1a87,
+	0xb8eff801, 0x8e6e8c6e,
+	0x866e6f6e, 0xbf85000a,
+	0x866eff6d, 0x00ff0000,
+	0xbf850007, 0xb8eef801,
+	0x866eff6e, 0x00000800,
+	0xbf850003, 0x866eff7b,
+	0x00000400, 0xbf85003a,
+	0xb8faf807, 0x867aff7a,
+	0x001f8000, 0x8e7a8b7a,
+	0x8977ff77, 0xfc000000,
+	0x87777a77, 0xba7ff807,
+	0x00000000, 0xb8faf812,
+	0xb8fbf813, 0x8efa887a,
+	0xbf0d8f7b, 0xbf840002,
+	0x877bff7b, 0xffff0000,
+	0xc0031bbd, 0x00000010,
+	0xbf8cc07f, 0x8e6e976e,
+	0x8977ff77, 0x00800000,
+	0x87776e77, 0xc0071bbd,
+	0x00000000, 0xbf8cc07f,
+	0xc0071ebd, 0x00000008,
+	0xbf8cc07f, 0x86ee6e6e,
+	0xbf840001, 0xbe801d6e,
+	0x866eff6d, 0x01ff0000,
+	0xbf850005, 0x8778ff78,
+	0x00002000, 0x80ec886c,
+	0x82ed806d, 0xbf820005,
+	0x866eff6d, 0x01000000,
+	0xbf850002, 0x806c846c,
+	0x826d806d, 0x866dff6d,
+	0x0000ffff, 0x8f7a8b77,
+	0x867aff7a, 0x001f8000,
+	0xb97af807, 0x86fe7e7e,
+	0x86ea6a6a, 0x8f6e8378,
+	0xb96ee0c2, 0xbf800002,
+	0xb9780002, 0xbe801f6c,
+	0x866dff6d, 0x0000ffff,
+	0xbefa0080, 0xb97a0283,
+	0xb8faf807, 0x867aff7a,
+	0x001f8000, 0x8e7a8b7a,
+	0x8977ff77, 0xfc000000,
+	0x87777a77, 0xba7ff807,
+	0x00000000, 0xbeee007e,
+	0xbeef007f, 0xbefe0180,
+	0xbf900004, 0x877a8478,
+	0xb97af802, 0xbf8e0002,
+	0xbf88fffe, 0xb8fa2a05,
+	0x807a817a, 0x8e7a8a7a,
+	0x8e7a817a, 0xb8fb1605,
+	0x807b817b, 0x8e7b867b,
+	0x807a7b7a, 0x807a7e7a,
+	0x827b807f, 0x867bff7b,
+	0x0000ffff, 0xc04b1c3d,
+	0x00000050, 0xbf8cc07f,
+	0xc04b1d3d, 0x00000060,
+	0xbf8cc07f, 0xc0431e7d,
+	0x00000074, 0xbf8cc07f,
+	0xbef4007e, 0x8675ff7f,
+	0x0000ffff, 0x8775ff75,
+	0x00040000, 0xbef60080,
+	0xbef700ff, 0x00807fac,
+	0xbef1007c, 0xbef00080,
+	0xb8f02a05, 0x80708170,
+	0x8e708a70, 0x8e708170,
+	0xb8fa1605, 0x807a817a,
+	0x8e7a867a, 0x80707a70,
+	0xbef60084, 0xbef600ff,
+	0x01000000, 0xbefe007c,
+	0xbefc0070, 0xc0611c7a,
+	0x0000007c, 0xbf8cc07f,
+	0x80708470, 0xbefc007e,
+	0xbefe007c, 0xbefc0070,
+	0xc0611b3a, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xbefe007c,
+	0xbefc0070, 0xc0611b7a,
+	0x0000007c, 0xbf8cc07f,
+	0x80708470, 0xbefc007e,
+	0xbefe007c, 0xbefc0070,
+	0xc0611bba, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xbefe007c,
+	0xbefc0070, 0xc0611bfa,
+	0x0000007c, 0xbf8cc07f,
+	0x80708470, 0xbefc007e,
+	0xbefe007c, 0xbefc0070,
+	0xc0611e3a, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xb8fbf803,
+	0xbefe007c, 0xbefc0070,
+	0xc0611efa, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xbefe007c,
+	0xbefc0070, 0xc0611a3a,
+	0x0000007c, 0xbf8cc07f,
+	0x80708470, 0xbefc007e,
+	0xbefe007c, 0xbefc0070,
+	0xc0611a7a, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xb8f1f801,
+	0xbefe007c, 0xbefc0070,
+	0xc0611c7a, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xbf108080,
+	0x867aff7f, 0x04000000,
+	0xbeef0080, 0x876f6f7a,
+	0xb8f02a05, 0x80708170,
+	0x8e708a70, 0x8e708170,
+	0xb8fb1605, 0x807b817b,
+	0x8e7b847b, 0x8e76827b,
+	0xbef600ff, 0x01000000,
+	0xbef20174, 0x80747074,
+	0x82758075, 0xbefc0080,
+	0xbf800000, 0xbe802b00,
+	0xbe822b02, 0xbe842b04,
+	0xbe862b06, 0xbe882b08,
+	0xbe8a2b0a, 0xbe8c2b0c,
+	0xbe8e2b0e, 0xc06b003a,
+	0x00000000, 0xbf8cc07f,
+	0xc06b013a, 0x00000010,
+	0xbf8cc07f, 0xc06b023a,
+	0x00000020, 0xbf8cc07f,
+	0xc06b033a, 0x00000030,
+	0xbf8cc07f, 0x8074c074,
+	0x82758075, 0x807c907c,
+	0xbf0a7b7c, 0xbf85ffe7,
+	0xbef40172, 0xbef00080,
+	0xbefe00c1, 0xbeff00c1,
+	0xbee80080, 0xbee90080,
+	0xbef600ff, 0x01000000,
+	0x867aff78, 0x00400000,
+	0xbf850003, 0xb8faf803,
+	0x897a7aff, 0x10000000,
+	0xbf85004d, 0xbe840080,
+	0xd2890000, 0x00000900,
+	0x80048104, 0xd2890001,
+	0x00000900, 0x80048104,
+	0xd2890002, 0x00000900,
+	0x80048104, 0xd2890003,
+	0x00000900, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000901, 0x80048104,
+	0xd2890001, 0x00000901,
+	0x80048104, 0xd2890002,
+	0x00000901, 0x80048104,
+	0xd2890003, 0x00000901,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000902,
+	0x80048104, 0xd2890001,
+	0x00000902, 0x80048104,
+	0xd2890002, 0x00000902,
+	0x80048104, 0xd2890003,
+	0x00000902, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000903, 0x80048104,
+	0xd2890001, 0x00000903,
+	0x80048104, 0xd2890002,
+	0x00000903, 0x80048104,
+	0xd2890003, 0x00000903,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbf820008,
+	0xe0724000, 0x701d0000,
+	0xe0724100, 0x701d0100,
+	0xe0724200, 0x701d0200,
+	0xe0724300, 0x701d0300,
+	0xbefe00c1, 0xbeff00c1,
+	0xb8fb4306, 0x867bc17b,
+	0xbf840064, 0xbf8a0000,
+	0x867aff6f, 0x04000000,
+	0xbf840060, 0x8e7b867b,
+	0x8e7b827b, 0xbef6007b,
+	0xb8f02a05, 0x80708170,
+	0x8e708a70, 0x8e708170,
+	0xb8fa1605, 0x807a817a,
+	0x8e7a867a, 0x80707a70,
+	0x8070ff70, 0x00000080,
+	0xbef600ff, 0x01000000,
+	0xbefc0080, 0xd28c0002,
+	0x000100c1, 0xd28d0003,
+	0x000204c1, 0x867aff78,
+	0x00400000, 0xbf850003,
+	0xb8faf803, 0x897a7aff,
+	0x10000000, 0xbf850030,
+	0x24040682, 0xd86e4000,
+	0x00000002, 0xbf8cc07f,
+	0xbe840080, 0xd2890000,
+	0x00000900, 0x80048104,
+	0xd2890001, 0x00000900,
+	0x80048104, 0xd2890002,
+	0x00000900, 0x80048104,
+	0xd2890003, 0x00000900,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000901,
+	0x80048104, 0xd2890001,
+	0x00000901, 0x80048104,
+	0xd2890002, 0x00000901,
+	0x80048104, 0xd2890003,
+	0x00000901, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0x680404ff, 0x00000200,
+	0xd0c9006a, 0x0000f702,
+	0xbf87ffd2, 0xbf820015,
+	0xd1060002, 0x00011103,
+	0x7e0602ff, 0x00000200,
+	0xbefc00ff, 0x00010000,
+	0xbe800077, 0x8677ff77,
+	0xff7fffff, 0x8777ff77,
+	0x00058000, 0xd8ec0000,
+	0x00000002, 0xbf8cc07f,
+	0xe0765000, 0x701d0002,
+	0x68040702, 0xd0c9006a,
+	0x0000f702, 0xbf87fff7,
+	0xbef70000, 0xbef000ff,
+	0x00000400, 0xbefe00c1,
+	0xbeff00c1, 0xb8fb2a05,
+	0x807b817b, 0x8e7b827b,
+	0xbef600ff, 0x01000000,
+	0xbefc0084, 0xbf0a7b7c,
+	0xbf84006d, 0xbf11017c,
+	0x807bff7b, 0x00001000,
+	0x867aff78, 0x00400000,
+	0xbf850003, 0xb8faf803,
+	0x897a7aff, 0x10000000,
+	0xbf850051, 0xbe840080,
+	0xd2890000, 0x00000900,
+	0x80048104, 0xd2890001,
+	0x00000900, 0x80048104,
+	0xd2890002, 0x00000900,
+	0x80048104, 0xd2890003,
+	0x00000900, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000901, 0x80048104,
+	0xd2890001, 0x00000901,
+	0x80048104, 0xd2890002,
+	0x00000901, 0x80048104,
+	0xd2890003, 0x00000901,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000902,
+	0x80048104, 0xd2890001,
+	0x00000902, 0x80048104,
+	0xd2890002, 0x00000902,
+	0x80048104, 0xd2890003,
+	0x00000902, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000903, 0x80048104,
+	0xd2890001, 0x00000903,
+	0x80048104, 0xd2890002,
+	0x00000903, 0x80048104,
+	0xd2890003, 0x00000903,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0x807c847c,
+	0xbf0a7b7c, 0xbf85ffb1,
+	0xbf9c0000, 0xbf820012,
+	0x7e000300, 0x7e020301,
+	0x7e040302, 0x7e060303,
+	0xe0724000, 0x701d0000,
+	0xe0724100, 0x701d0100,
+	0xe0724200, 0x701d0200,
+	0xe0724300, 0x701d0300,
+	0x807c847c, 0x8070ff70,
+	0x00000400, 0xbf0a7b7c,
+	0xbf85ffef, 0xbf9c0000,
+	0xbefc0080, 0xbf11017c,
+	0x867aff78, 0x00400000,
+	0xbf850003, 0xb8faf803,
+	0x897a7aff, 0x10000000,
+	0xbf850059, 0xd3d84000,
+	0x18000100, 0xd3d84001,
+	0x18000101, 0xd3d84002,
+	0x18000102, 0xd3d84003,
+	0x18000103, 0xbe840080,
+	0xd2890000, 0x00000900,
+	0x80048104, 0xd2890001,
+	0x00000900, 0x80048104,
+	0xd2890002, 0x00000900,
+	0x80048104, 0xd2890003,
+	0x00000900, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000901, 0x80048104,
+	0xd2890001, 0x00000901,
+	0x80048104, 0xd2890002,
+	0x00000901, 0x80048104,
+	0xd2890003, 0x00000901,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000902,
+	0x80048104, 0xd2890001,
+	0x00000902, 0x80048104,
+	0xd2890002, 0x00000902,
+	0x80048104, 0xd2890003,
+	0x00000902, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000903, 0x80048104,
+	0xd2890001, 0x00000903,
+	0x80048104, 0xd2890002,
+	0x00000903, 0x80048104,
+	0xd2890003, 0x00000903,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0x807c847c,
+	0xbf0a7b7c, 0xbf85ffa9,
+	0xbf9c0000, 0xbf820016,
+	0xd3d84000, 0x18000100,
+	0xd3d84001, 0x18000101,
+	0xd3d84002, 0x18000102,
+	0xd3d84003, 0x18000103,
+	0xe0724000, 0x701d0000,
+	0xe0724100, 0x701d0100,
+	0xe0724200, 0x701d0200,
+	0xe0724300, 0x701d0300,
+	0x807c847c, 0x8070ff70,
+	0x00000400, 0xbf0a7b7c,
+	0xbf85ffeb, 0xbf9c0000,
+	0xbf8200e3, 0xbef4007e,
+	0x8675ff7f, 0x0000ffff,
+	0x8775ff75, 0x00040000,
+	0xbef60080, 0xbef700ff,
+	0x00807fac, 0x866eff7f,
+	0x04000000, 0xbf84001f,
+	0xbefe00c1, 0xbeff00c1,
+	0xb8ef4306, 0x866fc16f,
+	0xbf84001a, 0x8e6f866f,
+	0x8e6f826f, 0xbef6006f,
+	0xb8f82a05, 0x80788178,
+	0x8e788a78, 0x8e788178,
+	0xb8ee1605, 0x806e816e,
+	0x8e6e866e, 0x80786e78,
+	0x8078ff78, 0x00000080,
+	0xbef600ff, 0x01000000,
+	0xbefc0080, 0xe0510000,
+	0x781d0000, 0xe0510100,
+	0x781d0000, 0x807cff7c,
+	0x00000200, 0x8078ff78,
+	0x00000200, 0xbf0a6f7c,
+	0xbf85fff6, 0xbefe00c1,
+	0xbeff00c1, 0xbef600ff,
+	0x01000000, 0xb8ef2a05,
+	0x806f816f, 0x8e6f826f,
+	0x806fff6f, 0x00008000,
+	0xbef80080, 0xbeee0078,
+	0x8078ff78, 0x00000400,
+	0xbefc0084, 0xbf11087c,
+	0xe0524000, 0x781d0000,
+	0xe0524100, 0x781d0100,
+	0xe0524200, 0x781d0200,
+	0xe0524300, 0x781d0300,
+	0xbf8c0f70, 0x7e000300,
+	0x7e020301, 0x7e040302,
+	0x7e060303, 0x807c847c,
+	0x8078ff78, 0x00000400,
+	0xbf0a6f7c, 0xbf85ffee,
+	0xbefc0080, 0xbf11087c,
+	0xe0524000, 0x781d0000,
+	0xe0524100, 0x781d0100,
+	0xe0524200, 0x781d0200,
+	0xe0524300, 0x781d0300,
+	0xbf8c0f70, 0xd3d94000,
+	0x18000100, 0xd3d94001,
+	0x18000101, 0xd3d94002,
+	0x18000102, 0xd3d94003,
+	0x18000103, 0x807c847c,
+	0x8078ff78, 0x00000400,
+	0xbf0a6f7c, 0xbf85ffea,
+	0xbf9c0000, 0xe0524000,
+	0x6e1d0000, 0xe0524100,
+	0x6e1d0100, 0xe0524200,
+	0x6e1d0200, 0xe0524300,
+	0x6e1d0300, 0xbf8c0f70,
+	0xb8f82a05, 0x80788178,
+	0x8e788a78, 0x8e788178,
+	0xb8ee1605, 0x806e816e,
+	0x8e6e866e, 0x80786e78,
+	0x80f8c078, 0xb8ef1605,
+	0x806f816f, 0x8e6f846f,
+	0x8e76826f, 0xbef600ff,
+	0x01000000, 0xbefc006f,
+	0xc031003a, 0x00000078,
+	0x80f8c078, 0xbf8cc07f,
+	0x80fc907c, 0xbf800000,
+	0xbe802d00, 0xbe822d02,
+	0xbe842d04, 0xbe862d06,
+	0xbe882d08, 0xbe8a2d0a,
+	0xbe8c2d0c, 0xbe8e2d0e,
+	0xbf06807c, 0xbf84fff0,
+	0xb8f82a05, 0x80788178,
+	0x8e788a78, 0x8e788178,
+	0xb8ee1605, 0x806e816e,
+	0x8e6e866e, 0x80786e78,
+	0xbef60084, 0xbef600ff,
+	0x01000000, 0xc0211bfa,
+	0x00000078, 0x80788478,
+	0xc0211b3a, 0x00000078,
+	0x80788478, 0xc0211b7a,
+	0x00000078, 0x80788478,
+	0xc0211c3a, 0x00000078,
+	0x80788478, 0xc0211c7a,
+	0x00000078, 0x80788478,
+	0xc0211eba, 0x00000078,
+	0x80788478, 0xc0211efa,
+	0x00000078, 0x80788478,
+	0xc0211a3a, 0x00000078,
+	0x80788478, 0xc0211a7a,
+	0x00000078, 0x80788478,
+	0xc0211cfa, 0x00000078,
+	0x80788478, 0xbf8cc07f,
+	0xbefc006f, 0xbefe0070,
+	0xbeff0071, 0x866f7bff,
+	0x000003ff, 0xb96f4803,
+	0x866f7bff, 0xfffff800,
+	0x8f6f8b6f, 0xb96fa2c3,
+	0xb973f801, 0xb8ee2a05,
+	0x806e816e, 0x8e6e8a6e,
+	0x8e6e816e, 0xb8ef1605,
+	0x806f816f, 0x8e6f866f,
+	0x806e6f6e, 0x806e746e,
+	0x826f8075, 0x866fff6f,
+	0x0000ffff, 0xc00b1c37,
+	0x00000050, 0xc00b1d37,
+	0x00000060, 0xc0031e77,
+	0x00000074, 0xbf8cc07f,
+	0x8f6e8b77, 0x866eff6e,
+	0x001f8000, 0xb96ef807,
+	0x866dff6d, 0x0000ffff,
+	0x86fe7e7e, 0x86ea6a6a,
+	0x8f6e837a, 0xb96ee0c2,
+	0xbf800002, 0xb97a0002,
+	0xbf8a0000, 0xbe801f6c,
+	0xbf9b0000, 0x00000000,
+};
+
+static const uint32_t cwsr_trap_aldebaran_hex[] = {
+	0xbf820001, 0xbf8202e0,
+	0xb8f8f802, 0x8978ff78,
+	0x00020006, 0xb8fbf803,
+	0x866eff78, 0x00002000,
+	0xbf840009, 0x866eff6d,
+	0x00ff0000, 0xbf85001e,
+	0x866eff7b, 0x00000400,
+	0xbf850055, 0xbf8e0010,
+	0xb8fbf803, 0xbf82fffa,
+	0x866eff7b, 0x03c00900,
+	0xbf850015, 0x866eff7b,
+	0x000071ff, 0xbf840008,
+	0x866fff7b, 0x00007080,
+	0xbf840001, 0xbeee1a87,
+	0xb8eff801, 0x8e6e8c6e,
+	0x866e6f6e, 0xbf85000a,
+	0x866eff6d, 0x00ff0000,
+	0xbf850007, 0xb8eef801,
+	0x866eff6e, 0x00000800,
+	0xbf850003, 0x866eff7b,
+	0x00000400, 0xbf85003a,
+	0xb8faf807, 0x867aff7a,
+	0x001f8000, 0x8e7a8b7a,
+	0x8977ff77, 0xfc000000,
+	0x87777a77, 0xba7ff807,
+	0x00000000, 0xb8faf812,
+	0xb8fbf813, 0x8efa887a,
+	0xbf0d8f7b, 0xbf840002,
+	0x877bff7b, 0xffff0000,
+	0xc0031bbd, 0x00000010,
+	0xbf8cc07f, 0x8e6e976e,
+	0x8977ff77, 0x00800000,
+	0x87776e77, 0xc0071bbd,
+	0x00000000, 0xbf8cc07f,
+	0xc0071ebd, 0x00000008,
+	0xbf8cc07f, 0x86ee6e6e,
+	0xbf840001, 0xbe801d6e,
+	0x866eff6d, 0x01ff0000,
+	0xbf850005, 0x8778ff78,
+	0x00002000, 0x80ec886c,
+	0x82ed806d, 0xbf820005,
+	0x866eff6d, 0x01000000,
+	0xbf850002, 0x806c846c,
+	0x826d806d, 0x866dff6d,
+	0x0000ffff, 0x8f7a8b77,
+	0x867aff7a, 0x001f8000,
+	0xb97af807, 0x86fe7e7e,
+	0x86ea6a6a, 0x8f6e8378,
+	0xb96ee0c2, 0xbf800002,
+	0xb9780002, 0xbe801f6c,
+	0x866dff6d, 0x0000ffff,
+	0xbefa0080, 0xb97a0283,
+	0xb8faf807, 0x867aff7a,
+	0x001f8000, 0x8e7a8b7a,
+	0x8977ff77, 0xfc000000,
+	0x87777a77, 0xba7ff807,
+	0x00000000, 0xbeee007e,
+	0xbeef007f, 0xbefe0180,
+	0xbf900004, 0x877a8478,
+	0xb97af802, 0xbf8e0002,
+	0xbf88fffe, 0xb8fa2985,
+	0x807a817a, 0x8e7a8a7a,
+	0x8e7a817a, 0xb8fb1605,
+	0x807b817b, 0x8e7b867b,
+	0x807a7b7a, 0x807a7e7a,
+	0x827b807f, 0x867bff7b,
+	0x0000ffff, 0xc04b1c3d,
+	0x00000050, 0xbf8cc07f,
+	0xc04b1d3d, 0x00000060,
+	0xbf8cc07f, 0xc0431e7d,
+	0x00000074, 0xbf8cc07f,
+	0xbef4007e, 0x8675ff7f,
+	0x0000ffff, 0x8775ff75,
+	0x00040000, 0xbef60080,
+	0xbef700ff, 0x00807fac,
+	0xbef1007c, 0xbef00080,
+	0xb8f02985, 0x80708170,
+	0x8e708a70, 0x8e708170,
+	0xb8fa1605, 0x807a817a,
+	0x8e7a867a, 0x80707a70,
+	0xbef60084, 0xbef600ff,
+	0x01000000, 0xbefe007c,
+	0xbefc0070, 0xc0611c7a,
+	0x0000007c, 0xbf8cc07f,
+	0x80708470, 0xbefc007e,
+	0xbefe007c, 0xbefc0070,
+	0xc0611b3a, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xbefe007c,
+	0xbefc0070, 0xc0611b7a,
+	0x0000007c, 0xbf8cc07f,
+	0x80708470, 0xbefc007e,
+	0xbefe007c, 0xbefc0070,
+	0xc0611bba, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xbefe007c,
+	0xbefc0070, 0xc0611bfa,
+	0x0000007c, 0xbf8cc07f,
+	0x80708470, 0xbefc007e,
+	0xbefe007c, 0xbefc0070,
+	0xc0611e3a, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xb8fbf803,
+	0xbefe007c, 0xbefc0070,
+	0xc0611efa, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xbefe007c,
+	0xbefc0070, 0xc0611a3a,
+	0x0000007c, 0xbf8cc07f,
+	0x80708470, 0xbefc007e,
+	0xbefe007c, 0xbefc0070,
+	0xc0611a7a, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xb8f1f801,
+	0xbefe007c, 0xbefc0070,
+	0xc0611c7a, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xbf108080,
+	0x867aff7f, 0x04000000,
+	0xbeef0080, 0x876f6f7a,
+	0xb8f02985, 0x80708170,
+	0x8e708a70, 0x8e708170,
+	0xb8fb1605, 0x807b817b,
+	0x8e7b847b, 0x8e76827b,
+	0xbef600ff, 0x01000000,
+	0xbef20174, 0x80747074,
+	0x82758075, 0xbefc0080,
+	0xbf800000, 0xbe802b00,
+	0xbe822b02, 0xbe842b04,
+	0xbe862b06, 0xbe882b08,
+	0xbe8a2b0a, 0xbe8c2b0c,
+	0xbe8e2b0e, 0xc06b003a,
+	0x00000000, 0xbf8cc07f,
+	0xc06b013a, 0x00000010,
+	0xbf8cc07f, 0xc06b023a,
+	0x00000020, 0xbf8cc07f,
+	0xc06b033a, 0x00000030,
+	0xbf8cc07f, 0x8074c074,
+	0x82758075, 0x807c907c,
+	0xbf0a7b7c, 0xbf85ffe7,
+	0xbef40172, 0xbef00080,
+	0xbefe00c1, 0xbeff00c1,
+	0xbee80080, 0xbee90080,
+	0xbef600ff, 0x01000000,
+	0x867aff78, 0x00400000,
+	0xbf850003, 0xb8faf803,
+	0x897a7aff, 0x10000000,
+	0xbf85004d, 0xbe840080,
+	0xd2890000, 0x00000900,
+	0x80048104, 0xd2890001,
+	0x00000900, 0x80048104,
+	0xd2890002, 0x00000900,
+	0x80048104, 0xd2890003,
+	0x00000900, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000901, 0x80048104,
+	0xd2890001, 0x00000901,
+	0x80048104, 0xd2890002,
+	0x00000901, 0x80048104,
+	0xd2890003, 0x00000901,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000902,
+	0x80048104, 0xd2890001,
+	0x00000902, 0x80048104,
+	0xd2890002, 0x00000902,
+	0x80048104, 0xd2890003,
+	0x00000902, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000903, 0x80048104,
+	0xd2890001, 0x00000903,
+	0x80048104, 0xd2890002,
+	0x00000903, 0x80048104,
+	0xd2890003, 0x00000903,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbf820008,
+	0xe0724000, 0x701d0000,
+	0xe0724100, 0x701d0100,
+	0xe0724200, 0x701d0200,
+	0xe0724300, 0x701d0300,
+	0xbefe00c1, 0xbeff00c1,
+	0xb8fb4306, 0x867bc17b,
+	0xbf840064, 0xbf8a0000,
+	0x867aff6f, 0x04000000,
+	0xbf840060, 0x8e7b867b,
+	0x8e7b827b, 0xbef6007b,
+	0xb8f02985, 0x80708170,
+	0x8e708a70, 0x8e708170,
+	0xb8fa1605, 0x807a817a,
+	0x8e7a867a, 0x80707a70,
+	0x8070ff70, 0x00000080,
+	0xbef600ff, 0x01000000,
+	0xbefc0080, 0xd28c0002,
+	0x000100c1, 0xd28d0003,
+	0x000204c1, 0x867aff78,
+	0x00400000, 0xbf850003,
+	0xb8faf803, 0x897a7aff,
+	0x10000000, 0xbf850030,
+	0x24040682, 0xd86e4000,
+	0x00000002, 0xbf8cc07f,
+	0xbe840080, 0xd2890000,
+	0x00000900, 0x80048104,
+	0xd2890001, 0x00000900,
+	0x80048104, 0xd2890002,
+	0x00000900, 0x80048104,
+	0xd2890003, 0x00000900,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000901,
+	0x80048104, 0xd2890001,
+	0x00000901, 0x80048104,
+	0xd2890002, 0x00000901,
+	0x80048104, 0xd2890003,
+	0x00000901, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0x680404ff, 0x00000200,
+	0xd0c9006a, 0x0000f702,
+	0xbf87ffd2, 0xbf820015,
+	0xd1060002, 0x00011103,
+	0x7e0602ff, 0x00000200,
+	0xbefc00ff, 0x00010000,
+	0xbe800077, 0x8677ff77,
+	0xff7fffff, 0x8777ff77,
+	0x00058000, 0xd8ec0000,
+	0x00000002, 0xbf8cc07f,
+	0xe0765000, 0x701d0002,
+	0x68040702, 0xd0c9006a,
+	0x0000f702, 0xbf87fff7,
+	0xbef70000, 0xbef000ff,
+	0x00000400, 0xbefe00c1,
+	0xbeff00c1, 0xb8fb2b05,
+	0x807b817b, 0x8e7b827b,
+	0xbef600ff, 0x01000000,
+	0xbefc0084, 0xbf0a7b7c,
+	0xbf84006d, 0xbf11017c,
+	0x807bff7b, 0x00001000,
+	0x867aff78, 0x00400000,
+	0xbf850003, 0xb8faf803,
+	0x897a7aff, 0x10000000,
+	0xbf850051, 0xbe840080,
+	0xd2890000, 0x00000900,
+	0x80048104, 0xd2890001,
+	0x00000900, 0x80048104,
+	0xd2890002, 0x00000900,
+	0x80048104, 0xd2890003,
+	0x00000900, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000901, 0x80048104,
+	0xd2890001, 0x00000901,
+	0x80048104, 0xd2890002,
+	0x00000901, 0x80048104,
+	0xd2890003, 0x00000901,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000902,
+	0x80048104, 0xd2890001,
+	0x00000902, 0x80048104,
+	0xd2890002, 0x00000902,
+	0x80048104, 0xd2890003,
+	0x00000902, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000903, 0x80048104,
+	0xd2890001, 0x00000903,
+	0x80048104, 0xd2890002,
+	0x00000903, 0x80048104,
+	0xd2890003, 0x00000903,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0x807c847c,
+	0xbf0a7b7c, 0xbf85ffb1,
+	0xbf9c0000, 0xbf820012,
+	0x7e000300, 0x7e020301,
+	0x7e040302, 0x7e060303,
+	0xe0724000, 0x701d0000,
+	0xe0724100, 0x701d0100,
+	0xe0724200, 0x701d0200,
+	0xe0724300, 0x701d0300,
+	0x807c847c, 0x8070ff70,
+	0x00000400, 0xbf0a7b7c,
+	0xbf85ffef, 0xbf9c0000,
+	0xb8fb2985, 0x807b817b,
+	0x8e7b837b, 0xb8fa2b05,
+	0x807a817a, 0x8e7a827a,
+	0x80fb7a7b, 0x867b7b7b,
+	0xbf84007a, 0x807bff7b,
+	0x00001000, 0xbefc0080,
+	0xbf11017c, 0x867aff78,
+	0x00400000, 0xbf850003,
+	0xb8faf803, 0x897a7aff,
+	0x10000000, 0xbf850059,
+	0xd3d84000, 0x18000100,
+	0xd3d84001, 0x18000101,
+	0xd3d84002, 0x18000102,
+	0xd3d84003, 0x18000103,
+	0xbe840080, 0xd2890000,
+	0x00000900, 0x80048104,
+	0xd2890001, 0x00000900,
+	0x80048104, 0xd2890002,
+	0x00000900, 0x80048104,
+	0xd2890003, 0x00000900,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000901,
+	0x80048104, 0xd2890001,
+	0x00000901, 0x80048104,
+	0xd2890002, 0x00000901,
+	0x80048104, 0xd2890003,
+	0x00000901, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000902, 0x80048104,
+	0xd2890001, 0x00000902,
+	0x80048104, 0xd2890002,
+	0x00000902, 0x80048104,
+	0xd2890003, 0x00000902,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000903,
+	0x80048104, 0xd2890001,
+	0x00000903, 0x80048104,
+	0xd2890002, 0x00000903,
+	0x80048104, 0xd2890003,
+	0x00000903, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0x807c847c, 0xbf0a7b7c,
+	0xbf85ffa9, 0xbf9c0000,
+	0xbf820016, 0xd3d84000,
+	0x18000100, 0xd3d84001,
+	0x18000101, 0xd3d84002,
+	0x18000102, 0xd3d84003,
+	0x18000103, 0xe0724000,
+	0x701d0000, 0xe0724100,
+	0x701d0100, 0xe0724200,
+	0x701d0200, 0xe0724300,
+	0x701d0300, 0x807c847c,
+	0x8070ff70, 0x00000400,
+	0xbf0a7b7c, 0xbf85ffeb,
+	0xbf9c0000, 0xbf8200ee,
+	0xbef4007e, 0x8675ff7f,
+	0x0000ffff, 0x8775ff75,
+	0x00040000, 0xbef60080,
+	0xbef700ff, 0x00807fac,
+	0x866eff7f, 0x04000000,
+	0xbf84001f, 0xbefe00c1,
+	0xbeff00c1, 0xb8ef4306,
+	0x866fc16f, 0xbf84001a,
+	0x8e6f866f, 0x8e6f826f,
+	0xbef6006f, 0xb8f82985,
+	0x80788178, 0x8e788a78,
+	0x8e788178, 0xb8ee1605,
+	0x806e816e, 0x8e6e866e,
+	0x80786e78, 0x8078ff78,
+	0x00000080, 0xbef600ff,
+	0x01000000, 0xbefc0080,
+	0xe0510000, 0x781d0000,
+	0xe0510100, 0x781d0000,
+	0x807cff7c, 0x00000200,
+	0x8078ff78, 0x00000200,
+	0xbf0a6f7c, 0xbf85fff6,
+	0xbefe00c1, 0xbeff00c1,
+	0xbef600ff, 0x01000000,
+	0xb8ef2b05, 0x806f816f,
+	0x8e6f826f, 0x806fff6f,
+	0x00008000, 0xbef80080,
+	0xbeee0078, 0x8078ff78,
+	0x00000400, 0xbefc0084,
+	0xbf11087c, 0xe0524000,
+	0x781d0000, 0xe0524100,
+	0x781d0100, 0xe0524200,
+	0x781d0200, 0xe0524300,
+	0x781d0300, 0xbf8c0f70,
+	0x7e000300, 0x7e020301,
+	0x7e040302, 0x7e060303,
+	0x807c847c, 0x8078ff78,
+	0x00000400, 0xbf0a6f7c,
+	0xbf85ffee, 0xb8ef2985,
+	0x806f816f, 0x8e6f836f,
+	0xb8f92b05, 0x80798179,
+	0x8e798279, 0x80ef796f,
+	0x866f6f6f, 0xbf84001a,
+	0x806fff6f, 0x00008000,
+	0xbefc0080, 0xbf11087c,
+	0xe0524000, 0x781d0000,
+	0xe0524100, 0x781d0100,
+	0xe0524200, 0x781d0200,
+	0xe0524300, 0x781d0300,
+	0xbf8c0f70, 0xd3d94000,
+	0x18000100, 0xd3d94001,
+	0x18000101, 0xd3d94002,
+	0x18000102, 0xd3d94003,
+	0x18000103, 0x807c847c,
+	0x8078ff78, 0x00000400,
+	0xbf0a6f7c, 0xbf85ffea,
+	0xbf9c0000, 0xe0524000,
+	0x6e1d0000, 0xe0524100,
+	0x6e1d0100, 0xe0524200,
+	0x6e1d0200, 0xe0524300,
+	0x6e1d0300, 0xbf8c0f70,
+	0xb8f82985, 0x80788178,
+	0x8e788a78, 0x8e788178,
+	0xb8ee1605, 0x806e816e,
+	0x8e6e866e, 0x80786e78,
+	0x80f8c078, 0xb8ef1605,
+	0x806f816f, 0x8e6f846f,
+	0x8e76826f, 0xbef600ff,
+	0x01000000, 0xbefc006f,
+	0xc031003a, 0x00000078,
+	0x80f8c078, 0xbf8cc07f,
+	0x80fc907c, 0xbf800000,
+	0xbe802d00, 0xbe822d02,
+	0xbe842d04, 0xbe862d06,
+	0xbe882d08, 0xbe8a2d0a,
+	0xbe8c2d0c, 0xbe8e2d0e,
+	0xbf06807c, 0xbf84fff0,
+	0xb8f82985, 0x80788178,
+	0x8e788a78, 0x8e788178,
+	0xb8ee1605, 0x806e816e,
+	0x8e6e866e, 0x80786e78,
+	0xbef60084, 0xbef600ff,
+	0x01000000, 0xc0211bfa,
+	0x00000078, 0x80788478,
+	0xc0211b3a, 0x00000078,
+	0x80788478, 0xc0211b7a,
+	0x00000078, 0x80788478,
+	0xc0211c3a, 0x00000078,
+	0x80788478, 0xc0211c7a,
+	0x00000078, 0x80788478,
+	0xc0211eba, 0x00000078,
+	0x80788478, 0xc0211efa,
+	0x00000078, 0x80788478,
+	0xc0211a3a, 0x00000078,
+	0x80788478, 0xc0211a7a,
+	0x00000078, 0x80788478,
+	0xc0211cfa, 0x00000078,
+	0x80788478, 0xbf8cc07f,
+	0xbefc006f, 0xbefe0070,
+	0xbeff0071, 0x866f7bff,
+	0x000003ff, 0xb96f4803,
+	0x866f7bff, 0xfffff800,
+	0x8f6f8b6f, 0xb96fa2c3,
+	0xb973f801, 0xb8ee2985,
+	0x806e816e, 0x8e6e8a6e,
+	0x8e6e816e, 0xb8ef1605,
+	0x806f816f, 0x8e6f866f,
+	0x806e6f6e, 0x806e746e,
+	0x826f8075, 0x866fff6f,
+	0x0000ffff, 0xc00b1c37,
+	0x00000050, 0xc00b1d37,
+	0x00000060, 0xc0031e77,
+	0x00000074, 0xbf8cc07f,
+	0x8f6e8b77, 0x866eff6e,
+	0x001f8000, 0xb96ef807,
+	0x866dff6d, 0x0000ffff,
+	0x86fe7e7e, 0x86ea6a6a,
+	0x8f6e837a, 0xb96ee0c2,
+	0xbf800002, 0xb97a0002,
+	0xbf8a0000, 0xbe801f6c,
+	0xbf9b0000, 0x00000000,
+};
+
+static const uint32_t cwsr_trap_gfx10_hex[] = {
+	0xbf820001, 0xbf820220,
+	0xb0804004, 0xb978f802,
+	0x8a78ff78, 0x00020006,
+	0xb97bf803, 0x876eff78,
+	0x00002000, 0xbf840009,
+	0x876eff6d, 0x00ff0000,
+	0xbf85001e, 0x876eff7b,
+	0x00000400, 0xbf850045,
+	0xbf8e0010, 0xb97bf803,
+	0xbf82fffa, 0x876eff7b,
+	0x00000900, 0xbf850015,
+	0x876eff7b, 0x000071ff,
+	0xbf840008, 0x876fff7b,
+	0x00007080, 0xbf840001,
+	0xbeee1d87, 0xb96ff801,
+	0x8f6e8c6e, 0x876e6f6e,
+	0xbf85000a, 0x876eff6d,
+	0x00ff0000, 0xbf850007,
+	0xb96ef801, 0x876eff6e,
+	0x00000800, 0xbf850003,
+	0x876eff7b, 0x00000400,
+	0xbf85002a, 0xb97af812,
+	0xb97bf813, 0x8ffa887a,
+	0xbf0d8f7b, 0xbf840002,
+	0x887bff7b, 0xffff0000,
+	0xf4011bbd, 0xfa000010,
+	0xbf8c0000, 0x8f6e976e,
+	0x8a77ff77, 0x00800000,
+	0x88776e77, 0xf4051bbd,
+	0xfa000000, 0xbf8c0000,
+	0xf4051ebd, 0xfa000008,
+	0xbf8c0000, 0x87ee6e6e,
+	0xbf840001, 0xbe80206e,
+	0x876eff6d, 0x00ff0000,
+	0xbf850008, 0x876eff6d,
+	0x01000000, 0xbf850007,
+	0x8878ff78, 0x00002000,
+	0x80ec886c, 0x82ed806d,
+	0xbf820002, 0x806c846c,
+	0x826d806d, 0x876dff6d,
+	0x0000ffff, 0x87fe7e7e,
+	0x87ea6a6a, 0xb9f8f802,
+	0xbe80226c, 0x876dff6d,
+	0x0000ffff, 0xbefa0380,
+	0xb9fa0283, 0xbeee037e,
+	0xbeef037f, 0xbefe0480,
+	0xbf900004, 0xbf8c0000,
+	0x877aff7f, 0x04000000,
+	0x8f7a857a, 0x886d7a6d,
+	0x7e008200, 0xbefa037e,
+	0x877bff7f, 0x0000ffff,
+	0xbefe03c1, 0xbeff03c1,
+	0xdc5f8000, 0x007a0000,
+	0x7e000280, 0xbefe037a,
+	0xbeff037b, 0xb97b02dc,
+	0x8f7b997b, 0xb97a3a05,
+	0x807a817a, 0xbf0d997b,
+	0xbf850002, 0x8f7a897a,
+	0xbf820001, 0x8f7a8a7a,
+	0xb97b1e06, 0x8f7b8a7b,
+	0x807a7b7a, 0x877bff7f,
+	0x0000ffff, 0x807aff7a,
+	0x00000200, 0x807a7e7a,
+	0x827b807b, 0xd7610000,
+	0x00010870, 0xd7610000,
+	0x00010a71, 0xd7610000,
+	0x00010c72, 0xd7610000,
+	0x00010e73, 0xd7610000,
+	0x00011074, 0xd7610000,
+	0x00011275, 0xd7610000,
+	0x00011476, 0xd7610000,
+	0x00011677, 0xd7610000,
+	0x00011a79, 0xd7610000,
+	0x00011c7e, 0xd7610000,
+	0x00011e7f, 0xbefe03ff,
+	0x00003fff, 0xbeff0380,
+	0xdc5f8040, 0x007a0000,
+	0xd760007a, 0x00011d00,
+	0xd760007b, 0x00011f00,
+	0xbefe037a, 0xbeff037b,
+	0xbef4037e, 0x8775ff7f,
+	0x0000ffff, 0x8875ff75,
+	0x00040000, 0xbef60380,
+	0xbef703ff, 0x10807fac,
+	0xbef1037c, 0xbef00380,
+	0xb97302dc, 0x8f739973,
+	0xbefe03c1, 0x907c9973,
+	0x877c817c, 0xbf06817c,
+	0xbf850002, 0xbeff0380,
+	0xbf820002, 0xbeff03c1,
+	0xbf820009, 0xbef603ff,
+	0x01000000, 0xe0704080,
+	0x705d0100, 0xe0704100,
+	0x705d0200, 0xe0704180,
+	0x705d0300, 0xbf820008,
+	0xbef603ff, 0x01000000,
+	0xe0704100, 0x705d0100,
+	0xe0704200, 0x705d0200,
+	0xe0704300, 0x705d0300,
+	0xb9703a05, 0x80708170,
+	0xbf0d9973, 0xbf850002,
+	0x8f708970, 0xbf820001,
+	0x8f708a70, 0xb97a1e06,
+	0x8f7a8a7a, 0x80707a70,
+	0x8070ff70, 0x00000200,
+	0xbef603ff, 0x01000000,
+	0x7e000280, 0x7e020280,
+	0x7e040280, 0xbefc0380,
+	0xd7610002, 0x0000f871,
+	0x807c817c, 0xd7610002,
+	0x0000f86c, 0x807c817c,
+	0x8a7aff6d, 0x80000000,
+	0xd7610002, 0x0000f87a,
+	0x807c817c, 0xd7610002,
+	0x0000f86e, 0x807c817c,
+	0xd7610002, 0x0000f86f,
+	0x807c817c, 0xd7610002,
+	0x0000f878, 0x807c817c,
+	0xb97af803, 0xd7610002,
+	0x0000f87a, 0x807c817c,
+	0xd7610002, 0x0000f87b,
+	0x807c817c, 0xb971f801,
+	0xd7610002, 0x0000f871,
+	0x807c817c, 0xb971f814,
+	0xd7610002, 0x0000f871,
+	0x807c817c, 0xb971f815,
+	0xd7610002, 0x0000f871,
+	0x807c817c, 0xbefe03ff,
+	0x0000ffff, 0xbeff0380,
+	0xe0704000, 0x705d0200,
+	0xbefe03c1, 0xb9703a05,
+	0x80708170, 0xbf0d9973,
+	0xbf850002, 0x8f708970,
+	0xbf820001, 0x8f708a70,
+	0xb97a1e06, 0x8f7a8a7a,
+	0x80707a70, 0xbef603ff,
+	0x01000000, 0xbef90380,
+	0xbefc0380, 0xbf800000,
+	0xbe802f00, 0xbe822f02,
+	0xbe842f04, 0xbe862f06,
+	0xbe882f08, 0xbe8a2f0a,
+	0xbe8c2f0c, 0xbe8e2f0e,
+	0xd7610002, 0x0000f200,
+	0x80798179, 0xd7610002,
+	0x0000f201, 0x80798179,
+	0xd7610002, 0x0000f202,
+	0x80798179, 0xd7610002,
+	0x0000f203, 0x80798179,
+	0xd7610002, 0x0000f204,
+	0x80798179, 0xd7610002,
+	0x0000f205, 0x80798179,
+	0xd7610002, 0x0000f206,
+	0x80798179, 0xd7610002,
+	0x0000f207, 0x80798179,
+	0xd7610002, 0x0000f208,
+	0x80798179, 0xd7610002,
+	0x0000f209, 0x80798179,
+	0xd7610002, 0x0000f20a,
+	0x80798179, 0xd7610002,
+	0x0000f20b, 0x80798179,
+	0xd7610002, 0x0000f20c,
+	0x80798179, 0xd7610002,
+	0x0000f20d, 0x80798179,
+	0xd7610002, 0x0000f20e,
+	0x80798179, 0xd7610002,
+	0x0000f20f, 0x80798179,
+	0xbf06a079, 0xbf840006,
+	0xe0704000, 0x705d0200,
+	0x8070ff70, 0x00000080,
+	0xbef90380, 0x7e040280,
+	0x807c907c, 0xbf0aff7c,
+	0x00000060, 0xbf85ffbc,
+	0xbe802f00, 0xbe822f02,
+	0xbe842f04, 0xbe862f06,
+	0xbe882f08, 0xbe8a2f0a,
+	0xd7610002, 0x0000f200,
+	0x80798179, 0xd7610002,
+	0x0000f201, 0x80798179,
+	0xd7610002, 0x0000f202,
+	0x80798179, 0xd7610002,
+	0x0000f203, 0x80798179,
+	0xd7610002, 0x0000f204,
+	0x80798179, 0xd7610002,
+	0x0000f205, 0x80798179,
+	0xd7610002, 0x0000f206,
+	0x80798179, 0xd7610002,
+	0x0000f207, 0x80798179,
+	0xd7610002, 0x0000f208,
+	0x80798179, 0xd7610002,
+	0x0000f209, 0x80798179,
+	0xd7610002, 0x0000f20a,
+	0x80798179, 0xd7610002,
+	0x0000f20b, 0x80798179,
+	0xe0704000, 0x705d0200,
+	0xbefe03c1, 0x907c9973,
+	0x877c817c, 0xbf06817c,
+	0xbf850002, 0xbeff0380,
+	0xbf820001, 0xbeff03c1,
+	0xb97b4306, 0x877bc17b,
+	0xbf840043, 0xbf8a0000,
+	0x877aff6d, 0x80000000,
+	0xbf84003f, 0x8f7b887b,
+	0xbef6037b, 0xb9703a05,
+	0x80708170, 0xbf0d9973,
+	0xbf850002, 0x8f708970,
+	0xbf820001, 0x8f708a70,
+	0xb97a1e06, 0x8f7a8a7a,
+	0x80707a70, 0x8070ff70,
+	0x00000200, 0x8070ff70,
+	0x00000080, 0xbef603ff,
+	0x01000000, 0xd7650000,
+	0x000100c1, 0xd7660000,
+	0x000200c1, 0x16000084,
+	0x907c9973, 0x877c817c,
+	0xbf06817c, 0xbefc0380,
+	0xbf850012, 0xbe8303ff,
+	0x00000080, 0xbf800000,
+	0xbf800000, 0xbf800000,
+	0xd8d80000, 0x01000000,
+	0xbf8c0000, 0xe0704000,
+	0x705d0100, 0x807c037c,
+	0x80700370, 0xd5250000,
+	0x0001ff00, 0x00000080,
+	0xbf0a7b7c, 0xbf85fff4,
+	0xbf820011, 0xbe8303ff,
+	0x00000100, 0xbf800000,
+	0xbf800000, 0xbf800000,
+	0xd8d80000, 0x01000000,
+	0xbf8c0000, 0xe0704000,
+	0x705d0100, 0x807c037c,
+	0x80700370, 0xd5250000,
+	0x0001ff00, 0x00000100,
+	0xbf0a7b7c, 0xbf85fff4,
+	0xbefe03c1, 0x907c9973,
+	0x877c817c, 0xbf06817c,
+	0xbf850004, 0xbef003ff,
+	0x00000200, 0xbeff0380,
+	0xbf820003, 0xbef003ff,
+	0x00000400, 0xbeff03c1,
+	0xb97b3a05, 0x807b817b,
+	0x8f7b827b, 0x907c9973,
+	0x877c817c, 0xbf06817c,
+	0xbf850017, 0xbef603ff,
+	0x01000000, 0xbefc0384,
+	0xbf0a7b7c, 0xbf840037,
+	0x7e008700, 0x7e028701,
+	0x7e048702, 0x7e068703,
+	0xe0704000, 0x705d0000,
+	0xe0704080, 0x705d0100,
+	0xe0704100, 0x705d0200,
+	0xe0704180, 0x705d0300,
+	0x807c847c, 0x8070ff70,
+	0x00000200, 0xbf0a7b7c,
+	0xbf85ffef, 0xbf820025,
+	0xbef603ff, 0x01000000,
+	0xbefc0384, 0xbf0a7b7c,
+	0xbf840011, 0x7e008700,
+	0x7e028701, 0x7e048702,
+	0x7e068703, 0xe0704000,
+	0x705d0000, 0xe0704100,
+	0x705d0100, 0xe0704200,
+	0x705d0200, 0xe0704300,
+	0x705d0300, 0x807c847c,
+	0x8070ff70, 0x00000400,
+	0xbf0a7b7c, 0xbf85ffef,
+	0xb97b1e06, 0x877bc17b,
+	0xbf84000c, 0x8f7b837b,
+	0x807b7c7b, 0xbefe03c1,
+	0xbeff0380, 0x7e008700,
+	0xe0704000, 0x705d0000,
+	0x807c817c, 0x8070ff70,
+	0x00000080, 0xbf0a7b7c,
+	0xbf85fff8, 0xbf820136,
+	0xbef4037e, 0x8775ff7f,
+	0x0000ffff, 0x8875ff75,
+	0x00040000, 0xbef60380,
+	0xbef703ff, 0x10807fac,
+	0xb97202dc, 0x8f729972,
+	0x876eff7f, 0x04000000,
+	0xbf840033, 0xbefe03c1,
+	0x907c9972, 0x877c817c,
+	0xbf06817c, 0xbf850002,
+	0xbeff0380, 0xbf820001,
+	0xbeff03c1, 0xb96f4306,
+	0x876fc16f, 0xbf840028,
+	0x8f6f886f, 0xbef6036f,
+	0xb9783a05, 0x80788178,
+	0xbf0d9972, 0xbf850002,
+	0x8f788978, 0xbf820001,
+	0x8f788a78, 0xb96e1e06,
+	0x8f6e8a6e, 0x80786e78,
+	0x8078ff78, 0x00000200,
+	0x8078ff78, 0x00000080,
+	0xbef603ff, 0x01000000,
+	0x907c9972, 0x877c817c,
+	0xbf06817c, 0xbefc0380,
+	0xbf850009, 0xe0310000,
+	0x781d0000, 0x807cff7c,
+	0x00000080, 0x8078ff78,
+	0x00000080, 0xbf0a6f7c,
+	0xbf85fff8, 0xbf820008,
+	0xe0310000, 0x781d0000,
+	0x807cff7c, 0x00000100,
+	0x8078ff78, 0x00000100,
+	0xbf0a6f7c, 0xbf85fff8,
+	0xbef80380, 0xbefe03c1,
+	0x907c9972, 0x877c817c,
+	0xbf06817c, 0xbf850002,
+	0xbeff0380, 0xbf820001,
+	0xbeff03c1, 0xb96f3a05,
+	0x806f816f, 0x8f6f826f,
+	0x907c9972, 0x877c817c,
+	0xbf06817c, 0xbf850024,
+	0xbef603ff, 0x01000000,
+	0xbeee0378, 0x8078ff78,
+	0x00000200, 0xbefc0384,
+	0xbf0a6f7c, 0xbf840050,
+	0xe0304000, 0x785d0000,
+	0xe0304080, 0x785d0100,
+	0xe0304100, 0x785d0200,
+	0xe0304180, 0x785d0300,
+	0xbf8c0000, 0x7e008500,
+	0x7e028501, 0x7e048502,
+	0x7e068503, 0x807c847c,
+	0x8078ff78, 0x00000200,
+	0xbf0a6f7c, 0xbf85ffee,
+	0xe0304000, 0x6e5d0000,
+	0xe0304080, 0x6e5d0100,
+	0xe0304100, 0x6e5d0200,
+	0xe0304180, 0x6e5d0300,
+	0xbf8c0000, 0xbf820034,
+	0xbef603ff, 0x01000000,
+	0xbeee0378, 0x8078ff78,
+	0x00000400, 0xbefc0384,
+	0xbf0a6f7c, 0xbf840012,
+	0xe0304000, 0x785d0000,
+	0xe0304100, 0x785d0100,
+	0xe0304200, 0x785d0200,
+	0xe0304300, 0x785d0300,
+	0xbf8c0000, 0x7e008500,
+	0x7e028501, 0x7e048502,
+	0x7e068503, 0x807c847c,
+	0x8078ff78, 0x00000400,
+	0xbf0a6f7c, 0xbf85ffee,
+	0xb96f1e06, 0x876fc16f,
+	0xbf84000e, 0x8f6f836f,
+	0x806f7c6f, 0xbefe03c1,
+	0xbeff0380, 0xe0304000,
+	0x785d0000, 0xbf8c0000,
+	0x7e008500, 0x807c817c,
+	0x8078ff78, 0x00000080,
+	0xbf0a6f7c, 0xbf85fff7,
+	0xbeff03c1, 0xe0304000,
+	0x6e5d0000, 0xe0304100,
+	0x6e5d0100, 0xe0304200,
+	0x6e5d0200, 0xe0304300,
+	0x6e5d0300, 0xbf8c0000,
+	0xb9783a05, 0x80788178,
+	0xbf0d9972, 0xbf850002,
+	0x8f788978, 0xbf820001,
+	0x8f788a78, 0xb96e1e06,
+	0x8f6e8a6e, 0x80786e78,
+	0x8078ff78, 0x00000200,
+	0x80f8ff78, 0x00000050,
+	0xbef603ff, 0x01000000,
+	0xbefc03ff, 0x0000006c,
+	0x80f89078, 0xf429003a,
+	0xf0000000, 0xbf8c0000,
+	0x80fc847c, 0xbf800000,
+	0xbe803100, 0xbe823102,
+	0x80f8a078, 0xf42d003a,
+	0xf0000000, 0xbf8c0000,
+	0x80fc887c, 0xbf800000,
+	0xbe803100, 0xbe823102,
+	0xbe843104, 0xbe863106,
+	0x80f8c078, 0xf431003a,
+	0xf0000000, 0xbf8c0000,
+	0x80fc907c, 0xbf800000,
+	0xbe803100, 0xbe823102,
+	0xbe843104, 0xbe863106,
+	0xbe883108, 0xbe8a310a,
+	0xbe8c310c, 0xbe8e310e,
+	0xbf06807c, 0xbf84fff0,
+	0xba80f801, 0x00000000,
+	0xbf8a0000, 0xb9783a05,
+	0x80788178, 0xbf0d9972,
+	0xbf850002, 0x8f788978,
+	0xbf820001, 0x8f788a78,
+	0xb96e1e06, 0x8f6e8a6e,
+	0x80786e78, 0x8078ff78,
+	0x00000200, 0xbef603ff,
+	0x01000000, 0xf4211bfa,
+	0xf0000000, 0x80788478,
+	0xf4211b3a, 0xf0000000,
+	0x80788478, 0xf4211b7a,
+	0xf0000000, 0x80788478,
+	0xf4211c3a, 0xf0000000,
+	0x80788478, 0xf4211c7a,
+	0xf0000000, 0x80788478,
+	0xf4211eba, 0xf0000000,
+	0x80788478, 0xf4211efa,
+	0xf0000000, 0x80788478,
+	0xf4211e7a, 0xf0000000,
+	0x80788478, 0xf4211cfa,
+	0xf0000000, 0x80788478,
+	0xf4211bba, 0xf0000000,
+	0x80788478, 0xbf8c0000,
+	0xb9eef814, 0xf4211bba,
+	0xf0000000, 0x80788478,
+	0xbf8c0000, 0xb9eef815,
+	0xbefc036f, 0xbefe0370,
+	0xbeff0371, 0xb9fb4803,
+	0x907b8b7b, 0xb9fba2c3,
+	0xb9f3f801, 0xb96e3a05,
+	0x806e816e, 0xbf0d9972,
+	0xbf850002, 0x8f6e896e,
+	0xbf820001, 0x8f6e8a6e,
+	0xb96f1e06, 0x8f6f8a6f,
+	0x806e6f6e, 0x806eff6e,
+	0x00000200, 0x806e746e,
+	0x826f8075, 0x876fff6f,
+	0x0000ffff, 0xf4091c37,
+	0xfa000050, 0xf4091d37,
+	0xfa000060, 0xf4011e77,
+	0xfa000074, 0xbf8c0000,
+	0x876dff6d, 0x0000ffff,
+	0x87fe7e7e, 0x87ea6a6a,
+	0xb9faf802, 0xbe80226c,
+	0xbf9b0000, 0xbf9f0000,
+	0xbf9f0000, 0xbf9f0000,
+	0xbf9f0000, 0xbf9f0000,
+};
+
+static const uint32_t cwsr_trap_gfx11_hex[] = {
+	0xbfa00001, 0xbfa00227,
+	0xb0804006, 0xb8f8f802,
+	0x9178ff78, 0x00020006,
+	0xb8fbf803, 0xbf0d9e6d,
+	0xbfa10001, 0xbfbd0000,
+	0xbf0d9f6d, 0xbfa20006,
+	0x8b6eff78, 0x00002000,
+	0xbfa10009, 0x8b6eff6d,
+	0x00ff0000, 0xbfa2001e,
+	0x8b6eff7b, 0x00000400,
+	0xbfa20045, 0xbf830010,
+	0xb8fbf803, 0xbfa0fffa,
+	0x8b6eff7b, 0x00160900,
+	0xbfa20015, 0x8b6eff7b,
+	0x000071ff, 0xbfa10008,
+	0x8b6fff7b, 0x00007080,
+	0xbfa10001, 0xbeee1287,
+	0xb8eff801, 0x846e8c6e,
+	0x8b6e6f6e, 0xbfa2000a,
+	0x8b6eff6d, 0x00ff0000,
+	0xbfa20007, 0xb8eef801,
+	0x8b6eff6e, 0x00000800,
+	0xbfa20003, 0x8b6eff7b,
+	0x00000400, 0xbfa2002a,
+	0xbefa4d82, 0xbf890000,
+	0x84fa887a, 0xbf0d8f7b,
+	0xbfa10002, 0x8c7bff7b,
+	0xffff0000, 0xf4005bbd,
+	0xf8000010, 0xbf890000,
+	0x846e976e, 0x9177ff77,
+	0x00800000, 0x8c776e77,
+	0xf4045bbd, 0xf8000000,
+	0xbf890000, 0xf4045ebd,
+	0xf8000008, 0xbf890000,
+	0x8bee6e6e, 0xbfa10001,
+	0xbe80486e, 0x8b6eff6d,
+	0x00ff0000, 0xbfa20008,
+	0x8b6eff6d, 0x01000000,
+	0xbfa20007, 0x8c78ff78,
+	0x00002000, 0x80ec886c,
+	0x82ed806d, 0xbfa00002,
+	0x806c846c, 0x826d806d,
+	0x8b6dff6d, 0x0000ffff,
+	0x8bfe7e7e, 0x8bea6a6a,
+	0xb978f802, 0xbe804a6c,
+	0xbf0d9878, 0xbfa10001,
+	0xbfb00000, 0x8b6dff6d,
+	0x0000ffff, 0xbefa0080,
+	0xb97a0283, 0xbeee007e,
+	0xbeef007f, 0xbefe0180,
+	0xbefe4d84, 0xbf890000,
+	0x8b7aff7f, 0x04000000,
+	0x847a857a, 0x8c6d7a6d,
+	0xbefa007e, 0x8b7bff7f,
+	0x0000ffff, 0xbefe00c1,
+	0xbeff00c1, 0xdca6c000,
+	0x007a0000, 0x7e000280,
+	0xbefe007a, 0xbeff007b,
+	0xb8fb02dc, 0x847b997b,
+	0xb8fa3b05, 0x807a817a,
+	0xbf0d997b, 0xbfa20002,
+	0x847a897a, 0xbfa00001,
+	0x847a8a7a, 0xb8fb1e06,
+	0x847b8a7b, 0x807a7b7a,
+	0x8b7bff7f, 0x0000ffff,
+	0x807aff7a, 0x00000200,
+	0x807a7e7a, 0x827b807b,
+	0xd7610000, 0x00010870,
+	0xd7610000, 0x00010a71,
+	0xd7610000, 0x00010c72,
+	0xd7610000, 0x00010e73,
+	0xd7610000, 0x00011074,
+	0xd7610000, 0x00011275,
+	0xd7610000, 0x00011476,
+	0xd7610000, 0x00011677,
+	0xd7610000, 0x00011a79,
+	0xd7610000, 0x00011c7e,
+	0xd7610000, 0x00011e7f,
+	0xbefe00ff, 0x00003fff,
+	0xbeff0080, 0xdca6c040,
+	0x007a0000, 0xd760007a,
+	0x00011d00, 0xd760007b,
+	0x00011f00, 0xbefe007a,
+	0xbeff007b, 0xbef4007e,
+	0x8b75ff7f, 0x0000ffff,
+	0x8c75ff75, 0x00040000,
+	0xbef60080, 0xbef700ff,
+	0x10807fac, 0xbef1007d,
+	0xbef00080, 0xb8f302dc,
+	0x84739973, 0xbefe00c1,
+	0x857d9973, 0x8b7d817d,
+	0xbf06817d, 0xbfa20002,
+	0xbeff0080, 0xbfa00002,
+	0xbeff00c1, 0xbfa00009,
+	0xbef600ff, 0x01000000,
+	0xe0685080, 0x701d0100,
+	0xe0685100, 0x701d0200,
+	0xe0685180, 0x701d0300,
+	0xbfa00008, 0xbef600ff,
+	0x01000000, 0xe0685100,
+	0x701d0100, 0xe0685200,
+	0x701d0200, 0xe0685300,
+	0x701d0300, 0xb8f03b05,
+	0x80708170, 0xbf0d9973,
+	0xbfa20002, 0x84708970,
+	0xbfa00001, 0x84708a70,
+	0xb8fa1e06, 0x847a8a7a,
+	0x80707a70, 0x8070ff70,
+	0x00000200, 0xbef600ff,
+	0x01000000, 0x7e000280,
+	0x7e020280, 0x7e040280,
+	0xbefd0080, 0xd7610002,
+	0x0000fa71, 0x807d817d,
+	0xd7610002, 0x0000fa6c,
+	0x807d817d, 0x917aff6d,
+	0x80000000, 0xd7610002,
+	0x0000fa7a, 0x807d817d,
+	0xd7610002, 0x0000fa6e,
+	0x807d817d, 0xd7610002,
+	0x0000fa6f, 0x807d817d,
+	0xd7610002, 0x0000fa78,
+	0x807d817d, 0xb8faf803,
+	0xd7610002, 0x0000fa7a,
+	0x807d817d, 0xd7610002,
+	0x0000fa7b, 0x807d817d,
+	0xb8f1f801, 0xd7610002,
+	0x0000fa71, 0x807d817d,
+	0xb8f1f814, 0xd7610002,
+	0x0000fa71, 0x807d817d,
+	0xb8f1f815, 0xd7610002,
+	0x0000fa71, 0x807d817d,
+	0xbefe00ff, 0x0000ffff,
+	0xbeff0080, 0xe0685000,
+	0x701d0200, 0xbefe00c1,
+	0xb8f03b05, 0x80708170,
+	0xbf0d9973, 0xbfa20002,
+	0x84708970, 0xbfa00001,
+	0x84708a70, 0xb8fa1e06,
+	0x847a8a7a, 0x80707a70,
+	0xbef600ff, 0x01000000,
+	0xbef90080, 0xbefd0080,
+	0xbf800000, 0xbe804100,
+	0xbe824102, 0xbe844104,
+	0xbe864106, 0xbe884108,
+	0xbe8a410a, 0xbe8c410c,
+	0xbe8e410e, 0xd7610002,
+	0x0000f200, 0x80798179,
+	0xd7610002, 0x0000f201,
+	0x80798179, 0xd7610002,
+	0x0000f202, 0x80798179,
+	0xd7610002, 0x0000f203,
+	0x80798179, 0xd7610002,
+	0x0000f204, 0x80798179,
+	0xd7610002, 0x0000f205,
+	0x80798179, 0xd7610002,
+	0x0000f206, 0x80798179,
+	0xd7610002, 0x0000f207,
+	0x80798179, 0xd7610002,
+	0x0000f208, 0x80798179,
+	0xd7610002, 0x0000f209,
+	0x80798179, 0xd7610002,
+	0x0000f20a, 0x80798179,
+	0xd7610002, 0x0000f20b,
+	0x80798179, 0xd7610002,
+	0x0000f20c, 0x80798179,
+	0xd7610002, 0x0000f20d,
+	0x80798179, 0xd7610002,
+	0x0000f20e, 0x80798179,
+	0xd7610002, 0x0000f20f,
+	0x80798179, 0xbf06a079,
+	0xbfa10006, 0xe0685000,
+	0x701d0200, 0x8070ff70,
+	0x00000080, 0xbef90080,
+	0x7e040280, 0x807d907d,
+	0xbf0aff7d, 0x00000060,
+	0xbfa2ffbc, 0xbe804100,
+	0xbe824102, 0xbe844104,
+	0xbe864106, 0xbe884108,
+	0xbe8a410a, 0xd7610002,
+	0x0000f200, 0x80798179,
+	0xd7610002, 0x0000f201,
+	0x80798179, 0xd7610002,
+	0x0000f202, 0x80798179,
+	0xd7610002, 0x0000f203,
+	0x80798179, 0xd7610002,
+	0x0000f204, 0x80798179,
+	0xd7610002, 0x0000f205,
+	0x80798179, 0xd7610002,
+	0x0000f206, 0x80798179,
+	0xd7610002, 0x0000f207,
+	0x80798179, 0xd7610002,
+	0x0000f208, 0x80798179,
+	0xd7610002, 0x0000f209,
+	0x80798179, 0xd7610002,
+	0x0000f20a, 0x80798179,
+	0xd7610002, 0x0000f20b,
+	0x80798179, 0xe0685000,
+	0x701d0200, 0xbefe00c1,
+	0x857d9973, 0x8b7d817d,
+	0xbf06817d, 0xbfa20002,
+	0xbeff0080, 0xbfa00001,
+	0xbeff00c1, 0xb8fb4306,
+	0x8b7bc17b, 0xbfa10043,
+	0xbfbd0000, 0x8b7aff6d,
+	0x80000000, 0xbfa1003f,
+	0x847b887b, 0xbef6007b,
+	0xb8f03b05, 0x80708170,
+	0xbf0d9973, 0xbfa20002,
+	0x84708970, 0xbfa00001,
+	0x84708a70, 0xb8fa1e06,
+	0x847a8a7a, 0x80707a70,
+	0x8070ff70, 0x00000200,
+	0x8070ff70, 0x00000080,
+	0xbef600ff, 0x01000000,
+	0xd71f0000, 0x000100c1,
+	0xd7200000, 0x000200c1,
+	0x16000084, 0x857d9973,
+	0x8b7d817d, 0xbf06817d,
+	0xbefd0080, 0xbfa20012,
+	0xbe8300ff, 0x00000080,
+	0xbf800000, 0xbf800000,
+	0xbf800000, 0xd8d80000,
+	0x01000000, 0xbf890000,
+	0xe0685000, 0x701d0100,
+	0x807d037d, 0x80700370,
+	0xd5250000, 0x0001ff00,
+	0x00000080, 0xbf0a7b7d,
+	0xbfa2fff4, 0xbfa00011,
+	0xbe8300ff, 0x00000100,
+	0xbf800000, 0xbf800000,
+	0xbf800000, 0xd8d80000,
+	0x01000000, 0xbf890000,
+	0xe0685000, 0x701d0100,
+	0x807d037d, 0x80700370,
+	0xd5250000, 0x0001ff00,
+	0x00000100, 0xbf0a7b7d,
+	0xbfa2fff4, 0xbefe00c1,
+	0x857d9973, 0x8b7d817d,
+	0xbf06817d, 0xbfa20004,
+	0xbef000ff, 0x00000200,
+	0xbeff0080, 0xbfa00003,
+	0xbef000ff, 0x00000400,
+	0xbeff00c1, 0xb8fb3b05,
+	0x807b817b, 0x847b827b,
+	0x857d9973, 0x8b7d817d,
+	0xbf06817d, 0xbfa20017,
+	0xbef600ff, 0x01000000,
+	0xbefd0084, 0xbf0a7b7d,
+	0xbfa10037, 0x7e008700,
+	0x7e028701, 0x7e048702,
+	0x7e068703, 0xe0685000,
+	0x701d0000, 0xe0685080,
+	0x701d0100, 0xe0685100,
+	0x701d0200, 0xe0685180,
+	0x701d0300, 0x807d847d,
+	0x8070ff70, 0x00000200,
+	0xbf0a7b7d, 0xbfa2ffef,
+	0xbfa00025, 0xbef600ff,
+	0x01000000, 0xbefd0084,
+	0xbf0a7b7d, 0xbfa10011,
+	0x7e008700, 0x7e028701,
+	0x7e048702, 0x7e068703,
+	0xe0685000, 0x701d0000,
+	0xe0685100, 0x701d0100,
+	0xe0685200, 0x701d0200,
+	0xe0685300, 0x701d0300,
+	0x807d847d, 0x8070ff70,
+	0x00000400, 0xbf0a7b7d,
+	0xbfa2ffef, 0xb8fb1e06,
+	0x8b7bc17b, 0xbfa1000c,
+	0x847b837b, 0x807b7d7b,
+	0xbefe00c1, 0xbeff0080,
+	0x7e008700, 0xe0685000,
+	0x701d0000, 0x807d817d,
+	0x8070ff70, 0x00000080,
+	0xbf0a7b7d, 0xbfa2fff8,
+	0xbfa00143, 0xbef4007e,
+	0x8b75ff7f, 0x0000ffff,
+	0x8c75ff75, 0x00040000,
+	0xbef60080, 0xbef700ff,
+	0x10807fac, 0xb8f202dc,
+	0x84729972, 0x8b6eff7f,
+	0x04000000, 0xbfa10039,
+	0xbefe00c1, 0x857d9972,
+	0x8b7d817d, 0xbf06817d,
+	0xbfa20002, 0xbeff0080,
+	0xbfa00001, 0xbeff00c1,
+	0xb8ef4306, 0x8b6fc16f,
+	0xbfa1002e, 0x846f886f,
+	0xbef6006f, 0xb8f83b05,
+	0x80788178, 0xbf0d9972,
+	0xbfa20002, 0x84788978,
+	0xbfa00001, 0x84788a78,
+	0xb8ee1e06, 0x846e8a6e,
+	0x80786e78, 0x8078ff78,
+	0x00000200, 0x8078ff78,
+	0x00000080, 0xbef600ff,
+	0x01000000, 0x857d9972,
+	0x8b7d817d, 0xbf06817d,
+	0xbefd0080, 0xbfa2000c,
+	0xe0500000, 0x781d0000,
+	0xbf890000, 0xdac00000,
+	0x00000000, 0x807dff7d,
+	0x00000080, 0x8078ff78,
+	0x00000080, 0xbf0a6f7d,
+	0xbfa2fff5, 0xbfa0000b,
+	0xe0500000, 0x781d0000,
+	0xbf890000, 0xdac00000,
+	0x00000000, 0x807dff7d,
+	0x00000100, 0x8078ff78,
+	0x00000100, 0xbf0a6f7d,
+	0xbfa2fff5, 0xbef80080,
+	0xbefe00c1, 0x857d9972,
+	0x8b7d817d, 0xbf06817d,
+	0xbfa20002, 0xbeff0080,
+	0xbfa00001, 0xbeff00c1,
+	0xb8ef3b05, 0x806f816f,
+	0x846f826f, 0x857d9972,
+	0x8b7d817d, 0xbf06817d,
+	0xbfa20024, 0xbef600ff,
+	0x01000000, 0xbeee0078,
+	0x8078ff78, 0x00000200,
+	0xbefd0084, 0xbf0a6f7d,
+	0xbfa10050, 0xe0505000,
+	0x781d0000, 0xe0505080,
+	0x781d0100, 0xe0505100,
+	0x781d0200, 0xe0505180,
+	0x781d0300, 0xbf890000,
+	0x7e008500, 0x7e028501,
+	0x7e048502, 0x7e068503,
+	0x807d847d, 0x8078ff78,
+	0x00000200, 0xbf0a6f7d,
+	0xbfa2ffee, 0xe0505000,
+	0x6e1d0000, 0xe0505080,
+	0x6e1d0100, 0xe0505100,
+	0x6e1d0200, 0xe0505180,
+	0x6e1d0300, 0xbf890000,
+	0xbfa00034, 0xbef600ff,
+	0x01000000, 0xbeee0078,
+	0x8078ff78, 0x00000400,
+	0xbefd0084, 0xbf0a6f7d,
+	0xbfa10012, 0xe0505000,
+	0x781d0000, 0xe0505100,
+	0x781d0100, 0xe0505200,
+	0x781d0200, 0xe0505300,
+	0x781d0300, 0xbf890000,
+	0x7e008500, 0x7e028501,
+	0x7e048502, 0x7e068503,
+	0x807d847d, 0x8078ff78,
+	0x00000400, 0xbf0a6f7d,
+	0xbfa2ffee, 0xb8ef1e06,
+	0x8b6fc16f, 0xbfa1000e,
+	0x846f836f, 0x806f7d6f,
+	0xbefe00c1, 0xbeff0080,
+	0xe0505000, 0x781d0000,
+	0xbf890000, 0x7e008500,
+	0x807d817d, 0x8078ff78,
+	0x00000080, 0xbf0a6f7d,
+	0xbfa2fff7, 0xbeff00c1,
+	0xe0505000, 0x6e1d0000,
+	0xe0505100, 0x6e1d0100,
+	0xe0505200, 0x6e1d0200,
+	0xe0505300, 0x6e1d0300,
+	0xbf890000, 0xb8f83b05,
+	0x80788178, 0xbf0d9972,
+	0xbfa20002, 0x84788978,
+	0xbfa00001, 0x84788a78,
+	0xb8ee1e06, 0x846e8a6e,
+	0x80786e78, 0x8078ff78,
+	0x00000200, 0x80f8ff78,
+	0x00000050, 0xbef600ff,
+	0x01000000, 0xbefd00ff,
+	0x0000006c, 0x80f89078,
+	0xf428403a, 0xf0000000,
+	0xbf890000, 0x80fd847d,
+	0xbf800000, 0xbe804300,
+	0xbe824302, 0x80f8a078,
+	0xf42c403a, 0xf0000000,
+	0xbf890000, 0x80fd887d,
+	0xbf800000, 0xbe804300,
+	0xbe824302, 0xbe844304,
+	0xbe864306, 0x80f8c078,
+	0xf430403a, 0xf0000000,
+	0xbf890000, 0x80fd907d,
+	0xbf800000, 0xbe804300,
+	0xbe824302, 0xbe844304,
+	0xbe864306, 0xbe884308,
+	0xbe8a430a, 0xbe8c430c,
+	0xbe8e430e, 0xbf06807d,
+	0xbfa1fff0, 0xb980f801,
+	0x00000000, 0xbfbd0000,
+	0xb8f83b05, 0x80788178,
+	0xbf0d9972, 0xbfa20002,
+	0x84788978, 0xbfa00001,
+	0x84788a78, 0xb8ee1e06,
+	0x846e8a6e, 0x80786e78,
+	0x8078ff78, 0x00000200,
+	0xbef600ff, 0x01000000,
+	0xf4205bfa, 0xf0000000,
+	0x80788478, 0xf4205b3a,
+	0xf0000000, 0x80788478,
+	0xf4205b7a, 0xf0000000,
+	0x80788478, 0xf4205c3a,
+	0xf0000000, 0x80788478,
+	0xf4205c7a, 0xf0000000,
+	0x80788478, 0xf4205eba,
+	0xf0000000, 0x80788478,
+	0xf4205efa, 0xf0000000,
+	0x80788478, 0xf4205e7a,
+	0xf0000000, 0x80788478,
+	0xf4205cfa, 0xf0000000,
+	0x80788478, 0xf4205bba,
+	0xf0000000, 0x80788478,
+	0xbf890000, 0xb96ef814,
+	0xf4205bba, 0xf0000000,
+	0x80788478, 0xbf890000,
+	0xb96ef815, 0xbefd006f,
+	0xbefe0070, 0xbeff0071,
+	0xb97b4803, 0x857b8b7b,
+	0xb97b22c3, 0x857b867b,
+	0xb97b7443, 0xb973f801,
+	0xb8ee3b05, 0x806e816e,
+	0xbf0d9972, 0xbfa20002,
+	0x846e896e, 0xbfa00001,
+	0x846e8a6e, 0xb8ef1e06,
+	0x846f8a6f, 0x806e6f6e,
+	0x806eff6e, 0x00000200,
+	0x806e746e, 0x826f8075,
+	0x8b6fff6f, 0x0000ffff,
+	0xf4085c37, 0xf8000050,
+	0xf4085d37, 0xf8000060,
+	0xf4005e77, 0xf8000074,
+	0xbf890000, 0x8b6dff6d,
+	0x0000ffff, 0x8bfe7e7e,
+	0x8bea6a6a, 0xb8eef802,
+	0xbf0d866e, 0xbfa20002,
+	0xb97af802, 0xbe80486c,
+	0xb97af802, 0xbe804a6c,
+	0xbfb10000, 0xbf9f0000,
+	0xbf9f0000, 0xbf9f0000,
+	0xbf9f0000, 0xbf9f0000,
+};
+
+static const uint32_t cwsr_trap_gfx9_4_3_hex[] = {
+	0xbf820001, 0xbf8202dc,
+	0xb8f8f802, 0x8978ff78,
+	0x00020006, 0xb8fbf803,
+	0x866eff78, 0x00002000,
+	0xbf840009, 0x866eff6d,
+	0x00ff0000, 0xbf85001a,
+	0x866eff7b, 0x00000400,
+	0xbf850051, 0xbf8e0010,
+	0xb8fbf803, 0xbf82fffa,
+	0x866eff7b, 0x03c00900,
+	0xbf850011, 0x866eff7b,
+	0x000071ff, 0xbf840008,
+	0x866fff7b, 0x00007080,
+	0xbf840001, 0xbeee1a87,
+	0xb8eff801, 0x8e6e8c6e,
+	0x866e6f6e, 0xbf850006,
+	0x866eff6d, 0x00ff0000,
+	0xbf850003, 0x866eff7b,
+	0x00000400, 0xbf85003a,
+	0xb8faf807, 0x867aff7a,
+	0x001f8000, 0x8e7a8b7a,
+	0x8979ff79, 0xfc000000,
+	0x87797a79, 0xba7ff807,
+	0x00000000, 0xb8faf812,
+	0xb8fbf813, 0x8efa887a,
+	0xbf0d8f7b, 0xbf840002,
+	0x877bff7b, 0xffff0000,
+	0xc0031bbd, 0x00000010,
+	0xbf8cc07f, 0x8e6e976e,
+	0x8979ff79, 0x00800000,
+	0x87796e79, 0xc0071bbd,
+	0x00000000, 0xbf8cc07f,
+	0xc0071ebd, 0x00000008,
+	0xbf8cc07f, 0x86ee6e6e,
+	0xbf840001, 0xbe801d6e,
+	0x866eff6d, 0x01ff0000,
+	0xbf850005, 0x8778ff78,
+	0x00002000, 0x80ec886c,
+	0x82ed806d, 0xbf820005,
+	0x866eff6d, 0x01000000,
+	0xbf850002, 0x806c846c,
+	0x826d806d, 0x866dff6d,
+	0x0000ffff, 0x8f7a8b79,
+	0x867aff7a, 0x001f8000,
+	0xb97af807, 0x86fe7e7e,
+	0x86ea6a6a, 0x8f6e8378,
+	0xb96ee0c2, 0xbf800002,
+	0xb9780002, 0xbe801f6c,
+	0x866dff6d, 0x0000ffff,
+	0xbefa0080, 0xb97a0283,
+	0xb8faf807, 0x867aff7a,
+	0x001f8000, 0x8e7a8b7a,
+	0x8979ff79, 0xfc000000,
+	0x87797a79, 0xba7ff807,
+	0x00000000, 0xbeee007e,
+	0xbeef007f, 0xbefe0180,
+	0xbf900004, 0x877a8478,
+	0xb97af802, 0xbf8e0002,
+	0xbf88fffe, 0xb8fa2985,
+	0x807a817a, 0x8e7a8a7a,
+	0x8e7a817a, 0xb8fb1605,
+	0x807b817b, 0x8e7b867b,
+	0x807a7b7a, 0x807a7e7a,
+	0x827b807f, 0x867bff7b,
+	0x0000ffff, 0xc04b1c3d,
+	0x00000050, 0xbf8cc07f,
+	0xc04b1d3d, 0x00000060,
+	0xbf8cc07f, 0xc0431e7d,
+	0x00000074, 0xbf8cc07f,
+	0xbef4007e, 0x8675ff7f,
+	0x0000ffff, 0x8775ff75,
+	0x00040000, 0xbef60080,
+	0xbef700ff, 0x00807fac,
+	0xbef1007c, 0xbef00080,
+	0xb8f02985, 0x80708170,
+	0x8e708a70, 0x8e708170,
+	0xb8fa1605, 0x807a817a,
+	0x8e7a867a, 0x80707a70,
+	0xbef60084, 0xbef600ff,
+	0x01000000, 0xbefe007c,
+	0xbefc0070, 0xc0611c7a,
+	0x0000007c, 0xbf8cc07f,
+	0x80708470, 0xbefc007e,
+	0xbefe007c, 0xbefc0070,
+	0xc0611b3a, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xbefe007c,
+	0xbefc0070, 0xc0611b7a,
+	0x0000007c, 0xbf8cc07f,
+	0x80708470, 0xbefc007e,
+	0xbefe007c, 0xbefc0070,
+	0xc0611bba, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xbefe007c,
+	0xbefc0070, 0xc0611bfa,
+	0x0000007c, 0xbf8cc07f,
+	0x80708470, 0xbefc007e,
+	0xbefe007c, 0xbefc0070,
+	0xc0611e3a, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xb8fbf803,
+	0xbefe007c, 0xbefc0070,
+	0xc0611efa, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xbefe007c,
+	0xbefc0070, 0xc0611a3a,
+	0x0000007c, 0xbf8cc07f,
+	0x80708470, 0xbefc007e,
+	0xbefe007c, 0xbefc0070,
+	0xc0611a7a, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xb8f1f801,
+	0xbefe007c, 0xbefc0070,
+	0xc0611c7a, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xbf108080,
+	0x867aff7f, 0x04000000,
+	0xbeef0080, 0x876f6f7a,
+	0xb8f02985, 0x80708170,
+	0x8e708a70, 0x8e708170,
+	0xb8fb1605, 0x807b817b,
+	0x8e7b847b, 0x8e76827b,
+	0xbef600ff, 0x01000000,
+	0xbef20174, 0x80747074,
+	0x82758075, 0xbefc0080,
+	0xbf800000, 0xbe802b00,
+	0xbe822b02, 0xbe842b04,
+	0xbe862b06, 0xbe882b08,
+	0xbe8a2b0a, 0xbe8c2b0c,
+	0xbe8e2b0e, 0xc06b003a,
+	0x00000000, 0xbf8cc07f,
+	0xc06b013a, 0x00000010,
+	0xbf8cc07f, 0xc06b023a,
+	0x00000020, 0xbf8cc07f,
+	0xc06b033a, 0x00000030,
+	0xbf8cc07f, 0x8074c074,
+	0x82758075, 0x807c907c,
+	0xbf0a7b7c, 0xbf85ffe7,
+	0xbef40172, 0xbef00080,
+	0xbefe00c1, 0xbeff00c1,
+	0xbee80080, 0xbee90080,
+	0xbef600ff, 0x01000000,
+	0x867aff78, 0x00400000,
+	0xbf850003, 0xb8faf803,
+	0x897a7aff, 0x10000000,
+	0xbf85004d, 0xbe840080,
+	0xd2890000, 0x00000900,
+	0x80048104, 0xd2890001,
+	0x00000900, 0x80048104,
+	0xd2890002, 0x00000900,
+	0x80048104, 0xd2890003,
+	0x00000900, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000901, 0x80048104,
+	0xd2890001, 0x00000901,
+	0x80048104, 0xd2890002,
+	0x00000901, 0x80048104,
+	0xd2890003, 0x00000901,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000902,
+	0x80048104, 0xd2890001,
+	0x00000902, 0x80048104,
+	0xd2890002, 0x00000902,
+	0x80048104, 0xd2890003,
+	0x00000902, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000903, 0x80048104,
+	0xd2890001, 0x00000903,
+	0x80048104, 0xd2890002,
+	0x00000903, 0x80048104,
+	0xd2890003, 0x00000903,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbf820008,
+	0xe0724000, 0x701d0000,
+	0xe0724100, 0x701d0100,
+	0xe0724200, 0x701d0200,
+	0xe0724300, 0x701d0300,
+	0xbefe00c1, 0xbeff00c1,
+	0xb8fb4306, 0x867bc17b,
+	0xbf840064, 0xbf8a0000,
+	0x867aff6f, 0x04000000,
+	0xbf840060, 0x8e7b867b,
+	0x8e7b827b, 0xbef6007b,
+	0xb8f02985, 0x80708170,
+	0x8e708a70, 0x8e708170,
+	0xb8fa1605, 0x807a817a,
+	0x8e7a867a, 0x80707a70,
+	0x8070ff70, 0x00000080,
+	0xbef600ff, 0x01000000,
+	0xbefc0080, 0xd28c0002,
+	0x000100c1, 0xd28d0003,
+	0x000204c1, 0x867aff78,
+	0x00400000, 0xbf850003,
+	0xb8faf803, 0x897a7aff,
+	0x10000000, 0xbf850030,
+	0x24040682, 0xd86e4000,
+	0x00000002, 0xbf8cc07f,
+	0xbe840080, 0xd2890000,
+	0x00000900, 0x80048104,
+	0xd2890001, 0x00000900,
+	0x80048104, 0xd2890002,
+	0x00000900, 0x80048104,
+	0xd2890003, 0x00000900,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000901,
+	0x80048104, 0xd2890001,
+	0x00000901, 0x80048104,
+	0xd2890002, 0x00000901,
+	0x80048104, 0xd2890003,
+	0x00000901, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0x680404ff, 0x00000200,
+	0xd0c9006a, 0x0000f702,
+	0xbf87ffd2, 0xbf820015,
+	0xd1060002, 0x00011103,
+	0x7e0602ff, 0x00000200,
+	0xbefc00ff, 0x00010000,
+	0xbe800077, 0x8677ff77,
+	0xff7fffff, 0x8777ff77,
+	0x00058000, 0xd8ec0000,
+	0x00000002, 0xbf8cc07f,
+	0xe0765000, 0x701d0002,
+	0x68040702, 0xd0c9006a,
+	0x0000f702, 0xbf87fff7,
+	0xbef70000, 0xbef000ff,
+	0x00000400, 0xbefe00c1,
+	0xbeff00c1, 0xb8fb2b05,
+	0x807b817b, 0x8e7b827b,
+	0xbef600ff, 0x01000000,
+	0xbefc0084, 0xbf0a7b7c,
+	0xbf84006d, 0xbf11017c,
+	0x807bff7b, 0x00001000,
+	0x867aff78, 0x00400000,
+	0xbf850003, 0xb8faf803,
+	0x897a7aff, 0x10000000,
+	0xbf850051, 0xbe840080,
+	0xd2890000, 0x00000900,
+	0x80048104, 0xd2890001,
+	0x00000900, 0x80048104,
+	0xd2890002, 0x00000900,
+	0x80048104, 0xd2890003,
+	0x00000900, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000901, 0x80048104,
+	0xd2890001, 0x00000901,
+	0x80048104, 0xd2890002,
+	0x00000901, 0x80048104,
+	0xd2890003, 0x00000901,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000902,
+	0x80048104, 0xd2890001,
+	0x00000902, 0x80048104,
+	0xd2890002, 0x00000902,
+	0x80048104, 0xd2890003,
+	0x00000902, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000903, 0x80048104,
+	0xd2890001, 0x00000903,
+	0x80048104, 0xd2890002,
+	0x00000903, 0x80048104,
+	0xd2890003, 0x00000903,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0x807c847c,
+	0xbf0a7b7c, 0xbf85ffb1,
+	0xbf9c0000, 0xbf820012,
+	0x7e000300, 0x7e020301,
+	0x7e040302, 0x7e060303,
+	0xe0724000, 0x701d0000,
+	0xe0724100, 0x701d0100,
+	0xe0724200, 0x701d0200,
+	0xe0724300, 0x701d0300,
+	0x807c847c, 0x8070ff70,
+	0x00000400, 0xbf0a7b7c,
+	0xbf85ffef, 0xbf9c0000,
+	0xb8fb2985, 0x807b817b,
+	0x8e7b837b, 0xb8fa2b05,
+	0x807a817a, 0x8e7a827a,
+	0x80fb7a7b, 0x867b7b7b,
+	0xbf84007a, 0x807bff7b,
+	0x00001000, 0xbefc0080,
+	0xbf11017c, 0x867aff78,
+	0x00400000, 0xbf850003,
+	0xb8faf803, 0x897a7aff,
+	0x10000000, 0xbf850059,
+	0xd3d84000, 0x18000100,
+	0xd3d84001, 0x18000101,
+	0xd3d84002, 0x18000102,
+	0xd3d84003, 0x18000103,
+	0xbe840080, 0xd2890000,
+	0x00000900, 0x80048104,
+	0xd2890001, 0x00000900,
+	0x80048104, 0xd2890002,
+	0x00000900, 0x80048104,
+	0xd2890003, 0x00000900,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000901,
+	0x80048104, 0xd2890001,
+	0x00000901, 0x80048104,
+	0xd2890002, 0x00000901,
+	0x80048104, 0xd2890003,
+	0x00000901, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000902, 0x80048104,
+	0xd2890001, 0x00000902,
+	0x80048104, 0xd2890002,
+	0x00000902, 0x80048104,
+	0xd2890003, 0x00000902,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000903,
+	0x80048104, 0xd2890001,
+	0x00000903, 0x80048104,
+	0xd2890002, 0x00000903,
+	0x80048104, 0xd2890003,
+	0x00000903, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0x807c847c, 0xbf0a7b7c,
+	0xbf85ffa9, 0xbf9c0000,
+	0xbf820016, 0xd3d84000,
+	0x18000100, 0xd3d84001,
+	0x18000101, 0xd3d84002,
+	0x18000102, 0xd3d84003,
+	0x18000103, 0xe0724000,
+	0x701d0000, 0xe0724100,
+	0x701d0100, 0xe0724200,
+	0x701d0200, 0xe0724300,
+	0x701d0300, 0x807c847c,
+	0x8070ff70, 0x00000400,
+	0xbf0a7b7c, 0xbf85ffeb,
+	0xbf9c0000, 0xbf8200ee,
+	0xbef4007e, 0x8675ff7f,
+	0x0000ffff, 0x8775ff75,
+	0x00040000, 0xbef60080,
+	0xbef700ff, 0x00807fac,
+	0x866eff7f, 0x04000000,
+	0xbf84001f, 0xbefe00c1,
+	0xbeff00c1, 0xb8ef4306,
+	0x866fc16f, 0xbf84001a,
+	0x8e6f866f, 0x8e6f826f,
+	0xbef6006f, 0xb8f82985,
+	0x80788178, 0x8e788a78,
+	0x8e788178, 0xb8ee1605,
+	0x806e816e, 0x8e6e866e,
+	0x80786e78, 0x8078ff78,
+	0x00000080, 0xbef600ff,
+	0x01000000, 0xbefc0080,
+	0xe0510000, 0x781d0000,
+	0xe0510100, 0x781d0000,
+	0x807cff7c, 0x00000200,
+	0x8078ff78, 0x00000200,
+	0xbf0a6f7c, 0xbf85fff6,
+	0xbefe00c1, 0xbeff00c1,
+	0xbef600ff, 0x01000000,
+	0xb8ef2b05, 0x806f816f,
+	0x8e6f826f, 0x806fff6f,
+	0x00008000, 0xbef80080,
+	0xbeee0078, 0x8078ff78,
+	0x00000400, 0xbefc0084,
+	0xbf11087c, 0xe0524000,
+	0x781d0000, 0xe0524100,
+	0x781d0100, 0xe0524200,
+	0x781d0200, 0xe0524300,
+	0x781d0300, 0xbf8c0f70,
+	0x7e000300, 0x7e020301,
+	0x7e040302, 0x7e060303,
+	0x807c847c, 0x8078ff78,
+	0x00000400, 0xbf0a6f7c,
+	0xbf85ffee, 0xb8ef2985,
+	0x806f816f, 0x8e6f836f,
+	0xb8f92b05, 0x80798179,
+	0x8e798279, 0x80ef796f,
+	0x866f6f6f, 0xbf84001a,
+	0x806fff6f, 0x00008000,
+	0xbefc0080, 0xbf11087c,
+	0xe0524000, 0x781d0000,
+	0xe0524100, 0x781d0100,
+	0xe0524200, 0x781d0200,
+	0xe0524300, 0x781d0300,
+	0xbf8c0f70, 0xd3d94000,
+	0x18000100, 0xd3d94001,
+	0x18000101, 0xd3d94002,
+	0x18000102, 0xd3d94003,
+	0x18000103, 0x807c847c,
+	0x8078ff78, 0x00000400,
+	0xbf0a6f7c, 0xbf85ffea,
+	0xbf9c0000, 0xe0524000,
+	0x6e1d0000, 0xe0524100,
+	0x6e1d0100, 0xe0524200,
+	0x6e1d0200, 0xe0524300,
+	0x6e1d0300, 0xbf8c0f70,
+	0xb8f82985, 0x80788178,
+	0x8e788a78, 0x8e788178,
+	0xb8ee1605, 0x806e816e,
+	0x8e6e866e, 0x80786e78,
+	0x80f8c078, 0xb8ef1605,
+	0x806f816f, 0x8e6f846f,
+	0x8e76826f, 0xbef600ff,
+	0x01000000, 0xbefc006f,
+	0xc031003a, 0x00000078,
+	0x80f8c078, 0xbf8cc07f,
+	0x80fc907c, 0xbf800000,
+	0xbe802d00, 0xbe822d02,
+	0xbe842d04, 0xbe862d06,
+	0xbe882d08, 0xbe8a2d0a,
+	0xbe8c2d0c, 0xbe8e2d0e,
+	0xbf06807c, 0xbf84fff0,
+	0xb8f82985, 0x80788178,
+	0x8e788a78, 0x8e788178,
+	0xb8ee1605, 0x806e816e,
+	0x8e6e866e, 0x80786e78,
+	0xbef60084, 0xbef600ff,
+	0x01000000, 0xc0211bfa,
+	0x00000078, 0x80788478,
+	0xc0211b3a, 0x00000078,
+	0x80788478, 0xc0211b7a,
+	0x00000078, 0x80788478,
+	0xc0211c3a, 0x00000078,
+	0x80788478, 0xc0211c7a,
+	0x00000078, 0x80788478,
+	0xc0211eba, 0x00000078,
+	0x80788478, 0xc0211efa,
+	0x00000078, 0x80788478,
+	0xc0211a3a, 0x00000078,
+	0x80788478, 0xc0211a7a,
+	0x00000078, 0x80788478,
+	0xc0211cfa, 0x00000078,
+	0x80788478, 0xbf8cc07f,
+	0xbefc006f, 0xbefe0070,
+	0xbeff0071, 0x866f7bff,
+	0x000003ff, 0xb96f4803,
+	0x866f7bff, 0xfffff800,
+	0x8f6f8b6f, 0xb96fa2c3,
+	0xb973f801, 0xb8ee2985,
+	0x806e816e, 0x8e6e8a6e,
+	0x8e6e816e, 0xb8ef1605,
+	0x806f816f, 0x8e6f866f,
+	0x806e6f6e, 0x806e746e,
+	0x826f8075, 0x866fff6f,
+	0x0000ffff, 0xc00b1c37,
+	0x00000050, 0xc00b1d37,
+	0x00000060, 0xc0031e77,
+	0x00000074, 0xbf8cc07f,
+	0x8f6e8b79, 0x866eff6e,
+	0x001f8000, 0xb96ef807,
+	0x866dff6d, 0x0000ffff,
+	0x86fe7e7e, 0x86ea6a6a,
+	0x8f6e837a, 0xb96ee0c2,
+	0xbf800002, 0xb97a0002,
+	0xbf8a0000, 0xbe801f6c,
+	0xbf9b0000, 0x00000000,
+};
+
+static const uint32_t cwsr_trap_gfx12_hex[] = {
+	0xbfa00001, 0xbfa002a2,
+	0xb0804009, 0xb8f8f804,
+	0x9178ff78, 0x00008c00,
+	0xb8fbf811, 0x8b6eff78,
+	0x00004000, 0xbfa10008,
+	0x8b6eff7b, 0x00000080,
+	0xbfa20018, 0x8b6ea07b,
+	0xbfa20042, 0xbf830010,
+	0xb8fbf811, 0xbfa0fffb,
+	0x8b6eff7b, 0x00000bd0,
+	0xbfa20010, 0xb8eef812,
+	0x8b6f8f7b, 0xbfa10002,
+	0x8c6eff6e, 0x00000080,
+	0xb8eff813, 0x8b6e6e6f,
+	0xbfa20008, 0x8b6eff6d,
+	0xf0000000, 0xbfa20005,
+	0x8b6fff6f, 0x00000200,
+	0xbfa20002, 0x8b6ea07b,
+	0xbfa2002c, 0xbefa4d82,
+	0xbf8a0000, 0x84fa887a,
+	0xbf0d8f7b, 0xbfa10002,
+	0x8c7bff7b, 0xffff0000,
+	0xf4601bbd, 0xf8000010,
+	0xbf8a0000, 0x846e976e,
+	0x9177ff77, 0x00800000,
+	0x8c776e77, 0xf4603bbd,
+	0xf8000000, 0xbf8a0000,
+	0xf4603ebd, 0xf8000008,
+	0xbf8a0000, 0x8bee6e6e,
+	0xbfa10001, 0xbe80486e,
+	0x8b6eff6d, 0xf0000000,
+	0xbfa20009, 0xb8eef811,
+	0x8b6eff6e, 0x00000080,
+	0xbfa20007, 0x8c78ff78,
+	0x00004000, 0x80ec886c,
+	0x82ed806d, 0xbfa00002,
+	0x806c846c, 0x826d806d,
+	0x8b6dff6d, 0x0000ffff,
+	0x8bfe7e7e, 0x8bea6a6a,
+	0x85788978, 0xb9783244,
+	0xbe804a6c, 0xb8faf802,
+	0xbf0d987a, 0xbfa10001,
+	0xbfb00000, 0x8b6dff6d,
+	0x0000ffff, 0xbefa0080,
+	0xb97a0151, 0xbeee007e,
+	0xbeef007f, 0xbefe0180,
+	0xbefe4d84, 0xbf8a0000,
+	0x8b7aff7f, 0x04000000,
+	0x847a857a, 0x8c6d7a6d,
+	0xbefa007e, 0x8b7bff7f,
+	0x0000ffff, 0xbefe00c1,
+	0xbeff00c1, 0xee0a407a,
+	0x000c0000, 0x00000000,
+	0x7e000280, 0xbefe007a,
+	0xbeff007b, 0xb8fb0742,
+	0x847b997b, 0xb8fa3b05,
+	0x807a817a, 0xbf0d997b,
+	0xbfa20002, 0x847a897a,
+	0xbfa00001, 0x847a8a7a,
+	0xb8fb1e06, 0x847b8a7b,
+	0x807a7b7a, 0x8b7bff7f,
+	0x0000ffff, 0x807aff7a,
+	0x00000200, 0x807a7e7a,
+	0x827b807b, 0xd7610000,
+	0x00010870, 0xd7610000,
+	0x00010a71, 0xd7610000,
+	0x00010c72, 0xd7610000,
+	0x00010e73, 0xd7610000,
+	0x00011074, 0xd7610000,
+	0x00011275, 0xd7610000,
+	0x00011476, 0xd7610000,
+	0x00011677, 0xd7610000,
+	0x00011a79, 0xd7610000,
+	0x00011c7e, 0xd7610000,
+	0x00011e7f, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xbefe00ff,
+	0x00003fff, 0xbeff0080,
+	0xee0a407a, 0x000c0000,
+	0x00004000, 0xd760007a,
+	0x00011d00, 0xd760007b,
+	0x00011f00, 0xbefe007a,
+	0xbeff007b, 0xbef4007e,
+	0x8b75ff7f, 0x0000ffff,
+	0x8c75ff75, 0x00040000,
+	0xbef60080, 0xbef700ff,
+	0x10807fac, 0xbef1007d,
+	0xbef00080, 0xb8f30742,
+	0x84739973, 0xbefe00c1,
+	0x857d9973, 0x8b7d817d,
+	0xbf06817d, 0xbfa20002,
+	0xbeff0080, 0xbfa00002,
+	0xbeff00c1, 0xbfa0000c,
+	0xbef600ff, 0x01000000,
+	0xc4068070, 0x008ce801,
+	0x00008000, 0xc4068070,
+	0x008ce802, 0x00010000,
+	0xc4068070, 0x008ce803,
+	0x00018000, 0xbfa0000b,
+	0xbef600ff, 0x01000000,
+	0xc4068070, 0x008ce801,
+	0x00010000, 0xc4068070,
+	0x008ce802, 0x00020000,
+	0xc4068070, 0x008ce803,
+	0x00030000, 0xb8f03b05,
+	0x80708170, 0xbf0d9973,
+	0xbfa20002, 0x84708970,
+	0xbfa00001, 0x84708a70,
+	0xb8fa1e06, 0x847a8a7a,
+	0x80707a70, 0x8070ff70,
+	0x00000200, 0xbef600ff,
+	0x01000000, 0x7e000280,
+	0x7e020280, 0x7e040280,
+	0xbe804ec2, 0xbf94fffe,
+	0xb8faf804, 0x8b7a847a,
+	0x91788478, 0x8c787a78,
+	0x917aff6d, 0x80000000,
+	0xd7610002, 0x00010071,
+	0xd7610002, 0x0001026c,
+	0xd7610002, 0x0001047a,
+	0xd7610002, 0x0001066e,
+	0xd7610002, 0x0001086f,
+	0xd7610002, 0x00010a78,
+	0xd7610002, 0x00010e7b,
+	0xd8500000, 0x00000000,
+	0xd8500000, 0x00000000,
+	0xd8500000, 0x00000000,
+	0xd8500000, 0x00000000,
+	0xd8500000, 0x00000000,
+	0xd8500000, 0x00000000,
+	0xd8500000, 0x00000000,
+	0xd8500000, 0x00000000,
+	0xb8faf811, 0xd7610002,
+	0x00010c7a, 0xb8faf801,
+	0xd7610002, 0x0001107a,
+	0xb8faf814, 0xd7610002,
+	0x0001127a, 0xb8faf815,
+	0xd7610002, 0x0001147a,
+	0xb8faf812, 0xd7610002,
+	0x0001167a, 0xb8faf813,
+	0xd7610002, 0x0001187a,
+	0xb8faf802, 0xd7610002,
+	0x00011a7a, 0xbefa50c1,
+	0xbfc70000, 0xd7610002,
+	0x00011c7a, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xbefe00ff,
+	0x0000ffff, 0xbeff0080,
+	0xc4068070, 0x008ce802,
+	0x00000000, 0xbefe00c1,
+	0xb8f03b05, 0x80708170,
+	0xbf0d9973, 0xbfa20002,
+	0x84708970, 0xbfa00001,
+	0x84708a70, 0xb8fa1e06,
+	0x847a8a7a, 0x80707a70,
+	0xbef600ff, 0x01000000,
+	0xbef90080, 0xbefd0080,
+	0xbf800000, 0xbe804100,
+	0xbe824102, 0xbe844104,
+	0xbe864106, 0xbe884108,
+	0xbe8a410a, 0xbe8c410c,
+	0xbe8e410e, 0xbf068079,
+	0xbfa10032, 0xd7610002,
+	0x00010000, 0xd7610002,
+	0x00010201, 0xd7610002,
+	0x00010402, 0xd7610002,
+	0x00010603, 0xd7610002,
+	0x00010804, 0xd7610002,
+	0x00010a05, 0xd7610002,
+	0x00010c06, 0xd7610002,
+	0x00010e07, 0xd7610002,
+	0x00011008, 0xd7610002,
+	0x00011209, 0xd7610002,
+	0x0001140a, 0xd7610002,
+	0x0001160b, 0xd7610002,
+	0x0001180c, 0xd7610002,
+	0x00011a0d, 0xd7610002,
+	0x00011c0e, 0xd7610002,
+	0x00011e0f, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0x80799079,
+	0xbfa00038, 0xd7610002,
+	0x00012000, 0xd7610002,
+	0x00012201, 0xd7610002,
+	0x00012402, 0xd7610002,
+	0x00012603, 0xd7610002,
+	0x00012804, 0xd7610002,
+	0x00012a05, 0xd7610002,
+	0x00012c06, 0xd7610002,
+	0x00012e07, 0xd7610002,
+	0x00013008, 0xd7610002,
+	0x00013209, 0xd7610002,
+	0x0001340a, 0xd7610002,
+	0x0001360b, 0xd7610002,
+	0x0001380c, 0xd7610002,
+	0x00013a0d, 0xd7610002,
+	0x00013c0e, 0xd7610002,
+	0x00013e0f, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0x80799079,
+	0xc4068070, 0x008ce802,
+	0x00000000, 0x8070ff70,
+	0x00000080, 0xbef90080,
+	0x7e040280, 0x807d907d,
+	0xbf0aff7d, 0x00000060,
+	0xbfa2ff88, 0xbe804100,
+	0xbe824102, 0xbe844104,
+	0xbe864106, 0xbe884108,
+	0xbe8a410a, 0xd7610002,
+	0x00010000, 0xd7610002,
+	0x00010201, 0xd7610002,
+	0x00010402, 0xd7610002,
+	0x00010603, 0xd7610002,
+	0x00010804, 0xd7610002,
+	0x00010a05, 0xd7610002,
+	0x00010c06, 0xd7610002,
+	0x00010e07, 0xd7610002,
+	0x00011008, 0xd7610002,
+	0x00011209, 0xd7610002,
+	0x0001140a, 0xd7610002,
+	0x0001160b, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xd8500000,
+	0x00000000, 0xc4068070,
+	0x008ce802, 0x00000000,
+	0xbefe00c1, 0x857d9973,
+	0x8b7d817d, 0xbf06817d,
+	0xbfa20002, 0xbeff0080,
+	0xbfa00001, 0xbeff00c1,
+	0xb8fb4306, 0x8b7bc17b,
+	0xbfa10044, 0x8b7aff6d,
+	0x80000000, 0xbfa10041,
+	0x847b897b, 0xbef6007b,
+	0xb8f03b05, 0x80708170,
+	0xbf0d9973, 0xbfa20002,
+	0x84708970, 0xbfa00001,
+	0x84708a70, 0xb8fa1e06,
+	0x847a8a7a, 0x80707a70,
+	0x8070ff70, 0x00000200,
+	0x8070ff70, 0x00000080,
+	0xbef600ff, 0x01000000,
+	0xd71f0000, 0x000100c1,
+	0xd7200000, 0x000200c1,
+	0x16000084, 0x857d9973,
+	0x8b7d817d, 0xbf06817d,
+	0xbefd0080, 0xbfa20013,
+	0xbe8300ff, 0x00000080,
+	0xbf800000, 0xbf800000,
+	0xbf800000, 0xd8d80000,
+	0x01000000, 0xbf8a0000,
+	0xc4068070, 0x008ce801,
+	0x00000000, 0x807d037d,
+	0x80700370, 0xd5250000,
+	0x0001ff00, 0x00000080,
+	0xbf0a7b7d, 0xbfa2fff3,
+	0xbfa00012, 0xbe8300ff,
+	0x00000100, 0xbf800000,
+	0xbf800000, 0xbf800000,
+	0xd8d80000, 0x01000000,
+	0xbf8a0000, 0xc4068070,
+	0x008ce801, 0x00000000,
+	0x807d037d, 0x80700370,
+	0xd5250000, 0x0001ff00,
+	0x00000100, 0xbf0a7b7d,
+	0xbfa2fff3, 0xbefe00c1,
+	0x857d9973, 0x8b7d817d,
+	0xbf06817d, 0xbfa20004,
+	0xbef000ff, 0x00000200,
+	0xbeff0080, 0xbfa00003,
+	0xbef000ff, 0x00000400,
+	0xbeff00c1, 0xb8fb3b05,
+	0x807b817b, 0x847b827b,
+	0x857d9973, 0x8b7d817d,
+	0xbf06817d, 0xbfa2001b,
+	0xbef600ff, 0x01000000,
+	0xbefd0084, 0xbf0a7b7d,
+	0xbfa10040, 0x7e008700,
+	0x7e028701, 0x7e048702,
+	0x7e068703, 0xc4068070,
+	0x008ce800, 0x00000000,
+	0xc4068070, 0x008ce801,
+	0x00008000, 0xc4068070,
+	0x008ce802, 0x00010000,
+	0xc4068070, 0x008ce803,
+	0x00018000, 0x807d847d,
+	0x8070ff70, 0x00000200,
+	0xbf0a7b7d, 0xbfa2ffeb,
+	0xbfa0002a, 0xbef600ff,
+	0x01000000, 0xbefd0084,
+	0xbf0a7b7d, 0xbfa10015,
+	0x7e008700, 0x7e028701,
+	0x7e048702, 0x7e068703,
+	0xc4068070, 0x008ce800,
+	0x00000000, 0xc4068070,
+	0x008ce801, 0x00010000,
+	0xc4068070, 0x008ce802,
+	0x00020000, 0xc4068070,
+	0x008ce803, 0x00030000,
+	0x807d847d, 0x8070ff70,
+	0x00000400, 0xbf0a7b7d,
+	0xbfa2ffeb, 0xb8fb1e06,
+	0x8b7bc17b, 0xbfa1000d,
+	0x847b837b, 0x807b7d7b,
+	0xbefe00c1, 0xbeff0080,
+	0x7e008700, 0xc4068070,
+	0x008ce800, 0x00000000,
+	0x807d817d, 0x8070ff70,
+	0x00000080, 0xbf0a7b7d,
+	0xbfa2fff7, 0xbfa0016e,
+	0xbef4007e, 0x8b75ff7f,
+	0x0000ffff, 0x8c75ff75,
+	0x00040000, 0xbef60080,
+	0xbef700ff, 0x10807fac,
+	0xbef1007f, 0xb8f20742,
+	0x84729972, 0x8b6eff7f,
+	0x04000000, 0xbfa1003b,
+	0xbefe00c1, 0x857d9972,
+	0x8b7d817d, 0xbf06817d,
+	0xbfa20002, 0xbeff0080,
+	0xbfa00001, 0xbeff00c1,
+	0xb8ef4306, 0x8b6fc16f,
+	0xbfa10030, 0x846f896f,
+	0xbef6006f, 0xb8f83b05,
+	0x80788178, 0xbf0d9972,
+	0xbfa20002, 0x84788978,
+	0xbfa00001, 0x84788a78,
+	0xb8ee1e06, 0x846e8a6e,
+	0x80786e78, 0x8078ff78,
+	0x00000200, 0x8078ff78,
+	0x00000080, 0xbef600ff,
+	0x01000000, 0x857d9972,
+	0x8b7d817d, 0xbf06817d,
+	0xbefd0080, 0xbfa2000d,
+	0xc4050078, 0x0080e800,
+	0x00000000, 0xbf8a0000,
+	0xdac00000, 0x00000000,
+	0x807dff7d, 0x00000080,
+	0x8078ff78, 0x00000080,
+	0xbf0a6f7d, 0xbfa2fff4,
+	0xbfa0000c, 0xc4050078,
+	0x0080e800, 0x00000000,
+	0xbf8a0000, 0xdac00000,
+	0x00000000, 0x807dff7d,
+	0x00000100, 0x8078ff78,
+	0x00000100, 0xbf0a6f7d,
+	0xbfa2fff4, 0xbef80080,
+	0xbefe00c1, 0x857d9972,
+	0x8b7d817d, 0xbf06817d,
+	0xbfa20002, 0xbeff0080,
+	0xbfa00001, 0xbeff00c1,
+	0xb8ef3b05, 0x806f816f,
+	0x846f826f, 0x857d9972,
+	0x8b7d817d, 0xbf06817d,
+	0xbfa2002c, 0xbef600ff,
+	0x01000000, 0xbeee0078,
+	0x8078ff78, 0x00000200,
+	0xbefd0084, 0xbf0a6f7d,
+	0xbfa10061, 0xc4050078,
+	0x008ce800, 0x00000000,
+	0xc4050078, 0x008ce801,
+	0x00008000, 0xc4050078,
+	0x008ce802, 0x00010000,
+	0xc4050078, 0x008ce803,
+	0x00018000, 0xbf8a0000,
+	0x7e008500, 0x7e028501,
+	0x7e048502, 0x7e068503,
+	0x807d847d, 0x8078ff78,
+	0x00000200, 0xbf0a6f7d,
+	0xbfa2ffea, 0xc405006e,
+	0x008ce800, 0x00000000,
+	0xc405006e, 0x008ce801,
+	0x00008000, 0xc405006e,
+	0x008ce802, 0x00010000,
+	0xc405006e, 0x008ce803,
+	0x00018000, 0xbf8a0000,
+	0xbfa0003d, 0xbef600ff,
+	0x01000000, 0xbeee0078,
+	0x8078ff78, 0x00000400,
+	0xbefd0084, 0xbf0a6f7d,
+	0xbfa10016, 0xc4050078,
+	0x008ce800, 0x00000000,
+	0xc4050078, 0x008ce801,
+	0x00010000, 0xc4050078,
+	0x008ce802, 0x00020000,
+	0xc4050078, 0x008ce803,
+	0x00030000, 0xbf8a0000,
+	0x7e008500, 0x7e028501,
+	0x7e048502, 0x7e068503,
+	0x807d847d, 0x8078ff78,
+	0x00000400, 0xbf0a6f7d,
+	0xbfa2ffea, 0xb8ef1e06,
+	0x8b6fc16f, 0xbfa1000f,
+	0x846f836f, 0x806f7d6f,
+	0xbefe00c1, 0xbeff0080,
+	0xc4050078, 0x008ce800,
+	0x00000000, 0xbf8a0000,
+	0x7e008500, 0x807d817d,
+	0x8078ff78, 0x00000080,
+	0xbf0a6f7d, 0xbfa2fff6,
+	0xbeff00c1, 0xc405006e,
+	0x008ce800, 0x00000000,
+	0xc405006e, 0x008ce801,
+	0x00010000, 0xc405006e,
+	0x008ce802, 0x00020000,
+	0xc405006e, 0x008ce803,
+	0x00030000, 0xbf8a0000,
+	0xb8f83b05, 0x80788178,
+	0xbf0d9972, 0xbfa20002,
+	0x84788978, 0xbfa00001,
+	0x84788a78, 0xb8ee1e06,
+	0x846e8a6e, 0x80786e78,
+	0x8078ff78, 0x00000200,
+	0x80f8ff78, 0x00000050,
+	0xbef600ff, 0x01000000,
+	0xbefd00ff, 0x0000006c,
+	0x80f89078, 0xf462403a,
+	0xf0000000, 0xbf8a0000,
+	0x80fd847d, 0xbf800000,
+	0xbe804300, 0xbe824302,
+	0x80f8a078, 0xf462603a,
+	0xf0000000, 0xbf8a0000,
+	0x80fd887d, 0xbf800000,
+	0xbe804300, 0xbe824302,
+	0xbe844304, 0xbe864306,
+	0x80f8c078, 0xf462803a,
+	0xf0000000, 0xbf8a0000,
+	0x80fd907d, 0xbf800000,
+	0xbe804300, 0xbe824302,
+	0xbe844304, 0xbe864306,
+	0xbe884308, 0xbe8a430a,
+	0xbe8c430c, 0xbe8e430e,
+	0xbf06807d, 0xbfa1fff0,
+	0xb980f801, 0x00000000,
+	0xb8f83b05, 0x80788178,
+	0xbf0d9972, 0xbfa20002,
+	0x84788978, 0xbfa00001,
+	0x84788a78, 0xb8ee1e06,
+	0x846e8a6e, 0x80786e78,
+	0x8078ff78, 0x00000200,
+	0xbef600ff, 0x01000000,
+	0xbeff0071, 0xf4621bfa,
+	0xf0000000, 0x80788478,
+	0xf4621b3a, 0xf0000000,
+	0x80788478, 0xf4621b7a,
+	0xf0000000, 0x80788478,
+	0xf4621c3a, 0xf0000000,
+	0x80788478, 0xf4621c7a,
+	0xf0000000, 0x80788478,
+	0xf4621eba, 0xf0000000,
+	0x80788478, 0xf4621efa,
+	0xf0000000, 0x80788478,
+	0xf4621e7a, 0xf0000000,
+	0x80788478, 0xf4621cfa,
+	0xf0000000, 0x80788478,
+	0xf4621bba, 0xf0000000,
+	0x80788478, 0xbf8a0000,
+	0xb96ef814, 0xf4621bba,
+	0xf0000000, 0x80788478,
+	0xbf8a0000, 0xb96ef815,
+	0xf4621bba, 0xf0000000,
+	0x80788478, 0xbf8a0000,
+	0xb96ef812, 0xf4621bba,
+	0xf0000000, 0x80788478,
+	0xbf8a0000, 0xb96ef813,
+	0x8b6eff7f, 0x04000000,
+	0xbfa1000d, 0x80788478,
+	0xf4621bba, 0xf0000000,
+	0x80788478, 0xbf8a0000,
+	0xbf0d806e, 0xbfa10006,
+	0x856e906e, 0x8b6e6e6e,
+	0xbfa10003, 0xbe804ec1,
+	0x816ec16e, 0xbfa0fffb,
+	0xbefd006f, 0xbefe0070,
+	0xbeff0071, 0xb97b2011,
+	0x857b867b, 0xb97b0191,
+	0x857b827b, 0xb97bba11,
+	0xb973f801, 0xb8ee3b05,
+	0x806e816e, 0xbf0d9972,
+	0xbfa20002, 0x846e896e,
+	0xbfa00001, 0x846e8a6e,
+	0xb8ef1e06, 0x846f8a6f,
+	0x806e6f6e, 0x806eff6e,
+	0x00000200, 0x806e746e,
+	0x826f8075, 0x8b6fff6f,
+	0x0000ffff, 0xf4605c37,
+	0xf8000050, 0xf4605d37,
+	0xf8000060, 0xf4601e77,
+	0xf8000074, 0xbf8a0000,
+	0x8b6dff6d, 0x0000ffff,
+	0x8bfe7e7e, 0x8bea6a6a,
+	0xb97af804, 0xbe804ec2,
+	0xbf94fffe, 0xbe804a6c,
+	0xbe804ec2, 0xbf94fffe,
+	0xbfb10000, 0xbf9f0000,
+	0xbf9f0000, 0xbf9f0000,
+	0xbf9f0000, 0xbf9f0000,
+};
+
+static const uint32_t cwsr_trap_gfx9_5_0_hex[] = {
+	0xbf820001, 0xbf8202ca,
+	0xb8f8f802, 0x8978ff78,
+	0x00020006, 0xb8fbf803,
+	0x866eff78, 0x00002000,
+	0xbf840009, 0x866eff6d,
+	0x00ff0000, 0xbf85001a,
+	0x866eff7b, 0x00000400,
+	0xbf850051, 0xbf8e0010,
+	0xb8fbf803, 0xbf82fffa,
+	0x866eff7b, 0x03c00900,
+	0xbf850011, 0x866eff7b,
+	0x000071ff, 0xbf840008,
+	0x866fff7b, 0x00007080,
+	0xbf840001, 0xbeee1a87,
+	0xb8eff801, 0x8e6e8c6e,
+	0x866e6f6e, 0xbf850006,
+	0x866eff6d, 0x00ff0000,
+	0xbf850003, 0x866eff7b,
+	0x00000400, 0xbf85003a,
+	0xb8faf807, 0x867aff7a,
+	0x001f8000, 0x8e7a8b7a,
+	0x8979ff79, 0xfc000000,
+	0x87797a79, 0xba7ff807,
+	0x00000000, 0xb8faf812,
+	0xb8fbf813, 0x8efa887a,
+	0xbf0d8f7b, 0xbf840002,
+	0x877bff7b, 0xffff0000,
+	0xc0031bbd, 0x00000010,
+	0xbf8cc07f, 0x8e6e976e,
+	0x8979ff79, 0x00800000,
+	0x87796e79, 0xc0071bbd,
+	0x00000000, 0xbf8cc07f,
+	0xc0071ebd, 0x00000008,
+	0xbf8cc07f, 0x86ee6e6e,
+	0xbf840001, 0xbe801d6e,
+	0x866eff6d, 0x01ff0000,
+	0xbf850005, 0x8778ff78,
+	0x00002000, 0x80ec886c,
+	0x82ed806d, 0xbf820005,
+	0x866eff6d, 0x01000000,
+	0xbf850002, 0x806c846c,
+	0x826d806d, 0x866dff6d,
+	0x0000ffff, 0x8f7a8b79,
+	0x867aff7a, 0x001f8000,
+	0xb97af807, 0x86fe7e7e,
+	0x86ea6a6a, 0x8f6e8378,
+	0xb96ee0c2, 0xbf800002,
+	0xb9780002, 0xbe801f6c,
+	0x866dff6d, 0x0000ffff,
+	0xbefa0080, 0xb97a0283,
+	0xb8faf807, 0x867aff7a,
+	0x001f8000, 0x8e7a8b7a,
+	0x8979ff79, 0xfc000000,
+	0x87797a79, 0xba7ff807,
+	0x00000000, 0xbeee007e,
+	0xbeef007f, 0xbefe0180,
+	0xbf900004, 0x877a8478,
+	0xb97af802, 0xbf8e0002,
+	0xbf88fffe, 0xb8fa2985,
+	0x807a817a, 0x8e7a8a7a,
+	0x8e7a817a, 0xb8fb1605,
+	0x807b817b, 0x8e7b867b,
+	0x807a7b7a, 0x807a7e7a,
+	0x827b807f, 0x867bff7b,
+	0x0000ffff, 0xc04b1c3d,
+	0x00000050, 0xbf8cc07f,
+	0xc04b1d3d, 0x00000060,
+	0xbf8cc07f, 0xc0431e7d,
+	0x00000074, 0xbf8cc07f,
+	0xbef4007e, 0x8675ff7f,
+	0x0000ffff, 0x8775ff75,
+	0x00040000, 0xbef60080,
+	0xbef700ff, 0x00807fac,
+	0xbef1007c, 0xbef00080,
+	0xb8f02985, 0x80708170,
+	0x8e708a70, 0x8e708170,
+	0xb8fa1605, 0x807a817a,
+	0x8e7a867a, 0x80707a70,
+	0xbef60084, 0xbef600ff,
+	0x01000000, 0xbefe007c,
+	0xbefc0070, 0xc0611c7a,
+	0x0000007c, 0xbf8cc07f,
+	0x80708470, 0xbefc007e,
+	0xbefe007c, 0xbefc0070,
+	0xc0611b3a, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xbefe007c,
+	0xbefc0070, 0xc0611b7a,
+	0x0000007c, 0xbf8cc07f,
+	0x80708470, 0xbefc007e,
+	0xbefe007c, 0xbefc0070,
+	0xc0611bba, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xbefe007c,
+	0xbefc0070, 0xc0611bfa,
+	0x0000007c, 0xbf8cc07f,
+	0x80708470, 0xbefc007e,
+	0xbefe007c, 0xbefc0070,
+	0xc0611e3a, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xb8fbf803,
+	0xbefe007c, 0xbefc0070,
+	0xc0611efa, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xbefe007c,
+	0xbefc0070, 0xc0611a3a,
+	0x0000007c, 0xbf8cc07f,
+	0x80708470, 0xbefc007e,
+	0xbefe007c, 0xbefc0070,
+	0xc0611a7a, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xb8f1f801,
+	0xbefe007c, 0xbefc0070,
+	0xc0611c7a, 0x0000007c,
+	0xbf8cc07f, 0x80708470,
+	0xbefc007e, 0xbf108080,
+	0x867aff7f, 0x04000000,
+	0xbeef0080, 0x876f6f7a,
+	0xb8f02985, 0x80708170,
+	0x8e708a70, 0x8e708170,
+	0xb8fb1605, 0x807b817b,
+	0x8e7b847b, 0x8e76827b,
+	0xbef600ff, 0x01000000,
+	0xbef20174, 0x80747074,
+	0x82758075, 0xbefc0080,
+	0xbf800000, 0xbe802b00,
+	0xbe822b02, 0xbe842b04,
+	0xbe862b06, 0xbe882b08,
+	0xbe8a2b0a, 0xbe8c2b0c,
+	0xbe8e2b0e, 0xc06b003a,
+	0x00000000, 0xbf8cc07f,
+	0xc06b013a, 0x00000010,
+	0xbf8cc07f, 0xc06b023a,
+	0x00000020, 0xbf8cc07f,
+	0xc06b033a, 0x00000030,
+	0xbf8cc07f, 0x8074c074,
+	0x82758075, 0x807c907c,
+	0xbf0a7b7c, 0xbf85ffe7,
+	0xbef40172, 0xbef00080,
+	0xbefe00c1, 0xbeff00c1,
+	0xbee80080, 0xbee90080,
+	0xbef600ff, 0x01000000,
+	0x867aff78, 0x00400000,
+	0xbf850003, 0xb8faf803,
+	0x897a7aff, 0x10000000,
+	0xbf85004d, 0xbe840080,
+	0xd2890000, 0x00000900,
+	0x80048104, 0xd2890001,
+	0x00000900, 0x80048104,
+	0xd2890002, 0x00000900,
+	0x80048104, 0xd2890003,
+	0x00000900, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000901, 0x80048104,
+	0xd2890001, 0x00000901,
+	0x80048104, 0xd2890002,
+	0x00000901, 0x80048104,
+	0xd2890003, 0x00000901,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000902,
+	0x80048104, 0xd2890001,
+	0x00000902, 0x80048104,
+	0xd2890002, 0x00000902,
+	0x80048104, 0xd2890003,
+	0x00000902, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000903, 0x80048104,
+	0xd2890001, 0x00000903,
+	0x80048104, 0xd2890002,
+	0x00000903, 0x80048104,
+	0xd2890003, 0x00000903,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbf820008,
+	0xe0724000, 0x701d0000,
+	0xe0724100, 0x701d0100,
+	0xe0724200, 0x701d0200,
+	0xe0724300, 0x701d0300,
+	0xbefe00c1, 0xbeff00c1,
+	0xb8fb5306, 0x867bc17b,
+	0xbf840052, 0xbf8a0000,
+	0x867aff6f, 0x04000000,
+	0xbf84004e, 0x8e7b867b,
+	0x8e7b827b, 0xbef6007b,
+	0xb8f02985, 0x80708170,
+	0x8e708a70, 0x8e708170,
+	0xb8fa1605, 0x807a817a,
+	0x8e7a867a, 0x80707a70,
+	0x8070ff70, 0x00000080,
+	0xbef600ff, 0x01000000,
+	0xbefc0080, 0xd28c0002,
+	0x000100c1, 0xd28d0003,
+	0x000204c1, 0x867aff78,
+	0x00400000, 0xbf850003,
+	0xb8faf803, 0x897a7aff,
+	0x10000000, 0xbf85001d,
+	0x24040682, 0xd86c0000,
+	0x00000002, 0xbf8cc07f,
+	0xbe840080, 0xd2890000,
+	0x00000900, 0x80048104,
+	0xd2890001, 0x00000900,
+	0x80048104, 0xd2890002,
+	0x00000900, 0x80048104,
+	0xd2890003, 0x00000900,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0x680404ff,
+	0x00000100, 0xd0c9006a,
+	0x0000f702, 0xbf87ffe5,
+	0xbf820016, 0xd1060002,
+	0x00011103, 0x7e0602ff,
+	0x00000200, 0xbefc00ff,
+	0x00010000, 0xbe800077,
+	0x8677ff77, 0xff7fffff,
+	0x8777ff77, 0x00058000,
+	0xd8ec0000, 0x00000002,
+	0xbf8cc07f, 0xe0765000,
+	0x701d0002, 0x68040702,
+	0xd0c9006a, 0x0000f702,
+	0xbefe016a, 0xbf87fff6,
+	0xbef70000, 0xbef000ff,
+	0x00000400, 0xbefe00c1,
+	0xbeff00c1, 0xb8fb2b05,
+	0x807b817b, 0x8e7b827b,
+	0xbef600ff, 0x01000000,
+	0xbefc0084, 0xbf0a7b7c,
+	0xbf84006d, 0xbf11017c,
+	0x807bff7b, 0x00001000,
+	0x867aff78, 0x00400000,
+	0xbf850003, 0xb8faf803,
+	0x897a7aff, 0x10000000,
+	0xbf850051, 0xbe840080,
+	0xd2890000, 0x00000900,
+	0x80048104, 0xd2890001,
+	0x00000900, 0x80048104,
+	0xd2890002, 0x00000900,
+	0x80048104, 0xd2890003,
+	0x00000900, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000901, 0x80048104,
+	0xd2890001, 0x00000901,
+	0x80048104, 0xd2890002,
+	0x00000901, 0x80048104,
+	0xd2890003, 0x00000901,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000902,
+	0x80048104, 0xd2890001,
+	0x00000902, 0x80048104,
+	0xd2890002, 0x00000902,
+	0x80048104, 0xd2890003,
+	0x00000902, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000903, 0x80048104,
+	0xd2890001, 0x00000903,
+	0x80048104, 0xd2890002,
+	0x00000903, 0x80048104,
+	0xd2890003, 0x00000903,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0x807c847c,
+	0xbf0a7b7c, 0xbf85ffb1,
+	0xbf9c0000, 0xbf820012,
+	0x7e000300, 0x7e020301,
+	0x7e040302, 0x7e060303,
+	0xe0724000, 0x701d0000,
+	0xe0724100, 0x701d0100,
+	0xe0724200, 0x701d0200,
+	0xe0724300, 0x701d0300,
+	0x807c847c, 0x8070ff70,
+	0x00000400, 0xbf0a7b7c,
+	0xbf85ffef, 0xbf9c0000,
+	0xb8fb2985, 0x807b817b,
+	0x8e7b837b, 0xb8fa2b05,
+	0x807a817a, 0x8e7a827a,
+	0x80fb7a7b, 0x867b7b7b,
+	0xbf84007a, 0x807bff7b,
+	0x00001000, 0xbefc0080,
+	0xbf11017c, 0x867aff78,
+	0x00400000, 0xbf850003,
+	0xb8faf803, 0x897a7aff,
+	0x10000000, 0xbf850059,
+	0xd3d84000, 0x18000100,
+	0xd3d84001, 0x18000101,
+	0xd3d84002, 0x18000102,
+	0xd3d84003, 0x18000103,
+	0xbe840080, 0xd2890000,
+	0x00000900, 0x80048104,
+	0xd2890001, 0x00000900,
+	0x80048104, 0xd2890002,
+	0x00000900, 0x80048104,
+	0xd2890003, 0x00000900,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000901,
+	0x80048104, 0xd2890001,
+	0x00000901, 0x80048104,
+	0xd2890002, 0x00000901,
+	0x80048104, 0xd2890003,
+	0x00000901, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000902, 0x80048104,
+	0xd2890001, 0x00000902,
+	0x80048104, 0xd2890002,
+	0x00000902, 0x80048104,
+	0xd2890003, 0x00000902,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000903,
+	0x80048104, 0xd2890001,
+	0x00000903, 0x80048104,
+	0xd2890002, 0x00000903,
+	0x80048104, 0xd2890003,
+	0x00000903, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0x807c847c, 0xbf0a7b7c,
+	0xbf85ffa9, 0xbf9c0000,
+	0xbf820016, 0xd3d84000,
+	0x18000100, 0xd3d84001,
+	0x18000101, 0xd3d84002,
+	0x18000102, 0xd3d84003,
+	0x18000103, 0xe0724000,
+	0x701d0000, 0xe0724100,
+	0x701d0100, 0xe0724200,
+	0x701d0200, 0xe0724300,
+	0x701d0300, 0x807c847c,
+	0x8070ff70, 0x00000400,
+	0xbf0a7b7c, 0xbf85ffeb,
+	0xbf9c0000, 0xbf8200f4,
+	0xbef4007e, 0x8675ff7f,
+	0x0000ffff, 0x8775ff75,
+	0x00040000, 0xbef60080,
+	0xbef700ff, 0x00807fac,
+	0x866eff7f, 0x04000000,
+	0xbf840025, 0xbefe00c1,
+	0xbeff00c1, 0xb8ef5306,
+	0x866fc16f, 0xbf840020,
+	0x8e6f866f, 0x8e6f826f,
+	0xbef6006f, 0xb8f82985,
+	0x80788178, 0x8e788a78,
+	0x8e788178, 0xb8ee1605,
+	0x806e816e, 0x8e6e866e,
+	0x80786e78, 0x8078ff78,
+	0x00000080, 0xbef600ff,
+	0x01000000, 0xbefc0080,
+	0xe0510000, 0x781d0000,
+	0xe0510100, 0x781d0000,
+	0xe0510200, 0x781d0000,
+	0xe0510300, 0x781d0000,
+	0xe0510400, 0x781d0000,
+	0x807cff7c, 0x00000500,
+	0x8078ff78, 0x00000500,
+	0xbf0a6f7c, 0xbf85fff0,
+	0xbefe00c1, 0xbeff00c1,
+	0xbef600ff, 0x01000000,
+	0xb8ef2b05, 0x806f816f,
+	0x8e6f826f, 0x806fff6f,
+	0x00008000, 0xbef80080,
+	0xbeee0078, 0x8078ff78,
+	0x00000400, 0xbefc0084,
+	0xbf11087c, 0xe0524000,
+	0x781d0000, 0xe0524100,
+	0x781d0100, 0xe0524200,
+	0x781d0200, 0xe0524300,
+	0x781d0300, 0xbf8c0f70,
+	0x7e000300, 0x7e020301,
+	0x7e040302, 0x7e060303,
+	0x807c847c, 0x8078ff78,
+	0x00000400, 0xbf0a6f7c,
+	0xbf85ffee, 0xb8ef2985,
+	0x806f816f, 0x8e6f836f,
+	0xb8f92b05, 0x80798179,
+	0x8e798279, 0x80ef796f,
+	0x866f6f6f, 0xbf84001a,
+	0x806fff6f, 0x00008000,
+	0xbefc0080, 0xbf11087c,
+	0xe0524000, 0x781d0000,
+	0xe0524100, 0x781d0100,
+	0xe0524200, 0x781d0200,
+	0xe0524300, 0x781d0300,
+	0xbf8c0f70, 0xd3d94000,
+	0x18000100, 0xd3d94001,
+	0x18000101, 0xd3d94002,
+	0x18000102, 0xd3d94003,
+	0x18000103, 0x807c847c,
+	0x8078ff78, 0x00000400,
+	0xbf0a6f7c, 0xbf85ffea,
+	0xbf9c0000, 0xe0524000,
+	0x6e1d0000, 0xe0524100,
+	0x6e1d0100, 0xe0524200,
+	0x6e1d0200, 0xe0524300,
+	0x6e1d0300, 0xbf8c0f70,
+	0xb8f82985, 0x80788178,
+	0x8e788a78, 0x8e788178,
+	0xb8ee1605, 0x806e816e,
+	0x8e6e866e, 0x80786e78,
+	0x80f8c078, 0xb8ef1605,
+	0x806f816f, 0x8e6f846f,
+	0x8e76826f, 0xbef600ff,
+	0x01000000, 0xbefc006f,
+	0xc031003a, 0x00000078,
+	0x80f8c078, 0xbf8cc07f,
+	0x80fc907c, 0xbf800000,
+	0xbe802d00, 0xbe822d02,
+	0xbe842d04, 0xbe862d06,
+	0xbe882d08, 0xbe8a2d0a,
+	0xbe8c2d0c, 0xbe8e2d0e,
+	0xbf06807c, 0xbf84fff0,
+	0xb8f82985, 0x80788178,
+	0x8e788a78, 0x8e788178,
+	0xb8ee1605, 0x806e816e,
+	0x8e6e866e, 0x80786e78,
+	0xbef60084, 0xbef600ff,
+	0x01000000, 0xc0211bfa,
+	0x00000078, 0x80788478,
+	0xc0211b3a, 0x00000078,
+	0x80788478, 0xc0211b7a,
+	0x00000078, 0x80788478,
+	0xc0211c3a, 0x00000078,
+	0x80788478, 0xc0211c7a,
+	0x00000078, 0x80788478,
+	0xc0211eba, 0x00000078,
+	0x80788478, 0xc0211efa,
+	0x00000078, 0x80788478,
+	0xc0211a3a, 0x00000078,
+	0x80788478, 0xc0211a7a,
+	0x00000078, 0x80788478,
+	0xc0211cfa, 0x00000078,
+	0x80788478, 0xbf8cc07f,
+	0xbefc006f, 0xbefe0070,
+	0xbeff0071, 0x866f7bff,
+	0x000003ff, 0xb96f4803,
+	0x866f7bff, 0xfffff800,
+	0x8f6f8b6f, 0xb96fa2c3,
+	0xb973f801, 0xb8ee2985,
+	0x806e816e, 0x8e6e8a6e,
+	0x8e6e816e, 0xb8ef1605,
+	0x806f816f, 0x8e6f866f,
+	0x806e6f6e, 0x806e746e,
+	0x826f8075, 0x866fff6f,
+	0x0000ffff, 0xc00b1c37,
+	0x00000050, 0xc00b1d37,
+	0x00000060, 0xc0031e77,
+	0x00000074, 0xbf8cc07f,
+	0x8f6e8b79, 0x866eff6e,
+	0x001f8000, 0xb96ef807,
+	0x866dff6d, 0x0000ffff,
+	0x86fe7e7e, 0x86ea6a6a,
+	0x8f6e837a, 0xb96ee0c2,
+	0xbf800002, 0xb97a0002,
+	0xbf8a0000, 0xbe801f6c,
+	0xbf9b0000, 0x00000000,
+};
diff --git a/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx10.asm b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx10.asm
new file mode 100644
index 000000000000..96fbb16ceb21
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx10.asm
@@ -0,0 +1,1434 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/* To compile this assembly code:
+ *
+ * Navi1x:
+ *   cpp -DASIC_FAMILY=CHIP_NAVI10 cwsr_trap_handler_gfx10.asm -P -o nv1x.sp3
+ *   sp3 nv1x.sp3 -hex nv1x.hex
+ *
+ * gfx10:
+ *   cpp -DASIC_FAMILY=CHIP_SIENNA_CICHLID cwsr_trap_handler_gfx10.asm -P -o gfx10.sp3
+ *   sp3 gfx10.sp3 -hex gfx10.hex
+ *
+ * gfx11:
+ *   cpp -DASIC_FAMILY=CHIP_PLUM_BONITO cwsr_trap_handler_gfx10.asm -P -o gfx11.sp3
+ *   sp3 gfx11.sp3 -hex gfx11.hex
+ *
+ */
+
+#define CHIP_NAVI10 26
+#define CHIP_SIENNA_CICHLID 30
+#define CHIP_PLUM_BONITO 36
+
+#define NO_SQC_STORE (ASIC_FAMILY >= CHIP_SIENNA_CICHLID)
+#define HAVE_XNACK (ASIC_FAMILY < CHIP_SIENNA_CICHLID)
+#define HAVE_SENDMSG_RTN (ASIC_FAMILY >= CHIP_PLUM_BONITO)
+#define HAVE_BUFFER_LDS_LOAD (ASIC_FAMILY < CHIP_PLUM_BONITO)
+#define SW_SA_TRAP (ASIC_FAMILY == CHIP_PLUM_BONITO)
+#define SAVE_AFTER_XNACK_ERROR (HAVE_XNACK && !NO_SQC_STORE) // workaround for TCP store failure after XNACK error when ALLOW_REPLAY=0, for debugger
+#define SINGLE_STEP_MISSED_WORKAROUND 1	//workaround for lost MODE.DEBUG_EN exception when SAVECTX raised
+
+#define S_COHERENCE glc:1
+#define V_COHERENCE slc:1 glc:1
+#define S_WAITCNT_0 s_waitcnt 0
+
+var SQ_WAVE_STATUS_SPI_PRIO_MASK		= 0x00000006
+var SQ_WAVE_STATUS_HALT_MASK			= 0x2000
+var SQ_WAVE_STATUS_ECC_ERR_MASK			= 0x20000
+var SQ_WAVE_STATUS_TRAP_EN_SHIFT		= 6
+var SQ_WAVE_IB_STS2_WAVE64_SHIFT		= 11
+var SQ_WAVE_IB_STS2_WAVE64_SIZE			= 1
+var SQ_WAVE_LDS_ALLOC_GRANULARITY		= 8
+var S_STATUS_HWREG				= HW_REG_STATUS
+var S_STATUS_ALWAYS_CLEAR_MASK			= SQ_WAVE_STATUS_SPI_PRIO_MASK|SQ_WAVE_STATUS_ECC_ERR_MASK
+var S_STATUS_HALT_MASK				= SQ_WAVE_STATUS_HALT_MASK
+var S_SAVE_PC_HI_TRAP_ID_MASK			= 0x00FF0000
+var S_SAVE_PC_HI_HT_MASK			= 0x01000000
+
+var SQ_WAVE_STATUS_NO_VGPRS_SHIFT		= 24
+var SQ_WAVE_LDS_ALLOC_LDS_SIZE_SHIFT		= 12
+var SQ_WAVE_LDS_ALLOC_LDS_SIZE_SIZE		= 9
+var SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SIZE		= 8
+var SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SHIFT	= 24
+var SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SIZE	= 4
+
+#if ASIC_FAMILY < CHIP_PLUM_BONITO
+var SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT		= 8
+#else
+var SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT		= 12
+#endif
+
+var SQ_WAVE_TRAPSTS_SAVECTX_MASK		= 0x400
+var SQ_WAVE_TRAPSTS_EXCP_MASK			= 0x1FF
+var SQ_WAVE_TRAPSTS_SAVECTX_SHIFT		= 10
+var SQ_WAVE_TRAPSTS_ADDR_WATCH_MASK		= 0x80
+var SQ_WAVE_TRAPSTS_ADDR_WATCH_SHIFT		= 7
+var SQ_WAVE_TRAPSTS_MEM_VIOL_MASK		= 0x100
+var SQ_WAVE_TRAPSTS_MEM_VIOL_SHIFT		= 8
+var SQ_WAVE_TRAPSTS_ILLEGAL_INST_MASK		= 0x800
+var SQ_WAVE_TRAPSTS_ILLEGAL_INST_SHIFT		= 11
+var SQ_WAVE_TRAPSTS_EXCP_HI_MASK		= 0x7000
+#if ASIC_FAMILY >= CHIP_PLUM_BONITO
+var SQ_WAVE_TRAPSTS_HOST_TRAP_SHIFT		= 16
+var SQ_WAVE_TRAPSTS_WAVE_START_MASK		= 0x20000
+var SQ_WAVE_TRAPSTS_WAVE_START_SHIFT		= 17
+var SQ_WAVE_TRAPSTS_WAVE_END_MASK		= 0x40000
+var SQ_WAVE_TRAPSTS_TRAP_AFTER_INST_MASK	= 0x100000
+#endif
+var SQ_WAVE_TRAPSTS_XNACK_ERROR_MASK		= 0x10000000
+
+var SQ_WAVE_MODE_EXCP_EN_SHIFT			= 12
+var SQ_WAVE_MODE_EXCP_EN_ADDR_WATCH_SHIFT	= 19
+
+var SQ_WAVE_IB_STS_FIRST_REPLAY_SHIFT		= 15
+var SQ_WAVE_IB_STS_REPLAY_W64H_SHIFT		= 25
+var SQ_WAVE_IB_STS_REPLAY_W64H_MASK		= 0x02000000
+var SQ_WAVE_IB_STS_RCNT_FIRST_REPLAY_MASK	= 0x003F8000
+
+var SQ_WAVE_MODE_DEBUG_EN_MASK			= 0x800
+
+var S_TRAPSTS_RESTORE_PART_1_SIZE		= SQ_WAVE_TRAPSTS_SAVECTX_SHIFT
+var S_TRAPSTS_RESTORE_PART_2_SHIFT		= SQ_WAVE_TRAPSTS_ILLEGAL_INST_SHIFT
+
+#if ASIC_FAMILY < CHIP_PLUM_BONITO
+var S_TRAPSTS_NON_MASKABLE_EXCP_MASK		= SQ_WAVE_TRAPSTS_MEM_VIOL_MASK|SQ_WAVE_TRAPSTS_ILLEGAL_INST_MASK
+var S_TRAPSTS_RESTORE_PART_2_SIZE		= 32 - S_TRAPSTS_RESTORE_PART_2_SHIFT
+var S_TRAPSTS_RESTORE_PART_3_SHIFT		= 0
+var S_TRAPSTS_RESTORE_PART_3_SIZE		= 0
+#else
+var S_TRAPSTS_NON_MASKABLE_EXCP_MASK		= SQ_WAVE_TRAPSTS_MEM_VIOL_MASK		|\
+						  SQ_WAVE_TRAPSTS_ILLEGAL_INST_MASK	|\
+						  SQ_WAVE_TRAPSTS_WAVE_START_MASK	|\
+						  SQ_WAVE_TRAPSTS_WAVE_END_MASK		|\
+						  SQ_WAVE_TRAPSTS_TRAP_AFTER_INST_MASK
+var S_TRAPSTS_RESTORE_PART_2_SIZE		= SQ_WAVE_TRAPSTS_HOST_TRAP_SHIFT - SQ_WAVE_TRAPSTS_ILLEGAL_INST_SHIFT
+var S_TRAPSTS_RESTORE_PART_3_SHIFT		= SQ_WAVE_TRAPSTS_WAVE_START_SHIFT
+var S_TRAPSTS_RESTORE_PART_3_SIZE		= 32 - S_TRAPSTS_RESTORE_PART_3_SHIFT
+#endif
+var S_TRAPSTS_HWREG				= HW_REG_TRAPSTS
+var S_TRAPSTS_SAVE_CONTEXT_MASK			= SQ_WAVE_TRAPSTS_SAVECTX_MASK
+var S_TRAPSTS_SAVE_CONTEXT_SHIFT		= SQ_WAVE_TRAPSTS_SAVECTX_SHIFT
+
+// bits [31:24] unused by SPI debug data
+var TTMP11_SAVE_REPLAY_W64H_SHIFT		= 31
+var TTMP11_SAVE_REPLAY_W64H_MASK		= 0x80000000
+var TTMP11_SAVE_RCNT_FIRST_REPLAY_SHIFT		= 24
+var TTMP11_SAVE_RCNT_FIRST_REPLAY_MASK		= 0x7F000000
+var TTMP11_DEBUG_TRAP_ENABLED_SHIFT		= 23
+var TTMP11_DEBUG_TRAP_ENABLED_MASK		= 0x800000
+
+// SQ_SEL_X/Y/Z/W, BUF_NUM_FORMAT_FLOAT, (0 for MUBUF stride[17:14]
+// when ADD_TID_ENABLE and BUF_DATA_FORMAT_32 for MTBUF), ADD_TID_ENABLE
+var S_SAVE_BUF_RSRC_WORD1_STRIDE		= 0x00040000
+var S_SAVE_BUF_RSRC_WORD3_MISC			= 0x10807FAC
+var S_SAVE_SPI_INIT_FIRST_WAVE_MASK		= 0x04000000
+var S_SAVE_SPI_INIT_FIRST_WAVE_SHIFT		= 26
+
+var S_SAVE_PC_HI_FIRST_WAVE_MASK		= 0x80000000
+var S_SAVE_PC_HI_FIRST_WAVE_SHIFT		= 31
+
+var s_sgpr_save_num				= 108
+
+var s_save_spi_init_lo				= exec_lo
+var s_save_spi_init_hi				= exec_hi
+var s_save_pc_lo				= ttmp0
+var s_save_pc_hi				= ttmp1
+var s_save_exec_lo				= ttmp2
+var s_save_exec_hi				= ttmp3
+var s_save_status				= ttmp12
+var s_save_trapsts				= ttmp15
+var s_save_xnack_mask				= s_save_trapsts
+var s_wave_size					= ttmp7
+var s_save_buf_rsrc0				= ttmp8
+var s_save_buf_rsrc1				= ttmp9
+var s_save_buf_rsrc2				= ttmp10
+var s_save_buf_rsrc3				= ttmp11
+var s_save_mem_offset				= ttmp4
+var s_save_alloc_size				= s_save_trapsts
+var s_save_tmp					= ttmp14
+var s_save_m0					= ttmp5
+var s_save_ttmps_lo				= s_save_tmp
+var s_save_ttmps_hi				= s_save_trapsts
+
+var S_RESTORE_BUF_RSRC_WORD1_STRIDE		= S_SAVE_BUF_RSRC_WORD1_STRIDE
+var S_RESTORE_BUF_RSRC_WORD3_MISC		= S_SAVE_BUF_RSRC_WORD3_MISC
+
+var S_RESTORE_SPI_INIT_FIRST_WAVE_MASK		= 0x04000000
+var S_RESTORE_SPI_INIT_FIRST_WAVE_SHIFT		= 26
+var S_WAVE_SIZE					= 25
+
+var s_restore_spi_init_lo			= exec_lo
+var s_restore_spi_init_hi			= exec_hi
+var s_restore_mem_offset			= ttmp12
+var s_restore_alloc_size			= ttmp3
+var s_restore_tmp				= ttmp2
+var s_restore_mem_offset_save			= s_restore_tmp
+var s_restore_m0				= s_restore_alloc_size
+var s_restore_mode				= ttmp7
+var s_restore_flat_scratch			= s_restore_tmp
+var s_restore_pc_lo				= ttmp0
+var s_restore_pc_hi				= ttmp1
+var s_restore_exec_lo				= ttmp4
+var s_restore_exec_hi				= ttmp5
+var s_restore_status				= ttmp14
+var s_restore_trapsts				= ttmp15
+var s_restore_xnack_mask			= ttmp13
+var s_restore_buf_rsrc0				= ttmp8
+var s_restore_buf_rsrc1				= ttmp9
+var s_restore_buf_rsrc2				= ttmp10
+var s_restore_buf_rsrc3				= ttmp11
+var s_restore_size				= ttmp6
+var s_restore_ttmps_lo				= s_restore_tmp
+var s_restore_ttmps_hi				= s_restore_alloc_size
+var s_restore_spi_init_hi_save			= s_restore_exec_hi
+
+shader main
+	asic(DEFAULT)
+	type(CS)
+	wave_size(32)
+
+	s_branch	L_SKIP_RESTORE						//NOT restore. might be a regular trap or save
+
+L_JUMP_TO_RESTORE:
+	s_branch	L_RESTORE
+
+L_SKIP_RESTORE:
+	s_getreg_b32	s_save_status, hwreg(S_STATUS_HWREG)			//save STATUS since we will change SCC
+
+	// Clear SPI_PRIO: do not save with elevated priority.
+	// Clear ECC_ERR: prevents SQC store and triggers FATAL_HALT if setreg'd.
+	s_andn2_b32	s_save_status, s_save_status, S_STATUS_ALWAYS_CLEAR_MASK
+
+	s_getreg_b32	s_save_trapsts, hwreg(S_TRAPSTS_HWREG)
+
+#if SW_SA_TRAP
+	// If ttmp1[30] is set then issue s_barrier to unblock dependent waves.
+	s_bitcmp1_b32	s_save_pc_hi, 30
+	s_cbranch_scc0	L_TRAP_NO_BARRIER
+	s_barrier
+
+L_TRAP_NO_BARRIER:
+	// If ttmp1[31] is set then trap may occur early.
+	// Spin wait until SAVECTX exception is raised.
+	s_bitcmp1_b32	s_save_pc_hi, 31
+	s_cbranch_scc1  L_CHECK_SAVE
+#endif
+
+	s_and_b32       ttmp2, s_save_status, S_STATUS_HALT_MASK
+	s_cbranch_scc0	L_NOT_HALTED
+
+L_HALTED:
+	// Host trap may occur while wave is halted.
+	s_and_b32	ttmp2, s_save_pc_hi, S_SAVE_PC_HI_TRAP_ID_MASK
+	s_cbranch_scc1	L_FETCH_2ND_TRAP
+
+L_CHECK_SAVE:
+	s_and_b32	ttmp2, s_save_trapsts, S_TRAPSTS_SAVE_CONTEXT_MASK
+	s_cbranch_scc1	L_SAVE
+
+	// Wave is halted but neither host trap nor SAVECTX is raised.
+	// Caused by instruction fetch memory violation.
+	// Spin wait until context saved to prevent interrupt storm.
+	s_sleep		0x10
+	s_getreg_b32	s_save_trapsts, hwreg(S_TRAPSTS_HWREG)
+	s_branch	L_CHECK_SAVE
+
+L_NOT_HALTED:
+	// Let second-level handle non-SAVECTX exception or trap.
+	// Any concurrent SAVECTX will be handled upon re-entry once halted.
+
+	// Check non-maskable exceptions. memory_violation, illegal_instruction
+	// and xnack_error exceptions always cause the wave to enter the trap
+	// handler.
+	s_and_b32	ttmp2, s_save_trapsts, S_TRAPSTS_NON_MASKABLE_EXCP_MASK
+	s_cbranch_scc1	L_FETCH_2ND_TRAP
+
+	// Check for maskable exceptions in trapsts.excp and trapsts.excp_hi.
+	// Maskable exceptions only cause the wave to enter the trap handler if
+	// their respective bit in mode.excp_en is set.
+	s_and_b32	ttmp2, s_save_trapsts, SQ_WAVE_TRAPSTS_EXCP_MASK|SQ_WAVE_TRAPSTS_EXCP_HI_MASK
+	s_cbranch_scc0	L_CHECK_TRAP_ID
+
+	s_and_b32	ttmp3, s_save_trapsts, SQ_WAVE_TRAPSTS_ADDR_WATCH_MASK|SQ_WAVE_TRAPSTS_EXCP_HI_MASK
+	s_cbranch_scc0	L_NOT_ADDR_WATCH
+	s_bitset1_b32	ttmp2, SQ_WAVE_TRAPSTS_ADDR_WATCH_SHIFT // Check all addr_watch[123] exceptions against excp_en.addr_watch
+
+L_NOT_ADDR_WATCH:
+	s_getreg_b32	ttmp3, hwreg(HW_REG_MODE)
+	s_lshl_b32	ttmp2, ttmp2, SQ_WAVE_MODE_EXCP_EN_SHIFT
+	s_and_b32	ttmp2, ttmp2, ttmp3
+	s_cbranch_scc1	L_FETCH_2ND_TRAP
+
+L_CHECK_TRAP_ID:
+	// Check trap_id != 0
+	s_and_b32	ttmp2, s_save_pc_hi, S_SAVE_PC_HI_TRAP_ID_MASK
+	s_cbranch_scc1	L_FETCH_2ND_TRAP
+
+#if SINGLE_STEP_MISSED_WORKAROUND
+	// Prioritize single step exception over context save.
+	// Second-level trap will halt wave and RFE, re-entering for SAVECTX.
+	s_getreg_b32	ttmp2, hwreg(HW_REG_MODE)
+	s_and_b32	ttmp2, ttmp2, SQ_WAVE_MODE_DEBUG_EN_MASK
+	s_cbranch_scc1	L_FETCH_2ND_TRAP
+#endif
+
+	s_and_b32	ttmp2, s_save_trapsts, S_TRAPSTS_SAVE_CONTEXT_MASK
+	s_cbranch_scc1	L_SAVE
+
+L_FETCH_2ND_TRAP:
+#if HAVE_XNACK
+	save_and_clear_ib_sts(ttmp14, ttmp15)
+#endif
+
+	// Read second-level TBA/TMA from first-level TMA and jump if available.
+	// ttmp[2:5] and ttmp12 can be used (others hold SPI-initialized debug data)
+	// ttmp12 holds SQ_WAVE_STATUS
+#if HAVE_SENDMSG_RTN
+	s_sendmsg_rtn_b64       [ttmp14, ttmp15], sendmsg(MSG_RTN_GET_TMA)
+	S_WAITCNT_0
+#else
+	s_getreg_b32	ttmp14, hwreg(HW_REG_SHADER_TMA_LO)
+	s_getreg_b32	ttmp15, hwreg(HW_REG_SHADER_TMA_HI)
+#endif
+	s_lshl_b64	[ttmp14, ttmp15], [ttmp14, ttmp15], 0x8
+
+	s_bitcmp1_b32	ttmp15, 0xF
+	s_cbranch_scc0	L_NO_SIGN_EXTEND_TMA
+	s_or_b32	ttmp15, ttmp15, 0xFFFF0000
+L_NO_SIGN_EXTEND_TMA:
+
+	s_load_dword    ttmp2, [ttmp14, ttmp15], 0x10 S_COHERENCE		// debug trap enabled flag
+	S_WAITCNT_0
+	s_lshl_b32      ttmp2, ttmp2, TTMP11_DEBUG_TRAP_ENABLED_SHIFT
+	s_andn2_b32     ttmp11, ttmp11, TTMP11_DEBUG_TRAP_ENABLED_MASK
+	s_or_b32        ttmp11, ttmp11, ttmp2
+
+	s_load_dwordx2	[ttmp2, ttmp3], [ttmp14, ttmp15], 0x0 S_COHERENCE	// second-level TBA
+	S_WAITCNT_0
+	s_load_dwordx2	[ttmp14, ttmp15], [ttmp14, ttmp15], 0x8 S_COHERENCE	// second-level TMA
+	S_WAITCNT_0
+
+	s_and_b64	[ttmp2, ttmp3], [ttmp2, ttmp3], [ttmp2, ttmp3]
+	s_cbranch_scc0	L_NO_NEXT_TRAP						// second-level trap handler not been set
+	s_setpc_b64	[ttmp2, ttmp3]						// jump to second-level trap handler
+
+L_NO_NEXT_TRAP:
+	// If not caused by trap then halt wave to prevent re-entry.
+	s_and_b32	ttmp2, s_save_pc_hi, S_SAVE_PC_HI_TRAP_ID_MASK
+	s_cbranch_scc1	L_TRAP_CASE
+
+	// Host trap will not cause trap re-entry.
+	s_and_b32	ttmp2, s_save_pc_hi, S_SAVE_PC_HI_HT_MASK
+	s_cbranch_scc1	L_EXIT_TRAP
+	s_or_b32	s_save_status, s_save_status, S_STATUS_HALT_MASK
+
+	// If the PC points to S_ENDPGM then context save will fail if STATUS.HALT is set.
+	// Rewind the PC to prevent this from occurring.
+	s_sub_u32	ttmp0, ttmp0, 0x8
+	s_subb_u32	ttmp1, ttmp1, 0x0
+
+	s_branch	L_EXIT_TRAP
+
+L_TRAP_CASE:
+	// Advance past trap instruction to prevent re-entry.
+	s_add_u32	ttmp0, ttmp0, 0x4
+	s_addc_u32	ttmp1, ttmp1, 0x0
+
+L_EXIT_TRAP:
+	s_and_b32	ttmp1, ttmp1, 0xFFFF
+
+#if HAVE_XNACK
+	restore_ib_sts(ttmp14, ttmp15)
+#endif
+
+	// Restore SQ_WAVE_STATUS.
+	s_and_b64	exec, exec, exec					// Restore STATUS.EXECZ, not writable by s_setreg_b32
+	s_and_b64	vcc, vcc, vcc						// Restore STATUS.VCCZ, not writable by s_setreg_b32
+
+	s_setreg_b32	hwreg(S_STATUS_HWREG), s_save_status
+	s_rfe_b64	[ttmp0, ttmp1]
+
+L_SAVE:
+	// If VGPRs have been deallocated then terminate the wavefront.
+	// It has no remaining program to run and cannot save without VGPRs.
+#if ASIC_FAMILY == CHIP_PLUM_BONITO
+	s_bitcmp1_b32	s_save_status, SQ_WAVE_STATUS_NO_VGPRS_SHIFT
+	s_cbranch_scc0	L_HAVE_VGPRS
+	s_endpgm
+L_HAVE_VGPRS:
+#endif
+	s_and_b32	s_save_pc_hi, s_save_pc_hi, 0x0000ffff			//pc[47:32]
+	s_mov_b32	s_save_tmp, 0
+	s_setreg_b32	hwreg(S_TRAPSTS_HWREG, S_TRAPSTS_SAVE_CONTEXT_SHIFT, 1), s_save_tmp	//clear saveCtx bit
+
+#if HAVE_XNACK
+	save_and_clear_ib_sts(s_save_tmp, s_save_trapsts)
+#endif
+
+	/* inform SPI the readiness and wait for SPI's go signal */
+	s_mov_b32	s_save_exec_lo, exec_lo					//save EXEC and use EXEC for the go signal from SPI
+	s_mov_b32	s_save_exec_hi, exec_hi
+	s_mov_b64	exec, 0x0						//clear EXEC to get ready to receive
+
+#if HAVE_SENDMSG_RTN
+	s_sendmsg_rtn_b64       [exec_lo, exec_hi], sendmsg(MSG_RTN_SAVE_WAVE)
+#else
+	s_sendmsg	sendmsg(MSG_SAVEWAVE)					//send SPI a message and wait for SPI's write to EXEC
+#endif
+
+#if ASIC_FAMILY < CHIP_SIENNA_CICHLID
+L_SLEEP:
+	// sleep 1 (64clk) is not enough for 8 waves per SIMD, which will cause
+	// SQ hang, since the 7,8th wave could not get arbit to exec inst, while
+	// other waves are stuck into the sleep-loop and waiting for wrexec!=0
+	s_sleep		0x2
+	s_cbranch_execz	L_SLEEP
+#else
+	S_WAITCNT_0
+#endif
+
+	// Save first_wave flag so we can clear high bits of save address.
+	s_and_b32	s_save_tmp, s_save_spi_init_hi, S_SAVE_SPI_INIT_FIRST_WAVE_MASK
+	s_lshl_b32	s_save_tmp, s_save_tmp, (S_SAVE_PC_HI_FIRST_WAVE_SHIFT - S_SAVE_SPI_INIT_FIRST_WAVE_SHIFT)
+	s_or_b32	s_save_pc_hi, s_save_pc_hi, s_save_tmp
+
+#if NO_SQC_STORE
+#if ASIC_FAMILY <= CHIP_SIENNA_CICHLID
+	// gfx10: If there was a VALU exception, the exception state must be
+	// cleared before executing the VALU instructions below.
+	v_clrexcp
+#endif
+
+	// Trap temporaries must be saved via VGPR but all VGPRs are in use.
+	// There is no ttmp space to hold the resource constant for VGPR save.
+	// Save v0 by itself since it requires only two SGPRs.
+	s_mov_b32	s_save_ttmps_lo, exec_lo
+	s_and_b32	s_save_ttmps_hi, exec_hi, 0xFFFF
+	s_mov_b32	exec_lo, 0xFFFFFFFF
+	s_mov_b32	exec_hi, 0xFFFFFFFF
+	global_store_dword_addtid	v0, [s_save_ttmps_lo, s_save_ttmps_hi] V_COHERENCE
+	v_mov_b32	v0, 0x0
+	s_mov_b32	exec_lo, s_save_ttmps_lo
+	s_mov_b32	exec_hi, s_save_ttmps_hi
+#endif
+
+	// Save trap temporaries 4-11, 13 initialized by SPI debug dispatch logic
+	// ttmp SR memory offset : size(VGPR)+size(SVGPR)+size(SGPR)+0x40
+	get_wave_size2(s_save_ttmps_hi)
+	get_vgpr_size_bytes(s_save_ttmps_lo, s_save_ttmps_hi)
+	get_svgpr_size_bytes(s_save_ttmps_hi)
+	s_add_u32	s_save_ttmps_lo, s_save_ttmps_lo, s_save_ttmps_hi
+	s_and_b32	s_save_ttmps_hi, s_save_spi_init_hi, 0xFFFF
+	s_add_u32	s_save_ttmps_lo, s_save_ttmps_lo, get_sgpr_size_bytes()
+	s_add_u32	s_save_ttmps_lo, s_save_ttmps_lo, s_save_spi_init_lo
+	s_addc_u32	s_save_ttmps_hi, s_save_ttmps_hi, 0x0
+
+#if NO_SQC_STORE
+	v_writelane_b32	v0, ttmp4, 0x4
+	v_writelane_b32	v0, ttmp5, 0x5
+	v_writelane_b32	v0, ttmp6, 0x6
+	v_writelane_b32	v0, ttmp7, 0x7
+	v_writelane_b32	v0, ttmp8, 0x8
+	v_writelane_b32	v0, ttmp9, 0x9
+	v_writelane_b32	v0, ttmp10, 0xA
+	v_writelane_b32	v0, ttmp11, 0xB
+	v_writelane_b32	v0, ttmp13, 0xD
+	v_writelane_b32	v0, exec_lo, 0xE
+	v_writelane_b32	v0, exec_hi, 0xF
+
+	s_mov_b32	exec_lo, 0x3FFF
+	s_mov_b32	exec_hi, 0x0
+	global_store_dword_addtid	v0, [s_save_ttmps_lo, s_save_ttmps_hi] inst_offset:0x40 V_COHERENCE
+	v_readlane_b32	ttmp14, v0, 0xE
+	v_readlane_b32	ttmp15, v0, 0xF
+	s_mov_b32	exec_lo, ttmp14
+	s_mov_b32	exec_hi, ttmp15
+#else
+	s_store_dwordx4	[ttmp4, ttmp5, ttmp6, ttmp7], [s_save_ttmps_lo, s_save_ttmps_hi], 0x50 S_COHERENCE
+	s_store_dwordx4	[ttmp8, ttmp9, ttmp10, ttmp11], [s_save_ttmps_lo, s_save_ttmps_hi], 0x60 S_COHERENCE
+	s_store_dword   ttmp13, [s_save_ttmps_lo, s_save_ttmps_hi], 0x74 S_COHERENCE
+#endif
+
+	/* setup Resource Contants */
+	s_mov_b32	s_save_buf_rsrc0, s_save_spi_init_lo			//base_addr_lo
+	s_and_b32	s_save_buf_rsrc1, s_save_spi_init_hi, 0x0000FFFF	//base_addr_hi
+	s_or_b32	s_save_buf_rsrc1, s_save_buf_rsrc1, S_SAVE_BUF_RSRC_WORD1_STRIDE
+	s_mov_b32	s_save_buf_rsrc2, 0					//NUM_RECORDS initial value = 0 (in bytes) although not neccessarily inited
+	s_mov_b32	s_save_buf_rsrc3, S_SAVE_BUF_RSRC_WORD3_MISC
+
+	s_mov_b32	s_save_m0, m0
+
+	/* global mem offset */
+	s_mov_b32	s_save_mem_offset, 0x0
+	get_wave_size2(s_wave_size)
+
+#if HAVE_XNACK
+	// Save and clear vector XNACK state late to free up SGPRs.
+	s_getreg_b32	s_save_xnack_mask, hwreg(HW_REG_SHADER_XNACK_MASK)
+	s_setreg_imm32_b32	hwreg(HW_REG_SHADER_XNACK_MASK), 0x0
+#endif
+
+	/* save first 4 VGPRs, needed for SGPR save */
+	s_mov_b32	exec_lo, 0xFFFFFFFF					//need every thread from now on
+	s_lshr_b32	m0, s_wave_size, S_WAVE_SIZE
+	s_and_b32	m0, m0, 1
+	s_cmp_eq_u32	m0, 1
+	s_cbranch_scc1	L_ENABLE_SAVE_4VGPR_EXEC_HI
+	s_mov_b32	exec_hi, 0x00000000
+	s_branch	L_SAVE_4VGPR_WAVE32
+L_ENABLE_SAVE_4VGPR_EXEC_HI:
+	s_mov_b32	exec_hi, 0xFFFFFFFF
+	s_branch	L_SAVE_4VGPR_WAVE64
+L_SAVE_4VGPR_WAVE32:
+	s_mov_b32	s_save_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+	// VGPR Allocated in 4-GPR granularity
+
+#if SAVE_AFTER_XNACK_ERROR
+	check_if_tcp_store_ok()
+	s_cbranch_scc1 L_SAVE_FIRST_VGPRS32_WITH_TCP
+
+	write_vgprs_to_mem_with_sqc_w32(v0, 4, s_save_buf_rsrc0, s_save_mem_offset)
+	s_branch L_SAVE_HWREG
+
+L_SAVE_FIRST_VGPRS32_WITH_TCP:
+#endif
+
+#if !NO_SQC_STORE
+	buffer_store_dword	v0, v0, s_save_buf_rsrc0, s_save_mem_offset V_COHERENCE
+#endif
+	buffer_store_dword	v1, v0, s_save_buf_rsrc0, s_save_mem_offset V_COHERENCE offset:128
+	buffer_store_dword	v2, v0, s_save_buf_rsrc0, s_save_mem_offset V_COHERENCE offset:128*2
+	buffer_store_dword	v3, v0, s_save_buf_rsrc0, s_save_mem_offset V_COHERENCE offset:128*3
+	s_branch	L_SAVE_HWREG
+
+L_SAVE_4VGPR_WAVE64:
+	s_mov_b32	s_save_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+	// VGPR Allocated in 4-GPR granularity
+
+#if  SAVE_AFTER_XNACK_ERROR
+	check_if_tcp_store_ok()
+	s_cbranch_scc1 L_SAVE_FIRST_VGPRS64_WITH_TCP
+
+	write_vgprs_to_mem_with_sqc_w64(v0, 4, s_save_buf_rsrc0, s_save_mem_offset)
+	s_branch L_SAVE_HWREG
+
+L_SAVE_FIRST_VGPRS64_WITH_TCP:
+#endif
+
+#if !NO_SQC_STORE
+	buffer_store_dword	v0, v0, s_save_buf_rsrc0, s_save_mem_offset V_COHERENCE
+#endif
+	buffer_store_dword	v1, v0, s_save_buf_rsrc0, s_save_mem_offset V_COHERENCE offset:256
+	buffer_store_dword	v2, v0, s_save_buf_rsrc0, s_save_mem_offset V_COHERENCE offset:256*2
+	buffer_store_dword	v3, v0, s_save_buf_rsrc0, s_save_mem_offset V_COHERENCE offset:256*3
+
+	/* save HW registers */
+
+L_SAVE_HWREG:
+	// HWREG SR memory offset : size(VGPR)+size(SVGPR)+size(SGPR)
+	get_vgpr_size_bytes(s_save_mem_offset, s_wave_size)
+	get_svgpr_size_bytes(s_save_tmp)
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, s_save_tmp
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, get_sgpr_size_bytes()
+
+	s_mov_b32	s_save_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+#if NO_SQC_STORE
+	v_mov_b32	v0, 0x0							//Offset[31:0] from buffer resource
+	v_mov_b32	v1, 0x0							//Offset[63:32] from buffer resource
+	v_mov_b32	v2, 0x0							//Set of SGPRs for TCP store
+	s_mov_b32	m0, 0x0							//Next lane of v2 to write to
+#endif
+
+	write_hwreg_to_mem(s_save_m0, s_save_buf_rsrc0, s_save_mem_offset)
+	write_hwreg_to_mem(s_save_pc_lo, s_save_buf_rsrc0, s_save_mem_offset)
+	s_andn2_b32	s_save_tmp, s_save_pc_hi, S_SAVE_PC_HI_FIRST_WAVE_MASK
+	write_hwreg_to_mem(s_save_tmp, s_save_buf_rsrc0, s_save_mem_offset)
+	write_hwreg_to_mem(s_save_exec_lo, s_save_buf_rsrc0, s_save_mem_offset)
+	write_hwreg_to_mem(s_save_exec_hi, s_save_buf_rsrc0, s_save_mem_offset)
+	write_hwreg_to_mem(s_save_status, s_save_buf_rsrc0, s_save_mem_offset)
+
+	s_getreg_b32	s_save_tmp, hwreg(S_TRAPSTS_HWREG)
+	write_hwreg_to_mem(s_save_tmp, s_save_buf_rsrc0, s_save_mem_offset)
+
+	// Not used on Sienna_Cichlid but keep layout same for debugger.
+	write_hwreg_to_mem(s_save_xnack_mask, s_save_buf_rsrc0, s_save_mem_offset)
+
+	s_getreg_b32	s_save_m0, hwreg(HW_REG_MODE)
+	write_hwreg_to_mem(s_save_m0, s_save_buf_rsrc0, s_save_mem_offset)
+
+	s_getreg_b32	s_save_m0, hwreg(HW_REG_SHADER_FLAT_SCRATCH_LO)
+	write_hwreg_to_mem(s_save_m0, s_save_buf_rsrc0, s_save_mem_offset)
+
+	s_getreg_b32	s_save_m0, hwreg(HW_REG_SHADER_FLAT_SCRATCH_HI)
+	write_hwreg_to_mem(s_save_m0, s_save_buf_rsrc0, s_save_mem_offset)
+
+#if NO_SQC_STORE
+	// Write HWREGs with 16 VGPR lanes. TTMPs occupy space after this.
+	s_mov_b32       exec_lo, 0xFFFF
+	s_mov_b32	exec_hi, 0x0
+	buffer_store_dword	v2, v0, s_save_buf_rsrc0, s_save_mem_offset V_COHERENCE
+
+	// Write SGPRs with 32 VGPR lanes. This works in wave32 and wave64 mode.
+	s_mov_b32       exec_lo, 0xFFFFFFFF
+#endif
+
+	/* save SGPRs */
+	// Save SGPR before LDS save, then the s0 to s4 can be used during LDS save...
+
+	// SGPR SR memory offset : size(VGPR)+size(SVGPR)
+	get_vgpr_size_bytes(s_save_mem_offset, s_wave_size)
+	get_svgpr_size_bytes(s_save_tmp)
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, s_save_tmp
+	s_mov_b32	s_save_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+#if NO_SQC_STORE
+	s_mov_b32	ttmp13, 0x0						//next VGPR lane to copy SGPR into
+#else
+	// backup s_save_buf_rsrc0,1 to s_save_pc_lo/hi, since write_16sgpr_to_mem function will change the rsrc0
+	s_mov_b32	s_save_xnack_mask, s_save_buf_rsrc0
+	s_add_u32	s_save_buf_rsrc0, s_save_buf_rsrc0, s_save_mem_offset
+	s_addc_u32	s_save_buf_rsrc1, s_save_buf_rsrc1, 0
+#endif
+
+	s_mov_b32	m0, 0x0							//SGPR initial index value =0
+	s_nop		0x0							//Manually inserted wait states
+L_SAVE_SGPR_LOOP:
+	// SGPR is allocated in 16 SGPR granularity
+	s_movrels_b64	s0, s0							//s0 = s[0+m0], s1 = s[1+m0]
+	s_movrels_b64	s2, s2							//s2 = s[2+m0], s3 = s[3+m0]
+	s_movrels_b64	s4, s4							//s4 = s[4+m0], s5 = s[5+m0]
+	s_movrels_b64	s6, s6							//s6 = s[6+m0], s7 = s[7+m0]
+	s_movrels_b64	s8, s8							//s8 = s[8+m0], s9 = s[9+m0]
+	s_movrels_b64	s10, s10						//s10 = s[10+m0], s11 = s[11+m0]
+	s_movrels_b64	s12, s12						//s12 = s[12+m0], s13 = s[13+m0]
+	s_movrels_b64	s14, s14						//s14 = s[14+m0], s15 = s[15+m0]
+
+	write_16sgpr_to_mem(s0, s_save_buf_rsrc0, s_save_mem_offset)
+
+#if NO_SQC_STORE
+	s_cmp_eq_u32	ttmp13, 0x20						//have 32 VGPR lanes filled?
+	s_cbranch_scc0	L_SAVE_SGPR_SKIP_TCP_STORE
+
+	buffer_store_dword	v2, v0, s_save_buf_rsrc0, s_save_mem_offset V_COHERENCE
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, 0x80
+	s_mov_b32	ttmp13, 0x0
+	v_mov_b32	v2, 0x0
+L_SAVE_SGPR_SKIP_TCP_STORE:
+#endif
+
+	s_add_u32	m0, m0, 16						//next sgpr index
+	s_cmp_lt_u32	m0, 96							//scc = (m0 < first 96 SGPR) ? 1 : 0
+	s_cbranch_scc1	L_SAVE_SGPR_LOOP					//first 96 SGPR save is complete?
+
+	//save the rest 12 SGPR
+	s_movrels_b64	s0, s0							//s0 = s[0+m0], s1 = s[1+m0]
+	s_movrels_b64	s2, s2							//s2 = s[2+m0], s3 = s[3+m0]
+	s_movrels_b64	s4, s4							//s4 = s[4+m0], s5 = s[5+m0]
+	s_movrels_b64	s6, s6							//s6 = s[6+m0], s7 = s[7+m0]
+	s_movrels_b64	s8, s8							//s8 = s[8+m0], s9 = s[9+m0]
+	s_movrels_b64	s10, s10						//s10 = s[10+m0], s11 = s[11+m0]
+	write_12sgpr_to_mem(s0, s_save_buf_rsrc0, s_save_mem_offset)
+
+#if NO_SQC_STORE
+	buffer_store_dword	v2, v0, s_save_buf_rsrc0, s_save_mem_offset V_COHERENCE
+#else
+	// restore s_save_buf_rsrc0,1
+	s_mov_b32	s_save_buf_rsrc0, s_save_xnack_mask
+#endif
+
+	/* save LDS */
+
+L_SAVE_LDS:
+	// Change EXEC to all threads...
+	s_mov_b32	exec_lo, 0xFFFFFFFF					//need every thread from now on
+	s_lshr_b32	m0, s_wave_size, S_WAVE_SIZE
+	s_and_b32	m0, m0, 1
+	s_cmp_eq_u32	m0, 1
+	s_cbranch_scc1	L_ENABLE_SAVE_LDS_EXEC_HI
+	s_mov_b32	exec_hi, 0x00000000
+	s_branch	L_SAVE_LDS_NORMAL
+L_ENABLE_SAVE_LDS_EXEC_HI:
+	s_mov_b32	exec_hi, 0xFFFFFFFF
+L_SAVE_LDS_NORMAL:
+	s_getreg_b32	s_save_alloc_size, hwreg(HW_REG_LDS_ALLOC,SQ_WAVE_LDS_ALLOC_LDS_SIZE_SHIFT,SQ_WAVE_LDS_ALLOC_LDS_SIZE_SIZE)
+	s_and_b32	s_save_alloc_size, s_save_alloc_size, 0xFFFFFFFF	//lds_size is zero?
+	s_cbranch_scc0	L_SAVE_LDS_DONE						//no lds used? jump to L_SAVE_DONE
+
+	s_barrier								//LDS is used? wait for other waves in the same TG
+	s_and_b32	s_save_tmp, s_save_pc_hi, S_SAVE_PC_HI_FIRST_WAVE_MASK
+	s_cbranch_scc0	L_SAVE_LDS_DONE
+
+	// first wave do LDS save;
+
+	s_lshl_b32	s_save_alloc_size, s_save_alloc_size, SQ_WAVE_LDS_ALLOC_GRANULARITY
+	s_mov_b32	s_save_buf_rsrc2, s_save_alloc_size			//NUM_RECORDS in bytes
+
+	// LDS at offset: size(VGPR)+size(SVGPR)+SIZE(SGPR)+SIZE(HWREG)
+	//
+	get_vgpr_size_bytes(s_save_mem_offset, s_wave_size)
+	get_svgpr_size_bytes(s_save_tmp)
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, s_save_tmp
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, get_sgpr_size_bytes()
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, get_hwreg_size_bytes()
+
+	s_mov_b32	s_save_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+	//load 0~63*4(byte address) to vgpr v0
+	v_mbcnt_lo_u32_b32	v0, -1, 0
+	v_mbcnt_hi_u32_b32	v0, -1, v0
+	v_mul_u32_u24	v0, 4, v0
+
+	s_lshr_b32	m0, s_wave_size, S_WAVE_SIZE
+	s_and_b32	m0, m0, 1
+	s_cmp_eq_u32	m0, 1
+	s_mov_b32	m0, 0x0
+	s_cbranch_scc1	L_SAVE_LDS_W64
+
+L_SAVE_LDS_W32:
+#if SAVE_AFTER_XNACK_ERROR
+	check_if_tcp_store_ok()
+	s_cbranch_scc1 L_SAVE_LDS_WITH_TCP_W32
+
+L_SAVE_LDS_LOOP_SQC_W32:
+	ds_read_b32	v1, v0
+	S_WAITCNT_0
+
+	write_vgprs_to_mem_with_sqc_w32(v1, 1, s_save_buf_rsrc0, s_save_mem_offset)
+
+	s_add_u32	m0, m0, 128						//every buffer_store_lds does 128 bytes
+	v_add_nc_u32	v0, v0, 128						//mem offset increased by 128 bytes
+	s_cmp_lt_u32	m0, s_save_alloc_size					//scc=(m0 < s_save_alloc_size) ? 1 : 0
+	s_cbranch_scc1	L_SAVE_LDS_LOOP_SQC_W32					//LDS save is complete?
+
+	s_branch	L_SAVE_LDS_DONE
+
+L_SAVE_LDS_WITH_TCP_W32:
+#endif
+
+	s_mov_b32	s3, 128
+	s_nop		0
+	s_nop		0
+	s_nop		0
+L_SAVE_LDS_LOOP_W32:
+	ds_read_b32	v1, v0
+	S_WAITCNT_0
+	buffer_store_dword	v1, v0, s_save_buf_rsrc0, s_save_mem_offset V_COHERENCE
+
+	s_add_u32	m0, m0, s3						//every buffer_store_lds does 128 bytes
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, s3
+	v_add_nc_u32	v0, v0, 128						//mem offset increased by 128 bytes
+	s_cmp_lt_u32	m0, s_save_alloc_size					//scc=(m0 < s_save_alloc_size) ? 1 : 0
+	s_cbranch_scc1	L_SAVE_LDS_LOOP_W32					//LDS save is complete?
+
+	s_branch	L_SAVE_LDS_DONE
+
+L_SAVE_LDS_W64:
+#if  SAVE_AFTER_XNACK_ERROR
+	check_if_tcp_store_ok()
+	s_cbranch_scc1 L_SAVE_LDS_WITH_TCP_W64
+
+L_SAVE_LDS_LOOP_SQC_W64:
+	ds_read_b32	v1, v0
+	S_WAITCNT_0
+
+	write_vgprs_to_mem_with_sqc_w64(v1, 1, s_save_buf_rsrc0, s_save_mem_offset)
+
+	s_add_u32	m0, m0, 256						//every buffer_store_lds does 256 bytes
+	v_add_nc_u32	v0, v0, 256						//mem offset increased by 256 bytes
+	s_cmp_lt_u32	m0, s_save_alloc_size					//scc=(m0 < s_save_alloc_size) ? 1 : 0
+	s_cbranch_scc1	L_SAVE_LDS_LOOP_SQC_W64					//LDS save is complete?
+
+	s_branch	L_SAVE_LDS_DONE
+
+L_SAVE_LDS_WITH_TCP_W64:
+#endif
+
+	s_mov_b32	s3, 256
+	s_nop		0
+	s_nop		0
+	s_nop		0
+L_SAVE_LDS_LOOP_W64:
+	ds_read_b32	v1, v0
+	S_WAITCNT_0
+	buffer_store_dword	v1, v0, s_save_buf_rsrc0, s_save_mem_offset V_COHERENCE
+
+	s_add_u32	m0, m0, s3						//every buffer_store_lds does 256 bytes
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, s3
+	v_add_nc_u32	v0, v0, 256						//mem offset increased by 256 bytes
+	s_cmp_lt_u32	m0, s_save_alloc_size					//scc=(m0 < s_save_alloc_size) ? 1 : 0
+	s_cbranch_scc1	L_SAVE_LDS_LOOP_W64					//LDS save is complete?
+
+L_SAVE_LDS_DONE:
+	/* save VGPRs  - set the Rest VGPRs */
+L_SAVE_VGPR:
+	// VGPR SR memory offset: 0
+	s_mov_b32	exec_lo, 0xFFFFFFFF					//need every thread from now on
+	s_lshr_b32	m0, s_wave_size, S_WAVE_SIZE
+	s_and_b32	m0, m0, 1
+	s_cmp_eq_u32	m0, 1
+	s_cbranch_scc1	L_ENABLE_SAVE_VGPR_EXEC_HI
+	s_mov_b32	s_save_mem_offset, (0+128*4)				// for the rest VGPRs
+	s_mov_b32	exec_hi, 0x00000000
+	s_branch	L_SAVE_VGPR_NORMAL
+L_ENABLE_SAVE_VGPR_EXEC_HI:
+	s_mov_b32	s_save_mem_offset, (0+256*4)				// for the rest VGPRs
+	s_mov_b32	exec_hi, 0xFFFFFFFF
+L_SAVE_VGPR_NORMAL:
+	s_getreg_b32	s_save_alloc_size, hwreg(HW_REG_GPR_ALLOC,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SIZE)
+	s_add_u32	s_save_alloc_size, s_save_alloc_size, 1
+	s_lshl_b32	s_save_alloc_size, s_save_alloc_size, 2			//Number of VGPRs = (vgpr_size + 1) * 4    (non-zero value)
+	//determine it is wave32 or wave64
+	s_lshr_b32	m0, s_wave_size, S_WAVE_SIZE
+	s_and_b32	m0, m0, 1
+	s_cmp_eq_u32	m0, 1
+	s_cbranch_scc1	L_SAVE_VGPR_WAVE64
+
+	s_mov_b32	s_save_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+	// VGPR Allocated in 4-GPR granularity
+
+	// VGPR store using dw burst
+	s_mov_b32	m0, 0x4							//VGPR initial index value =4
+	s_cmp_lt_u32	m0, s_save_alloc_size
+	s_cbranch_scc0	L_SAVE_VGPR_END
+
+#if  SAVE_AFTER_XNACK_ERROR
+	check_if_tcp_store_ok()
+	s_cbranch_scc1 L_SAVE_VGPR_W32_LOOP
+
+L_SAVE_VGPR_LOOP_SQC_W32:
+	v_movrels_b32	v0, v0							//v0 = v[0+m0]
+	v_movrels_b32	v1, v1							//v1 = v[1+m0]
+	v_movrels_b32	v2, v2							//v2 = v[2+m0]
+	v_movrels_b32	v3, v3							//v3 = v[3+m0]
+
+	write_vgprs_to_mem_with_sqc_w32(v0, 4, s_save_buf_rsrc0, s_save_mem_offset)
+
+	s_add_u32 m0, m0, 4
+	s_cmp_lt_u32 m0, s_save_alloc_size
+	s_cbranch_scc1 L_SAVE_VGPR_LOOP_SQC_W32
+
+	s_branch L_SAVE_VGPR_END
+#endif
+
+L_SAVE_VGPR_W32_LOOP:
+	v_movrels_b32	v0, v0							//v0 = v[0+m0]
+	v_movrels_b32	v1, v1							//v1 = v[1+m0]
+	v_movrels_b32	v2, v2							//v2 = v[2+m0]
+	v_movrels_b32	v3, v3							//v3 = v[3+m0]
+
+	buffer_store_dword	v0, v0, s_save_buf_rsrc0, s_save_mem_offset V_COHERENCE
+	buffer_store_dword	v1, v0, s_save_buf_rsrc0, s_save_mem_offset V_COHERENCE offset:128
+	buffer_store_dword	v2, v0, s_save_buf_rsrc0, s_save_mem_offset V_COHERENCE offset:128*2
+	buffer_store_dword	v3, v0, s_save_buf_rsrc0, s_save_mem_offset V_COHERENCE offset:128*3
+
+	s_add_u32	m0, m0, 4						//next vgpr index
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, 128*4		//every buffer_store_dword does 128 bytes
+	s_cmp_lt_u32	m0, s_save_alloc_size					//scc = (m0 < s_save_alloc_size) ? 1 : 0
+	s_cbranch_scc1	L_SAVE_VGPR_W32_LOOP					//VGPR save is complete?
+
+	s_branch	L_SAVE_VGPR_END
+
+L_SAVE_VGPR_WAVE64:
+	s_mov_b32	s_save_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+	// VGPR store using dw burst
+	s_mov_b32	m0, 0x4							//VGPR initial index value =4
+	s_cmp_lt_u32	m0, s_save_alloc_size
+	s_cbranch_scc0	L_SAVE_SHARED_VGPR
+
+#if  SAVE_AFTER_XNACK_ERROR
+	check_if_tcp_store_ok()
+	s_cbranch_scc1 L_SAVE_VGPR_W64_LOOP
+
+L_SAVE_VGPR_LOOP_SQC_W64:
+	v_movrels_b32	v0, v0							//v0 = v[0+m0]
+	v_movrels_b32	v1, v1							//v1 = v[1+m0]
+	v_movrels_b32	v2, v2							//v2 = v[2+m0]
+	v_movrels_b32	v3, v3							//v3 = v[3+m0]
+
+	write_vgprs_to_mem_with_sqc_w64(v0, 4, s_save_buf_rsrc0, s_save_mem_offset)
+
+	s_add_u32 m0, m0, 4
+	s_cmp_lt_u32 m0, s_save_alloc_size
+	s_cbranch_scc1 L_SAVE_VGPR_LOOP_SQC_W64
+
+	s_branch L_SAVE_VGPR_END
+#endif
+
+L_SAVE_VGPR_W64_LOOP:
+	v_movrels_b32	v0, v0							//v0 = v[0+m0]
+	v_movrels_b32	v1, v1							//v1 = v[1+m0]
+	v_movrels_b32	v2, v2							//v2 = v[2+m0]
+	v_movrels_b32	v3, v3							//v3 = v[3+m0]
+
+	buffer_store_dword	v0, v0, s_save_buf_rsrc0, s_save_mem_offset V_COHERENCE
+	buffer_store_dword	v1, v0, s_save_buf_rsrc0, s_save_mem_offset V_COHERENCE offset:256
+	buffer_store_dword	v2, v0, s_save_buf_rsrc0, s_save_mem_offset V_COHERENCE offset:256*2
+	buffer_store_dword	v3, v0, s_save_buf_rsrc0, s_save_mem_offset V_COHERENCE offset:256*3
+
+	s_add_u32	m0, m0, 4						//next vgpr index
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, 256*4		//every buffer_store_dword does 256 bytes
+	s_cmp_lt_u32	m0, s_save_alloc_size					//scc = (m0 < s_save_alloc_size) ? 1 : 0
+	s_cbranch_scc1	L_SAVE_VGPR_W64_LOOP					//VGPR save is complete?
+
+L_SAVE_SHARED_VGPR:
+	//Below part will be the save shared vgpr part (new for gfx10)
+	s_getreg_b32	s_save_alloc_size, hwreg(HW_REG_LDS_ALLOC,SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SHIFT,SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SIZE)
+	s_and_b32	s_save_alloc_size, s_save_alloc_size, 0xFFFFFFFF	//shared_vgpr_size is zero?
+	s_cbranch_scc0	L_SAVE_VGPR_END						//no shared_vgpr used? jump to L_SAVE_LDS
+	s_lshl_b32	s_save_alloc_size, s_save_alloc_size, 3			//Number of SHARED_VGPRs = shared_vgpr_size * 8    (non-zero value)
+	//m0 now has the value of normal vgpr count, just add the m0 with shared_vgpr count to get the total count.
+	//save shared_vgpr will start from the index of m0
+	s_add_u32	s_save_alloc_size, s_save_alloc_size, m0
+	s_mov_b32	exec_lo, 0xFFFFFFFF
+	s_mov_b32	exec_hi, 0x00000000
+
+#if  SAVE_AFTER_XNACK_ERROR
+	check_if_tcp_store_ok()
+	s_cbranch_scc1 L_SAVE_SHARED_VGPR_WAVE64_LOOP
+
+L_SAVE_SHARED_VGPR_WAVE64_LOOP_SQC:
+	v_movrels_b32	v0, v0
+
+	write_vgprs_to_mem_with_sqc_w64(v0, 1, s_save_buf_rsrc0, s_save_mem_offset)
+
+	s_add_u32 m0, m0, 1
+	s_cmp_lt_u32 m0, s_save_alloc_size
+	s_cbranch_scc1 L_SAVE_SHARED_VGPR_WAVE64_LOOP_SQC
+
+	s_branch L_SAVE_VGPR_END
+#endif
+
+L_SAVE_SHARED_VGPR_WAVE64_LOOP:
+	v_movrels_b32	v0, v0							//v0 = v[0+m0]
+	buffer_store_dword	v0, v0, s_save_buf_rsrc0, s_save_mem_offset V_COHERENCE
+	s_add_u32	m0, m0, 1						//next vgpr index
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, 128
+	s_cmp_lt_u32	m0, s_save_alloc_size					//scc = (m0 < s_save_alloc_size) ? 1 : 0
+	s_cbranch_scc1	L_SAVE_SHARED_VGPR_WAVE64_LOOP				//SHARED_VGPR save is complete?
+
+L_SAVE_VGPR_END:
+	s_branch	L_END_PGM
+
+L_RESTORE:
+	/* Setup Resource Contants */
+	s_mov_b32	s_restore_buf_rsrc0, s_restore_spi_init_lo		//base_addr_lo
+	s_and_b32	s_restore_buf_rsrc1, s_restore_spi_init_hi, 0x0000FFFF	//base_addr_hi
+	s_or_b32	s_restore_buf_rsrc1, s_restore_buf_rsrc1, S_RESTORE_BUF_RSRC_WORD1_STRIDE
+	s_mov_b32	s_restore_buf_rsrc2, 0					//NUM_RECORDS initial value = 0 (in bytes)
+	s_mov_b32	s_restore_buf_rsrc3, S_RESTORE_BUF_RSRC_WORD3_MISC
+
+	//determine it is wave32 or wave64
+	get_wave_size2(s_restore_size)
+
+	s_and_b32	s_restore_tmp, s_restore_spi_init_hi, S_RESTORE_SPI_INIT_FIRST_WAVE_MASK
+	s_cbranch_scc0	L_RESTORE_VGPR
+
+	/* restore LDS */
+L_RESTORE_LDS:
+	s_mov_b32	exec_lo, 0xFFFFFFFF					//need every thread from now on
+	s_lshr_b32	m0, s_restore_size, S_WAVE_SIZE
+	s_and_b32	m0, m0, 1
+	s_cmp_eq_u32	m0, 1
+	s_cbranch_scc1	L_ENABLE_RESTORE_LDS_EXEC_HI
+	s_mov_b32	exec_hi, 0x00000000
+	s_branch	L_RESTORE_LDS_NORMAL
+L_ENABLE_RESTORE_LDS_EXEC_HI:
+	s_mov_b32	exec_hi, 0xFFFFFFFF
+L_RESTORE_LDS_NORMAL:
+	s_getreg_b32	s_restore_alloc_size, hwreg(HW_REG_LDS_ALLOC,SQ_WAVE_LDS_ALLOC_LDS_SIZE_SHIFT,SQ_WAVE_LDS_ALLOC_LDS_SIZE_SIZE)
+	s_and_b32	s_restore_alloc_size, s_restore_alloc_size, 0xFFFFFFFF	//lds_size is zero?
+	s_cbranch_scc0	L_RESTORE_VGPR						//no lds used? jump to L_RESTORE_VGPR
+	s_lshl_b32	s_restore_alloc_size, s_restore_alloc_size, SQ_WAVE_LDS_ALLOC_GRANULARITY
+	s_mov_b32	s_restore_buf_rsrc2, s_restore_alloc_size		//NUM_RECORDS in bytes
+
+	// LDS at offset: size(VGPR)+size(SVGPR)+SIZE(SGPR)+SIZE(HWREG)
+	//
+	get_vgpr_size_bytes(s_restore_mem_offset, s_restore_size)
+	get_svgpr_size_bytes(s_restore_tmp)
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, s_restore_tmp
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, get_sgpr_size_bytes()
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, get_hwreg_size_bytes()
+
+	s_mov_b32	s_restore_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+	s_lshr_b32	m0, s_restore_size, S_WAVE_SIZE
+	s_and_b32	m0, m0, 1
+	s_cmp_eq_u32	m0, 1
+	s_mov_b32	m0, 0x0
+	s_cbranch_scc1	L_RESTORE_LDS_LOOP_W64
+
+L_RESTORE_LDS_LOOP_W32:
+#if HAVE_BUFFER_LDS_LOAD
+	buffer_load_dword	v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset lds:1	// first 64DW
+#else
+	buffer_load_dword       v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset
+	S_WAITCNT_0
+	ds_store_addtid_b32     v0
+#endif
+	s_add_u32	m0, m0, 128						// 128 DW
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, 128		//mem offset increased by 128DW
+	s_cmp_lt_u32	m0, s_restore_alloc_size				//scc=(m0 < s_restore_alloc_size) ? 1 : 0
+	s_cbranch_scc1	L_RESTORE_LDS_LOOP_W32					//LDS restore is complete?
+	s_branch	L_RESTORE_VGPR
+
+L_RESTORE_LDS_LOOP_W64:
+#if HAVE_BUFFER_LDS_LOAD
+	buffer_load_dword	v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset lds:1	// first 64DW
+#else
+	buffer_load_dword       v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset
+	S_WAITCNT_0
+	ds_store_addtid_b32     v0
+#endif
+	s_add_u32	m0, m0, 256						// 256 DW
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, 256		//mem offset increased by 256DW
+	s_cmp_lt_u32	m0, s_restore_alloc_size				//scc=(m0 < s_restore_alloc_size) ? 1 : 0
+	s_cbranch_scc1	L_RESTORE_LDS_LOOP_W64					//LDS restore is complete?
+
+	/* restore VGPRs */
+L_RESTORE_VGPR:
+	// VGPR SR memory offset : 0
+	s_mov_b32	s_restore_mem_offset, 0x0
+ 	s_mov_b32	exec_lo, 0xFFFFFFFF					//need every thread from now on
+	s_lshr_b32	m0, s_restore_size, S_WAVE_SIZE
+	s_and_b32	m0, m0, 1
+	s_cmp_eq_u32	m0, 1
+	s_cbranch_scc1	L_ENABLE_RESTORE_VGPR_EXEC_HI
+	s_mov_b32	exec_hi, 0x00000000
+	s_branch	L_RESTORE_VGPR_NORMAL
+L_ENABLE_RESTORE_VGPR_EXEC_HI:
+	s_mov_b32	exec_hi, 0xFFFFFFFF
+L_RESTORE_VGPR_NORMAL:
+	s_getreg_b32	s_restore_alloc_size, hwreg(HW_REG_GPR_ALLOC,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SIZE)
+	s_add_u32	s_restore_alloc_size, s_restore_alloc_size, 1
+	s_lshl_b32	s_restore_alloc_size, s_restore_alloc_size, 2		//Number of VGPRs = (vgpr_size + 1) * 4    (non-zero value)
+	//determine it is wave32 or wave64
+	s_lshr_b32	m0, s_restore_size, S_WAVE_SIZE
+	s_and_b32	m0, m0, 1
+	s_cmp_eq_u32	m0, 1
+	s_cbranch_scc1	L_RESTORE_VGPR_WAVE64
+
+	s_mov_b32	s_restore_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+	// VGPR load using dw burst
+	s_mov_b32	s_restore_mem_offset_save, s_restore_mem_offset		// restore start with v1, v0 will be the last
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, 128*4
+	s_mov_b32	m0, 4							//VGPR initial index value = 4
+	s_cmp_lt_u32	m0, s_restore_alloc_size
+	s_cbranch_scc0	L_RESTORE_SGPR
+
+L_RESTORE_VGPR_WAVE32_LOOP:
+	buffer_load_dword	v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset V_COHERENCE
+	buffer_load_dword	v1, v0, s_restore_buf_rsrc0, s_restore_mem_offset V_COHERENCE offset:128
+	buffer_load_dword	v2, v0, s_restore_buf_rsrc0, s_restore_mem_offset V_COHERENCE offset:128*2
+	buffer_load_dword	v3, v0, s_restore_buf_rsrc0, s_restore_mem_offset V_COHERENCE offset:128*3
+	S_WAITCNT_0
+	v_movreld_b32	v0, v0							//v[0+m0] = v0
+	v_movreld_b32	v1, v1
+	v_movreld_b32	v2, v2
+	v_movreld_b32	v3, v3
+	s_add_u32	m0, m0, 4						//next vgpr index
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, 128*4	//every buffer_load_dword does 128 bytes
+	s_cmp_lt_u32	m0, s_restore_alloc_size				//scc = (m0 < s_restore_alloc_size) ? 1 : 0
+	s_cbranch_scc1	L_RESTORE_VGPR_WAVE32_LOOP				//VGPR restore (except v0) is complete?
+
+	/* VGPR restore on v0 */
+	buffer_load_dword	v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save V_COHERENCE
+	buffer_load_dword	v1, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save V_COHERENCE offset:128
+	buffer_load_dword	v2, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save V_COHERENCE offset:128*2
+	buffer_load_dword	v3, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save V_COHERENCE offset:128*3
+	S_WAITCNT_0
+
+	s_branch	L_RESTORE_SGPR
+
+L_RESTORE_VGPR_WAVE64:
+	s_mov_b32	s_restore_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+	// VGPR load using dw burst
+	s_mov_b32	s_restore_mem_offset_save, s_restore_mem_offset		// restore start with v4, v0 will be the last
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, 256*4
+	s_mov_b32	m0, 4							//VGPR initial index value = 4
+	s_cmp_lt_u32	m0, s_restore_alloc_size
+	s_cbranch_scc0	L_RESTORE_SHARED_VGPR
+
+L_RESTORE_VGPR_WAVE64_LOOP:
+	buffer_load_dword	v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset V_COHERENCE
+	buffer_load_dword	v1, v0, s_restore_buf_rsrc0, s_restore_mem_offset V_COHERENCE offset:256
+	buffer_load_dword	v2, v0, s_restore_buf_rsrc0, s_restore_mem_offset V_COHERENCE offset:256*2
+	buffer_load_dword	v3, v0, s_restore_buf_rsrc0, s_restore_mem_offset V_COHERENCE offset:256*3
+	S_WAITCNT_0
+	v_movreld_b32	v0, v0							//v[0+m0] = v0
+	v_movreld_b32	v1, v1
+	v_movreld_b32	v2, v2
+	v_movreld_b32	v3, v3
+	s_add_u32	m0, m0, 4						//next vgpr index
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, 256*4	//every buffer_load_dword does 256 bytes
+	s_cmp_lt_u32	m0, s_restore_alloc_size				//scc = (m0 < s_restore_alloc_size) ? 1 : 0
+	s_cbranch_scc1	L_RESTORE_VGPR_WAVE64_LOOP				//VGPR restore (except v0) is complete?
+
+L_RESTORE_SHARED_VGPR:
+	//Below part will be the restore shared vgpr part (new for gfx10)
+	s_getreg_b32	s_restore_alloc_size, hwreg(HW_REG_LDS_ALLOC,SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SHIFT,SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SIZE)	//shared_vgpr_size
+	s_and_b32	s_restore_alloc_size, s_restore_alloc_size, 0xFFFFFFFF	//shared_vgpr_size is zero?
+	s_cbranch_scc0	L_RESTORE_V0						//no shared_vgpr used?
+	s_lshl_b32	s_restore_alloc_size, s_restore_alloc_size, 3		//Number of SHARED_VGPRs = shared_vgpr_size * 8    (non-zero value)
+	//m0 now has the value of normal vgpr count, just add the m0 with shared_vgpr count to get the total count.
+	//restore shared_vgpr will start from the index of m0
+	s_add_u32	s_restore_alloc_size, s_restore_alloc_size, m0
+	s_mov_b32	exec_lo, 0xFFFFFFFF
+	s_mov_b32	exec_hi, 0x00000000
+L_RESTORE_SHARED_VGPR_WAVE64_LOOP:
+	buffer_load_dword	v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset V_COHERENCE
+	S_WAITCNT_0
+	v_movreld_b32	v0, v0							//v[0+m0] = v0
+	s_add_u32	m0, m0, 1						//next vgpr index
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, 128
+	s_cmp_lt_u32	m0, s_restore_alloc_size				//scc = (m0 < s_restore_alloc_size) ? 1 : 0
+	s_cbranch_scc1	L_RESTORE_SHARED_VGPR_WAVE64_LOOP			//VGPR restore (except v0) is complete?
+
+	s_mov_b32	exec_hi, 0xFFFFFFFF					//restore back exec_hi before restoring V0!!
+
+	/* VGPR restore on v0 */
+L_RESTORE_V0:
+	buffer_load_dword	v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save V_COHERENCE
+	buffer_load_dword	v1, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save V_COHERENCE offset:256
+	buffer_load_dword	v2, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save V_COHERENCE offset:256*2
+	buffer_load_dword	v3, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save V_COHERENCE offset:256*3
+	S_WAITCNT_0
+
+	/* restore SGPRs */
+	//will be 2+8+16*6
+	// SGPR SR memory offset : size(VGPR)+size(SVGPR)
+L_RESTORE_SGPR:
+	get_vgpr_size_bytes(s_restore_mem_offset, s_restore_size)
+	get_svgpr_size_bytes(s_restore_tmp)
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, s_restore_tmp
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, get_sgpr_size_bytes()
+	s_sub_u32	s_restore_mem_offset, s_restore_mem_offset, 20*4	//s108~s127 is not saved
+
+	s_mov_b32	s_restore_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+	s_mov_b32	m0, s_sgpr_save_num
+
+	read_4sgpr_from_mem(s0, s_restore_buf_rsrc0, s_restore_mem_offset)
+	S_WAITCNT_0
+
+	s_sub_u32	m0, m0, 4						// Restore from S[0] to S[104]
+	s_nop		0							// hazard SALU M0=> S_MOVREL
+
+	s_movreld_b64	s0, s0							//s[0+m0] = s0
+	s_movreld_b64	s2, s2
+
+	read_8sgpr_from_mem(s0, s_restore_buf_rsrc0, s_restore_mem_offset)
+	S_WAITCNT_0
+
+	s_sub_u32	m0, m0, 8						// Restore from S[0] to S[96]
+	s_nop		0							// hazard SALU M0=> S_MOVREL
+
+	s_movreld_b64	s0, s0							//s[0+m0] = s0
+	s_movreld_b64	s2, s2
+	s_movreld_b64	s4, s4
+	s_movreld_b64	s6, s6
+
+ L_RESTORE_SGPR_LOOP:
+	read_16sgpr_from_mem(s0, s_restore_buf_rsrc0, s_restore_mem_offset)
+	S_WAITCNT_0
+
+	s_sub_u32	m0, m0, 16						// Restore from S[n] to S[0]
+	s_nop		0							// hazard SALU M0=> S_MOVREL
+
+	s_movreld_b64	s0, s0							//s[0+m0] = s0
+	s_movreld_b64	s2, s2
+	s_movreld_b64	s4, s4
+	s_movreld_b64	s6, s6
+	s_movreld_b64	s8, s8
+	s_movreld_b64	s10, s10
+	s_movreld_b64	s12, s12
+	s_movreld_b64	s14, s14
+
+	s_cmp_eq_u32	m0, 0							//scc = (m0 < s_sgpr_save_num) ? 1 : 0
+	s_cbranch_scc0	L_RESTORE_SGPR_LOOP
+
+	// s_barrier with MODE.DEBUG_EN=1, STATUS.PRIV=1 incorrectly asserts debug exception.
+	// Clear DEBUG_EN before and restore MODE after the barrier.
+	s_setreg_imm32_b32	hwreg(HW_REG_MODE), 0
+	s_barrier								//barrier to ensure the readiness of LDS before access attemps from any other wave in the same TG
+
+	/* restore HW registers */
+L_RESTORE_HWREG:
+	// HWREG SR memory offset : size(VGPR)+size(SVGPR)+size(SGPR)
+	get_vgpr_size_bytes(s_restore_mem_offset, s_restore_size)
+	get_svgpr_size_bytes(s_restore_tmp)
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, s_restore_tmp
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, get_sgpr_size_bytes()
+
+	s_mov_b32	s_restore_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+	read_hwreg_from_mem(s_restore_m0, s_restore_buf_rsrc0, s_restore_mem_offset)
+	read_hwreg_from_mem(s_restore_pc_lo, s_restore_buf_rsrc0, s_restore_mem_offset)
+	read_hwreg_from_mem(s_restore_pc_hi, s_restore_buf_rsrc0, s_restore_mem_offset)
+	read_hwreg_from_mem(s_restore_exec_lo, s_restore_buf_rsrc0, s_restore_mem_offset)
+	read_hwreg_from_mem(s_restore_exec_hi, s_restore_buf_rsrc0, s_restore_mem_offset)
+	read_hwreg_from_mem(s_restore_status, s_restore_buf_rsrc0, s_restore_mem_offset)
+	read_hwreg_from_mem(s_restore_trapsts, s_restore_buf_rsrc0, s_restore_mem_offset)
+	read_hwreg_from_mem(s_restore_xnack_mask, s_restore_buf_rsrc0, s_restore_mem_offset)
+	read_hwreg_from_mem(s_restore_mode, s_restore_buf_rsrc0, s_restore_mem_offset)
+	read_hwreg_from_mem(s_restore_flat_scratch, s_restore_buf_rsrc0, s_restore_mem_offset)
+	S_WAITCNT_0
+
+	s_setreg_b32	hwreg(HW_REG_SHADER_FLAT_SCRATCH_LO), s_restore_flat_scratch
+
+	read_hwreg_from_mem(s_restore_flat_scratch, s_restore_buf_rsrc0, s_restore_mem_offset)
+	S_WAITCNT_0
+
+	s_setreg_b32	hwreg(HW_REG_SHADER_FLAT_SCRATCH_HI), s_restore_flat_scratch
+
+	s_mov_b32	m0, s_restore_m0
+	s_mov_b32	exec_lo, s_restore_exec_lo
+	s_mov_b32	exec_hi, s_restore_exec_hi
+
+#if HAVE_XNACK
+	s_setreg_b32	hwreg(HW_REG_SHADER_XNACK_MASK), s_restore_xnack_mask
+#endif
+
+	// {TRAPSTS/EXCP_FLAG_PRIV}.SAVE_CONTEXT and HOST_TRAP may have changed.
+	// Only restore the other fields to avoid clobbering them.
+	s_setreg_b32	hwreg(S_TRAPSTS_HWREG, 0, S_TRAPSTS_RESTORE_PART_1_SIZE), s_restore_trapsts
+	s_lshr_b32	s_restore_trapsts, s_restore_trapsts, S_TRAPSTS_RESTORE_PART_2_SHIFT
+	s_setreg_b32	hwreg(S_TRAPSTS_HWREG, S_TRAPSTS_RESTORE_PART_2_SHIFT, S_TRAPSTS_RESTORE_PART_2_SIZE), s_restore_trapsts
+
+if S_TRAPSTS_RESTORE_PART_3_SIZE > 0
+	s_lshr_b32	s_restore_trapsts, s_restore_trapsts, S_TRAPSTS_RESTORE_PART_3_SHIFT - S_TRAPSTS_RESTORE_PART_2_SHIFT
+	s_setreg_b32	hwreg(S_TRAPSTS_HWREG, S_TRAPSTS_RESTORE_PART_3_SHIFT, S_TRAPSTS_RESTORE_PART_3_SIZE), s_restore_trapsts
+end
+
+	s_setreg_b32	hwreg(HW_REG_MODE), s_restore_mode
+
+	// Restore trap temporaries 4-11, 13 initialized by SPI debug dispatch logic
+	// ttmp SR memory offset : size(VGPR)+size(SVGPR)+size(SGPR)+0x40
+	get_vgpr_size_bytes(s_restore_ttmps_lo, s_restore_size)
+	get_svgpr_size_bytes(s_restore_ttmps_hi)
+	s_add_u32	s_restore_ttmps_lo, s_restore_ttmps_lo, s_restore_ttmps_hi
+	s_add_u32	s_restore_ttmps_lo, s_restore_ttmps_lo, get_sgpr_size_bytes()
+	s_add_u32	s_restore_ttmps_lo, s_restore_ttmps_lo, s_restore_buf_rsrc0
+	s_addc_u32	s_restore_ttmps_hi, s_restore_buf_rsrc1, 0x0
+	s_and_b32	s_restore_ttmps_hi, s_restore_ttmps_hi, 0xFFFF
+	s_load_dwordx4	[ttmp4, ttmp5, ttmp6, ttmp7], [s_restore_ttmps_lo, s_restore_ttmps_hi], 0x50 S_COHERENCE
+	s_load_dwordx4	[ttmp8, ttmp9, ttmp10, ttmp11], [s_restore_ttmps_lo, s_restore_ttmps_hi], 0x60 S_COHERENCE
+	s_load_dword	ttmp13, [s_restore_ttmps_lo, s_restore_ttmps_hi], 0x74 S_COHERENCE
+	S_WAITCNT_0
+
+#if HAVE_XNACK
+	restore_ib_sts(s_restore_tmp, s_restore_m0)
+#endif
+
+	s_and_b32	s_restore_pc_hi, s_restore_pc_hi, 0x0000ffff		//pc[47:32] //Do it here in order not to affect STATUS
+	s_and_b64	exec, exec, exec					// Restore STATUS.EXECZ, not writable by s_setreg_b32
+	s_and_b64	vcc, vcc, vcc						// Restore STATUS.VCCZ, not writable by s_setreg_b32
+
+#if SW_SA_TRAP
+	// If traps are enabled then return to the shader with PRIV=0.
+	// Otherwise retain PRIV=1 for subsequent context save requests.
+	s_getreg_b32	s_restore_tmp, hwreg(HW_REG_STATUS)
+	s_bitcmp1_b32	s_restore_tmp, SQ_WAVE_STATUS_TRAP_EN_SHIFT
+	s_cbranch_scc1	L_RETURN_WITHOUT_PRIV
+
+	s_setreg_b32	hwreg(HW_REG_STATUS), s_restore_status			// SCC is included, which is changed by previous salu
+	s_setpc_b64	[s_restore_pc_lo, s_restore_pc_hi]
+L_RETURN_WITHOUT_PRIV:
+#endif
+
+	s_setreg_b32	hwreg(S_STATUS_HWREG), s_restore_status			// SCC is included, which is changed by previous salu
+
+	s_rfe_b64	s_restore_pc_lo						//Return to the main shader program and resume execution
+
+L_END_PGM:
+	s_endpgm_saved
+end
+
+function write_hwreg_to_mem(s, s_rsrc, s_mem_offset)
+#if NO_SQC_STORE
+	// Copy into VGPR for later TCP store.
+	v_writelane_b32	v2, s, m0
+	s_add_u32	m0, m0, 0x1
+#else
+	s_mov_b32	exec_lo, m0
+	s_mov_b32	m0, s_mem_offset
+	s_buffer_store_dword	s, s_rsrc, m0 S_COHERENCE
+	s_add_u32	s_mem_offset, s_mem_offset, 4
+	s_mov_b32	m0, exec_lo
+#endif
+end
+
+
+function write_16sgpr_to_mem(s, s_rsrc, s_mem_offset)
+#if NO_SQC_STORE
+	// Copy into VGPR for later TCP store.
+	for var sgpr_idx = 0; sgpr_idx < 16; sgpr_idx ++
+		v_writelane_b32	v2, s[sgpr_idx], ttmp13
+		s_add_u32	ttmp13, ttmp13, 0x1
+	end
+#else
+	s_buffer_store_dwordx4	s[0], s_rsrc, 0 S_COHERENCE
+	s_buffer_store_dwordx4	s[4], s_rsrc, 16 S_COHERENCE
+	s_buffer_store_dwordx4	s[8], s_rsrc, 32 S_COHERENCE
+	s_buffer_store_dwordx4	s[12], s_rsrc, 48 S_COHERENCE
+	s_add_u32	s_rsrc[0], s_rsrc[0], 4*16
+	s_addc_u32	s_rsrc[1], s_rsrc[1], 0x0
+#endif
+end
+
+function write_12sgpr_to_mem(s, s_rsrc, s_mem_offset)
+#if NO_SQC_STORE
+	// Copy into VGPR for later TCP store.
+	for var sgpr_idx = 0; sgpr_idx < 12; sgpr_idx ++
+		v_writelane_b32	v2, s[sgpr_idx], ttmp13
+		s_add_u32	ttmp13, ttmp13, 0x1
+	end
+#else
+	s_buffer_store_dwordx4	s[0], s_rsrc, 0 S_COHERENCE
+	s_buffer_store_dwordx4	s[4], s_rsrc, 16 S_COHERENCE
+	s_buffer_store_dwordx4	s[8], s_rsrc, 32 S_COHERENCE
+	s_add_u32	s_rsrc[0], s_rsrc[0], 4*12
+	s_addc_u32	s_rsrc[1], s_rsrc[1], 0x0
+#endif
+end
+
+function read_hwreg_from_mem(s, s_rsrc, s_mem_offset)
+	s_buffer_load_dword	s, s_rsrc, s_mem_offset S_COHERENCE
+	s_add_u32	s_mem_offset, s_mem_offset, 4
+end
+
+function read_16sgpr_from_mem(s, s_rsrc, s_mem_offset)
+	s_sub_u32	s_mem_offset, s_mem_offset, 4*16
+	s_buffer_load_dwordx16	s, s_rsrc, s_mem_offset S_COHERENCE
+end
+
+function read_8sgpr_from_mem(s, s_rsrc, s_mem_offset)
+	s_sub_u32	s_mem_offset, s_mem_offset, 4*8
+	s_buffer_load_dwordx8	s, s_rsrc, s_mem_offset S_COHERENCE
+end
+
+function read_4sgpr_from_mem(s, s_rsrc, s_mem_offset)
+	s_sub_u32	s_mem_offset, s_mem_offset, 4*4
+	s_buffer_load_dwordx4	s, s_rsrc, s_mem_offset S_COHERENCE
+end
+
+#if SAVE_AFTER_XNACK_ERROR
+function check_if_tcp_store_ok
+	// If TRAPSTS.XNACK_ERROR=1 then TCP stores will fail.
+	s_getreg_b32 s_save_tmp, hwreg(HW_REG_TRAPSTS)
+	s_andn2_b32 s_save_tmp, SQ_WAVE_TRAPSTS_XNACK_ERROR_MASK, s_save_tmp
+
+L_TCP_STORE_CHECK_DONE:
+end
+
+function write_vgpr_to_mem_with_sqc(vgpr, n_lanes, s_rsrc, s_mem_offset)
+	s_mov_b32 s4, 0
+
+L_WRITE_VGPR_LANE_LOOP:
+	for var lane = 0; lane < 4; ++lane
+		v_readlane_b32 s[lane], vgpr, s4
+		s_add_u32 s4, s4, 1
+	end
+
+	s_buffer_store_dwordx4 s[0:3], s_rsrc, s_mem_offset glc:1
+
+	s_add_u32 s_mem_offset, s_mem_offset, 0x10
+	s_cmp_eq_u32 s4, n_lanes
+	s_cbranch_scc0 L_WRITE_VGPR_LANE_LOOP
+end
+
+function write_vgprs_to_mem_with_sqc_w32(vgpr0, n_vgprs, s_rsrc, s_mem_offset)
+	for var vgpr = 0; vgpr < n_vgprs; ++vgpr
+		write_vgpr_to_mem_with_sqc(vgpr0[vgpr], 32, s_rsrc, s_mem_offset)
+	end
+end
+
+function write_vgprs_to_mem_with_sqc_w64(vgpr0, n_vgprs, s_rsrc, s_mem_offset)
+	for var vgpr = 0; vgpr < n_vgprs; ++vgpr
+		write_vgpr_to_mem_with_sqc(vgpr0[vgpr], 64, s_rsrc, s_mem_offset)
+	end
+end
+#endif
+
+function get_vgpr_size_bytes(s_vgpr_size_byte, s_size)
+	s_getreg_b32	s_vgpr_size_byte, hwreg(HW_REG_GPR_ALLOC,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SIZE)
+	s_add_u32	s_vgpr_size_byte, s_vgpr_size_byte, 1
+	s_bitcmp1_b32	s_size, S_WAVE_SIZE
+	s_cbranch_scc1	L_ENABLE_SHIFT_W64
+	s_lshl_b32	s_vgpr_size_byte, s_vgpr_size_byte, (2+7)		//Number of VGPRs = (vgpr_size + 1) * 4 * 32 * 4   (non-zero value)
+	s_branch	L_SHIFT_DONE
+L_ENABLE_SHIFT_W64:
+	s_lshl_b32	s_vgpr_size_byte, s_vgpr_size_byte, (2+8)		//Number of VGPRs = (vgpr_size + 1) * 4 * 64 * 4   (non-zero value)
+L_SHIFT_DONE:
+end
+
+function get_svgpr_size_bytes(s_svgpr_size_byte)
+	s_getreg_b32	s_svgpr_size_byte, hwreg(HW_REG_LDS_ALLOC,SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SHIFT,SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SIZE)
+	s_lshl_b32	s_svgpr_size_byte, s_svgpr_size_byte, (3+7)
+end
+
+function get_sgpr_size_bytes
+	return 512
+end
+
+function get_hwreg_size_bytes
+	return 128
+end
+
+function get_wave_size2(s_reg)
+	s_getreg_b32	s_reg, hwreg(HW_REG_IB_STS2,SQ_WAVE_IB_STS2_WAVE64_SHIFT,SQ_WAVE_IB_STS2_WAVE64_SIZE)
+	s_lshl_b32	s_reg, s_reg, S_WAVE_SIZE
+end
+
+#if HAVE_XNACK
+function save_and_clear_ib_sts(tmp1, tmp2)
+	// Preserve and clear scalar XNACK state before issuing scalar loads.
+	// Save IB_STS.REPLAY_W64H[25], RCNT[21:16], FIRST_REPLAY[15] into
+	// unused space ttmp11[31:24].
+	s_andn2_b32	ttmp11, ttmp11, (TTMP11_SAVE_REPLAY_W64H_MASK | TTMP11_SAVE_RCNT_FIRST_REPLAY_MASK)
+	s_getreg_b32	tmp1, hwreg(HW_REG_IB_STS)
+	s_and_b32	tmp2, tmp1, SQ_WAVE_IB_STS_REPLAY_W64H_MASK
+	s_lshl_b32	tmp2, tmp2, (TTMP11_SAVE_REPLAY_W64H_SHIFT - SQ_WAVE_IB_STS_REPLAY_W64H_SHIFT)
+	s_or_b32	ttmp11, ttmp11, tmp2
+	s_and_b32	tmp2, tmp1, SQ_WAVE_IB_STS_RCNT_FIRST_REPLAY_MASK
+	s_lshl_b32	tmp2, tmp2, (TTMP11_SAVE_RCNT_FIRST_REPLAY_SHIFT - SQ_WAVE_IB_STS_FIRST_REPLAY_SHIFT)
+	s_or_b32	ttmp11, ttmp11, tmp2
+	s_andn2_b32	tmp1, tmp1, (SQ_WAVE_IB_STS_REPLAY_W64H_MASK | SQ_WAVE_IB_STS_RCNT_FIRST_REPLAY_MASK)
+	s_setreg_b32	hwreg(HW_REG_IB_STS), tmp1
+end
+
+function restore_ib_sts(tmp1, tmp2)
+	s_lshr_b32	tmp1, ttmp11, (TTMP11_SAVE_RCNT_FIRST_REPLAY_SHIFT - SQ_WAVE_IB_STS_FIRST_REPLAY_SHIFT)
+	s_and_b32	tmp2, tmp1, SQ_WAVE_IB_STS_RCNT_FIRST_REPLAY_MASK
+	s_lshr_b32	tmp1, ttmp11, (TTMP11_SAVE_REPLAY_W64H_SHIFT - SQ_WAVE_IB_STS_REPLAY_W64H_SHIFT)
+	s_and_b32	tmp1, tmp1, SQ_WAVE_IB_STS_REPLAY_W64H_MASK
+	s_or_b32	tmp1, tmp1, tmp2
+	s_setreg_b32	hwreg(HW_REG_IB_STS), tmp1
+end
+#endif
diff --git a/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx12.asm b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx12.asm
new file mode 100644
index 000000000000..5a1a1b1f897f
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx12.asm
@@ -0,0 +1,1136 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/* To compile this assembly code:
+ *
+ * gfx12:
+ *   cpp -DASIC_FAMILY=CHIP_GFX12 cwsr_trap_handler_gfx12.asm -P -o gfx12.sp3
+ *   sp3 gfx12.sp3 -hex gfx12.hex
+ */
+
+#define CHIP_GFX12 37
+
+#define SINGLE_STEP_MISSED_WORKAROUND 1	//workaround for lost TRAP_AFTER_INST exception when SAVECTX raised
+#define HAVE_VALU_SGPR_HAZARD (ASIC_FAMILY == CHIP_GFX12)
+
+var SQ_WAVE_STATE_PRIV_BARRIER_COMPLETE_MASK	= 0x4
+var SQ_WAVE_STATE_PRIV_SCC_SHIFT		= 9
+var SQ_WAVE_STATE_PRIV_SYS_PRIO_MASK		= 0xC00
+var SQ_WAVE_STATE_PRIV_HALT_MASK		= 0x4000
+var SQ_WAVE_STATE_PRIV_POISON_ERR_MASK		= 0x8000
+var SQ_WAVE_STATE_PRIV_POISON_ERR_SHIFT		= 15
+var SQ_WAVE_STATUS_WAVE64_SHIFT			= 29
+var SQ_WAVE_STATUS_WAVE64_SIZE			= 1
+var SQ_WAVE_STATUS_NO_VGPRS_SHIFT		= 24
+var SQ_WAVE_STATE_PRIV_ALWAYS_CLEAR_MASK	= SQ_WAVE_STATE_PRIV_SYS_PRIO_MASK|SQ_WAVE_STATE_PRIV_POISON_ERR_MASK
+var S_SAVE_PC_HI_TRAP_ID_MASK			= 0xF0000000
+
+var SQ_WAVE_LDS_ALLOC_LDS_SIZE_SHIFT		= 12
+var SQ_WAVE_LDS_ALLOC_LDS_SIZE_SIZE		= 9
+var SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SIZE		= 8
+var SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT		= 12
+var SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SHIFT	= 24
+var SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SIZE	= 4
+var SQ_WAVE_LDS_ALLOC_GRANULARITY		= 9
+
+var SQ_WAVE_EXCP_FLAG_PRIV_ADDR_WATCH_MASK	= 0xF
+var SQ_WAVE_EXCP_FLAG_PRIV_MEM_VIOL_MASK	= 0x10
+var SQ_WAVE_EXCP_FLAG_PRIV_SAVE_CONTEXT_SHIFT	= 5
+var SQ_WAVE_EXCP_FLAG_PRIV_SAVE_CONTEXT_MASK	= 0x20
+var SQ_WAVE_EXCP_FLAG_PRIV_ILLEGAL_INST_MASK	= 0x40
+var SQ_WAVE_EXCP_FLAG_PRIV_ILLEGAL_INST_SHIFT	= 6
+var SQ_WAVE_EXCP_FLAG_PRIV_HOST_TRAP_MASK	= 0x80
+var SQ_WAVE_EXCP_FLAG_PRIV_HOST_TRAP_SHIFT	= 7
+var SQ_WAVE_EXCP_FLAG_PRIV_WAVE_START_MASK	= 0x100
+var SQ_WAVE_EXCP_FLAG_PRIV_WAVE_START_SHIFT	= 8
+var SQ_WAVE_EXCP_FLAG_PRIV_WAVE_END_MASK	= 0x200
+var SQ_WAVE_EXCP_FLAG_PRIV_TRAP_AFTER_INST_MASK	= 0x800
+var SQ_WAVE_TRAP_CTRL_ADDR_WATCH_MASK		= 0x80
+var SQ_WAVE_TRAP_CTRL_TRAP_AFTER_INST_MASK	= 0x200
+
+var SQ_WAVE_EXCP_FLAG_PRIV_NON_MASKABLE_EXCP_MASK= SQ_WAVE_EXCP_FLAG_PRIV_MEM_VIOL_MASK		|\
+						  SQ_WAVE_EXCP_FLAG_PRIV_ILLEGAL_INST_MASK	|\
+						  SQ_WAVE_EXCP_FLAG_PRIV_HOST_TRAP_MASK		|\
+						  SQ_WAVE_EXCP_FLAG_PRIV_WAVE_START_MASK	|\
+						  SQ_WAVE_EXCP_FLAG_PRIV_WAVE_END_MASK		|\
+						  SQ_WAVE_EXCP_FLAG_PRIV_TRAP_AFTER_INST_MASK
+var SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_1_SIZE	= SQ_WAVE_EXCP_FLAG_PRIV_SAVE_CONTEXT_SHIFT
+var SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_2_SHIFT	= SQ_WAVE_EXCP_FLAG_PRIV_ILLEGAL_INST_SHIFT
+var SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_2_SIZE	= SQ_WAVE_EXCP_FLAG_PRIV_HOST_TRAP_SHIFT - SQ_WAVE_EXCP_FLAG_PRIV_ILLEGAL_INST_SHIFT
+var SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_3_SHIFT	= SQ_WAVE_EXCP_FLAG_PRIV_WAVE_START_SHIFT
+var SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_3_SIZE	= 32 - SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_3_SHIFT
+var BARRIER_STATE_SIGNAL_OFFSET			= 16
+var BARRIER_STATE_VALID_OFFSET			= 0
+
+var TTMP11_DEBUG_TRAP_ENABLED_SHIFT		= 23
+var TTMP11_DEBUG_TRAP_ENABLED_MASK		= 0x800000
+
+// SQ_SEL_X/Y/Z/W, BUF_NUM_FORMAT_FLOAT, (0 for MUBUF stride[17:14]
+// when ADD_TID_ENABLE and BUF_DATA_FORMAT_32 for MTBUF), ADD_TID_ENABLE
+var S_SAVE_BUF_RSRC_WORD1_STRIDE		= 0x00040000
+var S_SAVE_BUF_RSRC_WORD3_MISC			= 0x10807FAC
+var S_SAVE_SPI_INIT_FIRST_WAVE_MASK		= 0x04000000
+var S_SAVE_SPI_INIT_FIRST_WAVE_SHIFT		= 26
+
+var S_SAVE_PC_HI_FIRST_WAVE_MASK		= 0x80000000
+var S_SAVE_PC_HI_FIRST_WAVE_SHIFT		= 31
+
+var s_sgpr_save_num				= 108
+
+var s_save_spi_init_lo				= exec_lo
+var s_save_spi_init_hi				= exec_hi
+var s_save_pc_lo				= ttmp0
+var s_save_pc_hi				= ttmp1
+var s_save_exec_lo				= ttmp2
+var s_save_exec_hi				= ttmp3
+var s_save_state_priv				= ttmp12
+var s_save_excp_flag_priv			= ttmp15
+var s_save_xnack_mask				= s_save_excp_flag_priv
+var s_wave_size					= ttmp7
+var s_save_buf_rsrc0				= ttmp8
+var s_save_buf_rsrc1				= ttmp9
+var s_save_buf_rsrc2				= ttmp10
+var s_save_buf_rsrc3				= ttmp11
+var s_save_mem_offset				= ttmp4
+var s_save_alloc_size				= s_save_excp_flag_priv
+var s_save_tmp					= ttmp14
+var s_save_m0					= ttmp5
+var s_save_ttmps_lo				= s_save_tmp
+var s_save_ttmps_hi				= s_save_excp_flag_priv
+
+var S_RESTORE_BUF_RSRC_WORD1_STRIDE		= S_SAVE_BUF_RSRC_WORD1_STRIDE
+var S_RESTORE_BUF_RSRC_WORD3_MISC		= S_SAVE_BUF_RSRC_WORD3_MISC
+
+var S_RESTORE_SPI_INIT_FIRST_WAVE_MASK		= 0x04000000
+var S_RESTORE_SPI_INIT_FIRST_WAVE_SHIFT		= 26
+var S_WAVE_SIZE					= 25
+
+var s_restore_spi_init_lo			= exec_lo
+var s_restore_spi_init_hi			= exec_hi
+var s_restore_mem_offset			= ttmp12
+var s_restore_alloc_size			= ttmp3
+var s_restore_tmp				= ttmp2
+var s_restore_mem_offset_save			= s_restore_tmp
+var s_restore_m0				= s_restore_alloc_size
+var s_restore_mode				= ttmp7
+var s_restore_flat_scratch			= s_restore_tmp
+var s_restore_pc_lo				= ttmp0
+var s_restore_pc_hi				= ttmp1
+var s_restore_exec_lo				= ttmp4
+var s_restore_exec_hi				= ttmp5
+var s_restore_state_priv			= ttmp14
+var s_restore_excp_flag_priv			= ttmp15
+var s_restore_xnack_mask			= ttmp13
+var s_restore_buf_rsrc0				= ttmp8
+var s_restore_buf_rsrc1				= ttmp9
+var s_restore_buf_rsrc2				= ttmp10
+var s_restore_buf_rsrc3				= ttmp11
+var s_restore_size				= ttmp6
+var s_restore_ttmps_lo				= s_restore_tmp
+var s_restore_ttmps_hi				= s_restore_alloc_size
+var s_restore_spi_init_hi_save			= s_restore_exec_hi
+
+shader main
+	asic(DEFAULT)
+	type(CS)
+	wave_size(32)
+
+	s_branch	L_SKIP_RESTORE						//NOT restore. might be a regular trap or save
+
+L_JUMP_TO_RESTORE:
+	s_branch	L_RESTORE
+
+L_SKIP_RESTORE:
+	s_getreg_b32	s_save_state_priv, hwreg(HW_REG_WAVE_STATE_PRIV)	//save STATUS since we will change SCC
+
+	// Clear SPI_PRIO: do not save with elevated priority.
+	// Clear ECC_ERR: prevents SQC store and triggers FATAL_HALT if setreg'd.
+	s_andn2_b32	s_save_state_priv, s_save_state_priv, SQ_WAVE_STATE_PRIV_ALWAYS_CLEAR_MASK
+
+	s_getreg_b32	s_save_excp_flag_priv, hwreg(HW_REG_WAVE_EXCP_FLAG_PRIV)
+
+	s_and_b32       ttmp2, s_save_state_priv, SQ_WAVE_STATE_PRIV_HALT_MASK
+	s_cbranch_scc0	L_NOT_HALTED
+
+L_HALTED:
+	// Host trap may occur while wave is halted.
+	s_and_b32	ttmp2, s_save_excp_flag_priv, SQ_WAVE_EXCP_FLAG_PRIV_HOST_TRAP_MASK
+	s_cbranch_scc1	L_FETCH_2ND_TRAP
+
+L_CHECK_SAVE:
+	s_and_b32	ttmp2, s_save_excp_flag_priv, SQ_WAVE_EXCP_FLAG_PRIV_SAVE_CONTEXT_MASK
+	s_cbranch_scc1	L_SAVE
+
+	// Wave is halted but neither host trap nor SAVECTX is raised.
+	// Caused by instruction fetch memory violation.
+	// Spin wait until context saved to prevent interrupt storm.
+	s_sleep		0x10
+	s_getreg_b32	s_save_excp_flag_priv, hwreg(HW_REG_WAVE_EXCP_FLAG_PRIV)
+	s_branch	L_CHECK_SAVE
+
+L_NOT_HALTED:
+	// Let second-level handle non-SAVECTX exception or trap.
+	// Any concurrent SAVECTX will be handled upon re-entry once halted.
+
+	// Check non-maskable exceptions. memory_violation, illegal_instruction
+	// and xnack_error exceptions always cause the wave to enter the trap
+	// handler.
+	s_and_b32	ttmp2, s_save_excp_flag_priv, SQ_WAVE_EXCP_FLAG_PRIV_NON_MASKABLE_EXCP_MASK
+	s_cbranch_scc1	L_FETCH_2ND_TRAP
+
+	// Check for maskable exceptions in trapsts.excp and trapsts.excp_hi.
+	// Maskable exceptions only cause the wave to enter the trap handler if
+	// their respective bit in mode.excp_en is set.
+	s_getreg_b32	ttmp2, hwreg(HW_REG_WAVE_EXCP_FLAG_USER)
+	s_and_b32	ttmp3, s_save_excp_flag_priv, SQ_WAVE_EXCP_FLAG_PRIV_ADDR_WATCH_MASK
+	s_cbranch_scc0	L_NOT_ADDR_WATCH
+	s_or_b32	ttmp2, ttmp2, SQ_WAVE_TRAP_CTRL_ADDR_WATCH_MASK
+
+L_NOT_ADDR_WATCH:
+	s_getreg_b32	ttmp3, hwreg(HW_REG_WAVE_TRAP_CTRL)
+	s_and_b32	ttmp2, ttmp3, ttmp2
+	s_cbranch_scc1	L_FETCH_2ND_TRAP
+
+L_CHECK_TRAP_ID:
+	// Check trap_id != 0
+	s_and_b32	ttmp2, s_save_pc_hi, S_SAVE_PC_HI_TRAP_ID_MASK
+	s_cbranch_scc1	L_FETCH_2ND_TRAP
+
+#if SINGLE_STEP_MISSED_WORKAROUND
+	// Prioritize single step exception over context save.
+	// Second-level trap will halt wave and RFE, re-entering for SAVECTX.
+	// WAVE_TRAP_CTRL is already in ttmp3.
+	s_and_b32	ttmp3, ttmp3, SQ_WAVE_TRAP_CTRL_TRAP_AFTER_INST_MASK
+	s_cbranch_scc1	L_FETCH_2ND_TRAP
+#endif
+
+	s_and_b32	ttmp2, s_save_excp_flag_priv, SQ_WAVE_EXCP_FLAG_PRIV_SAVE_CONTEXT_MASK
+	s_cbranch_scc1	L_SAVE
+
+L_FETCH_2ND_TRAP:
+	// Read second-level TBA/TMA from first-level TMA and jump if available.
+	// ttmp[2:5] and ttmp12 can be used (others hold SPI-initialized debug data)
+	// ttmp12 holds SQ_WAVE_STATUS
+	s_sendmsg_rtn_b64       [ttmp14, ttmp15], sendmsg(MSG_RTN_GET_TMA)
+	s_wait_idle
+	s_lshl_b64	[ttmp14, ttmp15], [ttmp14, ttmp15], 0x8
+
+	s_bitcmp1_b32	ttmp15, 0xF
+	s_cbranch_scc0	L_NO_SIGN_EXTEND_TMA
+	s_or_b32	ttmp15, ttmp15, 0xFFFF0000
+L_NO_SIGN_EXTEND_TMA:
+
+	s_load_dword    ttmp2, [ttmp14, ttmp15], 0x10 scope:SCOPE_SYS		// debug trap enabled flag
+	s_wait_idle
+	s_lshl_b32      ttmp2, ttmp2, TTMP11_DEBUG_TRAP_ENABLED_SHIFT
+	s_andn2_b32     ttmp11, ttmp11, TTMP11_DEBUG_TRAP_ENABLED_MASK
+	s_or_b32        ttmp11, ttmp11, ttmp2
+
+	s_load_dwordx2	[ttmp2, ttmp3], [ttmp14, ttmp15], 0x0 scope:SCOPE_SYS	// second-level TBA
+	s_wait_idle
+	s_load_dwordx2	[ttmp14, ttmp15], [ttmp14, ttmp15], 0x8 scope:SCOPE_SYS	// second-level TMA
+	s_wait_idle
+
+	s_and_b64	[ttmp2, ttmp3], [ttmp2, ttmp3], [ttmp2, ttmp3]
+	s_cbranch_scc0	L_NO_NEXT_TRAP						// second-level trap handler not been set
+	s_setpc_b64	[ttmp2, ttmp3]						// jump to second-level trap handler
+
+L_NO_NEXT_TRAP:
+	// If not caused by trap then halt wave to prevent re-entry.
+	s_and_b32	ttmp2, s_save_pc_hi, S_SAVE_PC_HI_TRAP_ID_MASK
+	s_cbranch_scc1	L_TRAP_CASE
+
+	// Host trap will not cause trap re-entry.
+	s_getreg_b32	ttmp2, hwreg(HW_REG_WAVE_EXCP_FLAG_PRIV)
+	s_and_b32	ttmp2, ttmp2, SQ_WAVE_EXCP_FLAG_PRIV_HOST_TRAP_MASK
+	s_cbranch_scc1	L_EXIT_TRAP
+	s_or_b32	s_save_state_priv, s_save_state_priv, SQ_WAVE_STATE_PRIV_HALT_MASK
+
+	// If the PC points to S_ENDPGM then context save will fail if STATE_PRIV.HALT is set.
+	// Rewind the PC to prevent this from occurring.
+	s_sub_u32	ttmp0, ttmp0, 0x8
+	s_subb_u32	ttmp1, ttmp1, 0x0
+
+	s_branch	L_EXIT_TRAP
+
+L_TRAP_CASE:
+	// Advance past trap instruction to prevent re-entry.
+	s_add_u32	ttmp0, ttmp0, 0x4
+	s_addc_u32	ttmp1, ttmp1, 0x0
+
+L_EXIT_TRAP:
+	s_and_b32	ttmp1, ttmp1, 0xFFFF
+
+	// Restore SQ_WAVE_STATUS.
+	s_and_b64	exec, exec, exec					// Restore STATUS.EXECZ, not writable by s_setreg_b32
+	s_and_b64	vcc, vcc, vcc						// Restore STATUS.VCCZ, not writable by s_setreg_b32
+
+	// STATE_PRIV.BARRIER_COMPLETE may have changed since we read it.
+	// Only restore fields which the trap handler changes.
+	s_lshr_b32	s_save_state_priv, s_save_state_priv, SQ_WAVE_STATE_PRIV_SCC_SHIFT
+	s_setreg_b32	hwreg(HW_REG_WAVE_STATE_PRIV, SQ_WAVE_STATE_PRIV_SCC_SHIFT, \
+		SQ_WAVE_STATE_PRIV_POISON_ERR_SHIFT - SQ_WAVE_STATE_PRIV_SCC_SHIFT + 1), s_save_state_priv
+
+	s_rfe_b64	[ttmp0, ttmp1]
+
+L_SAVE:
+	// If VGPRs have been deallocated then terminate the wavefront.
+	// It has no remaining program to run and cannot save without VGPRs.
+	s_getreg_b32	s_save_tmp, hwreg(HW_REG_WAVE_STATUS)
+	s_bitcmp1_b32	s_save_tmp, SQ_WAVE_STATUS_NO_VGPRS_SHIFT
+	s_cbranch_scc0	L_HAVE_VGPRS
+	s_endpgm
+L_HAVE_VGPRS:
+
+	s_and_b32	s_save_pc_hi, s_save_pc_hi, 0x0000ffff			//pc[47:32]
+	s_mov_b32	s_save_tmp, 0
+	s_setreg_b32	hwreg(HW_REG_WAVE_EXCP_FLAG_PRIV, SQ_WAVE_EXCP_FLAG_PRIV_SAVE_CONTEXT_SHIFT, 1), s_save_tmp	//clear saveCtx bit
+
+	/* inform SPI the readiness and wait for SPI's go signal */
+	s_mov_b32	s_save_exec_lo, exec_lo					//save EXEC and use EXEC for the go signal from SPI
+	s_mov_b32	s_save_exec_hi, exec_hi
+	s_mov_b64	exec, 0x0						//clear EXEC to get ready to receive
+
+	s_sendmsg_rtn_b64       [exec_lo, exec_hi], sendmsg(MSG_RTN_SAVE_WAVE)
+	s_wait_idle
+
+	// Save first_wave flag so we can clear high bits of save address.
+	s_and_b32	s_save_tmp, s_save_spi_init_hi, S_SAVE_SPI_INIT_FIRST_WAVE_MASK
+	s_lshl_b32	s_save_tmp, s_save_tmp, (S_SAVE_PC_HI_FIRST_WAVE_SHIFT - S_SAVE_SPI_INIT_FIRST_WAVE_SHIFT)
+	s_or_b32	s_save_pc_hi, s_save_pc_hi, s_save_tmp
+
+	// Trap temporaries must be saved via VGPR but all VGPRs are in use.
+	// There is no ttmp space to hold the resource constant for VGPR save.
+	// Save v0 by itself since it requires only two SGPRs.
+	s_mov_b32	s_save_ttmps_lo, exec_lo
+	s_and_b32	s_save_ttmps_hi, exec_hi, 0xFFFF
+	s_mov_b32	exec_lo, 0xFFFFFFFF
+	s_mov_b32	exec_hi, 0xFFFFFFFF
+	global_store_dword_addtid	v0, [s_save_ttmps_lo, s_save_ttmps_hi] scope:SCOPE_SYS
+	v_mov_b32	v0, 0x0
+	s_mov_b32	exec_lo, s_save_ttmps_lo
+	s_mov_b32	exec_hi, s_save_ttmps_hi
+
+	// Save trap temporaries 4-11, 13 initialized by SPI debug dispatch logic
+	// ttmp SR memory offset : size(VGPR)+size(SVGPR)+size(SGPR)+0x40
+	get_wave_size2(s_save_ttmps_hi)
+	get_vgpr_size_bytes(s_save_ttmps_lo, s_save_ttmps_hi)
+	get_svgpr_size_bytes(s_save_ttmps_hi)
+	s_add_u32	s_save_ttmps_lo, s_save_ttmps_lo, s_save_ttmps_hi
+	s_and_b32	s_save_ttmps_hi, s_save_spi_init_hi, 0xFFFF
+	s_add_u32	s_save_ttmps_lo, s_save_ttmps_lo, get_sgpr_size_bytes()
+	s_add_u32	s_save_ttmps_lo, s_save_ttmps_lo, s_save_spi_init_lo
+	s_addc_u32	s_save_ttmps_hi, s_save_ttmps_hi, 0x0
+
+	v_writelane_b32	v0, ttmp4, 0x4
+	v_writelane_b32	v0, ttmp5, 0x5
+	v_writelane_b32	v0, ttmp6, 0x6
+	v_writelane_b32	v0, ttmp7, 0x7
+	v_writelane_b32	v0, ttmp8, 0x8
+	v_writelane_b32	v0, ttmp9, 0x9
+	v_writelane_b32	v0, ttmp10, 0xA
+	v_writelane_b32	v0, ttmp11, 0xB
+	v_writelane_b32	v0, ttmp13, 0xD
+	v_writelane_b32	v0, exec_lo, 0xE
+	v_writelane_b32	v0, exec_hi, 0xF
+	valu_sgpr_hazard()
+
+	s_mov_b32	exec_lo, 0x3FFF
+	s_mov_b32	exec_hi, 0x0
+	global_store_dword_addtid	v0, [s_save_ttmps_lo, s_save_ttmps_hi] offset:0x40 scope:SCOPE_SYS
+	v_readlane_b32	ttmp14, v0, 0xE
+	v_readlane_b32	ttmp15, v0, 0xF
+	s_mov_b32	exec_lo, ttmp14
+	s_mov_b32	exec_hi, ttmp15
+
+	/* setup Resource Contants */
+	s_mov_b32	s_save_buf_rsrc0, s_save_spi_init_lo			//base_addr_lo
+	s_and_b32	s_save_buf_rsrc1, s_save_spi_init_hi, 0x0000FFFF	//base_addr_hi
+	s_or_b32	s_save_buf_rsrc1, s_save_buf_rsrc1, S_SAVE_BUF_RSRC_WORD1_STRIDE
+	s_mov_b32	s_save_buf_rsrc2, 0					//NUM_RECORDS initial value = 0 (in bytes) although not neccessarily inited
+	s_mov_b32	s_save_buf_rsrc3, S_SAVE_BUF_RSRC_WORD3_MISC
+
+	s_mov_b32	s_save_m0, m0
+
+	/* global mem offset */
+	s_mov_b32	s_save_mem_offset, 0x0
+	get_wave_size2(s_wave_size)
+
+	/* save first 4 VGPRs, needed for SGPR save */
+	s_mov_b32	exec_lo, 0xFFFFFFFF					//need every thread from now on
+	s_lshr_b32	m0, s_wave_size, S_WAVE_SIZE
+	s_and_b32	m0, m0, 1
+	s_cmp_eq_u32	m0, 1
+	s_cbranch_scc1	L_ENABLE_SAVE_4VGPR_EXEC_HI
+	s_mov_b32	exec_hi, 0x00000000
+	s_branch	L_SAVE_4VGPR_WAVE32
+L_ENABLE_SAVE_4VGPR_EXEC_HI:
+	s_mov_b32	exec_hi, 0xFFFFFFFF
+	s_branch	L_SAVE_4VGPR_WAVE64
+L_SAVE_4VGPR_WAVE32:
+	s_mov_b32	s_save_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+	// VGPR Allocated in 4-GPR granularity
+
+	buffer_store_dword	v1, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:128
+	buffer_store_dword	v2, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:128*2
+	buffer_store_dword	v3, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:128*3
+	s_branch	L_SAVE_HWREG
+
+L_SAVE_4VGPR_WAVE64:
+	s_mov_b32	s_save_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+	// VGPR Allocated in 4-GPR granularity
+
+	buffer_store_dword	v1, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:256
+	buffer_store_dword	v2, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:256*2
+	buffer_store_dword	v3, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:256*3
+
+	/* save HW registers */
+
+L_SAVE_HWREG:
+	// HWREG SR memory offset : size(VGPR)+size(SVGPR)+size(SGPR)
+	get_vgpr_size_bytes(s_save_mem_offset, s_wave_size)
+	get_svgpr_size_bytes(s_save_tmp)
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, s_save_tmp
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, get_sgpr_size_bytes()
+
+	s_mov_b32	s_save_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+	v_mov_b32	v0, 0x0							//Offset[31:0] from buffer resource
+	v_mov_b32	v1, 0x0							//Offset[63:32] from buffer resource
+	v_mov_b32	v2, 0x0							//Set of SGPRs for TCP store
+
+	// Ensure no further changes to barrier or LDS state.
+	// STATE_PRIV.BARRIER_COMPLETE may change up to this point.
+	s_barrier_signal	-2
+	s_barrier_wait	-2
+
+	// Re-read final state of BARRIER_COMPLETE field for save.
+	s_getreg_b32	s_save_tmp, hwreg(HW_REG_WAVE_STATE_PRIV)
+	s_and_b32	s_save_tmp, s_save_tmp, SQ_WAVE_STATE_PRIV_BARRIER_COMPLETE_MASK
+	s_andn2_b32	s_save_state_priv, s_save_state_priv, SQ_WAVE_STATE_PRIV_BARRIER_COMPLETE_MASK
+	s_or_b32	s_save_state_priv, s_save_state_priv, s_save_tmp
+
+	s_andn2_b32	s_save_tmp, s_save_pc_hi, S_SAVE_PC_HI_FIRST_WAVE_MASK
+	v_writelane_b32	v2, s_save_m0, 0x0
+	v_writelane_b32	v2, s_save_pc_lo, 0x1
+	v_writelane_b32	v2, s_save_tmp, 0x2
+	v_writelane_b32	v2, s_save_exec_lo, 0x3
+	v_writelane_b32	v2, s_save_exec_hi, 0x4
+	v_writelane_b32	v2, s_save_state_priv, 0x5
+	v_writelane_b32	v2, s_save_xnack_mask, 0x7
+	valu_sgpr_hazard()
+
+	s_getreg_b32	s_save_tmp, hwreg(HW_REG_WAVE_EXCP_FLAG_PRIV)
+	v_writelane_b32	v2, s_save_tmp, 0x6
+
+	s_getreg_b32	s_save_tmp, hwreg(HW_REG_WAVE_MODE)
+	v_writelane_b32	v2, s_save_tmp, 0x8
+
+	s_getreg_b32	s_save_tmp, hwreg(HW_REG_WAVE_SCRATCH_BASE_LO)
+	v_writelane_b32	v2, s_save_tmp, 0x9
+
+	s_getreg_b32	s_save_tmp, hwreg(HW_REG_WAVE_SCRATCH_BASE_HI)
+	v_writelane_b32	v2, s_save_tmp, 0xA
+
+	s_getreg_b32	s_save_tmp, hwreg(HW_REG_WAVE_EXCP_FLAG_USER)
+	v_writelane_b32	v2, s_save_tmp, 0xB
+
+	s_getreg_b32	s_save_tmp, hwreg(HW_REG_WAVE_TRAP_CTRL)
+	v_writelane_b32	v2, s_save_tmp, 0xC
+
+	s_getreg_b32	s_save_tmp, hwreg(HW_REG_WAVE_STATUS)
+	v_writelane_b32	v2, s_save_tmp, 0xD
+
+	s_get_barrier_state s_save_tmp, -1
+	s_wait_kmcnt (0)
+	v_writelane_b32	v2, s_save_tmp, 0xE
+	valu_sgpr_hazard()
+
+	// Write HWREGs with 16 VGPR lanes. TTMPs occupy space after this.
+	s_mov_b32       exec_lo, 0xFFFF
+	s_mov_b32	exec_hi, 0x0
+	buffer_store_dword	v2, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS
+
+	// Write SGPRs with 32 VGPR lanes. This works in wave32 and wave64 mode.
+	s_mov_b32       exec_lo, 0xFFFFFFFF
+
+	/* save SGPRs */
+	// Save SGPR before LDS save, then the s0 to s4 can be used during LDS save...
+
+	// SGPR SR memory offset : size(VGPR)+size(SVGPR)
+	get_vgpr_size_bytes(s_save_mem_offset, s_wave_size)
+	get_svgpr_size_bytes(s_save_tmp)
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, s_save_tmp
+	s_mov_b32	s_save_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+	s_mov_b32	ttmp13, 0x0						//next VGPR lane to copy SGPR into
+
+	s_mov_b32	m0, 0x0							//SGPR initial index value =0
+	s_nop		0x0							//Manually inserted wait states
+L_SAVE_SGPR_LOOP:
+	// SGPR is allocated in 16 SGPR granularity
+	s_movrels_b64	s0, s0							//s0 = s[0+m0], s1 = s[1+m0]
+	s_movrels_b64	s2, s2							//s2 = s[2+m0], s3 = s[3+m0]
+	s_movrels_b64	s4, s4							//s4 = s[4+m0], s5 = s[5+m0]
+	s_movrels_b64	s6, s6							//s6 = s[6+m0], s7 = s[7+m0]
+	s_movrels_b64	s8, s8							//s8 = s[8+m0], s9 = s[9+m0]
+	s_movrels_b64	s10, s10						//s10 = s[10+m0], s11 = s[11+m0]
+	s_movrels_b64	s12, s12						//s12 = s[12+m0], s13 = s[13+m0]
+	s_movrels_b64	s14, s14						//s14 = s[14+m0], s15 = s[15+m0]
+
+	s_cmp_eq_u32	ttmp13, 0x0
+	s_cbranch_scc0	L_WRITE_V2_SECOND_HALF
+	write_16sgpr_to_v2(s0, 0x0)
+	s_branch	L_SAVE_SGPR_SKIP_TCP_STORE
+L_WRITE_V2_SECOND_HALF:
+	write_16sgpr_to_v2(s0, 0x10)
+
+	buffer_store_dword	v2, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, 0x80
+	s_mov_b32	ttmp13, 0x0
+	v_mov_b32	v2, 0x0
+L_SAVE_SGPR_SKIP_TCP_STORE:
+
+	s_add_u32	m0, m0, 16						//next sgpr index
+	s_cmp_lt_u32	m0, 96							//scc = (m0 < first 96 SGPR) ? 1 : 0
+	s_cbranch_scc1	L_SAVE_SGPR_LOOP					//first 96 SGPR save is complete?
+
+	//save the rest 12 SGPR
+	s_movrels_b64	s0, s0							//s0 = s[0+m0], s1 = s[1+m0]
+	s_movrels_b64	s2, s2							//s2 = s[2+m0], s3 = s[3+m0]
+	s_movrels_b64	s4, s4							//s4 = s[4+m0], s5 = s[5+m0]
+	s_movrels_b64	s6, s6							//s6 = s[6+m0], s7 = s[7+m0]
+	s_movrels_b64	s8, s8							//s8 = s[8+m0], s9 = s[9+m0]
+	s_movrels_b64	s10, s10						//s10 = s[10+m0], s11 = s[11+m0]
+	write_12sgpr_to_v2(s0)
+
+	buffer_store_dword	v2, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS
+
+	/* save LDS */
+
+L_SAVE_LDS:
+	// Change EXEC to all threads...
+	s_mov_b32	exec_lo, 0xFFFFFFFF					//need every thread from now on
+	s_lshr_b32	m0, s_wave_size, S_WAVE_SIZE
+	s_and_b32	m0, m0, 1
+	s_cmp_eq_u32	m0, 1
+	s_cbranch_scc1	L_ENABLE_SAVE_LDS_EXEC_HI
+	s_mov_b32	exec_hi, 0x00000000
+	s_branch	L_SAVE_LDS_NORMAL
+L_ENABLE_SAVE_LDS_EXEC_HI:
+	s_mov_b32	exec_hi, 0xFFFFFFFF
+L_SAVE_LDS_NORMAL:
+	s_getreg_b32	s_save_alloc_size, hwreg(HW_REG_WAVE_LDS_ALLOC,SQ_WAVE_LDS_ALLOC_LDS_SIZE_SHIFT,SQ_WAVE_LDS_ALLOC_LDS_SIZE_SIZE)
+	s_and_b32	s_save_alloc_size, s_save_alloc_size, 0xFFFFFFFF	//lds_size is zero?
+	s_cbranch_scc0	L_SAVE_LDS_DONE						//no lds used? jump to L_SAVE_DONE
+
+	s_and_b32	s_save_tmp, s_save_pc_hi, S_SAVE_PC_HI_FIRST_WAVE_MASK
+	s_cbranch_scc0	L_SAVE_LDS_DONE
+
+	// first wave do LDS save;
+
+	s_lshl_b32	s_save_alloc_size, s_save_alloc_size, SQ_WAVE_LDS_ALLOC_GRANULARITY
+	s_mov_b32	s_save_buf_rsrc2, s_save_alloc_size			//NUM_RECORDS in bytes
+
+	// LDS at offset: size(VGPR)+size(SVGPR)+SIZE(SGPR)+SIZE(HWREG)
+	//
+	get_vgpr_size_bytes(s_save_mem_offset, s_wave_size)
+	get_svgpr_size_bytes(s_save_tmp)
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, s_save_tmp
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, get_sgpr_size_bytes()
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, get_hwreg_size_bytes()
+
+	s_mov_b32	s_save_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+	//load 0~63*4(byte address) to vgpr v0
+	v_mbcnt_lo_u32_b32	v0, -1, 0
+	v_mbcnt_hi_u32_b32	v0, -1, v0
+	v_mul_u32_u24	v0, 4, v0
+
+	s_lshr_b32	m0, s_wave_size, S_WAVE_SIZE
+	s_and_b32	m0, m0, 1
+	s_cmp_eq_u32	m0, 1
+	s_mov_b32	m0, 0x0
+	s_cbranch_scc1	L_SAVE_LDS_W64
+
+L_SAVE_LDS_W32:
+	s_mov_b32	s3, 128
+	s_nop		0
+	s_nop		0
+	s_nop		0
+L_SAVE_LDS_LOOP_W32:
+	ds_read_b32	v1, v0
+	s_wait_idle
+	buffer_store_dword	v1, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS
+
+	s_add_u32	m0, m0, s3						//every buffer_store_lds does 128 bytes
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, s3
+	v_add_nc_u32	v0, v0, 128						//mem offset increased by 128 bytes
+	s_cmp_lt_u32	m0, s_save_alloc_size					//scc=(m0 < s_save_alloc_size) ? 1 : 0
+	s_cbranch_scc1	L_SAVE_LDS_LOOP_W32					//LDS save is complete?
+
+	s_branch	L_SAVE_LDS_DONE
+
+L_SAVE_LDS_W64:
+	s_mov_b32	s3, 256
+	s_nop		0
+	s_nop		0
+	s_nop		0
+L_SAVE_LDS_LOOP_W64:
+	ds_read_b32	v1, v0
+	s_wait_idle
+	buffer_store_dword	v1, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS
+
+	s_add_u32	m0, m0, s3						//every buffer_store_lds does 256 bytes
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, s3
+	v_add_nc_u32	v0, v0, 256						//mem offset increased by 256 bytes
+	s_cmp_lt_u32	m0, s_save_alloc_size					//scc=(m0 < s_save_alloc_size) ? 1 : 0
+	s_cbranch_scc1	L_SAVE_LDS_LOOP_W64					//LDS save is complete?
+
+L_SAVE_LDS_DONE:
+	/* save VGPRs  - set the Rest VGPRs */
+L_SAVE_VGPR:
+	// VGPR SR memory offset: 0
+	s_mov_b32	exec_lo, 0xFFFFFFFF					//need every thread from now on
+	s_lshr_b32	m0, s_wave_size, S_WAVE_SIZE
+	s_and_b32	m0, m0, 1
+	s_cmp_eq_u32	m0, 1
+	s_cbranch_scc1	L_ENABLE_SAVE_VGPR_EXEC_HI
+	s_mov_b32	s_save_mem_offset, (0+128*4)				// for the rest VGPRs
+	s_mov_b32	exec_hi, 0x00000000
+	s_branch	L_SAVE_VGPR_NORMAL
+L_ENABLE_SAVE_VGPR_EXEC_HI:
+	s_mov_b32	s_save_mem_offset, (0+256*4)				// for the rest VGPRs
+	s_mov_b32	exec_hi, 0xFFFFFFFF
+L_SAVE_VGPR_NORMAL:
+	s_getreg_b32	s_save_alloc_size, hwreg(HW_REG_WAVE_GPR_ALLOC,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SIZE)
+	s_add_u32	s_save_alloc_size, s_save_alloc_size, 1
+	s_lshl_b32	s_save_alloc_size, s_save_alloc_size, 2			//Number of VGPRs = (vgpr_size + 1) * 4    (non-zero value)
+	//determine it is wave32 or wave64
+	s_lshr_b32	m0, s_wave_size, S_WAVE_SIZE
+	s_and_b32	m0, m0, 1
+	s_cmp_eq_u32	m0, 1
+	s_cbranch_scc1	L_SAVE_VGPR_WAVE64
+
+	s_mov_b32	s_save_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+	// VGPR Allocated in 4-GPR granularity
+
+	// VGPR store using dw burst
+	s_mov_b32	m0, 0x4							//VGPR initial index value =4
+	s_cmp_lt_u32	m0, s_save_alloc_size
+	s_cbranch_scc0	L_SAVE_VGPR_END
+
+L_SAVE_VGPR_W32_LOOP:
+	v_movrels_b32	v0, v0							//v0 = v[0+m0]
+	v_movrels_b32	v1, v1							//v1 = v[1+m0]
+	v_movrels_b32	v2, v2							//v2 = v[2+m0]
+	v_movrels_b32	v3, v3							//v3 = v[3+m0]
+
+	buffer_store_dword	v0, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS
+	buffer_store_dword	v1, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:128
+	buffer_store_dword	v2, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:128*2
+	buffer_store_dword	v3, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:128*3
+
+	s_add_u32	m0, m0, 4						//next vgpr index
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, 128*4		//every buffer_store_dword does 128 bytes
+	s_cmp_lt_u32	m0, s_save_alloc_size					//scc = (m0 < s_save_alloc_size) ? 1 : 0
+	s_cbranch_scc1	L_SAVE_VGPR_W32_LOOP					//VGPR save is complete?
+
+	s_branch	L_SAVE_VGPR_END
+
+L_SAVE_VGPR_WAVE64:
+	s_mov_b32	s_save_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+	// VGPR store using dw burst
+	s_mov_b32	m0, 0x4							//VGPR initial index value =4
+	s_cmp_lt_u32	m0, s_save_alloc_size
+	s_cbranch_scc0	L_SAVE_SHARED_VGPR
+
+L_SAVE_VGPR_W64_LOOP:
+	v_movrels_b32	v0, v0							//v0 = v[0+m0]
+	v_movrels_b32	v1, v1							//v1 = v[1+m0]
+	v_movrels_b32	v2, v2							//v2 = v[2+m0]
+	v_movrels_b32	v3, v3							//v3 = v[3+m0]
+
+	buffer_store_dword	v0, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS
+	buffer_store_dword	v1, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:256
+	buffer_store_dword	v2, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:256*2
+	buffer_store_dword	v3, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:256*3
+
+	s_add_u32	m0, m0, 4						//next vgpr index
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, 256*4		//every buffer_store_dword does 256 bytes
+	s_cmp_lt_u32	m0, s_save_alloc_size					//scc = (m0 < s_save_alloc_size) ? 1 : 0
+	s_cbranch_scc1	L_SAVE_VGPR_W64_LOOP					//VGPR save is complete?
+
+L_SAVE_SHARED_VGPR:
+	s_getreg_b32	s_save_alloc_size, hwreg(HW_REG_WAVE_LDS_ALLOC,SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SHIFT,SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SIZE)
+	s_and_b32	s_save_alloc_size, s_save_alloc_size, 0xFFFFFFFF	//shared_vgpr_size is zero?
+	s_cbranch_scc0	L_SAVE_VGPR_END						//no shared_vgpr used? jump to L_SAVE_LDS
+	s_lshl_b32	s_save_alloc_size, s_save_alloc_size, 3			//Number of SHARED_VGPRs = shared_vgpr_size * 8    (non-zero value)
+	//m0 now has the value of normal vgpr count, just add the m0 with shared_vgpr count to get the total count.
+	//save shared_vgpr will start from the index of m0
+	s_add_u32	s_save_alloc_size, s_save_alloc_size, m0
+	s_mov_b32	exec_lo, 0xFFFFFFFF
+	s_mov_b32	exec_hi, 0x00000000
+
+L_SAVE_SHARED_VGPR_WAVE64_LOOP:
+	v_movrels_b32	v0, v0							//v0 = v[0+m0]
+	buffer_store_dword	v0, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS
+	s_add_u32	m0, m0, 1						//next vgpr index
+	s_add_u32	s_save_mem_offset, s_save_mem_offset, 128
+	s_cmp_lt_u32	m0, s_save_alloc_size					//scc = (m0 < s_save_alloc_size) ? 1 : 0
+	s_cbranch_scc1	L_SAVE_SHARED_VGPR_WAVE64_LOOP				//SHARED_VGPR save is complete?
+
+L_SAVE_VGPR_END:
+	s_branch	L_END_PGM
+
+L_RESTORE:
+	/* Setup Resource Contants */
+	s_mov_b32	s_restore_buf_rsrc0, s_restore_spi_init_lo		//base_addr_lo
+	s_and_b32	s_restore_buf_rsrc1, s_restore_spi_init_hi, 0x0000FFFF	//base_addr_hi
+	s_or_b32	s_restore_buf_rsrc1, s_restore_buf_rsrc1, S_RESTORE_BUF_RSRC_WORD1_STRIDE
+	s_mov_b32	s_restore_buf_rsrc2, 0					//NUM_RECORDS initial value = 0 (in bytes)
+	s_mov_b32	s_restore_buf_rsrc3, S_RESTORE_BUF_RSRC_WORD3_MISC
+
+	// Save s_restore_spi_init_hi for later use.
+	s_mov_b32 s_restore_spi_init_hi_save, s_restore_spi_init_hi
+
+	//determine it is wave32 or wave64
+	get_wave_size2(s_restore_size)
+
+	s_and_b32	s_restore_tmp, s_restore_spi_init_hi, S_RESTORE_SPI_INIT_FIRST_WAVE_MASK
+	s_cbranch_scc0	L_RESTORE_VGPR
+
+	/* restore LDS */
+L_RESTORE_LDS:
+	s_mov_b32	exec_lo, 0xFFFFFFFF					//need every thread from now on
+	s_lshr_b32	m0, s_restore_size, S_WAVE_SIZE
+	s_and_b32	m0, m0, 1
+	s_cmp_eq_u32	m0, 1
+	s_cbranch_scc1	L_ENABLE_RESTORE_LDS_EXEC_HI
+	s_mov_b32	exec_hi, 0x00000000
+	s_branch	L_RESTORE_LDS_NORMAL
+L_ENABLE_RESTORE_LDS_EXEC_HI:
+	s_mov_b32	exec_hi, 0xFFFFFFFF
+L_RESTORE_LDS_NORMAL:
+	s_getreg_b32	s_restore_alloc_size, hwreg(HW_REG_WAVE_LDS_ALLOC,SQ_WAVE_LDS_ALLOC_LDS_SIZE_SHIFT,SQ_WAVE_LDS_ALLOC_LDS_SIZE_SIZE)
+	s_and_b32	s_restore_alloc_size, s_restore_alloc_size, 0xFFFFFFFF	//lds_size is zero?
+	s_cbranch_scc0	L_RESTORE_VGPR						//no lds used? jump to L_RESTORE_VGPR
+	s_lshl_b32	s_restore_alloc_size, s_restore_alloc_size, SQ_WAVE_LDS_ALLOC_GRANULARITY
+	s_mov_b32	s_restore_buf_rsrc2, s_restore_alloc_size		//NUM_RECORDS in bytes
+
+	// LDS at offset: size(VGPR)+size(SVGPR)+SIZE(SGPR)+SIZE(HWREG)
+	//
+	get_vgpr_size_bytes(s_restore_mem_offset, s_restore_size)
+	get_svgpr_size_bytes(s_restore_tmp)
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, s_restore_tmp
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, get_sgpr_size_bytes()
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, get_hwreg_size_bytes()
+
+	s_mov_b32	s_restore_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+	s_lshr_b32	m0, s_restore_size, S_WAVE_SIZE
+	s_and_b32	m0, m0, 1
+	s_cmp_eq_u32	m0, 1
+	s_mov_b32	m0, 0x0
+	s_cbranch_scc1	L_RESTORE_LDS_LOOP_W64
+
+L_RESTORE_LDS_LOOP_W32:
+	buffer_load_dword       v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset
+	s_wait_idle
+	ds_store_addtid_b32     v0
+	s_add_u32	m0, m0, 128						// 128 DW
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, 128		//mem offset increased by 128DW
+	s_cmp_lt_u32	m0, s_restore_alloc_size				//scc=(m0 < s_restore_alloc_size) ? 1 : 0
+	s_cbranch_scc1	L_RESTORE_LDS_LOOP_W32					//LDS restore is complete?
+	s_branch	L_RESTORE_VGPR
+
+L_RESTORE_LDS_LOOP_W64:
+	buffer_load_dword       v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset
+	s_wait_idle
+	ds_store_addtid_b32     v0
+	s_add_u32	m0, m0, 256						// 256 DW
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, 256		//mem offset increased by 256DW
+	s_cmp_lt_u32	m0, s_restore_alloc_size				//scc=(m0 < s_restore_alloc_size) ? 1 : 0
+	s_cbranch_scc1	L_RESTORE_LDS_LOOP_W64					//LDS restore is complete?
+
+	/* restore VGPRs */
+L_RESTORE_VGPR:
+	// VGPR SR memory offset : 0
+	s_mov_b32	s_restore_mem_offset, 0x0
+	s_mov_b32	exec_lo, 0xFFFFFFFF					//need every thread from now on
+	s_lshr_b32	m0, s_restore_size, S_WAVE_SIZE
+	s_and_b32	m0, m0, 1
+	s_cmp_eq_u32	m0, 1
+	s_cbranch_scc1	L_ENABLE_RESTORE_VGPR_EXEC_HI
+	s_mov_b32	exec_hi, 0x00000000
+	s_branch	L_RESTORE_VGPR_NORMAL
+L_ENABLE_RESTORE_VGPR_EXEC_HI:
+	s_mov_b32	exec_hi, 0xFFFFFFFF
+L_RESTORE_VGPR_NORMAL:
+	s_getreg_b32	s_restore_alloc_size, hwreg(HW_REG_WAVE_GPR_ALLOC,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SIZE)
+	s_add_u32	s_restore_alloc_size, s_restore_alloc_size, 1
+	s_lshl_b32	s_restore_alloc_size, s_restore_alloc_size, 2		//Number of VGPRs = (vgpr_size + 1) * 4    (non-zero value)
+	//determine it is wave32 or wave64
+	s_lshr_b32	m0, s_restore_size, S_WAVE_SIZE
+	s_and_b32	m0, m0, 1
+	s_cmp_eq_u32	m0, 1
+	s_cbranch_scc1	L_RESTORE_VGPR_WAVE64
+
+	s_mov_b32	s_restore_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+	// VGPR load using dw burst
+	s_mov_b32	s_restore_mem_offset_save, s_restore_mem_offset		// restore start with v1, v0 will be the last
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, 128*4
+	s_mov_b32	m0, 4							//VGPR initial index value = 4
+	s_cmp_lt_u32	m0, s_restore_alloc_size
+	s_cbranch_scc0	L_RESTORE_SGPR
+
+L_RESTORE_VGPR_WAVE32_LOOP:
+	buffer_load_dword	v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset scope:SCOPE_SYS
+	buffer_load_dword	v1, v0, s_restore_buf_rsrc0, s_restore_mem_offset scope:SCOPE_SYS offset:128
+	buffer_load_dword	v2, v0, s_restore_buf_rsrc0, s_restore_mem_offset scope:SCOPE_SYS offset:128*2
+	buffer_load_dword	v3, v0, s_restore_buf_rsrc0, s_restore_mem_offset scope:SCOPE_SYS offset:128*3
+	s_wait_idle
+	v_movreld_b32	v0, v0							//v[0+m0] = v0
+	v_movreld_b32	v1, v1
+	v_movreld_b32	v2, v2
+	v_movreld_b32	v3, v3
+	s_add_u32	m0, m0, 4						//next vgpr index
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, 128*4	//every buffer_load_dword does 128 bytes
+	s_cmp_lt_u32	m0, s_restore_alloc_size				//scc = (m0 < s_restore_alloc_size) ? 1 : 0
+	s_cbranch_scc1	L_RESTORE_VGPR_WAVE32_LOOP				//VGPR restore (except v0) is complete?
+
+	/* VGPR restore on v0 */
+	buffer_load_dword	v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save scope:SCOPE_SYS
+	buffer_load_dword	v1, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save scope:SCOPE_SYS offset:128
+	buffer_load_dword	v2, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save scope:SCOPE_SYS offset:128*2
+	buffer_load_dword	v3, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save scope:SCOPE_SYS offset:128*3
+	s_wait_idle
+
+	s_branch	L_RESTORE_SGPR
+
+L_RESTORE_VGPR_WAVE64:
+	s_mov_b32	s_restore_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+	// VGPR load using dw burst
+	s_mov_b32	s_restore_mem_offset_save, s_restore_mem_offset		// restore start with v4, v0 will be the last
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, 256*4
+	s_mov_b32	m0, 4							//VGPR initial index value = 4
+	s_cmp_lt_u32	m0, s_restore_alloc_size
+	s_cbranch_scc0	L_RESTORE_SHARED_VGPR
+
+L_RESTORE_VGPR_WAVE64_LOOP:
+	buffer_load_dword	v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset scope:SCOPE_SYS
+	buffer_load_dword	v1, v0, s_restore_buf_rsrc0, s_restore_mem_offset scope:SCOPE_SYS offset:256
+	buffer_load_dword	v2, v0, s_restore_buf_rsrc0, s_restore_mem_offset scope:SCOPE_SYS offset:256*2
+	buffer_load_dword	v3, v0, s_restore_buf_rsrc0, s_restore_mem_offset scope:SCOPE_SYS offset:256*3
+	s_wait_idle
+	v_movreld_b32	v0, v0							//v[0+m0] = v0
+	v_movreld_b32	v1, v1
+	v_movreld_b32	v2, v2
+	v_movreld_b32	v3, v3
+	s_add_u32	m0, m0, 4						//next vgpr index
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, 256*4	//every buffer_load_dword does 256 bytes
+	s_cmp_lt_u32	m0, s_restore_alloc_size				//scc = (m0 < s_restore_alloc_size) ? 1 : 0
+	s_cbranch_scc1	L_RESTORE_VGPR_WAVE64_LOOP				//VGPR restore (except v0) is complete?
+
+L_RESTORE_SHARED_VGPR:
+	s_getreg_b32	s_restore_alloc_size, hwreg(HW_REG_WAVE_LDS_ALLOC,SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SHIFT,SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SIZE)	//shared_vgpr_size
+	s_and_b32	s_restore_alloc_size, s_restore_alloc_size, 0xFFFFFFFF	//shared_vgpr_size is zero?
+	s_cbranch_scc0	L_RESTORE_V0						//no shared_vgpr used?
+	s_lshl_b32	s_restore_alloc_size, s_restore_alloc_size, 3		//Number of SHARED_VGPRs = shared_vgpr_size * 8    (non-zero value)
+	//m0 now has the value of normal vgpr count, just add the m0 with shared_vgpr count to get the total count.
+	//restore shared_vgpr will start from the index of m0
+	s_add_u32	s_restore_alloc_size, s_restore_alloc_size, m0
+	s_mov_b32	exec_lo, 0xFFFFFFFF
+	s_mov_b32	exec_hi, 0x00000000
+L_RESTORE_SHARED_VGPR_WAVE64_LOOP:
+	buffer_load_dword	v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset scope:SCOPE_SYS
+	s_wait_idle
+	v_movreld_b32	v0, v0							//v[0+m0] = v0
+	s_add_u32	m0, m0, 1						//next vgpr index
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, 128
+	s_cmp_lt_u32	m0, s_restore_alloc_size				//scc = (m0 < s_restore_alloc_size) ? 1 : 0
+	s_cbranch_scc1	L_RESTORE_SHARED_VGPR_WAVE64_LOOP			//VGPR restore (except v0) is complete?
+
+	s_mov_b32	exec_hi, 0xFFFFFFFF					//restore back exec_hi before restoring V0!!
+
+	/* VGPR restore on v0 */
+L_RESTORE_V0:
+	buffer_load_dword	v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save scope:SCOPE_SYS
+	buffer_load_dword	v1, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save scope:SCOPE_SYS offset:256
+	buffer_load_dword	v2, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save scope:SCOPE_SYS offset:256*2
+	buffer_load_dword	v3, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save scope:SCOPE_SYS offset:256*3
+	s_wait_idle
+
+	/* restore SGPRs */
+	//will be 2+8+16*6
+	// SGPR SR memory offset : size(VGPR)+size(SVGPR)
+L_RESTORE_SGPR:
+	get_vgpr_size_bytes(s_restore_mem_offset, s_restore_size)
+	get_svgpr_size_bytes(s_restore_tmp)
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, s_restore_tmp
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, get_sgpr_size_bytes()
+	s_sub_u32	s_restore_mem_offset, s_restore_mem_offset, 20*4	//s108~s127 is not saved
+
+	s_mov_b32	s_restore_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+	s_mov_b32	m0, s_sgpr_save_num
+
+	read_4sgpr_from_mem(s0, s_restore_buf_rsrc0, s_restore_mem_offset)
+	s_wait_idle
+
+	s_sub_u32	m0, m0, 4						// Restore from S[0] to S[104]
+	s_nop		0							// hazard SALU M0=> S_MOVREL
+
+	s_movreld_b64	s0, s0							//s[0+m0] = s0
+	s_movreld_b64	s2, s2
+
+	read_8sgpr_from_mem(s0, s_restore_buf_rsrc0, s_restore_mem_offset)
+	s_wait_idle
+
+	s_sub_u32	m0, m0, 8						// Restore from S[0] to S[96]
+	s_nop		0							// hazard SALU M0=> S_MOVREL
+
+	s_movreld_b64	s0, s0							//s[0+m0] = s0
+	s_movreld_b64	s2, s2
+	s_movreld_b64	s4, s4
+	s_movreld_b64	s6, s6
+
+ L_RESTORE_SGPR_LOOP:
+	read_16sgpr_from_mem(s0, s_restore_buf_rsrc0, s_restore_mem_offset)
+	s_wait_idle
+
+	s_sub_u32	m0, m0, 16						// Restore from S[n] to S[0]
+	s_nop		0							// hazard SALU M0=> S_MOVREL
+
+	s_movreld_b64	s0, s0							//s[0+m0] = s0
+	s_movreld_b64	s2, s2
+	s_movreld_b64	s4, s4
+	s_movreld_b64	s6, s6
+	s_movreld_b64	s8, s8
+	s_movreld_b64	s10, s10
+	s_movreld_b64	s12, s12
+	s_movreld_b64	s14, s14
+
+	s_cmp_eq_u32	m0, 0							//scc = (m0 < s_sgpr_save_num) ? 1 : 0
+	s_cbranch_scc0	L_RESTORE_SGPR_LOOP
+
+	// s_barrier with STATE_PRIV.TRAP_AFTER_INST=1, STATUS.PRIV=1 incorrectly asserts debug exception.
+	// Clear DEBUG_EN before and restore MODE after the barrier.
+	s_setreg_imm32_b32	hwreg(HW_REG_WAVE_MODE), 0
+
+	/* restore HW registers */
+L_RESTORE_HWREG:
+	// HWREG SR memory offset : size(VGPR)+size(SVGPR)+size(SGPR)
+	get_vgpr_size_bytes(s_restore_mem_offset, s_restore_size)
+	get_svgpr_size_bytes(s_restore_tmp)
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, s_restore_tmp
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, get_sgpr_size_bytes()
+
+	s_mov_b32	s_restore_buf_rsrc2, 0x1000000				//NUM_RECORDS in bytes
+
+	// Restore s_restore_spi_init_hi before the saved value gets clobbered.
+	s_mov_b32 s_restore_spi_init_hi, s_restore_spi_init_hi_save
+
+	read_hwreg_from_mem(s_restore_m0, s_restore_buf_rsrc0, s_restore_mem_offset)
+	read_hwreg_from_mem(s_restore_pc_lo, s_restore_buf_rsrc0, s_restore_mem_offset)
+	read_hwreg_from_mem(s_restore_pc_hi, s_restore_buf_rsrc0, s_restore_mem_offset)
+	read_hwreg_from_mem(s_restore_exec_lo, s_restore_buf_rsrc0, s_restore_mem_offset)
+	read_hwreg_from_mem(s_restore_exec_hi, s_restore_buf_rsrc0, s_restore_mem_offset)
+	read_hwreg_from_mem(s_restore_state_priv, s_restore_buf_rsrc0, s_restore_mem_offset)
+	read_hwreg_from_mem(s_restore_excp_flag_priv, s_restore_buf_rsrc0, s_restore_mem_offset)
+	read_hwreg_from_mem(s_restore_xnack_mask, s_restore_buf_rsrc0, s_restore_mem_offset)
+	read_hwreg_from_mem(s_restore_mode, s_restore_buf_rsrc0, s_restore_mem_offset)
+	read_hwreg_from_mem(s_restore_flat_scratch, s_restore_buf_rsrc0, s_restore_mem_offset)
+	s_wait_idle
+
+	s_setreg_b32	hwreg(HW_REG_WAVE_SCRATCH_BASE_LO), s_restore_flat_scratch
+
+	read_hwreg_from_mem(s_restore_flat_scratch, s_restore_buf_rsrc0, s_restore_mem_offset)
+	s_wait_idle
+
+	s_setreg_b32	hwreg(HW_REG_WAVE_SCRATCH_BASE_HI), s_restore_flat_scratch
+
+	read_hwreg_from_mem(s_restore_tmp, s_restore_buf_rsrc0, s_restore_mem_offset)
+	s_wait_idle
+	s_setreg_b32	hwreg(HW_REG_WAVE_EXCP_FLAG_USER), s_restore_tmp
+
+	read_hwreg_from_mem(s_restore_tmp, s_restore_buf_rsrc0, s_restore_mem_offset)
+	s_wait_idle
+	s_setreg_b32	hwreg(HW_REG_WAVE_TRAP_CTRL), s_restore_tmp
+
+	// Only the first wave needs to restore the workgroup barrier.
+	s_and_b32	s_restore_tmp, s_restore_spi_init_hi, S_RESTORE_SPI_INIT_FIRST_WAVE_MASK
+	s_cbranch_scc0	L_SKIP_BARRIER_RESTORE
+
+	// Skip over WAVE_STATUS, since there is no state to restore from it
+	s_add_u32	s_restore_mem_offset, s_restore_mem_offset, 4
+
+	read_hwreg_from_mem(s_restore_tmp, s_restore_buf_rsrc0, s_restore_mem_offset)
+	s_wait_idle
+
+	s_bitcmp1_b32	s_restore_tmp, BARRIER_STATE_VALID_OFFSET
+	s_cbranch_scc0	L_SKIP_BARRIER_RESTORE
+
+	// extract the saved signal count from s_restore_tmp
+	s_lshr_b32	s_restore_tmp, s_restore_tmp, BARRIER_STATE_SIGNAL_OFFSET
+
+	// We need to call s_barrier_signal repeatedly to restore the signal
+	// count of the work group barrier.  The member count is already
+	// initialized with the number of waves in the work group.
+L_BARRIER_RESTORE_LOOP:
+	s_and_b32	s_restore_tmp, s_restore_tmp, s_restore_tmp
+	s_cbranch_scc0	L_SKIP_BARRIER_RESTORE
+	s_barrier_signal	-1
+	s_add_i32	s_restore_tmp, s_restore_tmp, -1
+	s_branch	L_BARRIER_RESTORE_LOOP
+
+L_SKIP_BARRIER_RESTORE:
+
+	s_mov_b32	m0, s_restore_m0
+	s_mov_b32	exec_lo, s_restore_exec_lo
+	s_mov_b32	exec_hi, s_restore_exec_hi
+
+	// EXCP_FLAG_PRIV.SAVE_CONTEXT and HOST_TRAP may have changed.
+	// Only restore the other fields to avoid clobbering them.
+	s_setreg_b32	hwreg(HW_REG_WAVE_EXCP_FLAG_PRIV, 0, SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_1_SIZE), s_restore_excp_flag_priv
+	s_lshr_b32	s_restore_excp_flag_priv, s_restore_excp_flag_priv, SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_2_SHIFT
+	s_setreg_b32	hwreg(HW_REG_WAVE_EXCP_FLAG_PRIV, SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_2_SHIFT, SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_2_SIZE), s_restore_excp_flag_priv
+	s_lshr_b32	s_restore_excp_flag_priv, s_restore_excp_flag_priv, SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_3_SHIFT - SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_2_SHIFT
+	s_setreg_b32	hwreg(HW_REG_WAVE_EXCP_FLAG_PRIV, SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_3_SHIFT, SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_3_SIZE), s_restore_excp_flag_priv
+
+	s_setreg_b32	hwreg(HW_REG_WAVE_MODE), s_restore_mode
+
+	// Restore trap temporaries 4-11, 13 initialized by SPI debug dispatch logic
+	// ttmp SR memory offset : size(VGPR)+size(SVGPR)+size(SGPR)+0x40
+	get_vgpr_size_bytes(s_restore_ttmps_lo, s_restore_size)
+	get_svgpr_size_bytes(s_restore_ttmps_hi)
+	s_add_u32	s_restore_ttmps_lo, s_restore_ttmps_lo, s_restore_ttmps_hi
+	s_add_u32	s_restore_ttmps_lo, s_restore_ttmps_lo, get_sgpr_size_bytes()
+	s_add_u32	s_restore_ttmps_lo, s_restore_ttmps_lo, s_restore_buf_rsrc0
+	s_addc_u32	s_restore_ttmps_hi, s_restore_buf_rsrc1, 0x0
+	s_and_b32	s_restore_ttmps_hi, s_restore_ttmps_hi, 0xFFFF
+	s_load_dwordx4	[ttmp4, ttmp5, ttmp6, ttmp7], [s_restore_ttmps_lo, s_restore_ttmps_hi], 0x50 scope:SCOPE_SYS
+	s_load_dwordx4	[ttmp8, ttmp9, ttmp10, ttmp11], [s_restore_ttmps_lo, s_restore_ttmps_hi], 0x60 scope:SCOPE_SYS
+	s_load_dword	ttmp13, [s_restore_ttmps_lo, s_restore_ttmps_hi], 0x74 scope:SCOPE_SYS
+	s_wait_idle
+
+	s_and_b32	s_restore_pc_hi, s_restore_pc_hi, 0x0000ffff		//pc[47:32] //Do it here in order not to affect STATUS
+	s_and_b64	exec, exec, exec					// Restore STATUS.EXECZ, not writable by s_setreg_b32
+	s_and_b64	vcc, vcc, vcc						// Restore STATUS.VCCZ, not writable by s_setreg_b32
+
+	s_setreg_b32	hwreg(HW_REG_WAVE_STATE_PRIV), s_restore_state_priv	// SCC is included, which is changed by previous salu
+
+	// Make barrier and LDS state visible to all waves in the group.
+	// STATE_PRIV.BARRIER_COMPLETE may change after this point.
+	s_barrier_signal	-2
+	s_barrier_wait	-2
+
+	s_rfe_b64	s_restore_pc_lo						//Return to the main shader program and resume execution
+
+L_END_PGM:
+	// Make sure that no wave of the workgroup can exit the trap handler
+	// before the workgroup barrier state is saved.
+	s_barrier_signal	-2
+	s_barrier_wait	-2
+	s_endpgm_saved
+end
+
+function write_16sgpr_to_v2(s, lane_offset)
+	// Copy into VGPR for later TCP store.
+	for var sgpr_idx = 0; sgpr_idx < 16; sgpr_idx ++
+		v_writelane_b32	v2, s[sgpr_idx], sgpr_idx + lane_offset
+	end
+	valu_sgpr_hazard()
+	s_add_u32	ttmp13, ttmp13, 0x10
+end
+
+function write_12sgpr_to_v2(s)
+	// Copy into VGPR for later TCP store.
+	for var sgpr_idx = 0; sgpr_idx < 12; sgpr_idx ++
+		v_writelane_b32	v2, s[sgpr_idx], sgpr_idx
+	end
+	valu_sgpr_hazard()
+end
+
+function read_hwreg_from_mem(s, s_rsrc, s_mem_offset)
+	s_buffer_load_dword	s, s_rsrc, s_mem_offset scope:SCOPE_SYS
+	s_add_u32	s_mem_offset, s_mem_offset, 4
+end
+
+function read_16sgpr_from_mem(s, s_rsrc, s_mem_offset)
+	s_sub_u32	s_mem_offset, s_mem_offset, 4*16
+	s_buffer_load_dwordx16	s, s_rsrc, s_mem_offset scope:SCOPE_SYS
+end
+
+function read_8sgpr_from_mem(s, s_rsrc, s_mem_offset)
+	s_sub_u32	s_mem_offset, s_mem_offset, 4*8
+	s_buffer_load_dwordx8	s, s_rsrc, s_mem_offset scope:SCOPE_SYS
+end
+
+function read_4sgpr_from_mem(s, s_rsrc, s_mem_offset)
+	s_sub_u32	s_mem_offset, s_mem_offset, 4*4
+	s_buffer_load_dwordx4	s, s_rsrc, s_mem_offset scope:SCOPE_SYS
+end
+
+function get_vgpr_size_bytes(s_vgpr_size_byte, s_size)
+	s_getreg_b32	s_vgpr_size_byte, hwreg(HW_REG_WAVE_GPR_ALLOC,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SIZE)
+	s_add_u32	s_vgpr_size_byte, s_vgpr_size_byte, 1
+	s_bitcmp1_b32	s_size, S_WAVE_SIZE
+	s_cbranch_scc1	L_ENABLE_SHIFT_W64
+	s_lshl_b32	s_vgpr_size_byte, s_vgpr_size_byte, (2+7)		//Number of VGPRs = (vgpr_size + 1) * 4 * 32 * 4   (non-zero value)
+	s_branch	L_SHIFT_DONE
+L_ENABLE_SHIFT_W64:
+	s_lshl_b32	s_vgpr_size_byte, s_vgpr_size_byte, (2+8)		//Number of VGPRs = (vgpr_size + 1) * 4 * 64 * 4   (non-zero value)
+L_SHIFT_DONE:
+end
+
+function get_svgpr_size_bytes(s_svgpr_size_byte)
+	s_getreg_b32	s_svgpr_size_byte, hwreg(HW_REG_WAVE_LDS_ALLOC,SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SHIFT,SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SIZE)
+	s_lshl_b32	s_svgpr_size_byte, s_svgpr_size_byte, (3+7)
+end
+
+function get_sgpr_size_bytes
+	return 512
+end
+
+function get_hwreg_size_bytes
+	return 128
+end
+
+function get_wave_size2(s_reg)
+	s_getreg_b32	s_reg, hwreg(HW_REG_WAVE_STATUS,SQ_WAVE_STATUS_WAVE64_SHIFT,SQ_WAVE_STATUS_WAVE64_SIZE)
+	s_lshl_b32	s_reg, s_reg, S_WAVE_SIZE
+end
+
+function valu_sgpr_hazard
+#if HAVE_VALU_SGPR_HAZARD
+	for var rep = 0; rep < 8; rep ++
+		ds_nop
+	end
+#endif
+end
diff --git a/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx8.asm b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx8.asm
new file mode 100644
index 000000000000..ac8edef09ca5
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx8.asm
@@ -0,0 +1,747 @@
+/*
+ * Copyright 2015-2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/* To compile this assembly code:
+ * PROJECT=vi ./sp3 cwsr_trap_handler_gfx8.asm -hex tmp.hex
+ */
+
+/**************************************************************************/
+/*                      variables                                         */
+/**************************************************************************/
+var SQ_WAVE_STATUS_INST_ATC_SHIFT  = 23
+var SQ_WAVE_STATUS_INST_ATC_MASK   = 0x00800000
+var SQ_WAVE_STATUS_SPI_PRIO_SHIFT  = 1
+var SQ_WAVE_STATUS_SPI_PRIO_MASK   = 0x00000006
+var SQ_WAVE_STATUS_PRE_SPI_PRIO_SHIFT   = 0
+var SQ_WAVE_STATUS_PRE_SPI_PRIO_SIZE    = 1
+var SQ_WAVE_STATUS_POST_SPI_PRIO_SHIFT  = 3
+var SQ_WAVE_STATUS_POST_SPI_PRIO_SIZE   = 29
+
+var SQ_WAVE_LDS_ALLOC_LDS_SIZE_SHIFT    = 12
+var SQ_WAVE_LDS_ALLOC_LDS_SIZE_SIZE     = 9
+var SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT   = 8
+var SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SIZE    = 6
+var SQ_WAVE_GPR_ALLOC_SGPR_SIZE_SHIFT   = 24
+var SQ_WAVE_GPR_ALLOC_SGPR_SIZE_SIZE    = 3                     //FIXME  sq.blk still has 4 bits at this time while SQ programming guide has 3 bits
+
+var SQ_WAVE_TRAPSTS_SAVECTX_MASK    =   0x400
+var SQ_WAVE_TRAPSTS_EXCE_MASK       =   0x1FF                   // Exception mask
+var SQ_WAVE_TRAPSTS_SAVECTX_SHIFT   =   10
+var SQ_WAVE_TRAPSTS_MEM_VIOL_MASK   =   0x100
+var SQ_WAVE_TRAPSTS_MEM_VIOL_SHIFT  =   8
+var SQ_WAVE_TRAPSTS_PRE_SAVECTX_MASK    =   0x3FF
+var SQ_WAVE_TRAPSTS_PRE_SAVECTX_SHIFT   =   0x0
+var SQ_WAVE_TRAPSTS_PRE_SAVECTX_SIZE    =   10
+var SQ_WAVE_TRAPSTS_POST_SAVECTX_MASK   =   0xFFFFF800
+var SQ_WAVE_TRAPSTS_POST_SAVECTX_SHIFT  =   11
+var SQ_WAVE_TRAPSTS_POST_SAVECTX_SIZE   =   21
+
+var SQ_WAVE_IB_STS_RCNT_SHIFT           =   16                  //FIXME
+var SQ_WAVE_IB_STS_RCNT_SIZE            =   4                   //FIXME
+var SQ_WAVE_IB_STS_FIRST_REPLAY_SHIFT   =   15                  //FIXME
+var SQ_WAVE_IB_STS_FIRST_REPLAY_SIZE    =   1                   //FIXME
+var SQ_WAVE_IB_STS_RCNT_FIRST_REPLAY_MASK_NEG   = 0x00007FFF    //FIXME
+
+var SQ_BUF_RSRC_WORD1_ATC_SHIFT     =   24
+var SQ_BUF_RSRC_WORD3_MTYPE_SHIFT   =   27
+
+
+/*      Save        */
+var S_SAVE_BUF_RSRC_WORD1_STRIDE        =   0x00040000          //stride is 4 bytes
+var S_SAVE_BUF_RSRC_WORD3_MISC          =   0x00807FAC          //SQ_SEL_X/Y/Z/W, BUF_NUM_FORMAT_FLOAT, (0 for MUBUF stride[17:14] when ADD_TID_ENABLE and BUF_DATA_FORMAT_32 for MTBUF), ADD_TID_ENABLE
+
+var S_SAVE_SPI_INIT_ATC_MASK            =   0x08000000          //bit[27]: ATC bit
+var S_SAVE_SPI_INIT_ATC_SHIFT           =   27
+var S_SAVE_SPI_INIT_MTYPE_MASK          =   0x70000000          //bit[30:28]: Mtype
+var S_SAVE_SPI_INIT_MTYPE_SHIFT         =   28
+var S_SAVE_SPI_INIT_FIRST_WAVE_MASK     =   0x04000000          //bit[26]: FirstWaveInTG
+var S_SAVE_SPI_INIT_FIRST_WAVE_SHIFT    =   26
+
+var S_SAVE_PC_HI_RCNT_SHIFT             =   28                  //FIXME  check with Brian to ensure all fields other than PC[47:0] can be used
+var S_SAVE_PC_HI_RCNT_MASK              =   0xF0000000          //FIXME
+var S_SAVE_PC_HI_FIRST_REPLAY_SHIFT     =   27                  //FIXME
+var S_SAVE_PC_HI_FIRST_REPLAY_MASK      =   0x08000000          //FIXME
+
+var s_save_spi_init_lo              =   exec_lo
+var s_save_spi_init_hi              =   exec_hi
+
+                                                //tba_lo and tba_hi need to be saved/restored
+var s_save_pc_lo            =   ttmp0           //{TTMP1, TTMP0} = {3'h0,pc_rewind[3:0], HT[0],trapID[7:0], PC[47:0]}
+var s_save_pc_hi            =   ttmp1
+var s_save_exec_lo          =   ttmp2
+var s_save_exec_hi          =   ttmp3
+var s_save_status           =   ttmp4
+var s_save_trapsts          =   ttmp5           //not really used until the end of the SAVE routine
+var s_save_xnack_mask_lo    =   ttmp6
+var s_save_xnack_mask_hi    =   ttmp7
+var s_save_buf_rsrc0        =   ttmp8
+var s_save_buf_rsrc1        =   ttmp9
+var s_save_buf_rsrc2        =   ttmp10
+var s_save_buf_rsrc3        =   ttmp11
+
+var s_save_mem_offset       =   tma_lo
+var s_save_alloc_size       =   s_save_trapsts          //conflict
+var s_save_tmp              =   s_save_buf_rsrc2        //shared with s_save_buf_rsrc2  (conflict: should not use mem access with s_save_tmp at the same time)
+var s_save_m0               =   tma_hi
+
+/*      Restore     */
+var S_RESTORE_BUF_RSRC_WORD1_STRIDE         =   S_SAVE_BUF_RSRC_WORD1_STRIDE
+var S_RESTORE_BUF_RSRC_WORD3_MISC           =   S_SAVE_BUF_RSRC_WORD3_MISC
+
+var S_RESTORE_SPI_INIT_ATC_MASK             =   0x08000000          //bit[27]: ATC bit
+var S_RESTORE_SPI_INIT_ATC_SHIFT            =   27
+var S_RESTORE_SPI_INIT_MTYPE_MASK           =   0x70000000          //bit[30:28]: Mtype
+var S_RESTORE_SPI_INIT_MTYPE_SHIFT          =   28
+var S_RESTORE_SPI_INIT_FIRST_WAVE_MASK      =   0x04000000          //bit[26]: FirstWaveInTG
+var S_RESTORE_SPI_INIT_FIRST_WAVE_SHIFT     =   26
+
+var S_RESTORE_PC_HI_RCNT_SHIFT              =   S_SAVE_PC_HI_RCNT_SHIFT
+var S_RESTORE_PC_HI_RCNT_MASK               =   S_SAVE_PC_HI_RCNT_MASK
+var S_RESTORE_PC_HI_FIRST_REPLAY_SHIFT      =   S_SAVE_PC_HI_FIRST_REPLAY_SHIFT
+var S_RESTORE_PC_HI_FIRST_REPLAY_MASK       =   S_SAVE_PC_HI_FIRST_REPLAY_MASK
+
+var s_restore_spi_init_lo                   =   exec_lo
+var s_restore_spi_init_hi                   =   exec_hi
+
+var s_restore_mem_offset        =   ttmp2
+var s_restore_alloc_size        =   ttmp3
+var s_restore_tmp               =   ttmp6               //tba_lo/hi need to be restored
+var s_restore_mem_offset_save   =   s_restore_tmp       //no conflict
+
+var s_restore_m0            =   s_restore_alloc_size    //no conflict
+
+var s_restore_mode          =   ttmp7
+
+var s_restore_pc_lo         =   ttmp0
+var s_restore_pc_hi         =   ttmp1
+var s_restore_exec_lo       =   tma_lo                  //no conflict
+var s_restore_exec_hi       =   tma_hi                  //no conflict
+var s_restore_status        =   ttmp4
+var s_restore_trapsts       =   ttmp5
+var s_restore_xnack_mask_lo =   xnack_mask_lo
+var s_restore_xnack_mask_hi =   xnack_mask_hi
+var s_restore_buf_rsrc0     =   ttmp8
+var s_restore_buf_rsrc1     =   ttmp9
+var s_restore_buf_rsrc2     =   ttmp10
+var s_restore_buf_rsrc3     =   ttmp11
+
+/**************************************************************************/
+/*                      trap handler entry points                         */
+/**************************************************************************/
+/* Shader Main*/
+
+shader main
+  asic(VI)
+  type(CS)
+
+
+        s_branch L_SKIP_RESTORE                                     //NOT restore. might be a regular trap or save
+
+L_JUMP_TO_RESTORE:
+    s_branch L_RESTORE                                              //restore
+
+L_SKIP_RESTORE:
+
+    s_getreg_b32    s_save_status, hwreg(HW_REG_STATUS)                             //save STATUS since we will change SCC
+    s_andn2_b32     s_save_status, s_save_status, SQ_WAVE_STATUS_SPI_PRIO_MASK      //check whether this is for save
+    s_getreg_b32    s_save_trapsts, hwreg(HW_REG_TRAPSTS)
+    s_and_b32       s_save_trapsts, s_save_trapsts, SQ_WAVE_TRAPSTS_SAVECTX_MASK    //check whether this is for save
+    s_cbranch_scc1  L_SAVE                                      //this is the operation for save
+
+    // *********    Handle non-CWSR traps       *******************
+
+    /* read tba and tma for next level trap handler, ttmp4 is used as s_save_status */
+    s_load_dwordx4  [ttmp8,ttmp9,ttmp10, ttmp11], [tma_lo,tma_hi], 0
+    s_waitcnt lgkmcnt(0)
+    s_or_b32        ttmp7, ttmp8, ttmp9
+    s_cbranch_scc0  L_NO_NEXT_TRAP //next level trap handler not been set
+    set_status_without_spi_prio(s_save_status, ttmp2) //restore HW status(SCC)
+    s_setpc_b64     [ttmp8,ttmp9] //jump to next level trap handler
+
+L_NO_NEXT_TRAP:
+    s_getreg_b32    s_save_trapsts, hwreg(HW_REG_TRAPSTS)
+    s_and_b32       s_save_trapsts, s_save_trapsts, SQ_WAVE_TRAPSTS_EXCE_MASK // Check whether it is an exception
+    s_cbranch_scc1  L_EXCP_CASE   // Exception, jump back to the shader program directly.
+    s_add_u32       ttmp0, ttmp0, 4   // S_TRAP case, add 4 to ttmp0
+    s_addc_u32  ttmp1, ttmp1, 0
+L_EXCP_CASE:
+    s_and_b32   ttmp1, ttmp1, 0xFFFF
+    set_status_without_spi_prio(s_save_status, ttmp2) //restore HW status(SCC)
+    s_rfe_b64       [ttmp0, ttmp1]
+
+    // *********        End handling of non-CWSR traps   *******************
+
+/**************************************************************************/
+/*                      save routine                                      */
+/**************************************************************************/
+
+L_SAVE:
+    s_mov_b32       s_save_tmp, 0                                                           //clear saveCtx bit
+    s_setreg_b32    hwreg(HW_REG_TRAPSTS, SQ_WAVE_TRAPSTS_SAVECTX_SHIFT, 1), s_save_tmp     //clear saveCtx bit
+
+    s_mov_b32       s_save_xnack_mask_lo,   xnack_mask_lo                                   //save XNACK_MASK
+    s_mov_b32       s_save_xnack_mask_hi,   xnack_mask_hi    //save XNACK must before any memory operation
+    s_getreg_b32    s_save_tmp, hwreg(HW_REG_IB_STS, SQ_WAVE_IB_STS_RCNT_SHIFT, SQ_WAVE_IB_STS_RCNT_SIZE)                   //save RCNT
+    s_lshl_b32      s_save_tmp, s_save_tmp, S_SAVE_PC_HI_RCNT_SHIFT
+    s_or_b32        s_save_pc_hi, s_save_pc_hi, s_save_tmp
+    s_getreg_b32    s_save_tmp, hwreg(HW_REG_IB_STS, SQ_WAVE_IB_STS_FIRST_REPLAY_SHIFT, SQ_WAVE_IB_STS_FIRST_REPLAY_SIZE)   //save FIRST_REPLAY
+    s_lshl_b32      s_save_tmp, s_save_tmp, S_SAVE_PC_HI_FIRST_REPLAY_SHIFT
+    s_or_b32        s_save_pc_hi, s_save_pc_hi, s_save_tmp
+    s_getreg_b32    s_save_tmp, hwreg(HW_REG_IB_STS)                                        //clear RCNT and FIRST_REPLAY in IB_STS
+    s_and_b32       s_save_tmp, s_save_tmp, SQ_WAVE_IB_STS_RCNT_FIRST_REPLAY_MASK_NEG
+
+    s_setreg_b32    hwreg(HW_REG_IB_STS), s_save_tmp
+
+    /*      inform SPI the readiness and wait for SPI's go signal */
+    s_mov_b32       s_save_exec_lo, exec_lo                                                 //save EXEC and use EXEC for the go signal from SPI
+    s_mov_b32       s_save_exec_hi, exec_hi
+    s_mov_b64       exec,   0x0                                                             //clear EXEC to get ready to receive
+
+        s_sendmsg   sendmsg(MSG_SAVEWAVE)  //send SPI a message and wait for SPI's write to EXEC
+
+    // Set SPI_PRIO=2 to avoid starving instruction fetch in the waves we're waiting for.
+    s_or_b32 s_save_tmp, s_save_status, (2 << SQ_WAVE_STATUS_SPI_PRIO_SHIFT)
+    s_setreg_b32 hwreg(HW_REG_STATUS), s_save_tmp
+
+  L_SLEEP:
+    s_sleep 0x2                // sleep 1 (64clk) is not enough for 8 waves per SIMD, which will cause SQ hang, since the 7,8th wave could not get arbit to exec inst, while other waves are stuck into the sleep-loop and waiting for wrexec!=0
+
+        s_cbranch_execz L_SLEEP
+
+    /*      setup Resource Contants    */
+    s_mov_b32       s_save_buf_rsrc0,   s_save_spi_init_lo                                                      //base_addr_lo
+    s_and_b32       s_save_buf_rsrc1,   s_save_spi_init_hi, 0x0000FFFF                                          //base_addr_hi
+    s_or_b32        s_save_buf_rsrc1,   s_save_buf_rsrc1,  S_SAVE_BUF_RSRC_WORD1_STRIDE
+    s_mov_b32       s_save_buf_rsrc2,   0                                                                       //NUM_RECORDS initial value = 0 (in bytes) although not neccessarily inited
+    s_mov_b32       s_save_buf_rsrc3,   S_SAVE_BUF_RSRC_WORD3_MISC
+    s_and_b32       s_save_tmp,         s_save_spi_init_hi, S_SAVE_SPI_INIT_ATC_MASK
+    s_lshr_b32      s_save_tmp,         s_save_tmp, (S_SAVE_SPI_INIT_ATC_SHIFT-SQ_BUF_RSRC_WORD1_ATC_SHIFT)         //get ATC bit into position
+    s_or_b32        s_save_buf_rsrc3,   s_save_buf_rsrc3,  s_save_tmp                                           //or ATC
+    s_and_b32       s_save_tmp,         s_save_spi_init_hi, S_SAVE_SPI_INIT_MTYPE_MASK
+    s_lshr_b32      s_save_tmp,         s_save_tmp, (S_SAVE_SPI_INIT_MTYPE_SHIFT-SQ_BUF_RSRC_WORD3_MTYPE_SHIFT)     //get MTYPE bits into position
+    s_or_b32        s_save_buf_rsrc3,   s_save_buf_rsrc3,  s_save_tmp                                           //or MTYPE
+
+    //FIXME  right now s_save_m0/s_save_mem_offset use tma_lo/tma_hi  (might need to save them before using them?)
+    s_mov_b32       s_save_m0,          m0                                                                  //save M0
+
+    /*      global mem offset           */
+    s_mov_b32       s_save_mem_offset,  0x0                                                                     //mem offset initial value = 0
+
+
+
+
+    /*      save HW registers   */
+    //////////////////////////////
+
+  L_SAVE_HWREG:
+        // HWREG SR memory offset : size(VGPR)+size(SGPR)
+       get_vgpr_size_bytes(s_save_mem_offset)
+       get_sgpr_size_bytes(s_save_tmp)
+       s_add_u32 s_save_mem_offset, s_save_mem_offset, s_save_tmp
+
+
+    s_mov_b32       s_save_buf_rsrc2, 0x4                               //NUM_RECORDS   in bytes
+        s_mov_b32       s_save_buf_rsrc2,  0x1000000                                //NUM_RECORDS in bytes
+
+
+    write_hwreg_to_mem(s_save_m0, s_save_buf_rsrc0, s_save_mem_offset)                  //M0
+    write_hwreg_to_mem(s_save_pc_lo, s_save_buf_rsrc0, s_save_mem_offset)                   //PC
+    write_hwreg_to_mem(s_save_pc_hi, s_save_buf_rsrc0, s_save_mem_offset)
+    write_hwreg_to_mem(s_save_exec_lo, s_save_buf_rsrc0, s_save_mem_offset)             //EXEC
+    write_hwreg_to_mem(s_save_exec_hi, s_save_buf_rsrc0, s_save_mem_offset)
+    write_hwreg_to_mem(s_save_status, s_save_buf_rsrc0, s_save_mem_offset)              //STATUS
+
+    //s_save_trapsts conflicts with s_save_alloc_size
+    s_getreg_b32    s_save_trapsts, hwreg(HW_REG_TRAPSTS)
+    write_hwreg_to_mem(s_save_trapsts, s_save_buf_rsrc0, s_save_mem_offset)             //TRAPSTS
+
+    write_hwreg_to_mem(s_save_xnack_mask_lo, s_save_buf_rsrc0, s_save_mem_offset)           //XNACK_MASK_LO
+    write_hwreg_to_mem(s_save_xnack_mask_hi, s_save_buf_rsrc0, s_save_mem_offset)           //XNACK_MASK_HI
+
+    //use s_save_tmp would introduce conflict here between s_save_tmp and s_save_buf_rsrc2
+    s_getreg_b32    s_save_m0, hwreg(HW_REG_MODE)                                                   //MODE
+    write_hwreg_to_mem(s_save_m0, s_save_buf_rsrc0, s_save_mem_offset)
+    write_hwreg_to_mem(tba_lo, s_save_buf_rsrc0, s_save_mem_offset)                     //TBA_LO
+    write_hwreg_to_mem(tba_hi, s_save_buf_rsrc0, s_save_mem_offset)                     //TBA_HI
+
+
+
+    /*      the first wave in the threadgroup    */
+        // save fist_wave bits in tba_hi unused bit.26
+    s_and_b32       s_save_tmp, s_save_spi_init_hi, S_SAVE_SPI_INIT_FIRST_WAVE_MASK     // extract fisrt wave bit
+    //s_or_b32        tba_hi, s_save_tmp, tba_hi                                        // save first wave bit to tba_hi.bits[26]
+    s_mov_b32        s_save_exec_hi, 0x0
+    s_or_b32         s_save_exec_hi, s_save_tmp, s_save_exec_hi                          // save first wave bit to s_save_exec_hi.bits[26]
+
+
+    /*          save SGPRs      */
+        // Save SGPR before LDS save, then the s0 to s4 can be used during LDS save...
+    //////////////////////////////
+
+    // SGPR SR memory offset : size(VGPR)
+    get_vgpr_size_bytes(s_save_mem_offset)
+    // TODO, change RSRC word to rearrange memory layout for SGPRS
+
+    s_getreg_b32    s_save_alloc_size, hwreg(HW_REG_GPR_ALLOC,SQ_WAVE_GPR_ALLOC_SGPR_SIZE_SHIFT,SQ_WAVE_GPR_ALLOC_SGPR_SIZE_SIZE)               //spgr_size
+    s_add_u32       s_save_alloc_size, s_save_alloc_size, 1
+    s_lshl_b32      s_save_alloc_size, s_save_alloc_size, 4                         //Number of SGPRs = (sgpr_size + 1) * 16   (non-zero value)
+
+        s_lshl_b32      s_save_buf_rsrc2,   s_save_alloc_size, 2                    //NUM_RECORDS in bytes
+        s_mov_b32       s_save_buf_rsrc2,  0x1000000                                //NUM_RECORDS in bytes
+
+    // backup s_save_buf_rsrc0,1 to s_save_pc_lo/hi, since write_16sgpr_to_mem function will change the rsrc0
+    //s_mov_b64 s_save_pc_lo, s_save_buf_rsrc0
+    s_mov_b64 s_save_xnack_mask_lo, s_save_buf_rsrc0
+    s_add_u32 s_save_buf_rsrc0, s_save_buf_rsrc0, s_save_mem_offset
+    s_addc_u32 s_save_buf_rsrc1, s_save_buf_rsrc1, 0
+
+    s_mov_b32       m0, 0x0                         //SGPR initial index value =0
+  L_SAVE_SGPR_LOOP:
+    // SGPR is allocated in 16 SGPR granularity
+    s_movrels_b64   s0, s0     //s0 = s[0+m0], s1 = s[1+m0]
+    s_movrels_b64   s2, s2     //s2 = s[2+m0], s3 = s[3+m0]
+    s_movrels_b64   s4, s4     //s4 = s[4+m0], s5 = s[5+m0]
+    s_movrels_b64   s6, s6     //s6 = s[6+m0], s7 = s[7+m0]
+    s_movrels_b64   s8, s8     //s8 = s[8+m0], s9 = s[9+m0]
+    s_movrels_b64   s10, s10   //s10 = s[10+m0], s11 = s[11+m0]
+    s_movrels_b64   s12, s12   //s12 = s[12+m0], s13 = s[13+m0]
+    s_movrels_b64   s14, s14   //s14 = s[14+m0], s15 = s[15+m0]
+
+    write_16sgpr_to_mem(s0, s_save_buf_rsrc0, s_save_mem_offset) //PV: the best performance should be using s_buffer_store_dwordx4
+    s_add_u32       m0, m0, 16                                                      //next sgpr index
+    s_cmp_lt_u32    m0, s_save_alloc_size                                           //scc = (m0 < s_save_alloc_size) ? 1 : 0
+    s_cbranch_scc1  L_SAVE_SGPR_LOOP                                    //SGPR save is complete?
+    // restore s_save_buf_rsrc0,1
+    //s_mov_b64 s_save_buf_rsrc0, s_save_pc_lo
+    s_mov_b64 s_save_buf_rsrc0, s_save_xnack_mask_lo
+
+
+
+
+    /*          save first 4 VGPR, then LDS save could use   */
+        // each wave will alloc 4 vgprs at least...
+    /////////////////////////////////////////////////////////////////////////////////////
+
+    s_mov_b32       s_save_mem_offset, 0
+    s_mov_b32       exec_lo, 0xFFFFFFFF                                             //need every thread from now on
+    s_mov_b32       exec_hi, 0xFFFFFFFF
+
+        s_mov_b32       s_save_buf_rsrc2,  0x1000000                                //NUM_RECORDS in bytes
+
+    // VGPR Allocated in 4-GPR granularity
+
+        buffer_store_dword v0, v0, s_save_buf_rsrc0, s_save_mem_offset slc:1 glc:1
+        buffer_store_dword v1, v0, s_save_buf_rsrc0, s_save_mem_offset slc:1 glc:1  offset:256
+        buffer_store_dword v2, v0, s_save_buf_rsrc0, s_save_mem_offset slc:1 glc:1  offset:256*2
+        buffer_store_dword v3, v0, s_save_buf_rsrc0, s_save_mem_offset slc:1 glc:1  offset:256*3
+
+
+
+    /*          save LDS        */
+    //////////////////////////////
+
+  L_SAVE_LDS:
+
+        // Change EXEC to all threads...
+    s_mov_b32       exec_lo, 0xFFFFFFFF   //need every thread from now on
+    s_mov_b32       exec_hi, 0xFFFFFFFF
+
+    s_getreg_b32    s_save_alloc_size, hwreg(HW_REG_LDS_ALLOC,SQ_WAVE_LDS_ALLOC_LDS_SIZE_SHIFT,SQ_WAVE_LDS_ALLOC_LDS_SIZE_SIZE)             //lds_size
+    s_and_b32       s_save_alloc_size, s_save_alloc_size, 0xFFFFFFFF                //lds_size is zero?
+    s_cbranch_scc0  L_SAVE_LDS_DONE                                                                            //no lds used? jump to L_SAVE_DONE
+
+    s_barrier               //LDS is used? wait for other waves in the same TG
+    //s_and_b32     s_save_tmp, tba_hi, S_SAVE_SPI_INIT_FIRST_WAVE_MASK                //exec is still used here
+    s_and_b32       s_save_tmp, s_save_exec_hi, S_SAVE_SPI_INIT_FIRST_WAVE_MASK                //exec is still used here
+    s_cbranch_scc0  L_SAVE_LDS_DONE
+
+        // first wave do LDS save;
+
+    s_lshl_b32      s_save_alloc_size, s_save_alloc_size, 6                         //LDS size in dwords = lds_size * 64dw
+    s_lshl_b32      s_save_alloc_size, s_save_alloc_size, 2                         //LDS size in bytes
+    s_mov_b32       s_save_buf_rsrc2,  s_save_alloc_size                            //NUM_RECORDS in bytes
+
+    // LDS at offset: size(VGPR)+SIZE(SGPR)+SIZE(HWREG)
+    //
+    get_vgpr_size_bytes(s_save_mem_offset)
+    get_sgpr_size_bytes(s_save_tmp)
+    s_add_u32  s_save_mem_offset, s_save_mem_offset, s_save_tmp
+    s_add_u32 s_save_mem_offset, s_save_mem_offset, get_hwreg_size_bytes()
+
+
+        s_mov_b32       s_save_buf_rsrc2,  0x1000000                  //NUM_RECORDS in bytes
+    s_mov_b32       m0, 0x0                                               //lds_offset initial value = 0
+
+
+      v_mbcnt_lo_u32_b32 v2, 0xffffffff, 0x0
+      v_mbcnt_hi_u32_b32 v3, 0xffffffff, v2     // tid
+      v_mul_i32_i24 v2, v3, 8   // tid*8
+      v_mov_b32 v3, 256*2
+      s_mov_b32 m0, 0x10000
+      s_mov_b32 s0, s_save_buf_rsrc3
+      s_and_b32 s_save_buf_rsrc3, s_save_buf_rsrc3, 0xFF7FFFFF    // disable add_tid
+      s_or_b32 s_save_buf_rsrc3, s_save_buf_rsrc3, 0x58000   //DFMT
+
+L_SAVE_LDS_LOOP_VECTOR:
+      ds_read_b64 v[0:1], v2    //x =LDS[a], byte address
+      s_waitcnt lgkmcnt(0)
+      buffer_store_dwordx2  v[0:1], v2, s_save_buf_rsrc0, s_save_mem_offset offen:1  glc:1  slc:1
+//      s_waitcnt vmcnt(0)
+      v_add_u32 v2, vcc[0:1], v2, v3
+      v_cmp_lt_u32 vcc[0:1], v2, s_save_alloc_size
+      s_cbranch_vccnz L_SAVE_LDS_LOOP_VECTOR
+
+      // restore rsrc3
+      s_mov_b32 s_save_buf_rsrc3, s0
+
+L_SAVE_LDS_DONE:
+
+
+    /*          save VGPRs  - set the Rest VGPRs        */
+    //////////////////////////////////////////////////////////////////////////////////////
+  L_SAVE_VGPR:
+    // VGPR SR memory offset: 0
+    // TODO rearrange the RSRC words to use swizzle for VGPR save...
+
+    s_mov_b32       s_save_mem_offset, (0+256*4)                                    // for the rest VGPRs
+    s_mov_b32       exec_lo, 0xFFFFFFFF                                             //need every thread from now on
+    s_mov_b32       exec_hi, 0xFFFFFFFF
+
+    s_getreg_b32    s_save_alloc_size, hwreg(HW_REG_GPR_ALLOC,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SIZE)                   //vpgr_size
+    s_add_u32       s_save_alloc_size, s_save_alloc_size, 1
+    s_lshl_b32      s_save_alloc_size, s_save_alloc_size, 2                         //Number of VGPRs = (vgpr_size + 1) * 4    (non-zero value)   //FIXME for GFX, zero is possible
+    s_lshl_b32      s_save_buf_rsrc2,  s_save_alloc_size, 8                         //NUM_RECORDS in bytes (64 threads*4)
+        s_mov_b32       s_save_buf_rsrc2,  0x1000000                                //NUM_RECORDS in bytes
+
+    // VGPR store using dw burst
+    s_mov_b32         m0, 0x4   //VGPR initial index value =0
+    s_cmp_lt_u32      m0, s_save_alloc_size
+    s_cbranch_scc0    L_SAVE_VGPR_END
+
+
+    s_set_gpr_idx_on    m0, 0x1 //M0[7:0] = M0[7:0] and M0[15:12] = 0x1
+    s_add_u32       s_save_alloc_size, s_save_alloc_size, 0x1000                    //add 0x1000 since we compare m0 against it later
+
+  L_SAVE_VGPR_LOOP:
+    v_mov_b32       v0, v0              //v0 = v[0+m0]
+    v_mov_b32       v1, v1              //v0 = v[0+m0]
+    v_mov_b32       v2, v2              //v0 = v[0+m0]
+    v_mov_b32       v3, v3              //v0 = v[0+m0]
+
+        buffer_store_dword v0, v0, s_save_buf_rsrc0, s_save_mem_offset slc:1 glc:1
+        buffer_store_dword v1, v0, s_save_buf_rsrc0, s_save_mem_offset slc:1 glc:1  offset:256
+        buffer_store_dword v2, v0, s_save_buf_rsrc0, s_save_mem_offset slc:1 glc:1  offset:256*2
+        buffer_store_dword v3, v0, s_save_buf_rsrc0, s_save_mem_offset slc:1 glc:1  offset:256*3
+
+    s_add_u32       m0, m0, 4                                                       //next vgpr index
+    s_add_u32       s_save_mem_offset, s_save_mem_offset, 256*4                     //every buffer_store_dword does 256 bytes
+    s_cmp_lt_u32    m0, s_save_alloc_size                                           //scc = (m0 < s_save_alloc_size) ? 1 : 0
+    s_cbranch_scc1  L_SAVE_VGPR_LOOP                                                //VGPR save is complete?
+    s_set_gpr_idx_off
+
+L_SAVE_VGPR_END:
+    s_branch    L_END_PGM
+
+
+
+/**************************************************************************/
+/*                      restore routine                                   */
+/**************************************************************************/
+
+L_RESTORE:
+    /*      Setup Resource Contants    */
+    s_mov_b32       s_restore_buf_rsrc0,    s_restore_spi_init_lo                                                           //base_addr_lo
+    s_and_b32       s_restore_buf_rsrc1,    s_restore_spi_init_hi, 0x0000FFFF                                               //base_addr_hi
+    s_or_b32        s_restore_buf_rsrc1,    s_restore_buf_rsrc1,  S_RESTORE_BUF_RSRC_WORD1_STRIDE
+    s_mov_b32       s_restore_buf_rsrc2,    0                                                                               //NUM_RECORDS initial value = 0 (in bytes)
+    s_mov_b32       s_restore_buf_rsrc3,    S_RESTORE_BUF_RSRC_WORD3_MISC
+    s_and_b32       s_restore_tmp,          s_restore_spi_init_hi, S_RESTORE_SPI_INIT_ATC_MASK
+    s_lshr_b32      s_restore_tmp,          s_restore_tmp, (S_RESTORE_SPI_INIT_ATC_SHIFT-SQ_BUF_RSRC_WORD1_ATC_SHIFT)       //get ATC bit into position
+    s_or_b32        s_restore_buf_rsrc3,    s_restore_buf_rsrc3,  s_restore_tmp                                             //or ATC
+    s_and_b32       s_restore_tmp,          s_restore_spi_init_hi, S_RESTORE_SPI_INIT_MTYPE_MASK
+    s_lshr_b32      s_restore_tmp,          s_restore_tmp, (S_RESTORE_SPI_INIT_MTYPE_SHIFT-SQ_BUF_RSRC_WORD3_MTYPE_SHIFT)   //get MTYPE bits into position
+    s_or_b32        s_restore_buf_rsrc3,    s_restore_buf_rsrc3,  s_restore_tmp                                             //or MTYPE
+
+    /*      global mem offset           */
+//  s_mov_b32       s_restore_mem_offset, 0x0                               //mem offset initial value = 0
+
+    /*      the first wave in the threadgroup    */
+    s_and_b32       s_restore_tmp, s_restore_spi_init_hi, S_RESTORE_SPI_INIT_FIRST_WAVE_MASK
+    s_cbranch_scc0  L_RESTORE_VGPR
+
+    /*          restore LDS     */
+    //////////////////////////////
+  L_RESTORE_LDS:
+
+    s_mov_b32       exec_lo, 0xFFFFFFFF                                                     //need every thread from now on   //be consistent with SAVE although can be moved ahead
+    s_mov_b32       exec_hi, 0xFFFFFFFF
+
+    s_getreg_b32    s_restore_alloc_size, hwreg(HW_REG_LDS_ALLOC,SQ_WAVE_LDS_ALLOC_LDS_SIZE_SHIFT,SQ_WAVE_LDS_ALLOC_LDS_SIZE_SIZE)              //lds_size
+    s_and_b32       s_restore_alloc_size, s_restore_alloc_size, 0xFFFFFFFF                  //lds_size is zero?
+    s_cbranch_scc0  L_RESTORE_VGPR                                                          //no lds used? jump to L_RESTORE_VGPR
+    s_lshl_b32      s_restore_alloc_size, s_restore_alloc_size, 6                           //LDS size in dwords = lds_size * 64dw
+    s_lshl_b32      s_restore_alloc_size, s_restore_alloc_size, 2                           //LDS size in bytes
+    s_mov_b32       s_restore_buf_rsrc2,    s_restore_alloc_size                            //NUM_RECORDS in bytes
+
+    // LDS at offset: size(VGPR)+SIZE(SGPR)+SIZE(HWREG)
+    //
+    get_vgpr_size_bytes(s_restore_mem_offset)
+    get_sgpr_size_bytes(s_restore_tmp)
+    s_add_u32  s_restore_mem_offset, s_restore_mem_offset, s_restore_tmp
+    s_add_u32  s_restore_mem_offset, s_restore_mem_offset, get_hwreg_size_bytes()            //FIXME, Check if offset overflow???
+
+
+        s_mov_b32       s_restore_buf_rsrc2,  0x1000000                                     //NUM_RECORDS in bytes
+    s_mov_b32       m0, 0x0                                                                 //lds_offset initial value = 0
+
+  L_RESTORE_LDS_LOOP:
+        buffer_load_dword   v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset lds:1                    // first 64DW
+        buffer_load_dword   v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset lds:1 offset:256         // second 64DW
+    s_add_u32       m0, m0, 256*2                                               // 128 DW
+    s_add_u32       s_restore_mem_offset, s_restore_mem_offset, 256*2           //mem offset increased by 128DW
+    s_cmp_lt_u32    m0, s_restore_alloc_size                                    //scc=(m0 < s_restore_alloc_size) ? 1 : 0
+    s_cbranch_scc1  L_RESTORE_LDS_LOOP                                                      //LDS restore is complete?
+
+
+    /*          restore VGPRs       */
+    //////////////////////////////
+  L_RESTORE_VGPR:
+        // VGPR SR memory offset : 0
+    s_mov_b32       s_restore_mem_offset, 0x0
+    s_mov_b32       exec_lo, 0xFFFFFFFF                                                     //need every thread from now on   //be consistent with SAVE although can be moved ahead
+    s_mov_b32       exec_hi, 0xFFFFFFFF
+
+    s_getreg_b32    s_restore_alloc_size, hwreg(HW_REG_GPR_ALLOC,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SIZE)    //vpgr_size
+    s_add_u32       s_restore_alloc_size, s_restore_alloc_size, 1
+    s_lshl_b32      s_restore_alloc_size, s_restore_alloc_size, 2                           //Number of VGPRs = (vgpr_size + 1) * 4    (non-zero value)
+    s_lshl_b32      s_restore_buf_rsrc2,  s_restore_alloc_size, 8                           //NUM_RECORDS in bytes (64 threads*4)
+        s_mov_b32       s_restore_buf_rsrc2,  0x1000000                                     //NUM_RECORDS in bytes
+
+    // VGPR load using dw burst
+    s_mov_b32       s_restore_mem_offset_save, s_restore_mem_offset     // restore start with v1, v0 will be the last
+    s_add_u32       s_restore_mem_offset, s_restore_mem_offset, 256*4
+    s_mov_b32       m0, 4                               //VGPR initial index value = 1
+    s_set_gpr_idx_on  m0, 0x8                       //M0[7:0] = M0[7:0] and M0[15:12] = 0x8
+    s_add_u32       s_restore_alloc_size, s_restore_alloc_size, 0x8000                      //add 0x8000 since we compare m0 against it later
+
+  L_RESTORE_VGPR_LOOP:
+        buffer_load_dword v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset slc:1 glc:1
+        buffer_load_dword v1, v0, s_restore_buf_rsrc0, s_restore_mem_offset slc:1 glc:1 offset:256
+        buffer_load_dword v2, v0, s_restore_buf_rsrc0, s_restore_mem_offset slc:1 glc:1 offset:256*2
+        buffer_load_dword v3, v0, s_restore_buf_rsrc0, s_restore_mem_offset slc:1 glc:1 offset:256*3
+    s_waitcnt       vmcnt(0)                                                                //ensure data ready
+    v_mov_b32       v0, v0                                                                  //v[0+m0] = v0
+    v_mov_b32       v1, v1
+    v_mov_b32       v2, v2
+    v_mov_b32       v3, v3
+    s_add_u32       m0, m0, 4                                                               //next vgpr index
+    s_add_u32       s_restore_mem_offset, s_restore_mem_offset, 256*4                           //every buffer_load_dword does 256 bytes
+    s_cmp_lt_u32    m0, s_restore_alloc_size                                                //scc = (m0 < s_restore_alloc_size) ? 1 : 0
+    s_cbranch_scc1  L_RESTORE_VGPR_LOOP                                                     //VGPR restore (except v0) is complete?
+    s_set_gpr_idx_off
+                                                                                            /* VGPR restore on v0 */
+        buffer_load_dword v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save    slc:1 glc:1
+        buffer_load_dword v1, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save    slc:1 glc:1 offset:256
+        buffer_load_dword v2, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save    slc:1 glc:1 offset:256*2
+        buffer_load_dword v3, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save    slc:1 glc:1 offset:256*3
+        s_waitcnt vmcnt(0)
+
+    /*          restore SGPRs       */
+    //////////////////////////////
+
+    // SGPR SR memory offset : size(VGPR)
+    get_vgpr_size_bytes(s_restore_mem_offset)
+    get_sgpr_size_bytes(s_restore_tmp)
+    s_add_u32 s_restore_mem_offset, s_restore_mem_offset, s_restore_tmp
+    s_sub_u32 s_restore_mem_offset, s_restore_mem_offset, 16*4     // restore SGPR from S[n] to S[0], by 16 sgprs group
+    // TODO, change RSRC word to rearrange memory layout for SGPRS
+
+    s_getreg_b32    s_restore_alloc_size, hwreg(HW_REG_GPR_ALLOC,SQ_WAVE_GPR_ALLOC_SGPR_SIZE_SHIFT,SQ_WAVE_GPR_ALLOC_SGPR_SIZE_SIZE)                //spgr_size
+    s_add_u32       s_restore_alloc_size, s_restore_alloc_size, 1
+    s_lshl_b32      s_restore_alloc_size, s_restore_alloc_size, 4                           //Number of SGPRs = (sgpr_size + 1) * 16   (non-zero value)
+
+        s_lshl_b32      s_restore_buf_rsrc2,    s_restore_alloc_size, 2                     //NUM_RECORDS in bytes
+        s_mov_b32       s_restore_buf_rsrc2,  0x1000000                                     //NUM_RECORDS in bytes
+
+    /* If 112 SGPRs ar allocated, 4 sgprs are not used TBA(108,109),TMA(110,111),
+       However, we are safe to restore these 4 SGPRs anyway, since TBA,TMA will later be restored by HWREG
+    */
+    s_mov_b32 m0, s_restore_alloc_size
+
+ L_RESTORE_SGPR_LOOP:
+    read_16sgpr_from_mem(s0, s_restore_buf_rsrc0, s_restore_mem_offset)  //PV: further performance improvement can be made
+    s_waitcnt       lgkmcnt(0)                                                              //ensure data ready
+
+    s_sub_u32 m0, m0, 16    // Restore from S[n] to S[0]
+
+    s_movreld_b64   s0, s0      //s[0+m0] = s0
+    s_movreld_b64   s2, s2
+    s_movreld_b64   s4, s4
+    s_movreld_b64   s6, s6
+    s_movreld_b64   s8, s8
+    s_movreld_b64   s10, s10
+    s_movreld_b64   s12, s12
+    s_movreld_b64   s14, s14
+
+    s_cmp_eq_u32    m0, 0               //scc = (m0 < s_restore_alloc_size) ? 1 : 0
+    s_cbranch_scc0  L_RESTORE_SGPR_LOOP             //SGPR restore (except s0) is complete?
+
+    /*      restore HW registers    */
+    //////////////////////////////
+  L_RESTORE_HWREG:
+
+    // HWREG SR memory offset : size(VGPR)+size(SGPR)
+    get_vgpr_size_bytes(s_restore_mem_offset)
+    get_sgpr_size_bytes(s_restore_tmp)
+    s_add_u32 s_restore_mem_offset, s_restore_mem_offset, s_restore_tmp
+
+
+    s_mov_b32       s_restore_buf_rsrc2, 0x4                                                //NUM_RECORDS   in bytes
+        s_mov_b32       s_restore_buf_rsrc2,  0x1000000                                     //NUM_RECORDS in bytes
+
+    read_hwreg_from_mem(s_restore_m0, s_restore_buf_rsrc0, s_restore_mem_offset)                    //M0
+    read_hwreg_from_mem(s_restore_pc_lo, s_restore_buf_rsrc0, s_restore_mem_offset)             //PC
+    read_hwreg_from_mem(s_restore_pc_hi, s_restore_buf_rsrc0, s_restore_mem_offset)
+    read_hwreg_from_mem(s_restore_exec_lo, s_restore_buf_rsrc0, s_restore_mem_offset)               //EXEC
+    read_hwreg_from_mem(s_restore_exec_hi, s_restore_buf_rsrc0, s_restore_mem_offset)
+    read_hwreg_from_mem(s_restore_status, s_restore_buf_rsrc0, s_restore_mem_offset)                //STATUS
+    read_hwreg_from_mem(s_restore_trapsts, s_restore_buf_rsrc0, s_restore_mem_offset)               //TRAPSTS
+    read_hwreg_from_mem(xnack_mask_lo, s_restore_buf_rsrc0, s_restore_mem_offset)                   //XNACK_MASK_LO
+    read_hwreg_from_mem(xnack_mask_hi, s_restore_buf_rsrc0, s_restore_mem_offset)                   //XNACK_MASK_HI
+    read_hwreg_from_mem(s_restore_mode, s_restore_buf_rsrc0, s_restore_mem_offset)              //MODE
+    read_hwreg_from_mem(tba_lo, s_restore_buf_rsrc0, s_restore_mem_offset)                      //TBA_LO
+    read_hwreg_from_mem(tba_hi, s_restore_buf_rsrc0, s_restore_mem_offset)                      //TBA_HI
+
+    s_waitcnt       lgkmcnt(0)                                                                                      //from now on, it is safe to restore STATUS and IB_STS
+
+    s_mov_b32       m0,         s_restore_m0
+    s_mov_b32       exec_lo,    s_restore_exec_lo
+    s_mov_b32       exec_hi,    s_restore_exec_hi
+
+    s_and_b32       s_restore_m0, SQ_WAVE_TRAPSTS_PRE_SAVECTX_MASK, s_restore_trapsts
+    s_setreg_b32    hwreg(HW_REG_TRAPSTS, SQ_WAVE_TRAPSTS_PRE_SAVECTX_SHIFT, SQ_WAVE_TRAPSTS_PRE_SAVECTX_SIZE), s_restore_m0
+    s_and_b32       s_restore_m0, SQ_WAVE_TRAPSTS_POST_SAVECTX_MASK, s_restore_trapsts
+    s_lshr_b32      s_restore_m0, s_restore_m0, SQ_WAVE_TRAPSTS_POST_SAVECTX_SHIFT
+    s_setreg_b32    hwreg(HW_REG_TRAPSTS, SQ_WAVE_TRAPSTS_POST_SAVECTX_SHIFT, SQ_WAVE_TRAPSTS_POST_SAVECTX_SIZE), s_restore_m0
+    //s_setreg_b32  hwreg(HW_REG_TRAPSTS),  s_restore_trapsts      //don't overwrite SAVECTX bit as it may be set through external SAVECTX during restore
+    s_setreg_b32    hwreg(HW_REG_MODE),     s_restore_mode
+    //reuse s_restore_m0 as a temp register
+    s_and_b32       s_restore_m0, s_restore_pc_hi, S_SAVE_PC_HI_RCNT_MASK
+    s_lshr_b32      s_restore_m0, s_restore_m0, S_SAVE_PC_HI_RCNT_SHIFT
+    s_lshl_b32      s_restore_m0, s_restore_m0, SQ_WAVE_IB_STS_RCNT_SHIFT
+    s_mov_b32       s_restore_tmp, 0x0                                                                              //IB_STS is zero
+    s_or_b32        s_restore_tmp, s_restore_tmp, s_restore_m0
+    s_and_b32       s_restore_m0, s_restore_pc_hi, S_SAVE_PC_HI_FIRST_REPLAY_MASK
+    s_lshr_b32      s_restore_m0, s_restore_m0, S_SAVE_PC_HI_FIRST_REPLAY_SHIFT
+    s_lshl_b32      s_restore_m0, s_restore_m0, SQ_WAVE_IB_STS_FIRST_REPLAY_SHIFT
+    s_or_b32        s_restore_tmp, s_restore_tmp, s_restore_m0
+    s_and_b32       s_restore_m0, s_restore_status, SQ_WAVE_STATUS_INST_ATC_MASK
+    s_lshr_b32      s_restore_m0, s_restore_m0, SQ_WAVE_STATUS_INST_ATC_SHIFT
+    s_setreg_b32    hwreg(HW_REG_IB_STS),   s_restore_tmp
+
+    s_and_b32 s_restore_pc_hi, s_restore_pc_hi, 0x0000ffff      //pc[47:32]        //Do it here in order not to affect STATUS
+    s_and_b64    exec, exec, exec  // Restore STATUS.EXECZ, not writable by s_setreg_b32
+    s_and_b64    vcc, vcc, vcc  // Restore STATUS.VCCZ, not writable by s_setreg_b32
+    set_status_without_spi_prio(s_restore_status, s_restore_tmp) // SCC is included, which is changed by previous salu
+
+    s_barrier                                                   //barrier to ensure the readiness of LDS before access attempts from any other wave in the same TG //FIXME not performance-optimal at this time
+
+//  s_rfe_b64 s_restore_pc_lo                                   //Return to the main shader program and resume execution
+    s_rfe_restore_b64  s_restore_pc_lo, s_restore_m0            // s_restore_m0[0] is used to set STATUS.inst_atc
+
+
+/**************************************************************************/
+/*                      the END                                           */
+/**************************************************************************/
+L_END_PGM:
+    s_endpgm
+
+end
+
+
+/**************************************************************************/
+/*                      the helper functions                              */
+/**************************************************************************/
+
+//Only for save hwreg to mem
+function write_hwreg_to_mem(s, s_rsrc, s_mem_offset)
+        s_mov_b32 exec_lo, m0                   //assuming exec_lo is not needed anymore from this point on
+        s_mov_b32 m0, s_mem_offset
+        s_buffer_store_dword s, s_rsrc, m0      glc:1
+        s_add_u32       s_mem_offset, s_mem_offset, 4
+        s_mov_b32   m0, exec_lo
+end
+
+
+// HWREG are saved before SGPRs, so all HWREG could be use.
+function write_16sgpr_to_mem(s, s_rsrc, s_mem_offset)
+
+        s_buffer_store_dwordx4 s[0], s_rsrc, 0  glc:1
+        s_buffer_store_dwordx4 s[4], s_rsrc, 16  glc:1
+        s_buffer_store_dwordx4 s[8], s_rsrc, 32  glc:1
+        s_buffer_store_dwordx4 s[12], s_rsrc, 48 glc:1
+        s_add_u32       s_rsrc[0], s_rsrc[0], 4*16
+        s_addc_u32      s_rsrc[1], s_rsrc[1], 0x0             // +scc
+end
+
+
+function read_hwreg_from_mem(s, s_rsrc, s_mem_offset)
+    s_buffer_load_dword s, s_rsrc, s_mem_offset     glc:1
+    s_add_u32       s_mem_offset, s_mem_offset, 4
+end
+
+function read_16sgpr_from_mem(s, s_rsrc, s_mem_offset)
+    s_buffer_load_dwordx16 s, s_rsrc, s_mem_offset      glc:1
+    s_sub_u32       s_mem_offset, s_mem_offset, 4*16
+end
+
+
+
+function get_lds_size_bytes(s_lds_size_byte)
+    // SQ LDS granularity is 64DW, while PGM_RSRC2.lds_size is in granularity 128DW
+    s_getreg_b32   s_lds_size_byte, hwreg(HW_REG_LDS_ALLOC, SQ_WAVE_LDS_ALLOC_LDS_SIZE_SHIFT, SQ_WAVE_LDS_ALLOC_LDS_SIZE_SIZE)          // lds_size
+    s_lshl_b32     s_lds_size_byte, s_lds_size_byte, 8                      //LDS size in dwords = lds_size * 64 *4Bytes    // granularity 64DW
+end
+
+function get_vgpr_size_bytes(s_vgpr_size_byte)
+    s_getreg_b32   s_vgpr_size_byte, hwreg(HW_REG_GPR_ALLOC,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SIZE)  //vpgr_size
+    s_add_u32      s_vgpr_size_byte, s_vgpr_size_byte, 1
+    s_lshl_b32     s_vgpr_size_byte, s_vgpr_size_byte, (2+8) //Number of VGPRs = (vgpr_size + 1) * 4 * 64 * 4   (non-zero value)   //FIXME for GFX, zero is possible
+end
+
+function get_sgpr_size_bytes(s_sgpr_size_byte)
+    s_getreg_b32   s_sgpr_size_byte, hwreg(HW_REG_GPR_ALLOC,SQ_WAVE_GPR_ALLOC_SGPR_SIZE_SHIFT,SQ_WAVE_GPR_ALLOC_SGPR_SIZE_SIZE)  //spgr_size
+    s_add_u32      s_sgpr_size_byte, s_sgpr_size_byte, 1
+    s_lshl_b32     s_sgpr_size_byte, s_sgpr_size_byte, 6 //Number of SGPRs = (sgpr_size + 1) * 16 *4   (non-zero value)
+end
+
+function get_hwreg_size_bytes
+    return 128 //HWREG size 128 bytes
+end
+
+function set_status_without_spi_prio(status, tmp)
+    // Do not restore STATUS.SPI_PRIO since scheduler may have raised it.
+    s_lshr_b32      tmp, status, SQ_WAVE_STATUS_POST_SPI_PRIO_SHIFT
+    s_setreg_b32    hwreg(HW_REG_STATUS, SQ_WAVE_STATUS_POST_SPI_PRIO_SHIFT, SQ_WAVE_STATUS_POST_SPI_PRIO_SIZE), tmp
+    s_nop           0x2 // avoid S_SETREG => S_SETREG hazard
+    s_setreg_b32    hwreg(HW_REG_STATUS, SQ_WAVE_STATUS_PRE_SPI_PRIO_SHIFT, SQ_WAVE_STATUS_PRE_SPI_PRIO_SIZE), status
+end
diff --git a/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx9.asm b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx9.asm
new file mode 100644
index 000000000000..6869e07a2fff
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx9.asm
@@ -0,0 +1,1135 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/* To compile this assembly code:
+ *
+ * gfx9:
+ *   cpp -DASIC_FAMILY=CHIP_VEGAM cwsr_trap_handler_gfx9.asm -P -o gfx9.sp3
+ *   sp3 gfx9.sp3 -hex gfx9.hex
+ *
+ * arcturus:
+ *   cpp -DASIC_FAMILY=CHIP_ARCTURUS cwsr_trap_handler_gfx9.asm -P -o arcturus.sp3
+ *   sp3 arcturus.sp3 -hex arcturus.hex
+ *
+ * aldebaran:
+ *   cpp -DASIC_FAMILY=CHIP_ALDEBARAN cwsr_trap_handler_gfx9.asm -P -o aldebaran.sp3
+ *   sp3 aldebaran.sp3 -hex aldebaran.hex
+ *
+ * gc_9_4_3:
+ *   cpp -DASIC_FAMILY=GC_9_4_3 cwsr_trap_handler_gfx9.asm -P -o gc_9_4_3.sp3
+ *   sp3 gc_9_4_3.sp3 -hex gc_9_4_3.hex
+ *
+ * gc_9_5_0:
+ *   cpp -DASIC_FAMILY=GC_9_5_0 cwsr_trap_handler_gfx9.asm -P -o gc_9_5_0.sp3
+ *   sp3 gc_9_5_0.sp3 -hex gc_9_5_0.hex
+ */
+
+#define CHIP_VEGAM 18
+#define CHIP_ARCTURUS 23
+#define CHIP_ALDEBARAN 25
+#define CHIP_GC_9_4_3 26
+#define CHIP_GC_9_5_0 27
+
+var ACK_SQC_STORE		    =	1		    //workaround for suspected SQC store bug causing incorrect stores under concurrency
+var SAVE_AFTER_XNACK_ERROR	    =	1		    //workaround for TCP store failure after XNACK error when ALLOW_REPLAY=0, for debugger
+var SINGLE_STEP_MISSED_WORKAROUND   =	(ASIC_FAMILY <= CHIP_ALDEBARAN)	//workaround for lost MODE.DEBUG_EN exception when SAVECTX raised
+
+#if ASIC_FAMILY < CHIP_GC_9_4_3
+#define VMEM_MODIFIERS slc:1 glc:1
+#else
+#define VMEM_MODIFIERS sc0:1 nt:1
+#endif
+
+/**************************************************************************/
+/*			variables					  */
+/**************************************************************************/
+var SQ_WAVE_STATUS_SPI_PRIO_SHIFT  = 1
+var SQ_WAVE_STATUS_SPI_PRIO_MASK   = 0x00000006
+var SQ_WAVE_STATUS_HALT_MASK       = 0x2000
+var SQ_WAVE_STATUS_PRE_SPI_PRIO_SHIFT   = 0
+var SQ_WAVE_STATUS_PRE_SPI_PRIO_SIZE    = 1
+var SQ_WAVE_STATUS_POST_SPI_PRIO_SHIFT  = 3
+var SQ_WAVE_STATUS_POST_SPI_PRIO_SIZE   = 29
+var SQ_WAVE_STATUS_ALLOW_REPLAY_MASK    = 0x400000
+var SQ_WAVE_STATUS_ECC_ERR_MASK         = 0x20000
+
+var SQ_WAVE_LDS_ALLOC_LDS_SIZE_SHIFT	= 12
+#if ASIC_FAMILY >= CHIP_GC_9_5_0
+var SQ_WAVE_LDS_ALLOC_LDS_SIZE_SIZE	= 11
+var LDS_RESTORE_GRANULARITY_BYTES	= 1280
+#else
+var SQ_WAVE_LDS_ALLOC_LDS_SIZE_SIZE	= 9
+var LDS_RESTORE_GRANULARITY_BYTES	= 512
+#endif
+var SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SIZE	= 6
+var SQ_WAVE_GPR_ALLOC_SGPR_SIZE_SIZE	= 3			//FIXME	 sq.blk still has 4 bits at this time while SQ programming guide has 3 bits
+var SQ_WAVE_GPR_ALLOC_SGPR_SIZE_SHIFT	= 24
+
+#if ASIC_FAMILY >= CHIP_ALDEBARAN
+var SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT	= 6
+var SQ_WAVE_GPR_ALLOC_ACCV_OFFSET_SHIFT	= 12
+var SQ_WAVE_GPR_ALLOC_ACCV_OFFSET_SIZE	= 6
+#else
+var SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT	= 8
+#endif
+
+var SQ_WAVE_TRAPSTS_SAVECTX_MASK    =	0x400
+var SQ_WAVE_TRAPSTS_EXCP_MASK	    =	0x1FF
+var SQ_WAVE_TRAPSTS_SAVECTX_SHIFT   =	10
+var SQ_WAVE_TRAPSTS_ADDR_WATCH_MASK =	0x80
+var SQ_WAVE_TRAPSTS_ADDR_WATCH_SHIFT =	7
+var SQ_WAVE_TRAPSTS_MEM_VIOL_MASK   =	0x100
+var SQ_WAVE_TRAPSTS_MEM_VIOL_SHIFT  =	8
+var SQ_WAVE_TRAPSTS_HOST_TRAP_MASK  =	0x400000
+var SQ_WAVE_TRAPSTS_WAVE_BEGIN_MASK =	0x800000
+var SQ_WAVE_TRAPSTS_WAVE_END_MASK   =	0x1000000
+var SQ_WAVE_TRAPSTS_TRAP_AFTER_INST_MASK =  0x2000000
+var SQ_WAVE_TRAPSTS_PRE_SAVECTX_MASK	=   0x3FF
+var SQ_WAVE_TRAPSTS_PRE_SAVECTX_SHIFT	=   0x0
+var SQ_WAVE_TRAPSTS_PRE_SAVECTX_SIZE	=   10
+var SQ_WAVE_TRAPSTS_POST_SAVECTX_MASK	=   0xFFFFF800
+var SQ_WAVE_TRAPSTS_POST_SAVECTX_SHIFT	=   11
+var SQ_WAVE_TRAPSTS_POST_SAVECTX_SIZE	=   21
+var SQ_WAVE_TRAPSTS_ILLEGAL_INST_MASK	=   0x800
+var SQ_WAVE_TRAPSTS_EXCP_HI_MASK	=   0x7000
+var SQ_WAVE_TRAPSTS_XNACK_ERROR_MASK	=   0x10000000
+
+var SQ_WAVE_MODE_EXCP_EN_SHIFT		=   12
+var SQ_WAVE_MODE_EXCP_EN_ADDR_WATCH_SHIFT	= 19
+
+var SQ_WAVE_IB_STS_FIRST_REPLAY_SHIFT	=   15			//FIXME
+var SQ_WAVE_IB_STS_RCNT_FIRST_REPLAY_MASK	= 0x1F8000
+
+var SQ_WAVE_MODE_DEBUG_EN_MASK		=   0x800
+
+var TTMP_SAVE_RCNT_FIRST_REPLAY_SHIFT	=   26			// bits [31:26] unused by SPI debug data
+var TTMP_SAVE_RCNT_FIRST_REPLAY_MASK	=   0xFC000000
+var TTMP_DEBUG_TRAP_ENABLED_SHIFT	=   23
+var TTMP_DEBUG_TRAP_ENABLED_MASK	=   0x800000
+
+/*	Save	    */
+var S_SAVE_BUF_RSRC_WORD1_STRIDE	=   0x00040000		//stride is 4 bytes
+var S_SAVE_BUF_RSRC_WORD3_MISC		=   0x00807FAC		//SQ_SEL_X/Y/Z/W, BUF_NUM_FORMAT_FLOAT, (0 for MUBUF stride[17:14] when ADD_TID_ENABLE and BUF_DATA_FORMAT_32 for MTBUF), ADD_TID_ENABLE
+var S_SAVE_PC_HI_TRAP_ID_MASK		=   0x00FF0000
+var S_SAVE_PC_HI_HT_MASK		=   0x01000000
+var S_SAVE_SPI_INIT_FIRST_WAVE_MASK	=   0x04000000		//bit[26]: FirstWaveInTG
+var S_SAVE_SPI_INIT_FIRST_WAVE_SHIFT	=   26
+
+var s_save_spi_init_lo		    =	exec_lo
+var s_save_spi_init_hi		    =	exec_hi
+
+var s_save_pc_lo	    =	ttmp0		//{TTMP1, TTMP0} = {3'h0,pc_rewind[3:0], HT[0],trapID[7:0], PC[47:0]}
+var s_save_pc_hi	    =	ttmp1
+var s_save_exec_lo	    =	ttmp2
+var s_save_exec_hi	    =	ttmp3
+var s_save_tmp		    =	ttmp14
+var s_save_trapsts	    =	ttmp15		//not really used until the end of the SAVE routine
+var s_save_xnack_mask_lo    =	ttmp6
+var s_save_xnack_mask_hi    =	ttmp7
+var s_save_buf_rsrc0	    =	ttmp8
+var s_save_buf_rsrc1	    =	ttmp9
+var s_save_buf_rsrc2	    =	ttmp10
+var s_save_buf_rsrc3	    =	ttmp11
+var s_save_status	    =	ttmp12
+var s_save_mem_offset	    =	ttmp4
+var s_save_alloc_size	    =	s_save_trapsts		//conflict
+var s_save_m0		    =	ttmp5
+var s_save_ttmps_lo	    =	s_save_tmp		//no conflict
+var s_save_ttmps_hi	    =	s_save_trapsts		//no conflict
+#if ASIC_FAMILY >= CHIP_GC_9_4_3
+var s_save_ib_sts       =	ttmp13
+#else
+var s_save_ib_sts       =	ttmp11
+#endif
+
+/*	Restore	    */
+var S_RESTORE_BUF_RSRC_WORD1_STRIDE	    =	S_SAVE_BUF_RSRC_WORD1_STRIDE
+var S_RESTORE_BUF_RSRC_WORD3_MISC	    =	S_SAVE_BUF_RSRC_WORD3_MISC
+
+var S_RESTORE_SPI_INIT_FIRST_WAVE_MASK	    =	0x04000000	    //bit[26]: FirstWaveInTG
+var S_RESTORE_SPI_INIT_FIRST_WAVE_SHIFT	    =	26
+
+var s_restore_spi_init_lo		    =	exec_lo
+var s_restore_spi_init_hi		    =	exec_hi
+
+var s_restore_mem_offset	=   ttmp12
+var s_restore_tmp2		=   ttmp13
+var s_restore_alloc_size	=   ttmp3
+var s_restore_tmp		=   ttmp2
+var s_restore_mem_offset_save	=   s_restore_tmp	//no conflict
+var s_restore_accvgpr_offset_save = ttmp7
+
+var s_restore_m0	    =	s_restore_alloc_size	//no conflict
+
+var s_restore_mode	    =	s_restore_accvgpr_offset_save
+
+var s_restore_pc_lo	    =	ttmp0
+var s_restore_pc_hi	    =	ttmp1
+var s_restore_exec_lo	    =	ttmp4
+var s_restore_exec_hi	    = 	ttmp5
+var s_restore_status	    =	ttmp14
+var s_restore_trapsts	    =	ttmp15
+var s_restore_xnack_mask_lo =	xnack_mask_lo
+var s_restore_xnack_mask_hi =	xnack_mask_hi
+var s_restore_buf_rsrc0	    =	ttmp8
+var s_restore_buf_rsrc1	    =	ttmp9
+var s_restore_buf_rsrc2	    =	ttmp10
+var s_restore_buf_rsrc3	    =	ttmp11
+var s_restore_ttmps_lo	    =	s_restore_tmp		//no conflict
+var s_restore_ttmps_hi	    =	s_restore_alloc_size	//no conflict
+
+/**************************************************************************/
+/*			trap handler entry points			  */
+/**************************************************************************/
+/* Shader Main*/
+
+shader main
+  asic(DEFAULT)
+  type(CS)
+
+
+	s_branch L_SKIP_RESTORE					    //NOT restore. might be a regular trap or save
+
+L_JUMP_TO_RESTORE:
+    s_branch L_RESTORE						    //restore
+
+L_SKIP_RESTORE:
+
+    s_getreg_b32    s_save_status, hwreg(HW_REG_STATUS)				    //save STATUS since we will change SCC
+
+    // Clear SPI_PRIO: do not save with elevated priority.
+    // Clear ECC_ERR: prevents SQC store and triggers FATAL_HALT if setreg'd.
+    s_andn2_b32     s_save_status, s_save_status, SQ_WAVE_STATUS_SPI_PRIO_MASK|SQ_WAVE_STATUS_ECC_ERR_MASK
+
+    s_getreg_b32    s_save_trapsts, hwreg(HW_REG_TRAPSTS)
+
+    s_and_b32       ttmp2, s_save_status, SQ_WAVE_STATUS_HALT_MASK
+    s_cbranch_scc0  L_NOT_HALTED
+
+L_HALTED:
+    // Host trap may occur while wave is halted.
+    s_and_b32       ttmp2, s_save_pc_hi, S_SAVE_PC_HI_TRAP_ID_MASK
+    s_cbranch_scc1  L_FETCH_2ND_TRAP
+
+L_CHECK_SAVE:
+    s_and_b32       ttmp2, s_save_trapsts, SQ_WAVE_TRAPSTS_SAVECTX_MASK    //check whether this is for save
+    s_cbranch_scc1  L_SAVE					//this is the operation for save
+
+    // Wave is halted but neither host trap nor SAVECTX is raised.
+    // Caused by instruction fetch memory violation.
+    // Spin wait until context saved to prevent interrupt storm.
+    s_sleep         0x10
+    s_getreg_b32    s_save_trapsts, hwreg(HW_REG_TRAPSTS)
+    s_branch        L_CHECK_SAVE
+
+L_NOT_HALTED:
+    // Let second-level handle non-SAVECTX exception or trap.
+    // Any concurrent SAVECTX will be handled upon re-entry once halted.
+
+    // Check non-maskable exceptions. memory_violation, illegal_instruction
+    // and debugger (host trap, wave start/end, trap after instruction)
+    // exceptions always cause the wave to enter the trap handler.
+    s_and_b32       ttmp2, s_save_trapsts,      \
+        SQ_WAVE_TRAPSTS_MEM_VIOL_MASK         | \
+        SQ_WAVE_TRAPSTS_ILLEGAL_INST_MASK     | \
+        SQ_WAVE_TRAPSTS_HOST_TRAP_MASK        | \
+        SQ_WAVE_TRAPSTS_WAVE_BEGIN_MASK       | \
+        SQ_WAVE_TRAPSTS_WAVE_END_MASK         | \
+        SQ_WAVE_TRAPSTS_TRAP_AFTER_INST_MASK
+    s_cbranch_scc1  L_FETCH_2ND_TRAP
+
+    // Check for maskable exceptions in trapsts.excp and trapsts.excp_hi.
+    // Maskable exceptions only cause the wave to enter the trap handler if
+    // their respective bit in mode.excp_en is set.
+    s_and_b32       ttmp2, s_save_trapsts, SQ_WAVE_TRAPSTS_EXCP_MASK|SQ_WAVE_TRAPSTS_EXCP_HI_MASK
+    s_cbranch_scc0  L_CHECK_TRAP_ID
+
+    s_and_b32       ttmp3, s_save_trapsts, SQ_WAVE_TRAPSTS_ADDR_WATCH_MASK|SQ_WAVE_TRAPSTS_EXCP_HI_MASK
+    s_cbranch_scc0  L_NOT_ADDR_WATCH
+    s_bitset1_b32   ttmp2, SQ_WAVE_TRAPSTS_ADDR_WATCH_SHIFT // Check all addr_watch[123] exceptions against excp_en.addr_watch
+
+L_NOT_ADDR_WATCH:
+    s_getreg_b32    ttmp3, hwreg(HW_REG_MODE)
+    s_lshl_b32      ttmp2, ttmp2, SQ_WAVE_MODE_EXCP_EN_SHIFT
+    s_and_b32       ttmp2, ttmp2, ttmp3
+    s_cbranch_scc1  L_FETCH_2ND_TRAP
+
+L_CHECK_TRAP_ID:
+    // Check trap_id != 0
+    s_and_b32       ttmp2, s_save_pc_hi, S_SAVE_PC_HI_TRAP_ID_MASK
+    s_cbranch_scc1  L_FETCH_2ND_TRAP
+
+if SINGLE_STEP_MISSED_WORKAROUND
+    // Prioritize single step exception over context save.
+    // Second-level trap will halt wave and RFE, re-entering for SAVECTX.
+    s_getreg_b32    ttmp2, hwreg(HW_REG_MODE)
+    s_and_b32       ttmp2, ttmp2, SQ_WAVE_MODE_DEBUG_EN_MASK
+    s_cbranch_scc1  L_FETCH_2ND_TRAP
+end
+
+    s_and_b32       ttmp2, s_save_trapsts, SQ_WAVE_TRAPSTS_SAVECTX_MASK
+    s_cbranch_scc1  L_SAVE
+
+L_FETCH_2ND_TRAP:
+    // Preserve and clear scalar XNACK state before issuing scalar reads.
+    save_and_clear_ib_sts(ttmp14)
+
+    // Read second-level TBA/TMA from first-level TMA and jump if available.
+    // ttmp[2:5] and ttmp12 can be used (others hold SPI-initialized debug data)
+    // ttmp12 holds SQ_WAVE_STATUS
+    s_getreg_b32    ttmp14, hwreg(HW_REG_SQ_SHADER_TMA_LO)
+    s_getreg_b32    ttmp15, hwreg(HW_REG_SQ_SHADER_TMA_HI)
+    s_lshl_b64      [ttmp14, ttmp15], [ttmp14, ttmp15], 0x8
+
+    s_bitcmp1_b32   ttmp15, 0xF
+    s_cbranch_scc0  L_NO_SIGN_EXTEND_TMA
+    s_or_b32        ttmp15, ttmp15, 0xFFFF0000
+L_NO_SIGN_EXTEND_TMA:
+
+    s_load_dword    ttmp2, [ttmp14, ttmp15], 0x10 glc:1 // debug trap enabled flag
+    s_waitcnt       lgkmcnt(0)
+    s_lshl_b32      ttmp2, ttmp2, TTMP_DEBUG_TRAP_ENABLED_SHIFT
+    s_andn2_b32     s_save_ib_sts, s_save_ib_sts, TTMP_DEBUG_TRAP_ENABLED_MASK
+    s_or_b32        s_save_ib_sts, s_save_ib_sts, ttmp2
+
+    s_load_dwordx2  [ttmp2, ttmp3], [ttmp14, ttmp15], 0x0 glc:1 // second-level TBA
+    s_waitcnt       lgkmcnt(0)
+    s_load_dwordx2  [ttmp14, ttmp15], [ttmp14, ttmp15], 0x8 glc:1 // second-level TMA
+    s_waitcnt       lgkmcnt(0)
+
+    s_and_b64       [ttmp2, ttmp3], [ttmp2, ttmp3], [ttmp2, ttmp3]
+    s_cbranch_scc0  L_NO_NEXT_TRAP // second-level trap handler not been set
+    s_setpc_b64     [ttmp2, ttmp3] // jump to second-level trap handler
+
+L_NO_NEXT_TRAP:
+    // If not caused by trap then halt wave to prevent re-entry.
+    s_and_b32       ttmp2, s_save_pc_hi, (S_SAVE_PC_HI_TRAP_ID_MASK|S_SAVE_PC_HI_HT_MASK)
+    s_cbranch_scc1  L_TRAP_CASE
+    s_or_b32        s_save_status, s_save_status, SQ_WAVE_STATUS_HALT_MASK
+
+    // If the PC points to S_ENDPGM then context save will fail if STATUS.HALT is set.
+    // Rewind the PC to prevent this from occurring.
+    s_sub_u32       ttmp0, ttmp0, 0x8
+    s_subb_u32      ttmp1, ttmp1, 0x0
+
+    s_branch        L_EXIT_TRAP
+
+L_TRAP_CASE:
+    // Host trap will not cause trap re-entry.
+    s_and_b32       ttmp2, s_save_pc_hi, S_SAVE_PC_HI_HT_MASK
+    s_cbranch_scc1  L_EXIT_TRAP
+
+    // Advance past trap instruction to prevent re-entry.
+    s_add_u32       ttmp0, ttmp0, 0x4
+    s_addc_u32      ttmp1, ttmp1, 0x0
+
+L_EXIT_TRAP:
+    s_and_b32	ttmp1, ttmp1, 0xFFFF
+
+    restore_ib_sts(ttmp14)
+
+    // Restore SQ_WAVE_STATUS.
+    s_and_b64       exec, exec, exec // Restore STATUS.EXECZ, not writable by s_setreg_b32
+    s_and_b64       vcc, vcc, vcc    // Restore STATUS.VCCZ, not writable by s_setreg_b32
+    set_status_without_spi_prio(s_save_status, ttmp2)
+
+    s_rfe_b64       [ttmp0, ttmp1]
+
+    // *********	End handling of non-CWSR traps	 *******************
+
+/**************************************************************************/
+/*			save routine					  */
+/**************************************************************************/
+
+L_SAVE:
+    s_and_b32	    s_save_pc_hi, s_save_pc_hi, 0x0000ffff    //pc[47:32]
+
+    s_mov_b32	    s_save_tmp, 0							    //clear saveCtx bit
+    s_setreg_b32    hwreg(HW_REG_TRAPSTS, SQ_WAVE_TRAPSTS_SAVECTX_SHIFT, 1), s_save_tmp	    //clear saveCtx bit
+
+    save_and_clear_ib_sts(s_save_tmp)
+
+    /*	    inform SPI the readiness and wait for SPI's go signal */
+    s_mov_b32	    s_save_exec_lo, exec_lo						    //save EXEC and use EXEC for the go signal from SPI
+    s_mov_b32	    s_save_exec_hi, exec_hi
+    s_mov_b64	    exec,   0x0								    //clear EXEC to get ready to receive
+
+	s_sendmsg   sendmsg(MSG_SAVEWAVE)  //send SPI a message and wait for SPI's write to EXEC
+
+    // Set SPI_PRIO=2 to avoid starving instruction fetch in the waves we're waiting for.
+    s_or_b32 s_save_tmp, s_save_status, (2 << SQ_WAVE_STATUS_SPI_PRIO_SHIFT)
+    s_setreg_b32 hwreg(HW_REG_STATUS), s_save_tmp
+
+  L_SLEEP:
+    s_sleep 0x2		       // sleep 1 (64clk) is not enough for 8 waves per SIMD, which will cause SQ hang, since the 7,8th wave could not get arbit to exec inst, while other waves are stuck into the sleep-loop and waiting for wrexec!=0
+
+	s_cbranch_execz L_SLEEP
+
+    // Save trap temporaries 4-11, 13 initialized by SPI debug dispatch logic
+    // ttmp SR memory offset : size(VGPR)+size(SGPR)+0x40
+    get_vgpr_size_bytes(s_save_ttmps_lo)
+    get_sgpr_size_bytes(s_save_ttmps_hi)
+    s_add_u32	    s_save_ttmps_lo, s_save_ttmps_lo, s_save_ttmps_hi
+    s_add_u32	    s_save_ttmps_lo, s_save_ttmps_lo, s_save_spi_init_lo
+    s_addc_u32	    s_save_ttmps_hi, s_save_spi_init_hi, 0x0
+    s_and_b32	    s_save_ttmps_hi, s_save_ttmps_hi, 0xFFFF
+    s_store_dwordx4 [ttmp4, ttmp5, ttmp6, ttmp7], [s_save_ttmps_lo, s_save_ttmps_hi], 0x50 glc:1
+    ack_sqc_store_workaround()
+    s_store_dwordx4 [ttmp8, ttmp9, ttmp10, ttmp11], [s_save_ttmps_lo, s_save_ttmps_hi], 0x60 glc:1
+    ack_sqc_store_workaround()
+    s_store_dword   ttmp13, [s_save_ttmps_lo, s_save_ttmps_hi], 0x74 glc:1
+    ack_sqc_store_workaround()
+
+    /*	    setup Resource Contants    */
+    s_mov_b32	    s_save_buf_rsrc0,	s_save_spi_init_lo							//base_addr_lo
+    s_and_b32	    s_save_buf_rsrc1,	s_save_spi_init_hi, 0x0000FFFF						//base_addr_hi
+    s_or_b32	    s_save_buf_rsrc1,	s_save_buf_rsrc1,  S_SAVE_BUF_RSRC_WORD1_STRIDE
+    s_mov_b32	    s_save_buf_rsrc2,	0									//NUM_RECORDS initial value = 0 (in bytes) although not neccessarily inited
+    s_mov_b32	    s_save_buf_rsrc3,	S_SAVE_BUF_RSRC_WORD3_MISC
+
+    //FIXME  right now s_save_m0/s_save_mem_offset use tma_lo/tma_hi  (might need to save them before using them?)
+    s_mov_b32	    s_save_m0,		m0								    //save M0
+
+    /*	    global mem offset		*/
+    s_mov_b32	    s_save_mem_offset,	0x0									//mem offset initial value = 0
+
+
+
+
+    /*	    save HW registers	*/
+    //////////////////////////////
+
+  L_SAVE_HWREG:
+	// HWREG SR memory offset : size(VGPR)+size(SGPR)
+       get_vgpr_size_bytes(s_save_mem_offset)
+       get_sgpr_size_bytes(s_save_tmp)
+       s_add_u32 s_save_mem_offset, s_save_mem_offset, s_save_tmp
+
+
+    s_mov_b32	    s_save_buf_rsrc2, 0x4				//NUM_RECORDS	in bytes
+	s_mov_b32	s_save_buf_rsrc2,  0x1000000				    //NUM_RECORDS in bytes
+
+
+    write_hwreg_to_mem(s_save_m0, s_save_buf_rsrc0, s_save_mem_offset)			//M0
+    write_hwreg_to_mem(s_save_pc_lo, s_save_buf_rsrc0, s_save_mem_offset)		    //PC
+    write_hwreg_to_mem(s_save_pc_hi, s_save_buf_rsrc0, s_save_mem_offset)
+    write_hwreg_to_mem(s_save_exec_lo, s_save_buf_rsrc0, s_save_mem_offset)		//EXEC
+    write_hwreg_to_mem(s_save_exec_hi, s_save_buf_rsrc0, s_save_mem_offset)
+    write_hwreg_to_mem(s_save_status, s_save_buf_rsrc0, s_save_mem_offset)		//STATUS
+
+    //s_save_trapsts conflicts with s_save_alloc_size
+    s_getreg_b32    s_save_trapsts, hwreg(HW_REG_TRAPSTS)
+    write_hwreg_to_mem(s_save_trapsts, s_save_buf_rsrc0, s_save_mem_offset)		//TRAPSTS
+
+    write_hwreg_to_mem(xnack_mask_lo, s_save_buf_rsrc0, s_save_mem_offset)	    //XNACK_MASK_LO
+    write_hwreg_to_mem(xnack_mask_hi, s_save_buf_rsrc0, s_save_mem_offset)	    //XNACK_MASK_HI
+
+    //use s_save_tmp would introduce conflict here between s_save_tmp and s_save_buf_rsrc2
+    s_getreg_b32    s_save_m0, hwreg(HW_REG_MODE)						    //MODE
+    write_hwreg_to_mem(s_save_m0, s_save_buf_rsrc0, s_save_mem_offset)
+
+    // Clear VSKIP state now that MODE.VSKIP has been saved.
+    // If user shader set it then vector instructions would be skipped.
+    s_setvskip	0,0
+
+    /*	    the first wave in the threadgroup	 */
+    s_and_b32	    s_save_tmp, s_save_spi_init_hi, S_SAVE_SPI_INIT_FIRST_WAVE_MASK	// extract fisrt wave bit
+    s_mov_b32	     s_save_exec_hi, 0x0
+    s_or_b32	     s_save_exec_hi, s_save_tmp, s_save_exec_hi				 // save first wave bit to s_save_exec_hi.bits[26]
+
+
+    /*		save SGPRs	*/
+	// Save SGPR before LDS save, then the s0 to s4 can be used during LDS save...
+    //////////////////////////////
+
+    // SGPR SR memory offset : size(VGPR)
+    get_vgpr_size_bytes(s_save_mem_offset)
+    // TODO, change RSRC word to rearrange memory layout for SGPRS
+
+    s_getreg_b32    s_save_alloc_size, hwreg(HW_REG_GPR_ALLOC,SQ_WAVE_GPR_ALLOC_SGPR_SIZE_SHIFT,SQ_WAVE_GPR_ALLOC_SGPR_SIZE_SIZE)		//spgr_size
+    s_add_u32	    s_save_alloc_size, s_save_alloc_size, 1
+    s_lshl_b32	    s_save_alloc_size, s_save_alloc_size, 4			    //Number of SGPRs = (sgpr_size + 1) * 16   (non-zero value)
+
+	s_lshl_b32	s_save_buf_rsrc2,   s_save_alloc_size, 2		    //NUM_RECORDS in bytes
+
+	s_mov_b32	s_save_buf_rsrc2,  0x1000000				    //NUM_RECORDS in bytes
+
+
+    // backup s_save_buf_rsrc0,1 to s_save_pc_lo/hi, since write_16sgpr_to_mem function will change the rsrc0
+    //s_mov_b64 s_save_pc_lo, s_save_buf_rsrc0
+    s_mov_b64 s_save_xnack_mask_lo, s_save_buf_rsrc0
+    s_add_u32 s_save_buf_rsrc0, s_save_buf_rsrc0, s_save_mem_offset
+    s_addc_u32 s_save_buf_rsrc1, s_save_buf_rsrc1, 0
+
+    s_mov_b32	    m0, 0x0			    //SGPR initial index value =0
+    s_nop	    0x0				    //Manually inserted wait states
+  L_SAVE_SGPR_LOOP:
+    // SGPR is allocated in 16 SGPR granularity
+    s_movrels_b64   s0, s0     //s0 = s[0+m0], s1 = s[1+m0]
+    s_movrels_b64   s2, s2     //s2 = s[2+m0], s3 = s[3+m0]
+    s_movrels_b64   s4, s4     //s4 = s[4+m0], s5 = s[5+m0]
+    s_movrels_b64   s6, s6     //s6 = s[6+m0], s7 = s[7+m0]
+    s_movrels_b64   s8, s8     //s8 = s[8+m0], s9 = s[9+m0]
+    s_movrels_b64   s10, s10   //s10 = s[10+m0], s11 = s[11+m0]
+    s_movrels_b64   s12, s12   //s12 = s[12+m0], s13 = s[13+m0]
+    s_movrels_b64   s14, s14   //s14 = s[14+m0], s15 = s[15+m0]
+
+    write_16sgpr_to_mem(s0, s_save_buf_rsrc0, s_save_mem_offset) //PV: the best performance should be using s_buffer_store_dwordx4
+    s_add_u32	    m0, m0, 16							    //next sgpr index
+    s_cmp_lt_u32    m0, s_save_alloc_size					    //scc = (m0 < s_save_alloc_size) ? 1 : 0
+    s_cbranch_scc1  L_SAVE_SGPR_LOOP					//SGPR save is complete?
+    // restore s_save_buf_rsrc0,1
+    //s_mov_b64 s_save_buf_rsrc0, s_save_pc_lo
+    s_mov_b64 s_save_buf_rsrc0, s_save_xnack_mask_lo
+
+
+
+
+    /*		save first 4 VGPR, then LDS save could use   */
+	// each wave will alloc 4 vgprs at least...
+    /////////////////////////////////////////////////////////////////////////////////////
+
+    s_mov_b32	    s_save_mem_offset, 0
+    s_mov_b32	    exec_lo, 0xFFFFFFFF						    //need every thread from now on
+    s_mov_b32	    exec_hi, 0xFFFFFFFF
+    s_mov_b32	    xnack_mask_lo, 0x0
+    s_mov_b32	    xnack_mask_hi, 0x0
+
+	s_mov_b32	s_save_buf_rsrc2,  0x1000000				    //NUM_RECORDS in bytes
+
+
+    // VGPR Allocated in 4-GPR granularity
+
+if SAVE_AFTER_XNACK_ERROR
+	check_if_tcp_store_ok()
+	s_cbranch_scc1 L_SAVE_FIRST_VGPRS_WITH_TCP
+
+	write_vgprs_to_mem_with_sqc(v0, 4, s_save_buf_rsrc0, s_save_mem_offset)
+	s_branch L_SAVE_LDS
+
+L_SAVE_FIRST_VGPRS_WITH_TCP:
+end
+
+    write_4vgprs_to_mem(s_save_buf_rsrc0, s_save_mem_offset)
+
+    /*		save LDS	*/
+    //////////////////////////////
+
+  L_SAVE_LDS:
+
+	// Change EXEC to all threads...
+    s_mov_b32	    exec_lo, 0xFFFFFFFF	  //need every thread from now on
+    s_mov_b32	    exec_hi, 0xFFFFFFFF
+
+    s_getreg_b32    s_save_alloc_size, hwreg(HW_REG_LDS_ALLOC,SQ_WAVE_LDS_ALLOC_LDS_SIZE_SHIFT,SQ_WAVE_LDS_ALLOC_LDS_SIZE_SIZE)		    //lds_size
+    s_and_b32	    s_save_alloc_size, s_save_alloc_size, 0xFFFFFFFF		    //lds_size is zero?
+    s_cbranch_scc0  L_SAVE_LDS_DONE									       //no lds used? jump to L_SAVE_DONE
+
+    s_barrier		    //LDS is used? wait for other waves in the same TG
+    s_and_b32	    s_save_tmp, s_save_exec_hi, S_SAVE_SPI_INIT_FIRST_WAVE_MASK		       //exec is still used here
+    s_cbranch_scc0  L_SAVE_LDS_DONE
+
+	// first wave do LDS save;
+
+    s_lshl_b32	    s_save_alloc_size, s_save_alloc_size, 6			    //LDS size in dwords = lds_size * 64dw
+    s_lshl_b32	    s_save_alloc_size, s_save_alloc_size, 2			    //LDS size in bytes
+    s_mov_b32	    s_save_buf_rsrc2,  s_save_alloc_size			    //NUM_RECORDS in bytes
+
+    // LDS at offset: size(VGPR)+SIZE(SGPR)+SIZE(HWREG)
+    //
+    get_vgpr_size_bytes(s_save_mem_offset)
+    get_sgpr_size_bytes(s_save_tmp)
+    s_add_u32  s_save_mem_offset, s_save_mem_offset, s_save_tmp
+    s_add_u32 s_save_mem_offset, s_save_mem_offset, get_hwreg_size_bytes()
+
+
+	s_mov_b32	s_save_buf_rsrc2,  0x1000000		      //NUM_RECORDS in bytes
+
+    s_mov_b32	    m0, 0x0						  //lds_offset initial value = 0
+
+
+      v_mbcnt_lo_u32_b32 v2, 0xffffffff, 0x0
+      v_mbcnt_hi_u32_b32 v3, 0xffffffff, v2	// tid
+
+if SAVE_AFTER_XNACK_ERROR
+	check_if_tcp_store_ok()
+	s_cbranch_scc1 L_SAVE_LDS_WITH_TCP
+
+	v_lshlrev_b32 v2, 2, v3
+L_SAVE_LDS_LOOP_SQC:
+#if ASIC_FAMILY < CHIP_GC_9_5_0
+	ds_read2_b32 v[0:1], v2 offset0:0 offset1:0x40
+	s_waitcnt lgkmcnt(0)
+	write_vgprs_to_mem_with_sqc(v0, 2, s_save_buf_rsrc0, s_save_mem_offset)
+
+	v_add_u32 v2, 0x200, v2
+#else
+	// gfx950 needs to save in multiple of 256 bytes.
+	ds_read_b32 v0, v2
+	s_waitcnt lgkmcnt(0)
+	write_vgprs_to_mem_with_sqc(v0, 1, s_save_buf_rsrc0, s_save_mem_offset)
+
+	v_add_u32 v2, 0x100, v2
+#endif
+
+	v_cmp_lt_u32 vcc[0:1], v2, s_save_alloc_size
+	s_cbranch_vccnz L_SAVE_LDS_LOOP_SQC
+
+	s_branch L_SAVE_LDS_DONE
+
+L_SAVE_LDS_WITH_TCP:
+end
+
+      v_mul_i32_i24 v2, v3, 8	// tid*8
+      v_mov_b32 v3, 256*2
+      s_mov_b32 m0, 0x10000
+      s_mov_b32 s0, s_save_buf_rsrc3
+      s_and_b32 s_save_buf_rsrc3, s_save_buf_rsrc3, 0xFF7FFFFF	  // disable add_tid
+      s_or_b32 s_save_buf_rsrc3, s_save_buf_rsrc3, 0x58000   //DFMT
+
+L_SAVE_LDS_LOOP_VECTOR:
+      ds_read_b64 v[0:1], v2	//x =LDS[a], byte address
+      s_waitcnt lgkmcnt(0)
+      buffer_store_dwordx2  v[0:1], v2, s_save_buf_rsrc0, s_save_mem_offset VMEM_MODIFIERS offen:1
+//	s_waitcnt vmcnt(0)
+//	v_add_u32 v2, vcc[0:1], v2, v3
+      v_add_u32 v2, v2, v3
+      v_cmp_lt_u32 vcc[0:1], v2, s_save_alloc_size
+#if ASIC_FAMILY >= CHIP_GC_9_5_0
+      s_mov_b64 exec, vcc
+#endif
+      s_cbranch_vccnz L_SAVE_LDS_LOOP_VECTOR
+
+      // restore rsrc3
+      s_mov_b32 s_save_buf_rsrc3, s0
+
+L_SAVE_LDS_DONE:
+
+
+    /*		save VGPRs  - set the Rest VGPRs	*/
+    //////////////////////////////////////////////////////////////////////////////////////
+  L_SAVE_VGPR:
+    // VGPR SR memory offset: 0
+    // TODO rearrange the RSRC words to use swizzle for VGPR save...
+
+    s_mov_b32	    s_save_mem_offset, (0+256*4)				    // for the rest VGPRs
+    s_mov_b32	    exec_lo, 0xFFFFFFFF						    //need every thread from now on
+    s_mov_b32	    exec_hi, 0xFFFFFFFF
+
+    get_num_arch_vgprs(s_save_alloc_size)
+    s_mov_b32	    s_save_buf_rsrc2,  0x1000000				    //NUM_RECORDS in bytes
+
+
+    // VGPR store using dw burst
+    s_mov_b32	      m0, 0x4	//VGPR initial index value =0
+    s_cmp_lt_u32      m0, s_save_alloc_size
+    s_cbranch_scc0    L_SAVE_VGPR_END
+
+
+    s_set_gpr_idx_on	m0, 0x1 //M0[7:0] = M0[7:0] and M0[15:12] = 0x1
+    s_add_u32	    s_save_alloc_size, s_save_alloc_size, 0x1000		    //add 0x1000 since we compare m0 against it later
+
+if SAVE_AFTER_XNACK_ERROR
+	check_if_tcp_store_ok()
+	s_cbranch_scc1 L_SAVE_VGPR_LOOP
+
+L_SAVE_VGPR_LOOP_SQC:
+	write_vgprs_to_mem_with_sqc(v0, 4, s_save_buf_rsrc0, s_save_mem_offset)
+
+	s_add_u32 m0, m0, 4
+	s_cmp_lt_u32 m0, s_save_alloc_size
+	s_cbranch_scc1 L_SAVE_VGPR_LOOP_SQC
+
+	s_set_gpr_idx_off
+	s_branch L_SAVE_VGPR_END
+end
+
+  L_SAVE_VGPR_LOOP:
+    v_mov_b32	    v0, v0		//v0 = v[0+m0]
+    v_mov_b32	    v1, v1		//v0 = v[0+m0]
+    v_mov_b32	    v2, v2		//v0 = v[0+m0]
+    v_mov_b32	    v3, v3		//v0 = v[0+m0]
+
+    write_4vgprs_to_mem(s_save_buf_rsrc0, s_save_mem_offset)
+
+    s_add_u32	    m0, m0, 4							    //next vgpr index
+    s_add_u32	    s_save_mem_offset, s_save_mem_offset, 256*4			    //every buffer_store_dword does 256 bytes
+    s_cmp_lt_u32    m0, s_save_alloc_size					    //scc = (m0 < s_save_alloc_size) ? 1 : 0
+    s_cbranch_scc1  L_SAVE_VGPR_LOOP						    //VGPR save is complete?
+    s_set_gpr_idx_off
+
+L_SAVE_VGPR_END:
+
+#if ASIC_FAMILY >= CHIP_ARCTURUS
+    // Save ACC VGPRs
+
+#if ASIC_FAMILY >= CHIP_ALDEBARAN
+    // ACC VGPR count may differ from ARCH VGPR count.
+    get_num_acc_vgprs(s_save_alloc_size, s_save_tmp)
+    s_and_b32       s_save_alloc_size, s_save_alloc_size, s_save_alloc_size
+    s_cbranch_scc0  L_SAVE_ACCVGPR_END
+    s_add_u32	    s_save_alloc_size, s_save_alloc_size, 0x1000		    //add 0x1000 since we compare m0 against it later
+#endif
+
+    s_mov_b32 m0, 0x0 //VGPR initial index value =0
+    s_set_gpr_idx_on m0, 0x1 //M0[7:0] = M0[7:0] and M0[15:12] = 0x1
+
+if SAVE_AFTER_XNACK_ERROR
+    check_if_tcp_store_ok()
+    s_cbranch_scc1 L_SAVE_ACCVGPR_LOOP
+
+L_SAVE_ACCVGPR_LOOP_SQC:
+    for var vgpr = 0; vgpr < 4; ++ vgpr
+        v_accvgpr_read v[vgpr], acc[vgpr]  // v[N] = acc[N+m0]
+    end
+
+    write_vgprs_to_mem_with_sqc(v0, 4, s_save_buf_rsrc0, s_save_mem_offset)
+
+    s_add_u32 m0, m0, 4
+    s_cmp_lt_u32 m0, s_save_alloc_size
+    s_cbranch_scc1 L_SAVE_ACCVGPR_LOOP_SQC
+
+    s_set_gpr_idx_off
+    s_branch L_SAVE_ACCVGPR_END
+end
+
+L_SAVE_ACCVGPR_LOOP:
+    for var vgpr = 0; vgpr < 4; ++ vgpr
+        v_accvgpr_read v[vgpr], acc[vgpr]  // v[N] = acc[N+m0]
+    end
+
+    write_4vgprs_to_mem(s_save_buf_rsrc0, s_save_mem_offset)
+
+    s_add_u32 m0, m0, 4
+    s_add_u32 s_save_mem_offset, s_save_mem_offset, 256*4
+    s_cmp_lt_u32 m0, s_save_alloc_size
+    s_cbranch_scc1 L_SAVE_ACCVGPR_LOOP
+    s_set_gpr_idx_off
+
+L_SAVE_ACCVGPR_END:
+#endif
+
+    s_branch	L_END_PGM
+
+
+
+/**************************************************************************/
+/*			restore routine					  */
+/**************************************************************************/
+
+L_RESTORE:
+    /*	    Setup Resource Contants    */
+    s_mov_b32	    s_restore_buf_rsrc0,    s_restore_spi_init_lo							    //base_addr_lo
+    s_and_b32	    s_restore_buf_rsrc1,    s_restore_spi_init_hi, 0x0000FFFF						    //base_addr_hi
+    s_or_b32	    s_restore_buf_rsrc1,    s_restore_buf_rsrc1,  S_RESTORE_BUF_RSRC_WORD1_STRIDE
+    s_mov_b32	    s_restore_buf_rsrc2,    0										    //NUM_RECORDS initial value = 0 (in bytes)
+    s_mov_b32	    s_restore_buf_rsrc3,    S_RESTORE_BUF_RSRC_WORD3_MISC
+
+    /*	    global mem offset		*/
+//  s_mov_b32	    s_restore_mem_offset, 0x0				    //mem offset initial value = 0
+
+    /*	    the first wave in the threadgroup	 */
+    s_and_b32	    s_restore_tmp, s_restore_spi_init_hi, S_RESTORE_SPI_INIT_FIRST_WAVE_MASK
+    s_cbranch_scc0  L_RESTORE_VGPR
+
+    /*		restore LDS	*/
+    //////////////////////////////
+  L_RESTORE_LDS:
+
+    s_mov_b32	    exec_lo, 0xFFFFFFFF							    //need every thread from now on   //be consistent with SAVE although can be moved ahead
+    s_mov_b32	    exec_hi, 0xFFFFFFFF
+
+    s_getreg_b32    s_restore_alloc_size, hwreg(HW_REG_LDS_ALLOC,SQ_WAVE_LDS_ALLOC_LDS_SIZE_SHIFT,SQ_WAVE_LDS_ALLOC_LDS_SIZE_SIZE)		//lds_size
+    s_and_b32	    s_restore_alloc_size, s_restore_alloc_size, 0xFFFFFFFF		    //lds_size is zero?
+    s_cbranch_scc0  L_RESTORE_VGPR							    //no lds used? jump to L_RESTORE_VGPR
+    s_lshl_b32	    s_restore_alloc_size, s_restore_alloc_size, 6			    //LDS size in dwords = lds_size * 64dw
+    s_lshl_b32	    s_restore_alloc_size, s_restore_alloc_size, 2			    //LDS size in bytes
+    s_mov_b32	    s_restore_buf_rsrc2,    s_restore_alloc_size			    //NUM_RECORDS in bytes
+
+    // LDS at offset: size(VGPR)+SIZE(SGPR)+SIZE(HWREG)
+    //
+    get_vgpr_size_bytes(s_restore_mem_offset)
+    get_sgpr_size_bytes(s_restore_tmp)
+    s_add_u32  s_restore_mem_offset, s_restore_mem_offset, s_restore_tmp
+    s_add_u32  s_restore_mem_offset, s_restore_mem_offset, get_hwreg_size_bytes()	     //FIXME, Check if offset overflow???
+
+
+	s_mov_b32	s_restore_buf_rsrc2,  0x1000000					    //NUM_RECORDS in bytes
+    s_mov_b32	    m0, 0x0								    //lds_offset initial value = 0
+
+  L_RESTORE_LDS_LOOP:
+	buffer_load_dword   v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset lds:1		       // first 64DW
+	buffer_load_dword   v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset lds:1 offset:256	       // second 64DW
+#if ASIC_FAMILY >= CHIP_GC_9_5_0
+	buffer_load_dword   v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset lds:1 offset:512	// third 64DW
+	buffer_load_dword   v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset lds:1 offset:768	// forth 64DW
+	buffer_load_dword   v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset lds:1 offset:1024	// fifth 64DW
+#endif
+    s_add_u32	    m0, m0, LDS_RESTORE_GRANULARITY_BYTES					// 128/320 DW
+    s_add_u32	    s_restore_mem_offset, s_restore_mem_offset, LDS_RESTORE_GRANULARITY_BYTES	//mem offset increased by 128/320 DW
+    s_cmp_lt_u32    m0, s_restore_alloc_size					//scc=(m0 < s_restore_alloc_size) ? 1 : 0
+    s_cbranch_scc1  L_RESTORE_LDS_LOOP							    //LDS restore is complete?
+
+
+    /*		restore VGPRs	    */
+    //////////////////////////////
+  L_RESTORE_VGPR:
+    s_mov_b32	    exec_lo, 0xFFFFFFFF							    //need every thread from now on   //be consistent with SAVE although can be moved ahead
+    s_mov_b32	    exec_hi, 0xFFFFFFFF
+    s_mov_b32	    s_restore_buf_rsrc2,  0x1000000					    //NUM_RECORDS in bytes
+
+    // Save ARCH VGPRs 4-N, then all ACC VGPRs, then ARCH VGPRs 0-3.
+    get_num_arch_vgprs(s_restore_alloc_size)
+    s_add_u32	    s_restore_alloc_size, s_restore_alloc_size, 0x8000			    //add 0x8000 since we compare m0 against it later
+
+    // ARCH VGPRs at offset: 0
+    s_mov_b32	    s_restore_mem_offset, 0x0
+    s_mov_b32	    s_restore_mem_offset_save, s_restore_mem_offset	// restore start with v1, v0 will be the last
+    s_add_u32	    s_restore_mem_offset, s_restore_mem_offset, 256*4
+    s_mov_b32	    m0, 4				//VGPR initial index value = 1
+    s_set_gpr_idx_on	m0, 0x8								    //M0[7:0] = M0[7:0] and M0[15:12] = 0x8
+
+  L_RESTORE_VGPR_LOOP:
+    read_4vgprs_from_mem(s_restore_buf_rsrc0, s_restore_mem_offset)
+    v_mov_b32	    v0, v0								    //v[0+m0] = v0
+    v_mov_b32	    v1, v1
+    v_mov_b32	    v2, v2
+    v_mov_b32	    v3, v3
+    s_add_u32	    m0, m0, 4								    //next vgpr index
+    s_add_u32	    s_restore_mem_offset, s_restore_mem_offset, 256*4				//every buffer_load_dword does 256 bytes
+    s_cmp_lt_u32    m0, s_restore_alloc_size						    //scc = (m0 < s_restore_alloc_size) ? 1 : 0
+    s_cbranch_scc1  L_RESTORE_VGPR_LOOP							    //VGPR restore (except v0) is complete?
+
+#if ASIC_FAMILY >= CHIP_ALDEBARAN
+    // ACC VGPR count may differ from ARCH VGPR count.
+    get_num_acc_vgprs(s_restore_alloc_size, s_restore_tmp2)
+    s_and_b32       s_restore_alloc_size, s_restore_alloc_size, s_restore_alloc_size
+    s_cbranch_scc0  L_RESTORE_ACCVGPR_END
+    s_add_u32	    s_restore_alloc_size, s_restore_alloc_size, 0x8000			    //add 0x8000 since we compare m0 against it later
+#endif
+
+#if ASIC_FAMILY >= CHIP_ARCTURUS
+    // ACC VGPRs at offset: size(ARCH VGPRs)
+    s_mov_b32	    m0, 0
+    s_set_gpr_idx_on	m0, 0x8								    //M0[7:0] = M0[7:0] and M0[15:12] = 0x8
+
+  L_RESTORE_ACCVGPR_LOOP:
+    read_4vgprs_from_mem(s_restore_buf_rsrc0, s_restore_mem_offset)
+
+    for var vgpr = 0; vgpr < 4; ++ vgpr
+        v_accvgpr_write acc[vgpr], v[vgpr]
+    end
+
+    s_add_u32	    m0, m0, 4								    //next vgpr index
+    s_add_u32	    s_restore_mem_offset, s_restore_mem_offset, 256*4			    //every buffer_load_dword does 256 bytes
+    s_cmp_lt_u32    m0, s_restore_alloc_size						    //scc = (m0 < s_restore_alloc_size) ? 1 : 0
+    s_cbranch_scc1  L_RESTORE_ACCVGPR_LOOP						    //VGPR restore (except v0) is complete?
+  L_RESTORE_ACCVGPR_END:
+#endif
+
+    s_set_gpr_idx_off
+
+    // Restore VGPRs 0-3 last, no longer needed.
+    read_4vgprs_from_mem(s_restore_buf_rsrc0, s_restore_mem_offset_save)
+
+    /*		restore SGPRs	    */
+    //////////////////////////////
+
+    // SGPR SR memory offset : size(VGPR)
+    get_vgpr_size_bytes(s_restore_mem_offset)
+    get_sgpr_size_bytes(s_restore_tmp)
+    s_add_u32 s_restore_mem_offset, s_restore_mem_offset, s_restore_tmp
+    s_sub_u32 s_restore_mem_offset, s_restore_mem_offset, 16*4	   // restore SGPR from S[n] to S[0], by 16 sgprs group
+    // TODO, change RSRC word to rearrange memory layout for SGPRS
+
+    s_getreg_b32    s_restore_alloc_size, hwreg(HW_REG_GPR_ALLOC,SQ_WAVE_GPR_ALLOC_SGPR_SIZE_SHIFT,SQ_WAVE_GPR_ALLOC_SGPR_SIZE_SIZE)		    //spgr_size
+    s_add_u32	    s_restore_alloc_size, s_restore_alloc_size, 1
+    s_lshl_b32	    s_restore_alloc_size, s_restore_alloc_size, 4			    //Number of SGPRs = (sgpr_size + 1) * 16   (non-zero value)
+
+	s_lshl_b32	s_restore_buf_rsrc2,	s_restore_alloc_size, 2			    //NUM_RECORDS in bytes
+	s_mov_b32	s_restore_buf_rsrc2,  0x1000000					    //NUM_RECORDS in bytes
+
+    s_mov_b32 m0, s_restore_alloc_size
+
+ L_RESTORE_SGPR_LOOP:
+    read_16sgpr_from_mem(s0, s_restore_buf_rsrc0, s_restore_mem_offset)	 //PV: further performance improvement can be made
+    s_waitcnt	    lgkmcnt(0)								    //ensure data ready
+
+    s_sub_u32 m0, m0, 16    // Restore from S[n] to S[0]
+    s_nop 0 // hazard SALU M0=> S_MOVREL
+
+    s_movreld_b64   s0, s0	//s[0+m0] = s0
+    s_movreld_b64   s2, s2
+    s_movreld_b64   s4, s4
+    s_movreld_b64   s6, s6
+    s_movreld_b64   s8, s8
+    s_movreld_b64   s10, s10
+    s_movreld_b64   s12, s12
+    s_movreld_b64   s14, s14
+
+    s_cmp_eq_u32    m0, 0		//scc = (m0 < s_restore_alloc_size) ? 1 : 0
+    s_cbranch_scc0  L_RESTORE_SGPR_LOOP		    //SGPR restore (except s0) is complete?
+
+    /*	    restore HW registers    */
+    //////////////////////////////
+  L_RESTORE_HWREG:
+
+
+    // HWREG SR memory offset : size(VGPR)+size(SGPR)
+    get_vgpr_size_bytes(s_restore_mem_offset)
+    get_sgpr_size_bytes(s_restore_tmp)
+    s_add_u32 s_restore_mem_offset, s_restore_mem_offset, s_restore_tmp
+
+
+    s_mov_b32	    s_restore_buf_rsrc2, 0x4						    //NUM_RECORDS   in bytes
+	s_mov_b32	s_restore_buf_rsrc2,  0x1000000					    //NUM_RECORDS in bytes
+
+    read_hwreg_from_mem(s_restore_m0, s_restore_buf_rsrc0, s_restore_mem_offset)		    //M0
+    read_hwreg_from_mem(s_restore_pc_lo, s_restore_buf_rsrc0, s_restore_mem_offset)		//PC
+    read_hwreg_from_mem(s_restore_pc_hi, s_restore_buf_rsrc0, s_restore_mem_offset)
+    read_hwreg_from_mem(s_restore_exec_lo, s_restore_buf_rsrc0, s_restore_mem_offset)		    //EXEC
+    read_hwreg_from_mem(s_restore_exec_hi, s_restore_buf_rsrc0, s_restore_mem_offset)
+    read_hwreg_from_mem(s_restore_status, s_restore_buf_rsrc0, s_restore_mem_offset)		    //STATUS
+    read_hwreg_from_mem(s_restore_trapsts, s_restore_buf_rsrc0, s_restore_mem_offset)		    //TRAPSTS
+    read_hwreg_from_mem(xnack_mask_lo, s_restore_buf_rsrc0, s_restore_mem_offset)		    //XNACK_MASK_LO
+    read_hwreg_from_mem(xnack_mask_hi, s_restore_buf_rsrc0, s_restore_mem_offset)		    //XNACK_MASK_HI
+    read_hwreg_from_mem(s_restore_mode, s_restore_buf_rsrc0, s_restore_mem_offset)		//MODE
+
+    s_waitcnt	    lgkmcnt(0)											    //from now on, it is safe to restore STATUS and IB_STS
+
+    s_mov_b32	    m0,		s_restore_m0
+    s_mov_b32	    exec_lo,	s_restore_exec_lo
+    s_mov_b32	    exec_hi,	s_restore_exec_hi
+
+    s_and_b32	    s_restore_m0, SQ_WAVE_TRAPSTS_PRE_SAVECTX_MASK, s_restore_trapsts
+    s_setreg_b32    hwreg(HW_REG_TRAPSTS, SQ_WAVE_TRAPSTS_PRE_SAVECTX_SHIFT, SQ_WAVE_TRAPSTS_PRE_SAVECTX_SIZE), s_restore_m0
+    s_and_b32	    s_restore_m0, SQ_WAVE_TRAPSTS_POST_SAVECTX_MASK, s_restore_trapsts
+    s_lshr_b32	    s_restore_m0, s_restore_m0, SQ_WAVE_TRAPSTS_POST_SAVECTX_SHIFT
+    s_setreg_b32    hwreg(HW_REG_TRAPSTS, SQ_WAVE_TRAPSTS_POST_SAVECTX_SHIFT, SQ_WAVE_TRAPSTS_POST_SAVECTX_SIZE), s_restore_m0
+    //s_setreg_b32  hwreg(HW_REG_TRAPSTS),  s_restore_trapsts	   //don't overwrite SAVECTX bit as it may be set through external SAVECTX during restore
+    s_setreg_b32    hwreg(HW_REG_MODE),	    s_restore_mode
+
+    // Restore trap temporaries 4-11, 13 initialized by SPI debug dispatch logic
+    // ttmp SR memory offset : size(VGPR)+size(SGPR)+0x40
+    get_vgpr_size_bytes(s_restore_ttmps_lo)
+    get_sgpr_size_bytes(s_restore_ttmps_hi)
+    s_add_u32	    s_restore_ttmps_lo, s_restore_ttmps_lo, s_restore_ttmps_hi
+    s_add_u32	    s_restore_ttmps_lo, s_restore_ttmps_lo, s_restore_buf_rsrc0
+    s_addc_u32	    s_restore_ttmps_hi, s_restore_buf_rsrc1, 0x0
+    s_and_b32	    s_restore_ttmps_hi, s_restore_ttmps_hi, 0xFFFF
+    s_load_dwordx4  [ttmp4, ttmp5, ttmp6, ttmp7], [s_restore_ttmps_lo, s_restore_ttmps_hi], 0x50 glc:1
+    s_load_dwordx4  [ttmp8, ttmp9, ttmp10, ttmp11], [s_restore_ttmps_lo, s_restore_ttmps_hi], 0x60 glc:1
+    s_load_dword    ttmp13, [s_restore_ttmps_lo, s_restore_ttmps_hi], 0x74 glc:1
+    s_waitcnt	    lgkmcnt(0)
+
+    restore_ib_sts(s_restore_tmp)
+
+    s_and_b32 s_restore_pc_hi, s_restore_pc_hi, 0x0000ffff	//pc[47:32]	   //Do it here in order not to affect STATUS
+    s_and_b64	 exec, exec, exec  // Restore STATUS.EXECZ, not writable by s_setreg_b32
+    s_and_b64	 vcc, vcc, vcc	// Restore STATUS.VCCZ, not writable by s_setreg_b32
+    set_status_without_spi_prio(s_restore_status, s_restore_tmp) // SCC is included, which is changed by previous salu
+
+    s_barrier							//barrier to ensure the readiness of LDS before access attempts from any other wave in the same TG //FIXME not performance-optimal at this time
+
+    s_rfe_b64 s_restore_pc_lo					//Return to the main shader program and resume execution
+
+
+/**************************************************************************/
+/*			the END						  */
+/**************************************************************************/
+L_END_PGM:
+    s_endpgm_saved
+
+end
+
+
+/**************************************************************************/
+/*			the helper functions				  */
+/**************************************************************************/
+
+//Only for save hwreg to mem
+function write_hwreg_to_mem(s, s_rsrc, s_mem_offset)
+	s_mov_b32 exec_lo, m0			//assuming exec_lo is not needed anymore from this point on
+	s_mov_b32 m0, s_mem_offset
+	s_buffer_store_dword s, s_rsrc, m0	glc:1
+	ack_sqc_store_workaround()
+	s_add_u32	s_mem_offset, s_mem_offset, 4
+	s_mov_b32   m0, exec_lo
+end
+
+
+// HWREG are saved before SGPRs, so all HWREG could be use.
+function write_16sgpr_to_mem(s, s_rsrc, s_mem_offset)
+
+	s_buffer_store_dwordx4 s[0], s_rsrc, 0	glc:1
+	ack_sqc_store_workaround()
+	s_buffer_store_dwordx4 s[4], s_rsrc, 16	 glc:1
+	ack_sqc_store_workaround()
+	s_buffer_store_dwordx4 s[8], s_rsrc, 32	 glc:1
+	ack_sqc_store_workaround()
+	s_buffer_store_dwordx4 s[12], s_rsrc, 48 glc:1
+	ack_sqc_store_workaround()
+	s_add_u32	s_rsrc[0], s_rsrc[0], 4*16
+	s_addc_u32	s_rsrc[1], s_rsrc[1], 0x0	      // +scc
+end
+
+
+function read_hwreg_from_mem(s, s_rsrc, s_mem_offset)
+    s_buffer_load_dword s, s_rsrc, s_mem_offset	    glc:1
+    s_add_u32	    s_mem_offset, s_mem_offset, 4
+end
+
+function read_16sgpr_from_mem(s, s_rsrc, s_mem_offset)
+    s_buffer_load_dwordx16 s, s_rsrc, s_mem_offset	glc:1
+    s_sub_u32	    s_mem_offset, s_mem_offset, 4*16
+end
+
+function check_if_tcp_store_ok
+	// If STATUS.ALLOW_REPLAY=0 and TRAPSTS.XNACK_ERROR=1 then TCP stores will fail.
+	s_and_b32 s_save_tmp, s_save_status, SQ_WAVE_STATUS_ALLOW_REPLAY_MASK
+	s_cbranch_scc1 L_TCP_STORE_CHECK_DONE
+
+	s_getreg_b32 s_save_tmp, hwreg(HW_REG_TRAPSTS)
+	s_andn2_b32 s_save_tmp, SQ_WAVE_TRAPSTS_XNACK_ERROR_MASK, s_save_tmp
+
+L_TCP_STORE_CHECK_DONE:
+end
+
+function write_4vgprs_to_mem(s_rsrc, s_mem_offset)
+	buffer_store_dword v0, v0, s_rsrc, s_mem_offset VMEM_MODIFIERS
+	buffer_store_dword v1, v0, s_rsrc, s_mem_offset VMEM_MODIFIERS offset:256
+	buffer_store_dword v2, v0, s_rsrc, s_mem_offset VMEM_MODIFIERS offset:256*2
+	buffer_store_dword v3, v0, s_rsrc, s_mem_offset VMEM_MODIFIERS offset:256*3
+end
+
+function read_4vgprs_from_mem(s_rsrc, s_mem_offset)
+	buffer_load_dword v0, v0, s_rsrc, s_mem_offset VMEM_MODIFIERS
+	buffer_load_dword v1, v0, s_rsrc, s_mem_offset VMEM_MODIFIERS offset:256
+	buffer_load_dword v2, v0, s_rsrc, s_mem_offset VMEM_MODIFIERS offset:256*2
+	buffer_load_dword v3, v0, s_rsrc, s_mem_offset VMEM_MODIFIERS offset:256*3
+	s_waitcnt vmcnt(0)
+end
+
+function write_vgpr_to_mem_with_sqc(v, s_rsrc, s_mem_offset)
+	s_mov_b32 s4, 0
+
+L_WRITE_VGPR_LANE_LOOP:
+	for var lane = 0; lane < 4; ++ lane
+		v_readlane_b32 s[lane], v, s4
+		s_add_u32 s4, s4, 1
+	end
+
+	s_buffer_store_dwordx4 s[0:3], s_rsrc, s_mem_offset glc:1
+	ack_sqc_store_workaround()
+
+	s_add_u32 s_mem_offset, s_mem_offset, 0x10
+	s_cmp_eq_u32 s4, 0x40
+	s_cbranch_scc0 L_WRITE_VGPR_LANE_LOOP
+end
+
+function write_vgprs_to_mem_with_sqc(v, n_vgprs, s_rsrc, s_mem_offset)
+	for var vgpr = 0; vgpr < n_vgprs; ++ vgpr
+		write_vgpr_to_mem_with_sqc(v[vgpr], s_rsrc, s_mem_offset)
+	end
+end
+
+function get_lds_size_bytes(s_lds_size_byte)
+    // SQ LDS granularity is 64DW, while PGM_RSRC2.lds_size is in granularity 128DW
+    s_getreg_b32   s_lds_size_byte, hwreg(HW_REG_LDS_ALLOC, SQ_WAVE_LDS_ALLOC_LDS_SIZE_SHIFT, SQ_WAVE_LDS_ALLOC_LDS_SIZE_SIZE)		// lds_size
+    s_lshl_b32	   s_lds_size_byte, s_lds_size_byte, 8			    //LDS size in dwords = lds_size * 64 *4Bytes    // granularity 64DW
+end
+
+function get_vgpr_size_bytes(s_vgpr_size_byte)
+    s_getreg_b32   s_vgpr_size_byte, hwreg(HW_REG_GPR_ALLOC,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SIZE)	 //vpgr_size
+    s_add_u32	   s_vgpr_size_byte, s_vgpr_size_byte, 1
+    s_lshl_b32	   s_vgpr_size_byte, s_vgpr_size_byte, (2+8) //Number of VGPRs = (vgpr_size + 1) * 4 * 64 * 4	(non-zero value)   //FIXME for GFX, zero is possible
+
+#if ASIC_FAMILY >= CHIP_ARCTURUS
+    s_lshl_b32     s_vgpr_size_byte, s_vgpr_size_byte, 1  // Double size for ACC VGPRs
+#endif
+end
+
+function get_sgpr_size_bytes(s_sgpr_size_byte)
+    s_getreg_b32   s_sgpr_size_byte, hwreg(HW_REG_GPR_ALLOC,SQ_WAVE_GPR_ALLOC_SGPR_SIZE_SHIFT,SQ_WAVE_GPR_ALLOC_SGPR_SIZE_SIZE)	 //spgr_size
+    s_add_u32	   s_sgpr_size_byte, s_sgpr_size_byte, 1
+    s_lshl_b32	   s_sgpr_size_byte, s_sgpr_size_byte, 6 //Number of SGPRs = (sgpr_size + 1) * 16 *4   (non-zero value)
+end
+
+function get_hwreg_size_bytes
+    return 128 //HWREG size 128 bytes
+end
+
+function get_num_arch_vgprs(s_num_arch_vgprs)
+#if ASIC_FAMILY >= CHIP_ALDEBARAN
+    // VGPR count includes ACC VGPRs, use ACC VGPR offset for ARCH VGPR count.
+    s_getreg_b32    s_num_arch_vgprs, hwreg(HW_REG_GPR_ALLOC,SQ_WAVE_GPR_ALLOC_ACCV_OFFSET_SHIFT,SQ_WAVE_GPR_ALLOC_ACCV_OFFSET_SIZE)
+#else
+    s_getreg_b32    s_num_arch_vgprs, hwreg(HW_REG_GPR_ALLOC,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SIZE)
+#endif
+
+    // Number of VGPRs = (vgpr_size + 1) * 4
+    s_add_u32	    s_num_arch_vgprs, s_num_arch_vgprs, 1
+    s_lshl_b32	    s_num_arch_vgprs, s_num_arch_vgprs, 2
+end
+
+#if ASIC_FAMILY >= CHIP_ALDEBARAN
+function get_num_acc_vgprs(s_num_acc_vgprs, s_tmp)
+    // VGPR count = (GPR_ALLOC.VGPR_SIZE + 1) * 8
+    s_getreg_b32    s_num_acc_vgprs, hwreg(HW_REG_GPR_ALLOC,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SIZE)
+    s_add_u32	    s_num_acc_vgprs, s_num_acc_vgprs, 1
+    s_lshl_b32	    s_num_acc_vgprs, s_num_acc_vgprs, 3
+
+    // ACC VGPR count = VGPR count - ARCH VGPR count.
+    get_num_arch_vgprs(s_tmp)
+    s_sub_u32	    s_num_acc_vgprs, s_num_acc_vgprs, s_tmp
+end
+#endif
+
+function ack_sqc_store_workaround
+    if ACK_SQC_STORE
+        s_waitcnt lgkmcnt(0)
+    end
+end
+
+function set_status_without_spi_prio(status, tmp)
+    // Do not restore STATUS.SPI_PRIO since scheduler may have raised it.
+    s_lshr_b32      tmp, status, SQ_WAVE_STATUS_POST_SPI_PRIO_SHIFT
+    s_setreg_b32    hwreg(HW_REG_STATUS, SQ_WAVE_STATUS_POST_SPI_PRIO_SHIFT, SQ_WAVE_STATUS_POST_SPI_PRIO_SIZE), tmp
+    s_nop           0x2 // avoid S_SETREG => S_SETREG hazard
+    s_setreg_b32    hwreg(HW_REG_STATUS, SQ_WAVE_STATUS_PRE_SPI_PRIO_SHIFT, SQ_WAVE_STATUS_PRE_SPI_PRIO_SIZE), status
+end
+
+function save_and_clear_ib_sts(tmp)
+    // Save IB_STS.FIRST_REPLAY[15] and IB_STS.RCNT[20:16] into unused space s_save_ib_sts[31:26].
+    s_getreg_b32    tmp, hwreg(HW_REG_IB_STS)
+    s_and_b32       tmp, tmp, SQ_WAVE_IB_STS_RCNT_FIRST_REPLAY_MASK
+    s_lshl_b32      tmp, tmp, (TTMP_SAVE_RCNT_FIRST_REPLAY_SHIFT - SQ_WAVE_IB_STS_FIRST_REPLAY_SHIFT)
+    s_andn2_b32     s_save_ib_sts, s_save_ib_sts, TTMP_SAVE_RCNT_FIRST_REPLAY_MASK
+    s_or_b32        s_save_ib_sts, s_save_ib_sts, tmp
+    s_setreg_imm32_b32 hwreg(HW_REG_IB_STS), 0x0
+end
+
+function restore_ib_sts(tmp)
+    s_lshr_b32      tmp, s_save_ib_sts, (TTMP_SAVE_RCNT_FIRST_REPLAY_SHIFT - SQ_WAVE_IB_STS_FIRST_REPLAY_SHIFT)
+    s_and_b32       tmp, tmp, SQ_WAVE_IB_STS_RCNT_FIRST_REPLAY_MASK
+    s_setreg_b32    hwreg(HW_REG_IB_STS), tmp
+end
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
index e4a8c2e52cb2..0f0719528bcc 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
@@ -1,5 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -21,9 +22,9 @@
  */
 
 #include <linux/device.h>
-#include <linux/export.h>
 #include <linux/err.h>
 #include <linux/fs.h>
+#include <linux/file.h>
 #include <linux/sched.h>
 #include <linux/slab.h>
 #include <linux/uaccess.h>
@@ -32,13 +33,20 @@
 #include <linux/time.h>
 #include <linux/mm.h>
 #include <linux/mman.h>
-#include <asm/processor.h>
+#include <linux/ptrace.h>
+#include <linux/dma-buf.h>
+#include <linux/processor.h>
 #include "kfd_priv.h"
 #include "kfd_device_queue_manager.h"
-#include "kfd_dbgmgr.h"
+#include "kfd_svm.h"
+#include "amdgpu_amdkfd.h"
+#include "kfd_smi_events.h"
+#include "amdgpu_dma_buf.h"
+#include "kfd_debug.h"
 
 static long kfd_ioctl(struct file *, unsigned int, unsigned long);
 static int kfd_open(struct inode *, struct file *);
+static int kfd_release(struct inode *, struct file *);
 static int kfd_mmap(struct file *, struct vm_area_struct *);
 
 static const char kfd_dev_name[] = "kfd";
@@ -46,14 +54,36 @@ static const char kfd_dev_name[] = "kfd";
 static const struct file_operations kfd_fops = {
 	.owner = THIS_MODULE,
 	.unlocked_ioctl = kfd_ioctl,
-	.compat_ioctl = kfd_ioctl,
+	.compat_ioctl = compat_ptr_ioctl,
 	.open = kfd_open,
+	.release = kfd_release,
 	.mmap = kfd_mmap,
 };
 
 static int kfd_char_dev_major = -1;
-static struct class *kfd_class;
 struct device *kfd_device;
+static const struct class kfd_class = {
+	.name = kfd_dev_name,
+};
+
+static inline struct kfd_process_device *kfd_lock_pdd_by_id(struct kfd_process *p, __u32 gpu_id)
+{
+	struct kfd_process_device *pdd;
+
+	mutex_lock(&p->mutex);
+	pdd = kfd_process_device_data_by_id(p, gpu_id);
+
+	if (pdd)
+		return pdd;
+
+	mutex_unlock(&p->mutex);
+	return NULL;
+}
+
+static inline void kfd_unlock_pdd(struct kfd_process_device *pdd)
+{
+	mutex_unlock(&pdd->process->mutex);
+}
 
 int kfd_chardev_init(void)
 {
@@ -64,14 +94,13 @@ int kfd_chardev_init(void)
 	if (err < 0)
 		goto err_register_chrdev;
 
-	kfd_class = class_create(THIS_MODULE, kfd_dev_name);
-	err = PTR_ERR(kfd_class);
-	if (IS_ERR(kfd_class))
+	err = class_register(&kfd_class);
+	if (err)
 		goto err_class_create;
 
-	kfd_device = device_create(kfd_class, NULL,
-					MKDEV(kfd_char_dev_major, 0),
-					NULL, kfd_dev_name);
+	kfd_device = device_create(&kfd_class, NULL,
+				   MKDEV(kfd_char_dev_major, 0),
+				   NULL, kfd_dev_name);
 	err = PTR_ERR(kfd_device);
 	if (IS_ERR(kfd_device))
 		goto err_device_create;
@@ -79,7 +108,7 @@ int kfd_chardev_init(void)
 	return 0;
 
 err_device_create:
-	class_destroy(kfd_class);
+	class_unregister(&kfd_class);
 err_class_create:
 	unregister_chrdev(kfd_char_dev_major, kfd_dev_name);
 err_register_chrdev:
@@ -88,14 +117,10 @@ err_register_chrdev:
 
 void kfd_chardev_exit(void)
 {
-	device_destroy(kfd_class, MKDEV(kfd_char_dev_major, 0));
-	class_destroy(kfd_class);
+	device_destroy(&kfd_class, MKDEV(kfd_char_dev_major, 0));
+	class_unregister(&kfd_class);
 	unregister_chrdev(kfd_char_dev_major, kfd_dev_name);
-}
-
-struct device *kfd_chardev(void)
-{
-	return kfd_device;
+	kfd_device = NULL;
 }
 
 
@@ -121,8 +146,26 @@ static int kfd_open(struct inode *inode, struct file *filep)
 	if (IS_ERR(process))
 		return PTR_ERR(process);
 
-	dev_dbg(kfd_device, "process %d opened, compat mode (32 bit) - %d\n",
-		process->pasid, process->is_32bit_user_mode);
+	if (kfd_process_init_cwsr_apu(process, filep)) {
+		kfd_unref_process(process);
+		return -EFAULT;
+	}
+
+	/* filep now owns the reference returned by kfd_create_process */
+	filep->private_data = process;
+
+	dev_dbg(kfd_device, "process pid %d opened kfd node, compat mode (32 bit) - %d\n",
+		process->lead_thread->pid, process->is_32bit_user_mode);
+
+	return 0;
+}
+
+static int kfd_release(struct inode *inode, struct file *filep)
+{
+	struct kfd_process *process = filep->private_data;
+
+	if (process)
+		kfd_unref_process(process);
 
 	return 0;
 }
@@ -141,7 +184,12 @@ static int kfd_ioctl_get_version(struct file *filep, struct kfd_process *p,
 static int set_queue_properties_from_user(struct queue_properties *q_properties,
 				struct kfd_ioctl_create_queue_args *args)
 {
-	if (args->queue_percentage > KFD_MAX_QUEUE_PERCENTAGE) {
+	/*
+	 * Repurpose queue percentage to accommodate new features:
+	 * bit 0-7: queue percentage
+	 * bit 8-15: pm4_target_xcc
+	 */
+	if ((args->queue_percentage & 0xFF) > KFD_MAX_QUEUE_PERCENTAGE) {
 		pr_err("Queue percentage must be between 0 to KFD_MAX_QUEUE_PERCENTAGE\n");
 		return -EINVAL;
 	}
@@ -152,8 +200,7 @@ static int set_queue_properties_from_user(struct queue_properties *q_properties,
 	}
 
 	if ((args->ring_base_address) &&
-		(!access_ok(VERIFY_WRITE,
-			(const void __user *) args->ring_base_address,
+		(!access_ok((const void __user *) args->ring_base_address,
 			sizeof(uint64_t)))) {
 		pr_err("Can't access ring base address\n");
 		return -EFAULT;
@@ -164,53 +211,63 @@ static int set_queue_properties_from_user(struct queue_properties *q_properties,
 		return -EINVAL;
 	}
 
-	if (!access_ok(VERIFY_WRITE,
-			(const void __user *) args->read_pointer_address,
+	if (args->ring_size < KFD_MIN_QUEUE_RING_SIZE) {
+		args->ring_size = KFD_MIN_QUEUE_RING_SIZE;
+		pr_debug("Size lower. clamped to KFD_MIN_QUEUE_RING_SIZE");
+	}
+
+	if (!access_ok((const void __user *) args->read_pointer_address,
 			sizeof(uint32_t))) {
 		pr_err("Can't access read pointer\n");
 		return -EFAULT;
 	}
 
-	if (!access_ok(VERIFY_WRITE,
-			(const void __user *) args->write_pointer_address,
+	if (!access_ok((const void __user *) args->write_pointer_address,
 			sizeof(uint32_t))) {
 		pr_err("Can't access write pointer\n");
 		return -EFAULT;
 	}
 
 	if (args->eop_buffer_address &&
-		!access_ok(VERIFY_WRITE,
-			(const void __user *) args->eop_buffer_address,
+		!access_ok((const void __user *) args->eop_buffer_address,
 			sizeof(uint32_t))) {
 		pr_debug("Can't access eop buffer");
 		return -EFAULT;
 	}
 
 	if (args->ctx_save_restore_address &&
-		!access_ok(VERIFY_WRITE,
-			(const void __user *) args->ctx_save_restore_address,
+		!access_ok((const void __user *) args->ctx_save_restore_address,
 			sizeof(uint32_t))) {
 		pr_debug("Can't access ctx save restore buffer");
 		return -EFAULT;
 	}
 
 	q_properties->is_interop = false;
-	q_properties->queue_percent = args->queue_percentage;
+	q_properties->is_gws = false;
+	q_properties->queue_percent = args->queue_percentage & 0xFF;
+	/* bit 8-15 are repurposed to be PM4 target XCC */
+	q_properties->pm4_target_xcc = (args->queue_percentage >> 8) & 0xFF;
 	q_properties->priority = args->queue_priority;
 	q_properties->queue_address = args->ring_base_address;
 	q_properties->queue_size = args->ring_size;
-	q_properties->read_ptr = (uint32_t *) args->read_pointer_address;
-	q_properties->write_ptr = (uint32_t *) args->write_pointer_address;
+	q_properties->read_ptr = (void __user *)args->read_pointer_address;
+	q_properties->write_ptr = (void __user *)args->write_pointer_address;
 	q_properties->eop_ring_buffer_address = args->eop_buffer_address;
 	q_properties->eop_ring_buffer_size = args->eop_buffer_size;
 	q_properties->ctx_save_restore_area_address =
 			args->ctx_save_restore_address;
 	q_properties->ctx_save_restore_area_size = args->ctx_save_restore_size;
+	q_properties->ctl_stack_size = args->ctl_stack_size;
+	q_properties->sdma_engine_id = args->sdma_engine_id;
 	if (args->queue_type == KFD_IOC_QUEUE_TYPE_COMPUTE ||
 		args->queue_type == KFD_IOC_QUEUE_TYPE_COMPUTE_AQL)
 		q_properties->type = KFD_QUEUE_TYPE_COMPUTE;
 	else if (args->queue_type == KFD_IOC_QUEUE_TYPE_SDMA)
 		q_properties->type = KFD_QUEUE_TYPE_SDMA;
+	else if (args->queue_type == KFD_IOC_QUEUE_TYPE_SDMA_XGMI)
+		q_properties->type = KFD_QUEUE_TYPE_SDMA_XGMI;
+	else if (args->queue_type == KFD_IOC_QUEUE_TYPE_SDMA_BY_ENG_ID)
+		q_properties->type = KFD_QUEUE_TYPE_SDMA_BY_ENG_ID;
 	else
 		return -ENOTSUPP;
 
@@ -231,7 +288,7 @@ static int set_queue_properties_from_user(struct queue_properties *q_properties,
 	pr_debug("Queue Size: 0x%llX, %u\n",
 			q_properties->queue_size, args->ring_size);
 
-	pr_debug("Queue r/w Pointers: %p, %p\n",
+	pr_debug("Queue r/w Pointers: %px, %px\n",
 			q_properties->read_ptr,
 			q_properties->write_ptr);
 
@@ -249,11 +306,12 @@ static int kfd_ioctl_create_queue(struct file *filep, struct kfd_process *p,
 					void *data)
 {
 	struct kfd_ioctl_create_queue_args *args = data;
-	struct kfd_dev *dev;
+	struct kfd_node *dev;
 	int err = 0;
 	unsigned int queue_id;
 	struct kfd_process_device *pdd;
 	struct queue_properties q_properties;
+	uint32_t doorbell_offset_in_process = 0;
 
 	memset(&q_properties, 0, sizeof(struct queue_properties));
 
@@ -264,26 +322,55 @@ static int kfd_ioctl_create_queue(struct file *filep, struct kfd_process *p,
 		return err;
 
 	pr_debug("Looking for gpu id 0x%x\n", args->gpu_id);
-	dev = kfd_device_by_id(args->gpu_id);
-	if (!dev) {
-		pr_debug("Could not find gpu id 0x%x\n", args->gpu_id);
-		return -EINVAL;
-	}
 
 	mutex_lock(&p->mutex);
 
+	pdd = kfd_process_device_data_by_id(p, args->gpu_id);
+	if (!pdd) {
+		pr_debug("Could not find gpu id 0x%x\n", args->gpu_id);
+		err = -EINVAL;
+		goto err_pdd;
+	}
+	dev = pdd->dev;
+
 	pdd = kfd_bind_process_to_device(dev, p);
 	if (IS_ERR(pdd)) {
 		err = -ESRCH;
 		goto err_bind_process;
 	}
 
-	pr_debug("Creating queue for PASID %d on gpu 0x%x\n",
-			p->pasid,
+	if (q_properties.type == KFD_QUEUE_TYPE_SDMA_BY_ENG_ID) {
+		int max_sdma_eng_id = kfd_get_num_sdma_engines(dev) +
+				      kfd_get_num_xgmi_sdma_engines(dev) - 1;
+
+		if (q_properties.sdma_engine_id > max_sdma_eng_id) {
+			err = -EINVAL;
+			pr_err("sdma_engine_id %i exceeds maximum id of %i\n",
+			       q_properties.sdma_engine_id, max_sdma_eng_id);
+			goto err_sdma_engine_id;
+		}
+	}
+
+	if (!pdd->qpd.proc_doorbells) {
+		err = kfd_alloc_process_doorbells(dev->kfd, pdd);
+		if (err) {
+			pr_debug("failed to allocate process doorbells\n");
+			goto err_bind_process;
+		}
+	}
+
+	err = kfd_queue_acquire_buffers(pdd, &q_properties);
+	if (err) {
+		pr_debug("failed to acquire user queue buffers\n");
+		goto err_acquire_queue_buf;
+	}
+
+	pr_debug("Creating queue for process pid %d on gpu 0x%x\n",
+			p->lead_thread->pid,
 			dev->id);
 
-	err = pqm_create_queue(&p->pqm, dev, filep, &q_properties,
-				0, q_properties.type, &queue_id);
+	err = pqm_create_queue(&p->pqm, dev, &q_properties, &queue_id,
+			NULL, NULL, NULL, &doorbell_offset_in_process);
 	if (err != 0)
 		goto err_create_queue;
 
@@ -291,8 +378,13 @@ static int kfd_ioctl_create_queue(struct file *filep, struct kfd_process *p,
 
 
 	/* Return gpu_id as doorbell offset for mmap usage */
-	args->doorbell_offset = (KFD_MMAP_DOORBELL_MASK | args->gpu_id);
-	args->doorbell_offset <<= PAGE_SHIFT;
+	args->doorbell_offset = KFD_MMAP_TYPE_DOORBELL;
+	args->doorbell_offset |= KFD_MMAP_GPU_ID(args->gpu_id);
+	if (KFD_IS_SOC15(dev))
+		/* On SOC15 ASICs, include the doorbell offset within the
+		 * process doorbell frame, which is 2 pages.
+		 */
+		args->doorbell_offset |= doorbell_offset_in_process;
 
 	mutex_unlock(&p->mutex);
 
@@ -307,10 +399,16 @@ static int kfd_ioctl_create_queue(struct file *filep, struct kfd_process *p,
 	pr_debug("Write ptr address   == 0x%016llX\n",
 			args->write_pointer_address);
 
+	kfd_dbg_ev_raise(KFD_EC_MASK(EC_QUEUE_NEW), p, dev, queue_id, false, NULL, 0);
 	return 0;
 
 err_create_queue:
+	kfd_queue_unref_bo_vas(pdd, &q_properties);
+	kfd_queue_release_buffers(pdd, &q_properties);
+err_acquire_queue_buf:
+err_sdma_engine_id:
 err_bind_process:
+err_pdd:
 	mutex_unlock(&p->mutex);
 	return err;
 }
@@ -321,9 +419,9 @@ static int kfd_ioctl_destroy_queue(struct file *filp, struct kfd_process *p,
 	int retval;
 	struct kfd_ioctl_destroy_queue_args *args = data;
 
-	pr_debug("Destroying queue id %d for pasid %d\n",
+	pr_debug("Destroying queue id %d for process pid %d\n",
 				args->queue_id,
-				p->pasid);
+				p->lead_thread->pid);
 
 	mutex_lock(&p->mutex);
 
@@ -340,7 +438,12 @@ static int kfd_ioctl_update_queue(struct file *filp, struct kfd_process *p,
 	struct kfd_ioctl_update_queue_args *args = data;
 	struct queue_properties properties;
 
-	if (args->queue_percentage > KFD_MAX_QUEUE_PERCENTAGE) {
+	/*
+	 * Repurpose queue percentage to accommodate new features:
+	 * bit 0-7: queue percentage
+	 * bit 8-15: pm4_target_xcc
+	 */
+	if ((args->queue_percentage & 0xFF) > KFD_MAX_QUEUE_PERCENTAGE) {
 		pr_err("Queue percentage must be between 0 to KFD_MAX_QUEUE_PERCENTAGE\n");
 		return -EINVAL;
 	}
@@ -351,8 +454,7 @@ static int kfd_ioctl_update_queue(struct file *filp, struct kfd_process *p,
 	}
 
 	if ((args->ring_base_address) &&
-		(!access_ok(VERIFY_WRITE,
-			(const void __user *) args->ring_base_address,
+		(!access_ok((const void __user *) args->ring_base_address,
 			sizeof(uint64_t)))) {
 		pr_err("Can't access ring base address\n");
 		return -EFAULT;
@@ -363,28 +465,100 @@ static int kfd_ioctl_update_queue(struct file *filp, struct kfd_process *p,
 		return -EINVAL;
 	}
 
+	if (args->ring_size < KFD_MIN_QUEUE_RING_SIZE) {
+		args->ring_size = KFD_MIN_QUEUE_RING_SIZE;
+		pr_debug("Size lower. clamped to KFD_MIN_QUEUE_RING_SIZE");
+	}
+
 	properties.queue_address = args->ring_base_address;
 	properties.queue_size = args->ring_size;
-	properties.queue_percent = args->queue_percentage;
+	properties.queue_percent = args->queue_percentage & 0xFF;
+	/* bit 8-15 are repurposed to be PM4 target XCC */
+	properties.pm4_target_xcc = (args->queue_percentage >> 8) & 0xFF;
 	properties.priority = args->queue_priority;
 
-	pr_debug("Updating queue id %d for pasid %d\n",
-			args->queue_id, p->pasid);
+	pr_debug("Updating queue id %d for process pid %d\n",
+			args->queue_id, p->lead_thread->pid);
 
 	mutex_lock(&p->mutex);
 
-	retval = pqm_update_queue(&p->pqm, args->queue_id, &properties);
+	retval = pqm_update_queue_properties(&p->pqm, args->queue_id, &properties);
 
 	mutex_unlock(&p->mutex);
 
 	return retval;
 }
 
+static int kfd_ioctl_set_cu_mask(struct file *filp, struct kfd_process *p,
+					void *data)
+{
+	int retval;
+	const int max_num_cus = 1024;
+	struct kfd_ioctl_set_cu_mask_args *args = data;
+	struct mqd_update_info minfo = {0};
+	uint32_t __user *cu_mask_ptr = (uint32_t __user *)args->cu_mask_ptr;
+	size_t cu_mask_size = sizeof(uint32_t) * (args->num_cu_mask / 32);
+
+	if ((args->num_cu_mask % 32) != 0) {
+		pr_debug("num_cu_mask 0x%x must be a multiple of 32",
+				args->num_cu_mask);
+		return -EINVAL;
+	}
+
+	minfo.cu_mask.count = args->num_cu_mask;
+	if (minfo.cu_mask.count == 0) {
+		pr_debug("CU mask cannot be 0");
+		return -EINVAL;
+	}
+
+	/* To prevent an unreasonably large CU mask size, set an arbitrary
+	 * limit of max_num_cus bits.  We can then just drop any CU mask bits
+	 * past max_num_cus bits and just use the first max_num_cus bits.
+	 */
+	if (minfo.cu_mask.count > max_num_cus) {
+		pr_debug("CU mask cannot be greater than 1024 bits");
+		minfo.cu_mask.count = max_num_cus;
+		cu_mask_size = sizeof(uint32_t) * (max_num_cus/32);
+	}
+
+	minfo.cu_mask.ptr = memdup_user(cu_mask_ptr, cu_mask_size);
+	if (IS_ERR(minfo.cu_mask.ptr)) {
+		pr_debug("Could not copy CU mask from userspace");
+		return PTR_ERR(minfo.cu_mask.ptr);
+	}
+
+	mutex_lock(&p->mutex);
+
+	retval = pqm_update_mqd(&p->pqm, args->queue_id, &minfo);
+
+	mutex_unlock(&p->mutex);
+
+	kfree(minfo.cu_mask.ptr);
+	return retval;
+}
+
+static int kfd_ioctl_get_queue_wave_state(struct file *filep,
+					  struct kfd_process *p, void *data)
+{
+	struct kfd_ioctl_get_queue_wave_state_args *args = data;
+	int r;
+
+	mutex_lock(&p->mutex);
+
+	r = pqm_get_wave_state(&p->pqm, args->queue_id,
+			       (void __user *)args->ctl_stack_address,
+			       &args->ctl_stack_used_size,
+			       &args->save_area_used_size);
+
+	mutex_unlock(&p->mutex);
+
+	return r;
+}
+
 static int kfd_ioctl_set_memory_policy(struct file *filep,
 					struct kfd_process *p, void *data)
 {
 	struct kfd_ioctl_set_memory_policy_args *args = data;
-	struct kfd_dev *dev;
 	int err = 0;
 	struct kfd_process_device *pdd;
 	enum cache_policy default_policy, alternate_policy;
@@ -399,13 +573,15 @@ static int kfd_ioctl_set_memory_policy(struct file *filep,
 		return -EINVAL;
 	}
 
-	dev = kfd_device_by_id(args->gpu_id);
-	if (!dev)
-		return -EINVAL;
-
 	mutex_lock(&p->mutex);
+	pdd = kfd_process_device_data_by_id(p, args->gpu_id);
+	if (!pdd) {
+		pr_debug("Could not find gpu id 0x%x\n", args->gpu_id);
+		err = -EINVAL;
+		goto err_pdd;
+	}
 
-	pdd = kfd_bind_process_to_device(dev, p);
+	pdd = kfd_bind_process_to_device(pdd->dev, p);
 	if (IS_ERR(pdd)) {
 		err = -ESRCH;
 		goto out;
@@ -418,317 +594,96 @@ static int kfd_ioctl_set_memory_policy(struct file *filep,
 		(args->alternate_policy == KFD_IOC_CACHE_POLICY_COHERENT)
 		   ? cache_policy_coherent : cache_policy_noncoherent;
 
-	if (!dev->dqm->ops.set_cache_memory_policy(dev->dqm,
+	if (!pdd->dev->dqm->ops.set_cache_memory_policy(pdd->dev->dqm,
 				&pdd->qpd,
 				default_policy,
 				alternate_policy,
 				(void __user *)args->alternate_aperture_base,
-				args->alternate_aperture_size))
+				args->alternate_aperture_size,
+				args->misc_process_flag))
 		err = -EINVAL;
 
 out:
+err_pdd:
 	mutex_unlock(&p->mutex);
 
 	return err;
 }
 
-static int kfd_ioctl_dbg_register(struct file *filep,
-				struct kfd_process *p, void *data)
+static int kfd_ioctl_set_trap_handler(struct file *filep,
+					struct kfd_process *p, void *data)
 {
-	struct kfd_ioctl_dbg_register_args *args = data;
-	struct kfd_dev *dev;
-	struct kfd_dbgmgr *dbgmgr_ptr;
+	struct kfd_ioctl_set_trap_handler_args *args = data;
+	int err = 0;
 	struct kfd_process_device *pdd;
-	bool create_ok;
-	long status = 0;
 
-	dev = kfd_device_by_id(args->gpu_id);
-	if (!dev)
-		return -EINVAL;
+	mutex_lock(&p->mutex);
 
-	if (dev->device_info->asic_family == CHIP_CARRIZO) {
-		pr_debug("kfd_ioctl_dbg_register not supported on CZ\n");
-		return -EINVAL;
+	pdd = kfd_process_device_data_by_id(p, args->gpu_id);
+	if (!pdd) {
+		err = -EINVAL;
+		goto err_pdd;
 	}
 
-	mutex_lock(kfd_get_dbgmgr_mutex());
-	mutex_lock(&p->mutex);
-
-	/*
-	 * make sure that we have pdd, if this the first queue created for
-	 * this process
-	 */
-	pdd = kfd_bind_process_to_device(dev, p);
+	pdd = kfd_bind_process_to_device(pdd->dev, p);
 	if (IS_ERR(pdd)) {
-		status = PTR_ERR(pdd);
+		err = -ESRCH;
 		goto out;
 	}
 
-	if (!dev->dbgmgr) {
-		/* In case of a legal call, we have no dbgmgr yet */
-		create_ok = kfd_dbgmgr_create(&dbgmgr_ptr, dev);
-		if (create_ok) {
-			status = kfd_dbgmgr_register(dbgmgr_ptr, p);
-			if (status != 0)
-				kfd_dbgmgr_destroy(dbgmgr_ptr);
-			else
-				dev->dbgmgr = dbgmgr_ptr;
-		}
-	} else {
-		pr_debug("debugger already registered\n");
-		status = -EINVAL;
-	}
+	kfd_process_set_trap_handler(&pdd->qpd, args->tba_addr, args->tma_addr);
 
 out:
+err_pdd:
 	mutex_unlock(&p->mutex);
-	mutex_unlock(kfd_get_dbgmgr_mutex());
 
-	return status;
+	return err;
 }
 
-static int kfd_ioctl_dbg_unregister(struct file *filep,
+static int kfd_ioctl_dbg_register(struct file *filep,
 				struct kfd_process *p, void *data)
 {
-	struct kfd_ioctl_dbg_unregister_args *args = data;
-	struct kfd_dev *dev;
-	long status;
-
-	dev = kfd_device_by_id(args->gpu_id);
-	if (!dev)
-		return -EINVAL;
-
-	if (dev->device_info->asic_family == CHIP_CARRIZO) {
-		pr_debug("kfd_ioctl_dbg_unregister not supported on CZ\n");
-		return -EINVAL;
-	}
-
-	mutex_lock(kfd_get_dbgmgr_mutex());
-
-	status = kfd_dbgmgr_unregister(dev->dbgmgr, p);
-	if (!status) {
-		kfd_dbgmgr_destroy(dev->dbgmgr);
-		dev->dbgmgr = NULL;
-	}
-
-	mutex_unlock(kfd_get_dbgmgr_mutex());
+	return -EPERM;
+}
 
-	return status;
+static int kfd_ioctl_dbg_unregister(struct file *filep,
+				struct kfd_process *p, void *data)
+{
+	return -EPERM;
 }
 
-/*
- * Parse and generate variable size data structure for address watch.
- * Total size of the buffer and # watch points is limited in order
- * to prevent kernel abuse. (no bearing to the much smaller HW limitation
- * which is enforced by dbgdev module)
- * please also note that the watch address itself are not "copied from user",
- * since it be set into the HW in user mode values.
- *
- */
 static int kfd_ioctl_dbg_address_watch(struct file *filep,
 					struct kfd_process *p, void *data)
 {
-	struct kfd_ioctl_dbg_address_watch_args *args = data;
-	struct kfd_dev *dev;
-	struct dbg_address_watch_info aw_info;
-	unsigned char *args_buff;
-	long status;
-	void __user *cmd_from_user;
-	uint64_t watch_mask_value = 0;
-	unsigned int args_idx = 0;
-
-	memset((void *) &aw_info, 0, sizeof(struct dbg_address_watch_info));
-
-	dev = kfd_device_by_id(args->gpu_id);
-	if (!dev)
-		return -EINVAL;
-
-	if (dev->device_info->asic_family == CHIP_CARRIZO) {
-		pr_debug("kfd_ioctl_dbg_wave_control not supported on CZ\n");
-		return -EINVAL;
-	}
-
-	cmd_from_user = (void __user *) args->content_ptr;
-
-	/* Validate arguments */
-
-	if ((args->buf_size_in_bytes > MAX_ALLOWED_AW_BUFF_SIZE) ||
-		(args->buf_size_in_bytes <= sizeof(*args) + sizeof(int) * 2) ||
-		(cmd_from_user == NULL))
-		return -EINVAL;
-
-	/* this is the actual buffer to work with */
-	args_buff = memdup_user(cmd_from_user,
-				args->buf_size_in_bytes - sizeof(*args));
-	if (IS_ERR(args_buff))
-		return PTR_ERR(args_buff);
-
-	aw_info.process = p;
-
-	aw_info.num_watch_points = *((uint32_t *)(&args_buff[args_idx]));
-	args_idx += sizeof(aw_info.num_watch_points);
-
-	aw_info.watch_mode = (enum HSA_DBG_WATCH_MODE *) &args_buff[args_idx];
-	args_idx += sizeof(enum HSA_DBG_WATCH_MODE) * aw_info.num_watch_points;
-
-	/*
-	 * set watch address base pointer to point on the array base
-	 * within args_buff
-	 */
-	aw_info.watch_address = (uint64_t *) &args_buff[args_idx];
-
-	/* skip over the addresses buffer */
-	args_idx += sizeof(aw_info.watch_address) * aw_info.num_watch_points;
-
-	if (args_idx >= args->buf_size_in_bytes - sizeof(*args)) {
-		status = -EINVAL;
-		goto out;
-	}
-
-	watch_mask_value = (uint64_t) args_buff[args_idx];
-
-	if (watch_mask_value > 0) {
-		/*
-		 * There is an array of masks.
-		 * set watch mask base pointer to point on the array base
-		 * within args_buff
-		 */
-		aw_info.watch_mask = (uint64_t *) &args_buff[args_idx];
-
-		/* skip over the masks buffer */
-		args_idx += sizeof(aw_info.watch_mask) *
-				aw_info.num_watch_points;
-	} else {
-		/* just the NULL mask, set to NULL and skip over it */
-		aw_info.watch_mask = NULL;
-		args_idx += sizeof(aw_info.watch_mask);
-	}
-
-	if (args_idx >= args->buf_size_in_bytes - sizeof(args)) {
-		status = -EINVAL;
-		goto out;
-	}
-
-	/* Currently HSA Event is not supported for DBG */
-	aw_info.watch_event = NULL;
-
-	mutex_lock(kfd_get_dbgmgr_mutex());
-
-	status = kfd_dbgmgr_address_watch(dev->dbgmgr, &aw_info);
-
-	mutex_unlock(kfd_get_dbgmgr_mutex());
-
-out:
-	kfree(args_buff);
-
-	return status;
+	return -EPERM;
 }
 
 /* Parse and generate fixed size data structure for wave control */
 static int kfd_ioctl_dbg_wave_control(struct file *filep,
 					struct kfd_process *p, void *data)
 {
-	struct kfd_ioctl_dbg_wave_control_args *args = data;
-	struct kfd_dev *dev;
-	struct dbg_wave_control_info wac_info;
-	unsigned char *args_buff;
-	uint32_t computed_buff_size;
-	long status;
-	void __user *cmd_from_user;
-	unsigned int args_idx = 0;
-
-	memset((void *) &wac_info, 0, sizeof(struct dbg_wave_control_info));
-
-	/* we use compact form, independent of the packing attribute value */
-	computed_buff_size = sizeof(*args) +
-				sizeof(wac_info.mode) +
-				sizeof(wac_info.operand) +
-				sizeof(wac_info.dbgWave_msg.DbgWaveMsg) +
-				sizeof(wac_info.dbgWave_msg.MemoryVA) +
-				sizeof(wac_info.trapId);
-
-	dev = kfd_device_by_id(args->gpu_id);
-	if (!dev)
-		return -EINVAL;
-
-	if (dev->device_info->asic_family == CHIP_CARRIZO) {
-		pr_debug("kfd_ioctl_dbg_wave_control not supported on CZ\n");
-		return -EINVAL;
-	}
-
-	/* input size must match the computed "compact" size */
-	if (args->buf_size_in_bytes != computed_buff_size) {
-		pr_debug("size mismatch, computed : actual %u : %u\n",
-				args->buf_size_in_bytes, computed_buff_size);
-		return -EINVAL;
-	}
-
-	cmd_from_user = (void __user *) args->content_ptr;
-
-	if (cmd_from_user == NULL)
-		return -EINVAL;
-
-	/* copy the entire buffer from user */
-
-	args_buff = memdup_user(cmd_from_user,
-				args->buf_size_in_bytes - sizeof(*args));
-	if (IS_ERR(args_buff))
-		return PTR_ERR(args_buff);
-
-	/* move ptr to the start of the "pay-load" area */
-	wac_info.process = p;
-
-	wac_info.operand = *((enum HSA_DBG_WAVEOP *)(&args_buff[args_idx]));
-	args_idx += sizeof(wac_info.operand);
-
-	wac_info.mode = *((enum HSA_DBG_WAVEMODE *)(&args_buff[args_idx]));
-	args_idx += sizeof(wac_info.mode);
-
-	wac_info.trapId = *((uint32_t *)(&args_buff[args_idx]));
-	args_idx += sizeof(wac_info.trapId);
-
-	wac_info.dbgWave_msg.DbgWaveMsg.WaveMsgInfoGen2.Value =
-					*((uint32_t *)(&args_buff[args_idx]));
-	wac_info.dbgWave_msg.MemoryVA = NULL;
-
-	mutex_lock(kfd_get_dbgmgr_mutex());
-
-	pr_debug("Calling dbg manager process %p, operand %u, mode %u, trapId %u, message %u\n",
-			wac_info.process, wac_info.operand,
-			wac_info.mode, wac_info.trapId,
-			wac_info.dbgWave_msg.DbgWaveMsg.WaveMsgInfoGen2.Value);
-
-	status = kfd_dbgmgr_wave_control(dev->dbgmgr, &wac_info);
-
-	pr_debug("Returned status of dbg manager is %ld\n", status);
-
-	mutex_unlock(kfd_get_dbgmgr_mutex());
-
-	kfree(args_buff);
-
-	return status;
+	return -EPERM;
 }
 
 static int kfd_ioctl_get_clock_counters(struct file *filep,
 				struct kfd_process *p, void *data)
 {
 	struct kfd_ioctl_get_clock_counters_args *args = data;
-	struct kfd_dev *dev;
-	struct timespec64 time;
-
-	dev = kfd_device_by_id(args->gpu_id);
-	if (dev == NULL)
-		return -EINVAL;
+	struct kfd_process_device *pdd;
 
-	/* Reading GPU clock counter from KGD */
-	args->gpu_clock_counter =
-		dev->kfd2kgd->get_gpu_clock_counter(dev->kgd);
+	mutex_lock(&p->mutex);
+	pdd = kfd_process_device_data_by_id(p, args->gpu_id);
+	mutex_unlock(&p->mutex);
+	if (pdd)
+		/* Reading GPU clock counter from KGD */
+		args->gpu_clock_counter = amdgpu_amdkfd_get_gpu_clock_counter(pdd->dev->adev);
+	else
+		/* Node without GPU resource */
+		args->gpu_clock_counter = 0;
 
 	/* No access to rdtsc. Using raw monotonic time */
-	getrawmonotonic64(&time);
-	args->cpu_clock_counter = (uint64_t)timespec64_to_ns(&time);
-
-	get_monotonic_boottime64(&time);
-	args->system_clock_counter = (uint64_t)timespec64_to_ns(&time);
+	args->cpu_clock_counter = ktime_get_raw_ns();
+	args->system_clock_counter = ktime_get_boottime_ns();
 
 	/* Since the counter is in nano-seconds we use 1GHz frequency */
 	args->system_clock_freq = 1000000000;
@@ -742,54 +697,128 @@ static int kfd_ioctl_get_process_apertures(struct file *filp,
 {
 	struct kfd_ioctl_get_process_apertures_args *args = data;
 	struct kfd_process_device_apertures *pAperture;
-	struct kfd_process_device *pdd;
+	int i;
 
-	dev_dbg(kfd_device, "get apertures for PASID %d", p->pasid);
+	dev_dbg(kfd_device, "get apertures for process pid %d", p->lead_thread->pid);
 
 	args->num_of_nodes = 0;
 
 	mutex_lock(&p->mutex);
+	/* Run over all pdd of the process */
+	for (i = 0; i < p->n_pdds; i++) {
+		struct kfd_process_device *pdd = p->pdds[i];
+
+		pAperture =
+			&args->process_apertures[args->num_of_nodes];
+		pAperture->gpu_id = pdd->dev->id;
+		pAperture->lds_base = pdd->lds_base;
+		pAperture->lds_limit = pdd->lds_limit;
+		pAperture->gpuvm_base = pdd->gpuvm_base;
+		pAperture->gpuvm_limit = pdd->gpuvm_limit;
+		pAperture->scratch_base = pdd->scratch_base;
+		pAperture->scratch_limit = pdd->scratch_limit;
+
+		dev_dbg(kfd_device,
+			"node id %u\n", args->num_of_nodes);
+		dev_dbg(kfd_device,
+			"gpu id %u\n", pdd->dev->id);
+		dev_dbg(kfd_device,
+			"lds_base %llX\n", pdd->lds_base);
+		dev_dbg(kfd_device,
+			"lds_limit %llX\n", pdd->lds_limit);
+		dev_dbg(kfd_device,
+			"gpuvm_base %llX\n", pdd->gpuvm_base);
+		dev_dbg(kfd_device,
+			"gpuvm_limit %llX\n", pdd->gpuvm_limit);
+		dev_dbg(kfd_device,
+			"scratch_base %llX\n", pdd->scratch_base);
+		dev_dbg(kfd_device,
+			"scratch_limit %llX\n", pdd->scratch_limit);
+
+		if (++args->num_of_nodes >= NUM_OF_SUPPORTED_GPUS)
+			break;
+	}
+	mutex_unlock(&p->mutex);
+
+	return 0;
+}
+
+static int kfd_ioctl_get_process_apertures_new(struct file *filp,
+				struct kfd_process *p, void *data)
+{
+	struct kfd_ioctl_get_process_apertures_new_args *args = data;
+	struct kfd_process_device_apertures *pa;
+	int ret;
+	int i;
+
+	dev_dbg(kfd_device, "get apertures for process pid %d",
+			p->lead_thread->pid);
+
+	if (args->num_of_nodes == 0) {
+		/* Return number of nodes, so that user space can alloacate
+		 * sufficient memory
+		 */
+		mutex_lock(&p->mutex);
+		args->num_of_nodes = p->n_pdds;
+		goto out_unlock;
+	}
+
+	/* Fill in process-aperture information for all available
+	 * nodes, but not more than args->num_of_nodes as that is
+	 * the amount of memory allocated by user
+	 */
+	pa = kcalloc(args->num_of_nodes, sizeof(struct kfd_process_device_apertures),
+		     GFP_KERNEL);
+	if (!pa)
+		return -ENOMEM;
+
+	mutex_lock(&p->mutex);
 
-	/*if the process-device list isn't empty*/
-	if (kfd_has_process_device_data(p)) {
-		/* Run over all pdd of the process */
-		pdd = kfd_get_first_process_device_data(p);
-		do {
-			pAperture =
-				&args->process_apertures[args->num_of_nodes];
-			pAperture->gpu_id = pdd->dev->id;
-			pAperture->lds_base = pdd->lds_base;
-			pAperture->lds_limit = pdd->lds_limit;
-			pAperture->gpuvm_base = pdd->gpuvm_base;
-			pAperture->gpuvm_limit = pdd->gpuvm_limit;
-			pAperture->scratch_base = pdd->scratch_base;
-			pAperture->scratch_limit = pdd->scratch_limit;
-
-			dev_dbg(kfd_device,
-				"node id %u\n", args->num_of_nodes);
-			dev_dbg(kfd_device,
-				"gpu id %u\n", pdd->dev->id);
-			dev_dbg(kfd_device,
-				"lds_base %llX\n", pdd->lds_base);
-			dev_dbg(kfd_device,
-				"lds_limit %llX\n", pdd->lds_limit);
-			dev_dbg(kfd_device,
-				"gpuvm_base %llX\n", pdd->gpuvm_base);
-			dev_dbg(kfd_device,
-				"gpuvm_limit %llX\n", pdd->gpuvm_limit);
-			dev_dbg(kfd_device,
-				"scratch_base %llX\n", pdd->scratch_base);
-			dev_dbg(kfd_device,
-				"scratch_limit %llX\n", pdd->scratch_limit);
-
-			args->num_of_nodes++;
-
-			pdd = kfd_get_next_process_device_data(p, pdd);
-		} while (pdd && (args->num_of_nodes < NUM_OF_SUPPORTED_GPUS));
+	if (!p->n_pdds) {
+		args->num_of_nodes = 0;
+		kfree(pa);
+		goto out_unlock;
 	}
 
+	/* Run over all pdd of the process */
+	for (i = 0; i < min(p->n_pdds, args->num_of_nodes); i++) {
+		struct kfd_process_device *pdd = p->pdds[i];
+
+		pa[i].gpu_id = pdd->dev->id;
+		pa[i].lds_base = pdd->lds_base;
+		pa[i].lds_limit = pdd->lds_limit;
+		pa[i].gpuvm_base = pdd->gpuvm_base;
+		pa[i].gpuvm_limit = pdd->gpuvm_limit;
+		pa[i].scratch_base = pdd->scratch_base;
+		pa[i].scratch_limit = pdd->scratch_limit;
+
+		dev_dbg(kfd_device,
+			"gpu id %u\n", pdd->dev->id);
+		dev_dbg(kfd_device,
+			"lds_base %llX\n", pdd->lds_base);
+		dev_dbg(kfd_device,
+			"lds_limit %llX\n", pdd->lds_limit);
+		dev_dbg(kfd_device,
+			"gpuvm_base %llX\n", pdd->gpuvm_base);
+		dev_dbg(kfd_device,
+			"gpuvm_limit %llX\n", pdd->gpuvm_limit);
+		dev_dbg(kfd_device,
+			"scratch_base %llX\n", pdd->scratch_base);
+		dev_dbg(kfd_device,
+			"scratch_limit %llX\n", pdd->scratch_limit);
+	}
 	mutex_unlock(&p->mutex);
 
+	args->num_of_nodes = i;
+	ret = copy_to_user(
+			(void __user *)args->kfd_process_device_apertures_ptr,
+			pa,
+			(i * sizeof(struct kfd_process_device_apertures)));
+	kfree(pa);
+	return ret ? -EFAULT : 0;
+
+out_unlock:
+	mutex_unlock(&p->mutex);
 	return 0;
 }
 
@@ -799,12 +828,25 @@ static int kfd_ioctl_create_event(struct file *filp, struct kfd_process *p,
 	struct kfd_ioctl_create_event_args *args = data;
 	int err;
 
+	/* For dGPUs the event page is allocated in user mode. The
+	 * handle is passed to KFD with the first call to this IOCTL
+	 * through the event_page_offset field.
+	 */
+	if (args->event_page_offset) {
+		mutex_lock(&p->mutex);
+		err = kfd_kmap_event_page(p, args->event_page_offset);
+		mutex_unlock(&p->mutex);
+		if (err)
+			return err;
+	}
+
 	err = kfd_event_create(filp, p, args->event_type,
 				args->auto_reset != 0, args->node_id,
 				&args->event_id, &args->event_trigger_data,
 				&args->event_page_offset,
 				&args->event_slot_index);
 
+	pr_debug("Created event (id:0x%08x) (%s)\n", args->event_id, __func__);
 	return err;
 }
 
@@ -836,31 +878,27 @@ static int kfd_ioctl_wait_events(struct file *filp, struct kfd_process *p,
 				void *data)
 {
 	struct kfd_ioctl_wait_events_args *args = data;
-	enum kfd_event_wait_result wait_result;
-	int err;
 
-	err = kfd_wait_on_events(p, args->num_events,
+	return kfd_wait_on_events(p, args->num_events,
 			(void __user *)args->events_ptr,
 			(args->wait_for_all != 0),
-			args->timeout, &wait_result);
-
-	args->wait_result = wait_result;
-
-	return err;
+			&args->timeout, &args->wait_result);
 }
 static int kfd_ioctl_set_scratch_backing_va(struct file *filep,
 					struct kfd_process *p, void *data)
 {
 	struct kfd_ioctl_set_scratch_backing_va_args *args = data;
 	struct kfd_process_device *pdd;
-	struct kfd_dev *dev;
+	struct kfd_node *dev;
 	long err;
 
-	dev = kfd_device_by_id(args->gpu_id);
-	if (!dev)
-		return -EINVAL;
-
 	mutex_lock(&p->mutex);
+	pdd = kfd_process_device_data_by_id(p, args->gpu_id);
+	if (!pdd) {
+		err = -EINVAL;
+		goto err_pdd;
+	}
+	dev = pdd->dev;
 
 	pdd = kfd_bind_process_to_device(dev, p);
 	if (IS_ERR(pdd)) {
@@ -872,13 +910,15 @@ static int kfd_ioctl_set_scratch_backing_va(struct file *filep,
 
 	mutex_unlock(&p->mutex);
 
-	if (sched_policy == KFD_SCHED_POLICY_NO_HWS && pdd->qpd.vmid != 0)
+	if (dev->dqm->sched_policy == KFD_SCHED_POLICY_NO_HWS &&
+	    pdd->qpd.vmid != 0 && dev->kfd2kgd->set_scratch_backing_va)
 		dev->kfd2kgd->set_scratch_backing_va(
-			dev->kgd, args->va_addr, pdd->qpd.vmid);
+			dev->adev, args->va_addr, pdd->qpd.vmid);
 
 	return 0;
 
 bind_process_to_device_fail:
+err_pdd:
 	mutex_unlock(&p->mutex);
 	return err;
 }
@@ -887,13 +927,17 @@ static int kfd_ioctl_get_tile_config(struct file *filep,
 		struct kfd_process *p, void *data)
 {
 	struct kfd_ioctl_get_tile_config_args *args = data;
-	struct kfd_dev *dev;
+	struct kfd_process_device *pdd;
 	struct tile_config config;
 	int err = 0;
 
-	dev = kfd_device_by_id(args->gpu_id);
+	mutex_lock(&p->mutex);
+	pdd = kfd_process_device_data_by_id(p, args->gpu_id);
+	mutex_unlock(&p->mutex);
+	if (!pdd)
+		return -EINVAL;
 
-	dev->kfd2kgd->get_tile_config(dev->kgd, &config);
+	amdgpu_amdkfd_get_tile_config(pdd->dev->adev, &config);
 
 	args->gb_addr_config = config.gb_addr_config;
 	args->num_banks = config.num_banks;
@@ -923,6 +967,2156 @@ static int kfd_ioctl_get_tile_config(struct file *filep,
 	return 0;
 }
 
+static int kfd_ioctl_acquire_vm(struct file *filep, struct kfd_process *p,
+				void *data)
+{
+	struct kfd_ioctl_acquire_vm_args *args = data;
+	struct kfd_process_device *pdd;
+	struct file *drm_file;
+	int ret;
+
+	drm_file = fget(args->drm_fd);
+	if (!drm_file)
+		return -EINVAL;
+
+	mutex_lock(&p->mutex);
+	pdd = kfd_process_device_data_by_id(p, args->gpu_id);
+	if (!pdd) {
+		ret = -EINVAL;
+		goto err_pdd;
+	}
+
+	if (pdd->drm_file) {
+		ret = pdd->drm_file == drm_file ? 0 : -EBUSY;
+		goto err_drm_file;
+	}
+
+	ret = kfd_process_device_init_vm(pdd, drm_file);
+	if (ret)
+		goto err_unlock;
+
+	/* On success, the PDD keeps the drm_file reference */
+	mutex_unlock(&p->mutex);
+
+	return 0;
+
+err_unlock:
+err_pdd:
+err_drm_file:
+	mutex_unlock(&p->mutex);
+	fput(drm_file);
+	return ret;
+}
+
+bool kfd_dev_is_large_bar(struct kfd_node *dev)
+{
+	if (dev->kfd->adev->debug_largebar) {
+		pr_debug("Simulate large-bar allocation on non large-bar machine\n");
+		return true;
+	}
+
+	if (dev->local_mem_info.local_mem_size_private == 0 &&
+	    dev->local_mem_info.local_mem_size_public > 0)
+		return true;
+
+	if (dev->local_mem_info.local_mem_size_public == 0 &&
+	    dev->kfd->adev->gmc.is_app_apu) {
+		pr_debug("APP APU, Consider like a large bar system\n");
+		return true;
+	}
+
+	return false;
+}
+
+static int kfd_ioctl_get_available_memory(struct file *filep,
+					  struct kfd_process *p, void *data)
+{
+	struct kfd_ioctl_get_available_memory_args *args = data;
+	struct kfd_process_device *pdd = kfd_lock_pdd_by_id(p, args->gpu_id);
+
+	if (!pdd)
+		return -EINVAL;
+	args->available = amdgpu_amdkfd_get_available_memory(pdd->dev->adev,
+							pdd->dev->node_id);
+	kfd_unlock_pdd(pdd);
+	return 0;
+}
+
+static int kfd_ioctl_alloc_memory_of_gpu(struct file *filep,
+					struct kfd_process *p, void *data)
+{
+	struct kfd_ioctl_alloc_memory_of_gpu_args *args = data;
+	struct kfd_process_device *pdd;
+	void *mem;
+	struct kfd_node *dev;
+	int idr_handle;
+	long err;
+	uint64_t offset = args->mmap_offset;
+	uint32_t flags = args->flags;
+
+	if (args->size == 0)
+		return -EINVAL;
+
+#if IS_ENABLED(CONFIG_HSA_AMD_SVM)
+	/* Flush pending deferred work to avoid racing with deferred actions
+	 * from previous memory map changes (e.g. munmap).
+	 */
+	svm_range_list_lock_and_flush_work(&p->svms, current->mm);
+	mutex_lock(&p->svms.lock);
+	mmap_write_unlock(current->mm);
+
+	/* Skip a special case that allocates VRAM without VA,
+	 * VA will be invalid of 0.
+	 */
+	if (!(!args->va_addr && (flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM)) &&
+	    interval_tree_iter_first(&p->svms.objects,
+				     args->va_addr >> PAGE_SHIFT,
+				     (args->va_addr + args->size - 1) >> PAGE_SHIFT)) {
+		pr_err("Address: 0x%llx already allocated by SVM\n",
+			args->va_addr);
+		mutex_unlock(&p->svms.lock);
+		return -EADDRINUSE;
+	}
+
+	/* When register user buffer check if it has been registered by svm by
+	 * buffer cpu virtual address.
+	 */
+	if ((flags & KFD_IOC_ALLOC_MEM_FLAGS_USERPTR) &&
+	    interval_tree_iter_first(&p->svms.objects,
+				     args->mmap_offset >> PAGE_SHIFT,
+				     (args->mmap_offset  + args->size - 1) >> PAGE_SHIFT)) {
+		pr_err("User Buffer Address: 0x%llx already allocated by SVM\n",
+			args->mmap_offset);
+		mutex_unlock(&p->svms.lock);
+		return -EADDRINUSE;
+	}
+
+	mutex_unlock(&p->svms.lock);
+#endif
+	mutex_lock(&p->mutex);
+	pdd = kfd_process_device_data_by_id(p, args->gpu_id);
+	if (!pdd) {
+		err = -EINVAL;
+		goto err_pdd;
+	}
+
+	dev = pdd->dev;
+
+	if ((flags & KFD_IOC_ALLOC_MEM_FLAGS_PUBLIC) &&
+		(flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM) &&
+		!kfd_dev_is_large_bar(dev)) {
+		pr_err("Alloc host visible vram on small bar is not allowed\n");
+		err = -EINVAL;
+		goto err_large_bar;
+	}
+
+	pdd = kfd_bind_process_to_device(dev, p);
+	if (IS_ERR(pdd)) {
+		err = PTR_ERR(pdd);
+		goto err_unlock;
+	}
+
+	if (flags & KFD_IOC_ALLOC_MEM_FLAGS_DOORBELL) {
+		if (args->size != kfd_doorbell_process_slice(dev->kfd)) {
+			err = -EINVAL;
+			goto err_unlock;
+		}
+		offset = kfd_get_process_doorbells(pdd);
+		if (!offset) {
+			err = -ENOMEM;
+			goto err_unlock;
+		}
+	} else if (flags & KFD_IOC_ALLOC_MEM_FLAGS_MMIO_REMAP) {
+		if (args->size != PAGE_SIZE) {
+			err = -EINVAL;
+			goto err_unlock;
+		}
+		offset = dev->adev->rmmio_remap.bus_addr;
+		if (!offset || (PAGE_SIZE > 4096)) {
+			err = -ENOMEM;
+			goto err_unlock;
+		}
+	}
+
+	err = amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
+		dev->adev, args->va_addr, args->size,
+		pdd->drm_priv, (struct kgd_mem **) &mem, &offset,
+		flags, false);
+
+	if (err)
+		goto err_unlock;
+
+	idr_handle = kfd_process_device_create_obj_handle(pdd, mem);
+	if (idr_handle < 0) {
+		err = -EFAULT;
+		goto err_free;
+	}
+
+	/* Update the VRAM usage count */
+	if (flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM) {
+		uint64_t size = args->size;
+
+		if (flags & KFD_IOC_ALLOC_MEM_FLAGS_AQL_QUEUE_MEM)
+			size >>= 1;
+		atomic64_add(PAGE_ALIGN(size), &pdd->vram_usage);
+	}
+
+	mutex_unlock(&p->mutex);
+
+	args->handle = MAKE_HANDLE(args->gpu_id, idr_handle);
+	args->mmap_offset = offset;
+
+	/* MMIO is mapped through kfd device
+	 * Generate a kfd mmap offset
+	 */
+	if (flags & KFD_IOC_ALLOC_MEM_FLAGS_MMIO_REMAP)
+		args->mmap_offset = KFD_MMAP_TYPE_MMIO
+					| KFD_MMAP_GPU_ID(args->gpu_id);
+
+	return 0;
+
+err_free:
+	amdgpu_amdkfd_gpuvm_free_memory_of_gpu(dev->adev, (struct kgd_mem *)mem,
+					       pdd->drm_priv, NULL);
+err_unlock:
+err_pdd:
+err_large_bar:
+	mutex_unlock(&p->mutex);
+	return err;
+}
+
+static int kfd_ioctl_free_memory_of_gpu(struct file *filep,
+					struct kfd_process *p, void *data)
+{
+	struct kfd_ioctl_free_memory_of_gpu_args *args = data;
+	struct kfd_process_device *pdd;
+	void *mem;
+	int ret;
+	uint64_t size = 0;
+
+	mutex_lock(&p->mutex);
+	/*
+	 * Safeguard to prevent user space from freeing signal BO.
+	 * It will be freed at process termination.
+	 */
+	if (p->signal_handle && (p->signal_handle == args->handle)) {
+		pr_err("Free signal BO is not allowed\n");
+		ret = -EPERM;
+		goto err_unlock;
+	}
+
+	pdd = kfd_process_device_data_by_id(p, GET_GPU_ID(args->handle));
+	if (!pdd) {
+		pr_err("Process device data doesn't exist\n");
+		ret = -EINVAL;
+		goto err_pdd;
+	}
+
+	mem = kfd_process_device_translate_handle(
+		pdd, GET_IDR_HANDLE(args->handle));
+	if (!mem) {
+		ret = -EINVAL;
+		goto err_unlock;
+	}
+
+	ret = amdgpu_amdkfd_gpuvm_free_memory_of_gpu(pdd->dev->adev,
+				(struct kgd_mem *)mem, pdd->drm_priv, &size);
+
+	/* If freeing the buffer failed, leave the handle in place for
+	 * clean-up during process tear-down.
+	 */
+	if (!ret)
+		kfd_process_device_remove_obj_handle(
+			pdd, GET_IDR_HANDLE(args->handle));
+
+	atomic64_sub(size, &pdd->vram_usage);
+
+err_unlock:
+err_pdd:
+	mutex_unlock(&p->mutex);
+	return ret;
+}
+
+static int kfd_ioctl_map_memory_to_gpu(struct file *filep,
+					struct kfd_process *p, void *data)
+{
+	struct kfd_ioctl_map_memory_to_gpu_args *args = data;
+	struct kfd_process_device *pdd, *peer_pdd;
+	void *mem;
+	struct kfd_node *dev;
+	long err = 0;
+	int i;
+	uint32_t *devices_arr = NULL;
+
+	if (!args->n_devices) {
+		pr_debug("Device IDs array empty\n");
+		return -EINVAL;
+	}
+	if (args->n_success > args->n_devices) {
+		pr_debug("n_success exceeds n_devices\n");
+		return -EINVAL;
+	}
+
+	devices_arr = kmalloc_array(args->n_devices, sizeof(*devices_arr),
+				    GFP_KERNEL);
+	if (!devices_arr)
+		return -ENOMEM;
+
+	err = copy_from_user(devices_arr,
+			     (void __user *)args->device_ids_array_ptr,
+			     args->n_devices * sizeof(*devices_arr));
+	if (err != 0) {
+		err = -EFAULT;
+		goto copy_from_user_failed;
+	}
+
+	mutex_lock(&p->mutex);
+	pdd = kfd_process_device_data_by_id(p, GET_GPU_ID(args->handle));
+	if (!pdd) {
+		err = -EINVAL;
+		goto get_process_device_data_failed;
+	}
+	dev = pdd->dev;
+
+	pdd = kfd_bind_process_to_device(dev, p);
+	if (IS_ERR(pdd)) {
+		err = PTR_ERR(pdd);
+		goto bind_process_to_device_failed;
+	}
+
+	mem = kfd_process_device_translate_handle(pdd,
+						GET_IDR_HANDLE(args->handle));
+	if (!mem) {
+		err = -ENOMEM;
+		goto get_mem_obj_from_handle_failed;
+	}
+
+	for (i = args->n_success; i < args->n_devices; i++) {
+		peer_pdd = kfd_process_device_data_by_id(p, devices_arr[i]);
+		if (!peer_pdd) {
+			pr_debug("Getting device by id failed for 0x%x\n",
+				 devices_arr[i]);
+			err = -EINVAL;
+			goto get_mem_obj_from_handle_failed;
+		}
+
+		peer_pdd = kfd_bind_process_to_device(peer_pdd->dev, p);
+		if (IS_ERR(peer_pdd)) {
+			err = PTR_ERR(peer_pdd);
+			goto get_mem_obj_from_handle_failed;
+		}
+
+		err = amdgpu_amdkfd_gpuvm_map_memory_to_gpu(
+			peer_pdd->dev->adev, (struct kgd_mem *)mem,
+			peer_pdd->drm_priv);
+		if (err) {
+			struct pci_dev *pdev = peer_pdd->dev->adev->pdev;
+
+			dev_err(dev->adev->dev,
+			       "Failed to map peer:%04x:%02x:%02x.%d mem_domain:%d\n",
+			       pci_domain_nr(pdev->bus),
+			       pdev->bus->number,
+			       PCI_SLOT(pdev->devfn),
+			       PCI_FUNC(pdev->devfn),
+			       ((struct kgd_mem *)mem)->domain);
+			goto map_memory_to_gpu_failed;
+		}
+		args->n_success = i+1;
+	}
+
+	err = amdgpu_amdkfd_gpuvm_sync_memory(dev->adev, (struct kgd_mem *) mem, true);
+	if (err) {
+		pr_debug("Sync memory failed, wait interrupted by user signal\n");
+		goto sync_memory_failed;
+	}
+
+	mutex_unlock(&p->mutex);
+
+	/* Flush TLBs after waiting for the page table updates to complete */
+	for (i = 0; i < args->n_devices; i++) {
+		peer_pdd = kfd_process_device_data_by_id(p, devices_arr[i]);
+		if (WARN_ON_ONCE(!peer_pdd))
+			continue;
+		kfd_flush_tlb(peer_pdd, TLB_FLUSH_LEGACY);
+	}
+	kfree(devices_arr);
+
+	return err;
+
+get_process_device_data_failed:
+bind_process_to_device_failed:
+get_mem_obj_from_handle_failed:
+map_memory_to_gpu_failed:
+sync_memory_failed:
+	mutex_unlock(&p->mutex);
+copy_from_user_failed:
+	kfree(devices_arr);
+
+	return err;
+}
+
+static int kfd_ioctl_unmap_memory_from_gpu(struct file *filep,
+					struct kfd_process *p, void *data)
+{
+	struct kfd_ioctl_unmap_memory_from_gpu_args *args = data;
+	struct kfd_process_device *pdd, *peer_pdd;
+	void *mem;
+	long err = 0;
+	uint32_t *devices_arr = NULL, i;
+	bool flush_tlb;
+
+	if (!args->n_devices) {
+		pr_debug("Device IDs array empty\n");
+		return -EINVAL;
+	}
+	if (args->n_success > args->n_devices) {
+		pr_debug("n_success exceeds n_devices\n");
+		return -EINVAL;
+	}
+
+	devices_arr = kmalloc_array(args->n_devices, sizeof(*devices_arr),
+				    GFP_KERNEL);
+	if (!devices_arr)
+		return -ENOMEM;
+
+	err = copy_from_user(devices_arr,
+			     (void __user *)args->device_ids_array_ptr,
+			     args->n_devices * sizeof(*devices_arr));
+	if (err != 0) {
+		err = -EFAULT;
+		goto copy_from_user_failed;
+	}
+
+	mutex_lock(&p->mutex);
+	pdd = kfd_process_device_data_by_id(p, GET_GPU_ID(args->handle));
+	if (!pdd) {
+		err = -EINVAL;
+		goto bind_process_to_device_failed;
+	}
+
+	mem = kfd_process_device_translate_handle(pdd,
+						GET_IDR_HANDLE(args->handle));
+	if (!mem) {
+		err = -ENOMEM;
+		goto get_mem_obj_from_handle_failed;
+	}
+
+	for (i = args->n_success; i < args->n_devices; i++) {
+		peer_pdd = kfd_process_device_data_by_id(p, devices_arr[i]);
+		if (!peer_pdd) {
+			err = -EINVAL;
+			goto get_mem_obj_from_handle_failed;
+		}
+		err = amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu(
+			peer_pdd->dev->adev, (struct kgd_mem *)mem, peer_pdd->drm_priv);
+		if (err) {
+			pr_debug("Failed to unmap from gpu %d/%d\n", i, args->n_devices);
+			goto unmap_memory_from_gpu_failed;
+		}
+		args->n_success = i+1;
+	}
+
+	flush_tlb = kfd_flush_tlb_after_unmap(pdd->dev->kfd);
+	if (flush_tlb) {
+		err = amdgpu_amdkfd_gpuvm_sync_memory(pdd->dev->adev,
+				(struct kgd_mem *) mem, true);
+		if (err) {
+			pr_debug("Sync memory failed, wait interrupted by user signal\n");
+			goto sync_memory_failed;
+		}
+	}
+
+	/* Flush TLBs after waiting for the page table updates to complete */
+	for (i = 0; i < args->n_devices; i++) {
+		peer_pdd = kfd_process_device_data_by_id(p, devices_arr[i]);
+		if (WARN_ON_ONCE(!peer_pdd))
+			continue;
+		if (flush_tlb)
+			kfd_flush_tlb(peer_pdd, TLB_FLUSH_HEAVYWEIGHT);
+
+		/* Remove dma mapping after tlb flush to avoid IO_PAGE_FAULT */
+		err = amdgpu_amdkfd_gpuvm_dmaunmap_mem(mem, peer_pdd->drm_priv);
+		if (err)
+			goto sync_memory_failed;
+	}
+
+	mutex_unlock(&p->mutex);
+
+	kfree(devices_arr);
+
+	return 0;
+
+bind_process_to_device_failed:
+get_mem_obj_from_handle_failed:
+unmap_memory_from_gpu_failed:
+sync_memory_failed:
+	mutex_unlock(&p->mutex);
+copy_from_user_failed:
+	kfree(devices_arr);
+	return err;
+}
+
+static int kfd_ioctl_alloc_queue_gws(struct file *filep,
+		struct kfd_process *p, void *data)
+{
+	int retval;
+	struct kfd_ioctl_alloc_queue_gws_args *args = data;
+	struct queue *q;
+	struct kfd_node *dev;
+
+	mutex_lock(&p->mutex);
+	q = pqm_get_user_queue(&p->pqm, args->queue_id);
+
+	if (q) {
+		dev = q->device;
+	} else {
+		retval = -EINVAL;
+		goto out_unlock;
+	}
+
+	if (!dev->gws) {
+		retval = -ENODEV;
+		goto out_unlock;
+	}
+
+	if (dev->dqm->sched_policy == KFD_SCHED_POLICY_NO_HWS) {
+		retval = -ENODEV;
+		goto out_unlock;
+	}
+
+	if (p->debug_trap_enabled && (!kfd_dbg_has_gws_support(dev) ||
+				      kfd_dbg_has_cwsr_workaround(dev))) {
+		retval = -EBUSY;
+		goto out_unlock;
+	}
+
+	retval = pqm_set_gws(&p->pqm, args->queue_id, args->num_gws ? dev->gws : NULL);
+	mutex_unlock(&p->mutex);
+
+	args->first_gws = 0;
+	return retval;
+
+out_unlock:
+	mutex_unlock(&p->mutex);
+	return retval;
+}
+
+static int kfd_ioctl_get_dmabuf_info(struct file *filep,
+		struct kfd_process *p, void *data)
+{
+	struct kfd_ioctl_get_dmabuf_info_args *args = data;
+	struct kfd_node *dev = NULL;
+	struct amdgpu_device *dmabuf_adev;
+	void *metadata_buffer = NULL;
+	uint32_t flags;
+	int8_t xcp_id;
+	unsigned int i;
+	int r;
+
+	/* Find a KFD GPU device that supports the get_dmabuf_info query */
+	for (i = 0; kfd_topology_enum_kfd_devices(i, &dev) == 0; i++)
+		if (dev && !kfd_devcgroup_check_permission(dev))
+			break;
+	if (!dev)
+		return -EINVAL;
+
+	if (args->metadata_ptr) {
+		metadata_buffer = kzalloc(args->metadata_size, GFP_KERNEL);
+		if (!metadata_buffer)
+			return -ENOMEM;
+	}
+
+	/* Get dmabuf info from KGD */
+	r = amdgpu_amdkfd_get_dmabuf_info(dev->adev, args->dmabuf_fd,
+					  &dmabuf_adev, &args->size,
+					  metadata_buffer, args->metadata_size,
+					  &args->metadata_size, &flags, &xcp_id);
+	if (r)
+		goto exit;
+
+	if (xcp_id >= 0)
+		args->gpu_id = dmabuf_adev->kfd.dev->nodes[xcp_id]->id;
+	else
+		args->gpu_id = dev->id;
+	args->flags = flags;
+
+	/* Copy metadata buffer to user mode */
+	if (metadata_buffer) {
+		r = copy_to_user((void __user *)args->metadata_ptr,
+				 metadata_buffer, args->metadata_size);
+		if (r != 0)
+			r = -EFAULT;
+	}
+
+exit:
+	kfree(metadata_buffer);
+
+	return r;
+}
+
+static int kfd_ioctl_import_dmabuf(struct file *filep,
+				   struct kfd_process *p, void *data)
+{
+	struct kfd_ioctl_import_dmabuf_args *args = data;
+	struct kfd_process_device *pdd;
+	int idr_handle;
+	uint64_t size;
+	void *mem;
+	int r;
+
+	mutex_lock(&p->mutex);
+	pdd = kfd_process_device_data_by_id(p, args->gpu_id);
+	if (!pdd) {
+		r = -EINVAL;
+		goto err_unlock;
+	}
+
+	pdd = kfd_bind_process_to_device(pdd->dev, p);
+	if (IS_ERR(pdd)) {
+		r = PTR_ERR(pdd);
+		goto err_unlock;
+	}
+
+	r = amdgpu_amdkfd_gpuvm_import_dmabuf_fd(pdd->dev->adev, args->dmabuf_fd,
+						 args->va_addr, pdd->drm_priv,
+						 (struct kgd_mem **)&mem, &size,
+						 NULL);
+	if (r)
+		goto err_unlock;
+
+	idr_handle = kfd_process_device_create_obj_handle(pdd, mem);
+	if (idr_handle < 0) {
+		r = -EFAULT;
+		goto err_free;
+	}
+
+	mutex_unlock(&p->mutex);
+
+	args->handle = MAKE_HANDLE(args->gpu_id, idr_handle);
+
+	return 0;
+
+err_free:
+	amdgpu_amdkfd_gpuvm_free_memory_of_gpu(pdd->dev->adev, (struct kgd_mem *)mem,
+					       pdd->drm_priv, NULL);
+err_unlock:
+	mutex_unlock(&p->mutex);
+	return r;
+}
+
+static int kfd_ioctl_export_dmabuf(struct file *filep,
+				   struct kfd_process *p, void *data)
+{
+	struct kfd_ioctl_export_dmabuf_args *args = data;
+	struct kfd_process_device *pdd;
+	struct dma_buf *dmabuf;
+	struct kfd_node *dev;
+	void *mem;
+	int ret = 0;
+
+	dev = kfd_device_by_id(GET_GPU_ID(args->handle));
+	if (!dev)
+		return -EINVAL;
+
+	mutex_lock(&p->mutex);
+
+	pdd = kfd_get_process_device_data(dev, p);
+	if (!pdd) {
+		ret = -EINVAL;
+		goto err_unlock;
+	}
+
+	mem = kfd_process_device_translate_handle(pdd,
+						GET_IDR_HANDLE(args->handle));
+	if (!mem) {
+		ret = -EINVAL;
+		goto err_unlock;
+	}
+
+	ret = amdgpu_amdkfd_gpuvm_export_dmabuf(mem, &dmabuf);
+	mutex_unlock(&p->mutex);
+	if (ret)
+		goto err_out;
+
+	ret = dma_buf_fd(dmabuf, args->flags);
+	if (ret < 0) {
+		dma_buf_put(dmabuf);
+		goto err_out;
+	}
+	/* dma_buf_fd assigns the reference count to the fd, no need to
+	 * put the reference here.
+	 */
+	args->dmabuf_fd = ret;
+
+	return 0;
+
+err_unlock:
+	mutex_unlock(&p->mutex);
+err_out:
+	return ret;
+}
+
+/* Handle requests for watching SMI events */
+static int kfd_ioctl_smi_events(struct file *filep,
+				struct kfd_process *p, void *data)
+{
+	struct kfd_ioctl_smi_events_args *args = data;
+	struct kfd_process_device *pdd;
+
+	mutex_lock(&p->mutex);
+
+	pdd = kfd_process_device_data_by_id(p, args->gpuid);
+	mutex_unlock(&p->mutex);
+	if (!pdd)
+		return -EINVAL;
+
+	return kfd_smi_event_open(pdd->dev, &args->anon_fd);
+}
+
+#if IS_ENABLED(CONFIG_HSA_AMD_SVM)
+
+static int kfd_ioctl_set_xnack_mode(struct file *filep,
+				    struct kfd_process *p, void *data)
+{
+	struct kfd_ioctl_set_xnack_mode_args *args = data;
+	int r = 0;
+
+	mutex_lock(&p->mutex);
+	if (args->xnack_enabled >= 0) {
+		if (!list_empty(&p->pqm.queues)) {
+			pr_debug("Process has user queues running\n");
+			r = -EBUSY;
+			goto out_unlock;
+		}
+
+		if (p->xnack_enabled == args->xnack_enabled)
+			goto out_unlock;
+
+		if (args->xnack_enabled && !kfd_process_xnack_mode(p, true)) {
+			r = -EPERM;
+			goto out_unlock;
+		}
+
+		r = svm_range_switch_xnack_reserve_mem(p, args->xnack_enabled);
+	} else {
+		args->xnack_enabled = p->xnack_enabled;
+	}
+
+out_unlock:
+	mutex_unlock(&p->mutex);
+
+	return r;
+}
+
+static int kfd_ioctl_svm(struct file *filep, struct kfd_process *p, void *data)
+{
+	struct kfd_ioctl_svm_args *args = data;
+	int r = 0;
+
+	pr_debug("start 0x%llx size 0x%llx op 0x%x nattr 0x%x\n",
+		 args->start_addr, args->size, args->op, args->nattr);
+
+	if ((args->start_addr & ~PAGE_MASK) || (args->size & ~PAGE_MASK))
+		return -EINVAL;
+	if (!args->start_addr || !args->size)
+		return -EINVAL;
+
+	r = svm_ioctl(p, args->op, args->start_addr, args->size, args->nattr,
+		      args->attrs);
+
+	return r;
+}
+#else
+static int kfd_ioctl_set_xnack_mode(struct file *filep,
+				    struct kfd_process *p, void *data)
+{
+	return -EPERM;
+}
+static int kfd_ioctl_svm(struct file *filep, struct kfd_process *p, void *data)
+{
+	return -EPERM;
+}
+#endif
+
+static int criu_checkpoint_process(struct kfd_process *p,
+			     uint8_t __user *user_priv_data,
+			     uint64_t *priv_offset)
+{
+	struct kfd_criu_process_priv_data process_priv;
+	int ret;
+
+	memset(&process_priv, 0, sizeof(process_priv));
+
+	process_priv.version = KFD_CRIU_PRIV_VERSION;
+	/* For CR, we don't consider negative xnack mode which is used for
+	 * querying without changing it, here 0 simply means disabled and 1
+	 * means enabled so retry for finding a valid PTE.
+	 */
+	process_priv.xnack_mode = p->xnack_enabled ? 1 : 0;
+
+	ret = copy_to_user(user_priv_data + *priv_offset,
+				&process_priv, sizeof(process_priv));
+
+	if (ret) {
+		pr_err("Failed to copy process information to user\n");
+		ret = -EFAULT;
+	}
+
+	*priv_offset += sizeof(process_priv);
+	return ret;
+}
+
+static int criu_checkpoint_devices(struct kfd_process *p,
+			     uint32_t num_devices,
+			     uint8_t __user *user_addr,
+			     uint8_t __user *user_priv_data,
+			     uint64_t *priv_offset)
+{
+	struct kfd_criu_device_priv_data *device_priv = NULL;
+	struct kfd_criu_device_bucket *device_buckets = NULL;
+	int ret = 0, i;
+
+	device_buckets = kvzalloc(num_devices * sizeof(*device_buckets), GFP_KERNEL);
+	if (!device_buckets) {
+		ret = -ENOMEM;
+		goto exit;
+	}
+
+	device_priv = kvzalloc(num_devices * sizeof(*device_priv), GFP_KERNEL);
+	if (!device_priv) {
+		ret = -ENOMEM;
+		goto exit;
+	}
+
+	for (i = 0; i < num_devices; i++) {
+		struct kfd_process_device *pdd = p->pdds[i];
+
+		device_buckets[i].user_gpu_id = pdd->user_gpu_id;
+		device_buckets[i].actual_gpu_id = pdd->dev->id;
+
+		/*
+		 * priv_data does not contain useful information for now and is reserved for
+		 * future use, so we do not set its contents.
+		 */
+	}
+
+	ret = copy_to_user(user_addr, device_buckets, num_devices * sizeof(*device_buckets));
+	if (ret) {
+		pr_err("Failed to copy device information to user\n");
+		ret = -EFAULT;
+		goto exit;
+	}
+
+	ret = copy_to_user(user_priv_data + *priv_offset,
+			   device_priv,
+			   num_devices * sizeof(*device_priv));
+	if (ret) {
+		pr_err("Failed to copy device information to user\n");
+		ret = -EFAULT;
+	}
+	*priv_offset += num_devices * sizeof(*device_priv);
+
+exit:
+	kvfree(device_buckets);
+	kvfree(device_priv);
+	return ret;
+}
+
+static uint32_t get_process_num_bos(struct kfd_process *p)
+{
+	uint32_t num_of_bos = 0;
+	int i;
+
+	/* Run over all PDDs of the process */
+	for (i = 0; i < p->n_pdds; i++) {
+		struct kfd_process_device *pdd = p->pdds[i];
+		void *mem;
+		int id;
+
+		idr_for_each_entry(&pdd->alloc_idr, mem, id) {
+			struct kgd_mem *kgd_mem = (struct kgd_mem *)mem;
+
+			if (!kgd_mem->va || kgd_mem->va > pdd->gpuvm_base)
+				num_of_bos++;
+		}
+	}
+	return num_of_bos;
+}
+
+static int criu_get_prime_handle(struct kgd_mem *mem,
+				 int flags, u32 *shared_fd,
+				 struct file **file)
+{
+	struct dma_buf *dmabuf;
+	int ret;
+
+	ret = amdgpu_amdkfd_gpuvm_export_dmabuf(mem, &dmabuf);
+	if (ret) {
+		pr_err("dmabuf export failed for the BO\n");
+		return ret;
+	}
+
+	ret = get_unused_fd_flags(flags);
+	if (ret < 0) {
+		pr_err("dmabuf create fd failed, ret:%d\n", ret);
+		goto out_free_dmabuf;
+	}
+
+	*shared_fd = ret;
+	*file = dmabuf->file;
+	return 0;
+
+out_free_dmabuf:
+	dma_buf_put(dmabuf);
+	return ret;
+}
+
+static void commit_files(struct file **files,
+			 struct kfd_criu_bo_bucket *bo_buckets,
+			 unsigned int count,
+			 int err)
+{
+	while (count--) {
+		struct file *file = files[count];
+
+		if (!file)
+			continue;
+		if (err) {
+			fput(file);
+			put_unused_fd(bo_buckets[count].dmabuf_fd);
+		} else {
+			fd_install(bo_buckets[count].dmabuf_fd, file);
+		}
+	}
+}
+
+static int criu_checkpoint_bos(struct kfd_process *p,
+			       uint32_t num_bos,
+			       uint8_t __user *user_bos,
+			       uint8_t __user *user_priv_data,
+			       uint64_t *priv_offset)
+{
+	struct kfd_criu_bo_bucket *bo_buckets;
+	struct kfd_criu_bo_priv_data *bo_privs;
+	struct file **files = NULL;
+	int ret = 0, pdd_index, bo_index = 0, id;
+	void *mem;
+
+	bo_buckets = kvzalloc(num_bos * sizeof(*bo_buckets), GFP_KERNEL);
+	if (!bo_buckets)
+		return -ENOMEM;
+
+	bo_privs = kvzalloc(num_bos * sizeof(*bo_privs), GFP_KERNEL);
+	if (!bo_privs) {
+		ret = -ENOMEM;
+		goto exit;
+	}
+
+	files = kvzalloc(num_bos * sizeof(struct file *), GFP_KERNEL);
+	if (!files) {
+		ret = -ENOMEM;
+		goto exit;
+	}
+
+	for (pdd_index = 0; pdd_index < p->n_pdds; pdd_index++) {
+		struct kfd_process_device *pdd = p->pdds[pdd_index];
+		struct amdgpu_bo *dumper_bo;
+		struct kgd_mem *kgd_mem;
+
+		idr_for_each_entry(&pdd->alloc_idr, mem, id) {
+			struct kfd_criu_bo_bucket *bo_bucket;
+			struct kfd_criu_bo_priv_data *bo_priv;
+			int i, dev_idx = 0;
+
+			kgd_mem = (struct kgd_mem *)mem;
+			dumper_bo = kgd_mem->bo;
+
+			/* Skip checkpointing BOs that are used for Trap handler
+			 * code and state. Currently, these BOs have a VA that
+			 * is less GPUVM Base
+			 */
+			if (kgd_mem->va && kgd_mem->va <= pdd->gpuvm_base)
+				continue;
+
+			bo_bucket = &bo_buckets[bo_index];
+			bo_priv = &bo_privs[bo_index];
+
+			bo_bucket->gpu_id = pdd->user_gpu_id;
+			bo_bucket->addr = (uint64_t)kgd_mem->va;
+			bo_bucket->size = amdgpu_bo_size(dumper_bo);
+			bo_bucket->alloc_flags = (uint32_t)kgd_mem->alloc_flags;
+			bo_priv->idr_handle = id;
+
+			if (bo_bucket->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_USERPTR) {
+				ret = amdgpu_ttm_tt_get_userptr(&dumper_bo->tbo,
+								&bo_priv->user_addr);
+				if (ret) {
+					pr_err("Failed to obtain user address for user-pointer bo\n");
+					goto exit;
+				}
+			}
+			if (bo_bucket->alloc_flags
+			    & (KFD_IOC_ALLOC_MEM_FLAGS_VRAM | KFD_IOC_ALLOC_MEM_FLAGS_GTT)) {
+				ret = criu_get_prime_handle(kgd_mem,
+						bo_bucket->alloc_flags &
+						KFD_IOC_ALLOC_MEM_FLAGS_WRITABLE ? DRM_RDWR : 0,
+						&bo_bucket->dmabuf_fd, &files[bo_index]);
+				if (ret)
+					goto exit;
+			} else {
+				bo_bucket->dmabuf_fd = KFD_INVALID_FD;
+			}
+
+			if (bo_bucket->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_DOORBELL)
+				bo_bucket->offset = KFD_MMAP_TYPE_DOORBELL |
+					KFD_MMAP_GPU_ID(pdd->dev->id);
+			else if (bo_bucket->alloc_flags &
+				KFD_IOC_ALLOC_MEM_FLAGS_MMIO_REMAP)
+				bo_bucket->offset = KFD_MMAP_TYPE_MMIO |
+					KFD_MMAP_GPU_ID(pdd->dev->id);
+			else
+				bo_bucket->offset = amdgpu_bo_mmap_offset(dumper_bo);
+
+			for (i = 0; i < p->n_pdds; i++) {
+				if (amdgpu_amdkfd_bo_mapped_to_dev(p->pdds[i]->drm_priv, kgd_mem))
+					bo_priv->mapped_gpuids[dev_idx++] = p->pdds[i]->user_gpu_id;
+			}
+
+			pr_debug("bo_size = 0x%llx, bo_addr = 0x%llx bo_offset = 0x%llx\n"
+					"gpu_id = 0x%x alloc_flags = 0x%x idr_handle = 0x%x",
+					bo_bucket->size,
+					bo_bucket->addr,
+					bo_bucket->offset,
+					bo_bucket->gpu_id,
+					bo_bucket->alloc_flags,
+					bo_priv->idr_handle);
+			bo_index++;
+		}
+	}
+
+	ret = copy_to_user(user_bos, bo_buckets, num_bos * sizeof(*bo_buckets));
+	if (ret) {
+		pr_err("Failed to copy BO information to user\n");
+		ret = -EFAULT;
+		goto exit;
+	}
+
+	ret = copy_to_user(user_priv_data + *priv_offset, bo_privs, num_bos * sizeof(*bo_privs));
+	if (ret) {
+		pr_err("Failed to copy BO priv information to user\n");
+		ret = -EFAULT;
+		goto exit;
+	}
+
+	*priv_offset += num_bos * sizeof(*bo_privs);
+
+exit:
+	commit_files(files, bo_buckets, bo_index, ret);
+	kvfree(files);
+	kvfree(bo_buckets);
+	kvfree(bo_privs);
+	return ret;
+}
+
+static int criu_get_process_object_info(struct kfd_process *p,
+					uint32_t *num_devices,
+					uint32_t *num_bos,
+					uint32_t *num_objects,
+					uint64_t *objs_priv_size)
+{
+	uint64_t queues_priv_data_size, svm_priv_data_size, priv_size;
+	uint32_t num_queues, num_events, num_svm_ranges;
+	int ret;
+
+	*num_devices = p->n_pdds;
+	*num_bos = get_process_num_bos(p);
+
+	ret = kfd_process_get_queue_info(p, &num_queues, &queues_priv_data_size);
+	if (ret)
+		return ret;
+
+	num_events = kfd_get_num_events(p);
+
+	svm_range_get_info(p, &num_svm_ranges, &svm_priv_data_size);
+
+	*num_objects = num_queues + num_events + num_svm_ranges;
+
+	if (objs_priv_size) {
+		priv_size = sizeof(struct kfd_criu_process_priv_data);
+		priv_size += *num_devices * sizeof(struct kfd_criu_device_priv_data);
+		priv_size += *num_bos * sizeof(struct kfd_criu_bo_priv_data);
+		priv_size += queues_priv_data_size;
+		priv_size += num_events * sizeof(struct kfd_criu_event_priv_data);
+		priv_size += svm_priv_data_size;
+		*objs_priv_size = priv_size;
+	}
+	return 0;
+}
+
+static int criu_checkpoint(struct file *filep,
+			   struct kfd_process *p,
+			   struct kfd_ioctl_criu_args *args)
+{
+	int ret;
+	uint32_t num_devices, num_bos, num_objects;
+	uint64_t priv_size, priv_offset = 0, bo_priv_offset;
+
+	if (!args->devices || !args->bos || !args->priv_data)
+		return -EINVAL;
+
+	mutex_lock(&p->mutex);
+
+	if (!p->n_pdds) {
+		pr_err("No pdd for given process\n");
+		ret = -ENODEV;
+		goto exit_unlock;
+	}
+
+	/* Confirm all process queues are evicted */
+	if (!p->queues_paused) {
+		pr_err("Cannot dump process when queues are not in evicted state\n");
+		/* CRIU plugin did not call op PROCESS_INFO before checkpointing */
+		ret = -EINVAL;
+		goto exit_unlock;
+	}
+
+	ret = criu_get_process_object_info(p, &num_devices, &num_bos, &num_objects, &priv_size);
+	if (ret)
+		goto exit_unlock;
+
+	if (num_devices != args->num_devices ||
+	    num_bos != args->num_bos ||
+	    num_objects != args->num_objects ||
+	    priv_size != args->priv_data_size) {
+
+		ret = -EINVAL;
+		goto exit_unlock;
+	}
+
+	/* each function will store private data inside priv_data and adjust priv_offset */
+	ret = criu_checkpoint_process(p, (uint8_t __user *)args->priv_data, &priv_offset);
+	if (ret)
+		goto exit_unlock;
+
+	ret = criu_checkpoint_devices(p, num_devices, (uint8_t __user *)args->devices,
+				(uint8_t __user *)args->priv_data, &priv_offset);
+	if (ret)
+		goto exit_unlock;
+
+	/* Leave room for BOs in the private data. They need to be restored
+	 * before events, but we checkpoint them last to simplify the error
+	 * handling.
+	 */
+	bo_priv_offset = priv_offset;
+	priv_offset += num_bos * sizeof(struct kfd_criu_bo_priv_data);
+
+	if (num_objects) {
+		ret = kfd_criu_checkpoint_queues(p, (uint8_t __user *)args->priv_data,
+						 &priv_offset);
+		if (ret)
+			goto exit_unlock;
+
+		ret = kfd_criu_checkpoint_events(p, (uint8_t __user *)args->priv_data,
+						 &priv_offset);
+		if (ret)
+			goto exit_unlock;
+
+		ret = kfd_criu_checkpoint_svm(p, (uint8_t __user *)args->priv_data, &priv_offset);
+		if (ret)
+			goto exit_unlock;
+	}
+
+	/* This must be the last thing in this function that can fail.
+	 * Otherwise we leak dmabuf file descriptors.
+	 */
+	ret = criu_checkpoint_bos(p, num_bos, (uint8_t __user *)args->bos,
+			   (uint8_t __user *)args->priv_data, &bo_priv_offset);
+
+exit_unlock:
+	mutex_unlock(&p->mutex);
+	if (ret)
+		pr_err("Failed to dump CRIU ret:%d\n", ret);
+	else
+		pr_debug("CRIU dump ret:%d\n", ret);
+
+	return ret;
+}
+
+static int criu_restore_process(struct kfd_process *p,
+				struct kfd_ioctl_criu_args *args,
+				uint64_t *priv_offset,
+				uint64_t max_priv_data_size)
+{
+	int ret = 0;
+	struct kfd_criu_process_priv_data process_priv;
+
+	if (*priv_offset + sizeof(process_priv) > max_priv_data_size)
+		return -EINVAL;
+
+	ret = copy_from_user(&process_priv,
+				(void __user *)(args->priv_data + *priv_offset),
+				sizeof(process_priv));
+	if (ret) {
+		pr_err("Failed to copy process private information from user\n");
+		ret = -EFAULT;
+		goto exit;
+	}
+	*priv_offset += sizeof(process_priv);
+
+	if (process_priv.version != KFD_CRIU_PRIV_VERSION) {
+		pr_err("Invalid CRIU API version (checkpointed:%d current:%d)\n",
+			process_priv.version, KFD_CRIU_PRIV_VERSION);
+		return -EINVAL;
+	}
+
+	pr_debug("Setting XNACK mode\n");
+	if (process_priv.xnack_mode && !kfd_process_xnack_mode(p, true)) {
+		pr_err("xnack mode cannot be set\n");
+		ret = -EPERM;
+		goto exit;
+	} else {
+		pr_debug("set xnack mode: %d\n", process_priv.xnack_mode);
+		p->xnack_enabled = process_priv.xnack_mode;
+	}
+
+exit:
+	return ret;
+}
+
+static int criu_restore_devices(struct kfd_process *p,
+				struct kfd_ioctl_criu_args *args,
+				uint64_t *priv_offset,
+				uint64_t max_priv_data_size)
+{
+	struct kfd_criu_device_bucket *device_buckets;
+	struct kfd_criu_device_priv_data *device_privs;
+	int ret = 0;
+	uint32_t i;
+
+	if (args->num_devices != p->n_pdds)
+		return -EINVAL;
+
+	if (*priv_offset + (args->num_devices * sizeof(*device_privs)) > max_priv_data_size)
+		return -EINVAL;
+
+	device_buckets = kmalloc_array(args->num_devices, sizeof(*device_buckets), GFP_KERNEL);
+	if (!device_buckets)
+		return -ENOMEM;
+
+	ret = copy_from_user(device_buckets, (void __user *)args->devices,
+				args->num_devices * sizeof(*device_buckets));
+	if (ret) {
+		pr_err("Failed to copy devices buckets from user\n");
+		ret = -EFAULT;
+		goto exit;
+	}
+
+	for (i = 0; i < args->num_devices; i++) {
+		struct kfd_node *dev;
+		struct kfd_process_device *pdd;
+		struct file *drm_file;
+
+		/* device private data is not currently used */
+
+		if (!device_buckets[i].user_gpu_id) {
+			pr_err("Invalid user gpu_id\n");
+			ret = -EINVAL;
+			goto exit;
+		}
+
+		dev = kfd_device_by_id(device_buckets[i].actual_gpu_id);
+		if (!dev) {
+			pr_err("Failed to find device with gpu_id = %x\n",
+				device_buckets[i].actual_gpu_id);
+			ret = -EINVAL;
+			goto exit;
+		}
+
+		pdd = kfd_get_process_device_data(dev, p);
+		if (!pdd) {
+			pr_err("Failed to get pdd for gpu_id = %x\n",
+					device_buckets[i].actual_gpu_id);
+			ret = -EINVAL;
+			goto exit;
+		}
+		pdd->user_gpu_id = device_buckets[i].user_gpu_id;
+
+		drm_file = fget(device_buckets[i].drm_fd);
+		if (!drm_file) {
+			pr_err("Invalid render node file descriptor sent from plugin (%d)\n",
+				device_buckets[i].drm_fd);
+			ret = -EINVAL;
+			goto exit;
+		}
+
+		if (pdd->drm_file) {
+			ret = -EINVAL;
+			goto exit;
+		}
+
+		/* create the vm using render nodes for kfd pdd */
+		if (kfd_process_device_init_vm(pdd, drm_file)) {
+			pr_err("could not init vm for given pdd\n");
+			/* On success, the PDD keeps the drm_file reference */
+			fput(drm_file);
+			ret = -EINVAL;
+			goto exit;
+		}
+		/*
+		 * pdd now already has the vm bound to render node so below api won't create a new
+		 * exclusive kfd mapping but use existing one with renderDXXX but is still needed
+		 * for iommu v2 binding  and runtime pm.
+		 */
+		pdd = kfd_bind_process_to_device(dev, p);
+		if (IS_ERR(pdd)) {
+			ret = PTR_ERR(pdd);
+			goto exit;
+		}
+
+		if (!pdd->qpd.proc_doorbells) {
+			ret = kfd_alloc_process_doorbells(dev->kfd, pdd);
+			if (ret)
+				goto exit;
+		}
+	}
+
+	/*
+	 * We are not copying device private data from user as we are not using the data for now,
+	 * but we still adjust for its private data.
+	 */
+	*priv_offset += args->num_devices * sizeof(*device_privs);
+
+exit:
+	kfree(device_buckets);
+	return ret;
+}
+
+static int criu_restore_memory_of_gpu(struct kfd_process_device *pdd,
+				      struct kfd_criu_bo_bucket *bo_bucket,
+				      struct kfd_criu_bo_priv_data *bo_priv,
+				      struct kgd_mem **kgd_mem)
+{
+	int idr_handle;
+	int ret;
+	const bool criu_resume = true;
+	u64 offset;
+
+	if (bo_bucket->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_DOORBELL) {
+		if (bo_bucket->size !=
+				kfd_doorbell_process_slice(pdd->dev->kfd))
+			return -EINVAL;
+
+		offset = kfd_get_process_doorbells(pdd);
+		if (!offset)
+			return -ENOMEM;
+	} else if (bo_bucket->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_MMIO_REMAP) {
+		/* MMIO BOs need remapped bus address */
+		if (bo_bucket->size != PAGE_SIZE) {
+			pr_err("Invalid page size\n");
+			return -EINVAL;
+		}
+		offset = pdd->dev->adev->rmmio_remap.bus_addr;
+		if (!offset || (PAGE_SIZE > 4096)) {
+			pr_err("amdgpu_amdkfd_get_mmio_remap_phys_addr failed\n");
+			return -ENOMEM;
+		}
+	} else if (bo_bucket->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_USERPTR) {
+		offset = bo_priv->user_addr;
+	}
+	/* Create the BO */
+	ret = amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(pdd->dev->adev, bo_bucket->addr,
+						      bo_bucket->size, pdd->drm_priv, kgd_mem,
+						      &offset, bo_bucket->alloc_flags, criu_resume);
+	if (ret) {
+		pr_err("Could not create the BO\n");
+		return ret;
+	}
+	pr_debug("New BO created: size:0x%llx addr:0x%llx offset:0x%llx\n",
+		 bo_bucket->size, bo_bucket->addr, offset);
+
+	/* Restore previous IDR handle */
+	pr_debug("Restoring old IDR handle for the BO");
+	idr_handle = idr_alloc(&pdd->alloc_idr, *kgd_mem, bo_priv->idr_handle,
+			       bo_priv->idr_handle + 1, GFP_KERNEL);
+
+	if (idr_handle < 0) {
+		pr_err("Could not allocate idr\n");
+		amdgpu_amdkfd_gpuvm_free_memory_of_gpu(pdd->dev->adev, *kgd_mem, pdd->drm_priv,
+						       NULL);
+		return -ENOMEM;
+	}
+
+	if (bo_bucket->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_DOORBELL)
+		bo_bucket->restored_offset = KFD_MMAP_TYPE_DOORBELL | KFD_MMAP_GPU_ID(pdd->dev->id);
+	if (bo_bucket->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_MMIO_REMAP) {
+		bo_bucket->restored_offset = KFD_MMAP_TYPE_MMIO | KFD_MMAP_GPU_ID(pdd->dev->id);
+	} else if (bo_bucket->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_GTT) {
+		bo_bucket->restored_offset = offset;
+	} else if (bo_bucket->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM) {
+		bo_bucket->restored_offset = offset;
+		/* Update the VRAM usage count */
+		atomic64_add(bo_bucket->size, &pdd->vram_usage);
+	}
+	return 0;
+}
+
+static int criu_restore_bo(struct kfd_process *p,
+			   struct kfd_criu_bo_bucket *bo_bucket,
+			   struct kfd_criu_bo_priv_data *bo_priv,
+			   struct file **file)
+{
+	struct kfd_process_device *pdd;
+	struct kgd_mem *kgd_mem;
+	int ret;
+	int j;
+
+	pr_debug("Restoring BO size:0x%llx addr:0x%llx gpu_id:0x%x flags:0x%x idr_handle:0x%x\n",
+		 bo_bucket->size, bo_bucket->addr, bo_bucket->gpu_id, bo_bucket->alloc_flags,
+		 bo_priv->idr_handle);
+
+	pdd = kfd_process_device_data_by_id(p, bo_bucket->gpu_id);
+	if (!pdd) {
+		pr_err("Failed to get pdd\n");
+		return -ENODEV;
+	}
+
+	ret = criu_restore_memory_of_gpu(pdd, bo_bucket, bo_priv, &kgd_mem);
+	if (ret)
+		return ret;
+
+	/* now map these BOs to GPU/s */
+	for (j = 0; j < p->n_pdds; j++) {
+		struct kfd_node *peer;
+		struct kfd_process_device *peer_pdd;
+
+		if (!bo_priv->mapped_gpuids[j])
+			break;
+
+		peer_pdd = kfd_process_device_data_by_id(p, bo_priv->mapped_gpuids[j]);
+		if (!peer_pdd)
+			return -EINVAL;
+
+		peer = peer_pdd->dev;
+
+		peer_pdd = kfd_bind_process_to_device(peer, p);
+		if (IS_ERR(peer_pdd))
+			return PTR_ERR(peer_pdd);
+
+		ret = amdgpu_amdkfd_gpuvm_map_memory_to_gpu(peer->adev, kgd_mem,
+							    peer_pdd->drm_priv);
+		if (ret) {
+			pr_err("Failed to map to gpu %d/%d\n", j, p->n_pdds);
+			return ret;
+		}
+	}
+
+	pr_debug("map memory was successful for the BO\n");
+	/* create the dmabuf object and export the bo */
+	if (bo_bucket->alloc_flags
+	    & (KFD_IOC_ALLOC_MEM_FLAGS_VRAM | KFD_IOC_ALLOC_MEM_FLAGS_GTT)) {
+		ret = criu_get_prime_handle(kgd_mem, DRM_RDWR,
+					    &bo_bucket->dmabuf_fd, file);
+		if (ret)
+			return ret;
+	} else {
+		bo_bucket->dmabuf_fd = KFD_INVALID_FD;
+	}
+
+	return 0;
+}
+
+static int criu_restore_bos(struct kfd_process *p,
+			    struct kfd_ioctl_criu_args *args,
+			    uint64_t *priv_offset,
+			    uint64_t max_priv_data_size)
+{
+	struct kfd_criu_bo_bucket *bo_buckets = NULL;
+	struct kfd_criu_bo_priv_data *bo_privs = NULL;
+	struct file **files = NULL;
+	int ret = 0;
+	uint32_t i = 0;
+
+	if (*priv_offset + (args->num_bos * sizeof(*bo_privs)) > max_priv_data_size)
+		return -EINVAL;
+
+	/* Prevent MMU notifications until stage-4 IOCTL (CRIU_RESUME) is received */
+	amdgpu_amdkfd_block_mmu_notifications(p->kgd_process_info);
+
+	bo_buckets = kvmalloc_array(args->num_bos, sizeof(*bo_buckets), GFP_KERNEL);
+	if (!bo_buckets)
+		return -ENOMEM;
+
+	files = kvzalloc(args->num_bos * sizeof(struct file *), GFP_KERNEL);
+	if (!files) {
+		ret = -ENOMEM;
+		goto exit;
+	}
+
+	ret = copy_from_user(bo_buckets, (void __user *)args->bos,
+			     args->num_bos * sizeof(*bo_buckets));
+	if (ret) {
+		pr_err("Failed to copy BOs information from user\n");
+		ret = -EFAULT;
+		goto exit;
+	}
+
+	bo_privs = kvmalloc_array(args->num_bos, sizeof(*bo_privs), GFP_KERNEL);
+	if (!bo_privs) {
+		ret = -ENOMEM;
+		goto exit;
+	}
+
+	ret = copy_from_user(bo_privs, (void __user *)args->priv_data + *priv_offset,
+			     args->num_bos * sizeof(*bo_privs));
+	if (ret) {
+		pr_err("Failed to copy BOs information from user\n");
+		ret = -EFAULT;
+		goto exit;
+	}
+	*priv_offset += args->num_bos * sizeof(*bo_privs);
+
+	/* Create and map new BOs */
+	for (; i < args->num_bos; i++) {
+		ret = criu_restore_bo(p, &bo_buckets[i], &bo_privs[i], &files[i]);
+		if (ret) {
+			pr_debug("Failed to restore BO[%d] ret%d\n", i, ret);
+			goto exit;
+		}
+	} /* done */
+
+	/* Copy only the buckets back so user can read bo_buckets[N].restored_offset */
+	ret = copy_to_user((void __user *)args->bos,
+				bo_buckets,
+				(args->num_bos * sizeof(*bo_buckets)));
+	if (ret)
+		ret = -EFAULT;
+
+exit:
+	commit_files(files, bo_buckets, i, ret);
+	kvfree(files);
+	kvfree(bo_buckets);
+	kvfree(bo_privs);
+	return ret;
+}
+
+static int criu_restore_objects(struct file *filep,
+				struct kfd_process *p,
+				struct kfd_ioctl_criu_args *args,
+				uint64_t *priv_offset,
+				uint64_t max_priv_data_size)
+{
+	int ret = 0;
+	uint32_t i;
+
+	BUILD_BUG_ON(offsetof(struct kfd_criu_queue_priv_data, object_type));
+	BUILD_BUG_ON(offsetof(struct kfd_criu_event_priv_data, object_type));
+	BUILD_BUG_ON(offsetof(struct kfd_criu_svm_range_priv_data, object_type));
+
+	for (i = 0; i < args->num_objects; i++) {
+		uint32_t object_type;
+
+		if (*priv_offset + sizeof(object_type) > max_priv_data_size) {
+			pr_err("Invalid private data size\n");
+			return -EINVAL;
+		}
+
+		ret = get_user(object_type, (uint32_t __user *)(args->priv_data + *priv_offset));
+		if (ret) {
+			pr_err("Failed to copy private information from user\n");
+			goto exit;
+		}
+
+		switch (object_type) {
+		case KFD_CRIU_OBJECT_TYPE_QUEUE:
+			ret = kfd_criu_restore_queue(p, (uint8_t __user *)args->priv_data,
+						     priv_offset, max_priv_data_size);
+			if (ret)
+				goto exit;
+			break;
+		case KFD_CRIU_OBJECT_TYPE_EVENT:
+			ret = kfd_criu_restore_event(filep, p, (uint8_t __user *)args->priv_data,
+						     priv_offset, max_priv_data_size);
+			if (ret)
+				goto exit;
+			break;
+		case KFD_CRIU_OBJECT_TYPE_SVM_RANGE:
+			ret = kfd_criu_restore_svm(p, (uint8_t __user *)args->priv_data,
+						     priv_offset, max_priv_data_size);
+			if (ret)
+				goto exit;
+			break;
+		default:
+			pr_err("Invalid object type:%u at index:%d\n", object_type, i);
+			ret = -EINVAL;
+			goto exit;
+		}
+	}
+exit:
+	return ret;
+}
+
+static int criu_restore(struct file *filep,
+			struct kfd_process *p,
+			struct kfd_ioctl_criu_args *args)
+{
+	uint64_t priv_offset = 0;
+	int ret = 0;
+
+	pr_debug("CRIU restore (num_devices:%u num_bos:%u num_objects:%u priv_data_size:%llu)\n",
+		 args->num_devices, args->num_bos, args->num_objects, args->priv_data_size);
+
+	if ((args->num_bos > 0 && !args->bos) || !args->devices || !args->priv_data ||
+	    !args->priv_data_size || !args->num_devices)
+		return -EINVAL;
+
+	mutex_lock(&p->mutex);
+
+	/*
+	 * Set the process to evicted state to avoid running any new queues before all the memory
+	 * mappings are ready.
+	 */
+	ret = kfd_process_evict_queues(p, KFD_QUEUE_EVICTION_CRIU_RESTORE);
+	if (ret)
+		goto exit_unlock;
+
+	/* Each function will adjust priv_offset based on how many bytes they consumed */
+	ret = criu_restore_process(p, args, &priv_offset, args->priv_data_size);
+	if (ret)
+		goto exit_unlock;
+
+	ret = criu_restore_devices(p, args, &priv_offset, args->priv_data_size);
+	if (ret)
+		goto exit_unlock;
+
+	ret = criu_restore_bos(p, args, &priv_offset, args->priv_data_size);
+	if (ret)
+		goto exit_unlock;
+
+	ret = criu_restore_objects(filep, p, args, &priv_offset, args->priv_data_size);
+	if (ret)
+		goto exit_unlock;
+
+	if (priv_offset != args->priv_data_size) {
+		pr_err("Invalid private data size\n");
+		ret = -EINVAL;
+	}
+
+exit_unlock:
+	mutex_unlock(&p->mutex);
+	if (ret)
+		pr_err("Failed to restore CRIU ret:%d\n", ret);
+	else
+		pr_debug("CRIU restore successful\n");
+
+	return ret;
+}
+
+static int criu_unpause(struct file *filep,
+			struct kfd_process *p,
+			struct kfd_ioctl_criu_args *args)
+{
+	int ret;
+
+	mutex_lock(&p->mutex);
+
+	if (!p->queues_paused) {
+		mutex_unlock(&p->mutex);
+		return -EINVAL;
+	}
+
+	ret = kfd_process_restore_queues(p);
+	if (ret)
+		pr_err("Failed to unpause queues ret:%d\n", ret);
+	else
+		p->queues_paused = false;
+
+	mutex_unlock(&p->mutex);
+
+	return ret;
+}
+
+static int criu_resume(struct file *filep,
+			struct kfd_process *p,
+			struct kfd_ioctl_criu_args *args)
+{
+	struct kfd_process *target = NULL;
+	struct pid *pid = NULL;
+	int ret = 0;
+
+	pr_debug("Inside %s, target pid for criu restore: %d\n", __func__,
+		 args->pid);
+
+	pid = find_get_pid(args->pid);
+	if (!pid) {
+		pr_err("Cannot find pid info for %i\n", args->pid);
+		return -ESRCH;
+	}
+
+	pr_debug("calling kfd_lookup_process_by_pid\n");
+	target = kfd_lookup_process_by_pid(pid);
+
+	put_pid(pid);
+
+	if (!target) {
+		pr_debug("Cannot find process info for %i\n", args->pid);
+		return -ESRCH;
+	}
+
+	mutex_lock(&target->mutex);
+	ret = kfd_criu_resume_svm(target);
+	if (ret) {
+		pr_err("kfd_criu_resume_svm failed for %i\n", args->pid);
+		goto exit;
+	}
+
+	ret =  amdgpu_amdkfd_criu_resume(target->kgd_process_info);
+	if (ret)
+		pr_err("amdgpu_amdkfd_criu_resume failed for %i\n", args->pid);
+
+exit:
+	mutex_unlock(&target->mutex);
+
+	kfd_unref_process(target);
+	return ret;
+}
+
+static int criu_process_info(struct file *filep,
+				struct kfd_process *p,
+				struct kfd_ioctl_criu_args *args)
+{
+	int ret = 0;
+
+	mutex_lock(&p->mutex);
+
+	if (!p->n_pdds) {
+		pr_err("No pdd for given process\n");
+		ret = -ENODEV;
+		goto err_unlock;
+	}
+
+	ret = kfd_process_evict_queues(p, KFD_QUEUE_EVICTION_CRIU_CHECKPOINT);
+	if (ret)
+		goto err_unlock;
+
+	p->queues_paused = true;
+
+	args->pid = task_pid_nr_ns(p->lead_thread,
+					task_active_pid_ns(p->lead_thread));
+
+	ret = criu_get_process_object_info(p, &args->num_devices, &args->num_bos,
+					   &args->num_objects, &args->priv_data_size);
+	if (ret)
+		goto err_unlock;
+
+	dev_dbg(kfd_device, "Num of devices:%u bos:%u objects:%u priv_data_size:%lld\n",
+				args->num_devices, args->num_bos, args->num_objects,
+				args->priv_data_size);
+
+err_unlock:
+	if (ret) {
+		kfd_process_restore_queues(p);
+		p->queues_paused = false;
+	}
+	mutex_unlock(&p->mutex);
+	return ret;
+}
+
+static int kfd_ioctl_criu(struct file *filep, struct kfd_process *p, void *data)
+{
+	struct kfd_ioctl_criu_args *args = data;
+	int ret;
+
+	dev_dbg(kfd_device, "CRIU operation: %d\n", args->op);
+	switch (args->op) {
+	case KFD_CRIU_OP_PROCESS_INFO:
+		ret = criu_process_info(filep, p, args);
+		break;
+	case KFD_CRIU_OP_CHECKPOINT:
+		ret = criu_checkpoint(filep, p, args);
+		break;
+	case KFD_CRIU_OP_UNPAUSE:
+		ret = criu_unpause(filep, p, args);
+		break;
+	case KFD_CRIU_OP_RESTORE:
+		ret = criu_restore(filep, p, args);
+		break;
+	case KFD_CRIU_OP_RESUME:
+		ret = criu_resume(filep, p, args);
+		break;
+	default:
+		dev_dbg(kfd_device, "Unsupported CRIU operation:%d\n", args->op);
+		ret = -EINVAL;
+		break;
+	}
+
+	if (ret)
+		dev_dbg(kfd_device, "CRIU operation:%d err:%d\n", args->op, ret);
+
+	return ret;
+}
+
+static int runtime_enable(struct kfd_process *p, uint64_t r_debug,
+			bool enable_ttmp_setup)
+{
+	int i = 0, ret = 0;
+
+	if (p->is_runtime_retry)
+		goto retry;
+
+	if (p->runtime_info.runtime_state != DEBUG_RUNTIME_STATE_DISABLED)
+		return -EBUSY;
+
+	for (i = 0; i < p->n_pdds; i++) {
+		struct kfd_process_device *pdd = p->pdds[i];
+
+		if (pdd->qpd.queue_count)
+			return -EEXIST;
+
+		/*
+		 * Setup TTMPs by default.
+		 * Note that this call must remain here for MES ADD QUEUE to
+		 * skip_process_ctx_clear unconditionally as the first call to
+		 * SET_SHADER_DEBUGGER clears any stale process context data
+		 * saved in MES.
+		 */
+		if (pdd->dev->kfd->shared_resources.enable_mes)
+			kfd_dbg_set_mes_debug_mode(pdd, !kfd_dbg_has_cwsr_workaround(pdd->dev));
+	}
+
+	p->runtime_info.runtime_state = DEBUG_RUNTIME_STATE_ENABLED;
+	p->runtime_info.r_debug = r_debug;
+	p->runtime_info.ttmp_setup = enable_ttmp_setup;
+
+	if (p->runtime_info.ttmp_setup) {
+		for (i = 0; i < p->n_pdds; i++) {
+			struct kfd_process_device *pdd = p->pdds[i];
+
+			if (!kfd_dbg_is_rlc_restore_supported(pdd->dev)) {
+				amdgpu_gfx_off_ctrl(pdd->dev->adev, false);
+				pdd->dev->kfd2kgd->enable_debug_trap(
+						pdd->dev->adev,
+						true,
+						pdd->dev->vm_info.last_vmid_kfd);
+			} else if (kfd_dbg_is_per_vmid_supported(pdd->dev)) {
+				pdd->spi_dbg_override = pdd->dev->kfd2kgd->enable_debug_trap(
+						pdd->dev->adev,
+						false,
+						0);
+			}
+		}
+	}
+
+retry:
+	if (p->debug_trap_enabled) {
+		if (!p->is_runtime_retry) {
+			kfd_dbg_trap_activate(p);
+			kfd_dbg_ev_raise(KFD_EC_MASK(EC_PROCESS_RUNTIME),
+					p, NULL, 0, false, NULL, 0);
+		}
+
+		mutex_unlock(&p->mutex);
+		ret = down_interruptible(&p->runtime_enable_sema);
+		mutex_lock(&p->mutex);
+
+		p->is_runtime_retry = !!ret;
+	}
+
+	return ret;
+}
+
+static int runtime_disable(struct kfd_process *p)
+{
+	int i = 0, ret;
+	bool was_enabled = p->runtime_info.runtime_state == DEBUG_RUNTIME_STATE_ENABLED;
+
+	p->runtime_info.runtime_state = DEBUG_RUNTIME_STATE_DISABLED;
+	p->runtime_info.r_debug = 0;
+
+	if (p->debug_trap_enabled) {
+		if (was_enabled)
+			kfd_dbg_trap_deactivate(p, false, 0);
+
+		if (!p->is_runtime_retry)
+			kfd_dbg_ev_raise(KFD_EC_MASK(EC_PROCESS_RUNTIME),
+					p, NULL, 0, false, NULL, 0);
+
+		mutex_unlock(&p->mutex);
+		ret = down_interruptible(&p->runtime_enable_sema);
+		mutex_lock(&p->mutex);
+
+		p->is_runtime_retry = !!ret;
+		if (ret)
+			return ret;
+	}
+
+	if (was_enabled && p->runtime_info.ttmp_setup) {
+		for (i = 0; i < p->n_pdds; i++) {
+			struct kfd_process_device *pdd = p->pdds[i];
+
+			if (!kfd_dbg_is_rlc_restore_supported(pdd->dev))
+				amdgpu_gfx_off_ctrl(pdd->dev->adev, true);
+		}
+	}
+
+	p->runtime_info.ttmp_setup = false;
+
+	/* disable ttmp setup */
+	for (i = 0; i < p->n_pdds; i++) {
+		struct kfd_process_device *pdd = p->pdds[i];
+
+		if (kfd_dbg_is_per_vmid_supported(pdd->dev)) {
+			pdd->spi_dbg_override =
+					pdd->dev->kfd2kgd->disable_debug_trap(
+					pdd->dev->adev,
+					false,
+					pdd->dev->vm_info.last_vmid_kfd);
+
+			if (!pdd->dev->kfd->shared_resources.enable_mes)
+				debug_refresh_runlist(pdd->dev->dqm);
+			else
+				kfd_dbg_set_mes_debug_mode(pdd,
+							   !kfd_dbg_has_cwsr_workaround(pdd->dev));
+		}
+	}
+
+	return 0;
+}
+
+static int kfd_ioctl_runtime_enable(struct file *filep, struct kfd_process *p, void *data)
+{
+	struct kfd_ioctl_runtime_enable_args *args = data;
+	int r;
+
+	mutex_lock(&p->mutex);
+
+	if (args->mode_mask & KFD_RUNTIME_ENABLE_MODE_ENABLE_MASK)
+		r = runtime_enable(p, args->r_debug,
+				!!(args->mode_mask & KFD_RUNTIME_ENABLE_MODE_TTMP_SAVE_MASK));
+	else
+		r = runtime_disable(p);
+
+	mutex_unlock(&p->mutex);
+
+	return r;
+}
+
+static int kfd_ioctl_set_debug_trap(struct file *filep, struct kfd_process *p, void *data)
+{
+	struct kfd_ioctl_dbg_trap_args *args = data;
+	struct task_struct *thread = NULL;
+	struct mm_struct *mm = NULL;
+	struct pid *pid = NULL;
+	struct kfd_process *target = NULL;
+	struct kfd_process_device *pdd = NULL;
+	int r = 0;
+
+	if (sched_policy == KFD_SCHED_POLICY_NO_HWS) {
+		pr_err("Debugging does not support sched_policy %i", sched_policy);
+		return -EINVAL;
+	}
+
+	pid = find_get_pid(args->pid);
+	if (!pid) {
+		pr_debug("Cannot find pid info for %i\n", args->pid);
+		r = -ESRCH;
+		goto out;
+	}
+
+	thread = get_pid_task(pid, PIDTYPE_PID);
+	if (!thread) {
+		r = -ESRCH;
+		goto out;
+	}
+
+	mm = get_task_mm(thread);
+	if (!mm) {
+		r = -ESRCH;
+		goto out;
+	}
+
+	if (args->op == KFD_IOC_DBG_TRAP_ENABLE) {
+		bool create_process;
+
+		rcu_read_lock();
+		create_process = thread && thread != current && ptrace_parent(thread) == current;
+		rcu_read_unlock();
+
+		target = create_process ? kfd_create_process(thread) :
+					kfd_lookup_process_by_pid(pid);
+	} else {
+		target = kfd_lookup_process_by_pid(pid);
+	}
+
+	if (IS_ERR_OR_NULL(target)) {
+		pr_debug("Cannot find process PID %i to debug\n", args->pid);
+		r = target ? PTR_ERR(target) : -ESRCH;
+		target = NULL;
+		goto out;
+	}
+
+	/* Check if target is still PTRACED. */
+	rcu_read_lock();
+	if (target != p && args->op != KFD_IOC_DBG_TRAP_DISABLE
+				&& ptrace_parent(target->lead_thread) != current) {
+		pr_err("PID %i is not PTRACED and cannot be debugged\n", args->pid);
+		r = -EPERM;
+	}
+	rcu_read_unlock();
+
+	if (r)
+		goto out;
+
+	mutex_lock(&target->mutex);
+
+	if (args->op != KFD_IOC_DBG_TRAP_ENABLE && !target->debug_trap_enabled) {
+		pr_err("PID %i not debug enabled for op %i\n", args->pid, args->op);
+		r = -EINVAL;
+		goto unlock_out;
+	}
+
+	if (target->runtime_info.runtime_state != DEBUG_RUNTIME_STATE_ENABLED &&
+			(args->op == KFD_IOC_DBG_TRAP_SET_WAVE_LAUNCH_OVERRIDE ||
+			 args->op == KFD_IOC_DBG_TRAP_SET_WAVE_LAUNCH_MODE ||
+			 args->op == KFD_IOC_DBG_TRAP_SUSPEND_QUEUES ||
+			 args->op == KFD_IOC_DBG_TRAP_RESUME_QUEUES ||
+			 args->op == KFD_IOC_DBG_TRAP_SET_NODE_ADDRESS_WATCH ||
+			 args->op == KFD_IOC_DBG_TRAP_CLEAR_NODE_ADDRESS_WATCH ||
+			 args->op == KFD_IOC_DBG_TRAP_SET_FLAGS)) {
+		r = -EPERM;
+		goto unlock_out;
+	}
+
+	if (args->op == KFD_IOC_DBG_TRAP_SET_NODE_ADDRESS_WATCH ||
+	    args->op == KFD_IOC_DBG_TRAP_CLEAR_NODE_ADDRESS_WATCH) {
+		int user_gpu_id = kfd_process_get_user_gpu_id(target,
+				args->op == KFD_IOC_DBG_TRAP_SET_NODE_ADDRESS_WATCH ?
+					args->set_node_address_watch.gpu_id :
+					args->clear_node_address_watch.gpu_id);
+
+		pdd = kfd_process_device_data_by_id(target, user_gpu_id);
+		if (user_gpu_id == -EINVAL || !pdd) {
+			r = -ENODEV;
+			goto unlock_out;
+		}
+	}
+
+	switch (args->op) {
+	case KFD_IOC_DBG_TRAP_ENABLE:
+		if (target != p)
+			target->debugger_process = p;
+
+		r = kfd_dbg_trap_enable(target,
+					args->enable.dbg_fd,
+					(void __user *)args->enable.rinfo_ptr,
+					&args->enable.rinfo_size);
+		if (!r)
+			target->exception_enable_mask = args->enable.exception_mask;
+
+		break;
+	case KFD_IOC_DBG_TRAP_DISABLE:
+		r = kfd_dbg_trap_disable(target);
+		break;
+	case KFD_IOC_DBG_TRAP_SEND_RUNTIME_EVENT:
+		r = kfd_dbg_send_exception_to_runtime(target,
+				args->send_runtime_event.gpu_id,
+				args->send_runtime_event.queue_id,
+				args->send_runtime_event.exception_mask);
+		break;
+	case KFD_IOC_DBG_TRAP_SET_EXCEPTIONS_ENABLED:
+		kfd_dbg_set_enabled_debug_exception_mask(target,
+				args->set_exceptions_enabled.exception_mask);
+		break;
+	case KFD_IOC_DBG_TRAP_SET_WAVE_LAUNCH_OVERRIDE:
+		r = kfd_dbg_trap_set_wave_launch_override(target,
+				args->launch_override.override_mode,
+				args->launch_override.enable_mask,
+				args->launch_override.support_request_mask,
+				&args->launch_override.enable_mask,
+				&args->launch_override.support_request_mask);
+		break;
+	case KFD_IOC_DBG_TRAP_SET_WAVE_LAUNCH_MODE:
+		r = kfd_dbg_trap_set_wave_launch_mode(target,
+				args->launch_mode.launch_mode);
+		break;
+	case KFD_IOC_DBG_TRAP_SUSPEND_QUEUES:
+		r = suspend_queues(target,
+				args->suspend_queues.num_queues,
+				args->suspend_queues.grace_period,
+				args->suspend_queues.exception_mask,
+				(uint32_t *)args->suspend_queues.queue_array_ptr);
+
+		break;
+	case KFD_IOC_DBG_TRAP_RESUME_QUEUES:
+		r = resume_queues(target, args->resume_queues.num_queues,
+				(uint32_t *)args->resume_queues.queue_array_ptr);
+		break;
+	case KFD_IOC_DBG_TRAP_SET_NODE_ADDRESS_WATCH:
+		r = kfd_dbg_trap_set_dev_address_watch(pdd,
+				args->set_node_address_watch.address,
+				args->set_node_address_watch.mask,
+				&args->set_node_address_watch.id,
+				args->set_node_address_watch.mode);
+		break;
+	case KFD_IOC_DBG_TRAP_CLEAR_NODE_ADDRESS_WATCH:
+		r = kfd_dbg_trap_clear_dev_address_watch(pdd,
+				args->clear_node_address_watch.id);
+		break;
+	case KFD_IOC_DBG_TRAP_SET_FLAGS:
+		r = kfd_dbg_trap_set_flags(target, &args->set_flags.flags);
+		break;
+	case KFD_IOC_DBG_TRAP_QUERY_DEBUG_EVENT:
+		r = kfd_dbg_ev_query_debug_event(target,
+				&args->query_debug_event.queue_id,
+				&args->query_debug_event.gpu_id,
+				args->query_debug_event.exception_mask,
+				&args->query_debug_event.exception_mask);
+		break;
+	case KFD_IOC_DBG_TRAP_QUERY_EXCEPTION_INFO:
+		r = kfd_dbg_trap_query_exception_info(target,
+				args->query_exception_info.source_id,
+				args->query_exception_info.exception_code,
+				args->query_exception_info.clear_exception,
+				(void __user *)args->query_exception_info.info_ptr,
+				&args->query_exception_info.info_size);
+		break;
+	case KFD_IOC_DBG_TRAP_GET_QUEUE_SNAPSHOT:
+		r = pqm_get_queue_snapshot(&target->pqm,
+				args->queue_snapshot.exception_mask,
+				(void __user *)args->queue_snapshot.snapshot_buf_ptr,
+				&args->queue_snapshot.num_queues,
+				&args->queue_snapshot.entry_size);
+		break;
+	case KFD_IOC_DBG_TRAP_GET_DEVICE_SNAPSHOT:
+		r = kfd_dbg_trap_device_snapshot(target,
+				args->device_snapshot.exception_mask,
+				(void __user *)args->device_snapshot.snapshot_buf_ptr,
+				&args->device_snapshot.num_devices,
+				&args->device_snapshot.entry_size);
+		break;
+	default:
+		pr_err("Invalid option: %i\n", args->op);
+		r = -EINVAL;
+	}
+
+unlock_out:
+	mutex_unlock(&target->mutex);
+
+out:
+	if (thread)
+		put_task_struct(thread);
+
+	if (mm)
+		mmput(mm);
+
+	if (pid)
+		put_pid(pid);
+
+	if (target)
+		kfd_unref_process(target);
+
+	return r;
+}
+
 #define AMDKFD_IOCTL_DEF(ioctl, _func, _flags) \
 	[_IOC_NR(ioctl)] = {.cmd = ioctl, .func = _func, .flags = _flags, \
 			    .cmd_drv = 0, .name = #ioctl}
@@ -965,23 +3159,82 @@ static const struct amdkfd_ioctl_desc amdkfd_ioctls[] = {
 	AMDKFD_IOCTL_DEF(AMDKFD_IOC_WAIT_EVENTS,
 			kfd_ioctl_wait_events, 0),
 
-	AMDKFD_IOCTL_DEF(AMDKFD_IOC_DBG_REGISTER,
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_DBG_REGISTER_DEPRECATED,
 			kfd_ioctl_dbg_register, 0),
 
-	AMDKFD_IOCTL_DEF(AMDKFD_IOC_DBG_UNREGISTER,
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_DBG_UNREGISTER_DEPRECATED,
 			kfd_ioctl_dbg_unregister, 0),
 
-	AMDKFD_IOCTL_DEF(AMDKFD_IOC_DBG_ADDRESS_WATCH,
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_DBG_ADDRESS_WATCH_DEPRECATED,
 			kfd_ioctl_dbg_address_watch, 0),
 
-	AMDKFD_IOCTL_DEF(AMDKFD_IOC_DBG_WAVE_CONTROL,
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_DBG_WAVE_CONTROL_DEPRECATED,
 			kfd_ioctl_dbg_wave_control, 0),
 
 	AMDKFD_IOCTL_DEF(AMDKFD_IOC_SET_SCRATCH_BACKING_VA,
 			kfd_ioctl_set_scratch_backing_va, 0),
 
 	AMDKFD_IOCTL_DEF(AMDKFD_IOC_GET_TILE_CONFIG,
-			kfd_ioctl_get_tile_config, 0)
+			kfd_ioctl_get_tile_config, 0),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_SET_TRAP_HANDLER,
+			kfd_ioctl_set_trap_handler, 0),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_GET_PROCESS_APERTURES_NEW,
+			kfd_ioctl_get_process_apertures_new, 0),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_ACQUIRE_VM,
+			kfd_ioctl_acquire_vm, 0),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_ALLOC_MEMORY_OF_GPU,
+			kfd_ioctl_alloc_memory_of_gpu, 0),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_FREE_MEMORY_OF_GPU,
+			kfd_ioctl_free_memory_of_gpu, 0),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_MAP_MEMORY_TO_GPU,
+			kfd_ioctl_map_memory_to_gpu, 0),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_UNMAP_MEMORY_FROM_GPU,
+			kfd_ioctl_unmap_memory_from_gpu, 0),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_SET_CU_MASK,
+			kfd_ioctl_set_cu_mask, 0),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_GET_QUEUE_WAVE_STATE,
+			kfd_ioctl_get_queue_wave_state, 0),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_GET_DMABUF_INFO,
+				kfd_ioctl_get_dmabuf_info, 0),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_IMPORT_DMABUF,
+				kfd_ioctl_import_dmabuf, 0),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_ALLOC_QUEUE_GWS,
+			kfd_ioctl_alloc_queue_gws, 0),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_SMI_EVENTS,
+			kfd_ioctl_smi_events, 0),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_SVM, kfd_ioctl_svm, 0),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_SET_XNACK_MODE,
+			kfd_ioctl_set_xnack_mode, 0),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_CRIU_OP,
+			kfd_ioctl_criu, KFD_IOC_FLAG_CHECKPOINT_RESTORE),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_AVAILABLE_MEMORY,
+			kfd_ioctl_get_available_memory, 0),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_EXPORT_DMABUF,
+				kfd_ioctl_export_dmabuf, 0),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_RUNTIME_ENABLE,
+			kfd_ioctl_runtime_enable, 0),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_DBG_TRAP,
+			kfd_ioctl_set_debug_trap, 0),
 };
 
 #define AMDKFD_CORE_IOCTL_COUNT	ARRAY_SIZE(amdkfd_ioctls)
@@ -996,9 +3249,12 @@ static long kfd_ioctl(struct file *filep, unsigned int cmd, unsigned long arg)
 	char *kdata = NULL;
 	unsigned int usize, asize;
 	int retcode = -EINVAL;
+	bool ptrace_attached = false;
 
-	if (nr >= AMDKFD_CORE_IOCTL_COUNT)
+	if (nr >= AMDKFD_CORE_IOCTL_COUNT) {
+		retcode = -ENOTTY;
 		goto err_i1;
+	}
 
 	if ((nr >= AMDKFD_COMMAND_START) && (nr < AMDKFD_COMMAND_END)) {
 		u32 amdkfd_size;
@@ -1011,14 +3267,29 @@ static long kfd_ioctl(struct file *filep, unsigned int cmd, unsigned long arg)
 			asize = amdkfd_size;
 
 		cmd = ioctl->cmd;
-	} else
+	} else {
+		retcode = -ENOTTY;
 		goto err_i1;
+	}
 
-	dev_dbg(kfd_device, "ioctl cmd 0x%x (#%d), arg 0x%lx\n", cmd, nr, arg);
+	dev_dbg(kfd_device, "ioctl cmd 0x%x (#0x%x), arg 0x%lx\n", cmd, nr, arg);
 
-	process = kfd_get_process(current);
-	if (IS_ERR(process)) {
-		dev_dbg(kfd_device, "no process\n");
+	/* Get the process struct from the filep. Only the process
+	 * that opened /dev/kfd can use the file descriptor. Child
+	 * processes need to create their own KFD device context.
+	 */
+	process = filep->private_data;
+
+	rcu_read_lock();
+	if ((ioctl->flags & KFD_IOC_FLAG_CHECKPOINT_RESTORE) &&
+	    ptrace_parent(process->lead_thread) == current)
+		ptrace_attached = true;
+	rcu_read_unlock();
+
+	if (process->lead_thread != current->group_leader
+	    && !ptrace_attached) {
+		dev_dbg(kfd_device, "Using KFD FD in wrong process\n");
+		retcode = -EBADF;
 		goto err_i1;
 	}
 
@@ -1031,6 +3302,19 @@ static long kfd_ioctl(struct file *filep, unsigned int cmd, unsigned long arg)
 		goto err_i1;
 	}
 
+	/*
+	 * Versions of docker shipped in Ubuntu 18.xx and 20.xx do not support
+	 * CAP_CHECKPOINT_RESTORE, so we also allow access if CAP_SYS_ADMIN as CAP_SYS_ADMIN is a
+	 * more priviledged access.
+	 */
+	if (unlikely(ioctl->flags & KFD_IOC_FLAG_CHECKPOINT_RESTORE)) {
+		if (!capable(CAP_CHECKPOINT_RESTORE) &&
+						!capable(CAP_SYS_ADMIN)) {
+			retcode = -EACCES;
+			goto err_i1;
+		}
+	}
+
 	if (cmd & (IOC_IN | IOC_OUT)) {
 		if (asize <= sizeof(stack_kdata)) {
 			kdata = stack_kdata;
@@ -1069,27 +3353,79 @@ err_i1:
 		kfree(kdata);
 
 	if (retcode)
-		dev_dbg(kfd_device, "ret = %d\n", retcode);
+		dev_dbg(kfd_device, "ioctl cmd (#0x%x), arg 0x%lx, ret = %d\n",
+				nr, arg, retcode);
 
 	return retcode;
 }
 
+static int kfd_mmio_mmap(struct kfd_node *dev, struct kfd_process *process,
+		      struct vm_area_struct *vma)
+{
+	phys_addr_t address;
+
+	if (vma->vm_end - vma->vm_start != PAGE_SIZE)
+		return -EINVAL;
+
+	if (PAGE_SIZE > 4096)
+		return -EINVAL;
+
+	address = dev->adev->rmmio_remap.bus_addr;
+
+	vm_flags_set(vma, VM_IO | VM_DONTCOPY | VM_DONTEXPAND | VM_NORESERVE |
+				VM_DONTDUMP | VM_PFNMAP);
+
+	vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
+
+	pr_debug("process pid %d mapping mmio page\n"
+		 "     target user address == 0x%08llX\n"
+		 "     physical address    == 0x%08llX\n"
+		 "     vm_flags            == 0x%04lX\n"
+		 "     size                == 0x%04lX\n",
+		 process->lead_thread->pid, (unsigned long long) vma->vm_start,
+		 address, vma->vm_flags, PAGE_SIZE);
+
+	return io_remap_pfn_range(vma,
+				vma->vm_start,
+				address >> PAGE_SHIFT,
+				PAGE_SIZE,
+				vma->vm_page_prot);
+}
+
+
 static int kfd_mmap(struct file *filp, struct vm_area_struct *vma)
 {
 	struct kfd_process *process;
+	struct kfd_node *dev = NULL;
+	unsigned long mmap_offset;
+	unsigned int gpu_id;
 
 	process = kfd_get_process(current);
 	if (IS_ERR(process))
 		return PTR_ERR(process);
 
-	if ((vma->vm_pgoff & KFD_MMAP_DOORBELL_MASK) ==
-			KFD_MMAP_DOORBELL_MASK) {
-		vma->vm_pgoff = vma->vm_pgoff ^ KFD_MMAP_DOORBELL_MASK;
-		return kfd_doorbell_mmap(process, vma);
-	} else if ((vma->vm_pgoff & KFD_MMAP_EVENTS_MASK) ==
-			KFD_MMAP_EVENTS_MASK) {
-		vma->vm_pgoff = vma->vm_pgoff ^ KFD_MMAP_EVENTS_MASK;
+	mmap_offset = vma->vm_pgoff << PAGE_SHIFT;
+	gpu_id = KFD_MMAP_GET_GPU_ID(mmap_offset);
+	if (gpu_id)
+		dev = kfd_device_by_id(gpu_id);
+
+	switch (mmap_offset & KFD_MMAP_TYPE_MASK) {
+	case KFD_MMAP_TYPE_DOORBELL:
+		if (!dev)
+			return -ENODEV;
+		return kfd_doorbell_mmap(dev, process, vma);
+
+	case KFD_MMAP_TYPE_EVENTS:
 		return kfd_event_mmap(process, vma);
+
+	case KFD_MMAP_TYPE_RESERVED_MEM:
+		if (!dev)
+			return -ENODEV;
+		return kfd_reserved_mem_mmap(dev, process, vma);
+	case KFD_MMAP_TYPE_MMIO:
+		if (!dev)
+			return -ENODEV;
+		return kfd_mmio_mmap(dev, process, vma);
 	}
 
 	return -EFAULT;
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_crat.c b/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
new file mode 100644
index 000000000000..4a7180b46b71
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
@@ -0,0 +1,2470 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
+/*
+ * Copyright 2015-2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include <linux/pci.h>
+#include <linux/acpi.h>
+#include "kfd_crat.h"
+#include "kfd_priv.h"
+#include "kfd_topology.h"
+#include "amdgpu.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_xgmi.h"
+
+/* GPU Processor ID base for dGPUs for which VCRAT needs to be created.
+ * GPU processor ID are expressed with Bit[31]=1.
+ * The base is set to 0x8000_0000 + 0x1000 to avoid collision with GPU IDs
+ * used in the CRAT.
+ */
+static uint32_t gpu_processor_id_low = 0x80001000;
+
+/* Return the next available gpu_processor_id and increment it for next GPU
+ *	@total_cu_count - Total CUs present in the GPU including ones
+ *			  masked off
+ */
+static inline unsigned int get_and_inc_gpu_processor_id(
+				unsigned int total_cu_count)
+{
+	int current_id = gpu_processor_id_low;
+
+	gpu_processor_id_low += total_cu_count;
+	return current_id;
+}
+
+
+static struct kfd_gpu_cache_info kaveri_cache_info[] = {
+	{
+		/* TCP L1 Cache per CU */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 1,
+	},
+	{
+		/* Scalar L1 Instruction Cache (in SQC module) per bank */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_INST_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* Scalar L1 Data Cache (in SQC module) per bank */
+		.cache_size = 8,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+
+	/* TODO: Add L2 Cache information */
+};
+
+
+static struct kfd_gpu_cache_info carrizo_cache_info[] = {
+	{
+		/* TCP L1 Cache per CU */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 1,
+	},
+	{
+		/* Scalar L1 Instruction Cache (in SQC module) per bank */
+		.cache_size = 32,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_INST_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 4,
+	},
+	{
+		/* Scalar L1 Data Cache (in SQC module) per bank. */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 4,
+	},
+
+	/* TODO: Add L2 Cache information */
+};
+
+#define hawaii_cache_info kaveri_cache_info
+#define tonga_cache_info carrizo_cache_info
+#define fiji_cache_info  carrizo_cache_info
+#define polaris10_cache_info carrizo_cache_info
+#define polaris11_cache_info carrizo_cache_info
+#define polaris12_cache_info carrizo_cache_info
+#define vegam_cache_info carrizo_cache_info
+
+/* NOTE: L1 cache information has been updated and L2/L3
+ * cache information has been added for Vega10 and
+ * newer ASICs. The unit for cache_size is KiB.
+ * In future,  check & update cache details
+ * for every new ASIC is required.
+ */
+
+static struct kfd_gpu_cache_info vega10_cache_info[] = {
+	{
+		/* TCP L1 Cache per CU */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 1,
+	},
+	{
+		/* Scalar L1 Instruction Cache per SQC */
+		.cache_size = 32,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_INST_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 3,
+	},
+	{
+		/* Scalar L1 Data Cache per SQC */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 3,
+	},
+	{
+		/* L2 Data Cache per GPU (Total Tex Cache) */
+		.cache_size = 4096,
+		.cache_level = 2,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 16,
+	},
+};
+
+static struct kfd_gpu_cache_info raven_cache_info[] = {
+	{
+		/* TCP L1 Cache per CU */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 1,
+	},
+	{
+		/* Scalar L1 Instruction Cache per SQC */
+		.cache_size = 32,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_INST_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 3,
+	},
+	{
+		/* Scalar L1 Data Cache per SQC */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 3,
+	},
+	{
+		/* L2 Data Cache per GPU (Total Tex Cache) */
+		.cache_size = 1024,
+		.cache_level = 2,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 11,
+	},
+};
+
+static struct kfd_gpu_cache_info renoir_cache_info[] = {
+	{
+		/* TCP L1 Cache per CU */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 1,
+	},
+	{
+		/* Scalar L1 Instruction Cache per SQC */
+		.cache_size = 32,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_INST_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 3,
+	},
+	{
+		/* Scalar L1 Data Cache per SQC */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 3,
+	},
+	{
+		/* L2 Data Cache per GPU (Total Tex Cache) */
+		.cache_size = 1024,
+		.cache_level = 2,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 8,
+	},
+};
+
+static struct kfd_gpu_cache_info vega12_cache_info[] = {
+	{
+		/* TCP L1 Cache per CU */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 1,
+	},
+	{
+		/* Scalar L1 Instruction Cache per SQC */
+		.cache_size = 32,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_INST_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 3,
+	},
+	{
+		/* Scalar L1 Data Cache per SQC */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 3,
+	},
+	{
+		/* L2 Data Cache per GPU (Total Tex Cache) */
+		.cache_size = 2048,
+		.cache_level = 2,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 5,
+	},
+};
+
+static struct kfd_gpu_cache_info vega20_cache_info[] = {
+	{
+		/* TCP L1 Cache per CU */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 1,
+	},
+	{
+		/* Scalar L1 Instruction Cache per SQC */
+		.cache_size = 32,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_INST_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 3,
+	},
+	{
+		/* Scalar L1 Data Cache per SQC */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 3,
+	},
+	{
+		/* L2 Data Cache per GPU (Total Tex Cache) */
+		.cache_size = 8192,
+		.cache_level = 2,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 16,
+	},
+};
+
+static struct kfd_gpu_cache_info aldebaran_cache_info[] = {
+	{
+		/* TCP L1 Cache per CU */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 1,
+	},
+	{
+		/* Scalar L1 Instruction Cache per SQC */
+		.cache_size = 32,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_INST_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* Scalar L1 Data Cache per SQC */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* L2 Data Cache per GPU (Total Tex Cache) */
+		.cache_size = 8192,
+		.cache_level = 2,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 14,
+	},
+};
+
+static struct kfd_gpu_cache_info navi10_cache_info[] = {
+	{
+		/* TCP L1 Cache per CU */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 1,
+	},
+	{
+		/* Scalar L1 Instruction Cache per SQC */
+		.cache_size = 32,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_INST_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* Scalar L1 Data Cache per SQC */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* GL1 Data Cache per SA */
+		.cache_size = 128,
+		.cache_level = 1,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 10,
+	},
+	{
+		/* L2 Data Cache per GPU (Total Tex Cache) */
+		.cache_size = 4096,
+		.cache_level = 2,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 10,
+	},
+};
+
+static struct kfd_gpu_cache_info vangogh_cache_info[] = {
+	{
+		/* TCP L1 Cache per CU */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 1,
+	},
+	{
+		/* Scalar L1 Instruction Cache per SQC */
+		.cache_size = 32,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_INST_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* Scalar L1 Data Cache per SQC */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* GL1 Data Cache per SA */
+		.cache_size = 128,
+		.cache_level = 1,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 8,
+	},
+	{
+		/* L2 Data Cache per GPU (Total Tex Cache) */
+		.cache_size = 1024,
+		.cache_level = 2,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 8,
+	},
+};
+
+static struct kfd_gpu_cache_info navi14_cache_info[] = {
+	{
+		/* TCP L1 Cache per CU */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 1,
+	},
+	{
+		/* Scalar L1 Instruction Cache per SQC */
+		.cache_size = 32,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_INST_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* Scalar L1 Data Cache per SQC */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* GL1 Data Cache per SA */
+		.cache_size = 128,
+		.cache_level = 1,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 12,
+	},
+	{
+		/* L2 Data Cache per GPU (Total Tex Cache) */
+		.cache_size = 2048,
+		.cache_level = 2,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 12,
+	},
+};
+
+static struct kfd_gpu_cache_info sienna_cichlid_cache_info[] = {
+	{
+		/* TCP L1 Cache per CU */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 1,
+	},
+	{
+		/* Scalar L1 Instruction Cache per SQC */
+		.cache_size = 32,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_INST_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* Scalar L1 Data Cache per SQC */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* GL1 Data Cache per SA */
+		.cache_size = 128,
+		.cache_level = 1,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 10,
+	},
+	{
+		/* L2 Data Cache per GPU (Total Tex Cache) */
+		.cache_size = 4096,
+		.cache_level = 2,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 10,
+	},
+	{
+		/* L3 Data Cache per GPU */
+		.cache_size = 128*1024,
+		.cache_level = 3,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 10,
+	},
+};
+
+static struct kfd_gpu_cache_info navy_flounder_cache_info[] = {
+	{
+		/* TCP L1 Cache per CU */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 1,
+	},
+	{
+		/* Scalar L1 Instruction Cache per SQC */
+		.cache_size = 32,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_INST_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* Scalar L1 Data Cache per SQC */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* GL1 Data Cache per SA */
+		.cache_size = 128,
+		.cache_level = 1,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 10,
+	},
+	{
+		/* L2 Data Cache per GPU (Total Tex Cache) */
+		.cache_size = 3072,
+		.cache_level = 2,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 10,
+	},
+	{
+		/* L3 Data Cache per GPU */
+		.cache_size = 96*1024,
+		.cache_level = 3,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 10,
+	},
+};
+
+static struct kfd_gpu_cache_info dimgrey_cavefish_cache_info[] = {
+	{
+		/* TCP L1 Cache per CU */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 1,
+	},
+	{
+		/* Scalar L1 Instruction Cache per SQC */
+		.cache_size = 32,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_INST_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* Scalar L1 Data Cache per SQC */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* GL1 Data Cache per SA */
+		.cache_size = 128,
+		.cache_level = 1,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 8,
+	},
+	{
+		/* L2 Data Cache per GPU (Total Tex Cache) */
+		.cache_size = 2048,
+		.cache_level = 2,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 8,
+	},
+	{
+		/* L3 Data Cache per GPU */
+		.cache_size = 32*1024,
+		.cache_level = 3,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 8,
+	},
+};
+
+static struct kfd_gpu_cache_info beige_goby_cache_info[] = {
+	{
+		/* TCP L1 Cache per CU */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 1,
+	},
+	{
+		/* Scalar L1 Instruction Cache per SQC */
+		.cache_size = 32,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_INST_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* Scalar L1 Data Cache per SQC */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* GL1 Data Cache per SA */
+		.cache_size = 128,
+		.cache_level = 1,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 8,
+	},
+	{
+		/* L2 Data Cache per GPU (Total Tex Cache) */
+		.cache_size = 1024,
+		.cache_level = 2,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 8,
+	},
+	{
+		/* L3 Data Cache per GPU */
+		.cache_size = 16*1024,
+		.cache_level = 3,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 8,
+	},
+};
+
+static struct kfd_gpu_cache_info yellow_carp_cache_info[] = {
+	{
+		/* TCP L1 Cache per CU */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 1,
+	},
+	{
+		/* Scalar L1 Instruction Cache per SQC */
+		.cache_size = 32,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_INST_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* Scalar L1 Data Cache per SQC */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* GL1 Data Cache per SA */
+		.cache_size = 128,
+		.cache_level = 1,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 6,
+	},
+	{
+		/* L2 Data Cache per GPU (Total Tex Cache) */
+		.cache_size = 2048,
+		.cache_level = 2,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 6,
+	},
+};
+
+static struct kfd_gpu_cache_info gfx1037_cache_info[] = {
+	{
+		/* TCP L1 Cache per CU */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 1,
+	},
+	{
+		/* Scalar L1 Instruction Cache per SQC */
+		.cache_size = 32,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_INST_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* Scalar L1 Data Cache per SQC */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* GL1 Data Cache per SA */
+		.cache_size = 128,
+		.cache_level = 1,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* L2 Data Cache per GPU (Total Tex Cache) */
+		.cache_size = 256,
+		.cache_level = 2,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+};
+
+static struct kfd_gpu_cache_info gc_10_3_6_cache_info[] = {
+	{
+		/* TCP L1 Cache per CU */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+			  CRAT_CACHE_FLAGS_DATA_CACHE |
+			  CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 1,
+	},
+	{
+		/* Scalar L1 Instruction Cache per SQC */
+		.cache_size = 32,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+			  CRAT_CACHE_FLAGS_INST_CACHE |
+			  CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* Scalar L1 Data Cache per SQC */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+			  CRAT_CACHE_FLAGS_DATA_CACHE |
+			  CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* GL1 Data Cache per SA */
+		.cache_size = 128,
+		.cache_level = 1,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+			  CRAT_CACHE_FLAGS_DATA_CACHE |
+			  CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* L2 Data Cache per GPU (Total Tex Cache) */
+		.cache_size = 256,
+		.cache_level = 2,
+		.cache_line_size = 128,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+			  CRAT_CACHE_FLAGS_DATA_CACHE |
+			  CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+};
+
+static struct kfd_gpu_cache_info dummy_cache_info[] = {
+	{
+		/* TCP L1 Cache per CU */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 1,
+	},
+	{
+		/* Scalar L1 Instruction Cache per SQC */
+		.cache_size = 32,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_INST_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* Scalar L1 Data Cache per SQC */
+		.cache_size = 16,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 2,
+	},
+	{
+		/* GL1 Data Cache per SA */
+		.cache_size = 128,
+		.cache_level = 1,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 6,
+	},
+	{
+		/* L2 Data Cache per GPU (Total Tex Cache) */
+		.cache_size = 2048,
+		.cache_level = 2,
+		.cache_line_size = 64,
+		.flags = (CRAT_CACHE_FLAGS_ENABLED |
+				CRAT_CACHE_FLAGS_DATA_CACHE |
+				CRAT_CACHE_FLAGS_SIMD_CACHE),
+		.num_cu_shared = 6,
+	},
+};
+
+static void kfd_populated_cu_info_cpu(struct kfd_topology_device *dev,
+		struct crat_subtype_computeunit *cu)
+{
+	dev->node_props.cpu_cores_count = cu->num_cpu_cores;
+	dev->node_props.cpu_core_id_base = cu->processor_id_low;
+	if (cu->hsa_capability & CRAT_CU_FLAGS_IOMMU_PRESENT)
+		dev->node_props.capability |= HSA_CAP_ATS_PRESENT;
+
+	pr_debug("CU CPU: cores=%d id_base=%d\n", cu->num_cpu_cores,
+			cu->processor_id_low);
+}
+
+static void kfd_populated_cu_info_gpu(struct kfd_topology_device *dev,
+		struct crat_subtype_computeunit *cu)
+{
+	dev->node_props.simd_id_base = cu->processor_id_low;
+	dev->node_props.simd_count = cu->num_simd_cores;
+	dev->node_props.lds_size_in_kb = cu->lds_size_in_kb;
+	dev->node_props.max_waves_per_simd = cu->max_waves_simd;
+	dev->node_props.wave_front_size = cu->wave_front_size;
+	dev->node_props.array_count = cu->array_count;
+	dev->node_props.cu_per_simd_array = cu->num_cu_per_array;
+	dev->node_props.simd_per_cu = cu->num_simd_per_cu;
+	dev->node_props.max_slots_scratch_cu = cu->max_slots_scatch_cu;
+	if (cu->hsa_capability & CRAT_CU_FLAGS_HOT_PLUGGABLE)
+		dev->node_props.capability |= HSA_CAP_HOT_PLUGGABLE;
+	pr_debug("CU GPU: id_base=%d\n", cu->processor_id_low);
+}
+
+/* kfd_parse_subtype_cu - parse compute unit subtypes and attach it to correct
+ * topology device present in the device_list
+ */
+static int kfd_parse_subtype_cu(struct crat_subtype_computeunit *cu,
+				struct list_head *device_list)
+{
+	struct kfd_topology_device *dev;
+
+	pr_debug("Found CU entry in CRAT table with proximity_domain=%d caps=%x\n",
+			cu->proximity_domain, cu->hsa_capability);
+	list_for_each_entry(dev, device_list, list) {
+		if (cu->proximity_domain == dev->proximity_domain) {
+			if (cu->flags & CRAT_CU_FLAGS_CPU_PRESENT)
+				kfd_populated_cu_info_cpu(dev, cu);
+
+			if (cu->flags & CRAT_CU_FLAGS_GPU_PRESENT)
+				kfd_populated_cu_info_gpu(dev, cu);
+			break;
+		}
+	}
+
+	return 0;
+}
+
+static struct kfd_mem_properties *
+find_subtype_mem(uint32_t heap_type, uint32_t flags, uint32_t width,
+		struct kfd_topology_device *dev)
+{
+	struct kfd_mem_properties *props;
+
+	list_for_each_entry(props, &dev->mem_props, list) {
+		if (props->heap_type == heap_type
+				&& props->flags == flags
+				&& props->width == width)
+			return props;
+	}
+
+	return NULL;
+}
+/* kfd_parse_subtype_mem - parse memory subtypes and attach it to correct
+ * topology device present in the device_list
+ */
+static int kfd_parse_subtype_mem(struct crat_subtype_memory *mem,
+				struct list_head *device_list)
+{
+	struct kfd_mem_properties *props;
+	struct kfd_topology_device *dev;
+	uint32_t heap_type;
+	uint64_t size_in_bytes;
+	uint32_t flags = 0;
+	uint32_t width;
+
+	pr_debug("Found memory entry in CRAT table with proximity_domain=%d\n",
+			mem->proximity_domain);
+	list_for_each_entry(dev, device_list, list) {
+		if (mem->proximity_domain == dev->proximity_domain) {
+			/* We're on GPU node */
+			if (dev->node_props.cpu_cores_count == 0) {
+				/* APU */
+				if (mem->visibility_type == 0)
+					heap_type =
+						HSA_MEM_HEAP_TYPE_FB_PRIVATE;
+				/* dGPU */
+				else
+					heap_type = mem->visibility_type;
+			} else
+				heap_type = HSA_MEM_HEAP_TYPE_SYSTEM;
+
+			if (mem->flags & CRAT_MEM_FLAGS_HOT_PLUGGABLE)
+				flags |= HSA_MEM_FLAGS_HOT_PLUGGABLE;
+			if (mem->flags & CRAT_MEM_FLAGS_NON_VOLATILE)
+				flags |= HSA_MEM_FLAGS_NON_VOLATILE;
+
+			size_in_bytes =
+				((uint64_t)mem->length_high << 32) +
+							mem->length_low;
+			width = mem->width;
+
+			/* Multiple banks of the same type are aggregated into
+			 * one. User mode doesn't care about multiple physical
+			 * memory segments. It's managed as a single virtual
+			 * heap for user mode.
+			 */
+			props = find_subtype_mem(heap_type, flags, width, dev);
+			if (props) {
+				props->size_in_bytes += size_in_bytes;
+				break;
+			}
+
+			props = kfd_alloc_struct(props);
+			if (!props)
+				return -ENOMEM;
+
+			props->heap_type = heap_type;
+			props->flags = flags;
+			props->size_in_bytes = size_in_bytes;
+			props->width = width;
+
+			dev->node_props.mem_banks_count++;
+			list_add_tail(&props->list, &dev->mem_props);
+
+			break;
+		}
+	}
+
+	return 0;
+}
+
+/* kfd_parse_subtype_cache - parse cache subtypes and attach it to correct
+ * topology device present in the device_list
+ */
+static int kfd_parse_subtype_cache(struct crat_subtype_cache *cache,
+			struct list_head *device_list)
+{
+	struct kfd_cache_properties *props;
+	struct kfd_topology_device *dev;
+	uint32_t id;
+	uint32_t total_num_of_cu;
+
+	id = cache->processor_id_low;
+
+	pr_debug("Found cache entry in CRAT table with processor_id=%d\n", id);
+	list_for_each_entry(dev, device_list, list) {
+		total_num_of_cu = (dev->node_props.array_count *
+					dev->node_props.cu_per_simd_array);
+
+		/* Cache infomration in CRAT doesn't have proximity_domain
+		 * information as it is associated with a CPU core or GPU
+		 * Compute Unit. So map the cache using CPU core Id or SIMD
+		 * (GPU) ID.
+		 * TODO: This works because currently we can safely assume that
+		 *  Compute Units are parsed before caches are parsed. In
+		 *  future, remove this dependency
+		 */
+		if ((id >= dev->node_props.cpu_core_id_base &&
+			id <= dev->node_props.cpu_core_id_base +
+				dev->node_props.cpu_cores_count) ||
+			(id >= dev->node_props.simd_id_base &&
+			id < dev->node_props.simd_id_base +
+				total_num_of_cu)) {
+			props = kfd_alloc_struct(props);
+			if (!props)
+				return -ENOMEM;
+
+			props->processor_id_low = id;
+			props->cache_level = cache->cache_level;
+			props->cache_size = cache->cache_size;
+			props->cacheline_size = cache->cache_line_size;
+			props->cachelines_per_tag = cache->lines_per_tag;
+			props->cache_assoc = cache->associativity;
+			props->cache_latency = cache->cache_latency;
+
+			memcpy(props->sibling_map, cache->sibling_map,
+					CRAT_SIBLINGMAP_SIZE);
+
+			/* set the sibling_map_size as 32 for CRAT from ACPI */
+			props->sibling_map_size = CRAT_SIBLINGMAP_SIZE;
+
+			if (cache->flags & CRAT_CACHE_FLAGS_DATA_CACHE)
+				props->cache_type |= HSA_CACHE_TYPE_DATA;
+			if (cache->flags & CRAT_CACHE_FLAGS_INST_CACHE)
+				props->cache_type |= HSA_CACHE_TYPE_INSTRUCTION;
+			if (cache->flags & CRAT_CACHE_FLAGS_CPU_CACHE)
+				props->cache_type |= HSA_CACHE_TYPE_CPU;
+			if (cache->flags & CRAT_CACHE_FLAGS_SIMD_CACHE)
+				props->cache_type |= HSA_CACHE_TYPE_HSACU;
+
+			dev->node_props.caches_count++;
+			list_add_tail(&props->list, &dev->cache_props);
+
+			break;
+		}
+	}
+
+	return 0;
+}
+
+/* kfd_parse_subtype_iolink - parse iolink subtypes and attach it to correct
+ * topology device present in the device_list
+ */
+static int kfd_parse_subtype_iolink(struct crat_subtype_iolink *iolink,
+					struct list_head *device_list)
+{
+	struct kfd_iolink_properties *props = NULL, *props2;
+	struct kfd_topology_device *dev, *to_dev;
+	uint32_t id_from;
+	uint32_t id_to;
+
+	id_from = iolink->proximity_domain_from;
+	id_to = iolink->proximity_domain_to;
+
+	pr_debug("Found IO link entry in CRAT table with id_from=%d, id_to %d\n",
+			id_from, id_to);
+	list_for_each_entry(dev, device_list, list) {
+		if (id_from == dev->proximity_domain) {
+			props = kfd_alloc_struct(props);
+			if (!props)
+				return -ENOMEM;
+
+			props->node_from = id_from;
+			props->node_to = id_to;
+			props->ver_maj = iolink->version_major;
+			props->ver_min = iolink->version_minor;
+			props->iolink_type = iolink->io_interface_type;
+
+			if (props->iolink_type == CRAT_IOLINK_TYPE_PCIEXPRESS)
+				props->weight = 20;
+			else if (props->iolink_type == CRAT_IOLINK_TYPE_XGMI)
+				props->weight = iolink->weight_xgmi;
+			else
+				props->weight = node_distance(id_from, id_to);
+
+			props->min_latency = iolink->minimum_latency;
+			props->max_latency = iolink->maximum_latency;
+			props->min_bandwidth = iolink->minimum_bandwidth_mbs;
+			props->max_bandwidth = iolink->maximum_bandwidth_mbs;
+			props->rec_transfer_size =
+					iolink->recommended_transfer_size;
+
+			dev->node_props.io_links_count++;
+			list_add_tail(&props->list, &dev->io_link_props);
+			break;
+		}
+	}
+
+	/* CPU topology is created before GPUs are detected, so CPU->GPU
+	 * links are not built at that time. If a PCIe type is discovered, it
+	 * means a GPU is detected and we are adding GPU->CPU to the topology.
+	 * At this time, also add the corresponded CPU->GPU link if GPU
+	 * is large bar.
+	 * For xGMI, we only added the link with one direction in the crat
+	 * table, add corresponded reversed direction link now.
+	 */
+	if (props && (iolink->flags & CRAT_IOLINK_FLAGS_BI_DIRECTIONAL)) {
+		to_dev = kfd_topology_device_by_proximity_domain_no_lock(id_to);
+		if (!to_dev)
+			return -ENODEV;
+		/* same everything but the other direction */
+		props2 = kmemdup(props, sizeof(*props2), GFP_KERNEL);
+		if (!props2)
+			return -ENOMEM;
+
+		props2->node_from = id_to;
+		props2->node_to = id_from;
+		props2->kobj = NULL;
+		to_dev->node_props.io_links_count++;
+		list_add_tail(&props2->list, &to_dev->io_link_props);
+	}
+
+	return 0;
+}
+
+/* kfd_parse_subtype - parse subtypes and attach it to correct topology device
+ * present in the device_list
+ *	@sub_type_hdr - subtype section of crat_image
+ *	@device_list - list of topology devices present in this crat_image
+ */
+static int kfd_parse_subtype(struct crat_subtype_generic *sub_type_hdr,
+				struct list_head *device_list)
+{
+	struct crat_subtype_computeunit *cu;
+	struct crat_subtype_memory *mem;
+	struct crat_subtype_cache *cache;
+	struct crat_subtype_iolink *iolink;
+	int ret = 0;
+
+	switch (sub_type_hdr->type) {
+	case CRAT_SUBTYPE_COMPUTEUNIT_AFFINITY:
+		cu = (struct crat_subtype_computeunit *)sub_type_hdr;
+		ret = kfd_parse_subtype_cu(cu, device_list);
+		break;
+	case CRAT_SUBTYPE_MEMORY_AFFINITY:
+		mem = (struct crat_subtype_memory *)sub_type_hdr;
+		ret = kfd_parse_subtype_mem(mem, device_list);
+		break;
+	case CRAT_SUBTYPE_CACHE_AFFINITY:
+		cache = (struct crat_subtype_cache *)sub_type_hdr;
+		ret = kfd_parse_subtype_cache(cache, device_list);
+		break;
+	case CRAT_SUBTYPE_TLB_AFFINITY:
+		/*
+		 * For now, nothing to do here
+		 */
+		pr_debug("Found TLB entry in CRAT table (not processing)\n");
+		break;
+	case CRAT_SUBTYPE_CCOMPUTE_AFFINITY:
+		/*
+		 * For now, nothing to do here
+		 */
+		pr_debug("Found CCOMPUTE entry in CRAT table (not processing)\n");
+		break;
+	case CRAT_SUBTYPE_IOLINK_AFFINITY:
+		iolink = (struct crat_subtype_iolink *)sub_type_hdr;
+		ret = kfd_parse_subtype_iolink(iolink, device_list);
+		break;
+	default:
+		pr_warn("Unknown subtype %d in CRAT\n",
+				sub_type_hdr->type);
+	}
+
+	return ret;
+}
+
+/* kfd_parse_crat_table - parse CRAT table. For each node present in CRAT
+ * create a kfd_topology_device and add in to device_list. Also parse
+ * CRAT subtypes and attach it to appropriate kfd_topology_device
+ *	@crat_image - input image containing CRAT
+ *	@device_list - [OUT] list of kfd_topology_device generated after
+ *		       parsing crat_image
+ *	@proximity_domain - Proximity domain of the first device in the table
+ *
+ *	Return - 0 if successful else -ve value
+ */
+int kfd_parse_crat_table(void *crat_image, struct list_head *device_list,
+			 uint32_t proximity_domain)
+{
+	struct kfd_topology_device *top_dev = NULL;
+	struct crat_subtype_generic *sub_type_hdr;
+	uint16_t node_id;
+	int ret = 0;
+	struct crat_header *crat_table = (struct crat_header *)crat_image;
+	uint16_t num_nodes;
+	uint32_t image_len;
+
+	if (!crat_image)
+		return -EINVAL;
+
+	if (!list_empty(device_list)) {
+		pr_warn("Error device list should be empty\n");
+		return -EINVAL;
+	}
+
+	num_nodes = crat_table->num_domains;
+	image_len = crat_table->length;
+
+	pr_debug("Parsing CRAT table with %d nodes\n", num_nodes);
+
+	for (node_id = 0; node_id < num_nodes; node_id++) {
+		top_dev = kfd_create_topology_device(device_list);
+		if (!top_dev)
+			break;
+		top_dev->proximity_domain = proximity_domain++;
+	}
+
+	if (!top_dev) {
+		ret = -ENOMEM;
+		goto err;
+	}
+
+	memcpy(top_dev->oem_id, crat_table->oem_id, CRAT_OEMID_LENGTH);
+	memcpy(top_dev->oem_table_id, crat_table->oem_table_id,
+			CRAT_OEMTABLEID_LENGTH);
+	top_dev->oem_revision = crat_table->oem_revision;
+
+	sub_type_hdr = (struct crat_subtype_generic *)(crat_table+1);
+	while ((char *)sub_type_hdr + sizeof(struct crat_subtype_generic) <
+			((char *)crat_image) + image_len) {
+		if (sub_type_hdr->flags & CRAT_SUBTYPE_FLAGS_ENABLED) {
+			ret = kfd_parse_subtype(sub_type_hdr, device_list);
+			if (ret)
+				break;
+		}
+
+		sub_type_hdr = (typeof(sub_type_hdr))((char *)sub_type_hdr +
+				sub_type_hdr->length);
+	}
+
+err:
+	if (ret)
+		kfd_release_topology_device_list(device_list);
+
+	return ret;
+}
+
+
+static int kfd_fill_gpu_cache_info_from_gfx_config(struct kfd_dev *kdev,
+						   bool cache_line_size_missing,
+						   struct kfd_gpu_cache_info *pcache_info)
+{
+	struct amdgpu_device *adev = kdev->adev;
+	int i = 0;
+
+	/* TCP L1 Cache per CU */
+	if (adev->gfx.config.gc_tcp_l1_size) {
+		pcache_info[i].cache_size = adev->gfx.config.gc_tcp_l1_size;
+		pcache_info[i].cache_level = 1;
+		pcache_info[i].flags = (CRAT_CACHE_FLAGS_ENABLED |
+					CRAT_CACHE_FLAGS_DATA_CACHE |
+					CRAT_CACHE_FLAGS_SIMD_CACHE);
+		pcache_info[i].num_cu_shared = adev->gfx.config.gc_num_tcp_per_wpg / 2;
+		pcache_info[i].cache_line_size = adev->gfx.config.gc_tcp_cache_line_size;
+		if (cache_line_size_missing && !pcache_info[i].cache_line_size)
+			pcache_info[i].cache_line_size = 128;
+		i++;
+	}
+	/* Scalar L1 Instruction Cache per SQC */
+	if (adev->gfx.config.gc_l1_instruction_cache_size_per_sqc) {
+		pcache_info[i].cache_size =
+			adev->gfx.config.gc_l1_instruction_cache_size_per_sqc;
+		pcache_info[i].cache_level = 1;
+		pcache_info[i].flags = (CRAT_CACHE_FLAGS_ENABLED |
+					CRAT_CACHE_FLAGS_INST_CACHE |
+					CRAT_CACHE_FLAGS_SIMD_CACHE);
+		pcache_info[i].num_cu_shared = adev->gfx.config.gc_num_sqc_per_wgp * 2;
+		pcache_info[i].cache_line_size = adev->gfx.config.gc_instruction_cache_line_size;
+		if (cache_line_size_missing && !pcache_info[i].cache_line_size)
+			pcache_info[i].cache_line_size = 128;
+		i++;
+	}
+	/* Scalar L1 Data Cache per SQC */
+	if (adev->gfx.config.gc_l1_data_cache_size_per_sqc) {
+		pcache_info[i].cache_size = adev->gfx.config.gc_l1_data_cache_size_per_sqc;
+		pcache_info[i].cache_level = 1;
+		pcache_info[i].flags = (CRAT_CACHE_FLAGS_ENABLED |
+					CRAT_CACHE_FLAGS_DATA_CACHE |
+					CRAT_CACHE_FLAGS_SIMD_CACHE);
+		pcache_info[i].num_cu_shared = adev->gfx.config.gc_num_sqc_per_wgp * 2;
+		pcache_info[i].cache_line_size = adev->gfx.config.gc_scalar_data_cache_line_size;
+		if (cache_line_size_missing && !pcache_info[i].cache_line_size)
+			pcache_info[i].cache_line_size = 64;
+		i++;
+	}
+	/* GL1 Data Cache per SA */
+	if (adev->gfx.config.gc_gl1c_per_sa &&
+	    adev->gfx.config.gc_gl1c_size_per_instance) {
+		pcache_info[i].cache_size = adev->gfx.config.gc_gl1c_per_sa *
+			adev->gfx.config.gc_gl1c_size_per_instance;
+		pcache_info[i].cache_level = 1;
+		pcache_info[i].flags = (CRAT_CACHE_FLAGS_ENABLED |
+					CRAT_CACHE_FLAGS_DATA_CACHE |
+					CRAT_CACHE_FLAGS_SIMD_CACHE);
+		pcache_info[i].num_cu_shared = adev->gfx.config.max_cu_per_sh;
+		if (cache_line_size_missing)
+			pcache_info[i].cache_line_size = 128;
+		i++;
+	}
+	/* L2 Data Cache per GPU (Total Tex Cache) */
+	if (adev->gfx.config.gc_gl2c_per_gpu) {
+		pcache_info[i].cache_size = adev->gfx.config.gc_gl2c_per_gpu;
+		pcache_info[i].cache_level = 2;
+		pcache_info[i].flags = (CRAT_CACHE_FLAGS_ENABLED |
+					CRAT_CACHE_FLAGS_DATA_CACHE |
+					CRAT_CACHE_FLAGS_SIMD_CACHE);
+		pcache_info[i].num_cu_shared = adev->gfx.config.max_cu_per_sh;
+		pcache_info[i].cache_line_size = adev->gfx.config.gc_tcc_cache_line_size;
+		if (cache_line_size_missing && !pcache_info[i].cache_line_size)
+			pcache_info[i].cache_line_size = 128;
+		i++;
+	}
+	/* L3 Data Cache per GPU */
+	if (adev->gmc.mall_size) {
+		pcache_info[i].cache_size = adev->gmc.mall_size / 1024;
+		pcache_info[i].cache_level = 3;
+		pcache_info[i].flags = (CRAT_CACHE_FLAGS_ENABLED |
+					CRAT_CACHE_FLAGS_DATA_CACHE |
+					CRAT_CACHE_FLAGS_SIMD_CACHE);
+		pcache_info[i].num_cu_shared = adev->gfx.config.max_cu_per_sh;
+		pcache_info[i].cache_line_size = 64;
+		i++;
+	}
+	return i;
+}
+
+static int kfd_fill_gpu_cache_info_from_gfx_config_v2(struct kfd_dev *kdev,
+						   struct kfd_gpu_cache_info *pcache_info)
+{
+	struct amdgpu_device *adev = kdev->adev;
+	int i = 0;
+
+	/* TCP L1 Cache per CU */
+	if (adev->gfx.config.gc_tcp_size_per_cu) {
+		pcache_info[i].cache_size = adev->gfx.config.gc_tcp_size_per_cu;
+		pcache_info[i].cache_level = 1;
+		/* Cacheline size not available in IP discovery for gc943,gc944 */
+		pcache_info[i].cache_line_size = 128;
+		pcache_info[i].flags = (CRAT_CACHE_FLAGS_ENABLED |
+					CRAT_CACHE_FLAGS_DATA_CACHE |
+					CRAT_CACHE_FLAGS_SIMD_CACHE);
+		pcache_info[i].num_cu_shared = 1;
+		i++;
+	}
+	/* Scalar L1 Instruction Cache per SQC */
+	if (adev->gfx.config.gc_l1_instruction_cache_size_per_sqc) {
+		pcache_info[i].cache_size =
+			adev->gfx.config.gc_l1_instruction_cache_size_per_sqc;
+		pcache_info[i].cache_level = 1;
+		pcache_info[i].cache_line_size = 64;
+		pcache_info[i].flags = (CRAT_CACHE_FLAGS_ENABLED |
+					CRAT_CACHE_FLAGS_INST_CACHE |
+					CRAT_CACHE_FLAGS_SIMD_CACHE);
+		pcache_info[i].num_cu_shared = adev->gfx.config.gc_num_cu_per_sqc;
+		i++;
+	}
+	/* Scalar L1 Data Cache per SQC */
+	if (adev->gfx.config.gc_l1_data_cache_size_per_sqc) {
+		pcache_info[i].cache_size = adev->gfx.config.gc_l1_data_cache_size_per_sqc;
+		pcache_info[i].cache_level = 1;
+		pcache_info[i].cache_line_size = 64;
+		pcache_info[i].flags = (CRAT_CACHE_FLAGS_ENABLED |
+					CRAT_CACHE_FLAGS_DATA_CACHE |
+					CRAT_CACHE_FLAGS_SIMD_CACHE);
+		pcache_info[i].num_cu_shared = adev->gfx.config.gc_num_cu_per_sqc;
+		i++;
+	}
+	/* L2 Data Cache per GPU (Total Tex Cache) */
+	if (adev->gfx.config.gc_tcc_size) {
+		pcache_info[i].cache_size = adev->gfx.config.gc_tcc_size;
+		pcache_info[i].cache_level = 2;
+		pcache_info[i].cache_line_size = 128;
+		pcache_info[i].flags = (CRAT_CACHE_FLAGS_ENABLED |
+					CRAT_CACHE_FLAGS_DATA_CACHE |
+					CRAT_CACHE_FLAGS_SIMD_CACHE);
+		pcache_info[i].num_cu_shared = adev->gfx.config.max_cu_per_sh;
+		i++;
+	}
+	/* L3 Data Cache per GPU */
+	if (adev->gmc.mall_size) {
+		pcache_info[i].cache_size = adev->gmc.mall_size / 1024;
+		pcache_info[i].cache_level = 3;
+		pcache_info[i].cache_line_size = 64;
+		pcache_info[i].flags = (CRAT_CACHE_FLAGS_ENABLED |
+					CRAT_CACHE_FLAGS_DATA_CACHE |
+					CRAT_CACHE_FLAGS_SIMD_CACHE);
+		pcache_info[i].num_cu_shared = adev->gfx.config.max_cu_per_sh;
+		i++;
+	}
+	return i;
+}
+
+int kfd_get_gpu_cache_info(struct kfd_node *kdev, struct kfd_gpu_cache_info **pcache_info)
+{
+	int num_of_cache_types = 0;
+	bool cache_line_size_missing = false;
+
+	switch (kdev->adev->asic_type) {
+	case CHIP_KAVERI:
+		*pcache_info = kaveri_cache_info;
+		num_of_cache_types = ARRAY_SIZE(kaveri_cache_info);
+		break;
+	case CHIP_HAWAII:
+		*pcache_info = hawaii_cache_info;
+		num_of_cache_types = ARRAY_SIZE(hawaii_cache_info);
+		break;
+	case CHIP_CARRIZO:
+		*pcache_info = carrizo_cache_info;
+		num_of_cache_types = ARRAY_SIZE(carrizo_cache_info);
+		break;
+	case CHIP_TONGA:
+		*pcache_info = tonga_cache_info;
+		num_of_cache_types = ARRAY_SIZE(tonga_cache_info);
+		break;
+	case CHIP_FIJI:
+		*pcache_info = fiji_cache_info;
+		num_of_cache_types = ARRAY_SIZE(fiji_cache_info);
+		break;
+	case CHIP_POLARIS10:
+		*pcache_info = polaris10_cache_info;
+		num_of_cache_types = ARRAY_SIZE(polaris10_cache_info);
+		break;
+	case CHIP_POLARIS11:
+		*pcache_info = polaris11_cache_info;
+		num_of_cache_types = ARRAY_SIZE(polaris11_cache_info);
+		break;
+	case CHIP_POLARIS12:
+		*pcache_info = polaris12_cache_info;
+		num_of_cache_types = ARRAY_SIZE(polaris12_cache_info);
+		break;
+	case CHIP_VEGAM:
+		*pcache_info = vegam_cache_info;
+		num_of_cache_types = ARRAY_SIZE(vegam_cache_info);
+		break;
+	default:
+		switch (KFD_GC_VERSION(kdev)) {
+		case IP_VERSION(9, 0, 1):
+			*pcache_info = vega10_cache_info;
+			num_of_cache_types = ARRAY_SIZE(vega10_cache_info);
+			break;
+		case IP_VERSION(9, 2, 1):
+			*pcache_info = vega12_cache_info;
+			num_of_cache_types = ARRAY_SIZE(vega12_cache_info);
+			break;
+		case IP_VERSION(9, 4, 0):
+		case IP_VERSION(9, 4, 1):
+			*pcache_info = vega20_cache_info;
+			num_of_cache_types = ARRAY_SIZE(vega20_cache_info);
+			break;
+		case IP_VERSION(9, 4, 2):
+			*pcache_info = aldebaran_cache_info;
+			num_of_cache_types = ARRAY_SIZE(aldebaran_cache_info);
+			break;
+		case IP_VERSION(9, 4, 3):
+		case IP_VERSION(9, 4, 4):
+		case IP_VERSION(9, 5, 0):
+			num_of_cache_types =
+				kfd_fill_gpu_cache_info_from_gfx_config_v2(kdev->kfd,
+									*pcache_info);
+			break;
+		case IP_VERSION(9, 1, 0):
+		case IP_VERSION(9, 2, 2):
+			*pcache_info = raven_cache_info;
+			num_of_cache_types = ARRAY_SIZE(raven_cache_info);
+			break;
+		case IP_VERSION(9, 3, 0):
+			*pcache_info = renoir_cache_info;
+			num_of_cache_types = ARRAY_SIZE(renoir_cache_info);
+			break;
+		case IP_VERSION(10, 1, 10):
+		case IP_VERSION(10, 1, 2):
+		case IP_VERSION(10, 1, 3):
+		case IP_VERSION(10, 1, 4):
+			*pcache_info = navi10_cache_info;
+			num_of_cache_types = ARRAY_SIZE(navi10_cache_info);
+			break;
+		case IP_VERSION(10, 1, 1):
+			*pcache_info = navi14_cache_info;
+			num_of_cache_types = ARRAY_SIZE(navi14_cache_info);
+			break;
+		case IP_VERSION(10, 3, 0):
+			*pcache_info = sienna_cichlid_cache_info;
+			num_of_cache_types = ARRAY_SIZE(sienna_cichlid_cache_info);
+			break;
+		case IP_VERSION(10, 3, 2):
+			*pcache_info = navy_flounder_cache_info;
+			num_of_cache_types = ARRAY_SIZE(navy_flounder_cache_info);
+			break;
+		case IP_VERSION(10, 3, 4):
+			*pcache_info = dimgrey_cavefish_cache_info;
+			num_of_cache_types = ARRAY_SIZE(dimgrey_cavefish_cache_info);
+			break;
+		case IP_VERSION(10, 3, 1):
+			*pcache_info = vangogh_cache_info;
+			num_of_cache_types = ARRAY_SIZE(vangogh_cache_info);
+			break;
+		case IP_VERSION(10, 3, 5):
+			*pcache_info = beige_goby_cache_info;
+			num_of_cache_types = ARRAY_SIZE(beige_goby_cache_info);
+			break;
+		case IP_VERSION(10, 3, 3):
+			*pcache_info = yellow_carp_cache_info;
+			num_of_cache_types = ARRAY_SIZE(yellow_carp_cache_info);
+			break;
+		case IP_VERSION(10, 3, 6):
+			*pcache_info = gc_10_3_6_cache_info;
+			num_of_cache_types = ARRAY_SIZE(gc_10_3_6_cache_info);
+			break;
+		case IP_VERSION(10, 3, 7):
+			*pcache_info = gfx1037_cache_info;
+			num_of_cache_types = ARRAY_SIZE(gfx1037_cache_info);
+			break;
+		case IP_VERSION(11, 0, 0):
+		case IP_VERSION(11, 0, 1):
+		case IP_VERSION(11, 0, 2):
+		case IP_VERSION(11, 0, 3):
+		case IP_VERSION(11, 0, 4):
+		case IP_VERSION(11, 5, 0):
+		case IP_VERSION(11, 5, 1):
+		case IP_VERSION(11, 5, 2):
+		case IP_VERSION(11, 5, 3):
+			/* Cacheline size not available in IP discovery for gc11.
+			 * kfd_fill_gpu_cache_info_from_gfx_config to hard code it
+			 */
+			cache_line_size_missing = true;
+			fallthrough;
+		case IP_VERSION(12, 0, 0):
+		case IP_VERSION(12, 0, 1):
+			num_of_cache_types =
+				kfd_fill_gpu_cache_info_from_gfx_config(kdev->kfd,
+									cache_line_size_missing,
+									*pcache_info);
+			break;
+		default:
+			*pcache_info = dummy_cache_info;
+			num_of_cache_types = ARRAY_SIZE(dummy_cache_info);
+			pr_warn("dummy cache info is used temporarily and real cache info need update later.\n");
+			break;
+		}
+	}
+	return num_of_cache_types;
+}
+
+/* Memory required to create Virtual CRAT.
+ * Since there is no easy way to predict the amount of memory required, the
+ * following amount is allocated for GPU Virtual CRAT. This is
+ * expected to cover all known conditions. But to be safe additional check
+ * is put in the code to ensure we don't overwrite.
+ */
+#define VCRAT_SIZE_FOR_GPU	(4 * PAGE_SIZE)
+
+/* kfd_fill_cu_for_cpu - Fill in Compute info for the given CPU NUMA node
+ *
+ *	@numa_node_id: CPU NUMA node id
+ *	@avail_size: Available size in the memory
+ *	@sub_type_hdr: Memory into which compute info will be filled in
+ *
+ *	Return 0 if successful else return -ve value
+ */
+static int kfd_fill_cu_for_cpu(int numa_node_id, int *avail_size,
+				int proximity_domain,
+				struct crat_subtype_computeunit *sub_type_hdr)
+{
+	const struct cpumask *cpumask;
+
+	*avail_size -= sizeof(struct crat_subtype_computeunit);
+	if (*avail_size < 0)
+		return -ENOMEM;
+
+	memset(sub_type_hdr, 0, sizeof(struct crat_subtype_computeunit));
+
+	/* Fill in subtype header data */
+	sub_type_hdr->type = CRAT_SUBTYPE_COMPUTEUNIT_AFFINITY;
+	sub_type_hdr->length = sizeof(struct crat_subtype_computeunit);
+	sub_type_hdr->flags = CRAT_SUBTYPE_FLAGS_ENABLED;
+
+	cpumask = cpumask_of_node(numa_node_id);
+
+	/* Fill in CU data */
+	sub_type_hdr->flags |= CRAT_CU_FLAGS_CPU_PRESENT;
+	sub_type_hdr->proximity_domain = proximity_domain;
+	sub_type_hdr->processor_id_low = kfd_numa_node_to_apic_id(numa_node_id);
+	if (sub_type_hdr->processor_id_low == -1)
+		return -EINVAL;
+
+	sub_type_hdr->num_cpu_cores = cpumask_weight(cpumask);
+
+	return 0;
+}
+
+/* kfd_fill_mem_info_for_cpu - Fill in Memory info for the given CPU NUMA node
+ *
+ *	@numa_node_id: CPU NUMA node id
+ *	@avail_size: Available size in the memory
+ *	@sub_type_hdr: Memory into which compute info will be filled in
+ *
+ *	Return 0 if successful else return -ve value
+ */
+static int kfd_fill_mem_info_for_cpu(int numa_node_id, int *avail_size,
+			int proximity_domain,
+			struct crat_subtype_memory *sub_type_hdr)
+{
+	uint64_t mem_in_bytes = 0;
+	pg_data_t *pgdat;
+	int zone_type;
+
+	*avail_size -= sizeof(struct crat_subtype_memory);
+	if (*avail_size < 0)
+		return -ENOMEM;
+
+	memset(sub_type_hdr, 0, sizeof(struct crat_subtype_memory));
+
+	/* Fill in subtype header data */
+	sub_type_hdr->type = CRAT_SUBTYPE_MEMORY_AFFINITY;
+	sub_type_hdr->length = sizeof(struct crat_subtype_memory);
+	sub_type_hdr->flags = CRAT_SUBTYPE_FLAGS_ENABLED;
+
+	/* Fill in Memory Subunit data */
+
+	/* Unlike si_meminfo, si_meminfo_node is not exported. So
+	 * the following lines are duplicated from si_meminfo_node
+	 * function
+	 */
+	pgdat = NODE_DATA(numa_node_id);
+	for (zone_type = 0; zone_type < MAX_NR_ZONES; zone_type++)
+		mem_in_bytes += zone_managed_pages(&pgdat->node_zones[zone_type]);
+	mem_in_bytes <<= PAGE_SHIFT;
+
+	sub_type_hdr->length_low = lower_32_bits(mem_in_bytes);
+	sub_type_hdr->length_high = upper_32_bits(mem_in_bytes);
+	sub_type_hdr->proximity_domain = proximity_domain;
+
+	return 0;
+}
+
+#ifdef CONFIG_X86_64
+static int kfd_fill_iolink_info_for_cpu(int numa_node_id, int *avail_size,
+				uint32_t *num_entries,
+				struct crat_subtype_iolink *sub_type_hdr)
+{
+	int nid;
+	struct cpuinfo_x86 *c = &cpu_data(0);
+	uint8_t link_type;
+
+	if (c->x86_vendor == X86_VENDOR_AMD)
+		link_type = CRAT_IOLINK_TYPE_HYPERTRANSPORT;
+	else
+		link_type = CRAT_IOLINK_TYPE_QPI_1_1;
+
+	*num_entries = 0;
+
+	/* Create IO links from this node to other CPU nodes */
+	for_each_online_node(nid) {
+		if (nid == numa_node_id) /* node itself */
+			continue;
+
+		*avail_size -= sizeof(struct crat_subtype_iolink);
+		if (*avail_size < 0)
+			return -ENOMEM;
+
+		memset(sub_type_hdr, 0, sizeof(struct crat_subtype_iolink));
+
+		/* Fill in subtype header data */
+		sub_type_hdr->type = CRAT_SUBTYPE_IOLINK_AFFINITY;
+		sub_type_hdr->length = sizeof(struct crat_subtype_iolink);
+		sub_type_hdr->flags = CRAT_SUBTYPE_FLAGS_ENABLED;
+
+		/* Fill in IO link data */
+		sub_type_hdr->proximity_domain_from = numa_node_id;
+		sub_type_hdr->proximity_domain_to = nid;
+		sub_type_hdr->io_interface_type = link_type;
+
+		(*num_entries)++;
+		sub_type_hdr++;
+	}
+
+	return 0;
+}
+#endif
+
+/* kfd_create_vcrat_image_cpu - Create Virtual CRAT for CPU
+ *
+ *	@pcrat_image: Fill in VCRAT for CPU
+ *	@size:	[IN] allocated size of crat_image.
+ *		[OUT] actual size of data filled in crat_image
+ */
+static int kfd_create_vcrat_image_cpu(void *pcrat_image, size_t *size)
+{
+	struct crat_header *crat_table = (struct crat_header *)pcrat_image;
+	struct acpi_table_header *acpi_table;
+	acpi_status status;
+	struct crat_subtype_generic *sub_type_hdr;
+	int avail_size = *size;
+	int numa_node_id;
+#ifdef CONFIG_X86_64
+	uint32_t entries = 0;
+#endif
+	int ret = 0;
+
+	if (!pcrat_image)
+		return -EINVAL;
+
+	/* Fill in CRAT Header.
+	 * Modify length and total_entries as subunits are added.
+	 */
+	avail_size -= sizeof(struct crat_header);
+	if (avail_size < 0)
+		return -ENOMEM;
+
+	memset(crat_table, 0, sizeof(struct crat_header));
+	memcpy(&crat_table->signature, CRAT_SIGNATURE,
+			sizeof(crat_table->signature));
+	crat_table->length = sizeof(struct crat_header);
+
+	status = acpi_get_table("DSDT", 0, &acpi_table);
+	if (status != AE_OK)
+		pr_warn("DSDT table not found for OEM information\n");
+	else {
+		crat_table->oem_revision = acpi_table->revision;
+		memcpy(crat_table->oem_id, acpi_table->oem_id,
+				CRAT_OEMID_LENGTH);
+		memcpy(crat_table->oem_table_id, acpi_table->oem_table_id,
+				CRAT_OEMTABLEID_LENGTH);
+		acpi_put_table(acpi_table);
+	}
+	crat_table->total_entries = 0;
+	crat_table->num_domains = 0;
+
+	sub_type_hdr = (struct crat_subtype_generic *)(crat_table+1);
+
+	for_each_online_node(numa_node_id) {
+		if (kfd_numa_node_to_apic_id(numa_node_id) == -1)
+			continue;
+
+		/* Fill in Subtype: Compute Unit */
+		ret = kfd_fill_cu_for_cpu(numa_node_id, &avail_size,
+			crat_table->num_domains,
+			(struct crat_subtype_computeunit *)sub_type_hdr);
+		if (ret < 0)
+			return ret;
+		crat_table->length += sub_type_hdr->length;
+		crat_table->total_entries++;
+
+		sub_type_hdr = (typeof(sub_type_hdr))((char *)sub_type_hdr +
+			sub_type_hdr->length);
+
+		/* Fill in Subtype: Memory */
+		ret = kfd_fill_mem_info_for_cpu(numa_node_id, &avail_size,
+			crat_table->num_domains,
+			(struct crat_subtype_memory *)sub_type_hdr);
+		if (ret < 0)
+			return ret;
+		crat_table->length += sub_type_hdr->length;
+		crat_table->total_entries++;
+
+		sub_type_hdr = (typeof(sub_type_hdr))((char *)sub_type_hdr +
+			sub_type_hdr->length);
+
+		/* Fill in Subtype: IO Link */
+#ifdef CONFIG_X86_64
+		ret = kfd_fill_iolink_info_for_cpu(numa_node_id, &avail_size,
+				&entries,
+				(struct crat_subtype_iolink *)sub_type_hdr);
+		if (ret < 0)
+			return ret;
+
+		if (entries) {
+			crat_table->length += (sub_type_hdr->length * entries);
+			crat_table->total_entries += entries;
+
+			sub_type_hdr = (typeof(sub_type_hdr))((char *)sub_type_hdr +
+					sub_type_hdr->length * entries);
+		}
+#else
+		pr_info("IO link not available for non x86 platforms\n");
+#endif
+
+		crat_table->num_domains++;
+	}
+
+	/* TODO: Add cache Subtype for CPU.
+	 * Currently, CPU cache information is available in function
+	 * detect_cache_attributes(cpu) defined in the file
+	 * ./arch/x86/kernel/cpu/intel_cacheinfo.c. This function is not
+	 * exported and to get the same information the code needs to be
+	 * duplicated.
+	 */
+
+	*size = crat_table->length;
+	pr_info("Virtual CRAT table created for CPU\n");
+
+	return 0;
+}
+
+static int kfd_fill_gpu_memory_affinity(int *avail_size,
+		struct kfd_node *kdev, uint8_t type, uint64_t size,
+		struct crat_subtype_memory *sub_type_hdr,
+		uint32_t proximity_domain,
+		const struct kfd_local_mem_info *local_mem_info)
+{
+	*avail_size -= sizeof(struct crat_subtype_memory);
+	if (*avail_size < 0)
+		return -ENOMEM;
+
+	memset((void *)sub_type_hdr, 0, sizeof(struct crat_subtype_memory));
+	sub_type_hdr->type = CRAT_SUBTYPE_MEMORY_AFFINITY;
+	sub_type_hdr->length = sizeof(struct crat_subtype_memory);
+	sub_type_hdr->flags |= CRAT_SUBTYPE_FLAGS_ENABLED;
+
+	sub_type_hdr->proximity_domain = proximity_domain;
+
+	pr_debug("Fill gpu memory affinity - type 0x%x size 0x%llx\n",
+			type, size);
+
+	sub_type_hdr->length_low = lower_32_bits(size);
+	sub_type_hdr->length_high = upper_32_bits(size);
+
+	sub_type_hdr->width = local_mem_info->vram_width;
+	sub_type_hdr->visibility_type = type;
+
+	return 0;
+}
+
+#ifdef CONFIG_ACPI_NUMA
+static void kfd_find_numa_node_in_srat(struct kfd_node *kdev)
+{
+	struct acpi_table_header *table_header = NULL;
+	struct acpi_subtable_header *sub_header = NULL;
+	unsigned long table_end, subtable_len;
+	u32 pci_id = pci_domain_nr(kdev->adev->pdev->bus) << 16 |
+			pci_dev_id(kdev->adev->pdev);
+	u32 bdf;
+	acpi_status status;
+	struct acpi_srat_cpu_affinity *cpu;
+	struct acpi_srat_generic_affinity *gpu;
+	int pxm = 0, max_pxm = 0;
+	int numa_node = NUMA_NO_NODE;
+	bool found = false;
+
+	/* Fetch the SRAT table from ACPI */
+	status = acpi_get_table(ACPI_SIG_SRAT, 0, &table_header);
+	if (status == AE_NOT_FOUND) {
+		pr_warn("SRAT table not found\n");
+		return;
+	} else if (ACPI_FAILURE(status)) {
+		const char *err = acpi_format_exception(status);
+		pr_err("SRAT table error: %s\n", err);
+		return;
+	}
+
+	table_end = (unsigned long)table_header + table_header->length;
+
+	/* Parse all entries looking for a match. */
+	sub_header = (struct acpi_subtable_header *)
+			((unsigned long)table_header +
+			sizeof(struct acpi_table_srat));
+	subtable_len = sub_header->length;
+
+	while (((unsigned long)sub_header) + subtable_len  < table_end) {
+		/*
+		 * If length is 0, break from this loop to avoid
+		 * infinite loop.
+		 */
+		if (subtable_len == 0) {
+			pr_err("SRAT invalid zero length\n");
+			break;
+		}
+
+		switch (sub_header->type) {
+		case ACPI_SRAT_TYPE_CPU_AFFINITY:
+			cpu = (struct acpi_srat_cpu_affinity *)sub_header;
+			pxm = *((u32 *)cpu->proximity_domain_hi) << 8 |
+					cpu->proximity_domain_lo;
+			if (pxm > max_pxm)
+				max_pxm = pxm;
+			break;
+		case ACPI_SRAT_TYPE_GENERIC_AFFINITY:
+			gpu = (struct acpi_srat_generic_affinity *)sub_header;
+			bdf = *((u16 *)(&gpu->device_handle[0])) << 16 |
+					*((u16 *)(&gpu->device_handle[2]));
+			if (bdf == pci_id) {
+				found = true;
+				numa_node = pxm_to_node(gpu->proximity_domain);
+			}
+			break;
+		default:
+			break;
+		}
+
+		if (found)
+			break;
+
+		sub_header = (struct acpi_subtable_header *)
+				((unsigned long)sub_header + subtable_len);
+		subtable_len = sub_header->length;
+	}
+
+	acpi_put_table(table_header);
+
+	/* Workaround bad cpu-gpu binding case */
+	if (found && (numa_node < 0 ||
+			numa_node > pxm_to_node(max_pxm)))
+		numa_node = 0;
+
+	if (numa_node != NUMA_NO_NODE)
+		set_dev_node(&kdev->adev->pdev->dev, numa_node);
+}
+#endif
+
+#define KFD_CRAT_INTRA_SOCKET_WEIGHT	13
+#define KFD_CRAT_XGMI_WEIGHT		15
+
+/* kfd_fill_gpu_direct_io_link - Fill in direct io link from GPU
+ * to its NUMA node
+ *	@avail_size: Available size in the memory
+ *	@kdev - [IN] GPU device
+ *	@sub_type_hdr: Memory into which io link info will be filled in
+ *	@proximity_domain - proximity domain of the GPU node
+ *
+ *	Return 0 if successful else return -ve value
+ */
+static int kfd_fill_gpu_direct_io_link_to_cpu(int *avail_size,
+			struct kfd_node *kdev,
+			struct crat_subtype_iolink *sub_type_hdr,
+			uint32_t proximity_domain)
+{
+	*avail_size -= sizeof(struct crat_subtype_iolink);
+	if (*avail_size < 0)
+		return -ENOMEM;
+
+	memset((void *)sub_type_hdr, 0, sizeof(struct crat_subtype_iolink));
+
+	/* Fill in subtype header data */
+	sub_type_hdr->type = CRAT_SUBTYPE_IOLINK_AFFINITY;
+	sub_type_hdr->length = sizeof(struct crat_subtype_iolink);
+	sub_type_hdr->flags |= CRAT_SUBTYPE_FLAGS_ENABLED;
+	if (kfd_dev_is_large_bar(kdev))
+		sub_type_hdr->flags |= CRAT_IOLINK_FLAGS_BI_DIRECTIONAL;
+
+	/* Fill in IOLINK subtype.
+	 * TODO: Fill-in other fields of iolink subtype
+	 */
+	if (kdev->adev->gmc.xgmi.connected_to_cpu ||
+	    (KFD_GC_VERSION(kdev) == IP_VERSION(9, 4, 3) &&
+	     kdev->adev->smuio.funcs->get_pkg_type(kdev->adev) ==
+	     AMDGPU_PKG_TYPE_APU)) {
+		bool ext_cpu = KFD_GC_VERSION(kdev) != IP_VERSION(9, 4, 3);
+		int mem_bw = 819200, weight = ext_cpu ? KFD_CRAT_XGMI_WEIGHT :
+							KFD_CRAT_INTRA_SOCKET_WEIGHT;
+		/*
+		 * with host gpu xgmi link, host can access gpu memory whether
+		 * or not pcie bar type is large, so always create bidirectional
+		 * io link.
+		 */
+		sub_type_hdr->flags |= CRAT_IOLINK_FLAGS_BI_DIRECTIONAL;
+		sub_type_hdr->io_interface_type = CRAT_IOLINK_TYPE_XGMI;
+		sub_type_hdr->weight_xgmi = weight;
+		if (ext_cpu) {
+			amdgpu_xgmi_get_bandwidth(kdev->adev, NULL,
+						  AMDGPU_XGMI_BW_MODE_PER_LINK,
+						  AMDGPU_XGMI_BW_UNIT_MBYTES,
+						  &sub_type_hdr->minimum_bandwidth_mbs,
+						  &sub_type_hdr->maximum_bandwidth_mbs);
+		} else {
+			sub_type_hdr->minimum_bandwidth_mbs = mem_bw;
+			sub_type_hdr->maximum_bandwidth_mbs = mem_bw;
+		}
+	} else {
+		sub_type_hdr->io_interface_type = CRAT_IOLINK_TYPE_PCIEXPRESS;
+		sub_type_hdr->minimum_bandwidth_mbs =
+				amdgpu_amdkfd_get_pcie_bandwidth_mbytes(kdev->adev, true);
+		sub_type_hdr->maximum_bandwidth_mbs =
+				amdgpu_amdkfd_get_pcie_bandwidth_mbytes(kdev->adev, false);
+	}
+
+	sub_type_hdr->proximity_domain_from = proximity_domain;
+
+#ifdef CONFIG_ACPI_NUMA
+	if (kdev->adev->pdev->dev.numa_node == NUMA_NO_NODE &&
+	    num_possible_nodes() > 1)
+		kfd_find_numa_node_in_srat(kdev);
+#endif
+#ifdef CONFIG_NUMA
+	if (kdev->adev->pdev->dev.numa_node == NUMA_NO_NODE)
+		sub_type_hdr->proximity_domain_to = 0;
+	else
+		sub_type_hdr->proximity_domain_to = kdev->adev->pdev->dev.numa_node;
+#else
+	sub_type_hdr->proximity_domain_to = 0;
+#endif
+	return 0;
+}
+
+static int kfd_fill_gpu_xgmi_link_to_gpu(int *avail_size,
+			struct kfd_node *kdev,
+			struct kfd_node *peer_kdev,
+			struct crat_subtype_iolink *sub_type_hdr,
+			uint32_t proximity_domain_from,
+			uint32_t proximity_domain_to)
+{
+	bool use_ta_info = kdev->kfd->num_nodes == 1;
+
+	*avail_size -= sizeof(struct crat_subtype_iolink);
+	if (*avail_size < 0)
+		return -ENOMEM;
+
+	memset((void *)sub_type_hdr, 0, sizeof(struct crat_subtype_iolink));
+
+	sub_type_hdr->type = CRAT_SUBTYPE_IOLINK_AFFINITY;
+	sub_type_hdr->length = sizeof(struct crat_subtype_iolink);
+	sub_type_hdr->flags |= CRAT_SUBTYPE_FLAGS_ENABLED |
+			       CRAT_IOLINK_FLAGS_BI_DIRECTIONAL;
+
+	sub_type_hdr->io_interface_type = CRAT_IOLINK_TYPE_XGMI;
+	sub_type_hdr->proximity_domain_from = proximity_domain_from;
+	sub_type_hdr->proximity_domain_to = proximity_domain_to;
+
+	if (use_ta_info) {
+		sub_type_hdr->weight_xgmi = KFD_CRAT_XGMI_WEIGHT *
+			amdgpu_xgmi_get_hops_count(kdev->adev, peer_kdev->adev);
+		amdgpu_xgmi_get_bandwidth(kdev->adev, peer_kdev->adev,
+					  AMDGPU_XGMI_BW_MODE_PER_PEER,
+					  AMDGPU_XGMI_BW_UNIT_MBYTES,
+					  &sub_type_hdr->minimum_bandwidth_mbs,
+					  &sub_type_hdr->maximum_bandwidth_mbs);
+	} else {
+		bool is_single_hop = kdev->kfd == peer_kdev->kfd;
+		int weight = is_single_hop ? KFD_CRAT_INTRA_SOCKET_WEIGHT :
+			(2 * KFD_CRAT_INTRA_SOCKET_WEIGHT) + KFD_CRAT_XGMI_WEIGHT;
+		int mem_bw = 819200;
+
+		sub_type_hdr->weight_xgmi = weight;
+		sub_type_hdr->maximum_bandwidth_mbs = is_single_hop ? mem_bw : 0;
+		sub_type_hdr->minimum_bandwidth_mbs = is_single_hop ? mem_bw : 0;
+	}
+
+	return 0;
+}
+
+/* kfd_create_vcrat_image_gpu - Create Virtual CRAT for CPU
+ *
+ *	@pcrat_image: Fill in VCRAT for GPU
+ *	@size:	[IN] allocated size of crat_image.
+ *		[OUT] actual size of data filled in crat_image
+ */
+static int kfd_create_vcrat_image_gpu(void *pcrat_image,
+				      size_t *size, struct kfd_node *kdev,
+				      uint32_t proximity_domain)
+{
+	struct crat_header *crat_table = (struct crat_header *)pcrat_image;
+	struct amdgpu_gfx_config *gfx_info = &kdev->adev->gfx.config;
+	struct amdgpu_cu_info *cu_info = &kdev->adev->gfx.cu_info;
+	struct crat_subtype_generic *sub_type_hdr;
+	struct kfd_local_mem_info local_mem_info;
+	struct kfd_topology_device *peer_dev;
+	struct crat_subtype_computeunit *cu;
+	int avail_size = *size;
+	uint32_t total_num_of_cu;
+	uint32_t nid = 0;
+	int ret = 0;
+
+	if (!pcrat_image || avail_size < VCRAT_SIZE_FOR_GPU)
+		return -EINVAL;
+
+	/* Fill the CRAT Header.
+	 * Modify length and total_entries as subunits are added.
+	 */
+	avail_size -= sizeof(struct crat_header);
+	memset(crat_table, 0, sizeof(struct crat_header));
+
+	memcpy(&crat_table->signature, CRAT_SIGNATURE,
+			sizeof(crat_table->signature));
+	/* Change length as we add more subtypes*/
+	crat_table->length = sizeof(struct crat_header);
+	crat_table->num_domains = 1;
+	crat_table->total_entries = 0;
+
+	/* Fill in Subtype: Compute Unit
+	 * First fill in the sub type header and then sub type data
+	 */
+	avail_size -= sizeof(struct crat_subtype_computeunit);
+	sub_type_hdr = (struct crat_subtype_generic *)(crat_table + 1);
+	memset(sub_type_hdr, 0, sizeof(struct crat_subtype_computeunit));
+
+	sub_type_hdr->type = CRAT_SUBTYPE_COMPUTEUNIT_AFFINITY;
+	sub_type_hdr->length = sizeof(struct crat_subtype_computeunit);
+	sub_type_hdr->flags = CRAT_SUBTYPE_FLAGS_ENABLED;
+
+	/* Fill CU subtype data */
+	cu = (struct crat_subtype_computeunit *)sub_type_hdr;
+	cu->flags |= CRAT_CU_FLAGS_GPU_PRESENT;
+	cu->proximity_domain = proximity_domain;
+
+	cu->num_simd_per_cu = cu_info->simd_per_cu;
+	cu->num_simd_cores = cu_info->simd_per_cu *
+			(cu_info->number / kdev->kfd->num_nodes);
+	cu->max_waves_simd = cu_info->max_waves_per_simd;
+
+	cu->wave_front_size = cu_info->wave_front_size;
+	cu->array_count = gfx_info->max_sh_per_se *
+		gfx_info->max_shader_engines;
+	total_num_of_cu = (cu->array_count * gfx_info->max_cu_per_sh);
+	cu->processor_id_low = get_and_inc_gpu_processor_id(total_num_of_cu);
+	cu->num_cu_per_array = gfx_info->max_cu_per_sh;
+	cu->max_slots_scatch_cu = cu_info->max_scratch_slots_per_cu;
+	cu->num_banks = gfx_info->max_shader_engines;
+	cu->lds_size_in_kb = cu_info->lds_size;
+
+	cu->hsa_capability = 0;
+
+	crat_table->length += sub_type_hdr->length;
+	crat_table->total_entries++;
+
+	/* Fill in Subtype: Memory. Only on systems with large BAR (no
+	 * private FB), report memory as public. On other systems
+	 * report the total FB size (public+private) as a single
+	 * private heap.
+	 */
+	local_mem_info = kdev->local_mem_info;
+	sub_type_hdr = (typeof(sub_type_hdr))((char *)sub_type_hdr +
+			sub_type_hdr->length);
+
+	if (kdev->adev->debug_largebar)
+		local_mem_info.local_mem_size_private = 0;
+
+	if (local_mem_info.local_mem_size_private == 0)
+		ret = kfd_fill_gpu_memory_affinity(&avail_size,
+				kdev, HSA_MEM_HEAP_TYPE_FB_PUBLIC,
+				local_mem_info.local_mem_size_public,
+				(struct crat_subtype_memory *)sub_type_hdr,
+				proximity_domain,
+				&local_mem_info);
+	else
+		ret = kfd_fill_gpu_memory_affinity(&avail_size,
+				kdev, HSA_MEM_HEAP_TYPE_FB_PRIVATE,
+				local_mem_info.local_mem_size_public +
+				local_mem_info.local_mem_size_private,
+				(struct crat_subtype_memory *)sub_type_hdr,
+				proximity_domain,
+				&local_mem_info);
+	if (ret < 0)
+		return ret;
+
+	crat_table->length += sizeof(struct crat_subtype_memory);
+	crat_table->total_entries++;
+
+	/* Fill in Subtype: IO_LINKS
+	 *  Only direct links are added here which is Link from GPU to
+	 *  its NUMA node. Indirect links are added by userspace.
+	 */
+	sub_type_hdr = (typeof(sub_type_hdr))((char *)sub_type_hdr +
+		sub_type_hdr->length);
+	ret = kfd_fill_gpu_direct_io_link_to_cpu(&avail_size, kdev,
+		(struct crat_subtype_iolink *)sub_type_hdr, proximity_domain);
+
+	if (ret < 0)
+		return ret;
+
+	crat_table->length += sub_type_hdr->length;
+	crat_table->total_entries++;
+
+
+	/* Fill in Subtype: IO_LINKS
+	 * Direct links from GPU to other GPUs through xGMI.
+	 * We will loop GPUs that already be processed (with lower value
+	 * of proximity_domain), add the link for the GPUs with same
+	 * hive id (from this GPU to other GPU) . The reversed iolink
+	 * (from other GPU to this GPU) will be added
+	 * in kfd_parse_subtype_iolink.
+	 */
+	if (kdev->kfd->hive_id) {
+		for (nid = 0; nid < proximity_domain; ++nid) {
+			peer_dev = kfd_topology_device_by_proximity_domain_no_lock(nid);
+			if (!peer_dev->gpu)
+				continue;
+			if (peer_dev->gpu->kfd->hive_id != kdev->kfd->hive_id)
+				continue;
+			if (!amdgpu_xgmi_get_is_sharing_enabled(kdev->adev, peer_dev->gpu->adev))
+				continue;
+			sub_type_hdr = (typeof(sub_type_hdr))(
+				(char *)sub_type_hdr +
+				sizeof(struct crat_subtype_iolink));
+			ret = kfd_fill_gpu_xgmi_link_to_gpu(
+				&avail_size, kdev, peer_dev->gpu,
+				(struct crat_subtype_iolink *)sub_type_hdr,
+				proximity_domain, nid);
+			if (ret < 0)
+				return ret;
+			crat_table->length += sub_type_hdr->length;
+			crat_table->total_entries++;
+		}
+	}
+	*size = crat_table->length;
+	pr_info("Virtual CRAT table created for GPU\n");
+
+	return ret;
+}
+
+/* kfd_create_crat_image_virtual - Allocates memory for CRAT image and
+ *		creates a Virtual CRAT (VCRAT) image
+ *
+ * NOTE: Call kfd_destroy_crat_image to free CRAT image memory
+ *
+ *	@crat_image: VCRAT image created because ACPI does not have a
+ *		     CRAT for this device
+ *	@size: [OUT] size of virtual crat_image
+ *	@flags:	COMPUTE_UNIT_CPU - Create VCRAT for CPU device
+ *		COMPUTE_UNIT_GPU - Create VCRAT for GPU
+ *		(COMPUTE_UNIT_CPU | COMPUTE_UNIT_GPU) - Create VCRAT for APU
+ *			-- this option is not currently implemented.
+ *			The assumption is that all AMD APUs will have CRAT
+ *	@kdev: Valid kfd_node required if flags contain COMPUTE_UNIT_GPU
+ *
+ *	Return 0 if successful else return -ve value
+ */
+int kfd_create_crat_image_virtual(void **crat_image, size_t *size,
+				  int flags, struct kfd_node *kdev,
+				  uint32_t proximity_domain)
+{
+	void *pcrat_image = NULL;
+	int ret = 0, num_nodes;
+	size_t dyn_size;
+
+	if (!crat_image)
+		return -EINVAL;
+
+	*crat_image = NULL;
+
+	/* Allocate the CPU Virtual CRAT size based on the number of online
+	 * nodes. Allocate VCRAT_SIZE_FOR_GPU for GPU virtual CRAT image.
+	 * This should cover all the current conditions. A check is put not
+	 * to overwrite beyond allocated size for GPUs
+	 */
+	switch (flags) {
+	case COMPUTE_UNIT_CPU:
+		num_nodes = num_online_nodes();
+		dyn_size = sizeof(struct crat_header) +
+			num_nodes * (sizeof(struct crat_subtype_computeunit) +
+			sizeof(struct crat_subtype_memory) +
+			(num_nodes - 1) * sizeof(struct crat_subtype_iolink));
+		pcrat_image = kvmalloc(dyn_size, GFP_KERNEL);
+		if (!pcrat_image)
+			return -ENOMEM;
+		*size = dyn_size;
+		pr_debug("CRAT size is %ld", dyn_size);
+		ret = kfd_create_vcrat_image_cpu(pcrat_image, size);
+		break;
+	case COMPUTE_UNIT_GPU:
+		if (!kdev)
+			return -EINVAL;
+		pcrat_image = kvmalloc(VCRAT_SIZE_FOR_GPU, GFP_KERNEL);
+		if (!pcrat_image)
+			return -ENOMEM;
+		*size = VCRAT_SIZE_FOR_GPU;
+		ret = kfd_create_vcrat_image_gpu(pcrat_image, size, kdev,
+						 proximity_domain);
+		break;
+	case (COMPUTE_UNIT_CPU | COMPUTE_UNIT_GPU):
+		/* TODO: */
+		ret = -EINVAL;
+		pr_err("VCRAT not implemented for APU\n");
+		break;
+	default:
+		ret = -EINVAL;
+	}
+
+	if (!ret)
+		*crat_image = pcrat_image;
+	else
+		kvfree(pcrat_image);
+
+	return ret;
+}
+
+
+/* kfd_destroy_crat_image
+ *
+ *	@crat_image: [IN] - crat_image from kfd_create_crat_image_xxx(..)
+ *
+ */
+void kfd_destroy_crat_image(void *crat_image)
+{
+	kvfree(crat_image);
+}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_crat.h b/drivers/gpu/drm/amd/amdkfd/kfd_crat.h
index a374fa3d3ee6..a8ca7ecb6d27 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_crat.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_crat.h
@@ -1,5 +1,6 @@
+/* SPDX-License-Identifier: GPL-2.0 OR MIT */
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -28,11 +29,10 @@
 #pragma pack(1)
 
 /*
- * 4CC signature values for the CRAT and CDIT ACPI tables
+ * 4CC signature value for the CRAT ACPI table
  */
 
 #define CRAT_SIGNATURE	"CRAT"
-#define CDIT_SIGNATURE	"CDIT"
 
 /*
  * Component Resource Association Table (CRAT)
@@ -42,7 +42,9 @@
 #define CRAT_OEMTABLEID_LENGTH	8
 #define CRAT_RESERVED_LENGTH	6
 
-#define CRAT_OEMID_64BIT_MASK ((1ULL << (CRAT_OEMID_LENGTH * 8)) - 1)
+/* Compute Unit flags */
+#define COMPUTE_UNIT_CPU	(1 << 0)  /* Create Virtual CRAT for CPU */
+#define COMPUTE_UNIT_GPU	(1 << 1)  /* Create Virtual CRAT for GPU */
 
 struct crat_header {
 	uint32_t	signature;
@@ -75,6 +77,10 @@ struct crat_header {
 #define CRAT_SUBTYPE_IOLINK_AFFINITY		5
 #define CRAT_SUBTYPE_MAX			6
 
+/*
+ * Do not change the value of CRAT_SIBLINGMAP_SIZE from 32
+ * as it breaks the ABI.
+ */
 #define CRAT_SIBLINGMAP_SIZE	32
 
 /*
@@ -105,7 +111,7 @@ struct crat_subtype_computeunit {
 	uint8_t		wave_front_size;
 	uint8_t		num_banks;
 	uint16_t	micro_engine_id;
-	uint8_t		num_arrays;
+	uint8_t		array_count;
 	uint8_t		num_cu_per_array;
 	uint8_t		num_simd_per_cu;
 	uint8_t		max_slots_scatch_cu;
@@ -127,13 +133,14 @@ struct crat_subtype_memory {
 	uint8_t		length;
 	uint16_t	reserved;
 	uint32_t	flags;
-	uint32_t	promixity_domain;
+	uint32_t	proximity_domain;
 	uint32_t	base_addr_low;
 	uint32_t	base_addr_high;
 	uint32_t	length_low;
 	uint32_t	length_high;
 	uint32_t	width;
-	uint8_t		reserved2[CRAT_MEMORY_RESERVED_LENGTH];
+	uint8_t		visibility_type; /* for virtual (dGPU) CRAT */
+	uint8_t		reserved2[CRAT_MEMORY_RESERVED_LENGTH - 1];
 };
 
 /*
@@ -222,9 +229,13 @@ struct crat_subtype_ccompute {
 /*
  * HSA IO Link Affinity structure and definitions
  */
-#define CRAT_IOLINK_FLAGS_ENABLED	0x00000001
-#define CRAT_IOLINK_FLAGS_COHERENCY	0x00000002
-#define CRAT_IOLINK_FLAGS_RESERVED	0xfffffffc
+#define CRAT_IOLINK_FLAGS_ENABLED		(1 << 0)
+#define CRAT_IOLINK_FLAGS_NON_COHERENT		(1 << 1)
+#define CRAT_IOLINK_FLAGS_NO_ATOMICS_32_BIT	(1 << 2)
+#define CRAT_IOLINK_FLAGS_NO_ATOMICS_64_BIT	(1 << 3)
+#define CRAT_IOLINK_FLAGS_NO_PEER_TO_PEER_DMA	(1 << 4)
+#define CRAT_IOLINK_FLAGS_BI_DIRECTIONAL	(1 << 31)
+#define CRAT_IOLINK_FLAGS_RESERVED_MASK		0x7fffffe0
 
 /*
  * IO interface types
@@ -232,10 +243,23 @@ struct crat_subtype_ccompute {
 #define CRAT_IOLINK_TYPE_UNDEFINED	0
 #define CRAT_IOLINK_TYPE_HYPERTRANSPORT	1
 #define CRAT_IOLINK_TYPE_PCIEXPRESS	2
-#define CRAT_IOLINK_TYPE_OTHER		3
+#define CRAT_IOLINK_TYPE_AMBA		3
+#define CRAT_IOLINK_TYPE_MIPI		4
+#define CRAT_IOLINK_TYPE_QPI_1_1	5
+#define CRAT_IOLINK_TYPE_RESERVED1	6
+#define CRAT_IOLINK_TYPE_RESERVED2	7
+#define CRAT_IOLINK_TYPE_RAPID_IO	8
+#define CRAT_IOLINK_TYPE_INFINIBAND	9
+#define CRAT_IOLINK_TYPE_RESERVED3	10
+#define CRAT_IOLINK_TYPE_XGMI		11
+#define CRAT_IOLINK_TYPE_XGOP		12
+#define CRAT_IOLINK_TYPE_GZ		13
+#define CRAT_IOLINK_TYPE_ETHERNET_RDMA	14
+#define CRAT_IOLINK_TYPE_RDMA_OTHER	15
+#define CRAT_IOLINK_TYPE_OTHER		16
 #define CRAT_IOLINK_TYPE_MAX		255
 
-#define CRAT_IOLINK_RESERVED_LENGTH 24
+#define CRAT_IOLINK_RESERVED_LENGTH	24
 
 struct crat_subtype_iolink {
 	uint8_t		type;
@@ -252,7 +276,8 @@ struct crat_subtype_iolink {
 	uint32_t	minimum_bandwidth_mbs;
 	uint32_t	maximum_bandwidth_mbs;
 	uint32_t	recommended_transfer_size;
-	uint8_t		reserved2[CRAT_IOLINK_RESERVED_LENGTH];
+	uint8_t		reserved2[CRAT_IOLINK_RESERVED_LENGTH - 1];
+	uint8_t		weight_xgmi;
 };
 
 /*
@@ -268,27 +293,28 @@ struct crat_subtype_generic {
 	uint32_t	flags;
 };
 
-/*
- * Component Locality Distance Information Table (CDIT)
- */
-#define CDIT_OEMID_LENGTH	6
-#define CDIT_OEMTABLEID_LENGTH	8
+#pragma pack()
 
-struct cdit_header {
-	uint32_t	signature;
-	uint32_t	length;
-	uint8_t		revision;
-	uint8_t		checksum;
-	uint8_t		oem_id[CDIT_OEMID_LENGTH];
-	uint8_t		oem_table_id[CDIT_OEMTABLEID_LENGTH];
-	uint32_t	oem_revision;
-	uint32_t	creator_id;
-	uint32_t	creator_revision;
-	uint32_t	total_entries;
-	uint16_t	num_domains;
-	uint8_t		entry[1];
-};
+struct kfd_node;
 
-#pragma pack()
+/* Static table to describe GPU Cache information */
+struct kfd_gpu_cache_info {
+	uint32_t	cache_size;
+	uint32_t	cache_level;
+	uint32_t	cache_line_size;
+	uint32_t	flags;
+	/* Indicates how many Compute Units share this cache
+	 * within a SA. Value = 1 indicates the cache is not shared
+	 */
+	uint32_t	num_cu_shared;
+};
+int kfd_get_gpu_cache_info(struct kfd_node *kdev, struct kfd_gpu_cache_info **pcache_info);
+
+void kfd_destroy_crat_image(void *crat_image);
+int kfd_parse_crat_table(void *crat_image, struct list_head *device_list,
+			 uint32_t proximity_domain);
+int kfd_create_crat_image_virtual(void **crat_image, size_t *size,
+				  int flags, struct kfd_node *kdev,
+				  uint32_t proximity_domain);
 
 #endif /* KFD_CRAT_H_INCLUDED */
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_dbgdev.c b/drivers/gpu/drm/amd/amdkfd/kfd_dbgdev.c
deleted file mode 100644
index 0aa021aa0aa1..000000000000
--- a/drivers/gpu/drm/amd/amdkfd/kfd_dbgdev.c
+++ /dev/null
@@ -1,854 +0,0 @@
-/*
- * Copyright 2014 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-
-#include <linux/types.h>
-#include <linux/kernel.h>
-#include <linux/log2.h>
-#include <linux/sched.h>
-#include <linux/slab.h>
-#include <linux/mutex.h>
-#include <linux/device.h>
-
-#include "kfd_pm4_headers.h"
-#include "kfd_pm4_headers_diq.h"
-#include "kfd_kernel_queue.h"
-#include "kfd_priv.h"
-#include "kfd_pm4_opcodes.h"
-#include "cik_regs.h"
-#include "kfd_dbgmgr.h"
-#include "kfd_dbgdev.h"
-#include "kfd_device_queue_manager.h"
-#include "../../radeon/cik_reg.h"
-
-static void dbgdev_address_watch_disable_nodiq(struct kfd_dev *dev)
-{
-	dev->kfd2kgd->address_watch_disable(dev->kgd);
-}
-
-static int dbgdev_diq_submit_ib(struct kfd_dbgdev *dbgdev,
-				unsigned int pasid, uint64_t vmid0_address,
-				uint32_t *packet_buff, size_t size_in_bytes)
-{
-	struct pm4__release_mem *rm_packet;
-	struct pm4__indirect_buffer_pasid *ib_packet;
-	struct kfd_mem_obj *mem_obj;
-	size_t pq_packets_size_in_bytes;
-	union ULARGE_INTEGER *largep;
-	union ULARGE_INTEGER addr;
-	struct kernel_queue *kq;
-	uint64_t *rm_state;
-	unsigned int *ib_packet_buff;
-	int status;
-
-	if (WARN_ON(!size_in_bytes))
-		return -EINVAL;
-
-	kq = dbgdev->kq;
-
-	pq_packets_size_in_bytes = sizeof(struct pm4__release_mem) +
-				sizeof(struct pm4__indirect_buffer_pasid);
-
-	/*
-	 * We acquire a buffer from DIQ
-	 * The receive packet buff will be sitting on the Indirect Buffer
-	 * and in the PQ we put the IB packet + sync packet(s).
-	 */
-	status = kq->ops.acquire_packet_buffer(kq,
-				pq_packets_size_in_bytes / sizeof(uint32_t),
-				&ib_packet_buff);
-	if (status) {
-		pr_err("acquire_packet_buffer failed\n");
-		return status;
-	}
-
-	memset(ib_packet_buff, 0, pq_packets_size_in_bytes);
-
-	ib_packet = (struct pm4__indirect_buffer_pasid *) (ib_packet_buff);
-
-	ib_packet->header.count = 3;
-	ib_packet->header.opcode = IT_INDIRECT_BUFFER_PASID;
-	ib_packet->header.type = PM4_TYPE_3;
-
-	largep = (union ULARGE_INTEGER *) &vmid0_address;
-
-	ib_packet->bitfields2.ib_base_lo = largep->u.low_part >> 2;
-	ib_packet->bitfields3.ib_base_hi = largep->u.high_part;
-
-	ib_packet->control = (1 << 23) | (1 << 31) |
-			((size_in_bytes / sizeof(uint32_t)) & 0xfffff);
-
-	ib_packet->bitfields5.pasid = pasid;
-
-	/*
-	 * for now we use release mem for GPU-CPU synchronization
-	 * Consider WaitRegMem + WriteData as a better alternative
-	 * we get a GART allocations ( gpu/cpu mapping),
-	 * for the sync variable, and wait until:
-	 * (a) Sync with HW
-	 * (b) Sync var is written by CP to mem.
-	 */
-	rm_packet = (struct pm4__release_mem *) (ib_packet_buff +
-			(sizeof(struct pm4__indirect_buffer_pasid) /
-					sizeof(unsigned int)));
-
-	status = kfd_gtt_sa_allocate(dbgdev->dev, sizeof(uint64_t),
-					&mem_obj);
-
-	if (status) {
-		pr_err("Failed to allocate GART memory\n");
-		kq->ops.rollback_packet(kq);
-		return status;
-	}
-
-	rm_state = (uint64_t *) mem_obj->cpu_ptr;
-
-	*rm_state = QUEUESTATE__ACTIVE_COMPLETION_PENDING;
-
-	rm_packet->header.opcode = IT_RELEASE_MEM;
-	rm_packet->header.type = PM4_TYPE_3;
-	rm_packet->header.count = sizeof(struct pm4__release_mem) /
-					sizeof(unsigned int) - 2;
-
-	rm_packet->bitfields2.event_type = CACHE_FLUSH_AND_INV_TS_EVENT;
-	rm_packet->bitfields2.event_index =
-				event_index___release_mem__end_of_pipe;
-
-	rm_packet->bitfields2.cache_policy = cache_policy___release_mem__lru;
-	rm_packet->bitfields2.atc = 0;
-	rm_packet->bitfields2.tc_wb_action_ena = 1;
-
-	addr.quad_part = mem_obj->gpu_addr;
-
-	rm_packet->bitfields4.address_lo_32b = addr.u.low_part >> 2;
-	rm_packet->address_hi = addr.u.high_part;
-
-	rm_packet->bitfields3.data_sel =
-				data_sel___release_mem__send_64_bit_data;
-
-	rm_packet->bitfields3.int_sel =
-			int_sel___release_mem__send_data_after_write_confirm;
-
-	rm_packet->bitfields3.dst_sel =
-			dst_sel___release_mem__memory_controller;
-
-	rm_packet->data_lo = QUEUESTATE__ACTIVE;
-
-	kq->ops.submit_packet(kq);
-
-	/* Wait till CP writes sync code: */
-	status = amdkfd_fence_wait_timeout(
-			(unsigned int *) rm_state,
-			QUEUESTATE__ACTIVE, 1500);
-
-	kfd_gtt_sa_free(dbgdev->dev, mem_obj);
-
-	return status;
-}
-
-static int dbgdev_register_nodiq(struct kfd_dbgdev *dbgdev)
-{
-	/*
-	 * no action is needed in this case,
-	 * just make sure diq will not be used
-	 */
-
-	dbgdev->kq = NULL;
-
-	return 0;
-}
-
-static int dbgdev_register_diq(struct kfd_dbgdev *dbgdev)
-{
-	struct queue_properties properties;
-	unsigned int qid;
-	struct kernel_queue *kq = NULL;
-	int status;
-
-	status = pqm_create_queue(dbgdev->pqm, dbgdev->dev, NULL,
-				&properties, 0, KFD_QUEUE_TYPE_DIQ,
-				&qid);
-
-	if (status) {
-		pr_err("Failed to create DIQ\n");
-		return status;
-	}
-
-	pr_debug("DIQ Created with queue id: %d\n", qid);
-
-	kq = pqm_get_kernel_queue(dbgdev->pqm, qid);
-
-	if (!kq) {
-		pr_err("Error getting DIQ\n");
-		pqm_destroy_queue(dbgdev->pqm, qid);
-		return -EFAULT;
-	}
-
-	dbgdev->kq = kq;
-
-	return status;
-}
-
-static int dbgdev_unregister_nodiq(struct kfd_dbgdev *dbgdev)
-{
-	/* disable watch address */
-	dbgdev_address_watch_disable_nodiq(dbgdev->dev);
-	return 0;
-}
-
-static int dbgdev_unregister_diq(struct kfd_dbgdev *dbgdev)
-{
-	/* todo - disable address watch */
-	int status;
-
-	status = pqm_destroy_queue(dbgdev->pqm,
-			dbgdev->kq->queue->properties.queue_id);
-	dbgdev->kq = NULL;
-
-	return status;
-}
-
-static void dbgdev_address_watch_set_registers(
-			const struct dbg_address_watch_info *adw_info,
-			union TCP_WATCH_ADDR_H_BITS *addrHi,
-			union TCP_WATCH_ADDR_L_BITS *addrLo,
-			union TCP_WATCH_CNTL_BITS *cntl,
-			unsigned int index, unsigned int vmid)
-{
-	union ULARGE_INTEGER addr;
-
-	addr.quad_part = 0;
-	addrHi->u32All = 0;
-	addrLo->u32All = 0;
-	cntl->u32All = 0;
-
-	if (adw_info->watch_mask)
-		cntl->bitfields.mask =
-			(uint32_t) (adw_info->watch_mask[index] &
-					ADDRESS_WATCH_REG_CNTL_DEFAULT_MASK);
-	else
-		cntl->bitfields.mask = ADDRESS_WATCH_REG_CNTL_DEFAULT_MASK;
-
-	addr.quad_part = (unsigned long long) adw_info->watch_address[index];
-
-	addrHi->bitfields.addr = addr.u.high_part &
-					ADDRESS_WATCH_REG_ADDHIGH_MASK;
-	addrLo->bitfields.addr =
-			(addr.u.low_part >> ADDRESS_WATCH_REG_ADDLOW_SHIFT);
-
-	cntl->bitfields.mode = adw_info->watch_mode[index];
-	cntl->bitfields.vmid = (uint32_t) vmid;
-	/* for now assume it is an ATC address */
-	cntl->u32All |= ADDRESS_WATCH_REG_CNTL_ATC_BIT;
-
-	pr_debug("\t\t%20s %08x\n", "set reg mask :", cntl->bitfields.mask);
-	pr_debug("\t\t%20s %08x\n", "set reg add high :",
-			addrHi->bitfields.addr);
-	pr_debug("\t\t%20s %08x\n", "set reg add low :",
-			addrLo->bitfields.addr);
-}
-
-static int dbgdev_address_watch_nodiq(struct kfd_dbgdev *dbgdev,
-				      struct dbg_address_watch_info *adw_info)
-{
-	union TCP_WATCH_ADDR_H_BITS addrHi;
-	union TCP_WATCH_ADDR_L_BITS addrLo;
-	union TCP_WATCH_CNTL_BITS cntl;
-	struct kfd_process_device *pdd;
-	unsigned int i;
-
-	/* taking the vmid for that process on the safe way using pdd */
-	pdd = kfd_get_process_device_data(dbgdev->dev,
-					adw_info->process);
-	if (!pdd) {
-		pr_err("Failed to get pdd for wave control no DIQ\n");
-		return -EFAULT;
-	}
-
-	addrHi.u32All = 0;
-	addrLo.u32All = 0;
-	cntl.u32All = 0;
-
-	if ((adw_info->num_watch_points > MAX_WATCH_ADDRESSES) ||
-			(adw_info->num_watch_points == 0)) {
-		pr_err("num_watch_points is invalid\n");
-		return -EINVAL;
-	}
-
-	if (!adw_info->watch_mode || !adw_info->watch_address) {
-		pr_err("adw_info fields are not valid\n");
-		return -EINVAL;
-	}
-
-	for (i = 0; i < adw_info->num_watch_points; i++) {
-		dbgdev_address_watch_set_registers(adw_info, &addrHi, &addrLo,
-						&cntl, i, pdd->qpd.vmid);
-
-		pr_debug("\t\t%30s\n", "* * * * * * * * * * * * * * * * * *");
-		pr_debug("\t\t%20s %08x\n", "register index :", i);
-		pr_debug("\t\t%20s %08x\n", "vmid is :", pdd->qpd.vmid);
-		pr_debug("\t\t%20s %08x\n", "Address Low is :",
-				addrLo.bitfields.addr);
-		pr_debug("\t\t%20s %08x\n", "Address high is :",
-				addrHi.bitfields.addr);
-		pr_debug("\t\t%20s %08x\n", "Address high is :",
-				addrHi.bitfields.addr);
-		pr_debug("\t\t%20s %08x\n", "Control Mask is :",
-				cntl.bitfields.mask);
-		pr_debug("\t\t%20s %08x\n", "Control Mode is :",
-				cntl.bitfields.mode);
-		pr_debug("\t\t%20s %08x\n", "Control Vmid is :",
-				cntl.bitfields.vmid);
-		pr_debug("\t\t%20s %08x\n", "Control atc  is :",
-				cntl.bitfields.atc);
-		pr_debug("\t\t%30s\n", "* * * * * * * * * * * * * * * * * *");
-
-		pdd->dev->kfd2kgd->address_watch_execute(
-						dbgdev->dev->kgd,
-						i,
-						cntl.u32All,
-						addrHi.u32All,
-						addrLo.u32All);
-	}
-
-	return 0;
-}
-
-static int dbgdev_address_watch_diq(struct kfd_dbgdev *dbgdev,
-				    struct dbg_address_watch_info *adw_info)
-{
-	struct pm4__set_config_reg *packets_vec;
-	union TCP_WATCH_ADDR_H_BITS addrHi;
-	union TCP_WATCH_ADDR_L_BITS addrLo;
-	union TCP_WATCH_CNTL_BITS cntl;
-	struct kfd_mem_obj *mem_obj;
-	unsigned int aw_reg_add_dword;
-	uint32_t *packet_buff_uint;
-	unsigned int i;
-	int status;
-	size_t ib_size = sizeof(struct pm4__set_config_reg) * 4;
-	/* we do not control the vmid in DIQ mode, just a place holder */
-	unsigned int vmid = 0;
-
-	addrHi.u32All = 0;
-	addrLo.u32All = 0;
-	cntl.u32All = 0;
-
-	if ((adw_info->num_watch_points > MAX_WATCH_ADDRESSES) ||
-			(adw_info->num_watch_points == 0)) {
-		pr_err("num_watch_points is invalid\n");
-		return -EINVAL;
-	}
-
-	if (!adw_info->watch_mode || !adw_info->watch_address) {
-		pr_err("adw_info fields are not valid\n");
-		return -EINVAL;
-	}
-
-	status = kfd_gtt_sa_allocate(dbgdev->dev, ib_size, &mem_obj);
-
-	if (status) {
-		pr_err("Failed to allocate GART memory\n");
-		return status;
-	}
-
-	packet_buff_uint = mem_obj->cpu_ptr;
-
-	memset(packet_buff_uint, 0, ib_size);
-
-	packets_vec = (struct pm4__set_config_reg *) (packet_buff_uint);
-
-	packets_vec[0].header.count = 1;
-	packets_vec[0].header.opcode = IT_SET_CONFIG_REG;
-	packets_vec[0].header.type = PM4_TYPE_3;
-	packets_vec[0].bitfields2.vmid_shift = ADDRESS_WATCH_CNTL_OFFSET;
-	packets_vec[0].bitfields2.insert_vmid = 1;
-	packets_vec[1].ordinal1 = packets_vec[0].ordinal1;
-	packets_vec[1].bitfields2.insert_vmid = 0;
-	packets_vec[2].ordinal1 = packets_vec[0].ordinal1;
-	packets_vec[2].bitfields2.insert_vmid = 0;
-	packets_vec[3].ordinal1 = packets_vec[0].ordinal1;
-	packets_vec[3].bitfields2.vmid_shift = ADDRESS_WATCH_CNTL_OFFSET;
-	packets_vec[3].bitfields2.insert_vmid = 1;
-
-	for (i = 0; i < adw_info->num_watch_points; i++) {
-		dbgdev_address_watch_set_registers(adw_info,
-						&addrHi,
-						&addrLo,
-						&cntl,
-						i,
-						vmid);
-
-		pr_debug("\t\t%30s\n", "* * * * * * * * * * * * * * * * * *");
-		pr_debug("\t\t%20s %08x\n", "register index :", i);
-		pr_debug("\t\t%20s %08x\n", "vmid is :", vmid);
-		pr_debug("\t\t%20s %p\n", "Add ptr is :",
-				adw_info->watch_address);
-		pr_debug("\t\t%20s %08llx\n", "Add     is :",
-				adw_info->watch_address[i]);
-		pr_debug("\t\t%20s %08x\n", "Address Low is :",
-				addrLo.bitfields.addr);
-		pr_debug("\t\t%20s %08x\n", "Address high is :",
-				addrHi.bitfields.addr);
-		pr_debug("\t\t%20s %08x\n", "Control Mask is :",
-				cntl.bitfields.mask);
-		pr_debug("\t\t%20s %08x\n", "Control Mode is :",
-				cntl.bitfields.mode);
-		pr_debug("\t\t%20s %08x\n", "Control Vmid is :",
-				cntl.bitfields.vmid);
-		pr_debug("\t\t%20s %08x\n", "Control atc  is :",
-				cntl.bitfields.atc);
-		pr_debug("\t\t%30s\n", "* * * * * * * * * * * * * * * * * *");
-
-		aw_reg_add_dword =
-				dbgdev->dev->kfd2kgd->address_watch_get_offset(
-					dbgdev->dev->kgd,
-					i,
-					ADDRESS_WATCH_REG_CNTL);
-
-		packets_vec[0].bitfields2.reg_offset =
-					aw_reg_add_dword - AMD_CONFIG_REG_BASE;
-
-		packets_vec[0].reg_data[0] = cntl.u32All;
-
-		aw_reg_add_dword =
-				dbgdev->dev->kfd2kgd->address_watch_get_offset(
-					dbgdev->dev->kgd,
-					i,
-					ADDRESS_WATCH_REG_ADDR_HI);
-
-		packets_vec[1].bitfields2.reg_offset =
-					aw_reg_add_dword - AMD_CONFIG_REG_BASE;
-		packets_vec[1].reg_data[0] = addrHi.u32All;
-
-		aw_reg_add_dword =
-				dbgdev->dev->kfd2kgd->address_watch_get_offset(
-					dbgdev->dev->kgd,
-					i,
-					ADDRESS_WATCH_REG_ADDR_LO);
-
-		packets_vec[2].bitfields2.reg_offset =
-				aw_reg_add_dword - AMD_CONFIG_REG_BASE;
-		packets_vec[2].reg_data[0] = addrLo.u32All;
-
-		/* enable watch flag if address is not zero*/
-		if (adw_info->watch_address[i] > 0)
-			cntl.bitfields.valid = 1;
-		else
-			cntl.bitfields.valid = 0;
-
-		aw_reg_add_dword =
-				dbgdev->dev->kfd2kgd->address_watch_get_offset(
-					dbgdev->dev->kgd,
-					i,
-					ADDRESS_WATCH_REG_CNTL);
-
-		packets_vec[3].bitfields2.reg_offset =
-					aw_reg_add_dword - AMD_CONFIG_REG_BASE;
-		packets_vec[3].reg_data[0] = cntl.u32All;
-
-		status = dbgdev_diq_submit_ib(
-					dbgdev,
-					adw_info->process->pasid,
-					mem_obj->gpu_addr,
-					packet_buff_uint,
-					ib_size);
-
-		if (status) {
-			pr_err("Failed to submit IB to DIQ\n");
-			break;
-		}
-	}
-
-	kfd_gtt_sa_free(dbgdev->dev, mem_obj);
-	return status;
-}
-
-static int dbgdev_wave_control_set_registers(
-				struct dbg_wave_control_info *wac_info,
-				union SQ_CMD_BITS *in_reg_sq_cmd,
-				union GRBM_GFX_INDEX_BITS *in_reg_gfx_index)
-{
-	int status = 0;
-	union SQ_CMD_BITS reg_sq_cmd;
-	union GRBM_GFX_INDEX_BITS reg_gfx_index;
-	struct HsaDbgWaveMsgAMDGen2 *pMsg;
-
-	reg_sq_cmd.u32All = 0;
-	reg_gfx_index.u32All = 0;
-	pMsg = &wac_info->dbgWave_msg.DbgWaveMsg.WaveMsgInfoGen2;
-
-	switch (wac_info->mode) {
-	/* Send command to single wave */
-	case HSA_DBG_WAVEMODE_SINGLE:
-		/*
-		 * Limit access to the process waves only,
-		 * by setting vmid check
-		 */
-		reg_sq_cmd.bits.check_vmid = 1;
-		reg_sq_cmd.bits.simd_id = pMsg->ui32.SIMD;
-		reg_sq_cmd.bits.wave_id = pMsg->ui32.WaveId;
-		reg_sq_cmd.bits.mode = SQ_IND_CMD_MODE_SINGLE;
-
-		reg_gfx_index.bits.sh_index = pMsg->ui32.ShaderArray;
-		reg_gfx_index.bits.se_index = pMsg->ui32.ShaderEngine;
-		reg_gfx_index.bits.instance_index = pMsg->ui32.HSACU;
-
-		break;
-
-	/* Send command to all waves with matching VMID */
-	case HSA_DBG_WAVEMODE_BROADCAST_PROCESS:
-
-		reg_gfx_index.bits.sh_broadcast_writes = 1;
-		reg_gfx_index.bits.se_broadcast_writes = 1;
-		reg_gfx_index.bits.instance_broadcast_writes = 1;
-
-		reg_sq_cmd.bits.mode = SQ_IND_CMD_MODE_BROADCAST;
-
-		break;
-
-	/* Send command to all CU waves with matching VMID */
-	case HSA_DBG_WAVEMODE_BROADCAST_PROCESS_CU:
-
-		reg_sq_cmd.bits.check_vmid = 1;
-		reg_sq_cmd.bits.mode = SQ_IND_CMD_MODE_BROADCAST;
-
-		reg_gfx_index.bits.sh_index = pMsg->ui32.ShaderArray;
-		reg_gfx_index.bits.se_index = pMsg->ui32.ShaderEngine;
-		reg_gfx_index.bits.instance_index = pMsg->ui32.HSACU;
-
-		break;
-
-	default:
-		return -EINVAL;
-	}
-
-	switch (wac_info->operand) {
-	case HSA_DBG_WAVEOP_HALT:
-		reg_sq_cmd.bits.cmd = SQ_IND_CMD_CMD_HALT;
-		break;
-
-	case HSA_DBG_WAVEOP_RESUME:
-		reg_sq_cmd.bits.cmd = SQ_IND_CMD_CMD_RESUME;
-		break;
-
-	case HSA_DBG_WAVEOP_KILL:
-		reg_sq_cmd.bits.cmd = SQ_IND_CMD_CMD_KILL;
-		break;
-
-	case HSA_DBG_WAVEOP_DEBUG:
-		reg_sq_cmd.bits.cmd = SQ_IND_CMD_CMD_DEBUG;
-		break;
-
-	case HSA_DBG_WAVEOP_TRAP:
-		if (wac_info->trapId < MAX_TRAPID) {
-			reg_sq_cmd.bits.cmd = SQ_IND_CMD_CMD_TRAP;
-			reg_sq_cmd.bits.trap_id = wac_info->trapId;
-		} else {
-			status = -EINVAL;
-		}
-		break;
-
-	default:
-		status = -EINVAL;
-		break;
-	}
-
-	if (status == 0) {
-		*in_reg_sq_cmd = reg_sq_cmd;
-		*in_reg_gfx_index = reg_gfx_index;
-	}
-
-	return status;
-}
-
-static int dbgdev_wave_control_diq(struct kfd_dbgdev *dbgdev,
-					struct dbg_wave_control_info *wac_info)
-{
-
-	int status;
-	union SQ_CMD_BITS reg_sq_cmd;
-	union GRBM_GFX_INDEX_BITS reg_gfx_index;
-	struct kfd_mem_obj *mem_obj;
-	uint32_t *packet_buff_uint;
-	struct pm4__set_config_reg *packets_vec;
-	size_t ib_size = sizeof(struct pm4__set_config_reg) * 3;
-
-	reg_sq_cmd.u32All = 0;
-
-	status = dbgdev_wave_control_set_registers(wac_info, &reg_sq_cmd,
-							&reg_gfx_index);
-	if (status) {
-		pr_err("Failed to set wave control registers\n");
-		return status;
-	}
-
-	/* we do not control the VMID in DIQ, so reset it to a known value */
-	reg_sq_cmd.bits.vm_id = 0;
-
-	pr_debug("\t\t %30s\n", "* * * * * * * * * * * * * * * * * *");
-
-	pr_debug("\t\t mode      is: %u\n", wac_info->mode);
-	pr_debug("\t\t operand   is: %u\n", wac_info->operand);
-	pr_debug("\t\t trap id   is: %u\n", wac_info->trapId);
-	pr_debug("\t\t msg value is: %u\n",
-			wac_info->dbgWave_msg.DbgWaveMsg.WaveMsgInfoGen2.Value);
-	pr_debug("\t\t vmid      is: N/A\n");
-
-	pr_debug("\t\t chk_vmid  is : %u\n", reg_sq_cmd.bitfields.check_vmid);
-	pr_debug("\t\t command   is : %u\n", reg_sq_cmd.bitfields.cmd);
-	pr_debug("\t\t queue id  is : %u\n", reg_sq_cmd.bitfields.queue_id);
-	pr_debug("\t\t simd id   is : %u\n", reg_sq_cmd.bitfields.simd_id);
-	pr_debug("\t\t mode      is : %u\n", reg_sq_cmd.bitfields.mode);
-	pr_debug("\t\t vm_id     is : %u\n", reg_sq_cmd.bitfields.vm_id);
-	pr_debug("\t\t wave_id   is : %u\n", reg_sq_cmd.bitfields.wave_id);
-
-	pr_debug("\t\t ibw       is : %u\n",
-			reg_gfx_index.bitfields.instance_broadcast_writes);
-	pr_debug("\t\t ii        is : %u\n",
-			reg_gfx_index.bitfields.instance_index);
-	pr_debug("\t\t sebw      is : %u\n",
-			reg_gfx_index.bitfields.se_broadcast_writes);
-	pr_debug("\t\t se_ind    is : %u\n", reg_gfx_index.bitfields.se_index);
-	pr_debug("\t\t sh_ind    is : %u\n", reg_gfx_index.bitfields.sh_index);
-	pr_debug("\t\t sbw       is : %u\n",
-			reg_gfx_index.bitfields.sh_broadcast_writes);
-
-	pr_debug("\t\t %30s\n", "* * * * * * * * * * * * * * * * * *");
-
-	status = kfd_gtt_sa_allocate(dbgdev->dev, ib_size, &mem_obj);
-
-	if (status != 0) {
-		pr_err("Failed to allocate GART memory\n");
-		return status;
-	}
-
-	packet_buff_uint = mem_obj->cpu_ptr;
-
-	memset(packet_buff_uint, 0, ib_size);
-
-	packets_vec =  (struct pm4__set_config_reg *) packet_buff_uint;
-	packets_vec[0].header.count = 1;
-	packets_vec[0].header.opcode = IT_SET_UCONFIG_REG;
-	packets_vec[0].header.type = PM4_TYPE_3;
-	packets_vec[0].bitfields2.reg_offset =
-			GRBM_GFX_INDEX / (sizeof(uint32_t)) -
-				USERCONFIG_REG_BASE;
-
-	packets_vec[0].bitfields2.insert_vmid = 0;
-	packets_vec[0].reg_data[0] = reg_gfx_index.u32All;
-
-	packets_vec[1].header.count = 1;
-	packets_vec[1].header.opcode = IT_SET_CONFIG_REG;
-	packets_vec[1].header.type = PM4_TYPE_3;
-	packets_vec[1].bitfields2.reg_offset = SQ_CMD / (sizeof(uint32_t)) -
-						AMD_CONFIG_REG_BASE;
-
-	packets_vec[1].bitfields2.vmid_shift = SQ_CMD_VMID_OFFSET;
-	packets_vec[1].bitfields2.insert_vmid = 1;
-	packets_vec[1].reg_data[0] = reg_sq_cmd.u32All;
-
-	/* Restore the GRBM_GFX_INDEX register */
-
-	reg_gfx_index.u32All = 0;
-	reg_gfx_index.bits.sh_broadcast_writes = 1;
-	reg_gfx_index.bits.instance_broadcast_writes = 1;
-	reg_gfx_index.bits.se_broadcast_writes = 1;
-
-
-	packets_vec[2].ordinal1 = packets_vec[0].ordinal1;
-	packets_vec[2].bitfields2.reg_offset =
-				GRBM_GFX_INDEX / (sizeof(uint32_t)) -
-					USERCONFIG_REG_BASE;
-
-	packets_vec[2].bitfields2.insert_vmid = 0;
-	packets_vec[2].reg_data[0] = reg_gfx_index.u32All;
-
-	status = dbgdev_diq_submit_ib(
-			dbgdev,
-			wac_info->process->pasid,
-			mem_obj->gpu_addr,
-			packet_buff_uint,
-			ib_size);
-
-	if (status)
-		pr_err("Failed to submit IB to DIQ\n");
-
-	kfd_gtt_sa_free(dbgdev->dev, mem_obj);
-
-	return status;
-}
-
-static int dbgdev_wave_control_nodiq(struct kfd_dbgdev *dbgdev,
-					struct dbg_wave_control_info *wac_info)
-{
-	int status;
-	union SQ_CMD_BITS reg_sq_cmd;
-	union GRBM_GFX_INDEX_BITS reg_gfx_index;
-	struct kfd_process_device *pdd;
-
-	reg_sq_cmd.u32All = 0;
-
-	/* taking the VMID for that process on the safe way using PDD */
-	pdd = kfd_get_process_device_data(dbgdev->dev, wac_info->process);
-
-	if (!pdd) {
-		pr_err("Failed to get pdd for wave control no DIQ\n");
-		return -EFAULT;
-	}
-	status = dbgdev_wave_control_set_registers(wac_info, &reg_sq_cmd,
-							&reg_gfx_index);
-	if (status) {
-		pr_err("Failed to set wave control registers\n");
-		return status;
-	}
-
-	/* for non DIQ we need to patch the VMID: */
-
-	reg_sq_cmd.bits.vm_id = pdd->qpd.vmid;
-
-	pr_debug("\t\t %30s\n", "* * * * * * * * * * * * * * * * * *");
-
-	pr_debug("\t\t mode      is: %u\n", wac_info->mode);
-	pr_debug("\t\t operand   is: %u\n", wac_info->operand);
-	pr_debug("\t\t trap id   is: %u\n", wac_info->trapId);
-	pr_debug("\t\t msg value is: %u\n",
-			wac_info->dbgWave_msg.DbgWaveMsg.WaveMsgInfoGen2.Value);
-	pr_debug("\t\t vmid      is: %u\n", pdd->qpd.vmid);
-
-	pr_debug("\t\t chk_vmid  is : %u\n", reg_sq_cmd.bitfields.check_vmid);
-	pr_debug("\t\t command   is : %u\n", reg_sq_cmd.bitfields.cmd);
-	pr_debug("\t\t queue id  is : %u\n", reg_sq_cmd.bitfields.queue_id);
-	pr_debug("\t\t simd id   is : %u\n", reg_sq_cmd.bitfields.simd_id);
-	pr_debug("\t\t mode      is : %u\n", reg_sq_cmd.bitfields.mode);
-	pr_debug("\t\t vm_id     is : %u\n", reg_sq_cmd.bitfields.vm_id);
-	pr_debug("\t\t wave_id   is : %u\n", reg_sq_cmd.bitfields.wave_id);
-
-	pr_debug("\t\t ibw       is : %u\n",
-			reg_gfx_index.bitfields.instance_broadcast_writes);
-	pr_debug("\t\t ii        is : %u\n",
-			reg_gfx_index.bitfields.instance_index);
-	pr_debug("\t\t sebw      is : %u\n",
-			reg_gfx_index.bitfields.se_broadcast_writes);
-	pr_debug("\t\t se_ind    is : %u\n", reg_gfx_index.bitfields.se_index);
-	pr_debug("\t\t sh_ind    is : %u\n", reg_gfx_index.bitfields.sh_index);
-	pr_debug("\t\t sbw       is : %u\n",
-			reg_gfx_index.bitfields.sh_broadcast_writes);
-
-	pr_debug("\t\t %30s\n", "* * * * * * * * * * * * * * * * * *");
-
-	return dbgdev->dev->kfd2kgd->wave_control_execute(dbgdev->dev->kgd,
-							reg_gfx_index.u32All,
-							reg_sq_cmd.u32All);
-}
-
-int dbgdev_wave_reset_wavefronts(struct kfd_dev *dev, struct kfd_process *p)
-{
-	int status = 0;
-	unsigned int vmid;
-	union SQ_CMD_BITS reg_sq_cmd;
-	union GRBM_GFX_INDEX_BITS reg_gfx_index;
-	struct kfd_process_device *pdd;
-	struct dbg_wave_control_info wac_info;
-	int temp;
-	int first_vmid_to_scan = 8;
-	int last_vmid_to_scan = 15;
-
-	first_vmid_to_scan = ffs(dev->shared_resources.compute_vmid_bitmap) - 1;
-	temp = dev->shared_resources.compute_vmid_bitmap >> first_vmid_to_scan;
-	last_vmid_to_scan = first_vmid_to_scan + ffz(temp);
-
-	reg_sq_cmd.u32All = 0;
-	status = 0;
-
-	wac_info.mode = HSA_DBG_WAVEMODE_BROADCAST_PROCESS;
-	wac_info.operand = HSA_DBG_WAVEOP_KILL;
-
-	pr_debug("Killing all process wavefronts\n");
-
-	/* Scan all registers in the range ATC_VMID8_PASID_MAPPING ..
-	 * ATC_VMID15_PASID_MAPPING
-	 * to check which VMID the current process is mapped to.
-	 */
-
-	for (vmid = first_vmid_to_scan; vmid <= last_vmid_to_scan; vmid++) {
-		if (dev->kfd2kgd->get_atc_vmid_pasid_mapping_valid
-				(dev->kgd, vmid)) {
-			if (dev->kfd2kgd->get_atc_vmid_pasid_mapping_pasid
-					(dev->kgd, vmid) == p->pasid) {
-				pr_debug("Killing wave fronts of vmid %d and pasid %d\n",
-						vmid, p->pasid);
-				break;
-			}
-		}
-	}
-
-	if (vmid > last_vmid_to_scan) {
-		pr_err("Didn't find vmid for pasid %d\n", p->pasid);
-		return -EFAULT;
-	}
-
-	/* taking the VMID for that process on the safe way using PDD */
-	pdd = kfd_get_process_device_data(dev, p);
-	if (!pdd)
-		return -EFAULT;
-
-	status = dbgdev_wave_control_set_registers(&wac_info, &reg_sq_cmd,
-			&reg_gfx_index);
-	if (status != 0)
-		return -EINVAL;
-
-	/* for non DIQ we need to patch the VMID: */
-	reg_sq_cmd.bits.vm_id = vmid;
-
-	dev->kfd2kgd->wave_control_execute(dev->kgd,
-					reg_gfx_index.u32All,
-					reg_sq_cmd.u32All);
-
-	return 0;
-}
-
-void kfd_dbgdev_init(struct kfd_dbgdev *pdbgdev, struct kfd_dev *pdev,
-			enum DBGDEV_TYPE type)
-{
-	pdbgdev->dev = pdev;
-	pdbgdev->kq = NULL;
-	pdbgdev->type = type;
-	pdbgdev->pqm = NULL;
-
-	switch (type) {
-	case DBGDEV_TYPE_NODIQ:
-		pdbgdev->dbgdev_register = dbgdev_register_nodiq;
-		pdbgdev->dbgdev_unregister = dbgdev_unregister_nodiq;
-		pdbgdev->dbgdev_wave_control = dbgdev_wave_control_nodiq;
-		pdbgdev->dbgdev_address_watch = dbgdev_address_watch_nodiq;
-		break;
-	case DBGDEV_TYPE_DIQ:
-	default:
-		pdbgdev->dbgdev_register = dbgdev_register_diq;
-		pdbgdev->dbgdev_unregister = dbgdev_unregister_diq;
-		pdbgdev->dbgdev_wave_control =  dbgdev_wave_control_diq;
-		pdbgdev->dbgdev_address_watch = dbgdev_address_watch_diq;
-		break;
-	}
-
-}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_dbgdev.h b/drivers/gpu/drm/amd/amdkfd/kfd_dbgdev.h
deleted file mode 100644
index 03424c20920c..000000000000
--- a/drivers/gpu/drm/amd/amdkfd/kfd_dbgdev.h
+++ /dev/null
@@ -1,193 +0,0 @@
-/*
- * Copyright 2014 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- */
-
-#ifndef KFD_DBGDEV_H_
-#define KFD_DBGDEV_H_
-
-enum {
-	SQ_CMD_VMID_OFFSET = 28,
-	ADDRESS_WATCH_CNTL_OFFSET = 24
-};
-
-enum {
-	PRIV_QUEUE_SYNC_TIME_MS = 200
-};
-
-/* CONTEXT reg space definition */
-enum {
-	CONTEXT_REG_BASE = 0xA000,
-	CONTEXT_REG_END = 0xA400,
-	CONTEXT_REG_SIZE = CONTEXT_REG_END - CONTEXT_REG_BASE
-};
-
-/* USER CONFIG reg space definition */
-enum {
-	USERCONFIG_REG_BASE = 0xC000,
-	USERCONFIG_REG_END = 0x10000,
-	USERCONFIG_REG_SIZE = USERCONFIG_REG_END - USERCONFIG_REG_BASE
-};
-
-/* CONFIG reg space definition */
-enum {
-	AMD_CONFIG_REG_BASE = 0x2000,	/* in dwords */
-	AMD_CONFIG_REG_END = 0x2B00,
-	AMD_CONFIG_REG_SIZE = AMD_CONFIG_REG_END - AMD_CONFIG_REG_BASE
-};
-
-/* SH reg space definition */
-enum {
-	SH_REG_BASE = 0x2C00,
-	SH_REG_END = 0x3000,
-	SH_REG_SIZE = SH_REG_END - SH_REG_BASE
-};
-
-enum SQ_IND_CMD_CMD {
-	SQ_IND_CMD_CMD_NULL = 0x00000000,
-	SQ_IND_CMD_CMD_HALT = 0x00000001,
-	SQ_IND_CMD_CMD_RESUME = 0x00000002,
-	SQ_IND_CMD_CMD_KILL = 0x00000003,
-	SQ_IND_CMD_CMD_DEBUG = 0x00000004,
-	SQ_IND_CMD_CMD_TRAP = 0x00000005,
-};
-
-enum SQ_IND_CMD_MODE {
-	SQ_IND_CMD_MODE_SINGLE = 0x00000000,
-	SQ_IND_CMD_MODE_BROADCAST = 0x00000001,
-	SQ_IND_CMD_MODE_BROADCAST_QUEUE = 0x00000002,
-	SQ_IND_CMD_MODE_BROADCAST_PIPE = 0x00000003,
-	SQ_IND_CMD_MODE_BROADCAST_ME = 0x00000004,
-};
-
-union SQ_IND_INDEX_BITS {
-	struct {
-		uint32_t wave_id:4;
-		uint32_t simd_id:2;
-		uint32_t thread_id:6;
-		 uint32_t:1;
-		uint32_t force_read:1;
-		uint32_t read_timeout:1;
-		uint32_t unindexed:1;
-		uint32_t index:16;
-
-	} bitfields, bits;
-	uint32_t u32All;
-	signed int i32All;
-	float f32All;
-};
-
-union SQ_IND_CMD_BITS {
-	struct {
-		uint32_t data:32;
-	} bitfields, bits;
-	uint32_t u32All;
-	signed int i32All;
-	float f32All;
-};
-
-union SQ_CMD_BITS {
-	struct {
-		uint32_t cmd:3;
-		 uint32_t:1;
-		uint32_t mode:3;
-		uint32_t check_vmid:1;
-		uint32_t trap_id:3;
-		 uint32_t:5;
-		uint32_t wave_id:4;
-		uint32_t simd_id:2;
-		 uint32_t:2;
-		uint32_t queue_id:3;
-		 uint32_t:1;
-		uint32_t vm_id:4;
-	} bitfields, bits;
-	uint32_t u32All;
-	signed int i32All;
-	float f32All;
-};
-
-union SQ_IND_DATA_BITS {
-	struct {
-		uint32_t data:32;
-	} bitfields, bits;
-	uint32_t u32All;
-	signed int i32All;
-	float f32All;
-};
-
-union GRBM_GFX_INDEX_BITS {
-	struct {
-		uint32_t instance_index:8;
-		uint32_t sh_index:8;
-		uint32_t se_index:8;
-		 uint32_t:5;
-		uint32_t sh_broadcast_writes:1;
-		uint32_t instance_broadcast_writes:1;
-		uint32_t se_broadcast_writes:1;
-	} bitfields, bits;
-	uint32_t u32All;
-	signed int i32All;
-	float f32All;
-};
-
-union TCP_WATCH_ADDR_H_BITS {
-	struct {
-		uint32_t addr:16;
-		 uint32_t:16;
-
-	} bitfields, bits;
-	uint32_t u32All;
-	signed int i32All;
-	float f32All;
-};
-
-union TCP_WATCH_ADDR_L_BITS {
-	struct {
-		uint32_t:6;
-		uint32_t addr:26;
-	} bitfields, bits;
-	uint32_t u32All;
-	signed int i32All;
-	float f32All;
-};
-
-enum {
-	QUEUESTATE__INVALID = 0, /* so by default we'll get invalid state */
-	QUEUESTATE__ACTIVE_COMPLETION_PENDING,
-	QUEUESTATE__ACTIVE
-};
-
-union ULARGE_INTEGER {
-	struct {
-		uint32_t low_part;
-		uint32_t high_part;
-	} u;
-	unsigned long long quad_part;
-};
-
-
-#define KFD_CIK_VMID_START_OFFSET (8)
-#define KFD_CIK_VMID_END_OFFSET (KFD_CIK_VMID_START_OFFSET + (8))
-
-
-void kfd_dbgdev_init(struct kfd_dbgdev *pdbgdev, struct kfd_dev *pdev,
-			enum DBGDEV_TYPE type);
-
-#endif	/* KFD_DBGDEV_H_ */
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_dbgmgr.c b/drivers/gpu/drm/amd/amdkfd/kfd_dbgmgr.c
deleted file mode 100644
index 3da25f7bda6b..000000000000
--- a/drivers/gpu/drm/amd/amdkfd/kfd_dbgmgr.c
+++ /dev/null
@@ -1,157 +0,0 @@
-/*
- * Copyright 2014 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-#include <linux/types.h>
-#include <linux/kernel.h>
-#include <linux/log2.h>
-#include <linux/sched.h>
-#include <linux/slab.h>
-#include <linux/device.h>
-
-#include "kfd_priv.h"
-#include "cik_regs.h"
-#include "kfd_pm4_headers.h"
-#include "kfd_pm4_headers_diq.h"
-#include "kfd_dbgmgr.h"
-#include "kfd_dbgdev.h"
-
-static DEFINE_MUTEX(kfd_dbgmgr_mutex);
-
-struct mutex *kfd_get_dbgmgr_mutex(void)
-{
-	return &kfd_dbgmgr_mutex;
-}
-
-
-static void kfd_dbgmgr_uninitialize(struct kfd_dbgmgr *pmgr)
-{
-	kfree(pmgr->dbgdev);
-
-	pmgr->dbgdev = NULL;
-	pmgr->pasid = 0;
-	pmgr->dev = NULL;
-}
-
-void kfd_dbgmgr_destroy(struct kfd_dbgmgr *pmgr)
-{
-	if (pmgr) {
-		kfd_dbgmgr_uninitialize(pmgr);
-		kfree(pmgr);
-	}
-}
-
-bool kfd_dbgmgr_create(struct kfd_dbgmgr **ppmgr, struct kfd_dev *pdev)
-{
-	enum DBGDEV_TYPE type = DBGDEV_TYPE_DIQ;
-	struct kfd_dbgmgr *new_buff;
-
-	if (WARN_ON(!pdev->init_complete))
-		return false;
-
-	new_buff = kfd_alloc_struct(new_buff);
-	if (!new_buff) {
-		pr_err("Failed to allocate dbgmgr instance\n");
-		return false;
-	}
-
-	new_buff->pasid = 0;
-	new_buff->dev = pdev;
-	new_buff->dbgdev = kfd_alloc_struct(new_buff->dbgdev);
-	if (!new_buff->dbgdev) {
-		pr_err("Failed to allocate dbgdev instance\n");
-		kfree(new_buff);
-		return false;
-	}
-
-	/* get actual type of DBGDevice cpsch or not */
-	if (sched_policy == KFD_SCHED_POLICY_NO_HWS)
-		type = DBGDEV_TYPE_NODIQ;
-
-	kfd_dbgdev_init(new_buff->dbgdev, pdev, type);
-	*ppmgr = new_buff;
-
-	return true;
-}
-
-long kfd_dbgmgr_register(struct kfd_dbgmgr *pmgr, struct kfd_process *p)
-{
-	if (pmgr->pasid != 0) {
-		pr_debug("H/W debugger is already active using pasid %d\n",
-				pmgr->pasid);
-		return -EBUSY;
-	}
-
-	/* remember pasid */
-	pmgr->pasid = p->pasid;
-
-	/* provide the pqm for diq generation */
-	pmgr->dbgdev->pqm = &p->pqm;
-
-	/* activate the actual registering */
-	pmgr->dbgdev->dbgdev_register(pmgr->dbgdev);
-
-	return 0;
-}
-
-long kfd_dbgmgr_unregister(struct kfd_dbgmgr *pmgr, struct kfd_process *p)
-{
-	/* Is the requests coming from the already registered process? */
-	if (pmgr->pasid != p->pasid) {
-		pr_debug("H/W debugger is not registered by calling pasid %d\n",
-				p->pasid);
-		return -EINVAL;
-	}
-
-	pmgr->dbgdev->dbgdev_unregister(pmgr->dbgdev);
-
-	pmgr->pasid = 0;
-
-	return 0;
-}
-
-long kfd_dbgmgr_wave_control(struct kfd_dbgmgr *pmgr,
-				struct dbg_wave_control_info *wac_info)
-{
-	/* Is the requests coming from the already registered process? */
-	if (pmgr->pasid != wac_info->process->pasid) {
-		pr_debug("H/W debugger support was not registered for requester pasid %d\n",
-				wac_info->process->pasid);
-		return -EINVAL;
-	}
-
-	return (long) pmgr->dbgdev->dbgdev_wave_control(pmgr->dbgdev, wac_info);
-}
-
-long kfd_dbgmgr_address_watch(struct kfd_dbgmgr *pmgr,
-				struct dbg_address_watch_info *adw_info)
-{
-	/* Is the requests coming from the already registered process? */
-	if (pmgr->pasid != adw_info->process->pasid) {
-		pr_debug("H/W debugger support was not registered for requester pasid %d\n",
-				adw_info->process->pasid);
-		return -EINVAL;
-	}
-
-	return (long) pmgr->dbgdev->dbgdev_address_watch(pmgr->dbgdev,
-							adw_info);
-}
-
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_dbgmgr.h b/drivers/gpu/drm/amd/amdkfd/kfd_dbgmgr.h
deleted file mode 100644
index a04a1fe1d0d9..000000000000
--- a/drivers/gpu/drm/amd/amdkfd/kfd_dbgmgr.h
+++ /dev/null
@@ -1,293 +0,0 @@
-/*
- * Copyright 2014 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-
-#ifndef KFD_DBGMGR_H_
-#define KFD_DBGMGR_H_
-
-#include "kfd_priv.h"
-
-/* must align with hsakmttypes definition */
-#pragma pack(push, 4)
-
-enum HSA_DBG_WAVEOP {
-	HSA_DBG_WAVEOP_HALT = 1,   /* Halts a wavefront */
-	HSA_DBG_WAVEOP_RESUME = 2, /* Resumes a wavefront */
-	HSA_DBG_WAVEOP_KILL = 3,   /* Kills a wavefront */
-	HSA_DBG_WAVEOP_DEBUG = 4,  /* Causes wavefront to enter dbg mode */
-	HSA_DBG_WAVEOP_TRAP = 5,   /* Causes wavefront to take a trap */
-	HSA_DBG_NUM_WAVEOP = 5,
-	HSA_DBG_MAX_WAVEOP = 0xFFFFFFFF
-};
-
-enum HSA_DBG_WAVEMODE {
-	/* send command to a single wave */
-	HSA_DBG_WAVEMODE_SINGLE = 0,
-	/*
-	 * Broadcast to all wavefronts of all processes is not
-	 * supported for HSA user mode
-	 */
-
-	/* send to waves within current process */
-	HSA_DBG_WAVEMODE_BROADCAST_PROCESS = 2,
-	/* send to waves within current process on CU  */
-	HSA_DBG_WAVEMODE_BROADCAST_PROCESS_CU = 3,
-	HSA_DBG_NUM_WAVEMODE = 3,
-	HSA_DBG_MAX_WAVEMODE = 0xFFFFFFFF
-};
-
-enum HSA_DBG_WAVEMSG_TYPE {
-	HSA_DBG_WAVEMSG_AUTO = 0,
-	HSA_DBG_WAVEMSG_USER = 1,
-	HSA_DBG_WAVEMSG_ERROR = 2,
-	HSA_DBG_NUM_WAVEMSG,
-	HSA_DBG_MAX_WAVEMSG = 0xFFFFFFFF
-};
-
-enum HSA_DBG_WATCH_MODE {
-	HSA_DBG_WATCH_READ = 0,		/* Read operations only */
-	HSA_DBG_WATCH_NONREAD = 1,	/* Write or Atomic operations only */
-	HSA_DBG_WATCH_ATOMIC = 2,	/* Atomic Operations only */
-	HSA_DBG_WATCH_ALL = 3,		/* Read, Write or Atomic operations */
-	HSA_DBG_WATCH_NUM,
-	HSA_DBG_WATCH_SIZE = 0xFFFFFFFF
-};
-
-/* This structure is hardware specific and may change in the future */
-struct HsaDbgWaveMsgAMDGen2 {
-	union {
-		struct ui32 {
-			uint32_t UserData:8;	/* user data */
-			uint32_t ShaderArray:1;	/* Shader array */
-			uint32_t Priv:1;	/* Privileged */
-			uint32_t Reserved0:4;	/* Reserved, should be 0 */
-			uint32_t WaveId:4;	/* wave id */
-			uint32_t SIMD:2;	/* SIMD id */
-			uint32_t HSACU:4;	/* Compute unit */
-			uint32_t ShaderEngine:2;/* Shader engine */
-			uint32_t MessageType:2;	/* see HSA_DBG_WAVEMSG_TYPE */
-			uint32_t Reserved1:4;	/* Reserved, should be 0 */
-		} ui32;
-		uint32_t Value;
-	};
-	uint32_t Reserved2;
-};
-
-union HsaDbgWaveMessageAMD {
-	struct HsaDbgWaveMsgAMDGen2 WaveMsgInfoGen2;
-	/* for future HsaDbgWaveMsgAMDGen3; */
-};
-
-struct HsaDbgWaveMessage {
-	void *MemoryVA;		/* ptr to associated host-accessible data */
-	union HsaDbgWaveMessageAMD DbgWaveMsg;
-};
-
-/*
- * TODO: This definitions to be MOVED to kfd_event, once it is implemented.
- *
- * HSA sync primitive, Event and HW Exception notification API definitions.
- * The API functions allow the runtime to define a so-called sync-primitive,
- * a SW object combining a user-mode provided "syncvar" and a scheduler event
- * that can be signaled through a defined GPU interrupt. A syncvar is
- * a process virtual memory location of a certain size that can be accessed
- * by CPU and GPU shader code within the process to set and query the content
- * within that memory. The definition of the content is determined by the HSA
- * runtime and potentially GPU shader code interfacing with the HSA runtime.
- * The syncvar values may be commonly written through an PM4 WRITE_DATA packet
- * in the user mode instruction stream. The OS scheduler event is typically
- * associated and signaled by an interrupt issued by the GPU, but other HSA
- * system interrupt conditions from other HW (e.g. IOMMUv2) may be surfaced
- * by the KFD by this mechanism, too.
- */
-
-/* these are the new definitions for events */
-enum HSA_EVENTTYPE {
-	HSA_EVENTTYPE_SIGNAL = 0,	/* user-mode generated GPU signal */
-	HSA_EVENTTYPE_NODECHANGE = 1,	/* HSA node change (attach/detach) */
-	HSA_EVENTTYPE_DEVICESTATECHANGE = 2,	/* HSA device state change
-						 * (start/stop)
-						 */
-	HSA_EVENTTYPE_HW_EXCEPTION = 3,	/* GPU shader exception event */
-	HSA_EVENTTYPE_SYSTEM_EVENT = 4,	/* GPU SYSCALL with parameter info */
-	HSA_EVENTTYPE_DEBUG_EVENT = 5,	/* GPU signal for debugging */
-	HSA_EVENTTYPE_PROFILE_EVENT = 6,/* GPU signal for profiling */
-	HSA_EVENTTYPE_QUEUE_EVENT = 7,	/* GPU signal queue idle state
-					 * (EOP pm4)
-					 */
-	/* ...  */
-	HSA_EVENTTYPE_MAXID,
-	HSA_EVENTTYPE_TYPE_SIZE = 0xFFFFFFFF
-};
-
-/* Sub-definitions for various event types: Syncvar */
-struct HsaSyncVar {
-	union SyncVar {
-		void *UserData;	/* pointer to user mode data */
-		uint64_t UserDataPtrValue; /* 64bit compatibility of value */
-	} SyncVar;
-	uint64_t SyncVarSize;
-};
-
-/* Sub-definitions for various event types: NodeChange */
-
-enum HSA_EVENTTYPE_NODECHANGE_FLAGS {
-	HSA_EVENTTYPE_NODECHANGE_ADD = 0,
-	HSA_EVENTTYPE_NODECHANGE_REMOVE = 1,
-	HSA_EVENTTYPE_NODECHANGE_SIZE = 0xFFFFFFFF
-};
-
-struct HsaNodeChange {
-	/* HSA node added/removed on the platform */
-	enum HSA_EVENTTYPE_NODECHANGE_FLAGS Flags;
-};
-
-/* Sub-definitions for various event types: DeviceStateChange */
-enum HSA_EVENTTYPE_DEVICESTATECHANGE_FLAGS {
-	/* device started (and available) */
-	HSA_EVENTTYPE_DEVICESTATUSCHANGE_START = 0,
-	/* device stopped (i.e. unavailable) */
-	HSA_EVENTTYPE_DEVICESTATUSCHANGE_STOP = 1,
-	HSA_EVENTTYPE_DEVICESTATUSCHANGE_SIZE = 0xFFFFFFFF
-};
-
-enum HSA_DEVICE {
-	HSA_DEVICE_CPU = 0,
-	HSA_DEVICE_GPU = 1,
-	MAX_HSA_DEVICE = 2
-};
-
-struct HsaDeviceStateChange {
-	uint32_t NodeId;	/* F-NUMA node that contains the device */
-	enum HSA_DEVICE Device;	/* device type: GPU or CPU */
-	enum HSA_EVENTTYPE_DEVICESTATECHANGE_FLAGS Flags; /* event flags */
-};
-
-struct HsaEventData {
-	enum HSA_EVENTTYPE EventType; /* event type */
-	union EventData {
-		/*
-		 * return data associated with HSA_EVENTTYPE_SIGNAL
-		 * and other events
-		 */
-		struct HsaSyncVar SyncVar;
-
-		/* data associated with HSA_EVENTTYPE_NODE_CHANGE */
-		struct HsaNodeChange NodeChangeState;
-
-		/* data associated with HSA_EVENTTYPE_DEVICE_STATE_CHANGE */
-		struct HsaDeviceStateChange DeviceState;
-	} EventData;
-
-	/* the following data entries are internal to the KFD & thunk itself */
-
-	/* internal thunk store for Event data (OsEventHandle) */
-	uint64_t HWData1;
-	/* internal thunk store for Event data (HWAddress) */
-	uint64_t HWData2;
-	/* internal thunk store for Event data (HWData) */
-	uint32_t HWData3;
-};
-
-struct HsaEventDescriptor {
-	/* event type to allocate */
-	enum HSA_EVENTTYPE EventType;
-	/* H-NUMA node containing GPU device that is event source */
-	uint32_t NodeId;
-	/* pointer to user mode syncvar data, syncvar->UserDataPtrValue
-	 * may be NULL
-	 */
-	struct HsaSyncVar SyncVar;
-};
-
-struct HsaEvent {
-	uint32_t EventId;
-	struct HsaEventData EventData;
-};
-
-#pragma pack(pop)
-
-enum DBGDEV_TYPE {
-	DBGDEV_TYPE_ILLEGAL = 0,
-	DBGDEV_TYPE_NODIQ = 1,
-	DBGDEV_TYPE_DIQ = 2,
-	DBGDEV_TYPE_TEST = 3
-};
-
-struct dbg_address_watch_info {
-	struct kfd_process *process;
-	enum HSA_DBG_WATCH_MODE *watch_mode;
-	uint64_t *watch_address;
-	uint64_t *watch_mask;
-	struct HsaEvent *watch_event;
-	uint32_t num_watch_points;
-};
-
-struct dbg_wave_control_info {
-	struct kfd_process *process;
-	uint32_t trapId;
-	enum HSA_DBG_WAVEOP operand;
-	enum HSA_DBG_WAVEMODE mode;
-	struct HsaDbgWaveMessage dbgWave_msg;
-};
-
-struct kfd_dbgdev {
-
-	/* The device that owns this data. */
-	struct kfd_dev *dev;
-
-	/* kernel queue for DIQ */
-	struct kernel_queue *kq;
-
-	/* a pointer to the pqm of the calling process */
-	struct process_queue_manager *pqm;
-
-	/* type of debug device ( DIQ, non DIQ, etc. ) */
-	enum DBGDEV_TYPE type;
-
-	/* virtualized function pointers to device dbg */
-	int (*dbgdev_register)(struct kfd_dbgdev *dbgdev);
-	int (*dbgdev_unregister)(struct kfd_dbgdev *dbgdev);
-	int (*dbgdev_address_watch)(struct kfd_dbgdev *dbgdev,
-				struct dbg_address_watch_info *adw_info);
-	int (*dbgdev_wave_control)(struct kfd_dbgdev *dbgdev,
-				struct dbg_wave_control_info *wac_info);
-
-};
-
-struct kfd_dbgmgr {
-	unsigned int pasid;
-	struct kfd_dev *dev;
-	struct kfd_dbgdev *dbgdev;
-};
-
-/* prototypes for debug manager functions */
-struct mutex *kfd_get_dbgmgr_mutex(void);
-void kfd_dbgmgr_destroy(struct kfd_dbgmgr *pmgr);
-bool kfd_dbgmgr_create(struct kfd_dbgmgr **ppmgr, struct kfd_dev *pdev);
-long kfd_dbgmgr_register(struct kfd_dbgmgr *pmgr, struct kfd_process *p);
-long kfd_dbgmgr_unregister(struct kfd_dbgmgr *pmgr, struct kfd_process *p);
-long kfd_dbgmgr_wave_control(struct kfd_dbgmgr *pmgr,
-				struct dbg_wave_control_info *wac_info);
-long kfd_dbgmgr_address_watch(struct kfd_dbgmgr *pmgr,
-			struct dbg_address_watch_info *adw_info);
-#endif /* KFD_DBGMGR_H_ */
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_debug.c b/drivers/gpu/drm/amd/amdkfd/kfd_debug.c
new file mode 100644
index 000000000000..ba99e0f258ae
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_debug.c
@@ -0,0 +1,1152 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include "kfd_debug.h"
+#include "kfd_device_queue_manager.h"
+#include "kfd_topology.h"
+#include <linux/file.h>
+#include <uapi/linux/kfd_ioctl.h>
+#include <uapi/linux/kfd_sysfs.h>
+
+#define MAX_WATCH_ADDRESSES	4
+
+int kfd_dbg_ev_query_debug_event(struct kfd_process *process,
+		      unsigned int *queue_id,
+		      unsigned int *gpu_id,
+		      uint64_t exception_clear_mask,
+		      uint64_t *event_status)
+{
+	struct process_queue_manager *pqm;
+	struct process_queue_node *pqn;
+	int i;
+
+	if (!(process && process->debug_trap_enabled))
+		return -ENODATA;
+
+	mutex_lock(&process->event_mutex);
+	*event_status = 0;
+	*queue_id = 0;
+	*gpu_id = 0;
+
+	/* find and report queue events */
+	pqm = &process->pqm;
+	list_for_each_entry(pqn, &pqm->queues, process_queue_list) {
+		uint64_t tmp = process->exception_enable_mask;
+
+		if (!pqn->q)
+			continue;
+
+		tmp &= pqn->q->properties.exception_status;
+
+		if (!tmp)
+			continue;
+
+		*event_status = pqn->q->properties.exception_status;
+		*queue_id = pqn->q->properties.queue_id;
+		*gpu_id = pqn->q->device->id;
+		pqn->q->properties.exception_status &= ~exception_clear_mask;
+		goto out;
+	}
+
+	/* find and report device events */
+	for (i = 0; i < process->n_pdds; i++) {
+		struct kfd_process_device *pdd = process->pdds[i];
+		uint64_t tmp = process->exception_enable_mask
+						& pdd->exception_status;
+
+		if (!tmp)
+			continue;
+
+		*event_status = pdd->exception_status;
+		*gpu_id = pdd->dev->id;
+		pdd->exception_status &= ~exception_clear_mask;
+		goto out;
+	}
+
+	/* report process events */
+	if (process->exception_enable_mask & process->exception_status) {
+		*event_status = process->exception_status;
+		process->exception_status &= ~exception_clear_mask;
+	}
+
+out:
+	mutex_unlock(&process->event_mutex);
+	return *event_status ? 0 : -EAGAIN;
+}
+
+void debug_event_write_work_handler(struct work_struct *work)
+{
+	struct kfd_process *process;
+
+	static const char write_data = '.';
+	loff_t pos = 0;
+
+	process = container_of(work,
+			struct kfd_process,
+			debug_event_workarea);
+
+	if (process->debug_trap_enabled && process->dbg_ev_file)
+		kernel_write(process->dbg_ev_file, &write_data, 1, &pos);
+}
+
+/* update process/device/queue exception status, write to descriptor
+ * only if exception_status is enabled.
+ */
+bool kfd_dbg_ev_raise(uint64_t event_mask,
+			struct kfd_process *process, struct kfd_node *dev,
+			unsigned int source_id, bool use_worker,
+			void *exception_data, size_t exception_data_size)
+{
+	struct process_queue_manager *pqm;
+	struct process_queue_node *pqn;
+	int i;
+	static const char write_data = '.';
+	loff_t pos = 0;
+	bool is_subscribed = true;
+
+	if (!(process && process->debug_trap_enabled))
+		return false;
+
+	mutex_lock(&process->event_mutex);
+
+	if (event_mask & KFD_EC_MASK_DEVICE) {
+		for (i = 0; i < process->n_pdds; i++) {
+			struct kfd_process_device *pdd = process->pdds[i];
+
+			if (pdd->dev != dev)
+				continue;
+
+			pdd->exception_status |= event_mask & KFD_EC_MASK_DEVICE;
+
+			if (event_mask & KFD_EC_MASK(EC_DEVICE_MEMORY_VIOLATION)) {
+				if (!pdd->vm_fault_exc_data) {
+					pdd->vm_fault_exc_data = kmemdup(
+							exception_data,
+							exception_data_size,
+							GFP_KERNEL);
+					if (!pdd->vm_fault_exc_data)
+						pr_debug("Failed to allocate exception data memory");
+				} else {
+					pr_debug("Debugger exception data not saved\n");
+					print_hex_dump_bytes("exception data: ",
+							DUMP_PREFIX_OFFSET,
+							exception_data,
+							exception_data_size);
+				}
+			}
+			break;
+		}
+	} else if (event_mask & KFD_EC_MASK_PROCESS) {
+		process->exception_status |= event_mask & KFD_EC_MASK_PROCESS;
+	} else {
+		pqm = &process->pqm;
+		list_for_each_entry(pqn, &pqm->queues,
+				process_queue_list) {
+			int target_id;
+
+			if (!pqn->q)
+				continue;
+
+			target_id = event_mask & KFD_EC_MASK(EC_QUEUE_NEW) ?
+					pqn->q->properties.queue_id :
+							pqn->q->doorbell_id;
+
+			if (pqn->q->device != dev || target_id != source_id)
+				continue;
+
+			pqn->q->properties.exception_status |= event_mask;
+			break;
+		}
+	}
+
+	if (process->exception_enable_mask & event_mask) {
+		if (use_worker)
+			schedule_work(&process->debug_event_workarea);
+		else
+			kernel_write(process->dbg_ev_file,
+					&write_data,
+					1,
+					&pos);
+	} else {
+		is_subscribed = false;
+	}
+
+	mutex_unlock(&process->event_mutex);
+
+	return is_subscribed;
+}
+
+/* set pending event queue entry from ring entry  */
+bool kfd_set_dbg_ev_from_interrupt(struct kfd_node *dev,
+				   unsigned int pasid,
+				   uint32_t doorbell_id,
+				   uint64_t trap_mask,
+				   void *exception_data,
+				   size_t exception_data_size)
+{
+	struct kfd_process *p;
+	struct kfd_process_device *pdd = NULL;
+	bool signaled_to_debugger_or_runtime = false;
+
+	p = kfd_lookup_process_by_pasid(pasid, &pdd);
+
+	if (!pdd)
+		return false;
+
+	if (!kfd_dbg_ev_raise(trap_mask, p, dev, doorbell_id, true,
+			      exception_data, exception_data_size)) {
+		struct process_queue_manager *pqm;
+		struct process_queue_node *pqn;
+
+		if (!!(trap_mask & KFD_EC_MASK_QUEUE) &&
+		       p->runtime_info.runtime_state == DEBUG_RUNTIME_STATE_ENABLED) {
+			mutex_lock(&p->mutex);
+
+			pqm = &p->pqm;
+			list_for_each_entry(pqn, &pqm->queues,
+							process_queue_list) {
+
+				if (!(pqn->q && pqn->q->device == dev &&
+				      pqn->q->doorbell_id == doorbell_id))
+					continue;
+
+				kfd_send_exception_to_runtime(p, pqn->q->properties.queue_id,
+							      trap_mask);
+
+				signaled_to_debugger_or_runtime = true;
+
+				break;
+			}
+
+			mutex_unlock(&p->mutex);
+		} else if (trap_mask & KFD_EC_MASK(EC_DEVICE_MEMORY_VIOLATION)) {
+			kfd_evict_process_device(pdd);
+			kfd_signal_vm_fault_event(pdd, NULL, exception_data);
+
+			signaled_to_debugger_or_runtime = true;
+		}
+	} else {
+		signaled_to_debugger_or_runtime = true;
+	}
+
+	kfd_unref_process(p);
+
+	return signaled_to_debugger_or_runtime;
+}
+
+int kfd_dbg_send_exception_to_runtime(struct kfd_process *p,
+					unsigned int dev_id,
+					unsigned int queue_id,
+					uint64_t error_reason)
+{
+	if (error_reason & KFD_EC_MASK(EC_DEVICE_MEMORY_VIOLATION)) {
+		struct kfd_process_device *pdd = NULL;
+		struct kfd_hsa_memory_exception_data *data;
+		int i;
+
+		for (i = 0; i < p->n_pdds; i++) {
+			if (p->pdds[i]->dev->id == dev_id) {
+				pdd = p->pdds[i];
+				break;
+			}
+		}
+
+		if (!pdd)
+			return -ENODEV;
+
+		data = (struct kfd_hsa_memory_exception_data *)
+						pdd->vm_fault_exc_data;
+
+		kfd_evict_process_device(pdd);
+		kfd_signal_vm_fault_event(pdd, NULL, data);
+		error_reason &= ~KFD_EC_MASK(EC_DEVICE_MEMORY_VIOLATION);
+	}
+
+	if (error_reason & (KFD_EC_MASK(EC_PROCESS_RUNTIME))) {
+		/*
+		 * block should only happen after the debugger receives runtime
+		 * enable notice.
+		 */
+		up(&p->runtime_enable_sema);
+		error_reason &= ~KFD_EC_MASK(EC_PROCESS_RUNTIME);
+	}
+
+	if (error_reason)
+		return kfd_send_exception_to_runtime(p, queue_id, error_reason);
+
+	return 0;
+}
+
+static int kfd_dbg_set_queue_workaround(struct queue *q, bool enable)
+{
+	struct mqd_update_info minfo = {0};
+	int err;
+
+	if (!q)
+		return 0;
+
+	if (!kfd_dbg_has_cwsr_workaround(q->device))
+		return 0;
+
+	if (enable && q->properties.is_user_cu_masked)
+		return -EBUSY;
+
+	minfo.update_flag = enable ? UPDATE_FLAG_DBG_WA_ENABLE : UPDATE_FLAG_DBG_WA_DISABLE;
+
+	q->properties.is_dbg_wa = enable;
+	err = q->device->dqm->ops.update_queue(q->device->dqm, q, &minfo);
+	if (err)
+		q->properties.is_dbg_wa = false;
+
+	return err;
+}
+
+static int kfd_dbg_set_workaround(struct kfd_process *target, bool enable)
+{
+	struct process_queue_manager *pqm = &target->pqm;
+	struct process_queue_node *pqn;
+	int r = 0;
+
+	list_for_each_entry(pqn, &pqm->queues, process_queue_list) {
+		r = kfd_dbg_set_queue_workaround(pqn->q, enable);
+		if (enable && r)
+			goto unwind;
+	}
+
+	return 0;
+
+unwind:
+	list_for_each_entry(pqn, &pqm->queues, process_queue_list)
+		kfd_dbg_set_queue_workaround(pqn->q, false);
+
+	if (enable)
+		target->runtime_info.runtime_state = r == -EBUSY ?
+				DEBUG_RUNTIME_STATE_ENABLED_BUSY :
+				DEBUG_RUNTIME_STATE_ENABLED_ERROR;
+
+	return r;
+}
+
+int kfd_dbg_set_mes_debug_mode(struct kfd_process_device *pdd, bool sq_trap_en)
+{
+	uint32_t spi_dbg_cntl = pdd->spi_dbg_override | pdd->spi_dbg_launch_mode;
+	uint32_t flags = pdd->process->dbg_flags;
+	struct amdgpu_device *adev = pdd->dev->adev;
+	int r;
+
+	if (!kfd_dbg_is_per_vmid_supported(pdd->dev))
+		return 0;
+
+	if (!pdd->proc_ctx_cpu_ptr) {
+		r = amdgpu_amdkfd_alloc_gtt_mem(adev,
+			AMDGPU_MES_PROC_CTX_SIZE,
+			&pdd->proc_ctx_bo,
+			&pdd->proc_ctx_gpu_addr,
+			&pdd->proc_ctx_cpu_ptr,
+			false);
+		if (r) {
+			dev_err(adev->dev,
+			"failed to allocate process context bo\n");
+			return r;
+		}
+		memset(pdd->proc_ctx_cpu_ptr, 0, AMDGPU_MES_PROC_CTX_SIZE);
+	}
+
+	return amdgpu_mes_set_shader_debugger(pdd->dev->adev, pdd->proc_ctx_gpu_addr, spi_dbg_cntl,
+						pdd->watch_points, flags, sq_trap_en);
+}
+
+#define KFD_DEBUGGER_INVALID_WATCH_POINT_ID -1
+static int kfd_dbg_get_dev_watch_id(struct kfd_process_device *pdd, int *watch_id)
+{
+	int i;
+
+	*watch_id = KFD_DEBUGGER_INVALID_WATCH_POINT_ID;
+
+	spin_lock(&pdd->dev->watch_points_lock);
+
+	for (i = 0; i < MAX_WATCH_ADDRESSES; i++) {
+		/* device watchpoint in use so skip */
+		if ((pdd->dev->alloc_watch_ids >> i) & 0x1)
+			continue;
+
+		pdd->alloc_watch_ids |= 0x1 << i;
+		pdd->dev->alloc_watch_ids |= 0x1 << i;
+		*watch_id = i;
+		spin_unlock(&pdd->dev->watch_points_lock);
+		return 0;
+	}
+
+	spin_unlock(&pdd->dev->watch_points_lock);
+
+	return -ENOMEM;
+}
+
+static void kfd_dbg_clear_dev_watch_id(struct kfd_process_device *pdd, int watch_id)
+{
+	spin_lock(&pdd->dev->watch_points_lock);
+
+	/* process owns device watch point so safe to clear */
+	if ((pdd->alloc_watch_ids >> watch_id) & 0x1) {
+		pdd->alloc_watch_ids &= ~(0x1 << watch_id);
+		pdd->dev->alloc_watch_ids &= ~(0x1 << watch_id);
+	}
+
+	spin_unlock(&pdd->dev->watch_points_lock);
+}
+
+static bool kfd_dbg_owns_dev_watch_id(struct kfd_process_device *pdd, int watch_id)
+{
+	bool owns_watch_id = false;
+
+	spin_lock(&pdd->dev->watch_points_lock);
+	owns_watch_id = watch_id < MAX_WATCH_ADDRESSES &&
+			((pdd->alloc_watch_ids >> watch_id) & 0x1);
+
+	spin_unlock(&pdd->dev->watch_points_lock);
+
+	return owns_watch_id;
+}
+
+int kfd_dbg_trap_clear_dev_address_watch(struct kfd_process_device *pdd,
+					uint32_t watch_id)
+{
+	int r;
+
+	if (!kfd_dbg_owns_dev_watch_id(pdd, watch_id))
+		return -EINVAL;
+
+	if (!pdd->dev->kfd->shared_resources.enable_mes) {
+		r = debug_lock_and_unmap(pdd->dev->dqm);
+		if (r)
+			return r;
+	}
+
+	amdgpu_gfx_off_ctrl(pdd->dev->adev, false);
+	pdd->watch_points[watch_id] = pdd->dev->kfd2kgd->clear_address_watch(
+							pdd->dev->adev,
+							watch_id);
+	amdgpu_gfx_off_ctrl(pdd->dev->adev, true);
+
+	if (!pdd->dev->kfd->shared_resources.enable_mes)
+		r = debug_map_and_unlock(pdd->dev->dqm);
+	else
+		r = kfd_dbg_set_mes_debug_mode(pdd, true);
+
+	kfd_dbg_clear_dev_watch_id(pdd, watch_id);
+
+	return r;
+}
+
+int kfd_dbg_trap_set_dev_address_watch(struct kfd_process_device *pdd,
+					uint64_t watch_address,
+					uint32_t watch_address_mask,
+					uint32_t *watch_id,
+					uint32_t watch_mode)
+{
+	int xcc_id, r = kfd_dbg_get_dev_watch_id(pdd, watch_id);
+	uint32_t xcc_mask = pdd->dev->xcc_mask;
+
+	if (r)
+		return r;
+
+	if (!pdd->dev->kfd->shared_resources.enable_mes) {
+		r = debug_lock_and_unmap(pdd->dev->dqm);
+		if (r) {
+			kfd_dbg_clear_dev_watch_id(pdd, *watch_id);
+			return r;
+		}
+	}
+
+	amdgpu_gfx_off_ctrl(pdd->dev->adev, false);
+	for_each_inst(xcc_id, xcc_mask)
+		pdd->watch_points[*watch_id] = pdd->dev->kfd2kgd->set_address_watch(
+				pdd->dev->adev,
+				watch_address,
+				watch_address_mask,
+				*watch_id,
+				watch_mode,
+				pdd->dev->vm_info.last_vmid_kfd,
+				xcc_id);
+	amdgpu_gfx_off_ctrl(pdd->dev->adev, true);
+
+	if (!pdd->dev->kfd->shared_resources.enable_mes)
+		r = debug_map_and_unlock(pdd->dev->dqm);
+	else
+		r = kfd_dbg_set_mes_debug_mode(pdd, true);
+
+	/* HWS is broken so no point in HW rollback but release the watchpoint anyways */
+	if (r)
+		kfd_dbg_clear_dev_watch_id(pdd, *watch_id);
+
+	return 0;
+}
+
+static void kfd_dbg_clear_process_address_watch(struct kfd_process *target)
+{
+	int i, j;
+
+	for (i = 0; i < target->n_pdds; i++)
+		for (j = 0; j < MAX_WATCH_ADDRESSES; j++)
+			kfd_dbg_trap_clear_dev_address_watch(target->pdds[i], j);
+}
+
+int kfd_dbg_trap_set_flags(struct kfd_process *target, uint32_t *flags)
+{
+	uint32_t prev_flags = target->dbg_flags;
+	int i, r = 0, rewind_count = 0;
+
+	for (i = 0; i < target->n_pdds; i++) {
+		struct kfd_topology_device *topo_dev =
+				kfd_topology_device_by_id(target->pdds[i]->dev->id);
+		uint32_t caps = topo_dev->node_props.capability;
+
+		if (!(caps & HSA_CAP_TRAP_DEBUG_PRECISE_MEMORY_OPERATIONS_SUPPORTED) &&
+			(*flags & KFD_DBG_TRAP_FLAG_SINGLE_MEM_OP)) {
+			*flags = prev_flags;
+			return -EACCES;
+		}
+
+		if (!(caps & HSA_CAP_TRAP_DEBUG_PRECISE_ALU_OPERATIONS_SUPPORTED) &&
+		    (*flags & KFD_DBG_TRAP_FLAG_SINGLE_ALU_OP)) {
+			*flags = prev_flags;
+			return -EACCES;
+		}
+	}
+
+	target->dbg_flags = *flags;
+	*flags = prev_flags;
+	for (i = 0; i < target->n_pdds; i++) {
+		struct kfd_process_device *pdd = target->pdds[i];
+
+		if (!kfd_dbg_is_per_vmid_supported(pdd->dev))
+			continue;
+
+		if (!pdd->dev->kfd->shared_resources.enable_mes)
+			r = debug_refresh_runlist(pdd->dev->dqm);
+		else
+			r = kfd_dbg_set_mes_debug_mode(pdd, true);
+
+		if (r) {
+			target->dbg_flags = prev_flags;
+			break;
+		}
+
+		rewind_count++;
+	}
+
+	/* Rewind flags */
+	if (r) {
+		target->dbg_flags = prev_flags;
+
+		for (i = 0; i < rewind_count; i++) {
+			struct kfd_process_device *pdd = target->pdds[i];
+
+			if (!kfd_dbg_is_per_vmid_supported(pdd->dev))
+				continue;
+
+			if (!pdd->dev->kfd->shared_resources.enable_mes)
+				debug_refresh_runlist(pdd->dev->dqm);
+			else
+				kfd_dbg_set_mes_debug_mode(pdd, true);
+		}
+	}
+
+	return r;
+}
+
+/* kfd_dbg_trap_deactivate:
+ *	target: target process
+ *	unwind: If this is unwinding a failed kfd_dbg_trap_enable()
+ *	unwind_count:
+ *		If unwind == true, how far down the pdd list we need
+ *				to unwind
+ *		else: ignored
+ */
+void kfd_dbg_trap_deactivate(struct kfd_process *target, bool unwind, int unwind_count)
+{
+	int i;
+
+	if (!unwind) {
+		uint32_t flags = 0;
+		int resume_count = resume_queues(target, 0, NULL);
+
+		if (resume_count)
+			pr_debug("Resumed %d queues\n", resume_count);
+
+		cancel_work_sync(&target->debug_event_workarea);
+		kfd_dbg_clear_process_address_watch(target);
+		kfd_dbg_trap_set_wave_launch_mode(target, 0);
+
+		kfd_dbg_trap_set_flags(target, &flags);
+	}
+
+	for (i = 0; i < target->n_pdds; i++) {
+		struct kfd_process_device *pdd = target->pdds[i];
+
+		/* If this is an unwind, and we have unwound the required
+		 * enable calls on the pdd list, we need to stop now
+		 * otherwise we may mess up another debugger session.
+		 */
+		if (unwind && i == unwind_count)
+			break;
+
+		kfd_process_set_trap_debug_flag(&pdd->qpd, false);
+
+		/* GFX off is already disabled by debug activate if not RLC restore supported. */
+		if (kfd_dbg_is_rlc_restore_supported(pdd->dev))
+			amdgpu_gfx_off_ctrl(pdd->dev->adev, false);
+		pdd->spi_dbg_override =
+				pdd->dev->kfd2kgd->disable_debug_trap(
+				pdd->dev->adev,
+				target->runtime_info.ttmp_setup,
+				pdd->dev->vm_info.last_vmid_kfd);
+		amdgpu_gfx_off_ctrl(pdd->dev->adev, true);
+
+		if (!kfd_dbg_is_per_vmid_supported(pdd->dev) &&
+				release_debug_trap_vmid(pdd->dev->dqm, &pdd->qpd))
+			pr_err("Failed to release debug vmid on [%i]\n", pdd->dev->id);
+
+		if (!pdd->dev->kfd->shared_resources.enable_mes)
+			debug_refresh_runlist(pdd->dev->dqm);
+		else
+			kfd_dbg_set_mes_debug_mode(pdd, !kfd_dbg_has_cwsr_workaround(pdd->dev));
+	}
+
+	kfd_dbg_set_workaround(target, false);
+}
+
+static void kfd_dbg_clean_exception_status(struct kfd_process *target)
+{
+	struct process_queue_manager *pqm;
+	struct process_queue_node *pqn;
+	int i;
+
+	for (i = 0; i < target->n_pdds; i++) {
+		struct kfd_process_device *pdd = target->pdds[i];
+
+		kfd_process_drain_interrupts(pdd);
+
+		pdd->exception_status = 0;
+	}
+
+	pqm = &target->pqm;
+	list_for_each_entry(pqn, &pqm->queues, process_queue_list) {
+		if (!pqn->q)
+			continue;
+
+		pqn->q->properties.exception_status = 0;
+	}
+
+	target->exception_status = 0;
+}
+
+int kfd_dbg_trap_disable(struct kfd_process *target)
+{
+	if (!target->debug_trap_enabled)
+		return 0;
+
+	/*
+	 * Defer deactivation to runtime if runtime not enabled otherwise reset
+	 * attached running target runtime state to enable for re-attach.
+	 */
+	if (target->runtime_info.runtime_state == DEBUG_RUNTIME_STATE_ENABLED)
+		kfd_dbg_trap_deactivate(target, false, 0);
+	else if (target->runtime_info.runtime_state != DEBUG_RUNTIME_STATE_DISABLED)
+		target->runtime_info.runtime_state = DEBUG_RUNTIME_STATE_ENABLED;
+
+	cancel_work_sync(&target->debug_event_workarea);
+	fput(target->dbg_ev_file);
+	target->dbg_ev_file = NULL;
+
+	if (target->debugger_process) {
+		atomic_dec(&target->debugger_process->debugged_process_count);
+		target->debugger_process = NULL;
+	}
+
+	target->debug_trap_enabled = false;
+	kfd_dbg_clean_exception_status(target);
+	kfd_unref_process(target);
+
+	return 0;
+}
+
+int kfd_dbg_trap_activate(struct kfd_process *target)
+{
+	int i, r = 0;
+
+	r = kfd_dbg_set_workaround(target, true);
+	if (r)
+		return r;
+
+	for (i = 0; i < target->n_pdds; i++) {
+		struct kfd_process_device *pdd = target->pdds[i];
+
+		if (!kfd_dbg_is_per_vmid_supported(pdd->dev)) {
+			r = reserve_debug_trap_vmid(pdd->dev->dqm, &pdd->qpd);
+
+			if (r) {
+				target->runtime_info.runtime_state = (r == -EBUSY) ?
+							DEBUG_RUNTIME_STATE_ENABLED_BUSY :
+							DEBUG_RUNTIME_STATE_ENABLED_ERROR;
+
+				goto unwind_err;
+			}
+		}
+
+		/* Disable GFX OFF to prevent garbage read/writes to debug registers.
+		 * If RLC restore of debug registers is not supported and runtime enable
+		 * hasn't done so already on ttmp setup request, restore the trap config registers.
+		 *
+		 * If RLC restore of debug registers is not supported, keep gfx off disabled for
+		 * the debug session.
+		 */
+		amdgpu_gfx_off_ctrl(pdd->dev->adev, false);
+		if (!(kfd_dbg_is_rlc_restore_supported(pdd->dev) ||
+						target->runtime_info.ttmp_setup))
+			pdd->dev->kfd2kgd->enable_debug_trap(pdd->dev->adev, true,
+								pdd->dev->vm_info.last_vmid_kfd);
+
+		pdd->spi_dbg_override = pdd->dev->kfd2kgd->enable_debug_trap(
+					pdd->dev->adev,
+					false,
+					pdd->dev->vm_info.last_vmid_kfd);
+
+		if (kfd_dbg_is_rlc_restore_supported(pdd->dev))
+			amdgpu_gfx_off_ctrl(pdd->dev->adev, true);
+
+		/*
+		 * Setting the debug flag in the trap handler requires that the TMA has been
+		 * allocated, which occurs during CWSR initialization.
+		 * In the event that CWSR has not been initialized at this point, setting the
+		 * flag will be called again during CWSR initialization if the target process
+		 * is still debug enabled.
+		 */
+		kfd_process_set_trap_debug_flag(&pdd->qpd, true);
+
+		if (!pdd->dev->kfd->shared_resources.enable_mes)
+			r = debug_refresh_runlist(pdd->dev->dqm);
+		else
+			r = kfd_dbg_set_mes_debug_mode(pdd, true);
+
+		if (r) {
+			target->runtime_info.runtime_state =
+					DEBUG_RUNTIME_STATE_ENABLED_ERROR;
+			goto unwind_err;
+		}
+	}
+
+	return 0;
+
+unwind_err:
+	/* Enabling debug failed, we need to disable on
+	 * all GPUs so the enable is all or nothing.
+	 */
+	kfd_dbg_trap_deactivate(target, true, i);
+	return r;
+}
+
+int kfd_dbg_trap_enable(struct kfd_process *target, uint32_t fd,
+			void __user *runtime_info, uint32_t *runtime_size)
+{
+	struct file *f;
+	uint32_t copy_size;
+	int i, r = 0;
+
+	if (target->debug_trap_enabled)
+		return -EALREADY;
+
+	/* Enable pre-checks */
+	for (i = 0; i < target->n_pdds; i++) {
+		struct kfd_process_device *pdd = target->pdds[i];
+
+		if (!KFD_IS_SOC15(pdd->dev))
+			return -ENODEV;
+
+		if (pdd->qpd.num_gws && (!kfd_dbg_has_gws_support(pdd->dev) ||
+					 kfd_dbg_has_cwsr_workaround(pdd->dev)))
+			return -EBUSY;
+	}
+
+	copy_size = min((size_t)(*runtime_size), sizeof(target->runtime_info));
+
+	f = fget(fd);
+	if (!f) {
+		pr_err("Failed to get file for (%i)\n", fd);
+		return -EBADF;
+	}
+
+	target->dbg_ev_file = f;
+
+	/* defer activation to runtime if not runtime enabled */
+	if (target->runtime_info.runtime_state == DEBUG_RUNTIME_STATE_ENABLED)
+		kfd_dbg_trap_activate(target);
+
+	/* We already hold the process reference but hold another one for the
+	 * debug session.
+	 */
+	kref_get(&target->ref);
+	target->debug_trap_enabled = true;
+
+	if (target->debugger_process)
+		atomic_inc(&target->debugger_process->debugged_process_count);
+
+	if (copy_to_user(runtime_info, (void *)&target->runtime_info, copy_size)) {
+		kfd_dbg_trap_deactivate(target, false, 0);
+		r = -EFAULT;
+	}
+
+	*runtime_size = sizeof(target->runtime_info);
+
+	return r;
+}
+
+static int kfd_dbg_validate_trap_override_request(struct kfd_process *p,
+						uint32_t trap_override,
+						uint32_t trap_mask_request,
+						uint32_t *trap_mask_supported)
+{
+	int i = 0;
+
+	*trap_mask_supported = 0xffffffff;
+
+	for (i = 0; i < p->n_pdds; i++) {
+		struct kfd_process_device *pdd = p->pdds[i];
+		int err = pdd->dev->kfd2kgd->validate_trap_override_request(
+								pdd->dev->adev,
+								trap_override,
+								trap_mask_supported);
+
+		if (err)
+			return err;
+	}
+
+	if (trap_mask_request & ~*trap_mask_supported)
+		return -EACCES;
+
+	return 0;
+}
+
+int kfd_dbg_trap_set_wave_launch_override(struct kfd_process *target,
+					uint32_t trap_override,
+					uint32_t trap_mask_bits,
+					uint32_t trap_mask_request,
+					uint32_t *trap_mask_prev,
+					uint32_t *trap_mask_supported)
+{
+	int r = 0, i;
+
+	r = kfd_dbg_validate_trap_override_request(target,
+						trap_override,
+						trap_mask_request,
+						trap_mask_supported);
+
+	if (r)
+		return r;
+
+	for (i = 0; i < target->n_pdds; i++) {
+		struct kfd_process_device *pdd = target->pdds[i];
+
+		amdgpu_gfx_off_ctrl(pdd->dev->adev, false);
+		pdd->spi_dbg_override = pdd->dev->kfd2kgd->set_wave_launch_trap_override(
+				pdd->dev->adev,
+				pdd->dev->vm_info.last_vmid_kfd,
+				trap_override,
+				trap_mask_bits,
+				trap_mask_request,
+				trap_mask_prev,
+				pdd->spi_dbg_override);
+		amdgpu_gfx_off_ctrl(pdd->dev->adev, true);
+
+		if (!pdd->dev->kfd->shared_resources.enable_mes)
+			r = debug_refresh_runlist(pdd->dev->dqm);
+		else
+			r = kfd_dbg_set_mes_debug_mode(pdd, true);
+
+		if (r)
+			break;
+	}
+
+	return r;
+}
+
+int kfd_dbg_trap_set_wave_launch_mode(struct kfd_process *target,
+					uint8_t wave_launch_mode)
+{
+	int r = 0, i;
+
+	if (wave_launch_mode != KFD_DBG_TRAP_WAVE_LAUNCH_MODE_NORMAL &&
+			wave_launch_mode != KFD_DBG_TRAP_WAVE_LAUNCH_MODE_HALT &&
+			wave_launch_mode != KFD_DBG_TRAP_WAVE_LAUNCH_MODE_DEBUG)
+		return -EINVAL;
+
+	for (i = 0; i < target->n_pdds; i++) {
+		struct kfd_process_device *pdd = target->pdds[i];
+
+		amdgpu_gfx_off_ctrl(pdd->dev->adev, false);
+		pdd->spi_dbg_launch_mode = pdd->dev->kfd2kgd->set_wave_launch_mode(
+				pdd->dev->adev,
+				wave_launch_mode,
+				pdd->dev->vm_info.last_vmid_kfd);
+		amdgpu_gfx_off_ctrl(pdd->dev->adev, true);
+
+		if (!pdd->dev->kfd->shared_resources.enable_mes)
+			r = debug_refresh_runlist(pdd->dev->dqm);
+		else
+			r = kfd_dbg_set_mes_debug_mode(pdd, true);
+
+		if (r)
+			break;
+	}
+
+	return r;
+}
+
+int kfd_dbg_trap_query_exception_info(struct kfd_process *target,
+		uint32_t source_id,
+		uint32_t exception_code,
+		bool clear_exception,
+		void __user *info,
+		uint32_t *info_size)
+{
+	bool found = false;
+	int r = 0;
+	uint32_t copy_size, actual_info_size = 0;
+	uint64_t *exception_status_ptr = NULL;
+
+	if (!target)
+		return -EINVAL;
+
+	if (!info || !info_size)
+		return -EINVAL;
+
+	mutex_lock(&target->event_mutex);
+
+	if (KFD_DBG_EC_TYPE_IS_QUEUE(exception_code)) {
+		/* Per queue exceptions */
+		struct queue *queue = NULL;
+		int i;
+
+		for (i = 0; i < target->n_pdds; i++) {
+			struct kfd_process_device *pdd = target->pdds[i];
+			struct qcm_process_device *qpd = &pdd->qpd;
+
+			list_for_each_entry(queue, &qpd->queues_list, list) {
+				if (!found && queue->properties.queue_id == source_id) {
+					found = true;
+					break;
+				}
+			}
+			if (found)
+				break;
+		}
+
+		if (!found) {
+			r = -EINVAL;
+			goto out;
+		}
+
+		if (!(queue->properties.exception_status & KFD_EC_MASK(exception_code))) {
+			r = -ENODATA;
+			goto out;
+		}
+		exception_status_ptr = &queue->properties.exception_status;
+	} else if (KFD_DBG_EC_TYPE_IS_DEVICE(exception_code)) {
+		/* Per device exceptions */
+		struct kfd_process_device *pdd = NULL;
+		int i;
+
+		for (i = 0; i < target->n_pdds; i++) {
+			pdd = target->pdds[i];
+			if (pdd->dev->id == source_id) {
+				found = true;
+				break;
+			}
+		}
+
+		if (!found) {
+			r = -EINVAL;
+			goto out;
+		}
+
+		if (!(pdd->exception_status & KFD_EC_MASK(exception_code))) {
+			r = -ENODATA;
+			goto out;
+		}
+
+		if (exception_code == EC_DEVICE_MEMORY_VIOLATION) {
+			copy_size = min((size_t)(*info_size), pdd->vm_fault_exc_data_size);
+
+			if (copy_to_user(info, pdd->vm_fault_exc_data, copy_size)) {
+				r = -EFAULT;
+				goto out;
+			}
+			actual_info_size = pdd->vm_fault_exc_data_size;
+			if (clear_exception) {
+				kfree(pdd->vm_fault_exc_data);
+				pdd->vm_fault_exc_data = NULL;
+				pdd->vm_fault_exc_data_size = 0;
+			}
+		}
+		exception_status_ptr = &pdd->exception_status;
+	} else if (KFD_DBG_EC_TYPE_IS_PROCESS(exception_code)) {
+		/* Per process exceptions */
+		if (!(target->exception_status & KFD_EC_MASK(exception_code))) {
+			r = -ENODATA;
+			goto out;
+		}
+
+		if (exception_code == EC_PROCESS_RUNTIME) {
+			copy_size = min((size_t)(*info_size), sizeof(target->runtime_info));
+
+			if (copy_to_user(info, (void *)&target->runtime_info, copy_size)) {
+				r = -EFAULT;
+				goto out;
+			}
+
+			actual_info_size = sizeof(target->runtime_info);
+		}
+
+		exception_status_ptr = &target->exception_status;
+	} else {
+		pr_debug("Bad exception type [%i]\n", exception_code);
+		r = -EINVAL;
+		goto out;
+	}
+
+	*info_size = actual_info_size;
+	if (clear_exception)
+		*exception_status_ptr &= ~KFD_EC_MASK(exception_code);
+out:
+	mutex_unlock(&target->event_mutex);
+	return r;
+}
+
+int kfd_dbg_trap_device_snapshot(struct kfd_process *target,
+		uint64_t exception_clear_mask,
+		void __user *user_info,
+		uint32_t *number_of_device_infos,
+		uint32_t *entry_size)
+{
+	struct kfd_dbg_device_info_entry device_info;
+	uint32_t tmp_entry_size, tmp_num_devices;
+	int i, r = 0;
+
+	if (!(target && user_info && number_of_device_infos && entry_size))
+		return -EINVAL;
+
+	tmp_entry_size = *entry_size;
+
+	tmp_num_devices = min_t(size_t, *number_of_device_infos, target->n_pdds);
+	*number_of_device_infos = target->n_pdds;
+	*entry_size = min_t(size_t, *entry_size, sizeof(device_info));
+
+	if (!tmp_num_devices)
+		return 0;
+
+	memset(&device_info, 0, sizeof(device_info));
+
+	mutex_lock(&target->event_mutex);
+
+	/* Run over all pdd of the process */
+	for (i = 0; i < tmp_num_devices; i++) {
+		struct kfd_process_device *pdd = target->pdds[i];
+		struct kfd_topology_device *topo_dev = kfd_topology_device_by_id(pdd->dev->id);
+
+		device_info.gpu_id = pdd->dev->id;
+		device_info.exception_status = pdd->exception_status;
+		device_info.lds_base = pdd->lds_base;
+		device_info.lds_limit = pdd->lds_limit;
+		device_info.scratch_base = pdd->scratch_base;
+		device_info.scratch_limit = pdd->scratch_limit;
+		device_info.gpuvm_base = pdd->gpuvm_base;
+		device_info.gpuvm_limit = pdd->gpuvm_limit;
+		device_info.location_id = topo_dev->node_props.location_id;
+		device_info.vendor_id = topo_dev->node_props.vendor_id;
+		device_info.device_id = topo_dev->node_props.device_id;
+		device_info.revision_id = pdd->dev->adev->pdev->revision;
+		device_info.subsystem_vendor_id = pdd->dev->adev->pdev->subsystem_vendor;
+		device_info.subsystem_device_id = pdd->dev->adev->pdev->subsystem_device;
+		device_info.fw_version = pdd->dev->kfd->mec_fw_version;
+		device_info.gfx_target_version =
+			topo_dev->node_props.gfx_target_version;
+		device_info.simd_count = topo_dev->node_props.simd_count;
+		device_info.max_waves_per_simd =
+			topo_dev->node_props.max_waves_per_simd;
+		device_info.array_count = topo_dev->node_props.array_count;
+		device_info.simd_arrays_per_engine =
+			topo_dev->node_props.simd_arrays_per_engine;
+		device_info.num_xcc = NUM_XCC(pdd->dev->xcc_mask);
+		device_info.capability = topo_dev->node_props.capability;
+		device_info.debug_prop = topo_dev->node_props.debug_prop;
+
+		if (exception_clear_mask)
+			pdd->exception_status &= ~exception_clear_mask;
+
+		if (copy_to_user(user_info, &device_info, *entry_size)) {
+			r = -EFAULT;
+			break;
+		}
+
+		user_info += tmp_entry_size;
+	}
+
+	mutex_unlock(&target->event_mutex);
+
+	return r;
+}
+
+void kfd_dbg_set_enabled_debug_exception_mask(struct kfd_process *target,
+					uint64_t exception_set_mask)
+{
+	uint64_t found_mask = 0;
+	struct process_queue_manager *pqm;
+	struct process_queue_node *pqn;
+	static const char write_data = '.';
+	loff_t pos = 0;
+	int i;
+
+	mutex_lock(&target->event_mutex);
+
+	found_mask |= target->exception_status;
+
+	pqm = &target->pqm;
+	list_for_each_entry(pqn, &pqm->queues, process_queue_list) {
+		if (!pqn->q)
+			continue;
+
+		found_mask |= pqn->q->properties.exception_status;
+	}
+
+	for (i = 0; i < target->n_pdds; i++) {
+		struct kfd_process_device *pdd = target->pdds[i];
+
+		found_mask |= pdd->exception_status;
+	}
+
+	if (exception_set_mask & found_mask)
+		kernel_write(target->dbg_ev_file, &write_data, 1, &pos);
+
+	target->exception_enable_mask = exception_set_mask;
+
+	mutex_unlock(&target->event_mutex);
+}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_debug.h b/drivers/gpu/drm/amd/amdkfd/kfd_debug.h
new file mode 100644
index 000000000000..27aa1a5b120f
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_debug.h
@@ -0,0 +1,142 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#ifndef KFD_DEBUG_EVENTS_H_INCLUDED
+#define KFD_DEBUG_EVENTS_H_INCLUDED
+
+#include "kfd_priv.h"
+
+void kfd_dbg_trap_deactivate(struct kfd_process *target, bool unwind, int unwind_count);
+int kfd_dbg_trap_activate(struct kfd_process *target);
+int kfd_dbg_ev_query_debug_event(struct kfd_process *process,
+			unsigned int *queue_id,
+			unsigned int *gpu_id,
+			uint64_t exception_clear_mask,
+			uint64_t *event_status);
+bool kfd_set_dbg_ev_from_interrupt(struct kfd_node *dev,
+				   unsigned int pasid,
+				   uint32_t doorbell_id,
+				   uint64_t trap_mask,
+				   void *exception_data,
+				   size_t exception_data_size);
+bool kfd_dbg_ev_raise(uint64_t event_mask,
+			struct kfd_process *process, struct kfd_node *dev,
+			unsigned int source_id, bool use_worker,
+			void *exception_data,
+			size_t exception_data_size);
+int kfd_dbg_trap_disable(struct kfd_process *target);
+int kfd_dbg_trap_enable(struct kfd_process *target, uint32_t fd,
+			void __user *runtime_info,
+			uint32_t *runtime_info_size);
+int kfd_dbg_trap_set_wave_launch_override(struct kfd_process *target,
+					uint32_t trap_override,
+					uint32_t trap_mask_bits,
+					uint32_t trap_mask_request,
+					uint32_t *trap_mask_prev,
+					uint32_t *trap_mask_supported);
+int kfd_dbg_trap_set_wave_launch_mode(struct kfd_process *target,
+					uint8_t wave_launch_mode);
+int kfd_dbg_trap_clear_dev_address_watch(struct kfd_process_device *pdd,
+					uint32_t watch_id);
+int kfd_dbg_trap_set_dev_address_watch(struct kfd_process_device *pdd,
+					uint64_t watch_address,
+					uint32_t watch_address_mask,
+					uint32_t *watch_id,
+					uint32_t watch_mode);
+int kfd_dbg_trap_set_flags(struct kfd_process *target, uint32_t *flags);
+int kfd_dbg_trap_query_exception_info(struct kfd_process *target,
+		uint32_t source_id,
+		uint32_t exception_code,
+		bool clear_exception,
+		void __user *info,
+		uint32_t *info_size);
+int kfd_dbg_send_exception_to_runtime(struct kfd_process *p,
+					unsigned int dev_id,
+					unsigned int queue_id,
+					uint64_t error_reason);
+
+static inline bool kfd_dbg_is_per_vmid_supported(struct kfd_node *dev)
+{
+	return (KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 2) ||
+		KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 3) ||
+		KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 4) ||
+		KFD_GC_VERSION(dev) == IP_VERSION(9, 5, 0) ||
+		KFD_GC_VERSION(dev) >= IP_VERSION(11, 0, 0));
+}
+
+void debug_event_write_work_handler(struct work_struct *work);
+int kfd_dbg_trap_device_snapshot(struct kfd_process *target,
+		uint64_t exception_clear_mask,
+		void __user *user_info,
+		uint32_t *number_of_device_infos,
+		uint32_t *entry_size);
+
+void kfd_dbg_set_enabled_debug_exception_mask(struct kfd_process *target,
+					uint64_t exception_set_mask);
+/*
+ * If GFX off is enabled, chips that do not support RLC restore for the debug
+ * registers will disable GFX off temporarily for the entire debug session.
+ * See disable_on_trap_action_entry and enable_on_trap_action_exit for details.
+ */
+static inline bool kfd_dbg_is_rlc_restore_supported(struct kfd_node *dev)
+{
+	return !(KFD_GC_VERSION(dev) == IP_VERSION(10, 1, 10) ||
+		 KFD_GC_VERSION(dev) == IP_VERSION(10, 1, 1));
+}
+
+static inline bool kfd_dbg_has_cwsr_workaround(struct kfd_node *dev)
+{
+	return KFD_GC_VERSION(dev) >= IP_VERSION(11, 0, 0) &&
+	       KFD_GC_VERSION(dev) <= IP_VERSION(11, 0, 3);
+}
+
+static inline bool kfd_dbg_has_gws_support(struct kfd_node *dev)
+{
+	if ((KFD_GC_VERSION(dev) == IP_VERSION(9, 0, 1)
+			&& dev->kfd->mec2_fw_version < 0x81b6) ||
+		(KFD_GC_VERSION(dev) >= IP_VERSION(9, 1, 0)
+			&& KFD_GC_VERSION(dev) <= IP_VERSION(9, 2, 2)
+			&& dev->kfd->mec2_fw_version < 0x1b6) ||
+		(KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 0)
+			&& dev->kfd->mec2_fw_version < 0x1b6) ||
+		(KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 1)
+			&& dev->kfd->mec2_fw_version < 0x30) ||
+		(KFD_GC_VERSION(dev) >= IP_VERSION(11, 0, 0) &&
+			KFD_GC_VERSION(dev) < IP_VERSION(12, 0, 0)))
+		return false;
+
+	/* Assume debugging and cooperative launch supported otherwise. */
+	return true;
+}
+
+int kfd_dbg_set_mes_debug_mode(struct kfd_process_device *pdd, bool sq_trap_en);
+
+static inline bool kfd_dbg_has_ttmps_always_setup(struct kfd_node *dev)
+{
+	return (KFD_GC_VERSION(dev) < IP_VERSION(11, 0, 0) &&
+			KFD_GC_VERSION(dev) != IP_VERSION(9, 4, 2)) ||
+	       (KFD_GC_VERSION(dev) >= IP_VERSION(11, 0, 0) &&
+			KFD_GC_VERSION(dev) < IP_VERSION(12, 0, 0) &&
+			(dev->adev->mes.sched_version & AMDGPU_MES_VERSION_MASK) >= 70) ||
+	       (KFD_GC_VERSION(dev) >= IP_VERSION(12, 0, 0));
+}
+#endif
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_debugfs.c b/drivers/gpu/drm/amd/amdkfd/kfd_debugfs.c
new file mode 100644
index 000000000000..9bde2c64540f
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_debugfs.c
@@ -0,0 +1,187 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
+/*
+ * Copyright 2016-2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include <linux/debugfs.h>
+#include <linux/uaccess.h>
+
+#include "kfd_priv.h"
+
+static struct dentry *debugfs_root;
+static struct dentry *debugfs_proc;
+static struct list_head procs;
+
+struct debugfs_proc_entry {
+	struct list_head list;
+	struct dentry *proc_dentry;
+	pid_t pid;
+};
+
+#define MAX_DEBUGFS_FILENAME_LEN 32
+
+static int kfd_debugfs_open(struct inode *inode, struct file *file)
+{
+	int (*show)(struct seq_file *, void *) = inode->i_private;
+
+	return single_open(file, show, NULL);
+}
+static int kfd_debugfs_hang_hws_read(struct seq_file *m, void *data)
+{
+	seq_puts(m, "echo gpu_id > hang_hws\n");
+	return 0;
+}
+
+static ssize_t kfd_debugfs_hang_hws_write(struct file *file,
+	const char __user *user_buf, size_t size, loff_t *ppos)
+{
+	struct kfd_node *dev;
+	char tmp[16];
+	uint32_t gpu_id;
+	int ret = -EINVAL;
+
+	memset(tmp, 0, 16);
+	if (size >= 16) {
+		pr_err("Invalid input for gpu id.\n");
+		goto out;
+	}
+	if (copy_from_user(tmp, user_buf, size)) {
+		ret = -EFAULT;
+		goto out;
+	}
+	if (kstrtoint(tmp, 10, &gpu_id)) {
+		pr_err("Invalid input for gpu id.\n");
+		goto out;
+	}
+	dev = kfd_device_by_id(gpu_id);
+	if (dev) {
+		kfd_debugfs_hang_hws(dev);
+		ret = size;
+	} else
+		pr_err("Cannot find device %d.\n", gpu_id);
+
+out:
+	return ret;
+}
+
+static const struct file_operations kfd_debugfs_fops = {
+	.owner = THIS_MODULE,
+	.open = kfd_debugfs_open,
+	.read = seq_read,
+	.llseek = seq_lseek,
+	.release = single_release,
+};
+
+static const struct file_operations kfd_debugfs_hang_hws_fops = {
+	.owner = THIS_MODULE,
+	.open = kfd_debugfs_open,
+	.read = seq_read,
+	.write = kfd_debugfs_hang_hws_write,
+	.llseek = seq_lseek,
+	.release = single_release,
+};
+
+void kfd_debugfs_init(void)
+{
+	debugfs_root = debugfs_create_dir("kfd", NULL);
+	debugfs_proc = debugfs_create_dir("proc", debugfs_root);
+	INIT_LIST_HEAD(&procs);
+
+	debugfs_create_file("mqds", S_IFREG | 0444, debugfs_root,
+			    kfd_debugfs_mqds_by_process, &kfd_debugfs_fops);
+	debugfs_create_file("hqds", S_IFREG | 0444, debugfs_root,
+			    kfd_debugfs_hqds_by_device, &kfd_debugfs_fops);
+	debugfs_create_file("rls", S_IFREG | 0444, debugfs_root,
+			    kfd_debugfs_rls_by_device, &kfd_debugfs_fops);
+	debugfs_create_file("hang_hws", S_IFREG | 0200, debugfs_root,
+			    kfd_debugfs_hang_hws_read, &kfd_debugfs_hang_hws_fops);
+	debugfs_create_file("mem_limit", S_IFREG | 0200, debugfs_root,
+			    kfd_debugfs_kfd_mem_limits, &kfd_debugfs_fops);
+}
+
+void kfd_debugfs_fini(void)
+{
+	debugfs_remove_recursive(debugfs_proc);
+	debugfs_remove_recursive(debugfs_root);
+}
+
+static ssize_t kfd_debugfs_pasid_read(struct file *file, char __user *buf,
+				      size_t count, loff_t *ppos)
+{
+	struct kfd_process_device *pdd = file_inode(file)->i_private;
+	char tmp[32];
+	int len;
+
+	len = snprintf(tmp, sizeof(tmp), "%u\n", pdd->pasid);
+
+	return simple_read_from_buffer(buf, count, ppos, tmp, len);
+}
+
+static const struct file_operations kfd_debugfs_pasid_fops = {
+	.owner = THIS_MODULE,
+	.read = kfd_debugfs_pasid_read,
+};
+
+void kfd_debugfs_add_process(struct kfd_process *p)
+{
+	int i;
+	char name[MAX_DEBUGFS_FILENAME_LEN];
+	struct debugfs_proc_entry *entry;
+
+	entry = kzalloc(sizeof(*entry), GFP_KERNEL);
+	if (!entry)
+		return;
+
+	list_add(&entry->list, &procs);
+	entry->pid = p->lead_thread->pid;
+	snprintf(name, MAX_DEBUGFS_FILENAME_LEN, "%d",
+		 (int)entry->pid);
+	entry->proc_dentry = debugfs_create_dir(name, debugfs_proc);
+
+	/* Create debugfs files for each GPU:
+	 * - proc/<pid>/pasid_<gpuid>
+	 */
+	for (i = 0; i < p->n_pdds; i++) {
+		struct kfd_process_device *pdd = p->pdds[i];
+
+		snprintf(name, MAX_DEBUGFS_FILENAME_LEN, "pasid_%u",
+			 pdd->dev->id);
+		debugfs_create_file((const char *)name, S_IFREG | 0444,
+				    entry->proc_dentry, pdd,
+				    &kfd_debugfs_pasid_fops);
+	}
+}
+
+void kfd_debugfs_remove_process(struct kfd_process *p)
+{
+	struct debugfs_proc_entry *entry, *next;
+
+	mutex_lock(&kfd_processes_mutex);
+	list_for_each_entry_safe(entry, next, &procs, list) {
+		if (entry->pid != p->lead_thread->pid)
+			continue;
+
+		debugfs_remove_recursive(entry->proc_dentry);
+		list_del(&entry->list);
+		kfree(entry);
+	}
+	mutex_unlock(&kfd_processes_mutex);
+}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device.c b/drivers/gpu/drm/amd/amdkfd/kfd_device.c
index 61fff25b4ce7..e9cfb80bd436 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device.c
@@ -1,5 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -20,105 +21,464 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  */
 
-#include <linux/amd-iommu.h>
 #include <linux/bsearch.h>
 #include <linux/pci.h>
 #include <linux/slab.h>
 #include "kfd_priv.h"
 #include "kfd_device_queue_manager.h"
 #include "kfd_pm4_headers_vi.h"
+#include "kfd_pm4_headers_aldebaran.h"
+#include "cwsr_trap_handler.h"
+#include "amdgpu_amdkfd.h"
+#include "kfd_smi_events.h"
+#include "kfd_svm.h"
+#include "kfd_migrate.h"
+#include "amdgpu.h"
+#include "amdgpu_xcp.h"
 
 #define MQD_SIZE_ALIGNED 768
 
-static const struct kfd_device_info kaveri_device_info = {
-	.asic_family = CHIP_KAVERI,
-	.max_pasid_bits = 16,
-	/* max num of queues for KV.TODO should be a dynamic value */
-	.max_no_of_hqd	= 24,
-	.ih_ring_entry_size = 4 * sizeof(uint32_t),
-	.event_interrupt_class = &event_interrupt_class_cik,
-	.num_of_watch_points = 4,
-	.mqd_size_aligned = MQD_SIZE_ALIGNED
-};
-
-static const struct kfd_device_info carrizo_device_info = {
-	.asic_family = CHIP_CARRIZO,
-	.max_pasid_bits = 16,
-	/* max num of queues for CZ.TODO should be a dynamic value */
-	.max_no_of_hqd	= 24,
-	.ih_ring_entry_size = 4 * sizeof(uint32_t),
-	.event_interrupt_class = &event_interrupt_class_cik,
-	.num_of_watch_points = 4,
-	.mqd_size_aligned = MQD_SIZE_ALIGNED
-};
-
-struct kfd_deviceid {
-	unsigned short did;
-	const struct kfd_device_info *device_info;
-};
-
-/* Please keep this sorted by increasing device id. */
-static const struct kfd_deviceid supported_devices[] = {
-	{ 0x1304, &kaveri_device_info },	/* Kaveri */
-	{ 0x1305, &kaveri_device_info },	/* Kaveri */
-	{ 0x1306, &kaveri_device_info },	/* Kaveri */
-	{ 0x1307, &kaveri_device_info },	/* Kaveri */
-	{ 0x1309, &kaveri_device_info },	/* Kaveri */
-	{ 0x130A, &kaveri_device_info },	/* Kaveri */
-	{ 0x130B, &kaveri_device_info },	/* Kaveri */
-	{ 0x130C, &kaveri_device_info },	/* Kaveri */
-	{ 0x130D, &kaveri_device_info },	/* Kaveri */
-	{ 0x130E, &kaveri_device_info },	/* Kaveri */
-	{ 0x130F, &kaveri_device_info },	/* Kaveri */
-	{ 0x1310, &kaveri_device_info },	/* Kaveri */
-	{ 0x1311, &kaveri_device_info },	/* Kaveri */
-	{ 0x1312, &kaveri_device_info },	/* Kaveri */
-	{ 0x1313, &kaveri_device_info },	/* Kaveri */
-	{ 0x1315, &kaveri_device_info },	/* Kaveri */
-	{ 0x1316, &kaveri_device_info },	/* Kaveri */
-	{ 0x1317, &kaveri_device_info },	/* Kaveri */
-	{ 0x1318, &kaveri_device_info },	/* Kaveri */
-	{ 0x131B, &kaveri_device_info },	/* Kaveri */
-	{ 0x131C, &kaveri_device_info },	/* Kaveri */
-	{ 0x131D, &kaveri_device_info },	/* Kaveri */
-	{ 0x9870, &carrizo_device_info },	/* Carrizo */
-	{ 0x9874, &carrizo_device_info },	/* Carrizo */
-	{ 0x9875, &carrizo_device_info },	/* Carrizo */
-	{ 0x9876, &carrizo_device_info },	/* Carrizo */
-	{ 0x9877, &carrizo_device_info }	/* Carrizo */
-};
+/*
+ * kfd_locked is used to lock the kfd driver during suspend or reset
+ * once locked, kfd driver will stop any further GPU execution.
+ * create process (open) will return -EAGAIN.
+ */
+static int kfd_locked;
+
+#ifdef CONFIG_DRM_AMDGPU_CIK
+extern const struct kfd2kgd_calls gfx_v7_kfd2kgd;
+#endif
+extern const struct kfd2kgd_calls gfx_v8_kfd2kgd;
+extern const struct kfd2kgd_calls gfx_v9_kfd2kgd;
+extern const struct kfd2kgd_calls arcturus_kfd2kgd;
+extern const struct kfd2kgd_calls aldebaran_kfd2kgd;
+extern const struct kfd2kgd_calls gc_9_4_3_kfd2kgd;
+extern const struct kfd2kgd_calls gfx_v10_kfd2kgd;
+extern const struct kfd2kgd_calls gfx_v10_3_kfd2kgd;
+extern const struct kfd2kgd_calls gfx_v11_kfd2kgd;
+extern const struct kfd2kgd_calls gfx_v12_kfd2kgd;
 
 static int kfd_gtt_sa_init(struct kfd_dev *kfd, unsigned int buf_size,
 				unsigned int chunk_size);
 static void kfd_gtt_sa_fini(struct kfd_dev *kfd);
 
-static const struct kfd_device_info *lookup_device_info(unsigned short did)
-{
-	size_t i;
+static int kfd_resume(struct kfd_node *kfd);
 
-	for (i = 0; i < ARRAY_SIZE(supported_devices); i++) {
-		if (supported_devices[i].did == did) {
-			WARN_ON(!supported_devices[i].device_info);
-			return supported_devices[i].device_info;
-		}
+static void kfd_device_info_set_sdma_info(struct kfd_dev *kfd)
+{
+	uint32_t sdma_version = amdgpu_ip_version(kfd->adev, SDMA0_HWIP, 0);
+
+	switch (sdma_version) {
+	case IP_VERSION(4, 0, 0):/* VEGA10 */
+	case IP_VERSION(4, 0, 1):/* VEGA12 */
+	case IP_VERSION(4, 1, 0):/* RAVEN */
+	case IP_VERSION(4, 1, 1):/* RAVEN */
+	case IP_VERSION(4, 1, 2):/* RENOIR */
+	case IP_VERSION(5, 2, 1):/* VANGOGH */
+	case IP_VERSION(5, 2, 3):/* YELLOW_CARP */
+	case IP_VERSION(5, 2, 6):/* GC 10.3.6 */
+	case IP_VERSION(5, 2, 7):/* GC 10.3.7 */
+		kfd->device_info.num_sdma_queues_per_engine = 2;
+		break;
+	case IP_VERSION(4, 2, 0):/* VEGA20 */
+	case IP_VERSION(4, 2, 2):/* ARCTURUS */
+	case IP_VERSION(4, 4, 0):/* ALDEBARAN */
+	case IP_VERSION(4, 4, 2):
+	case IP_VERSION(4, 4, 5):
+	case IP_VERSION(4, 4, 4):
+	case IP_VERSION(5, 0, 0):/* NAVI10 */
+	case IP_VERSION(5, 0, 1):/* CYAN_SKILLFISH */
+	case IP_VERSION(5, 0, 2):/* NAVI14 */
+	case IP_VERSION(5, 0, 5):/* NAVI12 */
+	case IP_VERSION(5, 2, 0):/* SIENNA_CICHLID */
+	case IP_VERSION(5, 2, 2):/* NAVY_FLOUNDER */
+	case IP_VERSION(5, 2, 4):/* DIMGREY_CAVEFISH */
+	case IP_VERSION(5, 2, 5):/* BEIGE_GOBY */
+	case IP_VERSION(6, 0, 0):
+	case IP_VERSION(6, 0, 1):
+	case IP_VERSION(6, 0, 2):
+	case IP_VERSION(6, 0, 3):
+	case IP_VERSION(6, 1, 0):
+	case IP_VERSION(6, 1, 1):
+	case IP_VERSION(6, 1, 2):
+	case IP_VERSION(6, 1, 3):
+	case IP_VERSION(7, 0, 0):
+	case IP_VERSION(7, 0, 1):
+		kfd->device_info.num_sdma_queues_per_engine = 8;
+		break;
+	default:
+		dev_warn(kfd_device,
+			"Default sdma queue per engine(8) is set due to mismatch of sdma ip block(SDMA_HWIP:0x%x).\n",
+			sdma_version);
+		kfd->device_info.num_sdma_queues_per_engine = 8;
 	}
 
-	dev_warn(kfd_device, "DID %04x is missing in supported_devices\n",
-		 did);
+	bitmap_zero(kfd->device_info.reserved_sdma_queues_bitmap, KFD_MAX_SDMA_QUEUES);
+
+	switch (sdma_version) {
+	case IP_VERSION(6, 0, 0):
+	case IP_VERSION(6, 0, 1):
+	case IP_VERSION(6, 0, 2):
+	case IP_VERSION(6, 0, 3):
+	case IP_VERSION(6, 1, 0):
+	case IP_VERSION(6, 1, 1):
+	case IP_VERSION(6, 1, 2):
+	case IP_VERSION(6, 1, 3):
+	case IP_VERSION(7, 0, 0):
+	case IP_VERSION(7, 0, 1):
+		/* Reserve 1 for paging and 1 for gfx */
+		kfd->device_info.num_reserved_sdma_queues_per_engine = 2;
+		/* BIT(0)=engine-0 queue-0; BIT(1)=engine-1 queue-0; BIT(2)=engine-0 queue-1; ... */
+		bitmap_set(kfd->device_info.reserved_sdma_queues_bitmap, 0,
+			   kfd->adev->sdma.num_instances *
+			   kfd->device_info.num_reserved_sdma_queues_per_engine);
+		break;
+	default:
+		break;
+	}
+}
 
-	return NULL;
+static void kfd_device_info_set_event_interrupt_class(struct kfd_dev *kfd)
+{
+	uint32_t gc_version = KFD_GC_VERSION(kfd);
+
+	switch (gc_version) {
+	case IP_VERSION(9, 0, 1): /* VEGA10 */
+	case IP_VERSION(9, 1, 0): /* RAVEN */
+	case IP_VERSION(9, 2, 1): /* VEGA12 */
+	case IP_VERSION(9, 2, 2): /* RAVEN */
+	case IP_VERSION(9, 3, 0): /* RENOIR */
+	case IP_VERSION(9, 4, 0): /* VEGA20 */
+	case IP_VERSION(9, 4, 1): /* ARCTURUS */
+	case IP_VERSION(9, 4, 2): /* ALDEBARAN */
+		kfd->device_info.event_interrupt_class = &event_interrupt_class_v9;
+		break;
+	case IP_VERSION(9, 4, 3): /* GC 9.4.3 */
+	case IP_VERSION(9, 4, 4): /* GC 9.4.4 */
+	case IP_VERSION(9, 5, 0): /* GC 9.5.0 */
+		kfd->device_info.event_interrupt_class =
+						&event_interrupt_class_v9_4_3;
+		break;
+	case IP_VERSION(10, 3, 1): /* VANGOGH */
+	case IP_VERSION(10, 3, 3): /* YELLOW_CARP */
+	case IP_VERSION(10, 3, 6): /* GC 10.3.6 */
+	case IP_VERSION(10, 3, 7): /* GC 10.3.7 */
+	case IP_VERSION(10, 1, 3): /* CYAN_SKILLFISH */
+	case IP_VERSION(10, 1, 4):
+	case IP_VERSION(10, 1, 10): /* NAVI10 */
+	case IP_VERSION(10, 1, 2): /* NAVI12 */
+	case IP_VERSION(10, 1, 1): /* NAVI14 */
+	case IP_VERSION(10, 3, 0): /* SIENNA_CICHLID */
+	case IP_VERSION(10, 3, 2): /* NAVY_FLOUNDER */
+	case IP_VERSION(10, 3, 4): /* DIMGREY_CAVEFISH */
+	case IP_VERSION(10, 3, 5): /* BEIGE_GOBY */
+		kfd->device_info.event_interrupt_class = &event_interrupt_class_v10;
+		break;
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 1):
+	case IP_VERSION(11, 0, 2):
+	case IP_VERSION(11, 0, 3):
+	case IP_VERSION(11, 0, 4):
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 1):
+	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
+		kfd->device_info.event_interrupt_class = &event_interrupt_class_v11;
+		break;
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 0, 1):
+		/* GFX12_TODO: Change to v12 version. */
+		kfd->device_info.event_interrupt_class = &event_interrupt_class_v11;
+		break;
+	default:
+		dev_warn(kfd_device, "v9 event interrupt handler is set due to "
+			"mismatch of gc ip block(GC_HWIP:0x%x).\n", gc_version);
+		kfd->device_info.event_interrupt_class = &event_interrupt_class_v9;
+	}
 }
 
-struct kfd_dev *kgd2kfd_probe(struct kgd_dev *kgd,
-	struct pci_dev *pdev, const struct kfd2kgd_calls *f2g)
+static void kfd_device_info_init(struct kfd_dev *kfd,
+				 bool vf, uint32_t gfx_target_version)
 {
-	struct kfd_dev *kfd;
+	uint32_t gc_version = KFD_GC_VERSION(kfd);
+	uint32_t asic_type = kfd->adev->asic_type;
+
+	kfd->device_info.max_pasid_bits = 16;
+	kfd->device_info.max_no_of_hqd = 24;
+	kfd->device_info.num_of_watch_points = 4;
+	kfd->device_info.mqd_size_aligned = MQD_SIZE_ALIGNED;
+	kfd->device_info.gfx_target_version = gfx_target_version;
+
+	if (KFD_IS_SOC15(kfd)) {
+		kfd->device_info.doorbell_size = 8;
+		kfd->device_info.ih_ring_entry_size = 8 * sizeof(uint32_t);
+		kfd->device_info.supports_cwsr = true;
+
+		kfd_device_info_set_sdma_info(kfd);
+
+		kfd_device_info_set_event_interrupt_class(kfd);
+
+		if (gc_version < IP_VERSION(11, 0, 0)) {
+			/* Navi2x+, Navi1x+ */
+			if (gc_version == IP_VERSION(10, 3, 6))
+				kfd->device_info.no_atomic_fw_version = 14;
+			else if (gc_version == IP_VERSION(10, 3, 7))
+				kfd->device_info.no_atomic_fw_version = 3;
+			else if (gc_version >= IP_VERSION(10, 3, 0))
+				kfd->device_info.no_atomic_fw_version = 92;
+			else if (gc_version >= IP_VERSION(10, 1, 1))
+				kfd->device_info.no_atomic_fw_version = 145;
+
+			/* Navi1x+ */
+			if (gc_version >= IP_VERSION(10, 1, 1))
+				kfd->device_info.needs_pci_atomics = true;
+		} else if (gc_version < IP_VERSION(12, 0, 0)) {
+			/*
+			 * PCIe atomics support acknowledgment in GFX11 RS64 CPFW requires
+			 * MEC version >= 509. Prior RS64 CPFW versions (and all F32) require
+			 * PCIe atomics support.
+			 */
+			kfd->device_info.needs_pci_atomics = true;
+			kfd->device_info.no_atomic_fw_version = kfd->adev->gfx.rs64_enable ? 509 : 0;
+		} else if (gc_version < IP_VERSION(13, 0, 0)) {
+			kfd->device_info.needs_pci_atomics = true;
+			kfd->device_info.no_atomic_fw_version = 2090;
+		} else {
+			kfd->device_info.needs_pci_atomics = true;
+		}
+	} else {
+		kfd->device_info.doorbell_size = 4;
+		kfd->device_info.ih_ring_entry_size = 4 * sizeof(uint32_t);
+		kfd->device_info.event_interrupt_class = &event_interrupt_class_cik;
+		kfd->device_info.num_sdma_queues_per_engine = 2;
+
+		if (asic_type != CHIP_KAVERI &&
+		    asic_type != CHIP_HAWAII &&
+		    asic_type != CHIP_TONGA)
+			kfd->device_info.supports_cwsr = true;
+
+		if (asic_type != CHIP_HAWAII && !vf)
+			kfd->device_info.needs_pci_atomics = true;
+	}
+}
 
-	const struct kfd_device_info *device_info =
-					lookup_device_info(pdev->device);
+struct kfd_dev *kgd2kfd_probe(struct amdgpu_device *adev, bool vf)
+{
+	struct kfd_dev *kfd = NULL;
+	const struct kfd2kgd_calls *f2g = NULL;
+	uint32_t gfx_target_version = 0;
+
+	switch (adev->asic_type) {
+#ifdef CONFIG_DRM_AMDGPU_CIK
+	case CHIP_KAVERI:
+		gfx_target_version = 70000;
+		if (!vf)
+			f2g = &gfx_v7_kfd2kgd;
+		break;
+#endif
+	case CHIP_CARRIZO:
+		gfx_target_version = 80001;
+		if (!vf)
+			f2g = &gfx_v8_kfd2kgd;
+		break;
+#ifdef CONFIG_DRM_AMDGPU_CIK
+	case CHIP_HAWAII:
+		gfx_target_version = 70001;
+		if (!amdgpu_exp_hw_support)
+			pr_info(
+	"KFD support on Hawaii is experimental. See modparam exp_hw_support\n"
+				);
+		else if (!vf)
+			f2g = &gfx_v7_kfd2kgd;
+		break;
+#endif
+	case CHIP_TONGA:
+		gfx_target_version = 80002;
+		if (!vf)
+			f2g = &gfx_v8_kfd2kgd;
+		break;
+	case CHIP_FIJI:
+	case CHIP_POLARIS10:
+		gfx_target_version = 80003;
+		f2g = &gfx_v8_kfd2kgd;
+		break;
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
+		gfx_target_version = 80003;
+		if (!vf)
+			f2g = &gfx_v8_kfd2kgd;
+		break;
+	default:
+		switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+		/* Vega 10 */
+		case IP_VERSION(9, 0, 1):
+			gfx_target_version = 90000;
+			f2g = &gfx_v9_kfd2kgd;
+			break;
+		/* Raven */
+		case IP_VERSION(9, 1, 0):
+		case IP_VERSION(9, 2, 2):
+			gfx_target_version = 90002;
+			if (!vf)
+				f2g = &gfx_v9_kfd2kgd;
+			break;
+		/* Vega12 */
+		case IP_VERSION(9, 2, 1):
+			gfx_target_version = 90004;
+			if (!vf)
+				f2g = &gfx_v9_kfd2kgd;
+			break;
+		/* Renoir */
+		case IP_VERSION(9, 3, 0):
+			gfx_target_version = 90012;
+			if (!vf)
+				f2g = &gfx_v9_kfd2kgd;
+			break;
+		/* Vega20 */
+		case IP_VERSION(9, 4, 0):
+			gfx_target_version = 90006;
+			if (!vf)
+				f2g = &gfx_v9_kfd2kgd;
+			break;
+		/* Arcturus */
+		case IP_VERSION(9, 4, 1):
+			gfx_target_version = 90008;
+			f2g = &arcturus_kfd2kgd;
+			break;
+		/* Aldebaran */
+		case IP_VERSION(9, 4, 2):
+			gfx_target_version = 90010;
+			f2g = &aldebaran_kfd2kgd;
+			break;
+		case IP_VERSION(9, 4, 3):
+		case IP_VERSION(9, 4, 4):
+			gfx_target_version = 90402;
+			f2g = &gc_9_4_3_kfd2kgd;
+			break;
+		case IP_VERSION(9, 5, 0):
+			gfx_target_version = 90500;
+			f2g = &gc_9_4_3_kfd2kgd;
+			break;
+		/* Navi10 */
+		case IP_VERSION(10, 1, 10):
+			gfx_target_version = 100100;
+			if (!vf)
+				f2g = &gfx_v10_kfd2kgd;
+			break;
+		/* Navi12 */
+		case IP_VERSION(10, 1, 2):
+			gfx_target_version = 100101;
+			f2g = &gfx_v10_kfd2kgd;
+			break;
+		/* Navi14 */
+		case IP_VERSION(10, 1, 1):
+			gfx_target_version = 100102;
+			if (!vf)
+				f2g = &gfx_v10_kfd2kgd;
+			break;
+		/* Cyan Skillfish */
+		case IP_VERSION(10, 1, 3):
+		case IP_VERSION(10, 1, 4):
+			gfx_target_version = 100103;
+			if (!vf)
+				f2g = &gfx_v10_kfd2kgd;
+			break;
+		/* Sienna Cichlid */
+		case IP_VERSION(10, 3, 0):
+			gfx_target_version = 100300;
+			f2g = &gfx_v10_3_kfd2kgd;
+			break;
+		/* Navy Flounder */
+		case IP_VERSION(10, 3, 2):
+			gfx_target_version = 100301;
+			f2g = &gfx_v10_3_kfd2kgd;
+			break;
+		/* Van Gogh */
+		case IP_VERSION(10, 3, 1):
+			gfx_target_version = 100303;
+			if (!vf)
+				f2g = &gfx_v10_3_kfd2kgd;
+			break;
+		/* Dimgrey Cavefish */
+		case IP_VERSION(10, 3, 4):
+			gfx_target_version = 100302;
+			f2g = &gfx_v10_3_kfd2kgd;
+			break;
+		/* Beige Goby */
+		case IP_VERSION(10, 3, 5):
+			gfx_target_version = 100304;
+			f2g = &gfx_v10_3_kfd2kgd;
+			break;
+		/* Yellow Carp */
+		case IP_VERSION(10, 3, 3):
+			gfx_target_version = 100305;
+			if (!vf)
+				f2g = &gfx_v10_3_kfd2kgd;
+			break;
+		case IP_VERSION(10, 3, 6):
+		case IP_VERSION(10, 3, 7):
+			gfx_target_version = 100306;
+			if (!vf)
+				f2g = &gfx_v10_3_kfd2kgd;
+			break;
+		case IP_VERSION(11, 0, 0):
+			gfx_target_version = 110000;
+			f2g = &gfx_v11_kfd2kgd;
+			break;
+		case IP_VERSION(11, 0, 1):
+		case IP_VERSION(11, 0, 4):
+			gfx_target_version = 110003;
+			f2g = &gfx_v11_kfd2kgd;
+			break;
+		case IP_VERSION(11, 0, 2):
+			gfx_target_version = 110002;
+			f2g = &gfx_v11_kfd2kgd;
+			break;
+		case IP_VERSION(11, 0, 3):
+			/* Note: Compiler version is 11.0.1 while HW version is 11.0.3 */
+			gfx_target_version = 110001;
+			f2g = &gfx_v11_kfd2kgd;
+			break;
+		case IP_VERSION(11, 5, 0):
+			gfx_target_version = 110500;
+			f2g = &gfx_v11_kfd2kgd;
+			break;
+		case IP_VERSION(11, 5, 1):
+			gfx_target_version = 110501;
+			f2g = &gfx_v11_kfd2kgd;
+			break;
+		case IP_VERSION(11, 5, 2):
+			gfx_target_version = 110502;
+			f2g = &gfx_v11_kfd2kgd;
+			break;
+		case IP_VERSION(11, 5, 3):
+			gfx_target_version = 110503;
+			f2g = &gfx_v11_kfd2kgd;
+			break;
+		case IP_VERSION(12, 0, 0):
+			gfx_target_version = 120000;
+			f2g = &gfx_v12_kfd2kgd;
+			break;
+		case IP_VERSION(12, 0, 1):
+			gfx_target_version = 120001;
+			f2g = &gfx_v12_kfd2kgd;
+			break;
+		default:
+			break;
+		}
+		break;
+	}
 
-	if (!device_info) {
-		dev_err(kfd_device, "kgd2kfd_probe failed\n");
+	if (!f2g) {
+		if (amdgpu_ip_version(adev, GC_HWIP, 0))
+			dev_info(kfd_device,
+				"GC IP %06x %s not supported in kfd\n",
+				amdgpu_ip_version(adev, GC_HWIP, 0),
+				vf ? "VF" : "");
+		else
+			dev_info(kfd_device, "%s %s not supported in kfd\n",
+				amdgpu_asic_name[adev->asic_type], vf ? "VF" : "");
 		return NULL;
 	}
 
@@ -126,120 +486,316 @@ struct kfd_dev *kgd2kfd_probe(struct kgd_dev *kgd,
 	if (!kfd)
 		return NULL;
 
-	kfd->kgd = kgd;
-	kfd->device_info = device_info;
-	kfd->pdev = pdev;
+	kfd->adev = adev;
+	kfd_device_info_init(kfd, vf, gfx_target_version);
 	kfd->init_complete = false;
 	kfd->kfd2kgd = f2g;
+	atomic_set(&kfd->compute_profile, 0);
 
 	mutex_init(&kfd->doorbell_mutex);
-	memset(&kfd->doorbell_available_index, 0,
-		sizeof(kfd->doorbell_available_index));
+
+	ida_init(&kfd->doorbell_ida);
+	atomic_set(&kfd->kfd_processes_count, 0);
 
 	return kfd;
 }
 
-static bool device_iommu_pasid_init(struct kfd_dev *kfd)
+static void kfd_cwsr_init(struct kfd_dev *kfd)
 {
-	const u32 required_iommu_flags = AMD_IOMMU_DEVICE_FLAG_ATS_SUP |
-					AMD_IOMMU_DEVICE_FLAG_PRI_SUP |
-					AMD_IOMMU_DEVICE_FLAG_PASID_SUP;
+	if (cwsr_enable && kfd->device_info.supports_cwsr) {
+		if (KFD_GC_VERSION(kfd) < IP_VERSION(9, 0, 1)) {
+			BUILD_BUG_ON(sizeof(cwsr_trap_gfx8_hex)
+					     > KFD_CWSR_TMA_OFFSET);
+			kfd->cwsr_isa = cwsr_trap_gfx8_hex;
+			kfd->cwsr_isa_size = sizeof(cwsr_trap_gfx8_hex);
+		} else if (KFD_GC_VERSION(kfd) == IP_VERSION(9, 4, 1)) {
+			BUILD_BUG_ON(sizeof(cwsr_trap_arcturus_hex)
+					     > KFD_CWSR_TMA_OFFSET);
+			kfd->cwsr_isa = cwsr_trap_arcturus_hex;
+			kfd->cwsr_isa_size = sizeof(cwsr_trap_arcturus_hex);
+		} else if (KFD_GC_VERSION(kfd) == IP_VERSION(9, 4, 2)) {
+			BUILD_BUG_ON(sizeof(cwsr_trap_aldebaran_hex)
+					     > KFD_CWSR_TMA_OFFSET);
+			kfd->cwsr_isa = cwsr_trap_aldebaran_hex;
+			kfd->cwsr_isa_size = sizeof(cwsr_trap_aldebaran_hex);
+		} else if (KFD_GC_VERSION(kfd) == IP_VERSION(9, 4, 3) ||
+			   KFD_GC_VERSION(kfd) == IP_VERSION(9, 4, 4)) {
+			BUILD_BUG_ON(sizeof(cwsr_trap_gfx9_4_3_hex)
+					     > KFD_CWSR_TMA_OFFSET);
+			kfd->cwsr_isa = cwsr_trap_gfx9_4_3_hex;
+			kfd->cwsr_isa_size = sizeof(cwsr_trap_gfx9_4_3_hex);
+		} else if (KFD_GC_VERSION(kfd) == IP_VERSION(9, 5, 0)) {
+			BUILD_BUG_ON(sizeof(cwsr_trap_gfx9_5_0_hex) > PAGE_SIZE);
+			kfd->cwsr_isa = cwsr_trap_gfx9_5_0_hex;
+			kfd->cwsr_isa_size = sizeof(cwsr_trap_gfx9_5_0_hex);
+		} else if (KFD_GC_VERSION(kfd) < IP_VERSION(10, 1, 1)) {
+			BUILD_BUG_ON(sizeof(cwsr_trap_gfx9_hex)
+					     > KFD_CWSR_TMA_OFFSET);
+			kfd->cwsr_isa = cwsr_trap_gfx9_hex;
+			kfd->cwsr_isa_size = sizeof(cwsr_trap_gfx9_hex);
+		} else if (KFD_GC_VERSION(kfd) < IP_VERSION(10, 3, 0)) {
+			BUILD_BUG_ON(sizeof(cwsr_trap_nv1x_hex)
+					     > KFD_CWSR_TMA_OFFSET);
+			kfd->cwsr_isa = cwsr_trap_nv1x_hex;
+			kfd->cwsr_isa_size = sizeof(cwsr_trap_nv1x_hex);
+		} else if (KFD_GC_VERSION(kfd) < IP_VERSION(11, 0, 0)) {
+			BUILD_BUG_ON(sizeof(cwsr_trap_gfx10_hex)
+					     > KFD_CWSR_TMA_OFFSET);
+			kfd->cwsr_isa = cwsr_trap_gfx10_hex;
+			kfd->cwsr_isa_size = sizeof(cwsr_trap_gfx10_hex);
+		} else if (KFD_GC_VERSION(kfd) < IP_VERSION(12, 0, 0)) {
+			/* The gfx11 cwsr trap handler must fit inside a single
+			   page. */
+			BUILD_BUG_ON(sizeof(cwsr_trap_gfx11_hex) > PAGE_SIZE);
+			kfd->cwsr_isa = cwsr_trap_gfx11_hex;
+			kfd->cwsr_isa_size = sizeof(cwsr_trap_gfx11_hex);
+		} else {
+			BUILD_BUG_ON(sizeof(cwsr_trap_gfx12_hex)
+					     > KFD_CWSR_TMA_OFFSET);
+			kfd->cwsr_isa = cwsr_trap_gfx12_hex;
+			kfd->cwsr_isa_size = sizeof(cwsr_trap_gfx12_hex);
+		}
 
-	struct amd_iommu_device_info iommu_info;
-	unsigned int pasid_limit;
-	int err;
+		kfd->cwsr_enabled = true;
+	}
+}
 
-	err = amd_iommu_device_info(kfd->pdev, &iommu_info);
-	if (err < 0) {
-		dev_err(kfd_device,
-			"error getting iommu info. is the iommu enabled?\n");
-		return false;
+static int kfd_gws_init(struct kfd_node *node)
+{
+	int ret = 0;
+	struct kfd_dev *kfd = node->kfd;
+	uint32_t mes_rev = node->adev->mes.sched_version & AMDGPU_MES_VERSION_MASK;
+
+	if (node->dqm->sched_policy == KFD_SCHED_POLICY_NO_HWS)
+		return 0;
+
+	if (hws_gws_support || (KFD_IS_SOC15(node) &&
+		((KFD_GC_VERSION(node) == IP_VERSION(9, 0, 1)
+			&& kfd->mec2_fw_version >= 0x81b3) ||
+		(KFD_GC_VERSION(node) <= IP_VERSION(9, 4, 0)
+			&& kfd->mec2_fw_version >= 0x1b3)  ||
+		(KFD_GC_VERSION(node) == IP_VERSION(9, 4, 1)
+			&& kfd->mec2_fw_version >= 0x30)   ||
+		(KFD_GC_VERSION(node) == IP_VERSION(9, 4, 2)
+			&& kfd->mec2_fw_version >= 0x28) ||
+		(KFD_GC_VERSION(node) == IP_VERSION(9, 4, 3) ||
+		 KFD_GC_VERSION(node) == IP_VERSION(9, 4, 4)) ||
+		(KFD_GC_VERSION(node) == IP_VERSION(9, 5, 0)) ||
+		(KFD_GC_VERSION(node) >= IP_VERSION(10, 3, 0)
+			&& KFD_GC_VERSION(node) < IP_VERSION(11, 0, 0)
+			&& kfd->mec2_fw_version >= 0x6b) ||
+		(KFD_GC_VERSION(node) >= IP_VERSION(11, 0, 0)
+			&& KFD_GC_VERSION(node) < IP_VERSION(12, 0, 0)
+			&& mes_rev >= 68) ||
+		(KFD_GC_VERSION(node) >= IP_VERSION(12, 0, 0))))) {
+		if (KFD_GC_VERSION(node) >= IP_VERSION(12, 0, 0))
+			node->adev->gds.gws_size = 64;
+		ret = amdgpu_amdkfd_alloc_gws(node->adev,
+				node->adev->gds.gws_size, &node->gws);
 	}
 
-	if ((iommu_info.flags & required_iommu_flags) != required_iommu_flags) {
-		dev_err(kfd_device, "error required iommu flags ats %i, pri %i, pasid %i\n",
-		       (iommu_info.flags & AMD_IOMMU_DEVICE_FLAG_ATS_SUP) != 0,
-		       (iommu_info.flags & AMD_IOMMU_DEVICE_FLAG_PRI_SUP) != 0,
-		       (iommu_info.flags & AMD_IOMMU_DEVICE_FLAG_PASID_SUP)
-									!= 0);
-		return false;
+	return ret;
+}
+
+static void kfd_smi_init(struct kfd_node *dev)
+{
+	INIT_LIST_HEAD(&dev->smi_clients);
+	spin_lock_init(&dev->smi_lock);
+}
+
+static int kfd_init_node(struct kfd_node *node)
+{
+	int err = -1;
+
+	if (kfd_interrupt_init(node)) {
+		dev_err(kfd_device, "Error initializing interrupts\n");
+		goto kfd_interrupt_error;
 	}
 
-	pasid_limit = min_t(unsigned int,
-			(unsigned int)(1 << kfd->device_info->max_pasid_bits),
-			iommu_info.max_pasids);
-	/*
-	 * last pasid is used for kernel queues doorbells
-	 * in the future the last pasid might be used for a kernel thread.
-	 */
-	pasid_limit = min_t(unsigned int,
-				pasid_limit,
-				kfd->doorbell_process_limit - 1);
+	node->dqm = device_queue_manager_init(node);
+	if (!node->dqm) {
+		dev_err(kfd_device, "Error initializing queue manager\n");
+		goto device_queue_manager_error;
+	}
 
-	err = amd_iommu_init_device(kfd->pdev, pasid_limit);
-	if (err < 0) {
-		dev_err(kfd_device, "error initializing iommu device\n");
-		return false;
+	if (kfd_gws_init(node)) {
+		dev_err(kfd_device, "Could not allocate %d gws\n",
+			node->adev->gds.gws_size);
+		goto gws_error;
 	}
 
-	if (!kfd_set_pasid_limit(pasid_limit)) {
-		dev_err(kfd_device, "error setting pasid limit\n");
-		amd_iommu_free_device(kfd->pdev);
-		return false;
+	if (kfd_resume(node))
+		goto kfd_resume_error;
+
+	if (kfd_topology_add_device(node)) {
+		dev_err(kfd_device, "Error adding device to topology\n");
+		goto kfd_topology_add_device_error;
 	}
 
-	return true;
-}
+	kfd_smi_init(node);
 
-static void iommu_pasid_shutdown_callback(struct pci_dev *pdev, int pasid)
-{
-	struct kfd_dev *dev = kfd_device_by_pci_dev(pdev);
+	return 0;
 
-	if (dev)
-		kfd_unbind_process_from_device(dev, pasid);
+kfd_topology_add_device_error:
+kfd_resume_error:
+gws_error:
+	device_queue_manager_uninit(node->dqm);
+device_queue_manager_error:
+	kfd_interrupt_exit(node);
+kfd_interrupt_error:
+	if (node->gws)
+		amdgpu_amdkfd_free_gws(node->adev, node->gws);
+
+	/* Cleanup the node memory here */
+	kfree(node);
+	return err;
 }
 
-/*
- * This function called by IOMMU driver on PPR failure
- */
-static int iommu_invalid_ppr_cb(struct pci_dev *pdev, int pasid,
-		unsigned long address, u16 flags)
+static void kfd_cleanup_nodes(struct kfd_dev *kfd, unsigned int num_nodes)
 {
-	struct kfd_dev *dev;
+	struct kfd_node *knode;
+	unsigned int i;
 
-	dev_warn(kfd_device,
-			"Invalid PPR device %x:%x.%x pasid %d address 0x%lX flags 0x%X",
-			PCI_BUS_NUM(pdev->devfn),
-			PCI_SLOT(pdev->devfn),
-			PCI_FUNC(pdev->devfn),
-			pasid,
-			address,
-			flags);
-
-	dev = kfd_device_by_pci_dev(pdev);
-	if (!WARN_ON(!dev))
-		kfd_signal_iommu_event(dev, pasid, address,
-			flags & PPR_FAULT_WRITE, flags & PPR_FAULT_EXEC);
+	/*
+	 * flush_work ensures that there are no outstanding
+	 * work-queue items that will access interrupt_ring. New work items
+	 * can't be created because we stopped interrupt handling above.
+	 */
+	flush_workqueue(kfd->ih_wq);
+	destroy_workqueue(kfd->ih_wq);
+
+	for (i = 0; i < num_nodes; i++) {
+		knode = kfd->nodes[i];
+		device_queue_manager_uninit(knode->dqm);
+		kfd_interrupt_exit(knode);
+		kfd_topology_remove_device(knode);
+		if (knode->gws)
+			amdgpu_amdkfd_free_gws(knode->adev, knode->gws);
+		kfree(knode);
+		kfd->nodes[i] = NULL;
+	}
+}
 
-	return AMD_IOMMU_INV_PRI_RSP_INVALID;
+static void kfd_setup_interrupt_bitmap(struct kfd_node *node,
+				       unsigned int kfd_node_idx)
+{
+	struct amdgpu_device *adev = node->adev;
+	uint32_t xcc_mask = node->xcc_mask;
+	uint32_t xcc, mapped_xcc;
+	/*
+	 * Interrupt bitmap is setup for processing interrupts from
+	 * different XCDs and AIDs.
+	 * Interrupt bitmap is defined as follows:
+	 * 1. Bits 0-15 - correspond to the NodeId field.
+	 *    Each bit corresponds to NodeId number. For example, if
+	 *    a KFD node has interrupt bitmap set to 0x7, then this
+	 *    KFD node will process interrupts with NodeId = 0, 1 and 2
+	 *    in the IH cookie.
+	 * 2. Bits 16-31 - unused.
+	 *
+	 * Please note that the kfd_node_idx argument passed to this
+	 * function is not related to NodeId field received in the
+	 * IH cookie.
+	 *
+	 * In CPX mode, a KFD node will process an interrupt if:
+	 * - the Node Id matches the corresponding bit set in
+	 *   Bits 0-15.
+	 * - AND VMID reported in the interrupt lies within the
+	 *   VMID range of the node.
+	 */
+	for_each_inst(xcc, xcc_mask) {
+		mapped_xcc = GET_INST(GC, xcc);
+		node->interrupt_bitmap |= (mapped_xcc % 2 ? 5 : 3) << (4 * (mapped_xcc / 2));
+	}
+	dev_info(kfd_device, "Node: %d, interrupt_bitmap: %x\n", kfd_node_idx,
+							node->interrupt_bitmap);
 }
 
 bool kgd2kfd_device_init(struct kfd_dev *kfd,
 			 const struct kgd2kfd_shared_resources *gpu_resources)
 {
-	unsigned int size;
-
+	unsigned int size, map_process_packet_size, i;
+	struct kfd_node *node;
+	uint32_t first_vmid_kfd, last_vmid_kfd, vmid_num_kfd;
+	unsigned int max_proc_per_quantum;
+	int partition_mode;
+	int xcp_idx;
+
+	kfd->mec_fw_version = amdgpu_amdkfd_get_fw_version(kfd->adev,
+			KGD_ENGINE_MEC1);
+	kfd->mec2_fw_version = amdgpu_amdkfd_get_fw_version(kfd->adev,
+			KGD_ENGINE_MEC2);
+	kfd->sdma_fw_version = amdgpu_amdkfd_get_fw_version(kfd->adev,
+			KGD_ENGINE_SDMA1);
 	kfd->shared_resources = *gpu_resources;
 
+	kfd->num_nodes = amdgpu_xcp_get_num_xcp(kfd->adev->xcp_mgr);
+
+	if (kfd->num_nodes == 0) {
+		dev_err(kfd_device,
+			"KFD num nodes cannot be 0, num_xcc_in_node: %d\n",
+			kfd->adev->gfx.num_xcc_per_xcp);
+		goto out;
+	}
+
+	/* Allow BIF to recode atomics to PCIe 3.0 AtomicOps.
+	 * 32 and 64-bit requests are possible and must be
+	 * supported.
+	 */
+	kfd->pci_atomic_requested = amdgpu_amdkfd_have_atomics_support(kfd->adev);
+	if (!kfd->pci_atomic_requested &&
+	    kfd->device_info.needs_pci_atomics &&
+	    (!kfd->device_info.no_atomic_fw_version ||
+	     kfd->mec_fw_version < kfd->device_info.no_atomic_fw_version)) {
+		dev_info(kfd_device,
+			 "skipped device %x:%x, PCI rejects atomics %d<%d\n",
+			 kfd->adev->pdev->vendor, kfd->adev->pdev->device,
+			 kfd->mec_fw_version,
+			 kfd->device_info.no_atomic_fw_version);
+		return false;
+	}
+
+	first_vmid_kfd = ffs(gpu_resources->compute_vmid_bitmap)-1;
+	last_vmid_kfd = fls(gpu_resources->compute_vmid_bitmap)-1;
+	vmid_num_kfd = last_vmid_kfd - first_vmid_kfd + 1;
+
+	/* For multi-partition capable GPUs, we need special handling for VMIDs
+	 * depending on partition mode.
+	 * In CPX mode, the VMID range needs to be shared between XCDs.
+	 * Additionally, there are 13 VMIDs (3-15) available for KFD. To
+	 * divide them equally, we change starting VMID to 4 and not use
+	 * VMID 3.
+	 * If the VMID range changes for multi-partition capable GPUs, then
+	 * this code MUST be revisited.
+	 */
+	if (kfd->adev->xcp_mgr) {
+		partition_mode = amdgpu_xcp_query_partition_mode(kfd->adev->xcp_mgr,
+								 AMDGPU_XCP_FL_LOCKED);
+		if (partition_mode == AMDGPU_CPX_PARTITION_MODE &&
+		    kfd->num_nodes != 1) {
+			vmid_num_kfd /= 2;
+			first_vmid_kfd = last_vmid_kfd + 1 - vmid_num_kfd*2;
+		}
+	}
+
+	/* Verify module parameters regarding mapped process number*/
+	if (hws_max_conc_proc >= 0)
+		max_proc_per_quantum = min((u32)hws_max_conc_proc, vmid_num_kfd);
+	else
+		max_proc_per_quantum = vmid_num_kfd;
+
 	/* calculate max size of mqds needed for queues */
 	size = max_num_of_queues_per_device *
-			kfd->device_info->mqd_size_aligned;
+			kfd->device_info.mqd_size_aligned;
 
 	/*
 	 * calculate max size of runlist packet.
 	 * There can be only 2 packets at once
 	 */
-	size += (KFD_MAX_NUM_OF_PROCESSES * sizeof(struct pm4_mes_map_process) +
+	map_process_packet_size = KFD_GC_VERSION(kfd) == IP_VERSION(9, 4, 2) ?
+				sizeof(struct pm4_mes_map_process_aldebaran) :
+				sizeof(struct pm4_mes_map_process);
+	size += (KFD_MAX_NUM_OF_PROCESSES * map_process_packet_size +
 		max_num_of_queues_per_device * sizeof(struct pm4_mes_map_queues)
 		+ sizeof(struct pm4_mes_runlist)) * 2;
 
@@ -249,11 +805,12 @@ bool kgd2kfd_device_init(struct kfd_dev *kfd,
 	/* add another 512KB for all other allocations on gart (HPD, fences) */
 	size += 512 * 1024;
 
-	if (kfd->kfd2kgd->init_gtt_mem_allocation(
-			kfd->kgd, size, &kfd->gtt_mem,
-			&kfd->gtt_start_gpu_addr, &kfd->gtt_start_cpu_ptr)){
+	if (amdgpu_amdkfd_alloc_gtt_mem(
+			kfd->adev, size, &kfd->gtt_mem,
+			&kfd->gtt_start_gpu_addr, &kfd->gtt_start_cpu_ptr,
+			false)) {
 		dev_err(kfd_device, "Could not allocate %d bytes\n", size);
-		goto out;
+		goto alloc_gtt_mem_failure;
 	}
 
 	dev_info(kfd_device, "Allocated %d bytes on gart\n", size);
@@ -270,67 +827,118 @@ bool kgd2kfd_device_init(struct kfd_dev *kfd,
 		goto kfd_doorbell_error;
 	}
 
-	if (kfd_topology_add_device(kfd)) {
-		dev_err(kfd_device, "Error adding device to topology\n");
-		goto kfd_topology_add_device_error;
-	}
+	if (amdgpu_use_xgmi_p2p)
+		kfd->hive_id = kfd->adev->gmc.xgmi.hive_id;
 
-	if (kfd_interrupt_init(kfd)) {
-		dev_err(kfd_device, "Error initializing interrupts\n");
-		goto kfd_interrupt_error;
-	}
+	/*
+	 * For multi-partition capable GPUs, the KFD abstracts all partitions
+	 * within a socket as xGMI connected in the topology so assign a unique
+	 * hive id per device based on the pci device location if device is in
+	 * PCIe mode.
+	 */
+	if (!kfd->hive_id && kfd->num_nodes > 1)
+		kfd->hive_id = pci_dev_id(kfd->adev->pdev);
+
+	kfd->noretry = kfd->adev->gmc.noretry;
+
+	kfd_cwsr_init(kfd);
+
+	dev_info(kfd_device, "Total number of KFD nodes to be created: %d\n",
+				kfd->num_nodes);
+
+	/* Allocate the KFD nodes */
+	for (i = 0, xcp_idx = 0; i < kfd->num_nodes; i++) {
+		node = kzalloc(sizeof(struct kfd_node), GFP_KERNEL);
+		if (!node)
+			goto node_alloc_error;
+
+		node->node_id = i;
+		node->adev = kfd->adev;
+		node->kfd = kfd;
+		node->kfd2kgd = kfd->kfd2kgd;
+		node->vm_info.vmid_num_kfd = vmid_num_kfd;
+		node->xcp = amdgpu_get_next_xcp(kfd->adev->xcp_mgr, &xcp_idx);
+		/* TODO : Check if error handling is needed */
+		if (node->xcp) {
+			amdgpu_xcp_get_inst_details(node->xcp, AMDGPU_XCP_GFX,
+						    &node->xcc_mask);
+			++xcp_idx;
+		} else {
+			node->xcc_mask =
+				(1U << NUM_XCC(kfd->adev->gfx.xcc_mask)) - 1;
+		}
 
-	if (!device_iommu_pasid_init(kfd)) {
-		dev_err(kfd_device,
-			"Error initializing iommuv2 for device %x:%x\n",
-			kfd->pdev->vendor, kfd->pdev->device);
-		goto device_iommu_pasid_error;
-	}
-	amd_iommu_set_invalidate_ctx_cb(kfd->pdev,
-						iommu_pasid_shutdown_callback);
-	amd_iommu_set_invalid_ppr_cb(kfd->pdev, iommu_invalid_ppr_cb);
+		if (node->xcp) {
+			dev_info(kfd_device, "KFD node %d partition %d size %lldM\n",
+				node->node_id, node->xcp->mem_id,
+				KFD_XCP_MEMORY_SIZE(node->adev, node->node_id) >> 20);
+		}
 
-	kfd->dqm = device_queue_manager_init(kfd);
-	if (!kfd->dqm) {
-		dev_err(kfd_device, "Error initializing queue manager\n");
-		goto device_queue_manager_error;
-	}
+		if (partition_mode == AMDGPU_CPX_PARTITION_MODE &&
+		    kfd->num_nodes != 1) {
+			/* For multi-partition capable GPUs and CPX mode, first
+			 * XCD gets VMID range 4-9 and second XCD gets VMID
+			 * range 10-15.
+			 */
+
+			node->vm_info.first_vmid_kfd = (i%2 == 0) ?
+						first_vmid_kfd :
+						first_vmid_kfd+vmid_num_kfd;
+			node->vm_info.last_vmid_kfd = (i%2 == 0) ?
+						last_vmid_kfd-vmid_num_kfd :
+						last_vmid_kfd;
+			node->compute_vmid_bitmap =
+				((0x1 << (node->vm_info.last_vmid_kfd + 1)) - 1) -
+				((0x1 << (node->vm_info.first_vmid_kfd)) - 1);
+		} else {
+			node->vm_info.first_vmid_kfd = first_vmid_kfd;
+			node->vm_info.last_vmid_kfd = last_vmid_kfd;
+			node->compute_vmid_bitmap =
+				gpu_resources->compute_vmid_bitmap;
+		}
+		node->max_proc_per_quantum = max_proc_per_quantum;
+		atomic_set(&node->sram_ecc_flag, 0);
 
-	if (kfd->dqm->ops.start(kfd->dqm)) {
-		dev_err(kfd_device,
-			"Error starting queue manager for device %x:%x\n",
-			kfd->pdev->vendor, kfd->pdev->device);
-		goto dqm_start_error;
+		amdgpu_amdkfd_get_local_mem_info(kfd->adev,
+					&node->local_mem_info, node->xcp);
+
+		if (kfd->adev->xcp_mgr)
+			kfd_setup_interrupt_bitmap(node, i);
+
+		/* Initialize the KFD node */
+		if (kfd_init_node(node)) {
+			dev_err(kfd_device, "Error initializing KFD node\n");
+			goto node_init_error;
+		}
+
+		spin_lock_init(&node->watch_points_lock);
+
+		kfd->nodes[i] = node;
 	}
 
-	kfd->dbgmgr = NULL;
+	svm_range_set_max_pages(kfd->adev);
 
 	kfd->init_complete = true;
-	dev_info(kfd_device, "added device %x:%x\n", kfd->pdev->vendor,
-		 kfd->pdev->device);
+	dev_info(kfd_device, "added device %x:%x\n", kfd->adev->pdev->vendor,
+		 kfd->adev->pdev->device);
 
 	pr_debug("Starting kfd with the following scheduling policy %d\n",
-		sched_policy);
+		node->dqm->sched_policy);
 
 	goto out;
 
-dqm_start_error:
-	device_queue_manager_uninit(kfd->dqm);
-device_queue_manager_error:
-	amd_iommu_free_device(kfd->pdev);
-device_iommu_pasid_error:
-	kfd_interrupt_exit(kfd);
-kfd_interrupt_error:
-	kfd_topology_remove_device(kfd);
-kfd_topology_add_device_error:
+node_init_error:
+node_alloc_error:
+	kfd_cleanup_nodes(kfd, i);
 	kfd_doorbell_fini(kfd);
 kfd_doorbell_error:
 	kfd_gtt_sa_fini(kfd);
 kfd_gtt_sa_init_error:
-	kfd->kfd2kgd->free_gtt_mem(kfd->kgd, kfd->gtt_mem);
+	amdgpu_amdkfd_free_gtt_mem(kfd->adev, &kfd->gtt_mem);
+alloc_gtt_mem_failure:
 	dev_err(kfd_device,
 		"device %x:%x NOT added due to errors\n",
-		kfd->pdev->vendor, kfd->pdev->device);
+		kfd->adev->pdev->vendor, kfd->adev->pdev->device);
 out:
 	return kfd->init_complete;
 }
@@ -338,72 +946,311 @@ out:
 void kgd2kfd_device_exit(struct kfd_dev *kfd)
 {
 	if (kfd->init_complete) {
-		device_queue_manager_uninit(kfd->dqm);
-		amd_iommu_free_device(kfd->pdev);
-		kfd_interrupt_exit(kfd);
-		kfd_topology_remove_device(kfd);
+		/* Cleanup KFD nodes */
+		kfd_cleanup_nodes(kfd, kfd->num_nodes);
+		/* Cleanup common/shared resources */
 		kfd_doorbell_fini(kfd);
+		ida_destroy(&kfd->doorbell_ida);
 		kfd_gtt_sa_fini(kfd);
-		kfd->kfd2kgd->free_gtt_mem(kfd->kgd, kfd->gtt_mem);
+		amdgpu_amdkfd_free_gtt_mem(kfd->adev, &kfd->gtt_mem);
 	}
 
 	kfree(kfd);
 }
 
-void kgd2kfd_suspend(struct kfd_dev *kfd)
+int kgd2kfd_pre_reset(struct kfd_dev *kfd,
+		      struct amdgpu_reset_context *reset_context)
 {
-	if (kfd->init_complete) {
-		kfd->dqm->ops.stop(kfd->dqm);
-		amd_iommu_set_invalidate_ctx_cb(kfd->pdev, NULL);
-		amd_iommu_set_invalid_ppr_cb(kfd->pdev, NULL);
-		amd_iommu_free_device(kfd->pdev);
+	struct kfd_node *node;
+	int i;
+
+	if (!kfd->init_complete)
+		return 0;
+
+	for (i = 0; i < kfd->num_nodes; i++) {
+		node = kfd->nodes[i];
+		kfd_smi_event_update_gpu_reset(node, false, reset_context);
 	}
+
+	kgd2kfd_suspend(kfd, true);
+
+	for (i = 0; i < kfd->num_nodes; i++)
+		kfd_signal_reset_event(kfd->nodes[i]);
+
+	return 0;
 }
 
-int kgd2kfd_resume(struct kfd_dev *kfd)
+/*
+ * Fix me. KFD won't be able to resume existing process for now.
+ * We will keep all existing process in a evicted state and
+ * wait the process to be terminated.
+ */
+
+int kgd2kfd_post_reset(struct kfd_dev *kfd)
 {
-	unsigned int pasid_limit;
-	int err;
+	int ret;
+	struct kfd_node *node;
+	int i;
 
-	pasid_limit = kfd_get_pasid_limit();
+	if (!kfd->init_complete)
+		return 0;
 
-	if (kfd->init_complete) {
-		err = amd_iommu_init_device(kfd->pdev, pasid_limit);
-		if (err < 0) {
-			dev_err(kfd_device, "failed to initialize iommu\n");
-			return -ENXIO;
-		}
+	for (i = 0; i < kfd->num_nodes; i++) {
+		ret = kfd_resume(kfd->nodes[i]);
+		if (ret)
+			return ret;
+	}
+
+	mutex_lock(&kfd_processes_mutex);
+	--kfd_locked;
+	mutex_unlock(&kfd_processes_mutex);
 
-		amd_iommu_set_invalidate_ctx_cb(kfd->pdev,
-						iommu_pasid_shutdown_callback);
-		amd_iommu_set_invalid_ppr_cb(kfd->pdev, iommu_invalid_ppr_cb);
-		kfd->dqm->ops.start(kfd->dqm);
+	for (i = 0; i < kfd->num_nodes; i++) {
+		node = kfd->nodes[i];
+		atomic_set(&node->sram_ecc_flag, 0);
+		kfd_smi_event_update_gpu_reset(node, true, NULL);
 	}
 
 	return 0;
 }
 
+bool kfd_is_locked(struct kfd_dev *kfd)
+{
+	uint8_t id  = 0;
+	struct kfd_node *dev;
+
+	lockdep_assert_held(&kfd_processes_mutex);
+
+	/* check reset/suspend lock */
+	if (kfd_locked > 0)
+		return true;
+
+	if (kfd)
+		return kfd->kfd_dev_lock > 0;
+
+	/* check lock on all cgroup accessible devices */
+	while (kfd_topology_enum_kfd_devices(id++, &dev) == 0) {
+		if (!dev || kfd_devcgroup_check_permission(dev))
+			continue;
+
+		if (dev->kfd->kfd_dev_lock > 0)
+			return true;
+	}
+
+	return false;
+}
+
+void kgd2kfd_suspend(struct kfd_dev *kfd, bool suspend_proc)
+{
+	struct kfd_node *node;
+	int i;
+
+	if (!kfd->init_complete)
+		return;
+
+	if (suspend_proc)
+		kgd2kfd_suspend_process(kfd);
+
+	for (i = 0; i < kfd->num_nodes; i++) {
+		node = kfd->nodes[i];
+		node->dqm->ops.stop(node->dqm);
+	}
+}
+
+int kgd2kfd_resume(struct kfd_dev *kfd, bool resume_proc)
+{
+	int ret, i;
+
+	if (!kfd->init_complete)
+		return 0;
+
+	for (i = 0; i < kfd->num_nodes; i++) {
+		ret = kfd_resume(kfd->nodes[i]);
+		if (ret)
+			return ret;
+	}
+
+	if (resume_proc)
+		ret = kgd2kfd_resume_process(kfd);
+
+	return ret;
+}
+
+void kgd2kfd_suspend_process(struct kfd_dev *kfd)
+{
+	if (!kfd->init_complete)
+		return;
+
+	mutex_lock(&kfd_processes_mutex);
+	/* For first KFD device suspend all the KFD processes */
+	if (++kfd_locked == 1)
+		kfd_suspend_all_processes();
+	mutex_unlock(&kfd_processes_mutex);
+}
+
+int kgd2kfd_resume_process(struct kfd_dev *kfd)
+{
+	int ret = 0;
+
+	if (!kfd->init_complete)
+		return 0;
+
+	mutex_lock(&kfd_processes_mutex);
+	if (--kfd_locked == 0)
+		ret = kfd_resume_all_processes();
+	WARN_ONCE(kfd_locked < 0, "KFD suspend / resume ref. error");
+	mutex_unlock(&kfd_processes_mutex);
+
+	return ret;
+}
+
+static int kfd_resume(struct kfd_node *node)
+{
+	int err = 0;
+
+	err = node->dqm->ops.start(node->dqm);
+	if (err)
+		dev_err(kfd_device,
+			"Error starting queue manager for device %x:%x\n",
+			node->adev->pdev->vendor, node->adev->pdev->device);
+
+	return err;
+}
+
 /* This is called directly from KGD at ISR. */
 void kgd2kfd_interrupt(struct kfd_dev *kfd, const void *ih_ring_entry)
 {
+	uint32_t patched_ihre[KFD_MAX_RING_ENTRY_SIZE], i;
+	bool is_patched = false;
+	unsigned long flags;
+	struct kfd_node *node;
+
 	if (!kfd->init_complete)
 		return;
 
-	spin_lock(&kfd->interrupt_lock);
+	if (kfd->device_info.ih_ring_entry_size > sizeof(patched_ihre)) {
+		dev_err_once(kfd_device, "Ring entry too small\n");
+		return;
+	}
+
+	for (i = 0; i < kfd->num_nodes; i++) {
+		/* Race if another thread in b/w
+		 * kfd_cleanup_nodes and kfree(kfd),
+		 * when kfd->nodes[i] = NULL
+		 */
+		if (kfd->nodes[i])
+			node = kfd->nodes[i];
+		else
+			return;
+
+		spin_lock_irqsave(&node->interrupt_lock, flags);
+
+		if (node->interrupts_active
+		    && interrupt_is_wanted(node, ih_ring_entry,
+			    	patched_ihre, &is_patched)
+		    && enqueue_ih_ring_entry(node,
+			    	is_patched ? patched_ihre : ih_ring_entry)) {
+			queue_work(node->kfd->ih_wq, &node->interrupt_work);
+			spin_unlock_irqrestore(&node->interrupt_lock, flags);
+			return;
+		}
+		spin_unlock_irqrestore(&node->interrupt_lock, flags);
+	}
+
+}
+
+int kgd2kfd_quiesce_mm(struct mm_struct *mm, uint32_t trigger)
+{
+	struct kfd_process *p;
+	int r;
+
+	/* Because we are called from arbitrary context (workqueue) as opposed
+	 * to process context, kfd_process could attempt to exit while we are
+	 * running so the lookup function increments the process ref count.
+	 */
+	p = kfd_lookup_process_by_mm(mm);
+	if (!p)
+		return -ESRCH;
+
+	WARN(debug_evictions, "Evicting pid %d", p->lead_thread->pid);
+	r = kfd_process_evict_queues(p, trigger);
+
+	kfd_unref_process(p);
+	return r;
+}
+
+int kgd2kfd_resume_mm(struct mm_struct *mm)
+{
+	struct kfd_process *p;
+	int r;
+
+	/* Because we are called from arbitrary context (workqueue) as opposed
+	 * to process context, kfd_process could attempt to exit while we are
+	 * running so the lookup function increments the process ref count.
+	 */
+	p = kfd_lookup_process_by_mm(mm);
+	if (!p)
+		return -ESRCH;
+
+	r = kfd_process_restore_queues(p);
+
+	kfd_unref_process(p);
+	return r;
+}
+
+/** kgd2kfd_schedule_evict_and_restore_process - Schedules work queue that will
+ *   prepare for safe eviction of KFD BOs that belong to the specified
+ *   process.
+ *
+ * @mm: mm_struct that identifies the specified KFD process
+ * @fence: eviction fence attached to KFD process BOs
+ *
+ */
+int kgd2kfd_schedule_evict_and_restore_process(struct mm_struct *mm,
+					       struct dma_fence *fence)
+{
+	struct kfd_process *p;
+	unsigned long active_time;
+	unsigned long delay_jiffies = msecs_to_jiffies(PROCESS_ACTIVE_TIME_MS);
+
+	if (!fence)
+		return -EINVAL;
+
+	if (dma_fence_is_signaled(fence))
+		return 0;
+
+	p = kfd_lookup_process_by_mm(mm);
+	if (!p)
+		return -ENODEV;
 
-	if (kfd->interrupts_active
-	    && interrupt_is_wanted(kfd, ih_ring_entry)
-	    && enqueue_ih_ring_entry(kfd, ih_ring_entry))
-		schedule_work(&kfd->interrupt_work);
+	if (fence->seqno == p->last_eviction_seqno)
+		goto out;
+
+	p->last_eviction_seqno = fence->seqno;
 
-	spin_unlock(&kfd->interrupt_lock);
+	/* Avoid KFD process starvation. Wait for at least
+	 * PROCESS_ACTIVE_TIME_MS before evicting the process again
+	 */
+	active_time = get_jiffies_64() - p->last_restore_timestamp;
+	if (delay_jiffies > active_time)
+		delay_jiffies -= active_time;
+	else
+		delay_jiffies = 0;
+
+	/* During process initialization eviction_work.dwork is initialized
+	 * to kfd_evict_bo_worker
+	 */
+	WARN(debug_evictions, "Scheduling eviction of pid %d in %ld jiffies",
+	     p->lead_thread->pid, delay_jiffies);
+	schedule_delayed_work(&p->eviction_work, delay_jiffies);
+out:
+	kfd_unref_process(p);
+	return 0;
 }
 
 static int kfd_gtt_sa_init(struct kfd_dev *kfd, unsigned int buf_size,
 				unsigned int chunk_size)
 {
-	unsigned int num_of_longs;
-
 	if (WARN_ON(buf_size < chunk_size))
 		return -EINVAL;
 	if (WARN_ON(buf_size == 0))
@@ -414,11 +1261,8 @@ static int kfd_gtt_sa_init(struct kfd_dev *kfd, unsigned int buf_size,
 	kfd->gtt_sa_chunk_size = chunk_size;
 	kfd->gtt_sa_num_of_chunks = buf_size / chunk_size;
 
-	num_of_longs = (kfd->gtt_sa_num_of_chunks + BITS_PER_LONG - 1) /
-		BITS_PER_LONG;
-
-	kfd->gtt_sa_bitmap = kcalloc(num_of_longs, sizeof(long), GFP_KERNEL);
-
+	kfd->gtt_sa_bitmap = bitmap_zalloc(kfd->gtt_sa_num_of_chunks,
+					   GFP_KERNEL);
 	if (!kfd->gtt_sa_bitmap)
 		return -ENOMEM;
 
@@ -428,13 +1272,12 @@ static int kfd_gtt_sa_init(struct kfd_dev *kfd, unsigned int buf_size,
 	mutex_init(&kfd->gtt_sa_lock);
 
 	return 0;
-
 }
 
 static void kfd_gtt_sa_fini(struct kfd_dev *kfd)
 {
 	mutex_destroy(&kfd->gtt_sa_lock);
-	kfree(kfd->gtt_sa_bitmap);
+	bitmap_free(kfd->gtt_sa_bitmap);
 }
 
 static inline uint64_t kfd_gtt_sa_calc_gpu_addr(uint64_t start_addr,
@@ -451,10 +1294,11 @@ static inline uint32_t *kfd_gtt_sa_calc_cpu_addr(void *start_addr,
 	return (uint32_t *) ((uint64_t) start_addr + bit_num * chunk_size);
 }
 
-int kfd_gtt_sa_allocate(struct kfd_dev *kfd, unsigned int size,
+int kfd_gtt_sa_allocate(struct kfd_node *node, unsigned int size,
 			struct kfd_mem_obj **mem_obj)
 {
 	unsigned int found, start_search, cur_size;
+	struct kfd_dev *kfd = node->kfd;
 
 	if (size == 0)
 		return -EINVAL;
@@ -462,8 +1306,8 @@ int kfd_gtt_sa_allocate(struct kfd_dev *kfd, unsigned int size,
 	if (size > kfd->gtt_sa_num_of_chunks * kfd->gtt_sa_chunk_size)
 		return -ENOMEM;
 
-	*mem_obj = kmalloc(sizeof(struct kfd_mem_obj), GFP_KERNEL);
-	if ((*mem_obj) == NULL)
+	*mem_obj = kzalloc(sizeof(struct kfd_mem_obj), GFP_KERNEL);
+	if (!(*mem_obj))
 		return -ENOMEM;
 
 	pr_debug("Allocated mem_obj = %p for size = %d\n", *mem_obj, size);
@@ -502,7 +1346,7 @@ kfd_gtt_restart_search:
 	/* If we need only one chunk, mark it as allocated and get out */
 	if (size <= kfd->gtt_sa_chunk_size) {
 		pr_debug("Single bit\n");
-		set_bit(found, kfd->gtt_sa_bitmap);
+		__set_bit(found, kfd->gtt_sa_bitmap);
 		goto kfd_gtt_out;
 	}
 
@@ -540,25 +1384,23 @@ kfd_gtt_restart_search:
 		(*mem_obj)->range_start, (*mem_obj)->range_end);
 
 	/* Mark the chunks as allocated */
-	for (found = (*mem_obj)->range_start;
-		found <= (*mem_obj)->range_end;
-		found++)
-		set_bit(found, kfd->gtt_sa_bitmap);
+	bitmap_set(kfd->gtt_sa_bitmap, (*mem_obj)->range_start,
+		   (*mem_obj)->range_end - (*mem_obj)->range_start + 1);
 
 kfd_gtt_out:
 	mutex_unlock(&kfd->gtt_sa_lock);
 	return 0;
 
 kfd_gtt_no_free_chunk:
-	pr_debug("Allocation failed with mem_obj = %p\n", mem_obj);
+	pr_debug("Allocation failed with mem_obj = %p\n", *mem_obj);
 	mutex_unlock(&kfd->gtt_sa_lock);
-	kfree(mem_obj);
+	kfree(*mem_obj);
 	return -ENOMEM;
 }
 
-int kfd_gtt_sa_free(struct kfd_dev *kfd, struct kfd_mem_obj *mem_obj)
+int kfd_gtt_sa_free(struct kfd_node *node, struct kfd_mem_obj *mem_obj)
 {
-	unsigned int bit;
+	struct kfd_dev *kfd = node->kfd;
 
 	/* Act like kfree when trying to free a NULL object */
 	if (!mem_obj)
@@ -570,13 +1412,318 @@ int kfd_gtt_sa_free(struct kfd_dev *kfd, struct kfd_mem_obj *mem_obj)
 	mutex_lock(&kfd->gtt_sa_lock);
 
 	/* Mark the chunks as free */
-	for (bit = mem_obj->range_start;
-		bit <= mem_obj->range_end;
-		bit++)
-		clear_bit(bit, kfd->gtt_sa_bitmap);
+	bitmap_clear(kfd->gtt_sa_bitmap, mem_obj->range_start,
+		     mem_obj->range_end - mem_obj->range_start + 1);
 
 	mutex_unlock(&kfd->gtt_sa_lock);
 
 	kfree(mem_obj);
 	return 0;
 }
+
+void kgd2kfd_set_sram_ecc_flag(struct kfd_dev *kfd)
+{
+	/*
+	 * TODO: Currently update SRAM ECC flag for first node.
+	 * This needs to be updated later when we can
+	 * identify SRAM ECC error on other nodes also.
+	 */
+	if (kfd)
+		atomic_inc(&kfd->nodes[0]->sram_ecc_flag);
+}
+
+void kfd_inc_compute_active(struct kfd_node *node)
+{
+	if (atomic_inc_return(&node->kfd->compute_profile) == 1)
+		amdgpu_amdkfd_set_compute_idle(node->adev, false);
+}
+
+void kfd_dec_compute_active(struct kfd_node *node)
+{
+	int count = atomic_dec_return(&node->kfd->compute_profile);
+
+	if (count == 0)
+		amdgpu_amdkfd_set_compute_idle(node->adev, true);
+	WARN_ONCE(count < 0, "Compute profile ref. count error");
+}
+
+static bool kfd_compute_active(struct kfd_node *node)
+{
+	if (atomic_read(&node->kfd->compute_profile))
+		return true;
+	return false;
+}
+
+void kgd2kfd_smi_event_throttle(struct kfd_dev *kfd, uint64_t throttle_bitmask)
+{
+	/*
+	 * TODO: For now, raise the throttling event only on first node.
+	 * This will need to change after we are able to determine
+	 * which node raised the throttling event.
+	 */
+	if (kfd && kfd->init_complete)
+		kfd_smi_event_update_thermal_throttling(kfd->nodes[0],
+							throttle_bitmask);
+}
+
+/* kfd_get_num_sdma_engines returns the number of PCIe optimized SDMA and
+ * kfd_get_num_xgmi_sdma_engines returns the number of XGMI SDMA.
+ * When the device has more than two engines, we reserve two for PCIe to enable
+ * full-duplex and the rest are used as XGMI.
+ */
+unsigned int kfd_get_num_sdma_engines(struct kfd_node *node)
+{
+	/* If XGMI is not supported, all SDMA engines are PCIe */
+	if (!node->adev->gmc.xgmi.supported)
+		return node->adev->sdma.num_instances/(int)node->kfd->num_nodes;
+
+	return min(node->adev->sdma.num_instances/(int)node->kfd->num_nodes, 2);
+}
+
+unsigned int kfd_get_num_xgmi_sdma_engines(struct kfd_node *node)
+{
+	/* After reserved for PCIe, the rest of engines are XGMI */
+	return node->adev->sdma.num_instances/(int)node->kfd->num_nodes -
+		kfd_get_num_sdma_engines(node);
+}
+
+int kgd2kfd_check_and_lock_kfd(struct kfd_dev *kfd)
+{
+	struct kfd_process *p;
+	int r = 0, temp, idx;
+
+	mutex_lock(&kfd_processes_mutex);
+
+	/* kfd_processes_count is per kfd_dev, return -EBUSY without
+	 * further check
+	 */
+	if (!!atomic_read(&kfd->kfd_processes_count)) {
+		pr_debug("process_wq_release not finished\n");
+		r = -EBUSY;
+		goto out;
+	}
+
+	if (hash_empty(kfd_processes_table) && !kfd_is_locked(kfd))
+		goto out;
+
+	/* fail under system reset/resume or kfd device is partition switching. */
+	if (kfd_is_locked(kfd)) {
+		r = -EBUSY;
+		goto out;
+	}
+
+	/*
+	 * ensure all running processes are cgroup excluded from device before mode switch.
+	 * i.e. no pdd was created on the process socket.
+	 */
+	idx = srcu_read_lock(&kfd_processes_srcu);
+	hash_for_each_rcu(kfd_processes_table, temp, p, kfd_processes) {
+		int i;
+
+		for (i = 0; i < p->n_pdds; i++) {
+			if (p->pdds[i]->dev->kfd != kfd)
+				continue;
+
+			r = -EBUSY;
+			goto proc_check_unlock;
+		}
+	}
+
+proc_check_unlock:
+	srcu_read_unlock(&kfd_processes_srcu, idx);
+out:
+	if (!r)
+		++kfd->kfd_dev_lock;
+	mutex_unlock(&kfd_processes_mutex);
+
+	return r;
+}
+
+void kgd2kfd_unlock_kfd(struct kfd_dev *kfd)
+{
+	mutex_lock(&kfd_processes_mutex);
+	--kfd->kfd_dev_lock;
+	mutex_unlock(&kfd_processes_mutex);
+}
+
+int kgd2kfd_start_sched(struct kfd_dev *kfd, uint32_t node_id)
+{
+	struct kfd_node *node;
+	int ret;
+
+	if (!kfd->init_complete)
+		return 0;
+
+	if (node_id >= kfd->num_nodes) {
+		dev_warn(kfd->adev->dev, "Invalid node ID: %u exceeds %u\n",
+			 node_id, kfd->num_nodes - 1);
+		return -EINVAL;
+	}
+	node = kfd->nodes[node_id];
+
+	ret = node->dqm->ops.unhalt(node->dqm);
+	if (ret)
+		dev_err(kfd_device, "Error in starting scheduler\n");
+
+	return ret;
+}
+
+int kgd2kfd_start_sched_all_nodes(struct kfd_dev *kfd)
+{
+	struct kfd_node *node;
+	int i, r;
+
+	if (!kfd->init_complete)
+		return 0;
+
+	for (i = 0; i < kfd->num_nodes; i++) {
+		node = kfd->nodes[i];
+		r = node->dqm->ops.unhalt(node->dqm);
+		if (r) {
+			dev_err(kfd_device, "Error in starting scheduler\n");
+			return r;
+		}
+	}
+	return 0;
+}
+
+int kgd2kfd_stop_sched(struct kfd_dev *kfd, uint32_t node_id)
+{
+	struct kfd_node *node;
+
+	if (!kfd->init_complete)
+		return 0;
+
+	if (node_id >= kfd->num_nodes) {
+		dev_warn(kfd->adev->dev, "Invalid node ID: %u exceeds %u\n",
+			 node_id, kfd->num_nodes - 1);
+		return -EINVAL;
+	}
+
+	node = kfd->nodes[node_id];
+	return node->dqm->ops.halt(node->dqm);
+}
+
+int kgd2kfd_stop_sched_all_nodes(struct kfd_dev *kfd)
+{
+	struct kfd_node *node;
+	int i, r;
+
+	if (!kfd->init_complete)
+		return 0;
+
+	for (i = 0; i < kfd->num_nodes; i++) {
+		node = kfd->nodes[i];
+		r = node->dqm->ops.halt(node->dqm);
+		if (r)
+			return r;
+	}
+	return 0;
+}
+
+bool kgd2kfd_compute_active(struct kfd_dev *kfd, uint32_t node_id)
+{
+	struct kfd_node *node;
+
+	if (!kfd->init_complete)
+		return false;
+
+	if (node_id >= kfd->num_nodes) {
+		dev_warn(kfd->adev->dev, "Invalid node ID: %u exceeds %u\n",
+			 node_id, kfd->num_nodes - 1);
+		return false;
+	}
+
+	node = kfd->nodes[node_id];
+
+	return kfd_compute_active(node);
+}
+
+/**
+ * kgd2kfd_vmfault_fast_path() - KFD vm page fault interrupt handling fast path for gmc v9
+ * @adev: amdgpu device
+ * @entry: vm fault interrupt vector
+ * @retry_fault: if this is retry fault
+ *
+ * retry fault -
+ *    with CAM enabled, adev primary ring
+ *                           |  gmc_v9_0_process_interrupt()
+ *                      adev soft_ring
+ *                           |  gmc_v9_0_process_interrupt() worker failed to recover page fault
+ *                      KFD node ih_fifo
+ *                           |  KFD interrupt_wq worker
+ *                      kfd_signal_vm_fault_event
+ *
+ *    without CAM,      adev primary ring1
+ *                           |  gmc_v9_0_process_interrupt worker failed to recvoer page fault
+ *                      KFD node ih_fifo
+ *                           |  KFD interrupt_wq worker
+ *                      kfd_signal_vm_fault_event
+ *
+ * no-retry fault -
+ *                      adev primary ring
+ *                           |  gmc_v9_0_process_interrupt()
+ *                      KFD node ih_fifo
+ *                           |  KFD interrupt_wq worker
+ *                      kfd_signal_vm_fault_event
+ *
+ * fast path - After kfd_signal_vm_fault_event, gmc_v9_0_process_interrupt drop the page fault
+ *            of same process, don't copy interrupt to KFD node ih_fifo.
+ *            With gdb debugger enabled, need convert the retry fault to no-retry fault for
+ *            debugger, cannot use the fast path.
+ *
+ * Return:
+ *   true - use the fast path to handle this fault
+ *   false - use normal path to handle it
+ */
+bool kgd2kfd_vmfault_fast_path(struct amdgpu_device *adev, struct amdgpu_iv_entry *entry,
+			       bool retry_fault)
+{
+	struct kfd_process *p;
+	u32 cam_index;
+
+	if (entry->ih == &adev->irq.ih_soft || entry->ih == &adev->irq.ih1) {
+		p = kfd_lookup_process_by_pasid(entry->pasid, NULL);
+		if (!p)
+			return true;
+
+		if (p->gpu_page_fault && !p->debug_trap_enabled) {
+			if (retry_fault && adev->irq.retry_cam_enabled) {
+				cam_index = entry->src_data[2] & 0x3ff;
+				WDOORBELL32(adev->irq.retry_cam_doorbell_index, cam_index);
+			}
+
+			kfd_unref_process(p);
+			return true;
+		}
+
+		/*
+		 * This is the first page fault, set flag and then signal user space
+		 */
+		p->gpu_page_fault = true;
+		kfd_unref_process(p);
+	}
+	return false;
+}
+
+#if defined(CONFIG_DEBUG_FS)
+
+/* This function will send a package to HIQ to hang the HWS
+ * which will trigger a GPU reset and bring the HWS back to normal state
+ */
+int kfd_debugfs_hang_hws(struct kfd_node *dev)
+{
+	if (dev->dqm->sched_policy != KFD_SCHED_POLICY_HWS) {
+		pr_err("HWS is not enabled");
+		return -EINVAL;
+	}
+
+	if (dev->kfd->shared_resources.enable_mes) {
+		dev_err(dev->adev->dev, "Inducing MES hang is not supported\n");
+		return -EINVAL;
+	}
+
+	return dqm_debugfs_hang_hws(dev->dqm);
+}
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
index 53a66e821624..6c5c7c1bf5ed 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
@@ -1,5 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -21,10 +22,11 @@
  *
  */
 
+#include <linux/ratelimit.h>
+#include <linux/printk.h>
 #include <linux/slab.h>
 #include <linux/list.h>
 #include <linux/types.h>
-#include <linux/printk.h>
 #include <linux/bitops.h>
 #include <linux/sched.h>
 #include "kfd_priv.h"
@@ -32,33 +34,48 @@
 #include "kfd_mqd_manager.h"
 #include "cik_regs.h"
 #include "kfd_kernel_queue.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_reset.h"
+#include "amdgpu_sdma.h"
+#include "mes_v11_api_def.h"
+#include "kfd_debug.h"
 
 /* Size of the per-pipe EOP queue */
 #define CIK_HPD_EOP_BYTES_LOG2 11
 #define CIK_HPD_EOP_BYTES (1U << CIK_HPD_EOP_BYTES_LOG2)
+/* See unmap_queues_cpsch() */
+#define USE_DEFAULT_GRACE_PERIOD 0xffffffff
 
 static int set_pasid_vmid_mapping(struct device_queue_manager *dqm,
-					unsigned int pasid, unsigned int vmid);
-
-static int create_compute_queue_nocpsch(struct device_queue_manager *dqm,
-					struct queue *q,
-					struct qcm_process_device *qpd);
+				  u32 pasid, unsigned int vmid);
 
-static int execute_queues_cpsch(struct device_queue_manager *dqm, bool lock);
-static int destroy_queues_cpsch(struct device_queue_manager *dqm,
-				bool preempt_static_queues, bool lock);
+static int execute_queues_cpsch(struct device_queue_manager *dqm,
+				enum kfd_unmap_queues_filter filter,
+				uint32_t filter_param,
+				uint32_t grace_period);
+static int unmap_queues_cpsch(struct device_queue_manager *dqm,
+				enum kfd_unmap_queues_filter filter,
+				uint32_t filter_param,
+				uint32_t grace_period,
+				bool reset);
 
-static int create_sdma_queue_nocpsch(struct device_queue_manager *dqm,
-					struct queue *q,
-					struct qcm_process_device *qpd);
+static int map_queues_cpsch(struct device_queue_manager *dqm);
 
 static void deallocate_sdma_queue(struct device_queue_manager *dqm,
-				unsigned int sdma_queue_id);
+				struct queue *q);
+
+static inline void deallocate_hqd(struct device_queue_manager *dqm,
+				struct queue *q);
+static int allocate_hqd(struct device_queue_manager *dqm, struct queue *q);
+static int allocate_sdma_queue(struct device_queue_manager *dqm,
+				struct queue *q, const uint32_t *restore_sdma_id);
+
+static int reset_queues_on_hws_hang(struct device_queue_manager *dqm, bool is_sdma);
 
 static inline
 enum KFD_MQD_TYPE get_mqd_type_from_queue_type(enum kfd_queue_type type)
 {
-	if (type == KFD_QUEUE_TYPE_SDMA)
+	if (type == KFD_QUEUE_TYPE_SDMA || type == KFD_QUEUE_TYPE_SDMA_XGMI)
 		return KFD_MQD_TYPE_SDMA;
 	return KFD_MQD_TYPE_CP;
 }
@@ -66,78 +83,536 @@ enum KFD_MQD_TYPE get_mqd_type_from_queue_type(enum kfd_queue_type type)
 static bool is_pipe_enabled(struct device_queue_manager *dqm, int mec, int pipe)
 {
 	int i;
-	int pipe_offset = mec * dqm->dev->shared_resources.num_pipe_per_mec
-		+ pipe * dqm->dev->shared_resources.num_queue_per_pipe;
+	int pipe_offset = (mec * dqm->dev->kfd->shared_resources.num_pipe_per_mec
+		+ pipe) * dqm->dev->kfd->shared_resources.num_queue_per_pipe;
 
 	/* queue is available for KFD usage if bit is 1 */
-	for (i = 0; i <  dqm->dev->shared_resources.num_queue_per_pipe; ++i)
+	for (i = 0; i <  dqm->dev->kfd->shared_resources.num_queue_per_pipe; ++i)
 		if (test_bit(pipe_offset + i,
-			      dqm->dev->shared_resources.queue_bitmap))
+			      dqm->dev->kfd->shared_resources.cp_queue_bitmap))
 			return true;
 	return false;
 }
 
-unsigned int get_queues_num(struct device_queue_manager *dqm)
+unsigned int get_cp_queues_num(struct device_queue_manager *dqm)
 {
-	return bitmap_weight(dqm->dev->shared_resources.queue_bitmap,
-				KGD_MAX_QUEUES);
+	return bitmap_weight(dqm->dev->kfd->shared_resources.cp_queue_bitmap,
+				AMDGPU_MAX_QUEUES);
 }
 
 unsigned int get_queues_per_pipe(struct device_queue_manager *dqm)
 {
-	return dqm->dev->shared_resources.num_queue_per_pipe;
+	return dqm->dev->kfd->shared_resources.num_queue_per_pipe;
 }
 
 unsigned int get_pipes_per_mec(struct device_queue_manager *dqm)
 {
-	return dqm->dev->shared_resources.num_pipe_per_mec;
+	return dqm->dev->kfd->shared_resources.num_pipe_per_mec;
+}
+
+static unsigned int get_num_all_sdma_engines(struct device_queue_manager *dqm)
+{
+	return kfd_get_num_sdma_engines(dqm->dev) +
+		kfd_get_num_xgmi_sdma_engines(dqm->dev);
+}
+
+unsigned int get_num_sdma_queues(struct device_queue_manager *dqm)
+{
+	return kfd_get_num_sdma_engines(dqm->dev) *
+		dqm->dev->kfd->device_info.num_sdma_queues_per_engine;
+}
+
+unsigned int get_num_xgmi_sdma_queues(struct device_queue_manager *dqm)
+{
+	return kfd_get_num_xgmi_sdma_engines(dqm->dev) *
+		dqm->dev->kfd->device_info.num_sdma_queues_per_engine;
+}
+
+static void init_sdma_bitmaps(struct device_queue_manager *dqm)
+{
+	bitmap_zero(dqm->sdma_bitmap, KFD_MAX_SDMA_QUEUES);
+	bitmap_set(dqm->sdma_bitmap, 0, get_num_sdma_queues(dqm));
+
+	bitmap_zero(dqm->xgmi_sdma_bitmap, KFD_MAX_SDMA_QUEUES);
+	bitmap_set(dqm->xgmi_sdma_bitmap, 0, get_num_xgmi_sdma_queues(dqm));
+
+	/* Mask out the reserved queues */
+	bitmap_andnot(dqm->sdma_bitmap, dqm->sdma_bitmap,
+		      dqm->dev->kfd->device_info.reserved_sdma_queues_bitmap,
+		      KFD_MAX_SDMA_QUEUES);
 }
 
 void program_sh_mem_settings(struct device_queue_manager *dqm,
 					struct qcm_process_device *qpd)
 {
-	return dqm->dev->kfd2kgd->program_sh_mem_settings(
-						dqm->dev->kgd, qpd->vmid,
-						qpd->sh_mem_config,
-						qpd->sh_mem_ape1_base,
-						qpd->sh_mem_ape1_limit,
-						qpd->sh_mem_bases);
+	uint32_t xcc_mask = dqm->dev->xcc_mask;
+	int xcc_id;
+
+	for_each_inst(xcc_id, xcc_mask)
+		dqm->dev->kfd2kgd->program_sh_mem_settings(
+			dqm->dev->adev, qpd->vmid, qpd->sh_mem_config,
+			qpd->sh_mem_ape1_base, qpd->sh_mem_ape1_limit,
+			qpd->sh_mem_bases, xcc_id);
+}
+
+static void kfd_hws_hang(struct device_queue_manager *dqm)
+{
+	struct device_process_node *cur;
+	struct qcm_process_device *qpd;
+	struct queue *q;
+
+	/* Mark all device queues as reset. */
+	list_for_each_entry(cur, &dqm->queues, list) {
+		qpd = cur->qpd;
+		list_for_each_entry(q, &qpd->queues_list, list) {
+			struct kfd_process_device *pdd = qpd_to_pdd(qpd);
+
+			pdd->has_reset_queue = true;
+		}
+	}
+
+	/*
+	 * Issue a GPU reset if HWS is unresponsive
+	 */
+	amdgpu_amdkfd_gpu_reset(dqm->dev->adev);
+}
+
+static int convert_to_mes_queue_type(int queue_type)
+{
+	int mes_queue_type;
+
+	switch (queue_type) {
+	case KFD_QUEUE_TYPE_COMPUTE:
+		mes_queue_type = MES_QUEUE_TYPE_COMPUTE;
+		break;
+	case KFD_QUEUE_TYPE_SDMA:
+		mes_queue_type = MES_QUEUE_TYPE_SDMA;
+		break;
+	default:
+		WARN(1, "Invalid queue type %d", queue_type);
+		mes_queue_type = -EINVAL;
+		break;
+	}
+
+	return mes_queue_type;
+}
+
+static int add_queue_mes(struct device_queue_manager *dqm, struct queue *q,
+			 struct qcm_process_device *qpd)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)dqm->dev->adev;
+	struct kfd_process_device *pdd = qpd_to_pdd(qpd);
+	struct mes_add_queue_input queue_input;
+	int r, queue_type;
+	uint64_t wptr_addr_off;
+
+	if (!dqm->sched_running || dqm->sched_halt)
+		return 0;
+	if (!down_read_trylock(&adev->reset_domain->sem))
+		return -EIO;
+
+	memset(&queue_input, 0x0, sizeof(struct mes_add_queue_input));
+	queue_input.process_id = pdd->pasid;
+	queue_input.page_table_base_addr =  qpd->page_table_base;
+	queue_input.process_va_start = 0;
+	queue_input.process_va_end = adev->vm_manager.max_pfn - 1;
+	/* MES unit for quantum is 100ns */
+	queue_input.process_quantum = KFD_MES_PROCESS_QUANTUM;  /* Equivalent to 10ms. */
+	queue_input.process_context_addr = pdd->proc_ctx_gpu_addr;
+	queue_input.gang_quantum = KFD_MES_GANG_QUANTUM; /* Equivalent to 1ms */
+	queue_input.gang_context_addr = q->gang_ctx_gpu_addr;
+	queue_input.inprocess_gang_priority = q->properties.priority;
+	queue_input.gang_global_priority_level =
+					AMDGPU_MES_PRIORITY_LEVEL_NORMAL;
+	queue_input.doorbell_offset = q->properties.doorbell_off;
+	queue_input.mqd_addr = q->gart_mqd_addr;
+	queue_input.wptr_addr = (uint64_t)q->properties.write_ptr;
+
+	wptr_addr_off = (uint64_t)q->properties.write_ptr & (PAGE_SIZE - 1);
+	queue_input.wptr_mc_addr = amdgpu_bo_gpu_offset(q->properties.wptr_bo) + wptr_addr_off;
+
+	queue_input.is_kfd_process = 1;
+	queue_input.is_aql_queue = (q->properties.format == KFD_QUEUE_FORMAT_AQL);
+	queue_input.queue_size = q->properties.queue_size >> 2;
+
+	queue_input.paging = false;
+	queue_input.tba_addr = qpd->tba_addr;
+	queue_input.tma_addr = qpd->tma_addr;
+	queue_input.trap_en = !kfd_dbg_has_cwsr_workaround(q->device);
+	queue_input.skip_process_ctx_clear =
+		qpd->pqm->process->runtime_info.runtime_state == DEBUG_RUNTIME_STATE_ENABLED &&
+						(qpd->pqm->process->debug_trap_enabled ||
+						 kfd_dbg_has_ttmps_always_setup(q->device));
+
+	queue_type = convert_to_mes_queue_type(q->properties.type);
+	if (queue_type < 0) {
+		dev_err(adev->dev, "Queue type not supported with MES, queue:%d\n",
+			q->properties.type);
+		up_read(&adev->reset_domain->sem);
+		return -EINVAL;
+	}
+	queue_input.queue_type = (uint32_t)queue_type;
+
+	queue_input.exclusively_scheduled = q->properties.is_gws;
+
+	amdgpu_mes_lock(&adev->mes);
+	r = adev->mes.funcs->add_hw_queue(&adev->mes, &queue_input);
+	amdgpu_mes_unlock(&adev->mes);
+	up_read(&adev->reset_domain->sem);
+	if (r) {
+		dev_err(adev->dev, "failed to add hardware queue to MES, doorbell=0x%x\n",
+			q->properties.doorbell_off);
+		dev_err(adev->dev, "MES might be in unrecoverable state, issue a GPU reset\n");
+		kfd_hws_hang(dqm);
+	}
+
+	return r;
+}
+
+static int remove_queue_mes(struct device_queue_manager *dqm, struct queue *q,
+			struct qcm_process_device *qpd)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)dqm->dev->adev;
+	int r;
+	struct mes_remove_queue_input queue_input;
+
+	if (!dqm->sched_running || dqm->sched_halt)
+		return 0;
+	if (!down_read_trylock(&adev->reset_domain->sem))
+		return -EIO;
+
+	memset(&queue_input, 0x0, sizeof(struct mes_remove_queue_input));
+	queue_input.doorbell_offset = q->properties.doorbell_off;
+	queue_input.gang_context_addr = q->gang_ctx_gpu_addr;
+
+	amdgpu_mes_lock(&adev->mes);
+	r = adev->mes.funcs->remove_hw_queue(&adev->mes, &queue_input);
+	amdgpu_mes_unlock(&adev->mes);
+	up_read(&adev->reset_domain->sem);
+
+	if (r) {
+		dev_err(adev->dev, "failed to remove hardware queue from MES, doorbell=0x%x\n",
+			q->properties.doorbell_off);
+		dev_err(adev->dev, "MES might be in unrecoverable state, issue a GPU reset\n");
+		kfd_hws_hang(dqm);
+	}
+
+	return r;
+}
+
+static int remove_all_kfd_queues_mes(struct device_queue_manager *dqm)
+{
+	struct device_process_node *cur;
+	struct device *dev = dqm->dev->adev->dev;
+	struct qcm_process_device *qpd;
+	struct queue *q;
+	int retval = 0;
+
+	list_for_each_entry(cur, &dqm->queues, list) {
+		qpd = cur->qpd;
+		list_for_each_entry(q, &qpd->queues_list, list) {
+			if (q->properties.is_active) {
+				retval = remove_queue_mes(dqm, q, qpd);
+				if (retval) {
+					dev_err(dev, "%s: Failed to remove queue %d for dev %d",
+						__func__,
+						q->properties.queue_id,
+						dqm->dev->id);
+					return retval;
+				}
+			}
+		}
+	}
+
+	return retval;
+}
+
+static int add_all_kfd_queues_mes(struct device_queue_manager *dqm)
+{
+	struct device_process_node *cur;
+	struct device *dev = dqm->dev->adev->dev;
+	struct qcm_process_device *qpd;
+	struct queue *q;
+	int retval = 0;
+
+	list_for_each_entry(cur, &dqm->queues, list) {
+		qpd = cur->qpd;
+		list_for_each_entry(q, &qpd->queues_list, list) {
+			if (!q->properties.is_active)
+				continue;
+			retval = add_queue_mes(dqm, q, qpd);
+			if (retval) {
+				dev_err(dev, "%s: Failed to add queue %d for dev %d",
+					__func__,
+					q->properties.queue_id,
+					dqm->dev->id);
+				return retval;
+			}
+		}
+	}
+
+	return retval;
+}
+
+static int suspend_all_queues_mes(struct device_queue_manager *dqm)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)dqm->dev->adev;
+	int r = 0;
+
+	if (!down_read_trylock(&adev->reset_domain->sem))
+		return -EIO;
+
+	r = amdgpu_mes_suspend(adev);
+	up_read(&adev->reset_domain->sem);
+
+	if (r) {
+		dev_err(adev->dev, "failed to suspend gangs from MES\n");
+		dev_err(adev->dev, "MES might be in unrecoverable state, issue a GPU reset\n");
+		kfd_hws_hang(dqm);
+	}
+
+	return r;
+}
+
+static int resume_all_queues_mes(struct device_queue_manager *dqm)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)dqm->dev->adev;
+	int r = 0;
+
+	if (!down_read_trylock(&adev->reset_domain->sem))
+		return -EIO;
+
+	r = amdgpu_mes_resume(adev);
+	up_read(&adev->reset_domain->sem);
+
+	if (r) {
+		dev_err(adev->dev, "failed to resume gangs from MES\n");
+		dev_err(adev->dev, "MES might be in unrecoverable state, issue a GPU reset\n");
+		kfd_hws_hang(dqm);
+	}
+
+	return r;
+}
+
+static void increment_queue_count(struct device_queue_manager *dqm,
+				  struct qcm_process_device *qpd,
+				  struct queue *q)
+{
+	dqm->active_queue_count++;
+	if (q->properties.type == KFD_QUEUE_TYPE_COMPUTE ||
+	    q->properties.type == KFD_QUEUE_TYPE_DIQ)
+		dqm->active_cp_queue_count++;
+
+	if (q->properties.is_gws) {
+		dqm->gws_queue_count++;
+		qpd->mapped_gws_queue = true;
+	}
+}
+
+static void decrement_queue_count(struct device_queue_manager *dqm,
+				  struct qcm_process_device *qpd,
+				  struct queue *q)
+{
+	dqm->active_queue_count--;
+	if (q->properties.type == KFD_QUEUE_TYPE_COMPUTE ||
+	    q->properties.type == KFD_QUEUE_TYPE_DIQ)
+		dqm->active_cp_queue_count--;
+
+	if (q->properties.is_gws) {
+		dqm->gws_queue_count--;
+		qpd->mapped_gws_queue = false;
+	}
+}
+
+/*
+ * Allocate a doorbell ID to this queue.
+ * If doorbell_id is passed in, make sure requested ID is valid then allocate it.
+ */
+static int allocate_doorbell(struct qcm_process_device *qpd,
+			     struct queue *q,
+			     uint32_t const *restore_id)
+{
+	struct kfd_node *dev = qpd->dqm->dev;
+
+	if (!KFD_IS_SOC15(dev)) {
+		/* On pre-SOC15 chips we need to use the queue ID to
+		 * preserve the user mode ABI.
+		 */
+
+		if (restore_id && *restore_id != q->properties.queue_id)
+			return -EINVAL;
+
+		q->doorbell_id = q->properties.queue_id;
+	} else if (q->properties.type == KFD_QUEUE_TYPE_SDMA ||
+			q->properties.type == KFD_QUEUE_TYPE_SDMA_XGMI) {
+		/* For SDMA queues on SOC15 with 8-byte doorbell, use static
+		 * doorbell assignments based on the engine and queue id.
+		 * The doobell index distance between RLC (2*i) and (2*i+1)
+		 * for a SDMA engine is 512.
+		 */
+
+		uint32_t *idx_offset = dev->kfd->shared_resources.sdma_doorbell_idx;
+
+		/*
+		 * q->properties.sdma_engine_id corresponds to the virtual
+		 * sdma engine number. However, for doorbell allocation,
+		 * we need the physical sdma engine id in order to get the
+		 * correct doorbell offset.
+		 */
+		uint32_t valid_id = idx_offset[qpd->dqm->dev->node_id *
+					       get_num_all_sdma_engines(qpd->dqm) +
+					       q->properties.sdma_engine_id]
+						+ (q->properties.sdma_queue_id & 1)
+						* KFD_QUEUE_DOORBELL_MIRROR_OFFSET
+						+ (q->properties.sdma_queue_id >> 1);
+
+		if (restore_id && *restore_id != valid_id)
+			return -EINVAL;
+		q->doorbell_id = valid_id;
+	} else {
+		/* For CP queues on SOC15 */
+		if (restore_id) {
+			/* make sure that ID is free  */
+			if (__test_and_set_bit(*restore_id, qpd->doorbell_bitmap))
+				return -EINVAL;
+
+			q->doorbell_id = *restore_id;
+		} else {
+			/* or reserve a free doorbell ID */
+			unsigned int found;
+
+			found = find_first_zero_bit(qpd->doorbell_bitmap,
+						    KFD_MAX_NUM_OF_QUEUES_PER_PROCESS);
+			if (found >= KFD_MAX_NUM_OF_QUEUES_PER_PROCESS) {
+				pr_debug("No doorbells available");
+				return -EBUSY;
+			}
+			set_bit(found, qpd->doorbell_bitmap);
+			q->doorbell_id = found;
+		}
+	}
+
+	q->properties.doorbell_off = amdgpu_doorbell_index_on_bar(dev->adev,
+								  qpd->proc_doorbells,
+								  q->doorbell_id,
+								  dev->kfd->device_info.doorbell_size);
+	return 0;
+}
+
+static void deallocate_doorbell(struct qcm_process_device *qpd,
+				struct queue *q)
+{
+	unsigned int old;
+	struct kfd_node *dev = qpd->dqm->dev;
+
+	if (!KFD_IS_SOC15(dev) ||
+	    q->properties.type == KFD_QUEUE_TYPE_SDMA ||
+	    q->properties.type == KFD_QUEUE_TYPE_SDMA_XGMI)
+		return;
+
+	old = test_and_clear_bit(q->doorbell_id, qpd->doorbell_bitmap);
+	WARN_ON(!old);
+}
+
+static void program_trap_handler_settings(struct device_queue_manager *dqm,
+				struct qcm_process_device *qpd)
+{
+	uint32_t xcc_mask = dqm->dev->xcc_mask;
+	int xcc_id;
+
+	if (dqm->dev->kfd2kgd->program_trap_handler_settings)
+		for_each_inst(xcc_id, xcc_mask)
+			dqm->dev->kfd2kgd->program_trap_handler_settings(
+				dqm->dev->adev, qpd->vmid, qpd->tba_addr,
+				qpd->tma_addr, xcc_id);
 }
 
 static int allocate_vmid(struct device_queue_manager *dqm,
 			struct qcm_process_device *qpd,
 			struct queue *q)
 {
-	int bit, allocated_vmid;
+	struct kfd_process_device *pdd = qpd_to_pdd(qpd);
+	struct device *dev = dqm->dev->adev->dev;
+	int allocated_vmid = -1, i;
+
+	for (i = dqm->dev->vm_info.first_vmid_kfd;
+			i <= dqm->dev->vm_info.last_vmid_kfd; i++) {
+		if (!dqm->vmid_pasid[i]) {
+			allocated_vmid = i;
+			break;
+		}
+	}
 
-	if (dqm->vmid_bitmap == 0)
-		return -ENOMEM;
+	if (allocated_vmid < 0) {
+		dev_err(dev, "no more vmid to allocate\n");
+		return -ENOSPC;
+	}
+
+	pr_debug("vmid allocated: %d\n", allocated_vmid);
 
-	bit = find_first_bit((unsigned long *)&dqm->vmid_bitmap, CIK_VMID_NUM);
-	clear_bit(bit, (unsigned long *)&dqm->vmid_bitmap);
+	dqm->vmid_pasid[allocated_vmid] = pdd->pasid;
+
+	set_pasid_vmid_mapping(dqm, pdd->pasid, allocated_vmid);
 
-	/* Kaveri kfd vmid's starts from vmid 8 */
-	allocated_vmid = bit + KFD_VMID_START_OFFSET;
-	pr_debug("vmid allocation %d\n", allocated_vmid);
 	qpd->vmid = allocated_vmid;
 	q->properties.vmid = allocated_vmid;
 
-	set_pasid_vmid_mapping(dqm, q->process->pasid, q->properties.vmid);
 	program_sh_mem_settings(dqm, qpd);
 
+	if (KFD_IS_SOC15(dqm->dev) && dqm->dev->kfd->cwsr_enabled)
+		program_trap_handler_settings(dqm, qpd);
+
+	/* qpd->page_table_base is set earlier when register_process()
+	 * is called, i.e. when the first queue is created.
+	 */
+	dqm->dev->kfd2kgd->set_vm_context_page_table_base(dqm->dev->adev,
+			qpd->vmid,
+			qpd->page_table_base);
+	/* invalidate the VM context after pasid and vmid mapping is set up */
+	kfd_flush_tlb(qpd_to_pdd(qpd), TLB_FLUSH_LEGACY);
+
+	if (dqm->dev->kfd2kgd->set_scratch_backing_va)
+		dqm->dev->kfd2kgd->set_scratch_backing_va(dqm->dev->adev,
+				qpd->sh_hidden_private_base, qpd->vmid);
+
 	return 0;
 }
 
+static int flush_texture_cache_nocpsch(struct kfd_node *kdev,
+				struct qcm_process_device *qpd)
+{
+	const struct packet_manager_funcs *pmf = qpd->dqm->packet_mgr.pmf;
+	int ret;
+
+	if (!qpd->ib_kaddr)
+		return -ENOMEM;
+
+	ret = pmf->release_mem(qpd->ib_base, (uint32_t *)qpd->ib_kaddr);
+	if (ret)
+		return ret;
+
+	return amdgpu_amdkfd_submit_ib(kdev->adev, KGD_ENGINE_MEC1, qpd->vmid,
+				qpd->ib_base, (uint32_t *)qpd->ib_kaddr,
+				pmf->release_mem_size / sizeof(uint32_t));
+}
+
 static void deallocate_vmid(struct device_queue_manager *dqm,
 				struct qcm_process_device *qpd,
 				struct queue *q)
 {
-	int bit = qpd->vmid - KFD_VMID_START_OFFSET;
+	struct device *dev = dqm->dev->adev->dev;
+
+	/* On GFX v7, CP doesn't flush TC at dequeue */
+	if (q->device->adev->asic_type == CHIP_HAWAII)
+		if (flush_texture_cache_nocpsch(q->device, qpd))
+			dev_err(dev, "Failed to flush TC\n");
+
+	kfd_flush_tlb(qpd_to_pdd(qpd), TLB_FLUSH_LEGACY);
 
 	/* Release the vmid mapping */
 	set_pasid_vmid_mapping(dqm, 0, qpd->vmid);
+	dqm->vmid_pasid[qpd->vmid] = 0;
 
-	set_bit(bit, (unsigned long *)&dqm->vmid_bitmap);
 	qpd->vmid = 0;
 	q->properties.vmid = 0;
 }
@@ -145,13 +620,13 @@ static void deallocate_vmid(struct device_queue_manager *dqm,
 static int create_queue_nocpsch(struct device_queue_manager *dqm,
 				struct queue *q,
 				struct qcm_process_device *qpd,
-				int *allocated_vmid)
+				const struct kfd_criu_queue_priv_data *qd,
+				const void *restore_mqd, const void *restore_ctl_stack)
 {
+	struct mqd_manager *mqd_mgr;
 	int retval;
 
-	print_queue(q);
-
-	mutex_lock(&dqm->lock);
+	dqm_lock(dqm);
 
 	if (dqm->total_queue_count >= max_num_of_queues_per_device) {
 		pr_warn("Can't create new usermode queue because %d queues were already created\n",
@@ -165,30 +640,76 @@ static int create_queue_nocpsch(struct device_queue_manager *dqm,
 		if (retval)
 			goto out_unlock;
 	}
-	*allocated_vmid = qpd->vmid;
 	q->properties.vmid = qpd->vmid;
+	/*
+	 * Eviction state logic: mark all queues as evicted, even ones
+	 * not currently active. Restoring inactive queues later only
+	 * updates the is_evicted flag but is a no-op otherwise.
+	 */
+	q->properties.is_evicted = !!qpd->evicted;
 
-	if (q->properties.type == KFD_QUEUE_TYPE_COMPUTE)
-		retval = create_compute_queue_nocpsch(dqm, q, qpd);
-	else if (q->properties.type == KFD_QUEUE_TYPE_SDMA)
-		retval = create_sdma_queue_nocpsch(dqm, q, qpd);
+	q->properties.tba_addr = qpd->tba_addr;
+	q->properties.tma_addr = qpd->tma_addr;
+
+	mqd_mgr = dqm->mqd_mgrs[get_mqd_type_from_queue_type(
+			q->properties.type)];
+	if (q->properties.type == KFD_QUEUE_TYPE_COMPUTE) {
+		retval = allocate_hqd(dqm, q);
+		if (retval)
+			goto deallocate_vmid;
+		pr_debug("Loading mqd to hqd on pipe %d, queue %d\n",
+			q->pipe, q->queue);
+	} else if (q->properties.type == KFD_QUEUE_TYPE_SDMA ||
+		q->properties.type == KFD_QUEUE_TYPE_SDMA_XGMI) {
+		retval = allocate_sdma_queue(dqm, q, qd ? &qd->sdma_id : NULL);
+		if (retval)
+			goto deallocate_vmid;
+		dqm->asic_ops.init_sdma_vm(dqm, q, qpd);
+	}
+
+	retval = allocate_doorbell(qpd, q, qd ? &qd->doorbell_id : NULL);
+	if (retval)
+		goto out_deallocate_hqd;
+
+	/* Temporarily release dqm lock to avoid a circular lock dependency */
+	dqm_unlock(dqm);
+	q->mqd_mem_obj = mqd_mgr->allocate_mqd(mqd_mgr->dev, &q->properties);
+	dqm_lock(dqm);
+
+	if (!q->mqd_mem_obj) {
+		retval = -ENOMEM;
+		goto out_deallocate_doorbell;
+	}
+
+	if (qd)
+		mqd_mgr->restore_mqd(mqd_mgr, &q->mqd, q->mqd_mem_obj, &q->gart_mqd_addr,
+				     &q->properties, restore_mqd, restore_ctl_stack,
+				     qd->ctl_stack_size);
 	else
-		retval = -EINVAL;
+		mqd_mgr->init_mqd(mqd_mgr, &q->mqd, q->mqd_mem_obj,
+					&q->gart_mqd_addr, &q->properties);
 
-	if (retval) {
-		if (list_empty(&qpd->queues_list)) {
-			deallocate_vmid(dqm, qpd, q);
-			*allocated_vmid = 0;
+	if (q->properties.is_active) {
+		if (!dqm->sched_running) {
+			WARN_ONCE(1, "Load non-HWS mqd while stopped\n");
+			goto add_queue_to_list;
 		}
-		goto out_unlock;
+
+		if (WARN(q->process->mm != current->mm,
+					"should only run in user thread"))
+			retval = -EFAULT;
+		else
+			retval = mqd_mgr->load_mqd(mqd_mgr, q->mqd, q->pipe,
+					q->queue, &q->properties, current->mm);
+		if (retval)
+			goto out_free_mqd;
 	}
 
+add_queue_to_list:
 	list_add(&q->list, &qpd->queues_list);
+	qpd->queue_count++;
 	if (q->properties.is_active)
-		dqm->queue_count++;
-
-	if (q->properties.type == KFD_QUEUE_TYPE_SDMA)
-		dqm->sdma_queue_count++;
+		increment_queue_count(dqm, qpd, q);
 
 	/*
 	 * Unconditionally increment this counter, regardless of the queue's
@@ -197,9 +718,23 @@ static int create_queue_nocpsch(struct device_queue_manager *dqm,
 	dqm->total_queue_count++;
 	pr_debug("Total of %d queues are accountable so far\n",
 			dqm->total_queue_count);
+	goto out_unlock;
 
+out_free_mqd:
+	mqd_mgr->free_mqd(mqd_mgr, q->mqd, q->mqd_mem_obj);
+out_deallocate_doorbell:
+	deallocate_doorbell(qpd, q);
+out_deallocate_hqd:
+	if (q->properties.type == KFD_QUEUE_TYPE_COMPUTE)
+		deallocate_hqd(dqm, q);
+	else if (q->properties.type == KFD_QUEUE_TYPE_SDMA ||
+		q->properties.type == KFD_QUEUE_TYPE_SDMA_XGMI)
+		deallocate_sdma_queue(dqm, q);
+deallocate_vmid:
+	if (list_empty(&qpd->queues_list))
+		deallocate_vmid(dqm, qpd, q);
 out_unlock:
-	mutex_unlock(&dqm->lock);
+	dqm_unlock(dqm);
 	return retval;
 }
 
@@ -218,12 +753,8 @@ static int allocate_hqd(struct device_queue_manager *dqm, struct queue *q)
 			continue;
 
 		if (dqm->allocated_queues[pipe] != 0) {
-			bit = find_first_bit(
-				(unsigned long *)&dqm->allocated_queues[pipe],
-				get_queues_per_pipe(dqm));
-
-			clear_bit(bit,
-				(unsigned long *)&dqm->allocated_queues[pipe]);
+			bit = ffs(dqm->allocated_queues[pipe]) - 1;
+			dqm->allocated_queues[pipe] &= ~(1 << bit);
 			q->pipe = pipe;
 			q->queue = bit;
 			set = true;
@@ -244,173 +775,625 @@ static int allocate_hqd(struct device_queue_manager *dqm, struct queue *q)
 static inline void deallocate_hqd(struct device_queue_manager *dqm,
 				struct queue *q)
 {
-	set_bit(q->queue, (unsigned long *)&dqm->allocated_queues[q->pipe]);
+	dqm->allocated_queues[q->pipe] |= (1 << q->queue);
 }
 
-static int create_compute_queue_nocpsch(struct device_queue_manager *dqm,
-					struct queue *q,
-					struct qcm_process_device *qpd)
+#define SQ_IND_CMD_CMD_KILL		0x00000003
+#define SQ_IND_CMD_MODE_BROADCAST	0x00000001
+
+static int dbgdev_wave_reset_wavefronts(struct kfd_node *dev, struct kfd_process *p)
 {
-	int retval;
-	struct mqd_manager *mqd;
+	int status = 0;
+	unsigned int vmid;
+	uint16_t queried_pasid;
+	union SQ_CMD_BITS reg_sq_cmd;
+	union GRBM_GFX_INDEX_BITS reg_gfx_index;
+	struct kfd_process_device *pdd;
+	int first_vmid_to_scan = dev->vm_info.first_vmid_kfd;
+	int last_vmid_to_scan = dev->vm_info.last_vmid_kfd;
+	uint32_t xcc_mask = dev->xcc_mask;
+	int xcc_id;
 
-	mqd = dqm->ops.get_mqd_manager(dqm, KFD_MQD_TYPE_COMPUTE);
-	if (!mqd)
-		return -ENOMEM;
+	reg_sq_cmd.u32All = 0;
+	reg_gfx_index.u32All = 0;
 
-	retval = allocate_hqd(dqm, q);
-	if (retval)
-		return retval;
+	pr_debug("Killing all process wavefronts\n");
 
-	retval = mqd->init_mqd(mqd, &q->mqd, &q->mqd_mem_obj,
-				&q->gart_mqd_addr, &q->properties);
-	if (retval)
-		goto out_deallocate_hqd;
+	if (!dev->kfd2kgd->get_atc_vmid_pasid_mapping_info) {
+		dev_err(dev->adev->dev, "no vmid pasid mapping supported\n");
+		return -EOPNOTSUPP;
+	}
 
-	pr_debug("Loading mqd to hqd on pipe %d, queue %d\n",
-			q->pipe, q->queue);
+	/* taking the VMID for that process on the safe way using PDD */
+	pdd = kfd_get_process_device_data(dev, p);
+	if (!pdd)
+		return -EFAULT;
 
-	dqm->dev->kfd2kgd->set_scratch_backing_va(
-			dqm->dev->kgd, qpd->sh_hidden_private_base, qpd->vmid);
+	/* Scan all registers in the range ATC_VMID8_PASID_MAPPING ..
+	 * ATC_VMID15_PASID_MAPPING
+	 * to check which VMID the current process is mapped to.
+	 */
 
-	retval = mqd->load_mqd(mqd, q->mqd, q->pipe, q->queue, &q->properties,
-			       q->process->mm);
-	if (retval)
-		goto out_uninit_mqd;
+	for (vmid = first_vmid_to_scan; vmid <= last_vmid_to_scan; vmid++) {
+		status = dev->kfd2kgd->get_atc_vmid_pasid_mapping_info
+				(dev->adev, vmid, &queried_pasid);
 
-	return 0;
+		if (status && queried_pasid == pdd->pasid) {
+			pr_debug("Killing wave fronts of vmid %d and process pid %d\n",
+					vmid, p->lead_thread->pid);
+			break;
+		}
+	}
 
-out_uninit_mqd:
-	mqd->uninit_mqd(mqd, q->mqd, q->mqd_mem_obj);
-out_deallocate_hqd:
-	deallocate_hqd(dqm, q);
+	if (vmid > last_vmid_to_scan) {
+		dev_err(dev->adev->dev, "Didn't find vmid for process pid %d\n",
+				p->lead_thread->pid);
+		return -EFAULT;
+	}
 
-	return retval;
+	reg_gfx_index.bits.sh_broadcast_writes = 1;
+	reg_gfx_index.bits.se_broadcast_writes = 1;
+	reg_gfx_index.bits.instance_broadcast_writes = 1;
+	reg_sq_cmd.bits.mode = SQ_IND_CMD_MODE_BROADCAST;
+	reg_sq_cmd.bits.cmd = SQ_IND_CMD_CMD_KILL;
+	reg_sq_cmd.bits.vm_id = vmid;
+
+	for_each_inst(xcc_id, xcc_mask)
+		dev->kfd2kgd->wave_control_execute(
+			dev->adev, reg_gfx_index.u32All,
+			reg_sq_cmd.u32All, xcc_id);
+
+	return 0;
 }
 
-static int destroy_queue_nocpsch(struct device_queue_manager *dqm,
+/* Access to DQM has to be locked before calling destroy_queue_nocpsch_locked
+ * to avoid asynchronized access
+ */
+static int destroy_queue_nocpsch_locked(struct device_queue_manager *dqm,
 				struct qcm_process_device *qpd,
 				struct queue *q)
 {
 	int retval;
-	struct mqd_manager *mqd;
-
-	retval = 0;
+	struct mqd_manager *mqd_mgr;
 
-	mutex_lock(&dqm->lock);
+	mqd_mgr = dqm->mqd_mgrs[get_mqd_type_from_queue_type(
+			q->properties.type)];
 
-	if (q->properties.type == KFD_QUEUE_TYPE_COMPUTE) {
-		mqd = dqm->ops.get_mqd_manager(dqm, KFD_MQD_TYPE_COMPUTE);
-		if (mqd == NULL) {
-			retval = -ENOMEM;
-			goto out;
-		}
+	if (q->properties.type == KFD_QUEUE_TYPE_COMPUTE)
 		deallocate_hqd(dqm, q);
-	} else if (q->properties.type == KFD_QUEUE_TYPE_SDMA) {
-		mqd = dqm->ops.get_mqd_manager(dqm, KFD_MQD_TYPE_SDMA);
-		if (mqd == NULL) {
-			retval = -ENOMEM;
-			goto out;
-		}
-		dqm->sdma_queue_count--;
-		deallocate_sdma_queue(dqm, q->sdma_id);
-	} else {
+	else if (q->properties.type == KFD_QUEUE_TYPE_SDMA)
+		deallocate_sdma_queue(dqm, q);
+	else if (q->properties.type == KFD_QUEUE_TYPE_SDMA_XGMI)
+		deallocate_sdma_queue(dqm, q);
+	else {
 		pr_debug("q->properties.type %d is invalid\n",
 				q->properties.type);
-		retval = -EINVAL;
-		goto out;
+		return -EINVAL;
 	}
+	dqm->total_queue_count--;
 
-	retval = mqd->destroy_mqd(mqd, q->mqd,
-				KFD_PREEMPT_TYPE_WAVEFRONT_RESET,
-				QUEUE_PREEMPT_DEFAULT_TIMEOUT_MS,
-				q->pipe, q->queue);
+	deallocate_doorbell(qpd, q);
 
-	if (retval)
-		goto out;
+	if (!dqm->sched_running) {
+		WARN_ONCE(1, "Destroy non-HWS queue while stopped\n");
+		return 0;
+	}
 
-	mqd->uninit_mqd(mqd, q->mqd, q->mqd_mem_obj);
+	retval = mqd_mgr->destroy_mqd(mqd_mgr, q->mqd,
+				KFD_PREEMPT_TYPE_WAVEFRONT_RESET,
+				KFD_UNMAP_LATENCY_MS,
+				q->pipe, q->queue);
+	if (retval == -ETIME)
+		qpd->reset_wavefronts = true;
 
 	list_del(&q->list);
-	if (list_empty(&qpd->queues_list))
+	if (list_empty(&qpd->queues_list)) {
+		if (qpd->reset_wavefronts) {
+			pr_warn("Resetting wave fronts (nocpsch) on dev %p\n",
+					dqm->dev);
+			/* dbgdev_wave_reset_wavefronts has to be called before
+			 * deallocate_vmid(), i.e. when vmid is still in use.
+			 */
+			dbgdev_wave_reset_wavefronts(dqm->dev,
+					qpd->pqm->process);
+			qpd->reset_wavefronts = false;
+		}
+
 		deallocate_vmid(dqm, qpd, q);
+	}
+	qpd->queue_count--;
 	if (q->properties.is_active)
-		dqm->queue_count--;
-
-	/*
-	 * Unconditionally decrement this counter, regardless of the queue's
-	 * type
-	 */
-	dqm->total_queue_count--;
-	pr_debug("Total of %d queues are accountable so far\n",
-			dqm->total_queue_count);
+		decrement_queue_count(dqm, qpd, q);
 
-out:
-	mutex_unlock(&dqm->lock);
 	return retval;
 }
 
-static int update_queue(struct device_queue_manager *dqm, struct queue *q)
+static int destroy_queue_nocpsch(struct device_queue_manager *dqm,
+				struct qcm_process_device *qpd,
+				struct queue *q)
 {
 	int retval;
-	struct mqd_manager *mqd;
+	uint64_t sdma_val = 0;
+	struct device *dev = dqm->dev->adev->dev;
+	struct kfd_process_device *pdd = qpd_to_pdd(qpd);
+	struct mqd_manager *mqd_mgr =
+		dqm->mqd_mgrs[get_mqd_type_from_queue_type(q->properties.type)];
+
+	/* Get the SDMA queue stats */
+	if ((q->properties.type == KFD_QUEUE_TYPE_SDMA) ||
+	    (q->properties.type == KFD_QUEUE_TYPE_SDMA_XGMI)) {
+		retval = read_sdma_queue_counter((uint64_t __user *)q->properties.read_ptr,
+							&sdma_val);
+		if (retval)
+			dev_err(dev, "Failed to read SDMA queue counter for queue: %d\n",
+				q->properties.queue_id);
+	}
+
+	dqm_lock(dqm);
+	retval = destroy_queue_nocpsch_locked(dqm, qpd, q);
+	if (!retval)
+		pdd->sdma_past_activity_counter += sdma_val;
+	dqm_unlock(dqm);
+
+	mqd_mgr->free_mqd(mqd_mgr, q->mqd, q->mqd_mem_obj);
+
+	return retval;
+}
+
+static int update_queue(struct device_queue_manager *dqm, struct queue *q,
+			struct mqd_update_info *minfo)
+{
+	int retval = 0;
+	struct device *dev = dqm->dev->adev->dev;
+	struct mqd_manager *mqd_mgr;
+	struct kfd_process_device *pdd;
 	bool prev_active = false;
 
-	mutex_lock(&dqm->lock);
-	mqd = dqm->ops.get_mqd_manager(dqm,
-			get_mqd_type_from_queue_type(q->properties.type));
-	if (!mqd) {
-		retval = -ENOMEM;
+	dqm_lock(dqm);
+	pdd = kfd_get_process_device_data(q->device, q->process);
+	if (!pdd) {
+		retval = -ENODEV;
 		goto out_unlock;
 	}
+	mqd_mgr = dqm->mqd_mgrs[get_mqd_type_from_queue_type(
+			q->properties.type)];
+
+	/* Save previous activity state for counters */
+	prev_active = q->properties.is_active;
+
+	/* Make sure the queue is unmapped before updating the MQD */
+	if (dqm->sched_policy != KFD_SCHED_POLICY_NO_HWS) {
+		if (!dqm->dev->kfd->shared_resources.enable_mes)
+			retval = unmap_queues_cpsch(dqm,
+						    KFD_UNMAP_QUEUES_FILTER_DYNAMIC_QUEUES, 0, USE_DEFAULT_GRACE_PERIOD, false);
+		else if (prev_active)
+			retval = remove_queue_mes(dqm, q, &pdd->qpd);
+
+		/* queue is reset so inaccessable  */
+		if (pdd->has_reset_queue) {
+			retval = -EACCES;
+			goto out_unlock;
+		}
 
-	if (q->properties.is_active)
-		prev_active = true;
+		if (retval) {
+			dev_err(dev, "unmap queue failed\n");
+			goto out_unlock;
+		}
+	} else if (prev_active &&
+		   (q->properties.type == KFD_QUEUE_TYPE_COMPUTE ||
+		    q->properties.type == KFD_QUEUE_TYPE_SDMA ||
+		    q->properties.type == KFD_QUEUE_TYPE_SDMA_XGMI)) {
+
+		if (!dqm->sched_running) {
+			WARN_ONCE(1, "Update non-HWS queue while stopped\n");
+			goto out_unlock;
+		}
+
+		retval = mqd_mgr->destroy_mqd(mqd_mgr, q->mqd,
+				(dqm->dev->kfd->cwsr_enabled ?
+				 KFD_PREEMPT_TYPE_WAVEFRONT_SAVE :
+				 KFD_PREEMPT_TYPE_WAVEFRONT_DRAIN),
+				KFD_UNMAP_LATENCY_MS, q->pipe, q->queue);
+		if (retval) {
+			dev_err(dev, "destroy mqd failed\n");
+			goto out_unlock;
+		}
+	}
+
+	mqd_mgr->update_mqd(mqd_mgr, q->mqd, &q->properties, minfo);
 
 	/*
-	 *
-	 * check active state vs. the previous state
-	 * and modify counter accordingly
+	 * check active state vs. the previous state and modify
+	 * counter accordingly. map_queues_cpsch uses the
+	 * dqm->active_queue_count to determine whether a new runlist must be
+	 * uploaded.
 	 */
-	retval = mqd->update_mqd(mqd, q->mqd, &q->properties);
-	if ((q->properties.is_active) && (!prev_active))
-		dqm->queue_count++;
-	else if (!q->properties.is_active && prev_active)
-		dqm->queue_count--;
+	if (q->properties.is_active && !prev_active) {
+		increment_queue_count(dqm, &pdd->qpd, q);
+	} else if (!q->properties.is_active && prev_active) {
+		decrement_queue_count(dqm, &pdd->qpd, q);
+	} else if (q->gws && !q->properties.is_gws) {
+		if (q->properties.is_active) {
+			dqm->gws_queue_count++;
+			pdd->qpd.mapped_gws_queue = true;
+		}
+		q->properties.is_gws = true;
+	} else if (!q->gws && q->properties.is_gws) {
+		if (q->properties.is_active) {
+			dqm->gws_queue_count--;
+			pdd->qpd.mapped_gws_queue = false;
+		}
+		q->properties.is_gws = false;
+	}
 
-	if (sched_policy != KFD_SCHED_POLICY_NO_HWS)
-		retval = execute_queues_cpsch(dqm, false);
+	if (dqm->sched_policy != KFD_SCHED_POLICY_NO_HWS) {
+		if (!dqm->dev->kfd->shared_resources.enable_mes)
+			retval = map_queues_cpsch(dqm);
+		else if (q->properties.is_active)
+			retval = add_queue_mes(dqm, q, &pdd->qpd);
+	} else if (q->properties.is_active &&
+		 (q->properties.type == KFD_QUEUE_TYPE_COMPUTE ||
+		  q->properties.type == KFD_QUEUE_TYPE_SDMA ||
+		  q->properties.type == KFD_QUEUE_TYPE_SDMA_XGMI)) {
+		if (WARN(q->process->mm != current->mm,
+			 "should only run in user thread"))
+			retval = -EFAULT;
+		else
+			retval = mqd_mgr->load_mqd(mqd_mgr, q->mqd,
+						   q->pipe, q->queue,
+						   &q->properties, current->mm);
+	}
 
 out_unlock:
-	mutex_unlock(&dqm->lock);
+	dqm_unlock(dqm);
 	return retval;
 }
 
-static struct mqd_manager *get_mqd_manager_nocpsch(
-		struct device_queue_manager *dqm, enum KFD_MQD_TYPE type)
+/* suspend_single_queue does not lock the dqm like the
+ * evict_process_queues_cpsch or evict_process_queues_nocpsch. You should
+ * lock the dqm before calling, and unlock after calling.
+ *
+ * The reason we don't lock the dqm is because this function may be
+ * called on multiple queues in a loop, so rather than locking/unlocking
+ * multiple times, we will just keep the dqm locked for all of the calls.
+ */
+static int suspend_single_queue(struct device_queue_manager *dqm,
+				      struct kfd_process_device *pdd,
+				      struct queue *q)
 {
-	struct mqd_manager *mqd;
+	bool is_new;
 
-	if (WARN_ON(type >= KFD_MQD_TYPE_MAX))
-		return NULL;
+	if (q->properties.is_suspended)
+		return 0;
+
+	pr_debug("Suspending process pid %d queue [%i]\n",
+			pdd->process->lead_thread->pid,
+			q->properties.queue_id);
 
-	pr_debug("mqd type %d\n", type);
+	is_new = q->properties.exception_status & KFD_EC_MASK(EC_QUEUE_NEW);
 
-	mqd = dqm->mqds[type];
-	if (!mqd) {
-		mqd = mqd_manager_init(type, dqm->dev);
-		if (!mqd)
-			pr_err("mqd manager is NULL");
-		dqm->mqds[type] = mqd;
+	if (is_new || q->properties.is_being_destroyed) {
+		pr_debug("Suspend: skip %s queue id %i\n",
+				is_new ? "new" : "destroyed",
+				q->properties.queue_id);
+		return -EBUSY;
 	}
 
-	return mqd;
+	q->properties.is_suspended = true;
+	if (q->properties.is_active) {
+		if (dqm->dev->kfd->shared_resources.enable_mes) {
+			int r = remove_queue_mes(dqm, q, &pdd->qpd);
+
+			if (r)
+				return r;
+		}
+
+		decrement_queue_count(dqm, &pdd->qpd, q);
+		q->properties.is_active = false;
+	}
+
+	return 0;
 }
 
-static int register_process_nocpsch(struct device_queue_manager *dqm,
+/* resume_single_queue does not lock the dqm like the functions
+ * restore_process_queues_cpsch or restore_process_queues_nocpsch. You should
+ * lock the dqm before calling, and unlock after calling.
+ *
+ * The reason we don't lock the dqm is because this function may be
+ * called on multiple queues in a loop, so rather than locking/unlocking
+ * multiple times, we will just keep the dqm locked for all of the calls.
+ */
+static int resume_single_queue(struct device_queue_manager *dqm,
+				      struct qcm_process_device *qpd,
+				      struct queue *q)
+{
+	struct kfd_process_device *pdd;
+
+	if (!q->properties.is_suspended)
+		return 0;
+
+	pdd = qpd_to_pdd(qpd);
+
+	pr_debug("Restoring from suspend process pid %d queue [%i]\n",
+			    pdd->process->lead_thread->pid,
+			    q->properties.queue_id);
+
+	q->properties.is_suspended = false;
+
+	if (QUEUE_IS_ACTIVE(q->properties)) {
+		if (dqm->dev->kfd->shared_resources.enable_mes) {
+			int r = add_queue_mes(dqm, q, &pdd->qpd);
+
+			if (r)
+				return r;
+		}
+
+		q->properties.is_active = true;
+		increment_queue_count(dqm, qpd, q);
+	}
+
+	return 0;
+}
+
+static int evict_process_queues_nocpsch(struct device_queue_manager *dqm,
+					struct qcm_process_device *qpd)
+{
+	struct queue *q;
+	struct mqd_manager *mqd_mgr;
+	struct kfd_process_device *pdd;
+	int retval, ret = 0;
+
+	dqm_lock(dqm);
+	if (qpd->evicted++ > 0) /* already evicted, do nothing */
+		goto out;
+
+	pdd = qpd_to_pdd(qpd);
+	pr_debug_ratelimited("Evicting process pid %d queues\n",
+			    pdd->process->lead_thread->pid);
+
+	pdd->last_evict_timestamp = get_jiffies_64();
+	/* Mark all queues as evicted. Deactivate all active queues on
+	 * the qpd.
+	 */
+	list_for_each_entry(q, &qpd->queues_list, list) {
+		q->properties.is_evicted = true;
+		if (!q->properties.is_active)
+			continue;
+
+		mqd_mgr = dqm->mqd_mgrs[get_mqd_type_from_queue_type(
+				q->properties.type)];
+		q->properties.is_active = false;
+		decrement_queue_count(dqm, qpd, q);
+
+		if (WARN_ONCE(!dqm->sched_running, "Evict when stopped\n"))
+			continue;
+
+		retval = mqd_mgr->destroy_mqd(mqd_mgr, q->mqd,
+				(dqm->dev->kfd->cwsr_enabled ?
+				 KFD_PREEMPT_TYPE_WAVEFRONT_SAVE :
+				 KFD_PREEMPT_TYPE_WAVEFRONT_DRAIN),
+				KFD_UNMAP_LATENCY_MS, q->pipe, q->queue);
+		if (retval && !ret)
+			/* Return the first error, but keep going to
+			 * maintain a consistent eviction state
+			 */
+			ret = retval;
+	}
+
+out:
+	dqm_unlock(dqm);
+	return ret;
+}
+
+static int evict_process_queues_cpsch(struct device_queue_manager *dqm,
+				      struct qcm_process_device *qpd)
+{
+	struct queue *q;
+	struct device *dev = dqm->dev->adev->dev;
+	struct kfd_process_device *pdd;
+	int retval = 0;
+
+	dqm_lock(dqm);
+	if (qpd->evicted++ > 0) /* already evicted, do nothing */
+		goto out;
+
+	pdd = qpd_to_pdd(qpd);
+
+	/* The debugger creates processes that temporarily have not acquired
+	 * all VMs for all devices and has no VMs itself.
+	 * Skip queue eviction on process eviction.
+	 */
+	if (!pdd->drm_priv)
+		goto out;
+
+	pr_debug_ratelimited("Evicting process pid %d queues\n",
+			    pdd->process->lead_thread->pid);
+
+	/* Mark all queues as evicted. Deactivate all active queues on
+	 * the qpd.
+	 */
+	list_for_each_entry(q, &qpd->queues_list, list) {
+		q->properties.is_evicted = true;
+		if (!q->properties.is_active)
+			continue;
+
+		q->properties.is_active = false;
+		decrement_queue_count(dqm, qpd, q);
+
+		if (dqm->dev->kfd->shared_resources.enable_mes) {
+			int err;
+
+			err = remove_queue_mes(dqm, q, qpd);
+			if (err) {
+				dev_err(dev, "Failed to evict queue %d\n",
+					q->properties.queue_id);
+				retval = err;
+			}
+		}
+	}
+	pdd->last_evict_timestamp = get_jiffies_64();
+	if (!dqm->dev->kfd->shared_resources.enable_mes)
+		retval = execute_queues_cpsch(dqm,
+					      qpd->is_debug ?
+					      KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES :
+					      KFD_UNMAP_QUEUES_FILTER_DYNAMIC_QUEUES, 0,
+					      USE_DEFAULT_GRACE_PERIOD);
+
+out:
+	dqm_unlock(dqm);
+	return retval;
+}
+
+static int restore_process_queues_nocpsch(struct device_queue_manager *dqm,
+					  struct qcm_process_device *qpd)
+{
+	struct mm_struct *mm = NULL;
+	struct queue *q;
+	struct mqd_manager *mqd_mgr;
+	struct kfd_process_device *pdd;
+	uint64_t pd_base;
+	uint64_t eviction_duration;
+	int retval, ret = 0;
+
+	pdd = qpd_to_pdd(qpd);
+	/* Retrieve PD base */
+	pd_base = amdgpu_amdkfd_gpuvm_get_process_page_dir(pdd->drm_priv);
+
+	dqm_lock(dqm);
+	if (WARN_ON_ONCE(!qpd->evicted)) /* already restored, do nothing */
+		goto out;
+	if (qpd->evicted > 1) { /* ref count still > 0, decrement & quit */
+		qpd->evicted--;
+		goto out;
+	}
+
+	pr_debug_ratelimited("Restoring process pid %d queues\n",
+			    pdd->process->lead_thread->pid);
+
+	/* Update PD Base in QPD */
+	qpd->page_table_base = pd_base;
+	pr_debug("Updated PD address to 0x%llx\n", pd_base);
+
+	if (!list_empty(&qpd->queues_list)) {
+		dqm->dev->kfd2kgd->set_vm_context_page_table_base(
+				dqm->dev->adev,
+				qpd->vmid,
+				qpd->page_table_base);
+		kfd_flush_tlb(pdd, TLB_FLUSH_LEGACY);
+	}
+
+	/* Take a safe reference to the mm_struct, which may otherwise
+	 * disappear even while the kfd_process is still referenced.
+	 */
+	mm = get_task_mm(pdd->process->lead_thread);
+	if (!mm) {
+		ret = -EFAULT;
+		goto out;
+	}
+
+	/* Remove the eviction flags. Activate queues that are not
+	 * inactive for other reasons.
+	 */
+	list_for_each_entry(q, &qpd->queues_list, list) {
+		q->properties.is_evicted = false;
+		if (!QUEUE_IS_ACTIVE(q->properties))
+			continue;
+
+		mqd_mgr = dqm->mqd_mgrs[get_mqd_type_from_queue_type(
+				q->properties.type)];
+		q->properties.is_active = true;
+		increment_queue_count(dqm, qpd, q);
+
+		if (WARN_ONCE(!dqm->sched_running, "Restore when stopped\n"))
+			continue;
+
+		retval = mqd_mgr->load_mqd(mqd_mgr, q->mqd, q->pipe,
+				       q->queue, &q->properties, mm);
+		if (retval && !ret)
+			/* Return the first error, but keep going to
+			 * maintain a consistent eviction state
+			 */
+			ret = retval;
+	}
+	qpd->evicted = 0;
+	eviction_duration = get_jiffies_64() - pdd->last_evict_timestamp;
+	atomic64_add(eviction_duration, &pdd->evict_duration_counter);
+out:
+	if (mm)
+		mmput(mm);
+	dqm_unlock(dqm);
+	return ret;
+}
+
+static int restore_process_queues_cpsch(struct device_queue_manager *dqm,
+					struct qcm_process_device *qpd)
+{
+	struct queue *q;
+	struct device *dev = dqm->dev->adev->dev;
+	struct kfd_process_device *pdd;
+	uint64_t eviction_duration;
+	int retval = 0;
+
+	pdd = qpd_to_pdd(qpd);
+
+	dqm_lock(dqm);
+	if (WARN_ON_ONCE(!qpd->evicted)) /* already restored, do nothing */
+		goto out;
+	if (qpd->evicted > 1) { /* ref count still > 0, decrement & quit */
+		qpd->evicted--;
+		goto out;
+	}
+
+	/* The debugger creates processes that temporarily have not acquired
+	 * all VMs for all devices and has no VMs itself.
+	 * Skip queue restore on process restore.
+	 */
+	if (!pdd->drm_priv)
+		goto vm_not_acquired;
+
+	pr_debug_ratelimited("Restoring process pid %d queues\n",
+			    pdd->process->lead_thread->pid);
+
+	/* Update PD Base in QPD */
+	qpd->page_table_base = amdgpu_amdkfd_gpuvm_get_process_page_dir(pdd->drm_priv);
+	pr_debug("Updated PD address to 0x%llx\n", qpd->page_table_base);
+
+	/* activate all active queues on the qpd */
+	list_for_each_entry(q, &qpd->queues_list, list) {
+		q->properties.is_evicted = false;
+		if (!QUEUE_IS_ACTIVE(q->properties))
+			continue;
+
+		q->properties.is_active = true;
+		increment_queue_count(dqm, &pdd->qpd, q);
+
+		if (dqm->dev->kfd->shared_resources.enable_mes) {
+			retval = add_queue_mes(dqm, q, qpd);
+			if (retval) {
+				dev_err(dev, "Failed to restore queue %d\n",
+					q->properties.queue_id);
+				goto out;
+			}
+		}
+	}
+	if (!dqm->dev->kfd->shared_resources.enable_mes)
+		retval = execute_queues_cpsch(dqm,
+					      KFD_UNMAP_QUEUES_FILTER_DYNAMIC_QUEUES, 0, USE_DEFAULT_GRACE_PERIOD);
+	eviction_duration = get_jiffies_64() - pdd->last_evict_timestamp;
+	atomic64_add(eviction_duration, &pdd->evict_duration_counter);
+vm_not_acquired:
+	qpd->evicted = 0;
+out:
+	dqm_unlock(dqm);
+	return retval;
+}
+
+static int register_process(struct device_queue_manager *dqm,
 					struct qcm_process_device *qpd)
 {
 	struct device_process_node *n;
+	struct kfd_process_device *pdd;
+	uint64_t pd_base;
 	int retval;
 
 	n = kzalloc(sizeof(*n), GFP_KERNEL);
@@ -419,19 +1402,32 @@ static int register_process_nocpsch(struct device_queue_manager *dqm,
 
 	n->qpd = qpd;
 
-	mutex_lock(&dqm->lock);
+	pdd = qpd_to_pdd(qpd);
+	/* Retrieve PD base */
+	pd_base = amdgpu_amdkfd_gpuvm_get_process_page_dir(pdd->drm_priv);
+
+	dqm_lock(dqm);
 	list_add(&n->list, &dqm->queues);
 
-	retval = dqm->ops_asic_specific.register_process(dqm, qpd);
+	/* Update PD Base in QPD */
+	qpd->page_table_base = pd_base;
+	pr_debug("Updated PD address to 0x%llx\n", pd_base);
+
+	retval = dqm->asic_ops.update_qpd(dqm, qpd);
 
 	dqm->processes_count++;
 
-	mutex_unlock(&dqm->lock);
+	dqm_unlock(dqm);
+
+	/* Outside the DQM lock because under the DQM lock we can't do
+	 * reclaim or take other locks that others hold while reclaiming.
+	 */
+	kfd_inc_compute_active(dqm->dev);
 
 	return retval;
 }
 
-static int unregister_process_nocpsch(struct device_queue_manager *dqm,
+static int unregister_process(struct device_queue_manager *dqm,
 					struct qcm_process_device *qpd)
 {
 	int retval;
@@ -441,7 +1437,7 @@ static int unregister_process_nocpsch(struct device_queue_manager *dqm,
 			list_empty(&qpd->queues_list) ? "empty" : "not empty");
 
 	retval = 0;
-	mutex_lock(&dqm->lock);
+	dqm_lock(dqm);
 
 	list_for_each_entry_safe(cur, next, &dqm->queues, list) {
 		if (qpd == cur->qpd) {
@@ -454,32 +1450,47 @@ static int unregister_process_nocpsch(struct device_queue_manager *dqm,
 	/* qpd not found in dqm list */
 	retval = 1;
 out:
-	mutex_unlock(&dqm->lock);
+	dqm_unlock(dqm);
+
+	/* Outside the DQM lock because under the DQM lock we can't do
+	 * reclaim or take other locks that others hold while reclaiming.
+	 */
+	if (!retval)
+		kfd_dec_compute_active(dqm->dev);
+
 	return retval;
 }
 
 static int
-set_pasid_vmid_mapping(struct device_queue_manager *dqm, unsigned int pasid,
+set_pasid_vmid_mapping(struct device_queue_manager *dqm, u32 pasid,
 			unsigned int vmid)
 {
-	uint32_t pasid_mapping;
+	uint32_t xcc_mask = dqm->dev->xcc_mask;
+	int xcc_id, ret;
 
-	pasid_mapping = (pasid == 0) ? 0 :
-		(uint32_t)pasid |
-		ATC_VMID_PASID_MAPPING_VALID;
+	for_each_inst(xcc_id, xcc_mask) {
+		ret = dqm->dev->kfd2kgd->set_pasid_vmid_mapping(
+			dqm->dev->adev, pasid, vmid, xcc_id);
+		if (ret)
+			break;
+	}
 
-	return dqm->dev->kfd2kgd->set_pasid_vmid_mapping(
-						dqm->dev->kgd, pasid_mapping,
-						vmid);
+	return ret;
 }
 
 static void init_interrupts(struct device_queue_manager *dqm)
 {
-	unsigned int i;
-
-	for (i = 0 ; i < get_pipes_per_mec(dqm) ; i++)
-		if (is_pipe_enabled(dqm, 0, i))
-			dqm->dev->kfd2kgd->init_interrupts(dqm->dev->kgd, i);
+	uint32_t xcc_mask = dqm->dev->xcc_mask;
+	unsigned int i, xcc_id;
+
+	for_each_inst(xcc_id, xcc_mask) {
+		for (i = 0 ; i < get_pipes_per_mec(dqm) ; i++) {
+			if (is_pipe_enabled(dqm, 0, i)) {
+				dqm->dev->kfd2kgd->init_interrupts(
+					dqm->dev->adev, i, xcc_id);
+			}
+		}
+	}
 }
 
 static int initialize_nocpsch(struct device_queue_manager *dqm)
@@ -493,115 +1504,190 @@ static int initialize_nocpsch(struct device_queue_manager *dqm)
 	if (!dqm->allocated_queues)
 		return -ENOMEM;
 
-	mutex_init(&dqm->lock);
+	mutex_init(&dqm->lock_hidden);
 	INIT_LIST_HEAD(&dqm->queues);
-	dqm->queue_count = dqm->next_pipe_to_allocate = 0;
-	dqm->sdma_queue_count = 0;
+	dqm->active_queue_count = dqm->next_pipe_to_allocate = 0;
+	dqm->active_cp_queue_count = 0;
+	dqm->gws_queue_count = 0;
 
 	for (pipe = 0; pipe < get_pipes_per_mec(dqm); pipe++) {
 		int pipe_offset = pipe * get_queues_per_pipe(dqm);
 
 		for (queue = 0; queue < get_queues_per_pipe(dqm); queue++)
 			if (test_bit(pipe_offset + queue,
-				     dqm->dev->shared_resources.queue_bitmap))
+				     dqm->dev->kfd->shared_resources.cp_queue_bitmap))
 				dqm->allocated_queues[pipe] |= 1 << queue;
 	}
 
-	dqm->vmid_bitmap = (1 << VMID_PER_DEVICE) - 1;
-	dqm->sdma_bitmap = (1 << CIK_SDMA_QUEUES) - 1;
+	memset(dqm->vmid_pasid, 0, sizeof(dqm->vmid_pasid));
+
+	init_sdma_bitmaps(dqm);
 
 	return 0;
 }
 
-static void uninitialize_nocpsch(struct device_queue_manager *dqm)
+static void uninitialize(struct device_queue_manager *dqm)
 {
 	int i;
 
-	WARN_ON(dqm->queue_count > 0 || dqm->processes_count > 0);
+	WARN_ON(dqm->active_queue_count > 0 || dqm->processes_count > 0);
 
 	kfree(dqm->allocated_queues);
 	for (i = 0 ; i < KFD_MQD_TYPE_MAX ; i++)
-		kfree(dqm->mqds[i]);
-	mutex_destroy(&dqm->lock);
-	kfd_gtt_sa_free(dqm->dev, dqm->pipeline_mem);
+		kfree(dqm->mqd_mgrs[i]);
+	mutex_destroy(&dqm->lock_hidden);
 }
 
 static int start_nocpsch(struct device_queue_manager *dqm)
 {
+	int r = 0;
+
+	pr_info("SW scheduler is used");
 	init_interrupts(dqm);
-	return 0;
+
+	if (dqm->dev->adev->asic_type == CHIP_HAWAII)
+		r = pm_init(&dqm->packet_mgr, dqm);
+	if (!r)
+		dqm->sched_running = true;
+
+	return r;
 }
 
 static int stop_nocpsch(struct device_queue_manager *dqm)
 {
+	dqm_lock(dqm);
+	if (!dqm->sched_running) {
+		dqm_unlock(dqm);
+		return 0;
+	}
+
+	if (dqm->dev->adev->asic_type == CHIP_HAWAII)
+		pm_uninit(&dqm->packet_mgr);
+	dqm->sched_running = false;
+	dqm_unlock(dqm);
+
 	return 0;
 }
 
 static int allocate_sdma_queue(struct device_queue_manager *dqm,
-				unsigned int *sdma_queue_id)
+				struct queue *q, const uint32_t *restore_sdma_id)
 {
+	struct device *dev = dqm->dev->adev->dev;
 	int bit;
 
-	if (dqm->sdma_bitmap == 0)
-		return -ENOMEM;
-
-	bit = find_first_bit((unsigned long *)&dqm->sdma_bitmap,
-				CIK_SDMA_QUEUES);
-
-	clear_bit(bit, (unsigned long *)&dqm->sdma_bitmap);
-	*sdma_queue_id = bit;
+	if (q->properties.type == KFD_QUEUE_TYPE_SDMA) {
+		if (bitmap_empty(dqm->sdma_bitmap, get_num_sdma_queues(dqm))) {
+			dev_warn(dev, "No more SDMA queue to allocate (%d total queues)\n",
+				 get_num_sdma_queues(dqm));
+			return -ENOMEM;
+		}
 
-	return 0;
-}
+		if (restore_sdma_id) {
+			/* Re-use existing sdma_id */
+			if (!test_bit(*restore_sdma_id, dqm->sdma_bitmap)) {
+				dev_err(dev, "SDMA queue already in use\n");
+				return -EBUSY;
+			}
+			clear_bit(*restore_sdma_id, dqm->sdma_bitmap);
+			q->sdma_id = *restore_sdma_id;
+		} else {
+			/* Find first available sdma_id */
+			bit = find_first_bit(dqm->sdma_bitmap,
+					     get_num_sdma_queues(dqm));
+			clear_bit(bit, dqm->sdma_bitmap);
+			q->sdma_id = bit;
+		}
 
-static void deallocate_sdma_queue(struct device_queue_manager *dqm,
-				unsigned int sdma_queue_id)
-{
-	if (sdma_queue_id >= CIK_SDMA_QUEUES)
-		return;
-	set_bit(sdma_queue_id, (unsigned long *)&dqm->sdma_bitmap);
-}
+		q->properties.sdma_engine_id =
+			q->sdma_id % kfd_get_num_sdma_engines(dqm->dev);
+		q->properties.sdma_queue_id = q->sdma_id /
+				kfd_get_num_sdma_engines(dqm->dev);
+	} else if (q->properties.type == KFD_QUEUE_TYPE_SDMA_XGMI) {
+		if (bitmap_empty(dqm->xgmi_sdma_bitmap, get_num_xgmi_sdma_queues(dqm))) {
+			dev_warn(dev, "No more XGMI SDMA queue to allocate (%d total queues)\n",
+				 get_num_xgmi_sdma_queues(dqm));
+			return -ENOMEM;
+		}
+		if (restore_sdma_id) {
+			/* Re-use existing sdma_id */
+			if (!test_bit(*restore_sdma_id, dqm->xgmi_sdma_bitmap)) {
+				dev_err(dev, "SDMA queue already in use\n");
+				return -EBUSY;
+			}
+			clear_bit(*restore_sdma_id, dqm->xgmi_sdma_bitmap);
+			q->sdma_id = *restore_sdma_id;
+		} else {
+			bit = find_first_bit(dqm->xgmi_sdma_bitmap,
+					     get_num_xgmi_sdma_queues(dqm));
+			clear_bit(bit, dqm->xgmi_sdma_bitmap);
+			q->sdma_id = bit;
+		}
+		/* sdma_engine_id is sdma id including
+		 * both PCIe-optimized SDMAs and XGMI-
+		 * optimized SDMAs. The calculation below
+		 * assumes the first N engines are always
+		 * PCIe-optimized ones
+		 */
+		q->properties.sdma_engine_id =
+			kfd_get_num_sdma_engines(dqm->dev) +
+			q->sdma_id % kfd_get_num_xgmi_sdma_engines(dqm->dev);
+		q->properties.sdma_queue_id = q->sdma_id /
+			kfd_get_num_xgmi_sdma_engines(dqm->dev);
+	} else if (q->properties.type == KFD_QUEUE_TYPE_SDMA_BY_ENG_ID) {
+		int i, num_queues, num_engines, eng_offset = 0, start_engine;
+		bool free_bit_found = false, is_xgmi = false;
+
+		if (q->properties.sdma_engine_id < kfd_get_num_sdma_engines(dqm->dev)) {
+			num_queues = get_num_sdma_queues(dqm);
+			num_engines = kfd_get_num_sdma_engines(dqm->dev);
+			q->properties.type = KFD_QUEUE_TYPE_SDMA;
+		} else {
+			num_queues = get_num_xgmi_sdma_queues(dqm);
+			num_engines = kfd_get_num_xgmi_sdma_engines(dqm->dev);
+			eng_offset = kfd_get_num_sdma_engines(dqm->dev);
+			q->properties.type = KFD_QUEUE_TYPE_SDMA_XGMI;
+			is_xgmi = true;
+		}
 
-static int create_sdma_queue_nocpsch(struct device_queue_manager *dqm,
-					struct queue *q,
-					struct qcm_process_device *qpd)
-{
-	struct mqd_manager *mqd;
-	int retval;
+		/* Scan available bit based on target engine ID. */
+		start_engine = q->properties.sdma_engine_id - eng_offset;
+		for (i = start_engine; i < num_queues; i += num_engines) {
 
-	mqd = dqm->ops.get_mqd_manager(dqm, KFD_MQD_TYPE_SDMA);
-	if (!mqd)
-		return -ENOMEM;
+			if (!test_bit(i, is_xgmi ? dqm->xgmi_sdma_bitmap : dqm->sdma_bitmap))
+				continue;
 
-	retval = allocate_sdma_queue(dqm, &q->sdma_id);
-	if (retval)
-		return retval;
+			clear_bit(i, is_xgmi ? dqm->xgmi_sdma_bitmap : dqm->sdma_bitmap);
+			q->sdma_id = i;
+			q->properties.sdma_queue_id = q->sdma_id / num_engines;
+			free_bit_found = true;
+			break;
+		}
 
-	q->properties.sdma_queue_id = q->sdma_id % CIK_SDMA_QUEUES_PER_ENGINE;
-	q->properties.sdma_engine_id = q->sdma_id / CIK_SDMA_ENGINE_NUM;
+		if (!free_bit_found) {
+			dev_warn(dev, "No more SDMA queue to allocate for target ID %i (%d total queues)\n",
+				 q->properties.sdma_engine_id, num_queues);
+			return -ENOMEM;
+		}
+	}
 
-	pr_debug("SDMA id is:    %d\n", q->sdma_id);
-	pr_debug("SDMA queue id: %d\n", q->properties.sdma_queue_id);
 	pr_debug("SDMA engine id: %d\n", q->properties.sdma_engine_id);
-
-	dqm->ops_asic_specific.init_sdma_vm(dqm, q, qpd);
-	retval = mqd->init_mqd(mqd, &q->mqd, &q->mqd_mem_obj,
-				&q->gart_mqd_addr, &q->properties);
-	if (retval)
-		goto out_deallocate_sdma_queue;
-
-	retval = mqd->load_mqd(mqd, q->mqd, 0, 0, &q->properties, NULL);
-	if (retval)
-		goto out_uninit_mqd;
+	pr_debug("SDMA queue id: %d\n", q->properties.sdma_queue_id);
 
 	return 0;
+}
 
-out_uninit_mqd:
-	mqd->uninit_mqd(mqd, q->mqd, q->mqd_mem_obj);
-out_deallocate_sdma_queue:
-	deallocate_sdma_queue(dqm, q->sdma_id);
-
-	return retval;
+static void deallocate_sdma_queue(struct device_queue_manager *dqm,
+				struct queue *q)
+{
+	if (q->properties.type == KFD_QUEUE_TYPE_SDMA) {
+		if (q->sdma_id >= get_num_sdma_queues(dqm))
+			return;
+		set_bit(q->sdma_id, dqm->sdma_bitmap);
+	} else if (q->properties.type == KFD_QUEUE_TYPE_SDMA_XGMI) {
+		if (q->sdma_id >= get_num_xgmi_sdma_queues(dqm))
+			return;
+		set_bit(q->sdma_id, dqm->xgmi_sdma_bitmap);
+	}
 }
 
 /*
@@ -612,16 +1698,16 @@ static int set_sched_resources(struct device_queue_manager *dqm)
 {
 	int i, mec;
 	struct scheduling_resources res;
+	struct device *dev = dqm->dev->adev->dev;
 
-	res.vmid_mask = (1 << VMID_PER_DEVICE) - 1;
-	res.vmid_mask <<= KFD_VMID_START_OFFSET;
+	res.vmid_mask = dqm->dev->compute_vmid_bitmap;
 
 	res.queue_mask = 0;
-	for (i = 0; i < KGD_MAX_QUEUES; ++i) {
-		mec = (i / dqm->dev->shared_resources.num_queue_per_pipe)
-			/ dqm->dev->shared_resources.num_pipe_per_mec;
+	for (i = 0; i < AMDGPU_MAX_QUEUES; ++i) {
+		mec = (i / dqm->dev->kfd->shared_resources.num_queue_per_pipe)
+			/ dqm->dev->kfd->shared_resources.num_pipe_per_mec;
 
-		if (!test_bit(i, dqm->dev->shared_resources.queue_bitmap))
+		if (!test_bit(i, dqm->dev->kfd->shared_resources.cp_queue_bitmap))
 			continue;
 
 		/* only acquire queues from the first MEC */
@@ -633,56 +1719,118 @@ static int set_sched_resources(struct device_queue_manager *dqm)
 		 * definition of res.queue_mask needs updating
 		 */
 		if (WARN_ON(i >= (sizeof(res.queue_mask)*8))) {
-			pr_err("Invalid queue enabled by amdgpu: %d\n", i);
+			dev_err(dev, "Invalid queue enabled by amdgpu: %d\n", i);
 			break;
 		}
 
-		res.queue_mask |= (1ull << i);
+		res.queue_mask |= 1ull
+			<< amdgpu_queue_mask_bit_to_set_resource_bit(
+				dqm->dev->adev, i);
 	}
-	res.gws_mask = res.oac_mask = res.gds_heap_base =
-						res.gds_heap_size = 0;
+	res.gws_mask = ~0ull;
+	res.oac_mask = res.gds_heap_base = res.gds_heap_size = 0;
 
 	pr_debug("Scheduling resources:\n"
 			"vmid mask: 0x%8X\n"
 			"queue mask: 0x%8llX\n",
 			res.vmid_mask, res.queue_mask);
 
-	return pm_send_set_resources(&dqm->packets, &res);
+	return pm_send_set_resources(&dqm->packet_mgr, &res);
 }
 
 static int initialize_cpsch(struct device_queue_manager *dqm)
 {
-	int retval;
-
 	pr_debug("num of pipes: %d\n", get_pipes_per_mec(dqm));
 
-	mutex_init(&dqm->lock);
+	mutex_init(&dqm->lock_hidden);
 	INIT_LIST_HEAD(&dqm->queues);
-	dqm->queue_count = dqm->processes_count = 0;
-	dqm->sdma_queue_count = 0;
+	dqm->active_queue_count = dqm->processes_count = 0;
+	dqm->active_cp_queue_count = 0;
+	dqm->gws_queue_count = 0;
 	dqm->active_runlist = false;
-	retval = dqm->ops_asic_specific.initialize(dqm);
-	if (retval)
-		mutex_destroy(&dqm->lock);
+	dqm->trap_debug_vmid = 0;
 
-	return retval;
+	init_sdma_bitmaps(dqm);
+
+	update_dqm_wait_times(dqm);
+	return 0;
+}
+
+/* halt_cpsch:
+ * Unmap queues so the schedule doesn't continue remaining jobs in the queue.
+ * Then set dqm->sched_halt so queues don't map to runlist until unhalt_cpsch
+ * is called.
+ */
+static int halt_cpsch(struct device_queue_manager *dqm)
+{
+	int ret = 0;
+
+	dqm_lock(dqm);
+	if (!dqm->sched_running) {
+		dqm_unlock(dqm);
+		return 0;
+	}
+
+	WARN_ONCE(dqm->sched_halt, "Scheduling is already on halt\n");
+
+	if (!dqm->is_hws_hang) {
+		if (!dqm->dev->kfd->shared_resources.enable_mes)
+			ret = unmap_queues_cpsch(dqm,
+						 KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES, 0,
+				USE_DEFAULT_GRACE_PERIOD, false);
+		else
+			ret = remove_all_kfd_queues_mes(dqm);
+	}
+	dqm->sched_halt = true;
+	dqm_unlock(dqm);
+
+	return ret;
+}
+
+/* unhalt_cpsch
+ * Unset dqm->sched_halt and map queues back to runlist
+ */
+static int unhalt_cpsch(struct device_queue_manager *dqm)
+{
+	int ret = 0;
+
+	dqm_lock(dqm);
+	if (!dqm->sched_running || !dqm->sched_halt) {
+		WARN_ONCE(!dqm->sched_halt, "Scheduling is not on halt.\n");
+		dqm_unlock(dqm);
+		return 0;
+	}
+	dqm->sched_halt = false;
+	if (!dqm->dev->kfd->shared_resources.enable_mes)
+		ret = execute_queues_cpsch(dqm,
+					   KFD_UNMAP_QUEUES_FILTER_DYNAMIC_QUEUES,
+			0, USE_DEFAULT_GRACE_PERIOD);
+	else
+		ret = add_all_kfd_queues_mes(dqm);
+
+	dqm_unlock(dqm);
+
+	return ret;
 }
 
 static int start_cpsch(struct device_queue_manager *dqm)
 {
-	struct device_process_node *node;
-	int retval;
+	struct device *dev = dqm->dev->adev->dev;
+	int retval, num_hw_queue_slots;
 
 	retval = 0;
 
-	retval = pm_init(&dqm->packets, dqm);
-	if (retval)
-		goto fail_packet_manager_init;
+	dqm_lock(dqm);
 
-	retval = set_sched_resources(dqm);
-	if (retval)
-		goto fail_set_sched_resources;
+	if (!dqm->dev->kfd->shared_resources.enable_mes) {
+		retval = pm_init(&dqm->packet_mgr, dqm);
+		if (retval)
+			goto fail_packet_manager_init;
 
+		retval = set_sched_resources(dqm);
+		if (retval)
+			goto fail_set_sched_resources;
+	}
 	pr_debug("Allocating fence memory\n");
 
 	/* allocate fence memory on the gart */
@@ -692,39 +1840,72 @@ static int start_cpsch(struct device_queue_manager *dqm)
 	if (retval)
 		goto fail_allocate_vidmem;
 
-	dqm->fence_addr = dqm->fence_mem->cpu_ptr;
+	dqm->fence_addr = (uint64_t *)dqm->fence_mem->cpu_ptr;
 	dqm->fence_gpu_addr = dqm->fence_mem->gpu_addr;
 
 	init_interrupts(dqm);
 
-	list_for_each_entry(node, &dqm->queues, list)
-		if (node->qpd->pqm->process && dqm->dev)
-			kfd_bind_process_to_device(dqm->dev,
-						node->qpd->pqm->process);
+	/* clear hang status when driver try to start the hw scheduler */
+	dqm->sched_running = true;
+
+	if (!dqm->dev->kfd->shared_resources.enable_mes) {
+		if (pm_config_dequeue_wait_counts(&dqm->packet_mgr,
+				KFD_DEQUEUE_WAIT_INIT, 0 /* unused */))
+			dev_err(dev, "Setting optimized dequeue wait failed. Using default values\n");
+		execute_queues_cpsch(dqm, KFD_UNMAP_QUEUES_FILTER_DYNAMIC_QUEUES, 0, USE_DEFAULT_GRACE_PERIOD);
+	}
+
+	/* setup per-queue reset detection buffer  */
+	num_hw_queue_slots =  dqm->dev->kfd->shared_resources.num_queue_per_pipe *
+			      dqm->dev->kfd->shared_resources.num_pipe_per_mec *
+			      NUM_XCC(dqm->dev->xcc_mask);
+
+	dqm->detect_hang_info_size = num_hw_queue_slots * sizeof(struct dqm_detect_hang_info);
+	dqm->detect_hang_info = kzalloc(dqm->detect_hang_info_size, GFP_KERNEL);
+
+	if (!dqm->detect_hang_info) {
+		retval = -ENOMEM;
+		goto fail_detect_hang_buffer;
+	}
 
-	execute_queues_cpsch(dqm, true);
+	dqm_unlock(dqm);
 
 	return 0;
+fail_detect_hang_buffer:
+	kfd_gtt_sa_free(dqm->dev, dqm->fence_mem);
 fail_allocate_vidmem:
 fail_set_sched_resources:
-	pm_uninit(&dqm->packets);
+	if (!dqm->dev->kfd->shared_resources.enable_mes)
+		pm_uninit(&dqm->packet_mgr);
 fail_packet_manager_init:
+	dqm_unlock(dqm);
 	return retval;
 }
 
 static int stop_cpsch(struct device_queue_manager *dqm)
 {
-	struct device_process_node *node;
-	struct kfd_process_device *pdd;
+	dqm_lock(dqm);
+	if (!dqm->sched_running) {
+		dqm_unlock(dqm);
+		return 0;
+	}
+
+	if (!dqm->dev->kfd->shared_resources.enable_mes)
+		unmap_queues_cpsch(dqm, KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES, 0, USE_DEFAULT_GRACE_PERIOD, false);
+	else
+		remove_all_kfd_queues_mes(dqm);
 
-	destroy_queues_cpsch(dqm, true, true);
+	dqm->sched_running = false;
+
+	if (!dqm->dev->kfd->shared_resources.enable_mes)
+		pm_release_ib(&dqm->packet_mgr);
 
-	list_for_each_entry(node, &dqm->queues, list) {
-		pdd = qpd_to_pdd(node->qpd);
-		pdd->bound = false;
-	}
 	kfd_gtt_sa_free(dqm->dev, dqm->fence_mem);
-	pm_uninit(&dqm->packets);
+	if (!dqm->dev->kfd->shared_resources.enable_mes)
+		pm_uninit(&dqm->packet_mgr);
+	kfree(dqm->detect_hang_info);
+	dqm->detect_hang_info = NULL;
+	dqm_unlock(dqm);
 
 	return 0;
 }
@@ -733,11 +1914,11 @@ static int create_kernel_queue_cpsch(struct device_queue_manager *dqm,
 					struct kernel_queue *kq,
 					struct qcm_process_device *qpd)
 {
-	mutex_lock(&dqm->lock);
+	dqm_lock(dqm);
 	if (dqm->total_queue_count >= max_num_of_queues_per_device) {
 		pr_warn("Can't create new kernel queue because %d queues were already created\n",
 				dqm->total_queue_count);
-		mutex_unlock(&dqm->lock);
+		dqm_unlock(dqm);
 		return -EPERM;
 	}
 
@@ -750,10 +1931,11 @@ static int create_kernel_queue_cpsch(struct device_queue_manager *dqm,
 			dqm->total_queue_count);
 
 	list_add(&kq->list, &qpd->priv_queue_list);
-	dqm->queue_count++;
+	increment_queue_count(dqm, qpd, kq->queue);
 	qpd->is_debug = true;
-	execute_queues_cpsch(dqm, false);
-	mutex_unlock(&dqm->lock);
+	execute_queues_cpsch(dqm, KFD_UNMAP_QUEUES_FILTER_DYNAMIC_QUEUES, 0,
+			USE_DEFAULT_GRACE_PERIOD);
+	dqm_unlock(dqm);
 
 	return 0;
 }
@@ -762,13 +1944,12 @@ static void destroy_kernel_queue_cpsch(struct device_queue_manager *dqm,
 					struct kernel_queue *kq,
 					struct qcm_process_device *qpd)
 {
-	mutex_lock(&dqm->lock);
-	/* here we actually preempt the DIQ */
-	destroy_queues_cpsch(dqm, true, false);
+	dqm_lock(dqm);
 	list_del(&kq->list);
-	dqm->queue_count--;
+	decrement_queue_count(dqm, qpd, kq->queue);
 	qpd->is_debug = false;
-	execute_queues_cpsch(dqm, false);
+	execute_queues_cpsch(dqm, KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES, 0,
+			USE_DEFAULT_GRACE_PERIOD);
 	/*
 	 * Unconditionally decrement this counter, regardless of the queue's
 	 * type.
@@ -776,29 +1957,16 @@ static void destroy_kernel_queue_cpsch(struct device_queue_manager *dqm,
 	dqm->total_queue_count--;
 	pr_debug("Total of %d queues are accountable so far\n",
 			dqm->total_queue_count);
-	mutex_unlock(&dqm->lock);
-}
-
-static void select_sdma_engine_id(struct queue *q)
-{
-	static int sdma_id;
-
-	q->sdma_id = sdma_id;
-	sdma_id = (sdma_id + 1) % 2;
+	dqm_unlock(dqm);
 }
 
 static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
-			struct qcm_process_device *qpd, int *allocate_vmid)
+			struct qcm_process_device *qpd,
+			const struct kfd_criu_queue_priv_data *qd,
+			const void *restore_mqd, const void *restore_ctl_stack)
 {
 	int retval;
-	struct mqd_manager *mqd;
-
-	retval = 0;
-
-	if (allocate_vmid)
-		*allocate_vmid = 0;
-
-	mutex_lock(&dqm->lock);
+	struct mqd_manager *mqd_mgr;
 
 	if (dqm->total_queue_count >= max_num_of_queues_per_device) {
 		pr_warn("Can't create new usermode queue because %d queues were already created\n",
@@ -807,31 +1975,67 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 		goto out;
 	}
 
-	if (q->properties.type == KFD_QUEUE_TYPE_SDMA)
-		select_sdma_engine_id(q);
+	if (q->properties.type == KFD_QUEUE_TYPE_SDMA ||
+		q->properties.type == KFD_QUEUE_TYPE_SDMA_XGMI ||
+		q->properties.type == KFD_QUEUE_TYPE_SDMA_BY_ENG_ID) {
+		dqm_lock(dqm);
+		retval = allocate_sdma_queue(dqm, q, qd ? &qd->sdma_id : NULL);
+		dqm_unlock(dqm);
+		if (retval)
+			goto out;
+	}
+
+	retval = allocate_doorbell(qpd, q, qd ? &qd->doorbell_id : NULL);
+	if (retval)
+		goto out_deallocate_sdma_queue;
 
-	mqd = dqm->ops.get_mqd_manager(dqm,
-			get_mqd_type_from_queue_type(q->properties.type));
+	mqd_mgr = dqm->mqd_mgrs[get_mqd_type_from_queue_type(
+			q->properties.type)];
 
-	if (!mqd) {
+	if (q->properties.type == KFD_QUEUE_TYPE_SDMA ||
+		q->properties.type == KFD_QUEUE_TYPE_SDMA_XGMI)
+		dqm->asic_ops.init_sdma_vm(dqm, q, qpd);
+	q->properties.tba_addr = qpd->tba_addr;
+	q->properties.tma_addr = qpd->tma_addr;
+	q->mqd_mem_obj = mqd_mgr->allocate_mqd(mqd_mgr->dev, &q->properties);
+	if (!q->mqd_mem_obj) {
 		retval = -ENOMEM;
-		goto out;
+		goto out_deallocate_doorbell;
 	}
 
-	dqm->ops_asic_specific.init_sdma_vm(dqm, q, qpd);
-	retval = mqd->init_mqd(mqd, &q->mqd, &q->mqd_mem_obj,
-				&q->gart_mqd_addr, &q->properties);
-	if (retval)
-		goto out;
+	dqm_lock(dqm);
+	/*
+	 * Eviction state logic: mark all queues as evicted, even ones
+	 * not currently active. Restoring inactive queues later only
+	 * updates the is_evicted flag but is a no-op otherwise.
+	 */
+	q->properties.is_evicted = !!qpd->evicted;
+	q->properties.is_dbg_wa = qpd->pqm->process->debug_trap_enabled &&
+				  kfd_dbg_has_cwsr_workaround(q->device);
+
+	if (qd)
+		mqd_mgr->restore_mqd(mqd_mgr, &q->mqd, q->mqd_mem_obj, &q->gart_mqd_addr,
+				     &q->properties, restore_mqd, restore_ctl_stack,
+				     qd->ctl_stack_size);
+	else
+		mqd_mgr->init_mqd(mqd_mgr, &q->mqd, q->mqd_mem_obj,
+					&q->gart_mqd_addr, &q->properties);
 
 	list_add(&q->list, &qpd->queues_list);
+	qpd->queue_count++;
+
 	if (q->properties.is_active) {
-		dqm->queue_count++;
-		retval = execute_queues_cpsch(dqm, false);
+		increment_queue_count(dqm, qpd, q);
+
+		if (!dqm->dev->kfd->shared_resources.enable_mes)
+			retval = execute_queues_cpsch(dqm,
+					KFD_UNMAP_QUEUES_FILTER_DYNAMIC_QUEUES, 0, USE_DEFAULT_GRACE_PERIOD);
+		else
+			retval = add_queue_mes(dqm, q, qpd);
+		if (retval)
+			goto cleanup_queue;
 	}
 
-	if (q->properties.type == KFD_QUEUE_TYPE_SDMA)
-		dqm->sdma_queue_count++;
 	/*
 	 * Unconditionally increment this counter, regardless of the queue's
 	 * type or whether the queue is active.
@@ -841,20 +2045,51 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 	pr_debug("Total of %d queues are accountable so far\n",
 			dqm->total_queue_count);
 
+	dqm_unlock(dqm);
+	return retval;
+
+cleanup_queue:
+	qpd->queue_count--;
+	list_del(&q->list);
+	if (q->properties.is_active)
+		decrement_queue_count(dqm, qpd, q);
+	mqd_mgr->free_mqd(mqd_mgr, q->mqd, q->mqd_mem_obj);
+	dqm_unlock(dqm);
+out_deallocate_doorbell:
+	deallocate_doorbell(qpd, q);
+out_deallocate_sdma_queue:
+	if (q->properties.type == KFD_QUEUE_TYPE_SDMA ||
+		q->properties.type == KFD_QUEUE_TYPE_SDMA_XGMI) {
+		dqm_lock(dqm);
+		deallocate_sdma_queue(dqm, q);
+		dqm_unlock(dqm);
+	}
 out:
-	mutex_unlock(&dqm->lock);
 	return retval;
 }
 
-int amdkfd_fence_wait_timeout(unsigned int *fence_addr,
-				unsigned int fence_value,
-				unsigned long timeout)
+int amdkfd_fence_wait_timeout(struct device_queue_manager *dqm,
+			      uint64_t fence_value,
+			      unsigned int timeout_ms)
 {
-	timeout += jiffies;
+	unsigned long end_jiffies = msecs_to_jiffies(timeout_ms) + jiffies;
+	struct device *dev = dqm->dev->adev->dev;
+	uint64_t *fence_addr = dqm->fence_addr;
 
 	while (*fence_addr != fence_value) {
-		if (time_after(jiffies, timeout)) {
-			pr_err("qcm fence wait loop timeout expired\n");
+		/* Fatal err detected, this response won't come */
+		if (amdgpu_amdkfd_is_fed(dqm->dev->adev))
+			return -EIO;
+
+		if (time_after(jiffies, end_jiffies)) {
+			dev_err(dev, "qcm fence wait loop timeout expired\n");
+			/* In HWS case, this is used to halt the driver thread
+			 * in order not to mess up CP states before doing
+			 * scandumps for FW debugging.
+			 */
+			while (halt_if_hws_hang)
+				schedule();
+
 			return -ETIME;
 		}
 		schedule();
@@ -863,100 +2098,408 @@ int amdkfd_fence_wait_timeout(unsigned int *fence_addr,
 	return 0;
 }
 
-static int destroy_sdma_queues(struct device_queue_manager *dqm,
-				unsigned int sdma_engine)
+/* dqm->lock mutex has to be locked before calling this function */
+static int map_queues_cpsch(struct device_queue_manager *dqm)
 {
-	return pm_send_unmap_queue(&dqm->packets, KFD_QUEUE_TYPE_SDMA,
-			KFD_PREEMPT_TYPE_FILTER_DYNAMIC_QUEUES, 0, false,
-			sdma_engine);
+	struct device *dev = dqm->dev->adev->dev;
+	int retval;
+
+	if (!dqm->sched_running || dqm->sched_halt)
+		return 0;
+	if (dqm->active_queue_count <= 0 || dqm->processes_count <= 0)
+		return 0;
+	if (dqm->active_runlist)
+		return 0;
+
+	retval = pm_send_runlist(&dqm->packet_mgr, &dqm->queues);
+	pr_debug("%s sent runlist\n", __func__);
+	if (retval) {
+		dev_err(dev, "failed to execute runlist\n");
+		return retval;
+	}
+	dqm->active_runlist = true;
+
+	return retval;
 }
 
-static int destroy_queues_cpsch(struct device_queue_manager *dqm,
-				bool preempt_static_queues, bool lock)
+static void set_queue_as_reset(struct device_queue_manager *dqm, struct queue *q,
+			       struct qcm_process_device *qpd)
 {
-	int retval;
-	enum kfd_preempt_type_filter preempt_type;
-	struct kfd_process_device *pdd;
+	struct kfd_process_device *pdd = qpd_to_pdd(qpd);
 
-	retval = 0;
+	dev_err(dqm->dev->adev->dev, "queue id 0x%0x at pasid %d is reset\n",
+		q->properties.queue_id, pdd->process->lead_thread->pid);
 
-	if (lock)
-		mutex_lock(&dqm->lock);
-	if (!dqm->active_runlist)
-		goto out;
+	pdd->has_reset_queue = true;
+	if (q->properties.is_active) {
+		q->properties.is_active = false;
+		decrement_queue_count(dqm, qpd, q);
+	}
+}
 
-	pr_debug("Before destroying queues, sdma queue count is : %u\n",
-		dqm->sdma_queue_count);
+static int detect_queue_hang(struct device_queue_manager *dqm)
+{
+	int i;
+
+	/* detect should be used only in dqm locked queue reset */
+	if (WARN_ON(dqm->detect_hang_count > 0))
+		return 0;
+
+	memset(dqm->detect_hang_info, 0, dqm->detect_hang_info_size);
+
+	for (i = 0; i < AMDGPU_MAX_QUEUES; ++i) {
+		uint32_t mec, pipe, queue;
+		int xcc_id;
+
+		mec = (i / dqm->dev->kfd->shared_resources.num_queue_per_pipe)
+			/ dqm->dev->kfd->shared_resources.num_pipe_per_mec;
+
+		if (mec || !test_bit(i, dqm->dev->kfd->shared_resources.cp_queue_bitmap))
+			continue;
 
-	if (dqm->sdma_queue_count > 0) {
-		destroy_sdma_queues(dqm, 0);
-		destroy_sdma_queues(dqm, 1);
+		amdgpu_queue_mask_bit_to_mec_queue(dqm->dev->adev, i, &mec, &pipe, &queue);
+
+		for_each_inst(xcc_id, dqm->dev->xcc_mask) {
+			uint64_t queue_addr = dqm->dev->kfd2kgd->hqd_get_pq_addr(
+						dqm->dev->adev, pipe, queue, xcc_id);
+			struct dqm_detect_hang_info hang_info;
+
+			if (!queue_addr)
+				continue;
+
+			hang_info.pipe_id = pipe;
+			hang_info.queue_id = queue;
+			hang_info.xcc_id = xcc_id;
+			hang_info.queue_address = queue_addr;
+
+			dqm->detect_hang_info[dqm->detect_hang_count] = hang_info;
+			dqm->detect_hang_count++;
+		}
 	}
 
-	preempt_type = preempt_static_queues ?
-			KFD_PREEMPT_TYPE_FILTER_ALL_QUEUES :
-			KFD_PREEMPT_TYPE_FILTER_DYNAMIC_QUEUES;
+	return dqm->detect_hang_count;
+}
 
-	retval = pm_send_unmap_queue(&dqm->packets, KFD_QUEUE_TYPE_COMPUTE,
-			preempt_type, 0, false, 0);
+static struct queue *find_queue_by_address(struct device_queue_manager *dqm, uint64_t queue_address)
+{
+	struct device_process_node *cur;
+	struct qcm_process_device *qpd;
+	struct queue *q;
+
+	list_for_each_entry(cur, &dqm->queues, list) {
+		qpd = cur->qpd;
+		list_for_each_entry(q, &qpd->queues_list, list) {
+			if (queue_address == q->properties.queue_address)
+				return q;
+		}
+	}
+
+	return NULL;
+}
+
+static int reset_hung_queues(struct device_queue_manager *dqm)
+{
+	int r = 0, reset_count = 0, i;
+
+	if (!dqm->detect_hang_info || dqm->is_hws_hang)
+		return -EIO;
+
+	/* assume dqm locked. */
+	if (!detect_queue_hang(dqm))
+		return -ENOTRECOVERABLE;
+
+	for (i = 0; i < dqm->detect_hang_count; i++) {
+		struct dqm_detect_hang_info hang_info = dqm->detect_hang_info[i];
+		struct queue *q = find_queue_by_address(dqm, hang_info.queue_address);
+		struct kfd_process_device *pdd;
+		uint64_t queue_addr = 0;
+
+		if (!q) {
+			r = -ENOTRECOVERABLE;
+			goto reset_fail;
+		}
+
+		pdd = kfd_get_process_device_data(dqm->dev, q->process);
+		if (!pdd) {
+			r = -ENOTRECOVERABLE;
+			goto reset_fail;
+		}
+
+		queue_addr = dqm->dev->kfd2kgd->hqd_reset(dqm->dev->adev,
+				hang_info.pipe_id, hang_info.queue_id, hang_info.xcc_id,
+				KFD_UNMAP_LATENCY_MS);
+
+		/* either reset failed or we reset an unexpected queue. */
+		if (queue_addr != q->properties.queue_address) {
+			r = -ENOTRECOVERABLE;
+			goto reset_fail;
+		}
+
+		set_queue_as_reset(dqm, q, &pdd->qpd);
+		reset_count++;
+	}
+
+	if (reset_count == dqm->detect_hang_count)
+		kfd_signal_reset_event(dqm->dev);
+	else
+		r = -ENOTRECOVERABLE;
+
+reset_fail:
+	dqm->detect_hang_count = 0;
+
+	return r;
+}
+
+static bool sdma_has_hang(struct device_queue_manager *dqm)
+{
+	int engine_start = dqm->dev->node_id * get_num_all_sdma_engines(dqm);
+	int engine_end = engine_start + get_num_all_sdma_engines(dqm);
+	int num_queues_per_eng =  dqm->dev->kfd->device_info.num_sdma_queues_per_engine;
+	int i, j;
+
+	for (i = engine_start; i < engine_end; i++) {
+		for (j = 0; j < num_queues_per_eng; j++) {
+			if (!dqm->dev->kfd2kgd->hqd_sdma_get_doorbell(dqm->dev->adev, i, j))
+				continue;
+
+			return true;
+		}
+	}
+
+	return false;
+}
+
+static bool set_sdma_queue_as_reset(struct device_queue_manager *dqm,
+				    uint32_t doorbell_off)
+{
+	struct device_process_node *cur;
+	struct qcm_process_device *qpd;
+	struct queue *q;
+
+	list_for_each_entry(cur, &dqm->queues, list) {
+		qpd = cur->qpd;
+		list_for_each_entry(q, &qpd->queues_list, list) {
+			if ((q->properties.type == KFD_QUEUE_TYPE_SDMA ||
+			     q->properties.type == KFD_QUEUE_TYPE_SDMA_XGMI) &&
+			     q->properties.doorbell_off == doorbell_off) {
+				set_queue_as_reset(dqm, q, qpd);
+				return true;
+			}
+		}
+	}
+
+	return false;
+}
+
+static int reset_hung_queues_sdma(struct device_queue_manager *dqm)
+{
+	int engine_start = dqm->dev->node_id * get_num_all_sdma_engines(dqm);
+	int engine_end = engine_start + get_num_all_sdma_engines(dqm);
+	int num_queues_per_eng =  dqm->dev->kfd->device_info.num_sdma_queues_per_engine;
+	int r = 0, i, j;
+
+	if (dqm->is_hws_hang)
+		return -EIO;
+
+	/* Scan for hung HW queues and reset engine. */
+	dqm->detect_hang_count = 0;
+	for (i = engine_start; i < engine_end; i++) {
+		for (j = 0; j < num_queues_per_eng; j++) {
+			uint32_t doorbell_off =
+				dqm->dev->kfd2kgd->hqd_sdma_get_doorbell(dqm->dev->adev, i, j);
+
+			if (!doorbell_off)
+				continue;
+
+			/* Reset engine and check. */
+			if (amdgpu_sdma_reset_engine(dqm->dev->adev, i, false) ||
+			    dqm->dev->kfd2kgd->hqd_sdma_get_doorbell(dqm->dev->adev, i, j) ||
+			    !set_sdma_queue_as_reset(dqm, doorbell_off)) {
+				r = -ENOTRECOVERABLE;
+				goto reset_fail;
+			}
+
+			/* Should only expect one queue active per engine */
+			dqm->detect_hang_count++;
+			break;
+		}
+	}
+
+	/* Signal process reset */
+	if (dqm->detect_hang_count)
+		kfd_signal_reset_event(dqm->dev);
+	else
+		r = -ENOTRECOVERABLE;
+
+reset_fail:
+	dqm->detect_hang_count = 0;
+
+	return r;
+}
+
+static int reset_queues_on_hws_hang(struct device_queue_manager *dqm, bool is_sdma)
+{
+	struct amdgpu_device *adev = dqm->dev->adev;
+
+	while (halt_if_hws_hang)
+		schedule();
+
+	if (adev->debug_disable_gpu_ring_reset) {
+		dev_info_once(adev->dev,
+			      "%s queue hung, but ring reset disabled",
+			      is_sdma ? "sdma" : "compute");
+
+		return -EPERM;
+	}
+	if (!amdgpu_gpu_recovery)
+		return -ENOTRECOVERABLE;
+
+	return is_sdma ? reset_hung_queues_sdma(dqm) : reset_hung_queues(dqm);
+}
+
+/* dqm->lock mutex has to be locked before calling this function
+ *
+ * @grace_period: If USE_DEFAULT_GRACE_PERIOD then default wait time
+ *   for context switch latency. Lower values are used by debugger
+ *   since context switching are triggered at high frequency.
+ *   This is configured by setting CP_IQ_WAIT_TIME2.SCH_WAVE
+ *
+ */
+static int unmap_queues_cpsch(struct device_queue_manager *dqm,
+				enum kfd_unmap_queues_filter filter,
+				uint32_t filter_param,
+				uint32_t grace_period,
+				bool reset)
+{
+	struct device *dev = dqm->dev->adev->dev;
+	struct mqd_manager *mqd_mgr;
+	int retval;
+
+	if (!dqm->sched_running)
+		return 0;
+	if (!dqm->active_runlist)
+		return 0;
+	if (!down_read_trylock(&dqm->dev->adev->reset_domain->sem))
+		return -EIO;
+
+	if (grace_period != USE_DEFAULT_GRACE_PERIOD) {
+		retval = pm_config_dequeue_wait_counts(&dqm->packet_mgr,
+				KFD_DEQUEUE_WAIT_SET_SCH_WAVE, grace_period);
+		if (retval)
+			goto out;
+	}
+
+	retval = pm_send_unmap_queue(&dqm->packet_mgr, filter, filter_param, reset);
 	if (retval)
 		goto out;
 
 	*dqm->fence_addr = KFD_FENCE_INIT;
-	pm_send_query_status(&dqm->packets, dqm->fence_gpu_addr,
+	mb();
+	pm_send_query_status(&dqm->packet_mgr, dqm->fence_gpu_addr,
 				KFD_FENCE_COMPLETED);
 	/* should be timed out */
-	retval = amdkfd_fence_wait_timeout(dqm->fence_addr, KFD_FENCE_COMPLETED,
-				QUEUE_PREEMPT_DEFAULT_TIMEOUT_MS);
+	retval = amdkfd_fence_wait_timeout(dqm, KFD_FENCE_COMPLETED,
+					   queue_preemption_timeout_ms);
 	if (retval) {
-		pdd = kfd_get_process_device_data(dqm->dev,
-				kfd_get_process(current));
-		pdd->reset_wavefronts = true;
+		dev_err(dev, "The cp might be in an unrecoverable state due to an unsuccessful queues preemption\n");
+		kfd_hws_hang(dqm);
 		goto out;
 	}
-	pm_release_ib(&dqm->packets);
-	dqm->active_runlist = false;
 
+	/* In the current MEC firmware implementation, if compute queue
+	 * doesn't response to the preemption request in time, HIQ will
+	 * abandon the unmap request without returning any timeout error
+	 * to driver. Instead, MEC firmware will log the doorbell of the
+	 * unresponding compute queue to HIQ.MQD.queue_doorbell_id fields.
+	 * To make sure the queue unmap was successful, driver need to
+	 * check those fields
+	 */
+	mqd_mgr = dqm->mqd_mgrs[KFD_MQD_TYPE_HIQ];
+	if (mqd_mgr->check_preemption_failed(mqd_mgr, dqm->packet_mgr.priv_queue->queue->mqd) &&
+	    reset_queues_on_hws_hang(dqm, false))
+		goto reset_fail;
+
+	/* Check for SDMA hang and attempt SDMA reset */
+	if (sdma_has_hang(dqm) && reset_queues_on_hws_hang(dqm, true))
+		goto reset_fail;
+
+	/* We need to reset the grace period value for this device */
+	if (grace_period != USE_DEFAULT_GRACE_PERIOD) {
+		if (pm_config_dequeue_wait_counts(&dqm->packet_mgr,
+				KFD_DEQUEUE_WAIT_RESET, 0 /* unused */))
+			dev_err(dev, "Failed to reset grace period\n");
+	}
+
+	pm_release_ib(&dqm->packet_mgr);
+	dqm->active_runlist = false;
 out:
-	if (lock)
-		mutex_unlock(&dqm->lock);
+	up_read(&dqm->dev->adev->reset_domain->sem);
 	return retval;
+
+reset_fail:
+	dqm->is_hws_hang = true;
+	kfd_hws_hang(dqm);
+	up_read(&dqm->dev->adev->reset_domain->sem);
+	return -ETIME;
 }
 
-static int execute_queues_cpsch(struct device_queue_manager *dqm, bool lock)
+/* only for compute queue */
+static int reset_queues_cpsch(struct device_queue_manager *dqm, uint16_t pasid)
 {
 	int retval;
 
-	if (lock)
-		mutex_lock(&dqm->lock);
+	dqm_lock(dqm);
 
-	retval = destroy_queues_cpsch(dqm, false, false);
-	if (retval) {
-		pr_err("The cp might be in an unrecoverable state due to an unsuccessful queues preemption");
-		goto out;
-	}
+	retval = unmap_queues_cpsch(dqm, KFD_UNMAP_QUEUES_FILTER_BY_PASID,
+			pasid, USE_DEFAULT_GRACE_PERIOD, true);
 
-	if (dqm->queue_count <= 0 || dqm->processes_count <= 0) {
-		retval = 0;
-		goto out;
-	}
+	dqm_unlock(dqm);
+	return retval;
+}
 
-	if (dqm->active_runlist) {
-		retval = 0;
-		goto out;
-	}
+/* dqm->lock mutex has to be locked before calling this function */
+static int execute_queues_cpsch(struct device_queue_manager *dqm,
+				enum kfd_unmap_queues_filter filter,
+				uint32_t filter_param,
+				uint32_t grace_period)
+{
+	int retval;
 
-	retval = pm_send_runlist(&dqm->packets, &dqm->queues);
-	if (retval) {
-		pr_err("failed to execute runlist");
-		goto out;
+	if (!down_read_trylock(&dqm->dev->adev->reset_domain->sem))
+		return -EIO;
+	retval = unmap_queues_cpsch(dqm, filter, filter_param, grace_period, false);
+	if (!retval)
+		retval = map_queues_cpsch(dqm);
+	up_read(&dqm->dev->adev->reset_domain->sem);
+	return retval;
+}
+
+static int wait_on_destroy_queue(struct device_queue_manager *dqm,
+				 struct queue *q)
+{
+	struct kfd_process_device *pdd = kfd_get_process_device_data(q->device,
+								q->process);
+	int ret = 0;
+
+	if (WARN_ON(!pdd))
+		return ret;
+
+	if (pdd->qpd.is_debug)
+		return ret;
+
+	q->properties.is_being_destroyed = true;
+
+	if (pdd->process->debug_trap_enabled && q->properties.is_suspended) {
+		dqm_unlock(dqm);
+		mutex_unlock(&q->process->mutex);
+		ret = wait_event_interruptible(dqm->destroy_wait,
+						!q->properties.is_suspended);
+
+		mutex_lock(&q->process->mutex);
+		dqm_lock(dqm);
 	}
-	dqm->active_runlist = true;
 
-out:
-	if (lock)
-		mutex_unlock(&dqm->lock);
-	return retval;
+	return ret;
 }
 
 static int destroy_queue_cpsch(struct device_queue_manager *dqm,
@@ -964,15 +2507,30 @@ static int destroy_queue_cpsch(struct device_queue_manager *dqm,
 				struct queue *q)
 {
 	int retval;
-	struct mqd_manager *mqd;
-	bool preempt_all_queues;
+	struct mqd_manager *mqd_mgr;
+	uint64_t sdma_val = 0;
+	struct kfd_process_device *pdd = qpd_to_pdd(qpd);
+	struct device *dev = dqm->dev->adev->dev;
+
+	/* Get the SDMA queue stats */
+	if ((q->properties.type == KFD_QUEUE_TYPE_SDMA) ||
+	    (q->properties.type == KFD_QUEUE_TYPE_SDMA_XGMI)) {
+		retval = read_sdma_queue_counter((uint64_t __user *)q->properties.read_ptr,
+							&sdma_val);
+		if (retval)
+			dev_err(dev, "Failed to read SDMA queue counter for queue: %d\n",
+				q->properties.queue_id);
+	}
 
-	preempt_all_queues = false;
+	/* remove queue from list to prevent rescheduling after preemption */
+	dqm_lock(dqm);
 
-	retval = 0;
+	retval = wait_on_destroy_queue(dqm, q);
 
-	/* remove queue from list to prevent rescheduling after preemption */
-	mutex_lock(&dqm->lock);
+	if (retval) {
+		dqm_unlock(dqm);
+		return retval;
+	}
 
 	if (qpd->is_debug) {
 		/*
@@ -984,23 +2542,32 @@ static int destroy_queue_cpsch(struct device_queue_manager *dqm,
 
 	}
 
-	mqd = dqm->ops.get_mqd_manager(dqm,
-			get_mqd_type_from_queue_type(q->properties.type));
-	if (!mqd) {
-		retval = -ENOMEM;
-		goto failed;
-	}
+	mqd_mgr = dqm->mqd_mgrs[get_mqd_type_from_queue_type(
+			q->properties.type)];
 
-	if (q->properties.type == KFD_QUEUE_TYPE_SDMA)
-		dqm->sdma_queue_count--;
-
-	list_del(&q->list);
-	if (q->properties.is_active)
-		dqm->queue_count--;
+	deallocate_doorbell(qpd, q);
 
-	execute_queues_cpsch(dqm, false);
+	if ((q->properties.type == KFD_QUEUE_TYPE_SDMA) ||
+	    (q->properties.type == KFD_QUEUE_TYPE_SDMA_XGMI)) {
+		deallocate_sdma_queue(dqm, q);
+		pdd->sdma_past_activity_counter += sdma_val;
+	}
 
-	mqd->uninit_mqd(mqd, q->mqd, q->mqd_mem_obj);
+	if (q->properties.is_active) {
+		decrement_queue_count(dqm, qpd, q);
+		q->properties.is_active = false;
+		if (!dqm->dev->kfd->shared_resources.enable_mes) {
+			retval = execute_queues_cpsch(dqm,
+						      KFD_UNMAP_QUEUES_FILTER_DYNAMIC_QUEUES, 0,
+						      USE_DEFAULT_GRACE_PERIOD);
+			if (retval == -ETIME)
+				qpd->reset_wavefronts = true;
+		} else {
+			retval = remove_queue_mes(dqm, q, qpd);
+		}
+	}
+	list_del(&q->list);
+	qpd->queue_count--;
 
 	/*
 	 * Unconditionally decrement this counter, regardless of the queue's
@@ -1010,73 +2577,54 @@ static int destroy_queue_cpsch(struct device_queue_manager *dqm,
 	pr_debug("Total of %d queues are accountable so far\n",
 			dqm->total_queue_count);
 
-	mutex_unlock(&dqm->lock);
+	dqm_unlock(dqm);
 
-	return 0;
+	/*
+	 * Do free_mqd and raise delete event after dqm_unlock(dqm) to avoid
+	 * circular locking
+	 */
+	kfd_dbg_ev_raise(KFD_EC_MASK(EC_DEVICE_QUEUE_DELETE),
+				qpd->pqm->process, q->device,
+				-1, false, NULL, 0);
+
+	mqd_mgr->free_mqd(mqd_mgr, q->mqd, q->mqd_mem_obj);
+
+	return retval;
 
-failed:
 failed_try_destroy_debugged_queue:
 
-	mutex_unlock(&dqm->lock);
+	dqm_unlock(dqm);
 	return retval;
 }
 
-/*
- * Low bits must be 0000/FFFF as required by HW, high bits must be 0 to
- * stay in user mode.
- */
-#define APE1_FIXED_BITS_MASK 0xFFFF80000000FFFFULL
-/* APE1 limit is inclusive and 64K aligned. */
-#define APE1_LIMIT_ALIGNMENT 0xFFFF
-
 static bool set_cache_memory_policy(struct device_queue_manager *dqm,
 				   struct qcm_process_device *qpd,
 				   enum cache_policy default_policy,
 				   enum cache_policy alternate_policy,
 				   void __user *alternate_aperture_base,
-				   uint64_t alternate_aperture_size)
+				   uint64_t alternate_aperture_size,
+				   u32 misc_process_properties)
 {
-	bool retval;
-
-	mutex_lock(&dqm->lock);
-
-	if (alternate_aperture_size == 0) {
-		/* base > limit disables APE1 */
-		qpd->sh_mem_ape1_base = 1;
-		qpd->sh_mem_ape1_limit = 0;
-	} else {
-		/*
-		 * In FSA64, APE1_Base[63:0] = { 16{SH_MEM_APE1_BASE[31]},
-		 *			SH_MEM_APE1_BASE[31:0], 0x0000 }
-		 * APE1_Limit[63:0] = { 16{SH_MEM_APE1_LIMIT[31]},
-		 *			SH_MEM_APE1_LIMIT[31:0], 0xFFFF }
-		 * Verify that the base and size parameters can be
-		 * represented in this format and convert them.
-		 * Additionally restrict APE1 to user-mode addresses.
-		 */
+	bool retval = true;
 
-		uint64_t base = (uintptr_t)alternate_aperture_base;
-		uint64_t limit = base + alternate_aperture_size - 1;
-
-		if (limit <= base || (base & APE1_FIXED_BITS_MASK) != 0 ||
-		   (limit & APE1_FIXED_BITS_MASK) != APE1_LIMIT_ALIGNMENT) {
-			retval = false;
-			goto out;
-		}
+	if (!dqm->asic_ops.set_cache_memory_policy)
+		return retval;
 
-		qpd->sh_mem_ape1_base = base >> 16;
-		qpd->sh_mem_ape1_limit = limit >> 16;
-	}
+	dqm_lock(dqm);
 
-	retval = dqm->ops_asic_specific.set_cache_memory_policy(
+	retval = dqm->asic_ops.set_cache_memory_policy(
 			dqm,
 			qpd,
 			default_policy,
 			alternate_policy,
 			alternate_aperture_base,
-			alternate_aperture_size);
+			alternate_aperture_size,
+			misc_process_properties);
+
+	if (retval)
+		goto out;
 
-	if ((sched_policy == KFD_SCHED_POLICY_NO_HWS) && (qpd->vmid != 0))
+	if ((dqm->sched_policy == KFD_SCHED_POLICY_NO_HWS) && (qpd->vmid != 0))
 		program_sh_mem_settings(dqm, qpd);
 
 	pr_debug("sh_mem_config: 0x%x, ape1_base: 0x%x, ape1_limit: 0x%x\n",
@@ -1084,11 +2632,277 @@ static bool set_cache_memory_policy(struct device_queue_manager *dqm,
 		qpd->sh_mem_ape1_limit);
 
 out:
-	mutex_unlock(&dqm->lock);
+	dqm_unlock(dqm);
+	return retval;
+}
+
+static int process_termination_nocpsch(struct device_queue_manager *dqm,
+		struct qcm_process_device *qpd)
+{
+	struct queue *q;
+	struct device_process_node *cur, *next_dpn;
+	int retval = 0;
+	bool found = false;
+
+	dqm_lock(dqm);
+
+	/* Clear all user mode queues */
+	while (!list_empty(&qpd->queues_list)) {
+		struct mqd_manager *mqd_mgr;
+		int ret;
+
+		q = list_first_entry(&qpd->queues_list, struct queue, list);
+		mqd_mgr = dqm->mqd_mgrs[get_mqd_type_from_queue_type(
+				q->properties.type)];
+		ret = destroy_queue_nocpsch_locked(dqm, qpd, q);
+		if (ret)
+			retval = ret;
+		dqm_unlock(dqm);
+		mqd_mgr->free_mqd(mqd_mgr, q->mqd, q->mqd_mem_obj);
+		dqm_lock(dqm);
+	}
+
+	/* Unregister process */
+	list_for_each_entry_safe(cur, next_dpn, &dqm->queues, list) {
+		if (qpd == cur->qpd) {
+			list_del(&cur->list);
+			kfree(cur);
+			dqm->processes_count--;
+			found = true;
+			break;
+		}
+	}
+
+	dqm_unlock(dqm);
+
+	/* Outside the DQM lock because under the DQM lock we can't do
+	 * reclaim or take other locks that others hold while reclaiming.
+	 */
+	if (found)
+		kfd_dec_compute_active(dqm->dev);
+
+	return retval;
+}
+
+static int get_wave_state(struct device_queue_manager *dqm,
+			  struct queue *q,
+			  void __user *ctl_stack,
+			  u32 *ctl_stack_used_size,
+			  u32 *save_area_used_size)
+{
+	struct mqd_manager *mqd_mgr;
+
+	dqm_lock(dqm);
+
+	mqd_mgr = dqm->mqd_mgrs[KFD_MQD_TYPE_CP];
+
+	if (q->properties.type != KFD_QUEUE_TYPE_COMPUTE ||
+	    q->properties.is_active || !q->device->kfd->cwsr_enabled ||
+	    !mqd_mgr->get_wave_state) {
+		dqm_unlock(dqm);
+		return -EINVAL;
+	}
+
+	dqm_unlock(dqm);
+
+	/*
+	 * get_wave_state is outside the dqm lock to prevent circular locking
+	 * and the queue should be protected against destruction by the process
+	 * lock.
+	 */
+	return mqd_mgr->get_wave_state(mqd_mgr, q->mqd, &q->properties,
+			ctl_stack, ctl_stack_used_size, save_area_used_size);
+}
+
+static void get_queue_checkpoint_info(struct device_queue_manager *dqm,
+			const struct queue *q,
+			u32 *mqd_size,
+			u32 *ctl_stack_size)
+{
+	struct mqd_manager *mqd_mgr;
+	enum KFD_MQD_TYPE mqd_type =
+			get_mqd_type_from_queue_type(q->properties.type);
+
+	dqm_lock(dqm);
+	mqd_mgr = dqm->mqd_mgrs[mqd_type];
+	*mqd_size = mqd_mgr->mqd_size * NUM_XCC(mqd_mgr->dev->xcc_mask);
+	*ctl_stack_size = 0;
+
+	if (q->properties.type == KFD_QUEUE_TYPE_COMPUTE && mqd_mgr->get_checkpoint_info)
+		mqd_mgr->get_checkpoint_info(mqd_mgr, q->mqd, ctl_stack_size);
+
+	dqm_unlock(dqm);
+}
+
+static int checkpoint_mqd(struct device_queue_manager *dqm,
+			  const struct queue *q,
+			  void *mqd,
+			  void *ctl_stack)
+{
+	struct mqd_manager *mqd_mgr;
+	int r = 0;
+	enum KFD_MQD_TYPE mqd_type =
+			get_mqd_type_from_queue_type(q->properties.type);
+
+	dqm_lock(dqm);
+
+	if (q->properties.is_active || !q->device->kfd->cwsr_enabled) {
+		r = -EINVAL;
+		goto dqm_unlock;
+	}
+
+	mqd_mgr = dqm->mqd_mgrs[mqd_type];
+	if (!mqd_mgr->checkpoint_mqd) {
+		r = -EOPNOTSUPP;
+		goto dqm_unlock;
+	}
+
+	mqd_mgr->checkpoint_mqd(mqd_mgr, q->mqd, mqd, ctl_stack);
+
+dqm_unlock:
+	dqm_unlock(dqm);
+	return r;
+}
+
+static int process_termination_cpsch(struct device_queue_manager *dqm,
+		struct qcm_process_device *qpd)
+{
+	int retval;
+	struct queue *q;
+	struct device *dev = dqm->dev->adev->dev;
+	struct kernel_queue *kq, *kq_next;
+	struct mqd_manager *mqd_mgr;
+	struct device_process_node *cur, *next_dpn;
+	enum kfd_unmap_queues_filter filter =
+		KFD_UNMAP_QUEUES_FILTER_DYNAMIC_QUEUES;
+	bool found = false;
+
+	retval = 0;
+
+	dqm_lock(dqm);
+
+	/* Clean all kernel queues */
+	list_for_each_entry_safe(kq, kq_next, &qpd->priv_queue_list, list) {
+		list_del(&kq->list);
+		decrement_queue_count(dqm, qpd, kq->queue);
+		qpd->is_debug = false;
+		dqm->total_queue_count--;
+		filter = KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES;
+	}
+
+	/* Clear all user mode queues */
+	list_for_each_entry(q, &qpd->queues_list, list) {
+		if (q->properties.type == KFD_QUEUE_TYPE_SDMA)
+			deallocate_sdma_queue(dqm, q);
+		else if (q->properties.type == KFD_QUEUE_TYPE_SDMA_XGMI)
+			deallocate_sdma_queue(dqm, q);
+
+		if (q->properties.is_active) {
+			decrement_queue_count(dqm, qpd, q);
+
+			if (dqm->dev->kfd->shared_resources.enable_mes) {
+				retval = remove_queue_mes(dqm, q, qpd);
+				if (retval)
+					dev_err(dev, "Failed to remove queue %d\n",
+						q->properties.queue_id);
+			}
+		}
+
+		dqm->total_queue_count--;
+	}
+
+	/* Unregister process */
+	list_for_each_entry_safe(cur, next_dpn, &dqm->queues, list) {
+		if (qpd == cur->qpd) {
+			list_del(&cur->list);
+			kfree(cur);
+			dqm->processes_count--;
+			found = true;
+			break;
+		}
+	}
+
+	if (!dqm->dev->kfd->shared_resources.enable_mes)
+		retval = execute_queues_cpsch(dqm, filter, 0, USE_DEFAULT_GRACE_PERIOD);
+
+	if ((retval || qpd->reset_wavefronts) &&
+	    down_read_trylock(&dqm->dev->adev->reset_domain->sem)) {
+		pr_warn("Resetting wave fronts (cpsch) on dev %p\n", dqm->dev);
+		dbgdev_wave_reset_wavefronts(dqm->dev, qpd->pqm->process);
+		qpd->reset_wavefronts = false;
+		up_read(&dqm->dev->adev->reset_domain->sem);
+	}
+
+	/* Lastly, free mqd resources.
+	 * Do free_mqd() after dqm_unlock to avoid circular locking.
+	 */
+	while (!list_empty(&qpd->queues_list)) {
+		q = list_first_entry(&qpd->queues_list, struct queue, list);
+		mqd_mgr = dqm->mqd_mgrs[get_mqd_type_from_queue_type(
+				q->properties.type)];
+		list_del(&q->list);
+		qpd->queue_count--;
+		dqm_unlock(dqm);
+		mqd_mgr->free_mqd(mqd_mgr, q->mqd, q->mqd_mem_obj);
+		dqm_lock(dqm);
+	}
+	dqm_unlock(dqm);
+
+	/* Outside the DQM lock because under the DQM lock we can't do
+	 * reclaim or take other locks that others hold while reclaiming.
+	 */
+	if (found)
+		kfd_dec_compute_active(dqm->dev);
+
 	return retval;
 }
 
-struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev)
+static int init_mqd_managers(struct device_queue_manager *dqm)
+{
+	int i, j;
+	struct device *dev = dqm->dev->adev->dev;
+	struct mqd_manager *mqd_mgr;
+
+	for (i = 0; i < KFD_MQD_TYPE_MAX; i++) {
+		mqd_mgr = dqm->asic_ops.mqd_manager_init(i, dqm->dev);
+		if (!mqd_mgr) {
+			dev_err(dev, "mqd manager [%d] initialization failed\n", i);
+			goto out_free;
+		}
+		dqm->mqd_mgrs[i] = mqd_mgr;
+	}
+
+	return 0;
+
+out_free:
+	for (j = 0; j < i; j++) {
+		kfree(dqm->mqd_mgrs[j]);
+		dqm->mqd_mgrs[j] = NULL;
+	}
+
+	return -ENOMEM;
+}
+
+/* Allocate one hiq mqd (HWS) and all SDMA mqd in a continuous trunk*/
+static int allocate_hiq_sdma_mqd(struct device_queue_manager *dqm)
+{
+	int retval;
+	struct kfd_node *dev = dqm->dev;
+	struct kfd_mem_obj *mem_obj = &dqm->hiq_sdma_mqd;
+	uint32_t size = dqm->mqd_mgrs[KFD_MQD_TYPE_SDMA]->mqd_size *
+		get_num_all_sdma_engines(dqm) *
+		dev->kfd->device_info.num_sdma_queues_per_engine +
+		(dqm->mqd_mgrs[KFD_MQD_TYPE_HIQ]->mqd_size *
+		NUM_XCC(dqm->dev->xcc_mask));
+
+	retval = amdgpu_amdkfd_alloc_gtt_mem(dev->adev, size,
+		&(mem_obj->gtt_mem), &(mem_obj->gpu_addr),
+		(void *)&(mem_obj->cpu_ptr), false);
+
+	return retval;
+}
+
+struct device_queue_manager *device_queue_manager_init(struct kfd_node *dev)
 {
 	struct device_queue_manager *dqm;
 
@@ -1098,8 +2912,24 @@ struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev)
 	if (!dqm)
 		return NULL;
 
+	switch (dev->adev->asic_type) {
+	/* HWS is not available on Hawaii. */
+	case CHIP_HAWAII:
+	/* HWS depends on CWSR for timely dequeue. CWSR is not
+	 * available on Tonga.
+	 *
+	 * FIXME: This argument also applies to Kaveri.
+	 */
+	case CHIP_TONGA:
+		dqm->sched_policy = KFD_SCHED_POLICY_NO_HWS;
+		break;
+	default:
+		dqm->sched_policy = sched_policy;
+		break;
+	}
+
 	dqm->dev = dev;
-	switch (sched_policy) {
+	switch (dqm->sched_policy) {
 	case KFD_SCHED_POLICY_HWS:
 	case KFD_SCHED_POLICY_HWS_NO_OVERSUBSCRIPTION:
 		/* initialize dqm for cp scheduling */
@@ -1107,15 +2937,23 @@ struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev)
 		dqm->ops.initialize = initialize_cpsch;
 		dqm->ops.start = start_cpsch;
 		dqm->ops.stop = stop_cpsch;
+		dqm->ops.halt = halt_cpsch;
+		dqm->ops.unhalt = unhalt_cpsch;
 		dqm->ops.destroy_queue = destroy_queue_cpsch;
 		dqm->ops.update_queue = update_queue;
-		dqm->ops.get_mqd_manager = get_mqd_manager_nocpsch;
-		dqm->ops.register_process = register_process_nocpsch;
-		dqm->ops.unregister_process = unregister_process_nocpsch;
-		dqm->ops.uninitialize = uninitialize_nocpsch;
+		dqm->ops.register_process = register_process;
+		dqm->ops.unregister_process = unregister_process;
+		dqm->ops.uninitialize = uninitialize;
 		dqm->ops.create_kernel_queue = create_kernel_queue_cpsch;
 		dqm->ops.destroy_kernel_queue = destroy_kernel_queue_cpsch;
 		dqm->ops.set_cache_memory_policy = set_cache_memory_policy;
+		dqm->ops.process_termination = process_termination_cpsch;
+		dqm->ops.evict_process_queues = evict_process_queues_cpsch;
+		dqm->ops.restore_process_queues = restore_process_queues_cpsch;
+		dqm->ops.get_wave_state = get_wave_state;
+		dqm->ops.reset_queues = reset_queues_cpsch;
+		dqm->ops.get_queue_checkpoint_info = get_queue_checkpoint_info;
+		dqm->ops.checkpoint_mqd = checkpoint_mqd;
 		break;
 	case KFD_SCHED_POLICY_NO_HWS:
 		/* initialize dqm for no cp scheduling */
@@ -1124,38 +2962,835 @@ struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev)
 		dqm->ops.create_queue = create_queue_nocpsch;
 		dqm->ops.destroy_queue = destroy_queue_nocpsch;
 		dqm->ops.update_queue = update_queue;
-		dqm->ops.get_mqd_manager = get_mqd_manager_nocpsch;
-		dqm->ops.register_process = register_process_nocpsch;
-		dqm->ops.unregister_process = unregister_process_nocpsch;
+		dqm->ops.register_process = register_process;
+		dqm->ops.unregister_process = unregister_process;
 		dqm->ops.initialize = initialize_nocpsch;
-		dqm->ops.uninitialize = uninitialize_nocpsch;
+		dqm->ops.uninitialize = uninitialize;
 		dqm->ops.set_cache_memory_policy = set_cache_memory_policy;
+		dqm->ops.process_termination = process_termination_nocpsch;
+		dqm->ops.evict_process_queues = evict_process_queues_nocpsch;
+		dqm->ops.restore_process_queues =
+			restore_process_queues_nocpsch;
+		dqm->ops.get_wave_state = get_wave_state;
+		dqm->ops.get_queue_checkpoint_info = get_queue_checkpoint_info;
+		dqm->ops.checkpoint_mqd = checkpoint_mqd;
 		break;
 	default:
-		pr_err("Invalid scheduling policy %d\n", sched_policy);
+		dev_err(dev->adev->dev, "Invalid scheduling policy %d\n", dqm->sched_policy);
 		goto out_free;
 	}
 
-	switch (dev->device_info->asic_family) {
-	case CHIP_CARRIZO:
-		device_queue_manager_init_vi(&dqm->ops_asic_specific);
+	switch (dev->adev->asic_type) {
+	case CHIP_KAVERI:
+	case CHIP_HAWAII:
+		device_queue_manager_init_cik(&dqm->asic_ops);
 		break;
 
-	case CHIP_KAVERI:
-		device_queue_manager_init_cik(&dqm->ops_asic_specific);
+	case CHIP_CARRIZO:
+	case CHIP_TONGA:
+	case CHIP_FIJI:
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
+		device_queue_manager_init_vi(&dqm->asic_ops);
 		break;
+
+	default:
+		if (KFD_GC_VERSION(dev) >= IP_VERSION(12, 0, 0))
+			device_queue_manager_init_v12(&dqm->asic_ops);
+		else if (KFD_GC_VERSION(dev) >= IP_VERSION(11, 0, 0))
+			device_queue_manager_init_v11(&dqm->asic_ops);
+		else if (KFD_GC_VERSION(dev) >= IP_VERSION(10, 1, 1))
+			device_queue_manager_init_v10(&dqm->asic_ops);
+		else if (KFD_GC_VERSION(dev) >= IP_VERSION(9, 0, 1))
+			device_queue_manager_init_v9(&dqm->asic_ops);
+		else {
+			WARN(1, "Unexpected ASIC family %u",
+			     dev->adev->asic_type);
+			goto out_free;
+		}
+	}
+
+	if (init_mqd_managers(dqm))
+		goto out_free;
+
+	if (!dev->kfd->shared_resources.enable_mes && allocate_hiq_sdma_mqd(dqm)) {
+		dev_err(dev->adev->dev, "Failed to allocate hiq sdma mqd trunk buffer\n");
+		goto out_free;
 	}
 
-	if (!dqm->ops.initialize(dqm))
+	if (!dqm->ops.initialize(dqm)) {
+		init_waitqueue_head(&dqm->destroy_wait);
 		return dqm;
+	}
 
 out_free:
 	kfree(dqm);
 	return NULL;
 }
 
+static void deallocate_hiq_sdma_mqd(struct kfd_node *dev,
+				    struct kfd_mem_obj *mqd)
+{
+	WARN(!mqd, "No hiq sdma mqd trunk to free");
+
+	amdgpu_amdkfd_free_gtt_mem(dev->adev, &mqd->gtt_mem);
+}
+
 void device_queue_manager_uninit(struct device_queue_manager *dqm)
 {
+	dqm->ops.stop(dqm);
 	dqm->ops.uninitialize(dqm);
+	if (!dqm->dev->kfd->shared_resources.enable_mes)
+		deallocate_hiq_sdma_mqd(dqm->dev, &dqm->hiq_sdma_mqd);
 	kfree(dqm);
 }
+
+int kfd_dqm_suspend_bad_queue_mes(struct kfd_node *knode, u32 pasid, u32 doorbell_id)
+{
+	struct kfd_process_device *pdd = NULL;
+	struct kfd_process *p = kfd_lookup_process_by_pasid(pasid, &pdd);
+	struct device_queue_manager *dqm = knode->dqm;
+	struct device *dev = dqm->dev->adev->dev;
+	struct qcm_process_device *qpd;
+	struct queue *q = NULL;
+	int ret = 0;
+
+	if (!pdd)
+		return -EINVAL;
+
+	dqm_lock(dqm);
+
+	if (pdd) {
+		qpd = &pdd->qpd;
+
+		list_for_each_entry(q, &qpd->queues_list, list) {
+			if (q->doorbell_id == doorbell_id && q->properties.is_active) {
+				ret = suspend_all_queues_mes(dqm);
+				if (ret) {
+					dev_err(dev, "Suspending all queues failed");
+					goto out;
+				}
+
+				q->properties.is_evicted = true;
+				q->properties.is_active = false;
+				decrement_queue_count(dqm, qpd, q);
+
+				ret = remove_queue_mes(dqm, q, qpd);
+				if (ret) {
+					dev_err(dev, "Removing bad queue failed");
+					goto out;
+				}
+
+				ret = resume_all_queues_mes(dqm);
+				if (ret)
+					dev_err(dev, "Resuming all queues failed");
+
+				break;
+			}
+		}
+	}
+
+out:
+	dqm_unlock(dqm);
+	kfd_unref_process(p);
+	return ret;
+}
+
+static int kfd_dqm_evict_pasid_mes(struct device_queue_manager *dqm,
+				   struct qcm_process_device *qpd)
+{
+	struct device *dev = dqm->dev->adev->dev;
+	int ret = 0;
+
+	/* Check if process is already evicted */
+	dqm_lock(dqm);
+	if (qpd->evicted) {
+		/* Increment the evicted count to make sure the
+		 * process stays evicted before its terminated.
+		 */
+		qpd->evicted++;
+		dqm_unlock(dqm);
+		goto out;
+	}
+	dqm_unlock(dqm);
+
+	ret = suspend_all_queues_mes(dqm);
+	if (ret) {
+		dev_err(dev, "Suspending all queues failed");
+		goto out;
+	}
+
+	ret = dqm->ops.evict_process_queues(dqm, qpd);
+	if (ret) {
+		dev_err(dev, "Evicting process queues failed");
+		goto out;
+	}
+
+	ret = resume_all_queues_mes(dqm);
+	if (ret)
+		dev_err(dev, "Resuming all queues failed");
+
+out:
+	return ret;
+}
+
+int kfd_evict_process_device(struct kfd_process_device *pdd)
+{
+	struct device_queue_manager *dqm;
+	struct kfd_process *p;
+	int ret = 0;
+
+	p = pdd->process;
+	dqm = pdd->dev->dqm;
+
+	WARN(debug_evictions, "Evicting pid %d", p->lead_thread->pid);
+
+	if (dqm->dev->kfd->shared_resources.enable_mes)
+		ret = kfd_dqm_evict_pasid_mes(dqm, &pdd->qpd);
+	else
+		ret = dqm->ops.evict_process_queues(dqm, &pdd->qpd);
+
+	return ret;
+}
+
+int reserve_debug_trap_vmid(struct device_queue_manager *dqm,
+				struct qcm_process_device *qpd)
+{
+	int r;
+	struct device *dev = dqm->dev->adev->dev;
+	int updated_vmid_mask;
+
+	if (dqm->sched_policy == KFD_SCHED_POLICY_NO_HWS) {
+		dev_err(dev, "Unsupported on sched_policy: %i\n", dqm->sched_policy);
+		return -EINVAL;
+	}
+
+	dqm_lock(dqm);
+
+	if (dqm->trap_debug_vmid != 0) {
+		dev_err(dev, "Trap debug id already reserved\n");
+		r = -EBUSY;
+		goto out_unlock;
+	}
+
+	r = unmap_queues_cpsch(dqm, KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES, 0,
+			USE_DEFAULT_GRACE_PERIOD, false);
+	if (r)
+		goto out_unlock;
+
+	updated_vmid_mask = dqm->dev->kfd->shared_resources.compute_vmid_bitmap;
+	updated_vmid_mask &= ~(1 << dqm->dev->vm_info.last_vmid_kfd);
+
+	dqm->dev->kfd->shared_resources.compute_vmid_bitmap = updated_vmid_mask;
+	dqm->trap_debug_vmid = dqm->dev->vm_info.last_vmid_kfd;
+	r = set_sched_resources(dqm);
+	if (r)
+		goto out_unlock;
+
+	r = map_queues_cpsch(dqm);
+	if (r)
+		goto out_unlock;
+
+	pr_debug("Reserved VMID for trap debug: %i\n", dqm->trap_debug_vmid);
+
+out_unlock:
+	dqm_unlock(dqm);
+	return r;
+}
+
+/*
+ * Releases vmid for the trap debugger
+ */
+int release_debug_trap_vmid(struct device_queue_manager *dqm,
+			struct qcm_process_device *qpd)
+{
+	struct device *dev = dqm->dev->adev->dev;
+	int r;
+	int updated_vmid_mask;
+	uint32_t trap_debug_vmid;
+
+	if (dqm->sched_policy == KFD_SCHED_POLICY_NO_HWS) {
+		dev_err(dev, "Unsupported on sched_policy: %i\n", dqm->sched_policy);
+		return -EINVAL;
+	}
+
+	dqm_lock(dqm);
+	trap_debug_vmid = dqm->trap_debug_vmid;
+	if (dqm->trap_debug_vmid == 0) {
+		dev_err(dev, "Trap debug id is not reserved\n");
+		r = -EINVAL;
+		goto out_unlock;
+	}
+
+	r = unmap_queues_cpsch(dqm, KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES, 0,
+			USE_DEFAULT_GRACE_PERIOD, false);
+	if (r)
+		goto out_unlock;
+
+	updated_vmid_mask = dqm->dev->kfd->shared_resources.compute_vmid_bitmap;
+	updated_vmid_mask |= (1 << dqm->dev->vm_info.last_vmid_kfd);
+
+	dqm->dev->kfd->shared_resources.compute_vmid_bitmap = updated_vmid_mask;
+	dqm->trap_debug_vmid = 0;
+	r = set_sched_resources(dqm);
+	if (r)
+		goto out_unlock;
+
+	r = map_queues_cpsch(dqm);
+	if (r)
+		goto out_unlock;
+
+	pr_debug("Released VMID for trap debug: %i\n", trap_debug_vmid);
+
+out_unlock:
+	dqm_unlock(dqm);
+	return r;
+}
+
+#define QUEUE_NOT_FOUND		-1
+/* invalidate queue operation in array */
+static void q_array_invalidate(uint32_t num_queues, uint32_t *queue_ids)
+{
+	int i;
+
+	for (i = 0; i < num_queues; i++)
+		queue_ids[i] |= KFD_DBG_QUEUE_INVALID_MASK;
+}
+
+/* find queue index in array */
+static int q_array_get_index(unsigned int queue_id,
+		uint32_t num_queues,
+		uint32_t *queue_ids)
+{
+	int i;
+
+	for (i = 0; i < num_queues; i++)
+		if (queue_id == (queue_ids[i] & ~KFD_DBG_QUEUE_INVALID_MASK))
+			return i;
+
+	return QUEUE_NOT_FOUND;
+}
+
+struct copy_context_work_handler_workarea {
+	struct work_struct copy_context_work;
+	struct kfd_process *p;
+};
+
+static void copy_context_work_handler(struct work_struct *work)
+{
+	struct copy_context_work_handler_workarea *workarea;
+	struct mqd_manager *mqd_mgr;
+	struct queue *q;
+	struct mm_struct *mm;
+	struct kfd_process *p;
+	uint32_t tmp_ctl_stack_used_size, tmp_save_area_used_size;
+	int i;
+
+	workarea = container_of(work,
+			struct copy_context_work_handler_workarea,
+			copy_context_work);
+
+	p = workarea->p;
+	mm = get_task_mm(p->lead_thread);
+
+	if (!mm)
+		return;
+
+	kthread_use_mm(mm);
+	for (i = 0; i < p->n_pdds; i++) {
+		struct kfd_process_device *pdd = p->pdds[i];
+		struct device_queue_manager *dqm = pdd->dev->dqm;
+		struct qcm_process_device *qpd = &pdd->qpd;
+
+		list_for_each_entry(q, &qpd->queues_list, list) {
+			if (q->properties.type != KFD_QUEUE_TYPE_COMPUTE)
+				continue;
+
+			mqd_mgr = dqm->mqd_mgrs[KFD_MQD_TYPE_CP];
+
+			/* We ignore the return value from get_wave_state
+			 * because
+			 * i) right now, it always returns 0, and
+			 * ii) if we hit an error, we would continue to the
+			 *      next queue anyway.
+			 */
+			mqd_mgr->get_wave_state(mqd_mgr,
+					q->mqd,
+					&q->properties,
+					(void __user *)	q->properties.ctx_save_restore_area_address,
+					&tmp_ctl_stack_used_size,
+					&tmp_save_area_used_size);
+		}
+	}
+	kthread_unuse_mm(mm);
+	mmput(mm);
+}
+
+static uint32_t *get_queue_ids(uint32_t num_queues, uint32_t *usr_queue_id_array)
+{
+	size_t array_size = num_queues * sizeof(uint32_t);
+
+	if (!usr_queue_id_array)
+		return NULL;
+
+	return memdup_user(usr_queue_id_array, array_size);
+}
+
+int resume_queues(struct kfd_process *p,
+		uint32_t num_queues,
+		uint32_t *usr_queue_id_array)
+{
+	uint32_t *queue_ids = NULL;
+	int total_resumed = 0;
+	int i;
+
+	if (usr_queue_id_array) {
+		queue_ids = get_queue_ids(num_queues, usr_queue_id_array);
+
+		if (IS_ERR(queue_ids))
+			return PTR_ERR(queue_ids);
+
+		/* mask all queues as invalid.  unmask per successful request */
+		q_array_invalidate(num_queues, queue_ids);
+	}
+
+	for (i = 0; i < p->n_pdds; i++) {
+		struct kfd_process_device *pdd = p->pdds[i];
+		struct device_queue_manager *dqm = pdd->dev->dqm;
+		struct device *dev = dqm->dev->adev->dev;
+		struct qcm_process_device *qpd = &pdd->qpd;
+		struct queue *q;
+		int r, per_device_resumed = 0;
+
+		dqm_lock(dqm);
+
+		/* unmask queues that resume or already resumed as valid */
+		list_for_each_entry(q, &qpd->queues_list, list) {
+			int q_idx = QUEUE_NOT_FOUND;
+
+			if (queue_ids)
+				q_idx = q_array_get_index(
+						q->properties.queue_id,
+						num_queues,
+						queue_ids);
+
+			if (!queue_ids || q_idx != QUEUE_NOT_FOUND) {
+				int err = resume_single_queue(dqm, &pdd->qpd, q);
+
+				if (queue_ids) {
+					if (!err) {
+						queue_ids[q_idx] &=
+							~KFD_DBG_QUEUE_INVALID_MASK;
+					} else {
+						queue_ids[q_idx] |=
+							KFD_DBG_QUEUE_ERROR_MASK;
+						break;
+					}
+				}
+
+				if (dqm->dev->kfd->shared_resources.enable_mes) {
+					wake_up_all(&dqm->destroy_wait);
+					if (!err)
+						total_resumed++;
+				} else {
+					per_device_resumed++;
+				}
+			}
+		}
+
+		if (!per_device_resumed) {
+			dqm_unlock(dqm);
+			continue;
+		}
+
+		r = execute_queues_cpsch(dqm,
+					KFD_UNMAP_QUEUES_FILTER_DYNAMIC_QUEUES,
+					0,
+					USE_DEFAULT_GRACE_PERIOD);
+		if (r) {
+			dev_err(dev, "Failed to resume process queues\n");
+			if (queue_ids) {
+				list_for_each_entry(q, &qpd->queues_list, list) {
+					int q_idx = q_array_get_index(
+							q->properties.queue_id,
+							num_queues,
+							queue_ids);
+
+					/* mask queue as error on resume fail */
+					if (q_idx != QUEUE_NOT_FOUND)
+						queue_ids[q_idx] |=
+							KFD_DBG_QUEUE_ERROR_MASK;
+				}
+			}
+		} else {
+			wake_up_all(&dqm->destroy_wait);
+			total_resumed += per_device_resumed;
+		}
+
+		dqm_unlock(dqm);
+	}
+
+	if (queue_ids) {
+		if (copy_to_user((void __user *)usr_queue_id_array, queue_ids,
+				num_queues * sizeof(uint32_t)))
+			pr_err("copy_to_user failed on queue resume\n");
+
+		kfree(queue_ids);
+	}
+
+	return total_resumed;
+}
+
+int suspend_queues(struct kfd_process *p,
+			uint32_t num_queues,
+			uint32_t grace_period,
+			uint64_t exception_clear_mask,
+			uint32_t *usr_queue_id_array)
+{
+	uint32_t *queue_ids = get_queue_ids(num_queues, usr_queue_id_array);
+	int total_suspended = 0;
+	int i;
+
+	if (IS_ERR(queue_ids))
+		return PTR_ERR(queue_ids);
+
+	/* mask all queues as invalid.  umask on successful request */
+	q_array_invalidate(num_queues, queue_ids);
+
+	for (i = 0; i < p->n_pdds; i++) {
+		struct kfd_process_device *pdd = p->pdds[i];
+		struct device_queue_manager *dqm = pdd->dev->dqm;
+		struct device *dev = dqm->dev->adev->dev;
+		struct qcm_process_device *qpd = &pdd->qpd;
+		struct queue *q;
+		int r, per_device_suspended = 0;
+
+		mutex_lock(&p->event_mutex);
+		dqm_lock(dqm);
+
+		/* unmask queues that suspend or already suspended */
+		list_for_each_entry(q, &qpd->queues_list, list) {
+			int q_idx = q_array_get_index(q->properties.queue_id,
+							num_queues,
+							queue_ids);
+
+			if (q_idx != QUEUE_NOT_FOUND) {
+				int err = suspend_single_queue(dqm, pdd, q);
+				bool is_mes = dqm->dev->kfd->shared_resources.enable_mes;
+
+				if (!err) {
+					queue_ids[q_idx] &= ~KFD_DBG_QUEUE_INVALID_MASK;
+					if (exception_clear_mask && is_mes)
+						q->properties.exception_status &=
+							~exception_clear_mask;
+
+					if (is_mes)
+						total_suspended++;
+					else
+						per_device_suspended++;
+				} else if (err != -EBUSY) {
+					r = err;
+					queue_ids[q_idx] |= KFD_DBG_QUEUE_ERROR_MASK;
+					break;
+				}
+			}
+		}
+
+		if (!per_device_suspended) {
+			dqm_unlock(dqm);
+			mutex_unlock(&p->event_mutex);
+			if (total_suspended)
+				amdgpu_amdkfd_debug_mem_fence(dqm->dev->adev);
+			continue;
+		}
+
+		r = execute_queues_cpsch(dqm,
+			KFD_UNMAP_QUEUES_FILTER_DYNAMIC_QUEUES, 0,
+			grace_period);
+
+		if (r)
+			dev_err(dev, "Failed to suspend process queues.\n");
+		else
+			total_suspended += per_device_suspended;
+
+		list_for_each_entry(q, &qpd->queues_list, list) {
+			int q_idx = q_array_get_index(q->properties.queue_id,
+						num_queues, queue_ids);
+
+			if (q_idx == QUEUE_NOT_FOUND)
+				continue;
+
+			/* mask queue as error on suspend fail */
+			if (r)
+				queue_ids[q_idx] |= KFD_DBG_QUEUE_ERROR_MASK;
+			else if (exception_clear_mask)
+				q->properties.exception_status &=
+							~exception_clear_mask;
+		}
+
+		dqm_unlock(dqm);
+		mutex_unlock(&p->event_mutex);
+		amdgpu_device_flush_hdp(dqm->dev->adev, NULL);
+	}
+
+	if (total_suspended) {
+		struct copy_context_work_handler_workarea copy_context_worker;
+
+		INIT_WORK_ONSTACK(
+				&copy_context_worker.copy_context_work,
+				copy_context_work_handler);
+
+		copy_context_worker.p = p;
+
+		schedule_work(&copy_context_worker.copy_context_work);
+
+
+		flush_work(&copy_context_worker.copy_context_work);
+		destroy_work_on_stack(&copy_context_worker.copy_context_work);
+	}
+
+	if (copy_to_user((void __user *)usr_queue_id_array, queue_ids,
+			num_queues * sizeof(uint32_t)))
+		pr_err("copy_to_user failed on queue suspend\n");
+
+	kfree(queue_ids);
+
+	return total_suspended;
+}
+
+static uint32_t set_queue_type_for_user(struct queue_properties *q_props)
+{
+	switch (q_props->type) {
+	case KFD_QUEUE_TYPE_COMPUTE:
+		return q_props->format == KFD_QUEUE_FORMAT_PM4
+					? KFD_IOC_QUEUE_TYPE_COMPUTE
+					: KFD_IOC_QUEUE_TYPE_COMPUTE_AQL;
+	case KFD_QUEUE_TYPE_SDMA:
+		return KFD_IOC_QUEUE_TYPE_SDMA;
+	case KFD_QUEUE_TYPE_SDMA_XGMI:
+		return KFD_IOC_QUEUE_TYPE_SDMA_XGMI;
+	default:
+		WARN_ONCE(true, "queue type not recognized!");
+		return 0xffffffff;
+	};
+}
+
+void set_queue_snapshot_entry(struct queue *q,
+			      uint64_t exception_clear_mask,
+			      struct kfd_queue_snapshot_entry *qss_entry)
+{
+	qss_entry->ring_base_address = q->properties.queue_address;
+	qss_entry->write_pointer_address = (uint64_t)q->properties.write_ptr;
+	qss_entry->read_pointer_address = (uint64_t)q->properties.read_ptr;
+	qss_entry->ctx_save_restore_address =
+				q->properties.ctx_save_restore_area_address;
+	qss_entry->ctx_save_restore_area_size =
+				q->properties.ctx_save_restore_area_size;
+	qss_entry->exception_status = q->properties.exception_status;
+	qss_entry->queue_id = q->properties.queue_id;
+	qss_entry->gpu_id = q->device->id;
+	qss_entry->ring_size = (uint32_t)q->properties.queue_size;
+	qss_entry->queue_type = set_queue_type_for_user(&q->properties);
+	q->properties.exception_status &= ~exception_clear_mask;
+}
+
+int debug_lock_and_unmap(struct device_queue_manager *dqm)
+{
+	struct device *dev = dqm->dev->adev->dev;
+	int r;
+
+	if (dqm->sched_policy == KFD_SCHED_POLICY_NO_HWS) {
+		dev_err(dev, "Unsupported on sched_policy: %i\n", dqm->sched_policy);
+		return -EINVAL;
+	}
+
+	if (!kfd_dbg_is_per_vmid_supported(dqm->dev))
+		return 0;
+
+	dqm_lock(dqm);
+
+	r = unmap_queues_cpsch(dqm, KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES, 0, 0, false);
+	if (r)
+		dqm_unlock(dqm);
+
+	return r;
+}
+
+int debug_map_and_unlock(struct device_queue_manager *dqm)
+{
+	struct device *dev = dqm->dev->adev->dev;
+	int r;
+
+	if (dqm->sched_policy == KFD_SCHED_POLICY_NO_HWS) {
+		dev_err(dev, "Unsupported on sched_policy: %i\n", dqm->sched_policy);
+		return -EINVAL;
+	}
+
+	if (!kfd_dbg_is_per_vmid_supported(dqm->dev))
+		return 0;
+
+	r = map_queues_cpsch(dqm);
+
+	dqm_unlock(dqm);
+
+	return r;
+}
+
+int debug_refresh_runlist(struct device_queue_manager *dqm)
+{
+	int r = debug_lock_and_unmap(dqm);
+
+	if (r)
+		return r;
+
+	return debug_map_and_unlock(dqm);
+}
+
+bool kfd_dqm_is_queue_in_process(struct device_queue_manager *dqm,
+				 struct qcm_process_device *qpd,
+				 int doorbell_off, u32 *queue_format)
+{
+	struct queue *q;
+	bool r = false;
+
+	if (!queue_format)
+		return r;
+
+	dqm_lock(dqm);
+
+	list_for_each_entry(q, &qpd->queues_list, list) {
+		if (q->properties.doorbell_off == doorbell_off) {
+			*queue_format = q->properties.format;
+			r = true;
+			goto out;
+		}
+	}
+
+out:
+	dqm_unlock(dqm);
+	return r;
+}
+#if defined(CONFIG_DEBUG_FS)
+
+static void seq_reg_dump(struct seq_file *m,
+			 uint32_t (*dump)[2], uint32_t n_regs)
+{
+	uint32_t i, count;
+
+	for (i = 0, count = 0; i < n_regs; i++) {
+		if (count == 0 ||
+		    dump[i-1][0] + sizeof(uint32_t) != dump[i][0]) {
+			seq_printf(m, "%s    %08x: %08x",
+				   i ? "\n" : "",
+				   dump[i][0], dump[i][1]);
+			count = 7;
+		} else {
+			seq_printf(m, " %08x", dump[i][1]);
+			count--;
+		}
+	}
+
+	seq_puts(m, "\n");
+}
+
+int dqm_debugfs_hqds(struct seq_file *m, void *data)
+{
+	struct device_queue_manager *dqm = data;
+	uint32_t xcc_mask = dqm->dev->xcc_mask;
+	uint32_t (*dump)[2], n_regs;
+	int pipe, queue;
+	int r = 0, xcc_id;
+	uint32_t sdma_engine_start;
+
+	if (!dqm->sched_running) {
+		seq_puts(m, " Device is stopped\n");
+		return 0;
+	}
+
+	for_each_inst(xcc_id, xcc_mask) {
+		r = dqm->dev->kfd2kgd->hqd_dump(dqm->dev->adev,
+						KFD_CIK_HIQ_PIPE,
+						KFD_CIK_HIQ_QUEUE, &dump,
+						&n_regs, xcc_id);
+		if (!r) {
+			seq_printf(
+				m,
+				"   Inst %d, HIQ on MEC %d Pipe %d Queue %d\n",
+				xcc_id,
+				KFD_CIK_HIQ_PIPE / get_pipes_per_mec(dqm) + 1,
+				KFD_CIK_HIQ_PIPE % get_pipes_per_mec(dqm),
+				KFD_CIK_HIQ_QUEUE);
+			seq_reg_dump(m, dump, n_regs);
+
+			kfree(dump);
+		}
+
+		for (pipe = 0; pipe < get_pipes_per_mec(dqm); pipe++) {
+			int pipe_offset = pipe * get_queues_per_pipe(dqm);
+
+			for (queue = 0; queue < get_queues_per_pipe(dqm); queue++) {
+				if (!test_bit(pipe_offset + queue,
+				      dqm->dev->kfd->shared_resources.cp_queue_bitmap))
+					continue;
+
+				r = dqm->dev->kfd2kgd->hqd_dump(dqm->dev->adev,
+								pipe, queue,
+								&dump, &n_regs,
+								xcc_id);
+				if (r)
+					break;
+
+				seq_printf(m,
+					   " Inst %d,  CP Pipe %d, Queue %d\n",
+					   xcc_id, pipe, queue);
+				seq_reg_dump(m, dump, n_regs);
+
+				kfree(dump);
+			}
+		}
+	}
+
+	sdma_engine_start = dqm->dev->node_id * get_num_all_sdma_engines(dqm);
+	for (pipe = sdma_engine_start;
+	     pipe < (sdma_engine_start + get_num_all_sdma_engines(dqm));
+	     pipe++) {
+		for (queue = 0;
+		     queue < dqm->dev->kfd->device_info.num_sdma_queues_per_engine;
+		     queue++) {
+			r = dqm->dev->kfd2kgd->hqd_sdma_dump(
+				dqm->dev->adev, pipe, queue, &dump, &n_regs);
+			if (r)
+				break;
+
+			seq_printf(m, "  SDMA Engine %d, RLC %d\n",
+				  pipe, queue);
+			seq_reg_dump(m, dump, n_regs);
+
+			kfree(dump);
+		}
+	}
+
+	return r;
+}
+
+int dqm_debugfs_hang_hws(struct device_queue_manager *dqm)
+{
+	int r = 0;
+
+	dqm_lock(dqm);
+	r = pm_debugfs_hang_hws(&dqm->packet_mgr);
+	if (r) {
+		dqm_unlock(dqm);
+		return r;
+	}
+	dqm->active_runlist = true;
+	r = execute_queues_cpsch(dqm, KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES,
+				0, USE_DEFAULT_GRACE_PERIOD);
+	dqm_unlock(dqm);
+
+	return r;
+}
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h
index faf820a06400..74a61b5b2f0b 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h
@@ -1,5 +1,6 @@
+/* SPDX-License-Identifier: GPL-2.0 OR MIT */
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -26,23 +27,57 @@
 
 #include <linux/rwsem.h>
 #include <linux/list.h>
+#include <linux/mutex.h>
+#include <linux/sched/mm.h>
 #include "kfd_priv.h"
 #include "kfd_mqd_manager.h"
 
-#define QUEUE_PREEMPT_DEFAULT_TIMEOUT_MS	(500)
-#define CIK_VMID_NUM				(8)
-#define KFD_VMID_START_OFFSET			(8)
-#define VMID_PER_DEVICE				CIK_VMID_NUM
-#define KFD_DQM_FIRST_PIPE			(0)
-#define CIK_SDMA_QUEUES				(4)
-#define CIK_SDMA_QUEUES_PER_ENGINE		(2)
-#define CIK_SDMA_ENGINE_NUM			(2)
+
+#define VMID_NUM 16
+
+#define KFD_MES_PROCESS_QUANTUM		100000
+#define KFD_MES_GANG_QUANTUM		10000
 
 struct device_process_node {
 	struct qcm_process_device *qpd;
 	struct list_head list;
 };
 
+union SQ_CMD_BITS {
+	struct {
+		uint32_t cmd:3;
+		uint32_t:1;
+		uint32_t mode:3;
+		uint32_t check_vmid:1;
+		uint32_t trap_id:3;
+		uint32_t:5;
+		uint32_t wave_id:4;
+		uint32_t simd_id:2;
+		uint32_t:2;
+		uint32_t queue_id:3;
+		uint32_t:1;
+		uint32_t vm_id:4;
+	} bitfields, bits;
+	uint32_t u32All;
+	signed int i32All;
+	float f32All;
+};
+
+union GRBM_GFX_INDEX_BITS {
+	struct {
+		uint32_t instance_index:8;
+		uint32_t sh_index:8;
+		uint32_t se_index:8;
+		uint32_t:5;
+		uint32_t sh_broadcast_writes:1;
+		uint32_t instance_broadcast_writes:1;
+		uint32_t se_broadcast_writes:1;
+	} bitfields, bits;
+	uint32_t u32All;
+	signed int i32All;
+	float f32All;
+};
+
 /**
  * struct device_queue_manager_ops
  *
@@ -52,8 +87,6 @@ struct device_process_node {
  *
  * @update_queue: Queue update routine.
  *
- * @get_mqd_manager: Returns the mqd manager according to the mqd type.
- *
  * @exeute_queues: Dispatches the queues list to the H/W.
  *
  * @register_process: This routine associates a specific process with device.
@@ -62,7 +95,7 @@ struct device_process_node {
  *
  * @initialize: Initializes the pipelines and memory module for that device.
  *
- * @start: Initializes the resources/modules the the device needs for queues
+ * @start: Initializes the resources/modules the device needs for queues
  * execution. This function is called on device initialization and after the
  * system woke up after suspension.
  *
@@ -72,6 +105,12 @@ struct device_process_node {
  * @uninitialize: Destroys all the device queue manager resources allocated in
  * initialize routine.
  *
+ * @halt: This routine unmaps queues from runlist and set halt status to true
+ * so no more queues will be mapped to runlist until unhalt.
+ *
+ * @unhalt: This routine unset halt status to flase and maps queues back to
+ * runlist.
+ *
  * @create_kernel_queue: Creates kernel queue. Used for debug queue.
  *
  * @destroy_kernel_queue: Destroys kernel queue. Used for debug queue.
@@ -79,24 +118,35 @@ struct device_process_node {
  * @set_cache_memory_policy: Sets memory policy (cached/ non cached) for the
  * memory apertures.
  *
+ * @process_termination: Clears all process queues belongs to that device.
+ *
+ * @evict_process_queues: Evict all active queues of a process
+ *
+ * @restore_process_queues: Restore all evicted queues of a process
+ *
+ * @get_wave_state: Retrieves context save state and optionally copies the
+ * control stack, if kept in the MQD, to the given userspace address.
+ *
+ * @reset_queues: reset queues which consume RAS poison
+ * @get_queue_checkpoint_info: Retrieves queue size information for CRIU checkpoint.
+ *
+ * @checkpoint_mqd: checkpoint queue MQD contents for CRIU.
  */
 
 struct device_queue_manager_ops {
 	int	(*create_queue)(struct device_queue_manager *dqm,
 				struct queue *q,
 				struct qcm_process_device *qpd,
-				int *allocate_vmid);
+				const struct kfd_criu_queue_priv_data *qd,
+				const void *restore_mqd,
+				const void *restore_ctl_stack);
 
 	int	(*destroy_queue)(struct device_queue_manager *dqm,
 				struct qcm_process_device *qpd,
 				struct queue *q);
 
 	int	(*update_queue)(struct device_queue_manager *dqm,
-				struct queue *q);
-
-	struct mqd_manager * (*get_mqd_manager)
-					(struct device_queue_manager *dqm,
-					enum KFD_MQD_TYPE type);
+				struct queue *q, struct mqd_update_info *minfo);
 
 	int	(*register_process)(struct device_queue_manager *dqm,
 					struct qcm_process_device *qpd);
@@ -108,6 +158,8 @@ struct device_queue_manager_ops {
 	int	(*start)(struct device_queue_manager *dqm);
 	int	(*stop)(struct device_queue_manager *dqm);
 	void	(*uninitialize)(struct device_queue_manager *dqm);
+	int     (*halt)(struct device_queue_manager *dqm);
+	int     (*unhalt)(struct device_queue_manager *dqm);
 	int	(*create_kernel_queue)(struct device_queue_manager *dqm,
 					struct kernel_queue *kq,
 					struct qcm_process_device *qpd);
@@ -121,22 +173,57 @@ struct device_queue_manager_ops {
 					   enum cache_policy default_policy,
 					   enum cache_policy alternate_policy,
 					   void __user *alternate_aperture_base,
-					   uint64_t alternate_aperture_size);
+					   uint64_t alternate_aperture_size,
+					   u32 misc_process_properties);
+
+	int (*process_termination)(struct device_queue_manager *dqm,
+			struct qcm_process_device *qpd);
+
+	int (*evict_process_queues)(struct device_queue_manager *dqm,
+				    struct qcm_process_device *qpd);
+	int (*restore_process_queues)(struct device_queue_manager *dqm,
+				      struct qcm_process_device *qpd);
+
+	int	(*get_wave_state)(struct device_queue_manager *dqm,
+				  struct queue *q,
+				  void __user *ctl_stack,
+				  u32 *ctl_stack_used_size,
+				  u32 *save_area_used_size);
+
+	int (*reset_queues)(struct device_queue_manager *dqm,
+					uint16_t pasid);
+	void	(*get_queue_checkpoint_info)(struct device_queue_manager *dqm,
+				  const struct queue *q, u32 *mqd_size,
+				  u32 *ctl_stack_size);
+
+	int	(*checkpoint_mqd)(struct device_queue_manager *dqm,
+				  const struct queue *q,
+				  void *mqd,
+				  void *ctl_stack);
 };
 
 struct device_queue_manager_asic_ops {
-	int	(*register_process)(struct device_queue_manager *dqm,
+	int	(*update_qpd)(struct device_queue_manager *dqm,
 					struct qcm_process_device *qpd);
-	int	(*initialize)(struct device_queue_manager *dqm);
 	bool	(*set_cache_memory_policy)(struct device_queue_manager *dqm,
 					   struct qcm_process_device *qpd,
 					   enum cache_policy default_policy,
 					   enum cache_policy alternate_policy,
 					   void __user *alternate_aperture_base,
-					   uint64_t alternate_aperture_size);
+					   uint64_t alternate_aperture_size,
+					   u32 misc_process_properties);
 	void	(*init_sdma_vm)(struct device_queue_manager *dqm,
 				struct queue *q,
 				struct qcm_process_device *qpd);
+	struct mqd_manager *	(*mqd_manager_init)(enum KFD_MQD_TYPE type,
+				 struct kfd_node *dev);
+};
+
+struct dqm_detect_hang_info {
+	int pipe_id;
+	int queue_id;
+	int xcc_id;
+	uint64_t queue_address;
 };
 
 /**
@@ -153,36 +240,93 @@ struct device_queue_manager_asic_ops {
 
 struct device_queue_manager {
 	struct device_queue_manager_ops ops;
-	struct device_queue_manager_asic_ops ops_asic_specific;
+	struct device_queue_manager_asic_ops asic_ops;
 
-	struct mqd_manager	*mqds[KFD_MQD_TYPE_MAX];
-	struct packet_manager	packets;
-	struct kfd_dev		*dev;
-	struct mutex		lock;
+	struct mqd_manager	*mqd_mgrs[KFD_MQD_TYPE_MAX];
+	struct packet_manager	packet_mgr;
+	struct kfd_node		*dev;
+	struct mutex		lock_hidden; /* use dqm_lock/unlock(dqm) */
 	struct list_head	queues;
+	unsigned int		saved_flags;
 	unsigned int		processes_count;
-	unsigned int		queue_count;
-	unsigned int		sdma_queue_count;
+	unsigned int		active_queue_count;
+	unsigned int		active_cp_queue_count;
+	unsigned int		gws_queue_count;
 	unsigned int		total_queue_count;
 	unsigned int		next_pipe_to_allocate;
 	unsigned int		*allocated_queues;
-	unsigned int		sdma_bitmap;
-	unsigned int		vmid_bitmap;
+	DECLARE_BITMAP(sdma_bitmap, KFD_MAX_SDMA_QUEUES);
+	DECLARE_BITMAP(xgmi_sdma_bitmap, KFD_MAX_SDMA_QUEUES);
+	/* the pasid mapping for each kfd vmid */
+	uint16_t		vmid_pasid[VMID_NUM];
 	uint64_t		pipelines_addr;
-	struct kfd_mem_obj	*pipeline_mem;
 	uint64_t		fence_gpu_addr;
-	unsigned int		*fence_addr;
+	uint64_t		*fence_addr;
 	struct kfd_mem_obj	*fence_mem;
 	bool			active_runlist;
+	int			sched_policy;
+	uint32_t		trap_debug_vmid;
+
+	/* hw exception  */
+	bool			is_hws_hang;
+	bool			is_resetting;
+	struct kfd_mem_obj	hiq_sdma_mqd;
+	bool			sched_running;
+	bool			sched_halt;
+
+	/* used for GFX 9.4.3 only */
+	uint32_t		current_logical_xcc_start;
+
+	uint32_t		wait_times;
+
+	wait_queue_head_t	destroy_wait;
+
+	/* for per-queue reset support */
+	struct dqm_detect_hang_info *detect_hang_info;
+	size_t detect_hang_info_size;
+	int detect_hang_count;
 };
 
-void device_queue_manager_init_cik(struct device_queue_manager_asic_ops *ops);
-void device_queue_manager_init_vi(struct device_queue_manager_asic_ops *ops);
+void device_queue_manager_init_cik(
+		struct device_queue_manager_asic_ops *asic_ops);
+void device_queue_manager_init_vi(
+		struct device_queue_manager_asic_ops *asic_ops);
+void device_queue_manager_init_v9(
+		struct device_queue_manager_asic_ops *asic_ops);
+void device_queue_manager_init_v10(
+		struct device_queue_manager_asic_ops *asic_ops);
+void device_queue_manager_init_v11(
+		struct device_queue_manager_asic_ops *asic_ops);
+void device_queue_manager_init_v12(
+		struct device_queue_manager_asic_ops *asic_ops);
 void program_sh_mem_settings(struct device_queue_manager *dqm,
 					struct qcm_process_device *qpd);
-unsigned int get_queues_num(struct device_queue_manager *dqm);
+unsigned int get_cp_queues_num(struct device_queue_manager *dqm);
 unsigned int get_queues_per_pipe(struct device_queue_manager *dqm);
 unsigned int get_pipes_per_mec(struct device_queue_manager *dqm);
+unsigned int get_num_sdma_queues(struct device_queue_manager *dqm);
+unsigned int get_num_xgmi_sdma_queues(struct device_queue_manager *dqm);
+int reserve_debug_trap_vmid(struct device_queue_manager *dqm,
+			struct qcm_process_device *qpd);
+int release_debug_trap_vmid(struct device_queue_manager *dqm,
+			struct qcm_process_device *qpd);
+int suspend_queues(struct kfd_process *p,
+			uint32_t num_queues,
+			uint32_t grace_period,
+			uint64_t exception_clear_mask,
+			uint32_t *usr_queue_id_array);
+int resume_queues(struct kfd_process *p,
+		uint32_t num_queues,
+		uint32_t *usr_queue_id_array);
+void set_queue_snapshot_entry(struct queue *q,
+			      uint64_t exception_clear_mask,
+			      struct kfd_queue_snapshot_entry *qss_entry);
+int debug_lock_and_unmap(struct device_queue_manager *dqm);
+int debug_map_and_unlock(struct device_queue_manager *dqm);
+int debug_refresh_runlist(struct device_queue_manager *dqm);
+bool kfd_dqm_is_queue_in_process(struct device_queue_manager *dqm,
+				 struct qcm_process_device *qpd,
+				 int doorbell_off, u32 *queue_format);
 
 static inline unsigned int get_sh_mem_bases_32(struct kfd_process_device *pdd)
 {
@@ -195,4 +339,34 @@ get_sh_mem_bases_nybble_64(struct kfd_process_device *pdd)
 	return (pdd->lds_base >> 60) & 0x0E;
 }
 
+/* The DQM lock can be taken in MMU notifiers. Make sure no reclaim-FS
+ * happens while holding this lock anywhere to prevent deadlocks when
+ * an MMU notifier runs in reclaim-FS context.
+ */
+static inline void dqm_lock(struct device_queue_manager *dqm)
+{
+	mutex_lock(&dqm->lock_hidden);
+	dqm->saved_flags = memalloc_noreclaim_save();
+}
+static inline void dqm_unlock(struct device_queue_manager *dqm)
+{
+	memalloc_noreclaim_restore(dqm->saved_flags);
+	mutex_unlock(&dqm->lock_hidden);
+}
+
+static inline int read_sdma_queue_counter(uint64_t __user *q_rptr, uint64_t *val)
+{
+	/* SDMA activity counter is stored at queue's RPTR + 0x8 location. */
+	return get_user(*val, q_rptr + 1);
+}
+
+static inline void update_dqm_wait_times(struct device_queue_manager *dqm)
+{
+	if (dqm->dev->kfd2kgd->get_iq_wait_times)
+		dqm->dev->kfd2kgd->get_iq_wait_times(dqm->dev->adev,
+					&dqm->wait_times,
+					ffs(dqm->dev->xcc_mask) - 1);
+}
+
+
 #endif /* KFD_DEVICE_QUEUE_MANAGER_H_ */
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_cik.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_cik.c
index 72c3cbabc0a7..0508ef5a41d7 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_cik.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_cik.c
@@ -1,5 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -26,24 +27,34 @@
 #include "oss/oss_2_4_sh_mask.h"
 #include "gca/gfx_7_2_sh_mask.h"
 
+/*
+ * Low bits must be 0000/FFFF as required by HW, high bits must be 0 to
+ * stay in user mode.
+ */
+#define APE1_FIXED_BITS_MASK 0xFFFF80000000FFFFULL
+/* APE1 limit is inclusive and 64K aligned. */
+#define APE1_LIMIT_ALIGNMENT 0xFFFF
+
 static bool set_cache_memory_policy_cik(struct device_queue_manager *dqm,
 				   struct qcm_process_device *qpd,
 				   enum cache_policy default_policy,
 				   enum cache_policy alternate_policy,
 				   void __user *alternate_aperture_base,
-				   uint64_t alternate_aperture_size);
-static int register_process_cik(struct device_queue_manager *dqm,
-					struct qcm_process_device *qpd);
-static int initialize_cpsch_cik(struct device_queue_manager *dqm);
-static void init_sdma_vm(struct device_queue_manager *dqm, struct queue *q,
-				struct qcm_process_device *qpd);
-
-void device_queue_manager_init_cik(struct device_queue_manager_asic_ops *ops)
+				   uint64_t alternate_aperture_size,
+				   u32 misc_process_properties);
+static int update_qpd_cik(struct device_queue_manager *dqm,
+			  struct qcm_process_device *qpd);
+static void init_sdma_vm(struct device_queue_manager *dqm,
+			 struct queue *q,
+			 struct qcm_process_device *qpd);
+
+void device_queue_manager_init_cik(
+	struct device_queue_manager_asic_ops *asic_ops)
 {
-	ops->set_cache_memory_policy = set_cache_memory_policy_cik;
-	ops->register_process = register_process_cik;
-	ops->initialize = initialize_cpsch_cik;
-	ops->init_sdma_vm = init_sdma_vm;
+	asic_ops->set_cache_memory_policy = set_cache_memory_policy_cik;
+	asic_ops->update_qpd = update_qpd_cik;
+	asic_ops->init_sdma_vm = init_sdma_vm;
+	asic_ops->mqd_manager_init = mqd_manager_init_cik;
 }
 
 static uint32_t compute_sh_mem_bases_64bit(unsigned int top_address_nybble)
@@ -78,10 +89,41 @@ static bool set_cache_memory_policy_cik(struct device_queue_manager *dqm,
 				   enum cache_policy default_policy,
 				   enum cache_policy alternate_policy,
 				   void __user *alternate_aperture_base,
-				   uint64_t alternate_aperture_size)
+				   uint64_t alternate_aperture_size,
+				   u32 misc_process_properties)
 {
 	uint32_t default_mtype;
 	uint32_t ape1_mtype;
+	unsigned int temp;
+	bool retval = true;
+
+	if (alternate_aperture_size == 0) {
+		/* base > limit disables APE1 */
+		qpd->sh_mem_ape1_base = 1;
+		qpd->sh_mem_ape1_limit = 0;
+	} else {
+		/*
+		 * In FSA64, APE1_Base[63:0] = { 16{SH_MEM_APE1_BASE[31]},
+		 *			SH_MEM_APE1_BASE[31:0], 0x0000 }
+		 * APE1_Limit[63:0] = { 16{SH_MEM_APE1_LIMIT[31]},
+		 *			SH_MEM_APE1_LIMIT[31:0], 0xFFFF }
+		 * Verify that the base and size parameters can be
+		 * represented in this format and convert them.
+		 * Additionally restrict APE1 to user-mode addresses.
+		 */
+
+		uint64_t base = (uintptr_t)alternate_aperture_base;
+		uint64_t limit = base + alternate_aperture_size - 1;
+
+		if (limit <= base || (base & APE1_FIXED_BITS_MASK) != 0 ||
+		   (limit & APE1_FIXED_BITS_MASK) != APE1_LIMIT_ALIGNMENT) {
+			retval = false;
+			goto out;
+		}
+
+		qpd->sh_mem_ape1_base = base >> 16;
+		qpd->sh_mem_ape1_limit = limit >> 16;
+	}
 
 	default_mtype = (default_policy == cache_policy_coherent) ?
 			MTYPE_NONCACHED :
@@ -95,61 +137,34 @@ static bool set_cache_memory_policy_cik(struct device_queue_manager *dqm,
 			| ALIGNMENT_MODE(SH_MEM_ALIGNMENT_MODE_UNALIGNED)
 			| DEFAULT_MTYPE(default_mtype)
 			| APE1_MTYPE(ape1_mtype);
-
-	return true;
-}
-
-static int register_process_cik(struct device_queue_manager *dqm,
-		struct qcm_process_device *qpd)
-{
-	struct kfd_process_device *pdd;
-	unsigned int temp;
-
-	pdd = qpd_to_pdd(qpd);
-
-	/* check if sh_mem_config register already configured */
-	if (qpd->sh_mem_config == 0) {
-		qpd->sh_mem_config =
-			ALIGNMENT_MODE(SH_MEM_ALIGNMENT_MODE_UNALIGNED) |
-			DEFAULT_MTYPE(MTYPE_NONCACHED) |
-			APE1_MTYPE(MTYPE_NONCACHED);
-		qpd->sh_mem_ape1_limit = 0;
-		qpd->sh_mem_ape1_base = 0;
-	}
-
-	if (qpd->pqm->process->is_32bit_user_mode) {
-		temp = get_sh_mem_bases_32(pdd);
-		qpd->sh_mem_bases = SHARED_BASE(temp);
-		qpd->sh_mem_config |= PTR32;
-	} else {
-		temp = get_sh_mem_bases_nybble_64(pdd);
-		qpd->sh_mem_bases = compute_sh_mem_bases_64bit(temp);
-		qpd->sh_mem_config |= 1  << SH_MEM_CONFIG__PRIVATE_ATC__SHIFT;
-	}
+	/* On dGPU we're always in GPUVM64 addressing mode with 64-bit
+	 * aperture addresses.
+	 */
+	temp = get_sh_mem_bases_nybble_64(qpd_to_pdd(qpd));
+	qpd->sh_mem_bases = compute_sh_mem_bases_64bit(temp);
 
 	pr_debug("is32bit process: %d sh_mem_bases nybble: 0x%X and register 0x%X\n",
 		qpd->pqm->process->is_32bit_user_mode, temp, qpd->sh_mem_bases);
 
-	return 0;
+out:
+	return retval;
 }
 
-static void init_sdma_vm(struct device_queue_manager *dqm, struct queue *q,
-				struct qcm_process_device *qpd)
+static int update_qpd_cik(struct device_queue_manager *dqm,
+			  struct qcm_process_device *qpd)
 {
-	uint32_t value = (1 << SDMA0_RLC0_VIRTUAL_ADDR__ATC__SHIFT);
-
-	if (q->process->is_32bit_user_mode)
-		value |= (1 << SDMA0_RLC0_VIRTUAL_ADDR__PTR32__SHIFT) |
-				get_sh_mem_bases_32(qpd_to_pdd(qpd));
-	else
-		value |= ((get_sh_mem_bases_nybble_64(qpd_to_pdd(qpd))) <<
-				SDMA0_RLC0_VIRTUAL_ADDR__SHARED_BASE__SHIFT) &
-				SDMA0_RLC0_VIRTUAL_ADDR__SHARED_BASE_MASK;
-
-	q->properties.sdma_vm_addr = value;
+	return 0;
 }
 
-static int initialize_cpsch_cik(struct device_queue_manager *dqm)
+static void init_sdma_vm(struct device_queue_manager *dqm,
+			 struct queue *q,
+			 struct qcm_process_device *qpd)
 {
-	return 0;
+	/* On dGPU we're always in GPUVM64 addressing mode with 64-bit
+	 * aperture addresses.
+	 */
+	q->properties.sdma_vm_addr =
+		((get_sh_mem_bases_nybble_64(qpd_to_pdd(qpd))) <<
+		 SDMA0_RLC0_VIRTUAL_ADDR__SHARED_BASE__SHIFT) &
+		SDMA0_RLC0_VIRTUAL_ADDR__SHARED_BASE_MASK;
 }
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v10.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v10.c
new file mode 100644
index 000000000000..ba6e3d747ccd
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v10.c
@@ -0,0 +1,90 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
+/*
+ * Copyright 2018-2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "kfd_device_queue_manager.h"
+#include "navi10_enum.h"
+#include "gc/gc_10_1_0_offset.h"
+#include "gc/gc_10_1_0_sh_mask.h"
+
+static int update_qpd_v10(struct device_queue_manager *dqm,
+			 struct qcm_process_device *qpd);
+static void init_sdma_vm_v10(struct device_queue_manager *dqm, struct queue *q,
+			    struct qcm_process_device *qpd);
+static bool set_cache_memory_policy_v10(struct device_queue_manager *dqm,
+				   struct qcm_process_device *qpd,
+				   enum cache_policy default_policy,
+				   enum cache_policy alternate_policy,
+				   void __user *alternate_aperture_base,
+				   uint64_t alternate_aperture_size,
+				   u32 misc_process_properties);
+
+void device_queue_manager_init_v10(
+	struct device_queue_manager_asic_ops *asic_ops)
+{
+	asic_ops->set_cache_memory_policy = set_cache_memory_policy_v10;
+	asic_ops->update_qpd = update_qpd_v10;
+	asic_ops->init_sdma_vm = init_sdma_vm_v10;
+	asic_ops->mqd_manager_init = mqd_manager_init_v10;
+}
+
+static uint32_t compute_sh_mem_bases_64bit(struct kfd_process_device *pdd)
+{
+	uint32_t shared_base = pdd->lds_base >> 48;
+	uint32_t private_base = pdd->scratch_base >> 48;
+
+	return (shared_base << SH_MEM_BASES__SHARED_BASE__SHIFT) |
+		private_base;
+}
+
+static bool set_cache_memory_policy_v10(struct device_queue_manager *dqm,
+				   struct qcm_process_device *qpd,
+				   enum cache_policy default_policy,
+				   enum cache_policy alternate_policy,
+				   void __user *alternate_aperture_base,
+				   uint64_t alternate_aperture_size,
+				   u32 misc_process_properties)
+{
+	qpd->sh_mem_config = (SH_MEM_ALIGNMENT_MODE_UNALIGNED <<
+			      SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) |
+			      (3 << SH_MEM_CONFIG__INITIAL_INST_PREFETCH__SHIFT);
+	qpd->sh_mem_ape1_limit = 0;
+	qpd->sh_mem_ape1_base = 0;
+	qpd->sh_mem_bases = compute_sh_mem_bases_64bit(qpd_to_pdd(qpd));
+
+	pr_debug("sh_mem_bases 0x%X\n", qpd->sh_mem_bases);
+	return true;
+}
+
+static int update_qpd_v10(struct device_queue_manager *dqm,
+			 struct qcm_process_device *qpd)
+{
+	return 0;
+}
+
+static void init_sdma_vm_v10(struct device_queue_manager *dqm, struct queue *q,
+			    struct qcm_process_device *qpd)
+{
+	/* Not needed on SDMAv4 onwards any more */
+	q->properties.sdma_vm_addr = 0;
+}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v11.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v11.c
new file mode 100644
index 000000000000..8b447d04558f
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v11.c
@@ -0,0 +1,90 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "kfd_device_queue_manager.h"
+#include "gc/gc_11_0_0_offset.h"
+#include "gc/gc_11_0_0_sh_mask.h"
+#include "soc21_enum.h"
+
+static int update_qpd_v11(struct device_queue_manager *dqm,
+			 struct qcm_process_device *qpd);
+static void init_sdma_vm_v11(struct device_queue_manager *dqm, struct queue *q,
+			    struct qcm_process_device *qpd);
+static bool set_cache_memory_policy_v11(struct device_queue_manager *dqm,
+				   struct qcm_process_device *qpd,
+				   enum cache_policy default_policy,
+				   enum cache_policy alternate_policy,
+				   void __user *alternate_aperture_base,
+				   uint64_t alternate_aperture_size,
+				   u32 misc_process_properties);
+
+void device_queue_manager_init_v11(
+	struct device_queue_manager_asic_ops *asic_ops)
+{
+	asic_ops->set_cache_memory_policy = set_cache_memory_policy_v11;
+	asic_ops->update_qpd = update_qpd_v11;
+	asic_ops->init_sdma_vm = init_sdma_vm_v11;
+	asic_ops->mqd_manager_init = mqd_manager_init_v11;
+}
+
+static uint32_t compute_sh_mem_bases_64bit(struct kfd_process_device *pdd)
+{
+	uint32_t shared_base = pdd->lds_base >> 48;
+	uint32_t private_base = pdd->scratch_base >> 48;
+
+	return (shared_base << SH_MEM_BASES__SHARED_BASE__SHIFT) |
+		private_base;
+}
+
+static bool set_cache_memory_policy_v11(struct device_queue_manager *dqm,
+				   struct qcm_process_device *qpd,
+				   enum cache_policy default_policy,
+				   enum cache_policy alternate_policy,
+				   void __user *alternate_aperture_base,
+				   uint64_t alternate_aperture_size,
+				   u32 misc_process_properties)
+{
+	qpd->sh_mem_config = (SH_MEM_ALIGNMENT_MODE_UNALIGNED <<
+			      SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) |
+			      (3 << SH_MEM_CONFIG__INITIAL_INST_PREFETCH__SHIFT);
+
+	qpd->sh_mem_ape1_limit = 0;
+	qpd->sh_mem_ape1_base = 0;
+	qpd->sh_mem_bases = compute_sh_mem_bases_64bit(qpd_to_pdd(qpd));
+
+	pr_debug("sh_mem_bases 0x%X\n", qpd->sh_mem_bases);
+	return true;
+}
+
+static int update_qpd_v11(struct device_queue_manager *dqm,
+			 struct qcm_process_device *qpd)
+{
+	return 0;
+}
+
+static void init_sdma_vm_v11(struct device_queue_manager *dqm, struct queue *q,
+			    struct qcm_process_device *qpd)
+{
+	/* Not needed on SDMAv4 onwards any more */
+	q->properties.sdma_vm_addr = 0;
+}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v12.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v12.c
new file mode 100644
index 000000000000..3550da3a46f9
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v12.c
@@ -0,0 +1,90 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "kfd_device_queue_manager.h"
+#include "gc/gc_12_0_0_sh_mask.h"
+#include "soc24_enum.h"
+
+static int update_qpd_v12(struct device_queue_manager *dqm,
+			 struct qcm_process_device *qpd);
+static void init_sdma_vm_v12(struct device_queue_manager *dqm, struct queue *q,
+			    struct qcm_process_device *qpd);
+static bool set_cache_memory_policy_v12(struct device_queue_manager *dqm,
+				   struct qcm_process_device *qpd,
+				   enum cache_policy default_policy,
+				   enum cache_policy alternate_policy,
+				   void __user *alternate_aperture_base,
+				   uint64_t alternate_aperture_size,
+				   u32 misc_process_properties);
+
+void device_queue_manager_init_v12(
+	struct device_queue_manager_asic_ops *asic_ops)
+{
+	asic_ops->set_cache_memory_policy = set_cache_memory_policy_v12;
+	asic_ops->update_qpd = update_qpd_v12;
+	asic_ops->init_sdma_vm = init_sdma_vm_v12;
+	asic_ops->mqd_manager_init = mqd_manager_init_v12;
+}
+
+static uint32_t compute_sh_mem_bases_64bit(struct kfd_process_device *pdd)
+{
+	uint32_t shared_base = pdd->lds_base >> 48;
+	uint32_t private_base = pdd->scratch_base >> 48;
+
+	return (shared_base << SH_MEM_BASES__SHARED_BASE__SHIFT) |
+		private_base;
+}
+
+static bool set_cache_memory_policy_v12(struct device_queue_manager *dqm,
+				   struct qcm_process_device *qpd,
+				   enum cache_policy default_policy,
+				   enum cache_policy alternate_policy,
+				   void __user *alternate_aperture_base,
+				   uint64_t alternate_aperture_size,
+				   u32 misc_process_properties)
+{
+	qpd->sh_mem_config = (SH_MEM_ALIGNMENT_MODE_UNALIGNED <<
+			      SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) |
+			      (3 << SH_MEM_CONFIG__INITIAL_INST_PREFETCH__SHIFT);
+
+	qpd->sh_mem_ape1_limit = 0;
+	qpd->sh_mem_ape1_base = 0;
+	qpd->sh_mem_bases = compute_sh_mem_bases_64bit(qpd_to_pdd(qpd));
+
+	pr_debug("sh_mem_bases 0x%X\n", qpd->sh_mem_bases);
+	return true;
+}
+
+static int update_qpd_v12(struct device_queue_manager *dqm,
+			 struct qcm_process_device *qpd)
+{
+	return 0;
+}
+
+static void init_sdma_vm_v12(struct device_queue_manager *dqm, struct queue *q,
+			    struct qcm_process_device *qpd)
+{
+	/* Not needed on SDMAv4 onwards any more */
+	q->properties.sdma_vm_addr = 0;
+}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c
new file mode 100644
index 000000000000..9fcc8c6e57b7
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c
@@ -0,0 +1,135 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
+/*
+ * Copyright 2016-2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "kfd_device_queue_manager.h"
+#include "vega10_enum.h"
+#include "gc/gc_9_4_3_sh_mask.h"
+
+static int update_qpd_v9(struct device_queue_manager *dqm,
+			 struct qcm_process_device *qpd);
+static void init_sdma_vm_v9(struct device_queue_manager *dqm, struct queue *q,
+			    struct qcm_process_device *qpd);
+static bool set_cache_memory_policy_v9(struct device_queue_manager *dqm,
+				   struct qcm_process_device *qpd,
+				   enum cache_policy default_policy,
+				   enum cache_policy alternate_policy,
+				   void __user *alternate_aperture_base,
+				   uint64_t alternate_aperture_size,
+				   u32 misc_process_properties);
+
+void device_queue_manager_init_v9(
+	struct device_queue_manager_asic_ops *asic_ops)
+{
+	asic_ops->set_cache_memory_policy = set_cache_memory_policy_v9;
+	asic_ops->update_qpd = update_qpd_v9;
+	asic_ops->init_sdma_vm = init_sdma_vm_v9;
+	asic_ops->mqd_manager_init = mqd_manager_init_v9;
+}
+
+static uint32_t compute_sh_mem_bases_64bit(struct kfd_process_device *pdd)
+{
+	uint32_t shared_base = pdd->lds_base >> 48;
+	uint32_t private_base = pdd->scratch_base >> 48;
+
+	return (shared_base << SH_MEM_BASES__SHARED_BASE__SHIFT) |
+		private_base;
+}
+
+static bool set_cache_memory_policy_v9(struct device_queue_manager *dqm,
+				   struct qcm_process_device *qpd,
+				   enum cache_policy default_policy,
+				   enum cache_policy alternate_policy,
+				   void __user *alternate_aperture_base,
+				   uint64_t alternate_aperture_size,
+				   u32 misc_process_properties)
+{
+	qpd->sh_mem_config = SH_MEM_ALIGNMENT_MODE_UNALIGNED <<
+				SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT;
+
+	if (dqm->dev->kfd->noretry)
+		qpd->sh_mem_config |= 1 << SH_MEM_CONFIG__RETRY_DISABLE__SHIFT;
+
+	if (KFD_GC_VERSION(dqm->dev->kfd) == IP_VERSION(9, 4, 3) ||
+		KFD_GC_VERSION(dqm->dev->kfd) == IP_VERSION(9, 4, 4))
+		qpd->sh_mem_config |= (1 << SH_MEM_CONFIG__F8_MODE__SHIFT);
+
+	if (KFD_GC_VERSION(dqm->dev->kfd) == IP_VERSION(9, 5, 0)) {
+		if (misc_process_properties & KFD_PROC_FLAG_MFMA_HIGH_PRECISION)
+			qpd->sh_mem_config |= 1 << SH_MEM_CONFIG__PRECISION_MODE__SHIFT;
+	}
+
+	qpd->sh_mem_ape1_limit = 0;
+	qpd->sh_mem_ape1_base = 0;
+	qpd->sh_mem_bases = compute_sh_mem_bases_64bit(qpd_to_pdd(qpd));
+
+	pr_debug("sh_mem_bases 0x%X sh_mem_config 0x%X\n", qpd->sh_mem_bases,
+		 qpd->sh_mem_config);
+	return true;
+}
+
+static int update_qpd_v9(struct device_queue_manager *dqm,
+			 struct qcm_process_device *qpd)
+{
+	struct kfd_process_device *pdd = qpd_to_pdd(qpd);
+
+	pdd = qpd_to_pdd(qpd);
+
+	/* check if sh_mem_config register already configured */
+	if (qpd->sh_mem_config == 0) {
+		qpd->sh_mem_config = SH_MEM_ALIGNMENT_MODE_UNALIGNED <<
+					SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT;
+
+		if (dqm->dev->kfd->noretry)
+			qpd->sh_mem_config |= 1 << SH_MEM_CONFIG__RETRY_DISABLE__SHIFT;
+
+		if (KFD_GC_VERSION(dqm->dev->kfd) == IP_VERSION(9, 4, 3) ||
+		    KFD_GC_VERSION(dqm->dev->kfd) == IP_VERSION(9, 4, 4))
+			qpd->sh_mem_config |=
+				(1 << SH_MEM_CONFIG__F8_MODE__SHIFT);
+
+		qpd->sh_mem_ape1_limit = 0;
+		qpd->sh_mem_ape1_base = 0;
+	}
+
+	if (KFD_SUPPORT_XNACK_PER_PROCESS(dqm->dev)) {
+		if (!pdd->process->xnack_enabled)
+			qpd->sh_mem_config |= 1 << SH_MEM_CONFIG__RETRY_DISABLE__SHIFT;
+		else
+			qpd->sh_mem_config &= ~(1 << SH_MEM_CONFIG__RETRY_DISABLE__SHIFT);
+	}
+
+	qpd->sh_mem_bases = compute_sh_mem_bases_64bit(pdd);
+
+	pr_debug("sh_mem_bases 0x%X sh_mem_config 0x%X\n", qpd->sh_mem_bases,
+		 qpd->sh_mem_config);
+
+	return 0;
+}
+
+static void init_sdma_vm_v9(struct device_queue_manager *dqm, struct queue *q,
+			    struct qcm_process_device *qpd)
+{
+	/* Not needed on SDMAv4 any more */
+	q->properties.sdma_vm_addr = 0;
+}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_vi.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_vi.c
index 40e9ddd096cd..dad83356e976 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_vi.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_vi.c
@@ -1,5 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -24,27 +25,36 @@
 #include "kfd_device_queue_manager.h"
 #include "gca/gfx_8_0_enum.h"
 #include "gca/gfx_8_0_sh_mask.h"
-#include "gca/gfx_8_0_enum.h"
 #include "oss/oss_3_0_sh_mask.h"
 
+/*
+ * Low bits must be 0000/FFFF as required by HW, high bits must be 0 to
+ * stay in user mode.
+ */
+#define APE1_FIXED_BITS_MASK 0xFFFF80000000FFFFULL
+/* APE1 limit is inclusive and 64K aligned. */
+#define APE1_LIMIT_ALIGNMENT 0xFFFF
+
 static bool set_cache_memory_policy_vi(struct device_queue_manager *dqm,
-				   struct qcm_process_device *qpd,
-				   enum cache_policy default_policy,
-				   enum cache_policy alternate_policy,
-				   void __user *alternate_aperture_base,
-				   uint64_t alternate_aperture_size);
-static int register_process_vi(struct device_queue_manager *dqm,
-					struct qcm_process_device *qpd);
-static int initialize_cpsch_vi(struct device_queue_manager *dqm);
-static void init_sdma_vm(struct device_queue_manager *dqm, struct queue *q,
-				struct qcm_process_device *qpd);
-
-void device_queue_manager_init_vi(struct device_queue_manager_asic_ops *ops)
+				       struct qcm_process_device *qpd,
+				       enum cache_policy default_policy,
+				       enum cache_policy alternate_policy,
+				       void __user *alternate_aperture_base,
+				       uint64_t alternate_aperture_size,
+				       u32 misc_process_properties);
+static int update_qpd_vi(struct device_queue_manager *dqm,
+			 struct qcm_process_device *qpd);
+static void init_sdma_vm(struct device_queue_manager *dqm,
+			 struct queue *q,
+			 struct qcm_process_device *qpd);
+
+void device_queue_manager_init_vi(
+	struct device_queue_manager_asic_ops *asic_ops)
 {
-	ops->set_cache_memory_policy = set_cache_memory_policy_vi;
-	ops->register_process = register_process_vi;
-	ops->initialize = initialize_cpsch_vi;
-	ops->init_sdma_vm = init_sdma_vm;
+	asic_ops->set_cache_memory_policy = set_cache_memory_policy_vi;
+	asic_ops->update_qpd = update_qpd_vi;
+	asic_ops->init_sdma_vm = init_sdma_vm;
+	asic_ops->mqd_manager_init = mqd_manager_init_vi;
 }
 
 static uint32_t compute_sh_mem_bases_64bit(unsigned int top_address_nybble)
@@ -76,92 +86,87 @@ static uint32_t compute_sh_mem_bases_64bit(unsigned int top_address_nybble)
 }
 
 static bool set_cache_memory_policy_vi(struct device_queue_manager *dqm,
-				   struct qcm_process_device *qpd,
-				   enum cache_policy default_policy,
-				   enum cache_policy alternate_policy,
-				   void __user *alternate_aperture_base,
-				   uint64_t alternate_aperture_size)
+		struct qcm_process_device *qpd,
+		enum cache_policy default_policy,
+		enum cache_policy alternate_policy,
+		void __user *alternate_aperture_base,
+		uint64_t alternate_aperture_size,
+		u32 misc_process_properties)
 {
 	uint32_t default_mtype;
 	uint32_t ape1_mtype;
+	unsigned int temp;
+	bool retval = true;
+
+	if (alternate_aperture_size == 0) {
+		/* base > limit disables APE1 */
+		qpd->sh_mem_ape1_base = 1;
+		qpd->sh_mem_ape1_limit = 0;
+	} else {
+		/*
+		 * In FSA64, APE1_Base[63:0] = { 16{SH_MEM_APE1_BASE[31]},
+		 *			SH_MEM_APE1_BASE[31:0], 0x0000 }
+		 * APE1_Limit[63:0] = { 16{SH_MEM_APE1_LIMIT[31]},
+		 *			SH_MEM_APE1_LIMIT[31:0], 0xFFFF }
+		 * Verify that the base and size parameters can be
+		 * represented in this format and convert them.
+		 * Additionally restrict APE1 to user-mode addresses.
+		 */
+
+		uint64_t base = (uintptr_t)alternate_aperture_base;
+		uint64_t limit = base + alternate_aperture_size - 1;
+
+		if (limit <= base || (base & APE1_FIXED_BITS_MASK) != 0 ||
+		   (limit & APE1_FIXED_BITS_MASK) != APE1_LIMIT_ALIGNMENT) {
+			retval = false;
+			goto out;
+		}
+
+		qpd->sh_mem_ape1_base = base >> 16;
+		qpd->sh_mem_ape1_limit = limit >> 16;
+	}
 
 	default_mtype = (default_policy == cache_policy_coherent) ?
-			MTYPE_CC :
+			MTYPE_UC :
 			MTYPE_NC;
 
 	ape1_mtype = (alternate_policy == cache_policy_coherent) ?
-			MTYPE_CC :
+			MTYPE_UC :
 			MTYPE_NC;
 
-	qpd->sh_mem_config = (qpd->sh_mem_config &
-			SH_MEM_CONFIG__ADDRESS_MODE_MASK) |
-		SH_MEM_ALIGNMENT_MODE_UNALIGNED <<
-				SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT |
-		default_mtype << SH_MEM_CONFIG__DEFAULT_MTYPE__SHIFT |
-		ape1_mtype << SH_MEM_CONFIG__APE1_MTYPE__SHIFT |
-		SH_MEM_CONFIG__PRIVATE_ATC_MASK;
-
-	return true;
-}
-
-static int register_process_vi(struct device_queue_manager *dqm,
-					struct qcm_process_device *qpd)
-{
-	struct kfd_process_device *pdd;
-	unsigned int temp;
-
-	pdd = qpd_to_pdd(qpd);
-
-	/* check if sh_mem_config register already configured */
-	if (qpd->sh_mem_config == 0) {
-		qpd->sh_mem_config =
+	qpd->sh_mem_config =
 			SH_MEM_ALIGNMENT_MODE_UNALIGNED <<
-				SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT |
-			MTYPE_CC << SH_MEM_CONFIG__DEFAULT_MTYPE__SHIFT |
-			MTYPE_CC << SH_MEM_CONFIG__APE1_MTYPE__SHIFT |
-			SH_MEM_CONFIG__PRIVATE_ATC_MASK;
-
-		qpd->sh_mem_ape1_limit = 0;
-		qpd->sh_mem_ape1_base = 0;
-	}
-
-	if (qpd->pqm->process->is_32bit_user_mode) {
-		temp = get_sh_mem_bases_32(pdd);
-		qpd->sh_mem_bases = temp << SH_MEM_BASES__SHARED_BASE__SHIFT;
-		qpd->sh_mem_config |= SH_MEM_ADDRESS_MODE_HSA32 <<
-					SH_MEM_CONFIG__ADDRESS_MODE__SHIFT;
-	} else {
-		temp = get_sh_mem_bases_nybble_64(pdd);
-		qpd->sh_mem_bases = compute_sh_mem_bases_64bit(temp);
-		qpd->sh_mem_config |= SH_MEM_ADDRESS_MODE_HSA64 <<
-			SH_MEM_CONFIG__ADDRESS_MODE__SHIFT;
-		qpd->sh_mem_config |= 1  <<
-			SH_MEM_CONFIG__PRIVATE_ATC__SHIFT;
-	}
+				   SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT |
+			default_mtype << SH_MEM_CONFIG__DEFAULT_MTYPE__SHIFT |
+			ape1_mtype << SH_MEM_CONFIG__APE1_MTYPE__SHIFT;
 
-	pr_debug("is32bit process: %d sh_mem_bases nybble: 0x%X and register 0x%X\n",
-		qpd->pqm->process->is_32bit_user_mode, temp, qpd->sh_mem_bases);
+	/* On dGPU we're always in GPUVM64 addressing mode with 64-bit
+	 * aperture addresses.
+	 */
+	temp = get_sh_mem_bases_nybble_64(qpd_to_pdd(qpd));
+	qpd->sh_mem_bases = compute_sh_mem_bases_64bit(temp);
 
-	return 0;
+	pr_debug("sh_mem_bases nybble: 0x%X and register 0x%X\n",
+		temp, qpd->sh_mem_bases);
+out:
+	return retval;
 }
 
-static void init_sdma_vm(struct device_queue_manager *dqm, struct queue *q,
-				struct qcm_process_device *qpd)
+static int update_qpd_vi(struct device_queue_manager *dqm,
+			 struct qcm_process_device *qpd)
 {
-	uint32_t value = (1 << SDMA0_RLC0_VIRTUAL_ADDR__ATC__SHIFT);
-
-	if (q->process->is_32bit_user_mode)
-		value |= (1 << SDMA0_RLC0_VIRTUAL_ADDR__PTR32__SHIFT) |
-				get_sh_mem_bases_32(qpd_to_pdd(qpd));
-	else
-		value |= ((get_sh_mem_bases_nybble_64(qpd_to_pdd(qpd))) <<
-				SDMA0_RLC0_VIRTUAL_ADDR__SHARED_BASE__SHIFT) &
-				SDMA0_RLC0_VIRTUAL_ADDR__SHARED_BASE_MASK;
-
-	q->properties.sdma_vm_addr = value;
+	return 0;
 }
 
-static int initialize_cpsch_vi(struct device_queue_manager *dqm)
+static void init_sdma_vm(struct device_queue_manager *dqm,
+			 struct queue *q,
+			 struct qcm_process_device *qpd)
 {
-	return 0;
+	/* On dGPU we're always in GPUVM64 addressing mode with 64-bit
+	 * aperture addresses.
+	 */
+	q->properties.sdma_vm_addr =
+		((get_sh_mem_bases_nybble_64(qpd_to_pdd(qpd))) <<
+		 SDMA0_RLC0_VIRTUAL_ADDR__SHARED_BASE__SHIFT) &
+		SDMA0_RLC0_VIRTUAL_ADDR__SHARED_BASE_MASK;
 }
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_doorbell.c b/drivers/gpu/drm/amd/amdkfd/kfd_doorbell.c
index acf4d2a977ad..05c74887fd6f 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_doorbell.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_doorbell.c
@@ -1,5 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -24,18 +25,13 @@
 #include <linux/mman.h>
 #include <linux/slab.h>
 #include <linux/io.h>
+#include <linux/idr.h>
 
 /*
- * This extension supports a kernel level doorbells management for
- * the kernel queues.
- * Basically the last doorbells page is devoted to kernel queues
- * and that's assures that any user process won't get access to the
- * kernel doorbells page
+ * This extension supports a kernel level doorbells management for the
+ * kernel queues using the first doorbell page reserved for the kernel.
  */
 
-#define KERNEL_DOORBELL_PASID 1
-#define KFD_SIZE_OF_DOORBELL_IN_BYTES 4
-
 /*
  * Each device exposes a doorbell aperture, a PCI MMIO aperture that
  * receives 32-bit writes that are passed to queues as wptr values.
@@ -51,103 +47,85 @@
  */
 
 /* # of doorbell bytes allocated for each process. */
-static inline size_t doorbell_process_allocation(void)
+size_t kfd_doorbell_process_slice(struct kfd_dev *kfd)
 {
-	return roundup(KFD_SIZE_OF_DOORBELL_IN_BYTES *
-			KFD_MAX_NUM_OF_QUEUES_PER_PROCESS,
-			PAGE_SIZE);
+	if (!kfd->shared_resources.enable_mes)
+		return roundup(kfd->device_info.doorbell_size *
+				KFD_MAX_NUM_OF_QUEUES_PER_PROCESS,
+				PAGE_SIZE);
+	else
+		return amdgpu_mes_doorbell_process_slice(
+					(struct amdgpu_device *)kfd->adev);
 }
 
 /* Doorbell calculations for device init. */
 int kfd_doorbell_init(struct kfd_dev *kfd)
 {
-	size_t doorbell_start_offset;
-	size_t doorbell_aperture_size;
-	size_t doorbell_process_limit;
+	int size = PAGE_SIZE;
+	int r;
 
 	/*
-	 * We start with calculations in bytes because the input data might
-	 * only be byte-aligned.
-	 * Only after we have done the rounding can we assume any alignment.
+	 * Todo: KFD kernel level operations need only one doorbell for
+	 * ring test/HWS. So instead of reserving a whole page here for
+	 * kernel, reserve and consume a doorbell from existing KGD kernel
+	 * doorbell page.
 	 */
 
-	doorbell_start_offset =
-			roundup(kfd->shared_resources.doorbell_start_offset,
-					doorbell_process_allocation());
-
-	doorbell_aperture_size =
-			rounddown(kfd->shared_resources.doorbell_aperture_size,
-					doorbell_process_allocation());
-
-	if (doorbell_aperture_size > doorbell_start_offset)
-		doorbell_process_limit =
-			(doorbell_aperture_size - doorbell_start_offset) /
-						doorbell_process_allocation();
-	else
-		doorbell_process_limit = 0;
-
-	kfd->doorbell_base = kfd->shared_resources.doorbell_physical_address +
-				doorbell_start_offset;
-
-	kfd->doorbell_id_offset = doorbell_start_offset / sizeof(u32);
-	kfd->doorbell_process_limit = doorbell_process_limit - 1;
-
-	kfd->doorbell_kernel_ptr = ioremap(kfd->doorbell_base,
-						doorbell_process_allocation());
-
-	if (!kfd->doorbell_kernel_ptr)
+	/* Bitmap to dynamically allocate doorbells from kernel page */
+	kfd->doorbell_bitmap = bitmap_zalloc(size / sizeof(u32), GFP_KERNEL);
+	if (!kfd->doorbell_bitmap) {
+		DRM_ERROR("Failed to allocate kernel doorbell bitmap\n");
 		return -ENOMEM;
+	}
 
-	pr_debug("Doorbell initialization:\n");
-	pr_debug("doorbell base           == 0x%08lX\n",
-			(uintptr_t)kfd->doorbell_base);
-
-	pr_debug("doorbell_id_offset      == 0x%08lX\n",
-			kfd->doorbell_id_offset);
-
-	pr_debug("doorbell_process_limit  == 0x%08lX\n",
-			doorbell_process_limit);
-
-	pr_debug("doorbell_kernel_offset  == 0x%08lX\n",
-			(uintptr_t)kfd->doorbell_base);
-
-	pr_debug("doorbell aperture size  == 0x%08lX\n",
-			kfd->shared_resources.doorbell_aperture_size);
-
-	pr_debug("doorbell kernel address == 0x%08lX\n",
-			(uintptr_t)kfd->doorbell_kernel_ptr);
+	/* Alloc a doorbell page for KFD kernel usages */
+	r = amdgpu_bo_create_kernel(kfd->adev,
+				    size,
+				    PAGE_SIZE,
+				    AMDGPU_GEM_DOMAIN_DOORBELL,
+				    &kfd->doorbells,
+				    NULL,
+				    (void **)&kfd->doorbell_kernel_ptr);
+	if (r) {
+		pr_err("failed to allocate kernel doorbells\n");
+		bitmap_free(kfd->doorbell_bitmap);
+		return r;
+	}
 
+	pr_debug("Doorbell kernel address == %p\n", kfd->doorbell_kernel_ptr);
 	return 0;
 }
 
 void kfd_doorbell_fini(struct kfd_dev *kfd)
 {
-	if (kfd->doorbell_kernel_ptr)
-		iounmap(kfd->doorbell_kernel_ptr);
+	bitmap_free(kfd->doorbell_bitmap);
+	amdgpu_bo_free_kernel(&kfd->doorbells, NULL,
+			     (void **)&kfd->doorbell_kernel_ptr);
 }
 
-int kfd_doorbell_mmap(struct kfd_process *process, struct vm_area_struct *vma)
+int kfd_doorbell_mmap(struct kfd_node *dev, struct kfd_process *process,
+		      struct vm_area_struct *vma)
 {
 	phys_addr_t address;
-	struct kfd_dev *dev;
+	struct kfd_process_device *pdd;
 
 	/*
 	 * For simplicitly we only allow mapping of the entire doorbell
 	 * allocation of a single device & process.
 	 */
-	if (vma->vm_end - vma->vm_start != doorbell_process_allocation())
+	if (vma->vm_end - vma->vm_start != kfd_doorbell_process_slice(dev->kfd))
 		return -EINVAL;
 
-	/* Find kfd device according to gpu id */
-	dev = kfd_device_by_id(vma->vm_pgoff);
-	if (!dev)
+	pdd = kfd_get_process_device_data(dev, process);
+	if (!pdd)
 		return -EINVAL;
 
 	/* Calculate physical address of doorbell */
-	address = kfd_get_process_doorbells(dev, process);
-
-	vma->vm_flags |= VM_IO | VM_DONTCOPY | VM_DONTEXPAND | VM_NORESERVE |
-				VM_DONTDUMP | VM_PFNMAP;
+	address = kfd_get_process_doorbells(pdd);
+	if (!address)
+		return -ENOMEM;
+	vm_flags_set(vma, VM_IO | VM_DONTCOPY | VM_DONTEXPAND | VM_NORESERVE |
+				VM_DONTDUMP | VM_PFNMAP);
 
 	vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
 
@@ -157,44 +135,42 @@ int kfd_doorbell_mmap(struct kfd_process *process, struct vm_area_struct *vma)
 		 "     vm_flags            == 0x%04lX\n"
 		 "     size                == 0x%04lX\n",
 		 (unsigned long long) vma->vm_start, address, vma->vm_flags,
-		 doorbell_process_allocation());
+		 kfd_doorbell_process_slice(dev->kfd));
 
 
 	return io_remap_pfn_range(vma,
 				vma->vm_start,
 				address >> PAGE_SHIFT,
-				doorbell_process_allocation(),
+				kfd_doorbell_process_slice(dev->kfd),
 				vma->vm_page_prot);
 }
 
 
 /* get kernel iomem pointer for a doorbell */
-u32 __iomem *kfd_get_kernel_doorbell(struct kfd_dev *kfd,
+void __iomem *kfd_get_kernel_doorbell(struct kfd_dev *kfd,
 					unsigned int *doorbell_off)
 {
 	u32 inx;
 
 	mutex_lock(&kfd->doorbell_mutex);
-	inx = find_first_zero_bit(kfd->doorbell_available_index,
-					KFD_MAX_NUM_OF_QUEUES_PER_PROCESS);
+	inx = find_first_zero_bit(kfd->doorbell_bitmap, PAGE_SIZE / sizeof(u32));
 
-	__set_bit(inx, kfd->doorbell_available_index);
+	__set_bit(inx, kfd->doorbell_bitmap);
 	mutex_unlock(&kfd->doorbell_mutex);
 
 	if (inx >= KFD_MAX_NUM_OF_QUEUES_PER_PROCESS)
 		return NULL;
 
-	/*
-	 * Calculating the kernel doorbell offset using "faked" kernel
-	 * pasid that allocated for kernel queues only
-	 */
-	*doorbell_off = KERNEL_DOORBELL_PASID * (doorbell_process_allocation() /
-							sizeof(u32)) + inx;
+	*doorbell_off = amdgpu_doorbell_index_on_bar(kfd->adev,
+						     kfd->doorbells,
+						     inx,
+						     kfd->device_info.doorbell_size);
+	inx *= 2;
 
 	pr_debug("Get kernel queue doorbell\n"
-			 "     doorbell offset   == 0x%08X\n"
-			 "     kernel address    == 0x%08lX\n",
-		*doorbell_off, (uintptr_t)(kfd->doorbell_kernel_ptr + inx));
+			"     doorbell offset   == 0x%08X\n"
+			"     doorbell index    == 0x%x\n",
+		*doorbell_off, inx);
 
 	return kfd->doorbell_kernel_ptr + inx;
 }
@@ -204,51 +180,124 @@ void kfd_release_kernel_doorbell(struct kfd_dev *kfd, u32 __iomem *db_addr)
 	unsigned int inx;
 
 	inx = (unsigned int)(db_addr - kfd->doorbell_kernel_ptr);
+	inx /= 2;
 
 	mutex_lock(&kfd->doorbell_mutex);
-	__clear_bit(inx, kfd->doorbell_available_index);
+	__clear_bit(inx, kfd->doorbell_bitmap);
 	mutex_unlock(&kfd->doorbell_mutex);
 }
 
-inline void write_kernel_doorbell(u32 __iomem *db, u32 value)
+void write_kernel_doorbell(void __iomem *db, u32 value)
 {
 	if (db) {
 		writel(value, db);
-		pr_debug("Writing %d to doorbell address 0x%p\n", value, db);
+		pr_debug("Writing %d to doorbell address %p\n", value, db);
 	}
 }
 
-/*
- * queue_ids are in the range [0,MAX_PROCESS_QUEUES) and are mapped 1:1
- * to doorbells with the process's doorbell page
- */
-unsigned int kfd_queue_id_to_doorbell(struct kfd_dev *kfd,
-					struct kfd_process *process,
-					unsigned int queue_id)
+void write_kernel_doorbell64(void __iomem *db, u64 value)
 {
-	/*
-	 * doorbell_id_offset accounts for doorbells taken by KGD.
-	 * pasid * doorbell_process_allocation/sizeof(u32) adjusts
-	 * to the process's doorbells
-	 */
-	return kfd->doorbell_id_offset +
-		process->pasid * (doorbell_process_allocation()/sizeof(u32)) +
-		queue_id;
+	if (db) {
+		WARN(((unsigned long)db & 7) != 0,
+		     "Unaligned 64-bit doorbell");
+		writeq(value, (u64 __iomem *)db);
+		pr_debug("writing %llu to doorbell address %p\n", value, db);
+	}
+}
+
+static int init_doorbell_bitmap(struct qcm_process_device *qpd,
+				struct kfd_dev *dev)
+{
+	unsigned int i;
+	int range_start = dev->shared_resources.non_cp_doorbells_start;
+	int range_end = dev->shared_resources.non_cp_doorbells_end;
+
+	if (!KFD_IS_SOC15(dev))
+		return 0;
+
+	/* Mask out doorbells reserved for SDMA, IH, and VCN on SOC15. */
+	pr_debug("reserved doorbell 0x%03x - 0x%03x\n", range_start, range_end);
+	pr_debug("reserved doorbell 0x%03x - 0x%03x\n",
+			range_start + KFD_QUEUE_DOORBELL_MIRROR_OFFSET,
+			range_end + KFD_QUEUE_DOORBELL_MIRROR_OFFSET);
+
+	for (i = 0; i < KFD_MAX_NUM_OF_QUEUES_PER_PROCESS / 2; i++) {
+		if (i >= range_start && i <= range_end) {
+			__set_bit(i, qpd->doorbell_bitmap);
+			__set_bit(i + KFD_QUEUE_DOORBELL_MIRROR_OFFSET,
+				  qpd->doorbell_bitmap);
+		}
+	}
+
+	return 0;
 }
 
-uint64_t kfd_get_number_elems(struct kfd_dev *kfd)
+phys_addr_t kfd_get_process_doorbells(struct kfd_process_device *pdd)
 {
-	uint64_t num_of_elems = (kfd->shared_resources.doorbell_aperture_size -
-				kfd->shared_resources.doorbell_start_offset) /
-					doorbell_process_allocation() + 1;
+	struct amdgpu_device *adev = pdd->dev->adev;
+	uint32_t first_db_index;
 
-	return num_of_elems;
+	if (!pdd->qpd.proc_doorbells) {
+		if (kfd_alloc_process_doorbells(pdd->dev->kfd, pdd))
+			/* phys_addr_t 0 is error */
+			return 0;
+	}
 
+	first_db_index = amdgpu_doorbell_index_on_bar(adev,
+						      pdd->qpd.proc_doorbells,
+						      0,
+						      pdd->dev->kfd->device_info.doorbell_size);
+	return adev->doorbell.base + first_db_index * sizeof(uint32_t);
 }
 
-phys_addr_t kfd_get_process_doorbells(struct kfd_dev *dev,
-					struct kfd_process *process)
+int kfd_alloc_process_doorbells(struct kfd_dev *kfd, struct kfd_process_device *pdd)
 {
-	return dev->doorbell_base +
-		process->pasid * doorbell_process_allocation();
+	int r;
+	struct qcm_process_device *qpd = &pdd->qpd;
+
+	/* Allocate bitmap for dynamic doorbell allocation */
+	qpd->doorbell_bitmap = bitmap_zalloc(KFD_MAX_NUM_OF_QUEUES_PER_PROCESS,
+					     GFP_KERNEL);
+	if (!qpd->doorbell_bitmap) {
+		DRM_ERROR("Failed to allocate process doorbell bitmap\n");
+		return -ENOMEM;
+	}
+
+	r = init_doorbell_bitmap(&pdd->qpd, kfd);
+	if (r) {
+		DRM_ERROR("Failed to initialize process doorbells\n");
+		r = -ENOMEM;
+		goto err;
+	}
+
+	/* Allocate doorbells for this process */
+	r = amdgpu_bo_create_kernel(kfd->adev,
+				    kfd_doorbell_process_slice(kfd),
+				    PAGE_SIZE,
+				    AMDGPU_GEM_DOMAIN_DOORBELL,
+				    &qpd->proc_doorbells,
+				    NULL,
+				    NULL);
+	if (r) {
+		DRM_ERROR("Failed to allocate process doorbells\n");
+		goto err;
+	}
+	return 0;
+
+err:
+	bitmap_free(qpd->doorbell_bitmap);
+	qpd->doorbell_bitmap = NULL;
+	return r;
+}
+
+void kfd_free_process_doorbells(struct kfd_dev *kfd, struct kfd_process_device *pdd)
+{
+	struct qcm_process_device *qpd = &pdd->qpd;
+
+	if (qpd->doorbell_bitmap) {
+		bitmap_free(qpd->doorbell_bitmap);
+		qpd->doorbell_bitmap = NULL;
+	}
+
+	amdgpu_bo_free_kernel(&qpd->proc_doorbells, NULL, NULL);
 }
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_events.c b/drivers/gpu/drm/amd/amdkfd/kfd_events.c
index 5979158c3f7b..82905f3e54dd 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_events.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_events.c
@@ -1,5 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -24,292 +25,189 @@
 #include <linux/slab.h>
 #include <linux/types.h>
 #include <linux/sched/signal.h>
+#include <linux/sched/mm.h>
 #include <linux/uaccess.h>
-#include <linux/mm.h>
 #include <linux/mman.h>
 #include <linux/memory.h>
 #include "kfd_priv.h"
 #include "kfd_events.h"
+#include "kfd_device_queue_manager.h"
 #include <linux/device.h>
 
 /*
- * A task can only be on a single wait_queue at a time, but we need to support
- * waiting on multiple events (any/all).
- * Instead of each event simply having a wait_queue with sleeping tasks, it
- * has a singly-linked list of tasks.
- * A thread that wants to sleep creates an array of these, one for each event
- * and adds one to each event's waiter chain.
+ * Wrapper around wait_queue_entry_t
  */
 struct kfd_event_waiter {
-	struct list_head waiters;
-	struct task_struct *sleeping_task;
-
-	/* Transitions to true when the event this belongs to is signaled. */
-	bool activated;
-
-	/* Event */
-	struct kfd_event *event;
-	uint32_t input_index;
+	wait_queue_entry_t wait;
+	struct kfd_event *event; /* Event to wait for */
+	bool activated;		 /* Becomes true when event is signaled */
+	bool event_age_enabled;  /* set to true when last_event_age is non-zero */
 };
 
 /*
- * Over-complicated pooled allocator for event notification slots.
- *
  * Each signal event needs a 64-bit signal slot where the signaler will write
- * a 1 before sending an interrupt.l (This is needed because some interrupts
+ * a 1 before sending an interrupt. (This is needed because some interrupts
  * do not contain enough spare data bits to identify an event.)
- * We get whole pages from vmalloc and map them to the process VA.
- * Individual signal events are then allocated a slot in a page.
+ * We get whole pages and map them to the process VA.
+ * Individual signal events use their event_id as slot index.
  */
-
-struct signal_page {
-	struct list_head event_pages;	/* kfd_process.signal_event_pages */
+struct kfd_signal_page {
 	uint64_t *kernel_address;
 	uint64_t __user *user_address;
-	uint32_t page_index;		/* Index into the mmap aperture. */
-	unsigned int free_slots;
-	unsigned long used_slot_bitmap[0];
+	bool need_to_free_pages;
 };
 
-#define SLOTS_PER_PAGE KFD_SIGNAL_EVENT_LIMIT
-#define SLOT_BITMAP_SIZE BITS_TO_LONGS(SLOTS_PER_PAGE)
-#define BITS_PER_PAGE (ilog2(SLOTS_PER_PAGE)+1)
-#define SIGNAL_PAGE_SIZE (sizeof(struct signal_page) + \
-				SLOT_BITMAP_SIZE * sizeof(long))
-
-/*
- * For signal events, the event ID is used as the interrupt user data.
- * For SQ s_sendmsg interrupts, this is limited to 8 bits.
- */
-
-#define INTERRUPT_DATA_BITS 8
-#define SIGNAL_EVENT_ID_SLOT_SHIFT 0
-
-static uint64_t *page_slots(struct signal_page *page)
+static uint64_t *page_slots(struct kfd_signal_page *page)
 {
 	return page->kernel_address;
 }
 
-static bool allocate_free_slot(struct kfd_process *process,
-				struct signal_page **out_page,
-				unsigned int *out_slot_index)
-{
-	struct signal_page *page;
-
-	list_for_each_entry(page, &process->signal_event_pages, event_pages) {
-		if (page->free_slots > 0) {
-			unsigned int slot =
-				find_first_zero_bit(page->used_slot_bitmap,
-							SLOTS_PER_PAGE);
-
-			__set_bit(slot, page->used_slot_bitmap);
-			page->free_slots--;
-
-			page_slots(page)[slot] = UNSIGNALED_EVENT_SLOT;
-
-			*out_page = page;
-			*out_slot_index = slot;
-
-			pr_debug("Allocated event signal slot in page %p, slot %d\n",
-					page, slot);
-
-			return true;
-		}
-	}
-
-	pr_debug("No free event signal slots were found for process %p\n",
-			process);
-
-	return false;
-}
-
-#define list_tail_entry(head, type, member) \
-	list_entry((head)->prev, type, member)
-
-static bool allocate_signal_page(struct file *devkfd, struct kfd_process *p)
+static struct kfd_signal_page *allocate_signal_page(struct kfd_process *p)
 {
 	void *backing_store;
-	struct signal_page *page;
+	struct kfd_signal_page *page;
 
-	page = kzalloc(SIGNAL_PAGE_SIZE, GFP_KERNEL);
+	page = kzalloc(sizeof(*page), GFP_KERNEL);
 	if (!page)
-		goto fail_alloc_signal_page;
+		return NULL;
 
-	page->free_slots = SLOTS_PER_PAGE;
-
-	backing_store = (void *) __get_free_pages(GFP_KERNEL | __GFP_ZERO,
+	backing_store = (void *) __get_free_pages(GFP_KERNEL,
 					get_order(KFD_SIGNAL_EVENT_LIMIT * 8));
 	if (!backing_store)
 		goto fail_alloc_signal_store;
 
-	/* prevent user-mode info leaks */
+	/* Initialize all events to unsignaled */
 	memset(backing_store, (uint8_t) UNSIGNALED_EVENT_SLOT,
-		KFD_SIGNAL_EVENT_LIMIT * 8);
+	       KFD_SIGNAL_EVENT_LIMIT * 8);
 
 	page->kernel_address = backing_store;
-
-	if (list_empty(&p->signal_event_pages))
-		page->page_index = 0;
-	else
-		page->page_index = list_tail_entry(&p->signal_event_pages,
-						   struct signal_page,
-						   event_pages)->page_index + 1;
-
+	page->need_to_free_pages = true;
 	pr_debug("Allocated new event signal page at %p, for process %p\n",
 			page, p);
-	pr_debug("Page index is %d\n", page->page_index);
-
-	list_add(&page->event_pages, &p->signal_event_pages);
 
-	return true;
+	return page;
 
 fail_alloc_signal_store:
 	kfree(page);
-fail_alloc_signal_page:
-	return false;
+	return NULL;
 }
 
-static bool allocate_event_notification_slot(struct file *devkfd,
-					struct kfd_process *p,
-					struct signal_page **page,
-					unsigned int *signal_slot_index)
+static int allocate_event_notification_slot(struct kfd_process *p,
+					    struct kfd_event *ev,
+					    const int *restore_id)
 {
-	bool ret;
-
-	ret = allocate_free_slot(p, page, signal_slot_index);
-	if (!ret) {
-		ret = allocate_signal_page(devkfd, p);
-		if (ret)
-			ret = allocate_free_slot(p, page, signal_slot_index);
+	int id;
+
+	if (!p->signal_page) {
+		p->signal_page = allocate_signal_page(p);
+		if (!p->signal_page)
+			return -ENOMEM;
+		/* Oldest user mode expects 256 event slots */
+		p->signal_mapped_size = 256*8;
 	}
 
-	return ret;
-}
-
-/* Assumes that the process's event_mutex is locked. */
-static void release_event_notification_slot(struct signal_page *page,
-						size_t slot_index)
-{
-	__clear_bit(slot_index, page->used_slot_bitmap);
-	page->free_slots++;
-
-	/* We don't free signal pages, they are retained by the process
-	 * and reused until it exits.
-	 */
-}
-
-static struct signal_page *lookup_signal_page_by_index(struct kfd_process *p,
-						unsigned int page_index)
-{
-	struct signal_page *page;
+	if (restore_id) {
+		id = idr_alloc(&p->event_idr, ev, *restore_id, *restore_id + 1,
+				GFP_KERNEL);
+	} else {
+		/*
+		 * Compatibility with old user mode: Only use signal slots
+		 * user mode has mapped, may be less than
+		 * KFD_SIGNAL_EVENT_LIMIT. This also allows future increase
+		 * of the event limit without breaking user mode.
+		 */
+		id = idr_alloc(&p->event_idr, ev, 0, p->signal_mapped_size / 8,
+				GFP_KERNEL);
+	}
+	if (id < 0)
+		return id;
 
-	/*
-	 * This is safe because we don't delete signal pages until the
-	 * process exits.
-	 */
-	list_for_each_entry(page, &p->signal_event_pages, event_pages)
-		if (page->page_index == page_index)
-			return page;
+	ev->event_id = id;
+	page_slots(p->signal_page)[id] = UNSIGNALED_EVENT_SLOT;
 
-	return NULL;
+	return 0;
 }
 
 /*
- * Assumes that p->event_mutex is held and of course that p is not going
- * away (current or locked).
+ * Assumes that p->event_mutex or rcu_readlock is held and of course that p is
+ * not going away.
  */
 static struct kfd_event *lookup_event_by_id(struct kfd_process *p, uint32_t id)
 {
-	struct kfd_event *ev;
-
-	hash_for_each_possible(p->events, ev, events, id)
-		if (ev->event_id == id)
-			return ev;
-
-	return NULL;
-}
-
-static u32 make_signal_event_id(struct signal_page *page,
-					 unsigned int signal_slot_index)
-{
-	return page->page_index |
-			(signal_slot_index << SIGNAL_EVENT_ID_SLOT_SHIFT);
+	return idr_find(&p->event_idr, id);
 }
 
-/*
- * Produce a kfd event id for a nonsignal event.
- * These are arbitrary numbers, so we do a sequential search through
- * the hash table for an unused number.
+/**
+ * lookup_signaled_event_by_partial_id - Lookup signaled event from partial ID
+ * @p:     Pointer to struct kfd_process
+ * @id:    ID to look up
+ * @bits:  Number of valid bits in @id
+ *
+ * Finds the first signaled event with a matching partial ID. If no
+ * matching signaled event is found, returns NULL. In that case the
+ * caller should assume that the partial ID is invalid and do an
+ * exhaustive search of all siglaned events.
+ *
+ * If multiple events with the same partial ID signal at the same
+ * time, they will be found one interrupt at a time, not necessarily
+ * in the same order the interrupts occurred. As long as the number of
+ * interrupts is correct, all signaled events will be seen by the
+ * driver.
  */
-static u32 make_nonsignal_event_id(struct kfd_process *p)
+static struct kfd_event *lookup_signaled_event_by_partial_id(
+	struct kfd_process *p, uint32_t id, uint32_t bits)
 {
-	u32 id;
-
-	for (id = p->next_nonsignal_event_id;
-		id < KFD_LAST_NONSIGNAL_EVENT_ID &&
-		lookup_event_by_id(p, id);
-		id++)
-		;
+	struct kfd_event *ev;
 
-	if (id < KFD_LAST_NONSIGNAL_EVENT_ID) {
+	if (!p->signal_page || id >= KFD_SIGNAL_EVENT_LIMIT)
+		return NULL;
 
-		/*
-		 * What if id == LAST_NONSIGNAL_EVENT_ID - 1?
-		 * Then next_nonsignal_event_id = LAST_NONSIGNAL_EVENT_ID so
-		 * the first loop fails immediately and we proceed with the
-		 * wraparound loop below.
-		 */
-		p->next_nonsignal_event_id = id + 1;
+	/* Fast path for the common case that @id is not a partial ID
+	 * and we only need a single lookup.
+	 */
+	if (bits > 31 || (1U << bits) >= KFD_SIGNAL_EVENT_LIMIT) {
+		if (page_slots(p->signal_page)[id] == UNSIGNALED_EVENT_SLOT)
+			return NULL;
 
-		return id;
+		return idr_find(&p->event_idr, id);
 	}
 
-	for (id = KFD_FIRST_NONSIGNAL_EVENT_ID;
-		id < KFD_LAST_NONSIGNAL_EVENT_ID &&
-		lookup_event_by_id(p, id);
-		id++)
-		;
-
+	/* General case for partial IDs: Iterate over all matching IDs
+	 * and find the first one that has signaled.
+	 */
+	for (ev = NULL; id < KFD_SIGNAL_EVENT_LIMIT && !ev; id += 1U << bits) {
+		if (page_slots(p->signal_page)[id] == UNSIGNALED_EVENT_SLOT)
+			continue;
 
-	if (id < KFD_LAST_NONSIGNAL_EVENT_ID) {
-		p->next_nonsignal_event_id = id + 1;
-		return id;
+		ev = idr_find(&p->event_idr, id);
 	}
 
-	p->next_nonsignal_event_id = KFD_FIRST_NONSIGNAL_EVENT_ID;
-	return 0;
+	return ev;
 }
 
-static struct kfd_event *lookup_event_by_page_slot(struct kfd_process *p,
-						struct signal_page *page,
-						unsigned int signal_slot)
+static int create_signal_event(struct file *devkfd, struct kfd_process *p,
+				struct kfd_event *ev, const int *restore_id)
 {
-	return lookup_event_by_id(p, make_signal_event_id(page, signal_slot));
-}
+	int ret;
 
-static int create_signal_event(struct file *devkfd,
-				struct kfd_process *p,
-				struct kfd_event *ev)
-{
-	if (p->signal_event_count == KFD_SIGNAL_EVENT_LIMIT) {
-		pr_warn("Signal event wasn't created because limit was reached\n");
-		return -ENOMEM;
+	if (p->signal_mapped_size &&
+	    p->signal_event_count == p->signal_mapped_size / 8) {
+		if (!p->signal_event_limit_reached) {
+			pr_debug("Signal event wasn't created because limit was reached\n");
+			p->signal_event_limit_reached = true;
+		}
+		return -ENOSPC;
 	}
 
-	if (!allocate_event_notification_slot(devkfd, p, &ev->signal_page,
-						&ev->signal_slot_index)) {
+	ret = allocate_event_notification_slot(p, ev, restore_id);
+	if (ret) {
 		pr_warn("Signal event wasn't created because out of kernel memory\n");
-		return -ENOMEM;
+		return ret;
 	}
 
 	p->signal_event_count++;
 
-	ev->user_signal_address =
-			&ev->signal_page->user_address[ev->signal_slot_index];
-
-	ev->event_id = make_signal_event_id(ev->signal_page,
-						ev->signal_slot_index);
-
+	ev->user_signal_address = &p->signal_page->user_address[ev->event_id];
 	pr_debug("Signal event number %zu created with id %d, address %p\n",
 			p->signal_event_count, ev->event_id,
 			ev->user_signal_address);
@@ -317,69 +215,93 @@ static int create_signal_event(struct file *devkfd,
 	return 0;
 }
 
-/*
- * No non-signal events are supported yet.
- * We create them as events that never signal.
- * Set event calls from user-mode are failed.
- */
-static int create_other_event(struct kfd_process *p, struct kfd_event *ev)
+static int create_other_event(struct kfd_process *p, struct kfd_event *ev, const int *restore_id)
 {
-	ev->event_id = make_nonsignal_event_id(p);
-	if (ev->event_id == 0)
-		return -ENOMEM;
+	int id;
+
+	if (restore_id)
+		id = idr_alloc(&p->event_idr, ev, *restore_id, *restore_id + 1,
+			GFP_KERNEL);
+	else
+		/* Cast KFD_LAST_NONSIGNAL_EVENT to uint32_t. This allows an
+		 * intentional integer overflow to -1 without a compiler
+		 * warning. idr_alloc treats a negative value as "maximum
+		 * signed integer".
+		 */
+		id = idr_alloc(&p->event_idr, ev, KFD_FIRST_NONSIGNAL_EVENT_ID,
+				(uint32_t)KFD_LAST_NONSIGNAL_EVENT_ID + 1,
+				GFP_KERNEL);
+
+	if (id < 0)
+		return id;
+	ev->event_id = id;
 
 	return 0;
 }
 
-void kfd_event_init_process(struct kfd_process *p)
+int kfd_event_init_process(struct kfd_process *p)
 {
+	int id;
+
 	mutex_init(&p->event_mutex);
-	hash_init(p->events);
-	INIT_LIST_HEAD(&p->signal_event_pages);
-	p->next_nonsignal_event_id = KFD_FIRST_NONSIGNAL_EVENT_ID;
-	p->signal_event_count = 0;
+	idr_init(&p->event_idr);
+	p->signal_page = NULL;
+	p->signal_event_count = 1;
+	/* Allocate event ID 0. It is used for a fast path to ignore bogus events
+	 * that are sent by the CP without a context ID
+	 */
+	id = idr_alloc(&p->event_idr, NULL, 0, 1, GFP_KERNEL);
+	if (id < 0) {
+		idr_destroy(&p->event_idr);
+		mutex_destroy(&p->event_mutex);
+		return id;
+	}
+	return 0;
 }
 
 static void destroy_event(struct kfd_process *p, struct kfd_event *ev)
 {
-	if (ev->signal_page) {
-		release_event_notification_slot(ev->signal_page,
-						ev->signal_slot_index);
-		p->signal_event_count--;
-	}
+	struct kfd_event_waiter *waiter;
 
-	/*
-	 * Abandon the list of waiters. Individual waiting threads will
-	 * clean up their own data.
-	 */
-	list_del(&ev->waiters);
+	/* Wake up pending waiters. They will return failure */
+	spin_lock(&ev->lock);
+	list_for_each_entry(waiter, &ev->wq.head, wait.entry)
+		WRITE_ONCE(waiter->event, NULL);
+	wake_up_all(&ev->wq);
+	spin_unlock(&ev->lock);
 
-	hash_del(&ev->events);
-	kfree(ev);
+	if (ev->type == KFD_EVENT_TYPE_SIGNAL ||
+	    ev->type == KFD_EVENT_TYPE_DEBUG)
+		p->signal_event_count--;
+
+	idr_remove(&p->event_idr, ev->event_id);
+	kfree_rcu(ev, rcu);
 }
 
 static void destroy_events(struct kfd_process *p)
 {
 	struct kfd_event *ev;
-	struct hlist_node *tmp;
-	unsigned int hash_bkt;
+	uint32_t id;
 
-	hash_for_each_safe(p->events, hash_bkt, tmp, ev, events)
-		destroy_event(p, ev);
+	idr_for_each_entry(&p->event_idr, ev, id)
+		if (ev)
+			destroy_event(p, ev);
+	idr_destroy(&p->event_idr);
+	mutex_destroy(&p->event_mutex);
 }
 
 /*
  * We assume that the process is being destroyed and there is no need to
  * unmap the pages or keep bookkeeping data in order.
  */
-static void shutdown_signal_pages(struct kfd_process *p)
+static void shutdown_signal_page(struct kfd_process *p)
 {
-	struct signal_page *page, *tmp;
+	struct kfd_signal_page *page = p->signal_page;
 
-	list_for_each_entry_safe(page, tmp, &p->signal_event_pages,
-					event_pages) {
-		free_pages((unsigned long)page->kernel_address,
-				get_order(KFD_SIGNAL_EVENT_LIMIT * 8));
+	if (page) {
+		if (page->need_to_free_pages)
+			free_pages((unsigned long)page->kernel_address,
+				   get_order(KFD_SIGNAL_EVENT_LIMIT * 8));
 		kfree(page);
 	}
 }
@@ -387,7 +309,7 @@ static void shutdown_signal_pages(struct kfd_process *p)
 void kfd_event_free_process(struct kfd_process *p)
 {
 	destroy_events(p);
-	shutdown_signal_pages(p);
+	shutdown_signal_page(p);
 }
 
 static bool event_can_be_gpu_signaled(const struct kfd_event *ev)
@@ -401,6 +323,76 @@ static bool event_can_be_cpu_signaled(const struct kfd_event *ev)
 	return ev->type == KFD_EVENT_TYPE_SIGNAL;
 }
 
+static int kfd_event_page_set(struct kfd_process *p, void *kernel_address,
+		       uint64_t size, uint64_t user_handle)
+{
+	struct kfd_signal_page *page;
+
+	if (p->signal_page)
+		return -EBUSY;
+
+	page = kzalloc(sizeof(*page), GFP_KERNEL);
+	if (!page)
+		return -ENOMEM;
+
+	/* Initialize all events to unsignaled */
+	memset(kernel_address, (uint8_t) UNSIGNALED_EVENT_SLOT,
+	       KFD_SIGNAL_EVENT_LIMIT * 8);
+
+	page->kernel_address = kernel_address;
+
+	p->signal_page = page;
+	p->signal_mapped_size = size;
+	p->signal_handle = user_handle;
+	return 0;
+}
+
+int kfd_kmap_event_page(struct kfd_process *p, uint64_t event_page_offset)
+{
+	struct kfd_node *kfd;
+	struct kfd_process_device *pdd;
+	void *mem, *kern_addr;
+	uint64_t size;
+	int err = 0;
+
+	if (p->signal_page) {
+		pr_err("Event page is already set\n");
+		return -EINVAL;
+	}
+
+	pdd = kfd_process_device_data_by_id(p, GET_GPU_ID(event_page_offset));
+	if (!pdd) {
+		pr_err("Getting device by id failed in %s\n", __func__);
+		return -EINVAL;
+	}
+	kfd = pdd->dev;
+
+	pdd = kfd_bind_process_to_device(kfd, p);
+	if (IS_ERR(pdd))
+		return PTR_ERR(pdd);
+
+	mem = kfd_process_device_translate_handle(pdd,
+			GET_IDR_HANDLE(event_page_offset));
+	if (!mem) {
+		pr_err("Can't find BO, offset is 0x%llx\n", event_page_offset);
+		return -EINVAL;
+	}
+
+	err = amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel(mem, &kern_addr, &size);
+	if (err) {
+		pr_err("Failed to map event page to kernel\n");
+		return err;
+	}
+
+	err = kfd_event_page_set(p, kern_addr, size, event_page_offset);
+	if (err) {
+		pr_err("Failed to set event page\n");
+		amdgpu_amdkfd_gpuvm_unmap_gtt_bo_from_kernel(mem);
+		return err;
+	}
+	return err;
+}
+
 int kfd_event_create(struct file *devkfd, struct kfd_process *p,
 		     uint32_t event_type, bool auto_reset, uint32_t node_id,
 		     uint32_t *event_id, uint32_t *event_trigger_data,
@@ -416,7 +408,8 @@ int kfd_event_create(struct file *devkfd, struct kfd_process *p,
 	ev->auto_reset = auto_reset;
 	ev->signaled = false;
 
-	INIT_LIST_HEAD(&ev->waiters);
+	spin_lock_init(&ev->lock);
+	init_waitqueue_head(&ev->wq);
 
 	*event_page_offset = 0;
 
@@ -425,24 +418,21 @@ int kfd_event_create(struct file *devkfd, struct kfd_process *p,
 	switch (event_type) {
 	case KFD_EVENT_TYPE_SIGNAL:
 	case KFD_EVENT_TYPE_DEBUG:
-		ret = create_signal_event(devkfd, p, ev);
+		ret = create_signal_event(devkfd, p, ev, NULL);
 		if (!ret) {
-			*event_page_offset = (ev->signal_page->page_index |
-					KFD_MMAP_EVENTS_MASK);
-			*event_page_offset <<= PAGE_SHIFT;
-			*event_slot_index = ev->signal_slot_index;
+			*event_page_offset = KFD_MMAP_TYPE_EVENTS;
+			*event_slot_index = ev->event_id;
 		}
 		break;
 	default:
-		ret = create_other_event(p, ev);
+		ret = create_other_event(p, ev, NULL);
 		break;
 	}
 
 	if (!ret) {
-		hash_add(p->events, &ev->events, ev->event_id);
-
 		*event_id = ev->event_id;
 		*event_trigger_data = ev->event_id;
+		ev->event_age = 1;
 	} else {
 		kfree(ev);
 	}
@@ -452,6 +442,166 @@ int kfd_event_create(struct file *devkfd, struct kfd_process *p,
 	return ret;
 }
 
+int kfd_criu_restore_event(struct file *devkfd,
+			   struct kfd_process *p,
+			   uint8_t __user *user_priv_ptr,
+			   uint64_t *priv_data_offset,
+			   uint64_t max_priv_data_size)
+{
+	struct kfd_criu_event_priv_data *ev_priv;
+	struct kfd_event *ev = NULL;
+	int ret = 0;
+
+	ev_priv = kmalloc(sizeof(*ev_priv), GFP_KERNEL);
+	if (!ev_priv)
+		return -ENOMEM;
+
+	ev = kzalloc(sizeof(*ev), GFP_KERNEL);
+	if (!ev) {
+		ret = -ENOMEM;
+		goto exit;
+	}
+
+	if (*priv_data_offset + sizeof(*ev_priv) > max_priv_data_size) {
+		ret = -EINVAL;
+		goto exit;
+	}
+
+	ret = copy_from_user(ev_priv, user_priv_ptr + *priv_data_offset, sizeof(*ev_priv));
+	if (ret) {
+		ret = -EFAULT;
+		goto exit;
+	}
+	*priv_data_offset += sizeof(*ev_priv);
+
+	if (ev_priv->user_handle) {
+		ret = kfd_kmap_event_page(p, ev_priv->user_handle);
+		if (ret)
+			goto exit;
+	}
+
+	ev->type = ev_priv->type;
+	ev->auto_reset = ev_priv->auto_reset;
+	ev->signaled = ev_priv->signaled;
+
+	spin_lock_init(&ev->lock);
+	init_waitqueue_head(&ev->wq);
+
+	mutex_lock(&p->event_mutex);
+	switch (ev->type) {
+	case KFD_EVENT_TYPE_SIGNAL:
+	case KFD_EVENT_TYPE_DEBUG:
+		ret = create_signal_event(devkfd, p, ev, &ev_priv->event_id);
+		break;
+	case KFD_EVENT_TYPE_MEMORY:
+		memcpy(&ev->memory_exception_data,
+			&ev_priv->memory_exception_data,
+			sizeof(struct kfd_hsa_memory_exception_data));
+
+		ret = create_other_event(p, ev, &ev_priv->event_id);
+		break;
+	case KFD_EVENT_TYPE_HW_EXCEPTION:
+		memcpy(&ev->hw_exception_data,
+			&ev_priv->hw_exception_data,
+			sizeof(struct kfd_hsa_hw_exception_data));
+
+		ret = create_other_event(p, ev, &ev_priv->event_id);
+		break;
+	}
+	mutex_unlock(&p->event_mutex);
+
+exit:
+	if (ret)
+		kfree(ev);
+
+	kfree(ev_priv);
+
+	return ret;
+}
+
+int kfd_criu_checkpoint_events(struct kfd_process *p,
+			 uint8_t __user *user_priv_data,
+			 uint64_t *priv_data_offset)
+{
+	struct kfd_criu_event_priv_data *ev_privs;
+	int i = 0;
+	int ret =  0;
+	struct kfd_event *ev;
+	uint32_t ev_id;
+
+	uint32_t num_events = kfd_get_num_events(p);
+
+	if (!num_events)
+		return 0;
+
+	ev_privs = kvzalloc(num_events * sizeof(*ev_privs), GFP_KERNEL);
+	if (!ev_privs)
+		return -ENOMEM;
+
+
+	idr_for_each_entry(&p->event_idr, ev, ev_id) {
+		struct kfd_criu_event_priv_data *ev_priv;
+
+		/*
+		 * Currently, all events have same size of private_data, but the current ioctl's
+		 * and CRIU plugin supports private_data of variable sizes
+		 */
+		ev_priv = &ev_privs[i];
+
+		ev_priv->object_type = KFD_CRIU_OBJECT_TYPE_EVENT;
+
+		/* We store the user_handle with the first event */
+		if (i == 0 && p->signal_page)
+			ev_priv->user_handle = p->signal_handle;
+
+		ev_priv->event_id = ev->event_id;
+		ev_priv->auto_reset = ev->auto_reset;
+		ev_priv->type = ev->type;
+		ev_priv->signaled = ev->signaled;
+
+		if (ev_priv->type == KFD_EVENT_TYPE_MEMORY)
+			memcpy(&ev_priv->memory_exception_data,
+				&ev->memory_exception_data,
+				sizeof(struct kfd_hsa_memory_exception_data));
+		else if (ev_priv->type == KFD_EVENT_TYPE_HW_EXCEPTION)
+			memcpy(&ev_priv->hw_exception_data,
+				&ev->hw_exception_data,
+				sizeof(struct kfd_hsa_hw_exception_data));
+
+		pr_debug("Checkpointed event[%d] id = 0x%08x auto_reset = %x type = %x signaled = %x\n",
+			  i,
+			  ev_priv->event_id,
+			  ev_priv->auto_reset,
+			  ev_priv->type,
+			  ev_priv->signaled);
+		i++;
+	}
+
+	ret = copy_to_user(user_priv_data + *priv_data_offset,
+			   ev_privs, num_events * sizeof(*ev_privs));
+	if (ret) {
+		pr_err("Failed to copy events priv to user\n");
+		ret = -EFAULT;
+	}
+
+	*priv_data_offset += num_events * sizeof(*ev_privs);
+
+	kvfree(ev_privs);
+	return ret;
+}
+
+int kfd_get_num_events(struct kfd_process *p)
+{
+	struct kfd_event *ev;
+	uint32_t id;
+	u32 num_events = 0;
+
+	idr_for_each_entry(&p->event_idr, ev, id)
+		num_events++;
+
+	return num_events;
+}
+
 /* Assumes that p is current. */
 int kfd_event_destroy(struct kfd_process *p, uint32_t event_id)
 {
@@ -474,19 +624,23 @@ int kfd_event_destroy(struct kfd_process *p, uint32_t event_id)
 static void set_event(struct kfd_event *ev)
 {
 	struct kfd_event_waiter *waiter;
-	struct kfd_event_waiter *next;
-
-	/* Auto reset if the list is non-empty and we're waking someone. */
-	ev->signaled = !ev->auto_reset || list_empty(&ev->waiters);
 
-	list_for_each_entry_safe(waiter, next, &ev->waiters, waiters) {
-		waiter->activated = true;
+	/* Auto reset if the list is non-empty and we're waking
+	 * someone. waitqueue_active is safe here because we're
+	 * protected by the ev->lock, which is also held when
+	 * updating the wait queues in kfd_wait_on_events.
+	 */
+	ev->signaled = !ev->auto_reset || !waitqueue_active(&ev->wq);
+	if (!(++ev->event_age)) {
+		/* Never wrap back to reserved/default event age 0/1 */
+		ev->event_age = 2;
+		WARN_ONCE(1, "event_age wrap back!");
+	}
 
-		/* _init because free_waiters will call list_del */
-		list_del_init(&waiter->waiters);
+	list_for_each_entry(waiter, &ev->wq.head, wait.entry)
+		WRITE_ONCE(waiter->activated, true);
 
-		wake_up_process(waiter->sleeping_task);
-	}
+	wake_up_all(&ev->wq);
 }
 
 /* Assumes that p is current. */
@@ -495,16 +649,23 @@ int kfd_set_event(struct kfd_process *p, uint32_t event_id)
 	int ret = 0;
 	struct kfd_event *ev;
 
-	mutex_lock(&p->event_mutex);
+	rcu_read_lock();
 
 	ev = lookup_event_by_id(p, event_id);
+	if (!ev) {
+		ret = -EINVAL;
+		goto unlock_rcu;
+	}
+	spin_lock(&ev->lock);
 
-	if (ev && event_can_be_cpu_signaled(ev))
+	if (event_can_be_cpu_signaled(ev))
 		set_event(ev);
 	else
 		ret = -EINVAL;
 
-	mutex_unlock(&p->event_mutex);
+	spin_unlock(&ev->lock);
+unlock_rcu:
+	rcu_read_unlock();
 	return ret;
 }
 
@@ -519,29 +680,30 @@ int kfd_reset_event(struct kfd_process *p, uint32_t event_id)
 	int ret = 0;
 	struct kfd_event *ev;
 
-	mutex_lock(&p->event_mutex);
+	rcu_read_lock();
 
 	ev = lookup_event_by_id(p, event_id);
+	if (!ev) {
+		ret = -EINVAL;
+		goto unlock_rcu;
+	}
+	spin_lock(&ev->lock);
 
-	if (ev && event_can_be_cpu_signaled(ev))
+	if (event_can_be_cpu_signaled(ev))
 		reset_event(ev);
 	else
 		ret = -EINVAL;
 
-	mutex_unlock(&p->event_mutex);
+	spin_unlock(&ev->lock);
+unlock_rcu:
+	rcu_read_unlock();
 	return ret;
 
 }
 
 static void acknowledge_signal(struct kfd_process *p, struct kfd_event *ev)
 {
-	page_slots(ev->signal_page)[ev->signal_slot_index] =
-						UNSIGNALED_EVENT_SLOT;
-}
-
-static bool is_slot_signaled(struct signal_page *page, unsigned int index)
-{
-	return page_slots(page)[index] != UNSIGNALED_EVENT_SLOT;
+	WRITE_ONCE(page_slots(p->signal_page)[ev->event_id], UNSIGNALED_EVENT_SLOT);
 }
 
 static void set_event_from_interrupt(struct kfd_process *p,
@@ -549,51 +711,84 @@ static void set_event_from_interrupt(struct kfd_process *p,
 {
 	if (ev && event_can_be_gpu_signaled(ev)) {
 		acknowledge_signal(p, ev);
+		spin_lock(&ev->lock);
 		set_event(ev);
+		spin_unlock(&ev->lock);
 	}
 }
 
-void kfd_signal_event_interrupt(unsigned int pasid, uint32_t partial_id,
+void kfd_signal_event_interrupt(u32 pasid, uint32_t partial_id,
 				uint32_t valid_id_bits)
 {
-	struct kfd_event *ev;
+	struct kfd_event *ev = NULL;
 
 	/*
 	 * Because we are called from arbitrary context (workqueue) as opposed
 	 * to process context, kfd_process could attempt to exit while we are
-	 * running so the lookup function returns a locked process.
+	 * running so the lookup function increments the process ref count.
 	 */
-	struct kfd_process *p = kfd_lookup_process_by_pasid(pasid);
+	struct kfd_process *p = kfd_lookup_process_by_pasid(pasid, NULL);
 
 	if (!p)
 		return; /* Presumably process exited. */
 
-	mutex_lock(&p->event_mutex);
+	rcu_read_lock();
 
-	if (valid_id_bits >= INTERRUPT_DATA_BITS) {
-		/* Partial ID is a full ID. */
-		ev = lookup_event_by_id(p, partial_id);
+	if (valid_id_bits)
+		ev = lookup_signaled_event_by_partial_id(p, partial_id,
+							 valid_id_bits);
+	if (ev) {
 		set_event_from_interrupt(p, ev);
-	} else {
+	} else if (p->signal_page) {
+		/*
+		 * Partial ID lookup failed. Assume that the event ID
+		 * in the interrupt payload was invalid and do an
+		 * exhaustive search of signaled events.
+		 */
+		uint64_t *slots = page_slots(p->signal_page);
+		uint32_t id;
+
 		/*
-		 * Partial ID is in fact partial. For now we completely
-		 * ignore it, but we could use any bits we did receive to
-		 * search faster.
+		 * If id is valid but slot is not signaled, GPU may signal the same event twice
+		 * before driver have chance to process the first interrupt, then signal slot is
+		 * auto-reset after set_event wakeup the user space, just drop the second event as
+		 * the application only need wakeup once.
 		 */
-		struct signal_page *page;
-		unsigned int i;
-
-		list_for_each_entry(page, &p->signal_event_pages, event_pages)
-			for (i = 0; i < SLOTS_PER_PAGE; i++)
-				if (is_slot_signaled(page, i)) {
-					ev = lookup_event_by_page_slot(p,
-								page, i);
+		if ((valid_id_bits > 31 || (1U << valid_id_bits) >= KFD_SIGNAL_EVENT_LIMIT) &&
+		    partial_id < KFD_SIGNAL_EVENT_LIMIT && slots[partial_id] == UNSIGNALED_EVENT_SLOT)
+			goto out_unlock;
+
+		if (valid_id_bits)
+			pr_debug_ratelimited("Partial ID invalid: %u (%u valid bits)\n",
+					     partial_id, valid_id_bits);
+
+		if (p->signal_event_count < KFD_SIGNAL_EVENT_LIMIT / 64) {
+			/* With relatively few events, it's faster to
+			 * iterate over the event IDR
+			 */
+			idr_for_each_entry(&p->event_idr, ev, id) {
+				if (id >= KFD_SIGNAL_EVENT_LIMIT)
+					break;
+
+				if (READ_ONCE(slots[id]) != UNSIGNALED_EVENT_SLOT)
+					set_event_from_interrupt(p, ev);
+			}
+		} else {
+			/* With relatively many events, it's faster to
+			 * iterate over the signal slots and lookup
+			 * only signaled events from the IDR.
+			 */
+			for (id = 1; id < KFD_SIGNAL_EVENT_LIMIT; id++)
+				if (READ_ONCE(slots[id]) != UNSIGNALED_EVENT_SLOT) {
+					ev = lookup_event_by_id(p, id);
 					set_event_from_interrupt(p, ev);
 				}
+		}
 	}
 
-	mutex_unlock(&p->event_mutex);
-	mutex_unlock(&p->mutex);
+out_unlock:
+	rcu_read_unlock();
+	kfd_unref_process(p);
 }
 
 static struct kfd_event_waiter *alloc_event_waiters(uint32_t num_events)
@@ -601,89 +796,120 @@ static struct kfd_event_waiter *alloc_event_waiters(uint32_t num_events)
 	struct kfd_event_waiter *event_waiters;
 	uint32_t i;
 
-	event_waiters = kmalloc_array(num_events,
-					sizeof(struct kfd_event_waiter),
-					GFP_KERNEL);
+	event_waiters = kcalloc(num_events, sizeof(struct kfd_event_waiter),
+				GFP_KERNEL);
+	if (!event_waiters)
+		return NULL;
 
-	for (i = 0; (event_waiters) && (i < num_events) ; i++) {
-		INIT_LIST_HEAD(&event_waiters[i].waiters);
-		event_waiters[i].sleeping_task = current;
-		event_waiters[i].activated = false;
-	}
+	for (i = 0; i < num_events; i++)
+		init_wait(&event_waiters[i].wait);
 
 	return event_waiters;
 }
 
 static int init_event_waiter(struct kfd_process *p,
 		struct kfd_event_waiter *waiter,
-		uint32_t event_id,
-		uint32_t input_index)
+		struct kfd_event_data *event_data)
 {
-	struct kfd_event *ev = lookup_event_by_id(p, event_id);
+	struct kfd_event *ev = lookup_event_by_id(p, event_data->event_id);
 
 	if (!ev)
 		return -EINVAL;
 
+	spin_lock(&ev->lock);
 	waiter->event = ev;
-	waiter->input_index = input_index;
 	waiter->activated = ev->signaled;
 	ev->signaled = ev->signaled && !ev->auto_reset;
 
-	list_add(&waiter->waiters, &ev->waiters);
+	/* last_event_age = 0 reserved for backward compatible */
+	if (waiter->event->type == KFD_EVENT_TYPE_SIGNAL &&
+		event_data->signal_event_data.last_event_age) {
+		waiter->event_age_enabled = true;
+		if (ev->event_age != event_data->signal_event_data.last_event_age)
+			waiter->activated = true;
+	}
+
+	if (!waiter->activated)
+		add_wait_queue(&ev->wq, &waiter->wait);
+	spin_unlock(&ev->lock);
 
 	return 0;
 }
 
-static bool test_event_condition(bool all, uint32_t num_events,
+/* test_event_condition - Test condition of events being waited for
+ * @all:           Return completion only if all events have signaled
+ * @num_events:    Number of events to wait for
+ * @event_waiters: Array of event waiters, one per event
+ *
+ * Returns KFD_IOC_WAIT_RESULT_COMPLETE if all (or one) event(s) have
+ * signaled. Returns KFD_IOC_WAIT_RESULT_TIMEOUT if no (or not all)
+ * events have signaled. Returns KFD_IOC_WAIT_RESULT_FAIL if any of
+ * the events have been destroyed.
+ */
+static uint32_t test_event_condition(bool all, uint32_t num_events,
 				struct kfd_event_waiter *event_waiters)
 {
 	uint32_t i;
 	uint32_t activated_count = 0;
 
 	for (i = 0; i < num_events; i++) {
-		if (event_waiters[i].activated) {
+		if (!READ_ONCE(event_waiters[i].event))
+			return KFD_IOC_WAIT_RESULT_FAIL;
+
+		if (READ_ONCE(event_waiters[i].activated)) {
 			if (!all)
-				return true;
+				return KFD_IOC_WAIT_RESULT_COMPLETE;
 
 			activated_count++;
 		}
 	}
 
-	return activated_count == num_events;
+	return activated_count == num_events ?
+		KFD_IOC_WAIT_RESULT_COMPLETE : KFD_IOC_WAIT_RESULT_TIMEOUT;
 }
 
 /*
  * Copy event specific data, if defined.
  * Currently only memory exception events have additional data to copy to user
  */
-static bool copy_signaled_event_data(uint32_t num_events,
+static int copy_signaled_event_data(uint32_t num_events,
 		struct kfd_event_waiter *event_waiters,
 		struct kfd_event_data __user *data)
 {
-	struct kfd_hsa_memory_exception_data *src;
-	struct kfd_hsa_memory_exception_data __user *dst;
+	void *src;
+	void __user *dst;
 	struct kfd_event_waiter *waiter;
 	struct kfd_event *event;
-	uint32_t i;
+	uint32_t i, size = 0;
 
 	for (i = 0; i < num_events; i++) {
 		waiter = &event_waiters[i];
 		event = waiter->event;
-		if (waiter->activated && event->type == KFD_EVENT_TYPE_MEMORY) {
-			dst = &data[waiter->input_index].memory_exception_data;
-			src = &event->memory_exception_data;
-			if (copy_to_user(dst, src,
-				sizeof(struct kfd_hsa_memory_exception_data)))
-				return false;
+		if (!event)
+			return -EINVAL; /* event was destroyed */
+		if (waiter->activated) {
+			if (event->type == KFD_EVENT_TYPE_MEMORY) {
+				dst = &data[i].memory_exception_data;
+				src = &event->memory_exception_data;
+				size = sizeof(struct kfd_hsa_memory_exception_data);
+			} else if (event->type == KFD_EVENT_TYPE_HW_EXCEPTION) {
+				dst = &data[i].memory_exception_data;
+				src = &event->hw_exception_data;
+				size = sizeof(struct kfd_hsa_hw_exception_data);
+			} else if (event->type == KFD_EVENT_TYPE_SIGNAL &&
+				waiter->event_age_enabled) {
+				dst = &data[i].signal_event_data.last_event_age;
+				src = &event->event_age;
+				size = sizeof(u64);
+			}
+			if (size && copy_to_user(dst, src, size))
+				return -EFAULT;
 		}
 	}
 
-	return true;
-
+	return 0;
 }
 
-
-
 static long user_timeout_to_jiffies(uint32_t user_timeout_ms)
 {
 	if (user_timeout_ms == KFD_EVENT_TIMEOUT_IMMEDIATE)
@@ -702,49 +928,74 @@ static long user_timeout_to_jiffies(uint32_t user_timeout_ms)
 	return msecs_to_jiffies(user_timeout_ms) + 1;
 }
 
-static void free_waiters(uint32_t num_events, struct kfd_event_waiter *waiters)
+static void free_waiters(uint32_t num_events, struct kfd_event_waiter *waiters,
+			 bool undo_auto_reset)
 {
 	uint32_t i;
 
 	for (i = 0; i < num_events; i++)
-		list_del(&waiters[i].waiters);
+		if (waiters[i].event) {
+			spin_lock(&waiters[i].event->lock);
+			remove_wait_queue(&waiters[i].event->wq,
+					  &waiters[i].wait);
+			if (undo_auto_reset && waiters[i].activated &&
+			    waiters[i].event && waiters[i].event->auto_reset)
+				set_event(waiters[i].event);
+			spin_unlock(&waiters[i].event->lock);
+		}
 
 	kfree(waiters);
 }
 
 int kfd_wait_on_events(struct kfd_process *p,
 		       uint32_t num_events, void __user *data,
-		       bool all, uint32_t user_timeout_ms,
-		       enum kfd_event_wait_result *wait_result)
+		       bool all, uint32_t *user_timeout_ms,
+		       uint32_t *wait_result)
 {
 	struct kfd_event_data __user *events =
 			(struct kfd_event_data __user *) data;
 	uint32_t i;
 	int ret = 0;
-	struct kfd_event_waiter *event_waiters = NULL;
-	long timeout = user_timeout_to_jiffies(user_timeout_ms);
 
-	mutex_lock(&p->event_mutex);
+	struct kfd_event_waiter *event_waiters = NULL;
+	long timeout = user_timeout_to_jiffies(*user_timeout_ms);
 
 	event_waiters = alloc_event_waiters(num_events);
 	if (!event_waiters) {
 		ret = -ENOMEM;
-		goto fail;
+		goto out;
 	}
 
+	/* Use p->event_mutex here to protect against concurrent creation and
+	 * destruction of events while we initialize event_waiters.
+	 */
+	mutex_lock(&p->event_mutex);
+
 	for (i = 0; i < num_events; i++) {
 		struct kfd_event_data event_data;
 
 		if (copy_from_user(&event_data, &events[i],
 				sizeof(struct kfd_event_data))) {
 			ret = -EFAULT;
-			goto fail;
+			goto out_unlock;
 		}
 
-		ret = init_event_waiter(p, &event_waiters[i],
-				event_data.event_id, i);
+		ret = init_event_waiter(p, &event_waiters[i], &event_data);
 		if (ret)
-			goto fail;
+			goto out_unlock;
+	}
+
+	/* Check condition once. */
+	*wait_result = test_event_condition(all, num_events, event_waiters);
+	if (*wait_result == KFD_IOC_WAIT_RESULT_COMPLETE) {
+		ret = copy_signaled_event_data(num_events,
+					       event_waiters, events);
+		goto out_unlock;
+	} else if (WARN_ON(*wait_result == KFD_IOC_WAIT_RESULT_FAIL)) {
+		/* This should not happen. Events shouldn't be
+		 * destroyed while we're holding the event_mutex
+		 */
+		goto out_unlock;
 	}
 
 	mutex_unlock(&p->event_mutex);
@@ -756,82 +1007,87 @@ int kfd_wait_on_events(struct kfd_process *p,
 		}
 
 		if (signal_pending(current)) {
-			/*
-			 * This is wrong when a nonzero, non-infinite timeout
-			 * is specified. We need to use
-			 * ERESTARTSYS_RESTARTBLOCK, but struct restart_block
-			 * contains a union with data for each user and it's
-			 * in generic kernel code that I don't want to
-			 * touch yet.
-			 */
 			ret = -ERESTARTSYS;
+			if (*user_timeout_ms != KFD_EVENT_TIMEOUT_IMMEDIATE &&
+			    *user_timeout_ms != KFD_EVENT_TIMEOUT_INFINITE)
+				*user_timeout_ms = jiffies_to_msecs(
+					max(0l, timeout-1));
 			break;
 		}
 
-		if (test_event_condition(all, num_events, event_waiters)) {
-			if (copy_signaled_event_data(num_events,
-					event_waiters, events))
-				*wait_result = KFD_WAIT_COMPLETE;
-			else
-				*wait_result = KFD_WAIT_ERROR;
+		/* Set task state to interruptible sleep before
+		 * checking wake-up conditions. A concurrent wake-up
+		 * will put the task back into runnable state. In that
+		 * case schedule_timeout will not put the task to
+		 * sleep and we'll get a chance to re-check the
+		 * updated conditions almost immediately. Otherwise,
+		 * this race condition would lead to a soft hang or a
+		 * very long sleep.
+		 */
+		set_current_state(TASK_INTERRUPTIBLE);
+
+		*wait_result = test_event_condition(all, num_events,
+						    event_waiters);
+		if (*wait_result != KFD_IOC_WAIT_RESULT_TIMEOUT)
 			break;
-		}
 
-		if (timeout <= 0) {
-			*wait_result = KFD_WAIT_TIMEOUT;
+		if (timeout <= 0)
 			break;
-		}
 
-		timeout = schedule_timeout_interruptible(timeout);
+		timeout = schedule_timeout(timeout);
 	}
 	__set_current_state(TASK_RUNNING);
 
 	mutex_lock(&p->event_mutex);
-	free_waiters(num_events, event_waiters);
-	mutex_unlock(&p->event_mutex);
-
-	return ret;
-
-fail:
-	if (event_waiters)
-		free_waiters(num_events, event_waiters);
+	/* copy_signaled_event_data may sleep. So this has to happen
+	 * after the task state is set back to RUNNING.
+	 *
+	 * The event may also have been destroyed after signaling. So
+	 * copy_signaled_event_data also must confirm that the event
+	 * still exists. Therefore this must be under the p->event_mutex
+	 * which is also held when events are destroyed.
+	 */
+	if (!ret && *wait_result == KFD_IOC_WAIT_RESULT_COMPLETE)
+		ret = copy_signaled_event_data(num_events,
+					       event_waiters, events);
 
+out_unlock:
+	free_waiters(num_events, event_waiters, ret == -ERESTARTSYS);
 	mutex_unlock(&p->event_mutex);
-
-	*wait_result = KFD_WAIT_ERROR;
+out:
+	if (ret)
+		*wait_result = KFD_IOC_WAIT_RESULT_FAIL;
+	else if (*wait_result == KFD_IOC_WAIT_RESULT_FAIL)
+		ret = -EIO;
 
 	return ret;
 }
 
 int kfd_event_mmap(struct kfd_process *p, struct vm_area_struct *vma)
 {
-
-	unsigned int page_index;
 	unsigned long pfn;
-	struct signal_page *page;
+	struct kfd_signal_page *page;
+	int ret;
 
-	/* check required size is logical */
-	if (get_order(KFD_SIGNAL_EVENT_LIMIT * 8) !=
+	/* check required size doesn't exceed the allocated size */
+	if (get_order(KFD_SIGNAL_EVENT_LIMIT * 8) <
 			get_order(vma->vm_end - vma->vm_start)) {
 		pr_err("Event page mmap requested illegal size\n");
 		return -EINVAL;
 	}
 
-	page_index = vma->vm_pgoff;
-
-	page = lookup_signal_page_by_index(p, page_index);
+	page = p->signal_page;
 	if (!page) {
 		/* Probably KFD bug, but mmap is user-accessible. */
-		pr_debug("Signal page could not be found for page_index %u\n",
-				page_index);
+		pr_debug("Signal page could not be found\n");
 		return -EINVAL;
 	}
 
 	pfn = __pa(page->kernel_address);
 	pfn >>= PAGE_SHIFT;
 
-	vma->vm_flags |= VM_IO | VM_DONTCOPY | VM_DONTEXPAND | VM_NORESERVE
-		       | VM_DONTDUMP | VM_PFNMAP;
+	vm_flags_set(vma, VM_IO | VM_DONTCOPY | VM_DONTEXPAND | VM_NORESERVE
+		       | VM_DONTDUMP | VM_PFNMAP);
 
 	pr_debug("Mapping signal page\n");
 	pr_debug("     start user address  == 0x%08lx\n", vma->vm_start);
@@ -844,125 +1100,294 @@ int kfd_event_mmap(struct kfd_process *p, struct vm_area_struct *vma)
 	page->user_address = (uint64_t __user *)vma->vm_start;
 
 	/* mapping the page to user process */
-	return remap_pfn_range(vma, vma->vm_start, pfn,
+	ret = remap_pfn_range(vma, vma->vm_start, pfn,
 			vma->vm_end - vma->vm_start, vma->vm_page_prot);
+	if (!ret)
+		p->signal_mapped_size = vma->vm_end - vma->vm_start;
+
+	return ret;
 }
 
 /*
- * Assumes that p->event_mutex is held and of course
- * that p is not going away (current or locked).
+ * Assumes that p is not going away.
  */
 static void lookup_events_by_type_and_signal(struct kfd_process *p,
 		int type, void *event_data)
 {
 	struct kfd_hsa_memory_exception_data *ev_data;
 	struct kfd_event *ev;
-	int bkt;
+	uint32_t id;
 	bool send_signal = true;
 
 	ev_data = (struct kfd_hsa_memory_exception_data *) event_data;
 
-	hash_for_each(p->events, bkt, ev, events)
+	rcu_read_lock();
+
+	id = KFD_FIRST_NONSIGNAL_EVENT_ID;
+	idr_for_each_entry_continue(&p->event_idr, ev, id)
 		if (ev->type == type) {
 			send_signal = false;
 			dev_dbg(kfd_device,
 					"Event found: id %X type %d",
 					ev->event_id, ev->type);
+			spin_lock(&ev->lock);
 			set_event(ev);
 			if (ev->type == KFD_EVENT_TYPE_MEMORY && ev_data)
 				ev->memory_exception_data = *ev_data;
+			spin_unlock(&ev->lock);
 		}
 
+	if (type == KFD_EVENT_TYPE_MEMORY) {
+		dev_warn(kfd_device,
+			"Sending SIGSEGV to process pid %d",
+				p->lead_thread->pid);
+		send_sig(SIGSEGV, p->lead_thread, 0);
+	}
+
 	/* Send SIGTERM no event of type "type" has been found*/
 	if (send_signal) {
 		if (send_sigterm) {
 			dev_warn(kfd_device,
-				"Sending SIGTERM to HSA Process with PID %d ",
+				"Sending SIGTERM to process pid %d",
 					p->lead_thread->pid);
 			send_sig(SIGTERM, p->lead_thread, 0);
 		} else {
 			dev_err(kfd_device,
-				"HSA Process (PID %d) got unhandled exception",
+				"Process pid %d got unhandled exception",
 				p->lead_thread->pid);
 		}
 	}
+
+	rcu_read_unlock();
 }
 
-void kfd_signal_iommu_event(struct kfd_dev *dev, unsigned int pasid,
-		unsigned long address, bool is_write_requested,
-		bool is_execute_requested)
+void kfd_signal_hw_exception_event(u32 pasid)
 {
-	struct kfd_hsa_memory_exception_data memory_exception_data;
-	struct vm_area_struct *vma;
-
 	/*
 	 * Because we are called from arbitrary context (workqueue) as opposed
 	 * to process context, kfd_process could attempt to exit while we are
-	 * running so the lookup function returns a locked process.
+	 * running so the lookup function increments the process ref count.
 	 */
-	struct kfd_process *p = kfd_lookup_process_by_pasid(pasid);
+	struct kfd_process *p = kfd_lookup_process_by_pasid(pasid, NULL);
 
 	if (!p)
 		return; /* Presumably process exited. */
 
-	memset(&memory_exception_data, 0, sizeof(memory_exception_data));
+	lookup_events_by_type_and_signal(p, KFD_EVENT_TYPE_HW_EXCEPTION, NULL);
+	kfd_unref_process(p);
+}
 
-	down_read(&p->mm->mmap_sem);
-	vma = find_vma(p->mm, address);
-
-	memory_exception_data.gpu_id = dev->id;
-	memory_exception_data.va = address;
-	/* Set failure reason */
-	memory_exception_data.failure.NotPresent = 1;
-	memory_exception_data.failure.NoExecute = 0;
-	memory_exception_data.failure.ReadOnly = 0;
-	if (vma) {
-		if (vma->vm_start > address) {
-			memory_exception_data.failure.NotPresent = 1;
-			memory_exception_data.failure.NoExecute = 0;
-			memory_exception_data.failure.ReadOnly = 0;
-		} else {
-			memory_exception_data.failure.NotPresent = 0;
-			if (is_write_requested && !(vma->vm_flags & VM_WRITE))
-				memory_exception_data.failure.ReadOnly = 1;
-			else
-				memory_exception_data.failure.ReadOnly = 0;
-			if (is_execute_requested && !(vma->vm_flags & VM_EXEC))
-				memory_exception_data.failure.NoExecute = 1;
-			else
-				memory_exception_data.failure.NoExecute = 0;
+void kfd_signal_vm_fault_event_with_userptr(struct kfd_process *p, uint64_t gpu_va)
+{
+	struct kfd_process_device *pdd;
+	struct kfd_hsa_memory_exception_data exception_data;
+	int i;
+
+	memset(&exception_data, 0, sizeof(exception_data));
+	exception_data.va = gpu_va;
+	exception_data.failure.NotPresent = 1;
+
+	// Send VM seg fault to all kfd process device
+	for (i = 0; i < p->n_pdds; i++) {
+		pdd = p->pdds[i];
+		exception_data.gpu_id = pdd->user_gpu_id;
+		kfd_evict_process_device(pdd);
+		kfd_signal_vm_fault_event(pdd, NULL, &exception_data);
+	}
+}
+
+void kfd_signal_vm_fault_event(struct kfd_process_device *pdd,
+				struct kfd_vm_fault_info *info,
+				struct kfd_hsa_memory_exception_data *data)
+{
+	struct kfd_event *ev;
+	uint32_t id;
+	struct kfd_process *p = pdd->process;
+	struct kfd_hsa_memory_exception_data memory_exception_data;
+	int user_gpu_id;
+
+	user_gpu_id = kfd_process_get_user_gpu_id(p, pdd->dev->id);
+	if (unlikely(user_gpu_id == -EINVAL)) {
+		WARN_ONCE(1, "Could not get user_gpu_id from dev->id:%x\n",
+			  pdd->dev->id);
+		return;
+	}
+
+	/* SoC15 chips and onwards will pass in data from now on. */
+	if (!data) {
+		memset(&memory_exception_data, 0, sizeof(memory_exception_data));
+		memory_exception_data.gpu_id = user_gpu_id;
+		memory_exception_data.failure.imprecise = true;
+
+		/* Set failure reason */
+		if (info) {
+			memory_exception_data.va = (info->page_addr) <<
+								PAGE_SHIFT;
+			memory_exception_data.failure.NotPresent =
+				info->prot_valid ? 1 : 0;
+			memory_exception_data.failure.NoExecute =
+				info->prot_exec ? 1 : 0;
+			memory_exception_data.failure.ReadOnly =
+				info->prot_write ? 1 : 0;
+			memory_exception_data.failure.imprecise = 0;
 		}
 	}
 
-	up_read(&p->mm->mmap_sem);
+	rcu_read_lock();
 
-	mutex_lock(&p->event_mutex);
+	id = KFD_FIRST_NONSIGNAL_EVENT_ID;
+	idr_for_each_entry_continue(&p->event_idr, ev, id)
+		if (ev->type == KFD_EVENT_TYPE_MEMORY) {
+			spin_lock(&ev->lock);
+			ev->memory_exception_data = data ? *data :
+							memory_exception_data;
+			set_event(ev);
+			spin_unlock(&ev->lock);
+		}
 
-	/* Lookup events by type and signal them */
-	lookup_events_by_type_and_signal(p, KFD_EVENT_TYPE_MEMORY,
-			&memory_exception_data);
+	rcu_read_unlock();
+}
 
-	mutex_unlock(&p->event_mutex);
-	mutex_unlock(&p->mutex);
+void kfd_signal_reset_event(struct kfd_node *dev)
+{
+	struct kfd_hsa_hw_exception_data hw_exception_data;
+	struct kfd_hsa_memory_exception_data memory_exception_data;
+	struct kfd_process *p;
+	struct kfd_event *ev;
+	unsigned int temp;
+	uint32_t id, idx;
+	int reset_cause = atomic_read(&dev->sram_ecc_flag) ?
+			KFD_HW_EXCEPTION_ECC :
+			KFD_HW_EXCEPTION_GPU_HANG;
+
+	/* Whole gpu reset caused by GPU hang and memory is lost */
+	memset(&hw_exception_data, 0, sizeof(hw_exception_data));
+	hw_exception_data.memory_lost = 1;
+	hw_exception_data.reset_cause = reset_cause;
+
+	memset(&memory_exception_data, 0, sizeof(memory_exception_data));
+	memory_exception_data.ErrorType = KFD_MEM_ERR_SRAM_ECC;
+	memory_exception_data.failure.imprecise = true;
+
+	idx = srcu_read_lock(&kfd_processes_srcu);
+	hash_for_each_rcu(kfd_processes_table, temp, p, kfd_processes) {
+		int user_gpu_id = kfd_process_get_user_gpu_id(p, dev->id);
+		struct kfd_process_device *pdd = kfd_get_process_device_data(dev, p);
+
+		if (unlikely(user_gpu_id == -EINVAL)) {
+			WARN_ONCE(1, "Could not get user_gpu_id from dev->id:%x\n", dev->id);
+			continue;
+		}
+
+		if (unlikely(!pdd)) {
+			WARN_ONCE(1, "Could not get device data from process pid:%d\n",
+				  p->lead_thread->pid);
+			continue;
+		}
+
+		if (dev->dqm->detect_hang_count && !pdd->has_reset_queue)
+			continue;
+
+		if (dev->dqm->detect_hang_count) {
+			struct amdgpu_task_info *ti;
+			struct amdgpu_fpriv *drv_priv;
+
+			if (unlikely(amdgpu_file_to_fpriv(pdd->drm_file, &drv_priv))) {
+				WARN_ONCE(1, "Could not get vm for device %x from pid:%d\n",
+					  dev->id, p->lead_thread->pid);
+				continue;
+			}
+
+			ti = amdgpu_vm_get_task_info_vm(&drv_priv->vm);
+			if (ti) {
+				dev_err(dev->adev->dev,
+					"Queues reset on process %s tid %d thread %s pid %d\n",
+					ti->process_name, ti->tgid, ti->task.comm, ti->task.pid);
+				amdgpu_vm_put_task_info(ti);
+			}
+		}
+
+		rcu_read_lock();
+
+		id = KFD_FIRST_NONSIGNAL_EVENT_ID;
+		idr_for_each_entry_continue(&p->event_idr, ev, id) {
+			if (ev->type == KFD_EVENT_TYPE_HW_EXCEPTION) {
+				spin_lock(&ev->lock);
+				ev->hw_exception_data = hw_exception_data;
+				ev->hw_exception_data.gpu_id = user_gpu_id;
+				set_event(ev);
+				spin_unlock(&ev->lock);
+			}
+			if (ev->type == KFD_EVENT_TYPE_MEMORY &&
+			    reset_cause == KFD_HW_EXCEPTION_ECC) {
+				spin_lock(&ev->lock);
+				ev->memory_exception_data = memory_exception_data;
+				ev->memory_exception_data.gpu_id = user_gpu_id;
+				set_event(ev);
+				spin_unlock(&ev->lock);
+			}
+		}
+
+		rcu_read_unlock();
+	}
+	srcu_read_unlock(&kfd_processes_srcu, idx);
 }
 
-void kfd_signal_hw_exception_event(unsigned int pasid)
+void kfd_signal_poison_consumed_event(struct kfd_node *dev, u32 pasid)
 {
-	/*
-	 * Because we are called from arbitrary context (workqueue) as opposed
-	 * to process context, kfd_process could attempt to exit while we are
-	 * running so the lookup function returns a locked process.
-	 */
-	struct kfd_process *p = kfd_lookup_process_by_pasid(pasid);
+	struct kfd_process *p = kfd_lookup_process_by_pasid(pasid, NULL);
+	struct kfd_hsa_memory_exception_data memory_exception_data;
+	struct kfd_hsa_hw_exception_data hw_exception_data;
+	struct kfd_event *ev;
+	uint32_t id = KFD_FIRST_NONSIGNAL_EVENT_ID;
+	int user_gpu_id;
 
-	if (!p)
+	if (!p) {
+		dev_warn(dev->adev->dev, "Not find process with pasid:%d\n", pasid);
 		return; /* Presumably process exited. */
+	}
 
-	mutex_lock(&p->event_mutex);
+	user_gpu_id = kfd_process_get_user_gpu_id(p, dev->id);
+	if (unlikely(user_gpu_id == -EINVAL)) {
+		WARN_ONCE(1, "Could not get user_gpu_id from dev->id:%x\n", dev->id);
+		kfd_unref_process(p);
+		return;
+	}
 
-	/* Lookup events by type and signal them */
-	lookup_events_by_type_and_signal(p, KFD_EVENT_TYPE_HW_EXCEPTION, NULL);
+	memset(&hw_exception_data, 0, sizeof(hw_exception_data));
+	hw_exception_data.gpu_id = user_gpu_id;
+	hw_exception_data.memory_lost = 1;
+	hw_exception_data.reset_cause = KFD_HW_EXCEPTION_ECC;
 
-	mutex_unlock(&p->event_mutex);
-	mutex_unlock(&p->mutex);
+	memset(&memory_exception_data, 0, sizeof(memory_exception_data));
+	memory_exception_data.ErrorType = KFD_MEM_ERR_POISON_CONSUMED;
+	memory_exception_data.gpu_id = user_gpu_id;
+	memory_exception_data.failure.imprecise = true;
+
+	rcu_read_lock();
+
+	idr_for_each_entry_continue(&p->event_idr, ev, id) {
+		if (ev->type == KFD_EVENT_TYPE_HW_EXCEPTION) {
+			spin_lock(&ev->lock);
+			ev->hw_exception_data = hw_exception_data;
+			set_event(ev);
+			spin_unlock(&ev->lock);
+		}
+
+		if (ev->type == KFD_EVENT_TYPE_MEMORY) {
+			spin_lock(&ev->lock);
+			ev->memory_exception_data = memory_exception_data;
+			set_event(ev);
+			spin_unlock(&ev->lock);
+		}
+	}
+
+	dev_warn(dev->adev->dev, "Send SIGBUS to process %s(pasid:%d)\n",
+		p->lead_thread->comm, pasid);
+	rcu_read_unlock();
+
+	/* user application will handle SIGBUS signal */
+	send_sig(SIGBUS, p->lead_thread, 0);
+
+	kfd_unref_process(p);
 }
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_events.h b/drivers/gpu/drm/amd/amdkfd/kfd_events.h
index 28f6838b1f4c..52ccfd397c2b 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_events.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_events.h
@@ -1,5 +1,6 @@
+/* SPDX-License-Identifier: GPL-2.0 OR MIT */
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -27,12 +28,17 @@
 #include <linux/hashtable.h>
 #include <linux/types.h>
 #include <linux/list.h>
+#include <linux/wait.h>
 #include "kfd_priv.h"
 #include <uapi/linux/kfd_ioctl.h>
 
-#define KFD_EVENT_ID_NONSIGNAL_MASK 0x80000000U
-#define KFD_FIRST_NONSIGNAL_EVENT_ID KFD_EVENT_ID_NONSIGNAL_MASK
-#define KFD_LAST_NONSIGNAL_EVENT_ID UINT_MAX
+/*
+ * IDR supports non-negative integer IDs. Small IDs are used for
+ * signal events to match their signal slot. Use the upper half of the
+ * ID space for non-signal events.
+ */
+#define KFD_FIRST_NONSIGNAL_EVENT_ID ((INT_MAX >> 1) + 1)
+#define KFD_LAST_NONSIGNAL_EVENT_ID INT_MAX
 
 /*
  * Written into kfd_signal_slot_t to indicate that the event is not signaled.
@@ -46,27 +52,27 @@ struct kfd_event_waiter;
 struct signal_page;
 
 struct kfd_event {
-	/* All events in process, rooted at kfd_process.events. */
-	struct hlist_node events;
-
 	u32 event_id;
+	u64 event_age;
 
 	bool signaled;
 	bool auto_reset;
 
 	int type;
 
-	struct list_head waiters; /* List of kfd_event_waiter by waiters. */
+	spinlock_t lock;
+	wait_queue_head_t wq; /* List of event waiters. */
 
 	/* Only for signal events. */
-	struct signal_page *signal_page;
-	unsigned int signal_slot_index;
 	uint64_t __user *user_signal_address;
 
 	/* type specific data */
 	union {
 		struct kfd_hsa_memory_exception_data memory_exception_data;
+		struct kfd_hsa_hw_exception_data hw_exception_data;
 	};
+
+	struct rcu_head rcu; /* for asynchronous kfree_rcu */
 };
 
 #define KFD_EVENT_TIMEOUT_IMMEDIATE 0
@@ -78,7 +84,7 @@ struct kfd_event {
 #define KFD_EVENT_TYPE_DEBUG 5
 #define KFD_EVENT_TYPE_MEMORY 8
 
-extern void kfd_signal_event_interrupt(unsigned int pasid, uint32_t partial_id,
-					uint32_t valid_id_bits);
+extern void kfd_signal_event_interrupt(u32 pasid, uint32_t partial_id,
+				       uint32_t valid_id_bits);
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_flat_memory.c b/drivers/gpu/drm/amd/amdkfd/kfd_flat_memory.c
index c59384bbbc5f..1d170dc50df3 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_flat_memory.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_flat_memory.c
@@ -1,5 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -22,7 +23,6 @@
  */
 
 #include <linux/device.h>
-#include <linux/export.h>
 #include <linux/err.h>
 #include <linux/fs.h>
 #include <linux/sched.h>
@@ -34,7 +34,8 @@
 #include "kfd_priv.h"
 #include <linux/mm.h>
 #include <linux/mman.h>
-#include <asm/processor.h>
+#include <linux/processor.h>
+#include "amdgpu_vm.h"
 
 /*
  * The primary memory I/O features being added for revisions of gfxip
@@ -275,41 +276,115 @@
  * for FLAT_* / S_LOAD operations.
  */
 
-#define MAKE_GPUVM_APP_BASE(gpu_num) \
+#define MAKE_GPUVM_APP_BASE_VI(gpu_num) \
 	(((uint64_t)(gpu_num) << 61) + 0x1000000000000L)
 
-#define MAKE_GPUVM_APP_LIMIT(base) \
-	(((uint64_t)(base) & \
-		0xFFFFFF0000000000UL) | 0xFFFFFFFFFFL)
+#define MAKE_GPUVM_APP_LIMIT(base, size) \
+	(((uint64_t)(base) & 0xFFFFFF0000000000UL) + (size) - 1)
 
-#define MAKE_SCRATCH_APP_BASE(gpu_num) \
-	(((uint64_t)(gpu_num) << 61) + 0x100000000L)
+#define MAKE_SCRATCH_APP_BASE_VI() \
+	(((uint64_t)(0x1UL) << 61) + 0x100000000L)
 
 #define MAKE_SCRATCH_APP_LIMIT(base) \
 	(((uint64_t)base & 0xFFFFFFFF00000000UL) | 0xFFFFFFFF)
 
-#define MAKE_LDS_APP_BASE(gpu_num) \
-	(((uint64_t)(gpu_num) << 61) + 0x0)
+#define MAKE_LDS_APP_BASE_VI() \
+	(((uint64_t)(0x1UL) << 61) + 0x0)
 #define MAKE_LDS_APP_LIMIT(base) \
 	(((uint64_t)(base) & 0xFFFFFFFF00000000UL) | 0xFFFFFFFF)
 
+/* On GFXv9 the LDS and scratch apertures are programmed independently
+ * using the high 16 bits of the 64-bit virtual address. They must be
+ * in the hole, which will be the case as long as the high 16 bits are
+ * not 0.
+ *
+ * The aperture sizes are still 4GB implicitly.
+ *
+ * A GPUVM aperture is not applicable on GFXv9.
+ */
+#define MAKE_LDS_APP_BASE_V9() ((uint64_t)(0x1UL) << 48)
+#define MAKE_SCRATCH_APP_BASE_V9() ((uint64_t)(0x2UL) << 48)
+
+/* User mode manages most of the SVM aperture address space. The low
+ * 16MB are reserved for kernel use (CWSR trap handler and kernel IB
+ * for now).
+ */
+#define SVM_USER_BASE (u64)(KFD_CWSR_TBA_TMA_SIZE + 2*PAGE_SIZE)
+#define SVM_CWSR_BASE (SVM_USER_BASE - KFD_CWSR_TBA_TMA_SIZE)
+#define SVM_IB_BASE   (SVM_CWSR_BASE - PAGE_SIZE)
+
+static void kfd_init_apertures_vi(struct kfd_process_device *pdd, uint8_t id)
+{
+	/*
+	 * node id couldn't be 0 - the three MSB bits of
+	 * aperture shouldn't be 0
+	 */
+	pdd->lds_base = MAKE_LDS_APP_BASE_VI();
+	pdd->lds_limit = MAKE_LDS_APP_LIMIT(pdd->lds_base);
+
+	/* dGPUs: SVM aperture starting at 0
+	 * with small reserved space for kernel.
+	 * Set them to CANONICAL addresses.
+	 */
+	pdd->gpuvm_base = max(SVM_USER_BASE, AMDGPU_VA_RESERVED_BOTTOM);
+	pdd->gpuvm_limit =
+		pdd->dev->kfd->shared_resources.gpuvm_size - 1;
+
+	/* dGPUs: the reserved space for kernel
+	 * before SVM
+	 */
+	pdd->qpd.cwsr_base = SVM_CWSR_BASE;
+	pdd->qpd.ib_base = SVM_IB_BASE;
+
+	pdd->scratch_base = MAKE_SCRATCH_APP_BASE_VI();
+	pdd->scratch_limit = MAKE_SCRATCH_APP_LIMIT(pdd->scratch_base);
+}
+
+static void kfd_init_apertures_v9(struct kfd_process_device *pdd, uint8_t id)
+{
+	pdd->lds_base = MAKE_LDS_APP_BASE_V9();
+	pdd->lds_limit = MAKE_LDS_APP_LIMIT(pdd->lds_base);
+
+	pdd->gpuvm_base = AMDGPU_VA_RESERVED_BOTTOM;
+	pdd->gpuvm_limit =
+		pdd->dev->kfd->shared_resources.gpuvm_size - 1;
+
+	pdd->scratch_base = MAKE_SCRATCH_APP_BASE_V9();
+	pdd->scratch_limit = MAKE_SCRATCH_APP_LIMIT(pdd->scratch_base);
+
+	/*
+	 * Place TBA/TMA on opposite side of VM hole to prevent
+	 * stray faults from triggering SVM on these pages.
+	 */
+	pdd->qpd.cwsr_base = AMDGPU_VA_RESERVED_TRAP_START(pdd->dev->adev);
+}
+
 int kfd_init_apertures(struct kfd_process *process)
 {
 	uint8_t id  = 0;
-	struct kfd_dev *dev;
+	struct kfd_node *dev;
 	struct kfd_process_device *pdd;
 
 	/*Iterating over all devices*/
-	while ((dev = kfd_topology_enum_kfd_devices(id)) != NULL &&
-		id < NUM_OF_SUPPORTED_GPUS) {
+	while (kfd_topology_enum_kfd_devices(id, &dev) == 0) {
+		if (!dev || kfd_devcgroup_check_permission(dev)) {
+			/* Skip non GPU devices and devices to which the
+			 * current process have no access to. Access can be
+			 * limited by placing the process in a specific
+			 * cgroup hierarchy
+			 */
+			id++;
+			continue;
+		}
 
 		pdd = kfd_create_process_device_data(dev, process);
 		if (!pdd) {
-			pr_err("Failed to create process device data\n");
-			return -1;
+			dev_err(dev->adev->dev,
+				"Failed to create process device data\n");
+			return -ENOMEM;
 		}
 		/*
-		 * For 64 bit process aperture will be statically reserved in
+		 * For 64 bit process apertures will be statically reserved in
 		 * the x86_64 non canonical process address space
 		 * amdkfd doesn't currently support apertures for 32 bit process
 		 */
@@ -318,23 +393,27 @@ int kfd_init_apertures(struct kfd_process *process)
 			pdd->gpuvm_base = pdd->gpuvm_limit = 0;
 			pdd->scratch_base = pdd->scratch_limit = 0;
 		} else {
-			/*
-			 * node id couldn't be 0 - the three MSB bits of
-			 * aperture shoudn't be 0
-			 */
-			pdd->lds_base = MAKE_LDS_APP_BASE(id + 1);
-
-			pdd->lds_limit = MAKE_LDS_APP_LIMIT(pdd->lds_base);
-
-			pdd->gpuvm_base = MAKE_GPUVM_APP_BASE(id + 1);
-
-			pdd->gpuvm_limit =
-					MAKE_GPUVM_APP_LIMIT(pdd->gpuvm_base);
-
-			pdd->scratch_base = MAKE_SCRATCH_APP_BASE(id + 1);
-
-			pdd->scratch_limit =
-				MAKE_SCRATCH_APP_LIMIT(pdd->scratch_base);
+			switch (dev->adev->asic_type) {
+			case CHIP_KAVERI:
+			case CHIP_HAWAII:
+			case CHIP_CARRIZO:
+			case CHIP_TONGA:
+			case CHIP_FIJI:
+			case CHIP_POLARIS10:
+			case CHIP_POLARIS11:
+			case CHIP_POLARIS12:
+			case CHIP_VEGAM:
+				kfd_init_apertures_vi(pdd, id);
+				break;
+			default:
+				if (KFD_GC_VERSION(dev) >= IP_VERSION(9, 0, 1))
+					kfd_init_apertures_v9(pdd, id);
+				else {
+					WARN(1, "Unexpected ASIC family %u",
+					     dev->adev->asic_type);
+					return -EINVAL;
+				}
+			}
 		}
 
 		dev_dbg(kfd_device, "node id %u\n", id);
@@ -351,5 +430,3 @@ int kfd_init_apertures(struct kfd_process *process)
 
 	return 0;
 }
-
-
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v10.c b/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v10.c
new file mode 100644
index 000000000000..3e1ad8974797
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v10.c
@@ -0,0 +1,386 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include "kfd_events.h"
+#include "kfd_debug.h"
+#include "soc15_int.h"
+#include "kfd_device_queue_manager.h"
+
+/*
+ * GFX10 SQ Interrupts
+ *
+ * There are 3 encoding types of interrupts sourced from SQ sent as a 44-bit
+ * packet to the Interrupt Handler:
+ * Auto - Generated by the SQG (various cmd overflows, timestamps etc)
+ * Wave - Generated by S_SENDMSG through a shader program
+ * Error - HW generated errors (Illegal instructions, Memviols, EDC etc)
+ *
+ * The 44-bit packet is mapped as {context_id1[7:0],context_id0[31:0]} plus
+ * 4-bits for VMID (SOC15_VMID_FROM_IH_ENTRY) as such:
+ *
+ * - context_id1[7:6]
+ * Encoding type (0 = Auto, 1 = Wave, 2 = Error)
+ *
+ * - context_id0[24]
+ * PRIV bit indicates that Wave S_SEND or error occurred within trap
+ *
+ * - context_id0[22:0]
+ * 23-bit data with the following layout per encoding type:
+ * Auto - only context_id0[8:0] is used, which reports various interrupts
+ * generated by SQG.  The rest is 0.
+ * Wave - user data sent from m0 via S_SENDMSG
+ * Error - Error type (context_id0[22:19]), Error Details (rest of bits)
+ *
+ * The other context_id bits show coordinates (SE/SH/CU/SIMD/WGP) for wave
+ * S_SENDMSG and Errors.  These are 0 for Auto.
+ */
+
+enum SQ_INTERRUPT_WORD_ENCODING {
+	SQ_INTERRUPT_WORD_ENCODING_AUTO = 0x0,
+	SQ_INTERRUPT_WORD_ENCODING_INST,
+	SQ_INTERRUPT_WORD_ENCODING_ERROR,
+};
+
+enum SQ_INTERRUPT_ERROR_TYPE {
+	SQ_INTERRUPT_ERROR_TYPE_EDC_FUE = 0x0,
+	SQ_INTERRUPT_ERROR_TYPE_ILLEGAL_INST,
+	SQ_INTERRUPT_ERROR_TYPE_MEMVIOL,
+	SQ_INTERRUPT_ERROR_TYPE_EDC_FED,
+};
+
+/* SQ_INTERRUPT_WORD_AUTO_CTXID */
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__THREAD_TRACE__SHIFT 0
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__WLT__SHIFT 1
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__THREAD_TRACE_BUF0_FULL__SHIFT 2
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__THREAD_TRACE_BUF1_FULL__SHIFT 3
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__THREAD_TRACE_UTC_ERROR__SHIFT 7
+#define SQ_INTERRUPT_WORD_AUTO_CTXID1__SE_ID__SHIFT 4
+#define SQ_INTERRUPT_WORD_AUTO_CTXID1__ENCODING__SHIFT 6
+
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__THREAD_TRACE_MASK 0x00000001
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__WLT_MASK 0x00000002
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__THREAD_TRACE_BUF0_FULL_MASK 0x00000004
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__THREAD_TRACE_BUF1_FULL_MASK 0x00000008
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__THREAD_TRACE_UTC_ERROR_MASK 0x00000080
+#define SQ_INTERRUPT_WORD_AUTO_CTXID1__SE_ID_MASK 0x030
+#define SQ_INTERRUPT_WORD_AUTO_CTXID1__ENCODING_MASK 0x0c0
+
+/* SQ_INTERRUPT_WORD_WAVE_CTXID */
+#define SQ_INTERRUPT_WORD_WAVE_CTXID0__DATA__SHIFT 0
+#define SQ_INTERRUPT_WORD_WAVE_CTXID0__SA_ID__SHIFT 23
+#define SQ_INTERRUPT_WORD_WAVE_CTXID0__PRIV__SHIFT 24
+#define SQ_INTERRUPT_WORD_WAVE_CTXID0__WAVE_ID__SHIFT 25
+#define SQ_INTERRUPT_WORD_WAVE_CTXID0__SIMD_ID__SHIFT 30
+#define SQ_INTERRUPT_WORD_WAVE_CTXID1__WGP_ID__SHIFT 0
+#define SQ_INTERRUPT_WORD_WAVE_CTXID1__SE_ID__SHIFT 4
+#define SQ_INTERRUPT_WORD_WAVE_CTXID1__ENCODING__SHIFT 6
+
+#define SQ_INTERRUPT_WORD_WAVE_CTXID0__DATA_MASK 0x000007fffff
+#define SQ_INTERRUPT_WORD_WAVE_CTXID0__SA_ID_MASK 0x0000800000
+#define SQ_INTERRUPT_WORD_WAVE_CTXID0__PRIV_MASK 0x00001000000
+#define SQ_INTERRUPT_WORD_WAVE_CTXID0__WAVE_ID_MASK 0x0003e000000
+#define SQ_INTERRUPT_WORD_WAVE_CTXID0__SIMD_ID_MASK 0x000c0000000
+#define SQ_INTERRUPT_WORD_WAVE_CTXID1__WGP_ID_MASK 0x00f
+#define SQ_INTERRUPT_WORD_WAVE_CTXID1__SE_ID_MASK 0x030
+#define SQ_INTERRUPT_WORD_WAVE_CTXID1__ENCODING_MASK 0x0c0
+
+#define KFD_CTXID0__ERR_TYPE_MASK 0x780000
+#define KFD_CTXID0__ERR_TYPE__SHIFT 19
+
+/* GFX10 SQ interrupt ENC type bit (context_id1[7:6]) for wave s_sendmsg */
+#define KFD_CONTEXT_ID1_ENC_TYPE_WAVE_MASK	0x40
+/* GFX10 SQ interrupt PRIV bit (context_id0[24]) for s_sendmsg inside trap */
+#define KFD_CONTEXT_ID0_PRIV_MASK		0x1000000
+/*
+ * The debugger will send user data(m0) with PRIV=1 to indicate it requires
+ * notification from the KFD with the following queue id (DOORBELL_ID) and
+ * trap code (TRAP_CODE).
+ */
+#define KFD_CONTEXT_ID0_DEBUG_DOORBELL_MASK	0x0003ff
+#define KFD_CONTEXT_ID0_DEBUG_TRAP_CODE_SHIFT	10
+#define KFD_CONTEXT_ID0_DEBUG_TRAP_CODE_MASK	0x07fc00
+#define KFD_DEBUG_DOORBELL_ID(ctxid0)	((ctxid0) &	\
+				KFD_CONTEXT_ID0_DEBUG_DOORBELL_MASK)
+#define KFD_DEBUG_TRAP_CODE(ctxid0)	(((ctxid0) &	\
+				KFD_CONTEXT_ID0_DEBUG_TRAP_CODE_MASK)	\
+				>> KFD_CONTEXT_ID0_DEBUG_TRAP_CODE_SHIFT)
+#define KFD_DEBUG_CP_BAD_OP_ECODE_MASK		0x3fffc00
+#define KFD_DEBUG_CP_BAD_OP_ECODE_SHIFT		10
+#define KFD_DEBUG_CP_BAD_OP_ECODE(ctxid0) (((ctxid0) &			\
+				KFD_DEBUG_CP_BAD_OP_ECODE_MASK)		\
+				>> KFD_DEBUG_CP_BAD_OP_ECODE_SHIFT)
+
+static bool event_interrupt_isr_v10(struct kfd_node *dev,
+					const uint32_t *ih_ring_entry,
+					uint32_t *patched_ihre,
+					bool *patched_flag)
+{
+	uint16_t source_id, client_id, pasid, vmid;
+	const uint32_t *data = ih_ring_entry;
+
+	source_id = SOC15_SOURCE_ID_FROM_IH_ENTRY(ih_ring_entry);
+	client_id = SOC15_CLIENT_ID_FROM_IH_ENTRY(ih_ring_entry);
+
+	/* Only handle interrupts from KFD VMIDs */
+	vmid = SOC15_VMID_FROM_IH_ENTRY(ih_ring_entry);
+	if (!KFD_IRQ_IS_FENCE(client_id, source_id) &&
+	   (vmid < dev->vm_info.first_vmid_kfd ||
+	    vmid > dev->vm_info.last_vmid_kfd))
+		return false;
+
+	pasid = SOC15_PASID_FROM_IH_ENTRY(ih_ring_entry);
+
+	/* Only handle clients we care about */
+	if (client_id != SOC15_IH_CLIENTID_GRBM_CP &&
+	    client_id != SOC15_IH_CLIENTID_SDMA0 &&
+	    client_id != SOC15_IH_CLIENTID_SDMA1 &&
+	    client_id != SOC15_IH_CLIENTID_SDMA2 &&
+	    client_id != SOC15_IH_CLIENTID_SDMA3 &&
+	    client_id != SOC15_IH_CLIENTID_SDMA4 &&
+	    client_id != SOC15_IH_CLIENTID_SDMA5 &&
+	    client_id != SOC15_IH_CLIENTID_SDMA6 &&
+	    client_id != SOC15_IH_CLIENTID_SDMA7 &&
+	    client_id != SOC15_IH_CLIENTID_VMC &&
+	    client_id != SOC15_IH_CLIENTID_VMC1 &&
+	    client_id != SOC15_IH_CLIENTID_UTCL2 &&
+	    client_id != SOC15_IH_CLIENTID_SE0SH &&
+	    client_id != SOC15_IH_CLIENTID_SE1SH &&
+	    client_id != SOC15_IH_CLIENTID_SE2SH &&
+	    client_id != SOC15_IH_CLIENTID_SE3SH)
+		return false;
+
+	dev_dbg(dev->adev->dev,
+		"client id 0x%x, source id %d, vmid %d, pasid 0x%x. raw data:\n",
+		client_id, source_id, vmid, pasid);
+	dev_dbg(dev->adev->dev, "%8X, %8X, %8X, %8X, %8X, %8X, %8X, %8X.\n",
+		data[0], data[1], data[2], data[3], data[4], data[5], data[6],
+		data[7]);
+
+	if (pasid == 0)
+		return 0;
+
+	/* Interrupt types we care about: various signals and faults.
+	 * They will be forwarded to a work queue (see below).
+	 */
+	return source_id == SOC15_INTSRC_CP_END_OF_PIPE ||
+		source_id == SOC15_INTSRC_SDMA_TRAP ||
+		source_id == SOC15_INTSRC_SQ_INTERRUPT_MSG ||
+		source_id == SOC15_INTSRC_CP_BAD_OPCODE ||
+		client_id == SOC15_IH_CLIENTID_VMC ||
+		client_id == SOC15_IH_CLIENTID_VMC1 ||
+		client_id == SOC15_IH_CLIENTID_UTCL2 ||
+		KFD_IRQ_IS_FENCE(client_id, source_id);
+}
+
+static void event_interrupt_wq_v10(struct kfd_node *dev,
+					const uint32_t *ih_ring_entry)
+{
+	uint16_t source_id, client_id, pasid, vmid;
+	uint32_t context_id0, context_id1;
+	uint32_t encoding, sq_intr_err_type;
+
+	source_id = SOC15_SOURCE_ID_FROM_IH_ENTRY(ih_ring_entry);
+	client_id = SOC15_CLIENT_ID_FROM_IH_ENTRY(ih_ring_entry);
+	pasid = SOC15_PASID_FROM_IH_ENTRY(ih_ring_entry);
+	vmid = SOC15_VMID_FROM_IH_ENTRY(ih_ring_entry);
+	context_id0 = SOC15_CONTEXT_ID0_FROM_IH_ENTRY(ih_ring_entry);
+	context_id1 = SOC15_CONTEXT_ID1_FROM_IH_ENTRY(ih_ring_entry);
+
+	if (client_id == SOC15_IH_CLIENTID_GRBM_CP ||
+	    client_id == SOC15_IH_CLIENTID_SE0SH ||
+	    client_id == SOC15_IH_CLIENTID_SE1SH ||
+	    client_id == SOC15_IH_CLIENTID_SE2SH ||
+	    client_id == SOC15_IH_CLIENTID_SE3SH) {
+		if (source_id == SOC15_INTSRC_CP_END_OF_PIPE)
+			kfd_signal_event_interrupt(pasid, context_id0, 32);
+		else if (source_id == SOC15_INTSRC_SQ_INTERRUPT_MSG) {
+			encoding = REG_GET_FIELD(context_id1,
+						SQ_INTERRUPT_WORD_WAVE_CTXID1, ENCODING);
+			switch (encoding) {
+			case SQ_INTERRUPT_WORD_ENCODING_AUTO:
+				dev_dbg_ratelimited(
+					dev->adev->dev,
+					"sq_intr: auto, se %d, ttrace %d, wlt %d, ttrac_buf0_full %d, ttrac_buf1_full %d, ttrace_utc_err %d\n",
+					REG_GET_FIELD(
+						context_id1,
+						SQ_INTERRUPT_WORD_AUTO_CTXID1,
+						SE_ID),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_AUTO_CTXID0,
+						THREAD_TRACE),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_AUTO_CTXID0,
+						WLT),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_AUTO_CTXID0,
+						THREAD_TRACE_BUF0_FULL),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_AUTO_CTXID0,
+						THREAD_TRACE_BUF1_FULL),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_AUTO_CTXID0,
+						THREAD_TRACE_UTC_ERROR));
+				break;
+			case SQ_INTERRUPT_WORD_ENCODING_INST:
+				dev_dbg_ratelimited(
+					dev->adev->dev,
+					"sq_intr: inst, se %d, data 0x%x, sa %d, priv %d, wave_id %d, simd_id %d, wgp_id %d\n",
+					REG_GET_FIELD(
+						context_id1,
+						SQ_INTERRUPT_WORD_WAVE_CTXID1,
+						SE_ID),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID0,
+						DATA),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID0,
+						SA_ID),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID0,
+						PRIV),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID0,
+						WAVE_ID),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID0,
+						SIMD_ID),
+					REG_GET_FIELD(
+						context_id1,
+						SQ_INTERRUPT_WORD_WAVE_CTXID1,
+						WGP_ID));
+				if (context_id0 & SQ_INTERRUPT_WORD_WAVE_CTXID0__PRIV_MASK) {
+					if (kfd_set_dbg_ev_from_interrupt(dev, pasid,
+							KFD_DEBUG_DOORBELL_ID(context_id0),
+							KFD_DEBUG_TRAP_CODE(context_id0),
+							NULL, 0))
+						return;
+				}
+				break;
+			case SQ_INTERRUPT_WORD_ENCODING_ERROR:
+				sq_intr_err_type = REG_GET_FIELD(context_id0, KFD_CTXID0,
+								ERR_TYPE);
+				dev_warn_ratelimited(
+					dev->adev->dev,
+					"sq_intr: error, se %d, data 0x%x, sa %d, priv %d, wave_id %d, simd_id %d, wgp_id %d, err_type %d\n",
+					REG_GET_FIELD(
+						context_id1,
+						SQ_INTERRUPT_WORD_WAVE_CTXID1,
+						SE_ID),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID0,
+						DATA),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID0,
+						SA_ID),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID0,
+						PRIV),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID0,
+						WAVE_ID),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID0,
+						SIMD_ID),
+					REG_GET_FIELD(
+						context_id1,
+						SQ_INTERRUPT_WORD_WAVE_CTXID1,
+						WGP_ID),
+					sq_intr_err_type);
+				break;
+			default:
+				break;
+			}
+			kfd_signal_event_interrupt(pasid, context_id0 & 0x7fffff, 23);
+		} else if (source_id == SOC15_INTSRC_CP_BAD_OPCODE &&
+			   KFD_DBG_EC_TYPE_IS_PACKET(KFD_DEBUG_CP_BAD_OP_ECODE(context_id0))) {
+			kfd_set_dbg_ev_from_interrupt(dev, pasid,
+				KFD_DEBUG_DOORBELL_ID(context_id0),
+				KFD_EC_MASK(KFD_DEBUG_CP_BAD_OP_ECODE(context_id0)),
+				NULL,
+				0);
+		}
+	} else if (client_id == SOC15_IH_CLIENTID_SDMA0 ||
+		   client_id == SOC15_IH_CLIENTID_SDMA1 ||
+		   client_id == SOC15_IH_CLIENTID_SDMA2 ||
+		   client_id == SOC15_IH_CLIENTID_SDMA3 ||
+		   (client_id == SOC15_IH_CLIENTID_SDMA3_Sienna_Cichlid &&
+		    KFD_GC_VERSION(dev) == IP_VERSION(10, 3, 0)) ||
+		   client_id == SOC15_IH_CLIENTID_SDMA4 ||
+		   client_id == SOC15_IH_CLIENTID_SDMA5 ||
+		   client_id == SOC15_IH_CLIENTID_SDMA6 ||
+		   client_id == SOC15_IH_CLIENTID_SDMA7) {
+		if (source_id == SOC15_INTSRC_SDMA_TRAP) {
+			kfd_signal_event_interrupt(pasid, context_id0 & 0xfffffff, 28);
+		}
+	} else if (client_id == SOC15_IH_CLIENTID_VMC ||
+		   client_id == SOC15_IH_CLIENTID_VMC1 ||
+		   client_id == SOC15_IH_CLIENTID_UTCL2) {
+		struct kfd_vm_fault_info info = {0};
+		uint16_t ring_id = SOC15_RING_ID_FROM_IH_ENTRY(ih_ring_entry);
+		struct kfd_hsa_memory_exception_data exception_data;
+
+		info.vmid = vmid;
+		info.mc_id = client_id;
+		info.page_addr = ih_ring_entry[4] |
+			(uint64_t)(ih_ring_entry[5] & 0xf) << 32;
+		info.prot_valid = ring_id & 0x08;
+		info.prot_read  = ring_id & 0x10;
+		info.prot_write = ring_id & 0x20;
+
+		memset(&exception_data, 0, sizeof(exception_data));
+		exception_data.gpu_id = dev->id;
+		exception_data.va = (info.page_addr) << PAGE_SHIFT;
+		exception_data.failure.NotPresent = info.prot_valid ? 1 : 0;
+		exception_data.failure.NoExecute = info.prot_exec ? 1 : 0;
+		exception_data.failure.ReadOnly = info.prot_write ? 1 : 0;
+		exception_data.failure.imprecise = 0;
+
+		kfd_set_dbg_ev_from_interrupt(dev,
+						pasid,
+						-1,
+						KFD_EC_MASK(EC_DEVICE_MEMORY_VIOLATION),
+						&exception_data,
+						sizeof(exception_data));
+	} else if (KFD_IRQ_IS_FENCE(client_id, source_id)) {
+		kfd_process_close_interrupt_drain(pasid);
+	}
+}
+
+const struct kfd_event_interrupt_class event_interrupt_class_v10 = {
+	.interrupt_isr = event_interrupt_isr_v10,
+	.interrupt_wq = event_interrupt_wq_v10,
+};
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v11.c b/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v11.c
new file mode 100644
index 000000000000..2788a52714d1
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v11.c
@@ -0,0 +1,418 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include "kfd_priv.h"
+#include "kfd_events.h"
+#include "soc15_int.h"
+#include "kfd_device_queue_manager.h"
+#include "ivsrcid/vmc/irqsrcs_vmc_1_0.h"
+#include "kfd_smi_events.h"
+#include "kfd_debug.h"
+
+/*
+ * GFX11 SQ Interrupts
+ *
+ * There are 3 encoding types of interrupts sourced from SQ sent as a 44-bit
+ * packet to the Interrupt Handler:
+ * Auto - Generated by the SQG (various cmd overflows, timestamps etc)
+ * Wave - Generated by S_SENDMSG through a shader program
+ * Error - HW generated errors (Illegal instructions, Memviols, EDC etc)
+ *
+ * The 44-bit packet is mapped as {context_id1[7:0],context_id0[31:0]} plus
+ * 4-bits for VMID (SOC15_VMID_FROM_IH_ENTRY) as such:
+ *
+ * - context_id1[7:6]
+ * Encoding type (0 = Auto, 1 = Wave, 2 = Error)
+ *
+ * - context_id0[26]
+ * PRIV bit indicates that Wave S_SEND or error occurred within trap
+ *
+ * - context_id0[24:0]
+ * 25-bit data with the following layout per encoding type:
+ * Auto - only context_id0[8:0] is used, which reports various interrupts
+ * generated by SQG.  The rest is 0.
+ * Wave - user data sent from m0 via S_SENDMSG (context_id0[23:0])
+ * Error - Error Type (context_id0[24:21]), Error Details (context_id0[20:0])
+ *
+ * The other context_id bits show coordinates (SE/SH/CU/SIMD/WGP) for wave
+ * S_SENDMSG and Errors.  These are 0 for Auto.
+ */
+
+enum SQ_INTERRUPT_WORD_ENCODING {
+	SQ_INTERRUPT_WORD_ENCODING_AUTO = 0x0,
+	SQ_INTERRUPT_WORD_ENCODING_INST,
+	SQ_INTERRUPT_WORD_ENCODING_ERROR,
+};
+
+enum SQ_INTERRUPT_ERROR_TYPE {
+	SQ_INTERRUPT_ERROR_TYPE_EDC_FUE = 0x0,
+	SQ_INTERRUPT_ERROR_TYPE_ILLEGAL_INST,
+	SQ_INTERRUPT_ERROR_TYPE_MEMVIOL,
+	SQ_INTERRUPT_ERROR_TYPE_EDC_FED,
+};
+
+/* SQ_INTERRUPT_WORD_AUTO_CTXID */
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__THREAD_TRACE__SHIFT		0
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__WLT__SHIFT			1
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__THREAD_TRACE_BUF_FULL__SHIFT	2
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__REG_TIMESTAMP__SHIFT		3
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__CMD_TIMESTAMP__SHIFT		4
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__HOST_CMD_OVERFLOW__SHIFT		5
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__HOST_REG_OVERFLOW__SHIFT		6
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__IMMED_OVERFLOW__SHIFT		7
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__THREAD_TRACE_UTC_ERROR__SHIFT	8
+#define SQ_INTERRUPT_WORD_AUTO_CTXID1__ENCODING__SHIFT			6
+
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__THREAD_TRACE_MASK		0x00000001
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__WLT_MASK				0x00000002
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__THREAD_TRACE_BUF_FULL_MASK	0x00000004
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__REG_TIMESTAMP_MASK		0x00000008
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__CMD_TIMESTAMP_MASK		0x00000010
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__HOST_CMD_OVERFLOW_MASK		0x00000020
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__HOST_REG_OVERFLOW_MASK		0x00000040
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__IMMED_OVERFLOW_MASK		0x00000080
+#define SQ_INTERRUPT_WORD_AUTO_CTXID0__THREAD_TRACE_UTC_ERROR_MASK	0x00000100
+#define SQ_INTERRUPT_WORD_AUTO_CTXID1__ENCODING_MASK			0x000000c0
+
+/* SQ_INTERRUPT_WORD_WAVE_CTXID */
+#define SQ_INTERRUPT_WORD_WAVE_CTXID0__DATA__SHIFT	0
+#define SQ_INTERRUPT_WORD_WAVE_CTXID0__SH_ID__SHIFT	25
+#define SQ_INTERRUPT_WORD_WAVE_CTXID0__PRIV__SHIFT	26
+#define SQ_INTERRUPT_WORD_WAVE_CTXID0__WAVE_ID__SHIFT	27
+#define SQ_INTERRUPT_WORD_WAVE_CTXID1__SIMD_ID__SHIFT	0
+#define SQ_INTERRUPT_WORD_WAVE_CTXID1__WGP_ID__SHIFT	2
+#define SQ_INTERRUPT_WORD_WAVE_CTXID1__ENCODING__SHIFT	6
+
+#define SQ_INTERRUPT_WORD_WAVE_CTXID0__DATA_MASK	0x00ffffff /* [23:0] */
+#define SQ_INTERRUPT_WORD_WAVE_CTXID0__SH_ID_MASK	0x02000000 /* [25] */
+#define SQ_INTERRUPT_WORD_WAVE_CTXID0__PRIV_MASK	0x04000000 /* [26] */
+#define SQ_INTERRUPT_WORD_WAVE_CTXID0__WAVE_ID_MASK	0xf8000000 /* [31:27] */
+#define SQ_INTERRUPT_WORD_WAVE_CTXID1__SIMD_ID_MASK	0x00000003 /* [33:32] */
+#define SQ_INTERRUPT_WORD_WAVE_CTXID1__WGP_ID_MASK	0x0000003c /* [37:34] */
+#define SQ_INTERRUPT_WORD_WAVE_CTXID1__ENCODING_MASK	0x000000c0 /* [39:38] */
+
+/* SQ_INTERRUPT_WORD_ERROR_CTXID */
+#define SQ_INTERRUPT_WORD_ERROR_CTXID0__DETAIL__SHIFT	0
+#define SQ_INTERRUPT_WORD_ERROR_CTXID0__TYPE__SHIFT	21
+#define SQ_INTERRUPT_WORD_ERROR_CTXID0__SH_ID__SHIFT	25
+#define SQ_INTERRUPT_WORD_ERROR_CTXID0__PRIV__SHIFT	26
+#define SQ_INTERRUPT_WORD_ERROR_CTXID0__WAVE_ID__SHIFT	27
+#define SQ_INTERRUPT_WORD_ERROR_CTXID1__SIMD_ID__SHIFT	0
+#define SQ_INTERRUPT_WORD_ERROR_CTXID1__WGP_ID__SHIFT	2
+#define SQ_INTERRUPT_WORD_ERROR_CTXID1__ENCODING__SHIFT	6
+
+#define SQ_INTERRUPT_WORD_ERROR_CTXID0__DETAIL_MASK	0x001fffff /* [20:0] */
+#define SQ_INTERRUPT_WORD_ERROR_CTXID0__TYPE_MASK	0x01e00000 /* [24:21] */
+#define SQ_INTERRUPT_WORD_ERROR_CTXID0__SH_ID_MASK	0x02000000 /* [25] */
+#define SQ_INTERRUPT_WORD_ERROR_CTXID0__PRIV_MASK	0x04000000 /* [26] */
+#define SQ_INTERRUPT_WORD_ERROR_CTXID0__WAVE_ID_MASK	0xf8000000 /* [31:27] */
+#define SQ_INTERRUPT_WORD_ERROR_CTXID1__SIMD_ID_MASK	0x00000003 /* [33:32] */
+#define SQ_INTERRUPT_WORD_ERROR_CTXID1__WGP_ID_MASK	0x0000003c /* [37:34] */
+#define SQ_INTERRUPT_WORD_ERROR_CTXID1__ENCODING_MASK	0x000000c0 /* [39:38] */
+
+/*
+ * The debugger will send user data(m0) with PRIV=1 to indicate it requires
+ * notification from the KFD with the following queue id (DOORBELL_ID) and
+ * trap code (TRAP_CODE).
+ */
+#define KFD_CTXID0_TRAP_CODE_SHIFT	10
+#define KFD_CTXID0_TRAP_CODE_MASK	0xfffc00
+#define KFD_CTXID0_CP_BAD_OP_ECODE_MASK	0x3ffffff
+#define KFD_CTXID0_DOORBELL_ID_MASK	0x0003ff
+
+#define KFD_CTXID0_TRAP_CODE(ctxid0)		(((ctxid0) &  \
+				KFD_CTXID0_TRAP_CODE_MASK) >> \
+				KFD_CTXID0_TRAP_CODE_SHIFT)
+#define KFD_CTXID0_CP_BAD_OP_ECODE(ctxid0)	(((ctxid0) &        \
+				KFD_CTXID0_CP_BAD_OP_ECODE_MASK) >> \
+				KFD_CTXID0_TRAP_CODE_SHIFT)
+#define KFD_CTXID0_DOORBELL_ID(ctxid0)		((ctxid0) & \
+				KFD_CTXID0_DOORBELL_ID_MASK)
+
+static void print_sq_intr_info_auto(struct kfd_node *dev, uint32_t context_id0,
+				    uint32_t context_id1)
+{
+	dev_dbg_ratelimited(
+		dev->adev->dev,
+		"sq_intr: auto, ttrace %d, wlt %d, ttrace_buf_full %d, reg_tms %d, cmd_tms %d, host_cmd_ovf %d, host_reg_ovf %d, immed_ovf %d, ttrace_utc_err %d\n",
+		REG_GET_FIELD(context_id0, SQ_INTERRUPT_WORD_AUTO_CTXID0,
+			      THREAD_TRACE),
+		REG_GET_FIELD(context_id0, SQ_INTERRUPT_WORD_AUTO_CTXID0, WLT),
+		REG_GET_FIELD(context_id0, SQ_INTERRUPT_WORD_AUTO_CTXID0,
+			      THREAD_TRACE_BUF_FULL),
+		REG_GET_FIELD(context_id0, SQ_INTERRUPT_WORD_AUTO_CTXID0,
+			      REG_TIMESTAMP),
+		REG_GET_FIELD(context_id0, SQ_INTERRUPT_WORD_AUTO_CTXID0,
+			      CMD_TIMESTAMP),
+		REG_GET_FIELD(context_id0, SQ_INTERRUPT_WORD_AUTO_CTXID0,
+			      HOST_CMD_OVERFLOW),
+		REG_GET_FIELD(context_id0, SQ_INTERRUPT_WORD_AUTO_CTXID0,
+			      HOST_REG_OVERFLOW),
+		REG_GET_FIELD(context_id0, SQ_INTERRUPT_WORD_AUTO_CTXID0,
+			      IMMED_OVERFLOW),
+		REG_GET_FIELD(context_id0, SQ_INTERRUPT_WORD_AUTO_CTXID0,
+			      THREAD_TRACE_UTC_ERROR));
+}
+
+static void print_sq_intr_info_inst(struct kfd_node *dev, uint32_t context_id0,
+				    uint32_t context_id1)
+{
+	dev_dbg_ratelimited(
+		dev->adev->dev,
+		"sq_intr: inst, data 0x%08x, sh %d, priv %d, wave_id %d, simd_id %d, wgp_id %d\n",
+		REG_GET_FIELD(context_id0, SQ_INTERRUPT_WORD_WAVE_CTXID0, DATA),
+		REG_GET_FIELD(context_id0, SQ_INTERRUPT_WORD_WAVE_CTXID0,
+			      SH_ID),
+		REG_GET_FIELD(context_id0, SQ_INTERRUPT_WORD_WAVE_CTXID0, PRIV),
+		REG_GET_FIELD(context_id0, SQ_INTERRUPT_WORD_WAVE_CTXID0,
+			      WAVE_ID),
+		REG_GET_FIELD(context_id1, SQ_INTERRUPT_WORD_WAVE_CTXID1,
+			      SIMD_ID),
+		REG_GET_FIELD(context_id1, SQ_INTERRUPT_WORD_WAVE_CTXID1,
+			      WGP_ID));
+}
+
+static void print_sq_intr_info_error(struct kfd_node *dev, uint32_t context_id0,
+				     uint32_t context_id1)
+{
+	dev_warn_ratelimited(
+		dev->adev->dev,
+		"sq_intr: error, detail 0x%08x, type %d, sh %d, priv %d, wave_id %d, simd_id %d, wgp_id %d\n",
+		REG_GET_FIELD(context_id0, SQ_INTERRUPT_WORD_ERROR_CTXID0,
+			      DETAIL),
+		REG_GET_FIELD(context_id0, SQ_INTERRUPT_WORD_ERROR_CTXID0,
+			      TYPE),
+		REG_GET_FIELD(context_id0, SQ_INTERRUPT_WORD_ERROR_CTXID0,
+			      SH_ID),
+		REG_GET_FIELD(context_id0, SQ_INTERRUPT_WORD_ERROR_CTXID0,
+			      PRIV),
+		REG_GET_FIELD(context_id0, SQ_INTERRUPT_WORD_ERROR_CTXID0,
+			      WAVE_ID),
+		REG_GET_FIELD(context_id0, SQ_INTERRUPT_WORD_ERROR_CTXID1,
+			      SIMD_ID),
+		REG_GET_FIELD(context_id0, SQ_INTERRUPT_WORD_ERROR_CTXID1,
+			      WGP_ID));
+}
+
+static void event_interrupt_poison_consumption_v11(struct kfd_node *dev,
+				uint16_t pasid, uint16_t source_id)
+{
+	enum amdgpu_ras_block block = 0;
+	int ret = -EINVAL;
+	uint32_t reset = 0;
+	struct kfd_process *p = kfd_lookup_process_by_pasid(pasid, NULL);
+
+	if (!p)
+		return;
+
+	/* all queues of a process will be unmapped in one time */
+	if (atomic_read(&p->poison)) {
+		kfd_unref_process(p);
+		return;
+	}
+
+	atomic_set(&p->poison, 1);
+	kfd_unref_process(p);
+
+	switch (source_id) {
+	case SOC15_INTSRC_SQ_INTERRUPT_MSG:
+		if (dev->dqm->ops.reset_queues)
+			ret = dev->dqm->ops.reset_queues(dev->dqm, pasid);
+		block = AMDGPU_RAS_BLOCK__GFX;
+		if (ret)
+			reset = AMDGPU_RAS_GPU_RESET_MODE2_RESET;
+		break;
+	case SOC21_INTSRC_SDMA_ECC:
+	default:
+		block = AMDGPU_RAS_BLOCK__GFX;
+		reset = AMDGPU_RAS_GPU_RESET_MODE2_RESET;
+		break;
+	}
+
+	kfd_signal_poison_consumed_event(dev, pasid);
+
+	/* resetting queue passes, do page retirement without gpu reset
+	   resetting queue fails, fallback to gpu reset solution */
+	amdgpu_amdkfd_ras_poison_consumption_handler(dev->adev, block, reset);
+}
+
+static bool event_interrupt_isr_v11(struct kfd_node *dev,
+					const uint32_t *ih_ring_entry,
+					uint32_t *patched_ihre,
+					bool *patched_flag)
+{
+	uint16_t source_id, client_id, pasid, vmid;
+	const uint32_t *data = ih_ring_entry;
+	uint32_t context_id0;
+
+	source_id = SOC15_SOURCE_ID_FROM_IH_ENTRY(ih_ring_entry);
+	client_id = SOC15_CLIENT_ID_FROM_IH_ENTRY(ih_ring_entry);
+	/* Only handle interrupts from KFD VMIDs */
+	vmid = SOC15_VMID_FROM_IH_ENTRY(ih_ring_entry);
+	if (!KFD_IRQ_IS_FENCE(client_id, source_id) &&
+	    (vmid < dev->vm_info.first_vmid_kfd ||
+	    vmid > dev->vm_info.last_vmid_kfd))
+		return false;
+
+	pasid = SOC15_PASID_FROM_IH_ENTRY(ih_ring_entry);
+	context_id0 = SOC15_CONTEXT_ID0_FROM_IH_ENTRY(ih_ring_entry);
+
+	if ((source_id == SOC15_INTSRC_CP_END_OF_PIPE) &&
+	    (context_id0 & AMDGPU_FENCE_MES_QUEUE_FLAG))
+		return false;
+
+	dev_dbg(dev->adev->dev,
+		"client id 0x%x, source id %d, vmid %d, pasid 0x%x. raw data:\n",
+		client_id, source_id, vmid, pasid);
+	dev_dbg(dev->adev->dev, "%8X, %8X, %8X, %8X, %8X, %8X, %8X, %8X.\n",
+		data[0], data[1], data[2], data[3], data[4], data[5], data[6],
+		data[7]);
+
+	if (pasid == 0)
+		return false;
+
+	/* Interrupt types we care about: various signals and faults.
+	 * They will be forwarded to a work queue (see below).
+	 */
+	return source_id == SOC15_INTSRC_CP_END_OF_PIPE ||
+		source_id == SOC15_INTSRC_SQ_INTERRUPT_MSG ||
+		source_id == SOC15_INTSRC_CP_BAD_OPCODE ||
+		source_id == SOC21_INTSRC_SDMA_TRAP ||
+		KFD_IRQ_IS_FENCE(client_id, source_id) ||
+		(((client_id == SOC21_IH_CLIENTID_VMC) ||
+		 ((client_id == SOC21_IH_CLIENTID_GFX) &&
+		  (source_id == UTCL2_1_0__SRCID__FAULT))) &&
+		  !amdgpu_no_queue_eviction_on_vm_fault);
+}
+
+static void event_interrupt_wq_v11(struct kfd_node *dev,
+					const uint32_t *ih_ring_entry)
+{
+	uint16_t source_id, client_id, ring_id, pasid, vmid;
+	uint32_t context_id0, context_id1;
+	uint8_t sq_int_enc, sq_int_priv, sq_int_errtype;
+	struct kfd_vm_fault_info info = {0};
+	struct kfd_hsa_memory_exception_data exception_data;
+
+	source_id = SOC15_SOURCE_ID_FROM_IH_ENTRY(ih_ring_entry);
+	client_id = SOC15_CLIENT_ID_FROM_IH_ENTRY(ih_ring_entry);
+	ring_id = SOC15_RING_ID_FROM_IH_ENTRY(ih_ring_entry);
+	pasid = SOC15_PASID_FROM_IH_ENTRY(ih_ring_entry);
+	vmid = SOC15_VMID_FROM_IH_ENTRY(ih_ring_entry);
+	context_id0 = SOC15_CONTEXT_ID0_FROM_IH_ENTRY(ih_ring_entry);
+	context_id1 = SOC15_CONTEXT_ID1_FROM_IH_ENTRY(ih_ring_entry);
+
+	/* VMC, UTCL2 */
+	if (client_id == SOC21_IH_CLIENTID_VMC ||
+	     ((client_id == SOC21_IH_CLIENTID_GFX) &&
+	     (source_id == UTCL2_1_0__SRCID__FAULT))) {
+
+		info.vmid = vmid;
+		info.mc_id = client_id;
+		info.page_addr = ih_ring_entry[4] |
+			(uint64_t)(ih_ring_entry[5] & 0xf) << 32;
+		info.prot_valid = ring_id & 0x08;
+		info.prot_read  = ring_id & 0x10;
+		info.prot_write = ring_id & 0x20;
+
+		memset(&exception_data, 0, sizeof(exception_data));
+		exception_data.gpu_id = dev->id;
+		exception_data.va = (info.page_addr) << PAGE_SHIFT;
+		exception_data.failure.NotPresent = info.prot_valid ? 1 : 0;
+		exception_data.failure.NoExecute = info.prot_exec ? 1 : 0;
+		exception_data.failure.ReadOnly = info.prot_write ? 1 : 0;
+		exception_data.failure.imprecise = 0;
+
+		kfd_set_dbg_ev_from_interrupt(dev, pasid, -1,
+					      KFD_EC_MASK(EC_DEVICE_MEMORY_VIOLATION),
+					      &exception_data, sizeof(exception_data));
+		kfd_smi_event_update_vmfault(dev, pasid);
+
+	/* GRBM, SDMA, SE, PMM */
+	} else if (client_id == SOC21_IH_CLIENTID_GRBM_CP ||
+		   client_id == SOC21_IH_CLIENTID_GFX) {
+
+		/* CP */
+		if (source_id == SOC15_INTSRC_CP_END_OF_PIPE)
+			kfd_signal_event_interrupt(pasid, context_id0, 32);
+		else if (source_id == SOC15_INTSRC_CP_BAD_OPCODE &&
+			 KFD_DBG_EC_TYPE_IS_PACKET(KFD_CTXID0_CP_BAD_OP_ECODE(context_id0))) {
+			u32 doorbell_id = KFD_CTXID0_DOORBELL_ID(context_id0);
+
+			kfd_set_dbg_ev_from_interrupt(dev, pasid, doorbell_id,
+				KFD_EC_MASK(KFD_CTXID0_CP_BAD_OP_ECODE(context_id0)),
+				NULL, 0);
+			kfd_dqm_suspend_bad_queue_mes(dev, pasid, doorbell_id);
+		}
+
+		/* SDMA */
+		else if (source_id == SOC21_INTSRC_SDMA_TRAP)
+			kfd_signal_event_interrupt(pasid, context_id0 & 0xfffffff, 28);
+		else if (source_id == SOC21_INTSRC_SDMA_ECC) {
+			event_interrupt_poison_consumption_v11(dev, pasid, source_id);
+			return;
+		}
+
+		/* SQ */
+		else if (source_id == SOC15_INTSRC_SQ_INTERRUPT_MSG) {
+			sq_int_enc = REG_GET_FIELD(context_id1,
+					SQ_INTERRUPT_WORD_WAVE_CTXID1, ENCODING);
+			switch (sq_int_enc) {
+			case SQ_INTERRUPT_WORD_ENCODING_AUTO:
+				print_sq_intr_info_auto(dev, context_id0, context_id1);
+				break;
+			case SQ_INTERRUPT_WORD_ENCODING_INST:
+				print_sq_intr_info_inst(dev, context_id0, context_id1);
+				sq_int_priv = REG_GET_FIELD(context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID0, PRIV);
+				if (sq_int_priv && (kfd_set_dbg_ev_from_interrupt(dev, pasid,
+						KFD_CTXID0_DOORBELL_ID(context_id0),
+						KFD_CTXID0_TRAP_CODE(context_id0),
+						NULL, 0)))
+					return;
+				break;
+			case SQ_INTERRUPT_WORD_ENCODING_ERROR:
+				print_sq_intr_info_error(dev, context_id0, context_id1);
+				sq_int_errtype = REG_GET_FIELD(context_id0,
+						SQ_INTERRUPT_WORD_ERROR_CTXID0, TYPE);
+				if (sq_int_errtype != SQ_INTERRUPT_ERROR_TYPE_ILLEGAL_INST &&
+				    sq_int_errtype != SQ_INTERRUPT_ERROR_TYPE_MEMVIOL) {
+					event_interrupt_poison_consumption_v11(
+							dev, pasid, source_id);
+					return;
+				}
+				break;
+			default:
+				break;
+			}
+			kfd_signal_event_interrupt(pasid, context_id0 & 0xffffff, 24);
+		}
+
+	} else if (KFD_IRQ_IS_FENCE(client_id, source_id)) {
+		kfd_process_close_interrupt_drain(pasid);
+	}
+}
+
+const struct kfd_event_interrupt_class event_interrupt_class_v11 = {
+	.interrupt_isr = event_interrupt_isr_v11,
+	.interrupt_wq = event_interrupt_wq_v11,
+};
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v9.c
new file mode 100644
index 000000000000..4ceb251312a6
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v9.c
@@ -0,0 +1,605 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
+/*
+ * Copyright 2016-2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include "kfd_priv.h"
+#include "kfd_events.h"
+#include "kfd_debug.h"
+#include "soc15_int.h"
+#include "kfd_device_queue_manager.h"
+#include "kfd_smi_events.h"
+#include "amdgpu_ras.h"
+
+/*
+ * GFX9 SQ Interrupts
+ *
+ * There are 3 encoding types of interrupts sourced from SQ sent as a 44-bit
+ * packet to the Interrupt Handler:
+ * Auto - Generated by the SQG (various cmd overflows, timestamps etc)
+ * Wave - Generated by S_SENDMSG through a shader program
+ * Error - HW generated errors (Illegal instructions, Memviols, EDC etc)
+ *
+ * The 44-bit packet is mapped as {context_id1[7:0],context_id0[31:0]} plus
+ * 4-bits for VMID (SOC15_VMID_FROM_IH_ENTRY) as such:
+ *
+ * - context_id0[27:26]
+ * Encoding type (0 = Auto, 1 = Wave, 2 = Error)
+ *
+ * - context_id0[13]
+ * PRIV bit indicates that Wave S_SEND or error occurred within trap
+ *
+ * - {context_id1[7:0],context_id0[31:28],context_id0[11:0]}
+ * 24-bit data with the following layout per encoding type:
+ * Auto - only context_id0[8:0] is used, which reports various interrupts
+ * generated by SQG.  The rest is 0.
+ * Wave - user data sent from m0 via S_SENDMSG
+ * Error - Error type (context_id1[7:4]), Error Details (rest of bits)
+ *
+ * The other context_id bits show coordinates (SE/SH/CU/SIMD/WAVE) for wave
+ * S_SENDMSG and Errors.  These are 0 for Auto.
+ */
+
+enum SQ_INTERRUPT_WORD_ENCODING {
+	SQ_INTERRUPT_WORD_ENCODING_AUTO = 0x0,
+	SQ_INTERRUPT_WORD_ENCODING_INST,
+	SQ_INTERRUPT_WORD_ENCODING_ERROR,
+};
+
+enum SQ_INTERRUPT_ERROR_TYPE {
+	SQ_INTERRUPT_ERROR_TYPE_EDC_FUE = 0x0,
+	SQ_INTERRUPT_ERROR_TYPE_ILLEGAL_INST,
+	SQ_INTERRUPT_ERROR_TYPE_MEMVIOL,
+	SQ_INTERRUPT_ERROR_TYPE_EDC_FED,
+};
+
+/* SQ_INTERRUPT_WORD_AUTO_CTXID */
+#define SQ_INTERRUPT_WORD_AUTO_CTXID__THREAD_TRACE__SHIFT 0
+#define SQ_INTERRUPT_WORD_AUTO_CTXID__WLT__SHIFT 1
+#define SQ_INTERRUPT_WORD_AUTO_CTXID__THREAD_TRACE_BUF_FULL__SHIFT 2
+#define SQ_INTERRUPT_WORD_AUTO_CTXID__REG_TIMESTAMP__SHIFT 3
+#define SQ_INTERRUPT_WORD_AUTO_CTXID__CMD_TIMESTAMP__SHIFT 4
+#define SQ_INTERRUPT_WORD_AUTO_CTXID__HOST_CMD_OVERFLOW__SHIFT 5
+#define SQ_INTERRUPT_WORD_AUTO_CTXID__HOST_REG_OVERFLOW__SHIFT 6
+#define SQ_INTERRUPT_WORD_AUTO_CTXID__IMMED_OVERFLOW__SHIFT 7
+#define SQ_INTERRUPT_WORD_AUTO_CTXID__THREAD_TRACE_UTC_ERROR__SHIFT 8
+#define SQ_INTERRUPT_WORD_AUTO_CTXID__SE_ID__SHIFT 24
+#define SQ_INTERRUPT_WORD_AUTO_CTXID__ENCODING__SHIFT 26
+
+#define SQ_INTERRUPT_WORD_AUTO_CTXID__THREAD_TRACE_MASK 0x00000001
+#define SQ_INTERRUPT_WORD_AUTO_CTXID__WLT_MASK 0x00000002
+#define SQ_INTERRUPT_WORD_AUTO_CTXID__THREAD_TRACE_BUF_FULL_MASK 0x00000004
+#define SQ_INTERRUPT_WORD_AUTO_CTXID__REG_TIMESTAMP_MASK 0x00000008
+#define SQ_INTERRUPT_WORD_AUTO_CTXID__CMD_TIMESTAMP_MASK 0x00000010
+#define SQ_INTERRUPT_WORD_AUTO_CTXID__HOST_CMD_OVERFLOW_MASK 0x00000020
+#define SQ_INTERRUPT_WORD_AUTO_CTXID__HOST_REG_OVERFLOW_MASK 0x00000040
+#define SQ_INTERRUPT_WORD_AUTO_CTXID__IMMED_OVERFLOW_MASK 0x00000080
+#define SQ_INTERRUPT_WORD_AUTO_CTXID__THREAD_TRACE_UTC_ERROR_MASK 0x00000100
+#define SQ_INTERRUPT_WORD_AUTO_CTXID__SE_ID_MASK 0x03000000
+#define SQ_INTERRUPT_WORD_AUTO_CTXID__ENCODING_MASK 0x0c000000
+
+/* SQ_INTERRUPT_WORD_WAVE_CTXID */
+#define SQ_INTERRUPT_WORD_WAVE_CTXID__DATA__SHIFT 0
+#define SQ_INTERRUPT_WORD_WAVE_CTXID__SH_ID__SHIFT 12
+#define SQ_INTERRUPT_WORD_WAVE_CTXID__PRIV__SHIFT 13
+#define SQ_INTERRUPT_WORD_WAVE_CTXID__WAVE_ID__SHIFT 14
+#define SQ_INTERRUPT_WORD_WAVE_CTXID__SIMD_ID__SHIFT 18
+#define SQ_INTERRUPT_WORD_WAVE_CTXID__CU_ID__SHIFT 20
+#define SQ_INTERRUPT_WORD_WAVE_CTXID__SE_ID__SHIFT 24
+#define SQ_INTERRUPT_WORD_WAVE_CTXID__ENCODING__SHIFT 26
+
+#define SQ_INTERRUPT_WORD_WAVE_CTXID__DATA_MASK 0x00000fff
+#define SQ_INTERRUPT_WORD_WAVE_CTXID__SH_ID_MASK 0x00001000
+#define SQ_INTERRUPT_WORD_WAVE_CTXID__PRIV_MASK 0x00002000
+#define SQ_INTERRUPT_WORD_WAVE_CTXID__WAVE_ID_MASK 0x0003c000
+#define SQ_INTERRUPT_WORD_WAVE_CTXID__SIMD_ID_MASK 0x000c0000
+#define SQ_INTERRUPT_WORD_WAVE_CTXID__CU_ID_MASK 0x00f00000
+#define SQ_INTERRUPT_WORD_WAVE_CTXID__SE_ID_MASK 0x03000000
+#define SQ_INTERRUPT_WORD_WAVE_CTXID__ENCODING_MASK 0x0c000000
+
+/* GFX9 SQ interrupt 24-bit data from context_id<0,1> */
+#define KFD_CONTEXT_ID_GET_SQ_INT_DATA(ctx0, ctx1)                             \
+	((ctx0 & 0xfff) | ((ctx0 >> 16) & 0xf000) | ((ctx1 << 16) & 0xff0000))
+
+#define KFD_SQ_INT_DATA__ERR_TYPE_MASK 0xF00000
+#define KFD_SQ_INT_DATA__ERR_TYPE__SHIFT 20
+
+/*
+ * The debugger will send user data(m0) with PRIV=1 to indicate it requires
+ * notification from the KFD with the following queue id (DOORBELL_ID) and
+ * trap code (TRAP_CODE).
+ */
+#define KFD_INT_DATA_DEBUG_DOORBELL_MASK	0x0003ff
+#define KFD_INT_DATA_DEBUG_TRAP_CODE_SHIFT	10
+#define KFD_INT_DATA_DEBUG_TRAP_CODE_MASK	0x07fc00
+#define KFD_DEBUG_DOORBELL_ID(sq_int_data)	((sq_int_data) &	\
+				KFD_INT_DATA_DEBUG_DOORBELL_MASK)
+#define KFD_DEBUG_TRAP_CODE(sq_int_data)	(((sq_int_data) &	\
+				KFD_INT_DATA_DEBUG_TRAP_CODE_MASK)	\
+				>> KFD_INT_DATA_DEBUG_TRAP_CODE_SHIFT)
+#define KFD_DEBUG_CP_BAD_OP_ECODE_MASK		0x3fffc00
+#define KFD_DEBUG_CP_BAD_OP_ECODE_SHIFT		10
+#define KFD_DEBUG_CP_BAD_OP_ECODE(ctxid0)	(((ctxid0) &		\
+				KFD_DEBUG_CP_BAD_OP_ECODE_MASK)		\
+				>> KFD_DEBUG_CP_BAD_OP_ECODE_SHIFT)
+
+static void event_interrupt_poison_consumption_v9(struct kfd_node *dev,
+				uint16_t pasid, uint16_t client_id)
+{
+	enum amdgpu_ras_block block = 0;
+	uint32_t reset = 0;
+	struct kfd_process *p = kfd_lookup_process_by_pasid(pasid, NULL);
+	enum ras_event_type type = RAS_EVENT_TYPE_POISON_CONSUMPTION;
+	u64 event_id;
+	int old_poison, ret;
+
+	if (!p)
+		return;
+
+	/* all queues of a process will be unmapped in one time */
+	old_poison = atomic_cmpxchg(&p->poison, 0, 1);
+	kfd_unref_process(p);
+	if (old_poison)
+		return;
+
+	switch (client_id) {
+	case SOC15_IH_CLIENTID_SE0SH:
+	case SOC15_IH_CLIENTID_SE1SH:
+	case SOC15_IH_CLIENTID_SE2SH:
+	case SOC15_IH_CLIENTID_SE3SH:
+	case SOC15_IH_CLIENTID_UTCL2:
+		block = AMDGPU_RAS_BLOCK__GFX;
+		if (amdgpu_ip_version(dev->adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3)) {
+			/* driver mode-2 for gfx poison is only supported by
+			 * pmfw 0x00557300 and onwards */
+			if (dev->adev->pm.fw_version < 0x00557300)
+				reset = AMDGPU_RAS_GPU_RESET_MODE1_RESET;
+			else
+				reset = AMDGPU_RAS_GPU_RESET_MODE2_RESET;
+		} else if (amdgpu_ip_version(dev->adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4)) {
+			/* driver mode-2 for gfx poison is only supported by
+			 * pmfw 0x05550C00 and onwards */
+			if (dev->adev->pm.fw_version < 0x05550C00)
+				reset = AMDGPU_RAS_GPU_RESET_MODE1_RESET;
+			else
+				reset = AMDGPU_RAS_GPU_RESET_MODE2_RESET;
+		} else {
+			reset = AMDGPU_RAS_GPU_RESET_MODE2_RESET;
+		}
+		amdgpu_ras_set_err_poison(dev->adev, AMDGPU_RAS_BLOCK__GFX);
+		break;
+	case SOC15_IH_CLIENTID_VMC:
+	case SOC15_IH_CLIENTID_VMC1:
+		block = AMDGPU_RAS_BLOCK__MMHUB;
+		reset = AMDGPU_RAS_GPU_RESET_MODE1_RESET;
+		break;
+	case SOC15_IH_CLIENTID_SDMA0:
+	case SOC15_IH_CLIENTID_SDMA1:
+	case SOC15_IH_CLIENTID_SDMA2:
+	case SOC15_IH_CLIENTID_SDMA3:
+	case SOC15_IH_CLIENTID_SDMA4:
+		block = AMDGPU_RAS_BLOCK__SDMA;
+		if (amdgpu_ip_version(dev->adev, SDMA0_HWIP, 0) == IP_VERSION(4, 4, 2)) {
+			/* driver mode-2 for gfx poison is only supported by
+			 * pmfw 0x00557300 and onwards */
+			if (dev->adev->pm.fw_version < 0x00557300)
+				reset = AMDGPU_RAS_GPU_RESET_MODE1_RESET;
+			else
+				reset = AMDGPU_RAS_GPU_RESET_MODE2_RESET;
+		} else if (amdgpu_ip_version(dev->adev, SDMA0_HWIP, 0) == IP_VERSION(4, 4, 5)) {
+			/* driver mode-2 for gfx poison is only supported by
+			 * pmfw 0x05550C00 and onwards */
+			if (dev->adev->pm.fw_version < 0x05550C00)
+				reset = AMDGPU_RAS_GPU_RESET_MODE1_RESET;
+			else
+				reset = AMDGPU_RAS_GPU_RESET_MODE2_RESET;
+		} else {
+			reset = AMDGPU_RAS_GPU_RESET_MODE2_RESET;
+		}
+		amdgpu_ras_set_err_poison(dev->adev, AMDGPU_RAS_BLOCK__SDMA);
+		break;
+	default:
+		dev_warn(dev->adev->dev,
+			 "client %d does not support poison consumption\n", client_id);
+		return;
+	}
+
+	ret = amdgpu_ras_mark_ras_event(dev->adev, type);
+	if (ret)
+		return;
+
+	kfd_signal_poison_consumed_event(dev, pasid);
+
+	event_id = amdgpu_ras_acquire_event_id(dev->adev, type);
+
+	RAS_EVENT_LOG(dev->adev, event_id,
+		      "poison is consumed by client %d, kick off gpu reset flow\n", client_id);
+
+	amdgpu_amdkfd_ras_pasid_poison_consumption_handler(dev->adev,
+		block, pasid, NULL, NULL, reset);
+}
+
+static bool context_id_expected(struct kfd_dev *dev)
+{
+	switch (KFD_GC_VERSION(dev)) {
+	case IP_VERSION(9, 0, 1):
+		return dev->mec_fw_version >= 0x817a;
+	case IP_VERSION(9, 1, 0):
+	case IP_VERSION(9, 2, 1):
+	case IP_VERSION(9, 2, 2):
+	case IP_VERSION(9, 3, 0):
+	case IP_VERSION(9, 4, 0):
+		return dev->mec_fw_version >= 0x17a;
+	default:
+		/* Other GFXv9 and later GPUs always sent valid context IDs
+		 * on legitimate events
+		 */
+		return KFD_GC_VERSION(dev) >= IP_VERSION(9, 4, 1);
+	}
+}
+
+static bool event_interrupt_isr_v9(struct kfd_node *dev,
+					const uint32_t *ih_ring_entry,
+					uint32_t *patched_ihre,
+					bool *patched_flag)
+{
+	uint16_t source_id, client_id, pasid, vmid;
+	const uint32_t *data = ih_ring_entry;
+
+	source_id = SOC15_SOURCE_ID_FROM_IH_ENTRY(ih_ring_entry);
+	client_id = SOC15_CLIENT_ID_FROM_IH_ENTRY(ih_ring_entry);
+
+	/* Only handle interrupts from KFD VMIDs */
+	vmid = SOC15_VMID_FROM_IH_ENTRY(ih_ring_entry);
+	if (!KFD_IRQ_IS_FENCE(client_id, source_id) &&
+	   (vmid < dev->vm_info.first_vmid_kfd ||
+	    vmid > dev->vm_info.last_vmid_kfd))
+		return false;
+
+	pasid = SOC15_PASID_FROM_IH_ENTRY(ih_ring_entry);
+
+	/* Only handle clients we care about */
+	if (client_id != SOC15_IH_CLIENTID_GRBM_CP &&
+	    client_id != SOC15_IH_CLIENTID_SDMA0 &&
+	    client_id != SOC15_IH_CLIENTID_SDMA1 &&
+	    client_id != SOC15_IH_CLIENTID_SDMA2 &&
+	    client_id != SOC15_IH_CLIENTID_SDMA3 &&
+	    client_id != SOC15_IH_CLIENTID_SDMA4 &&
+	    client_id != SOC15_IH_CLIENTID_SDMA5 &&
+	    client_id != SOC15_IH_CLIENTID_SDMA6 &&
+	    client_id != SOC15_IH_CLIENTID_SDMA7 &&
+	    client_id != SOC15_IH_CLIENTID_VMC &&
+	    client_id != SOC15_IH_CLIENTID_VMC1 &&
+	    client_id != SOC15_IH_CLIENTID_UTCL2 &&
+	    client_id != SOC15_IH_CLIENTID_SE0SH &&
+	    client_id != SOC15_IH_CLIENTID_SE1SH &&
+	    client_id != SOC15_IH_CLIENTID_SE2SH &&
+	    client_id != SOC15_IH_CLIENTID_SE3SH &&
+	    !KFD_IRQ_IS_FENCE(client_id, source_id))
+		return false;
+
+	/* This is a known issue for gfx9. Under non HWS, pasid is not set
+	 * in the interrupt payload, so we need to find out the pasid on our
+	 * own.
+	 */
+	if (!pasid && dev->dqm->sched_policy == KFD_SCHED_POLICY_NO_HWS) {
+		const uint32_t pasid_mask = 0xffff;
+
+		*patched_flag = true;
+		memcpy(patched_ihre, ih_ring_entry,
+				dev->kfd->device_info.ih_ring_entry_size);
+
+		pasid = dev->dqm->vmid_pasid[vmid];
+
+		/* Patch the pasid field */
+		patched_ihre[3] = cpu_to_le32((le32_to_cpu(patched_ihre[3])
+					& ~pasid_mask) | pasid);
+	}
+
+	dev_dbg(dev->adev->dev,
+		"client id 0x%x, source id %d, vmid %d, pasid 0x%x. raw data:\n",
+		client_id, source_id, vmid, pasid);
+	dev_dbg(dev->adev->dev, "%8X, %8X, %8X, %8X, %8X, %8X, %8X, %8X.\n",
+		data[0], data[1], data[2], data[3], data[4], data[5], data[6],
+		data[7]);
+
+	/* If there is no valid PASID, it's likely a bug */
+	if (WARN_ONCE(pasid == 0, "Bug: No PASID in KFD interrupt"))
+		return false;
+
+	/* Workaround CP firmware sending bogus signals with 0 context_id.
+	 * Those can be safely ignored on hardware and firmware versions that
+	 * include a valid context_id on legitimate signals. This avoids the
+	 * slow path in kfd_signal_event_interrupt that scans all event slots
+	 * for signaled events.
+	 */
+	if (source_id == SOC15_INTSRC_CP_END_OF_PIPE) {
+		uint32_t context_id =
+			SOC15_CONTEXT_ID0_FROM_IH_ENTRY(ih_ring_entry);
+
+		if (context_id == 0 && context_id_expected(dev->kfd))
+			return false;
+	}
+
+	/* Interrupt types we care about: various signals and faults.
+	 * They will be forwarded to a work queue (see below).
+	 */
+	return source_id == SOC15_INTSRC_CP_END_OF_PIPE ||
+		source_id == SOC15_INTSRC_SDMA_TRAP ||
+		source_id == SOC15_INTSRC_SDMA_ECC ||
+		source_id == SOC15_INTSRC_SQ_INTERRUPT_MSG ||
+		source_id == SOC15_INTSRC_CP_BAD_OPCODE ||
+		KFD_IRQ_IS_FENCE(client_id, source_id) ||
+		((client_id == SOC15_IH_CLIENTID_VMC ||
+		client_id == SOC15_IH_CLIENTID_VMC1 ||
+		client_id == SOC15_IH_CLIENTID_UTCL2) &&
+		!amdgpu_no_queue_eviction_on_vm_fault);
+}
+
+static void event_interrupt_wq_v9(struct kfd_node *dev,
+					const uint32_t *ih_ring_entry)
+{
+	uint16_t source_id, client_id, pasid, vmid;
+	uint32_t context_id0, context_id1;
+	uint32_t sq_intr_err, sq_int_data, encoding;
+
+	source_id = SOC15_SOURCE_ID_FROM_IH_ENTRY(ih_ring_entry);
+	client_id = SOC15_CLIENT_ID_FROM_IH_ENTRY(ih_ring_entry);
+	pasid = SOC15_PASID_FROM_IH_ENTRY(ih_ring_entry);
+	vmid = SOC15_VMID_FROM_IH_ENTRY(ih_ring_entry);
+	context_id0 = SOC15_CONTEXT_ID0_FROM_IH_ENTRY(ih_ring_entry);
+	context_id1 = SOC15_CONTEXT_ID1_FROM_IH_ENTRY(ih_ring_entry);
+
+	if (client_id == SOC15_IH_CLIENTID_GRBM_CP ||
+	    client_id == SOC15_IH_CLIENTID_SE0SH ||
+	    client_id == SOC15_IH_CLIENTID_SE1SH ||
+	    client_id == SOC15_IH_CLIENTID_SE2SH ||
+	    client_id == SOC15_IH_CLIENTID_SE3SH) {
+		if (source_id == SOC15_INTSRC_CP_END_OF_PIPE)
+			kfd_signal_event_interrupt(pasid, context_id0, 32);
+		else if (source_id == SOC15_INTSRC_SQ_INTERRUPT_MSG) {
+			sq_int_data = KFD_CONTEXT_ID_GET_SQ_INT_DATA(context_id0, context_id1);
+			encoding = REG_GET_FIELD(context_id0, SQ_INTERRUPT_WORD_WAVE_CTXID, ENCODING);
+			switch (encoding) {
+			case SQ_INTERRUPT_WORD_ENCODING_AUTO:
+				dev_dbg_ratelimited(
+					dev->adev->dev,
+					"sq_intr: auto, se %d, ttrace %d, wlt %d, ttrac_buf_full %d, reg_tms %d, cmd_tms %d, host_cmd_ovf %d, host_reg_ovf %d, immed_ovf %d, ttrace_utc_err %d\n",
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_AUTO_CTXID,
+						SE_ID),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_AUTO_CTXID,
+						THREAD_TRACE),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_AUTO_CTXID,
+						WLT),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_AUTO_CTXID,
+						THREAD_TRACE_BUF_FULL),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_AUTO_CTXID,
+						REG_TIMESTAMP),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_AUTO_CTXID,
+						CMD_TIMESTAMP),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_AUTO_CTXID,
+						HOST_CMD_OVERFLOW),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_AUTO_CTXID,
+						HOST_REG_OVERFLOW),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_AUTO_CTXID,
+						IMMED_OVERFLOW),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_AUTO_CTXID,
+						THREAD_TRACE_UTC_ERROR));
+				break;
+			case SQ_INTERRUPT_WORD_ENCODING_INST:
+				dev_dbg_ratelimited(
+					dev->adev->dev,
+					"sq_intr: inst, se %d, data 0x%x, sh %d, priv %d, wave_id %d, simd_id %d, cu_id %d, intr_data 0x%x\n",
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID,
+						SE_ID),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID,
+						DATA),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID,
+						SH_ID),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID,
+						PRIV),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID,
+						WAVE_ID),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID,
+						SIMD_ID),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID,
+						CU_ID),
+					sq_int_data);
+				if (context_id0 & SQ_INTERRUPT_WORD_WAVE_CTXID__PRIV_MASK) {
+					if (kfd_set_dbg_ev_from_interrupt(dev, pasid,
+							KFD_DEBUG_DOORBELL_ID(sq_int_data),
+							KFD_DEBUG_TRAP_CODE(sq_int_data),
+							NULL, 0))
+						return;
+				}
+				break;
+			case SQ_INTERRUPT_WORD_ENCODING_ERROR:
+				sq_intr_err = REG_GET_FIELD(sq_int_data, KFD_SQ_INT_DATA, ERR_TYPE);
+				dev_warn_ratelimited(
+					dev->adev->dev,
+					"sq_intr: error, se %d, data 0x%x, sh %d, priv %d, wave_id %d, simd_id %d, cu_id %d, err_type %d\n",
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID,
+						SE_ID),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID,
+						DATA),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID,
+						SH_ID),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID,
+						PRIV),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID,
+						WAVE_ID),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID,
+						SIMD_ID),
+					REG_GET_FIELD(
+						context_id0,
+						SQ_INTERRUPT_WORD_WAVE_CTXID,
+						CU_ID),
+					sq_intr_err);
+				if (sq_intr_err != SQ_INTERRUPT_ERROR_TYPE_ILLEGAL_INST &&
+					sq_intr_err != SQ_INTERRUPT_ERROR_TYPE_MEMVIOL) {
+					event_interrupt_poison_consumption_v9(dev, pasid, client_id);
+					return;
+				}
+				break;
+			default:
+				break;
+			}
+			kfd_signal_event_interrupt(pasid, sq_int_data, 24);
+		} else if (source_id == SOC15_INTSRC_CP_BAD_OPCODE &&
+			   KFD_DBG_EC_TYPE_IS_PACKET(KFD_DEBUG_CP_BAD_OP_ECODE(context_id0))) {
+			kfd_set_dbg_ev_from_interrupt(dev, pasid,
+				KFD_DEBUG_DOORBELL_ID(context_id0),
+				KFD_EC_MASK(KFD_DEBUG_CP_BAD_OP_ECODE(context_id0)),
+				NULL, 0);
+		}
+	} else if (client_id == SOC15_IH_CLIENTID_SDMA0 ||
+		   client_id == SOC15_IH_CLIENTID_SDMA1 ||
+		   client_id == SOC15_IH_CLIENTID_SDMA2 ||
+		   client_id == SOC15_IH_CLIENTID_SDMA3 ||
+		   client_id == SOC15_IH_CLIENTID_SDMA4 ||
+		   client_id == SOC15_IH_CLIENTID_SDMA5 ||
+		   client_id == SOC15_IH_CLIENTID_SDMA6 ||
+		   client_id == SOC15_IH_CLIENTID_SDMA7) {
+		if (source_id == SOC15_INTSRC_SDMA_TRAP) {
+			kfd_signal_event_interrupt(pasid, context_id0 & 0xfffffff, 28);
+		} else if (source_id == SOC15_INTSRC_SDMA_ECC) {
+			event_interrupt_poison_consumption_v9(dev, pasid, client_id);
+			return;
+		}
+	} else if (client_id == SOC15_IH_CLIENTID_VMC ||
+		   client_id == SOC15_IH_CLIENTID_VMC1 ||
+		   client_id == SOC15_IH_CLIENTID_UTCL2) {
+		struct kfd_vm_fault_info info = {0};
+		uint16_t ring_id = SOC15_RING_ID_FROM_IH_ENTRY(ih_ring_entry);
+		struct kfd_hsa_memory_exception_data exception_data;
+
+		if (source_id == SOC15_INTSRC_VMC_UTCL2_POISON) {
+			event_interrupt_poison_consumption_v9(dev, pasid, client_id);
+			return;
+		}
+
+		info.vmid = vmid;
+		info.mc_id = client_id;
+		info.page_addr = ih_ring_entry[4] |
+			(uint64_t)(ih_ring_entry[5] & 0xf) << 32;
+		info.prot_valid = ring_id & 0x08;
+		info.prot_read  = ring_id & 0x10;
+		info.prot_write = ring_id & 0x20;
+
+		memset(&exception_data, 0, sizeof(exception_data));
+		exception_data.gpu_id = dev->id;
+		exception_data.va = (info.page_addr) << PAGE_SHIFT;
+		exception_data.failure.NotPresent = info.prot_valid ? 1 : 0;
+		exception_data.failure.NoExecute = info.prot_exec ? 1 : 0;
+		exception_data.failure.ReadOnly = info.prot_write ? 1 : 0;
+		exception_data.failure.imprecise = 0;
+
+		kfd_set_dbg_ev_from_interrupt(dev,
+						pasid,
+						-1,
+						KFD_EC_MASK(EC_DEVICE_MEMORY_VIOLATION),
+						&exception_data,
+						sizeof(exception_data));
+		kfd_smi_event_update_vmfault(dev, pasid);
+	} else if (KFD_IRQ_IS_FENCE(client_id, source_id)) {
+		kfd_process_close_interrupt_drain(pasid);
+	}
+}
+
+static bool event_interrupt_isr_v9_4_3(struct kfd_node *node,
+				const uint32_t *ih_ring_entry,
+				uint32_t *patched_ihre,
+				bool *patched_flag)
+{
+	uint16_t node_id, vmid;
+
+	/*
+	 * For GFX 9.4.3, process the interrupt if:
+	 * - NodeID field in IH entry matches the corresponding bit
+	 *   set in interrupt_bitmap Bits 0-15.
+	 *   OR
+	 * - If partition mode is CPX and interrupt came from
+	 *   Node_id 0,4,8,12, then check if the Bit (16 + client id)
+	 *   is set in interrupt bitmap Bits 16-31.
+	 */
+	node_id = SOC15_NODEID_FROM_IH_ENTRY(ih_ring_entry);
+	vmid = SOC15_VMID_FROM_IH_ENTRY(ih_ring_entry);
+	if (kfd_irq_is_from_node(node, node_id, vmid))
+		return event_interrupt_isr_v9(node, ih_ring_entry,
+					patched_ihre, patched_flag);
+	return false;
+}
+
+const struct kfd_event_interrupt_class event_interrupt_class_v9 = {
+	.interrupt_isr = event_interrupt_isr_v9,
+	.interrupt_wq = event_interrupt_wq_v9,
+};
+
+const struct kfd_event_interrupt_class event_interrupt_class_v9_4_3 = {
+	.interrupt_isr = event_interrupt_isr_v9_4_3,
+	.interrupt_wq = event_interrupt_wq_v9,
+};
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_interrupt.c b/drivers/gpu/drm/amd/amdkfd/kfd_interrupt.c
index 70b3a99cffc2..783c2f5a04e4 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_interrupt.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_interrupt.c
@@ -1,5 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -42,31 +43,39 @@
 
 #include <linux/slab.h>
 #include <linux/device.h>
+#include <linux/kfifo.h>
 #include "kfd_priv.h"
 
-#define KFD_INTERRUPT_RING_SIZE 1024
+#define KFD_IH_NUM_ENTRIES 16384
 
 static void interrupt_wq(struct work_struct *);
 
-int kfd_interrupt_init(struct kfd_dev *kfd)
+int kfd_interrupt_init(struct kfd_node *node)
 {
-	void *interrupt_ring = kmalloc_array(KFD_INTERRUPT_RING_SIZE,
-					kfd->device_info->ih_ring_entry_size,
-					GFP_KERNEL);
-	if (!interrupt_ring)
-		return -ENOMEM;
-
-	kfd->interrupt_ring = interrupt_ring;
-	kfd->interrupt_ring_size =
-		KFD_INTERRUPT_RING_SIZE * kfd->device_info->ih_ring_entry_size;
-	atomic_set(&kfd->interrupt_ring_wptr, 0);
-	atomic_set(&kfd->interrupt_ring_rptr, 0);
+	int r;
+
+	r = kfifo_alloc(&node->ih_fifo,
+		KFD_IH_NUM_ENTRIES * node->kfd->device_info.ih_ring_entry_size,
+		GFP_KERNEL);
+	if (r) {
+		dev_err(node->adev->dev, "Failed to allocate IH fifo\n");
+		return r;
+	}
 
-	spin_lock_init(&kfd->interrupt_lock);
+	if (!node->kfd->ih_wq) {
+		node->kfd->ih_wq = alloc_workqueue("KFD IH", WQ_HIGHPRI | WQ_UNBOUND,
+						   node->kfd->num_nodes);
+		if (unlikely(!node->kfd->ih_wq)) {
+			kfifo_free(&node->ih_fifo);
+			dev_err(node->adev->dev, "Failed to allocate KFD IH workqueue\n");
+			return -ENOMEM;
+		}
+	}
+	spin_lock_init(&node->interrupt_lock);
 
-	INIT_WORK(&kfd->interrupt_work, interrupt_wq);
+	INIT_WORK(&node->interrupt_work, interrupt_wq);
 
-	kfd->interrupts_active = true;
+	node->interrupts_active = true;
 
 	/*
 	 * After this function returns, the interrupt will be enabled. This
@@ -78,7 +87,7 @@ int kfd_interrupt_init(struct kfd_dev *kfd)
 	return 0;
 }
 
-void kfd_interrupt_exit(struct kfd_dev *kfd)
+void kfd_interrupt_exit(struct kfd_node *node)
 {
 	/*
 	 * Stop the interrupt handler from writing to the ring and scheduling
@@ -87,102 +96,73 @@ void kfd_interrupt_exit(struct kfd_dev *kfd)
 	 */
 	unsigned long flags;
 
-	spin_lock_irqsave(&kfd->interrupt_lock, flags);
-	kfd->interrupts_active = false;
-	spin_unlock_irqrestore(&kfd->interrupt_lock, flags);
-
-	/*
-	 * Flush_scheduled_work ensures that there are no outstanding
-	 * work-queue items that will access interrupt_ring. New work items
-	 * can't be created because we stopped interrupt handling above.
-	 */
-	flush_scheduled_work();
-
-	kfree(kfd->interrupt_ring);
+	spin_lock_irqsave(&node->interrupt_lock, flags);
+	node->interrupts_active = false;
+	spin_unlock_irqrestore(&node->interrupt_lock, flags);
+	kfifo_free(&node->ih_fifo);
 }
 
 /*
- * This assumes that it can't be called concurrently with itself
- * but only with dequeue_ih_ring_entry.
+ * Assumption: single reader/writer. This function is not re-entrant
  */
-bool enqueue_ih_ring_entry(struct kfd_dev *kfd,	const void *ih_ring_entry)
+bool enqueue_ih_ring_entry(struct kfd_node *node, const void *ih_ring_entry)
 {
-	unsigned int rptr = atomic_read(&kfd->interrupt_ring_rptr);
-	unsigned int wptr = atomic_read(&kfd->interrupt_ring_wptr);
-
-	if ((rptr - wptr) % kfd->interrupt_ring_size ==
-					kfd->device_info->ih_ring_entry_size) {
-		/* This is very bad, the system is likely to hang. */
-		dev_err_ratelimited(kfd_chardev(),
-			"Interrupt ring overflow, dropping interrupt.\n");
+	if (kfifo_is_full(&node->ih_fifo)) {
+		dev_warn_ratelimited(node->adev->dev, "KFD node %d ih_fifo overflow\n",
+				     node->node_id);
 		return false;
 	}
 
-	memcpy(kfd->interrupt_ring + wptr, ih_ring_entry,
-			kfd->device_info->ih_ring_entry_size);
-
-	wptr = (wptr + kfd->device_info->ih_ring_entry_size) %
-			kfd->interrupt_ring_size;
-	smp_wmb(); /* Ensure memcpy'd data is visible before wptr update. */
-	atomic_set(&kfd->interrupt_ring_wptr, wptr);
-
+	kfifo_in(&node->ih_fifo, ih_ring_entry, node->kfd->device_info.ih_ring_entry_size);
 	return true;
 }
 
 /*
- * This assumes that it can't be called concurrently with itself
- * but only with enqueue_ih_ring_entry.
+ * Assumption: single reader/writer. This function is not re-entrant
  */
-static bool dequeue_ih_ring_entry(struct kfd_dev *kfd, void *ih_ring_entry)
+static bool dequeue_ih_ring_entry(struct kfd_node *node, u32 **ih_ring_entry)
 {
-	/*
-	 * Assume that wait queues have an implicit barrier, i.e. anything that
-	 * happened in the ISR before it queued work is visible.
-	 */
+	int count;
 
-	unsigned int wptr = atomic_read(&kfd->interrupt_ring_wptr);
-	unsigned int rptr = atomic_read(&kfd->interrupt_ring_rptr);
-
-	if (rptr == wptr)
+	if (kfifo_is_empty(&node->ih_fifo))
 		return false;
 
-	memcpy(ih_ring_entry, kfd->interrupt_ring + rptr,
-			kfd->device_info->ih_ring_entry_size);
-
-	rptr = (rptr + kfd->device_info->ih_ring_entry_size) %
-			kfd->interrupt_ring_size;
-
-	/*
-	 * Ensure the rptr write update is not visible until
-	 * memcpy has finished reading.
-	 */
-	smp_mb();
-	atomic_set(&kfd->interrupt_ring_rptr, rptr);
-
-	return true;
+	count = kfifo_out_linear_ptr(&node->ih_fifo, ih_ring_entry,
+				     node->kfd->device_info.ih_ring_entry_size);
+	WARN_ON(count != node->kfd->device_info.ih_ring_entry_size);
+	return count == node->kfd->device_info.ih_ring_entry_size;
 }
 
 static void interrupt_wq(struct work_struct *work)
 {
-	struct kfd_dev *dev = container_of(work, struct kfd_dev,
-						interrupt_work);
-
-	uint32_t ih_ring_entry[DIV_ROUND_UP(
-				dev->device_info->ih_ring_entry_size,
-				sizeof(uint32_t))];
+	struct kfd_node *dev = container_of(work, struct kfd_node, interrupt_work);
+	uint32_t *ih_ring_entry;
+	unsigned long start_jiffies = jiffies;
 
-	while (dequeue_ih_ring_entry(dev, ih_ring_entry))
-		dev->device_info->event_interrupt_class->interrupt_wq(dev,
+	while (dequeue_ih_ring_entry(dev, &ih_ring_entry)) {
+		dev->kfd->device_info.event_interrupt_class->interrupt_wq(dev,
 								ih_ring_entry);
+		kfifo_skip_count(&dev->ih_fifo, dev->kfd->device_info.ih_ring_entry_size);
+
+		if (time_is_before_jiffies(start_jiffies + HZ)) {
+			/* If we spent more than a second processing signals,
+			 * reschedule the worker to avoid soft-lockup warnings
+			 */
+			queue_work(dev->kfd->ih_wq, &dev->interrupt_work);
+			break;
+		}
+	}
 }
 
-bool interrupt_is_wanted(struct kfd_dev *dev, const uint32_t *ih_ring_entry)
+bool interrupt_is_wanted(struct kfd_node *dev,
+			const uint32_t *ih_ring_entry,
+			uint32_t *patched_ihre, bool *flag)
 {
 	/* integer and bitwise OR so there is no boolean short-circuiting */
 	unsigned int wanted = 0;
 
-	wanted |= dev->device_info->event_interrupt_class->interrupt_isr(dev,
-								ih_ring_entry);
+	wanted |= dev->kfd->device_info.event_interrupt_class->interrupt_isr(dev,
+					 ih_ring_entry, patched_ihre, flag);
 
 	return wanted != 0;
 }
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c b/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c
index 681b639f5133..fb3129883a4c 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c
@@ -1,5 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -31,21 +32,21 @@
 #include "kfd_device_queue_manager.h"
 #include "kfd_pm4_headers.h"
 #include "kfd_pm4_opcodes.h"
+#include "amdgpu_reset.h"
 
 #define PM4_COUNT_ZERO (((1 << 15) - 1) << 16)
 
-static bool initialize(struct kernel_queue *kq, struct kfd_dev *dev,
+/* Initialize a kernel queue, including allocations of GART memory
+ * needed for the queue.
+ */
+static bool kq_initialize(struct kernel_queue *kq, struct kfd_node *dev,
 		enum kfd_queue_type type, unsigned int queue_size)
 {
 	struct queue_properties prop;
 	int retval;
 	union PM4_MES_TYPE_3_HEADER nop;
 
-	if (WARN_ON(type != KFD_QUEUE_TYPE_DIQ && type != KFD_QUEUE_TYPE_HIQ))
-		return false;
-
-	pr_debug("Initializing queue type %d size %d\n", KFD_QUEUE_TYPE_HIQ,
-			queue_size);
+	pr_debug("Initializing queue type %d size %d\n", type, queue_size);
 
 	memset(&prop, 0, sizeof(prop));
 	memset(&nop, 0, sizeof(nop));
@@ -58,37 +59,48 @@ static bool initialize(struct kernel_queue *kq, struct kfd_dev *dev,
 	kq->nop_packet = nop.u32all;
 	switch (type) {
 	case KFD_QUEUE_TYPE_DIQ:
+		kq->mqd_mgr = dev->dqm->mqd_mgrs[KFD_MQD_TYPE_DIQ];
+		break;
 	case KFD_QUEUE_TYPE_HIQ:
-		kq->mqd = dev->dqm->ops.get_mqd_manager(dev->dqm,
-						KFD_MQD_TYPE_HIQ);
+		kq->mqd_mgr = dev->dqm->mqd_mgrs[KFD_MQD_TYPE_HIQ];
 		break;
 	default:
-		pr_err("Invalid queue type %d\n", type);
+		WARN(1, "Invalid queue type %d\n", type);
+		dev_err(dev->adev->dev, "Invalid queue type %d\n", type);
 		return false;
 	}
 
-	if (!kq->mqd)
+	if (!kq->mqd_mgr)
 		return false;
 
-	prop.doorbell_ptr = kfd_get_kernel_doorbell(dev, &prop.doorbell_off);
+	prop.doorbell_ptr = kfd_get_kernel_doorbell(dev->kfd, &prop.doorbell_off);
 
 	if (!prop.doorbell_ptr) {
-		pr_err("Failed to initialize doorbell");
+		dev_err(dev->adev->dev, "Failed to initialize doorbell");
 		goto err_get_kernel_doorbell;
 	}
 
 	retval = kfd_gtt_sa_allocate(dev, queue_size, &kq->pq);
 	if (retval != 0) {
-		pr_err("Failed to init pq queues size %d\n", queue_size);
+		dev_err(dev->adev->dev, "Failed to init pq queues size %d\n",
+			queue_size);
 		goto err_pq_allocate_vidmem;
 	}
 
 	kq->pq_kernel_addr = kq->pq->cpu_ptr;
 	kq->pq_gpu_addr = kq->pq->gpu_addr;
 
-	retval = kq->ops_asic_specific.initialize(kq, dev, type, queue_size);
-	if (!retval)
-		goto err_eop_allocate_vidmem;
+	/* For CIK family asics, kq->eop_mem is not needed */
+	if (dev->adev->asic_type > CHIP_MULLINS) {
+		retval = kfd_gtt_sa_allocate(dev, PAGE_SIZE, &kq->eop_mem);
+		if (retval != 0)
+			goto err_eop_allocate_vidmem;
+
+		kq->eop_gpu_addr = kq->eop_mem->gpu_addr;
+		kq->eop_kernel_addr = kq->eop_mem->cpu_ptr;
+
+		memset(kq->eop_kernel_addr, 0, PAGE_SIZE);
+	}
 
 	retval = kfd_gtt_sa_allocate(dev, sizeof(*kq->rptr_kernel),
 					&kq->rptr_mem);
@@ -99,7 +111,7 @@ static bool initialize(struct kernel_queue *kq, struct kfd_dev *dev,
 	kq->rptr_kernel = kq->rptr_mem->cpu_ptr;
 	kq->rptr_gpu_addr = kq->rptr_mem->gpu_addr;
 
-	retval = kfd_gtt_sa_allocate(dev, sizeof(*kq->wptr_kernel),
+	retval = kfd_gtt_sa_allocate(dev, dev->kfd->device_info.doorbell_size,
 					&kq->wptr_mem);
 
 	if (retval != 0)
@@ -110,10 +122,11 @@ static bool initialize(struct kernel_queue *kq, struct kfd_dev *dev,
 
 	memset(kq->pq_kernel_addr, 0, queue_size);
 	memset(kq->rptr_kernel, 0, sizeof(*kq->rptr_kernel));
-	memset(kq->wptr_kernel, 0, sizeof(*kq->wptr_kernel));
+	memset(kq->wptr_kernel, 0, dev->kfd->device_info.doorbell_size);
 
 	prop.queue_size = queue_size;
 	prop.is_interop = false;
+	prop.is_gws = false;
 	prop.priority = 1;
 	prop.queue_percent = 100;
 	prop.type = type;
@@ -130,21 +143,22 @@ static bool initialize(struct kernel_queue *kq, struct kfd_dev *dev,
 	kq->queue->device = dev;
 	kq->queue->process = kfd_get_process(current);
 
-	retval = kq->mqd->init_mqd(kq->mqd, &kq->queue->mqd,
-					&kq->queue->mqd_mem_obj,
+	kq->queue->mqd_mem_obj = kq->mqd_mgr->allocate_mqd(kq->mqd_mgr->dev,
+					&kq->queue->properties);
+	if (!kq->queue->mqd_mem_obj)
+		goto err_allocate_mqd;
+	kq->mqd_mgr->init_mqd(kq->mqd_mgr, &kq->queue->mqd,
+					kq->queue->mqd_mem_obj,
 					&kq->queue->gart_mqd_addr,
 					&kq->queue->properties);
-	if (retval != 0)
-		goto err_init_mqd;
-
 	/* assign HIQ to HQD */
 	if (type == KFD_QUEUE_TYPE_HIQ) {
 		pr_debug("Assigning hiq to hqd\n");
 		kq->queue->pipe = KFD_CIK_HIQ_PIPE;
 		kq->queue->queue = KFD_CIK_HIQ_QUEUE;
-		kq->mqd->load_mqd(kq->mqd, kq->queue->mqd, kq->queue->pipe,
-				  kq->queue->queue, &kq->queue->properties,
-				  NULL);
+		kq->mqd_mgr->load_mqd(kq->mqd_mgr, kq->queue->mqd,
+				kq->queue->pipe, kq->queue->queue,
+				&kq->queue->properties, NULL);
 	} else {
 		/* allocate fence for DIQ */
 
@@ -162,7 +176,8 @@ static bool initialize(struct kernel_queue *kq, struct kfd_dev *dev,
 
 	return true;
 err_alloc_fence:
-err_init_mqd:
+	kq->mqd_mgr->free_mqd(kq->mqd_mgr, kq->queue->mqd, kq->queue->mqd_mem_obj);
+err_allocate_mqd:
 	uninit_queue(kq->queue);
 err_init_queue:
 	kfd_gtt_sa_free(dev, kq->wptr_mem);
@@ -173,79 +188,106 @@ err_rptr_allocate_vidmem:
 err_eop_allocate_vidmem:
 	kfd_gtt_sa_free(dev, kq->pq);
 err_pq_allocate_vidmem:
-	kfd_release_kernel_doorbell(dev, prop.doorbell_ptr);
+	kfd_release_kernel_doorbell(dev->kfd, prop.doorbell_ptr);
 err_get_kernel_doorbell:
 	return false;
 
 }
 
-static void uninitialize(struct kernel_queue *kq)
+/* Uninitialize a kernel queue and free all its memory usages. */
+static void kq_uninitialize(struct kernel_queue *kq)
 {
-	if (kq->queue->properties.type == KFD_QUEUE_TYPE_HIQ)
-		kq->mqd->destroy_mqd(kq->mqd,
-					NULL,
-					false,
-					QUEUE_PREEMPT_DEFAULT_TIMEOUT_MS,
+	if (kq->queue->properties.type == KFD_QUEUE_TYPE_HIQ && down_read_trylock(&kq->dev->adev->reset_domain->sem)) {
+		kq->mqd_mgr->destroy_mqd(kq->mqd_mgr,
+					kq->queue->mqd,
+					KFD_PREEMPT_TYPE_WAVEFRONT_RESET,
+					KFD_UNMAP_LATENCY_MS,
 					kq->queue->pipe,
 					kq->queue->queue);
+		up_read(&kq->dev->adev->reset_domain->sem);
+	}
 	else if (kq->queue->properties.type == KFD_QUEUE_TYPE_DIQ)
 		kfd_gtt_sa_free(kq->dev, kq->fence_mem_obj);
 
-	kq->mqd->uninit_mqd(kq->mqd, kq->queue->mqd, kq->queue->mqd_mem_obj);
+	kq->mqd_mgr->free_mqd(kq->mqd_mgr, kq->queue->mqd,
+				kq->queue->mqd_mem_obj);
 
 	kfd_gtt_sa_free(kq->dev, kq->rptr_mem);
 	kfd_gtt_sa_free(kq->dev, kq->wptr_mem);
-	kq->ops_asic_specific.uninitialize(kq);
+
+	/* For CIK family asics, kq->eop_mem is Null, kfd_gtt_sa_free()
+	 * is able to handle NULL properly.
+	 */
+	kfd_gtt_sa_free(kq->dev, kq->eop_mem);
+
 	kfd_gtt_sa_free(kq->dev, kq->pq);
-	kfd_release_kernel_doorbell(kq->dev,
+	kfd_release_kernel_doorbell(kq->dev->kfd,
 					kq->queue->properties.doorbell_ptr);
 	uninit_queue(kq->queue);
 }
 
-static int acquire_packet_buffer(struct kernel_queue *kq,
+int kq_acquire_packet_buffer(struct kernel_queue *kq,
 		size_t packet_size_in_dwords, unsigned int **buffer_ptr)
 {
 	size_t available_size;
 	size_t queue_size_dwords;
 	uint32_t wptr, rptr;
+	uint64_t wptr64;
 	unsigned int *queue_address;
 
+	/* When rptr == wptr, the buffer is empty.
+	 * When rptr == wptr + 1, the buffer is full.
+	 * It is always rptr that advances to the position of wptr, rather than
+	 * the opposite. So we can only use up to queue_size_dwords - 1 dwords.
+	 */
 	rptr = *kq->rptr_kernel;
-	wptr = *kq->wptr_kernel;
+	wptr = kq->pending_wptr;
+	wptr64 = kq->pending_wptr64;
 	queue_address = (unsigned int *)kq->pq_kernel_addr;
-	queue_size_dwords = kq->queue->properties.queue_size / sizeof(uint32_t);
+	queue_size_dwords = kq->queue->properties.queue_size / 4;
 
 	pr_debug("rptr: %d\n", rptr);
 	pr_debug("wptr: %d\n", wptr);
 	pr_debug("queue_address 0x%p\n", queue_address);
 
-	available_size = (rptr - 1 - wptr + queue_size_dwords) %
+	available_size = (rptr + queue_size_dwords - 1 - wptr) %
 							queue_size_dwords;
 
-	if (packet_size_in_dwords >= queue_size_dwords ||
-			packet_size_in_dwords >= available_size) {
+	if (packet_size_in_dwords > available_size) {
 		/*
 		 * make sure calling functions know
 		 * acquire_packet_buffer() failed
 		 */
-		*buffer_ptr = NULL;
-		return -ENOMEM;
+		goto err_no_space;
 	}
 
 	if (wptr + packet_size_in_dwords >= queue_size_dwords) {
+		/* make sure after rolling back to position 0, there is
+		 * still enough space.
+		 */
+		if (packet_size_in_dwords >= rptr)
+			goto err_no_space;
+
+		/* fill nops, roll back and start at position 0 */
 		while (wptr > 0) {
 			queue_address[wptr] = kq->nop_packet;
 			wptr = (wptr + 1) % queue_size_dwords;
+			wptr64++;
 		}
 	}
 
 	*buffer_ptr = &queue_address[wptr];
 	kq->pending_wptr = wptr + packet_size_in_dwords;
+	kq->pending_wptr64 = wptr64 + packet_size_in_dwords;
 
 	return 0;
+
+err_no_space:
+	*buffer_ptr = NULL;
+	return -ENOMEM;
 }
 
-static void submit_packet(struct kernel_queue *kq)
+int kq_submit_packet(struct kernel_queue *kq)
 {
 #ifdef DEBUG
 	int i;
@@ -257,18 +299,40 @@ static void submit_packet(struct kernel_queue *kq)
 	}
 	pr_debug("\n");
 #endif
+	/* Fatal err detected, packet submission won't go through */
+	if (amdgpu_amdkfd_is_fed(kq->dev->adev))
+		return -EIO;
+
+	/* Make sure ring buffer is updated before wptr updated */
+	mb();
+
+	if (kq->dev->kfd->device_info.doorbell_size == 8) {
+		*kq->wptr64_kernel = kq->pending_wptr64;
+		mb(); /* Make sure wptr updated before ring doorbell */
+		write_kernel_doorbell64(kq->queue->properties.doorbell_ptr,
+					kq->pending_wptr64);
+	} else {
+		*kq->wptr_kernel = kq->pending_wptr;
+		mb(); /* Make sure wptr updated before ring doorbell */
+		write_kernel_doorbell(kq->queue->properties.doorbell_ptr,
+					kq->pending_wptr);
+	}
 
-	*kq->wptr_kernel = kq->pending_wptr;
-	write_kernel_doorbell(kq->queue->properties.doorbell_ptr,
-				kq->pending_wptr);
+	return 0;
 }
 
-static void rollback_packet(struct kernel_queue *kq)
+void kq_rollback_packet(struct kernel_queue *kq)
 {
-	kq->pending_wptr = *kq->queue->properties.write_ptr;
+	if (kq->dev->kfd->device_info.doorbell_size == 8) {
+		kq->pending_wptr64 = *kq->wptr64_kernel;
+		kq->pending_wptr = *kq->wptr_kernel %
+			(kq->queue->properties.queue_size / 4);
+	} else {
+		kq->pending_wptr = *kq->wptr_kernel;
+	}
 }
 
-struct kernel_queue *kernel_queue_init(struct kfd_dev *dev,
+struct kernel_queue *kernel_queue_init(struct kfd_node *dev,
 					enum kfd_queue_type type)
 {
 	struct kernel_queue *kq;
@@ -277,63 +341,48 @@ struct kernel_queue *kernel_queue_init(struct kfd_dev *dev,
 	if (!kq)
 		return NULL;
 
-	kq->ops.initialize = initialize;
-	kq->ops.uninitialize = uninitialize;
-	kq->ops.acquire_packet_buffer = acquire_packet_buffer;
-	kq->ops.submit_packet = submit_packet;
-	kq->ops.rollback_packet = rollback_packet;
+	if (kq_initialize(kq, dev, type, KFD_KERNEL_QUEUE_SIZE))
+		return kq;
 
-	switch (dev->device_info->asic_family) {
-	case CHIP_CARRIZO:
-		kernel_queue_init_vi(&kq->ops_asic_specific);
-		break;
-
-	case CHIP_KAVERI:
-		kernel_queue_init_cik(&kq->ops_asic_specific);
-		break;
-	}
+	dev_err(dev->adev->dev, "Failed to init kernel queue\n");
 
-	if (!kq->ops.initialize(kq, dev, type, KFD_KERNEL_QUEUE_SIZE)) {
-		pr_err("Failed to init kernel queue\n");
-		kfree(kq);
-		return NULL;
-	}
-	return kq;
+	kfree(kq);
+	return NULL;
 }
 
 void kernel_queue_uninit(struct kernel_queue *kq)
 {
-	kq->ops.uninitialize(kq);
+	kq_uninitialize(kq);
 	kfree(kq);
 }
 
 /* FIXME: Can this test be removed? */
-static __attribute__((unused)) void test_kq(struct kfd_dev *dev)
+static __attribute__((unused)) void test_kq(struct kfd_node *dev)
 {
 	struct kernel_queue *kq;
 	uint32_t *buffer, i;
 	int retval;
 
-	pr_err("Starting kernel queue test\n");
+	dev_err(dev->adev->dev, "Starting kernel queue test\n");
 
 	kq = kernel_queue_init(dev, KFD_QUEUE_TYPE_HIQ);
 	if (unlikely(!kq)) {
-		pr_err("  Failed to initialize HIQ\n");
-		pr_err("Kernel queue test failed\n");
+		dev_err(dev->adev->dev, "  Failed to initialize HIQ\n");
+		dev_err(dev->adev->dev, "Kernel queue test failed\n");
 		return;
 	}
 
-	retval = kq->ops.acquire_packet_buffer(kq, 5, &buffer);
+	retval = kq_acquire_packet_buffer(kq, 5, &buffer);
 	if (unlikely(retval != 0)) {
-		pr_err("  Failed to acquire packet buffer\n");
-		pr_err("Kernel queue test failed\n");
+		dev_err(dev->adev->dev, "  Failed to acquire packet buffer\n");
+		dev_err(dev->adev->dev, "Kernel queue test failed\n");
 		return;
 	}
 	for (i = 0; i < 5; i++)
 		buffer[i] = kq->nop_packet;
-	kq->ops.submit_packet(kq);
+	kq_submit_packet(kq);
 
-	pr_err("Ending kernel queue test\n");
+	dev_err(dev->adev->dev, "Ending kernel queue test\n");
 }
 
 
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.h b/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.h
index 594053136ee4..e24ee50acdf0 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.h
@@ -1,5 +1,6 @@
+/* SPDX-License-Identifier: GPL-2.0 OR MIT */
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -29,49 +30,33 @@
 #include "kfd_priv.h"
 
 /**
- * struct kernel_queue_ops
- *
- * @initialize: Initialize a kernel queue, including allocations of GART memory
- * needed for the queue.
- *
- * @uninitialize: Uninitialize a kernel queue and free all its memory usages.
- *
- * @acquire_packet_buffer: Returns a pointer to the location in the kernel
+ * kq_acquire_packet_buffer: Returns a pointer to the location in the kernel
  * queue ring buffer where the calling function can write its packet. It is
  * Guaranteed that there is enough space for that packet. It also updates the
  * pending write pointer to that location so subsequent calls to
  * acquire_packet_buffer will get a correct write pointer
  *
- * @submit_packet: Update the write pointer and doorbell of a kernel queue.
- *
- * @sync_with_hw: Wait until the write pointer and the read pointer of a kernel
- * queue are equal, which means the CP has read all the submitted packets.
+ * kq_submit_packet: Update the write pointer and doorbell of a kernel queue.
  *
- * @rollback_packet: This routine is called if we failed to build an acquired
+ * kq_rollback_packet: This routine is called if we failed to build an acquired
  * packet for some reason. It just overwrites the pending wptr with the current
  * one
  *
  */
-struct kernel_queue_ops {
-	bool	(*initialize)(struct kernel_queue *kq, struct kfd_dev *dev,
-			enum kfd_queue_type type, unsigned int queue_size);
-	void	(*uninitialize)(struct kernel_queue *kq);
-	int	(*acquire_packet_buffer)(struct kernel_queue *kq,
-					size_t packet_size_in_dwords,
-					unsigned int **buffer_ptr);
 
-	void	(*submit_packet)(struct kernel_queue *kq);
-	void	(*rollback_packet)(struct kernel_queue *kq);
-};
+int kq_acquire_packet_buffer(struct kernel_queue *kq,
+				size_t packet_size_in_dwords,
+				unsigned int **buffer_ptr);
+int kq_submit_packet(struct kernel_queue *kq);
+void kq_rollback_packet(struct kernel_queue *kq);
 
-struct kernel_queue {
-	struct kernel_queue_ops ops;
-	struct kernel_queue_ops ops_asic_specific;
 
+struct kernel_queue {
 	/* data */
-	struct kfd_dev		*dev;
-	struct mqd_manager	*mqd;
+	struct kfd_node		*dev;
+	struct mqd_manager	*mqd_mgr;
 	struct queue		*queue;
+	uint64_t		pending_wptr64;
 	uint32_t		pending_wptr;
 	unsigned int		nop_packet;
 
@@ -79,7 +64,10 @@ struct kernel_queue {
 	uint32_t		*rptr_kernel;
 	uint64_t		rptr_gpu_addr;
 	struct kfd_mem_obj	*wptr_mem;
-	uint32_t		*wptr_kernel;
+	union {
+		uint64_t	*wptr64_kernel;
+		uint32_t	*wptr_kernel;
+	};
 	uint64_t		wptr_gpu_addr;
 	struct kfd_mem_obj	*pq;
 	uint64_t		pq_gpu_addr;
@@ -95,7 +83,4 @@ struct kernel_queue {
 	struct list_head	list;
 };
 
-void kernel_queue_init_cik(struct kernel_queue_ops *ops);
-void kernel_queue_init_vi(struct kernel_queue_ops *ops);
-
 #endif /* KFD_KERNEL_QUEUE_H_ */
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue_cik.c b/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue_cik.c
deleted file mode 100644
index a90eb440b1fb..000000000000
--- a/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue_cik.c
+++ /dev/null
@@ -1,44 +0,0 @@
-/*
- * Copyright 2014 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-
-#include "kfd_kernel_queue.h"
-
-static bool initialize_cik(struct kernel_queue *kq, struct kfd_dev *dev,
-			enum kfd_queue_type type, unsigned int queue_size);
-static void uninitialize_cik(struct kernel_queue *kq);
-
-void kernel_queue_init_cik(struct kernel_queue_ops *ops)
-{
-	ops->initialize = initialize_cik;
-	ops->uninitialize = uninitialize_cik;
-}
-
-static bool initialize_cik(struct kernel_queue *kq, struct kfd_dev *dev,
-			enum kfd_queue_type type, unsigned int queue_size)
-{
-	return true;
-}
-
-static void uninitialize_cik(struct kernel_queue *kq)
-{
-}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue_vi.c b/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue_vi.c
deleted file mode 100644
index f1d48281e322..000000000000
--- a/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue_vi.c
+++ /dev/null
@@ -1,56 +0,0 @@
-/*
- * Copyright 2014 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-
-#include "kfd_kernel_queue.h"
-
-static bool initialize_vi(struct kernel_queue *kq, struct kfd_dev *dev,
-			enum kfd_queue_type type, unsigned int queue_size);
-static void uninitialize_vi(struct kernel_queue *kq);
-
-void kernel_queue_init_vi(struct kernel_queue_ops *ops)
-{
-	ops->initialize = initialize_vi;
-	ops->uninitialize = uninitialize_vi;
-}
-
-static bool initialize_vi(struct kernel_queue *kq, struct kfd_dev *dev,
-			enum kfd_queue_type type, unsigned int queue_size)
-{
-	int retval;
-
-	retval = kfd_gtt_sa_allocate(dev, PAGE_SIZE, &kq->eop_mem);
-	if (retval != 0)
-		return false;
-
-	kq->eop_gpu_addr = kq->eop_mem->gpu_addr;
-	kq->eop_kernel_addr = kq->eop_mem->cpu_ptr;
-
-	memset(kq->eop_kernel_addr, 0, PAGE_SIZE);
-
-	return true;
-}
-
-static void uninitialize_vi(struct kernel_queue *kq)
-{
-	kfd_gtt_sa_free(kq->dev, kq->eop_mem);
-}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c
new file mode 100644
index 000000000000..59a5a3fea65d
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c
@@ -0,0 +1,1079 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
+/*
+ * Copyright 2020-2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+#include <linux/types.h>
+#include <linux/hmm.h>
+#include <linux/dma-direction.h>
+#include <linux/dma-mapping.h>
+#include <linux/migrate.h>
+#include "amdgpu_sync.h"
+#include "amdgpu_object.h"
+#include "amdgpu_vm.h"
+#include "amdgpu_res_cursor.h"
+#include "kfd_priv.h"
+#include "kfd_svm.h"
+#include "kfd_migrate.h"
+#include "kfd_smi_events.h"
+
+#ifdef dev_fmt
+#undef dev_fmt
+#endif
+#define dev_fmt(fmt) "kfd_migrate: " fmt
+
+static u64
+svm_migrate_direct_mapping_addr(struct amdgpu_device *adev, u64 addr)
+{
+	return addr + amdgpu_ttm_domain_start(adev, TTM_PL_VRAM);
+}
+
+static int
+svm_migrate_gart_map(struct amdgpu_ring *ring, u64 npages,
+		     dma_addr_t *addr, u64 *gart_addr, u64 flags)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_job *job;
+	unsigned int num_dw, num_bytes;
+	struct dma_fence *fence;
+	u64 src_addr, dst_addr;
+	u64 pte_flags;
+	void *cpu_addr;
+	int r;
+
+	/* use gart window 0 */
+	*gart_addr = adev->gmc.gart_start;
+
+	num_dw = ALIGN(adev->mman.buffer_funcs->copy_num_dw, 8);
+	num_bytes = npages * 8;
+
+	r = amdgpu_job_alloc_with_ib(adev, &adev->mman.high_pr,
+				     AMDGPU_FENCE_OWNER_UNDEFINED,
+				     num_dw * 4 + num_bytes,
+				     AMDGPU_IB_POOL_DELAYED,
+				     &job,
+				     AMDGPU_KERNEL_JOB_ID_KFD_GART_MAP);
+	if (r)
+		return r;
+
+	src_addr = num_dw * 4;
+	src_addr += job->ibs[0].gpu_addr;
+
+	dst_addr = amdgpu_bo_gpu_offset(adev->gart.bo);
+	amdgpu_emit_copy_buffer(adev, &job->ibs[0], src_addr,
+				dst_addr, num_bytes, 0);
+
+	amdgpu_ring_pad_ib(ring, &job->ibs[0]);
+	WARN_ON(job->ibs[0].length_dw > num_dw);
+
+	pte_flags = AMDGPU_PTE_VALID | AMDGPU_PTE_READABLE;
+	pte_flags |= AMDGPU_PTE_SYSTEM | AMDGPU_PTE_SNOOPED;
+	if (!(flags & KFD_IOCTL_SVM_FLAG_GPU_RO))
+		pte_flags |= AMDGPU_PTE_WRITEABLE;
+	pte_flags |= adev->gart.gart_pte_flags;
+
+	cpu_addr = &job->ibs[0].ptr[num_dw];
+
+	amdgpu_gart_map(adev, 0, npages, addr, pte_flags, cpu_addr);
+	fence = amdgpu_job_submit(job);
+	dma_fence_put(fence);
+
+	return r;
+}
+
+/**
+ * svm_migrate_copy_memory_gart - sdma copy data between ram and vram
+ *
+ * @adev: amdgpu device the sdma ring running
+ * @sys: system DMA pointer to be copied
+ * @vram: vram destination DMA pointer
+ * @npages: number of pages to copy
+ * @direction: enum MIGRATION_COPY_DIR
+ * @mfence: output, sdma fence to signal after sdma is done
+ *
+ * ram address uses GART table continuous entries mapping to ram pages,
+ * vram address uses direct mapping of vram pages, which must have npages
+ * number of continuous pages.
+ * GART update and sdma uses same buf copy function ring, sdma is splited to
+ * multiple GTT_MAX_PAGES transfer, all sdma operations are serialized, wait for
+ * the last sdma finish fence which is returned to check copy memory is done.
+ *
+ * Context: Process context, takes and releases gtt_window_lock
+ *
+ * Return:
+ * 0 - OK, otherwise error code
+ */
+
+static int
+svm_migrate_copy_memory_gart(struct amdgpu_device *adev, dma_addr_t *sys,
+			     u64 *vram, u64 npages,
+			     enum MIGRATION_COPY_DIR direction,
+			     struct dma_fence **mfence)
+{
+	const u64 GTT_MAX_PAGES = AMDGPU_GTT_MAX_TRANSFER_SIZE;
+	struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
+	u64 gart_s, gart_d;
+	struct dma_fence *next;
+	u64 size;
+	int r;
+
+	mutex_lock(&adev->mman.gtt_window_lock);
+
+	while (npages) {
+		size = min(GTT_MAX_PAGES, npages);
+
+		if (direction == FROM_VRAM_TO_RAM) {
+			gart_s = svm_migrate_direct_mapping_addr(adev, *vram);
+			r = svm_migrate_gart_map(ring, size, sys, &gart_d, 0);
+
+		} else if (direction == FROM_RAM_TO_VRAM) {
+			r = svm_migrate_gart_map(ring, size, sys, &gart_s,
+						 KFD_IOCTL_SVM_FLAG_GPU_RO);
+			gart_d = svm_migrate_direct_mapping_addr(adev, *vram);
+		}
+		if (r) {
+			dev_err(adev->dev, "fail %d create gart mapping\n", r);
+			goto out_unlock;
+		}
+
+		r = amdgpu_copy_buffer(ring, gart_s, gart_d, size * PAGE_SIZE,
+				       NULL, &next, false, true, 0);
+		if (r) {
+			dev_err(adev->dev, "fail %d to copy memory\n", r);
+			goto out_unlock;
+		}
+
+		dma_fence_put(*mfence);
+		*mfence = next;
+		npages -= size;
+		if (npages) {
+			sys += size;
+			vram += size;
+		}
+	}
+
+out_unlock:
+	mutex_unlock(&adev->mman.gtt_window_lock);
+
+	return r;
+}
+
+/**
+ * svm_migrate_copy_done - wait for memory copy sdma is done
+ *
+ * @adev: amdgpu device the sdma memory copy is executing on
+ * @mfence: migrate fence
+ *
+ * Wait for dma fence is signaled, if the copy ssplit into multiple sdma
+ * operations, this is the last sdma operation fence.
+ *
+ * Context: called after svm_migrate_copy_memory
+ *
+ * Return:
+ * 0		- success
+ * otherwise	- error code from dma fence signal
+ */
+static int
+svm_migrate_copy_done(struct amdgpu_device *adev, struct dma_fence *mfence)
+{
+	int r = 0;
+
+	if (mfence) {
+		r = dma_fence_wait(mfence, false);
+		dma_fence_put(mfence);
+		pr_debug("sdma copy memory fence done\n");
+	}
+
+	return r;
+}
+
+unsigned long
+svm_migrate_addr_to_pfn(struct amdgpu_device *adev, unsigned long addr)
+{
+	return (addr + adev->kfd.pgmap.range.start) >> PAGE_SHIFT;
+}
+
+static void
+svm_migrate_get_vram_page(struct svm_range *prange, unsigned long pfn)
+{
+	struct page *page;
+
+	page = pfn_to_page(pfn);
+	svm_range_bo_ref(prange->svm_bo);
+	page->zone_device_data = prange->svm_bo;
+	zone_device_page_init(page);
+}
+
+static void
+svm_migrate_put_vram_page(struct amdgpu_device *adev, unsigned long addr)
+{
+	struct page *page;
+
+	page = pfn_to_page(svm_migrate_addr_to_pfn(adev, addr));
+	unlock_page(page);
+	put_page(page);
+}
+
+static unsigned long
+svm_migrate_addr(struct amdgpu_device *adev, struct page *page)
+{
+	unsigned long addr;
+
+	addr = page_to_pfn(page) << PAGE_SHIFT;
+	return (addr - adev->kfd.pgmap.range.start);
+}
+
+static struct page *
+svm_migrate_get_sys_page(struct vm_area_struct *vma, unsigned long addr)
+{
+	struct page *page;
+
+	page = alloc_page_vma(GFP_HIGHUSER, vma, addr);
+	if (page)
+		lock_page(page);
+
+	return page;
+}
+
+static void svm_migrate_put_sys_page(unsigned long addr)
+{
+	struct page *page;
+
+	page = pfn_to_page(addr >> PAGE_SHIFT);
+	unlock_page(page);
+	put_page(page);
+}
+
+static unsigned long svm_migrate_successful_pages(struct migrate_vma *migrate)
+{
+	unsigned long mpages = 0;
+	unsigned long i;
+
+	for (i = 0; i < migrate->npages; i++) {
+		if (migrate->dst[i] & MIGRATE_PFN_VALID &&
+		    migrate->src[i] & MIGRATE_PFN_MIGRATE)
+			mpages++;
+	}
+	return mpages;
+}
+
+static int
+svm_migrate_copy_to_vram(struct kfd_node *node, struct svm_range *prange,
+			 struct migrate_vma *migrate, struct dma_fence **mfence,
+			 dma_addr_t *scratch, u64 ttm_res_offset)
+{
+	u64 npages = migrate->npages;
+	struct amdgpu_device *adev = node->adev;
+	struct device *dev = adev->dev;
+	struct amdgpu_res_cursor cursor;
+	u64 mpages = 0;
+	dma_addr_t *src;
+	u64 *dst;
+	u64 i, j;
+	int r;
+
+	pr_debug("svms 0x%p [0x%lx 0x%lx 0x%llx]\n", prange->svms, prange->start,
+		 prange->last, ttm_res_offset);
+
+	src = scratch;
+	dst = (u64 *)(scratch + npages);
+
+	amdgpu_res_first(prange->ttm_res, ttm_res_offset,
+			 npages << PAGE_SHIFT, &cursor);
+	for (i = j = 0; (i < npages) && (mpages < migrate->cpages); i++) {
+		struct page *spage;
+
+		if (migrate->src[i] & MIGRATE_PFN_MIGRATE) {
+			dst[i] = cursor.start + (j << PAGE_SHIFT);
+			migrate->dst[i] = svm_migrate_addr_to_pfn(adev, dst[i]);
+			svm_migrate_get_vram_page(prange, migrate->dst[i]);
+			migrate->dst[i] = migrate_pfn(migrate->dst[i]);
+			mpages++;
+		}
+		spage = migrate_pfn_to_page(migrate->src[i]);
+		if (spage && !is_zone_device_page(spage)) {
+			src[i] = dma_map_page(dev, spage, 0, PAGE_SIZE,
+					      DMA_BIDIRECTIONAL);
+			r = dma_mapping_error(dev, src[i]);
+			if (r) {
+				dev_err(dev, "%s: fail %d dma_map_page\n",
+					__func__, r);
+				goto out_free_vram_pages;
+			}
+		} else {
+			if (j) {
+				r = svm_migrate_copy_memory_gart(
+						adev, src + i - j,
+						dst + i - j, j,
+						FROM_RAM_TO_VRAM,
+						mfence);
+				if (r)
+					goto out_free_vram_pages;
+				amdgpu_res_next(&cursor, (j + 1) << PAGE_SHIFT);
+				j = 0;
+			} else {
+				amdgpu_res_next(&cursor, PAGE_SIZE);
+			}
+			continue;
+		}
+
+		pr_debug_ratelimited("dma mapping src to 0x%llx, pfn 0x%lx\n",
+				     src[i] >> PAGE_SHIFT, page_to_pfn(spage));
+
+		if (j >= (cursor.size >> PAGE_SHIFT) - 1 && i < npages - 1) {
+			r = svm_migrate_copy_memory_gart(adev, src + i - j,
+							 dst + i - j, j + 1,
+							 FROM_RAM_TO_VRAM,
+							 mfence);
+			if (r)
+				goto out_free_vram_pages;
+			amdgpu_res_next(&cursor, (j + 1) * PAGE_SIZE);
+			j = 0;
+		} else {
+			j++;
+		}
+	}
+
+	r = svm_migrate_copy_memory_gart(adev, src + i - j, dst + i - j, j,
+					 FROM_RAM_TO_VRAM, mfence);
+
+out_free_vram_pages:
+	if (r) {
+		pr_debug("failed %d to copy memory to vram\n", r);
+		for (i = 0; i < npages && mpages; i++) {
+			if (!dst[i])
+				continue;
+			svm_migrate_put_vram_page(adev, dst[i]);
+			migrate->dst[i] = 0;
+			mpages--;
+		}
+	}
+
+#ifdef DEBUG_FORCE_MIXED_DOMAINS
+	for (i = 0, j = 0; i < npages; i += 4, j++) {
+		if (j & 1)
+			continue;
+		svm_migrate_put_vram_page(adev, dst[i]);
+		migrate->dst[i] = 0;
+		svm_migrate_put_vram_page(adev, dst[i + 1]);
+		migrate->dst[i + 1] = 0;
+		svm_migrate_put_vram_page(adev, dst[i + 2]);
+		migrate->dst[i + 2] = 0;
+		svm_migrate_put_vram_page(adev, dst[i + 3]);
+		migrate->dst[i + 3] = 0;
+	}
+#endif
+
+	return r;
+}
+
+static long
+svm_migrate_vma_to_vram(struct kfd_node *node, struct svm_range *prange,
+			struct vm_area_struct *vma, u64 start,
+			u64 end, uint32_t trigger, u64 ttm_res_offset)
+{
+	struct kfd_process *p = container_of(prange->svms, struct kfd_process, svms);
+	u64 npages = (end - start) >> PAGE_SHIFT;
+	struct amdgpu_device *adev = node->adev;
+	struct kfd_process_device *pdd;
+	struct dma_fence *mfence = NULL;
+	struct migrate_vma migrate = { 0 };
+	unsigned long cpages = 0;
+	unsigned long mpages = 0;
+	dma_addr_t *scratch;
+	void *buf;
+	int r = -ENOMEM;
+
+	memset(&migrate, 0, sizeof(migrate));
+	migrate.vma = vma;
+	migrate.start = start;
+	migrate.end = end;
+	migrate.flags = MIGRATE_VMA_SELECT_SYSTEM;
+	migrate.pgmap_owner = SVM_ADEV_PGMAP_OWNER(adev);
+
+	buf = kvcalloc(npages,
+		       2 * sizeof(*migrate.src) + sizeof(u64) + sizeof(dma_addr_t),
+		       GFP_KERNEL);
+	if (!buf)
+		goto out;
+
+	migrate.src = buf;
+	migrate.dst = migrate.src + npages;
+	scratch = (dma_addr_t *)(migrate.dst + npages);
+
+	kfd_smi_event_migration_start(node, p->lead_thread->pid,
+				      start >> PAGE_SHIFT, end >> PAGE_SHIFT,
+				      0, node->id, prange->prefetch_loc,
+				      prange->preferred_loc, trigger);
+
+	r = migrate_vma_setup(&migrate);
+	if (r) {
+		dev_err(adev->dev, "%s: vma setup fail %d range [0x%lx 0x%lx]\n",
+			__func__, r, prange->start, prange->last);
+		goto out_free;
+	}
+
+	cpages = migrate.cpages;
+	if (!cpages) {
+		pr_debug("failed collect migrate sys pages [0x%lx 0x%lx]\n",
+			 prange->start, prange->last);
+		goto out_free;
+	}
+	if (cpages != npages)
+		pr_debug("partial migration, 0x%lx/0x%llx pages collected\n",
+			 cpages, npages);
+	else
+		pr_debug("0x%lx pages collected\n", cpages);
+
+	r = svm_migrate_copy_to_vram(node, prange, &migrate, &mfence, scratch, ttm_res_offset);
+	migrate_vma_pages(&migrate);
+
+	svm_migrate_copy_done(adev, mfence);
+	migrate_vma_finalize(&migrate);
+
+	mpages = svm_migrate_successful_pages(&migrate);
+	pr_debug("migrated/collected/requested 0x%lx/0x%lx/0x%lx\n",
+		 mpages, cpages, migrate.npages);
+
+	svm_range_dma_unmap_dev(adev->dev, scratch, 0, npages);
+
+out_free:
+	kvfree(buf);
+	kfd_smi_event_migration_end(node, p->lead_thread->pid,
+				    start >> PAGE_SHIFT, end >> PAGE_SHIFT,
+				    0, node->id, trigger, r);
+out:
+	if (!r && mpages) {
+		pdd = svm_range_get_pdd_by_node(prange, node);
+		if (pdd)
+			WRITE_ONCE(pdd->page_in, pdd->page_in + mpages);
+
+		return mpages;
+	}
+	return r;
+}
+
+/**
+ * svm_migrate_ram_to_vram - migrate svm range from system to device
+ * @prange: range structure
+ * @best_loc: the device to migrate to
+ * @start_mgr: start page to migrate
+ * @last_mgr: last page to migrate
+ * @mm: the process mm structure
+ * @trigger: reason of migration
+ *
+ * Context: Process context, caller hold mmap read lock, svms lock, prange lock
+ *
+ * Return:
+ * 0 - OK, otherwise error code
+ */
+static int
+svm_migrate_ram_to_vram(struct svm_range *prange, uint32_t best_loc,
+			unsigned long start_mgr, unsigned long last_mgr,
+			struct mm_struct *mm, uint32_t trigger)
+{
+	unsigned long addr, start, end;
+	struct vm_area_struct *vma;
+	u64 ttm_res_offset;
+	struct kfd_node *node;
+	unsigned long mpages = 0;
+	long r = 0;
+
+	if (start_mgr < prange->start || last_mgr > prange->last) {
+		pr_debug("range [0x%lx 0x%lx] out prange [0x%lx 0x%lx]\n",
+			 start_mgr, last_mgr, prange->start, prange->last);
+		return -EFAULT;
+	}
+
+	node = svm_range_get_node_by_id(prange, best_loc);
+	if (!node) {
+		pr_debug("failed to get kfd node by id 0x%x\n", best_loc);
+		return -ENODEV;
+	}
+
+	pr_debug("svms 0x%p [0x%lx 0x%lx] in [0x%lx 0x%lx] to gpu 0x%x\n",
+		prange->svms, start_mgr, last_mgr, prange->start, prange->last,
+		best_loc);
+
+	start = start_mgr << PAGE_SHIFT;
+	end = (last_mgr + 1) << PAGE_SHIFT;
+
+	r = amdgpu_amdkfd_reserve_mem_limit(node->adev,
+					prange->npages * PAGE_SIZE,
+					KFD_IOC_ALLOC_MEM_FLAGS_VRAM,
+					node->xcp ? node->xcp->id : 0);
+	if (r) {
+		dev_dbg(node->adev->dev, "failed to reserve VRAM, r: %ld\n", r);
+		return -ENOSPC;
+	}
+
+	r = svm_range_vram_node_new(node, prange, true);
+	if (r) {
+		dev_dbg(node->adev->dev, "fail %ld to alloc vram\n", r);
+		goto out;
+	}
+	ttm_res_offset = (start_mgr - prange->start + prange->offset) << PAGE_SHIFT;
+
+	for (addr = start; addr < end;) {
+		unsigned long next;
+
+		vma = vma_lookup(mm, addr);
+		if (!vma)
+			break;
+
+		next = min(vma->vm_end, end);
+		r = svm_migrate_vma_to_vram(node, prange, vma, addr, next, trigger, ttm_res_offset);
+		if (r < 0) {
+			pr_debug("failed %ld to migrate\n", r);
+			break;
+		} else {
+			mpages += r;
+		}
+		ttm_res_offset += next - addr;
+		addr = next;
+	}
+
+	if (mpages) {
+		prange->actual_loc = best_loc;
+		prange->vram_pages += mpages;
+	} else if (!prange->actual_loc) {
+		/* if no page migrated and all pages from prange are at
+		 * sys ram drop svm_bo got from svm_range_vram_node_new
+		 */
+		svm_range_vram_node_free(prange);
+	}
+
+out:
+	amdgpu_amdkfd_unreserve_mem_limit(node->adev,
+					prange->npages * PAGE_SIZE,
+					KFD_IOC_ALLOC_MEM_FLAGS_VRAM,
+					node->xcp ? node->xcp->id : 0);
+	return r < 0 ? r : 0;
+}
+
+static void svm_migrate_page_free(struct page *page)
+{
+	struct svm_range_bo *svm_bo = page->zone_device_data;
+
+	if (svm_bo) {
+		pr_debug_ratelimited("ref: %d\n", kref_read(&svm_bo->kref));
+		svm_range_bo_unref_async(svm_bo);
+	}
+}
+
+static int
+svm_migrate_copy_to_ram(struct amdgpu_device *adev, struct svm_range *prange,
+			struct migrate_vma *migrate, struct dma_fence **mfence,
+			dma_addr_t *scratch, u64 npages)
+{
+	struct device *dev = adev->dev;
+	u64 *src;
+	dma_addr_t *dst;
+	struct page *dpage;
+	u64 i = 0, j;
+	u64 addr;
+	int r = 0;
+
+	pr_debug("svms 0x%p [0x%lx 0x%lx]\n", prange->svms, prange->start,
+		 prange->last);
+
+	addr = migrate->start;
+
+	src = (u64 *)(scratch + npages);
+	dst = scratch;
+
+	for (i = 0, j = 0; i < npages; i++, addr += PAGE_SIZE) {
+		struct page *spage;
+
+		spage = migrate_pfn_to_page(migrate->src[i]);
+		if (!spage || !is_zone_device_page(spage)) {
+			pr_debug("invalid page. Could be in CPU already svms 0x%p [0x%lx 0x%lx]\n",
+				 prange->svms, prange->start, prange->last);
+			if (j) {
+				r = svm_migrate_copy_memory_gart(adev, dst + i - j,
+								 src + i - j, j,
+								 FROM_VRAM_TO_RAM,
+								 mfence);
+				if (r)
+					goto out_oom;
+				j = 0;
+			}
+			continue;
+		}
+		src[i] = svm_migrate_addr(adev, spage);
+		if (j > 0 && src[i] != src[i - 1] + PAGE_SIZE) {
+			r = svm_migrate_copy_memory_gart(adev, dst + i - j,
+							 src + i - j, j,
+							 FROM_VRAM_TO_RAM,
+							 mfence);
+			if (r)
+				goto out_oom;
+			j = 0;
+		}
+
+		dpage = svm_migrate_get_sys_page(migrate->vma, addr);
+		if (!dpage) {
+			pr_debug("failed get page svms 0x%p [0x%lx 0x%lx]\n",
+				 prange->svms, prange->start, prange->last);
+			r = -ENOMEM;
+			goto out_oom;
+		}
+
+		dst[i] = dma_map_page(dev, dpage, 0, PAGE_SIZE, DMA_BIDIRECTIONAL);
+		r = dma_mapping_error(dev, dst[i]);
+		if (r) {
+			dev_err(adev->dev, "%s: fail %d dma_map_page\n", __func__, r);
+			goto out_oom;
+		}
+
+		pr_debug_ratelimited("dma mapping dst to 0x%llx, pfn 0x%lx\n",
+				     dst[i] >> PAGE_SHIFT, page_to_pfn(dpage));
+
+		migrate->dst[i] = migrate_pfn(page_to_pfn(dpage));
+		j++;
+	}
+
+	r = svm_migrate_copy_memory_gart(adev, dst + i - j, src + i - j, j,
+					 FROM_VRAM_TO_RAM, mfence);
+
+out_oom:
+	if (r) {
+		pr_debug("failed %d copy to ram\n", r);
+		while (i--) {
+			svm_migrate_put_sys_page(dst[i]);
+			migrate->dst[i] = 0;
+		}
+	}
+
+	return r;
+}
+
+/**
+ * svm_migrate_vma_to_ram - migrate range inside one vma from device to system
+ *
+ * @prange: svm range structure
+ * @vma: vm_area_struct that range [start, end] belongs to
+ * @start: range start virtual address in pages
+ * @end: range end virtual address in pages
+ * @node: kfd node device to migrate from
+ * @trigger: reason of migration
+ * @fault_page: is from vmf->page, svm_migrate_to_ram(), this is CPU page fault callback
+ *
+ * Context: Process context, caller hold mmap read lock, prange->migrate_mutex
+ *
+ * Return:
+ *   negative values - indicate error
+ *   positive values or zero - number of pages got migrated
+ */
+static long
+svm_migrate_vma_to_ram(struct kfd_node *node, struct svm_range *prange,
+		       struct vm_area_struct *vma, u64 start, u64 end,
+		       uint32_t trigger, struct page *fault_page)
+{
+	struct kfd_process *p = container_of(prange->svms, struct kfd_process, svms);
+	u64 npages = (end - start) >> PAGE_SHIFT;
+	unsigned long cpages = 0;
+	unsigned long mpages = 0;
+	struct amdgpu_device *adev = node->adev;
+	struct kfd_process_device *pdd;
+	struct dma_fence *mfence = NULL;
+	struct migrate_vma migrate = { 0 };
+	dma_addr_t *scratch;
+	void *buf;
+	int r = -ENOMEM;
+
+	memset(&migrate, 0, sizeof(migrate));
+	migrate.vma = vma;
+	migrate.start = start;
+	migrate.end = end;
+	migrate.pgmap_owner = SVM_ADEV_PGMAP_OWNER(adev);
+	if (adev->gmc.xgmi.connected_to_cpu)
+		migrate.flags = MIGRATE_VMA_SELECT_DEVICE_COHERENT;
+	else
+		migrate.flags = MIGRATE_VMA_SELECT_DEVICE_PRIVATE;
+
+	buf = kvcalloc(npages,
+		       2 * sizeof(*migrate.src) + sizeof(u64) + sizeof(dma_addr_t),
+		       GFP_KERNEL);
+	if (!buf)
+		goto out;
+
+	migrate.src = buf;
+	migrate.dst = migrate.src + npages;
+	migrate.fault_page = fault_page;
+	scratch = (dma_addr_t *)(migrate.dst + npages);
+
+	kfd_smi_event_migration_start(node, p->lead_thread->pid,
+				      start >> PAGE_SHIFT, end >> PAGE_SHIFT,
+				      node->id, 0, prange->prefetch_loc,
+				      prange->preferred_loc, trigger);
+
+	r = migrate_vma_setup(&migrate);
+	if (r) {
+		dev_err(adev->dev, "%s: vma setup fail %d range [0x%lx 0x%lx]\n",
+			__func__, r, prange->start, prange->last);
+		goto out_free;
+	}
+
+	cpages = migrate.cpages;
+	if (!cpages) {
+		pr_debug("failed collect migrate device pages [0x%lx 0x%lx]\n",
+			 prange->start, prange->last);
+		goto out_free;
+	}
+	if (cpages != npages)
+		pr_debug("partial migration, 0x%lx/0x%llx pages collected\n",
+			 cpages, npages);
+	else
+		pr_debug("0x%lx pages collected\n", cpages);
+
+	r = svm_migrate_copy_to_ram(adev, prange, &migrate, &mfence,
+				    scratch, npages);
+	migrate_vma_pages(&migrate);
+
+	mpages = svm_migrate_successful_pages(&migrate);
+	pr_debug("migrated/collected/requested 0x%lx/0x%lx/0x%lx\n",
+		 mpages, cpages, migrate.npages);
+
+	svm_migrate_copy_done(adev, mfence);
+	migrate_vma_finalize(&migrate);
+
+	svm_range_dma_unmap_dev(adev->dev, scratch, 0, npages);
+
+out_free:
+	kvfree(buf);
+	kfd_smi_event_migration_end(node, p->lead_thread->pid,
+				    start >> PAGE_SHIFT, end >> PAGE_SHIFT,
+				    node->id, 0, trigger, r);
+out:
+	if (!r && mpages) {
+		pdd = svm_range_get_pdd_by_node(prange, node);
+		if (pdd)
+			WRITE_ONCE(pdd->page_out, pdd->page_out + mpages);
+	}
+
+	return r ? r : mpages;
+}
+
+/**
+ * svm_migrate_vram_to_ram - migrate svm range from device to system
+ * @prange: range structure
+ * @mm: process mm, use current->mm if NULL
+ * @start_mgr: start page need be migrated to sys ram
+ * @last_mgr: last page need be migrated to sys ram
+ * @trigger: reason of migration
+ * @fault_page: is from vmf->page, svm_migrate_to_ram(), this is CPU page fault callback
+ *
+ * Context: Process context, caller hold mmap read lock, prange->migrate_mutex
+ *
+ * Return:
+ * 0 - OK, otherwise error code
+ */
+int svm_migrate_vram_to_ram(struct svm_range *prange, struct mm_struct *mm,
+			    unsigned long start_mgr, unsigned long last_mgr,
+			    uint32_t trigger, struct page *fault_page)
+{
+	struct kfd_node *node;
+	struct vm_area_struct *vma;
+	unsigned long addr;
+	unsigned long start;
+	unsigned long end;
+	unsigned long mpages = 0;
+	long r = 0;
+
+	/* this pragne has no any vram page to migrate to sys ram */
+	if (!prange->actual_loc) {
+		pr_debug("[0x%lx 0x%lx] already migrated to ram\n",
+			 prange->start, prange->last);
+		return 0;
+	}
+
+	if (start_mgr < prange->start || last_mgr > prange->last) {
+		pr_debug("range [0x%lx 0x%lx] out prange [0x%lx 0x%lx]\n",
+			 start_mgr, last_mgr, prange->start, prange->last);
+		return -EFAULT;
+	}
+
+	node = svm_range_get_node_by_id(prange, prange->actual_loc);
+	if (!node) {
+		pr_debug("failed to get kfd node by id 0x%x\n", prange->actual_loc);
+		return -ENODEV;
+	}
+	pr_debug("svms 0x%p prange 0x%p [0x%lx 0x%lx] from gpu 0x%x to ram\n",
+		 prange->svms, prange, start_mgr, last_mgr,
+		 prange->actual_loc);
+
+	start = start_mgr << PAGE_SHIFT;
+	end = (last_mgr + 1) << PAGE_SHIFT;
+
+	for (addr = start; addr < end;) {
+		unsigned long next;
+
+		vma = vma_lookup(mm, addr);
+		if (!vma) {
+			pr_debug("failed to find vma for prange %p\n", prange);
+			r = -EFAULT;
+			break;
+		}
+
+		next = min(vma->vm_end, end);
+		r = svm_migrate_vma_to_ram(node, prange, vma, addr, next, trigger,
+			fault_page);
+		if (r < 0) {
+			pr_debug("failed %ld to migrate prange %p\n", r, prange);
+			break;
+		} else {
+			mpages += r;
+		}
+		addr = next;
+	}
+
+	if (r >= 0) {
+		WARN_ONCE(prange->vram_pages < mpages,
+			  "Recorded vram pages(0x%llx) should not be less than migration pages(0x%lx).",
+			  prange->vram_pages, mpages);
+		prange->vram_pages -= mpages;
+
+		/* prange does not have vram page set its actual_loc to system
+		 * and drop its svm_bo ref
+		 */
+		if (prange->vram_pages == 0 && prange->ttm_res) {
+			prange->actual_loc = 0;
+			svm_range_vram_node_free(prange);
+		}
+	}
+
+	return r < 0 ? r : 0;
+}
+
+/**
+ * svm_migrate_vram_to_vram - migrate svm range from device to device
+ * @prange: range structure
+ * @best_loc: the device to migrate to
+ * @start: start page need be migrated to sys ram
+ * @last: last page need be migrated to sys ram
+ * @mm: process mm, use current->mm if NULL
+ * @trigger: reason of migration
+ *
+ * Context: Process context, caller hold mmap read lock, svms lock, prange lock
+ *
+ * migrate all vram pages in prange to sys ram, then migrate
+ * [start, last] pages from sys ram to gpu node best_loc.
+ *
+ * Return:
+ * 0 - OK, otherwise error code
+ */
+static int
+svm_migrate_vram_to_vram(struct svm_range *prange, uint32_t best_loc,
+			unsigned long start, unsigned long last,
+			struct mm_struct *mm, uint32_t trigger)
+{
+	int r, retries = 3;
+
+	/*
+	 * TODO: for both devices with PCIe large bar or on same xgmi hive, skip
+	 * system memory as migration bridge
+	 */
+
+	pr_debug("from gpu 0x%x to gpu 0x%x\n", prange->actual_loc, best_loc);
+
+	do {
+		r = svm_migrate_vram_to_ram(prange, mm, prange->start, prange->last,
+					    trigger, NULL);
+		if (r)
+			return r;
+	} while (prange->actual_loc && --retries);
+
+	if (prange->actual_loc)
+		return -EDEADLK;
+
+	return svm_migrate_ram_to_vram(prange, best_loc, start, last, mm, trigger);
+}
+
+int
+svm_migrate_to_vram(struct svm_range *prange, uint32_t best_loc,
+		    unsigned long start, unsigned long last,
+		    struct mm_struct *mm, uint32_t trigger)
+{
+	if  (!prange->actual_loc || prange->actual_loc == best_loc)
+		return svm_migrate_ram_to_vram(prange, best_loc, start, last,
+					       mm, trigger);
+
+	else
+		return svm_migrate_vram_to_vram(prange, best_loc, start, last,
+						mm, trigger);
+
+}
+
+/**
+ * svm_migrate_to_ram - CPU page fault handler
+ * @vmf: CPU vm fault vma, address
+ *
+ * Context: vm fault handler, caller holds the mmap read lock
+ *
+ * Return:
+ * 0 - OK
+ * VM_FAULT_SIGBUS - notice application to have SIGBUS page fault
+ */
+static vm_fault_t svm_migrate_to_ram(struct vm_fault *vmf)
+{
+	unsigned long start, last, size;
+	unsigned long addr = vmf->address;
+	struct svm_range_bo *svm_bo;
+	struct svm_range *prange;
+	struct kfd_process *p;
+	struct mm_struct *mm;
+	int r = 0;
+
+	svm_bo = vmf->page->zone_device_data;
+	if (!svm_bo) {
+		pr_debug("failed get device page at addr 0x%lx\n", addr);
+		return VM_FAULT_SIGBUS;
+	}
+	if (!mmget_not_zero(svm_bo->eviction_fence->mm)) {
+		pr_debug("addr 0x%lx of process mm is destroyed\n", addr);
+		return VM_FAULT_SIGBUS;
+	}
+
+	mm = svm_bo->eviction_fence->mm;
+	if (mm != vmf->vma->vm_mm)
+		pr_debug("addr 0x%lx is COW mapping in child process\n", addr);
+
+	p = kfd_lookup_process_by_mm(mm);
+	if (!p) {
+		pr_debug("failed find process at fault address 0x%lx\n", addr);
+		r = VM_FAULT_SIGBUS;
+		goto out_mmput;
+	}
+	if (READ_ONCE(p->svms.faulting_task) == current) {
+		pr_debug("skipping ram migration\n");
+		r = 0;
+		goto out_unref_process;
+	}
+
+	pr_debug("CPU page fault svms 0x%p address 0x%lx\n", &p->svms, addr);
+	addr >>= PAGE_SHIFT;
+
+	mutex_lock(&p->svms.lock);
+
+	prange = svm_range_from_addr(&p->svms, addr, NULL);
+	if (!prange) {
+		pr_debug("failed get range svms 0x%p addr 0x%lx\n", &p->svms, addr);
+		r = -EFAULT;
+		goto out_unlock_svms;
+	}
+
+	mutex_lock(&prange->migrate_mutex);
+
+	if (!prange->actual_loc)
+		goto out_unlock_prange;
+
+	/* Align migration range start and size to granularity size */
+	size = 1UL << prange->granularity;
+	start = max(ALIGN_DOWN(addr, size), prange->start);
+	last = min(ALIGN(addr + 1, size) - 1, prange->last);
+
+	r = svm_migrate_vram_to_ram(prange, vmf->vma->vm_mm, start, last,
+				    KFD_MIGRATE_TRIGGER_PAGEFAULT_CPU, vmf->page);
+	if (r)
+		pr_debug("failed %d migrate svms 0x%p range 0x%p [0x%lx 0x%lx]\n",
+			r, prange->svms, prange, start, last);
+
+out_unlock_prange:
+	mutex_unlock(&prange->migrate_mutex);
+out_unlock_svms:
+	mutex_unlock(&p->svms.lock);
+out_unref_process:
+	pr_debug("CPU fault svms 0x%p address 0x%lx done\n", &p->svms, addr);
+	kfd_unref_process(p);
+out_mmput:
+	mmput(mm);
+	return r ? VM_FAULT_SIGBUS : 0;
+}
+
+static const struct dev_pagemap_ops svm_migrate_pgmap_ops = {
+	.page_free		= svm_migrate_page_free,
+	.migrate_to_ram		= svm_migrate_to_ram,
+};
+
+/* Each VRAM page uses sizeof(struct page) on system memory */
+#define SVM_HMM_PAGE_STRUCT_SIZE(size) ((size)/PAGE_SIZE * sizeof(struct page))
+
+int kgd2kfd_init_zone_device(struct amdgpu_device *adev)
+{
+	struct amdgpu_kfd_dev *kfddev = &adev->kfd;
+	struct dev_pagemap *pgmap;
+	struct resource *res = NULL;
+	unsigned long size;
+	void *r;
+
+	/* Page migration works on gfx9 or newer */
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(9, 0, 1))
+		return -EINVAL;
+
+	if (adev->apu_prefer_gtt)
+		return 0;
+
+	pgmap = &kfddev->pgmap;
+	memset(pgmap, 0, sizeof(*pgmap));
+
+	/* TODO: register all vram to HMM for now.
+	 * should remove reserved size
+	 */
+	size = ALIGN(adev->gmc.real_vram_size, 2ULL << 20);
+	if (adev->gmc.xgmi.connected_to_cpu) {
+		pgmap->range.start = adev->gmc.aper_base;
+		pgmap->range.end = adev->gmc.aper_base + adev->gmc.aper_size - 1;
+		pgmap->type = MEMORY_DEVICE_COHERENT;
+	} else {
+		res = devm_request_free_mem_region(adev->dev, &iomem_resource, size);
+		if (IS_ERR(res))
+			return PTR_ERR(res);
+		pgmap->range.start = res->start;
+		pgmap->range.end = res->end;
+		pgmap->type = MEMORY_DEVICE_PRIVATE;
+	}
+
+	pgmap->nr_range = 1;
+	pgmap->ops = &svm_migrate_pgmap_ops;
+	pgmap->owner = SVM_ADEV_PGMAP_OWNER(adev);
+	pgmap->flags = 0;
+	/* Device manager releases device-specific resources, memory region and
+	 * pgmap when driver disconnects from device.
+	 */
+	r = devm_memremap_pages(adev->dev, pgmap);
+	if (IS_ERR(r)) {
+		pr_err("failed to register HMM device memory\n");
+		if (pgmap->type == MEMORY_DEVICE_PRIVATE)
+			devm_release_mem_region(adev->dev, res->start, resource_size(res));
+		/* Disable SVM support capability */
+		pgmap->type = 0;
+		return PTR_ERR(r);
+	}
+
+	pr_debug("reserve %ldMB system memory for VRAM pages struct\n",
+		 SVM_HMM_PAGE_STRUCT_SIZE(size) >> 20);
+
+	amdgpu_amdkfd_reserve_system_mem(SVM_HMM_PAGE_STRUCT_SIZE(size));
+
+	pr_info("HMM registered %ldMB device memory\n", size >> 20);
+
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.h b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.h
new file mode 100644
index 000000000000..2eebf67f9c2c
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.h
@@ -0,0 +1,56 @@
+/* SPDX-License-Identifier: GPL-2.0 OR MIT */
+/*
+ * Copyright 2020-2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef KFD_MIGRATE_H_
+#define KFD_MIGRATE_H_
+
+#if IS_ENABLED(CONFIG_HSA_AMD_SVM)
+
+#include <linux/rwsem.h>
+#include <linux/list.h>
+#include <linux/mutex.h>
+#include <linux/sched/mm.h>
+#include <linux/hmm.h>
+#include "kfd_priv.h"
+#include "kfd_svm.h"
+
+enum MIGRATION_COPY_DIR {
+	FROM_RAM_TO_VRAM = 0,
+	FROM_VRAM_TO_RAM
+};
+
+int svm_migrate_to_vram(struct svm_range *prange,  uint32_t best_loc,
+			unsigned long start, unsigned long last,
+			struct mm_struct *mm, uint32_t trigger);
+
+int svm_migrate_vram_to_ram(struct svm_range *prange, struct mm_struct *mm,
+			    unsigned long start, unsigned long last,
+			    uint32_t trigger, struct page *fault_page);
+
+unsigned long
+svm_migrate_addr_to_pfn(struct amdgpu_device *adev, unsigned long addr);
+
+#endif /* IS_ENABLED(CONFIG_HSA_AMD_SVM) */
+
+#endif /* KFD_MIGRATE_H_ */
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_module.c b/drivers/gpu/drm/amd/amdkfd/kfd_module.c
index 0d73bea22c45..33aa23450b3f 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_module.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_module.c
@@ -1,5 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -20,71 +21,12 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  */
 
-#include <linux/module.h>
 #include <linux/sched.h>
-#include <linux/moduleparam.h>
 #include <linux/device.h>
 #include "kfd_priv.h"
+#include "amdgpu_amdkfd.h"
 
-#define KFD_DRIVER_AUTHOR	"AMD Inc. and others"
-
-#define KFD_DRIVER_DESC		"Standalone HSA driver for AMD's GPUs"
-#define KFD_DRIVER_DATE		"20150421"
-#define KFD_DRIVER_MAJOR	0
-#define KFD_DRIVER_MINOR	7
-#define KFD_DRIVER_PATCHLEVEL	2
-
-static const struct kgd2kfd_calls kgd2kfd = {
-	.exit		= kgd2kfd_exit,
-	.probe		= kgd2kfd_probe,
-	.device_init	= kgd2kfd_device_init,
-	.device_exit	= kgd2kfd_device_exit,
-	.interrupt	= kgd2kfd_interrupt,
-	.suspend	= kgd2kfd_suspend,
-	.resume		= kgd2kfd_resume,
-};
-
-int sched_policy = KFD_SCHED_POLICY_HWS;
-module_param(sched_policy, int, 0444);
-MODULE_PARM_DESC(sched_policy,
-	"Scheduling policy (0 = HWS (Default), 1 = HWS without over-subscription, 2 = Non-HWS (Used for debugging only)");
-
-int max_num_of_queues_per_device = KFD_MAX_NUM_OF_QUEUES_PER_DEVICE_DEFAULT;
-module_param(max_num_of_queues_per_device, int, 0444);
-MODULE_PARM_DESC(max_num_of_queues_per_device,
-	"Maximum number of supported queues per device (1 = Minimum, 4096 = default)");
-
-int send_sigterm;
-module_param(send_sigterm, int, 0444);
-MODULE_PARM_DESC(send_sigterm,
-	"Send sigterm to HSA process on unhandled exception (0 = disable, 1 = enable)");
-
-static int amdkfd_init_completed;
-
-int kgd2kfd_init(unsigned int interface_version,
-		const struct kgd2kfd_calls **g2f)
-{
-	if (!amdkfd_init_completed)
-		return -EPROBE_DEFER;
-
-	/*
-	 * Only one interface version is supported,
-	 * no kfd/kgd version skew allowed.
-	 */
-	if (interface_version != KFD_INTERFACE_VERSION)
-		return -EINVAL;
-
-	*g2f = &kgd2kfd;
-
-	return 0;
-}
-EXPORT_SYMBOL(kgd2kfd_init);
-
-void kgd2kfd_exit(void)
-{
-}
-
-static int __init kfd_module_init(void)
+static int kfd_init(void)
 {
 	int err;
 
@@ -92,7 +34,7 @@ static int __init kfd_module_init(void)
 	if ((sched_policy < KFD_SCHED_POLICY_HWS) ||
 		(sched_policy > KFD_SCHED_POLICY_NO_HWS)) {
 		pr_err("sched_policy has invalid value\n");
-		return -1;
+		return -EINVAL;
 	}
 
 	/* Verify module parameters */
@@ -100,13 +42,9 @@ static int __init kfd_module_init(void)
 		(max_num_of_queues_per_device >
 			KFD_MAX_NUM_OF_QUEUES_PER_DEVICE)) {
 		pr_err("max_num_of_queues_per_device must be between 1 to KFD_MAX_NUM_OF_QUEUES_PER_DEVICE\n");
-		return -1;
+		return -EINVAL;
 	}
 
-	err = kfd_pasid_init();
-	if (err < 0)
-		return err;
-
 	err = kfd_chardev_init();
 	if (err < 0)
 		goto err_ioctl;
@@ -115,38 +53,44 @@ static int __init kfd_module_init(void)
 	if (err < 0)
 		goto err_topology;
 
-	kfd_process_create_wq();
+	err = kfd_process_create_wq();
+	if (err < 0)
+		goto err_create_wq;
 
-	amdkfd_init_completed = 1;
+	/* Ignore the return value, so that we can continue
+	 * to init the KFD, even if procfs isn't craated
+	 */
+	kfd_procfs_init();
 
-	dev_info(kfd_device, "Initialized module\n");
+	kfd_debugfs_init();
 
 	return 0;
 
+err_create_wq:
+	kfd_topology_shutdown();
 err_topology:
 	kfd_chardev_exit();
 err_ioctl:
-	kfd_pasid_exit();
+	pr_err("KFD is disabled due to module initialization failure\n");
 	return err;
 }
 
-static void __exit kfd_module_exit(void)
+static void kfd_exit(void)
 {
-	amdkfd_init_completed = 0;
-
+	kfd_cleanup_processes();
 	kfd_process_destroy_wq();
+	kfd_debugfs_fini();
+	kfd_procfs_shutdown();
 	kfd_topology_shutdown();
 	kfd_chardev_exit();
-	kfd_pasid_exit();
-	dev_info(kfd_device, "Removed module\n");
 }
 
-module_init(kfd_module_init);
-module_exit(kfd_module_exit);
+int kgd2kfd_init(void)
+{
+	return kfd_init();
+}
 
-MODULE_AUTHOR(KFD_DRIVER_AUTHOR);
-MODULE_DESCRIPTION(KFD_DRIVER_DESC);
-MODULE_LICENSE("GPL and additional rights");
-MODULE_VERSION(__stringify(KFD_DRIVER_MAJOR) "."
-	       __stringify(KFD_DRIVER_MINOR) "."
-	       __stringify(KFD_DRIVER_PATCHLEVEL));
+void kgd2kfd_exit(void)
+{
+	kfd_exit();
+}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.c
index b1ef1368c3bb..d9ae854b6908 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.c
@@ -1,5 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -21,17 +22,291 @@
  *
  */
 
-#include "kfd_priv.h"
+#include "kfd_mqd_manager.h"
+#include "amdgpu_amdkfd.h"
+#include "kfd_device_queue_manager.h"
 
-struct mqd_manager *mqd_manager_init(enum KFD_MQD_TYPE type,
-					struct kfd_dev *dev)
+/* Mapping queue priority to pipe priority, indexed by queue priority */
+int pipe_priority_map[] = {
+	KFD_PIPE_PRIORITY_CS_LOW,
+	KFD_PIPE_PRIORITY_CS_LOW,
+	KFD_PIPE_PRIORITY_CS_LOW,
+	KFD_PIPE_PRIORITY_CS_LOW,
+	KFD_PIPE_PRIORITY_CS_LOW,
+	KFD_PIPE_PRIORITY_CS_LOW,
+	KFD_PIPE_PRIORITY_CS_LOW,
+	KFD_PIPE_PRIORITY_CS_MEDIUM,
+	KFD_PIPE_PRIORITY_CS_MEDIUM,
+	KFD_PIPE_PRIORITY_CS_MEDIUM,
+	KFD_PIPE_PRIORITY_CS_MEDIUM,
+	KFD_PIPE_PRIORITY_CS_HIGH,
+	KFD_PIPE_PRIORITY_CS_HIGH,
+	KFD_PIPE_PRIORITY_CS_HIGH,
+	KFD_PIPE_PRIORITY_CS_HIGH,
+	KFD_PIPE_PRIORITY_CS_HIGH
+};
+
+struct kfd_mem_obj *allocate_hiq_mqd(struct kfd_node *dev, struct queue_properties *q)
+{
+	struct kfd_mem_obj *mqd_mem_obj;
+
+	mqd_mem_obj = kzalloc(sizeof(struct kfd_mem_obj), GFP_KERNEL);
+	if (!mqd_mem_obj)
+		return NULL;
+
+	mqd_mem_obj->gtt_mem = dev->dqm->hiq_sdma_mqd.gtt_mem;
+	mqd_mem_obj->gpu_addr = dev->dqm->hiq_sdma_mqd.gpu_addr;
+	mqd_mem_obj->cpu_ptr = dev->dqm->hiq_sdma_mqd.cpu_ptr;
+
+	return mqd_mem_obj;
+}
+
+struct kfd_mem_obj *allocate_sdma_mqd(struct kfd_node *dev,
+					struct queue_properties *q)
+{
+	struct kfd_mem_obj *mqd_mem_obj;
+	uint64_t offset;
+
+	mqd_mem_obj = kzalloc(sizeof(struct kfd_mem_obj), GFP_KERNEL);
+	if (!mqd_mem_obj)
+		return NULL;
+
+	offset = (q->sdma_engine_id *
+		dev->kfd->device_info.num_sdma_queues_per_engine +
+		q->sdma_queue_id) *
+		dev->dqm->mqd_mgrs[KFD_MQD_TYPE_SDMA]->mqd_size;
+
+	offset += dev->dqm->mqd_mgrs[KFD_MQD_TYPE_HIQ]->mqd_size *
+		  NUM_XCC(dev->xcc_mask);
+
+	mqd_mem_obj->gtt_mem = (void *)((uint64_t)dev->dqm->hiq_sdma_mqd.gtt_mem
+				+ offset);
+	mqd_mem_obj->gpu_addr = dev->dqm->hiq_sdma_mqd.gpu_addr + offset;
+	mqd_mem_obj->cpu_ptr = (uint32_t *)((uint64_t)
+				dev->dqm->hiq_sdma_mqd.cpu_ptr + offset);
+
+	return mqd_mem_obj;
+}
+
+void free_mqd_hiq_sdma(struct mqd_manager *mm, void *mqd,
+			struct kfd_mem_obj *mqd_mem_obj)
+{
+	WARN_ON(!mqd_mem_obj->gtt_mem);
+	kfree(mqd_mem_obj);
+}
+
+void mqd_symmetrically_map_cu_mask(struct mqd_manager *mm,
+		const uint32_t *cu_mask, uint32_t cu_mask_count,
+		uint32_t *se_mask, uint32_t inst)
+{
+	struct amdgpu_cu_info *cu_info = &mm->dev->adev->gfx.cu_info;
+	struct amdgpu_gfx_config *gfx_info = &mm->dev->adev->gfx.config;
+	uint32_t cu_per_sh[KFD_MAX_NUM_SE][KFD_MAX_NUM_SH_PER_SE] = {0};
+	bool wgp_mode_req = KFD_GC_VERSION(mm->dev) >= IP_VERSION(10, 0, 0);
+	uint32_t en_mask = wgp_mode_req ? 0x3 : 0x1;
+	int i, se, sh, cu, cu_bitmap_sh_mul, cu_inc = wgp_mode_req ? 2 : 1;
+	uint32_t cu_active_per_node;
+	int inc = cu_inc * NUM_XCC(mm->dev->xcc_mask);
+	int xcc_inst = inst + ffs(mm->dev->xcc_mask) - 1;
+
+	cu_active_per_node = cu_info->number / mm->dev->kfd->num_nodes;
+	if (cu_mask_count > cu_active_per_node)
+		cu_mask_count = cu_active_per_node;
+
+	/* Exceeding these bounds corrupts the stack and indicates a coding error.
+	 * Returning with no CU's enabled will hang the queue, which should be
+	 * attention grabbing.
+	 */
+	if (gfx_info->max_shader_engines > KFD_MAX_NUM_SE) {
+		dev_err(mm->dev->adev->dev,
+			"Exceeded KFD_MAX_NUM_SE, chip reports %d\n",
+			gfx_info->max_shader_engines);
+		return;
+	}
+	if (gfx_info->max_sh_per_se > KFD_MAX_NUM_SH_PER_SE) {
+		dev_err(mm->dev->adev->dev,
+			"Exceeded KFD_MAX_NUM_SH, chip reports %d\n",
+			gfx_info->max_sh_per_se * gfx_info->max_shader_engines);
+		return;
+	}
+
+	cu_bitmap_sh_mul = (KFD_GC_VERSION(mm->dev) >= IP_VERSION(11, 0, 0) &&
+			    KFD_GC_VERSION(mm->dev) < IP_VERSION(13, 0, 0)) ? 2 : 1;
+
+	/* Count active CUs per SH.
+	 *
+	 * Some CUs in an SH may be disabled.	HW expects disabled CUs to be
+	 * represented in the high bits of each SH's enable mask (the upper and lower
+	 * 16 bits of se_mask) and will take care of the actual distribution of
+	 * disabled CUs within each SH automatically.
+	 * Each half of se_mask must be filled only on bits 0-cu_per_sh[se][sh]-1.
+	 *
+	 * See note on Arcturus cu_bitmap layout in gfx_v9_0_get_cu_info.
+	 * See note on GFX11 cu_bitmap layout in gfx_v11_0_get_cu_info.
+	 */
+	for (se = 0; se < gfx_info->max_shader_engines; se++)
+		for (sh = 0; sh < gfx_info->max_sh_per_se; sh++)
+			cu_per_sh[se][sh] = hweight32(
+				cu_info->bitmap[xcc_inst][se % 4][sh + (se / 4) *
+				cu_bitmap_sh_mul]);
+
+	/* Symmetrically map cu_mask to all SEs & SHs:
+	 * se_mask programs up to 2 SH in the upper and lower 16 bits.
+	 *
+	 * Examples
+	 * Assuming 1 SH/SE, 4 SEs:
+	 * cu_mask[0] bit0 -> se_mask[0] bit0
+	 * cu_mask[0] bit1 -> se_mask[1] bit0
+	 * ...
+	 * cu_mask[0] bit4 -> se_mask[0] bit1
+	 * ...
+	 *
+	 * Assuming 2 SH/SE, 4 SEs
+	 * cu_mask[0] bit0 -> se_mask[0] bit0 (SE0,SH0,CU0)
+	 * cu_mask[0] bit1 -> se_mask[1] bit0 (SE1,SH0,CU0)
+	 * ...
+	 * cu_mask[0] bit4 -> se_mask[0] bit16 (SE0,SH1,CU0)
+	 * cu_mask[0] bit5 -> se_mask[1] bit16 (SE1,SH1,CU0)
+	 * ...
+	 * cu_mask[0] bit8 -> se_mask[0] bit1 (SE0,SH0,CU1)
+	 * ...
+	 *
+	 * For GFX 9.4.3, the following code only looks at a
+	 * subset of the cu_mask corresponding to the inst parameter.
+	 * If we have n XCCs under one GPU node
+	 * cu_mask[0] bit0 -> XCC0 se_mask[0] bit0 (XCC0,SE0,SH0,CU0)
+	 * cu_mask[0] bit1 -> XCC1 se_mask[0] bit0 (XCC1,SE0,SH0,CU0)
+	 * ..
+	 * cu_mask[0] bitn -> XCCn se_mask[0] bit0 (XCCn,SE0,SH0,CU0)
+	 * cu_mask[0] bit n+1 -> XCC0 se_mask[1] bit0 (XCC0,SE1,SH0,CU0)
+	 *
+	 * For example, if there are 6 XCCs under 1 KFD node, this code
+	 * running for each inst, will look at the bits as:
+	 * inst, inst + 6, inst + 12...
+	 *
+	 * First ensure all CUs are disabled, then enable user specified CUs.
+	 */
+	for (i = 0; i < gfx_info->max_shader_engines; i++)
+		se_mask[i] = 0;
+
+	i = inst;
+	for (cu = 0; cu < 16; cu += cu_inc) {
+		for (sh = 0; sh < gfx_info->max_sh_per_se; sh++) {
+			for (se = 0; se < gfx_info->max_shader_engines; se++) {
+				if (cu_per_sh[se][sh] > cu) {
+					if (cu_mask[i / 32] & (en_mask << (i % 32)))
+						se_mask[se] |= en_mask << (cu + sh * 16);
+					i += inc;
+					if (i >= cu_mask_count)
+						return;
+				}
+			}
+		}
+	}
+}
+
+int kfd_hiq_load_mqd_kiq(struct mqd_manager *mm, void *mqd,
+		     uint32_t pipe_id, uint32_t queue_id,
+		     struct queue_properties *p, struct mm_struct *mms)
+{
+	return mm->dev->kfd2kgd->hiq_mqd_load(mm->dev->adev, mqd, pipe_id,
+					      queue_id, p->doorbell_off, 0);
+}
+
+int kfd_destroy_mqd_cp(struct mqd_manager *mm, void *mqd,
+		enum kfd_preempt_type type, unsigned int timeout,
+		uint32_t pipe_id, uint32_t queue_id)
 {
-	switch (dev->device_info->asic_family) {
-	case CHIP_KAVERI:
-		return mqd_manager_init_cik(type, dev);
-	case CHIP_CARRIZO:
-		return mqd_manager_init_vi(type, dev);
+	return mm->dev->kfd2kgd->hqd_destroy(mm->dev->adev, mqd, type, timeout,
+						pipe_id, queue_id, 0);
+}
+
+void kfd_free_mqd_cp(struct mqd_manager *mm, void *mqd,
+	      struct kfd_mem_obj *mqd_mem_obj)
+{
+	if (mqd_mem_obj->gtt_mem) {
+		amdgpu_amdkfd_free_gtt_mem(mm->dev->adev, &mqd_mem_obj->gtt_mem);
+		kfree(mqd_mem_obj);
+	} else {
+		kfd_gtt_sa_free(mm->dev, mqd_mem_obj);
+	}
+}
+
+bool kfd_is_occupied_cp(struct mqd_manager *mm, void *mqd,
+		 uint64_t queue_address, uint32_t pipe_id,
+		 uint32_t queue_id)
+{
+	return mm->dev->kfd2kgd->hqd_is_occupied(mm->dev->adev, queue_address,
+						pipe_id, queue_id, 0);
+}
+
+int kfd_load_mqd_sdma(struct mqd_manager *mm, void *mqd,
+		  uint32_t pipe_id, uint32_t queue_id,
+		  struct queue_properties *p, struct mm_struct *mms)
+{
+	return mm->dev->kfd2kgd->hqd_sdma_load(mm->dev->adev, mqd,
+						(uint32_t __user *)p->write_ptr,
+						mms);
+}
+
+/*
+ * preempt type here is ignored because there is only one way
+ * to preempt sdma queue
+ */
+int kfd_destroy_mqd_sdma(struct mqd_manager *mm, void *mqd,
+		     enum kfd_preempt_type type,
+		     unsigned int timeout, uint32_t pipe_id,
+		     uint32_t queue_id)
+{
+	return mm->dev->kfd2kgd->hqd_sdma_destroy(mm->dev->adev, mqd, timeout);
+}
+
+bool kfd_is_occupied_sdma(struct mqd_manager *mm, void *mqd,
+		      uint64_t queue_address, uint32_t pipe_id,
+		      uint32_t queue_id)
+{
+	return mm->dev->kfd2kgd->hqd_sdma_is_occupied(mm->dev->adev, mqd);
+}
+
+uint64_t kfd_hiq_mqd_stride(struct kfd_node *dev)
+{
+	return dev->dqm->mqd_mgrs[KFD_MQD_TYPE_HIQ]->mqd_size;
+}
+
+void kfd_get_hiq_xcc_mqd(struct kfd_node *dev, struct kfd_mem_obj *mqd_mem_obj,
+		     uint32_t virtual_xcc_id)
+{
+	uint64_t offset;
+
+	offset = kfd_hiq_mqd_stride(dev) * virtual_xcc_id;
+
+	mqd_mem_obj->gtt_mem = (virtual_xcc_id == 0) ?
+			dev->dqm->hiq_sdma_mqd.gtt_mem : NULL;
+	mqd_mem_obj->gpu_addr = dev->dqm->hiq_sdma_mqd.gpu_addr + offset;
+	mqd_mem_obj->cpu_ptr = (uint32_t *)((uintptr_t)
+				dev->dqm->hiq_sdma_mqd.cpu_ptr + offset);
+}
+
+uint64_t kfd_mqd_stride(struct mqd_manager *mm,
+			struct queue_properties *q)
+{
+	return mm->mqd_size;
+}
+
+bool kfd_check_hiq_mqd_doorbell_id(struct kfd_node *node, uint32_t doorbell_id,
+				   uint32_t inst)
+{
+	if (doorbell_id) {
+		struct device *dev = node->adev->dev;
+
+		if (node->adev->xcp_mgr && node->adev->xcp_mgr->num_xcps > 0)
+			dev_err(dev, "XCC %d: Queue preemption failed for queue with doorbell_id: %x\n",
+							inst, doorbell_id);
+		else
+			dev_err(dev, "Queue preemption failed for queue with doorbell_id: %x\n",
+							doorbell_id);
+		return true;
 	}
 
-	return NULL;
+	return false;
 }
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.h b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.h
index 1f3a6ba7eed2..17cc1f25c8d0 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.h
@@ -1,5 +1,6 @@
+/* SPDX-License-Identifier: GPL-2.0 OR MIT */
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -26,6 +27,9 @@
 
 #include "kfd_priv.h"
 
+#define KFD_MAX_NUM_SE 8
+#define KFD_MAX_NUM_SH_PER_SE 2
+
 /**
  * struct mqd_manager
  *
@@ -39,10 +43,13 @@
  * @destroy_mqd: Destroys the HQD slot and by that preempt the relevant queue.
  * Used only for no cp scheduling.
  *
- * @uninit_mqd: Releases the mqd buffer from local gpu memory.
+ * @free_mqd: Releases the mqd buffer from local gpu memory.
  *
  * @is_occupied: Checks if the relevant HQD slot is occupied.
  *
+ * @get_wave_state: Retrieves context save state and optionally copies the
+ * control stack, if kept in the MQD, to the given userspace address.
+ *
  * @mqd_mutex: Mqd manager mutex.
  *
  * @dev: The kfd device structure coupled with this module.
@@ -59,10 +66,13 @@
  * per KFD_MQD_TYPE for each device.
  *
  */
-
+extern int pipe_priority_map[];
 struct mqd_manager {
-	int	(*init_mqd)(struct mqd_manager *mm, void **mqd,
-			struct kfd_mem_obj **mqd_mem_obj, uint64_t *gart_addr,
+	struct kfd_mem_obj*	(*allocate_mqd)(struct kfd_node *kfd,
+		struct queue_properties *q);
+
+	void	(*init_mqd)(struct mqd_manager *mm, void **mqd,
+			struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
 			struct queue_properties *q);
 
 	int	(*load_mqd)(struct mqd_manager *mm, void *mqd,
@@ -70,23 +80,124 @@ struct mqd_manager {
 				struct queue_properties *p,
 				struct mm_struct *mms);
 
-	int	(*update_mqd)(struct mqd_manager *mm, void *mqd,
-				struct queue_properties *q);
+	void	(*update_mqd)(struct mqd_manager *mm, void *mqd,
+				struct queue_properties *q,
+				struct mqd_update_info *minfo);
 
 	int	(*destroy_mqd)(struct mqd_manager *mm, void *mqd,
 				enum kfd_preempt_type type,
 				unsigned int timeout, uint32_t pipe_id,
 				uint32_t queue_id);
 
-	void	(*uninit_mqd)(struct mqd_manager *mm, void *mqd,
+	void	(*free_mqd)(struct mqd_manager *mm, void *mqd,
 				struct kfd_mem_obj *mqd_mem_obj);
 
 	bool	(*is_occupied)(struct mqd_manager *mm, void *mqd,
 				uint64_t queue_address,	uint32_t pipe_id,
 				uint32_t queue_id);
 
+	int	(*get_wave_state)(struct mqd_manager *mm, void *mqd,
+				  struct queue_properties *q,
+				  void __user *ctl_stack,
+				  u32 *ctl_stack_used_size,
+				  u32 *save_area_used_size);
+
+	void	(*get_checkpoint_info)(struct mqd_manager *mm, void *mqd, uint32_t *ctl_stack_size);
+
+	void	(*checkpoint_mqd)(struct mqd_manager *mm,
+				  void *mqd,
+				  void *mqd_dst,
+				  void *ctl_stack_dst);
+
+	void	(*restore_mqd)(struct mqd_manager *mm, void **mqd,
+				struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+				struct queue_properties *p,
+				const void *mqd_src,
+				const void *ctl_stack_src,
+				const u32 ctl_stack_size);
+
+#if defined(CONFIG_DEBUG_FS)
+	int	(*debugfs_show_mqd)(struct seq_file *m, void *data);
+#endif
+	bool (*check_preemption_failed)(struct mqd_manager *mm, void *mqd);
+	uint64_t (*mqd_stride)(struct mqd_manager *mm,
+				struct queue_properties *p);
+
 	struct mutex	mqd_mutex;
-	struct kfd_dev	*dev;
+	struct kfd_node	*dev;
+	uint32_t mqd_size;
+};
+
+struct mqd_user_context_save_area_header {
+	/* Byte offset from start of user context
+	 * save area to the last saved top (lowest
+	 * address) of control stack data. Must be
+	 * 4 byte aligned.
+	 */
+	uint32_t control_stack_offset;
+
+	/* Byte size of the last saved control stack
+	 * data. Must be 4 byte aligned.
+	 */
+	uint32_t control_stack_size;
+
+	/* Byte offset from start of user context save
+	 * area to the last saved base (lowest address)
+	 * of wave state data. Must be 4 byte aligned.
+	 */
+	uint32_t wave_state_offset;
+
+	/* Byte size of the last saved wave state data.
+	 * Must be 4 byte aligned.
+	 */
+	uint32_t wave_state_size;
 };
 
+struct kfd_mem_obj *allocate_hiq_mqd(struct kfd_node *dev,
+				struct queue_properties *q);
+
+struct kfd_mem_obj *allocate_sdma_mqd(struct kfd_node *dev,
+					struct queue_properties *q);
+void free_mqd_hiq_sdma(struct mqd_manager *mm, void *mqd,
+				struct kfd_mem_obj *mqd_mem_obj);
+
+void mqd_symmetrically_map_cu_mask(struct mqd_manager *mm,
+		const uint32_t *cu_mask, uint32_t cu_mask_count,
+		uint32_t *se_mask, uint32_t inst);
+
+int kfd_hiq_load_mqd_kiq(struct mqd_manager *mm, void *mqd,
+		uint32_t pipe_id, uint32_t queue_id,
+		struct queue_properties *p, struct mm_struct *mms);
+
+int kfd_destroy_mqd_cp(struct mqd_manager *mm, void *mqd,
+		enum kfd_preempt_type type, unsigned int timeout,
+		uint32_t pipe_id, uint32_t queue_id);
+
+void kfd_free_mqd_cp(struct mqd_manager *mm, void *mqd,
+		struct kfd_mem_obj *mqd_mem_obj);
+
+bool kfd_is_occupied_cp(struct mqd_manager *mm, void *mqd,
+		 uint64_t queue_address, uint32_t pipe_id,
+		 uint32_t queue_id);
+
+int kfd_load_mqd_sdma(struct mqd_manager *mm, void *mqd,
+		uint32_t pipe_id, uint32_t queue_id,
+		struct queue_properties *p, struct mm_struct *mms);
+
+int kfd_destroy_mqd_sdma(struct mqd_manager *mm, void *mqd,
+		enum kfd_preempt_type type, unsigned int timeout,
+		uint32_t pipe_id, uint32_t queue_id);
+
+bool kfd_is_occupied_sdma(struct mqd_manager *mm, void *mqd,
+		uint64_t queue_address, uint32_t pipe_id,
+		uint32_t queue_id);
+
+void kfd_get_hiq_xcc_mqd(struct kfd_node *dev,
+		struct kfd_mem_obj *mqd_mem_obj, uint32_t virtual_xcc_id);
+
+uint64_t kfd_hiq_mqd_stride(struct kfd_node *dev);
+uint64_t kfd_mqd_stride(struct mqd_manager *mm,
+			struct queue_properties *q);
+bool kfd_check_hiq_mqd_doorbell_id(struct kfd_node *node, uint32_t doorbell_id,
+				   uint32_t inst);
 #endif /* KFD_MQD_MANAGER_H_ */
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_cik.c b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_cik.c
index 44ffd23348fc..05f3ac2eaef9 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_cik.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_cik.c
@@ -1,5 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -36,22 +37,63 @@ static inline struct cik_mqd *get_mqd(void *mqd)
 	return (struct cik_mqd *)mqd;
 }
 
-static int init_mqd(struct mqd_manager *mm, void **mqd,
-		struct kfd_mem_obj **mqd_mem_obj, uint64_t *gart_addr,
-		struct queue_properties *q)
+static inline struct cik_sdma_rlc_registers *get_sdma_mqd(void *mqd)
+{
+	return (struct cik_sdma_rlc_registers *)mqd;
+}
+
+static void update_cu_mask(struct mqd_manager *mm, void *mqd,
+			struct mqd_update_info *minfo)
 {
-	uint64_t addr;
 	struct cik_mqd *m;
-	int retval;
+	uint32_t se_mask[4] = {0}; /* 4 is the max # of SEs */
+
+	if (!minfo || !minfo->cu_mask.ptr)
+		return;
+
+	mqd_symmetrically_map_cu_mask(mm,
+		minfo->cu_mask.ptr, minfo->cu_mask.count, se_mask, 0);
+
+	m = get_mqd(mqd);
+	m->compute_static_thread_mgmt_se0 = se_mask[0];
+	m->compute_static_thread_mgmt_se1 = se_mask[1];
+	m->compute_static_thread_mgmt_se2 = se_mask[2];
+	m->compute_static_thread_mgmt_se3 = se_mask[3];
+
+	pr_debug("Update cu mask to %#x %#x %#x %#x\n",
+		m->compute_static_thread_mgmt_se0,
+		m->compute_static_thread_mgmt_se1,
+		m->compute_static_thread_mgmt_se2,
+		m->compute_static_thread_mgmt_se3);
+}
+
+static void set_priority(struct cik_mqd *m, struct queue_properties *q)
+{
+	m->cp_hqd_pipe_priority = pipe_priority_map[q->priority];
+	m->cp_hqd_queue_priority = q->priority;
+}
+
+static struct kfd_mem_obj *allocate_mqd(struct kfd_node *kfd,
+					struct queue_properties *q)
+{
+	struct kfd_mem_obj *mqd_mem_obj;
+
+	if (kfd_gtt_sa_allocate(kfd, sizeof(struct cik_mqd),
+			&mqd_mem_obj))
+		return NULL;
 
-	retval = kfd_gtt_sa_allocate(mm->dev, sizeof(struct cik_mqd),
-					mqd_mem_obj);
+	return mqd_mem_obj;
+}
 
-	if (retval != 0)
-		return -ENOMEM;
+static void init_mqd(struct mqd_manager *mm, void **mqd,
+		struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+		struct queue_properties *q)
+{
+	uint64_t addr;
+	struct cik_mqd *m;
 
-	m = (struct cik_mqd *) (*mqd_mem_obj)->cpu_ptr;
-	addr = (*mqd_mem_obj)->gpu_addr;
+	m = (struct cik_mqd *) mqd_mem_obj->cpu_ptr;
+	addr = mqd_mem_obj->gpu_addr;
 
 	memset(m, 0, ALIGN(sizeof(struct cik_mqd), 256));
 
@@ -74,10 +116,6 @@ static int init_mqd(struct mqd_manager *mm, void **mqd,
 	m->cp_mqd_base_addr_lo        = lower_32_bits(addr);
 	m->cp_mqd_base_addr_hi        = upper_32_bits(addr);
 
-	m->cp_hqd_ib_control = DEFAULT_MIN_IB_AVAIL_SIZE | IB_ATC_EN;
-	/* Although WinKFD writes this, I suspect it should not be necessary */
-	m->cp_hqd_ib_control = IB_ATC_EN | DEFAULT_MIN_IB_AVAIL_SIZE;
-
 	m->cp_hqd_quantum = QUANTUM_EN | QUANTUM_SCALE_1MS |
 				QUANTUM_DURATION(10);
 
@@ -90,8 +128,7 @@ static int init_mqd(struct mqd_manager *mm, void **mqd,
 	 * 1 = CS_MEDIUM (typically between HP3D and GFX
 	 * 2 = CS_HIGH (typically above HP3D)
 	 */
-	m->cp_hqd_pipe_priority = 1;
-	m->cp_hqd_queue_priority = 15;
+	set_priority(m, q);
 
 	if (q->format == KFD_QUEUE_FORMAT_AQL)
 		m->cp_hqd_iq_rptr = AQL_ENABLE;
@@ -99,48 +136,24 @@ static int init_mqd(struct mqd_manager *mm, void **mqd,
 	*mqd = m;
 	if (gart_addr)
 		*gart_addr = addr;
-	retval = mm->update_mqd(mm, m, q);
-
-	return retval;
+	mm->update_mqd(mm, m, q, NULL);
 }
 
-static int init_mqd_sdma(struct mqd_manager *mm, void **mqd,
-			struct kfd_mem_obj **mqd_mem_obj, uint64_t *gart_addr,
+static void init_mqd_sdma(struct mqd_manager *mm, void **mqd,
+			struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
 			struct queue_properties *q)
 {
-	int retval;
 	struct cik_sdma_rlc_registers *m;
 
-	retval = kfd_gtt_sa_allocate(mm->dev,
-					sizeof(struct cik_sdma_rlc_registers),
-					mqd_mem_obj);
-
-	if (retval != 0)
-		return -ENOMEM;
-
-	m = (struct cik_sdma_rlc_registers *) (*mqd_mem_obj)->cpu_ptr;
+	m = (struct cik_sdma_rlc_registers *) mqd_mem_obj->cpu_ptr;
 
 	memset(m, 0, sizeof(struct cik_sdma_rlc_registers));
 
 	*mqd = m;
 	if (gart_addr)
-		*gart_addr = (*mqd_mem_obj)->gpu_addr;
+		*gart_addr = mqd_mem_obj->gpu_addr;
 
-	retval = mm->update_mqd(mm, m, q);
-
-	return retval;
-}
-
-static void uninit_mqd(struct mqd_manager *mm, void *mqd,
-			struct kfd_mem_obj *mqd_mem_obj)
-{
-	kfd_gtt_sa_free(mm->dev, mqd_mem_obj);
-}
-
-static void uninit_mqd_sdma(struct mqd_manager *mm, void *mqd,
-				struct kfd_mem_obj *mqd_mem_obj)
-{
-	kfd_gtt_sa_free(mm->dev, mqd_mem_obj);
+	mm->update_mqd(mm, m, q, NULL);
 }
 
 static int load_mqd(struct mqd_manager *mm, void *mqd, uint32_t pipe_id,
@@ -149,35 +162,33 @@ static int load_mqd(struct mqd_manager *mm, void *mqd, uint32_t pipe_id,
 {
 	/* AQL write pointer counts in 64B packets, PM4/CP counts in dwords. */
 	uint32_t wptr_shift = (p->format == KFD_QUEUE_FORMAT_AQL ? 4 : 0);
-	uint32_t wptr_mask = (uint32_t)((p->queue_size / sizeof(uint32_t)) - 1);
+	uint32_t wptr_mask = (uint32_t)((p->queue_size / 4) - 1);
 
-	return mm->dev->kfd2kgd->hqd_load(mm->dev->kgd, mqd, pipe_id, queue_id,
+	return mm->dev->kfd2kgd->hqd_load(mm->dev->adev, mqd, pipe_id, queue_id,
 					  (uint32_t __user *)p->write_ptr,
-					  wptr_shift, wptr_mask, mms);
+					  wptr_shift, wptr_mask, mms, 0);
 }
 
-static int load_mqd_sdma(struct mqd_manager *mm, void *mqd,
-			 uint32_t pipe_id, uint32_t queue_id,
-			 struct queue_properties *p, struct mm_struct *mms)
-{
-	return mm->dev->kfd2kgd->hqd_sdma_load(mm->dev->kgd, mqd);
-}
-
-static int update_mqd(struct mqd_manager *mm, void *mqd,
-			struct queue_properties *q)
+static void __update_mqd(struct mqd_manager *mm, void *mqd,
+			struct queue_properties *q, struct mqd_update_info *minfo,
+			unsigned int atc_bit)
 {
 	struct cik_mqd *m;
 
 	m = get_mqd(mqd);
 	m->cp_hqd_pq_control = DEFAULT_RPTR_BLOCK_SIZE |
-				DEFAULT_MIN_AVAIL_SIZE | PQ_ATC_EN;
+				DEFAULT_MIN_AVAIL_SIZE;
+	m->cp_hqd_ib_control = DEFAULT_MIN_IB_AVAIL_SIZE;
+	if (atc_bit) {
+		m->cp_hqd_pq_control |= PQ_ATC_EN;
+		m->cp_hqd_ib_control |= IB_ATC_EN;
+	}
 
 	/*
 	 * Calculating queue size which is log base 2 of actual queue size -1
 	 * dwords and another -1 for ffs
 	 */
-	m->cp_hqd_pq_control |= ffs(q->queue_size / sizeof(unsigned int))
-								- 1 - 1;
+	m->cp_hqd_pq_control |= order_base_2(q->queue_size / 4) - 1;
 	m->cp_hqd_pq_base_lo = lower_32_bits((uint64_t)q->queue_address >> 8);
 	m->cp_hqd_pq_base_hi = upper_32_bits((uint64_t)q->queue_address >> 8);
 	m->cp_hqd_pq_rptr_report_addr_lo = lower_32_bits((uint64_t)q->read_ptr);
@@ -189,24 +200,35 @@ static int update_mqd(struct mqd_manager *mm, void *mqd,
 	if (q->format == KFD_QUEUE_FORMAT_AQL)
 		m->cp_hqd_pq_control |= NO_UPDATE_RPTR;
 
-	q->is_active = false;
-	if (q->queue_size > 0 &&
-			q->queue_address != 0 &&
-			q->queue_percent > 0) {
-		q->is_active = true;
-	}
+	update_cu_mask(mm, mqd, minfo);
+	set_priority(m, q);
 
-	return 0;
+	q->is_active = QUEUE_IS_ACTIVE(*q);
 }
 
-static int update_mqd_sdma(struct mqd_manager *mm, void *mqd,
-				struct queue_properties *q)
+static bool check_preemption_failed(struct mqd_manager *mm, void *mqd)
+{
+	struct cik_mqd *m = (struct cik_mqd *)mqd;
+
+	return kfd_check_hiq_mqd_doorbell_id(mm->dev, m->queue_doorbell_id0, 0);
+}
+
+static void update_mqd(struct mqd_manager *mm, void *mqd,
+		       struct queue_properties *q,
+		       struct mqd_update_info *minfo)
+{
+	__update_mqd(mm, mqd, q, minfo, 0);
+}
+
+static void update_mqd_sdma(struct mqd_manager *mm, void *mqd,
+			struct queue_properties *q,
+			struct mqd_update_info *minfo)
 {
 	struct cik_sdma_rlc_registers *m;
 
 	m = get_sdma_mqd(mqd);
-	m->sdma_rlc_rb_cntl = ffs(q->queue_size / sizeof(unsigned int)) <<
-			SDMA0_RLC0_RB_CNTL__RB_SIZE__SHIFT |
+	m->sdma_rlc_rb_cntl = order_base_2(q->queue_size / 4)
+			<< SDMA0_RLC0_RB_CNTL__RB_SIZE__SHIFT |
 			q->vmid << SDMA0_RLC0_RB_CNTL__RB_VMID__SHIFT |
 			1 << SDMA0_RLC0_RB_CNTL__RPTR_WRITEBACK_ENABLE__SHIFT |
 			6 << SDMA0_RLC0_RB_CNTL__RPTR_WRITEBACK_TIMER__SHIFT;
@@ -215,131 +237,104 @@ static int update_mqd_sdma(struct mqd_manager *mm, void *mqd,
 	m->sdma_rlc_rb_base_hi = upper_32_bits(q->queue_address >> 8);
 	m->sdma_rlc_rb_rptr_addr_lo = lower_32_bits((uint64_t)q->read_ptr);
 	m->sdma_rlc_rb_rptr_addr_hi = upper_32_bits((uint64_t)q->read_ptr);
-	m->sdma_rlc_doorbell = q->doorbell_off <<
-			SDMA0_RLC0_DOORBELL__OFFSET__SHIFT |
-			1 << SDMA0_RLC0_DOORBELL__ENABLE__SHIFT;
+	m->sdma_rlc_doorbell =
+		q->doorbell_off << SDMA0_RLC0_DOORBELL__OFFSET__SHIFT;
 
 	m->sdma_rlc_virtual_addr = q->sdma_vm_addr;
 
 	m->sdma_engine_id = q->sdma_engine_id;
 	m->sdma_queue_id = q->sdma_queue_id;
 
-	q->is_active = false;
-	if (q->queue_size > 0 &&
-			q->queue_address != 0 &&
-			q->queue_percent > 0) {
-		m->sdma_rlc_rb_cntl |=
-				1 << SDMA0_RLC0_RB_CNTL__RB_ENABLE__SHIFT;
-
-		q->is_active = true;
-	}
-
-	return 0;
+	q->is_active = QUEUE_IS_ACTIVE(*q);
 }
 
-static int destroy_mqd(struct mqd_manager *mm, void *mqd,
-			enum kfd_preempt_type type,
-			unsigned int timeout, uint32_t pipe_id,
-			uint32_t queue_id)
+static void checkpoint_mqd(struct mqd_manager *mm, void *mqd, void *mqd_dst, void *ctl_stack_dst)
 {
-	return mm->dev->kfd2kgd->hqd_destroy(mm->dev->kgd, mqd, type, timeout,
-					pipe_id, queue_id);
-}
+	struct cik_mqd *m;
 
-/*
- * preempt type here is ignored because there is only one way
- * to preempt sdma queue
- */
-static int destroy_mqd_sdma(struct mqd_manager *mm, void *mqd,
-				enum kfd_preempt_type type,
-				unsigned int timeout, uint32_t pipe_id,
-				uint32_t queue_id)
-{
-	return mm->dev->kfd2kgd->hqd_sdma_destroy(mm->dev->kgd, mqd, timeout);
+	m = get_mqd(mqd);
+
+	memcpy(mqd_dst, m, sizeof(struct cik_mqd));
 }
 
-static bool is_occupied(struct mqd_manager *mm, void *mqd,
-			uint64_t queue_address,	uint32_t pipe_id,
-			uint32_t queue_id)
+static void restore_mqd(struct mqd_manager *mm, void **mqd,
+			struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+			struct queue_properties *qp,
+			const void *mqd_src,
+			const void *ctl_stack_src, const u32 ctl_stack_size)
 {
+	uint64_t addr;
+	struct cik_mqd *m;
 
-	return mm->dev->kfd2kgd->hqd_is_occupied(mm->dev->kgd, queue_address,
-					pipe_id, queue_id);
-
-}
+	m = (struct cik_mqd *) mqd_mem_obj->cpu_ptr;
+	addr = mqd_mem_obj->gpu_addr;
 
-static bool is_occupied_sdma(struct mqd_manager *mm, void *mqd,
-			uint64_t queue_address,	uint32_t pipe_id,
-			uint32_t queue_id)
-{
-	return mm->dev->kfd2kgd->hqd_sdma_is_occupied(mm->dev->kgd, mqd);
-}
+	memcpy(m, mqd_src, sizeof(*m));
 
-/*
- * HIQ MQD Implementation, concrete implementation for HIQ MQD implementation.
- * The HIQ queue in Kaveri is using the same MQD structure as all the user mode
- * queues but with different initial values.
- */
+	*mqd = m;
+	if (gart_addr)
+		*gart_addr = addr;
 
-static int init_mqd_hiq(struct mqd_manager *mm, void **mqd,
-		struct kfd_mem_obj **mqd_mem_obj, uint64_t *gart_addr,
-		struct queue_properties *q)
-{
-	uint64_t addr;
-	struct cik_mqd *m;
-	int retval;
+	m->cp_hqd_pq_doorbell_control = DOORBELL_OFFSET(qp->doorbell_off);
 
-	retval = kfd_gtt_sa_allocate(mm->dev, sizeof(struct cik_mqd),
-					mqd_mem_obj);
+	pr_debug("cp_hqd_pq_doorbell_control 0x%x\n",
+			m->cp_hqd_pq_doorbell_control);
 
-	if (retval != 0)
-		return -ENOMEM;
+	qp->is_active = 0;
+}
 
-	m = (struct cik_mqd *) (*mqd_mem_obj)->cpu_ptr;
-	addr = (*mqd_mem_obj)->gpu_addr;
+static void checkpoint_mqd_sdma(struct mqd_manager *mm,
+				void *mqd,
+				void *mqd_dst,
+				void *ctl_stack_dst)
+{
+	struct cik_sdma_rlc_registers *m;
 
-	memset(m, 0, ALIGN(sizeof(struct cik_mqd), 256));
+	m = get_sdma_mqd(mqd);
 
-	m->header = 0xC0310800;
-	m->compute_pipelinestat_enable = 1;
-	m->compute_static_thread_mgmt_se0 = 0xFFFFFFFF;
-	m->compute_static_thread_mgmt_se1 = 0xFFFFFFFF;
-	m->compute_static_thread_mgmt_se2 = 0xFFFFFFFF;
-	m->compute_static_thread_mgmt_se3 = 0xFFFFFFFF;
+	memcpy(mqd_dst, m, sizeof(struct cik_sdma_rlc_registers));
+}
 
-	m->cp_hqd_persistent_state = DEFAULT_CP_HQD_PERSISTENT_STATE |
-					PRELOAD_REQ;
-	m->cp_hqd_quantum = QUANTUM_EN | QUANTUM_SCALE_1MS |
-				QUANTUM_DURATION(10);
+static void restore_mqd_sdma(struct mqd_manager *mm, void **mqd,
+				struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+				struct queue_properties *qp,
+				const void *mqd_src,
+				const void *ctl_stack_src, const u32 ctl_stack_size)
+{
+	uint64_t addr;
+	struct cik_sdma_rlc_registers *m;
 
-	m->cp_mqd_control             = MQD_CONTROL_PRIV_STATE_EN;
-	m->cp_mqd_base_addr_lo        = lower_32_bits(addr);
-	m->cp_mqd_base_addr_hi        = upper_32_bits(addr);
+	m = (struct cik_sdma_rlc_registers *) mqd_mem_obj->cpu_ptr;
+	addr = mqd_mem_obj->gpu_addr;
 
-	m->cp_hqd_ib_control = DEFAULT_MIN_IB_AVAIL_SIZE;
+	memcpy(m, mqd_src, sizeof(*m));
 
-	/*
-	 * Pipe Priority
-	 * Identifies the pipe relative priority when this queue is connected
-	 * to the pipeline. The pipe priority is against the GFX pipe and HP3D.
-	 * In KFD we are using a fixed pipe priority set to CS_MEDIUM.
-	 * 0 = CS_LOW (typically below GFX)
-	 * 1 = CS_MEDIUM (typically between HP3D and GFX
-	 * 2 = CS_HIGH (typically above HP3D)
-	 */
-	m->cp_hqd_pipe_priority = 1;
-	m->cp_hqd_queue_priority = 15;
+	m->sdma_rlc_doorbell =
+		qp->doorbell_off << SDMA0_RLC0_DOORBELL__OFFSET__SHIFT;
 
 	*mqd = m;
 	if (gart_addr)
 		*gart_addr = addr;
-	retval = mm->update_mqd(mm, m, q);
 
-	return retval;
+	qp->is_active = 0;
+}
+
+/*
+ * HIQ MQD Implementation, concrete implementation for HIQ MQD implementation.
+ * The HIQ queue in Kaveri is using the same MQD structure as all the user mode
+ * queues but with different initial values.
+ */
+
+static void init_mqd_hiq(struct mqd_manager *mm, void **mqd,
+		struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+		struct queue_properties *q)
+{
+	init_mqd(mm, mqd, mqd_mem_obj, gart_addr, q);
 }
 
-static int update_mqd_hiq(struct mqd_manager *mm, void *mqd,
-				struct queue_properties *q)
+static void update_mqd_hiq(struct mqd_manager *mm, void *mqd,
+			struct queue_properties *q,
+			struct mqd_update_info *minfo)
 {
 	struct cik_mqd *m;
 
@@ -353,40 +348,40 @@ static int update_mqd_hiq(struct mqd_manager *mm, void *mqd,
 	 * Calculating queue size which is log base 2 of actual queue
 	 * size -1 dwords
 	 */
-	m->cp_hqd_pq_control |= ffs(q->queue_size / sizeof(unsigned int))
-								- 1 - 1;
+	m->cp_hqd_pq_control |= order_base_2(q->queue_size / 4) - 1;
 	m->cp_hqd_pq_base_lo = lower_32_bits((uint64_t)q->queue_address >> 8);
 	m->cp_hqd_pq_base_hi = upper_32_bits((uint64_t)q->queue_address >> 8);
 	m->cp_hqd_pq_rptr_report_addr_lo = lower_32_bits((uint64_t)q->read_ptr);
 	m->cp_hqd_pq_rptr_report_addr_hi = upper_32_bits((uint64_t)q->read_ptr);
-	m->cp_hqd_pq_doorbell_control = DOORBELL_EN |
-					DOORBELL_OFFSET(q->doorbell_off);
+	m->cp_hqd_pq_doorbell_control = DOORBELL_OFFSET(q->doorbell_off);
 
 	m->cp_hqd_vmid = q->vmid;
 
-	m->cp_hqd_active = 0;
-	q->is_active = false;
-	if (q->queue_size > 0 &&
-			q->queue_address != 0 &&
-			q->queue_percent > 0) {
-		m->cp_hqd_active = 1;
-		q->is_active = true;
-	}
+	q->is_active = QUEUE_IS_ACTIVE(*q);
 
-	return 0;
+	set_priority(m, q);
 }
 
-struct cik_sdma_rlc_registers *get_sdma_mqd(void *mqd)
-{
-	struct cik_sdma_rlc_registers *m;
+#if defined(CONFIG_DEBUG_FS)
 
-	m = (struct cik_sdma_rlc_registers *)mqd;
+static int debugfs_show_mqd(struct seq_file *m, void *data)
+{
+	seq_hex_dump(m, "    ", DUMP_PREFIX_OFFSET, 32, 4,
+		     data, sizeof(struct cik_mqd), false);
+	return 0;
+}
 
-	return m;
+static int debugfs_show_mqd_sdma(struct seq_file *m, void *data)
+{
+	seq_hex_dump(m, "    ", DUMP_PREFIX_OFFSET, 32, 4,
+		     data, sizeof(struct cik_sdma_rlc_registers), false);
+	return 0;
 }
 
+#endif
+
 struct mqd_manager *mqd_manager_init_cik(enum KFD_MQD_TYPE type,
-		struct kfd_dev *dev)
+		struct kfd_node *dev)
 {
 	struct mqd_manager *mqd;
 
@@ -401,29 +396,64 @@ struct mqd_manager *mqd_manager_init_cik(enum KFD_MQD_TYPE type,
 
 	switch (type) {
 	case KFD_MQD_TYPE_CP:
-	case KFD_MQD_TYPE_COMPUTE:
+		mqd->allocate_mqd = allocate_mqd;
 		mqd->init_mqd = init_mqd;
-		mqd->uninit_mqd = uninit_mqd;
+		mqd->free_mqd = kfd_free_mqd_cp;
 		mqd->load_mqd = load_mqd;
 		mqd->update_mqd = update_mqd;
-		mqd->destroy_mqd = destroy_mqd;
-		mqd->is_occupied = is_occupied;
+		mqd->destroy_mqd = kfd_destroy_mqd_cp;
+		mqd->is_occupied = kfd_is_occupied_cp;
+		mqd->checkpoint_mqd = checkpoint_mqd;
+		mqd->restore_mqd = restore_mqd;
+		mqd->mqd_size = sizeof(struct cik_mqd);
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd;
+#endif
 		break;
 	case KFD_MQD_TYPE_HIQ:
+		mqd->allocate_mqd = allocate_hiq_mqd;
+		mqd->init_mqd = init_mqd_hiq;
+		mqd->free_mqd = free_mqd_hiq_sdma;
+		mqd->load_mqd = load_mqd;
+		mqd->update_mqd = update_mqd_hiq;
+		mqd->destroy_mqd = kfd_destroy_mqd_cp;
+		mqd->is_occupied = kfd_is_occupied_cp;
+		mqd->mqd_size = sizeof(struct cik_mqd);
+		mqd->mqd_stride = kfd_mqd_stride;
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd;
+#endif
+		mqd->check_preemption_failed = check_preemption_failed;
+		break;
+	case KFD_MQD_TYPE_DIQ:
+		mqd->allocate_mqd = allocate_mqd;
 		mqd->init_mqd = init_mqd_hiq;
-		mqd->uninit_mqd = uninit_mqd;
+		mqd->free_mqd = kfd_free_mqd_cp;
 		mqd->load_mqd = load_mqd;
 		mqd->update_mqd = update_mqd_hiq;
-		mqd->destroy_mqd = destroy_mqd;
-		mqd->is_occupied = is_occupied;
+		mqd->destroy_mqd = kfd_destroy_mqd_cp;
+		mqd->is_occupied = kfd_is_occupied_cp;
+		mqd->mqd_size = sizeof(struct cik_mqd);
+		mqd->mqd_stride = kfd_mqd_stride;
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd;
+#endif
 		break;
 	case KFD_MQD_TYPE_SDMA:
+		mqd->allocate_mqd = allocate_sdma_mqd;
 		mqd->init_mqd = init_mqd_sdma;
-		mqd->uninit_mqd = uninit_mqd_sdma;
-		mqd->load_mqd = load_mqd_sdma;
+		mqd->free_mqd = free_mqd_hiq_sdma;
+		mqd->load_mqd = kfd_load_mqd_sdma;
 		mqd->update_mqd = update_mqd_sdma;
-		mqd->destroy_mqd = destroy_mqd_sdma;
-		mqd->is_occupied = is_occupied_sdma;
+		mqd->destroy_mqd = kfd_destroy_mqd_sdma;
+		mqd->is_occupied = kfd_is_occupied_sdma;
+		mqd->checkpoint_mqd = checkpoint_mqd_sdma;
+		mqd->restore_mqd = restore_mqd_sdma;
+		mqd->mqd_size = sizeof(struct cik_sdma_rlc_registers);
+		mqd->mqd_stride = kfd_mqd_stride;
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd_sdma;
+#endif
 		break;
 	default:
 		kfree(mqd);
@@ -432,4 +462,3 @@ struct mqd_manager *mqd_manager_init_cik(enum KFD_MQD_TYPE type,
 
 	return mqd;
 }
-
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v10.c b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v10.c
new file mode 100644
index 000000000000..1695dd78ede8
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v10.c
@@ -0,0 +1,533 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
+/*
+ * Copyright 2018-2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/printk.h>
+#include <linux/slab.h>
+#include <linux/uaccess.h>
+#include "kfd_priv.h"
+#include "kfd_mqd_manager.h"
+#include "v10_structs.h"
+#include "gc/gc_10_1_0_offset.h"
+#include "gc/gc_10_1_0_sh_mask.h"
+#include "amdgpu_amdkfd.h"
+
+static inline struct v10_compute_mqd *get_mqd(void *mqd)
+{
+	return (struct v10_compute_mqd *)mqd;
+}
+
+static inline struct v10_sdma_mqd *get_sdma_mqd(void *mqd)
+{
+	return (struct v10_sdma_mqd *)mqd;
+}
+
+static void update_cu_mask(struct mqd_manager *mm, void *mqd,
+			struct mqd_update_info *minfo)
+{
+	struct v10_compute_mqd *m;
+	uint32_t se_mask[4] = {0}; /* 4 is the max # of SEs */
+
+	if (!minfo || !minfo->cu_mask.ptr)
+		return;
+
+	mqd_symmetrically_map_cu_mask(mm,
+		minfo->cu_mask.ptr, minfo->cu_mask.count, se_mask, 0);
+
+	m = get_mqd(mqd);
+	m->compute_static_thread_mgmt_se0 = se_mask[0];
+	m->compute_static_thread_mgmt_se1 = se_mask[1];
+	m->compute_static_thread_mgmt_se2 = se_mask[2];
+	m->compute_static_thread_mgmt_se3 = se_mask[3];
+
+	pr_debug("update cu mask to %#x %#x %#x %#x\n",
+		m->compute_static_thread_mgmt_se0,
+		m->compute_static_thread_mgmt_se1,
+		m->compute_static_thread_mgmt_se2,
+		m->compute_static_thread_mgmt_se3);
+}
+
+static void set_priority(struct v10_compute_mqd *m, struct queue_properties *q)
+{
+	m->cp_hqd_pipe_priority = pipe_priority_map[q->priority];
+	m->cp_hqd_queue_priority = q->priority;
+}
+
+static struct kfd_mem_obj *allocate_mqd(struct kfd_node *kfd,
+		struct queue_properties *q)
+{
+	struct kfd_mem_obj *mqd_mem_obj;
+
+	if (kfd_gtt_sa_allocate(kfd, sizeof(struct v10_compute_mqd),
+			&mqd_mem_obj))
+		return NULL;
+
+	return mqd_mem_obj;
+}
+
+static void init_mqd(struct mqd_manager *mm, void **mqd,
+			struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+			struct queue_properties *q)
+{
+	uint64_t addr;
+	struct v10_compute_mqd *m;
+
+	m = (struct v10_compute_mqd *) mqd_mem_obj->cpu_ptr;
+	addr = mqd_mem_obj->gpu_addr;
+
+	memset(m, 0, sizeof(struct v10_compute_mqd));
+
+	m->header = 0xC0310800;
+	m->compute_pipelinestat_enable = 1;
+	m->compute_static_thread_mgmt_se0 = 0xFFFFFFFF;
+	m->compute_static_thread_mgmt_se1 = 0xFFFFFFFF;
+	m->compute_static_thread_mgmt_se2 = 0xFFFFFFFF;
+	m->compute_static_thread_mgmt_se3 = 0xFFFFFFFF;
+
+	m->cp_hqd_persistent_state = CP_HQD_PERSISTENT_STATE__PRELOAD_REQ_MASK |
+			0x53 << CP_HQD_PERSISTENT_STATE__PRELOAD_SIZE__SHIFT;
+
+	m->cp_hqd_pq_control = 5 << CP_HQD_PQ_CONTROL__RPTR_BLOCK_SIZE__SHIFT;
+	m->cp_hqd_pq_control |= CP_HQD_PQ_CONTROL__UNORD_DISPATCH_MASK;
+	m->cp_mqd_control = 1 << CP_MQD_CONTROL__PRIV_STATE__SHIFT;
+
+	m->cp_mqd_base_addr_lo        = lower_32_bits(addr);
+	m->cp_mqd_base_addr_hi        = upper_32_bits(addr);
+
+	m->cp_hqd_quantum = 1 << CP_HQD_QUANTUM__QUANTUM_EN__SHIFT |
+			1 << CP_HQD_QUANTUM__QUANTUM_SCALE__SHIFT |
+			1 << CP_HQD_QUANTUM__QUANTUM_DURATION__SHIFT;
+
+	/* Set cp_hqd_hq_scheduler0 bit 14 to 1 to have the CP set up the
+	 * DISPATCH_PTR.  This is required for the kfd debugger
+	 */
+	m->cp_hqd_hq_scheduler0 = 1 << 14;
+
+	if (q->format == KFD_QUEUE_FORMAT_AQL) {
+		m->cp_hqd_aql_control =
+			1 << CP_HQD_AQL_CONTROL__CONTROL0__SHIFT;
+	}
+
+	if (mm->dev->kfd->cwsr_enabled) {
+		m->cp_hqd_persistent_state |=
+			(1 << CP_HQD_PERSISTENT_STATE__QSWITCH_MODE__SHIFT);
+		m->cp_hqd_ctx_save_base_addr_lo =
+			lower_32_bits(q->ctx_save_restore_area_address);
+		m->cp_hqd_ctx_save_base_addr_hi =
+			upper_32_bits(q->ctx_save_restore_area_address);
+		m->cp_hqd_ctx_save_size = q->ctx_save_restore_area_size;
+		m->cp_hqd_cntl_stack_size = q->ctl_stack_size;
+		m->cp_hqd_cntl_stack_offset = q->ctl_stack_size;
+		m->cp_hqd_wg_state_offset = q->ctl_stack_size;
+	}
+
+	*mqd = m;
+	if (gart_addr)
+		*gart_addr = addr;
+	mm->update_mqd(mm, m, q, NULL);
+}
+
+static int load_mqd(struct mqd_manager *mm, void *mqd,
+			uint32_t pipe_id, uint32_t queue_id,
+			struct queue_properties *p, struct mm_struct *mms)
+{
+	int r = 0;
+	/* AQL write pointer counts in 64B packets, PM4/CP counts in dwords. */
+	uint32_t wptr_shift = (p->format == KFD_QUEUE_FORMAT_AQL ? 4 : 0);
+
+	r = mm->dev->kfd2kgd->hqd_load(mm->dev->adev, mqd, pipe_id, queue_id,
+					  (uint32_t __user *)p->write_ptr,
+					  wptr_shift, 0, mms, 0);
+	return r;
+}
+
+static void update_mqd(struct mqd_manager *mm, void *mqd,
+			struct queue_properties *q,
+			struct mqd_update_info *minfo)
+{
+	struct v10_compute_mqd *m;
+
+	m = get_mqd(mqd);
+
+	m->cp_hqd_pq_control &= ~CP_HQD_PQ_CONTROL__QUEUE_SIZE_MASK;
+	m->cp_hqd_pq_control |=
+			ffs(q->queue_size / sizeof(unsigned int)) - 1 - 1;
+
+	pr_debug("cp_hqd_pq_control 0x%x\n", m->cp_hqd_pq_control);
+
+	m->cp_hqd_pq_base_lo = lower_32_bits((uint64_t)q->queue_address >> 8);
+	m->cp_hqd_pq_base_hi = upper_32_bits((uint64_t)q->queue_address >> 8);
+
+	m->cp_hqd_pq_rptr_report_addr_lo = lower_32_bits((uint64_t)q->read_ptr);
+	m->cp_hqd_pq_rptr_report_addr_hi = upper_32_bits((uint64_t)q->read_ptr);
+	m->cp_hqd_pq_wptr_poll_addr_lo = lower_32_bits((uint64_t)q->write_ptr);
+	m->cp_hqd_pq_wptr_poll_addr_hi = upper_32_bits((uint64_t)q->write_ptr);
+
+	m->cp_hqd_pq_doorbell_control =
+		q->doorbell_off <<
+			CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_OFFSET__SHIFT;
+	pr_debug("cp_hqd_pq_doorbell_control 0x%x\n",
+			m->cp_hqd_pq_doorbell_control);
+
+	m->cp_hqd_ib_control = 3 << CP_HQD_IB_CONTROL__MIN_IB_AVAIL_SIZE__SHIFT;
+
+	/*
+	 * HW does not clamp this field correctly. Maximum EOP queue size
+	 * is constrained by per-SE EOP done signal count, which is 8-bit.
+	 * Limit is 0xFF EOP entries (= 0x7F8 dwords). CP will not submit
+	 * more than (EOP entry count - 1) so a queue size of 0x800 dwords
+	 * is safe, giving a maximum field value of 0xA.
+	 */
+	m->cp_hqd_eop_control = min(0xA,
+		ffs(q->eop_ring_buffer_size / sizeof(unsigned int)) - 1 - 1);
+	m->cp_hqd_eop_base_addr_lo =
+			lower_32_bits(q->eop_ring_buffer_address >> 8);
+	m->cp_hqd_eop_base_addr_hi =
+			upper_32_bits(q->eop_ring_buffer_address >> 8);
+
+	m->cp_hqd_iq_timer = 0;
+
+	m->cp_hqd_vmid = q->vmid;
+
+	if (q->format == KFD_QUEUE_FORMAT_AQL) {
+		/* GC 10 removed WPP_CLAMP from PQ Control */
+		m->cp_hqd_pq_control |= CP_HQD_PQ_CONTROL__NO_UPDATE_RPTR_MASK |
+				2 << CP_HQD_PQ_CONTROL__SLOT_BASED_WPTR__SHIFT |
+				1 << CP_HQD_PQ_CONTROL__QUEUE_FULL_EN__SHIFT;
+		m->cp_hqd_pq_doorbell_control |=
+			1 << CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_BIF_DROP__SHIFT;
+	}
+	if (mm->dev->kfd->cwsr_enabled)
+		m->cp_hqd_ctx_save_control = 0;
+
+	update_cu_mask(mm, mqd, minfo);
+	set_priority(m, q);
+
+	q->is_active = QUEUE_IS_ACTIVE(*q);
+}
+
+static bool check_preemption_failed(struct mqd_manager *mm, void *mqd)
+{
+	struct v10_compute_mqd *m = (struct v10_compute_mqd *)mqd;
+
+	return kfd_check_hiq_mqd_doorbell_id(mm->dev, m->queue_doorbell_id0, 0);
+}
+
+static int get_wave_state(struct mqd_manager *mm, void *mqd,
+			  struct queue_properties *q,
+			  void __user *ctl_stack,
+			  u32 *ctl_stack_used_size,
+			  u32 *save_area_used_size)
+{
+	struct v10_compute_mqd *m;
+	struct kfd_context_save_area_header header;
+
+	m = get_mqd(mqd);
+
+	/* Control stack is written backwards, while workgroup context data
+	 * is written forwards. Both starts from m->cp_hqd_cntl_stack_size.
+	 * Current position is at m->cp_hqd_cntl_stack_offset and
+	 * m->cp_hqd_wg_state_offset, respectively.
+	 */
+	*ctl_stack_used_size = m->cp_hqd_cntl_stack_size -
+		m->cp_hqd_cntl_stack_offset;
+	*save_area_used_size = m->cp_hqd_wg_state_offset -
+		m->cp_hqd_cntl_stack_size;
+
+	/* Control stack is not copied to user mode for GFXv10 because
+	 * it's part of the context save area that is already
+	 * accessible to user mode
+	 */
+
+	header.wave_state.control_stack_size = *ctl_stack_used_size;
+	header.wave_state.wave_state_size = *save_area_used_size;
+
+	header.wave_state.wave_state_offset = m->cp_hqd_wg_state_offset;
+	header.wave_state.control_stack_offset = m->cp_hqd_cntl_stack_offset;
+
+	if (copy_to_user(ctl_stack, &header, sizeof(header.wave_state)))
+		return -EFAULT;
+
+	return 0;
+}
+
+static void checkpoint_mqd(struct mqd_manager *mm, void *mqd, void *mqd_dst, void *ctl_stack_dst)
+{
+	struct v10_compute_mqd *m;
+
+	m = get_mqd(mqd);
+
+	memcpy(mqd_dst, m, sizeof(struct v10_compute_mqd));
+}
+
+static void restore_mqd(struct mqd_manager *mm, void **mqd,
+			struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+			struct queue_properties *qp,
+			const void *mqd_src,
+			const void *ctl_stack_src, const u32 ctl_stack_size)
+{
+	uint64_t addr;
+	struct v10_compute_mqd *m;
+
+	m = (struct v10_compute_mqd *) mqd_mem_obj->cpu_ptr;
+	addr = mqd_mem_obj->gpu_addr;
+
+	memcpy(m, mqd_src, sizeof(*m));
+
+	*mqd = m;
+	if (gart_addr)
+		*gart_addr = addr;
+
+	m->cp_hqd_pq_doorbell_control =
+		qp->doorbell_off <<
+			CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_OFFSET__SHIFT;
+	pr_debug("cp_hqd_pq_doorbell_control 0x%x\n",
+			m->cp_hqd_pq_doorbell_control);
+
+	qp->is_active = 0;
+}
+
+static void init_mqd_hiq(struct mqd_manager *mm, void **mqd,
+			struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+			struct queue_properties *q)
+{
+	struct v10_compute_mqd *m;
+
+	init_mqd(mm, mqd, mqd_mem_obj, gart_addr, q);
+
+	m = get_mqd(*mqd);
+
+	m->cp_hqd_pq_control |= 1 << CP_HQD_PQ_CONTROL__PRIV_STATE__SHIFT |
+			1 << CP_HQD_PQ_CONTROL__KMD_QUEUE__SHIFT;
+}
+
+static int destroy_hiq_mqd(struct mqd_manager *mm, void *mqd,
+			enum kfd_preempt_type type, unsigned int timeout,
+			uint32_t pipe_id, uint32_t queue_id)
+{
+	int err;
+	struct v10_compute_mqd *m;
+	u32 doorbell_off;
+
+	m = get_mqd(mqd);
+
+	doorbell_off = m->cp_hqd_pq_doorbell_control >>
+			CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_OFFSET__SHIFT;
+
+	err = amdgpu_amdkfd_unmap_hiq(mm->dev->adev, doorbell_off, 0);
+	if (err)
+		pr_debug("Destroy HIQ MQD failed: %d\n", err);
+
+	return err;
+}
+
+static void init_mqd_sdma(struct mqd_manager *mm, void **mqd,
+		struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+		struct queue_properties *q)
+{
+	struct v10_sdma_mqd *m;
+
+	m = (struct v10_sdma_mqd *) mqd_mem_obj->cpu_ptr;
+
+	memset(m, 0, sizeof(struct v10_sdma_mqd));
+
+	*mqd = m;
+	if (gart_addr)
+		*gart_addr = mqd_mem_obj->gpu_addr;
+
+	mm->update_mqd(mm, m, q, NULL);
+}
+
+#define SDMA_RLC_DUMMY_DEFAULT 0xf
+
+static void update_mqd_sdma(struct mqd_manager *mm, void *mqd,
+			struct queue_properties *q,
+			struct mqd_update_info *minfo)
+{
+	struct v10_sdma_mqd *m;
+
+	m = get_sdma_mqd(mqd);
+	m->sdmax_rlcx_rb_cntl = (ffs(q->queue_size / sizeof(unsigned int)) - 1)
+		<< SDMA0_RLC0_RB_CNTL__RB_SIZE__SHIFT |
+		q->vmid << SDMA0_RLC0_RB_CNTL__RB_VMID__SHIFT |
+		1 << SDMA0_RLC0_RB_CNTL__RPTR_WRITEBACK_ENABLE__SHIFT |
+		6 << SDMA0_RLC0_RB_CNTL__RPTR_WRITEBACK_TIMER__SHIFT;
+
+	m->sdmax_rlcx_rb_base = lower_32_bits(q->queue_address >> 8);
+	m->sdmax_rlcx_rb_base_hi = upper_32_bits(q->queue_address >> 8);
+	m->sdmax_rlcx_rb_rptr_addr_lo = lower_32_bits((uint64_t)q->read_ptr);
+	m->sdmax_rlcx_rb_rptr_addr_hi = upper_32_bits((uint64_t)q->read_ptr);
+	m->sdmax_rlcx_doorbell_offset =
+		q->doorbell_off << SDMA0_RLC0_DOORBELL_OFFSET__OFFSET__SHIFT;
+
+	m->sdma_engine_id = q->sdma_engine_id;
+	m->sdma_queue_id = q->sdma_queue_id;
+	m->sdmax_rlcx_dummy_reg = SDMA_RLC_DUMMY_DEFAULT;
+
+	q->is_active = QUEUE_IS_ACTIVE(*q);
+}
+
+static void checkpoint_mqd_sdma(struct mqd_manager *mm,
+				void *mqd,
+				void *mqd_dst,
+				void *ctl_stack_dst)
+{
+	struct v10_sdma_mqd *m;
+
+	m = get_sdma_mqd(mqd);
+
+	memcpy(mqd_dst, m, sizeof(struct v10_sdma_mqd));
+}
+
+static void restore_mqd_sdma(struct mqd_manager *mm, void **mqd,
+			     struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+			     struct queue_properties *qp,
+			     const void *mqd_src,
+			     const void *ctl_stack_src,
+			     const u32 ctl_stack_size)
+{
+	uint64_t addr;
+	struct v10_sdma_mqd *m;
+
+	m = (struct v10_sdma_mqd *) mqd_mem_obj->cpu_ptr;
+	addr = mqd_mem_obj->gpu_addr;
+
+	memcpy(m, mqd_src, sizeof(*m));
+
+	m->sdmax_rlcx_doorbell_offset =
+		qp->doorbell_off << SDMA0_RLC0_DOORBELL_OFFSET__OFFSET__SHIFT;
+
+	*mqd = m;
+	if (gart_addr)
+		*gart_addr = addr;
+
+	qp->is_active = 0;
+}
+
+#if defined(CONFIG_DEBUG_FS)
+
+static int debugfs_show_mqd(struct seq_file *m, void *data)
+{
+	seq_hex_dump(m, "    ", DUMP_PREFIX_OFFSET, 32, 4,
+		     data, sizeof(struct v10_compute_mqd), false);
+	return 0;
+}
+
+static int debugfs_show_mqd_sdma(struct seq_file *m, void *data)
+{
+	seq_hex_dump(m, "    ", DUMP_PREFIX_OFFSET, 32, 4,
+		     data, sizeof(struct v10_sdma_mqd), false);
+	return 0;
+}
+
+#endif
+
+struct mqd_manager *mqd_manager_init_v10(enum KFD_MQD_TYPE type,
+		struct kfd_node *dev)
+{
+	struct mqd_manager *mqd;
+
+	if (WARN_ON(type >= KFD_MQD_TYPE_MAX))
+		return NULL;
+
+	mqd = kzalloc(sizeof(*mqd), GFP_KERNEL);
+	if (!mqd)
+		return NULL;
+
+	mqd->dev = dev;
+
+	switch (type) {
+	case KFD_MQD_TYPE_CP:
+		pr_debug("%s@%i\n", __func__, __LINE__);
+		mqd->allocate_mqd = allocate_mqd;
+		mqd->init_mqd = init_mqd;
+		mqd->free_mqd = kfd_free_mqd_cp;
+		mqd->load_mqd = load_mqd;
+		mqd->update_mqd = update_mqd;
+		mqd->destroy_mqd = kfd_destroy_mqd_cp;
+		mqd->is_occupied = kfd_is_occupied_cp;
+		mqd->mqd_size = sizeof(struct v10_compute_mqd);
+		mqd->get_wave_state = get_wave_state;
+		mqd->checkpoint_mqd = checkpoint_mqd;
+		mqd->restore_mqd = restore_mqd;
+		mqd->mqd_stride = kfd_mqd_stride;
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd;
+#endif
+		pr_debug("%s@%i\n", __func__, __LINE__);
+		break;
+	case KFD_MQD_TYPE_HIQ:
+		pr_debug("%s@%i\n", __func__, __LINE__);
+		mqd->allocate_mqd = allocate_hiq_mqd;
+		mqd->init_mqd = init_mqd_hiq;
+		mqd->free_mqd = free_mqd_hiq_sdma;
+		mqd->load_mqd = kfd_hiq_load_mqd_kiq;
+		mqd->update_mqd = update_mqd;
+		mqd->destroy_mqd = destroy_hiq_mqd;
+		mqd->is_occupied = kfd_is_occupied_cp;
+		mqd->mqd_size = sizeof(struct v10_compute_mqd);
+		mqd->mqd_stride = kfd_mqd_stride;
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd;
+#endif
+		mqd->check_preemption_failed = check_preemption_failed;
+		pr_debug("%s@%i\n", __func__, __LINE__);
+		break;
+	case KFD_MQD_TYPE_DIQ:
+		mqd->allocate_mqd = allocate_mqd;
+		mqd->init_mqd = init_mqd_hiq;
+		mqd->free_mqd = kfd_free_mqd_cp;
+		mqd->load_mqd = load_mqd;
+		mqd->update_mqd = update_mqd;
+		mqd->destroy_mqd = kfd_destroy_mqd_cp;
+		mqd->is_occupied = kfd_is_occupied_cp;
+		mqd->mqd_size = sizeof(struct v10_compute_mqd);
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd;
+#endif
+		break;
+	case KFD_MQD_TYPE_SDMA:
+		pr_debug("%s@%i\n", __func__, __LINE__);
+		mqd->allocate_mqd = allocate_sdma_mqd;
+		mqd->init_mqd = init_mqd_sdma;
+		mqd->free_mqd = free_mqd_hiq_sdma;
+		mqd->load_mqd = kfd_load_mqd_sdma;
+		mqd->update_mqd = update_mqd_sdma;
+		mqd->destroy_mqd = kfd_destroy_mqd_sdma;
+		mqd->is_occupied = kfd_is_occupied_sdma;
+		mqd->checkpoint_mqd = checkpoint_mqd_sdma;
+		mqd->restore_mqd = restore_mqd_sdma;
+		mqd->mqd_size = sizeof(struct v10_sdma_mqd);
+		mqd->mqd_stride = kfd_mqd_stride;
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd_sdma;
+#endif
+		pr_debug("%s@%i\n", __func__, __LINE__);
+		break;
+	default:
+		kfree(mqd);
+		return NULL;
+	}
+
+	return mqd;
+}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v11.c b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v11.c
new file mode 100644
index 000000000000..3c0ae28c5923
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v11.c
@@ -0,0 +1,569 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/printk.h>
+#include <linux/slab.h>
+#include <linux/uaccess.h>
+#include "kfd_priv.h"
+#include "kfd_mqd_manager.h"
+#include "v11_structs.h"
+#include "gc/gc_11_0_0_offset.h"
+#include "gc/gc_11_0_0_sh_mask.h"
+#include "amdgpu_amdkfd.h"
+
+static inline struct v11_compute_mqd *get_mqd(void *mqd)
+{
+	return (struct v11_compute_mqd *)mqd;
+}
+
+static inline struct v11_sdma_mqd *get_sdma_mqd(void *mqd)
+{
+	return (struct v11_sdma_mqd *)mqd;
+}
+
+static void update_cu_mask(struct mqd_manager *mm, void *mqd,
+			   struct mqd_update_info *minfo)
+{
+	struct v11_compute_mqd *m;
+	uint32_t se_mask[KFD_MAX_NUM_SE] = {0};
+	bool has_wa_flag = minfo && (minfo->update_flag & (UPDATE_FLAG_DBG_WA_ENABLE |
+			UPDATE_FLAG_DBG_WA_DISABLE));
+
+	if (!minfo || !(has_wa_flag || minfo->cu_mask.ptr))
+		return;
+
+	m = get_mqd(mqd);
+
+	if (has_wa_flag) {
+		uint32_t wa_mask =
+			(minfo->update_flag & UPDATE_FLAG_DBG_WA_ENABLE) ? 0xffff : 0xffffffff;
+
+		m->compute_static_thread_mgmt_se0 = wa_mask;
+		m->compute_static_thread_mgmt_se1 = wa_mask;
+		m->compute_static_thread_mgmt_se2 = wa_mask;
+		m->compute_static_thread_mgmt_se3 = wa_mask;
+		m->compute_static_thread_mgmt_se4 = wa_mask;
+		m->compute_static_thread_mgmt_se5 = wa_mask;
+		m->compute_static_thread_mgmt_se6 = wa_mask;
+		m->compute_static_thread_mgmt_se7 = wa_mask;
+
+		return;
+	}
+
+	mqd_symmetrically_map_cu_mask(mm,
+		minfo->cu_mask.ptr, minfo->cu_mask.count, se_mask, 0);
+
+	m->compute_static_thread_mgmt_se0 = se_mask[0];
+	m->compute_static_thread_mgmt_se1 = se_mask[1];
+	m->compute_static_thread_mgmt_se2 = se_mask[2];
+	m->compute_static_thread_mgmt_se3 = se_mask[3];
+	m->compute_static_thread_mgmt_se4 = se_mask[4];
+	m->compute_static_thread_mgmt_se5 = se_mask[5];
+	m->compute_static_thread_mgmt_se6 = se_mask[6];
+	m->compute_static_thread_mgmt_se7 = se_mask[7];
+
+	pr_debug("update cu mask to %#x %#x %#x %#x %#x %#x %#x %#x\n",
+		m->compute_static_thread_mgmt_se0,
+		m->compute_static_thread_mgmt_se1,
+		m->compute_static_thread_mgmt_se2,
+		m->compute_static_thread_mgmt_se3,
+		m->compute_static_thread_mgmt_se4,
+		m->compute_static_thread_mgmt_se5,
+		m->compute_static_thread_mgmt_se6,
+		m->compute_static_thread_mgmt_se7);
+}
+
+static void set_priority(struct v11_compute_mqd *m, struct queue_properties *q)
+{
+	m->cp_hqd_pipe_priority = pipe_priority_map[q->priority];
+	m->cp_hqd_queue_priority = q->priority;
+}
+
+static struct kfd_mem_obj *allocate_mqd(struct kfd_node *node,
+		struct queue_properties *q)
+{
+	struct kfd_mem_obj *mqd_mem_obj;
+	int size;
+
+	/*
+	 * MES write to areas beyond MQD size. So allocate
+	 * 1 PAGE_SIZE memory for MQD is MES is enabled.
+	 */
+	if (node->kfd->shared_resources.enable_mes)
+		size = PAGE_SIZE;
+	else
+		size = sizeof(struct v11_compute_mqd);
+
+	if (kfd_gtt_sa_allocate(node, size, &mqd_mem_obj))
+		return NULL;
+
+	return mqd_mem_obj;
+}
+
+static void init_mqd(struct mqd_manager *mm, void **mqd,
+			struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+			struct queue_properties *q)
+{
+	uint64_t addr;
+	struct v11_compute_mqd *m;
+	int size;
+	uint32_t wa_mask = q->is_dbg_wa ? 0xffff : 0xffffffff;
+
+	m = (struct v11_compute_mqd *) mqd_mem_obj->cpu_ptr;
+	addr = mqd_mem_obj->gpu_addr;
+
+	if (mm->dev->kfd->shared_resources.enable_mes)
+		size = PAGE_SIZE;
+	else
+		size = sizeof(struct v11_compute_mqd);
+
+	memset(m, 0, size);
+
+	m->header = 0xC0310800;
+	m->compute_pipelinestat_enable = 1;
+
+	m->compute_static_thread_mgmt_se0 = wa_mask;
+	m->compute_static_thread_mgmt_se1 = wa_mask;
+	m->compute_static_thread_mgmt_se2 = wa_mask;
+	m->compute_static_thread_mgmt_se3 = wa_mask;
+	m->compute_static_thread_mgmt_se4 = wa_mask;
+	m->compute_static_thread_mgmt_se5 = wa_mask;
+	m->compute_static_thread_mgmt_se6 = wa_mask;
+	m->compute_static_thread_mgmt_se7 = wa_mask;
+
+	m->cp_hqd_persistent_state = CP_HQD_PERSISTENT_STATE__PRELOAD_REQ_MASK |
+			0x55 << CP_HQD_PERSISTENT_STATE__PRELOAD_SIZE__SHIFT;
+
+	m->cp_hqd_pq_control = 5 << CP_HQD_PQ_CONTROL__RPTR_BLOCK_SIZE__SHIFT;
+	m->cp_hqd_pq_control |= CP_HQD_PQ_CONTROL__UNORD_DISPATCH_MASK;
+	m->cp_mqd_control = 1 << CP_MQD_CONTROL__PRIV_STATE__SHIFT;
+
+	m->cp_mqd_base_addr_lo        = lower_32_bits(addr);
+	m->cp_mqd_base_addr_hi        = upper_32_bits(addr);
+
+	m->cp_hqd_quantum = 1 << CP_HQD_QUANTUM__QUANTUM_EN__SHIFT |
+			1 << CP_HQD_QUANTUM__QUANTUM_SCALE__SHIFT |
+			1 << CP_HQD_QUANTUM__QUANTUM_DURATION__SHIFT;
+
+	/* Set cp_hqd_hq_scheduler0 bit 14 to 1 to have the CP set up the
+	 * DISPATCH_PTR.  This is required for the kfd debugger
+	 */
+	m->cp_hqd_hq_status0 = 1 << 14;
+
+	/*
+	 * GFX11 RS64 CPFW version >= 509 supports PCIe atomics support
+	 * acknowledgment.
+	 */
+	if (amdgpu_amdkfd_have_atomics_support(mm->dev->adev))
+		m->cp_hqd_hq_status0 |= 1 << 29;
+
+	if (q->format == KFD_QUEUE_FORMAT_AQL) {
+		m->cp_hqd_aql_control =
+			1 << CP_HQD_AQL_CONTROL__CONTROL0__SHIFT;
+	}
+
+	if (mm->dev->kfd->cwsr_enabled) {
+		m->cp_hqd_persistent_state |=
+			(1 << CP_HQD_PERSISTENT_STATE__QSWITCH_MODE__SHIFT);
+		m->cp_hqd_ctx_save_base_addr_lo =
+			lower_32_bits(q->ctx_save_restore_area_address);
+		m->cp_hqd_ctx_save_base_addr_hi =
+			upper_32_bits(q->ctx_save_restore_area_address);
+		m->cp_hqd_ctx_save_size = q->ctx_save_restore_area_size;
+		m->cp_hqd_cntl_stack_size = q->ctl_stack_size;
+		m->cp_hqd_cntl_stack_offset = q->ctl_stack_size;
+		m->cp_hqd_wg_state_offset = q->ctl_stack_size;
+	}
+
+	*mqd = m;
+	if (gart_addr)
+		*gart_addr = addr;
+	mm->update_mqd(mm, m, q, NULL);
+}
+
+static int load_mqd(struct mqd_manager *mm, void *mqd,
+			uint32_t pipe_id, uint32_t queue_id,
+			struct queue_properties *p, struct mm_struct *mms)
+{
+	int r = 0;
+	/* AQL write pointer counts in 64B packets, PM4/CP counts in dwords. */
+	uint32_t wptr_shift = (p->format == KFD_QUEUE_FORMAT_AQL ? 4 : 0);
+
+	r = mm->dev->kfd2kgd->hqd_load(mm->dev->adev, mqd, pipe_id, queue_id,
+					  (uint32_t __user *)p->write_ptr,
+					  wptr_shift, 0, mms, 0);
+	return r;
+}
+
+static void update_mqd(struct mqd_manager *mm, void *mqd,
+		       struct queue_properties *q,
+		       struct mqd_update_info *minfo)
+{
+	struct v11_compute_mqd *m;
+
+	m = get_mqd(mqd);
+
+	m->cp_hqd_pq_control &= ~CP_HQD_PQ_CONTROL__QUEUE_SIZE_MASK;
+	m->cp_hqd_pq_control |=
+			ffs(q->queue_size / sizeof(unsigned int)) - 1 - 1;
+	pr_debug("cp_hqd_pq_control 0x%x\n", m->cp_hqd_pq_control);
+
+	m->cp_hqd_pq_base_lo = lower_32_bits((uint64_t)q->queue_address >> 8);
+	m->cp_hqd_pq_base_hi = upper_32_bits((uint64_t)q->queue_address >> 8);
+
+	m->cp_hqd_pq_rptr_report_addr_lo = lower_32_bits((uint64_t)q->read_ptr);
+	m->cp_hqd_pq_rptr_report_addr_hi = upper_32_bits((uint64_t)q->read_ptr);
+	m->cp_hqd_pq_wptr_poll_addr_lo = lower_32_bits((uint64_t)q->write_ptr);
+	m->cp_hqd_pq_wptr_poll_addr_hi = upper_32_bits((uint64_t)q->write_ptr);
+
+	m->cp_hqd_pq_doorbell_control =
+		q->doorbell_off <<
+			CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_OFFSET__SHIFT;
+	pr_debug("cp_hqd_pq_doorbell_control 0x%x\n",
+			m->cp_hqd_pq_doorbell_control);
+
+	m->cp_hqd_ib_control = 3 << CP_HQD_IB_CONTROL__MIN_IB_AVAIL_SIZE__SHIFT;
+
+	/*
+	 * HW does not clamp this field correctly. Maximum EOP queue size
+	 * is constrained by per-SE EOP done signal count, which is 8-bit.
+	 * Limit is 0xFF EOP entries (= 0x7F8 dwords). CP will not submit
+	 * more than (EOP entry count - 1) so a queue size of 0x800 dwords
+	 * is safe, giving a maximum field value of 0xA.
+	 */
+	m->cp_hqd_eop_control = min(0xA,
+		ffs(q->eop_ring_buffer_size / sizeof(unsigned int)) - 1 - 1);
+	m->cp_hqd_eop_base_addr_lo =
+			lower_32_bits(q->eop_ring_buffer_address >> 8);
+	m->cp_hqd_eop_base_addr_hi =
+			upper_32_bits(q->eop_ring_buffer_address >> 8);
+
+	m->cp_hqd_iq_timer = 0;
+
+	m->cp_hqd_vmid = q->vmid;
+
+	if (q->format == KFD_QUEUE_FORMAT_AQL) {
+		/* GC 10 removed WPP_CLAMP from PQ Control */
+		m->cp_hqd_pq_control |= CP_HQD_PQ_CONTROL__NO_UPDATE_RPTR_MASK |
+				2 << CP_HQD_PQ_CONTROL__SLOT_BASED_WPTR__SHIFT |
+				1 << CP_HQD_PQ_CONTROL__QUEUE_FULL_EN__SHIFT ;
+		m->cp_hqd_pq_doorbell_control |=
+			1 << CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_BIF_DROP__SHIFT;
+	}
+	if (mm->dev->kfd->cwsr_enabled)
+		m->cp_hqd_ctx_save_control = 0;
+
+	update_cu_mask(mm, mqd, minfo);
+	set_priority(m, q);
+
+	q->is_active = QUEUE_IS_ACTIVE(*q);
+}
+
+static bool check_preemption_failed(struct mqd_manager *mm, void *mqd)
+{
+	struct v11_compute_mqd *m = (struct v11_compute_mqd *)mqd;
+
+	return kfd_check_hiq_mqd_doorbell_id(mm->dev, m->queue_doorbell_id0, 0);
+}
+
+static int get_wave_state(struct mqd_manager *mm, void *mqd,
+			  struct queue_properties *q,
+			  void __user *ctl_stack,
+			  u32 *ctl_stack_used_size,
+			  u32 *save_area_used_size)
+{
+	struct v11_compute_mqd *m;
+	struct kfd_context_save_area_header header;
+
+	m = get_mqd(mqd);
+
+	/* Control stack is written backwards, while workgroup context data
+	 * is written forwards. Both starts from m->cp_hqd_cntl_stack_size.
+	 * Current position is at m->cp_hqd_cntl_stack_offset and
+	 * m->cp_hqd_wg_state_offset, respectively.
+	 */
+	*ctl_stack_used_size = m->cp_hqd_cntl_stack_size -
+		m->cp_hqd_cntl_stack_offset;
+	*save_area_used_size = m->cp_hqd_wg_state_offset -
+		m->cp_hqd_cntl_stack_size;
+
+	/* Control stack is not copied to user mode for GFXv11 because
+	 * it's part of the context save area that is already
+	 * accessible to user mode
+	 */
+	header.wave_state.control_stack_size = *ctl_stack_used_size;
+	header.wave_state.wave_state_size = *save_area_used_size;
+
+	header.wave_state.wave_state_offset = m->cp_hqd_wg_state_offset;
+	header.wave_state.control_stack_offset = m->cp_hqd_cntl_stack_offset;
+
+	if (copy_to_user(ctl_stack, &header, sizeof(header.wave_state)))
+		return -EFAULT;
+
+	return 0;
+}
+
+static void checkpoint_mqd(struct mqd_manager *mm, void *mqd, void *mqd_dst, void *ctl_stack_dst)
+{
+	struct v11_compute_mqd *m;
+
+	m = get_mqd(mqd);
+
+	memcpy(mqd_dst, m, sizeof(struct v11_compute_mqd));
+}
+
+static void restore_mqd(struct mqd_manager *mm, void **mqd,
+			struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+			struct queue_properties *qp,
+			const void *mqd_src,
+			const void *ctl_stack_src, const u32 ctl_stack_size)
+{
+	uint64_t addr;
+	struct v11_compute_mqd *m;
+
+	m = (struct v11_compute_mqd *) mqd_mem_obj->cpu_ptr;
+	addr = mqd_mem_obj->gpu_addr;
+
+	memcpy(m, mqd_src, sizeof(*m));
+
+	*mqd = m;
+	if (gart_addr)
+		*gart_addr = addr;
+
+	m->cp_hqd_pq_doorbell_control =
+		qp->doorbell_off <<
+			CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_OFFSET__SHIFT;
+	pr_debug("cp_hqd_pq_doorbell_control 0x%x\n",
+			m->cp_hqd_pq_doorbell_control);
+
+	qp->is_active = 0;
+}
+
+
+static void init_mqd_hiq(struct mqd_manager *mm, void **mqd,
+			struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+			struct queue_properties *q)
+{
+	struct v11_compute_mqd *m;
+
+	init_mqd(mm, mqd, mqd_mem_obj, gart_addr, q);
+
+	m = get_mqd(*mqd);
+
+	m->cp_hqd_pq_control |= 1 << CP_HQD_PQ_CONTROL__PRIV_STATE__SHIFT |
+			1 << CP_HQD_PQ_CONTROL__KMD_QUEUE__SHIFT;
+}
+
+static int destroy_hiq_mqd(struct mqd_manager *mm, void *mqd,
+			enum kfd_preempt_type type, unsigned int timeout,
+			uint32_t pipe_id, uint32_t queue_id)
+{
+	int err;
+	struct v11_compute_mqd *m;
+	u32 doorbell_off;
+
+	m = get_mqd(mqd);
+
+	doorbell_off = m->cp_hqd_pq_doorbell_control >>
+			CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_OFFSET__SHIFT;
+
+	err = amdgpu_amdkfd_unmap_hiq(mm->dev->adev, doorbell_off, 0);
+	if (err)
+		pr_debug("Destroy HIQ MQD failed: %d\n", err);
+
+	return err;
+}
+
+static void init_mqd_sdma(struct mqd_manager *mm, void **mqd,
+		struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+		struct queue_properties *q)
+{
+	struct v11_sdma_mqd *m;
+	int size;
+
+	m = (struct v11_sdma_mqd *) mqd_mem_obj->cpu_ptr;
+
+	if (mm->dev->kfd->shared_resources.enable_mes)
+		size = PAGE_SIZE;
+	else
+		size = sizeof(struct v11_sdma_mqd);
+
+	memset(m, 0, size);
+	*mqd = m;
+	if (gart_addr)
+		*gart_addr = mqd_mem_obj->gpu_addr;
+
+	mm->update_mqd(mm, m, q, NULL);
+}
+
+#define SDMA_RLC_DUMMY_DEFAULT 0xf
+
+static void update_mqd_sdma(struct mqd_manager *mm, void *mqd,
+		struct queue_properties *q,
+		struct mqd_update_info *minfo)
+{
+	struct v11_sdma_mqd *m;
+
+	m = get_sdma_mqd(mqd);
+	m->sdmax_rlcx_rb_cntl = (ffs(q->queue_size / sizeof(unsigned int)) - 1)
+		<< SDMA0_QUEUE0_RB_CNTL__RB_SIZE__SHIFT |
+		q->vmid << SDMA0_QUEUE0_RB_CNTL__RB_VMID__SHIFT |
+		1 << SDMA0_QUEUE0_RB_CNTL__RPTR_WRITEBACK_ENABLE__SHIFT |
+		6 << SDMA0_QUEUE0_RB_CNTL__RPTR_WRITEBACK_TIMER__SHIFT |
+		1 << SDMA0_QUEUE0_RB_CNTL__F32_WPTR_POLL_ENABLE__SHIFT;
+
+	m->sdmax_rlcx_rb_base = lower_32_bits(q->queue_address >> 8);
+	m->sdmax_rlcx_rb_base_hi = upper_32_bits(q->queue_address >> 8);
+	m->sdmax_rlcx_rb_rptr_addr_lo = lower_32_bits((uint64_t)q->read_ptr);
+	m->sdmax_rlcx_rb_rptr_addr_hi = upper_32_bits((uint64_t)q->read_ptr);
+	m->sdmax_rlcx_rb_wptr_poll_addr_lo = lower_32_bits((uint64_t)q->write_ptr);
+	m->sdmax_rlcx_rb_wptr_poll_addr_hi = upper_32_bits((uint64_t)q->write_ptr);
+	m->sdmax_rlcx_doorbell_offset =
+		q->doorbell_off << SDMA0_QUEUE0_DOORBELL_OFFSET__OFFSET__SHIFT;
+
+	m->sdmax_rlcx_sched_cntl = (amdgpu_sdma_phase_quantum
+		<< SDMA0_QUEUE0_SCHEDULE_CNTL__CONTEXT_QUANTUM__SHIFT)
+		 & SDMA0_QUEUE0_SCHEDULE_CNTL__CONTEXT_QUANTUM_MASK;
+
+	m->sdma_engine_id = q->sdma_engine_id;
+	m->sdma_queue_id = q->sdma_queue_id;
+	m->sdmax_rlcx_dummy_reg = SDMA_RLC_DUMMY_DEFAULT;
+
+	q->is_active = QUEUE_IS_ACTIVE(*q);
+}
+
+#if defined(CONFIG_DEBUG_FS)
+
+static int debugfs_show_mqd(struct seq_file *m, void *data)
+{
+	seq_hex_dump(m, "    ", DUMP_PREFIX_OFFSET, 32, 4,
+		     data, sizeof(struct v11_compute_mqd), false);
+	return 0;
+}
+
+static int debugfs_show_mqd_sdma(struct seq_file *m, void *data)
+{
+	seq_hex_dump(m, "    ", DUMP_PREFIX_OFFSET, 32, 4,
+		     data, sizeof(struct v11_sdma_mqd), false);
+	return 0;
+}
+
+#endif
+
+struct mqd_manager *mqd_manager_init_v11(enum KFD_MQD_TYPE type,
+		struct kfd_node *dev)
+{
+	struct mqd_manager *mqd;
+
+	if (WARN_ON(type >= KFD_MQD_TYPE_MAX))
+		return NULL;
+
+	mqd = kzalloc(sizeof(*mqd), GFP_KERNEL);
+	if (!mqd)
+		return NULL;
+
+	mqd->dev = dev;
+
+	switch (type) {
+	case KFD_MQD_TYPE_CP:
+		pr_debug("%s@%i\n", __func__, __LINE__);
+		mqd->allocate_mqd = allocate_mqd;
+		mqd->init_mqd = init_mqd;
+		mqd->free_mqd = kfd_free_mqd_cp;
+		mqd->load_mqd = load_mqd;
+		mqd->update_mqd = update_mqd;
+		mqd->destroy_mqd = kfd_destroy_mqd_cp;
+		mqd->is_occupied = kfd_is_occupied_cp;
+		mqd->mqd_size = sizeof(struct v11_compute_mqd);
+		mqd->get_wave_state = get_wave_state;
+		mqd->mqd_stride = kfd_mqd_stride;
+		mqd->checkpoint_mqd = checkpoint_mqd;
+		mqd->restore_mqd = restore_mqd;
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd;
+#endif
+		pr_debug("%s@%i\n", __func__, __LINE__);
+		break;
+	case KFD_MQD_TYPE_HIQ:
+		pr_debug("%s@%i\n", __func__, __LINE__);
+		mqd->allocate_mqd = allocate_hiq_mqd;
+		mqd->init_mqd = init_mqd_hiq;
+		mqd->free_mqd = free_mqd_hiq_sdma;
+		mqd->load_mqd = kfd_hiq_load_mqd_kiq;
+		mqd->update_mqd = update_mqd;
+		mqd->destroy_mqd = destroy_hiq_mqd;
+		mqd->is_occupied = kfd_is_occupied_cp;
+		mqd->mqd_size = sizeof(struct v11_compute_mqd);
+		mqd->mqd_stride = kfd_mqd_stride;
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd;
+#endif
+		mqd->check_preemption_failed = check_preemption_failed;
+		pr_debug("%s@%i\n", __func__, __LINE__);
+		break;
+	case KFD_MQD_TYPE_DIQ:
+		mqd->allocate_mqd = allocate_mqd;
+		mqd->init_mqd = init_mqd_hiq;
+		mqd->free_mqd = kfd_free_mqd_cp;
+		mqd->load_mqd = load_mqd;
+		mqd->update_mqd = update_mqd;
+		mqd->destroy_mqd = kfd_destroy_mqd_cp;
+		mqd->is_occupied = kfd_is_occupied_cp;
+		mqd->mqd_size = sizeof(struct v11_compute_mqd);
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd;
+#endif
+		break;
+	case KFD_MQD_TYPE_SDMA:
+		pr_debug("%s@%i\n", __func__, __LINE__);
+		mqd->allocate_mqd = allocate_sdma_mqd;
+		mqd->init_mqd = init_mqd_sdma;
+		mqd->free_mqd = free_mqd_hiq_sdma;
+		mqd->load_mqd = kfd_load_mqd_sdma;
+		mqd->update_mqd = update_mqd_sdma;
+		mqd->destroy_mqd = kfd_destroy_mqd_sdma;
+		mqd->is_occupied = kfd_is_occupied_sdma;
+		mqd->checkpoint_mqd = checkpoint_mqd;
+		mqd->restore_mqd = restore_mqd;
+		mqd->mqd_size = sizeof(struct v11_sdma_mqd);
+		mqd->mqd_stride = kfd_mqd_stride;
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd_sdma;
+#endif
+		/*
+		 * To allocate SDMA MQDs by generic functions
+		 * when MES is enabled.
+		 */
+		if (dev->kfd->shared_resources.enable_mes) {
+			mqd->allocate_mqd = allocate_mqd;
+			mqd->free_mqd = kfd_free_mqd_cp;
+		}
+		pr_debug("%s@%i\n", __func__, __LINE__);
+		break;
+	default:
+		kfree(mqd);
+		return NULL;
+	}
+
+	return mqd;
+}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v12.c b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v12.c
new file mode 100644
index 000000000000..565858b9044d
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v12.c
@@ -0,0 +1,459 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/printk.h>
+#include <linux/slab.h>
+#include <linux/uaccess.h>
+#include "kfd_priv.h"
+#include "kfd_mqd_manager.h"
+#include "v12_structs.h"
+#include "gc/gc_12_0_0_sh_mask.h"
+#include "amdgpu_amdkfd.h"
+
+static inline struct v12_compute_mqd *get_mqd(void *mqd)
+{
+	return (struct v12_compute_mqd *)mqd;
+}
+
+static inline struct v12_sdma_mqd *get_sdma_mqd(void *mqd)
+{
+	return (struct v12_sdma_mqd *)mqd;
+}
+
+static void update_cu_mask(struct mqd_manager *mm, void *mqd,
+			   struct mqd_update_info *minfo)
+{
+	struct v12_compute_mqd *m;
+	uint32_t se_mask[KFD_MAX_NUM_SE] = {0};
+
+	if (!minfo || !minfo->cu_mask.ptr)
+		return;
+
+	mqd_symmetrically_map_cu_mask(mm,
+		minfo->cu_mask.ptr, minfo->cu_mask.count, se_mask, 0);
+
+	m = get_mqd(mqd);
+	m->compute_static_thread_mgmt_se0 = se_mask[0];
+	m->compute_static_thread_mgmt_se1 = se_mask[1];
+	m->compute_static_thread_mgmt_se2 = se_mask[2];
+	m->compute_static_thread_mgmt_se3 = se_mask[3];
+	m->compute_static_thread_mgmt_se4 = se_mask[4];
+	m->compute_static_thread_mgmt_se5 = se_mask[5];
+	m->compute_static_thread_mgmt_se6 = se_mask[6];
+	m->compute_static_thread_mgmt_se7 = se_mask[7];
+
+	pr_debug("update cu mask to %#x %#x %#x %#x %#x %#x %#x %#x\n",
+		m->compute_static_thread_mgmt_se0,
+		m->compute_static_thread_mgmt_se1,
+		m->compute_static_thread_mgmt_se2,
+		m->compute_static_thread_mgmt_se3,
+		m->compute_static_thread_mgmt_se4,
+		m->compute_static_thread_mgmt_se5,
+		m->compute_static_thread_mgmt_se6,
+		m->compute_static_thread_mgmt_se7);
+}
+
+static void set_priority(struct v12_compute_mqd *m, struct queue_properties *q)
+{
+	m->cp_hqd_pipe_priority = pipe_priority_map[q->priority];
+	m->cp_hqd_queue_priority = q->priority;
+}
+
+static struct kfd_mem_obj *allocate_mqd(struct kfd_node *node,
+		struct queue_properties *q)
+{
+	struct kfd_mem_obj *mqd_mem_obj;
+
+	/*
+	 * Allocate one PAGE_SIZE memory for MQD as MES writes to areas beyond
+	 * struct MQD size.
+	 */
+	if (kfd_gtt_sa_allocate(node, PAGE_SIZE, &mqd_mem_obj))
+		return NULL;
+
+	return mqd_mem_obj;
+}
+
+static void init_mqd(struct mqd_manager *mm, void **mqd,
+			struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+			struct queue_properties *q)
+{
+	uint64_t addr;
+	struct v12_compute_mqd *m;
+
+	m = (struct v12_compute_mqd *) mqd_mem_obj->cpu_ptr;
+	addr = mqd_mem_obj->gpu_addr;
+
+	memset(m, 0, PAGE_SIZE);
+
+	m->header = 0xC0310800;
+	m->compute_pipelinestat_enable = 1;
+	m->compute_static_thread_mgmt_se0 = 0xFFFFFFFF;
+	m->compute_static_thread_mgmt_se1 = 0xFFFFFFFF;
+	m->compute_static_thread_mgmt_se2 = 0xFFFFFFFF;
+	m->compute_static_thread_mgmt_se3 = 0xFFFFFFFF;
+	m->compute_static_thread_mgmt_se4 = 0xFFFFFFFF;
+	m->compute_static_thread_mgmt_se5 = 0xFFFFFFFF;
+	m->compute_static_thread_mgmt_se6 = 0xFFFFFFFF;
+	m->compute_static_thread_mgmt_se7 = 0xFFFFFFFF;
+
+	m->cp_hqd_persistent_state = CP_HQD_PERSISTENT_STATE__PRELOAD_REQ_MASK |
+			0x55 << CP_HQD_PERSISTENT_STATE__PRELOAD_SIZE__SHIFT;
+
+	m->cp_hqd_pq_control = 5 << CP_HQD_PQ_CONTROL__RPTR_BLOCK_SIZE__SHIFT;
+	m->cp_hqd_pq_control |= CP_HQD_PQ_CONTROL__UNORD_DISPATCH_MASK;
+	m->cp_mqd_control = 1 << CP_MQD_CONTROL__PRIV_STATE__SHIFT;
+
+	m->cp_mqd_base_addr_lo        = lower_32_bits(addr);
+	m->cp_mqd_base_addr_hi        = upper_32_bits(addr);
+
+	m->cp_hqd_quantum = 1 << CP_HQD_QUANTUM__QUANTUM_EN__SHIFT |
+			1 << CP_HQD_QUANTUM__QUANTUM_SCALE__SHIFT |
+			1 << CP_HQD_QUANTUM__QUANTUM_DURATION__SHIFT;
+
+	/* Set cp_hqd_hq_status0.c_queue_debug_en to 1 to have the CP set up the
+	 * DISPATCH_PTR.  This is required for the kfd debugger
+	 */
+	m->cp_hqd_hq_status0 = 1 << 14;
+
+	if (amdgpu_amdkfd_have_atomics_support(mm->dev->adev))
+		m->cp_hqd_hq_status0 |= 1 << 29;
+
+	if (q->format == KFD_QUEUE_FORMAT_AQL) {
+		m->cp_hqd_aql_control =
+			1 << CP_HQD_AQL_CONTROL__CONTROL0__SHIFT;
+	}
+
+	if (mm->dev->kfd->cwsr_enabled) {
+		m->cp_hqd_persistent_state |=
+			(1 << CP_HQD_PERSISTENT_STATE__QSWITCH_MODE__SHIFT);
+		m->cp_hqd_ctx_save_base_addr_lo =
+			lower_32_bits(q->ctx_save_restore_area_address);
+		m->cp_hqd_ctx_save_base_addr_hi =
+			upper_32_bits(q->ctx_save_restore_area_address);
+		m->cp_hqd_ctx_save_size = q->ctx_save_restore_area_size;
+		m->cp_hqd_cntl_stack_size = q->ctl_stack_size;
+		m->cp_hqd_cntl_stack_offset = q->ctl_stack_size;
+		m->cp_hqd_wg_state_offset = q->ctl_stack_size;
+	}
+
+	*mqd = m;
+	if (gart_addr)
+		*gart_addr = addr;
+	mm->update_mqd(mm, m, q, NULL);
+}
+
+static int load_mqd(struct mqd_manager *mm, void *mqd,
+			uint32_t pipe_id, uint32_t queue_id,
+			struct queue_properties *p, struct mm_struct *mms)
+{
+	int r = 0;
+	/* AQL write pointer counts in 64B packets, PM4/CP counts in dwords. */
+	uint32_t wptr_shift = (p->format == KFD_QUEUE_FORMAT_AQL ? 4 : 0);
+
+	r = mm->dev->kfd2kgd->hqd_load(mm->dev->adev, mqd, pipe_id, queue_id,
+					  (uint32_t __user *)p->write_ptr,
+					  wptr_shift, 0, mms, 0);
+	return r;
+}
+
+static void update_mqd(struct mqd_manager *mm, void *mqd,
+		       struct queue_properties *q,
+		       struct mqd_update_info *minfo)
+{
+	struct v12_compute_mqd *m;
+
+	m = get_mqd(mqd);
+
+	m->cp_hqd_pq_control &= ~CP_HQD_PQ_CONTROL__QUEUE_SIZE_MASK;
+	m->cp_hqd_pq_control |=
+			ffs(q->queue_size / sizeof(unsigned int)) - 1 - 1;
+	pr_debug("cp_hqd_pq_control 0x%x\n", m->cp_hqd_pq_control);
+
+	m->cp_hqd_pq_base_lo = lower_32_bits((uint64_t)q->queue_address >> 8);
+	m->cp_hqd_pq_base_hi = upper_32_bits((uint64_t)q->queue_address >> 8);
+
+	m->cp_hqd_pq_rptr_report_addr_lo = lower_32_bits((uint64_t)q->read_ptr);
+	m->cp_hqd_pq_rptr_report_addr_hi = upper_32_bits((uint64_t)q->read_ptr);
+	m->cp_hqd_pq_wptr_poll_addr_lo = lower_32_bits((uint64_t)q->write_ptr);
+	m->cp_hqd_pq_wptr_poll_addr_hi = upper_32_bits((uint64_t)q->write_ptr);
+
+	m->cp_hqd_pq_doorbell_control =
+		q->doorbell_off <<
+			CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_OFFSET__SHIFT;
+	pr_debug("cp_hqd_pq_doorbell_control 0x%x\n",
+			m->cp_hqd_pq_doorbell_control);
+
+	m->cp_hqd_ib_control = 3 << CP_HQD_IB_CONTROL__MIN_IB_AVAIL_SIZE__SHIFT;
+
+	/*
+	 * HW does not clamp this field correctly. Maximum EOP queue size
+	 * is constrained by per-SE EOP done signal count, which is 8-bit.
+	 * Limit is 0xFF EOP entries (= 0x7F8 dwords). CP will not submit
+	 * more than (EOP entry count - 1) so a queue size of 0x800 dwords
+	 * is safe, giving a maximum field value of 0xA.
+	 */
+	m->cp_hqd_eop_control = min(0xA,
+		ffs(q->eop_ring_buffer_size / sizeof(unsigned int)) - 1 - 1);
+	m->cp_hqd_eop_base_addr_lo =
+			lower_32_bits(q->eop_ring_buffer_address >> 8);
+	m->cp_hqd_eop_base_addr_hi =
+			upper_32_bits(q->eop_ring_buffer_address >> 8);
+
+	m->cp_hqd_iq_timer = 0;
+
+	m->cp_hqd_vmid = q->vmid;
+
+	if (q->format == KFD_QUEUE_FORMAT_AQL) {
+		/* GC 10 removed WPP_CLAMP from PQ Control */
+		m->cp_hqd_pq_control |= CP_HQD_PQ_CONTROL__NO_UPDATE_RPTR_MASK |
+				2 << CP_HQD_PQ_CONTROL__SLOT_BASED_WPTR__SHIFT |
+				1 << CP_HQD_PQ_CONTROL__QUEUE_FULL_EN__SHIFT;
+		m->cp_hqd_pq_doorbell_control |=
+			1 << CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_BIF_DROP__SHIFT;
+	}
+	if (mm->dev->kfd->cwsr_enabled)
+		m->cp_hqd_ctx_save_control = 0;
+
+	update_cu_mask(mm, mqd, minfo);
+	set_priority(m, q);
+
+	q->is_active = QUEUE_IS_ACTIVE(*q);
+}
+
+static bool check_preemption_failed(struct mqd_manager *mm, void *mqd)
+{
+	struct v12_compute_mqd *m = (struct v12_compute_mqd *)mqd;
+
+	return kfd_check_hiq_mqd_doorbell_id(mm->dev, m->queue_doorbell_id0, 0);
+}
+
+static int get_wave_state(struct mqd_manager *mm, void *mqd,
+			  struct queue_properties *q,
+			  void __user *ctl_stack,
+			  u32 *ctl_stack_used_size,
+			  u32 *save_area_used_size)
+{
+	struct v12_compute_mqd *m;
+	struct mqd_user_context_save_area_header header;
+
+	m = get_mqd(mqd);
+
+	/* Control stack is written backwards, while workgroup context data
+	 * is written forwards. Both starts from m->cp_hqd_cntl_stack_size.
+	 * Current position is at m->cp_hqd_cntl_stack_offset and
+	 * m->cp_hqd_wg_state_offset, respectively.
+	 */
+	*ctl_stack_used_size = m->cp_hqd_cntl_stack_size -
+		m->cp_hqd_cntl_stack_offset;
+	*save_area_used_size = m->cp_hqd_wg_state_offset -
+		m->cp_hqd_cntl_stack_size;
+
+	/* Control stack is not copied to user mode for GFXv12 because
+	 * it's part of the context save area that is already
+	 * accessible to user mode
+	 */
+	header.control_stack_size = *ctl_stack_used_size;
+	header.wave_state_size = *save_area_used_size;
+
+	header.wave_state_offset = m->cp_hqd_wg_state_offset;
+	header.control_stack_offset = m->cp_hqd_cntl_stack_offset;
+
+	if (copy_to_user(ctl_stack, &header, sizeof(header)))
+		return -EFAULT;
+
+	return 0;
+}
+
+static void init_mqd_hiq(struct mqd_manager *mm, void **mqd,
+			struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+			struct queue_properties *q)
+{
+	struct v12_compute_mqd *m;
+
+	init_mqd(mm, mqd, mqd_mem_obj, gart_addr, q);
+
+	m = get_mqd(*mqd);
+
+	m->cp_hqd_pq_control |= 1 << CP_HQD_PQ_CONTROL__PRIV_STATE__SHIFT |
+			1 << CP_HQD_PQ_CONTROL__KMD_QUEUE__SHIFT;
+}
+
+static void init_mqd_sdma(struct mqd_manager *mm, void **mqd,
+		struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+		struct queue_properties *q)
+{
+	struct v12_sdma_mqd *m;
+
+	m = (struct v12_sdma_mqd *) mqd_mem_obj->cpu_ptr;
+
+	memset(m, 0, sizeof(struct v12_sdma_mqd));
+
+	*mqd = m;
+	if (gart_addr)
+		*gart_addr = mqd_mem_obj->gpu_addr;
+
+	mm->update_mqd(mm, m, q, NULL);
+}
+
+#define SDMA_RLC_DUMMY_DEFAULT 0xf
+
+static void update_mqd_sdma(struct mqd_manager *mm, void *mqd,
+		struct queue_properties *q,
+		struct mqd_update_info *minfo)
+{
+	struct v12_sdma_mqd *m;
+
+	m = get_sdma_mqd(mqd);
+	m->sdmax_rlcx_rb_cntl = (ffs(q->queue_size / sizeof(unsigned int)) - 1)
+		<< SDMA0_QUEUE0_RB_CNTL__RB_SIZE__SHIFT |
+		q->vmid << SDMA0_QUEUE0_RB_CNTL__RB_VMID__SHIFT |
+		1 << SDMA0_QUEUE0_RB_CNTL__RPTR_WRITEBACK_ENABLE__SHIFT |
+		6 << SDMA0_QUEUE0_RB_CNTL__RPTR_WRITEBACK_TIMER__SHIFT |
+		1 << SDMA0_QUEUE0_RB_CNTL__MCU_WPTR_POLL_ENABLE__SHIFT;
+
+	m->sdmax_rlcx_rb_base = lower_32_bits(q->queue_address >> 8);
+	m->sdmax_rlcx_rb_base_hi = upper_32_bits(q->queue_address >> 8);
+	m->sdmax_rlcx_rb_rptr_addr_lo = lower_32_bits((uint64_t)q->read_ptr);
+	m->sdmax_rlcx_rb_rptr_addr_hi = upper_32_bits((uint64_t)q->read_ptr);
+	m->sdmax_rlcx_rb_wptr_poll_addr_lo = lower_32_bits((uint64_t)q->write_ptr);
+	m->sdmax_rlcx_rb_wptr_poll_addr_hi = upper_32_bits((uint64_t)q->write_ptr);
+	m->sdmax_rlcx_doorbell_offset =
+		q->doorbell_off << SDMA0_QUEUE0_DOORBELL_OFFSET__OFFSET__SHIFT;
+
+	m->sdmax_rlcx_sched_cntl = (amdgpu_sdma_phase_quantum
+		<< SDMA0_QUEUE0_SCHEDULE_CNTL__CONTEXT_QUANTUM__SHIFT)
+		 & SDMA0_QUEUE0_SCHEDULE_CNTL__CONTEXT_QUANTUM_MASK;
+
+	m->sdma_engine_id = q->sdma_engine_id;
+	m->sdma_queue_id = q->sdma_queue_id;
+
+	m->sdmax_rlcx_dummy_reg = SDMA_RLC_DUMMY_DEFAULT;
+
+	q->is_active = QUEUE_IS_ACTIVE(*q);
+}
+
+#if defined(CONFIG_DEBUG_FS)
+
+static int debugfs_show_mqd(struct seq_file *m, void *data)
+{
+	seq_hex_dump(m, "    ", DUMP_PREFIX_OFFSET, 32, 4,
+		     data, sizeof(struct v12_compute_mqd), false);
+	return 0;
+}
+
+static int debugfs_show_mqd_sdma(struct seq_file *m, void *data)
+{
+	seq_hex_dump(m, "    ", DUMP_PREFIX_OFFSET, 32, 4,
+		     data, sizeof(struct v12_sdma_mqd), false);
+	return 0;
+}
+
+#endif
+
+struct mqd_manager *mqd_manager_init_v12(enum KFD_MQD_TYPE type,
+		struct kfd_node *dev)
+{
+	struct mqd_manager *mqd;
+
+	if (WARN_ON(type >= KFD_MQD_TYPE_MAX))
+		return NULL;
+
+	mqd = kzalloc(sizeof(*mqd), GFP_KERNEL);
+	if (!mqd)
+		return NULL;
+
+	mqd->dev = dev;
+
+	switch (type) {
+	case KFD_MQD_TYPE_CP:
+		pr_debug("%s@%i\n", __func__, __LINE__);
+		mqd->allocate_mqd = allocate_mqd;
+		mqd->init_mqd = init_mqd;
+		mqd->free_mqd = kfd_free_mqd_cp;
+		mqd->load_mqd = load_mqd;
+		mqd->update_mqd = update_mqd;
+		mqd->destroy_mqd = kfd_destroy_mqd_cp;
+		mqd->is_occupied = kfd_is_occupied_cp;
+		mqd->mqd_size = sizeof(struct v12_compute_mqd);
+		mqd->get_wave_state = get_wave_state;
+		mqd->mqd_stride = kfd_mqd_stride;
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd;
+#endif
+		pr_debug("%s@%i\n", __func__, __LINE__);
+		break;
+	case KFD_MQD_TYPE_HIQ:
+		pr_debug("%s@%i\n", __func__, __LINE__);
+		mqd->allocate_mqd = allocate_hiq_mqd;
+		mqd->init_mqd = init_mqd_hiq;
+		mqd->free_mqd = free_mqd_hiq_sdma;
+		mqd->load_mqd = kfd_hiq_load_mqd_kiq;
+		mqd->update_mqd = update_mqd;
+		mqd->destroy_mqd = kfd_destroy_mqd_cp;
+		mqd->is_occupied = kfd_is_occupied_cp;
+		mqd->mqd_size = sizeof(struct v12_compute_mqd);
+		mqd->mqd_stride = kfd_mqd_stride;
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd;
+#endif
+		mqd->check_preemption_failed = check_preemption_failed;
+		pr_debug("%s@%i\n", __func__, __LINE__);
+		break;
+	case KFD_MQD_TYPE_DIQ:
+		mqd->allocate_mqd = allocate_mqd;
+		mqd->init_mqd = init_mqd_hiq;
+		mqd->free_mqd = kfd_free_mqd_cp;
+		mqd->load_mqd = load_mqd;
+		mqd->update_mqd = update_mqd;
+		mqd->destroy_mqd = kfd_destroy_mqd_cp;
+		mqd->is_occupied = kfd_is_occupied_cp;
+		mqd->mqd_size = sizeof(struct v12_compute_mqd);
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd;
+#endif
+		break;
+	case KFD_MQD_TYPE_SDMA:
+		pr_debug("%s@%i\n", __func__, __LINE__);
+		mqd->allocate_mqd = allocate_mqd;
+		mqd->init_mqd = init_mqd_sdma;
+		mqd->free_mqd = kfd_free_mqd_cp;
+		mqd->load_mqd = kfd_load_mqd_sdma;
+		mqd->update_mqd = update_mqd_sdma;
+		mqd->destroy_mqd = kfd_destroy_mqd_sdma;
+		mqd->is_occupied = kfd_is_occupied_sdma;
+		mqd->mqd_size = sizeof(struct v12_sdma_mqd);
+		mqd->mqd_stride = kfd_mqd_stride;
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd_sdma;
+#endif
+		pr_debug("%s@%i\n", __func__, __LINE__);
+		break;
+	default:
+		kfree(mqd);
+		return NULL;
+	}
+
+	return mqd;
+}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c
new file mode 100644
index 000000000000..f2dee320fada
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c
@@ -0,0 +1,1034 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
+/*
+ * Copyright 2016-2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/printk.h>
+#include <linux/slab.h>
+#include <linux/uaccess.h>
+#include "kfd_priv.h"
+#include "kfd_mqd_manager.h"
+#include "v9_structs.h"
+#include "gc/gc_9_0_offset.h"
+#include "gc/gc_9_0_sh_mask.h"
+#include "sdma0/sdma0_4_0_sh_mask.h"
+#include "amdgpu_amdkfd.h"
+#include "kfd_device_queue_manager.h"
+
+static void update_mqd(struct mqd_manager *mm, void *mqd,
+		       struct queue_properties *q,
+		       struct mqd_update_info *minfo);
+
+static uint64_t mqd_stride_v9(struct mqd_manager *mm,
+				struct queue_properties *q)
+{
+	if (mm->dev->kfd->cwsr_enabled &&
+	    q->type == KFD_QUEUE_TYPE_COMPUTE)
+		return ALIGN(q->ctl_stack_size, PAGE_SIZE) +
+			ALIGN(sizeof(struct v9_mqd), PAGE_SIZE);
+
+	return mm->mqd_size;
+}
+
+static inline struct v9_mqd *get_mqd(void *mqd)
+{
+	return (struct v9_mqd *)mqd;
+}
+
+static inline struct v9_sdma_mqd *get_sdma_mqd(void *mqd)
+{
+	return (struct v9_sdma_mqd *)mqd;
+}
+
+static void update_cu_mask(struct mqd_manager *mm, void *mqd,
+			struct mqd_update_info *minfo, uint32_t inst)
+{
+	struct v9_mqd *m;
+	uint32_t se_mask[KFD_MAX_NUM_SE] = {0};
+
+	if (!minfo || !minfo->cu_mask.ptr)
+		return;
+
+	mqd_symmetrically_map_cu_mask(mm,
+		minfo->cu_mask.ptr, minfo->cu_mask.count, se_mask, inst);
+
+	m = get_mqd(mqd);
+
+	m->compute_static_thread_mgmt_se0 = se_mask[0];
+	m->compute_static_thread_mgmt_se1 = se_mask[1];
+	m->compute_static_thread_mgmt_se2 = se_mask[2];
+	m->compute_static_thread_mgmt_se3 = se_mask[3];
+	if (KFD_GC_VERSION(mm->dev) != IP_VERSION(9, 4, 3) &&
+	    KFD_GC_VERSION(mm->dev) != IP_VERSION(9, 4, 4) &&
+	    KFD_GC_VERSION(mm->dev) != IP_VERSION(9, 5, 0)) {
+		m->compute_static_thread_mgmt_se4 = se_mask[4];
+		m->compute_static_thread_mgmt_se5 = se_mask[5];
+		m->compute_static_thread_mgmt_se6 = se_mask[6];
+		m->compute_static_thread_mgmt_se7 = se_mask[7];
+
+		pr_debug("update cu mask to %#x %#x %#x %#x %#x %#x %#x %#x\n",
+			m->compute_static_thread_mgmt_se0,
+			m->compute_static_thread_mgmt_se1,
+			m->compute_static_thread_mgmt_se2,
+			m->compute_static_thread_mgmt_se3,
+			m->compute_static_thread_mgmt_se4,
+			m->compute_static_thread_mgmt_se5,
+			m->compute_static_thread_mgmt_se6,
+			m->compute_static_thread_mgmt_se7);
+	} else {
+		pr_debug("inst: %u, update cu mask to %#x %#x %#x %#x\n",
+			inst, m->compute_static_thread_mgmt_se0,
+			m->compute_static_thread_mgmt_se1,
+			m->compute_static_thread_mgmt_se2,
+			m->compute_static_thread_mgmt_se3);
+	}
+}
+
+static void set_priority(struct v9_mqd *m, struct queue_properties *q)
+{
+	m->cp_hqd_pipe_priority = pipe_priority_map[q->priority];
+	m->cp_hqd_queue_priority = q->priority;
+}
+
+static struct kfd_mem_obj *allocate_mqd(struct kfd_node *node,
+		struct queue_properties *q)
+{
+	int retval;
+	struct kfd_mem_obj *mqd_mem_obj = NULL;
+
+	/* For V9 only, due to a HW bug, the control stack of a user mode
+	 * compute queue needs to be allocated just behind the page boundary
+	 * of its regular MQD buffer. So we allocate an enlarged MQD buffer:
+	 * the first page of the buffer serves as the regular MQD buffer
+	 * purpose and the remaining is for control stack. Although the two
+	 * parts are in the same buffer object, they need different memory
+	 * types: MQD part needs UC (uncached) as usual, while control stack
+	 * needs NC (non coherent), which is different from the UC type which
+	 * is used when control stack is allocated in user space.
+	 *
+	 * Because of all those, we use the gtt allocation function instead
+	 * of sub-allocation function for this enlarged MQD buffer. Moreover,
+	 * in order to achieve two memory types in a single buffer object, we
+	 * pass a special bo flag AMDGPU_GEM_CREATE_CP_MQD_GFX9 to instruct
+	 * amdgpu memory functions to do so.
+	 */
+	if (node->kfd->cwsr_enabled && (q->type == KFD_QUEUE_TYPE_COMPUTE)) {
+		mqd_mem_obj = kzalloc(sizeof(struct kfd_mem_obj), GFP_KERNEL);
+		if (!mqd_mem_obj)
+			return NULL;
+		retval = amdgpu_amdkfd_alloc_gtt_mem(node->adev,
+			(ALIGN(q->ctl_stack_size, PAGE_SIZE) +
+			ALIGN(sizeof(struct v9_mqd), PAGE_SIZE)) *
+			NUM_XCC(node->xcc_mask),
+			&(mqd_mem_obj->gtt_mem),
+			&(mqd_mem_obj->gpu_addr),
+			(void *)&(mqd_mem_obj->cpu_ptr), true);
+
+		if (retval) {
+			kfree(mqd_mem_obj);
+			return NULL;
+		}
+	} else {
+		retval = kfd_gtt_sa_allocate(node, sizeof(struct v9_mqd),
+				&mqd_mem_obj);
+		if (retval)
+			return NULL;
+	}
+
+	return mqd_mem_obj;
+}
+
+static void init_mqd(struct mqd_manager *mm, void **mqd,
+			struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+			struct queue_properties *q)
+{
+	uint64_t addr;
+	struct v9_mqd *m;
+
+	m = (struct v9_mqd *) mqd_mem_obj->cpu_ptr;
+	addr = mqd_mem_obj->gpu_addr;
+
+	memset(m, 0, sizeof(struct v9_mqd));
+
+	m->header = 0xC0310800;
+	m->compute_pipelinestat_enable = 1;
+	m->compute_static_thread_mgmt_se0 = 0xFFFFFFFF;
+	m->compute_static_thread_mgmt_se1 = 0xFFFFFFFF;
+	m->compute_static_thread_mgmt_se2 = 0xFFFFFFFF;
+	m->compute_static_thread_mgmt_se3 = 0xFFFFFFFF;
+	m->compute_static_thread_mgmt_se4 = 0xFFFFFFFF;
+	m->compute_static_thread_mgmt_se5 = 0xFFFFFFFF;
+	m->compute_static_thread_mgmt_se6 = 0xFFFFFFFF;
+	m->compute_static_thread_mgmt_se7 = 0xFFFFFFFF;
+
+	m->cp_hqd_persistent_state = CP_HQD_PERSISTENT_STATE__PRELOAD_REQ_MASK |
+			0x53 << CP_HQD_PERSISTENT_STATE__PRELOAD_SIZE__SHIFT;
+
+	m->cp_hqd_pq_control = 5 << CP_HQD_PQ_CONTROL__RPTR_BLOCK_SIZE__SHIFT;
+	m->cp_hqd_pq_control |= CP_HQD_PQ_CONTROL__UNORD_DISPATCH_MASK;
+
+	m->cp_mqd_control = 1 << CP_MQD_CONTROL__PRIV_STATE__SHIFT;
+
+	m->cp_mqd_base_addr_lo        = lower_32_bits(addr);
+	m->cp_mqd_base_addr_hi        = upper_32_bits(addr);
+
+	m->cp_hqd_quantum = 1 << CP_HQD_QUANTUM__QUANTUM_EN__SHIFT |
+			1 << CP_HQD_QUANTUM__QUANTUM_SCALE__SHIFT |
+			1 << CP_HQD_QUANTUM__QUANTUM_DURATION__SHIFT;
+
+	/* Set cp_hqd_hq_scheduler0 bit 14 to 1 to have the CP set up the
+	 * DISPATCH_PTR.  This is required for the kfd debugger
+	 */
+	m->cp_hqd_hq_status0 = 1 << 14;
+
+	if (q->format == KFD_QUEUE_FORMAT_AQL)
+		m->cp_hqd_aql_control =
+			1 << CP_HQD_AQL_CONTROL__CONTROL0__SHIFT;
+
+	if (q->tba_addr) {
+		m->compute_pgm_rsrc2 |=
+			(1 << COMPUTE_PGM_RSRC2__TRAP_PRESENT__SHIFT);
+	}
+
+	if (mm->dev->kfd->cwsr_enabled && q->ctx_save_restore_area_address) {
+		m->cp_hqd_persistent_state |=
+			(1 << CP_HQD_PERSISTENT_STATE__QSWITCH_MODE__SHIFT);
+		m->cp_hqd_ctx_save_base_addr_lo =
+			lower_32_bits(q->ctx_save_restore_area_address);
+		m->cp_hqd_ctx_save_base_addr_hi =
+			upper_32_bits(q->ctx_save_restore_area_address);
+		m->cp_hqd_ctx_save_size = q->ctx_save_restore_area_size;
+		m->cp_hqd_cntl_stack_size = q->ctl_stack_size;
+		m->cp_hqd_cntl_stack_offset = q->ctl_stack_size;
+		m->cp_hqd_wg_state_offset = q->ctl_stack_size;
+	}
+
+	*mqd = m;
+	if (gart_addr)
+		*gart_addr = addr;
+	update_mqd(mm, m, q, NULL);
+}
+
+static int load_mqd(struct mqd_manager *mm, void *mqd,
+			uint32_t pipe_id, uint32_t queue_id,
+			struct queue_properties *p, struct mm_struct *mms)
+{
+	/* AQL write pointer counts in 64B packets, PM4/CP counts in dwords. */
+	uint32_t wptr_shift = (p->format == KFD_QUEUE_FORMAT_AQL ? 4 : 0);
+
+	return mm->dev->kfd2kgd->hqd_load(mm->dev->adev, mqd, pipe_id, queue_id,
+					  (uint32_t __user *)p->write_ptr,
+					  wptr_shift, 0, mms, 0);
+}
+
+static void update_mqd(struct mqd_manager *mm, void *mqd,
+			struct queue_properties *q,
+			struct mqd_update_info *minfo)
+{
+	struct v9_mqd *m;
+
+	m = get_mqd(mqd);
+
+	m->cp_hqd_pq_control &= ~CP_HQD_PQ_CONTROL__QUEUE_SIZE_MASK;
+	m->cp_hqd_pq_control |= order_base_2(q->queue_size / 4) - 1;
+	pr_debug("cp_hqd_pq_control 0x%x\n", m->cp_hqd_pq_control);
+
+	m->cp_hqd_pq_base_lo = lower_32_bits((uint64_t)q->queue_address >> 8);
+	m->cp_hqd_pq_base_hi = upper_32_bits((uint64_t)q->queue_address >> 8);
+
+	m->cp_hqd_pq_rptr_report_addr_lo = lower_32_bits((uint64_t)q->read_ptr);
+	m->cp_hqd_pq_rptr_report_addr_hi = upper_32_bits((uint64_t)q->read_ptr);
+	m->cp_hqd_pq_wptr_poll_addr_lo = lower_32_bits((uint64_t)q->write_ptr);
+	m->cp_hqd_pq_wptr_poll_addr_hi = upper_32_bits((uint64_t)q->write_ptr);
+
+	m->cp_hqd_pq_doorbell_control =
+		q->doorbell_off <<
+			CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_OFFSET__SHIFT;
+	pr_debug("cp_hqd_pq_doorbell_control 0x%x\n",
+			m->cp_hqd_pq_doorbell_control);
+
+	m->cp_hqd_ib_control =
+		3 << CP_HQD_IB_CONTROL__MIN_IB_AVAIL_SIZE__SHIFT |
+		1 << CP_HQD_IB_CONTROL__IB_EXE_DISABLE__SHIFT;
+
+	/*
+	 * HW does not clamp this field correctly. Maximum EOP queue size
+	 * is constrained by per-SE EOP done signal count, which is 8-bit.
+	 * Limit is 0xFF EOP entries (= 0x7F8 dwords). CP will not submit
+	 * more than (EOP entry count - 1) so a queue size of 0x800 dwords
+	 * is safe, giving a maximum field value of 0xA.
+	 *
+	 * Also, do calculation only if EOP is used (size > 0), otherwise
+	 * the order_base_2 calculation provides incorrect result.
+	 *
+	 */
+	m->cp_hqd_eop_control = q->eop_ring_buffer_size ?
+		min(0xA, order_base_2(q->eop_ring_buffer_size / 4) - 1) : 0;
+
+	m->cp_hqd_eop_base_addr_lo =
+			lower_32_bits(q->eop_ring_buffer_address >> 8);
+	m->cp_hqd_eop_base_addr_hi =
+			upper_32_bits(q->eop_ring_buffer_address >> 8);
+
+	m->cp_hqd_iq_timer = 0;
+
+	m->cp_hqd_vmid = q->vmid;
+
+	if (q->format == KFD_QUEUE_FORMAT_AQL) {
+		m->cp_hqd_pq_control |= CP_HQD_PQ_CONTROL__NO_UPDATE_RPTR_MASK |
+				2 << CP_HQD_PQ_CONTROL__SLOT_BASED_WPTR__SHIFT |
+				1 << CP_HQD_PQ_CONTROL__QUEUE_FULL_EN__SHIFT |
+				1 << CP_HQD_PQ_CONTROL__WPP_CLAMP_EN__SHIFT;
+		m->cp_hqd_pq_doorbell_control |= 1 <<
+			CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_BIF_DROP__SHIFT;
+	}
+	if (mm->dev->kfd->cwsr_enabled && q->ctx_save_restore_area_address)
+		m->cp_hqd_ctx_save_control = 0;
+
+	if (KFD_GC_VERSION(mm->dev) != IP_VERSION(9, 4, 3) &&
+	    KFD_GC_VERSION(mm->dev) != IP_VERSION(9, 4, 4) &&
+	    KFD_GC_VERSION(mm->dev) != IP_VERSION(9, 5, 0))
+		update_cu_mask(mm, mqd, minfo, 0);
+	set_priority(m, q);
+
+	if (minfo && KFD_GC_VERSION(mm->dev) >= IP_VERSION(9, 4, 2)) {
+		if (minfo->update_flag & UPDATE_FLAG_IS_GWS)
+			m->compute_resource_limits |=
+				COMPUTE_RESOURCE_LIMITS__FORCE_SIMD_DIST_MASK;
+		else
+			m->compute_resource_limits &=
+				~COMPUTE_RESOURCE_LIMITS__FORCE_SIMD_DIST_MASK;
+	}
+
+	q->is_active = QUEUE_IS_ACTIVE(*q);
+}
+
+
+static bool check_preemption_failed(struct mqd_manager *mm, void *mqd)
+{
+	struct v9_mqd *m = (struct v9_mqd *)mqd;
+	uint32_t doorbell_id = m->queue_doorbell_id0;
+
+	m->queue_doorbell_id0 = 0;
+
+	return kfd_check_hiq_mqd_doorbell_id(mm->dev, doorbell_id, 0);
+}
+
+static int get_wave_state(struct mqd_manager *mm, void *mqd,
+			  struct queue_properties *q,
+			  void __user *ctl_stack,
+			  u32 *ctl_stack_used_size,
+			  u32 *save_area_used_size)
+{
+	struct v9_mqd *m;
+	struct kfd_context_save_area_header header;
+
+	/* Control stack is located one page after MQD. */
+	void *mqd_ctl_stack = (void *)((uintptr_t)mqd + PAGE_SIZE);
+
+	m = get_mqd(mqd);
+
+	*ctl_stack_used_size = m->cp_hqd_cntl_stack_size -
+		m->cp_hqd_cntl_stack_offset;
+	*save_area_used_size = m->cp_hqd_wg_state_offset -
+		m->cp_hqd_cntl_stack_size;
+
+	header.wave_state.control_stack_size = *ctl_stack_used_size;
+	header.wave_state.wave_state_size = *save_area_used_size;
+
+	header.wave_state.wave_state_offset = m->cp_hqd_wg_state_offset;
+	header.wave_state.control_stack_offset = m->cp_hqd_cntl_stack_offset;
+
+	if (copy_to_user(ctl_stack, &header, sizeof(header.wave_state)))
+		return -EFAULT;
+
+	if (copy_to_user(ctl_stack + m->cp_hqd_cntl_stack_offset,
+				mqd_ctl_stack + m->cp_hqd_cntl_stack_offset,
+				*ctl_stack_used_size))
+		return -EFAULT;
+
+	return 0;
+}
+
+static void get_checkpoint_info(struct mqd_manager *mm, void *mqd, u32 *ctl_stack_size)
+{
+	struct v9_mqd *m = get_mqd(mqd);
+
+	*ctl_stack_size = m->cp_hqd_cntl_stack_size * NUM_XCC(mm->dev->xcc_mask);
+}
+
+static void checkpoint_mqd(struct mqd_manager *mm, void *mqd, void *mqd_dst, void *ctl_stack_dst)
+{
+	struct v9_mqd *m;
+	/* Control stack is located one page after MQD. */
+	void *ctl_stack = (void *)((uintptr_t)mqd + PAGE_SIZE);
+
+	m = get_mqd(mqd);
+
+	memcpy(mqd_dst, m, sizeof(struct v9_mqd));
+	memcpy(ctl_stack_dst, ctl_stack, m->cp_hqd_cntl_stack_size);
+}
+
+static void checkpoint_mqd_v9_4_3(struct mqd_manager *mm,
+								  void *mqd,
+								  void *mqd_dst,
+								  void *ctl_stack_dst)
+{
+	struct v9_mqd *m;
+	int xcc;
+	uint64_t size = get_mqd(mqd)->cp_mqd_stride_size;
+
+	for (xcc = 0; xcc < NUM_XCC(mm->dev->xcc_mask); xcc++) {
+		m = get_mqd(mqd + size * xcc);
+
+		checkpoint_mqd(mm, m,
+				(uint8_t *)mqd_dst + sizeof(*m) * xcc,
+				(uint8_t *)ctl_stack_dst + m->cp_hqd_cntl_stack_size * xcc);
+	}
+}
+
+static void restore_mqd(struct mqd_manager *mm, void **mqd,
+			struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+			struct queue_properties *qp,
+			const void *mqd_src,
+			const void *ctl_stack_src, u32 ctl_stack_size)
+{
+	uint64_t addr;
+	struct v9_mqd *m;
+	void *ctl_stack;
+
+	m = (struct v9_mqd *) mqd_mem_obj->cpu_ptr;
+	addr = mqd_mem_obj->gpu_addr;
+
+	memcpy(m, mqd_src, sizeof(*m));
+
+	*mqd = m;
+	if (gart_addr)
+		*gart_addr = addr;
+
+	/* Control stack is located one page after MQD. */
+	ctl_stack = (void *)((uintptr_t)*mqd + PAGE_SIZE);
+	memcpy(ctl_stack, ctl_stack_src, ctl_stack_size);
+
+	m->cp_hqd_pq_doorbell_control =
+		qp->doorbell_off <<
+			CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_OFFSET__SHIFT;
+	pr_debug("cp_hqd_pq_doorbell_control 0x%x\n",
+				m->cp_hqd_pq_doorbell_control);
+
+	qp->is_active = 0;
+}
+
+static void init_mqd_hiq(struct mqd_manager *mm, void **mqd,
+			struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+			struct queue_properties *q)
+{
+	struct v9_mqd *m;
+
+	init_mqd(mm, mqd, mqd_mem_obj, gart_addr, q);
+
+	m = get_mqd(*mqd);
+
+	m->cp_hqd_pq_control |= 1 << CP_HQD_PQ_CONTROL__PRIV_STATE__SHIFT |
+			1 << CP_HQD_PQ_CONTROL__KMD_QUEUE__SHIFT;
+}
+
+static int destroy_hiq_mqd(struct mqd_manager *mm, void *mqd,
+			enum kfd_preempt_type type, unsigned int timeout,
+			uint32_t pipe_id, uint32_t queue_id)
+{
+	int err;
+	struct v9_mqd *m;
+	u32 doorbell_off;
+
+	m = get_mqd(mqd);
+
+	doorbell_off = m->cp_hqd_pq_doorbell_control >>
+			CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_OFFSET__SHIFT;
+	err = amdgpu_amdkfd_unmap_hiq(mm->dev->adev, doorbell_off, 0);
+	if (err)
+		pr_debug("Destroy HIQ MQD failed: %d\n", err);
+
+	return err;
+}
+
+static void init_mqd_sdma(struct mqd_manager *mm, void **mqd,
+		struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+		struct queue_properties *q)
+{
+	struct v9_sdma_mqd *m;
+
+	m = (struct v9_sdma_mqd *) mqd_mem_obj->cpu_ptr;
+
+	memset(m, 0, sizeof(struct v9_sdma_mqd));
+
+	*mqd = m;
+	if (gart_addr)
+		*gart_addr = mqd_mem_obj->gpu_addr;
+
+	mm->update_mqd(mm, m, q, NULL);
+}
+
+#define SDMA_RLC_DUMMY_DEFAULT 0xf
+
+static void update_mqd_sdma(struct mqd_manager *mm, void *mqd,
+			struct queue_properties *q,
+			struct mqd_update_info *minfo)
+{
+	struct v9_sdma_mqd *m;
+
+	m = get_sdma_mqd(mqd);
+	m->sdmax_rlcx_rb_cntl = order_base_2(q->queue_size / 4)
+		<< SDMA0_RLC0_RB_CNTL__RB_SIZE__SHIFT |
+		q->vmid << SDMA0_RLC0_RB_CNTL__RB_VMID__SHIFT |
+		1 << SDMA0_RLC0_RB_CNTL__RPTR_WRITEBACK_ENABLE__SHIFT |
+		6 << SDMA0_RLC0_RB_CNTL__RPTR_WRITEBACK_TIMER__SHIFT;
+
+	m->sdmax_rlcx_rb_base = lower_32_bits(q->queue_address >> 8);
+	m->sdmax_rlcx_rb_base_hi = upper_32_bits(q->queue_address >> 8);
+	m->sdmax_rlcx_rb_rptr_addr_lo = lower_32_bits((uint64_t)q->read_ptr);
+	m->sdmax_rlcx_rb_rptr_addr_hi = upper_32_bits((uint64_t)q->read_ptr);
+	m->sdmax_rlcx_doorbell_offset =
+		q->doorbell_off << SDMA0_RLC0_DOORBELL_OFFSET__OFFSET__SHIFT;
+
+	m->sdma_engine_id = q->sdma_engine_id;
+	m->sdma_queue_id = q->sdma_queue_id;
+	m->sdmax_rlcx_dummy_reg = SDMA_RLC_DUMMY_DEFAULT;
+	/* Allow context switch so we don't cross-process starve with a massive
+	 * command buffer of long-running SDMA commands
+	 */
+	m->sdmax_rlcx_ib_cntl |= SDMA0_GFX_IB_CNTL__SWITCH_INSIDE_IB_MASK;
+
+	q->is_active = QUEUE_IS_ACTIVE(*q);
+}
+
+static void checkpoint_mqd_sdma(struct mqd_manager *mm,
+				void *mqd,
+				void *mqd_dst,
+				void *ctl_stack_dst)
+{
+	struct v9_sdma_mqd *m;
+
+	m = get_sdma_mqd(mqd);
+
+	memcpy(mqd_dst, m, sizeof(struct v9_sdma_mqd));
+}
+
+static void restore_mqd_sdma(struct mqd_manager *mm, void **mqd,
+			     struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+			     struct queue_properties *qp,
+			     const void *mqd_src,
+			     const void *ctl_stack_src, const u32 ctl_stack_size)
+{
+	uint64_t addr;
+	struct v9_sdma_mqd *m;
+
+	m = (struct v9_sdma_mqd *) mqd_mem_obj->cpu_ptr;
+	addr = mqd_mem_obj->gpu_addr;
+
+	memcpy(m, mqd_src, sizeof(*m));
+
+	m->sdmax_rlcx_doorbell_offset =
+		qp->doorbell_off << SDMA0_RLC0_DOORBELL_OFFSET__OFFSET__SHIFT;
+
+	*mqd = m;
+	if (gart_addr)
+		*gart_addr = addr;
+
+	qp->is_active = 0;
+}
+
+static void init_mqd_hiq_v9_4_3(struct mqd_manager *mm, void **mqd,
+			struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+			struct queue_properties *q)
+{
+	struct v9_mqd *m;
+	int xcc = 0;
+	struct kfd_mem_obj xcc_mqd_mem_obj;
+	uint64_t xcc_gart_addr = 0;
+
+	memset(&xcc_mqd_mem_obj, 0x0, sizeof(struct kfd_mem_obj));
+
+	for (xcc = 0; xcc < NUM_XCC(mm->dev->xcc_mask); xcc++) {
+		kfd_get_hiq_xcc_mqd(mm->dev, &xcc_mqd_mem_obj, xcc);
+
+		init_mqd(mm, (void **)&m, &xcc_mqd_mem_obj, &xcc_gart_addr, q);
+
+		m->cp_hqd_pq_control |= CP_HQD_PQ_CONTROL__NO_UPDATE_RPTR_MASK |
+					1 << CP_HQD_PQ_CONTROL__PRIV_STATE__SHIFT |
+					1 << CP_HQD_PQ_CONTROL__KMD_QUEUE__SHIFT;
+		if (amdgpu_sriov_multi_vf_mode(mm->dev->adev))
+			m->cp_hqd_pq_doorbell_control |= 1 <<
+				CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_MODE__SHIFT;
+		m->cp_mqd_stride_size = kfd_hiq_mqd_stride(mm->dev);
+		if (xcc == 0) {
+			/* Set no_update_rptr = 0 in Master XCC */
+			m->cp_hqd_pq_control &= ~CP_HQD_PQ_CONTROL__NO_UPDATE_RPTR_MASK;
+
+			/* Set the MQD pointer and gart address to XCC0 MQD */
+			*mqd = m;
+			*gart_addr = xcc_gart_addr;
+		}
+	}
+}
+
+static int hiq_load_mqd_kiq_v9_4_3(struct mqd_manager *mm, void *mqd,
+			uint32_t pipe_id, uint32_t queue_id,
+			struct queue_properties *p, struct mm_struct *mms)
+{
+	uint32_t xcc_mask = mm->dev->xcc_mask;
+	int xcc_id, err, inst = 0;
+	void *xcc_mqd;
+	uint64_t hiq_mqd_size = kfd_hiq_mqd_stride(mm->dev);
+
+	for_each_inst(xcc_id, xcc_mask) {
+		xcc_mqd = mqd + hiq_mqd_size * inst;
+		err = mm->dev->kfd2kgd->hiq_mqd_load(mm->dev->adev, xcc_mqd,
+						     pipe_id, queue_id,
+						     p->doorbell_off, xcc_id);
+		if (err) {
+			pr_debug("Failed to load HIQ MQD for XCC: %d\n", inst);
+			break;
+		}
+		++inst;
+	}
+
+	return err;
+}
+
+static int destroy_hiq_mqd_v9_4_3(struct mqd_manager *mm, void *mqd,
+			enum kfd_preempt_type type, unsigned int timeout,
+			uint32_t pipe_id, uint32_t queue_id)
+{
+	uint32_t xcc_mask = mm->dev->xcc_mask;
+	int xcc_id, err, inst = 0;
+	uint64_t hiq_mqd_size = kfd_hiq_mqd_stride(mm->dev);
+	struct v9_mqd *m;
+	u32 doorbell_off;
+
+	for_each_inst(xcc_id, xcc_mask) {
+		m = get_mqd(mqd + hiq_mqd_size * inst);
+
+		doorbell_off = m->cp_hqd_pq_doorbell_control >>
+				CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_OFFSET__SHIFT;
+
+		err = amdgpu_amdkfd_unmap_hiq(mm->dev->adev, doorbell_off, xcc_id);
+		if (err) {
+			pr_debug("Destroy HIQ MQD failed for xcc: %d\n", inst);
+			break;
+		}
+		++inst;
+	}
+
+	return err;
+}
+
+static bool check_preemption_failed_v9_4_3(struct mqd_manager *mm, void *mqd)
+{
+	uint64_t hiq_mqd_size = kfd_hiq_mqd_stride(mm->dev);
+	uint32_t xcc_mask = mm->dev->xcc_mask;
+	int inst = 0, xcc_id;
+	struct v9_mqd *m;
+	bool ret = false;
+
+	for_each_inst(xcc_id, xcc_mask) {
+		m = get_mqd(mqd + hiq_mqd_size * inst);
+		ret |= kfd_check_hiq_mqd_doorbell_id(mm->dev,
+					m->queue_doorbell_id0, inst);
+		m->queue_doorbell_id0 = 0;
+		++inst;
+	}
+
+	return ret;
+}
+
+static void get_xcc_mqd(struct kfd_mem_obj *mqd_mem_obj,
+			       struct kfd_mem_obj *xcc_mqd_mem_obj,
+			       uint64_t offset)
+{
+	xcc_mqd_mem_obj->gtt_mem = (offset == 0) ?
+					mqd_mem_obj->gtt_mem : NULL;
+	xcc_mqd_mem_obj->gpu_addr = mqd_mem_obj->gpu_addr + offset;
+	xcc_mqd_mem_obj->cpu_ptr = (uint32_t *)((uintptr_t)mqd_mem_obj->cpu_ptr
+						+ offset);
+}
+
+static void init_mqd_v9_4_3(struct mqd_manager *mm, void **mqd,
+			struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+			struct queue_properties *q)
+{
+	struct v9_mqd *m;
+	int xcc = 0;
+	struct kfd_mem_obj xcc_mqd_mem_obj;
+	uint64_t xcc_gart_addr = 0;
+	uint64_t xcc_ctx_save_restore_area_address;
+	uint64_t offset = mm->mqd_stride(mm, q);
+	uint32_t local_xcc_start = mm->dev->dqm->current_logical_xcc_start++;
+
+	memset(&xcc_mqd_mem_obj, 0x0, sizeof(struct kfd_mem_obj));
+	for (xcc = 0; xcc < NUM_XCC(mm->dev->xcc_mask); xcc++) {
+		get_xcc_mqd(mqd_mem_obj, &xcc_mqd_mem_obj, offset*xcc);
+
+		init_mqd(mm, (void **)&m, &xcc_mqd_mem_obj, &xcc_gart_addr, q);
+		if (amdgpu_sriov_multi_vf_mode(mm->dev->adev))
+				m->cp_hqd_pq_doorbell_control |= 1 <<
+					CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_MODE__SHIFT;
+		m->cp_mqd_stride_size = offset;
+
+		/*
+		 * Update the CWSR address for each XCC if CWSR is enabled
+		 * and CWSR area is allocated in thunk
+		 */
+		if (mm->dev->kfd->cwsr_enabled &&
+		    q->ctx_save_restore_area_address) {
+			xcc_ctx_save_restore_area_address =
+				q->ctx_save_restore_area_address +
+				(xcc * q->ctx_save_restore_area_size);
+
+			m->cp_hqd_ctx_save_base_addr_lo =
+				lower_32_bits(xcc_ctx_save_restore_area_address);
+			m->cp_hqd_ctx_save_base_addr_hi =
+				upper_32_bits(xcc_ctx_save_restore_area_address);
+		}
+
+		if (q->format == KFD_QUEUE_FORMAT_AQL) {
+			m->compute_tg_chunk_size = 1;
+			m->compute_current_logic_xcc_id =
+					(local_xcc_start + xcc) %
+					NUM_XCC(mm->dev->xcc_mask);
+
+			switch (xcc) {
+			case 0:
+				/* Master XCC */
+				m->cp_hqd_pq_control &=
+					~CP_HQD_PQ_CONTROL__NO_UPDATE_RPTR_MASK;
+				break;
+			default:
+				break;
+			}
+		} else {
+			/* PM4 Queue */
+			m->compute_current_logic_xcc_id = 0;
+			m->compute_tg_chunk_size = 0;
+			m->pm4_target_xcc_in_xcp = q->pm4_target_xcc;
+		}
+
+		if (xcc == 0) {
+			/* Set the MQD pointer and gart address to XCC0 MQD */
+			*mqd = m;
+			*gart_addr = xcc_gart_addr;
+		}
+	}
+}
+
+static void update_mqd_v9_4_3(struct mqd_manager *mm, void *mqd,
+		      struct queue_properties *q, struct mqd_update_info *minfo)
+{
+	struct v9_mqd *m;
+	int xcc = 0;
+	uint64_t size = mm->mqd_stride(mm, q);
+
+	for (xcc = 0; xcc < NUM_XCC(mm->dev->xcc_mask); xcc++) {
+		m = get_mqd(mqd + size * xcc);
+		update_mqd(mm, m, q, minfo);
+
+		if (amdgpu_sriov_multi_vf_mode(mm->dev->adev))
+				m->cp_hqd_pq_doorbell_control |= 1 <<
+					CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_MODE__SHIFT;
+		update_cu_mask(mm, m, minfo, xcc);
+
+		if (q->format == KFD_QUEUE_FORMAT_AQL) {
+			switch (xcc) {
+			case 0:
+				/* Master XCC */
+				m->cp_hqd_pq_control &=
+					~CP_HQD_PQ_CONTROL__NO_UPDATE_RPTR_MASK;
+				break;
+			default:
+				break;
+			}
+			m->compute_tg_chunk_size = 1;
+		} else {
+			/* PM4 Queue */
+			m->compute_current_logic_xcc_id = 0;
+			m->compute_tg_chunk_size = 0;
+			m->pm4_target_xcc_in_xcp = q->pm4_target_xcc;
+		}
+	}
+}
+
+static void restore_mqd_v9_4_3(struct mqd_manager *mm, void **mqd,
+			struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+			struct queue_properties *qp,
+			const void *mqd_src,
+			const void *ctl_stack_src, u32 ctl_stack_size)
+{
+	struct kfd_mem_obj xcc_mqd_mem_obj;
+	u32 mqd_ctl_stack_size;
+	struct v9_mqd *m;
+	u32 num_xcc;
+	int xcc;
+
+	uint64_t offset = mm->mqd_stride(mm, qp);
+
+	mm->dev->dqm->current_logical_xcc_start++;
+
+	num_xcc = NUM_XCC(mm->dev->xcc_mask);
+	mqd_ctl_stack_size = ctl_stack_size / num_xcc;
+
+	memset(&xcc_mqd_mem_obj, 0x0, sizeof(struct kfd_mem_obj));
+
+	/* Set the MQD pointer and gart address to XCC0 MQD */
+	*mqd = mqd_mem_obj->cpu_ptr;
+	if (gart_addr)
+		*gart_addr = mqd_mem_obj->gpu_addr;
+
+	for (xcc = 0; xcc < num_xcc; xcc++) {
+		get_xcc_mqd(mqd_mem_obj, &xcc_mqd_mem_obj, offset * xcc);
+		restore_mqd(mm, (void **)&m,
+					&xcc_mqd_mem_obj,
+					NULL,
+					qp,
+					(uint8_t *)mqd_src + xcc * sizeof(*m),
+					(uint8_t *)ctl_stack_src + xcc *  mqd_ctl_stack_size,
+					mqd_ctl_stack_size);
+	}
+}
+static int destroy_mqd_v9_4_3(struct mqd_manager *mm, void *mqd,
+		   enum kfd_preempt_type type, unsigned int timeout,
+		   uint32_t pipe_id, uint32_t queue_id)
+{
+	uint32_t xcc_mask = mm->dev->xcc_mask;
+	int xcc_id, err, inst = 0;
+	void *xcc_mqd;
+	struct v9_mqd *m;
+	uint64_t mqd_offset;
+
+	m = get_mqd(mqd);
+	mqd_offset = m->cp_mqd_stride_size;
+
+	for_each_inst(xcc_id, xcc_mask) {
+		xcc_mqd = mqd + mqd_offset * inst;
+		err = mm->dev->kfd2kgd->hqd_destroy(mm->dev->adev, xcc_mqd,
+						    type, timeout, pipe_id,
+						    queue_id, xcc_id);
+		if (err) {
+			pr_debug("Destroy MQD failed for xcc: %d\n", inst);
+			break;
+		}
+		++inst;
+	}
+
+	return err;
+}
+
+static int load_mqd_v9_4_3(struct mqd_manager *mm, void *mqd,
+			uint32_t pipe_id, uint32_t queue_id,
+			struct queue_properties *p, struct mm_struct *mms)
+{
+	/* AQL write pointer counts in 64B packets, PM4/CP counts in dwords. */
+	uint32_t wptr_shift = (p->format == KFD_QUEUE_FORMAT_AQL ? 4 : 0);
+	uint32_t xcc_mask = mm->dev->xcc_mask;
+	int xcc_id, err, inst = 0;
+	void *xcc_mqd;
+	uint64_t mqd_stride_size = mm->mqd_stride(mm, p);
+
+	for_each_inst(xcc_id, xcc_mask) {
+		xcc_mqd = mqd + mqd_stride_size * inst;
+		err = mm->dev->kfd2kgd->hqd_load(
+			mm->dev->adev, xcc_mqd, pipe_id, queue_id,
+			(uint32_t __user *)p->write_ptr, wptr_shift, 0, mms,
+			xcc_id);
+		if (err) {
+			pr_debug("Load MQD failed for xcc: %d\n", inst);
+			break;
+		}
+		++inst;
+	}
+
+	return err;
+}
+
+static int get_wave_state_v9_4_3(struct mqd_manager *mm, void *mqd,
+				 struct queue_properties *q,
+				 void __user *ctl_stack,
+				 u32 *ctl_stack_used_size,
+				 u32 *save_area_used_size)
+{
+	int xcc, err = 0;
+	void *xcc_mqd;
+	void __user *xcc_ctl_stack;
+	uint64_t mqd_stride_size = mm->mqd_stride(mm, q);
+	u32 tmp_ctl_stack_used_size = 0, tmp_save_area_used_size = 0;
+
+	for (xcc = 0; xcc < NUM_XCC(mm->dev->xcc_mask); xcc++) {
+		xcc_mqd = mqd + mqd_stride_size * xcc;
+		xcc_ctl_stack = (void __user *)((uintptr_t)ctl_stack +
+					q->ctx_save_restore_area_size * xcc);
+
+		err = get_wave_state(mm, xcc_mqd, q, xcc_ctl_stack,
+				     &tmp_ctl_stack_used_size,
+				     &tmp_save_area_used_size);
+		if (err)
+			break;
+
+		/*
+		 * Set the ctl_stack_used_size and save_area_used_size to
+		 * ctl_stack_used_size and save_area_used_size of XCC 0 when
+		 * passing the info the user-space.
+		 * For multi XCC, user-space would have to look at the header
+		 * info of each Control stack area to determine the control
+		 * stack size and save area used.
+		 */
+		if (xcc == 0) {
+			*ctl_stack_used_size = tmp_ctl_stack_used_size;
+			*save_area_used_size = tmp_save_area_used_size;
+		}
+	}
+
+	return err;
+}
+
+#if defined(CONFIG_DEBUG_FS)
+
+static int debugfs_show_mqd(struct seq_file *m, void *data)
+{
+	seq_hex_dump(m, "    ", DUMP_PREFIX_OFFSET, 32, 4,
+		     data, sizeof(struct v9_mqd), false);
+	return 0;
+}
+
+static int debugfs_show_mqd_sdma(struct seq_file *m, void *data)
+{
+	seq_hex_dump(m, "    ", DUMP_PREFIX_OFFSET, 32, 4,
+		     data, sizeof(struct v9_sdma_mqd), false);
+	return 0;
+}
+
+#endif
+
+struct mqd_manager *mqd_manager_init_v9(enum KFD_MQD_TYPE type,
+		struct kfd_node *dev)
+{
+	struct mqd_manager *mqd;
+
+	if (WARN_ON(type >= KFD_MQD_TYPE_MAX))
+		return NULL;
+
+	mqd = kzalloc(sizeof(*mqd), GFP_KERNEL);
+	if (!mqd)
+		return NULL;
+
+	mqd->dev = dev;
+
+	switch (type) {
+	case KFD_MQD_TYPE_CP:
+		mqd->allocate_mqd = allocate_mqd;
+		mqd->free_mqd = kfd_free_mqd_cp;
+		mqd->is_occupied = kfd_is_occupied_cp;
+		mqd->get_checkpoint_info = get_checkpoint_info;
+		mqd->mqd_size = sizeof(struct v9_mqd);
+		mqd->mqd_stride = mqd_stride_v9;
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd;
+#endif
+		if (KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 3) ||
+		    KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 4) ||
+		    KFD_GC_VERSION(dev) == IP_VERSION(9, 5, 0)) {
+			mqd->init_mqd = init_mqd_v9_4_3;
+			mqd->load_mqd = load_mqd_v9_4_3;
+			mqd->update_mqd = update_mqd_v9_4_3;
+			mqd->destroy_mqd = destroy_mqd_v9_4_3;
+			mqd->get_wave_state = get_wave_state_v9_4_3;
+			mqd->checkpoint_mqd = checkpoint_mqd_v9_4_3;
+			mqd->restore_mqd = restore_mqd_v9_4_3;
+		} else {
+			mqd->init_mqd = init_mqd;
+			mqd->load_mqd = load_mqd;
+			mqd->update_mqd = update_mqd;
+			mqd->destroy_mqd = kfd_destroy_mqd_cp;
+			mqd->get_wave_state = get_wave_state;
+			mqd->checkpoint_mqd = checkpoint_mqd;
+			mqd->restore_mqd = restore_mqd;
+		}
+		break;
+	case KFD_MQD_TYPE_HIQ:
+		mqd->allocate_mqd = allocate_hiq_mqd;
+		mqd->free_mqd = free_mqd_hiq_sdma;
+		mqd->update_mqd = update_mqd;
+		mqd->is_occupied = kfd_is_occupied_cp;
+		mqd->mqd_size = sizeof(struct v9_mqd);
+		mqd->mqd_stride = kfd_mqd_stride;
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd;
+#endif
+		mqd->check_preemption_failed = check_preemption_failed;
+		if (KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 3) ||
+		    KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 4) ||
+		    KFD_GC_VERSION(dev) == IP_VERSION(9, 5, 0)) {
+			mqd->init_mqd = init_mqd_hiq_v9_4_3;
+			mqd->load_mqd = hiq_load_mqd_kiq_v9_4_3;
+			mqd->destroy_mqd = destroy_hiq_mqd_v9_4_3;
+			mqd->check_preemption_failed = check_preemption_failed_v9_4_3;
+		} else {
+			mqd->init_mqd = init_mqd_hiq;
+			mqd->load_mqd = kfd_hiq_load_mqd_kiq;
+			mqd->destroy_mqd = destroy_hiq_mqd;
+			mqd->check_preemption_failed = check_preemption_failed;
+		}
+		break;
+	case KFD_MQD_TYPE_DIQ:
+		mqd->allocate_mqd = allocate_mqd;
+		mqd->init_mqd = init_mqd_hiq;
+		mqd->free_mqd = kfd_free_mqd_cp;
+		mqd->load_mqd = load_mqd;
+		mqd->update_mqd = update_mqd;
+		mqd->destroy_mqd = kfd_destroy_mqd_cp;
+		mqd->is_occupied = kfd_is_occupied_cp;
+		mqd->mqd_size = sizeof(struct v9_mqd);
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd;
+#endif
+		break;
+	case KFD_MQD_TYPE_SDMA:
+		mqd->allocate_mqd = allocate_sdma_mqd;
+		mqd->init_mqd = init_mqd_sdma;
+		mqd->free_mqd = free_mqd_hiq_sdma;
+		mqd->load_mqd = kfd_load_mqd_sdma;
+		mqd->update_mqd = update_mqd_sdma;
+		mqd->destroy_mqd = kfd_destroy_mqd_sdma;
+		mqd->is_occupied = kfd_is_occupied_sdma;
+		mqd->checkpoint_mqd = checkpoint_mqd_sdma;
+		mqd->restore_mqd = restore_mqd_sdma;
+		mqd->mqd_size = sizeof(struct v9_sdma_mqd);
+		mqd->mqd_stride = kfd_mqd_stride;
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd_sdma;
+#endif
+		break;
+	default:
+		kfree(mqd);
+		return NULL;
+	}
+
+	return mqd;
+}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_vi.c b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_vi.c
index 73cbfe186dd2..c1fafc502515 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_vi.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_vi.c
@@ -1,5 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -30,6 +31,7 @@
 #include "vi_structs.h"
 #include "gca/gfx_8_0_sh_mask.h"
 #include "gca/gfx_8_0_enum.h"
+#include "oss/oss_3_0_sh_mask.h"
 
 #define CP_MQD_CONTROL__PRIV_STATE__SHIFT 0x8
 
@@ -38,21 +40,63 @@ static inline struct vi_mqd *get_mqd(void *mqd)
 	return (struct vi_mqd *)mqd;
 }
 
-static int init_mqd(struct mqd_manager *mm, void **mqd,
-			struct kfd_mem_obj **mqd_mem_obj, uint64_t *gart_addr,
+static inline struct vi_sdma_mqd *get_sdma_mqd(void *mqd)
+{
+	return (struct vi_sdma_mqd *)mqd;
+}
+
+static void update_cu_mask(struct mqd_manager *mm, void *mqd,
+			struct mqd_update_info *minfo)
+{
+	struct vi_mqd *m;
+	uint32_t se_mask[4] = {0}; /* 4 is the max # of SEs */
+
+	if (!minfo || !minfo->cu_mask.ptr)
+		return;
+
+	mqd_symmetrically_map_cu_mask(mm,
+		minfo->cu_mask.ptr, minfo->cu_mask.count, se_mask, 0);
+
+	m = get_mqd(mqd);
+	m->compute_static_thread_mgmt_se0 = se_mask[0];
+	m->compute_static_thread_mgmt_se1 = se_mask[1];
+	m->compute_static_thread_mgmt_se2 = se_mask[2];
+	m->compute_static_thread_mgmt_se3 = se_mask[3];
+
+	pr_debug("Update cu mask to %#x %#x %#x %#x\n",
+		m->compute_static_thread_mgmt_se0,
+		m->compute_static_thread_mgmt_se1,
+		m->compute_static_thread_mgmt_se2,
+		m->compute_static_thread_mgmt_se3);
+}
+
+static void set_priority(struct vi_mqd *m, struct queue_properties *q)
+{
+	m->cp_hqd_pipe_priority = pipe_priority_map[q->priority];
+	m->cp_hqd_queue_priority = q->priority;
+}
+
+static struct kfd_mem_obj *allocate_mqd(struct kfd_node *kfd,
+					struct queue_properties *q)
+{
+	struct kfd_mem_obj *mqd_mem_obj;
+
+	if (kfd_gtt_sa_allocate(kfd, sizeof(struct vi_mqd),
+			&mqd_mem_obj))
+		return NULL;
+
+	return mqd_mem_obj;
+}
+
+static void init_mqd(struct mqd_manager *mm, void **mqd,
+			struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
 			struct queue_properties *q)
 {
-	int retval;
 	uint64_t addr;
 	struct vi_mqd *m;
 
-	retval = kfd_gtt_sa_allocate(mm->dev, sizeof(struct vi_mqd),
-			mqd_mem_obj);
-	if (retval != 0)
-		return -ENOMEM;
-
-	m = (struct vi_mqd *) (*mqd_mem_obj)->cpu_ptr;
-	addr = (*mqd_mem_obj)->gpu_addr;
+	m = (struct vi_mqd *) mqd_mem_obj->cpu_ptr;
+	addr = mqd_mem_obj->gpu_addr;
 
 	memset(m, 0, sizeof(struct vi_mqd));
 
@@ -74,22 +118,40 @@ static int init_mqd(struct mqd_manager *mm, void **mqd,
 
 	m->cp_hqd_quantum = 1 << CP_HQD_QUANTUM__QUANTUM_EN__SHIFT |
 			1 << CP_HQD_QUANTUM__QUANTUM_SCALE__SHIFT |
-			10 << CP_HQD_QUANTUM__QUANTUM_DURATION__SHIFT;
-
-	m->cp_hqd_pipe_priority = 1;
-	m->cp_hqd_queue_priority = 15;
+			1 << CP_HQD_QUANTUM__QUANTUM_DURATION__SHIFT;
 
+	set_priority(m, q);
 	m->cp_hqd_eop_rptr = 1 << CP_HQD_EOP_RPTR__INIT_FETCHER__SHIFT;
 
 	if (q->format == KFD_QUEUE_FORMAT_AQL)
 		m->cp_hqd_iq_rptr = 1;
 
+	if (q->tba_addr) {
+		m->compute_tba_lo = lower_32_bits(q->tba_addr >> 8);
+		m->compute_tba_hi = upper_32_bits(q->tba_addr >> 8);
+		m->compute_tma_lo = lower_32_bits(q->tma_addr >> 8);
+		m->compute_tma_hi = upper_32_bits(q->tma_addr >> 8);
+		m->compute_pgm_rsrc2 |=
+			(1 << COMPUTE_PGM_RSRC2__TRAP_PRESENT__SHIFT);
+	}
+
+	if (mm->dev->kfd->cwsr_enabled && q->ctx_save_restore_area_address) {
+		m->cp_hqd_persistent_state |=
+			(1 << CP_HQD_PERSISTENT_STATE__QSWITCH_MODE__SHIFT);
+		m->cp_hqd_ctx_save_base_addr_lo =
+			lower_32_bits(q->ctx_save_restore_area_address);
+		m->cp_hqd_ctx_save_base_addr_hi =
+			upper_32_bits(q->ctx_save_restore_area_address);
+		m->cp_hqd_ctx_save_size = q->ctx_save_restore_area_size;
+		m->cp_hqd_cntl_stack_size = q->ctl_stack_size;
+		m->cp_hqd_cntl_stack_offset = q->ctl_stack_size;
+		m->cp_hqd_wg_state_offset = q->ctl_stack_size;
+	}
+
 	*mqd = m;
 	if (gart_addr)
 		*gart_addr = addr;
-	retval = mm->update_mqd(mm, m, q);
-
-	return retval;
+	mm->update_mqd(mm, m, q, NULL);
 }
 
 static int load_mqd(struct mqd_manager *mm, void *mqd,
@@ -98,16 +160,16 @@ static int load_mqd(struct mqd_manager *mm, void *mqd,
 {
 	/* AQL write pointer counts in 64B packets, PM4/CP counts in dwords. */
 	uint32_t wptr_shift = (p->format == KFD_QUEUE_FORMAT_AQL ? 4 : 0);
-	uint32_t wptr_mask = (uint32_t)((p->queue_size / sizeof(uint32_t)) - 1);
+	uint32_t wptr_mask = (uint32_t)((p->queue_size / 4) - 1);
 
-	return mm->dev->kfd2kgd->hqd_load(mm->dev->kgd, mqd, pipe_id, queue_id,
+	return mm->dev->kfd2kgd->hqd_load(mm->dev->adev, mqd, pipe_id, queue_id,
 					  (uint32_t __user *)p->write_ptr,
-					  wptr_shift, wptr_mask, mms);
+					  wptr_shift, wptr_mask, mms, 0);
 }
 
-static int __update_mqd(struct mqd_manager *mm, void *mqd,
-			struct queue_properties *q, unsigned int mtype,
-			unsigned int atc_bit)
+static void __update_mqd(struct mqd_manager *mm, void *mqd,
+			struct queue_properties *q, struct mqd_update_info *minfo,
+			unsigned int mtype, unsigned int atc_bit)
 {
 	struct vi_mqd *m;
 
@@ -116,8 +178,7 @@ static int __update_mqd(struct mqd_manager *mm, void *mqd,
 	m->cp_hqd_pq_control = 5 << CP_HQD_PQ_CONTROL__RPTR_BLOCK_SIZE__SHIFT |
 			atc_bit << CP_HQD_PQ_CONTROL__PQ_ATC__SHIFT |
 			mtype << CP_HQD_PQ_CONTROL__MTYPE__SHIFT;
-	m->cp_hqd_pq_control |=
-			ffs(q->queue_size / sizeof(unsigned int)) - 1 - 1;
+	m->cp_hqd_pq_control |=	order_base_2(q->queue_size / 4) - 1;
 	pr_debug("cp_hqd_pq_control 0x%x\n", m->cp_hqd_pq_control);
 
 	m->cp_hqd_pq_base_lo = lower_32_bits((uint64_t)q->queue_address >> 8);
@@ -125,6 +186,8 @@ static int __update_mqd(struct mqd_manager *mm, void *mqd,
 
 	m->cp_hqd_pq_rptr_report_addr_lo = lower_32_bits((uint64_t)q->read_ptr);
 	m->cp_hqd_pq_rptr_report_addr_hi = upper_32_bits((uint64_t)q->read_ptr);
+	m->cp_hqd_pq_wptr_poll_addr_lo = lower_32_bits((uint64_t)q->write_ptr);
+	m->cp_hqd_pq_wptr_poll_addr_hi = upper_32_bits((uint64_t)q->write_ptr);
 
 	m->cp_hqd_pq_doorbell_control =
 		q->doorbell_off <<
@@ -147,7 +210,7 @@ static int __update_mqd(struct mqd_manager *mm, void *mqd,
 	 * is safe, giving a maximum field value of 0xA.
 	 */
 	m->cp_hqd_eop_control |= min(0xA,
-		ffs(q->eop_ring_buffer_size / sizeof(unsigned int)) - 1 - 1);
+		order_base_2(q->eop_ring_buffer_size / 4) - 1);
 	m->cp_hqd_eop_base_addr_lo =
 			lower_32_bits(q->eop_ring_buffer_address >> 8);
 	m->cp_hqd_eop_base_addr_hi =
@@ -163,82 +226,219 @@ static int __update_mqd(struct mqd_manager *mm, void *mqd,
 				2 << CP_HQD_PQ_CONTROL__SLOT_BASED_WPTR__SHIFT;
 	}
 
-	q->is_active = false;
-	if (q->queue_size > 0 &&
-			q->queue_address != 0 &&
-			q->queue_percent > 0) {
-		q->is_active = true;
-	}
+	if (mm->dev->kfd->cwsr_enabled && q->ctx_save_restore_area_address)
+		m->cp_hqd_ctx_save_control =
+			atc_bit << CP_HQD_CTX_SAVE_CONTROL__ATC__SHIFT |
+			mtype << CP_HQD_CTX_SAVE_CONTROL__MTYPE__SHIFT;
 
-	return 0;
+	update_cu_mask(mm, mqd, minfo);
+	set_priority(m, q);
+
+	q->is_active = QUEUE_IS_ACTIVE(*q);
 }
 
+static bool check_preemption_failed(struct mqd_manager *mm, void *mqd)
+{
+	struct vi_mqd *m = (struct vi_mqd *)mqd;
 
-static int update_mqd(struct mqd_manager *mm, void *mqd,
-			struct queue_properties *q)
+	return kfd_check_hiq_mqd_doorbell_id(mm->dev, m->queue_doorbell_id0, 0);
+}
+
+static void update_mqd(struct mqd_manager *mm, void *mqd,
+		       struct queue_properties *q,
+		       struct mqd_update_info *minfo)
 {
-	return __update_mqd(mm, mqd, q, MTYPE_CC, 1);
+	__update_mqd(mm, mqd, q, minfo, MTYPE_UC, 0);
 }
 
-static int destroy_mqd(struct mqd_manager *mm, void *mqd,
-			enum kfd_preempt_type type,
-			unsigned int timeout, uint32_t pipe_id,
-			uint32_t queue_id)
+static int get_wave_state(struct mqd_manager *mm, void *mqd,
+			  struct queue_properties *q,
+			  void __user *ctl_stack,
+			  u32 *ctl_stack_used_size,
+			  u32 *save_area_used_size)
 {
-	return mm->dev->kfd2kgd->hqd_destroy
-		(mm->dev->kgd, mqd, type, timeout,
-		pipe_id, queue_id);
+	struct vi_mqd *m;
+
+	m = get_mqd(mqd);
+
+	*ctl_stack_used_size = m->cp_hqd_cntl_stack_size -
+		m->cp_hqd_cntl_stack_offset;
+	*save_area_used_size = m->cp_hqd_wg_state_offset -
+		m->cp_hqd_cntl_stack_size;
+
+	/* Control stack is not copied to user mode for GFXv8 because
+	 * it's part of the context save area that is already
+	 * accessible to user mode
+	 */
+
+	return 0;
 }
 
-static void uninit_mqd(struct mqd_manager *mm, void *mqd,
-			struct kfd_mem_obj *mqd_mem_obj)
+static void get_checkpoint_info(struct mqd_manager *mm, void *mqd, u32 *ctl_stack_size)
 {
-	kfd_gtt_sa_free(mm->dev, mqd_mem_obj);
+	/* Control stack is stored in user mode */
+	*ctl_stack_size = 0;
 }
 
-static bool is_occupied(struct mqd_manager *mm, void *mqd,
-			uint64_t queue_address,	uint32_t pipe_id,
-			uint32_t queue_id)
+static void checkpoint_mqd(struct mqd_manager *mm, void *mqd, void *mqd_dst, void *ctl_stack_dst)
 {
-	return mm->dev->kfd2kgd->hqd_is_occupied(
-		mm->dev->kgd, queue_address,
-		pipe_id, queue_id);
+	struct vi_mqd *m;
+
+	m = get_mqd(mqd);
+
+	memcpy(mqd_dst, m, sizeof(struct vi_mqd));
 }
 
-static int init_mqd_hiq(struct mqd_manager *mm, void **mqd,
-			struct kfd_mem_obj **mqd_mem_obj, uint64_t *gart_addr,
+static void restore_mqd(struct mqd_manager *mm, void **mqd,
+			struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+			struct queue_properties *qp,
+			const void *mqd_src,
+			const void *ctl_stack_src, const u32 ctl_stack_size)
+{
+	uint64_t addr;
+	struct vi_mqd *m;
+
+	m = (struct vi_mqd *) mqd_mem_obj->cpu_ptr;
+	addr = mqd_mem_obj->gpu_addr;
+
+	memcpy(m, mqd_src, sizeof(*m));
+
+	*mqd = m;
+	if (gart_addr)
+		*gart_addr = addr;
+
+	m->cp_hqd_pq_doorbell_control =
+		qp->doorbell_off <<
+			CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_OFFSET__SHIFT;
+	pr_debug("cp_hqd_pq_doorbell_control 0x%x\n",
+			m->cp_hqd_pq_doorbell_control);
+
+	qp->is_active = 0;
+}
+
+static void init_mqd_hiq(struct mqd_manager *mm, void **mqd,
+			struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
 			struct queue_properties *q)
 {
 	struct vi_mqd *m;
-	int retval = init_mqd(mm, mqd, mqd_mem_obj, gart_addr, q);
 
-	if (retval != 0)
-		return retval;
+	init_mqd(mm, mqd, mqd_mem_obj, gart_addr, q);
 
 	m = get_mqd(*mqd);
 
 	m->cp_hqd_pq_control |= 1 << CP_HQD_PQ_CONTROL__PRIV_STATE__SHIFT |
 			1 << CP_HQD_PQ_CONTROL__KMD_QUEUE__SHIFT;
+}
 
-	return retval;
+static void update_mqd_hiq(struct mqd_manager *mm, void *mqd,
+			struct queue_properties *q,
+			struct mqd_update_info *minfo)
+{
+	__update_mqd(mm, mqd, q, minfo, MTYPE_UC, 0);
 }
 
-static int update_mqd_hiq(struct mqd_manager *mm, void *mqd,
-			struct queue_properties *q)
+static void init_mqd_sdma(struct mqd_manager *mm, void **mqd,
+		struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+		struct queue_properties *q)
 {
-	struct vi_mqd *m;
-	int retval = __update_mqd(mm, mqd, q, MTYPE_UC, 0);
+	struct vi_sdma_mqd *m;
 
-	if (retval != 0)
-		return retval;
+	m = (struct vi_sdma_mqd *) mqd_mem_obj->cpu_ptr;
 
-	m = get_mqd(mqd);
-	m->cp_hqd_vmid = q->vmid;
-	return retval;
+	memset(m, 0, sizeof(struct vi_sdma_mqd));
+
+	*mqd = m;
+	if (gart_addr)
+		*gart_addr = mqd_mem_obj->gpu_addr;
+
+	mm->update_mqd(mm, m, q, NULL);
+}
+
+static void update_mqd_sdma(struct mqd_manager *mm, void *mqd,
+			struct queue_properties *q,
+			struct mqd_update_info *minfo)
+{
+	struct vi_sdma_mqd *m;
+
+	m = get_sdma_mqd(mqd);
+	m->sdmax_rlcx_rb_cntl = order_base_2(q->queue_size / 4)
+		<< SDMA0_RLC0_RB_CNTL__RB_SIZE__SHIFT |
+		q->vmid << SDMA0_RLC0_RB_CNTL__RB_VMID__SHIFT |
+		1 << SDMA0_RLC0_RB_CNTL__RPTR_WRITEBACK_ENABLE__SHIFT |
+		6 << SDMA0_RLC0_RB_CNTL__RPTR_WRITEBACK_TIMER__SHIFT;
+
+	m->sdmax_rlcx_rb_base = lower_32_bits(q->queue_address >> 8);
+	m->sdmax_rlcx_rb_base_hi = upper_32_bits(q->queue_address >> 8);
+	m->sdmax_rlcx_rb_rptr_addr_lo = lower_32_bits((uint64_t)q->read_ptr);
+	m->sdmax_rlcx_rb_rptr_addr_hi = upper_32_bits((uint64_t)q->read_ptr);
+	m->sdmax_rlcx_doorbell =
+		q->doorbell_off << SDMA0_RLC0_DOORBELL__OFFSET__SHIFT;
+
+	m->sdmax_rlcx_virtual_addr = q->sdma_vm_addr;
+
+	m->sdma_engine_id = q->sdma_engine_id;
+	m->sdma_queue_id = q->sdma_queue_id;
+
+	q->is_active = QUEUE_IS_ACTIVE(*q);
+}
+
+static void checkpoint_mqd_sdma(struct mqd_manager *mm,
+				void *mqd,
+				void *mqd_dst,
+				void *ctl_stack_dst)
+{
+	struct vi_sdma_mqd *m;
+
+	m = get_sdma_mqd(mqd);
+
+	memcpy(mqd_dst, m, sizeof(struct vi_sdma_mqd));
+}
+
+static void restore_mqd_sdma(struct mqd_manager *mm, void **mqd,
+			     struct kfd_mem_obj *mqd_mem_obj, uint64_t *gart_addr,
+			     struct queue_properties *qp,
+			     const void *mqd_src,
+			     const void *ctl_stack_src, const u32 ctl_stack_size)
+{
+	uint64_t addr;
+	struct vi_sdma_mqd *m;
+
+	m = (struct vi_sdma_mqd *) mqd_mem_obj->cpu_ptr;
+	addr = mqd_mem_obj->gpu_addr;
+
+	memcpy(m, mqd_src, sizeof(*m));
+
+	m->sdmax_rlcx_doorbell =
+		qp->doorbell_off << SDMA0_RLC0_DOORBELL__OFFSET__SHIFT;
+
+	*mqd = m;
+	if (gart_addr)
+		*gart_addr = addr;
+
+	qp->is_active = 0;
+}
+
+#if defined(CONFIG_DEBUG_FS)
+
+
+static int debugfs_show_mqd(struct seq_file *m, void *data)
+{
+	seq_hex_dump(m, "    ", DUMP_PREFIX_OFFSET, 32, 4,
+		     data, sizeof(struct vi_mqd), false);
+	return 0;
+}
+
+static int debugfs_show_mqd_sdma(struct seq_file *m, void *data)
+{
+	seq_hex_dump(m, "    ", DUMP_PREFIX_OFFSET, 32, 4,
+		     data, sizeof(struct vi_sdma_mqd), false);
+	return 0;
 }
 
+#endif
+
 struct mqd_manager *mqd_manager_init_vi(enum KFD_MQD_TYPE type,
-		struct kfd_dev *dev)
+		struct kfd_node *dev)
 {
 	struct mqd_manager *mqd;
 
@@ -253,23 +453,66 @@ struct mqd_manager *mqd_manager_init_vi(enum KFD_MQD_TYPE type,
 
 	switch (type) {
 	case KFD_MQD_TYPE_CP:
-	case KFD_MQD_TYPE_COMPUTE:
+		mqd->allocate_mqd = allocate_mqd;
 		mqd->init_mqd = init_mqd;
-		mqd->uninit_mqd = uninit_mqd;
+		mqd->free_mqd = kfd_free_mqd_cp;
 		mqd->load_mqd = load_mqd;
 		mqd->update_mqd = update_mqd;
-		mqd->destroy_mqd = destroy_mqd;
-		mqd->is_occupied = is_occupied;
+		mqd->destroy_mqd = kfd_destroy_mqd_cp;
+		mqd->is_occupied = kfd_is_occupied_cp;
+		mqd->get_wave_state = get_wave_state;
+		mqd->get_checkpoint_info = get_checkpoint_info;
+		mqd->checkpoint_mqd = checkpoint_mqd;
+		mqd->restore_mqd = restore_mqd;
+		mqd->mqd_size = sizeof(struct vi_mqd);
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd;
+#endif
 		break;
 	case KFD_MQD_TYPE_HIQ:
+		mqd->allocate_mqd = allocate_hiq_mqd;
+		mqd->init_mqd = init_mqd_hiq;
+		mqd->free_mqd = free_mqd_hiq_sdma;
+		mqd->load_mqd = load_mqd;
+		mqd->update_mqd = update_mqd_hiq;
+		mqd->destroy_mqd = kfd_destroy_mqd_cp;
+		mqd->is_occupied = kfd_is_occupied_cp;
+		mqd->mqd_size = sizeof(struct vi_mqd);
+		mqd->mqd_stride = kfd_mqd_stride;
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd;
+#endif
+		mqd->check_preemption_failed = check_preemption_failed;
+		break;
+	case KFD_MQD_TYPE_DIQ:
+		mqd->allocate_mqd = allocate_mqd;
 		mqd->init_mqd = init_mqd_hiq;
-		mqd->uninit_mqd = uninit_mqd;
+		mqd->free_mqd = kfd_free_mqd_cp;
 		mqd->load_mqd = load_mqd;
 		mqd->update_mqd = update_mqd_hiq;
-		mqd->destroy_mqd = destroy_mqd;
-		mqd->is_occupied = is_occupied;
+		mqd->destroy_mqd = kfd_destroy_mqd_cp;
+		mqd->is_occupied = kfd_is_occupied_cp;
+		mqd->mqd_size = sizeof(struct vi_mqd);
+		mqd->mqd_stride = kfd_mqd_stride;
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd;
+#endif
 		break;
 	case KFD_MQD_TYPE_SDMA:
+		mqd->allocate_mqd = allocate_sdma_mqd;
+		mqd->init_mqd = init_mqd_sdma;
+		mqd->free_mqd = free_mqd_hiq_sdma;
+		mqd->load_mqd = kfd_load_mqd_sdma;
+		mqd->update_mqd = update_mqd_sdma;
+		mqd->destroy_mqd = kfd_destroy_mqd_sdma;
+		mqd->is_occupied = kfd_is_occupied_sdma;
+		mqd->checkpoint_mqd = checkpoint_mqd_sdma;
+		mqd->restore_mqd = restore_mqd_sdma;
+		mqd->mqd_size = sizeof(struct vi_sdma_mqd);
+		mqd->mqd_stride = kfd_mqd_stride;
+#if defined(CONFIG_DEBUG_FS)
+		mqd->debugfs_show_mqd = debugfs_show_mqd_sdma;
+#endif
 		break;
 	default:
 		kfree(mqd);
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c
index 1d312603de9f..b1a6eb349bb3 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c
@@ -1,5 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -26,8 +27,11 @@
 #include "kfd_device_queue_manager.h"
 #include "kfd_kernel_queue.h"
 #include "kfd_priv.h"
-#include "kfd_pm4_headers_vi.h"
-#include "kfd_pm4_opcodes.h"
+
+#define OVER_SUBSCRIPTION_PROCESS_COUNT (1 << 0)
+#define OVER_SUBSCRIPTION_COMPUTE_QUEUE_COUNT (1 << 1)
+#define OVER_SUBSCRIPTION_GWS_QUEUE_COUNT (1 << 2)
+#define OVER_SUBSCRIPTION_XNACK_CONFLICT (1 << 3)
 
 static inline void inc_wptr(unsigned int *wptr, unsigned int increment_bytes,
 				unsigned int buffer_size_bytes)
@@ -39,38 +43,47 @@ static inline void inc_wptr(unsigned int *wptr, unsigned int increment_bytes,
 	*wptr = temp;
 }
 
-static unsigned int build_pm4_header(unsigned int opcode, size_t packet_size)
-{
-	union PM4_MES_TYPE_3_HEADER header;
-
-	header.u32All = 0;
-	header.opcode = opcode;
-	header.count = packet_size/sizeof(uint32_t) - 2;
-	header.type = PM4_TYPE_3;
-
-	return header.u32All;
-}
-
 static void pm_calc_rlib_size(struct packet_manager *pm,
 				unsigned int *rlib_size,
-				bool *over_subscription)
+				int *over_subscription,
+				int xnack_conflict)
 {
-	unsigned int process_count, queue_count;
+	unsigned int process_count, queue_count, compute_queue_count, gws_queue_count;
 	unsigned int map_queue_size;
+	unsigned int max_proc_per_quantum = 1;
+	struct kfd_node *node = pm->dqm->dev;
+	struct device *dev = node->adev->dev;
 
 	process_count = pm->dqm->processes_count;
-	queue_count = pm->dqm->queue_count;
+	queue_count = pm->dqm->active_queue_count;
+	compute_queue_count = pm->dqm->active_cp_queue_count;
+	gws_queue_count = pm->dqm->gws_queue_count;
+
+	/* check if there is over subscription
+	 * Note: the arbitration between the number of VMIDs and
+	 * hws_max_conc_proc has been done in
+	 * kgd2kfd_device_init().
+	 */
+	*over_subscription = 0;
 
-	/* check if there is over subscription*/
-	*over_subscription = false;
-	if ((process_count > 1) || queue_count > get_queues_num(pm->dqm)) {
-		*over_subscription = true;
-		pr_debug("Over subscribed runlist\n");
-	}
+	if (node->max_proc_per_quantum > 1)
+		max_proc_per_quantum = node->max_proc_per_quantum;
+
+	if (process_count > max_proc_per_quantum)
+		*over_subscription |= OVER_SUBSCRIPTION_PROCESS_COUNT;
+	if (compute_queue_count > get_cp_queues_num(pm->dqm))
+		*over_subscription |= OVER_SUBSCRIPTION_COMPUTE_QUEUE_COUNT;
+	if (gws_queue_count > 1)
+		*over_subscription |= OVER_SUBSCRIPTION_GWS_QUEUE_COUNT;
+	if (xnack_conflict && (node->adev->gmc.xnack_flags & AMDGPU_GMC_XNACK_FLAG_CHAIN))
+		*over_subscription |= OVER_SUBSCRIPTION_XNACK_CONFLICT;
 
-	map_queue_size = sizeof(struct pm4_mes_map_queues);
+	if (*over_subscription)
+		dev_dbg(dev, "Over subscribed runlist\n");
+
+	map_queue_size = pm->pmf->map_queues_size;
 	/* calculate run list ib allocation size */
-	*rlib_size = process_count * sizeof(struct pm4_mes_map_process) +
+	*rlib_size = process_count * pm->pmf->map_process_size +
 		     queue_count * map_queue_size;
 
 	/*
@@ -78,30 +91,35 @@ static void pm_calc_rlib_size(struct packet_manager *pm,
 	 * when over subscription
 	 */
 	if (*over_subscription)
-		*rlib_size += sizeof(struct pm4_mes_runlist);
+		*rlib_size += pm->pmf->runlist_size;
 
-	pr_debug("runlist ib size %d\n", *rlib_size);
+	dev_dbg(dev, "runlist ib size %d\n", *rlib_size);
 }
 
 static int pm_allocate_runlist_ib(struct packet_manager *pm,
 				unsigned int **rl_buffer,
 				uint64_t *rl_gpu_buffer,
 				unsigned int *rl_buffer_size,
-				bool *is_over_subscription)
+				int *is_over_subscription,
+				int xnack_conflict)
 {
+	struct kfd_node *node = pm->dqm->dev;
+	struct device *dev = node->adev->dev;
 	int retval;
 
 	if (WARN_ON(pm->allocated))
 		return -EINVAL;
 
-	pm_calc_rlib_size(pm, rl_buffer_size, is_over_subscription);
+	pm_calc_rlib_size(pm, rl_buffer_size, is_over_subscription,
+				xnack_conflict);
 
-	retval = kfd_gtt_sa_allocate(pm->dqm->dev, *rl_buffer_size,
-					&pm->ib_buffer_obj);
+	mutex_lock(&pm->lock);
+
+	retval = kfd_gtt_sa_allocate(node, *rl_buffer_size, &pm->ib_buffer_obj);
 
 	if (retval) {
-		pr_err("Failed to allocate runlist IB\n");
-		return retval;
+		dev_err(dev, "Failed to allocate runlist IB\n");
+		goto out;
 	}
 
 	*(void **)rl_buffer = pm->ib_buffer_obj->cpu_ptr;
@@ -109,129 +127,10 @@ static int pm_allocate_runlist_ib(struct packet_manager *pm,
 
 	memset(*rl_buffer, 0, *rl_buffer_size);
 	pm->allocated = true;
-	return retval;
-}
-
-static int pm_create_runlist(struct packet_manager *pm, uint32_t *buffer,
-			uint64_t ib, size_t ib_size_in_dwords, bool chain)
-{
-	struct pm4_mes_runlist *packet;
-
-	if (WARN_ON(!ib))
-		return -EFAULT;
-
-	packet = (struct pm4_mes_runlist *)buffer;
-
-	memset(buffer, 0, sizeof(struct pm4_mes_runlist));
-	packet->header.u32All = build_pm4_header(IT_RUN_LIST,
-						sizeof(struct pm4_mes_runlist));
-
-	packet->bitfields4.ib_size = ib_size_in_dwords;
-	packet->bitfields4.chain = chain ? 1 : 0;
-	packet->bitfields4.offload_polling = 0;
-	packet->bitfields4.valid = 1;
-	packet->ordinal2 = lower_32_bits(ib);
-	packet->bitfields3.ib_base_hi = upper_32_bits(ib);
-
-	return 0;
-}
 
-static int pm_create_map_process(struct packet_manager *pm, uint32_t *buffer,
-				struct qcm_process_device *qpd)
-{
-	struct pm4_mes_map_process *packet;
-	struct queue *cur;
-	uint32_t num_queues;
-
-	packet = (struct pm4_mes_map_process *)buffer;
-
-	memset(buffer, 0, sizeof(struct pm4_mes_map_process));
-
-	packet->header.u32All = build_pm4_header(IT_MAP_PROCESS,
-					sizeof(struct pm4_mes_map_process));
-	packet->bitfields2.diq_enable = (qpd->is_debug) ? 1 : 0;
-	packet->bitfields2.process_quantum = 1;
-	packet->bitfields2.pasid = qpd->pqm->process->pasid;
-	packet->bitfields3.page_table_base = qpd->page_table_base;
-	packet->bitfields10.gds_size = qpd->gds_size;
-	packet->bitfields10.num_gws = qpd->num_gws;
-	packet->bitfields10.num_oac = qpd->num_oac;
-	num_queues = 0;
-	list_for_each_entry(cur, &qpd->queues_list, list)
-		num_queues++;
-	packet->bitfields10.num_queues = (qpd->is_debug) ? 0 : num_queues;
-
-	packet->sh_mem_config = qpd->sh_mem_config;
-	packet->sh_mem_bases = qpd->sh_mem_bases;
-	packet->sh_mem_ape1_base = qpd->sh_mem_ape1_base;
-	packet->sh_mem_ape1_limit = qpd->sh_mem_ape1_limit;
-
-	/* TODO: scratch support */
-	packet->sh_hidden_private_base_vmid = 0;
-
-	packet->gds_addr_lo = lower_32_bits(qpd->gds_context_area);
-	packet->gds_addr_hi = upper_32_bits(qpd->gds_context_area);
-
-	return 0;
-}
-
-static int pm_create_map_queue(struct packet_manager *pm, uint32_t *buffer,
-		struct queue *q, bool is_static)
-{
-	struct pm4_mes_map_queues *packet;
-	bool use_static = is_static;
-
-	packet = (struct pm4_mes_map_queues *)buffer;
-	memset(buffer, 0, sizeof(struct pm4_mes_map_queues));
-
-	packet->header.u32All = build_pm4_header(IT_MAP_QUEUES,
-						sizeof(struct pm4_mes_map_queues));
-	packet->bitfields2.alloc_format =
-		alloc_format__mes_map_queues__one_per_pipe_vi;
-	packet->bitfields2.num_queues = 1;
-	packet->bitfields2.queue_sel =
-		queue_sel__mes_map_queues__map_to_hws_determined_queue_slots_vi;
-
-	packet->bitfields2.engine_sel =
-		engine_sel__mes_map_queues__compute_vi;
-	packet->bitfields2.queue_type =
-		queue_type__mes_map_queues__normal_compute_vi;
-
-	switch (q->properties.type) {
-	case KFD_QUEUE_TYPE_COMPUTE:
-		if (use_static)
-			packet->bitfields2.queue_type =
-		queue_type__mes_map_queues__normal_latency_static_queue_vi;
-		break;
-	case KFD_QUEUE_TYPE_DIQ:
-		packet->bitfields2.queue_type =
-			queue_type__mes_map_queues__debug_interface_queue_vi;
-		break;
-	case KFD_QUEUE_TYPE_SDMA:
-		packet->bitfields2.engine_sel =
-				engine_sel__mes_map_queues__sdma0_vi;
-		use_static = false; /* no static queues under SDMA */
-		break;
-	default:
-		WARN(1, "queue type %d", q->properties.type);
-		return -EINVAL;
-	}
-	packet->bitfields3.doorbell_offset =
-			q->properties.doorbell_off;
-
-	packet->mqd_addr_lo =
-			lower_32_bits(q->gart_mqd_addr);
-
-	packet->mqd_addr_hi =
-			upper_32_bits(q->gart_mqd_addr);
-
-	packet->wptr_addr_lo =
-			lower_32_bits((uint64_t)q->properties.write_ptr);
-
-	packet->wptr_addr_hi =
-			upper_32_bits((uint64_t)q->properties.write_ptr);
-
-	return 0;
+out:
+	mutex_unlock(&pm->lock);
+	return retval;
 }
 
 static int pm_create_runlist_ib(struct packet_manager *pm,
@@ -241,51 +140,75 @@ static int pm_create_runlist_ib(struct packet_manager *pm,
 {
 	unsigned int alloc_size_bytes;
 	unsigned int *rl_buffer, rl_wptr, i;
-	int retval, proccesses_mapped;
+	struct kfd_node *node = pm->dqm->dev;
+	struct device *dev = node->adev->dev;
+	int retval, processes_mapped;
 	struct device_process_node *cur;
 	struct qcm_process_device *qpd;
 	struct queue *q;
 	struct kernel_queue *kq;
-	bool is_over_subscription;
+	int is_over_subscription;
+	int xnack_enabled = -1;
+	bool xnack_conflict = 0;
 
-	rl_wptr = retval = proccesses_mapped = 0;
+	rl_wptr = retval = processes_mapped = 0;
+
+	/* Check if processes set different xnack modes */
+	list_for_each_entry(cur, queues, list) {
+		qpd = cur->qpd;
+		if (xnack_enabled < 0)
+			/* First process */
+			xnack_enabled = qpd->pqm->process->xnack_enabled;
+		else if (qpd->pqm->process->xnack_enabled != xnack_enabled) {
+			/* Found a process with a different xnack mode */
+			xnack_conflict = 1;
+			break;
+		}
+	}
 
 	retval = pm_allocate_runlist_ib(pm, &rl_buffer, rl_gpu_addr,
-				&alloc_size_bytes, &is_over_subscription);
+				&alloc_size_bytes, &is_over_subscription,
+				xnack_conflict);
 	if (retval)
 		return retval;
 
 	*rl_size_bytes = alloc_size_bytes;
+	pm->ib_size_bytes = alloc_size_bytes;
 
-	pr_debug("Building runlist ib process count: %d queues count %d\n",
-		pm->dqm->processes_count, pm->dqm->queue_count);
+	dev_dbg(dev, "Building runlist ib process count: %d queues count %d\n",
+		pm->dqm->processes_count, pm->dqm->active_queue_count);
 
+build_runlist_ib:
 	/* build the run list ib packet */
 	list_for_each_entry(cur, queues, list) {
 		qpd = cur->qpd;
+		/* group processes with the same xnack mode together */
+		if (qpd->pqm->process->xnack_enabled != xnack_enabled)
+			continue;
 		/* build map process packet */
-		if (proccesses_mapped >= pm->dqm->processes_count) {
-			pr_debug("Not enough space left in runlist IB\n");
+		if (processes_mapped >= pm->dqm->processes_count) {
+			dev_dbg(dev, "Not enough space left in runlist IB\n");
 			pm_release_ib(pm);
 			return -ENOMEM;
 		}
 
-		retval = pm_create_map_process(pm, &rl_buffer[rl_wptr], qpd);
+		retval = pm->pmf->map_process(pm, &rl_buffer[rl_wptr], qpd);
 		if (retval)
 			return retval;
 
-		proccesses_mapped++;
-		inc_wptr(&rl_wptr, sizeof(struct pm4_mes_map_process),
+		processes_mapped++;
+		inc_wptr(&rl_wptr, pm->pmf->map_process_size,
 				alloc_size_bytes);
 
 		list_for_each_entry(kq, &qpd->priv_queue_list, list) {
 			if (!kq->queue->properties.is_active)
 				continue;
 
-			pr_debug("static_queue, mapping kernel q %d, is debug status %d\n",
+			dev_dbg(dev,
+				"static_queue, mapping kernel q %d, is debug status %d\n",
 				kq->queue->queue, qpd->is_debug);
 
-			retval = pm_create_map_queue(pm,
+			retval = pm->pmf->map_queues(pm,
 						&rl_buffer[rl_wptr],
 						kq->queue,
 						qpd->is_debug);
@@ -293,7 +216,7 @@ static int pm_create_runlist_ib(struct packet_manager *pm,
 				return retval;
 
 			inc_wptr(&rl_wptr,
-				sizeof(struct pm4_mes_map_queues),
+				pm->pmf->map_queues_size,
 				alloc_size_bytes);
 		}
 
@@ -301,10 +224,11 @@ static int pm_create_runlist_ib(struct packet_manager *pm,
 			if (!q->properties.is_active)
 				continue;
 
-			pr_debug("static_queue, mapping user queue %d, is debug status %d\n",
+			dev_dbg(dev,
+				"static_queue, mapping user queue %d, is debug status %d\n",
 				q->queue, qpd->is_debug);
 
-			retval = pm_create_map_queue(pm,
+			retval = pm->pmf->map_queues(pm,
 						&rl_buffer[rl_wptr],
 						q,
 						qpd->is_debug);
@@ -313,18 +237,37 @@ static int pm_create_runlist_ib(struct packet_manager *pm,
 				return retval;
 
 			inc_wptr(&rl_wptr,
-				sizeof(struct pm4_mes_map_queues),
+				pm->pmf->map_queues_size,
 				alloc_size_bytes);
 		}
 	}
+	if (xnack_conflict) {
+		/* pick up processes with the other xnack mode */
+		xnack_enabled = !xnack_enabled;
+		xnack_conflict = 0;
+		goto build_runlist_ib;
+	}
 
-	pr_debug("Finished map process and queues to runlist\n");
-
-	if (is_over_subscription)
-		retval = pm_create_runlist(pm, &rl_buffer[rl_wptr],
+	dev_dbg(dev, "Finished map process and queues to runlist\n");
+
+	if (is_over_subscription) {
+		if (!pm->is_over_subscription)
+			dev_warn(dev, "Runlist is getting oversubscribed due to%s%s%s%s. Expect reduced ROCm performance.\n",
+				is_over_subscription & OVER_SUBSCRIPTION_PROCESS_COUNT ?
+				" too many processes" : "",
+				is_over_subscription & OVER_SUBSCRIPTION_COMPUTE_QUEUE_COUNT ?
+				" too many queues" : "",
+				is_over_subscription & OVER_SUBSCRIPTION_GWS_QUEUE_COUNT ?
+				" multiple processes using cooperative launch" : "",
+				is_over_subscription & OVER_SUBSCRIPTION_XNACK_CONFLICT ?
+				" xnack on/off processes mixed on gfx9" : "");
+
+		retval = pm->pmf->runlist(pm, &rl_buffer[rl_wptr],
 					*rl_gpu_addr,
 					alloc_size_bytes / sizeof(uint32_t),
 					true);
+	}
+	pm->is_over_subscription = !!is_over_subscription;
 
 	for (i = 0; i < alloc_size_bytes / sizeof(uint32_t); i++)
 		pr_debug("0x%2X ", rl_buffer[i]);
@@ -335,6 +278,34 @@ static int pm_create_runlist_ib(struct packet_manager *pm,
 
 int pm_init(struct packet_manager *pm, struct device_queue_manager *dqm)
 {
+	switch (dqm->dev->adev->asic_type) {
+	case CHIP_KAVERI:
+	case CHIP_HAWAII:
+		/* PM4 packet structures on CIK are the same as on VI */
+	case CHIP_CARRIZO:
+	case CHIP_TONGA:
+	case CHIP_FIJI:
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
+		pm->pmf = &kfd_vi_pm_funcs;
+		break;
+	default:
+		if (KFD_GC_VERSION(dqm->dev) == IP_VERSION(9, 4, 2) ||
+		    KFD_GC_VERSION(dqm->dev) == IP_VERSION(9, 4, 3) ||
+		    KFD_GC_VERSION(dqm->dev) == IP_VERSION(9, 4, 4) ||
+		    KFD_GC_VERSION(dqm->dev) == IP_VERSION(9, 5, 0))
+			pm->pmf = &kfd_aldebaran_pm_funcs;
+		else if (KFD_GC_VERSION(dqm->dev) >= IP_VERSION(9, 0, 1))
+			pm->pmf = &kfd_v9_pm_funcs;
+		else {
+			WARN(1, "Unexpected ASIC family %u",
+			     dqm->dev->adev->asic_type);
+			return -EINVAL;
+		}
+	}
+
 	pm->dqm = dqm;
 	mutex_init(&pm->lock);
 	pm->priv_queue = kernel_queue_init(dqm->dev, KFD_QUEUE_TYPE_HIQ);
@@ -351,43 +322,33 @@ void pm_uninit(struct packet_manager *pm)
 {
 	mutex_destroy(&pm->lock);
 	kernel_queue_uninit(pm->priv_queue);
+	pm->priv_queue = NULL;
 }
 
 int pm_send_set_resources(struct packet_manager *pm,
 				struct scheduling_resources *res)
 {
-	struct pm4_mes_set_resources *packet;
+	struct kfd_node *node = pm->dqm->dev;
+	struct device *dev = node->adev->dev;
+	uint32_t *buffer, size;
 	int retval = 0;
 
+	size = pm->pmf->set_resources_size;
 	mutex_lock(&pm->lock);
-	pm->priv_queue->ops.acquire_packet_buffer(pm->priv_queue,
-					sizeof(*packet) / sizeof(uint32_t),
-					(unsigned int **)&packet);
-	if (!packet) {
-		pr_err("Failed to allocate buffer on kernel queue\n");
+	kq_acquire_packet_buffer(pm->priv_queue,
+					size / sizeof(uint32_t),
+					(unsigned int **)&buffer);
+	if (!buffer) {
+		dev_err(dev, "Failed to allocate buffer on kernel queue\n");
 		retval = -ENOMEM;
 		goto out;
 	}
 
-	memset(packet, 0, sizeof(struct pm4_mes_set_resources));
-	packet->header.u32All = build_pm4_header(IT_SET_RESOURCES,
-					sizeof(struct pm4_mes_set_resources));
-
-	packet->bitfields2.queue_type =
-			queue_type__mes_set_resources__hsa_interface_queue_hiq;
-	packet->bitfields2.vmid_mask = res->vmid_mask;
-	packet->bitfields2.unmap_latency = KFD_UNMAP_LATENCY;
-	packet->bitfields7.oac_mask = res->oac_mask;
-	packet->bitfields8.gds_heap_base = res->gds_heap_base;
-	packet->bitfields8.gds_heap_size = res->gds_heap_size;
-
-	packet->gws_mask_lo = lower_32_bits(res->gws_mask);
-	packet->gws_mask_hi = upper_32_bits(res->gws_mask);
-
-	packet->queue_mask_lo = lower_32_bits(res->queue_mask);
-	packet->queue_mask_hi = upper_32_bits(res->queue_mask);
-
-	pm->priv_queue->ops.submit_packet(pm->priv_queue);
+	retval = pm->pmf->set_resources(pm, buffer, res);
+	if (!retval)
+		retval = kq_submit_packet(pm->priv_queue);
+	else
+		kq_rollback_packet(pm->priv_queue);
 
 out:
 	mutex_unlock(&pm->lock);
@@ -409,27 +370,27 @@ int pm_send_runlist(struct packet_manager *pm, struct list_head *dqm_queues)
 
 	pr_debug("runlist IB address: 0x%llX\n", rl_gpu_ib_addr);
 
-	packet_size_dwords = sizeof(struct pm4_mes_runlist) / sizeof(uint32_t);
+	packet_size_dwords = pm->pmf->runlist_size / sizeof(uint32_t);
 	mutex_lock(&pm->lock);
 
-	retval = pm->priv_queue->ops.acquire_packet_buffer(pm->priv_queue,
+	retval = kq_acquire_packet_buffer(pm->priv_queue,
 					packet_size_dwords, &rl_buffer);
 	if (retval)
 		goto fail_acquire_packet_buffer;
 
-	retval = pm_create_runlist(pm, rl_buffer, rl_gpu_ib_addr,
+	retval = pm->pmf->runlist(pm, rl_buffer, rl_gpu_ib_addr,
 					rl_ib_size / sizeof(uint32_t), false);
 	if (retval)
 		goto fail_create_runlist;
 
-	pm->priv_queue->ops.submit_packet(pm->priv_queue);
+	retval = kq_submit_packet(pm->priv_queue);
 
 	mutex_unlock(&pm->lock);
 
 	return retval;
 
 fail_create_runlist:
-	pm->priv_queue->ops.rollback_packet(pm->priv_queue);
+	kq_rollback_packet(pm->priv_queue);
 fail_acquire_packet_buffer:
 	mutex_unlock(&pm->lock);
 fail_create_runlist_ib:
@@ -438,124 +399,122 @@ fail_create_runlist_ib:
 }
 
 int pm_send_query_status(struct packet_manager *pm, uint64_t fence_address,
-			uint32_t fence_value)
+			uint64_t fence_value)
 {
-	int retval;
-	struct pm4_mes_query_status *packet;
+	struct kfd_node *node = pm->dqm->dev;
+	struct device *dev = node->adev->dev;
+	uint32_t *buffer, size;
+	int retval = 0;
 
 	if (WARN_ON(!fence_address))
 		return -EFAULT;
 
+	size = pm->pmf->query_status_size;
 	mutex_lock(&pm->lock);
-	retval = pm->priv_queue->ops.acquire_packet_buffer(
-			pm->priv_queue,
-			sizeof(struct pm4_mes_query_status) / sizeof(uint32_t),
-			(unsigned int **)&packet);
-	if (retval)
-		goto fail_acquire_packet_buffer;
+	kq_acquire_packet_buffer(pm->priv_queue,
+			size / sizeof(uint32_t), (unsigned int **)&buffer);
+	if (!buffer) {
+		dev_err(dev, "Failed to allocate buffer on kernel queue\n");
+		retval = -ENOMEM;
+		goto out;
+	}
+
+	retval = pm->pmf->query_status(pm, buffer, fence_address, fence_value);
+	if (!retval)
+		retval = kq_submit_packet(pm->priv_queue);
+	else
+		kq_rollback_packet(pm->priv_queue);
+
+out:
+	mutex_unlock(&pm->lock);
+	return retval;
+}
 
-	packet->header.u32All = build_pm4_header(IT_QUERY_STATUS,
-					sizeof(struct pm4_mes_query_status));
+/* pm_config_dequeue_wait_counts: Configure dequeue timer Wait Counts
+ *  by writing to CP_IQ_WAIT_TIME2 registers.
+ *
+ *  @cmd: See emum kfd_config_dequeue_wait_counts_cmd definition
+ *  @value: Depends on the cmd. This parameter is unused for
+ *    KFD_DEQUEUE_WAIT_INIT and KFD_DEQUEUE_WAIT_RESET. For
+ *    KFD_DEQUEUE_WAIT_SET_SCH_WAVE it holds value to be set
+ *
+ */
+int pm_config_dequeue_wait_counts(struct packet_manager *pm,
+		enum kfd_config_dequeue_wait_counts_cmd cmd,
+		uint32_t value)
+{
+	struct kfd_node *node = pm->dqm->dev;
+	struct device *dev = node->adev->dev;
+	int retval = 0;
+	uint32_t *buffer, size;
 
-	packet->bitfields2.context_id = 0;
-	packet->bitfields2.interrupt_sel =
-			interrupt_sel__mes_query_status__completion_status;
-	packet->bitfields2.command =
-			command__mes_query_status__fence_only_after_write_ack;
+	if (!pm->pmf->config_dequeue_wait_counts ||
+	    !pm->pmf->config_dequeue_wait_counts_size)
+		return 0;
 
-	packet->addr_hi = upper_32_bits((uint64_t)fence_address);
-	packet->addr_lo = lower_32_bits((uint64_t)fence_address);
-	packet->data_hi = upper_32_bits((uint64_t)fence_value);
-	packet->data_lo = lower_32_bits((uint64_t)fence_value);
+	if (cmd == KFD_DEQUEUE_WAIT_INIT && (KFD_GC_VERSION(pm->dqm->dev) < IP_VERSION(9, 4, 1) ||
+	   KFD_GC_VERSION(pm->dqm->dev) >= IP_VERSION(10, 0, 0)))
+		return 0;
 
-	pm->priv_queue->ops.submit_packet(pm->priv_queue);
+	size = pm->pmf->config_dequeue_wait_counts_size;
 
-fail_acquire_packet_buffer:
+	mutex_lock(&pm->lock);
+
+	if (size) {
+		kq_acquire_packet_buffer(pm->priv_queue,
+			size / sizeof(uint32_t),
+			(unsigned int **)&buffer);
+
+		if (!buffer) {
+			dev_err(dev,
+				"Failed to allocate buffer on kernel queue\n");
+			retval = -ENOMEM;
+			goto out;
+		}
+
+		retval = pm->pmf->config_dequeue_wait_counts(pm, buffer,
+							     cmd, value);
+		if (!retval) {
+			retval = kq_submit_packet(pm->priv_queue);
+
+			/* If default value is modified, cache that in dqm->wait_times */
+			if (!retval && cmd == KFD_DEQUEUE_WAIT_INIT)
+				update_dqm_wait_times(pm->dqm);
+		} else {
+			kq_rollback_packet(pm->priv_queue);
+		}
+	}
+out:
 	mutex_unlock(&pm->lock);
 	return retval;
 }
 
-int pm_send_unmap_queue(struct packet_manager *pm, enum kfd_queue_type type,
-			enum kfd_preempt_type_filter mode,
-			uint32_t filter_param, bool reset,
-			unsigned int sdma_engine)
+int pm_send_unmap_queue(struct packet_manager *pm,
+			enum kfd_unmap_queues_filter filter,
+			uint32_t filter_param, bool reset)
 {
-	int retval;
-	uint32_t *buffer;
-	struct pm4_mes_unmap_queues *packet;
+	struct kfd_node *node = pm->dqm->dev;
+	struct device *dev = node->adev->dev;
+	uint32_t *buffer, size;
+	int retval = 0;
 
+	size = pm->pmf->unmap_queues_size;
 	mutex_lock(&pm->lock);
-	retval = pm->priv_queue->ops.acquire_packet_buffer(
-			pm->priv_queue,
-			sizeof(struct pm4_mes_unmap_queues) / sizeof(uint32_t),
-			&buffer);
-	if (retval)
-		goto err_acquire_packet_buffer;
-
-	packet = (struct pm4_mes_unmap_queues *)buffer;
-	memset(buffer, 0, sizeof(struct pm4_mes_unmap_queues));
-	pr_debug("static_queue: unmapping queues: mode is %d , reset is %d , type is %d\n",
-		mode, reset, type);
-	packet->header.u32All = build_pm4_header(IT_UNMAP_QUEUES,
-					sizeof(struct pm4_mes_unmap_queues));
-	switch (type) {
-	case KFD_QUEUE_TYPE_COMPUTE:
-	case KFD_QUEUE_TYPE_DIQ:
-		packet->bitfields2.engine_sel =
-			engine_sel__mes_unmap_queues__compute;
-		break;
-	case KFD_QUEUE_TYPE_SDMA:
-		packet->bitfields2.engine_sel =
-			engine_sel__mes_unmap_queues__sdma0 + sdma_engine;
-		break;
-	default:
-		WARN(1, "queue type %d", type);
-		retval = -EINVAL;
-		goto err_invalid;
+	kq_acquire_packet_buffer(pm->priv_queue,
+			size / sizeof(uint32_t), (unsigned int **)&buffer);
+	if (!buffer) {
+		dev_err(dev, "Failed to allocate buffer on kernel queue\n");
+		retval = -ENOMEM;
+		goto out;
 	}
 
-	if (reset)
-		packet->bitfields2.action =
-				action__mes_unmap_queues__reset_queues;
+	retval = pm->pmf->unmap_queues(pm, buffer, filter, filter_param, reset);
+	if (!retval)
+		retval = kq_submit_packet(pm->priv_queue);
 	else
-		packet->bitfields2.action =
-				action__mes_unmap_queues__preempt_queues;
-
-	switch (mode) {
-	case KFD_PREEMPT_TYPE_FILTER_SINGLE_QUEUE:
-		packet->bitfields2.queue_sel =
-				queue_sel__mes_unmap_queues__perform_request_on_specified_queues;
-		packet->bitfields2.num_queues = 1;
-		packet->bitfields3b.doorbell_offset0 = filter_param;
-		break;
-	case KFD_PREEMPT_TYPE_FILTER_BY_PASID:
-		packet->bitfields2.queue_sel =
-				queue_sel__mes_unmap_queues__perform_request_on_pasid_queues;
-		packet->bitfields3a.pasid = filter_param;
-		break;
-	case KFD_PREEMPT_TYPE_FILTER_ALL_QUEUES:
-		packet->bitfields2.queue_sel =
-				queue_sel__mes_unmap_queues__unmap_all_queues;
-		break;
-	case KFD_PREEMPT_TYPE_FILTER_DYNAMIC_QUEUES:
-		/* in this case, we do not preempt static queues */
-		packet->bitfields2.queue_sel =
-				queue_sel__mes_unmap_queues__unmap_all_non_static_queues;
-		break;
-	default:
-		WARN(1, "filter %d", mode);
-		retval = -EINVAL;
-		goto err_invalid;
-	}
+		kq_rollback_packet(pm->priv_queue);
 
-	pm->priv_queue->ops.submit_packet(pm->priv_queue);
-
-	mutex_unlock(&pm->lock);
-	return 0;
-
-err_invalid:
-	pm->priv_queue->ops.rollback_packet(pm->priv_queue);
-err_acquire_packet_buffer:
+out:
 	mutex_unlock(&pm->lock);
 	return retval;
 }
@@ -569,3 +528,57 @@ void pm_release_ib(struct packet_manager *pm)
 	}
 	mutex_unlock(&pm->lock);
 }
+
+#if defined(CONFIG_DEBUG_FS)
+
+int pm_debugfs_runlist(struct seq_file *m, void *data)
+{
+	struct packet_manager *pm = data;
+
+	mutex_lock(&pm->lock);
+
+	if (!pm->allocated) {
+		seq_puts(m, "  No active runlist\n");
+		goto out;
+	}
+
+	seq_hex_dump(m, "  ", DUMP_PREFIX_OFFSET, 32, 4,
+		     pm->ib_buffer_obj->cpu_ptr, pm->ib_size_bytes, false);
+
+out:
+	mutex_unlock(&pm->lock);
+	return 0;
+}
+
+int pm_debugfs_hang_hws(struct packet_manager *pm)
+{
+	struct kfd_node *node = pm->dqm->dev;
+	struct device *dev = node->adev->dev;
+	uint32_t *buffer, size;
+	int r = 0;
+
+	if (!pm->priv_queue)
+		return -EAGAIN;
+
+	size = pm->pmf->query_status_size;
+	mutex_lock(&pm->lock);
+	kq_acquire_packet_buffer(pm->priv_queue,
+			size / sizeof(uint32_t), (unsigned int **)&buffer);
+	if (!buffer) {
+		dev_err(dev, "Failed to allocate buffer on kernel queue\n");
+		r = -ENOMEM;
+		goto out;
+	}
+	memset(buffer, 0x55, size);
+	kq_submit_packet(pm->priv_queue);
+
+	dev_info(dev, "Submitting %x %x %x %x %x %x %x to HIQ to hang the HWS.",
+		 buffer[0], buffer[1], buffer[2], buffer[3], buffer[4],
+		 buffer[5], buffer[6]);
+out:
+	mutex_unlock(&pm->lock);
+	return r;
+}
+
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c
new file mode 100644
index 000000000000..505036968a77
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c
@@ -0,0 +1,507 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
+/*
+ * Copyright 2016-2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "kfd_kernel_queue.h"
+#include "kfd_device_queue_manager.h"
+#include "kfd_pm4_headers_ai.h"
+#include "kfd_pm4_headers_aldebaran.h"
+#include "kfd_pm4_opcodes.h"
+#include "gc/gc_10_1_0_sh_mask.h"
+
+static int pm_map_process_v9(struct packet_manager *pm,
+		uint32_t *buffer, struct qcm_process_device *qpd)
+{
+	struct pm4_mes_map_process *packet;
+	uint64_t vm_page_table_base_addr = qpd->page_table_base;
+	struct kfd_node *kfd = pm->dqm->dev;
+	struct kfd_process_device *pdd =
+			container_of(qpd, struct kfd_process_device, qpd);
+	struct amdgpu_device *adev = kfd->adev;
+
+	packet = (struct pm4_mes_map_process *)buffer;
+	memset(buffer, 0, sizeof(struct pm4_mes_map_process));
+	packet->header.u32All = pm_build_pm4_header(IT_MAP_PROCESS,
+					sizeof(struct pm4_mes_map_process));
+	if (adev->enforce_isolation[kfd->node_id] == AMDGPU_ENFORCE_ISOLATION_ENABLE)
+		packet->bitfields2.exec_cleaner_shader = 1;
+	packet->bitfields2.diq_enable = (qpd->is_debug) ? 1 : 0;
+	packet->bitfields2.process_quantum = 10;
+	packet->bitfields2.pasid = pdd->pasid;
+	packet->bitfields14.gds_size = qpd->gds_size & 0x3F;
+	packet->bitfields14.gds_size_hi = (qpd->gds_size >> 6) & 0xF;
+	packet->bitfields14.num_gws = (qpd->mapped_gws_queue) ? qpd->num_gws : 0;
+	packet->bitfields14.num_oac = qpd->num_oac;
+	packet->bitfields14.sdma_enable = 1;
+	packet->bitfields14.num_queues = (qpd->is_debug) ? 0 : qpd->queue_count;
+
+	if (kfd->dqm->trap_debug_vmid && pdd->process->debug_trap_enabled &&
+			pdd->process->runtime_info.runtime_state == DEBUG_RUNTIME_STATE_ENABLED) {
+		packet->bitfields2.debug_vmid = kfd->dqm->trap_debug_vmid;
+		packet->bitfields2.new_debug = 1;
+	}
+
+	packet->sh_mem_config = qpd->sh_mem_config;
+	packet->sh_mem_bases = qpd->sh_mem_bases;
+	if (qpd->tba_addr) {
+		packet->sq_shader_tba_lo = lower_32_bits(qpd->tba_addr >> 8);
+		/* On GFX9, unlike GFX10, bit TRAP_EN of SQ_SHADER_TBA_HI is
+		 * not defined, so setting it won't do any harm.
+		 */
+		packet->sq_shader_tba_hi = upper_32_bits(qpd->tba_addr >> 8)
+				| 1 << SQ_SHADER_TBA_HI__TRAP_EN__SHIFT;
+
+		packet->sq_shader_tma_lo = lower_32_bits(qpd->tma_addr >> 8);
+		packet->sq_shader_tma_hi = upper_32_bits(qpd->tma_addr >> 8);
+	}
+
+	packet->gds_addr_lo = lower_32_bits(qpd->gds_context_area);
+	packet->gds_addr_hi = upper_32_bits(qpd->gds_context_area);
+
+	packet->vm_context_page_table_base_addr_lo32 =
+			lower_32_bits(vm_page_table_base_addr);
+	packet->vm_context_page_table_base_addr_hi32 =
+			upper_32_bits(vm_page_table_base_addr);
+
+	return 0;
+}
+
+static int pm_map_process_aldebaran(struct packet_manager *pm,
+		uint32_t *buffer, struct qcm_process_device *qpd)
+{
+	struct pm4_mes_map_process_aldebaran *packet;
+	uint64_t vm_page_table_base_addr = qpd->page_table_base;
+	struct kfd_dev *kfd = pm->dqm->dev->kfd;
+	struct kfd_node *knode = pm->dqm->dev;
+	struct kfd_process_device *pdd =
+			container_of(qpd, struct kfd_process_device, qpd);
+	int i;
+	struct amdgpu_device *adev = kfd->adev;
+
+	packet = (struct pm4_mes_map_process_aldebaran *)buffer;
+	memset(buffer, 0, sizeof(struct pm4_mes_map_process_aldebaran));
+	packet->header.u32All = pm_build_pm4_header(IT_MAP_PROCESS,
+			sizeof(struct pm4_mes_map_process_aldebaran));
+	if (adev->enforce_isolation[knode->node_id] ==
+	    AMDGPU_ENFORCE_ISOLATION_ENABLE)
+		packet->bitfields2.exec_cleaner_shader = 1;
+	packet->bitfields2.diq_enable = (qpd->is_debug) ? 1 : 0;
+	packet->bitfields2.process_quantum = 10;
+	packet->bitfields2.pasid = pdd->pasid;
+	packet->bitfields14.gds_size = qpd->gds_size & 0x3F;
+	packet->bitfields14.gds_size_hi = (qpd->gds_size >> 6) & 0xF;
+	packet->bitfields14.num_gws = (qpd->mapped_gws_queue) ? qpd->num_gws : 0;
+	packet->bitfields14.num_oac = qpd->num_oac;
+	packet->bitfields14.sdma_enable = 1;
+	packet->bitfields14.num_queues = (qpd->is_debug) ? 0 : qpd->queue_count;
+	packet->spi_gdbg_per_vmid_cntl = pdd->spi_dbg_override |
+						pdd->spi_dbg_launch_mode;
+
+	if (pdd->process->debug_trap_enabled) {
+		for (i = 0; i < kfd->device_info.num_of_watch_points; i++)
+			packet->tcp_watch_cntl[i] = pdd->watch_points[i];
+
+		packet->bitfields2.single_memops =
+				!!(pdd->process->dbg_flags & KFD_DBG_TRAP_FLAG_SINGLE_MEM_OP);
+	}
+
+	packet->sh_mem_config = qpd->sh_mem_config;
+	packet->sh_mem_bases = qpd->sh_mem_bases;
+	if (qpd->tba_addr) {
+		packet->sq_shader_tba_lo = lower_32_bits(qpd->tba_addr >> 8);
+		packet->sq_shader_tba_hi = upper_32_bits(qpd->tba_addr >> 8);
+		packet->sq_shader_tma_lo = lower_32_bits(qpd->tma_addr >> 8);
+		packet->sq_shader_tma_hi = upper_32_bits(qpd->tma_addr >> 8);
+	}
+
+	packet->gds_addr_lo = lower_32_bits(qpd->gds_context_area);
+	packet->gds_addr_hi = upper_32_bits(qpd->gds_context_area);
+
+	packet->vm_context_page_table_base_addr_lo32 =
+			lower_32_bits(vm_page_table_base_addr);
+	packet->vm_context_page_table_base_addr_hi32 =
+			upper_32_bits(vm_page_table_base_addr);
+
+	return 0;
+}
+
+static int pm_runlist_v9(struct packet_manager *pm, uint32_t *buffer,
+			uint64_t ib, size_t ib_size_in_dwords, bool chain)
+{
+	struct pm4_mes_runlist *packet;
+
+	int concurrent_proc_cnt = 0;
+	struct kfd_node *kfd = pm->dqm->dev;
+	struct amdgpu_device *adev = kfd->adev;
+
+	/* Determine the number of processes to map together to HW:
+	 * it can not exceed the number of VMIDs available to the
+	 * scheduler, and it is determined by the smaller of the number
+	 * of processes in the runlist and kfd module parameter
+	 * hws_max_conc_proc.
+	 * However, if enforce_isolation is set (toggle LDS/VGPRs/SGPRs
+	 * cleaner between process switch), enable single-process mode
+	 * in HWS.
+	 * Note: the arbitration between the number of VMIDs and
+	 * hws_max_conc_proc has been done in
+	 * kgd2kfd_device_init().
+	 */
+	concurrent_proc_cnt = (adev->enforce_isolation[kfd->node_id] ==
+			       AMDGPU_ENFORCE_ISOLATION_ENABLE) ?
+		1 : min(pm->dqm->processes_count, kfd->max_proc_per_quantum);
+
+	packet = (struct pm4_mes_runlist *)buffer;
+
+	memset(buffer, 0, sizeof(struct pm4_mes_runlist));
+	packet->header.u32All = pm_build_pm4_header(IT_RUN_LIST,
+						sizeof(struct pm4_mes_runlist));
+
+	packet->bitfields4.ib_size = ib_size_in_dwords;
+	packet->bitfields4.chain = chain ? 1 : 0;
+	packet->bitfields4.offload_polling = 0;
+	packet->bitfields4.chained_runlist_idle_disable = chain ? 1 : 0;
+	packet->bitfields4.valid = 1;
+	packet->bitfields4.process_cnt = concurrent_proc_cnt;
+	packet->ordinal2 = lower_32_bits(ib);
+	packet->ib_base_hi = upper_32_bits(ib);
+
+	return 0;
+}
+
+static int pm_set_resources_v9(struct packet_manager *pm, uint32_t *buffer,
+				struct scheduling_resources *res)
+{
+	struct pm4_mes_set_resources *packet;
+
+	packet = (struct pm4_mes_set_resources *)buffer;
+	memset(buffer, 0, sizeof(struct pm4_mes_set_resources));
+
+	packet->header.u32All = pm_build_pm4_header(IT_SET_RESOURCES,
+					sizeof(struct pm4_mes_set_resources));
+
+	packet->bitfields2.queue_type =
+			queue_type__mes_set_resources__hsa_interface_queue_hiq;
+	packet->bitfields2.vmid_mask = res->vmid_mask;
+	packet->bitfields2.unmap_latency = KFD_UNMAP_LATENCY_MS / 100;
+	if (pm->dqm->dev->adev->gmc.xnack_flags & AMDGPU_GMC_XNACK_FLAG_CHAIN)
+		packet->bitfields2.enb_xnack_retry_disable_check = 1;
+	packet->bitfields7.oac_mask = res->oac_mask;
+	packet->bitfields8.gds_heap_base = res->gds_heap_base;
+	packet->bitfields8.gds_heap_size = res->gds_heap_size;
+
+	packet->gws_mask_lo = lower_32_bits(res->gws_mask);
+	packet->gws_mask_hi = upper_32_bits(res->gws_mask);
+
+	packet->queue_mask_lo = lower_32_bits(res->queue_mask);
+	packet->queue_mask_hi = upper_32_bits(res->queue_mask);
+
+	return 0;
+}
+
+static inline bool pm_use_ext_eng(struct kfd_dev *dev)
+{
+	return amdgpu_ip_version(dev->adev, SDMA0_HWIP, 0) >=
+	       IP_VERSION(5, 2, 0);
+}
+
+static int pm_map_queues_v9(struct packet_manager *pm, uint32_t *buffer,
+		struct queue *q, bool is_static)
+{
+	struct pm4_mes_map_queues *packet;
+
+	packet = (struct pm4_mes_map_queues *)buffer;
+	memset(buffer, 0, sizeof(struct pm4_mes_map_queues));
+
+	packet->header.u32All = pm_build_pm4_header(IT_MAP_QUEUES,
+					sizeof(struct pm4_mes_map_queues));
+	packet->bitfields2.num_queues = 1;
+	packet->bitfields2.queue_sel =
+		queue_sel__mes_map_queues__map_to_hws_determined_queue_slots_vi;
+
+	packet->bitfields2.engine_sel =
+		engine_sel__mes_map_queues__compute_vi;
+	packet->bitfields2.gws_control_queue = q->properties.is_gws ? 1 : 0;
+	packet->bitfields2.extended_engine_sel =
+		extended_engine_sel__mes_map_queues__legacy_engine_sel;
+	packet->bitfields2.queue_type =
+		queue_type__mes_map_queues__normal_compute_vi;
+
+	switch (q->properties.type) {
+	case KFD_QUEUE_TYPE_COMPUTE:
+		if (is_static)
+			packet->bitfields2.queue_type =
+		queue_type__mes_map_queues__normal_latency_static_queue_vi;
+		break;
+	case KFD_QUEUE_TYPE_DIQ:
+		packet->bitfields2.queue_type =
+			queue_type__mes_map_queues__debug_interface_queue_vi;
+		break;
+	case KFD_QUEUE_TYPE_SDMA:
+	case KFD_QUEUE_TYPE_SDMA_XGMI:
+		if (q->properties.sdma_engine_id < 2 &&
+		    !pm_use_ext_eng(q->device->kfd))
+			packet->bitfields2.engine_sel = q->properties.sdma_engine_id +
+				engine_sel__mes_map_queues__sdma0_vi;
+		else {
+			/*
+			 * For GFX9.4.3, SDMA engine id can be greater than 8.
+			 * For such cases, set extended_engine_sel to 2 and
+			 * ensure engine_sel lies between 0-7.
+			 */
+			if (q->properties.sdma_engine_id >= 8)
+				packet->bitfields2.extended_engine_sel =
+					extended_engine_sel__mes_map_queues__sdma8_to_15_sel;
+			else
+				packet->bitfields2.extended_engine_sel =
+					extended_engine_sel__mes_map_queues__sdma0_to_7_sel;
+
+			packet->bitfields2.engine_sel = q->properties.sdma_engine_id % 8;
+		}
+		break;
+	default:
+		WARN(1, "queue type %d", q->properties.type);
+		return -EINVAL;
+	}
+	packet->bitfields3.doorbell_offset =
+			q->properties.doorbell_off;
+
+	packet->mqd_addr_lo =
+			lower_32_bits(q->gart_mqd_addr);
+
+	packet->mqd_addr_hi =
+			upper_32_bits(q->gart_mqd_addr);
+
+	packet->wptr_addr_lo =
+			lower_32_bits((uint64_t)q->properties.write_ptr);
+
+	packet->wptr_addr_hi =
+			upper_32_bits((uint64_t)q->properties.write_ptr);
+
+	return 0;
+}
+
+static inline void pm_build_dequeue_wait_counts_packet_info(struct packet_manager *pm,
+			uint32_t sch_value, uint32_t que_sleep, uint32_t *reg_offset,
+			uint32_t *reg_data)
+{
+	pm->dqm->dev->kfd2kgd->build_dequeue_wait_counts_packet_info(
+		pm->dqm->dev->adev,
+		pm->dqm->wait_times,
+		sch_value,
+		que_sleep,
+		reg_offset,
+		reg_data);
+}
+
+/* pm_config_dequeue_wait_counts_v9: Builds WRITE_DATA packet with
+ *    register/value for configuring dequeue wait counts
+ *
+ * @return: -ve for failure and 0 for success and buffer is
+ *  filled in with packet
+ *
+ **/
+static int pm_config_dequeue_wait_counts_v9(struct packet_manager *pm,
+		uint32_t *buffer,
+		enum kfd_config_dequeue_wait_counts_cmd cmd,
+		uint32_t value)
+{
+	struct pm4_mec_write_data_mmio *packet;
+	uint32_t reg_offset = 0;
+	uint32_t reg_data = 0;
+
+	switch (cmd) {
+	case KFD_DEQUEUE_WAIT_INIT: {
+		uint32_t sch_wave = 0, que_sleep = 1;
+
+		/* For all gfx9 ASICs > gfx941,
+		 * Reduce CP_IQ_WAIT_TIME2.QUE_SLEEP to 0x1 from default 0x40.
+		 * On a 1GHz machine this is roughly 1 microsecond, which is
+		 * about how long it takes to load data out of memory during
+		 * queue connect
+		 * QUE_SLEEP: Wait Count for Dequeue Retry.
+		 *
+		 * Set CWSR grace period to 1x1000 cycle for GFX9.4.3 APU
+		 */
+		if (KFD_GC_VERSION(pm->dqm->dev) < IP_VERSION(9, 4, 1) ||
+		    KFD_GC_VERSION(pm->dqm->dev) >= IP_VERSION(10, 0, 0))
+			return -EPERM;
+
+		if (amdgpu_emu_mode == 0 && pm->dqm->dev->adev->gmc.is_app_apu &&
+		    (KFD_GC_VERSION(pm->dqm->dev) == IP_VERSION(9, 4, 3)))
+			sch_wave = 1;
+
+		pm_build_dequeue_wait_counts_packet_info(pm, sch_wave, que_sleep,
+			&reg_offset, &reg_data);
+
+		break;
+	}
+	case KFD_DEQUEUE_WAIT_RESET:
+		/* reg_data would be set to dqm->wait_times */
+		pm_build_dequeue_wait_counts_packet_info(pm, 0, 0, &reg_offset, &reg_data);
+		break;
+
+	case KFD_DEQUEUE_WAIT_SET_SCH_WAVE:
+		/* The CP cannot handle value 0 and it will result in
+		 * an infinite grace period being set so set to 1 to prevent this. Also
+		 * avoid debugger API breakage as it sets 0 and expects a low value.
+		 */
+		if (!value)
+			value = 1;
+		pm_build_dequeue_wait_counts_packet_info(pm, value, 0, &reg_offset, &reg_data);
+		break;
+	default:
+		pr_err("Invalid dequeue wait cmd\n");
+		return -EINVAL;
+	}
+
+	packet = (struct pm4_mec_write_data_mmio *)buffer;
+	memset(buffer, 0, sizeof(struct pm4_mec_write_data_mmio));
+
+	packet->header.u32All = pm_build_pm4_header(IT_WRITE_DATA,
+					sizeof(struct pm4_mec_write_data_mmio));
+
+	packet->bitfields2.dst_sel  = dst_sel___write_data__mem_mapped_register;
+	packet->bitfields2.addr_incr =
+			addr_incr___write_data__do_not_increment_address;
+
+	packet->bitfields3.dst_mmreg_addr = reg_offset;
+
+	packet->data = reg_data;
+
+	return 0;
+}
+
+static int pm_unmap_queues_v9(struct packet_manager *pm, uint32_t *buffer,
+			enum kfd_unmap_queues_filter filter,
+			uint32_t filter_param, bool reset)
+{
+	struct pm4_mes_unmap_queues *packet;
+
+	packet = (struct pm4_mes_unmap_queues *)buffer;
+	memset(buffer, 0, sizeof(struct pm4_mes_unmap_queues));
+
+	packet->header.u32All = pm_build_pm4_header(IT_UNMAP_QUEUES,
+					sizeof(struct pm4_mes_unmap_queues));
+
+	packet->bitfields2.extended_engine_sel =
+				pm_use_ext_eng(pm->dqm->dev->kfd) ?
+		extended_engine_sel__mes_unmap_queues__sdma0_to_7_sel :
+		extended_engine_sel__mes_unmap_queues__legacy_engine_sel;
+
+	packet->bitfields2.engine_sel =
+		engine_sel__mes_unmap_queues__compute;
+
+	if (reset)
+		packet->bitfields2.action =
+			action__mes_unmap_queues__reset_queues;
+	else
+		packet->bitfields2.action =
+			action__mes_unmap_queues__preempt_queues;
+
+	switch (filter) {
+	case KFD_UNMAP_QUEUES_FILTER_BY_PASID:
+		packet->bitfields2.queue_sel =
+			queue_sel__mes_unmap_queues__perform_request_on_pasid_queues;
+		packet->bitfields3a.pasid = filter_param;
+		break;
+	case KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES:
+		packet->bitfields2.queue_sel =
+			queue_sel__mes_unmap_queues__unmap_all_queues;
+		break;
+	case KFD_UNMAP_QUEUES_FILTER_DYNAMIC_QUEUES:
+		/* in this case, we do not preempt static queues */
+		packet->bitfields2.queue_sel =
+			queue_sel__mes_unmap_queues__unmap_all_non_static_queues;
+		break;
+	default:
+		WARN(1, "filter %d", filter);
+		return -EINVAL;
+	}
+
+	return 0;
+
+}
+
+static int pm_query_status_v9(struct packet_manager *pm, uint32_t *buffer,
+			uint64_t fence_address,	uint64_t fence_value)
+{
+	struct pm4_mes_query_status *packet;
+
+	packet = (struct pm4_mes_query_status *)buffer;
+	memset(buffer, 0, sizeof(struct pm4_mes_query_status));
+
+
+	packet->header.u32All = pm_build_pm4_header(IT_QUERY_STATUS,
+					sizeof(struct pm4_mes_query_status));
+
+	packet->bitfields2.context_id = 0;
+	packet->bitfields2.interrupt_sel =
+			interrupt_sel__mes_query_status__completion_status;
+	packet->bitfields2.command =
+			command__mes_query_status__fence_only_after_write_ack;
+
+	packet->addr_hi = upper_32_bits((uint64_t)fence_address);
+	packet->addr_lo = lower_32_bits((uint64_t)fence_address);
+	packet->data_hi = upper_32_bits((uint64_t)fence_value);
+	packet->data_lo = lower_32_bits((uint64_t)fence_value);
+
+	return 0;
+}
+
+const struct packet_manager_funcs kfd_v9_pm_funcs = {
+	.map_process		= pm_map_process_v9,
+	.runlist		= pm_runlist_v9,
+	.set_resources		= pm_set_resources_v9,
+	.map_queues		= pm_map_queues_v9,
+	.unmap_queues		= pm_unmap_queues_v9,
+	.config_dequeue_wait_counts = pm_config_dequeue_wait_counts_v9,
+	.query_status		= pm_query_status_v9,
+	.release_mem		= NULL,
+	.map_process_size	= sizeof(struct pm4_mes_map_process),
+	.runlist_size		= sizeof(struct pm4_mes_runlist),
+	.set_resources_size	= sizeof(struct pm4_mes_set_resources),
+	.map_queues_size	= sizeof(struct pm4_mes_map_queues),
+	.unmap_queues_size	= sizeof(struct pm4_mes_unmap_queues),
+	.config_dequeue_wait_counts_size  = sizeof(struct pm4_mec_write_data_mmio),
+	.query_status_size	= sizeof(struct pm4_mes_query_status),
+	.release_mem_size	= 0,
+};
+
+const struct packet_manager_funcs kfd_aldebaran_pm_funcs = {
+	.map_process		= pm_map_process_aldebaran,
+	.runlist		= pm_runlist_v9,
+	.set_resources		= pm_set_resources_v9,
+	.map_queues		= pm_map_queues_v9,
+	.unmap_queues		= pm_unmap_queues_v9,
+	.config_dequeue_wait_counts = pm_config_dequeue_wait_counts_v9,
+	.query_status		= pm_query_status_v9,
+	.release_mem		= NULL,
+	.map_process_size	= sizeof(struct pm4_mes_map_process_aldebaran),
+	.runlist_size		= sizeof(struct pm4_mes_runlist),
+	.set_resources_size	= sizeof(struct pm4_mes_set_resources),
+	.map_queues_size	= sizeof(struct pm4_mes_map_queues),
+	.unmap_queues_size	= sizeof(struct pm4_mes_unmap_queues),
+	.config_dequeue_wait_counts_size  = sizeof(struct pm4_mec_write_data_mmio),
+	.query_status_size	= sizeof(struct pm4_mes_query_status),
+	.release_mem_size	= 0,
+};
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_vi.c b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_vi.c
new file mode 100644
index 000000000000..a1de5d7e173a
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_vi.c
@@ -0,0 +1,318 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
+/*
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "kfd_kernel_queue.h"
+#include "kfd_device_queue_manager.h"
+#include "kfd_pm4_headers_vi.h"
+#include "kfd_pm4_opcodes.h"
+
+unsigned int pm_build_pm4_header(unsigned int opcode, size_t packet_size)
+{
+	union PM4_MES_TYPE_3_HEADER header;
+
+	header.u32All = 0;
+	header.opcode = opcode;
+	header.count = packet_size / 4 - 2;
+	header.type = PM4_TYPE_3;
+
+	return header.u32All;
+}
+
+static int pm_map_process_vi(struct packet_manager *pm, uint32_t *buffer,
+				struct qcm_process_device *qpd)
+{
+	struct kfd_process_device *pdd = qpd_to_pdd(qpd);
+	struct pm4_mes_map_process *packet;
+
+	packet = (struct pm4_mes_map_process *)buffer;
+
+	memset(buffer, 0, sizeof(struct pm4_mes_map_process));
+
+	packet->header.u32All = pm_build_pm4_header(IT_MAP_PROCESS,
+					sizeof(struct pm4_mes_map_process));
+	packet->bitfields2.diq_enable = (qpd->is_debug) ? 1 : 0;
+	packet->bitfields2.process_quantum = 10;
+	packet->bitfields2.pasid = pdd->pasid;
+	packet->bitfields3.page_table_base = qpd->page_table_base;
+	packet->bitfields10.gds_size = qpd->gds_size;
+	packet->bitfields10.num_gws = qpd->num_gws;
+	packet->bitfields10.num_oac = qpd->num_oac;
+	packet->bitfields10.num_queues = (qpd->is_debug) ? 0 : qpd->queue_count;
+
+	packet->sh_mem_config = qpd->sh_mem_config;
+	packet->sh_mem_bases = qpd->sh_mem_bases;
+	packet->sh_mem_ape1_base = qpd->sh_mem_ape1_base;
+	packet->sh_mem_ape1_limit = qpd->sh_mem_ape1_limit;
+
+	packet->sh_hidden_private_base_vmid = qpd->sh_hidden_private_base;
+
+	packet->gds_addr_lo = lower_32_bits(qpd->gds_context_area);
+	packet->gds_addr_hi = upper_32_bits(qpd->gds_context_area);
+
+	return 0;
+}
+
+static int pm_runlist_vi(struct packet_manager *pm, uint32_t *buffer,
+			uint64_t ib, size_t ib_size_in_dwords, bool chain)
+{
+	struct pm4_mes_runlist *packet;
+	int concurrent_proc_cnt = 0;
+	struct kfd_node *kfd = pm->dqm->dev;
+
+	if (WARN_ON(!ib))
+		return -EFAULT;
+
+	/* Determine the number of processes to map together to HW:
+	 * it can not exceed the number of VMIDs available to the
+	 * scheduler, and it is determined by the smaller of the number
+	 * of processes in the runlist and kfd module parameter
+	 * hws_max_conc_proc.
+	 * Note: the arbitration between the number of VMIDs and
+	 * hws_max_conc_proc has been done in
+	 * kgd2kfd_device_init().
+	 */
+	concurrent_proc_cnt = min(pm->dqm->processes_count,
+			kfd->max_proc_per_quantum);
+
+	packet = (struct pm4_mes_runlist *)buffer;
+
+	memset(buffer, 0, sizeof(struct pm4_mes_runlist));
+	packet->header.u32All = pm_build_pm4_header(IT_RUN_LIST,
+						sizeof(struct pm4_mes_runlist));
+
+	packet->bitfields4.ib_size = ib_size_in_dwords;
+	packet->bitfields4.chain = chain ? 1 : 0;
+	packet->bitfields4.offload_polling = 0;
+	packet->bitfields4.valid = 1;
+	packet->bitfields4.process_cnt = concurrent_proc_cnt;
+	packet->ordinal2 = lower_32_bits(ib);
+	packet->bitfields3.ib_base_hi = upper_32_bits(ib);
+
+	return 0;
+}
+
+static int pm_set_resources_vi(struct packet_manager *pm, uint32_t *buffer,
+			       struct scheduling_resources *res)
+{
+	struct pm4_mes_set_resources *packet;
+
+	packet = (struct pm4_mes_set_resources *)buffer;
+	memset(buffer, 0, sizeof(struct pm4_mes_set_resources));
+
+	packet->header.u32All = pm_build_pm4_header(IT_SET_RESOURCES,
+					sizeof(struct pm4_mes_set_resources));
+
+	packet->bitfields2.queue_type =
+			queue_type__mes_set_resources__hsa_interface_queue_hiq;
+	packet->bitfields2.vmid_mask = res->vmid_mask;
+	packet->bitfields2.unmap_latency = KFD_UNMAP_LATENCY_MS / 100;
+	packet->bitfields7.oac_mask = res->oac_mask;
+	packet->bitfields8.gds_heap_base = res->gds_heap_base;
+	packet->bitfields8.gds_heap_size = res->gds_heap_size;
+
+	packet->gws_mask_lo = lower_32_bits(res->gws_mask);
+	packet->gws_mask_hi = upper_32_bits(res->gws_mask);
+
+	packet->queue_mask_lo = lower_32_bits(res->queue_mask);
+	packet->queue_mask_hi = upper_32_bits(res->queue_mask);
+
+	return 0;
+}
+
+static int pm_map_queues_vi(struct packet_manager *pm, uint32_t *buffer,
+		struct queue *q, bool is_static)
+{
+	struct pm4_mes_map_queues *packet;
+	bool use_static = is_static;
+
+	packet = (struct pm4_mes_map_queues *)buffer;
+	memset(buffer, 0, sizeof(struct pm4_mes_map_queues));
+
+	packet->header.u32All = pm_build_pm4_header(IT_MAP_QUEUES,
+					sizeof(struct pm4_mes_map_queues));
+	packet->bitfields2.num_queues = 1;
+	packet->bitfields2.queue_sel =
+		queue_sel__mes_map_queues__map_to_hws_determined_queue_slots_vi;
+
+	packet->bitfields2.engine_sel =
+		engine_sel__mes_map_queues__compute_vi;
+	packet->bitfields2.queue_type =
+		queue_type__mes_map_queues__normal_compute_vi;
+
+	switch (q->properties.type) {
+	case KFD_QUEUE_TYPE_COMPUTE:
+		if (use_static)
+			packet->bitfields2.queue_type =
+		queue_type__mes_map_queues__normal_latency_static_queue_vi;
+		break;
+	case KFD_QUEUE_TYPE_DIQ:
+		packet->bitfields2.queue_type =
+			queue_type__mes_map_queues__debug_interface_queue_vi;
+		break;
+	case KFD_QUEUE_TYPE_SDMA:
+	case KFD_QUEUE_TYPE_SDMA_XGMI:
+		packet->bitfields2.engine_sel = q->properties.sdma_engine_id +
+				engine_sel__mes_map_queues__sdma0_vi;
+		use_static = false; /* no static queues under SDMA */
+		break;
+	default:
+		WARN(1, "queue type %d", q->properties.type);
+		return -EINVAL;
+	}
+	packet->bitfields3.doorbell_offset =
+			q->properties.doorbell_off;
+
+	packet->mqd_addr_lo =
+			lower_32_bits(q->gart_mqd_addr);
+
+	packet->mqd_addr_hi =
+			upper_32_bits(q->gart_mqd_addr);
+
+	packet->wptr_addr_lo =
+			lower_32_bits((uint64_t)q->properties.write_ptr);
+
+	packet->wptr_addr_hi =
+			upper_32_bits((uint64_t)q->properties.write_ptr);
+
+	return 0;
+}
+
+static int pm_unmap_queues_vi(struct packet_manager *pm, uint32_t *buffer,
+			enum kfd_unmap_queues_filter filter,
+			uint32_t filter_param, bool reset)
+{
+	struct pm4_mes_unmap_queues *packet;
+
+	packet = (struct pm4_mes_unmap_queues *)buffer;
+	memset(buffer, 0, sizeof(struct pm4_mes_unmap_queues));
+
+	packet->header.u32All = pm_build_pm4_header(IT_UNMAP_QUEUES,
+					sizeof(struct pm4_mes_unmap_queues));
+
+	packet->bitfields2.engine_sel =
+			engine_sel__mes_unmap_queues__compute;
+
+	if (reset)
+		packet->bitfields2.action =
+			action__mes_unmap_queues__reset_queues;
+	else
+		packet->bitfields2.action =
+			action__mes_unmap_queues__preempt_queues;
+
+	switch (filter) {
+	case KFD_UNMAP_QUEUES_FILTER_BY_PASID:
+		packet->bitfields2.queue_sel =
+			queue_sel__mes_unmap_queues__perform_request_on_pasid_queues;
+		packet->bitfields3a.pasid = filter_param;
+		break;
+	case KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES:
+		packet->bitfields2.queue_sel =
+			queue_sel__mes_unmap_queues__unmap_all_queues;
+		break;
+	case KFD_UNMAP_QUEUES_FILTER_DYNAMIC_QUEUES:
+		/* in this case, we do not preempt static queues */
+		packet->bitfields2.queue_sel =
+			queue_sel__mes_unmap_queues__unmap_all_non_static_queues;
+		break;
+	default:
+		WARN(1, "filter %d", filter);
+		return -EINVAL;
+	}
+
+	return 0;
+
+}
+
+static int pm_query_status_vi(struct packet_manager *pm, uint32_t *buffer,
+			uint64_t fence_address,	uint64_t fence_value)
+{
+	struct pm4_mes_query_status *packet;
+
+	packet = (struct pm4_mes_query_status *)buffer;
+	memset(buffer, 0, sizeof(struct pm4_mes_query_status));
+
+	packet->header.u32All = pm_build_pm4_header(IT_QUERY_STATUS,
+					sizeof(struct pm4_mes_query_status));
+
+	packet->bitfields2.context_id = 0;
+	packet->bitfields2.interrupt_sel =
+			interrupt_sel__mes_query_status__completion_status;
+	packet->bitfields2.command =
+			command__mes_query_status__fence_only_after_write_ack;
+
+	packet->addr_hi = upper_32_bits((uint64_t)fence_address);
+	packet->addr_lo = lower_32_bits((uint64_t)fence_address);
+	packet->data_hi = upper_32_bits((uint64_t)fence_value);
+	packet->data_lo = lower_32_bits((uint64_t)fence_value);
+
+	return 0;
+}
+
+static int pm_release_mem_vi(uint64_t gpu_addr, uint32_t *buffer)
+{
+	struct pm4_mec_release_mem *packet;
+
+	packet = (struct pm4_mec_release_mem *)buffer;
+	memset(buffer, 0, sizeof(*packet));
+
+	packet->header.u32All = pm_build_pm4_header(IT_RELEASE_MEM,
+						 sizeof(*packet));
+
+	packet->bitfields2.event_type = CACHE_FLUSH_AND_INV_TS_EVENT;
+	packet->bitfields2.event_index = event_index___release_mem__end_of_pipe;
+	packet->bitfields2.tcl1_action_ena = 1;
+	packet->bitfields2.tc_action_ena = 1;
+	packet->bitfields2.cache_policy = cache_policy___release_mem__lru;
+	packet->bitfields2.atc = 0;
+
+	packet->bitfields3.data_sel = data_sel___release_mem__send_32_bit_low;
+	packet->bitfields3.int_sel =
+		int_sel___release_mem__send_interrupt_after_write_confirm;
+
+	packet->bitfields4.address_lo_32b = (gpu_addr & 0xffffffff) >> 2;
+	packet->address_hi = upper_32_bits(gpu_addr);
+
+	packet->data_lo = 0;
+
+	return 0;
+}
+
+const struct packet_manager_funcs kfd_vi_pm_funcs = {
+	.map_process		= pm_map_process_vi,
+	.runlist		= pm_runlist_vi,
+	.set_resources		= pm_set_resources_vi,
+	.map_queues		= pm_map_queues_vi,
+	.unmap_queues		= pm_unmap_queues_vi,
+	.config_dequeue_wait_counts	= NULL,
+	.query_status		= pm_query_status_vi,
+	.release_mem		= pm_release_mem_vi,
+	.map_process_size	= sizeof(struct pm4_mes_map_process),
+	.runlist_size		= sizeof(struct pm4_mes_runlist),
+	.set_resources_size	= sizeof(struct pm4_mes_set_resources),
+	.map_queues_size	= sizeof(struct pm4_mes_map_queues),
+	.unmap_queues_size	= sizeof(struct pm4_mes_unmap_queues),
+	.config_dequeue_wait_counts_size	= 0,
+	.query_status_size	= sizeof(struct pm4_mes_query_status),
+	.release_mem_size	= sizeof(struct pm4_mec_release_mem)
+};
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_pasid.c b/drivers/gpu/drm/amd/amdkfd/kfd_pasid.c
deleted file mode 100644
index 1e06de0bc673..000000000000
--- a/drivers/gpu/drm/amd/amdkfd/kfd_pasid.c
+++ /dev/null
@@ -1,97 +0,0 @@
-/*
- * Copyright 2014 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- */
-
-#include <linux/slab.h>
-#include <linux/types.h>
-#include "kfd_priv.h"
-
-static unsigned long *pasid_bitmap;
-static unsigned int pasid_limit;
-static DEFINE_MUTEX(pasid_mutex);
-
-int kfd_pasid_init(void)
-{
-	pasid_limit = KFD_MAX_NUM_OF_PROCESSES;
-
-	pasid_bitmap = kcalloc(BITS_TO_LONGS(pasid_limit), sizeof(long),
-				GFP_KERNEL);
-	if (!pasid_bitmap)
-		return -ENOMEM;
-
-	set_bit(0, pasid_bitmap); /* PASID 0 is reserved. */
-
-	return 0;
-}
-
-void kfd_pasid_exit(void)
-{
-	kfree(pasid_bitmap);
-}
-
-bool kfd_set_pasid_limit(unsigned int new_limit)
-{
-	if (new_limit < pasid_limit) {
-		bool ok;
-
-		mutex_lock(&pasid_mutex);
-
-		/* ensure that no pasids >= new_limit are in-use */
-		ok = (find_next_bit(pasid_bitmap, pasid_limit, new_limit) ==
-								pasid_limit);
-		if (ok)
-			pasid_limit = new_limit;
-
-		mutex_unlock(&pasid_mutex);
-
-		return ok;
-	}
-
-	return true;
-}
-
-inline unsigned int kfd_get_pasid_limit(void)
-{
-	return pasid_limit;
-}
-
-unsigned int kfd_pasid_alloc(void)
-{
-	unsigned int found;
-
-	mutex_lock(&pasid_mutex);
-
-	found = find_first_zero_bit(pasid_bitmap, pasid_limit);
-	if (found == pasid_limit)
-		found = 0;
-	else
-		set_bit(found, pasid_bitmap);
-
-	mutex_unlock(&pasid_mutex);
-
-	return found;
-}
-
-void kfd_pasid_free(unsigned int pasid)
-{
-	if (!WARN_ON(pasid == 0 || pasid >= pasid_limit))
-		clear_bit(pasid, pasid_bitmap);
-}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_pm4_headers.h b/drivers/gpu/drm/amd/amdkfd/kfd_pm4_headers.h
index e50f73d25de6..7274edfd3f38 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_pm4_headers.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_pm4_headers.h
@@ -1,5 +1,6 @@
+/* SPDX-License-Identifier: GPL-2.0 OR MIT */
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_pm4_headers_ai.h b/drivers/gpu/drm/amd/amdkfd/kfd_pm4_headers_ai.h
new file mode 100644
index 000000000000..e356a207d03c
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_pm4_headers_ai.h
@@ -0,0 +1,661 @@
+/* SPDX-License-Identifier: GPL-2.0 OR MIT */
+/*
+ * Copyright 2016-2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef F32_MES_PM4_PACKETS_H
+#define F32_MES_PM4_PACKETS_H
+
+#ifndef PM4_MES_HEADER_DEFINED
+#define PM4_MES_HEADER_DEFINED
+union PM4_MES_TYPE_3_HEADER {
+	struct {
+		uint32_t reserved1 : 8; /* < reserved */
+		uint32_t opcode    : 8; /* < IT opcode */
+		uint32_t count     : 14;/* < number of DWORDs - 1 in the
+					 *   information body.
+					 */
+		uint32_t type      : 2; /* < packet identifier.
+					 *   It should be 3 for type 3 packets
+					 */
+	};
+	uint32_t u32All;
+};
+#endif /* PM4_MES_HEADER_DEFINED */
+
+/*--------------------MES_SET_RESOURCES--------------------*/
+
+#ifndef PM4_MES_SET_RESOURCES_DEFINED
+#define PM4_MES_SET_RESOURCES_DEFINED
+enum mes_set_resources_queue_type_enum {
+	queue_type__mes_set_resources__kernel_interface_queue_kiq = 0,
+	queue_type__mes_set_resources__hsa_interface_queue_hiq = 1,
+	queue_type__mes_set_resources__hsa_debug_interface_queue = 4
+};
+
+
+struct pm4_mes_set_resources {
+	union {
+		union PM4_MES_TYPE_3_HEADER	header;		/* header */
+		uint32_t			ordinal1;
+	};
+
+	union {
+		struct {
+			uint32_t vmid_mask:16;
+			uint32_t unmap_latency:8;
+			uint32_t reserved1:4;
+			uint32_t enb_xnack_retry_disable_check:1;
+			enum mes_set_resources_queue_type_enum queue_type:3;
+		} bitfields2;
+		uint32_t ordinal2;
+	};
+
+	uint32_t queue_mask_lo;
+	uint32_t queue_mask_hi;
+	uint32_t gws_mask_lo;
+	uint32_t gws_mask_hi;
+
+	union {
+		struct {
+			uint32_t oac_mask:16;
+			uint32_t reserved2:16;
+		} bitfields7;
+		uint32_t ordinal7;
+	};
+
+	union {
+		struct {
+		uint32_t gds_heap_base:10;
+		uint32_t reserved3:1;
+		uint32_t gds_heap_size:10;
+		uint32_t reserved4:11;
+		} bitfields8;
+		uint32_t ordinal8;
+	};
+
+};
+#endif
+
+/*--------------------MES_RUN_LIST--------------------*/
+
+#ifndef PM4_MES_RUN_LIST_DEFINED
+#define PM4_MES_RUN_LIST_DEFINED
+
+struct pm4_mes_runlist {
+	union {
+		union PM4_MES_TYPE_3_HEADER header; /* header */
+		uint32_t ordinal1;
+	};
+
+	union {
+		struct {
+			uint32_t reserved1:2;
+			uint32_t ib_base_lo:30;
+		} bitfields2;
+		uint32_t ordinal2;
+	};
+
+	uint32_t ib_base_hi;
+
+	union {
+		struct {
+			uint32_t ib_size:20;
+			uint32_t chain:1;
+			uint32_t offload_polling:1;
+			uint32_t chained_runlist_idle_disable:1;
+			uint32_t valid:1;
+			uint32_t process_cnt:4;
+			uint32_t reserved3:4;
+		} bitfields4;
+		uint32_t ordinal4;
+	};
+
+};
+#endif
+
+/*--------------------MES_MAP_PROCESS--------------------*/
+
+#ifndef PM4_MES_MAP_PROCESS_DEFINED
+#define PM4_MES_MAP_PROCESS_DEFINED
+
+struct pm4_mes_map_process {
+	union {
+		union PM4_MES_TYPE_3_HEADER header;	/* header */
+		uint32_t ordinal1;
+	};
+
+	union {
+		struct {
+			uint32_t pasid:16;		/* 0 - 15  */
+			uint32_t reserved1:1;		/* 16      */
+			uint32_t exec_cleaner_shader:1;	/* 17      */
+			uint32_t debug_vmid:4;
+			uint32_t new_debug:1;
+			uint32_t reserved2:1;
+			uint32_t diq_enable:1;
+			uint32_t process_quantum:7;
+		} bitfields2;
+		uint32_t ordinal2;
+	};
+
+	uint32_t vm_context_page_table_base_addr_lo32;
+
+	uint32_t vm_context_page_table_base_addr_hi32;
+
+	uint32_t sh_mem_bases;
+
+	uint32_t sh_mem_config;
+
+	uint32_t sq_shader_tba_lo;
+
+	uint32_t sq_shader_tba_hi;
+
+	uint32_t sq_shader_tma_lo;
+
+	uint32_t sq_shader_tma_hi;
+
+	uint32_t reserved6;
+
+	uint32_t gds_addr_lo;
+
+	uint32_t gds_addr_hi;
+
+	union {
+		struct {
+			uint32_t num_gws:7;
+			uint32_t sdma_enable:1;
+			uint32_t num_oac:4;
+			uint32_t gds_size_hi:4;
+			uint32_t gds_size:6;
+			uint32_t num_queues:10;
+		} bitfields14;
+		uint32_t ordinal14;
+	};
+
+	uint32_t completion_signal_lo;
+
+	uint32_t completion_signal_hi;
+
+};
+
+#endif
+
+/*--------------------MES_MAP_PROCESS_VM--------------------*/
+
+#ifndef PM4_MES_MAP_PROCESS_VM_DEFINED
+#define PM4_MES_MAP_PROCESS_VM_DEFINED
+
+struct PM4_MES_MAP_PROCESS_VM {
+	union {
+		union PM4_MES_TYPE_3_HEADER header;	/* header */
+		uint32_t ordinal1;
+	};
+
+	uint32_t reserved1;
+
+	uint32_t vm_context_cntl;
+
+	uint32_t reserved2;
+
+	uint32_t vm_context_page_table_end_addr_lo32;
+
+	uint32_t vm_context_page_table_end_addr_hi32;
+
+	uint32_t vm_context_page_table_start_addr_lo32;
+
+	uint32_t vm_context_page_table_start_addr_hi32;
+
+	uint32_t reserved3;
+
+	uint32_t reserved4;
+
+	uint32_t reserved5;
+
+	uint32_t reserved6;
+
+	uint32_t reserved7;
+
+	uint32_t reserved8;
+
+	uint32_t completion_signal_lo32;
+
+	uint32_t completion_signal_hi32;
+
+};
+#endif
+
+/*--------------------MES_MAP_QUEUES--------------------*/
+
+#ifndef PM4_MES_MAP_QUEUES_VI_DEFINED
+#define PM4_MES_MAP_QUEUES_VI_DEFINED
+enum mes_map_queues_queue_sel_enum {
+	queue_sel__mes_map_queues__map_to_specified_queue_slots_vi = 0,
+queue_sel__mes_map_queues__map_to_hws_determined_queue_slots_vi = 1
+};
+
+enum mes_map_queues_queue_type_enum {
+	queue_type__mes_map_queues__normal_compute_vi = 0,
+	queue_type__mes_map_queues__debug_interface_queue_vi = 1,
+	queue_type__mes_map_queues__normal_latency_static_queue_vi = 2,
+queue_type__mes_map_queues__low_latency_static_queue_vi = 3
+};
+
+enum mes_map_queues_engine_sel_enum {
+	engine_sel__mes_map_queues__compute_vi = 0,
+	engine_sel__mes_map_queues__sdma0_vi = 2,
+	engine_sel__mes_map_queues__sdma1_vi = 3
+};
+
+enum mes_map_queues_extended_engine_sel_enum {
+	extended_engine_sel__mes_map_queues__legacy_engine_sel = 0,
+	extended_engine_sel__mes_map_queues__sdma0_to_7_sel  = 1,
+	extended_engine_sel__mes_map_queues__sdma8_to_15_sel = 2
+};
+
+struct pm4_mes_map_queues {
+	union {
+		union PM4_MES_TYPE_3_HEADER   header;            /* header */
+		uint32_t            ordinal1;
+	};
+
+	union {
+		struct {
+			uint32_t reserved1:2;
+			enum mes_map_queues_extended_engine_sel_enum extended_engine_sel:2;
+			enum mes_map_queues_queue_sel_enum queue_sel:2;
+			uint32_t reserved5:6;
+			uint32_t gws_control_queue:1;
+			uint32_t reserved2:8;
+			enum mes_map_queues_queue_type_enum queue_type:3;
+			uint32_t reserved3:2;
+			enum mes_map_queues_engine_sel_enum engine_sel:3;
+			uint32_t num_queues:3;
+		} bitfields2;
+		uint32_t ordinal2;
+	};
+
+	union {
+		struct {
+			uint32_t reserved3:1;
+			uint32_t check_disable:1;
+			uint32_t doorbell_offset:26;
+			uint32_t reserved4:4;
+		} bitfields3;
+		uint32_t ordinal3;
+	};
+
+	uint32_t mqd_addr_lo;
+	uint32_t mqd_addr_hi;
+	uint32_t wptr_addr_lo;
+	uint32_t wptr_addr_hi;
+};
+#endif
+
+/*--------------------MES_QUERY_STATUS--------------------*/
+
+#ifndef PM4_MES_QUERY_STATUS_DEFINED
+#define PM4_MES_QUERY_STATUS_DEFINED
+enum mes_query_status_interrupt_sel_enum {
+	interrupt_sel__mes_query_status__completion_status = 0,
+	interrupt_sel__mes_query_status__process_status = 1,
+	interrupt_sel__mes_query_status__queue_status = 2
+};
+
+enum mes_query_status_command_enum {
+	command__mes_query_status__interrupt_only = 0,
+	command__mes_query_status__fence_only_immediate = 1,
+	command__mes_query_status__fence_only_after_write_ack = 2,
+	command__mes_query_status__fence_wait_for_write_ack_send_interrupt = 3
+};
+
+enum mes_query_status_engine_sel_enum {
+	engine_sel__mes_query_status__compute = 0,
+	engine_sel__mes_query_status__sdma0_queue = 2,
+	engine_sel__mes_query_status__sdma1_queue = 3
+};
+
+struct pm4_mes_query_status {
+	union {
+		union PM4_MES_TYPE_3_HEADER   header;            /* header */
+		uint32_t            ordinal1;
+	};
+
+	union {
+		struct {
+			uint32_t context_id:28;
+			enum mes_query_status_interrupt_sel_enum	interrupt_sel:2;
+			enum mes_query_status_command_enum command:2;
+		} bitfields2;
+		uint32_t ordinal2;
+	};
+
+	union {
+		struct {
+			uint32_t pasid:16;
+			uint32_t reserved1:16;
+		} bitfields3a;
+		struct {
+			uint32_t reserved2:2;
+			uint32_t doorbell_offset:26;
+			enum mes_query_status_engine_sel_enum engine_sel:3;
+			uint32_t reserved3:1;
+		} bitfields3b;
+		uint32_t ordinal3;
+	};
+
+	uint32_t addr_lo;
+	uint32_t addr_hi;
+	uint32_t data_lo;
+	uint32_t data_hi;
+};
+#endif
+
+/*--------------------MES_UNMAP_QUEUES--------------------*/
+
+#ifndef PM4_MES_UNMAP_QUEUES_DEFINED
+#define PM4_MES_UNMAP_QUEUES_DEFINED
+enum mes_unmap_queues_action_enum {
+	action__mes_unmap_queues__preempt_queues = 0,
+	action__mes_unmap_queues__reset_queues = 1,
+	action__mes_unmap_queues__disable_process_queues = 2,
+	action__mes_unmap_queues__reserved = 3
+};
+
+enum mes_unmap_queues_queue_sel_enum {
+	queue_sel__mes_unmap_queues__perform_request_on_specified_queues = 0,
+	queue_sel__mes_unmap_queues__perform_request_on_pasid_queues = 1,
+	queue_sel__mes_unmap_queues__unmap_all_queues = 2,
+	queue_sel__mes_unmap_queues__unmap_all_non_static_queues = 3
+};
+
+enum mes_unmap_queues_engine_sel_enum {
+	engine_sel__mes_unmap_queues__compute = 0,
+	engine_sel__mes_unmap_queues__sdma0 = 2,
+	engine_sel__mes_unmap_queues__sdmal = 3
+};
+
+enum mes_unmap_queues_extended_engine_sel_enum {
+	extended_engine_sel__mes_unmap_queues__legacy_engine_sel = 0,
+	extended_engine_sel__mes_unmap_queues__sdma0_to_7_sel = 1
+};
+
+struct pm4_mes_unmap_queues {
+	union {
+		union PM4_MES_TYPE_3_HEADER   header;            /* header */
+		uint32_t            ordinal1;
+	};
+
+	union {
+		struct {
+			enum mes_unmap_queues_action_enum action:2;
+			enum mes_unmap_queues_extended_engine_sel_enum extended_engine_sel:2;
+			enum mes_unmap_queues_queue_sel_enum queue_sel:2;
+			uint32_t reserved2:20;
+			enum mes_unmap_queues_engine_sel_enum engine_sel:3;
+			uint32_t num_queues:3;
+		} bitfields2;
+		uint32_t ordinal2;
+	};
+
+	union {
+		struct {
+			uint32_t pasid:16;
+			uint32_t reserved3:16;
+		} bitfields3a;
+		struct {
+			uint32_t reserved4:2;
+			uint32_t doorbell_offset0:26;
+			int32_t reserved5:4;
+		} bitfields3b;
+		uint32_t ordinal3;
+	};
+
+	union {
+	struct {
+			uint32_t reserved6:2;
+			uint32_t doorbell_offset1:26;
+			uint32_t reserved7:4;
+		} bitfields4;
+		uint32_t ordinal4;
+	};
+
+	union {
+		struct {
+			uint32_t reserved8:2;
+			uint32_t doorbell_offset2:26;
+			uint32_t reserved9:4;
+		} bitfields5;
+		uint32_t ordinal5;
+	};
+
+	union {
+		struct {
+			uint32_t reserved10:2;
+			uint32_t doorbell_offset3:26;
+			uint32_t reserved11:4;
+		} bitfields6;
+		uint32_t ordinal6;
+	};
+};
+#endif
+
+#ifndef PM4_MEC_RELEASE_MEM_DEFINED
+#define PM4_MEC_RELEASE_MEM_DEFINED
+
+enum mec_release_mem_event_index_enum {
+	event_index__mec_release_mem__end_of_pipe = 5,
+	event_index__mec_release_mem__shader_done = 6
+};
+
+enum mec_release_mem_cache_policy_enum {
+	cache_policy__mec_release_mem__lru = 0,
+	cache_policy__mec_release_mem__stream = 1
+};
+
+enum mec_release_mem_pq_exe_status_enum {
+	pq_exe_status__mec_release_mem__default = 0,
+	pq_exe_status__mec_release_mem__phase_update = 1
+};
+
+enum mec_release_mem_dst_sel_enum {
+	dst_sel__mec_release_mem__memory_controller = 0,
+	dst_sel__mec_release_mem__tc_l2 = 1,
+	dst_sel__mec_release_mem__queue_write_pointer_register = 2,
+	dst_sel__mec_release_mem__queue_write_pointer_poll_mask_bit = 3
+};
+
+enum mec_release_mem_int_sel_enum {
+	int_sel__mec_release_mem__none = 0,
+	int_sel__mec_release_mem__send_interrupt_only = 1,
+	int_sel__mec_release_mem__send_interrupt_after_write_confirm = 2,
+	int_sel__mec_release_mem__send_data_after_write_confirm = 3,
+	int_sel__mec_release_mem__unconditionally_send_int_ctxid = 4,
+	int_sel__mec_release_mem__conditionally_send_int_ctxid_based_on_32_bit_compare = 5,
+	int_sel__mec_release_mem__conditionally_send_int_ctxid_based_on_64_bit_compare = 6
+};
+
+enum mec_release_mem_data_sel_enum {
+	data_sel__mec_release_mem__none = 0,
+	data_sel__mec_release_mem__send_32_bit_low = 1,
+	data_sel__mec_release_mem__send_64_bit_data = 2,
+	data_sel__mec_release_mem__send_gpu_clock_counter = 3,
+	data_sel__mec_release_mem__send_cp_perfcounter_hi_lo = 4,
+	data_sel__mec_release_mem__store_gds_data_to_memory = 5
+};
+
+struct pm4_mec_release_mem {
+	union {
+		union PM4_MES_TYPE_3_HEADER header;     /*header */
+		unsigned int ordinal1;
+	};
+
+	union {
+		struct {
+			unsigned int event_type:6;
+			unsigned int reserved1:2;
+			enum mec_release_mem_event_index_enum event_index:4;
+			unsigned int tcl1_vol_action_ena:1;
+			unsigned int tc_vol_action_ena:1;
+			unsigned int reserved2:1;
+			unsigned int tc_wb_action_ena:1;
+			unsigned int tcl1_action_ena:1;
+			unsigned int tc_action_ena:1;
+			uint32_t reserved3:1;
+			uint32_t tc_nc_action_ena:1;
+			uint32_t tc_wc_action_ena:1;
+			uint32_t tc_md_action_ena:1;
+			uint32_t reserved4:3;
+			enum mec_release_mem_cache_policy_enum cache_policy:2;
+			uint32_t reserved5:2;
+			enum mec_release_mem_pq_exe_status_enum pq_exe_status:1;
+			uint32_t reserved6:2;
+		} bitfields2;
+		unsigned int ordinal2;
+	};
+
+	union {
+		struct {
+			uint32_t reserved7:16;
+			enum mec_release_mem_dst_sel_enum dst_sel:2;
+			uint32_t reserved8:6;
+			enum mec_release_mem_int_sel_enum int_sel:3;
+			uint32_t reserved9:2;
+			enum mec_release_mem_data_sel_enum data_sel:3;
+		} bitfields3;
+		unsigned int ordinal3;
+	};
+
+	union {
+		struct {
+			uint32_t reserved10:2;
+			unsigned int address_lo_32b:30;
+		} bitfields4;
+		struct {
+			uint32_t reserved11:3;
+			uint32_t address_lo_64b:29;
+		} bitfields4b;
+		uint32_t reserved12;
+		unsigned int ordinal4;
+	};
+
+	union {
+		uint32_t address_hi;
+		uint32_t reserved13;
+		uint32_t ordinal5;
+	};
+
+	union {
+		uint32_t data_lo;
+		uint32_t cmp_data_lo;
+		struct {
+			uint32_t dw_offset:16;
+			uint32_t num_dwords:16;
+		} bitfields6c;
+		uint32_t reserved14;
+		uint32_t ordinal6;
+	};
+
+	union {
+		uint32_t data_hi;
+		uint32_t cmp_data_hi;
+		uint32_t reserved15;
+		uint32_t reserved16;
+		uint32_t ordinal7;
+	};
+
+	uint32_t int_ctxid;
+
+};
+
+#endif
+
+#ifndef PM4_MEC_WRITE_DATA_DEFINED
+#define PM4_MEC_WRITE_DATA_DEFINED
+
+enum WRITE_DATA_dst_sel_enum {
+	dst_sel___write_data__mem_mapped_register = 0,
+	dst_sel___write_data__tc_l2 = 2,
+	dst_sel___write_data__gds = 3,
+	dst_sel___write_data__memory = 5,
+	dst_sel___write_data__memory_mapped_adc_persistent_state = 6,
+};
+
+enum WRITE_DATA_addr_incr_enum {
+	addr_incr___write_data__increment_address = 0,
+	addr_incr___write_data__do_not_increment_address = 1
+};
+
+enum WRITE_DATA_wr_confirm_enum {
+	wr_confirm___write_data__do_not_wait_for_write_confirmation = 0,
+	wr_confirm___write_data__wait_for_write_confirmation = 1
+};
+
+enum WRITE_DATA_cache_policy_enum {
+	cache_policy___write_data__lru = 0,
+	cache_policy___write_data__stream = 1
+};
+
+
+struct pm4_mec_write_data_mmio {
+	union {
+		union PM4_MES_TYPE_3_HEADER header;     /*header */
+		unsigned int ordinal1;
+	};
+
+	union {
+		struct {
+			unsigned int reserved1:8;
+			unsigned int dst_sel:4;
+			unsigned int reserved2:4;
+			unsigned int addr_incr:1;
+			unsigned int reserved3:2;
+			unsigned int resume_vf:1;
+			unsigned int wr_confirm:1;
+			unsigned int reserved4:4;
+			unsigned int cache_policy:2;
+			unsigned int reserved5:5;
+		} bitfields2;
+		unsigned int ordinal2;
+	};
+
+	union {
+		struct {
+			unsigned int dst_mmreg_addr:18;
+			unsigned int reserved6:14;
+		} bitfields3;
+		unsigned int ordinal3;
+	};
+
+	uint32_t reserved7;
+
+	uint32_t data;
+
+};
+
+#endif
+
+enum {
+	CACHE_FLUSH_AND_INV_TS_EVENT = 0x00000014
+};
+#endif
+
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_pm4_headers_aldebaran.h b/drivers/gpu/drm/amd/amdkfd/kfd_pm4_headers_aldebaran.h
new file mode 100644
index 000000000000..e0ed62c4ade0
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_pm4_headers_aldebaran.h
@@ -0,0 +1,94 @@
+/* SPDX-License-Identifier: GPL-2.0 OR MIT */
+/*
+ * Copyright 2020-2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+/*--------------------MES_MAP_PROCESS (PER DEBUG VMID)--------------------*/
+
+#ifndef PM4_MES_MAP_PROCESS_PER_DEBUG_VMID_DEFINED
+#define PM4_MES_MAP_PROCESS_PER_DEBUG_VMID_DEFINED
+
+struct pm4_mes_map_process_aldebaran {
+	union {
+		union PM4_MES_TYPE_3_HEADER header;	/* header */
+		uint32_t ordinal1;
+	};
+
+	union {
+		struct {
+			uint32_t pasid:16;	    /* 0 - 15  */
+			uint32_t single_memops:1;   /* 16      */
+			uint32_t exec_cleaner_shader:1;	    /* 17      */
+			uint32_t debug_vmid:4;	    /* 18 - 21 */
+			uint32_t new_debug:1;	    /* 22      */
+			uint32_t tmz:1;		    /* 23      */
+			uint32_t diq_enable:1;      /* 24      */
+			uint32_t process_quantum:7; /* 25 - 31 */
+		} bitfields2;
+		uint32_t ordinal2;
+	};
+
+	uint32_t vm_context_page_table_base_addr_lo32;
+
+	uint32_t vm_context_page_table_base_addr_hi32;
+
+	uint32_t sh_mem_bases;
+
+	uint32_t sh_mem_config;
+
+	uint32_t sq_shader_tba_lo;
+
+	uint32_t sq_shader_tba_hi;
+
+	uint32_t sq_shader_tma_lo;
+
+	uint32_t sq_shader_tma_hi;
+
+	uint32_t reserved6;
+
+	uint32_t gds_addr_lo;
+
+	uint32_t gds_addr_hi;
+
+	union {
+		struct {
+			uint32_t num_gws:7;
+			uint32_t sdma_enable:1;
+			uint32_t num_oac:4;
+			uint32_t gds_size_hi:4;
+			uint32_t gds_size:6;
+			uint32_t num_queues:10;
+		} bitfields14;
+		uint32_t ordinal14;
+	};
+
+	uint32_t spi_gdbg_per_vmid_cntl;
+
+	uint32_t tcp_watch_cntl[4];
+
+	uint32_t completion_signal_lo;
+
+	uint32_t completion_signal_hi;
+
+};
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_pm4_headers_diq.h b/drivers/gpu/drm/amd/amdkfd/kfd_pm4_headers_diq.h
deleted file mode 100644
index a0ff34878163..000000000000
--- a/drivers/gpu/drm/amd/amdkfd/kfd_pm4_headers_diq.h
+++ /dev/null
@@ -1,290 +0,0 @@
-/*
- * Copyright 2014 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-
-#ifndef KFD_PM4_HEADERS_DIQ_H_
-#define KFD_PM4_HEADERS_DIQ_H_
-
-/*--------------------_INDIRECT_BUFFER-------------------- */
-
-#ifndef _PM4__INDIRECT_BUFFER_DEFINED
-#define _PM4__INDIRECT_BUFFER_DEFINED
-enum _INDIRECT_BUFFER_cache_policy_enum {
-	cache_policy___indirect_buffer__lru = 0,
-	cache_policy___indirect_buffer__stream = 1,
-	cache_policy___indirect_buffer__bypass = 2
-};
-
-enum {
-	IT_INDIRECT_BUFFER_PASID = 0x5C
-};
-
-struct pm4__indirect_buffer_pasid {
-	union {
-		union PM4_MES_TYPE_3_HEADER header;	/* header */
-		unsigned int ordinal1;
-	};
-
-	union {
-		struct {
-			unsigned int reserved1:2;
-			unsigned int ib_base_lo:30;
-		} bitfields2;
-		unsigned int ordinal2;
-	};
-
-	union {
-		struct {
-			unsigned int ib_base_hi:16;
-			unsigned int reserved2:16;
-		} bitfields3;
-		unsigned int ordinal3;
-	};
-
-	union {
-		unsigned int control;
-		unsigned int ordinal4;
-	};
-
-	union {
-		struct {
-			unsigned int pasid:10;
-			unsigned int reserved4:22;
-		} bitfields5;
-		unsigned int ordinal5;
-	};
-
-};
-
-#endif
-
-/*--------------------_RELEASE_MEM-------------------- */
-
-#ifndef _PM4__RELEASE_MEM_DEFINED
-#define _PM4__RELEASE_MEM_DEFINED
-enum _RELEASE_MEM_event_index_enum {
-	event_index___release_mem__end_of_pipe = 5,
-	event_index___release_mem__shader_done = 6
-};
-
-enum _RELEASE_MEM_cache_policy_enum {
-	cache_policy___release_mem__lru = 0,
-	cache_policy___release_mem__stream = 1,
-	cache_policy___release_mem__bypass = 2
-};
-
-enum _RELEASE_MEM_dst_sel_enum {
-	dst_sel___release_mem__memory_controller = 0,
-	dst_sel___release_mem__tc_l2 = 1,
-	dst_sel___release_mem__queue_write_pointer_register = 2,
-	dst_sel___release_mem__queue_write_pointer_poll_mask_bit = 3
-};
-
-enum _RELEASE_MEM_int_sel_enum {
-	int_sel___release_mem__none = 0,
-	int_sel___release_mem__send_interrupt_only = 1,
-	int_sel___release_mem__send_interrupt_after_write_confirm = 2,
-	int_sel___release_mem__send_data_after_write_confirm = 3
-};
-
-enum _RELEASE_MEM_data_sel_enum {
-	data_sel___release_mem__none = 0,
-	data_sel___release_mem__send_32_bit_low = 1,
-	data_sel___release_mem__send_64_bit_data = 2,
-	data_sel___release_mem__send_gpu_clock_counter = 3,
-	data_sel___release_mem__send_cp_perfcounter_hi_lo = 4,
-	data_sel___release_mem__store_gds_data_to_memory = 5
-};
-
-struct pm4__release_mem {
-	union {
-		union PM4_MES_TYPE_3_HEADER header;	/*header */
-		unsigned int ordinal1;
-	};
-
-	union {
-		struct {
-			unsigned int event_type:6;
-			unsigned int reserved1:2;
-			enum _RELEASE_MEM_event_index_enum event_index:4;
-			unsigned int tcl1_vol_action_ena:1;
-			unsigned int tc_vol_action_ena:1;
-			unsigned int reserved2:1;
-			unsigned int tc_wb_action_ena:1;
-			unsigned int tcl1_action_ena:1;
-			unsigned int tc_action_ena:1;
-			unsigned int reserved3:6;
-			unsigned int atc:1;
-			enum _RELEASE_MEM_cache_policy_enum cache_policy:2;
-			unsigned int reserved4:5;
-		} bitfields2;
-		unsigned int ordinal2;
-	};
-
-	union {
-		struct {
-			unsigned int reserved5:16;
-			enum _RELEASE_MEM_dst_sel_enum dst_sel:2;
-			unsigned int reserved6:6;
-			enum _RELEASE_MEM_int_sel_enum int_sel:3;
-			unsigned int reserved7:2;
-			enum _RELEASE_MEM_data_sel_enum data_sel:3;
-		} bitfields3;
-		unsigned int ordinal3;
-	};
-
-	union {
-		struct {
-			unsigned int reserved8:2;
-			unsigned int address_lo_32b:30;
-		} bitfields4;
-		struct {
-			unsigned int reserved9:3;
-			unsigned int address_lo_64b:29;
-		} bitfields5;
-		unsigned int ordinal4;
-	};
-
-	unsigned int address_hi;
-
-	unsigned int data_lo;
-
-	unsigned int data_hi;
-
-};
-#endif
-
-
-/*--------------------_SET_CONFIG_REG-------------------- */
-
-#ifndef _PM4__SET_CONFIG_REG_DEFINED
-#define _PM4__SET_CONFIG_REG_DEFINED
-
-struct pm4__set_config_reg {
-	union {
-		union PM4_MES_TYPE_3_HEADER header;	/*header */
-		unsigned int ordinal1;
-	};
-
-	union {
-		struct {
-			unsigned int reg_offset:16;
-			unsigned int reserved1:7;
-			unsigned int vmid_shift:5;
-			unsigned int insert_vmid:1;
-			unsigned int reserved2:3;
-		} bitfields2;
-		unsigned int ordinal2;
-	};
-
-	unsigned int reg_data[1];	/*1..N of these fields */
-
-};
-#endif
-
-/*--------------------_WAIT_REG_MEM-------------------- */
-
-#ifndef _PM4__WAIT_REG_MEM_DEFINED
-#define _PM4__WAIT_REG_MEM_DEFINED
-enum _WAIT_REG_MEM_function_enum {
-	function___wait_reg_mem__always_pass = 0,
-	function___wait_reg_mem__less_than_ref_value = 1,
-	function___wait_reg_mem__less_than_equal_to_the_ref_value = 2,
-	function___wait_reg_mem__equal_to_the_reference_value = 3,
-	function___wait_reg_mem__not_equal_reference_value = 4,
-	function___wait_reg_mem__greater_than_or_equal_reference_value = 5,
-	function___wait_reg_mem__greater_than_reference_value = 6,
-	function___wait_reg_mem__reserved = 7
-};
-
-enum _WAIT_REG_MEM_mem_space_enum {
-	mem_space___wait_reg_mem__register_space = 0,
-	mem_space___wait_reg_mem__memory_space = 1
-};
-
-enum _WAIT_REG_MEM_operation_enum {
-	operation___wait_reg_mem__wait_reg_mem = 0,
-	operation___wait_reg_mem__wr_wait_wr_reg = 1
-};
-
-struct pm4__wait_reg_mem {
-	union {
-		union PM4_MES_TYPE_3_HEADER header;	/*header */
-		unsigned int ordinal1;
-	};
-
-	union {
-		struct {
-			enum _WAIT_REG_MEM_function_enum function:3;
-			unsigned int reserved1:1;
-			enum _WAIT_REG_MEM_mem_space_enum mem_space:2;
-			enum _WAIT_REG_MEM_operation_enum operation:2;
-			unsigned int reserved2:24;
-		} bitfields2;
-		unsigned int ordinal2;
-	};
-
-	union {
-		struct {
-			unsigned int reserved3:2;
-			unsigned int memory_poll_addr_lo:30;
-		} bitfields3;
-		struct {
-			unsigned int register_poll_addr:16;
-			unsigned int reserved4:16;
-		} bitfields4;
-		struct {
-			unsigned int register_write_addr:16;
-			unsigned int reserved5:16;
-		} bitfields5;
-		unsigned int ordinal3;
-	};
-
-	union {
-		struct {
-			unsigned int poll_address_hi:16;
-			unsigned int reserved6:16;
-		} bitfields6;
-		struct {
-			unsigned int register_write_addr:16;
-			unsigned int reserved7:16;
-		} bitfields7;
-		unsigned int ordinal4;
-	};
-
-	unsigned int reference;
-
-	unsigned int mask;
-
-	union {
-		struct {
-			unsigned int poll_interval:16;
-			unsigned int reserved8:16;
-		} bitfields8;
-		unsigned int ordinal7;
-	};
-
-};
-#endif
-
-
-#endif /* KFD_PM4_HEADERS_DIQ_H_ */
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_pm4_headers_vi.h b/drivers/gpu/drm/amd/amdkfd/kfd_pm4_headers_vi.h
index 7c8d9b357749..8147395c083b 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_pm4_headers_vi.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_pm4_headers_vi.h
@@ -1,5 +1,6 @@
+/* SPDX-License-Identifier: GPL-2.0 OR MIT */
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -101,8 +102,8 @@ struct pm4_mes_set_resources {
 
 struct pm4_mes_runlist {
 	union {
-	    union PM4_MES_TYPE_3_HEADER   header;            /* header */
-	    uint32_t            ordinal1;
+		union PM4_MES_TYPE_3_HEADER   header;            /* header */
+		uint32_t            ordinal1;
 	};
 
 	union {
@@ -216,11 +217,6 @@ enum mes_map_queues_queue_type_vi_enum {
 queue_type__mes_map_queues__low_latency_static_queue_vi = 3
 };
 
-enum mes_map_queues_alloc_format_vi_enum {
-	alloc_format__mes_map_queues__one_per_pipe_vi = 0,
-alloc_format__mes_map_queues__all_on_one_pipe_vi = 1
-};
-
 enum mes_map_queues_engine_sel_vi_enum {
 	engine_sel__mes_map_queues__compute_vi = 0,
 	engine_sel__mes_map_queues__sdma0_vi = 2,
@@ -240,7 +236,7 @@ struct pm4_mes_map_queues {
 			enum mes_map_queues_queue_sel_vi_enum queue_sel:2;
 			uint32_t reserved2:15;
 			enum mes_map_queues_queue_type_vi_enum queue_type:3;
-			enum mes_map_queues_alloc_format_vi_enum alloc_format:2;
+			uint32_t reserved3:2;
 			enum mes_map_queues_engine_sel_vi_enum engine_sel:3;
 			uint32_t num_queues:3;
 		} bitfields2;
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_pm4_opcodes.h b/drivers/gpu/drm/amd/amdkfd/kfd_pm4_opcodes.h
index b72fa3b8c2d4..5bfd0f9cbe23 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_pm4_opcodes.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_pm4_opcodes.h
@@ -1,5 +1,6 @@
+/* SPDX-License-Identifier: GPL-2.0 OR MIT */
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
index b397ec726400..70ef051511bb 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
@@ -1,5 +1,6 @@
+/* SPDX-License-Identifier: GPL-2.0 OR MIT */
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -25,22 +26,62 @@
 
 #include <linux/hashtable.h>
 #include <linux/mmu_notifier.h>
+#include <linux/memremap.h>
 #include <linux/mutex.h>
 #include <linux/types.h>
 #include <linux/atomic.h>
 #include <linux/workqueue.h>
 #include <linux/spinlock.h>
-#include <linux/kfd_ioctl.h>
+#include <uapi/linux/kfd_ioctl.h>
+#include <linux/idr.h>
+#include <linux/kfifo.h>
+#include <linux/seq_file.h>
+#include <linux/kref.h>
+#include <linux/sysfs.h>
+#include <linux/device_cgroup.h>
+#include <drm/drm_file.h>
+#include <drm/drm_drv.h>
+#include <drm/drm_device.h>
+#include <drm/drm_ioctl.h>
 #include <kgd_kfd_interface.h>
+#include <linux/swap.h>
+
+#include "amd_shared.h"
+#include "amdgpu.h"
+
+#define KFD_MAX_RING_ENTRY_SIZE	8
 
 #define KFD_SYSFS_FILE_MODE 0444
 
-#define KFD_MMAP_DOORBELL_MASK 0x8000000000000
-#define KFD_MMAP_EVENTS_MASK 0x4000000000000
+/* GPU ID hash width in bits */
+#define KFD_GPU_ID_HASH_WIDTH 16
+
+/* Use upper bits of mmap offset to store KFD driver specific information.
+ * BITS[63:62] - Encode MMAP type
+ * BITS[61:46] - Encode gpu_id. To identify to which GPU the offset belongs to
+ * BITS[45:0]  - MMAP offset value
+ *
+ * NOTE: struct vm_area_struct.vm_pgoff uses offset in pages. Hence, these
+ *  defines are w.r.t to PAGE_SIZE
+ */
+#define KFD_MMAP_TYPE_SHIFT	62
+#define KFD_MMAP_TYPE_MASK	(0x3ULL << KFD_MMAP_TYPE_SHIFT)
+#define KFD_MMAP_TYPE_DOORBELL	(0x3ULL << KFD_MMAP_TYPE_SHIFT)
+#define KFD_MMAP_TYPE_EVENTS	(0x2ULL << KFD_MMAP_TYPE_SHIFT)
+#define KFD_MMAP_TYPE_RESERVED_MEM	(0x1ULL << KFD_MMAP_TYPE_SHIFT)
+#define KFD_MMAP_TYPE_MMIO	(0x0ULL << KFD_MMAP_TYPE_SHIFT)
+
+#define KFD_MMAP_GPU_ID_SHIFT 46
+#define KFD_MMAP_GPU_ID_MASK (((1ULL << KFD_GPU_ID_HASH_WIDTH) - 1) \
+				<< KFD_MMAP_GPU_ID_SHIFT)
+#define KFD_MMAP_GPU_ID(gpu_id) ((((uint64_t)gpu_id) << KFD_MMAP_GPU_ID_SHIFT)\
+				& KFD_MMAP_GPU_ID_MASK)
+#define KFD_MMAP_GET_GPU_ID(offset)    ((offset & KFD_MMAP_GPU_ID_MASK) \
+				>> KFD_MMAP_GPU_ID_SHIFT)
 
 /*
  * When working with cp scheduler we should assign the HIQ manually or via
- * the radeon driver to a fixed hqd slot, here are the fixed HIQ hqd slot
+ * the amdgpu driver to a fixed hqd slot, here are the fixed HIQ hqd slot
  * definitions for Kaveri. In Kaveri only the first ME queues participates
  * in the cp scheduling taking that in mind we set the HIQ slot in the
  * second ME.
@@ -48,9 +89,6 @@
 #define KFD_CIK_HIQ_PIPE 4
 #define KFD_CIK_HIQ_QUEUE 0
 
-/* GPU ID hash width in bits */
-#define KFD_GPU_ID_HASH_WIDTH 16
-
 /* Macro for allocating structures */
 #define kfd_alloc_struct(ptr_to_struct)	\
 	((typeof(ptr_to_struct)) kzalloc(sizeof(*ptr_to_struct), GFP_KERNEL))
@@ -59,107 +97,236 @@
 #define KFD_MAX_NUM_OF_QUEUES_PER_PROCESS 1024
 
 /*
- * Kernel module parameter to specify maximum number of supported queues per
- * device
+ * Size of the per-process TBA+TMA buffer: 2 pages
+ *
+ * The first chunk is the TBA used for the CWSR ISA code. The second
+ * chunk is used as TMA for user-mode trap handler setup in daisy-chain mode.
  */
-extern int max_num_of_queues_per_device;
+#define KFD_CWSR_TBA_TMA_SIZE (PAGE_SIZE * 2)
+#define KFD_CWSR_TMA_OFFSET (PAGE_SIZE + 2048)
 
-#define KFD_MAX_NUM_OF_QUEUES_PER_DEVICE_DEFAULT 4096
 #define KFD_MAX_NUM_OF_QUEUES_PER_DEVICE		\
 	(KFD_MAX_NUM_OF_PROCESSES *			\
 			KFD_MAX_NUM_OF_QUEUES_PER_PROCESS)
 
 #define KFD_KERNEL_QUEUE_SIZE 2048
 
+/*  KFD_UNMAP_LATENCY_MS is the timeout CP waiting for SDMA preemption. One XCC
+ *  can be associated to 2 SDMA engines. queue_preemption_timeout_ms is the time
+ *  driver waiting for CP returning the UNMAP_QUEUE fence. Thus the math is
+ *  queue_preemption_timeout_ms = sdma_preemption_time * 2 + cp workload
+ *  The format here makes CP workload 10% of total timeout
+ */
+#define KFD_UNMAP_LATENCY_MS	\
+	((queue_preemption_timeout_ms - queue_preemption_timeout_ms / 10) >> 1)
+
+#define KFD_MAX_SDMA_QUEUES	128
+
+/*
+ * 512 = 0x200
+ * The doorbell index distance between SDMA RLC (2*i) and (2*i+1) in the
+ * same SDMA engine on SOC15, which has 8-byte doorbells for SDMA.
+ * 512 8-byte doorbell distance (i.e. one page away) ensures that SDMA RLC
+ * (2*i+1) doorbells (in terms of the lower 12 bit address) lie exactly in
+ * the OFFSET and SIZE set in registers like BIF_SDMA0_DOORBELL_RANGE.
+ */
+#define KFD_QUEUE_DOORBELL_MIRROR_OFFSET 512
+
+/**
+ * enum kfd_ioctl_flags - KFD ioctl flags
+ * Various flags that can be set in &amdkfd_ioctl_desc.flags to control how
+ * userspace can use a given ioctl.
+ */
+enum kfd_ioctl_flags {
+	/*
+	 * @KFD_IOC_FLAG_CHECKPOINT_RESTORE:
+	 * Certain KFD ioctls such as AMDKFD_IOC_CRIU_OP can potentially
+	 * perform privileged operations and load arbitrary data into MQDs and
+	 * eventually HQD registers when the queue is mapped by HWS. In order to
+	 * prevent this we should perform additional security checks.
+	 *
+	 * This is equivalent to callers with the CHECKPOINT_RESTORE capability.
+	 *
+	 * Note: Since earlier versions of docker do not support CHECKPOINT_RESTORE,
+	 * we also allow ioctls with SYS_ADMIN capability.
+	 */
+	KFD_IOC_FLAG_CHECKPOINT_RESTORE = BIT(0),
+};
+/*
+ * Kernel module parameter to specify maximum number of supported queues per
+ * device
+ */
+extern int max_num_of_queues_per_device;
+
+
 /* Kernel module parameter to specify the scheduling policy */
 extern int sched_policy;
 
 /*
+ * Kernel module parameter to specify the maximum process
+ * number per HW scheduler
+ */
+extern int hws_max_conc_proc;
+
+extern int cwsr_enable;
+
+/*
  * Kernel module parameter to specify whether to send sigterm to HSA process on
  * unhandled exception
  */
 extern int send_sigterm;
 
-/**
- * enum kfd_sched_policy
- *
- * @KFD_SCHED_POLICY_HWS: H/W scheduling policy known as command processor (cp)
- * scheduling. In this scheduling mode we're using the firmware code to
- * schedule the user mode queues and kernel queues such as HIQ and DIQ.
- * the HIQ queue is used as a special queue that dispatches the configuration
- * to the cp and the user mode queues list that are currently running.
- * the DIQ queue is a debugging queue that dispatches debugging commands to the
- * firmware.
- * in this scheduling mode user mode queues over subscription feature is
- * enabled.
- *
- * @KFD_SCHED_POLICY_HWS_NO_OVERSUBSCRIPTION: The same as above but the over
- * subscription feature disabled.
- *
- * @KFD_SCHED_POLICY_NO_HWS: no H/W scheduling policy is a mode which directly
- * set the command processor registers and sets the queues "manually". This
- * mode is used *ONLY* for debugging proposes.
- *
+/*
+ * This kernel module is used to simulate large bar machine on non-large bar
+ * enabled machines.
  */
-enum kfd_sched_policy {
-	KFD_SCHED_POLICY_HWS = 0,
-	KFD_SCHED_POLICY_HWS_NO_OVERSUBSCRIPTION,
-	KFD_SCHED_POLICY_NO_HWS
-};
+extern int debug_largebar;
+
+/* Set sh_mem_config.retry_disable on GFX v9 */
+extern int amdgpu_noretry;
+
+/* Halt if HWS hang is detected */
+extern int halt_if_hws_hang;
+
+/* Whether MEC FW support GWS barriers */
+extern bool hws_gws_support;
+
+/* Queue preemption timeout in ms */
+extern int queue_preemption_timeout_ms;
+
+/*
+ * Don't evict process queues on vm fault
+ */
+extern int amdgpu_no_queue_eviction_on_vm_fault;
+
+/* Enable eviction debug messages */
+extern bool debug_evictions;
+
+extern struct mutex kfd_processes_mutex;
 
 enum cache_policy {
 	cache_policy_coherent,
 	cache_policy_noncoherent
 };
 
-enum asic_family_type {
-	CHIP_KAVERI = 0,
-	CHIP_CARRIZO
-};
+#define KFD_GC_VERSION(dev) (amdgpu_ip_version((dev)->adev, GC_HWIP, 0))
+#define KFD_IS_SOC15(dev)   ((KFD_GC_VERSION(dev)) >= (IP_VERSION(9, 0, 1)))
+#define KFD_SUPPORT_XNACK_PER_PROCESS(dev)\
+	((KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 2)) ||	\
+	 (KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 3)) ||	\
+	 (KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 4)) ||	\
+	 (KFD_GC_VERSION(dev) == IP_VERSION(9, 5, 0)))
+
+struct kfd_node;
 
 struct kfd_event_interrupt_class {
-	bool (*interrupt_isr)(struct kfd_dev *dev,
-				const uint32_t *ih_ring_entry);
-	void (*interrupt_wq)(struct kfd_dev *dev,
-				const uint32_t *ih_ring_entry);
+	bool (*interrupt_isr)(struct kfd_node *dev,
+			const uint32_t *ih_ring_entry, uint32_t *patched_ihre,
+			bool *patched_flag);
+	void (*interrupt_wq)(struct kfd_node *dev,
+			const uint32_t *ih_ring_entry);
 };
 
 struct kfd_device_info {
-	unsigned int asic_family;
+	uint32_t gfx_target_version;
 	const struct kfd_event_interrupt_class *event_interrupt_class;
 	unsigned int max_pasid_bits;
 	unsigned int max_no_of_hqd;
+	unsigned int doorbell_size;
 	size_t ih_ring_entry_size;
 	uint8_t num_of_watch_points;
 	uint16_t mqd_size_aligned;
+	bool supports_cwsr;
+	bool needs_pci_atomics;
+	uint32_t no_atomic_fw_version;
+	unsigned int num_sdma_queues_per_engine;
+	unsigned int num_reserved_sdma_queues_per_engine;
+	DECLARE_BITMAP(reserved_sdma_queues_bitmap, KFD_MAX_SDMA_QUEUES);
 };
 
+unsigned int kfd_get_num_sdma_engines(struct kfd_node *kdev);
+unsigned int kfd_get_num_xgmi_sdma_engines(struct kfd_node *kdev);
+
 struct kfd_mem_obj {
 	uint32_t range_start;
 	uint32_t range_end;
 	uint64_t gpu_addr;
 	uint32_t *cpu_ptr;
+	void *gtt_mem;
 };
 
-struct kfd_dev {
-	struct kgd_dev *kgd;
+struct kfd_vmid_info {
+	uint32_t first_vmid_kfd;
+	uint32_t last_vmid_kfd;
+	uint32_t vmid_num_kfd;
+};
 
-	const struct kfd_device_info *device_info;
-	struct pci_dev *pdev;
+#define MAX_KFD_NODES	8
 
-	unsigned int id;		/* topology stub index */
+struct kfd_dev;
 
-	phys_addr_t doorbell_base;	/* Start of actual doorbells used by
-					 * KFD. It is aligned for mapping
-					 * into user mode
-					 */
-	size_t doorbell_id_offset;	/* Doorbell offset (from KFD doorbell
-					 * to HW doorbell, GFX reserved some
-					 * at the start)
-					 */
-	size_t doorbell_process_limit;	/* Number of processes we have doorbell
-					 * space for.
+struct kfd_node {
+	unsigned int node_id;
+	struct amdgpu_device *adev;     /* Duplicated here along with keeping
+					 * a copy in kfd_dev to save a hop
 					 */
+	const struct kfd2kgd_calls *kfd2kgd; /* Duplicated here along with
+					      * keeping a copy in kfd_dev to
+					      * save a hop
+					      */
+	struct kfd_vmid_info vm_info;
+	unsigned int id;                /* topology stub index */
+	uint32_t xcc_mask; /* Instance mask of XCCs present */
+	struct amdgpu_xcp *xcp;
+
+	/* Interrupts */
+	struct kfifo ih_fifo;
+	struct work_struct interrupt_work;
+	spinlock_t interrupt_lock;
+
+	/*
+	 * Interrupts of interest to KFD are copied
+	 * from the HW ring into a SW ring.
+	 */
+	bool interrupts_active;
+	uint32_t interrupt_bitmap; /* Only used for GFX 9.4.3 */
+
+	/* QCM Device instance */
+	struct device_queue_manager *dqm;
+
+	/* Global GWS resource shared between processes */
+	void *gws;
+
+	/* Clients watching SMI events */
+	struct list_head smi_clients;
+	spinlock_t smi_lock;
+	uint32_t reset_seq_num;
+
+	/* SRAM ECC flag */
+	atomic_t sram_ecc_flag;
+
+	/*spm process id */
+	unsigned int spm_pasid;
+
+	/* Maximum process number mapped to HW scheduler */
+	unsigned int max_proc_per_quantum;
+
+	unsigned int compute_vmid_bitmap;
+
+	struct kfd_local_mem_info local_mem_info;
+
+	struct kfd_dev *kfd;
+
+	/* Track per device allocated watch points */
+	uint32_t alloc_watch_ids;
+	spinlock_t watch_points_lock;
+};
+
+struct kfd_dev {
+	struct amdgpu_device *adev;
+
+	struct kfd_device_info device_info;
+
 	u32 __iomem *doorbell_kernel_ptr; /* This is a pointer for a doorbells
 					   * page used by kernel queue
 					   */
@@ -168,8 +335,6 @@ struct kfd_dev {
 
 	const struct kfd2kgd_calls *kfd2kgd;
 	struct mutex doorbell_mutex;
-	DECLARE_BITMAP(doorbell_available_index,
-			KFD_MAX_NUM_OF_QUEUES_PER_PROCESS);
 
 	void *gtt_mem;
 	uint64_t gtt_start_gpu_addr;
@@ -179,35 +344,47 @@ struct kfd_dev {
 	unsigned int gtt_sa_chunk_size;
 	unsigned int gtt_sa_num_of_chunks;
 
-	/* Interrupts */
-	void *interrupt_ring;
-	size_t interrupt_ring_size;
-	atomic_t interrupt_ring_rptr;
-	atomic_t interrupt_ring_wptr;
-	struct work_struct interrupt_work;
-	spinlock_t interrupt_lock;
+	bool init_complete;
 
-	/* QCM Device instance */
-	struct device_queue_manager *dqm;
+	/* Firmware versions */
+	uint16_t mec_fw_version;
+	uint16_t mec2_fw_version;
+	uint16_t sdma_fw_version;
 
-	bool init_complete;
-	/*
-	 * Interrupts of interest to KFD are copied
-	 * from the HW ring into a SW ring.
-	 */
-	bool interrupts_active;
+	/* CWSR */
+	bool cwsr_enabled;
+	const void *cwsr_isa;
+	unsigned int cwsr_isa_size;
 
-	/* Debug manager */
-	struct kfd_dbgmgr           *dbgmgr;
-};
+	/* xGMI */
+	uint64_t hive_id;
+
+	bool pci_atomic_requested;
+
+	/* Compute Profile ref. count */
+	atomic_t compute_profile;
+
+	struct ida doorbell_ida;
+	unsigned int max_doorbell_slices;
 
-/* KGD2KFD callbacks */
-void kgd2kfd_exit(void);
-struct kfd_dev *kgd2kfd_probe(struct kgd_dev *kgd,
-			struct pci_dev *pdev, const struct kfd2kgd_calls *f2g);
-bool kgd2kfd_device_init(struct kfd_dev *kfd,
-			const struct kgd2kfd_shared_resources *gpu_resources);
-void kgd2kfd_device_exit(struct kfd_dev *kfd);
+	int noretry;
+
+	struct kfd_node *nodes[MAX_KFD_NODES];
+	unsigned int num_nodes;
+
+	struct workqueue_struct *ih_wq;
+
+	/* Kernel doorbells for KFD device */
+	struct amdgpu_bo *doorbells;
+
+	/* bitmap for dynamic doorbell allocation from doorbell object */
+	unsigned long *doorbell_bitmap;
+
+	/* for dynamic partitioning */
+	int kfd_dev_lock;
+
+	atomic_t kfd_processes_count;
+};
 
 enum kfd_mempool {
 	KFD_MEMPOOL_SYSTEM_CACHEABLE = 1,
@@ -218,43 +395,48 @@ enum kfd_mempool {
 /* Character device interface */
 int kfd_chardev_init(void);
 void kfd_chardev_exit(void);
-struct device *kfd_chardev(void);
 
 /**
- * enum kfd_preempt_type_filter
+ * enum kfd_unmap_queues_filter - Enum for queue filters.
  *
- * @KFD_PREEMPT_TYPE_FILTER_SINGLE_QUEUE: Preempts single queue.
- *
- * @KFD_PRERMPT_TYPE_FILTER_ALL_QUEUES: Preempts all queues in the
+ * @KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES: Preempts all queues in the
  *						running queues list.
  *
- * @KFD_PRERMPT_TYPE_FILTER_BY_PASID: Preempts queues that belongs to
+ * @KFD_UNMAP_QUEUES_FILTER_DYNAMIC_QUEUES: Preempts all non-static queues
+ *						in the run list.
+ *
+ * @KFD_UNMAP_QUEUES_FILTER_BY_PASID: Preempts queues that belongs to
  *						specific process.
  *
  */
-enum kfd_preempt_type_filter {
-	KFD_PREEMPT_TYPE_FILTER_SINGLE_QUEUE,
-	KFD_PREEMPT_TYPE_FILTER_ALL_QUEUES,
-	KFD_PREEMPT_TYPE_FILTER_DYNAMIC_QUEUES,
-	KFD_PREEMPT_TYPE_FILTER_BY_PASID
+enum kfd_unmap_queues_filter {
+	KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES = 1,
+	KFD_UNMAP_QUEUES_FILTER_DYNAMIC_QUEUES = 2,
+	KFD_UNMAP_QUEUES_FILTER_BY_PASID = 3
 };
 
 /**
- * enum kfd_queue_type
+ * enum kfd_queue_type - Enum for various queue types.
  *
  * @KFD_QUEUE_TYPE_COMPUTE: Regular user mode queue type.
  *
- * @KFD_QUEUE_TYPE_SDMA: Sdma user mode queue type.
+ * @KFD_QUEUE_TYPE_SDMA: SDMA user mode queue type.
  *
  * @KFD_QUEUE_TYPE_HIQ: HIQ queue type.
  *
  * @KFD_QUEUE_TYPE_DIQ: DIQ queue type.
+ *
+ * @KFD_QUEUE_TYPE_SDMA_XGMI: Special SDMA queue for XGMI interface.
+ *
+ * @KFD_QUEUE_TYPE_SDMA_BY_ENG_ID:  SDMA user mode queue with target SDMA engine ID.
  */
 enum kfd_queue_type  {
 	KFD_QUEUE_TYPE_COMPUTE,
 	KFD_QUEUE_TYPE_SDMA,
 	KFD_QUEUE_TYPE_HIQ,
-	KFD_QUEUE_TYPE_DIQ
+	KFD_QUEUE_TYPE_DIQ,
+	KFD_QUEUE_TYPE_SDMA_XGMI,
+	KFD_QUEUE_TYPE_SDMA_BY_ENG_ID
 };
 
 enum kfd_queue_format {
@@ -262,6 +444,11 @@ enum kfd_queue_format {
 	KFD_QUEUE_FORMAT_AQL
 };
 
+enum KFD_QUEUE_PRIORITY {
+	KFD_QUEUE_PRIORITY_MINIMUM = 0,
+	KFD_QUEUE_PRIORITY_MAXIMUM = 15
+};
+
 /**
  * struct queue_properties
  *
@@ -288,16 +475,24 @@ enum kfd_queue_format {
  *
  * @write_ptr: Defines the number of dwords written to the ring buffer.
  *
- * @doorbell_ptr: This field aim is to notify the H/W of new packet written to
- * the queue ring buffer. This field should be similar to write_ptr and the
- * user should update this field after he updated the write_ptr.
+ * @doorbell_ptr: Notifies the H/W of new packet written to the queue ring
+ * buffer. This field should be similar to write_ptr and the user should
+ * update this field after updating the write_ptr.
  *
  * @doorbell_off: The doorbell offset in the doorbell pci-bar.
  *
  * @is_interop: Defines if this is a interop queue. Interop queue means that
  * the queue can access both graphics and compute resources.
  *
- * @is_active: Defines if the queue is active or not.
+ * @is_evicted: Defines if the queue is evicted. Only active queues
+ * are evicted, rendering them inactive.
+ *
+ * @is_active: Defines if the queue is active or not. @is_active and
+ * @is_evicted are protected by the DQM lock.
+ *
+ * @is_gws: Defines if the queue has been updated to be GWS-capable or not.
+ * @is_gws should be protected by the DQM lock, since changing it can yield the
+ * possibility of updating DQM state on number of GWS queues.
  *
  * @vmid: If the scheduling mode is no cp scheduling the field defines the vmid
  * of the queue.
@@ -306,6 +501,7 @@ enum kfd_queue_format {
  * it's user mode or kernel mode queue.
  *
  */
+
 struct queue_properties {
 	enum kfd_queue_type type;
 	enum kfd_queue_format format;
@@ -314,12 +510,19 @@ struct queue_properties {
 	uint64_t  queue_size;
 	uint32_t priority;
 	uint32_t queue_percent;
-	uint32_t *read_ptr;
-	uint32_t *write_ptr;
-	uint32_t __iomem *doorbell_ptr;
+	void __user *read_ptr;
+	void __user *write_ptr;
+	void __iomem *doorbell_ptr;
 	uint32_t doorbell_off;
 	bool is_interop;
+	bool is_evicted;
+	bool is_suspended;
+	bool is_being_destroyed;
 	bool is_active;
+	bool is_gws;
+	uint32_t pm4_target_xcc;
+	bool is_dbg_wa;
+	bool is_user_cu_masked;
 	/* Not relevant for user mode queues in cp scheduling */
 	unsigned int vmid;
 	/* Relevant only for sdma queues*/
@@ -331,6 +534,38 @@ struct queue_properties {
 	uint32_t eop_ring_buffer_size;
 	uint64_t ctx_save_restore_area_address;
 	uint32_t ctx_save_restore_area_size;
+	uint32_t ctl_stack_size;
+	uint64_t tba_addr;
+	uint64_t tma_addr;
+	uint64_t exception_status;
+
+	struct amdgpu_bo *wptr_bo;
+	struct amdgpu_bo *rptr_bo;
+	struct amdgpu_bo *ring_bo;
+	struct amdgpu_bo *eop_buf_bo;
+	struct amdgpu_bo *cwsr_bo;
+};
+
+#define QUEUE_IS_ACTIVE(q) ((q).queue_size > 0 &&	\
+			    (q).queue_address != 0 &&	\
+			    (q).queue_percent > 0 &&	\
+			    !(q).is_evicted &&		\
+			    !(q).is_suspended)
+
+enum mqd_update_flag {
+	UPDATE_FLAG_DBG_WA_ENABLE = 1,
+	UPDATE_FLAG_DBG_WA_DISABLE = 2,
+	UPDATE_FLAG_IS_GWS = 4, /* quirk for gfx9 IP */
+};
+
+struct mqd_update_info {
+	union {
+		struct {
+			uint32_t count; /* Must be a multiple of 32 */
+			uint32_t *ptr;
+		} cu_mask;
+	};
+	enum mqd_update_flag update_flag;
 };
 
 /**
@@ -338,7 +573,7 @@ struct queue_properties {
  *
  * @list: Queue linked list.
  *
- * @mqd: The queue MQD.
+ * @mqd: The queue MQD (memory queue descriptor).
  *
  * @mqd_mem_obj: The MQD local gpu memory object.
  *
@@ -347,7 +582,7 @@ struct queue_properties {
  * @properties: The queue properties.
  *
  * @mec: Used only in no cp scheduling mode and identifies to micro engine id
- *	 that the queue should be execute on.
+ *	 that the queue should be executed on.
  *
  * @pipe: Used only in no cp scheduling mode and identifies the queue's pipe
  *	  id.
@@ -358,6 +593,9 @@ struct queue_properties {
  *
  * @device: The kfd device that created this queue.
  *
+ * @gws: Pointing to gws kgd_mem if this is a gws control queue; NULL
+ * otherwise.
+ *
  * This structure represents user mode compute queues.
  * It contains all the necessary data to handle such queues.
  *
@@ -375,22 +613,36 @@ struct queue {
 	uint32_t queue;
 
 	unsigned int sdma_id;
+	unsigned int doorbell_id;
 
 	struct kfd_process	*process;
-	struct kfd_dev		*device;
+	struct kfd_node		*device;
+	void *gws;
+
+	/* procfs */
+	struct kobject kobj;
+
+	void *gang_ctx_bo;
+	uint64_t gang_ctx_gpu_addr;
+	void *gang_ctx_cpu_ptr;
+
+	struct amdgpu_bo *wptr_bo_gart;
 };
 
-/*
- * Please read the kfd_mqd_manager.h description.
- */
 enum KFD_MQD_TYPE {
-	KFD_MQD_TYPE_COMPUTE = 0,	/* for no cp scheduling */
-	KFD_MQD_TYPE_HIQ,		/* for hiq */
+	KFD_MQD_TYPE_HIQ = 0,		/* for hiq */
 	KFD_MQD_TYPE_CP,		/* for cp queues and diq */
 	KFD_MQD_TYPE_SDMA,		/* for sdma queues */
+	KFD_MQD_TYPE_DIQ,		/* for diq */
 	KFD_MQD_TYPE_MAX
 };
 
+enum KFD_PIPE_PRIORITY {
+	KFD_PIPE_PRIORITY_CS_LOW = 0,
+	KFD_PIPE_PRIORITY_CS_MEDIUM,
+	KFD_PIPE_PRIORITY_CS_HIGH
+};
+
 struct scheduling_resources {
 	unsigned int vmid_mask;
 	enum kfd_queue_type type;
@@ -404,7 +656,6 @@ struct scheduling_resources {
 struct process_queue_manager {
 	/* data */
 	struct kfd_process	*process;
-	unsigned int		num_concurrent_processes;
 	struct list_head	queues;
 	unsigned long		*queue_slot_bitmap;
 };
@@ -420,32 +671,92 @@ struct qcm_process_device {
 	unsigned int queue_count;
 	unsigned int vmid;
 	bool is_debug;
-	/*
-	 * All the memory management data should be here too
+	unsigned int evicted; /* eviction counter, 0=active */
+
+	/* This flag tells if we should reset all wavefronts on
+	 * process termination
+	 */
+	bool reset_wavefronts;
+
+	/* This flag tells us if this process has a GWS-capable
+	 * queue that will be mapped into the runlist. It's
+	 * possible to request a GWS BO, but not have the queue
+	 * currently mapped, and this changes how the MAP_PROCESS
+	 * PM4 packet is configured.
 	 */
+	bool mapped_gws_queue;
+
+	/* All the memory management data should be here too */
 	uint64_t gds_context_area;
+	/* Contains page table flags such as AMDGPU_PTE_VALID since gfx9 */
+	uint64_t page_table_base;
 	uint32_t sh_mem_config;
 	uint32_t sh_mem_bases;
 	uint32_t sh_mem_ape1_base;
 	uint32_t sh_mem_ape1_limit;
-	uint32_t page_table_base;
 	uint32_t gds_size;
 	uint32_t num_gws;
 	uint32_t num_oac;
 	uint32_t sh_hidden_private_base;
+
+	/* CWSR memory */
+	struct kgd_mem *cwsr_mem;
+	void *cwsr_kaddr;
+	uint64_t cwsr_base;
+	uint64_t tba_addr;
+	uint64_t tma_addr;
+
+	/* IB memory */
+	struct kgd_mem *ib_mem;
+	uint64_t ib_base;
+	void *ib_kaddr;
+
+	/* doorbells for kfd process */
+	struct amdgpu_bo *proc_doorbells;
+
+	/* bitmap for dynamic doorbell allocation from the bo */
+	unsigned long *doorbell_bitmap;
 };
 
+/* KFD Memory Eviction */
+
+/* Approx. wait time before attempting to restore evicted BOs */
+#define PROCESS_RESTORE_TIME_MS 100
+/* Approx. back off time if restore fails due to lack of memory */
+#define PROCESS_BACK_OFF_TIME_MS 100
+/* Approx. time before evicting the process again */
+#define PROCESS_ACTIVE_TIME_MS 10
+
+/* 8 byte handle containing GPU ID in the most significant 4 bytes and
+ * idr_handle in the least significant 4 bytes
+ */
+#define MAKE_HANDLE(gpu_id, idr_handle) \
+	(((uint64_t)(gpu_id) << 32) + idr_handle)
+#define GET_GPU_ID(handle) (handle >> 32)
+#define GET_IDR_HANDLE(handle) (handle & 0xFFFFFFFF)
+
+enum kfd_pdd_bound {
+	PDD_UNBOUND = 0,
+	PDD_BOUND,
+	PDD_BOUND_SUSPENDED,
+};
+
+#define MAX_SYSFS_FILENAME_LEN 15
+
+/*
+ * SDMA counter runs at 100MHz frequency.
+ * We display SDMA activity in microsecond granularity in sysfs.
+ * As a result, the divisor is 100.
+ */
+#define SDMA_ACTIVITY_DIVISOR  100
+
 /* Data that is per-process-per device. */
 struct kfd_process_device {
-	/*
-	 * List of all per-device data for a process.
-	 * Starts from kfd_process.per_device_data.
-	 */
-	struct list_head per_device_list;
-
 	/* The device that owns this data. */
-	struct kfd_dev *dev;
+	struct kfd_node *dev;
 
+	/* The process that owns this kfd_process_device. */
+	struct kfd_process *process;
 
 	/* per-process-per device QCM data structure */
 	struct qcm_process_device qpd;
@@ -458,17 +769,128 @@ struct kfd_process_device {
 	uint64_t scratch_base;
 	uint64_t scratch_limit;
 
+	/* VM context for GPUVM allocations */
+	struct file *drm_file;
+	void *drm_priv;
+
+	/* GPUVM allocations storage */
+	struct idr alloc_idr;
+
+	/* Flag used to tell the pdd has dequeued from the dqm.
+	 * This is used to prevent dev->dqm->ops.process_termination() from
+	 * being called twice when it is already called in IOMMU callback
+	 * function.
+	 */
+	bool already_dequeued;
+	bool runtime_inuse;
+
 	/* Is this process/pasid bound to this device? (amd_iommu_bind_pasid) */
-	bool bound;
+	enum kfd_pdd_bound bound;
+
+	/* VRAM usage */
+	atomic64_t vram_usage;
+	struct attribute attr_vram;
+	char vram_filename[MAX_SYSFS_FILENAME_LEN];
 
-	/* This flag tells if we should reset all
-	 * wavefronts on process termination
+	/* SDMA activity tracking */
+	uint64_t sdma_past_activity_counter;
+	struct attribute attr_sdma;
+	char sdma_filename[MAX_SYSFS_FILENAME_LEN];
+
+	/* Eviction activity tracking */
+	uint64_t last_evict_timestamp;
+	atomic64_t evict_duration_counter;
+	struct attribute attr_evict;
+
+	struct kobject *kobj_stats;
+
+	/*
+	 * @cu_occupancy: Reports occupancy of Compute Units (CU) of a process
+	 * that is associated with device encoded by "this" struct instance. The
+	 * value reflects CU usage by all of the waves launched by this process
+	 * on this device. A very important property of occupancy parameter is
+	 * that its value is a snapshot of current use.
+	 *
+	 * Following is to be noted regarding how this parameter is reported:
+	 *
+	 *  The number of waves that a CU can launch is limited by couple of
+	 *  parameters. These are encoded by struct amdgpu_cu_info instance
+	 *  that is part of every device definition. For GFX9 devices this
+	 *  translates to 40 waves (simd_per_cu * max_waves_per_simd) when waves
+	 *  do not use scratch memory and 32 waves (max_scratch_slots_per_cu)
+	 *  when they do use scratch memory. This could change for future
+	 *  devices and therefore this example should be considered as a guide.
+	 *
+	 *  All CU's of a device are available for the process. This may not be true
+	 *  under certain conditions - e.g. CU masking.
+	 *
+	 *  Finally number of CU's that are occupied by a process is affected by both
+	 *  number of CU's a device has along with number of other competing processes
 	 */
-	bool reset_wavefronts;
+	struct attribute attr_cu_occupancy;
+
+	/* sysfs counters for GPU retry fault and page migration tracking */
+	struct kobject *kobj_counters;
+	struct attribute attr_faults;
+	struct attribute attr_page_in;
+	struct attribute attr_page_out;
+	uint64_t faults;
+	uint64_t page_in;
+	uint64_t page_out;
+
+	/* Exception code status*/
+	uint64_t exception_status;
+	void *vm_fault_exc_data;
+	size_t vm_fault_exc_data_size;
+
+	/* Tracks debug per-vmid request settings */
+	uint32_t spi_dbg_override;
+	uint32_t spi_dbg_launch_mode;
+	uint32_t watch_points[4];
+	uint32_t alloc_watch_ids;
+
+	/*
+	 * If this process has been checkpointed before, then the user
+	 * application will use the original gpu_id on the
+	 * checkpointed node to refer to this device.
+	 */
+	uint32_t user_gpu_id;
+
+	void *proc_ctx_bo;
+	uint64_t proc_ctx_gpu_addr;
+	void *proc_ctx_cpu_ptr;
+
+	/* Tracks queue reset status */
+	bool has_reset_queue;
+
+	u32 pasid;
 };
 
 #define qpd_to_pdd(x) container_of(x, struct kfd_process_device, qpd)
 
+struct svm_range_list {
+	struct mutex			lock;
+	struct rb_root_cached		objects;
+	struct list_head		list;
+	struct work_struct		deferred_list_work;
+	struct list_head		deferred_range_list;
+	struct list_head                criu_svm_metadata_list;
+	spinlock_t			deferred_list_lock;
+	atomic_t			evicted_ranges;
+	atomic_t			drain_pagefaults;
+	struct delayed_work		restore_work;
+	DECLARE_BITMAP(bitmap_supported, MAX_GPU_INSTANCE);
+	struct task_struct		*faulting_task;
+	/* check point ts decides if page fault recovery need be dropped */
+	uint64_t			checkpoint_ts[MAX_GPU_INSTANCE];
+
+	/* Default granularity to use in buffer migration
+	 * and restoration of backing memory while handling
+	 * recoverable page faults
+	 */
+	uint8_t default_granularity;
+};
+
 /* Process data */
 struct kfd_process {
 	/*
@@ -477,7 +899,15 @@ struct kfd_process {
 	 */
 	struct hlist_node kfd_processes;
 
-	struct mm_struct *mm;
+	/*
+	 * Opaque pointer to mm_struct. We don't hold a reference to
+	 * it so it should never be dereferenced from here. This is
+	 * only used for looking up processes by their mm.
+	 */
+	void *mm;
+
+	struct kref ref;
+	struct work_struct release_work;
 
 	struct mutex mutex;
 
@@ -485,50 +915,123 @@ struct kfd_process {
 	 * In any process, the thread that started main() is the lead
 	 * thread and outlives the rest.
 	 * It is here because amd_iommu_bind_pasid wants a task_struct.
+	 * It can also be used for safely getting a reference to the
+	 * mm_struct of the process.
 	 */
 	struct task_struct *lead_thread;
 
 	/* We want to receive a notification when the mm_struct is destroyed */
 	struct mmu_notifier mmu_notifier;
 
-	/* Use for delayed freeing of kfd_process structure */
-	struct rcu_head	rcu;
-
-	unsigned int pasid;
-
 	/*
-	 * List of kfd_process_device structures,
+	 * Array of kfd_process_device pointers,
 	 * one for each device the process is using.
 	 */
-	struct list_head per_device_data;
+	struct kfd_process_device *pdds[MAX_GPU_INSTANCE];
+	uint32_t n_pdds;
 
 	struct process_queue_manager pqm;
 
-	/* The process's queues. */
-	size_t queue_array_size;
-
-	/* Size is queue_array_size, up to MAX_PROCESS_QUEUES. */
-	struct kfd_queue **queues;
-
 	/*Is the user space process 32 bit?*/
 	bool is_32bit_user_mode;
 
 	/* Event-related data */
 	struct mutex event_mutex;
-	/* All events in process hashed by ID, linked on kfd_event.events. */
-	DECLARE_HASHTABLE(events, 4);
-	/* struct slot_page_header.event_pages */
-	struct list_head signal_event_pages;
-	u32 next_nonsignal_event_id;
+	/* Event ID allocator and lookup */
+	struct idr event_idr;
+	/* Event page */
+	u64 signal_handle;
+	struct kfd_signal_page *signal_page;
+	size_t signal_mapped_size;
 	size_t signal_event_count;
+	bool signal_event_limit_reached;
+
+	/* Information used for memory eviction */
+	void *kgd_process_info;
+	/* Eviction fence that is attached to all the BOs of this process. The
+	 * fence will be triggered during eviction and new one will be created
+	 * during restore
+	 */
+	struct dma_fence __rcu *ef;
+
+	/* Work items for evicting and restoring BOs */
+	struct delayed_work eviction_work;
+	struct delayed_work restore_work;
+	/* seqno of the last scheduled eviction */
+	unsigned int last_eviction_seqno;
+	/* Approx. the last timestamp (in jiffies) when the process was
+	 * restored after an eviction
+	 */
+	unsigned long last_restore_timestamp;
+
+	/* Indicates device process is debug attached with reserved vmid. */
+	bool debug_trap_enabled;
+
+	/* per-process-per device debug event fd file */
+	struct file *dbg_ev_file;
+
+	/* If the process is a kfd debugger, we need to know so we can clean
+	 * up at exit time.  If a process enables debugging on itself, it does
+	 * its own clean-up, so we don't set the flag here.  We track this by
+	 * counting the number of processes this process is debugging.
+	 */
+	atomic_t debugged_process_count;
+
+	/* If the process is a debugged, this is the debugger process */
+	struct kfd_process *debugger_process;
+
+	/* Kobj for our procfs */
+	struct kobject *kobj;
+	struct kobject *kobj_queues;
+	struct attribute attr_pasid;
+
+	/* Keep track cwsr init */
+	bool has_cwsr;
+
+	/* Exception code enable mask and status */
+	uint64_t exception_enable_mask;
+	uint64_t exception_status;
+
+	/* Used to drain stale interrupts */
+	wait_queue_head_t wait_irq_drain;
+	bool irq_drain_is_open;
+
+	/* shared virtual memory registered by this process */
+	struct svm_range_list svms;
+
+	bool xnack_enabled;
+
+	/* Work area for debugger event writer worker. */
+	struct work_struct debug_event_workarea;
+
+	/* Tracks debug per-vmid request for debug flags */
+	u32 dbg_flags;
+
+	atomic_t poison;
+	/* Queues are in paused stated because we are in the process of doing a CRIU checkpoint */
+	bool queues_paused;
+
+	/* Tracks runtime enable status */
+	struct semaphore runtime_enable_sema;
+	bool is_runtime_retry;
+	struct kfd_runtime_info runtime_info;
+
+	/* if gpu page fault sent to KFD */
+	bool gpu_page_fault;
 };
 
+#define KFD_PROCESS_TABLE_SIZE 5 /* bits: 32 entries */
+extern DECLARE_HASHTABLE(kfd_processes_table, KFD_PROCESS_TABLE_SIZE);
+extern struct srcu_struct kfd_processes_srcu;
+
 /**
- * Ioctl function type.
+ * typedef amdkfd_ioctl_t - typedef for ioctl function pointer.
+ *
+ * @filep: pointer to file structure.
+ * @p: amdkfd process pointer.
+ * @data: pointer to arg that was copied from user.
  *
- * \param filep pointer to file structure.
- * \param p amdkfd process pointer.
- * \param data pointer to arg that was copied from user.
+ * Return: returns ioctl completion code.
  */
 typedef int amdkfd_ioctl_t(struct file *filep, struct kfd_process *p,
 				void *data);
@@ -540,101 +1043,311 @@ struct amdkfd_ioctl_desc {
 	unsigned int cmd_drv;
 	const char *name;
 };
+bool kfd_dev_is_large_bar(struct kfd_node *dev);
 
-void kfd_process_create_wq(void);
+int kfd_process_create_wq(void);
 void kfd_process_destroy_wq(void);
-struct kfd_process *kfd_create_process(const struct task_struct *);
-struct kfd_process *kfd_get_process(const struct task_struct *);
-struct kfd_process *kfd_lookup_process_by_pasid(unsigned int pasid);
-
-struct kfd_process_device *kfd_bind_process_to_device(struct kfd_dev *dev,
-							struct kfd_process *p);
-void kfd_unbind_process_from_device(struct kfd_dev *dev, unsigned int pasid);
-struct kfd_process_device *kfd_get_process_device_data(struct kfd_dev *dev,
+void kfd_cleanup_processes(void);
+struct kfd_process *kfd_create_process(struct task_struct *thread);
+struct kfd_process *kfd_get_process(const struct task_struct *task);
+struct kfd_process *kfd_lookup_process_by_pasid(u32 pasid,
+						 struct kfd_process_device **pdd);
+struct kfd_process *kfd_lookup_process_by_mm(const struct mm_struct *mm);
+
+int kfd_process_gpuidx_from_gpuid(struct kfd_process *p, uint32_t gpu_id);
+int kfd_process_gpuid_from_node(struct kfd_process *p, struct kfd_node *node,
+				uint32_t *gpuid, uint32_t *gpuidx);
+static inline int kfd_process_gpuid_from_gpuidx(struct kfd_process *p,
+				uint32_t gpuidx, uint32_t *gpuid) {
+	return gpuidx < p->n_pdds ? p->pdds[gpuidx]->dev->id : -EINVAL;
+}
+static inline struct kfd_process_device *kfd_process_device_from_gpuidx(
+				struct kfd_process *p, uint32_t gpuidx) {
+	return gpuidx < p->n_pdds ? p->pdds[gpuidx] : NULL;
+}
+
+void kfd_unref_process(struct kfd_process *p);
+int kfd_process_evict_queues(struct kfd_process *p, uint32_t trigger);
+int kfd_process_restore_queues(struct kfd_process *p);
+void kfd_suspend_all_processes(void);
+int kfd_resume_all_processes(void);
+
+struct kfd_process_device *kfd_process_device_data_by_id(struct kfd_process *process,
+							 uint32_t gpu_id);
+
+int kfd_process_get_user_gpu_id(struct kfd_process *p, uint32_t actual_gpu_id);
+
+int kfd_process_device_init_vm(struct kfd_process_device *pdd,
+			       struct file *drm_file);
+struct kfd_process_device *kfd_bind_process_to_device(struct kfd_node *dev,
+						struct kfd_process *p);
+struct kfd_process_device *kfd_get_process_device_data(struct kfd_node *dev,
 							struct kfd_process *p);
-struct kfd_process_device *kfd_create_process_device_data(struct kfd_dev *dev,
+struct kfd_process_device *kfd_create_process_device_data(struct kfd_node *dev,
 							struct kfd_process *p);
 
-/* Process device data iterator */
-struct kfd_process_device *kfd_get_first_process_device_data(
-							struct kfd_process *p);
-struct kfd_process_device *kfd_get_next_process_device_data(
-						struct kfd_process *p,
-						struct kfd_process_device *pdd);
-bool kfd_has_process_device_data(struct kfd_process *p);
+bool kfd_process_xnack_mode(struct kfd_process *p, bool supported);
+
+int kfd_reserved_mem_mmap(struct kfd_node *dev, struct kfd_process *process,
+			  struct vm_area_struct *vma);
+
+/* KFD process API for creating and translating handles */
+int kfd_process_device_create_obj_handle(struct kfd_process_device *pdd,
+					void *mem);
+void *kfd_process_device_translate_handle(struct kfd_process_device *p,
+					int handle);
+void kfd_process_device_remove_obj_handle(struct kfd_process_device *pdd,
+					int handle);
+struct kfd_process *kfd_lookup_process_by_pid(struct pid *pid);
 
 /* PASIDs */
 int kfd_pasid_init(void);
 void kfd_pasid_exit(void);
-bool kfd_set_pasid_limit(unsigned int new_limit);
-unsigned int kfd_get_pasid_limit(void);
-unsigned int kfd_pasid_alloc(void);
-void kfd_pasid_free(unsigned int pasid);
+u32 kfd_pasid_alloc(void);
+void kfd_pasid_free(u32 pasid);
 
 /* Doorbells */
+size_t kfd_doorbell_process_slice(struct kfd_dev *kfd);
 int kfd_doorbell_init(struct kfd_dev *kfd);
 void kfd_doorbell_fini(struct kfd_dev *kfd);
-int kfd_doorbell_mmap(struct kfd_process *process, struct vm_area_struct *vma);
-u32 __iomem *kfd_get_kernel_doorbell(struct kfd_dev *kfd,
+int kfd_doorbell_mmap(struct kfd_node *dev, struct kfd_process *process,
+		      struct vm_area_struct *vma);
+void __iomem *kfd_get_kernel_doorbell(struct kfd_dev *kfd,
 					unsigned int *doorbell_off);
 void kfd_release_kernel_doorbell(struct kfd_dev *kfd, u32 __iomem *db_addr);
 u32 read_kernel_doorbell(u32 __iomem *db);
-void write_kernel_doorbell(u32 __iomem *db, u32 value);
-unsigned int kfd_queue_id_to_doorbell(struct kfd_dev *kfd,
-					struct kfd_process *process,
-					unsigned int queue_id);
-
+void write_kernel_doorbell(void __iomem *db, u32 value);
+void write_kernel_doorbell64(void __iomem *db, u64 value);
+unsigned int kfd_get_doorbell_dw_offset_in_bar(struct kfd_dev *kfd,
+					struct kfd_process_device *pdd,
+					unsigned int doorbell_id);
+phys_addr_t kfd_get_process_doorbells(struct kfd_process_device *pdd);
+int kfd_alloc_process_doorbells(struct kfd_dev *kfd,
+				struct kfd_process_device *pdd);
+void kfd_free_process_doorbells(struct kfd_dev *kfd,
+				struct kfd_process_device *pdd);
 /* GTT Sub-Allocator */
 
-int kfd_gtt_sa_allocate(struct kfd_dev *kfd, unsigned int size,
+int kfd_gtt_sa_allocate(struct kfd_node *node, unsigned int size,
 			struct kfd_mem_obj **mem_obj);
 
-int kfd_gtt_sa_free(struct kfd_dev *kfd, struct kfd_mem_obj *mem_obj);
+int kfd_gtt_sa_free(struct kfd_node *node, struct kfd_mem_obj *mem_obj);
 
 extern struct device *kfd_device;
 
+/* KFD's procfs */
+void kfd_procfs_init(void);
+void kfd_procfs_shutdown(void);
+int kfd_procfs_add_queue(struct queue *q);
+void kfd_procfs_del_queue(struct queue *q);
+
 /* Topology */
 int kfd_topology_init(void);
 void kfd_topology_shutdown(void);
-int kfd_topology_add_device(struct kfd_dev *gpu);
-int kfd_topology_remove_device(struct kfd_dev *gpu);
-struct kfd_dev *kfd_device_by_id(uint32_t gpu_id);
-struct kfd_dev *kfd_device_by_pci_dev(const struct pci_dev *pdev);
-struct kfd_dev *kfd_topology_enum_kfd_devices(uint8_t idx);
+int kfd_topology_add_device(struct kfd_node *gpu);
+int kfd_topology_remove_device(struct kfd_node *gpu);
+struct kfd_topology_device *kfd_topology_device_by_proximity_domain(
+						uint32_t proximity_domain);
+struct kfd_topology_device *kfd_topology_device_by_proximity_domain_no_lock(
+						uint32_t proximity_domain);
+struct kfd_topology_device *kfd_topology_device_by_id(uint32_t gpu_id);
+struct kfd_node *kfd_device_by_id(uint32_t gpu_id);
+static inline bool kfd_irq_is_from_node(struct kfd_node *node, uint32_t node_id,
+					uint32_t vmid)
+{
+	return (node->interrupt_bitmap & (1 << node_id)) != 0 &&
+	       (node->compute_vmid_bitmap & (1 << vmid)) != 0;
+}
+static inline struct kfd_node *kfd_node_by_irq_ids(struct amdgpu_device *adev,
+					uint32_t node_id, uint32_t vmid) {
+	struct kfd_dev *dev = adev->kfd.dev;
+	uint32_t i;
+
+	if (KFD_GC_VERSION(dev) != IP_VERSION(9, 4, 3) &&
+	    KFD_GC_VERSION(dev) != IP_VERSION(9, 4, 4) &&
+	    KFD_GC_VERSION(dev) != IP_VERSION(9, 5, 0))
+		return dev->nodes[0];
+
+	for (i = 0; i < dev->num_nodes; i++)
+		if (kfd_irq_is_from_node(dev->nodes[i], node_id, vmid))
+			return dev->nodes[i];
+
+	return NULL;
+}
+int kfd_topology_enum_kfd_devices(uint8_t idx, struct kfd_node **kdev);
+int kfd_numa_node_to_apic_id(int numa_node_id);
 
 /* Interrupts */
-int kfd_interrupt_init(struct kfd_dev *dev);
-void kfd_interrupt_exit(struct kfd_dev *dev);
-void kgd2kfd_interrupt(struct kfd_dev *kfd, const void *ih_ring_entry);
-bool enqueue_ih_ring_entry(struct kfd_dev *kfd,	const void *ih_ring_entry);
-bool interrupt_is_wanted(struct kfd_dev *dev, const uint32_t *ih_ring_entry);
-
-/* Power Management */
-void kgd2kfd_suspend(struct kfd_dev *kfd);
-int kgd2kfd_resume(struct kfd_dev *kfd);
+#define	KFD_IRQ_FENCE_CLIENTID	0xff
+#define	KFD_IRQ_FENCE_SOURCEID	0xff
+#define	KFD_IRQ_IS_FENCE(client, source)				\
+				((client) == KFD_IRQ_FENCE_CLIENTID &&	\
+				(source) == KFD_IRQ_FENCE_SOURCEID)
+int kfd_interrupt_init(struct kfd_node *dev);
+void kfd_interrupt_exit(struct kfd_node *dev);
+bool enqueue_ih_ring_entry(struct kfd_node *kfd, const void *ih_ring_entry);
+bool interrupt_is_wanted(struct kfd_node *dev,
+				const uint32_t *ih_ring_entry,
+				uint32_t *patched_ihre, bool *flag);
+int kfd_process_drain_interrupts(struct kfd_process_device *pdd);
+void kfd_process_close_interrupt_drain(unsigned int pasid);
 
 /* amdkfd Apertures */
 int kfd_init_apertures(struct kfd_process *process);
 
-/* Queue Context Management */
-struct cik_sdma_rlc_registers *get_sdma_mqd(void *mqd);
+void kfd_process_set_trap_handler(struct qcm_process_device *qpd,
+				  uint64_t tba_addr,
+				  uint64_t tma_addr);
+void kfd_process_set_trap_debug_flag(struct qcm_process_device *qpd,
+				     bool enabled);
+
+/* CWSR initialization */
+int kfd_process_init_cwsr_apu(struct kfd_process *process, struct file *filep);
+
+/* CRIU */
+/*
+ * Need to increment KFD_CRIU_PRIV_VERSION each time a change is made to any of the CRIU private
+ * structures:
+ * kfd_criu_process_priv_data
+ * kfd_criu_device_priv_data
+ * kfd_criu_bo_priv_data
+ * kfd_criu_queue_priv_data
+ * kfd_criu_event_priv_data
+ * kfd_criu_svm_range_priv_data
+ */
+
+#define KFD_CRIU_PRIV_VERSION 1
+
+struct kfd_criu_process_priv_data {
+	uint32_t version;
+	uint32_t xnack_mode;
+};
+
+struct kfd_criu_device_priv_data {
+	/* For future use */
+	uint64_t reserved;
+};
+
+struct kfd_criu_bo_priv_data {
+	uint64_t user_addr;
+	uint32_t idr_handle;
+	uint32_t mapped_gpuids[MAX_GPU_INSTANCE];
+};
+
+/*
+ * The first 4 bytes of kfd_criu_queue_priv_data, kfd_criu_event_priv_data,
+ * kfd_criu_svm_range_priv_data is the object type
+ */
+enum kfd_criu_object_type {
+	KFD_CRIU_OBJECT_TYPE_QUEUE,
+	KFD_CRIU_OBJECT_TYPE_EVENT,
+	KFD_CRIU_OBJECT_TYPE_SVM_RANGE,
+};
+
+struct kfd_criu_svm_range_priv_data {
+	uint32_t object_type;
+	uint64_t start_addr;
+	uint64_t size;
+	/* Variable length array of attributes */
+	struct kfd_ioctl_svm_attribute attrs[];
+};
+
+struct kfd_criu_queue_priv_data {
+	uint32_t object_type;
+	uint64_t q_address;
+	uint64_t q_size;
+	uint64_t read_ptr_addr;
+	uint64_t write_ptr_addr;
+	uint64_t doorbell_off;
+	uint64_t eop_ring_buffer_address;
+	uint64_t ctx_save_restore_area_address;
+	uint32_t gpu_id;
+	uint32_t type;
+	uint32_t format;
+	uint32_t q_id;
+	uint32_t priority;
+	uint32_t q_percent;
+	uint32_t doorbell_id;
+	uint32_t gws;
+	uint32_t sdma_id;
+	uint32_t eop_ring_buffer_size;
+	uint32_t ctx_save_restore_area_size;
+	uint32_t ctl_stack_size;
+	uint32_t mqd_size;
+};
+
+struct kfd_criu_event_priv_data {
+	uint32_t object_type;
+	uint64_t user_handle;
+	uint32_t event_id;
+	uint32_t auto_reset;
+	uint32_t type;
+	uint32_t signaled;
+
+	union {
+		struct kfd_hsa_memory_exception_data memory_exception_data;
+		struct kfd_hsa_hw_exception_data hw_exception_data;
+	};
+};
+
+int kfd_process_get_queue_info(struct kfd_process *p,
+			       uint32_t *num_queues,
+			       uint64_t *priv_data_sizes);
+
+int kfd_criu_checkpoint_queues(struct kfd_process *p,
+			 uint8_t __user *user_priv_data,
+			 uint64_t *priv_data_offset);
+
+int kfd_criu_restore_queue(struct kfd_process *p,
+			   uint8_t __user *user_priv_data,
+			   uint64_t *priv_data_offset,
+			   uint64_t max_priv_data_size);
 
+int kfd_criu_checkpoint_events(struct kfd_process *p,
+			 uint8_t __user *user_priv_data,
+			 uint64_t *priv_data_offset);
+
+int kfd_criu_restore_event(struct file *devkfd,
+			   struct kfd_process *p,
+			   uint8_t __user *user_priv_data,
+			   uint64_t *priv_data_offset,
+			   uint64_t max_priv_data_size);
+/* CRIU - End */
+
+/* Queue Context Management */
 int init_queue(struct queue **q, const struct queue_properties *properties);
 void uninit_queue(struct queue *q);
 void print_queue_properties(struct queue_properties *q);
 void print_queue(struct queue *q);
+int kfd_queue_buffer_get(struct amdgpu_vm *vm, void __user *addr, struct amdgpu_bo **pbo,
+			 u64 expected_size);
+void kfd_queue_buffer_put(struct amdgpu_bo **bo);
+int kfd_queue_acquire_buffers(struct kfd_process_device *pdd, struct queue_properties *properties);
+int kfd_queue_release_buffers(struct kfd_process_device *pdd, struct queue_properties *properties);
+void kfd_queue_unref_bo_va(struct amdgpu_vm *vm, struct amdgpu_bo **bo);
+int kfd_queue_unref_bo_vas(struct kfd_process_device *pdd,
+			   struct queue_properties *properties);
+void kfd_queue_ctx_save_restore_size(struct kfd_topology_device *dev);
 
-struct mqd_manager *mqd_manager_init(enum KFD_MQD_TYPE type,
-					struct kfd_dev *dev);
 struct mqd_manager *mqd_manager_init_cik(enum KFD_MQD_TYPE type,
-		struct kfd_dev *dev);
+		struct kfd_node *dev);
 struct mqd_manager *mqd_manager_init_vi(enum KFD_MQD_TYPE type,
-		struct kfd_dev *dev);
-struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev);
+		struct kfd_node *dev);
+struct mqd_manager *mqd_manager_init_v9(enum KFD_MQD_TYPE type,
+		struct kfd_node *dev);
+struct mqd_manager *mqd_manager_init_v10(enum KFD_MQD_TYPE type,
+		struct kfd_node *dev);
+struct mqd_manager *mqd_manager_init_v11(enum KFD_MQD_TYPE type,
+		struct kfd_node *dev);
+struct mqd_manager *mqd_manager_init_v12(enum KFD_MQD_TYPE type,
+		struct kfd_node *dev);
+struct device_queue_manager *device_queue_manager_init(struct kfd_node *dev);
 void device_queue_manager_uninit(struct device_queue_manager *dqm);
-struct kernel_queue *kernel_queue_init(struct kfd_dev *dev,
+struct kernel_queue *kernel_queue_init(struct kfd_node *dev,
 					enum kfd_queue_type type);
 void kernel_queue_uninit(struct kernel_queue *kq);
+int kfd_evict_process_device(struct kfd_process_device *pdd);
+int kfd_dqm_suspend_bad_queue_mes(struct kfd_node *knode, u32 pasid, u32 doorbell_id);
 
 /* Process Queue Manager */
 struct process_queue_node {
@@ -643,32 +1356,68 @@ struct process_queue_node {
 	struct list_head process_queue_list;
 };
 
+void kfd_process_dequeue_from_device(struct kfd_process_device *pdd);
+void kfd_process_dequeue_from_all_devices(struct kfd_process *p);
 int pqm_init(struct process_queue_manager *pqm, struct kfd_process *p);
 void pqm_uninit(struct process_queue_manager *pqm);
 int pqm_create_queue(struct process_queue_manager *pqm,
-			    struct kfd_dev *dev,
-			    struct file *f,
+			    struct kfd_node *dev,
 			    struct queue_properties *properties,
-			    unsigned int flags,
-			    enum kfd_queue_type type,
-			    unsigned int *qid);
+			    unsigned int *qid,
+			    const struct kfd_criu_queue_priv_data *q_data,
+			    const void *restore_mqd,
+			    const void *restore_ctl_stack,
+			    uint32_t *p_doorbell_offset_in_process);
 int pqm_destroy_queue(struct process_queue_manager *pqm, unsigned int qid);
-int pqm_update_queue(struct process_queue_manager *pqm, unsigned int qid,
+int pqm_update_queue_properties(struct process_queue_manager *pqm, unsigned int qid,
 			struct queue_properties *p);
-struct kernel_queue *pqm_get_kernel_queue(struct process_queue_manager *pqm,
+int pqm_update_mqd(struct process_queue_manager *pqm, unsigned int qid,
+			struct mqd_update_info *minfo);
+int pqm_set_gws(struct process_queue_manager *pqm, unsigned int qid,
+			void *gws);
+struct queue *pqm_get_user_queue(struct process_queue_manager *pqm,
 						unsigned int qid);
-
-int amdkfd_fence_wait_timeout(unsigned int *fence_addr,
-				unsigned int fence_value,
-				unsigned long timeout);
-
+int pqm_get_wave_state(struct process_queue_manager *pqm,
+		       unsigned int qid,
+		       void __user *ctl_stack,
+		       u32 *ctl_stack_used_size,
+		       u32 *save_area_used_size);
+int pqm_get_queue_snapshot(struct process_queue_manager *pqm,
+			   uint64_t exception_clear_mask,
+			   void __user *buf,
+			   int *num_qss_entries,
+			   uint32_t *entry_size);
+
+int amdkfd_fence_wait_timeout(struct device_queue_manager *dqm,
+			      uint64_t fence_value,
+			      unsigned int timeout_ms);
+
+int pqm_get_queue_checkpoint_info(struct process_queue_manager *pqm,
+				  unsigned int qid,
+				  u32 *mqd_size,
+				  u32 *ctl_stack_size);
 /* Packet Manager */
 
-#define KFD_HIQ_TIMEOUT (500)
-
 #define KFD_FENCE_COMPLETED (100)
 #define KFD_FENCE_INIT   (10)
-#define KFD_UNMAP_LATENCY (150)
+
+/**
+ * enum kfd_config_dequeue_wait_counts_cmd - Command for configuring
+ *  dequeue wait counts.
+ *
+ * @KFD_DEQUEUE_WAIT_INIT: Set optimized dequeue wait counts for a
+ *	certain ASICs. For these ASICs, this is default value used by RESET
+ * @KFD_DEQUEUE_WAIT_RESET: Reset dequeue wait counts to the optimized value
+ *	for certain ASICs. For others set it to default hardware reset value
+ * @KFD_DEQUEUE_WAIT_SET_SCH_WAVE: Set context switch latency wait
+ *
+ */
+enum kfd_config_dequeue_wait_counts_cmd {
+	KFD_DEQUEUE_WAIT_INIT = 1,
+	KFD_DEQUEUE_WAIT_RESET = 2,
+	KFD_DEQUEUE_WAIT_SET_SCH_WAVE = 3
+};
+
 
 struct packet_manager {
 	struct device_queue_manager *dqm;
@@ -676,58 +1425,186 @@ struct packet_manager {
 	struct mutex lock;
 	bool allocated;
 	struct kfd_mem_obj *ib_buffer_obj;
+	unsigned int ib_size_bytes;
+	bool is_over_subscription;
+
+	const struct packet_manager_funcs *pmf;
+};
+
+struct packet_manager_funcs {
+	/* Support ASIC-specific packet formats for PM4 packets */
+	int (*map_process)(struct packet_manager *pm, uint32_t *buffer,
+			struct qcm_process_device *qpd);
+	int (*runlist)(struct packet_manager *pm, uint32_t *buffer,
+			uint64_t ib, size_t ib_size_in_dwords, bool chain);
+	int (*set_resources)(struct packet_manager *pm, uint32_t *buffer,
+			struct scheduling_resources *res);
+	int (*map_queues)(struct packet_manager *pm, uint32_t *buffer,
+			struct queue *q, bool is_static);
+	int (*unmap_queues)(struct packet_manager *pm, uint32_t *buffer,
+			enum kfd_unmap_queues_filter mode,
+			uint32_t filter_param, bool reset);
+	int (*config_dequeue_wait_counts)(struct packet_manager *pm, uint32_t *buffer,
+			enum kfd_config_dequeue_wait_counts_cmd cmd, uint32_t value);
+	int (*query_status)(struct packet_manager *pm, uint32_t *buffer,
+			uint64_t fence_address,	uint64_t fence_value);
+	int (*release_mem)(uint64_t gpu_addr, uint32_t *buffer);
+
+	/* Packet sizes */
+	int map_process_size;
+	int runlist_size;
+	int set_resources_size;
+	int map_queues_size;
+	int unmap_queues_size;
+	int config_dequeue_wait_counts_size;
+	int query_status_size;
+	int release_mem_size;
 };
 
+extern const struct packet_manager_funcs kfd_vi_pm_funcs;
+extern const struct packet_manager_funcs kfd_v9_pm_funcs;
+extern const struct packet_manager_funcs kfd_aldebaran_pm_funcs;
+
 int pm_init(struct packet_manager *pm, struct device_queue_manager *dqm);
 void pm_uninit(struct packet_manager *pm);
 int pm_send_set_resources(struct packet_manager *pm,
 				struct scheduling_resources *res);
 int pm_send_runlist(struct packet_manager *pm, struct list_head *dqm_queues);
 int pm_send_query_status(struct packet_manager *pm, uint64_t fence_address,
-				uint32_t fence_value);
+				uint64_t fence_value);
 
-int pm_send_unmap_queue(struct packet_manager *pm, enum kfd_queue_type type,
-			enum kfd_preempt_type_filter mode,
-			uint32_t filter_param, bool reset,
-			unsigned int sdma_engine);
+int pm_send_unmap_queue(struct packet_manager *pm,
+			enum kfd_unmap_queues_filter mode,
+			uint32_t filter_param, bool reset);
 
 void pm_release_ib(struct packet_manager *pm);
 
+int pm_config_dequeue_wait_counts(struct packet_manager *pm,
+			enum kfd_config_dequeue_wait_counts_cmd cmd,
+			uint32_t wait_counts_config);
+
+/* Following PM funcs can be shared among VI and AI */
+unsigned int pm_build_pm4_header(unsigned int opcode, size_t packet_size);
+
 uint64_t kfd_get_number_elems(struct kfd_dev *kfd);
-phys_addr_t kfd_get_process_doorbells(struct kfd_dev *dev,
-					struct kfd_process *process);
 
 /* Events */
 extern const struct kfd_event_interrupt_class event_interrupt_class_cik;
-extern const struct kfd_device_global_init_class device_global_init_class_cik;
+extern const struct kfd_event_interrupt_class event_interrupt_class_v9;
+extern const struct kfd_event_interrupt_class event_interrupt_class_v9_4_3;
+extern const struct kfd_event_interrupt_class event_interrupt_class_v10;
+extern const struct kfd_event_interrupt_class event_interrupt_class_v11;
 
-enum kfd_event_wait_result {
-	KFD_WAIT_COMPLETE,
-	KFD_WAIT_TIMEOUT,
-	KFD_WAIT_ERROR
-};
+extern const struct kfd_device_global_init_class device_global_init_class_cik;
 
-void kfd_event_init_process(struct kfd_process *p);
+int kfd_event_init_process(struct kfd_process *p);
 void kfd_event_free_process(struct kfd_process *p);
 int kfd_event_mmap(struct kfd_process *process, struct vm_area_struct *vma);
 int kfd_wait_on_events(struct kfd_process *p,
 		       uint32_t num_events, void __user *data,
-		       bool all, uint32_t user_timeout_ms,
-		       enum kfd_event_wait_result *wait_result);
-void kfd_signal_event_interrupt(unsigned int pasid, uint32_t partial_id,
+		       bool all, uint32_t *user_timeout_ms,
+		       uint32_t *wait_result);
+void kfd_signal_event_interrupt(u32 pasid, uint32_t partial_id,
 				uint32_t valid_id_bits);
-void kfd_signal_iommu_event(struct kfd_dev *dev,
-		unsigned int pasid, unsigned long address,
-		bool is_write_requested, bool is_execute_requested);
-void kfd_signal_hw_exception_event(unsigned int pasid);
+void kfd_signal_hw_exception_event(u32 pasid);
 int kfd_set_event(struct kfd_process *p, uint32_t event_id);
 int kfd_reset_event(struct kfd_process *p, uint32_t event_id);
+int kfd_kmap_event_page(struct kfd_process *p, uint64_t event_page_offset);
+
 int kfd_event_create(struct file *devkfd, struct kfd_process *p,
 		     uint32_t event_type, bool auto_reset, uint32_t node_id,
 		     uint32_t *event_id, uint32_t *event_trigger_data,
 		     uint64_t *event_page_offset, uint32_t *event_slot_index);
+
+int kfd_get_num_events(struct kfd_process *p);
 int kfd_event_destroy(struct kfd_process *p, uint32_t event_id);
 
-int dbgdev_wave_reset_wavefronts(struct kfd_dev *dev, struct kfd_process *p);
+void kfd_signal_vm_fault_event_with_userptr(struct kfd_process *p, uint64_t gpu_va);
+
+void kfd_signal_vm_fault_event(struct kfd_process_device *pdd,
+				struct kfd_vm_fault_info *info,
+				struct kfd_hsa_memory_exception_data *data);
+
+void kfd_signal_reset_event(struct kfd_node *dev);
+
+void kfd_signal_poison_consumed_event(struct kfd_node *dev, u32 pasid);
+
+static inline void kfd_flush_tlb(struct kfd_process_device *pdd,
+				 enum TLB_FLUSH_TYPE type)
+{
+	struct amdgpu_device *adev = pdd->dev->adev;
+	struct amdgpu_vm *vm = drm_priv_to_vm(pdd->drm_priv);
+
+	amdgpu_vm_flush_compute_tlb(adev, vm, type, pdd->dev->xcc_mask);
+}
+
+static inline bool kfd_flush_tlb_after_unmap(struct kfd_dev *dev)
+{
+	return KFD_GC_VERSION(dev) >= IP_VERSION(9, 4, 2) ||
+	       (KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 1) && dev->sdma_fw_version >= 18) ||
+	       KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 0);
+}
+
+int kfd_send_exception_to_runtime(struct kfd_process *p,
+				unsigned int queue_id,
+				uint64_t error_reason);
+bool kfd_is_locked(struct kfd_dev *kfd);
+
+/* Compute profile */
+void kfd_inc_compute_active(struct kfd_node *dev);
+void kfd_dec_compute_active(struct kfd_node *dev);
+
+/* Cgroup Support */
+/* Check with device cgroup if @kfd device is accessible */
+static inline int kfd_devcgroup_check_permission(struct kfd_node *node)
+{
+#if defined(CONFIG_CGROUP_DEVICE) || defined(CONFIG_CGROUP_BPF)
+	struct drm_device *ddev;
+
+	if (node->xcp)
+		ddev = node->xcp->ddev;
+	else
+		ddev = adev_to_drm(node->adev);
+
+	return devcgroup_check_permission(DEVCG_DEV_CHAR, DRM_MAJOR,
+					  ddev->render->index,
+					  DEVCG_ACC_WRITE | DEVCG_ACC_READ);
+#else
+	return 0;
+#endif
+}
+
+static inline bool kfd_is_first_node(struct kfd_node *node)
+{
+	return (node == node->kfd->nodes[0]);
+}
+
+/* Debugfs */
+#if defined(CONFIG_DEBUG_FS)
+
+void kfd_debugfs_init(void);
+void kfd_debugfs_fini(void);
+int kfd_debugfs_mqds_by_process(struct seq_file *m, void *data);
+int pqm_debugfs_mqds(struct seq_file *m, void *data);
+int kfd_debugfs_hqds_by_device(struct seq_file *m, void *data);
+int dqm_debugfs_hqds(struct seq_file *m, void *data);
+int kfd_debugfs_rls_by_device(struct seq_file *m, void *data);
+int pm_debugfs_runlist(struct seq_file *m, void *data);
+
+int kfd_debugfs_hang_hws(struct kfd_node *dev);
+int pm_debugfs_hang_hws(struct packet_manager *pm);
+int dqm_debugfs_hang_hws(struct device_queue_manager *dqm);
+
+void kfd_debugfs_add_process(struct kfd_process *p);
+void kfd_debugfs_remove_process(struct kfd_process *p);
+
+#else
+
+static inline void kfd_debugfs_init(void) {}
+static inline void kfd_debugfs_fini(void) {}
+static inline void kfd_debugfs_add_process(struct kfd_process *p) {}
+static inline void kfd_debugfs_remove_process(struct kfd_process *p) {}
+
+#endif
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
index c74cf22a1ed9..ddfe30c13e9d 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
@@ -1,5 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -24,47 +25,682 @@
 #include <linux/log2.h>
 #include <linux/sched.h>
 #include <linux/sched/mm.h>
+#include <linux/sched/task.h>
+#include <linux/mmu_context.h>
 #include <linux/slab.h>
-#include <linux/amd-iommu.h>
 #include <linux/notifier.h>
 #include <linux/compat.h>
+#include <linux/mman.h>
+#include <linux/file.h>
+#include <linux/pm_runtime.h>
+#include "amdgpu_amdkfd.h"
+#include "amdgpu.h"
+#include "amdgpu_reset.h"
 
 struct mm_struct;
 
 #include "kfd_priv.h"
-#include "kfd_dbgmgr.h"
-
-/*
- * Initial size for the array of queues.
- * The allocated size is doubled each time
- * it is exceeded up to MAX_PROCESS_QUEUES.
- */
-#define INITIAL_QUEUE_ARRAY_SIZE 16
+#include "kfd_device_queue_manager.h"
+#include "kfd_svm.h"
+#include "kfd_smi_events.h"
+#include "kfd_debug.h"
 
 /*
  * List of struct kfd_process (field kfd_process).
  * Unique/indexed by mm_struct*
  */
-#define KFD_PROCESS_TABLE_SIZE 5 /* bits: 32 entries */
-static DEFINE_HASHTABLE(kfd_processes_table, KFD_PROCESS_TABLE_SIZE);
-static DEFINE_MUTEX(kfd_processes_mutex);
+DEFINE_HASHTABLE(kfd_processes_table, KFD_PROCESS_TABLE_SIZE);
+DEFINE_MUTEX(kfd_processes_mutex);
 
-DEFINE_STATIC_SRCU(kfd_processes_srcu);
+DEFINE_SRCU(kfd_processes_srcu);
 
+/* For process termination handling */
 static struct workqueue_struct *kfd_process_wq;
 
-struct kfd_process_release_work {
-	struct work_struct kfd_work;
-	struct kfd_process *p;
-};
+/* Ordered, single-threaded workqueue for restoring evicted
+ * processes. Restoring multiple processes concurrently under memory
+ * pressure can lead to processes blocking each other from validating
+ * their BOs and result in a live-lock situation where processes
+ * remain evicted indefinitely.
+ */
+static struct workqueue_struct *kfd_restore_wq;
 
-static struct kfd_process *find_process(const struct task_struct *thread);
+static struct kfd_process *find_process(const struct task_struct *thread,
+					bool ref);
+static void kfd_process_ref_release(struct kref *ref);
 static struct kfd_process *create_process(const struct task_struct *thread);
 
-void kfd_process_create_wq(void)
+static void evict_process_worker(struct work_struct *work);
+static void restore_process_worker(struct work_struct *work);
+
+static void kfd_process_device_destroy_cwsr_dgpu(struct kfd_process_device *pdd);
+
+struct kfd_procfs_tree {
+	struct kobject *kobj;
+};
+
+static struct kfd_procfs_tree procfs;
+
+/*
+ * Structure for SDMA activity tracking
+ */
+struct kfd_sdma_activity_handler_workarea {
+	struct work_struct sdma_activity_work;
+	struct kfd_process_device *pdd;
+	uint64_t sdma_activity_counter;
+};
+
+struct temp_sdma_queue_list {
+	uint64_t __user *rptr;
+	uint64_t sdma_val;
+	unsigned int queue_id;
+	struct list_head list;
+};
+
+static void kfd_sdma_activity_worker(struct work_struct *work)
+{
+	struct kfd_sdma_activity_handler_workarea *workarea;
+	struct kfd_process_device *pdd;
+	uint64_t val;
+	struct mm_struct *mm;
+	struct queue *q;
+	struct qcm_process_device *qpd;
+	struct device_queue_manager *dqm;
+	int ret = 0;
+	struct temp_sdma_queue_list sdma_q_list;
+	struct temp_sdma_queue_list *sdma_q, *next;
+
+	workarea = container_of(work, struct kfd_sdma_activity_handler_workarea,
+				sdma_activity_work);
+
+	pdd = workarea->pdd;
+	if (!pdd)
+		return;
+	dqm = pdd->dev->dqm;
+	qpd = &pdd->qpd;
+	if (!dqm || !qpd)
+		return;
+	/*
+	 * Total SDMA activity is current SDMA activity + past SDMA activity
+	 * Past SDMA count is stored in pdd.
+	 * To get the current activity counters for all active SDMA queues,
+	 * we loop over all SDMA queues and get their counts from user-space.
+	 *
+	 * We cannot call get_user() with dqm_lock held as it can cause
+	 * a circular lock dependency situation. To read the SDMA stats,
+	 * we need to do the following:
+	 *
+	 * 1. Create a temporary list of SDMA queue nodes from the qpd->queues_list,
+	 *    with dqm_lock/dqm_unlock().
+	 * 2. Call get_user() for each node in temporary list without dqm_lock.
+	 *    Save the SDMA count for each node and also add the count to the total
+	 *    SDMA count counter.
+	 *    Its possible, during this step, a few SDMA queue nodes got deleted
+	 *    from the qpd->queues_list.
+	 * 3. Do a second pass over qpd->queues_list to check if any nodes got deleted.
+	 *    If any node got deleted, its SDMA count would be captured in the sdma
+	 *    past activity counter. So subtract the SDMA counter stored in step 2
+	 *    for this node from the total SDMA count.
+	 */
+	INIT_LIST_HEAD(&sdma_q_list.list);
+
+	/*
+	 * Create the temp list of all SDMA queues
+	 */
+	dqm_lock(dqm);
+
+	list_for_each_entry(q, &qpd->queues_list, list) {
+		if ((q->properties.type != KFD_QUEUE_TYPE_SDMA) &&
+		    (q->properties.type != KFD_QUEUE_TYPE_SDMA_XGMI))
+			continue;
+
+		sdma_q = kzalloc(sizeof(struct temp_sdma_queue_list), GFP_KERNEL);
+		if (!sdma_q) {
+			dqm_unlock(dqm);
+			goto cleanup;
+		}
+
+		INIT_LIST_HEAD(&sdma_q->list);
+		sdma_q->rptr = (uint64_t __user *)q->properties.read_ptr;
+		sdma_q->queue_id = q->properties.queue_id;
+		list_add_tail(&sdma_q->list, &sdma_q_list.list);
+	}
+
+	/*
+	 * If the temp list is empty, then no SDMA queues nodes were found in
+	 * qpd->queues_list. Return the past activity count as the total sdma
+	 * count
+	 */
+	if (list_empty(&sdma_q_list.list)) {
+		workarea->sdma_activity_counter = pdd->sdma_past_activity_counter;
+		dqm_unlock(dqm);
+		return;
+	}
+
+	dqm_unlock(dqm);
+
+	/*
+	 * Get the usage count for each SDMA queue in temp_list.
+	 */
+	mm = get_task_mm(pdd->process->lead_thread);
+	if (!mm)
+		goto cleanup;
+
+	kthread_use_mm(mm);
+
+	list_for_each_entry(sdma_q, &sdma_q_list.list, list) {
+		val = 0;
+		ret = read_sdma_queue_counter(sdma_q->rptr, &val);
+		if (ret) {
+			pr_debug("Failed to read SDMA queue active counter for queue id: %d",
+				 sdma_q->queue_id);
+		} else {
+			sdma_q->sdma_val = val;
+			workarea->sdma_activity_counter += val;
+		}
+	}
+
+	kthread_unuse_mm(mm);
+	mmput(mm);
+
+	/*
+	 * Do a second iteration over qpd_queues_list to check if any SDMA
+	 * nodes got deleted while fetching SDMA counter.
+	 */
+	dqm_lock(dqm);
+
+	workarea->sdma_activity_counter += pdd->sdma_past_activity_counter;
+
+	list_for_each_entry(q, &qpd->queues_list, list) {
+		if (list_empty(&sdma_q_list.list))
+			break;
+
+		if ((q->properties.type != KFD_QUEUE_TYPE_SDMA) &&
+		    (q->properties.type != KFD_QUEUE_TYPE_SDMA_XGMI))
+			continue;
+
+		list_for_each_entry_safe(sdma_q, next, &sdma_q_list.list, list) {
+			if (((uint64_t __user *)q->properties.read_ptr == sdma_q->rptr) &&
+			     (sdma_q->queue_id == q->properties.queue_id)) {
+				list_del(&sdma_q->list);
+				kfree(sdma_q);
+				break;
+			}
+		}
+	}
+
+	dqm_unlock(dqm);
+
+	/*
+	 * If temp list is not empty, it implies some queues got deleted
+	 * from qpd->queues_list during SDMA usage read. Subtract the SDMA
+	 * count for each node from the total SDMA count.
+	 */
+	list_for_each_entry_safe(sdma_q, next, &sdma_q_list.list, list) {
+		workarea->sdma_activity_counter -= sdma_q->sdma_val;
+		list_del(&sdma_q->list);
+		kfree(sdma_q);
+	}
+
+	return;
+
+cleanup:
+	list_for_each_entry_safe(sdma_q, next, &sdma_q_list.list, list) {
+		list_del(&sdma_q->list);
+		kfree(sdma_q);
+	}
+}
+
+/**
+ * kfd_get_cu_occupancy - Collect number of waves in-flight on this device
+ * by current process. Translates acquired wave count into number of compute units
+ * that are occupied.
+ *
+ * @attr: Handle of attribute that allows reporting of wave count. The attribute
+ * handle encapsulates GPU device it is associated with, thereby allowing collection
+ * of waves in flight, etc
+ * @buffer: Handle of user provided buffer updated with wave count
+ *
+ * Return: Number of bytes written to user buffer or an error value
+ */
+static int kfd_get_cu_occupancy(struct attribute *attr, char *buffer)
+{
+	int cu_cnt;
+	int wave_cnt;
+	int max_waves_per_cu;
+	struct kfd_node *dev = NULL;
+	struct kfd_process *proc = NULL;
+	struct kfd_process_device *pdd = NULL;
+	int i;
+	struct kfd_cu_occupancy *cu_occupancy;
+	u32 queue_format;
+
+	pdd = container_of(attr, struct kfd_process_device, attr_cu_occupancy);
+	dev = pdd->dev;
+	if (dev->kfd2kgd->get_cu_occupancy == NULL)
+		return -EINVAL;
+
+	cu_cnt = 0;
+	proc = pdd->process;
+	if (pdd->qpd.queue_count == 0) {
+		pr_debug("Gpu-Id: %d has no active queues for process pid %d\n",
+			 dev->id, (int)proc->lead_thread->pid);
+		return snprintf(buffer, PAGE_SIZE, "%d\n", cu_cnt);
+	}
+
+	/* Collect wave count from device if it supports */
+	wave_cnt = 0;
+	max_waves_per_cu = 0;
+
+	cu_occupancy = kcalloc(AMDGPU_MAX_QUEUES, sizeof(*cu_occupancy), GFP_KERNEL);
+	if (!cu_occupancy)
+		return -ENOMEM;
+
+	/*
+	 * For GFX 9.4.3, fetch the CU occupancy from the first XCC in the partition.
+	 * For AQL queues, because of cooperative dispatch we multiply the wave count
+	 * by number of XCCs in the partition to get the total wave counts across all
+	 * XCCs in the partition.
+	 * For PM4 queues, there is no cooperative dispatch so wave_cnt stay as it is.
+	 */
+	dev->kfd2kgd->get_cu_occupancy(dev->adev, cu_occupancy,
+			&max_waves_per_cu, ffs(dev->xcc_mask) - 1);
+
+	for (i = 0; i < AMDGPU_MAX_QUEUES; i++) {
+		if (cu_occupancy[i].wave_cnt != 0 &&
+		    kfd_dqm_is_queue_in_process(dev->dqm, &pdd->qpd,
+						cu_occupancy[i].doorbell_off,
+						&queue_format)) {
+			if (unlikely(queue_format == KFD_QUEUE_FORMAT_PM4))
+				wave_cnt += cu_occupancy[i].wave_cnt;
+			else
+				wave_cnt += (NUM_XCC(dev->xcc_mask) *
+						cu_occupancy[i].wave_cnt);
+		}
+	}
+
+	/* Translate wave count to number of compute units */
+	cu_cnt = (wave_cnt + (max_waves_per_cu - 1)) / max_waves_per_cu;
+	kfree(cu_occupancy);
+	return snprintf(buffer, PAGE_SIZE, "%d\n", cu_cnt);
+}
+
+static ssize_t kfd_procfs_show(struct kobject *kobj, struct attribute *attr,
+			       char *buffer)
+{
+	if (strcmp(attr->name, "pasid") == 0)
+		return snprintf(buffer, PAGE_SIZE, "%d\n", 0);
+	else if (strncmp(attr->name, "vram_", 5) == 0) {
+		struct kfd_process_device *pdd = container_of(attr, struct kfd_process_device,
+							      attr_vram);
+		return snprintf(buffer, PAGE_SIZE, "%llu\n", atomic64_read(&pdd->vram_usage));
+	} else if (strncmp(attr->name, "sdma_", 5) == 0) {
+		struct kfd_process_device *pdd = container_of(attr, struct kfd_process_device,
+							      attr_sdma);
+		struct kfd_sdma_activity_handler_workarea sdma_activity_work_handler;
+
+		INIT_WORK_ONSTACK(&sdma_activity_work_handler.sdma_activity_work,
+				  kfd_sdma_activity_worker);
+
+		sdma_activity_work_handler.pdd = pdd;
+		sdma_activity_work_handler.sdma_activity_counter = 0;
+
+		schedule_work(&sdma_activity_work_handler.sdma_activity_work);
+
+		flush_work(&sdma_activity_work_handler.sdma_activity_work);
+		destroy_work_on_stack(&sdma_activity_work_handler.sdma_activity_work);
+
+		return snprintf(buffer, PAGE_SIZE, "%llu\n",
+				(sdma_activity_work_handler.sdma_activity_counter)/
+				 SDMA_ACTIVITY_DIVISOR);
+	} else {
+		pr_err("Invalid attribute");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static void kfd_procfs_kobj_release(struct kobject *kobj)
+{
+	kfree(kobj);
+}
+
+static const struct sysfs_ops kfd_procfs_ops = {
+	.show = kfd_procfs_show,
+};
+
+static const struct kobj_type procfs_type = {
+	.release = kfd_procfs_kobj_release,
+	.sysfs_ops = &kfd_procfs_ops,
+};
+
+void kfd_procfs_init(void)
+{
+	int ret = 0;
+
+	procfs.kobj = kfd_alloc_struct(procfs.kobj);
+	if (!procfs.kobj)
+		return;
+
+	ret = kobject_init_and_add(procfs.kobj, &procfs_type,
+				   &kfd_device->kobj, "proc");
+	if (ret) {
+		pr_warn("Could not create procfs proc folder");
+		/* If we fail to create the procfs, clean up */
+		kfd_procfs_shutdown();
+	}
+}
+
+void kfd_procfs_shutdown(void)
+{
+	if (procfs.kobj) {
+		kobject_del(procfs.kobj);
+		kobject_put(procfs.kobj);
+		procfs.kobj = NULL;
+	}
+}
+
+static ssize_t kfd_procfs_queue_show(struct kobject *kobj,
+				     struct attribute *attr, char *buffer)
+{
+	struct queue *q = container_of(kobj, struct queue, kobj);
+
+	if (!strcmp(attr->name, "size"))
+		return snprintf(buffer, PAGE_SIZE, "%llu",
+				q->properties.queue_size);
+	else if (!strcmp(attr->name, "type"))
+		return snprintf(buffer, PAGE_SIZE, "%d", q->properties.type);
+	else if (!strcmp(attr->name, "gpuid"))
+		return snprintf(buffer, PAGE_SIZE, "%u", q->device->id);
+	else
+		pr_err("Invalid attribute");
+
+	return 0;
+}
+
+static ssize_t kfd_procfs_stats_show(struct kobject *kobj,
+				     struct attribute *attr, char *buffer)
+{
+	if (strcmp(attr->name, "evicted_ms") == 0) {
+		struct kfd_process_device *pdd = container_of(attr,
+				struct kfd_process_device,
+				attr_evict);
+		uint64_t evict_jiffies;
+
+		evict_jiffies = atomic64_read(&pdd->evict_duration_counter);
+
+		return snprintf(buffer,
+				PAGE_SIZE,
+				"%llu\n",
+				jiffies64_to_msecs(evict_jiffies));
+
+	/* Sysfs handle that gets CU occupancy is per device */
+	} else if (strcmp(attr->name, "cu_occupancy") == 0) {
+		return kfd_get_cu_occupancy(attr, buffer);
+	} else {
+		pr_err("Invalid attribute");
+	}
+
+	return 0;
+}
+
+static ssize_t kfd_sysfs_counters_show(struct kobject *kobj,
+				       struct attribute *attr, char *buf)
+{
+	struct kfd_process_device *pdd;
+
+	if (!strcmp(attr->name, "faults")) {
+		pdd = container_of(attr, struct kfd_process_device,
+				   attr_faults);
+		return sysfs_emit(buf, "%llu\n", READ_ONCE(pdd->faults));
+	}
+	if (!strcmp(attr->name, "page_in")) {
+		pdd = container_of(attr, struct kfd_process_device,
+				   attr_page_in);
+		return sysfs_emit(buf, "%llu\n", READ_ONCE(pdd->page_in));
+	}
+	if (!strcmp(attr->name, "page_out")) {
+		pdd = container_of(attr, struct kfd_process_device,
+				   attr_page_out);
+		return sysfs_emit(buf, "%llu\n", READ_ONCE(pdd->page_out));
+	}
+	return 0;
+}
+
+static struct attribute attr_queue_size = {
+	.name = "size",
+	.mode = KFD_SYSFS_FILE_MODE
+};
+
+static struct attribute attr_queue_type = {
+	.name = "type",
+	.mode = KFD_SYSFS_FILE_MODE
+};
+
+static struct attribute attr_queue_gpuid = {
+	.name = "gpuid",
+	.mode = KFD_SYSFS_FILE_MODE
+};
+
+static struct attribute *procfs_queue_attrs[] = {
+	&attr_queue_size,
+	&attr_queue_type,
+	&attr_queue_gpuid,
+	NULL
+};
+ATTRIBUTE_GROUPS(procfs_queue);
+
+static const struct sysfs_ops procfs_queue_ops = {
+	.show = kfd_procfs_queue_show,
+};
+
+static const struct kobj_type procfs_queue_type = {
+	.sysfs_ops = &procfs_queue_ops,
+	.default_groups = procfs_queue_groups,
+};
+
+static const struct sysfs_ops procfs_stats_ops = {
+	.show = kfd_procfs_stats_show,
+};
+
+static const struct kobj_type procfs_stats_type = {
+	.sysfs_ops = &procfs_stats_ops,
+	.release = kfd_procfs_kobj_release,
+};
+
+static const struct sysfs_ops sysfs_counters_ops = {
+	.show = kfd_sysfs_counters_show,
+};
+
+static const struct kobj_type sysfs_counters_type = {
+	.sysfs_ops = &sysfs_counters_ops,
+	.release = kfd_procfs_kobj_release,
+};
+
+int kfd_procfs_add_queue(struct queue *q)
+{
+	struct kfd_process *proc;
+	int ret;
+
+	if (!q || !q->process)
+		return -EINVAL;
+	proc = q->process;
+
+	/* Create proc/<pid>/queues/<queue id> folder */
+	if (!proc->kobj_queues)
+		return -EFAULT;
+	ret = kobject_init_and_add(&q->kobj, &procfs_queue_type,
+			proc->kobj_queues, "%u", q->properties.queue_id);
+	if (ret < 0) {
+		pr_warn("Creating proc/<pid>/queues/%u failed",
+			q->properties.queue_id);
+		kobject_put(&q->kobj);
+		return ret;
+	}
+
+	return 0;
+}
+
+static void kfd_sysfs_create_file(struct kobject *kobj, struct attribute *attr,
+				 char *name)
+{
+	int ret;
+
+	if (!kobj || !attr || !name)
+		return;
+
+	attr->name = name;
+	attr->mode = KFD_SYSFS_FILE_MODE;
+	sysfs_attr_init(attr);
+
+	ret = sysfs_create_file(kobj, attr);
+	if (ret)
+		pr_warn("Create sysfs %s/%s failed %d", kobj->name, name, ret);
+}
+
+static void kfd_procfs_add_sysfs_stats(struct kfd_process *p)
+{
+	int ret;
+	int i;
+	char stats_dir_filename[MAX_SYSFS_FILENAME_LEN];
+
+	if (!p || !p->kobj)
+		return;
+
+	/*
+	 * Create sysfs files for each GPU:
+	 * - proc/<pid>/stats_<gpuid>/
+	 * - proc/<pid>/stats_<gpuid>/evicted_ms
+	 * - proc/<pid>/stats_<gpuid>/cu_occupancy
+	 */
+	for (i = 0; i < p->n_pdds; i++) {
+		struct kfd_process_device *pdd = p->pdds[i];
+
+		snprintf(stats_dir_filename, MAX_SYSFS_FILENAME_LEN,
+				"stats_%u", pdd->dev->id);
+		pdd->kobj_stats = kfd_alloc_struct(pdd->kobj_stats);
+		if (!pdd->kobj_stats)
+			return;
+
+		ret = kobject_init_and_add(pdd->kobj_stats,
+					   &procfs_stats_type,
+					   p->kobj,
+					   stats_dir_filename);
+
+		if (ret) {
+			pr_warn("Creating KFD proc/stats_%s folder failed",
+				stats_dir_filename);
+			kobject_put(pdd->kobj_stats);
+			pdd->kobj_stats = NULL;
+			return;
+		}
+
+		kfd_sysfs_create_file(pdd->kobj_stats, &pdd->attr_evict,
+				      "evicted_ms");
+		/* Add sysfs file to report compute unit occupancy */
+		if (pdd->dev->kfd2kgd->get_cu_occupancy)
+			kfd_sysfs_create_file(pdd->kobj_stats,
+					      &pdd->attr_cu_occupancy,
+					      "cu_occupancy");
+	}
+}
+
+static void kfd_procfs_add_sysfs_counters(struct kfd_process *p)
+{
+	int ret = 0;
+	int i;
+	char counters_dir_filename[MAX_SYSFS_FILENAME_LEN];
+
+	if (!p || !p->kobj)
+		return;
+
+	/*
+	 * Create sysfs files for each GPU which supports SVM
+	 * - proc/<pid>/counters_<gpuid>/
+	 * - proc/<pid>/counters_<gpuid>/faults
+	 * - proc/<pid>/counters_<gpuid>/page_in
+	 * - proc/<pid>/counters_<gpuid>/page_out
+	 */
+	for_each_set_bit(i, p->svms.bitmap_supported, p->n_pdds) {
+		struct kfd_process_device *pdd = p->pdds[i];
+		struct kobject *kobj_counters;
+
+		snprintf(counters_dir_filename, MAX_SYSFS_FILENAME_LEN,
+			"counters_%u", pdd->dev->id);
+		kobj_counters = kfd_alloc_struct(kobj_counters);
+		if (!kobj_counters)
+			return;
+
+		ret = kobject_init_and_add(kobj_counters, &sysfs_counters_type,
+					   p->kobj, counters_dir_filename);
+		if (ret) {
+			pr_warn("Creating KFD proc/%s folder failed",
+				counters_dir_filename);
+			kobject_put(kobj_counters);
+			return;
+		}
+
+		pdd->kobj_counters = kobj_counters;
+		kfd_sysfs_create_file(kobj_counters, &pdd->attr_faults,
+				      "faults");
+		kfd_sysfs_create_file(kobj_counters, &pdd->attr_page_in,
+				      "page_in");
+		kfd_sysfs_create_file(kobj_counters, &pdd->attr_page_out,
+				      "page_out");
+	}
+}
+
+static void kfd_procfs_add_sysfs_files(struct kfd_process *p)
+{
+	int i;
+
+	if (!p || !p->kobj)
+		return;
+
+	/*
+	 * Create sysfs files for each GPU:
+	 * - proc/<pid>/vram_<gpuid>
+	 * - proc/<pid>/sdma_<gpuid>
+	 */
+	for (i = 0; i < p->n_pdds; i++) {
+		struct kfd_process_device *pdd = p->pdds[i];
+
+		snprintf(pdd->vram_filename, MAX_SYSFS_FILENAME_LEN, "vram_%u",
+			 pdd->dev->id);
+		kfd_sysfs_create_file(p->kobj, &pdd->attr_vram,
+				      pdd->vram_filename);
+
+		snprintf(pdd->sdma_filename, MAX_SYSFS_FILENAME_LEN, "sdma_%u",
+			 pdd->dev->id);
+		kfd_sysfs_create_file(p->kobj, &pdd->attr_sdma,
+					    pdd->sdma_filename);
+	}
+}
+
+void kfd_procfs_del_queue(struct queue *q)
+{
+	if (!q)
+		return;
+
+	kobject_del(&q->kobj);
+	kobject_put(&q->kobj);
+}
+
+int kfd_process_create_wq(void)
 {
 	if (!kfd_process_wq)
 		kfd_process_wq = alloc_workqueue("kfd_process_wq", 0, 0);
+	if (!kfd_restore_wq)
+		kfd_restore_wq = alloc_ordered_workqueue("kfd_restore_wq",
+							 WQ_FREEZABLE);
+
+	if (!kfd_process_wq || !kfd_restore_wq) {
+		kfd_process_destroy_wq();
+		return -ENOMEM;
+	}
+
+	return 0;
 }
 
 void kfd_process_destroy_wq(void)
@@ -73,21 +709,143 @@ void kfd_process_destroy_wq(void)
 		destroy_workqueue(kfd_process_wq);
 		kfd_process_wq = NULL;
 	}
+	if (kfd_restore_wq) {
+		destroy_workqueue(kfd_restore_wq);
+		kfd_restore_wq = NULL;
+	}
 }
 
-struct kfd_process *kfd_create_process(const struct task_struct *thread)
+static void kfd_process_free_gpuvm(struct kgd_mem *mem,
+			struct kfd_process_device *pdd, void **kptr)
+{
+	struct kfd_node *dev = pdd->dev;
+
+	if (kptr && *kptr) {
+		amdgpu_amdkfd_gpuvm_unmap_gtt_bo_from_kernel(mem);
+		*kptr = NULL;
+	}
+
+	amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu(dev->adev, mem, pdd->drm_priv);
+	amdgpu_amdkfd_gpuvm_free_memory_of_gpu(dev->adev, mem, pdd->drm_priv,
+					       NULL);
+}
+
+/* kfd_process_alloc_gpuvm - Allocate GPU VM for the KFD process
+ *	This function should be only called right after the process
+ *	is created and when kfd_processes_mutex is still being held
+ *	to avoid concurrency. Because of that exclusiveness, we do
+ *	not need to take p->mutex.
+ */
+static int kfd_process_alloc_gpuvm(struct kfd_process_device *pdd,
+				   uint64_t gpu_va, uint32_t size,
+				   uint32_t flags, struct kgd_mem **mem, void **kptr)
+{
+	struct kfd_node *kdev = pdd->dev;
+	int err;
+
+	err = amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(kdev->adev, gpu_va, size,
+						 pdd->drm_priv, mem, NULL,
+						 flags, false);
+	if (err)
+		goto err_alloc_mem;
+
+	err = amdgpu_amdkfd_gpuvm_map_memory_to_gpu(kdev->adev, *mem,
+			pdd->drm_priv);
+	if (err)
+		goto err_map_mem;
+
+	err = amdgpu_amdkfd_gpuvm_sync_memory(kdev->adev, *mem, true);
+	if (err) {
+		pr_debug("Sync memory failed, wait interrupted by user signal\n");
+		goto sync_memory_failed;
+	}
+
+	if (kptr) {
+		err = amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel(
+				(struct kgd_mem *)*mem, kptr, NULL);
+		if (err) {
+			pr_debug("Map GTT BO to kernel failed\n");
+			goto sync_memory_failed;
+		}
+	}
+
+	return err;
+
+sync_memory_failed:
+	amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu(kdev->adev, *mem, pdd->drm_priv);
+
+err_map_mem:
+	amdgpu_amdkfd_gpuvm_free_memory_of_gpu(kdev->adev, *mem, pdd->drm_priv,
+					       NULL);
+err_alloc_mem:
+	*mem = NULL;
+	*kptr = NULL;
+	return err;
+}
+
+/* kfd_process_device_reserve_ib_mem - Reserve memory inside the
+ *	process for IB usage The memory reserved is for KFD to submit
+ *	IB to AMDGPU from kernel.  If the memory is reserved
+ *	successfully, ib_kaddr will have the CPU/kernel
+ *	address. Check ib_kaddr before accessing the memory.
+ */
+static int kfd_process_device_reserve_ib_mem(struct kfd_process_device *pdd)
+{
+	struct qcm_process_device *qpd = &pdd->qpd;
+	uint32_t flags = KFD_IOC_ALLOC_MEM_FLAGS_GTT |
+			KFD_IOC_ALLOC_MEM_FLAGS_NO_SUBSTITUTE |
+			KFD_IOC_ALLOC_MEM_FLAGS_WRITABLE |
+			KFD_IOC_ALLOC_MEM_FLAGS_EXECUTABLE;
+	struct kgd_mem *mem;
+	void *kaddr;
+	int ret;
+
+	if (qpd->ib_kaddr || !qpd->ib_base)
+		return 0;
+
+	/* ib_base is only set for dGPU */
+	ret = kfd_process_alloc_gpuvm(pdd, qpd->ib_base, PAGE_SIZE, flags,
+				      &mem, &kaddr);
+	if (ret)
+		return ret;
+
+	qpd->ib_mem = mem;
+	qpd->ib_kaddr = kaddr;
+
+	return 0;
+}
+
+static void kfd_process_device_destroy_ib_mem(struct kfd_process_device *pdd)
+{
+	struct qcm_process_device *qpd = &pdd->qpd;
+
+	if (!qpd->ib_kaddr || !qpd->ib_base)
+		return;
+
+	kfd_process_free_gpuvm(qpd->ib_mem, pdd, &qpd->ib_kaddr);
+}
+
+struct kfd_process *kfd_create_process(struct task_struct *thread)
 {
 	struct kfd_process *process;
+	int ret;
 
-	if (!thread->mm)
+	if (!(thread->mm && mmget_not_zero(thread->mm)))
 		return ERR_PTR(-EINVAL);
 
 	/* Only the pthreads threading model is supported. */
-	if (thread->group_leader->mm != thread->mm)
+	if (thread->group_leader->mm != thread->mm) {
+		mmput(thread->mm);
 		return ERR_PTR(-EINVAL);
+	}
 
-	/* Take mmap_sem because we call __mmu_notifier_register inside */
-	down_write(&thread->mm->mmap_sem);
+	/* If the process just called exec(3), it is possible that the
+	 * cleanup of the kfd_process (following the release of the mm
+	 * of the old process image) is still in the cleanup work queue.
+	 * Make sure to drain any job before trying to recreate any
+	 * resource for this process.
+	 */
+	flush_workqueue(kfd_process_wq);
 
 	/*
 	 * take kfd processes mutex before starting of process creation
@@ -96,17 +854,59 @@ struct kfd_process *kfd_create_process(const struct task_struct *thread)
 	 */
 	mutex_lock(&kfd_processes_mutex);
 
-	/* A prior open of /dev/kfd could have already created the process. */
-	process = find_process(thread);
-	if (process)
-		pr_debug("Process already found\n");
+	if (kfd_is_locked(NULL)) {
+		pr_debug("KFD is locked! Cannot create process");
+		process = ERR_PTR(-EINVAL);
+		goto out;
+	}
 
-	if (!process)
+	/* A prior open of /dev/kfd could have already created the process.
+	 * find_process will increase process kref in this case
+	 */
+	process = find_process(thread, true);
+	if (process) {
+		pr_debug("Process already found\n");
+	} else {
 		process = create_process(thread);
+		if (IS_ERR(process))
+			goto out;
 
-	mutex_unlock(&kfd_processes_mutex);
+		if (!procfs.kobj)
+			goto out;
+
+		process->kobj = kfd_alloc_struct(process->kobj);
+		if (!process->kobj) {
+			pr_warn("Creating procfs kobject failed");
+			goto out;
+		}
+		ret = kobject_init_and_add(process->kobj, &procfs_type,
+					   procfs.kobj, "%d",
+					   (int)process->lead_thread->pid);
+		if (ret) {
+			pr_warn("Creating procfs pid directory failed");
+			kobject_put(process->kobj);
+			goto out;
+		}
 
-	up_write(&thread->mm->mmap_sem);
+		kfd_sysfs_create_file(process->kobj, &process->attr_pasid,
+				      "pasid");
+
+		process->kobj_queues = kobject_create_and_add("queues",
+							process->kobj);
+		if (!process->kobj_queues)
+			pr_warn("Creating KFD proc/queues folder failed");
+
+		kfd_procfs_add_sysfs_stats(process);
+		kfd_procfs_add_sysfs_files(process);
+		kfd_procfs_add_sysfs_counters(process);
+
+		kfd_debugfs_add_process(process);
+
+		init_waitqueue_head(&process->wait_irq_drain);
+	}
+out:
+	mutex_unlock(&kfd_processes_mutex);
+	mmput(thread->mm);
 
 	return process;
 }
@@ -122,7 +922,9 @@ struct kfd_process *kfd_get_process(const struct task_struct *thread)
 	if (thread->group_leader->mm != thread->mm)
 		return ERR_PTR(-EINVAL);
 
-	process = find_process(thread);
+	process = find_process(thread, false);
+	if (!process)
+		return ERR_PTR(-EINVAL);
 
 	return process;
 }
@@ -139,86 +941,328 @@ static struct kfd_process *find_process_by_mm(const struct mm_struct *mm)
 	return NULL;
 }
 
-static struct kfd_process *find_process(const struct task_struct *thread)
+static struct kfd_process *find_process(const struct task_struct *thread,
+					bool ref)
 {
 	struct kfd_process *p;
 	int idx;
 
 	idx = srcu_read_lock(&kfd_processes_srcu);
 	p = find_process_by_mm(thread->mm);
+	if (p && ref)
+		kref_get(&p->ref);
 	srcu_read_unlock(&kfd_processes_srcu, idx);
 
 	return p;
 }
 
-static void kfd_process_wq_release(struct work_struct *work)
+void kfd_unref_process(struct kfd_process *p)
 {
-	struct kfd_process_release_work *my_work;
-	struct kfd_process_device *pdd, *temp;
-	struct kfd_process *p;
+	kref_put(&p->ref, kfd_process_ref_release);
+}
 
-	my_work = (struct kfd_process_release_work *) work;
+/* This increments the process->ref counter. */
+struct kfd_process *kfd_lookup_process_by_pid(struct pid *pid)
+{
+	struct task_struct *task = NULL;
+	struct kfd_process *p    = NULL;
+
+	if (!pid) {
+		task = current;
+		get_task_struct(task);
+	} else {
+		task = get_pid_task(pid, PIDTYPE_PID);
+	}
 
-	p = my_work->p;
+	if (task) {
+		p = find_process(task, true);
+		put_task_struct(task);
+	}
 
-	pr_debug("Releasing process (pasid %d) in workqueue\n",
-			p->pasid);
+	return p;
+}
+
+static void kfd_process_device_free_bos(struct kfd_process_device *pdd)
+{
+	struct kfd_process *p = pdd->process;
+	void *mem;
+	int id;
+	int i;
+
+	/*
+	 * Remove all handles from idr and release appropriate
+	 * local memory object
+	 */
+	idr_for_each_entry(&pdd->alloc_idr, mem, id) {
+
+		for (i = 0; i < p->n_pdds; i++) {
+			struct kfd_process_device *peer_pdd = p->pdds[i];
+
+			if (!peer_pdd->drm_priv)
+				continue;
+			amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu(
+				peer_pdd->dev->adev, mem, peer_pdd->drm_priv);
+		}
+
+		amdgpu_amdkfd_gpuvm_free_memory_of_gpu(pdd->dev->adev, mem,
+						       pdd->drm_priv, NULL);
+		kfd_process_device_remove_obj_handle(pdd, id);
+	}
+}
+
+/*
+ * Just kunmap and unpin signal BO here. It will be freed in
+ * kfd_process_free_outstanding_kfd_bos()
+ */
+static void kfd_process_kunmap_signal_bo(struct kfd_process *p)
+{
+	struct kfd_process_device *pdd;
+	struct kfd_node *kdev;
+	void *mem;
+
+	kdev = kfd_device_by_id(GET_GPU_ID(p->signal_handle));
+	if (!kdev)
+		return;
 
 	mutex_lock(&p->mutex);
 
-	list_for_each_entry_safe(pdd, temp, &p->per_device_data,
-							per_device_list) {
-		pr_debug("Releasing pdd (topology id %d) for process (pasid %d) in workqueue\n",
-				pdd->dev->id, p->pasid);
+	pdd = kfd_get_process_device_data(kdev, p);
+	if (!pdd)
+		goto out;
+
+	mem = kfd_process_device_translate_handle(
+		pdd, GET_IDR_HANDLE(p->signal_handle));
+	if (!mem)
+		goto out;
+
+	amdgpu_amdkfd_gpuvm_unmap_gtt_bo_from_kernel(mem);
 
-		if (pdd->reset_wavefronts)
-			dbgdev_wave_reset_wavefronts(pdd->dev, p);
+out:
+	mutex_unlock(&p->mutex);
+}
+
+static void kfd_process_free_outstanding_kfd_bos(struct kfd_process *p)
+{
+	int i;
+
+	for (i = 0; i < p->n_pdds; i++)
+		kfd_process_device_free_bos(p->pdds[i]);
+}
+
+static void kfd_process_destroy_pdds(struct kfd_process *p)
+{
+	int i;
+
+	for (i = 0; i < p->n_pdds; i++) {
+		struct kfd_process_device *pdd = p->pdds[i];
+
+		kfd_smi_event_process(pdd, false);
+
+		pr_debug("Releasing pdd (topology id %d, for pid %d)\n",
+			pdd->dev->id, p->lead_thread->pid);
+		kfd_process_device_destroy_cwsr_dgpu(pdd);
+		kfd_process_device_destroy_ib_mem(pdd);
 
-		amd_iommu_unbind_pasid(pdd->dev->pdev, p->pasid);
-		list_del(&pdd->per_device_list);
+		if (pdd->drm_file)
+			fput(pdd->drm_file);
+
+		if (pdd->qpd.cwsr_kaddr && !pdd->qpd.cwsr_base)
+			free_pages((unsigned long)pdd->qpd.cwsr_kaddr,
+				get_order(KFD_CWSR_TBA_TMA_SIZE));
+
+		idr_destroy(&pdd->alloc_idr);
+
+		kfd_free_process_doorbells(pdd->dev->kfd, pdd);
+
+		if (pdd->dev->kfd->shared_resources.enable_mes &&
+			pdd->proc_ctx_cpu_ptr)
+			amdgpu_amdkfd_free_gtt_mem(pdd->dev->adev,
+						   &pdd->proc_ctx_bo);
+		/*
+		 * before destroying pdd, make sure to report availability
+		 * for auto suspend
+		 */
+		if (pdd->runtime_inuse) {
+			pm_runtime_mark_last_busy(adev_to_drm(pdd->dev->adev)->dev);
+			pm_runtime_put_autosuspend(adev_to_drm(pdd->dev->adev)->dev);
+			pdd->runtime_inuse = false;
+		}
+
+		atomic_dec(&pdd->dev->kfd->kfd_processes_count);
 
 		kfree(pdd);
+		p->pdds[i] = NULL;
 	}
+	p->n_pdds = 0;
+}
 
-	kfd_event_free_process(p);
+static void kfd_process_remove_sysfs(struct kfd_process *p)
+{
+	struct kfd_process_device *pdd;
+	int i;
 
-	kfd_pasid_free(p->pasid);
+	if (!p->kobj)
+		return;
 
-	mutex_unlock(&p->mutex);
+	sysfs_remove_file(p->kobj, &p->attr_pasid);
+	kobject_del(p->kobj_queues);
+	kobject_put(p->kobj_queues);
+	p->kobj_queues = NULL;
+
+	for (i = 0; i < p->n_pdds; i++) {
+		pdd = p->pdds[i];
+
+		sysfs_remove_file(p->kobj, &pdd->attr_vram);
+		sysfs_remove_file(p->kobj, &pdd->attr_sdma);
+
+		sysfs_remove_file(pdd->kobj_stats, &pdd->attr_evict);
+		if (pdd->dev->kfd2kgd->get_cu_occupancy)
+			sysfs_remove_file(pdd->kobj_stats,
+					  &pdd->attr_cu_occupancy);
+		kobject_del(pdd->kobj_stats);
+		kobject_put(pdd->kobj_stats);
+		pdd->kobj_stats = NULL;
+	}
+
+	for_each_set_bit(i, p->svms.bitmap_supported, p->n_pdds) {
+		pdd = p->pdds[i];
+
+		sysfs_remove_file(pdd->kobj_counters, &pdd->attr_faults);
+		sysfs_remove_file(pdd->kobj_counters, &pdd->attr_page_in);
+		sysfs_remove_file(pdd->kobj_counters, &pdd->attr_page_out);
+		kobject_del(pdd->kobj_counters);
+		kobject_put(pdd->kobj_counters);
+		pdd->kobj_counters = NULL;
+	}
+
+	kobject_del(p->kobj);
+	kobject_put(p->kobj);
+	p->kobj = NULL;
+}
+
+/*
+ * If any GPU is ongoing reset, wait for reset complete.
+ */
+static void kfd_process_wait_gpu_reset_complete(struct kfd_process *p)
+{
+	int i;
+
+	for (i = 0; i < p->n_pdds; i++)
+		flush_workqueue(p->pdds[i]->dev->adev->reset_domain->wq);
+}
+
+/* No process locking is needed in this function, because the process
+ * is not findable any more. We must assume that no other thread is
+ * using it any more, otherwise we couldn't safely free the process
+ * structure in the end.
+ */
+static void kfd_process_wq_release(struct work_struct *work)
+{
+	struct kfd_process *p = container_of(work, struct kfd_process,
+					     release_work);
+	struct dma_fence *ef;
+
+	kfd_process_dequeue_from_all_devices(p);
+	pqm_uninit(&p->pqm);
+
+	/*
+	 * If GPU in reset, user queues may still running, wait for reset complete.
+	 */
+	kfd_process_wait_gpu_reset_complete(p);
+
+	/* Signal the eviction fence after user mode queues are
+	 * destroyed. This allows any BOs to be freed without
+	 * triggering pointless evictions or waiting for fences.
+	 */
+	synchronize_rcu();
+	ef = rcu_access_pointer(p->ef);
+	if (ef)
+		dma_fence_signal(ef);
+
+	kfd_process_remove_sysfs(p);
+	kfd_debugfs_remove_process(p);
+
+	kfd_process_kunmap_signal_bo(p);
+	kfd_process_free_outstanding_kfd_bos(p);
+	svm_range_list_fini(p);
+
+	kfd_process_destroy_pdds(p);
+	dma_fence_put(ef);
+
+	kfd_event_free_process(p);
 
 	mutex_destroy(&p->mutex);
 
-	kfree(p->queues);
+	put_task_struct(p->lead_thread);
 
 	kfree(p);
+}
+
+static void kfd_process_ref_release(struct kref *ref)
+{
+	struct kfd_process *p = container_of(ref, struct kfd_process, ref);
 
-	kfree(work);
+	INIT_WORK(&p->release_work, kfd_process_wq_release);
+	queue_work(kfd_process_wq, &p->release_work);
 }
 
-static void kfd_process_destroy_delayed(struct rcu_head *rcu)
+static struct mmu_notifier *kfd_process_alloc_notifier(struct mm_struct *mm)
 {
-	struct kfd_process_release_work *work;
-	struct kfd_process *p;
+	/* This increments p->ref counter if kfd process p exists */
+	struct kfd_process *p = kfd_lookup_process_by_mm(mm);
+
+	return p ? &p->mmu_notifier : ERR_PTR(-ESRCH);
+}
+
+static void kfd_process_free_notifier(struct mmu_notifier *mn)
+{
+	kfd_unref_process(container_of(mn, struct kfd_process, mmu_notifier));
+}
+
+static void kfd_process_notifier_release_internal(struct kfd_process *p)
+{
+	int i;
 
-	p = container_of(rcu, struct kfd_process, rcu);
-	WARN_ON(atomic_read(&p->mm->mm_count) <= 0);
+	cancel_delayed_work_sync(&p->eviction_work);
+	cancel_delayed_work_sync(&p->restore_work);
 
-	mmdrop(p->mm);
+	for (i = 0; i < p->n_pdds; i++) {
+		struct kfd_process_device *pdd = p->pdds[i];
 
-	work = kmalloc(sizeof(struct kfd_process_release_work), GFP_ATOMIC);
+		/* re-enable GFX OFF since runtime enable with ttmp setup disabled it. */
+		if (!kfd_dbg_is_rlc_restore_supported(pdd->dev) && p->runtime_info.ttmp_setup)
+			amdgpu_gfx_off_ctrl(pdd->dev->adev, true);
+	}
+
+	/* Indicate to other users that MM is no longer valid */
+	p->mm = NULL;
+	kfd_dbg_trap_disable(p);
+
+	if (atomic_read(&p->debugged_process_count) > 0) {
+		struct kfd_process *target;
+		unsigned int temp;
+		int idx = srcu_read_lock(&kfd_processes_srcu);
+
+		hash_for_each_rcu(kfd_processes_table, temp, target, kfd_processes) {
+			if (target->debugger_process && target->debugger_process == p) {
+				mutex_lock_nested(&target->mutex, 1);
+				kfd_dbg_trap_disable(target);
+				mutex_unlock(&target->mutex);
+				if (atomic_read(&p->debugged_process_count) == 0)
+					break;
+			}
+		}
 
-	if (work) {
-		INIT_WORK((struct work_struct *) work, kfd_process_wq_release);
-		work->p = p;
-		queue_work(kfd_process_wq, (struct work_struct *) work);
+		srcu_read_unlock(&kfd_processes_srcu, idx);
 	}
+
+	mmu_notifier_put(&p->mmu_notifier);
 }
 
 static void kfd_process_notifier_release(struct mmu_notifier *mn,
 					struct mm_struct *mm)
 {
 	struct kfd_process *p;
-	struct kfd_process_device *pdd = NULL;
 
 	/*
 	 * The kfd_process structure can not be free because the
@@ -229,145 +1273,472 @@ static void kfd_process_notifier_release(struct mmu_notifier *mn,
 		return;
 
 	mutex_lock(&kfd_processes_mutex);
+	/*
+	 * Do early return if table is empty.
+	 *
+	 * This could potentially happen if this function is called concurrently
+	 * by mmu_notifier and by kfd_cleanup_pocesses.
+	 *
+	 */
+	if (hash_empty(kfd_processes_table)) {
+		mutex_unlock(&kfd_processes_mutex);
+		return;
+	}
 	hash_del_rcu(&p->kfd_processes);
 	mutex_unlock(&kfd_processes_mutex);
 	synchronize_srcu(&kfd_processes_srcu);
 
-	mutex_lock(&p->mutex);
+	kfd_process_notifier_release_internal(p);
+}
 
-	/* In case our notifier is called before IOMMU notifier */
-	pqm_uninit(&p->pqm);
+static const struct mmu_notifier_ops kfd_process_mmu_notifier_ops = {
+	.release = kfd_process_notifier_release,
+	.alloc_notifier = kfd_process_alloc_notifier,
+	.free_notifier = kfd_process_free_notifier,
+};
 
-	/* Iterate over all process device data structure and check
-	 * if we should delete debug managers and reset all wavefronts
+/*
+ * This code handles the case when driver is being unloaded before all
+ * mm_struct are released.  We need to safely free the kfd_process and
+ * avoid race conditions with mmu_notifier that might try to free them.
+ *
+ */
+void kfd_cleanup_processes(void)
+{
+	struct kfd_process *p;
+	struct hlist_node *p_temp;
+	unsigned int temp;
+	HLIST_HEAD(cleanup_list);
+
+	/*
+	 * Move all remaining kfd_process from the process table to a
+	 * temp list for processing.   Once done, callback from mmu_notifier
+	 * release will not see the kfd_process in the table and do early return,
+	 * avoiding double free issues.
 	 */
-	list_for_each_entry(pdd, &p->per_device_data, per_device_list) {
-		if ((pdd->dev->dbgmgr) &&
-				(pdd->dev->dbgmgr->pasid == p->pasid))
-			kfd_dbgmgr_destroy(pdd->dev->dbgmgr);
-
-		if (pdd->reset_wavefronts) {
-			pr_warn("Resetting all wave fronts\n");
-			dbgdev_wave_reset_wavefronts(pdd->dev, p);
-			pdd->reset_wavefronts = false;
-		}
+	mutex_lock(&kfd_processes_mutex);
+	hash_for_each_safe(kfd_processes_table, temp, p_temp, p, kfd_processes) {
+		hash_del_rcu(&p->kfd_processes);
+		synchronize_srcu(&kfd_processes_srcu);
+		hlist_add_head(&p->kfd_processes, &cleanup_list);
 	}
+	mutex_unlock(&kfd_processes_mutex);
 
-	mutex_unlock(&p->mutex);
+	hlist_for_each_entry_safe(p, p_temp, &cleanup_list, kfd_processes)
+		kfd_process_notifier_release_internal(p);
 
 	/*
-	 * Because we drop mm_count inside kfd_process_destroy_delayed
-	 * and because the mmu_notifier_unregister function also drop
-	 * mm_count we need to take an extra count here.
+	 * Ensures that all outstanding free_notifier get called, triggering
+	 * the release of the kfd_process struct.
 	 */
-	mmgrab(p->mm);
-	mmu_notifier_unregister_no_release(&p->mmu_notifier, p->mm);
-	mmu_notifier_call_srcu(&p->rcu, &kfd_process_destroy_delayed);
+	mmu_notifier_synchronize();
 }
 
-static const struct mmu_notifier_ops kfd_process_mmu_notifier_ops = {
-	.release = kfd_process_notifier_release,
-};
+int kfd_process_init_cwsr_apu(struct kfd_process *p, struct file *filep)
+{
+	unsigned long  offset;
+	int i;
 
-static struct kfd_process *create_process(const struct task_struct *thread)
+	if (p->has_cwsr)
+		return 0;
+
+	for (i = 0; i < p->n_pdds; i++) {
+		struct kfd_node *dev = p->pdds[i]->dev;
+		struct qcm_process_device *qpd = &p->pdds[i]->qpd;
+
+		if (!dev->kfd->cwsr_enabled || qpd->cwsr_kaddr || qpd->cwsr_base)
+			continue;
+
+		offset = KFD_MMAP_TYPE_RESERVED_MEM | KFD_MMAP_GPU_ID(dev->id);
+		qpd->tba_addr = (int64_t)vm_mmap(filep, 0,
+			KFD_CWSR_TBA_TMA_SIZE, PROT_READ | PROT_EXEC,
+			MAP_SHARED, offset);
+
+		if (IS_ERR_VALUE(qpd->tba_addr)) {
+			int err = qpd->tba_addr;
+
+			dev_err(dev->adev->dev,
+				"Failure to set tba address. error %d.\n", err);
+			qpd->tba_addr = 0;
+			qpd->cwsr_kaddr = NULL;
+			return err;
+		}
+
+		memcpy(qpd->cwsr_kaddr, dev->kfd->cwsr_isa, dev->kfd->cwsr_isa_size);
+
+		kfd_process_set_trap_debug_flag(qpd, p->debug_trap_enabled);
+
+		qpd->tma_addr = qpd->tba_addr + KFD_CWSR_TMA_OFFSET;
+		pr_debug("set tba :0x%llx, tma:0x%llx, cwsr_kaddr:%p for pqm.\n",
+			qpd->tba_addr, qpd->tma_addr, qpd->cwsr_kaddr);
+	}
+
+	p->has_cwsr = true;
+
+	return 0;
+}
+
+static int kfd_process_device_init_cwsr_dgpu(struct kfd_process_device *pdd)
 {
-	struct kfd_process *process;
-	int err = -ENOMEM;
+	struct kfd_node *dev = pdd->dev;
+	struct qcm_process_device *qpd = &pdd->qpd;
+	uint32_t flags = KFD_IOC_ALLOC_MEM_FLAGS_GTT
+			| KFD_IOC_ALLOC_MEM_FLAGS_NO_SUBSTITUTE
+			| KFD_IOC_ALLOC_MEM_FLAGS_EXECUTABLE;
+	struct kgd_mem *mem;
+	void *kaddr;
+	int ret;
+
+	if (!dev->kfd->cwsr_enabled || qpd->cwsr_kaddr || !qpd->cwsr_base)
+		return 0;
+
+	/* cwsr_base is only set for dGPU */
+	ret = kfd_process_alloc_gpuvm(pdd, qpd->cwsr_base,
+				      KFD_CWSR_TBA_TMA_SIZE, flags, &mem, &kaddr);
+	if (ret)
+		return ret;
+
+	qpd->cwsr_mem = mem;
+	qpd->cwsr_kaddr = kaddr;
+	qpd->tba_addr = qpd->cwsr_base;
+
+	memcpy(qpd->cwsr_kaddr, dev->kfd->cwsr_isa, dev->kfd->cwsr_isa_size);
+
+	kfd_process_set_trap_debug_flag(&pdd->qpd,
+					pdd->process->debug_trap_enabled);
+
+	qpd->tma_addr = qpd->tba_addr + KFD_CWSR_TMA_OFFSET;
+	pr_debug("set tba :0x%llx, tma:0x%llx, cwsr_kaddr:%p for pqm.\n",
+		 qpd->tba_addr, qpd->tma_addr, qpd->cwsr_kaddr);
+
+	return 0;
+}
 
-	process = kzalloc(sizeof(*process), GFP_KERNEL);
+static void kfd_process_device_destroy_cwsr_dgpu(struct kfd_process_device *pdd)
+{
+	struct kfd_node *dev = pdd->dev;
+	struct qcm_process_device *qpd = &pdd->qpd;
 
-	if (!process)
-		goto err_alloc_process;
+	if (!dev->kfd->cwsr_enabled || !qpd->cwsr_kaddr || !qpd->cwsr_base)
+		return;
 
-	process->queues = kmalloc_array(INITIAL_QUEUE_ARRAY_SIZE,
-					sizeof(process->queues[0]), GFP_KERNEL);
-	if (!process->queues)
-		goto err_alloc_queues;
+	kfd_process_free_gpuvm(qpd->cwsr_mem, pdd, &qpd->cwsr_kaddr);
+}
 
-	process->pasid = kfd_pasid_alloc();
-	if (process->pasid == 0)
-		goto err_alloc_pasid;
+void kfd_process_set_trap_handler(struct qcm_process_device *qpd,
+				  uint64_t tba_addr,
+				  uint64_t tma_addr)
+{
+	if (qpd->cwsr_kaddr) {
+		/* KFD trap handler is bound, record as second-level TBA/TMA
+		 * in first-level TMA. First-level trap will jump to second.
+		 */
+		uint64_t *tma =
+			(uint64_t *)(qpd->cwsr_kaddr + KFD_CWSR_TMA_OFFSET);
+		tma[0] = tba_addr;
+		tma[1] = tma_addr;
+	} else {
+		/* No trap handler bound, bind as first-level TBA/TMA. */
+		qpd->tba_addr = tba_addr;
+		qpd->tma_addr = tma_addr;
+	}
+}
 
-	mutex_init(&process->mutex);
+bool kfd_process_xnack_mode(struct kfd_process *p, bool supported)
+{
+	int i;
+
+	/* On most GFXv9 GPUs, the retry mode in the SQ must match the
+	 * boot time retry setting. Mixing processes with different
+	 * XNACK/retry settings can hang the GPU.
+	 *
+	 * Different GPUs can have different noretry settings depending
+	 * on HW bugs or limitations. We need to find at least one
+	 * XNACK mode for this process that's compatible with all GPUs.
+	 * Fortunately GPUs with retry enabled (noretry=0) can run code
+	 * built for XNACK-off. On GFXv9 it may perform slower.
+	 *
+	 * Therefore applications built for XNACK-off can always be
+	 * supported and will be our fallback if any GPU does not
+	 * support retry.
+	 */
+	for (i = 0; i < p->n_pdds; i++) {
+		struct kfd_node *dev = p->pdds[i]->dev;
+
+		/* Only consider GFXv9 and higher GPUs. Older GPUs don't
+		 * support the SVM APIs and don't need to be considered
+		 * for the XNACK mode selection.
+		 */
+		if (!KFD_IS_SOC15(dev))
+			continue;
+		/* Aldebaran can always support XNACK because it can support
+		 * per-process XNACK mode selection. But let the dev->noretry
+		 * setting still influence the default XNACK mode.
+		 */
+		if (supported && KFD_SUPPORT_XNACK_PER_PROCESS(dev)) {
+			if (!amdgpu_sriov_xnack_support(dev->kfd->adev)) {
+				pr_debug("SRIOV platform xnack not supported\n");
+				return false;
+			}
+			continue;
+		}
 
-	process->mm = thread->mm;
+		/* GFXv10 and later GPUs do not support shader preemption
+		 * during page faults. This can lead to poor QoS for queue
+		 * management and memory-manager-related preemptions or
+		 * even deadlocks.
+		 */
+		if (KFD_GC_VERSION(dev) >= IP_VERSION(10, 1, 1))
+			return false;
 
-	/* register notifier */
-	process->mmu_notifier.ops = &kfd_process_mmu_notifier_ops;
-	err = __mmu_notifier_register(&process->mmu_notifier, process->mm);
-	if (err)
-		goto err_mmu_notifier;
+		if (dev->kfd->noretry)
+			return false;
+	}
 
-	hash_add_rcu(kfd_processes_table, &process->kfd_processes,
-			(uintptr_t)process->mm);
+	return true;
+}
 
-	process->lead_thread = thread->group_leader;
+void kfd_process_set_trap_debug_flag(struct qcm_process_device *qpd,
+				     bool enabled)
+{
+	if (qpd->cwsr_kaddr) {
+		uint64_t *tma =
+			(uint64_t *)(qpd->cwsr_kaddr + KFD_CWSR_TMA_OFFSET);
+		tma[2] = enabled;
+	}
+}
 
-	process->queue_array_size = INITIAL_QUEUE_ARRAY_SIZE;
+/*
+ * On return the kfd_process is fully operational and will be freed when the
+ * mm is released
+ */
+static struct kfd_process *create_process(const struct task_struct *thread)
+{
+	struct kfd_process *process;
+	struct mmu_notifier *mn;
+	int err = -ENOMEM;
 
-	INIT_LIST_HEAD(&process->per_device_data);
+	process = kzalloc(sizeof(*process), GFP_KERNEL);
+	if (!process)
+		goto err_alloc_process;
 
-	kfd_event_init_process(process);
+	kref_init(&process->ref);
+	mutex_init(&process->mutex);
+	process->mm = thread->mm;
+	process->lead_thread = thread->group_leader;
+	process->n_pdds = 0;
+	process->queues_paused = false;
+	INIT_DELAYED_WORK(&process->eviction_work, evict_process_worker);
+	INIT_DELAYED_WORK(&process->restore_work, restore_process_worker);
+	process->last_restore_timestamp = get_jiffies_64();
+	err = kfd_event_init_process(process);
+	if (err)
+		goto err_event_init;
+	process->is_32bit_user_mode = in_compat_syscall();
+	process->debug_trap_enabled = false;
+	process->debugger_process = NULL;
+	process->exception_enable_mask = 0;
+	atomic_set(&process->debugged_process_count, 0);
+	sema_init(&process->runtime_enable_sema, 0);
 
 	err = pqm_init(&process->pqm, process);
 	if (err != 0)
 		goto err_process_pqm_init;
 
 	/* init process apertures*/
-	process->is_32bit_user_mode = in_compat_syscall();
 	err = kfd_init_apertures(process);
 	if (err != 0)
 		goto err_init_apertures;
 
+	/* Check XNACK support after PDDs are created in kfd_init_apertures */
+	process->xnack_enabled = kfd_process_xnack_mode(process, false);
+
+	err = svm_range_list_init(process);
+	if (err)
+		goto err_init_svm_range_list;
+
+	/* alloc_notifier needs to find the process in the hash table */
+	hash_add_rcu(kfd_processes_table, &process->kfd_processes,
+			(uintptr_t)process->mm);
+
+	/* Avoid free_notifier to start kfd_process_wq_release if
+	 * mmu_notifier_get failed because of pending signal.
+	 */
+	kref_get(&process->ref);
+
+	/* MMU notifier registration must be the last call that can fail
+	 * because after this point we cannot unwind the process creation.
+	 * After this point, mmu_notifier_put will trigger the cleanup by
+	 * dropping the last process reference in the free_notifier.
+	 */
+	mn = mmu_notifier_get(&kfd_process_mmu_notifier_ops, process->mm);
+	if (IS_ERR(mn)) {
+		err = PTR_ERR(mn);
+		goto err_register_notifier;
+	}
+	BUG_ON(mn != &process->mmu_notifier);
+
+	kfd_unref_process(process);
+	get_task_struct(process->lead_thread);
+
+	INIT_WORK(&process->debug_event_workarea, debug_event_write_work_handler);
+
 	return process;
 
+err_register_notifier:
+	hash_del_rcu(&process->kfd_processes);
+	svm_range_list_fini(process);
+err_init_svm_range_list:
+	kfd_process_free_outstanding_kfd_bos(process);
+	kfd_process_destroy_pdds(process);
 err_init_apertures:
 	pqm_uninit(&process->pqm);
 err_process_pqm_init:
-	hash_del_rcu(&process->kfd_processes);
-	synchronize_rcu();
-	mmu_notifier_unregister_no_release(&process->mmu_notifier, process->mm);
-err_mmu_notifier:
+	kfd_event_free_process(process);
+err_event_init:
 	mutex_destroy(&process->mutex);
-	kfd_pasid_free(process->pasid);
-err_alloc_pasid:
-	kfree(process->queues);
-err_alloc_queues:
 	kfree(process);
 err_alloc_process:
 	return ERR_PTR(err);
 }
 
-struct kfd_process_device *kfd_get_process_device_data(struct kfd_dev *dev,
+struct kfd_process_device *kfd_get_process_device_data(struct kfd_node *dev,
 							struct kfd_process *p)
 {
-	struct kfd_process_device *pdd = NULL;
+	int i;
 
-	list_for_each_entry(pdd, &p->per_device_data, per_device_list)
-		if (pdd->dev == dev)
-			break;
+	for (i = 0; i < p->n_pdds; i++)
+		if (p->pdds[i]->dev == dev)
+			return p->pdds[i];
 
-	return pdd;
+	return NULL;
 }
 
-struct kfd_process_device *kfd_create_process_device_data(struct kfd_dev *dev,
+struct kfd_process_device *kfd_create_process_device_data(struct kfd_node *dev,
 							struct kfd_process *p)
 {
 	struct kfd_process_device *pdd = NULL;
 
+	if (WARN_ON_ONCE(p->n_pdds >= MAX_GPU_INSTANCE))
+		return NULL;
 	pdd = kzalloc(sizeof(*pdd), GFP_KERNEL);
-	if (pdd != NULL) {
-		pdd->dev = dev;
-		INIT_LIST_HEAD(&pdd->qpd.queues_list);
-		INIT_LIST_HEAD(&pdd->qpd.priv_queue_list);
-		pdd->qpd.dqm = dev->dqm;
-		pdd->reset_wavefronts = false;
-		list_add(&pdd->per_device_list, &p->per_device_data);
-	}
+	if (!pdd)
+		return NULL;
+
+	pdd->dev = dev;
+	INIT_LIST_HEAD(&pdd->qpd.queues_list);
+	INIT_LIST_HEAD(&pdd->qpd.priv_queue_list);
+	pdd->qpd.dqm = dev->dqm;
+	pdd->qpd.pqm = &p->pqm;
+	pdd->qpd.evicted = 0;
+	pdd->qpd.mapped_gws_queue = false;
+	pdd->process = p;
+	pdd->bound = PDD_UNBOUND;
+	pdd->already_dequeued = false;
+	pdd->runtime_inuse = false;
+	atomic64_set(&pdd->vram_usage, 0);
+	pdd->sdma_past_activity_counter = 0;
+	pdd->user_gpu_id = dev->id;
+	atomic64_set(&pdd->evict_duration_counter, 0);
+
+	p->pdds[p->n_pdds++] = pdd;
+	if (kfd_dbg_is_per_vmid_supported(pdd->dev))
+		pdd->spi_dbg_override = pdd->dev->kfd2kgd->disable_debug_trap(
+							pdd->dev->adev,
+							false,
+							0);
+
+	/* Init idr used for memory handle translation */
+	idr_init(&pdd->alloc_idr);
+
+	atomic_inc(&dev->kfd->kfd_processes_count);
 
 	return pdd;
 }
 
+/**
+ * kfd_process_device_init_vm - Initialize a VM for a process-device
+ *
+ * @pdd: The process-device
+ * @drm_file: Optional pointer to a DRM file descriptor
+ *
+ * If @drm_file is specified, it will be used to acquire the VM from
+ * that file descriptor. If successful, the @pdd takes ownership of
+ * the file descriptor.
+ *
+ * If @drm_file is NULL, a new VM is created.
+ *
+ * Returns 0 on success, -errno on failure.
+ */
+int kfd_process_device_init_vm(struct kfd_process_device *pdd,
+			       struct file *drm_file)
+{
+	struct amdgpu_fpriv *drv_priv;
+	struct amdgpu_vm *avm;
+	struct kfd_process *p;
+	struct dma_fence *ef;
+	struct kfd_node *dev;
+	int ret;
+
+	if (!drm_file)
+		return -EINVAL;
+
+	if (pdd->drm_priv)
+		return -EBUSY;
+
+	ret = amdgpu_file_to_fpriv(drm_file, &drv_priv);
+	if (ret)
+		return ret;
+	avm = &drv_priv->vm;
+
+	p = pdd->process;
+	dev = pdd->dev;
+
+	ret = amdgpu_amdkfd_gpuvm_acquire_process_vm(dev->adev, avm,
+						     &p->kgd_process_info,
+						     p->ef ? NULL : &ef);
+	if (ret) {
+		dev_err(dev->adev->dev, "Failed to create process VM object\n");
+		return ret;
+	}
+
+	if (!p->ef)
+		RCU_INIT_POINTER(p->ef, ef);
+
+	pdd->drm_priv = drm_file->private_data;
+
+	ret = kfd_process_device_reserve_ib_mem(pdd);
+	if (ret)
+		goto err_reserve_ib_mem;
+	ret = kfd_process_device_init_cwsr_dgpu(pdd);
+	if (ret)
+		goto err_init_cwsr;
+
+	if (unlikely(!avm->pasid)) {
+		dev_warn(pdd->dev->adev->dev, "WARN: vm %p has no pasid associated",
+				 avm);
+		ret = -EINVAL;
+		goto err_get_pasid;
+	}
+
+	pdd->pasid = avm->pasid;
+	pdd->drm_file = drm_file;
+
+	kfd_smi_event_process(pdd, true);
+
+	return 0;
+
+err_get_pasid:
+	kfd_process_device_destroy_cwsr_dgpu(pdd);
+err_init_cwsr:
+	kfd_process_device_destroy_ib_mem(pdd);
+err_reserve_ib_mem:
+	pdd->drm_priv = NULL;
+	amdgpu_amdkfd_gpuvm_destroy_cb(dev->adev, avm);
+
+	return ret;
+}
+
 /*
  * Direct the IOMMU to bind the process (specifically the pasid->mm)
  * to the device.
@@ -375,7 +1746,7 @@ struct kfd_process_device *kfd_create_process_device_data(struct kfd_dev *dev,
  *
  * Assumes that the process lock is held.
  */
-struct kfd_process_device *kfd_bind_process_to_device(struct kfd_dev *dev,
+struct kfd_process_device *kfd_bind_process_to_device(struct kfd_node *dev,
 							struct kfd_process *p)
 {
 	struct kfd_process_device *pdd;
@@ -383,105 +1754,586 @@ struct kfd_process_device *kfd_bind_process_to_device(struct kfd_dev *dev,
 
 	pdd = kfd_get_process_device_data(dev, p);
 	if (!pdd) {
-		pr_err("Process device data doesn't exist\n");
+		dev_err(dev->adev->dev, "Process device data doesn't exist\n");
 		return ERR_PTR(-ENOMEM);
 	}
 
-	if (pdd->bound)
-		return pdd;
+	if (!pdd->drm_priv)
+		return ERR_PTR(-ENODEV);
 
-	err = amd_iommu_bind_pasid(dev->pdev, p->pasid, p->lead_thread);
-	if (err < 0)
-		return ERR_PTR(err);
+	/*
+	 * signal runtime-pm system to auto resume and prevent
+	 * further runtime suspend once device pdd is created until
+	 * pdd is destroyed.
+	 */
+	if (!pdd->runtime_inuse) {
+		err = pm_runtime_get_sync(adev_to_drm(dev->adev)->dev);
+		if (err < 0) {
+			pm_runtime_put_autosuspend(adev_to_drm(dev->adev)->dev);
+			return ERR_PTR(err);
+		}
+	}
 
-	pdd->bound = true;
+	/*
+	 * make sure that runtime_usage counter is incremented just once
+	 * per pdd
+	 */
+	pdd->runtime_inuse = true;
 
 	return pdd;
 }
 
-void kfd_unbind_process_from_device(struct kfd_dev *dev, unsigned int pasid)
+/* Create specific handle mapped to mem from process local memory idr
+ * Assumes that the process lock is held.
+ */
+int kfd_process_device_create_obj_handle(struct kfd_process_device *pdd,
+					void *mem)
+{
+	return idr_alloc(&pdd->alloc_idr, mem, 0, 0, GFP_KERNEL);
+}
+
+/* Translate specific handle from process local memory idr
+ * Assumes that the process lock is held.
+ */
+void *kfd_process_device_translate_handle(struct kfd_process_device *pdd,
+					int handle)
+{
+	if (handle < 0)
+		return NULL;
+
+	return idr_find(&pdd->alloc_idr, handle);
+}
+
+/* Remove specific handle from process local memory idr
+ * Assumes that the process lock is held.
+ */
+void kfd_process_device_remove_obj_handle(struct kfd_process_device *pdd,
+					int handle)
 {
+	if (handle >= 0)
+		idr_remove(&pdd->alloc_idr, handle);
+}
+
+static struct kfd_process_device *kfd_lookup_process_device_by_pasid(u32 pasid)
+{
+	struct kfd_process_device *ret_p = NULL;
 	struct kfd_process *p;
+	unsigned int temp;
+	int i;
+
+	hash_for_each_rcu(kfd_processes_table, temp, p, kfd_processes) {
+		for (i = 0; i < p->n_pdds; i++) {
+			if (p->pdds[i]->pasid == pasid) {
+				ret_p = p->pdds[i];
+				break;
+			}
+		}
+		if (ret_p)
+			break;
+	}
+	return ret_p;
+}
+
+/* This increments the process->ref counter. */
+struct kfd_process *kfd_lookup_process_by_pasid(u32 pasid,
+						struct kfd_process_device **pdd)
+{
+	struct kfd_process_device *ret_p;
+
+	int idx = srcu_read_lock(&kfd_processes_srcu);
+
+	ret_p = kfd_lookup_process_device_by_pasid(pasid);
+	if (ret_p) {
+		if (pdd)
+			*pdd = ret_p;
+		kref_get(&ret_p->process->ref);
+
+		srcu_read_unlock(&kfd_processes_srcu, idx);
+		return ret_p->process;
+	}
+
+	srcu_read_unlock(&kfd_processes_srcu, idx);
+
+	if (pdd)
+		*pdd = NULL;
+
+	return NULL;
+}
+
+/* This increments the process->ref counter. */
+struct kfd_process *kfd_lookup_process_by_mm(const struct mm_struct *mm)
+{
+	struct kfd_process *p;
+
+	int idx = srcu_read_lock(&kfd_processes_srcu);
+
+	p = find_process_by_mm(mm);
+	if (p)
+		kref_get(&p->ref);
+
+	srcu_read_unlock(&kfd_processes_srcu, idx);
+
+	return p;
+}
+
+/* kfd_process_evict_queues - Evict all user queues of a process
+ *
+ * Eviction is reference-counted per process-device. This means multiple
+ * evictions from different sources can be nested safely.
+ */
+int kfd_process_evict_queues(struct kfd_process *p, uint32_t trigger)
+{
+	int r = 0;
+	int i;
+	unsigned int n_evicted = 0;
+
+	for (i = 0; i < p->n_pdds; i++) {
+		struct kfd_process_device *pdd = p->pdds[i];
+		struct device *dev = pdd->dev->adev->dev;
+
+		kfd_smi_event_queue_eviction(pdd->dev, p->lead_thread->pid,
+					     trigger);
+
+		r = pdd->dev->dqm->ops.evict_process_queues(pdd->dev->dqm,
+							    &pdd->qpd);
+		/* evict return -EIO if HWS is hang or asic is resetting, in this case
+		 * we would like to set all the queues to be in evicted state to prevent
+		 * them been add back since they actually not be saved right now.
+		 */
+		if (r && r != -EIO) {
+			dev_err(dev, "Failed to evict process queues\n");
+			goto fail;
+		}
+		n_evicted++;
+
+		pdd->dev->dqm->is_hws_hang = false;
+	}
+
+	return r;
+
+fail:
+	/* To keep state consistent, roll back partial eviction by
+	 * restoring queues
+	 */
+	for (i = 0; i < p->n_pdds; i++) {
+		struct kfd_process_device *pdd = p->pdds[i];
+
+		if (n_evicted == 0)
+			break;
+
+		kfd_smi_event_queue_restore(pdd->dev, p->lead_thread->pid);
+
+		if (pdd->dev->dqm->ops.restore_process_queues(pdd->dev->dqm,
+							      &pdd->qpd))
+			dev_err(pdd->dev->adev->dev,
+				"Failed to restore queues\n");
+
+		n_evicted--;
+	}
+
+	return r;
+}
+
+/* kfd_process_restore_queues - Restore all user queues of a process */
+int kfd_process_restore_queues(struct kfd_process *p)
+{
+	int r, ret = 0;
+	int i;
+
+	for (i = 0; i < p->n_pdds; i++) {
+		struct kfd_process_device *pdd = p->pdds[i];
+		struct device *dev = pdd->dev->adev->dev;
+
+		kfd_smi_event_queue_restore(pdd->dev, p->lead_thread->pid);
+
+		r = pdd->dev->dqm->ops.restore_process_queues(pdd->dev->dqm,
+							      &pdd->qpd);
+		if (r) {
+			dev_err(dev, "Failed to restore process queues\n");
+			if (!ret)
+				ret = r;
+		}
+	}
+
+	return ret;
+}
+
+int kfd_process_gpuidx_from_gpuid(struct kfd_process *p, uint32_t gpu_id)
+{
+	int i;
+
+	for (i = 0; i < p->n_pdds; i++)
+		if (p->pdds[i] && gpu_id == p->pdds[i]->user_gpu_id)
+			return i;
+	return -EINVAL;
+}
+
+int
+kfd_process_gpuid_from_node(struct kfd_process *p, struct kfd_node *node,
+			    uint32_t *gpuid, uint32_t *gpuidx)
+{
+	int i;
+
+	for (i = 0; i < p->n_pdds; i++)
+		if (p->pdds[i] && p->pdds[i]->dev == node) {
+			*gpuid = p->pdds[i]->user_gpu_id;
+			*gpuidx = i;
+			return 0;
+		}
+	return -EINVAL;
+}
+
+static int signal_eviction_fence(struct kfd_process *p)
+{
+	struct dma_fence *ef;
+	int ret;
+
+	rcu_read_lock();
+	ef = dma_fence_get_rcu_safe(&p->ef);
+	rcu_read_unlock();
+	if (!ef)
+		return -EINVAL;
+
+	ret = dma_fence_signal(ef);
+	dma_fence_put(ef);
+
+	return ret;
+}
+
+static void evict_process_worker(struct work_struct *work)
+{
+	int ret;
+	struct kfd_process *p;
+	struct delayed_work *dwork;
+
+	dwork = to_delayed_work(work);
+
+	/* Process termination destroys this worker thread. So during the
+	 * lifetime of this thread, kfd_process p will be valid
+	 */
+	p = container_of(dwork, struct kfd_process, eviction_work);
+
+	pr_debug("Started evicting process pid %d\n", p->lead_thread->pid);
+	ret = kfd_process_evict_queues(p, KFD_QUEUE_EVICTION_TRIGGER_TTM);
+	if (!ret) {
+		/* If another thread already signaled the eviction fence,
+		 * they are responsible stopping the queues and scheduling
+		 * the restore work.
+		 */
+		if (signal_eviction_fence(p) ||
+		    mod_delayed_work(kfd_restore_wq, &p->restore_work,
+				     msecs_to_jiffies(PROCESS_RESTORE_TIME_MS)))
+			kfd_process_restore_queues(p);
+
+		pr_debug("Finished evicting process pid %d\n", p->lead_thread->pid);
+	} else
+		pr_err("Failed to evict queues of process pid %d\n", p->lead_thread->pid);
+}
+
+static int restore_process_helper(struct kfd_process *p)
+{
+	int ret = 0;
+
+	/* VMs may not have been acquired yet during debugging. */
+	if (p->kgd_process_info) {
+		ret = amdgpu_amdkfd_gpuvm_restore_process_bos(
+			p->kgd_process_info, &p->ef);
+		if (ret)
+			return ret;
+	}
+
+	ret = kfd_process_restore_queues(p);
+	if (!ret)
+		pr_debug("Finished restoring process pid %d\n",
+			p->lead_thread->pid);
+	else
+		pr_err("Failed to restore queues of process pid %d\n",
+		      p->lead_thread->pid);
+
+	return ret;
+}
+
+static void restore_process_worker(struct work_struct *work)
+{
+	struct delayed_work *dwork;
+	struct kfd_process *p;
+	int ret = 0;
+
+	dwork = to_delayed_work(work);
+
+	/* Process termination destroys this worker thread. So during the
+	 * lifetime of this thread, kfd_process p will be valid
+	 */
+	p = container_of(dwork, struct kfd_process, restore_work);
+	pr_debug("Started restoring process pasid %d\n", (int)p->lead_thread->pid);
+
+	/* Setting last_restore_timestamp before successful restoration.
+	 * Otherwise this would have to be set by KGD (restore_process_bos)
+	 * before KFD BOs are unreserved. If not, the process can be evicted
+	 * again before the timestamp is set.
+	 * If restore fails, the timestamp will be set again in the next
+	 * attempt. This would mean that the minimum GPU quanta would be
+	 * PROCESS_ACTIVE_TIME_MS - (time to execute the following two
+	 * functions)
+	 */
+
+	p->last_restore_timestamp = get_jiffies_64();
+
+	ret = restore_process_helper(p);
+	if (ret) {
+		pr_debug("Failed to restore BOs of process pid %d, retry after %d ms\n",
+			 p->lead_thread->pid, PROCESS_BACK_OFF_TIME_MS);
+		if (mod_delayed_work(kfd_restore_wq, &p->restore_work,
+				     msecs_to_jiffies(PROCESS_RESTORE_TIME_MS)))
+			kfd_process_restore_queues(p);
+	}
+}
+
+void kfd_suspend_all_processes(void)
+{
+	struct kfd_process *p;
+	unsigned int temp;
+	int idx = srcu_read_lock(&kfd_processes_srcu);
+
+	WARN(debug_evictions, "Evicting all processes");
+	hash_for_each_rcu(kfd_processes_table, temp, p, kfd_processes) {
+		if (kfd_process_evict_queues(p, KFD_QUEUE_EVICTION_TRIGGER_SUSPEND))
+			pr_err("Failed to suspend process pid %d\n", p->lead_thread->pid);
+		signal_eviction_fence(p);
+	}
+	srcu_read_unlock(&kfd_processes_srcu, idx);
+}
+
+int kfd_resume_all_processes(void)
+{
+	struct kfd_process *p;
+	unsigned int temp;
+	int ret = 0, idx = srcu_read_lock(&kfd_processes_srcu);
+
+	hash_for_each_rcu(kfd_processes_table, temp, p, kfd_processes) {
+		if (restore_process_helper(p)) {
+			pr_err("Restore process pid %d failed during resume\n",
+			      p->lead_thread->pid);
+			ret = -EFAULT;
+		}
+	}
+	srcu_read_unlock(&kfd_processes_srcu, idx);
+	return ret;
+}
+
+int kfd_reserved_mem_mmap(struct kfd_node *dev, struct kfd_process *process,
+			  struct vm_area_struct *vma)
+{
 	struct kfd_process_device *pdd;
+	struct qcm_process_device *qpd;
+
+	if ((vma->vm_end - vma->vm_start) != KFD_CWSR_TBA_TMA_SIZE) {
+		dev_err(dev->adev->dev, "Incorrect CWSR mapping size.\n");
+		return -EINVAL;
+	}
+
+	pdd = kfd_get_process_device_data(dev, process);
+	if (!pdd)
+		return -EINVAL;
+	qpd = &pdd->qpd;
+
+	qpd->cwsr_kaddr = (void *)__get_free_pages(GFP_KERNEL | __GFP_ZERO,
+					get_order(KFD_CWSR_TBA_TMA_SIZE));
+	if (!qpd->cwsr_kaddr) {
+		dev_err(dev->adev->dev,
+			"Error allocating per process CWSR buffer.\n");
+		return -ENOMEM;
+	}
+
+	vm_flags_set(vma, VM_IO | VM_DONTCOPY | VM_DONTEXPAND
+		| VM_NORESERVE | VM_DONTDUMP | VM_PFNMAP);
+	/* Mapping pages to user process */
+	return remap_pfn_range(vma, vma->vm_start,
+			       PFN_DOWN(__pa(qpd->cwsr_kaddr)),
+			       KFD_CWSR_TBA_TMA_SIZE, vma->vm_page_prot);
+}
+
+/* assumes caller holds process lock. */
+int kfd_process_drain_interrupts(struct kfd_process_device *pdd)
+{
+	uint32_t irq_drain_fence[8];
+	uint8_t node_id = 0;
+	int r = 0;
+
+	if (!KFD_IS_SOC15(pdd->dev))
+		return 0;
+
+	pdd->process->irq_drain_is_open = true;
+
+	memset(irq_drain_fence, 0, sizeof(irq_drain_fence));
+	irq_drain_fence[0] = (KFD_IRQ_FENCE_SOURCEID << 8) |
+							KFD_IRQ_FENCE_CLIENTID;
+	irq_drain_fence[3] = pdd->pasid;
 
 	/*
-	 * Look for the process that matches the pasid. If there is no such
-	 * process, we either released it in amdkfd's own notifier, or there
-	 * is a bug. Unfortunately, there is no way to tell...
+	 * For GFX 9.4.3/9.5.0, send the NodeId also in IH cookie DW[3]
 	 */
-	p = kfd_lookup_process_by_pasid(pasid);
+	if (KFD_GC_VERSION(pdd->dev->kfd) == IP_VERSION(9, 4, 3) ||
+	    KFD_GC_VERSION(pdd->dev->kfd) == IP_VERSION(9, 4, 4) ||
+	    KFD_GC_VERSION(pdd->dev->kfd) == IP_VERSION(9, 5, 0)) {
+		node_id = ffs(pdd->dev->interrupt_bitmap) - 1;
+		irq_drain_fence[3] |= node_id << 16;
+	}
+
+	/* ensure stale irqs scheduled KFD interrupts and send drain fence. */
+	if (amdgpu_amdkfd_send_close_event_drain_irq(pdd->dev->adev,
+						     irq_drain_fence)) {
+		pdd->process->irq_drain_is_open = false;
+		return 0;
+	}
+
+	r = wait_event_interruptible(pdd->process->wait_irq_drain,
+				     !READ_ONCE(pdd->process->irq_drain_is_open));
+	if (r)
+		pdd->process->irq_drain_is_open = false;
+
+	return r;
+}
+
+void kfd_process_close_interrupt_drain(unsigned int pasid)
+{
+	struct kfd_process *p;
+
+	p = kfd_lookup_process_by_pasid(pasid, NULL);
+
 	if (!p)
 		return;
 
-	pr_debug("Unbinding process %d from IOMMU\n", pasid);
+	WRITE_ONCE(p->irq_drain_is_open, false);
+	wake_up_all(&p->wait_irq_drain);
+	kfd_unref_process(p);
+}
 
-	if ((dev->dbgmgr) && (dev->dbgmgr->pasid == p->pasid))
-		kfd_dbgmgr_destroy(dev->dbgmgr);
+struct send_exception_work_handler_workarea {
+	struct work_struct work;
+	struct kfd_process *p;
+	unsigned int queue_id;
+	uint64_t error_reason;
+};
 
-	pqm_uninit(&p->pqm);
+static void send_exception_work_handler(struct work_struct *work)
+{
+	struct send_exception_work_handler_workarea *workarea;
+	struct kfd_process *p;
+	struct queue *q;
+	struct mm_struct *mm;
+	struct kfd_context_save_area_header __user *csa_header;
+	uint64_t __user *err_payload_ptr;
+	uint64_t cur_err;
+	uint32_t ev_id;
 
-	pdd = kfd_get_process_device_data(dev, p);
+	workarea = container_of(work,
+				struct send_exception_work_handler_workarea,
+				work);
+	p = workarea->p;
 
-	if (!pdd) {
-		mutex_unlock(&p->mutex);
+	mm = get_task_mm(p->lead_thread);
+
+	if (!mm)
 		return;
-	}
 
-	if (pdd->reset_wavefronts) {
-		dbgdev_wave_reset_wavefronts(pdd->dev, p);
-		pdd->reset_wavefronts = false;
-	}
+	kthread_use_mm(mm);
 
-	/*
-	 * Just mark pdd as unbound, because we still need it
-	 * to call amd_iommu_unbind_pasid() in when the
-	 * process exits.
-	 * We don't call amd_iommu_unbind_pasid() here
-	 * because the IOMMU called us.
-	 */
-	pdd->bound = false;
+	q = pqm_get_user_queue(&p->pqm, workarea->queue_id);
 
-	mutex_unlock(&p->mutex);
+	if (!q)
+		goto out;
+
+	csa_header = (void __user *)q->properties.ctx_save_restore_area_address;
+
+	get_user(err_payload_ptr, (uint64_t __user **)&csa_header->err_payload_addr);
+	get_user(cur_err, err_payload_ptr);
+	cur_err |= workarea->error_reason;
+	put_user(cur_err, err_payload_ptr);
+	get_user(ev_id, &csa_header->err_event_id);
+
+	kfd_set_event(p, ev_id);
+
+out:
+	kthread_unuse_mm(mm);
+	mmput(mm);
 }
 
-struct kfd_process_device *kfd_get_first_process_device_data(
-						struct kfd_process *p)
+int kfd_send_exception_to_runtime(struct kfd_process *p,
+			unsigned int queue_id,
+			uint64_t error_reason)
 {
-	return list_first_entry(&p->per_device_data,
-				struct kfd_process_device,
-				per_device_list);
+	struct send_exception_work_handler_workarea worker;
+
+	INIT_WORK_ONSTACK(&worker.work, send_exception_work_handler);
+
+	worker.p = p;
+	worker.queue_id = queue_id;
+	worker.error_reason = error_reason;
+
+	schedule_work(&worker.work);
+	flush_work(&worker.work);
+	destroy_work_on_stack(&worker.work);
+
+	return 0;
 }
 
-struct kfd_process_device *kfd_get_next_process_device_data(
-						struct kfd_process *p,
-						struct kfd_process_device *pdd)
+struct kfd_process_device *kfd_process_device_data_by_id(struct kfd_process *p, uint32_t gpu_id)
 {
-	if (list_is_last(&pdd->per_device_list, &p->per_device_data))
-		return NULL;
-	return list_next_entry(pdd, per_device_list);
+	int i;
+
+	if (gpu_id) {
+		for (i = 0; i < p->n_pdds; i++) {
+			struct kfd_process_device *pdd = p->pdds[i];
+
+			if (pdd->user_gpu_id == gpu_id)
+				return pdd;
+		}
+	}
+	return NULL;
 }
 
-bool kfd_has_process_device_data(struct kfd_process *p)
+int kfd_process_get_user_gpu_id(struct kfd_process *p, uint32_t actual_gpu_id)
 {
-	return !(list_empty(&p->per_device_data));
+	int i;
+
+	if (!actual_gpu_id)
+		return 0;
+
+	for (i = 0; i < p->n_pdds; i++) {
+		struct kfd_process_device *pdd = p->pdds[i];
+
+		if (pdd->dev->id == actual_gpu_id)
+			return pdd->user_gpu_id;
+	}
+	return -EINVAL;
 }
 
-/* This returns with process->mutex locked. */
-struct kfd_process *kfd_lookup_process_by_pasid(unsigned int pasid)
+#if defined(CONFIG_DEBUG_FS)
+
+int kfd_debugfs_mqds_by_process(struct seq_file *m, void *data)
 {
 	struct kfd_process *p;
 	unsigned int temp;
+	int r = 0;
 
 	int idx = srcu_read_lock(&kfd_processes_srcu);
 
 	hash_for_each_rcu(kfd_processes_table, temp, p, kfd_processes) {
-		if (p->pasid == pasid) {
-			mutex_lock(&p->mutex);
+		seq_printf(m, "Process %d PASID %d:\n",
+			   p->lead_thread->tgid, p->lead_thread->pid);
+
+		mutex_lock(&p->mutex);
+		r = pqm_debugfs_mqds(m, &p->pqm);
+		mutex_unlock(&p->mutex);
+
+		if (r)
 			break;
-		}
 	}
 
 	srcu_read_unlock(&kfd_processes_srcu, idx);
 
-	return p;
+	return r;
 }
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c
index 1cae95e2b13a..7fbb5c274ccc 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c
@@ -1,5 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -26,6 +27,8 @@
 #include "kfd_device_queue_manager.h"
 #include "kfd_priv.h"
 #include "kfd_kernel_queue.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_reset.h"
 
 static inline struct process_queue_node *get_queue_by_qid(
 			struct process_queue_manager *pqm, unsigned int qid)
@@ -41,6 +44,20 @@ static inline struct process_queue_node *get_queue_by_qid(
 	return NULL;
 }
 
+static int assign_queue_slot_by_qid(struct process_queue_manager *pqm,
+				    unsigned int qid)
+{
+	if (qid >= KFD_MAX_NUM_OF_QUEUES_PER_PROCESS)
+		return -EINVAL;
+
+	if (__test_and_set_bit(qid, pqm->queue_slot_bitmap)) {
+		pr_err("Cannot create new queue because requested qid(%u) is in use\n", qid);
+		return -ENOSPC;
+	}
+
+	return 0;
+}
+
 static int find_available_queue_slot(struct process_queue_manager *pqm,
 					unsigned int *qid)
 {
@@ -52,8 +69,8 @@ static int find_available_queue_slot(struct process_queue_manager *pqm,
 	pr_debug("The new slot id %lu\n", found);
 
 	if (found >= KFD_MAX_NUM_OF_QUEUES_PER_PROCESS) {
-		pr_info("Cannot open more queues for process with pasid %d\n",
-				pqm->process->pasid);
+		pr_info("Cannot open more queues for process with pid %d\n",
+			pqm->process->lead_thread->pid);
 		return -ENOMEM;
 	}
 
@@ -63,12 +80,103 @@ static int find_available_queue_slot(struct process_queue_manager *pqm,
 	return 0;
 }
 
+void kfd_process_dequeue_from_device(struct kfd_process_device *pdd)
+{
+	struct kfd_node *dev = pdd->dev;
+
+	if (pdd->already_dequeued)
+		return;
+	/* The MES context flush needs to filter out the case which the
+	 * KFD process is created without setting up the MES context and
+	 * queue for creating a compute queue.
+	 */
+	dev->dqm->ops.process_termination(dev->dqm, &pdd->qpd);
+	if (dev->kfd->shared_resources.enable_mes && !!pdd->proc_ctx_gpu_addr &&
+	    down_read_trylock(&dev->adev->reset_domain->sem)) {
+		amdgpu_mes_flush_shader_debugger(dev->adev,
+						 pdd->proc_ctx_gpu_addr);
+		up_read(&dev->adev->reset_domain->sem);
+	}
+	pdd->already_dequeued = true;
+}
+
+int pqm_set_gws(struct process_queue_manager *pqm, unsigned int qid,
+			void *gws)
+{
+	struct mqd_update_info minfo = {0};
+	struct kfd_node *dev = NULL;
+	struct process_queue_node *pqn;
+	struct kfd_process_device *pdd;
+	struct kgd_mem *mem = NULL;
+	int ret;
+
+	pqn = get_queue_by_qid(pqm, qid);
+	if (!pqn) {
+		pr_err("Queue id does not match any known queue\n");
+		return -EINVAL;
+	}
+
+	if (pqn->q)
+		dev = pqn->q->device;
+	if (WARN_ON(!dev))
+		return -ENODEV;
+
+	pdd = kfd_get_process_device_data(dev, pqm->process);
+	if (!pdd) {
+		pr_err("Process device data doesn't exist\n");
+		return -EINVAL;
+	}
+
+	/* Only allow one queue per process can have GWS assigned */
+	if (gws && pdd->qpd.num_gws)
+		return -EBUSY;
+
+	if (!gws && pdd->qpd.num_gws == 0)
+		return -EINVAL;
+
+	if ((KFD_GC_VERSION(dev) != IP_VERSION(9, 4, 3) &&
+	     KFD_GC_VERSION(dev) != IP_VERSION(9, 4, 4) &&
+	     KFD_GC_VERSION(dev) != IP_VERSION(9, 5, 0)) &&
+	    !dev->kfd->shared_resources.enable_mes) {
+		if (gws)
+			ret = amdgpu_amdkfd_add_gws_to_process(pdd->process->kgd_process_info,
+				gws, &mem);
+		else
+			ret = amdgpu_amdkfd_remove_gws_from_process(pdd->process->kgd_process_info,
+				pqn->q->gws);
+		if (unlikely(ret))
+			return ret;
+		pqn->q->gws = mem;
+	} else {
+		/*
+		 * Intentionally set GWS to a non-NULL value
+		 * for devices that do not use GWS for global wave
+		 * synchronization but require the formality
+		 * of setting GWS for cooperative groups.
+		 */
+		pqn->q->gws = gws ? ERR_PTR(-ENOMEM) : NULL;
+	}
+
+	pdd->qpd.num_gws = gws ? dev->adev->gds.gws_size : 0;
+	minfo.update_flag = gws ? UPDATE_FLAG_IS_GWS : 0;
+
+	return pqn->q->device->dqm->ops.update_queue(pqn->q->device->dqm,
+							pqn->q, &minfo);
+}
+
+void kfd_process_dequeue_from_all_devices(struct kfd_process *p)
+{
+	int i;
+
+	for (i = 0; i < p->n_pdds; i++)
+		kfd_process_dequeue_from_device(p->pdds[i]);
+}
+
 int pqm_init(struct process_queue_manager *pqm, struct kfd_process *p)
 {
 	INIT_LIST_HEAD(&pqm->queues);
-	pqm->queue_slot_bitmap =
-			kzalloc(DIV_ROUND_UP(KFD_MAX_NUM_OF_QUEUES_PER_PROCESS,
-					BITS_PER_BYTE), GFP_KERNEL);
+	pqm->queue_slot_bitmap = bitmap_zalloc(KFD_MAX_NUM_OF_QUEUES_PER_PROCESS,
+					       GFP_KERNEL);
 	if (!pqm->queue_slot_bitmap)
 		return -ENOMEM;
 	pqm->process = p;
@@ -76,39 +184,73 @@ int pqm_init(struct process_queue_manager *pqm, struct kfd_process *p)
 	return 0;
 }
 
+static void pqm_clean_queue_resource(struct process_queue_manager *pqm,
+				     struct process_queue_node *pqn)
+{
+	struct kfd_node *dev;
+	struct kfd_process_device *pdd;
+
+	dev = pqn->q->device;
+
+	pdd = kfd_get_process_device_data(dev, pqm->process);
+	if (!pdd) {
+		pr_err("Process device data doesn't exist\n");
+		return;
+	}
+
+	if (pqn->q->gws) {
+		if (KFD_GC_VERSION(pqn->q->device) != IP_VERSION(9, 4, 3) &&
+		    KFD_GC_VERSION(pqn->q->device) != IP_VERSION(9, 4, 4) &&
+		    KFD_GC_VERSION(pqn->q->device) != IP_VERSION(9, 5, 0) &&
+		    !dev->kfd->shared_resources.enable_mes)
+			amdgpu_amdkfd_remove_gws_from_process(
+				pqm->process->kgd_process_info, pqn->q->gws);
+		pdd->qpd.num_gws = 0;
+	}
+
+	if (dev->kfd->shared_resources.enable_mes) {
+		amdgpu_amdkfd_free_gtt_mem(dev->adev, &pqn->q->gang_ctx_bo);
+		amdgpu_amdkfd_free_gtt_mem(dev->adev, (void **)&pqn->q->wptr_bo_gart);
+	}
+}
+
 void pqm_uninit(struct process_queue_manager *pqm)
 {
-	int retval;
 	struct process_queue_node *pqn, *next;
 
 	list_for_each_entry_safe(pqn, next, &pqm->queues, process_queue_list) {
-		retval = pqm_destroy_queue(
-				pqm,
-				(pqn->q != NULL) ?
-					pqn->q->properties.queue_id :
-					pqn->kq->queue->properties.queue_id);
-
-		if (retval != 0) {
-			pr_err("failed to destroy queue\n");
-			return;
+		if (pqn->q) {
+			struct kfd_process_device *pdd = kfd_get_process_device_data(pqn->q->device,
+										     pqm->process);
+			if (pdd) {
+				kfd_queue_unref_bo_vas(pdd, &pqn->q->properties);
+				kfd_queue_release_buffers(pdd, &pqn->q->properties);
+			} else {
+				WARN_ON(!pdd);
+			}
+			pqm_clean_queue_resource(pqm, pqn);
 		}
+
+		kfd_procfs_del_queue(pqn->q);
+		uninit_queue(pqn->q);
+		list_del(&pqn->process_queue_list);
+		kfree(pqn);
 	}
-	kfree(pqm->queue_slot_bitmap);
+
+	bitmap_free(pqm->queue_slot_bitmap);
 	pqm->queue_slot_bitmap = NULL;
 }
 
-static int create_cp_queue(struct process_queue_manager *pqm,
-				struct kfd_dev *dev, struct queue **q,
+static int init_user_queue(struct process_queue_manager *pqm,
+				struct kfd_node *dev, struct queue **q,
 				struct queue_properties *q_properties,
-				struct file *f, unsigned int qid)
+				unsigned int qid)
 {
 	int retval;
 
 	/* Doorbell initialized in user space*/
 	q_properties->doorbell_ptr = NULL;
-
-	q_properties->doorbell_off =
-			kfd_queue_id_to_doorbell(dev, pqm->process, qid);
+	q_properties->exception_status = KFD_EC_MASK(EC_QUEUE_NEW);
 
 	/* let DQM handle it*/
 	q_properties->vmid = 0;
@@ -121,30 +263,73 @@ static int create_cp_queue(struct process_queue_manager *pqm,
 	(*q)->device = dev;
 	(*q)->process = pqm->process;
 
+	if (dev->kfd->shared_resources.enable_mes) {
+		retval = amdgpu_amdkfd_alloc_gtt_mem(dev->adev,
+						AMDGPU_MES_GANG_CTX_SIZE,
+						&(*q)->gang_ctx_bo,
+						&(*q)->gang_ctx_gpu_addr,
+						&(*q)->gang_ctx_cpu_ptr,
+						false);
+		if (retval) {
+			pr_err("failed to allocate gang context bo\n");
+			goto cleanup;
+		}
+		memset((*q)->gang_ctx_cpu_ptr, 0, AMDGPU_MES_GANG_CTX_SIZE);
+
+		/* Starting with GFX11, wptr BOs must be mapped to GART for MES to determine work
+		 * on unmapped queues for usermode queue oversubscription (no aggregated doorbell)
+		 */
+		if (dev->adev != amdgpu_ttm_adev(q_properties->wptr_bo->tbo.bdev)) {
+			pr_err("Queue memory allocated to wrong device\n");
+			retval = -EINVAL;
+			goto free_gang_ctx_bo;
+		}
+
+		retval = amdgpu_amdkfd_map_gtt_bo_to_gart(q_properties->wptr_bo,
+							  &(*q)->wptr_bo_gart);
+		if (retval) {
+			pr_err("Failed to map wptr bo to GART\n");
+			goto free_gang_ctx_bo;
+		}
+	}
+
 	pr_debug("PQM After init queue");
+	return 0;
 
+free_gang_ctx_bo:
+	amdgpu_amdkfd_free_gtt_mem(dev->adev, &(*q)->gang_ctx_bo);
+cleanup:
+	uninit_queue(*q);
+	*q = NULL;
 	return retval;
 }
 
 int pqm_create_queue(struct process_queue_manager *pqm,
-			    struct kfd_dev *dev,
-			    struct file *f,
+			    struct kfd_node *dev,
 			    struct queue_properties *properties,
-			    unsigned int flags,
-			    enum kfd_queue_type type,
-			    unsigned int *qid)
+			    unsigned int *qid,
+			    const struct kfd_criu_queue_priv_data *q_data,
+			    const void *restore_mqd,
+			    const void *restore_ctl_stack,
+			    uint32_t *p_doorbell_offset_in_process)
 {
 	int retval;
 	struct kfd_process_device *pdd;
-	struct queue_properties q_properties;
 	struct queue *q;
 	struct process_queue_node *pqn;
 	struct kernel_queue *kq;
-	int num_queues = 0;
-	struct queue *cur;
+	enum kfd_queue_type type = properties->type;
+	unsigned int max_queues = 127; /* HWS limit */
+
+	/*
+	 * On GFX 9.4.3/9.5.0, increase the number of queues that
+	 * can be created to 255. No HWS limit on GFX 9.4.3/9.5.0.
+	 */
+	if (KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 3) ||
+	    KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 4) ||
+	    KFD_GC_VERSION(dev) == IP_VERSION(9, 5, 0))
+		max_queues = 255;
 
-	memset(&q_properties, 0, sizeof(struct queue_properties));
-	memcpy(&q_properties, properties, sizeof(struct queue_properties));
 	q = NULL;
 	kq = NULL;
 
@@ -160,21 +345,39 @@ int pqm_create_queue(struct process_queue_manager *pqm,
 	 * If we are just about to create DIQ, the is_debug flag is not set yet
 	 * Hence we also check the type as well
 	 */
-	if ((pdd->qpd.is_debug) ||
-		(type == KFD_QUEUE_TYPE_DIQ)) {
-		list_for_each_entry(cur, &pdd->qpd.queues_list, list)
-			num_queues++;
-		if (num_queues >= dev->device_info->max_no_of_hqd/2)
-			return -ENOSPC;
-	}
+	if ((pdd->qpd.is_debug) || (type == KFD_QUEUE_TYPE_DIQ))
+		max_queues = dev->kfd->device_info.max_no_of_hqd/2;
+
+	if (pdd->qpd.queue_count >= max_queues)
+		return -ENOSPC;
+
+	if (q_data) {
+		retval = assign_queue_slot_by_qid(pqm, q_data->q_id);
+		*qid = q_data->q_id;
+	} else
+		retval = find_available_queue_slot(pqm, qid);
 
-	retval = find_available_queue_slot(pqm, qid);
 	if (retval != 0)
 		return retval;
 
-	if (list_empty(&pqm->queues)) {
-		pdd->qpd.pqm = pqm;
+	/* Register process if this is the first queue */
+	if (list_empty(&pdd->qpd.queues_list) &&
+	    list_empty(&pdd->qpd.priv_queue_list))
 		dev->dqm->ops.register_process(dev->dqm, &pdd->qpd);
+
+	/* Allocate proc_ctx_bo only if MES is enabled and this is the first queue */
+	if (!pdd->proc_ctx_cpu_ptr && dev->kfd->shared_resources.enable_mes) {
+		retval = amdgpu_amdkfd_alloc_gtt_mem(dev->adev,
+						     AMDGPU_MES_PROC_CTX_SIZE,
+						     &pdd->proc_ctx_bo,
+						     &pdd->proc_ctx_gpu_addr,
+						     &pdd->proc_ctx_cpu_ptr,
+						     false);
+		if (retval) {
+			dev_err(dev->adev->dev, "failed to allocate process context bo\n");
+			return retval;
+		}
+		memset(pdd->proc_ctx_cpu_ptr, 0, AMDGPU_MES_PROC_CTX_SIZE);
 	}
 
 	pqn = kzalloc(sizeof(*pqn), GFP_KERNEL);
@@ -185,24 +388,42 @@ int pqm_create_queue(struct process_queue_manager *pqm,
 
 	switch (type) {
 	case KFD_QUEUE_TYPE_SDMA:
+	case KFD_QUEUE_TYPE_SDMA_XGMI:
+	case KFD_QUEUE_TYPE_SDMA_BY_ENG_ID:
+		/* SDMA queues are always allocated statically no matter
+		 * which scheduler mode is used. We also do not need to
+		 * check whether a SDMA queue can be allocated here, because
+		 * allocate_sdma_queue() in create_queue() has the
+		 * corresponding check logic.
+		 */
+		retval = init_user_queue(pqm, dev, &q, properties, *qid);
+		if (retval != 0)
+			goto err_create_queue;
+		pqn->q = q;
+		pqn->kq = NULL;
+		retval = dev->dqm->ops.create_queue(dev->dqm, q, &pdd->qpd, q_data,
+						    restore_mqd, restore_ctl_stack);
+		print_queue(q);
+		break;
+
 	case KFD_QUEUE_TYPE_COMPUTE:
 		/* check if there is over subscription */
-		if ((sched_policy == KFD_SCHED_POLICY_HWS_NO_OVERSUBSCRIPTION) &&
-		((dev->dqm->processes_count >= VMID_PER_DEVICE) ||
-		(dev->dqm->queue_count >= get_queues_num(dev->dqm)))) {
-			pr_err("Over-subscription is not allowed in radeon_kfd.sched_policy == 1\n");
+		if ((dev->dqm->sched_policy ==
+		     KFD_SCHED_POLICY_HWS_NO_OVERSUBSCRIPTION) &&
+		((dev->dqm->processes_count >= dev->vm_info.vmid_num_kfd) ||
+		(dev->dqm->active_queue_count >= get_cp_queues_num(dev->dqm)))) {
+			pr_debug("Over-subscription is not allowed when amdkfd.sched_policy == 1\n");
 			retval = -EPERM;
 			goto err_create_queue;
 		}
 
-		retval = create_cp_queue(pqm, dev, &q, &q_properties, f, *qid);
+		retval = init_user_queue(pqm, dev, &q, properties, *qid);
 		if (retval != 0)
 			goto err_create_queue;
 		pqn->q = q;
 		pqn->kq = NULL;
-		retval = dev->dqm->ops.create_queue(dev->dqm, q, &pdd->qpd,
-						&q->properties.vmid);
-		pr_debug("DQM returned %d for create_queue\n", retval);
+		retval = dev->dqm->ops.create_queue(dev->dqm, q, &pdd->qpd, q_data,
+						    restore_mqd, restore_ctl_stack);
 		print_queue(q);
 		break;
 	case KFD_QUEUE_TYPE_DIQ:
@@ -214,6 +435,10 @@ int pqm_create_queue(struct process_queue_manager *pqm,
 		kq->queue->properties.queue_id = *qid;
 		pqn->kq = kq;
 		pqn->q = NULL;
+		retval = kfd_process_drain_interrupts(pdd);
+		if (retval)
+			break;
+
 		retval = dev->dqm->ops.create_kernel_queue(dev->dqm,
 							kq, &pdd->qpd);
 		break;
@@ -223,28 +448,56 @@ int pqm_create_queue(struct process_queue_manager *pqm,
 	}
 
 	if (retval != 0) {
-		pr_err("DQM create queue failed\n");
+		if ((type == KFD_QUEUE_TYPE_SDMA ||
+		    type == KFD_QUEUE_TYPE_SDMA_XGMI ||
+		    type == KFD_QUEUE_TYPE_SDMA_BY_ENG_ID) &&
+		    retval == -ENOMEM)
+			pr_warn("process pid %d DQM create queue type %d failed. ret %d\n",
+				pqm->process->lead_thread->pid, type, retval);
+		else
+			pr_err("process pid %d DQM create queue type %d failed. ret %d\n",
+				pqm->process->lead_thread->pid, type, retval);
 		goto err_create_queue;
 	}
 
+	if (q && p_doorbell_offset_in_process) {
+		/* Return the doorbell offset within the doorbell page
+		 * to the caller so it can be passed up to user mode
+		 * (in bytes).
+		 * relative doorbell index = Absolute doorbell index -
+		 * absolute index of first doorbell in the page.
+		 */
+		uint32_t first_db_index = amdgpu_doorbell_index_on_bar(pdd->dev->adev,
+								       pdd->qpd.proc_doorbells,
+								       0,
+								       pdd->dev->kfd->device_info.doorbell_size);
+
+		*p_doorbell_offset_in_process = (q->properties.doorbell_off
+						- first_db_index) * sizeof(uint32_t);
+	}
+
 	pr_debug("PQM After DQM create queue\n");
 
 	list_add(&pqn->process_queue_list, &pqm->queues);
 
 	if (q) {
-		*properties = q->properties;
 		pr_debug("PQM done creating queue\n");
-		print_queue_properties(properties);
+		kfd_procfs_add_queue(q);
+		print_queue_properties(&q->properties);
 	}
 
 	return retval;
 
 err_create_queue:
+	uninit_queue(q);
+	if (kq)
+		kernel_queue_uninit(kq);
 	kfree(pqn);
 err_allocate_pqn:
 	/* check if queues list is empty unregister process from device */
 	clear_bit(*qid, pqm->queue_slot_bitmap);
-	if (list_empty(&pqm->queues))
+	if (list_empty(&pdd->qpd.queues_list) &&
+	    list_empty(&pdd->qpd.priv_queue_list))
 		dev->dqm->ops.unregister_process(dev->dqm, &pdd->qpd);
 	return retval;
 }
@@ -254,7 +507,7 @@ int pqm_destroy_queue(struct process_queue_manager *pqm, unsigned int qid)
 	struct process_queue_node *pqn;
 	struct kfd_process_device *pdd;
 	struct device_queue_manager *dqm;
-	struct kfd_dev *dev;
+	struct kfd_node *dev;
 	int retval;
 
 	dqm = NULL;
@@ -289,11 +542,22 @@ int pqm_destroy_queue(struct process_queue_manager *pqm, unsigned int qid)
 	}
 
 	if (pqn->q) {
+		retval = kfd_queue_unref_bo_vas(pdd, &pqn->q->properties);
+		if (retval)
+			goto err_destroy_queue;
+
 		dqm = pqn->q->device->dqm;
 		retval = dqm->ops.destroy_queue(dqm, &pdd->qpd, pqn->q);
-		if (retval != 0)
-			return retval;
-
+		if (retval) {
+			pr_err("Pasid 0x%x destroy queue %d failed, ret %d\n",
+				pdd->pasid,
+				pqn->q->properties.queue_id, retval);
+			if (retval != -ETIME && retval != -EIO)
+				goto err_destroy_queue;
+		}
+		kfd_procfs_del_queue(pqn->q);
+		kfd_queue_release_buffers(pdd, &pqn->q->properties);
+		pqm_clean_queue_resource(pqm, pqn);
 		uninit_queue(pqn->q);
 	}
 
@@ -301,48 +565,609 @@ int pqm_destroy_queue(struct process_queue_manager *pqm, unsigned int qid)
 	kfree(pqn);
 	clear_bit(qid, pqm->queue_slot_bitmap);
 
-	if (list_empty(&pqm->queues))
+	if (list_empty(&pdd->qpd.queues_list) &&
+	    list_empty(&pdd->qpd.priv_queue_list))
 		dqm->ops.unregister_process(dqm, &pdd->qpd);
 
+err_destroy_queue:
 	return retval;
 }
 
-int pqm_update_queue(struct process_queue_manager *pqm, unsigned int qid,
-			struct queue_properties *p)
+int pqm_update_queue_properties(struct process_queue_manager *pqm,
+				unsigned int qid, struct queue_properties *p)
 {
 	int retval;
 	struct process_queue_node *pqn;
 
 	pqn = get_queue_by_qid(pqm, qid);
-	if (!pqn) {
+	if (!pqn || !pqn->q) {
 		pr_debug("No queue %d exists for update operation\n", qid);
 		return -EFAULT;
 	}
 
+	/*
+	 * Update with NULL ring address is used to disable the queue
+	 */
+	if (p->queue_address && p->queue_size) {
+		struct kfd_process_device *pdd;
+		struct amdgpu_vm *vm;
+		struct queue *q = pqn->q;
+		int err;
+
+		pdd = kfd_get_process_device_data(q->device, q->process);
+		if (!pdd)
+			return -ENODEV;
+		vm = drm_priv_to_vm(pdd->drm_priv);
+		err = amdgpu_bo_reserve(vm->root.bo, false);
+		if (err)
+			return err;
+
+		if (kfd_queue_buffer_get(vm, (void *)p->queue_address, &p->ring_bo,
+					 p->queue_size)) {
+			pr_debug("ring buf 0x%llx size 0x%llx not mapped on GPU\n",
+				 p->queue_address, p->queue_size);
+			return -EFAULT;
+		}
+
+		kfd_queue_unref_bo_va(vm, &pqn->q->properties.ring_bo);
+		kfd_queue_buffer_put(&pqn->q->properties.ring_bo);
+		amdgpu_bo_unreserve(vm->root.bo);
+
+		pqn->q->properties.ring_bo = p->ring_bo;
+	}
+
 	pqn->q->properties.queue_address = p->queue_address;
 	pqn->q->properties.queue_size = p->queue_size;
 	pqn->q->properties.queue_percent = p->queue_percent;
 	pqn->q->properties.priority = p->priority;
+	pqn->q->properties.pm4_target_xcc = p->pm4_target_xcc;
 
 	retval = pqn->q->device->dqm->ops.update_queue(pqn->q->device->dqm,
-							pqn->q);
+							pqn->q, NULL);
 	if (retval != 0)
 		return retval;
 
 	return 0;
 }
 
-struct kernel_queue *pqm_get_kernel_queue(
-					struct process_queue_manager *pqm,
+int pqm_update_mqd(struct process_queue_manager *pqm,
+				unsigned int qid, struct mqd_update_info *minfo)
+{
+	int retval;
+	struct process_queue_node *pqn;
+
+	pqn = get_queue_by_qid(pqm, qid);
+	if (!pqn) {
+		pr_debug("No queue %d exists for update operation\n", qid);
+		return -EFAULT;
+	}
+
+	/* CUs are masked for debugger requirements so deny user mask  */
+	if (pqn->q->properties.is_dbg_wa && minfo && minfo->cu_mask.ptr)
+		return -EBUSY;
+
+	/* ASICs that have WGPs must enforce pairwise enabled mask checks. */
+	if (minfo && minfo->cu_mask.ptr &&
+			KFD_GC_VERSION(pqn->q->device) >= IP_VERSION(10, 0, 0)) {
+		int i;
+
+		for (i = 0; i < minfo->cu_mask.count; i += 2) {
+			uint32_t cu_pair = (minfo->cu_mask.ptr[i / 32] >> (i % 32)) & 0x3;
+
+			if (cu_pair && cu_pair != 0x3) {
+				pr_debug("CUs must be adjacent pairwise enabled.\n");
+				return -EINVAL;
+			}
+		}
+	}
+
+	retval = pqn->q->device->dqm->ops.update_queue(pqn->q->device->dqm,
+							pqn->q, minfo);
+	if (retval != 0)
+		return retval;
+
+	if (minfo && minfo->cu_mask.ptr)
+		pqn->q->properties.is_user_cu_masked = true;
+
+	return 0;
+}
+
+struct queue *pqm_get_user_queue(struct process_queue_manager *pqm,
 					unsigned int qid)
 {
 	struct process_queue_node *pqn;
 
 	pqn = get_queue_by_qid(pqm, qid);
-	if (pqn && pqn->kq)
-		return pqn->kq;
+	return pqn ? pqn->q : NULL;
+}
 
-	return NULL;
+int pqm_get_wave_state(struct process_queue_manager *pqm,
+		       unsigned int qid,
+		       void __user *ctl_stack,
+		       u32 *ctl_stack_used_size,
+		       u32 *save_area_used_size)
+{
+	struct process_queue_node *pqn;
+
+	pqn = get_queue_by_qid(pqm, qid);
+	if (!pqn) {
+		pr_debug("amdkfd: No queue %d exists for operation\n",
+			 qid);
+		return -EFAULT;
+	}
+
+	return pqn->q->device->dqm->ops.get_wave_state(pqn->q->device->dqm,
+						       pqn->q,
+						       ctl_stack,
+						       ctl_stack_used_size,
+						       save_area_used_size);
+}
+
+int pqm_get_queue_snapshot(struct process_queue_manager *pqm,
+			   uint64_t exception_clear_mask,
+			   void __user *buf,
+			   int *num_qss_entries,
+			   uint32_t *entry_size)
+{
+	struct process_queue_node *pqn;
+	struct kfd_queue_snapshot_entry src;
+	uint32_t tmp_entry_size = *entry_size, tmp_qss_entries = *num_qss_entries;
+	int r = 0;
+
+	*num_qss_entries = 0;
+	if (!(*entry_size))
+		return -EINVAL;
+
+	*entry_size = min_t(size_t, *entry_size, sizeof(struct kfd_queue_snapshot_entry));
+	mutex_lock(&pqm->process->event_mutex);
+
+	memset(&src, 0, sizeof(src));
+
+	list_for_each_entry(pqn, &pqm->queues, process_queue_list) {
+		if (!pqn->q)
+			continue;
+
+		if (*num_qss_entries < tmp_qss_entries) {
+			set_queue_snapshot_entry(pqn->q, exception_clear_mask, &src);
+
+			if (copy_to_user(buf, &src, *entry_size)) {
+				r = -EFAULT;
+				break;
+			}
+			buf += tmp_entry_size;
+		}
+		*num_qss_entries += 1;
+	}
+
+	mutex_unlock(&pqm->process->event_mutex);
+	return r;
+}
+
+static int get_queue_data_sizes(struct kfd_process_device *pdd,
+				struct queue *q,
+				uint32_t *mqd_size,
+				uint32_t *ctl_stack_size)
+{
+	int ret;
+
+	ret = pqm_get_queue_checkpoint_info(&pdd->process->pqm,
+					    q->properties.queue_id,
+					    mqd_size,
+					    ctl_stack_size);
+	if (ret)
+		pr_err("Failed to get queue dump info (%d)\n", ret);
+
+	return ret;
+}
+
+int kfd_process_get_queue_info(struct kfd_process *p,
+			       uint32_t *num_queues,
+			       uint64_t *priv_data_sizes)
+{
+	uint32_t extra_data_sizes = 0;
+	struct queue *q;
+	int i;
+	int ret;
+
+	*num_queues = 0;
+
+	/* Run over all PDDs of the process */
+	for (i = 0; i < p->n_pdds; i++) {
+		struct kfd_process_device *pdd = p->pdds[i];
+
+		list_for_each_entry(q, &pdd->qpd.queues_list, list) {
+			if (q->properties.type == KFD_QUEUE_TYPE_COMPUTE ||
+				q->properties.type == KFD_QUEUE_TYPE_SDMA ||
+				q->properties.type == KFD_QUEUE_TYPE_SDMA_XGMI) {
+				uint32_t mqd_size, ctl_stack_size;
+
+				*num_queues = *num_queues + 1;
+
+				ret = get_queue_data_sizes(pdd, q, &mqd_size, &ctl_stack_size);
+				if (ret)
+					return ret;
+
+				extra_data_sizes += mqd_size + ctl_stack_size;
+			} else {
+				pr_err("Unsupported queue type (%d)\n", q->properties.type);
+				return -EOPNOTSUPP;
+			}
+		}
+	}
+	*priv_data_sizes = extra_data_sizes +
+				(*num_queues * sizeof(struct kfd_criu_queue_priv_data));
+
+	return 0;
+}
+
+static int pqm_checkpoint_mqd(struct process_queue_manager *pqm,
+			      unsigned int qid,
+			      void *mqd,
+			      void *ctl_stack)
+{
+	struct process_queue_node *pqn;
+
+	pqn = get_queue_by_qid(pqm, qid);
+	if (!pqn) {
+		pr_debug("amdkfd: No queue %d exists for operation\n", qid);
+		return -EFAULT;
+	}
+
+	if (!pqn->q->device->dqm->ops.checkpoint_mqd) {
+		pr_err("amdkfd: queue dumping not supported on this device\n");
+		return -EOPNOTSUPP;
+	}
+
+	return pqn->q->device->dqm->ops.checkpoint_mqd(pqn->q->device->dqm,
+						       pqn->q, mqd, ctl_stack);
+}
+
+static int criu_checkpoint_queue(struct kfd_process_device *pdd,
+			   struct queue *q,
+			   struct kfd_criu_queue_priv_data *q_data)
+{
+	uint8_t *mqd, *ctl_stack;
+	int ret;
+
+	mqd = (void *)(q_data + 1);
+	ctl_stack = mqd + q_data->mqd_size;
+
+	q_data->gpu_id = pdd->user_gpu_id;
+	q_data->type = q->properties.type;
+	q_data->format = q->properties.format;
+	q_data->q_id =  q->properties.queue_id;
+	q_data->q_address = q->properties.queue_address;
+	q_data->q_size = q->properties.queue_size;
+	q_data->priority = q->properties.priority;
+	q_data->q_percent = q->properties.queue_percent;
+	q_data->read_ptr_addr = (uint64_t)q->properties.read_ptr;
+	q_data->write_ptr_addr = (uint64_t)q->properties.write_ptr;
+	q_data->doorbell_id = q->doorbell_id;
+
+	q_data->sdma_id = q->sdma_id;
+
+	q_data->eop_ring_buffer_address =
+		q->properties.eop_ring_buffer_address;
+
+	q_data->eop_ring_buffer_size = q->properties.eop_ring_buffer_size;
+
+	q_data->ctx_save_restore_area_address =
+		q->properties.ctx_save_restore_area_address;
+
+	q_data->ctx_save_restore_area_size =
+		q->properties.ctx_save_restore_area_size;
+
+	q_data->gws = !!q->gws;
+
+	ret = pqm_checkpoint_mqd(&pdd->process->pqm, q->properties.queue_id, mqd, ctl_stack);
+	if (ret) {
+		pr_err("Failed checkpoint queue_mqd (%d)\n", ret);
+		return ret;
+	}
+
+	pr_debug("Dumping Queue: gpu_id:%x queue_id:%u\n", q_data->gpu_id, q_data->q_id);
+	return ret;
 }
 
+static int criu_checkpoint_queues_device(struct kfd_process_device *pdd,
+				   uint8_t __user *user_priv,
+				   unsigned int *q_index,
+				   uint64_t *queues_priv_data_offset)
+{
+	unsigned int q_private_data_size = 0;
+	uint8_t *q_private_data = NULL; /* Local buffer to store individual queue private data */
+	struct queue *q;
+	int ret = 0;
+
+	list_for_each_entry(q, &pdd->qpd.queues_list, list) {
+		struct kfd_criu_queue_priv_data *q_data;
+		uint64_t q_data_size;
+		uint32_t mqd_size;
+		uint32_t ctl_stack_size;
+
+		if (q->properties.type != KFD_QUEUE_TYPE_COMPUTE &&
+			q->properties.type != KFD_QUEUE_TYPE_SDMA &&
+			q->properties.type != KFD_QUEUE_TYPE_SDMA_XGMI) {
+
+			pr_err("Unsupported queue type (%d)\n", q->properties.type);
+			ret = -EOPNOTSUPP;
+			break;
+		}
+
+		ret = get_queue_data_sizes(pdd, q, &mqd_size, &ctl_stack_size);
+		if (ret)
+			break;
+
+		q_data_size = sizeof(*q_data) + mqd_size + ctl_stack_size;
+
+		/* Increase local buffer space if needed */
+		if (q_private_data_size < q_data_size) {
+			kfree(q_private_data);
+
+			q_private_data = kzalloc(q_data_size, GFP_KERNEL);
+			if (!q_private_data) {
+				ret = -ENOMEM;
+				break;
+			}
+			q_private_data_size = q_data_size;
+		}
+
+		q_data = (struct kfd_criu_queue_priv_data *)q_private_data;
+
+		/*
+		 * data stored in this order:
+		 * priv_data, mqd[xcc0], mqd[xcc1],..., ctl_stack[xcc0], ctl_stack[xcc1]...
+		 */
+		q_data->mqd_size = mqd_size;
+		q_data->ctl_stack_size = ctl_stack_size;
+
+		ret = criu_checkpoint_queue(pdd, q, q_data);
+		if (ret)
+			break;
+
+		q_data->object_type = KFD_CRIU_OBJECT_TYPE_QUEUE;
+
+		ret = copy_to_user(user_priv + *queues_priv_data_offset,
+				q_data, q_data_size);
+		if (ret) {
+			ret = -EFAULT;
+			break;
+		}
+		*queues_priv_data_offset += q_data_size;
+		*q_index = *q_index + 1;
+	}
+
+	kfree(q_private_data);
+
+	return ret;
+}
+
+int kfd_criu_checkpoint_queues(struct kfd_process *p,
+			 uint8_t __user *user_priv_data,
+			 uint64_t *priv_data_offset)
+{
+	int ret = 0, pdd_index, q_index = 0;
+
+	for (pdd_index = 0; pdd_index < p->n_pdds; pdd_index++) {
+		struct kfd_process_device *pdd = p->pdds[pdd_index];
+
+		/*
+		 * criu_checkpoint_queues_device will copy data to user and update q_index and
+		 * queues_priv_data_offset
+		 */
+		ret = criu_checkpoint_queues_device(pdd, user_priv_data, &q_index,
+					      priv_data_offset);
+
+		if (ret)
+			break;
+	}
+
+	return ret;
+}
+
+static void set_queue_properties_from_criu(struct queue_properties *qp,
+					  struct kfd_criu_queue_priv_data *q_data, uint32_t num_xcc)
+{
+	qp->is_interop = false;
+	qp->queue_percent = q_data->q_percent;
+	qp->priority = q_data->priority;
+	qp->queue_address = q_data->q_address;
+	qp->queue_size = q_data->q_size;
+	qp->read_ptr = (uint32_t *) q_data->read_ptr_addr;
+	qp->write_ptr = (uint32_t *) q_data->write_ptr_addr;
+	qp->eop_ring_buffer_address = q_data->eop_ring_buffer_address;
+	qp->eop_ring_buffer_size = q_data->eop_ring_buffer_size;
+	qp->ctx_save_restore_area_address = q_data->ctx_save_restore_area_address;
+	qp->ctx_save_restore_area_size = q_data->ctx_save_restore_area_size;
+	if (q_data->type == KFD_QUEUE_TYPE_COMPUTE)
+		qp->ctl_stack_size = q_data->ctl_stack_size / num_xcc;
+	else
+		qp->ctl_stack_size = q_data->ctl_stack_size;
+
+	qp->type = q_data->type;
+	qp->format = q_data->format;
+}
+
+int kfd_criu_restore_queue(struct kfd_process *p,
+			   uint8_t __user *user_priv_ptr,
+			   uint64_t *priv_data_offset,
+			   uint64_t max_priv_data_size)
+{
+	uint8_t *mqd, *ctl_stack, *q_extra_data = NULL;
+	struct kfd_criu_queue_priv_data *q_data;
+	struct kfd_process_device *pdd;
+	uint64_t q_extra_data_size;
+	struct queue_properties qp;
+	unsigned int queue_id;
+	int ret = 0;
+
+	if (*priv_data_offset + sizeof(*q_data) > max_priv_data_size)
+		return -EINVAL;
+
+	q_data = kmalloc(sizeof(*q_data), GFP_KERNEL);
+	if (!q_data)
+		return -ENOMEM;
+
+	ret = copy_from_user(q_data, user_priv_ptr + *priv_data_offset, sizeof(*q_data));
+	if (ret) {
+		ret = -EFAULT;
+		goto exit;
+	}
+
+	*priv_data_offset += sizeof(*q_data);
+	q_extra_data_size = (uint64_t)q_data->ctl_stack_size + q_data->mqd_size;
+
+	if (*priv_data_offset + q_extra_data_size > max_priv_data_size) {
+		ret = -EINVAL;
+		goto exit;
+	}
+
+	q_extra_data = kmalloc(q_extra_data_size, GFP_KERNEL);
+	if (!q_extra_data) {
+		ret = -ENOMEM;
+		goto exit;
+	}
+
+	ret = copy_from_user(q_extra_data, user_priv_ptr + *priv_data_offset, q_extra_data_size);
+	if (ret) {
+		ret = -EFAULT;
+		goto exit;
+	}
+
+	*priv_data_offset += q_extra_data_size;
+
+	pdd = kfd_process_device_data_by_id(p, q_data->gpu_id);
+	if (!pdd) {
+		pr_err("Failed to get pdd\n");
+		ret = -EINVAL;
+		goto exit;
+	}
+
+	/*
+	 * data stored in this order:
+	 * mqd[xcc0], mqd[xcc1],..., ctl_stack[xcc0], ctl_stack[xcc1]...
+	 */
+	mqd = q_extra_data;
+	ctl_stack = mqd + q_data->mqd_size;
+
+	memset(&qp, 0, sizeof(qp));
+	set_queue_properties_from_criu(&qp, q_data, NUM_XCC(pdd->dev->adev->gfx.xcc_mask));
+
+	print_queue_properties(&qp);
+
+	ret = pqm_create_queue(&p->pqm, pdd->dev, &qp, &queue_id, q_data, mqd, ctl_stack, NULL);
+	if (ret) {
+		pr_err("Failed to create new queue err:%d\n", ret);
+		goto exit;
+	}
+
+	if (q_data->gws)
+		ret = pqm_set_gws(&p->pqm, q_data->q_id, pdd->dev->gws);
+
+exit:
+	if (ret)
+		pr_err("Failed to restore queue (%d)\n", ret);
+	else
+		pr_debug("Queue id %d was restored successfully\n", queue_id);
+
+	kfree(q_data);
+	kfree(q_extra_data);
+
+	return ret;
+}
+
+int pqm_get_queue_checkpoint_info(struct process_queue_manager *pqm,
+				  unsigned int qid,
+				  uint32_t *mqd_size,
+				  uint32_t *ctl_stack_size)
+{
+	struct process_queue_node *pqn;
+
+	pqn = get_queue_by_qid(pqm, qid);
+	if (!pqn) {
+		pr_debug("amdkfd: No queue %d exists for operation\n", qid);
+		return -EFAULT;
+	}
+
+	if (!pqn->q->device->dqm->ops.get_queue_checkpoint_info) {
+		pr_err("amdkfd: queue dumping not supported on this device\n");
+		return -EOPNOTSUPP;
+	}
+
+	pqn->q->device->dqm->ops.get_queue_checkpoint_info(pqn->q->device->dqm,
+						       pqn->q, mqd_size,
+						       ctl_stack_size);
+	return 0;
+}
+
+#if defined(CONFIG_DEBUG_FS)
+
+int pqm_debugfs_mqds(struct seq_file *m, void *data)
+{
+	struct process_queue_manager *pqm = data;
+	struct process_queue_node *pqn;
+	struct queue *q;
+	enum KFD_MQD_TYPE mqd_type;
+	struct mqd_manager *mqd_mgr;
+	int r = 0, xcc, num_xccs = 1;
+	void *mqd;
+	uint64_t size = 0;
+
+	list_for_each_entry(pqn, &pqm->queues, process_queue_list) {
+		if (pqn->q) {
+			q = pqn->q;
+			switch (q->properties.type) {
+			case KFD_QUEUE_TYPE_SDMA:
+			case KFD_QUEUE_TYPE_SDMA_XGMI:
+				seq_printf(m, "  SDMA queue on device %x\n",
+					   q->device->id);
+				mqd_type = KFD_MQD_TYPE_SDMA;
+				break;
+			case KFD_QUEUE_TYPE_COMPUTE:
+				seq_printf(m, "  Compute queue on device %x\n",
+					   q->device->id);
+				mqd_type = KFD_MQD_TYPE_CP;
+				num_xccs = NUM_XCC(q->device->xcc_mask);
+				break;
+			default:
+				seq_printf(m,
+				"  Bad user queue type %d on device %x\n",
+					   q->properties.type, q->device->id);
+				continue;
+			}
+			mqd_mgr = q->device->dqm->mqd_mgrs[mqd_type];
+			size = mqd_mgr->mqd_stride(mqd_mgr,
+							&q->properties);
+		} else if (pqn->kq) {
+			q = pqn->kq->queue;
+			mqd_mgr = pqn->kq->mqd_mgr;
+			switch (q->properties.type) {
+			case KFD_QUEUE_TYPE_DIQ:
+				seq_printf(m, "  DIQ on device %x\n",
+					   pqn->kq->dev->id);
+				break;
+			default:
+				seq_printf(m,
+				"  Bad kernel queue type %d on device %x\n",
+					   q->properties.type,
+					   pqn->kq->dev->id);
+				continue;
+			}
+		} else {
+			seq_printf(m,
+		"  Weird: Queue node with neither kernel nor user queue\n");
+			continue;
+		}
+
+		for (xcc = 0; xcc < num_xccs; xcc++) {
+			mqd = q->mqd + size * xcc;
+			r = mqd_mgr->debugfs_show_mqd(m, mqd);
+			if (r != 0)
+				break;
+		}
+	}
+
+	return r;
+}
 
+#endif
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_queue.c b/drivers/gpu/drm/amd/amdkfd/kfd_queue.c
index a5315d4f1c95..a65c67cf56ff 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_queue.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_queue.c
@@ -1,5 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -23,6 +24,8 @@
 
 #include <linux/slab.h>
 #include "kfd_priv.h"
+#include "kfd_topology.h"
+#include "kfd_svm.h"
 
 void print_queue_properties(struct queue_properties *q)
 {
@@ -36,8 +39,8 @@ void print_queue_properties(struct queue_properties *q)
 	pr_debug("Queue Address: 0x%llX\n", q->queue_address);
 	pr_debug("Queue Id: %u\n", q->queue_id);
 	pr_debug("Queue Process Vmid: %u\n", q->vmid);
-	pr_debug("Queue Read Pointer: 0x%p\n", q->read_ptr);
-	pr_debug("Queue Write Pointer: 0x%p\n", q->write_ptr);
+	pr_debug("Queue Read Pointer: 0x%px\n", q->read_ptr);
+	pr_debug("Queue Write Pointer: 0x%px\n", q->write_ptr);
 	pr_debug("Queue Doorbell Pointer: 0x%p\n", q->doorbell_ptr);
 	pr_debug("Queue Doorbell Offset: %u\n", q->doorbell_off);
 }
@@ -53,8 +56,8 @@ void print_queue(struct queue *q)
 	pr_debug("Queue Address: 0x%llX\n", q->properties.queue_address);
 	pr_debug("Queue Id: %u\n", q->properties.queue_id);
 	pr_debug("Queue Process Vmid: %u\n", q->properties.vmid);
-	pr_debug("Queue Read Pointer: 0x%p\n", q->properties.read_ptr);
-	pr_debug("Queue Write Pointer: 0x%p\n", q->properties.write_ptr);
+	pr_debug("Queue Read Pointer: 0x%px\n", q->properties.read_ptr);
+	pr_debug("Queue Write Pointer: 0x%px\n", q->properties.write_ptr);
 	pr_debug("Queue Doorbell Pointer: 0x%p\n", q->properties.doorbell_ptr);
 	pr_debug("Queue Doorbell Offset: %u\n", q->properties.doorbell_off);
 	pr_debug("Queue MQD Address: 0x%p\n", q->mqd);
@@ -81,3 +84,386 @@ void uninit_queue(struct queue *q)
 {
 	kfree(q);
 }
+
+#if IS_ENABLED(CONFIG_HSA_AMD_SVM)
+
+static int kfd_queue_buffer_svm_get(struct kfd_process_device *pdd, u64 addr, u64 size)
+{
+	struct kfd_process *p = pdd->process;
+	struct list_head update_list;
+	struct svm_range *prange;
+	int ret = -EINVAL;
+
+	INIT_LIST_HEAD(&update_list);
+	addr >>= PAGE_SHIFT;
+	size >>= PAGE_SHIFT;
+
+	mutex_lock(&p->svms.lock);
+
+	/*
+	 * range may split to multiple svm pranges aligned to granularity boundaery.
+	 */
+	while (size) {
+		uint32_t gpuid, gpuidx;
+		int r;
+
+		prange = svm_range_from_addr(&p->svms, addr, NULL);
+		if (!prange)
+			break;
+
+		if (!prange->mapped_to_gpu)
+			break;
+
+		r = kfd_process_gpuid_from_node(p, pdd->dev, &gpuid, &gpuidx);
+		if (r < 0)
+			break;
+		if (!test_bit(gpuidx, prange->bitmap_access) &&
+		    !test_bit(gpuidx, prange->bitmap_aip))
+			break;
+
+		if (!(prange->flags & KFD_IOCTL_SVM_FLAG_GPU_ALWAYS_MAPPED))
+			break;
+
+		list_add(&prange->update_list, &update_list);
+
+		if (prange->last - prange->start + 1 >= size) {
+			size = 0;
+			break;
+		}
+
+		size -= prange->last - prange->start + 1;
+		addr += prange->last - prange->start + 1;
+	}
+	if (size) {
+		pr_debug("[0x%llx 0x%llx] not registered\n", addr, addr + size - 1);
+		goto out_unlock;
+	}
+
+	list_for_each_entry(prange, &update_list, update_list)
+		atomic_inc(&prange->queue_refcount);
+	ret = 0;
+
+out_unlock:
+	mutex_unlock(&p->svms.lock);
+	return ret;
+}
+
+static void kfd_queue_buffer_svm_put(struct kfd_process_device *pdd, u64 addr, u64 size)
+{
+	struct kfd_process *p = pdd->process;
+	struct svm_range *prange, *pchild;
+	struct interval_tree_node *node;
+	unsigned long last;
+
+	addr >>= PAGE_SHIFT;
+	last = addr + (size >> PAGE_SHIFT) - 1;
+
+	mutex_lock(&p->svms.lock);
+
+	node = interval_tree_iter_first(&p->svms.objects, addr, last);
+	while (node) {
+		struct interval_tree_node *next_node;
+		unsigned long next_start;
+
+		prange = container_of(node, struct svm_range, it_node);
+		next_node = interval_tree_iter_next(node, addr, last);
+		next_start = min(node->last, last) + 1;
+
+		if (atomic_add_unless(&prange->queue_refcount, -1, 0)) {
+			list_for_each_entry(pchild, &prange->child_list, child_list)
+				atomic_add_unless(&pchild->queue_refcount, -1, 0);
+		}
+
+		node = next_node;
+		addr = next_start;
+	}
+
+	mutex_unlock(&p->svms.lock);
+}
+#else
+
+static int kfd_queue_buffer_svm_get(struct kfd_process_device *pdd, u64 addr, u64 size)
+{
+	return -EINVAL;
+}
+
+static void kfd_queue_buffer_svm_put(struct kfd_process_device *pdd, u64 addr, u64 size)
+{
+}
+
+#endif
+
+int kfd_queue_buffer_get(struct amdgpu_vm *vm, void __user *addr, struct amdgpu_bo **pbo,
+			 u64 expected_size)
+{
+	struct amdgpu_bo_va_mapping *mapping;
+	u64 user_addr;
+	u64 size;
+
+	user_addr = (u64)addr >> AMDGPU_GPU_PAGE_SHIFT;
+	size = expected_size >> AMDGPU_GPU_PAGE_SHIFT;
+
+	mapping = amdgpu_vm_bo_lookup_mapping(vm, user_addr);
+	if (!mapping)
+		goto out_err;
+
+	if (user_addr != mapping->start ||
+	    (size != 0 && user_addr + size - 1 != mapping->last)) {
+		pr_debug("expected size 0x%llx not equal to mapping addr 0x%llx size 0x%llx\n",
+			expected_size, mapping->start << AMDGPU_GPU_PAGE_SHIFT,
+			(mapping->last - mapping->start + 1) << AMDGPU_GPU_PAGE_SHIFT);
+		goto out_err;
+	}
+
+	*pbo = amdgpu_bo_ref(mapping->bo_va->base.bo);
+	mapping->bo_va->queue_refcount++;
+	return 0;
+
+out_err:
+	*pbo = NULL;
+	return -EINVAL;
+}
+
+/* FIXME: remove this function, just call amdgpu_bo_unref directly */
+void kfd_queue_buffer_put(struct amdgpu_bo **bo)
+{
+	amdgpu_bo_unref(bo);
+}
+
+int kfd_queue_acquire_buffers(struct kfd_process_device *pdd, struct queue_properties *properties)
+{
+	struct kfd_topology_device *topo_dev;
+	u64 expected_queue_size;
+	struct amdgpu_vm *vm;
+	u32 total_cwsr_size;
+	int err;
+
+	topo_dev = kfd_topology_device_by_id(pdd->dev->id);
+	if (!topo_dev)
+		return -EINVAL;
+
+	/* AQL queues on GFX7 and GFX8 appear twice their actual size */
+	if (properties->type == KFD_QUEUE_TYPE_COMPUTE &&
+	    properties->format == KFD_QUEUE_FORMAT_AQL &&
+	    topo_dev->node_props.gfx_target_version >= 70000 &&
+	    topo_dev->node_props.gfx_target_version < 90000)
+		expected_queue_size = properties->queue_size / 2;
+	else
+		expected_queue_size = properties->queue_size;
+
+	vm = drm_priv_to_vm(pdd->drm_priv);
+	err = amdgpu_bo_reserve(vm->root.bo, false);
+	if (err)
+		return err;
+
+	err = kfd_queue_buffer_get(vm, properties->write_ptr, &properties->wptr_bo, PAGE_SIZE);
+	if (err)
+		goto out_err_unreserve;
+
+	err = kfd_queue_buffer_get(vm, properties->read_ptr, &properties->rptr_bo, PAGE_SIZE);
+	if (err)
+		goto out_err_unreserve;
+
+	err = kfd_queue_buffer_get(vm, (void *)properties->queue_address,
+				   &properties->ring_bo, expected_queue_size);
+	if (err)
+		goto out_err_unreserve;
+
+	/* only compute queue requires EOP buffer and CWSR area */
+	if (properties->type != KFD_QUEUE_TYPE_COMPUTE)
+		goto out_unreserve;
+
+	/* EOP buffer is not required for all ASICs */
+	if (properties->eop_ring_buffer_address) {
+		if (properties->eop_ring_buffer_size != topo_dev->node_props.eop_buffer_size) {
+			pr_debug("queue eop bo size 0x%x not equal to node eop buf size 0x%x\n",
+				properties->eop_ring_buffer_size,
+				topo_dev->node_props.eop_buffer_size);
+			err = -EINVAL;
+			goto out_err_unreserve;
+		}
+		err = kfd_queue_buffer_get(vm, (void *)properties->eop_ring_buffer_address,
+					   &properties->eop_buf_bo,
+					   properties->eop_ring_buffer_size);
+		if (err)
+			goto out_err_unreserve;
+	}
+
+	if (properties->ctl_stack_size != topo_dev->node_props.ctl_stack_size) {
+		pr_debug("queue ctl stack size 0x%x not equal to node ctl stack size 0x%x\n",
+			properties->ctl_stack_size,
+			topo_dev->node_props.ctl_stack_size);
+		err = -EINVAL;
+		goto out_err_unreserve;
+	}
+
+	if (properties->ctx_save_restore_area_size != topo_dev->node_props.cwsr_size) {
+		pr_debug("queue cwsr size 0x%x not equal to node cwsr size 0x%x\n",
+			properties->ctx_save_restore_area_size,
+			topo_dev->node_props.cwsr_size);
+		err = -EINVAL;
+		goto out_err_unreserve;
+	}
+
+	total_cwsr_size = (topo_dev->node_props.cwsr_size + topo_dev->node_props.debug_memory_size)
+			  * NUM_XCC(pdd->dev->xcc_mask);
+	total_cwsr_size = ALIGN(total_cwsr_size, PAGE_SIZE);
+
+	err = kfd_queue_buffer_get(vm, (void *)properties->ctx_save_restore_area_address,
+				   &properties->cwsr_bo, total_cwsr_size);
+	if (!err)
+		goto out_unreserve;
+
+	amdgpu_bo_unreserve(vm->root.bo);
+
+	err = kfd_queue_buffer_svm_get(pdd, properties->ctx_save_restore_area_address,
+				       total_cwsr_size);
+	if (err)
+		goto out_err_release;
+
+	return 0;
+
+out_unreserve:
+	amdgpu_bo_unreserve(vm->root.bo);
+	return 0;
+
+out_err_unreserve:
+	amdgpu_bo_unreserve(vm->root.bo);
+out_err_release:
+	/* FIXME: make a _locked version of this that can be called before
+	 * dropping the VM reservation.
+	 */
+	kfd_queue_unref_bo_vas(pdd, properties);
+	kfd_queue_release_buffers(pdd, properties);
+	return err;
+}
+
+int kfd_queue_release_buffers(struct kfd_process_device *pdd, struct queue_properties *properties)
+{
+	struct kfd_topology_device *topo_dev;
+	u32 total_cwsr_size;
+
+	kfd_queue_buffer_put(&properties->wptr_bo);
+	kfd_queue_buffer_put(&properties->rptr_bo);
+	kfd_queue_buffer_put(&properties->ring_bo);
+	kfd_queue_buffer_put(&properties->eop_buf_bo);
+	kfd_queue_buffer_put(&properties->cwsr_bo);
+
+	topo_dev = kfd_topology_device_by_id(pdd->dev->id);
+	if (!topo_dev)
+		return -EINVAL;
+	total_cwsr_size = (topo_dev->node_props.cwsr_size + topo_dev->node_props.debug_memory_size)
+			  * NUM_XCC(pdd->dev->xcc_mask);
+	total_cwsr_size = ALIGN(total_cwsr_size, PAGE_SIZE);
+
+	kfd_queue_buffer_svm_put(pdd, properties->ctx_save_restore_area_address, total_cwsr_size);
+	return 0;
+}
+
+void kfd_queue_unref_bo_va(struct amdgpu_vm *vm, struct amdgpu_bo **bo)
+{
+	if (*bo) {
+		struct amdgpu_bo_va *bo_va;
+
+		bo_va = amdgpu_vm_bo_find(vm, *bo);
+		if (bo_va && bo_va->queue_refcount)
+			bo_va->queue_refcount--;
+	}
+}
+
+int kfd_queue_unref_bo_vas(struct kfd_process_device *pdd,
+			   struct queue_properties *properties)
+{
+	struct amdgpu_vm *vm;
+	int err;
+
+	vm = drm_priv_to_vm(pdd->drm_priv);
+	err = amdgpu_bo_reserve(vm->root.bo, false);
+	if (err)
+		return err;
+
+	kfd_queue_unref_bo_va(vm, &properties->wptr_bo);
+	kfd_queue_unref_bo_va(vm, &properties->rptr_bo);
+	kfd_queue_unref_bo_va(vm, &properties->ring_bo);
+	kfd_queue_unref_bo_va(vm, &properties->eop_buf_bo);
+	kfd_queue_unref_bo_va(vm, &properties->cwsr_bo);
+
+	amdgpu_bo_unreserve(vm->root.bo);
+	return 0;
+}
+
+#define SGPR_SIZE_PER_CU	0x4000
+#define LDS_SIZE_PER_CU		0x10000
+#define HWREG_SIZE_PER_CU	0x1000
+#define DEBUGGER_BYTES_ALIGN	64
+#define DEBUGGER_BYTES_PER_WAVE	32
+
+static u32 kfd_get_vgpr_size_per_cu(u32 gfxv)
+{
+	u32 vgpr_size = 0x40000;
+
+	if (gfxv == 90402 ||			/* GFX_VERSION_AQUA_VANJARAM */
+	    gfxv == 90010 ||			/* GFX_VERSION_ALDEBARAN */
+	    gfxv == 90008 ||			/* GFX_VERSION_ARCTURUS */
+	    gfxv == 90500)
+		vgpr_size = 0x80000;
+	else if (gfxv == 110000 ||		/* GFX_VERSION_PLUM_BONITO */
+		 gfxv == 110001 ||		/* GFX_VERSION_WHEAT_NAS */
+		 gfxv == 120000 ||		/* GFX_VERSION_GFX1200 */
+		 gfxv == 120001)		/* GFX_VERSION_GFX1201 */
+		vgpr_size = 0x60000;
+
+	return vgpr_size;
+}
+
+#define WG_CONTEXT_DATA_SIZE_PER_CU(gfxv, props)	\
+	(kfd_get_vgpr_size_per_cu(gfxv) + SGPR_SIZE_PER_CU +\
+	 (((gfxv) == 90500) ? (props->lds_size_in_kb << 10) : LDS_SIZE_PER_CU) +\
+	 HWREG_SIZE_PER_CU)
+
+#define CNTL_STACK_BYTES_PER_WAVE(gfxv)	\
+	((gfxv) >= 100100 ? 12 : 8)	/* GFX_VERSION_NAVI10*/
+
+#define SIZEOF_HSA_USER_CONTEXT_SAVE_AREA_HEADER 40
+
+void kfd_queue_ctx_save_restore_size(struct kfd_topology_device *dev)
+{
+	struct kfd_node_properties *props = &dev->node_props;
+	u32 gfxv = props->gfx_target_version;
+	u32 ctl_stack_size;
+	u32 wg_data_size;
+	u32 wave_num;
+	u32 cu_num;
+
+	if (gfxv < 80001)	/* GFX_VERSION_CARRIZO */
+		return;
+
+	cu_num = props->simd_count / props->simd_per_cu / NUM_XCC(dev->gpu->xcc_mask);
+	wave_num = (gfxv < 100100) ?	/* GFX_VERSION_NAVI10 */
+		    min(cu_num * 40, props->array_count / props->simd_arrays_per_engine * 512)
+		    : cu_num * 32;
+
+	wg_data_size = ALIGN(cu_num * WG_CONTEXT_DATA_SIZE_PER_CU(gfxv, props), PAGE_SIZE);
+	ctl_stack_size = wave_num * CNTL_STACK_BYTES_PER_WAVE(gfxv) + 8;
+	ctl_stack_size = ALIGN(SIZEOF_HSA_USER_CONTEXT_SAVE_AREA_HEADER + ctl_stack_size,
+			       PAGE_SIZE);
+
+	if ((gfxv / 10000 * 10000) == 100000) {
+		/* HW design limits control stack size to 0x7000.
+		 * This is insufficient for theoretical PM4 cases
+		 * but sufficient for AQL, limited by SPI events.
+		 */
+		ctl_stack_size = min(ctl_stack_size, 0x7000);
+	}
+
+	props->ctl_stack_size = ctl_stack_size;
+	props->debug_memory_size = ALIGN(wave_num * DEBUGGER_BYTES_PER_WAVE, DEBUGGER_BYTES_ALIGN);
+	props->cwsr_size = ctl_stack_size + wg_data_size;
+
+	if (gfxv == 80002)	/* GFX_VERSION_TONGA */
+		props->eop_buffer_size = 0x8000;
+	else if (gfxv == 90402)	/* GFX_VERSION_AQUA_VANJARAM */
+		props->eop_buffer_size = 4096;
+	else if (gfxv >= 80000)
+		props->eop_buffer_size = 4096;
+}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_smi_events.c b/drivers/gpu/drm/amd/amdkfd/kfd_smi_events.c
new file mode 100644
index 000000000000..a499449fcb06
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_smi_events.c
@@ -0,0 +1,411 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
+/*
+ * Copyright 2020-2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include <linux/poll.h>
+#include <linux/wait.h>
+#include <linux/anon_inodes.h>
+#include <uapi/linux/kfd_ioctl.h>
+#include "amdgpu.h"
+#include "amdgpu_vm.h"
+#include "kfd_priv.h"
+#include "kfd_smi_events.h"
+#include "amdgpu_reset.h"
+
+struct kfd_smi_client {
+	struct list_head list;
+	struct kfifo fifo;
+	wait_queue_head_t wait_queue;
+	/* events enabled */
+	uint64_t events;
+	struct kfd_node *dev;
+	spinlock_t lock;
+	struct rcu_head rcu;
+	pid_t pid;
+	bool suser;
+};
+
+#define KFD_MAX_KFIFO_SIZE	8192
+
+static __poll_t kfd_smi_ev_poll(struct file *, struct poll_table_struct *);
+static ssize_t kfd_smi_ev_read(struct file *, char __user *, size_t, loff_t *);
+static ssize_t kfd_smi_ev_write(struct file *, const char __user *, size_t,
+				loff_t *);
+static int kfd_smi_ev_release(struct inode *, struct file *);
+
+static const char kfd_smi_name[] = "kfd_smi_ev";
+
+static const struct file_operations kfd_smi_ev_fops = {
+	.owner = THIS_MODULE,
+	.poll = kfd_smi_ev_poll,
+	.read = kfd_smi_ev_read,
+	.write = kfd_smi_ev_write,
+	.release = kfd_smi_ev_release
+};
+
+static __poll_t kfd_smi_ev_poll(struct file *filep,
+				struct poll_table_struct *wait)
+{
+	struct kfd_smi_client *client = filep->private_data;
+	__poll_t mask = 0;
+
+	poll_wait(filep, &client->wait_queue, wait);
+
+	spin_lock(&client->lock);
+	if (!kfifo_is_empty(&client->fifo))
+		mask = EPOLLIN | EPOLLRDNORM;
+	spin_unlock(&client->lock);
+
+	return mask;
+}
+
+static ssize_t kfd_smi_ev_read(struct file *filep, char __user *user,
+			       size_t size, loff_t *offset)
+{
+	int ret;
+	size_t to_copy;
+	struct kfd_smi_client *client = filep->private_data;
+	unsigned char *buf;
+
+	size = min_t(size_t, size, KFD_MAX_KFIFO_SIZE);
+	buf = kmalloc(size, GFP_KERNEL);
+	if (!buf)
+		return -ENOMEM;
+
+	/* kfifo_to_user can sleep so we can't use spinlock protection around
+	 * it. Instead, we kfifo out as spinlocked then copy them to the user.
+	 */
+	spin_lock(&client->lock);
+	to_copy = kfifo_len(&client->fifo);
+	if (!to_copy) {
+		spin_unlock(&client->lock);
+		ret = -EAGAIN;
+		goto ret_err;
+	}
+	to_copy = min(size, to_copy);
+	ret = kfifo_out(&client->fifo, buf, to_copy);
+	spin_unlock(&client->lock);
+	if (ret <= 0) {
+		ret = -EAGAIN;
+		goto ret_err;
+	}
+
+	ret = copy_to_user(user, buf, to_copy);
+	if (ret) {
+		ret = -EFAULT;
+		goto ret_err;
+	}
+
+	kfree(buf);
+	return to_copy;
+
+ret_err:
+	kfree(buf);
+	return ret;
+}
+
+static ssize_t kfd_smi_ev_write(struct file *filep, const char __user *user,
+				size_t size, loff_t *offset)
+{
+	struct kfd_smi_client *client = filep->private_data;
+	uint64_t events;
+
+	if (!access_ok(user, size) || size < sizeof(events))
+		return -EFAULT;
+	if (copy_from_user(&events, user, sizeof(events)))
+		return -EFAULT;
+
+	WRITE_ONCE(client->events, events);
+
+	return sizeof(events);
+}
+
+static void kfd_smi_ev_client_free(struct rcu_head *p)
+{
+	struct kfd_smi_client *ev = container_of(p, struct kfd_smi_client, rcu);
+
+	kfifo_free(&ev->fifo);
+	kfree(ev);
+}
+
+static int kfd_smi_ev_release(struct inode *inode, struct file *filep)
+{
+	struct kfd_smi_client *client = filep->private_data;
+	struct kfd_node *dev = client->dev;
+
+	spin_lock(&dev->smi_lock);
+	list_del_rcu(&client->list);
+	spin_unlock(&dev->smi_lock);
+
+	call_rcu(&client->rcu, kfd_smi_ev_client_free);
+	return 0;
+}
+
+static bool kfd_smi_ev_enabled(pid_t pid, struct kfd_smi_client *client,
+			       unsigned int event)
+{
+	uint64_t events = READ_ONCE(client->events);
+
+	if (pid && client->pid != pid && !client->suser)
+		return false;
+
+	return events & KFD_SMI_EVENT_MASK_FROM_INDEX(event);
+}
+
+static void add_event_to_kfifo(pid_t pid, struct kfd_node *dev,
+			       unsigned int smi_event, char *event_msg, int len)
+{
+	struct kfd_smi_client *client;
+
+	rcu_read_lock();
+
+	list_for_each_entry_rcu(client, &dev->smi_clients, list) {
+		if (!kfd_smi_ev_enabled(pid, client, smi_event))
+			continue;
+		spin_lock(&client->lock);
+		if (kfifo_avail(&client->fifo) >= len) {
+			kfifo_in(&client->fifo, event_msg, len);
+			wake_up_all(&client->wait_queue);
+		} else {
+			pr_debug("smi_event(EventID: %u): no space left\n",
+					smi_event);
+		}
+		spin_unlock(&client->lock);
+	}
+
+	rcu_read_unlock();
+}
+
+__printf(4, 5)
+static void kfd_smi_event_add(pid_t pid, struct kfd_node *dev,
+			      unsigned int event, char *fmt, ...)
+{
+	char fifo_in[KFD_SMI_EVENT_MSG_SIZE];
+	int len;
+	va_list args;
+
+	if (list_empty(&dev->smi_clients))
+		return;
+
+	len = snprintf(fifo_in, sizeof(fifo_in), "%x ", event);
+
+	va_start(args, fmt);
+	len += vsnprintf(fifo_in + len, sizeof(fifo_in) - len, fmt, args);
+	va_end(args);
+
+	add_event_to_kfifo(pid, dev, event, fifo_in, len);
+}
+
+void kfd_smi_event_update_gpu_reset(struct kfd_node *dev, bool post_reset,
+				    struct amdgpu_reset_context *reset_context)
+{
+	unsigned int event;
+	char reset_cause[64];
+
+	if (post_reset) {
+		event = KFD_SMI_EVENT_GPU_POST_RESET;
+	} else {
+		event = KFD_SMI_EVENT_GPU_PRE_RESET;
+		++(dev->reset_seq_num);
+	}
+
+	memset(reset_cause, 0, sizeof(reset_cause));
+
+	if (reset_context)
+		amdgpu_reset_get_desc(reset_context, reset_cause,
+				      sizeof(reset_cause));
+
+	kfd_smi_event_add(0, dev, event, KFD_EVENT_FMT_UPDATE_GPU_RESET(
+			  dev->reset_seq_num, reset_cause));
+}
+
+void kfd_smi_event_update_thermal_throttling(struct kfd_node *dev,
+					     uint64_t throttle_bitmask)
+{
+	kfd_smi_event_add(0, dev, KFD_SMI_EVENT_THERMAL_THROTTLE, KFD_EVENT_FMT_THERMAL_THROTTLING(
+			  throttle_bitmask,
+			  amdgpu_dpm_get_thermal_throttling_counter(dev->adev)));
+}
+
+void kfd_smi_event_update_vmfault(struct kfd_node *dev, uint16_t pasid)
+{
+	struct amdgpu_task_info *task_info;
+
+	task_info = amdgpu_vm_get_task_info_pasid(dev->adev, pasid);
+	if (task_info) {
+		/* Report VM faults from user applications, not retry from kernel */
+		if (task_info->task.pid)
+			kfd_smi_event_add(0, dev, KFD_SMI_EVENT_VMFAULT, KFD_EVENT_FMT_VMFAULT(
+					  task_info->task.pid, task_info->task.comm));
+		amdgpu_vm_put_task_info(task_info);
+	}
+}
+
+void kfd_smi_event_page_fault_start(struct kfd_node *node, pid_t pid,
+				    unsigned long address, bool write_fault,
+				    ktime_t ts)
+{
+	kfd_smi_event_add(pid, node, KFD_SMI_EVENT_PAGE_FAULT_START,
+			  KFD_EVENT_FMT_PAGEFAULT_START(ktime_to_ns(ts), pid,
+			  address, node->id, write_fault ? 'W' : 'R'));
+}
+
+void kfd_smi_event_page_fault_end(struct kfd_node *node, pid_t pid,
+				  unsigned long address, bool migration)
+{
+	kfd_smi_event_add(pid, node, KFD_SMI_EVENT_PAGE_FAULT_END,
+			  KFD_EVENT_FMT_PAGEFAULT_END(ktime_get_boottime_ns(),
+			  pid, address, node->id, migration ? 'M' : 'U'));
+}
+
+void kfd_smi_event_migration_start(struct kfd_node *node, pid_t pid,
+				   unsigned long start, unsigned long end,
+				   uint32_t from, uint32_t to,
+				   uint32_t prefetch_loc, uint32_t preferred_loc,
+				   uint32_t trigger)
+{
+	kfd_smi_event_add(pid, node, KFD_SMI_EVENT_MIGRATE_START,
+			  KFD_EVENT_FMT_MIGRATE_START(
+			  ktime_get_boottime_ns(), pid, start, end - start,
+			  from, to, prefetch_loc, preferred_loc, trigger));
+}
+
+void kfd_smi_event_migration_end(struct kfd_node *node, pid_t pid,
+				 unsigned long start, unsigned long end,
+				 uint32_t from, uint32_t to, uint32_t trigger,
+				 int error_code)
+{
+	kfd_smi_event_add(pid, node, KFD_SMI_EVENT_MIGRATE_END,
+			  KFD_EVENT_FMT_MIGRATE_END(
+			  ktime_get_boottime_ns(), pid, start, end - start,
+			  from, to, trigger, error_code));
+}
+
+void kfd_smi_event_queue_eviction(struct kfd_node *node, pid_t pid,
+				  uint32_t trigger)
+{
+	kfd_smi_event_add(pid, node, KFD_SMI_EVENT_QUEUE_EVICTION,
+			  KFD_EVENT_FMT_QUEUE_EVICTION(ktime_get_boottime_ns(), pid,
+			  node->id, trigger));
+}
+
+void kfd_smi_event_queue_restore(struct kfd_node *node, pid_t pid)
+{
+	kfd_smi_event_add(pid, node, KFD_SMI_EVENT_QUEUE_RESTORE,
+			  KFD_EVENT_FMT_QUEUE_RESTORE(ktime_get_boottime_ns(), pid,
+			  node->id, 0));
+}
+
+void kfd_smi_event_queue_restore_rescheduled(struct mm_struct *mm)
+{
+	struct kfd_process *p;
+	int i;
+
+	p = kfd_lookup_process_by_mm(mm);
+	if (!p)
+		return;
+
+	for (i = 0; i < p->n_pdds; i++) {
+		struct kfd_process_device *pdd = p->pdds[i];
+
+		kfd_smi_event_add(p->lead_thread->pid, pdd->dev,
+				  KFD_SMI_EVENT_QUEUE_RESTORE,
+				  KFD_EVENT_FMT_QUEUE_RESTORE(ktime_get_boottime_ns(),
+				  p->lead_thread->pid, pdd->dev->id, 'R'));
+	}
+	kfd_unref_process(p);
+}
+
+void kfd_smi_event_unmap_from_gpu(struct kfd_node *node, pid_t pid,
+				  unsigned long address, unsigned long last,
+				  uint32_t trigger)
+{
+	kfd_smi_event_add(pid, node, KFD_SMI_EVENT_UNMAP_FROM_GPU,
+			  KFD_EVENT_FMT_UNMAP_FROM_GPU(ktime_get_boottime_ns(),
+			  pid, address, last - address + 1, node->id, trigger));
+}
+
+void kfd_smi_event_process(struct kfd_process_device *pdd, bool start)
+{
+	struct amdgpu_task_info *task_info;
+	struct amdgpu_vm *avm;
+
+	if (!pdd->drm_priv)
+		return;
+
+	avm = drm_priv_to_vm(pdd->drm_priv);
+	task_info = amdgpu_vm_get_task_info_vm(avm);
+
+	if (task_info) {
+		kfd_smi_event_add(0, pdd->dev,
+				  start ? KFD_SMI_EVENT_PROCESS_START :
+				  KFD_SMI_EVENT_PROCESS_END,
+				  KFD_EVENT_FMT_PROCESS(task_info->task.pid,
+				  task_info->task.comm));
+		amdgpu_vm_put_task_info(task_info);
+	}
+}
+
+int kfd_smi_event_open(struct kfd_node *dev, uint32_t *fd)
+{
+	struct kfd_smi_client *client;
+	int ret;
+
+	client = kzalloc(sizeof(struct kfd_smi_client), GFP_KERNEL);
+	if (!client)
+		return -ENOMEM;
+	INIT_LIST_HEAD(&client->list);
+
+	ret = kfifo_alloc(&client->fifo, KFD_MAX_KFIFO_SIZE, GFP_KERNEL);
+	if (ret) {
+		kfree(client);
+		return ret;
+	}
+
+	init_waitqueue_head(&client->wait_queue);
+	spin_lock_init(&client->lock);
+	client->events = 0;
+	client->dev = dev;
+	client->pid = current->tgid;
+	client->suser = capable(CAP_SYS_ADMIN);
+
+	spin_lock(&dev->smi_lock);
+	list_add_rcu(&client->list, &dev->smi_clients);
+	spin_unlock(&dev->smi_lock);
+
+	ret = anon_inode_getfd(kfd_smi_name, &kfd_smi_ev_fops, (void *)client,
+			       O_RDWR);
+	if (ret < 0) {
+		spin_lock(&dev->smi_lock);
+		list_del_rcu(&client->list);
+		spin_unlock(&dev->smi_lock);
+
+		synchronize_rcu();
+
+		kfifo_free(&client->fifo);
+		kfree(client);
+		return ret;
+	}
+	*fd = ret;
+
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_smi_events.h b/drivers/gpu/drm/amd/amdkfd/kfd_smi_events.h
new file mode 100644
index 000000000000..bb4d72b57387
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_smi_events.h
@@ -0,0 +1,57 @@
+/* SPDX-License-Identifier: GPL-2.0 OR MIT */
+/*
+ * Copyright 2020-2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#ifndef KFD_SMI_EVENTS_H_INCLUDED
+#define KFD_SMI_EVENTS_H_INCLUDED
+
+struct amdgpu_reset_context;
+
+int kfd_smi_event_open(struct kfd_node *dev, uint32_t *fd);
+void kfd_smi_event_update_vmfault(struct kfd_node *dev, uint16_t pasid);
+void kfd_smi_event_update_thermal_throttling(struct kfd_node *dev,
+					     uint64_t throttle_bitmask);
+void kfd_smi_event_update_gpu_reset(struct kfd_node *dev, bool post_reset,
+				    struct amdgpu_reset_context *reset_context);
+void kfd_smi_event_page_fault_start(struct kfd_node *node, pid_t pid,
+				    unsigned long address, bool write_fault,
+				    ktime_t ts);
+void kfd_smi_event_page_fault_end(struct kfd_node *node, pid_t pid,
+				  unsigned long address, bool migration);
+void kfd_smi_event_migration_start(struct kfd_node *node, pid_t pid,
+			     unsigned long start, unsigned long end,
+			     uint32_t from, uint32_t to,
+			     uint32_t prefetch_loc, uint32_t preferred_loc,
+			     uint32_t trigger);
+void kfd_smi_event_migration_end(struct kfd_node *node, pid_t pid,
+			     unsigned long start, unsigned long end,
+			     uint32_t from, uint32_t to, uint32_t trigger,
+			     int error_code);
+void kfd_smi_event_queue_eviction(struct kfd_node *node, pid_t pid,
+				  uint32_t trigger);
+void kfd_smi_event_queue_restore(struct kfd_node *node, pid_t pid);
+void kfd_smi_event_queue_restore_rescheduled(struct mm_struct *mm);
+void kfd_smi_event_unmap_from_gpu(struct kfd_node *node, pid_t pid,
+				  unsigned long address, unsigned long last,
+				  uint32_t trigger);
+void kfd_smi_event_process(struct kfd_process_device *pdd, bool start);
+#endif
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
new file mode 100644
index 000000000000..9d72411c3379
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
@@ -0,0 +1,4271 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
+/*
+ * Copyright 2020-2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include <linux/types.h>
+#include <linux/sched/task.h>
+#include <linux/dynamic_debug.h>
+#include <drm/ttm/ttm_tt.h>
+#include <drm/drm_exec.h>
+
+#include "amdgpu_sync.h"
+#include "amdgpu_object.h"
+#include "amdgpu_vm.h"
+#include "amdgpu_hmm.h"
+#include "amdgpu.h"
+#include "amdgpu_xgmi.h"
+#include "kfd_priv.h"
+#include "kfd_svm.h"
+#include "kfd_migrate.h"
+#include "kfd_smi_events.h"
+
+#ifdef dev_fmt
+#undef dev_fmt
+#endif
+#define dev_fmt(fmt) "kfd_svm: %s: " fmt, __func__
+
+#define AMDGPU_SVM_RANGE_RESTORE_DELAY_MS 1
+
+/* Long enough to ensure no retry fault comes after svm range is restored and
+ * page table is updated.
+ */
+#define AMDGPU_SVM_RANGE_RETRY_FAULT_PENDING	(2UL * NSEC_PER_MSEC)
+#if IS_ENABLED(CONFIG_DYNAMIC_DEBUG)
+#define dynamic_svm_range_dump(svms) \
+	_dynamic_func_call_no_desc("svm_range_dump", svm_range_debug_dump, svms)
+#else
+#define dynamic_svm_range_dump(svms) \
+	do { if (0) svm_range_debug_dump(svms); } while (0)
+#endif
+
+/* Giant svm range split into smaller ranges based on this, it is decided using
+ * minimum of all dGPU/APU 1/32 VRAM size, between 2MB to 1GB and alignment to
+ * power of 2MB.
+ */
+static uint64_t max_svm_range_pages;
+
+struct criu_svm_metadata {
+	struct list_head list;
+	struct kfd_criu_svm_range_priv_data data;
+};
+
+static void svm_range_evict_svm_bo_worker(struct work_struct *work);
+static bool
+svm_range_cpu_invalidate_pagetables(struct mmu_interval_notifier *mni,
+				    const struct mmu_notifier_range *range,
+				    unsigned long cur_seq);
+static int
+svm_range_check_vm(struct kfd_process *p, uint64_t start, uint64_t last,
+		   uint64_t *bo_s, uint64_t *bo_l);
+static const struct mmu_interval_notifier_ops svm_range_mn_ops = {
+	.invalidate = svm_range_cpu_invalidate_pagetables,
+};
+
+/**
+ * svm_range_unlink - unlink svm_range from lists and interval tree
+ * @prange: svm range structure to be removed
+ *
+ * Remove the svm_range from the svms and svm_bo lists and the svms
+ * interval tree.
+ *
+ * Context: The caller must hold svms->lock
+ */
+static void svm_range_unlink(struct svm_range *prange)
+{
+	pr_debug("svms 0x%p prange 0x%p [0x%lx 0x%lx]\n", prange->svms,
+		 prange, prange->start, prange->last);
+
+	if (prange->svm_bo) {
+		spin_lock(&prange->svm_bo->list_lock);
+		list_del(&prange->svm_bo_list);
+		spin_unlock(&prange->svm_bo->list_lock);
+	}
+
+	list_del(&prange->list);
+	if (prange->it_node.start != 0 && prange->it_node.last != 0)
+		interval_tree_remove(&prange->it_node, &prange->svms->objects);
+}
+
+static void
+svm_range_add_notifier_locked(struct mm_struct *mm, struct svm_range *prange)
+{
+	pr_debug("svms 0x%p prange 0x%p [0x%lx 0x%lx]\n", prange->svms,
+		 prange, prange->start, prange->last);
+
+	mmu_interval_notifier_insert_locked(&prange->notifier, mm,
+				     prange->start << PAGE_SHIFT,
+				     prange->npages << PAGE_SHIFT,
+				     &svm_range_mn_ops);
+}
+
+/**
+ * svm_range_add_to_svms - add svm range to svms
+ * @prange: svm range structure to be added
+ *
+ * Add the svm range to svms interval tree and link list
+ *
+ * Context: The caller must hold svms->lock
+ */
+static void svm_range_add_to_svms(struct svm_range *prange)
+{
+	pr_debug("svms 0x%p prange 0x%p [0x%lx 0x%lx]\n", prange->svms,
+		 prange, prange->start, prange->last);
+
+	list_move_tail(&prange->list, &prange->svms->list);
+	prange->it_node.start = prange->start;
+	prange->it_node.last = prange->last;
+	interval_tree_insert(&prange->it_node, &prange->svms->objects);
+}
+
+static void svm_range_remove_notifier(struct svm_range *prange)
+{
+	pr_debug("remove notifier svms 0x%p prange 0x%p [0x%lx 0x%lx]\n",
+		 prange->svms, prange,
+		 prange->notifier.interval_tree.start >> PAGE_SHIFT,
+		 prange->notifier.interval_tree.last >> PAGE_SHIFT);
+
+	if (prange->notifier.interval_tree.start != 0 &&
+	    prange->notifier.interval_tree.last != 0)
+		mmu_interval_notifier_remove(&prange->notifier);
+}
+
+static bool
+svm_is_valid_dma_mapping_addr(struct device *dev, dma_addr_t dma_addr)
+{
+	return dma_addr && !dma_mapping_error(dev, dma_addr) &&
+	       !(dma_addr & SVM_RANGE_VRAM_DOMAIN);
+}
+
+static int
+svm_range_dma_map_dev(struct amdgpu_device *adev, struct svm_range *prange,
+		      unsigned long offset, unsigned long npages,
+		      unsigned long *hmm_pfns, uint32_t gpuidx)
+{
+	enum dma_data_direction dir = DMA_BIDIRECTIONAL;
+	dma_addr_t *addr = prange->dma_addr[gpuidx];
+	struct device *dev = adev->dev;
+	struct page *page;
+	int i, r;
+
+	if (!addr) {
+		addr = kvcalloc(prange->npages, sizeof(*addr), GFP_KERNEL);
+		if (!addr)
+			return -ENOMEM;
+		prange->dma_addr[gpuidx] = addr;
+	}
+
+	addr += offset;
+	for (i = 0; i < npages; i++) {
+		if (svm_is_valid_dma_mapping_addr(dev, addr[i]))
+			dma_unmap_page(dev, addr[i], PAGE_SIZE, dir);
+
+		page = hmm_pfn_to_page(hmm_pfns[i]);
+		if (is_zone_device_page(page)) {
+			struct amdgpu_device *bo_adev = prange->svm_bo->node->adev;
+
+			addr[i] = (hmm_pfns[i] << PAGE_SHIFT) +
+				   bo_adev->vm_manager.vram_base_offset -
+				   bo_adev->kfd.pgmap.range.start;
+			addr[i] |= SVM_RANGE_VRAM_DOMAIN;
+			pr_debug_ratelimited("vram address: 0x%llx\n", addr[i]);
+			continue;
+		}
+		addr[i] = dma_map_page(dev, page, 0, PAGE_SIZE, dir);
+		r = dma_mapping_error(dev, addr[i]);
+		if (r) {
+			dev_err(dev, "failed %d dma_map_page\n", r);
+			return r;
+		}
+		pr_debug_ratelimited("dma mapping 0x%llx for page addr 0x%lx\n",
+				     addr[i] >> PAGE_SHIFT, page_to_pfn(page));
+	}
+
+	return 0;
+}
+
+static int
+svm_range_dma_map(struct svm_range *prange, unsigned long *bitmap,
+		  unsigned long offset, unsigned long npages,
+		  unsigned long *hmm_pfns)
+{
+	struct kfd_process *p;
+	uint32_t gpuidx;
+	int r;
+
+	p = container_of(prange->svms, struct kfd_process, svms);
+
+	for_each_set_bit(gpuidx, bitmap, MAX_GPU_INSTANCE) {
+		struct kfd_process_device *pdd;
+
+		pr_debug("mapping to gpu idx 0x%x\n", gpuidx);
+		pdd = kfd_process_device_from_gpuidx(p, gpuidx);
+		if (!pdd) {
+			pr_debug("failed to find device idx %d\n", gpuidx);
+			return -EINVAL;
+		}
+
+		r = svm_range_dma_map_dev(pdd->dev->adev, prange, offset, npages,
+					  hmm_pfns, gpuidx);
+		if (r)
+			break;
+	}
+
+	return r;
+}
+
+void svm_range_dma_unmap_dev(struct device *dev, dma_addr_t *dma_addr,
+			 unsigned long offset, unsigned long npages)
+{
+	enum dma_data_direction dir = DMA_BIDIRECTIONAL;
+	int i;
+
+	if (!dma_addr)
+		return;
+
+	for (i = offset; i < offset + npages; i++) {
+		if (!svm_is_valid_dma_mapping_addr(dev, dma_addr[i]))
+			continue;
+		pr_debug_ratelimited("unmap 0x%llx\n", dma_addr[i] >> PAGE_SHIFT);
+		dma_unmap_page(dev, dma_addr[i], PAGE_SIZE, dir);
+		dma_addr[i] = 0;
+	}
+}
+
+void svm_range_dma_unmap(struct svm_range *prange)
+{
+	struct kfd_process_device *pdd;
+	dma_addr_t *dma_addr;
+	struct device *dev;
+	struct kfd_process *p;
+	uint32_t gpuidx;
+
+	p = container_of(prange->svms, struct kfd_process, svms);
+
+	for (gpuidx = 0; gpuidx < MAX_GPU_INSTANCE; gpuidx++) {
+		dma_addr = prange->dma_addr[gpuidx];
+		if (!dma_addr)
+			continue;
+
+		pdd = kfd_process_device_from_gpuidx(p, gpuidx);
+		if (!pdd) {
+			pr_debug("failed to find device idx %d\n", gpuidx);
+			continue;
+		}
+		dev = &pdd->dev->adev->pdev->dev;
+
+		svm_range_dma_unmap_dev(dev, dma_addr, 0, prange->npages);
+	}
+}
+
+static void svm_range_free(struct svm_range *prange, bool do_unmap)
+{
+	uint64_t size = (prange->last - prange->start + 1) << PAGE_SHIFT;
+	struct kfd_process *p = container_of(prange->svms, struct kfd_process, svms);
+	uint32_t gpuidx;
+
+	pr_debug("svms 0x%p prange 0x%p [0x%lx 0x%lx]\n", prange->svms, prange,
+		 prange->start, prange->last);
+
+	svm_range_vram_node_free(prange);
+	if (do_unmap)
+		svm_range_dma_unmap(prange);
+
+	if (do_unmap && !p->xnack_enabled) {
+		pr_debug("unreserve prange 0x%p size: 0x%llx\n", prange, size);
+		amdgpu_amdkfd_unreserve_mem_limit(NULL, size,
+					KFD_IOC_ALLOC_MEM_FLAGS_USERPTR, 0);
+	}
+
+	/* free dma_addr array for each gpu */
+	for (gpuidx = 0; gpuidx < MAX_GPU_INSTANCE; gpuidx++) {
+		if (prange->dma_addr[gpuidx]) {
+			kvfree(prange->dma_addr[gpuidx]);
+			prange->dma_addr[gpuidx] = NULL;
+		}
+	}
+
+	mutex_destroy(&prange->lock);
+	mutex_destroy(&prange->migrate_mutex);
+	kfree(prange);
+}
+
+static void
+svm_range_set_default_attributes(struct svm_range_list *svms, int32_t *location,
+				 int32_t *prefetch_loc, uint8_t *granularity,
+				 uint32_t *flags)
+{
+	*location = KFD_IOCTL_SVM_LOCATION_UNDEFINED;
+	*prefetch_loc = KFD_IOCTL_SVM_LOCATION_UNDEFINED;
+	*granularity = svms->default_granularity;
+	*flags =
+		KFD_IOCTL_SVM_FLAG_HOST_ACCESS | KFD_IOCTL_SVM_FLAG_COHERENT;
+}
+
+static struct
+svm_range *svm_range_new(struct svm_range_list *svms, uint64_t start,
+			 uint64_t last, bool update_mem_usage)
+{
+	uint64_t size = last - start + 1;
+	struct svm_range *prange;
+	struct kfd_process *p;
+
+	prange = kzalloc(sizeof(*prange), GFP_KERNEL);
+	if (!prange)
+		return NULL;
+
+	p = container_of(svms, struct kfd_process, svms);
+	if (!p->xnack_enabled && update_mem_usage &&
+	    amdgpu_amdkfd_reserve_mem_limit(NULL, size << PAGE_SHIFT,
+				    KFD_IOC_ALLOC_MEM_FLAGS_USERPTR, 0)) {
+		pr_info("SVM mapping failed, exceeds resident system memory limit\n");
+		kfree(prange);
+		return NULL;
+	}
+	prange->npages = size;
+	prange->svms = svms;
+	prange->start = start;
+	prange->last = last;
+	INIT_LIST_HEAD(&prange->list);
+	INIT_LIST_HEAD(&prange->update_list);
+	INIT_LIST_HEAD(&prange->svm_bo_list);
+	INIT_LIST_HEAD(&prange->deferred_list);
+	INIT_LIST_HEAD(&prange->child_list);
+	atomic_set(&prange->invalid, 0);
+	prange->validate_timestamp = 0;
+	prange->vram_pages = 0;
+	mutex_init(&prange->migrate_mutex);
+	mutex_init(&prange->lock);
+
+	if (p->xnack_enabled)
+		bitmap_copy(prange->bitmap_access, svms->bitmap_supported,
+			    MAX_GPU_INSTANCE);
+
+	svm_range_set_default_attributes(svms, &prange->preferred_loc,
+					 &prange->prefetch_loc,
+					 &prange->granularity, &prange->flags);
+
+	pr_debug("svms 0x%p [0x%llx 0x%llx]\n", svms, start, last);
+
+	return prange;
+}
+
+static bool svm_bo_ref_unless_zero(struct svm_range_bo *svm_bo)
+{
+	if (!svm_bo || !kref_get_unless_zero(&svm_bo->kref))
+		return false;
+
+	return true;
+}
+
+static void svm_range_bo_release(struct kref *kref)
+{
+	struct svm_range_bo *svm_bo;
+
+	svm_bo = container_of(kref, struct svm_range_bo, kref);
+	pr_debug("svm_bo 0x%p\n", svm_bo);
+
+	spin_lock(&svm_bo->list_lock);
+	while (!list_empty(&svm_bo->range_list)) {
+		struct svm_range *prange =
+				list_first_entry(&svm_bo->range_list,
+						struct svm_range, svm_bo_list);
+		/* list_del_init tells a concurrent svm_range_vram_node_new when
+		 * it's safe to reuse the svm_bo pointer and svm_bo_list head.
+		 */
+		list_del_init(&prange->svm_bo_list);
+		spin_unlock(&svm_bo->list_lock);
+
+		pr_debug("svms 0x%p [0x%lx 0x%lx]\n", prange->svms,
+			 prange->start, prange->last);
+		mutex_lock(&prange->lock);
+		prange->svm_bo = NULL;
+		/* prange should not hold vram page now */
+		WARN_ONCE(prange->actual_loc, "prange should not hold vram page");
+		mutex_unlock(&prange->lock);
+
+		spin_lock(&svm_bo->list_lock);
+	}
+	spin_unlock(&svm_bo->list_lock);
+
+	if (mmget_not_zero(svm_bo->eviction_fence->mm)) {
+		struct kfd_process_device *pdd;
+		struct kfd_process *p;
+		struct mm_struct *mm;
+
+		mm = svm_bo->eviction_fence->mm;
+		/*
+		 * The forked child process takes svm_bo device pages ref, svm_bo could be
+		 * released after parent process is gone.
+		 */
+		p = kfd_lookup_process_by_mm(mm);
+		if (p) {
+			pdd = kfd_get_process_device_data(svm_bo->node, p);
+			if (pdd)
+				atomic64_sub(amdgpu_bo_size(svm_bo->bo), &pdd->vram_usage);
+			kfd_unref_process(p);
+		}
+		mmput(mm);
+	}
+
+	if (!dma_fence_is_signaled(&svm_bo->eviction_fence->base))
+		/* We're not in the eviction worker. Signal the fence. */
+		dma_fence_signal(&svm_bo->eviction_fence->base);
+	dma_fence_put(&svm_bo->eviction_fence->base);
+	amdgpu_bo_unref(&svm_bo->bo);
+	kfree(svm_bo);
+}
+
+static void svm_range_bo_wq_release(struct work_struct *work)
+{
+	struct svm_range_bo *svm_bo;
+
+	svm_bo = container_of(work, struct svm_range_bo, release_work);
+	svm_range_bo_release(&svm_bo->kref);
+}
+
+static void svm_range_bo_release_async(struct kref *kref)
+{
+	struct svm_range_bo *svm_bo;
+
+	svm_bo = container_of(kref, struct svm_range_bo, kref);
+	pr_debug("svm_bo 0x%p\n", svm_bo);
+	INIT_WORK(&svm_bo->release_work, svm_range_bo_wq_release);
+	schedule_work(&svm_bo->release_work);
+}
+
+void svm_range_bo_unref_async(struct svm_range_bo *svm_bo)
+{
+	kref_put(&svm_bo->kref, svm_range_bo_release_async);
+}
+
+static void svm_range_bo_unref(struct svm_range_bo *svm_bo)
+{
+	if (svm_bo)
+		kref_put(&svm_bo->kref, svm_range_bo_release);
+}
+
+static bool
+svm_range_validate_svm_bo(struct kfd_node *node, struct svm_range *prange)
+{
+	mutex_lock(&prange->lock);
+	if (!prange->svm_bo) {
+		mutex_unlock(&prange->lock);
+		return false;
+	}
+	if (prange->ttm_res) {
+		/* We still have a reference, all is well */
+		mutex_unlock(&prange->lock);
+		return true;
+	}
+	if (svm_bo_ref_unless_zero(prange->svm_bo)) {
+		/*
+		 * Migrate from GPU to GPU, remove range from source svm_bo->node
+		 * range list, and return false to allocate svm_bo from destination
+		 * node.
+		 */
+		if (prange->svm_bo->node != node) {
+			mutex_unlock(&prange->lock);
+
+			spin_lock(&prange->svm_bo->list_lock);
+			list_del_init(&prange->svm_bo_list);
+			spin_unlock(&prange->svm_bo->list_lock);
+
+			svm_range_bo_unref(prange->svm_bo);
+			return false;
+		}
+		if (READ_ONCE(prange->svm_bo->evicting)) {
+			struct dma_fence *f;
+			struct svm_range_bo *svm_bo;
+			/* The BO is getting evicted,
+			 * we need to get a new one
+			 */
+			mutex_unlock(&prange->lock);
+			svm_bo = prange->svm_bo;
+			f = dma_fence_get(&svm_bo->eviction_fence->base);
+			svm_range_bo_unref(prange->svm_bo);
+			/* wait for the fence to avoid long spin-loop
+			 * at list_empty_careful
+			 */
+			dma_fence_wait(f, false);
+			dma_fence_put(f);
+		} else {
+			/* The BO was still around and we got
+			 * a new reference to it
+			 */
+			mutex_unlock(&prange->lock);
+			pr_debug("reuse old bo svms 0x%p [0x%lx 0x%lx]\n",
+				 prange->svms, prange->start, prange->last);
+
+			prange->ttm_res = prange->svm_bo->bo->tbo.resource;
+			return true;
+		}
+
+	} else {
+		mutex_unlock(&prange->lock);
+	}
+
+	/* We need a new svm_bo. Spin-loop to wait for concurrent
+	 * svm_range_bo_release to finish removing this range from
+	 * its range list and set prange->svm_bo to null. After this,
+	 * it is safe to reuse the svm_bo pointer and svm_bo_list head.
+	 */
+	while (!list_empty_careful(&prange->svm_bo_list) || prange->svm_bo)
+		cond_resched();
+
+	return false;
+}
+
+static struct svm_range_bo *svm_range_bo_new(void)
+{
+	struct svm_range_bo *svm_bo;
+
+	svm_bo = kzalloc(sizeof(*svm_bo), GFP_KERNEL);
+	if (!svm_bo)
+		return NULL;
+
+	kref_init(&svm_bo->kref);
+	INIT_LIST_HEAD(&svm_bo->range_list);
+	spin_lock_init(&svm_bo->list_lock);
+
+	return svm_bo;
+}
+
+int
+svm_range_vram_node_new(struct kfd_node *node, struct svm_range *prange,
+			bool clear)
+{
+	struct kfd_process_device *pdd;
+	struct amdgpu_bo_param bp;
+	struct svm_range_bo *svm_bo;
+	struct amdgpu_bo_user *ubo;
+	struct amdgpu_bo *bo;
+	struct kfd_process *p;
+	struct mm_struct *mm;
+	int r;
+
+	p = container_of(prange->svms, struct kfd_process, svms);
+	pr_debug("process pid: %d svms 0x%p [0x%lx 0x%lx]\n",
+		 p->lead_thread->pid, prange->svms,
+		 prange->start, prange->last);
+
+	if (svm_range_validate_svm_bo(node, prange))
+		return 0;
+
+	svm_bo = svm_range_bo_new();
+	if (!svm_bo) {
+		pr_debug("failed to alloc svm bo\n");
+		return -ENOMEM;
+	}
+	mm = get_task_mm(p->lead_thread);
+	if (!mm) {
+		pr_debug("failed to get mm\n");
+		kfree(svm_bo);
+		return -ESRCH;
+	}
+	svm_bo->node = node;
+	svm_bo->eviction_fence =
+		amdgpu_amdkfd_fence_create(dma_fence_context_alloc(1),
+					   mm,
+					   svm_bo);
+	mmput(mm);
+	INIT_WORK(&svm_bo->eviction_work, svm_range_evict_svm_bo_worker);
+	svm_bo->evicting = 0;
+	memset(&bp, 0, sizeof(bp));
+	bp.size = prange->npages * PAGE_SIZE;
+	bp.byte_align = PAGE_SIZE;
+	bp.domain = AMDGPU_GEM_DOMAIN_VRAM;
+	bp.flags = AMDGPU_GEM_CREATE_NO_CPU_ACCESS;
+	bp.flags |= clear ? AMDGPU_GEM_CREATE_VRAM_CLEARED : 0;
+	bp.flags |= AMDGPU_GEM_CREATE_DISCARDABLE;
+	bp.type = ttm_bo_type_device;
+	bp.resv = NULL;
+	if (node->xcp)
+		bp.xcp_id_plus1 = node->xcp->id + 1;
+
+	r = amdgpu_bo_create_user(node->adev, &bp, &ubo);
+	if (r) {
+		pr_debug("failed %d to create bo\n", r);
+		goto create_bo_failed;
+	}
+	bo = &ubo->bo;
+
+	pr_debug("alloc bo at offset 0x%lx size 0x%lx on partition %d\n",
+		 bo->tbo.resource->start << PAGE_SHIFT, bp.size,
+		 bp.xcp_id_plus1 - 1);
+
+	r = amdgpu_bo_reserve(bo, true);
+	if (r) {
+		pr_debug("failed %d to reserve bo\n", r);
+		goto reserve_bo_failed;
+	}
+
+	if (clear) {
+		r = amdgpu_bo_sync_wait(bo, AMDGPU_FENCE_OWNER_KFD, false);
+		if (r) {
+			pr_debug("failed %d to sync bo\n", r);
+			amdgpu_bo_unreserve(bo);
+			goto reserve_bo_failed;
+		}
+	}
+
+	r = dma_resv_reserve_fences(bo->tbo.base.resv, 1);
+	if (r) {
+		pr_debug("failed %d to reserve bo\n", r);
+		amdgpu_bo_unreserve(bo);
+		goto reserve_bo_failed;
+	}
+	amdgpu_bo_fence(bo, &svm_bo->eviction_fence->base, true);
+
+	amdgpu_bo_unreserve(bo);
+
+	svm_bo->bo = bo;
+	prange->svm_bo = svm_bo;
+	prange->ttm_res = bo->tbo.resource;
+	prange->offset = 0;
+
+	spin_lock(&svm_bo->list_lock);
+	list_add(&prange->svm_bo_list, &svm_bo->range_list);
+	spin_unlock(&svm_bo->list_lock);
+
+	pdd = svm_range_get_pdd_by_node(prange, node);
+	if (pdd)
+		atomic64_add(amdgpu_bo_size(bo), &pdd->vram_usage);
+
+	return 0;
+
+reserve_bo_failed:
+	amdgpu_bo_unref(&bo);
+create_bo_failed:
+	dma_fence_put(&svm_bo->eviction_fence->base);
+	kfree(svm_bo);
+	prange->ttm_res = NULL;
+
+	return r;
+}
+
+void svm_range_vram_node_free(struct svm_range *prange)
+{
+	/* serialize prange->svm_bo unref */
+	mutex_lock(&prange->lock);
+	/* prange->svm_bo has not been unref */
+	if (prange->ttm_res) {
+		prange->ttm_res = NULL;
+		mutex_unlock(&prange->lock);
+		svm_range_bo_unref(prange->svm_bo);
+	} else
+		mutex_unlock(&prange->lock);
+}
+
+struct kfd_node *
+svm_range_get_node_by_id(struct svm_range *prange, uint32_t gpu_id)
+{
+	struct kfd_process *p;
+	struct kfd_process_device *pdd;
+
+	p = container_of(prange->svms, struct kfd_process, svms);
+	pdd = kfd_process_device_data_by_id(p, gpu_id);
+	if (!pdd) {
+		pr_debug("failed to get kfd process device by id 0x%x\n", gpu_id);
+		return NULL;
+	}
+
+	return pdd->dev;
+}
+
+struct kfd_process_device *
+svm_range_get_pdd_by_node(struct svm_range *prange, struct kfd_node *node)
+{
+	struct kfd_process *p;
+
+	p = container_of(prange->svms, struct kfd_process, svms);
+
+	return kfd_get_process_device_data(node, p);
+}
+
+static int svm_range_bo_validate(void *param, struct amdgpu_bo *bo)
+{
+	struct ttm_operation_ctx ctx = { false, false };
+
+	amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_VRAM);
+
+	return ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+}
+
+static int
+svm_range_check_attr(struct kfd_process *p,
+		     uint32_t nattr, struct kfd_ioctl_svm_attribute *attrs)
+{
+	uint32_t i;
+
+	for (i = 0; i < nattr; i++) {
+		uint32_t val = attrs[i].value;
+		int gpuidx = MAX_GPU_INSTANCE;
+
+		switch (attrs[i].type) {
+		case KFD_IOCTL_SVM_ATTR_PREFERRED_LOC:
+			if (val != KFD_IOCTL_SVM_LOCATION_SYSMEM &&
+			    val != KFD_IOCTL_SVM_LOCATION_UNDEFINED)
+				gpuidx = kfd_process_gpuidx_from_gpuid(p, val);
+			break;
+		case KFD_IOCTL_SVM_ATTR_PREFETCH_LOC:
+			if (val != KFD_IOCTL_SVM_LOCATION_SYSMEM)
+				gpuidx = kfd_process_gpuidx_from_gpuid(p, val);
+			break;
+		case KFD_IOCTL_SVM_ATTR_ACCESS:
+		case KFD_IOCTL_SVM_ATTR_ACCESS_IN_PLACE:
+		case KFD_IOCTL_SVM_ATTR_NO_ACCESS:
+			gpuidx = kfd_process_gpuidx_from_gpuid(p, val);
+			break;
+		case KFD_IOCTL_SVM_ATTR_SET_FLAGS:
+			break;
+		case KFD_IOCTL_SVM_ATTR_CLR_FLAGS:
+			break;
+		case KFD_IOCTL_SVM_ATTR_GRANULARITY:
+			break;
+		default:
+			pr_debug("unknown attr type 0x%x\n", attrs[i].type);
+			return -EINVAL;
+		}
+
+		if (gpuidx < 0) {
+			pr_debug("no GPU 0x%x found\n", val);
+			return -EINVAL;
+		} else if (gpuidx < MAX_GPU_INSTANCE &&
+			   !test_bit(gpuidx, p->svms.bitmap_supported)) {
+			pr_debug("GPU 0x%x not supported\n", val);
+			return -EINVAL;
+		}
+	}
+
+	return 0;
+}
+
+static void
+svm_range_apply_attrs(struct kfd_process *p, struct svm_range *prange,
+		      uint32_t nattr, struct kfd_ioctl_svm_attribute *attrs,
+		      bool *update_mapping)
+{
+	uint32_t i;
+	int gpuidx;
+
+	for (i = 0; i < nattr; i++) {
+		switch (attrs[i].type) {
+		case KFD_IOCTL_SVM_ATTR_PREFERRED_LOC:
+			prange->preferred_loc = attrs[i].value;
+			break;
+		case KFD_IOCTL_SVM_ATTR_PREFETCH_LOC:
+			prange->prefetch_loc = attrs[i].value;
+			break;
+		case KFD_IOCTL_SVM_ATTR_ACCESS:
+		case KFD_IOCTL_SVM_ATTR_ACCESS_IN_PLACE:
+		case KFD_IOCTL_SVM_ATTR_NO_ACCESS:
+			if (!p->xnack_enabled)
+				*update_mapping = true;
+
+			gpuidx = kfd_process_gpuidx_from_gpuid(p,
+							       attrs[i].value);
+			if (attrs[i].type == KFD_IOCTL_SVM_ATTR_NO_ACCESS) {
+				bitmap_clear(prange->bitmap_access, gpuidx, 1);
+				bitmap_clear(prange->bitmap_aip, gpuidx, 1);
+			} else if (attrs[i].type == KFD_IOCTL_SVM_ATTR_ACCESS) {
+				bitmap_set(prange->bitmap_access, gpuidx, 1);
+				bitmap_clear(prange->bitmap_aip, gpuidx, 1);
+			} else {
+				bitmap_clear(prange->bitmap_access, gpuidx, 1);
+				bitmap_set(prange->bitmap_aip, gpuidx, 1);
+			}
+			break;
+		case KFD_IOCTL_SVM_ATTR_SET_FLAGS:
+			*update_mapping = true;
+			prange->flags |= attrs[i].value;
+			break;
+		case KFD_IOCTL_SVM_ATTR_CLR_FLAGS:
+			*update_mapping = true;
+			prange->flags &= ~attrs[i].value;
+			break;
+		case KFD_IOCTL_SVM_ATTR_GRANULARITY:
+			prange->granularity = min_t(uint32_t, attrs[i].value, 0x3F);
+			break;
+		default:
+			WARN_ONCE(1, "svm_range_check_attrs wasn't called?");
+		}
+	}
+}
+
+static bool
+svm_range_is_same_attrs(struct kfd_process *p, struct svm_range *prange,
+			uint32_t nattr, struct kfd_ioctl_svm_attribute *attrs)
+{
+	uint32_t i;
+	int gpuidx;
+
+	for (i = 0; i < nattr; i++) {
+		switch (attrs[i].type) {
+		case KFD_IOCTL_SVM_ATTR_PREFERRED_LOC:
+			if (prange->preferred_loc != attrs[i].value)
+				return false;
+			break;
+		case KFD_IOCTL_SVM_ATTR_PREFETCH_LOC:
+			/* Prefetch should always trigger a migration even
+			 * if the value of the attribute didn't change.
+			 */
+			return false;
+		case KFD_IOCTL_SVM_ATTR_ACCESS:
+		case KFD_IOCTL_SVM_ATTR_ACCESS_IN_PLACE:
+		case KFD_IOCTL_SVM_ATTR_NO_ACCESS:
+			gpuidx = kfd_process_gpuidx_from_gpuid(p,
+							       attrs[i].value);
+			if (attrs[i].type == KFD_IOCTL_SVM_ATTR_NO_ACCESS) {
+				if (test_bit(gpuidx, prange->bitmap_access) ||
+				    test_bit(gpuidx, prange->bitmap_aip))
+					return false;
+			} else if (attrs[i].type == KFD_IOCTL_SVM_ATTR_ACCESS) {
+				if (!test_bit(gpuidx, prange->bitmap_access))
+					return false;
+			} else {
+				if (!test_bit(gpuidx, prange->bitmap_aip))
+					return false;
+			}
+			break;
+		case KFD_IOCTL_SVM_ATTR_SET_FLAGS:
+			if ((prange->flags & attrs[i].value) != attrs[i].value)
+				return false;
+			break;
+		case KFD_IOCTL_SVM_ATTR_CLR_FLAGS:
+			if ((prange->flags & attrs[i].value) != 0)
+				return false;
+			break;
+		case KFD_IOCTL_SVM_ATTR_GRANULARITY:
+			if (prange->granularity != attrs[i].value)
+				return false;
+			break;
+		default:
+			WARN_ONCE(1, "svm_range_check_attrs wasn't called?");
+		}
+	}
+
+	return true;
+}
+
+/**
+ * svm_range_debug_dump - print all range information from svms
+ * @svms: svm range list header
+ *
+ * debug output svm range start, end, prefetch location from svms
+ * interval tree and link list
+ *
+ * Context: The caller must hold svms->lock
+ */
+static void svm_range_debug_dump(struct svm_range_list *svms)
+{
+	struct interval_tree_node *node;
+	struct svm_range *prange;
+
+	pr_debug("dump svms 0x%p list\n", svms);
+	pr_debug("range\tstart\tpage\tend\t\tlocation\n");
+
+	list_for_each_entry(prange, &svms->list, list) {
+		pr_debug("0x%p 0x%lx\t0x%llx\t0x%llx\t0x%x\n",
+			 prange, prange->start, prange->npages,
+			 prange->start + prange->npages - 1,
+			 prange->actual_loc);
+	}
+
+	pr_debug("dump svms 0x%p interval tree\n", svms);
+	pr_debug("range\tstart\tpage\tend\t\tlocation\n");
+	node = interval_tree_iter_first(&svms->objects, 0, ~0ULL);
+	while (node) {
+		prange = container_of(node, struct svm_range, it_node);
+		pr_debug("0x%p 0x%lx\t0x%llx\t0x%llx\t0x%x\n",
+			 prange, prange->start, prange->npages,
+			 prange->start + prange->npages - 1,
+			 prange->actual_loc);
+		node = interval_tree_iter_next(node, 0, ~0ULL);
+	}
+}
+
+static void *
+svm_range_copy_array(void *psrc, size_t size, uint64_t num_elements,
+		     uint64_t offset, uint64_t *vram_pages)
+{
+	unsigned char *src = (unsigned char *)psrc + offset;
+	unsigned char *dst;
+	uint64_t i;
+
+	dst = kvmalloc_array(num_elements, size, GFP_KERNEL);
+	if (!dst)
+		return NULL;
+
+	if (!vram_pages) {
+		memcpy(dst, src, num_elements * size);
+		return (void *)dst;
+	}
+
+	*vram_pages = 0;
+	for (i = 0; i < num_elements; i++) {
+		dma_addr_t *temp;
+		temp = (dma_addr_t *)dst + i;
+		*temp = *((dma_addr_t *)src + i);
+		if (*temp&SVM_RANGE_VRAM_DOMAIN)
+			(*vram_pages)++;
+	}
+
+	return (void *)dst;
+}
+
+static int
+svm_range_copy_dma_addrs(struct svm_range *dst, struct svm_range *src)
+{
+	int i;
+
+	for (i = 0; i < MAX_GPU_INSTANCE; i++) {
+		if (!src->dma_addr[i])
+			continue;
+		dst->dma_addr[i] = svm_range_copy_array(src->dma_addr[i],
+					sizeof(*src->dma_addr[i]), src->npages, 0, NULL);
+		if (!dst->dma_addr[i])
+			return -ENOMEM;
+	}
+
+	return 0;
+}
+
+static int
+svm_range_split_array(void *ppnew, void *ppold, size_t size,
+		      uint64_t old_start, uint64_t old_n,
+		      uint64_t new_start, uint64_t new_n, uint64_t *new_vram_pages)
+{
+	unsigned char *new, *old, *pold;
+	uint64_t d;
+
+	if (!ppold)
+		return 0;
+	pold = *(unsigned char **)ppold;
+	if (!pold)
+		return 0;
+
+	d = (new_start - old_start) * size;
+	/* get dma addr array for new range and calculte its vram page number */
+	new = svm_range_copy_array(pold, size, new_n, d, new_vram_pages);
+	if (!new)
+		return -ENOMEM;
+	d = (new_start == old_start) ? new_n * size : 0;
+	old = svm_range_copy_array(pold, size, old_n, d, NULL);
+	if (!old) {
+		kvfree(new);
+		return -ENOMEM;
+	}
+	kvfree(pold);
+	*(void **)ppold = old;
+	*(void **)ppnew = new;
+
+	return 0;
+}
+
+static int
+svm_range_split_pages(struct svm_range *new, struct svm_range *old,
+		      uint64_t start, uint64_t last)
+{
+	uint64_t npages = last - start + 1;
+	int i, r;
+
+	for (i = 0; i < MAX_GPU_INSTANCE; i++) {
+		r = svm_range_split_array(&new->dma_addr[i], &old->dma_addr[i],
+					  sizeof(*old->dma_addr[i]), old->start,
+					  npages, new->start, new->npages,
+					  old->actual_loc ? &new->vram_pages : NULL);
+		if (r)
+			return r;
+	}
+	if (old->actual_loc)
+		old->vram_pages -= new->vram_pages;
+
+	return 0;
+}
+
+static int
+svm_range_split_nodes(struct svm_range *new, struct svm_range *old,
+		      uint64_t start, uint64_t last)
+{
+	uint64_t npages = last - start + 1;
+
+	pr_debug("svms 0x%p new prange 0x%p start 0x%lx [0x%llx 0x%llx]\n",
+		 new->svms, new, new->start, start, last);
+
+	if (new->start == old->start) {
+		new->offset = old->offset;
+		old->offset += new->npages;
+	} else {
+		new->offset = old->offset + npages;
+	}
+
+	new->svm_bo = svm_range_bo_ref(old->svm_bo);
+	new->ttm_res = old->ttm_res;
+
+	spin_lock(&new->svm_bo->list_lock);
+	list_add(&new->svm_bo_list, &new->svm_bo->range_list);
+	spin_unlock(&new->svm_bo->list_lock);
+
+	return 0;
+}
+
+/**
+ * svm_range_split_adjust - split range and adjust
+ *
+ * @new: new range
+ * @old: the old range
+ * @start: the old range adjust to start address in pages
+ * @last: the old range adjust to last address in pages
+ *
+ * Copy system memory dma_addr or vram ttm_res in old range to new
+ * range from new_start up to size new->npages, the remaining old range is from
+ * start to last
+ *
+ * Return:
+ * 0 - OK, -ENOMEM - out of memory
+ */
+static int
+svm_range_split_adjust(struct svm_range *new, struct svm_range *old,
+		      uint64_t start, uint64_t last)
+{
+	int r;
+
+	pr_debug("svms 0x%p new 0x%lx old [0x%lx 0x%lx] => [0x%llx 0x%llx]\n",
+		 new->svms, new->start, old->start, old->last, start, last);
+
+	if (new->start < old->start ||
+	    new->last > old->last) {
+		WARN_ONCE(1, "invalid new range start or last\n");
+		return -EINVAL;
+	}
+
+	r = svm_range_split_pages(new, old, start, last);
+	if (r)
+		return r;
+
+	if (old->actual_loc && old->ttm_res) {
+		r = svm_range_split_nodes(new, old, start, last);
+		if (r)
+			return r;
+	}
+
+	old->npages = last - start + 1;
+	old->start = start;
+	old->last = last;
+	new->flags = old->flags;
+	new->preferred_loc = old->preferred_loc;
+	new->prefetch_loc = old->prefetch_loc;
+	new->actual_loc = old->actual_loc;
+	new->granularity = old->granularity;
+	new->mapped_to_gpu = old->mapped_to_gpu;
+	bitmap_copy(new->bitmap_access, old->bitmap_access, MAX_GPU_INSTANCE);
+	bitmap_copy(new->bitmap_aip, old->bitmap_aip, MAX_GPU_INSTANCE);
+	atomic_set(&new->queue_refcount, atomic_read(&old->queue_refcount));
+
+	return 0;
+}
+
+/**
+ * svm_range_split - split a range in 2 ranges
+ *
+ * @prange: the svm range to split
+ * @start: the remaining range start address in pages
+ * @last: the remaining range last address in pages
+ * @new: the result new range generated
+ *
+ * Two cases only:
+ * case 1: if start == prange->start
+ *         prange ==> prange[start, last]
+ *         new range [last + 1, prange->last]
+ *
+ * case 2: if last == prange->last
+ *         prange ==> prange[start, last]
+ *         new range [prange->start, start - 1]
+ *
+ * Return:
+ * 0 - OK, -ENOMEM - out of memory, -EINVAL - invalid start, last
+ */
+static int
+svm_range_split(struct svm_range *prange, uint64_t start, uint64_t last,
+		struct svm_range **new)
+{
+	uint64_t old_start = prange->start;
+	uint64_t old_last = prange->last;
+	struct svm_range_list *svms;
+	int r = 0;
+
+	pr_debug("svms 0x%p [0x%llx 0x%llx] to [0x%llx 0x%llx]\n", prange->svms,
+		 old_start, old_last, start, last);
+
+	if (old_start != start && old_last != last)
+		return -EINVAL;
+	if (start < old_start || last > old_last)
+		return -EINVAL;
+
+	svms = prange->svms;
+	if (old_start == start)
+		*new = svm_range_new(svms, last + 1, old_last, false);
+	else
+		*new = svm_range_new(svms, old_start, start - 1, false);
+	if (!*new)
+		return -ENOMEM;
+
+	r = svm_range_split_adjust(*new, prange, start, last);
+	if (r) {
+		pr_debug("failed %d split [0x%llx 0x%llx] to [0x%llx 0x%llx]\n",
+			 r, old_start, old_last, start, last);
+		svm_range_free(*new, false);
+		*new = NULL;
+	}
+
+	return r;
+}
+
+static int
+svm_range_split_tail(struct svm_range *prange, uint64_t new_last,
+		     struct list_head *insert_list, struct list_head *remap_list)
+{
+	struct svm_range *tail = NULL;
+	int r = svm_range_split(prange, prange->start, new_last, &tail);
+
+	if (!r) {
+		list_add(&tail->list, insert_list);
+		if (!IS_ALIGNED(new_last + 1, 1UL << prange->granularity))
+			list_add(&tail->update_list, remap_list);
+	}
+	return r;
+}
+
+static int
+svm_range_split_head(struct svm_range *prange, uint64_t new_start,
+		     struct list_head *insert_list, struct list_head *remap_list)
+{
+	struct svm_range *head = NULL;
+	int r = svm_range_split(prange, new_start, prange->last, &head);
+
+	if (!r) {
+		list_add(&head->list, insert_list);
+		if (!IS_ALIGNED(new_start, 1UL << prange->granularity))
+			list_add(&head->update_list, remap_list);
+	}
+	return r;
+}
+
+static void
+svm_range_add_child(struct svm_range *prange, struct svm_range *pchild, enum svm_work_list_ops op)
+{
+	pr_debug("add child 0x%p [0x%lx 0x%lx] to prange 0x%p child list %d\n",
+		 pchild, pchild->start, pchild->last, prange, op);
+
+	pchild->work_item.mm = NULL;
+	pchild->work_item.op = op;
+	list_add_tail(&pchild->child_list, &prange->child_list);
+}
+
+static bool
+svm_nodes_in_same_hive(struct kfd_node *node_a, struct kfd_node *node_b)
+{
+	return (node_a->adev == node_b->adev ||
+		amdgpu_xgmi_same_hive(node_a->adev, node_b->adev));
+}
+
+static uint64_t
+svm_range_get_pte_flags(struct kfd_node *node, struct amdgpu_vm *vm,
+			struct svm_range *prange, int domain)
+{
+	struct kfd_node *bo_node;
+	uint32_t flags = prange->flags;
+	uint32_t mapping_flags = 0;
+	uint32_t gc_ip_version = KFD_GC_VERSION(node);
+	uint64_t pte_flags;
+	bool snoop = (domain != SVM_RANGE_VRAM_DOMAIN);
+	bool coherent = flags & (KFD_IOCTL_SVM_FLAG_COHERENT | KFD_IOCTL_SVM_FLAG_EXT_COHERENT);
+	bool ext_coherent = flags & KFD_IOCTL_SVM_FLAG_EXT_COHERENT;
+	unsigned int mtype_local;
+
+	if (domain == SVM_RANGE_VRAM_DOMAIN)
+		bo_node = prange->svm_bo->node;
+
+	switch (gc_ip_version) {
+	case IP_VERSION(9, 4, 1):
+		if (domain == SVM_RANGE_VRAM_DOMAIN) {
+			if (bo_node == node) {
+				mapping_flags |= coherent ?
+					AMDGPU_VM_MTYPE_CC : AMDGPU_VM_MTYPE_RW;
+			} else {
+				mapping_flags |= coherent ?
+					AMDGPU_VM_MTYPE_UC : AMDGPU_VM_MTYPE_NC;
+				if (svm_nodes_in_same_hive(node, bo_node))
+					snoop = true;
+			}
+		} else {
+			mapping_flags |= coherent ?
+				AMDGPU_VM_MTYPE_UC : AMDGPU_VM_MTYPE_NC;
+		}
+		break;
+	case IP_VERSION(9, 4, 2):
+		if (domain == SVM_RANGE_VRAM_DOMAIN) {
+			if (bo_node == node) {
+				mapping_flags |= coherent ?
+					AMDGPU_VM_MTYPE_CC : AMDGPU_VM_MTYPE_RW;
+				if (node->adev->gmc.xgmi.connected_to_cpu)
+					snoop = true;
+			} else {
+				mapping_flags |= coherent ?
+					AMDGPU_VM_MTYPE_UC : AMDGPU_VM_MTYPE_NC;
+				if (svm_nodes_in_same_hive(node, bo_node))
+					snoop = true;
+			}
+		} else {
+			mapping_flags |= coherent ?
+				AMDGPU_VM_MTYPE_UC : AMDGPU_VM_MTYPE_NC;
+		}
+		break;
+	case IP_VERSION(9, 4, 3):
+	case IP_VERSION(9, 4, 4):
+	case IP_VERSION(9, 5, 0):
+		if (ext_coherent)
+			mtype_local = AMDGPU_VM_MTYPE_CC;
+		else
+			mtype_local = amdgpu_mtype_local == 1 ? AMDGPU_VM_MTYPE_NC :
+				amdgpu_mtype_local == 2 ? AMDGPU_VM_MTYPE_CC : AMDGPU_VM_MTYPE_RW;
+		snoop = true;
+		if (domain == SVM_RANGE_VRAM_DOMAIN) {
+			/* local HBM region close to partition */
+			if (bo_node->adev == node->adev &&
+			    (!bo_node->xcp || !node->xcp || bo_node->xcp->mem_id == node->xcp->mem_id))
+				mapping_flags |= mtype_local;
+			/* local HBM region far from partition or remote XGMI GPU
+			 * with regular system scope coherence
+			 */
+			else if (svm_nodes_in_same_hive(bo_node, node) && !ext_coherent)
+				mapping_flags |= AMDGPU_VM_MTYPE_NC;
+			/* PCIe P2P on GPUs pre-9.5.0 */
+			else if (gc_ip_version < IP_VERSION(9, 5, 0) &&
+				 !svm_nodes_in_same_hive(bo_node, node))
+				mapping_flags |= AMDGPU_VM_MTYPE_UC;
+			/* Other remote memory */
+			else
+				mapping_flags |= ext_coherent ? AMDGPU_VM_MTYPE_UC : AMDGPU_VM_MTYPE_NC;
+		/* system memory accessed by the APU */
+		} else if (node->adev->flags & AMD_IS_APU) {
+			/* On NUMA systems, locality is determined per-page
+			 * in amdgpu_gmc_override_vm_pte_flags
+			 */
+			if (num_possible_nodes() <= 1)
+				mapping_flags |= mtype_local;
+			else
+				mapping_flags |= ext_coherent ? AMDGPU_VM_MTYPE_UC : AMDGPU_VM_MTYPE_NC;
+		/* system memory accessed by the dGPU */
+		} else {
+			if (gc_ip_version < IP_VERSION(9, 5, 0) || ext_coherent)
+				mapping_flags |= AMDGPU_VM_MTYPE_UC;
+			else
+				mapping_flags |= AMDGPU_VM_MTYPE_NC;
+		}
+		break;
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 0, 1):
+		mapping_flags |= AMDGPU_VM_MTYPE_NC;
+		break;
+	default:
+		mapping_flags |= coherent ?
+			AMDGPU_VM_MTYPE_UC : AMDGPU_VM_MTYPE_NC;
+	}
+
+	if (flags & KFD_IOCTL_SVM_FLAG_GPU_EXEC)
+		mapping_flags |= AMDGPU_VM_PAGE_EXECUTABLE;
+
+	pte_flags = AMDGPU_PTE_VALID;
+	pte_flags |= (domain == SVM_RANGE_VRAM_DOMAIN) ? 0 : AMDGPU_PTE_SYSTEM;
+	pte_flags |= snoop ? AMDGPU_PTE_SNOOPED : 0;
+	if (gc_ip_version >= IP_VERSION(12, 0, 0))
+		pte_flags |= AMDGPU_PTE_IS_PTE;
+
+	amdgpu_gmc_get_vm_pte(node->adev, vm, NULL, mapping_flags, &pte_flags);
+	pte_flags |= AMDGPU_PTE_READABLE;
+	if (!(flags & KFD_IOCTL_SVM_FLAG_GPU_RO))
+		pte_flags |= AMDGPU_PTE_WRITEABLE;
+	return pte_flags;
+}
+
+static int
+svm_range_unmap_from_gpu(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			 uint64_t start, uint64_t last,
+			 struct dma_fence **fence)
+{
+	uint64_t init_pte_value = 0;
+
+	pr_debug("[0x%llx 0x%llx]\n", start, last);
+
+	return amdgpu_vm_update_range(adev, vm, false, true, true, false, NULL, start,
+				      last, init_pte_value, 0, 0, NULL, NULL,
+				      fence);
+}
+
+static int
+svm_range_unmap_from_gpus(struct svm_range *prange, unsigned long start,
+			  unsigned long last, uint32_t trigger)
+{
+	DECLARE_BITMAP(bitmap, MAX_GPU_INSTANCE);
+	struct kfd_process_device *pdd;
+	struct dma_fence *fence = NULL;
+	struct kfd_process *p;
+	uint32_t gpuidx;
+	int r = 0;
+
+	if (!prange->mapped_to_gpu) {
+		pr_debug("prange 0x%p [0x%lx 0x%lx] not mapped to GPU\n",
+			 prange, prange->start, prange->last);
+		return 0;
+	}
+
+	if (prange->start == start && prange->last == last) {
+		pr_debug("unmap svms 0x%p prange 0x%p\n", prange->svms, prange);
+		prange->mapped_to_gpu = false;
+	}
+
+	bitmap_or(bitmap, prange->bitmap_access, prange->bitmap_aip,
+		  MAX_GPU_INSTANCE);
+	p = container_of(prange->svms, struct kfd_process, svms);
+
+	for_each_set_bit(gpuidx, bitmap, MAX_GPU_INSTANCE) {
+		pr_debug("unmap from gpu idx 0x%x\n", gpuidx);
+		pdd = kfd_process_device_from_gpuidx(p, gpuidx);
+		if (!pdd) {
+			pr_debug("failed to find device idx %d\n", gpuidx);
+			return -EINVAL;
+		}
+
+		kfd_smi_event_unmap_from_gpu(pdd->dev, p->lead_thread->pid,
+					     start, last, trigger);
+
+		r = svm_range_unmap_from_gpu(pdd->dev->adev,
+					     drm_priv_to_vm(pdd->drm_priv),
+					     start, last, &fence);
+		if (r)
+			break;
+
+		if (fence) {
+			r = dma_fence_wait(fence, false);
+			dma_fence_put(fence);
+			fence = NULL;
+			if (r)
+				break;
+		}
+		kfd_flush_tlb(pdd, TLB_FLUSH_HEAVYWEIGHT);
+	}
+
+	return r;
+}
+
+static int
+svm_range_map_to_gpu(struct kfd_process_device *pdd, struct svm_range *prange,
+		     unsigned long offset, unsigned long npages, bool readonly,
+		     dma_addr_t *dma_addr, struct amdgpu_device *bo_adev,
+		     struct dma_fence **fence, bool flush_tlb)
+{
+	struct amdgpu_device *adev = pdd->dev->adev;
+	struct amdgpu_vm *vm = drm_priv_to_vm(pdd->drm_priv);
+	uint64_t pte_flags;
+	unsigned long last_start;
+	int last_domain;
+	int r = 0;
+	int64_t i, j;
+
+	last_start = prange->start + offset;
+
+	pr_debug("svms 0x%p [0x%lx 0x%lx] readonly %d\n", prange->svms,
+		 last_start, last_start + npages - 1, readonly);
+
+	for (i = offset; i < offset + npages; i++) {
+		last_domain = dma_addr[i] & SVM_RANGE_VRAM_DOMAIN;
+		dma_addr[i] &= ~SVM_RANGE_VRAM_DOMAIN;
+
+		/* Collect all pages in the same address range and memory domain
+		 * that can be mapped with a single call to update mapping.
+		 */
+		if (i < offset + npages - 1 &&
+		    last_domain == (dma_addr[i + 1] & SVM_RANGE_VRAM_DOMAIN))
+			continue;
+
+		pr_debug("Mapping range [0x%lx 0x%llx] on domain: %s\n",
+			 last_start, prange->start + i, last_domain ? "GPU" : "CPU");
+
+		pte_flags = svm_range_get_pte_flags(pdd->dev, vm, prange, last_domain);
+		if (readonly)
+			pte_flags &= ~AMDGPU_PTE_WRITEABLE;
+
+		pr_debug("svms 0x%p map [0x%lx 0x%llx] vram %d PTE 0x%llx\n",
+			 prange->svms, last_start, prange->start + i,
+			 (last_domain == SVM_RANGE_VRAM_DOMAIN) ? 1 : 0,
+			 pte_flags);
+
+		/* For dGPU mode, we use same vm_manager to allocate VRAM for
+		 * different memory partition based on fpfn/lpfn, we should use
+		 * same vm_manager.vram_base_offset regardless memory partition.
+		 */
+		r = amdgpu_vm_update_range(adev, vm, false, false, flush_tlb, true,
+					   NULL, last_start, prange->start + i,
+					   pte_flags,
+					   (last_start - prange->start) << PAGE_SHIFT,
+					   bo_adev ? bo_adev->vm_manager.vram_base_offset : 0,
+					   NULL, dma_addr, &vm->last_update);
+
+		for (j = last_start - prange->start; j <= i; j++)
+			dma_addr[j] |= last_domain;
+
+		if (r) {
+			pr_debug("failed %d to map to gpu 0x%lx\n", r, prange->start);
+			goto out;
+		}
+		last_start = prange->start + i + 1;
+	}
+
+	r = amdgpu_vm_update_pdes(adev, vm, false);
+	if (r) {
+		pr_debug("failed %d to update directories 0x%lx\n", r,
+			 prange->start);
+		goto out;
+	}
+
+	if (fence)
+		*fence = dma_fence_get(vm->last_update);
+
+out:
+	return r;
+}
+
+static int
+svm_range_map_to_gpus(struct svm_range *prange, unsigned long offset,
+		      unsigned long npages, bool readonly,
+		      unsigned long *bitmap, bool wait, bool flush_tlb)
+{
+	struct kfd_process_device *pdd;
+	struct amdgpu_device *bo_adev = NULL;
+	struct kfd_process *p;
+	struct dma_fence *fence = NULL;
+	uint32_t gpuidx;
+	int r = 0;
+
+	if (prange->svm_bo && prange->ttm_res)
+		bo_adev = prange->svm_bo->node->adev;
+
+	p = container_of(prange->svms, struct kfd_process, svms);
+	for_each_set_bit(gpuidx, bitmap, MAX_GPU_INSTANCE) {
+		pr_debug("mapping to gpu idx 0x%x\n", gpuidx);
+		pdd = kfd_process_device_from_gpuidx(p, gpuidx);
+		if (!pdd) {
+			pr_debug("failed to find device idx %d\n", gpuidx);
+			return -EINVAL;
+		}
+
+		pdd = kfd_bind_process_to_device(pdd->dev, p);
+		if (IS_ERR(pdd))
+			return -EINVAL;
+
+		if (bo_adev && pdd->dev->adev != bo_adev &&
+		    !amdgpu_xgmi_same_hive(pdd->dev->adev, bo_adev)) {
+			pr_debug("cannot map to device idx %d\n", gpuidx);
+			continue;
+		}
+
+		r = svm_range_map_to_gpu(pdd, prange, offset, npages, readonly,
+					 prange->dma_addr[gpuidx],
+					 bo_adev, wait ? &fence : NULL,
+					 flush_tlb);
+		if (r)
+			break;
+
+		if (fence) {
+			r = dma_fence_wait(fence, false);
+			dma_fence_put(fence);
+			fence = NULL;
+			if (r) {
+				pr_debug("failed %d to dma fence wait\n", r);
+				break;
+			}
+		}
+
+		kfd_flush_tlb(pdd, TLB_FLUSH_LEGACY);
+	}
+
+	return r;
+}
+
+struct svm_validate_context {
+	struct kfd_process *process;
+	struct svm_range *prange;
+	bool intr;
+	DECLARE_BITMAP(bitmap, MAX_GPU_INSTANCE);
+	struct drm_exec exec;
+};
+
+static int svm_range_reserve_bos(struct svm_validate_context *ctx, bool intr)
+{
+	struct kfd_process_device *pdd;
+	struct amdgpu_vm *vm;
+	uint32_t gpuidx;
+	int r;
+
+	drm_exec_init(&ctx->exec, intr ? DRM_EXEC_INTERRUPTIBLE_WAIT: 0, 0);
+	drm_exec_until_all_locked(&ctx->exec) {
+		for_each_set_bit(gpuidx, ctx->bitmap, MAX_GPU_INSTANCE) {
+			pdd = kfd_process_device_from_gpuidx(ctx->process, gpuidx);
+			if (!pdd) {
+				pr_debug("failed to find device idx %d\n", gpuidx);
+				r = -EINVAL;
+				goto unreserve_out;
+			}
+			vm = drm_priv_to_vm(pdd->drm_priv);
+
+			r = amdgpu_vm_lock_pd(vm, &ctx->exec, 2);
+			drm_exec_retry_on_contention(&ctx->exec);
+			if (unlikely(r)) {
+				pr_debug("failed %d to reserve bo\n", r);
+				goto unreserve_out;
+			}
+		}
+	}
+
+	for_each_set_bit(gpuidx, ctx->bitmap, MAX_GPU_INSTANCE) {
+		pdd = kfd_process_device_from_gpuidx(ctx->process, gpuidx);
+		if (!pdd) {
+			pr_debug("failed to find device idx %d\n", gpuidx);
+			r = -EINVAL;
+			goto unreserve_out;
+		}
+
+		r = amdgpu_vm_validate(pdd->dev->adev,
+				       drm_priv_to_vm(pdd->drm_priv), NULL,
+				       svm_range_bo_validate, NULL);
+		if (r) {
+			pr_debug("failed %d validate pt bos\n", r);
+			goto unreserve_out;
+		}
+	}
+
+	return 0;
+
+unreserve_out:
+	drm_exec_fini(&ctx->exec);
+	return r;
+}
+
+static void svm_range_unreserve_bos(struct svm_validate_context *ctx)
+{
+	drm_exec_fini(&ctx->exec);
+}
+
+static void *kfd_svm_page_owner(struct kfd_process *p, int32_t gpuidx)
+{
+	struct kfd_process_device *pdd;
+
+	pdd = kfd_process_device_from_gpuidx(p, gpuidx);
+	if (!pdd)
+		return NULL;
+
+	return SVM_ADEV_PGMAP_OWNER(pdd->dev->adev);
+}
+
+/*
+ * Validation+GPU mapping with concurrent invalidation (MMU notifiers)
+ *
+ * To prevent concurrent destruction or change of range attributes, the
+ * svm_read_lock must be held. The caller must not hold the svm_write_lock
+ * because that would block concurrent evictions and lead to deadlocks. To
+ * serialize concurrent migrations or validations of the same range, the
+ * prange->migrate_mutex must be held.
+ *
+ * For VRAM ranges, the SVM BO must be allocated and valid (protected by its
+ * eviction fence.
+ *
+ * The following sequence ensures race-free validation and GPU mapping:
+ *
+ * 1. Reserve page table (and SVM BO if range is in VRAM)
+ * 2. hmm_range_fault to get page addresses (if system memory)
+ * 3. DMA-map pages (if system memory)
+ * 4-a. Take notifier lock
+ * 4-b. Check that pages still valid (mmu_interval_read_retry)
+ * 4-c. Check that the range was not split or otherwise invalidated
+ * 4-d. Update GPU page table
+ * 4.e. Release notifier lock
+ * 5. Release page table (and SVM BO) reservation
+ */
+static int svm_range_validate_and_map(struct mm_struct *mm,
+				      unsigned long map_start, unsigned long map_last,
+				      struct svm_range *prange, int32_t gpuidx,
+				      bool intr, bool wait, bool flush_tlb)
+{
+	struct svm_validate_context *ctx;
+	unsigned long start, end, addr;
+	struct kfd_process *p;
+	void *owner;
+	int32_t idx;
+	int r = 0;
+
+	ctx = kzalloc(sizeof(struct svm_validate_context), GFP_KERNEL);
+	if (!ctx)
+		return -ENOMEM;
+	ctx->process = container_of(prange->svms, struct kfd_process, svms);
+	ctx->prange = prange;
+	ctx->intr = intr;
+
+	if (gpuidx < MAX_GPU_INSTANCE) {
+		bitmap_zero(ctx->bitmap, MAX_GPU_INSTANCE);
+		bitmap_set(ctx->bitmap, gpuidx, 1);
+	} else if (ctx->process->xnack_enabled) {
+		bitmap_copy(ctx->bitmap, prange->bitmap_aip, MAX_GPU_INSTANCE);
+
+		/* If prefetch range to GPU, or GPU retry fault migrate range to
+		 * GPU, which has ACCESS attribute to the range, create mapping
+		 * on that GPU.
+		 */
+		if (prange->actual_loc) {
+			gpuidx = kfd_process_gpuidx_from_gpuid(ctx->process,
+							prange->actual_loc);
+			if (gpuidx < 0) {
+				WARN_ONCE(1, "failed get device by id 0x%x\n",
+					 prange->actual_loc);
+				r = -EINVAL;
+				goto free_ctx;
+			}
+			if (test_bit(gpuidx, prange->bitmap_access))
+				bitmap_set(ctx->bitmap, gpuidx, 1);
+		}
+
+		/*
+		 * If prange is already mapped or with always mapped flag,
+		 * update mapping on GPUs with ACCESS attribute
+		 */
+		if (bitmap_empty(ctx->bitmap, MAX_GPU_INSTANCE)) {
+			if (prange->mapped_to_gpu ||
+			    prange->flags & KFD_IOCTL_SVM_FLAG_GPU_ALWAYS_MAPPED)
+				bitmap_copy(ctx->bitmap, prange->bitmap_access, MAX_GPU_INSTANCE);
+		}
+	} else {
+		bitmap_or(ctx->bitmap, prange->bitmap_access,
+			  prange->bitmap_aip, MAX_GPU_INSTANCE);
+	}
+
+	if (bitmap_empty(ctx->bitmap, MAX_GPU_INSTANCE)) {
+		r = 0;
+		goto free_ctx;
+	}
+
+	if (prange->actual_loc && !prange->ttm_res) {
+		/* This should never happen. actual_loc gets set by
+		 * svm_migrate_ram_to_vram after allocating a BO.
+		 */
+		WARN_ONCE(1, "VRAM BO missing during validation\n");
+		r = -EINVAL;
+		goto free_ctx;
+	}
+
+	r = svm_range_reserve_bos(ctx, intr);
+	if (r)
+		goto free_ctx;
+
+	p = container_of(prange->svms, struct kfd_process, svms);
+	owner = kfd_svm_page_owner(p, find_first_bit(ctx->bitmap,
+						MAX_GPU_INSTANCE));
+	for_each_set_bit(idx, ctx->bitmap, MAX_GPU_INSTANCE) {
+		if (kfd_svm_page_owner(p, idx) != owner) {
+			owner = NULL;
+			break;
+		}
+	}
+
+	start = map_start << PAGE_SHIFT;
+	end = (map_last + 1) << PAGE_SHIFT;
+	for (addr = start; !r && addr < end; ) {
+		struct hmm_range *hmm_range = NULL;
+		unsigned long map_start_vma;
+		unsigned long map_last_vma;
+		struct vm_area_struct *vma;
+		unsigned long next = 0;
+		unsigned long offset;
+		unsigned long npages;
+		bool readonly;
+
+		vma = vma_lookup(mm, addr);
+		if (vma) {
+			readonly = !(vma->vm_flags & VM_WRITE);
+
+			next = min(vma->vm_end, end);
+			npages = (next - addr) >> PAGE_SHIFT;
+			/* HMM requires at least READ permissions. If provided with PROT_NONE,
+			 * unmap the memory. If it's not already mapped, this is a no-op
+			 * If PROT_WRITE is provided without READ, warn first then unmap
+			 */
+			if (!(vma->vm_flags & VM_READ)) {
+				unsigned long e, s;
+
+				svm_range_lock(prange);
+				if (vma->vm_flags & VM_WRITE)
+					pr_debug("VM_WRITE without VM_READ is not supported");
+				s = max(start, prange->start);
+				e = min(end, prange->last);
+				if (e >= s)
+					r = svm_range_unmap_from_gpus(prange, s, e,
+						       KFD_SVM_UNMAP_TRIGGER_UNMAP_FROM_CPU);
+				svm_range_unlock(prange);
+				/* If unmap returns non-zero, we'll bail on the next for loop
+				 * iteration, so just leave r and continue
+				 */
+				addr = next;
+				continue;
+			}
+
+			WRITE_ONCE(p->svms.faulting_task, current);
+			r = amdgpu_hmm_range_get_pages(&prange->notifier, addr, npages,
+						       readonly, owner,
+						       &hmm_range);
+			WRITE_ONCE(p->svms.faulting_task, NULL);
+			if (r)
+				pr_debug("failed %d to get svm range pages\n", r);
+		} else {
+			r = -EFAULT;
+		}
+
+		if (!r) {
+			offset = (addr >> PAGE_SHIFT) - prange->start;
+			r = svm_range_dma_map(prange, ctx->bitmap, offset, npages,
+					      hmm_range->hmm_pfns);
+			if (r)
+				pr_debug("failed %d to dma map range\n", r);
+		}
+
+		svm_range_lock(prange);
+
+		/* Free backing memory of hmm_range if it was initialized
+		 * Overrride return value to TRY AGAIN only if prior returns
+		 * were successful
+		 */
+		if (hmm_range && amdgpu_hmm_range_get_pages_done(hmm_range) && !r) {
+			pr_debug("hmm update the range, need validate again\n");
+			r = -EAGAIN;
+		}
+
+		if (!r && !list_empty(&prange->child_list)) {
+			pr_debug("range split by unmap in parallel, validate again\n");
+			r = -EAGAIN;
+		}
+
+		if (!r) {
+			map_start_vma = max(map_start, prange->start + offset);
+			map_last_vma = min(map_last, prange->start + offset + npages - 1);
+			if (map_start_vma <= map_last_vma) {
+				offset = map_start_vma - prange->start;
+				npages = map_last_vma - map_start_vma + 1;
+				r = svm_range_map_to_gpus(prange, offset, npages, readonly,
+							  ctx->bitmap, wait, flush_tlb);
+			}
+		}
+
+		if (!r && next == end)
+			prange->mapped_to_gpu = true;
+
+		svm_range_unlock(prange);
+
+		addr = next;
+	}
+
+	svm_range_unreserve_bos(ctx);
+	if (!r)
+		prange->validate_timestamp = ktime_get_boottime();
+
+free_ctx:
+	kfree(ctx);
+
+	return r;
+}
+
+/**
+ * svm_range_list_lock_and_flush_work - flush pending deferred work
+ *
+ * @svms: the svm range list
+ * @mm: the mm structure
+ *
+ * Context: Returns with mmap write lock held, pending deferred work flushed
+ *
+ */
+void
+svm_range_list_lock_and_flush_work(struct svm_range_list *svms,
+				   struct mm_struct *mm)
+{
+retry_flush_work:
+	flush_work(&svms->deferred_list_work);
+	mmap_write_lock(mm);
+
+	if (list_empty(&svms->deferred_range_list))
+		return;
+	mmap_write_unlock(mm);
+	pr_debug("retry flush\n");
+	goto retry_flush_work;
+}
+
+static void svm_range_restore_work(struct work_struct *work)
+{
+	struct delayed_work *dwork = to_delayed_work(work);
+	struct amdkfd_process_info *process_info;
+	struct svm_range_list *svms;
+	struct svm_range *prange;
+	struct kfd_process *p;
+	struct mm_struct *mm;
+	int evicted_ranges;
+	int invalid;
+	int r;
+
+	svms = container_of(dwork, struct svm_range_list, restore_work);
+	evicted_ranges = atomic_read(&svms->evicted_ranges);
+	if (!evicted_ranges)
+		return;
+
+	pr_debug("restore svm ranges\n");
+
+	p = container_of(svms, struct kfd_process, svms);
+	process_info = p->kgd_process_info;
+
+	/* Keep mm reference when svm_range_validate_and_map ranges */
+	mm = get_task_mm(p->lead_thread);
+	if (!mm) {
+		pr_debug("svms 0x%p process mm gone\n", svms);
+		return;
+	}
+
+	mutex_lock(&process_info->lock);
+	svm_range_list_lock_and_flush_work(svms, mm);
+	mutex_lock(&svms->lock);
+
+	evicted_ranges = atomic_read(&svms->evicted_ranges);
+
+	list_for_each_entry(prange, &svms->list, list) {
+		invalid = atomic_read(&prange->invalid);
+		if (!invalid)
+			continue;
+
+		pr_debug("restoring svms 0x%p prange 0x%p [0x%lx %lx] inv %d\n",
+			 prange->svms, prange, prange->start, prange->last,
+			 invalid);
+
+		/*
+		 * If range is migrating, wait for migration is done.
+		 */
+		mutex_lock(&prange->migrate_mutex);
+
+		r = svm_range_validate_and_map(mm, prange->start, prange->last, prange,
+					       MAX_GPU_INSTANCE, false, true, false);
+		if (r)
+			pr_debug("failed %d to map 0x%lx to gpus\n", r,
+				 prange->start);
+
+		mutex_unlock(&prange->migrate_mutex);
+		if (r)
+			goto out_reschedule;
+
+		if (atomic_cmpxchg(&prange->invalid, invalid, 0) != invalid)
+			goto out_reschedule;
+	}
+
+	if (atomic_cmpxchg(&svms->evicted_ranges, evicted_ranges, 0) !=
+	    evicted_ranges)
+		goto out_reschedule;
+
+	evicted_ranges = 0;
+
+	r = kgd2kfd_resume_mm(mm);
+	if (r) {
+		/* No recovery from this failure. Probably the CP is
+		 * hanging. No point trying again.
+		 */
+		pr_debug("failed %d to resume KFD\n", r);
+	}
+
+	pr_debug("restore svm ranges successfully\n");
+
+out_reschedule:
+	mutex_unlock(&svms->lock);
+	mmap_write_unlock(mm);
+	mutex_unlock(&process_info->lock);
+
+	/* If validation failed, reschedule another attempt */
+	if (evicted_ranges) {
+		pr_debug("reschedule to restore svm range\n");
+		queue_delayed_work(system_freezable_wq, &svms->restore_work,
+			msecs_to_jiffies(AMDGPU_SVM_RANGE_RESTORE_DELAY_MS));
+
+		kfd_smi_event_queue_restore_rescheduled(mm);
+	}
+	mmput(mm);
+}
+
+/**
+ * svm_range_evict - evict svm range
+ * @prange: svm range structure
+ * @mm: current process mm_struct
+ * @start: starting process queue number
+ * @last: last process queue number
+ * @event: mmu notifier event when range is evicted or migrated
+ *
+ * Stop all queues of the process to ensure GPU doesn't access the memory, then
+ * return to let CPU evict the buffer and proceed CPU pagetable update.
+ *
+ * Don't need use lock to sync cpu pagetable invalidation with GPU execution.
+ * If invalidation happens while restore work is running, restore work will
+ * restart to ensure to get the latest CPU pages mapping to GPU, then start
+ * the queues.
+ */
+static int
+svm_range_evict(struct svm_range *prange, struct mm_struct *mm,
+		unsigned long start, unsigned long last,
+		enum mmu_notifier_event event)
+{
+	struct svm_range_list *svms = prange->svms;
+	struct svm_range *pchild;
+	struct kfd_process *p;
+	int r = 0;
+
+	p = container_of(svms, struct kfd_process, svms);
+
+	pr_debug("invalidate svms 0x%p prange [0x%lx 0x%lx] [0x%lx 0x%lx]\n",
+		 svms, prange->start, prange->last, start, last);
+
+	if (!p->xnack_enabled ||
+	    (prange->flags & KFD_IOCTL_SVM_FLAG_GPU_ALWAYS_MAPPED)) {
+		int evicted_ranges;
+		bool mapped = prange->mapped_to_gpu;
+
+		list_for_each_entry(pchild, &prange->child_list, child_list) {
+			if (!pchild->mapped_to_gpu)
+				continue;
+			mapped = true;
+			mutex_lock_nested(&pchild->lock, 1);
+			if (pchild->start <= last && pchild->last >= start) {
+				pr_debug("increment pchild invalid [0x%lx 0x%lx]\n",
+					 pchild->start, pchild->last);
+				atomic_inc(&pchild->invalid);
+			}
+			mutex_unlock(&pchild->lock);
+		}
+
+		if (!mapped)
+			return r;
+
+		if (prange->start <= last && prange->last >= start)
+			atomic_inc(&prange->invalid);
+
+		evicted_ranges = atomic_inc_return(&svms->evicted_ranges);
+		if (evicted_ranges != 1)
+			return r;
+
+		pr_debug("evicting svms 0x%p range [0x%lx 0x%lx]\n",
+			 prange->svms, prange->start, prange->last);
+
+		/* First eviction, stop the queues */
+		r = kgd2kfd_quiesce_mm(mm, KFD_QUEUE_EVICTION_TRIGGER_SVM);
+		if (r)
+			pr_debug("failed to quiesce KFD\n");
+
+		pr_debug("schedule to restore svm %p ranges\n", svms);
+		queue_delayed_work(system_freezable_wq, &svms->restore_work,
+			msecs_to_jiffies(AMDGPU_SVM_RANGE_RESTORE_DELAY_MS));
+	} else {
+		unsigned long s, l;
+		uint32_t trigger;
+
+		if (event == MMU_NOTIFY_MIGRATE)
+			trigger = KFD_SVM_UNMAP_TRIGGER_MMU_NOTIFY_MIGRATE;
+		else
+			trigger = KFD_SVM_UNMAP_TRIGGER_MMU_NOTIFY;
+
+		pr_debug("invalidate unmap svms 0x%p [0x%lx 0x%lx] from GPUs\n",
+			 prange->svms, start, last);
+		list_for_each_entry(pchild, &prange->child_list, child_list) {
+			mutex_lock_nested(&pchild->lock, 1);
+			s = max(start, pchild->start);
+			l = min(last, pchild->last);
+			if (l >= s)
+				svm_range_unmap_from_gpus(pchild, s, l, trigger);
+			mutex_unlock(&pchild->lock);
+		}
+		s = max(start, prange->start);
+		l = min(last, prange->last);
+		if (l >= s)
+			svm_range_unmap_from_gpus(prange, s, l, trigger);
+	}
+
+	return r;
+}
+
+static struct svm_range *svm_range_clone(struct svm_range *old)
+{
+	struct svm_range *new;
+
+	new = svm_range_new(old->svms, old->start, old->last, false);
+	if (!new)
+		return NULL;
+	if (svm_range_copy_dma_addrs(new, old)) {
+		svm_range_free(new, false);
+		return NULL;
+	}
+	if (old->svm_bo) {
+		new->ttm_res = old->ttm_res;
+		new->offset = old->offset;
+		new->svm_bo = svm_range_bo_ref(old->svm_bo);
+		spin_lock(&new->svm_bo->list_lock);
+		list_add(&new->svm_bo_list, &new->svm_bo->range_list);
+		spin_unlock(&new->svm_bo->list_lock);
+	}
+	new->flags = old->flags;
+	new->preferred_loc = old->preferred_loc;
+	new->prefetch_loc = old->prefetch_loc;
+	new->actual_loc = old->actual_loc;
+	new->granularity = old->granularity;
+	new->mapped_to_gpu = old->mapped_to_gpu;
+	new->vram_pages = old->vram_pages;
+	bitmap_copy(new->bitmap_access, old->bitmap_access, MAX_GPU_INSTANCE);
+	bitmap_copy(new->bitmap_aip, old->bitmap_aip, MAX_GPU_INSTANCE);
+	atomic_set(&new->queue_refcount, atomic_read(&old->queue_refcount));
+
+	return new;
+}
+
+void svm_range_set_max_pages(struct amdgpu_device *adev)
+{
+	uint64_t max_pages;
+	uint64_t pages, _pages;
+	uint64_t min_pages = 0;
+	int i, id;
+
+	for (i = 0; i < adev->kfd.dev->num_nodes; i++) {
+		if (adev->kfd.dev->nodes[i]->xcp)
+			id = adev->kfd.dev->nodes[i]->xcp->id;
+		else
+			id = -1;
+		pages = KFD_XCP_MEMORY_SIZE(adev, id) >> 17;
+		pages = clamp(pages, 1ULL << 9, 1ULL << 18);
+		pages = rounddown_pow_of_two(pages);
+		min_pages = min_not_zero(min_pages, pages);
+	}
+
+	do {
+		max_pages = READ_ONCE(max_svm_range_pages);
+		_pages = min_not_zero(max_pages, min_pages);
+	} while (cmpxchg(&max_svm_range_pages, max_pages, _pages) != max_pages);
+}
+
+static int
+svm_range_split_new(struct svm_range_list *svms, uint64_t start, uint64_t last,
+		    uint64_t max_pages, struct list_head *insert_list,
+		    struct list_head *update_list)
+{
+	struct svm_range *prange;
+	uint64_t l;
+
+	pr_debug("max_svm_range_pages 0x%llx adding [0x%llx 0x%llx]\n",
+		 max_pages, start, last);
+
+	while (last >= start) {
+		l = min(last, ALIGN_DOWN(start + max_pages, max_pages) - 1);
+
+		prange = svm_range_new(svms, start, l, true);
+		if (!prange)
+			return -ENOMEM;
+		list_add(&prange->list, insert_list);
+		list_add(&prange->update_list, update_list);
+
+		start = l + 1;
+	}
+	return 0;
+}
+
+/**
+ * svm_range_add - add svm range and handle overlap
+ * @p: the range add to this process svms
+ * @start: page size aligned
+ * @size: page size aligned
+ * @nattr: number of attributes
+ * @attrs: array of attributes
+ * @update_list: output, the ranges need validate and update GPU mapping
+ * @insert_list: output, the ranges need insert to svms
+ * @remove_list: output, the ranges are replaced and need remove from svms
+ * @remap_list: output, remap unaligned svm ranges
+ *
+ * Check if the virtual address range has overlap with any existing ranges,
+ * split partly overlapping ranges and add new ranges in the gaps. All changes
+ * should be applied to the range_list and interval tree transactionally. If
+ * any range split or allocation fails, the entire update fails. Therefore any
+ * existing overlapping svm_ranges are cloned and the original svm_ranges left
+ * unchanged.
+ *
+ * If the transaction succeeds, the caller can update and insert clones and
+ * new ranges, then free the originals.
+ *
+ * Otherwise the caller can free the clones and new ranges, while the old
+ * svm_ranges remain unchanged.
+ *
+ * Context: Process context, caller must hold svms->lock
+ *
+ * Return:
+ * 0 - OK, otherwise error code
+ */
+static int
+svm_range_add(struct kfd_process *p, uint64_t start, uint64_t size,
+	      uint32_t nattr, struct kfd_ioctl_svm_attribute *attrs,
+	      struct list_head *update_list, struct list_head *insert_list,
+	      struct list_head *remove_list, struct list_head *remap_list)
+{
+	unsigned long last = start + size - 1UL;
+	struct svm_range_list *svms = &p->svms;
+	struct interval_tree_node *node;
+	struct svm_range *prange;
+	struct svm_range *tmp;
+	struct list_head new_list;
+	int r = 0;
+
+	pr_debug("svms 0x%p [0x%llx 0x%lx]\n", &p->svms, start, last);
+
+	INIT_LIST_HEAD(update_list);
+	INIT_LIST_HEAD(insert_list);
+	INIT_LIST_HEAD(remove_list);
+	INIT_LIST_HEAD(&new_list);
+	INIT_LIST_HEAD(remap_list);
+
+	node = interval_tree_iter_first(&svms->objects, start, last);
+	while (node) {
+		struct interval_tree_node *next;
+		unsigned long next_start;
+
+		pr_debug("found overlap node [0x%lx 0x%lx]\n", node->start,
+			 node->last);
+
+		prange = container_of(node, struct svm_range, it_node);
+		next = interval_tree_iter_next(node, start, last);
+		next_start = min(node->last, last) + 1;
+
+		if (svm_range_is_same_attrs(p, prange, nattr, attrs) &&
+		    prange->mapped_to_gpu) {
+			/* nothing to do */
+		} else if (node->start < start || node->last > last) {
+			/* node intersects the update range and its attributes
+			 * will change. Clone and split it, apply updates only
+			 * to the overlapping part
+			 */
+			struct svm_range *old = prange;
+
+			prange = svm_range_clone(old);
+			if (!prange) {
+				r = -ENOMEM;
+				goto out;
+			}
+
+			list_add(&old->update_list, remove_list);
+			list_add(&prange->list, insert_list);
+			list_add(&prange->update_list, update_list);
+
+			if (node->start < start) {
+				pr_debug("change old range start\n");
+				r = svm_range_split_head(prange, start,
+							 insert_list, remap_list);
+				if (r)
+					goto out;
+			}
+			if (node->last > last) {
+				pr_debug("change old range last\n");
+				r = svm_range_split_tail(prange, last,
+							 insert_list, remap_list);
+				if (r)
+					goto out;
+			}
+		} else {
+			/* The node is contained within start..last,
+			 * just update it
+			 */
+			list_add(&prange->update_list, update_list);
+		}
+
+		/* insert a new node if needed */
+		if (node->start > start) {
+			r = svm_range_split_new(svms, start, node->start - 1,
+						READ_ONCE(max_svm_range_pages),
+						&new_list, update_list);
+			if (r)
+				goto out;
+		}
+
+		node = next;
+		start = next_start;
+	}
+
+	/* add a final range at the end if needed */
+	if (start <= last)
+		r = svm_range_split_new(svms, start, last,
+					READ_ONCE(max_svm_range_pages),
+					&new_list, update_list);
+
+out:
+	if (r) {
+		list_for_each_entry_safe(prange, tmp, insert_list, list)
+			svm_range_free(prange, false);
+		list_for_each_entry_safe(prange, tmp, &new_list, list)
+			svm_range_free(prange, true);
+	} else {
+		list_splice(&new_list, insert_list);
+	}
+
+	return r;
+}
+
+static void
+svm_range_update_notifier_and_interval_tree(struct mm_struct *mm,
+					    struct svm_range *prange)
+{
+	unsigned long start;
+	unsigned long last;
+
+	start = prange->notifier.interval_tree.start >> PAGE_SHIFT;
+	last = prange->notifier.interval_tree.last >> PAGE_SHIFT;
+
+	if (prange->start == start && prange->last == last)
+		return;
+
+	pr_debug("up notifier 0x%p prange 0x%p [0x%lx 0x%lx] [0x%lx 0x%lx]\n",
+		  prange->svms, prange, start, last, prange->start,
+		  prange->last);
+
+	if (start != 0 && last != 0) {
+		interval_tree_remove(&prange->it_node, &prange->svms->objects);
+		svm_range_remove_notifier(prange);
+	}
+	prange->it_node.start = prange->start;
+	prange->it_node.last = prange->last;
+
+	interval_tree_insert(&prange->it_node, &prange->svms->objects);
+	svm_range_add_notifier_locked(mm, prange);
+}
+
+static void
+svm_range_handle_list_op(struct svm_range_list *svms, struct svm_range *prange,
+			 struct mm_struct *mm)
+{
+	switch (prange->work_item.op) {
+	case SVM_OP_NULL:
+		pr_debug("NULL OP 0x%p prange 0x%p [0x%lx 0x%lx]\n",
+			 svms, prange, prange->start, prange->last);
+		break;
+	case SVM_OP_UNMAP_RANGE:
+		pr_debug("remove 0x%p prange 0x%p [0x%lx 0x%lx]\n",
+			 svms, prange, prange->start, prange->last);
+		svm_range_unlink(prange);
+		svm_range_remove_notifier(prange);
+		svm_range_free(prange, true);
+		break;
+	case SVM_OP_UPDATE_RANGE_NOTIFIER:
+		pr_debug("update notifier 0x%p prange 0x%p [0x%lx 0x%lx]\n",
+			 svms, prange, prange->start, prange->last);
+		svm_range_update_notifier_and_interval_tree(mm, prange);
+		break;
+	case SVM_OP_UPDATE_RANGE_NOTIFIER_AND_MAP:
+		pr_debug("update and map 0x%p prange 0x%p [0x%lx 0x%lx]\n",
+			 svms, prange, prange->start, prange->last);
+		svm_range_update_notifier_and_interval_tree(mm, prange);
+		/* TODO: implement deferred validation and mapping */
+		break;
+	case SVM_OP_ADD_RANGE:
+		pr_debug("add 0x%p prange 0x%p [0x%lx 0x%lx]\n", svms, prange,
+			 prange->start, prange->last);
+		svm_range_add_to_svms(prange);
+		svm_range_add_notifier_locked(mm, prange);
+		break;
+	case SVM_OP_ADD_RANGE_AND_MAP:
+		pr_debug("add and map 0x%p prange 0x%p [0x%lx 0x%lx]\n", svms,
+			 prange, prange->start, prange->last);
+		svm_range_add_to_svms(prange);
+		svm_range_add_notifier_locked(mm, prange);
+		/* TODO: implement deferred validation and mapping */
+		break;
+	default:
+		WARN_ONCE(1, "Unknown prange 0x%p work op %d\n", prange,
+			 prange->work_item.op);
+	}
+}
+
+static void svm_range_drain_retry_fault(struct svm_range_list *svms)
+{
+	struct kfd_process_device *pdd;
+	struct kfd_process *p;
+	uint32_t i;
+
+	p = container_of(svms, struct kfd_process, svms);
+
+	for_each_set_bit(i, svms->bitmap_supported, p->n_pdds) {
+		pdd = p->pdds[i];
+		if (!pdd)
+			continue;
+
+		pr_debug("drain retry fault gpu %d svms %p\n", i, svms);
+
+		amdgpu_ih_wait_on_checkpoint_process_ts(pdd->dev->adev,
+				pdd->dev->adev->irq.retry_cam_enabled ?
+				&pdd->dev->adev->irq.ih :
+				&pdd->dev->adev->irq.ih1);
+
+		if (pdd->dev->adev->irq.retry_cam_enabled)
+			amdgpu_ih_wait_on_checkpoint_process_ts(pdd->dev->adev,
+				&pdd->dev->adev->irq.ih_soft);
+
+
+		pr_debug("drain retry fault gpu %d svms 0x%p done\n", i, svms);
+	}
+}
+
+static void svm_range_deferred_list_work(struct work_struct *work)
+{
+	struct svm_range_list *svms;
+	struct svm_range *prange;
+	struct mm_struct *mm;
+
+	svms = container_of(work, struct svm_range_list, deferred_list_work);
+	pr_debug("enter svms 0x%p\n", svms);
+
+	spin_lock(&svms->deferred_list_lock);
+	while (!list_empty(&svms->deferred_range_list)) {
+		prange = list_first_entry(&svms->deferred_range_list,
+					  struct svm_range, deferred_list);
+		spin_unlock(&svms->deferred_list_lock);
+
+		pr_debug("prange 0x%p [0x%lx 0x%lx] op %d\n", prange,
+			 prange->start, prange->last, prange->work_item.op);
+
+		mm = prange->work_item.mm;
+
+		mmap_write_lock(mm);
+
+		/* Remove from deferred_list must be inside mmap write lock, for
+		 * two race cases:
+		 * 1. unmap_from_cpu may change work_item.op and add the range
+		 *    to deferred_list again, cause use after free bug.
+		 * 2. svm_range_list_lock_and_flush_work may hold mmap write
+		 *    lock and continue because deferred_list is empty, but
+		 *    deferred_list work is actually waiting for mmap lock.
+		 */
+		spin_lock(&svms->deferred_list_lock);
+		list_del_init(&prange->deferred_list);
+		spin_unlock(&svms->deferred_list_lock);
+
+		mutex_lock(&svms->lock);
+		mutex_lock(&prange->migrate_mutex);
+		while (!list_empty(&prange->child_list)) {
+			struct svm_range *pchild;
+
+			pchild = list_first_entry(&prange->child_list,
+						struct svm_range, child_list);
+			pr_debug("child prange 0x%p op %d\n", pchild,
+				 pchild->work_item.op);
+			list_del_init(&pchild->child_list);
+			svm_range_handle_list_op(svms, pchild, mm);
+		}
+		mutex_unlock(&prange->migrate_mutex);
+
+		svm_range_handle_list_op(svms, prange, mm);
+		mutex_unlock(&svms->lock);
+		mmap_write_unlock(mm);
+
+		/* Pairs with mmget in svm_range_add_list_work. If dropping the
+		 * last mm refcount, schedule release work to avoid circular locking
+		 */
+		mmput_async(mm);
+
+		spin_lock(&svms->deferred_list_lock);
+	}
+	spin_unlock(&svms->deferred_list_lock);
+	pr_debug("exit svms 0x%p\n", svms);
+}
+
+void
+svm_range_add_list_work(struct svm_range_list *svms, struct svm_range *prange,
+			struct mm_struct *mm, enum svm_work_list_ops op)
+{
+	spin_lock(&svms->deferred_list_lock);
+	/* if prange is on the deferred list */
+	if (!list_empty(&prange->deferred_list)) {
+		pr_debug("update exist prange 0x%p work op %d\n", prange, op);
+		WARN_ONCE(prange->work_item.mm != mm, "unmatch mm\n");
+		if (op != SVM_OP_NULL &&
+		    prange->work_item.op != SVM_OP_UNMAP_RANGE)
+			prange->work_item.op = op;
+	} else {
+		/* Pairs with mmput in deferred_list_work.
+		 * If process is exiting and mm is gone, don't update mmu notifier.
+		 */
+		if (mmget_not_zero(mm)) {
+			prange->work_item.mm = mm;
+			prange->work_item.op = op;
+			list_add_tail(&prange->deferred_list,
+				      &prange->svms->deferred_range_list);
+			pr_debug("add prange 0x%p [0x%lx 0x%lx] to work list op %d\n",
+				 prange, prange->start, prange->last, op);
+		}
+	}
+	spin_unlock(&svms->deferred_list_lock);
+}
+
+void schedule_deferred_list_work(struct svm_range_list *svms)
+{
+	spin_lock(&svms->deferred_list_lock);
+	if (!list_empty(&svms->deferred_range_list))
+		schedule_work(&svms->deferred_list_work);
+	spin_unlock(&svms->deferred_list_lock);
+}
+
+static void
+svm_range_unmap_split(struct svm_range *parent, struct svm_range *prange, unsigned long start,
+		      unsigned long last)
+{
+	struct svm_range *head;
+	struct svm_range *tail;
+
+	if (prange->work_item.op == SVM_OP_UNMAP_RANGE) {
+		pr_debug("prange 0x%p [0x%lx 0x%lx] is already freed\n", prange,
+			 prange->start, prange->last);
+		return;
+	}
+	if (start > prange->last || last < prange->start)
+		return;
+
+	head = tail = prange;
+	if (start > prange->start)
+		svm_range_split(prange, prange->start, start - 1, &tail);
+	if (last < tail->last)
+		svm_range_split(tail, last + 1, tail->last, &head);
+
+	if (head != prange && tail != prange) {
+		svm_range_add_child(parent, head, SVM_OP_UNMAP_RANGE);
+		svm_range_add_child(parent, tail, SVM_OP_ADD_RANGE);
+	} else if (tail != prange) {
+		svm_range_add_child(parent, tail, SVM_OP_UNMAP_RANGE);
+	} else if (head != prange) {
+		svm_range_add_child(parent, head, SVM_OP_UNMAP_RANGE);
+	} else if (parent != prange) {
+		prange->work_item.op = SVM_OP_UNMAP_RANGE;
+	}
+}
+
+static void
+svm_range_unmap_from_cpu(struct mm_struct *mm, struct svm_range *prange,
+			 unsigned long start, unsigned long last)
+{
+	uint32_t trigger = KFD_SVM_UNMAP_TRIGGER_UNMAP_FROM_CPU;
+	struct svm_range_list *svms;
+	struct svm_range *pchild;
+	struct kfd_process *p;
+	unsigned long s, l;
+	bool unmap_parent;
+	uint32_t i;
+
+	if (atomic_read(&prange->queue_refcount)) {
+		int r;
+
+		pr_warn("Freeing queue vital buffer 0x%lx, queue evicted\n",
+			prange->start << PAGE_SHIFT);
+		r = kgd2kfd_quiesce_mm(mm, KFD_QUEUE_EVICTION_TRIGGER_SVM);
+		if (r)
+			pr_debug("failed %d to quiesce KFD queues\n", r);
+	}
+
+	p = kfd_lookup_process_by_mm(mm);
+	if (!p)
+		return;
+	svms = &p->svms;
+
+	pr_debug("svms 0x%p prange 0x%p [0x%lx 0x%lx] [0x%lx 0x%lx]\n", svms,
+		 prange, prange->start, prange->last, start, last);
+
+	/* calculate time stamps that are used to decide which page faults need be
+	 * dropped or handled before unmap pages from gpu vm
+	 */
+	for_each_set_bit(i, svms->bitmap_supported, p->n_pdds) {
+		struct kfd_process_device *pdd;
+		struct amdgpu_device *adev;
+		struct amdgpu_ih_ring *ih;
+		uint32_t checkpoint_wptr;
+
+		pdd = p->pdds[i];
+		if (!pdd)
+			continue;
+
+		adev = pdd->dev->adev;
+
+		/* Check and drain ih1 ring if cam not available */
+		if (adev->irq.ih1.ring_size) {
+			ih = &adev->irq.ih1;
+			checkpoint_wptr = amdgpu_ih_get_wptr(adev, ih);
+			if (ih->rptr != checkpoint_wptr) {
+				svms->checkpoint_ts[i] =
+					amdgpu_ih_decode_iv_ts(adev, ih, checkpoint_wptr, -1);
+				continue;
+			}
+		}
+
+		/* check if dev->irq.ih_soft is not empty */
+		ih = &adev->irq.ih_soft;
+		checkpoint_wptr = amdgpu_ih_get_wptr(adev, ih);
+		if (ih->rptr != checkpoint_wptr)
+			svms->checkpoint_ts[i] = amdgpu_ih_decode_iv_ts(adev, ih, checkpoint_wptr, -1);
+	}
+
+	unmap_parent = start <= prange->start && last >= prange->last;
+
+	list_for_each_entry(pchild, &prange->child_list, child_list) {
+		mutex_lock_nested(&pchild->lock, 1);
+		s = max(start, pchild->start);
+		l = min(last, pchild->last);
+		if (l >= s)
+			svm_range_unmap_from_gpus(pchild, s, l, trigger);
+		svm_range_unmap_split(prange, pchild, start, last);
+		mutex_unlock(&pchild->lock);
+	}
+	s = max(start, prange->start);
+	l = min(last, prange->last);
+	if (l >= s)
+		svm_range_unmap_from_gpus(prange, s, l, trigger);
+	svm_range_unmap_split(prange, prange, start, last);
+
+	if (unmap_parent)
+		svm_range_add_list_work(svms, prange, mm, SVM_OP_UNMAP_RANGE);
+	else
+		svm_range_add_list_work(svms, prange, mm,
+					SVM_OP_UPDATE_RANGE_NOTIFIER);
+	schedule_deferred_list_work(svms);
+
+	kfd_unref_process(p);
+}
+
+/**
+ * svm_range_cpu_invalidate_pagetables - interval notifier callback
+ * @mni: mmu_interval_notifier struct
+ * @range: mmu_notifier_range struct
+ * @cur_seq: value to pass to mmu_interval_set_seq()
+ *
+ * If event is MMU_NOTIFY_UNMAP, this is from CPU unmap range, otherwise, it
+ * is from migration, or CPU page invalidation callback.
+ *
+ * For unmap event, unmap range from GPUs, remove prange from svms in a delayed
+ * work thread, and split prange if only part of prange is unmapped.
+ *
+ * For invalidation event, if GPU retry fault is not enabled, evict the queues,
+ * then schedule svm_range_restore_work to update GPU mapping and resume queues.
+ * If GPU retry fault is enabled, unmap the svm range from GPU, retry fault will
+ * update GPU mapping to recover.
+ *
+ * Context: mmap lock, notifier_invalidate_start lock are held
+ *          for invalidate event, prange lock is held if this is from migration
+ */
+static bool
+svm_range_cpu_invalidate_pagetables(struct mmu_interval_notifier *mni,
+				    const struct mmu_notifier_range *range,
+				    unsigned long cur_seq)
+{
+	struct svm_range *prange;
+	unsigned long start;
+	unsigned long last;
+
+	if (range->event == MMU_NOTIFY_RELEASE)
+		return true;
+
+	start = mni->interval_tree.start;
+	last = mni->interval_tree.last;
+	start = max(start, range->start) >> PAGE_SHIFT;
+	last = min(last, range->end - 1) >> PAGE_SHIFT;
+	pr_debug("[0x%lx 0x%lx] range[0x%lx 0x%lx] notifier[0x%lx 0x%lx] %d\n",
+		 start, last, range->start >> PAGE_SHIFT,
+		 (range->end - 1) >> PAGE_SHIFT,
+		 mni->interval_tree.start >> PAGE_SHIFT,
+		 mni->interval_tree.last >> PAGE_SHIFT, range->event);
+
+	prange = container_of(mni, struct svm_range, notifier);
+
+	svm_range_lock(prange);
+	mmu_interval_set_seq(mni, cur_seq);
+
+	switch (range->event) {
+	case MMU_NOTIFY_UNMAP:
+		svm_range_unmap_from_cpu(mni->mm, prange, start, last);
+		break;
+	default:
+		svm_range_evict(prange, mni->mm, start, last, range->event);
+		break;
+	}
+
+	svm_range_unlock(prange);
+
+	return true;
+}
+
+/**
+ * svm_range_from_addr - find svm range from fault address
+ * @svms: svm range list header
+ * @addr: address to search range interval tree, in pages
+ * @parent: parent range if range is on child list
+ *
+ * Context: The caller must hold svms->lock
+ *
+ * Return: the svm_range found or NULL
+ */
+struct svm_range *
+svm_range_from_addr(struct svm_range_list *svms, unsigned long addr,
+		    struct svm_range **parent)
+{
+	struct interval_tree_node *node;
+	struct svm_range *prange;
+	struct svm_range *pchild;
+
+	node = interval_tree_iter_first(&svms->objects, addr, addr);
+	if (!node)
+		return NULL;
+
+	prange = container_of(node, struct svm_range, it_node);
+	pr_debug("address 0x%lx prange [0x%lx 0x%lx] node [0x%lx 0x%lx]\n",
+		 addr, prange->start, prange->last, node->start, node->last);
+
+	if (addr >= prange->start && addr <= prange->last) {
+		if (parent)
+			*parent = prange;
+		return prange;
+	}
+	list_for_each_entry(pchild, &prange->child_list, child_list)
+		if (addr >= pchild->start && addr <= pchild->last) {
+			pr_debug("found address 0x%lx pchild [0x%lx 0x%lx]\n",
+				 addr, pchild->start, pchild->last);
+			if (parent)
+				*parent = prange;
+			return pchild;
+		}
+
+	return NULL;
+}
+
+/* svm_range_best_restore_location - decide the best fault restore location
+ * @prange: svm range structure
+ * @adev: the GPU on which vm fault happened
+ *
+ * This is only called when xnack is on, to decide the best location to restore
+ * the range mapping after GPU vm fault. Caller uses the best location to do
+ * migration if actual loc is not best location, then update GPU page table
+ * mapping to the best location.
+ *
+ * If the preferred loc is accessible by faulting GPU, use preferred loc.
+ * If vm fault gpu idx is on range ACCESSIBLE bitmap, best_loc is vm fault gpu
+ * If vm fault gpu idx is on range ACCESSIBLE_IN_PLACE bitmap, then
+ *    if range actual loc is cpu, best_loc is cpu
+ *    if vm fault gpu is on xgmi same hive of range actual loc gpu, best_loc is
+ *    range actual loc.
+ * Otherwise, GPU no access, best_loc is -1.
+ *
+ * Return:
+ * -1 means vm fault GPU no access
+ * 0 for CPU or GPU id
+ */
+static int32_t
+svm_range_best_restore_location(struct svm_range *prange,
+				struct kfd_node *node,
+				int32_t *gpuidx)
+{
+	struct kfd_node *bo_node, *preferred_node;
+	struct kfd_process *p;
+	uint32_t gpuid;
+	int r;
+
+	p = container_of(prange->svms, struct kfd_process, svms);
+
+	r = kfd_process_gpuid_from_node(p, node, &gpuid, gpuidx);
+	if (r < 0) {
+		pr_debug("failed to get gpuid from kgd\n");
+		return -1;
+	}
+
+	if (node->adev->apu_prefer_gtt)
+		return 0;
+
+	if (prange->preferred_loc == gpuid ||
+	    prange->preferred_loc == KFD_IOCTL_SVM_LOCATION_SYSMEM) {
+		return prange->preferred_loc;
+	} else if (prange->preferred_loc != KFD_IOCTL_SVM_LOCATION_UNDEFINED) {
+		preferred_node = svm_range_get_node_by_id(prange, prange->preferred_loc);
+		if (preferred_node && svm_nodes_in_same_hive(node, preferred_node))
+			return prange->preferred_loc;
+		/* fall through */
+	}
+
+	if (test_bit(*gpuidx, prange->bitmap_access))
+		return gpuid;
+
+	if (test_bit(*gpuidx, prange->bitmap_aip)) {
+		if (!prange->actual_loc)
+			return 0;
+
+		bo_node = svm_range_get_node_by_id(prange, prange->actual_loc);
+		if (bo_node && svm_nodes_in_same_hive(node, bo_node))
+			return prange->actual_loc;
+		else
+			return 0;
+	}
+
+	return -1;
+}
+
+static int
+svm_range_get_range_boundaries(struct kfd_process *p, int64_t addr,
+			       unsigned long *start, unsigned long *last,
+			       bool *is_heap_stack)
+{
+	struct vm_area_struct *vma;
+	struct interval_tree_node *node;
+	struct rb_node *rb_node;
+	unsigned long start_limit, end_limit;
+
+	vma = vma_lookup(p->mm, addr << PAGE_SHIFT);
+	if (!vma) {
+		pr_debug("VMA does not exist in address [0x%llx]\n", addr);
+		return -EFAULT;
+	}
+
+	*is_heap_stack = vma_is_initial_heap(vma) || vma_is_initial_stack(vma);
+
+	start_limit = max(vma->vm_start >> PAGE_SHIFT,
+		      (unsigned long)ALIGN_DOWN(addr, 1UL << p->svms.default_granularity));
+	end_limit = min(vma->vm_end >> PAGE_SHIFT,
+		    (unsigned long)ALIGN(addr + 1, 1UL << p->svms.default_granularity));
+
+	/* First range that starts after the fault address */
+	node = interval_tree_iter_first(&p->svms.objects, addr + 1, ULONG_MAX);
+	if (node) {
+		end_limit = min(end_limit, node->start);
+		/* Last range that ends before the fault address */
+		rb_node = rb_prev(&node->rb);
+	} else {
+		/* Last range must end before addr because
+		 * there was no range after addr
+		 */
+		rb_node = rb_last(&p->svms.objects.rb_root);
+	}
+	if (rb_node) {
+		node = container_of(rb_node, struct interval_tree_node, rb);
+		if (node->last >= addr) {
+			WARN(1, "Overlap with prev node and page fault addr\n");
+			return -EFAULT;
+		}
+		start_limit = max(start_limit, node->last + 1);
+	}
+
+	*start = start_limit;
+	*last = end_limit - 1;
+
+	pr_debug("vma [0x%lx 0x%lx] range [0x%lx 0x%lx] is_heap_stack %d\n",
+		 vma->vm_start >> PAGE_SHIFT, vma->vm_end >> PAGE_SHIFT,
+		 *start, *last, *is_heap_stack);
+
+	return 0;
+}
+
+static int
+svm_range_check_vm_userptr(struct kfd_process *p, uint64_t start, uint64_t last,
+			   uint64_t *bo_s, uint64_t *bo_l)
+{
+	struct amdgpu_bo_va_mapping *mapping;
+	struct interval_tree_node *node;
+	struct amdgpu_bo *bo = NULL;
+	unsigned long userptr;
+	uint32_t i;
+	int r;
+
+	for (i = 0; i < p->n_pdds; i++) {
+		struct amdgpu_vm *vm;
+
+		if (!p->pdds[i]->drm_priv)
+			continue;
+
+		vm = drm_priv_to_vm(p->pdds[i]->drm_priv);
+		r = amdgpu_bo_reserve(vm->root.bo, false);
+		if (r)
+			return r;
+
+		/* Check userptr by searching entire vm->va interval tree */
+		node = interval_tree_iter_first(&vm->va, 0, ~0ULL);
+		while (node) {
+			mapping = container_of((struct rb_node *)node,
+					       struct amdgpu_bo_va_mapping, rb);
+			bo = mapping->bo_va->base.bo;
+
+			if (!amdgpu_ttm_tt_affect_userptr(bo->tbo.ttm,
+							 start << PAGE_SHIFT,
+							 last << PAGE_SHIFT,
+							 &userptr)) {
+				node = interval_tree_iter_next(node, 0, ~0ULL);
+				continue;
+			}
+
+			pr_debug("[0x%llx 0x%llx] already userptr mapped\n",
+				 start, last);
+			if (bo_s && bo_l) {
+				*bo_s = userptr >> PAGE_SHIFT;
+				*bo_l = *bo_s + bo->tbo.ttm->num_pages - 1;
+			}
+			amdgpu_bo_unreserve(vm->root.bo);
+			return -EADDRINUSE;
+		}
+		amdgpu_bo_unreserve(vm->root.bo);
+	}
+	return 0;
+}
+
+static struct
+svm_range *svm_range_create_unregistered_range(struct kfd_node *node,
+						struct kfd_process *p,
+						struct mm_struct *mm,
+						int64_t addr)
+{
+	struct svm_range *prange = NULL;
+	unsigned long start, last;
+	uint32_t gpuid, gpuidx;
+	bool is_heap_stack;
+	uint64_t bo_s = 0;
+	uint64_t bo_l = 0;
+	int r;
+
+	if (svm_range_get_range_boundaries(p, addr, &start, &last,
+					   &is_heap_stack))
+		return NULL;
+
+	r = svm_range_check_vm(p, start, last, &bo_s, &bo_l);
+	if (r != -EADDRINUSE)
+		r = svm_range_check_vm_userptr(p, start, last, &bo_s, &bo_l);
+
+	if (r == -EADDRINUSE) {
+		if (addr >= bo_s && addr <= bo_l)
+			return NULL;
+
+		/* Create one page svm range if 2MB range overlapping */
+		start = addr;
+		last = addr;
+	}
+
+	prange = svm_range_new(&p->svms, start, last, true);
+	if (!prange) {
+		pr_debug("Failed to create prange in address [0x%llx]\n", addr);
+		return NULL;
+	}
+	if (kfd_process_gpuid_from_node(p, node, &gpuid, &gpuidx)) {
+		pr_debug("failed to get gpuid from kgd\n");
+		svm_range_free(prange, true);
+		return NULL;
+	}
+
+	if (is_heap_stack)
+		prange->preferred_loc = KFD_IOCTL_SVM_LOCATION_SYSMEM;
+
+	svm_range_add_to_svms(prange);
+	svm_range_add_notifier_locked(mm, prange);
+
+	return prange;
+}
+
+/* svm_range_skip_recover - decide if prange can be recovered
+ * @prange: svm range structure
+ *
+ * GPU vm retry fault handle skip recover the range for cases:
+ * 1. prange is on deferred list to be removed after unmap, it is stale fault,
+ *    deferred list work will drain the stale fault before free the prange.
+ * 2. prange is on deferred list to add interval notifier after split, or
+ * 3. prange is child range, it is split from parent prange, recover later
+ *    after interval notifier is added.
+ *
+ * Return: true to skip recover, false to recover
+ */
+static bool svm_range_skip_recover(struct svm_range *prange)
+{
+	struct svm_range_list *svms = prange->svms;
+
+	spin_lock(&svms->deferred_list_lock);
+	if (list_empty(&prange->deferred_list) &&
+	    list_empty(&prange->child_list)) {
+		spin_unlock(&svms->deferred_list_lock);
+		return false;
+	}
+	spin_unlock(&svms->deferred_list_lock);
+
+	if (prange->work_item.op == SVM_OP_UNMAP_RANGE) {
+		pr_debug("svms 0x%p prange 0x%p [0x%lx 0x%lx] unmapped\n",
+			 svms, prange, prange->start, prange->last);
+		return true;
+	}
+	if (prange->work_item.op == SVM_OP_ADD_RANGE_AND_MAP ||
+	    prange->work_item.op == SVM_OP_ADD_RANGE) {
+		pr_debug("svms 0x%p prange 0x%p [0x%lx 0x%lx] not added yet\n",
+			 svms, prange, prange->start, prange->last);
+		return true;
+	}
+	return false;
+}
+
+static void
+svm_range_count_fault(struct kfd_node *node, struct kfd_process *p,
+		      int32_t gpuidx)
+{
+	struct kfd_process_device *pdd;
+
+	/* fault is on different page of same range
+	 * or fault is skipped to recover later
+	 * or fault is on invalid virtual address
+	 */
+	if (gpuidx == MAX_GPU_INSTANCE) {
+		uint32_t gpuid;
+		int r;
+
+		r = kfd_process_gpuid_from_node(p, node, &gpuid, &gpuidx);
+		if (r < 0)
+			return;
+	}
+
+	/* fault is recovered
+	 * or fault cannot recover because GPU no access on the range
+	 */
+	pdd = kfd_process_device_from_gpuidx(p, gpuidx);
+	if (pdd)
+		WRITE_ONCE(pdd->faults, pdd->faults + 1);
+}
+
+static bool
+svm_fault_allowed(struct vm_area_struct *vma, bool write_fault)
+{
+	unsigned long requested = VM_READ;
+
+	if (write_fault)
+		requested |= VM_WRITE;
+
+	pr_debug("requested 0x%lx, vma permission flags 0x%lx\n", requested,
+		vma->vm_flags);
+	return (vma->vm_flags & requested) == requested;
+}
+
+int
+svm_range_restore_pages(struct amdgpu_device *adev, unsigned int pasid,
+			uint32_t vmid, uint32_t node_id,
+			uint64_t addr, uint64_t ts, bool write_fault)
+{
+	unsigned long start, last, size;
+	struct mm_struct *mm = NULL;
+	struct svm_range_list *svms;
+	struct svm_range *prange;
+	struct kfd_process *p;
+	ktime_t timestamp = ktime_get_boottime();
+	struct kfd_node *node;
+	int32_t best_loc;
+	int32_t gpuid, gpuidx = MAX_GPU_INSTANCE;
+	bool write_locked = false;
+	struct vm_area_struct *vma;
+	bool migration = false;
+	int r = 0;
+
+	if (!KFD_IS_SVM_API_SUPPORTED(adev)) {
+		pr_debug("device does not support SVM\n");
+		return -EFAULT;
+	}
+
+	p = kfd_lookup_process_by_pasid(pasid, NULL);
+	if (!p) {
+		pr_debug("kfd process not founded pasid 0x%x\n", pasid);
+		return 0;
+	}
+	svms = &p->svms;
+
+	pr_debug("restoring svms 0x%p fault address 0x%llx\n", svms, addr);
+
+	if (atomic_read(&svms->drain_pagefaults)) {
+		pr_debug("page fault handling disabled, drop fault 0x%llx\n", addr);
+		r = 0;
+		goto out;
+	}
+
+	node = kfd_node_by_irq_ids(adev, node_id, vmid);
+	if (!node) {
+		pr_debug("kfd node does not exist node_id: %d, vmid: %d\n", node_id,
+			 vmid);
+		r = -EFAULT;
+		goto out;
+	}
+
+	if (kfd_process_gpuid_from_node(p, node, &gpuid, &gpuidx)) {
+		pr_debug("failed to get gpuid/gpuidex for node_id: %d\n", node_id);
+		r = -EFAULT;
+		goto out;
+	}
+
+	if (!p->xnack_enabled) {
+		pr_debug("XNACK not enabled for pasid 0x%x\n", pasid);
+		r = -EFAULT;
+		goto out;
+	}
+
+	/* p->lead_thread is available as kfd_process_wq_release flush the work
+	 * before releasing task ref.
+	 */
+	mm = get_task_mm(p->lead_thread);
+	if (!mm) {
+		pr_debug("svms 0x%p failed to get mm\n", svms);
+		r = 0;
+		goto out;
+	}
+
+	mmap_read_lock(mm);
+retry_write_locked:
+	mutex_lock(&svms->lock);
+
+	/* check if this page fault time stamp is before svms->checkpoint_ts */
+	if (svms->checkpoint_ts[gpuidx] != 0) {
+		if (amdgpu_ih_ts_after_or_equal(ts,  svms->checkpoint_ts[gpuidx])) {
+			pr_debug("draining retry fault, drop fault 0x%llx\n", addr);
+			if (write_locked)
+				mmap_write_downgrade(mm);
+			r = -EAGAIN;
+			goto out_unlock_svms;
+		} else {
+			/* ts is after svms->checkpoint_ts now, reset svms->checkpoint_ts
+			 * to zero to avoid following ts wrap around give wrong comparing
+			 */
+			svms->checkpoint_ts[gpuidx] = 0;
+		}
+	}
+
+	prange = svm_range_from_addr(svms, addr, NULL);
+	if (!prange) {
+		pr_debug("failed to find prange svms 0x%p address [0x%llx]\n",
+			 svms, addr);
+		if (!write_locked) {
+			/* Need the write lock to create new range with MMU notifier.
+			 * Also flush pending deferred work to make sure the interval
+			 * tree is up to date before we add a new range
+			 */
+			mutex_unlock(&svms->lock);
+			mmap_read_unlock(mm);
+			mmap_write_lock(mm);
+			write_locked = true;
+			goto retry_write_locked;
+		}
+		prange = svm_range_create_unregistered_range(node, p, mm, addr);
+		if (!prange) {
+			pr_debug("failed to create unregistered range svms 0x%p address [0x%llx]\n",
+				 svms, addr);
+			mmap_write_downgrade(mm);
+			r = -EFAULT;
+			goto out_unlock_svms;
+		}
+	}
+	if (write_locked)
+		mmap_write_downgrade(mm);
+
+	mutex_lock(&prange->migrate_mutex);
+
+	if (svm_range_skip_recover(prange)) {
+		amdgpu_gmc_filter_faults_remove(node->adev, addr, pasid);
+		r = 0;
+		goto out_unlock_range;
+	}
+
+	/* skip duplicate vm fault on different pages of same range */
+	if (ktime_before(timestamp, ktime_add_ns(prange->validate_timestamp,
+				AMDGPU_SVM_RANGE_RETRY_FAULT_PENDING))) {
+		pr_debug("svms 0x%p [0x%lx %lx] already restored\n",
+			 svms, prange->start, prange->last);
+		r = 0;
+		goto out_unlock_range;
+	}
+
+	/* __do_munmap removed VMA, return success as we are handling stale
+	 * retry fault.
+	 */
+	vma = vma_lookup(mm, addr << PAGE_SHIFT);
+	if (!vma) {
+		pr_debug("address 0x%llx VMA is removed\n", addr);
+		r = 0;
+		goto out_unlock_range;
+	}
+
+	if (!svm_fault_allowed(vma, write_fault)) {
+		pr_debug("fault addr 0x%llx no %s permission\n", addr,
+			write_fault ? "write" : "read");
+		r = -EPERM;
+		goto out_unlock_range;
+	}
+
+	best_loc = svm_range_best_restore_location(prange, node, &gpuidx);
+	if (best_loc == -1) {
+		pr_debug("svms %p failed get best restore loc [0x%lx 0x%lx]\n",
+			 svms, prange->start, prange->last);
+		r = -EACCES;
+		goto out_unlock_range;
+	}
+
+	pr_debug("svms %p [0x%lx 0x%lx] best restore 0x%x, actual loc 0x%x\n",
+		 svms, prange->start, prange->last, best_loc,
+		 prange->actual_loc);
+
+	kfd_smi_event_page_fault_start(node, p->lead_thread->pid, addr,
+				       write_fault, timestamp);
+
+	/* Align migration range start and size to granularity size */
+	size = 1UL << prange->granularity;
+	start = max_t(unsigned long, ALIGN_DOWN(addr, size), prange->start);
+	last = min_t(unsigned long, ALIGN(addr + 1, size) - 1, prange->last);
+	if (prange->actual_loc != 0 || best_loc != 0) {
+		if (best_loc) {
+			r = svm_migrate_to_vram(prange, best_loc, start, last,
+					mm, KFD_MIGRATE_TRIGGER_PAGEFAULT_GPU);
+			if (r) {
+				pr_debug("svm_migrate_to_vram failed (%d) at %llx, falling back to system memory\n",
+					 r, addr);
+				/* Fallback to system memory if migration to
+				 * VRAM failed
+				 */
+				if (prange->actual_loc && prange->actual_loc != best_loc)
+					r = svm_migrate_vram_to_ram(prange, mm, start, last,
+						KFD_MIGRATE_TRIGGER_PAGEFAULT_GPU, NULL);
+				else
+					r = 0;
+			}
+		} else {
+			r = svm_migrate_vram_to_ram(prange, mm, start, last,
+					KFD_MIGRATE_TRIGGER_PAGEFAULT_GPU, NULL);
+		}
+		if (r) {
+			pr_debug("failed %d to migrate svms %p [0x%lx 0x%lx]\n",
+				 r, svms, start, last);
+			goto out_migrate_fail;
+		} else {
+			migration = true;
+		}
+	}
+
+	r = svm_range_validate_and_map(mm, start, last, prange, gpuidx, false,
+				       false, false);
+	if (r)
+		pr_debug("failed %d to map svms 0x%p [0x%lx 0x%lx] to gpus\n",
+			 r, svms, start, last);
+
+out_migrate_fail:
+	kfd_smi_event_page_fault_end(node, p->lead_thread->pid, addr,
+				     migration);
+
+out_unlock_range:
+	mutex_unlock(&prange->migrate_mutex);
+out_unlock_svms:
+	mutex_unlock(&svms->lock);
+	mmap_read_unlock(mm);
+
+	if (r != -EAGAIN)
+		svm_range_count_fault(node, p, gpuidx);
+
+	mmput(mm);
+out:
+	kfd_unref_process(p);
+
+	if (r == -EAGAIN) {
+		pr_debug("recover vm fault later\n");
+		amdgpu_gmc_filter_faults_remove(node->adev, addr, pasid);
+		r = 0;
+	}
+	return r;
+}
+
+int
+svm_range_switch_xnack_reserve_mem(struct kfd_process *p, bool xnack_enabled)
+{
+	struct svm_range *prange, *pchild;
+	uint64_t reserved_size = 0;
+	uint64_t size;
+	int r = 0;
+
+	pr_debug("switching xnack from %d to %d\n", p->xnack_enabled, xnack_enabled);
+
+	mutex_lock(&p->svms.lock);
+
+	list_for_each_entry(prange, &p->svms.list, list) {
+		svm_range_lock(prange);
+		list_for_each_entry(pchild, &prange->child_list, child_list) {
+			size = (pchild->last - pchild->start + 1) << PAGE_SHIFT;
+			if (xnack_enabled) {
+				amdgpu_amdkfd_unreserve_mem_limit(NULL, size,
+					KFD_IOC_ALLOC_MEM_FLAGS_USERPTR, 0);
+			} else {
+				r = amdgpu_amdkfd_reserve_mem_limit(NULL, size,
+					KFD_IOC_ALLOC_MEM_FLAGS_USERPTR, 0);
+				if (r)
+					goto out_unlock;
+				reserved_size += size;
+			}
+		}
+
+		size = (prange->last - prange->start + 1) << PAGE_SHIFT;
+		if (xnack_enabled) {
+			amdgpu_amdkfd_unreserve_mem_limit(NULL, size,
+					KFD_IOC_ALLOC_MEM_FLAGS_USERPTR, 0);
+		} else {
+			r = amdgpu_amdkfd_reserve_mem_limit(NULL, size,
+					KFD_IOC_ALLOC_MEM_FLAGS_USERPTR, 0);
+			if (r)
+				goto out_unlock;
+			reserved_size += size;
+		}
+out_unlock:
+		svm_range_unlock(prange);
+		if (r)
+			break;
+	}
+
+	if (r)
+		amdgpu_amdkfd_unreserve_mem_limit(NULL, reserved_size,
+					KFD_IOC_ALLOC_MEM_FLAGS_USERPTR, 0);
+	else
+		/* Change xnack mode must be inside svms lock, to avoid race with
+		 * svm_range_deferred_list_work unreserve memory in parallel.
+		 */
+		p->xnack_enabled = xnack_enabled;
+
+	mutex_unlock(&p->svms.lock);
+	return r;
+}
+
+void svm_range_list_fini(struct kfd_process *p)
+{
+	struct svm_range *prange;
+	struct svm_range *next;
+
+	pr_debug("process pid %d svms 0x%p\n", p->lead_thread->pid,
+		 &p->svms);
+
+	cancel_delayed_work_sync(&p->svms.restore_work);
+
+	/* Ensure list work is finished before process is destroyed */
+	flush_work(&p->svms.deferred_list_work);
+
+	/*
+	 * Ensure no retry fault comes in afterwards, as page fault handler will
+	 * not find kfd process and take mm lock to recover fault.
+	 * stop kfd page fault handing, then wait pending page faults got drained
+	 */
+	atomic_set(&p->svms.drain_pagefaults, 1);
+	svm_range_drain_retry_fault(&p->svms);
+
+	list_for_each_entry_safe(prange, next, &p->svms.list, list) {
+		svm_range_unlink(prange);
+		svm_range_remove_notifier(prange);
+		svm_range_free(prange, true);
+	}
+
+	mutex_destroy(&p->svms.lock);
+
+	pr_debug("process pid %d svms 0x%p done\n",
+		p->lead_thread->pid, &p->svms);
+}
+
+int svm_range_list_init(struct kfd_process *p)
+{
+	struct svm_range_list *svms = &p->svms;
+	int i;
+
+	svms->objects = RB_ROOT_CACHED;
+	mutex_init(&svms->lock);
+	INIT_LIST_HEAD(&svms->list);
+	atomic_set(&svms->evicted_ranges, 0);
+	atomic_set(&svms->drain_pagefaults, 0);
+	INIT_DELAYED_WORK(&svms->restore_work, svm_range_restore_work);
+	INIT_WORK(&svms->deferred_list_work, svm_range_deferred_list_work);
+	INIT_LIST_HEAD(&svms->deferred_range_list);
+	INIT_LIST_HEAD(&svms->criu_svm_metadata_list);
+	spin_lock_init(&svms->deferred_list_lock);
+
+	for (i = 0; i < p->n_pdds; i++)
+		if (KFD_IS_SVM_API_SUPPORTED(p->pdds[i]->dev->adev))
+			bitmap_set(svms->bitmap_supported, i, 1);
+
+	 /* Value of default granularity cannot exceed 0x1B, the
+	  * number of pages supported by a 4-level paging table
+	  */
+	svms->default_granularity = min_t(u8, amdgpu_svm_default_granularity, 0x1B);
+	pr_debug("Default SVM Granularity to use: %d\n", svms->default_granularity);
+
+	return 0;
+}
+
+/**
+ * svm_range_check_vm - check if virtual address range mapped already
+ * @p: current kfd_process
+ * @start: range start address, in pages
+ * @last: range last address, in pages
+ * @bo_s: mapping start address in pages if address range already mapped
+ * @bo_l: mapping last address in pages if address range already mapped
+ *
+ * The purpose is to avoid virtual address ranges already allocated by
+ * kfd_ioctl_alloc_memory_of_gpu ioctl.
+ * It looks for each pdd in the kfd_process.
+ *
+ * Context: Process context
+ *
+ * Return 0 - OK, if the range is not mapped.
+ * Otherwise error code:
+ * -EADDRINUSE - if address is mapped already by kfd_ioctl_alloc_memory_of_gpu
+ * -ERESTARTSYS - A wait for the buffer to become unreserved was interrupted by
+ * a signal. Release all buffer reservations and return to user-space.
+ */
+static int
+svm_range_check_vm(struct kfd_process *p, uint64_t start, uint64_t last,
+		   uint64_t *bo_s, uint64_t *bo_l)
+{
+	struct amdgpu_bo_va_mapping *mapping;
+	struct interval_tree_node *node;
+	uint32_t i;
+	int r;
+
+	for (i = 0; i < p->n_pdds; i++) {
+		struct amdgpu_vm *vm;
+
+		if (!p->pdds[i]->drm_priv)
+			continue;
+
+		vm = drm_priv_to_vm(p->pdds[i]->drm_priv);
+		r = amdgpu_bo_reserve(vm->root.bo, false);
+		if (r)
+			return r;
+
+		node = interval_tree_iter_first(&vm->va, start, last);
+		if (node) {
+			pr_debug("range [0x%llx 0x%llx] already TTM mapped\n",
+				 start, last);
+			mapping = container_of((struct rb_node *)node,
+					       struct amdgpu_bo_va_mapping, rb);
+			if (bo_s && bo_l) {
+				*bo_s = mapping->start;
+				*bo_l = mapping->last;
+			}
+			amdgpu_bo_unreserve(vm->root.bo);
+			return -EADDRINUSE;
+		}
+		amdgpu_bo_unreserve(vm->root.bo);
+	}
+
+	return 0;
+}
+
+/**
+ * svm_range_is_valid - check if virtual address range is valid
+ * @p: current kfd_process
+ * @start: range start address, in pages
+ * @size: range size, in pages
+ *
+ * Valid virtual address range means it belongs to one or more VMAs
+ *
+ * Context: Process context
+ *
+ * Return:
+ *  0 - OK, otherwise error code
+ */
+static int
+svm_range_is_valid(struct kfd_process *p, uint64_t start, uint64_t size)
+{
+	const unsigned long device_vma = VM_IO | VM_PFNMAP | VM_MIXEDMAP;
+	struct vm_area_struct *vma;
+	unsigned long end;
+	unsigned long start_unchg = start;
+
+	start <<= PAGE_SHIFT;
+	end = start + (size << PAGE_SHIFT);
+	do {
+		vma = vma_lookup(p->mm, start);
+		if (!vma || (vma->vm_flags & device_vma))
+			return -EFAULT;
+		start = min(end, vma->vm_end);
+	} while (start < end);
+
+	return svm_range_check_vm(p, start_unchg, (end - 1) >> PAGE_SHIFT, NULL,
+				  NULL);
+}
+
+/**
+ * svm_range_best_prefetch_location - decide the best prefetch location
+ * @prange: svm range structure
+ *
+ * For xnack off:
+ * If range map to single GPU, the best prefetch location is prefetch_loc, which
+ * can be CPU or GPU.
+ *
+ * If range is ACCESS or ACCESS_IN_PLACE by mGPUs, only if mGPU connection on
+ * XGMI same hive, the best prefetch location is prefetch_loc GPU, othervise
+ * the best prefetch location is always CPU, because GPU can not have coherent
+ * mapping VRAM of other GPUs even with large-BAR PCIe connection.
+ *
+ * For xnack on:
+ * If range is not ACCESS_IN_PLACE by mGPUs, the best prefetch location is
+ * prefetch_loc, other GPU access will generate vm fault and trigger migration.
+ *
+ * If range is ACCESS_IN_PLACE by mGPUs, only if mGPU connection on XGMI same
+ * hive, the best prefetch location is prefetch_loc GPU, otherwise the best
+ * prefetch location is always CPU.
+ *
+ * Context: Process context
+ *
+ * Return:
+ * 0 for CPU or GPU id
+ */
+static uint32_t
+svm_range_best_prefetch_location(struct svm_range *prange)
+{
+	DECLARE_BITMAP(bitmap, MAX_GPU_INSTANCE);
+	uint32_t best_loc = prange->prefetch_loc;
+	struct kfd_process_device *pdd;
+	struct kfd_node *bo_node;
+	struct kfd_process *p;
+	uint32_t gpuidx;
+
+	p = container_of(prange->svms, struct kfd_process, svms);
+
+	if (!best_loc || best_loc == KFD_IOCTL_SVM_LOCATION_UNDEFINED)
+		goto out;
+
+	bo_node = svm_range_get_node_by_id(prange, best_loc);
+	if (!bo_node) {
+		WARN_ONCE(1, "failed to get valid kfd node at id%x\n", best_loc);
+		best_loc = 0;
+		goto out;
+	}
+
+	if (bo_node->adev->apu_prefer_gtt) {
+		best_loc = 0;
+		goto out;
+	}
+
+	if (p->xnack_enabled)
+		bitmap_copy(bitmap, prange->bitmap_aip, MAX_GPU_INSTANCE);
+	else
+		bitmap_or(bitmap, prange->bitmap_access, prange->bitmap_aip,
+			  MAX_GPU_INSTANCE);
+
+	for_each_set_bit(gpuidx, bitmap, MAX_GPU_INSTANCE) {
+		pdd = kfd_process_device_from_gpuidx(p, gpuidx);
+		if (!pdd) {
+			pr_debug("failed to get device by idx 0x%x\n", gpuidx);
+			continue;
+		}
+
+		if (pdd->dev->adev == bo_node->adev)
+			continue;
+
+		if (!svm_nodes_in_same_hive(pdd->dev, bo_node)) {
+			best_loc = 0;
+			break;
+		}
+	}
+
+out:
+	pr_debug("xnack %d svms 0x%p [0x%lx 0x%lx] best loc 0x%x\n",
+		 p->xnack_enabled, &p->svms, prange->start, prange->last,
+		 best_loc);
+
+	return best_loc;
+}
+
+/* svm_range_trigger_migration - start page migration if prefetch loc changed
+ * @mm: current process mm_struct
+ * @prange: svm range structure
+ * @migrated: output, true if migration is triggered
+ *
+ * If range perfetch_loc is GPU, actual loc is cpu 0, then migrate the range
+ * from ram to vram.
+ * If range prefetch_loc is cpu 0, actual loc is GPU, then migrate the range
+ * from vram to ram.
+ *
+ * If GPU vm fault retry is not enabled, migration interact with MMU notifier
+ * and restore work:
+ * 1. migrate_vma_setup invalidate pages, MMU notifier callback svm_range_evict
+ *    stops all queues, schedule restore work
+ * 2. svm_range_restore_work wait for migration is done by
+ *    a. svm_range_validate_vram takes prange->migrate_mutex
+ *    b. svm_range_validate_ram HMM get pages wait for CPU fault handle returns
+ * 3. restore work update mappings of GPU, resume all queues.
+ *
+ * Context: Process context
+ *
+ * Return:
+ * 0 - OK, otherwise - error code of migration
+ */
+static int
+svm_range_trigger_migration(struct mm_struct *mm, struct svm_range *prange,
+			    bool *migrated)
+{
+	uint32_t best_loc;
+	int r = 0;
+
+	*migrated = false;
+	best_loc = svm_range_best_prefetch_location(prange);
+
+	/* when best_loc is a gpu node and same as prange->actual_loc
+	 * we still need do migration as prange->actual_loc !=0 does
+	 * not mean all pages in prange are vram. hmm migrate will pick
+	 * up right pages during migration.
+	 */
+	if ((best_loc == KFD_IOCTL_SVM_LOCATION_UNDEFINED) ||
+	    (best_loc == 0 && prange->actual_loc == 0))
+		return 0;
+
+	if (!best_loc) {
+		r = svm_migrate_vram_to_ram(prange, mm, prange->start, prange->last,
+					KFD_MIGRATE_TRIGGER_PREFETCH, NULL);
+		*migrated = !r;
+		return r;
+	}
+
+	r = svm_migrate_to_vram(prange, best_loc, prange->start, prange->last,
+				mm, KFD_MIGRATE_TRIGGER_PREFETCH);
+	*migrated = !r;
+
+	return 0;
+}
+
+int svm_range_schedule_evict_svm_bo(struct amdgpu_amdkfd_fence *fence)
+{
+	/* Dereferencing fence->svm_bo is safe here because the fence hasn't
+	 * signaled yet and we're under the protection of the fence->lock.
+	 * After the fence is signaled in svm_range_bo_release, we cannot get
+	 * here any more.
+	 *
+	 * Reference is dropped in svm_range_evict_svm_bo_worker.
+	 */
+	if (svm_bo_ref_unless_zero(fence->svm_bo)) {
+		WRITE_ONCE(fence->svm_bo->evicting, 1);
+		schedule_work(&fence->svm_bo->eviction_work);
+	}
+
+	return 0;
+}
+
+static void svm_range_evict_svm_bo_worker(struct work_struct *work)
+{
+	struct svm_range_bo *svm_bo;
+	struct mm_struct *mm;
+	int r = 0;
+
+	svm_bo = container_of(work, struct svm_range_bo, eviction_work);
+
+	if (mmget_not_zero(svm_bo->eviction_fence->mm)) {
+		mm = svm_bo->eviction_fence->mm;
+	} else {
+		svm_range_bo_unref(svm_bo);
+		return;
+	}
+
+	mmap_read_lock(mm);
+	spin_lock(&svm_bo->list_lock);
+	while (!list_empty(&svm_bo->range_list) && !r) {
+		struct svm_range *prange =
+				list_first_entry(&svm_bo->range_list,
+						struct svm_range, svm_bo_list);
+		int retries = 3;
+
+		list_del_init(&prange->svm_bo_list);
+		spin_unlock(&svm_bo->list_lock);
+
+		pr_debug("svms 0x%p [0x%lx 0x%lx]\n", prange->svms,
+			 prange->start, prange->last);
+
+		mutex_lock(&prange->migrate_mutex);
+		do {
+			/* migrate all vram pages in this prange to sys ram
+			 * after that prange->actual_loc should be zero
+			 */
+			r = svm_migrate_vram_to_ram(prange, mm,
+					prange->start, prange->last,
+					KFD_MIGRATE_TRIGGER_TTM_EVICTION, NULL);
+		} while (!r && prange->actual_loc && --retries);
+
+		if (!r && prange->actual_loc)
+			pr_info_once("Migration failed during eviction");
+
+		if (!prange->actual_loc) {
+			mutex_lock(&prange->lock);
+			prange->svm_bo = NULL;
+			mutex_unlock(&prange->lock);
+		}
+		mutex_unlock(&prange->migrate_mutex);
+
+		spin_lock(&svm_bo->list_lock);
+	}
+	spin_unlock(&svm_bo->list_lock);
+	mmap_read_unlock(mm);
+	mmput(mm);
+
+	dma_fence_signal(&svm_bo->eviction_fence->base);
+
+	/* This is the last reference to svm_bo, after svm_range_vram_node_free
+	 * has been called in svm_migrate_vram_to_ram
+	 */
+	WARN_ONCE(!r && kref_read(&svm_bo->kref) != 1, "This was not the last reference\n");
+	svm_range_bo_unref(svm_bo);
+}
+
+static int
+svm_range_set_attr(struct kfd_process *p, struct mm_struct *mm,
+		   uint64_t start, uint64_t size, uint32_t nattr,
+		   struct kfd_ioctl_svm_attribute *attrs)
+{
+	struct amdkfd_process_info *process_info = p->kgd_process_info;
+	struct list_head update_list;
+	struct list_head insert_list;
+	struct list_head remove_list;
+	struct list_head remap_list;
+	struct svm_range_list *svms;
+	struct svm_range *prange;
+	struct svm_range *next;
+	bool update_mapping = false;
+	bool flush_tlb;
+	int r, ret = 0;
+
+	pr_debug("process pid %d svms 0x%p [0x%llx 0x%llx] pages 0x%llx\n",
+		 p->lead_thread->pid, &p->svms, start, start + size - 1, size);
+
+	r = svm_range_check_attr(p, nattr, attrs);
+	if (r)
+		return r;
+
+	svms = &p->svms;
+
+	mutex_lock(&process_info->lock);
+
+	svm_range_list_lock_and_flush_work(svms, mm);
+
+	r = svm_range_is_valid(p, start, size);
+	if (r) {
+		pr_debug("invalid range r=%d\n", r);
+		mmap_write_unlock(mm);
+		goto out;
+	}
+
+	mutex_lock(&svms->lock);
+
+	/* Add new range and split existing ranges as needed */
+	r = svm_range_add(p, start, size, nattr, attrs, &update_list,
+			  &insert_list, &remove_list, &remap_list);
+	if (r) {
+		mutex_unlock(&svms->lock);
+		mmap_write_unlock(mm);
+		goto out;
+	}
+	/* Apply changes as a transaction */
+	list_for_each_entry_safe(prange, next, &insert_list, list) {
+		svm_range_add_to_svms(prange);
+		svm_range_add_notifier_locked(mm, prange);
+	}
+	list_for_each_entry(prange, &update_list, update_list) {
+		svm_range_apply_attrs(p, prange, nattr, attrs, &update_mapping);
+		/* TODO: unmap ranges from GPU that lost access */
+	}
+	list_for_each_entry_safe(prange, next, &remove_list, update_list) {
+		pr_debug("unlink old 0x%p prange 0x%p [0x%lx 0x%lx]\n",
+			 prange->svms, prange, prange->start,
+			 prange->last);
+		svm_range_unlink(prange);
+		svm_range_remove_notifier(prange);
+		svm_range_free(prange, false);
+	}
+
+	mmap_write_downgrade(mm);
+	/* Trigger migrations and revalidate and map to GPUs as needed. If
+	 * this fails we may be left with partially completed actions. There
+	 * is no clean way of rolling back to the previous state in such a
+	 * case because the rollback wouldn't be guaranteed to work either.
+	 */
+	list_for_each_entry(prange, &update_list, update_list) {
+		bool migrated;
+
+		mutex_lock(&prange->migrate_mutex);
+
+		r = svm_range_trigger_migration(mm, prange, &migrated);
+		if (r)
+			goto out_unlock_range;
+
+		if (migrated && (!p->xnack_enabled ||
+		    (prange->flags & KFD_IOCTL_SVM_FLAG_GPU_ALWAYS_MAPPED)) &&
+		    prange->mapped_to_gpu) {
+			pr_debug("restore_work will update mappings of GPUs\n");
+			mutex_unlock(&prange->migrate_mutex);
+			continue;
+		}
+
+		if (!migrated && !update_mapping) {
+			mutex_unlock(&prange->migrate_mutex);
+			continue;
+		}
+
+		flush_tlb = !migrated && update_mapping && prange->mapped_to_gpu;
+
+		r = svm_range_validate_and_map(mm, prange->start, prange->last, prange,
+					       MAX_GPU_INSTANCE, true, true, flush_tlb);
+		if (r)
+			pr_debug("failed %d to map svm range\n", r);
+
+out_unlock_range:
+		mutex_unlock(&prange->migrate_mutex);
+		if (r)
+			ret = r;
+	}
+
+	list_for_each_entry(prange, &remap_list, update_list) {
+		pr_debug("Remapping prange 0x%p [0x%lx 0x%lx]\n",
+			 prange, prange->start, prange->last);
+		mutex_lock(&prange->migrate_mutex);
+		r = svm_range_validate_and_map(mm,  prange->start, prange->last, prange,
+					       MAX_GPU_INSTANCE, true, true, prange->mapped_to_gpu);
+		if (r)
+			pr_debug("failed %d on remap svm range\n", r);
+		mutex_unlock(&prange->migrate_mutex);
+		if (r)
+			ret = r;
+	}
+
+	dynamic_svm_range_dump(svms);
+
+	mutex_unlock(&svms->lock);
+	mmap_read_unlock(mm);
+out:
+	mutex_unlock(&process_info->lock);
+
+	pr_debug("process pid %d svms 0x%p [0x%llx 0x%llx] done, r=%d\n",
+		 p->lead_thread->pid, &p->svms, start, start + size - 1, r);
+
+	return ret ? ret : r;
+}
+
+static int
+svm_range_get_attr(struct kfd_process *p, struct mm_struct *mm,
+		   uint64_t start, uint64_t size, uint32_t nattr,
+		   struct kfd_ioctl_svm_attribute *attrs)
+{
+	DECLARE_BITMAP(bitmap_access, MAX_GPU_INSTANCE);
+	DECLARE_BITMAP(bitmap_aip, MAX_GPU_INSTANCE);
+	bool get_preferred_loc = false;
+	bool get_prefetch_loc = false;
+	bool get_granularity = false;
+	bool get_accessible = false;
+	bool get_flags = false;
+	uint64_t last = start + size - 1UL;
+	uint8_t granularity = 0xff;
+	struct interval_tree_node *node;
+	struct svm_range_list *svms;
+	struct svm_range *prange;
+	uint32_t prefetch_loc = KFD_IOCTL_SVM_LOCATION_UNDEFINED;
+	uint32_t location = KFD_IOCTL_SVM_LOCATION_UNDEFINED;
+	uint32_t flags_and = 0xffffffff;
+	uint32_t flags_or = 0;
+	int gpuidx;
+	uint32_t i;
+	int r = 0;
+
+	pr_debug("svms 0x%p [0x%llx 0x%llx] nattr 0x%x\n", &p->svms, start,
+		 start + size - 1, nattr);
+
+	/* Flush pending deferred work to avoid racing with deferred actions from
+	 * previous memory map changes (e.g. munmap). Concurrent memory map changes
+	 * can still race with get_attr because we don't hold the mmap lock. But that
+	 * would be a race condition in the application anyway, and undefined
+	 * behaviour is acceptable in that case.
+	 */
+	flush_work(&p->svms.deferred_list_work);
+
+	mmap_read_lock(mm);
+	r = svm_range_is_valid(p, start, size);
+	mmap_read_unlock(mm);
+	if (r) {
+		pr_debug("invalid range r=%d\n", r);
+		return r;
+	}
+
+	for (i = 0; i < nattr; i++) {
+		switch (attrs[i].type) {
+		case KFD_IOCTL_SVM_ATTR_PREFERRED_LOC:
+			get_preferred_loc = true;
+			break;
+		case KFD_IOCTL_SVM_ATTR_PREFETCH_LOC:
+			get_prefetch_loc = true;
+			break;
+		case KFD_IOCTL_SVM_ATTR_ACCESS:
+			get_accessible = true;
+			break;
+		case KFD_IOCTL_SVM_ATTR_SET_FLAGS:
+		case KFD_IOCTL_SVM_ATTR_CLR_FLAGS:
+			get_flags = true;
+			break;
+		case KFD_IOCTL_SVM_ATTR_GRANULARITY:
+			get_granularity = true;
+			break;
+		case KFD_IOCTL_SVM_ATTR_ACCESS_IN_PLACE:
+		case KFD_IOCTL_SVM_ATTR_NO_ACCESS:
+			fallthrough;
+		default:
+			pr_debug("get invalid attr type 0x%x\n", attrs[i].type);
+			return -EINVAL;
+		}
+	}
+
+	svms = &p->svms;
+
+	mutex_lock(&svms->lock);
+
+	node = interval_tree_iter_first(&svms->objects, start, last);
+	if (!node) {
+		pr_debug("range attrs not found return default values\n");
+		svm_range_set_default_attributes(svms, &location, &prefetch_loc,
+						 &granularity, &flags_and);
+		flags_or = flags_and;
+		if (p->xnack_enabled)
+			bitmap_copy(bitmap_access, svms->bitmap_supported,
+				    MAX_GPU_INSTANCE);
+		else
+			bitmap_zero(bitmap_access, MAX_GPU_INSTANCE);
+		bitmap_zero(bitmap_aip, MAX_GPU_INSTANCE);
+		goto fill_values;
+	}
+	bitmap_copy(bitmap_access, svms->bitmap_supported, MAX_GPU_INSTANCE);
+	bitmap_copy(bitmap_aip, svms->bitmap_supported, MAX_GPU_INSTANCE);
+
+	while (node) {
+		struct interval_tree_node *next;
+
+		prange = container_of(node, struct svm_range, it_node);
+		next = interval_tree_iter_next(node, start, last);
+
+		if (get_preferred_loc) {
+			if (prange->preferred_loc ==
+					KFD_IOCTL_SVM_LOCATION_UNDEFINED ||
+			    (location != KFD_IOCTL_SVM_LOCATION_UNDEFINED &&
+			     location != prange->preferred_loc)) {
+				location = KFD_IOCTL_SVM_LOCATION_UNDEFINED;
+				get_preferred_loc = false;
+			} else {
+				location = prange->preferred_loc;
+			}
+		}
+		if (get_prefetch_loc) {
+			if (prange->prefetch_loc ==
+					KFD_IOCTL_SVM_LOCATION_UNDEFINED ||
+			    (prefetch_loc != KFD_IOCTL_SVM_LOCATION_UNDEFINED &&
+			     prefetch_loc != prange->prefetch_loc)) {
+				prefetch_loc = KFD_IOCTL_SVM_LOCATION_UNDEFINED;
+				get_prefetch_loc = false;
+			} else {
+				prefetch_loc = prange->prefetch_loc;
+			}
+		}
+		if (get_accessible) {
+			bitmap_and(bitmap_access, bitmap_access,
+				   prange->bitmap_access, MAX_GPU_INSTANCE);
+			bitmap_and(bitmap_aip, bitmap_aip,
+				   prange->bitmap_aip, MAX_GPU_INSTANCE);
+		}
+		if (get_flags) {
+			flags_and &= prange->flags;
+			flags_or |= prange->flags;
+		}
+
+		if (get_granularity && prange->granularity < granularity)
+			granularity = prange->granularity;
+
+		node = next;
+	}
+fill_values:
+	mutex_unlock(&svms->lock);
+
+	for (i = 0; i < nattr; i++) {
+		switch (attrs[i].type) {
+		case KFD_IOCTL_SVM_ATTR_PREFERRED_LOC:
+			attrs[i].value = location;
+			break;
+		case KFD_IOCTL_SVM_ATTR_PREFETCH_LOC:
+			attrs[i].value = prefetch_loc;
+			break;
+		case KFD_IOCTL_SVM_ATTR_ACCESS:
+			gpuidx = kfd_process_gpuidx_from_gpuid(p,
+							       attrs[i].value);
+			if (gpuidx < 0) {
+				pr_debug("invalid gpuid %x\n", attrs[i].value);
+				return -EINVAL;
+			}
+			if (test_bit(gpuidx, bitmap_access))
+				attrs[i].type = KFD_IOCTL_SVM_ATTR_ACCESS;
+			else if (test_bit(gpuidx, bitmap_aip))
+				attrs[i].type =
+					KFD_IOCTL_SVM_ATTR_ACCESS_IN_PLACE;
+			else
+				attrs[i].type = KFD_IOCTL_SVM_ATTR_NO_ACCESS;
+			break;
+		case KFD_IOCTL_SVM_ATTR_SET_FLAGS:
+			attrs[i].value = flags_and;
+			break;
+		case KFD_IOCTL_SVM_ATTR_CLR_FLAGS:
+			attrs[i].value = ~flags_or;
+			break;
+		case KFD_IOCTL_SVM_ATTR_GRANULARITY:
+			attrs[i].value = (uint32_t)granularity;
+			break;
+		}
+	}
+
+	return 0;
+}
+
+int kfd_criu_resume_svm(struct kfd_process *p)
+{
+	struct kfd_ioctl_svm_attribute *set_attr_new, *set_attr = NULL;
+	int nattr_common = 4, nattr_accessibility = 1;
+	struct criu_svm_metadata *criu_svm_md = NULL;
+	struct svm_range_list *svms = &p->svms;
+	struct criu_svm_metadata *next = NULL;
+	uint32_t set_flags = 0xffffffff;
+	int i, j, num_attrs, ret = 0;
+	uint64_t set_attr_size;
+	struct mm_struct *mm;
+
+	if (list_empty(&svms->criu_svm_metadata_list)) {
+		pr_debug("No SVM data from CRIU restore stage 2\n");
+		return ret;
+	}
+
+	mm = get_task_mm(p->lead_thread);
+	if (!mm) {
+		pr_err("failed to get mm for the target process\n");
+		return -ESRCH;
+	}
+
+	num_attrs = nattr_common + (nattr_accessibility * p->n_pdds);
+
+	i = j = 0;
+	list_for_each_entry(criu_svm_md, &svms->criu_svm_metadata_list, list) {
+		pr_debug("criu_svm_md[%d]\n\tstart: 0x%llx size: 0x%llx (npages)\n",
+			 i, criu_svm_md->data.start_addr, criu_svm_md->data.size);
+
+		for (j = 0; j < num_attrs; j++) {
+			pr_debug("\ncriu_svm_md[%d]->attrs[%d].type : 0x%x\ncriu_svm_md[%d]->attrs[%d].value : 0x%x\n",
+				 i, j, criu_svm_md->data.attrs[j].type,
+				 i, j, criu_svm_md->data.attrs[j].value);
+			switch (criu_svm_md->data.attrs[j].type) {
+			/* During Checkpoint operation, the query for
+			 * KFD_IOCTL_SVM_ATTR_PREFETCH_LOC attribute might
+			 * return KFD_IOCTL_SVM_LOCATION_UNDEFINED if they were
+			 * not used by the range which was checkpointed. Care
+			 * must be taken to not restore with an invalid value
+			 * otherwise the gpuidx value will be invalid and
+			 * set_attr would eventually fail so just replace those
+			 * with another dummy attribute such as
+			 * KFD_IOCTL_SVM_ATTR_SET_FLAGS.
+			 */
+			case KFD_IOCTL_SVM_ATTR_PREFETCH_LOC:
+				if (criu_svm_md->data.attrs[j].value ==
+				    KFD_IOCTL_SVM_LOCATION_UNDEFINED) {
+					criu_svm_md->data.attrs[j].type =
+						KFD_IOCTL_SVM_ATTR_SET_FLAGS;
+					criu_svm_md->data.attrs[j].value = 0;
+				}
+				break;
+			case KFD_IOCTL_SVM_ATTR_SET_FLAGS:
+				set_flags = criu_svm_md->data.attrs[j].value;
+				break;
+			default:
+				break;
+			}
+		}
+
+		/* CLR_FLAGS is not available via get_attr during checkpoint but
+		 * it needs to be inserted before restoring the ranges so
+		 * allocate extra space for it before calling set_attr
+		 */
+		set_attr_size = sizeof(struct kfd_ioctl_svm_attribute) *
+						(num_attrs + 1);
+		set_attr_new = krealloc(set_attr, set_attr_size,
+					    GFP_KERNEL);
+		if (!set_attr_new) {
+			ret = -ENOMEM;
+			goto exit;
+		}
+		set_attr = set_attr_new;
+
+		memcpy(set_attr, criu_svm_md->data.attrs, num_attrs *
+					sizeof(struct kfd_ioctl_svm_attribute));
+		set_attr[num_attrs].type = KFD_IOCTL_SVM_ATTR_CLR_FLAGS;
+		set_attr[num_attrs].value = ~set_flags;
+
+		ret = svm_range_set_attr(p, mm, criu_svm_md->data.start_addr,
+					 criu_svm_md->data.size, num_attrs + 1,
+					 set_attr);
+		if (ret) {
+			pr_err("CRIU: failed to set range attributes\n");
+			goto exit;
+		}
+
+		i++;
+	}
+exit:
+	kfree(set_attr);
+	list_for_each_entry_safe(criu_svm_md, next, &svms->criu_svm_metadata_list, list) {
+		pr_debug("freeing criu_svm_md[]\n\tstart: 0x%llx\n",
+						criu_svm_md->data.start_addr);
+		kfree(criu_svm_md);
+	}
+
+	mmput(mm);
+	return ret;
+
+}
+
+int kfd_criu_restore_svm(struct kfd_process *p,
+			 uint8_t __user *user_priv_ptr,
+			 uint64_t *priv_data_offset,
+			 uint64_t max_priv_data_size)
+{
+	uint64_t svm_priv_data_size, svm_object_md_size, svm_attrs_size;
+	int nattr_common = 4, nattr_accessibility = 1;
+	struct criu_svm_metadata *criu_svm_md = NULL;
+	struct svm_range_list *svms = &p->svms;
+	uint32_t num_devices;
+	int ret = 0;
+
+	num_devices = p->n_pdds;
+	/* Handle one SVM range object at a time, also the number of gpus are
+	 * assumed to be same on the restore node, checking must be done while
+	 * evaluating the topology earlier
+	 */
+
+	svm_attrs_size = sizeof(struct kfd_ioctl_svm_attribute) *
+		(nattr_common + nattr_accessibility * num_devices);
+	svm_object_md_size = sizeof(struct criu_svm_metadata) + svm_attrs_size;
+
+	svm_priv_data_size = sizeof(struct kfd_criu_svm_range_priv_data) +
+								svm_attrs_size;
+
+	criu_svm_md = kzalloc(svm_object_md_size, GFP_KERNEL);
+	if (!criu_svm_md) {
+		pr_err("failed to allocate memory to store svm metadata\n");
+		return -ENOMEM;
+	}
+	if (*priv_data_offset + svm_priv_data_size > max_priv_data_size) {
+		ret = -EINVAL;
+		goto exit;
+	}
+
+	ret = copy_from_user(&criu_svm_md->data, user_priv_ptr + *priv_data_offset,
+			     svm_priv_data_size);
+	if (ret) {
+		ret = -EFAULT;
+		goto exit;
+	}
+	*priv_data_offset += svm_priv_data_size;
+
+	list_add_tail(&criu_svm_md->list, &svms->criu_svm_metadata_list);
+
+	return 0;
+
+
+exit:
+	kfree(criu_svm_md);
+	return ret;
+}
+
+void svm_range_get_info(struct kfd_process *p, uint32_t *num_svm_ranges,
+			uint64_t *svm_priv_data_size)
+{
+	uint64_t total_size, accessibility_size, common_attr_size;
+	int nattr_common = 4, nattr_accessibility = 1;
+	int num_devices = p->n_pdds;
+	struct svm_range_list *svms;
+	struct svm_range *prange;
+	uint32_t count = 0;
+
+	*svm_priv_data_size = 0;
+
+	svms = &p->svms;
+
+	mutex_lock(&svms->lock);
+	list_for_each_entry(prange, &svms->list, list) {
+		pr_debug("prange: 0x%p start: 0x%lx\t npages: 0x%llx\t end: 0x%llx\n",
+			 prange, prange->start, prange->npages,
+			 prange->start + prange->npages - 1);
+		count++;
+	}
+	mutex_unlock(&svms->lock);
+
+	*num_svm_ranges = count;
+	/* Only the accessbility attributes need to be queried for all the gpus
+	 * individually, remaining ones are spanned across the entire process
+	 * regardless of the various gpu nodes. Of the remaining attributes,
+	 * KFD_IOCTL_SVM_ATTR_CLR_FLAGS need not be saved.
+	 *
+	 * KFD_IOCTL_SVM_ATTR_PREFERRED_LOC
+	 * KFD_IOCTL_SVM_ATTR_PREFETCH_LOC
+	 * KFD_IOCTL_SVM_ATTR_SET_FLAGS
+	 * KFD_IOCTL_SVM_ATTR_GRANULARITY
+	 *
+	 * ** ACCESSBILITY ATTRIBUTES **
+	 * (Considered as one, type is altered during query, value is gpuid)
+	 * KFD_IOCTL_SVM_ATTR_ACCESS
+	 * KFD_IOCTL_SVM_ATTR_ACCESS_IN_PLACE
+	 * KFD_IOCTL_SVM_ATTR_NO_ACCESS
+	 */
+	if (*num_svm_ranges > 0) {
+		common_attr_size = sizeof(struct kfd_ioctl_svm_attribute) *
+			nattr_common;
+		accessibility_size = sizeof(struct kfd_ioctl_svm_attribute) *
+			nattr_accessibility * num_devices;
+
+		total_size = sizeof(struct kfd_criu_svm_range_priv_data) +
+			common_attr_size + accessibility_size;
+
+		*svm_priv_data_size = *num_svm_ranges * total_size;
+	}
+
+	pr_debug("num_svm_ranges %u total_priv_size %llu\n", *num_svm_ranges,
+		 *svm_priv_data_size);
+}
+
+int kfd_criu_checkpoint_svm(struct kfd_process *p,
+			    uint8_t __user *user_priv_data,
+			    uint64_t *priv_data_offset)
+{
+	struct kfd_criu_svm_range_priv_data *svm_priv = NULL;
+	struct kfd_ioctl_svm_attribute *query_attr = NULL;
+	uint64_t svm_priv_data_size, query_attr_size = 0;
+	int index, nattr_common = 4, ret = 0;
+	struct svm_range_list *svms;
+	int num_devices = p->n_pdds;
+	struct svm_range *prange;
+	struct mm_struct *mm;
+
+	svms = &p->svms;
+
+	mm = get_task_mm(p->lead_thread);
+	if (!mm) {
+		pr_err("failed to get mm for the target process\n");
+		return -ESRCH;
+	}
+
+	query_attr_size = sizeof(struct kfd_ioctl_svm_attribute) *
+				(nattr_common + num_devices);
+
+	query_attr = kzalloc(query_attr_size, GFP_KERNEL);
+	if (!query_attr) {
+		ret = -ENOMEM;
+		goto exit;
+	}
+
+	query_attr[0].type = KFD_IOCTL_SVM_ATTR_PREFERRED_LOC;
+	query_attr[1].type = KFD_IOCTL_SVM_ATTR_PREFETCH_LOC;
+	query_attr[2].type = KFD_IOCTL_SVM_ATTR_SET_FLAGS;
+	query_attr[3].type = KFD_IOCTL_SVM_ATTR_GRANULARITY;
+
+	for (index = 0; index < num_devices; index++) {
+		struct kfd_process_device *pdd = p->pdds[index];
+
+		query_attr[index + nattr_common].type =
+			KFD_IOCTL_SVM_ATTR_ACCESS;
+		query_attr[index + nattr_common].value = pdd->user_gpu_id;
+	}
+
+	svm_priv_data_size = sizeof(*svm_priv) + query_attr_size;
+
+	svm_priv = kzalloc(svm_priv_data_size, GFP_KERNEL);
+	if (!svm_priv) {
+		ret = -ENOMEM;
+		goto exit_query;
+	}
+
+	index = 0;
+	list_for_each_entry(prange, &svms->list, list) {
+
+		svm_priv->object_type = KFD_CRIU_OBJECT_TYPE_SVM_RANGE;
+		svm_priv->start_addr = prange->start;
+		svm_priv->size = prange->npages;
+		memcpy(&svm_priv->attrs, query_attr, query_attr_size);
+		pr_debug("CRIU: prange: 0x%p start: 0x%lx\t npages: 0x%llx end: 0x%llx\t size: 0x%llx\n",
+			 prange, prange->start, prange->npages,
+			 prange->start + prange->npages - 1,
+			 prange->npages * PAGE_SIZE);
+
+		ret = svm_range_get_attr(p, mm, svm_priv->start_addr,
+					 svm_priv->size,
+					 (nattr_common + num_devices),
+					 svm_priv->attrs);
+		if (ret) {
+			pr_err("CRIU: failed to obtain range attributes\n");
+			goto exit_priv;
+		}
+
+		if (copy_to_user(user_priv_data + *priv_data_offset, svm_priv,
+				 svm_priv_data_size)) {
+			pr_err("Failed to copy svm priv to user\n");
+			ret = -EFAULT;
+			goto exit_priv;
+		}
+
+		*priv_data_offset += svm_priv_data_size;
+
+	}
+
+
+exit_priv:
+	kfree(svm_priv);
+exit_query:
+	kfree(query_attr);
+exit:
+	mmput(mm);
+	return ret;
+}
+
+int
+svm_ioctl(struct kfd_process *p, enum kfd_ioctl_svm_op op, uint64_t start,
+	  uint64_t size, uint32_t nattrs, struct kfd_ioctl_svm_attribute *attrs)
+{
+	struct mm_struct *mm = current->mm;
+	int r;
+
+	start >>= PAGE_SHIFT;
+	size >>= PAGE_SHIFT;
+
+	switch (op) {
+	case KFD_IOCTL_SVM_OP_SET_ATTR:
+		r = svm_range_set_attr(p, mm, start, size, nattrs, attrs);
+		break;
+	case KFD_IOCTL_SVM_OP_GET_ATTR:
+		r = svm_range_get_attr(p, mm, start, size, nattrs, attrs);
+		break;
+	default:
+		r = -EINVAL;
+		break;
+	}
+
+	return r;
+}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.h b/drivers/gpu/drm/amd/amdkfd/kfd_svm.h
new file mode 100644
index 000000000000..01c7a4877904
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.h
@@ -0,0 +1,276 @@
+/* SPDX-License-Identifier: GPL-2.0 OR MIT */
+/*
+ * Copyright 2020-2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef KFD_SVM_H_
+#define KFD_SVM_H_
+
+#if IS_ENABLED(CONFIG_HSA_AMD_SVM)
+
+#include <linux/rwsem.h>
+#include <linux/list.h>
+#include <linux/mutex.h>
+#include <linux/sched/mm.h>
+#include <linux/hmm.h>
+#include "amdgpu.h"
+#include "kfd_priv.h"
+
+#define SVM_RANGE_VRAM_DOMAIN (1UL << 0)
+#define SVM_ADEV_PGMAP_OWNER(adev)\
+			((adev)->hive ? (void *)(adev)->hive : (void *)(adev))
+
+struct svm_range_bo {
+	struct amdgpu_bo		*bo;
+	struct kref			kref;
+	struct list_head		range_list; /* all svm ranges shared this bo */
+	spinlock_t			list_lock;
+	struct amdgpu_amdkfd_fence	*eviction_fence;
+	struct work_struct		eviction_work;
+	uint32_t			evicting;
+	struct work_struct		release_work;
+	struct kfd_node			*node;
+};
+
+enum svm_work_list_ops {
+	SVM_OP_NULL,
+	SVM_OP_UNMAP_RANGE,
+	SVM_OP_UPDATE_RANGE_NOTIFIER,
+	SVM_OP_UPDATE_RANGE_NOTIFIER_AND_MAP,
+	SVM_OP_ADD_RANGE,
+	SVM_OP_ADD_RANGE_AND_MAP
+};
+
+struct svm_work_list_item {
+	enum svm_work_list_ops op;
+	struct mm_struct *mm;
+};
+
+/**
+ * struct svm_range - shared virtual memory range
+ *
+ * @svms:       list of svm ranges, structure defined in kfd_process
+ * @migrate_mutex: to serialize range migration, validation and mapping update
+ * @start:      range start address in pages
+ * @last:       range last address in pages
+ * @it_node:    node [start, last] stored in interval tree, start, last are page
+ *              aligned, page size is (last - start + 1)
+ * @list:       link list node, used to scan all ranges of svms
+ * @update_list:link list node used to add to update_list
+ * @mapping:    bo_va mapping structure to create and update GPU page table
+ * @npages:     number of pages
+ * @vram_pages: vram pages number in this svm_range
+ * @dma_addr:   dma mapping address on each GPU for system memory physical page
+ * @ttm_res:    vram ttm resource map
+ * @offset:     range start offset within mm_nodes
+ * @svm_bo:     struct to manage splited amdgpu_bo
+ * @svm_bo_list:link list node, to scan all ranges which share same svm_bo
+ * @lock:       protect prange start, last, child_list, svm_bo_list
+ * @saved_flags:save/restore current PF_MEMALLOC flags
+ * @flags:      flags defined as KFD_IOCTL_SVM_FLAG_*
+ * @perferred_loc: perferred location, 0 for CPU, or GPU id
+ * @perfetch_loc: last prefetch location, 0 for CPU, or GPU id
+ * @actual_loc: this svm_range location. 0: all pages are from sys ram;
+ *              GPU id: this svm_range may include vram pages from GPU with
+ *              id actual_loc.
+ * @granularity:migration granularity, log2 num pages
+ * @invalid:    not 0 means cpu page table is invalidated
+ * @validate_timestamp: system timestamp when range is validated
+ * @notifier:   register mmu interval notifier
+ * @work_item:  deferred work item information
+ * @deferred_list: list header used to add range to deferred list
+ * @child_list: list header for split ranges which are not added to svms yet
+ * @bitmap_access: index bitmap of GPUs which can access the range
+ * @bitmap_aip: index bitmap of GPUs which can access the range in place
+ *
+ * Data structure for virtual memory range shared by CPU and GPUs, it can be
+ * allocated from system memory ram or device vram, and migrate from ram to vram
+ * or from vram to ram.
+ */
+struct svm_range {
+	struct svm_range_list		*svms;
+	struct mutex			migrate_mutex;
+	unsigned long			start;
+	unsigned long			last;
+	struct interval_tree_node	it_node;
+	struct list_head		list;
+	struct list_head		update_list;
+	uint64_t			npages;
+	uint64_t			vram_pages;
+	dma_addr_t			*dma_addr[MAX_GPU_INSTANCE];
+	struct ttm_resource		*ttm_res;
+	uint64_t			offset;
+	struct svm_range_bo		*svm_bo;
+	struct list_head		svm_bo_list;
+	struct mutex                    lock;
+	unsigned int                    saved_flags;
+	uint32_t			flags;
+	uint32_t			preferred_loc;
+	uint32_t			prefetch_loc;
+	uint32_t			actual_loc;
+	uint8_t				granularity;
+	atomic_t			invalid;
+	ktime_t				validate_timestamp;
+	struct mmu_interval_notifier	notifier;
+	struct svm_work_list_item	work_item;
+	struct list_head		deferred_list;
+	struct list_head		child_list;
+	DECLARE_BITMAP(bitmap_access, MAX_GPU_INSTANCE);
+	DECLARE_BITMAP(bitmap_aip, MAX_GPU_INSTANCE);
+	bool				mapped_to_gpu;
+	atomic_t			queue_refcount;
+};
+
+static inline void svm_range_lock(struct svm_range *prange)
+{
+	mutex_lock(&prange->lock);
+	prange->saved_flags = memalloc_noreclaim_save();
+
+}
+static inline void svm_range_unlock(struct svm_range *prange)
+{
+	memalloc_noreclaim_restore(prange->saved_flags);
+	mutex_unlock(&prange->lock);
+}
+
+static inline struct svm_range_bo *svm_range_bo_ref(struct svm_range_bo *svm_bo)
+{
+	if (svm_bo)
+		kref_get(&svm_bo->kref);
+
+	return svm_bo;
+}
+
+int svm_range_list_init(struct kfd_process *p);
+void svm_range_list_fini(struct kfd_process *p);
+int svm_ioctl(struct kfd_process *p, enum kfd_ioctl_svm_op op, uint64_t start,
+	      uint64_t size, uint32_t nattrs,
+	      struct kfd_ioctl_svm_attribute *attrs);
+struct svm_range *svm_range_from_addr(struct svm_range_list *svms,
+				      unsigned long addr,
+				      struct svm_range **parent);
+struct kfd_node *svm_range_get_node_by_id(struct svm_range *prange,
+					  uint32_t gpu_id);
+int svm_range_vram_node_new(struct kfd_node *node, struct svm_range *prange,
+			    bool clear);
+void svm_range_vram_node_free(struct svm_range *prange);
+int svm_range_restore_pages(struct amdgpu_device *adev, unsigned int pasid,
+			    uint32_t vmid, uint32_t node_id, uint64_t addr, uint64_t ts,
+			    bool write_fault);
+int svm_range_schedule_evict_svm_bo(struct amdgpu_amdkfd_fence *fence);
+void svm_range_add_list_work(struct svm_range_list *svms,
+			     struct svm_range *prange, struct mm_struct *mm,
+			     enum svm_work_list_ops op);
+void schedule_deferred_list_work(struct svm_range_list *svms);
+void svm_range_dma_unmap_dev(struct device *dev, dma_addr_t *dma_addr,
+			 unsigned long offset, unsigned long npages);
+void svm_range_dma_unmap(struct svm_range *prange);
+void svm_range_get_info(struct kfd_process *p, uint32_t *num_svm_ranges,
+			uint64_t *svm_priv_data_size);
+int kfd_criu_checkpoint_svm(struct kfd_process *p,
+			    uint8_t __user *user_priv_data,
+			    uint64_t *priv_offset);
+int kfd_criu_restore_svm(struct kfd_process *p,
+			 uint8_t __user *user_priv_ptr,
+			 uint64_t *priv_data_offset,
+			 uint64_t max_priv_data_size);
+int kfd_criu_resume_svm(struct kfd_process *p);
+struct kfd_process_device *
+svm_range_get_pdd_by_node(struct svm_range *prange, struct kfd_node *node);
+void svm_range_list_lock_and_flush_work(struct svm_range_list *svms, struct mm_struct *mm);
+
+/* SVM API and HMM page migration work together, device memory type
+ * is initialized to not 0 when page migration register device memory.
+ */
+#define KFD_IS_SVM_API_SUPPORTED(adev) ((adev)->kfd.pgmap.type != 0 ||\
+					((adev)->apu_prefer_gtt))
+
+void svm_range_bo_unref_async(struct svm_range_bo *svm_bo);
+
+void svm_range_set_max_pages(struct amdgpu_device *adev);
+int svm_range_switch_xnack_reserve_mem(struct kfd_process *p, bool xnack_enabled);
+
+#else
+
+struct kfd_process;
+
+static inline int svm_range_list_init(struct kfd_process *p)
+{
+	return 0;
+}
+static inline void svm_range_list_fini(struct kfd_process *p)
+{
+	/* empty */
+}
+
+static inline int svm_range_restore_pages(struct amdgpu_device *adev,
+					  unsigned int pasid,
+					  uint32_t client_id, uint32_t node_id,
+					  uint64_t addr, uint64_t ts, bool write_fault)
+{
+	return -EFAULT;
+}
+
+static inline int svm_range_schedule_evict_svm_bo(
+		struct amdgpu_amdkfd_fence *fence)
+{
+	WARN_ONCE(1, "SVM eviction fence triggered, but SVM is disabled");
+	return -EINVAL;
+}
+
+static inline void svm_range_get_info(struct kfd_process *p,
+				      uint32_t *num_svm_ranges,
+				      uint64_t *svm_priv_data_size)
+{
+	*num_svm_ranges = 0;
+	*svm_priv_data_size = 0;
+}
+
+static inline int kfd_criu_checkpoint_svm(struct kfd_process *p,
+					  uint8_t __user *user_priv_data,
+					  uint64_t *priv_offset)
+{
+	return 0;
+}
+
+static inline int kfd_criu_restore_svm(struct kfd_process *p,
+				       uint8_t __user *user_priv_ptr,
+				       uint64_t *priv_data_offset,
+				       uint64_t max_priv_data_size)
+{
+	return -EINVAL;
+}
+
+static inline int kfd_criu_resume_svm(struct kfd_process *p)
+{
+	return 0;
+}
+
+static inline void svm_range_set_max_pages(struct amdgpu_device *adev)
+{
+}
+
+#define KFD_IS_SVM_API_SUPPORTED(dev) false
+
+#endif /* IS_ENABLED(CONFIG_HSA_AMD_SVM) */
+
+#endif /* KFD_SVM_H_ */
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
index 19ce59028d6b..5c98746eb72d 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
@@ -1,5 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -28,329 +29,93 @@
 #include <linux/hash.h>
 #include <linux/cpufreq.h>
 #include <linux/log2.h>
+#include <linux/dmi.h>
+#include <linux/atomic.h>
+#include <linux/crc16.h>
 
 #include "kfd_priv.h"
 #include "kfd_crat.h"
 #include "kfd_topology.h"
-
+#include "kfd_device_queue_manager.h"
+#include "kfd_svm.h"
+#include "kfd_debug.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_ras.h"
+#include "amdgpu.h"
+
+/* topology_device_list - Master list of all topology devices */
 static struct list_head topology_device_list;
-static int topology_crat_parsed;
 static struct kfd_system_properties sys_props;
 
 static DECLARE_RWSEM(topology_lock);
+static uint32_t topology_crat_proximity_domain;
 
-struct kfd_dev *kfd_device_by_id(uint32_t gpu_id)
+struct kfd_topology_device *kfd_topology_device_by_proximity_domain_no_lock(
+						uint32_t proximity_domain)
 {
 	struct kfd_topology_device *top_dev;
-	struct kfd_dev *device = NULL;
-
-	down_read(&topology_lock);
+	struct kfd_topology_device *device = NULL;
 
 	list_for_each_entry(top_dev, &topology_device_list, list)
-		if (top_dev->gpu_id == gpu_id) {
-			device = top_dev->gpu;
+		if (top_dev->proximity_domain == proximity_domain) {
+			device = top_dev;
 			break;
 		}
 
-	up_read(&topology_lock);
-
 	return device;
 }
 
-struct kfd_dev *kfd_device_by_pci_dev(const struct pci_dev *pdev)
+struct kfd_topology_device *kfd_topology_device_by_proximity_domain(
+						uint32_t proximity_domain)
 {
-	struct kfd_topology_device *top_dev;
-	struct kfd_dev *device = NULL;
+	struct kfd_topology_device *device = NULL;
 
 	down_read(&topology_lock);
 
-	list_for_each_entry(top_dev, &topology_device_list, list)
-		if (top_dev->gpu->pdev == pdev) {
-			device = top_dev->gpu;
-			break;
-		}
-
+	device = kfd_topology_device_by_proximity_domain_no_lock(
+							proximity_domain);
 	up_read(&topology_lock);
 
 	return device;
 }
 
-static int kfd_topology_get_crat_acpi(void *crat_image, size_t *size)
+struct kfd_topology_device *kfd_topology_device_by_id(uint32_t gpu_id)
 {
-	struct acpi_table_header *crat_table;
-	acpi_status status;
-
-	if (!size)
-		return -EINVAL;
-
-	/*
-	 * Fetch the CRAT table from ACPI
-	 */
-	status = acpi_get_table(CRAT_SIGNATURE, 0, &crat_table);
-	if (status == AE_NOT_FOUND) {
-		pr_warn("CRAT table not found\n");
-		return -ENODATA;
-	} else if (ACPI_FAILURE(status)) {
-		const char *err = acpi_format_exception(status);
-
-		pr_err("CRAT table error: %s\n", err);
-		return -EINVAL;
-	}
-
-	if (*size >= crat_table->length && crat_image != NULL)
-		memcpy(crat_image, crat_table, crat_table->length);
-
-	*size = crat_table->length;
-
-	return 0;
-}
-
-static void kfd_populated_cu_info_cpu(struct kfd_topology_device *dev,
-		struct crat_subtype_computeunit *cu)
-{
-	dev->node_props.cpu_cores_count = cu->num_cpu_cores;
-	dev->node_props.cpu_core_id_base = cu->processor_id_low;
-	if (cu->hsa_capability & CRAT_CU_FLAGS_IOMMU_PRESENT)
-		dev->node_props.capability |= HSA_CAP_ATS_PRESENT;
-
-	pr_info("CU CPU: cores=%d id_base=%d\n", cu->num_cpu_cores,
-			cu->processor_id_low);
-}
-
-static void kfd_populated_cu_info_gpu(struct kfd_topology_device *dev,
-		struct crat_subtype_computeunit *cu)
-{
-	dev->node_props.simd_id_base = cu->processor_id_low;
-	dev->node_props.simd_count = cu->num_simd_cores;
-	dev->node_props.lds_size_in_kb = cu->lds_size_in_kb;
-	dev->node_props.max_waves_per_simd = cu->max_waves_simd;
-	dev->node_props.wave_front_size = cu->wave_front_size;
-	dev->node_props.mem_banks_count = cu->num_banks;
-	dev->node_props.array_count = cu->num_arrays;
-	dev->node_props.cu_per_simd_array = cu->num_cu_per_array;
-	dev->node_props.simd_per_cu = cu->num_simd_per_cu;
-	dev->node_props.max_slots_scratch_cu = cu->max_slots_scatch_cu;
-	if (cu->hsa_capability & CRAT_CU_FLAGS_HOT_PLUGGABLE)
-		dev->node_props.capability |= HSA_CAP_HOT_PLUGGABLE;
-	pr_info("CU GPU: simds=%d id_base=%d\n", cu->num_simd_cores,
-				cu->processor_id_low);
-}
-
-/* kfd_parse_subtype_cu is called when the topology mutex is already acquired */
-static int kfd_parse_subtype_cu(struct crat_subtype_computeunit *cu)
-{
-	struct kfd_topology_device *dev;
-	int i = 0;
-
-	pr_info("Found CU entry in CRAT table with proximity_domain=%d caps=%x\n",
-			cu->proximity_domain, cu->hsa_capability);
-	list_for_each_entry(dev, &topology_device_list, list) {
-		if (cu->proximity_domain == i) {
-			if (cu->flags & CRAT_CU_FLAGS_CPU_PRESENT)
-				kfd_populated_cu_info_cpu(dev, cu);
-
-			if (cu->flags & CRAT_CU_FLAGS_GPU_PRESENT)
-				kfd_populated_cu_info_gpu(dev, cu);
-			break;
-		}
-		i++;
-	}
+	struct kfd_topology_device *top_dev = NULL;
+	struct kfd_topology_device *ret = NULL;
 
-	return 0;
-}
-
-/*
- * kfd_parse_subtype_mem is called when the topology mutex is
- * already acquired
- */
-static int kfd_parse_subtype_mem(struct crat_subtype_memory *mem)
-{
-	struct kfd_mem_properties *props;
-	struct kfd_topology_device *dev;
-	int i = 0;
-
-	pr_info("Found memory entry in CRAT table with proximity_domain=%d\n",
-			mem->promixity_domain);
-	list_for_each_entry(dev, &topology_device_list, list) {
-		if (mem->promixity_domain == i) {
-			props = kfd_alloc_struct(props);
-			if (props == NULL)
-				return -ENOMEM;
-
-			if (dev->node_props.cpu_cores_count == 0)
-				props->heap_type = HSA_MEM_HEAP_TYPE_FB_PRIVATE;
-			else
-				props->heap_type = HSA_MEM_HEAP_TYPE_SYSTEM;
-
-			if (mem->flags & CRAT_MEM_FLAGS_HOT_PLUGGABLE)
-				props->flags |= HSA_MEM_FLAGS_HOT_PLUGGABLE;
-			if (mem->flags & CRAT_MEM_FLAGS_NON_VOLATILE)
-				props->flags |= HSA_MEM_FLAGS_NON_VOLATILE;
-
-			props->size_in_bytes =
-				((uint64_t)mem->length_high << 32) +
-							mem->length_low;
-			props->width = mem->width;
-
-			dev->mem_bank_count++;
-			list_add_tail(&props->list, &dev->mem_props);
-
-			break;
-		}
-		i++;
-	}
-
-	return 0;
-}
-
-/*
- * kfd_parse_subtype_cache is called when the topology mutex
- * is already acquired
- */
-static int kfd_parse_subtype_cache(struct crat_subtype_cache *cache)
-{
-	struct kfd_cache_properties *props;
-	struct kfd_topology_device *dev;
-	uint32_t id;
-
-	id = cache->processor_id_low;
-
-	pr_info("Found cache entry in CRAT table with processor_id=%d\n", id);
-	list_for_each_entry(dev, &topology_device_list, list)
-		if (id == dev->node_props.cpu_core_id_base ||
-		    id == dev->node_props.simd_id_base) {
-			props = kfd_alloc_struct(props);
-			if (props == NULL)
-				return -ENOMEM;
-
-			props->processor_id_low = id;
-			props->cache_level = cache->cache_level;
-			props->cache_size = cache->cache_size;
-			props->cacheline_size = cache->cache_line_size;
-			props->cachelines_per_tag = cache->lines_per_tag;
-			props->cache_assoc = cache->associativity;
-			props->cache_latency = cache->cache_latency;
-
-			if (cache->flags & CRAT_CACHE_FLAGS_DATA_CACHE)
-				props->cache_type |= HSA_CACHE_TYPE_DATA;
-			if (cache->flags & CRAT_CACHE_FLAGS_INST_CACHE)
-				props->cache_type |= HSA_CACHE_TYPE_INSTRUCTION;
-			if (cache->flags & CRAT_CACHE_FLAGS_CPU_CACHE)
-				props->cache_type |= HSA_CACHE_TYPE_CPU;
-			if (cache->flags & CRAT_CACHE_FLAGS_SIMD_CACHE)
-				props->cache_type |= HSA_CACHE_TYPE_HSACU;
-
-			dev->cache_count++;
-			dev->node_props.caches_count++;
-			list_add_tail(&props->list, &dev->cache_props);
+	down_read(&topology_lock);
 
+	list_for_each_entry(top_dev, &topology_device_list, list)
+		if (top_dev->gpu_id == gpu_id) {
+			ret = top_dev;
 			break;
 		}
 
-	return 0;
-}
-
-/*
- * kfd_parse_subtype_iolink is called when the topology mutex
- * is already acquired
- */
-static int kfd_parse_subtype_iolink(struct crat_subtype_iolink *iolink)
-{
-	struct kfd_iolink_properties *props;
-	struct kfd_topology_device *dev;
-	uint32_t i = 0;
-	uint32_t id_from;
-	uint32_t id_to;
-
-	id_from = iolink->proximity_domain_from;
-	id_to = iolink->proximity_domain_to;
-
-	pr_info("Found IO link entry in CRAT table with id_from=%d\n", id_from);
-	list_for_each_entry(dev, &topology_device_list, list) {
-		if (id_from == i) {
-			props = kfd_alloc_struct(props);
-			if (props == NULL)
-				return -ENOMEM;
-
-			props->node_from = id_from;
-			props->node_to = id_to;
-			props->ver_maj = iolink->version_major;
-			props->ver_min = iolink->version_minor;
-
-			/*
-			 * weight factor (derived from CDIR), currently always 1
-			 */
-			props->weight = 1;
-
-			props->min_latency = iolink->minimum_latency;
-			props->max_latency = iolink->maximum_latency;
-			props->min_bandwidth = iolink->minimum_bandwidth_mbs;
-			props->max_bandwidth = iolink->maximum_bandwidth_mbs;
-			props->rec_transfer_size =
-					iolink->recommended_transfer_size;
-
-			dev->io_link_count++;
-			dev->node_props.io_links_count++;
-			list_add_tail(&props->list, &dev->io_link_props);
-
-			break;
-		}
-		i++;
-	}
+	up_read(&topology_lock);
 
-	return 0;
+	return ret;
 }
 
-static int kfd_parse_subtype(struct crat_subtype_generic *sub_type_hdr)
+struct kfd_node *kfd_device_by_id(uint32_t gpu_id)
 {
-	struct crat_subtype_computeunit *cu;
-	struct crat_subtype_memory *mem;
-	struct crat_subtype_cache *cache;
-	struct crat_subtype_iolink *iolink;
-	int ret = 0;
+	struct kfd_topology_device *top_dev;
 
-	switch (sub_type_hdr->type) {
-	case CRAT_SUBTYPE_COMPUTEUNIT_AFFINITY:
-		cu = (struct crat_subtype_computeunit *)sub_type_hdr;
-		ret = kfd_parse_subtype_cu(cu);
-		break;
-	case CRAT_SUBTYPE_MEMORY_AFFINITY:
-		mem = (struct crat_subtype_memory *)sub_type_hdr;
-		ret = kfd_parse_subtype_mem(mem);
-		break;
-	case CRAT_SUBTYPE_CACHE_AFFINITY:
-		cache = (struct crat_subtype_cache *)sub_type_hdr;
-		ret = kfd_parse_subtype_cache(cache);
-		break;
-	case CRAT_SUBTYPE_TLB_AFFINITY:
-		/*
-		 * For now, nothing to do here
-		 */
-		pr_info("Found TLB entry in CRAT table (not processing)\n");
-		break;
-	case CRAT_SUBTYPE_CCOMPUTE_AFFINITY:
-		/*
-		 * For now, nothing to do here
-		 */
-		pr_info("Found CCOMPUTE entry in CRAT table (not processing)\n");
-		break;
-	case CRAT_SUBTYPE_IOLINK_AFFINITY:
-		iolink = (struct crat_subtype_iolink *)sub_type_hdr;
-		ret = kfd_parse_subtype_iolink(iolink);
-		break;
-	default:
-		pr_warn("Unknown subtype (%d) in CRAT\n",
-				sub_type_hdr->type);
-	}
+	top_dev = kfd_topology_device_by_id(gpu_id);
+	if (!top_dev)
+		return NULL;
 
-	return ret;
+	return top_dev->gpu;
 }
 
+/* Called with write topology_lock acquired */
 static void kfd_release_topology_device(struct kfd_topology_device *dev)
 {
 	struct kfd_mem_properties *mem;
 	struct kfd_cache_properties *cache;
 	struct kfd_iolink_properties *iolink;
+	struct kfd_iolink_properties *p2plink;
+	struct kfd_perf_properties *perf;
 
 	list_del(&dev->list);
 
@@ -375,25 +140,42 @@ static void kfd_release_topology_device(struct kfd_topology_device *dev)
 		kfree(iolink);
 	}
 
-	kfree(dev);
+	while (dev->p2p_link_props.next != &dev->p2p_link_props) {
+		p2plink = container_of(dev->p2p_link_props.next,
+				struct kfd_iolink_properties, list);
+		list_del(&p2plink->list);
+		kfree(p2plink);
+	}
+
+	while (dev->perf_props.next != &dev->perf_props) {
+		perf = container_of(dev->perf_props.next,
+				struct kfd_perf_properties, list);
+		list_del(&perf->list);
+		kfree(perf);
+	}
 
-	sys_props.num_devices--;
+	kfree(dev);
 }
 
-static void kfd_release_live_view(void)
+void kfd_release_topology_device_list(struct list_head *device_list)
 {
 	struct kfd_topology_device *dev;
 
-	while (topology_device_list.next != &topology_device_list) {
-		dev = container_of(topology_device_list.next,
-				 struct kfd_topology_device, list);
+	while (!list_empty(device_list)) {
+		dev = list_first_entry(device_list,
+				       struct kfd_topology_device, list);
 		kfd_release_topology_device(dev);
+	}
 }
 
+static void kfd_release_live_view(void)
+{
+	kfd_release_topology_device_list(&topology_device_list);
 	memset(&sys_props, 0, sizeof(sys_props));
 }
 
-static struct kfd_topology_device *kfd_create_topology_device(void)
+struct kfd_topology_device *kfd_create_topology_device(
+				struct list_head *device_list)
 {
 	struct kfd_topology_device *dev;
 
@@ -406,216 +188,224 @@ static struct kfd_topology_device *kfd_create_topology_device(void)
 	INIT_LIST_HEAD(&dev->mem_props);
 	INIT_LIST_HEAD(&dev->cache_props);
 	INIT_LIST_HEAD(&dev->io_link_props);
+	INIT_LIST_HEAD(&dev->p2p_link_props);
+	INIT_LIST_HEAD(&dev->perf_props);
 
-	list_add_tail(&dev->list, &topology_device_list);
-	sys_props.num_devices++;
+	list_add_tail(&dev->list, device_list);
 
 	return dev;
 }
 
-static int kfd_parse_crat_table(void *crat_image)
-{
-	struct kfd_topology_device *top_dev;
-	struct crat_subtype_generic *sub_type_hdr;
-	uint16_t node_id;
-	int ret;
-	struct crat_header *crat_table = (struct crat_header *)crat_image;
-	uint16_t num_nodes;
-	uint32_t image_len;
-
-	if (!crat_image)
-		return -EINVAL;
-
-	num_nodes = crat_table->num_domains;
-	image_len = crat_table->length;
-
-	pr_info("Parsing CRAT table with %d nodes\n", num_nodes);
-
-	for (node_id = 0; node_id < num_nodes; node_id++) {
-		top_dev = kfd_create_topology_device();
-		if (!top_dev) {
-			kfd_release_live_view();
-			return -ENOMEM;
-		}
-	}
-
-	sys_props.platform_id =
-		(*((uint64_t *)crat_table->oem_id)) & CRAT_OEMID_64BIT_MASK;
-	sys_props.platform_oem = *((uint64_t *)crat_table->oem_table_id);
-	sys_props.platform_rev = crat_table->revision;
-
-	sub_type_hdr = (struct crat_subtype_generic *)(crat_table+1);
-	while ((char *)sub_type_hdr + sizeof(struct crat_subtype_generic) <
-			((char *)crat_image) + image_len) {
-		if (sub_type_hdr->flags & CRAT_SUBTYPE_FLAGS_ENABLED) {
-			ret = kfd_parse_subtype(sub_type_hdr);
-			if (ret != 0) {
-				kfd_release_live_view();
-				return ret;
-			}
-		}
-
-		sub_type_hdr = (typeof(sub_type_hdr))((char *)sub_type_hdr +
-				sub_type_hdr->length);
-	}
-
-	sys_props.generation_count++;
-	topology_crat_parsed = 1;
-
-	return 0;
-}
-
 
-#define sysfs_show_gen_prop(buffer, fmt, ...) \
-		snprintf(buffer, PAGE_SIZE, "%s"fmt, buffer, __VA_ARGS__)
-#define sysfs_show_32bit_prop(buffer, name, value) \
-		sysfs_show_gen_prop(buffer, "%s %u\n", name, value)
-#define sysfs_show_64bit_prop(buffer, name, value) \
-		sysfs_show_gen_prop(buffer, "%s %llu\n", name, value)
-#define sysfs_show_32bit_val(buffer, value) \
-		sysfs_show_gen_prop(buffer, "%u\n", value)
-#define sysfs_show_str_val(buffer, value) \
-		sysfs_show_gen_prop(buffer, "%s\n", value)
+#define sysfs_show_gen_prop(buffer, offs, fmt, ...)		\
+		(offs += snprintf(buffer+offs, PAGE_SIZE-offs,	\
+				  fmt, __VA_ARGS__))
+#define sysfs_show_32bit_prop(buffer, offs, name, value) \
+		sysfs_show_gen_prop(buffer, offs, "%s %u\n", name, value)
+#define sysfs_show_64bit_prop(buffer, offs, name, value) \
+		sysfs_show_gen_prop(buffer, offs, "%s %llu\n", name, value)
+#define sysfs_show_32bit_val(buffer, offs, value) \
+		sysfs_show_gen_prop(buffer, offs, "%u\n", value)
+#define sysfs_show_str_val(buffer, offs, value) \
+		sysfs_show_gen_prop(buffer, offs, "%s\n", value)
 
 static ssize_t sysprops_show(struct kobject *kobj, struct attribute *attr,
 		char *buffer)
 {
-	ssize_t ret;
+	int offs = 0;
 
 	/* Making sure that the buffer is an empty string */
 	buffer[0] = 0;
 
 	if (attr == &sys_props.attr_genid) {
-		ret = sysfs_show_32bit_val(buffer, sys_props.generation_count);
+		sysfs_show_32bit_val(buffer, offs,
+				     sys_props.generation_count);
 	} else if (attr == &sys_props.attr_props) {
-		sysfs_show_64bit_prop(buffer, "platform_oem",
-				sys_props.platform_oem);
-		sysfs_show_64bit_prop(buffer, "platform_id",
-				sys_props.platform_id);
-		ret = sysfs_show_64bit_prop(buffer, "platform_rev",
-				sys_props.platform_rev);
+		sysfs_show_64bit_prop(buffer, offs, "platform_oem",
+				      sys_props.platform_oem);
+		sysfs_show_64bit_prop(buffer, offs, "platform_id",
+				      sys_props.platform_id);
+		sysfs_show_64bit_prop(buffer, offs, "platform_rev",
+				      sys_props.platform_rev);
 	} else {
-		ret = -EINVAL;
+		offs = -EINVAL;
 	}
 
-	return ret;
+	return offs;
+}
+
+static void kfd_topology_kobj_release(struct kobject *kobj)
+{
+	kfree(kobj);
 }
 
 static const struct sysfs_ops sysprops_ops = {
 	.show = sysprops_show,
 };
 
-static struct kobj_type sysprops_type = {
+static const struct kobj_type sysprops_type = {
+	.release = kfd_topology_kobj_release,
 	.sysfs_ops = &sysprops_ops,
 };
 
 static ssize_t iolink_show(struct kobject *kobj, struct attribute *attr,
 		char *buffer)
 {
-	ssize_t ret;
+	int offs = 0;
 	struct kfd_iolink_properties *iolink;
 
 	/* Making sure that the buffer is an empty string */
 	buffer[0] = 0;
 
 	iolink = container_of(attr, struct kfd_iolink_properties, attr);
-	sysfs_show_32bit_prop(buffer, "type", iolink->iolink_type);
-	sysfs_show_32bit_prop(buffer, "version_major", iolink->ver_maj);
-	sysfs_show_32bit_prop(buffer, "version_minor", iolink->ver_min);
-	sysfs_show_32bit_prop(buffer, "node_from", iolink->node_from);
-	sysfs_show_32bit_prop(buffer, "node_to", iolink->node_to);
-	sysfs_show_32bit_prop(buffer, "weight", iolink->weight);
-	sysfs_show_32bit_prop(buffer, "min_latency", iolink->min_latency);
-	sysfs_show_32bit_prop(buffer, "max_latency", iolink->max_latency);
-	sysfs_show_32bit_prop(buffer, "min_bandwidth", iolink->min_bandwidth);
-	sysfs_show_32bit_prop(buffer, "max_bandwidth", iolink->max_bandwidth);
-	sysfs_show_32bit_prop(buffer, "recommended_transfer_size",
-			iolink->rec_transfer_size);
-	ret = sysfs_show_32bit_prop(buffer, "flags", iolink->flags);
-
-	return ret;
+	if (iolink->gpu && kfd_devcgroup_check_permission(iolink->gpu))
+		return -EPERM;
+	sysfs_show_32bit_prop(buffer, offs, "type", iolink->iolink_type);
+	sysfs_show_32bit_prop(buffer, offs, "version_major", iolink->ver_maj);
+	sysfs_show_32bit_prop(buffer, offs, "version_minor", iolink->ver_min);
+	sysfs_show_32bit_prop(buffer, offs, "node_from", iolink->node_from);
+	sysfs_show_32bit_prop(buffer, offs, "node_to", iolink->node_to);
+	sysfs_show_32bit_prop(buffer, offs, "weight", iolink->weight);
+	sysfs_show_32bit_prop(buffer, offs, "min_latency", iolink->min_latency);
+	sysfs_show_32bit_prop(buffer, offs, "max_latency", iolink->max_latency);
+	sysfs_show_32bit_prop(buffer, offs, "min_bandwidth",
+			      iolink->min_bandwidth);
+	sysfs_show_32bit_prop(buffer, offs, "max_bandwidth",
+			      iolink->max_bandwidth);
+	sysfs_show_32bit_prop(buffer, offs, "recommended_transfer_size",
+			      iolink->rec_transfer_size);
+	sysfs_show_32bit_prop(buffer, offs, "recommended_sdma_engine_id_mask",
+			      iolink->rec_sdma_eng_id_mask);
+	sysfs_show_32bit_prop(buffer, offs, "flags", iolink->flags);
+
+	return offs;
 }
 
 static const struct sysfs_ops iolink_ops = {
 	.show = iolink_show,
 };
 
-static struct kobj_type iolink_type = {
+static const struct kobj_type iolink_type = {
+	.release = kfd_topology_kobj_release,
 	.sysfs_ops = &iolink_ops,
 };
 
 static ssize_t mem_show(struct kobject *kobj, struct attribute *attr,
 		char *buffer)
 {
-	ssize_t ret;
+	int offs = 0;
 	struct kfd_mem_properties *mem;
 
 	/* Making sure that the buffer is an empty string */
 	buffer[0] = 0;
 
 	mem = container_of(attr, struct kfd_mem_properties, attr);
-	sysfs_show_32bit_prop(buffer, "heap_type", mem->heap_type);
-	sysfs_show_64bit_prop(buffer, "size_in_bytes", mem->size_in_bytes);
-	sysfs_show_32bit_prop(buffer, "flags", mem->flags);
-	sysfs_show_32bit_prop(buffer, "width", mem->width);
-	ret = sysfs_show_32bit_prop(buffer, "mem_clk_max", mem->mem_clk_max);
-
-	return ret;
+	if (mem->gpu && kfd_devcgroup_check_permission(mem->gpu))
+		return -EPERM;
+	sysfs_show_32bit_prop(buffer, offs, "heap_type", mem->heap_type);
+	sysfs_show_64bit_prop(buffer, offs, "size_in_bytes",
+			      mem->size_in_bytes);
+	sysfs_show_32bit_prop(buffer, offs, "flags", mem->flags);
+	sysfs_show_32bit_prop(buffer, offs, "width", mem->width);
+	sysfs_show_32bit_prop(buffer, offs, "mem_clk_max",
+			      mem->mem_clk_max);
+
+	return offs;
 }
 
 static const struct sysfs_ops mem_ops = {
 	.show = mem_show,
 };
 
-static struct kobj_type mem_type = {
+static const struct kobj_type mem_type = {
+	.release = kfd_topology_kobj_release,
 	.sysfs_ops = &mem_ops,
 };
 
 static ssize_t kfd_cache_show(struct kobject *kobj, struct attribute *attr,
 		char *buffer)
 {
-	ssize_t ret;
-	uint32_t i;
+	int offs = 0;
+	uint32_t i, j;
 	struct kfd_cache_properties *cache;
 
 	/* Making sure that the buffer is an empty string */
 	buffer[0] = 0;
-
 	cache = container_of(attr, struct kfd_cache_properties, attr);
-	sysfs_show_32bit_prop(buffer, "processor_id_low",
+	if (cache->gpu && kfd_devcgroup_check_permission(cache->gpu))
+		return -EPERM;
+	sysfs_show_32bit_prop(buffer, offs, "processor_id_low",
 			cache->processor_id_low);
-	sysfs_show_32bit_prop(buffer, "level", cache->cache_level);
-	sysfs_show_32bit_prop(buffer, "size", cache->cache_size);
-	sysfs_show_32bit_prop(buffer, "cache_line_size", cache->cacheline_size);
-	sysfs_show_32bit_prop(buffer, "cache_lines_per_tag",
-			cache->cachelines_per_tag);
-	sysfs_show_32bit_prop(buffer, "association", cache->cache_assoc);
-	sysfs_show_32bit_prop(buffer, "latency", cache->cache_latency);
-	sysfs_show_32bit_prop(buffer, "type", cache->cache_type);
-	snprintf(buffer, PAGE_SIZE, "%ssibling_map ", buffer);
-	for (i = 0; i < KFD_TOPOLOGY_CPU_SIBLINGS; i++)
-		ret = snprintf(buffer, PAGE_SIZE, "%s%d%s",
-				buffer, cache->sibling_map[i],
-				(i == KFD_TOPOLOGY_CPU_SIBLINGS-1) ?
-						"\n" : ",");
-
-	return ret;
+	sysfs_show_32bit_prop(buffer, offs, "level", cache->cache_level);
+	sysfs_show_32bit_prop(buffer, offs, "size", cache->cache_size);
+	sysfs_show_32bit_prop(buffer, offs, "cache_line_size",
+			      cache->cacheline_size);
+	sysfs_show_32bit_prop(buffer, offs, "cache_lines_per_tag",
+			      cache->cachelines_per_tag);
+	sysfs_show_32bit_prop(buffer, offs, "association", cache->cache_assoc);
+	sysfs_show_32bit_prop(buffer, offs, "latency", cache->cache_latency);
+	sysfs_show_32bit_prop(buffer, offs, "type", cache->cache_type);
+
+	offs += snprintf(buffer+offs, PAGE_SIZE-offs, "sibling_map ");
+	for (i = 0; i < cache->sibling_map_size; i++)
+		for (j = 0; j < sizeof(cache->sibling_map[0])*8; j++)
+			/* Check each bit */
+			offs += snprintf(buffer+offs, PAGE_SIZE-offs, "%d,",
+						(cache->sibling_map[i] >> j) & 1);
+
+	/* Replace the last "," with end of line */
+	buffer[offs-1] = '\n';
+	return offs;
 }
 
 static const struct sysfs_ops cache_ops = {
 	.show = kfd_cache_show,
 };
 
-static struct kobj_type cache_type = {
+static const struct kobj_type cache_type = {
+	.release = kfd_topology_kobj_release,
 	.sysfs_ops = &cache_ops,
 };
 
+/****** Sysfs of Performance Counters ******/
+
+struct kfd_perf_attr {
+	struct kobj_attribute attr;
+	uint32_t data;
+};
+
+static ssize_t perf_show(struct kobject *kobj, struct kobj_attribute *attrs,
+			char *buf)
+{
+	int offs = 0;
+	struct kfd_perf_attr *attr;
+
+	buf[0] = 0;
+	attr = container_of(attrs, struct kfd_perf_attr, attr);
+	if (!attr->data) /* invalid data for PMC */
+		return 0;
+	else
+		return sysfs_show_32bit_val(buf, offs, attr->data);
+}
+
+#define KFD_PERF_DESC(_name, _data)			\
+{							\
+	.attr  = __ATTR(_name, 0444, perf_show, NULL),	\
+	.data = _data,					\
+}
+
+static struct kfd_perf_attr perf_attr_iommu[] = {
+	KFD_PERF_DESC(max_concurrent, 0),
+	KFD_PERF_DESC(num_counters, 0),
+	KFD_PERF_DESC(counter_ids, 0),
+};
+/****************************************/
+
 static ssize_t node_show(struct kobject *kobj, struct attribute *attr,
 		char *buffer)
 {
+	int offs = 0;
 	struct kfd_topology_device *dev;
-	char public_name[KFD_TOPOLOGY_PUBLIC_NAME_SIZE];
-	uint32_t i;
 	uint32_t log_max_watch_addr;
 
 	/* Making sure that the buffer is an empty string */
@@ -624,76 +414,87 @@ static ssize_t node_show(struct kobject *kobj, struct attribute *attr,
 	if (strcmp(attr->name, "gpu_id") == 0) {
 		dev = container_of(attr, struct kfd_topology_device,
 				attr_gpuid);
-		return sysfs_show_32bit_val(buffer, dev->gpu_id);
+		if (dev->gpu && kfd_devcgroup_check_permission(dev->gpu))
+			return -EPERM;
+		return sysfs_show_32bit_val(buffer, offs, dev->gpu_id);
 	}
 
 	if (strcmp(attr->name, "name") == 0) {
 		dev = container_of(attr, struct kfd_topology_device,
 				attr_name);
-		for (i = 0; i < KFD_TOPOLOGY_PUBLIC_NAME_SIZE; i++) {
-			public_name[i] =
-					(char)dev->node_props.marketing_name[i];
-			if (dev->node_props.marketing_name[i] == 0)
-				break;
-		}
-		public_name[KFD_TOPOLOGY_PUBLIC_NAME_SIZE-1] = 0x0;
-		return sysfs_show_str_val(buffer, public_name);
+
+		if (dev->gpu && kfd_devcgroup_check_permission(dev->gpu))
+			return -EPERM;
+		return sysfs_show_str_val(buffer, offs, dev->node_props.name);
 	}
 
 	dev = container_of(attr, struct kfd_topology_device,
 			attr_props);
-	sysfs_show_32bit_prop(buffer, "cpu_cores_count",
-			dev->node_props.cpu_cores_count);
-	sysfs_show_32bit_prop(buffer, "simd_count",
-			dev->node_props.simd_count);
-
-	if (dev->mem_bank_count < dev->node_props.mem_banks_count) {
-		pr_info_once("mem_banks_count truncated from %d to %d\n",
-				dev->node_props.mem_banks_count,
-				dev->mem_bank_count);
-		sysfs_show_32bit_prop(buffer, "mem_banks_count",
-				dev->mem_bank_count);
-	} else {
-		sysfs_show_32bit_prop(buffer, "mem_banks_count",
-				dev->node_props.mem_banks_count);
-	}
-
-	sysfs_show_32bit_prop(buffer, "caches_count",
-			dev->node_props.caches_count);
-	sysfs_show_32bit_prop(buffer, "io_links_count",
-			dev->node_props.io_links_count);
-	sysfs_show_32bit_prop(buffer, "cpu_core_id_base",
-			dev->node_props.cpu_core_id_base);
-	sysfs_show_32bit_prop(buffer, "simd_id_base",
-			dev->node_props.simd_id_base);
-	sysfs_show_32bit_prop(buffer, "max_waves_per_simd",
-			dev->node_props.max_waves_per_simd);
-	sysfs_show_32bit_prop(buffer, "lds_size_in_kb",
-			dev->node_props.lds_size_in_kb);
-	sysfs_show_32bit_prop(buffer, "gds_size_in_kb",
-			dev->node_props.gds_size_in_kb);
-	sysfs_show_32bit_prop(buffer, "wave_front_size",
-			dev->node_props.wave_front_size);
-	sysfs_show_32bit_prop(buffer, "array_count",
-			dev->node_props.array_count);
-	sysfs_show_32bit_prop(buffer, "simd_arrays_per_engine",
-			dev->node_props.simd_arrays_per_engine);
-	sysfs_show_32bit_prop(buffer, "cu_per_simd_array",
-			dev->node_props.cu_per_simd_array);
-	sysfs_show_32bit_prop(buffer, "simd_per_cu",
-			dev->node_props.simd_per_cu);
-	sysfs_show_32bit_prop(buffer, "max_slots_scratch_cu",
-			dev->node_props.max_slots_scratch_cu);
-	sysfs_show_32bit_prop(buffer, "vendor_id",
-			dev->node_props.vendor_id);
-	sysfs_show_32bit_prop(buffer, "device_id",
-			dev->node_props.device_id);
-	sysfs_show_32bit_prop(buffer, "location_id",
-			dev->node_props.location_id);
+	if (dev->gpu && kfd_devcgroup_check_permission(dev->gpu))
+		return -EPERM;
+	sysfs_show_32bit_prop(buffer, offs, "cpu_cores_count",
+			      dev->node_props.cpu_cores_count);
+	sysfs_show_32bit_prop(buffer, offs, "simd_count",
+			      dev->gpu ? dev->node_props.simd_count : 0);
+	sysfs_show_32bit_prop(buffer, offs, "mem_banks_count",
+			      dev->node_props.mem_banks_count);
+	sysfs_show_32bit_prop(buffer, offs, "caches_count",
+			      dev->node_props.caches_count);
+	sysfs_show_32bit_prop(buffer, offs, "io_links_count",
+			      dev->node_props.io_links_count);
+	sysfs_show_32bit_prop(buffer, offs, "p2p_links_count",
+			      dev->node_props.p2p_links_count);
+	sysfs_show_32bit_prop(buffer, offs, "cpu_core_id_base",
+			      dev->node_props.cpu_core_id_base);
+	sysfs_show_32bit_prop(buffer, offs, "simd_id_base",
+			      dev->node_props.simd_id_base);
+	sysfs_show_32bit_prop(buffer, offs, "max_waves_per_simd",
+			      dev->node_props.max_waves_per_simd);
+	sysfs_show_32bit_prop(buffer, offs, "lds_size_in_kb",
+			      dev->node_props.lds_size_in_kb);
+	sysfs_show_32bit_prop(buffer, offs, "gds_size_in_kb",
+			      dev->node_props.gds_size_in_kb);
+	sysfs_show_32bit_prop(buffer, offs, "num_gws",
+			      dev->node_props.num_gws);
+	sysfs_show_32bit_prop(buffer, offs, "wave_front_size",
+			      dev->node_props.wave_front_size);
+	sysfs_show_32bit_prop(buffer, offs, "array_count",
+			      dev->gpu ? (dev->node_props.array_count *
+					  NUM_XCC(dev->gpu->xcc_mask)) : 0);
+	sysfs_show_32bit_prop(buffer, offs, "simd_arrays_per_engine",
+			      dev->node_props.simd_arrays_per_engine);
+	sysfs_show_32bit_prop(buffer, offs, "cu_per_simd_array",
+			      dev->node_props.cu_per_simd_array);
+	sysfs_show_32bit_prop(buffer, offs, "simd_per_cu",
+			      dev->node_props.simd_per_cu);
+	sysfs_show_32bit_prop(buffer, offs, "max_slots_scratch_cu",
+			      dev->node_props.max_slots_scratch_cu);
+	sysfs_show_32bit_prop(buffer, offs, "gfx_target_version",
+			      dev->node_props.gfx_target_version);
+	sysfs_show_32bit_prop(buffer, offs, "vendor_id",
+			      dev->node_props.vendor_id);
+	sysfs_show_32bit_prop(buffer, offs, "device_id",
+			      dev->node_props.device_id);
+	sysfs_show_32bit_prop(buffer, offs, "location_id",
+			      dev->node_props.location_id);
+	sysfs_show_32bit_prop(buffer, offs, "domain",
+			      dev->node_props.domain);
+	sysfs_show_32bit_prop(buffer, offs, "drm_render_minor",
+			      dev->node_props.drm_render_minor);
+	sysfs_show_64bit_prop(buffer, offs, "hive_id",
+			      dev->node_props.hive_id);
+	sysfs_show_32bit_prop(buffer, offs, "num_sdma_engines",
+			      dev->node_props.num_sdma_engines);
+	sysfs_show_32bit_prop(buffer, offs, "num_sdma_xgmi_engines",
+			      dev->node_props.num_sdma_xgmi_engines);
+	sysfs_show_32bit_prop(buffer, offs, "num_sdma_queues_per_engine",
+			      dev->node_props.num_sdma_queues_per_engine);
+	sysfs_show_32bit_prop(buffer, offs, "num_cp_queues",
+			      dev->node_props.num_cp_queues);
 
 	if (dev->gpu) {
 		log_max_watch_addr =
-			__ilog2_u32(dev->gpu->device_info->num_of_watch_points);
+			__ilog2_u32(dev->gpu->kfd->device_info.num_of_watch_points);
 
 		if (log_max_watch_addr) {
 			dev->node_props.capability |=
@@ -705,30 +506,47 @@ static ssize_t node_show(struct kobject *kobj, struct attribute *attr,
 				HSA_CAP_WATCH_POINTS_TOTALBITS_MASK);
 		}
 
-		sysfs_show_32bit_prop(buffer, "max_engine_clk_fcompute",
-			dev->gpu->kfd2kgd->get_max_engine_clock_in_mhz(
-					dev->gpu->kgd));
-
-		sysfs_show_64bit_prop(buffer, "local_mem_size",
-				(unsigned long long int) 0);
-
-		sysfs_show_32bit_prop(buffer, "fw_version",
-			dev->gpu->kfd2kgd->get_fw_version(
-						dev->gpu->kgd,
-						KGD_ENGINE_MEC1));
-		sysfs_show_32bit_prop(buffer, "capability",
-				dev->node_props.capability);
+		if (dev->gpu->adev->asic_type == CHIP_TONGA)
+			dev->node_props.capability |=
+					HSA_CAP_AQL_QUEUE_DOUBLE_MAP;
+
+		if (KFD_GC_VERSION(dev->gpu) < IP_VERSION(10, 0, 0) &&
+			(dev->gpu->adev->sdma.supported_reset & AMDGPU_RESET_TYPE_PER_QUEUE))
+				dev->node_props.capability2 |= HSA_CAP2_PER_SDMA_QUEUE_RESET_SUPPORTED;
+
+		sysfs_show_32bit_prop(buffer, offs, "max_engine_clk_fcompute",
+			dev->node_props.max_engine_clk_fcompute);
+
+		sysfs_show_64bit_prop(buffer, offs, "local_mem_size", 0ULL);
+
+		sysfs_show_32bit_prop(buffer, offs, "fw_version",
+				      dev->gpu->kfd->mec_fw_version);
+		sysfs_show_32bit_prop(buffer, offs, "capability",
+				      dev->node_props.capability);
+		sysfs_show_32bit_prop(buffer, offs, "capability2",
+				      dev->node_props.capability2);
+		sysfs_show_64bit_prop(buffer, offs, "debug_prop",
+				      dev->node_props.debug_prop);
+		sysfs_show_32bit_prop(buffer, offs, "sdma_fw_version",
+				      dev->gpu->kfd->sdma_fw_version);
+		sysfs_show_64bit_prop(buffer, offs, "unique_id",
+				      dev->gpu->xcp ?
+				      dev->gpu->xcp->unique_id :
+				      dev->gpu->adev->unique_id);
+		sysfs_show_32bit_prop(buffer, offs, "num_xcc",
+				      NUM_XCC(dev->gpu->xcc_mask));
 	}
 
-	return sysfs_show_32bit_prop(buffer, "max_engine_clk_ccompute",
-					cpufreq_quick_get_max(0)/1000);
+	return sysfs_show_32bit_prop(buffer, offs, "max_engine_clk_ccompute",
+				     cpufreq_quick_get_max(0)/1000);
 }
 
 static const struct sysfs_ops node_ops = {
 	.show = node_show,
 };
 
-static struct kobj_type node_type = {
+static const struct kobj_type node_type = {
+	.release = kfd_topology_kobj_release,
 	.sysfs_ops = &node_ops,
 };
 
@@ -741,9 +559,11 @@ static void kfd_remove_sysfs_file(struct kobject *kobj, struct attribute *attr)
 
 static void kfd_remove_sysfs_node_entry(struct kfd_topology_device *dev)
 {
+	struct kfd_iolink_properties *p2plink;
 	struct kfd_iolink_properties *iolink;
 	struct kfd_cache_properties *cache;
 	struct kfd_mem_properties *mem;
+	struct kfd_perf_properties *perf;
 
 	if (dev->kobj_iolink) {
 		list_for_each_entry(iolink, &dev->io_link_props, list)
@@ -757,6 +577,18 @@ static void kfd_remove_sysfs_node_entry(struct kfd_topology_device *dev)
 		dev->kobj_iolink = NULL;
 	}
 
+	if (dev->kobj_p2plink) {
+		list_for_each_entry(p2plink, &dev->p2p_link_props, list)
+			if (p2plink->kobj) {
+				kfd_remove_sysfs_file(p2plink->kobj,
+							&p2plink->attr);
+				p2plink->kobj = NULL;
+			}
+		kobject_del(dev->kobj_p2plink);
+		kobject_put(dev->kobj_p2plink);
+		dev->kobj_p2plink = NULL;
+	}
+
 	if (dev->kobj_cache) {
 		list_for_each_entry(cache, &dev->cache_props, list)
 			if (cache->kobj) {
@@ -780,6 +612,16 @@ static void kfd_remove_sysfs_node_entry(struct kfd_topology_device *dev)
 		dev->kobj_mem = NULL;
 	}
 
+	if (dev->kobj_perf) {
+		list_for_each_entry(perf, &dev->perf_props, list) {
+			kfree(perf->attr_group);
+			perf->attr_group = NULL;
+		}
+		kobject_del(dev->kobj_perf);
+		kobject_put(dev->kobj_perf);
+		dev->kobj_perf = NULL;
+	}
+
 	if (dev->kobj_node) {
 		sysfs_remove_file(dev->kobj_node, &dev->attr_gpuid);
 		sysfs_remove_file(dev->kobj_node, &dev->attr_name);
@@ -793,11 +635,14 @@ static void kfd_remove_sysfs_node_entry(struct kfd_topology_device *dev)
 static int kfd_build_sysfs_node_entry(struct kfd_topology_device *dev,
 		uint32_t id)
 {
+	struct kfd_iolink_properties *p2plink;
 	struct kfd_iolink_properties *iolink;
 	struct kfd_cache_properties *cache;
 	struct kfd_mem_properties *mem;
+	struct kfd_perf_properties *perf;
 	int ret;
-	uint32_t i;
+	uint32_t i, num_attrs;
+	struct attribute **attrs;
 
 	if (WARN_ON(dev->kobj_node))
 		return -EEXIST;
@@ -811,8 +656,10 @@ static int kfd_build_sysfs_node_entry(struct kfd_topology_device *dev,
 
 	ret = kobject_init_and_add(dev->kobj_node, &node_type,
 			sys_props.kobj_nodes, "%d", id);
-	if (ret < 0)
+	if (ret < 0) {
+		kobject_put(dev->kobj_node);
 		return ret;
+	}
 
 	dev->kobj_mem = kobject_create_and_add("mem_banks", dev->kobj_node);
 	if (!dev->kobj_mem)
@@ -826,6 +673,14 @@ static int kfd_build_sysfs_node_entry(struct kfd_topology_device *dev,
 	if (!dev->kobj_iolink)
 		return -ENOMEM;
 
+	dev->kobj_p2plink = kobject_create_and_add("p2p_links", dev->kobj_node);
+	if (!dev->kobj_p2plink)
+		return -ENOMEM;
+
+	dev->kobj_perf = kobject_create_and_add("perf", dev->kobj_node);
+	if (!dev->kobj_perf)
+		return -ENOMEM;
+
 	/*
 	 * Creating sysfs files for node properties
 	 */
@@ -855,8 +710,10 @@ static int kfd_build_sysfs_node_entry(struct kfd_topology_device *dev,
 			return -ENOMEM;
 		ret = kobject_init_and_add(mem->kobj, &mem_type,
 				dev->kobj_mem, "%d", i);
-		if (ret < 0)
+		if (ret < 0) {
+			kobject_put(mem->kobj);
 			return ret;
+		}
 
 		mem->attr.name = "properties";
 		mem->attr.mode = KFD_SYSFS_FILE_MODE;
@@ -874,8 +731,10 @@ static int kfd_build_sysfs_node_entry(struct kfd_topology_device *dev,
 			return -ENOMEM;
 		ret = kobject_init_and_add(cache->kobj, &cache_type,
 				dev->kobj_cache, "%d", i);
-		if (ret < 0)
+		if (ret < 0) {
+			kobject_put(cache->kobj);
 			return ret;
+		}
 
 		cache->attr.name = "properties";
 		cache->attr.mode = KFD_SYSFS_FILE_MODE;
@@ -893,8 +752,10 @@ static int kfd_build_sysfs_node_entry(struct kfd_topology_device *dev,
 			return -ENOMEM;
 		ret = kobject_init_and_add(iolink->kobj, &iolink_type,
 				dev->kobj_iolink, "%d", i);
-		if (ret < 0)
+		if (ret < 0) {
+			kobject_put(iolink->kobj);
 			return ret;
+		}
 
 		iolink->attr.name = "properties";
 		iolink->attr.mode = KFD_SYSFS_FILE_MODE;
@@ -903,11 +764,59 @@ static int kfd_build_sysfs_node_entry(struct kfd_topology_device *dev,
 		if (ret < 0)
 			return ret;
 		i++;
-}
+	}
+
+	i = 0;
+	list_for_each_entry(p2plink, &dev->p2p_link_props, list) {
+		p2plink->kobj = kzalloc(sizeof(struct kobject), GFP_KERNEL);
+		if (!p2plink->kobj)
+			return -ENOMEM;
+		ret = kobject_init_and_add(p2plink->kobj, &iolink_type,
+				dev->kobj_p2plink, "%d", i);
+		if (ret < 0) {
+			kobject_put(p2plink->kobj);
+			return ret;
+		}
+
+		p2plink->attr.name = "properties";
+		p2plink->attr.mode = KFD_SYSFS_FILE_MODE;
+		sysfs_attr_init(&p2plink->attr);
+		ret = sysfs_create_file(p2plink->kobj, &p2plink->attr);
+		if (ret < 0)
+			return ret;
+		i++;
+	}
+
+	/* All hardware blocks have the same number of attributes. */
+	num_attrs = ARRAY_SIZE(perf_attr_iommu);
+	list_for_each_entry(perf, &dev->perf_props, list) {
+		perf->attr_group = kzalloc(sizeof(struct kfd_perf_attr)
+			* num_attrs + sizeof(struct attribute_group),
+			GFP_KERNEL);
+		if (!perf->attr_group)
+			return -ENOMEM;
+
+		attrs = (struct attribute **)(perf->attr_group + 1);
+		if (!strcmp(perf->block_name, "iommu")) {
+		/* Information of IOMMU's num_counters and counter_ids is shown
+		 * under /sys/bus/event_source/devices/amd_iommu. We don't
+		 * duplicate here.
+		 */
+			perf_attr_iommu[0].data = perf->max_concurrent;
+			for (i = 0; i < num_attrs; i++)
+				attrs[i] = &perf_attr_iommu[i].attr.attr;
+		}
+		perf->attr_group->name = perf->block_name;
+		perf->attr_group->attrs = attrs;
+		ret = sysfs_create_group(dev->kobj_perf, perf->attr_group);
+		if (ret < 0)
+			return ret;
+	}
 
 	return 0;
 }
 
+/* Called with write topology lock acquired */
 static int kfd_build_sysfs_node_tree(void)
 {
 	struct kfd_topology_device *dev;
@@ -924,6 +833,7 @@ static int kfd_build_sysfs_node_tree(void)
 	return 0;
 }
 
+/* Called with write topology lock acquired */
 static void kfd_remove_sysfs_node_tree(void)
 {
 	struct kfd_topology_device *dev;
@@ -936,7 +846,6 @@ static int kfd_topology_update_sysfs(void)
 {
 	int ret;
 
-	pr_info("Creating topology SYSFS entries\n");
 	if (!sys_props.kobj_topology) {
 		sys_props.kobj_topology =
 				kfd_alloc_struct(sys_props.kobj_topology);
@@ -946,8 +855,10 @@ static int kfd_topology_update_sysfs(void)
 		ret = kobject_init_and_add(sys_props.kobj_topology,
 				&sysprops_type,  &kfd_device->kobj,
 				"topology");
-		if (ret < 0)
+		if (ret < 0) {
+			kobject_put(sys_props.kobj_topology);
 			return ret;
+		}
 
 		sys_props.kobj_nodes = kobject_create_and_add("nodes",
 				sys_props.kobj_topology);
@@ -995,115 +906,271 @@ static void kfd_topology_release_sysfs(void)
 	}
 }
 
+/* Called with write topology_lock acquired */
+static void kfd_topology_update_device_list(struct list_head *temp_list,
+					struct list_head *master_list)
+{
+	while (!list_empty(temp_list)) {
+		list_move_tail(temp_list->next, master_list);
+		sys_props.num_devices++;
+	}
+}
+
+static void kfd_debug_print_topology(void)
+{
+	struct kfd_topology_device *dev;
+
+	down_read(&topology_lock);
+
+	dev = list_last_entry(&topology_device_list,
+			struct kfd_topology_device, list);
+	if (dev) {
+		if (dev->node_props.cpu_cores_count &&
+				dev->node_props.simd_count) {
+			pr_info("Topology: Add APU node [0x%0x:0x%0x]\n",
+				dev->node_props.device_id,
+				dev->node_props.vendor_id);
+		} else if (dev->node_props.cpu_cores_count)
+			pr_info("Topology: Add CPU node\n");
+		else if (dev->node_props.simd_count)
+			pr_info("Topology: Add dGPU node [0x%0x:0x%0x]\n",
+				dev->node_props.device_id,
+				dev->node_props.vendor_id);
+	}
+	up_read(&topology_lock);
+}
+
+/* Helper function for intializing platform_xx members of
+ * kfd_system_properties. Uses OEM info from the last CPU/APU node.
+ */
+static void kfd_update_system_properties(void)
+{
+	struct kfd_topology_device *dev;
+
+	down_read(&topology_lock);
+	dev = list_last_entry(&topology_device_list,
+			struct kfd_topology_device, list);
+	if (dev) {
+		sys_props.platform_id = dev->oem_id64;
+		sys_props.platform_oem = *((uint64_t *)dev->oem_table_id);
+		sys_props.platform_rev = dev->oem_revision;
+	}
+	up_read(&topology_lock);
+}
+
+static void find_system_memory(const struct dmi_header *dm, void *private)
+{
+	struct dmi_mem_device *memdev = container_of(dm, struct dmi_mem_device, header);
+	struct kfd_mem_properties *mem;
+	struct kfd_topology_device *kdev =
+		(struct kfd_topology_device *)private;
+
+	if (memdev->header.type != DMI_ENTRY_MEM_DEVICE)
+		return;
+	if (memdev->header.length < sizeof(struct dmi_mem_device))
+		return;
+
+	list_for_each_entry(mem, &kdev->mem_props, list) {
+		if (memdev->total_width != 0xFFFF && memdev->total_width != 0)
+			mem->width = memdev->total_width;
+		if (memdev->speed != 0)
+			mem->mem_clk_max = memdev->speed;
+	}
+}
+
+/* kfd_add_non_crat_information - Add information that is not currently
+ *	defined in CRAT but is necessary for KFD topology
+ * @dev - topology device to which addition info is added
+ */
+static void kfd_add_non_crat_information(struct kfd_topology_device *kdev)
+{
+	/* Check if CPU only node. */
+	if (!kdev->gpu) {
+		/* Add system memory information */
+		dmi_walk(find_system_memory, kdev);
+	}
+	/* TODO: For GPU node, rearrange code from kfd_topology_add_device */
+}
+
 int kfd_topology_init(void)
 {
 	void *crat_image = NULL;
 	size_t image_size = 0;
 	int ret;
-
-	/*
-	 * Initialize the head for the topology device list
+	struct list_head temp_topology_device_list;
+	int cpu_only_node = 0;
+	struct kfd_topology_device *kdev;
+	int proximity_domain;
+
+	/* topology_device_list - Master list of all topology devices
+	 * temp_topology_device_list - temporary list created while parsing CRAT
+	 * or VCRAT. Once parsing is complete the contents of list is moved to
+	 * topology_device_list
 	 */
+
+	/* Initialize the head for the both the lists */
 	INIT_LIST_HEAD(&topology_device_list);
+	INIT_LIST_HEAD(&temp_topology_device_list);
 	init_rwsem(&topology_lock);
-	topology_crat_parsed = 0;
 
 	memset(&sys_props, 0, sizeof(sys_props));
 
-	/*
-	 * Get the CRAT image from the ACPI
+	/* Proximity domains in ACPI CRAT tables start counting at
+	 * 0. The same should be true for virtual CRAT tables created
+	 * at this stage. GPUs added later in kfd_topology_add_device
+	 * use a counter.
 	 */
-	ret = kfd_topology_get_crat_acpi(crat_image, &image_size);
-	if (ret == 0 && image_size > 0) {
-		pr_info("Found CRAT image with size=%zd\n", image_size);
-		crat_image = kmalloc(image_size, GFP_KERNEL);
-		if (!crat_image) {
-			ret = -ENOMEM;
-			pr_err("No memory for allocating CRAT image\n");
-			goto err;
-		}
-		ret = kfd_topology_get_crat_acpi(crat_image, &image_size);
-
-		if (ret == 0) {
-			down_write(&topology_lock);
-			ret = kfd_parse_crat_table(crat_image);
-			if (ret == 0)
-				ret = kfd_topology_update_sysfs();
-			up_write(&topology_lock);
-		} else {
-			pr_err("Couldn't get CRAT table size from ACPI\n");
-		}
-		kfree(crat_image);
-	} else if (ret == -ENODATA) {
-		ret = 0;
-	} else {
-		pr_err("Couldn't get CRAT table size from ACPI\n");
+	proximity_domain = 0;
+
+	ret = kfd_create_crat_image_virtual(&crat_image, &image_size,
+					    COMPUTE_UNIT_CPU, NULL,
+					    proximity_domain);
+	cpu_only_node = 1;
+	if (ret) {
+		pr_err("Error creating VCRAT table for CPU\n");
+		return ret;
+	}
+
+	ret = kfd_parse_crat_table(crat_image,
+				   &temp_topology_device_list,
+				   proximity_domain);
+	if (ret) {
+		pr_err("Error parsing VCRAT table for CPU\n");
+		goto err;
+	}
+
+	kdev = list_first_entry(&temp_topology_device_list,
+				struct kfd_topology_device, list);
+
+	down_write(&topology_lock);
+	kfd_topology_update_device_list(&temp_topology_device_list,
+					&topology_device_list);
+	topology_crat_proximity_domain = sys_props.num_devices-1;
+	ret = kfd_topology_update_sysfs();
+	up_write(&topology_lock);
+
+	if (!ret) {
+		sys_props.generation_count++;
+		kfd_update_system_properties();
+		kfd_debug_print_topology();
+	} else
+		pr_err("Failed to update topology in sysfs ret=%d\n", ret);
+
+	/* For nodes with GPU, this information gets added
+	 * when GPU is detected (kfd_topology_add_device).
+	 */
+	if (cpu_only_node) {
+		/* Add additional information to CPU only node created above */
+		down_write(&topology_lock);
+		kdev = list_first_entry(&topology_device_list,
+				struct kfd_topology_device, list);
+		up_write(&topology_lock);
+		kfd_add_non_crat_information(kdev);
 	}
 
 err:
-	pr_info("Finished initializing topology ret=%d\n", ret);
+	kfd_destroy_crat_image(crat_image);
 	return ret;
 }
 
 void kfd_topology_shutdown(void)
 {
+	down_write(&topology_lock);
 	kfd_topology_release_sysfs();
 	kfd_release_live_view();
+	up_write(&topology_lock);
 }
 
-static void kfd_debug_print_topology(void)
-{
-	struct kfd_topology_device *dev;
-	uint32_t i = 0;
-
-	pr_info("DEBUG PRINT OF TOPOLOGY:");
-	list_for_each_entry(dev, &topology_device_list, list) {
-		pr_info("Node: %d\n", i);
-		pr_info("\tGPU assigned: %s\n", (dev->gpu ? "yes" : "no"));
-		pr_info("\tCPU count: %d\n", dev->node_props.cpu_cores_count);
-		pr_info("\tSIMD count: %d", dev->node_props.simd_count);
-		i++;
-	}
-}
-
-static uint32_t kfd_generate_gpu_id(struct kfd_dev *gpu)
+static uint32_t kfd_generate_gpu_id(struct kfd_node *gpu)
 {
-	uint32_t hashout;
-	uint32_t buf[7];
+	uint32_t gpu_id;
+	uint32_t buf[8];
 	uint64_t local_mem_size;
-	int i;
+	struct kfd_topology_device *dev;
+	bool is_unique;
+	uint8_t *crc_buf;
 
 	if (!gpu)
 		return 0;
 
-	local_mem_size = gpu->kfd2kgd->get_vmem_size(gpu->kgd);
-
-	buf[0] = gpu->pdev->devfn;
-	buf[1] = gpu->pdev->subsystem_vendor;
-	buf[2] = gpu->pdev->subsystem_device;
-	buf[3] = gpu->pdev->device;
-	buf[4] = gpu->pdev->bus->number;
+	crc_buf = (uint8_t *)&buf;
+	local_mem_size = gpu->local_mem_info.local_mem_size_private +
+			gpu->local_mem_info.local_mem_size_public;
+	buf[0] = gpu->adev->pdev->devfn;
+	buf[1] = gpu->adev->pdev->subsystem_vendor |
+		(gpu->adev->pdev->subsystem_device << 16);
+	buf[2] = pci_domain_nr(gpu->adev->pdev->bus);
+	buf[3] = gpu->adev->pdev->device;
+	buf[4] = gpu->adev->pdev->bus->number;
 	buf[5] = lower_32_bits(local_mem_size);
 	buf[6] = upper_32_bits(local_mem_size);
+	buf[7] = (ffs(gpu->xcc_mask) - 1) | (NUM_XCC(gpu->xcc_mask) << 16);
 
-	for (i = 0, hashout = 0; i < 7; i++)
-		hashout ^= hash_32(buf[i], KFD_GPU_ID_HASH_WIDTH);
+	gpu_id = crc16(0, crc_buf, sizeof(buf)) &
+		 ((1 << KFD_GPU_ID_HASH_WIDTH) - 1);
 
-	return hashout;
-}
+	/* There is a very small possibility when generating a
+	 * 16 (KFD_GPU_ID_HASH_WIDTH) bit value from 8 word buffer
+	 * that the value could be 0 or non-unique. So, check if
+	 * it is unique and non-zero. If not unique increment till
+	 * unique one is found. In case of overflow, restart from 1
+	 */
+
+	down_read(&topology_lock);
+	do {
+		is_unique = true;
+		if (!gpu_id)
+			gpu_id = 1;
+		list_for_each_entry(dev, &topology_device_list, list) {
+			if (dev->gpu && dev->gpu_id == gpu_id) {
+				is_unique = false;
+				break;
+			}
+		}
+		if (unlikely(!is_unique))
+			gpu_id = (gpu_id + 1) &
+				  ((1 << KFD_GPU_ID_HASH_WIDTH) - 1);
+	} while (!is_unique);
+	up_read(&topology_lock);
 
-static struct kfd_topology_device *kfd_assign_gpu(struct kfd_dev *gpu)
+	return gpu_id;
+}
+/* kfd_assign_gpu - Attach @gpu to the correct kfd topology device. If
+ *		the GPU device is not already present in the topology device
+ *		list then return NULL. This means a new topology device has to
+ *		be created for this GPU.
+ */
+static struct kfd_topology_device *kfd_assign_gpu(struct kfd_node *gpu)
 {
 	struct kfd_topology_device *dev;
 	struct kfd_topology_device *out_dev = NULL;
+	struct kfd_mem_properties *mem;
+	struct kfd_cache_properties *cache;
+	struct kfd_iolink_properties *iolink;
+	struct kfd_iolink_properties *p2plink;
+
+	list_for_each_entry(dev, &topology_device_list, list) {
+		/* Discrete GPUs need their own topology device list
+		 * entries. Don't assign them to CPU/APU nodes.
+		 */
+		if (dev->node_props.cpu_cores_count)
+			continue;
 
-	list_for_each_entry(dev, &topology_device_list, list)
 		if (!dev->gpu && (dev->node_props.simd_count > 0)) {
 			dev->gpu = gpu;
 			out_dev = dev;
+
+			list_for_each_entry(mem, &dev->mem_props, list)
+				mem->gpu = dev->gpu;
+			list_for_each_entry(cache, &dev->cache_props, list)
+				cache->gpu = dev->gpu;
+			list_for_each_entry(iolink, &dev->io_link_props, list)
+				iolink->gpu = dev->gpu;
+			list_for_each_entry(p2plink, &dev->p2p_link_props, list)
+				p2plink->gpu = dev->gpu;
 			break;
 		}
-
+	}
 	return out_dev;
 }
 
@@ -1115,114 +1182,1133 @@ static void kfd_notify_gpu_change(uint32_t gpu_id, int arrival)
 	 */
 }
 
-int kfd_topology_add_device(struct kfd_dev *gpu)
+/* kfd_fill_mem_clk_max_info - Since CRAT doesn't have memory clock info,
+ *		patch this after CRAT parsing.
+ */
+static void kfd_fill_mem_clk_max_info(struct kfd_topology_device *dev)
+{
+	struct kfd_mem_properties *mem;
+	struct kfd_local_mem_info local_mem_info;
+
+	if (!dev)
+		return;
+
+	/* Currently, amdgpu driver (amdgpu_mc) deals only with GPUs with
+	 * single bank of VRAM local memory.
+	 * for dGPUs - VCRAT reports only one bank of Local Memory
+	 * for APUs - If CRAT from ACPI reports more than one bank, then
+	 *	all the banks will report the same mem_clk_max information
+	 */
+	amdgpu_amdkfd_get_local_mem_info(dev->gpu->adev, &local_mem_info,
+					 dev->gpu->xcp);
+
+	list_for_each_entry(mem, &dev->mem_props, list)
+		mem->mem_clk_max = local_mem_info.mem_clk_max;
+}
+
+static void kfd_set_iolink_no_atomics(struct kfd_topology_device *dev,
+					struct kfd_topology_device *target_gpu_dev,
+					struct kfd_iolink_properties *link)
+{
+	/* xgmi always supports atomics between links. */
+	if (link->iolink_type == CRAT_IOLINK_TYPE_XGMI)
+		return;
+
+	/* check pcie support to set cpu(dev) flags for target_gpu_dev link. */
+	if (target_gpu_dev) {
+		uint32_t cap;
+
+		pcie_capability_read_dword(target_gpu_dev->gpu->adev->pdev,
+				PCI_EXP_DEVCAP2, &cap);
+
+		if (!(cap & (PCI_EXP_DEVCAP2_ATOMIC_COMP32 |
+			     PCI_EXP_DEVCAP2_ATOMIC_COMP64)))
+			link->flags |= CRAT_IOLINK_FLAGS_NO_ATOMICS_32_BIT |
+				CRAT_IOLINK_FLAGS_NO_ATOMICS_64_BIT;
+	/* set gpu (dev) flags. */
+	} else {
+		if (!dev->gpu->kfd->pci_atomic_requested ||
+				dev->gpu->adev->asic_type == CHIP_HAWAII)
+			link->flags |= CRAT_IOLINK_FLAGS_NO_ATOMICS_32_BIT |
+				CRAT_IOLINK_FLAGS_NO_ATOMICS_64_BIT;
+	}
+}
+
+static void kfd_set_iolink_non_coherent(struct kfd_topology_device *to_dev,
+		struct kfd_iolink_properties *outbound_link,
+		struct kfd_iolink_properties *inbound_link)
+{
+	/* CPU -> GPU with PCIe */
+	if (!to_dev->gpu &&
+	    inbound_link->iolink_type == CRAT_IOLINK_TYPE_PCIEXPRESS)
+		inbound_link->flags |= CRAT_IOLINK_FLAGS_NON_COHERENT;
+
+	if (to_dev->gpu) {
+		/* GPU <-> GPU with PCIe and
+		 * Vega20 with XGMI
+		 */
+		if (inbound_link->iolink_type == CRAT_IOLINK_TYPE_PCIEXPRESS ||
+		    (inbound_link->iolink_type == CRAT_IOLINK_TYPE_XGMI &&
+		    KFD_GC_VERSION(to_dev->gpu) == IP_VERSION(9, 4, 0))) {
+			outbound_link->flags |= CRAT_IOLINK_FLAGS_NON_COHERENT;
+			inbound_link->flags |= CRAT_IOLINK_FLAGS_NON_COHERENT;
+		}
+	}
+}
+
+#define REC_SDMA_NUM_GPU	8
+static const int rec_sdma_eng_map[REC_SDMA_NUM_GPU][REC_SDMA_NUM_GPU] = {
+							{ -1, 14, 12, 2, 4, 8, 10, 6 },
+							{ 14, -1, 2, 10, 8, 4, 6, 12 },
+							{ 10, 2, -1, 12, 14, 6, 4, 8 },
+							{ 2, 12, 10, -1, 6, 14, 8, 4 },
+							{ 4, 8, 14, 6, -1, 10, 12, 2 },
+							{ 8, 4, 6, 14, 12, -1, 2, 10 },
+							{ 10, 6, 4, 8, 12, 2, -1, 14 },
+							{ 6, 12, 8, 4, 2, 10, 14, -1 }};
+
+static void kfd_set_recommended_sdma_engines(struct kfd_topology_device *to_dev,
+					     struct kfd_iolink_properties *outbound_link,
+					     struct kfd_iolink_properties *inbound_link)
+{
+	struct kfd_node *gpu = outbound_link->gpu;
+	struct amdgpu_device *adev = gpu->adev;
+	unsigned int num_xgmi_nodes = adev->gmc.xgmi.num_physical_nodes;
+	unsigned int num_xgmi_sdma_engines = kfd_get_num_xgmi_sdma_engines(gpu);
+	unsigned int num_sdma_engines = kfd_get_num_sdma_engines(gpu);
+	uint32_t sdma_eng_id_mask = (1 << num_sdma_engines) - 1;
+	uint32_t xgmi_sdma_eng_id_mask =
+			((1 << num_xgmi_sdma_engines) - 1) << num_sdma_engines;
+
+	bool support_rec_eng = !amdgpu_sriov_vf(adev) && to_dev->gpu &&
+		adev->aid_mask && num_xgmi_nodes && gpu->kfd->num_nodes == 1 &&
+		num_xgmi_sdma_engines >= 6 && (!(adev->flags & AMD_IS_APU) &&
+		num_xgmi_nodes == 8);
+
+	if (support_rec_eng) {
+		int src_socket_id = adev->gmc.xgmi.physical_node_id;
+		int dst_socket_id = to_dev->gpu->adev->gmc.xgmi.physical_node_id;
+		unsigned int reshift = num_xgmi_sdma_engines == 6 ? 1 : 0;
+
+		outbound_link->rec_sdma_eng_id_mask =
+			1 << (rec_sdma_eng_map[src_socket_id][dst_socket_id] >> reshift);
+		inbound_link->rec_sdma_eng_id_mask =
+			1 << (rec_sdma_eng_map[dst_socket_id][src_socket_id] >> reshift);
+
+		/* If recommended engine is out of range, need to reset the mask */
+		if (outbound_link->rec_sdma_eng_id_mask & sdma_eng_id_mask)
+			outbound_link->rec_sdma_eng_id_mask = xgmi_sdma_eng_id_mask;
+		if (inbound_link->rec_sdma_eng_id_mask & sdma_eng_id_mask)
+			inbound_link->rec_sdma_eng_id_mask = xgmi_sdma_eng_id_mask;
+
+	} else {
+		uint32_t engine_mask = (outbound_link->iolink_type == CRAT_IOLINK_TYPE_XGMI &&
+				num_xgmi_sdma_engines && to_dev->gpu) ? xgmi_sdma_eng_id_mask :
+				sdma_eng_id_mask;
+
+		outbound_link->rec_sdma_eng_id_mask = engine_mask;
+		inbound_link->rec_sdma_eng_id_mask = engine_mask;
+	}
+}
+
+static void kfd_fill_iolink_non_crat_info(struct kfd_topology_device *dev)
+{
+	struct kfd_iolink_properties *link, *inbound_link;
+	struct kfd_topology_device *peer_dev;
+
+	if (!dev || !dev->gpu)
+		return;
+
+	/* GPU only creates direct links so apply flags setting to all */
+	list_for_each_entry(link, &dev->io_link_props, list) {
+		link->flags = CRAT_IOLINK_FLAGS_ENABLED;
+		kfd_set_iolink_no_atomics(dev, NULL, link);
+		peer_dev = kfd_topology_device_by_proximity_domain(
+				link->node_to);
+
+		if (!peer_dev)
+			continue;
+
+		/* Include the CPU peer in GPU hive if connected over xGMI. */
+		if (!peer_dev->gpu &&
+		    link->iolink_type == CRAT_IOLINK_TYPE_XGMI) {
+			/*
+			 * If the GPU is not part of a GPU hive, use its pci
+			 * device location as the hive ID to bind with the CPU.
+			 */
+			if (!dev->node_props.hive_id)
+				dev->node_props.hive_id = pci_dev_id(dev->gpu->adev->pdev);
+			peer_dev->node_props.hive_id = dev->node_props.hive_id;
+		}
+
+		list_for_each_entry(inbound_link, &peer_dev->io_link_props,
+									list) {
+			if (inbound_link->node_to != link->node_from)
+				continue;
+
+			inbound_link->flags = CRAT_IOLINK_FLAGS_ENABLED;
+			kfd_set_iolink_no_atomics(peer_dev, dev, inbound_link);
+			kfd_set_iolink_non_coherent(peer_dev, link, inbound_link);
+			kfd_set_recommended_sdma_engines(peer_dev, link, inbound_link);
+		}
+	}
+
+	/* Create indirect links so apply flags setting to all */
+	list_for_each_entry(link, &dev->p2p_link_props, list) {
+		link->flags = CRAT_IOLINK_FLAGS_ENABLED;
+		kfd_set_iolink_no_atomics(dev, NULL, link);
+		peer_dev = kfd_topology_device_by_proximity_domain(
+				link->node_to);
+
+		if (!peer_dev)
+			continue;
+
+		list_for_each_entry(inbound_link, &peer_dev->p2p_link_props,
+									list) {
+			if (inbound_link->node_to != link->node_from)
+				continue;
+
+			inbound_link->flags = CRAT_IOLINK_FLAGS_ENABLED;
+			kfd_set_iolink_no_atomics(peer_dev, dev, inbound_link);
+			kfd_set_iolink_non_coherent(peer_dev, link, inbound_link);
+		}
+	}
+}
+
+static int kfd_build_p2p_node_entry(struct kfd_topology_device *dev,
+				struct kfd_iolink_properties *p2plink)
+{
+	int ret;
+
+	p2plink->kobj = kzalloc(sizeof(struct kobject), GFP_KERNEL);
+	if (!p2plink->kobj)
+		return -ENOMEM;
+
+	ret = kobject_init_and_add(p2plink->kobj, &iolink_type,
+			dev->kobj_p2plink, "%d", dev->node_props.p2p_links_count - 1);
+	if (ret < 0) {
+		kobject_put(p2plink->kobj);
+		return ret;
+	}
+
+	p2plink->attr.name = "properties";
+	p2plink->attr.mode = KFD_SYSFS_FILE_MODE;
+	sysfs_attr_init(&p2plink->attr);
+	ret = sysfs_create_file(p2plink->kobj, &p2plink->attr);
+	if (ret < 0)
+		return ret;
+
+	return 0;
+}
+
+static int kfd_create_indirect_link_prop(struct kfd_topology_device *kdev, int gpu_node)
+{
+	struct kfd_iolink_properties *gpu_link, *tmp_link, *cpu_link;
+	struct kfd_iolink_properties *props = NULL, *props2 = NULL;
+	struct kfd_topology_device *cpu_dev;
+	int ret = 0;
+	int i, num_cpu;
+
+	num_cpu = 0;
+	list_for_each_entry(cpu_dev, &topology_device_list, list) {
+		if (cpu_dev->gpu)
+			break;
+		num_cpu++;
+	}
+
+	if (list_empty(&kdev->io_link_props))
+		return -ENODATA;
+
+	gpu_link = list_first_entry(&kdev->io_link_props,
+				    struct kfd_iolink_properties, list);
+
+	for (i = 0; i < num_cpu; i++) {
+		/* CPU <--> GPU */
+		if (gpu_link->node_to == i)
+			continue;
+
+		/* find CPU <-->  CPU links */
+		cpu_link = NULL;
+		cpu_dev = kfd_topology_device_by_proximity_domain(i);
+		if (cpu_dev) {
+			list_for_each_entry(tmp_link,
+					&cpu_dev->io_link_props, list) {
+				if (tmp_link->node_to == gpu_link->node_to) {
+					cpu_link = tmp_link;
+					break;
+				}
+			}
+		}
+
+		if (!cpu_link)
+			return -ENOMEM;
+
+		/* CPU <--> CPU <--> GPU, GPU node*/
+		props = kfd_alloc_struct(props);
+		if (!props)
+			return -ENOMEM;
+
+		memcpy(props, gpu_link, sizeof(struct kfd_iolink_properties));
+		props->weight = gpu_link->weight + cpu_link->weight;
+		props->min_latency = gpu_link->min_latency + cpu_link->min_latency;
+		props->max_latency = gpu_link->max_latency + cpu_link->max_latency;
+		props->min_bandwidth = min(gpu_link->min_bandwidth, cpu_link->min_bandwidth);
+		props->max_bandwidth = min(gpu_link->max_bandwidth, cpu_link->max_bandwidth);
+
+		props->node_from = gpu_node;
+		props->node_to = i;
+		kdev->node_props.p2p_links_count++;
+		list_add_tail(&props->list, &kdev->p2p_link_props);
+		ret = kfd_build_p2p_node_entry(kdev, props);
+		if (ret < 0)
+			return ret;
+
+		/* for small Bar, no CPU --> GPU in-direct links */
+		if (kfd_dev_is_large_bar(kdev->gpu)) {
+			/* CPU <--> CPU <--> GPU, CPU node*/
+			props2 = kfd_alloc_struct(props2);
+			if (!props2)
+				return -ENOMEM;
+
+			memcpy(props2, props, sizeof(struct kfd_iolink_properties));
+			props2->node_from = i;
+			props2->node_to = gpu_node;
+			props2->kobj = NULL;
+			cpu_dev->node_props.p2p_links_count++;
+			list_add_tail(&props2->list, &cpu_dev->p2p_link_props);
+			ret = kfd_build_p2p_node_entry(cpu_dev, props2);
+			if (ret < 0)
+				return ret;
+		}
+	}
+	return ret;
+}
+
+#if defined(CONFIG_HSA_AMD_P2P)
+static int kfd_add_peer_prop(struct kfd_topology_device *kdev,
+		struct kfd_topology_device *peer, int from, int to)
+{
+	struct kfd_iolink_properties *props = NULL;
+	struct kfd_iolink_properties *iolink1, *iolink2, *iolink3;
+	struct kfd_topology_device *cpu_dev;
+	int ret = 0;
+
+	if (!amdgpu_device_is_peer_accessible(
+				kdev->gpu->adev,
+				peer->gpu->adev))
+		return ret;
+
+	if (list_empty(&kdev->io_link_props))
+		return -ENODATA;
+
+	iolink1 = list_first_entry(&kdev->io_link_props,
+				   struct kfd_iolink_properties, list);
+
+	if (list_empty(&peer->io_link_props))
+		return -ENODATA;
+
+	iolink2 = list_first_entry(&peer->io_link_props,
+				   struct kfd_iolink_properties, list);
+
+	props = kfd_alloc_struct(props);
+	if (!props)
+		return -ENOMEM;
+
+	memcpy(props, iolink1, sizeof(struct kfd_iolink_properties));
+
+	props->weight = iolink1->weight + iolink2->weight;
+	props->min_latency = iolink1->min_latency + iolink2->min_latency;
+	props->max_latency = iolink1->max_latency + iolink2->max_latency;
+	props->min_bandwidth = min(iolink1->min_bandwidth, iolink2->min_bandwidth);
+	props->max_bandwidth = min(iolink2->max_bandwidth, iolink2->max_bandwidth);
+
+	if (iolink1->node_to != iolink2->node_to) {
+		/* CPU->CPU  link*/
+		cpu_dev = kfd_topology_device_by_proximity_domain(iolink1->node_to);
+		if (cpu_dev) {
+			list_for_each_entry(iolink3, &cpu_dev->io_link_props, list) {
+				if (iolink3->node_to != iolink2->node_to)
+					continue;
+
+				props->weight += iolink3->weight;
+				props->min_latency += iolink3->min_latency;
+				props->max_latency += iolink3->max_latency;
+				props->min_bandwidth = min(props->min_bandwidth,
+							   iolink3->min_bandwidth);
+				props->max_bandwidth = min(props->max_bandwidth,
+							   iolink3->max_bandwidth);
+				break;
+			}
+		} else {
+			WARN(1, "CPU node not found");
+		}
+	}
+
+	props->node_from = from;
+	props->node_to = to;
+	peer->node_props.p2p_links_count++;
+	list_add_tail(&props->list, &peer->p2p_link_props);
+	ret = kfd_build_p2p_node_entry(peer, props);
+
+	return ret;
+}
+#endif
+
+static int kfd_dev_create_p2p_links(void)
 {
-	uint32_t gpu_id;
 	struct kfd_topology_device *dev;
-	int res;
+	struct kfd_topology_device *new_dev;
+#if defined(CONFIG_HSA_AMD_P2P)
+	uint32_t i;
+#endif
+	uint32_t k;
+	int ret = 0;
 
-	gpu_id = kfd_generate_gpu_id(gpu);
+	k = 0;
+	list_for_each_entry(dev, &topology_device_list, list)
+		k++;
+	if (k < 2)
+		return 0;
 
-	pr_debug("Adding new GPU (ID: 0x%x) to topology\n", gpu_id);
+	new_dev = list_last_entry(&topology_device_list, struct kfd_topology_device, list);
+	if (WARN_ON(!new_dev->gpu))
+		return 0;
 
-	down_write(&topology_lock);
-	/*
-	 * Try to assign the GPU to existing topology device (generated from
-	 * CRAT table
+	k--;
+
+	/* create in-direct links */
+	ret = kfd_create_indirect_link_prop(new_dev, k);
+	if (ret < 0)
+		goto out;
+
+	/* create p2p links */
+#if defined(CONFIG_HSA_AMD_P2P)
+	i = 0;
+	list_for_each_entry(dev, &topology_device_list, list) {
+		if (dev == new_dev)
+			break;
+		if (!dev->gpu || !dev->gpu->adev ||
+		    (dev->gpu->kfd->hive_id &&
+		     dev->gpu->kfd->hive_id == new_dev->gpu->kfd->hive_id &&
+		     amdgpu_xgmi_get_is_sharing_enabled(dev->gpu->adev, new_dev->gpu->adev)))
+			goto next;
+
+		/* check if node(s) is/are peer accessible in one direction or bi-direction */
+		ret = kfd_add_peer_prop(new_dev, dev, i, k);
+		if (ret < 0)
+			goto out;
+
+		ret = kfd_add_peer_prop(dev, new_dev, k, i);
+		if (ret < 0)
+			goto out;
+next:
+		i++;
+	}
+#endif
+
+out:
+	return ret;
+}
+
+/* Helper function. See kfd_fill_gpu_cache_info for parameter description */
+static int fill_in_l1_pcache(struct kfd_cache_properties **props_ext,
+				struct kfd_gpu_cache_info *pcache_info,
+				int cu_bitmask,
+				int cache_type, unsigned int cu_processor_id,
+				int cu_block)
+{
+	unsigned int cu_sibling_map_mask;
+	int first_active_cu;
+	struct kfd_cache_properties *pcache = NULL;
+
+	cu_sibling_map_mask = cu_bitmask;
+	cu_sibling_map_mask >>= cu_block;
+	cu_sibling_map_mask &= ((1 << pcache_info[cache_type].num_cu_shared) - 1);
+	first_active_cu = ffs(cu_sibling_map_mask);
+
+	/* CU could be inactive. In case of shared cache find the first active
+	 * CU. and incase of non-shared cache check if the CU is inactive. If
+	 * inactive active skip it
 	 */
-	dev = kfd_assign_gpu(gpu);
-	if (!dev) {
-		pr_info("GPU was not found in the current topology. Extending.\n");
-		kfd_debug_print_topology();
-		dev = kfd_create_topology_device();
-		if (!dev) {
-			res = -ENOMEM;
-			goto err;
-		}
-		dev->gpu = gpu;
+	if (first_active_cu) {
+		pcache = kfd_alloc_struct(pcache);
+		if (!pcache)
+			return -ENOMEM;
 
-		/*
-		 * TODO: Make a call to retrieve topology information from the
-		 * GPU vBIOS
+		memset(pcache, 0, sizeof(struct kfd_cache_properties));
+		pcache->processor_id_low = cu_processor_id + (first_active_cu - 1);
+		pcache->cache_level = pcache_info[cache_type].cache_level;
+		pcache->cache_size = pcache_info[cache_type].cache_size;
+		pcache->cacheline_size = pcache_info[cache_type].cache_line_size;
+
+		if (pcache_info[cache_type].flags & CRAT_CACHE_FLAGS_DATA_CACHE)
+			pcache->cache_type |= HSA_CACHE_TYPE_DATA;
+		if (pcache_info[cache_type].flags & CRAT_CACHE_FLAGS_INST_CACHE)
+			pcache->cache_type |= HSA_CACHE_TYPE_INSTRUCTION;
+		if (pcache_info[cache_type].flags & CRAT_CACHE_FLAGS_CPU_CACHE)
+			pcache->cache_type |= HSA_CACHE_TYPE_CPU;
+		if (pcache_info[cache_type].flags & CRAT_CACHE_FLAGS_SIMD_CACHE)
+			pcache->cache_type |= HSA_CACHE_TYPE_HSACU;
+
+		/* Sibling map is w.r.t processor_id_low, so shift out
+		 * inactive CU
 		 */
+		cu_sibling_map_mask =
+			cu_sibling_map_mask >> (first_active_cu - 1);
+
+		pcache->sibling_map[0] = (uint8_t)(cu_sibling_map_mask & 0xFF);
+		pcache->sibling_map[1] =
+				(uint8_t)((cu_sibling_map_mask >> 8) & 0xFF);
+		pcache->sibling_map[2] =
+				(uint8_t)((cu_sibling_map_mask >> 16) & 0xFF);
+		pcache->sibling_map[3] =
+				(uint8_t)((cu_sibling_map_mask >> 24) & 0xFF);
+
+		pcache->sibling_map_size = 4;
+		*props_ext = pcache;
+
+		return 0;
+	}
+	return 1;
+}
+
+/* Helper function. See kfd_fill_gpu_cache_info for parameter description */
+static int fill_in_l2_l3_pcache(struct kfd_cache_properties **props_ext,
+				struct kfd_gpu_cache_info *pcache_info,
+				struct amdgpu_cu_info *cu_info,
+				struct amdgpu_gfx_config *gfx_info,
+				int cache_type, unsigned int cu_processor_id,
+				struct kfd_node *knode)
+{
+	unsigned int cu_sibling_map_mask = 0;
+	int first_active_cu;
+	int i, j, k, xcc, start, end;
+	int num_xcc = NUM_XCC(knode->xcc_mask);
+	struct kfd_cache_properties *pcache = NULL;
+	enum amdgpu_memory_partition mode;
+	struct amdgpu_device *adev = knode->adev;
+	bool found = false;
+
+	start = ffs(knode->xcc_mask) - 1;
+	end = start + num_xcc;
+
+	/* To find the bitmap in the first active cu in the first
+	 * xcc, it is based on the assumption that evrey xcc must
+	 * have at least one active cu.
+	 */
+	for (i = 0; i < gfx_info->max_shader_engines && !found; i++) {
+		for (j = 0; j < gfx_info->max_sh_per_se && !found; j++) {
+			if (cu_info->bitmap[start][i % 4][j % 4]) {
+				cu_sibling_map_mask =
+					cu_info->bitmap[start][i % 4][j % 4];
+				found = true;
+			}
+		}
+	}
+
+	cu_sibling_map_mask &=
+		((1 << pcache_info[cache_type].num_cu_shared) - 1);
+	first_active_cu = ffs(cu_sibling_map_mask);
+
+	/* CU could be inactive. In case of shared cache find the first active
+	 * CU. and incase of non-shared cache check if the CU is inactive. If
+	 * inactive active skip it
+	 */
+	if (first_active_cu) {
+		pcache = kfd_alloc_struct(pcache);
+		if (!pcache)
+			return -ENOMEM;
 
-		/* Update the SYSFS tree, since we added another topology
-		 * device
+		memset(pcache, 0, sizeof(struct kfd_cache_properties));
+		pcache->processor_id_low = cu_processor_id
+					+ (first_active_cu - 1);
+		pcache->cache_level = pcache_info[cache_type].cache_level;
+		pcache->cacheline_size = pcache_info[cache_type].cache_line_size;
+
+		if (KFD_GC_VERSION(knode) == IP_VERSION(9, 4, 3) ||
+		    KFD_GC_VERSION(knode) == IP_VERSION(9, 4, 4) ||
+		    KFD_GC_VERSION(knode) == IP_VERSION(9, 5, 0))
+			mode = adev->gmc.gmc_funcs->query_mem_partition_mode(adev);
+		else
+			mode = UNKNOWN_MEMORY_PARTITION_MODE;
+
+		pcache->cache_size = pcache_info[cache_type].cache_size;
+		/* Partition mode only affects L3 cache size */
+		if (mode && pcache->cache_level == 3)
+			pcache->cache_size /= mode;
+
+		if (pcache_info[cache_type].flags & CRAT_CACHE_FLAGS_DATA_CACHE)
+			pcache->cache_type |= HSA_CACHE_TYPE_DATA;
+		if (pcache_info[cache_type].flags & CRAT_CACHE_FLAGS_INST_CACHE)
+			pcache->cache_type |= HSA_CACHE_TYPE_INSTRUCTION;
+		if (pcache_info[cache_type].flags & CRAT_CACHE_FLAGS_CPU_CACHE)
+			pcache->cache_type |= HSA_CACHE_TYPE_CPU;
+		if (pcache_info[cache_type].flags & CRAT_CACHE_FLAGS_SIMD_CACHE)
+			pcache->cache_type |= HSA_CACHE_TYPE_HSACU;
+
+		/* Sibling map is w.r.t processor_id_low, so shift out
+		 * inactive CU
 		 */
-		if (kfd_topology_update_sysfs() < 0)
-			kfd_topology_release_sysfs();
+		cu_sibling_map_mask = cu_sibling_map_mask >> (first_active_cu - 1);
+		k = 0;
+
+		for (xcc = start; xcc < end; xcc++) {
+			for (i = 0; i < gfx_info->max_shader_engines; i++) {
+				for (j = 0; j < gfx_info->max_sh_per_se; j++) {
+					pcache->sibling_map[k] = (uint8_t)(cu_sibling_map_mask & 0xFF);
+					pcache->sibling_map[k+1] = (uint8_t)((cu_sibling_map_mask >> 8) & 0xFF);
+					pcache->sibling_map[k+2] = (uint8_t)((cu_sibling_map_mask >> 16) & 0xFF);
+					pcache->sibling_map[k+3] = (uint8_t)((cu_sibling_map_mask >> 24) & 0xFF);
+					k += 4;
+
+					cu_sibling_map_mask = cu_info->bitmap[xcc][i % 4][j + i / 4];
+					cu_sibling_map_mask &= ((1 << pcache_info[cache_type].num_cu_shared) - 1);
+				}
+			}
+		}
+		pcache->sibling_map_size = k;
+		*props_ext = pcache;
+		return 0;
+	}
+	return 1;
+}
+
+#define KFD_MAX_CACHE_TYPES 6
 
+/* kfd_fill_cache_non_crat_info - Fill GPU cache info using kfd_gpu_cache_info
+ * tables
+ */
+static void kfd_fill_cache_non_crat_info(struct kfd_topology_device *dev, struct kfd_node *kdev)
+{
+	struct kfd_gpu_cache_info *pcache_info = NULL;
+	int i, j, k, xcc, start, end;
+	int ct = 0;
+	unsigned int cu_processor_id;
+	int ret;
+	unsigned int num_cu_shared;
+	struct amdgpu_cu_info *cu_info = &kdev->adev->gfx.cu_info;
+	struct amdgpu_gfx_config *gfx_info = &kdev->adev->gfx.config;
+	int gpu_processor_id;
+	struct kfd_cache_properties *props_ext = NULL;
+	int num_of_entries = 0;
+	int num_of_cache_types = 0;
+	struct kfd_gpu_cache_info cache_info[KFD_MAX_CACHE_TYPES];
+
+
+	gpu_processor_id = dev->node_props.simd_id_base;
+
+	memset(cache_info, 0, sizeof(cache_info));
+	pcache_info = cache_info;
+	num_of_cache_types = kfd_get_gpu_cache_info(kdev, &pcache_info);
+	if (!num_of_cache_types) {
+		pr_warn("no cache info found\n");
+		return;
 	}
 
+	/* For each type of cache listed in the kfd_gpu_cache_info table,
+	 * go through all available Compute Units.
+	 * The [i,j,k] loop will
+	 *		if kfd_gpu_cache_info.num_cu_shared = 1
+	 *			will parse through all available CU
+	 *		If (kfd_gpu_cache_info.num_cu_shared != 1)
+	 *			then it will consider only one CU from
+	 *			the shared unit
+	 */
+	start = ffs(kdev->xcc_mask) - 1;
+	end = start + NUM_XCC(kdev->xcc_mask);
+
+	for (ct = 0; ct < num_of_cache_types; ct++) {
+		cu_processor_id = gpu_processor_id;
+		if (pcache_info[ct].cache_level == 1) {
+			for (xcc = start; xcc < end; xcc++) {
+				for (i = 0; i < gfx_info->max_shader_engines; i++) {
+					for (j = 0; j < gfx_info->max_sh_per_se; j++) {
+						for (k = 0; k < gfx_info->max_cu_per_sh; k += pcache_info[ct].num_cu_shared) {
+
+							ret = fill_in_l1_pcache(&props_ext, pcache_info,
+										cu_info->bitmap[xcc][i % 4][j + i / 4], ct,
+										cu_processor_id, k);
+
+							if (ret < 0)
+								break;
+
+							if (!ret) {
+								num_of_entries++;
+								list_add_tail(&props_ext->list, &dev->cache_props);
+							}
+
+							/* Move to next CU block */
+							num_cu_shared = ((k + pcache_info[ct].num_cu_shared) <=
+								gfx_info->max_cu_per_sh) ?
+								pcache_info[ct].num_cu_shared :
+								(gfx_info->max_cu_per_sh - k);
+							cu_processor_id += num_cu_shared;
+						}
+					}
+				}
+			}
+		} else {
+			ret = fill_in_l2_l3_pcache(&props_ext, pcache_info,
+						   cu_info, gfx_info, ct, cu_processor_id, kdev);
+
+			if (ret < 0)
+				break;
+
+			if (!ret) {
+				num_of_entries++;
+				list_add_tail(&props_ext->list, &dev->cache_props);
+			}
+		}
+	}
+	dev->node_props.caches_count += num_of_entries;
+	pr_debug("Added [%d] GPU cache entries\n", num_of_entries);
+}
+
+static int kfd_topology_add_device_locked(struct kfd_node *gpu,
+					  struct kfd_topology_device **dev)
+{
+	int proximity_domain = ++topology_crat_proximity_domain;
+	struct list_head temp_topology_device_list;
+	void *crat_image = NULL;
+	size_t image_size = 0;
+	int res;
+
+	res = kfd_create_crat_image_virtual(&crat_image, &image_size,
+					    COMPUTE_UNIT_GPU, gpu,
+					    proximity_domain);
+	if (res) {
+		dev_err(gpu->adev->dev, "Error creating VCRAT\n");
+		topology_crat_proximity_domain--;
+		goto err;
+	}
+
+	INIT_LIST_HEAD(&temp_topology_device_list);
+
+	res = kfd_parse_crat_table(crat_image,
+				   &temp_topology_device_list,
+				   proximity_domain);
+	if (res) {
+		dev_err(gpu->adev->dev, "Error parsing VCRAT\n");
+		topology_crat_proximity_domain--;
+		goto err;
+	}
+
+	kfd_topology_update_device_list(&temp_topology_device_list,
+					&topology_device_list);
+
+	*dev = kfd_assign_gpu(gpu);
+	if (WARN_ON(!*dev)) {
+		res = -ENODEV;
+		goto err;
+	}
+
+	/* Fill the cache affinity information here for the GPUs
+	 * using VCRAT
+	 */
+	kfd_fill_cache_non_crat_info(*dev, gpu);
+
+	/* Update the SYSFS tree, since we added another topology
+	 * device
+	 */
+	res = kfd_topology_update_sysfs();
+	if (!res)
+		sys_props.generation_count++;
+	else
+		dev_err(gpu->adev->dev, "Failed to update GPU to sysfs topology. res=%d\n",
+			res);
+
+err:
+	kfd_destroy_crat_image(crat_image);
+	return res;
+}
+
+static void kfd_topology_set_dbg_firmware_support(struct kfd_topology_device *dev)
+{
+	bool firmware_supported = true;
+
+	if (KFD_GC_VERSION(dev->gpu) >= IP_VERSION(11, 0, 0) &&
+			KFD_GC_VERSION(dev->gpu) < IP_VERSION(12, 0, 0)) {
+		uint32_t mes_api_rev = (dev->gpu->adev->mes.sched_version &
+						AMDGPU_MES_API_VERSION_MASK) >>
+						AMDGPU_MES_API_VERSION_SHIFT;
+		uint32_t mes_rev = dev->gpu->adev->mes.sched_version &
+						AMDGPU_MES_VERSION_MASK;
+
+		firmware_supported = (mes_api_rev >= 14) && (mes_rev >= 64);
+		goto out;
+	}
+
+	/*
+	 * Note: Any unlisted devices here are assumed to support exception handling.
+	 * Add additional checks here as needed.
+	 */
+	switch (KFD_GC_VERSION(dev->gpu)) {
+	case IP_VERSION(9, 0, 1):
+		firmware_supported = dev->gpu->kfd->mec_fw_version >= 459 + 32768;
+		break;
+	case IP_VERSION(9, 1, 0):
+	case IP_VERSION(9, 2, 1):
+	case IP_VERSION(9, 2, 2):
+	case IP_VERSION(9, 3, 0):
+	case IP_VERSION(9, 4, 0):
+		firmware_supported = dev->gpu->kfd->mec_fw_version >= 459;
+		break;
+	case IP_VERSION(9, 4, 1):
+		firmware_supported = dev->gpu->kfd->mec_fw_version >= 60;
+		break;
+	case IP_VERSION(9, 4, 2):
+		firmware_supported = dev->gpu->kfd->mec_fw_version >= 51;
+		break;
+	case IP_VERSION(10, 1, 10):
+	case IP_VERSION(10, 1, 2):
+	case IP_VERSION(10, 1, 1):
+		firmware_supported = dev->gpu->kfd->mec_fw_version >= 144;
+		break;
+	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 1):
+	case IP_VERSION(10, 3, 4):
+	case IP_VERSION(10, 3, 5):
+		firmware_supported = dev->gpu->kfd->mec_fw_version >= 89;
+		break;
+	case IP_VERSION(10, 1, 3):
+	case IP_VERSION(10, 3, 3):
+		firmware_supported = false;
+		break;
+	default:
+		break;
+	}
+
+out:
+	if (firmware_supported)
+		dev->node_props.capability |= HSA_CAP_TRAP_DEBUG_FIRMWARE_SUPPORTED;
+}
+
+static void kfd_topology_set_capabilities(struct kfd_topology_device *dev)
+{
+	dev->node_props.capability |= ((HSA_CAP_DOORBELL_TYPE_2_0 <<
+				HSA_CAP_DOORBELL_TYPE_TOTALBITS_SHIFT) &
+				HSA_CAP_DOORBELL_TYPE_TOTALBITS_MASK);
+
+	dev->node_props.capability |= HSA_CAP_TRAP_DEBUG_SUPPORT |
+			HSA_CAP_TRAP_DEBUG_WAVE_LAUNCH_TRAP_OVERRIDE_SUPPORTED |
+			HSA_CAP_TRAP_DEBUG_WAVE_LAUNCH_MODE_SUPPORTED;
+
+	if (kfd_dbg_has_ttmps_always_setup(dev->gpu))
+		dev->node_props.debug_prop |= HSA_DBG_DISPATCH_INFO_ALWAYS_VALID;
+
+	if (KFD_GC_VERSION(dev->gpu) < IP_VERSION(10, 0, 0)) {
+		if (KFD_GC_VERSION(dev->gpu) == IP_VERSION(9, 4, 3) ||
+		    KFD_GC_VERSION(dev->gpu) == IP_VERSION(9, 4, 4))
+			dev->node_props.debug_prop |=
+				HSA_DBG_WATCH_ADDR_MASK_LO_BIT_GFX9_4_3 |
+				HSA_DBG_WATCH_ADDR_MASK_HI_BIT_GFX9_4_3;
+		else
+			dev->node_props.debug_prop |=
+				HSA_DBG_WATCH_ADDR_MASK_LO_BIT_GFX9 |
+				HSA_DBG_WATCH_ADDR_MASK_HI_BIT;
+
+		if (KFD_GC_VERSION(dev->gpu) >= IP_VERSION(9, 4, 2))
+			dev->node_props.capability |=
+				HSA_CAP_TRAP_DEBUG_PRECISE_MEMORY_OPERATIONS_SUPPORTED;
+
+		if (!amdgpu_sriov_vf(dev->gpu->adev))
+			dev->node_props.capability |= HSA_CAP_PER_QUEUE_RESET_SUPPORTED;
+
+	} else {
+		dev->node_props.debug_prop |= HSA_DBG_WATCH_ADDR_MASK_LO_BIT_GFX10 |
+					HSA_DBG_WATCH_ADDR_MASK_HI_BIT;
+
+		if (KFD_GC_VERSION(dev->gpu) >= IP_VERSION(12, 0, 0))
+			dev->node_props.capability |=
+				HSA_CAP_TRAP_DEBUG_PRECISE_ALU_OPERATIONS_SUPPORTED;
+	}
+
+	kfd_topology_set_dbg_firmware_support(dev);
+}
+
+int kfd_topology_add_device(struct kfd_node *gpu)
+{
+	uint32_t gpu_id;
+	struct kfd_topology_device *dev;
+	int res = 0;
+	int i;
+	const char *asic_name = amdgpu_asic_name[gpu->adev->asic_type];
+	struct amdgpu_gfx_config *gfx_info = &gpu->adev->gfx.config;
+	struct amdgpu_cu_info *cu_info = &gpu->adev->gfx.cu_info;
+
+	if (gpu->xcp && !gpu->xcp->ddev) {
+		dev_warn(gpu->adev->dev,
+			 "Won't add GPU to topology since it has no drm node assigned.");
+		return 0;
+	} else {
+		dev_dbg(gpu->adev->dev, "Adding new GPU to topology\n");
+	}
+
+	/* Check to see if this gpu device exists in the topology_device_list.
+	 * If so, assign the gpu to that device,
+	 * else create a Virtual CRAT for this gpu device and then parse that
+	 * CRAT to create a new topology device. Once created assign the gpu to
+	 * that topology device
+	 */
+	down_write(&topology_lock);
+	dev = kfd_assign_gpu(gpu);
+	if (!dev)
+		res = kfd_topology_add_device_locked(gpu, &dev);
+	up_write(&topology_lock);
+	if (res)
+		return res;
+
+	gpu_id = kfd_generate_gpu_id(gpu);
 	dev->gpu_id = gpu_id;
 	gpu->id = gpu_id;
-	dev->node_props.vendor_id = gpu->pdev->vendor;
-	dev->node_props.device_id = gpu->pdev->device;
-	dev->node_props.location_id = (gpu->pdev->bus->number << 24) +
-			(gpu->pdev->devfn & 0xffffff);
+
+	kfd_dev_create_p2p_links();
+
+	/* TODO: Move the following lines to function
+	 *	kfd_add_non_crat_information
+	 */
+
+	/* Fill-in additional information that is not available in CRAT but
+	 * needed for the topology
+	 */
+	for (i = 0; i < KFD_TOPOLOGY_PUBLIC_NAME_SIZE-1; i++) {
+		dev->node_props.name[i] = __tolower(asic_name[i]);
+		if (asic_name[i] == '\0')
+			break;
+	}
+	dev->node_props.name[i] = '\0';
+
+	dev->node_props.simd_arrays_per_engine =
+		gfx_info->max_sh_per_se;
+
+	dev->node_props.gfx_target_version =
+				gpu->kfd->device_info.gfx_target_version;
+	dev->node_props.vendor_id = gpu->adev->pdev->vendor;
+	dev->node_props.device_id = gpu->adev->pdev->device;
+	dev->node_props.capability |=
+		((dev->gpu->adev->rev_id << HSA_CAP_ASIC_REVISION_SHIFT) &
+			HSA_CAP_ASIC_REVISION_MASK);
+
+	dev->node_props.location_id = pci_dev_id(gpu->adev->pdev);
+	if (gpu->kfd->num_nodes > 1)
+		dev->node_props.location_id |= dev->gpu->node_id;
+
+	dev->node_props.domain = pci_domain_nr(gpu->adev->pdev->bus);
+	dev->node_props.max_engine_clk_fcompute =
+		amdgpu_amdkfd_get_max_engine_clock_in_mhz(dev->gpu->adev);
+	dev->node_props.max_engine_clk_ccompute =
+		cpufreq_quick_get_max(0) / 1000;
+
+	if (gpu->xcp)
+		dev->node_props.drm_render_minor = gpu->xcp->ddev->render->index;
+	else
+		dev->node_props.drm_render_minor =
+				gpu->kfd->shared_resources.drm_render_minor;
+
+	dev->node_props.hive_id = gpu->kfd->hive_id;
+	dev->node_props.num_sdma_engines = kfd_get_num_sdma_engines(gpu);
+	dev->node_props.num_sdma_xgmi_engines =
+					kfd_get_num_xgmi_sdma_engines(gpu);
+	dev->node_props.num_sdma_queues_per_engine =
+				gpu->kfd->device_info.num_sdma_queues_per_engine -
+				gpu->kfd->device_info.num_reserved_sdma_queues_per_engine;
+	dev->node_props.num_gws = (dev->gpu->gws &&
+		dev->gpu->dqm->sched_policy != KFD_SCHED_POLICY_NO_HWS) ?
+		dev->gpu->adev->gds.gws_size : 0;
+	dev->node_props.num_cp_queues = get_cp_queues_num(dev->gpu->dqm);
+
+	kfd_fill_mem_clk_max_info(dev);
+	kfd_fill_iolink_non_crat_info(dev);
+
+	switch (dev->gpu->adev->asic_type) {
+	case CHIP_KAVERI:
+	case CHIP_HAWAII:
+	case CHIP_TONGA:
+		dev->node_props.capability |= ((HSA_CAP_DOORBELL_TYPE_PRE_1_0 <<
+			HSA_CAP_DOORBELL_TYPE_TOTALBITS_SHIFT) &
+			HSA_CAP_DOORBELL_TYPE_TOTALBITS_MASK);
+		break;
+	case CHIP_CARRIZO:
+	case CHIP_FIJI:
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
+		pr_debug("Adding doorbell packet type capability\n");
+		dev->node_props.capability |= ((HSA_CAP_DOORBELL_TYPE_1_0 <<
+			HSA_CAP_DOORBELL_TYPE_TOTALBITS_SHIFT) &
+			HSA_CAP_DOORBELL_TYPE_TOTALBITS_MASK);
+		break;
+	default:
+		if (KFD_GC_VERSION(dev->gpu) < IP_VERSION(9, 0, 1))
+			WARN(1, "Unexpected ASIC family %u",
+			     dev->gpu->adev->asic_type);
+		else
+			kfd_topology_set_capabilities(dev);
+	}
+
 	/*
-	 * TODO: Retrieve max engine clock values from KGD
+	 * Overwrite ATS capability according to needs_iommu_device to fix
+	 * potential missing corresponding bit in CRAT of BIOS.
 	 */
+	dev->node_props.capability &= ~HSA_CAP_ATS_PRESENT;
 
-	if (dev->gpu->device_info->asic_family == CHIP_CARRIZO) {
-		dev->node_props.capability |= HSA_CAP_DOORBELL_PACKET_TYPE;
-		pr_info("Adding doorbell packet type capability\n");
+	/* Fix errors in CZ CRAT.
+	 * simd_count: Carrizo CRAT reports wrong simd_count, probably
+	 *		because it doesn't consider masked out CUs
+	 * max_waves_per_simd: Carrizo reports wrong max_waves_per_simd
+	 */
+	if (dev->gpu->adev->asic_type == CHIP_CARRIZO) {
+		dev->node_props.simd_count =
+			cu_info->simd_per_cu * cu_info->number;
+		dev->node_props.max_waves_per_simd = 10;
 	}
 
-	res = 0;
+	/* kfd only concerns sram ecc on GFX and HBM ecc on UMC */
+	dev->node_props.capability |=
+		((dev->gpu->adev->ras_enabled & BIT(AMDGPU_RAS_BLOCK__GFX)) != 0) ?
+		HSA_CAP_SRAM_EDCSUPPORTED : 0;
+	dev->node_props.capability |=
+		((dev->gpu->adev->ras_enabled & BIT(AMDGPU_RAS_BLOCK__UMC)) != 0) ?
+		HSA_CAP_MEM_EDCSUPPORTED : 0;
 
-err:
-	up_write(&topology_lock);
+	if (KFD_GC_VERSION(dev->gpu) != IP_VERSION(9, 0, 1))
+		dev->node_props.capability |= (dev->gpu->adev->ras_enabled != 0) ?
+			HSA_CAP_RASEVENTNOTIFY : 0;
 
-	if (res == 0)
-		kfd_notify_gpu_change(gpu_id, 1);
+	if (KFD_IS_SVM_API_SUPPORTED(dev->gpu->adev))
+		dev->node_props.capability |= HSA_CAP_SVMAPI_SUPPORTED;
 
-	return res;
+	if (dev->gpu->adev->gmc.is_app_apu ||
+		dev->gpu->adev->gmc.xgmi.connected_to_cpu)
+		dev->node_props.capability |= HSA_CAP_FLAGS_COHERENTHOSTACCESS;
+
+	kfd_queue_ctx_save_restore_size(dev);
+
+	kfd_debug_print_topology();
+
+	kfd_notify_gpu_change(gpu_id, 1);
+
+	return 0;
 }
 
-int kfd_topology_remove_device(struct kfd_dev *gpu)
+/**
+ * kfd_topology_update_io_links() - Update IO links after device removal.
+ * @proximity_domain: Proximity domain value of the dev being removed.
+ *
+ * The topology list currently is arranged in increasing order of
+ * proximity domain.
+ *
+ * Two things need to be done when a device is removed:
+ * 1. All the IO links to this device need to be removed.
+ * 2. All nodes after the current device node need to move
+ *    up once this device node is removed from the topology
+ *    list. As a result, the proximity domain values for
+ *    all nodes after the node being deleted reduce by 1.
+ *    This would also cause the proximity domain values for
+ *    io links to be updated based on new proximity domain
+ *    values.
+ *
+ * Context: The caller must hold write topology_lock.
+ */
+static void kfd_topology_update_io_links(int proximity_domain)
 {
 	struct kfd_topology_device *dev;
+	struct kfd_iolink_properties *iolink, *p2plink, *tmp;
+
+	list_for_each_entry(dev, &topology_device_list, list) {
+		if (dev->proximity_domain > proximity_domain)
+			dev->proximity_domain--;
+
+		list_for_each_entry_safe(iolink, tmp, &dev->io_link_props, list) {
+			/*
+			 * If there is an io link to the dev being deleted
+			 * then remove that IO link also.
+			 */
+			if (iolink->node_to == proximity_domain) {
+				list_del(&iolink->list);
+				dev->node_props.io_links_count--;
+			} else {
+				if (iolink->node_from > proximity_domain)
+					iolink->node_from--;
+				if (iolink->node_to > proximity_domain)
+					iolink->node_to--;
+			}
+		}
+
+		list_for_each_entry_safe(p2plink, tmp, &dev->p2p_link_props, list) {
+			/*
+			 * If there is a p2p link to the dev being deleted
+			 * then remove that p2p link also.
+			 */
+			if (p2plink->node_to == proximity_domain) {
+				list_del(&p2plink->list);
+				dev->node_props.p2p_links_count--;
+			} else {
+				if (p2plink->node_from > proximity_domain)
+					p2plink->node_from--;
+				if (p2plink->node_to > proximity_domain)
+					p2plink->node_to--;
+			}
+		}
+	}
+}
+
+int kfd_topology_remove_device(struct kfd_node *gpu)
+{
+	struct kfd_topology_device *dev, *tmp;
 	uint32_t gpu_id;
 	int res = -ENODEV;
+	int i = 0;
 
 	down_write(&topology_lock);
 
-	list_for_each_entry(dev, &topology_device_list, list)
+	list_for_each_entry_safe(dev, tmp, &topology_device_list, list) {
 		if (dev->gpu == gpu) {
 			gpu_id = dev->gpu_id;
 			kfd_remove_sysfs_node_entry(dev);
 			kfd_release_topology_device(dev);
+			sys_props.num_devices--;
+			kfd_topology_update_io_links(i);
+			topology_crat_proximity_domain = sys_props.num_devices-1;
+			sys_props.generation_count++;
 			res = 0;
 			if (kfd_topology_update_sysfs() < 0)
 				kfd_topology_release_sysfs();
 			break;
 		}
+		i++;
+	}
 
 	up_write(&topology_lock);
 
-	if (res == 0)
+	if (!res)
 		kfd_notify_gpu_change(gpu_id, 0);
 
 	return res;
 }
 
-/*
- * When idx is out of bounds, the function will return NULL
+/* kfd_topology_enum_kfd_devices - Enumerate through all devices in KFD
+ *	topology. If GPU device is found @idx, then valid kfd_dev pointer is
+ *	returned through @kdev
+ * Return -	0: On success (@kdev will be NULL for non GPU nodes)
+ *		-1: If end of list
  */
-struct kfd_dev *kfd_topology_enum_kfd_devices(uint8_t idx)
+int kfd_topology_enum_kfd_devices(uint8_t idx, struct kfd_node **kdev)
 {
 
 	struct kfd_topology_device *top_dev;
-	struct kfd_dev *device = NULL;
 	uint8_t device_idx = 0;
 
+	*kdev = NULL;
 	down_read(&topology_lock);
 
 	list_for_each_entry(top_dev, &topology_device_list, list) {
 		if (device_idx == idx) {
-			device = top_dev->gpu;
-			break;
+			*kdev = top_dev->gpu;
+			up_read(&topology_lock);
+			return 0;
 		}
 
 		device_idx++;
@@ -1230,6 +2316,89 @@ struct kfd_dev *kfd_topology_enum_kfd_devices(uint8_t idx)
 
 	up_read(&topology_lock);
 
-	return device;
+	return -1;
 
 }
+
+static int kfd_cpumask_to_apic_id(const struct cpumask *cpumask)
+{
+	int first_cpu_of_numa_node;
+
+	if (!cpumask || cpumask == cpu_none_mask)
+		return -1;
+	first_cpu_of_numa_node = cpumask_first(cpumask);
+	if (first_cpu_of_numa_node >= nr_cpu_ids)
+		return -1;
+#ifdef CONFIG_X86_64
+	return cpu_data(first_cpu_of_numa_node).topo.apicid;
+#else
+	return first_cpu_of_numa_node;
+#endif
+}
+
+/* kfd_numa_node_to_apic_id - Returns the APIC ID of the first logical processor
+ *	of the given NUMA node (numa_node_id)
+ * Return -1 on failure
+ */
+int kfd_numa_node_to_apic_id(int numa_node_id)
+{
+	if (numa_node_id == -1) {
+		pr_warn("Invalid NUMA Node. Use online CPU mask\n");
+		return kfd_cpumask_to_apic_id(cpu_online_mask);
+	}
+	return kfd_cpumask_to_apic_id(cpumask_of_node(numa_node_id));
+}
+
+#if defined(CONFIG_DEBUG_FS)
+
+int kfd_debugfs_hqds_by_device(struct seq_file *m, void *data)
+{
+	struct kfd_topology_device *dev;
+	unsigned int i = 0;
+	int r = 0;
+
+	down_read(&topology_lock);
+
+	list_for_each_entry(dev, &topology_device_list, list) {
+		if (!dev->gpu) {
+			i++;
+			continue;
+		}
+
+		seq_printf(m, "Node %u, gpu_id %x:\n", i++, dev->gpu->id);
+		r = dqm_debugfs_hqds(m, dev->gpu->dqm);
+		if (r)
+			break;
+	}
+
+	up_read(&topology_lock);
+
+	return r;
+}
+
+int kfd_debugfs_rls_by_device(struct seq_file *m, void *data)
+{
+	struct kfd_topology_device *dev;
+	unsigned int i = 0;
+	int r = 0;
+
+	down_read(&topology_lock);
+
+	list_for_each_entry(dev, &topology_device_list, list) {
+		if (!dev->gpu) {
+			i++;
+			continue;
+		}
+
+		seq_printf(m, "Node %u, gpu_id %x:\n", i++, dev->gpu->id);
+		r = pm_debugfs_runlist(m, &dev->gpu->dqm->packet_mgr);
+		if (r)
+			break;
+	}
+
+	up_read(&topology_lock);
+
+	return r;
+}
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_topology.h b/drivers/gpu/drm/amd/amdkfd/kfd_topology.h
index c3ddb9b95ff8..3de8ec0043bb 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_topology.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_topology.h
@@ -1,5 +1,6 @@
+/* SPDX-License-Identifier: GPL-2.0 OR MIT */
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -23,37 +24,39 @@
 #ifndef __KFD_TOPOLOGY_H__
 #define __KFD_TOPOLOGY_H__
 
+#include <linux/dmi.h>
 #include <linux/types.h>
 #include <linux/list.h>
-#include "kfd_priv.h"
+#include <linux/kfd_sysfs.h>
+#include "kfd_crat.h"
 
-#define KFD_TOPOLOGY_PUBLIC_NAME_SIZE 128
+#define KFD_TOPOLOGY_PUBLIC_NAME_SIZE 32
 
-#define HSA_CAP_HOT_PLUGGABLE			0x00000001
-#define HSA_CAP_ATS_PRESENT			0x00000002
-#define HSA_CAP_SHARED_WITH_GRAPHICS		0x00000004
-#define HSA_CAP_QUEUE_SIZE_POW2			0x00000008
-#define HSA_CAP_QUEUE_SIZE_32BIT		0x00000010
-#define HSA_CAP_QUEUE_IDLE_EVENT		0x00000020
-#define HSA_CAP_VA_LIMIT			0x00000040
-#define HSA_CAP_WATCH_POINTS_SUPPORTED		0x00000080
-#define HSA_CAP_WATCH_POINTS_TOTALBITS_MASK	0x00000f00
-#define HSA_CAP_WATCH_POINTS_TOTALBITS_SHIFT	8
-#define HSA_CAP_RESERVED			0xfffff000
-#define HSA_CAP_DOORBELL_PACKET_TYPE		0x00001000
+#define HSA_DBG_WATCH_ADDR_MASK_LO_BIT_GFX9	6
+#define HSA_DBG_WATCH_ADDR_MASK_LO_BIT_GFX9_4_3 7
+#define HSA_DBG_WATCH_ADDR_MASK_LO_BIT_GFX10	7
+#define HSA_DBG_WATCH_ADDR_MASK_HI_BIT  \
+			(29 << HSA_DBG_WATCH_ADDR_MASK_HI_BIT_SHIFT)
+#define HSA_DBG_WATCH_ADDR_MASK_HI_BIT_GFX9_4_3 \
+			(30 << HSA_DBG_WATCH_ADDR_MASK_HI_BIT_SHIFT)
 
 struct kfd_node_properties {
+	uint64_t hive_id;
 	uint32_t cpu_cores_count;
 	uint32_t simd_count;
 	uint32_t mem_banks_count;
 	uint32_t caches_count;
 	uint32_t io_links_count;
+	uint32_t p2p_links_count;
 	uint32_t cpu_core_id_base;
 	uint32_t simd_id_base;
 	uint32_t capability;
+	uint32_t capability2;
+	uint64_t debug_prop;
 	uint32_t max_waves_per_simd;
 	uint32_t lds_size_in_kb;
 	uint32_t gds_size_in_kb;
+	uint32_t num_gws;
 	uint32_t wave_front_size;
 	uint32_t array_count;
 	uint32_t simd_arrays_per_engine;
@@ -61,25 +64,25 @@ struct kfd_node_properties {
 	uint32_t simd_per_cu;
 	uint32_t max_slots_scratch_cu;
 	uint32_t engine_id;
+	uint32_t gfx_target_version;
 	uint32_t vendor_id;
 	uint32_t device_id;
 	uint32_t location_id;
+	uint32_t domain;
 	uint32_t max_engine_clk_fcompute;
 	uint32_t max_engine_clk_ccompute;
-	uint16_t marketing_name[KFD_TOPOLOGY_PUBLIC_NAME_SIZE];
+	int32_t  drm_render_minor;
+	uint32_t num_sdma_engines;
+	uint32_t num_sdma_xgmi_engines;
+	uint32_t num_sdma_queues_per_engine;
+	uint32_t num_cp_queues;
+	uint32_t cwsr_size;
+	uint32_t ctl_stack_size;
+	uint32_t eop_buffer_size;
+	uint32_t debug_memory_size;
+	char name[KFD_TOPOLOGY_PUBLIC_NAME_SIZE];
 };
 
-#define HSA_MEM_HEAP_TYPE_SYSTEM	0
-#define HSA_MEM_HEAP_TYPE_FB_PUBLIC	1
-#define HSA_MEM_HEAP_TYPE_FB_PRIVATE	2
-#define HSA_MEM_HEAP_TYPE_GPU_GDS	3
-#define HSA_MEM_HEAP_TYPE_GPU_LDS	4
-#define HSA_MEM_HEAP_TYPE_GPU_SCRATCH	5
-
-#define HSA_MEM_FLAGS_HOT_PLUGGABLE	0x00000001
-#define HSA_MEM_FLAGS_NON_VOLATILE	0x00000002
-#define HSA_MEM_FLAGS_RESERVED		0xfffffffc
-
 struct kfd_mem_properties {
 	struct list_head	list;
 	uint32_t		heap_type;
@@ -87,17 +90,12 @@ struct kfd_mem_properties {
 	uint32_t		flags;
 	uint32_t		width;
 	uint32_t		mem_clk_max;
+	struct kfd_node		*gpu;
 	struct kobject		*kobj;
 	struct attribute	attr;
 };
 
-#define KFD_TOPOLOGY_CPU_SIBLINGS 256
-
-#define HSA_CACHE_TYPE_DATA		0x00000001
-#define HSA_CACHE_TYPE_INSTRUCTION	0x00000002
-#define HSA_CACHE_TYPE_CPU		0x00000004
-#define HSA_CACHE_TYPE_HSACU		0x00000008
-#define HSA_CACHE_TYPE_RESERVED		0xfffffff0
+#define CACHE_SIBLINGMAP_SIZE 128
 
 struct kfd_cache_properties {
 	struct list_head	list;
@@ -109,9 +107,11 @@ struct kfd_cache_properties {
 	uint32_t		cache_assoc;
 	uint32_t		cache_latency;
 	uint32_t		cache_type;
-	uint8_t			sibling_map[KFD_TOPOLOGY_CPU_SIBLINGS];
+	uint8_t			sibling_map[CACHE_SIBLINGMAP_SIZE];
+	struct kfd_node		*gpu;
 	struct kobject		*kobj;
 	struct attribute	attr;
+	uint32_t		sibling_map_size;
 };
 
 struct kfd_iolink_properties {
@@ -127,29 +127,46 @@ struct kfd_iolink_properties {
 	uint32_t		min_bandwidth;
 	uint32_t		max_bandwidth;
 	uint32_t		rec_transfer_size;
+	uint32_t		rec_sdma_eng_id_mask;
 	uint32_t		flags;
+	struct kfd_node		*gpu;
 	struct kobject		*kobj;
 	struct attribute	attr;
 };
 
+struct kfd_perf_properties {
+	struct list_head	list;
+	char			block_name[16];
+	uint32_t		max_concurrent;
+	struct attribute_group	*attr_group;
+};
+
 struct kfd_topology_device {
 	struct list_head		list;
 	uint32_t			gpu_id;
+	uint32_t			proximity_domain;
 	struct kfd_node_properties	node_props;
-	uint32_t			mem_bank_count;
 	struct list_head		mem_props;
-	uint32_t			cache_count;
 	struct list_head		cache_props;
-	uint32_t			io_link_count;
 	struct list_head		io_link_props;
-	struct kfd_dev			*gpu;
+	struct list_head		p2p_link_props;
+	struct list_head		perf_props;
+	struct kfd_node			*gpu;
 	struct kobject			*kobj_node;
 	struct kobject			*kobj_mem;
 	struct kobject			*kobj_cache;
 	struct kobject			*kobj_iolink;
+	struct kobject			*kobj_p2plink;
+	struct kobject			*kobj_perf;
 	struct attribute		attr_gpuid;
 	struct attribute		attr_name;
 	struct attribute		attr_props;
+	union {
+		uint8_t				oem_id[CRAT_OEMID_LENGTH];
+		uint64_t			oem_id64;
+	};
+	uint8_t				oem_table_id[CRAT_OEMTABLEID_LENGTH];
+	uint32_t			oem_revision;
 };
 
 struct kfd_system_properties {
@@ -164,6 +181,24 @@ struct kfd_system_properties {
 	struct attribute	attr_props;
 };
 
-
+struct dmi_mem_device {
+	struct dmi_header header;
+	u16 physical_handle;
+	u16 error_handle;
+	u16 total_width;
+	u16 data_width;
+	u16 size;
+	u8 form_factor;
+	u8 device_set;
+	u8 device_locator;
+	u8 bank_locator;
+	u8 memory_type;
+	u16 type_detail;
+	u16 speed;
+} __packed;
+
+struct kfd_topology_device *kfd_create_topology_device(
+		struct list_head *device_list);
+void kfd_release_topology_device_list(struct list_head *device_list);
 
 #endif /* __KFD_TOPOLOGY_H__ */
diff --git a/drivers/gpu/drm/amd/amdkfd/soc15_int.h b/drivers/gpu/drm/amd/amdkfd/soc15_int.h
new file mode 100644
index 000000000000..e5c0205f2618
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdkfd/soc15_int.h
@@ -0,0 +1,51 @@
+/*
+ * Copyright 2016-2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#ifndef HSA_SOC15_INT_H_INCLUDED
+#define HSA_SOC15_INT_H_INCLUDED
+
+#include "soc15_ih_clientid.h"
+
+#define SOC15_INTSRC_CP_END_OF_PIPE	181
+#define SOC15_INTSRC_CP_BAD_OPCODE	183
+#define SOC15_INTSRC_SQ_INTERRUPT_MSG	239
+#define SOC15_INTSRC_VMC_FAULT		0
+#define SOC15_INTSRC_VMC_UTCL2_POISON	1
+#define SOC15_INTSRC_SDMA_TRAP		224
+#define SOC15_INTSRC_SDMA_ECC		220
+#define SOC21_INTSRC_SDMA_TRAP		49
+#define SOC21_INTSRC_SDMA_ECC		62
+
+#define SOC15_CLIENT_ID_FROM_IH_ENTRY(entry) (le32_to_cpu(entry[0]) & 0xff)
+#define SOC15_SOURCE_ID_FROM_IH_ENTRY(entry) (le32_to_cpu(entry[0]) >> 8 & 0xff)
+#define SOC15_RING_ID_FROM_IH_ENTRY(entry) (le32_to_cpu(entry[0]) >> 16 & 0xff)
+#define SOC15_VMID_FROM_IH_ENTRY(entry) (le32_to_cpu(entry[0]) >> 24 & 0xf)
+#define SOC15_VMID_TYPE_FROM_IH_ENTRY(entry) (le32_to_cpu(entry[0]) >> 31 & 0x1)
+#define SOC15_PASID_FROM_IH_ENTRY(entry) (le32_to_cpu(entry[3]) & 0xffff)
+#define SOC15_NODEID_FROM_IH_ENTRY(entry) (le32_to_cpu(entry[3]) >> 16 & 0xff)
+#define SOC15_CONTEXT_ID0_FROM_IH_ENTRY(entry) (le32_to_cpu(entry[4]))
+#define SOC15_CONTEXT_ID1_FROM_IH_ENTRY(entry) (le32_to_cpu(entry[5]))
+#define SOC15_CONTEXT_ID2_FROM_IH_ENTRY(entry) (le32_to_cpu(entry[6]))
+#define SOC15_CONTEXT_ID3_FROM_IH_ENTRY(entry) (le32_to_cpu(entry[7]))
+
+#endif
+
diff --git a/drivers/gpu/drm/amd/amdxcp/Makefile b/drivers/gpu/drm/amd/amdxcp/Makefile
new file mode 100644
index 000000000000..870501a4bb8c
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdxcp/Makefile
@@ -0,0 +1,25 @@
+#
+# Copyright 2023 Advanced Micro Devices, Inc.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+
+amdxcp-y := amdgpu_xcp_drv.o
+
+obj-$(CONFIG_DRM_AMDGPU) += amdxcp.o
diff --git a/drivers/gpu/drm/amd/amdxcp/amdgpu_xcp_drv.c b/drivers/gpu/drm/amd/amdxcp/amdgpu_xcp_drv.c
new file mode 100644
index 000000000000..44009aa8216e
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdxcp/amdgpu_xcp_drv.c
@@ -0,0 +1,152 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/export.h>
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/platform_device.h>
+
+#include <drm/drm_drv.h>
+
+#include "amdgpu_xcp_drv.h"
+
+#define MAX_XCP_PLATFORM_DEVICE 64
+
+struct xcp_device {
+	struct drm_device drm;
+	struct platform_device *pdev;
+};
+
+static const struct drm_driver amdgpu_xcp_driver = {
+	.driver_features = DRIVER_GEM | DRIVER_RENDER,
+	.name = "amdgpu_xcp_drv",
+	.major = 1,
+	.minor = 0,
+};
+
+static int8_t pdev_num;
+static struct xcp_device *xcp_dev[MAX_XCP_PLATFORM_DEVICE];
+static DEFINE_MUTEX(xcp_mutex);
+
+int amdgpu_xcp_drm_dev_alloc(struct drm_device **ddev)
+{
+	struct platform_device *pdev;
+	struct xcp_device *pxcp_dev;
+	char dev_name[20];
+	int ret, i;
+
+	guard(mutex)(&xcp_mutex);
+
+	if (pdev_num >= MAX_XCP_PLATFORM_DEVICE)
+		return -ENODEV;
+
+	for (i = 0; i < MAX_XCP_PLATFORM_DEVICE; i++) {
+		if (!xcp_dev[i])
+			break;
+	}
+
+	if (i >= MAX_XCP_PLATFORM_DEVICE)
+		return -ENODEV;
+
+	snprintf(dev_name, sizeof(dev_name), "amdgpu_xcp_%d", i);
+	pdev = platform_device_register_simple(dev_name, -1, NULL, 0);
+	if (IS_ERR(pdev))
+		return PTR_ERR(pdev);
+
+	if (!devres_open_group(&pdev->dev, NULL, GFP_KERNEL)) {
+		ret = -ENOMEM;
+		goto out_unregister;
+	}
+
+	pxcp_dev = devm_drm_dev_alloc(&pdev->dev, &amdgpu_xcp_driver, struct xcp_device, drm);
+	if (IS_ERR(pxcp_dev)) {
+		ret = PTR_ERR(pxcp_dev);
+		goto out_devres;
+	}
+
+	xcp_dev[i] = pxcp_dev;
+	xcp_dev[i]->pdev = pdev;
+	*ddev = &pxcp_dev->drm;
+	pdev_num++;
+
+	return 0;
+
+out_devres:
+	devres_release_group(&pdev->dev, NULL);
+out_unregister:
+	platform_device_unregister(pdev);
+
+	return ret;
+}
+EXPORT_SYMBOL(amdgpu_xcp_drm_dev_alloc);
+
+static void free_xcp_dev(int8_t index)
+{
+	if ((index < MAX_XCP_PLATFORM_DEVICE) && (xcp_dev[index])) {
+		struct platform_device *pdev = xcp_dev[index]->pdev;
+
+		devres_release_group(&pdev->dev, NULL);
+		platform_device_unregister(pdev);
+
+		xcp_dev[index] = NULL;
+		pdev_num--;
+	}
+}
+
+void amdgpu_xcp_drm_dev_free(struct drm_device *ddev)
+{
+	int8_t i;
+
+	guard(mutex)(&xcp_mutex);
+
+	for (i = 0; i < MAX_XCP_PLATFORM_DEVICE; i++) {
+		if ((xcp_dev[i]) && (&xcp_dev[i]->drm == ddev)) {
+			free_xcp_dev(i);
+			break;
+		}
+	}
+}
+EXPORT_SYMBOL(amdgpu_xcp_drm_dev_free);
+
+void amdgpu_xcp_drv_release(void)
+{
+	int8_t i;
+
+	guard(mutex)(&xcp_mutex);
+
+	for (i = 0; pdev_num && i < MAX_XCP_PLATFORM_DEVICE; i++) {
+		free_xcp_dev(i);
+	}
+}
+EXPORT_SYMBOL(amdgpu_xcp_drv_release);
+
+static void __exit amdgpu_xcp_drv_exit(void)
+{
+	amdgpu_xcp_drv_release();
+}
+
+module_exit(amdgpu_xcp_drv_exit);
+
+MODULE_AUTHOR("AMD linux driver team");
+MODULE_DESCRIPTION("AMD XCP PLATFORM DEVICES");
+MODULE_LICENSE("GPL and additional rights");
diff --git a/drivers/gpu/drm/amd/amdxcp/amdgpu_xcp_drv.h b/drivers/gpu/drm/amd/amdxcp/amdgpu_xcp_drv.h
new file mode 100644
index 000000000000..580a1602c8e3
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdxcp/amdgpu_xcp_drv.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef _AMDGPU_XCP_DRV_H_
+#define _AMDGPU_XCP_DRV_H_
+
+int amdgpu_xcp_drm_dev_alloc(struct drm_device **ddev);
+void amdgpu_xcp_drm_dev_free(struct drm_device *ddev);
+void amdgpu_xcp_drv_release(void);
+#endif /* _AMDGPU_XCP_DRV_H_ */
diff --git a/drivers/gpu/drm/amd/display/Kconfig b/drivers/gpu/drm/amd/display/Kconfig
new file mode 100644
index 000000000000..abd3b6564373
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/Kconfig
@@ -0,0 +1,59 @@
+# SPDX-License-Identifier: MIT
+# Copyright © 2019-2024 Advanced Micro Devices, Inc. All rights reserved.
+
+menu "Display Engine Configuration"
+	depends on DRM && DRM_AMDGPU
+
+config DRM_AMD_DC
+	bool "AMD DC - Enable new display engine"
+	default y
+	depends on BROKEN || !CC_IS_CLANG || ARM64 || LOONGARCH || RISCV || SPARC64 || X86_64
+	select CEC_CORE
+	select CEC_NOTIFIER
+	select SND_HDA_COMPONENT if SND_HDA_CORE
+	# !CC_IS_CLANG: https://github.com/ClangBuiltLinux/linux/issues/1752
+	select DRM_AMD_DC_FP if ARCH_HAS_KERNEL_FPU_SUPPORT && !(CC_IS_CLANG && (ARM64 || LOONGARCH || RISCV))
+	help
+	  Choose this option if you want to use the new display engine
+	  support for AMDGPU. This adds required support for Vega and
+	  Raven ASICs.
+
+	  calculate_bandwidth() is presently broken on all !(X86_64 || SPARC64 ||
+	  ARM64 || LOONGARCH || RISCV) architectures built with Clang (all released
+	  versions), whereby the stack frame gets blown up to well over 5k.  This
+	  would cause an immediate kernel panic on most architectures.  We'll revert
+	  this when the following bug report has been resolved:
+	  https://github.com/llvm/llvm-project/issues/41896.
+
+config DRM_AMD_DC_FP
+	def_bool n
+	help
+	  Floating point support, required for DCN-based SoCs
+
+config DRM_AMD_DC_SI
+	bool "AMD DC support for Southern Islands ASICs"
+	depends on DRM_AMDGPU_SI
+	depends on DRM_AMD_DC
+	help
+	  Choose this option to enable new AMD DC support for SI asics
+	  by default. This includes Tahiti, Pitcairn, Cape Verde, Oland.
+	  Hainan is not supported by AMD DC and it has no physical DCE6.
+
+config DEBUG_KERNEL_DC
+	bool "Enable kgdb break in DC"
+	depends on DRM_AMD_DC
+	depends on KGDB
+	help
+	  Choose this option if you want to hit kdgb_break in assert.
+
+config DRM_AMD_SECURE_DISPLAY
+	bool "Enable secure display support"
+	depends on DEBUG_FS
+	depends on DRM_AMD_DC_FP
+	help
+	  Choose this option if you want to support secure display
+
+	  This option enables the calculation of crc of specific region via
+	  debugfs. Cooperate with specific DMCU FW.
+
+endmenu
diff --git a/drivers/gpu/drm/amd/display/Makefile b/drivers/gpu/drm/amd/display/Makefile
new file mode 100644
index 000000000000..0084a8d55254
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/Makefile
@@ -0,0 +1,66 @@
+#
+# Copyright 2017 Advanced Micro Devices, Inc.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+#
+# Makefile for the DAL (Display Abstract Layer), which is a  sub-component
+# of the AMDGPU drm driver.
+# It provides the HW control for display related functionalities.
+
+AMDDALPATH = $(RELATIVE_AMD_DISPLAY_PATH)
+
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/inc/
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/inc/hw
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/clk_mgr
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/hwss
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/resource
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dsc
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/optc
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dpp
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/hubbub
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dccg
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/hubp
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dio
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dwb
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/hpo
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/mmhubbub
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/mpc
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/opp
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/pg
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/soc_and_ip_translator
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/modules/inc
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/modules/freesync
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/modules/color
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/modules/info_packet
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/modules/power
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dmub/inc
+
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/modules/hdcp
+
+#TODO: remove when Timing Sync feature is complete
+subdir-ccflags-y += -DBUILD_FEATURE_TIMING_SYNC=0
+
+DAL_LIBS = amdgpu_dm dc	modules/freesync modules/color modules/info_packet modules/power dmub/src
+
+DAL_LIBS += modules/hdcp
+
+AMD_DAL = $(addsuffix /Makefile, $(addprefix $(FULL_AMD_DISPLAY_PATH)/,$(DAL_LIBS)))
+
+include $(AMD_DAL)
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/Makefile b/drivers/gpu/drm/amd/display/amdgpu_dm/Makefile
new file mode 100644
index 000000000000..7329b8cc2576
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/Makefile
@@ -0,0 +1,59 @@
+#
+# Copyright 2017 Advanced Micro Devices, Inc.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+#
+# Makefile for the 'dm' sub-component of DAL.
+# It provides the control and status of dm blocks.
+
+
+
+ifneq ($(CONFIG_DRM_AMD_DC),)
+AMDGPUDM = \
+	amdgpu_dm.o \
+	amdgpu_dm_plane.o \
+	amdgpu_dm_crtc.o \
+	amdgpu_dm_irq.o \
+	amdgpu_dm_mst_types.o \
+	amdgpu_dm_color.o \
+	amdgpu_dm_services.o \
+	amdgpu_dm_helpers.o \
+	amdgpu_dm_pp_smu.o \
+	amdgpu_dm_psr.o \
+	amdgpu_dm_replay.o \
+	amdgpu_dm_quirks.o \
+	amdgpu_dm_wb.o
+
+ifdef CONFIG_DRM_AMD_DC_FP
+AMDGPUDM += dc_fpu.o
+endif
+
+AMDGPUDM += amdgpu_dm_hdcp.o
+
+ifneq ($(CONFIG_DEBUG_FS),)
+AMDGPUDM += amdgpu_dm_crc.o amdgpu_dm_debugfs.o
+endif
+
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc
+
+AMDGPU_DM = $(addprefix $(AMDDALPATH)/amdgpu_dm/,$(AMDGPUDM))
+
+AMD_DISPLAY_FILES += $(AMDGPU_DM)
+endif
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
new file mode 100644
index 000000000000..0d03e324d5b9
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
@@ -0,0 +1,13123 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+/* The caprices of the preprocessor require that this be declared right here */
+#define CREATE_TRACE_POINTS
+
+#include "dm_services_types.h"
+#include "dc.h"
+#include "link_enc_cfg.h"
+#include "dc/inc/core_types.h"
+#include "dal_asic_id.h"
+#include "dmub/dmub_srv.h"
+#include "dc/inc/hw/dmcu.h"
+#include "dc/inc/hw/abm.h"
+#include "dc/dc_dmub_srv.h"
+#include "dc/dc_edid_parser.h"
+#include "dc/dc_stat.h"
+#include "dc/dc_state.h"
+#include "amdgpu_dm_trace.h"
+#include "link/protocols/link_dpcd.h"
+#include "link_service_types.h"
+#include "link/protocols/link_dp_capability.h"
+#include "link/protocols/link_ddc.h"
+
+#include "amdgpu.h"
+#include "amdgpu_display.h"
+#include "amdgpu_ucode.h"
+#include "atom.h"
+#include "amdgpu_dm.h"
+#include "amdgpu_dm_plane.h"
+#include "amdgpu_dm_crtc.h"
+#include "amdgpu_dm_hdcp.h"
+#include <drm/display/drm_hdcp_helper.h>
+#include "amdgpu_dm_wb.h"
+#include "amdgpu_atombios.h"
+
+#include "amd_shared.h"
+#include "amdgpu_dm_irq.h"
+#include "dm_helpers.h"
+#include "amdgpu_dm_mst_types.h"
+#if defined(CONFIG_DEBUG_FS)
+#include "amdgpu_dm_debugfs.h"
+#endif
+#include "amdgpu_dm_psr.h"
+#include "amdgpu_dm_replay.h"
+
+#include "ivsrcid/ivsrcid_vislands30.h"
+
+#include <linux/backlight.h>
+#include <linux/module.h>
+#include <linux/moduleparam.h>
+#include <linux/types.h>
+#include <linux/pm_runtime.h>
+#include <linux/pci.h>
+#include <linux/power_supply.h>
+#include <linux/firmware.h>
+#include <linux/component.h>
+#include <linux/sort.h>
+
+#include <drm/drm_privacy_screen_consumer.h>
+#include <drm/display/drm_dp_mst_helper.h>
+#include <drm/display/drm_hdmi_helper.h>
+#include <drm/drm_atomic.h>
+#include <drm/drm_atomic_uapi.h>
+#include <drm/drm_atomic_helper.h>
+#include <drm/drm_blend.h>
+#include <drm/drm_fixed.h>
+#include <drm/drm_fourcc.h>
+#include <drm/drm_edid.h>
+#include <drm/drm_eld.h>
+#include <drm/drm_utils.h>
+#include <drm/drm_vblank.h>
+#include <drm/drm_audio_component.h>
+#include <drm/drm_gem_atomic_helper.h>
+
+#include <media/cec-notifier.h>
+#include <acpi/video.h>
+
+#include "ivsrcid/dcn/irqsrcs_dcn_1_0.h"
+
+#include "modules/inc/mod_freesync.h"
+#include "modules/power/power_helpers.h"
+
+static_assert(AMDGPU_DMUB_NOTIFICATION_MAX == DMUB_NOTIFICATION_MAX, "AMDGPU_DMUB_NOTIFICATION_MAX mismatch");
+
+#define FIRMWARE_RENOIR_DMUB "amdgpu/renoir_dmcub.bin"
+MODULE_FIRMWARE(FIRMWARE_RENOIR_DMUB);
+#define FIRMWARE_SIENNA_CICHLID_DMUB "amdgpu/sienna_cichlid_dmcub.bin"
+MODULE_FIRMWARE(FIRMWARE_SIENNA_CICHLID_DMUB);
+#define FIRMWARE_NAVY_FLOUNDER_DMUB "amdgpu/navy_flounder_dmcub.bin"
+MODULE_FIRMWARE(FIRMWARE_NAVY_FLOUNDER_DMUB);
+#define FIRMWARE_GREEN_SARDINE_DMUB "amdgpu/green_sardine_dmcub.bin"
+MODULE_FIRMWARE(FIRMWARE_GREEN_SARDINE_DMUB);
+#define FIRMWARE_VANGOGH_DMUB "amdgpu/vangogh_dmcub.bin"
+MODULE_FIRMWARE(FIRMWARE_VANGOGH_DMUB);
+#define FIRMWARE_DIMGREY_CAVEFISH_DMUB "amdgpu/dimgrey_cavefish_dmcub.bin"
+MODULE_FIRMWARE(FIRMWARE_DIMGREY_CAVEFISH_DMUB);
+#define FIRMWARE_BEIGE_GOBY_DMUB "amdgpu/beige_goby_dmcub.bin"
+MODULE_FIRMWARE(FIRMWARE_BEIGE_GOBY_DMUB);
+#define FIRMWARE_YELLOW_CARP_DMUB "amdgpu/yellow_carp_dmcub.bin"
+MODULE_FIRMWARE(FIRMWARE_YELLOW_CARP_DMUB);
+#define FIRMWARE_DCN_314_DMUB "amdgpu/dcn_3_1_4_dmcub.bin"
+MODULE_FIRMWARE(FIRMWARE_DCN_314_DMUB);
+#define FIRMWARE_DCN_315_DMUB "amdgpu/dcn_3_1_5_dmcub.bin"
+MODULE_FIRMWARE(FIRMWARE_DCN_315_DMUB);
+#define FIRMWARE_DCN316_DMUB "amdgpu/dcn_3_1_6_dmcub.bin"
+MODULE_FIRMWARE(FIRMWARE_DCN316_DMUB);
+
+#define FIRMWARE_DCN_V3_2_0_DMCUB "amdgpu/dcn_3_2_0_dmcub.bin"
+MODULE_FIRMWARE(FIRMWARE_DCN_V3_2_0_DMCUB);
+#define FIRMWARE_DCN_V3_2_1_DMCUB "amdgpu/dcn_3_2_1_dmcub.bin"
+MODULE_FIRMWARE(FIRMWARE_DCN_V3_2_1_DMCUB);
+
+#define FIRMWARE_RAVEN_DMCU		"amdgpu/raven_dmcu.bin"
+MODULE_FIRMWARE(FIRMWARE_RAVEN_DMCU);
+
+#define FIRMWARE_NAVI12_DMCU            "amdgpu/navi12_dmcu.bin"
+MODULE_FIRMWARE(FIRMWARE_NAVI12_DMCU);
+
+#define FIRMWARE_DCN_35_DMUB "amdgpu/dcn_3_5_dmcub.bin"
+MODULE_FIRMWARE(FIRMWARE_DCN_35_DMUB);
+
+#define FIRMWARE_DCN_351_DMUB "amdgpu/dcn_3_5_1_dmcub.bin"
+MODULE_FIRMWARE(FIRMWARE_DCN_351_DMUB);
+
+#define FIRMWARE_DCN_36_DMUB "amdgpu/dcn_3_6_dmcub.bin"
+MODULE_FIRMWARE(FIRMWARE_DCN_36_DMUB);
+
+#define FIRMWARE_DCN_401_DMUB "amdgpu/dcn_4_0_1_dmcub.bin"
+MODULE_FIRMWARE(FIRMWARE_DCN_401_DMUB);
+
+/* Number of bytes in PSP header for firmware. */
+#define PSP_HEADER_BYTES 0x100
+
+/* Number of bytes in PSP footer for firmware. */
+#define PSP_FOOTER_BYTES 0x100
+
+/**
+ * DOC: overview
+ *
+ * The AMDgpu display manager, **amdgpu_dm** (or even simpler,
+ * **dm**) sits between DRM and DC. It acts as a liaison, converting DRM
+ * requests into DC requests, and DC responses into DRM responses.
+ *
+ * The root control structure is &struct amdgpu_display_manager.
+ */
+
+/* basic init/fini API */
+static int amdgpu_dm_init(struct amdgpu_device *adev);
+static void amdgpu_dm_fini(struct amdgpu_device *adev);
+static bool is_freesync_video_mode(const struct drm_display_mode *mode, struct amdgpu_dm_connector *aconnector);
+static void reset_freesync_config_for_crtc(struct dm_crtc_state *new_crtc_state);
+static struct amdgpu_i2c_adapter *
+create_i2c(struct ddc_service *ddc_service, bool oem);
+
+static enum drm_mode_subconnector get_subconnector_type(struct dc_link *link)
+{
+	switch (link->dpcd_caps.dongle_type) {
+	case DISPLAY_DONGLE_NONE:
+		return DRM_MODE_SUBCONNECTOR_Native;
+	case DISPLAY_DONGLE_DP_VGA_CONVERTER:
+		return DRM_MODE_SUBCONNECTOR_VGA;
+	case DISPLAY_DONGLE_DP_DVI_CONVERTER:
+	case DISPLAY_DONGLE_DP_DVI_DONGLE:
+		return DRM_MODE_SUBCONNECTOR_DVID;
+	case DISPLAY_DONGLE_DP_HDMI_CONVERTER:
+	case DISPLAY_DONGLE_DP_HDMI_DONGLE:
+		return DRM_MODE_SUBCONNECTOR_HDMIA;
+	case DISPLAY_DONGLE_DP_HDMI_MISMATCHED_DONGLE:
+	default:
+		return DRM_MODE_SUBCONNECTOR_Unknown;
+	}
+}
+
+static void update_subconnector_property(struct amdgpu_dm_connector *aconnector)
+{
+	struct dc_link *link = aconnector->dc_link;
+	struct drm_connector *connector = &aconnector->base;
+	enum drm_mode_subconnector subconnector = DRM_MODE_SUBCONNECTOR_Unknown;
+
+	if (connector->connector_type != DRM_MODE_CONNECTOR_DisplayPort)
+		return;
+
+	if (aconnector->dc_sink)
+		subconnector = get_subconnector_type(link);
+
+	drm_object_property_set_value(&connector->base,
+			connector->dev->mode_config.dp_subconnector_property,
+			subconnector);
+}
+
+/*
+ * initializes drm_device display related structures, based on the information
+ * provided by DAL. The drm strcutures are: drm_crtc, drm_connector,
+ * drm_encoder, drm_mode_config
+ *
+ * Returns 0 on success
+ */
+static int amdgpu_dm_initialize_drm_device(struct amdgpu_device *adev);
+/* removes and deallocates the drm structures, created by the above function */
+static void amdgpu_dm_destroy_drm_device(struct amdgpu_display_manager *dm);
+
+static int amdgpu_dm_connector_init(struct amdgpu_display_manager *dm,
+				    struct amdgpu_dm_connector *amdgpu_dm_connector,
+				    u32 link_index,
+				    struct amdgpu_encoder *amdgpu_encoder);
+static int amdgpu_dm_encoder_init(struct drm_device *dev,
+				  struct amdgpu_encoder *aencoder,
+				  uint32_t link_index);
+
+static int amdgpu_dm_connector_get_modes(struct drm_connector *connector);
+
+static int amdgpu_dm_atomic_setup_commit(struct drm_atomic_state *state);
+static void amdgpu_dm_atomic_commit_tail(struct drm_atomic_state *state);
+
+static int amdgpu_dm_atomic_check(struct drm_device *dev,
+				  struct drm_atomic_state *state);
+
+static void handle_hpd_irq_helper(struct amdgpu_dm_connector *aconnector);
+static void handle_hpd_rx_irq(void *param);
+
+static void amdgpu_dm_backlight_set_level(struct amdgpu_display_manager *dm,
+					 int bl_idx,
+					 u32 user_brightness);
+
+static bool
+is_timing_unchanged_for_freesync(struct drm_crtc_state *old_crtc_state,
+				 struct drm_crtc_state *new_crtc_state);
+/*
+ * dm_vblank_get_counter
+ *
+ * @brief
+ * Get counter for number of vertical blanks
+ *
+ * @param
+ * struct amdgpu_device *adev - [in] desired amdgpu device
+ * int disp_idx - [in] which CRTC to get the counter from
+ *
+ * @return
+ * Counter for vertical blanks
+ */
+static u32 dm_vblank_get_counter(struct amdgpu_device *adev, int crtc)
+{
+	struct amdgpu_crtc *acrtc = NULL;
+
+	if (crtc >= adev->mode_info.num_crtc)
+		return 0;
+
+	acrtc = adev->mode_info.crtcs[crtc];
+
+	if (!acrtc->dm_irq_params.stream) {
+		drm_err(adev_to_drm(adev), "dc_stream_state is NULL for crtc '%d'!\n",
+			  crtc);
+		return 0;
+	}
+
+	return dc_stream_get_vblank_counter(acrtc->dm_irq_params.stream);
+}
+
+static int dm_crtc_get_scanoutpos(struct amdgpu_device *adev, int crtc,
+				  u32 *vbl, u32 *position)
+{
+	u32 v_blank_start = 0, v_blank_end = 0, h_position = 0, v_position = 0;
+	struct amdgpu_crtc *acrtc = NULL;
+	struct dc *dc = adev->dm.dc;
+
+	if ((crtc < 0) || (crtc >= adev->mode_info.num_crtc))
+		return -EINVAL;
+
+	acrtc = adev->mode_info.crtcs[crtc];
+
+	if (!acrtc->dm_irq_params.stream) {
+		drm_err(adev_to_drm(adev), "dc_stream_state is NULL for crtc '%d'!\n",
+			  crtc);
+		return 0;
+	}
+
+	if (dc && dc->caps.ips_support && dc->idle_optimizations_allowed)
+		dc_allow_idle_optimizations(dc, false);
+
+	/*
+	 * TODO rework base driver to use values directly.
+	 * for now parse it back into reg-format
+	 */
+	dc_stream_get_scanoutpos(acrtc->dm_irq_params.stream,
+				 &v_blank_start,
+				 &v_blank_end,
+				 &h_position,
+				 &v_position);
+
+	*position = v_position | (h_position << 16);
+	*vbl = v_blank_start | (v_blank_end << 16);
+
+	return 0;
+}
+
+static bool dm_is_idle(struct amdgpu_ip_block *ip_block)
+{
+	/* XXX todo */
+	return true;
+}
+
+static int dm_wait_for_idle(struct amdgpu_ip_block *ip_block)
+{
+	/* XXX todo */
+	return 0;
+}
+
+static bool dm_check_soft_reset(struct amdgpu_ip_block *ip_block)
+{
+	return false;
+}
+
+static int dm_soft_reset(struct amdgpu_ip_block *ip_block)
+{
+	/* XXX todo */
+	return 0;
+}
+
+static struct amdgpu_crtc *
+get_crtc_by_otg_inst(struct amdgpu_device *adev,
+		     int otg_inst)
+{
+	struct drm_device *dev = adev_to_drm(adev);
+	struct drm_crtc *crtc;
+	struct amdgpu_crtc *amdgpu_crtc;
+
+	if (WARN_ON(otg_inst == -1))
+		return adev->mode_info.crtcs[0];
+
+	list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
+		amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+		if (amdgpu_crtc->otg_inst == otg_inst)
+			return amdgpu_crtc;
+	}
+
+	return NULL;
+}
+
+static inline bool is_dc_timing_adjust_needed(struct dm_crtc_state *old_state,
+					      struct dm_crtc_state *new_state)
+{
+	if (new_state->stream->adjust.timing_adjust_pending)
+		return true;
+	if (new_state->freesync_config.state ==  VRR_STATE_ACTIVE_FIXED)
+		return true;
+	else if (amdgpu_dm_crtc_vrr_active(old_state) != amdgpu_dm_crtc_vrr_active(new_state))
+		return true;
+	else
+		return false;
+}
+
+/*
+ * DC will program planes with their z-order determined by their ordering
+ * in the dc_surface_updates array. This comparator is used to sort them
+ * by descending zpos.
+ */
+static int dm_plane_layer_index_cmp(const void *a, const void *b)
+{
+	const struct dc_surface_update *sa = (struct dc_surface_update *)a;
+	const struct dc_surface_update *sb = (struct dc_surface_update *)b;
+
+	/* Sort by descending dc_plane layer_index (i.e. normalized_zpos) */
+	return sb->surface->layer_index - sa->surface->layer_index;
+}
+
+/**
+ * update_planes_and_stream_adapter() - Send planes to be updated in DC
+ *
+ * DC has a generic way to update planes and stream via
+ * dc_update_planes_and_stream function; however, DM might need some
+ * adjustments and preparation before calling it. This function is a wrapper
+ * for the dc_update_planes_and_stream that does any required configuration
+ * before passing control to DC.
+ *
+ * @dc: Display Core control structure
+ * @update_type: specify whether it is FULL/MEDIUM/FAST update
+ * @planes_count: planes count to update
+ * @stream: stream state
+ * @stream_update: stream update
+ * @array_of_surface_update: dc surface update pointer
+ *
+ */
+static inline bool update_planes_and_stream_adapter(struct dc *dc,
+						    int update_type,
+						    int planes_count,
+						    struct dc_stream_state *stream,
+						    struct dc_stream_update *stream_update,
+						    struct dc_surface_update *array_of_surface_update)
+{
+	sort(array_of_surface_update, planes_count,
+	     sizeof(*array_of_surface_update), dm_plane_layer_index_cmp, NULL);
+
+	/*
+	 * Previous frame finished and HW is ready for optimization.
+	 */
+	dc_post_update_surfaces_to_stream(dc);
+
+	return dc_update_planes_and_stream(dc,
+					   array_of_surface_update,
+					   planes_count,
+					   stream,
+					   stream_update);
+}
+
+/**
+ * dm_pflip_high_irq() - Handle pageflip interrupt
+ * @interrupt_params: ignored
+ *
+ * Handles the pageflip interrupt by notifying all interested parties
+ * that the pageflip has been completed.
+ */
+static void dm_pflip_high_irq(void *interrupt_params)
+{
+	struct amdgpu_crtc *amdgpu_crtc;
+	struct common_irq_params *irq_params = interrupt_params;
+	struct amdgpu_device *adev = irq_params->adev;
+	struct drm_device *dev = adev_to_drm(adev);
+	unsigned long flags;
+	struct drm_pending_vblank_event *e;
+	u32 vpos, hpos, v_blank_start, v_blank_end;
+	bool vrr_active;
+
+	amdgpu_crtc = get_crtc_by_otg_inst(adev, irq_params->irq_src - IRQ_TYPE_PFLIP);
+
+	/* IRQ could occur when in initial stage */
+	/* TODO work and BO cleanup */
+	if (amdgpu_crtc == NULL) {
+		drm_dbg_state(dev, "CRTC is null, returning.\n");
+		return;
+	}
+
+	spin_lock_irqsave(&adev_to_drm(adev)->event_lock, flags);
+
+	if (amdgpu_crtc->pflip_status != AMDGPU_FLIP_SUBMITTED) {
+		drm_dbg_state(dev,
+			      "amdgpu_crtc->pflip_status = %d != AMDGPU_FLIP_SUBMITTED(%d) on crtc:%d[%p]\n",
+			      amdgpu_crtc->pflip_status, AMDGPU_FLIP_SUBMITTED,
+			      amdgpu_crtc->crtc_id, amdgpu_crtc);
+		spin_unlock_irqrestore(&adev_to_drm(adev)->event_lock, flags);
+		return;
+	}
+
+	/* page flip completed. */
+	e = amdgpu_crtc->event;
+	amdgpu_crtc->event = NULL;
+
+	WARN_ON(!e);
+
+	vrr_active = amdgpu_dm_crtc_vrr_active_irq(amdgpu_crtc);
+
+	/* Fixed refresh rate, or VRR scanout position outside front-porch? */
+	if (!vrr_active ||
+	    !dc_stream_get_scanoutpos(amdgpu_crtc->dm_irq_params.stream, &v_blank_start,
+				      &v_blank_end, &hpos, &vpos) ||
+	    (vpos < v_blank_start)) {
+		/* Update to correct count and vblank timestamp if racing with
+		 * vblank irq. This also updates to the correct vblank timestamp
+		 * even in VRR mode, as scanout is past the front-porch atm.
+		 */
+		drm_crtc_accurate_vblank_count(&amdgpu_crtc->base);
+
+		/* Wake up userspace by sending the pageflip event with proper
+		 * count and timestamp of vblank of flip completion.
+		 */
+		if (e) {
+			drm_crtc_send_vblank_event(&amdgpu_crtc->base, e);
+
+			/* Event sent, so done with vblank for this flip */
+			drm_crtc_vblank_put(&amdgpu_crtc->base);
+		}
+	} else if (e) {
+		/* VRR active and inside front-porch: vblank count and
+		 * timestamp for pageflip event will only be up to date after
+		 * drm_crtc_handle_vblank() has been executed from late vblank
+		 * irq handler after start of back-porch (vline 0). We queue the
+		 * pageflip event for send-out by drm_crtc_handle_vblank() with
+		 * updated timestamp and count, once it runs after us.
+		 *
+		 * We need to open-code this instead of using the helper
+		 * drm_crtc_arm_vblank_event(), as that helper would
+		 * call drm_crtc_accurate_vblank_count(), which we must
+		 * not call in VRR mode while we are in front-porch!
+		 */
+
+		/* sequence will be replaced by real count during send-out. */
+		e->sequence = drm_crtc_vblank_count(&amdgpu_crtc->base);
+		e->pipe = amdgpu_crtc->crtc_id;
+
+		list_add_tail(&e->base.link, &adev_to_drm(adev)->vblank_event_list);
+		e = NULL;
+	}
+
+	/* Keep track of vblank of this flip for flip throttling. We use the
+	 * cooked hw counter, as that one incremented at start of this vblank
+	 * of pageflip completion, so last_flip_vblank is the forbidden count
+	 * for queueing new pageflips if vsync + VRR is enabled.
+	 */
+	amdgpu_crtc->dm_irq_params.last_flip_vblank =
+		amdgpu_get_vblank_counter_kms(&amdgpu_crtc->base);
+
+	amdgpu_crtc->pflip_status = AMDGPU_FLIP_NONE;
+	spin_unlock_irqrestore(&adev_to_drm(adev)->event_lock, flags);
+
+	drm_dbg_state(dev,
+		      "crtc:%d[%p], pflip_stat:AMDGPU_FLIP_NONE, vrr[%d]-fp %d\n",
+		      amdgpu_crtc->crtc_id, amdgpu_crtc, vrr_active, (int)!e);
+}
+
+static void dm_handle_vmin_vmax_update(struct work_struct *offload_work)
+{
+	struct vupdate_offload_work *work = container_of(offload_work, struct vupdate_offload_work, work);
+	struct amdgpu_device *adev = work->adev;
+	struct dc_stream_state *stream = work->stream;
+	struct dc_crtc_timing_adjust *adjust = work->adjust;
+
+	mutex_lock(&adev->dm.dc_lock);
+	dc_stream_adjust_vmin_vmax(adev->dm.dc, stream, adjust);
+	mutex_unlock(&adev->dm.dc_lock);
+
+	dc_stream_release(stream);
+	kfree(work->adjust);
+	kfree(work);
+}
+
+static void schedule_dc_vmin_vmax(struct amdgpu_device *adev,
+	struct dc_stream_state *stream,
+	struct dc_crtc_timing_adjust *adjust)
+{
+	struct vupdate_offload_work *offload_work = kzalloc(sizeof(*offload_work), GFP_KERNEL);
+	if (!offload_work) {
+		drm_dbg_driver(adev_to_drm(adev), "Failed to allocate vupdate_offload_work\n");
+		return;
+	}
+
+	struct dc_crtc_timing_adjust *adjust_copy = kzalloc(sizeof(*adjust_copy), GFP_KERNEL);
+	if (!adjust_copy) {
+		drm_dbg_driver(adev_to_drm(adev), "Failed to allocate adjust_copy\n");
+		kfree(offload_work);
+		return;
+	}
+
+	dc_stream_retain(stream);
+	memcpy(adjust_copy, adjust, sizeof(*adjust_copy));
+
+	INIT_WORK(&offload_work->work, dm_handle_vmin_vmax_update);
+	offload_work->adev = adev;
+	offload_work->stream = stream;
+	offload_work->adjust = adjust_copy;
+
+	queue_work(system_wq, &offload_work->work);
+}
+
+static void dm_vupdate_high_irq(void *interrupt_params)
+{
+	struct common_irq_params *irq_params = interrupt_params;
+	struct amdgpu_device *adev = irq_params->adev;
+	struct amdgpu_crtc *acrtc;
+	struct drm_device *drm_dev;
+	struct drm_vblank_crtc *vblank;
+	ktime_t frame_duration_ns, previous_timestamp;
+	unsigned long flags;
+	int vrr_active;
+
+	acrtc = get_crtc_by_otg_inst(adev, irq_params->irq_src - IRQ_TYPE_VUPDATE);
+
+	if (acrtc) {
+		vrr_active = amdgpu_dm_crtc_vrr_active_irq(acrtc);
+		drm_dev = acrtc->base.dev;
+		vblank = drm_crtc_vblank_crtc(&acrtc->base);
+		previous_timestamp = atomic64_read(&irq_params->previous_timestamp);
+		frame_duration_ns = vblank->time - previous_timestamp;
+
+		if (frame_duration_ns > 0) {
+			trace_amdgpu_refresh_rate_track(acrtc->base.index,
+						frame_duration_ns,
+						ktime_divns(NSEC_PER_SEC, frame_duration_ns));
+			atomic64_set(&irq_params->previous_timestamp, vblank->time);
+		}
+
+		drm_dbg_vbl(drm_dev,
+			    "crtc:%d, vupdate-vrr:%d\n", acrtc->crtc_id,
+			    vrr_active);
+
+		/* Core vblank handling is done here after end of front-porch in
+		 * vrr mode, as vblank timestamping will give valid results
+		 * while now done after front-porch. This will also deliver
+		 * page-flip completion events that have been queued to us
+		 * if a pageflip happened inside front-porch.
+		 */
+		if (vrr_active && acrtc->dm_irq_params.stream) {
+			bool replay_en = acrtc->dm_irq_params.stream->link->replay_settings.replay_feature_enabled;
+			bool psr_en = acrtc->dm_irq_params.stream->link->psr_settings.psr_feature_enabled;
+			bool fs_active_var_en = acrtc->dm_irq_params.freesync_config.state
+				== VRR_STATE_ACTIVE_VARIABLE;
+
+			amdgpu_dm_crtc_handle_vblank(acrtc);
+
+			/* BTR processing for pre-DCE12 ASICs */
+			if (adev->family < AMDGPU_FAMILY_AI) {
+				spin_lock_irqsave(&adev_to_drm(adev)->event_lock, flags);
+				mod_freesync_handle_v_update(
+				    adev->dm.freesync_module,
+				    acrtc->dm_irq_params.stream,
+				    &acrtc->dm_irq_params.vrr_params);
+
+				if (fs_active_var_en || (!fs_active_var_en && !replay_en && !psr_en)) {
+					schedule_dc_vmin_vmax(adev,
+						acrtc->dm_irq_params.stream,
+						&acrtc->dm_irq_params.vrr_params.adjust);
+				}
+				spin_unlock_irqrestore(&adev_to_drm(adev)->event_lock, flags);
+			}
+		}
+	}
+}
+
+/**
+ * dm_crtc_high_irq() - Handles CRTC interrupt
+ * @interrupt_params: used for determining the CRTC instance
+ *
+ * Handles the CRTC/VSYNC interrupt by notfying DRM's VBLANK
+ * event handler.
+ */
+static void dm_crtc_high_irq(void *interrupt_params)
+{
+	struct common_irq_params *irq_params = interrupt_params;
+	struct amdgpu_device *adev = irq_params->adev;
+	struct drm_writeback_job *job;
+	struct amdgpu_crtc *acrtc;
+	unsigned long flags;
+	int vrr_active;
+
+	acrtc = get_crtc_by_otg_inst(adev, irq_params->irq_src - IRQ_TYPE_VBLANK);
+	if (!acrtc)
+		return;
+
+	if (acrtc->wb_conn) {
+		spin_lock_irqsave(&acrtc->wb_conn->job_lock, flags);
+
+		if (acrtc->wb_pending) {
+			job = list_first_entry_or_null(&acrtc->wb_conn->job_queue,
+						       struct drm_writeback_job,
+						       list_entry);
+			acrtc->wb_pending = false;
+			spin_unlock_irqrestore(&acrtc->wb_conn->job_lock, flags);
+
+			if (job) {
+				unsigned int v_total, refresh_hz;
+				struct dc_stream_state *stream = acrtc->dm_irq_params.stream;
+
+				v_total = stream->adjust.v_total_max ?
+					  stream->adjust.v_total_max : stream->timing.v_total;
+				refresh_hz = div_u64((uint64_t) stream->timing.pix_clk_100hz *
+					     100LL, (v_total * stream->timing.h_total));
+				mdelay(1000 / refresh_hz);
+
+				drm_writeback_signal_completion(acrtc->wb_conn, 0);
+				dc_stream_fc_disable_writeback(adev->dm.dc,
+							       acrtc->dm_irq_params.stream, 0);
+			}
+		} else
+			spin_unlock_irqrestore(&acrtc->wb_conn->job_lock, flags);
+	}
+
+	vrr_active = amdgpu_dm_crtc_vrr_active_irq(acrtc);
+
+	drm_dbg_vbl(adev_to_drm(adev),
+		    "crtc:%d, vupdate-vrr:%d, planes:%d\n", acrtc->crtc_id,
+		    vrr_active, acrtc->dm_irq_params.active_planes);
+
+	/**
+	 * Core vblank handling at start of front-porch is only possible
+	 * in non-vrr mode, as only there vblank timestamping will give
+	 * valid results while done in front-porch. Otherwise defer it
+	 * to dm_vupdate_high_irq after end of front-porch.
+	 */
+	if (!vrr_active)
+		amdgpu_dm_crtc_handle_vblank(acrtc);
+
+	/**
+	 * Following stuff must happen at start of vblank, for crc
+	 * computation and below-the-range btr support in vrr mode.
+	 */
+	amdgpu_dm_crtc_handle_crc_irq(&acrtc->base);
+
+	/* BTR updates need to happen before VUPDATE on Vega and above. */
+	if (adev->family < AMDGPU_FAMILY_AI)
+		return;
+
+	spin_lock_irqsave(&adev_to_drm(adev)->event_lock, flags);
+
+	if (acrtc->dm_irq_params.stream &&
+		acrtc->dm_irq_params.vrr_params.supported) {
+		bool replay_en = acrtc->dm_irq_params.stream->link->replay_settings.replay_feature_enabled;
+		bool psr_en = acrtc->dm_irq_params.stream->link->psr_settings.psr_feature_enabled;
+		bool fs_active_var_en = acrtc->dm_irq_params.freesync_config.state == VRR_STATE_ACTIVE_VARIABLE;
+
+		mod_freesync_handle_v_update(adev->dm.freesync_module,
+					     acrtc->dm_irq_params.stream,
+					     &acrtc->dm_irq_params.vrr_params);
+
+		/* update vmin_vmax only if freesync is enabled, or only if PSR and REPLAY are disabled */
+		if (fs_active_var_en || (!fs_active_var_en && !replay_en && !psr_en)) {
+			schedule_dc_vmin_vmax(adev, acrtc->dm_irq_params.stream,
+					&acrtc->dm_irq_params.vrr_params.adjust);
+		}
+	}
+
+	/*
+	 * If there aren't any active_planes then DCH HUBP may be clock-gated.
+	 * In that case, pageflip completion interrupts won't fire and pageflip
+	 * completion events won't get delivered. Prevent this by sending
+	 * pending pageflip events from here if a flip is still pending.
+	 *
+	 * If any planes are enabled, use dm_pflip_high_irq() instead, to
+	 * avoid race conditions between flip programming and completion,
+	 * which could cause too early flip completion events.
+	 */
+	if (adev->family >= AMDGPU_FAMILY_RV &&
+	    acrtc->pflip_status == AMDGPU_FLIP_SUBMITTED &&
+	    acrtc->dm_irq_params.active_planes == 0) {
+		if (acrtc->event) {
+			drm_crtc_send_vblank_event(&acrtc->base, acrtc->event);
+			acrtc->event = NULL;
+			drm_crtc_vblank_put(&acrtc->base);
+		}
+		acrtc->pflip_status = AMDGPU_FLIP_NONE;
+	}
+
+	spin_unlock_irqrestore(&adev_to_drm(adev)->event_lock, flags);
+}
+
+#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY)
+/**
+ * dm_dcn_vertical_interrupt0_high_irq() - Handles OTG Vertical interrupt0 for
+ * DCN generation ASICs
+ * @interrupt_params: interrupt parameters
+ *
+ * Used to set crc window/read out crc value at vertical line 0 position
+ */
+static void dm_dcn_vertical_interrupt0_high_irq(void *interrupt_params)
+{
+	struct common_irq_params *irq_params = interrupt_params;
+	struct amdgpu_device *adev = irq_params->adev;
+	struct amdgpu_crtc *acrtc;
+
+	acrtc = get_crtc_by_otg_inst(adev, irq_params->irq_src - IRQ_TYPE_VLINE0);
+
+	if (!acrtc)
+		return;
+
+	amdgpu_dm_crtc_handle_crc_window_irq(&acrtc->base);
+}
+#endif /* CONFIG_DRM_AMD_SECURE_DISPLAY */
+
+/**
+ * dmub_aux_setconfig_callback - Callback for AUX or SET_CONFIG command.
+ * @adev: amdgpu_device pointer
+ * @notify: dmub notification structure
+ *
+ * Dmub AUX or SET_CONFIG command completion processing callback
+ * Copies dmub notification to DM which is to be read by AUX command.
+ * issuing thread and also signals the event to wake up the thread.
+ */
+static void dmub_aux_setconfig_callback(struct amdgpu_device *adev,
+					struct dmub_notification *notify)
+{
+	if (adev->dm.dmub_notify)
+		memcpy(adev->dm.dmub_notify, notify, sizeof(struct dmub_notification));
+	if (notify->type == DMUB_NOTIFICATION_AUX_REPLY)
+		complete(&adev->dm.dmub_aux_transfer_done);
+}
+
+static void dmub_aux_fused_io_callback(struct amdgpu_device *adev,
+					struct dmub_notification *notify)
+{
+	if (!adev || !notify) {
+		ASSERT(false);
+		return;
+	}
+
+	const struct dmub_cmd_fused_request *req = &notify->fused_request;
+	const uint8_t ddc_line = req->u.aux.ddc_line;
+
+	if (ddc_line >= ARRAY_SIZE(adev->dm.fused_io)) {
+		ASSERT(false);
+		return;
+	}
+
+	struct fused_io_sync *sync = &adev->dm.fused_io[ddc_line];
+
+	static_assert(sizeof(*req) <= sizeof(sync->reply_data), "Size mismatch");
+	memcpy(sync->reply_data, req, sizeof(*req));
+	complete(&sync->replied);
+}
+
+/**
+ * dmub_hpd_callback - DMUB HPD interrupt processing callback.
+ * @adev: amdgpu_device pointer
+ * @notify: dmub notification structure
+ *
+ * Dmub Hpd interrupt processing callback. Gets displayindex through the
+ * ink index and calls helper to do the processing.
+ */
+static void dmub_hpd_callback(struct amdgpu_device *adev,
+			      struct dmub_notification *notify)
+{
+	struct amdgpu_dm_connector *aconnector;
+	struct amdgpu_dm_connector *hpd_aconnector = NULL;
+	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
+	struct dc_link *link;
+	u8 link_index = 0;
+	struct drm_device *dev;
+
+	if (adev == NULL)
+		return;
+
+	if (notify == NULL) {
+		drm_err(adev_to_drm(adev), "DMUB HPD callback notification was NULL");
+		return;
+	}
+
+	if (notify->link_index > adev->dm.dc->link_count) {
+		drm_err(adev_to_drm(adev), "DMUB HPD index (%u)is abnormal", notify->link_index);
+		return;
+	}
+
+	/* Skip DMUB HPD IRQ in suspend/resume. We will probe them later. */
+	if (notify->type == DMUB_NOTIFICATION_HPD && adev->in_suspend) {
+		drm_info(adev_to_drm(adev), "Skip DMUB HPD IRQ callback in suspend/resume\n");
+		return;
+	}
+
+	link_index = notify->link_index;
+	link = adev->dm.dc->links[link_index];
+	dev = adev->dm.ddev;
+
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_WRITEBACK)
+			continue;
+
+		aconnector = to_amdgpu_dm_connector(connector);
+		if (link && aconnector->dc_link == link) {
+			if (notify->type == DMUB_NOTIFICATION_HPD)
+				drm_info(adev_to_drm(adev), "DMUB HPD IRQ callback: link_index=%u\n", link_index);
+			else if (notify->type == DMUB_NOTIFICATION_HPD_IRQ)
+				drm_info(adev_to_drm(adev), "DMUB HPD RX IRQ callback: link_index=%u\n", link_index);
+			else
+				drm_warn(adev_to_drm(adev), "DMUB Unknown HPD callback type %d, link_index=%u\n",
+						notify->type, link_index);
+
+			hpd_aconnector = aconnector;
+			break;
+		}
+	}
+	drm_connector_list_iter_end(&iter);
+
+	if (hpd_aconnector) {
+		if (notify->type == DMUB_NOTIFICATION_HPD) {
+			if (hpd_aconnector->dc_link->hpd_status == (notify->hpd_status == DP_HPD_PLUG))
+				drm_warn(adev_to_drm(adev), "DMUB reported hpd status unchanged. link_index=%u\n", link_index);
+			handle_hpd_irq_helper(hpd_aconnector);
+		} else if (notify->type == DMUB_NOTIFICATION_HPD_IRQ) {
+			handle_hpd_rx_irq(hpd_aconnector);
+		}
+	}
+}
+
+/**
+ * dmub_hpd_sense_callback - DMUB HPD sense processing callback.
+ * @adev: amdgpu_device pointer
+ * @notify: dmub notification structure
+ *
+ * HPD sense changes can occur during low power states and need to be
+ * notified from firmware to driver.
+ */
+static void dmub_hpd_sense_callback(struct amdgpu_device *adev,
+			      struct dmub_notification *notify)
+{
+	drm_dbg_driver(adev_to_drm(adev), "DMUB HPD SENSE callback.\n");
+}
+
+/**
+ * register_dmub_notify_callback - Sets callback for DMUB notify
+ * @adev: amdgpu_device pointer
+ * @type: Type of dmub notification
+ * @callback: Dmub interrupt callback function
+ * @dmub_int_thread_offload: offload indicator
+ *
+ * API to register a dmub callback handler for a dmub notification
+ * Also sets indicator whether callback processing to be offloaded.
+ * to dmub interrupt handling thread
+ * Return: true if successfully registered, false if there is existing registration
+ */
+static bool register_dmub_notify_callback(struct amdgpu_device *adev,
+					  enum dmub_notification_type type,
+					  dmub_notify_interrupt_callback_t callback,
+					  bool dmub_int_thread_offload)
+{
+	if (callback != NULL && type < ARRAY_SIZE(adev->dm.dmub_thread_offload)) {
+		adev->dm.dmub_callback[type] = callback;
+		adev->dm.dmub_thread_offload[type] = dmub_int_thread_offload;
+	} else
+		return false;
+
+	return true;
+}
+
+static void dm_handle_hpd_work(struct work_struct *work)
+{
+	struct dmub_hpd_work *dmub_hpd_wrk;
+
+	dmub_hpd_wrk = container_of(work, struct dmub_hpd_work, handle_hpd_work);
+
+	if (!dmub_hpd_wrk->dmub_notify) {
+		drm_err(adev_to_drm(dmub_hpd_wrk->adev), "dmub_hpd_wrk dmub_notify is NULL");
+		return;
+	}
+
+	if (dmub_hpd_wrk->dmub_notify->type < ARRAY_SIZE(dmub_hpd_wrk->adev->dm.dmub_callback)) {
+		dmub_hpd_wrk->adev->dm.dmub_callback[dmub_hpd_wrk->dmub_notify->type](dmub_hpd_wrk->adev,
+		dmub_hpd_wrk->dmub_notify);
+	}
+
+	kfree(dmub_hpd_wrk->dmub_notify);
+	kfree(dmub_hpd_wrk);
+
+}
+
+static const char *dmub_notification_type_str(enum dmub_notification_type e)
+{
+	switch (e) {
+	case DMUB_NOTIFICATION_NO_DATA:
+		return "NO_DATA";
+	case DMUB_NOTIFICATION_AUX_REPLY:
+		return "AUX_REPLY";
+	case DMUB_NOTIFICATION_HPD:
+		return "HPD";
+	case DMUB_NOTIFICATION_HPD_IRQ:
+		return "HPD_IRQ";
+	case DMUB_NOTIFICATION_SET_CONFIG_REPLY:
+		return "SET_CONFIG_REPLY";
+	case DMUB_NOTIFICATION_DPIA_NOTIFICATION:
+		return "DPIA_NOTIFICATION";
+	case DMUB_NOTIFICATION_HPD_SENSE_NOTIFY:
+		return "HPD_SENSE_NOTIFY";
+	case DMUB_NOTIFICATION_FUSED_IO:
+		return "FUSED_IO";
+	default:
+		return "<unknown>";
+	}
+}
+
+#define DMUB_TRACE_MAX_READ 64
+/**
+ * dm_dmub_outbox1_low_irq() - Handles Outbox interrupt
+ * @interrupt_params: used for determining the Outbox instance
+ *
+ * Handles the Outbox Interrupt
+ * event handler.
+ */
+static void dm_dmub_outbox1_low_irq(void *interrupt_params)
+{
+	struct dmub_notification notify = {0};
+	struct common_irq_params *irq_params = interrupt_params;
+	struct amdgpu_device *adev = irq_params->adev;
+	struct amdgpu_display_manager *dm = &adev->dm;
+	struct dmcub_trace_buf_entry entry = { 0 };
+	u32 count = 0;
+	struct dmub_hpd_work *dmub_hpd_wrk;
+
+	do {
+		if (dc_dmub_srv_get_dmub_outbox0_msg(dm->dc, &entry)) {
+			trace_amdgpu_dmub_trace_high_irq(entry.trace_code, entry.tick_count,
+							entry.param0, entry.param1);
+
+			drm_dbg_driver(adev_to_drm(adev), "trace_code:%u, tick_count:%u, param0:%u, param1:%u\n",
+				 entry.trace_code, entry.tick_count, entry.param0, entry.param1);
+		} else
+			break;
+
+		count++;
+
+	} while (count <= DMUB_TRACE_MAX_READ);
+
+	if (count > DMUB_TRACE_MAX_READ)
+		drm_dbg_driver(adev_to_drm(adev), "Warning : count > DMUB_TRACE_MAX_READ");
+
+	if (dc_enable_dmub_notifications(adev->dm.dc) &&
+		irq_params->irq_src == DC_IRQ_SOURCE_DMCUB_OUTBOX) {
+
+		do {
+			dc_stat_get_dmub_notification(adev->dm.dc, &notify);
+			if (notify.type >= ARRAY_SIZE(dm->dmub_thread_offload)) {
+				drm_err(adev_to_drm(adev), "DM: notify type %d invalid!", notify.type);
+				continue;
+			}
+			if (!dm->dmub_callback[notify.type]) {
+				drm_warn(adev_to_drm(adev), "DMUB notification skipped due to no handler: type=%s\n",
+					dmub_notification_type_str(notify.type));
+				continue;
+			}
+			if (dm->dmub_thread_offload[notify.type] == true) {
+				dmub_hpd_wrk = kzalloc(sizeof(*dmub_hpd_wrk), GFP_ATOMIC);
+				if (!dmub_hpd_wrk) {
+					drm_err(adev_to_drm(adev), "Failed to allocate dmub_hpd_wrk");
+					return;
+				}
+				dmub_hpd_wrk->dmub_notify = kmemdup(&notify, sizeof(struct dmub_notification),
+								    GFP_ATOMIC);
+				if (!dmub_hpd_wrk->dmub_notify) {
+					kfree(dmub_hpd_wrk);
+					drm_err(adev_to_drm(adev), "Failed to allocate dmub_hpd_wrk->dmub_notify");
+					return;
+				}
+				INIT_WORK(&dmub_hpd_wrk->handle_hpd_work, dm_handle_hpd_work);
+				dmub_hpd_wrk->adev = adev;
+				queue_work(adev->dm.delayed_hpd_wq, &dmub_hpd_wrk->handle_hpd_work);
+			} else {
+				dm->dmub_callback[notify.type](adev, &notify);
+			}
+		} while (notify.pending_notification);
+	}
+}
+
+static int dm_set_clockgating_state(struct amdgpu_ip_block *ip_block,
+		  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int dm_set_powergating_state(struct amdgpu_ip_block *ip_block,
+		  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+/* Prototypes of private functions */
+static int dm_early_init(struct amdgpu_ip_block *ip_block);
+
+/* Allocate memory for FBC compressed data  */
+static void amdgpu_dm_fbc_init(struct drm_connector *connector)
+{
+	struct amdgpu_device *adev = drm_to_adev(connector->dev);
+	struct dm_compressor_info *compressor = &adev->dm.compressor;
+	struct amdgpu_dm_connector *aconn = to_amdgpu_dm_connector(connector);
+	struct drm_display_mode *mode;
+	unsigned long max_size = 0;
+
+	if (adev->dm.dc->fbc_compressor == NULL)
+		return;
+
+	if (aconn->dc_link->connector_signal != SIGNAL_TYPE_EDP)
+		return;
+
+	if (compressor->bo_ptr)
+		return;
+
+
+	list_for_each_entry(mode, &connector->modes, head) {
+		if (max_size < (unsigned long) mode->htotal * mode->vtotal)
+			max_size = (unsigned long) mode->htotal * mode->vtotal;
+	}
+
+	if (max_size) {
+		int r = amdgpu_bo_create_kernel(adev, max_size * 4, PAGE_SIZE,
+			    AMDGPU_GEM_DOMAIN_GTT, &compressor->bo_ptr,
+			    &compressor->gpu_addr, &compressor->cpu_addr);
+
+		if (r)
+			drm_err(adev_to_drm(adev), "DM: Failed to initialize FBC\n");
+		else {
+			adev->dm.dc->ctx->fbc_gpu_addr = compressor->gpu_addr;
+			drm_info(adev_to_drm(adev), "DM: FBC alloc %lu\n", max_size*4);
+		}
+
+	}
+
+}
+
+static int amdgpu_dm_audio_component_get_eld(struct device *kdev, int port,
+					  int pipe, bool *enabled,
+					  unsigned char *buf, int max_bytes)
+{
+	struct drm_device *dev = dev_get_drvdata(kdev);
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	struct drm_connector *connector;
+	struct drm_connector_list_iter conn_iter;
+	struct amdgpu_dm_connector *aconnector;
+	int ret = 0;
+
+	*enabled = false;
+
+	mutex_lock(&adev->dm.audio_lock);
+
+	drm_connector_list_iter_begin(dev, &conn_iter);
+	drm_for_each_connector_iter(connector, &conn_iter) {
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_WRITEBACK)
+			continue;
+
+		aconnector = to_amdgpu_dm_connector(connector);
+		if (aconnector->audio_inst != port)
+			continue;
+
+		*enabled = true;
+		mutex_lock(&connector->eld_mutex);
+		ret = drm_eld_size(connector->eld);
+		memcpy(buf, connector->eld, min(max_bytes, ret));
+		mutex_unlock(&connector->eld_mutex);
+
+		break;
+	}
+	drm_connector_list_iter_end(&conn_iter);
+
+	mutex_unlock(&adev->dm.audio_lock);
+
+	DRM_DEBUG_KMS("Get ELD : idx=%d ret=%d en=%d\n", port, ret, *enabled);
+
+	return ret;
+}
+
+static const struct drm_audio_component_ops amdgpu_dm_audio_component_ops = {
+	.get_eld = amdgpu_dm_audio_component_get_eld,
+};
+
+static int amdgpu_dm_audio_component_bind(struct device *kdev,
+				       struct device *hda_kdev, void *data)
+{
+	struct drm_device *dev = dev_get_drvdata(kdev);
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	struct drm_audio_component *acomp = data;
+
+	acomp->ops = &amdgpu_dm_audio_component_ops;
+	acomp->dev = kdev;
+	adev->dm.audio_component = acomp;
+
+	return 0;
+}
+
+static void amdgpu_dm_audio_component_unbind(struct device *kdev,
+					  struct device *hda_kdev, void *data)
+{
+	struct amdgpu_device *adev = drm_to_adev(dev_get_drvdata(kdev));
+	struct drm_audio_component *acomp = data;
+
+	acomp->ops = NULL;
+	acomp->dev = NULL;
+	adev->dm.audio_component = NULL;
+}
+
+static const struct component_ops amdgpu_dm_audio_component_bind_ops = {
+	.bind	= amdgpu_dm_audio_component_bind,
+	.unbind	= amdgpu_dm_audio_component_unbind,
+};
+
+static int amdgpu_dm_audio_init(struct amdgpu_device *adev)
+{
+	int i, ret;
+
+	if (!amdgpu_audio)
+		return 0;
+
+	adev->mode_info.audio.enabled = true;
+
+	adev->mode_info.audio.num_pins = adev->dm.dc->res_pool->audio_count;
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		adev->mode_info.audio.pin[i].channels = -1;
+		adev->mode_info.audio.pin[i].rate = -1;
+		adev->mode_info.audio.pin[i].bits_per_sample = -1;
+		adev->mode_info.audio.pin[i].status_bits = 0;
+		adev->mode_info.audio.pin[i].category_code = 0;
+		adev->mode_info.audio.pin[i].connected = false;
+		adev->mode_info.audio.pin[i].id =
+			adev->dm.dc->res_pool->audios[i]->inst;
+		adev->mode_info.audio.pin[i].offset = 0;
+	}
+
+	ret = component_add(adev->dev, &amdgpu_dm_audio_component_bind_ops);
+	if (ret < 0)
+		return ret;
+
+	adev->dm.audio_registered = true;
+
+	return 0;
+}
+
+static void amdgpu_dm_audio_fini(struct amdgpu_device *adev)
+{
+	if (!amdgpu_audio)
+		return;
+
+	if (!adev->mode_info.audio.enabled)
+		return;
+
+	if (adev->dm.audio_registered) {
+		component_del(adev->dev, &amdgpu_dm_audio_component_bind_ops);
+		adev->dm.audio_registered = false;
+	}
+
+	/* TODO: Disable audio? */
+
+	adev->mode_info.audio.enabled = false;
+}
+
+static  void amdgpu_dm_audio_eld_notify(struct amdgpu_device *adev, int pin)
+{
+	struct drm_audio_component *acomp = adev->dm.audio_component;
+
+	if (acomp && acomp->audio_ops && acomp->audio_ops->pin_eld_notify) {
+		DRM_DEBUG_KMS("Notify ELD: %d\n", pin);
+
+		acomp->audio_ops->pin_eld_notify(acomp->audio_ops->audio_ptr,
+						 pin, -1);
+	}
+}
+
+static int dm_dmub_hw_init(struct amdgpu_device *adev)
+{
+	const struct dmcub_firmware_header_v1_0 *hdr;
+	struct dmub_srv *dmub_srv = adev->dm.dmub_srv;
+	struct dmub_srv_fb_info *fb_info = adev->dm.dmub_fb_info;
+	const struct firmware *dmub_fw = adev->dm.dmub_fw;
+	struct dmcu *dmcu = adev->dm.dc->res_pool->dmcu;
+	struct abm *abm = adev->dm.dc->res_pool->abm;
+	struct dc_context *ctx = adev->dm.dc->ctx;
+	struct dmub_srv_hw_params hw_params;
+	enum dmub_status status;
+	const unsigned char *fw_inst_const, *fw_bss_data;
+	u32 i, fw_inst_const_size, fw_bss_data_size;
+	bool has_hw_support;
+
+	if (!dmub_srv)
+		/* DMUB isn't supported on the ASIC. */
+		return 0;
+
+	if (!fb_info) {
+		drm_err(adev_to_drm(adev), "No framebuffer info for DMUB service.\n");
+		return -EINVAL;
+	}
+
+	if (!dmub_fw) {
+		/* Firmware required for DMUB support. */
+		drm_err(adev_to_drm(adev), "No firmware provided for DMUB.\n");
+		return -EINVAL;
+	}
+
+	/* initialize register offsets for ASICs with runtime initialization available */
+	if (dmub_srv->hw_funcs.init_reg_offsets)
+		dmub_srv->hw_funcs.init_reg_offsets(dmub_srv, ctx);
+
+	status = dmub_srv_has_hw_support(dmub_srv, &has_hw_support);
+	if (status != DMUB_STATUS_OK) {
+		drm_err(adev_to_drm(adev), "Error checking HW support for DMUB: %d\n", status);
+		return -EINVAL;
+	}
+
+	if (!has_hw_support) {
+		drm_info(adev_to_drm(adev), "DMUB unsupported on ASIC\n");
+		return 0;
+	}
+
+	/* Reset DMCUB if it was previously running - before we overwrite its memory. */
+	status = dmub_srv_hw_reset(dmub_srv);
+	if (status != DMUB_STATUS_OK)
+		drm_warn(adev_to_drm(adev), "Error resetting DMUB HW: %d\n", status);
+
+	hdr = (const struct dmcub_firmware_header_v1_0 *)dmub_fw->data;
+
+	fw_inst_const = dmub_fw->data +
+			le32_to_cpu(hdr->header.ucode_array_offset_bytes) +
+			PSP_HEADER_BYTES;
+
+	fw_bss_data = dmub_fw->data +
+		      le32_to_cpu(hdr->header.ucode_array_offset_bytes) +
+		      le32_to_cpu(hdr->inst_const_bytes);
+
+	/* Copy firmware and bios info into FB memory. */
+	fw_inst_const_size = le32_to_cpu(hdr->inst_const_bytes) -
+			     PSP_HEADER_BYTES - PSP_FOOTER_BYTES;
+
+	fw_bss_data_size = le32_to_cpu(hdr->bss_data_bytes);
+
+	/* if adev->firmware.load_type == AMDGPU_FW_LOAD_PSP,
+	 * amdgpu_ucode_init_single_fw will load dmub firmware
+	 * fw_inst_const part to cw0; otherwise, the firmware back door load
+	 * will be done by dm_dmub_hw_init
+	 */
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
+		memcpy(fb_info->fb[DMUB_WINDOW_0_INST_CONST].cpu_addr, fw_inst_const,
+				fw_inst_const_size);
+	}
+
+	if (fw_bss_data_size)
+		memcpy(fb_info->fb[DMUB_WINDOW_2_BSS_DATA].cpu_addr,
+		       fw_bss_data, fw_bss_data_size);
+
+	/* Copy firmware bios info into FB memory. */
+	memcpy(fb_info->fb[DMUB_WINDOW_3_VBIOS].cpu_addr, adev->bios,
+	       adev->bios_size);
+
+	/* Reset regions that need to be reset. */
+	memset(fb_info->fb[DMUB_WINDOW_4_MAILBOX].cpu_addr, 0,
+	fb_info->fb[DMUB_WINDOW_4_MAILBOX].size);
+
+	memset(fb_info->fb[DMUB_WINDOW_5_TRACEBUFF].cpu_addr, 0,
+	       fb_info->fb[DMUB_WINDOW_5_TRACEBUFF].size);
+
+	memset(fb_info->fb[DMUB_WINDOW_6_FW_STATE].cpu_addr, 0,
+	       fb_info->fb[DMUB_WINDOW_6_FW_STATE].size);
+
+	memset(fb_info->fb[DMUB_WINDOW_SHARED_STATE].cpu_addr, 0,
+	       fb_info->fb[DMUB_WINDOW_SHARED_STATE].size);
+
+	/* Initialize hardware. */
+	memset(&hw_params, 0, sizeof(hw_params));
+	hw_params.fb_base = adev->gmc.fb_start;
+	hw_params.fb_offset = adev->vm_manager.vram_base_offset;
+
+	/* backdoor load firmware and trigger dmub running */
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
+		hw_params.load_inst_const = true;
+
+	if (dmcu)
+		hw_params.psp_version = dmcu->psp_version;
+
+	for (i = 0; i < fb_info->num_fb; ++i)
+		hw_params.fb[i] = &fb_info->fb[i];
+
+	switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) {
+	case IP_VERSION(3, 1, 3):
+	case IP_VERSION(3, 1, 4):
+	case IP_VERSION(3, 5, 0):
+	case IP_VERSION(3, 5, 1):
+	case IP_VERSION(3, 6, 0):
+	case IP_VERSION(4, 0, 1):
+		hw_params.dpia_supported = true;
+		hw_params.disable_dpia = adev->dm.dc->debug.dpia_debug.bits.disable_dpia;
+		break;
+	default:
+		break;
+	}
+
+	switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) {
+	case IP_VERSION(3, 5, 0):
+	case IP_VERSION(3, 5, 1):
+	case IP_VERSION(3, 6, 0):
+		hw_params.ips_sequential_ono = adev->external_rev_id > 0x10;
+		hw_params.lower_hbr3_phy_ssc = true;
+		break;
+	default:
+		break;
+	}
+
+	status = dmub_srv_hw_init(dmub_srv, &hw_params);
+	if (status != DMUB_STATUS_OK) {
+		drm_err(adev_to_drm(adev), "Error initializing DMUB HW: %d\n", status);
+		return -EINVAL;
+	}
+
+	/* Wait for firmware load to finish. */
+	status = dmub_srv_wait_for_auto_load(dmub_srv, 100000);
+	if (status != DMUB_STATUS_OK)
+		drm_warn(adev_to_drm(adev), "Wait for DMUB auto-load failed: %d\n", status);
+
+	/* Init DMCU and ABM if available. */
+	if (dmcu && abm) {
+		dmcu->funcs->dmcu_init(dmcu);
+		abm->dmcu_is_running = dmcu->funcs->is_dmcu_initialized(dmcu);
+	}
+
+	if (!adev->dm.dc->ctx->dmub_srv)
+		adev->dm.dc->ctx->dmub_srv = dc_dmub_srv_create(adev->dm.dc, dmub_srv);
+	if (!adev->dm.dc->ctx->dmub_srv) {
+		drm_err(adev_to_drm(adev), "Couldn't allocate DC DMUB server!\n");
+		return -ENOMEM;
+	}
+
+	drm_info(adev_to_drm(adev), "DMUB hardware initialized: version=0x%08X\n",
+		 adev->dm.dmcub_fw_version);
+
+	/* Keeping sanity checks off if
+	 * DCN31 >= 4.0.59.0
+	 * DCN314 >= 8.0.16.0
+	 * Otherwise, turn on sanity checks
+	 */
+	switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) {
+	case IP_VERSION(3, 1, 2):
+	case IP_VERSION(3, 1, 3):
+		if (adev->dm.dmcub_fw_version &&
+			adev->dm.dmcub_fw_version >= DMUB_FW_VERSION(4, 0, 0) &&
+			adev->dm.dmcub_fw_version < DMUB_FW_VERSION(4, 0, 59))
+				adev->dm.dc->debug.sanity_checks = true;
+		break;
+	case IP_VERSION(3, 1, 4):
+		if (adev->dm.dmcub_fw_version &&
+			adev->dm.dmcub_fw_version >= DMUB_FW_VERSION(4, 0, 0) &&
+			adev->dm.dmcub_fw_version < DMUB_FW_VERSION(8, 0, 16))
+				adev->dm.dc->debug.sanity_checks = true;
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static void dm_dmub_hw_resume(struct amdgpu_device *adev)
+{
+	struct dmub_srv *dmub_srv = adev->dm.dmub_srv;
+	enum dmub_status status;
+	bool init;
+	int r;
+
+	if (!dmub_srv) {
+		/* DMUB isn't supported on the ASIC. */
+		return;
+	}
+
+	status = dmub_srv_is_hw_init(dmub_srv, &init);
+	if (status != DMUB_STATUS_OK)
+		drm_warn(adev_to_drm(adev), "DMUB hardware init check failed: %d\n", status);
+
+	if (status == DMUB_STATUS_OK && init) {
+		/* Wait for firmware load to finish. */
+		status = dmub_srv_wait_for_auto_load(dmub_srv, 100000);
+		if (status != DMUB_STATUS_OK)
+			drm_warn(adev_to_drm(adev), "Wait for DMUB auto-load failed: %d\n", status);
+	} else {
+		/* Perform the full hardware initialization. */
+		r = dm_dmub_hw_init(adev);
+		if (r)
+			drm_err(adev_to_drm(adev), "DMUB interface failed to initialize: status=%d\n", r);
+	}
+}
+
+static void mmhub_read_system_context(struct amdgpu_device *adev, struct dc_phy_addr_space_config *pa_config)
+{
+	u64 pt_base;
+	u32 logical_addr_low;
+	u32 logical_addr_high;
+	u32 agp_base, agp_bot, agp_top;
+	PHYSICAL_ADDRESS_LOC page_table_start, page_table_end, page_table_base;
+
+	memset(pa_config, 0, sizeof(*pa_config));
+
+	agp_base = 0;
+	agp_bot = adev->gmc.agp_start >> 24;
+	agp_top = adev->gmc.agp_end >> 24;
+
+	/* AGP aperture is disabled */
+	if (agp_bot > agp_top) {
+		logical_addr_low = adev->gmc.fb_start >> 18;
+		if (adev->apu_flags & (AMD_APU_IS_RAVEN2 |
+				       AMD_APU_IS_RENOIR |
+				       AMD_APU_IS_GREEN_SARDINE))
+			/*
+			 * Raven2 has a HW issue that it is unable to use the vram which
+			 * is out of MC_VM_SYSTEM_APERTURE_HIGH_ADDR. So here is the
+			 * workaround that increase system aperture high address (add 1)
+			 * to get rid of the VM fault and hardware hang.
+			 */
+			logical_addr_high = (adev->gmc.fb_end >> 18) + 0x1;
+		else
+			logical_addr_high = adev->gmc.fb_end >> 18;
+	} else {
+		logical_addr_low = min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18;
+		if (adev->apu_flags & (AMD_APU_IS_RAVEN2 |
+				       AMD_APU_IS_RENOIR |
+				       AMD_APU_IS_GREEN_SARDINE))
+			/*
+			 * Raven2 has a HW issue that it is unable to use the vram which
+			 * is out of MC_VM_SYSTEM_APERTURE_HIGH_ADDR. So here is the
+			 * workaround that increase system aperture high address (add 1)
+			 * to get rid of the VM fault and hardware hang.
+			 */
+			logical_addr_high = max((adev->gmc.fb_end >> 18) + 0x1, adev->gmc.agp_end >> 18);
+		else
+			logical_addr_high = max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18;
+	}
+
+	pt_base = amdgpu_gmc_pd_addr(adev->gart.bo);
+
+	page_table_start.high_part = upper_32_bits(adev->gmc.gart_start >>
+						   AMDGPU_GPU_PAGE_SHIFT);
+	page_table_start.low_part = lower_32_bits(adev->gmc.gart_start >>
+						  AMDGPU_GPU_PAGE_SHIFT);
+	page_table_end.high_part = upper_32_bits(adev->gmc.gart_end >>
+						 AMDGPU_GPU_PAGE_SHIFT);
+	page_table_end.low_part = lower_32_bits(adev->gmc.gart_end >>
+						AMDGPU_GPU_PAGE_SHIFT);
+	page_table_base.high_part = upper_32_bits(pt_base);
+	page_table_base.low_part = lower_32_bits(pt_base);
+
+	pa_config->system_aperture.start_addr = (uint64_t)logical_addr_low << 18;
+	pa_config->system_aperture.end_addr = (uint64_t)logical_addr_high << 18;
+
+	pa_config->system_aperture.agp_base = (uint64_t)agp_base << 24;
+	pa_config->system_aperture.agp_bot = (uint64_t)agp_bot << 24;
+	pa_config->system_aperture.agp_top = (uint64_t)agp_top << 24;
+
+	pa_config->system_aperture.fb_base = adev->gmc.fb_start;
+	pa_config->system_aperture.fb_offset = adev->vm_manager.vram_base_offset;
+	pa_config->system_aperture.fb_top = adev->gmc.fb_end;
+
+	pa_config->gart_config.page_table_start_addr = page_table_start.quad_part << 12;
+	pa_config->gart_config.page_table_end_addr = page_table_end.quad_part << 12;
+	pa_config->gart_config.page_table_base_addr = page_table_base.quad_part;
+
+	pa_config->is_hvm_enabled = adev->mode_info.gpu_vm_support;
+
+}
+
+static void force_connector_state(
+	struct amdgpu_dm_connector *aconnector,
+	enum drm_connector_force force_state)
+{
+	struct drm_connector *connector = &aconnector->base;
+
+	mutex_lock(&connector->dev->mode_config.mutex);
+	aconnector->base.force = force_state;
+	mutex_unlock(&connector->dev->mode_config.mutex);
+
+	mutex_lock(&aconnector->hpd_lock);
+	drm_kms_helper_connector_hotplug_event(connector);
+	mutex_unlock(&aconnector->hpd_lock);
+}
+
+static void dm_handle_hpd_rx_offload_work(struct work_struct *work)
+{
+	struct hpd_rx_irq_offload_work *offload_work;
+	struct amdgpu_dm_connector *aconnector;
+	struct dc_link *dc_link;
+	struct amdgpu_device *adev;
+	enum dc_connection_type new_connection_type = dc_connection_none;
+	unsigned long flags;
+	union test_response test_response;
+
+	memset(&test_response, 0, sizeof(test_response));
+
+	offload_work = container_of(work, struct hpd_rx_irq_offload_work, work);
+	aconnector = offload_work->offload_wq->aconnector;
+	adev = offload_work->adev;
+
+	if (!aconnector) {
+		drm_err(adev_to_drm(adev), "Can't retrieve aconnector in hpd_rx_irq_offload_work");
+		goto skip;
+	}
+
+	dc_link = aconnector->dc_link;
+
+	mutex_lock(&aconnector->hpd_lock);
+	if (!dc_link_detect_connection_type(dc_link, &new_connection_type))
+		drm_err(adev_to_drm(adev), "KMS: Failed to detect connector\n");
+	mutex_unlock(&aconnector->hpd_lock);
+
+	if (new_connection_type == dc_connection_none)
+		goto skip;
+
+	if (amdgpu_in_reset(adev))
+		goto skip;
+
+	if (offload_work->data.bytes.device_service_irq.bits.UP_REQ_MSG_RDY ||
+		offload_work->data.bytes.device_service_irq.bits.DOWN_REP_MSG_RDY) {
+		dm_handle_mst_sideband_msg_ready_event(&aconnector->mst_mgr, DOWN_OR_UP_MSG_RDY_EVENT);
+		spin_lock_irqsave(&offload_work->offload_wq->offload_lock, flags);
+		offload_work->offload_wq->is_handling_mst_msg_rdy_event = false;
+		spin_unlock_irqrestore(&offload_work->offload_wq->offload_lock, flags);
+		goto skip;
+	}
+
+	mutex_lock(&adev->dm.dc_lock);
+	if (offload_work->data.bytes.device_service_irq.bits.AUTOMATED_TEST) {
+		dc_link_dp_handle_automated_test(dc_link);
+
+		if (aconnector->timing_changed) {
+			/* force connector disconnect and reconnect */
+			force_connector_state(aconnector, DRM_FORCE_OFF);
+			msleep(100);
+			force_connector_state(aconnector, DRM_FORCE_UNSPECIFIED);
+		}
+
+		test_response.bits.ACK = 1;
+
+		core_link_write_dpcd(
+		dc_link,
+		DP_TEST_RESPONSE,
+		&test_response.raw,
+		sizeof(test_response));
+	} else if ((dc_link->connector_signal != SIGNAL_TYPE_EDP) &&
+			dc_link_check_link_loss_status(dc_link, &offload_work->data) &&
+			dc_link_dp_allow_hpd_rx_irq(dc_link)) {
+		/* offload_work->data is from handle_hpd_rx_irq->
+		 * schedule_hpd_rx_offload_work.this is defer handle
+		 * for hpd short pulse. upon here, link status may be
+		 * changed, need get latest link status from dpcd
+		 * registers. if link status is good, skip run link
+		 * training again.
+		 */
+		union hpd_irq_data irq_data;
+
+		memset(&irq_data, 0, sizeof(irq_data));
+
+		/* before dc_link_dp_handle_link_loss, allow new link lost handle
+		 * request be added to work queue if link lost at end of dc_link_
+		 * dp_handle_link_loss
+		 */
+		spin_lock_irqsave(&offload_work->offload_wq->offload_lock, flags);
+		offload_work->offload_wq->is_handling_link_loss = false;
+		spin_unlock_irqrestore(&offload_work->offload_wq->offload_lock, flags);
+
+		if ((dc_link_dp_read_hpd_rx_irq_data(dc_link, &irq_data) == DC_OK) &&
+			dc_link_check_link_loss_status(dc_link, &irq_data))
+			dc_link_dp_handle_link_loss(dc_link);
+	}
+	mutex_unlock(&adev->dm.dc_lock);
+
+skip:
+	kfree(offload_work);
+
+}
+
+static struct hpd_rx_irq_offload_work_queue *hpd_rx_irq_create_workqueue(struct amdgpu_device *adev)
+{
+	struct dc *dc = adev->dm.dc;
+	int max_caps = dc->caps.max_links;
+	int i = 0;
+	struct hpd_rx_irq_offload_work_queue *hpd_rx_offload_wq = NULL;
+
+	hpd_rx_offload_wq = kcalloc(max_caps, sizeof(*hpd_rx_offload_wq), GFP_KERNEL);
+
+	if (!hpd_rx_offload_wq)
+		return NULL;
+
+
+	for (i = 0; i < max_caps; i++) {
+		hpd_rx_offload_wq[i].wq =
+				    create_singlethread_workqueue("amdgpu_dm_hpd_rx_offload_wq");
+
+		if (hpd_rx_offload_wq[i].wq == NULL) {
+			drm_err(adev_to_drm(adev), "create amdgpu_dm_hpd_rx_offload_wq fail!");
+			goto out_err;
+		}
+
+		spin_lock_init(&hpd_rx_offload_wq[i].offload_lock);
+	}
+
+	return hpd_rx_offload_wq;
+
+out_err:
+	for (i = 0; i < max_caps; i++) {
+		if (hpd_rx_offload_wq[i].wq)
+			destroy_workqueue(hpd_rx_offload_wq[i].wq);
+	}
+	kfree(hpd_rx_offload_wq);
+	return NULL;
+}
+
+struct amdgpu_stutter_quirk {
+	u16 chip_vendor;
+	u16 chip_device;
+	u16 subsys_vendor;
+	u16 subsys_device;
+	u8 revision;
+};
+
+static const struct amdgpu_stutter_quirk amdgpu_stutter_quirk_list[] = {
+	/* https://bugzilla.kernel.org/show_bug.cgi?id=214417 */
+	{ 0x1002, 0x15dd, 0x1002, 0x15dd, 0xc8 },
+	{ 0, 0, 0, 0, 0 },
+};
+
+static bool dm_should_disable_stutter(struct pci_dev *pdev)
+{
+	const struct amdgpu_stutter_quirk *p = amdgpu_stutter_quirk_list;
+
+	while (p && p->chip_device != 0) {
+		if (pdev->vendor == p->chip_vendor &&
+		    pdev->device == p->chip_device &&
+		    pdev->subsystem_vendor == p->subsys_vendor &&
+		    pdev->subsystem_device == p->subsys_device &&
+		    pdev->revision == p->revision) {
+			return true;
+		}
+		++p;
+	}
+	return false;
+}
+
+
+void*
+dm_allocate_gpu_mem(
+		struct amdgpu_device *adev,
+		enum dc_gpu_mem_alloc_type type,
+		size_t size,
+		long long *addr)
+{
+	struct dal_allocation *da;
+	u32 domain = (type == DC_MEM_ALLOC_TYPE_GART) ?
+		AMDGPU_GEM_DOMAIN_GTT : AMDGPU_GEM_DOMAIN_VRAM;
+	int ret;
+
+	da = kzalloc(sizeof(struct dal_allocation), GFP_KERNEL);
+	if (!da)
+		return NULL;
+
+	ret = amdgpu_bo_create_kernel(adev, size, PAGE_SIZE,
+				      domain, &da->bo,
+				      &da->gpu_addr, &da->cpu_ptr);
+
+	*addr = da->gpu_addr;
+
+	if (ret) {
+		kfree(da);
+		return NULL;
+	}
+
+	/* add da to list in dm */
+	list_add(&da->list, &adev->dm.da_list);
+
+	return da->cpu_ptr;
+}
+
+void
+dm_free_gpu_mem(
+		struct amdgpu_device *adev,
+		enum dc_gpu_mem_alloc_type type,
+		void *pvMem)
+{
+	struct dal_allocation *da;
+
+	/* walk the da list in DM */
+	list_for_each_entry(da, &adev->dm.da_list, list) {
+		if (pvMem == da->cpu_ptr) {
+			amdgpu_bo_free_kernel(&da->bo, &da->gpu_addr, &da->cpu_ptr);
+			list_del(&da->list);
+			kfree(da);
+			break;
+		}
+	}
+
+}
+
+static enum dmub_status
+dm_dmub_send_vbios_gpint_command(struct amdgpu_device *adev,
+				 enum dmub_gpint_command command_code,
+				 uint16_t param,
+				 uint32_t timeout_us)
+{
+	union dmub_gpint_data_register reg, test;
+	uint32_t i;
+
+	/* Assume that VBIOS DMUB is ready to take commands */
+
+	reg.bits.status = 1;
+	reg.bits.command_code = command_code;
+	reg.bits.param = param;
+
+	cgs_write_register(adev->dm.cgs_device, 0x34c0 + 0x01f8, reg.all);
+
+	for (i = 0; i < timeout_us; ++i) {
+		udelay(1);
+
+		/* Check if our GPINT got acked */
+		reg.bits.status = 0;
+		test = (union dmub_gpint_data_register)
+			cgs_read_register(adev->dm.cgs_device, 0x34c0 + 0x01f8);
+
+		if (test.all == reg.all)
+			return DMUB_STATUS_OK;
+	}
+
+	return DMUB_STATUS_TIMEOUT;
+}
+
+static void *dm_dmub_get_vbios_bounding_box(struct amdgpu_device *adev)
+{
+	void *bb;
+	long long addr;
+	unsigned int bb_size;
+	int i = 0;
+	uint16_t chunk;
+	enum dmub_gpint_command send_addrs[] = {
+		DMUB_GPINT__SET_BB_ADDR_WORD0,
+		DMUB_GPINT__SET_BB_ADDR_WORD1,
+		DMUB_GPINT__SET_BB_ADDR_WORD2,
+		DMUB_GPINT__SET_BB_ADDR_WORD3,
+	};
+	enum dmub_status ret;
+
+	switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) {
+	case IP_VERSION(4, 0, 1):
+		bb_size = sizeof(struct dml2_soc_bb);
+		break;
+	default:
+		return NULL;
+	}
+
+	bb =  dm_allocate_gpu_mem(adev,
+				  DC_MEM_ALLOC_TYPE_GART,
+				  bb_size,
+				  &addr);
+	if (!bb)
+		return NULL;
+
+	for (i = 0; i < 4; i++) {
+		/* Extract 16-bit chunk */
+		chunk = ((uint64_t) addr >> (i * 16)) & 0xFFFF;
+		/* Send the chunk */
+		ret = dm_dmub_send_vbios_gpint_command(adev, send_addrs[i], chunk, 30000);
+		if (ret != DMUB_STATUS_OK)
+			goto free_bb;
+	}
+
+	/* Now ask DMUB to copy the bb */
+	ret = dm_dmub_send_vbios_gpint_command(adev, DMUB_GPINT__BB_COPY, 1, 200000);
+	if (ret != DMUB_STATUS_OK)
+		goto free_bb;
+
+	return bb;
+
+free_bb:
+	dm_free_gpu_mem(adev, DC_MEM_ALLOC_TYPE_GART, (void *) bb);
+	return NULL;
+
+}
+
+static enum dmub_ips_disable_type dm_get_default_ips_mode(
+	struct amdgpu_device *adev)
+{
+	enum dmub_ips_disable_type ret = DMUB_IPS_ENABLE;
+
+	switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) {
+	case IP_VERSION(3, 5, 0):
+	case IP_VERSION(3, 6, 0):
+	case IP_VERSION(3, 5, 1):
+		ret =  DMUB_IPS_RCG_IN_ACTIVE_IPS2_IN_OFF;
+		break;
+	default:
+		/* ASICs older than DCN35 do not have IPSs */
+		if (amdgpu_ip_version(adev, DCE_HWIP, 0) < IP_VERSION(3, 5, 0))
+			ret = DMUB_IPS_DISABLE_ALL;
+		break;
+	}
+
+	return ret;
+}
+
+static int amdgpu_dm_init(struct amdgpu_device *adev)
+{
+	struct dc_init_data init_data;
+	struct dc_callback_init init_params;
+	int r;
+
+	adev->dm.ddev = adev_to_drm(adev);
+	adev->dm.adev = adev;
+
+	/* Zero all the fields */
+	memset(&init_data, 0, sizeof(init_data));
+	memset(&init_params, 0, sizeof(init_params));
+
+	mutex_init(&adev->dm.dpia_aux_lock);
+	mutex_init(&adev->dm.dc_lock);
+	mutex_init(&adev->dm.audio_lock);
+
+	if (amdgpu_dm_irq_init(adev)) {
+		drm_err(adev_to_drm(adev), "failed to initialize DM IRQ support.\n");
+		goto error;
+	}
+
+	init_data.asic_id.chip_family = adev->family;
+
+	init_data.asic_id.pci_revision_id = adev->pdev->revision;
+	init_data.asic_id.hw_internal_rev = adev->external_rev_id;
+	init_data.asic_id.chip_id = adev->pdev->device;
+
+	init_data.asic_id.vram_width = adev->gmc.vram_width;
+	/* TODO: initialize init_data.asic_id.vram_type here!!!! */
+	init_data.asic_id.atombios_base_address =
+		adev->mode_info.atom_context->bios;
+
+	init_data.driver = adev;
+
+	/* cgs_device was created in dm_sw_init() */
+	init_data.cgs_device = adev->dm.cgs_device;
+
+	init_data.dce_environment = DCE_ENV_PRODUCTION_DRV;
+
+	switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) {
+	case IP_VERSION(2, 1, 0):
+		switch (adev->dm.dmcub_fw_version) {
+		case 0: /* development */
+		case 0x1: /* linux-firmware.git hash 6d9f399 */
+		case 0x01000000: /* linux-firmware.git hash 9a0b0f4 */
+			init_data.flags.disable_dmcu = false;
+			break;
+		default:
+			init_data.flags.disable_dmcu = true;
+		}
+		break;
+	case IP_VERSION(2, 0, 3):
+		init_data.flags.disable_dmcu = true;
+		break;
+	default:
+		break;
+	}
+
+	/* APU support S/G display by default except:
+	 * ASICs before Carrizo,
+	 * RAVEN1 (Users reported stability issue)
+	 */
+
+	if (adev->asic_type < CHIP_CARRIZO) {
+		init_data.flags.gpu_vm_support = false;
+	} else if (adev->asic_type == CHIP_RAVEN) {
+		if (adev->apu_flags & AMD_APU_IS_RAVEN)
+			init_data.flags.gpu_vm_support = false;
+		else
+			init_data.flags.gpu_vm_support = (amdgpu_sg_display != 0);
+	} else {
+		if (amdgpu_ip_version(adev, DCE_HWIP, 0) == IP_VERSION(2, 0, 3))
+			init_data.flags.gpu_vm_support = (amdgpu_sg_display == 1);
+		else
+			init_data.flags.gpu_vm_support =
+				(amdgpu_sg_display != 0) && (adev->flags & AMD_IS_APU);
+	}
+
+	adev->mode_info.gpu_vm_support = init_data.flags.gpu_vm_support;
+
+	if (amdgpu_dc_feature_mask & DC_FBC_MASK)
+		init_data.flags.fbc_support = true;
+
+	if (amdgpu_dc_feature_mask & DC_MULTI_MON_PP_MCLK_SWITCH_MASK)
+		init_data.flags.multi_mon_pp_mclk_switch = true;
+
+	if (amdgpu_dc_feature_mask & DC_DISABLE_FRACTIONAL_PWM_MASK)
+		init_data.flags.disable_fractional_pwm = true;
+
+	if (amdgpu_dc_feature_mask & DC_EDP_NO_POWER_SEQUENCING)
+		init_data.flags.edp_no_power_sequencing = true;
+
+	if (amdgpu_dc_feature_mask & DC_DISABLE_LTTPR_DP1_4A)
+		init_data.flags.allow_lttpr_non_transparent_mode.bits.DP1_4A = true;
+	if (amdgpu_dc_feature_mask & DC_DISABLE_LTTPR_DP2_0)
+		init_data.flags.allow_lttpr_non_transparent_mode.bits.DP2_0 = true;
+
+	init_data.flags.seamless_boot_edp_requested = false;
+
+	if (amdgpu_device_seamless_boot_supported(adev)) {
+		init_data.flags.seamless_boot_edp_requested = true;
+		init_data.flags.allow_seamless_boot_optimization = true;
+		drm_dbg(adev->dm.ddev, "Seamless boot requested\n");
+	}
+
+	init_data.flags.enable_mipi_converter_optimization = true;
+
+	init_data.dcn_reg_offsets = adev->reg_offset[DCE_HWIP][0];
+	init_data.nbio_reg_offsets = adev->reg_offset[NBIO_HWIP][0];
+	init_data.clk_reg_offsets = adev->reg_offset[CLK_HWIP][0];
+
+	if (amdgpu_dc_debug_mask & DC_DISABLE_IPS)
+		init_data.flags.disable_ips = DMUB_IPS_DISABLE_ALL;
+	else if (amdgpu_dc_debug_mask & DC_DISABLE_IPS_DYNAMIC)
+		init_data.flags.disable_ips = DMUB_IPS_DISABLE_DYNAMIC;
+	else if (amdgpu_dc_debug_mask & DC_DISABLE_IPS2_DYNAMIC)
+		init_data.flags.disable_ips = DMUB_IPS_RCG_IN_ACTIVE_IPS2_IN_OFF;
+	else if (amdgpu_dc_debug_mask & DC_FORCE_IPS_ENABLE)
+		init_data.flags.disable_ips = DMUB_IPS_ENABLE;
+	else
+		init_data.flags.disable_ips = dm_get_default_ips_mode(adev);
+
+	init_data.flags.disable_ips_in_vpb = 0;
+
+	/* DCN35 and above supports dynamic DTBCLK switch */
+	if (amdgpu_ip_version(adev, DCE_HWIP, 0) >= IP_VERSION(3, 5, 0))
+		init_data.flags.allow_0_dtb_clk = true;
+
+	/* Enable DWB for tested platforms only */
+	if (amdgpu_ip_version(adev, DCE_HWIP, 0) >= IP_VERSION(3, 0, 0))
+		init_data.num_virtual_links = 1;
+
+	retrieve_dmi_info(&adev->dm);
+	if (adev->dm.edp0_on_dp1_quirk)
+		init_data.flags.support_edp0_on_dp1 = true;
+
+	if (adev->dm.bb_from_dmub)
+		init_data.bb_from_dmub = adev->dm.bb_from_dmub;
+	else
+		init_data.bb_from_dmub = NULL;
+
+	/* Display Core create. */
+	adev->dm.dc = dc_create(&init_data);
+
+	if (adev->dm.dc) {
+		drm_info(adev_to_drm(adev), "Display Core v%s initialized on %s\n", DC_VER,
+			 dce_version_to_string(adev->dm.dc->ctx->dce_version));
+	} else {
+		drm_info(adev_to_drm(adev), "Display Core failed to initialize with v%s!\n", DC_VER);
+		goto error;
+	}
+
+	if (amdgpu_dc_debug_mask & DC_DISABLE_PIPE_SPLIT) {
+		adev->dm.dc->debug.force_single_disp_pipe_split = false;
+		adev->dm.dc->debug.pipe_split_policy = MPC_SPLIT_AVOID;
+	}
+
+	if (adev->asic_type != CHIP_CARRIZO && adev->asic_type != CHIP_STONEY)
+		adev->dm.dc->debug.disable_stutter = amdgpu_pp_feature_mask & PP_STUTTER_MODE ? false : true;
+	if (dm_should_disable_stutter(adev->pdev))
+		adev->dm.dc->debug.disable_stutter = true;
+
+	if (amdgpu_dc_debug_mask & DC_DISABLE_STUTTER)
+		adev->dm.dc->debug.disable_stutter = true;
+
+	if (amdgpu_dc_debug_mask & DC_DISABLE_DSC)
+		adev->dm.dc->debug.disable_dsc = true;
+
+	if (amdgpu_dc_debug_mask & DC_DISABLE_CLOCK_GATING)
+		adev->dm.dc->debug.disable_clock_gate = true;
+
+	if (amdgpu_dc_debug_mask & DC_FORCE_SUBVP_MCLK_SWITCH)
+		adev->dm.dc->debug.force_subvp_mclk_switch = true;
+
+	if (amdgpu_dc_debug_mask & DC_DISABLE_SUBVP_FAMS) {
+		adev->dm.dc->debug.force_disable_subvp = true;
+		adev->dm.dc->debug.fams2_config.bits.enable = false;
+	}
+
+	if (amdgpu_dc_debug_mask & DC_ENABLE_DML2) {
+		adev->dm.dc->debug.using_dml2 = true;
+		adev->dm.dc->debug.using_dml21 = true;
+	}
+
+	if (amdgpu_dc_debug_mask & DC_HDCP_LC_FORCE_FW_ENABLE)
+		adev->dm.dc->debug.hdcp_lc_force_fw_enable = true;
+
+	if (amdgpu_dc_debug_mask & DC_HDCP_LC_ENABLE_SW_FALLBACK)
+		adev->dm.dc->debug.hdcp_lc_enable_sw_fallback = true;
+
+	if (amdgpu_dc_debug_mask & DC_SKIP_DETECTION_LT)
+		adev->dm.dc->debug.skip_detection_link_training = true;
+
+	adev->dm.dc->debug.visual_confirm = amdgpu_dc_visual_confirm;
+
+	/* TODO: Remove after DP2 receiver gets proper support of Cable ID feature */
+	adev->dm.dc->debug.ignore_cable_id = true;
+
+	if (adev->dm.dc->caps.dp_hdmi21_pcon_support)
+		drm_info(adev_to_drm(adev), "DP-HDMI FRL PCON supported\n");
+
+	r = dm_dmub_hw_init(adev);
+	if (r) {
+		drm_err(adev_to_drm(adev), "DMUB interface failed to initialize: status=%d\n", r);
+		goto error;
+	}
+
+	dc_hardware_init(adev->dm.dc);
+
+	adev->dm.restore_backlight = true;
+
+	adev->dm.hpd_rx_offload_wq = hpd_rx_irq_create_workqueue(adev);
+	if (!adev->dm.hpd_rx_offload_wq) {
+		drm_err(adev_to_drm(adev), "failed to create hpd rx offload workqueue.\n");
+		goto error;
+	}
+
+	if ((adev->flags & AMD_IS_APU) && (adev->asic_type >= CHIP_CARRIZO)) {
+		struct dc_phy_addr_space_config pa_config;
+
+		mmhub_read_system_context(adev, &pa_config);
+
+		// Call the DC init_memory func
+		dc_setup_system_context(adev->dm.dc, &pa_config);
+	}
+
+	adev->dm.freesync_module = mod_freesync_create(adev->dm.dc);
+	if (!adev->dm.freesync_module) {
+		drm_err(adev_to_drm(adev),
+		"failed to initialize freesync_module.\n");
+	} else
+		drm_dbg_driver(adev_to_drm(adev), "amdgpu: freesync_module init done %p.\n",
+				adev->dm.freesync_module);
+
+	amdgpu_dm_init_color_mod();
+
+	if (adev->dm.dc->caps.max_links > 0) {
+		adev->dm.vblank_control_workqueue =
+			create_singlethread_workqueue("dm_vblank_control_workqueue");
+		if (!adev->dm.vblank_control_workqueue)
+			drm_err(adev_to_drm(adev), "failed to initialize vblank_workqueue.\n");
+	}
+
+	if (adev->dm.dc->caps.ips_support &&
+	    adev->dm.dc->config.disable_ips != DMUB_IPS_DISABLE_ALL)
+		adev->dm.idle_workqueue = idle_create_workqueue(adev);
+
+	if (adev->dm.dc->caps.max_links > 0 && adev->family >= AMDGPU_FAMILY_RV) {
+		adev->dm.hdcp_workqueue = hdcp_create_workqueue(adev, &init_params.cp_psp, adev->dm.dc);
+
+		if (!adev->dm.hdcp_workqueue)
+			drm_err(adev_to_drm(adev), "failed to initialize hdcp_workqueue.\n");
+		else
+			drm_dbg_driver(adev_to_drm(adev), "amdgpu: hdcp_workqueue init done %p.\n", adev->dm.hdcp_workqueue);
+
+		dc_init_callbacks(adev->dm.dc, &init_params);
+	}
+	if (dc_is_dmub_outbox_supported(adev->dm.dc)) {
+		init_completion(&adev->dm.dmub_aux_transfer_done);
+		adev->dm.dmub_notify = kzalloc(sizeof(struct dmub_notification), GFP_KERNEL);
+		if (!adev->dm.dmub_notify) {
+			drm_info(adev_to_drm(adev), "fail to allocate adev->dm.dmub_notify");
+			goto error;
+		}
+
+		adev->dm.delayed_hpd_wq = create_singlethread_workqueue("amdgpu_dm_hpd_wq");
+		if (!adev->dm.delayed_hpd_wq) {
+			drm_err(adev_to_drm(adev), "failed to create hpd offload workqueue.\n");
+			goto error;
+		}
+
+		amdgpu_dm_outbox_init(adev);
+		if (!register_dmub_notify_callback(adev, DMUB_NOTIFICATION_AUX_REPLY,
+			dmub_aux_setconfig_callback, false)) {
+			drm_err(adev_to_drm(adev), "fail to register dmub aux callback");
+			goto error;
+		}
+
+		for (size_t i = 0; i < ARRAY_SIZE(adev->dm.fused_io); i++)
+			init_completion(&adev->dm.fused_io[i].replied);
+
+		if (!register_dmub_notify_callback(adev, DMUB_NOTIFICATION_FUSED_IO,
+			dmub_aux_fused_io_callback, false)) {
+			drm_err(adev_to_drm(adev), "fail to register dmub fused io callback");
+			goto error;
+		}
+		/* Enable outbox notification only after IRQ handlers are registered and DMUB is alive.
+		 * It is expected that DMUB will resend any pending notifications at this point. Note
+		 * that hpd and hpd_irq handler registration are deferred to register_hpd_handlers() to
+		 * align legacy interface initialization sequence. Connection status will be proactivly
+		 * detected once in the amdgpu_dm_initialize_drm_device.
+		 */
+		dc_enable_dmub_outbox(adev->dm.dc);
+
+		/* DPIA trace goes to dmesg logs only if outbox is enabled */
+		if (amdgpu_dc_debug_mask & DC_ENABLE_DPIA_TRACE)
+			dc_dmub_srv_enable_dpia_trace(adev->dm.dc);
+	}
+
+	if (amdgpu_dm_initialize_drm_device(adev)) {
+		drm_err(adev_to_drm(adev),
+		"failed to initialize sw for display support.\n");
+		goto error;
+	}
+
+	/* create fake encoders for MST */
+	dm_dp_create_fake_mst_encoders(adev);
+
+	/* TODO: Add_display_info? */
+
+	/* TODO use dynamic cursor width */
+	adev_to_drm(adev)->mode_config.cursor_width = adev->dm.dc->caps.max_cursor_size;
+	adev_to_drm(adev)->mode_config.cursor_height = adev->dm.dc->caps.max_cursor_size;
+
+	if (drm_vblank_init(adev_to_drm(adev), adev->dm.display_indexes_num)) {
+		drm_err(adev_to_drm(adev),
+		"failed to initialize vblank for display support.\n");
+		goto error;
+	}
+
+#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY)
+	amdgpu_dm_crtc_secure_display_create_contexts(adev);
+	if (!adev->dm.secure_display_ctx.crtc_ctx)
+		drm_err(adev_to_drm(adev), "failed to initialize secure display contexts.\n");
+
+	if (amdgpu_ip_version(adev, DCE_HWIP, 0) >= IP_VERSION(4, 0, 1))
+		adev->dm.secure_display_ctx.support_mul_roi = true;
+
+#endif
+
+	drm_dbg_driver(adev_to_drm(adev), "KMS initialized.\n");
+
+	return 0;
+error:
+	amdgpu_dm_fini(adev);
+
+	return -EINVAL;
+}
+
+static int amdgpu_dm_early_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	amdgpu_dm_audio_fini(adev);
+
+	return 0;
+}
+
+static void amdgpu_dm_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (adev->dm.vblank_control_workqueue) {
+		destroy_workqueue(adev->dm.vblank_control_workqueue);
+		adev->dm.vblank_control_workqueue = NULL;
+	}
+
+	if (adev->dm.idle_workqueue) {
+		if (adev->dm.idle_workqueue->running) {
+			adev->dm.idle_workqueue->enable = false;
+			flush_work(&adev->dm.idle_workqueue->work);
+		}
+
+		kfree(adev->dm.idle_workqueue);
+		adev->dm.idle_workqueue = NULL;
+	}
+
+	amdgpu_dm_destroy_drm_device(&adev->dm);
+
+#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY)
+	if (adev->dm.secure_display_ctx.crtc_ctx) {
+		for (i = 0; i < adev->mode_info.num_crtc; i++) {
+			if (adev->dm.secure_display_ctx.crtc_ctx[i].crtc) {
+				flush_work(&adev->dm.secure_display_ctx.crtc_ctx[i].notify_ta_work);
+				flush_work(&adev->dm.secure_display_ctx.crtc_ctx[i].forward_roi_work);
+			}
+		}
+		kfree(adev->dm.secure_display_ctx.crtc_ctx);
+		adev->dm.secure_display_ctx.crtc_ctx = NULL;
+	}
+#endif
+	if (adev->dm.hdcp_workqueue) {
+		hdcp_destroy(&adev->dev->kobj, adev->dm.hdcp_workqueue);
+		adev->dm.hdcp_workqueue = NULL;
+	}
+
+	if (adev->dm.dc) {
+		dc_deinit_callbacks(adev->dm.dc);
+		dc_dmub_srv_destroy(&adev->dm.dc->ctx->dmub_srv);
+		if (dc_enable_dmub_notifications(adev->dm.dc)) {
+			kfree(adev->dm.dmub_notify);
+			adev->dm.dmub_notify = NULL;
+			destroy_workqueue(adev->dm.delayed_hpd_wq);
+			adev->dm.delayed_hpd_wq = NULL;
+		}
+	}
+
+	if (adev->dm.dmub_bo)
+		amdgpu_bo_free_kernel(&adev->dm.dmub_bo,
+				      &adev->dm.dmub_bo_gpu_addr,
+				      &adev->dm.dmub_bo_cpu_addr);
+
+	if (adev->dm.hpd_rx_offload_wq && adev->dm.dc) {
+		for (i = 0; i < adev->dm.dc->caps.max_links; i++) {
+			if (adev->dm.hpd_rx_offload_wq[i].wq) {
+				destroy_workqueue(adev->dm.hpd_rx_offload_wq[i].wq);
+				adev->dm.hpd_rx_offload_wq[i].wq = NULL;
+			}
+		}
+
+		kfree(adev->dm.hpd_rx_offload_wq);
+		adev->dm.hpd_rx_offload_wq = NULL;
+	}
+
+	/* DC Destroy TODO: Replace destroy DAL */
+	if (adev->dm.dc)
+		dc_destroy(&adev->dm.dc);
+	/*
+	 * TODO: pageflip, vlank interrupt
+	 *
+	 * amdgpu_dm_irq_fini(adev);
+	 */
+
+	if (adev->dm.cgs_device) {
+		amdgpu_cgs_destroy_device(adev->dm.cgs_device);
+		adev->dm.cgs_device = NULL;
+	}
+	if (adev->dm.freesync_module) {
+		mod_freesync_destroy(adev->dm.freesync_module);
+		adev->dm.freesync_module = NULL;
+	}
+
+	mutex_destroy(&adev->dm.audio_lock);
+	mutex_destroy(&adev->dm.dc_lock);
+	mutex_destroy(&adev->dm.dpia_aux_lock);
+}
+
+static int load_dmcu_fw(struct amdgpu_device *adev)
+{
+	const char *fw_name_dmcu = NULL;
+	int r;
+	const struct dmcu_firmware_header_v1_0 *hdr;
+
+	switch (adev->asic_type) {
+#if defined(CONFIG_DRM_AMD_DC_SI)
+	case CHIP_TAHITI:
+	case CHIP_PITCAIRN:
+	case CHIP_VERDE:
+	case CHIP_OLAND:
+#endif
+	case CHIP_BONAIRE:
+	case CHIP_HAWAII:
+	case CHIP_KAVERI:
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+	case CHIP_TONGA:
+	case CHIP_FIJI:
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
+	case CHIP_VEGA10:
+	case CHIP_VEGA12:
+	case CHIP_VEGA20:
+		return 0;
+	case CHIP_NAVI12:
+		fw_name_dmcu = FIRMWARE_NAVI12_DMCU;
+		break;
+	case CHIP_RAVEN:
+		if (ASICREV_IS_PICASSO(adev->external_rev_id))
+			fw_name_dmcu = FIRMWARE_RAVEN_DMCU;
+		else if (ASICREV_IS_RAVEN2(adev->external_rev_id))
+			fw_name_dmcu = FIRMWARE_RAVEN_DMCU;
+		else
+			return 0;
+		break;
+	default:
+		switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) {
+		case IP_VERSION(2, 0, 2):
+		case IP_VERSION(2, 0, 3):
+		case IP_VERSION(2, 0, 0):
+		case IP_VERSION(2, 1, 0):
+		case IP_VERSION(3, 0, 0):
+		case IP_VERSION(3, 0, 2):
+		case IP_VERSION(3, 0, 3):
+		case IP_VERSION(3, 0, 1):
+		case IP_VERSION(3, 1, 2):
+		case IP_VERSION(3, 1, 3):
+		case IP_VERSION(3, 1, 4):
+		case IP_VERSION(3, 1, 5):
+		case IP_VERSION(3, 1, 6):
+		case IP_VERSION(3, 2, 0):
+		case IP_VERSION(3, 2, 1):
+		case IP_VERSION(3, 5, 0):
+		case IP_VERSION(3, 5, 1):
+		case IP_VERSION(3, 6, 0):
+		case IP_VERSION(4, 0, 1):
+			return 0;
+		default:
+			break;
+		}
+		drm_err(adev_to_drm(adev), "Unsupported ASIC type: 0x%X\n", adev->asic_type);
+		return -EINVAL;
+	}
+
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
+		DRM_DEBUG_KMS("dm: DMCU firmware not supported on direct or SMU loading\n");
+		return 0;
+	}
+
+	r = amdgpu_ucode_request(adev, &adev->dm.fw_dmcu, AMDGPU_UCODE_REQUIRED,
+				 "%s", fw_name_dmcu);
+	if (r == -ENODEV) {
+		/* DMCU firmware is not necessary, so don't raise a fuss if it's missing */
+		DRM_DEBUG_KMS("dm: DMCU firmware not found\n");
+		adev->dm.fw_dmcu = NULL;
+		return 0;
+	}
+	if (r) {
+		drm_err(adev_to_drm(adev), "amdgpu_dm: Can't validate firmware \"%s\"\n",
+			fw_name_dmcu);
+		amdgpu_ucode_release(&adev->dm.fw_dmcu);
+		return r;
+	}
+
+	hdr = (const struct dmcu_firmware_header_v1_0 *)adev->dm.fw_dmcu->data;
+	adev->firmware.ucode[AMDGPU_UCODE_ID_DMCU_ERAM].ucode_id = AMDGPU_UCODE_ID_DMCU_ERAM;
+	adev->firmware.ucode[AMDGPU_UCODE_ID_DMCU_ERAM].fw = adev->dm.fw_dmcu;
+	adev->firmware.fw_size +=
+		ALIGN(le32_to_cpu(hdr->header.ucode_size_bytes) - le32_to_cpu(hdr->intv_size_bytes), PAGE_SIZE);
+
+	adev->firmware.ucode[AMDGPU_UCODE_ID_DMCU_INTV].ucode_id = AMDGPU_UCODE_ID_DMCU_INTV;
+	adev->firmware.ucode[AMDGPU_UCODE_ID_DMCU_INTV].fw = adev->dm.fw_dmcu;
+	adev->firmware.fw_size +=
+		ALIGN(le32_to_cpu(hdr->intv_size_bytes), PAGE_SIZE);
+
+	adev->dm.dmcu_fw_version = le32_to_cpu(hdr->header.ucode_version);
+
+	DRM_DEBUG_KMS("PSP loading DMCU firmware\n");
+
+	return 0;
+}
+
+static uint32_t amdgpu_dm_dmub_reg_read(void *ctx, uint32_t address)
+{
+	struct amdgpu_device *adev = ctx;
+
+	return dm_read_reg(adev->dm.dc->ctx, address);
+}
+
+static void amdgpu_dm_dmub_reg_write(void *ctx, uint32_t address,
+				     uint32_t value)
+{
+	struct amdgpu_device *adev = ctx;
+
+	return dm_write_reg(adev->dm.dc->ctx, address, value);
+}
+
+static int dm_dmub_sw_init(struct amdgpu_device *adev)
+{
+	struct dmub_srv_create_params create_params;
+	struct dmub_srv_region_params region_params;
+	struct dmub_srv_region_info region_info;
+	struct dmub_srv_memory_params memory_params;
+	struct dmub_srv_fb_info *fb_info;
+	struct dmub_srv *dmub_srv;
+	const struct dmcub_firmware_header_v1_0 *hdr;
+	enum dmub_asic dmub_asic;
+	enum dmub_status status;
+	static enum dmub_window_memory_type window_memory_type[DMUB_WINDOW_TOTAL] = {
+		DMUB_WINDOW_MEMORY_TYPE_FB,		//DMUB_WINDOW_0_INST_CONST
+		DMUB_WINDOW_MEMORY_TYPE_FB,		//DMUB_WINDOW_1_STACK
+		DMUB_WINDOW_MEMORY_TYPE_FB,		//DMUB_WINDOW_2_BSS_DATA
+		DMUB_WINDOW_MEMORY_TYPE_FB,		//DMUB_WINDOW_3_VBIOS
+		DMUB_WINDOW_MEMORY_TYPE_FB,		//DMUB_WINDOW_4_MAILBOX
+		DMUB_WINDOW_MEMORY_TYPE_FB,		//DMUB_WINDOW_5_TRACEBUFF
+		DMUB_WINDOW_MEMORY_TYPE_FB,		//DMUB_WINDOW_6_FW_STATE
+		DMUB_WINDOW_MEMORY_TYPE_FB,		//DMUB_WINDOW_7_SCRATCH_MEM
+		DMUB_WINDOW_MEMORY_TYPE_FB,		//DMUB_WINDOW_IB_MEM
+		DMUB_WINDOW_MEMORY_TYPE_FB,		//DMUB_WINDOW_SHARED_STATE
+	};
+	int r;
+
+	switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) {
+	case IP_VERSION(2, 1, 0):
+		dmub_asic = DMUB_ASIC_DCN21;
+		break;
+	case IP_VERSION(3, 0, 0):
+		dmub_asic = DMUB_ASIC_DCN30;
+		break;
+	case IP_VERSION(3, 0, 1):
+		dmub_asic = DMUB_ASIC_DCN301;
+		break;
+	case IP_VERSION(3, 0, 2):
+		dmub_asic = DMUB_ASIC_DCN302;
+		break;
+	case IP_VERSION(3, 0, 3):
+		dmub_asic = DMUB_ASIC_DCN303;
+		break;
+	case IP_VERSION(3, 1, 2):
+	case IP_VERSION(3, 1, 3):
+		dmub_asic = (adev->external_rev_id == YELLOW_CARP_B0) ? DMUB_ASIC_DCN31B : DMUB_ASIC_DCN31;
+		break;
+	case IP_VERSION(3, 1, 4):
+		dmub_asic = DMUB_ASIC_DCN314;
+		break;
+	case IP_VERSION(3, 1, 5):
+		dmub_asic = DMUB_ASIC_DCN315;
+		break;
+	case IP_VERSION(3, 1, 6):
+		dmub_asic = DMUB_ASIC_DCN316;
+		break;
+	case IP_VERSION(3, 2, 0):
+		dmub_asic = DMUB_ASIC_DCN32;
+		break;
+	case IP_VERSION(3, 2, 1):
+		dmub_asic = DMUB_ASIC_DCN321;
+		break;
+	case IP_VERSION(3, 5, 0):
+	case IP_VERSION(3, 5, 1):
+		dmub_asic = DMUB_ASIC_DCN35;
+		break;
+	case IP_VERSION(3, 6, 0):
+		dmub_asic = DMUB_ASIC_DCN36;
+		break;
+	case IP_VERSION(4, 0, 1):
+		dmub_asic = DMUB_ASIC_DCN401;
+		break;
+
+	default:
+		/* ASIC doesn't support DMUB. */
+		return 0;
+	}
+
+	hdr = (const struct dmcub_firmware_header_v1_0 *)adev->dm.dmub_fw->data;
+	adev->dm.dmcub_fw_version = le32_to_cpu(hdr->header.ucode_version);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		adev->firmware.ucode[AMDGPU_UCODE_ID_DMCUB].ucode_id =
+			AMDGPU_UCODE_ID_DMCUB;
+		adev->firmware.ucode[AMDGPU_UCODE_ID_DMCUB].fw =
+			adev->dm.dmub_fw;
+		adev->firmware.fw_size +=
+			ALIGN(le32_to_cpu(hdr->inst_const_bytes), PAGE_SIZE);
+
+		drm_info(adev_to_drm(adev), "Loading DMUB firmware via PSP: version=0x%08X\n",
+			 adev->dm.dmcub_fw_version);
+	}
+
+
+	adev->dm.dmub_srv = kzalloc(sizeof(*adev->dm.dmub_srv), GFP_KERNEL);
+	dmub_srv = adev->dm.dmub_srv;
+
+	if (!dmub_srv) {
+		drm_err(adev_to_drm(adev), "Failed to allocate DMUB service!\n");
+		return -ENOMEM;
+	}
+
+	memset(&create_params, 0, sizeof(create_params));
+	create_params.user_ctx = adev;
+	create_params.funcs.reg_read = amdgpu_dm_dmub_reg_read;
+	create_params.funcs.reg_write = amdgpu_dm_dmub_reg_write;
+	create_params.asic = dmub_asic;
+
+	/* Create the DMUB service. */
+	status = dmub_srv_create(dmub_srv, &create_params);
+	if (status != DMUB_STATUS_OK) {
+		drm_err(adev_to_drm(adev), "Error creating DMUB service: %d\n", status);
+		return -EINVAL;
+	}
+
+	/* Calculate the size of all the regions for the DMUB service. */
+	memset(&region_params, 0, sizeof(region_params));
+
+	region_params.inst_const_size = le32_to_cpu(hdr->inst_const_bytes) -
+					PSP_HEADER_BYTES - PSP_FOOTER_BYTES;
+	region_params.bss_data_size = le32_to_cpu(hdr->bss_data_bytes);
+	region_params.vbios_size = adev->bios_size;
+	region_params.fw_bss_data = region_params.bss_data_size ?
+		adev->dm.dmub_fw->data +
+		le32_to_cpu(hdr->header.ucode_array_offset_bytes) +
+		le32_to_cpu(hdr->inst_const_bytes) : NULL;
+	region_params.fw_inst_const =
+		adev->dm.dmub_fw->data +
+		le32_to_cpu(hdr->header.ucode_array_offset_bytes) +
+		PSP_HEADER_BYTES;
+	region_params.window_memory_type = window_memory_type;
+
+	status = dmub_srv_calc_region_info(dmub_srv, &region_params,
+					   &region_info);
+
+	if (status != DMUB_STATUS_OK) {
+		drm_err(adev_to_drm(adev), "Error calculating DMUB region info: %d\n", status);
+		return -EINVAL;
+	}
+
+	/*
+	 * Allocate a framebuffer based on the total size of all the regions.
+	 * TODO: Move this into GART.
+	 */
+	r = amdgpu_bo_create_kernel(adev, region_info.fb_size, PAGE_SIZE,
+				    AMDGPU_GEM_DOMAIN_VRAM |
+				    AMDGPU_GEM_DOMAIN_GTT,
+				    &adev->dm.dmub_bo,
+				    &adev->dm.dmub_bo_gpu_addr,
+				    &adev->dm.dmub_bo_cpu_addr);
+	if (r)
+		return r;
+
+	/* Rebase the regions on the framebuffer address. */
+	memset(&memory_params, 0, sizeof(memory_params));
+	memory_params.cpu_fb_addr = adev->dm.dmub_bo_cpu_addr;
+	memory_params.gpu_fb_addr = adev->dm.dmub_bo_gpu_addr;
+	memory_params.region_info = &region_info;
+	memory_params.window_memory_type = window_memory_type;
+
+	adev->dm.dmub_fb_info =
+		kzalloc(sizeof(*adev->dm.dmub_fb_info), GFP_KERNEL);
+	fb_info = adev->dm.dmub_fb_info;
+
+	if (!fb_info) {
+		drm_err(adev_to_drm(adev),
+			"Failed to allocate framebuffer info for DMUB service!\n");
+		return -ENOMEM;
+	}
+
+	status = dmub_srv_calc_mem_info(dmub_srv, &memory_params, fb_info);
+	if (status != DMUB_STATUS_OK) {
+		drm_err(adev_to_drm(adev), "Error calculating DMUB FB info: %d\n", status);
+		return -EINVAL;
+	}
+
+	adev->dm.bb_from_dmub = dm_dmub_get_vbios_bounding_box(adev);
+
+	return 0;
+}
+
+static int dm_sw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	adev->dm.cgs_device = amdgpu_cgs_create_device(adev);
+
+	if (!adev->dm.cgs_device) {
+		drm_err(adev_to_drm(adev), "failed to create cgs device.\n");
+		return -EINVAL;
+	}
+
+	/* Moved from dm init since we need to use allocations for storing bounding box data */
+	INIT_LIST_HEAD(&adev->dm.da_list);
+
+	r = dm_dmub_sw_init(adev);
+	if (r)
+		return r;
+
+	return load_dmcu_fw(adev);
+}
+
+static int dm_sw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct dal_allocation *da;
+
+	list_for_each_entry(da, &adev->dm.da_list, list) {
+		if (adev->dm.bb_from_dmub == (void *) da->cpu_ptr) {
+			amdgpu_bo_free_kernel(&da->bo, &da->gpu_addr, &da->cpu_ptr);
+			list_del(&da->list);
+			kfree(da);
+			adev->dm.bb_from_dmub = NULL;
+			break;
+		}
+	}
+
+
+	kfree(adev->dm.dmub_fb_info);
+	adev->dm.dmub_fb_info = NULL;
+
+	if (adev->dm.dmub_srv) {
+		dmub_srv_destroy(adev->dm.dmub_srv);
+		kfree(adev->dm.dmub_srv);
+		adev->dm.dmub_srv = NULL;
+	}
+
+	amdgpu_ucode_release(&adev->dm.dmub_fw);
+	amdgpu_ucode_release(&adev->dm.fw_dmcu);
+
+	return 0;
+}
+
+static int detect_mst_link_for_all_connectors(struct drm_device *dev)
+{
+	struct amdgpu_dm_connector *aconnector;
+	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
+	int ret = 0;
+
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_WRITEBACK)
+			continue;
+
+		aconnector = to_amdgpu_dm_connector(connector);
+		if (aconnector->dc_link->type == dc_connection_mst_branch &&
+		    aconnector->mst_mgr.aux) {
+			drm_dbg_kms(dev, "DM_MST: starting TM on aconnector: %p [id: %d]\n",
+					 aconnector,
+					 aconnector->base.base.id);
+
+			ret = drm_dp_mst_topology_mgr_set_mst(&aconnector->mst_mgr, true);
+			if (ret < 0) {
+				drm_err(dev, "DM_MST: Failed to start MST\n");
+				aconnector->dc_link->type =
+					dc_connection_single;
+				ret = dm_helpers_dp_mst_stop_top_mgr(aconnector->dc_link->ctx,
+								     aconnector->dc_link);
+				break;
+			}
+		}
+	}
+	drm_connector_list_iter_end(&iter);
+
+	return ret;
+}
+
+static int dm_late_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	struct dmcu_iram_parameters params;
+	unsigned int linear_lut[16];
+	int i;
+	struct dmcu *dmcu = NULL;
+
+	dmcu = adev->dm.dc->res_pool->dmcu;
+
+	for (i = 0; i < 16; i++)
+		linear_lut[i] = 0xFFFF * i / 15;
+
+	params.set = 0;
+	params.backlight_ramping_override = false;
+	params.backlight_ramping_start = 0xCCCC;
+	params.backlight_ramping_reduction = 0xCCCCCCCC;
+	params.backlight_lut_array_size = 16;
+	params.backlight_lut_array = linear_lut;
+
+	/* Min backlight level after ABM reduction,  Don't allow below 1%
+	 * 0xFFFF x 0.01 = 0x28F
+	 */
+	params.min_abm_backlight = 0x28F;
+	/* In the case where abm is implemented on dmcub,
+	 * dmcu object will be null.
+	 * ABM 2.4 and up are implemented on dmcub.
+	 */
+	if (dmcu) {
+		if (!dmcu_load_iram(dmcu, params))
+			return -EINVAL;
+	} else if (adev->dm.dc->ctx->dmub_srv) {
+		struct dc_link *edp_links[MAX_NUM_EDP];
+		int edp_num;
+
+		dc_get_edp_links(adev->dm.dc, edp_links, &edp_num);
+		for (i = 0; i < edp_num; i++) {
+			if (!dmub_init_abm_config(adev->dm.dc->res_pool, params, i))
+				return -EINVAL;
+		}
+	}
+
+	return detect_mst_link_for_all_connectors(adev_to_drm(adev));
+}
+
+static void resume_mst_branch_status(struct drm_dp_mst_topology_mgr *mgr)
+{
+	u8 buf[UUID_SIZE];
+	guid_t guid;
+	int ret;
+
+	mutex_lock(&mgr->lock);
+	if (!mgr->mst_primary)
+		goto out_fail;
+
+	if (drm_dp_read_dpcd_caps(mgr->aux, mgr->dpcd) < 0) {
+		drm_dbg_kms(mgr->dev, "dpcd read failed - undocked during suspend?\n");
+		goto out_fail;
+	}
+
+	ret = drm_dp_dpcd_writeb(mgr->aux, DP_MSTM_CTRL,
+				 DP_MST_EN |
+				 DP_UP_REQ_EN |
+				 DP_UPSTREAM_IS_SRC);
+	if (ret < 0) {
+		drm_dbg_kms(mgr->dev, "mst write failed - undocked during suspend?\n");
+		goto out_fail;
+	}
+
+	/* Some hubs forget their guids after they resume */
+	ret = drm_dp_dpcd_read(mgr->aux, DP_GUID, buf, sizeof(buf));
+	if (ret != sizeof(buf)) {
+		drm_dbg_kms(mgr->dev, "dpcd read failed - undocked during suspend?\n");
+		goto out_fail;
+	}
+
+	import_guid(&guid, buf);
+
+	if (guid_is_null(&guid)) {
+		guid_gen(&guid);
+		export_guid(buf, &guid);
+
+		ret = drm_dp_dpcd_write(mgr->aux, DP_GUID, buf, sizeof(buf));
+
+		if (ret != sizeof(buf)) {
+			drm_dbg_kms(mgr->dev, "check mstb guid failed - undocked during suspend?\n");
+			goto out_fail;
+		}
+	}
+
+	guid_copy(&mgr->mst_primary->guid, &guid);
+
+out_fail:
+	mutex_unlock(&mgr->lock);
+}
+
+void hdmi_cec_unset_edid(struct amdgpu_dm_connector *aconnector)
+{
+	struct cec_notifier *n = aconnector->notifier;
+
+	if (!n)
+		return;
+
+	cec_notifier_phys_addr_invalidate(n);
+}
+
+void hdmi_cec_set_edid(struct amdgpu_dm_connector *aconnector)
+{
+	struct drm_connector *connector = &aconnector->base;
+	struct cec_notifier *n = aconnector->notifier;
+
+	if (!n)
+		return;
+
+	cec_notifier_set_phys_addr(n,
+				   connector->display_info.source_physical_address);
+}
+
+static void s3_handle_hdmi_cec(struct drm_device *ddev, bool suspend)
+{
+	struct amdgpu_dm_connector *aconnector;
+	struct drm_connector *connector;
+	struct drm_connector_list_iter conn_iter;
+
+	drm_connector_list_iter_begin(ddev, &conn_iter);
+	drm_for_each_connector_iter(connector, &conn_iter) {
+		if (connector->connector_type == DRM_MODE_CONNECTOR_WRITEBACK)
+			continue;
+
+		aconnector = to_amdgpu_dm_connector(connector);
+		if (suspend)
+			hdmi_cec_unset_edid(aconnector);
+		else
+			hdmi_cec_set_edid(aconnector);
+	}
+	drm_connector_list_iter_end(&conn_iter);
+}
+
+static void s3_handle_mst(struct drm_device *dev, bool suspend)
+{
+	struct amdgpu_dm_connector *aconnector;
+	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
+	struct drm_dp_mst_topology_mgr *mgr;
+
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_WRITEBACK)
+			continue;
+
+		aconnector = to_amdgpu_dm_connector(connector);
+		if (aconnector->dc_link->type != dc_connection_mst_branch ||
+		    aconnector->mst_root)
+			continue;
+
+		mgr = &aconnector->mst_mgr;
+
+		if (suspend) {
+			drm_dp_mst_topology_mgr_suspend(mgr);
+		} else {
+			/* if extended timeout is supported in hardware,
+			 * default to LTTPR timeout (3.2ms) first as a W/A for DP link layer
+			 * CTS 4.2.1.1 regression introduced by CTS specs requirement update.
+			 */
+			try_to_configure_aux_timeout(aconnector->dc_link->ddc, LINK_AUX_DEFAULT_LTTPR_TIMEOUT_PERIOD);
+			if (!dp_is_lttpr_present(aconnector->dc_link))
+				try_to_configure_aux_timeout(aconnector->dc_link->ddc, LINK_AUX_DEFAULT_TIMEOUT_PERIOD);
+
+			/* TODO: move resume_mst_branch_status() into drm mst resume again
+			 * once topology probing work is pulled out from mst resume into mst
+			 * resume 2nd step. mst resume 2nd step should be called after old
+			 * state getting restored (i.e. drm_atomic_helper_resume()).
+			 */
+			resume_mst_branch_status(mgr);
+		}
+	}
+	drm_connector_list_iter_end(&iter);
+}
+
+static int amdgpu_dm_smu_write_watermarks_table(struct amdgpu_device *adev)
+{
+	int ret = 0;
+
+	/* This interface is for dGPU Navi1x.Linux dc-pplib interface depends
+	 * on window driver dc implementation.
+	 * For Navi1x, clock settings of dcn watermarks are fixed. the settings
+	 * should be passed to smu during boot up and resume from s3.
+	 * boot up: dc calculate dcn watermark clock settings within dc_create,
+	 * dcn20_resource_construct
+	 * then call pplib functions below to pass the settings to smu:
+	 * smu_set_watermarks_for_clock_ranges
+	 * smu_set_watermarks_table
+	 * navi10_set_watermarks_table
+	 * smu_write_watermarks_table
+	 *
+	 * For Renoir, clock settings of dcn watermark are also fixed values.
+	 * dc has implemented different flow for window driver:
+	 * dc_hardware_init / dc_set_power_state
+	 * dcn10_init_hw
+	 * notify_wm_ranges
+	 * set_wm_ranges
+	 * -- Linux
+	 * smu_set_watermarks_for_clock_ranges
+	 * renoir_set_watermarks_table
+	 * smu_write_watermarks_table
+	 *
+	 * For Linux,
+	 * dc_hardware_init -> amdgpu_dm_init
+	 * dc_set_power_state --> dm_resume
+	 *
+	 * therefore, this function apply to navi10/12/14 but not Renoir
+	 * *
+	 */
+	switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) {
+	case IP_VERSION(2, 0, 2):
+	case IP_VERSION(2, 0, 0):
+		break;
+	default:
+		return 0;
+	}
+
+	ret = amdgpu_dpm_write_watermarks_table(adev);
+	if (ret) {
+		drm_err(adev_to_drm(adev), "Failed to update WMTABLE!\n");
+		return ret;
+	}
+
+	return 0;
+}
+
+static int dm_oem_i2c_hw_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_display_manager *dm = &adev->dm;
+	struct amdgpu_i2c_adapter *oem_i2c;
+	struct ddc_service *oem_ddc_service;
+	int r;
+
+	oem_ddc_service = dc_get_oem_i2c_device(adev->dm.dc);
+	if (oem_ddc_service) {
+		oem_i2c = create_i2c(oem_ddc_service, true);
+		if (!oem_i2c) {
+			drm_info(adev_to_drm(adev), "Failed to create oem i2c adapter data\n");
+			return -ENOMEM;
+		}
+
+		r = devm_i2c_add_adapter(adev->dev, &oem_i2c->base);
+		if (r) {
+			drm_info(adev_to_drm(adev), "Failed to register oem i2c\n");
+			kfree(oem_i2c);
+			return r;
+		}
+		dm->oem_i2c = oem_i2c;
+	}
+
+	return 0;
+}
+
+/**
+ * dm_hw_init() - Initialize DC device
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Initialize the &struct amdgpu_display_manager device. This involves calling
+ * the initializers of each DM component, then populating the struct with them.
+ *
+ * Although the function implies hardware initialization, both hardware and
+ * software are initialized here. Splitting them out to their relevant init
+ * hooks is a future TODO item.
+ *
+ * Some notable things that are initialized here:
+ *
+ * - Display Core, both software and hardware
+ * - DC modules that we need (freesync and color management)
+ * - DRM software states
+ * - Interrupt sources and handlers
+ * - Vblank support
+ * - Debug FS entries, if enabled
+ */
+static int dm_hw_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int r;
+
+	/* Create DAL display manager */
+	r = amdgpu_dm_init(adev);
+	if (r)
+		return r;
+	amdgpu_dm_hpd_init(adev);
+
+	r = dm_oem_i2c_hw_init(adev);
+	if (r)
+		drm_info(adev_to_drm(adev), "Failed to add OEM i2c bus\n");
+
+	return 0;
+}
+
+/**
+ * dm_hw_fini() - Teardown DC device
+ * @ip_block: Pointer to the amdgpu_ip_block for this hw instance.
+ *
+ * Teardown components within &struct amdgpu_display_manager that require
+ * cleanup. This involves cleaning up the DRM device, DC, and any modules that
+ * were loaded. Also flush IRQ workqueues and disable them.
+ */
+static int dm_hw_fini(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+
+	amdgpu_dm_hpd_fini(adev);
+
+	amdgpu_dm_irq_fini(adev);
+	amdgpu_dm_fini(adev);
+	return 0;
+}
+
+
+static void dm_gpureset_toggle_interrupts(struct amdgpu_device *adev,
+				 struct dc_state *state, bool enable)
+{
+	enum dc_irq_source irq_source;
+	struct amdgpu_crtc *acrtc;
+	int rc = -EBUSY;
+	int i = 0;
+
+	for (i = 0; i < state->stream_count; i++) {
+		acrtc = get_crtc_by_otg_inst(
+				adev, state->stream_status[i].primary_otg_inst);
+
+		if (acrtc && state->stream_status[i].plane_count != 0) {
+			irq_source = IRQ_TYPE_PFLIP + acrtc->otg_inst;
+			rc = dc_interrupt_set(adev->dm.dc, irq_source, enable) ? 0 : -EBUSY;
+			if (rc)
+				drm_warn(adev_to_drm(adev), "Failed to %s pflip interrupts\n",
+					 enable ? "enable" : "disable");
+
+			if (dc_supports_vrr(adev->dm.dc->ctx->dce_version)) {
+				if (enable) {
+					if (amdgpu_dm_crtc_vrr_active(
+							to_dm_crtc_state(acrtc->base.state)))
+						rc = amdgpu_dm_crtc_set_vupdate_irq(
+							&acrtc->base, true);
+				} else
+					rc = amdgpu_dm_crtc_set_vupdate_irq(
+							&acrtc->base, false);
+
+				if (rc)
+					drm_warn(adev_to_drm(adev), "Failed to %sable vupdate interrupt\n",
+						enable ? "en" : "dis");
+			}
+
+			irq_source = IRQ_TYPE_VBLANK + acrtc->otg_inst;
+			/* During gpu-reset we disable and then enable vblank irq, so
+			 * don't use amdgpu_irq_get/put() to avoid refcount change.
+			 */
+			if (!dc_interrupt_set(adev->dm.dc, irq_source, enable))
+				drm_warn(adev_to_drm(adev), "Failed to %sable vblank interrupt\n", enable ? "en" : "dis");
+		}
+	}
+
+}
+
+DEFINE_FREE(state_release, struct dc_state *, if (_T) dc_state_release(_T))
+
+static enum dc_status amdgpu_dm_commit_zero_streams(struct dc *dc)
+{
+	struct dc_state *context __free(state_release) = NULL;
+	int i;
+	struct dc_stream_state *del_streams[MAX_PIPES];
+	int del_streams_count = 0;
+	struct dc_commit_streams_params params = {};
+
+	memset(del_streams, 0, sizeof(del_streams));
+
+	context = dc_state_create_current_copy(dc);
+	if (context == NULL)
+		return DC_ERROR_UNEXPECTED;
+
+	/* First remove from context all streams */
+	for (i = 0; i < context->stream_count; i++) {
+		struct dc_stream_state *stream = context->streams[i];
+
+		del_streams[del_streams_count++] = stream;
+	}
+
+	/* Remove all planes for removed streams and then remove the streams */
+	for (i = 0; i < del_streams_count; i++) {
+		enum dc_status res;
+
+		if (!dc_state_rem_all_planes_for_stream(dc, del_streams[i], context))
+			return DC_FAIL_DETACH_SURFACES;
+
+		res = dc_state_remove_stream(dc, context, del_streams[i]);
+		if (res != DC_OK)
+			return res;
+	}
+
+	params.streams = context->streams;
+	params.stream_count = context->stream_count;
+
+	return dc_commit_streams(dc, &params);
+}
+
+static void hpd_rx_irq_work_suspend(struct amdgpu_display_manager *dm)
+{
+	int i;
+
+	if (dm->hpd_rx_offload_wq) {
+		for (i = 0; i < dm->dc->caps.max_links; i++)
+			flush_workqueue(dm->hpd_rx_offload_wq[i].wq);
+	}
+}
+
+static int dm_cache_state(struct amdgpu_device *adev)
+{
+	int r;
+
+	adev->dm.cached_state = drm_atomic_helper_suspend(adev_to_drm(adev));
+	if (IS_ERR(adev->dm.cached_state)) {
+		r = PTR_ERR(adev->dm.cached_state);
+		adev->dm.cached_state = NULL;
+	}
+
+	return adev->dm.cached_state ? 0 : r;
+}
+
+static void dm_destroy_cached_state(struct amdgpu_device *adev)
+{
+	struct amdgpu_display_manager *dm = &adev->dm;
+	struct drm_device *ddev = adev_to_drm(adev);
+	struct dm_plane_state *dm_new_plane_state;
+	struct drm_plane_state *new_plane_state;
+	struct dm_crtc_state *dm_new_crtc_state;
+	struct drm_crtc_state *new_crtc_state;
+	struct drm_plane *plane;
+	struct drm_crtc *crtc;
+	int i;
+
+	if (!dm->cached_state)
+		return;
+
+	/* Force mode set in atomic commit */
+	for_each_new_crtc_in_state(dm->cached_state, crtc, new_crtc_state, i) {
+		new_crtc_state->active_changed = true;
+		dm_new_crtc_state = to_dm_crtc_state(new_crtc_state);
+		reset_freesync_config_for_crtc(dm_new_crtc_state);
+	}
+
+	/*
+	 * atomic_check is expected to create the dc states. We need to release
+	 * them here, since they were duplicated as part of the suspend
+	 * procedure.
+	 */
+	for_each_new_crtc_in_state(dm->cached_state, crtc, new_crtc_state, i) {
+		dm_new_crtc_state = to_dm_crtc_state(new_crtc_state);
+		if (dm_new_crtc_state->stream) {
+			WARN_ON(kref_read(&dm_new_crtc_state->stream->refcount) > 1);
+			dc_stream_release(dm_new_crtc_state->stream);
+			dm_new_crtc_state->stream = NULL;
+		}
+		dm_new_crtc_state->base.color_mgmt_changed = true;
+	}
+
+	for_each_new_plane_in_state(dm->cached_state, plane, new_plane_state, i) {
+		dm_new_plane_state = to_dm_plane_state(new_plane_state);
+		if (dm_new_plane_state->dc_state) {
+			WARN_ON(kref_read(&dm_new_plane_state->dc_state->refcount) > 1);
+			dc_plane_state_release(dm_new_plane_state->dc_state);
+			dm_new_plane_state->dc_state = NULL;
+		}
+	}
+
+	drm_atomic_helper_resume(ddev, dm->cached_state);
+
+	dm->cached_state = NULL;
+}
+
+static int dm_suspend(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_display_manager *dm = &adev->dm;
+
+	if (amdgpu_in_reset(adev)) {
+		enum dc_status res;
+
+		mutex_lock(&dm->dc_lock);
+
+		dc_allow_idle_optimizations(adev->dm.dc, false);
+
+		dm->cached_dc_state = dc_state_create_copy(dm->dc->current_state);
+
+		if (dm->cached_dc_state)
+			dm_gpureset_toggle_interrupts(adev, dm->cached_dc_state, false);
+
+		res = amdgpu_dm_commit_zero_streams(dm->dc);
+		if (res != DC_OK) {
+			drm_err(adev_to_drm(adev), "Failed to commit zero streams: %d\n", res);
+			return -EINVAL;
+		}
+
+		amdgpu_dm_irq_suspend(adev);
+
+		hpd_rx_irq_work_suspend(dm);
+
+		return 0;
+	}
+
+	if (!adev->dm.cached_state) {
+		int r = dm_cache_state(adev);
+
+		if (r)
+			return r;
+	}
+
+	s3_handle_hdmi_cec(adev_to_drm(adev), true);
+
+	s3_handle_mst(adev_to_drm(adev), true);
+
+	amdgpu_dm_irq_suspend(adev);
+
+	hpd_rx_irq_work_suspend(dm);
+
+	dc_set_power_state(dm->dc, DC_ACPI_CM_POWER_STATE_D3);
+
+	if (dm->dc->caps.ips_support && adev->in_s0ix)
+		dc_allow_idle_optimizations(dm->dc, true);
+
+	dc_dmub_srv_set_power_state(dm->dc->ctx->dmub_srv, DC_ACPI_CM_POWER_STATE_D3);
+
+	return 0;
+}
+
+struct drm_connector *
+amdgpu_dm_find_first_crtc_matching_connector(struct drm_atomic_state *state,
+					     struct drm_crtc *crtc)
+{
+	u32 i;
+	struct drm_connector_state *new_con_state;
+	struct drm_connector *connector;
+	struct drm_crtc *crtc_from_state;
+
+	for_each_new_connector_in_state(state, connector, new_con_state, i) {
+		crtc_from_state = new_con_state->crtc;
+
+		if (crtc_from_state == crtc)
+			return connector;
+	}
+
+	return NULL;
+}
+
+static void emulated_link_detect(struct dc_link *link)
+{
+	struct dc_sink_init_data sink_init_data = { 0 };
+	struct display_sink_capability sink_caps = { 0 };
+	enum dc_edid_status edid_status;
+	struct dc_context *dc_ctx = link->ctx;
+	struct drm_device *dev = adev_to_drm(dc_ctx->driver_context);
+	struct dc_sink *sink = NULL;
+	struct dc_sink *prev_sink = NULL;
+
+	link->type = dc_connection_none;
+	prev_sink = link->local_sink;
+
+	if (prev_sink)
+		dc_sink_release(prev_sink);
+
+	switch (link->connector_signal) {
+	case SIGNAL_TYPE_HDMI_TYPE_A: {
+		sink_caps.transaction_type = DDC_TRANSACTION_TYPE_I2C;
+		sink_caps.signal = SIGNAL_TYPE_HDMI_TYPE_A;
+		break;
+	}
+
+	case SIGNAL_TYPE_DVI_SINGLE_LINK: {
+		sink_caps.transaction_type = DDC_TRANSACTION_TYPE_I2C;
+		sink_caps.signal = SIGNAL_TYPE_DVI_SINGLE_LINK;
+		break;
+	}
+
+	case SIGNAL_TYPE_DVI_DUAL_LINK: {
+		sink_caps.transaction_type = DDC_TRANSACTION_TYPE_I2C;
+		sink_caps.signal = SIGNAL_TYPE_DVI_DUAL_LINK;
+		break;
+	}
+
+	case SIGNAL_TYPE_LVDS: {
+		sink_caps.transaction_type = DDC_TRANSACTION_TYPE_I2C;
+		sink_caps.signal = SIGNAL_TYPE_LVDS;
+		break;
+	}
+
+	case SIGNAL_TYPE_EDP: {
+		sink_caps.transaction_type =
+			DDC_TRANSACTION_TYPE_I2C_OVER_AUX;
+		sink_caps.signal = SIGNAL_TYPE_EDP;
+		break;
+	}
+
+	case SIGNAL_TYPE_DISPLAY_PORT: {
+		sink_caps.transaction_type =
+			DDC_TRANSACTION_TYPE_I2C_OVER_AUX;
+		sink_caps.signal = SIGNAL_TYPE_VIRTUAL;
+		break;
+	}
+
+	default:
+		drm_err(dev, "Invalid connector type! signal:%d\n",
+			link->connector_signal);
+		return;
+	}
+
+	sink_init_data.link = link;
+	sink_init_data.sink_signal = sink_caps.signal;
+
+	sink = dc_sink_create(&sink_init_data);
+	if (!sink) {
+		drm_err(dev, "Failed to create sink!\n");
+		return;
+	}
+
+	/* dc_sink_create returns a new reference */
+	link->local_sink = sink;
+
+	edid_status = dm_helpers_read_local_edid(
+			link->ctx,
+			link,
+			sink);
+
+	if (edid_status != EDID_OK)
+		drm_err(dev, "Failed to read EDID\n");
+
+}
+
+static void dm_gpureset_commit_state(struct dc_state *dc_state,
+				     struct amdgpu_display_manager *dm)
+{
+	struct {
+		struct dc_surface_update surface_updates[MAX_SURFACES];
+		struct dc_plane_info plane_infos[MAX_SURFACES];
+		struct dc_scaling_info scaling_infos[MAX_SURFACES];
+		struct dc_flip_addrs flip_addrs[MAX_SURFACES];
+		struct dc_stream_update stream_update;
+	} *bundle __free(kfree);
+	int k, m;
+
+	bundle = kzalloc(sizeof(*bundle), GFP_KERNEL);
+
+	if (!bundle) {
+		drm_err(dm->ddev, "Failed to allocate update bundle\n");
+		return;
+	}
+
+	for (k = 0; k < dc_state->stream_count; k++) {
+		bundle->stream_update.stream = dc_state->streams[k];
+
+		for (m = 0; m < dc_state->stream_status[k].plane_count; m++) {
+			bundle->surface_updates[m].surface =
+				dc_state->stream_status[k].plane_states[m];
+			bundle->surface_updates[m].surface->force_full_update =
+				true;
+		}
+
+		update_planes_and_stream_adapter(dm->dc,
+					 UPDATE_TYPE_FULL,
+					 dc_state->stream_status[k].plane_count,
+					 dc_state->streams[k],
+					 &bundle->stream_update,
+					 bundle->surface_updates);
+	}
+}
+
+static void apply_delay_after_dpcd_poweroff(struct amdgpu_device *adev,
+					    struct dc_sink *sink)
+{
+	struct dc_panel_patch *ppatch = NULL;
+
+	if (!sink)
+		return;
+
+	ppatch = &sink->edid_caps.panel_patch;
+	if (ppatch->wait_after_dpcd_poweroff_ms) {
+		msleep(ppatch->wait_after_dpcd_poweroff_ms);
+		drm_dbg_driver(adev_to_drm(adev),
+			       "%s: adding a %ds delay as w/a for panel\n",
+			       __func__,
+			       ppatch->wait_after_dpcd_poweroff_ms / 1000);
+	}
+}
+
+static int dm_resume(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct drm_device *ddev = adev_to_drm(adev);
+	struct amdgpu_display_manager *dm = &adev->dm;
+	struct amdgpu_dm_connector *aconnector;
+	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
+	struct dm_atomic_state *dm_state = to_dm_atomic_state(dm->atomic_obj.state);
+	enum dc_connection_type new_connection_type = dc_connection_none;
+	struct dc_state *dc_state;
+	int i, r, j;
+	struct dc_commit_streams_params commit_params = {};
+
+	if (dm->dc->caps.ips_support) {
+		dc_dmub_srv_apply_idle_power_optimizations(dm->dc, false);
+	}
+
+	if (amdgpu_in_reset(adev)) {
+		dc_state = dm->cached_dc_state;
+
+		/*
+		 * The dc->current_state is backed up into dm->cached_dc_state
+		 * before we commit 0 streams.
+		 *
+		 * DC will clear link encoder assignments on the real state
+		 * but the changes won't propagate over to the copy we made
+		 * before the 0 streams commit.
+		 *
+		 * DC expects that link encoder assignments are *not* valid
+		 * when committing a state, so as a workaround we can copy
+		 * off of the current state.
+		 *
+		 * We lose the previous assignments, but we had already
+		 * commit 0 streams anyway.
+		 */
+		link_enc_cfg_copy(adev->dm.dc->current_state, dc_state);
+
+		r = dm_dmub_hw_init(adev);
+		if (r) {
+			drm_err(adev_to_drm(adev), "DMUB interface failed to initialize: status=%d\n", r);
+			return r;
+		}
+
+		dc_dmub_srv_set_power_state(dm->dc->ctx->dmub_srv, DC_ACPI_CM_POWER_STATE_D0);
+		dc_set_power_state(dm->dc, DC_ACPI_CM_POWER_STATE_D0);
+
+		dc_resume(dm->dc);
+		adev->dm.restore_backlight = true;
+
+		amdgpu_dm_irq_resume_early(adev);
+
+		for (i = 0; i < dc_state->stream_count; i++) {
+			dc_state->streams[i]->mode_changed = true;
+			for (j = 0; j < dc_state->stream_status[i].plane_count; j++) {
+				dc_state->stream_status[i].plane_states[j]->update_flags.raw
+					= 0xffffffff;
+			}
+		}
+
+		if (dc_is_dmub_outbox_supported(adev->dm.dc)) {
+			amdgpu_dm_outbox_init(adev);
+			dc_enable_dmub_outbox(adev->dm.dc);
+		}
+
+		commit_params.streams = dc_state->streams;
+		commit_params.stream_count = dc_state->stream_count;
+		dc_exit_ips_for_hw_access(dm->dc);
+		WARN_ON(!dc_commit_streams(dm->dc, &commit_params));
+
+		dm_gpureset_commit_state(dm->cached_dc_state, dm);
+
+		dm_gpureset_toggle_interrupts(adev, dm->cached_dc_state, true);
+
+		dc_state_release(dm->cached_dc_state);
+		dm->cached_dc_state = NULL;
+
+		amdgpu_dm_irq_resume_late(adev);
+
+		mutex_unlock(&dm->dc_lock);
+
+		/* set the backlight after a reset */
+		for (i = 0; i < dm->num_of_edps; i++) {
+			if (dm->backlight_dev[i])
+				amdgpu_dm_backlight_set_level(dm, i, dm->brightness[i]);
+		}
+
+		return 0;
+	}
+	/* Recreate dc_state - DC invalidates it when setting power state to S3. */
+	dc_state_release(dm_state->context);
+	dm_state->context = dc_state_create(dm->dc, NULL);
+	/* TODO: Remove dc_state->dccg, use dc->dccg directly. */
+
+	/* Before powering on DC we need to re-initialize DMUB. */
+	dm_dmub_hw_resume(adev);
+
+	/* Re-enable outbox interrupts for DPIA. */
+	if (dc_is_dmub_outbox_supported(adev->dm.dc)) {
+		amdgpu_dm_outbox_init(adev);
+		dc_enable_dmub_outbox(adev->dm.dc);
+	}
+
+	/* power on hardware */
+	dc_dmub_srv_set_power_state(dm->dc->ctx->dmub_srv, DC_ACPI_CM_POWER_STATE_D0);
+	dc_set_power_state(dm->dc, DC_ACPI_CM_POWER_STATE_D0);
+
+	/* program HPD filter */
+	dc_resume(dm->dc);
+
+	/*
+	 * early enable HPD Rx IRQ, should be done before set mode as short
+	 * pulse interrupts are used for MST
+	 */
+	amdgpu_dm_irq_resume_early(adev);
+
+	s3_handle_hdmi_cec(ddev, false);
+
+	/* On resume we need to rewrite the MSTM control bits to enable MST*/
+	s3_handle_mst(ddev, false);
+
+	/* Do detection*/
+	drm_connector_list_iter_begin(ddev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
+		bool ret;
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_WRITEBACK)
+			continue;
+
+		aconnector = to_amdgpu_dm_connector(connector);
+
+		if (!aconnector->dc_link)
+			continue;
+
+		/*
+		 * this is the case when traversing through already created end sink
+		 * MST connectors, should be skipped
+		 */
+		if (aconnector->mst_root)
+			continue;
+
+		guard(mutex)(&aconnector->hpd_lock);
+		if (!dc_link_detect_connection_type(aconnector->dc_link, &new_connection_type))
+			drm_err(adev_to_drm(adev), "KMS: Failed to detect connector\n");
+
+		if (aconnector->base.force && new_connection_type == dc_connection_none) {
+			emulated_link_detect(aconnector->dc_link);
+		} else {
+			guard(mutex)(&dm->dc_lock);
+			dc_exit_ips_for_hw_access(dm->dc);
+			ret = dc_link_detect(aconnector->dc_link, DETECT_REASON_RESUMEFROMS3S4);
+			if (ret) {
+				/* w/a delay for certain panels */
+				apply_delay_after_dpcd_poweroff(adev, aconnector->dc_sink);
+			}
+		}
+
+		if (aconnector->fake_enable && aconnector->dc_link->local_sink)
+			aconnector->fake_enable = false;
+
+		if (aconnector->dc_sink)
+			dc_sink_release(aconnector->dc_sink);
+		aconnector->dc_sink = NULL;
+		amdgpu_dm_update_connector_after_detect(aconnector);
+	}
+	drm_connector_list_iter_end(&iter);
+
+	dm_destroy_cached_state(adev);
+
+	/* Do mst topology probing after resuming cached state*/
+	drm_connector_list_iter_begin(ddev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_WRITEBACK)
+			continue;
+
+		aconnector = to_amdgpu_dm_connector(connector);
+		if (aconnector->dc_link->type != dc_connection_mst_branch ||
+		    aconnector->mst_root)
+			continue;
+
+		drm_dp_mst_topology_queue_probe(&aconnector->mst_mgr);
+	}
+	drm_connector_list_iter_end(&iter);
+
+	amdgpu_dm_irq_resume_late(adev);
+
+	amdgpu_dm_smu_write_watermarks_table(adev);
+
+	drm_kms_helper_hotplug_event(ddev);
+
+	return 0;
+}
+
+/**
+ * DOC: DM Lifecycle
+ *
+ * DM (and consequently DC) is registered in the amdgpu base driver as a IP
+ * block. When CONFIG_DRM_AMD_DC is enabled, the DM device IP block is added to
+ * the base driver's device list to be initialized and torn down accordingly.
+ *
+ * The functions to do so are provided as hooks in &struct amd_ip_funcs.
+ */
+
+static const struct amd_ip_funcs amdgpu_dm_funcs = {
+	.name = "dm",
+	.early_init = dm_early_init,
+	.late_init = dm_late_init,
+	.sw_init = dm_sw_init,
+	.sw_fini = dm_sw_fini,
+	.early_fini = amdgpu_dm_early_fini,
+	.hw_init = dm_hw_init,
+	.hw_fini = dm_hw_fini,
+	.suspend = dm_suspend,
+	.resume = dm_resume,
+	.is_idle = dm_is_idle,
+	.wait_for_idle = dm_wait_for_idle,
+	.check_soft_reset = dm_check_soft_reset,
+	.soft_reset = dm_soft_reset,
+	.set_clockgating_state = dm_set_clockgating_state,
+	.set_powergating_state = dm_set_powergating_state,
+};
+
+const struct amdgpu_ip_block_version dm_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_DCE,
+	.major = 1,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &amdgpu_dm_funcs,
+};
+
+
+/**
+ * DOC: atomic
+ *
+ * *WIP*
+ */
+
+static const struct drm_mode_config_funcs amdgpu_dm_mode_funcs = {
+	.fb_create = amdgpu_display_user_framebuffer_create,
+	.get_format_info = amdgpu_dm_plane_get_format_info,
+	.atomic_check = amdgpu_dm_atomic_check,
+	.atomic_commit = drm_atomic_helper_commit,
+};
+
+static struct drm_mode_config_helper_funcs amdgpu_dm_mode_config_helperfuncs = {
+	.atomic_commit_tail = amdgpu_dm_atomic_commit_tail,
+	.atomic_commit_setup = amdgpu_dm_atomic_setup_commit,
+};
+
+static void update_connector_ext_caps(struct amdgpu_dm_connector *aconnector)
+{
+	const struct drm_panel_backlight_quirk *panel_backlight_quirk;
+	struct amdgpu_dm_backlight_caps *caps;
+	struct drm_connector *conn_base;
+	struct amdgpu_device *adev;
+	struct drm_luminance_range_info *luminance_range;
+	struct drm_device *drm;
+
+	if (aconnector->bl_idx == -1 ||
+	    aconnector->dc_link->connector_signal != SIGNAL_TYPE_EDP)
+		return;
+
+	conn_base = &aconnector->base;
+	drm = conn_base->dev;
+	adev = drm_to_adev(drm);
+
+	caps = &adev->dm.backlight_caps[aconnector->bl_idx];
+	caps->ext_caps = &aconnector->dc_link->dpcd_sink_ext_caps;
+	caps->aux_support = false;
+
+	if (caps->ext_caps->bits.oled == 1
+	    /*
+	     * ||
+	     * caps->ext_caps->bits.sdr_aux_backlight_control == 1 ||
+	     * caps->ext_caps->bits.hdr_aux_backlight_control == 1
+	     */)
+		caps->aux_support = true;
+
+	if (amdgpu_backlight == 0)
+		caps->aux_support = false;
+	else if (amdgpu_backlight == 1)
+		caps->aux_support = true;
+	if (caps->aux_support)
+		aconnector->dc_link->backlight_control_type = BACKLIGHT_CONTROL_AMD_AUX;
+
+	luminance_range = &conn_base->display_info.luminance_range;
+
+	if (luminance_range->max_luminance)
+		caps->aux_max_input_signal = luminance_range->max_luminance;
+	else
+		caps->aux_max_input_signal = 512;
+
+	if (luminance_range->min_luminance)
+		caps->aux_min_input_signal = luminance_range->min_luminance;
+	else
+		caps->aux_min_input_signal = 1;
+
+	panel_backlight_quirk =
+		drm_get_panel_backlight_quirk(aconnector->drm_edid);
+	if (!IS_ERR_OR_NULL(panel_backlight_quirk)) {
+		if (panel_backlight_quirk->min_brightness) {
+			caps->min_input_signal =
+				panel_backlight_quirk->min_brightness - 1;
+			drm_info(drm,
+				 "Applying panel backlight quirk, min_brightness: %d\n",
+				 caps->min_input_signal);
+		}
+		if (panel_backlight_quirk->brightness_mask) {
+			drm_info(drm,
+				 "Applying panel backlight quirk, brightness_mask: 0x%X\n",
+				 panel_backlight_quirk->brightness_mask);
+			caps->brightness_mask =
+				panel_backlight_quirk->brightness_mask;
+		}
+	}
+}
+
+DEFINE_FREE(sink_release, struct dc_sink *, if (_T) dc_sink_release(_T))
+
+void amdgpu_dm_update_connector_after_detect(
+		struct amdgpu_dm_connector *aconnector)
+{
+	struct drm_connector *connector = &aconnector->base;
+	struct dc_sink *sink __free(sink_release) = NULL;
+	struct drm_device *dev = connector->dev;
+
+	/* MST handled by drm_mst framework */
+	if (aconnector->mst_mgr.mst_state == true)
+		return;
+
+	sink = aconnector->dc_link->local_sink;
+	if (sink)
+		dc_sink_retain(sink);
+
+	/*
+	 * Edid mgmt connector gets first update only in mode_valid hook and then
+	 * the connector sink is set to either fake or physical sink depends on link status.
+	 * Skip if already done during boot.
+	 */
+	if (aconnector->base.force != DRM_FORCE_UNSPECIFIED
+			&& aconnector->dc_em_sink) {
+
+		/*
+		 * For S3 resume with headless use eml_sink to fake stream
+		 * because on resume connector->sink is set to NULL
+		 */
+		guard(mutex)(&dev->mode_config.mutex);
+
+		if (sink) {
+			if (aconnector->dc_sink) {
+				amdgpu_dm_update_freesync_caps(connector, NULL);
+				/*
+				 * retain and release below are used to
+				 * bump up refcount for sink because the link doesn't point
+				 * to it anymore after disconnect, so on next crtc to connector
+				 * reshuffle by UMD we will get into unwanted dc_sink release
+				 */
+				dc_sink_release(aconnector->dc_sink);
+			}
+			aconnector->dc_sink = sink;
+			dc_sink_retain(aconnector->dc_sink);
+			amdgpu_dm_update_freesync_caps(connector,
+					aconnector->drm_edid);
+		} else {
+			amdgpu_dm_update_freesync_caps(connector, NULL);
+			if (!aconnector->dc_sink) {
+				aconnector->dc_sink = aconnector->dc_em_sink;
+				dc_sink_retain(aconnector->dc_sink);
+			}
+		}
+
+		return;
+	}
+
+	/*
+	 * TODO: temporary guard to look for proper fix
+	 * if this sink is MST sink, we should not do anything
+	 */
+	if (sink && sink->sink_signal == SIGNAL_TYPE_DISPLAY_PORT_MST)
+		return;
+
+	if (aconnector->dc_sink == sink) {
+		/*
+		 * We got a DP short pulse (Link Loss, DP CTS, etc...).
+		 * Do nothing!!
+		 */
+		drm_dbg_kms(dev, "DCHPD: connector_id=%d: dc_sink didn't change.\n",
+				 aconnector->connector_id);
+		return;
+	}
+
+	drm_dbg_kms(dev, "DCHPD: connector_id=%d: Old sink=%p New sink=%p\n",
+		    aconnector->connector_id, aconnector->dc_sink, sink);
+
+	guard(mutex)(&dev->mode_config.mutex);
+
+	/*
+	 * 1. Update status of the drm connector
+	 * 2. Send an event and let userspace tell us what to do
+	 */
+	if (sink) {
+		/*
+		 * TODO: check if we still need the S3 mode update workaround.
+		 * If yes, put it here.
+		 */
+		if (aconnector->dc_sink) {
+			amdgpu_dm_update_freesync_caps(connector, NULL);
+			dc_sink_release(aconnector->dc_sink);
+		}
+
+		aconnector->dc_sink = sink;
+		dc_sink_retain(aconnector->dc_sink);
+		if (sink->dc_edid.length == 0) {
+			aconnector->drm_edid = NULL;
+			hdmi_cec_unset_edid(aconnector);
+			if (aconnector->dc_link->aux_mode) {
+				drm_dp_cec_unset_edid(&aconnector->dm_dp_aux.aux);
+			}
+		} else {
+			const struct edid *edid = (const struct edid *)sink->dc_edid.raw_edid;
+
+			aconnector->drm_edid = drm_edid_alloc(edid, sink->dc_edid.length);
+			drm_edid_connector_update(connector, aconnector->drm_edid);
+
+			hdmi_cec_set_edid(aconnector);
+			if (aconnector->dc_link->aux_mode)
+				drm_dp_cec_attach(&aconnector->dm_dp_aux.aux,
+						  connector->display_info.source_physical_address);
+		}
+
+		if (!aconnector->timing_requested) {
+			aconnector->timing_requested =
+				kzalloc(sizeof(struct dc_crtc_timing), GFP_KERNEL);
+			if (!aconnector->timing_requested)
+				drm_err(dev,
+					"failed to create aconnector->requested_timing\n");
+		}
+
+		amdgpu_dm_update_freesync_caps(connector, aconnector->drm_edid);
+		update_connector_ext_caps(aconnector);
+	} else {
+		hdmi_cec_unset_edid(aconnector);
+		drm_dp_cec_unset_edid(&aconnector->dm_dp_aux.aux);
+		amdgpu_dm_update_freesync_caps(connector, NULL);
+		aconnector->num_modes = 0;
+		dc_sink_release(aconnector->dc_sink);
+		aconnector->dc_sink = NULL;
+		drm_edid_free(aconnector->drm_edid);
+		aconnector->drm_edid = NULL;
+		kfree(aconnector->timing_requested);
+		aconnector->timing_requested = NULL;
+		/* Set CP to DESIRED if it was ENABLED, so we can re-enable it again on hotplug */
+		if (connector->state->content_protection == DRM_MODE_CONTENT_PROTECTION_ENABLED)
+			connector->state->content_protection = DRM_MODE_CONTENT_PROTECTION_DESIRED;
+	}
+
+	update_subconnector_property(aconnector);
+}
+
+static void handle_hpd_irq_helper(struct amdgpu_dm_connector *aconnector)
+{
+	struct drm_connector *connector = &aconnector->base;
+	struct drm_device *dev = connector->dev;
+	enum dc_connection_type new_connection_type = dc_connection_none;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	struct dm_connector_state *dm_con_state = to_dm_connector_state(connector->state);
+	struct dc *dc = aconnector->dc_link->ctx->dc;
+	bool ret = false;
+
+	if (adev->dm.disable_hpd_irq)
+		return;
+
+	/*
+	 * In case of failure or MST no need to update connector status or notify the OS
+	 * since (for MST case) MST does this in its own context.
+	 */
+	guard(mutex)(&aconnector->hpd_lock);
+
+	if (adev->dm.hdcp_workqueue) {
+		hdcp_reset_display(adev->dm.hdcp_workqueue, aconnector->dc_link->link_index);
+		dm_con_state->update_hdcp = true;
+	}
+	if (aconnector->fake_enable)
+		aconnector->fake_enable = false;
+
+	aconnector->timing_changed = false;
+
+	if (!dc_link_detect_connection_type(aconnector->dc_link, &new_connection_type))
+		drm_err(adev_to_drm(adev), "KMS: Failed to detect connector\n");
+
+	if (aconnector->base.force && new_connection_type == dc_connection_none) {
+		emulated_link_detect(aconnector->dc_link);
+
+		drm_modeset_lock_all(dev);
+		dm_restore_drm_connector_state(dev, connector);
+		drm_modeset_unlock_all(dev);
+
+		if (aconnector->base.force == DRM_FORCE_UNSPECIFIED)
+			drm_kms_helper_connector_hotplug_event(connector);
+	} else {
+		scoped_guard(mutex, &adev->dm.dc_lock) {
+			dc_exit_ips_for_hw_access(dc);
+			ret = dc_link_detect(aconnector->dc_link, DETECT_REASON_HPD);
+		}
+		if (ret) {
+			/* w/a delay for certain panels */
+			apply_delay_after_dpcd_poweroff(adev, aconnector->dc_sink);
+			amdgpu_dm_update_connector_after_detect(aconnector);
+
+			drm_modeset_lock_all(dev);
+			dm_restore_drm_connector_state(dev, connector);
+			drm_modeset_unlock_all(dev);
+
+			if (aconnector->base.force == DRM_FORCE_UNSPECIFIED)
+				drm_kms_helper_connector_hotplug_event(connector);
+		}
+	}
+}
+
+static void handle_hpd_irq(void *param)
+{
+	struct amdgpu_dm_connector *aconnector = (struct amdgpu_dm_connector *)param;
+
+	handle_hpd_irq_helper(aconnector);
+
+}
+
+static void schedule_hpd_rx_offload_work(struct amdgpu_device *adev, struct hpd_rx_irq_offload_work_queue *offload_wq,
+							union hpd_irq_data hpd_irq_data)
+{
+	struct hpd_rx_irq_offload_work *offload_work =
+				kzalloc(sizeof(*offload_work), GFP_KERNEL);
+
+	if (!offload_work) {
+		drm_err(adev_to_drm(adev), "Failed to allocate hpd_rx_irq_offload_work.\n");
+		return;
+	}
+
+	INIT_WORK(&offload_work->work, dm_handle_hpd_rx_offload_work);
+	offload_work->data = hpd_irq_data;
+	offload_work->offload_wq = offload_wq;
+	offload_work->adev = adev;
+
+	queue_work(offload_wq->wq, &offload_work->work);
+	DRM_DEBUG_KMS("queue work to handle hpd_rx offload work");
+}
+
+static void handle_hpd_rx_irq(void *param)
+{
+	struct amdgpu_dm_connector *aconnector = (struct amdgpu_dm_connector *)param;
+	struct drm_connector *connector = &aconnector->base;
+	struct drm_device *dev = connector->dev;
+	struct dc_link *dc_link = aconnector->dc_link;
+	bool is_mst_root_connector = aconnector->mst_mgr.mst_state;
+	bool result = false;
+	enum dc_connection_type new_connection_type = dc_connection_none;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	union hpd_irq_data hpd_irq_data;
+	bool link_loss = false;
+	bool has_left_work = false;
+	int idx = dc_link->link_index;
+	struct hpd_rx_irq_offload_work_queue *offload_wq = &adev->dm.hpd_rx_offload_wq[idx];
+	struct dc *dc = aconnector->dc_link->ctx->dc;
+
+	memset(&hpd_irq_data, 0, sizeof(hpd_irq_data));
+
+	if (adev->dm.disable_hpd_irq)
+		return;
+
+	/*
+	 * TODO:Temporary add mutex to protect hpd interrupt not have a gpio
+	 * conflict, after implement i2c helper, this mutex should be
+	 * retired.
+	 */
+	mutex_lock(&aconnector->hpd_lock);
+
+	result = dc_link_handle_hpd_rx_irq(dc_link, &hpd_irq_data,
+						&link_loss, true, &has_left_work);
+
+	if (!has_left_work)
+		goto out;
+
+	if (hpd_irq_data.bytes.device_service_irq.bits.AUTOMATED_TEST) {
+		schedule_hpd_rx_offload_work(adev, offload_wq, hpd_irq_data);
+		goto out;
+	}
+
+	if (dc_link_dp_allow_hpd_rx_irq(dc_link)) {
+		if (hpd_irq_data.bytes.device_service_irq.bits.UP_REQ_MSG_RDY ||
+			hpd_irq_data.bytes.device_service_irq.bits.DOWN_REP_MSG_RDY) {
+			bool skip = false;
+
+			/*
+			 * DOWN_REP_MSG_RDY is also handled by polling method
+			 * mgr->cbs->poll_hpd_irq()
+			 */
+			spin_lock(&offload_wq->offload_lock);
+			skip = offload_wq->is_handling_mst_msg_rdy_event;
+
+			if (!skip)
+				offload_wq->is_handling_mst_msg_rdy_event = true;
+
+			spin_unlock(&offload_wq->offload_lock);
+
+			if (!skip)
+				schedule_hpd_rx_offload_work(adev, offload_wq, hpd_irq_data);
+
+			goto out;
+		}
+
+		if (link_loss) {
+			bool skip = false;
+
+			spin_lock(&offload_wq->offload_lock);
+			skip = offload_wq->is_handling_link_loss;
+
+			if (!skip)
+				offload_wq->is_handling_link_loss = true;
+
+			spin_unlock(&offload_wq->offload_lock);
+
+			if (!skip)
+				schedule_hpd_rx_offload_work(adev, offload_wq, hpd_irq_data);
+
+			goto out;
+		}
+	}
+
+out:
+	if (result && !is_mst_root_connector) {
+		/* Downstream Port status changed. */
+		if (!dc_link_detect_connection_type(dc_link, &new_connection_type))
+			drm_err(adev_to_drm(adev), "KMS: Failed to detect connector\n");
+
+		if (aconnector->base.force && new_connection_type == dc_connection_none) {
+			emulated_link_detect(dc_link);
+
+			if (aconnector->fake_enable)
+				aconnector->fake_enable = false;
+
+			amdgpu_dm_update_connector_after_detect(aconnector);
+
+
+			drm_modeset_lock_all(dev);
+			dm_restore_drm_connector_state(dev, connector);
+			drm_modeset_unlock_all(dev);
+
+			drm_kms_helper_connector_hotplug_event(connector);
+		} else {
+			bool ret = false;
+
+			mutex_lock(&adev->dm.dc_lock);
+			dc_exit_ips_for_hw_access(dc);
+			ret = dc_link_detect(dc_link, DETECT_REASON_HPDRX);
+			mutex_unlock(&adev->dm.dc_lock);
+
+			if (ret) {
+				if (aconnector->fake_enable)
+					aconnector->fake_enable = false;
+
+				amdgpu_dm_update_connector_after_detect(aconnector);
+
+				drm_modeset_lock_all(dev);
+				dm_restore_drm_connector_state(dev, connector);
+				drm_modeset_unlock_all(dev);
+
+				drm_kms_helper_connector_hotplug_event(connector);
+			}
+		}
+	}
+	if (hpd_irq_data.bytes.device_service_irq.bits.CP_IRQ) {
+		if (adev->dm.hdcp_workqueue)
+			hdcp_handle_cpirq(adev->dm.hdcp_workqueue,  aconnector->base.index);
+	}
+
+	if (dc_link->type != dc_connection_mst_branch)
+		drm_dp_cec_irq(&aconnector->dm_dp_aux.aux);
+
+	mutex_unlock(&aconnector->hpd_lock);
+}
+
+static int register_hpd_handlers(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev_to_drm(adev);
+	struct drm_connector *connector;
+	struct amdgpu_dm_connector *aconnector;
+	const struct dc_link *dc_link;
+	struct dc_interrupt_params int_params = {0};
+
+	int_params.requested_polarity = INTERRUPT_POLARITY_DEFAULT;
+	int_params.current_polarity = INTERRUPT_POLARITY_DEFAULT;
+
+	if (dc_is_dmub_outbox_supported(adev->dm.dc)) {
+		if (!register_dmub_notify_callback(adev, DMUB_NOTIFICATION_HPD,
+			dmub_hpd_callback, true)) {
+			drm_err(adev_to_drm(adev), "fail to register dmub hpd callback");
+			return -EINVAL;
+		}
+
+		if (!register_dmub_notify_callback(adev, DMUB_NOTIFICATION_HPD_IRQ,
+			dmub_hpd_callback, true)) {
+			drm_err(adev_to_drm(adev), "fail to register dmub hpd callback");
+			return -EINVAL;
+		}
+
+		if (!register_dmub_notify_callback(adev, DMUB_NOTIFICATION_HPD_SENSE_NOTIFY,
+			dmub_hpd_sense_callback, true)) {
+			drm_err(adev_to_drm(adev), "fail to register dmub hpd sense callback");
+			return -EINVAL;
+		}
+	}
+
+	list_for_each_entry(connector,
+			&dev->mode_config.connector_list, head)	{
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_WRITEBACK)
+			continue;
+
+		aconnector = to_amdgpu_dm_connector(connector);
+		dc_link = aconnector->dc_link;
+
+		if (dc_link->irq_source_hpd != DC_IRQ_SOURCE_INVALID) {
+			int_params.int_context = INTERRUPT_LOW_IRQ_CONTEXT;
+			int_params.irq_source = dc_link->irq_source_hpd;
+
+			if (int_params.irq_source == DC_IRQ_SOURCE_INVALID ||
+				int_params.irq_source  < DC_IRQ_SOURCE_HPD1 ||
+				int_params.irq_source  > DC_IRQ_SOURCE_HPD6) {
+				drm_err(adev_to_drm(adev), "Failed to register hpd irq!\n");
+				return -EINVAL;
+			}
+
+			if (!amdgpu_dm_irq_register_interrupt(adev, &int_params,
+				handle_hpd_irq, (void *) aconnector))
+				return -ENOMEM;
+		}
+
+		if (dc_link->irq_source_hpd_rx != DC_IRQ_SOURCE_INVALID) {
+
+			/* Also register for DP short pulse (hpd_rx). */
+			int_params.int_context = INTERRUPT_LOW_IRQ_CONTEXT;
+			int_params.irq_source =	dc_link->irq_source_hpd_rx;
+
+			if (int_params.irq_source == DC_IRQ_SOURCE_INVALID ||
+				int_params.irq_source  < DC_IRQ_SOURCE_HPD1RX ||
+				int_params.irq_source  > DC_IRQ_SOURCE_HPD6RX) {
+				drm_err(adev_to_drm(adev), "Failed to register hpd rx irq!\n");
+				return -EINVAL;
+			}
+
+			if (!amdgpu_dm_irq_register_interrupt(adev, &int_params,
+				handle_hpd_rx_irq, (void *) aconnector))
+				return -ENOMEM;
+		}
+	}
+	return 0;
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+/* Register IRQ sources and initialize IRQ callbacks */
+static int dce60_register_irq_handlers(struct amdgpu_device *adev)
+{
+	struct dc *dc = adev->dm.dc;
+	struct common_irq_params *c_irq_params;
+	struct dc_interrupt_params int_params = {0};
+	int r;
+	int i;
+	unsigned int client_id = AMDGPU_IRQ_CLIENTID_LEGACY;
+
+	int_params.requested_polarity = INTERRUPT_POLARITY_DEFAULT;
+	int_params.current_polarity = INTERRUPT_POLARITY_DEFAULT;
+
+	/*
+	 * Actions of amdgpu_irq_add_id():
+	 * 1. Register a set() function with base driver.
+	 *    Base driver will call set() function to enable/disable an
+	 *    interrupt in DC hardware.
+	 * 2. Register amdgpu_dm_irq_handler().
+	 *    Base driver will call amdgpu_dm_irq_handler() for ALL interrupts
+	 *    coming from DC hardware.
+	 *    amdgpu_dm_irq_handler() will re-direct the interrupt to DC
+	 *    for acknowledging and handling.
+	 */
+
+	/* Use VBLANK interrupt */
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		r = amdgpu_irq_add_id(adev, client_id, i + 1, &adev->crtc_irq);
+		if (r) {
+			drm_err(adev_to_drm(adev), "Failed to add crtc irq id!\n");
+			return r;
+		}
+
+		int_params.int_context = INTERRUPT_HIGH_IRQ_CONTEXT;
+		int_params.irq_source =
+			dc_interrupt_to_irq_source(dc, i + 1, 0);
+
+		if (int_params.irq_source == DC_IRQ_SOURCE_INVALID ||
+			int_params.irq_source  < DC_IRQ_SOURCE_VBLANK1 ||
+			int_params.irq_source  > DC_IRQ_SOURCE_VBLANK6) {
+			drm_err(adev_to_drm(adev), "Failed to register vblank irq!\n");
+			return -EINVAL;
+		}
+
+		c_irq_params = &adev->dm.vblank_params[int_params.irq_source - DC_IRQ_SOURCE_VBLANK1];
+
+		c_irq_params->adev = adev;
+		c_irq_params->irq_src = int_params.irq_source;
+
+		if (!amdgpu_dm_irq_register_interrupt(adev, &int_params,
+			dm_crtc_high_irq, c_irq_params))
+			return -ENOMEM;
+	}
+
+	/* Use GRPH_PFLIP interrupt */
+	for (i = VISLANDS30_IV_SRCID_D1_GRPH_PFLIP;
+			i <= VISLANDS30_IV_SRCID_D6_GRPH_PFLIP; i += 2) {
+		r = amdgpu_irq_add_id(adev, client_id, i, &adev->pageflip_irq);
+		if (r) {
+			drm_err(adev_to_drm(adev), "Failed to add page flip irq id!\n");
+			return r;
+		}
+
+		int_params.int_context = INTERRUPT_HIGH_IRQ_CONTEXT;
+		int_params.irq_source =
+			dc_interrupt_to_irq_source(dc, i, 0);
+
+		if (int_params.irq_source == DC_IRQ_SOURCE_INVALID ||
+			int_params.irq_source  < DC_IRQ_SOURCE_PFLIP_FIRST ||
+			int_params.irq_source  > DC_IRQ_SOURCE_PFLIP_LAST) {
+			drm_err(adev_to_drm(adev), "Failed to register pflip irq!\n");
+			return -EINVAL;
+		}
+
+		c_irq_params = &adev->dm.pflip_params[int_params.irq_source - DC_IRQ_SOURCE_PFLIP_FIRST];
+
+		c_irq_params->adev = adev;
+		c_irq_params->irq_src = int_params.irq_source;
+
+		if (!amdgpu_dm_irq_register_interrupt(adev, &int_params,
+			dm_pflip_high_irq, c_irq_params))
+			return -ENOMEM;
+	}
+
+	/* HPD */
+	r = amdgpu_irq_add_id(adev, client_id,
+			VISLANDS30_IV_SRCID_HOTPLUG_DETECT_A, &adev->hpd_irq);
+	if (r) {
+		drm_err(adev_to_drm(adev), "Failed to add hpd irq id!\n");
+		return r;
+	}
+
+	r = register_hpd_handlers(adev);
+
+	return r;
+}
+#endif
+
+/* Register IRQ sources and initialize IRQ callbacks */
+static int dce110_register_irq_handlers(struct amdgpu_device *adev)
+{
+	struct dc *dc = adev->dm.dc;
+	struct common_irq_params *c_irq_params;
+	struct dc_interrupt_params int_params = {0};
+	int r;
+	int i;
+	unsigned int client_id = AMDGPU_IRQ_CLIENTID_LEGACY;
+
+	if (adev->family >= AMDGPU_FAMILY_AI)
+		client_id = SOC15_IH_CLIENTID_DCE;
+
+	int_params.requested_polarity = INTERRUPT_POLARITY_DEFAULT;
+	int_params.current_polarity = INTERRUPT_POLARITY_DEFAULT;
+
+	/*
+	 * Actions of amdgpu_irq_add_id():
+	 * 1. Register a set() function with base driver.
+	 *    Base driver will call set() function to enable/disable an
+	 *    interrupt in DC hardware.
+	 * 2. Register amdgpu_dm_irq_handler().
+	 *    Base driver will call amdgpu_dm_irq_handler() for ALL interrupts
+	 *    coming from DC hardware.
+	 *    amdgpu_dm_irq_handler() will re-direct the interrupt to DC
+	 *    for acknowledging and handling.
+	 */
+
+	/* Use VBLANK interrupt */
+	for (i = VISLANDS30_IV_SRCID_D1_VERTICAL_INTERRUPT0; i <= VISLANDS30_IV_SRCID_D6_VERTICAL_INTERRUPT0; i++) {
+		r = amdgpu_irq_add_id(adev, client_id, i, &adev->crtc_irq);
+		if (r) {
+			drm_err(adev_to_drm(adev), "Failed to add crtc irq id!\n");
+			return r;
+		}
+
+		int_params.int_context = INTERRUPT_HIGH_IRQ_CONTEXT;
+		int_params.irq_source =
+			dc_interrupt_to_irq_source(dc, i, 0);
+
+		if (int_params.irq_source == DC_IRQ_SOURCE_INVALID ||
+			int_params.irq_source  < DC_IRQ_SOURCE_VBLANK1 ||
+			int_params.irq_source  > DC_IRQ_SOURCE_VBLANK6) {
+			drm_err(adev_to_drm(adev), "Failed to register vblank irq!\n");
+			return -EINVAL;
+		}
+
+		c_irq_params = &adev->dm.vblank_params[int_params.irq_source - DC_IRQ_SOURCE_VBLANK1];
+
+		c_irq_params->adev = adev;
+		c_irq_params->irq_src = int_params.irq_source;
+
+		if (!amdgpu_dm_irq_register_interrupt(adev, &int_params,
+			dm_crtc_high_irq, c_irq_params))
+			return -ENOMEM;
+	}
+
+	/* Use VUPDATE interrupt */
+	for (i = VISLANDS30_IV_SRCID_D1_V_UPDATE_INT; i <= VISLANDS30_IV_SRCID_D6_V_UPDATE_INT; i += 2) {
+		r = amdgpu_irq_add_id(adev, client_id, i, &adev->vupdate_irq);
+		if (r) {
+			drm_err(adev_to_drm(adev), "Failed to add vupdate irq id!\n");
+			return r;
+		}
+
+		int_params.int_context = INTERRUPT_HIGH_IRQ_CONTEXT;
+		int_params.irq_source =
+			dc_interrupt_to_irq_source(dc, i, 0);
+
+		if (int_params.irq_source == DC_IRQ_SOURCE_INVALID ||
+			int_params.irq_source  < DC_IRQ_SOURCE_VUPDATE1 ||
+			int_params.irq_source  > DC_IRQ_SOURCE_VUPDATE6) {
+			drm_err(adev_to_drm(adev), "Failed to register vupdate irq!\n");
+			return -EINVAL;
+		}
+
+		c_irq_params = &adev->dm.vupdate_params[int_params.irq_source - DC_IRQ_SOURCE_VUPDATE1];
+
+		c_irq_params->adev = adev;
+		c_irq_params->irq_src = int_params.irq_source;
+
+		if (!amdgpu_dm_irq_register_interrupt(adev, &int_params,
+			dm_vupdate_high_irq, c_irq_params))
+			return -ENOMEM;
+	}
+
+	/* Use GRPH_PFLIP interrupt */
+	for (i = VISLANDS30_IV_SRCID_D1_GRPH_PFLIP;
+			i <= VISLANDS30_IV_SRCID_D6_GRPH_PFLIP; i += 2) {
+		r = amdgpu_irq_add_id(adev, client_id, i, &adev->pageflip_irq);
+		if (r) {
+			drm_err(adev_to_drm(adev), "Failed to add page flip irq id!\n");
+			return r;
+		}
+
+		int_params.int_context = INTERRUPT_HIGH_IRQ_CONTEXT;
+		int_params.irq_source =
+			dc_interrupt_to_irq_source(dc, i, 0);
+
+		if (int_params.irq_source == DC_IRQ_SOURCE_INVALID ||
+			int_params.irq_source  < DC_IRQ_SOURCE_PFLIP_FIRST ||
+			int_params.irq_source  > DC_IRQ_SOURCE_PFLIP_LAST) {
+			drm_err(adev_to_drm(adev), "Failed to register pflip irq!\n");
+			return -EINVAL;
+		}
+
+		c_irq_params = &adev->dm.pflip_params[int_params.irq_source - DC_IRQ_SOURCE_PFLIP_FIRST];
+
+		c_irq_params->adev = adev;
+		c_irq_params->irq_src = int_params.irq_source;
+
+		if (!amdgpu_dm_irq_register_interrupt(adev, &int_params,
+			dm_pflip_high_irq, c_irq_params))
+			return -ENOMEM;
+	}
+
+	/* HPD */
+	r = amdgpu_irq_add_id(adev, client_id,
+			VISLANDS30_IV_SRCID_HOTPLUG_DETECT_A, &adev->hpd_irq);
+	if (r) {
+		drm_err(adev_to_drm(adev), "Failed to add hpd irq id!\n");
+		return r;
+	}
+
+	r = register_hpd_handlers(adev);
+
+	return r;
+}
+
+/* Register IRQ sources and initialize IRQ callbacks */
+static int dcn10_register_irq_handlers(struct amdgpu_device *adev)
+{
+	struct dc *dc = adev->dm.dc;
+	struct common_irq_params *c_irq_params;
+	struct dc_interrupt_params int_params = {0};
+	int r;
+	int i;
+#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY)
+	static const unsigned int vrtl_int_srcid[] = {
+		DCN_1_0__SRCID__OTG1_VERTICAL_INTERRUPT0_CONTROL,
+		DCN_1_0__SRCID__OTG2_VERTICAL_INTERRUPT0_CONTROL,
+		DCN_1_0__SRCID__OTG3_VERTICAL_INTERRUPT0_CONTROL,
+		DCN_1_0__SRCID__OTG4_VERTICAL_INTERRUPT0_CONTROL,
+		DCN_1_0__SRCID__OTG5_VERTICAL_INTERRUPT0_CONTROL,
+		DCN_1_0__SRCID__OTG6_VERTICAL_INTERRUPT0_CONTROL
+	};
+#endif
+
+	int_params.requested_polarity = INTERRUPT_POLARITY_DEFAULT;
+	int_params.current_polarity = INTERRUPT_POLARITY_DEFAULT;
+
+	/*
+	 * Actions of amdgpu_irq_add_id():
+	 * 1. Register a set() function with base driver.
+	 *    Base driver will call set() function to enable/disable an
+	 *    interrupt in DC hardware.
+	 * 2. Register amdgpu_dm_irq_handler().
+	 *    Base driver will call amdgpu_dm_irq_handler() for ALL interrupts
+	 *    coming from DC hardware.
+	 *    amdgpu_dm_irq_handler() will re-direct the interrupt to DC
+	 *    for acknowledging and handling.
+	 */
+
+	/* Use VSTARTUP interrupt */
+	for (i = DCN_1_0__SRCID__DC_D1_OTG_VSTARTUP;
+			i <= DCN_1_0__SRCID__DC_D1_OTG_VSTARTUP + adev->mode_info.num_crtc - 1;
+			i++) {
+		r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_DCE, i, &adev->crtc_irq);
+
+		if (r) {
+			drm_err(adev_to_drm(adev), "Failed to add crtc irq id!\n");
+			return r;
+		}
+
+		int_params.int_context = INTERRUPT_HIGH_IRQ_CONTEXT;
+		int_params.irq_source =
+			dc_interrupt_to_irq_source(dc, i, 0);
+
+		if (int_params.irq_source == DC_IRQ_SOURCE_INVALID ||
+			int_params.irq_source  < DC_IRQ_SOURCE_VBLANK1 ||
+			int_params.irq_source  > DC_IRQ_SOURCE_VBLANK6) {
+			drm_err(adev_to_drm(adev), "Failed to register vblank irq!\n");
+			return -EINVAL;
+		}
+
+		c_irq_params = &adev->dm.vblank_params[int_params.irq_source - DC_IRQ_SOURCE_VBLANK1];
+
+		c_irq_params->adev = adev;
+		c_irq_params->irq_src = int_params.irq_source;
+
+		if (!amdgpu_dm_irq_register_interrupt(adev, &int_params,
+			dm_crtc_high_irq, c_irq_params))
+			return -ENOMEM;
+	}
+
+	/* Use otg vertical line interrupt */
+#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY)
+	for (i = 0; i <= adev->mode_info.num_crtc - 1; i++) {
+		r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_DCE,
+				vrtl_int_srcid[i], &adev->vline0_irq);
+
+		if (r) {
+			drm_err(adev_to_drm(adev), "Failed to add vline0 irq id!\n");
+			return r;
+		}
+
+		int_params.int_context = INTERRUPT_HIGH_IRQ_CONTEXT;
+		int_params.irq_source =
+			dc_interrupt_to_irq_source(dc, vrtl_int_srcid[i], 0);
+
+		if (int_params.irq_source == DC_IRQ_SOURCE_INVALID ||
+			int_params.irq_source < DC_IRQ_SOURCE_DC1_VLINE0 ||
+			int_params.irq_source > DC_IRQ_SOURCE_DC6_VLINE0) {
+			drm_err(adev_to_drm(adev), "Failed to register vline0 irq!\n");
+			return -EINVAL;
+		}
+
+		c_irq_params = &adev->dm.vline0_params[int_params.irq_source
+					- DC_IRQ_SOURCE_DC1_VLINE0];
+
+		c_irq_params->adev = adev;
+		c_irq_params->irq_src = int_params.irq_source;
+
+		if (!amdgpu_dm_irq_register_interrupt(adev, &int_params,
+			dm_dcn_vertical_interrupt0_high_irq,
+			c_irq_params))
+			return -ENOMEM;
+	}
+#endif
+
+	/* Use VUPDATE_NO_LOCK interrupt on DCN, which seems to correspond to
+	 * the regular VUPDATE interrupt on DCE. We want DC_IRQ_SOURCE_VUPDATEx
+	 * to trigger at end of each vblank, regardless of state of the lock,
+	 * matching DCE behaviour.
+	 */
+	for (i = DCN_1_0__SRCID__OTG0_IHC_V_UPDATE_NO_LOCK_INTERRUPT;
+	     i <= DCN_1_0__SRCID__OTG0_IHC_V_UPDATE_NO_LOCK_INTERRUPT + adev->mode_info.num_crtc - 1;
+	     i++) {
+		r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_DCE, i, &adev->vupdate_irq);
+
+		if (r) {
+			drm_err(adev_to_drm(adev), "Failed to add vupdate irq id!\n");
+			return r;
+		}
+
+		int_params.int_context = INTERRUPT_HIGH_IRQ_CONTEXT;
+		int_params.irq_source =
+			dc_interrupt_to_irq_source(dc, i, 0);
+
+		if (int_params.irq_source == DC_IRQ_SOURCE_INVALID ||
+			int_params.irq_source  < DC_IRQ_SOURCE_VUPDATE1 ||
+			int_params.irq_source  > DC_IRQ_SOURCE_VUPDATE6) {
+			drm_err(adev_to_drm(adev), "Failed to register vupdate irq!\n");
+			return -EINVAL;
+		}
+
+		c_irq_params = &adev->dm.vupdate_params[int_params.irq_source - DC_IRQ_SOURCE_VUPDATE1];
+
+		c_irq_params->adev = adev;
+		c_irq_params->irq_src = int_params.irq_source;
+
+		if (!amdgpu_dm_irq_register_interrupt(adev, &int_params,
+			dm_vupdate_high_irq, c_irq_params))
+			return -ENOMEM;
+	}
+
+	/* Use GRPH_PFLIP interrupt */
+	for (i = DCN_1_0__SRCID__HUBP0_FLIP_INTERRUPT;
+			i <= DCN_1_0__SRCID__HUBP0_FLIP_INTERRUPT + dc->caps.max_otg_num - 1;
+			i++) {
+		r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_DCE, i, &adev->pageflip_irq);
+		if (r) {
+			drm_err(adev_to_drm(adev), "Failed to add page flip irq id!\n");
+			return r;
+		}
+
+		int_params.int_context = INTERRUPT_HIGH_IRQ_CONTEXT;
+		int_params.irq_source =
+			dc_interrupt_to_irq_source(dc, i, 0);
+
+		if (int_params.irq_source == DC_IRQ_SOURCE_INVALID ||
+			int_params.irq_source  < DC_IRQ_SOURCE_PFLIP_FIRST ||
+			int_params.irq_source  > DC_IRQ_SOURCE_PFLIP_LAST) {
+			drm_err(adev_to_drm(adev), "Failed to register pflip irq!\n");
+			return -EINVAL;
+		}
+
+		c_irq_params = &adev->dm.pflip_params[int_params.irq_source - DC_IRQ_SOURCE_PFLIP_FIRST];
+
+		c_irq_params->adev = adev;
+		c_irq_params->irq_src = int_params.irq_source;
+
+		if (!amdgpu_dm_irq_register_interrupt(adev, &int_params,
+			dm_pflip_high_irq, c_irq_params))
+			return -ENOMEM;
+	}
+
+	/* HPD */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_DCE, DCN_1_0__SRCID__DC_HPD1_INT,
+			&adev->hpd_irq);
+	if (r) {
+		drm_err(adev_to_drm(adev), "Failed to add hpd irq id!\n");
+		return r;
+	}
+
+	r = register_hpd_handlers(adev);
+
+	return r;
+}
+/* Register Outbox IRQ sources and initialize IRQ callbacks */
+static int register_outbox_irq_handlers(struct amdgpu_device *adev)
+{
+	struct dc *dc = adev->dm.dc;
+	struct common_irq_params *c_irq_params;
+	struct dc_interrupt_params int_params = {0};
+	int r, i;
+
+	int_params.requested_polarity = INTERRUPT_POLARITY_DEFAULT;
+	int_params.current_polarity = INTERRUPT_POLARITY_DEFAULT;
+
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_DCE, DCN_1_0__SRCID__DMCUB_OUTBOX_LOW_PRIORITY_READY_INT,
+			&adev->dmub_outbox_irq);
+	if (r) {
+		drm_err(adev_to_drm(adev), "Failed to add outbox irq id!\n");
+		return r;
+	}
+
+	if (dc->ctx->dmub_srv) {
+		i = DCN_1_0__SRCID__DMCUB_OUTBOX_LOW_PRIORITY_READY_INT;
+		int_params.int_context = INTERRUPT_LOW_IRQ_CONTEXT;
+		int_params.irq_source =
+		dc_interrupt_to_irq_source(dc, i, 0);
+
+		c_irq_params = &adev->dm.dmub_outbox_params[0];
+
+		c_irq_params->adev = adev;
+		c_irq_params->irq_src = int_params.irq_source;
+
+		if (!amdgpu_dm_irq_register_interrupt(adev, &int_params,
+			dm_dmub_outbox1_low_irq, c_irq_params))
+			return -ENOMEM;
+	}
+
+	return 0;
+}
+
+/*
+ * Acquires the lock for the atomic state object and returns
+ * the new atomic state.
+ *
+ * This should only be called during atomic check.
+ */
+int dm_atomic_get_state(struct drm_atomic_state *state,
+			struct dm_atomic_state **dm_state)
+{
+	struct drm_device *dev = state->dev;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	struct amdgpu_display_manager *dm = &adev->dm;
+	struct drm_private_state *priv_state;
+
+	if (*dm_state)
+		return 0;
+
+	priv_state = drm_atomic_get_private_obj_state(state, &dm->atomic_obj);
+	if (IS_ERR(priv_state))
+		return PTR_ERR(priv_state);
+
+	*dm_state = to_dm_atomic_state(priv_state);
+
+	return 0;
+}
+
+static struct dm_atomic_state *
+dm_atomic_get_new_state(struct drm_atomic_state *state)
+{
+	struct drm_device *dev = state->dev;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	struct amdgpu_display_manager *dm = &adev->dm;
+	struct drm_private_obj *obj;
+	struct drm_private_state *new_obj_state;
+	int i;
+
+	for_each_new_private_obj_in_state(state, obj, new_obj_state, i) {
+		if (obj->funcs == dm->atomic_obj.funcs)
+			return to_dm_atomic_state(new_obj_state);
+	}
+
+	return NULL;
+}
+
+static struct drm_private_state *
+dm_atomic_duplicate_state(struct drm_private_obj *obj)
+{
+	struct dm_atomic_state *old_state, *new_state;
+
+	new_state = kzalloc(sizeof(*new_state), GFP_KERNEL);
+	if (!new_state)
+		return NULL;
+
+	__drm_atomic_helper_private_obj_duplicate_state(obj, &new_state->base);
+
+	old_state = to_dm_atomic_state(obj->state);
+
+	if (old_state && old_state->context)
+		new_state->context = dc_state_create_copy(old_state->context);
+
+	if (!new_state->context) {
+		kfree(new_state);
+		return NULL;
+	}
+
+	return &new_state->base;
+}
+
+static void dm_atomic_destroy_state(struct drm_private_obj *obj,
+				    struct drm_private_state *state)
+{
+	struct dm_atomic_state *dm_state = to_dm_atomic_state(state);
+
+	if (dm_state && dm_state->context)
+		dc_state_release(dm_state->context);
+
+	kfree(dm_state);
+}
+
+static struct drm_private_state_funcs dm_atomic_state_funcs = {
+	.atomic_duplicate_state = dm_atomic_duplicate_state,
+	.atomic_destroy_state = dm_atomic_destroy_state,
+};
+
+static int amdgpu_dm_mode_config_init(struct amdgpu_device *adev)
+{
+	struct dm_atomic_state *state;
+	int r;
+
+	adev->mode_info.mode_config_initialized = true;
+
+	adev_to_drm(adev)->mode_config.funcs = (void *)&amdgpu_dm_mode_funcs;
+	adev_to_drm(adev)->mode_config.helper_private = &amdgpu_dm_mode_config_helperfuncs;
+
+	adev_to_drm(adev)->mode_config.max_width = 16384;
+	adev_to_drm(adev)->mode_config.max_height = 16384;
+
+	adev_to_drm(adev)->mode_config.preferred_depth = 24;
+	if (adev->asic_type == CHIP_HAWAII)
+		/* disable prefer shadow for now due to hibernation issues */
+		adev_to_drm(adev)->mode_config.prefer_shadow = 0;
+	else
+		adev_to_drm(adev)->mode_config.prefer_shadow = 1;
+	/* indicates support for immediate flip */
+	adev_to_drm(adev)->mode_config.async_page_flip = true;
+
+	state = kzalloc(sizeof(*state), GFP_KERNEL);
+	if (!state)
+		return -ENOMEM;
+
+	state->context = dc_state_create_current_copy(adev->dm.dc);
+	if (!state->context) {
+		kfree(state);
+		return -ENOMEM;
+	}
+
+	drm_atomic_private_obj_init(adev_to_drm(adev),
+				    &adev->dm.atomic_obj,
+				    &state->base,
+				    &dm_atomic_state_funcs);
+
+	r = amdgpu_display_modeset_create_props(adev);
+	if (r) {
+		dc_state_release(state->context);
+		kfree(state);
+		return r;
+	}
+
+#ifdef AMD_PRIVATE_COLOR
+	if (amdgpu_dm_create_color_properties(adev)) {
+		dc_state_release(state->context);
+		kfree(state);
+		return -ENOMEM;
+	}
+#endif
+
+	r = amdgpu_dm_audio_init(adev);
+	if (r) {
+		dc_state_release(state->context);
+		kfree(state);
+		return r;
+	}
+
+	return 0;
+}
+
+#define AMDGPU_DM_DEFAULT_MIN_BACKLIGHT 12
+#define AMDGPU_DM_DEFAULT_MAX_BACKLIGHT 255
+#define AMDGPU_DM_MIN_SPREAD ((AMDGPU_DM_DEFAULT_MAX_BACKLIGHT - AMDGPU_DM_DEFAULT_MIN_BACKLIGHT) / 2)
+#define AUX_BL_DEFAULT_TRANSITION_TIME_MS 50
+
+static void amdgpu_dm_update_backlight_caps(struct amdgpu_display_manager *dm,
+					    int bl_idx)
+{
+	struct amdgpu_dm_backlight_caps *caps = &dm->backlight_caps[bl_idx];
+
+	if (caps->caps_valid)
+		return;
+
+#if defined(CONFIG_ACPI)
+	amdgpu_acpi_get_backlight_caps(caps);
+
+	/* validate the firmware value is sane */
+	if (caps->caps_valid) {
+		int spread = caps->max_input_signal - caps->min_input_signal;
+
+		if (caps->max_input_signal > AMDGPU_DM_DEFAULT_MAX_BACKLIGHT ||
+		    caps->min_input_signal < 0 ||
+		    spread > AMDGPU_DM_DEFAULT_MAX_BACKLIGHT ||
+		    spread < AMDGPU_DM_MIN_SPREAD) {
+			DRM_DEBUG_KMS("DM: Invalid backlight caps: min=%d, max=%d\n",
+				      caps->min_input_signal, caps->max_input_signal);
+			caps->caps_valid = false;
+		}
+	}
+
+	if (!caps->caps_valid) {
+		caps->min_input_signal = AMDGPU_DM_DEFAULT_MIN_BACKLIGHT;
+		caps->max_input_signal = AMDGPU_DM_DEFAULT_MAX_BACKLIGHT;
+		caps->caps_valid = true;
+	}
+#else
+	if (caps->aux_support)
+		return;
+
+	caps->min_input_signal = AMDGPU_DM_DEFAULT_MIN_BACKLIGHT;
+	caps->max_input_signal = AMDGPU_DM_DEFAULT_MAX_BACKLIGHT;
+	caps->caps_valid = true;
+#endif
+}
+
+static int get_brightness_range(const struct amdgpu_dm_backlight_caps *caps,
+				unsigned int *min, unsigned int *max)
+{
+	if (!caps)
+		return 0;
+
+	if (caps->aux_support) {
+		// Firmware limits are in nits, DC API wants millinits.
+		*max = 1000 * caps->aux_max_input_signal;
+		*min = 1000 * caps->aux_min_input_signal;
+	} else {
+		// Firmware limits are 8-bit, PWM control is 16-bit.
+		*max = 0x101 * caps->max_input_signal;
+		*min = 0x101 * caps->min_input_signal;
+	}
+	return 1;
+}
+
+/* Rescale from [min..max] to [0..AMDGPU_MAX_BL_LEVEL] */
+static inline u32 scale_input_to_fw(int min, int max, u64 input)
+{
+	return DIV_ROUND_CLOSEST_ULL(input * AMDGPU_MAX_BL_LEVEL, max - min);
+}
+
+/* Rescale from [0..AMDGPU_MAX_BL_LEVEL] to [min..max] */
+static inline u32 scale_fw_to_input(int min, int max, u64 input)
+{
+	return min + DIV_ROUND_CLOSEST_ULL(input * (max - min), AMDGPU_MAX_BL_LEVEL);
+}
+
+static void convert_custom_brightness(const struct amdgpu_dm_backlight_caps *caps,
+				      unsigned int min, unsigned int max,
+				      uint32_t *user_brightness)
+{
+	u32 brightness = scale_input_to_fw(min, max, *user_brightness);
+	u8 lower_signal, upper_signal, upper_lum, lower_lum, lum;
+	int left, right;
+
+	if (amdgpu_dc_debug_mask & DC_DISABLE_CUSTOM_BRIGHTNESS_CURVE)
+		return;
+
+	if (!caps->data_points)
+		return;
+
+	/*
+	 * Handle the case where brightness is below the first data point
+	 * Interpolate between (0,0) and (first_signal, first_lum)
+	 */
+	if (brightness < caps->luminance_data[0].input_signal) {
+		lum = DIV_ROUND_CLOSEST(caps->luminance_data[0].luminance * brightness,
+					caps->luminance_data[0].input_signal);
+		goto scale;
+	}
+
+	left = 0;
+	right = caps->data_points - 1;
+	while (left <= right) {
+		int mid = left + (right - left) / 2;
+		u8 signal = caps->luminance_data[mid].input_signal;
+
+		/* Exact match found */
+		if (signal == brightness) {
+			lum = caps->luminance_data[mid].luminance;
+			goto scale;
+		}
+
+		if (signal < brightness)
+			left = mid + 1;
+		else
+			right = mid - 1;
+	}
+
+	/* verify bound */
+	if (left >= caps->data_points)
+		left = caps->data_points - 1;
+
+	/* At this point, left > right */
+	lower_signal = caps->luminance_data[right].input_signal;
+	upper_signal = caps->luminance_data[left].input_signal;
+	lower_lum = caps->luminance_data[right].luminance;
+	upper_lum = caps->luminance_data[left].luminance;
+
+	/* interpolate */
+	if (right == left || !lower_lum)
+		lum = upper_lum;
+	else
+		lum = lower_lum + DIV_ROUND_CLOSEST((upper_lum - lower_lum) *
+						    (brightness - lower_signal),
+						    upper_signal - lower_signal);
+scale:
+	*user_brightness = scale_fw_to_input(min, max,
+					     DIV_ROUND_CLOSEST(lum * brightness, 101));
+}
+
+static u32 convert_brightness_from_user(const struct amdgpu_dm_backlight_caps *caps,
+					uint32_t brightness)
+{
+	unsigned int min, max;
+
+	if (!get_brightness_range(caps, &min, &max))
+		return brightness;
+
+	convert_custom_brightness(caps, min, max, &brightness);
+
+	// Rescale 0..max to min..max
+	return min + DIV_ROUND_CLOSEST_ULL((u64)(max - min) * brightness, max);
+}
+
+static u32 convert_brightness_to_user(const struct amdgpu_dm_backlight_caps *caps,
+				      uint32_t brightness)
+{
+	unsigned int min, max;
+
+	if (!get_brightness_range(caps, &min, &max))
+		return brightness;
+
+	if (brightness < min)
+		return 0;
+	// Rescale min..max to 0..max
+	return DIV_ROUND_CLOSEST_ULL((u64)max * (brightness - min),
+				 max - min);
+}
+
+static void amdgpu_dm_backlight_set_level(struct amdgpu_display_manager *dm,
+					 int bl_idx,
+					 u32 user_brightness)
+{
+	struct amdgpu_dm_backlight_caps *caps;
+	struct dc_link *link;
+	u32 brightness;
+	bool rc, reallow_idle = false;
+
+	amdgpu_dm_update_backlight_caps(dm, bl_idx);
+	caps = &dm->backlight_caps[bl_idx];
+
+	dm->brightness[bl_idx] = user_brightness;
+	/* update scratch register */
+	if (bl_idx == 0)
+		amdgpu_atombios_scratch_regs_set_backlight_level(dm->adev, dm->brightness[bl_idx]);
+	brightness = convert_brightness_from_user(caps, dm->brightness[bl_idx]);
+	link = (struct dc_link *)dm->backlight_link[bl_idx];
+
+	/* Apply brightness quirk */
+	if (caps->brightness_mask)
+		brightness |= caps->brightness_mask;
+
+	/* Change brightness based on AUX property */
+	mutex_lock(&dm->dc_lock);
+	if (dm->dc->caps.ips_support && dm->dc->ctx->dmub_srv->idle_allowed) {
+		dc_allow_idle_optimizations(dm->dc, false);
+		reallow_idle = true;
+	}
+
+	if (trace_amdgpu_dm_brightness_enabled()) {
+		trace_amdgpu_dm_brightness(__builtin_return_address(0),
+					   user_brightness,
+					   brightness,
+					   caps->aux_support,
+					   power_supply_is_system_supplied() > 0);
+	}
+
+	if (caps->aux_support) {
+		rc = dc_link_set_backlight_level_nits(link, true, brightness,
+						      AUX_BL_DEFAULT_TRANSITION_TIME_MS);
+		if (!rc)
+			DRM_DEBUG("DM: Failed to update backlight via AUX on eDP[%d]\n", bl_idx);
+	} else {
+		struct set_backlight_level_params backlight_level_params = { 0 };
+
+		backlight_level_params.backlight_pwm_u16_16 = brightness;
+		backlight_level_params.transition_time_in_ms = 0;
+
+		rc = dc_link_set_backlight_level(link, &backlight_level_params);
+		if (!rc)
+			DRM_DEBUG("DM: Failed to update backlight on eDP[%d]\n", bl_idx);
+	}
+
+	if (dm->dc->caps.ips_support && reallow_idle)
+		dc_allow_idle_optimizations(dm->dc, true);
+
+	mutex_unlock(&dm->dc_lock);
+
+	if (rc)
+		dm->actual_brightness[bl_idx] = user_brightness;
+}
+
+static int amdgpu_dm_backlight_update_status(struct backlight_device *bd)
+{
+	struct amdgpu_display_manager *dm = bl_get_data(bd);
+	int i;
+
+	for (i = 0; i < dm->num_of_edps; i++) {
+		if (bd == dm->backlight_dev[i])
+			break;
+	}
+	if (i >= AMDGPU_DM_MAX_NUM_EDP)
+		i = 0;
+	amdgpu_dm_backlight_set_level(dm, i, bd->props.brightness);
+
+	return 0;
+}
+
+static u32 amdgpu_dm_backlight_get_level(struct amdgpu_display_manager *dm,
+					 int bl_idx)
+{
+	int ret;
+	struct amdgpu_dm_backlight_caps caps;
+	struct dc_link *link = (struct dc_link *)dm->backlight_link[bl_idx];
+
+	amdgpu_dm_update_backlight_caps(dm, bl_idx);
+	caps = dm->backlight_caps[bl_idx];
+
+	if (caps.aux_support) {
+		u32 avg, peak;
+
+		if (!dc_link_get_backlight_level_nits(link, &avg, &peak))
+			return dm->brightness[bl_idx];
+		return convert_brightness_to_user(&caps, avg);
+	}
+
+	ret = dc_link_get_backlight_level(link);
+
+	if (ret == DC_ERROR_UNEXPECTED)
+		return dm->brightness[bl_idx];
+
+	return convert_brightness_to_user(&caps, ret);
+}
+
+static int amdgpu_dm_backlight_get_brightness(struct backlight_device *bd)
+{
+	struct amdgpu_display_manager *dm = bl_get_data(bd);
+	int i;
+
+	for (i = 0; i < dm->num_of_edps; i++) {
+		if (bd == dm->backlight_dev[i])
+			break;
+	}
+	if (i >= AMDGPU_DM_MAX_NUM_EDP)
+		i = 0;
+	return amdgpu_dm_backlight_get_level(dm, i);
+}
+
+static const struct backlight_ops amdgpu_dm_backlight_ops = {
+	.options = BL_CORE_SUSPENDRESUME,
+	.get_brightness = amdgpu_dm_backlight_get_brightness,
+	.update_status	= amdgpu_dm_backlight_update_status,
+};
+
+static void
+amdgpu_dm_register_backlight_device(struct amdgpu_dm_connector *aconnector)
+{
+	struct drm_device *drm = aconnector->base.dev;
+	struct amdgpu_display_manager *dm = &drm_to_adev(drm)->dm;
+	struct backlight_properties props = { 0 };
+	struct amdgpu_dm_backlight_caps *caps;
+	char bl_name[16];
+	int min, max;
+
+	if (aconnector->bl_idx == -1)
+		return;
+
+	if (!acpi_video_backlight_use_native()) {
+		drm_info(drm, "Skipping amdgpu DM backlight registration\n");
+		/* Try registering an ACPI video backlight device instead. */
+		acpi_video_register_backlight();
+		return;
+	}
+
+	caps = &dm->backlight_caps[aconnector->bl_idx];
+	if (get_brightness_range(caps, &min, &max)) {
+		if (power_supply_is_system_supplied() > 0)
+			props.brightness = DIV_ROUND_CLOSEST((max - min) * caps->ac_level, 100);
+		else
+			props.brightness = DIV_ROUND_CLOSEST((max - min) * caps->dc_level, 100);
+		/* min is zero, so max needs to be adjusted */
+		props.max_brightness = max - min;
+		drm_dbg(drm, "Backlight caps: min: %d, max: %d, ac %d, dc %d\n", min, max,
+			caps->ac_level, caps->dc_level);
+	} else
+		props.brightness = props.max_brightness = MAX_BACKLIGHT_LEVEL;
+
+	if (caps->data_points && !(amdgpu_dc_debug_mask & DC_DISABLE_CUSTOM_BRIGHTNESS_CURVE)) {
+		drm_info(drm, "Using custom brightness curve\n");
+		props.scale = BACKLIGHT_SCALE_NON_LINEAR;
+	} else
+		props.scale = BACKLIGHT_SCALE_LINEAR;
+	props.type = BACKLIGHT_RAW;
+
+	snprintf(bl_name, sizeof(bl_name), "amdgpu_bl%d",
+		 drm->primary->index + aconnector->bl_idx);
+
+	dm->backlight_dev[aconnector->bl_idx] =
+		backlight_device_register(bl_name, aconnector->base.kdev, dm,
+					  &amdgpu_dm_backlight_ops, &props);
+	dm->brightness[aconnector->bl_idx] = props.brightness;
+
+	if (IS_ERR(dm->backlight_dev[aconnector->bl_idx])) {
+		drm_err(drm, "DM: Backlight registration failed!\n");
+		dm->backlight_dev[aconnector->bl_idx] = NULL;
+	} else
+		drm_dbg_driver(drm, "DM: Registered Backlight device: %s\n", bl_name);
+}
+
+static int initialize_plane(struct amdgpu_display_manager *dm,
+			    struct amdgpu_mode_info *mode_info, int plane_id,
+			    enum drm_plane_type plane_type,
+			    const struct dc_plane_cap *plane_cap)
+{
+	struct drm_plane *plane;
+	unsigned long possible_crtcs;
+	int ret = 0;
+
+	plane = kzalloc(sizeof(struct drm_plane), GFP_KERNEL);
+	if (!plane) {
+		drm_err(adev_to_drm(dm->adev), "KMS: Failed to allocate plane\n");
+		return -ENOMEM;
+	}
+	plane->type = plane_type;
+
+	/*
+	 * HACK: IGT tests expect that the primary plane for a CRTC
+	 * can only have one possible CRTC. Only expose support for
+	 * any CRTC if they're not going to be used as a primary plane
+	 * for a CRTC - like overlay or underlay planes.
+	 */
+	possible_crtcs = 1 << plane_id;
+	if (plane_id >= dm->dc->caps.max_streams)
+		possible_crtcs = 0xff;
+
+	ret = amdgpu_dm_plane_init(dm, plane, possible_crtcs, plane_cap);
+
+	if (ret) {
+		drm_err(adev_to_drm(dm->adev), "KMS: Failed to initialize plane\n");
+		kfree(plane);
+		return ret;
+	}
+
+	if (mode_info)
+		mode_info->planes[plane_id] = plane;
+
+	return ret;
+}
+
+
+static void setup_backlight_device(struct amdgpu_display_manager *dm,
+				   struct amdgpu_dm_connector *aconnector)
+{
+	struct dc_link *link = aconnector->dc_link;
+	int bl_idx = dm->num_of_edps;
+
+	if (!(link->connector_signal & (SIGNAL_TYPE_EDP | SIGNAL_TYPE_LVDS)) ||
+	    link->type == dc_connection_none)
+		return;
+
+	if (dm->num_of_edps >= AMDGPU_DM_MAX_NUM_EDP) {
+		drm_warn(adev_to_drm(dm->adev), "Too much eDP connections, skipping backlight setup for additional eDPs\n");
+		return;
+	}
+
+	aconnector->bl_idx = bl_idx;
+
+	amdgpu_dm_update_backlight_caps(dm, bl_idx);
+	dm->backlight_link[bl_idx] = link;
+	dm->num_of_edps++;
+
+	update_connector_ext_caps(aconnector);
+}
+
+static void amdgpu_set_panel_orientation(struct drm_connector *connector);
+
+/*
+ * In this architecture, the association
+ * connector -> encoder -> crtc
+ * id not really requried. The crtc and connector will hold the
+ * display_index as an abstraction to use with DAL component
+ *
+ * Returns 0 on success
+ */
+static int amdgpu_dm_initialize_drm_device(struct amdgpu_device *adev)
+{
+	struct amdgpu_display_manager *dm = &adev->dm;
+	s32 i;
+	struct amdgpu_dm_connector *aconnector = NULL;
+	struct amdgpu_encoder *aencoder = NULL;
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	u32 link_cnt;
+	s32 primary_planes;
+	enum dc_connection_type new_connection_type = dc_connection_none;
+	const struct dc_plane_cap *plane;
+	bool psr_feature_enabled = false;
+	bool replay_feature_enabled = false;
+	int max_overlay = dm->dc->caps.max_slave_planes;
+
+	dm->display_indexes_num = dm->dc->caps.max_streams;
+	/* Update the actual used number of crtc */
+	adev->mode_info.num_crtc = adev->dm.display_indexes_num;
+
+	amdgpu_dm_set_irq_funcs(adev);
+
+	link_cnt = dm->dc->caps.max_links;
+	if (amdgpu_dm_mode_config_init(dm->adev)) {
+		drm_err(adev_to_drm(adev), "DM: Failed to initialize mode config\n");
+		return -EINVAL;
+	}
+
+	/* There is one primary plane per CRTC */
+	primary_planes = dm->dc->caps.max_streams;
+	if (primary_planes > AMDGPU_MAX_PLANES) {
+		drm_err(adev_to_drm(adev), "DM: Plane nums out of 6 planes\n");
+		return -EINVAL;
+	}
+
+	/*
+	 * Initialize primary planes, implicit planes for legacy IOCTLS.
+	 * Order is reversed to match iteration order in atomic check.
+	 */
+	for (i = (primary_planes - 1); i >= 0; i--) {
+		plane = &dm->dc->caps.planes[i];
+
+		if (initialize_plane(dm, mode_info, i,
+				     DRM_PLANE_TYPE_PRIMARY, plane)) {
+			drm_err(adev_to_drm(adev), "KMS: Failed to initialize primary plane\n");
+			goto fail;
+		}
+	}
+
+	/*
+	 * Initialize overlay planes, index starting after primary planes.
+	 * These planes have a higher DRM index than the primary planes since
+	 * they should be considered as having a higher z-order.
+	 * Order is reversed to match iteration order in atomic check.
+	 *
+	 * Only support DCN for now, and only expose one so we don't encourage
+	 * userspace to use up all the pipes.
+	 */
+	for (i = 0; i < dm->dc->caps.max_planes; ++i) {
+		struct dc_plane_cap *plane = &dm->dc->caps.planes[i];
+
+		/* Do not create overlay if MPO disabled */
+		if (amdgpu_dc_debug_mask & DC_DISABLE_MPO)
+			break;
+
+		if (plane->type != DC_PLANE_TYPE_DCN_UNIVERSAL)
+			continue;
+
+		if (!plane->pixel_format_support.argb8888)
+			continue;
+
+		if (max_overlay-- == 0)
+			break;
+
+		if (initialize_plane(dm, NULL, primary_planes + i,
+				     DRM_PLANE_TYPE_OVERLAY, plane)) {
+			drm_err(adev_to_drm(adev), "KMS: Failed to initialize overlay plane\n");
+			goto fail;
+		}
+	}
+
+	for (i = 0; i < dm->dc->caps.max_streams; i++)
+		if (amdgpu_dm_crtc_init(dm, mode_info->planes[i], i)) {
+			drm_err(adev_to_drm(adev), "KMS: Failed to initialize crtc\n");
+			goto fail;
+		}
+
+	/* Use Outbox interrupt */
+	switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) {
+	case IP_VERSION(3, 0, 0):
+	case IP_VERSION(3, 1, 2):
+	case IP_VERSION(3, 1, 3):
+	case IP_VERSION(3, 1, 4):
+	case IP_VERSION(3, 1, 5):
+	case IP_VERSION(3, 1, 6):
+	case IP_VERSION(3, 2, 0):
+	case IP_VERSION(3, 2, 1):
+	case IP_VERSION(2, 1, 0):
+	case IP_VERSION(3, 5, 0):
+	case IP_VERSION(3, 5, 1):
+	case IP_VERSION(3, 6, 0):
+	case IP_VERSION(4, 0, 1):
+		if (register_outbox_irq_handlers(dm->adev)) {
+			drm_err(adev_to_drm(adev), "DM: Failed to initialize IRQ\n");
+			goto fail;
+		}
+		break;
+	default:
+		DRM_DEBUG_KMS("Unsupported DCN IP version for outbox: 0x%X\n",
+			      amdgpu_ip_version(adev, DCE_HWIP, 0));
+	}
+
+	/* Determine whether to enable PSR support by default. */
+	if (!(amdgpu_dc_debug_mask & DC_DISABLE_PSR)) {
+		switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) {
+		case IP_VERSION(3, 1, 2):
+		case IP_VERSION(3, 1, 3):
+		case IP_VERSION(3, 1, 4):
+		case IP_VERSION(3, 1, 5):
+		case IP_VERSION(3, 1, 6):
+		case IP_VERSION(3, 2, 0):
+		case IP_VERSION(3, 2, 1):
+		case IP_VERSION(3, 5, 0):
+		case IP_VERSION(3, 5, 1):
+		case IP_VERSION(3, 6, 0):
+		case IP_VERSION(4, 0, 1):
+			psr_feature_enabled = true;
+			break;
+		default:
+			psr_feature_enabled = amdgpu_dc_feature_mask & DC_PSR_MASK;
+			break;
+		}
+	}
+
+	/* Determine whether to enable Replay support by default. */
+	if (!(amdgpu_dc_debug_mask & DC_DISABLE_REPLAY)) {
+		switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) {
+		case IP_VERSION(3, 1, 4):
+		case IP_VERSION(3, 2, 0):
+		case IP_VERSION(3, 2, 1):
+		case IP_VERSION(3, 5, 0):
+		case IP_VERSION(3, 5, 1):
+		case IP_VERSION(3, 6, 0):
+			replay_feature_enabled = true;
+			break;
+
+		default:
+			replay_feature_enabled = amdgpu_dc_feature_mask & DC_REPLAY_MASK;
+			break;
+		}
+	}
+
+	if (link_cnt > MAX_LINKS) {
+		drm_err(adev_to_drm(adev),
+			"KMS: Cannot support more than %d display indexes\n",
+				MAX_LINKS);
+		goto fail;
+	}
+
+	/* loops over all connectors on the board */
+	for (i = 0; i < link_cnt; i++) {
+		struct dc_link *link = NULL;
+
+		link = dc_get_link_at_index(dm->dc, i);
+
+		if (link->connector_signal == SIGNAL_TYPE_VIRTUAL) {
+			struct amdgpu_dm_wb_connector *wbcon = kzalloc(sizeof(*wbcon), GFP_KERNEL);
+
+			if (!wbcon) {
+				drm_err(adev_to_drm(adev), "KMS: Failed to allocate writeback connector\n");
+				continue;
+			}
+
+			if (amdgpu_dm_wb_connector_init(dm, wbcon, i)) {
+				drm_err(adev_to_drm(adev), "KMS: Failed to initialize writeback connector\n");
+				kfree(wbcon);
+				continue;
+			}
+
+			link->psr_settings.psr_feature_enabled = false;
+			link->psr_settings.psr_version = DC_PSR_VERSION_UNSUPPORTED;
+
+			continue;
+		}
+
+		aconnector = kzalloc(sizeof(*aconnector), GFP_KERNEL);
+		if (!aconnector)
+			goto fail;
+
+		aencoder = kzalloc(sizeof(*aencoder), GFP_KERNEL);
+		if (!aencoder)
+			goto fail;
+
+		if (amdgpu_dm_encoder_init(dm->ddev, aencoder, i)) {
+			drm_err(adev_to_drm(adev), "KMS: Failed to initialize encoder\n");
+			goto fail;
+		}
+
+		if (amdgpu_dm_connector_init(dm, aconnector, i, aencoder)) {
+			drm_err(adev_to_drm(adev), "KMS: Failed to initialize connector\n");
+			goto fail;
+		}
+
+		if (dm->hpd_rx_offload_wq)
+			dm->hpd_rx_offload_wq[aconnector->base.index].aconnector =
+				aconnector;
+
+		if (!dc_link_detect_connection_type(link, &new_connection_type))
+			drm_err(adev_to_drm(adev), "KMS: Failed to detect connector\n");
+
+		if (aconnector->base.force && new_connection_type == dc_connection_none) {
+			emulated_link_detect(link);
+			amdgpu_dm_update_connector_after_detect(aconnector);
+		} else {
+			bool ret = false;
+
+			mutex_lock(&dm->dc_lock);
+			dc_exit_ips_for_hw_access(dm->dc);
+			ret = dc_link_detect(link, DETECT_REASON_BOOT);
+			mutex_unlock(&dm->dc_lock);
+
+			if (ret) {
+				amdgpu_dm_update_connector_after_detect(aconnector);
+				setup_backlight_device(dm, aconnector);
+
+				/* Disable PSR if Replay can be enabled */
+				if (replay_feature_enabled)
+					if (amdgpu_dm_set_replay_caps(link, aconnector))
+						psr_feature_enabled = false;
+
+				if (psr_feature_enabled) {
+					amdgpu_dm_set_psr_caps(link);
+					drm_info(adev_to_drm(adev), "PSR support %d, DC PSR ver %d, sink PSR ver %d DPCD caps 0x%x su_y_granularity %d\n",
+						 link->psr_settings.psr_feature_enabled,
+						 link->psr_settings.psr_version,
+						 link->dpcd_caps.psr_info.psr_version,
+						 link->dpcd_caps.psr_info.psr_dpcd_caps.raw,
+						 link->dpcd_caps.psr_info.psr2_su_y_granularity_cap);
+				}
+			}
+		}
+		amdgpu_set_panel_orientation(&aconnector->base);
+	}
+
+	/* Software is initialized. Now we can register interrupt handlers. */
+	switch (adev->asic_type) {
+#if defined(CONFIG_DRM_AMD_DC_SI)
+	case CHIP_TAHITI:
+	case CHIP_PITCAIRN:
+	case CHIP_VERDE:
+	case CHIP_OLAND:
+		if (dce60_register_irq_handlers(dm->adev)) {
+			drm_err(adev_to_drm(adev), "DM: Failed to initialize IRQ\n");
+			goto fail;
+		}
+		break;
+#endif
+	case CHIP_BONAIRE:
+	case CHIP_HAWAII:
+	case CHIP_KAVERI:
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+	case CHIP_TONGA:
+	case CHIP_FIJI:
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
+	case CHIP_VEGA10:
+	case CHIP_VEGA12:
+	case CHIP_VEGA20:
+		if (dce110_register_irq_handlers(dm->adev)) {
+			drm_err(adev_to_drm(adev), "DM: Failed to initialize IRQ\n");
+			goto fail;
+		}
+		break;
+	default:
+		switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) {
+		case IP_VERSION(1, 0, 0):
+		case IP_VERSION(1, 0, 1):
+		case IP_VERSION(2, 0, 2):
+		case IP_VERSION(2, 0, 3):
+		case IP_VERSION(2, 0, 0):
+		case IP_VERSION(2, 1, 0):
+		case IP_VERSION(3, 0, 0):
+		case IP_VERSION(3, 0, 2):
+		case IP_VERSION(3, 0, 3):
+		case IP_VERSION(3, 0, 1):
+		case IP_VERSION(3, 1, 2):
+		case IP_VERSION(3, 1, 3):
+		case IP_VERSION(3, 1, 4):
+		case IP_VERSION(3, 1, 5):
+		case IP_VERSION(3, 1, 6):
+		case IP_VERSION(3, 2, 0):
+		case IP_VERSION(3, 2, 1):
+		case IP_VERSION(3, 5, 0):
+		case IP_VERSION(3, 5, 1):
+		case IP_VERSION(3, 6, 0):
+		case IP_VERSION(4, 0, 1):
+			if (dcn10_register_irq_handlers(dm->adev)) {
+				drm_err(adev_to_drm(adev), "DM: Failed to initialize IRQ\n");
+				goto fail;
+			}
+			break;
+		default:
+			drm_err(adev_to_drm(adev), "Unsupported DCE IP versions: 0x%X\n",
+					amdgpu_ip_version(adev, DCE_HWIP, 0));
+			goto fail;
+		}
+		break;
+	}
+
+	return 0;
+fail:
+	kfree(aencoder);
+	kfree(aconnector);
+
+	return -EINVAL;
+}
+
+static void amdgpu_dm_destroy_drm_device(struct amdgpu_display_manager *dm)
+{
+	if (dm->atomic_obj.state)
+		drm_atomic_private_obj_fini(&dm->atomic_obj);
+}
+
+/******************************************************************************
+ * amdgpu_display_funcs functions
+ *****************************************************************************/
+
+/*
+ * dm_bandwidth_update - program display watermarks
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Calculate and program the display watermarks and line buffer allocation.
+ */
+static void dm_bandwidth_update(struct amdgpu_device *adev)
+{
+	/* TODO: implement later */
+}
+
+static const struct amdgpu_display_funcs dm_display_funcs = {
+	.bandwidth_update = dm_bandwidth_update, /* called unconditionally */
+	.vblank_get_counter = dm_vblank_get_counter,/* called unconditionally */
+	.backlight_set_level = NULL, /* never called for DC */
+	.backlight_get_level = NULL, /* never called for DC */
+	.hpd_sense = NULL,/* called unconditionally */
+	.hpd_set_polarity = NULL, /* called unconditionally */
+	.hpd_get_gpio_reg = NULL, /* VBIOS parsing. DAL does it. */
+	.page_flip_get_scanoutpos =
+		dm_crtc_get_scanoutpos,/* called unconditionally */
+	.add_encoder = NULL, /* VBIOS parsing. DAL does it. */
+	.add_connector = NULL, /* VBIOS parsing. DAL does it. */
+};
+
+#if defined(CONFIG_DEBUG_KERNEL_DC)
+
+static ssize_t s3_debug_store(struct device *device,
+			      struct device_attribute *attr,
+			      const char *buf,
+			      size_t count)
+{
+	int ret;
+	int s3_state;
+	struct drm_device *drm_dev = dev_get_drvdata(device);
+	struct amdgpu_device *adev = drm_to_adev(drm_dev);
+	struct amdgpu_ip_block *ip_block;
+
+	ip_block = amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_DCE);
+	if (!ip_block)
+		return -EINVAL;
+
+	ret = kstrtoint(buf, 0, &s3_state);
+
+	if (ret == 0) {
+		if (s3_state) {
+			dm_resume(ip_block);
+			drm_kms_helper_hotplug_event(adev_to_drm(adev));
+		} else
+			dm_suspend(ip_block);
+	}
+
+	return ret == 0 ? count : 0;
+}
+
+DEVICE_ATTR_WO(s3_debug);
+
+#endif
+
+static int dm_init_microcode(struct amdgpu_device *adev)
+{
+	char *fw_name_dmub;
+	int r;
+
+	switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) {
+	case IP_VERSION(2, 1, 0):
+		fw_name_dmub = FIRMWARE_RENOIR_DMUB;
+		if (ASICREV_IS_GREEN_SARDINE(adev->external_rev_id))
+			fw_name_dmub = FIRMWARE_GREEN_SARDINE_DMUB;
+		break;
+	case IP_VERSION(3, 0, 0):
+		if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 3, 0))
+			fw_name_dmub = FIRMWARE_SIENNA_CICHLID_DMUB;
+		else
+			fw_name_dmub = FIRMWARE_NAVY_FLOUNDER_DMUB;
+		break;
+	case IP_VERSION(3, 0, 1):
+		fw_name_dmub = FIRMWARE_VANGOGH_DMUB;
+		break;
+	case IP_VERSION(3, 0, 2):
+		fw_name_dmub = FIRMWARE_DIMGREY_CAVEFISH_DMUB;
+		break;
+	case IP_VERSION(3, 0, 3):
+		fw_name_dmub = FIRMWARE_BEIGE_GOBY_DMUB;
+		break;
+	case IP_VERSION(3, 1, 2):
+	case IP_VERSION(3, 1, 3):
+		fw_name_dmub = FIRMWARE_YELLOW_CARP_DMUB;
+		break;
+	case IP_VERSION(3, 1, 4):
+		fw_name_dmub = FIRMWARE_DCN_314_DMUB;
+		break;
+	case IP_VERSION(3, 1, 5):
+		fw_name_dmub = FIRMWARE_DCN_315_DMUB;
+		break;
+	case IP_VERSION(3, 1, 6):
+		fw_name_dmub = FIRMWARE_DCN316_DMUB;
+		break;
+	case IP_VERSION(3, 2, 0):
+		fw_name_dmub = FIRMWARE_DCN_V3_2_0_DMCUB;
+		break;
+	case IP_VERSION(3, 2, 1):
+		fw_name_dmub = FIRMWARE_DCN_V3_2_1_DMCUB;
+		break;
+	case IP_VERSION(3, 5, 0):
+		fw_name_dmub = FIRMWARE_DCN_35_DMUB;
+		break;
+	case IP_VERSION(3, 5, 1):
+		fw_name_dmub = FIRMWARE_DCN_351_DMUB;
+		break;
+	case IP_VERSION(3, 6, 0):
+		fw_name_dmub = FIRMWARE_DCN_36_DMUB;
+		break;
+	case IP_VERSION(4, 0, 1):
+		fw_name_dmub = FIRMWARE_DCN_401_DMUB;
+		break;
+	default:
+		/* ASIC doesn't support DMUB. */
+		return 0;
+	}
+	r = amdgpu_ucode_request(adev, &adev->dm.dmub_fw, AMDGPU_UCODE_REQUIRED,
+				 "%s", fw_name_dmub);
+	return r;
+}
+
+static int dm_early_init(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	struct atom_context *ctx = mode_info->atom_context;
+	int index = GetIndexIntoMasterTable(DATA, Object_Header);
+	u16 data_offset;
+
+	/* if there is no object header, skip DM */
+	if (!amdgpu_atom_parse_data_header(ctx, index, NULL, NULL, NULL, &data_offset)) {
+		adev->harvest_ip_mask |= AMD_HARVEST_IP_DMU_MASK;
+		drm_info(adev_to_drm(adev), "No object header, skipping DM\n");
+		return -ENOENT;
+	}
+
+	switch (adev->asic_type) {
+#if defined(CONFIG_DRM_AMD_DC_SI)
+	case CHIP_TAHITI:
+	case CHIP_PITCAIRN:
+	case CHIP_VERDE:
+		adev->mode_info.num_crtc = 6;
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 6;
+		break;
+	case CHIP_OLAND:
+		adev->mode_info.num_crtc = 2;
+		adev->mode_info.num_hpd = 2;
+		adev->mode_info.num_dig = 2;
+		break;
+#endif
+	case CHIP_BONAIRE:
+	case CHIP_HAWAII:
+		adev->mode_info.num_crtc = 6;
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 6;
+		break;
+	case CHIP_KAVERI:
+		adev->mode_info.num_crtc = 4;
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 7;
+		break;
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+		adev->mode_info.num_crtc = 2;
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 6;
+		break;
+	case CHIP_FIJI:
+	case CHIP_TONGA:
+		adev->mode_info.num_crtc = 6;
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 7;
+		break;
+	case CHIP_CARRIZO:
+		adev->mode_info.num_crtc = 3;
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 9;
+		break;
+	case CHIP_STONEY:
+		adev->mode_info.num_crtc = 2;
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 9;
+		break;
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+		adev->mode_info.num_crtc = 5;
+		adev->mode_info.num_hpd = 5;
+		adev->mode_info.num_dig = 5;
+		break;
+	case CHIP_POLARIS10:
+	case CHIP_VEGAM:
+		adev->mode_info.num_crtc = 6;
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 6;
+		break;
+	case CHIP_VEGA10:
+	case CHIP_VEGA12:
+	case CHIP_VEGA20:
+		adev->mode_info.num_crtc = 6;
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 6;
+		break;
+	default:
+
+		switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) {
+		case IP_VERSION(2, 0, 2):
+		case IP_VERSION(3, 0, 0):
+			adev->mode_info.num_crtc = 6;
+			adev->mode_info.num_hpd = 6;
+			adev->mode_info.num_dig = 6;
+			break;
+		case IP_VERSION(2, 0, 0):
+		case IP_VERSION(3, 0, 2):
+			adev->mode_info.num_crtc = 5;
+			adev->mode_info.num_hpd = 5;
+			adev->mode_info.num_dig = 5;
+			break;
+		case IP_VERSION(2, 0, 3):
+		case IP_VERSION(3, 0, 3):
+			adev->mode_info.num_crtc = 2;
+			adev->mode_info.num_hpd = 2;
+			adev->mode_info.num_dig = 2;
+			break;
+		case IP_VERSION(1, 0, 0):
+		case IP_VERSION(1, 0, 1):
+		case IP_VERSION(3, 0, 1):
+		case IP_VERSION(2, 1, 0):
+		case IP_VERSION(3, 1, 2):
+		case IP_VERSION(3, 1, 3):
+		case IP_VERSION(3, 1, 4):
+		case IP_VERSION(3, 1, 5):
+		case IP_VERSION(3, 1, 6):
+		case IP_VERSION(3, 2, 0):
+		case IP_VERSION(3, 2, 1):
+		case IP_VERSION(3, 5, 0):
+		case IP_VERSION(3, 5, 1):
+		case IP_VERSION(3, 6, 0):
+		case IP_VERSION(4, 0, 1):
+			adev->mode_info.num_crtc = 4;
+			adev->mode_info.num_hpd = 4;
+			adev->mode_info.num_dig = 4;
+			break;
+		default:
+			drm_err(adev_to_drm(adev), "Unsupported DCE IP versions: 0x%x\n",
+					amdgpu_ip_version(adev, DCE_HWIP, 0));
+			return -EINVAL;
+		}
+		break;
+	}
+
+	if (adev->mode_info.funcs == NULL)
+		adev->mode_info.funcs = &dm_display_funcs;
+
+	/*
+	 * Note: Do NOT change adev->audio_endpt_rreg and
+	 * adev->audio_endpt_wreg because they are initialised in
+	 * amdgpu_device_init()
+	 */
+#if defined(CONFIG_DEBUG_KERNEL_DC)
+	device_create_file(
+		adev_to_drm(adev)->dev,
+		&dev_attr_s3_debug);
+#endif
+	adev->dc_enabled = true;
+
+	return dm_init_microcode(adev);
+}
+
+static bool modereset_required(struct drm_crtc_state *crtc_state)
+{
+	return !crtc_state->active && drm_atomic_crtc_needs_modeset(crtc_state);
+}
+
+static void amdgpu_dm_encoder_destroy(struct drm_encoder *encoder)
+{
+	drm_encoder_cleanup(encoder);
+	kfree(encoder);
+}
+
+static const struct drm_encoder_funcs amdgpu_dm_encoder_funcs = {
+	.destroy = amdgpu_dm_encoder_destroy,
+};
+
+static int
+fill_plane_color_attributes(const struct drm_plane_state *plane_state,
+			    const enum surface_pixel_format format,
+			    enum dc_color_space *color_space)
+{
+	bool full_range;
+
+	*color_space = COLOR_SPACE_SRGB;
+
+	/* DRM color properties only affect non-RGB formats. */
+	if (format < SURFACE_PIXEL_FORMAT_VIDEO_BEGIN)
+		return 0;
+
+	full_range = (plane_state->color_range == DRM_COLOR_YCBCR_FULL_RANGE);
+
+	switch (plane_state->color_encoding) {
+	case DRM_COLOR_YCBCR_BT601:
+		if (full_range)
+			*color_space = COLOR_SPACE_YCBCR601;
+		else
+			*color_space = COLOR_SPACE_YCBCR601_LIMITED;
+		break;
+
+	case DRM_COLOR_YCBCR_BT709:
+		if (full_range)
+			*color_space = COLOR_SPACE_YCBCR709;
+		else
+			*color_space = COLOR_SPACE_YCBCR709_LIMITED;
+		break;
+
+	case DRM_COLOR_YCBCR_BT2020:
+		if (full_range)
+			*color_space = COLOR_SPACE_2020_YCBCR_FULL;
+		else
+			*color_space = COLOR_SPACE_2020_YCBCR_LIMITED;
+		break;
+
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int
+fill_dc_plane_info_and_addr(struct amdgpu_device *adev,
+			    const struct drm_plane_state *plane_state,
+			    const u64 tiling_flags,
+			    struct dc_plane_info *plane_info,
+			    struct dc_plane_address *address,
+			    bool tmz_surface)
+{
+	const struct drm_framebuffer *fb = plane_state->fb;
+	const struct amdgpu_framebuffer *afb =
+		to_amdgpu_framebuffer(plane_state->fb);
+	int ret;
+
+	memset(plane_info, 0, sizeof(*plane_info));
+
+	switch (fb->format->format) {
+	case DRM_FORMAT_C8:
+		plane_info->format =
+			SURFACE_PIXEL_FORMAT_GRPH_PALETA_256_COLORS;
+		break;
+	case DRM_FORMAT_RGB565:
+		plane_info->format = SURFACE_PIXEL_FORMAT_GRPH_RGB565;
+		break;
+	case DRM_FORMAT_XRGB8888:
+	case DRM_FORMAT_ARGB8888:
+		plane_info->format = SURFACE_PIXEL_FORMAT_GRPH_ARGB8888;
+		break;
+	case DRM_FORMAT_XRGB2101010:
+	case DRM_FORMAT_ARGB2101010:
+		plane_info->format = SURFACE_PIXEL_FORMAT_GRPH_ARGB2101010;
+		break;
+	case DRM_FORMAT_XBGR2101010:
+	case DRM_FORMAT_ABGR2101010:
+		plane_info->format = SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010;
+		break;
+	case DRM_FORMAT_XBGR8888:
+	case DRM_FORMAT_ABGR8888:
+		plane_info->format = SURFACE_PIXEL_FORMAT_GRPH_ABGR8888;
+		break;
+	case DRM_FORMAT_NV21:
+		plane_info->format = SURFACE_PIXEL_FORMAT_VIDEO_420_YCbCr;
+		break;
+	case DRM_FORMAT_NV12:
+		plane_info->format = SURFACE_PIXEL_FORMAT_VIDEO_420_YCrCb;
+		break;
+	case DRM_FORMAT_P010:
+		plane_info->format = SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCrCb;
+		break;
+	case DRM_FORMAT_XRGB16161616F:
+	case DRM_FORMAT_ARGB16161616F:
+		plane_info->format = SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616F;
+		break;
+	case DRM_FORMAT_XBGR16161616F:
+	case DRM_FORMAT_ABGR16161616F:
+		plane_info->format = SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F;
+		break;
+	case DRM_FORMAT_XRGB16161616:
+	case DRM_FORMAT_ARGB16161616:
+		plane_info->format = SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616;
+		break;
+	case DRM_FORMAT_XBGR16161616:
+	case DRM_FORMAT_ABGR16161616:
+		plane_info->format = SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616;
+		break;
+	default:
+		drm_err(adev_to_drm(adev),
+			"Unsupported screen format %p4cc\n",
+			&fb->format->format);
+		return -EINVAL;
+	}
+
+	switch (plane_state->rotation & DRM_MODE_ROTATE_MASK) {
+	case DRM_MODE_ROTATE_0:
+		plane_info->rotation = ROTATION_ANGLE_0;
+		break;
+	case DRM_MODE_ROTATE_90:
+		plane_info->rotation = ROTATION_ANGLE_90;
+		break;
+	case DRM_MODE_ROTATE_180:
+		plane_info->rotation = ROTATION_ANGLE_180;
+		break;
+	case DRM_MODE_ROTATE_270:
+		plane_info->rotation = ROTATION_ANGLE_270;
+		break;
+	default:
+		plane_info->rotation = ROTATION_ANGLE_0;
+		break;
+	}
+
+
+	plane_info->visible = true;
+	plane_info->stereo_format = PLANE_STEREO_FORMAT_NONE;
+
+	plane_info->layer_index = plane_state->normalized_zpos;
+
+	ret = fill_plane_color_attributes(plane_state, plane_info->format,
+					  &plane_info->color_space);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_dm_plane_fill_plane_buffer_attributes(adev, afb, plane_info->format,
+					   plane_info->rotation, tiling_flags,
+					   &plane_info->tiling_info,
+					   &plane_info->plane_size,
+					   &plane_info->dcc, address,
+					   tmz_surface);
+	if (ret)
+		return ret;
+
+	amdgpu_dm_plane_fill_blending_from_plane_state(
+		plane_state, &plane_info->per_pixel_alpha, &plane_info->pre_multiplied_alpha,
+		&plane_info->global_alpha, &plane_info->global_alpha_value);
+
+	return 0;
+}
+
+static int fill_dc_plane_attributes(struct amdgpu_device *adev,
+				    struct dc_plane_state *dc_plane_state,
+				    struct drm_plane_state *plane_state,
+				    struct drm_crtc_state *crtc_state)
+{
+	struct dm_crtc_state *dm_crtc_state = to_dm_crtc_state(crtc_state);
+	struct amdgpu_framebuffer *afb = (struct amdgpu_framebuffer *)plane_state->fb;
+	struct dc_scaling_info scaling_info;
+	struct dc_plane_info plane_info;
+	int ret;
+
+	ret = amdgpu_dm_plane_fill_dc_scaling_info(adev, plane_state, &scaling_info);
+	if (ret)
+		return ret;
+
+	dc_plane_state->src_rect = scaling_info.src_rect;
+	dc_plane_state->dst_rect = scaling_info.dst_rect;
+	dc_plane_state->clip_rect = scaling_info.clip_rect;
+	dc_plane_state->scaling_quality = scaling_info.scaling_quality;
+
+	ret = fill_dc_plane_info_and_addr(adev, plane_state,
+					  afb->tiling_flags,
+					  &plane_info,
+					  &dc_plane_state->address,
+					  afb->tmz_surface);
+	if (ret)
+		return ret;
+
+	dc_plane_state->format = plane_info.format;
+	dc_plane_state->color_space = plane_info.color_space;
+	dc_plane_state->format = plane_info.format;
+	dc_plane_state->plane_size = plane_info.plane_size;
+	dc_plane_state->rotation = plane_info.rotation;
+	dc_plane_state->horizontal_mirror = plane_info.horizontal_mirror;
+	dc_plane_state->stereo_format = plane_info.stereo_format;
+	dc_plane_state->tiling_info = plane_info.tiling_info;
+	dc_plane_state->visible = plane_info.visible;
+	dc_plane_state->per_pixel_alpha = plane_info.per_pixel_alpha;
+	dc_plane_state->pre_multiplied_alpha = plane_info.pre_multiplied_alpha;
+	dc_plane_state->global_alpha = plane_info.global_alpha;
+	dc_plane_state->global_alpha_value = plane_info.global_alpha_value;
+	dc_plane_state->dcc = plane_info.dcc;
+	dc_plane_state->layer_index = plane_info.layer_index;
+	dc_plane_state->flip_int_enabled = true;
+
+	/*
+	 * Always set input transfer function, since plane state is refreshed
+	 * every time.
+	 */
+	ret = amdgpu_dm_update_plane_color_mgmt(dm_crtc_state,
+						plane_state,
+						dc_plane_state);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static inline void fill_dc_dirty_rect(struct drm_plane *plane,
+				      struct rect *dirty_rect, int32_t x,
+				      s32 y, s32 width, s32 height,
+				      int *i, bool ffu)
+{
+	WARN_ON(*i >= DC_MAX_DIRTY_RECTS);
+
+	dirty_rect->x = x;
+	dirty_rect->y = y;
+	dirty_rect->width = width;
+	dirty_rect->height = height;
+
+	if (ffu)
+		drm_dbg(plane->dev,
+			"[PLANE:%d] PSR FFU dirty rect size (%d, %d)\n",
+			plane->base.id, width, height);
+	else
+		drm_dbg(plane->dev,
+			"[PLANE:%d] PSR SU dirty rect at (%d, %d) size (%d, %d)",
+			plane->base.id, x, y, width, height);
+
+	(*i)++;
+}
+
+/**
+ * fill_dc_dirty_rects() - Fill DC dirty regions for PSR selective updates
+ *
+ * @plane: DRM plane containing dirty regions that need to be flushed to the eDP
+ *         remote fb
+ * @old_plane_state: Old state of @plane
+ * @new_plane_state: New state of @plane
+ * @crtc_state: New state of CRTC connected to the @plane
+ * @flip_addrs: DC flip tracking struct, which also tracts dirty rects
+ * @is_psr_su: Flag indicating whether Panel Self Refresh Selective Update (PSR SU) is enabled.
+ *             If PSR SU is enabled and damage clips are available, only the regions of the screen
+ *             that have changed will be updated. If PSR SU is not enabled,
+ *             or if damage clips are not available, the entire screen will be updated.
+ * @dirty_regions_changed: dirty regions changed
+ *
+ * For PSR SU, DC informs the DMUB uController of dirty rectangle regions
+ * (referred to as "damage clips" in DRM nomenclature) that require updating on
+ * the eDP remote buffer. The responsibility of specifying the dirty regions is
+ * amdgpu_dm's.
+ *
+ * A damage-aware DRM client should fill the FB_DAMAGE_CLIPS property on the
+ * plane with regions that require flushing to the eDP remote buffer. In
+ * addition, certain use cases - such as cursor and multi-plane overlay (MPO) -
+ * implicitly provide damage clips without any client support via the plane
+ * bounds.
+ */
+static void fill_dc_dirty_rects(struct drm_plane *plane,
+				struct drm_plane_state *old_plane_state,
+				struct drm_plane_state *new_plane_state,
+				struct drm_crtc_state *crtc_state,
+				struct dc_flip_addrs *flip_addrs,
+				bool is_psr_su,
+				bool *dirty_regions_changed)
+{
+	struct dm_crtc_state *dm_crtc_state = to_dm_crtc_state(crtc_state);
+	struct rect *dirty_rects = flip_addrs->dirty_rects;
+	u32 num_clips;
+	struct drm_mode_rect *clips;
+	bool bb_changed;
+	bool fb_changed;
+	u32 i = 0;
+	*dirty_regions_changed = false;
+
+	/*
+	 * Cursor plane has it's own dirty rect update interface. See
+	 * dcn10_dmub_update_cursor_data and dmub_cmd_update_cursor_info_data
+	 */
+	if (plane->type == DRM_PLANE_TYPE_CURSOR)
+		return;
+
+	if (new_plane_state->rotation != DRM_MODE_ROTATE_0)
+		goto ffu;
+
+	num_clips = drm_plane_get_damage_clips_count(new_plane_state);
+	clips = drm_plane_get_damage_clips(new_plane_state);
+
+	if (num_clips && (!amdgpu_damage_clips || (amdgpu_damage_clips < 0 &&
+						   is_psr_su)))
+		goto ffu;
+
+	if (!dm_crtc_state->mpo_requested) {
+		if (!num_clips || num_clips > DC_MAX_DIRTY_RECTS)
+			goto ffu;
+
+		for (; flip_addrs->dirty_rect_count < num_clips; clips++)
+			fill_dc_dirty_rect(new_plane_state->plane,
+					   &dirty_rects[flip_addrs->dirty_rect_count],
+					   clips->x1, clips->y1,
+					   clips->x2 - clips->x1, clips->y2 - clips->y1,
+					   &flip_addrs->dirty_rect_count,
+					   false);
+		return;
+	}
+
+	/*
+	 * MPO is requested. Add entire plane bounding box to dirty rects if
+	 * flipped to or damaged.
+	 *
+	 * If plane is moved or resized, also add old bounding box to dirty
+	 * rects.
+	 */
+	fb_changed = old_plane_state->fb->base.id !=
+		     new_plane_state->fb->base.id;
+	bb_changed = (old_plane_state->crtc_x != new_plane_state->crtc_x ||
+		      old_plane_state->crtc_y != new_plane_state->crtc_y ||
+		      old_plane_state->crtc_w != new_plane_state->crtc_w ||
+		      old_plane_state->crtc_h != new_plane_state->crtc_h);
+
+	drm_dbg(plane->dev,
+		"[PLANE:%d] PSR bb_changed:%d fb_changed:%d num_clips:%d\n",
+		new_plane_state->plane->base.id,
+		bb_changed, fb_changed, num_clips);
+
+	*dirty_regions_changed = bb_changed;
+
+	if ((num_clips + (bb_changed ? 2 : 0)) > DC_MAX_DIRTY_RECTS)
+		goto ffu;
+
+	if (bb_changed) {
+		fill_dc_dirty_rect(new_plane_state->plane, &dirty_rects[i],
+				   new_plane_state->crtc_x,
+				   new_plane_state->crtc_y,
+				   new_plane_state->crtc_w,
+				   new_plane_state->crtc_h, &i, false);
+
+		/* Add old plane bounding-box if plane is moved or resized */
+		fill_dc_dirty_rect(new_plane_state->plane, &dirty_rects[i],
+				   old_plane_state->crtc_x,
+				   old_plane_state->crtc_y,
+				   old_plane_state->crtc_w,
+				   old_plane_state->crtc_h, &i, false);
+	}
+
+	if (num_clips) {
+		for (; i < num_clips; clips++)
+			fill_dc_dirty_rect(new_plane_state->plane,
+					   &dirty_rects[i], clips->x1,
+					   clips->y1, clips->x2 - clips->x1,
+					   clips->y2 - clips->y1, &i, false);
+	} else if (fb_changed && !bb_changed) {
+		fill_dc_dirty_rect(new_plane_state->plane, &dirty_rects[i],
+				   new_plane_state->crtc_x,
+				   new_plane_state->crtc_y,
+				   new_plane_state->crtc_w,
+				   new_plane_state->crtc_h, &i, false);
+	}
+
+	flip_addrs->dirty_rect_count = i;
+	return;
+
+ffu:
+	fill_dc_dirty_rect(new_plane_state->plane, &dirty_rects[0], 0, 0,
+			   dm_crtc_state->base.mode.crtc_hdisplay,
+			   dm_crtc_state->base.mode.crtc_vdisplay,
+			   &flip_addrs->dirty_rect_count, true);
+}
+
+static void update_stream_scaling_settings(const struct drm_display_mode *mode,
+					   const struct dm_connector_state *dm_state,
+					   struct dc_stream_state *stream)
+{
+	enum amdgpu_rmx_type rmx_type;
+
+	struct rect src = { 0 }; /* viewport in composition space*/
+	struct rect dst = { 0 }; /* stream addressable area */
+
+	/* no mode. nothing to be done */
+	if (!mode)
+		return;
+
+	/* Full screen scaling by default */
+	src.width = mode->hdisplay;
+	src.height = mode->vdisplay;
+	dst.width = stream->timing.h_addressable;
+	dst.height = stream->timing.v_addressable;
+
+	if (dm_state) {
+		rmx_type = dm_state->scaling;
+		if (rmx_type == RMX_ASPECT || rmx_type == RMX_OFF) {
+			if (src.width * dst.height <
+					src.height * dst.width) {
+				/* height needs less upscaling/more downscaling */
+				dst.width = src.width *
+						dst.height / src.height;
+			} else {
+				/* width needs less upscaling/more downscaling */
+				dst.height = src.height *
+						dst.width / src.width;
+			}
+		} else if (rmx_type == RMX_CENTER) {
+			dst = src;
+		}
+
+		dst.x = (stream->timing.h_addressable - dst.width) / 2;
+		dst.y = (stream->timing.v_addressable - dst.height) / 2;
+
+		if (dm_state->underscan_enable) {
+			dst.x += dm_state->underscan_hborder / 2;
+			dst.y += dm_state->underscan_vborder / 2;
+			dst.width -= dm_state->underscan_hborder;
+			dst.height -= dm_state->underscan_vborder;
+		}
+	}
+
+	stream->src = src;
+	stream->dst = dst;
+
+	DRM_DEBUG_KMS("Destination Rectangle x:%d  y:%d  width:%d  height:%d\n",
+		      dst.x, dst.y, dst.width, dst.height);
+
+}
+
+static enum dc_color_depth
+convert_color_depth_from_display_info(const struct drm_connector *connector,
+				      bool is_y420, int requested_bpc)
+{
+	u8 bpc;
+
+	if (is_y420) {
+		bpc = 8;
+
+		/* Cap display bpc based on HDMI 2.0 HF-VSDB */
+		if (connector->display_info.hdmi.y420_dc_modes & DRM_EDID_YCBCR420_DC_48)
+			bpc = 16;
+		else if (connector->display_info.hdmi.y420_dc_modes & DRM_EDID_YCBCR420_DC_36)
+			bpc = 12;
+		else if (connector->display_info.hdmi.y420_dc_modes & DRM_EDID_YCBCR420_DC_30)
+			bpc = 10;
+	} else {
+		bpc = (uint8_t)connector->display_info.bpc;
+		/* Assume 8 bpc by default if no bpc is specified. */
+		bpc = bpc ? bpc : 8;
+	}
+
+	if (requested_bpc > 0) {
+		/*
+		 * Cap display bpc based on the user requested value.
+		 *
+		 * The value for state->max_bpc may not correctly updated
+		 * depending on when the connector gets added to the state
+		 * or if this was called outside of atomic check, so it
+		 * can't be used directly.
+		 */
+		bpc = min_t(u8, bpc, requested_bpc);
+
+		/* Round down to the nearest even number. */
+		bpc = bpc - (bpc & 1);
+	}
+
+	switch (bpc) {
+	case 0:
+		/*
+		 * Temporary Work around, DRM doesn't parse color depth for
+		 * EDID revision before 1.4
+		 * TODO: Fix edid parsing
+		 */
+		return COLOR_DEPTH_888;
+	case 6:
+		return COLOR_DEPTH_666;
+	case 8:
+		return COLOR_DEPTH_888;
+	case 10:
+		return COLOR_DEPTH_101010;
+	case 12:
+		return COLOR_DEPTH_121212;
+	case 14:
+		return COLOR_DEPTH_141414;
+	case 16:
+		return COLOR_DEPTH_161616;
+	default:
+		return COLOR_DEPTH_UNDEFINED;
+	}
+}
+
+static enum dc_aspect_ratio
+get_aspect_ratio(const struct drm_display_mode *mode_in)
+{
+	/* 1-1 mapping, since both enums follow the HDMI spec. */
+	return (enum dc_aspect_ratio) mode_in->picture_aspect_ratio;
+}
+
+static enum dc_color_space
+get_output_color_space(const struct dc_crtc_timing *dc_crtc_timing,
+		       const struct drm_connector_state *connector_state)
+{
+	enum dc_color_space color_space = COLOR_SPACE_SRGB;
+
+	switch (connector_state->colorspace) {
+	case DRM_MODE_COLORIMETRY_BT601_YCC:
+		if (dc_crtc_timing->flags.Y_ONLY)
+			color_space = COLOR_SPACE_YCBCR601_LIMITED;
+		else
+			color_space = COLOR_SPACE_YCBCR601;
+		break;
+	case DRM_MODE_COLORIMETRY_BT709_YCC:
+		if (dc_crtc_timing->flags.Y_ONLY)
+			color_space = COLOR_SPACE_YCBCR709_LIMITED;
+		else
+			color_space = COLOR_SPACE_YCBCR709;
+		break;
+	case DRM_MODE_COLORIMETRY_OPRGB:
+		color_space = COLOR_SPACE_ADOBERGB;
+		break;
+	case DRM_MODE_COLORIMETRY_BT2020_RGB:
+	case DRM_MODE_COLORIMETRY_BT2020_YCC:
+		if (dc_crtc_timing->pixel_encoding == PIXEL_ENCODING_RGB)
+			color_space = COLOR_SPACE_2020_RGB_FULLRANGE;
+		else
+			color_space = COLOR_SPACE_2020_YCBCR_LIMITED;
+		break;
+	case DRM_MODE_COLORIMETRY_DEFAULT: // ITU601
+	default:
+		if (dc_crtc_timing->pixel_encoding == PIXEL_ENCODING_RGB) {
+			color_space = COLOR_SPACE_SRGB;
+			if (connector_state->hdmi.broadcast_rgb == DRM_HDMI_BROADCAST_RGB_LIMITED)
+				color_space = COLOR_SPACE_SRGB_LIMITED;
+		/*
+		 * 27030khz is the separation point between HDTV and SDTV
+		 * according to HDMI spec, we use YCbCr709 and YCbCr601
+		 * respectively
+		 */
+		} else if (dc_crtc_timing->pix_clk_100hz > 270300) {
+			if (dc_crtc_timing->flags.Y_ONLY)
+				color_space =
+					COLOR_SPACE_YCBCR709_LIMITED;
+			else
+				color_space = COLOR_SPACE_YCBCR709;
+		} else {
+			if (dc_crtc_timing->flags.Y_ONLY)
+				color_space =
+					COLOR_SPACE_YCBCR601_LIMITED;
+			else
+				color_space = COLOR_SPACE_YCBCR601;
+		}
+		break;
+	}
+
+	return color_space;
+}
+
+static enum display_content_type
+get_output_content_type(const struct drm_connector_state *connector_state)
+{
+	switch (connector_state->content_type) {
+	default:
+	case DRM_MODE_CONTENT_TYPE_NO_DATA:
+		return DISPLAY_CONTENT_TYPE_NO_DATA;
+	case DRM_MODE_CONTENT_TYPE_GRAPHICS:
+		return DISPLAY_CONTENT_TYPE_GRAPHICS;
+	case DRM_MODE_CONTENT_TYPE_PHOTO:
+		return DISPLAY_CONTENT_TYPE_PHOTO;
+	case DRM_MODE_CONTENT_TYPE_CINEMA:
+		return DISPLAY_CONTENT_TYPE_CINEMA;
+	case DRM_MODE_CONTENT_TYPE_GAME:
+		return DISPLAY_CONTENT_TYPE_GAME;
+	}
+}
+
+static bool adjust_colour_depth_from_display_info(
+	struct dc_crtc_timing *timing_out,
+	const struct drm_display_info *info)
+{
+	enum dc_color_depth depth = timing_out->display_color_depth;
+	int normalized_clk;
+
+	do {
+		normalized_clk = timing_out->pix_clk_100hz / 10;
+		/* YCbCr 4:2:0 requires additional adjustment of 1/2 */
+		if (timing_out->pixel_encoding == PIXEL_ENCODING_YCBCR420)
+			normalized_clk /= 2;
+		/* Adjusting pix clock following on HDMI spec based on colour depth */
+		switch (depth) {
+		case COLOR_DEPTH_888:
+			break;
+		case COLOR_DEPTH_101010:
+			normalized_clk = (normalized_clk * 30) / 24;
+			break;
+		case COLOR_DEPTH_121212:
+			normalized_clk = (normalized_clk * 36) / 24;
+			break;
+		case COLOR_DEPTH_161616:
+			normalized_clk = (normalized_clk * 48) / 24;
+			break;
+		default:
+			/* The above depths are the only ones valid for HDMI. */
+			return false;
+		}
+		if (normalized_clk <= info->max_tmds_clock) {
+			timing_out->display_color_depth = depth;
+			return true;
+		}
+	} while (--depth > COLOR_DEPTH_666);
+	return false;
+}
+
+static void fill_stream_properties_from_drm_display_mode(
+	struct dc_stream_state *stream,
+	const struct drm_display_mode *mode_in,
+	const struct drm_connector *connector,
+	const struct drm_connector_state *connector_state,
+	const struct dc_stream_state *old_stream,
+	int requested_bpc)
+{
+	struct dc_crtc_timing *timing_out = &stream->timing;
+	const struct drm_display_info *info = &connector->display_info;
+	struct amdgpu_dm_connector *aconnector = NULL;
+	struct hdmi_vendor_infoframe hv_frame;
+	struct hdmi_avi_infoframe avi_frame;
+	ssize_t err;
+
+	if (connector->connector_type != DRM_MODE_CONNECTOR_WRITEBACK)
+		aconnector = to_amdgpu_dm_connector(connector);
+
+	memset(&hv_frame, 0, sizeof(hv_frame));
+	memset(&avi_frame, 0, sizeof(avi_frame));
+
+	timing_out->h_border_left = 0;
+	timing_out->h_border_right = 0;
+	timing_out->v_border_top = 0;
+	timing_out->v_border_bottom = 0;
+	/* TODO: un-hardcode */
+	if (drm_mode_is_420_only(info, mode_in)
+			&& stream->signal == SIGNAL_TYPE_HDMI_TYPE_A)
+		timing_out->pixel_encoding = PIXEL_ENCODING_YCBCR420;
+	else if (drm_mode_is_420_also(info, mode_in)
+			&& aconnector
+			&& aconnector->force_yuv420_output)
+		timing_out->pixel_encoding = PIXEL_ENCODING_YCBCR420;
+	else if ((connector->display_info.color_formats & DRM_COLOR_FORMAT_YCBCR422)
+			&& aconnector
+			&& aconnector->force_yuv422_output)
+		timing_out->pixel_encoding = PIXEL_ENCODING_YCBCR422;
+	else if ((connector->display_info.color_formats & DRM_COLOR_FORMAT_YCBCR444)
+			&& stream->signal == SIGNAL_TYPE_HDMI_TYPE_A)
+		timing_out->pixel_encoding = PIXEL_ENCODING_YCBCR444;
+	else
+		timing_out->pixel_encoding = PIXEL_ENCODING_RGB;
+
+	timing_out->timing_3d_format = TIMING_3D_FORMAT_NONE;
+	timing_out->display_color_depth = convert_color_depth_from_display_info(
+		connector,
+		(timing_out->pixel_encoding == PIXEL_ENCODING_YCBCR420),
+		requested_bpc);
+	timing_out->scan_type = SCANNING_TYPE_NODATA;
+	timing_out->hdmi_vic = 0;
+
+	if (old_stream) {
+		timing_out->vic = old_stream->timing.vic;
+		timing_out->flags.HSYNC_POSITIVE_POLARITY = old_stream->timing.flags.HSYNC_POSITIVE_POLARITY;
+		timing_out->flags.VSYNC_POSITIVE_POLARITY = old_stream->timing.flags.VSYNC_POSITIVE_POLARITY;
+	} else {
+		timing_out->vic = drm_match_cea_mode(mode_in);
+		if (mode_in->flags & DRM_MODE_FLAG_PHSYNC)
+			timing_out->flags.HSYNC_POSITIVE_POLARITY = 1;
+		if (mode_in->flags & DRM_MODE_FLAG_PVSYNC)
+			timing_out->flags.VSYNC_POSITIVE_POLARITY = 1;
+	}
+
+	if (stream->signal == SIGNAL_TYPE_HDMI_TYPE_A) {
+		err = drm_hdmi_avi_infoframe_from_display_mode(&avi_frame,
+							       (struct drm_connector *)connector,
+							       mode_in);
+		if (err < 0)
+			drm_warn_once(connector->dev, "Failed to setup avi infoframe on connector %s: %zd\n",
+				      connector->name, err);
+		timing_out->vic = avi_frame.video_code;
+		err = drm_hdmi_vendor_infoframe_from_display_mode(&hv_frame,
+								  (struct drm_connector *)connector,
+								  mode_in);
+		if (err < 0)
+			drm_warn_once(connector->dev, "Failed to setup vendor infoframe on connector %s: %zd\n",
+				      connector->name, err);
+		timing_out->hdmi_vic = hv_frame.vic;
+	}
+
+	if (aconnector && is_freesync_video_mode(mode_in, aconnector)) {
+		timing_out->h_addressable = mode_in->hdisplay;
+		timing_out->h_total = mode_in->htotal;
+		timing_out->h_sync_width = mode_in->hsync_end - mode_in->hsync_start;
+		timing_out->h_front_porch = mode_in->hsync_start - mode_in->hdisplay;
+		timing_out->v_total = mode_in->vtotal;
+		timing_out->v_addressable = mode_in->vdisplay;
+		timing_out->v_front_porch = mode_in->vsync_start - mode_in->vdisplay;
+		timing_out->v_sync_width = mode_in->vsync_end - mode_in->vsync_start;
+		timing_out->pix_clk_100hz = mode_in->clock * 10;
+	} else {
+		timing_out->h_addressable = mode_in->crtc_hdisplay;
+		timing_out->h_total = mode_in->crtc_htotal;
+		timing_out->h_sync_width = mode_in->crtc_hsync_end - mode_in->crtc_hsync_start;
+		timing_out->h_front_porch = mode_in->crtc_hsync_start - mode_in->crtc_hdisplay;
+		timing_out->v_total = mode_in->crtc_vtotal;
+		timing_out->v_addressable = mode_in->crtc_vdisplay;
+		timing_out->v_front_porch = mode_in->crtc_vsync_start - mode_in->crtc_vdisplay;
+		timing_out->v_sync_width = mode_in->crtc_vsync_end - mode_in->crtc_vsync_start;
+		timing_out->pix_clk_100hz = mode_in->crtc_clock * 10;
+	}
+
+	timing_out->aspect_ratio = get_aspect_ratio(mode_in);
+
+	stream->out_transfer_func.type = TF_TYPE_PREDEFINED;
+	stream->out_transfer_func.tf = TRANSFER_FUNCTION_SRGB;
+	if (stream->signal == SIGNAL_TYPE_HDMI_TYPE_A) {
+		if (!adjust_colour_depth_from_display_info(timing_out, info) &&
+		    drm_mode_is_420_also(info, mode_in) &&
+		    timing_out->pixel_encoding != PIXEL_ENCODING_YCBCR420) {
+			timing_out->pixel_encoding = PIXEL_ENCODING_YCBCR420;
+			adjust_colour_depth_from_display_info(timing_out, info);
+		}
+	}
+
+	stream->output_color_space = get_output_color_space(timing_out, connector_state);
+	stream->content_type = get_output_content_type(connector_state);
+}
+
+static void fill_audio_info(struct audio_info *audio_info,
+			    const struct drm_connector *drm_connector,
+			    const struct dc_sink *dc_sink)
+{
+	int i = 0;
+	int cea_revision = 0;
+	const struct dc_edid_caps *edid_caps = &dc_sink->edid_caps;
+
+	audio_info->manufacture_id = edid_caps->manufacturer_id;
+	audio_info->product_id = edid_caps->product_id;
+
+	cea_revision = drm_connector->display_info.cea_rev;
+
+	strscpy(audio_info->display_name,
+		edid_caps->display_name,
+		AUDIO_INFO_DISPLAY_NAME_SIZE_IN_CHARS);
+
+	if (cea_revision >= 3) {
+		audio_info->mode_count = edid_caps->audio_mode_count;
+
+		for (i = 0; i < audio_info->mode_count; ++i) {
+			audio_info->modes[i].format_code =
+					(enum audio_format_code)
+					(edid_caps->audio_modes[i].format_code);
+			audio_info->modes[i].channel_count =
+					edid_caps->audio_modes[i].channel_count;
+			audio_info->modes[i].sample_rates.all =
+					edid_caps->audio_modes[i].sample_rate;
+			audio_info->modes[i].sample_size =
+					edid_caps->audio_modes[i].sample_size;
+		}
+	}
+
+	audio_info->flags.all = edid_caps->speaker_flags;
+
+	/* TODO: We only check for the progressive mode, check for interlace mode too */
+	if (drm_connector->latency_present[0]) {
+		audio_info->video_latency = drm_connector->video_latency[0];
+		audio_info->audio_latency = drm_connector->audio_latency[0];
+	}
+
+	/* TODO: For DP, video and audio latency should be calculated from DPCD caps */
+
+}
+
+static void
+copy_crtc_timing_for_drm_display_mode(const struct drm_display_mode *src_mode,
+				      struct drm_display_mode *dst_mode)
+{
+	dst_mode->crtc_hdisplay = src_mode->crtc_hdisplay;
+	dst_mode->crtc_vdisplay = src_mode->crtc_vdisplay;
+	dst_mode->crtc_clock = src_mode->crtc_clock;
+	dst_mode->crtc_hblank_start = src_mode->crtc_hblank_start;
+	dst_mode->crtc_hblank_end = src_mode->crtc_hblank_end;
+	dst_mode->crtc_hsync_start =  src_mode->crtc_hsync_start;
+	dst_mode->crtc_hsync_end = src_mode->crtc_hsync_end;
+	dst_mode->crtc_htotal = src_mode->crtc_htotal;
+	dst_mode->crtc_hskew = src_mode->crtc_hskew;
+	dst_mode->crtc_vblank_start = src_mode->crtc_vblank_start;
+	dst_mode->crtc_vblank_end = src_mode->crtc_vblank_end;
+	dst_mode->crtc_vsync_start = src_mode->crtc_vsync_start;
+	dst_mode->crtc_vsync_end = src_mode->crtc_vsync_end;
+	dst_mode->crtc_vtotal = src_mode->crtc_vtotal;
+}
+
+static void
+decide_crtc_timing_for_drm_display_mode(struct drm_display_mode *drm_mode,
+					const struct drm_display_mode *native_mode,
+					bool scale_enabled)
+{
+	if (scale_enabled || (
+	    native_mode->clock == drm_mode->clock &&
+	    native_mode->htotal == drm_mode->htotal &&
+	    native_mode->vtotal == drm_mode->vtotal)) {
+		if (native_mode->crtc_clock)
+			copy_crtc_timing_for_drm_display_mode(native_mode, drm_mode);
+	} else {
+		/* no scaling nor amdgpu inserted, no need to patch */
+	}
+}
+
+static struct dc_sink *
+create_fake_sink(struct drm_device *dev, struct dc_link *link)
+{
+	struct dc_sink_init_data sink_init_data = { 0 };
+	struct dc_sink *sink = NULL;
+
+	sink_init_data.link = link;
+	sink_init_data.sink_signal = link->connector_signal;
+
+	sink = dc_sink_create(&sink_init_data);
+	if (!sink) {
+		drm_err(dev, "Failed to create sink!\n");
+		return NULL;
+	}
+	sink->sink_signal = SIGNAL_TYPE_VIRTUAL;
+
+	return sink;
+}
+
+static void set_multisync_trigger_params(
+		struct dc_stream_state *stream)
+{
+	struct dc_stream_state *master = NULL;
+
+	if (stream->triggered_crtc_reset.enabled) {
+		master = stream->triggered_crtc_reset.event_source;
+		stream->triggered_crtc_reset.event =
+			master->timing.flags.VSYNC_POSITIVE_POLARITY ?
+			CRTC_EVENT_VSYNC_RISING : CRTC_EVENT_VSYNC_FALLING;
+		stream->triggered_crtc_reset.delay = TRIGGER_DELAY_NEXT_PIXEL;
+	}
+}
+
+static void set_master_stream(struct dc_stream_state *stream_set[],
+			      int stream_count)
+{
+	int j, highest_rfr = 0, master_stream = 0;
+
+	for (j = 0;  j < stream_count; j++) {
+		if (stream_set[j] && stream_set[j]->triggered_crtc_reset.enabled) {
+			int refresh_rate = 0;
+
+			refresh_rate = (stream_set[j]->timing.pix_clk_100hz*100)/
+				(stream_set[j]->timing.h_total*stream_set[j]->timing.v_total);
+			if (refresh_rate > highest_rfr) {
+				highest_rfr = refresh_rate;
+				master_stream = j;
+			}
+		}
+	}
+	for (j = 0;  j < stream_count; j++) {
+		if (stream_set[j])
+			stream_set[j]->triggered_crtc_reset.event_source = stream_set[master_stream];
+	}
+}
+
+static void dm_enable_per_frame_crtc_master_sync(struct dc_state *context)
+{
+	int i = 0;
+	struct dc_stream_state *stream;
+
+	if (context->stream_count < 2)
+		return;
+	for (i = 0; i < context->stream_count ; i++) {
+		if (!context->streams[i])
+			continue;
+		/*
+		 * TODO: add a function to read AMD VSDB bits and set
+		 * crtc_sync_master.multi_sync_enabled flag
+		 * For now it's set to false
+		 */
+	}
+
+	set_master_stream(context->streams, context->stream_count);
+
+	for (i = 0; i < context->stream_count ; i++) {
+		stream = context->streams[i];
+
+		if (!stream)
+			continue;
+
+		set_multisync_trigger_params(stream);
+	}
+}
+
+/**
+ * DOC: FreeSync Video
+ *
+ * When a userspace application wants to play a video, the content follows a
+ * standard format definition that usually specifies the FPS for that format.
+ * The below list illustrates some video format and the expected FPS,
+ * respectively:
+ *
+ * - TV/NTSC (23.976 FPS)
+ * - Cinema (24 FPS)
+ * - TV/PAL (25 FPS)
+ * - TV/NTSC (29.97 FPS)
+ * - TV/NTSC (30 FPS)
+ * - Cinema HFR (48 FPS)
+ * - TV/PAL (50 FPS)
+ * - Commonly used (60 FPS)
+ * - Multiples of 24 (48,72,96 FPS)
+ *
+ * The list of standards video format is not huge and can be added to the
+ * connector modeset list beforehand. With that, userspace can leverage
+ * FreeSync to extends the front porch in order to attain the target refresh
+ * rate. Such a switch will happen seamlessly, without screen blanking or
+ * reprogramming of the output in any other way. If the userspace requests a
+ * modesetting change compatible with FreeSync modes that only differ in the
+ * refresh rate, DC will skip the full update and avoid blink during the
+ * transition. For example, the video player can change the modesetting from
+ * 60Hz to 30Hz for playing TV/NTSC content when it goes full screen without
+ * causing any display blink. This same concept can be applied to a mode
+ * setting change.
+ */
+static struct drm_display_mode *
+get_highest_refresh_rate_mode(struct amdgpu_dm_connector *aconnector,
+		bool use_probed_modes)
+{
+	struct drm_display_mode *m, *m_pref = NULL;
+	u16 current_refresh, highest_refresh;
+	struct list_head *list_head = use_probed_modes ?
+		&aconnector->base.probed_modes :
+		&aconnector->base.modes;
+
+	if (aconnector->base.connector_type == DRM_MODE_CONNECTOR_WRITEBACK)
+		return NULL;
+
+	if (aconnector->freesync_vid_base.clock != 0)
+		return &aconnector->freesync_vid_base;
+
+	/* Find the preferred mode */
+	list_for_each_entry(m, list_head, head) {
+		if (m->type & DRM_MODE_TYPE_PREFERRED) {
+			m_pref = m;
+			break;
+		}
+	}
+
+	if (!m_pref) {
+		/* Probably an EDID with no preferred mode. Fallback to first entry */
+		m_pref = list_first_entry_or_null(
+				&aconnector->base.modes, struct drm_display_mode, head);
+		if (!m_pref) {
+			drm_dbg_driver(aconnector->base.dev, "No preferred mode found in EDID\n");
+			return NULL;
+		}
+	}
+
+	highest_refresh = drm_mode_vrefresh(m_pref);
+
+	/*
+	 * Find the mode with highest refresh rate with same resolution.
+	 * For some monitors, preferred mode is not the mode with highest
+	 * supported refresh rate.
+	 */
+	list_for_each_entry(m, list_head, head) {
+		current_refresh  = drm_mode_vrefresh(m);
+
+		if (m->hdisplay == m_pref->hdisplay &&
+		    m->vdisplay == m_pref->vdisplay &&
+		    highest_refresh < current_refresh) {
+			highest_refresh = current_refresh;
+			m_pref = m;
+		}
+	}
+
+	drm_mode_copy(&aconnector->freesync_vid_base, m_pref);
+	return m_pref;
+}
+
+static bool is_freesync_video_mode(const struct drm_display_mode *mode,
+		struct amdgpu_dm_connector *aconnector)
+{
+	struct drm_display_mode *high_mode;
+	int timing_diff;
+
+	high_mode = get_highest_refresh_rate_mode(aconnector, false);
+	if (!high_mode || !mode)
+		return false;
+
+	timing_diff = high_mode->vtotal - mode->vtotal;
+
+	if (high_mode->clock == 0 || high_mode->clock != mode->clock ||
+	    high_mode->hdisplay != mode->hdisplay ||
+	    high_mode->vdisplay != mode->vdisplay ||
+	    high_mode->hsync_start != mode->hsync_start ||
+	    high_mode->hsync_end != mode->hsync_end ||
+	    high_mode->htotal != mode->htotal ||
+	    high_mode->hskew != mode->hskew ||
+	    high_mode->vscan != mode->vscan ||
+	    high_mode->vsync_start - mode->vsync_start != timing_diff ||
+	    high_mode->vsync_end - mode->vsync_end != timing_diff)
+		return false;
+	else
+		return true;
+}
+
+#if defined(CONFIG_DRM_AMD_DC_FP)
+static void update_dsc_caps(struct amdgpu_dm_connector *aconnector,
+			    struct dc_sink *sink, struct dc_stream_state *stream,
+			    struct dsc_dec_dpcd_caps *dsc_caps)
+{
+	stream->timing.flags.DSC = 0;
+	dsc_caps->is_dsc_supported = false;
+
+	if (aconnector->dc_link && (sink->sink_signal == SIGNAL_TYPE_DISPLAY_PORT ||
+	    sink->sink_signal == SIGNAL_TYPE_EDP)) {
+		if (sink->link->dpcd_caps.dongle_type == DISPLAY_DONGLE_NONE ||
+			sink->link->dpcd_caps.dongle_type == DISPLAY_DONGLE_DP_HDMI_CONVERTER)
+			dc_dsc_parse_dsc_dpcd(aconnector->dc_link->ctx->dc,
+				aconnector->dc_link->dpcd_caps.dsc_caps.dsc_basic_caps.raw,
+				aconnector->dc_link->dpcd_caps.dsc_caps.dsc_branch_decoder_caps.raw,
+				dsc_caps);
+	}
+}
+
+static void apply_dsc_policy_for_edp(struct amdgpu_dm_connector *aconnector,
+				    struct dc_sink *sink, struct dc_stream_state *stream,
+				    struct dsc_dec_dpcd_caps *dsc_caps,
+				    uint32_t max_dsc_target_bpp_limit_override)
+{
+	const struct dc_link_settings *verified_link_cap = NULL;
+	u32 link_bw_in_kbps;
+	u32 edp_min_bpp_x16, edp_max_bpp_x16;
+	struct dc *dc = sink->ctx->dc;
+	struct dc_dsc_bw_range bw_range = {0};
+	struct dc_dsc_config dsc_cfg = {0};
+	struct dc_dsc_config_options dsc_options = {0};
+
+	dc_dsc_get_default_config_option(dc, &dsc_options);
+	dsc_options.max_target_bpp_limit_override_x16 = max_dsc_target_bpp_limit_override * 16;
+
+	verified_link_cap = dc_link_get_link_cap(stream->link);
+	link_bw_in_kbps = dc_link_bandwidth_kbps(stream->link, verified_link_cap);
+	edp_min_bpp_x16 = 8 * 16;
+	edp_max_bpp_x16 = 8 * 16;
+
+	if (edp_max_bpp_x16 > dsc_caps->edp_max_bits_per_pixel)
+		edp_max_bpp_x16 = dsc_caps->edp_max_bits_per_pixel;
+
+	if (edp_max_bpp_x16 < edp_min_bpp_x16)
+		edp_min_bpp_x16 = edp_max_bpp_x16;
+
+	if (dc_dsc_compute_bandwidth_range(dc->res_pool->dscs[0],
+				dc->debug.dsc_min_slice_height_override,
+				edp_min_bpp_x16, edp_max_bpp_x16,
+				dsc_caps,
+				&stream->timing,
+				dc_link_get_highest_encoding_format(aconnector->dc_link),
+				&bw_range)) {
+
+		if (bw_range.max_kbps < link_bw_in_kbps) {
+			if (dc_dsc_compute_config(dc->res_pool->dscs[0],
+					dsc_caps,
+					&dsc_options,
+					0,
+					&stream->timing,
+					dc_link_get_highest_encoding_format(aconnector->dc_link),
+					&dsc_cfg)) {
+				stream->timing.dsc_cfg = dsc_cfg;
+				stream->timing.flags.DSC = 1;
+				stream->timing.dsc_cfg.bits_per_pixel = edp_max_bpp_x16;
+			}
+			return;
+		}
+	}
+
+	if (dc_dsc_compute_config(dc->res_pool->dscs[0],
+				dsc_caps,
+				&dsc_options,
+				link_bw_in_kbps,
+				&stream->timing,
+				dc_link_get_highest_encoding_format(aconnector->dc_link),
+				&dsc_cfg)) {
+		stream->timing.dsc_cfg = dsc_cfg;
+		stream->timing.flags.DSC = 1;
+	}
+}
+
+static void apply_dsc_policy_for_stream(struct amdgpu_dm_connector *aconnector,
+					struct dc_sink *sink, struct dc_stream_state *stream,
+					struct dsc_dec_dpcd_caps *dsc_caps)
+{
+	struct drm_connector *drm_connector = &aconnector->base;
+	u32 link_bandwidth_kbps;
+	struct dc *dc = sink->ctx->dc;
+	u32 max_supported_bw_in_kbps, timing_bw_in_kbps;
+	u32 dsc_max_supported_bw_in_kbps;
+	u32 max_dsc_target_bpp_limit_override =
+		drm_connector->display_info.max_dsc_bpp;
+	struct dc_dsc_config_options dsc_options = {0};
+
+	dc_dsc_get_default_config_option(dc, &dsc_options);
+	dsc_options.max_target_bpp_limit_override_x16 = max_dsc_target_bpp_limit_override * 16;
+
+	link_bandwidth_kbps = dc_link_bandwidth_kbps(aconnector->dc_link,
+							dc_link_get_link_cap(aconnector->dc_link));
+
+	/* Set DSC policy according to dsc_clock_en */
+	dc_dsc_policy_set_enable_dsc_when_not_needed(
+		aconnector->dsc_settings.dsc_force_enable == DSC_CLK_FORCE_ENABLE);
+
+	if (sink->sink_signal == SIGNAL_TYPE_EDP &&
+	    !aconnector->dc_link->panel_config.dsc.disable_dsc_edp &&
+	    dc->caps.edp_dsc_support && aconnector->dsc_settings.dsc_force_enable != DSC_CLK_FORCE_DISABLE) {
+
+		apply_dsc_policy_for_edp(aconnector, sink, stream, dsc_caps, max_dsc_target_bpp_limit_override);
+
+	} else if (sink->sink_signal == SIGNAL_TYPE_DISPLAY_PORT) {
+		if (sink->link->dpcd_caps.dongle_type == DISPLAY_DONGLE_NONE) {
+			if (dc_dsc_compute_config(aconnector->dc_link->ctx->dc->res_pool->dscs[0],
+						dsc_caps,
+						&dsc_options,
+						link_bandwidth_kbps,
+						&stream->timing,
+						dc_link_get_highest_encoding_format(aconnector->dc_link),
+						&stream->timing.dsc_cfg)) {
+				stream->timing.flags.DSC = 1;
+				drm_dbg_driver(drm_connector->dev, "%s: SST_DSC [%s] DSC is selected from SST RX\n",
+							__func__, drm_connector->name);
+			}
+		} else if (sink->link->dpcd_caps.dongle_type == DISPLAY_DONGLE_DP_HDMI_CONVERTER) {
+			timing_bw_in_kbps = dc_bandwidth_in_kbps_from_timing(&stream->timing,
+					dc_link_get_highest_encoding_format(aconnector->dc_link));
+			max_supported_bw_in_kbps = link_bandwidth_kbps;
+			dsc_max_supported_bw_in_kbps = link_bandwidth_kbps;
+
+			if (timing_bw_in_kbps > max_supported_bw_in_kbps &&
+					max_supported_bw_in_kbps > 0 &&
+					dsc_max_supported_bw_in_kbps > 0)
+				if (dc_dsc_compute_config(aconnector->dc_link->ctx->dc->res_pool->dscs[0],
+						dsc_caps,
+						&dsc_options,
+						dsc_max_supported_bw_in_kbps,
+						&stream->timing,
+						dc_link_get_highest_encoding_format(aconnector->dc_link),
+						&stream->timing.dsc_cfg)) {
+					stream->timing.flags.DSC = 1;
+					drm_dbg_driver(drm_connector->dev, "%s: SST_DSC [%s] DSC is selected from DP-HDMI PCON\n",
+									 __func__, drm_connector->name);
+				}
+		}
+	}
+
+	/* Overwrite the stream flag if DSC is enabled through debugfs */
+	if (aconnector->dsc_settings.dsc_force_enable == DSC_CLK_FORCE_ENABLE)
+		stream->timing.flags.DSC = 1;
+
+	if (stream->timing.flags.DSC && aconnector->dsc_settings.dsc_num_slices_h)
+		stream->timing.dsc_cfg.num_slices_h = aconnector->dsc_settings.dsc_num_slices_h;
+
+	if (stream->timing.flags.DSC && aconnector->dsc_settings.dsc_num_slices_v)
+		stream->timing.dsc_cfg.num_slices_v = aconnector->dsc_settings.dsc_num_slices_v;
+
+	if (stream->timing.flags.DSC && aconnector->dsc_settings.dsc_bits_per_pixel)
+		stream->timing.dsc_cfg.bits_per_pixel = aconnector->dsc_settings.dsc_bits_per_pixel;
+}
+#endif
+
+static struct dc_stream_state *
+create_stream_for_sink(struct drm_connector *connector,
+		       const struct drm_display_mode *drm_mode,
+		       const struct dm_connector_state *dm_state,
+		       const struct dc_stream_state *old_stream,
+		       int requested_bpc)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_dm_connector *aconnector = NULL;
+	struct drm_display_mode *preferred_mode = NULL;
+	const struct drm_connector_state *con_state = &dm_state->base;
+	struct dc_stream_state *stream = NULL;
+	struct drm_display_mode mode;
+	struct drm_display_mode saved_mode;
+	struct drm_display_mode *freesync_mode = NULL;
+	bool native_mode_found = false;
+	bool recalculate_timing = false;
+	bool scale = dm_state->scaling != RMX_OFF;
+	int mode_refresh;
+	int preferred_refresh = 0;
+	enum color_transfer_func tf = TRANSFER_FUNC_UNKNOWN;
+#if defined(CONFIG_DRM_AMD_DC_FP)
+	struct dsc_dec_dpcd_caps dsc_caps;
+#endif
+	struct dc_link *link = NULL;
+	struct dc_sink *sink = NULL;
+
+	drm_mode_init(&mode, drm_mode);
+	memset(&saved_mode, 0, sizeof(saved_mode));
+
+	if (connector->connector_type != DRM_MODE_CONNECTOR_WRITEBACK) {
+		aconnector = NULL;
+		aconnector = to_amdgpu_dm_connector(connector);
+		link = aconnector->dc_link;
+	} else {
+		struct drm_writeback_connector *wbcon = NULL;
+		struct amdgpu_dm_wb_connector *dm_wbcon = NULL;
+
+		wbcon = drm_connector_to_writeback(connector);
+		dm_wbcon = to_amdgpu_dm_wb_connector(wbcon);
+		link = dm_wbcon->link;
+	}
+
+	if (!aconnector || !aconnector->dc_sink) {
+		sink = create_fake_sink(dev, link);
+		if (!sink)
+			return stream;
+
+	} else {
+		sink = aconnector->dc_sink;
+		dc_sink_retain(sink);
+	}
+
+	stream = dc_create_stream_for_sink(sink);
+
+	if (stream == NULL) {
+		drm_err(dev, "Failed to create stream for sink!\n");
+		goto finish;
+	}
+
+	/* We leave this NULL for writeback connectors */
+	stream->dm_stream_context = aconnector;
+
+	stream->timing.flags.LTE_340MCSC_SCRAMBLE =
+		connector->display_info.hdmi.scdc.scrambling.low_rates;
+
+	list_for_each_entry(preferred_mode, &connector->modes, head) {
+		/* Search for preferred mode */
+		if (preferred_mode->type & DRM_MODE_TYPE_PREFERRED) {
+			native_mode_found = true;
+			break;
+		}
+	}
+	if (!native_mode_found)
+		preferred_mode = list_first_entry_or_null(
+				&connector->modes,
+				struct drm_display_mode,
+				head);
+
+	mode_refresh = drm_mode_vrefresh(&mode);
+
+	if (preferred_mode == NULL) {
+		/*
+		 * This may not be an error, the use case is when we have no
+		 * usermode calls to reset and set mode upon hotplug. In this
+		 * case, we call set mode ourselves to restore the previous mode
+		 * and the modelist may not be filled in time.
+		 */
+		drm_dbg_driver(dev, "No preferred mode found\n");
+	} else if (aconnector) {
+		recalculate_timing = amdgpu_freesync_vid_mode &&
+				 is_freesync_video_mode(&mode, aconnector);
+		if (recalculate_timing) {
+			freesync_mode = get_highest_refresh_rate_mode(aconnector, false);
+			drm_mode_copy(&saved_mode, &mode);
+			saved_mode.picture_aspect_ratio = mode.picture_aspect_ratio;
+			drm_mode_copy(&mode, freesync_mode);
+			mode.picture_aspect_ratio = saved_mode.picture_aspect_ratio;
+		} else {
+			decide_crtc_timing_for_drm_display_mode(
+					&mode, preferred_mode, scale);
+
+			preferred_refresh = drm_mode_vrefresh(preferred_mode);
+		}
+	}
+
+	if (recalculate_timing)
+		drm_mode_set_crtcinfo(&saved_mode, 0);
+
+	/*
+	 * If scaling is enabled and refresh rate didn't change
+	 * we copy the vic and polarities of the old timings
+	 */
+	if (!scale || mode_refresh != preferred_refresh)
+		fill_stream_properties_from_drm_display_mode(
+			stream, &mode, connector, con_state, NULL,
+			requested_bpc);
+	else
+		fill_stream_properties_from_drm_display_mode(
+			stream, &mode, connector, con_state, old_stream,
+			requested_bpc);
+
+	/* The rest isn't needed for writeback connectors */
+	if (!aconnector)
+		goto finish;
+
+	if (aconnector->timing_changed) {
+		drm_dbg(aconnector->base.dev,
+			"overriding timing for automated test, bpc %d, changing to %d\n",
+			stream->timing.display_color_depth,
+			aconnector->timing_requested->display_color_depth);
+		stream->timing = *aconnector->timing_requested;
+	}
+
+#if defined(CONFIG_DRM_AMD_DC_FP)
+	/* SST DSC determination policy */
+	update_dsc_caps(aconnector, sink, stream, &dsc_caps);
+	if (aconnector->dsc_settings.dsc_force_enable != DSC_CLK_FORCE_DISABLE && dsc_caps.is_dsc_supported)
+		apply_dsc_policy_for_stream(aconnector, sink, stream, &dsc_caps);
+#endif
+
+	update_stream_scaling_settings(&mode, dm_state, stream);
+
+	fill_audio_info(
+		&stream->audio_info,
+		connector,
+		sink);
+
+	update_stream_signal(stream, sink);
+
+	if (stream->signal == SIGNAL_TYPE_HDMI_TYPE_A)
+		mod_build_hf_vsif_infopacket(stream, &stream->vsp_infopacket);
+
+	if (stream->signal == SIGNAL_TYPE_DISPLAY_PORT ||
+	    stream->signal == SIGNAL_TYPE_DISPLAY_PORT_MST ||
+	    stream->signal == SIGNAL_TYPE_EDP) {
+		const struct dc_edid_caps *edid_caps;
+		unsigned int disable_colorimetry = 0;
+
+		if (aconnector->dc_sink) {
+			edid_caps = &aconnector->dc_sink->edid_caps;
+			disable_colorimetry = edid_caps->panel_patch.disable_colorimetry;
+		}
+
+		//
+		// should decide stream support vsc sdp colorimetry capability
+		// before building vsc info packet
+		//
+		stream->use_vsc_sdp_for_colorimetry = stream->link->dpcd_caps.dpcd_rev.raw >= 0x14 &&
+						      stream->link->dpcd_caps.dprx_feature.bits.VSC_SDP_COLORIMETRY_SUPPORTED &&
+						      !disable_colorimetry;
+
+		if (stream->out_transfer_func.tf == TRANSFER_FUNCTION_GAMMA22)
+			tf = TRANSFER_FUNC_GAMMA_22;
+		mod_build_vsc_infopacket(stream, &stream->vsc_infopacket, stream->output_color_space, tf);
+		aconnector->sr_skip_count = AMDGPU_DM_PSR_ENTRY_DELAY;
+
+	}
+finish:
+	dc_sink_release(sink);
+
+	return stream;
+}
+
+static enum drm_connector_status
+amdgpu_dm_connector_detect(struct drm_connector *connector, bool force)
+{
+	bool connected;
+	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector);
+
+	/*
+	 * Notes:
+	 * 1. This interface is NOT called in context of HPD irq.
+	 * 2. This interface *is called* in context of user-mode ioctl. Which
+	 * makes it a bad place for *any* MST-related activity.
+	 */
+
+	if (aconnector->base.force == DRM_FORCE_UNSPECIFIED &&
+	    !aconnector->fake_enable)
+		connected = (aconnector->dc_sink != NULL);
+	else
+		connected = (aconnector->base.force == DRM_FORCE_ON ||
+				aconnector->base.force == DRM_FORCE_ON_DIGITAL);
+
+	update_subconnector_property(aconnector);
+
+	return (connected ? connector_status_connected :
+			connector_status_disconnected);
+}
+
+int amdgpu_dm_connector_atomic_set_property(struct drm_connector *connector,
+					    struct drm_connector_state *connector_state,
+					    struct drm_property *property,
+					    uint64_t val)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	struct dm_connector_state *dm_old_state =
+		to_dm_connector_state(connector->state);
+	struct dm_connector_state *dm_new_state =
+		to_dm_connector_state(connector_state);
+
+	int ret = -EINVAL;
+
+	if (property == dev->mode_config.scaling_mode_property) {
+		enum amdgpu_rmx_type rmx_type;
+
+		switch (val) {
+		case DRM_MODE_SCALE_CENTER:
+			rmx_type = RMX_CENTER;
+			break;
+		case DRM_MODE_SCALE_ASPECT:
+			rmx_type = RMX_ASPECT;
+			break;
+		case DRM_MODE_SCALE_FULLSCREEN:
+			rmx_type = RMX_FULL;
+			break;
+		case DRM_MODE_SCALE_NONE:
+		default:
+			rmx_type = RMX_OFF;
+			break;
+		}
+
+		if (dm_old_state->scaling == rmx_type)
+			return 0;
+
+		dm_new_state->scaling = rmx_type;
+		ret = 0;
+	} else if (property == adev->mode_info.underscan_hborder_property) {
+		dm_new_state->underscan_hborder = val;
+		ret = 0;
+	} else if (property == adev->mode_info.underscan_vborder_property) {
+		dm_new_state->underscan_vborder = val;
+		ret = 0;
+	} else if (property == adev->mode_info.underscan_property) {
+		dm_new_state->underscan_enable = val;
+		ret = 0;
+	}
+
+	return ret;
+}
+
+int amdgpu_dm_connector_atomic_get_property(struct drm_connector *connector,
+					    const struct drm_connector_state *state,
+					    struct drm_property *property,
+					    uint64_t *val)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	struct dm_connector_state *dm_state =
+		to_dm_connector_state(state);
+	int ret = -EINVAL;
+
+	if (property == dev->mode_config.scaling_mode_property) {
+		switch (dm_state->scaling) {
+		case RMX_CENTER:
+			*val = DRM_MODE_SCALE_CENTER;
+			break;
+		case RMX_ASPECT:
+			*val = DRM_MODE_SCALE_ASPECT;
+			break;
+		case RMX_FULL:
+			*val = DRM_MODE_SCALE_FULLSCREEN;
+			break;
+		case RMX_OFF:
+		default:
+			*val = DRM_MODE_SCALE_NONE;
+			break;
+		}
+		ret = 0;
+	} else if (property == adev->mode_info.underscan_hborder_property) {
+		*val = dm_state->underscan_hborder;
+		ret = 0;
+	} else if (property == adev->mode_info.underscan_vborder_property) {
+		*val = dm_state->underscan_vborder;
+		ret = 0;
+	} else if (property == adev->mode_info.underscan_property) {
+		*val = dm_state->underscan_enable;
+		ret = 0;
+	}
+
+	return ret;
+}
+
+/**
+ * DOC: panel power savings
+ *
+ * The display manager allows you to set your desired **panel power savings**
+ * level (between 0-4, with 0 representing off), e.g. using the following::
+ *
+ *   # echo 3 > /sys/class/drm/card0-eDP-1/amdgpu/panel_power_savings
+ *
+ * Modifying this value can have implications on color accuracy, so tread
+ * carefully.
+ */
+
+static ssize_t panel_power_savings_show(struct device *device,
+					struct device_attribute *attr,
+					char *buf)
+{
+	struct drm_connector *connector = dev_get_drvdata(device);
+	struct drm_device *dev = connector->dev;
+	u8 val;
+
+	drm_modeset_lock(&dev->mode_config.connection_mutex, NULL);
+	val = to_dm_connector_state(connector->state)->abm_level ==
+		ABM_LEVEL_IMMEDIATE_DISABLE ? 0 :
+		to_dm_connector_state(connector->state)->abm_level;
+	drm_modeset_unlock(&dev->mode_config.connection_mutex);
+
+	return sysfs_emit(buf, "%u\n", val);
+}
+
+static ssize_t panel_power_savings_store(struct device *device,
+					 struct device_attribute *attr,
+					 const char *buf, size_t count)
+{
+	struct drm_connector *connector = dev_get_drvdata(device);
+	struct drm_device *dev = connector->dev;
+	long val;
+	int ret;
+
+	ret = kstrtol(buf, 0, &val);
+
+	if (ret)
+		return ret;
+
+	if (val < 0 || val > 4)
+		return -EINVAL;
+
+	drm_modeset_lock(&dev->mode_config.connection_mutex, NULL);
+	to_dm_connector_state(connector->state)->abm_level = val ?:
+		ABM_LEVEL_IMMEDIATE_DISABLE;
+	drm_modeset_unlock(&dev->mode_config.connection_mutex);
+
+	drm_kms_helper_hotplug_event(dev);
+
+	return count;
+}
+
+static DEVICE_ATTR_RW(panel_power_savings);
+
+static struct attribute *amdgpu_attrs[] = {
+	&dev_attr_panel_power_savings.attr,
+	NULL
+};
+
+static const struct attribute_group amdgpu_group = {
+	.name = "amdgpu",
+	.attrs = amdgpu_attrs
+};
+
+static bool
+amdgpu_dm_should_create_sysfs(struct amdgpu_dm_connector *amdgpu_dm_connector)
+{
+	if (amdgpu_dm_abm_level >= 0)
+		return false;
+
+	if (amdgpu_dm_connector->base.connector_type != DRM_MODE_CONNECTOR_eDP)
+		return false;
+
+	/* check for OLED panels */
+	if (amdgpu_dm_connector->bl_idx >= 0) {
+		struct drm_device *drm = amdgpu_dm_connector->base.dev;
+		struct amdgpu_display_manager *dm = &drm_to_adev(drm)->dm;
+		struct amdgpu_dm_backlight_caps *caps;
+
+		caps = &dm->backlight_caps[amdgpu_dm_connector->bl_idx];
+		if (caps->aux_support)
+			return false;
+	}
+
+	return true;
+}
+
+static void amdgpu_dm_connector_unregister(struct drm_connector *connector)
+{
+	struct amdgpu_dm_connector *amdgpu_dm_connector = to_amdgpu_dm_connector(connector);
+
+	if (amdgpu_dm_should_create_sysfs(amdgpu_dm_connector))
+		sysfs_remove_group(&connector->kdev->kobj, &amdgpu_group);
+
+	cec_notifier_conn_unregister(amdgpu_dm_connector->notifier);
+	drm_dp_aux_unregister(&amdgpu_dm_connector->dm_dp_aux.aux);
+}
+
+static void amdgpu_dm_connector_destroy(struct drm_connector *connector)
+{
+	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector);
+	struct amdgpu_device *adev = drm_to_adev(connector->dev);
+	struct amdgpu_display_manager *dm = &adev->dm;
+
+	/*
+	 * Call only if mst_mgr was initialized before since it's not done
+	 * for all connector types.
+	 */
+	if (aconnector->mst_mgr.dev)
+		drm_dp_mst_topology_mgr_destroy(&aconnector->mst_mgr);
+
+	if (aconnector->bl_idx != -1) {
+		backlight_device_unregister(dm->backlight_dev[aconnector->bl_idx]);
+		dm->backlight_dev[aconnector->bl_idx] = NULL;
+	}
+
+	if (aconnector->dc_em_sink)
+		dc_sink_release(aconnector->dc_em_sink);
+	aconnector->dc_em_sink = NULL;
+	if (aconnector->dc_sink)
+		dc_sink_release(aconnector->dc_sink);
+	aconnector->dc_sink = NULL;
+
+	drm_dp_cec_unregister_connector(&aconnector->dm_dp_aux.aux);
+	drm_connector_unregister(connector);
+	drm_connector_cleanup(connector);
+	kfree(aconnector->dm_dp_aux.aux.name);
+
+	kfree(connector);
+}
+
+void amdgpu_dm_connector_funcs_reset(struct drm_connector *connector)
+{
+	struct dm_connector_state *state =
+		to_dm_connector_state(connector->state);
+
+	if (connector->state)
+		__drm_atomic_helper_connector_destroy_state(connector->state);
+
+	kfree(state);
+
+	state = kzalloc(sizeof(*state), GFP_KERNEL);
+
+	if (state) {
+		state->scaling = RMX_OFF;
+		state->underscan_enable = false;
+		state->underscan_hborder = 0;
+		state->underscan_vborder = 0;
+		state->base.max_requested_bpc = 8;
+		state->vcpi_slots = 0;
+		state->pbn = 0;
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_eDP) {
+			if (amdgpu_dm_abm_level <= 0)
+				state->abm_level = ABM_LEVEL_IMMEDIATE_DISABLE;
+			else
+				state->abm_level = amdgpu_dm_abm_level;
+		}
+
+		__drm_atomic_helper_connector_reset(connector, &state->base);
+	}
+}
+
+struct drm_connector_state *
+amdgpu_dm_connector_atomic_duplicate_state(struct drm_connector *connector)
+{
+	struct dm_connector_state *state =
+		to_dm_connector_state(connector->state);
+
+	struct dm_connector_state *new_state =
+			kmemdup(state, sizeof(*state), GFP_KERNEL);
+
+	if (!new_state)
+		return NULL;
+
+	__drm_atomic_helper_connector_duplicate_state(connector, &new_state->base);
+
+	new_state->freesync_capable = state->freesync_capable;
+	new_state->abm_level = state->abm_level;
+	new_state->scaling = state->scaling;
+	new_state->underscan_enable = state->underscan_enable;
+	new_state->underscan_hborder = state->underscan_hborder;
+	new_state->underscan_vborder = state->underscan_vborder;
+	new_state->vcpi_slots = state->vcpi_slots;
+	new_state->pbn = state->pbn;
+	return &new_state->base;
+}
+
+static int
+amdgpu_dm_connector_late_register(struct drm_connector *connector)
+{
+	struct amdgpu_dm_connector *amdgpu_dm_connector =
+		to_amdgpu_dm_connector(connector);
+	int r;
+
+	if (amdgpu_dm_should_create_sysfs(amdgpu_dm_connector)) {
+		r = sysfs_create_group(&connector->kdev->kobj,
+				       &amdgpu_group);
+		if (r)
+			return r;
+	}
+
+	amdgpu_dm_register_backlight_device(amdgpu_dm_connector);
+
+	if ((connector->connector_type == DRM_MODE_CONNECTOR_DisplayPort) ||
+	    (connector->connector_type == DRM_MODE_CONNECTOR_eDP)) {
+		amdgpu_dm_connector->dm_dp_aux.aux.dev = connector->kdev;
+		r = drm_dp_aux_register(&amdgpu_dm_connector->dm_dp_aux.aux);
+		if (r)
+			return r;
+	}
+
+#if defined(CONFIG_DEBUG_FS)
+	connector_debugfs_init(amdgpu_dm_connector);
+#endif
+
+	return 0;
+}
+
+static void amdgpu_dm_connector_funcs_force(struct drm_connector *connector)
+{
+	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector);
+	struct dc_link *dc_link = aconnector->dc_link;
+	struct dc_sink *dc_em_sink = aconnector->dc_em_sink;
+	const struct drm_edid *drm_edid;
+	struct i2c_adapter *ddc;
+	struct drm_device *dev = connector->dev;
+
+	if (dc_link && dc_link->aux_mode)
+		ddc = &aconnector->dm_dp_aux.aux.ddc;
+	else
+		ddc = &aconnector->i2c->base;
+
+	drm_edid = drm_edid_read_ddc(connector, ddc);
+	drm_edid_connector_update(connector, drm_edid);
+	if (!drm_edid) {
+		drm_err(dev, "No EDID found on connector: %s.\n", connector->name);
+		return;
+	}
+
+	aconnector->drm_edid = drm_edid;
+	/* Update emulated (virtual) sink's EDID */
+	if (dc_em_sink && dc_link) {
+		// FIXME: Get rid of drm_edid_raw()
+		const struct edid *edid = drm_edid_raw(drm_edid);
+
+		memset(&dc_em_sink->edid_caps, 0, sizeof(struct dc_edid_caps));
+		memmove(dc_em_sink->dc_edid.raw_edid, edid,
+			(edid->extensions + 1) * EDID_LENGTH);
+		dm_helpers_parse_edid_caps(
+			dc_link,
+			&dc_em_sink->dc_edid,
+			&dc_em_sink->edid_caps);
+	}
+}
+
+static const struct drm_connector_funcs amdgpu_dm_connector_funcs = {
+	.reset = amdgpu_dm_connector_funcs_reset,
+	.detect = amdgpu_dm_connector_detect,
+	.fill_modes = drm_helper_probe_single_connector_modes,
+	.destroy = amdgpu_dm_connector_destroy,
+	.atomic_duplicate_state = amdgpu_dm_connector_atomic_duplicate_state,
+	.atomic_destroy_state = drm_atomic_helper_connector_destroy_state,
+	.atomic_set_property = amdgpu_dm_connector_atomic_set_property,
+	.atomic_get_property = amdgpu_dm_connector_atomic_get_property,
+	.late_register = amdgpu_dm_connector_late_register,
+	.early_unregister = amdgpu_dm_connector_unregister,
+	.force = amdgpu_dm_connector_funcs_force
+};
+
+static int get_modes(struct drm_connector *connector)
+{
+	return amdgpu_dm_connector_get_modes(connector);
+}
+
+static void create_eml_sink(struct amdgpu_dm_connector *aconnector)
+{
+	struct drm_connector *connector = &aconnector->base;
+	struct dc_link *dc_link = aconnector->dc_link;
+	struct dc_sink_init_data init_params = {
+			.link = aconnector->dc_link,
+			.sink_signal = SIGNAL_TYPE_VIRTUAL
+	};
+	const struct drm_edid *drm_edid;
+	const struct edid *edid;
+	struct i2c_adapter *ddc;
+
+	if (dc_link && dc_link->aux_mode)
+		ddc = &aconnector->dm_dp_aux.aux.ddc;
+	else
+		ddc = &aconnector->i2c->base;
+
+	drm_edid = drm_edid_read_ddc(connector, ddc);
+	drm_edid_connector_update(connector, drm_edid);
+	if (!drm_edid) {
+		drm_err(connector->dev, "No EDID found on connector: %s.\n", connector->name);
+		return;
+	}
+
+	if (connector->display_info.is_hdmi)
+		init_params.sink_signal = SIGNAL_TYPE_HDMI_TYPE_A;
+
+	aconnector->drm_edid = drm_edid;
+
+	edid = drm_edid_raw(drm_edid); // FIXME: Get rid of drm_edid_raw()
+	aconnector->dc_em_sink = dc_link_add_remote_sink(
+		aconnector->dc_link,
+		(uint8_t *)edid,
+		(edid->extensions + 1) * EDID_LENGTH,
+		&init_params);
+
+	if (aconnector->base.force == DRM_FORCE_ON) {
+		aconnector->dc_sink = aconnector->dc_link->local_sink ?
+		aconnector->dc_link->local_sink :
+		aconnector->dc_em_sink;
+		if (aconnector->dc_sink)
+			dc_sink_retain(aconnector->dc_sink);
+	}
+}
+
+static void handle_edid_mgmt(struct amdgpu_dm_connector *aconnector)
+{
+	struct dc_link *link = (struct dc_link *)aconnector->dc_link;
+
+	/*
+	 * In case of headless boot with force on for DP managed connector
+	 * Those settings have to be != 0 to get initial modeset
+	 */
+	if (link->connector_signal == SIGNAL_TYPE_DISPLAY_PORT) {
+		link->verified_link_cap.lane_count = LANE_COUNT_FOUR;
+		link->verified_link_cap.link_rate = LINK_RATE_HIGH2;
+	}
+
+	create_eml_sink(aconnector);
+}
+
+static enum dc_status dm_validate_stream_and_context(struct dc *dc,
+						struct dc_stream_state *stream)
+{
+	enum dc_status dc_result = DC_ERROR_UNEXPECTED;
+	struct dc_plane_state *dc_plane_state = NULL;
+	struct dc_state *dc_state = NULL;
+
+	if (!stream)
+		goto cleanup;
+
+	dc_plane_state = dc_create_plane_state(dc);
+	if (!dc_plane_state)
+		goto cleanup;
+
+	dc_state = dc_state_create(dc, NULL);
+	if (!dc_state)
+		goto cleanup;
+
+	/* populate stream to plane */
+	dc_plane_state->src_rect.height  = stream->src.height;
+	dc_plane_state->src_rect.width   = stream->src.width;
+	dc_plane_state->dst_rect.height  = stream->src.height;
+	dc_plane_state->dst_rect.width   = stream->src.width;
+	dc_plane_state->clip_rect.height = stream->src.height;
+	dc_plane_state->clip_rect.width  = stream->src.width;
+	dc_plane_state->plane_size.surface_pitch = ((stream->src.width + 255) / 256) * 256;
+	dc_plane_state->plane_size.surface_size.height = stream->src.height;
+	dc_plane_state->plane_size.surface_size.width  = stream->src.width;
+	dc_plane_state->plane_size.chroma_size.height  = stream->src.height;
+	dc_plane_state->plane_size.chroma_size.width   = stream->src.width;
+	dc_plane_state->format = SURFACE_PIXEL_FORMAT_GRPH_ARGB8888;
+	dc_plane_state->tiling_info.gfx9.swizzle = DC_SW_UNKNOWN;
+	dc_plane_state->rotation = ROTATION_ANGLE_0;
+	dc_plane_state->is_tiling_rotated = false;
+	dc_plane_state->tiling_info.gfx8.array_mode = DC_ARRAY_LINEAR_GENERAL;
+
+	dc_result = dc_validate_stream(dc, stream);
+	if (dc_result == DC_OK)
+		dc_result = dc_validate_plane(dc, dc_plane_state);
+
+	if (dc_result == DC_OK)
+		dc_result = dc_state_add_stream(dc, dc_state, stream);
+
+	if (dc_result == DC_OK && !dc_state_add_plane(
+						dc,
+						stream,
+						dc_plane_state,
+						dc_state))
+		dc_result = DC_FAIL_ATTACH_SURFACES;
+
+	if (dc_result == DC_OK)
+		dc_result = dc_validate_global_state(dc, dc_state, DC_VALIDATE_MODE_ONLY);
+
+cleanup:
+	if (dc_state)
+		dc_state_release(dc_state);
+
+	if (dc_plane_state)
+		dc_plane_state_release(dc_plane_state);
+
+	return dc_result;
+}
+
+struct dc_stream_state *
+create_validate_stream_for_sink(struct drm_connector *connector,
+				const struct drm_display_mode *drm_mode,
+				const struct dm_connector_state *dm_state,
+				const struct dc_stream_state *old_stream)
+{
+	struct amdgpu_dm_connector *aconnector = NULL;
+	struct amdgpu_device *adev = drm_to_adev(connector->dev);
+	struct dc_stream_state *stream;
+	const struct drm_connector_state *drm_state = dm_state ? &dm_state->base : NULL;
+	int requested_bpc = drm_state ? drm_state->max_requested_bpc : 8;
+	enum dc_status dc_result = DC_OK;
+	uint8_t bpc_limit = 6;
+
+	if (!dm_state)
+		return NULL;
+
+	if (connector->connector_type != DRM_MODE_CONNECTOR_WRITEBACK)
+		aconnector = to_amdgpu_dm_connector(connector);
+
+	if (aconnector &&
+	    (aconnector->dc_link->connector_signal == SIGNAL_TYPE_HDMI_TYPE_A ||
+	     aconnector->dc_link->dpcd_caps.dongle_type == DISPLAY_DONGLE_DP_HDMI_CONVERTER))
+		bpc_limit = 8;
+
+	do {
+		drm_dbg_kms(connector->dev, "Trying with %d bpc\n", requested_bpc);
+		stream = create_stream_for_sink(connector, drm_mode,
+						dm_state, old_stream,
+						requested_bpc);
+		if (stream == NULL) {
+			drm_err(adev_to_drm(adev), "Failed to create stream for sink!\n");
+			break;
+		}
+
+		dc_result = dc_validate_stream(adev->dm.dc, stream);
+
+		if (!aconnector) /* writeback connector */
+			return stream;
+
+		if (dc_result == DC_OK && stream->signal == SIGNAL_TYPE_DISPLAY_PORT_MST)
+			dc_result = dm_dp_mst_is_port_support_mode(aconnector, stream);
+
+		if (dc_result == DC_OK)
+			dc_result = dm_validate_stream_and_context(adev->dm.dc, stream);
+
+		if (dc_result != DC_OK) {
+			DRM_DEBUG_KMS("Pruned mode %d x %d (clk %d) %s %s -- %s\n",
+				      drm_mode->hdisplay,
+				      drm_mode->vdisplay,
+				      drm_mode->clock,
+				      dc_pixel_encoding_to_str(stream->timing.pixel_encoding),
+				      dc_color_depth_to_str(stream->timing.display_color_depth),
+				      dc_status_to_str(dc_result));
+
+			dc_stream_release(stream);
+			stream = NULL;
+			requested_bpc -= 2; /* lower bpc to retry validation */
+		}
+
+	} while (stream == NULL && requested_bpc >= bpc_limit);
+
+	switch (dc_result) {
+	/*
+	 * If we failed to validate DP bandwidth stream with the requested RGB color depth,
+	 * we try to fallback and configure in order:
+	 * YUV422 (8bpc, 6bpc)
+	 * YUV420 (8bpc, 6bpc)
+	 */
+	case DC_FAIL_ENC_VALIDATE:
+	case DC_EXCEED_DONGLE_CAP:
+	case DC_NO_DP_LINK_BANDWIDTH:
+		/* recursively entered twice and already tried both YUV422 and YUV420 */
+		if (aconnector->force_yuv422_output && aconnector->force_yuv420_output)
+			break;
+		/* first failure; try YUV422 */
+		if (!aconnector->force_yuv422_output) {
+			drm_dbg_kms(connector->dev, "%s:%d Validation failed with %d, retrying w/ YUV422\n",
+				    __func__, __LINE__, dc_result);
+			aconnector->force_yuv422_output = true;
+		/* recursively entered and YUV422 failed, try YUV420 */
+		} else if (!aconnector->force_yuv420_output) {
+			drm_dbg_kms(connector->dev, "%s:%d Validation failed with %d, retrying w/ YUV420\n",
+				    __func__, __LINE__, dc_result);
+			aconnector->force_yuv420_output = true;
+		}
+		stream = create_validate_stream_for_sink(connector, drm_mode,
+							 dm_state, old_stream);
+		aconnector->force_yuv422_output = false;
+		aconnector->force_yuv420_output = false;
+		break;
+	case DC_OK:
+		break;
+	default:
+		drm_dbg_kms(connector->dev, "%s:%d Unhandled validation failure %d\n",
+			    __func__, __LINE__, dc_result);
+		break;
+	}
+
+	return stream;
+}
+
+enum drm_mode_status amdgpu_dm_connector_mode_valid(struct drm_connector *connector,
+				   const struct drm_display_mode *mode)
+{
+	int result = MODE_ERROR;
+	struct dc_sink *dc_sink;
+	struct drm_display_mode *test_mode;
+	/* TODO: Unhardcode stream count */
+	struct dc_stream_state *stream;
+	/* we always have an amdgpu_dm_connector here since we got
+	 * here via the amdgpu_dm_connector_helper_funcs
+	 */
+	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector);
+
+	if ((mode->flags & DRM_MODE_FLAG_INTERLACE) ||
+			(mode->flags & DRM_MODE_FLAG_DBLSCAN))
+		return result;
+
+	/*
+	 * Only run this the first time mode_valid is called to initilialize
+	 * EDID mgmt
+	 */
+	if (aconnector->base.force != DRM_FORCE_UNSPECIFIED &&
+		!aconnector->dc_em_sink)
+		handle_edid_mgmt(aconnector);
+
+	dc_sink = to_amdgpu_dm_connector(connector)->dc_sink;
+
+	if (dc_sink == NULL && aconnector->base.force != DRM_FORCE_ON_DIGITAL &&
+				aconnector->base.force != DRM_FORCE_ON) {
+		drm_err(connector->dev, "dc_sink is NULL!\n");
+		goto fail;
+	}
+
+	test_mode = drm_mode_duplicate(connector->dev, mode);
+	if (!test_mode)
+		goto fail;
+
+	drm_mode_set_crtcinfo(test_mode, 0);
+
+	stream = create_validate_stream_for_sink(connector, test_mode,
+						 to_dm_connector_state(connector->state),
+						 NULL);
+	drm_mode_destroy(connector->dev, test_mode);
+	if (stream) {
+		dc_stream_release(stream);
+		result = MODE_OK;
+	}
+
+fail:
+	/* TODO: error handling*/
+	return result;
+}
+
+static int fill_hdr_info_packet(const struct drm_connector_state *state,
+				struct dc_info_packet *out)
+{
+	struct hdmi_drm_infoframe frame;
+	unsigned char buf[30]; /* 26 + 4 */
+	ssize_t len;
+	int ret, i;
+
+	memset(out, 0, sizeof(*out));
+
+	if (!state->hdr_output_metadata)
+		return 0;
+
+	ret = drm_hdmi_infoframe_set_hdr_metadata(&frame, state);
+	if (ret)
+		return ret;
+
+	len = hdmi_drm_infoframe_pack_only(&frame, buf, sizeof(buf));
+	if (len < 0)
+		return (int)len;
+
+	/* Static metadata is a fixed 26 bytes + 4 byte header. */
+	if (len != 30)
+		return -EINVAL;
+
+	/* Prepare the infopacket for DC. */
+	switch (state->connector->connector_type) {
+	case DRM_MODE_CONNECTOR_HDMIA:
+		out->hb0 = 0x87; /* type */
+		out->hb1 = 0x01; /* version */
+		out->hb2 = 0x1A; /* length */
+		out->sb[0] = buf[3]; /* checksum */
+		i = 1;
+		break;
+
+	case DRM_MODE_CONNECTOR_DisplayPort:
+	case DRM_MODE_CONNECTOR_eDP:
+		out->hb0 = 0x00; /* sdp id, zero */
+		out->hb1 = 0x87; /* type */
+		out->hb2 = 0x1D; /* payload len - 1 */
+		out->hb3 = (0x13 << 2); /* sdp version */
+		out->sb[0] = 0x01; /* version */
+		out->sb[1] = 0x1A; /* length */
+		i = 2;
+		break;
+
+	default:
+		return -EINVAL;
+	}
+
+	memcpy(&out->sb[i], &buf[4], 26);
+	out->valid = true;
+
+	print_hex_dump(KERN_DEBUG, "HDR SB:", DUMP_PREFIX_NONE, 16, 1, out->sb,
+		       sizeof(out->sb), false);
+
+	return 0;
+}
+
+static int
+amdgpu_dm_connector_atomic_check(struct drm_connector *conn,
+				 struct drm_atomic_state *state)
+{
+	struct drm_connector_state *new_con_state =
+		drm_atomic_get_new_connector_state(state, conn);
+	struct drm_connector_state *old_con_state =
+		drm_atomic_get_old_connector_state(state, conn);
+	struct drm_crtc *crtc = new_con_state->crtc;
+	struct drm_crtc_state *new_crtc_state;
+	struct amdgpu_dm_connector *aconn = to_amdgpu_dm_connector(conn);
+	int ret;
+
+	if (WARN_ON(unlikely(!old_con_state || !new_con_state)))
+		return -EINVAL;
+
+	trace_amdgpu_dm_connector_atomic_check(new_con_state);
+
+	if (conn->connector_type == DRM_MODE_CONNECTOR_DisplayPort) {
+		ret = drm_dp_mst_root_conn_atomic_check(new_con_state, &aconn->mst_mgr);
+		if (ret < 0)
+			return ret;
+	}
+
+	if (!crtc)
+		return 0;
+
+	if (new_con_state->privacy_screen_sw_state != old_con_state->privacy_screen_sw_state) {
+		new_crtc_state = drm_atomic_get_crtc_state(state, crtc);
+		if (IS_ERR(new_crtc_state))
+			return PTR_ERR(new_crtc_state);
+
+		new_crtc_state->mode_changed = true;
+	}
+
+	if (new_con_state->colorspace != old_con_state->colorspace) {
+		new_crtc_state = drm_atomic_get_crtc_state(state, crtc);
+		if (IS_ERR(new_crtc_state))
+			return PTR_ERR(new_crtc_state);
+
+		new_crtc_state->mode_changed = true;
+	}
+
+	if (new_con_state->content_type != old_con_state->content_type) {
+		new_crtc_state = drm_atomic_get_crtc_state(state, crtc);
+		if (IS_ERR(new_crtc_state))
+			return PTR_ERR(new_crtc_state);
+
+		new_crtc_state->mode_changed = true;
+	}
+
+	if (!drm_connector_atomic_hdr_metadata_equal(old_con_state, new_con_state)) {
+		struct dc_info_packet hdr_infopacket;
+
+		ret = fill_hdr_info_packet(new_con_state, &hdr_infopacket);
+		if (ret)
+			return ret;
+
+		new_crtc_state = drm_atomic_get_crtc_state(state, crtc);
+		if (IS_ERR(new_crtc_state))
+			return PTR_ERR(new_crtc_state);
+
+		/*
+		 * DC considers the stream backends changed if the
+		 * static metadata changes. Forcing the modeset also
+		 * gives a simple way for userspace to switch from
+		 * 8bpc to 10bpc when setting the metadata to enter
+		 * or exit HDR.
+		 *
+		 * Changing the static metadata after it's been
+		 * set is permissible, however. So only force a
+		 * modeset if we're entering or exiting HDR.
+		 */
+		new_crtc_state->mode_changed = new_crtc_state->mode_changed ||
+			!old_con_state->hdr_output_metadata ||
+			!new_con_state->hdr_output_metadata;
+	}
+
+	return 0;
+}
+
+static const struct drm_connector_helper_funcs
+amdgpu_dm_connector_helper_funcs = {
+	/*
+	 * If hotplugging a second bigger display in FB Con mode, bigger resolution
+	 * modes will be filtered by drm_mode_validate_size(), and those modes
+	 * are missing after user start lightdm. So we need to renew modes list.
+	 * in get_modes call back, not just return the modes count
+	 */
+	.get_modes = get_modes,
+	.mode_valid = amdgpu_dm_connector_mode_valid,
+	.atomic_check = amdgpu_dm_connector_atomic_check,
+};
+
+static void dm_encoder_helper_disable(struct drm_encoder *encoder)
+{
+
+}
+
+int convert_dc_color_depth_into_bpc(enum dc_color_depth display_color_depth)
+{
+	switch (display_color_depth) {
+	case COLOR_DEPTH_666:
+		return 6;
+	case COLOR_DEPTH_888:
+		return 8;
+	case COLOR_DEPTH_101010:
+		return 10;
+	case COLOR_DEPTH_121212:
+		return 12;
+	case COLOR_DEPTH_141414:
+		return 14;
+	case COLOR_DEPTH_161616:
+		return 16;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int dm_encoder_helper_atomic_check(struct drm_encoder *encoder,
+					  struct drm_crtc_state *crtc_state,
+					  struct drm_connector_state *conn_state)
+{
+	struct drm_atomic_state *state = crtc_state->state;
+	struct drm_connector *connector = conn_state->connector;
+	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector);
+	struct dm_connector_state *dm_new_connector_state = to_dm_connector_state(conn_state);
+	const struct drm_display_mode *adjusted_mode = &crtc_state->adjusted_mode;
+	struct drm_dp_mst_topology_mgr *mst_mgr;
+	struct drm_dp_mst_port *mst_port;
+	struct drm_dp_mst_topology_state *mst_state;
+	enum dc_color_depth color_depth;
+	int clock, bpp = 0;
+	bool is_y420 = false;
+
+	if ((connector->connector_type == DRM_MODE_CONNECTOR_eDP) ||
+	    (connector->connector_type == DRM_MODE_CONNECTOR_LVDS)) {
+		struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+		struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode;
+		enum drm_mode_status result;
+
+		result = drm_crtc_helper_mode_valid_fixed(encoder->crtc, adjusted_mode, native_mode);
+		if (result != MODE_OK && dm_new_connector_state->scaling == RMX_OFF) {
+			drm_dbg_driver(encoder->dev,
+				       "mode %dx%d@%dHz is not native, enabling scaling\n",
+				       adjusted_mode->hdisplay, adjusted_mode->vdisplay,
+				       drm_mode_vrefresh(adjusted_mode));
+			dm_new_connector_state->scaling = RMX_FULL;
+		}
+		return 0;
+	}
+
+	if (!aconnector->mst_output_port)
+		return 0;
+
+	mst_port = aconnector->mst_output_port;
+	mst_mgr = &aconnector->mst_root->mst_mgr;
+
+	if (!crtc_state->connectors_changed && !crtc_state->mode_changed)
+		return 0;
+
+	mst_state = drm_atomic_get_mst_topology_state(state, mst_mgr);
+	if (IS_ERR(mst_state))
+		return PTR_ERR(mst_state);
+
+	mst_state->pbn_div.full = dm_mst_get_pbn_divider(aconnector->mst_root->dc_link);
+
+	if (!state->duplicated) {
+		int max_bpc = conn_state->max_requested_bpc;
+
+		is_y420 = drm_mode_is_420_also(&connector->display_info, adjusted_mode) &&
+			  aconnector->force_yuv420_output;
+		color_depth = convert_color_depth_from_display_info(connector,
+								    is_y420,
+								    max_bpc);
+		bpp = convert_dc_color_depth_into_bpc(color_depth) * 3;
+		clock = adjusted_mode->clock;
+		dm_new_connector_state->pbn = drm_dp_calc_pbn_mode(clock, bpp << 4);
+	}
+
+	dm_new_connector_state->vcpi_slots =
+		drm_dp_atomic_find_time_slots(state, mst_mgr, mst_port,
+					      dm_new_connector_state->pbn);
+	if (dm_new_connector_state->vcpi_slots < 0) {
+		DRM_DEBUG_ATOMIC("failed finding vcpi slots: %d\n", (int)dm_new_connector_state->vcpi_slots);
+		return dm_new_connector_state->vcpi_slots;
+	}
+	return 0;
+}
+
+const struct drm_encoder_helper_funcs amdgpu_dm_encoder_helper_funcs = {
+	.disable = dm_encoder_helper_disable,
+	.atomic_check = dm_encoder_helper_atomic_check
+};
+
+static int dm_update_mst_vcpi_slots_for_dsc(struct drm_atomic_state *state,
+					    struct dc_state *dc_state,
+					    struct dsc_mst_fairness_vars *vars)
+{
+	struct dc_stream_state *stream = NULL;
+	struct drm_connector *connector;
+	struct drm_connector_state *new_con_state;
+	struct amdgpu_dm_connector *aconnector;
+	struct dm_connector_state *dm_conn_state;
+	int i, j, ret;
+	int vcpi, pbn_div, pbn = 0, slot_num = 0;
+
+	for_each_new_connector_in_state(state, connector, new_con_state, i) {
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_WRITEBACK)
+			continue;
+
+		aconnector = to_amdgpu_dm_connector(connector);
+
+		if (!aconnector->mst_output_port)
+			continue;
+
+		if (!new_con_state || !new_con_state->crtc)
+			continue;
+
+		dm_conn_state = to_dm_connector_state(new_con_state);
+
+		for (j = 0; j < dc_state->stream_count; j++) {
+			stream = dc_state->streams[j];
+			if (!stream)
+				continue;
+
+			if ((struct amdgpu_dm_connector *)stream->dm_stream_context == aconnector)
+				break;
+
+			stream = NULL;
+		}
+
+		if (!stream)
+			continue;
+
+		pbn_div = dm_mst_get_pbn_divider(stream->link);
+		/* pbn is calculated by compute_mst_dsc_configs_for_state*/
+		for (j = 0; j < dc_state->stream_count; j++) {
+			if (vars[j].aconnector == aconnector) {
+				pbn = vars[j].pbn;
+				break;
+			}
+		}
+
+		if (j == dc_state->stream_count || pbn_div == 0)
+			continue;
+
+		slot_num = DIV_ROUND_UP(pbn, pbn_div);
+
+		if (stream->timing.flags.DSC != 1) {
+			dm_conn_state->pbn = pbn;
+			dm_conn_state->vcpi_slots = slot_num;
+
+			ret = drm_dp_mst_atomic_enable_dsc(state, aconnector->mst_output_port,
+							   dm_conn_state->pbn, false);
+			if (ret < 0)
+				return ret;
+
+			continue;
+		}
+
+		vcpi = drm_dp_mst_atomic_enable_dsc(state, aconnector->mst_output_port, pbn, true);
+		if (vcpi < 0)
+			return vcpi;
+
+		dm_conn_state->pbn = pbn;
+		dm_conn_state->vcpi_slots = vcpi;
+	}
+	return 0;
+}
+
+static int to_drm_connector_type(enum signal_type st)
+{
+	switch (st) {
+	case SIGNAL_TYPE_HDMI_TYPE_A:
+		return DRM_MODE_CONNECTOR_HDMIA;
+	case SIGNAL_TYPE_EDP:
+		return DRM_MODE_CONNECTOR_eDP;
+	case SIGNAL_TYPE_LVDS:
+		return DRM_MODE_CONNECTOR_LVDS;
+	case SIGNAL_TYPE_RGB:
+		return DRM_MODE_CONNECTOR_VGA;
+	case SIGNAL_TYPE_DISPLAY_PORT:
+	case SIGNAL_TYPE_DISPLAY_PORT_MST:
+		return DRM_MODE_CONNECTOR_DisplayPort;
+	case SIGNAL_TYPE_DVI_DUAL_LINK:
+	case SIGNAL_TYPE_DVI_SINGLE_LINK:
+		return DRM_MODE_CONNECTOR_DVID;
+	case SIGNAL_TYPE_VIRTUAL:
+		return DRM_MODE_CONNECTOR_VIRTUAL;
+
+	default:
+		return DRM_MODE_CONNECTOR_Unknown;
+	}
+}
+
+static struct drm_encoder *amdgpu_dm_connector_to_encoder(struct drm_connector *connector)
+{
+	struct drm_encoder *encoder;
+
+	/* There is only one encoder per connector */
+	drm_connector_for_each_possible_encoder(connector, encoder)
+		return encoder;
+
+	return NULL;
+}
+
+static void amdgpu_dm_get_native_mode(struct drm_connector *connector)
+{
+	struct drm_encoder *encoder;
+	struct amdgpu_encoder *amdgpu_encoder;
+
+	encoder = amdgpu_dm_connector_to_encoder(connector);
+
+	if (encoder == NULL)
+		return;
+
+	amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+	amdgpu_encoder->native_mode.clock = 0;
+
+	if (!list_empty(&connector->probed_modes)) {
+		struct drm_display_mode *preferred_mode = NULL;
+
+		list_for_each_entry(preferred_mode,
+				    &connector->probed_modes,
+				    head) {
+			if (preferred_mode->type & DRM_MODE_TYPE_PREFERRED)
+				amdgpu_encoder->native_mode = *preferred_mode;
+
+			break;
+		}
+
+	}
+}
+
+static struct drm_display_mode *
+amdgpu_dm_create_common_mode(struct drm_encoder *encoder,
+			     char *name,
+			     int hdisplay, int vdisplay)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_display_mode *mode = NULL;
+	struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode;
+
+	mode = drm_mode_duplicate(dev, native_mode);
+
+	if (mode == NULL)
+		return NULL;
+
+	mode->hdisplay = hdisplay;
+	mode->vdisplay = vdisplay;
+	mode->type &= ~DRM_MODE_TYPE_PREFERRED;
+	strscpy(mode->name, name, DRM_DISPLAY_MODE_LEN);
+
+	return mode;
+
+}
+
+static void amdgpu_dm_connector_add_common_modes(struct drm_encoder *encoder,
+						 struct drm_connector *connector)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_display_mode *mode = NULL;
+	struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode;
+	struct amdgpu_dm_connector *amdgpu_dm_connector =
+				to_amdgpu_dm_connector(connector);
+	int i;
+	int n;
+	struct mode_size {
+		char name[DRM_DISPLAY_MODE_LEN];
+		int w;
+		int h;
+	} common_modes[] = {
+		{  "640x480",  640,  480},
+		{  "800x600",  800,  600},
+		{ "1024x768", 1024,  768},
+		{ "1280x720", 1280,  720},
+		{ "1280x800", 1280,  800},
+		{"1280x1024", 1280, 1024},
+		{ "1440x900", 1440,  900},
+		{"1680x1050", 1680, 1050},
+		{"1600x1200", 1600, 1200},
+		{"1920x1080", 1920, 1080},
+		{"1920x1200", 1920, 1200}
+	};
+
+	if ((connector->connector_type != DRM_MODE_CONNECTOR_eDP) &&
+	    (connector->connector_type != DRM_MODE_CONNECTOR_LVDS))
+		return;
+
+	n = ARRAY_SIZE(common_modes);
+
+	for (i = 0; i < n; i++) {
+		struct drm_display_mode *curmode = NULL;
+		bool mode_existed = false;
+
+		if (common_modes[i].w > native_mode->hdisplay ||
+		    common_modes[i].h > native_mode->vdisplay ||
+		   (common_modes[i].w == native_mode->hdisplay &&
+		    common_modes[i].h == native_mode->vdisplay))
+			continue;
+
+		list_for_each_entry(curmode, &connector->probed_modes, head) {
+			if (common_modes[i].w == curmode->hdisplay &&
+			    common_modes[i].h == curmode->vdisplay) {
+				mode_existed = true;
+				break;
+			}
+		}
+
+		if (mode_existed)
+			continue;
+
+		mode = amdgpu_dm_create_common_mode(encoder,
+				common_modes[i].name, common_modes[i].w,
+				common_modes[i].h);
+		if (!mode)
+			continue;
+
+		drm_mode_probed_add(connector, mode);
+		amdgpu_dm_connector->num_modes++;
+	}
+}
+
+static void amdgpu_set_panel_orientation(struct drm_connector *connector)
+{
+	struct drm_encoder *encoder;
+	struct amdgpu_encoder *amdgpu_encoder;
+	const struct drm_display_mode *native_mode;
+
+	if (connector->connector_type != DRM_MODE_CONNECTOR_eDP &&
+	    connector->connector_type != DRM_MODE_CONNECTOR_LVDS)
+		return;
+
+	mutex_lock(&connector->dev->mode_config.mutex);
+	amdgpu_dm_connector_get_modes(connector);
+	mutex_unlock(&connector->dev->mode_config.mutex);
+
+	encoder = amdgpu_dm_connector_to_encoder(connector);
+	if (!encoder)
+		return;
+
+	amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+	native_mode = &amdgpu_encoder->native_mode;
+	if (native_mode->hdisplay == 0 || native_mode->vdisplay == 0)
+		return;
+
+	drm_connector_set_panel_orientation_with_quirk(connector,
+						       DRM_MODE_PANEL_ORIENTATION_UNKNOWN,
+						       native_mode->hdisplay,
+						       native_mode->vdisplay);
+}
+
+static void amdgpu_dm_connector_ddc_get_modes(struct drm_connector *connector,
+					      const struct drm_edid *drm_edid)
+{
+	struct amdgpu_dm_connector *amdgpu_dm_connector =
+			to_amdgpu_dm_connector(connector);
+
+	if (drm_edid) {
+		/* empty probed_modes */
+		INIT_LIST_HEAD(&connector->probed_modes);
+		amdgpu_dm_connector->num_modes =
+				drm_edid_connector_add_modes(connector);
+
+		/* sorting the probed modes before calling function
+		 * amdgpu_dm_get_native_mode() since EDID can have
+		 * more than one preferred mode. The modes that are
+		 * later in the probed mode list could be of higher
+		 * and preferred resolution. For example, 3840x2160
+		 * resolution in base EDID preferred timing and 4096x2160
+		 * preferred resolution in DID extension block later.
+		 */
+		drm_mode_sort(&connector->probed_modes);
+		amdgpu_dm_get_native_mode(connector);
+
+		/* Freesync capabilities are reset by calling
+		 * drm_edid_connector_add_modes() and need to be
+		 * restored here.
+		 */
+		amdgpu_dm_update_freesync_caps(connector, drm_edid);
+	} else {
+		amdgpu_dm_connector->num_modes = 0;
+	}
+}
+
+static bool is_duplicate_mode(struct amdgpu_dm_connector *aconnector,
+			      struct drm_display_mode *mode)
+{
+	struct drm_display_mode *m;
+
+	list_for_each_entry(m, &aconnector->base.probed_modes, head) {
+		if (drm_mode_equal(m, mode))
+			return true;
+	}
+
+	return false;
+}
+
+static uint add_fs_modes(struct amdgpu_dm_connector *aconnector)
+{
+	const struct drm_display_mode *m;
+	struct drm_display_mode *new_mode;
+	uint i;
+	u32 new_modes_count = 0;
+
+	/* Standard FPS values
+	 *
+	 * 23.976       - TV/NTSC
+	 * 24           - Cinema
+	 * 25           - TV/PAL
+	 * 29.97        - TV/NTSC
+	 * 30           - TV/NTSC
+	 * 48           - Cinema HFR
+	 * 50           - TV/PAL
+	 * 60           - Commonly used
+	 * 48,72,96,120 - Multiples of 24
+	 */
+	static const u32 common_rates[] = {
+		23976, 24000, 25000, 29970, 30000,
+		48000, 50000, 60000, 72000, 96000, 120000
+	};
+
+	/*
+	 * Find mode with highest refresh rate with the same resolution
+	 * as the preferred mode. Some monitors report a preferred mode
+	 * with lower resolution than the highest refresh rate supported.
+	 */
+
+	m = get_highest_refresh_rate_mode(aconnector, true);
+	if (!m)
+		return 0;
+
+	for (i = 0; i < ARRAY_SIZE(common_rates); i++) {
+		u64 target_vtotal, target_vtotal_diff;
+		u64 num, den;
+
+		if (drm_mode_vrefresh(m) * 1000 < common_rates[i])
+			continue;
+
+		if (common_rates[i] < aconnector->min_vfreq * 1000 ||
+		    common_rates[i] > aconnector->max_vfreq * 1000)
+			continue;
+
+		num = (unsigned long long)m->clock * 1000 * 1000;
+		den = common_rates[i] * (unsigned long long)m->htotal;
+		target_vtotal = div_u64(num, den);
+		target_vtotal_diff = target_vtotal - m->vtotal;
+
+		/* Check for illegal modes */
+		if (m->vsync_start + target_vtotal_diff < m->vdisplay ||
+		    m->vsync_end + target_vtotal_diff < m->vsync_start ||
+		    m->vtotal + target_vtotal_diff < m->vsync_end)
+			continue;
+
+		new_mode = drm_mode_duplicate(aconnector->base.dev, m);
+		if (!new_mode)
+			goto out;
+
+		new_mode->vtotal += (u16)target_vtotal_diff;
+		new_mode->vsync_start += (u16)target_vtotal_diff;
+		new_mode->vsync_end += (u16)target_vtotal_diff;
+		new_mode->type &= ~DRM_MODE_TYPE_PREFERRED;
+		new_mode->type |= DRM_MODE_TYPE_DRIVER;
+
+		if (!is_duplicate_mode(aconnector, new_mode)) {
+			drm_mode_probed_add(&aconnector->base, new_mode);
+			new_modes_count += 1;
+		} else
+			drm_mode_destroy(aconnector->base.dev, new_mode);
+	}
+ out:
+	return new_modes_count;
+}
+
+static void amdgpu_dm_connector_add_freesync_modes(struct drm_connector *connector,
+						   const struct drm_edid *drm_edid)
+{
+	struct amdgpu_dm_connector *amdgpu_dm_connector =
+		to_amdgpu_dm_connector(connector);
+
+	if (!(amdgpu_freesync_vid_mode && drm_edid))
+		return;
+
+	if (amdgpu_dm_connector->max_vfreq - amdgpu_dm_connector->min_vfreq > 10)
+		amdgpu_dm_connector->num_modes +=
+			add_fs_modes(amdgpu_dm_connector);
+}
+
+static int amdgpu_dm_connector_get_modes(struct drm_connector *connector)
+{
+	struct amdgpu_dm_connector *amdgpu_dm_connector =
+			to_amdgpu_dm_connector(connector);
+	struct drm_encoder *encoder;
+	const struct drm_edid *drm_edid = amdgpu_dm_connector->drm_edid;
+	struct dc_link_settings *verified_link_cap =
+			&amdgpu_dm_connector->dc_link->verified_link_cap;
+	const struct dc *dc = amdgpu_dm_connector->dc_link->dc;
+
+	encoder = amdgpu_dm_connector_to_encoder(connector);
+
+	if (!drm_edid) {
+		amdgpu_dm_connector->num_modes =
+				drm_add_modes_noedid(connector, 640, 480);
+		if (dc->link_srv->dp_get_encoding_format(verified_link_cap) == DP_128b_132b_ENCODING)
+			amdgpu_dm_connector->num_modes +=
+				drm_add_modes_noedid(connector, 1920, 1080);
+	} else {
+		amdgpu_dm_connector_ddc_get_modes(connector, drm_edid);
+		if (encoder)
+			amdgpu_dm_connector_add_common_modes(encoder, connector);
+		amdgpu_dm_connector_add_freesync_modes(connector, drm_edid);
+	}
+	amdgpu_dm_fbc_init(connector);
+
+	return amdgpu_dm_connector->num_modes;
+}
+
+static const u32 supported_colorspaces =
+	BIT(DRM_MODE_COLORIMETRY_BT709_YCC) |
+	BIT(DRM_MODE_COLORIMETRY_OPRGB) |
+	BIT(DRM_MODE_COLORIMETRY_BT2020_RGB) |
+	BIT(DRM_MODE_COLORIMETRY_BT2020_YCC);
+
+void amdgpu_dm_connector_init_helper(struct amdgpu_display_manager *dm,
+				     struct amdgpu_dm_connector *aconnector,
+				     int connector_type,
+				     struct dc_link *link,
+				     int link_index)
+{
+	struct amdgpu_device *adev = drm_to_adev(dm->ddev);
+
+	/*
+	 * Some of the properties below require access to state, like bpc.
+	 * Allocate some default initial connector state with our reset helper.
+	 */
+	if (aconnector->base.funcs->reset)
+		aconnector->base.funcs->reset(&aconnector->base);
+
+	aconnector->connector_id = link_index;
+	aconnector->bl_idx = -1;
+	aconnector->dc_link = link;
+	aconnector->base.interlace_allowed = false;
+	aconnector->base.doublescan_allowed = false;
+	aconnector->base.stereo_allowed = false;
+	aconnector->base.dpms = DRM_MODE_DPMS_OFF;
+	aconnector->hpd.hpd = AMDGPU_HPD_NONE; /* not used */
+	aconnector->audio_inst = -1;
+	aconnector->pack_sdp_v1_3 = false;
+	aconnector->as_type = ADAPTIVE_SYNC_TYPE_NONE;
+	memset(&aconnector->vsdb_info, 0, sizeof(aconnector->vsdb_info));
+	mutex_init(&aconnector->hpd_lock);
+	mutex_init(&aconnector->handle_mst_msg_ready);
+
+	/*
+	 * configure support HPD hot plug connector_>polled default value is 0
+	 * which means HPD hot plug not supported
+	 */
+	switch (connector_type) {
+	case DRM_MODE_CONNECTOR_HDMIA:
+		aconnector->base.polled = DRM_CONNECTOR_POLL_HPD;
+		aconnector->base.ycbcr_420_allowed =
+			link->link_enc->features.hdmi_ycbcr420_supported ? true : false;
+		break;
+	case DRM_MODE_CONNECTOR_DisplayPort:
+		aconnector->base.polled = DRM_CONNECTOR_POLL_HPD;
+		link->link_enc = link_enc_cfg_get_link_enc(link);
+		ASSERT(link->link_enc);
+		if (link->link_enc)
+			aconnector->base.ycbcr_420_allowed =
+			link->link_enc->features.dp_ycbcr420_supported ? true : false;
+		break;
+	case DRM_MODE_CONNECTOR_DVID:
+		aconnector->base.polled = DRM_CONNECTOR_POLL_HPD;
+		break;
+	default:
+		break;
+	}
+
+	drm_object_attach_property(&aconnector->base.base,
+				dm->ddev->mode_config.scaling_mode_property,
+				DRM_MODE_SCALE_NONE);
+
+	if (connector_type == DRM_MODE_CONNECTOR_HDMIA
+		|| (connector_type == DRM_MODE_CONNECTOR_DisplayPort && !aconnector->mst_root))
+		drm_connector_attach_broadcast_rgb_property(&aconnector->base);
+
+	drm_object_attach_property(&aconnector->base.base,
+				adev->mode_info.underscan_property,
+				UNDERSCAN_OFF);
+	drm_object_attach_property(&aconnector->base.base,
+				adev->mode_info.underscan_hborder_property,
+				0);
+	drm_object_attach_property(&aconnector->base.base,
+				adev->mode_info.underscan_vborder_property,
+				0);
+
+	if (!aconnector->mst_root)
+		drm_connector_attach_max_bpc_property(&aconnector->base, 8, 16);
+
+	aconnector->base.state->max_bpc = 16;
+	aconnector->base.state->max_requested_bpc = aconnector->base.state->max_bpc;
+
+	if (connector_type == DRM_MODE_CONNECTOR_HDMIA) {
+		/* Content Type is currently only implemented for HDMI. */
+		drm_connector_attach_content_type_property(&aconnector->base);
+	}
+
+	if (connector_type == DRM_MODE_CONNECTOR_HDMIA) {
+		if (!drm_mode_create_hdmi_colorspace_property(&aconnector->base, supported_colorspaces))
+			drm_connector_attach_colorspace_property(&aconnector->base);
+	} else if ((connector_type == DRM_MODE_CONNECTOR_DisplayPort && !aconnector->mst_root) ||
+		   connector_type == DRM_MODE_CONNECTOR_eDP) {
+		if (!drm_mode_create_dp_colorspace_property(&aconnector->base, supported_colorspaces))
+			drm_connector_attach_colorspace_property(&aconnector->base);
+	}
+
+	if (connector_type == DRM_MODE_CONNECTOR_HDMIA ||
+	    connector_type == DRM_MODE_CONNECTOR_DisplayPort ||
+	    connector_type == DRM_MODE_CONNECTOR_eDP) {
+		drm_connector_attach_hdr_output_metadata_property(&aconnector->base);
+
+		if (!aconnector->mst_root)
+			drm_connector_attach_vrr_capable_property(&aconnector->base);
+
+		if (adev->dm.hdcp_workqueue)
+			drm_connector_attach_content_protection_property(&aconnector->base, true);
+	}
+
+	if (connector_type == DRM_MODE_CONNECTOR_eDP) {
+		struct drm_privacy_screen *privacy_screen;
+
+		privacy_screen = drm_privacy_screen_get(adev_to_drm(adev)->dev, NULL);
+		if (!IS_ERR(privacy_screen)) {
+			drm_connector_attach_privacy_screen_provider(&aconnector->base,
+								     privacy_screen);
+		} else if (PTR_ERR(privacy_screen) != -ENODEV) {
+			drm_warn(adev_to_drm(adev), "Error getting privacy-screen\n");
+		}
+	}
+}
+
+static int amdgpu_dm_i2c_xfer(struct i2c_adapter *i2c_adap,
+			      struct i2c_msg *msgs, int num)
+{
+	struct amdgpu_i2c_adapter *i2c = i2c_get_adapdata(i2c_adap);
+	struct ddc_service *ddc_service = i2c->ddc_service;
+	struct i2c_command cmd;
+	int i;
+	int result = -EIO;
+
+	if (!ddc_service->ddc_pin)
+		return result;
+
+	cmd.payloads = kcalloc(num, sizeof(struct i2c_payload), GFP_KERNEL);
+
+	if (!cmd.payloads)
+		return result;
+
+	cmd.number_of_payloads = num;
+	cmd.engine = I2C_COMMAND_ENGINE_DEFAULT;
+	cmd.speed = 100;
+
+	for (i = 0; i < num; i++) {
+		cmd.payloads[i].write = !(msgs[i].flags & I2C_M_RD);
+		cmd.payloads[i].address = msgs[i].addr;
+		cmd.payloads[i].length = msgs[i].len;
+		cmd.payloads[i].data = msgs[i].buf;
+	}
+
+	if (i2c->oem) {
+		if (dc_submit_i2c_oem(
+			    ddc_service->ctx->dc,
+			    &cmd))
+			result = num;
+	} else {
+		if (dc_submit_i2c(
+			    ddc_service->ctx->dc,
+			    ddc_service->link->link_index,
+			    &cmd))
+			result = num;
+	}
+
+	kfree(cmd.payloads);
+	return result;
+}
+
+static u32 amdgpu_dm_i2c_func(struct i2c_adapter *adap)
+{
+	return I2C_FUNC_I2C | I2C_FUNC_SMBUS_EMUL;
+}
+
+static const struct i2c_algorithm amdgpu_dm_i2c_algo = {
+	.master_xfer = amdgpu_dm_i2c_xfer,
+	.functionality = amdgpu_dm_i2c_func,
+};
+
+static struct amdgpu_i2c_adapter *
+create_i2c(struct ddc_service *ddc_service, bool oem)
+{
+	struct amdgpu_device *adev = ddc_service->ctx->driver_context;
+	struct amdgpu_i2c_adapter *i2c;
+
+	i2c = kzalloc(sizeof(struct amdgpu_i2c_adapter), GFP_KERNEL);
+	if (!i2c)
+		return NULL;
+	i2c->base.owner = THIS_MODULE;
+	i2c->base.dev.parent = &adev->pdev->dev;
+	i2c->base.algo = &amdgpu_dm_i2c_algo;
+	if (oem)
+		snprintf(i2c->base.name, sizeof(i2c->base.name), "AMDGPU DM i2c OEM bus");
+	else
+		snprintf(i2c->base.name, sizeof(i2c->base.name), "AMDGPU DM i2c hw bus %d",
+			 ddc_service->link->link_index);
+	i2c_set_adapdata(&i2c->base, i2c);
+	i2c->ddc_service = ddc_service;
+	i2c->oem = oem;
+
+	return i2c;
+}
+
+int amdgpu_dm_initialize_hdmi_connector(struct amdgpu_dm_connector *aconnector)
+{
+	struct cec_connector_info conn_info;
+	struct drm_device *ddev = aconnector->base.dev;
+	struct device *hdmi_dev = ddev->dev;
+
+	if (amdgpu_dc_debug_mask & DC_DISABLE_HDMI_CEC) {
+		drm_info(ddev, "HDMI-CEC feature masked\n");
+		return -EINVAL;
+	}
+
+	cec_fill_conn_info_from_drm(&conn_info, &aconnector->base);
+	aconnector->notifier =
+		cec_notifier_conn_register(hdmi_dev, NULL, &conn_info);
+	if (!aconnector->notifier) {
+		drm_err(ddev, "Failed to create cec notifier\n");
+		return -ENOMEM;
+	}
+
+	return 0;
+}
+
+/*
+ * Note: this function assumes that dc_link_detect() was called for the
+ * dc_link which will be represented by this aconnector.
+ */
+static int amdgpu_dm_connector_init(struct amdgpu_display_manager *dm,
+				    struct amdgpu_dm_connector *aconnector,
+				    u32 link_index,
+				    struct amdgpu_encoder *aencoder)
+{
+	int res = 0;
+	int connector_type;
+	struct dc *dc = dm->dc;
+	struct dc_link *link = dc_get_link_at_index(dc, link_index);
+	struct amdgpu_i2c_adapter *i2c;
+
+	/* Not needed for writeback connector */
+	link->priv = aconnector;
+
+
+	i2c = create_i2c(link->ddc, false);
+	if (!i2c) {
+		drm_err(adev_to_drm(dm->adev), "Failed to create i2c adapter data\n");
+		return -ENOMEM;
+	}
+
+	aconnector->i2c = i2c;
+	res = devm_i2c_add_adapter(dm->adev->dev, &i2c->base);
+
+	if (res) {
+		drm_err(adev_to_drm(dm->adev), "Failed to register hw i2c %d\n", link->link_index);
+		goto out_free;
+	}
+
+	connector_type = to_drm_connector_type(link->connector_signal);
+
+	res = drm_connector_init_with_ddc(
+			dm->ddev,
+			&aconnector->base,
+			&amdgpu_dm_connector_funcs,
+			connector_type,
+			&i2c->base);
+
+	if (res) {
+		drm_err(adev_to_drm(dm->adev), "connector_init failed\n");
+		aconnector->connector_id = -1;
+		goto out_free;
+	}
+
+	drm_connector_helper_add(
+			&aconnector->base,
+			&amdgpu_dm_connector_helper_funcs);
+
+	amdgpu_dm_connector_init_helper(
+		dm,
+		aconnector,
+		connector_type,
+		link,
+		link_index);
+
+	drm_connector_attach_encoder(
+		&aconnector->base, &aencoder->base);
+
+	if (connector_type == DRM_MODE_CONNECTOR_HDMIA ||
+	    connector_type == DRM_MODE_CONNECTOR_HDMIB)
+		amdgpu_dm_initialize_hdmi_connector(aconnector);
+
+	if (connector_type == DRM_MODE_CONNECTOR_DisplayPort
+		|| connector_type == DRM_MODE_CONNECTOR_eDP)
+		amdgpu_dm_initialize_dp_connector(dm, aconnector, link->link_index);
+
+out_free:
+	if (res) {
+		kfree(i2c);
+		aconnector->i2c = NULL;
+	}
+	return res;
+}
+
+int amdgpu_dm_get_encoder_crtc_mask(struct amdgpu_device *adev)
+{
+	switch (adev->mode_info.num_crtc) {
+	case 1:
+		return 0x1;
+	case 2:
+		return 0x3;
+	case 3:
+		return 0x7;
+	case 4:
+		return 0xf;
+	case 5:
+		return 0x1f;
+	case 6:
+	default:
+		return 0x3f;
+	}
+}
+
+static int amdgpu_dm_encoder_init(struct drm_device *dev,
+				  struct amdgpu_encoder *aencoder,
+				  uint32_t link_index)
+{
+	struct amdgpu_device *adev = drm_to_adev(dev);
+
+	int res = drm_encoder_init(dev,
+				   &aencoder->base,
+				   &amdgpu_dm_encoder_funcs,
+				   DRM_MODE_ENCODER_TMDS,
+				   NULL);
+
+	aencoder->base.possible_crtcs = amdgpu_dm_get_encoder_crtc_mask(adev);
+
+	if (!res)
+		aencoder->encoder_id = link_index;
+	else
+		aencoder->encoder_id = -1;
+
+	drm_encoder_helper_add(&aencoder->base, &amdgpu_dm_encoder_helper_funcs);
+
+	return res;
+}
+
+static void manage_dm_interrupts(struct amdgpu_device *adev,
+				 struct amdgpu_crtc *acrtc,
+				 struct dm_crtc_state *acrtc_state)
+{	/*
+	 * We cannot be sure that the frontend index maps to the same
+	 * backend index - some even map to more than one.
+	 * So we have to go through the CRTC to find the right IRQ.
+	 */
+	int irq_type = amdgpu_display_crtc_idx_to_irq_type(
+			adev,
+			acrtc->crtc_id);
+	struct drm_device *dev = adev_to_drm(adev);
+
+	struct drm_vblank_crtc_config config = {0};
+	struct dc_crtc_timing *timing;
+	int offdelay;
+
+	if (acrtc_state) {
+		timing = &acrtc_state->stream->timing;
+
+		/*
+		 * Depending on when the HW latching event of double-buffered
+		 * registers happen relative to the PSR SDP deadline, and how
+		 * bad the Panel clock has drifted since the last ALPM off
+		 * event, there can be up to 3 frames of delay between sending
+		 * the PSR exit cmd to DMUB fw, and when the panel starts
+		 * displaying live frames.
+		 *
+		 * We can set:
+		 *
+		 * 20/100 * offdelay_ms = 3_frames_ms
+		 * => offdelay_ms = 5 * 3_frames_ms
+		 *
+		 * This ensures that `3_frames_ms` will only be experienced as a
+		 * 20% delay on top how long the display has been static, and
+		 * thus make the delay less perceivable.
+		 */
+		if (acrtc_state->stream->link->psr_settings.psr_version <
+		    DC_PSR_VERSION_UNSUPPORTED) {
+			offdelay = DIV64_U64_ROUND_UP((u64)5 * 3 * 10 *
+						      timing->v_total *
+						      timing->h_total,
+						      timing->pix_clk_100hz);
+			config.offdelay_ms = offdelay ?: 30;
+		} else if (amdgpu_ip_version(adev, DCE_HWIP, 0) <
+			   IP_VERSION(3, 5, 0) ||
+			   !(adev->flags & AMD_IS_APU)) {
+			/*
+			 * Older HW and DGPU have issues with instant off;
+			 * use a 2 frame offdelay.
+			 */
+			offdelay = DIV64_U64_ROUND_UP((u64)20 *
+						      timing->v_total *
+						      timing->h_total,
+						      timing->pix_clk_100hz);
+
+			config.offdelay_ms = offdelay ?: 30;
+		} else {
+			/* offdelay_ms = 0 will never disable vblank */
+			config.offdelay_ms = 1;
+			config.disable_immediate = true;
+		}
+
+		drm_crtc_vblank_on_config(&acrtc->base,
+					  &config);
+		/* Allow RX6xxx, RX7700, RX7800 GPUs to call amdgpu_irq_get.*/
+		switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) {
+		case IP_VERSION(3, 0, 0):
+		case IP_VERSION(3, 0, 2):
+		case IP_VERSION(3, 0, 3):
+		case IP_VERSION(3, 2, 0):
+			if (amdgpu_irq_get(adev, &adev->pageflip_irq, irq_type))
+				drm_err(dev, "DM_IRQ: Cannot get pageflip irq!\n");
+#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY)
+			if (amdgpu_irq_get(adev, &adev->vline0_irq, irq_type))
+				drm_err(dev, "DM_IRQ: Cannot get vline0 irq!\n");
+#endif
+		}
+
+	} else {
+		/* Allow RX6xxx, RX7700, RX7800 GPUs to call amdgpu_irq_put.*/
+		switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) {
+		case IP_VERSION(3, 0, 0):
+		case IP_VERSION(3, 0, 2):
+		case IP_VERSION(3, 0, 3):
+		case IP_VERSION(3, 2, 0):
+#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY)
+			if (amdgpu_irq_put(adev, &adev->vline0_irq, irq_type))
+				drm_err(dev, "DM_IRQ: Cannot put vline0 irq!\n");
+#endif
+			if (amdgpu_irq_put(adev, &adev->pageflip_irq, irq_type))
+				drm_err(dev, "DM_IRQ: Cannot put pageflip irq!\n");
+		}
+
+		drm_crtc_vblank_off(&acrtc->base);
+	}
+}
+
+static void dm_update_pflip_irq_state(struct amdgpu_device *adev,
+				      struct amdgpu_crtc *acrtc)
+{
+	int irq_type =
+		amdgpu_display_crtc_idx_to_irq_type(adev, acrtc->crtc_id);
+
+	/**
+	 * This reads the current state for the IRQ and force reapplies
+	 * the setting to hardware.
+	 */
+	amdgpu_irq_update(adev, &adev->pageflip_irq, irq_type);
+}
+
+static bool
+is_scaling_state_different(const struct dm_connector_state *dm_state,
+			   const struct dm_connector_state *old_dm_state)
+{
+	if (dm_state->scaling != old_dm_state->scaling)
+		return true;
+	if (!dm_state->underscan_enable && old_dm_state->underscan_enable) {
+		if (old_dm_state->underscan_hborder != 0 && old_dm_state->underscan_vborder != 0)
+			return true;
+	} else  if (dm_state->underscan_enable && !old_dm_state->underscan_enable) {
+		if (dm_state->underscan_hborder != 0 && dm_state->underscan_vborder != 0)
+			return true;
+	} else if (dm_state->underscan_hborder != old_dm_state->underscan_hborder ||
+		   dm_state->underscan_vborder != old_dm_state->underscan_vborder)
+		return true;
+	return false;
+}
+
+static bool is_content_protection_different(struct drm_crtc_state *new_crtc_state,
+					    struct drm_crtc_state *old_crtc_state,
+					    struct drm_connector_state *new_conn_state,
+					    struct drm_connector_state *old_conn_state,
+					    const struct drm_connector *connector,
+					    struct hdcp_workqueue *hdcp_w)
+{
+	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector);
+	struct dm_connector_state *dm_con_state = to_dm_connector_state(connector->state);
+
+	pr_debug("[HDCP_DM] connector->index: %x connect_status: %x dpms: %x\n",
+		connector->index, connector->status, connector->dpms);
+	pr_debug("[HDCP_DM] state protection old: %x new: %x\n",
+		old_conn_state->content_protection, new_conn_state->content_protection);
+
+	if (old_crtc_state)
+		pr_debug("[HDCP_DM] old crtc en: %x a: %x m: %x a-chg: %x c-chg: %x\n",
+		old_crtc_state->enable,
+		old_crtc_state->active,
+		old_crtc_state->mode_changed,
+		old_crtc_state->active_changed,
+		old_crtc_state->connectors_changed);
+
+	if (new_crtc_state)
+		pr_debug("[HDCP_DM] NEW crtc en: %x a: %x m: %x a-chg: %x c-chg: %x\n",
+		new_crtc_state->enable,
+		new_crtc_state->active,
+		new_crtc_state->mode_changed,
+		new_crtc_state->active_changed,
+		new_crtc_state->connectors_changed);
+
+	/* hdcp content type change */
+	if (old_conn_state->hdcp_content_type != new_conn_state->hdcp_content_type &&
+	    new_conn_state->content_protection != DRM_MODE_CONTENT_PROTECTION_UNDESIRED) {
+		new_conn_state->content_protection = DRM_MODE_CONTENT_PROTECTION_DESIRED;
+		pr_debug("[HDCP_DM] Type0/1 change %s :true\n", __func__);
+		return true;
+	}
+
+	/* CP is being re enabled, ignore this */
+	if (old_conn_state->content_protection == DRM_MODE_CONTENT_PROTECTION_ENABLED &&
+	    new_conn_state->content_protection == DRM_MODE_CONTENT_PROTECTION_DESIRED) {
+		if (new_crtc_state && new_crtc_state->mode_changed) {
+			new_conn_state->content_protection = DRM_MODE_CONTENT_PROTECTION_DESIRED;
+			pr_debug("[HDCP_DM] ENABLED->DESIRED & mode_changed %s :true\n", __func__);
+			return true;
+		}
+		new_conn_state->content_protection = DRM_MODE_CONTENT_PROTECTION_ENABLED;
+		pr_debug("[HDCP_DM] ENABLED -> DESIRED %s :false\n", __func__);
+		return false;
+	}
+
+	/* S3 resume case, since old state will always be 0 (UNDESIRED) and the restored state will be ENABLED
+	 *
+	 * Handles:	UNDESIRED -> ENABLED
+	 */
+	if (old_conn_state->content_protection == DRM_MODE_CONTENT_PROTECTION_UNDESIRED &&
+	    new_conn_state->content_protection == DRM_MODE_CONTENT_PROTECTION_ENABLED)
+		new_conn_state->content_protection = DRM_MODE_CONTENT_PROTECTION_DESIRED;
+
+	/* Stream removed and re-enabled
+	 *
+	 * Can sometimes overlap with the HPD case,
+	 * thus set update_hdcp to false to avoid
+	 * setting HDCP multiple times.
+	 *
+	 * Handles:	DESIRED -> DESIRED (Special case)
+	 */
+	if (!(old_conn_state->crtc && old_conn_state->crtc->enabled) &&
+		new_conn_state->crtc && new_conn_state->crtc->enabled &&
+		connector->state->content_protection == DRM_MODE_CONTENT_PROTECTION_DESIRED) {
+		dm_con_state->update_hdcp = false;
+		pr_debug("[HDCP_DM] DESIRED->DESIRED (Stream removed and re-enabled) %s :true\n",
+			__func__);
+		return true;
+	}
+
+	/* Hot-plug, headless s3, dpms
+	 *
+	 * Only start HDCP if the display is connected/enabled.
+	 * update_hdcp flag will be set to false until the next
+	 * HPD comes in.
+	 *
+	 * Handles:	DESIRED -> DESIRED (Special case)
+	 */
+	if (dm_con_state->update_hdcp &&
+	new_conn_state->content_protection == DRM_MODE_CONTENT_PROTECTION_DESIRED &&
+	connector->dpms == DRM_MODE_DPMS_ON && aconnector->dc_sink != NULL) {
+		dm_con_state->update_hdcp = false;
+		pr_debug("[HDCP_DM] DESIRED->DESIRED (Hot-plug, headless s3, dpms) %s :true\n",
+			__func__);
+		return true;
+	}
+
+	if (old_conn_state->content_protection == new_conn_state->content_protection) {
+		if (new_conn_state->content_protection >= DRM_MODE_CONTENT_PROTECTION_DESIRED) {
+			if (new_crtc_state && new_crtc_state->mode_changed) {
+				pr_debug("[HDCP_DM] DESIRED->DESIRED or ENABLE->ENABLE mode_change %s :true\n",
+					__func__);
+				return true;
+			}
+			pr_debug("[HDCP_DM] DESIRED->DESIRED & ENABLE->ENABLE %s :false\n",
+				__func__);
+			return false;
+		}
+
+		pr_debug("[HDCP_DM] UNDESIRED->UNDESIRED %s :false\n", __func__);
+		return false;
+	}
+
+	if (new_conn_state->content_protection != DRM_MODE_CONTENT_PROTECTION_ENABLED) {
+		pr_debug("[HDCP_DM] UNDESIRED->DESIRED or DESIRED->UNDESIRED or ENABLED->UNDESIRED %s :true\n",
+			__func__);
+		return true;
+	}
+
+	pr_debug("[HDCP_DM] DESIRED->ENABLED %s :false\n", __func__);
+	return false;
+}
+
+static void remove_stream(struct amdgpu_device *adev,
+			  struct amdgpu_crtc *acrtc,
+			  struct dc_stream_state *stream)
+{
+	/* this is the update mode case */
+
+	acrtc->otg_inst = -1;
+	acrtc->enabled = false;
+}
+
+static void prepare_flip_isr(struct amdgpu_crtc *acrtc)
+{
+
+	assert_spin_locked(&acrtc->base.dev->event_lock);
+	WARN_ON(acrtc->event);
+
+	acrtc->event = acrtc->base.state->event;
+
+	/* Set the flip status */
+	acrtc->pflip_status = AMDGPU_FLIP_SUBMITTED;
+
+	/* Mark this event as consumed */
+	acrtc->base.state->event = NULL;
+
+	drm_dbg_state(acrtc->base.dev,
+		      "crtc:%d, pflip_stat:AMDGPU_FLIP_SUBMITTED\n",
+		      acrtc->crtc_id);
+}
+
+static void update_freesync_state_on_stream(
+	struct amdgpu_display_manager *dm,
+	struct dm_crtc_state *new_crtc_state,
+	struct dc_stream_state *new_stream,
+	struct dc_plane_state *surface,
+	u32 flip_timestamp_in_us)
+{
+	struct mod_vrr_params vrr_params;
+	struct dc_info_packet vrr_infopacket = {0};
+	struct amdgpu_device *adev = dm->adev;
+	struct amdgpu_crtc *acrtc = to_amdgpu_crtc(new_crtc_state->base.crtc);
+	unsigned long flags;
+	bool pack_sdp_v1_3 = false;
+	struct amdgpu_dm_connector *aconn;
+	enum vrr_packet_type packet_type = PACKET_TYPE_VRR;
+
+	if (!new_stream)
+		return;
+
+	/*
+	 * TODO: Determine why min/max totals and vrefresh can be 0 here.
+	 * For now it's sufficient to just guard against these conditions.
+	 */
+
+	if (!new_stream->timing.h_total || !new_stream->timing.v_total)
+		return;
+
+	spin_lock_irqsave(&adev_to_drm(adev)->event_lock, flags);
+	vrr_params = acrtc->dm_irq_params.vrr_params;
+
+	if (surface) {
+		mod_freesync_handle_preflip(
+			dm->freesync_module,
+			surface,
+			new_stream,
+			flip_timestamp_in_us,
+			&vrr_params);
+
+		if (adev->family < AMDGPU_FAMILY_AI &&
+		    amdgpu_dm_crtc_vrr_active(new_crtc_state)) {
+			mod_freesync_handle_v_update(dm->freesync_module,
+						     new_stream, &vrr_params);
+
+			/* Need to call this before the frame ends. */
+			dc_stream_adjust_vmin_vmax(dm->dc,
+						   new_crtc_state->stream,
+						   &vrr_params.adjust);
+		}
+	}
+
+	aconn = (struct amdgpu_dm_connector *)new_stream->dm_stream_context;
+
+	if (aconn && (aconn->as_type == FREESYNC_TYPE_PCON_IN_WHITELIST || aconn->vsdb_info.replay_mode)) {
+		pack_sdp_v1_3 = aconn->pack_sdp_v1_3;
+
+		if (aconn->vsdb_info.amd_vsdb_version == 1)
+			packet_type = PACKET_TYPE_FS_V1;
+		else if (aconn->vsdb_info.amd_vsdb_version == 2)
+			packet_type = PACKET_TYPE_FS_V2;
+		else if (aconn->vsdb_info.amd_vsdb_version == 3)
+			packet_type = PACKET_TYPE_FS_V3;
+
+		mod_build_adaptive_sync_infopacket(new_stream, aconn->as_type, NULL,
+					&new_stream->adaptive_sync_infopacket);
+	}
+
+	mod_freesync_build_vrr_infopacket(
+		dm->freesync_module,
+		new_stream,
+		&vrr_params,
+		packet_type,
+		TRANSFER_FUNC_UNKNOWN,
+		&vrr_infopacket,
+		pack_sdp_v1_3);
+
+	new_crtc_state->freesync_vrr_info_changed |=
+		(memcmp(&new_crtc_state->vrr_infopacket,
+			&vrr_infopacket,
+			sizeof(vrr_infopacket)) != 0);
+
+	acrtc->dm_irq_params.vrr_params = vrr_params;
+	new_crtc_state->vrr_infopacket = vrr_infopacket;
+
+	new_stream->vrr_infopacket = vrr_infopacket;
+	new_stream->allow_freesync = mod_freesync_get_freesync_enabled(&vrr_params);
+
+	if (new_crtc_state->freesync_vrr_info_changed)
+		DRM_DEBUG_KMS("VRR packet update: crtc=%u enabled=%d state=%d",
+			      new_crtc_state->base.crtc->base.id,
+			      (int)new_crtc_state->base.vrr_enabled,
+			      (int)vrr_params.state);
+
+	spin_unlock_irqrestore(&adev_to_drm(adev)->event_lock, flags);
+}
+
+static void update_stream_irq_parameters(
+	struct amdgpu_display_manager *dm,
+	struct dm_crtc_state *new_crtc_state)
+{
+	struct dc_stream_state *new_stream = new_crtc_state->stream;
+	struct mod_vrr_params vrr_params;
+	struct mod_freesync_config config = new_crtc_state->freesync_config;
+	struct amdgpu_device *adev = dm->adev;
+	struct amdgpu_crtc *acrtc = to_amdgpu_crtc(new_crtc_state->base.crtc);
+	unsigned long flags;
+
+	if (!new_stream)
+		return;
+
+	/*
+	 * TODO: Determine why min/max totals and vrefresh can be 0 here.
+	 * For now it's sufficient to just guard against these conditions.
+	 */
+	if (!new_stream->timing.h_total || !new_stream->timing.v_total)
+		return;
+
+	spin_lock_irqsave(&adev_to_drm(adev)->event_lock, flags);
+	vrr_params = acrtc->dm_irq_params.vrr_params;
+
+	if (new_crtc_state->vrr_supported &&
+	    config.min_refresh_in_uhz &&
+	    config.max_refresh_in_uhz) {
+		/*
+		 * if freesync compatible mode was set, config.state will be set
+		 * in atomic check
+		 */
+		if (config.state == VRR_STATE_ACTIVE_FIXED && config.fixed_refresh_in_uhz &&
+		    (!drm_atomic_crtc_needs_modeset(&new_crtc_state->base) ||
+		     new_crtc_state->freesync_config.state == VRR_STATE_ACTIVE_FIXED)) {
+			vrr_params.max_refresh_in_uhz = config.max_refresh_in_uhz;
+			vrr_params.min_refresh_in_uhz = config.min_refresh_in_uhz;
+			vrr_params.fixed_refresh_in_uhz = config.fixed_refresh_in_uhz;
+			vrr_params.state = VRR_STATE_ACTIVE_FIXED;
+		} else {
+			config.state = new_crtc_state->base.vrr_enabled ?
+						     VRR_STATE_ACTIVE_VARIABLE :
+						     VRR_STATE_INACTIVE;
+		}
+	} else {
+		config.state = VRR_STATE_UNSUPPORTED;
+	}
+
+	mod_freesync_build_vrr_params(dm->freesync_module,
+				      new_stream,
+				      &config, &vrr_params);
+
+	new_crtc_state->freesync_config = config;
+	/* Copy state for access from DM IRQ handler */
+	acrtc->dm_irq_params.freesync_config = config;
+	acrtc->dm_irq_params.active_planes = new_crtc_state->active_planes;
+	acrtc->dm_irq_params.vrr_params = vrr_params;
+	spin_unlock_irqrestore(&adev_to_drm(adev)->event_lock, flags);
+}
+
+static void amdgpu_dm_handle_vrr_transition(struct dm_crtc_state *old_state,
+					    struct dm_crtc_state *new_state)
+{
+	bool old_vrr_active = amdgpu_dm_crtc_vrr_active(old_state);
+	bool new_vrr_active = amdgpu_dm_crtc_vrr_active(new_state);
+
+	if (!old_vrr_active && new_vrr_active) {
+		/* Transition VRR inactive -> active:
+		 * While VRR is active, we must not disable vblank irq, as a
+		 * reenable after disable would compute bogus vblank/pflip
+		 * timestamps if it likely happened inside display front-porch.
+		 *
+		 * We also need vupdate irq for the actual core vblank handling
+		 * at end of vblank.
+		 */
+		WARN_ON(amdgpu_dm_crtc_set_vupdate_irq(new_state->base.crtc, true) != 0);
+		WARN_ON(drm_crtc_vblank_get(new_state->base.crtc) != 0);
+		drm_dbg_driver(new_state->base.crtc->dev, "%s: crtc=%u VRR off->on: Get vblank ref\n",
+				 __func__, new_state->base.crtc->base.id);
+	} else if (old_vrr_active && !new_vrr_active) {
+		/* Transition VRR active -> inactive:
+		 * Allow vblank irq disable again for fixed refresh rate.
+		 */
+		WARN_ON(amdgpu_dm_crtc_set_vupdate_irq(new_state->base.crtc, false) != 0);
+		drm_crtc_vblank_put(new_state->base.crtc);
+		drm_dbg_driver(new_state->base.crtc->dev, "%s: crtc=%u VRR on->off: Drop vblank ref\n",
+				 __func__, new_state->base.crtc->base.id);
+	}
+}
+
+static void amdgpu_dm_commit_cursors(struct drm_atomic_state *state)
+{
+	struct drm_plane *plane;
+	struct drm_plane_state *old_plane_state;
+	int i;
+
+	/*
+	 * TODO: Make this per-stream so we don't issue redundant updates for
+	 * commits with multiple streams.
+	 */
+	for_each_old_plane_in_state(state, plane, old_plane_state, i)
+		if (plane->type == DRM_PLANE_TYPE_CURSOR)
+			amdgpu_dm_plane_handle_cursor_update(plane, old_plane_state);
+}
+
+static inline uint32_t get_mem_type(struct drm_framebuffer *fb)
+{
+	struct amdgpu_bo *abo = gem_to_amdgpu_bo(fb->obj[0]);
+
+	return abo->tbo.resource ? abo->tbo.resource->mem_type : 0;
+}
+
+static void amdgpu_dm_update_cursor(struct drm_plane *plane,
+				    struct drm_plane_state *old_plane_state,
+				    struct dc_stream_update *update)
+{
+	struct amdgpu_device *adev = drm_to_adev(plane->dev);
+	struct amdgpu_framebuffer *afb = to_amdgpu_framebuffer(plane->state->fb);
+	struct drm_crtc *crtc = afb ? plane->state->crtc : old_plane_state->crtc;
+	struct dm_crtc_state *crtc_state = crtc ? to_dm_crtc_state(crtc->state) : NULL;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	uint64_t address = afb ? afb->address : 0;
+	struct dc_cursor_position position = {0};
+	struct dc_cursor_attributes attributes;
+	int ret;
+
+	if (!plane->state->fb && !old_plane_state->fb)
+		return;
+
+	drm_dbg_atomic(plane->dev, "crtc_id=%d with size %d to %d\n",
+		       amdgpu_crtc->crtc_id, plane->state->crtc_w,
+		       plane->state->crtc_h);
+
+	ret = amdgpu_dm_plane_get_cursor_position(plane, crtc, &position);
+	if (ret)
+		return;
+
+	if (!position.enable) {
+		/* turn off cursor */
+		if (crtc_state && crtc_state->stream) {
+			dc_stream_set_cursor_position(crtc_state->stream,
+						      &position);
+			update->cursor_position = &crtc_state->stream->cursor_position;
+		}
+		return;
+	}
+
+	amdgpu_crtc->cursor_width = plane->state->crtc_w;
+	amdgpu_crtc->cursor_height = plane->state->crtc_h;
+
+	memset(&attributes, 0, sizeof(attributes));
+	attributes.address.high_part = upper_32_bits(address);
+	attributes.address.low_part  = lower_32_bits(address);
+	attributes.width             = plane->state->crtc_w;
+	attributes.height            = plane->state->crtc_h;
+	attributes.color_format      = CURSOR_MODE_COLOR_PRE_MULTIPLIED_ALPHA;
+	attributes.rotation_angle    = 0;
+	attributes.attribute_flags.value = 0;
+
+	/* Enable cursor degamma ROM on DCN3+ for implicit sRGB degamma in DRM
+	 * legacy gamma setup.
+	 */
+	if (crtc_state->cm_is_degamma_srgb &&
+	    adev->dm.dc->caps.color.dpp.gamma_corr)
+		attributes.attribute_flags.bits.ENABLE_CURSOR_DEGAMMA = 1;
+
+	if (afb)
+		attributes.pitch = afb->base.pitches[0] / afb->base.format->cpp[0];
+
+	if (crtc_state->stream) {
+		if (!dc_stream_set_cursor_attributes(crtc_state->stream,
+						     &attributes))
+			drm_err(adev_to_drm(adev), "DC failed to set cursor attributes\n");
+
+		update->cursor_attributes = &crtc_state->stream->cursor_attributes;
+
+		if (!dc_stream_set_cursor_position(crtc_state->stream,
+						   &position))
+			drm_err(adev_to_drm(adev), "DC failed to set cursor position\n");
+
+		update->cursor_position = &crtc_state->stream->cursor_position;
+	}
+}
+
+static void amdgpu_dm_enable_self_refresh(struct amdgpu_crtc *acrtc_attach,
+					  const struct dm_crtc_state *acrtc_state,
+					  const u64 current_ts)
+{
+	struct psr_settings *psr = &acrtc_state->stream->link->psr_settings;
+	struct replay_settings *pr = &acrtc_state->stream->link->replay_settings;
+	struct amdgpu_dm_connector *aconn =
+		(struct amdgpu_dm_connector *)acrtc_state->stream->dm_stream_context;
+	bool vrr_active = amdgpu_dm_crtc_vrr_active(acrtc_state);
+
+	if (acrtc_state->update_type > UPDATE_TYPE_FAST) {
+		if (pr->config.replay_supported && !pr->replay_feature_enabled)
+			amdgpu_dm_link_setup_replay(acrtc_state->stream->link, aconn);
+		else if (psr->psr_version != DC_PSR_VERSION_UNSUPPORTED &&
+			     !psr->psr_feature_enabled)
+			if (!aconn->disallow_edp_enter_psr)
+				amdgpu_dm_link_setup_psr(acrtc_state->stream);
+	}
+
+	/* Decrement skip count when SR is enabled and we're doing fast updates. */
+	if (acrtc_state->update_type == UPDATE_TYPE_FAST &&
+	    (psr->psr_feature_enabled || pr->config.replay_supported)) {
+		if (aconn->sr_skip_count > 0)
+			aconn->sr_skip_count--;
+
+		/* Allow SR when skip count is 0. */
+		acrtc_attach->dm_irq_params.allow_sr_entry = !aconn->sr_skip_count;
+
+		/*
+		 * If sink supports PSR SU/Panel Replay, there is no need to rely on
+		 * a vblank event disable request to enable PSR/RP. PSR SU/RP
+		 * can be enabled immediately once OS demonstrates an
+		 * adequate number of fast atomic commits to notify KMD
+		 * of update events. See `vblank_control_worker()`.
+		 */
+		if (!vrr_active &&
+		    acrtc_attach->dm_irq_params.allow_sr_entry &&
+#ifdef CONFIG_DRM_AMD_SECURE_DISPLAY
+		    !amdgpu_dm_crc_window_is_activated(acrtc_state->base.crtc) &&
+#endif
+		    (current_ts - psr->psr_dirty_rects_change_timestamp_ns) > 500000000) {
+			if (pr->replay_feature_enabled && !pr->replay_allow_active)
+				amdgpu_dm_replay_enable(acrtc_state->stream, true);
+			if (psr->psr_version == DC_PSR_VERSION_SU_1 &&
+			    !psr->psr_allow_active && !aconn->disallow_edp_enter_psr)
+				amdgpu_dm_psr_enable(acrtc_state->stream);
+		}
+	} else {
+		acrtc_attach->dm_irq_params.allow_sr_entry = false;
+	}
+}
+
+static void amdgpu_dm_commit_planes(struct drm_atomic_state *state,
+				    struct drm_device *dev,
+				    struct amdgpu_display_manager *dm,
+				    struct drm_crtc *pcrtc,
+				    bool wait_for_vblank)
+{
+	u32 i;
+	u64 timestamp_ns = ktime_get_ns();
+	struct drm_plane *plane;
+	struct drm_plane_state *old_plane_state, *new_plane_state;
+	struct amdgpu_crtc *acrtc_attach = to_amdgpu_crtc(pcrtc);
+	struct drm_crtc_state *new_pcrtc_state =
+			drm_atomic_get_new_crtc_state(state, pcrtc);
+	struct dm_crtc_state *acrtc_state = to_dm_crtc_state(new_pcrtc_state);
+	struct dm_crtc_state *dm_old_crtc_state =
+			to_dm_crtc_state(drm_atomic_get_old_crtc_state(state, pcrtc));
+	int planes_count = 0, vpos, hpos;
+	unsigned long flags;
+	u32 target_vblank, last_flip_vblank;
+	bool vrr_active = amdgpu_dm_crtc_vrr_active(acrtc_state);
+	bool cursor_update = false;
+	bool pflip_present = false;
+	bool dirty_rects_changed = false;
+	bool updated_planes_and_streams = false;
+	struct {
+		struct dc_surface_update surface_updates[MAX_SURFACES];
+		struct dc_plane_info plane_infos[MAX_SURFACES];
+		struct dc_scaling_info scaling_infos[MAX_SURFACES];
+		struct dc_flip_addrs flip_addrs[MAX_SURFACES];
+		struct dc_stream_update stream_update;
+	} *bundle;
+
+	bundle = kzalloc(sizeof(*bundle), GFP_KERNEL);
+
+	if (!bundle) {
+		drm_err(dev, "Failed to allocate update bundle\n");
+		goto cleanup;
+	}
+
+	/*
+	 * Disable the cursor first if we're disabling all the planes.
+	 * It'll remain on the screen after the planes are re-enabled
+	 * if we don't.
+	 *
+	 * If the cursor is transitioning from native to overlay mode, the
+	 * native cursor needs to be disabled first.
+	 */
+	if (acrtc_state->cursor_mode == DM_CURSOR_OVERLAY_MODE &&
+	    dm_old_crtc_state->cursor_mode == DM_CURSOR_NATIVE_MODE) {
+		struct dc_cursor_position cursor_position = {0};
+
+		if (!dc_stream_set_cursor_position(acrtc_state->stream,
+						   &cursor_position))
+			drm_err(dev, "DC failed to disable native cursor\n");
+
+		bundle->stream_update.cursor_position =
+				&acrtc_state->stream->cursor_position;
+	}
+
+	if (acrtc_state->active_planes == 0 &&
+	    dm_old_crtc_state->cursor_mode == DM_CURSOR_NATIVE_MODE)
+		amdgpu_dm_commit_cursors(state);
+
+	/* update planes when needed */
+	for_each_oldnew_plane_in_state(state, plane, old_plane_state, new_plane_state, i) {
+		struct drm_crtc *crtc = new_plane_state->crtc;
+		struct drm_crtc_state *new_crtc_state;
+		struct drm_framebuffer *fb = new_plane_state->fb;
+		struct amdgpu_framebuffer *afb = (struct amdgpu_framebuffer *)fb;
+		bool plane_needs_flip;
+		struct dc_plane_state *dc_plane;
+		struct dm_plane_state *dm_new_plane_state = to_dm_plane_state(new_plane_state);
+
+		/* Cursor plane is handled after stream updates */
+		if (plane->type == DRM_PLANE_TYPE_CURSOR &&
+		    acrtc_state->cursor_mode == DM_CURSOR_NATIVE_MODE) {
+			if ((fb && crtc == pcrtc) ||
+			    (old_plane_state->fb && old_plane_state->crtc == pcrtc)) {
+				cursor_update = true;
+				if (amdgpu_ip_version(dm->adev, DCE_HWIP, 0) != 0)
+					amdgpu_dm_update_cursor(plane, old_plane_state, &bundle->stream_update);
+			}
+
+			continue;
+		}
+
+		if (!fb || !crtc || pcrtc != crtc)
+			continue;
+
+		new_crtc_state = drm_atomic_get_new_crtc_state(state, crtc);
+		if (!new_crtc_state->active)
+			continue;
+
+		dc_plane = dm_new_plane_state->dc_state;
+		if (!dc_plane)
+			continue;
+
+		bundle->surface_updates[planes_count].surface = dc_plane;
+		if (new_pcrtc_state->color_mgmt_changed) {
+			bundle->surface_updates[planes_count].gamma = &dc_plane->gamma_correction;
+			bundle->surface_updates[planes_count].in_transfer_func = &dc_plane->in_transfer_func;
+			bundle->surface_updates[planes_count].gamut_remap_matrix = &dc_plane->gamut_remap_matrix;
+			bundle->surface_updates[planes_count].hdr_mult = dc_plane->hdr_mult;
+			bundle->surface_updates[planes_count].func_shaper = &dc_plane->in_shaper_func;
+			bundle->surface_updates[planes_count].lut3d_func = &dc_plane->lut3d_func;
+			bundle->surface_updates[planes_count].blend_tf = &dc_plane->blend_tf;
+		}
+
+		amdgpu_dm_plane_fill_dc_scaling_info(dm->adev, new_plane_state,
+				     &bundle->scaling_infos[planes_count]);
+
+		bundle->surface_updates[planes_count].scaling_info =
+			&bundle->scaling_infos[planes_count];
+
+		plane_needs_flip = old_plane_state->fb && new_plane_state->fb;
+
+		pflip_present = pflip_present || plane_needs_flip;
+
+		if (!plane_needs_flip) {
+			planes_count += 1;
+			continue;
+		}
+
+		fill_dc_plane_info_and_addr(
+			dm->adev, new_plane_state,
+			afb->tiling_flags,
+			&bundle->plane_infos[planes_count],
+			&bundle->flip_addrs[planes_count].address,
+			afb->tmz_surface);
+
+		drm_dbg_state(state->dev, "plane: id=%d dcc_en=%d\n",
+				 new_plane_state->plane->index,
+				 bundle->plane_infos[planes_count].dcc.enable);
+
+		bundle->surface_updates[planes_count].plane_info =
+			&bundle->plane_infos[planes_count];
+
+		if (acrtc_state->stream->link->psr_settings.psr_feature_enabled ||
+		    acrtc_state->stream->link->replay_settings.replay_feature_enabled) {
+			fill_dc_dirty_rects(plane, old_plane_state,
+					    new_plane_state, new_crtc_state,
+					    &bundle->flip_addrs[planes_count],
+					    acrtc_state->stream->link->psr_settings.psr_version ==
+					    DC_PSR_VERSION_SU_1,
+					    &dirty_rects_changed);
+
+			/*
+			 * If the dirty regions changed, PSR-SU need to be disabled temporarily
+			 * and enabled it again after dirty regions are stable to avoid video glitch.
+			 * PSR-SU will be enabled in vblank_control_worker() if user pause the video
+			 * during the PSR-SU was disabled.
+			 */
+			if (acrtc_state->stream->link->psr_settings.psr_version >= DC_PSR_VERSION_SU_1 &&
+			    acrtc_attach->dm_irq_params.allow_sr_entry &&
+#ifdef CONFIG_DRM_AMD_SECURE_DISPLAY
+			    !amdgpu_dm_crc_window_is_activated(acrtc_state->base.crtc) &&
+#endif
+			    dirty_rects_changed) {
+				mutex_lock(&dm->dc_lock);
+				acrtc_state->stream->link->psr_settings.psr_dirty_rects_change_timestamp_ns =
+				timestamp_ns;
+				if (acrtc_state->stream->link->psr_settings.psr_allow_active)
+					amdgpu_dm_psr_disable(acrtc_state->stream, true);
+				mutex_unlock(&dm->dc_lock);
+			}
+		}
+
+		/*
+		 * Only allow immediate flips for fast updates that don't
+		 * change memory domain, FB pitch, DCC state, rotation or
+		 * mirroring.
+		 *
+		 * dm_crtc_helper_atomic_check() only accepts async flips with
+		 * fast updates.
+		 */
+		if (crtc->state->async_flip &&
+		    (acrtc_state->update_type != UPDATE_TYPE_FAST ||
+		     get_mem_type(old_plane_state->fb) != get_mem_type(fb)))
+			drm_warn_once(state->dev,
+				      "[PLANE:%d:%s] async flip with non-fast update\n",
+				      plane->base.id, plane->name);
+
+		bundle->flip_addrs[planes_count].flip_immediate =
+			crtc->state->async_flip &&
+			acrtc_state->update_type == UPDATE_TYPE_FAST &&
+			get_mem_type(old_plane_state->fb) == get_mem_type(fb);
+
+		timestamp_ns = ktime_get_ns();
+		bundle->flip_addrs[planes_count].flip_timestamp_in_us = div_u64(timestamp_ns, 1000);
+		bundle->surface_updates[planes_count].flip_addr = &bundle->flip_addrs[planes_count];
+		bundle->surface_updates[planes_count].surface = dc_plane;
+
+		if (!bundle->surface_updates[planes_count].surface) {
+			drm_err(dev, "No surface for CRTC: id=%d\n",
+					acrtc_attach->crtc_id);
+			continue;
+		}
+
+		if (plane == pcrtc->primary)
+			update_freesync_state_on_stream(
+				dm,
+				acrtc_state,
+				acrtc_state->stream,
+				dc_plane,
+				bundle->flip_addrs[planes_count].flip_timestamp_in_us);
+
+		drm_dbg_state(state->dev, "%s Flipping to hi: 0x%x, low: 0x%x\n",
+				 __func__,
+				 bundle->flip_addrs[planes_count].address.grph.addr.high_part,
+				 bundle->flip_addrs[planes_count].address.grph.addr.low_part);
+
+		planes_count += 1;
+
+	}
+
+	if (pflip_present) {
+		if (!vrr_active) {
+			/* Use old throttling in non-vrr fixed refresh rate mode
+			 * to keep flip scheduling based on target vblank counts
+			 * working in a backwards compatible way, e.g., for
+			 * clients using the GLX_OML_sync_control extension or
+			 * DRI3/Present extension with defined target_msc.
+			 */
+			last_flip_vblank = amdgpu_get_vblank_counter_kms(pcrtc);
+		} else {
+			/* For variable refresh rate mode only:
+			 * Get vblank of last completed flip to avoid > 1 vrr
+			 * flips per video frame by use of throttling, but allow
+			 * flip programming anywhere in the possibly large
+			 * variable vrr vblank interval for fine-grained flip
+			 * timing control and more opportunity to avoid stutter
+			 * on late submission of flips.
+			 */
+			spin_lock_irqsave(&pcrtc->dev->event_lock, flags);
+			last_flip_vblank = acrtc_attach->dm_irq_params.last_flip_vblank;
+			spin_unlock_irqrestore(&pcrtc->dev->event_lock, flags);
+		}
+
+		target_vblank = last_flip_vblank + wait_for_vblank;
+
+		/*
+		 * Wait until we're out of the vertical blank period before the one
+		 * targeted by the flip
+		 */
+		while ((acrtc_attach->enabled &&
+			(amdgpu_display_get_crtc_scanoutpos(dm->ddev, acrtc_attach->crtc_id,
+							    0, &vpos, &hpos, NULL,
+							    NULL, &pcrtc->hwmode)
+			 & (DRM_SCANOUTPOS_VALID | DRM_SCANOUTPOS_IN_VBLANK)) ==
+			(DRM_SCANOUTPOS_VALID | DRM_SCANOUTPOS_IN_VBLANK) &&
+			(int)(target_vblank -
+			  amdgpu_get_vblank_counter_kms(pcrtc)) > 0)) {
+			usleep_range(1000, 1100);
+		}
+
+		/**
+		 * Prepare the flip event for the pageflip interrupt to handle.
+		 *
+		 * This only works in the case where we've already turned on the
+		 * appropriate hardware blocks (eg. HUBP) so in the transition case
+		 * from 0 -> n planes we have to skip a hardware generated event
+		 * and rely on sending it from software.
+		 */
+		if (acrtc_attach->base.state->event &&
+		    acrtc_state->active_planes > 0) {
+			drm_crtc_vblank_get(pcrtc);
+
+			spin_lock_irqsave(&pcrtc->dev->event_lock, flags);
+
+			WARN_ON(acrtc_attach->pflip_status != AMDGPU_FLIP_NONE);
+			prepare_flip_isr(acrtc_attach);
+
+			spin_unlock_irqrestore(&pcrtc->dev->event_lock, flags);
+		}
+
+		if (acrtc_state->stream) {
+			if (acrtc_state->freesync_vrr_info_changed)
+				bundle->stream_update.vrr_infopacket =
+					&acrtc_state->stream->vrr_infopacket;
+		}
+	} else if (cursor_update && acrtc_state->active_planes > 0) {
+		spin_lock_irqsave(&pcrtc->dev->event_lock, flags);
+		if (acrtc_attach->base.state->event) {
+			drm_crtc_vblank_get(pcrtc);
+			acrtc_attach->event = acrtc_attach->base.state->event;
+			acrtc_attach->base.state->event = NULL;
+		}
+		spin_unlock_irqrestore(&pcrtc->dev->event_lock, flags);
+	}
+
+	/* Update the planes if changed or disable if we don't have any. */
+	if ((planes_count || acrtc_state->active_planes == 0) &&
+		acrtc_state->stream) {
+		/*
+		 * If PSR or idle optimizations are enabled then flush out
+		 * any pending work before hardware programming.
+		 */
+		if (dm->vblank_control_workqueue)
+			flush_workqueue(dm->vblank_control_workqueue);
+
+		bundle->stream_update.stream = acrtc_state->stream;
+		if (new_pcrtc_state->mode_changed) {
+			bundle->stream_update.src = acrtc_state->stream->src;
+			bundle->stream_update.dst = acrtc_state->stream->dst;
+		}
+
+		if (new_pcrtc_state->color_mgmt_changed) {
+			/*
+			 * TODO: This isn't fully correct since we've actually
+			 * already modified the stream in place.
+			 */
+			bundle->stream_update.gamut_remap =
+				&acrtc_state->stream->gamut_remap_matrix;
+			bundle->stream_update.output_csc_transform =
+				&acrtc_state->stream->csc_color_matrix;
+			bundle->stream_update.out_transfer_func =
+				&acrtc_state->stream->out_transfer_func;
+			bundle->stream_update.lut3d_func =
+				(struct dc_3dlut *) acrtc_state->stream->lut3d_func;
+			bundle->stream_update.func_shaper =
+				(struct dc_transfer_func *) acrtc_state->stream->func_shaper;
+		}
+
+		acrtc_state->stream->abm_level = acrtc_state->abm_level;
+		if (acrtc_state->abm_level != dm_old_crtc_state->abm_level)
+			bundle->stream_update.abm_level = &acrtc_state->abm_level;
+
+		mutex_lock(&dm->dc_lock);
+		if ((acrtc_state->update_type > UPDATE_TYPE_FAST) || vrr_active) {
+			if (acrtc_state->stream->link->replay_settings.replay_allow_active)
+				amdgpu_dm_replay_disable(acrtc_state->stream);
+			if (acrtc_state->stream->link->psr_settings.psr_allow_active)
+				amdgpu_dm_psr_disable(acrtc_state->stream, true);
+		}
+		mutex_unlock(&dm->dc_lock);
+
+		/*
+		 * If FreeSync state on the stream has changed then we need to
+		 * re-adjust the min/max bounds now that DC doesn't handle this
+		 * as part of commit.
+		 */
+		if (is_dc_timing_adjust_needed(dm_old_crtc_state, acrtc_state)) {
+			spin_lock_irqsave(&pcrtc->dev->event_lock, flags);
+			dc_stream_adjust_vmin_vmax(
+				dm->dc, acrtc_state->stream,
+				&acrtc_attach->dm_irq_params.vrr_params.adjust);
+			spin_unlock_irqrestore(&pcrtc->dev->event_lock, flags);
+		}
+		mutex_lock(&dm->dc_lock);
+		update_planes_and_stream_adapter(dm->dc,
+					 acrtc_state->update_type,
+					 planes_count,
+					 acrtc_state->stream,
+					 &bundle->stream_update,
+					 bundle->surface_updates);
+		updated_planes_and_streams = true;
+
+		/**
+		 * Enable or disable the interrupts on the backend.
+		 *
+		 * Most pipes are put into power gating when unused.
+		 *
+		 * When power gating is enabled on a pipe we lose the
+		 * interrupt enablement state when power gating is disabled.
+		 *
+		 * So we need to update the IRQ control state in hardware
+		 * whenever the pipe turns on (since it could be previously
+		 * power gated) or off (since some pipes can't be power gated
+		 * on some ASICs).
+		 */
+		if (dm_old_crtc_state->active_planes != acrtc_state->active_planes)
+			dm_update_pflip_irq_state(drm_to_adev(dev),
+						  acrtc_attach);
+
+		amdgpu_dm_enable_self_refresh(acrtc_attach, acrtc_state, timestamp_ns);
+		mutex_unlock(&dm->dc_lock);
+	}
+
+	/*
+	 * Update cursor state *after* programming all the planes.
+	 * This avoids redundant programming in the case where we're going
+	 * to be disabling a single plane - those pipes are being disabled.
+	 */
+	if (acrtc_state->active_planes &&
+	    (!updated_planes_and_streams || amdgpu_ip_version(dm->adev, DCE_HWIP, 0) == 0) &&
+	    acrtc_state->cursor_mode == DM_CURSOR_NATIVE_MODE)
+		amdgpu_dm_commit_cursors(state);
+
+cleanup:
+	kfree(bundle);
+}
+
+static void amdgpu_dm_commit_audio(struct drm_device *dev,
+				   struct drm_atomic_state *state)
+{
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	struct amdgpu_dm_connector *aconnector;
+	struct drm_connector *connector;
+	struct drm_connector_state *old_con_state, *new_con_state;
+	struct drm_crtc_state *new_crtc_state;
+	struct dm_crtc_state *new_dm_crtc_state;
+	const struct dc_stream_status *status;
+	int i, inst;
+
+	/* Notify device removals. */
+	for_each_oldnew_connector_in_state(state, connector, old_con_state, new_con_state, i) {
+		if (old_con_state->crtc != new_con_state->crtc) {
+			/* CRTC changes require notification. */
+			goto notify;
+		}
+
+		if (!new_con_state->crtc)
+			continue;
+
+		new_crtc_state = drm_atomic_get_new_crtc_state(
+			state, new_con_state->crtc);
+
+		if (!new_crtc_state)
+			continue;
+
+		if (!drm_atomic_crtc_needs_modeset(new_crtc_state))
+			continue;
+
+notify:
+		if (connector->connector_type == DRM_MODE_CONNECTOR_WRITEBACK)
+			continue;
+
+		aconnector = to_amdgpu_dm_connector(connector);
+
+		mutex_lock(&adev->dm.audio_lock);
+		inst = aconnector->audio_inst;
+		aconnector->audio_inst = -1;
+		mutex_unlock(&adev->dm.audio_lock);
+
+		amdgpu_dm_audio_eld_notify(adev, inst);
+	}
+
+	/* Notify audio device additions. */
+	for_each_new_connector_in_state(state, connector, new_con_state, i) {
+		if (!new_con_state->crtc)
+			continue;
+
+		new_crtc_state = drm_atomic_get_new_crtc_state(
+			state, new_con_state->crtc);
+
+		if (!new_crtc_state)
+			continue;
+
+		if (!drm_atomic_crtc_needs_modeset(new_crtc_state))
+			continue;
+
+		new_dm_crtc_state = to_dm_crtc_state(new_crtc_state);
+		if (!new_dm_crtc_state->stream)
+			continue;
+
+		status = dc_stream_get_status(new_dm_crtc_state->stream);
+		if (!status)
+			continue;
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_WRITEBACK)
+			continue;
+
+		aconnector = to_amdgpu_dm_connector(connector);
+
+		mutex_lock(&adev->dm.audio_lock);
+		inst = status->audio_inst;
+		aconnector->audio_inst = inst;
+		mutex_unlock(&adev->dm.audio_lock);
+
+		amdgpu_dm_audio_eld_notify(adev, inst);
+	}
+}
+
+/*
+ * amdgpu_dm_crtc_copy_transient_flags - copy mirrored flags from DRM to DC
+ * @crtc_state: the DRM CRTC state
+ * @stream_state: the DC stream state.
+ *
+ * Copy the mirrored transient state flags from DRM, to DC. It is used to bring
+ * a dc_stream_state's flags in sync with a drm_crtc_state's flags.
+ */
+static void amdgpu_dm_crtc_copy_transient_flags(struct drm_crtc_state *crtc_state,
+						struct dc_stream_state *stream_state)
+{
+	stream_state->mode_changed = drm_atomic_crtc_needs_modeset(crtc_state);
+}
+
+static void dm_clear_writeback(struct amdgpu_display_manager *dm,
+			      struct dm_crtc_state *crtc_state)
+{
+	dc_stream_remove_writeback(dm->dc, crtc_state->stream, 0);
+}
+
+static void amdgpu_dm_commit_streams(struct drm_atomic_state *state,
+					struct dc_state *dc_state)
+{
+	struct drm_device *dev = state->dev;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	struct amdgpu_display_manager *dm = &adev->dm;
+	struct drm_crtc *crtc;
+	struct drm_crtc_state *old_crtc_state, *new_crtc_state;
+	struct dm_crtc_state *dm_old_crtc_state, *dm_new_crtc_state;
+	struct drm_connector_state *old_con_state;
+	struct drm_connector *connector;
+	bool mode_set_reset_required = false;
+	u32 i;
+	struct dc_commit_streams_params params = {dc_state->streams, dc_state->stream_count};
+
+	/* Disable writeback */
+	for_each_old_connector_in_state(state, connector, old_con_state, i) {
+		struct dm_connector_state *dm_old_con_state;
+		struct amdgpu_crtc *acrtc;
+
+		if (connector->connector_type != DRM_MODE_CONNECTOR_WRITEBACK)
+			continue;
+
+		old_crtc_state = NULL;
+
+		dm_old_con_state = to_dm_connector_state(old_con_state);
+		if (!dm_old_con_state->base.crtc)
+			continue;
+
+		acrtc = to_amdgpu_crtc(dm_old_con_state->base.crtc);
+		if (acrtc)
+			old_crtc_state = drm_atomic_get_old_crtc_state(state, &acrtc->base);
+
+		if (!acrtc || !acrtc->wb_enabled)
+			continue;
+
+		dm_old_crtc_state = to_dm_crtc_state(old_crtc_state);
+
+		dm_clear_writeback(dm, dm_old_crtc_state);
+		acrtc->wb_enabled = false;
+	}
+
+	for_each_oldnew_crtc_in_state(state, crtc, old_crtc_state,
+				      new_crtc_state, i) {
+		struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
+
+		dm_old_crtc_state = to_dm_crtc_state(old_crtc_state);
+
+		if (old_crtc_state->active &&
+		    (!new_crtc_state->active ||
+		     drm_atomic_crtc_needs_modeset(new_crtc_state))) {
+			manage_dm_interrupts(adev, acrtc, NULL);
+			dc_stream_release(dm_old_crtc_state->stream);
+		}
+	}
+
+	drm_atomic_helper_calc_timestamping_constants(state);
+
+	/* update changed items */
+	for_each_oldnew_crtc_in_state(state, crtc, old_crtc_state, new_crtc_state, i) {
+		struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
+
+		dm_new_crtc_state = to_dm_crtc_state(new_crtc_state);
+		dm_old_crtc_state = to_dm_crtc_state(old_crtc_state);
+
+		drm_dbg_state(state->dev,
+			"amdgpu_crtc id:%d crtc_state_flags: enable:%d, active:%d, planes_changed:%d, mode_changed:%d,active_changed:%d,connectors_changed:%d\n",
+			acrtc->crtc_id,
+			new_crtc_state->enable,
+			new_crtc_state->active,
+			new_crtc_state->planes_changed,
+			new_crtc_state->mode_changed,
+			new_crtc_state->active_changed,
+			new_crtc_state->connectors_changed);
+
+		/* Disable cursor if disabling crtc */
+		if (old_crtc_state->active && !new_crtc_state->active) {
+			struct dc_cursor_position position;
+
+			memset(&position, 0, sizeof(position));
+			mutex_lock(&dm->dc_lock);
+			dc_exit_ips_for_hw_access(dm->dc);
+			dc_stream_program_cursor_position(dm_old_crtc_state->stream, &position);
+			mutex_unlock(&dm->dc_lock);
+		}
+
+		/* Copy all transient state flags into dc state */
+		if (dm_new_crtc_state->stream) {
+			amdgpu_dm_crtc_copy_transient_flags(&dm_new_crtc_state->base,
+							    dm_new_crtc_state->stream);
+		}
+
+		/* handles headless hotplug case, updating new_state and
+		 * aconnector as needed
+		 */
+
+		if (amdgpu_dm_crtc_modeset_required(new_crtc_state, dm_new_crtc_state->stream, dm_old_crtc_state->stream)) {
+
+			drm_dbg_atomic(dev,
+				       "Atomic commit: SET crtc id %d: [%p]\n",
+				       acrtc->crtc_id, acrtc);
+
+			if (!dm_new_crtc_state->stream) {
+				/*
+				 * this could happen because of issues with
+				 * userspace notifications delivery.
+				 * In this case userspace tries to set mode on
+				 * display which is disconnected in fact.
+				 * dc_sink is NULL in this case on aconnector.
+				 * We expect reset mode will come soon.
+				 *
+				 * This can also happen when unplug is done
+				 * during resume sequence ended
+				 *
+				 * In this case, we want to pretend we still
+				 * have a sink to keep the pipe running so that
+				 * hw state is consistent with the sw state
+				 */
+				drm_dbg_atomic(dev,
+					       "Failed to create new stream for crtc %d\n",
+						acrtc->base.base.id);
+				continue;
+			}
+
+			if (dm_old_crtc_state->stream)
+				remove_stream(adev, acrtc, dm_old_crtc_state->stream);
+
+			pm_runtime_get_noresume(dev->dev);
+
+			acrtc->enabled = true;
+			acrtc->hw_mode = new_crtc_state->mode;
+			crtc->hwmode = new_crtc_state->mode;
+			mode_set_reset_required = true;
+		} else if (modereset_required(new_crtc_state)) {
+			drm_dbg_atomic(dev,
+				       "Atomic commit: RESET. crtc id %d:[%p]\n",
+				       acrtc->crtc_id, acrtc);
+			/* i.e. reset mode */
+			if (dm_old_crtc_state->stream)
+				remove_stream(adev, acrtc, dm_old_crtc_state->stream);
+
+			mode_set_reset_required = true;
+		}
+	} /* for_each_crtc_in_state() */
+
+	/* if there mode set or reset, disable eDP PSR, Replay */
+	if (mode_set_reset_required) {
+		if (dm->vblank_control_workqueue)
+			flush_workqueue(dm->vblank_control_workqueue);
+
+		amdgpu_dm_replay_disable_all(dm);
+		amdgpu_dm_psr_disable_all(dm);
+	}
+
+	dm_enable_per_frame_crtc_master_sync(dc_state);
+	mutex_lock(&dm->dc_lock);
+	dc_exit_ips_for_hw_access(dm->dc);
+	WARN_ON(!dc_commit_streams(dm->dc, &params));
+
+	/* Allow idle optimization when vblank count is 0 for display off */
+	if ((dm->active_vblank_irq_count == 0) && amdgpu_dm_is_headless(dm->adev))
+		dc_allow_idle_optimizations(dm->dc, true);
+	mutex_unlock(&dm->dc_lock);
+
+	for_each_new_crtc_in_state(state, crtc, new_crtc_state, i) {
+		struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
+
+		dm_new_crtc_state = to_dm_crtc_state(new_crtc_state);
+
+		if (dm_new_crtc_state->stream != NULL) {
+			const struct dc_stream_status *status =
+					dc_stream_get_status(dm_new_crtc_state->stream);
+
+			if (!status)
+				status = dc_state_get_stream_status(dc_state,
+									 dm_new_crtc_state->stream);
+			if (!status)
+				drm_err(dev,
+					"got no status for stream %p on acrtc%p\n",
+					dm_new_crtc_state->stream, acrtc);
+			else
+				acrtc->otg_inst = status->primary_otg_inst;
+		}
+	}
+
+	/* During boot up and resume the DC layer will reset the panel brightness
+	 * to fix a flicker issue.
+	 * It will cause the dm->actual_brightness is not the current panel brightness
+	 * level. (the dm->brightness is the correct panel level)
+	 * So we set the backlight level with dm->brightness value after initial
+	 * set mode. Use restore_backlight flag to avoid setting backlight level
+	 * for every subsequent mode set.
+	 */
+	if (dm->restore_backlight) {
+		for (i = 0; i < dm->num_of_edps; i++) {
+			if (dm->backlight_dev[i])
+				amdgpu_dm_backlight_set_level(dm, i, dm->brightness[i]);
+		}
+		dm->restore_backlight = false;
+	}
+}
+
+static void dm_set_writeback(struct amdgpu_display_manager *dm,
+			      struct dm_crtc_state *crtc_state,
+			      struct drm_connector *connector,
+			      struct drm_connector_state *new_con_state)
+{
+	struct drm_writeback_connector *wb_conn = drm_connector_to_writeback(connector);
+	struct amdgpu_device *adev = dm->adev;
+	struct amdgpu_crtc *acrtc;
+	struct dc_writeback_info *wb_info;
+	struct pipe_ctx *pipe = NULL;
+	struct amdgpu_framebuffer *afb;
+	int i = 0;
+
+	wb_info = kzalloc(sizeof(*wb_info), GFP_KERNEL);
+	if (!wb_info) {
+		drm_err(adev_to_drm(adev), "Failed to allocate wb_info\n");
+		return;
+	}
+
+	acrtc = to_amdgpu_crtc(wb_conn->encoder.crtc);
+	if (!acrtc) {
+		drm_err(adev_to_drm(adev), "no amdgpu_crtc found\n");
+		kfree(wb_info);
+		return;
+	}
+
+	afb = to_amdgpu_framebuffer(new_con_state->writeback_job->fb);
+	if (!afb) {
+		drm_err(adev_to_drm(adev), "No amdgpu_framebuffer found\n");
+		kfree(wb_info);
+		return;
+	}
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		if (dm->dc->current_state->res_ctx.pipe_ctx[i].stream == crtc_state->stream) {
+			pipe = &dm->dc->current_state->res_ctx.pipe_ctx[i];
+			break;
+		}
+	}
+
+	/* fill in wb_info */
+	wb_info->wb_enabled = true;
+
+	wb_info->dwb_pipe_inst = 0;
+	wb_info->dwb_params.dwbscl_black_color = 0;
+	wb_info->dwb_params.hdr_mult = 0x1F000;
+	wb_info->dwb_params.csc_params.gamut_adjust_type = CM_GAMUT_ADJUST_TYPE_BYPASS;
+	wb_info->dwb_params.csc_params.gamut_coef_format = CM_GAMUT_REMAP_COEF_FORMAT_S2_13;
+	wb_info->dwb_params.output_depth = DWB_OUTPUT_PIXEL_DEPTH_10BPC;
+	wb_info->dwb_params.cnv_params.cnv_out_bpc = DWB_CNV_OUT_BPC_10BPC;
+
+	/* width & height from crtc */
+	wb_info->dwb_params.cnv_params.src_width = acrtc->base.mode.crtc_hdisplay;
+	wb_info->dwb_params.cnv_params.src_height = acrtc->base.mode.crtc_vdisplay;
+	wb_info->dwb_params.dest_width = acrtc->base.mode.crtc_hdisplay;
+	wb_info->dwb_params.dest_height = acrtc->base.mode.crtc_vdisplay;
+
+	wb_info->dwb_params.cnv_params.crop_en = false;
+	wb_info->dwb_params.stereo_params.stereo_enabled = false;
+
+	wb_info->dwb_params.cnv_params.out_max_pix_val = 0x3ff;	// 10 bits
+	wb_info->dwb_params.cnv_params.out_min_pix_val = 0;
+	wb_info->dwb_params.cnv_params.fc_out_format = DWB_OUT_FORMAT_32BPP_ARGB;
+	wb_info->dwb_params.cnv_params.out_denorm_mode = DWB_OUT_DENORM_BYPASS;
+
+	wb_info->dwb_params.out_format = dwb_scaler_mode_bypass444;
+
+	wb_info->dwb_params.capture_rate = dwb_capture_rate_0;
+
+	wb_info->dwb_params.scaler_taps.h_taps = 4;
+	wb_info->dwb_params.scaler_taps.v_taps = 4;
+	wb_info->dwb_params.scaler_taps.h_taps_c = 2;
+	wb_info->dwb_params.scaler_taps.v_taps_c = 2;
+	wb_info->dwb_params.subsample_position = DWB_INTERSTITIAL_SUBSAMPLING;
+
+	wb_info->mcif_buf_params.luma_pitch = afb->base.pitches[0];
+	wb_info->mcif_buf_params.chroma_pitch = afb->base.pitches[1];
+
+	for (i = 0; i < DWB_MCIF_BUF_COUNT; i++) {
+		wb_info->mcif_buf_params.luma_address[i] = afb->address;
+		wb_info->mcif_buf_params.chroma_address[i] = 0;
+	}
+
+	wb_info->mcif_buf_params.p_vmid = 1;
+	if (amdgpu_ip_version(adev, DCE_HWIP, 0) >= IP_VERSION(3, 0, 0)) {
+		wb_info->mcif_warmup_params.start_address.quad_part = afb->address;
+		wb_info->mcif_warmup_params.region_size =
+			wb_info->mcif_buf_params.luma_pitch * wb_info->dwb_params.dest_height;
+	}
+	wb_info->mcif_warmup_params.p_vmid = 1;
+	wb_info->writeback_source_plane = pipe->plane_state;
+
+	dc_stream_add_writeback(dm->dc, crtc_state->stream, wb_info);
+
+	acrtc->wb_pending = true;
+	acrtc->wb_conn = wb_conn;
+	drm_writeback_queue_job(wb_conn, new_con_state);
+}
+
+static void amdgpu_dm_update_hdcp(struct drm_atomic_state *state)
+{
+	struct drm_connector_state *old_con_state, *new_con_state;
+	struct drm_device *dev = state->dev;
+	struct drm_connector *connector;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	int i;
+
+	if (!adev->dm.hdcp_workqueue)
+		return;
+
+	for_each_oldnew_connector_in_state(state, connector, old_con_state, new_con_state, i) {
+		struct dm_connector_state *dm_new_con_state = to_dm_connector_state(new_con_state);
+		struct amdgpu_crtc *acrtc = to_amdgpu_crtc(dm_new_con_state->base.crtc);
+		struct drm_crtc_state *old_crtc_state, *new_crtc_state;
+		struct dm_crtc_state *dm_new_crtc_state;
+		struct amdgpu_dm_connector *aconnector;
+
+		if (!connector || connector->connector_type == DRM_MODE_CONNECTOR_WRITEBACK)
+			continue;
+
+		aconnector = to_amdgpu_dm_connector(connector);
+
+		drm_dbg(dev, "[HDCP_DM] -------------- i : %x ----------\n", i);
+
+		drm_dbg(dev, "[HDCP_DM] connector->index: %x connect_status: %x dpms: %x\n",
+			connector->index, connector->status, connector->dpms);
+		drm_dbg(dev, "[HDCP_DM] state protection old: %x new: %x\n",
+			old_con_state->content_protection, new_con_state->content_protection);
+
+		if (aconnector->dc_sink) {
+			if (aconnector->dc_sink->sink_signal != SIGNAL_TYPE_VIRTUAL &&
+				aconnector->dc_sink->sink_signal != SIGNAL_TYPE_NONE) {
+				drm_dbg(dev, "[HDCP_DM] pipe_ctx dispname=%s\n",
+				aconnector->dc_sink->edid_caps.display_name);
+			}
+		}
+
+		new_crtc_state = NULL;
+		old_crtc_state = NULL;
+
+		if (acrtc) {
+			new_crtc_state = drm_atomic_get_new_crtc_state(state, &acrtc->base);
+			old_crtc_state = drm_atomic_get_old_crtc_state(state, &acrtc->base);
+		}
+
+		if (old_crtc_state)
+			drm_dbg(dev, "old crtc en: %x a: %x m: %x a-chg: %x c-chg: %x\n",
+			old_crtc_state->enable,
+			old_crtc_state->active,
+			old_crtc_state->mode_changed,
+			old_crtc_state->active_changed,
+			old_crtc_state->connectors_changed);
+
+		if (new_crtc_state)
+			drm_dbg(dev, "NEW crtc en: %x a: %x m: %x a-chg: %x c-chg: %x\n",
+			new_crtc_state->enable,
+			new_crtc_state->active,
+			new_crtc_state->mode_changed,
+			new_crtc_state->active_changed,
+			new_crtc_state->connectors_changed);
+
+
+		dm_new_crtc_state = to_dm_crtc_state(new_crtc_state);
+
+		if (dm_new_crtc_state && dm_new_crtc_state->stream == NULL &&
+		    connector->state->content_protection == DRM_MODE_CONTENT_PROTECTION_ENABLED) {
+			hdcp_reset_display(adev->dm.hdcp_workqueue, aconnector->dc_link->link_index);
+			new_con_state->content_protection = DRM_MODE_CONTENT_PROTECTION_DESIRED;
+			dm_new_con_state->update_hdcp = true;
+			continue;
+		}
+
+		if (is_content_protection_different(new_crtc_state, old_crtc_state, new_con_state,
+											old_con_state, connector, adev->dm.hdcp_workqueue)) {
+			/* when display is unplugged from mst hub, connctor will
+			 * be destroyed within dm_dp_mst_connector_destroy. connector
+			 * hdcp perperties, like type, undesired, desired, enabled,
+			 * will be lost. So, save hdcp properties into hdcp_work within
+			 * amdgpu_dm_atomic_commit_tail. if the same display is
+			 * plugged back with same display index, its hdcp properties
+			 * will be retrieved from hdcp_work within dm_dp_mst_get_modes
+			 */
+
+			bool enable_encryption = false;
+
+			if (new_con_state->content_protection == DRM_MODE_CONTENT_PROTECTION_DESIRED)
+				enable_encryption = true;
+
+			if (aconnector->dc_link && aconnector->dc_sink &&
+				aconnector->dc_link->type == dc_connection_mst_branch) {
+				struct hdcp_workqueue *hdcp_work = adev->dm.hdcp_workqueue;
+				struct hdcp_workqueue *hdcp_w =
+					&hdcp_work[aconnector->dc_link->link_index];
+
+				hdcp_w->hdcp_content_type[connector->index] =
+					new_con_state->hdcp_content_type;
+				hdcp_w->content_protection[connector->index] =
+					new_con_state->content_protection;
+			}
+
+			if (new_crtc_state && new_crtc_state->mode_changed &&
+				new_con_state->content_protection >= DRM_MODE_CONTENT_PROTECTION_DESIRED)
+				enable_encryption = true;
+
+			drm_info(dev, "[HDCP_DM] hdcp_update_display enable_encryption = %x\n", enable_encryption);
+
+			if (aconnector->dc_link)
+				hdcp_update_display(
+					adev->dm.hdcp_workqueue, aconnector->dc_link->link_index, aconnector,
+					new_con_state->hdcp_content_type, enable_encryption);
+		}
+	}
+}
+
+static int amdgpu_dm_atomic_setup_commit(struct drm_atomic_state *state)
+{
+	struct drm_crtc *crtc;
+	struct drm_crtc_state *old_crtc_state, *new_crtc_state;
+	struct dm_crtc_state *dm_old_crtc_state, *dm_new_crtc_state;
+	int i, ret;
+
+	ret = drm_dp_mst_atomic_setup_commit(state);
+	if (ret)
+		return ret;
+
+	for_each_oldnew_crtc_in_state(state, crtc, old_crtc_state, new_crtc_state, i) {
+		dm_old_crtc_state = to_dm_crtc_state(old_crtc_state);
+		dm_new_crtc_state = to_dm_crtc_state(new_crtc_state);
+		/*
+		 * Color management settings. We also update color properties
+		 * when a modeset is needed, to ensure it gets reprogrammed.
+		 */
+		if (dm_new_crtc_state->base.active && dm_new_crtc_state->stream &&
+		    (dm_new_crtc_state->base.color_mgmt_changed ||
+		     dm_old_crtc_state->regamma_tf != dm_new_crtc_state->regamma_tf ||
+		     drm_atomic_crtc_needs_modeset(new_crtc_state))) {
+			ret = amdgpu_dm_update_crtc_color_mgmt(dm_new_crtc_state);
+			if (ret) {
+				drm_dbg_atomic(state->dev, "Failed to update color state\n");
+				return ret;
+			}
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_dm_atomic_commit_tail() - AMDgpu DM's commit tail implementation.
+ * @state: The atomic state to commit
+ *
+ * This will tell DC to commit the constructed DC state from atomic_check,
+ * programming the hardware. Any failures here implies a hardware failure, since
+ * atomic check should have filtered anything non-kosher.
+ */
+static void amdgpu_dm_atomic_commit_tail(struct drm_atomic_state *state)
+{
+	struct drm_device *dev = state->dev;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	struct amdgpu_display_manager *dm = &adev->dm;
+	struct dm_atomic_state *dm_state;
+	struct dc_state *dc_state = NULL;
+	u32 i, j;
+	struct drm_crtc *crtc;
+	struct drm_crtc_state *old_crtc_state, *new_crtc_state;
+	unsigned long flags;
+	bool wait_for_vblank = true;
+	struct drm_connector *connector;
+	struct drm_connector_state *old_con_state = NULL, *new_con_state = NULL;
+	struct dm_crtc_state *dm_old_crtc_state, *dm_new_crtc_state;
+	int crtc_disable_count = 0;
+
+	trace_amdgpu_dm_atomic_commit_tail_begin(state);
+
+	drm_atomic_helper_update_legacy_modeset_state(dev, state);
+	drm_dp_mst_atomic_wait_for_dependencies(state);
+
+	dm_state = dm_atomic_get_new_state(state);
+	if (dm_state && dm_state->context) {
+		dc_state = dm_state->context;
+		amdgpu_dm_commit_streams(state, dc_state);
+	}
+
+	amdgpu_dm_update_hdcp(state);
+
+	/* Handle connector state changes */
+	for_each_oldnew_connector_in_state(state, connector, old_con_state, new_con_state, i) {
+		struct dm_connector_state *dm_new_con_state = to_dm_connector_state(new_con_state);
+		struct dm_connector_state *dm_old_con_state = to_dm_connector_state(old_con_state);
+		struct amdgpu_crtc *acrtc = to_amdgpu_crtc(dm_new_con_state->base.crtc);
+		struct dc_surface_update *dummy_updates;
+		struct dc_stream_update stream_update;
+		struct dc_info_packet hdr_packet;
+		struct dc_stream_status *status = NULL;
+		bool abm_changed, hdr_changed, scaling_changed, output_color_space_changed = false;
+
+		memset(&stream_update, 0, sizeof(stream_update));
+
+		if (acrtc) {
+			new_crtc_state = drm_atomic_get_new_crtc_state(state, &acrtc->base);
+			old_crtc_state = drm_atomic_get_old_crtc_state(state, &acrtc->base);
+		}
+
+		/* Skip any modesets/resets */
+		if (!acrtc || drm_atomic_crtc_needs_modeset(new_crtc_state))
+			continue;
+
+		dm_new_crtc_state = to_dm_crtc_state(new_crtc_state);
+		dm_old_crtc_state = to_dm_crtc_state(old_crtc_state);
+
+		scaling_changed = is_scaling_state_different(dm_new_con_state,
+							     dm_old_con_state);
+
+		if ((new_con_state->hdmi.broadcast_rgb != old_con_state->hdmi.broadcast_rgb) &&
+			(dm_old_crtc_state->stream->output_color_space !=
+				get_output_color_space(&dm_new_crtc_state->stream->timing, new_con_state)))
+			output_color_space_changed = true;
+
+		abm_changed = dm_new_crtc_state->abm_level !=
+			      dm_old_crtc_state->abm_level;
+
+		hdr_changed =
+			!drm_connector_atomic_hdr_metadata_equal(old_con_state, new_con_state);
+
+		if (!scaling_changed && !abm_changed && !hdr_changed && !output_color_space_changed)
+			continue;
+
+		stream_update.stream = dm_new_crtc_state->stream;
+		if (scaling_changed) {
+			update_stream_scaling_settings(&dm_new_con_state->base.crtc->mode,
+					dm_new_con_state, dm_new_crtc_state->stream);
+
+			stream_update.src = dm_new_crtc_state->stream->src;
+			stream_update.dst = dm_new_crtc_state->stream->dst;
+		}
+
+		if (output_color_space_changed) {
+			dm_new_crtc_state->stream->output_color_space
+				= get_output_color_space(&dm_new_crtc_state->stream->timing, new_con_state);
+
+			stream_update.output_color_space = &dm_new_crtc_state->stream->output_color_space;
+		}
+
+		if (abm_changed) {
+			dm_new_crtc_state->stream->abm_level = dm_new_crtc_state->abm_level;
+
+			stream_update.abm_level = &dm_new_crtc_state->abm_level;
+		}
+
+		if (hdr_changed) {
+			fill_hdr_info_packet(new_con_state, &hdr_packet);
+			stream_update.hdr_static_metadata = &hdr_packet;
+		}
+
+		status = dc_stream_get_status(dm_new_crtc_state->stream);
+
+		if (WARN_ON(!status))
+			continue;
+
+		WARN_ON(!status->plane_count);
+
+		/*
+		 * TODO: DC refuses to perform stream updates without a dc_surface_update.
+		 * Here we create an empty update on each plane.
+		 * To fix this, DC should permit updating only stream properties.
+		 */
+		dummy_updates = kzalloc(sizeof(struct dc_surface_update) * MAX_SURFACES, GFP_ATOMIC);
+		if (!dummy_updates) {
+			drm_err(adev_to_drm(adev), "Failed to allocate memory for dummy_updates.\n");
+			continue;
+		}
+		for (j = 0; j < status->plane_count; j++)
+			dummy_updates[j].surface = status->plane_states[0];
+
+		sort(dummy_updates, status->plane_count,
+		     sizeof(*dummy_updates), dm_plane_layer_index_cmp, NULL);
+
+		mutex_lock(&dm->dc_lock);
+		dc_exit_ips_for_hw_access(dm->dc);
+		dc_update_planes_and_stream(dm->dc,
+					    dummy_updates,
+					    status->plane_count,
+					    dm_new_crtc_state->stream,
+					    &stream_update);
+		mutex_unlock(&dm->dc_lock);
+		kfree(dummy_updates);
+
+		drm_connector_update_privacy_screen(new_con_state);
+	}
+
+	/**
+	 * Enable interrupts for CRTCs that are newly enabled or went through
+	 * a modeset. It was intentionally deferred until after the front end
+	 * state was modified to wait until the OTG was on and so the IRQ
+	 * handlers didn't access stale or invalid state.
+	 */
+	for_each_oldnew_crtc_in_state(state, crtc, old_crtc_state, new_crtc_state, i) {
+		struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
+#ifdef CONFIG_DEBUG_FS
+		enum amdgpu_dm_pipe_crc_source cur_crc_src;
+#endif
+		/* Count number of newly disabled CRTCs for dropping PM refs later. */
+		if (old_crtc_state->active && !new_crtc_state->active)
+			crtc_disable_count++;
+
+		dm_new_crtc_state = to_dm_crtc_state(new_crtc_state);
+		dm_old_crtc_state = to_dm_crtc_state(old_crtc_state);
+
+		/* For freesync config update on crtc state and params for irq */
+		update_stream_irq_parameters(dm, dm_new_crtc_state);
+
+#ifdef CONFIG_DEBUG_FS
+		spin_lock_irqsave(&adev_to_drm(adev)->event_lock, flags);
+		cur_crc_src = acrtc->dm_irq_params.crc_src;
+		spin_unlock_irqrestore(&adev_to_drm(adev)->event_lock, flags);
+#endif
+
+		if (new_crtc_state->active &&
+		    (!old_crtc_state->active ||
+		     drm_atomic_crtc_needs_modeset(new_crtc_state))) {
+			dc_stream_retain(dm_new_crtc_state->stream);
+			acrtc->dm_irq_params.stream = dm_new_crtc_state->stream;
+			manage_dm_interrupts(adev, acrtc, dm_new_crtc_state);
+		}
+		/* Handle vrr on->off / off->on transitions */
+		amdgpu_dm_handle_vrr_transition(dm_old_crtc_state, dm_new_crtc_state);
+
+#ifdef CONFIG_DEBUG_FS
+		if (new_crtc_state->active &&
+		    (!old_crtc_state->active ||
+		     drm_atomic_crtc_needs_modeset(new_crtc_state))) {
+			/**
+			 * Frontend may have changed so reapply the CRC capture
+			 * settings for the stream.
+			 */
+			if (amdgpu_dm_is_valid_crc_source(cur_crc_src)) {
+#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY)
+				if (amdgpu_dm_crc_window_is_activated(crtc)) {
+					uint8_t cnt;
+
+					spin_lock_irqsave(&adev_to_drm(adev)->event_lock, flags);
+					for (cnt = 0; cnt < MAX_CRC_WINDOW_NUM; cnt++) {
+						if (acrtc->dm_irq_params.window_param[cnt].enable) {
+							acrtc->dm_irq_params.window_param[cnt].update_win = true;
+
+							/**
+							 * It takes 2 frames for HW to stably generate CRC when
+							 * resuming from suspend, so we set skip_frame_cnt 2.
+							 */
+							acrtc->dm_irq_params.window_param[cnt].skip_frame_cnt = 2;
+						}
+					}
+					spin_unlock_irqrestore(&adev_to_drm(adev)->event_lock, flags);
+				}
+#endif
+				if (amdgpu_dm_crtc_configure_crc_source(
+					crtc, dm_new_crtc_state, cur_crc_src))
+					drm_dbg_atomic(dev, "Failed to configure crc source");
+			}
+		}
+#endif
+	}
+
+	for_each_new_crtc_in_state(state, crtc, new_crtc_state, j)
+		if (new_crtc_state->async_flip)
+			wait_for_vblank = false;
+
+	/* update planes when needed per crtc*/
+	for_each_new_crtc_in_state(state, crtc, new_crtc_state, j) {
+		dm_new_crtc_state = to_dm_crtc_state(new_crtc_state);
+
+		if (dm_new_crtc_state->stream)
+			amdgpu_dm_commit_planes(state, dev, dm, crtc, wait_for_vblank);
+	}
+
+	/* Enable writeback */
+	for_each_new_connector_in_state(state, connector, new_con_state, i) {
+		struct dm_connector_state *dm_new_con_state = to_dm_connector_state(new_con_state);
+		struct amdgpu_crtc *acrtc = to_amdgpu_crtc(dm_new_con_state->base.crtc);
+
+		if (connector->connector_type != DRM_MODE_CONNECTOR_WRITEBACK)
+			continue;
+
+		if (!new_con_state->writeback_job)
+			continue;
+
+		new_crtc_state = drm_atomic_get_new_crtc_state(state, &acrtc->base);
+
+		if (!new_crtc_state)
+			continue;
+
+		if (acrtc->wb_enabled)
+			continue;
+
+		dm_new_crtc_state = to_dm_crtc_state(new_crtc_state);
+
+		dm_set_writeback(dm, dm_new_crtc_state, connector, new_con_state);
+		acrtc->wb_enabled = true;
+	}
+
+	/* Update audio instances for each connector. */
+	amdgpu_dm_commit_audio(dev, state);
+
+	/* restore the backlight level */
+	for (i = 0; i < dm->num_of_edps; i++) {
+		if (dm->backlight_dev[i] &&
+		    (dm->actual_brightness[i] != dm->brightness[i]))
+			amdgpu_dm_backlight_set_level(dm, i, dm->brightness[i]);
+	}
+
+	/*
+	 * send vblank event on all events not handled in flip and
+	 * mark consumed event for drm_atomic_helper_commit_hw_done
+	 */
+	spin_lock_irqsave(&adev_to_drm(adev)->event_lock, flags);
+	for_each_new_crtc_in_state(state, crtc, new_crtc_state, i) {
+
+		if (new_crtc_state->event)
+			drm_send_event_locked(dev, &new_crtc_state->event->base);
+
+		new_crtc_state->event = NULL;
+	}
+	spin_unlock_irqrestore(&adev_to_drm(adev)->event_lock, flags);
+
+	/* Signal HW programming completion */
+	drm_atomic_helper_commit_hw_done(state);
+
+	if (wait_for_vblank)
+		drm_atomic_helper_wait_for_flip_done(dev, state);
+
+	drm_atomic_helper_cleanup_planes(dev, state);
+
+	/* Don't free the memory if we are hitting this as part of suspend.
+	 * This way we don't free any memory during suspend; see
+	 * amdgpu_bo_free_kernel().  The memory will be freed in the first
+	 * non-suspend modeset or when the driver is torn down.
+	 */
+	if (!adev->in_suspend) {
+		/* return the stolen vga memory back to VRAM */
+		if (!adev->mman.keep_stolen_vga_memory)
+			amdgpu_bo_free_kernel(&adev->mman.stolen_vga_memory, NULL, NULL);
+		amdgpu_bo_free_kernel(&adev->mman.stolen_extended_memory, NULL, NULL);
+	}
+
+	/*
+	 * Finally, drop a runtime PM reference for each newly disabled CRTC,
+	 * so we can put the GPU into runtime suspend if we're not driving any
+	 * displays anymore
+	 */
+	for (i = 0; i < crtc_disable_count; i++)
+		pm_runtime_put_autosuspend(dev->dev);
+	pm_runtime_mark_last_busy(dev->dev);
+
+	trace_amdgpu_dm_atomic_commit_tail_finish(state);
+}
+
+static int dm_force_atomic_commit(struct drm_connector *connector)
+{
+	int ret = 0;
+	struct drm_device *ddev = connector->dev;
+	struct drm_atomic_state *state = drm_atomic_state_alloc(ddev);
+	struct amdgpu_crtc *disconnected_acrtc = to_amdgpu_crtc(connector->encoder->crtc);
+	struct drm_plane *plane = disconnected_acrtc->base.primary;
+	struct drm_connector_state *conn_state;
+	struct drm_crtc_state *crtc_state;
+	struct drm_plane_state *plane_state;
+
+	if (!state)
+		return -ENOMEM;
+
+	state->acquire_ctx = ddev->mode_config.acquire_ctx;
+
+	/* Construct an atomic state to restore previous display setting */
+
+	/*
+	 * Attach connectors to drm_atomic_state
+	 */
+	conn_state = drm_atomic_get_connector_state(state, connector);
+
+	/* Check for error in getting connector state */
+	if (IS_ERR(conn_state)) {
+		ret = PTR_ERR(conn_state);
+		goto out;
+	}
+
+	/* Attach crtc to drm_atomic_state*/
+	crtc_state = drm_atomic_get_crtc_state(state, &disconnected_acrtc->base);
+
+	/* Check for error in getting crtc state */
+	if (IS_ERR(crtc_state)) {
+		ret = PTR_ERR(crtc_state);
+		goto out;
+	}
+
+	/* force a restore */
+	crtc_state->mode_changed = true;
+
+	/* Attach plane to drm_atomic_state */
+	plane_state = drm_atomic_get_plane_state(state, plane);
+
+	/* Check for error in getting plane state */
+	if (IS_ERR(plane_state)) {
+		ret = PTR_ERR(plane_state);
+		goto out;
+	}
+
+	/* Call commit internally with the state we just constructed */
+	ret = drm_atomic_commit(state);
+
+out:
+	drm_atomic_state_put(state);
+	if (ret)
+		drm_err(ddev, "Restoring old state failed with %i\n", ret);
+
+	return ret;
+}
+
+/*
+ * This function handles all cases when set mode does not come upon hotplug.
+ * This includes when a display is unplugged then plugged back into the
+ * same port and when running without usermode desktop manager supprot
+ */
+void dm_restore_drm_connector_state(struct drm_device *dev,
+				    struct drm_connector *connector)
+{
+	struct amdgpu_dm_connector *aconnector;
+	struct amdgpu_crtc *disconnected_acrtc;
+	struct dm_crtc_state *acrtc_state;
+
+	if (connector->connector_type == DRM_MODE_CONNECTOR_WRITEBACK)
+		return;
+
+	aconnector = to_amdgpu_dm_connector(connector);
+
+	if (!aconnector->dc_sink || !connector->state || !connector->encoder)
+		return;
+
+	disconnected_acrtc = to_amdgpu_crtc(connector->encoder->crtc);
+	if (!disconnected_acrtc)
+		return;
+
+	acrtc_state = to_dm_crtc_state(disconnected_acrtc->base.state);
+	if (!acrtc_state->stream)
+		return;
+
+	/*
+	 * If the previous sink is not released and different from the current,
+	 * we deduce we are in a state where we can not rely on usermode call
+	 * to turn on the display, so we do it here
+	 */
+	if (acrtc_state->stream->sink != aconnector->dc_sink)
+		dm_force_atomic_commit(&aconnector->base);
+}
+
+/*
+ * Grabs all modesetting locks to serialize against any blocking commits,
+ * Waits for completion of all non blocking commits.
+ */
+static int do_aquire_global_lock(struct drm_device *dev,
+				 struct drm_atomic_state *state)
+{
+	struct drm_crtc *crtc;
+	struct drm_crtc_commit *commit;
+	long ret;
+
+	/*
+	 * Adding all modeset locks to aquire_ctx will
+	 * ensure that when the framework release it the
+	 * extra locks we are locking here will get released to
+	 */
+	ret = drm_modeset_lock_all_ctx(dev, state->acquire_ctx);
+	if (ret)
+		return ret;
+
+	list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
+		spin_lock(&crtc->commit_lock);
+		commit = list_first_entry_or_null(&crtc->commit_list,
+				struct drm_crtc_commit, commit_entry);
+		if (commit)
+			drm_crtc_commit_get(commit);
+		spin_unlock(&crtc->commit_lock);
+
+		if (!commit)
+			continue;
+
+		/*
+		 * Make sure all pending HW programming completed and
+		 * page flips done
+		 */
+		ret = wait_for_completion_interruptible_timeout(&commit->hw_done, 10*HZ);
+
+		if (ret > 0)
+			ret = wait_for_completion_interruptible_timeout(
+					&commit->flip_done, 10*HZ);
+
+		if (ret == 0)
+			drm_err(dev, "[CRTC:%d:%s] hw_done or flip_done timed out\n",
+				  crtc->base.id, crtc->name);
+
+		drm_crtc_commit_put(commit);
+	}
+
+	return ret < 0 ? ret : 0;
+}
+
+static void get_freesync_config_for_crtc(
+	struct dm_crtc_state *new_crtc_state,
+	struct dm_connector_state *new_con_state)
+{
+	struct mod_freesync_config config = {0};
+	struct amdgpu_dm_connector *aconnector;
+	struct drm_display_mode *mode = &new_crtc_state->base.mode;
+	int vrefresh = drm_mode_vrefresh(mode);
+	bool fs_vid_mode = false;
+
+	if (new_con_state->base.connector->connector_type == DRM_MODE_CONNECTOR_WRITEBACK)
+		return;
+
+	aconnector = to_amdgpu_dm_connector(new_con_state->base.connector);
+
+	new_crtc_state->vrr_supported = new_con_state->freesync_capable &&
+					vrefresh >= aconnector->min_vfreq &&
+					vrefresh <= aconnector->max_vfreq;
+
+	if (new_crtc_state->vrr_supported) {
+		new_crtc_state->stream->ignore_msa_timing_param = true;
+		fs_vid_mode = new_crtc_state->freesync_config.state == VRR_STATE_ACTIVE_FIXED;
+
+		config.min_refresh_in_uhz = aconnector->min_vfreq * 1000000;
+		config.max_refresh_in_uhz = aconnector->max_vfreq * 1000000;
+		config.vsif_supported = true;
+		config.btr = true;
+
+		if (fs_vid_mode) {
+			config.state = VRR_STATE_ACTIVE_FIXED;
+			config.fixed_refresh_in_uhz = new_crtc_state->freesync_config.fixed_refresh_in_uhz;
+			goto out;
+		} else if (new_crtc_state->base.vrr_enabled) {
+			config.state = VRR_STATE_ACTIVE_VARIABLE;
+		} else {
+			config.state = VRR_STATE_INACTIVE;
+		}
+	} else {
+		config.state = VRR_STATE_UNSUPPORTED;
+	}
+out:
+	new_crtc_state->freesync_config = config;
+}
+
+static void reset_freesync_config_for_crtc(
+	struct dm_crtc_state *new_crtc_state)
+{
+	new_crtc_state->vrr_supported = false;
+
+	memset(&new_crtc_state->vrr_infopacket, 0,
+	       sizeof(new_crtc_state->vrr_infopacket));
+}
+
+static bool
+is_timing_unchanged_for_freesync(struct drm_crtc_state *old_crtc_state,
+				 struct drm_crtc_state *new_crtc_state)
+{
+	const struct drm_display_mode *old_mode, *new_mode;
+
+	if (!old_crtc_state || !new_crtc_state)
+		return false;
+
+	old_mode = &old_crtc_state->mode;
+	new_mode = &new_crtc_state->mode;
+
+	if (old_mode->clock       == new_mode->clock &&
+	    old_mode->hdisplay    == new_mode->hdisplay &&
+	    old_mode->vdisplay    == new_mode->vdisplay &&
+	    old_mode->htotal      == new_mode->htotal &&
+	    old_mode->vtotal      != new_mode->vtotal &&
+	    old_mode->hsync_start == new_mode->hsync_start &&
+	    old_mode->vsync_start != new_mode->vsync_start &&
+	    old_mode->hsync_end   == new_mode->hsync_end &&
+	    old_mode->vsync_end   != new_mode->vsync_end &&
+	    old_mode->hskew       == new_mode->hskew &&
+	    old_mode->vscan       == new_mode->vscan &&
+	    (old_mode->vsync_end - old_mode->vsync_start) ==
+	    (new_mode->vsync_end - new_mode->vsync_start))
+		return true;
+
+	return false;
+}
+
+static void set_freesync_fixed_config(struct dm_crtc_state *dm_new_crtc_state)
+{
+	u64 num, den, res;
+	struct drm_crtc_state *new_crtc_state = &dm_new_crtc_state->base;
+
+	dm_new_crtc_state->freesync_config.state = VRR_STATE_ACTIVE_FIXED;
+
+	num = (unsigned long long)new_crtc_state->mode.clock * 1000 * 1000000;
+	den = (unsigned long long)new_crtc_state->mode.htotal *
+	      (unsigned long long)new_crtc_state->mode.vtotal;
+
+	res = div_u64(num, den);
+	dm_new_crtc_state->freesync_config.fixed_refresh_in_uhz = res;
+}
+
+static int dm_update_crtc_state(struct amdgpu_display_manager *dm,
+			 struct drm_atomic_state *state,
+			 struct drm_crtc *crtc,
+			 struct drm_crtc_state *old_crtc_state,
+			 struct drm_crtc_state *new_crtc_state,
+			 bool enable,
+			 bool *lock_and_validation_needed)
+{
+	struct dm_atomic_state *dm_state = NULL;
+	struct dm_crtc_state *dm_old_crtc_state, *dm_new_crtc_state;
+	struct dc_stream_state *new_stream;
+	struct amdgpu_device *adev = dm->adev;
+	int ret = 0;
+
+	/*
+	 * TODO Move this code into dm_crtc_atomic_check once we get rid of dc_validation_set
+	 * update changed items
+	 */
+	struct amdgpu_crtc *acrtc = NULL;
+	struct drm_connector *connector = NULL;
+	struct amdgpu_dm_connector *aconnector = NULL;
+	struct drm_connector_state *drm_new_conn_state = NULL, *drm_old_conn_state = NULL;
+	struct dm_connector_state *dm_new_conn_state = NULL, *dm_old_conn_state = NULL;
+
+	new_stream = NULL;
+
+	dm_old_crtc_state = to_dm_crtc_state(old_crtc_state);
+	dm_new_crtc_state = to_dm_crtc_state(new_crtc_state);
+	acrtc = to_amdgpu_crtc(crtc);
+	connector = amdgpu_dm_find_first_crtc_matching_connector(state, crtc);
+	if (connector)
+		aconnector = to_amdgpu_dm_connector(connector);
+
+	/* TODO This hack should go away */
+	if (connector && enable) {
+		/* Make sure fake sink is created in plug-in scenario */
+		drm_new_conn_state = drm_atomic_get_new_connector_state(state,
+									connector);
+		drm_old_conn_state = drm_atomic_get_old_connector_state(state,
+									connector);
+
+		if (WARN_ON(!drm_new_conn_state)) {
+			ret = -EINVAL;
+			goto fail;
+		}
+
+		dm_new_conn_state = to_dm_connector_state(drm_new_conn_state);
+		dm_old_conn_state = to_dm_connector_state(drm_old_conn_state);
+
+		if (!drm_atomic_crtc_needs_modeset(new_crtc_state))
+			goto skip_modeset;
+
+		new_stream = create_validate_stream_for_sink(connector,
+							     &new_crtc_state->mode,
+							     dm_new_conn_state,
+							     dm_old_crtc_state->stream);
+
+		/*
+		 * we can have no stream on ACTION_SET if a display
+		 * was disconnected during S3, in this case it is not an
+		 * error, the OS will be updated after detection, and
+		 * will do the right thing on next atomic commit
+		 */
+
+		if (!new_stream) {
+			drm_dbg_driver(adev_to_drm(adev), "%s: Failed to create new stream for crtc %d\n",
+					__func__, acrtc->base.base.id);
+			ret = -ENOMEM;
+			goto fail;
+		}
+
+		/*
+		 * TODO: Check VSDB bits to decide whether this should
+		 * be enabled or not.
+		 */
+		new_stream->triggered_crtc_reset.enabled =
+			dm->force_timing_sync;
+
+		dm_new_crtc_state->abm_level = dm_new_conn_state->abm_level;
+
+		ret = fill_hdr_info_packet(drm_new_conn_state,
+					   &new_stream->hdr_static_metadata);
+		if (ret)
+			goto fail;
+
+		/*
+		 * If we already removed the old stream from the context
+		 * (and set the new stream to NULL) then we can't reuse
+		 * the old stream even if the stream and scaling are unchanged.
+		 * We'll hit the BUG_ON and black screen.
+		 *
+		 * TODO: Refactor this function to allow this check to work
+		 * in all conditions.
+		 */
+		if (amdgpu_freesync_vid_mode &&
+		    dm_new_crtc_state->stream &&
+		    is_timing_unchanged_for_freesync(new_crtc_state, old_crtc_state))
+			goto skip_modeset;
+
+		if (dm_new_crtc_state->stream &&
+		    dc_is_stream_unchanged(new_stream, dm_old_crtc_state->stream) &&
+		    dc_is_stream_scaling_unchanged(new_stream, dm_old_crtc_state->stream)) {
+			new_crtc_state->mode_changed = false;
+			drm_dbg_driver(adev_to_drm(adev), "Mode change not required, setting mode_changed to %d",
+					 new_crtc_state->mode_changed);
+		}
+	}
+
+	/* mode_changed flag may get updated above, need to check again */
+	if (!drm_atomic_crtc_needs_modeset(new_crtc_state))
+		goto skip_modeset;
+
+	drm_dbg_state(state->dev,
+		"amdgpu_crtc id:%d crtc_state_flags: enable:%d, active:%d, planes_changed:%d, mode_changed:%d,active_changed:%d,connectors_changed:%d\n",
+		acrtc->crtc_id,
+		new_crtc_state->enable,
+		new_crtc_state->active,
+		new_crtc_state->planes_changed,
+		new_crtc_state->mode_changed,
+		new_crtc_state->active_changed,
+		new_crtc_state->connectors_changed);
+
+	/* Remove stream for any changed/disabled CRTC */
+	if (!enable) {
+
+		if (!dm_old_crtc_state->stream)
+			goto skip_modeset;
+
+		/* Unset freesync video if it was active before */
+		if (dm_old_crtc_state->freesync_config.state == VRR_STATE_ACTIVE_FIXED) {
+			dm_new_crtc_state->freesync_config.state = VRR_STATE_INACTIVE;
+			dm_new_crtc_state->freesync_config.fixed_refresh_in_uhz = 0;
+		}
+
+		/* Now check if we should set freesync video mode */
+		if (amdgpu_freesync_vid_mode && dm_new_crtc_state->stream &&
+		    dc_is_stream_unchanged(new_stream, dm_old_crtc_state->stream) &&
+		    dc_is_stream_scaling_unchanged(new_stream, dm_old_crtc_state->stream) &&
+		    is_timing_unchanged_for_freesync(new_crtc_state,
+						     old_crtc_state)) {
+			new_crtc_state->mode_changed = false;
+			drm_dbg_driver(adev_to_drm(adev),
+				"Mode change not required for front porch change, setting mode_changed to %d",
+				new_crtc_state->mode_changed);
+
+			set_freesync_fixed_config(dm_new_crtc_state);
+
+			goto skip_modeset;
+		} else if (amdgpu_freesync_vid_mode && aconnector &&
+			   is_freesync_video_mode(&new_crtc_state->mode,
+						  aconnector)) {
+			struct drm_display_mode *high_mode;
+
+			high_mode = get_highest_refresh_rate_mode(aconnector, false);
+			if (!drm_mode_equal(&new_crtc_state->mode, high_mode))
+				set_freesync_fixed_config(dm_new_crtc_state);
+		}
+
+		ret = dm_atomic_get_state(state, &dm_state);
+		if (ret)
+			goto fail;
+
+		drm_dbg_driver(adev_to_drm(adev), "Disabling DRM crtc: %d\n",
+				crtc->base.id);
+
+		/* i.e. reset mode */
+		if (dc_state_remove_stream(
+				dm->dc,
+				dm_state->context,
+				dm_old_crtc_state->stream) != DC_OK) {
+			ret = -EINVAL;
+			goto fail;
+		}
+
+		dc_stream_release(dm_old_crtc_state->stream);
+		dm_new_crtc_state->stream = NULL;
+
+		reset_freesync_config_for_crtc(dm_new_crtc_state);
+
+		*lock_and_validation_needed = true;
+
+	} else {/* Add stream for any updated/enabled CRTC */
+		/*
+		 * Quick fix to prevent NULL pointer on new_stream when
+		 * added MST connectors not found in existing crtc_state in the chained mode
+		 * TODO: need to dig out the root cause of that
+		 */
+		if (!connector)
+			goto skip_modeset;
+
+		if (modereset_required(new_crtc_state))
+			goto skip_modeset;
+
+		if (amdgpu_dm_crtc_modeset_required(new_crtc_state, new_stream,
+				     dm_old_crtc_state->stream)) {
+
+			WARN_ON(dm_new_crtc_state->stream);
+
+			ret = dm_atomic_get_state(state, &dm_state);
+			if (ret)
+				goto fail;
+
+			dm_new_crtc_state->stream = new_stream;
+
+			dc_stream_retain(new_stream);
+
+			DRM_DEBUG_ATOMIC("Enabling DRM crtc: %d\n",
+					 crtc->base.id);
+
+			if (dc_state_add_stream(
+					dm->dc,
+					dm_state->context,
+					dm_new_crtc_state->stream) != DC_OK) {
+				ret = -EINVAL;
+				goto fail;
+			}
+
+			*lock_and_validation_needed = true;
+		}
+	}
+
+skip_modeset:
+	/* Release extra reference */
+	if (new_stream)
+		dc_stream_release(new_stream);
+
+	/*
+	 * We want to do dc stream updates that do not require a
+	 * full modeset below.
+	 */
+	if (!(enable && connector && new_crtc_state->active))
+		return 0;
+	/*
+	 * Given above conditions, the dc state cannot be NULL because:
+	 * 1. We're in the process of enabling CRTCs (just been added
+	 *    to the dc context, or already is on the context)
+	 * 2. Has a valid connector attached, and
+	 * 3. Is currently active and enabled.
+	 * => The dc stream state currently exists.
+	 */
+	BUG_ON(dm_new_crtc_state->stream == NULL);
+
+	/* Scaling or underscan settings */
+	if (is_scaling_state_different(dm_old_conn_state, dm_new_conn_state) ||
+				drm_atomic_crtc_needs_modeset(new_crtc_state))
+		update_stream_scaling_settings(
+			&new_crtc_state->mode, dm_new_conn_state, dm_new_crtc_state->stream);
+
+	/* ABM settings */
+	dm_new_crtc_state->abm_level = dm_new_conn_state->abm_level;
+
+	/*
+	 * Color management settings. We also update color properties
+	 * when a modeset is needed, to ensure it gets reprogrammed.
+	 */
+	if (dm_new_crtc_state->base.color_mgmt_changed ||
+	    dm_old_crtc_state->regamma_tf != dm_new_crtc_state->regamma_tf ||
+	    drm_atomic_crtc_needs_modeset(new_crtc_state)) {
+		ret = amdgpu_dm_check_crtc_color_mgmt(dm_new_crtc_state, true);
+		if (ret)
+			goto fail;
+	}
+
+	/* Update Freesync settings. */
+	get_freesync_config_for_crtc(dm_new_crtc_state,
+				     dm_new_conn_state);
+
+	return ret;
+
+fail:
+	if (new_stream)
+		dc_stream_release(new_stream);
+	return ret;
+}
+
+static bool should_reset_plane(struct drm_atomic_state *state,
+			       struct drm_plane *plane,
+			       struct drm_plane_state *old_plane_state,
+			       struct drm_plane_state *new_plane_state)
+{
+	struct drm_plane *other;
+	struct drm_plane_state *old_other_state, *new_other_state;
+	struct drm_crtc_state *old_crtc_state, *new_crtc_state;
+	struct dm_crtc_state *old_dm_crtc_state, *new_dm_crtc_state;
+	struct amdgpu_device *adev = drm_to_adev(plane->dev);
+	int i;
+
+	/*
+	 * TODO: Remove this hack for all asics once it proves that the
+	 * fast updates works fine on DCN3.2+.
+	 */
+	if (amdgpu_ip_version(adev, DCE_HWIP, 0) < IP_VERSION(3, 2, 0) &&
+	    state->allow_modeset)
+		return true;
+
+	if (amdgpu_in_reset(adev) && state->allow_modeset)
+		return true;
+
+	/* Exit early if we know that we're adding or removing the plane. */
+	if (old_plane_state->crtc != new_plane_state->crtc)
+		return true;
+
+	/* old crtc == new_crtc == NULL, plane not in context. */
+	if (!new_plane_state->crtc)
+		return false;
+
+	new_crtc_state =
+		drm_atomic_get_new_crtc_state(state, new_plane_state->crtc);
+	old_crtc_state =
+		drm_atomic_get_old_crtc_state(state, old_plane_state->crtc);
+
+	if (!new_crtc_state)
+		return true;
+
+	/*
+	 * A change in cursor mode means a new dc pipe needs to be acquired or
+	 * released from the state
+	 */
+	old_dm_crtc_state = to_dm_crtc_state(old_crtc_state);
+	new_dm_crtc_state = to_dm_crtc_state(new_crtc_state);
+	if (plane->type == DRM_PLANE_TYPE_CURSOR &&
+	    old_dm_crtc_state != NULL &&
+	    old_dm_crtc_state->cursor_mode != new_dm_crtc_state->cursor_mode) {
+		return true;
+	}
+
+	/* CRTC Degamma changes currently require us to recreate planes. */
+	if (new_crtc_state->color_mgmt_changed)
+		return true;
+
+	/*
+	 * On zpos change, planes need to be reordered by removing and re-adding
+	 * them one by one to the dc state, in order of descending zpos.
+	 *
+	 * TODO: We can likely skip bandwidth validation if the only thing that
+	 * changed about the plane was it'z z-ordering.
+	 */
+	if (old_plane_state->normalized_zpos != new_plane_state->normalized_zpos)
+		return true;
+
+	if (drm_atomic_crtc_needs_modeset(new_crtc_state))
+		return true;
+
+	/*
+	 * If there are any new primary or overlay planes being added or
+	 * removed then the z-order can potentially change. To ensure
+	 * correct z-order and pipe acquisition the current DC architecture
+	 * requires us to remove and recreate all existing planes.
+	 *
+	 * TODO: Come up with a more elegant solution for this.
+	 */
+	for_each_oldnew_plane_in_state(state, other, old_other_state, new_other_state, i) {
+		struct amdgpu_framebuffer *old_afb, *new_afb;
+		struct dm_plane_state *dm_new_other_state, *dm_old_other_state;
+
+		dm_new_other_state = to_dm_plane_state(new_other_state);
+		dm_old_other_state = to_dm_plane_state(old_other_state);
+
+		if (other->type == DRM_PLANE_TYPE_CURSOR)
+			continue;
+
+		if (old_other_state->crtc != new_plane_state->crtc &&
+		    new_other_state->crtc != new_plane_state->crtc)
+			continue;
+
+		if (old_other_state->crtc != new_other_state->crtc)
+			return true;
+
+		/* Src/dst size and scaling updates. */
+		if (old_other_state->src_w != new_other_state->src_w ||
+		    old_other_state->src_h != new_other_state->src_h ||
+		    old_other_state->crtc_w != new_other_state->crtc_w ||
+		    old_other_state->crtc_h != new_other_state->crtc_h)
+			return true;
+
+		/* Rotation / mirroring updates. */
+		if (old_other_state->rotation != new_other_state->rotation)
+			return true;
+
+		/* Blending updates. */
+		if (old_other_state->pixel_blend_mode !=
+		    new_other_state->pixel_blend_mode)
+			return true;
+
+		/* Alpha updates. */
+		if (old_other_state->alpha != new_other_state->alpha)
+			return true;
+
+		/* Colorspace changes. */
+		if (old_other_state->color_range != new_other_state->color_range ||
+		    old_other_state->color_encoding != new_other_state->color_encoding)
+			return true;
+
+		/* HDR/Transfer Function changes. */
+		if (dm_old_other_state->degamma_tf != dm_new_other_state->degamma_tf ||
+		    dm_old_other_state->degamma_lut != dm_new_other_state->degamma_lut ||
+		    dm_old_other_state->hdr_mult != dm_new_other_state->hdr_mult ||
+		    dm_old_other_state->ctm != dm_new_other_state->ctm ||
+		    dm_old_other_state->shaper_lut != dm_new_other_state->shaper_lut ||
+		    dm_old_other_state->shaper_tf != dm_new_other_state->shaper_tf ||
+		    dm_old_other_state->lut3d != dm_new_other_state->lut3d ||
+		    dm_old_other_state->blend_lut != dm_new_other_state->blend_lut ||
+		    dm_old_other_state->blend_tf != dm_new_other_state->blend_tf)
+			return true;
+
+		/* Framebuffer checks fall at the end. */
+		if (!old_other_state->fb || !new_other_state->fb)
+			continue;
+
+		/* Pixel format changes can require bandwidth updates. */
+		if (old_other_state->fb->format != new_other_state->fb->format)
+			return true;
+
+		old_afb = (struct amdgpu_framebuffer *)old_other_state->fb;
+		new_afb = (struct amdgpu_framebuffer *)new_other_state->fb;
+
+		/* Tiling and DCC changes also require bandwidth updates. */
+		if (old_afb->tiling_flags != new_afb->tiling_flags ||
+		    old_afb->base.modifier != new_afb->base.modifier)
+			return true;
+	}
+
+	return false;
+}
+
+static int dm_check_cursor_fb(struct amdgpu_crtc *new_acrtc,
+			      struct drm_plane_state *new_plane_state,
+			      struct drm_framebuffer *fb)
+{
+	struct amdgpu_device *adev = drm_to_adev(new_acrtc->base.dev);
+	struct amdgpu_framebuffer *afb = to_amdgpu_framebuffer(fb);
+	unsigned int pitch;
+	bool linear;
+
+	if (fb->width > new_acrtc->max_cursor_width ||
+	    fb->height > new_acrtc->max_cursor_height) {
+		DRM_DEBUG_ATOMIC("Bad cursor FB size %dx%d\n",
+				 new_plane_state->fb->width,
+				 new_plane_state->fb->height);
+		return -EINVAL;
+	}
+	if (new_plane_state->src_w != fb->width << 16 ||
+	    new_plane_state->src_h != fb->height << 16) {
+		DRM_DEBUG_ATOMIC("Cropping not supported for cursor plane\n");
+		return -EINVAL;
+	}
+
+	/* Pitch in pixels */
+	pitch = fb->pitches[0] / fb->format->cpp[0];
+
+	if (fb->width != pitch) {
+		DRM_DEBUG_ATOMIC("Cursor FB width %d doesn't match pitch %d",
+				 fb->width, pitch);
+		return -EINVAL;
+	}
+
+	switch (pitch) {
+	case 64:
+	case 128:
+	case 256:
+		/* FB pitch is supported by cursor plane */
+		break;
+	default:
+		DRM_DEBUG_ATOMIC("Bad cursor FB pitch %d px\n", pitch);
+		return -EINVAL;
+	}
+
+	/* Core DRM takes care of checking FB modifiers, so we only need to
+	 * check tiling flags when the FB doesn't have a modifier.
+	 */
+	if (!(fb->flags & DRM_MODE_FB_MODIFIERS)) {
+		if (adev->family >= AMDGPU_FAMILY_GC_12_0_0) {
+			linear = AMDGPU_TILING_GET(afb->tiling_flags, GFX12_SWIZZLE_MODE) == 0;
+		} else if (adev->family >= AMDGPU_FAMILY_AI) {
+			linear = AMDGPU_TILING_GET(afb->tiling_flags, SWIZZLE_MODE) == 0;
+		} else {
+			linear = AMDGPU_TILING_GET(afb->tiling_flags, ARRAY_MODE) != DC_ARRAY_2D_TILED_THIN1 &&
+				 AMDGPU_TILING_GET(afb->tiling_flags, ARRAY_MODE) != DC_ARRAY_1D_TILED_THIN1 &&
+				 AMDGPU_TILING_GET(afb->tiling_flags, MICRO_TILE_MODE) == 0;
+		}
+		if (!linear) {
+			DRM_DEBUG_ATOMIC("Cursor FB not linear");
+			return -EINVAL;
+		}
+	}
+
+	return 0;
+}
+
+/*
+ * Helper function for checking the cursor in native mode
+ */
+static int dm_check_native_cursor_state(struct drm_crtc *new_plane_crtc,
+					struct drm_plane *plane,
+					struct drm_plane_state *new_plane_state,
+					bool enable)
+{
+
+	struct amdgpu_crtc *new_acrtc;
+	int ret;
+
+	if (!enable || !new_plane_crtc ||
+	    drm_atomic_plane_disabling(plane->state, new_plane_state))
+		return 0;
+
+	new_acrtc = to_amdgpu_crtc(new_plane_crtc);
+
+	if (new_plane_state->src_x != 0 || new_plane_state->src_y != 0) {
+		DRM_DEBUG_ATOMIC("Cropping not supported for cursor plane\n");
+		return -EINVAL;
+	}
+
+	if (new_plane_state->fb) {
+		ret = dm_check_cursor_fb(new_acrtc, new_plane_state,
+						new_plane_state->fb);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static bool dm_should_update_native_cursor(struct drm_atomic_state *state,
+					   struct drm_crtc *old_plane_crtc,
+					   struct drm_crtc *new_plane_crtc,
+					   bool enable)
+{
+	struct drm_crtc_state *old_crtc_state, *new_crtc_state;
+	struct dm_crtc_state *dm_old_crtc_state, *dm_new_crtc_state;
+
+	if (!enable) {
+		if (old_plane_crtc == NULL)
+			return true;
+
+		old_crtc_state = drm_atomic_get_old_crtc_state(
+			state, old_plane_crtc);
+		dm_old_crtc_state = to_dm_crtc_state(old_crtc_state);
+
+		return dm_old_crtc_state->cursor_mode == DM_CURSOR_NATIVE_MODE;
+	} else {
+		if (new_plane_crtc == NULL)
+			return true;
+
+		new_crtc_state = drm_atomic_get_new_crtc_state(
+			state, new_plane_crtc);
+		dm_new_crtc_state = to_dm_crtc_state(new_crtc_state);
+
+		return dm_new_crtc_state->cursor_mode == DM_CURSOR_NATIVE_MODE;
+	}
+}
+
+static int dm_update_plane_state(struct dc *dc,
+				 struct drm_atomic_state *state,
+				 struct drm_plane *plane,
+				 struct drm_plane_state *old_plane_state,
+				 struct drm_plane_state *new_plane_state,
+				 bool enable,
+				 bool *lock_and_validation_needed,
+				 bool *is_top_most_overlay)
+{
+
+	struct dm_atomic_state *dm_state = NULL;
+	struct drm_crtc *new_plane_crtc, *old_plane_crtc;
+	struct drm_crtc_state *old_crtc_state, *new_crtc_state;
+	struct dm_crtc_state *dm_new_crtc_state, *dm_old_crtc_state;
+	struct dm_plane_state *dm_new_plane_state, *dm_old_plane_state;
+	bool needs_reset, update_native_cursor;
+	int ret = 0;
+
+
+	new_plane_crtc = new_plane_state->crtc;
+	old_plane_crtc = old_plane_state->crtc;
+	dm_new_plane_state = to_dm_plane_state(new_plane_state);
+	dm_old_plane_state = to_dm_plane_state(old_plane_state);
+
+	update_native_cursor = dm_should_update_native_cursor(state,
+							      old_plane_crtc,
+							      new_plane_crtc,
+							      enable);
+
+	if (plane->type == DRM_PLANE_TYPE_CURSOR && update_native_cursor) {
+		ret = dm_check_native_cursor_state(new_plane_crtc, plane,
+						    new_plane_state, enable);
+		if (ret)
+			return ret;
+
+		return 0;
+	}
+
+	needs_reset = should_reset_plane(state, plane, old_plane_state,
+					 new_plane_state);
+
+	/* Remove any changed/removed planes */
+	if (!enable) {
+		if (!needs_reset)
+			return 0;
+
+		if (!old_plane_crtc)
+			return 0;
+
+		old_crtc_state = drm_atomic_get_old_crtc_state(
+				state, old_plane_crtc);
+		dm_old_crtc_state = to_dm_crtc_state(old_crtc_state);
+
+		if (!dm_old_crtc_state->stream)
+			return 0;
+
+		DRM_DEBUG_ATOMIC("Disabling DRM plane: %d on DRM crtc %d\n",
+				plane->base.id, old_plane_crtc->base.id);
+
+		ret = dm_atomic_get_state(state, &dm_state);
+		if (ret)
+			return ret;
+
+		if (!dc_state_remove_plane(
+				dc,
+				dm_old_crtc_state->stream,
+				dm_old_plane_state->dc_state,
+				dm_state->context)) {
+
+			return -EINVAL;
+		}
+
+		if (dm_old_plane_state->dc_state)
+			dc_plane_state_release(dm_old_plane_state->dc_state);
+
+		dm_new_plane_state->dc_state = NULL;
+
+		*lock_and_validation_needed = true;
+
+	} else { /* Add new planes */
+		struct dc_plane_state *dc_new_plane_state;
+
+		if (drm_atomic_plane_disabling(plane->state, new_plane_state))
+			return 0;
+
+		if (!new_plane_crtc)
+			return 0;
+
+		new_crtc_state = drm_atomic_get_new_crtc_state(state, new_plane_crtc);
+		dm_new_crtc_state = to_dm_crtc_state(new_crtc_state);
+
+		if (!dm_new_crtc_state->stream)
+			return 0;
+
+		if (!needs_reset)
+			return 0;
+
+		ret = amdgpu_dm_plane_helper_check_state(new_plane_state, new_crtc_state);
+		if (ret)
+			goto out;
+
+		WARN_ON(dm_new_plane_state->dc_state);
+
+		dc_new_plane_state = dc_create_plane_state(dc);
+		if (!dc_new_plane_state) {
+			ret = -ENOMEM;
+			goto out;
+		}
+
+		DRM_DEBUG_ATOMIC("Enabling DRM plane: %d on DRM crtc %d\n",
+				 plane->base.id, new_plane_crtc->base.id);
+
+		ret = fill_dc_plane_attributes(
+			drm_to_adev(new_plane_crtc->dev),
+			dc_new_plane_state,
+			new_plane_state,
+			new_crtc_state);
+		if (ret) {
+			dc_plane_state_release(dc_new_plane_state);
+			goto out;
+		}
+
+		ret = dm_atomic_get_state(state, &dm_state);
+		if (ret) {
+			dc_plane_state_release(dc_new_plane_state);
+			goto out;
+		}
+
+		/*
+		 * Any atomic check errors that occur after this will
+		 * not need a release. The plane state will be attached
+		 * to the stream, and therefore part of the atomic
+		 * state. It'll be released when the atomic state is
+		 * cleaned.
+		 */
+		if (!dc_state_add_plane(
+				dc,
+				dm_new_crtc_state->stream,
+				dc_new_plane_state,
+				dm_state->context)) {
+
+			dc_plane_state_release(dc_new_plane_state);
+			ret = -EINVAL;
+			goto out;
+		}
+
+		dm_new_plane_state->dc_state = dc_new_plane_state;
+
+		dm_new_crtc_state->mpo_requested |= (plane->type == DRM_PLANE_TYPE_OVERLAY);
+
+		/* Tell DC to do a full surface update every time there
+		 * is a plane change. Inefficient, but works for now.
+		 */
+		dm_new_plane_state->dc_state->update_flags.bits.full_update = 1;
+
+		*lock_and_validation_needed = true;
+	}
+
+out:
+	/* If enabling cursor overlay failed, attempt fallback to native mode */
+	if (enable && ret == -EINVAL && plane->type == DRM_PLANE_TYPE_CURSOR) {
+		ret = dm_check_native_cursor_state(new_plane_crtc, plane,
+						    new_plane_state, enable);
+		if (ret)
+			return ret;
+
+		dm_new_crtc_state->cursor_mode = DM_CURSOR_NATIVE_MODE;
+	}
+
+	return ret;
+}
+
+static void dm_get_oriented_plane_size(struct drm_plane_state *plane_state,
+				       int *src_w, int *src_h)
+{
+	switch (plane_state->rotation & DRM_MODE_ROTATE_MASK) {
+	case DRM_MODE_ROTATE_90:
+	case DRM_MODE_ROTATE_270:
+		*src_w = plane_state->src_h >> 16;
+		*src_h = plane_state->src_w >> 16;
+		break;
+	case DRM_MODE_ROTATE_0:
+	case DRM_MODE_ROTATE_180:
+	default:
+		*src_w = plane_state->src_w >> 16;
+		*src_h = plane_state->src_h >> 16;
+		break;
+	}
+}
+
+static void
+dm_get_plane_scale(struct drm_plane_state *plane_state,
+		   int *out_plane_scale_w, int *out_plane_scale_h)
+{
+	int plane_src_w, plane_src_h;
+
+	dm_get_oriented_plane_size(plane_state, &plane_src_w, &plane_src_h);
+	*out_plane_scale_w = plane_src_w ? plane_state->crtc_w * 1000 / plane_src_w : 0;
+	*out_plane_scale_h = plane_src_h ? plane_state->crtc_h * 1000 / plane_src_h : 0;
+}
+
+/*
+ * The normalized_zpos value cannot be used by this iterator directly. It's only
+ * calculated for enabled planes, potentially causing normalized_zpos collisions
+ * between enabled/disabled planes in the atomic state. We need a unique value
+ * so that the iterator will not generate the same object twice, or loop
+ * indefinitely.
+ */
+static inline struct __drm_planes_state *__get_next_zpos(
+	struct drm_atomic_state *state,
+	struct __drm_planes_state *prev)
+{
+	unsigned int highest_zpos = 0, prev_zpos = 256;
+	uint32_t highest_id = 0, prev_id = UINT_MAX;
+	struct drm_plane_state *new_plane_state;
+	struct drm_plane *plane;
+	int i, highest_i = -1;
+
+	if (prev != NULL) {
+		prev_zpos = prev->new_state->zpos;
+		prev_id = prev->ptr->base.id;
+	}
+
+	for_each_new_plane_in_state(state, plane, new_plane_state, i) {
+		/* Skip planes with higher zpos than the previously returned */
+		if (new_plane_state->zpos > prev_zpos ||
+		    (new_plane_state->zpos == prev_zpos &&
+		     plane->base.id >= prev_id))
+			continue;
+
+		/* Save the index of the plane with highest zpos */
+		if (new_plane_state->zpos > highest_zpos ||
+		    (new_plane_state->zpos == highest_zpos &&
+		     plane->base.id > highest_id)) {
+			highest_zpos = new_plane_state->zpos;
+			highest_id = plane->base.id;
+			highest_i = i;
+		}
+	}
+
+	if (highest_i < 0)
+		return NULL;
+
+	return &state->planes[highest_i];
+}
+
+/*
+ * Use the uniqueness of the plane's (zpos, drm obj ID) combination to iterate
+ * by descending zpos, as read from the new plane state. This is the same
+ * ordering as defined by drm_atomic_normalize_zpos().
+ */
+#define for_each_oldnew_plane_in_descending_zpos(__state, plane, old_plane_state, new_plane_state) \
+	for (struct __drm_planes_state *__i = __get_next_zpos((__state), NULL); \
+	     __i != NULL; __i = __get_next_zpos((__state), __i))		\
+		for_each_if(((plane) = __i->ptr,				\
+			     (void)(plane) /* Only to avoid unused-but-set-variable warning */, \
+			     (old_plane_state) = __i->old_state,		\
+			     (new_plane_state) = __i->new_state, 1))
+
+static int add_affected_mst_dsc_crtcs(struct drm_atomic_state *state, struct drm_crtc *crtc)
+{
+	struct drm_connector *connector;
+	struct drm_connector_state *conn_state, *old_conn_state;
+	struct amdgpu_dm_connector *aconnector = NULL;
+	int i;
+
+	for_each_oldnew_connector_in_state(state, connector, old_conn_state, conn_state, i) {
+		if (!conn_state->crtc)
+			conn_state = old_conn_state;
+
+		if (conn_state->crtc != crtc)
+			continue;
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_WRITEBACK)
+			continue;
+
+		aconnector = to_amdgpu_dm_connector(connector);
+		if (!aconnector->mst_output_port || !aconnector->mst_root)
+			aconnector = NULL;
+		else
+			break;
+	}
+
+	if (!aconnector)
+		return 0;
+
+	return drm_dp_mst_add_affected_dsc_crtcs(state, &aconnector->mst_root->mst_mgr);
+}
+
+/**
+ * DOC: Cursor Modes - Native vs Overlay
+ *
+ * In native mode, the cursor uses a integrated cursor pipe within each DCN hw
+ * plane. It does not require a dedicated hw plane to enable, but it is
+ * subjected to the same z-order and scaling as the hw plane. It also has format
+ * restrictions, a RGB cursor in native mode cannot be enabled within a non-RGB
+ * hw plane.
+ *
+ * In overlay mode, the cursor uses a separate DCN hw plane, and thus has its
+ * own scaling and z-pos. It also has no blending restrictions. It lends to a
+ * cursor behavior more akin to a DRM client's expectations. However, it does
+ * occupy an extra DCN plane, and therefore will only be used if a DCN plane is
+ * available.
+ */
+
+/**
+ * dm_crtc_get_cursor_mode() - Determine the required cursor mode on crtc
+ * @adev: amdgpu device
+ * @state: DRM atomic state
+ * @dm_crtc_state: amdgpu state for the CRTC containing the cursor
+ * @cursor_mode: Returns the required cursor mode on dm_crtc_state
+ *
+ * Get whether the cursor should be enabled in native mode, or overlay mode, on
+ * the dm_crtc_state.
+ *
+ * The cursor should be enabled in overlay mode if there exists an underlying
+ * plane - on which the cursor may be blended - that is either YUV formatted, or
+ * scaled differently from the cursor.
+ *
+ * Since zpos info is required, drm_atomic_normalize_zpos must be called before
+ * calling this function.
+ *
+ * Return: 0 on success, or an error code if getting the cursor plane state
+ * failed.
+ */
+static int dm_crtc_get_cursor_mode(struct amdgpu_device *adev,
+				   struct drm_atomic_state *state,
+				   struct dm_crtc_state *dm_crtc_state,
+				   enum amdgpu_dm_cursor_mode *cursor_mode)
+{
+	struct drm_plane_state *old_plane_state, *plane_state, *cursor_state;
+	struct drm_crtc_state *crtc_state = &dm_crtc_state->base;
+	struct drm_plane *plane;
+	bool consider_mode_change = false;
+	bool entire_crtc_covered = false;
+	bool cursor_changed = false;
+	int underlying_scale_w, underlying_scale_h;
+	int cursor_scale_w, cursor_scale_h;
+	int i;
+
+	/* Overlay cursor not supported on HW before DCN
+	 * DCN401 does not have the cursor-on-scaled-plane or cursor-on-yuv-plane restrictions
+	 * as previous DCN generations, so enable native mode on DCN401 in addition to DCE
+	 */
+	if (amdgpu_ip_version(adev, DCE_HWIP, 0) == 0 ||
+	    amdgpu_ip_version(adev, DCE_HWIP, 0) == IP_VERSION(4, 0, 1)) {
+		*cursor_mode = DM_CURSOR_NATIVE_MODE;
+		return 0;
+	}
+
+	/* Init cursor_mode to be the same as current */
+	*cursor_mode = dm_crtc_state->cursor_mode;
+
+	/*
+	 * Cursor mode can change if a plane's format changes, scale changes, is
+	 * enabled/disabled, or z-order changes.
+	 */
+	for_each_oldnew_plane_in_state(state, plane, old_plane_state, plane_state, i) {
+		int new_scale_w, new_scale_h, old_scale_w, old_scale_h;
+
+		/* Only care about planes on this CRTC */
+		if ((drm_plane_mask(plane) & crtc_state->plane_mask) == 0)
+			continue;
+
+		if (plane->type == DRM_PLANE_TYPE_CURSOR)
+			cursor_changed = true;
+
+		if (drm_atomic_plane_enabling(old_plane_state, plane_state) ||
+		    drm_atomic_plane_disabling(old_plane_state, plane_state) ||
+		    old_plane_state->fb->format != plane_state->fb->format) {
+			consider_mode_change = true;
+			break;
+		}
+
+		dm_get_plane_scale(plane_state, &new_scale_w, &new_scale_h);
+		dm_get_plane_scale(old_plane_state, &old_scale_w, &old_scale_h);
+		if (new_scale_w != old_scale_w || new_scale_h != old_scale_h) {
+			consider_mode_change = true;
+			break;
+		}
+	}
+
+	if (!consider_mode_change && !crtc_state->zpos_changed)
+		return 0;
+
+	/*
+	 * If no cursor change on this CRTC, and not enabled on this CRTC, then
+	 * no need to set cursor mode. This avoids needlessly locking the cursor
+	 * state.
+	 */
+	if (!cursor_changed &&
+	    !(drm_plane_mask(crtc_state->crtc->cursor) & crtc_state->plane_mask)) {
+		return 0;
+	}
+
+	cursor_state = drm_atomic_get_plane_state(state,
+						  crtc_state->crtc->cursor);
+	if (IS_ERR(cursor_state))
+		return PTR_ERR(cursor_state);
+
+	/* Cursor is disabled */
+	if (!cursor_state->fb)
+		return 0;
+
+	/* For all planes in descending z-order (all of which are below cursor
+	 * as per zpos definitions), check their scaling and format
+	 */
+	for_each_oldnew_plane_in_descending_zpos(state, plane, old_plane_state, plane_state) {
+
+		/* Only care about non-cursor planes on this CRTC */
+		if ((drm_plane_mask(plane) & crtc_state->plane_mask) == 0 ||
+		    plane->type == DRM_PLANE_TYPE_CURSOR)
+			continue;
+
+		/* Underlying plane is YUV format - use overlay cursor */
+		if (amdgpu_dm_plane_is_video_format(plane_state->fb->format->format)) {
+			*cursor_mode = DM_CURSOR_OVERLAY_MODE;
+			return 0;
+		}
+
+		dm_get_plane_scale(plane_state,
+				   &underlying_scale_w, &underlying_scale_h);
+		dm_get_plane_scale(cursor_state,
+				   &cursor_scale_w, &cursor_scale_h);
+
+		/* Underlying plane has different scale - use overlay cursor */
+		if (cursor_scale_w != underlying_scale_w &&
+		    cursor_scale_h != underlying_scale_h) {
+			*cursor_mode = DM_CURSOR_OVERLAY_MODE;
+			return 0;
+		}
+
+		/* If this plane covers the whole CRTC, no need to check planes underneath */
+		if (plane_state->crtc_x <= 0 && plane_state->crtc_y <= 0 &&
+		    plane_state->crtc_x + plane_state->crtc_w >= crtc_state->mode.hdisplay &&
+		    plane_state->crtc_y + plane_state->crtc_h >= crtc_state->mode.vdisplay) {
+			entire_crtc_covered = true;
+			break;
+		}
+	}
+
+	/* If planes do not cover the entire CRTC, use overlay mode to enable
+	 * cursor over holes
+	 */
+	if (entire_crtc_covered)
+		*cursor_mode = DM_CURSOR_NATIVE_MODE;
+	else
+		*cursor_mode = DM_CURSOR_OVERLAY_MODE;
+
+	return 0;
+}
+
+static bool amdgpu_dm_crtc_mem_type_changed(struct drm_device *dev,
+					    struct drm_atomic_state *state,
+					    struct drm_crtc_state *crtc_state)
+{
+	struct drm_plane *plane;
+	struct drm_plane_state *new_plane_state, *old_plane_state;
+
+	drm_for_each_plane_mask(plane, dev, crtc_state->plane_mask) {
+		new_plane_state = drm_atomic_get_plane_state(state, plane);
+		old_plane_state = drm_atomic_get_plane_state(state, plane);
+
+		if (IS_ERR(new_plane_state) || IS_ERR(old_plane_state)) {
+			drm_err(dev, "Failed to get plane state for plane %s\n", plane->name);
+			return false;
+		}
+
+		if (old_plane_state->fb && new_plane_state->fb &&
+		    get_mem_type(old_plane_state->fb) != get_mem_type(new_plane_state->fb))
+			return true;
+	}
+
+	return false;
+}
+
+/**
+ * amdgpu_dm_atomic_check() - Atomic check implementation for AMDgpu DM.
+ *
+ * @dev: The DRM device
+ * @state: The atomic state to commit
+ *
+ * Validate that the given atomic state is programmable by DC into hardware.
+ * This involves constructing a &struct dc_state reflecting the new hardware
+ * state we wish to commit, then querying DC to see if it is programmable. It's
+ * important not to modify the existing DC state. Otherwise, atomic_check
+ * may unexpectedly commit hardware changes.
+ *
+ * When validating the DC state, it's important that the right locks are
+ * acquired. For full updates case which removes/adds/updates streams on one
+ * CRTC while flipping on another CRTC, acquiring global lock will guarantee
+ * that any such full update commit will wait for completion of any outstanding
+ * flip using DRMs synchronization events.
+ *
+ * Note that DM adds the affected connectors for all CRTCs in state, when that
+ * might not seem necessary. This is because DC stream creation requires the
+ * DC sink, which is tied to the DRM connector state. Cleaning this up should
+ * be possible but non-trivial - a possible TODO item.
+ *
+ * Return: -Error code if validation failed.
+ */
+static int amdgpu_dm_atomic_check(struct drm_device *dev,
+				  struct drm_atomic_state *state)
+{
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	struct dm_atomic_state *dm_state = NULL;
+	struct dc *dc = adev->dm.dc;
+	struct drm_connector *connector;
+	struct drm_connector_state *old_con_state, *new_con_state;
+	struct drm_crtc *crtc;
+	struct drm_crtc_state *old_crtc_state, *new_crtc_state;
+	struct drm_plane *plane;
+	struct drm_plane_state *old_plane_state, *new_plane_state, *new_cursor_state;
+	enum dc_status status;
+	int ret, i;
+	bool lock_and_validation_needed = false;
+	bool is_top_most_overlay = true;
+	struct dm_crtc_state *dm_old_crtc_state, *dm_new_crtc_state;
+	struct drm_dp_mst_topology_mgr *mgr;
+	struct drm_dp_mst_topology_state *mst_state;
+	struct dsc_mst_fairness_vars vars[MAX_PIPES] = {0};
+
+	trace_amdgpu_dm_atomic_check_begin(state);
+
+	ret = drm_atomic_helper_check_modeset(dev, state);
+	if (ret) {
+		drm_dbg_atomic(dev, "drm_atomic_helper_check_modeset() failed\n");
+		goto fail;
+	}
+
+	/* Check connector changes */
+	for_each_oldnew_connector_in_state(state, connector, old_con_state, new_con_state, i) {
+		struct dm_connector_state *dm_old_con_state = to_dm_connector_state(old_con_state);
+		struct dm_connector_state *dm_new_con_state = to_dm_connector_state(new_con_state);
+
+		/* Skip connectors that are disabled or part of modeset already. */
+		if (!new_con_state->crtc)
+			continue;
+
+		new_crtc_state = drm_atomic_get_crtc_state(state, new_con_state->crtc);
+		if (IS_ERR(new_crtc_state)) {
+			drm_dbg_atomic(dev, "drm_atomic_get_crtc_state() failed\n");
+			ret = PTR_ERR(new_crtc_state);
+			goto fail;
+		}
+
+		if (dm_old_con_state->abm_level != dm_new_con_state->abm_level ||
+		    dm_old_con_state->scaling != dm_new_con_state->scaling)
+			new_crtc_state->connectors_changed = true;
+	}
+
+	if (dc_resource_is_dsc_encoding_supported(dc)) {
+		for_each_oldnew_crtc_in_state(state, crtc, old_crtc_state, new_crtc_state, i) {
+			if (drm_atomic_crtc_needs_modeset(new_crtc_state)) {
+				ret = add_affected_mst_dsc_crtcs(state, crtc);
+				if (ret) {
+					drm_dbg_atomic(dev, "add_affected_mst_dsc_crtcs() failed\n");
+					goto fail;
+				}
+			}
+		}
+	}
+	for_each_oldnew_crtc_in_state(state, crtc, old_crtc_state, new_crtc_state, i) {
+		dm_old_crtc_state = to_dm_crtc_state(old_crtc_state);
+
+		if (!drm_atomic_crtc_needs_modeset(new_crtc_state) &&
+		    !new_crtc_state->color_mgmt_changed &&
+		    old_crtc_state->vrr_enabled == new_crtc_state->vrr_enabled &&
+			dm_old_crtc_state->dsc_force_changed == false)
+			continue;
+
+		ret = amdgpu_dm_verify_lut_sizes(new_crtc_state);
+		if (ret) {
+			drm_dbg_atomic(dev, "amdgpu_dm_verify_lut_sizes() failed\n");
+			goto fail;
+		}
+
+		if (!new_crtc_state->enable)
+			continue;
+
+		ret = drm_atomic_add_affected_connectors(state, crtc);
+		if (ret) {
+			drm_dbg_atomic(dev, "drm_atomic_add_affected_connectors() failed\n");
+			goto fail;
+		}
+
+		ret = drm_atomic_add_affected_planes(state, crtc);
+		if (ret) {
+			drm_dbg_atomic(dev, "drm_atomic_add_affected_planes() failed\n");
+			goto fail;
+		}
+
+		if (dm_old_crtc_state->dsc_force_changed)
+			new_crtc_state->mode_changed = true;
+	}
+
+	/*
+	 * Add all primary and overlay planes on the CRTC to the state
+	 * whenever a plane is enabled to maintain correct z-ordering
+	 * and to enable fast surface updates.
+	 */
+	drm_for_each_crtc(crtc, dev) {
+		bool modified = false;
+
+		for_each_oldnew_plane_in_state(state, plane, old_plane_state, new_plane_state, i) {
+			if (plane->type == DRM_PLANE_TYPE_CURSOR)
+				continue;
+
+			if (new_plane_state->crtc == crtc ||
+			    old_plane_state->crtc == crtc) {
+				modified = true;
+				break;
+			}
+		}
+
+		if (!modified)
+			continue;
+
+		drm_for_each_plane_mask(plane, state->dev, crtc->state->plane_mask) {
+			if (plane->type == DRM_PLANE_TYPE_CURSOR)
+				continue;
+
+			new_plane_state =
+				drm_atomic_get_plane_state(state, plane);
+
+			if (IS_ERR(new_plane_state)) {
+				ret = PTR_ERR(new_plane_state);
+				drm_dbg_atomic(dev, "new_plane_state is BAD\n");
+				goto fail;
+			}
+		}
+	}
+
+	/*
+	 * DC consults the zpos (layer_index in DC terminology) to determine the
+	 * hw plane on which to enable the hw cursor (see
+	 * `dcn10_can_pipe_disable_cursor`). By now, all modified planes are in
+	 * atomic state, so call drm helper to normalize zpos.
+	 */
+	ret = drm_atomic_normalize_zpos(dev, state);
+	if (ret) {
+		drm_dbg(dev, "drm_atomic_normalize_zpos() failed\n");
+		goto fail;
+	}
+
+	/*
+	 * Determine whether cursors on each CRTC should be enabled in native or
+	 * overlay mode.
+	 */
+	for_each_new_crtc_in_state(state, crtc, new_crtc_state, i) {
+		dm_new_crtc_state = to_dm_crtc_state(new_crtc_state);
+
+		ret = dm_crtc_get_cursor_mode(adev, state, dm_new_crtc_state,
+					      &dm_new_crtc_state->cursor_mode);
+		if (ret) {
+			drm_dbg(dev, "Failed to determine cursor mode\n");
+			goto fail;
+		}
+
+		/*
+		 * If overlay cursor is needed, DC cannot go through the
+		 * native cursor update path. All enabled planes on the CRTC
+		 * need to be added for DC to not disable a plane by mistake
+		 */
+		if (dm_new_crtc_state->cursor_mode == DM_CURSOR_OVERLAY_MODE) {
+			ret = drm_atomic_add_affected_planes(state, crtc);
+			if (ret)
+				goto fail;
+		}
+	}
+
+	/* Remove exiting planes if they are modified */
+	for_each_oldnew_plane_in_descending_zpos(state, plane, old_plane_state, new_plane_state) {
+
+		ret = dm_update_plane_state(dc, state, plane,
+					    old_plane_state,
+					    new_plane_state,
+					    false,
+					    &lock_and_validation_needed,
+					    &is_top_most_overlay);
+		if (ret) {
+			drm_dbg_atomic(dev, "dm_update_plane_state() failed\n");
+			goto fail;
+		}
+	}
+
+	/* Disable all crtcs which require disable */
+	for_each_oldnew_crtc_in_state(state, crtc, old_crtc_state, new_crtc_state, i) {
+		ret = dm_update_crtc_state(&adev->dm, state, crtc,
+					   old_crtc_state,
+					   new_crtc_state,
+					   false,
+					   &lock_and_validation_needed);
+		if (ret) {
+			drm_dbg_atomic(dev, "DISABLE: dm_update_crtc_state() failed\n");
+			goto fail;
+		}
+	}
+
+	/* Enable all crtcs which require enable */
+	for_each_oldnew_crtc_in_state(state, crtc, old_crtc_state, new_crtc_state, i) {
+		ret = dm_update_crtc_state(&adev->dm, state, crtc,
+					   old_crtc_state,
+					   new_crtc_state,
+					   true,
+					   &lock_and_validation_needed);
+		if (ret) {
+			drm_dbg_atomic(dev, "ENABLE: dm_update_crtc_state() failed\n");
+			goto fail;
+		}
+	}
+
+	/* Add new/modified planes */
+	for_each_oldnew_plane_in_descending_zpos(state, plane, old_plane_state, new_plane_state) {
+		ret = dm_update_plane_state(dc, state, plane,
+					    old_plane_state,
+					    new_plane_state,
+					    true,
+					    &lock_and_validation_needed,
+					    &is_top_most_overlay);
+		if (ret) {
+			drm_dbg_atomic(dev, "dm_update_plane_state() failed\n");
+			goto fail;
+		}
+	}
+
+#if defined(CONFIG_DRM_AMD_DC_FP)
+	if (dc_resource_is_dsc_encoding_supported(dc)) {
+		ret = pre_validate_dsc(state, &dm_state, vars);
+		if (ret != 0)
+			goto fail;
+	}
+#endif
+
+	/* Run this here since we want to validate the streams we created */
+	ret = drm_atomic_helper_check_planes(dev, state);
+	if (ret) {
+		drm_dbg_atomic(dev, "drm_atomic_helper_check_planes() failed\n");
+		goto fail;
+	}
+
+	for_each_new_crtc_in_state(state, crtc, new_crtc_state, i) {
+		dm_new_crtc_state = to_dm_crtc_state(new_crtc_state);
+		if (dm_new_crtc_state->mpo_requested)
+			drm_dbg_atomic(dev, "MPO enablement requested on crtc:[%p]\n", crtc);
+	}
+
+	/* Check cursor restrictions */
+	for_each_new_crtc_in_state(state, crtc, new_crtc_state, i) {
+		enum amdgpu_dm_cursor_mode required_cursor_mode;
+		int is_rotated, is_scaled;
+
+		/* Overlay cusor not subject to native cursor restrictions */
+		dm_new_crtc_state = to_dm_crtc_state(new_crtc_state);
+		if (dm_new_crtc_state->cursor_mode == DM_CURSOR_OVERLAY_MODE)
+			continue;
+
+		/* Check if rotation or scaling is enabled on DCN401 */
+		if ((drm_plane_mask(crtc->cursor) & new_crtc_state->plane_mask) &&
+		    amdgpu_ip_version(adev, DCE_HWIP, 0) == IP_VERSION(4, 0, 1)) {
+			new_cursor_state = drm_atomic_get_new_plane_state(state, crtc->cursor);
+
+			is_rotated = new_cursor_state &&
+				((new_cursor_state->rotation & DRM_MODE_ROTATE_MASK) != DRM_MODE_ROTATE_0);
+			is_scaled = new_cursor_state && ((new_cursor_state->src_w >> 16 != new_cursor_state->crtc_w) ||
+				(new_cursor_state->src_h >> 16 != new_cursor_state->crtc_h));
+
+			if (is_rotated || is_scaled) {
+				drm_dbg_driver(
+					crtc->dev,
+					"[CRTC:%d:%s] cannot enable hardware cursor due to rotation/scaling\n",
+					crtc->base.id, crtc->name);
+				ret = -EINVAL;
+				goto fail;
+			}
+		}
+
+		/* If HW can only do native cursor, check restrictions again */
+		ret = dm_crtc_get_cursor_mode(adev, state, dm_new_crtc_state,
+					      &required_cursor_mode);
+		if (ret) {
+			drm_dbg_driver(crtc->dev,
+				       "[CRTC:%d:%s] Checking cursor mode failed\n",
+				       crtc->base.id, crtc->name);
+			goto fail;
+		} else if (required_cursor_mode == DM_CURSOR_OVERLAY_MODE) {
+			drm_dbg_driver(crtc->dev,
+				       "[CRTC:%d:%s] Cannot enable native cursor due to scaling or YUV restrictions\n",
+				       crtc->base.id, crtc->name);
+			ret = -EINVAL;
+			goto fail;
+		}
+	}
+
+	if (state->legacy_cursor_update) {
+		/*
+		 * This is a fast cursor update coming from the plane update
+		 * helper, check if it can be done asynchronously for better
+		 * performance.
+		 */
+		state->async_update =
+			!drm_atomic_helper_async_check(dev, state);
+
+		/*
+		 * Skip the remaining global validation if this is an async
+		 * update. Cursor updates can be done without affecting
+		 * state or bandwidth calcs and this avoids the performance
+		 * penalty of locking the private state object and
+		 * allocating a new dc_state.
+		 */
+		if (state->async_update)
+			return 0;
+	}
+
+	/* Check scaling and underscan changes*/
+	/* TODO Removed scaling changes validation due to inability to commit
+	 * new stream into context w\o causing full reset. Need to
+	 * decide how to handle.
+	 */
+	for_each_oldnew_connector_in_state(state, connector, old_con_state, new_con_state, i) {
+		struct dm_connector_state *dm_old_con_state = to_dm_connector_state(old_con_state);
+		struct dm_connector_state *dm_new_con_state = to_dm_connector_state(new_con_state);
+		struct amdgpu_crtc *acrtc = to_amdgpu_crtc(dm_new_con_state->base.crtc);
+
+		/* Skip any modesets/resets */
+		if (!acrtc || drm_atomic_crtc_needs_modeset(
+				drm_atomic_get_new_crtc_state(state, &acrtc->base)))
+			continue;
+
+		/* Skip any thing not scale or underscan changes */
+		if (!is_scaling_state_different(dm_new_con_state, dm_old_con_state))
+			continue;
+
+		lock_and_validation_needed = true;
+	}
+
+	/* set the slot info for each mst_state based on the link encoding format */
+	for_each_new_mst_mgr_in_state(state, mgr, mst_state, i) {
+		struct amdgpu_dm_connector *aconnector;
+		struct drm_connector *connector;
+		struct drm_connector_list_iter iter;
+		u8 link_coding_cap;
+
+		drm_connector_list_iter_begin(dev, &iter);
+		drm_for_each_connector_iter(connector, &iter) {
+			if (connector->index == mst_state->mgr->conn_base_id) {
+				aconnector = to_amdgpu_dm_connector(connector);
+				link_coding_cap = dc_link_dp_mst_decide_link_encoding_format(aconnector->dc_link);
+				drm_dp_mst_update_slots(mst_state, link_coding_cap);
+
+				break;
+			}
+		}
+		drm_connector_list_iter_end(&iter);
+	}
+
+	/**
+	 * Streams and planes are reset when there are changes that affect
+	 * bandwidth. Anything that affects bandwidth needs to go through
+	 * DC global validation to ensure that the configuration can be applied
+	 * to hardware.
+	 *
+	 * We have to currently stall out here in atomic_check for outstanding
+	 * commits to finish in this case because our IRQ handlers reference
+	 * DRM state directly - we can end up disabling interrupts too early
+	 * if we don't.
+	 *
+	 * TODO: Remove this stall and drop DM state private objects.
+	 */
+	if (lock_and_validation_needed) {
+		ret = dm_atomic_get_state(state, &dm_state);
+		if (ret) {
+			drm_dbg_atomic(dev, "dm_atomic_get_state() failed\n");
+			goto fail;
+		}
+
+		ret = do_aquire_global_lock(dev, state);
+		if (ret) {
+			drm_dbg_atomic(dev, "do_aquire_global_lock() failed\n");
+			goto fail;
+		}
+
+#if defined(CONFIG_DRM_AMD_DC_FP)
+		if (dc_resource_is_dsc_encoding_supported(dc)) {
+			ret = compute_mst_dsc_configs_for_state(state, dm_state->context, vars);
+			if (ret) {
+				drm_dbg_atomic(dev, "MST_DSC compute_mst_dsc_configs_for_state() failed\n");
+				ret = -EINVAL;
+				goto fail;
+			}
+		}
+#endif
+
+		ret = dm_update_mst_vcpi_slots_for_dsc(state, dm_state->context, vars);
+		if (ret) {
+			drm_dbg_atomic(dev, "dm_update_mst_vcpi_slots_for_dsc() failed\n");
+			goto fail;
+		}
+
+		/*
+		 * Perform validation of MST topology in the state:
+		 * We need to perform MST atomic check before calling
+		 * dc_validate_global_state(), or there is a chance
+		 * to get stuck in an infinite loop and hang eventually.
+		 */
+		ret = drm_dp_mst_atomic_check(state);
+		if (ret) {
+			drm_dbg_atomic(dev, "MST drm_dp_mst_atomic_check() failed\n");
+			goto fail;
+		}
+		status = dc_validate_global_state(dc, dm_state->context, DC_VALIDATE_MODE_ONLY);
+		if (status != DC_OK) {
+			drm_dbg_atomic(dev, "DC global validation failure: %s (%d)",
+				       dc_status_to_str(status), status);
+			ret = -EINVAL;
+			goto fail;
+		}
+	} else {
+		/*
+		 * The commit is a fast update. Fast updates shouldn't change
+		 * the DC context, affect global validation, and can have their
+		 * commit work done in parallel with other commits not touching
+		 * the same resource. If we have a new DC context as part of
+		 * the DM atomic state from validation we need to free it and
+		 * retain the existing one instead.
+		 *
+		 * Furthermore, since the DM atomic state only contains the DC
+		 * context and can safely be annulled, we can free the state
+		 * and clear the associated private object now to free
+		 * some memory and avoid a possible use-after-free later.
+		 */
+
+		for (i = 0; i < state->num_private_objs; i++) {
+			struct drm_private_obj *obj = state->private_objs[i].ptr;
+
+			if (obj->funcs == adev->dm.atomic_obj.funcs) {
+				int j = state->num_private_objs-1;
+
+				dm_atomic_destroy_state(obj,
+						state->private_objs[i].state);
+
+				/* If i is not at the end of the array then the
+				 * last element needs to be moved to where i was
+				 * before the array can safely be truncated.
+				 */
+				if (i != j)
+					state->private_objs[i] =
+						state->private_objs[j];
+
+				state->private_objs[j].ptr = NULL;
+				state->private_objs[j].state = NULL;
+				state->private_objs[j].old_state = NULL;
+				state->private_objs[j].new_state = NULL;
+
+				state->num_private_objs = j;
+				break;
+			}
+		}
+	}
+
+	/* Store the overall update type for use later in atomic check. */
+	for_each_new_crtc_in_state(state, crtc, new_crtc_state, i) {
+		struct dm_crtc_state *dm_new_crtc_state =
+			to_dm_crtc_state(new_crtc_state);
+
+		/*
+		 * Only allow async flips for fast updates that don't change
+		 * the FB pitch, the DCC state, rotation, mem_type, etc.
+		 */
+		if (new_crtc_state->async_flip &&
+		    (lock_and_validation_needed ||
+		     amdgpu_dm_crtc_mem_type_changed(dev, state, new_crtc_state))) {
+			drm_dbg_atomic(crtc->dev,
+				       "[CRTC:%d:%s] async flips are only supported for fast updates\n",
+				       crtc->base.id, crtc->name);
+			ret = -EINVAL;
+			goto fail;
+		}
+
+		dm_new_crtc_state->update_type = lock_and_validation_needed ?
+			UPDATE_TYPE_FULL : UPDATE_TYPE_FAST;
+	}
+
+	/* Must be success */
+	WARN_ON(ret);
+
+	trace_amdgpu_dm_atomic_check_finish(state, ret);
+
+	return ret;
+
+fail:
+	if (ret == -EDEADLK)
+		drm_dbg_atomic(dev, "Atomic check stopped to avoid deadlock.\n");
+	else if (ret == -EINTR || ret == -EAGAIN || ret == -ERESTARTSYS)
+		drm_dbg_atomic(dev, "Atomic check stopped due to signal.\n");
+	else
+		drm_dbg_atomic(dev, "Atomic check failed with err: %d\n", ret);
+
+	trace_amdgpu_dm_atomic_check_finish(state, ret);
+
+	return ret;
+}
+
+static bool dm_edid_parser_send_cea(struct amdgpu_display_manager *dm,
+		unsigned int offset,
+		unsigned int total_length,
+		u8 *data,
+		unsigned int length,
+		struct amdgpu_hdmi_vsdb_info *vsdb)
+{
+	bool res;
+	union dmub_rb_cmd cmd;
+	struct dmub_cmd_send_edid_cea *input;
+	struct dmub_cmd_edid_cea_output *output;
+
+	if (length > DMUB_EDID_CEA_DATA_CHUNK_BYTES)
+		return false;
+
+	memset(&cmd, 0, sizeof(cmd));
+
+	input = &cmd.edid_cea.data.input;
+
+	cmd.edid_cea.header.type = DMUB_CMD__EDID_CEA;
+	cmd.edid_cea.header.sub_type = 0;
+	cmd.edid_cea.header.payload_bytes =
+		sizeof(cmd.edid_cea) - sizeof(cmd.edid_cea.header);
+	input->offset = offset;
+	input->length = length;
+	input->cea_total_length = total_length;
+	memcpy(input->payload, data, length);
+
+	res = dc_wake_and_execute_dmub_cmd(dm->dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY);
+	if (!res) {
+		drm_err(adev_to_drm(dm->adev), "EDID CEA parser failed\n");
+		return false;
+	}
+
+	output = &cmd.edid_cea.data.output;
+
+	if (output->type == DMUB_CMD__EDID_CEA_ACK) {
+		if (!output->ack.success) {
+			drm_err(adev_to_drm(dm->adev), "EDID CEA ack failed at offset %d\n",
+					output->ack.offset);
+		}
+	} else if (output->type == DMUB_CMD__EDID_CEA_AMD_VSDB) {
+		if (!output->amd_vsdb.vsdb_found)
+			return false;
+
+		vsdb->freesync_supported = output->amd_vsdb.freesync_supported;
+		vsdb->amd_vsdb_version = output->amd_vsdb.amd_vsdb_version;
+		vsdb->min_refresh_rate_hz = output->amd_vsdb.min_frame_rate;
+		vsdb->max_refresh_rate_hz = output->amd_vsdb.max_frame_rate;
+	} else {
+		drm_warn(adev_to_drm(dm->adev), "Unknown EDID CEA parser results\n");
+		return false;
+	}
+
+	return true;
+}
+
+static bool parse_edid_cea_dmcu(struct amdgpu_display_manager *dm,
+		u8 *edid_ext, int len,
+		struct amdgpu_hdmi_vsdb_info *vsdb_info)
+{
+	int i;
+
+	/* send extension block to DMCU for parsing */
+	for (i = 0; i < len; i += 8) {
+		bool res;
+		int offset;
+
+		/* send 8 bytes a time */
+		if (!dc_edid_parser_send_cea(dm->dc, i, len, &edid_ext[i], 8))
+			return false;
+
+		if (i+8 == len) {
+			/* EDID block sent completed, expect result */
+			int version, min_rate, max_rate;
+
+			res = dc_edid_parser_recv_amd_vsdb(dm->dc, &version, &min_rate, &max_rate);
+			if (res) {
+				/* amd vsdb found */
+				vsdb_info->freesync_supported = 1;
+				vsdb_info->amd_vsdb_version = version;
+				vsdb_info->min_refresh_rate_hz = min_rate;
+				vsdb_info->max_refresh_rate_hz = max_rate;
+				return true;
+			}
+			/* not amd vsdb */
+			return false;
+		}
+
+		/* check for ack*/
+		res = dc_edid_parser_recv_cea_ack(dm->dc, &offset);
+		if (!res)
+			return false;
+	}
+
+	return false;
+}
+
+static bool parse_edid_cea_dmub(struct amdgpu_display_manager *dm,
+		u8 *edid_ext, int len,
+		struct amdgpu_hdmi_vsdb_info *vsdb_info)
+{
+	int i;
+
+	/* send extension block to DMCU for parsing */
+	for (i = 0; i < len; i += 8) {
+		/* send 8 bytes a time */
+		if (!dm_edid_parser_send_cea(dm, i, len, &edid_ext[i], 8, vsdb_info))
+			return false;
+	}
+
+	return vsdb_info->freesync_supported;
+}
+
+static bool parse_edid_cea(struct amdgpu_dm_connector *aconnector,
+		u8 *edid_ext, int len,
+		struct amdgpu_hdmi_vsdb_info *vsdb_info)
+{
+	struct amdgpu_device *adev = drm_to_adev(aconnector->base.dev);
+	bool ret;
+
+	mutex_lock(&adev->dm.dc_lock);
+	if (adev->dm.dmub_srv)
+		ret = parse_edid_cea_dmub(&adev->dm, edid_ext, len, vsdb_info);
+	else
+		ret = parse_edid_cea_dmcu(&adev->dm, edid_ext, len, vsdb_info);
+	mutex_unlock(&adev->dm.dc_lock);
+	return ret;
+}
+
+static void parse_edid_displayid_vrr(struct drm_connector *connector,
+				     const struct edid *edid)
+{
+	u8 *edid_ext = NULL;
+	int i;
+	int j = 0;
+	u16 min_vfreq;
+	u16 max_vfreq;
+
+	if (edid == NULL || edid->extensions == 0)
+		return;
+
+	/* Find DisplayID extension */
+	for (i = 0; i < edid->extensions; i++) {
+		edid_ext = (void *)(edid + (i + 1));
+		if (edid_ext[0] == DISPLAYID_EXT)
+			break;
+	}
+
+	if (edid_ext == NULL)
+		return;
+
+	while (j < EDID_LENGTH) {
+		/* Get dynamic video timing range from DisplayID if available */
+		if (EDID_LENGTH - j > 13 && edid_ext[j] == 0x25	&&
+		    (edid_ext[j+1] & 0xFE) == 0 && (edid_ext[j+2] == 9)) {
+			min_vfreq = edid_ext[j+9];
+			if (edid_ext[j+1] & 7)
+				max_vfreq = edid_ext[j+10] + ((edid_ext[j+11] & 3) << 8);
+			else
+				max_vfreq = edid_ext[j+10];
+
+			if (max_vfreq && min_vfreq) {
+				connector->display_info.monitor_range.max_vfreq = max_vfreq;
+				connector->display_info.monitor_range.min_vfreq = min_vfreq;
+
+				return;
+			}
+		}
+		j++;
+	}
+}
+
+static int parse_amd_vsdb(struct amdgpu_dm_connector *aconnector,
+			  const struct edid *edid, struct amdgpu_hdmi_vsdb_info *vsdb_info)
+{
+	u8 *edid_ext = NULL;
+	int i;
+	int j = 0;
+
+	if (edid == NULL || edid->extensions == 0)
+		return -ENODEV;
+
+	/* Find DisplayID extension */
+	for (i = 0; i < edid->extensions; i++) {
+		edid_ext = (void *)(edid + (i + 1));
+		if (edid_ext[0] == DISPLAYID_EXT)
+			break;
+	}
+
+	while (j < EDID_LENGTH - sizeof(struct amd_vsdb_block)) {
+		struct amd_vsdb_block *amd_vsdb = (struct amd_vsdb_block *)&edid_ext[j];
+		unsigned int ieeeId = (amd_vsdb->ieee_id[2] << 16) | (amd_vsdb->ieee_id[1] << 8) | (amd_vsdb->ieee_id[0]);
+
+		if (ieeeId == HDMI_AMD_VENDOR_SPECIFIC_DATA_BLOCK_IEEE_REGISTRATION_ID &&
+				amd_vsdb->version == HDMI_AMD_VENDOR_SPECIFIC_DATA_BLOCK_VERSION_3) {
+			vsdb_info->replay_mode = (amd_vsdb->feature_caps & AMD_VSDB_VERSION_3_FEATURECAP_REPLAYMODE) ? true : false;
+			vsdb_info->amd_vsdb_version = HDMI_AMD_VENDOR_SPECIFIC_DATA_BLOCK_VERSION_3;
+			DRM_DEBUG_KMS("Panel supports Replay Mode: %d\n", vsdb_info->replay_mode);
+
+			return true;
+		}
+		j++;
+	}
+
+	return false;
+}
+
+static int parse_hdmi_amd_vsdb(struct amdgpu_dm_connector *aconnector,
+			       const struct edid *edid,
+			       struct amdgpu_hdmi_vsdb_info *vsdb_info)
+{
+	u8 *edid_ext = NULL;
+	int i;
+	bool valid_vsdb_found = false;
+
+	/*----- drm_find_cea_extension() -----*/
+	/* No EDID or EDID extensions */
+	if (edid == NULL || edid->extensions == 0)
+		return -ENODEV;
+
+	/* Find CEA extension */
+	for (i = 0; i < edid->extensions; i++) {
+		edid_ext = (uint8_t *)edid + EDID_LENGTH * (i + 1);
+		if (edid_ext[0] == CEA_EXT)
+			break;
+	}
+
+	if (i == edid->extensions)
+		return -ENODEV;
+
+	/*----- cea_db_offsets() -----*/
+	if (edid_ext[0] != CEA_EXT)
+		return -ENODEV;
+
+	valid_vsdb_found = parse_edid_cea(aconnector, edid_ext, EDID_LENGTH, vsdb_info);
+
+	return valid_vsdb_found ? i : -ENODEV;
+}
+
+/**
+ * amdgpu_dm_update_freesync_caps - Update Freesync capabilities
+ *
+ * @connector: Connector to query.
+ * @drm_edid: DRM EDID from monitor
+ *
+ * Amdgpu supports Freesync in DP and HDMI displays, and it is required to keep
+ * track of some of the display information in the internal data struct used by
+ * amdgpu_dm. This function checks which type of connector we need to set the
+ * FreeSync parameters.
+ */
+void amdgpu_dm_update_freesync_caps(struct drm_connector *connector,
+				    const struct drm_edid *drm_edid)
+{
+	int i = 0;
+	struct amdgpu_dm_connector *amdgpu_dm_connector =
+			to_amdgpu_dm_connector(connector);
+	struct dm_connector_state *dm_con_state = NULL;
+	struct dc_sink *sink;
+	struct amdgpu_device *adev = drm_to_adev(connector->dev);
+	struct amdgpu_hdmi_vsdb_info vsdb_info = {0};
+	const struct edid *edid;
+	bool freesync_capable = false;
+	enum adaptive_sync_type as_type = ADAPTIVE_SYNC_TYPE_NONE;
+
+	if (!connector->state) {
+		drm_err(adev_to_drm(adev), "%s - Connector has no state", __func__);
+		goto update;
+	}
+
+	sink = amdgpu_dm_connector->dc_sink ?
+		amdgpu_dm_connector->dc_sink :
+		amdgpu_dm_connector->dc_em_sink;
+
+	drm_edid_connector_update(connector, drm_edid);
+
+	if (!drm_edid || !sink) {
+		dm_con_state = to_dm_connector_state(connector->state);
+
+		amdgpu_dm_connector->min_vfreq = 0;
+		amdgpu_dm_connector->max_vfreq = 0;
+		freesync_capable = false;
+
+		goto update;
+	}
+
+	dm_con_state = to_dm_connector_state(connector->state);
+
+	if (!adev->dm.freesync_module || !dc_supports_vrr(sink->ctx->dce_version))
+		goto update;
+
+	edid = drm_edid_raw(drm_edid); // FIXME: Get rid of drm_edid_raw()
+
+	/* Some eDP panels only have the refresh rate range info in DisplayID */
+	if ((connector->display_info.monitor_range.min_vfreq == 0 ||
+	     connector->display_info.monitor_range.max_vfreq == 0))
+		parse_edid_displayid_vrr(connector, edid);
+
+	if (edid && (sink->sink_signal == SIGNAL_TYPE_DISPLAY_PORT ||
+		     sink->sink_signal == SIGNAL_TYPE_EDP)) {
+		if (amdgpu_dm_connector->dc_link &&
+		    amdgpu_dm_connector->dc_link->dpcd_caps.allow_invalid_MSA_timing_param) {
+			amdgpu_dm_connector->min_vfreq = connector->display_info.monitor_range.min_vfreq;
+			amdgpu_dm_connector->max_vfreq = connector->display_info.monitor_range.max_vfreq;
+			if (amdgpu_dm_connector->max_vfreq - amdgpu_dm_connector->min_vfreq > 10)
+				freesync_capable = true;
+		}
+
+		parse_amd_vsdb(amdgpu_dm_connector, edid, &vsdb_info);
+
+		if (vsdb_info.replay_mode) {
+			amdgpu_dm_connector->vsdb_info.replay_mode = vsdb_info.replay_mode;
+			amdgpu_dm_connector->vsdb_info.amd_vsdb_version = vsdb_info.amd_vsdb_version;
+			amdgpu_dm_connector->as_type = ADAPTIVE_SYNC_TYPE_EDP;
+		}
+
+	} else if (drm_edid && sink->sink_signal == SIGNAL_TYPE_HDMI_TYPE_A) {
+		i = parse_hdmi_amd_vsdb(amdgpu_dm_connector, edid, &vsdb_info);
+		if (i >= 0 && vsdb_info.freesync_supported) {
+			amdgpu_dm_connector->min_vfreq = vsdb_info.min_refresh_rate_hz;
+			amdgpu_dm_connector->max_vfreq = vsdb_info.max_refresh_rate_hz;
+			if (amdgpu_dm_connector->max_vfreq - amdgpu_dm_connector->min_vfreq > 10)
+				freesync_capable = true;
+
+			connector->display_info.monitor_range.min_vfreq = vsdb_info.min_refresh_rate_hz;
+			connector->display_info.monitor_range.max_vfreq = vsdb_info.max_refresh_rate_hz;
+		}
+	}
+
+	if (amdgpu_dm_connector->dc_link)
+		as_type = dm_get_adaptive_sync_support_type(amdgpu_dm_connector->dc_link);
+
+	if (as_type == FREESYNC_TYPE_PCON_IN_WHITELIST) {
+		i = parse_hdmi_amd_vsdb(amdgpu_dm_connector, edid, &vsdb_info);
+		if (i >= 0 && vsdb_info.freesync_supported && vsdb_info.amd_vsdb_version > 0) {
+
+			amdgpu_dm_connector->pack_sdp_v1_3 = true;
+			amdgpu_dm_connector->as_type = as_type;
+			amdgpu_dm_connector->vsdb_info = vsdb_info;
+
+			amdgpu_dm_connector->min_vfreq = vsdb_info.min_refresh_rate_hz;
+			amdgpu_dm_connector->max_vfreq = vsdb_info.max_refresh_rate_hz;
+			if (amdgpu_dm_connector->max_vfreq - amdgpu_dm_connector->min_vfreq > 10)
+				freesync_capable = true;
+
+			connector->display_info.monitor_range.min_vfreq = vsdb_info.min_refresh_rate_hz;
+			connector->display_info.monitor_range.max_vfreq = vsdb_info.max_refresh_rate_hz;
+		}
+	}
+
+update:
+	if (dm_con_state)
+		dm_con_state->freesync_capable = freesync_capable;
+
+	if (connector->state && amdgpu_dm_connector->dc_link && !freesync_capable &&
+	    amdgpu_dm_connector->dc_link->replay_settings.config.replay_supported) {
+		amdgpu_dm_connector->dc_link->replay_settings.config.replay_supported = false;
+		amdgpu_dm_connector->dc_link->replay_settings.replay_feature_enabled = false;
+	}
+
+	if (connector->vrr_capable_property)
+		drm_connector_set_vrr_capable_property(connector,
+						       freesync_capable);
+}
+
+void amdgpu_dm_trigger_timing_sync(struct drm_device *dev)
+{
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	struct dc *dc = adev->dm.dc;
+	int i;
+
+	mutex_lock(&adev->dm.dc_lock);
+	if (dc->current_state) {
+		for (i = 0; i < dc->current_state->stream_count; ++i)
+			dc->current_state->streams[i]
+				->triggered_crtc_reset.enabled =
+				adev->dm.force_timing_sync;
+
+		dm_enable_per_frame_crtc_master_sync(dc->current_state);
+		dc_trigger_sync(dc, dc->current_state);
+	}
+	mutex_unlock(&adev->dm.dc_lock);
+}
+
+static inline void amdgpu_dm_exit_ips_for_hw_access(struct dc *dc)
+{
+	if (dc->ctx->dmub_srv && !dc->ctx->dmub_srv->idle_exit_counter)
+		dc_exit_ips_for_hw_access(dc);
+}
+
+void dm_write_reg_func(const struct dc_context *ctx, uint32_t address,
+		       u32 value, const char *func_name)
+{
+#ifdef DM_CHECK_ADDR_0
+	if (address == 0) {
+		drm_err(adev_to_drm(ctx->driver_context),
+			"invalid register write. address = 0");
+		return;
+	}
+#endif
+
+	amdgpu_dm_exit_ips_for_hw_access(ctx->dc);
+	cgs_write_register(ctx->cgs_device, address, value);
+	trace_amdgpu_dc_wreg(&ctx->perf_trace->write_count, address, value);
+}
+
+uint32_t dm_read_reg_func(const struct dc_context *ctx, uint32_t address,
+			  const char *func_name)
+{
+	u32 value;
+#ifdef DM_CHECK_ADDR_0
+	if (address == 0) {
+		drm_err(adev_to_drm(ctx->driver_context),
+			"invalid register read; address = 0\n");
+		return 0;
+	}
+#endif
+
+	if (ctx->dmub_srv &&
+	    ctx->dmub_srv->reg_helper_offload.gather_in_progress &&
+	    !ctx->dmub_srv->reg_helper_offload.should_burst_write) {
+		ASSERT(false);
+		return 0;
+	}
+
+	amdgpu_dm_exit_ips_for_hw_access(ctx->dc);
+
+	value = cgs_read_register(ctx->cgs_device, address);
+
+	trace_amdgpu_dc_rreg(&ctx->perf_trace->read_count, address, value);
+
+	return value;
+}
+
+int amdgpu_dm_process_dmub_aux_transfer_sync(
+		struct dc_context *ctx,
+		unsigned int link_index,
+		struct aux_payload *payload,
+		enum aux_return_code_type *operation_result)
+{
+	struct amdgpu_device *adev = ctx->driver_context;
+	struct dmub_notification *p_notify = adev->dm.dmub_notify;
+	int ret = -1;
+
+	mutex_lock(&adev->dm.dpia_aux_lock);
+	if (!dc_process_dmub_aux_transfer_async(ctx->dc, link_index, payload)) {
+		*operation_result = AUX_RET_ERROR_ENGINE_ACQUIRE;
+		goto out;
+	}
+
+	if (!wait_for_completion_timeout(&adev->dm.dmub_aux_transfer_done, 10 * HZ)) {
+		drm_err(adev_to_drm(adev), "wait_for_completion_timeout timeout!");
+		*operation_result = AUX_RET_ERROR_TIMEOUT;
+		goto out;
+	}
+
+	if (p_notify->result != AUX_RET_SUCCESS) {
+		/*
+		 * Transient states before tunneling is enabled could
+		 * lead to this error. We can ignore this for now.
+		 */
+		if (p_notify->result == AUX_RET_ERROR_PROTOCOL_ERROR) {
+			drm_warn(adev_to_drm(adev), "DPIA AUX failed on 0x%x(%d), error %d\n",
+					payload->address, payload->length,
+					p_notify->result);
+		}
+		*operation_result = p_notify->result;
+		goto out;
+	}
+
+	payload->reply[0] = adev->dm.dmub_notify->aux_reply.command & 0xF;
+	if (adev->dm.dmub_notify->aux_reply.command & 0xF0)
+		/* The reply is stored in the top nibble of the command. */
+		payload->reply[0] = (adev->dm.dmub_notify->aux_reply.command >> 4) & 0xF;
+
+	/*write req may receive a byte indicating partially written number as well*/
+	if (p_notify->aux_reply.length)
+		memcpy(payload->data, p_notify->aux_reply.data,
+				p_notify->aux_reply.length);
+
+	/* success */
+	ret = p_notify->aux_reply.length;
+	*operation_result = p_notify->result;
+out:
+	reinit_completion(&adev->dm.dmub_aux_transfer_done);
+	mutex_unlock(&adev->dm.dpia_aux_lock);
+	return ret;
+}
+
+static void abort_fused_io(
+		struct dc_context *ctx,
+		const struct dmub_cmd_fused_request *request
+)
+{
+	union dmub_rb_cmd command = { 0 };
+	struct dmub_rb_cmd_fused_io *io = &command.fused_io;
+
+	io->header.type = DMUB_CMD__FUSED_IO;
+	io->header.sub_type = DMUB_CMD__FUSED_IO_ABORT;
+	io->header.payload_bytes = sizeof(*io) - sizeof(io->header);
+	io->request = *request;
+	dm_execute_dmub_cmd(ctx, &command, DM_DMUB_WAIT_TYPE_NO_WAIT);
+}
+
+static bool execute_fused_io(
+		struct amdgpu_device *dev,
+		struct dc_context *ctx,
+		union dmub_rb_cmd *commands,
+		uint8_t count,
+		uint32_t timeout_us
+)
+{
+	const uint8_t ddc_line = commands[0].fused_io.request.u.aux.ddc_line;
+
+	if (ddc_line >= ARRAY_SIZE(dev->dm.fused_io))
+		return false;
+
+	struct fused_io_sync *sync = &dev->dm.fused_io[ddc_line];
+	struct dmub_rb_cmd_fused_io *first = &commands[0].fused_io;
+	const bool result = dm_execute_dmub_cmd_list(ctx, count, commands, DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY)
+			&& first->header.ret_status
+			&& first->request.status == FUSED_REQUEST_STATUS_SUCCESS;
+
+	if (!result)
+		return false;
+
+	while (wait_for_completion_timeout(&sync->replied, usecs_to_jiffies(timeout_us))) {
+		reinit_completion(&sync->replied);
+
+		struct dmub_cmd_fused_request *reply = (struct dmub_cmd_fused_request *) sync->reply_data;
+
+		static_assert(sizeof(*reply) <= sizeof(sync->reply_data), "Size mismatch");
+
+		if (reply->identifier == first->request.identifier) {
+			first->request = *reply;
+			return true;
+		}
+	}
+
+	reinit_completion(&sync->replied);
+	first->request.status = FUSED_REQUEST_STATUS_TIMEOUT;
+	abort_fused_io(ctx, &first->request);
+	return false;
+}
+
+bool amdgpu_dm_execute_fused_io(
+		struct amdgpu_device *dev,
+		struct dc_link *link,
+		union dmub_rb_cmd *commands,
+		uint8_t count,
+		uint32_t timeout_us)
+{
+	struct amdgpu_display_manager *dm = &dev->dm;
+
+	mutex_lock(&dm->dpia_aux_lock);
+
+	const bool result = execute_fused_io(dev, link->ctx, commands, count, timeout_us);
+
+	mutex_unlock(&dm->dpia_aux_lock);
+	return result;
+}
+
+int amdgpu_dm_process_dmub_set_config_sync(
+		struct dc_context *ctx,
+		unsigned int link_index,
+		struct set_config_cmd_payload *payload,
+		enum set_config_status *operation_result)
+{
+	struct amdgpu_device *adev = ctx->driver_context;
+	bool is_cmd_complete;
+	int ret;
+
+	mutex_lock(&adev->dm.dpia_aux_lock);
+	is_cmd_complete = dc_process_dmub_set_config_async(ctx->dc,
+			link_index, payload, adev->dm.dmub_notify);
+
+	if (is_cmd_complete || wait_for_completion_timeout(&adev->dm.dmub_aux_transfer_done, 10 * HZ)) {
+		ret = 0;
+		*operation_result = adev->dm.dmub_notify->sc_status;
+	} else {
+		drm_err(adev_to_drm(adev), "wait_for_completion_timeout timeout!");
+		ret = -1;
+		*operation_result = SET_CONFIG_UNKNOWN_ERROR;
+	}
+
+	if (!is_cmd_complete)
+		reinit_completion(&adev->dm.dmub_aux_transfer_done);
+	mutex_unlock(&adev->dm.dpia_aux_lock);
+	return ret;
+}
+
+bool dm_execute_dmub_cmd(const struct dc_context *ctx, union dmub_rb_cmd *cmd, enum dm_dmub_wait_type wait_type)
+{
+	return dc_dmub_srv_cmd_run(ctx->dmub_srv, cmd, wait_type);
+}
+
+bool dm_execute_dmub_cmd_list(const struct dc_context *ctx, unsigned int count, union dmub_rb_cmd *cmd, enum dm_dmub_wait_type wait_type)
+{
+	return dc_dmub_srv_cmd_run_list(ctx->dmub_srv, count, cmd, wait_type);
+}
+
+void dm_acpi_process_phy_transition_interlock(
+	const struct dc_context *ctx,
+	struct dm_process_phy_transition_init_params process_phy_transition_init_params)
+{
+	// Not yet implemented
+}
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
new file mode 100644
index 000000000000..009f206226f0
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
@@ -0,0 +1,1114 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright (C) 2015-2020 Advanced Micro Devices, Inc. All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __AMDGPU_DM_H__
+#define __AMDGPU_DM_H__
+
+#include <drm/display/drm_dp_mst_helper.h>
+#include <drm/drm_atomic.h>
+#include <drm/drm_connector.h>
+#include <drm/drm_crtc.h>
+#include <drm/drm_plane.h>
+#include "link_service_types.h"
+#include <drm/drm_writeback.h>
+
+/*
+ * This file contains the definition for amdgpu_display_manager
+ * and its API for amdgpu driver's use.
+ * This component provides all the display related functionality
+ * and this is the only component that calls DAL API.
+ * The API contained here intended for amdgpu driver use.
+ * The API that is called directly from KMS framework is located
+ * in amdgpu_dm_kms.h file
+ */
+
+#define AMDGPU_DM_MAX_DISPLAY_INDEX 31
+
+#define AMDGPU_DM_MAX_CRTC 6
+
+#define AMDGPU_DM_MAX_NUM_EDP 2
+
+#define AMDGPU_DMUB_NOTIFICATION_MAX 8
+
+#define HDMI_AMD_VENDOR_SPECIFIC_DATA_BLOCK_IEEE_REGISTRATION_ID 0x00001A
+#define AMD_VSDB_VERSION_3_FEATURECAP_REPLAYMODE 0x40
+#define HDMI_AMD_VENDOR_SPECIFIC_DATA_BLOCK_VERSION_3 0x3
+
+#define AMDGPU_HDR_MULT_DEFAULT (0x100000000LL)
+
+/*
+#include "include/amdgpu_dal_power_if.h"
+#include "amdgpu_dm_irq.h"
+*/
+
+#include "irq_types.h"
+#include "signal_types.h"
+#include "amdgpu_dm_crc.h"
+#include "mod_info_packet.h"
+struct aux_payload;
+struct set_config_cmd_payload;
+enum aux_return_code_type;
+enum set_config_status;
+
+/* Forward declarations */
+struct amdgpu_device;
+struct amdgpu_crtc;
+struct drm_device;
+struct dc;
+struct amdgpu_bo;
+struct dmub_srv;
+struct dc_plane_state;
+struct dmub_notification;
+struct dmub_cmd_fused_request;
+
+struct amd_vsdb_block {
+	unsigned char ieee_id[3];
+	unsigned char version;
+	unsigned char feature_caps;
+};
+
+struct common_irq_params {
+	struct amdgpu_device *adev;
+	enum dc_irq_source irq_src;
+	atomic64_t previous_timestamp;
+};
+
+/**
+ * struct dm_compressor_info - Buffer info used by frame buffer compression
+ * @cpu_addr: MMIO cpu addr
+ * @bo_ptr: Pointer to the buffer object
+ * @gpu_addr: MMIO gpu addr
+ */
+struct dm_compressor_info {
+	void *cpu_addr;
+	struct amdgpu_bo *bo_ptr;
+	uint64_t gpu_addr;
+};
+
+typedef void (*dmub_notify_interrupt_callback_t)(struct amdgpu_device *adev, struct dmub_notification *notify);
+
+/**
+ * struct dmub_hpd_work - Handle time consuming work in low priority outbox IRQ
+ *
+ * @handle_hpd_work: Work to be executed in a separate thread to handle hpd_low_irq
+ * @dmub_notify:  notification for callback function
+ * @adev: amdgpu_device pointer
+ */
+struct dmub_hpd_work {
+	struct work_struct handle_hpd_work;
+	struct dmub_notification *dmub_notify;
+	struct amdgpu_device *adev;
+};
+
+/**
+ * struct vblank_control_work - Work data for vblank control
+ * @work: Kernel work data for the work event
+ * @dm: amdgpu display manager device
+ * @acrtc: amdgpu CRTC instance for which the event has occurred
+ * @stream: DC stream for which the event has occurred
+ * @enable: true if enabling vblank
+ */
+struct vblank_control_work {
+	struct work_struct work;
+	struct amdgpu_display_manager *dm;
+	struct amdgpu_crtc *acrtc;
+	struct dc_stream_state *stream;
+	bool enable;
+};
+
+/**
+ * struct idle_workqueue - Work data for periodic action in idle
+ * @work: Kernel work data for the work event
+ * @dm: amdgpu display manager device
+ * @enable: true if idle worker is enabled
+ * @running: true if idle worker is running
+ */
+struct idle_workqueue {
+	struct work_struct work;
+	struct amdgpu_display_manager *dm;
+	bool enable;
+	bool running;
+};
+
+/**
+ * struct vupdate_offload_work - Work data for offloading task from vupdate handler
+ * @work: Kernel work data for the work event
+ * @adev: amdgpu_device back pointer
+ * @stream: DC stream associated with the crtc
+ * @adjust: DC CRTC timing adjust to be applied to the crtc
+ */
+struct vupdate_offload_work {
+	struct work_struct work;
+	struct amdgpu_device *adev;
+	struct dc_stream_state *stream;
+	struct dc_crtc_timing_adjust *adjust;
+};
+
+#define MAX_LUMINANCE_DATA_POINTS 99
+
+/**
+ * struct amdgpu_dm_luminance_data - Custom luminance data
+ * @luminance: Luminance in percent
+ * @input_signal: Input signal in range 0-255
+ */
+struct amdgpu_dm_luminance_data {
+	u8 luminance;
+	u8 input_signal;
+} __packed;
+
+/**
+ * struct amdgpu_dm_backlight_caps - Information about backlight
+ *
+ * Describe the backlight support for ACPI or eDP AUX.
+ */
+struct amdgpu_dm_backlight_caps {
+	/**
+	 * @ext_caps: Keep the data struct with all the information about the
+	 * display support for HDR.
+	 */
+	union dpcd_sink_ext_caps *ext_caps;
+	/**
+	 * @aux_min_input_signal: Min brightness value supported by the display
+	 */
+	u32 aux_min_input_signal;
+	/**
+	 * @aux_max_input_signal: Max brightness value supported by the display
+	 * in nits.
+	 */
+	u32 aux_max_input_signal;
+	/**
+	 * @min_input_signal: minimum possible input in range 0-255.
+	 */
+	int min_input_signal;
+	/**
+	 * @max_input_signal: maximum possible input in range 0-255.
+	 */
+	int max_input_signal;
+	/**
+	 * @caps_valid: true if these values are from the ACPI interface.
+	 */
+	bool caps_valid;
+	/**
+	 * @aux_support: Describes if the display supports AUX backlight.
+	 */
+	bool aux_support;
+	/**
+	 * @brightness_mask: After deriving brightness, OR it with this mask.
+	 * Workaround for panels with issues with certain brightness values.
+	 */
+	u32 brightness_mask;
+	/**
+	 * @ac_level: the default brightness if booted on AC
+	 */
+	u8 ac_level;
+	/**
+	 * @dc_level: the default brightness if booted on DC
+	 */
+	u8 dc_level;
+	/**
+	 * @data_points: the number of custom luminance data points
+	 */
+	u8 data_points;
+	/**
+	 * @luminance_data: custom luminance data
+	 */
+	struct amdgpu_dm_luminance_data luminance_data[MAX_LUMINANCE_DATA_POINTS];
+};
+
+/**
+ * struct dal_allocation - Tracks mapped FB memory for SMU communication
+ * @list: list of dal allocations
+ * @bo: GPU buffer object
+ * @cpu_ptr: CPU virtual address of the GPU buffer object
+ * @gpu_addr: GPU virtual address of the GPU buffer object
+ */
+struct dal_allocation {
+	struct list_head list;
+	struct amdgpu_bo *bo;
+	void *cpu_ptr;
+	u64 gpu_addr;
+};
+
+/**
+ * struct hpd_rx_irq_offload_work_queue - Work queue to handle hpd_rx_irq
+ * offload work
+ */
+struct hpd_rx_irq_offload_work_queue {
+	/**
+	 * @wq: workqueue structure to queue offload work.
+	 */
+	struct workqueue_struct *wq;
+	/**
+	 * @offload_lock: To protect fields of offload work queue.
+	 */
+	spinlock_t offload_lock;
+	/**
+	 * @is_handling_link_loss: Used to prevent inserting link loss event when
+	 * we're handling link loss
+	 */
+	bool is_handling_link_loss;
+	/**
+	 * @is_handling_mst_msg_rdy_event: Used to prevent inserting mst message
+	 * ready event when we're already handling mst message ready event
+	 */
+	bool is_handling_mst_msg_rdy_event;
+	/**
+	 * @aconnector: The aconnector that this work queue is attached to
+	 */
+	struct amdgpu_dm_connector *aconnector;
+};
+
+/**
+ * struct hpd_rx_irq_offload_work - hpd_rx_irq offload work structure
+ */
+struct hpd_rx_irq_offload_work {
+	/**
+	 * @work: offload work
+	 */
+	struct work_struct work;
+	/**
+	 * @data: reference irq data which is used while handling offload work
+	 */
+	union hpd_irq_data data;
+	/**
+	 * @offload_wq: offload work queue that this work is queued to
+	 */
+	struct hpd_rx_irq_offload_work_queue *offload_wq;
+	/**
+	 * @adev: amdgpu_device pointer
+	 */
+	struct amdgpu_device *adev;
+};
+
+/**
+ * struct amdgpu_display_manager - Central amdgpu display manager device
+ *
+ * @dc: Display Core control structure
+ * @adev: AMDGPU base driver structure
+ * @ddev: DRM base driver structure
+ * @display_indexes_num: Max number of display streams supported
+ * @irq_handler_list_table_lock: Synchronizes access to IRQ tables
+ * @backlight_dev: Backlight control device
+ * @backlight_link: Link on which to control backlight
+ * @backlight_caps: Capabilities of the backlight device
+ * @freesync_module: Module handling freesync calculations
+ * @hdcp_workqueue: AMDGPU content protection queue
+ * @fw_dmcu: Reference to DMCU firmware
+ * @dmcu_fw_version: Version of the DMCU firmware
+ * @soc_bounding_box: SOC bounding box values provided by gpu_info FW
+ * @cached_state: Caches device atomic state for suspend/resume
+ * @cached_dc_state: Cached state of content streams
+ * @compressor: Frame buffer compression buffer. See &struct dm_compressor_info
+ * @force_timing_sync: set via debugfs. When set, indicates that all connected
+ *		       displays will be forced to synchronize.
+ * @dmcub_trace_event_en: enable dmcub trace events
+ * @dmub_outbox_params: DMUB Outbox parameters
+ * @num_of_edps: number of backlight eDPs
+ * @disable_hpd_irq: disables all HPD and HPD RX interrupt handling in the
+ *		     driver when true
+ * @dmub_aux_transfer_done: struct completion used to indicate when DMUB
+ * 			    transfers are done
+ * @delayed_hpd_wq: work queue used to delay DMUB HPD work
+ */
+struct amdgpu_display_manager {
+
+	struct dc *dc;
+
+	/**
+	 * @dmub_srv:
+	 *
+	 * DMUB service, used for controlling the DMUB on hardware
+	 * that supports it. The pointer to the dmub_srv will be
+	 * NULL on hardware that does not support it.
+	 */
+	struct dmub_srv *dmub_srv;
+
+	/**
+	 * @dmub_notify:
+	 *
+	 * Notification from DMUB.
+	 */
+
+	struct dmub_notification *dmub_notify;
+
+	/**
+	 * @dmub_callback:
+	 *
+	 * Callback functions to handle notification from DMUB.
+	 */
+
+	dmub_notify_interrupt_callback_t dmub_callback[AMDGPU_DMUB_NOTIFICATION_MAX];
+
+	/**
+	 * @dmub_thread_offload:
+	 *
+	 * Flag to indicate if callback is offload.
+	 */
+
+	bool dmub_thread_offload[AMDGPU_DMUB_NOTIFICATION_MAX];
+
+	/**
+	 * @dmub_fb_info:
+	 *
+	 * Framebuffer regions for the DMUB.
+	 */
+	struct dmub_srv_fb_info *dmub_fb_info;
+
+	/**
+	 * @dmub_fw:
+	 *
+	 * DMUB firmware, required on hardware that has DMUB support.
+	 */
+	const struct firmware *dmub_fw;
+
+	/**
+	 * @dmub_bo:
+	 *
+	 * Buffer object for the DMUB.
+	 */
+	struct amdgpu_bo *dmub_bo;
+
+	/**
+	 * @dmub_bo_gpu_addr:
+	 *
+	 * GPU virtual address for the DMUB buffer object.
+	 */
+	u64 dmub_bo_gpu_addr;
+
+	/**
+	 * @dmub_bo_cpu_addr:
+	 *
+	 * CPU address for the DMUB buffer object.
+	 */
+	void *dmub_bo_cpu_addr;
+
+	/**
+	 * @dmcub_fw_version:
+	 *
+	 * DMCUB firmware version.
+	 */
+	uint32_t dmcub_fw_version;
+
+	/**
+	 * @cgs_device:
+	 *
+	 * The Common Graphics Services device. It provides an interface for
+	 * accessing registers.
+	 */
+	struct cgs_device *cgs_device;
+
+	struct amdgpu_device *adev;
+	struct drm_device *ddev;
+	u16 display_indexes_num;
+
+	/**
+	 * @atomic_obj:
+	 *
+	 * In combination with &dm_atomic_state it helps manage
+	 * global atomic state that doesn't map cleanly into existing
+	 * drm resources, like &dc_context.
+	 */
+	struct drm_private_obj atomic_obj;
+
+	/**
+	 * @dc_lock:
+	 *
+	 * Guards access to DC functions that can issue register write
+	 * sequences.
+	 */
+	struct mutex dc_lock;
+
+	/**
+	 * @audio_lock:
+	 *
+	 * Guards access to audio instance changes.
+	 */
+	struct mutex audio_lock;
+
+	/**
+	 * @audio_component:
+	 *
+	 * Used to notify ELD changes to sound driver.
+	 */
+	struct drm_audio_component *audio_component;
+
+	/**
+	 * @audio_registered:
+	 *
+	 * True if the audio component has been registered
+	 * successfully, false otherwise.
+	 */
+	bool audio_registered;
+
+	/**
+	 * @irq_handler_list_low_tab:
+	 *
+	 * Low priority IRQ handler table.
+	 *
+	 * It is a n*m table consisting of n IRQ sources, and m handlers per IRQ
+	 * source. Low priority IRQ handlers are deferred to a workqueue to be
+	 * processed. Hence, they can sleep.
+	 *
+	 * Note that handlers are called in the same order as they were
+	 * registered (FIFO).
+	 */
+	struct list_head irq_handler_list_low_tab[DAL_IRQ_SOURCES_NUMBER];
+
+	/**
+	 * @irq_handler_list_high_tab:
+	 *
+	 * High priority IRQ handler table.
+	 *
+	 * It is a n*m table, same as &irq_handler_list_low_tab. However,
+	 * handlers in this table are not deferred and are called immediately.
+	 */
+	struct list_head irq_handler_list_high_tab[DAL_IRQ_SOURCES_NUMBER];
+
+	/**
+	 * @pflip_params:
+	 *
+	 * Page flip IRQ parameters, passed to registered handlers when
+	 * triggered.
+	 */
+	struct common_irq_params
+	pflip_params[DC_IRQ_SOURCE_PFLIP_LAST - DC_IRQ_SOURCE_PFLIP_FIRST + 1];
+
+	/**
+	 * @vblank_params:
+	 *
+	 * Vertical blanking IRQ parameters, passed to registered handlers when
+	 * triggered.
+	 */
+	struct common_irq_params
+	vblank_params[DC_IRQ_SOURCE_VBLANK6 - DC_IRQ_SOURCE_VBLANK1 + 1];
+
+	/**
+	 * @vline0_params:
+	 *
+	 * OTG vertical interrupt0 IRQ parameters, passed to registered
+	 * handlers when triggered.
+	 */
+	struct common_irq_params
+	vline0_params[DC_IRQ_SOURCE_DC6_VLINE0 - DC_IRQ_SOURCE_DC1_VLINE0 + 1];
+
+	/**
+	 * @vupdate_params:
+	 *
+	 * Vertical update IRQ parameters, passed to registered handlers when
+	 * triggered.
+	 */
+	struct common_irq_params
+	vupdate_params[DC_IRQ_SOURCE_VUPDATE6 - DC_IRQ_SOURCE_VUPDATE1 + 1];
+
+	/**
+	 * @dmub_trace_params:
+	 *
+	 * DMUB trace event IRQ parameters, passed to registered handlers when
+	 * triggered.
+	 */
+	struct common_irq_params
+	dmub_trace_params[1];
+
+	struct common_irq_params
+	dmub_outbox_params[1];
+
+	spinlock_t irq_handler_list_table_lock;
+
+	struct backlight_device *backlight_dev[AMDGPU_DM_MAX_NUM_EDP];
+
+	const struct dc_link *backlight_link[AMDGPU_DM_MAX_NUM_EDP];
+
+	uint8_t num_of_edps;
+
+	struct amdgpu_dm_backlight_caps backlight_caps[AMDGPU_DM_MAX_NUM_EDP];
+
+	struct mod_freesync *freesync_module;
+	struct hdcp_workqueue *hdcp_workqueue;
+
+	/**
+	 * @vblank_control_workqueue:
+	 *
+	 * Deferred work for vblank control events.
+	 */
+	struct workqueue_struct *vblank_control_workqueue;
+
+	/**
+	 * @idle_workqueue:
+	 *
+	 * Periodic work for idle events.
+	 */
+	struct idle_workqueue *idle_workqueue;
+
+	struct drm_atomic_state *cached_state;
+	struct dc_state *cached_dc_state;
+
+	struct dm_compressor_info compressor;
+
+	const struct firmware *fw_dmcu;
+	uint32_t dmcu_fw_version;
+	/**
+	 * @soc_bounding_box:
+	 *
+	 * gpu_info FW provided soc bounding box struct or 0 if not
+	 * available in FW
+	 */
+	const struct gpu_info_soc_bounding_box_v1_0 *soc_bounding_box;
+
+	/**
+	 * @active_vblank_irq_count:
+	 *
+	 * number of currently active vblank irqs
+	 */
+	uint32_t active_vblank_irq_count;
+
+#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY)
+	/**
+	 * @secure_display_ctx:
+	 *
+	 * Store secure display relevant info. e.g. the ROI information
+	 * , the work_struct to command dmub, etc.
+	 */
+	struct secure_display_context secure_display_ctx;
+#endif
+	/**
+	 * @hpd_rx_offload_wq:
+	 *
+	 * Work queue to offload works of hpd_rx_irq
+	 */
+	struct hpd_rx_irq_offload_work_queue *hpd_rx_offload_wq;
+	/**
+	 * @mst_encoders:
+	 *
+	 * fake encoders used for DP MST.
+	 */
+	struct amdgpu_encoder mst_encoders[AMDGPU_DM_MAX_CRTC];
+	bool force_timing_sync;
+	bool disable_hpd_irq;
+	bool dmcub_trace_event_en;
+	/**
+	 * @da_list:
+	 *
+	 * DAL fb memory allocation list, for communication with SMU.
+	 */
+	struct list_head da_list;
+	struct completion dmub_aux_transfer_done;
+	struct workqueue_struct *delayed_hpd_wq;
+
+	/**
+	 * @brightness:
+	 *
+	 * cached backlight values.
+	 */
+	u32 brightness[AMDGPU_DM_MAX_NUM_EDP];
+	/**
+	 * @actual_brightness:
+	 *
+	 * last successfully applied backlight values.
+	 */
+	u32 actual_brightness[AMDGPU_DM_MAX_NUM_EDP];
+
+	/**
+	 * @restore_backlight:
+	 *
+	 * Flag to indicate whether to restore backlight after modeset.
+	 */
+	bool restore_backlight;
+
+	/**
+	 * @aux_hpd_discon_quirk:
+	 *
+	 * quirk for hpd discon while aux is on-going.
+	 * occurred on certain intel platform
+	 */
+	bool aux_hpd_discon_quirk;
+
+	/**
+	 * @edp0_on_dp1_quirk:
+	 *
+	 * quirk for platforms that put edp0 on DP1.
+	 */
+	bool edp0_on_dp1_quirk;
+
+	/**
+	 * @dpia_aux_lock:
+	 *
+	 * Guards access to DPIA AUX
+	 */
+	struct mutex dpia_aux_lock;
+
+	/**
+	 * @bb_from_dmub:
+	 *
+	 * Bounding box data read from dmub during early initialization for DCN4+
+	 * Data is stored as a byte array that should be casted to the appropriate bb struct
+	 */
+	void *bb_from_dmub;
+
+	/**
+	 * @oem_i2c:
+	 *
+	 * OEM i2c bus
+	 */
+	struct amdgpu_i2c_adapter *oem_i2c;
+
+	/**
+	 * @fused_io:
+	 *
+	 * dmub fused io interface
+	 */
+	struct fused_io_sync {
+		struct completion replied;
+		char reply_data[0x40];  // Cannot include dmub_cmd here
+	} fused_io[8];
+};
+
+enum dsc_clock_force_state {
+	DSC_CLK_FORCE_DEFAULT = 0,
+	DSC_CLK_FORCE_ENABLE,
+	DSC_CLK_FORCE_DISABLE,
+};
+
+struct dsc_preferred_settings {
+	enum dsc_clock_force_state dsc_force_enable;
+	uint32_t dsc_num_slices_v;
+	uint32_t dsc_num_slices_h;
+	uint32_t dsc_bits_per_pixel;
+	bool dsc_force_disable_passthrough;
+};
+
+enum mst_progress_status {
+	MST_STATUS_DEFAULT = 0,
+	MST_PROBE = BIT(0),
+	MST_REMOTE_EDID = BIT(1),
+	MST_ALLOCATE_NEW_PAYLOAD = BIT(2),
+	MST_CLEAR_ALLOCATED_PAYLOAD = BIT(3),
+};
+
+/**
+ * struct amdgpu_hdmi_vsdb_info - Keep track of the VSDB info
+ *
+ * AMDGPU supports FreeSync over HDMI by using the VSDB section, and this
+ * struct is useful to keep track of the display-specific information about
+ * FreeSync.
+ */
+struct amdgpu_hdmi_vsdb_info {
+	/**
+	 * @amd_vsdb_version: Vendor Specific Data Block Version, should be
+	 * used to determine which Vendor Specific InfoFrame (VSIF) to send.
+	 */
+	unsigned int amd_vsdb_version;
+
+	/**
+	 * @freesync_supported: FreeSync Supported.
+	 */
+	bool freesync_supported;
+
+	/**
+	 * @min_refresh_rate_hz: FreeSync Minimum Refresh Rate in Hz.
+	 */
+	unsigned int min_refresh_rate_hz;
+
+	/**
+	 * @max_refresh_rate_hz: FreeSync Maximum Refresh Rate in Hz
+	 */
+	unsigned int max_refresh_rate_hz;
+
+	/**
+	 * @replay_mode: Replay supported
+	 */
+	bool replay_mode;
+};
+
+struct amdgpu_dm_connector {
+
+	struct drm_connector base;
+	uint32_t connector_id;
+	int bl_idx;
+
+	struct cec_notifier *notifier;
+
+	/* we need to mind the EDID between detect
+	   and get modes due to analog/digital/tvencoder */
+	const struct drm_edid *drm_edid;
+
+	/* shared with amdgpu */
+	struct amdgpu_hpd hpd;
+
+	/* number of modes generated from EDID at 'dc_sink' */
+	int num_modes;
+
+	/* The 'old' sink - before an HPD.
+	 * The 'current' sink is in dc_link->sink. */
+	struct dc_sink *dc_sink;
+	struct dc_link *dc_link;
+
+	/**
+	 * @dc_em_sink: Reference to the emulated (virtual) sink.
+	 */
+	struct dc_sink *dc_em_sink;
+
+	/* DM only */
+	struct drm_dp_mst_topology_mgr mst_mgr;
+	struct amdgpu_dm_dp_aux dm_dp_aux;
+	struct drm_dp_mst_port *mst_output_port;
+	struct amdgpu_dm_connector *mst_root;
+	struct drm_dp_aux *dsc_aux;
+	uint32_t mst_local_bw;
+	uint16_t vc_full_pbn;
+	struct mutex handle_mst_msg_ready;
+
+	/* branch device specific data */
+	uint32_t branch_ieee_oui;
+
+	/* TODO see if we can merge with ddc_bus or make a dm_connector */
+	struct amdgpu_i2c_adapter *i2c;
+
+	/* Monitor range limits */
+	/**
+	 * @min_vfreq: Minimal frequency supported by the display in Hz. This
+	 * value is set to zero when there is no FreeSync support.
+	 */
+	int min_vfreq;
+
+	/**
+	 * @max_vfreq: Maximum frequency supported by the display in Hz. This
+	 * value is set to zero when there is no FreeSync support.
+	 */
+	int max_vfreq ;
+
+	/* Audio instance - protected by audio_lock. */
+	int audio_inst;
+
+	struct mutex hpd_lock;
+
+	bool fake_enable;
+	bool force_yuv420_output;
+	bool force_yuv422_output;
+	struct dsc_preferred_settings dsc_settings;
+	union dp_downstream_port_present mst_downstream_port_present;
+	/* Cached display modes */
+	struct drm_display_mode freesync_vid_base;
+
+	int sr_skip_count;
+	bool disallow_edp_enter_psr;
+
+	/* Record progress status of mst*/
+	uint8_t mst_status;
+
+	/* Automated testing */
+	bool timing_changed;
+	struct dc_crtc_timing *timing_requested;
+
+	/* Adaptive Sync */
+	bool pack_sdp_v1_3;
+	enum adaptive_sync_type as_type;
+	struct amdgpu_hdmi_vsdb_info vsdb_info;
+};
+
+static inline void amdgpu_dm_set_mst_status(uint8_t *status,
+		uint8_t flags, bool set)
+{
+	if (set)
+		*status |= flags;
+	else
+		*status &= ~flags;
+}
+
+#define to_amdgpu_dm_connector(x) container_of(x, struct amdgpu_dm_connector, base)
+
+struct amdgpu_dm_wb_connector {
+	struct drm_writeback_connector base;
+	struct dc_link *link;
+};
+
+#define to_amdgpu_dm_wb_connector(x) container_of(x, struct amdgpu_dm_wb_connector, base)
+
+extern const struct amdgpu_ip_block_version dm_ip_block;
+
+/* enum amdgpu_transfer_function: pre-defined transfer function supported by AMD.
+ *
+ * It includes standardized transfer functions and pure power functions. The
+ * transfer function coefficients are available at modules/color/color_gamma.c
+ */
+enum amdgpu_transfer_function {
+	AMDGPU_TRANSFER_FUNCTION_DEFAULT,
+	AMDGPU_TRANSFER_FUNCTION_SRGB_EOTF,
+	AMDGPU_TRANSFER_FUNCTION_BT709_INV_OETF,
+	AMDGPU_TRANSFER_FUNCTION_PQ_EOTF,
+	AMDGPU_TRANSFER_FUNCTION_IDENTITY,
+	AMDGPU_TRANSFER_FUNCTION_GAMMA22_EOTF,
+	AMDGPU_TRANSFER_FUNCTION_GAMMA24_EOTF,
+	AMDGPU_TRANSFER_FUNCTION_GAMMA26_EOTF,
+	AMDGPU_TRANSFER_FUNCTION_SRGB_INV_EOTF,
+	AMDGPU_TRANSFER_FUNCTION_BT709_OETF,
+	AMDGPU_TRANSFER_FUNCTION_PQ_INV_EOTF,
+	AMDGPU_TRANSFER_FUNCTION_GAMMA22_INV_EOTF,
+	AMDGPU_TRANSFER_FUNCTION_GAMMA24_INV_EOTF,
+	AMDGPU_TRANSFER_FUNCTION_GAMMA26_INV_EOTF,
+	AMDGPU_TRANSFER_FUNCTION_COUNT
+};
+
+struct dm_plane_state {
+	struct drm_plane_state base;
+	struct dc_plane_state *dc_state;
+
+	/* Plane color mgmt */
+	/**
+	 * @degamma_lut:
+	 *
+	 * 1D LUT for mapping framebuffer/plane pixel data before sampling or
+	 * blending operations. It's usually applied to linearize input space.
+	 * The blob (if not NULL) is an array of &struct drm_color_lut.
+	 */
+	struct drm_property_blob *degamma_lut;
+	/**
+	 * @degamma_tf:
+	 *
+	 * Predefined transfer function to tell DC driver the input space to
+	 * linearize.
+	 */
+	enum amdgpu_transfer_function degamma_tf;
+	/**
+	 * @hdr_mult:
+	 *
+	 * Multiplier to 'gain' the plane.  When PQ is decoded using the fixed
+	 * func transfer function to the internal FP16 fb, 1.0 -> 80 nits (on
+	 * AMD at least). When sRGB is decoded, 1.0 -> 1.0, obviously.
+	 * Therefore, 1.0 multiplier = 80 nits for SDR content.  So if you
+	 * want, 203 nits for SDR content, pass in (203.0 / 80.0).  Format is
+	 * S31.32 sign-magnitude.
+	 *
+	 * HDR multiplier can wide range beyond [0.0, 1.0]. This means that PQ
+	 * TF is needed for any subsequent linear-to-non-linear transforms.
+	 */
+	__u64 hdr_mult;
+	/**
+	 * @ctm:
+	 *
+	 * Color transformation matrix. The blob (if not NULL) is a &struct
+	 * drm_color_ctm_3x4.
+	 */
+	struct drm_property_blob *ctm;
+	/**
+	 * @shaper_lut: shaper lookup table blob. The blob (if not NULL) is an
+	 * array of &struct drm_color_lut.
+	 */
+	struct drm_property_blob *shaper_lut;
+	/**
+	 * @shaper_tf:
+	 *
+	 * Predefined transfer function to delinearize color space.
+	 */
+	enum amdgpu_transfer_function shaper_tf;
+	/**
+	 * @lut3d: 3D lookup table blob. The blob (if not NULL) is an array of
+	 * &struct drm_color_lut.
+	 */
+	struct drm_property_blob *lut3d;
+	/**
+	 * @blend_lut: blend lut lookup table blob. The blob (if not NULL) is an
+	 * array of &struct drm_color_lut.
+	 */
+	struct drm_property_blob *blend_lut;
+	/**
+	 * @blend_tf:
+	 *
+	 * Pre-defined transfer function for converting plane pixel data before
+	 * applying blend LUT.
+	 */
+	enum amdgpu_transfer_function blend_tf;
+};
+
+enum amdgpu_dm_cursor_mode {
+	DM_CURSOR_NATIVE_MODE = 0,
+	DM_CURSOR_OVERLAY_MODE,
+};
+
+struct dm_crtc_state {
+	struct drm_crtc_state base;
+	struct dc_stream_state *stream;
+
+	bool cm_has_degamma;
+	bool cm_is_degamma_srgb;
+
+	bool mpo_requested;
+
+	int update_type;
+	int active_planes;
+
+	int crc_skip_count;
+
+	bool freesync_vrr_info_changed;
+
+	bool dsc_force_changed;
+	bool vrr_supported;
+	struct mod_freesync_config freesync_config;
+	struct dc_info_packet vrr_infopacket;
+
+	int abm_level;
+
+	/**
+	 * @regamma_tf:
+	 *
+	 * Pre-defined transfer function for converting internal FB -> wire
+	 * encoding.
+	 */
+	enum amdgpu_transfer_function regamma_tf;
+
+	enum amdgpu_dm_cursor_mode cursor_mode;
+};
+
+#define to_dm_crtc_state(x) container_of(x, struct dm_crtc_state, base)
+
+struct dm_atomic_state {
+	struct drm_private_state base;
+
+	struct dc_state *context;
+};
+
+#define to_dm_atomic_state(x) container_of(x, struct dm_atomic_state, base)
+
+struct dm_connector_state {
+	struct drm_connector_state base;
+
+	enum amdgpu_rmx_type scaling;
+	uint8_t underscan_vborder;
+	uint8_t underscan_hborder;
+	bool underscan_enable;
+	bool freesync_capable;
+	bool update_hdcp;
+	uint8_t abm_level;
+	int vcpi_slots;
+	uint64_t pbn;
+};
+
+#define to_dm_connector_state(x)\
+	container_of((x), struct dm_connector_state, base)
+
+void amdgpu_dm_connector_funcs_reset(struct drm_connector *connector);
+struct drm_connector_state *
+amdgpu_dm_connector_atomic_duplicate_state(struct drm_connector *connector);
+int amdgpu_dm_connector_atomic_set_property(struct drm_connector *connector,
+					    struct drm_connector_state *state,
+					    struct drm_property *property,
+					    uint64_t val);
+
+int amdgpu_dm_connector_atomic_get_property(struct drm_connector *connector,
+					    const struct drm_connector_state *state,
+					    struct drm_property *property,
+					    uint64_t *val);
+
+int amdgpu_dm_get_encoder_crtc_mask(struct amdgpu_device *adev);
+
+void amdgpu_dm_connector_init_helper(struct amdgpu_display_manager *dm,
+				     struct amdgpu_dm_connector *aconnector,
+				     int connector_type,
+				     struct dc_link *link,
+				     int link_index);
+
+enum drm_mode_status amdgpu_dm_connector_mode_valid(struct drm_connector *connector,
+				   const struct drm_display_mode *mode);
+
+void dm_restore_drm_connector_state(struct drm_device *dev,
+				    struct drm_connector *connector);
+
+void amdgpu_dm_update_freesync_caps(struct drm_connector *connector,
+				    const struct drm_edid *drm_edid);
+
+void amdgpu_dm_trigger_timing_sync(struct drm_device *dev);
+
+/* 3D LUT max size is 17x17x17 (4913 entries) */
+#define MAX_COLOR_3DLUT_SIZE 17
+#define MAX_COLOR_3DLUT_BITDEPTH 12
+int amdgpu_dm_verify_lut3d_size(struct amdgpu_device *adev,
+				struct drm_plane_state *plane_state);
+/* 1D LUT size */
+#define MAX_COLOR_LUT_ENTRIES 4096
+/* Legacy gamm LUT users such as X doesn't like large LUT sizes */
+#define MAX_COLOR_LEGACY_LUT_ENTRIES 256
+
+void amdgpu_dm_init_color_mod(void);
+int amdgpu_dm_create_color_properties(struct amdgpu_device *adev);
+int amdgpu_dm_verify_lut_sizes(const struct drm_crtc_state *crtc_state);
+int amdgpu_dm_update_crtc_color_mgmt(struct dm_crtc_state *crtc);
+int amdgpu_dm_check_crtc_color_mgmt(struct dm_crtc_state *crtc,
+				    bool check_only);
+int amdgpu_dm_update_plane_color_mgmt(struct dm_crtc_state *crtc,
+				      struct drm_plane_state *plane_state,
+				      struct dc_plane_state *dc_plane_state);
+
+void amdgpu_dm_update_connector_after_detect(
+		struct amdgpu_dm_connector *aconnector);
+
+extern const struct drm_encoder_helper_funcs amdgpu_dm_encoder_helper_funcs;
+
+int amdgpu_dm_process_dmub_aux_transfer_sync(struct dc_context *ctx, unsigned int link_index,
+					struct aux_payload *payload, enum aux_return_code_type *operation_result);
+
+bool amdgpu_dm_execute_fused_io(
+		struct amdgpu_device *dev,
+		struct dc_link *link,
+		union dmub_rb_cmd *commands,
+		uint8_t count,
+		uint32_t timeout_us
+);
+
+int amdgpu_dm_process_dmub_set_config_sync(struct dc_context *ctx, unsigned int link_index,
+					struct set_config_cmd_payload *payload, enum set_config_status *operation_result);
+
+struct dc_stream_state *
+	create_validate_stream_for_sink(struct drm_connector *connector,
+					const struct drm_display_mode *drm_mode,
+					const struct dm_connector_state *dm_state,
+					const struct dc_stream_state *old_stream);
+
+int dm_atomic_get_state(struct drm_atomic_state *state,
+			struct dm_atomic_state **dm_state);
+
+struct drm_connector *
+amdgpu_dm_find_first_crtc_matching_connector(struct drm_atomic_state *state,
+					     struct drm_crtc *crtc);
+
+int convert_dc_color_depth_into_bpc(enum dc_color_depth display_color_depth);
+struct idle_workqueue *idle_create_workqueue(struct amdgpu_device *adev);
+
+void *dm_allocate_gpu_mem(struct amdgpu_device *adev,
+						  enum dc_gpu_mem_alloc_type type,
+						  size_t size,
+						  long long *addr);
+void dm_free_gpu_mem(struct amdgpu_device *adev,
+						  enum dc_gpu_mem_alloc_type type,
+						  void *addr);
+
+bool amdgpu_dm_is_headless(struct amdgpu_device *adev);
+
+void hdmi_cec_set_edid(struct amdgpu_dm_connector *aconnector);
+void hdmi_cec_unset_edid(struct amdgpu_dm_connector *aconnector);
+int amdgpu_dm_initialize_hdmi_connector(struct amdgpu_dm_connector *aconnector);
+
+void retrieve_dmi_info(struct amdgpu_display_manager *dm);
+
+#endif /* __AMDGPU_DM_H__ */
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c
new file mode 100644
index 000000000000..a4ac6d442278
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c
@@ -0,0 +1,1327 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+#include "amdgpu.h"
+#include "amdgpu_mode.h"
+#include "amdgpu_dm.h"
+#include "dc.h"
+#include "modules/color/color_gamma.h"
+
+/**
+ * DOC: overview
+ *
+ * The DC interface to HW gives us the following color management blocks
+ * per pipe (surface):
+ *
+ * - Input gamma LUT (de-normalized)
+ * - Input CSC (normalized)
+ * - Surface degamma LUT (normalized)
+ * - Surface CSC (normalized)
+ * - Surface regamma LUT (normalized)
+ * - Output CSC (normalized)
+ *
+ * But these aren't a direct mapping to DRM color properties. The current DRM
+ * interface exposes CRTC degamma, CRTC CTM and CRTC regamma while our hardware
+ * is essentially giving:
+ *
+ * Plane CTM -> Plane degamma -> Plane CTM -> Plane regamma -> Plane CTM
+ *
+ * The input gamma LUT block isn't really applicable here since it operates
+ * on the actual input data itself rather than the HW fp representation. The
+ * input and output CSC blocks are technically available to use as part of
+ * the DC interface but are typically used internally by DC for conversions
+ * between color spaces. These could be blended together with user
+ * adjustments in the future but for now these should remain untouched.
+ *
+ * The pipe blending also happens after these blocks so we don't actually
+ * support any CRTC props with correct blending with multiple planes - but we
+ * can still support CRTC color management properties in DM in most single
+ * plane cases correctly with clever management of the DC interface in DM.
+ *
+ * As per DRM documentation, blocks should be in hardware bypass when their
+ * respective property is set to NULL. A linear DGM/RGM LUT should also
+ * considered as putting the respective block into bypass mode.
+ *
+ * This means that the following
+ * configuration is assumed to be the default:
+ *
+ * Plane DGM Bypass -> Plane CTM Bypass -> Plane RGM Bypass -> ...
+ * CRTC DGM Bypass -> CRTC CTM Bypass -> CRTC RGM Bypass
+ */
+
+#define MAX_DRM_LUT_VALUE 0xFFFF
+#define SDR_WHITE_LEVEL_INIT_VALUE 80
+
+/**
+ * amdgpu_dm_init_color_mod - Initialize the color module.
+ *
+ * We're not using the full color module, only certain components.
+ * Only call setup functions for components that we need.
+ */
+void amdgpu_dm_init_color_mod(void)
+{
+	setup_x_points_distribution();
+}
+
+static inline struct fixed31_32 amdgpu_dm_fixpt_from_s3132(__u64 x)
+{
+	struct fixed31_32 val;
+
+	/* If negative, convert to 2's complement. */
+	if (x & (1ULL << 63))
+		x = -(x & ~(1ULL << 63));
+
+	val.value = x;
+	return val;
+}
+
+#ifdef AMD_PRIVATE_COLOR
+/* Pre-defined Transfer Functions (TF)
+ *
+ * AMD driver supports pre-defined mathematical functions for transferring
+ * between encoded values and optical/linear space. Depending on HW color caps,
+ * ROMs and curves built by the AMD color module support these transforms.
+ *
+ * The driver-specific color implementation exposes properties for pre-blending
+ * degamma TF, shaper TF (before 3D LUT), and blend(dpp.ogam) TF and
+ * post-blending regamma (mpc.ogam) TF. However, only pre-blending degamma
+ * supports ROM curves. AMD color module uses pre-defined coefficients to build
+ * curves for the other blocks. What can be done by each color block is
+ * described by struct dpp_color_capsand struct mpc_color_caps.
+ *
+ * AMD driver-specific color API exposes the following pre-defined transfer
+ * functions:
+ *
+ * - Identity: linear/identity relationship between pixel value and
+ *   luminance value;
+ * - Gamma 2.2, Gamma 2.4, Gamma 2.6: pure power functions;
+ * - sRGB: 2.4: The piece-wise transfer function from IEC 61966-2-1:1999;
+ * - BT.709: has a linear segment in the bottom part and then a power function
+ *   with a 0.45 (~1/2.22) gamma for the rest of the range; standardized by
+ *   ITU-R BT.709-6;
+ * - PQ (Perceptual Quantizer): used for HDR display, allows luminance range
+ *   capability of 0 to 10,000 nits; standardized by SMPTE ST 2084.
+ *
+ * The AMD color model is designed with an assumption that SDR (sRGB, BT.709,
+ * Gamma 2.2, etc.) peak white maps (normalized to 1.0 FP) to 80 nits in the PQ
+ * system. This has the implication that PQ EOTF (non-linear to linear) maps to
+ * [0.0..125.0] where 125.0 = 10,000 nits / 80 nits.
+ *
+ * Non-linear and linear forms are described in the table below:
+ *
+ * ┌───────────┬─────────────────────┬──────────────────────┐
+ * │           │     Non-linear      │   Linear             │
+ * ├───────────┼─────────────────────┼──────────────────────┤
+ * │      sRGB │ UNORM or [0.0, 1.0] │ [0.0, 1.0]           │
+ * ├───────────┼─────────────────────┼──────────────────────┤
+ * │     BT709 │ UNORM or [0.0, 1.0] │ [0.0, 1.0]           │
+ * ├───────────┼─────────────────────┼──────────────────────┤
+ * │ Gamma 2.x │ UNORM or [0.0, 1.0] │ [0.0, 1.0]           │
+ * ├───────────┼─────────────────────┼──────────────────────┤
+ * │        PQ │ UNORM or FP16 CCCS* │ [0.0, 125.0]         │
+ * ├───────────┼─────────────────────┼──────────────────────┤
+ * │  Identity │ UNORM or FP16 CCCS* │ [0.0, 1.0] or CCCS** │
+ * └───────────┴─────────────────────┴──────────────────────┘
+ * * CCCS: Windows canonical composition color space
+ * ** Respectively
+ *
+ * In the driver-specific API, color block names attached to TF properties
+ * suggest the intention regarding non-linear encoding pixel's luminance
+ * values. As some newer encodings don't use gamma curve, we make encoding and
+ * decoding explicit by defining an enum list of transfer functions supported
+ * in terms of EOTF and inverse EOTF, where:
+ *
+ * - EOTF (electro-optical transfer function): is the transfer function to go
+ *   from the encoded value to an optical (linear) value. De-gamma functions
+ *   traditionally do this.
+ * - Inverse EOTF (simply the inverse of the EOTF): is usually intended to go
+ *   from an optical/linear space (which might have been used for blending)
+ *   back to the encoded values. Gamma functions traditionally do this.
+ */
+static const char * const
+amdgpu_transfer_function_names[] = {
+	[AMDGPU_TRANSFER_FUNCTION_DEFAULT]		= "Default",
+	[AMDGPU_TRANSFER_FUNCTION_IDENTITY]		= "Identity",
+	[AMDGPU_TRANSFER_FUNCTION_SRGB_EOTF]		= "sRGB EOTF",
+	[AMDGPU_TRANSFER_FUNCTION_BT709_INV_OETF]	= "BT.709 inv_OETF",
+	[AMDGPU_TRANSFER_FUNCTION_PQ_EOTF]		= "PQ EOTF",
+	[AMDGPU_TRANSFER_FUNCTION_GAMMA22_EOTF]		= "Gamma 2.2 EOTF",
+	[AMDGPU_TRANSFER_FUNCTION_GAMMA24_EOTF]		= "Gamma 2.4 EOTF",
+	[AMDGPU_TRANSFER_FUNCTION_GAMMA26_EOTF]		= "Gamma 2.6 EOTF",
+	[AMDGPU_TRANSFER_FUNCTION_SRGB_INV_EOTF]	= "sRGB inv_EOTF",
+	[AMDGPU_TRANSFER_FUNCTION_BT709_OETF]		= "BT.709 OETF",
+	[AMDGPU_TRANSFER_FUNCTION_PQ_INV_EOTF]		= "PQ inv_EOTF",
+	[AMDGPU_TRANSFER_FUNCTION_GAMMA22_INV_EOTF]	= "Gamma 2.2 inv_EOTF",
+	[AMDGPU_TRANSFER_FUNCTION_GAMMA24_INV_EOTF]	= "Gamma 2.4 inv_EOTF",
+	[AMDGPU_TRANSFER_FUNCTION_GAMMA26_INV_EOTF]	= "Gamma 2.6 inv_EOTF",
+};
+
+static const u32 amdgpu_eotf =
+	BIT(AMDGPU_TRANSFER_FUNCTION_SRGB_EOTF) |
+	BIT(AMDGPU_TRANSFER_FUNCTION_BT709_INV_OETF) |
+	BIT(AMDGPU_TRANSFER_FUNCTION_PQ_EOTF) |
+	BIT(AMDGPU_TRANSFER_FUNCTION_GAMMA22_EOTF) |
+	BIT(AMDGPU_TRANSFER_FUNCTION_GAMMA24_EOTF) |
+	BIT(AMDGPU_TRANSFER_FUNCTION_GAMMA26_EOTF);
+
+static const u32 amdgpu_inv_eotf =
+	BIT(AMDGPU_TRANSFER_FUNCTION_SRGB_INV_EOTF) |
+	BIT(AMDGPU_TRANSFER_FUNCTION_BT709_OETF) |
+	BIT(AMDGPU_TRANSFER_FUNCTION_PQ_INV_EOTF) |
+	BIT(AMDGPU_TRANSFER_FUNCTION_GAMMA22_INV_EOTF) |
+	BIT(AMDGPU_TRANSFER_FUNCTION_GAMMA24_INV_EOTF) |
+	BIT(AMDGPU_TRANSFER_FUNCTION_GAMMA26_INV_EOTF);
+
+static struct drm_property *
+amdgpu_create_tf_property(struct drm_device *dev,
+			  const char *name,
+			  u32 supported_tf)
+{
+	u32 transfer_functions = supported_tf |
+				 BIT(AMDGPU_TRANSFER_FUNCTION_DEFAULT) |
+				 BIT(AMDGPU_TRANSFER_FUNCTION_IDENTITY);
+	struct drm_prop_enum_list enum_list[AMDGPU_TRANSFER_FUNCTION_COUNT];
+	int i, len;
+
+	len = 0;
+	for (i = 0; i < AMDGPU_TRANSFER_FUNCTION_COUNT; i++) {
+		if ((transfer_functions & BIT(i)) == 0)
+			continue;
+
+		enum_list[len].type = i;
+		enum_list[len].name = amdgpu_transfer_function_names[i];
+		len++;
+	}
+
+	return drm_property_create_enum(dev, DRM_MODE_PROP_ENUM,
+					name, enum_list, len);
+}
+
+int
+amdgpu_dm_create_color_properties(struct amdgpu_device *adev)
+{
+	struct drm_property *prop;
+
+	prop = drm_property_create(adev_to_drm(adev),
+				   DRM_MODE_PROP_BLOB,
+				   "AMD_PLANE_DEGAMMA_LUT", 0);
+	if (!prop)
+		return -ENOMEM;
+	adev->mode_info.plane_degamma_lut_property = prop;
+
+	prop = drm_property_create_range(adev_to_drm(adev),
+					 DRM_MODE_PROP_IMMUTABLE,
+					 "AMD_PLANE_DEGAMMA_LUT_SIZE",
+					 0, UINT_MAX);
+	if (!prop)
+		return -ENOMEM;
+	adev->mode_info.plane_degamma_lut_size_property = prop;
+
+	prop = amdgpu_create_tf_property(adev_to_drm(adev),
+					 "AMD_PLANE_DEGAMMA_TF",
+					 amdgpu_eotf);
+	if (!prop)
+		return -ENOMEM;
+	adev->mode_info.plane_degamma_tf_property = prop;
+
+	prop = drm_property_create_range(adev_to_drm(adev),
+					 0, "AMD_PLANE_HDR_MULT", 0, U64_MAX);
+	if (!prop)
+		return -ENOMEM;
+	adev->mode_info.plane_hdr_mult_property = prop;
+
+	prop = drm_property_create(adev_to_drm(adev),
+				   DRM_MODE_PROP_BLOB,
+				   "AMD_PLANE_CTM", 0);
+	if (!prop)
+		return -ENOMEM;
+	adev->mode_info.plane_ctm_property = prop;
+
+	prop = drm_property_create(adev_to_drm(adev),
+				   DRM_MODE_PROP_BLOB,
+				   "AMD_PLANE_SHAPER_LUT", 0);
+	if (!prop)
+		return -ENOMEM;
+	adev->mode_info.plane_shaper_lut_property = prop;
+
+	prop = drm_property_create_range(adev_to_drm(adev),
+					 DRM_MODE_PROP_IMMUTABLE,
+					 "AMD_PLANE_SHAPER_LUT_SIZE", 0, UINT_MAX);
+	if (!prop)
+		return -ENOMEM;
+	adev->mode_info.plane_shaper_lut_size_property = prop;
+
+	prop = amdgpu_create_tf_property(adev_to_drm(adev),
+					 "AMD_PLANE_SHAPER_TF",
+					 amdgpu_inv_eotf);
+	if (!prop)
+		return -ENOMEM;
+	adev->mode_info.plane_shaper_tf_property = prop;
+
+	prop = drm_property_create(adev_to_drm(adev),
+				   DRM_MODE_PROP_BLOB,
+				   "AMD_PLANE_LUT3D", 0);
+	if (!prop)
+		return -ENOMEM;
+	adev->mode_info.plane_lut3d_property = prop;
+
+	prop = drm_property_create_range(adev_to_drm(adev),
+					 DRM_MODE_PROP_IMMUTABLE,
+					 "AMD_PLANE_LUT3D_SIZE", 0, UINT_MAX);
+	if (!prop)
+		return -ENOMEM;
+	adev->mode_info.plane_lut3d_size_property = prop;
+
+	prop = drm_property_create(adev_to_drm(adev),
+				   DRM_MODE_PROP_BLOB,
+				   "AMD_PLANE_BLEND_LUT", 0);
+	if (!prop)
+		return -ENOMEM;
+	adev->mode_info.plane_blend_lut_property = prop;
+
+	prop = drm_property_create_range(adev_to_drm(adev),
+					 DRM_MODE_PROP_IMMUTABLE,
+					 "AMD_PLANE_BLEND_LUT_SIZE", 0, UINT_MAX);
+	if (!prop)
+		return -ENOMEM;
+	adev->mode_info.plane_blend_lut_size_property = prop;
+
+	prop = amdgpu_create_tf_property(adev_to_drm(adev),
+					 "AMD_PLANE_BLEND_TF",
+					 amdgpu_eotf);
+	if (!prop)
+		return -ENOMEM;
+	adev->mode_info.plane_blend_tf_property = prop;
+
+	prop = amdgpu_create_tf_property(adev_to_drm(adev),
+					 "AMD_CRTC_REGAMMA_TF",
+					 amdgpu_inv_eotf);
+	if (!prop)
+		return -ENOMEM;
+	adev->mode_info.regamma_tf_property = prop;
+
+	return 0;
+}
+#endif
+
+/**
+ * __extract_blob_lut - Extracts the DRM lut and lut size from a blob.
+ * @blob: DRM color mgmt property blob
+ * @size: lut size
+ *
+ * Returns:
+ * DRM LUT or NULL
+ */
+static const struct drm_color_lut *
+__extract_blob_lut(const struct drm_property_blob *blob, uint32_t *size)
+{
+	*size = blob ? drm_color_lut_size(blob) : 0;
+	return blob ? (struct drm_color_lut *)blob->data : NULL;
+}
+
+/**
+ * __is_lut_linear - check if the given lut is a linear mapping of values
+ * @lut: given lut to check values
+ * @size: lut size
+ *
+ * It is considered linear if the lut represents:
+ * f(a) = (0xFF00/MAX_COLOR_LUT_ENTRIES-1)a; for integer a in [0,
+ * MAX_COLOR_LUT_ENTRIES)
+ *
+ * Returns:
+ * True if the given lut is a linear mapping of values, i.e. it acts like a
+ * bypass LUT. Otherwise, false.
+ */
+static bool __is_lut_linear(const struct drm_color_lut *lut, uint32_t size)
+{
+	int i;
+	uint32_t expected;
+	int delta;
+
+	for (i = 0; i < size; i++) {
+		/* All color values should equal */
+		if ((lut[i].red != lut[i].green) || (lut[i].green != lut[i].blue))
+			return false;
+
+		expected = i * MAX_DRM_LUT_VALUE / (size-1);
+
+		/* Allow a +/-1 error. */
+		delta = lut[i].red - expected;
+		if (delta < -1 || 1 < delta)
+			return false;
+	}
+	return true;
+}
+
+/**
+ * __drm_lut_to_dc_gamma - convert the drm_color_lut to dc_gamma.
+ * @lut: DRM lookup table for color conversion
+ * @gamma: DC gamma to set entries
+ * @is_legacy: legacy or atomic gamma
+ *
+ * The conversion depends on the size of the lut - whether or not it's legacy.
+ */
+static void __drm_lut_to_dc_gamma(const struct drm_color_lut *lut,
+				  struct dc_gamma *gamma, bool is_legacy)
+{
+	uint32_t r, g, b;
+	int i;
+
+	if (is_legacy) {
+		for (i = 0; i < MAX_COLOR_LEGACY_LUT_ENTRIES; i++) {
+			r = drm_color_lut_extract(lut[i].red, 16);
+			g = drm_color_lut_extract(lut[i].green, 16);
+			b = drm_color_lut_extract(lut[i].blue, 16);
+
+			gamma->entries.red[i] = dc_fixpt_from_int(r);
+			gamma->entries.green[i] = dc_fixpt_from_int(g);
+			gamma->entries.blue[i] = dc_fixpt_from_int(b);
+		}
+		return;
+	}
+
+	/* else */
+	for (i = 0; i < MAX_COLOR_LUT_ENTRIES; i++) {
+		r = drm_color_lut_extract(lut[i].red, 16);
+		g = drm_color_lut_extract(lut[i].green, 16);
+		b = drm_color_lut_extract(lut[i].blue, 16);
+
+		gamma->entries.red[i] = dc_fixpt_from_fraction(r, MAX_DRM_LUT_VALUE);
+		gamma->entries.green[i] = dc_fixpt_from_fraction(g, MAX_DRM_LUT_VALUE);
+		gamma->entries.blue[i] = dc_fixpt_from_fraction(b, MAX_DRM_LUT_VALUE);
+	}
+}
+
+/**
+ * __drm_ctm_to_dc_matrix - converts a DRM CTM to a DC CSC float matrix
+ * @ctm: DRM color transformation matrix
+ * @matrix: DC CSC float matrix
+ *
+ * The matrix needs to be a 3x4 (12 entry) matrix.
+ */
+static void __drm_ctm_to_dc_matrix(const struct drm_color_ctm *ctm,
+				   struct fixed31_32 *matrix)
+{
+	int i;
+
+	/*
+	 * DRM gives a 3x3 matrix, but DC wants 3x4. Assuming we're operating
+	 * with homogeneous coordinates, augment the matrix with 0's.
+	 *
+	 * The format provided is S31.32, using signed-magnitude representation.
+	 * Our fixed31_32 is also S31.32, but is using 2's complement. We have
+	 * to convert from signed-magnitude to 2's complement.
+	 */
+	for (i = 0; i < 12; i++) {
+		/* Skip 4th element */
+		if (i % 4 == 3) {
+			matrix[i] = dc_fixpt_zero;
+			continue;
+		}
+
+		/* gamut_remap_matrix[i] = ctm[i - floor(i/4)] */
+		matrix[i] = amdgpu_dm_fixpt_from_s3132(ctm->matrix[i - (i / 4)]);
+	}
+}
+
+/**
+ * __drm_ctm_3x4_to_dc_matrix - converts a DRM CTM 3x4 to a DC CSC float matrix
+ * @ctm: DRM color transformation matrix with 3x4 dimensions
+ * @matrix: DC CSC float matrix
+ *
+ * The matrix needs to be a 3x4 (12 entry) matrix.
+ */
+static void __drm_ctm_3x4_to_dc_matrix(const struct drm_color_ctm_3x4 *ctm,
+				       struct fixed31_32 *matrix)
+{
+	int i;
+
+	/* The format provided is S31.32, using signed-magnitude representation.
+	 * Our fixed31_32 is also S31.32, but is using 2's complement. We have
+	 * to convert from signed-magnitude to 2's complement.
+	 */
+	for (i = 0; i < 12; i++) {
+		/* gamut_remap_matrix[i] = ctm[i - floor(i/4)] */
+		matrix[i] = amdgpu_dm_fixpt_from_s3132(ctm->matrix[i]);
+	}
+}
+
+/**
+ * __set_legacy_tf - Calculates the legacy transfer function
+ * @func: transfer function
+ * @lut: lookup table that defines the color space
+ * @lut_size: size of respective lut
+ * @has_rom: if ROM can be used for hardcoded curve
+ *
+ * Only for sRGB input space
+ *
+ * Returns:
+ * 0 in case of success, -ENOMEM if fails
+ */
+static int __set_legacy_tf(struct dc_transfer_func *func,
+			   const struct drm_color_lut *lut, uint32_t lut_size,
+			   bool has_rom)
+{
+	struct dc_gamma *gamma = NULL;
+	struct calculate_buffer cal_buffer = {0};
+	bool res;
+
+	ASSERT(lut && lut_size == MAX_COLOR_LEGACY_LUT_ENTRIES);
+
+	cal_buffer.buffer_index = -1;
+
+	gamma = dc_create_gamma();
+	if (!gamma)
+		return -ENOMEM;
+
+	gamma->type = GAMMA_RGB_256;
+	gamma->num_entries = lut_size;
+	__drm_lut_to_dc_gamma(lut, gamma, true);
+
+	res = mod_color_calculate_regamma_params(func, gamma, true, has_rom,
+						 NULL, &cal_buffer);
+
+	dc_gamma_release(&gamma);
+
+	return res ? 0 : -ENOMEM;
+}
+
+/**
+ * __set_output_tf - calculates the output transfer function based on expected input space.
+ * @func: transfer function
+ * @lut: lookup table that defines the color space
+ * @lut_size: size of respective lut
+ * @has_rom: if ROM can be used for hardcoded curve
+ *
+ * Returns:
+ * 0 in case of success. -ENOMEM if fails.
+ */
+static int __set_output_tf(struct dc_transfer_func *func,
+			   const struct drm_color_lut *lut, uint32_t lut_size,
+			   bool has_rom)
+{
+	struct dc_gamma *gamma = NULL;
+	struct calculate_buffer cal_buffer = {0};
+	bool res;
+
+	cal_buffer.buffer_index = -1;
+
+	if (lut_size) {
+		ASSERT(lut && lut_size == MAX_COLOR_LUT_ENTRIES);
+
+		gamma = dc_create_gamma();
+		if (!gamma)
+			return -ENOMEM;
+
+		gamma->num_entries = lut_size;
+		__drm_lut_to_dc_gamma(lut, gamma, false);
+	}
+
+	if (func->tf == TRANSFER_FUNCTION_LINEAR) {
+		/*
+		 * Color module doesn't like calculating regamma params
+		 * on top of a linear input. But degamma params can be used
+		 * instead to simulate this.
+		 */
+		if (gamma)
+			gamma->type = GAMMA_CUSTOM;
+		res = mod_color_calculate_degamma_params(NULL, func,
+							 gamma, gamma != NULL);
+	} else {
+		/*
+		 * Assume sRGB. The actual mapping will depend on whether the
+		 * input was legacy or not.
+		 */
+		if (gamma)
+			gamma->type = GAMMA_CS_TFM_1D;
+		res = mod_color_calculate_regamma_params(func, gamma, gamma != NULL,
+							 has_rom, NULL, &cal_buffer);
+	}
+
+	if (gamma)
+		dc_gamma_release(&gamma);
+
+	return res ? 0 : -ENOMEM;
+}
+
+static int amdgpu_dm_set_atomic_regamma(struct dc_transfer_func *out_tf,
+					const struct drm_color_lut *regamma_lut,
+					uint32_t regamma_size, bool has_rom,
+					enum dc_transfer_func_predefined tf)
+{
+	int ret = 0;
+
+	if (regamma_size || tf != TRANSFER_FUNCTION_LINEAR) {
+		/*
+		 * CRTC RGM goes into RGM LUT.
+		 *
+		 * Note: there is no implicit sRGB regamma here. We are using
+		 * degamma calculation from color module to calculate the curve
+		 * from a linear base if gamma TF is not set. However, if gamma
+		 * TF (!= Linear) and LUT are set at the same time, we will use
+		 * regamma calculation, and the color module will combine the
+		 * pre-defined TF and the custom LUT values into the LUT that's
+		 * actually programmed.
+		 */
+		out_tf->type = TF_TYPE_DISTRIBUTED_POINTS;
+		out_tf->tf = tf;
+		out_tf->sdr_ref_white_level = SDR_WHITE_LEVEL_INIT_VALUE;
+
+		ret = __set_output_tf(out_tf, regamma_lut, regamma_size, has_rom);
+	} else {
+		/*
+		 * No CRTC RGM means we can just put the block into bypass
+		 * since we don't have any plane level adjustments using it.
+		 */
+		out_tf->type = TF_TYPE_BYPASS;
+		out_tf->tf = TRANSFER_FUNCTION_LINEAR;
+	}
+
+	return ret;
+}
+
+/**
+ * __set_input_tf - calculates the input transfer function based on expected
+ * input space.
+ * @caps: dc color capabilities
+ * @func: transfer function
+ * @lut: lookup table that defines the color space
+ * @lut_size: size of respective lut.
+ *
+ * Returns:
+ * 0 in case of success. -ENOMEM if fails.
+ */
+static int __set_input_tf(struct dc_color_caps *caps, struct dc_transfer_func *func,
+			  const struct drm_color_lut *lut, uint32_t lut_size)
+{
+	struct dc_gamma *gamma = NULL;
+	bool res;
+
+	if (lut_size) {
+		gamma = dc_create_gamma();
+		if (!gamma)
+			return -ENOMEM;
+
+		gamma->type = GAMMA_CUSTOM;
+		gamma->num_entries = lut_size;
+
+		__drm_lut_to_dc_gamma(lut, gamma, false);
+	}
+
+	res = mod_color_calculate_degamma_params(caps, func, gamma, gamma != NULL);
+
+	if (gamma)
+		dc_gamma_release(&gamma);
+
+	return res ? 0 : -ENOMEM;
+}
+
+static enum dc_transfer_func_predefined
+amdgpu_tf_to_dc_tf(enum amdgpu_transfer_function tf)
+{
+	switch (tf) {
+	default:
+	case AMDGPU_TRANSFER_FUNCTION_DEFAULT:
+	case AMDGPU_TRANSFER_FUNCTION_IDENTITY:
+		return TRANSFER_FUNCTION_LINEAR;
+	case AMDGPU_TRANSFER_FUNCTION_SRGB_EOTF:
+	case AMDGPU_TRANSFER_FUNCTION_SRGB_INV_EOTF:
+		return TRANSFER_FUNCTION_SRGB;
+	case AMDGPU_TRANSFER_FUNCTION_BT709_OETF:
+	case AMDGPU_TRANSFER_FUNCTION_BT709_INV_OETF:
+		return TRANSFER_FUNCTION_BT709;
+	case AMDGPU_TRANSFER_FUNCTION_PQ_EOTF:
+	case AMDGPU_TRANSFER_FUNCTION_PQ_INV_EOTF:
+		return TRANSFER_FUNCTION_PQ;
+	case AMDGPU_TRANSFER_FUNCTION_GAMMA22_EOTF:
+	case AMDGPU_TRANSFER_FUNCTION_GAMMA22_INV_EOTF:
+		return TRANSFER_FUNCTION_GAMMA22;
+	case AMDGPU_TRANSFER_FUNCTION_GAMMA24_EOTF:
+	case AMDGPU_TRANSFER_FUNCTION_GAMMA24_INV_EOTF:
+		return TRANSFER_FUNCTION_GAMMA24;
+	case AMDGPU_TRANSFER_FUNCTION_GAMMA26_EOTF:
+	case AMDGPU_TRANSFER_FUNCTION_GAMMA26_INV_EOTF:
+		return TRANSFER_FUNCTION_GAMMA26;
+	}
+}
+
+static void __to_dc_lut3d_color(struct dc_rgb *rgb,
+				const struct drm_color_lut lut,
+				int bit_precision)
+{
+	rgb->red = drm_color_lut_extract(lut.red, bit_precision);
+	rgb->green = drm_color_lut_extract(lut.green, bit_precision);
+	rgb->blue  = drm_color_lut_extract(lut.blue, bit_precision);
+}
+
+static void __drm_3dlut_to_dc_3dlut(const struct drm_color_lut *lut,
+				    uint32_t lut3d_size,
+				    struct tetrahedral_params *params,
+				    bool use_tetrahedral_9,
+				    int bit_depth)
+{
+	struct dc_rgb *lut0;
+	struct dc_rgb *lut1;
+	struct dc_rgb *lut2;
+	struct dc_rgb *lut3;
+	int lut_i, i;
+
+
+	if (use_tetrahedral_9) {
+		lut0 = params->tetrahedral_9.lut0;
+		lut1 = params->tetrahedral_9.lut1;
+		lut2 = params->tetrahedral_9.lut2;
+		lut3 = params->tetrahedral_9.lut3;
+	} else {
+		lut0 = params->tetrahedral_17.lut0;
+		lut1 = params->tetrahedral_17.lut1;
+		lut2 = params->tetrahedral_17.lut2;
+		lut3 = params->tetrahedral_17.lut3;
+	}
+
+	for (lut_i = 0, i = 0; i < lut3d_size - 4; lut_i++, i += 4) {
+		/*
+		 * We should consider the 3D LUT RGB values are distributed
+		 * along four arrays lut0-3 where the first sizes 1229 and the
+		 * other 1228. The bit depth supported for 3dlut channel is
+		 * 12-bit, but DC also supports 10-bit.
+		 *
+		 * TODO: improve color pipeline API to enable the userspace set
+		 * bit depth and 3D LUT size/stride, as specified by VA-API.
+		 */
+		__to_dc_lut3d_color(&lut0[lut_i], lut[i], bit_depth);
+		__to_dc_lut3d_color(&lut1[lut_i], lut[i + 1], bit_depth);
+		__to_dc_lut3d_color(&lut2[lut_i], lut[i + 2], bit_depth);
+		__to_dc_lut3d_color(&lut3[lut_i], lut[i + 3], bit_depth);
+	}
+	/* lut0 has 1229 points (lut_size/4 + 1) */
+	__to_dc_lut3d_color(&lut0[lut_i], lut[i], bit_depth);
+}
+
+/* amdgpu_dm_atomic_lut3d - set DRM 3D LUT to DC stream
+ * @drm_lut3d: user 3D LUT
+ * @drm_lut3d_size: size of 3D LUT
+ * @lut3d: DC 3D LUT
+ *
+ * Map user 3D LUT data to DC 3D LUT and all necessary bits to program it
+ * on DCN accordingly.
+ */
+static void amdgpu_dm_atomic_lut3d(const struct drm_color_lut *drm_lut3d,
+				   uint32_t drm_lut3d_size,
+				   struct dc_3dlut *lut)
+{
+	if (!drm_lut3d_size) {
+		lut->state.bits.initialized = 0;
+	} else {
+		/* Stride and bit depth are not programmable by API yet.
+		 * Therefore, only supports 17x17x17 3D LUT (12-bit).
+		 */
+		lut->lut_3d.use_tetrahedral_9 = false;
+		lut->lut_3d.use_12bits = true;
+		lut->state.bits.initialized = 1;
+		__drm_3dlut_to_dc_3dlut(drm_lut3d, drm_lut3d_size, &lut->lut_3d,
+					lut->lut_3d.use_tetrahedral_9,
+					MAX_COLOR_3DLUT_BITDEPTH);
+	}
+}
+
+static int amdgpu_dm_atomic_shaper_lut(const struct drm_color_lut *shaper_lut,
+				       bool has_rom,
+				       enum dc_transfer_func_predefined tf,
+				       uint32_t shaper_size,
+				       struct dc_transfer_func *func_shaper)
+{
+	int ret = 0;
+
+	if (shaper_size || tf != TRANSFER_FUNCTION_LINEAR) {
+		/*
+		 * If user shaper LUT is set, we assume a linear color space
+		 * (linearized by degamma 1D LUT or not).
+		 */
+		func_shaper->type = TF_TYPE_DISTRIBUTED_POINTS;
+		func_shaper->tf = tf;
+		func_shaper->sdr_ref_white_level = SDR_WHITE_LEVEL_INIT_VALUE;
+
+		ret = __set_output_tf(func_shaper, shaper_lut, shaper_size, has_rom);
+	} else {
+		func_shaper->type = TF_TYPE_BYPASS;
+		func_shaper->tf = TRANSFER_FUNCTION_LINEAR;
+	}
+
+	return ret;
+}
+
+static int amdgpu_dm_atomic_blend_lut(const struct drm_color_lut *blend_lut,
+				       bool has_rom,
+				       enum dc_transfer_func_predefined tf,
+				       uint32_t blend_size,
+				       struct dc_transfer_func *func_blend)
+{
+	int ret = 0;
+
+	if (blend_size || tf != TRANSFER_FUNCTION_LINEAR) {
+		/*
+		 * DRM plane gamma LUT or TF means we are linearizing color
+		 * space before blending (similar to degamma programming). As
+		 * we don't have hardcoded curve support, or we use AMD color
+		 * module to fill the parameters that will be translated to HW
+		 * points.
+		 */
+		func_blend->type = TF_TYPE_DISTRIBUTED_POINTS;
+		func_blend->tf = tf;
+		func_blend->sdr_ref_white_level = SDR_WHITE_LEVEL_INIT_VALUE;
+
+		ret = __set_input_tf(NULL, func_blend, blend_lut, blend_size);
+	} else {
+		func_blend->type = TF_TYPE_BYPASS;
+		func_blend->tf = TRANSFER_FUNCTION_LINEAR;
+	}
+
+	return ret;
+}
+
+/**
+ * amdgpu_dm_verify_lut3d_size - verifies if 3D LUT is supported and if user
+ * shaper and 3D LUTs match the hw supported size
+ * @adev: amdgpu device
+ * @plane_state: the DRM plane state
+ *
+ * Verifies if pre-blending (DPP) 3D LUT is supported by the HW (DCN 2.0 or
+ * newer) and if the user shaper and 3D LUTs match the supported size.
+ *
+ * Returns:
+ * 0 on success. -EINVAL if lut size are invalid.
+ */
+int amdgpu_dm_verify_lut3d_size(struct amdgpu_device *adev,
+				struct drm_plane_state *plane_state)
+{
+	struct dm_plane_state *dm_plane_state = to_dm_plane_state(plane_state);
+	const struct drm_color_lut *shaper = NULL, *lut3d = NULL;
+	uint32_t exp_size, size, dim_size = MAX_COLOR_3DLUT_SIZE;
+	bool has_3dlut = adev->dm.dc->caps.color.dpp.hw_3d_lut || adev->dm.dc->caps.color.mpc.preblend;
+
+	/* shaper LUT is only available if 3D LUT color caps */
+	exp_size = has_3dlut ? MAX_COLOR_LUT_ENTRIES : 0;
+	shaper = __extract_blob_lut(dm_plane_state->shaper_lut, &size);
+
+	if (shaper && size != exp_size) {
+		drm_dbg(&adev->ddev,
+			"Invalid Shaper LUT size. Should be %u but got %u.\n",
+			exp_size, size);
+		return -EINVAL;
+	}
+
+	/* The number of 3D LUT entries is the dimension size cubed */
+	exp_size = has_3dlut ? dim_size * dim_size * dim_size : 0;
+	lut3d = __extract_blob_lut(dm_plane_state->lut3d, &size);
+
+	if (lut3d && size != exp_size) {
+		drm_dbg(&adev->ddev,
+			"Invalid 3D LUT size. Should be %u but got %u.\n",
+			exp_size, size);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_dm_verify_lut_sizes - verifies if DRM luts match the hw supported sizes
+ * @crtc_state: the DRM CRTC state
+ *
+ * Verifies that the Degamma and Gamma LUTs attached to the &crtc_state
+ * are of the expected size.
+ *
+ * Returns:
+ * 0 on success. -EINVAL if any lut sizes are invalid.
+ */
+int amdgpu_dm_verify_lut_sizes(const struct drm_crtc_state *crtc_state)
+{
+	const struct drm_color_lut *lut = NULL;
+	uint32_t size = 0;
+
+	lut = __extract_blob_lut(crtc_state->degamma_lut, &size);
+	if (lut && size != MAX_COLOR_LUT_ENTRIES) {
+		DRM_DEBUG_DRIVER(
+			"Invalid Degamma LUT size. Should be %u but got %u.\n",
+			MAX_COLOR_LUT_ENTRIES, size);
+		return -EINVAL;
+	}
+
+	lut = __extract_blob_lut(crtc_state->gamma_lut, &size);
+	if (lut && size != MAX_COLOR_LUT_ENTRIES &&
+	    size != MAX_COLOR_LEGACY_LUT_ENTRIES) {
+		DRM_DEBUG_DRIVER(
+			"Invalid Gamma LUT size. Should be %u (or %u for legacy) but got %u.\n",
+			MAX_COLOR_LUT_ENTRIES, MAX_COLOR_LEGACY_LUT_ENTRIES,
+			size);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_dm_check_crtc_color_mgmt: Check if DRM color props are programmable by DC.
+ * @crtc: amdgpu_dm crtc state
+ * @check_only: only check color state without update dc stream
+ *
+ * This function just verifies CRTC LUT sizes, if there is enough space for
+ * output transfer function and if its parameters can be calculated by AMD
+ * color module. It also adjusts some settings for programming CRTC degamma at
+ * plane stage, using plane DGM block.
+ *
+ * The RGM block is typically more fully featured and accurate across
+ * all ASICs - DCE can't support a custom non-linear CRTC DGM.
+ *
+ * For supporting both plane level color management and CRTC level color
+ * management at once we have to either restrict the usage of some CRTC
+ * properties or blend adjustments together.
+ *
+ * Returns:
+ * 0 on success. Error code if validation fails.
+ */
+
+int amdgpu_dm_check_crtc_color_mgmt(struct dm_crtc_state *crtc,
+				    bool check_only)
+{
+	struct dc_stream_state *stream = crtc->stream;
+	struct amdgpu_device *adev = drm_to_adev(crtc->base.state->dev);
+	bool has_rom = adev->asic_type <= CHIP_RAVEN;
+	struct dc_transfer_func *out_tf;
+	const struct drm_color_lut *degamma_lut, *regamma_lut;
+	uint32_t degamma_size, regamma_size;
+	bool has_regamma, has_degamma;
+	enum dc_transfer_func_predefined tf = TRANSFER_FUNCTION_LINEAR;
+	bool is_legacy;
+	int r;
+
+	tf = amdgpu_tf_to_dc_tf(crtc->regamma_tf);
+
+	r = amdgpu_dm_verify_lut_sizes(&crtc->base);
+	if (r)
+		return r;
+
+	degamma_lut = __extract_blob_lut(crtc->base.degamma_lut, &degamma_size);
+	regamma_lut = __extract_blob_lut(crtc->base.gamma_lut, &regamma_size);
+
+	has_degamma =
+		degamma_lut && !__is_lut_linear(degamma_lut, degamma_size);
+
+	has_regamma =
+		regamma_lut && !__is_lut_linear(regamma_lut, regamma_size);
+
+	is_legacy = regamma_size == MAX_COLOR_LEGACY_LUT_ENTRIES;
+
+	/* Reset all adjustments. */
+	crtc->cm_has_degamma = false;
+	crtc->cm_is_degamma_srgb = false;
+
+	if (check_only) {
+		out_tf = kvzalloc(sizeof(*out_tf), GFP_KERNEL);
+		if (!out_tf)
+			return -ENOMEM;
+	} else {
+		out_tf = &stream->out_transfer_func;
+	}
+
+	/* Setup regamma and degamma. */
+	if (is_legacy) {
+		/*
+		 * Legacy regamma forces us to use the sRGB RGM as a base.
+		 * This also means we can't use linear DGM since DGM needs
+		 * to use sRGB as a base as well, resulting in incorrect CRTC
+		 * DGM and CRTC CTM.
+		 *
+		 * TODO: Just map this to the standard regamma interface
+		 * instead since this isn't really right. One of the cases
+		 * where this setup currently fails is trying to do an
+		 * inverse color ramp in legacy userspace.
+		 */
+		crtc->cm_is_degamma_srgb = true;
+		out_tf->type = TF_TYPE_DISTRIBUTED_POINTS;
+		out_tf->tf = TRANSFER_FUNCTION_SRGB;
+		/*
+		 * Note: although we pass has_rom as parameter here, we never
+		 * actually use ROM because the color module only takes the ROM
+		 * path if transfer_func->type == PREDEFINED.
+		 *
+		 * See more in mod_color_calculate_regamma_params()
+		 */
+		r = __set_legacy_tf(out_tf, regamma_lut,
+				    regamma_size, has_rom);
+	} else {
+		regamma_size = has_regamma ? regamma_size : 0;
+		r = amdgpu_dm_set_atomic_regamma(out_tf, regamma_lut,
+						 regamma_size, has_rom, tf);
+	}
+
+	/*
+	 * CRTC DGM goes into DGM LUT. It would be nice to place it
+	 * into the RGM since it's a more featured block but we'd
+	 * have to place the CTM in the OCSC in that case.
+	 */
+	crtc->cm_has_degamma = has_degamma;
+	if (check_only)
+		kvfree(out_tf);
+
+	return r;
+}
+
+/**
+ * amdgpu_dm_update_crtc_color_mgmt: Maps DRM color management to DC stream.
+ * @crtc: amdgpu_dm crtc state
+ *
+ * With no plane level color management properties we're free to use any
+ * of the HW blocks as long as the CRTC CTM always comes before the
+ * CRTC RGM and after the CRTC DGM.
+ *
+ * - The CRTC RGM block will be placed in the RGM LUT block if it is non-linear.
+ * - The CRTC DGM block will be placed in the DGM LUT block if it is non-linear.
+ * - The CRTC CTM will be placed in the gamut remap block if it is non-linear.
+ *
+ * The RGM block is typically more fully featured and accurate across
+ * all ASICs - DCE can't support a custom non-linear CRTC DGM.
+ *
+ * For supporting both plane level color management and CRTC level color
+ * management at once we have to either restrict the usage of CRTC properties
+ * or blend adjustments together.
+ *
+ * Returns:
+ * 0 on success. Error code if setup fails.
+ */
+int amdgpu_dm_update_crtc_color_mgmt(struct dm_crtc_state *crtc)
+{
+	struct dc_stream_state *stream = crtc->stream;
+	struct drm_color_ctm *ctm = NULL;
+	int ret;
+
+	ret = amdgpu_dm_check_crtc_color_mgmt(crtc, false);
+	if (ret)
+		return ret;
+
+	/* Setup CRTC CTM. */
+	if (crtc->base.ctm) {
+		ctm = (struct drm_color_ctm *)crtc->base.ctm->data;
+
+		/*
+		 * Gamut remapping must be used for gamma correction
+		 * since it comes before the regamma correction.
+		 *
+		 * OCSC could be used for gamma correction, but we'd need to
+		 * blend the adjustments together with the required output
+		 * conversion matrix - so just use the gamut remap block
+		 * for now.
+		 */
+		__drm_ctm_to_dc_matrix(ctm, stream->gamut_remap_matrix.matrix);
+
+		stream->gamut_remap_matrix.enable_remap = true;
+		stream->csc_color_matrix.enable_adjustment = false;
+	} else {
+		/* Bypass CTM. */
+		stream->gamut_remap_matrix.enable_remap = false;
+		stream->csc_color_matrix.enable_adjustment = false;
+	}
+
+	return 0;
+}
+
+static int
+map_crtc_degamma_to_dc_plane(struct dm_crtc_state *crtc,
+			     struct dc_plane_state *dc_plane_state,
+			     struct dc_color_caps *caps)
+{
+	const struct drm_color_lut *degamma_lut;
+	enum dc_transfer_func_predefined tf = TRANSFER_FUNCTION_SRGB;
+	uint32_t degamma_size;
+	int r;
+
+	/* Get the correct base transfer function for implicit degamma. */
+	switch (dc_plane_state->format) {
+	case SURFACE_PIXEL_FORMAT_VIDEO_420_YCbCr:
+	case SURFACE_PIXEL_FORMAT_VIDEO_420_YCrCb:
+		/* DC doesn't have a transfer function for BT601 specifically. */
+		tf = TRANSFER_FUNCTION_BT709;
+		break;
+	default:
+		break;
+	}
+
+	if (crtc->cm_has_degamma) {
+		degamma_lut = __extract_blob_lut(crtc->base.degamma_lut,
+						 &degamma_size);
+		ASSERT(degamma_size == MAX_COLOR_LUT_ENTRIES);
+
+		dc_plane_state->in_transfer_func.type = TF_TYPE_DISTRIBUTED_POINTS;
+
+		/*
+		 * This case isn't fully correct, but also fairly
+		 * uncommon. This is userspace trying to use a
+		 * legacy gamma LUT + atomic degamma LUT
+		 * at the same time.
+		 *
+		 * Legacy gamma requires the input to be in linear
+		 * space, so that means we need to apply an sRGB
+		 * degamma. But color module also doesn't support
+		 * a user ramp in this case so the degamma will
+		 * be lost.
+		 *
+		 * Even if we did support it, it's still not right:
+		 *
+		 * Input -> CRTC DGM -> sRGB DGM -> CRTC CTM ->
+		 * sRGB RGM -> CRTC RGM -> Output
+		 *
+		 * The CSC will be done in the wrong space since
+		 * we're applying an sRGB DGM on top of the CRTC
+		 * DGM.
+		 *
+		 * TODO: Don't use the legacy gamma interface and just
+		 * map these to the atomic one instead.
+		 */
+		if (crtc->cm_is_degamma_srgb)
+			dc_plane_state->in_transfer_func.tf = tf;
+		else
+			dc_plane_state->in_transfer_func.tf =
+				TRANSFER_FUNCTION_LINEAR;
+
+		r = __set_input_tf(caps, &dc_plane_state->in_transfer_func,
+				   degamma_lut, degamma_size);
+		if (r)
+			return r;
+	} else {
+		/*
+		 * For legacy gamma support we need the regamma input
+		 * in linear space. Assume that the input is sRGB.
+		 */
+		dc_plane_state->in_transfer_func.type = TF_TYPE_PREDEFINED;
+		dc_plane_state->in_transfer_func.tf = tf;
+
+		if (tf != TRANSFER_FUNCTION_SRGB &&
+		    !mod_color_calculate_degamma_params(caps,
+							&dc_plane_state->in_transfer_func,
+							NULL, false))
+			return -ENOMEM;
+	}
+
+	return 0;
+}
+
+static int
+__set_dm_plane_degamma(struct drm_plane_state *plane_state,
+		       struct dc_plane_state *dc_plane_state,
+		       struct dc_color_caps *color_caps)
+{
+	struct dm_plane_state *dm_plane_state = to_dm_plane_state(plane_state);
+	const struct drm_color_lut *degamma_lut;
+	enum amdgpu_transfer_function tf = AMDGPU_TRANSFER_FUNCTION_DEFAULT;
+	uint32_t degamma_size;
+	bool has_degamma_lut;
+	int ret;
+
+	degamma_lut = __extract_blob_lut(dm_plane_state->degamma_lut,
+					 &degamma_size);
+
+	has_degamma_lut = degamma_lut &&
+			  !__is_lut_linear(degamma_lut, degamma_size);
+
+	tf = dm_plane_state->degamma_tf;
+
+	/* If we don't have plane degamma LUT nor TF to set on DC, we have
+	 * nothing to do here, return.
+	 */
+	if (!has_degamma_lut && tf == AMDGPU_TRANSFER_FUNCTION_DEFAULT)
+		return -EINVAL;
+
+	dc_plane_state->in_transfer_func.tf = amdgpu_tf_to_dc_tf(tf);
+
+	if (has_degamma_lut) {
+		ASSERT(degamma_size == MAX_COLOR_LUT_ENTRIES);
+
+		dc_plane_state->in_transfer_func.type =
+			TF_TYPE_DISTRIBUTED_POINTS;
+
+		ret = __set_input_tf(color_caps, &dc_plane_state->in_transfer_func,
+				     degamma_lut, degamma_size);
+		if (ret)
+			return ret;
+       } else {
+		dc_plane_state->in_transfer_func.type =
+			TF_TYPE_PREDEFINED;
+
+		if (!mod_color_calculate_degamma_params(color_caps,
+		    &dc_plane_state->in_transfer_func, NULL, false))
+			return -ENOMEM;
+	}
+	return 0;
+}
+
+static int
+amdgpu_dm_plane_set_color_properties(struct drm_plane_state *plane_state,
+				     struct dc_plane_state *dc_plane_state)
+{
+	struct dm_plane_state *dm_plane_state = to_dm_plane_state(plane_state);
+	enum amdgpu_transfer_function shaper_tf = AMDGPU_TRANSFER_FUNCTION_DEFAULT;
+	enum amdgpu_transfer_function blend_tf = AMDGPU_TRANSFER_FUNCTION_DEFAULT;
+	const struct drm_color_lut *shaper_lut, *lut3d, *blend_lut;
+	uint32_t shaper_size, lut3d_size, blend_size;
+	int ret;
+
+	dc_plane_state->hdr_mult = amdgpu_dm_fixpt_from_s3132(dm_plane_state->hdr_mult);
+
+	shaper_lut = __extract_blob_lut(dm_plane_state->shaper_lut, &shaper_size);
+	shaper_size = shaper_lut != NULL ? shaper_size : 0;
+	shaper_tf = dm_plane_state->shaper_tf;
+	lut3d = __extract_blob_lut(dm_plane_state->lut3d, &lut3d_size);
+	lut3d_size = lut3d != NULL ? lut3d_size : 0;
+
+	amdgpu_dm_atomic_lut3d(lut3d, lut3d_size, &dc_plane_state->lut3d_func);
+	ret = amdgpu_dm_atomic_shaper_lut(shaper_lut, false,
+					  amdgpu_tf_to_dc_tf(shaper_tf),
+					  shaper_size,
+					  &dc_plane_state->in_shaper_func);
+	if (ret) {
+		drm_dbg_kms(plane_state->plane->dev,
+			    "setting plane %d shaper LUT failed.\n",
+			    plane_state->plane->index);
+
+		return ret;
+	}
+
+	blend_tf = dm_plane_state->blend_tf;
+	blend_lut = __extract_blob_lut(dm_plane_state->blend_lut, &blend_size);
+	blend_size = blend_lut != NULL ? blend_size : 0;
+
+	ret = amdgpu_dm_atomic_blend_lut(blend_lut, false,
+					 amdgpu_tf_to_dc_tf(blend_tf),
+					 blend_size, &dc_plane_state->blend_tf);
+	if (ret) {
+		drm_dbg_kms(plane_state->plane->dev,
+			    "setting plane %d gamma lut failed.\n",
+			    plane_state->plane->index);
+
+		return ret;
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_dm_update_plane_color_mgmt: Maps DRM color management to DC plane.
+ * @crtc: amdgpu_dm crtc state
+ * @plane_state: DRM plane state
+ * @dc_plane_state: target DC surface
+ *
+ * Update the underlying dc_stream_state's input transfer function (ITF) in
+ * preparation for hardware commit. The transfer function used depends on
+ * the preparation done on the stream for color management.
+ *
+ * Returns:
+ * 0 on success. -ENOMEM if mem allocation fails.
+ */
+int amdgpu_dm_update_plane_color_mgmt(struct dm_crtc_state *crtc,
+				      struct drm_plane_state *plane_state,
+				      struct dc_plane_state *dc_plane_state)
+{
+	struct amdgpu_device *adev = drm_to_adev(crtc->base.state->dev);
+	struct dm_plane_state *dm_plane_state = to_dm_plane_state(plane_state);
+	struct drm_color_ctm_3x4 *ctm = NULL;
+	struct dc_color_caps *color_caps = NULL;
+	bool has_crtc_cm_degamma;
+	int ret;
+
+	ret = amdgpu_dm_verify_lut3d_size(adev, plane_state);
+	if (ret) {
+		drm_dbg_driver(&adev->ddev, "amdgpu_dm_verify_lut3d_size() failed\n");
+		return ret;
+	}
+
+	if (dc_plane_state->ctx && dc_plane_state->ctx->dc)
+		color_caps = &dc_plane_state->ctx->dc->caps.color;
+
+	/* Initially, we can just bypass the DGM block. */
+	dc_plane_state->in_transfer_func.type = TF_TYPE_BYPASS;
+	dc_plane_state->in_transfer_func.tf = TRANSFER_FUNCTION_LINEAR;
+
+	/* After, we start to update values according to color props */
+	has_crtc_cm_degamma = (crtc->cm_has_degamma || crtc->cm_is_degamma_srgb);
+
+	ret = __set_dm_plane_degamma(plane_state, dc_plane_state, color_caps);
+	if (ret == -ENOMEM)
+		return ret;
+
+	/* We only have one degamma block available (pre-blending) for the
+	 * whole color correction pipeline, so that we can't actually perform
+	 * plane and CRTC degamma at the same time. Explicitly reject atomic
+	 * updates when userspace sets both plane and CRTC degamma properties.
+	 */
+	if (has_crtc_cm_degamma && ret != -EINVAL) {
+		drm_dbg_kms(crtc->base.crtc->dev,
+			    "doesn't support plane and CRTC degamma at the same time\n");
+		return -EINVAL;
+	}
+
+	/* If we are here, it means we don't have plane degamma settings, check
+	 * if we have CRTC degamma waiting for mapping to pre-blending degamma
+	 * block
+	 */
+	if (has_crtc_cm_degamma) {
+		/*
+		 * AMD HW doesn't have post-blending degamma caps. When DRM
+		 * CRTC atomic degamma is set, we maps it to DPP degamma block
+		 * (pre-blending) or, on legacy gamma, we use DPP degamma to
+		 * linearize (implicit degamma) from sRGB/BT709 according to
+		 * the input space.
+		 */
+		ret = map_crtc_degamma_to_dc_plane(crtc, dc_plane_state, color_caps);
+		if (ret)
+			return ret;
+	}
+
+	/* Setup CRTC CTM. */
+	if (dm_plane_state->ctm) {
+		ctm = (struct drm_color_ctm_3x4 *)dm_plane_state->ctm->data;
+		/*
+		 * DCN2 and older don't support both pre-blending and
+		 * post-blending gamut remap. For this HW family, if we have
+		 * the plane and CRTC CTMs simultaneously, CRTC CTM takes
+		 * priority, and we discard plane CTM, as implemented in
+		 * dcn10_program_gamut_remap(). However, DCN3+ has DPP
+		 * (pre-blending) and MPC (post-blending) `gamut remap` blocks;
+		 * therefore, we can program plane and CRTC CTMs together by
+		 * mapping CRTC CTM to MPC and keeping plane CTM setup at DPP,
+		 * as it's done by dcn30_program_gamut_remap().
+		 */
+		__drm_ctm_3x4_to_dc_matrix(ctm, dc_plane_state->gamut_remap_matrix.matrix);
+
+		dc_plane_state->gamut_remap_matrix.enable_remap = true;
+		dc_plane_state->input_csc_color_matrix.enable_adjustment = false;
+	} else {
+		/* Bypass CTM. */
+		dc_plane_state->gamut_remap_matrix.enable_remap = false;
+		dc_plane_state->input_csc_color_matrix.enable_adjustment = false;
+	}
+
+	return amdgpu_dm_plane_set_color_properties(plane_state, dc_plane_state);
+}
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c
new file mode 100644
index 000000000000..e20aa7438066
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c
@@ -0,0 +1,951 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include <drm/drm_crtc.h>
+#include <drm/drm_vblank.h>
+
+#include "amdgpu.h"
+#include "amdgpu_dm.h"
+#include "dc.h"
+#include "amdgpu_securedisplay.h"
+#include "amdgpu_dm_psr.h"
+
+static const char *const pipe_crc_sources[] = {
+	"none",
+	"crtc",
+	"crtc dither",
+	"dprx",
+	"dprx dither",
+	"auto",
+};
+
+static enum amdgpu_dm_pipe_crc_source dm_parse_crc_source(const char *source)
+{
+	if (!source || !strcmp(source, "none"))
+		return AMDGPU_DM_PIPE_CRC_SOURCE_NONE;
+	if (!strcmp(source, "auto") || !strcmp(source, "crtc"))
+		return AMDGPU_DM_PIPE_CRC_SOURCE_CRTC;
+	if (!strcmp(source, "dprx"))
+		return AMDGPU_DM_PIPE_CRC_SOURCE_DPRX;
+	if (!strcmp(source, "crtc dither"))
+		return AMDGPU_DM_PIPE_CRC_SOURCE_CRTC_DITHER;
+	if (!strcmp(source, "dprx dither"))
+		return AMDGPU_DM_PIPE_CRC_SOURCE_DPRX_DITHER;
+
+	return AMDGPU_DM_PIPE_CRC_SOURCE_INVALID;
+}
+
+static bool dm_is_crc_source_crtc(enum amdgpu_dm_pipe_crc_source src)
+{
+	return (src == AMDGPU_DM_PIPE_CRC_SOURCE_CRTC) ||
+	       (src == AMDGPU_DM_PIPE_CRC_SOURCE_CRTC_DITHER);
+}
+
+static bool dm_is_crc_source_dprx(enum amdgpu_dm_pipe_crc_source src)
+{
+	return (src == AMDGPU_DM_PIPE_CRC_SOURCE_DPRX) ||
+	       (src == AMDGPU_DM_PIPE_CRC_SOURCE_DPRX_DITHER);
+}
+
+static bool dm_need_crc_dither(enum amdgpu_dm_pipe_crc_source src)
+{
+	return (src == AMDGPU_DM_PIPE_CRC_SOURCE_CRTC_DITHER) ||
+	       (src == AMDGPU_DM_PIPE_CRC_SOURCE_DPRX_DITHER) ||
+	       (src == AMDGPU_DM_PIPE_CRC_SOURCE_NONE);
+}
+
+const char *const *amdgpu_dm_crtc_get_crc_sources(struct drm_crtc *crtc,
+						  size_t *count)
+{
+	*count = ARRAY_SIZE(pipe_crc_sources);
+	return pipe_crc_sources;
+}
+
+#ifdef CONFIG_DRM_AMD_SECURE_DISPLAY
+static void update_phy_id_mapping(struct amdgpu_device *adev)
+{
+	struct drm_device *ddev = adev_to_drm(adev);
+	struct amdgpu_display_manager *dm = &adev->dm;
+	struct drm_connector *connector;
+	struct amdgpu_dm_connector *aconnector;
+	struct amdgpu_dm_connector *sort_connector[AMDGPU_DM_MAX_CRTC] = {NULL};
+	struct drm_connector_list_iter iter;
+	uint8_t idx = 0, idx_2 = 0, connector_cnt = 0;
+
+	dm->secure_display_ctx.phy_mapping_updated = false;
+
+	mutex_lock(&ddev->mode_config.mutex);
+	drm_connector_list_iter_begin(ddev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
+
+		if (connector->status != connector_status_connected)
+			continue;
+
+		if (idx >= AMDGPU_DM_MAX_CRTC) {
+			DRM_WARN("%s connected connectors exceed max crtc\n", __func__);
+			mutex_unlock(&ddev->mode_config.mutex);
+			return;
+		}
+
+		aconnector = to_amdgpu_dm_connector(connector);
+
+		sort_connector[idx] = aconnector;
+		idx++;
+		connector_cnt++;
+	}
+	drm_connector_list_iter_end(&iter);
+
+	/* sort connectors by link_enc_hw_instance first */
+	for (idx = connector_cnt; idx > 1 ; idx--) {
+		for (idx_2 = 0; idx_2 < (idx - 1); idx_2++) {
+			if (sort_connector[idx_2]->dc_link->link_enc_hw_inst >
+			    sort_connector[idx_2 + 1]->dc_link->link_enc_hw_inst)
+				swap(sort_connector[idx_2], sort_connector[idx_2 + 1]);
+		}
+	}
+
+	/*
+	 * Sort mst connectors by RAD. mst connectors with the same enc_hw_instance are already
+	 * sorted together above.
+	 */
+	for (idx = 0; idx < connector_cnt; /*Do nothing*/) {
+		if (sort_connector[idx]->mst_root) {
+			uint8_t i, j, k;
+			uint8_t mst_con_cnt = 1;
+
+			for (idx_2 = (idx + 1); idx_2 < connector_cnt; idx_2++) {
+				if (sort_connector[idx_2]->mst_root == sort_connector[idx]->mst_root)
+					mst_con_cnt++;
+				else
+					break;
+			}
+
+			for (i = mst_con_cnt; i > 1; i--) {
+				for (j = idx; j < (idx + i - 2); j++) {
+					int mstb_lct = sort_connector[j]->mst_output_port->parent->lct;
+					int next_mstb_lct = sort_connector[j + 1]->mst_output_port->parent->lct;
+					u8 *rad;
+					u8 *next_rad;
+					bool swap = false;
+
+					/* Sort by mst tree depth first. Then compare RAD if depth is the same*/
+					if (mstb_lct > next_mstb_lct) {
+						swap = true;
+					} else if (mstb_lct == next_mstb_lct) {
+						if (mstb_lct == 1) {
+							if (sort_connector[j]->mst_output_port->port_num > sort_connector[j + 1]->mst_output_port->port_num)
+								swap = true;
+						} else if (mstb_lct > 1) {
+							rad = sort_connector[j]->mst_output_port->parent->rad;
+							next_rad = sort_connector[j + 1]->mst_output_port->parent->rad;
+
+							for (k = 0; k < mstb_lct - 1; k++) {
+								int shift = (k % 2) ? 0 : 4;
+								int port_num = (rad[k / 2] >> shift) & 0xf;
+								int next_port_num = (next_rad[k / 2] >> shift) & 0xf;
+
+								if (port_num > next_port_num) {
+									swap = true;
+									break;
+								}
+							}
+						} else {
+							DRM_ERROR("MST LCT shouldn't be set as < 1");
+							mutex_unlock(&ddev->mode_config.mutex);
+							return;
+						}
+					}
+
+					if (swap)
+						swap(sort_connector[j], sort_connector[j + 1]);
+				}
+			}
+
+			idx += mst_con_cnt;
+		} else {
+			idx++;
+		}
+	}
+
+	/* Complete sorting. Assign relavant result to dm->secure_display_ctx.phy_id_mapping[]*/
+	memset(dm->secure_display_ctx.phy_id_mapping, 0, sizeof(dm->secure_display_ctx.phy_id_mapping));
+	for (idx = 0; idx < connector_cnt; idx++) {
+		aconnector = sort_connector[idx];
+
+		dm->secure_display_ctx.phy_id_mapping[idx].assigned = true;
+		dm->secure_display_ctx.phy_id_mapping[idx].is_mst = false;
+		dm->secure_display_ctx.phy_id_mapping[idx].enc_hw_inst = aconnector->dc_link->link_enc_hw_inst;
+
+		if (sort_connector[idx]->mst_root) {
+			dm->secure_display_ctx.phy_id_mapping[idx].is_mst = true;
+			dm->secure_display_ctx.phy_id_mapping[idx].lct = aconnector->mst_output_port->parent->lct;
+			dm->secure_display_ctx.phy_id_mapping[idx].port_num = aconnector->mst_output_port->port_num;
+			memcpy(dm->secure_display_ctx.phy_id_mapping[idx].rad,
+				aconnector->mst_output_port->parent->rad, sizeof(aconnector->mst_output_port->parent->rad));
+		}
+	}
+	mutex_unlock(&ddev->mode_config.mutex);
+
+	dm->secure_display_ctx.phy_id_mapping_cnt = connector_cnt;
+	dm->secure_display_ctx.phy_mapping_updated = true;
+}
+
+static bool get_phy_id(struct amdgpu_display_manager *dm,
+			struct amdgpu_dm_connector *aconnector, uint8_t *phy_id)
+{
+	int idx, idx_2;
+	bool found = false;
+
+	/*
+	 * Assume secure display start after all connectors are probed. The connection
+	 * config is static as well
+	 */
+	if (!dm->secure_display_ctx.phy_mapping_updated) {
+		DRM_WARN("%s Should update the phy id table before get it's value", __func__);
+		return false;
+	}
+
+	for (idx = 0; idx < dm->secure_display_ctx.phy_id_mapping_cnt; idx++) {
+		if (!dm->secure_display_ctx.phy_id_mapping[idx].assigned) {
+			DRM_ERROR("phy_id_mapping[%d] should be assigned", idx);
+			return false;
+		}
+
+		if (aconnector->dc_link->link_enc_hw_inst ==
+				dm->secure_display_ctx.phy_id_mapping[idx].enc_hw_inst) {
+			if (!dm->secure_display_ctx.phy_id_mapping[idx].is_mst) {
+				found = true;
+				goto out;
+			} else {
+				/* Could caused by wrongly pass mst root connector */
+				if (!aconnector->mst_output_port) {
+					DRM_ERROR("%s Check mst case but connector without a port assigned", __func__);
+					return false;
+				}
+
+				if (aconnector->mst_root &&
+					aconnector->mst_root->mst_mgr.mst_primary == NULL) {
+					DRM_WARN("%s pass in a stale mst connector", __func__);
+				}
+
+				if (aconnector->mst_output_port->parent->lct == dm->secure_display_ctx.phy_id_mapping[idx].lct &&
+					aconnector->mst_output_port->port_num == dm->secure_display_ctx.phy_id_mapping[idx].port_num) {
+					if (aconnector->mst_output_port->parent->lct == 1) {
+						found = true;
+						goto out;
+					} else if (aconnector->mst_output_port->parent->lct > 1) {
+						/* Check RAD */
+						for (idx_2 = 0; idx_2 < aconnector->mst_output_port->parent->lct - 1; idx_2++) {
+							int shift = (idx_2 % 2) ? 0 : 4;
+							int port_num = (aconnector->mst_output_port->parent->rad[idx_2 / 2] >> shift) & 0xf;
+							int port_num2 = (dm->secure_display_ctx.phy_id_mapping[idx].rad[idx_2 / 2] >> shift) & 0xf;
+
+							if (port_num != port_num2)
+								break;
+						}
+
+						if (idx_2 == aconnector->mst_output_port->parent->lct - 1) {
+							found = true;
+							goto out;
+						}
+					} else {
+						DRM_ERROR("lCT should be >= 1");
+						return false;
+					}
+				}
+			}
+		}
+	}
+
+out:
+	if (found) {
+		DRM_DEBUG_DRIVER("Associated secure display PHY ID as %d", idx);
+		*phy_id = idx;
+	} else {
+		DRM_WARN("Can't find associated phy ID");
+		return false;
+	}
+
+	return true;
+}
+
+static void amdgpu_dm_set_crc_window_default(struct drm_crtc *crtc, struct dc_stream_state *stream)
+{
+	struct drm_device *drm_dev = crtc->dev;
+	struct amdgpu_display_manager *dm = &drm_to_adev(drm_dev)->dm;
+	struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_dm_connector *aconnector;
+	bool was_activated;
+	uint8_t phy_id;
+	unsigned long flags;
+	int i;
+
+	spin_lock_irqsave(&drm_dev->event_lock, flags);
+	was_activated = acrtc->dm_irq_params.crc_window_activated;
+	for (i = 0; i < MAX_CRC_WINDOW_NUM; i++) {
+		acrtc->dm_irq_params.window_param[i].x_start = 0;
+		acrtc->dm_irq_params.window_param[i].y_start = 0;
+		acrtc->dm_irq_params.window_param[i].x_end = 0;
+		acrtc->dm_irq_params.window_param[i].y_end = 0;
+		acrtc->dm_irq_params.window_param[i].enable = false;
+		acrtc->dm_irq_params.window_param[i].update_win = false;
+		acrtc->dm_irq_params.window_param[i].skip_frame_cnt = 0;
+	}
+	acrtc->dm_irq_params.crc_window_activated = false;
+	spin_unlock_irqrestore(&drm_dev->event_lock, flags);
+
+	/* Disable secure_display if it was enabled */
+	if (was_activated && dm->secure_display_ctx.op_mode == LEGACY_MODE) {
+		/* stop ROI update on this crtc */
+		flush_work(&dm->secure_display_ctx.crtc_ctx[crtc->index].notify_ta_work);
+		flush_work(&dm->secure_display_ctx.crtc_ctx[crtc->index].forward_roi_work);
+		aconnector = (struct amdgpu_dm_connector *)stream->dm_stream_context;
+
+		if (aconnector && get_phy_id(dm, aconnector, &phy_id)) {
+			if (dm->secure_display_ctx.support_mul_roi)
+				dc_stream_forward_multiple_crc_window(stream, NULL, phy_id, true);
+			else
+				dc_stream_forward_crc_window(stream, NULL, phy_id, true);
+		} else {
+			DRM_DEBUG_DRIVER("%s Can't find matching phy id", __func__);
+		}
+	}
+}
+
+static void amdgpu_dm_crtc_notify_ta_to_read(struct work_struct *work)
+{
+	struct secure_display_crtc_context *crtc_ctx;
+	struct psp_context *psp;
+	struct ta_securedisplay_cmd *securedisplay_cmd;
+	struct drm_crtc *crtc;
+	struct dc_stream_state *stream;
+	struct amdgpu_dm_connector *aconnector;
+	uint8_t phy_inst;
+	struct amdgpu_display_manager *dm;
+	struct crc_data crc_cpy[MAX_CRC_WINDOW_NUM];
+	unsigned long flags;
+	uint8_t roi_idx = 0;
+	int ret;
+	int i;
+
+	crtc_ctx = container_of(work, struct secure_display_crtc_context, notify_ta_work);
+	crtc = crtc_ctx->crtc;
+
+	if (!crtc)
+		return;
+
+	psp = &drm_to_adev(crtc->dev)->psp;
+
+	if (!psp->securedisplay_context.context.initialized) {
+		DRM_DEBUG_DRIVER("Secure Display fails to notify PSP TA\n");
+		return;
+	}
+
+	dm = &drm_to_adev(crtc->dev)->dm;
+	stream = to_amdgpu_crtc(crtc)->dm_irq_params.stream;
+	aconnector = (struct amdgpu_dm_connector *)stream->dm_stream_context;
+	if (!aconnector)
+		return;
+
+	mutex_lock(&crtc->dev->mode_config.mutex);
+	if (!get_phy_id(dm, aconnector, &phy_inst)) {
+		DRM_WARN("%s Can't find mapping phy id!", __func__);
+		mutex_unlock(&crtc->dev->mode_config.mutex);
+		return;
+	}
+	mutex_unlock(&crtc->dev->mode_config.mutex);
+
+	spin_lock_irqsave(&crtc->dev->event_lock, flags);
+	memcpy(crc_cpy, crtc_ctx->crc_info.crc, sizeof(struct crc_data) * MAX_CRC_WINDOW_NUM);
+	spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
+
+	/* need lock for multiple crtcs to use the command buffer */
+	mutex_lock(&psp->securedisplay_context.mutex);
+	/* PSP TA is expected to finish data transmission over I2C within current frame,
+	 * even there are up to 4 crtcs request to send in this frame.
+	 */
+	if (dm->secure_display_ctx.support_mul_roi) {
+		psp_prep_securedisplay_cmd_buf(psp, &securedisplay_cmd,
+							TA_SECUREDISPLAY_COMMAND__SEND_ROI_CRC_V2);
+
+		securedisplay_cmd->securedisplay_in_message.send_roi_crc_v2.phy_id = phy_inst;
+
+		for (i = 0; i < MAX_CRC_WINDOW_NUM; i++) {
+			if (crc_cpy[i].crc_ready)
+				roi_idx |= 1 << i;
+		}
+		securedisplay_cmd->securedisplay_in_message.send_roi_crc_v2.roi_idx = roi_idx;
+
+		ret = psp_securedisplay_invoke(psp, TA_SECUREDISPLAY_COMMAND__SEND_ROI_CRC_V2);
+	} else {
+		psp_prep_securedisplay_cmd_buf(psp, &securedisplay_cmd,
+							TA_SECUREDISPLAY_COMMAND__SEND_ROI_CRC);
+
+		securedisplay_cmd->securedisplay_in_message.send_roi_crc.phy_id = phy_inst;
+
+		ret = psp_securedisplay_invoke(psp, TA_SECUREDISPLAY_COMMAND__SEND_ROI_CRC);
+	}
+
+	if (!ret) {
+		if (securedisplay_cmd->status != TA_SECUREDISPLAY_STATUS__SUCCESS)
+			psp_securedisplay_parse_resp_status(psp, securedisplay_cmd->status);
+	}
+
+	mutex_unlock(&psp->securedisplay_context.mutex);
+}
+
+static void
+amdgpu_dm_forward_crc_window(struct work_struct *work)
+{
+	struct secure_display_crtc_context *crtc_ctx;
+	struct amdgpu_display_manager *dm;
+	struct drm_crtc *crtc;
+	struct dc_stream_state *stream;
+	struct amdgpu_dm_connector *aconnector;
+	struct crc_window roi_cpy[MAX_CRC_WINDOW_NUM];
+	unsigned long flags;
+	uint8_t phy_id;
+
+	crtc_ctx = container_of(work, struct secure_display_crtc_context, forward_roi_work);
+	crtc = crtc_ctx->crtc;
+
+	if (!crtc)
+		return;
+
+	dm = &drm_to_adev(crtc->dev)->dm;
+	stream = to_amdgpu_crtc(crtc)->dm_irq_params.stream;
+	aconnector = (struct amdgpu_dm_connector *)stream->dm_stream_context;
+
+	if (!aconnector)
+		return;
+
+	mutex_lock(&crtc->dev->mode_config.mutex);
+	if (!get_phy_id(dm, aconnector, &phy_id)) {
+		DRM_WARN("%s Can't find mapping phy id!", __func__);
+		mutex_unlock(&crtc->dev->mode_config.mutex);
+		return;
+	}
+	mutex_unlock(&crtc->dev->mode_config.mutex);
+
+	spin_lock_irqsave(&crtc->dev->event_lock, flags);
+	memcpy(roi_cpy, crtc_ctx->roi, sizeof(struct crc_window) * MAX_CRC_WINDOW_NUM);
+	spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
+
+	mutex_lock(&dm->dc_lock);
+	if (dm->secure_display_ctx.support_mul_roi)
+		dc_stream_forward_multiple_crc_window(stream, roi_cpy,
+			phy_id, false);
+	else
+		dc_stream_forward_crc_window(stream, &roi_cpy[0].rect,
+			phy_id, false);
+	mutex_unlock(&dm->dc_lock);
+}
+
+bool amdgpu_dm_crc_window_is_activated(struct drm_crtc *crtc)
+{
+	struct drm_device *drm_dev = crtc->dev;
+	struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
+	bool ret = false;
+
+	spin_lock_irq(&drm_dev->event_lock);
+	ret = acrtc->dm_irq_params.crc_window_activated;
+	spin_unlock_irq(&drm_dev->event_lock);
+
+	return ret;
+}
+#endif
+
+int
+amdgpu_dm_crtc_verify_crc_source(struct drm_crtc *crtc, const char *src_name,
+				 size_t *values_cnt)
+{
+	enum amdgpu_dm_pipe_crc_source source = dm_parse_crc_source(src_name);
+
+	if (source < 0) {
+		DRM_DEBUG_DRIVER("Unknown CRC source %s for CRTC%d\n",
+				 src_name, crtc->index);
+		return -EINVAL;
+	}
+
+	*values_cnt = 3;
+	return 0;
+}
+
+int amdgpu_dm_crtc_configure_crc_source(struct drm_crtc *crtc,
+					struct dm_crtc_state *dm_crtc_state,
+					enum amdgpu_dm_pipe_crc_source source)
+{
+	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
+	struct dc_stream_state *stream_state = dm_crtc_state->stream;
+	bool enable = amdgpu_dm_is_valid_crc_source(source);
+	int ret = 0;
+
+	/* Configuration will be deferred to stream enable. */
+	if (!stream_state)
+		return -EINVAL;
+
+	mutex_lock(&adev->dm.dc_lock);
+
+	/* For PSR1, check that the panel has exited PSR */
+	if (stream_state->link->psr_settings.psr_version < DC_PSR_VERSION_SU_1)
+		amdgpu_dm_psr_wait_disable(stream_state);
+
+	/* Enable or disable CRTC CRC generation */
+	if (dm_is_crc_source_crtc(source) || source == AMDGPU_DM_PIPE_CRC_SOURCE_NONE) {
+		if (!dc_stream_configure_crc(stream_state->ctx->dc,
+					     stream_state, NULL, enable, enable, 0, true)) {
+			ret = -EINVAL;
+			goto unlock;
+		}
+	}
+
+	/* Configure dithering */
+	if (!dm_need_crc_dither(source)) {
+		dc_stream_set_dither_option(stream_state, DITHER_OPTION_TRUN8);
+		dc_stream_set_dyn_expansion(stream_state->ctx->dc, stream_state,
+					    DYN_EXPANSION_DISABLE);
+	} else {
+		dc_stream_set_dither_option(stream_state,
+					    DITHER_OPTION_DEFAULT);
+		dc_stream_set_dyn_expansion(stream_state->ctx->dc, stream_state,
+					    DYN_EXPANSION_AUTO);
+	}
+
+unlock:
+	mutex_unlock(&adev->dm.dc_lock);
+
+	return ret;
+}
+
+int amdgpu_dm_crtc_set_crc_source(struct drm_crtc *crtc, const char *src_name)
+{
+	enum amdgpu_dm_pipe_crc_source source = dm_parse_crc_source(src_name);
+	enum amdgpu_dm_pipe_crc_source cur_crc_src;
+	struct drm_crtc_commit *commit;
+	struct dm_crtc_state *crtc_state;
+	struct drm_device *drm_dev = crtc->dev;
+#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY)
+	struct amdgpu_device *adev = drm_to_adev(drm_dev);
+	struct amdgpu_display_manager *dm = &adev->dm;
+#endif
+	struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
+	struct drm_dp_aux *aux = NULL;
+	bool enable = false;
+	bool enabled = false;
+	int ret = 0;
+
+	if (source < 0) {
+		DRM_DEBUG_DRIVER("Unknown CRC source %s for CRTC%d\n",
+				 src_name, crtc->index);
+		return -EINVAL;
+	}
+
+	ret = drm_modeset_lock(&crtc->mutex, NULL);
+	if (ret)
+		return ret;
+
+	spin_lock(&crtc->commit_lock);
+	commit = list_first_entry_or_null(&crtc->commit_list,
+					  struct drm_crtc_commit, commit_entry);
+	if (commit)
+		drm_crtc_commit_get(commit);
+	spin_unlock(&crtc->commit_lock);
+
+	if (commit) {
+		/*
+		 * Need to wait for all outstanding programming to complete
+		 * in commit tail since it can modify CRC related fields and
+		 * hardware state. Since we're holding the CRTC lock we're
+		 * guaranteed that no other commit work can be queued off
+		 * before we modify the state below.
+		 */
+		ret = wait_for_completion_interruptible_timeout(
+			&commit->hw_done, 10 * HZ);
+		if (ret)
+			goto cleanup;
+	}
+
+	enable = amdgpu_dm_is_valid_crc_source(source);
+	crtc_state = to_dm_crtc_state(crtc->state);
+	spin_lock_irq(&drm_dev->event_lock);
+	cur_crc_src = acrtc->dm_irq_params.crc_src;
+	spin_unlock_irq(&drm_dev->event_lock);
+
+	/*
+	 * USER REQ SRC | CURRENT SRC | BEHAVIOR
+	 * -----------------------------
+	 * None         | None        | Do nothing
+	 * None         | CRTC        | Disable CRTC CRC, set default to dither
+	 * None         | DPRX        | Disable DPRX CRC, need 'aux', set default to dither
+	 * None         | CRTC DITHER | Disable CRTC CRC
+	 * None         | DPRX DITHER | Disable DPRX CRC, need 'aux'
+	 * CRTC         | XXXX        | Enable CRTC CRC, no dither
+	 * DPRX         | XXXX        | Enable DPRX CRC, need 'aux', no dither
+	 * CRTC DITHER  | XXXX        | Enable CRTC CRC, set dither
+	 * DPRX DITHER  | XXXX        | Enable DPRX CRC, need 'aux', set dither
+	 */
+	if (dm_is_crc_source_dprx(source) ||
+	    (source == AMDGPU_DM_PIPE_CRC_SOURCE_NONE &&
+	     dm_is_crc_source_dprx(cur_crc_src))) {
+		struct amdgpu_dm_connector *aconn = NULL;
+		struct drm_connector *connector;
+		struct drm_connector_list_iter conn_iter;
+
+		drm_connector_list_iter_begin(crtc->dev, &conn_iter);
+		drm_for_each_connector_iter(connector, &conn_iter) {
+			if (!connector->state || connector->state->crtc != crtc)
+				continue;
+
+			if (connector->connector_type == DRM_MODE_CONNECTOR_WRITEBACK)
+				continue;
+
+			aconn = to_amdgpu_dm_connector(connector);
+			break;
+		}
+		drm_connector_list_iter_end(&conn_iter);
+
+		if (!aconn) {
+			DRM_DEBUG_DRIVER("No amd connector matching CRTC-%d\n", crtc->index);
+			ret = -EINVAL;
+			goto cleanup;
+		}
+
+		aux = (aconn->mst_output_port) ? &aconn->mst_output_port->aux : &aconn->dm_dp_aux.aux;
+
+		if (!aux) {
+			DRM_DEBUG_DRIVER("No dp aux for amd connector\n");
+			ret = -EINVAL;
+			goto cleanup;
+		}
+
+		if ((aconn->base.connector_type != DRM_MODE_CONNECTOR_DisplayPort) &&
+				(aconn->base.connector_type != DRM_MODE_CONNECTOR_eDP)) {
+			DRM_DEBUG_DRIVER("No DP connector available for CRC source\n");
+			ret = -EINVAL;
+			goto cleanup;
+		}
+
+	}
+
+	/*
+	 * Reading the CRC requires the vblank interrupt handler to be
+	 * enabled. Keep a reference until CRC capture stops.
+	 */
+	enabled = amdgpu_dm_is_valid_crc_source(cur_crc_src);
+	if (!enabled && enable) {
+		ret = drm_crtc_vblank_get(crtc);
+		if (ret)
+			goto cleanup;
+	}
+
+#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY)
+	/* Reset secure_display when we change crc source from debugfs */
+	amdgpu_dm_set_crc_window_default(crtc, crtc_state->stream);
+#endif
+
+	if (amdgpu_dm_crtc_configure_crc_source(crtc, crtc_state, source)) {
+		ret = -EINVAL;
+		goto cleanup;
+	}
+
+	if (!enabled && enable) {
+		if (dm_is_crc_source_dprx(source)) {
+			if (drm_dp_start_crc(aux, crtc)) {
+				DRM_DEBUG_DRIVER("dp start crc failed\n");
+				ret = -EINVAL;
+				goto cleanup;
+			}
+		}
+	} else if (enabled && !enable) {
+		drm_crtc_vblank_put(crtc);
+		if (dm_is_crc_source_dprx(source)) {
+			if (drm_dp_stop_crc(aux)) {
+				DRM_DEBUG_DRIVER("dp stop crc failed\n");
+				ret = -EINVAL;
+				goto cleanup;
+			}
+		}
+	}
+
+	spin_lock_irq(&drm_dev->event_lock);
+	acrtc->dm_irq_params.crc_src = source;
+	spin_unlock_irq(&drm_dev->event_lock);
+
+	/* Reset crc_skipped on dm state */
+	crtc_state->crc_skip_count = 0;
+
+#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY)
+	/* Initialize phy id mapping table for secure display*/
+	if (dm->secure_display_ctx.op_mode == LEGACY_MODE &&
+		!dm->secure_display_ctx.phy_mapping_updated)
+		update_phy_id_mapping(adev);
+#endif
+
+cleanup:
+	if (commit)
+		drm_crtc_commit_put(commit);
+
+	drm_modeset_unlock(&crtc->mutex);
+
+	return ret;
+}
+
+/**
+ * amdgpu_dm_crtc_handle_crc_irq: Report to DRM the CRC on given CRTC.
+ * @crtc: DRM CRTC object.
+ *
+ * This function should be called at the end of a vblank, when the fb has been
+ * fully processed through the pipe.
+ */
+void amdgpu_dm_crtc_handle_crc_irq(struct drm_crtc *crtc)
+{
+	struct dm_crtc_state *crtc_state;
+	struct dc_stream_state *stream_state;
+	struct drm_device *drm_dev = NULL;
+	enum amdgpu_dm_pipe_crc_source cur_crc_src;
+	struct amdgpu_crtc *acrtc = NULL;
+	uint32_t crcs[3];
+	unsigned long flags;
+
+	if (crtc == NULL)
+		return;
+
+	crtc_state = to_dm_crtc_state(crtc->state);
+	stream_state = crtc_state->stream;
+	acrtc = to_amdgpu_crtc(crtc);
+	drm_dev = crtc->dev;
+
+	spin_lock_irqsave(&drm_dev->event_lock, flags);
+	cur_crc_src = acrtc->dm_irq_params.crc_src;
+	spin_unlock_irqrestore(&drm_dev->event_lock, flags);
+
+	/* Early return if CRC capture is not enabled. */
+	if (!amdgpu_dm_is_valid_crc_source(cur_crc_src))
+		return;
+
+	/*
+	 * Since flipping and crc enablement happen asynchronously, we - more
+	 * often than not - will be returning an 'uncooked' crc on first frame.
+	 * Probably because hw isn't ready yet. For added security, skip the
+	 * first two CRC values.
+	 */
+	if (crtc_state->crc_skip_count < 2) {
+		crtc_state->crc_skip_count += 1;
+		return;
+	}
+
+	if (dm_is_crc_source_crtc(cur_crc_src)) {
+		if (!dc_stream_get_crc(stream_state->ctx->dc, stream_state, 0,
+				       &crcs[0], &crcs[1], &crcs[2]))
+			return;
+
+		drm_crtc_add_crc_entry(crtc, true,
+				       drm_crtc_accurate_vblank_count(crtc), crcs);
+	}
+}
+
+#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY)
+void amdgpu_dm_crtc_handle_crc_window_irq(struct drm_crtc *crtc)
+{
+	struct drm_device *drm_dev = NULL;
+	enum amdgpu_dm_pipe_crc_source cur_crc_src;
+	struct amdgpu_crtc *acrtc = NULL;
+	struct amdgpu_device *adev = NULL;
+	struct secure_display_crtc_context *crtc_ctx = NULL;
+	bool reset_crc_frame_count[MAX_CRC_WINDOW_NUM] = {false};
+	uint32_t crc_r[MAX_CRC_WINDOW_NUM] = {0};
+	uint32_t crc_g[MAX_CRC_WINDOW_NUM] = {0};
+	uint32_t crc_b[MAX_CRC_WINDOW_NUM] = {0};
+	unsigned long flags1;
+	bool forward_roi_change = false;
+	bool notify_ta = false;
+	bool all_crc_ready = true;
+	struct dc_stream_state *stream_state;
+	int i;
+
+	if (crtc == NULL)
+		return;
+
+	acrtc = to_amdgpu_crtc(crtc);
+	adev = drm_to_adev(crtc->dev);
+	drm_dev = crtc->dev;
+	stream_state = to_dm_crtc_state(crtc->state)->stream;
+
+	spin_lock_irqsave(&drm_dev->event_lock, flags1);
+	cur_crc_src = acrtc->dm_irq_params.crc_src;
+
+	/* Early return if CRC capture is not enabled. */
+	if (!amdgpu_dm_is_valid_crc_source(cur_crc_src) ||
+	    !dm_is_crc_source_crtc(cur_crc_src)) {
+		spin_unlock_irqrestore(&drm_dev->event_lock, flags1);
+		return;
+	}
+
+	if (!acrtc->dm_irq_params.crc_window_activated) {
+		spin_unlock_irqrestore(&drm_dev->event_lock, flags1);
+		return;
+	}
+
+	crtc_ctx = &adev->dm.secure_display_ctx.crtc_ctx[acrtc->crtc_id];
+	if (WARN_ON(crtc_ctx->crtc != crtc)) {
+		/* We have set the crtc when creating secure_display_crtc_context,
+		 * don't expect it to be changed here.
+		 */
+		crtc_ctx->crtc = crtc;
+	}
+
+	for (i = 0; i < MAX_CRC_WINDOW_NUM; i++) {
+		struct crc_params crc_window = {
+			.windowa_x_start = acrtc->dm_irq_params.window_param[i].x_start,
+			.windowa_y_start = acrtc->dm_irq_params.window_param[i].y_start,
+			.windowa_x_end = acrtc->dm_irq_params.window_param[i].x_end,
+			.windowa_y_end = acrtc->dm_irq_params.window_param[i].y_end,
+			.windowb_x_start = acrtc->dm_irq_params.window_param[i].x_start,
+			.windowb_y_start = acrtc->dm_irq_params.window_param[i].y_start,
+			.windowb_x_end = acrtc->dm_irq_params.window_param[i].x_end,
+			.windowb_y_end = acrtc->dm_irq_params.window_param[i].y_end,
+		};
+
+		crtc_ctx->roi[i].enable = acrtc->dm_irq_params.window_param[i].enable;
+
+		if (!acrtc->dm_irq_params.window_param[i].enable) {
+			crtc_ctx->crc_info.crc[i].crc_ready = false;
+			continue;
+		}
+
+		if (acrtc->dm_irq_params.window_param[i].skip_frame_cnt) {
+			acrtc->dm_irq_params.window_param[i].skip_frame_cnt -= 1;
+			crtc_ctx->crc_info.crc[i].crc_ready = false;
+			continue;
+		}
+
+		if (acrtc->dm_irq_params.window_param[i].update_win) {
+			crtc_ctx->roi[i].rect.x = crc_window.windowa_x_start;
+			crtc_ctx->roi[i].rect.y = crc_window.windowa_y_start;
+			crtc_ctx->roi[i].rect.width = crc_window.windowa_x_end -
+						crc_window.windowa_x_start;
+			crtc_ctx->roi[i].rect.height = crc_window.windowa_y_end -
+						crc_window.windowa_y_start;
+
+			if (adev->dm.secure_display_ctx.op_mode == LEGACY_MODE)
+				/* forward task to dmub to update ROI */
+				forward_roi_change = true;
+			else if (adev->dm.secure_display_ctx.op_mode == DISPLAY_CRC_MODE)
+				/* update ROI via dm*/
+				dc_stream_configure_crc(stream_state->ctx->dc, stream_state,
+					&crc_window, true, true, i, false);
+
+			reset_crc_frame_count[i] = true;
+
+			acrtc->dm_irq_params.window_param[i].update_win = false;
+
+			/* Statically skip 1 frame, because we may need to wait below things
+			 * before sending ROI to dmub:
+			 * 1. We defer the work by using system workqueue.
+			 * 2. We may need to wait for dc_lock before accessing dmub.
+			 */
+			acrtc->dm_irq_params.window_param[i].skip_frame_cnt = 1;
+			crtc_ctx->crc_info.crc[i].crc_ready = false;
+		} else {
+			if (!dc_stream_get_crc(stream_state->ctx->dc, stream_state, i,
+						&crc_r[i], &crc_g[i], &crc_b[i]))
+				DRM_ERROR("Secure Display: fail to get crc from engine %d\n", i);
+
+			if (adev->dm.secure_display_ctx.op_mode == LEGACY_MODE)
+				/* forward task to psp to read ROI/CRC and output via I2C */
+				notify_ta = true;
+			else if (adev->dm.secure_display_ctx.op_mode == DISPLAY_CRC_MODE)
+				/* Avoid ROI window get changed, keep overwriting. */
+				dc_stream_configure_crc(stream_state->ctx->dc, stream_state,
+						&crc_window, true, true, i, false);
+
+			/* crc ready for psp to read out */
+			crtc_ctx->crc_info.crc[i].crc_ready = true;
+		}
+	}
+
+	spin_unlock_irqrestore(&drm_dev->event_lock, flags1);
+
+	if (forward_roi_change)
+		schedule_work(&crtc_ctx->forward_roi_work);
+
+	if (notify_ta)
+		schedule_work(&crtc_ctx->notify_ta_work);
+
+	spin_lock_irqsave(&crtc_ctx->crc_info.lock, flags1);
+	for (i = 0; i < MAX_CRC_WINDOW_NUM; i++) {
+		crtc_ctx->crc_info.crc[i].crc_R = crc_r[i];
+		crtc_ctx->crc_info.crc[i].crc_G = crc_g[i];
+		crtc_ctx->crc_info.crc[i].crc_B = crc_b[i];
+
+		if (!crtc_ctx->roi[i].enable) {
+			crtc_ctx->crc_info.crc[i].frame_count = 0;
+			continue;
+		}
+
+		if (!crtc_ctx->crc_info.crc[i].crc_ready)
+			all_crc_ready = false;
+
+		if (reset_crc_frame_count[i] || crtc_ctx->crc_info.crc[i].frame_count == UINT_MAX)
+			/* Reset the reference frame count after user update the ROI
+			 * or it reaches the maximum value.
+			 */
+			crtc_ctx->crc_info.crc[i].frame_count = 0;
+		else
+			crtc_ctx->crc_info.crc[i].frame_count += 1;
+	}
+	spin_unlock_irqrestore(&crtc_ctx->crc_info.lock, flags1);
+
+	if (all_crc_ready)
+		complete_all(&crtc_ctx->crc_info.completion);
+}
+
+void amdgpu_dm_crtc_secure_display_create_contexts(struct amdgpu_device *adev)
+{
+	struct secure_display_crtc_context *crtc_ctx = NULL;
+	int i;
+
+	crtc_ctx = kcalloc(adev->mode_info.num_crtc,
+				      sizeof(struct secure_display_crtc_context),
+				      GFP_KERNEL);
+
+	if (!crtc_ctx) {
+		adev->dm.secure_display_ctx.crtc_ctx = NULL;
+		return;
+	}
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		INIT_WORK(&crtc_ctx[i].forward_roi_work, amdgpu_dm_forward_crc_window);
+		INIT_WORK(&crtc_ctx[i].notify_ta_work, amdgpu_dm_crtc_notify_ta_to_read);
+		crtc_ctx[i].crtc = &adev->mode_info.crtcs[i]->base;
+		spin_lock_init(&crtc_ctx[i].crc_info.lock);
+	}
+
+	adev->dm.secure_display_ctx.crtc_ctx = crtc_ctx;
+
+	adev->dm.secure_display_ctx.op_mode = DISPLAY_CRC_MODE;
+}
+#endif
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.h
new file mode 100644
index 000000000000..95bdb8699d7f
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.h
@@ -0,0 +1,152 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef AMD_DAL_DEV_AMDGPU_DM_AMDGPU_DM_CRC_H_
+#define AMD_DAL_DEV_AMDGPU_DM_AMDGPU_DM_CRC_H_
+
+struct drm_crtc;
+struct dm_crtc_state;
+
+enum amdgpu_dm_pipe_crc_source {
+	AMDGPU_DM_PIPE_CRC_SOURCE_NONE = 0,
+	AMDGPU_DM_PIPE_CRC_SOURCE_CRTC,
+	AMDGPU_DM_PIPE_CRC_SOURCE_CRTC_DITHER,
+	AMDGPU_DM_PIPE_CRC_SOURCE_DPRX,
+	AMDGPU_DM_PIPE_CRC_SOURCE_DPRX_DITHER,
+	AMDGPU_DM_PIPE_CRC_SOURCE_MAX,
+	AMDGPU_DM_PIPE_CRC_SOURCE_INVALID = -1,
+};
+
+#ifdef CONFIG_DRM_AMD_SECURE_DISPLAY
+#define MAX_CRTC 6
+
+enum secure_display_mode {
+	/* via dmub + psp */
+	LEGACY_MODE = 0,
+	/* driver directly */
+	DISPLAY_CRC_MODE,
+	SECURE_DISPLAY_MODE_MAX,
+};
+
+struct phy_id_mapping {
+	bool assigned;
+	bool is_mst;
+	uint8_t enc_hw_inst;
+	u8 lct;
+	u8 port_num;
+	u8 rad[8];
+};
+
+struct crc_data {
+	uint32_t crc_R;
+	uint32_t crc_G;
+	uint32_t crc_B;
+	uint32_t frame_count;
+	bool crc_ready;
+};
+
+struct crc_info {
+	struct crc_data crc[MAX_CRC_WINDOW_NUM];
+	struct completion completion;
+	spinlock_t lock;
+};
+
+struct crc_window_param {
+	uint16_t x_start;
+	uint16_t y_start;
+	uint16_t x_end;
+	uint16_t y_end;
+	/* CRC window is activated or not*/
+	bool enable;
+	/* Update crc window during vertical blank or not */
+	bool update_win;
+	/* skip reading/writing for few frames */
+	int skip_frame_cnt;
+};
+
+struct secure_display_crtc_context {
+	/* work to notify PSP TA*/
+	struct work_struct notify_ta_work;
+
+	/* work to forward ROI to dmcu/dmub */
+	struct work_struct forward_roi_work;
+
+	struct drm_crtc *crtc;
+
+	/* Region of Interest (ROI) */
+	struct crc_window roi[MAX_CRC_WINDOW_NUM];
+
+	struct crc_info crc_info;
+};
+
+struct secure_display_context {
+
+	struct secure_display_crtc_context *crtc_ctx;
+    /* Whether dmub support multiple ROI setting */
+	bool support_mul_roi;
+	enum secure_display_mode op_mode;
+	bool phy_mapping_updated;
+	int phy_id_mapping_cnt;
+	struct phy_id_mapping phy_id_mapping[MAX_CRTC];
+};
+#endif
+
+static inline bool amdgpu_dm_is_valid_crc_source(enum amdgpu_dm_pipe_crc_source source)
+{
+	return (source > AMDGPU_DM_PIPE_CRC_SOURCE_NONE) &&
+	       (source < AMDGPU_DM_PIPE_CRC_SOURCE_MAX);
+}
+
+/* amdgpu_dm_crc.c */
+#ifdef CONFIG_DEBUG_FS
+int amdgpu_dm_crtc_configure_crc_source(struct drm_crtc *crtc,
+					struct dm_crtc_state *dm_crtc_state,
+					enum amdgpu_dm_pipe_crc_source source);
+int amdgpu_dm_crtc_set_crc_source(struct drm_crtc *crtc, const char *src_name);
+int amdgpu_dm_crtc_verify_crc_source(struct drm_crtc *crtc,
+				     const char *src_name,
+				     size_t *values_cnt);
+const char *const *amdgpu_dm_crtc_get_crc_sources(struct drm_crtc *crtc,
+						  size_t *count);
+void amdgpu_dm_crtc_handle_crc_irq(struct drm_crtc *crtc);
+#else
+#define amdgpu_dm_crtc_set_crc_source NULL
+#define amdgpu_dm_crtc_verify_crc_source NULL
+#define amdgpu_dm_crtc_get_crc_sources NULL
+#define amdgpu_dm_crtc_handle_crc_irq(x)
+#endif
+
+#ifdef CONFIG_DRM_AMD_SECURE_DISPLAY
+bool amdgpu_dm_crc_window_is_activated(struct drm_crtc *crtc);
+void amdgpu_dm_crtc_handle_crc_window_irq(struct drm_crtc *crtc);
+void amdgpu_dm_crtc_secure_display_create_contexts(struct amdgpu_device *adev);
+#else
+#define amdgpu_dm_crc_window_is_activated(x)
+#define amdgpu_dm_crtc_handle_crc_window_irq(x)
+#define amdgpu_dm_crtc_secure_display_create_contexts(x)
+#endif
+
+#endif /* AMD_DAL_DEV_AMDGPU_DM_AMDGPU_DM_CRC_H_ */
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c
new file mode 100644
index 000000000000..1ec9d03ad747
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c
@@ -0,0 +1,791 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+#include <drm/drm_vblank.h>
+#include <drm/drm_atomic_helper.h>
+
+#include "dc.h"
+#include "amdgpu.h"
+#include "amdgpu_dm_psr.h"
+#include "amdgpu_dm_replay.h"
+#include "amdgpu_dm_crtc.h"
+#include "amdgpu_dm_plane.h"
+#include "amdgpu_dm_trace.h"
+#include "amdgpu_dm_debugfs.h"
+
+#define HPD_DETECTION_PERIOD_uS 2000000
+#define HPD_DETECTION_TIME_uS 100000
+
+void amdgpu_dm_crtc_handle_vblank(struct amdgpu_crtc *acrtc)
+{
+	struct drm_crtc *crtc = &acrtc->base;
+	struct drm_device *dev = crtc->dev;
+	unsigned long flags;
+
+	drm_crtc_handle_vblank(crtc);
+
+	spin_lock_irqsave(&dev->event_lock, flags);
+
+	/* Send completion event for cursor-only commits */
+	if (acrtc->event && acrtc->pflip_status != AMDGPU_FLIP_SUBMITTED) {
+		drm_crtc_send_vblank_event(crtc, acrtc->event);
+		drm_crtc_vblank_put(crtc);
+		acrtc->event = NULL;
+	}
+
+	spin_unlock_irqrestore(&dev->event_lock, flags);
+}
+
+bool amdgpu_dm_crtc_modeset_required(struct drm_crtc_state *crtc_state,
+			     struct dc_stream_state *new_stream,
+			     struct dc_stream_state *old_stream)
+{
+	return crtc_state->active && drm_atomic_crtc_needs_modeset(crtc_state);
+}
+
+bool amdgpu_dm_crtc_vrr_active_irq(struct amdgpu_crtc *acrtc)
+
+{
+	return acrtc->dm_irq_params.freesync_config.state ==
+		       VRR_STATE_ACTIVE_VARIABLE ||
+	       acrtc->dm_irq_params.freesync_config.state ==
+		       VRR_STATE_ACTIVE_FIXED;
+}
+
+int amdgpu_dm_crtc_set_vupdate_irq(struct drm_crtc *crtc, bool enable)
+{
+	enum dc_irq_source irq_source;
+	struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
+	int rc;
+
+	if (acrtc->otg_inst == -1)
+		return 0;
+
+	irq_source = IRQ_TYPE_VUPDATE + acrtc->otg_inst;
+
+	rc = dc_interrupt_set(adev->dm.dc, irq_source, enable) ? 0 : -EBUSY;
+
+	DRM_DEBUG_VBL("crtc %d - vupdate irq %sabling: r=%d\n",
+		      acrtc->crtc_id, enable ? "en" : "dis", rc);
+	return rc;
+}
+
+bool amdgpu_dm_crtc_vrr_active(const struct dm_crtc_state *dm_state)
+{
+	return dm_state->freesync_config.state == VRR_STATE_ACTIVE_VARIABLE ||
+	       dm_state->freesync_config.state == VRR_STATE_ACTIVE_FIXED;
+}
+
+/**
+ * amdgpu_dm_crtc_set_panel_sr_feature() - Manage panel self-refresh features.
+ *
+ * @vblank_work:    is a pointer to a struct vblank_control_work object.
+ * @vblank_enabled: indicates whether the DRM vblank counter is currently
+ *                  enabled (true) or disabled (false).
+ * @allow_sr_entry: represents whether entry into the self-refresh mode is
+ *                  allowed (true) or not allowed (false).
+ *
+ * The DRM vblank counter enable/disable action is used as the trigger to enable
+ * or disable various panel self-refresh features:
+ *
+ * Panel Replay and PSR SU
+ * - Enable when:
+ *      - VRR is disabled
+ *      - vblank counter is disabled
+ *      - entry is allowed: usermode demonstrates an adequate number of fast
+ *        commits)
+ *     - CRC capture window isn't active
+ * - Keep enabled even when vblank counter gets enabled
+ *
+ * PSR1
+ * - Enable condition same as above
+ * - Disable when vblank counter is enabled
+ */
+static void amdgpu_dm_crtc_set_panel_sr_feature(
+	struct vblank_control_work *vblank_work,
+	bool vblank_enabled, bool allow_sr_entry)
+{
+	struct dc_link *link = vblank_work->stream->link;
+	bool is_sr_active = (link->replay_settings.replay_allow_active ||
+				 link->psr_settings.psr_allow_active);
+	bool is_crc_window_active = false;
+	bool vrr_active = amdgpu_dm_crtc_vrr_active_irq(vblank_work->acrtc);
+
+#ifdef CONFIG_DRM_AMD_SECURE_DISPLAY
+	is_crc_window_active =
+		amdgpu_dm_crc_window_is_activated(&vblank_work->acrtc->base);
+#endif
+
+	if (link->replay_settings.replay_feature_enabled && !vrr_active &&
+		allow_sr_entry && !is_sr_active && !is_crc_window_active) {
+		amdgpu_dm_replay_enable(vblank_work->stream, true);
+	} else if (vblank_enabled) {
+		if (link->psr_settings.psr_version < DC_PSR_VERSION_SU_1 && is_sr_active)
+			amdgpu_dm_psr_disable(vblank_work->stream, false);
+	} else if (link->psr_settings.psr_feature_enabled && !vrr_active &&
+		allow_sr_entry && !is_sr_active && !is_crc_window_active) {
+
+		struct amdgpu_dm_connector *aconn =
+			(struct amdgpu_dm_connector *) vblank_work->stream->dm_stream_context;
+
+		if (!aconn->disallow_edp_enter_psr) {
+			struct amdgpu_display_manager *dm = vblank_work->dm;
+
+			amdgpu_dm_psr_enable(vblank_work->stream);
+			if (dm->idle_workqueue &&
+			    (dm->dc->config.disable_ips == DMUB_IPS_ENABLE) &&
+			    dm->dc->idle_optimizations_allowed &&
+			    dm->idle_workqueue->enable &&
+			    !dm->idle_workqueue->running)
+				schedule_work(&dm->idle_workqueue->work);
+		}
+	}
+}
+
+bool amdgpu_dm_is_headless(struct amdgpu_device *adev)
+{
+	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
+	struct drm_device *dev;
+	bool is_headless = true;
+
+	if (adev == NULL)
+		return true;
+
+	dev = adev->dm.ddev;
+
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_WRITEBACK)
+			continue;
+
+		if (connector->status == connector_status_connected) {
+			is_headless = false;
+			break;
+		}
+	}
+	drm_connector_list_iter_end(&iter);
+	return is_headless;
+}
+
+static void amdgpu_dm_idle_worker(struct work_struct *work)
+{
+	struct idle_workqueue *idle_work;
+
+	idle_work = container_of(work, struct idle_workqueue, work);
+	idle_work->dm->idle_workqueue->running = true;
+
+	while (idle_work->enable) {
+		fsleep(HPD_DETECTION_PERIOD_uS);
+		mutex_lock(&idle_work->dm->dc_lock);
+		if (!idle_work->dm->dc->idle_optimizations_allowed) {
+			mutex_unlock(&idle_work->dm->dc_lock);
+			break;
+		}
+		dc_allow_idle_optimizations(idle_work->dm->dc, false);
+
+		mutex_unlock(&idle_work->dm->dc_lock);
+		fsleep(HPD_DETECTION_TIME_uS);
+		mutex_lock(&idle_work->dm->dc_lock);
+
+		if (!amdgpu_dm_is_headless(idle_work->dm->adev) &&
+		    !amdgpu_dm_psr_is_active_allowed(idle_work->dm)) {
+			mutex_unlock(&idle_work->dm->dc_lock);
+			break;
+		}
+
+		if (idle_work->enable) {
+			dc_post_update_surfaces_to_stream(idle_work->dm->dc);
+			dc_allow_idle_optimizations(idle_work->dm->dc, true);
+		}
+		mutex_unlock(&idle_work->dm->dc_lock);
+	}
+	idle_work->dm->idle_workqueue->running = false;
+}
+
+struct idle_workqueue *idle_create_workqueue(struct amdgpu_device *adev)
+{
+	struct idle_workqueue *idle_work;
+
+	idle_work = kzalloc(sizeof(*idle_work), GFP_KERNEL);
+	if (ZERO_OR_NULL_PTR(idle_work))
+		return NULL;
+
+	idle_work->dm = &adev->dm;
+	idle_work->enable = false;
+	idle_work->running = false;
+	INIT_WORK(&idle_work->work, amdgpu_dm_idle_worker);
+
+	return idle_work;
+}
+
+static void amdgpu_dm_crtc_vblank_control_worker(struct work_struct *work)
+{
+	struct vblank_control_work *vblank_work =
+		container_of(work, struct vblank_control_work, work);
+	struct amdgpu_display_manager *dm = vblank_work->dm;
+
+	mutex_lock(&dm->dc_lock);
+
+	if (vblank_work->enable)
+		dm->active_vblank_irq_count++;
+	else if (dm->active_vblank_irq_count)
+		dm->active_vblank_irq_count--;
+
+	if (dm->active_vblank_irq_count > 0)
+		dc_allow_idle_optimizations(dm->dc, false);
+
+	/*
+	 * Control PSR based on vblank requirements from OS
+	 *
+	 * If panel supports PSR SU, there's no need to disable PSR when OS is
+	 * submitting fast atomic commits (we infer this by whether the OS
+	 * requests vblank events). Fast atomic commits will simply trigger a
+	 * full-frame-update (FFU); a specific case of selective-update (SU)
+	 * where the SU region is the full hactive*vactive region. See
+	 * fill_dc_dirty_rects().
+	 */
+	if (vblank_work->stream && vblank_work->stream->link && vblank_work->acrtc) {
+		amdgpu_dm_crtc_set_panel_sr_feature(
+			vblank_work, vblank_work->enable,
+			vblank_work->acrtc->dm_irq_params.allow_sr_entry);
+	}
+
+	if (dm->active_vblank_irq_count == 0) {
+		dc_post_update_surfaces_to_stream(dm->dc);
+		dc_allow_idle_optimizations(dm->dc, true);
+	}
+
+	mutex_unlock(&dm->dc_lock);
+
+	dc_stream_release(vblank_work->stream);
+
+	kfree(vblank_work);
+}
+
+static inline int amdgpu_dm_crtc_set_vblank(struct drm_crtc *crtc, bool enable)
+{
+	struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
+	struct dm_crtc_state *acrtc_state = to_dm_crtc_state(crtc->state);
+	struct amdgpu_display_manager *dm = &adev->dm;
+	struct vblank_control_work *work;
+	int irq_type;
+	int rc = 0;
+
+	if (acrtc->otg_inst == -1)
+		goto skip;
+
+	irq_type = amdgpu_display_crtc_idx_to_irq_type(adev, acrtc->crtc_id);
+
+	if (enable) {
+		struct dc *dc = adev->dm.dc;
+		struct drm_vblank_crtc *vblank = drm_crtc_vblank_crtc(crtc);
+		struct psr_settings *psr = &acrtc_state->stream->link->psr_settings;
+		struct replay_settings *pr = &acrtc_state->stream->link->replay_settings;
+		bool sr_supported = (psr->psr_version != DC_PSR_VERSION_UNSUPPORTED) ||
+								pr->config.replay_supported;
+
+		/*
+		 * IPS & self-refresh feature can cause vblank counter resets between
+		 * vblank disable and enable.
+		 * It may cause system stuck due to waiting for the vblank counter.
+		 * Call this function to estimate missed vblanks by using timestamps and
+		 * update the vblank counter in DRM.
+		 */
+		if (dc->caps.ips_support &&
+			dc->config.disable_ips != DMUB_IPS_DISABLE_ALL &&
+			sr_supported && vblank->config.disable_immediate)
+			drm_crtc_vblank_restore(crtc);
+	}
+
+	if (dc_supports_vrr(dm->dc->ctx->dce_version)) {
+		if (enable) {
+			/* vblank irq on -> Only need vupdate irq in vrr mode */
+			if (amdgpu_dm_crtc_vrr_active(acrtc_state))
+				rc = amdgpu_dm_crtc_set_vupdate_irq(crtc, true);
+		} else {
+			/* vblank irq off -> vupdate irq off */
+			rc = amdgpu_dm_crtc_set_vupdate_irq(crtc, false);
+		}
+	}
+
+	if (rc)
+		return rc;
+
+	/* crtc vblank or vstartup interrupt */
+	if (enable) {
+		rc = amdgpu_irq_get(adev, &adev->crtc_irq, irq_type);
+		drm_dbg_vbl(crtc->dev, "Get crtc_irq ret=%d\n", rc);
+	} else {
+		rc = amdgpu_irq_put(adev, &adev->crtc_irq, irq_type);
+		drm_dbg_vbl(crtc->dev, "Put crtc_irq ret=%d\n", rc);
+	}
+
+	if (rc)
+		return rc;
+
+	/*
+	 * hubp surface flip interrupt
+	 *
+	 * We have no guarantee that the frontend index maps to the same
+	 * backend index - some even map to more than one.
+	 *
+	 * TODO: Use a different interrupt or check DC itself for the mapping.
+	 */
+	if (enable) {
+		rc = amdgpu_irq_get(adev, &adev->pageflip_irq, irq_type);
+		drm_dbg_vbl(crtc->dev, "Get pageflip_irq ret=%d\n", rc);
+	} else {
+		rc = amdgpu_irq_put(adev, &adev->pageflip_irq, irq_type);
+		drm_dbg_vbl(crtc->dev, "Put pageflip_irq ret=%d\n", rc);
+	}
+
+	if (rc)
+		return rc;
+
+#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY)
+	/* crtc vline0 interrupt, only available on DCN+ */
+	if (amdgpu_ip_version(adev, DCE_HWIP, 0) != 0) {
+		if (enable) {
+			rc = amdgpu_irq_get(adev, &adev->vline0_irq, irq_type);
+			drm_dbg_vbl(crtc->dev, "Get vline0_irq ret=%d\n", rc);
+		} else {
+			rc = amdgpu_irq_put(adev, &adev->vline0_irq, irq_type);
+			drm_dbg_vbl(crtc->dev, "Put vline0_irq ret=%d\n", rc);
+		}
+
+		if (rc)
+			return rc;
+	}
+#endif
+skip:
+	if (amdgpu_in_reset(adev))
+		return 0;
+
+	if (dm->vblank_control_workqueue) {
+		work = kzalloc(sizeof(*work), GFP_ATOMIC);
+		if (!work)
+			return -ENOMEM;
+
+		INIT_WORK(&work->work, amdgpu_dm_crtc_vblank_control_worker);
+		work->dm = dm;
+		work->acrtc = acrtc;
+		work->enable = enable;
+
+		if (acrtc_state->stream) {
+			dc_stream_retain(acrtc_state->stream);
+			work->stream = acrtc_state->stream;
+		}
+
+		queue_work(dm->vblank_control_workqueue, &work->work);
+	}
+
+	return 0;
+}
+
+int amdgpu_dm_crtc_enable_vblank(struct drm_crtc *crtc)
+{
+	return amdgpu_dm_crtc_set_vblank(crtc, true);
+}
+
+void amdgpu_dm_crtc_disable_vblank(struct drm_crtc *crtc)
+{
+	amdgpu_dm_crtc_set_vblank(crtc, false);
+}
+
+static void amdgpu_dm_crtc_destroy_state(struct drm_crtc *crtc,
+				  struct drm_crtc_state *state)
+{
+	struct dm_crtc_state *cur = to_dm_crtc_state(state);
+
+	/* TODO Destroy dc_stream objects are stream object is flattened */
+	if (cur->stream)
+		dc_stream_release(cur->stream);
+
+
+	__drm_atomic_helper_crtc_destroy_state(state);
+
+
+	kfree(state);
+}
+
+static struct drm_crtc_state *amdgpu_dm_crtc_duplicate_state(struct drm_crtc *crtc)
+{
+	struct dm_crtc_state *state, *cur;
+
+	cur = to_dm_crtc_state(crtc->state);
+
+	if (WARN_ON(!crtc->state))
+		return NULL;
+
+	state = kzalloc(sizeof(*state), GFP_KERNEL);
+	if (!state)
+		return NULL;
+
+	__drm_atomic_helper_crtc_duplicate_state(crtc, &state->base);
+
+	if (cur->stream) {
+		state->stream = cur->stream;
+		dc_stream_retain(state->stream);
+	}
+
+	state->active_planes = cur->active_planes;
+	state->vrr_infopacket = cur->vrr_infopacket;
+	state->abm_level = cur->abm_level;
+	state->vrr_supported = cur->vrr_supported;
+	state->freesync_config = cur->freesync_config;
+	state->cm_has_degamma = cur->cm_has_degamma;
+	state->cm_is_degamma_srgb = cur->cm_is_degamma_srgb;
+	state->regamma_tf = cur->regamma_tf;
+	state->crc_skip_count = cur->crc_skip_count;
+	state->mpo_requested = cur->mpo_requested;
+	state->cursor_mode = cur->cursor_mode;
+	/* TODO Duplicate dc_stream after objects are stream object is flattened */
+
+	return &state->base;
+}
+
+static void amdgpu_dm_crtc_destroy(struct drm_crtc *crtc)
+{
+	drm_crtc_cleanup(crtc);
+	kfree(crtc);
+}
+
+static void amdgpu_dm_crtc_reset_state(struct drm_crtc *crtc)
+{
+	struct dm_crtc_state *state;
+
+	if (crtc->state)
+		amdgpu_dm_crtc_destroy_state(crtc, crtc->state);
+
+	state = kzalloc(sizeof(*state), GFP_KERNEL);
+	if (WARN_ON(!state))
+		return;
+
+	__drm_atomic_helper_crtc_reset(crtc, &state->base);
+}
+
+#ifdef CONFIG_DEBUG_FS
+static int amdgpu_dm_crtc_late_register(struct drm_crtc *crtc)
+{
+	crtc_debugfs_init(crtc);
+
+	return 0;
+}
+#endif
+
+#ifdef AMD_PRIVATE_COLOR
+/**
+ * dm_crtc_additional_color_mgmt - enable additional color properties
+ * @crtc: DRM CRTC
+ *
+ * This function lets the driver enable post-blending CRTC regamma transfer
+ * function property in addition to DRM CRTC gamma LUT. Default value means
+ * linear transfer function, which is the default CRTC gamma LUT behaviour
+ * without this property.
+ */
+static void
+dm_crtc_additional_color_mgmt(struct drm_crtc *crtc)
+{
+	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
+
+	if (adev->dm.dc->caps.color.mpc.ogam_ram)
+		drm_object_attach_property(&crtc->base,
+					   adev->mode_info.regamma_tf_property,
+					   AMDGPU_TRANSFER_FUNCTION_DEFAULT);
+}
+
+static int
+amdgpu_dm_atomic_crtc_set_property(struct drm_crtc *crtc,
+				   struct drm_crtc_state *state,
+				   struct drm_property *property,
+				   uint64_t val)
+{
+	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
+	struct dm_crtc_state *acrtc_state = to_dm_crtc_state(state);
+
+	if (property == adev->mode_info.regamma_tf_property) {
+		if (acrtc_state->regamma_tf != val) {
+			acrtc_state->regamma_tf = val;
+			acrtc_state->base.color_mgmt_changed |= 1;
+		}
+	} else {
+		drm_dbg_atomic(crtc->dev,
+			       "[CRTC:%d:%s] unknown property [PROP:%d:%s]]\n",
+			       crtc->base.id, crtc->name,
+			       property->base.id, property->name);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int
+amdgpu_dm_atomic_crtc_get_property(struct drm_crtc *crtc,
+				   const struct drm_crtc_state *state,
+				   struct drm_property *property,
+				   uint64_t *val)
+{
+	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
+	struct dm_crtc_state *acrtc_state = to_dm_crtc_state(state);
+
+	if (property == adev->mode_info.regamma_tf_property)
+		*val = acrtc_state->regamma_tf;
+	else
+		return -EINVAL;
+
+	return 0;
+}
+#endif
+
+/* Implemented only the options currently available for the driver */
+static const struct drm_crtc_funcs amdgpu_dm_crtc_funcs = {
+	.reset = amdgpu_dm_crtc_reset_state,
+	.destroy = amdgpu_dm_crtc_destroy,
+	.set_config = drm_atomic_helper_set_config,
+	.page_flip = drm_atomic_helper_page_flip,
+	.atomic_duplicate_state = amdgpu_dm_crtc_duplicate_state,
+	.atomic_destroy_state = amdgpu_dm_crtc_destroy_state,
+	.set_crc_source = amdgpu_dm_crtc_set_crc_source,
+	.verify_crc_source = amdgpu_dm_crtc_verify_crc_source,
+	.get_crc_sources = amdgpu_dm_crtc_get_crc_sources,
+	.get_vblank_counter = amdgpu_get_vblank_counter_kms,
+	.enable_vblank = amdgpu_dm_crtc_enable_vblank,
+	.disable_vblank = amdgpu_dm_crtc_disable_vblank,
+	.get_vblank_timestamp = drm_crtc_vblank_helper_get_vblank_timestamp,
+#if defined(CONFIG_DEBUG_FS)
+	.late_register = amdgpu_dm_crtc_late_register,
+#endif
+#ifdef AMD_PRIVATE_COLOR
+	.atomic_set_property = amdgpu_dm_atomic_crtc_set_property,
+	.atomic_get_property = amdgpu_dm_atomic_crtc_get_property,
+#endif
+};
+
+static void amdgpu_dm_crtc_helper_disable(struct drm_crtc *crtc)
+{
+}
+
+static int amdgpu_dm_crtc_count_crtc_active_planes(struct drm_crtc_state *new_crtc_state)
+{
+	struct drm_atomic_state *state = new_crtc_state->state;
+	struct drm_plane *plane;
+	int num_active = 0;
+
+	drm_for_each_plane_mask(plane, state->dev, new_crtc_state->plane_mask) {
+		struct drm_plane_state *new_plane_state;
+
+		/* Cursor planes are "fake". */
+		if (plane->type == DRM_PLANE_TYPE_CURSOR)
+			continue;
+
+		new_plane_state = drm_atomic_get_new_plane_state(state, plane);
+
+		if (!new_plane_state) {
+			/*
+			 * The plane is enable on the CRTC and hasn't changed
+			 * state. This means that it previously passed
+			 * validation and is therefore enabled.
+			 */
+			num_active += 1;
+			continue;
+		}
+
+		/* We need a framebuffer to be considered enabled. */
+		num_active += (new_plane_state->fb != NULL);
+	}
+
+	return num_active;
+}
+
+static void amdgpu_dm_crtc_update_crtc_active_planes(struct drm_crtc *crtc,
+						     struct drm_crtc_state *new_crtc_state)
+{
+	struct dm_crtc_state *dm_new_crtc_state =
+		to_dm_crtc_state(new_crtc_state);
+
+	dm_new_crtc_state->active_planes = 0;
+
+	if (!dm_new_crtc_state->stream)
+		return;
+
+	dm_new_crtc_state->active_planes =
+		amdgpu_dm_crtc_count_crtc_active_planes(new_crtc_state);
+}
+
+static bool amdgpu_dm_crtc_helper_mode_fixup(struct drm_crtc *crtc,
+				      const struct drm_display_mode *mode,
+				      struct drm_display_mode *adjusted_mode)
+{
+	return true;
+}
+
+static int amdgpu_dm_crtc_helper_atomic_check(struct drm_crtc *crtc,
+					      struct drm_atomic_state *state)
+{
+	struct drm_crtc_state *crtc_state = drm_atomic_get_new_crtc_state(state,
+										crtc);
+	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
+	struct dc *dc = adev->dm.dc;
+	struct dm_crtc_state *dm_crtc_state = to_dm_crtc_state(crtc_state);
+	int ret = -EINVAL;
+
+	trace_amdgpu_dm_crtc_atomic_check(crtc_state);
+
+	amdgpu_dm_crtc_update_crtc_active_planes(crtc, crtc_state);
+
+	if (WARN_ON(unlikely(!dm_crtc_state->stream &&
+			amdgpu_dm_crtc_modeset_required(crtc_state, NULL, dm_crtc_state->stream)))) {
+		return ret;
+	}
+
+	/*
+	 * We require the primary plane to be enabled whenever the CRTC is, otherwise
+	 * drm_mode_cursor_universal may end up trying to enable the cursor plane while all other
+	 * planes are disabled, which is not supported by the hardware. And there is legacy
+	 * userspace which stops using the HW cursor altogether in response to the resulting EINVAL.
+	 */
+	if (crtc_state->enable &&
+		!(crtc_state->plane_mask & drm_plane_mask(crtc->primary))) {
+		DRM_DEBUG_ATOMIC("Can't enable a CRTC without enabling the primary plane\n");
+		return -EINVAL;
+	}
+
+	/*
+	 * Only allow async flips for fast updates that don't change the FB
+	 * pitch, the DCC state, rotation, etc.
+	 */
+	if (crtc_state->async_flip &&
+	    dm_crtc_state->update_type != UPDATE_TYPE_FAST) {
+		drm_dbg_atomic(crtc->dev,
+			       "[CRTC:%d:%s] async flips are only supported for fast updates\n",
+			       crtc->base.id, crtc->name);
+		return -EINVAL;
+	}
+
+	if (!state->legacy_cursor_update && amdgpu_dm_crtc_vrr_active(dm_crtc_state)) {
+		struct drm_plane_state *primary_state;
+
+		/* Pull in primary plane for correct VRR handling */
+		primary_state = drm_atomic_get_plane_state(state, crtc->primary);
+		if (IS_ERR(primary_state))
+			return PTR_ERR(primary_state);
+	}
+
+	/* In some use cases, like reset, no stream is attached */
+	if (!dm_crtc_state->stream)
+		return 0;
+
+	if (dc_validate_stream(dc, dm_crtc_state->stream) == DC_OK)
+		return 0;
+
+	DRM_DEBUG_ATOMIC("Failed DC stream validation\n");
+	return ret;
+}
+
+static const struct drm_crtc_helper_funcs amdgpu_dm_crtc_helper_funcs = {
+	.disable = amdgpu_dm_crtc_helper_disable,
+	.atomic_check = amdgpu_dm_crtc_helper_atomic_check,
+	.mode_fixup = amdgpu_dm_crtc_helper_mode_fixup,
+	.get_scanout_position = amdgpu_crtc_get_scanout_position,
+};
+
+int amdgpu_dm_crtc_init(struct amdgpu_display_manager *dm,
+			       struct drm_plane *plane,
+			       uint32_t crtc_index)
+{
+	struct amdgpu_crtc *acrtc = NULL;
+	struct drm_plane *cursor_plane;
+	bool is_dcn;
+	int res = -ENOMEM;
+
+	cursor_plane = kzalloc(sizeof(*cursor_plane), GFP_KERNEL);
+	if (!cursor_plane)
+		goto fail;
+
+	cursor_plane->type = DRM_PLANE_TYPE_CURSOR;
+	res = amdgpu_dm_plane_init(dm, cursor_plane, 0, NULL);
+
+	acrtc = kzalloc(sizeof(struct amdgpu_crtc), GFP_KERNEL);
+	if (!acrtc)
+		goto fail;
+
+	res = drm_crtc_init_with_planes(
+			dm->ddev,
+			&acrtc->base,
+			plane,
+			cursor_plane,
+			&amdgpu_dm_crtc_funcs, NULL);
+
+	if (res)
+		goto fail;
+
+	drm_crtc_helper_add(&acrtc->base, &amdgpu_dm_crtc_helper_funcs);
+
+	/* Create (reset) the plane state */
+	if (acrtc->base.funcs->reset)
+		acrtc->base.funcs->reset(&acrtc->base);
+
+	acrtc->max_cursor_width = dm->adev->dm.dc->caps.max_cursor_size;
+	acrtc->max_cursor_height = dm->adev->dm.dc->caps.max_cursor_size;
+
+	acrtc->crtc_id = crtc_index;
+	acrtc->base.enabled = false;
+	acrtc->otg_inst = -1;
+
+	dm->adev->mode_info.crtcs[crtc_index] = acrtc;
+
+	/* Don't enable DRM CRTC degamma property for DCE since it doesn't
+	 * support programmable degamma anywhere.
+	 */
+	is_dcn = dm->adev->dm.dc->caps.color.dpp.dcn_arch;
+	/* Dont't enable DRM CRTC degamma property for DCN401 since the
+	 * pre-blending degamma LUT doesn't apply to cursor, and therefore
+	 * can't work similar to a post-blending degamma LUT as in other hw
+	 * versions.
+	 * TODO: revisit it once KMS plane color API is merged.
+	 */
+	drm_crtc_enable_color_mgmt(&acrtc->base,
+				   (is_dcn &&
+				    dm->adev->dm.dc->ctx->dce_version != DCN_VERSION_4_01) ?
+				     MAX_COLOR_LUT_ENTRIES : 0,
+				   true, MAX_COLOR_LUT_ENTRIES);
+
+	drm_mode_crtc_set_gamma_size(&acrtc->base, MAX_COLOR_LEGACY_LUT_ENTRIES);
+
+#ifdef AMD_PRIVATE_COLOR
+	dm_crtc_additional_color_mgmt(&acrtc->base);
+#endif
+	return 0;
+
+fail:
+	kfree(acrtc);
+	kfree(cursor_plane);
+	return res;
+}
+
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.h
new file mode 100644
index 000000000000..c1212947a77b
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.h
@@ -0,0 +1,51 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __AMDGPU_DM_CRTC_H__
+#define __AMDGPU_DM_CRTC_H__
+
+void amdgpu_dm_crtc_handle_vblank(struct amdgpu_crtc *acrtc);
+
+bool amdgpu_dm_crtc_modeset_required(struct drm_crtc_state *crtc_state,
+		      struct dc_stream_state *new_stream,
+		      struct dc_stream_state *old_stream);
+
+int amdgpu_dm_crtc_set_vupdate_irq(struct drm_crtc *crtc, bool enable);
+
+bool amdgpu_dm_crtc_vrr_active_irq(struct amdgpu_crtc *acrtc);
+
+bool amdgpu_dm_crtc_vrr_active(const struct dm_crtc_state *dm_state);
+
+int amdgpu_dm_crtc_enable_vblank(struct drm_crtc *crtc);
+
+void amdgpu_dm_crtc_disable_vblank(struct drm_crtc *crtc);
+
+int amdgpu_dm_crtc_init(struct amdgpu_display_manager *dm,
+			struct drm_plane *plane,
+			uint32_t link_index);
+
+#endif
+
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c
new file mode 100644
index 000000000000..f263e1a4537e
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c
@@ -0,0 +1,4275 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include <linux/string_helpers.h>
+#include <linux/uaccess.h>
+#include <media/cec-notifier.h>
+
+#include "dc.h"
+#include "amdgpu.h"
+#include "amdgpu_dm.h"
+#include "amdgpu_dm_debugfs.h"
+#include "amdgpu_dm_replay.h"
+#include "dm_helpers.h"
+#include "dmub/dmub_srv.h"
+#include "resource.h"
+#include "dsc.h"
+#include "link_hwss.h"
+#include "dc/dc_dmub_srv.h"
+#include "link/protocols/link_dp_capability.h"
+#include "inc/hw/dchubbub.h"
+
+#ifdef CONFIG_DRM_AMD_SECURE_DISPLAY
+#include "amdgpu_dm_psr.h"
+#endif
+
+struct dmub_debugfs_trace_header {
+	uint32_t entry_count;
+	uint32_t reserved[3];
+};
+
+struct dmub_debugfs_trace_entry {
+	uint32_t trace_code;
+	uint32_t tick_count;
+	uint32_t param0;
+	uint32_t param1;
+};
+
+static const char *const mst_progress_status[] = {
+	"probe",
+	"remote_edid",
+	"allocate_new_payload",
+	"clear_allocated_payload",
+};
+
+/* parse_write_buffer_into_params - Helper function to parse debugfs write buffer into an array
+ *
+ * Function takes in attributes passed to debugfs write entry
+ * and writes into param array.
+ * The user passes max_param_num to identify maximum number of
+ * parameters that could be parsed.
+ *
+ */
+static int parse_write_buffer_into_params(char *wr_buf, uint32_t wr_buf_size,
+					  long *param, const char __user *buf,
+					  int max_param_num,
+					  uint8_t *param_nums)
+{
+	char *wr_buf_ptr = NULL;
+	uint32_t wr_buf_count = 0;
+	int r;
+	char *sub_str = NULL;
+	const char delimiter[3] = {' ', '\n', '\0'};
+	uint8_t param_index = 0;
+
+	*param_nums = 0;
+
+	wr_buf_ptr = wr_buf;
+
+	/* r is bytes not be copied */
+	if (copy_from_user(wr_buf_ptr, buf, wr_buf_size)) {
+		DRM_DEBUG_DRIVER("user data could not be read successfully\n");
+		return -EFAULT;
+	}
+
+	/* check number of parameters. isspace could not differ space and \n */
+	while ((*wr_buf_ptr != 0xa) && (wr_buf_count < wr_buf_size)) {
+		/* skip space*/
+		while (isspace(*wr_buf_ptr) && (wr_buf_count < wr_buf_size)) {
+			wr_buf_ptr++;
+			wr_buf_count++;
+			}
+
+		if (wr_buf_count == wr_buf_size)
+			break;
+
+		/* skip non-space*/
+		while ((!isspace(*wr_buf_ptr)) && (wr_buf_count < wr_buf_size)) {
+			wr_buf_ptr++;
+			wr_buf_count++;
+		}
+
+		(*param_nums)++;
+
+		if (wr_buf_count == wr_buf_size)
+			break;
+	}
+
+	if (*param_nums > max_param_num)
+		*param_nums = max_param_num;
+
+	wr_buf_ptr = wr_buf; /* reset buf pointer */
+	wr_buf_count = 0; /* number of char already checked */
+
+	while (isspace(*wr_buf_ptr) && (wr_buf_count < wr_buf_size)) {
+		wr_buf_ptr++;
+		wr_buf_count++;
+	}
+
+	while (param_index < *param_nums) {
+		/* after strsep, wr_buf_ptr will be moved to after space */
+		sub_str = strsep(&wr_buf_ptr, delimiter);
+
+		r = kstrtol(sub_str, 16, &(param[param_index]));
+
+		if (r)
+			DRM_DEBUG_DRIVER("string to int convert error code: %d\n", r);
+
+		param_index++;
+	}
+
+	return 0;
+}
+
+/* function description
+ * get/ set DP configuration: lane_count, link_rate, spread_spectrum
+ *
+ * valid lane count value: 1, 2, 4
+ * valid link rate value:
+ * 06h = 1.62Gbps per lane
+ * 0Ah = 2.7Gbps per lane
+ * 0Ch = 3.24Gbps per lane
+ * 14h = 5.4Gbps per lane
+ * 1Eh = 8.1Gbps per lane
+ *
+ * debugfs is located at /sys/kernel/debug/dri/0/DP-x/link_settings
+ *
+ * --- to get dp configuration
+ *
+ * cat /sys/kernel/debug/dri/0/DP-x/link_settings
+ *
+ * It will list current, verified, reported, preferred dp configuration.
+ * current -- for current video mode
+ * verified --- maximum configuration which pass link training
+ * reported --- DP rx report caps (DPCD register offset 0, 1 2)
+ * preferred --- user force settings
+ *
+ * --- set (or force) dp configuration
+ *
+ * echo <lane_count>  <link_rate> > link_settings
+ *
+ * for example, to force to  2 lane, 2.7GHz,
+ * echo 4 0xa > /sys/kernel/debug/dri/0/DP-x/link_settings
+ *
+ * spread_spectrum could not be changed dynamically.
+ *
+ * in case invalid lane count, link rate are force, no hw programming will be
+ * done. please check link settings after force operation to see if HW get
+ * programming.
+ *
+ * cat /sys/kernel/debug/dri/0/DP-x/link_settings
+ *
+ * check current and preferred settings.
+ *
+ */
+static ssize_t dp_link_settings_read(struct file *f, char __user *buf,
+				 size_t size, loff_t *pos)
+{
+	struct amdgpu_dm_connector *connector = file_inode(f)->i_private;
+	struct dc_link *link = connector->dc_link;
+	char *rd_buf = NULL;
+	char *rd_buf_ptr = NULL;
+	const uint32_t rd_buf_size = 100;
+	uint32_t result = 0;
+	uint8_t str_len = 0;
+	int r;
+
+	if (*pos & 3 || size & 3)
+		return -EINVAL;
+
+	rd_buf = kcalloc(rd_buf_size, sizeof(char), GFP_KERNEL);
+	if (!rd_buf)
+		return 0;
+
+	rd_buf_ptr = rd_buf;
+
+	str_len = strlen("Current:  %d  0x%x  %d  ");
+	snprintf(rd_buf_ptr, str_len, "Current:  %d  0x%x  %d  ",
+			link->cur_link_settings.lane_count,
+			link->cur_link_settings.link_rate,
+			link->cur_link_settings.link_spread);
+	rd_buf_ptr += str_len;
+
+	str_len = strlen("Verified:  %d  0x%x  %d  ");
+	snprintf(rd_buf_ptr, str_len, "Verified:  %d  0x%x  %d  ",
+			link->verified_link_cap.lane_count,
+			link->verified_link_cap.link_rate,
+			link->verified_link_cap.link_spread);
+	rd_buf_ptr += str_len;
+
+	str_len = strlen("Reported:  %d  0x%x  %d  ");
+	snprintf(rd_buf_ptr, str_len, "Reported:  %d  0x%x  %d  ",
+			link->reported_link_cap.lane_count,
+			link->reported_link_cap.link_rate,
+			link->reported_link_cap.link_spread);
+	rd_buf_ptr += str_len;
+
+	str_len = strlen("Preferred:  %d  0x%x  %d  ");
+	snprintf(rd_buf_ptr, str_len, "Preferred:  %d  0x%x  %d\n",
+			link->preferred_link_setting.lane_count,
+			link->preferred_link_setting.link_rate,
+			link->preferred_link_setting.link_spread);
+
+	while (size) {
+		if (*pos >= rd_buf_size)
+			break;
+
+		r = put_user(*(rd_buf + result), buf);
+		if (r) {
+			kfree(rd_buf);
+			return r; /* r = -EFAULT */
+		}
+
+		buf += 1;
+		size -= 1;
+		*pos += 1;
+		result += 1;
+	}
+
+	kfree(rd_buf);
+	return result;
+}
+
+static ssize_t dp_link_settings_write(struct file *f, const char __user *buf,
+				 size_t size, loff_t *pos)
+{
+	struct amdgpu_dm_connector *connector = file_inode(f)->i_private;
+	struct dc_link *link = connector->dc_link;
+	struct amdgpu_device *adev = drm_to_adev(connector->base.dev);
+	struct dc *dc = (struct dc *)link->dc;
+	struct dc_link_settings prefer_link_settings = {0};
+	char *wr_buf = NULL;
+	const uint32_t wr_buf_size = 40;
+	/* 0: lane_count; 1: link_rate */
+	int max_param_num = 2;
+	uint8_t param_nums = 0;
+	long param[2];
+	bool valid_input = true;
+
+	if (size == 0)
+		return -EINVAL;
+
+	wr_buf = kcalloc(wr_buf_size, sizeof(char), GFP_KERNEL);
+	if (!wr_buf)
+		return -ENOSPC;
+
+	if (parse_write_buffer_into_params(wr_buf, wr_buf_size,
+					   (long *)param, buf,
+					   max_param_num,
+					   &param_nums)) {
+		kfree(wr_buf);
+		return -EINVAL;
+	}
+
+	if (param_nums <= 0) {
+		kfree(wr_buf);
+		DRM_DEBUG_DRIVER("user data not be read\n");
+		return -EINVAL;
+	}
+
+	switch (param[0]) {
+	case LANE_COUNT_ONE:
+	case LANE_COUNT_TWO:
+	case LANE_COUNT_FOUR:
+		break;
+	default:
+		valid_input = false;
+		break;
+	}
+
+	switch (param[1]) {
+	case LINK_RATE_LOW:
+	case LINK_RATE_HIGH:
+	case LINK_RATE_RBR2:
+	case LINK_RATE_HIGH2:
+	case LINK_RATE_HIGH3:
+	case LINK_RATE_UHBR10:
+	case LINK_RATE_UHBR13_5:
+	case LINK_RATE_UHBR20:
+		break;
+	default:
+		valid_input = false;
+		break;
+	}
+
+	if (!valid_input) {
+		kfree(wr_buf);
+		DRM_DEBUG_DRIVER("Invalid Input value No HW will be programmed\n");
+		mutex_lock(&adev->dm.dc_lock);
+		dc_link_set_preferred_training_settings(dc, NULL, NULL, link, false);
+		mutex_unlock(&adev->dm.dc_lock);
+		return size;
+	}
+
+	/* save user force lane_count, link_rate to preferred settings
+	 * spread spectrum will not be changed
+	 */
+	prefer_link_settings.link_spread = link->cur_link_settings.link_spread;
+	prefer_link_settings.use_link_rate_set = false;
+	prefer_link_settings.lane_count = param[0];
+	prefer_link_settings.link_rate = param[1];
+
+	mutex_lock(&adev->dm.dc_lock);
+	dc_link_set_preferred_training_settings(dc, &prefer_link_settings, NULL, link, false);
+	mutex_unlock(&adev->dm.dc_lock);
+
+	kfree(wr_buf);
+	return size;
+}
+
+static bool dp_mst_is_end_device(struct amdgpu_dm_connector *aconnector)
+{
+	bool is_end_device = false;
+	struct drm_dp_mst_topology_mgr *mgr = NULL;
+	struct drm_dp_mst_port *port = NULL;
+
+	if (aconnector->mst_root && aconnector->mst_root->mst_mgr.mst_state) {
+		mgr = &aconnector->mst_root->mst_mgr;
+		port = aconnector->mst_output_port;
+
+		drm_modeset_lock(&mgr->base.lock, NULL);
+		if (port->pdt == DP_PEER_DEVICE_SST_SINK ||
+			port->pdt == DP_PEER_DEVICE_DP_LEGACY_CONV)
+			is_end_device = true;
+		drm_modeset_unlock(&mgr->base.lock);
+	}
+
+	return is_end_device;
+}
+
+/* Change MST link setting
+ *
+ * valid lane count value: 1, 2, 4
+ * valid link rate value:
+ * 06h = 1.62Gbps per lane
+ * 0Ah = 2.7Gbps per lane
+ * 0Ch = 3.24Gbps per lane
+ * 14h = 5.4Gbps per lane
+ * 1Eh = 8.1Gbps per lane
+ * 3E8h = 10.0Gbps per lane
+ * 546h = 13.5Gbps per lane
+ * 7D0h = 20.0Gbps per lane
+ *
+ * debugfs is located at /sys/kernel/debug/dri/0/DP-x/mst_link_settings
+ *
+ * for example, to force to  2 lane, 10.0GHz,
+ * echo 2 0x3e8 > /sys/kernel/debug/dri/0/DP-x/mst_link_settings
+ *
+ * Valid input will trigger hotplug event to get new link setting applied
+ * Invalid input will trigger training setting reset
+ *
+ * The usage can be referred to link_settings entry
+ *
+ */
+static ssize_t dp_mst_link_setting(struct file *f, const char __user *buf,
+				 size_t size, loff_t *pos)
+{
+	struct amdgpu_dm_connector *aconnector = file_inode(f)->i_private;
+	struct dc_link *link = aconnector->dc_link;
+	struct amdgpu_device *adev = drm_to_adev(aconnector->base.dev);
+	struct dc *dc = (struct dc *)link->dc;
+	struct dc_link_settings prefer_link_settings = {0};
+	char *wr_buf = NULL;
+	const uint32_t wr_buf_size = 40;
+	/* 0: lane_count; 1: link_rate */
+	int max_param_num = 2;
+	uint8_t param_nums = 0;
+	long param[2];
+	bool valid_input = true;
+
+	if (!dp_mst_is_end_device(aconnector))
+		return -EINVAL;
+
+	if (size == 0)
+		return -EINVAL;
+
+	wr_buf = kcalloc(wr_buf_size, sizeof(char), GFP_KERNEL);
+	if (!wr_buf)
+		return -ENOSPC;
+
+	if (parse_write_buffer_into_params(wr_buf, wr_buf_size,
+					   (long *)param, buf,
+					   max_param_num,
+					   &param_nums)) {
+		kfree(wr_buf);
+		return -EINVAL;
+	}
+
+	if (param_nums <= 0) {
+		kfree(wr_buf);
+		DRM_DEBUG_DRIVER("user data not be read\n");
+		return -EINVAL;
+	}
+
+	switch (param[0]) {
+	case LANE_COUNT_ONE:
+	case LANE_COUNT_TWO:
+	case LANE_COUNT_FOUR:
+		break;
+	default:
+		valid_input = false;
+		break;
+	}
+
+	switch (param[1]) {
+	case LINK_RATE_LOW:
+	case LINK_RATE_HIGH:
+	case LINK_RATE_RBR2:
+	case LINK_RATE_HIGH2:
+	case LINK_RATE_HIGH3:
+	case LINK_RATE_UHBR10:
+	case LINK_RATE_UHBR13_5:
+	case LINK_RATE_UHBR20:
+		break;
+	default:
+		valid_input = false;
+		break;
+	}
+
+	if (!valid_input) {
+		kfree(wr_buf);
+		DRM_DEBUG_DRIVER("Invalid Input value No HW will be programmed\n");
+		mutex_lock(&adev->dm.dc_lock);
+		dc_link_set_preferred_training_settings(dc, NULL, NULL, link, false);
+		mutex_unlock(&adev->dm.dc_lock);
+		return -EINVAL;
+	}
+
+	/* save user force lane_count, link_rate to preferred settings
+	 * spread spectrum will not be changed
+	 */
+	prefer_link_settings.link_spread = link->cur_link_settings.link_spread;
+	prefer_link_settings.use_link_rate_set = false;
+	prefer_link_settings.lane_count = param[0];
+	prefer_link_settings.link_rate = param[1];
+
+	/* skip immediate retrain, and train to new link setting after hotplug event triggered */
+	mutex_lock(&adev->dm.dc_lock);
+	dc_link_set_preferred_training_settings(dc, &prefer_link_settings, NULL, link, true);
+	mutex_unlock(&adev->dm.dc_lock);
+
+	mutex_lock(&aconnector->base.dev->mode_config.mutex);
+	aconnector->base.force = DRM_FORCE_OFF;
+	mutex_unlock(&aconnector->base.dev->mode_config.mutex);
+	drm_kms_helper_hotplug_event(aconnector->base.dev);
+
+	msleep(100);
+
+	mutex_lock(&aconnector->base.dev->mode_config.mutex);
+	aconnector->base.force = DRM_FORCE_UNSPECIFIED;
+	mutex_unlock(&aconnector->base.dev->mode_config.mutex);
+	drm_kms_helper_hotplug_event(aconnector->base.dev);
+
+	kfree(wr_buf);
+	return size;
+}
+
+/* function: get current DP PHY settings: voltage swing, pre-emphasis,
+ * post-cursor2 (defined by VESA DP specification)
+ *
+ * valid values
+ * voltage swing: 0,1,2,3
+ * pre-emphasis : 0,1,2,3
+ * post cursor2 : 0,1,2,3
+ *
+ *
+ * how to use this debugfs
+ *
+ * debugfs is located at /sys/kernel/debug/dri/0/DP-x
+ *
+ * there will be directories, like DP-1, DP-2,DP-3, etc. for DP display
+ *
+ * To figure out which DP-x is the display for DP to be check,
+ * cd DP-x
+ * ls -ll
+ * There should be debugfs file, like link_settings, phy_settings.
+ * cat link_settings
+ * from lane_count, link_rate to figure which DP-x is for display to be worked
+ * on
+ *
+ * To get current DP PHY settings,
+ * cat phy_settings
+ *
+ * To change DP PHY settings,
+ * echo <voltage_swing> <pre-emphasis> <post_cursor2> > phy_settings
+ * for examle, to change voltage swing to 2, pre-emphasis to 3, post_cursor2 to
+ * 0,
+ * echo 2 3 0 > phy_settings
+ *
+ * To check if change be applied, get current phy settings by
+ * cat phy_settings
+ *
+ * In case invalid values are set by user, like
+ * echo 1 4 0 > phy_settings
+ *
+ * HW will NOT be programmed by these settings.
+ * cat phy_settings will show the previous valid settings.
+ */
+static ssize_t dp_phy_settings_read(struct file *f, char __user *buf,
+				 size_t size, loff_t *pos)
+{
+	struct amdgpu_dm_connector *connector = file_inode(f)->i_private;
+	struct dc_link *link = connector->dc_link;
+	char *rd_buf = NULL;
+	const uint32_t rd_buf_size = 20;
+	uint32_t result = 0;
+	int r;
+
+	if (*pos & 3 || size & 3)
+		return -EINVAL;
+
+	rd_buf = kcalloc(rd_buf_size, sizeof(char), GFP_KERNEL);
+	if (!rd_buf)
+		return -EINVAL;
+
+	snprintf(rd_buf, rd_buf_size, "  %d  %d  %d\n",
+			link->cur_lane_setting[0].VOLTAGE_SWING,
+			link->cur_lane_setting[0].PRE_EMPHASIS,
+			link->cur_lane_setting[0].POST_CURSOR2);
+
+	while (size) {
+		if (*pos >= rd_buf_size)
+			break;
+
+		r = put_user((*(rd_buf + result)), buf);
+		if (r) {
+			kfree(rd_buf);
+			return r; /* r = -EFAULT */
+		}
+
+		buf += 1;
+		size -= 1;
+		*pos += 1;
+		result += 1;
+	}
+
+	kfree(rd_buf);
+	return result;
+}
+
+static int dp_lttpr_status_show(struct seq_file *m, void *unused)
+{
+	struct drm_connector *connector = m->private;
+	struct amdgpu_dm_connector *aconnector =
+		to_amdgpu_dm_connector(connector);
+	struct dc_lttpr_caps caps = aconnector->dc_link->dpcd_caps.lttpr_caps;
+
+	if (connector->status != connector_status_connected)
+		return -ENODEV;
+
+	seq_printf(m, "phy repeater count: %u (raw: 0x%x)\n",
+		   dp_parse_lttpr_repeater_count(caps.phy_repeater_cnt),
+		   caps.phy_repeater_cnt);
+
+	seq_puts(m, "phy repeater mode: ");
+
+	switch (caps.mode) {
+	case DP_PHY_REPEATER_MODE_TRANSPARENT:
+		seq_puts(m, "transparent");
+		break;
+	case DP_PHY_REPEATER_MODE_NON_TRANSPARENT:
+		seq_puts(m, "non-transparent");
+		break;
+	case 0x00:
+		seq_puts(m, "non lttpr");
+		break;
+	default:
+		seq_printf(m, "read error (raw: 0x%x)", caps.mode);
+		break;
+	}
+
+	seq_puts(m, "\n");
+	return 0;
+}
+
+static ssize_t dp_phy_settings_write(struct file *f, const char __user *buf,
+				 size_t size, loff_t *pos)
+{
+	struct amdgpu_dm_connector *connector = file_inode(f)->i_private;
+	struct dc_link *link = connector->dc_link;
+	struct dc *dc = (struct dc *)link->dc;
+	char *wr_buf = NULL;
+	uint32_t wr_buf_size = 40;
+	long param[3];
+	bool use_prefer_link_setting;
+	struct link_training_settings link_lane_settings = {0};
+	int max_param_num = 3;
+	uint8_t param_nums = 0;
+	int r = 0;
+
+
+	if (size == 0)
+		return -EINVAL;
+
+	wr_buf = kcalloc(wr_buf_size, sizeof(char), GFP_KERNEL);
+	if (!wr_buf)
+		return -ENOSPC;
+
+	if (parse_write_buffer_into_params(wr_buf, wr_buf_size,
+					   (long *)param, buf,
+					   max_param_num,
+					   &param_nums)) {
+		kfree(wr_buf);
+		return -EINVAL;
+	}
+
+	if (param_nums <= 0) {
+		kfree(wr_buf);
+		DRM_DEBUG_DRIVER("user data not be read\n");
+		return -EINVAL;
+	}
+
+	if ((param[0] > VOLTAGE_SWING_MAX_LEVEL) ||
+			(param[1] > PRE_EMPHASIS_MAX_LEVEL) ||
+			(param[2] > POST_CURSOR2_MAX_LEVEL)) {
+		kfree(wr_buf);
+		DRM_DEBUG_DRIVER("Invalid Input No HW will be programmed\n");
+		return size;
+	}
+
+	/* get link settings: lane count, link rate */
+	use_prefer_link_setting =
+		((link->preferred_link_setting.link_rate != LINK_RATE_UNKNOWN) &&
+		(link->test_pattern_enabled));
+
+	memset(&link_lane_settings, 0, sizeof(link_lane_settings));
+
+	if (use_prefer_link_setting) {
+		link_lane_settings.link_settings.lane_count =
+				link->preferred_link_setting.lane_count;
+		link_lane_settings.link_settings.link_rate =
+				link->preferred_link_setting.link_rate;
+		link_lane_settings.link_settings.link_spread =
+				link->preferred_link_setting.link_spread;
+	} else {
+		link_lane_settings.link_settings.lane_count =
+				link->cur_link_settings.lane_count;
+		link_lane_settings.link_settings.link_rate =
+				link->cur_link_settings.link_rate;
+		link_lane_settings.link_settings.link_spread =
+				link->cur_link_settings.link_spread;
+	}
+
+	/* apply phy settings from user */
+	for (r = 0; r < link_lane_settings.link_settings.lane_count; r++) {
+		link_lane_settings.hw_lane_settings[r].VOLTAGE_SWING =
+				(enum dc_voltage_swing) (param[0]);
+		link_lane_settings.hw_lane_settings[r].PRE_EMPHASIS =
+				(enum dc_pre_emphasis) (param[1]);
+		link_lane_settings.hw_lane_settings[r].POST_CURSOR2 =
+				(enum dc_post_cursor2) (param[2]);
+	}
+
+	/* program ASIC registers and DPCD registers */
+	dc_link_set_drive_settings(dc, &link_lane_settings, link);
+
+	kfree(wr_buf);
+	return size;
+}
+
+/* function description
+ *
+ * set PHY layer or Link layer test pattern
+ * PHY test pattern is used for PHY SI check.
+ * Link layer test will not affect PHY SI.
+ *
+ * Reset Test Pattern:
+ * 0 = DP_TEST_PATTERN_VIDEO_MODE
+ *
+ * PHY test pattern supported:
+ * 1 = DP_TEST_PATTERN_D102
+ * 2 = DP_TEST_PATTERN_SYMBOL_ERROR
+ * 3 = DP_TEST_PATTERN_PRBS7
+ * 4 = DP_TEST_PATTERN_80BIT_CUSTOM
+ * 5 = DP_TEST_PATTERN_CP2520_1
+ * 6 = DP_TEST_PATTERN_CP2520_2 = DP_TEST_PATTERN_HBR2_COMPLIANCE_EYE
+ * 7 = DP_TEST_PATTERN_CP2520_3
+ *
+ * DP PHY Link Training Patterns
+ * 8 = DP_TEST_PATTERN_TRAINING_PATTERN1
+ * 9 = DP_TEST_PATTERN_TRAINING_PATTERN2
+ * a = DP_TEST_PATTERN_TRAINING_PATTERN3
+ * b = DP_TEST_PATTERN_TRAINING_PATTERN4
+ *
+ * DP Link Layer Test pattern
+ * c = DP_TEST_PATTERN_COLOR_SQUARES
+ * d = DP_TEST_PATTERN_COLOR_SQUARES_CEA
+ * e = DP_TEST_PATTERN_VERTICAL_BARS
+ * f = DP_TEST_PATTERN_HORIZONTAL_BARS
+ * 10= DP_TEST_PATTERN_COLOR_RAMP
+ *
+ * debugfs phy_test_pattern is located at /syskernel/debug/dri/0/DP-x
+ *
+ * --- set test pattern
+ * echo <test pattern #> > test_pattern
+ *
+ * If test pattern # is not supported, NO HW programming will be done.
+ * for DP_TEST_PATTERN_80BIT_CUSTOM, it needs extra 10 bytes of data
+ * for the user pattern. input 10 bytes data are separated by space
+ *
+ * echo 0x4 0x11 0x22 0x33 0x44 0x55 0x66 0x77 0x88 0x99 0xaa > test_pattern
+ *
+ * --- reset test pattern
+ * echo 0 > test_pattern
+ *
+ * --- HPD detection is disabled when set PHY test pattern
+ *
+ * when PHY test pattern (pattern # within [1,7]) is set, HPD pin of HW ASIC
+ * is disable. User could unplug DP display from DP connected and plug scope to
+ * check test pattern PHY SI.
+ * If there is need unplug scope and plug DP display back, do steps below:
+ * echo 0 > phy_test_pattern
+ * unplug scope
+ * plug DP display.
+ *
+ * "echo 0 > phy_test_pattern" will re-enable HPD pin again so that video sw
+ * driver could detect "unplug scope" and "plug DP display"
+ */
+static ssize_t dp_phy_test_pattern_debugfs_write(struct file *f, const char __user *buf,
+				 size_t size, loff_t *pos)
+{
+	struct amdgpu_dm_connector *connector = file_inode(f)->i_private;
+	struct dc_link *link = connector->dc_link;
+	char *wr_buf = NULL;
+	uint32_t wr_buf_size = 100;
+	long param[11] = {0x0};
+	int max_param_num = 11;
+	enum dp_test_pattern test_pattern = DP_TEST_PATTERN_UNSUPPORTED;
+	bool disable_hpd = false;
+	bool valid_test_pattern = false;
+	uint8_t param_nums = 0;
+	/* init with default 80bit custom pattern */
+	uint8_t custom_pattern[10] = {
+			0x1f, 0x7c, 0xf0, 0xc1, 0x07,
+			0x1f, 0x7c, 0xf0, 0xc1, 0x07
+			};
+	struct dc_link_settings prefer_link_settings = {LANE_COUNT_UNKNOWN,
+			LINK_RATE_UNKNOWN, LINK_SPREAD_DISABLED};
+	struct dc_link_settings cur_link_settings = {LANE_COUNT_UNKNOWN,
+			LINK_RATE_UNKNOWN, LINK_SPREAD_DISABLED};
+	struct link_training_settings link_training_settings = {0};
+	int i;
+
+	if (size == 0)
+		return -EINVAL;
+
+	wr_buf = kcalloc(wr_buf_size, sizeof(char), GFP_KERNEL);
+	if (!wr_buf)
+		return -ENOSPC;
+
+	if (parse_write_buffer_into_params(wr_buf, wr_buf_size,
+					   (long *)param, buf,
+					   max_param_num,
+					   &param_nums)) {
+		kfree(wr_buf);
+		return -EINVAL;
+	}
+
+	if (param_nums <= 0) {
+		kfree(wr_buf);
+		DRM_DEBUG_DRIVER("user data not be read\n");
+		return -EINVAL;
+	}
+
+
+	test_pattern = param[0];
+
+	switch (test_pattern) {
+	case DP_TEST_PATTERN_VIDEO_MODE:
+	case DP_TEST_PATTERN_COLOR_SQUARES:
+	case DP_TEST_PATTERN_COLOR_SQUARES_CEA:
+	case DP_TEST_PATTERN_VERTICAL_BARS:
+	case DP_TEST_PATTERN_HORIZONTAL_BARS:
+	case DP_TEST_PATTERN_COLOR_RAMP:
+		valid_test_pattern = true;
+		break;
+
+	case DP_TEST_PATTERN_D102:
+	case DP_TEST_PATTERN_SYMBOL_ERROR:
+	case DP_TEST_PATTERN_PRBS7:
+	case DP_TEST_PATTERN_80BIT_CUSTOM:
+	case DP_TEST_PATTERN_HBR2_COMPLIANCE_EYE:
+	case DP_TEST_PATTERN_TRAINING_PATTERN4:
+		disable_hpd = true;
+		valid_test_pattern = true;
+		break;
+
+	default:
+		valid_test_pattern = false;
+		test_pattern = DP_TEST_PATTERN_UNSUPPORTED;
+		break;
+	}
+
+	if (!valid_test_pattern) {
+		kfree(wr_buf);
+		DRM_DEBUG_DRIVER("Invalid Test Pattern Parameters\n");
+		return size;
+	}
+
+	if (test_pattern == DP_TEST_PATTERN_80BIT_CUSTOM) {
+		for (i = 0; i < 10; i++) {
+			if ((uint8_t) param[i + 1] != 0x0)
+				break;
+		}
+
+		if (i < 10) {
+			/* not use default value */
+			for (i = 0; i < 10; i++)
+				custom_pattern[i] = (uint8_t) param[i + 1];
+		}
+	}
+
+	/* Usage: set DP physical test pattern using debugfs with normal DP
+	 * panel. Then plug out DP panel and connect a scope to measure
+	 * For normal video mode and test pattern generated from CRCT,
+	 * they are visibile to user. So do not disable HPD.
+	 * Video Mode is also set to clear the test pattern, so enable HPD
+	 * because it might have been disabled after a test pattern was set.
+	 * AUX depends on HPD * sequence dependent, do not move!
+	 */
+	if (!disable_hpd)
+		dc_link_enable_hpd(link);
+
+	prefer_link_settings.lane_count = link->verified_link_cap.lane_count;
+	prefer_link_settings.link_rate = link->verified_link_cap.link_rate;
+	prefer_link_settings.link_spread = link->verified_link_cap.link_spread;
+
+	cur_link_settings.lane_count = link->cur_link_settings.lane_count;
+	cur_link_settings.link_rate = link->cur_link_settings.link_rate;
+	cur_link_settings.link_spread = link->cur_link_settings.link_spread;
+
+	link_training_settings.link_settings = cur_link_settings;
+
+
+	if (test_pattern != DP_TEST_PATTERN_VIDEO_MODE) {
+		if (prefer_link_settings.lane_count != LANE_COUNT_UNKNOWN &&
+			prefer_link_settings.link_rate !=  LINK_RATE_UNKNOWN &&
+			(prefer_link_settings.lane_count != cur_link_settings.lane_count ||
+			prefer_link_settings.link_rate != cur_link_settings.link_rate))
+			link_training_settings.link_settings = prefer_link_settings;
+	}
+
+	for (i = 0; i < (unsigned int)(link_training_settings.link_settings.lane_count); i++)
+		link_training_settings.hw_lane_settings[i] = link->cur_lane_setting[i];
+
+	dc_link_dp_set_test_pattern(
+		link,
+		test_pattern,
+		DP_TEST_PATTERN_COLOR_SPACE_RGB,
+		&link_training_settings,
+		custom_pattern,
+		10);
+
+	/* Usage: Set DP physical test pattern using AMDDP with normal DP panel
+	 * Then plug out DP panel and connect a scope to measure DP PHY signal.
+	 * Need disable interrupt to avoid SW driver disable DP output. This is
+	 * done after the test pattern is set.
+	 */
+	if (valid_test_pattern && disable_hpd)
+		dc_link_disable_hpd(link);
+
+	kfree(wr_buf);
+
+	return size;
+}
+
+/*
+ * Returns the DMCUB tracebuffer contents.
+ * Example usage: cat /sys/kernel/debug/dri/0/amdgpu_dm_dmub_tracebuffer
+ */
+static int dmub_tracebuffer_show(struct seq_file *m, void *data)
+{
+	struct amdgpu_device *adev = m->private;
+	struct dmub_srv_fb_info *fb_info = adev->dm.dmub_fb_info;
+	struct dmub_fw_meta_info *fw_meta_info = NULL;
+	struct dmub_debugfs_trace_entry *entries;
+	uint8_t *tbuf_base;
+	uint32_t tbuf_size, max_entries, num_entries, first_entry, i;
+
+	if (!fb_info)
+		return 0;
+
+	tbuf_base = (uint8_t *)fb_info->fb[DMUB_WINDOW_5_TRACEBUFF].cpu_addr;
+	if (!tbuf_base)
+		return 0;
+
+	if (adev->dm.dmub_srv)
+		fw_meta_info = &adev->dm.dmub_srv->meta_info;
+
+	tbuf_size = fw_meta_info ? fw_meta_info->trace_buffer_size :
+				   DMUB_TRACE_BUFFER_SIZE;
+	max_entries = (tbuf_size - sizeof(struct dmub_debugfs_trace_header)) /
+		      sizeof(struct dmub_debugfs_trace_entry);
+
+	num_entries =
+		((struct dmub_debugfs_trace_header *)tbuf_base)->entry_count;
+
+	/* DMCUB tracebuffer is a ring. If it rolled over, print a hint that
+	 * entries are being overwritten.
+	 */
+	if (num_entries > max_entries)
+		seq_printf(m, "...\n");
+
+	first_entry = num_entries % max_entries;
+	num_entries = min(num_entries, max_entries);
+
+	entries = (struct dmub_debugfs_trace_entry
+			   *)(tbuf_base +
+			      sizeof(struct dmub_debugfs_trace_header));
+
+	/* To print entries chronologically, start from the first entry till the
+	 * top of buffer, then from base of buffer to first entry.
+	 */
+	for (i = first_entry; i < num_entries; ++i) {
+		struct dmub_debugfs_trace_entry *entry = &entries[i];
+
+		seq_printf(m,
+			   "trace_code=%u tick_count=%u param0=%u param1=%u\n",
+			   entry->trace_code, entry->tick_count, entry->param0,
+			   entry->param1);
+	}
+	for (i = 0; i < first_entry; ++i) {
+		struct dmub_debugfs_trace_entry *entry = &entries[i];
+
+		seq_printf(m,
+			   "trace_code=%u tick_count=%u param0=%u param1=%u\n",
+			   entry->trace_code, entry->tick_count, entry->param0,
+			   entry->param1);
+	}
+
+	return 0;
+}
+
+/*
+ * Returns the DMCUB firmware state contents.
+ * Example usage: cat /sys/kernel/debug/dri/0/amdgpu_dm_dmub_fw_state
+ */
+static int dmub_fw_state_show(struct seq_file *m, void *data)
+{
+	struct amdgpu_device *adev = m->private;
+	struct dmub_srv_fb_info *fb_info = adev->dm.dmub_fb_info;
+	uint8_t *state_base;
+	uint32_t state_size;
+
+	if (!fb_info)
+		return 0;
+
+	state_base = (uint8_t *)fb_info->fb[DMUB_WINDOW_6_FW_STATE].cpu_addr;
+	if (!state_base)
+		return 0;
+
+	state_size = fb_info->fb[DMUB_WINDOW_6_FW_STATE].size;
+
+	return seq_write(m, state_base, state_size);
+}
+
+/* replay_capability_show() - show eDP panel replay capability
+ *
+ * The read function: replay_capability_show
+ * Shows if sink and driver has Replay capability or not.
+ *
+ *	cat /sys/kernel/debug/dri/0/eDP-X/replay_capability
+ *
+ * Expected output:
+ * "Sink support: no\n" - if panel doesn't support Replay
+ * "Sink support: yes\n" - if panel supports Replay
+ * "Driver support: no\n" - if driver doesn't support Replay
+ * "Driver support: yes\n" - if driver supports Replay
+ */
+static int replay_capability_show(struct seq_file *m, void *data)
+{
+	struct drm_connector *connector = m->private;
+	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector);
+	struct dc_link *link = aconnector->dc_link;
+	bool sink_support_replay = false;
+	bool driver_support_replay = false;
+
+	if (!link)
+		return -ENODEV;
+
+	if (link->type == dc_connection_none)
+		return -ENODEV;
+
+	if (!(link->connector_signal & SIGNAL_TYPE_EDP))
+		return -ENODEV;
+
+	/* If Replay is already set to support, skip the checks */
+	if (link->replay_settings.config.replay_supported) {
+		sink_support_replay = true;
+		driver_support_replay = true;
+	} else if ((amdgpu_dc_debug_mask & DC_DISABLE_REPLAY)) {
+		sink_support_replay = amdgpu_dm_link_supports_replay(link, aconnector);
+	} else {
+		struct dc *dc = link->ctx->dc;
+
+		sink_support_replay = amdgpu_dm_link_supports_replay(link, aconnector);
+		if (dc->ctx->dmub_srv && dc->ctx->dmub_srv->dmub)
+			driver_support_replay =
+				(bool)dc->ctx->dmub_srv->dmub->feature_caps.replay_supported;
+	}
+
+	seq_printf(m, "Sink support: %s\n", str_yes_no(sink_support_replay));
+	seq_printf(m, "Driver support: %s\n", str_yes_no(driver_support_replay));
+	seq_printf(m, "Config support: %s\n", str_yes_no(link->replay_settings.config.replay_supported));
+
+	return 0;
+}
+
+/* psr_capability_show() - show eDP panel PSR capability
+ *
+ * The read function: sink_psr_capability_show
+ * Shows if sink has PSR capability or not.
+ * If yes - the PSR version is appended
+ *
+ *	cat /sys/kernel/debug/dri/0/eDP-X/psr_capability
+ *
+ * Expected output:
+ * "Sink support: no\n" - if panel doesn't support PSR
+ * "Sink support: yes [0x01]\n" - if panel supports PSR1
+ * "Driver support: no\n" - if driver doesn't support PSR
+ * "Driver support: yes [0x01]\n" - if driver supports PSR1
+ */
+static int psr_capability_show(struct seq_file *m, void *data)
+{
+	struct drm_connector *connector = m->private;
+	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector);
+	struct dc_link *link = aconnector->dc_link;
+
+	if (!link)
+		return -ENODEV;
+
+	if (link->type == dc_connection_none)
+		return -ENODEV;
+
+	if (!(link->connector_signal & SIGNAL_TYPE_EDP))
+		return -ENODEV;
+
+	seq_printf(m, "Sink support: %s", str_yes_no(link->dpcd_caps.psr_info.psr_version != 0));
+	if (link->dpcd_caps.psr_info.psr_version)
+		seq_printf(m, " [0x%02x]", link->dpcd_caps.psr_info.psr_version);
+	seq_puts(m, "\n");
+
+	seq_printf(m, "Driver support: %s", str_yes_no(link->psr_settings.psr_feature_enabled));
+	if (link->psr_settings.psr_version)
+		seq_printf(m, " [0x%02x]", link->psr_settings.psr_version);
+	seq_puts(m, "\n");
+
+	return 0;
+}
+
+/*
+ * Returns the current bpc for the crtc.
+ * Example usage: cat /sys/kernel/debug/dri/0/crtc-0/amdgpu_current_bpc
+ */
+static int amdgpu_current_bpc_show(struct seq_file *m, void *data)
+{
+	struct drm_crtc *crtc = m->private;
+	struct drm_device *dev = crtc->dev;
+	struct dm_crtc_state *dm_crtc_state = NULL;
+	int res = -ENODEV;
+	unsigned int bpc;
+
+	mutex_lock(&dev->mode_config.mutex);
+	drm_modeset_lock(&crtc->mutex, NULL);
+	if (crtc->state == NULL)
+		goto unlock;
+
+	dm_crtc_state = to_dm_crtc_state(crtc->state);
+	if (dm_crtc_state->stream == NULL)
+		goto unlock;
+
+	switch (dm_crtc_state->stream->timing.display_color_depth) {
+	case COLOR_DEPTH_666:
+		bpc = 6;
+		break;
+	case COLOR_DEPTH_888:
+		bpc = 8;
+		break;
+	case COLOR_DEPTH_101010:
+		bpc = 10;
+		break;
+	case COLOR_DEPTH_121212:
+		bpc = 12;
+		break;
+	case COLOR_DEPTH_161616:
+		bpc = 16;
+		break;
+	default:
+		goto unlock;
+	}
+
+	seq_printf(m, "Current: %u\n", bpc);
+	res = 0;
+
+unlock:
+	drm_modeset_unlock(&crtc->mutex);
+	mutex_unlock(&dev->mode_config.mutex);
+
+	return res;
+}
+DEFINE_SHOW_ATTRIBUTE(amdgpu_current_bpc);
+
+/*
+ * Returns the current colorspace for the crtc.
+ * Example usage: cat /sys/kernel/debug/dri/0/crtc-0/amdgpu_current_colorspace
+ */
+static int amdgpu_current_colorspace_show(struct seq_file *m, void *data)
+{
+	struct drm_crtc *crtc = m->private;
+	struct drm_device *dev = crtc->dev;
+	struct dm_crtc_state *dm_crtc_state = NULL;
+	int res = -ENODEV;
+
+	mutex_lock(&dev->mode_config.mutex);
+	drm_modeset_lock(&crtc->mutex, NULL);
+	if (crtc->state == NULL)
+		goto unlock;
+
+	dm_crtc_state = to_dm_crtc_state(crtc->state);
+	if (dm_crtc_state->stream == NULL)
+		goto unlock;
+
+	switch (dm_crtc_state->stream->output_color_space) {
+	case COLOR_SPACE_SRGB:
+		seq_puts(m, "sRGB");
+		break;
+	case COLOR_SPACE_YCBCR601:
+	case COLOR_SPACE_YCBCR601_LIMITED:
+		seq_puts(m, "BT601_YCC");
+		break;
+	case COLOR_SPACE_YCBCR709:
+	case COLOR_SPACE_YCBCR709_LIMITED:
+		seq_puts(m, "BT709_YCC");
+		break;
+	case COLOR_SPACE_ADOBERGB:
+		seq_puts(m, "opRGB");
+		break;
+	case COLOR_SPACE_2020_RGB_FULLRANGE:
+		seq_puts(m, "BT2020_RGB");
+		break;
+	case COLOR_SPACE_2020_YCBCR_LIMITED:
+		seq_puts(m, "BT2020_YCC");
+		break;
+	default:
+		goto unlock;
+	}
+	res = 0;
+
+unlock:
+	drm_modeset_unlock(&crtc->mutex);
+	mutex_unlock(&dev->mode_config.mutex);
+
+	return res;
+}
+DEFINE_SHOW_ATTRIBUTE(amdgpu_current_colorspace);
+
+
+/*
+ * Example usage:
+ * Disable dsc passthrough, i.e.,: have dsc decoding at converver, not external RX
+ *   echo 1 /sys/kernel/debug/dri/0/DP-1/dsc_disable_passthrough
+ * Enable dsc passthrough, i.e.,: have dsc passthrough to external RX
+ *   echo 0 /sys/kernel/debug/dri/0/DP-1/dsc_disable_passthrough
+ */
+static ssize_t dp_dsc_passthrough_set(struct file *f, const char __user *buf,
+				 size_t size, loff_t *pos)
+{
+	struct amdgpu_dm_connector *aconnector = file_inode(f)->i_private;
+	char *wr_buf = NULL;
+	uint32_t wr_buf_size = 42;
+	int max_param_num = 1;
+	long param;
+	uint8_t param_nums = 0;
+
+	if (size == 0)
+		return -EINVAL;
+
+	wr_buf = kcalloc(wr_buf_size, sizeof(char), GFP_KERNEL);
+
+	if (!wr_buf) {
+		DRM_DEBUG_DRIVER("no memory to allocate write buffer\n");
+		return -ENOSPC;
+	}
+
+	if (parse_write_buffer_into_params(wr_buf, wr_buf_size,
+					   &param, buf,
+					   max_param_num,
+					   &param_nums)) {
+		kfree(wr_buf);
+		return -EINVAL;
+	}
+
+	aconnector->dsc_settings.dsc_force_disable_passthrough = param;
+
+	kfree(wr_buf);
+	return 0;
+}
+
+/*
+ * Returns the HDCP capability of the Display (1.4 for now).
+ *
+ * NOTE* Not all HDMI displays report their HDCP caps even when they are capable.
+ * Since its rare for a display to not be HDCP 1.4 capable, we set HDMI as always capable.
+ *
+ * Example usage: cat /sys/kernel/debug/dri/0/DP-1/hdcp_sink_capability
+ *		or cat /sys/kernel/debug/dri/0/HDMI-A-1/hdcp_sink_capability
+ */
+static int hdcp_sink_capability_show(struct seq_file *m, void *data)
+{
+	struct drm_connector *connector = m->private;
+	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector);
+	bool hdcp_cap, hdcp2_cap;
+
+	if (connector->status != connector_status_connected)
+		return -ENODEV;
+
+	seq_printf(m, "%s:%d HDCP version: ", connector->name, connector->base.id);
+
+	hdcp_cap = dc_link_is_hdcp14(aconnector->dc_link, aconnector->dc_sink->sink_signal);
+	hdcp2_cap = dc_link_is_hdcp22(aconnector->dc_link, aconnector->dc_sink->sink_signal);
+
+
+	if (hdcp_cap)
+		seq_printf(m, "%s ", "HDCP1.4");
+	if (hdcp2_cap)
+		seq_printf(m, "%s ", "HDCP2.2");
+
+	if (!hdcp_cap && !hdcp2_cap)
+		seq_printf(m, "%s ", "None");
+
+	seq_puts(m, "\n");
+
+	return 0;
+}
+
+/*
+ * Returns whether the connected display is internal and not hotpluggable.
+ * Example usage: cat /sys/kernel/debug/dri/0/DP-1/internal_display
+ */
+static int internal_display_show(struct seq_file *m, void *data)
+{
+	struct drm_connector *connector = m->private;
+	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector);
+	struct dc_link *link = aconnector->dc_link;
+
+	seq_printf(m, "Internal: %u\n", link->is_internal_display);
+
+	return 0;
+}
+
+/*
+ * Returns the number of segments used if ODM Combine mode is enabled.
+ * Example usage: cat /sys/kernel/debug/dri/0/DP-1/odm_combine_segments
+ */
+static int odm_combine_segments_show(struct seq_file *m, void *unused)
+{
+	struct drm_connector *connector = m->private;
+	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector);
+	struct dc_link *link = aconnector->dc_link;
+	struct pipe_ctx *pipe_ctx = NULL;
+	int i, segments = -EOPNOTSUPP;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe_ctx = &link->dc->current_state->res_ctx.pipe_ctx[i];
+		if (pipe_ctx->stream &&
+		    pipe_ctx->stream->link == link)
+			break;
+	}
+
+	if (connector->status != connector_status_connected)
+		return -ENODEV;
+
+	if (pipe_ctx != NULL && pipe_ctx->stream_res.tg->funcs->get_odm_combine_segments)
+		pipe_ctx->stream_res.tg->funcs->get_odm_combine_segments(pipe_ctx->stream_res.tg, &segments);
+
+	seq_printf(m, "%d\n", segments);
+	return 0;
+}
+
+/* function description
+ *
+ * generic SDP message access for testing
+ *
+ * debugfs sdp_message is located at /syskernel/debug/dri/0/DP-x
+ *
+ * SDP header
+ * Hb0 : Secondary-Data Packet ID
+ * Hb1 : Secondary-Data Packet type
+ * Hb2 : Secondary-Data-packet-specific header, Byte 0
+ * Hb3 : Secondary-Data-packet-specific header, Byte 1
+ *
+ * for using custom sdp message: input 4 bytes SDP header and 32 bytes raw data
+ */
+static ssize_t dp_sdp_message_debugfs_write(struct file *f, const char __user *buf,
+				 size_t size, loff_t *pos)
+{
+	int r;
+	uint8_t data[36] = {0};
+	struct amdgpu_dm_connector *connector = file_inode(f)->i_private;
+	struct dm_crtc_state *acrtc_state;
+	uint32_t write_size = 36;
+
+	if (connector->base.status != connector_status_connected)
+		return -ENODEV;
+
+	if (size == 0)
+		return 0;
+
+	acrtc_state = to_dm_crtc_state(connector->base.state->crtc->state);
+
+	r = copy_from_user(data, buf, write_size);
+
+	write_size -= r;
+
+	dc_stream_send_dp_sdp(acrtc_state->stream, data, write_size);
+
+	return write_size;
+}
+
+/* function: Read link's DSC & FEC capabilities
+ *
+ *
+ * Access it with the following command (you need to specify
+ * connector like DP-1):
+ *
+ *	cat /sys/kernel/debug/dri/0/DP-X/dp_dsc_fec_support
+ *
+ */
+static int dp_dsc_fec_support_show(struct seq_file *m, void *data)
+{
+	struct drm_connector *connector = m->private;
+	struct drm_modeset_acquire_ctx ctx;
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector);
+	int ret = 0;
+	bool try_again = false;
+	bool is_fec_supported = false;
+	bool is_dsc_supported = false;
+	struct dpcd_caps dpcd_caps;
+
+	drm_modeset_acquire_init(&ctx, DRM_MODESET_ACQUIRE_INTERRUPTIBLE);
+	do {
+		try_again = false;
+		ret = drm_modeset_lock(&dev->mode_config.connection_mutex, &ctx);
+		if (ret) {
+			if (ret == -EDEADLK) {
+				ret = drm_modeset_backoff(&ctx);
+				if (!ret) {
+					try_again = true;
+					continue;
+				}
+			}
+			break;
+		}
+		if (connector->status != connector_status_connected) {
+			ret = -ENODEV;
+			break;
+		}
+		dpcd_caps = aconnector->dc_link->dpcd_caps;
+		if (aconnector->mst_output_port) {
+			/* aconnector sets dsc_aux during get_modes call
+			 * if MST connector has it means it can either
+			 * enable DSC on the sink device or on MST branch
+			 * its connected to.
+			 */
+			if (aconnector->dsc_aux) {
+				is_fec_supported = true;
+				is_dsc_supported = true;
+			}
+		} else {
+			is_fec_supported = dpcd_caps.fec_cap.raw & 0x1;
+			is_dsc_supported = dpcd_caps.dsc_caps.dsc_basic_caps.raw[0] & 0x1;
+		}
+	} while (try_again);
+
+	drm_modeset_drop_locks(&ctx);
+	drm_modeset_acquire_fini(&ctx);
+
+	seq_printf(m, "FEC_Sink_Support: %s\n", str_yes_no(is_fec_supported));
+	seq_printf(m, "DSC_Sink_Support: %s\n", str_yes_no(is_dsc_supported));
+
+	return ret;
+}
+
+/* function: Trigger virtual HPD redetection on connector
+ *
+ * This function will perform link rediscovery, link disable
+ * and enable, and dm connector state update.
+ *
+ * Retrigger HPD on an existing connector by echoing 1 into
+ * its respectful "trigger_hotplug" debugfs entry:
+ *
+ *	echo 1 > /sys/kernel/debug/dri/0/DP-X/trigger_hotplug
+ *
+ * This function can perform HPD unplug:
+ *
+ *	echo 0 > /sys/kernel/debug/dri/0/DP-X/trigger_hotplug
+ *
+ */
+static ssize_t trigger_hotplug(struct file *f, const char __user *buf,
+							size_t size, loff_t *pos)
+{
+	struct amdgpu_dm_connector *aconnector = file_inode(f)->i_private;
+	struct drm_connector *connector = &aconnector->base;
+	struct dc_link *link = NULL;
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	enum dc_connection_type new_connection_type = dc_connection_none;
+	char *wr_buf = NULL;
+	uint32_t wr_buf_size = 42;
+	int max_param_num = 1;
+	long param[1] = {0};
+	uint8_t param_nums = 0;
+	bool ret = false;
+
+	if (!aconnector->dc_link)
+		return -EINVAL;
+
+	if (size == 0)
+		return -EINVAL;
+
+	wr_buf = kcalloc(wr_buf_size, sizeof(char), GFP_KERNEL);
+
+	if (!wr_buf) {
+		DRM_DEBUG_DRIVER("no memory to allocate write buffer\n");
+		return -ENOSPC;
+	}
+
+	if (parse_write_buffer_into_params(wr_buf, wr_buf_size,
+						(long *)param, buf,
+						max_param_num,
+						&param_nums)) {
+		kfree(wr_buf);
+		return -EINVAL;
+	}
+
+	kfree(wr_buf);
+
+	if (param_nums <= 0) {
+		DRM_DEBUG_DRIVER("user data not be read\n");
+		return -EINVAL;
+	}
+
+	mutex_lock(&aconnector->hpd_lock);
+
+	/* Don't support for mst end device*/
+	if (aconnector->mst_root) {
+		mutex_unlock(&aconnector->hpd_lock);
+		return -EINVAL;
+	}
+
+	if (param[0] == 1) {
+
+		if (!dc_link_detect_connection_type(aconnector->dc_link, &new_connection_type) &&
+			new_connection_type != dc_connection_none)
+			goto unlock;
+
+		mutex_lock(&adev->dm.dc_lock);
+		ret = dc_link_detect(aconnector->dc_link, DETECT_REASON_HPD);
+		mutex_unlock(&adev->dm.dc_lock);
+
+		if (!ret)
+			goto unlock;
+
+		amdgpu_dm_update_connector_after_detect(aconnector);
+
+		drm_modeset_lock_all(dev);
+		dm_restore_drm_connector_state(dev, connector);
+		drm_modeset_unlock_all(dev);
+
+		drm_kms_helper_connector_hotplug_event(connector);
+	} else if (param[0] == 0) {
+		if (!aconnector->dc_link)
+			goto unlock;
+
+		link = aconnector->dc_link;
+
+		if (link->local_sink) {
+			dc_sink_release(link->local_sink);
+			link->local_sink = NULL;
+		}
+
+		link->dpcd_sink_count = 0;
+		link->type = dc_connection_none;
+		link->dongle_max_pix_clk = 0;
+
+		amdgpu_dm_update_connector_after_detect(aconnector);
+
+		/* If the aconnector is the root node in mst topology */
+		if (aconnector->mst_mgr.mst_state == true)
+			dc_link_reset_cur_dp_mst_topology(link);
+
+		drm_modeset_lock_all(dev);
+		dm_restore_drm_connector_state(dev, connector);
+		drm_modeset_unlock_all(dev);
+
+		drm_kms_helper_connector_hotplug_event(connector);
+	}
+
+unlock:
+	mutex_unlock(&aconnector->hpd_lock);
+
+	return size;
+}
+
+/* function: read DSC status on the connector
+ *
+ * The read function: dp_dsc_clock_en_read
+ * returns current status of DSC clock on the connector.
+ * The return is a boolean flag: 1 or 0.
+ *
+ * Access it with the following command (you need to specify
+ * connector like DP-1):
+ *
+ *	cat /sys/kernel/debug/dri/0/DP-X/dsc_clock_en
+ *
+ * Expected output:
+ * 1 - means that DSC is currently enabled
+ * 0 - means that DSC is disabled
+ */
+static ssize_t dp_dsc_clock_en_read(struct file *f, char __user *buf,
+				    size_t size, loff_t *pos)
+{
+	char *rd_buf = NULL;
+	struct amdgpu_dm_connector *aconnector = file_inode(f)->i_private;
+	struct display_stream_compressor *dsc;
+	struct dcn_dsc_state dsc_state = {0};
+	const uint32_t rd_buf_size = 10;
+	struct pipe_ctx *pipe_ctx;
+	ssize_t result = 0;
+	int i, r, str_len = 10;
+
+	rd_buf = kcalloc(rd_buf_size, sizeof(char), GFP_KERNEL);
+
+	if (!rd_buf)
+		return -ENOMEM;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe_ctx = &aconnector->dc_link->dc->current_state->res_ctx.pipe_ctx[i];
+		if (pipe_ctx->stream &&
+		    pipe_ctx->stream->link == aconnector->dc_link &&
+		    pipe_ctx->stream->sink &&
+		    pipe_ctx->stream->sink == aconnector->dc_sink)
+			break;
+	}
+
+	dsc = pipe_ctx->stream_res.dsc;
+	if (dsc)
+		dsc->funcs->dsc_read_state(dsc, &dsc_state);
+
+	snprintf(rd_buf, str_len,
+		"%d\n",
+		dsc_state.dsc_clock_en);
+
+	while (size) {
+		if (*pos >= rd_buf_size)
+			break;
+
+		r = put_user(*(rd_buf + result), buf);
+		if (r) {
+			kfree(rd_buf);
+			return r; /* r = -EFAULT */
+		}
+
+		buf += 1;
+		size -= 1;
+		*pos += 1;
+		result += 1;
+	}
+
+	kfree(rd_buf);
+	return result;
+}
+
+/* function: write force DSC on the connector
+ *
+ * The write function: dp_dsc_clock_en_write
+ * enables to force DSC on the connector.
+ * User can write to either force enable or force disable DSC
+ * on the next modeset or set it to driver default
+ *
+ * Accepted inputs:
+ * 0 - default DSC enablement policy
+ * 1 - force enable DSC on the connector
+ * 2 - force disable DSC on the connector (might cause fail in atomic_check)
+ *
+ * Writing DSC settings is done with the following command:
+ * - To force enable DSC (you need to specify
+ * connector like DP-1):
+ *
+ *	echo 0x1 > /sys/kernel/debug/dri/0/DP-X/dsc_clock_en
+ *
+ * - To return to default state set the flag to zero and
+ * let driver deal with DSC automatically
+ * (you need to specify connector like DP-1):
+ *
+ *	echo 0x0 > /sys/kernel/debug/dri/0/DP-X/dsc_clock_en
+ *
+ */
+static ssize_t dp_dsc_clock_en_write(struct file *f, const char __user *buf,
+				     size_t size, loff_t *pos)
+{
+	struct amdgpu_dm_connector *aconnector = file_inode(f)->i_private;
+	struct drm_connector *connector = &aconnector->base;
+	struct drm_device *dev = connector->dev;
+	struct drm_crtc *crtc = NULL;
+	struct dm_crtc_state *dm_crtc_state = NULL;
+	struct pipe_ctx *pipe_ctx;
+	int i;
+	char *wr_buf = NULL;
+	uint32_t wr_buf_size = 42;
+	int max_param_num = 1;
+	long param[1] = {0};
+	uint8_t param_nums = 0;
+
+	if (size == 0)
+		return -EINVAL;
+
+	wr_buf = kcalloc(wr_buf_size, sizeof(char), GFP_KERNEL);
+
+	if (!wr_buf) {
+		DRM_DEBUG_DRIVER("no memory to allocate write buffer\n");
+		return -ENOSPC;
+	}
+
+	if (parse_write_buffer_into_params(wr_buf, wr_buf_size,
+					    (long *)param, buf,
+					    max_param_num,
+					    &param_nums)) {
+		kfree(wr_buf);
+		return -EINVAL;
+	}
+
+	if (param_nums <= 0) {
+		DRM_DEBUG_DRIVER("user data not be read\n");
+		kfree(wr_buf);
+		return -EINVAL;
+	}
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe_ctx = &aconnector->dc_link->dc->current_state->res_ctx.pipe_ctx[i];
+		if (pipe_ctx->stream &&
+		    pipe_ctx->stream->link == aconnector->dc_link &&
+		    pipe_ctx->stream->sink &&
+		    pipe_ctx->stream->sink == aconnector->dc_sink)
+			break;
+	}
+
+	if (!pipe_ctx->stream)
+		goto done;
+
+	// Get CRTC state
+	mutex_lock(&dev->mode_config.mutex);
+	drm_modeset_lock(&dev->mode_config.connection_mutex, NULL);
+
+	if (connector->state == NULL)
+		goto unlock;
+
+	crtc = connector->state->crtc;
+	if (crtc == NULL)
+		goto unlock;
+
+	drm_modeset_lock(&crtc->mutex, NULL);
+	if (crtc->state == NULL)
+		goto unlock;
+
+	dm_crtc_state = to_dm_crtc_state(crtc->state);
+	if (dm_crtc_state->stream == NULL)
+		goto unlock;
+
+	if (param[0] == 1)
+		aconnector->dsc_settings.dsc_force_enable = DSC_CLK_FORCE_ENABLE;
+	else if (param[0] == 2)
+		aconnector->dsc_settings.dsc_force_enable = DSC_CLK_FORCE_DISABLE;
+	else
+		aconnector->dsc_settings.dsc_force_enable = DSC_CLK_FORCE_DEFAULT;
+
+	dm_crtc_state->dsc_force_changed = true;
+
+unlock:
+	if (crtc)
+		drm_modeset_unlock(&crtc->mutex);
+	drm_modeset_unlock(&dev->mode_config.connection_mutex);
+	mutex_unlock(&dev->mode_config.mutex);
+
+done:
+	kfree(wr_buf);
+	return size;
+}
+
+/* function: read DSC slice width parameter on the connector
+ *
+ * The read function: dp_dsc_slice_width_read
+ * returns dsc slice width used in the current configuration
+ * The return is an integer: 0 or other positive number
+ *
+ * Access the status with the following command:
+ *
+ *	cat /sys/kernel/debug/dri/0/DP-X/dsc_slice_width
+ *
+ * 0 - means that DSC is disabled
+ *
+ * Any other number more than zero represents the
+ * slice width currently used by DSC in pixels
+ *
+ */
+static ssize_t dp_dsc_slice_width_read(struct file *f, char __user *buf,
+				    size_t size, loff_t *pos)
+{
+	char *rd_buf = NULL;
+	struct amdgpu_dm_connector *aconnector = file_inode(f)->i_private;
+	struct display_stream_compressor *dsc;
+	struct dcn_dsc_state dsc_state = {0};
+	const uint32_t rd_buf_size = 100;
+	struct pipe_ctx *pipe_ctx;
+	ssize_t result = 0;
+	int i, r, str_len = 30;
+
+	rd_buf = kcalloc(rd_buf_size, sizeof(char), GFP_KERNEL);
+
+	if (!rd_buf)
+		return -ENOMEM;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe_ctx = &aconnector->dc_link->dc->current_state->res_ctx.pipe_ctx[i];
+		if (pipe_ctx->stream &&
+		    pipe_ctx->stream->link == aconnector->dc_link &&
+		    pipe_ctx->stream->sink &&
+		    pipe_ctx->stream->sink == aconnector->dc_sink)
+			break;
+	}
+
+	dsc = pipe_ctx->stream_res.dsc;
+	if (dsc)
+		dsc->funcs->dsc_read_state(dsc, &dsc_state);
+
+	snprintf(rd_buf, str_len,
+		"%d\n",
+		dsc_state.dsc_slice_width);
+
+	while (size) {
+		if (*pos >= rd_buf_size)
+			break;
+
+		r = put_user(*(rd_buf + result), buf);
+		if (r) {
+			kfree(rd_buf);
+			return r; /* r = -EFAULT */
+		}
+
+		buf += 1;
+		size -= 1;
+		*pos += 1;
+		result += 1;
+	}
+
+	kfree(rd_buf);
+	return result;
+}
+
+/* function: write DSC slice width parameter
+ *
+ * The write function: dp_dsc_slice_width_write
+ * overwrites automatically generated DSC configuration
+ * of slice width.
+ *
+ * The user has to write the slice width divisible by the
+ * picture width.
+ *
+ * Also the user has to write width in hexidecimal
+ * rather than in decimal.
+ *
+ * Writing DSC settings is done with the following command:
+ * - To force overwrite slice width: (example sets to 1920 pixels)
+ *
+ *	echo 0x780 > /sys/kernel/debug/dri/0/DP-X/dsc_slice_width
+ *
+ *  - To stop overwriting and let driver find the optimal size,
+ * set the width to zero:
+ *
+ *	echo 0x0 > /sys/kernel/debug/dri/0/DP-X/dsc_slice_width
+ *
+ */
+static ssize_t dp_dsc_slice_width_write(struct file *f, const char __user *buf,
+				     size_t size, loff_t *pos)
+{
+	struct amdgpu_dm_connector *aconnector = file_inode(f)->i_private;
+	struct pipe_ctx *pipe_ctx;
+	struct drm_connector *connector = &aconnector->base;
+	struct drm_device *dev = connector->dev;
+	struct drm_crtc *crtc = NULL;
+	struct dm_crtc_state *dm_crtc_state = NULL;
+	int i;
+	char *wr_buf = NULL;
+	uint32_t wr_buf_size = 42;
+	int max_param_num = 1;
+	long param[1] = {0};
+	uint8_t param_nums = 0;
+
+	if (size == 0)
+		return -EINVAL;
+
+	wr_buf = kcalloc(wr_buf_size, sizeof(char), GFP_KERNEL);
+
+	if (!wr_buf) {
+		DRM_DEBUG_DRIVER("no memory to allocate write buffer\n");
+		return -ENOSPC;
+	}
+
+	if (parse_write_buffer_into_params(wr_buf, wr_buf_size,
+					    (long *)param, buf,
+					    max_param_num,
+					    &param_nums)) {
+		kfree(wr_buf);
+		return -EINVAL;
+	}
+
+	if (param_nums <= 0) {
+		DRM_DEBUG_DRIVER("user data not be read\n");
+		kfree(wr_buf);
+		return -EINVAL;
+	}
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe_ctx = &aconnector->dc_link->dc->current_state->res_ctx.pipe_ctx[i];
+		if (pipe_ctx->stream &&
+		    pipe_ctx->stream->link == aconnector->dc_link &&
+		    pipe_ctx->stream->sink &&
+		    pipe_ctx->stream->sink == aconnector->dc_sink)
+			break;
+	}
+
+	if (!pipe_ctx->stream)
+		goto done;
+
+	// Safely get CRTC state
+	mutex_lock(&dev->mode_config.mutex);
+	drm_modeset_lock(&dev->mode_config.connection_mutex, NULL);
+
+	if (connector->state == NULL)
+		goto unlock;
+
+	crtc = connector->state->crtc;
+	if (crtc == NULL)
+		goto unlock;
+
+	drm_modeset_lock(&crtc->mutex, NULL);
+	if (crtc->state == NULL)
+		goto unlock;
+
+	dm_crtc_state = to_dm_crtc_state(crtc->state);
+	if (dm_crtc_state->stream == NULL)
+		goto unlock;
+
+	if (param[0] > 0)
+		aconnector->dsc_settings.dsc_num_slices_h = DIV_ROUND_UP(
+					pipe_ctx->stream->timing.h_addressable,
+					param[0]);
+	else
+		aconnector->dsc_settings.dsc_num_slices_h = 0;
+
+	dm_crtc_state->dsc_force_changed = true;
+
+unlock:
+	if (crtc)
+		drm_modeset_unlock(&crtc->mutex);
+	drm_modeset_unlock(&dev->mode_config.connection_mutex);
+	mutex_unlock(&dev->mode_config.mutex);
+
+done:
+	kfree(wr_buf);
+	return size;
+}
+
+/* function: read DSC slice height parameter on the connector
+ *
+ * The read function: dp_dsc_slice_height_read
+ * returns dsc slice height used in the current configuration
+ * The return is an integer: 0 or other positive number
+ *
+ * Access the status with the following command:
+ *
+ *	cat /sys/kernel/debug/dri/0/DP-X/dsc_slice_height
+ *
+ * 0 - means that DSC is disabled
+ *
+ * Any other number more than zero represents the
+ * slice height currently used by DSC in pixels
+ *
+ */
+static ssize_t dp_dsc_slice_height_read(struct file *f, char __user *buf,
+				    size_t size, loff_t *pos)
+{
+	char *rd_buf = NULL;
+	struct amdgpu_dm_connector *aconnector = file_inode(f)->i_private;
+	struct display_stream_compressor *dsc;
+	struct dcn_dsc_state dsc_state = {0};
+	const uint32_t rd_buf_size = 100;
+	struct pipe_ctx *pipe_ctx;
+	ssize_t result = 0;
+	int i, r, str_len = 30;
+
+	rd_buf = kcalloc(rd_buf_size, sizeof(char), GFP_KERNEL);
+
+	if (!rd_buf)
+		return -ENOMEM;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe_ctx = &aconnector->dc_link->dc->current_state->res_ctx.pipe_ctx[i];
+		if (pipe_ctx->stream &&
+		    pipe_ctx->stream->link == aconnector->dc_link &&
+		    pipe_ctx->stream->sink &&
+		    pipe_ctx->stream->sink == aconnector->dc_sink)
+			break;
+	}
+
+	dsc = pipe_ctx->stream_res.dsc;
+	if (dsc)
+		dsc->funcs->dsc_read_state(dsc, &dsc_state);
+
+	snprintf(rd_buf, str_len,
+		"%d\n",
+		dsc_state.dsc_slice_height);
+
+	while (size) {
+		if (*pos >= rd_buf_size)
+			break;
+
+		r = put_user(*(rd_buf + result), buf);
+		if (r) {
+			kfree(rd_buf);
+			return r; /* r = -EFAULT */
+		}
+
+		buf += 1;
+		size -= 1;
+		*pos += 1;
+		result += 1;
+	}
+
+	kfree(rd_buf);
+	return result;
+}
+
+/* function: write DSC slice height parameter
+ *
+ * The write function: dp_dsc_slice_height_write
+ * overwrites automatically generated DSC configuration
+ * of slice height.
+ *
+ * The user has to write the slice height divisible by the
+ * picture height.
+ *
+ * Also the user has to write height in hexidecimal
+ * rather than in decimal.
+ *
+ * Writing DSC settings is done with the following command:
+ * - To force overwrite slice height (example sets to 128 pixels):
+ *
+ *	echo 0x80 > /sys/kernel/debug/dri/0/DP-X/dsc_slice_height
+ *
+ *  - To stop overwriting and let driver find the optimal size,
+ * set the height to zero:
+ *
+ *	echo 0x0 > /sys/kernel/debug/dri/0/DP-X/dsc_slice_height
+ *
+ */
+static ssize_t dp_dsc_slice_height_write(struct file *f, const char __user *buf,
+				     size_t size, loff_t *pos)
+{
+	struct amdgpu_dm_connector *aconnector = file_inode(f)->i_private;
+	struct drm_connector *connector = &aconnector->base;
+	struct drm_device *dev = connector->dev;
+	struct drm_crtc *crtc = NULL;
+	struct dm_crtc_state *dm_crtc_state = NULL;
+	struct pipe_ctx *pipe_ctx;
+	int i;
+	char *wr_buf = NULL;
+	uint32_t wr_buf_size = 42;
+	int max_param_num = 1;
+	uint8_t param_nums = 0;
+	long param[1] = {0};
+
+	if (size == 0)
+		return -EINVAL;
+
+	wr_buf = kcalloc(wr_buf_size, sizeof(char), GFP_KERNEL);
+
+	if (!wr_buf) {
+		DRM_DEBUG_DRIVER("no memory to allocate write buffer\n");
+		return -ENOSPC;
+	}
+
+	if (parse_write_buffer_into_params(wr_buf, wr_buf_size,
+					    (long *)param, buf,
+					    max_param_num,
+					    &param_nums)) {
+		kfree(wr_buf);
+		return -EINVAL;
+	}
+
+	if (param_nums <= 0) {
+		DRM_DEBUG_DRIVER("user data not be read\n");
+		kfree(wr_buf);
+		return -EINVAL;
+	}
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe_ctx = &aconnector->dc_link->dc->current_state->res_ctx.pipe_ctx[i];
+		if (pipe_ctx->stream &&
+		    pipe_ctx->stream->link == aconnector->dc_link &&
+		    pipe_ctx->stream->sink &&
+		    pipe_ctx->stream->sink == aconnector->dc_sink)
+			break;
+	}
+
+	if (!pipe_ctx->stream)
+		goto done;
+
+	// Get CRTC state
+	mutex_lock(&dev->mode_config.mutex);
+	drm_modeset_lock(&dev->mode_config.connection_mutex, NULL);
+
+	if (connector->state == NULL)
+		goto unlock;
+
+	crtc = connector->state->crtc;
+	if (crtc == NULL)
+		goto unlock;
+
+	drm_modeset_lock(&crtc->mutex, NULL);
+	if (crtc->state == NULL)
+		goto unlock;
+
+	dm_crtc_state = to_dm_crtc_state(crtc->state);
+	if (dm_crtc_state->stream == NULL)
+		goto unlock;
+
+	if (param[0] > 0)
+		aconnector->dsc_settings.dsc_num_slices_v = DIV_ROUND_UP(
+					pipe_ctx->stream->timing.v_addressable,
+					param[0]);
+	else
+		aconnector->dsc_settings.dsc_num_slices_v = 0;
+
+	dm_crtc_state->dsc_force_changed = true;
+
+unlock:
+	if (crtc)
+		drm_modeset_unlock(&crtc->mutex);
+	drm_modeset_unlock(&dev->mode_config.connection_mutex);
+	mutex_unlock(&dev->mode_config.mutex);
+
+done:
+	kfree(wr_buf);
+	return size;
+}
+
+/* function: read DSC target rate on the connector in bits per pixel
+ *
+ * The read function: dp_dsc_bits_per_pixel_read
+ * returns target rate of compression in bits per pixel
+ * The return is an integer: 0 or other positive integer
+ *
+ * Access it with the following command:
+ *
+ *	cat /sys/kernel/debug/dri/0/DP-X/dsc_bits_per_pixel
+ *
+ *  0 - means that DSC is disabled
+ */
+static ssize_t dp_dsc_bits_per_pixel_read(struct file *f, char __user *buf,
+				    size_t size, loff_t *pos)
+{
+	char *rd_buf = NULL;
+	struct amdgpu_dm_connector *aconnector = file_inode(f)->i_private;
+	struct display_stream_compressor *dsc;
+	struct dcn_dsc_state dsc_state = {0};
+	const uint32_t rd_buf_size = 100;
+	struct pipe_ctx *pipe_ctx;
+	ssize_t result = 0;
+	int i, r, str_len = 30;
+
+	rd_buf = kcalloc(rd_buf_size, sizeof(char), GFP_KERNEL);
+
+	if (!rd_buf)
+		return -ENOMEM;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe_ctx = &aconnector->dc_link->dc->current_state->res_ctx.pipe_ctx[i];
+		if (pipe_ctx->stream &&
+		    pipe_ctx->stream->link == aconnector->dc_link &&
+		    pipe_ctx->stream->sink &&
+		    pipe_ctx->stream->sink == aconnector->dc_sink)
+			break;
+	}
+
+	dsc = pipe_ctx->stream_res.dsc;
+	if (dsc)
+		dsc->funcs->dsc_read_state(dsc, &dsc_state);
+
+	snprintf(rd_buf, str_len,
+		"%d\n",
+		dsc_state.dsc_bits_per_pixel);
+
+	while (size) {
+		if (*pos >= rd_buf_size)
+			break;
+
+		r = put_user(*(rd_buf + result), buf);
+		if (r) {
+			kfree(rd_buf);
+			return r; /* r = -EFAULT */
+		}
+
+		buf += 1;
+		size -= 1;
+		*pos += 1;
+		result += 1;
+	}
+
+	kfree(rd_buf);
+	return result;
+}
+
+/* function: write DSC target rate in bits per pixel
+ *
+ * The write function: dp_dsc_bits_per_pixel_write
+ * overwrites automatically generated DSC configuration
+ * of DSC target bit rate.
+ *
+ * Also the user has to write bpp in hexidecimal
+ * rather than in decimal.
+ *
+ * Writing DSC settings is done with the following command:
+ * - To force overwrite rate (example sets to 256 bpp x 1/16):
+ *
+ *	echo 0x100 > /sys/kernel/debug/dri/0/DP-X/dsc_bits_per_pixel
+ *
+ *  - To stop overwriting and let driver find the optimal rate,
+ * set the rate to zero:
+ *
+ *	echo 0x0 > /sys/kernel/debug/dri/0/DP-X/dsc_bits_per_pixel
+ *
+ */
+static ssize_t dp_dsc_bits_per_pixel_write(struct file *f, const char __user *buf,
+				     size_t size, loff_t *pos)
+{
+	struct amdgpu_dm_connector *aconnector = file_inode(f)->i_private;
+	struct drm_connector *connector = &aconnector->base;
+	struct drm_device *dev = connector->dev;
+	struct drm_crtc *crtc = NULL;
+	struct dm_crtc_state *dm_crtc_state = NULL;
+	struct pipe_ctx *pipe_ctx;
+	int i;
+	char *wr_buf = NULL;
+	uint32_t wr_buf_size = 42;
+	int max_param_num = 1;
+	uint8_t param_nums = 0;
+	long param[1] = {0};
+
+	if (size == 0)
+		return -EINVAL;
+
+	wr_buf = kcalloc(wr_buf_size, sizeof(char), GFP_KERNEL);
+
+	if (!wr_buf) {
+		DRM_DEBUG_DRIVER("no memory to allocate write buffer\n");
+		return -ENOSPC;
+	}
+
+	if (parse_write_buffer_into_params(wr_buf, wr_buf_size,
+					    (long *)param, buf,
+					    max_param_num,
+					    &param_nums)) {
+		kfree(wr_buf);
+		return -EINVAL;
+	}
+
+	if (param_nums <= 0) {
+		DRM_DEBUG_DRIVER("user data not be read\n");
+		kfree(wr_buf);
+		return -EINVAL;
+	}
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe_ctx = &aconnector->dc_link->dc->current_state->res_ctx.pipe_ctx[i];
+		if (pipe_ctx->stream &&
+		    pipe_ctx->stream->link == aconnector->dc_link &&
+		    pipe_ctx->stream->sink &&
+		    pipe_ctx->stream->sink == aconnector->dc_sink)
+			break;
+	}
+
+	if (!pipe_ctx->stream)
+		goto done;
+
+	// Get CRTC state
+	mutex_lock(&dev->mode_config.mutex);
+	drm_modeset_lock(&dev->mode_config.connection_mutex, NULL);
+
+	if (connector->state == NULL)
+		goto unlock;
+
+	crtc = connector->state->crtc;
+	if (crtc == NULL)
+		goto unlock;
+
+	drm_modeset_lock(&crtc->mutex, NULL);
+	if (crtc->state == NULL)
+		goto unlock;
+
+	dm_crtc_state = to_dm_crtc_state(crtc->state);
+	if (dm_crtc_state->stream == NULL)
+		goto unlock;
+
+	aconnector->dsc_settings.dsc_bits_per_pixel = param[0];
+
+	dm_crtc_state->dsc_force_changed = true;
+
+unlock:
+	if (crtc)
+		drm_modeset_unlock(&crtc->mutex);
+	drm_modeset_unlock(&dev->mode_config.connection_mutex);
+	mutex_unlock(&dev->mode_config.mutex);
+
+done:
+	kfree(wr_buf);
+	return size;
+}
+
+/* function: read DSC picture width parameter on the connector
+ *
+ * The read function: dp_dsc_pic_width_read
+ * returns dsc picture width used in the current configuration
+ * It is the same as h_addressable of the current
+ * display's timing
+ * The return is an integer: 0 or other positive integer
+ * If 0 then DSC is disabled.
+ *
+ * Access it with the following command:
+ *
+ *	cat /sys/kernel/debug/dri/0/DP-X/dsc_pic_width
+ *
+ * 0 - means that DSC is disabled
+ */
+static ssize_t dp_dsc_pic_width_read(struct file *f, char __user *buf,
+				    size_t size, loff_t *pos)
+{
+	char *rd_buf = NULL;
+	struct amdgpu_dm_connector *aconnector = file_inode(f)->i_private;
+	struct display_stream_compressor *dsc;
+	struct dcn_dsc_state dsc_state = {0};
+	const uint32_t rd_buf_size = 100;
+	struct pipe_ctx *pipe_ctx;
+	ssize_t result = 0;
+	int i, r, str_len = 30;
+
+	rd_buf = kcalloc(rd_buf_size, sizeof(char), GFP_KERNEL);
+
+	if (!rd_buf)
+		return -ENOMEM;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe_ctx = &aconnector->dc_link->dc->current_state->res_ctx.pipe_ctx[i];
+		if (pipe_ctx->stream &&
+		    pipe_ctx->stream->link == aconnector->dc_link &&
+		    pipe_ctx->stream->sink &&
+		    pipe_ctx->stream->sink == aconnector->dc_sink)
+			break;
+	}
+
+	dsc = pipe_ctx->stream_res.dsc;
+	if (dsc)
+		dsc->funcs->dsc_read_state(dsc, &dsc_state);
+
+	snprintf(rd_buf, str_len,
+		"%d\n",
+		dsc_state.dsc_pic_width);
+
+	while (size) {
+		if (*pos >= rd_buf_size)
+			break;
+
+		r = put_user(*(rd_buf + result), buf);
+		if (r) {
+			kfree(rd_buf);
+			return r; /* r = -EFAULT */
+		}
+
+		buf += 1;
+		size -= 1;
+		*pos += 1;
+		result += 1;
+	}
+
+	kfree(rd_buf);
+	return result;
+}
+
+static ssize_t dp_dsc_pic_height_read(struct file *f, char __user *buf,
+				    size_t size, loff_t *pos)
+{
+	char *rd_buf = NULL;
+	struct amdgpu_dm_connector *aconnector = file_inode(f)->i_private;
+	struct display_stream_compressor *dsc;
+	struct dcn_dsc_state dsc_state = {0};
+	const uint32_t rd_buf_size = 100;
+	struct pipe_ctx *pipe_ctx;
+	ssize_t result = 0;
+	int i, r, str_len = 30;
+
+	rd_buf = kcalloc(rd_buf_size, sizeof(char), GFP_KERNEL);
+
+	if (!rd_buf)
+		return -ENOMEM;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe_ctx = &aconnector->dc_link->dc->current_state->res_ctx.pipe_ctx[i];
+		if (pipe_ctx->stream &&
+		    pipe_ctx->stream->link == aconnector->dc_link &&
+		    pipe_ctx->stream->sink &&
+		    pipe_ctx->stream->sink == aconnector->dc_sink)
+			break;
+	}
+
+	dsc = pipe_ctx->stream_res.dsc;
+	if (dsc)
+		dsc->funcs->dsc_read_state(dsc, &dsc_state);
+
+	snprintf(rd_buf, str_len,
+		"%d\n",
+		dsc_state.dsc_pic_height);
+
+	while (size) {
+		if (*pos >= rd_buf_size)
+			break;
+
+		r = put_user(*(rd_buf + result), buf);
+		if (r) {
+			kfree(rd_buf);
+			return r; /* r = -EFAULT */
+		}
+
+		buf += 1;
+		size -= 1;
+		*pos += 1;
+		result += 1;
+	}
+
+	kfree(rd_buf);
+	return result;
+}
+
+/* function: read DSC chunk size parameter on the connector
+ *
+ * The read function: dp_dsc_chunk_size_read
+ * returns dsc chunk size set in the current configuration
+ * The value is calculated automatically by DSC code
+ * and depends on slice parameters and bpp target rate
+ * The return is an integer: 0 or other positive integer
+ * If 0 then DSC is disabled.
+ *
+ * Access it with the following command:
+ *
+ *	cat /sys/kernel/debug/dri/0/DP-X/dsc_chunk_size
+ *
+ * 0 - means that DSC is disabled
+ */
+static ssize_t dp_dsc_chunk_size_read(struct file *f, char __user *buf,
+				    size_t size, loff_t *pos)
+{
+	char *rd_buf = NULL;
+	struct amdgpu_dm_connector *aconnector = file_inode(f)->i_private;
+	struct display_stream_compressor *dsc;
+	struct dcn_dsc_state dsc_state = {0};
+	const uint32_t rd_buf_size = 100;
+	struct pipe_ctx *pipe_ctx;
+	ssize_t result = 0;
+	int i, r, str_len = 30;
+
+	rd_buf = kcalloc(rd_buf_size, sizeof(char), GFP_KERNEL);
+
+	if (!rd_buf)
+		return -ENOMEM;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe_ctx = &aconnector->dc_link->dc->current_state->res_ctx.pipe_ctx[i];
+		if (pipe_ctx->stream &&
+		    pipe_ctx->stream->link == aconnector->dc_link &&
+		    pipe_ctx->stream->sink &&
+		    pipe_ctx->stream->sink == aconnector->dc_sink)
+			break;
+	}
+
+	dsc = pipe_ctx->stream_res.dsc;
+	if (dsc)
+		dsc->funcs->dsc_read_state(dsc, &dsc_state);
+
+	snprintf(rd_buf, str_len,
+		"%d\n",
+		dsc_state.dsc_chunk_size);
+
+	while (size) {
+		if (*pos >= rd_buf_size)
+			break;
+
+		r = put_user(*(rd_buf + result), buf);
+		if (r) {
+			kfree(rd_buf);
+			return r; /* r = -EFAULT */
+		}
+
+		buf += 1;
+		size -= 1;
+		*pos += 1;
+		result += 1;
+	}
+
+	kfree(rd_buf);
+	return result;
+}
+
+/* function: read DSC slice bpg offset on the connector
+ *
+ * The read function: dp_dsc_slice_bpg_offset_read
+ * returns dsc bpg slice offset set in the current configuration
+ * The value is calculated automatically by DSC code
+ * and depends on slice parameters and bpp target rate
+ * The return is an integer: 0 or other positive integer
+ * If 0 then DSC is disabled.
+ *
+ * Access it with the following command:
+ *
+ *	cat /sys/kernel/debug/dri/0/DP-X/dsc_slice_bpg_offset
+ *
+ * 0 - means that DSC is disabled
+ */
+static ssize_t dp_dsc_slice_bpg_offset_read(struct file *f, char __user *buf,
+				    size_t size, loff_t *pos)
+{
+	char *rd_buf = NULL;
+	struct amdgpu_dm_connector *aconnector = file_inode(f)->i_private;
+	struct display_stream_compressor *dsc;
+	struct dcn_dsc_state dsc_state = {0};
+	const uint32_t rd_buf_size = 100;
+	struct pipe_ctx *pipe_ctx;
+	ssize_t result = 0;
+	int i, r, str_len = 30;
+
+	rd_buf = kcalloc(rd_buf_size, sizeof(char), GFP_KERNEL);
+
+	if (!rd_buf)
+		return -ENOMEM;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe_ctx = &aconnector->dc_link->dc->current_state->res_ctx.pipe_ctx[i];
+		if (pipe_ctx->stream &&
+		    pipe_ctx->stream->link == aconnector->dc_link &&
+		    pipe_ctx->stream->sink &&
+		    pipe_ctx->stream->sink == aconnector->dc_sink)
+			break;
+	}
+
+	dsc = pipe_ctx->stream_res.dsc;
+	if (dsc)
+		dsc->funcs->dsc_read_state(dsc, &dsc_state);
+
+	snprintf(rd_buf, str_len,
+		"%d\n",
+		dsc_state.dsc_slice_bpg_offset);
+
+	while (size) {
+		if (*pos >= rd_buf_size)
+			break;
+
+		r = put_user(*(rd_buf + result), buf);
+		if (r) {
+			kfree(rd_buf);
+			return r; /* r = -EFAULT */
+		}
+
+		buf += 1;
+		size -= 1;
+		*pos += 1;
+		result += 1;
+	}
+
+	kfree(rd_buf);
+	return result;
+}
+
+
+/*
+ * function description: Read max_requested_bpc property from the connector
+ *
+ * Access it with the following command:
+ *
+ *	cat /sys/kernel/debug/dri/0/DP-X/max_bpc
+ *
+ */
+static ssize_t dp_max_bpc_read(struct file *f, char __user *buf,
+		size_t size, loff_t *pos)
+{
+	struct amdgpu_dm_connector *aconnector = file_inode(f)->i_private;
+	struct drm_connector *connector = &aconnector->base;
+	struct drm_device *dev = connector->dev;
+	struct dm_connector_state *state;
+	ssize_t result = 0;
+	char *rd_buf = NULL;
+	char *rd_buf_ptr = NULL;
+	const uint32_t rd_buf_size = 10;
+	int r;
+
+	rd_buf = kcalloc(rd_buf_size, sizeof(char), GFP_KERNEL);
+
+	if (!rd_buf)
+		return -ENOMEM;
+
+	mutex_lock(&dev->mode_config.mutex);
+	drm_modeset_lock(&dev->mode_config.connection_mutex, NULL);
+
+	if (connector->state == NULL)
+		goto unlock;
+
+	state = to_dm_connector_state(connector->state);
+
+	rd_buf_ptr = rd_buf;
+	snprintf(rd_buf_ptr, rd_buf_size,
+		"%u\n",
+		state->base.max_requested_bpc);
+
+	while (size) {
+		if (*pos >= rd_buf_size)
+			break;
+
+		r = put_user(*(rd_buf + result), buf);
+		if (r) {
+			result = r; /* r = -EFAULT */
+			goto unlock;
+		}
+		buf += 1;
+		size -= 1;
+		*pos += 1;
+		result += 1;
+	}
+unlock:
+	drm_modeset_unlock(&dev->mode_config.connection_mutex);
+	mutex_unlock(&dev->mode_config.mutex);
+	kfree(rd_buf);
+	return result;
+}
+
+
+/*
+ * function description: Set max_requested_bpc property on the connector
+ *
+ * This function will not force the input BPC on connector, it will only
+ * change the max value. This is equivalent to setting max_bpc through
+ * xrandr.
+ *
+ * The BPC value written must be >= 6 and <= 16. Values outside of this
+ * range will result in errors.
+ *
+ * BPC values:
+ *	0x6 - 6 BPC
+ *	0x8 - 8 BPC
+ *	0xa - 10 BPC
+ *	0xc - 12 BPC
+ *	0x10 - 16 BPC
+ *
+ * Write the max_bpc in the following way:
+ *
+ * echo 0x6 > /sys/kernel/debug/dri/0/DP-X/max_bpc
+ *
+ */
+static ssize_t dp_max_bpc_write(struct file *f, const char __user *buf,
+				     size_t size, loff_t *pos)
+{
+	struct amdgpu_dm_connector *aconnector = file_inode(f)->i_private;
+	struct drm_connector *connector = &aconnector->base;
+	struct dm_connector_state *state;
+	struct drm_device *dev = connector->dev;
+	char *wr_buf = NULL;
+	uint32_t wr_buf_size = 42;
+	int max_param_num = 1;
+	long param[1] = {0};
+	uint8_t param_nums = 0;
+
+	if (size == 0)
+		return -EINVAL;
+
+	wr_buf = kcalloc(wr_buf_size, sizeof(char), GFP_KERNEL);
+
+	if (!wr_buf) {
+		DRM_DEBUG_DRIVER("no memory to allocate write buffer\n");
+		return -ENOSPC;
+	}
+
+	if (parse_write_buffer_into_params(wr_buf, wr_buf_size,
+					   (long *)param, buf,
+					   max_param_num,
+					   &param_nums)) {
+		kfree(wr_buf);
+		return -EINVAL;
+	}
+
+	if (param_nums <= 0) {
+		DRM_DEBUG_DRIVER("user data not be read\n");
+		kfree(wr_buf);
+		return -EINVAL;
+	}
+
+	if (param[0] < 6 || param[0] > 16) {
+		DRM_DEBUG_DRIVER("bad max_bpc value\n");
+		kfree(wr_buf);
+		return -EINVAL;
+	}
+
+	mutex_lock(&dev->mode_config.mutex);
+	drm_modeset_lock(&dev->mode_config.connection_mutex, NULL);
+
+	if (connector->state == NULL)
+		goto unlock;
+
+	state = to_dm_connector_state(connector->state);
+	state->base.max_requested_bpc = param[0];
+unlock:
+	drm_modeset_unlock(&dev->mode_config.connection_mutex);
+	mutex_unlock(&dev->mode_config.mutex);
+
+	kfree(wr_buf);
+	return size;
+}
+
+/*
+ * IPS status.  Read only.
+ *
+ * Example usage: cat /sys/kernel/debug/dri/0/amdgpu_dm_ips_status
+ */
+static int ips_status_show(struct seq_file *m, void *unused)
+{
+	struct amdgpu_device *adev = m->private;
+	struct dc *dc = adev->dm.dc;
+	struct dc_dmub_srv *dc_dmub_srv;
+
+	seq_printf(m, "IPS config: %d\n", dc->config.disable_ips);
+	seq_printf(m, "Idle optimization: %d\n", dc->idle_optimizations_allowed);
+
+	if (adev->dm.idle_workqueue) {
+		seq_printf(m, "Idle workqueue - enabled: %d\n", adev->dm.idle_workqueue->enable);
+		seq_printf(m, "Idle workqueue - running: %d\n", adev->dm.idle_workqueue->running);
+	}
+
+	dc_dmub_srv = dc->ctx->dmub_srv;
+	if (dc_dmub_srv && dc_dmub_srv->dmub) {
+		uint32_t rcg_count, ips1_count, ips2_count;
+		volatile const struct dmub_shared_state_ips_fw *ips_fw =
+			&dc_dmub_srv->dmub->shared_state[DMUB_SHARED_SHARE_FEATURE__IPS_FW].data.ips_fw;
+		rcg_count = ips_fw->rcg_entry_count;
+		ips1_count = ips_fw->ips1_entry_count;
+		ips2_count = ips_fw->ips2_entry_count;
+		seq_printf(m, "entry counts: rcg=%u ips1=%u ips2=%u\n",
+			   rcg_count,
+			   ips1_count,
+			   ips2_count);
+		rcg_count = ips_fw->rcg_exit_count;
+		ips1_count = ips_fw->ips1_exit_count;
+		ips2_count = ips_fw->ips2_exit_count;
+		seq_printf(m, "exit counts: rcg=%u ips1=%u ips2=%u",
+			   rcg_count,
+			   ips1_count,
+			   ips2_count);
+		seq_puts(m, "\n");
+	}
+	return 0;
+}
+
+/*
+ * Backlight at this moment.  Read only.
+ * As written to display, taking ABM and backlight lut into account.
+ * Ranges from 0x0 to 0x10000 (= 100% PWM)
+ *
+ * Example usage: cat /sys/kernel/debug/dri/0/eDP-1/current_backlight
+ */
+static int current_backlight_show(struct seq_file *m, void *unused)
+{
+	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(m->private);
+	struct dc_link *link = aconnector->dc_link;
+	unsigned int backlight;
+
+	backlight = dc_link_get_backlight_level(link);
+	seq_printf(m, "0x%x\n", backlight);
+
+	return 0;
+}
+
+/*
+ * Backlight value that is being approached.  Read only.
+ * As written to display, taking ABM and backlight lut into account.
+ * Ranges from 0x0 to 0x10000 (= 100% PWM)
+ *
+ * Example usage: cat /sys/kernel/debug/dri/0/eDP-1/target_backlight
+ */
+static int target_backlight_show(struct seq_file *m, void *unused)
+{
+	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(m->private);
+	struct dc_link *link = aconnector->dc_link;
+	unsigned int backlight;
+
+	backlight = dc_link_get_target_backlight_pwm(link);
+	seq_printf(m, "0x%x\n", backlight);
+
+	return 0;
+}
+
+/*
+ * function description: Determine if the connector is mst connector
+ *
+ * This function helps to determine whether a connector is a mst connector.
+ * - "root" stands for the root connector of the topology
+ * - "branch" stands for branch device of the topology
+ * - "end" stands for leaf node connector of the topology
+ * - "no" stands for the connector is not a device of a mst topology
+ * Access it with the following command:
+ *
+ *	cat /sys/kernel/debug/dri/0/DP-X/is_mst_connector
+ *
+ */
+static int dp_is_mst_connector_show(struct seq_file *m, void *unused)
+{
+	struct drm_connector *connector = m->private;
+	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector);
+	struct drm_dp_mst_topology_mgr *mgr = NULL;
+	struct drm_dp_mst_port *port = NULL;
+	char *role = NULL;
+
+	mutex_lock(&aconnector->hpd_lock);
+
+	if (aconnector->mst_mgr.mst_state) {
+		role = "root";
+	} else if (aconnector->mst_root &&
+		aconnector->mst_root->mst_mgr.mst_state) {
+
+		role = "end";
+
+		mgr = &aconnector->mst_root->mst_mgr;
+		port = aconnector->mst_output_port;
+
+		drm_modeset_lock(&mgr->base.lock, NULL);
+		if (port->pdt == DP_PEER_DEVICE_MST_BRANCHING &&
+			port->mcs)
+			role = "branch";
+		drm_modeset_unlock(&mgr->base.lock);
+
+	} else {
+		role = "no";
+	}
+
+	seq_printf(m, "%s\n", role);
+
+	mutex_unlock(&aconnector->hpd_lock);
+
+	return 0;
+}
+
+/*
+ * function description: Read out the mst progress status
+ *
+ * This function helps to determine the mst progress status of
+ * a mst connector.
+ *
+ * Access it with the following command:
+ *
+ *	cat /sys/kernel/debug/dri/0/DP-X/mst_progress_status
+ *
+ */
+static int dp_mst_progress_status_show(struct seq_file *m, void *unused)
+{
+	struct drm_connector *connector = m->private;
+	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector);
+	struct amdgpu_device *adev = drm_to_adev(connector->dev);
+	int i;
+
+	mutex_lock(&aconnector->hpd_lock);
+	mutex_lock(&adev->dm.dc_lock);
+
+	if (aconnector->mst_status == MST_STATUS_DEFAULT) {
+		seq_puts(m, "disabled\n");
+	} else {
+		for (i = 0; i < sizeof(mst_progress_status)/sizeof(char *); i++)
+			seq_printf(m, "%s:%s\n",
+				mst_progress_status[i],
+				aconnector->mst_status & BIT(i) ? "done" : "not_done");
+	}
+
+	mutex_unlock(&adev->dm.dc_lock);
+	mutex_unlock(&aconnector->hpd_lock);
+
+	return 0;
+}
+
+/*
+ * Reports whether the connected display is a USB4 DPIA tunneled display
+ * Example usage: cat /sys/kernel/debug/dri/0/DP-8/is_dpia_link
+ */
+static int is_dpia_link_show(struct seq_file *m, void *data)
+{
+	struct drm_connector *connector = m->private;
+	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector);
+	struct dc_link *link = aconnector->dc_link;
+
+	if (connector->status != connector_status_connected)
+		return -ENODEV;
+
+	seq_printf(m, "%s\n", (link->ep_type == DISPLAY_ENDPOINT_USB4_DPIA) ? "yes" :
+				(link->ep_type == DISPLAY_ENDPOINT_PHY) ? "no" : "unknown");
+
+	return 0;
+}
+
+/**
+ * hdmi_cec_state_show - Read out the HDMI-CEC feature status
+ * @m: sequence file.
+ * @data: unused.
+ *
+ * Return 0 on success
+ */
+static int hdmi_cec_state_show(struct seq_file *m, void *data)
+{
+	struct drm_connector *connector = m->private;
+	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector);
+
+	seq_printf(m, "%s:%d\n", connector->name, connector->base.id);
+	seq_printf(m, "HDMI-CEC status: %d\n", aconnector->notifier ? 1 : 0);
+
+	return 0;
+}
+
+/**
+ * hdmi_cec_state_write - Enable/Disable HDMI-CEC feature from driver side
+ * @f: file structure.
+ * @buf: userspace buffer. set to '1' to enable; '0' to disable cec feature.
+ * @size: size of buffer from userpsace.
+ * @pos: unused.
+ *
+ * Return size on success, error code on failure
+ */
+static ssize_t hdmi_cec_state_write(struct file *f, const char __user *buf,
+				    size_t size, loff_t *pos)
+{
+	int ret;
+	bool enable;
+	struct amdgpu_dm_connector *aconnector = file_inode(f)->i_private;
+	struct drm_device *ddev = aconnector->base.dev;
+
+	if (size == 0)
+		return -EINVAL;
+
+	ret = kstrtobool_from_user(buf, size, &enable);
+	if (ret) {
+		drm_dbg_driver(ddev, "invalid user data !\n");
+		return ret;
+	}
+
+	if (enable) {
+		if (aconnector->notifier)
+			return -EINVAL;
+		ret = amdgpu_dm_initialize_hdmi_connector(aconnector);
+		if (ret)
+			return ret;
+		hdmi_cec_set_edid(aconnector);
+	} else {
+		if (!aconnector->notifier)
+			return -EINVAL;
+		cec_notifier_conn_unregister(aconnector->notifier);
+		aconnector->notifier = NULL;
+	}
+
+	return size;
+}
+
+DEFINE_SHOW_ATTRIBUTE(dp_dsc_fec_support);
+DEFINE_SHOW_ATTRIBUTE(dmub_fw_state);
+DEFINE_SHOW_ATTRIBUTE(dmub_tracebuffer);
+DEFINE_SHOW_ATTRIBUTE(dp_lttpr_status);
+DEFINE_SHOW_ATTRIBUTE(hdcp_sink_capability);
+DEFINE_SHOW_ATTRIBUTE(internal_display);
+DEFINE_SHOW_ATTRIBUTE(odm_combine_segments);
+DEFINE_SHOW_ATTRIBUTE(replay_capability);
+DEFINE_SHOW_ATTRIBUTE(psr_capability);
+DEFINE_SHOW_ATTRIBUTE(dp_is_mst_connector);
+DEFINE_SHOW_ATTRIBUTE(dp_mst_progress_status);
+DEFINE_SHOW_ATTRIBUTE(is_dpia_link);
+DEFINE_SHOW_STORE_ATTRIBUTE(hdmi_cec_state);
+
+static const struct file_operations dp_dsc_clock_en_debugfs_fops = {
+	.owner = THIS_MODULE,
+	.read = dp_dsc_clock_en_read,
+	.write = dp_dsc_clock_en_write,
+	.llseek = default_llseek
+};
+
+static const struct file_operations dp_dsc_slice_width_debugfs_fops = {
+	.owner = THIS_MODULE,
+	.read = dp_dsc_slice_width_read,
+	.write = dp_dsc_slice_width_write,
+	.llseek = default_llseek
+};
+
+static const struct file_operations dp_dsc_slice_height_debugfs_fops = {
+	.owner = THIS_MODULE,
+	.read = dp_dsc_slice_height_read,
+	.write = dp_dsc_slice_height_write,
+	.llseek = default_llseek
+};
+
+static const struct file_operations dp_dsc_bits_per_pixel_debugfs_fops = {
+	.owner = THIS_MODULE,
+	.read = dp_dsc_bits_per_pixel_read,
+	.write = dp_dsc_bits_per_pixel_write,
+	.llseek = default_llseek
+};
+
+static const struct file_operations dp_dsc_pic_width_debugfs_fops = {
+	.owner = THIS_MODULE,
+	.read = dp_dsc_pic_width_read,
+	.llseek = default_llseek
+};
+
+static const struct file_operations dp_dsc_pic_height_debugfs_fops = {
+	.owner = THIS_MODULE,
+	.read = dp_dsc_pic_height_read,
+	.llseek = default_llseek
+};
+
+static const struct file_operations dp_dsc_chunk_size_debugfs_fops = {
+	.owner = THIS_MODULE,
+	.read = dp_dsc_chunk_size_read,
+	.llseek = default_llseek
+};
+
+static const struct file_operations dp_dsc_slice_bpg_offset_debugfs_fops = {
+	.owner = THIS_MODULE,
+	.read = dp_dsc_slice_bpg_offset_read,
+	.llseek = default_llseek
+};
+
+static const struct file_operations trigger_hotplug_debugfs_fops = {
+	.owner = THIS_MODULE,
+	.write = trigger_hotplug,
+	.llseek = default_llseek
+};
+
+static const struct file_operations dp_link_settings_debugfs_fops = {
+	.owner = THIS_MODULE,
+	.read = dp_link_settings_read,
+	.write = dp_link_settings_write,
+	.llseek = default_llseek
+};
+
+static const struct file_operations dp_phy_settings_debugfs_fop = {
+	.owner = THIS_MODULE,
+	.read = dp_phy_settings_read,
+	.write = dp_phy_settings_write,
+	.llseek = default_llseek
+};
+
+static const struct file_operations dp_phy_test_pattern_fops = {
+	.owner = THIS_MODULE,
+	.write = dp_phy_test_pattern_debugfs_write,
+	.llseek = default_llseek
+};
+
+static const struct file_operations sdp_message_fops = {
+	.owner = THIS_MODULE,
+	.write = dp_sdp_message_debugfs_write,
+	.llseek = default_llseek
+};
+
+static const struct file_operations dp_max_bpc_debugfs_fops = {
+	.owner = THIS_MODULE,
+	.read = dp_max_bpc_read,
+	.write = dp_max_bpc_write,
+	.llseek = default_llseek
+};
+
+static const struct file_operations dp_dsc_disable_passthrough_debugfs_fops = {
+	.owner = THIS_MODULE,
+	.write = dp_dsc_passthrough_set,
+	.llseek = default_llseek
+};
+
+static const struct file_operations dp_mst_link_settings_debugfs_fops = {
+	.owner = THIS_MODULE,
+	.write = dp_mst_link_setting,
+	.llseek = default_llseek
+};
+
+static const struct {
+	char *name;
+	const struct file_operations *fops;
+} dp_debugfs_entries[] = {
+		{"link_settings", &dp_link_settings_debugfs_fops},
+		{"phy_settings", &dp_phy_settings_debugfs_fop},
+		{"lttpr_status", &dp_lttpr_status_fops},
+		{"test_pattern", &dp_phy_test_pattern_fops},
+		{"hdcp_sink_capability", &hdcp_sink_capability_fops},
+		{"sdp_message", &sdp_message_fops},
+		{"dsc_clock_en", &dp_dsc_clock_en_debugfs_fops},
+		{"dsc_slice_width", &dp_dsc_slice_width_debugfs_fops},
+		{"dsc_slice_height", &dp_dsc_slice_height_debugfs_fops},
+		{"dsc_bits_per_pixel", &dp_dsc_bits_per_pixel_debugfs_fops},
+		{"dsc_pic_width", &dp_dsc_pic_width_debugfs_fops},
+		{"dsc_pic_height", &dp_dsc_pic_height_debugfs_fops},
+		{"dsc_chunk_size", &dp_dsc_chunk_size_debugfs_fops},
+		{"dsc_slice_bpg", &dp_dsc_slice_bpg_offset_debugfs_fops},
+		{"dp_dsc_fec_support", &dp_dsc_fec_support_fops},
+		{"max_bpc", &dp_max_bpc_debugfs_fops},
+		{"dsc_disable_passthrough", &dp_dsc_disable_passthrough_debugfs_fops},
+		{"is_mst_connector", &dp_is_mst_connector_fops},
+		{"mst_progress_status", &dp_mst_progress_status_fops},
+		{"is_dpia_link", &is_dpia_link_fops},
+		{"mst_link_settings", &dp_mst_link_settings_debugfs_fops}
+};
+
+static const struct {
+	char *name;
+	const struct file_operations *fops;
+} hdmi_debugfs_entries[] = {
+		{"hdcp_sink_capability", &hdcp_sink_capability_fops},
+		{"hdmi_cec_state", &hdmi_cec_state_fops}
+};
+
+/*
+ * Force YUV420 output if available from the given mode
+ */
+static int force_yuv420_output_set(void *data, u64 val)
+{
+	struct amdgpu_dm_connector *connector = data;
+
+	connector->force_yuv420_output = (bool)val;
+
+	return 0;
+}
+
+/*
+ * Check if YUV420 is forced when available from the given mode
+ */
+static int force_yuv420_output_get(void *data, u64 *val)
+{
+	struct amdgpu_dm_connector *connector = data;
+
+	*val = connector->force_yuv420_output;
+
+	return 0;
+}
+
+DEFINE_DEBUGFS_ATTRIBUTE(force_yuv420_output_fops, force_yuv420_output_get,
+			 force_yuv420_output_set, "%llu\n");
+
+/*
+ *  Read Replay state
+ */
+static int replay_get_state(void *data, u64 *val)
+{
+	struct amdgpu_dm_connector *connector = data;
+	struct dc_link *link = connector->dc_link;
+	uint64_t state = REPLAY_STATE_INVALID;
+
+	dc_link_get_replay_state(link, &state);
+
+	*val = state;
+
+	return 0;
+}
+
+/*
+ *  Start / Stop capture Replay residency
+ */
+static int replay_set_residency(void *data, u64 val)
+{
+	struct amdgpu_dm_connector *connector = data;
+	struct dc_link *link = connector->dc_link;
+	bool is_start = (val != 0);
+	u32 residency = 0;
+
+	link->dc->link_srv->edp_replay_residency(link, &residency, is_start, PR_RESIDENCY_MODE_PHY);
+	return 0;
+}
+
+/*
+ *  Read Replay residency
+ */
+static int replay_get_residency(void *data, u64 *val)
+{
+	struct amdgpu_dm_connector *connector = data;
+	struct dc_link *link = connector->dc_link;
+	u32 residency = 0;
+
+	link->dc->link_srv->edp_replay_residency(link, &residency, false, PR_RESIDENCY_MODE_PHY);
+	*val = (u64)residency;
+
+	return 0;
+}
+
+/*
+ *  Read PSR state
+ */
+static int psr_get(void *data, u64 *val)
+{
+	struct amdgpu_dm_connector *connector = data;
+	struct dc_link *link = connector->dc_link;
+	enum dc_psr_state state = PSR_STATE0;
+
+	dc_link_get_psr_state(link, &state);
+
+	*val = state;
+
+	return 0;
+}
+
+/*
+ *  Read PSR state residency
+ */
+static int psr_read_residency(void *data, u64 *val)
+{
+	struct amdgpu_dm_connector *connector = data;
+	struct dc_link *link = connector->dc_link;
+	u32 residency = 0;
+
+	link->dc->link_srv->edp_get_psr_residency(link, &residency, PSR_RESIDENCY_MODE_PHY);
+
+	*val = (u64)residency;
+
+	return 0;
+}
+
+/* read allow_edp_hotplug_detection */
+static int allow_edp_hotplug_detection_get(void *data, u64 *val)
+{
+	struct amdgpu_dm_connector *aconnector = data;
+	struct drm_connector *connector = &aconnector->base;
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+
+	*val = adev->dm.dc->config.allow_edp_hotplug_detection;
+
+	return 0;
+}
+
+/* set allow_edp_hotplug_detection */
+static int allow_edp_hotplug_detection_set(void *data, u64 val)
+{
+	struct amdgpu_dm_connector *aconnector = data;
+	struct drm_connector *connector = &aconnector->base;
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+
+	adev->dm.dc->config.allow_edp_hotplug_detection = (uint32_t) val;
+
+	return 0;
+}
+
+/* check if kernel disallow eDP enter psr state
+ * cat /sys/kernel/debug/dri/0/eDP-X/disallow_edp_enter_psr
+ * 0: allow edp enter psr; 1: disallow
+ */
+static int disallow_edp_enter_psr_get(void *data, u64 *val)
+{
+	struct amdgpu_dm_connector *aconnector = data;
+
+	*val = (u64) aconnector->disallow_edp_enter_psr;
+	return 0;
+}
+
+/* set kernel disallow eDP enter psr state
+ * echo 0x0 /sys/kernel/debug/dri/0/eDP-X/disallow_edp_enter_psr
+ * 0: allow edp enter psr; 1: disallow
+ *
+ * usage: test app read crc from PSR eDP rx.
+ *
+ * during kernel boot up, kernel write dpcd 0x170 = 5.
+ * this notify eDP rx psr enable and let rx check crc.
+ * rx fw will start checking crc for rx internal logic.
+ * crc read count within dpcd 0x246 is not updated and
+ * value is 0. when eDP tx driver wants to read rx crc
+ * from dpcd 0x246, 0x270, read count 0 lead tx driver
+ * timeout.
+ *
+ * to avoid this, we add this debugfs to let test app to disbable
+ * rx crc checking for rx internal logic. then test app can read
+ * non-zero crc read count.
+ *
+ * expected app sequence is as below:
+ * 1. disable eDP PHY and notify eDP rx with dpcd 0x600 = 2.
+ * 2. echo 0x1 /sys/kernel/debug/dri/0/eDP-X/disallow_edp_enter_psr
+ * 3. enable eDP PHY and notify eDP rx with dpcd 0x600 = 1 but
+ *    without dpcd 0x170 = 5.
+ * 4. read crc from rx dpcd 0x270, 0x246, etc.
+ * 5. echo 0x0 /sys/kernel/debug/dri/0/eDP-X/disallow_edp_enter_psr.
+ *    this will let eDP back to normal with psr setup dpcd 0x170 = 5.
+ */
+static int disallow_edp_enter_psr_set(void *data, u64 val)
+{
+	struct amdgpu_dm_connector *aconnector = data;
+
+	aconnector->disallow_edp_enter_psr = val ? true : false;
+	return 0;
+}
+
+static int dmub_trace_mask_set(void *data, u64 val)
+{
+	struct amdgpu_device *adev = data;
+	struct dmub_srv *srv = adev->dm.dc->ctx->dmub_srv->dmub;
+	enum dmub_gpint_command cmd;
+	u64 mask = 0xffff;
+	u8 shift = 0;
+	u32 res;
+	int i;
+
+	if (!srv->fw_version)
+		return -EINVAL;
+
+	for (i = 0;  i < 4; i++) {
+		res = (val & mask) >> shift;
+
+		switch (i) {
+		case 0:
+			cmd = DMUB_GPINT__SET_TRACE_BUFFER_MASK_WORD0;
+			break;
+		case 1:
+			cmd = DMUB_GPINT__SET_TRACE_BUFFER_MASK_WORD1;
+			break;
+		case 2:
+			cmd = DMUB_GPINT__SET_TRACE_BUFFER_MASK_WORD2;
+			break;
+		case 3:
+			cmd = DMUB_GPINT__SET_TRACE_BUFFER_MASK_WORD3;
+			break;
+		}
+
+		if (!dc_wake_and_execute_gpint(adev->dm.dc->ctx, cmd, res, NULL, DM_DMUB_WAIT_TYPE_WAIT))
+			return -EIO;
+
+		usleep_range(100, 1000);
+
+		mask <<= 16;
+		shift += 16;
+	}
+
+	return 0;
+}
+
+static int dmub_trace_mask_show(void *data, u64 *val)
+{
+	enum dmub_gpint_command cmd = DMUB_GPINT__GET_TRACE_BUFFER_MASK_WORD0;
+	struct amdgpu_device *adev = data;
+	struct dmub_srv *srv = adev->dm.dc->ctx->dmub_srv->dmub;
+	u8 shift = 0;
+	u64 raw = 0;
+	u64 res = 0;
+	int i = 0;
+
+	if (!srv->fw_version)
+		return -EINVAL;
+
+	while (i < 4) {
+		uint32_t response;
+
+		if (!dc_wake_and_execute_gpint(adev->dm.dc->ctx, cmd, 0, &response, DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY))
+			return -EIO;
+
+		raw = response;
+		usleep_range(100, 1000);
+
+		cmd++;
+		res |= (raw << shift);
+		shift += 16;
+		i++;
+	}
+
+	*val = res;
+
+	return 0;
+}
+
+DEFINE_DEBUGFS_ATTRIBUTE(dmub_trace_mask_fops, dmub_trace_mask_show,
+			 dmub_trace_mask_set, "0x%llx\n");
+
+/*
+ * Set dmcub trace event IRQ enable or disable.
+ * Usage to enable dmcub trace event IRQ: echo 1 > /sys/kernel/debug/dri/0/amdgpu_dm_dmcub_trace_event_en
+ * Usage to disable dmcub trace event IRQ: echo 0 > /sys/kernel/debug/dri/0/amdgpu_dm_dmcub_trace_event_en
+ */
+static int dmcub_trace_event_state_set(void *data, u64 val)
+{
+	struct amdgpu_device *adev = data;
+
+	if (val == 1 || val == 0) {
+		dc_dmub_trace_event_control(adev->dm.dc, val);
+		adev->dm.dmcub_trace_event_en = (bool)val;
+	} else
+		return 0;
+
+	return 0;
+}
+
+/*
+ * The interface doesn't need get function, so it will return the
+ * value of zero
+ * Usage: cat /sys/kernel/debug/dri/0/amdgpu_dm_dmcub_trace_event_en
+ */
+static int dmcub_trace_event_state_get(void *data, u64 *val)
+{
+	struct amdgpu_device *adev = data;
+
+	*val = adev->dm.dmcub_trace_event_en;
+	return 0;
+}
+
+DEFINE_DEBUGFS_ATTRIBUTE(dmcub_trace_event_state_fops, dmcub_trace_event_state_get,
+			 dmcub_trace_event_state_set, "%llu\n");
+
+DEFINE_DEBUGFS_ATTRIBUTE(replay_state_fops, replay_get_state, NULL, "%llu\n");
+DEFINE_DEBUGFS_ATTRIBUTE(replay_residency_fops, replay_get_residency, replay_set_residency,
+			 "%llu\n");
+DEFINE_DEBUGFS_ATTRIBUTE(psr_fops, psr_get, NULL, "%llu\n");
+DEFINE_DEBUGFS_ATTRIBUTE(psr_residency_fops, psr_read_residency, NULL,
+			 "%llu\n");
+
+DEFINE_DEBUGFS_ATTRIBUTE(allow_edp_hotplug_detection_fops,
+			allow_edp_hotplug_detection_get,
+			allow_edp_hotplug_detection_set, "%llu\n");
+
+DEFINE_DEBUGFS_ATTRIBUTE(disallow_edp_enter_psr_fops,
+			disallow_edp_enter_psr_get,
+			disallow_edp_enter_psr_set, "%llu\n");
+
+DEFINE_SHOW_ATTRIBUTE(current_backlight);
+DEFINE_SHOW_ATTRIBUTE(target_backlight);
+DEFINE_SHOW_ATTRIBUTE(ips_status);
+
+static const struct {
+	char *name;
+	const struct file_operations *fops;
+} connector_debugfs_entries[] = {
+		{"force_yuv420_output", &force_yuv420_output_fops},
+		{"trigger_hotplug", &trigger_hotplug_debugfs_fops},
+		{"internal_display", &internal_display_fops},
+		{"odm_combine_segments", &odm_combine_segments_fops}
+};
+
+/*
+ * Returns supported customized link rates by this eDP panel.
+ * Example usage: cat /sys/kernel/debug/dri/0/eDP-x/ilr_setting
+ */
+static int edp_ilr_show(struct seq_file *m, void *unused)
+{
+	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(m->private);
+	struct dc_link *link = aconnector->dc_link;
+	uint8_t supported_link_rates[16];
+	uint32_t link_rate_in_khz;
+	uint32_t entry = 0;
+	uint8_t dpcd_rev;
+
+	memset(supported_link_rates, 0, sizeof(supported_link_rates));
+	dm_helpers_dp_read_dpcd(link->ctx, link, DP_SUPPORTED_LINK_RATES,
+		supported_link_rates, sizeof(supported_link_rates));
+
+	dpcd_rev = link->dpcd_caps.dpcd_rev.raw;
+
+	if (dpcd_rev >= DP_DPCD_REV_13 &&
+		(supported_link_rates[entry+1] != 0 || supported_link_rates[entry] != 0)) {
+
+		for (entry = 0; entry < 16; entry += 2) {
+			link_rate_in_khz = (supported_link_rates[entry+1] * 0x100 +
+										supported_link_rates[entry]) * 200;
+			seq_printf(m, "[%d] %d kHz\n", entry/2, link_rate_in_khz);
+		}
+	} else {
+		seq_puts(m, "ILR is not supported by this eDP panel.\n");
+	}
+
+	return 0;
+}
+
+/*
+ * Set supported customized link rate to eDP panel.
+ *
+ * echo <lane_count>  <link_rate option> > ilr_setting
+ *
+ * for example, supported ILR : [0] 1620000 kHz [1] 2160000 kHz [2] 2430000 kHz ...
+ * echo 4 1 > /sys/kernel/debug/dri/0/eDP-x/ilr_setting
+ * to set 4 lanes and 2.16 GHz
+ */
+static ssize_t edp_ilr_write(struct file *f, const char __user *buf,
+				 size_t size, loff_t *pos)
+{
+	struct amdgpu_dm_connector *connector = file_inode(f)->i_private;
+	struct dc_link *link = connector->dc_link;
+	struct amdgpu_device *adev = drm_to_adev(connector->base.dev);
+	struct dc *dc = (struct dc *)link->dc;
+	struct dc_link_settings prefer_link_settings;
+	char *wr_buf = NULL;
+	const uint32_t wr_buf_size = 40;
+	/* 0: lane_count; 1: link_rate */
+	int max_param_num = 2;
+	uint8_t param_nums = 0;
+	long param[2];
+	bool valid_input = true;
+
+	if (size == 0)
+		return -EINVAL;
+
+	wr_buf = kcalloc(wr_buf_size, sizeof(char), GFP_KERNEL);
+	if (!wr_buf)
+		return -ENOMEM;
+
+	if (parse_write_buffer_into_params(wr_buf, wr_buf_size,
+					   (long *)param, buf,
+					   max_param_num,
+					   &param_nums)) {
+		kfree(wr_buf);
+		return -EINVAL;
+	}
+
+	if (param_nums <= 0) {
+		kfree(wr_buf);
+		return -EINVAL;
+	}
+
+	switch (param[0]) {
+	case LANE_COUNT_ONE:
+	case LANE_COUNT_TWO:
+	case LANE_COUNT_FOUR:
+		break;
+	default:
+		valid_input = false;
+		break;
+	}
+
+	if (param[1] >= link->dpcd_caps.edp_supported_link_rates_count)
+		valid_input = false;
+
+	if (!valid_input) {
+		kfree(wr_buf);
+		DRM_DEBUG_DRIVER("Invalid Input value. No HW will be programmed\n");
+		prefer_link_settings.use_link_rate_set = false;
+		mutex_lock(&adev->dm.dc_lock);
+		dc_link_set_preferred_training_settings(dc, NULL, NULL, link, false);
+		mutex_unlock(&adev->dm.dc_lock);
+		return size;
+	}
+
+	/* save user force lane_count, link_rate to preferred settings
+	 * spread spectrum will not be changed
+	 */
+	prefer_link_settings.link_spread = link->cur_link_settings.link_spread;
+	prefer_link_settings.lane_count = param[0];
+	prefer_link_settings.use_link_rate_set = true;
+	prefer_link_settings.link_rate_set = param[1];
+	prefer_link_settings.link_rate = link->dpcd_caps.edp_supported_link_rates[param[1]];
+
+	mutex_lock(&adev->dm.dc_lock);
+	dc_link_set_preferred_training_settings(dc, &prefer_link_settings,
+						NULL, link, false);
+	mutex_unlock(&adev->dm.dc_lock);
+
+	kfree(wr_buf);
+	return size;
+}
+
+static int edp_ilr_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, edp_ilr_show, inode->i_private);
+}
+
+static const struct file_operations edp_ilr_debugfs_fops = {
+	.owner = THIS_MODULE,
+	.open = edp_ilr_open,
+	.read = seq_read,
+	.llseek = seq_lseek,
+	.release = single_release,
+	.write = edp_ilr_write
+};
+
+void connector_debugfs_init(struct amdgpu_dm_connector *connector)
+{
+	int i;
+	struct dentry *dir = connector->base.debugfs_entry;
+
+	if (connector->base.connector_type == DRM_MODE_CONNECTOR_DisplayPort ||
+	    connector->base.connector_type == DRM_MODE_CONNECTOR_eDP) {
+		for (i = 0; i < ARRAY_SIZE(dp_debugfs_entries); i++) {
+			debugfs_create_file(dp_debugfs_entries[i].name,
+					    0644, dir, connector,
+					    dp_debugfs_entries[i].fops);
+		}
+	}
+	if (connector->base.connector_type == DRM_MODE_CONNECTOR_eDP) {
+		debugfs_create_file("replay_capability", 0444, dir, connector,
+					&replay_capability_fops);
+		debugfs_create_file("replay_state", 0444, dir, connector, &replay_state_fops);
+		debugfs_create_file_unsafe("replay_residency", 0444, dir,
+					   connector, &replay_residency_fops);
+		debugfs_create_file_unsafe("psr_capability", 0444, dir, connector, &psr_capability_fops);
+		debugfs_create_file_unsafe("psr_state", 0444, dir, connector, &psr_fops);
+		debugfs_create_file_unsafe("psr_residency", 0444, dir,
+					   connector, &psr_residency_fops);
+		debugfs_create_file("amdgpu_current_backlight_pwm", 0444, dir, connector,
+				    &current_backlight_fops);
+		debugfs_create_file("amdgpu_target_backlight_pwm", 0444, dir, connector,
+				    &target_backlight_fops);
+		debugfs_create_file("ilr_setting", 0644, dir, connector,
+					&edp_ilr_debugfs_fops);
+		debugfs_create_file("allow_edp_hotplug_detection", 0644, dir, connector,
+					&allow_edp_hotplug_detection_fops);
+		debugfs_create_file("disallow_edp_enter_psr", 0644, dir, connector,
+					&disallow_edp_enter_psr_fops);
+	}
+
+	for (i = 0; i < ARRAY_SIZE(connector_debugfs_entries); i++) {
+		debugfs_create_file(connector_debugfs_entries[i].name,
+				    0644, dir, connector,
+				    connector_debugfs_entries[i].fops);
+	}
+
+	if (connector->base.connector_type == DRM_MODE_CONNECTOR_HDMIA) {
+		for (i = 0; i < ARRAY_SIZE(hdmi_debugfs_entries); i++) {
+			debugfs_create_file(hdmi_debugfs_entries[i].name,
+					    0644, dir, connector,
+					    hdmi_debugfs_entries[i].fops);
+		}
+	}
+}
+
+#ifdef CONFIG_DRM_AMD_SECURE_DISPLAY
+/*
+ * Set crc window coordinate x start
+ */
+static int crc_win_x_start_set(void *data, u64 val)
+{
+	struct drm_crtc *crtc = data;
+	struct drm_device *drm_dev = crtc->dev;
+	struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
+
+	spin_lock_irq(&drm_dev->event_lock);
+	acrtc->dm_irq_params.window_param[0].x_start = (uint16_t) val;
+	acrtc->dm_irq_params.window_param[0].update_win = false;
+	spin_unlock_irq(&drm_dev->event_lock);
+
+	return 0;
+}
+
+/*
+ * Get crc window coordinate x start
+ */
+static int crc_win_x_start_get(void *data, u64 *val)
+{
+	struct drm_crtc *crtc = data;
+	struct drm_device *drm_dev = crtc->dev;
+	struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
+
+	spin_lock_irq(&drm_dev->event_lock);
+	*val = acrtc->dm_irq_params.window_param[0].x_start;
+	spin_unlock_irq(&drm_dev->event_lock);
+
+	return 0;
+}
+
+DEFINE_DEBUGFS_ATTRIBUTE(crc_win_x_start_fops, crc_win_x_start_get,
+			 crc_win_x_start_set, "%llu\n");
+
+
+/*
+ * Set crc window coordinate y start
+ */
+static int crc_win_y_start_set(void *data, u64 val)
+{
+	struct drm_crtc *crtc = data;
+	struct drm_device *drm_dev = crtc->dev;
+	struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
+
+	spin_lock_irq(&drm_dev->event_lock);
+	acrtc->dm_irq_params.window_param[0].y_start = (uint16_t) val;
+	acrtc->dm_irq_params.window_param[0].update_win = false;
+	spin_unlock_irq(&drm_dev->event_lock);
+
+	return 0;
+}
+
+/*
+ * Get crc window coordinate y start
+ */
+static int crc_win_y_start_get(void *data, u64 *val)
+{
+	struct drm_crtc *crtc = data;
+	struct drm_device *drm_dev = crtc->dev;
+	struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
+
+	spin_lock_irq(&drm_dev->event_lock);
+	*val = acrtc->dm_irq_params.window_param[0].y_start;
+	spin_unlock_irq(&drm_dev->event_lock);
+
+	return 0;
+}
+
+DEFINE_DEBUGFS_ATTRIBUTE(crc_win_y_start_fops, crc_win_y_start_get,
+			 crc_win_y_start_set, "%llu\n");
+
+/*
+ * Set crc window coordinate x end
+ */
+static int crc_win_x_end_set(void *data, u64 val)
+{
+	struct drm_crtc *crtc = data;
+	struct drm_device *drm_dev = crtc->dev;
+	struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
+
+	spin_lock_irq(&drm_dev->event_lock);
+	acrtc->dm_irq_params.window_param[0].x_end = (uint16_t) val;
+	acrtc->dm_irq_params.window_param[0].update_win = false;
+	spin_unlock_irq(&drm_dev->event_lock);
+
+	return 0;
+}
+
+/*
+ * Get crc window coordinate x end
+ */
+static int crc_win_x_end_get(void *data, u64 *val)
+{
+	struct drm_crtc *crtc = data;
+	struct drm_device *drm_dev = crtc->dev;
+	struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
+
+	spin_lock_irq(&drm_dev->event_lock);
+	*val = acrtc->dm_irq_params.window_param[0].x_end;
+	spin_unlock_irq(&drm_dev->event_lock);
+
+	return 0;
+}
+
+DEFINE_DEBUGFS_ATTRIBUTE(crc_win_x_end_fops, crc_win_x_end_get,
+			 crc_win_x_end_set, "%llu\n");
+
+/*
+ * Set crc window coordinate y end
+ */
+static int crc_win_y_end_set(void *data, u64 val)
+{
+	struct drm_crtc *crtc = data;
+	struct drm_device *drm_dev = crtc->dev;
+	struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
+
+	spin_lock_irq(&drm_dev->event_lock);
+	acrtc->dm_irq_params.window_param[0].y_end = (uint16_t) val;
+	acrtc->dm_irq_params.window_param[0].update_win = false;
+	spin_unlock_irq(&drm_dev->event_lock);
+
+	return 0;
+}
+
+/*
+ * Get crc window coordinate y end
+ */
+static int crc_win_y_end_get(void *data, u64 *val)
+{
+	struct drm_crtc *crtc = data;
+	struct drm_device *drm_dev = crtc->dev;
+	struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
+
+	spin_lock_irq(&drm_dev->event_lock);
+	*val = acrtc->dm_irq_params.window_param[0].y_end;
+	spin_unlock_irq(&drm_dev->event_lock);
+
+	return 0;
+}
+
+DEFINE_DEBUGFS_ATTRIBUTE(crc_win_y_end_fops, crc_win_y_end_get,
+			 crc_win_y_end_set, "%llu\n");
+/*
+ * Trigger to commit crc window
+ */
+static int crc_win_update_set(void *data, u64 val)
+{
+	struct drm_crtc *crtc = data;
+	struct amdgpu_crtc *acrtc;
+	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
+
+	if (val) {
+		acrtc = to_amdgpu_crtc(crtc);
+		mutex_lock(&adev->dm.dc_lock);
+		/* PSR may write to OTG CRC window control register,
+		 * so close it before starting secure_display.
+		 */
+		amdgpu_dm_psr_disable(acrtc->dm_irq_params.stream, true);
+
+		spin_lock_irq(&adev_to_drm(adev)->event_lock);
+
+		acrtc->dm_irq_params.window_param[0].enable = true;
+		acrtc->dm_irq_params.window_param[0].update_win = true;
+		acrtc->dm_irq_params.window_param[0].skip_frame_cnt = 0;
+		acrtc->dm_irq_params.crc_window_activated = true;
+
+		spin_unlock_irq(&adev_to_drm(adev)->event_lock);
+		mutex_unlock(&adev->dm.dc_lock);
+	}
+
+	return 0;
+}
+
+/*
+ * Get crc window update flag
+ */
+static int crc_win_update_get(void *data, u64 *val)
+{
+	*val = 0;
+	return 0;
+}
+
+DEFINE_DEBUGFS_ATTRIBUTE(crc_win_update_fops, crc_win_update_get,
+			 crc_win_update_set, "%llu\n");
+#endif
+void crtc_debugfs_init(struct drm_crtc *crtc)
+{
+#ifdef CONFIG_DRM_AMD_SECURE_DISPLAY
+	struct dentry *dir = debugfs_lookup("crc", crtc->debugfs_entry);
+
+	if (!dir)
+		return;
+
+	debugfs_create_file_unsafe("crc_win_x_start", 0644, dir, crtc,
+				   &crc_win_x_start_fops);
+	debugfs_create_file_unsafe("crc_win_y_start", 0644, dir, crtc,
+				   &crc_win_y_start_fops);
+	debugfs_create_file_unsafe("crc_win_x_end", 0644, dir, crtc,
+				   &crc_win_x_end_fops);
+	debugfs_create_file_unsafe("crc_win_y_end", 0644, dir, crtc,
+				   &crc_win_y_end_fops);
+	debugfs_create_file_unsafe("crc_win_update", 0644, dir, crtc,
+				   &crc_win_update_fops);
+	dput(dir);
+#endif
+	debugfs_create_file("amdgpu_current_bpc", 0644, crtc->debugfs_entry,
+			    crtc, &amdgpu_current_bpc_fops);
+	debugfs_create_file("amdgpu_current_colorspace", 0644, crtc->debugfs_entry,
+			    crtc, &amdgpu_current_colorspace_fops);
+}
+
+/*
+ * Writes DTN log state to the user supplied buffer.
+ * Example usage: cat /sys/kernel/debug/dri/0/amdgpu_dm_dtn_log
+ */
+static ssize_t dtn_log_read(
+	struct file *f,
+	char __user *buf,
+	size_t size,
+	loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	struct dc *dc = adev->dm.dc;
+	struct dc_log_buffer_ctx log_ctx = { 0 };
+	ssize_t result = 0;
+
+	if (!buf || !size)
+		return -EINVAL;
+
+	if (!dc->hwss.log_hw_state)
+		return 0;
+
+	dc->hwss.log_hw_state(dc, &log_ctx);
+
+	if (*pos < log_ctx.pos) {
+		size_t to_copy = log_ctx.pos - *pos;
+
+		to_copy = min(to_copy, size);
+
+		if (!copy_to_user(buf, log_ctx.buf + *pos, to_copy)) {
+			*pos += to_copy;
+			result = to_copy;
+		}
+	}
+
+	kfree(log_ctx.buf);
+
+	return result;
+}
+
+/*
+ * Writes DTN log state to dmesg when triggered via a write.
+ * Example usage: echo 1 > /sys/kernel/debug/dri/0/amdgpu_dm_dtn_log
+ */
+static ssize_t dtn_log_write(
+	struct file *f,
+	const char __user *buf,
+	size_t size,
+	loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	struct dc *dc = adev->dm.dc;
+
+	/* Write triggers log output via dmesg. */
+	if (size == 0)
+		return 0;
+
+	if (dc->hwss.log_hw_state)
+		dc->hwss.log_hw_state(dc, NULL);
+
+	return size;
+}
+
+static int mst_topo_show(struct seq_file *m, void *unused)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)m->private;
+	struct drm_device *dev = adev_to_drm(adev);
+	struct drm_connector *connector;
+	struct drm_connector_list_iter conn_iter;
+	struct amdgpu_dm_connector *aconnector;
+
+	drm_connector_list_iter_begin(dev, &conn_iter);
+	drm_for_each_connector_iter(connector, &conn_iter) {
+		if (connector->connector_type != DRM_MODE_CONNECTOR_DisplayPort)
+			continue;
+
+		aconnector = to_amdgpu_dm_connector(connector);
+
+		/* Ensure we're only dumping the topology of a root mst node */
+		if (!aconnector->mst_mgr.mst_state)
+			continue;
+
+		seq_printf(m, "\nMST topology for connector %d\n", aconnector->connector_id);
+		drm_dp_mst_dump_topology(m, &aconnector->mst_mgr);
+	}
+	drm_connector_list_iter_end(&conn_iter);
+
+	return 0;
+}
+
+/*
+ * Sets trigger hpd for MST topologies.
+ * All connected connectors will be rediscovered and re started as needed if val of 1 is sent.
+ * All topologies will be disconnected if val of 0 is set .
+ * Usage to enable topologies: echo 1 > /sys/kernel/debug/dri/0/amdgpu_dm_trigger_hpd_mst
+ * Usage to disable topologies: echo 0 > /sys/kernel/debug/dri/0/amdgpu_dm_trigger_hpd_mst
+ */
+static int trigger_hpd_mst_set(void *data, u64 val)
+{
+	struct amdgpu_device *adev = data;
+	struct drm_device *dev = adev_to_drm(adev);
+	struct drm_connector_list_iter iter;
+	struct amdgpu_dm_connector *aconnector;
+	struct drm_connector *connector;
+	struct dc_link *link = NULL;
+	int ret;
+
+	if (val == 1) {
+		drm_connector_list_iter_begin(dev, &iter);
+		drm_for_each_connector_iter(connector, &iter) {
+			aconnector = to_amdgpu_dm_connector(connector);
+			if (aconnector->dc_link->type == dc_connection_mst_branch &&
+			    aconnector->mst_mgr.aux) {
+				mutex_lock(&adev->dm.dc_lock);
+				ret = dc_link_detect(aconnector->dc_link, DETECT_REASON_HPD);
+				mutex_unlock(&adev->dm.dc_lock);
+
+				if (!ret)
+					DRM_ERROR("DM_MST: Failed to detect dc link!");
+
+				ret = drm_dp_mst_topology_mgr_set_mst(&aconnector->mst_mgr, true);
+				if (ret < 0)
+					DRM_ERROR("DM_MST: Failed to set the device into MST mode!");
+			}
+		}
+	} else if (val == 0) {
+		drm_connector_list_iter_begin(dev, &iter);
+		drm_for_each_connector_iter(connector, &iter) {
+			aconnector = to_amdgpu_dm_connector(connector);
+			if (!aconnector->dc_link)
+				continue;
+
+			if (!aconnector->mst_root)
+				continue;
+
+			link = aconnector->dc_link;
+			dc_link_dp_receiver_power_ctrl(link, false);
+			drm_dp_mst_topology_mgr_set_mst(&aconnector->mst_root->mst_mgr, false);
+			link->mst_stream_alloc_table.stream_count = 0;
+			memset(link->mst_stream_alloc_table.stream_allocations, 0,
+					sizeof(link->mst_stream_alloc_table.stream_allocations));
+		}
+	} else {
+		return 0;
+	}
+	drm_kms_helper_hotplug_event(dev);
+
+	return 0;
+}
+
+/*
+ * The interface doesn't need get function, so it will return the
+ * value of zero
+ * Usage: cat /sys/kernel/debug/dri/0/amdgpu_dm_trigger_hpd_mst
+ */
+static int trigger_hpd_mst_get(void *data, u64 *val)
+{
+	*val = 0;
+	return 0;
+}
+
+DEFINE_DEBUGFS_ATTRIBUTE(trigger_hpd_mst_ops, trigger_hpd_mst_get,
+			 trigger_hpd_mst_set, "%llu\n");
+
+
+/*
+ * Sets the force_timing_sync debug option from the given string.
+ * All connected displays will be force synchronized immediately.
+ * Usage: echo 1 > /sys/kernel/debug/dri/0/amdgpu_dm_force_timing_sync
+ */
+static int force_timing_sync_set(void *data, u64 val)
+{
+	struct amdgpu_device *adev = data;
+
+	adev->dm.force_timing_sync = (bool)val;
+
+	amdgpu_dm_trigger_timing_sync(adev_to_drm(adev));
+
+	return 0;
+}
+
+/*
+ * Gets the force_timing_sync debug option value into the given buffer.
+ * Usage: cat /sys/kernel/debug/dri/0/amdgpu_dm_force_timing_sync
+ */
+static int force_timing_sync_get(void *data, u64 *val)
+{
+	struct amdgpu_device *adev = data;
+
+	*val = adev->dm.force_timing_sync;
+
+	return 0;
+}
+
+DEFINE_DEBUGFS_ATTRIBUTE(force_timing_sync_ops, force_timing_sync_get,
+			 force_timing_sync_set, "%llu\n");
+
+
+/*
+ * Disables all HPD and HPD RX interrupt handling in the
+ * driver when set to 1. Default is 0.
+ */
+static int disable_hpd_set(void *data, u64 val)
+{
+	struct amdgpu_device *adev = data;
+
+	adev->dm.disable_hpd_irq = (bool)val;
+
+	return 0;
+}
+
+
+/*
+ * Returns 1 if HPD and HPRX interrupt handling is disabled,
+ * 0 otherwise.
+ */
+static int disable_hpd_get(void *data, u64 *val)
+{
+	struct amdgpu_device *adev = data;
+
+	*val = adev->dm.disable_hpd_irq;
+
+	return 0;
+}
+
+DEFINE_DEBUGFS_ATTRIBUTE(disable_hpd_ops, disable_hpd_get,
+			 disable_hpd_set, "%llu\n");
+
+/*
+ * Prints hardware capabilities. These are used for IGT testing.
+ */
+static int capabilities_show(struct seq_file *m, void *unused)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)m->private;
+	struct dc *dc = adev->dm.dc;
+	bool mall_supported = dc->caps.mall_size_total;
+	bool subvp_supported = dc->caps.subvp_fw_processing_delay_us;
+	unsigned int mall_in_use = false;
+	unsigned int subvp_in_use = false;
+
+	struct hubbub *hubbub = dc->res_pool->hubbub;
+
+	if (hubbub && hubbub->funcs->get_mall_en)
+		hubbub->funcs->get_mall_en(hubbub, &mall_in_use);
+
+	if (dc->cap_funcs.get_subvp_en)
+		subvp_in_use = dc->cap_funcs.get_subvp_en(dc, dc->current_state);
+
+	seq_printf(m, "mall supported: %s, enabled: %s\n",
+			   mall_supported ? "yes" : "no", mall_in_use ? "yes" : "no");
+	seq_printf(m, "sub-viewport supported: %s, enabled: %s\n",
+			   subvp_supported ? "yes" : "no", subvp_in_use ? "yes" : "no");
+
+	return 0;
+}
+
+DEFINE_SHOW_ATTRIBUTE(capabilities);
+
+/*
+ * Temporary w/a to force sst sequence in M42D DP2 mst receiver
+ * Example usage: echo 1 > /sys/kernel/debug/dri/0/amdgpu_dm_dp_set_mst_en_for_sst
+ */
+static int dp_force_sst_set(void *data, u64 val)
+{
+	struct amdgpu_device *adev = data;
+
+	adev->dm.dc->debug.set_mst_en_for_sst = val;
+
+	return 0;
+}
+
+static int dp_force_sst_get(void *data, u64 *val)
+{
+	struct amdgpu_device *adev = data;
+
+	*val = adev->dm.dc->debug.set_mst_en_for_sst;
+
+	return 0;
+}
+DEFINE_DEBUGFS_ATTRIBUTE(dp_set_mst_en_for_sst_ops, dp_force_sst_get,
+			 dp_force_sst_set, "%llu\n");
+
+/*
+ * Force DP2 sequence without VESA certified cable.
+ * Example usage: echo 1 > /sys/kernel/debug/dri/0/amdgpu_dm_dp_ignore_cable_id
+ */
+static int dp_ignore_cable_id_set(void *data, u64 val)
+{
+	struct amdgpu_device *adev = data;
+
+	adev->dm.dc->debug.ignore_cable_id = val;
+
+	return 0;
+}
+
+static int dp_ignore_cable_id_get(void *data, u64 *val)
+{
+	struct amdgpu_device *adev = data;
+
+	*val = adev->dm.dc->debug.ignore_cable_id;
+
+	return 0;
+}
+DEFINE_DEBUGFS_ATTRIBUTE(dp_ignore_cable_id_ops, dp_ignore_cable_id_get,
+			 dp_ignore_cable_id_set, "%llu\n");
+
+/*
+ * Sets the DC visual confirm debug option from the given string.
+ * Example usage: echo 1 > /sys/kernel/debug/dri/0/amdgpu_visual_confirm
+ */
+static int visual_confirm_set(void *data, u64 val)
+{
+	struct amdgpu_device *adev = data;
+
+	adev->dm.dc->debug.visual_confirm = (enum visual_confirm)val;
+
+	return 0;
+}
+
+/*
+ * Reads the DC visual confirm debug option value into the given buffer.
+ * Example usage: cat /sys/kernel/debug/dri/0/amdgpu_dm_visual_confirm
+ */
+static int visual_confirm_get(void *data, u64 *val)
+{
+	struct amdgpu_device *adev = data;
+
+	*val = adev->dm.dc->debug.visual_confirm;
+
+	return 0;
+}
+
+DEFINE_SHOW_ATTRIBUTE(mst_topo);
+DEFINE_DEBUGFS_ATTRIBUTE(visual_confirm_fops, visual_confirm_get,
+			 visual_confirm_set, "%llu\n");
+
+
+/*
+ * Sets the DC skip_detection_link_training debug option from the given string.
+ * Example usage: echo 1 > /sys/kernel/debug/dri/0/amdgpu_skip_detection_link_training
+ */
+static int skip_detection_link_training_set(void *data, u64 val)
+{
+	struct amdgpu_device *adev = data;
+
+	if (val == 0)
+		adev->dm.dc->debug.skip_detection_link_training = false;
+	else
+		adev->dm.dc->debug.skip_detection_link_training = true;
+
+	return 0;
+}
+
+/*
+ * Reads the DC skip_detection_link_training debug option value into the given buffer.
+ * Example usage: cat /sys/kernel/debug/dri/0/amdgpu_dm_skip_detection_link_training
+ */
+static int skip_detection_link_training_get(void *data, u64 *val)
+{
+	struct amdgpu_device *adev = data;
+
+	*val = adev->dm.dc->debug.skip_detection_link_training;
+
+	return 0;
+}
+
+DEFINE_DEBUGFS_ATTRIBUTE(skip_detection_link_training_fops,
+			 skip_detection_link_training_get,
+			 skip_detection_link_training_set, "%llu\n");
+
+/*
+ * Dumps the DCC_EN bit for each pipe.
+ * Example usage: cat /sys/kernel/debug/dri/0/amdgpu_dm_dcc_en
+ */
+static ssize_t dcc_en_bits_read(
+	struct file *f,
+	char __user *buf,
+	size_t size,
+	loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	struct dc *dc = adev->dm.dc;
+	char *rd_buf = NULL;
+	const uint32_t rd_buf_size = 32;
+	uint32_t result = 0;
+	int offset = 0;
+	int num_pipes = dc->res_pool->pipe_count;
+	int *dcc_en_bits;
+	int i, r;
+
+	dcc_en_bits = kcalloc(num_pipes, sizeof(int), GFP_KERNEL);
+	if (!dcc_en_bits)
+		return -ENOMEM;
+
+	if (!dc->hwss.get_dcc_en_bits) {
+		kfree(dcc_en_bits);
+		return 0;
+	}
+
+	dc->hwss.get_dcc_en_bits(dc, dcc_en_bits);
+
+	rd_buf = kcalloc(rd_buf_size, sizeof(char), GFP_KERNEL);
+	if (!rd_buf) {
+		kfree(dcc_en_bits);
+		return -ENOMEM;
+	}
+
+	for (i = 0; i < num_pipes; i++)
+		offset += snprintf(rd_buf + offset, rd_buf_size - offset,
+				   "%d  ", dcc_en_bits[i]);
+	rd_buf[strlen(rd_buf)] = '\n';
+
+	kfree(dcc_en_bits);
+
+	while (size) {
+		if (*pos >= rd_buf_size)
+			break;
+		r = put_user(*(rd_buf + result), buf);
+		if (r) {
+			kfree(rd_buf);
+			return r; /* r = -EFAULT */
+		}
+		buf += 1;
+		size -= 1;
+		*pos += 1;
+		result += 1;
+	}
+
+	kfree(rd_buf);
+	return result;
+}
+
+void dtn_debugfs_init(struct amdgpu_device *adev)
+{
+	static const struct file_operations dtn_log_fops = {
+		.owner = THIS_MODULE,
+		.read = dtn_log_read,
+		.write = dtn_log_write,
+		.llseek = default_llseek
+	};
+	static const struct file_operations dcc_en_bits_fops = {
+		.owner = THIS_MODULE,
+		.read = dcc_en_bits_read,
+		.llseek = default_llseek
+	};
+
+	struct drm_minor *minor = adev_to_drm(adev)->primary;
+	struct dentry *root = minor->debugfs_root;
+
+	debugfs_create_file("amdgpu_mst_topology", 0444, root,
+			    adev, &mst_topo_fops);
+	debugfs_create_file("amdgpu_dm_capabilities", 0444, root,
+			    adev, &capabilities_fops);
+	debugfs_create_file("amdgpu_dm_dtn_log", 0644, root, adev,
+			    &dtn_log_fops);
+	debugfs_create_file("amdgpu_dm_dp_set_mst_en_for_sst", 0644, root, adev,
+				&dp_set_mst_en_for_sst_ops);
+	debugfs_create_file("amdgpu_dm_dp_ignore_cable_id", 0644, root, adev,
+				&dp_ignore_cable_id_ops);
+
+	debugfs_create_file_unsafe("amdgpu_dm_visual_confirm", 0644, root, adev,
+				   &visual_confirm_fops);
+
+	debugfs_create_file_unsafe("amdgpu_dm_skip_detection_link_training", 0644, root, adev,
+				   &skip_detection_link_training_fops);
+
+	debugfs_create_file_unsafe("amdgpu_dm_dmub_tracebuffer", 0644, root,
+				   adev, &dmub_tracebuffer_fops);
+
+	debugfs_create_file_unsafe("amdgpu_dm_dmub_fw_state", 0644, root,
+				   adev, &dmub_fw_state_fops);
+
+	debugfs_create_file_unsafe("amdgpu_dm_force_timing_sync", 0644, root,
+				   adev, &force_timing_sync_ops);
+
+	debugfs_create_file_unsafe("amdgpu_dm_dmub_trace_mask", 0644, root,
+				   adev, &dmub_trace_mask_fops);
+
+	debugfs_create_file_unsafe("amdgpu_dm_dmcub_trace_event_en", 0644, root,
+				   adev, &dmcub_trace_event_state_fops);
+
+	debugfs_create_file_unsafe("amdgpu_dm_trigger_hpd_mst", 0644, root,
+				   adev, &trigger_hpd_mst_ops);
+
+	debugfs_create_file_unsafe("amdgpu_dm_dcc_en", 0644, root, adev,
+				   &dcc_en_bits_fops);
+
+	debugfs_create_file_unsafe("amdgpu_dm_disable_hpd", 0644, root, adev,
+				   &disable_hpd_ops);
+
+	if (adev->dm.dc->caps.ips_support)
+		debugfs_create_file_unsafe("amdgpu_dm_ips_status", 0644, root, adev,
+					   &ips_status_fops);
+}
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.h
new file mode 100644
index 000000000000..122cdc124b3b
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.h
@@ -0,0 +1,37 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __AMDGPU_DM_DEBUGFS_H__
+#define __AMDGPU_DM_DEBUGFS_H__
+
+#include "amdgpu.h"
+#include "amdgpu_dm.h"
+
+void connector_debugfs_init(struct amdgpu_dm_connector *connector);
+void dtn_debugfs_init(struct amdgpu_device *adev);
+void crtc_debugfs_init(struct drm_crtc *crtc);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c
new file mode 100644
index 000000000000..19038f336155
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c
@@ -0,0 +1,827 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "amdgpu_dm_hdcp.h"
+#include "amdgpu.h"
+#include "amdgpu_dm.h"
+#include "dc_fused_io.h"
+#include "dm_helpers.h"
+#include <drm/display/drm_hdcp_helper.h>
+#include "hdcp_psp.h"
+
+/*
+ * If the SRM version being loaded is less than or equal to the
+ * currently loaded SRM, psp will return 0xFFFF as the version
+ */
+#define PSP_SRM_VERSION_MAX 0xFFFF
+
+static bool
+lp_write_i2c(void *handle, uint32_t address, const uint8_t *data, uint32_t size)
+{
+	struct dc_link *link = handle;
+	struct i2c_payload i2c_payloads[] = {{true, address, size, (void *)data} };
+	struct i2c_command cmd = {i2c_payloads, 1, I2C_COMMAND_ENGINE_HW,
+				  link->dc->caps.i2c_speed_in_khz};
+
+	return dm_helpers_submit_i2c(link->ctx, link, &cmd);
+}
+
+static bool
+lp_read_i2c(void *handle, uint32_t address, uint8_t offset, uint8_t *data, uint32_t size)
+{
+	struct dc_link *link = handle;
+
+	struct i2c_payload i2c_payloads[] = {{true, address, 1, &offset},
+					     {false, address, size, data} };
+	struct i2c_command cmd = {i2c_payloads, 2, I2C_COMMAND_ENGINE_HW,
+				  link->dc->caps.i2c_speed_in_khz};
+
+	return dm_helpers_submit_i2c(link->ctx, link, &cmd);
+}
+
+static bool
+lp_write_dpcd(void *handle, uint32_t address, const uint8_t *data, uint32_t size)
+{
+	struct dc_link *link = handle;
+
+	return dm_helpers_dp_write_dpcd(link->ctx, link, address, data, size);
+}
+
+static bool
+lp_read_dpcd(void *handle, uint32_t address, uint8_t *data, uint32_t size)
+{
+	struct dc_link *link = handle;
+
+	return dm_helpers_dp_read_dpcd(link->ctx, link, address, data, size);
+}
+
+static bool lp_atomic_write_poll_read_i2c(
+		void *handle,
+		const struct mod_hdcp_atomic_op_i2c *write,
+		const struct mod_hdcp_atomic_op_i2c *poll,
+		struct mod_hdcp_atomic_op_i2c *read,
+		uint32_t poll_timeout_us,
+		uint8_t poll_mask_msb
+)
+{
+	struct dc_link *link = handle;
+
+	return dm_atomic_write_poll_read_i2c(link, write, poll, read, poll_timeout_us, poll_mask_msb);
+}
+
+static bool lp_atomic_write_poll_read_aux(
+		void *handle,
+		const struct mod_hdcp_atomic_op_aux *write,
+		const struct mod_hdcp_atomic_op_aux *poll,
+		struct mod_hdcp_atomic_op_aux *read,
+		uint32_t poll_timeout_us,
+		uint8_t poll_mask_msb
+)
+{
+	struct dc_link *link = handle;
+
+	return dm_atomic_write_poll_read_aux(link, write, poll, read, poll_timeout_us, poll_mask_msb);
+}
+
+static uint8_t *psp_get_srm(struct psp_context *psp, uint32_t *srm_version, uint32_t *srm_size)
+{
+	struct ta_hdcp_shared_memory *hdcp_cmd;
+
+	if (!psp->hdcp_context.context.initialized) {
+		DRM_WARN("Failed to get hdcp srm. HDCP TA is not initialized.");
+		return NULL;
+	}
+
+	hdcp_cmd = (struct ta_hdcp_shared_memory *)psp->hdcp_context.context.mem_context.shared_buf;
+	memset(hdcp_cmd, 0, sizeof(struct ta_hdcp_shared_memory));
+
+	hdcp_cmd->cmd_id = TA_HDCP_COMMAND__HDCP_GET_SRM;
+	psp_hdcp_invoke(psp, hdcp_cmd->cmd_id);
+
+	if (hdcp_cmd->hdcp_status != TA_HDCP_STATUS__SUCCESS)
+		return NULL;
+
+	*srm_version = hdcp_cmd->out_msg.hdcp_get_srm.srm_version;
+	*srm_size = hdcp_cmd->out_msg.hdcp_get_srm.srm_buf_size;
+
+	return hdcp_cmd->out_msg.hdcp_get_srm.srm_buf;
+}
+
+static int psp_set_srm(struct psp_context *psp,
+		       u8 *srm, uint32_t srm_size, uint32_t *srm_version)
+{
+	struct ta_hdcp_shared_memory *hdcp_cmd;
+
+	if (!psp->hdcp_context.context.initialized) {
+		DRM_WARN("Failed to get hdcp srm. HDCP TA is not initialized.");
+		return -EINVAL;
+	}
+
+	hdcp_cmd = (struct ta_hdcp_shared_memory *)psp->hdcp_context.context.mem_context.shared_buf;
+	memset(hdcp_cmd, 0, sizeof(struct ta_hdcp_shared_memory));
+
+	memcpy(hdcp_cmd->in_msg.hdcp_set_srm.srm_buf, srm, srm_size);
+	hdcp_cmd->in_msg.hdcp_set_srm.srm_buf_size = srm_size;
+	hdcp_cmd->cmd_id = TA_HDCP_COMMAND__HDCP_SET_SRM;
+
+	psp_hdcp_invoke(psp, hdcp_cmd->cmd_id);
+
+	if (hdcp_cmd->hdcp_status != TA_HDCP_STATUS__SUCCESS ||
+	    hdcp_cmd->out_msg.hdcp_set_srm.valid_signature != 1 ||
+	    hdcp_cmd->out_msg.hdcp_set_srm.srm_version == PSP_SRM_VERSION_MAX)
+		return -EINVAL;
+
+	*srm_version = hdcp_cmd->out_msg.hdcp_set_srm.srm_version;
+	return 0;
+}
+
+static void process_output(struct hdcp_workqueue *hdcp_work)
+{
+	struct mod_hdcp_output output = hdcp_work->output;
+
+	if (output.callback_stop)
+		cancel_delayed_work(&hdcp_work->callback_dwork);
+
+	if (output.callback_needed)
+		schedule_delayed_work(&hdcp_work->callback_dwork,
+				      msecs_to_jiffies(output.callback_delay));
+
+	if (output.watchdog_timer_stop)
+		cancel_delayed_work(&hdcp_work->watchdog_timer_dwork);
+
+	if (output.watchdog_timer_needed)
+		schedule_delayed_work(&hdcp_work->watchdog_timer_dwork,
+				      msecs_to_jiffies(output.watchdog_timer_delay));
+
+	schedule_delayed_work(&hdcp_work->property_validate_dwork, msecs_to_jiffies(0));
+}
+
+static void link_lock(struct hdcp_workqueue *work, bool lock)
+{
+	int i = 0;
+
+	for (i = 0; i < work->max_link; i++) {
+		if (lock)
+			mutex_lock(&work[i].mutex);
+		else
+			mutex_unlock(&work[i].mutex);
+	}
+}
+
+void hdcp_update_display(struct hdcp_workqueue *hdcp_work,
+			 unsigned int link_index,
+			 struct amdgpu_dm_connector *aconnector,
+			 u8 content_type,
+			 bool enable_encryption)
+{
+	struct hdcp_workqueue *hdcp_w = &hdcp_work[link_index];
+	struct mod_hdcp_link_adjustment link_adjust;
+	struct mod_hdcp_display_adjustment display_adjust;
+	unsigned int conn_index = aconnector->base.index;
+
+	guard(mutex)(&hdcp_w->mutex);
+	drm_connector_get(&aconnector->base);
+	if (hdcp_w->aconnector[conn_index])
+		drm_connector_put(&hdcp_w->aconnector[conn_index]->base);
+	hdcp_w->aconnector[conn_index] = aconnector;
+
+	memset(&link_adjust, 0, sizeof(link_adjust));
+	memset(&display_adjust, 0, sizeof(display_adjust));
+
+	if (enable_encryption) {
+		/* Explicitly set the saved SRM as sysfs call will be after we already enabled hdcp
+		 * (s3 resume case)
+		 */
+		if (hdcp_work->srm_size > 0)
+			psp_set_srm(hdcp_work->hdcp.config.psp.handle, hdcp_work->srm,
+				    hdcp_work->srm_size,
+				    &hdcp_work->srm_version);
+
+		display_adjust.disable = MOD_HDCP_DISPLAY_NOT_DISABLE;
+
+		link_adjust.auth_delay = 2;
+		link_adjust.retry_limit = MAX_NUM_OF_ATTEMPTS;
+
+		if (content_type == DRM_MODE_HDCP_CONTENT_TYPE0) {
+			link_adjust.hdcp2.force_type = MOD_HDCP_FORCE_TYPE_0;
+		} else if (content_type == DRM_MODE_HDCP_CONTENT_TYPE1) {
+			link_adjust.hdcp1.disable = 1;
+			link_adjust.hdcp2.force_type = MOD_HDCP_FORCE_TYPE_1;
+		}
+
+		schedule_delayed_work(&hdcp_w->property_validate_dwork,
+				      msecs_to_jiffies(DRM_HDCP_CHECK_PERIOD_MS));
+	} else {
+		display_adjust.disable = MOD_HDCP_DISPLAY_DISABLE_AUTHENTICATION;
+		hdcp_w->encryption_status[conn_index] = MOD_HDCP_ENCRYPTION_STATUS_HDCP_OFF;
+		cancel_delayed_work(&hdcp_w->property_validate_dwork);
+	}
+
+	mod_hdcp_update_display(&hdcp_w->hdcp, conn_index, &link_adjust, &display_adjust, &hdcp_w->output);
+
+	process_output(hdcp_w);
+}
+
+static void hdcp_remove_display(struct hdcp_workqueue *hdcp_work,
+				unsigned int link_index,
+			 struct amdgpu_dm_connector *aconnector)
+{
+	struct hdcp_workqueue *hdcp_w = &hdcp_work[link_index];
+	struct drm_connector_state *conn_state = aconnector->base.state;
+	unsigned int conn_index = aconnector->base.index;
+
+	guard(mutex)(&hdcp_w->mutex);
+
+	/* the removal of display will invoke auth reset -> hdcp destroy and
+	 * we'd expect the Content Protection (CP) property changed back to
+	 * DESIRED if at the time ENABLED. CP property change should occur
+	 * before the element removed from linked list.
+	 */
+	if (conn_state && conn_state->content_protection == DRM_MODE_CONTENT_PROTECTION_ENABLED) {
+		conn_state->content_protection = DRM_MODE_CONTENT_PROTECTION_DESIRED;
+
+		DRM_DEBUG_DRIVER("[HDCP_DM] display %d, CP 2 -> 1, type %u, DPMS %u\n",
+				 aconnector->base.index, conn_state->hdcp_content_type,
+				 aconnector->base.dpms);
+	}
+
+	mod_hdcp_remove_display(&hdcp_w->hdcp, aconnector->base.index, &hdcp_w->output);
+	if (hdcp_w->aconnector[conn_index]) {
+		drm_connector_put(&hdcp_w->aconnector[conn_index]->base);
+		hdcp_w->aconnector[conn_index] = NULL;
+	}
+	process_output(hdcp_w);
+}
+
+void hdcp_reset_display(struct hdcp_workqueue *hdcp_work, unsigned int link_index)
+{
+	struct hdcp_workqueue *hdcp_w = &hdcp_work[link_index];
+	unsigned int conn_index;
+
+	guard(mutex)(&hdcp_w->mutex);
+
+	mod_hdcp_reset_connection(&hdcp_w->hdcp,  &hdcp_w->output);
+
+	cancel_delayed_work(&hdcp_w->property_validate_dwork);
+
+	for (conn_index = 0; conn_index < AMDGPU_DM_MAX_DISPLAY_INDEX; conn_index++) {
+		hdcp_w->encryption_status[conn_index] =
+			MOD_HDCP_ENCRYPTION_STATUS_HDCP_OFF;
+		if (hdcp_w->aconnector[conn_index]) {
+			drm_connector_put(&hdcp_w->aconnector[conn_index]->base);
+			hdcp_w->aconnector[conn_index] = NULL;
+		}
+	}
+
+	process_output(hdcp_w);
+}
+
+void hdcp_handle_cpirq(struct hdcp_workqueue *hdcp_work, unsigned int link_index)
+{
+	struct hdcp_workqueue *hdcp_w = &hdcp_work[link_index];
+
+	schedule_work(&hdcp_w->cpirq_work);
+}
+
+static void event_callback(struct work_struct *work)
+{
+	struct hdcp_workqueue *hdcp_work;
+
+	hdcp_work = container_of(to_delayed_work(work), struct hdcp_workqueue,
+				 callback_dwork);
+
+	guard(mutex)(&hdcp_work->mutex);
+
+	cancel_delayed_work(&hdcp_work->callback_dwork);
+
+	mod_hdcp_process_event(&hdcp_work->hdcp, MOD_HDCP_EVENT_CALLBACK,
+			       &hdcp_work->output);
+
+	process_output(hdcp_work);
+}
+
+static void event_property_update(struct work_struct *work)
+{
+	struct hdcp_workqueue *hdcp_work = container_of(work, struct hdcp_workqueue,
+							property_update_work);
+	struct amdgpu_dm_connector *aconnector = NULL;
+	struct drm_device *dev;
+	long ret;
+	unsigned int conn_index;
+	struct drm_connector *connector;
+	struct drm_connector_state *conn_state;
+
+	for (conn_index = 0; conn_index < AMDGPU_DM_MAX_DISPLAY_INDEX; conn_index++) {
+		aconnector = hdcp_work->aconnector[conn_index];
+
+		if (!aconnector)
+			continue;
+
+		connector = &aconnector->base;
+
+		/* check if display connected */
+		if (connector->status != connector_status_connected)
+			continue;
+
+		conn_state = aconnector->base.state;
+
+		if (!conn_state)
+			continue;
+
+		dev = connector->dev;
+
+		if (!dev)
+			continue;
+
+		drm_modeset_lock(&dev->mode_config.connection_mutex, NULL);
+		guard(mutex)(&hdcp_work->mutex);
+
+		if (conn_state->commit) {
+			ret = wait_for_completion_interruptible_timeout(&conn_state->commit->hw_done,
+									10 * HZ);
+			if (ret == 0) {
+				DRM_ERROR("HDCP state unknown! Setting it to DESIRED\n");
+				hdcp_work->encryption_status[conn_index] =
+					MOD_HDCP_ENCRYPTION_STATUS_HDCP_OFF;
+			}
+		}
+		if (hdcp_work->encryption_status[conn_index] !=
+			MOD_HDCP_ENCRYPTION_STATUS_HDCP_OFF) {
+			if (conn_state->hdcp_content_type ==
+				DRM_MODE_HDCP_CONTENT_TYPE0 &&
+				hdcp_work->encryption_status[conn_index] <=
+				MOD_HDCP_ENCRYPTION_STATUS_HDCP2_TYPE0_ON) {
+				DRM_DEBUG_DRIVER("[HDCP_DM] DRM_MODE_CONTENT_PROTECTION_ENABLED\n");
+				drm_hdcp_update_content_protection(connector,
+								   DRM_MODE_CONTENT_PROTECTION_ENABLED);
+			} else if (conn_state->hdcp_content_type ==
+					DRM_MODE_HDCP_CONTENT_TYPE1 &&
+					hdcp_work->encryption_status[conn_index] ==
+					MOD_HDCP_ENCRYPTION_STATUS_HDCP2_TYPE1_ON) {
+				drm_hdcp_update_content_protection(connector,
+								   DRM_MODE_CONTENT_PROTECTION_ENABLED);
+			}
+		} else {
+			DRM_DEBUG_DRIVER("[HDCP_DM] DRM_MODE_CONTENT_PROTECTION_DESIRED\n");
+			drm_hdcp_update_content_protection(connector,
+							   DRM_MODE_CONTENT_PROTECTION_DESIRED);
+		}
+		drm_modeset_unlock(&dev->mode_config.connection_mutex);
+	}
+}
+
+static void event_property_validate(struct work_struct *work)
+{
+	struct hdcp_workqueue *hdcp_work =
+		container_of(to_delayed_work(work), struct hdcp_workqueue, property_validate_dwork);
+	struct mod_hdcp_display_query query;
+	struct amdgpu_dm_connector *aconnector;
+	unsigned int conn_index;
+
+	guard(mutex)(&hdcp_work->mutex);
+
+	for (conn_index = 0; conn_index < AMDGPU_DM_MAX_DISPLAY_INDEX;
+	     conn_index++) {
+		aconnector = hdcp_work->aconnector[conn_index];
+
+		if (!aconnector)
+			continue;
+
+		/* check if display connected */
+		if (aconnector->base.status != connector_status_connected)
+			continue;
+
+		if (!aconnector->base.state)
+			continue;
+
+		query.encryption_status = MOD_HDCP_ENCRYPTION_STATUS_HDCP_OFF;
+		mod_hdcp_query_display(&hdcp_work->hdcp, aconnector->base.index,
+				       &query);
+
+		DRM_DEBUG_DRIVER("[HDCP_DM] disp %d, connector->CP %u, (query, work): (%d, %d)\n",
+				 aconnector->base.index,
+			aconnector->base.state->content_protection,
+			query.encryption_status,
+			hdcp_work->encryption_status[conn_index]);
+
+		if (query.encryption_status !=
+		    hdcp_work->encryption_status[conn_index]) {
+			DRM_DEBUG_DRIVER("[HDCP_DM] encryption_status change from %x to %x\n",
+					 hdcp_work->encryption_status[conn_index],
+					 query.encryption_status);
+
+			hdcp_work->encryption_status[conn_index] =
+				query.encryption_status;
+
+			DRM_DEBUG_DRIVER("[HDCP_DM] trigger property_update_work\n");
+
+			schedule_work(&hdcp_work->property_update_work);
+		}
+	}
+}
+
+static void event_watchdog_timer(struct work_struct *work)
+{
+	struct hdcp_workqueue *hdcp_work;
+
+	hdcp_work = container_of(to_delayed_work(work),
+				 struct hdcp_workqueue,
+				      watchdog_timer_dwork);
+
+	guard(mutex)(&hdcp_work->mutex);
+
+	cancel_delayed_work(&hdcp_work->watchdog_timer_dwork);
+
+	mod_hdcp_process_event(&hdcp_work->hdcp,
+			       MOD_HDCP_EVENT_WATCHDOG_TIMEOUT,
+			       &hdcp_work->output);
+
+	process_output(hdcp_work);
+}
+
+static void event_cpirq(struct work_struct *work)
+{
+	struct hdcp_workqueue *hdcp_work;
+
+	hdcp_work = container_of(work, struct hdcp_workqueue, cpirq_work);
+
+	guard(mutex)(&hdcp_work->mutex);
+
+	mod_hdcp_process_event(&hdcp_work->hdcp, MOD_HDCP_EVENT_CPIRQ, &hdcp_work->output);
+
+	process_output(hdcp_work);
+}
+
+void hdcp_destroy(struct kobject *kobj, struct hdcp_workqueue *hdcp_work)
+{
+	int i = 0;
+
+	for (i = 0; i < hdcp_work->max_link; i++) {
+		cancel_delayed_work_sync(&hdcp_work[i].callback_dwork);
+		cancel_delayed_work_sync(&hdcp_work[i].watchdog_timer_dwork);
+		cancel_delayed_work_sync(&hdcp_work[i].property_validate_dwork);
+	}
+
+	sysfs_remove_bin_file(kobj, &hdcp_work[0].attr);
+	kfree(hdcp_work->srm);
+	kfree(hdcp_work->srm_temp);
+	kfree(hdcp_work);
+}
+
+static bool enable_assr(void *handle, struct dc_link *link)
+{
+	struct hdcp_workqueue *hdcp_work = handle;
+	struct mod_hdcp hdcp = hdcp_work->hdcp;
+	struct psp_context *psp = hdcp.config.psp.handle;
+	struct ta_dtm_shared_memory *dtm_cmd;
+
+	if (!psp->dtm_context.context.initialized) {
+		DRM_INFO("Failed to enable ASSR, DTM TA is not initialized.");
+		return false;
+	}
+
+	dtm_cmd = (struct ta_dtm_shared_memory *)psp->dtm_context.context.mem_context.shared_buf;
+
+	guard(mutex)(&psp->dtm_context.mutex);
+	memset(dtm_cmd, 0, sizeof(struct ta_dtm_shared_memory));
+
+	dtm_cmd->cmd_id = TA_DTM_COMMAND__TOPOLOGY_ASSR_ENABLE;
+	dtm_cmd->dtm_in_message.topology_assr_enable.display_topology_dig_be_index =
+		link->link_enc_hw_inst;
+	dtm_cmd->dtm_status = TA_DTM_STATUS__GENERIC_FAILURE;
+
+	psp_dtm_invoke(psp, dtm_cmd->cmd_id);
+
+	if (dtm_cmd->dtm_status != TA_DTM_STATUS__SUCCESS) {
+		DRM_INFO("Failed to enable ASSR");
+		return false;
+	}
+
+	return true;
+}
+
+static void update_config(void *handle, struct cp_psp_stream_config *config)
+{
+	struct hdcp_workqueue *hdcp_work = handle;
+	struct amdgpu_dm_connector *aconnector = config->dm_stream_ctx;
+	int link_index = aconnector->dc_link->link_index;
+	unsigned int conn_index = aconnector->base.index;
+	struct mod_hdcp_display *display = &hdcp_work[link_index].display;
+	struct mod_hdcp_link *link = &hdcp_work[link_index].link;
+	struct hdcp_workqueue *hdcp_w = &hdcp_work[link_index];
+	struct dc_sink *sink = NULL;
+	bool link_is_hdcp14 = false;
+
+	if (config->dpms_off) {
+		hdcp_remove_display(hdcp_work, link_index, aconnector);
+		return;
+	}
+
+	memset(display, 0, sizeof(*display));
+	memset(link, 0, sizeof(*link));
+
+	display->index = aconnector->base.index;
+	display->state = MOD_HDCP_DISPLAY_ACTIVE;
+
+	if (aconnector->dc_sink)
+		sink = aconnector->dc_sink;
+	else if (aconnector->dc_em_sink)
+		sink = aconnector->dc_em_sink;
+
+	if (sink)
+		link->mode = mod_hdcp_signal_type_to_operation_mode(sink->sink_signal);
+
+	display->controller = CONTROLLER_ID_D0 + config->otg_inst;
+	display->dig_fe = config->dig_fe;
+	link->dig_be = config->dig_be;
+	link->ddc_line = aconnector->dc_link->ddc_hw_inst + 1;
+	display->stream_enc_idx = config->stream_enc_idx;
+	link->link_enc_idx = config->link_enc_idx;
+	link->dio_output_id = config->dio_output_idx;
+	link->phy_idx = config->phy_idx;
+
+	if (sink)
+		link_is_hdcp14 = dc_link_is_hdcp14(aconnector->dc_link, sink->sink_signal);
+	link->hdcp_supported_informational = link_is_hdcp14;
+	link->dp.rev = aconnector->dc_link->dpcd_caps.dpcd_rev.raw;
+	link->dp.assr_enabled = config->assr_enabled;
+	link->dp.mst_enabled = config->mst_enabled;
+	link->dp.dp2_enabled = config->dp2_enabled;
+	link->dp.usb4_enabled = config->usb4_enabled;
+	display->adjust.disable = MOD_HDCP_DISPLAY_DISABLE_AUTHENTICATION;
+	link->adjust.auth_delay = 2;
+	link->adjust.retry_limit = MAX_NUM_OF_ATTEMPTS;
+	link->adjust.hdcp1.disable = 0;
+	hdcp_w->encryption_status[display->index] = MOD_HDCP_ENCRYPTION_STATUS_HDCP_OFF;
+
+	DRM_DEBUG_DRIVER("[HDCP_DM] display %d, CP %d, type %d\n", aconnector->base.index,
+			 (!!aconnector->base.state) ?
+			 aconnector->base.state->content_protection : -1,
+			 (!!aconnector->base.state) ?
+			 aconnector->base.state->hdcp_content_type : -1);
+
+	guard(mutex)(&hdcp_w->mutex);
+
+	mod_hdcp_add_display(&hdcp_w->hdcp, link, display, &hdcp_w->output);
+	drm_connector_get(&aconnector->base);
+	if (hdcp_w->aconnector[conn_index])
+		drm_connector_put(&hdcp_w->aconnector[conn_index]->base);
+	hdcp_w->aconnector[conn_index] = aconnector;
+	process_output(hdcp_w);
+}
+
+/**
+ * DOC: Add sysfs interface for set/get srm
+ *
+ * NOTE: From the usermodes prospective you only need to call write *ONCE*, the kernel
+ *      will automatically call once or twice depending on the size
+ *
+ * call: "cat file > /sys/class/drm/card0/device/hdcp_srm" from usermode no matter what the size is
+ *
+ * The kernel can only send PAGE_SIZE at once and since MAX_SRM_FILE(5120) > PAGE_SIZE(4096),
+ * srm_data_write can be called multiple times.
+ *
+ * sysfs interface doesn't tell us the size we will get so we are sending partial SRMs to psp and on
+ * the last call we will send the full SRM. PSP will fail on every call before the last.
+ *
+ * This means we don't know if the SRM is good until the last call. And because of this
+ * limitation we cannot throw errors early as it will stop the kernel from writing to sysfs
+ *
+ * Example 1:
+ *	Good SRM size = 5096
+ *	first call to write 4096 -> PSP fails
+ *	Second call to write 1000 -> PSP Pass -> SRM is set
+ *
+ * Example 2:
+ *	Bad SRM size = 4096
+ *	first call to write 4096 -> PSP fails (This is the same as above, but we don't know if this
+ *	is the last call)
+ *
+ * Solution?:
+ *	1: Parse the SRM? -> It is signed so we don't know the EOF
+ *	2: We can have another sysfs that passes the size before calling set. -> simpler solution
+ *	below
+ *
+ * Easy Solution:
+ * Always call get after Set to verify if set was successful.
+ * +----------------------+
+ * |   Why it works:      |
+ * +----------------------+
+ * PSP will only update its srm if its older than the one we are trying to load.
+ * Always do set first than get.
+ *	-if we try to "1. SET" a older version PSP will reject it and we can "2. GET" the newer
+ *	version and save it
+ *
+ *	-if we try to "1. SET" a newer version PSP will accept it and we can "2. GET" the
+ *	same(newer) version back and save it
+ *
+ *	-if we try to "1. SET" a newer version and PSP rejects it. That means the format is
+ *	incorrect/corrupted and we should correct our SRM by getting it from PSP
+ */
+static ssize_t srm_data_write(struct file *filp, struct kobject *kobj,
+			      const struct bin_attribute *bin_attr, char *buffer,
+			      loff_t pos, size_t count)
+{
+	struct hdcp_workqueue *work;
+	u32 srm_version = 0;
+
+	work = container_of(bin_attr, struct hdcp_workqueue, attr);
+	link_lock(work, true);
+
+	memcpy(work->srm_temp + pos, buffer, count);
+
+	if (!psp_set_srm(work->hdcp.config.psp.handle, work->srm_temp, pos + count, &srm_version)) {
+		DRM_DEBUG_DRIVER("HDCP SRM SET version 0x%X", srm_version);
+		memcpy(work->srm, work->srm_temp, pos + count);
+		work->srm_size = pos + count;
+		work->srm_version = srm_version;
+	}
+
+	link_lock(work, false);
+
+	return count;
+}
+
+static ssize_t srm_data_read(struct file *filp, struct kobject *kobj,
+			     const struct bin_attribute *bin_attr, char *buffer,
+			     loff_t pos, size_t count)
+{
+	struct hdcp_workqueue *work;
+	u8 *srm = NULL;
+	u32 srm_version;
+	u32 srm_size;
+	size_t ret = count;
+
+	work = container_of(bin_attr, struct hdcp_workqueue, attr);
+
+	link_lock(work, true);
+
+	srm = psp_get_srm(work->hdcp.config.psp.handle, &srm_version, &srm_size);
+
+	if (!srm) {
+		ret = -EINVAL;
+		goto ret;
+	}
+
+	if (pos >= srm_size)
+		ret = 0;
+
+	if (srm_size - pos < count) {
+		memcpy(buffer, srm + pos, srm_size - pos);
+		ret = srm_size - pos;
+		goto ret;
+	}
+
+	memcpy(buffer, srm + pos, count);
+
+ret:
+	link_lock(work, false);
+	return ret;
+}
+
+/* From the hdcp spec (5.Renewability) SRM needs to be stored in a non-volatile memory.
+ *
+ * For example,
+ *	if Application "A" sets the SRM (ver 2) and we reboot/suspend and later when Application "B"
+ *	needs to use HDCP, the version in PSP should be SRM(ver 2). So SRM should be persistent
+ *	across boot/reboots/suspend/resume/shutdown
+ *
+ * Currently when the system goes down (suspend/shutdown) the SRM is cleared from PSP. For HDCP
+ * we need to make the SRM persistent.
+ *
+ * -PSP owns the checking of SRM but doesn't have the ability to store it in a non-volatile memory.
+ * -The kernel cannot write to the file systems.
+ * -So we need usermode to do this for us, which is why an interface for usermode is needed
+ *
+ *
+ *
+ * Usermode can read/write to/from PSP using the sysfs interface
+ * For example:
+ *	to save SRM from PSP to storage : cat /sys/class/drm/card0/device/hdcp_srm > srmfile
+ *	to load from storage to PSP: cat srmfile > /sys/class/drm/card0/device/hdcp_srm
+ */
+static const struct bin_attribute data_attr = {
+	.attr = {.name = "hdcp_srm", .mode = 0664},
+	.size = PSP_HDCP_SRM_FIRST_GEN_MAX_SIZE, /* Limit SRM size */
+	.write = srm_data_write,
+	.read = srm_data_read,
+};
+
+struct hdcp_workqueue *hdcp_create_workqueue(struct amdgpu_device *adev,
+					     struct cp_psp *cp_psp, struct dc *dc)
+{
+	int max_caps = dc->caps.max_links;
+	struct hdcp_workqueue *hdcp_work;
+	int i = 0;
+
+	hdcp_work = kcalloc(max_caps, sizeof(*hdcp_work), GFP_KERNEL);
+	if (ZERO_OR_NULL_PTR(hdcp_work))
+		return NULL;
+
+	hdcp_work->srm = kcalloc(PSP_HDCP_SRM_FIRST_GEN_MAX_SIZE,
+				 sizeof(*hdcp_work->srm), GFP_KERNEL);
+
+	if (!hdcp_work->srm)
+		goto fail_alloc_context;
+
+	hdcp_work->srm_temp = kcalloc(PSP_HDCP_SRM_FIRST_GEN_MAX_SIZE,
+				      sizeof(*hdcp_work->srm_temp), GFP_KERNEL);
+
+	if (!hdcp_work->srm_temp)
+		goto fail_alloc_context;
+
+	hdcp_work->max_link = max_caps;
+
+	for (i = 0; i < max_caps; i++) {
+		mutex_init(&hdcp_work[i].mutex);
+
+		INIT_WORK(&hdcp_work[i].cpirq_work, event_cpirq);
+		INIT_WORK(&hdcp_work[i].property_update_work, event_property_update);
+		INIT_DELAYED_WORK(&hdcp_work[i].callback_dwork, event_callback);
+		INIT_DELAYED_WORK(&hdcp_work[i].watchdog_timer_dwork, event_watchdog_timer);
+		INIT_DELAYED_WORK(&hdcp_work[i].property_validate_dwork, event_property_validate);
+
+		struct mod_hdcp_config *config = &hdcp_work[i].hdcp.config;
+		struct mod_hdcp_ddc_funcs *ddc_funcs = &config->ddc.funcs;
+
+		config->psp.handle = &adev->psp;
+		if (dc->ctx->dce_version == DCN_VERSION_3_1  ||
+		    dc->ctx->dce_version == DCN_VERSION_3_14 ||
+		    dc->ctx->dce_version == DCN_VERSION_3_15 ||
+		    dc->ctx->dce_version == DCN_VERSION_3_16 ||
+		    dc->ctx->dce_version == DCN_VERSION_3_2  ||
+		    dc->ctx->dce_version == DCN_VERSION_3_21 ||
+		    dc->ctx->dce_version == DCN_VERSION_3_5  ||
+		    dc->ctx->dce_version == DCN_VERSION_3_51 ||
+		    dc->ctx->dce_version == DCN_VERSION_3_6  ||
+		    dc->ctx->dce_version == DCN_VERSION_4_01)
+			config->psp.caps.dtm_v3_supported = 1;
+
+		config->ddc.handle = dc_get_link_at_index(dc, i);
+
+		ddc_funcs->write_i2c = lp_write_i2c;
+		ddc_funcs->read_i2c = lp_read_i2c;
+		ddc_funcs->write_dpcd = lp_write_dpcd;
+		ddc_funcs->read_dpcd = lp_read_dpcd;
+
+		config->debug.lc_enable_sw_fallback = dc->debug.hdcp_lc_enable_sw_fallback;
+		if (dc->caps.fused_io_supported || dc->debug.hdcp_lc_force_fw_enable) {
+			ddc_funcs->atomic_write_poll_read_i2c = lp_atomic_write_poll_read_i2c;
+			ddc_funcs->atomic_write_poll_read_aux = lp_atomic_write_poll_read_aux;
+		} else {
+			ddc_funcs->atomic_write_poll_read_i2c = NULL;
+			ddc_funcs->atomic_write_poll_read_aux = NULL;
+		}
+
+		memset(hdcp_work[i].aconnector, 0,
+		       sizeof(struct amdgpu_dm_connector *) *
+			       AMDGPU_DM_MAX_DISPLAY_INDEX);
+		memset(hdcp_work[i].encryption_status, 0,
+		       sizeof(enum mod_hdcp_encryption_status) *
+			       AMDGPU_DM_MAX_DISPLAY_INDEX);
+	}
+
+	cp_psp->funcs.update_stream_config = update_config;
+	cp_psp->funcs.enable_assr = enable_assr;
+	cp_psp->handle = hdcp_work;
+
+	/* File created at /sys/class/drm/card0/device/hdcp_srm*/
+	hdcp_work[0].attr = data_attr;
+	sysfs_bin_attr_init(&hdcp_work[0].attr);
+
+	if (sysfs_create_bin_file(&adev->dev->kobj, &hdcp_work[0].attr))
+		DRM_WARN("Failed to create device file hdcp_srm");
+
+	return hdcp_work;
+
+fail_alloc_context:
+	kfree(hdcp_work->srm);
+	kfree(hdcp_work->srm_temp);
+	kfree(hdcp_work);
+
+	return NULL;
+}
+
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.h
new file mode 100644
index 000000000000..4faa344f196e
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.h
@@ -0,0 +1,90 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef AMDGPU_DM_AMDGPU_DM_HDCP_H_
+#define AMDGPU_DM_AMDGPU_DM_HDCP_H_
+
+#include "mod_hdcp.h"
+#include "hdcp.h"
+#include "dc.h"
+#include "dm_cp_psp.h"
+#include "amdgpu.h"
+
+struct mod_hdcp;
+struct mod_hdcp_link;
+struct mod_hdcp_display;
+struct cp_psp;
+
+struct hdcp_workqueue {
+	struct work_struct cpirq_work;
+	struct work_struct property_update_work;
+	struct delayed_work callback_dwork;
+	struct delayed_work watchdog_timer_dwork;
+	struct delayed_work property_validate_dwork;
+	struct amdgpu_dm_connector *aconnector[AMDGPU_DM_MAX_DISPLAY_INDEX];
+	struct mutex mutex;
+
+	struct mod_hdcp hdcp;
+	struct mod_hdcp_output output;
+	struct mod_hdcp_display display;
+	struct mod_hdcp_link link;
+
+	enum mod_hdcp_encryption_status encryption_status[AMDGPU_DM_MAX_DISPLAY_INDEX];
+	/* when display is unplugged from mst hub, connctor will be
+	 * destroyed within dm_dp_mst_connector_destroy. connector
+	 * hdcp perperties, like type, undesired, desired, enabled,
+	 * will be lost. So, save hdcp properties into hdcp_work within
+	 * amdgpu_dm_atomic_commit_tail. if the same display is
+	 * plugged back with same display index, its hdcp properties
+	 * will be retrieved from hdcp_work within dm_dp_mst_get_modes
+	 */
+	/* un-desired, desired, enabled */
+	unsigned int content_protection[AMDGPU_DM_MAX_DISPLAY_INDEX];
+	/* hdcp1.x, hdcp2.x */
+	unsigned int hdcp_content_type[AMDGPU_DM_MAX_DISPLAY_INDEX];
+
+	uint8_t max_link;
+
+	uint8_t *srm;
+	uint8_t *srm_temp;
+	uint32_t srm_version;
+	uint32_t srm_size;
+	struct bin_attribute attr;
+};
+
+void hdcp_update_display(struct hdcp_workqueue *hdcp_work,
+			 unsigned int link_index,
+			 struct amdgpu_dm_connector *aconnector,
+			 uint8_t content_type,
+			 bool enable_encryption);
+
+void hdcp_reset_display(struct hdcp_workqueue *work, unsigned int link_index);
+void hdcp_handle_cpirq(struct hdcp_workqueue *work, unsigned int link_index);
+void hdcp_destroy(struct kobject *kobj, struct hdcp_workqueue *work);
+
+struct hdcp_workqueue *hdcp_create_workqueue(struct amdgpu_device *adev, struct cp_psp *cp_psp, struct dc *dc);
+
+#endif /* AMDGPU_DM_AMDGPU_DM_HDCP_H_ */
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c
new file mode 100644
index 000000000000..fe100e4c9801
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c
@@ -0,0 +1,1414 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include <acpi/video.h>
+
+#include <linux/string.h>
+#include <linux/acpi.h>
+#include <linux/i2c.h>
+
+#include <drm/drm_atomic.h>
+#include <drm/drm_probe_helper.h>
+#include <drm/amdgpu_drm.h>
+#include <drm/drm_edid.h>
+#include <drm/drm_fixed.h>
+
+#include "dm_services.h"
+#include "amdgpu.h"
+#include "dc.h"
+#include "amdgpu_dm.h"
+#include "amdgpu_dm_irq.h"
+#include "amdgpu_dm_mst_types.h"
+#include "dpcd_defs.h"
+#include "dc/inc/core_types.h"
+
+#include "dm_helpers.h"
+#include "ddc_service_types.h"
+#include "clk_mgr.h"
+
+static u32 edid_extract_panel_id(struct edid *edid)
+{
+	return (u32)edid->mfg_id[0] << 24   |
+	       (u32)edid->mfg_id[1] << 16   |
+	       (u32)EDID_PRODUCT_ID(edid);
+}
+
+static void apply_edid_quirks(struct drm_device *dev, struct edid *edid, struct dc_edid_caps *edid_caps)
+{
+	uint32_t panel_id = edid_extract_panel_id(edid);
+
+	switch (panel_id) {
+	/* Workaround for monitors that need a delay after detecting the link */
+	case drm_edid_encode_panel_id('G', 'B', 'T', 0x3215):
+		drm_dbg_driver(dev, "Add 10s delay for link detection for panel id %X\n", panel_id);
+		edid_caps->panel_patch.wait_after_dpcd_poweroff_ms = 10000;
+		break;
+	/* Workaround for some monitors which does not work well with FAMS */
+	case drm_edid_encode_panel_id('S', 'A', 'M', 0x0E5E):
+	case drm_edid_encode_panel_id('S', 'A', 'M', 0x7053):
+	case drm_edid_encode_panel_id('S', 'A', 'M', 0x71AC):
+		drm_dbg_driver(dev, "Disabling FAMS on monitor with panel id %X\n", panel_id);
+		edid_caps->panel_patch.disable_fams = true;
+		break;
+	/* Workaround for some monitors that do not clear DPCD 0x317 if FreeSync is unsupported */
+	case drm_edid_encode_panel_id('A', 'U', 'O', 0xA7AB):
+	case drm_edid_encode_panel_id('A', 'U', 'O', 0xE69B):
+	case drm_edid_encode_panel_id('B', 'O', 'E', 0x092A):
+	case drm_edid_encode_panel_id('L', 'G', 'D', 0x06D1):
+	case drm_edid_encode_panel_id('M', 'S', 'F', 0x1003):
+		drm_dbg_driver(dev, "Clearing DPCD 0x317 on monitor with panel id %X\n", panel_id);
+		edid_caps->panel_patch.remove_sink_ext_caps = true;
+		break;
+	case drm_edid_encode_panel_id('S', 'D', 'C', 0x4154):
+		drm_dbg_driver(dev, "Disabling VSC on monitor with panel id %X\n", panel_id);
+		edid_caps->panel_patch.disable_colorimetry = true;
+		break;
+	default:
+		return;
+	}
+}
+
+/**
+ * dm_helpers_parse_edid_caps() - Parse edid caps
+ *
+ * @link: current detected link
+ * @edid:	[in] pointer to edid
+ * @edid_caps:	[in] pointer to edid caps
+ *
+ * Return: void
+ */
+enum dc_edid_status dm_helpers_parse_edid_caps(
+		struct dc_link *link,
+		const struct dc_edid *edid,
+		struct dc_edid_caps *edid_caps)
+{
+	struct amdgpu_dm_connector *aconnector = link->priv;
+	struct drm_connector *connector = &aconnector->base;
+	struct drm_device *dev = connector->dev;
+	struct edid *edid_buf = edid ? (struct edid *) edid->raw_edid : NULL;
+	struct cea_sad *sads;
+	int sad_count = -1;
+	int sadb_count = -1;
+	int i = 0;
+	uint8_t *sadb = NULL;
+
+	enum dc_edid_status result = EDID_OK;
+
+	if (!edid_caps || !edid)
+		return EDID_BAD_INPUT;
+
+	if (!drm_edid_is_valid(edid_buf))
+		result = EDID_BAD_CHECKSUM;
+
+	edid_caps->manufacturer_id = (uint16_t) edid_buf->mfg_id[0] |
+					((uint16_t) edid_buf->mfg_id[1])<<8;
+	edid_caps->product_id = (uint16_t) edid_buf->prod_code[0] |
+					((uint16_t) edid_buf->prod_code[1])<<8;
+	edid_caps->serial_number = edid_buf->serial;
+	edid_caps->manufacture_week = edid_buf->mfg_week;
+	edid_caps->manufacture_year = edid_buf->mfg_year;
+
+	drm_edid_get_monitor_name(edid_buf,
+				  edid_caps->display_name,
+				  AUDIO_INFO_DISPLAY_NAME_SIZE_IN_CHARS);
+
+	edid_caps->edid_hdmi = connector->display_info.is_hdmi;
+
+	apply_edid_quirks(dev, edid_buf, edid_caps);
+
+	sad_count = drm_edid_to_sad((struct edid *) edid->raw_edid, &sads);
+	if (sad_count <= 0)
+		return result;
+
+	edid_caps->audio_mode_count = min(sad_count, DC_MAX_AUDIO_DESC_COUNT);
+	for (i = 0; i < edid_caps->audio_mode_count; ++i) {
+		struct cea_sad *sad = &sads[i];
+
+		edid_caps->audio_modes[i].format_code = sad->format;
+		edid_caps->audio_modes[i].channel_count = sad->channels + 1;
+		edid_caps->audio_modes[i].sample_rate = sad->freq;
+		edid_caps->audio_modes[i].sample_size = sad->byte2;
+	}
+
+	sadb_count = drm_edid_to_speaker_allocation((struct edid *) edid->raw_edid, &sadb);
+
+	if (sadb_count < 0) {
+		DRM_ERROR("Couldn't read Speaker Allocation Data Block: %d\n", sadb_count);
+		sadb_count = 0;
+	}
+
+	if (sadb_count)
+		edid_caps->speaker_flags = sadb[0];
+	else
+		edid_caps->speaker_flags = DEFAULT_SPEAKER_LOCATION;
+
+	kfree(sads);
+	kfree(sadb);
+
+	return result;
+}
+
+static void
+fill_dc_mst_payload_table_from_drm(struct dc_link *link,
+				   bool enable,
+				   struct drm_dp_mst_atomic_payload *target_payload,
+				   struct dc_dp_mst_stream_allocation_table *table)
+{
+	struct dc_dp_mst_stream_allocation_table new_table = { 0 };
+	struct dc_dp_mst_stream_allocation *sa;
+	struct link_mst_stream_allocation_table copy_of_link_table =
+										link->mst_stream_alloc_table;
+
+	int i;
+	int current_hw_table_stream_cnt = copy_of_link_table.stream_count;
+	struct link_mst_stream_allocation *dc_alloc;
+
+	/* TODO: refactor to set link->mst_stream_alloc_table directly if possible.*/
+	if (enable) {
+		dc_alloc =
+		&copy_of_link_table.stream_allocations[current_hw_table_stream_cnt];
+		dc_alloc->vcp_id = target_payload->vcpi;
+		dc_alloc->slot_count = target_payload->time_slots;
+	} else {
+		for (i = 0; i < copy_of_link_table.stream_count; i++) {
+			dc_alloc =
+			&copy_of_link_table.stream_allocations[i];
+
+			if (dc_alloc->vcp_id == target_payload->vcpi) {
+				dc_alloc->vcp_id = 0;
+				dc_alloc->slot_count = 0;
+				break;
+			}
+		}
+		ASSERT(i != copy_of_link_table.stream_count);
+	}
+
+	/* Fill payload info*/
+	for (i = 0; i < MAX_CONTROLLER_NUM; i++) {
+		dc_alloc =
+			&copy_of_link_table.stream_allocations[i];
+		if (dc_alloc->vcp_id > 0 && dc_alloc->slot_count > 0) {
+			sa = &new_table.stream_allocations[new_table.stream_count];
+			sa->slot_count = dc_alloc->slot_count;
+			sa->vcp_id = dc_alloc->vcp_id;
+			new_table.stream_count++;
+		}
+	}
+
+	/* Overwrite the old table */
+	*table = new_table;
+}
+
+void dm_helpers_dp_update_branch_info(
+	struct dc_context *ctx,
+	const struct dc_link *link)
+{}
+
+static void dm_helpers_construct_old_payload(
+			struct drm_dp_mst_topology_mgr *mgr,
+			struct drm_dp_mst_topology_state *mst_state,
+			struct drm_dp_mst_atomic_payload *new_payload,
+			struct drm_dp_mst_atomic_payload *old_payload)
+{
+	struct drm_dp_mst_atomic_payload *pos;
+	int pbn_per_slot = dfixed_trunc(mst_state->pbn_div);
+	u8 next_payload_vc_start = mgr->next_start_slot;
+	u8 payload_vc_start = new_payload->vc_start_slot;
+	u8 allocated_time_slots;
+
+	*old_payload = *new_payload;
+
+	/* Set correct time_slots/PBN of old payload.
+	 * other fields (delete & dsc_enabled) in
+	 * struct drm_dp_mst_atomic_payload are don't care fields
+	 * while calling drm_dp_remove_payload_part2()
+	 */
+	list_for_each_entry(pos, &mst_state->payloads, next) {
+		if (pos != new_payload &&
+		    pos->vc_start_slot > payload_vc_start &&
+		    pos->vc_start_slot < next_payload_vc_start)
+			next_payload_vc_start = pos->vc_start_slot;
+	}
+
+	allocated_time_slots = next_payload_vc_start - payload_vc_start;
+
+	old_payload->time_slots = allocated_time_slots;
+	old_payload->pbn = allocated_time_slots * pbn_per_slot;
+}
+
+/*
+ * Writes payload allocation table in immediate downstream device.
+ */
+bool dm_helpers_dp_mst_write_payload_allocation_table(
+		struct dc_context *ctx,
+		const struct dc_stream_state *stream,
+		struct dc_dp_mst_stream_allocation_table *proposed_table,
+		bool enable)
+{
+	struct amdgpu_dm_connector *aconnector;
+	struct drm_dp_mst_topology_state *mst_state;
+	struct drm_dp_mst_atomic_payload *target_payload, *new_payload, old_payload;
+	struct drm_dp_mst_topology_mgr *mst_mgr;
+
+	aconnector = (struct amdgpu_dm_connector *)stream->dm_stream_context;
+	/* Accessing the connector state is required for vcpi_slots allocation
+	 * and directly relies on behaviour in commit check
+	 * that blocks before commit guaranteeing that the state
+	 * is not gonna be swapped while still in use in commit tail
+	 */
+
+	if (!aconnector || !aconnector->mst_root)
+		return false;
+
+	mst_mgr = &aconnector->mst_root->mst_mgr;
+	mst_state = to_drm_dp_mst_topology_state(mst_mgr->base.state);
+	new_payload = drm_atomic_get_mst_payload_state(mst_state, aconnector->mst_output_port);
+
+	if (enable) {
+		target_payload = new_payload;
+
+		/* It's OK for this to fail */
+		drm_dp_add_payload_part1(mst_mgr, mst_state, new_payload);
+	} else {
+		/* construct old payload by VCPI*/
+		dm_helpers_construct_old_payload(mst_mgr, mst_state,
+						 new_payload, &old_payload);
+		target_payload = &old_payload;
+
+		drm_dp_remove_payload_part1(mst_mgr, mst_state, new_payload);
+	}
+
+	/* mst_mgr->->payloads are VC payload notify MST branch using DPCD or
+	 * AUX message. The sequence is slot 1-63 allocated sequence for each
+	 * stream. AMD ASIC stream slot allocation should follow the same
+	 * sequence. copy DRM MST allocation to dc
+	 */
+	fill_dc_mst_payload_table_from_drm(stream->link, enable, target_payload, proposed_table);
+
+	return true;
+}
+
+/*
+ * poll pending down reply
+ */
+void dm_helpers_dp_mst_poll_pending_down_reply(
+	struct dc_context *ctx,
+	const struct dc_link *link)
+{}
+
+/*
+ * Clear payload allocation table before enable MST DP link.
+ */
+void dm_helpers_dp_mst_clear_payload_allocation_table(
+	struct dc_context *ctx,
+	const struct dc_link *link)
+{}
+
+/*
+ * Polls for ACT (allocation change trigger) handled and sends
+ * ALLOCATE_PAYLOAD message.
+ */
+enum act_return_status dm_helpers_dp_mst_poll_for_allocation_change_trigger(
+		struct dc_context *ctx,
+		const struct dc_stream_state *stream)
+{
+	struct amdgpu_dm_connector *aconnector;
+	struct drm_dp_mst_topology_mgr *mst_mgr;
+	int ret;
+
+	aconnector = (struct amdgpu_dm_connector *)stream->dm_stream_context;
+
+	if (!aconnector || !aconnector->mst_root)
+		return ACT_FAILED;
+
+	mst_mgr = &aconnector->mst_root->mst_mgr;
+
+	if (!mst_mgr->mst_state)
+		return ACT_FAILED;
+
+	ret = drm_dp_check_act_status(mst_mgr);
+
+	if (ret)
+		return ACT_FAILED;
+
+	return ACT_SUCCESS;
+}
+
+void dm_helpers_dp_mst_send_payload_allocation(
+		struct dc_context *ctx,
+		const struct dc_stream_state *stream)
+{
+	struct amdgpu_dm_connector *aconnector;
+	struct drm_dp_mst_topology_state *mst_state;
+	struct drm_dp_mst_topology_mgr *mst_mgr;
+	struct drm_dp_mst_atomic_payload *new_payload;
+	enum mst_progress_status set_flag = MST_ALLOCATE_NEW_PAYLOAD;
+	enum mst_progress_status clr_flag = MST_CLEAR_ALLOCATED_PAYLOAD;
+	int ret = 0;
+
+	aconnector = (struct amdgpu_dm_connector *)stream->dm_stream_context;
+
+	if (!aconnector || !aconnector->mst_root)
+		return;
+
+	mst_mgr = &aconnector->mst_root->mst_mgr;
+	mst_state = to_drm_dp_mst_topology_state(mst_mgr->base.state);
+	new_payload = drm_atomic_get_mst_payload_state(mst_state, aconnector->mst_output_port);
+
+	ret = drm_dp_add_payload_part2(mst_mgr, new_payload);
+
+	if (ret) {
+		amdgpu_dm_set_mst_status(&aconnector->mst_status,
+			set_flag, false);
+	} else {
+		amdgpu_dm_set_mst_status(&aconnector->mst_status,
+			set_flag, true);
+		amdgpu_dm_set_mst_status(&aconnector->mst_status,
+			clr_flag, false);
+	}
+}
+
+void dm_helpers_dp_mst_update_mst_mgr_for_deallocation(
+		struct dc_context *ctx,
+		const struct dc_stream_state *stream)
+{
+	struct amdgpu_dm_connector *aconnector;
+	struct drm_dp_mst_topology_state *mst_state;
+	struct drm_dp_mst_topology_mgr *mst_mgr;
+	struct drm_dp_mst_atomic_payload *new_payload, old_payload;
+	enum mst_progress_status set_flag = MST_CLEAR_ALLOCATED_PAYLOAD;
+	enum mst_progress_status clr_flag = MST_ALLOCATE_NEW_PAYLOAD;
+
+	aconnector = (struct amdgpu_dm_connector *)stream->dm_stream_context;
+
+	if (!aconnector || !aconnector->mst_root)
+		return;
+
+	mst_mgr = &aconnector->mst_root->mst_mgr;
+	mst_state = to_drm_dp_mst_topology_state(mst_mgr->base.state);
+	new_payload = drm_atomic_get_mst_payload_state(mst_state, aconnector->mst_output_port);
+	dm_helpers_construct_old_payload(mst_mgr, mst_state,
+					 new_payload, &old_payload);
+
+	drm_dp_remove_payload_part2(mst_mgr, mst_state, &old_payload, new_payload);
+
+	amdgpu_dm_set_mst_status(&aconnector->mst_status, set_flag, true);
+	amdgpu_dm_set_mst_status(&aconnector->mst_status, clr_flag, false);
+ }
+
+void dm_dtn_log_begin(struct dc_context *ctx,
+	struct dc_log_buffer_ctx *log_ctx)
+{
+	static const char msg[] = "[dtn begin]\n";
+
+	if (!log_ctx) {
+		pr_info("%s", msg);
+		return;
+	}
+
+	dm_dtn_log_append_v(ctx, log_ctx, "%s", msg);
+}
+
+__printf(3, 4)
+void dm_dtn_log_append_v(struct dc_context *ctx,
+	struct dc_log_buffer_ctx *log_ctx,
+	const char *msg, ...)
+{
+	va_list args;
+	size_t total;
+	int n;
+
+	if (!log_ctx) {
+		/* No context, redirect to dmesg. */
+		struct va_format vaf;
+
+		vaf.fmt = msg;
+		vaf.va = &args;
+
+		va_start(args, msg);
+		pr_info("%pV", &vaf);
+		va_end(args);
+
+		return;
+	}
+
+	/* Measure the output. */
+	va_start(args, msg);
+	n = vsnprintf(NULL, 0, msg, args);
+	va_end(args);
+
+	if (n <= 0)
+		return;
+
+	/* Reallocate the string buffer as needed. */
+	total = log_ctx->pos + n + 1;
+
+	if (total > log_ctx->size) {
+		char *buf = kvcalloc(total, sizeof(char), GFP_KERNEL);
+
+		if (buf) {
+			memcpy(buf, log_ctx->buf, log_ctx->pos);
+			kfree(log_ctx->buf);
+
+			log_ctx->buf = buf;
+			log_ctx->size = total;
+		}
+	}
+
+	if (!log_ctx->buf)
+		return;
+
+	/* Write the formatted string to the log buffer. */
+	va_start(args, msg);
+	n = vscnprintf(
+		log_ctx->buf + log_ctx->pos,
+		log_ctx->size - log_ctx->pos,
+		msg,
+		args);
+	va_end(args);
+
+	if (n > 0)
+		log_ctx->pos += n;
+}
+
+void dm_dtn_log_end(struct dc_context *ctx,
+	struct dc_log_buffer_ctx *log_ctx)
+{
+	static const char msg[] = "[dtn end]\n";
+
+	if (!log_ctx) {
+		pr_info("%s", msg);
+		return;
+	}
+
+	dm_dtn_log_append_v(ctx, log_ctx, "%s", msg);
+}
+
+bool dm_helpers_dp_mst_start_top_mgr(
+		struct dc_context *ctx,
+		const struct dc_link *link,
+		bool boot)
+{
+	struct amdgpu_dm_connector *aconnector = link->priv;
+	int ret;
+
+	if (!aconnector) {
+		DRM_ERROR("Failed to find connector for link!");
+		return false;
+	}
+
+	if (boot) {
+		DRM_INFO("DM_MST: Differing MST start on aconnector: %p [id: %d]\n",
+					aconnector, aconnector->base.base.id);
+		return true;
+	}
+
+	DRM_INFO("DM_MST: starting TM on aconnector: %p [id: %d]\n",
+			aconnector, aconnector->base.base.id);
+
+	ret = drm_dp_mst_topology_mgr_set_mst(&aconnector->mst_mgr, true);
+	if (ret < 0) {
+		DRM_ERROR("DM_MST: Failed to set the device into MST mode!");
+		return false;
+	}
+
+	DRM_INFO("DM_MST: DP%x, %d-lane link detected\n", aconnector->mst_mgr.dpcd[0],
+		aconnector->mst_mgr.dpcd[2] & DP_MAX_LANE_COUNT_MASK);
+
+	return true;
+}
+
+bool dm_helpers_dp_mst_stop_top_mgr(
+		struct dc_context *ctx,
+		struct dc_link *link)
+{
+	struct amdgpu_dm_connector *aconnector = link->priv;
+
+	if (!aconnector) {
+		DRM_ERROR("Failed to find connector for link!");
+		return false;
+	}
+
+	DRM_INFO("DM_MST: stopping TM on aconnector: %p [id: %d]\n",
+			aconnector, aconnector->base.base.id);
+
+	if (aconnector->mst_mgr.mst_state == true) {
+		drm_dp_mst_topology_mgr_set_mst(&aconnector->mst_mgr, false);
+		link->cur_link_settings.lane_count = 0;
+	}
+
+	return false;
+}
+
+bool dm_helpers_dp_read_dpcd(
+		struct dc_context *ctx,
+		const struct dc_link *link,
+		uint32_t address,
+		uint8_t *data,
+		uint32_t size)
+{
+
+	struct amdgpu_dm_connector *aconnector = link->priv;
+
+	if (!aconnector)
+		return false;
+
+	return drm_dp_dpcd_read(&aconnector->dm_dp_aux.aux, address, data,
+				size) == size;
+}
+
+bool dm_helpers_dp_write_dpcd(
+		struct dc_context *ctx,
+		const struct dc_link *link,
+		uint32_t address,
+		const uint8_t *data,
+		uint32_t size)
+{
+	struct amdgpu_dm_connector *aconnector = link->priv;
+
+	if (!aconnector)
+		return false;
+
+	return drm_dp_dpcd_write(&aconnector->dm_dp_aux.aux,
+			address, (uint8_t *)data, size) > 0;
+}
+
+bool dm_helpers_submit_i2c(
+		struct dc_context *ctx,
+		const struct dc_link *link,
+		struct i2c_command *cmd)
+{
+	struct amdgpu_dm_connector *aconnector = link->priv;
+	struct i2c_msg *msgs;
+	int i = 0;
+	int num = cmd->number_of_payloads;
+	bool result;
+
+	if (!aconnector) {
+		DRM_ERROR("Failed to find connector for link!");
+		return false;
+	}
+
+	msgs = kcalloc(num, sizeof(struct i2c_msg), GFP_KERNEL);
+
+	if (!msgs)
+		return false;
+
+	for (i = 0; i < num; i++) {
+		msgs[i].flags = cmd->payloads[i].write ? 0 : I2C_M_RD;
+		msgs[i].addr = cmd->payloads[i].address;
+		msgs[i].len = cmd->payloads[i].length;
+		msgs[i].buf = cmd->payloads[i].data;
+	}
+
+	result = i2c_transfer(&aconnector->i2c->base, msgs, num) == num;
+
+	kfree(msgs);
+
+	return result;
+}
+
+bool dm_helpers_execute_fused_io(
+		struct dc_context *ctx,
+		struct dc_link *link,
+		union dmub_rb_cmd *commands,
+		uint8_t count,
+		uint32_t timeout_us
+)
+{
+	struct amdgpu_device *dev = ctx->driver_context;
+
+	return amdgpu_dm_execute_fused_io(dev, link, commands, count, timeout_us);
+}
+
+static bool execute_synaptics_rc_command(struct drm_dp_aux *aux,
+		bool is_write_cmd,
+		unsigned char cmd,
+		unsigned int length,
+		unsigned int offset,
+		unsigned char *data)
+{
+	bool success = false;
+	unsigned char rc_data[16] = {0};
+	unsigned char rc_offset[4] = {0};
+	unsigned char rc_length[2] = {0};
+	unsigned char rc_cmd = 0;
+	unsigned char rc_result = 0xFF;
+	unsigned char i = 0;
+	int ret;
+
+	if (is_write_cmd) {
+		// write rc data
+		memmove(rc_data, data, length);
+		ret = drm_dp_dpcd_write(aux, SYNAPTICS_RC_DATA, rc_data, sizeof(rc_data));
+		if (ret < 0)
+			goto err;
+	}
+
+	// write rc offset
+	rc_offset[0] = (unsigned char) offset & 0xFF;
+	rc_offset[1] = (unsigned char) (offset >> 8) & 0xFF;
+	rc_offset[2] = (unsigned char) (offset >> 16) & 0xFF;
+	rc_offset[3] = (unsigned char) (offset >> 24) & 0xFF;
+	ret = drm_dp_dpcd_write(aux, SYNAPTICS_RC_OFFSET, rc_offset, sizeof(rc_offset));
+	if (ret < 0)
+		goto err;
+
+	// write rc length
+	rc_length[0] = (unsigned char) length & 0xFF;
+	rc_length[1] = (unsigned char) (length >> 8) & 0xFF;
+	ret = drm_dp_dpcd_write(aux, SYNAPTICS_RC_LENGTH, rc_length, sizeof(rc_length));
+	if (ret < 0)
+		goto err;
+
+	// write rc cmd
+	rc_cmd = cmd | 0x80;
+	ret = drm_dp_dpcd_write(aux, SYNAPTICS_RC_COMMAND, &rc_cmd, sizeof(rc_cmd));
+	if (ret < 0)
+		goto err;
+
+	// poll until active is 0
+	for (i = 0; i < 10; i++) {
+		drm_dp_dpcd_read(aux, SYNAPTICS_RC_COMMAND, &rc_cmd, sizeof(rc_cmd));
+		if (rc_cmd == cmd)
+			// active is 0
+			break;
+		msleep(10);
+	}
+
+	// read rc result
+	drm_dp_dpcd_read(aux, SYNAPTICS_RC_RESULT, &rc_result, sizeof(rc_result));
+	success = (rc_result == 0);
+
+	if (success && !is_write_cmd) {
+		// read rc data
+		drm_dp_dpcd_read(aux, SYNAPTICS_RC_DATA, data, length);
+	}
+
+	drm_dbg_dp(aux->drm_dev, "success = %d\n", success);
+
+	return success;
+
+err:
+	DRM_ERROR("%s: write cmd ..., err = %d\n",  __func__, ret);
+	return false;
+}
+
+static void apply_synaptics_fifo_reset_wa(struct drm_dp_aux *aux)
+{
+	unsigned char data[16] = {0};
+
+	drm_dbg_dp(aux->drm_dev, "Start\n");
+
+	// Step 2
+	data[0] = 'P';
+	data[1] = 'R';
+	data[2] = 'I';
+	data[3] = 'U';
+	data[4] = 'S';
+
+	if (!execute_synaptics_rc_command(aux, true, 0x01, 5, 0, data))
+		return;
+
+	// Step 3 and 4
+	if (!execute_synaptics_rc_command(aux, false, 0x31, 4, 0x220998, data))
+		return;
+
+	data[0] &= (~(1 << 1)); // set bit 1 to 0
+	if (!execute_synaptics_rc_command(aux, true, 0x21, 4, 0x220998, data))
+		return;
+
+	if (!execute_synaptics_rc_command(aux, false, 0x31, 4, 0x220D98, data))
+		return;
+
+	data[0] &= (~(1 << 1)); // set bit 1 to 0
+	if (!execute_synaptics_rc_command(aux, true, 0x21, 4, 0x220D98, data))
+		return;
+
+	if (!execute_synaptics_rc_command(aux, false, 0x31, 4, 0x221198, data))
+		return;
+
+	data[0] &= (~(1 << 1)); // set bit 1 to 0
+	if (!execute_synaptics_rc_command(aux, true, 0x21, 4, 0x221198, data))
+		return;
+
+	// Step 3 and 5
+	if (!execute_synaptics_rc_command(aux, false, 0x31, 4, 0x220998, data))
+		return;
+
+	data[0] |= (1 << 1); // set bit 1 to 1
+	if (!execute_synaptics_rc_command(aux, true, 0x21, 4, 0x220998, data))
+		return;
+
+	if (!execute_synaptics_rc_command(aux, false, 0x31, 4, 0x220D98, data))
+		return;
+
+	data[0] |= (1 << 1); // set bit 1 to 1
+
+	if (!execute_synaptics_rc_command(aux, false, 0x31, 4, 0x221198, data))
+		return;
+
+	data[0] |= (1 << 1); // set bit 1 to 1
+	if (!execute_synaptics_rc_command(aux, true, 0x21, 4, 0x221198, data))
+		return;
+
+	// Step 6
+	if (!execute_synaptics_rc_command(aux, true, 0x02, 0, 0, NULL))
+		return;
+
+	drm_dbg_dp(aux->drm_dev, "Done\n");
+}
+
+/* MST Dock */
+static const uint8_t SYNAPTICS_DEVICE_ID[] = "SYNA";
+
+static uint8_t write_dsc_enable_synaptics_non_virtual_dpcd_mst(
+		struct drm_dp_aux *aux,
+		const struct dc_stream_state *stream,
+		bool enable)
+{
+	uint8_t ret = 0;
+
+	drm_dbg_dp(aux->drm_dev,
+		   "MST_DSC Configure DSC to non-virtual dpcd synaptics\n");
+
+	if (enable) {
+		/* When DSC is enabled on previous boot and reboot with the hub,
+		 * there is a chance that Synaptics hub gets stuck during reboot sequence.
+		 * Applying a workaround to reset Synaptics SDP fifo before enabling the first stream
+		 */
+		if (!stream->link->link_status.link_active &&
+			memcmp(stream->link->dpcd_caps.branch_dev_name,
+				(int8_t *)SYNAPTICS_DEVICE_ID, 4) == 0)
+			apply_synaptics_fifo_reset_wa(aux);
+
+		ret = drm_dp_dpcd_write(aux, DP_DSC_ENABLE, &enable, 1);
+		DRM_INFO("MST_DSC Send DSC enable to synaptics\n");
+
+	} else {
+		/* Synaptics hub not support virtual dpcd,
+		 * external monitor occur garbage while disable DSC,
+		 * Disable DSC only when entire link status turn to false,
+		 */
+		if (!stream->link->link_status.link_active) {
+			ret = drm_dp_dpcd_write(aux, DP_DSC_ENABLE, &enable, 1);
+			DRM_INFO("MST_DSC Send DSC disable to synaptics\n");
+		}
+	}
+
+	return ret;
+}
+
+bool dm_helpers_dp_write_dsc_enable(
+		struct dc_context *ctx,
+		const struct dc_stream_state *stream,
+		bool enable)
+{
+	static const uint8_t DSC_DISABLE;
+	static const uint8_t DSC_DECODING = 0x01;
+	static const uint8_t DSC_PASSTHROUGH = 0x02;
+
+	struct amdgpu_dm_connector *aconnector =
+		(struct amdgpu_dm_connector *)stream->dm_stream_context;
+	struct drm_device *dev = aconnector->base.dev;
+	struct drm_dp_mst_port *port;
+	uint8_t enable_dsc = enable ? DSC_DECODING : DSC_DISABLE;
+	uint8_t enable_passthrough = enable ? DSC_PASSTHROUGH : DSC_DISABLE;
+	uint8_t ret = 0;
+
+	if (stream->signal == SIGNAL_TYPE_DISPLAY_PORT_MST) {
+		if (!aconnector->dsc_aux)
+			return false;
+
+		// apply w/a to synaptics
+		if (needs_dsc_aux_workaround(aconnector->dc_link) &&
+		    (aconnector->mst_downstream_port_present.byte & 0x7) != 0x3)
+			return write_dsc_enable_synaptics_non_virtual_dpcd_mst(
+				aconnector->dsc_aux, stream, enable_dsc);
+
+		port = aconnector->mst_output_port;
+
+		if (enable) {
+			if (port->passthrough_aux) {
+				ret = drm_dp_dpcd_write(port->passthrough_aux,
+							DP_DSC_ENABLE,
+							&enable_passthrough, 1);
+				drm_dbg_dp(dev,
+					   "MST_DSC Sent DSC pass-through enable to virtual dpcd port, ret = %u\n",
+					   ret);
+			}
+
+			ret = drm_dp_dpcd_write(aconnector->dsc_aux,
+						DP_DSC_ENABLE, &enable_dsc, 1);
+			drm_dbg_dp(dev,
+				   "MST_DSC Sent DSC decoding enable to %s port, ret = %u\n",
+				   (port->passthrough_aux) ? "remote RX" :
+				   "virtual dpcd",
+				   ret);
+		} else {
+			ret = drm_dp_dpcd_write(aconnector->dsc_aux,
+						DP_DSC_ENABLE, &enable_dsc, 1);
+			drm_dbg_dp(dev,
+				   "MST_DSC Sent DSC decoding disable to %s port, ret = %u\n",
+				   (port->passthrough_aux) ? "remote RX" :
+				   "virtual dpcd",
+				   ret);
+
+			if (port->passthrough_aux) {
+				ret = drm_dp_dpcd_write(port->passthrough_aux,
+							DP_DSC_ENABLE,
+							&enable_passthrough, 1);
+				drm_dbg_dp(dev,
+					   "MST_DSC Sent DSC pass-through disable to virtual dpcd port, ret = %u\n",
+					   ret);
+			}
+		}
+	}
+
+	if (stream->signal == SIGNAL_TYPE_DISPLAY_PORT || stream->signal == SIGNAL_TYPE_EDP) {
+		if (stream->sink->link->dpcd_caps.dongle_type == DISPLAY_DONGLE_NONE) {
+			ret = dm_helpers_dp_write_dpcd(ctx, stream->link, DP_DSC_ENABLE, &enable_dsc, 1);
+			drm_dbg_dp(dev,
+				   "SST_DSC Send DSC %s to SST RX\n",
+				   enable_dsc ? "enable" : "disable");
+		} else if (stream->sink->link->dpcd_caps.dongle_type == DISPLAY_DONGLE_DP_HDMI_CONVERTER) {
+			ret = dm_helpers_dp_write_dpcd(ctx, stream->link, DP_DSC_ENABLE, &enable_dsc, 1);
+			drm_dbg_dp(dev,
+				   "SST_DSC Send DSC %s to DP-HDMI PCON\n",
+				   enable_dsc ? "enable" : "disable");
+		}
+	}
+
+	return ret;
+}
+
+bool dm_helpers_dp_write_hblank_reduction(struct dc_context *ctx, const struct dc_stream_state *stream)
+{
+	// TODO
+	return false;
+}
+
+bool dm_helpers_is_dp_sink_present(struct dc_link *link)
+{
+	bool dp_sink_present;
+	struct amdgpu_dm_connector *aconnector = link->priv;
+
+	if (!aconnector) {
+		BUG_ON("Failed to find connector for link!");
+		return true;
+	}
+
+	mutex_lock(&aconnector->dm_dp_aux.aux.hw_mutex);
+	dp_sink_present = dc_link_is_dp_sink_present(link);
+	mutex_unlock(&aconnector->dm_dp_aux.aux.hw_mutex);
+	return dp_sink_present;
+}
+
+static int
+dm_helpers_probe_acpi_edid(void *data, u8 *buf, unsigned int block, size_t len)
+{
+	struct drm_connector *connector = data;
+	struct acpi_device *acpidev = ACPI_COMPANION(connector->dev->dev);
+	unsigned short start = block * EDID_LENGTH;
+	struct edid *edid;
+	int r;
+
+	if (!acpidev)
+		return -ENODEV;
+
+	/* fetch the entire edid from BIOS */
+	r = acpi_video_get_edid(acpidev, ACPI_VIDEO_DISPLAY_LCD, -1, (void *)&edid);
+	if (r < 0) {
+		drm_dbg(connector->dev, "Failed to get EDID from ACPI: %d\n", r);
+		return r;
+	}
+	if (len > r || start > r || start + len > r) {
+		r = -EINVAL;
+		goto cleanup;
+	}
+
+	/* sanity check */
+	if (edid->revision < 4 || !(edid->input & DRM_EDID_INPUT_DIGITAL) ||
+	    (edid->input & DRM_EDID_DIGITAL_TYPE_MASK) == DRM_EDID_DIGITAL_TYPE_UNDEF) {
+		r = -EINVAL;
+		goto cleanup;
+	}
+
+	memcpy(buf, (void *)edid + start, len);
+	r = 0;
+
+cleanup:
+	kfree(edid);
+
+	return r;
+}
+
+static const struct drm_edid *
+dm_helpers_read_acpi_edid(struct amdgpu_dm_connector *aconnector)
+{
+	struct drm_connector *connector = &aconnector->base;
+
+	if (amdgpu_dc_debug_mask & DC_DISABLE_ACPI_EDID)
+		return NULL;
+
+	switch (connector->connector_type) {
+	case DRM_MODE_CONNECTOR_LVDS:
+	case DRM_MODE_CONNECTOR_eDP:
+		break;
+	default:
+		return NULL;
+	}
+
+	if (connector->force == DRM_FORCE_OFF)
+		return NULL;
+
+	return drm_edid_read_custom(connector, dm_helpers_probe_acpi_edid, connector);
+}
+
+enum dc_edid_status dm_helpers_read_local_edid(
+		struct dc_context *ctx,
+		struct dc_link *link,
+		struct dc_sink *sink)
+{
+	struct amdgpu_dm_connector *aconnector = link->priv;
+	struct drm_connector *connector = &aconnector->base;
+	struct i2c_adapter *ddc;
+	int retry = 3;
+	enum dc_edid_status edid_status;
+	const struct drm_edid *drm_edid;
+	const struct edid *edid;
+
+	if (link->aux_mode)
+		ddc = &aconnector->dm_dp_aux.aux.ddc;
+	else
+		ddc = &aconnector->i2c->base;
+
+	/* some dongles read edid incorrectly the first time,
+	 * do check sum and retry to make sure read correct edid.
+	 */
+	do {
+		drm_edid = dm_helpers_read_acpi_edid(aconnector);
+		if (drm_edid)
+			drm_info(connector->dev, "Using ACPI provided EDID for %s\n", connector->name);
+		else
+			drm_edid = drm_edid_read_ddc(connector, ddc);
+		drm_edid_connector_update(connector, drm_edid);
+
+		/* DP Compliance Test 4.2.2.6 */
+		if (link->aux_mode && connector->edid_corrupt)
+			drm_dp_send_real_edid_checksum(&aconnector->dm_dp_aux.aux, connector->real_edid_checksum);
+
+		if (!drm_edid && connector->edid_corrupt) {
+			connector->edid_corrupt = false;
+			return EDID_BAD_CHECKSUM;
+		}
+
+		if (!drm_edid)
+			return EDID_NO_RESPONSE;
+
+		edid = drm_edid_raw(drm_edid); // FIXME: Get rid of drm_edid_raw()
+		if (!edid ||
+		    edid->extensions >= sizeof(sink->dc_edid.raw_edid) / EDID_LENGTH)
+			return EDID_BAD_INPUT;
+
+		sink->dc_edid.length = EDID_LENGTH * (edid->extensions + 1);
+		memmove(sink->dc_edid.raw_edid, (uint8_t *)edid, sink->dc_edid.length);
+
+		/* We don't need the original edid anymore */
+		drm_edid_free(drm_edid);
+
+		edid_status = dm_helpers_parse_edid_caps(
+						link,
+						&sink->dc_edid,
+						&sink->edid_caps);
+
+	} while (edid_status == EDID_BAD_CHECKSUM && --retry > 0);
+
+	if (edid_status != EDID_OK)
+		DRM_ERROR("EDID err: %d, on connector: %s",
+				edid_status,
+				aconnector->base.name);
+	if (link->aux_mode) {
+		union test_request test_request = {0};
+		union test_response test_response = {0};
+
+		dm_helpers_dp_read_dpcd(ctx,
+					link,
+					DP_TEST_REQUEST,
+					&test_request.raw,
+					sizeof(union test_request));
+
+		if (!test_request.bits.EDID_READ)
+			return edid_status;
+
+		test_response.bits.EDID_CHECKSUM_WRITE = 1;
+
+		dm_helpers_dp_write_dpcd(ctx,
+					link,
+					DP_TEST_EDID_CHECKSUM,
+					&sink->dc_edid.raw_edid[sink->dc_edid.length-1],
+					1);
+
+		dm_helpers_dp_write_dpcd(ctx,
+					link,
+					DP_TEST_RESPONSE,
+					&test_response.raw,
+					sizeof(test_response));
+
+	}
+
+	return edid_status;
+}
+int dm_helper_dmub_aux_transfer_sync(
+		struct dc_context *ctx,
+		const struct dc_link *link,
+		struct aux_payload *payload,
+		enum aux_return_code_type *operation_result)
+{
+	if (!link->hpd_status) {
+		*operation_result = AUX_RET_ERROR_HPD_DISCON;
+		return -1;
+	}
+
+	return amdgpu_dm_process_dmub_aux_transfer_sync(ctx, link->link_index, payload,
+			operation_result);
+}
+
+int dm_helpers_dmub_set_config_sync(struct dc_context *ctx,
+		const struct dc_link *link,
+		struct set_config_cmd_payload *payload,
+		enum set_config_status *operation_result)
+{
+	return amdgpu_dm_process_dmub_set_config_sync(ctx, link->link_index, payload,
+			operation_result);
+}
+
+void dm_set_dcn_clocks(struct dc_context *ctx, struct dc_clocks *clks)
+{
+	/* TODO: something */
+}
+
+void dm_helpers_smu_timeout(struct dc_context *ctx, unsigned int msg_id, unsigned int param, unsigned int timeout_us)
+{
+	// TODO:
+	//amdgpu_device_gpu_recover(dc_context->driver-context, NULL);
+}
+
+void dm_helpers_init_panel_settings(
+	struct dc_context *ctx,
+	struct dc_panel_config *panel_config,
+	struct dc_sink *sink)
+{
+	// Extra Panel Power Sequence
+	panel_config->pps.extra_t3_ms = sink->edid_caps.panel_patch.extra_t3_ms;
+	panel_config->pps.extra_t7_ms = sink->edid_caps.panel_patch.extra_t7_ms;
+	panel_config->pps.extra_delay_backlight_off = sink->edid_caps.panel_patch.extra_delay_backlight_off;
+	panel_config->pps.extra_post_t7_ms = 0;
+	panel_config->pps.extra_pre_t11_ms = 0;
+	panel_config->pps.extra_t12_ms = sink->edid_caps.panel_patch.extra_t12_ms;
+	panel_config->pps.extra_post_OUI_ms = 0;
+	// Feature DSC
+	panel_config->dsc.disable_dsc_edp = false;
+	panel_config->dsc.force_dsc_edp_policy = 0;
+}
+
+void dm_helpers_override_panel_settings(
+	struct dc_context *ctx,
+	struct dc_panel_config *panel_config)
+{
+	// Feature DSC
+	if (amdgpu_dc_debug_mask & DC_DISABLE_DSC)
+		panel_config->dsc.disable_dsc_edp = true;
+}
+
+void *dm_helpers_allocate_gpu_mem(
+		struct dc_context *ctx,
+		enum dc_gpu_mem_alloc_type type,
+		size_t size,
+		long long *addr)
+{
+	struct amdgpu_device *adev = ctx->driver_context;
+
+	return dm_allocate_gpu_mem(adev, type, size, addr);
+}
+
+void dm_helpers_free_gpu_mem(
+		struct dc_context *ctx,
+		enum dc_gpu_mem_alloc_type type,
+		void *pvMem)
+{
+	struct amdgpu_device *adev = ctx->driver_context;
+
+	dm_free_gpu_mem(adev, type, pvMem);
+}
+
+bool dm_helpers_dmub_outbox_interrupt_control(struct dc_context *ctx, bool enable)
+{
+	enum dc_irq_source irq_source;
+	bool ret;
+
+	irq_source = DC_IRQ_SOURCE_DMCUB_OUTBOX;
+
+	ret = dc_interrupt_set(ctx->dc, irq_source, enable);
+
+	DRM_DEBUG_DRIVER("Dmub trace irq %sabling: r=%d\n",
+			 enable ? "en" : "dis", ret);
+	return ret;
+}
+
+void dm_helpers_mst_enable_stream_features(const struct dc_stream_state *stream)
+{
+	/* TODO: virtual DPCD */
+	struct dc_link *link = stream->link;
+	union down_spread_ctrl old_downspread;
+	union down_spread_ctrl new_downspread;
+
+	if (link->aux_access_disabled)
+		return;
+
+	if (!dm_helpers_dp_read_dpcd(link->ctx, link, DP_DOWNSPREAD_CTRL,
+				     &old_downspread.raw,
+				     sizeof(old_downspread)))
+		return;
+
+	new_downspread.raw = old_downspread.raw;
+	new_downspread.bits.IGNORE_MSA_TIMING_PARAM =
+		(stream->ignore_msa_timing_param) ? 1 : 0;
+
+	if (new_downspread.raw != old_downspread.raw)
+		dm_helpers_dp_write_dpcd(link->ctx, link, DP_DOWNSPREAD_CTRL,
+					 &new_downspread.raw,
+					 sizeof(new_downspread));
+}
+
+bool dm_helpers_dp_handle_test_pattern_request(
+		struct dc_context *ctx,
+		const struct dc_link *link,
+		union link_test_pattern dpcd_test_pattern,
+		union test_misc dpcd_test_params)
+{
+	enum dp_test_pattern test_pattern;
+	enum dp_test_pattern_color_space test_pattern_color_space =
+			DP_TEST_PATTERN_COLOR_SPACE_UNDEFINED;
+	enum dc_color_depth requestColorDepth = COLOR_DEPTH_UNDEFINED;
+	enum dc_pixel_encoding requestPixelEncoding = PIXEL_ENCODING_UNDEFINED;
+	struct pipe_ctx *pipes = link->dc->current_state->res_ctx.pipe_ctx;
+	struct pipe_ctx *pipe_ctx = NULL;
+	struct amdgpu_dm_connector *aconnector = link->priv;
+	struct drm_device *dev = aconnector->base.dev;
+	struct dc_state *dc_state = ctx->dc->current_state;
+	struct clk_mgr *clk_mgr = ctx->dc->clk_mgr;
+	int i;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		if (pipes[i].stream == NULL)
+			continue;
+
+		if (pipes[i].stream->link == link && !pipes[i].top_pipe &&
+			!pipes[i].prev_odm_pipe) {
+			pipe_ctx = &pipes[i];
+			break;
+		}
+	}
+
+	if (pipe_ctx == NULL)
+		return false;
+
+	switch (dpcd_test_pattern.bits.PATTERN) {
+	case LINK_TEST_PATTERN_COLOR_RAMP:
+		test_pattern = DP_TEST_PATTERN_COLOR_RAMP;
+	break;
+	case LINK_TEST_PATTERN_VERTICAL_BARS:
+		test_pattern = DP_TEST_PATTERN_VERTICAL_BARS;
+	break; /* black and white */
+	case LINK_TEST_PATTERN_COLOR_SQUARES:
+		test_pattern = (dpcd_test_params.bits.DYN_RANGE ==
+				TEST_DYN_RANGE_VESA ?
+				DP_TEST_PATTERN_COLOR_SQUARES :
+				DP_TEST_PATTERN_COLOR_SQUARES_CEA);
+	break;
+	default:
+		test_pattern = DP_TEST_PATTERN_VIDEO_MODE;
+	break;
+	}
+
+	if (dpcd_test_params.bits.CLR_FORMAT == 0)
+		test_pattern_color_space = DP_TEST_PATTERN_COLOR_SPACE_RGB;
+	else
+		test_pattern_color_space = dpcd_test_params.bits.YCBCR_COEFS ?
+				DP_TEST_PATTERN_COLOR_SPACE_YCBCR709 :
+				DP_TEST_PATTERN_COLOR_SPACE_YCBCR601;
+
+	switch (dpcd_test_params.bits.BPC) {
+	case 0: // 6 bits
+		requestColorDepth = COLOR_DEPTH_666;
+		break;
+	case 1: // 8 bits
+		requestColorDepth = COLOR_DEPTH_888;
+		break;
+	case 2: // 10 bits
+		requestColorDepth = COLOR_DEPTH_101010;
+		break;
+	case 3: // 12 bits
+		requestColorDepth = COLOR_DEPTH_121212;
+		break;
+	default:
+		break;
+	}
+
+	switch (dpcd_test_params.bits.CLR_FORMAT) {
+	case 0:
+		requestPixelEncoding = PIXEL_ENCODING_RGB;
+		break;
+	case 1:
+		requestPixelEncoding = PIXEL_ENCODING_YCBCR422;
+		break;
+	case 2:
+		requestPixelEncoding = PIXEL_ENCODING_YCBCR444;
+		break;
+	default:
+		requestPixelEncoding = PIXEL_ENCODING_RGB;
+		break;
+	}
+
+	if ((requestColorDepth != COLOR_DEPTH_UNDEFINED
+		&& pipe_ctx->stream->timing.display_color_depth != requestColorDepth)
+		|| (requestPixelEncoding != PIXEL_ENCODING_UNDEFINED
+		&& pipe_ctx->stream->timing.pixel_encoding != requestPixelEncoding)) {
+		drm_dbg(dev,
+			"original bpc %d pix encoding %d, changing to %d  %d\n",
+			pipe_ctx->stream->timing.display_color_depth,
+			pipe_ctx->stream->timing.pixel_encoding,
+			requestColorDepth,
+			requestPixelEncoding);
+		pipe_ctx->stream->timing.display_color_depth = requestColorDepth;
+		pipe_ctx->stream->timing.pixel_encoding = requestPixelEncoding;
+
+		dc_link_update_dsc_config(pipe_ctx);
+
+		aconnector->timing_changed = true;
+		/* store current timing */
+		if (aconnector->timing_requested)
+			*aconnector->timing_requested = pipe_ctx->stream->timing;
+		else
+			drm_err(dev, "timing storage failed\n");
+
+	}
+
+	pipe_ctx->stream->test_pattern.type = test_pattern;
+	pipe_ctx->stream->test_pattern.color_space = test_pattern_color_space;
+
+	/* Temp W/A for compliance test failure */
+	dc_state->bw_ctx.bw.dcn.clk.p_state_change_support = false;
+	dc_state->bw_ctx.bw.dcn.clk.dramclk_khz = clk_mgr->dc_mode_softmax_enabled ?
+		clk_mgr->bw_params->dc_mode_softmax_memclk : clk_mgr->bw_params->max_memclk_mhz;
+	dc_state->bw_ctx.bw.dcn.clk.idle_dramclk_khz = dc_state->bw_ctx.bw.dcn.clk.dramclk_khz;
+	ctx->dc->clk_mgr->funcs->update_clocks(
+			ctx->dc->clk_mgr,
+			dc_state,
+			false);
+
+	dc_link_dp_set_test_pattern(
+		(struct dc_link *) link,
+		test_pattern,
+		test_pattern_color_space,
+		NULL,
+		NULL,
+		0);
+
+	return false;
+}
+
+void dm_set_phyd32clk(struct dc_context *ctx, int freq_khz)
+{
+       // TODO
+}
+
+void dm_helpers_enable_periodic_detection(struct dc_context *ctx, bool enable)
+{
+	struct amdgpu_device *adev = ctx->driver_context;
+
+	if (adev->dm.idle_workqueue) {
+		adev->dm.idle_workqueue->enable = enable;
+		if (enable && !adev->dm.idle_workqueue->running && amdgpu_dm_is_headless(adev))
+			schedule_work(&adev->dm.idle_workqueue->work);
+	}
+}
+
+void dm_helpers_dp_mst_update_branch_bandwidth(
+		struct dc_context *ctx,
+		struct dc_link *link)
+{
+	// TODO
+}
+
+static bool dm_is_freesync_pcon_whitelist(const uint32_t branch_dev_id)
+{
+	bool ret_val = false;
+
+	switch (branch_dev_id) {
+	case DP_BRANCH_DEVICE_ID_0060AD:
+	case DP_BRANCH_DEVICE_ID_00E04C:
+	case DP_BRANCH_DEVICE_ID_90CC24:
+		ret_val = true;
+		break;
+	default:
+		break;
+	}
+
+	return ret_val;
+}
+
+enum adaptive_sync_type dm_get_adaptive_sync_support_type(struct dc_link *link)
+{
+	struct dpcd_caps *dpcd_caps = &link->dpcd_caps;
+	enum adaptive_sync_type as_type = ADAPTIVE_SYNC_TYPE_NONE;
+
+	switch (dpcd_caps->dongle_type) {
+	case DISPLAY_DONGLE_DP_HDMI_CONVERTER:
+		if (dpcd_caps->adaptive_sync_caps.dp_adap_sync_caps.bits.ADAPTIVE_SYNC_SDP_SUPPORT == true &&
+			dpcd_caps->allow_invalid_MSA_timing_param == true &&
+			dm_is_freesync_pcon_whitelist(dpcd_caps->branch_dev_id))
+			as_type = FREESYNC_TYPE_PCON_IN_WHITELIST;
+		break;
+	default:
+		break;
+	}
+
+	return as_type;
+}
+
+bool dm_helpers_is_fullscreen(struct dc_context *ctx, struct dc_stream_state *stream)
+{
+	// TODO
+	return false;
+}
+
+bool dm_helpers_is_hdr_on(struct dc_context *ctx, struct dc_stream_state *stream)
+{
+	// TODO
+	return false;
+}
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c
new file mode 100644
index 000000000000..a1c722112c22
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c
@@ -0,0 +1,1000 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services_types.h"
+#include "dc.h"
+
+#include "amdgpu.h"
+#include "amdgpu_dm.h"
+#include "amdgpu_dm_irq.h"
+
+/**
+ * DOC: overview
+ *
+ * DM provides another layer of IRQ management on top of what the base driver
+ * already provides. This is something that could be cleaned up, and is a
+ * future TODO item.
+ *
+ * The base driver provides IRQ source registration with DRM, handler
+ * registration into the base driver's IRQ table, and a handler callback
+ * amdgpu_irq_handler(), with which DRM calls on interrupts. This generic
+ * handler looks up the IRQ table, and calls the respective
+ * &amdgpu_irq_src_funcs.process hookups.
+ *
+ * What DM provides on top are two IRQ tables specifically for top-half and
+ * bottom-half IRQ handling, with the bottom-half implementing workqueues:
+ *
+ * - &amdgpu_display_manager.irq_handler_list_high_tab
+ * - &amdgpu_display_manager.irq_handler_list_low_tab
+ *
+ * They override the base driver's IRQ table, and the effect can be seen
+ * in the hooks that DM provides for &amdgpu_irq_src_funcs.process. They
+ * are all set to the DM generic handler amdgpu_dm_irq_handler(), which looks up
+ * DM's IRQ tables. However, in order for base driver to recognize this hook, DM
+ * still needs to register the IRQ with the base driver. See
+ * dce110_register_irq_handlers() and dcn10_register_irq_handlers().
+ *
+ * To expose DC's hardware interrupt toggle to the base driver, DM implements
+ * &amdgpu_irq_src_funcs.set hooks. Base driver calls it through
+ * amdgpu_irq_update() to enable or disable the interrupt.
+ */
+
+/******************************************************************************
+ * Private declarations.
+ *****************************************************************************/
+
+/**
+ * struct amdgpu_dm_irq_handler_data - Data for DM interrupt handlers.
+ *
+ * @list: Linked list entry referencing the next/previous handler
+ * @handler: Handler function
+ * @handler_arg: Argument passed to the handler when triggered
+ * @dm: DM which this handler belongs to
+ * @irq_source: DC interrupt source that this handler is registered for
+ * @work: work struct
+ */
+struct amdgpu_dm_irq_handler_data {
+	struct list_head list;
+	interrupt_handler handler;
+	void *handler_arg;
+
+	struct amdgpu_display_manager *dm;
+	/* DAL irq source which registered for this interrupt. */
+	enum dc_irq_source irq_source;
+	struct work_struct work;
+};
+
+#define DM_IRQ_TABLE_LOCK(adev, flags) \
+	spin_lock_irqsave(&adev->dm.irq_handler_list_table_lock, flags)
+
+#define DM_IRQ_TABLE_UNLOCK(adev, flags) \
+	spin_unlock_irqrestore(&adev->dm.irq_handler_list_table_lock, flags)
+
+/******************************************************************************
+ * Private functions.
+ *****************************************************************************/
+
+static void init_handler_common_data(struct amdgpu_dm_irq_handler_data *hcd,
+				     void (*ih)(void *),
+				     void *args,
+				     struct amdgpu_display_manager *dm)
+{
+	hcd->handler = ih;
+	hcd->handler_arg = args;
+	hcd->dm = dm;
+}
+
+/**
+ * dm_irq_work_func() - Handle an IRQ outside of the interrupt handler proper.
+ *
+ * @work: work struct
+ */
+static void dm_irq_work_func(struct work_struct *work)
+{
+	struct amdgpu_dm_irq_handler_data *handler_data =
+		container_of(work, struct amdgpu_dm_irq_handler_data, work);
+
+	handler_data->handler(handler_data->handler_arg);
+
+	/* Call a DAL subcomponent which registered for interrupt notification
+	 * at INTERRUPT_LOW_IRQ_CONTEXT.
+	 * (The most common use is HPD interrupt)
+	 */
+}
+
+/*
+ * Remove a handler and return a pointer to handler list from which the
+ * handler was removed.
+ */
+static struct list_head *remove_irq_handler(struct amdgpu_device *adev,
+					    void *ih,
+					    const struct dc_interrupt_params *int_params)
+{
+	struct list_head *hnd_list;
+	struct list_head *entry, *tmp;
+	struct amdgpu_dm_irq_handler_data *handler;
+	unsigned long irq_table_flags;
+	bool handler_removed = false;
+	enum dc_irq_source irq_source;
+
+	DM_IRQ_TABLE_LOCK(adev, irq_table_flags);
+
+	irq_source = int_params->irq_source;
+
+	switch (int_params->int_context) {
+	case INTERRUPT_HIGH_IRQ_CONTEXT:
+		hnd_list = &adev->dm.irq_handler_list_high_tab[irq_source];
+		break;
+	case INTERRUPT_LOW_IRQ_CONTEXT:
+	default:
+		hnd_list = &adev->dm.irq_handler_list_low_tab[irq_source];
+		break;
+	}
+
+	list_for_each_safe(entry, tmp, hnd_list) {
+
+		handler = list_entry(entry, struct amdgpu_dm_irq_handler_data,
+				     list);
+
+		if (handler == NULL)
+			continue;
+
+		if (ih == handler->handler) {
+			/* Found our handler. Remove it from the list. */
+			list_del(&handler->list);
+			handler_removed = true;
+			break;
+		}
+	}
+
+	DM_IRQ_TABLE_UNLOCK(adev, irq_table_flags);
+
+	if (handler_removed == false) {
+		/* Not necessarily an error - caller may not
+		 * know the context.
+		 */
+		return NULL;
+	}
+
+	kfree(handler);
+
+	DRM_DEBUG_KMS(
+	"DM_IRQ: removed irq handler: %p for: dal_src=%d, irq context=%d\n",
+		ih, int_params->irq_source, int_params->int_context);
+
+	return hnd_list;
+}
+
+/**
+ * unregister_all_irq_handlers() - Cleans up handlers from the DM IRQ table
+ * @adev: The base driver device containing the DM device
+ *
+ * Go through low and high context IRQ tables and deallocate handlers.
+ */
+static void unregister_all_irq_handlers(struct amdgpu_device *adev)
+{
+	struct list_head *hnd_list_low;
+	struct list_head *hnd_list_high;
+	struct list_head *entry, *tmp;
+	struct amdgpu_dm_irq_handler_data *handler;
+	unsigned long irq_table_flags;
+	int i;
+
+	DM_IRQ_TABLE_LOCK(adev, irq_table_flags);
+
+	for (i = 0; i < DAL_IRQ_SOURCES_NUMBER; i++) {
+		hnd_list_low = &adev->dm.irq_handler_list_low_tab[i];
+		hnd_list_high = &adev->dm.irq_handler_list_high_tab[i];
+
+		list_for_each_safe(entry, tmp, hnd_list_low) {
+
+			handler = list_entry(entry, struct amdgpu_dm_irq_handler_data,
+					     list);
+
+			if (handler == NULL || handler->handler == NULL)
+				continue;
+
+			list_del(&handler->list);
+			kfree(handler);
+		}
+
+		list_for_each_safe(entry, tmp, hnd_list_high) {
+
+			handler = list_entry(entry, struct amdgpu_dm_irq_handler_data,
+					     list);
+
+			if (handler == NULL || handler->handler == NULL)
+				continue;
+
+			list_del(&handler->list);
+			kfree(handler);
+		}
+	}
+
+	DM_IRQ_TABLE_UNLOCK(adev, irq_table_flags);
+}
+
+static bool
+validate_irq_registration_params(struct dc_interrupt_params *int_params,
+				 void (*ih)(void *))
+{
+	if (NULL == int_params || NULL == ih) {
+		DRM_ERROR("DM_IRQ: invalid input!\n");
+		return false;
+	}
+
+	if (int_params->int_context >= INTERRUPT_CONTEXT_NUMBER) {
+		DRM_ERROR("DM_IRQ: invalid context: %d!\n",
+				int_params->int_context);
+		return false;
+	}
+
+	if (!DAL_VALID_IRQ_SRC_NUM(int_params->irq_source)) {
+		DRM_ERROR("DM_IRQ: invalid irq_source: %d!\n",
+				int_params->irq_source);
+		return false;
+	}
+
+	return true;
+}
+
+static bool validate_irq_unregistration_params(enum dc_irq_source irq_source,
+					       irq_handler_idx handler_idx)
+{
+	if (handler_idx == DAL_INVALID_IRQ_HANDLER_IDX) {
+		DRM_ERROR("DM_IRQ: invalid handler_idx==NULL!\n");
+		return false;
+	}
+
+	if (!DAL_VALID_IRQ_SRC_NUM(irq_source)) {
+		DRM_ERROR("DM_IRQ: invalid irq_source:%d!\n", irq_source);
+		return false;
+	}
+
+	return true;
+}
+/******************************************************************************
+ * Public functions.
+ *
+ * Note: caller is responsible for input validation.
+ *****************************************************************************/
+
+/**
+ * amdgpu_dm_irq_register_interrupt() - Register a handler within DM.
+ * @adev: The base driver device containing the DM device.
+ * @int_params: Interrupt parameters containing the source, and handler context
+ * @ih: Function pointer to the interrupt handler to register
+ * @handler_args: Arguments passed to the handler when the interrupt occurs
+ *
+ * Register an interrupt handler for the given IRQ source, under the given
+ * context. The context can either be high or low. High context handlers are
+ * executed directly within ISR context, while low context is executed within a
+ * workqueue, thereby allowing operations that sleep.
+ *
+ * Registered handlers are called in a FIFO manner, i.e. the most recently
+ * registered handler will be called first.
+ *
+ * Return: Handler data &struct amdgpu_dm_irq_handler_data containing the IRQ
+ *         source, handler function, and args
+ */
+void *amdgpu_dm_irq_register_interrupt(struct amdgpu_device *adev,
+				       struct dc_interrupt_params *int_params,
+				       void (*ih)(void *),
+				       void *handler_args)
+{
+	struct list_head *hnd_list;
+	struct amdgpu_dm_irq_handler_data *handler_data;
+	unsigned long irq_table_flags;
+	enum dc_irq_source irq_source;
+
+	if (false == validate_irq_registration_params(int_params, ih))
+		return DAL_INVALID_IRQ_HANDLER_IDX;
+
+	handler_data = kzalloc(sizeof(*handler_data), GFP_KERNEL);
+	if (!handler_data) {
+		DRM_ERROR("DM_IRQ: failed to allocate irq handler!\n");
+		return DAL_INVALID_IRQ_HANDLER_IDX;
+	}
+
+	init_handler_common_data(handler_data, ih, handler_args, &adev->dm);
+
+	irq_source = int_params->irq_source;
+
+	handler_data->irq_source = irq_source;
+
+	/* Lock the list, add the handler. */
+	DM_IRQ_TABLE_LOCK(adev, irq_table_flags);
+
+	switch (int_params->int_context) {
+	case INTERRUPT_HIGH_IRQ_CONTEXT:
+		hnd_list = &adev->dm.irq_handler_list_high_tab[irq_source];
+		break;
+	case INTERRUPT_LOW_IRQ_CONTEXT:
+	default:
+		hnd_list = &adev->dm.irq_handler_list_low_tab[irq_source];
+		INIT_WORK(&handler_data->work, dm_irq_work_func);
+		break;
+	}
+
+	list_add_tail(&handler_data->list, hnd_list);
+
+	DM_IRQ_TABLE_UNLOCK(adev, irq_table_flags);
+
+	/* This pointer will be stored by code which requested interrupt
+	 * registration.
+	 * The same pointer will be needed in order to unregister the
+	 * interrupt.
+	 */
+
+	DRM_DEBUG_KMS(
+		"DM_IRQ: added irq handler: %p for: dal_src=%d, irq context=%d\n",
+		handler_data,
+		irq_source,
+		int_params->int_context);
+
+	return handler_data;
+}
+
+/**
+ * amdgpu_dm_irq_unregister_interrupt() - Remove a handler from the DM IRQ table
+ * @adev: The base driver device containing the DM device
+ * @irq_source: IRQ source to remove the given handler from
+ * @ih: Function pointer to the interrupt handler to unregister
+ *
+ * Go through both low and high context IRQ tables, and find the given handler
+ * for the given irq source. If found, remove it. Otherwise, do nothing.
+ */
+void amdgpu_dm_irq_unregister_interrupt(struct amdgpu_device *adev,
+					enum dc_irq_source irq_source,
+					void *ih)
+{
+	struct list_head *handler_list;
+	struct dc_interrupt_params int_params;
+	int i;
+
+	if (false == validate_irq_unregistration_params(irq_source, ih))
+		return;
+
+	memset(&int_params, 0, sizeof(int_params));
+
+	int_params.irq_source = irq_source;
+
+	for (i = 0; i < INTERRUPT_CONTEXT_NUMBER; i++) {
+
+		int_params.int_context = i;
+
+		handler_list = remove_irq_handler(adev, ih, &int_params);
+
+		if (handler_list != NULL)
+			break;
+	}
+
+	if (handler_list == NULL) {
+		/* If we got here, it means we searched all irq contexts
+		 * for this irq source, but the handler was not found.
+		 */
+		DRM_ERROR(
+		"DM_IRQ: failed to find irq handler:%p for irq_source:%d!\n",
+			ih, irq_source);
+	}
+}
+
+/**
+ * amdgpu_dm_irq_init() - Initialize DM IRQ management
+ * @adev:  The base driver device containing the DM device
+ *
+ * Initialize DM's high and low context IRQ tables.
+ *
+ * The N by M table contains N IRQ sources, with M
+ * &struct amdgpu_dm_irq_handler_data hooked together in a linked list. The
+ * list_heads are initialized here. When an interrupt n is triggered, all m
+ * handlers are called in sequence, FIFO according to registration order.
+ *
+ * The low context table requires special steps to initialize, since handlers
+ * will be deferred to a workqueue. See &struct irq_list_head.
+ */
+int amdgpu_dm_irq_init(struct amdgpu_device *adev)
+{
+	int src;
+	struct list_head *lh;
+
+	DRM_DEBUG_KMS("DM_IRQ\n");
+
+	spin_lock_init(&adev->dm.irq_handler_list_table_lock);
+
+	for (src = 0; src < DAL_IRQ_SOURCES_NUMBER; src++) {
+		/* low context handler list init */
+		lh = &adev->dm.irq_handler_list_low_tab[src];
+		INIT_LIST_HEAD(lh);
+		/* high context handler init */
+		INIT_LIST_HEAD(&adev->dm.irq_handler_list_high_tab[src]);
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_dm_irq_fini() - Tear down DM IRQ management
+ * @adev: The base driver device containing the DM device
+ *
+ * Flush all work within the low context IRQ table.
+ */
+void amdgpu_dm_irq_fini(struct amdgpu_device *adev)
+{
+	int src;
+	struct list_head *lh;
+	struct list_head *entry, *tmp;
+	struct amdgpu_dm_irq_handler_data *handler;
+	unsigned long irq_table_flags;
+
+	DRM_DEBUG_KMS("DM_IRQ: releasing resources.\n");
+	for (src = 0; src < DAL_IRQ_SOURCES_NUMBER; src++) {
+		DM_IRQ_TABLE_LOCK(adev, irq_table_flags);
+		/* The handler was removed from the table,
+		 * it means it is safe to flush all the 'work'
+		 * (because no code can schedule a new one).
+		 */
+		lh = &adev->dm.irq_handler_list_low_tab[src];
+		DM_IRQ_TABLE_UNLOCK(adev, irq_table_flags);
+
+		if (!list_empty(lh)) {
+			list_for_each_safe(entry, tmp, lh) {
+				handler = list_entry(
+					entry,
+					struct amdgpu_dm_irq_handler_data,
+					list);
+				flush_work(&handler->work);
+			}
+		}
+	}
+	/* Deallocate handlers from the table. */
+	unregister_all_irq_handlers(adev);
+}
+
+void amdgpu_dm_irq_suspend(struct amdgpu_device *adev)
+{
+	int src;
+	struct list_head *hnd_list_h;
+	struct list_head *hnd_list_l;
+	unsigned long irq_table_flags;
+	struct list_head *entry, *tmp;
+	struct amdgpu_dm_irq_handler_data *handler;
+
+	DM_IRQ_TABLE_LOCK(adev, irq_table_flags);
+
+	DRM_DEBUG_KMS("DM_IRQ: suspend\n");
+
+	/**
+	 * Disable HW interrupt  for HPD and HPDRX only since FLIP and VBLANK
+	 * will be disabled from manage_dm_interrupts on disable CRTC.
+	 */
+	for (src = DC_IRQ_SOURCE_HPD1; src <= DC_IRQ_SOURCE_HPD6RX; src++) {
+		hnd_list_l = &adev->dm.irq_handler_list_low_tab[src];
+		hnd_list_h = &adev->dm.irq_handler_list_high_tab[src];
+		if (!list_empty(hnd_list_l) || !list_empty(hnd_list_h))
+			dc_interrupt_set(adev->dm.dc, src, false);
+
+		DM_IRQ_TABLE_UNLOCK(adev, irq_table_flags);
+
+		if (!list_empty(hnd_list_l)) {
+			list_for_each_safe(entry, tmp, hnd_list_l) {
+				handler = list_entry(
+					entry,
+					struct amdgpu_dm_irq_handler_data,
+					list);
+				flush_work(&handler->work);
+			}
+		}
+		DM_IRQ_TABLE_LOCK(adev, irq_table_flags);
+	}
+
+	DM_IRQ_TABLE_UNLOCK(adev, irq_table_flags);
+}
+
+void amdgpu_dm_irq_resume_early(struct amdgpu_device *adev)
+{
+	int src;
+	struct list_head *hnd_list_h, *hnd_list_l;
+	unsigned long irq_table_flags;
+
+	DM_IRQ_TABLE_LOCK(adev, irq_table_flags);
+
+	drm_dbg(adev_to_drm(adev), "DM_IRQ: early resume\n");
+
+	/* re-enable short pulse interrupts HW interrupt */
+	for (src = DC_IRQ_SOURCE_HPD1RX; src <= DC_IRQ_SOURCE_HPD6RX; src++) {
+		hnd_list_l = &adev->dm.irq_handler_list_low_tab[src];
+		hnd_list_h = &adev->dm.irq_handler_list_high_tab[src];
+		if (!list_empty(hnd_list_l) || !list_empty(hnd_list_h))
+			dc_interrupt_set(adev->dm.dc, src, true);
+	}
+
+	DM_IRQ_TABLE_UNLOCK(adev, irq_table_flags);
+}
+
+void amdgpu_dm_irq_resume_late(struct amdgpu_device *adev)
+{
+	int src;
+	struct list_head *hnd_list_h, *hnd_list_l;
+	unsigned long irq_table_flags;
+
+	DM_IRQ_TABLE_LOCK(adev, irq_table_flags);
+
+	drm_dbg(adev_to_drm(adev), "DM_IRQ: resume\n");
+
+	/**
+	 * Renable HW interrupt  for HPD and only since FLIP and VBLANK
+	 * will be enabled from manage_dm_interrupts on enable CRTC.
+	 */
+	for (src = DC_IRQ_SOURCE_HPD1; src <= DC_IRQ_SOURCE_HPD6; src++) {
+		hnd_list_l = &adev->dm.irq_handler_list_low_tab[src];
+		hnd_list_h = &adev->dm.irq_handler_list_high_tab[src];
+		if (!list_empty(hnd_list_l) || !list_empty(hnd_list_h))
+			dc_interrupt_set(adev->dm.dc, src, true);
+	}
+
+	DM_IRQ_TABLE_UNLOCK(adev, irq_table_flags);
+}
+
+/*
+ * amdgpu_dm_irq_schedule_work - schedule all work items registered for the
+ * "irq_source".
+ */
+static void amdgpu_dm_irq_schedule_work(struct amdgpu_device *adev,
+					enum dc_irq_source irq_source)
+{
+	struct  list_head *handler_list = &adev->dm.irq_handler_list_low_tab[irq_source];
+	struct  amdgpu_dm_irq_handler_data *handler_data;
+	bool    work_queued = false;
+
+	if (list_empty(handler_list))
+		return;
+
+	list_for_each_entry(handler_data, handler_list, list) {
+		if (queue_work(system_highpri_wq, &handler_data->work)) {
+			work_queued = true;
+			break;
+		}
+	}
+
+	if (!work_queued) {
+		struct  amdgpu_dm_irq_handler_data *handler_data_add;
+		/*get the amdgpu_dm_irq_handler_data of first item pointed by handler_list*/
+		handler_data = container_of(handler_list->next, struct amdgpu_dm_irq_handler_data, list);
+
+		/*allocate a new amdgpu_dm_irq_handler_data*/
+		handler_data_add = kzalloc(sizeof(*handler_data), GFP_ATOMIC);
+		if (!handler_data_add) {
+			DRM_ERROR("DM_IRQ: failed to allocate irq handler!\n");
+			return;
+		}
+
+		/*copy new amdgpu_dm_irq_handler_data members from handler_data*/
+		handler_data_add->handler       = handler_data->handler;
+		handler_data_add->handler_arg   = handler_data->handler_arg;
+		handler_data_add->dm            = handler_data->dm;
+		handler_data_add->irq_source    = irq_source;
+
+		list_add_tail(&handler_data_add->list, handler_list);
+
+		INIT_WORK(&handler_data_add->work, dm_irq_work_func);
+
+		if (queue_work(system_highpri_wq, &handler_data_add->work))
+			DRM_DEBUG("Queued work for handling interrupt from "
+				  "display for IRQ source %d\n",
+				  irq_source);
+		else
+			DRM_ERROR("Failed to queue work for handling interrupt "
+				  "from display for IRQ source %d\n",
+				  irq_source);
+	}
+}
+
+/*
+ * amdgpu_dm_irq_immediate_work
+ * Callback high irq work immediately, don't send to work queue
+ */
+static void amdgpu_dm_irq_immediate_work(struct amdgpu_device *adev,
+					 enum dc_irq_source irq_source)
+{
+	struct amdgpu_dm_irq_handler_data *handler_data;
+	unsigned long irq_table_flags;
+
+	DM_IRQ_TABLE_LOCK(adev, irq_table_flags);
+
+	list_for_each_entry(handler_data,
+			    &adev->dm.irq_handler_list_high_tab[irq_source],
+			    list) {
+		/* Call a subcomponent which registered for immediate
+		 * interrupt notification
+		 */
+		handler_data->handler(handler_data->handler_arg);
+	}
+
+	DM_IRQ_TABLE_UNLOCK(adev, irq_table_flags);
+}
+
+/**
+ * amdgpu_dm_irq_handler - Generic DM IRQ handler
+ * @adev: amdgpu base driver device containing the DM device
+ * @source: Unused
+ * @entry: Data about the triggered interrupt
+ *
+ * Calls all registered high irq work immediately, and schedules work for low
+ * irq. The DM IRQ table is used to find the corresponding handlers.
+ */
+static int amdgpu_dm_irq_handler(struct amdgpu_device *adev,
+				 struct amdgpu_irq_src *source,
+				 struct amdgpu_iv_entry *entry)
+{
+
+	enum dc_irq_source src =
+		dc_interrupt_to_irq_source(
+			adev->dm.dc,
+			entry->src_id,
+			entry->src_data[0]);
+
+	dc_interrupt_ack(adev->dm.dc, src);
+
+	/* Call high irq work immediately */
+	amdgpu_dm_irq_immediate_work(adev, src);
+	/*Schedule low_irq work */
+	amdgpu_dm_irq_schedule_work(adev, src);
+
+	return 0;
+}
+
+static enum dc_irq_source amdgpu_dm_hpd_to_dal_irq_source(unsigned int type)
+{
+	switch (type) {
+	case AMDGPU_HPD_1:
+		return DC_IRQ_SOURCE_HPD1;
+	case AMDGPU_HPD_2:
+		return DC_IRQ_SOURCE_HPD2;
+	case AMDGPU_HPD_3:
+		return DC_IRQ_SOURCE_HPD3;
+	case AMDGPU_HPD_4:
+		return DC_IRQ_SOURCE_HPD4;
+	case AMDGPU_HPD_5:
+		return DC_IRQ_SOURCE_HPD5;
+	case AMDGPU_HPD_6:
+		return DC_IRQ_SOURCE_HPD6;
+	default:
+		return DC_IRQ_SOURCE_INVALID;
+	}
+}
+
+static int amdgpu_dm_set_hpd_irq_state(struct amdgpu_device *adev,
+				       struct amdgpu_irq_src *source,
+				       unsigned int type,
+				       enum amdgpu_interrupt_state state)
+{
+	enum dc_irq_source src = amdgpu_dm_hpd_to_dal_irq_source(type);
+	bool st = (state == AMDGPU_IRQ_STATE_ENABLE);
+
+	dc_interrupt_set(adev->dm.dc, src, st);
+	return 0;
+}
+
+static inline int dm_irq_state(struct amdgpu_device *adev,
+			       struct amdgpu_irq_src *source,
+			       unsigned int crtc_id,
+			       enum amdgpu_interrupt_state state,
+			       const enum irq_type dal_irq_type,
+			       const char *func)
+{
+	bool st;
+	enum dc_irq_source irq_source;
+	struct dc *dc = adev->dm.dc;
+	struct amdgpu_crtc *acrtc = adev->mode_info.crtcs[crtc_id];
+
+	if (!acrtc) {
+		DRM_ERROR(
+			"%s: crtc is NULL at id :%d\n",
+			func,
+			crtc_id);
+		return 0;
+	}
+
+	if (acrtc->otg_inst == -1)
+		return 0;
+
+	irq_source = dal_irq_type + acrtc->otg_inst;
+
+	st = (state == AMDGPU_IRQ_STATE_ENABLE);
+
+	if (dc && dc->caps.ips_support && dc->idle_optimizations_allowed)
+		dc_allow_idle_optimizations(dc, false);
+
+	dc_interrupt_set(adev->dm.dc, irq_source, st);
+	return 0;
+}
+
+static int amdgpu_dm_set_pflip_irq_state(struct amdgpu_device *adev,
+					 struct amdgpu_irq_src *source,
+					 unsigned int crtc_id,
+					 enum amdgpu_interrupt_state state)
+{
+	return dm_irq_state(
+		adev,
+		source,
+		crtc_id,
+		state,
+		IRQ_TYPE_PFLIP,
+		__func__);
+}
+
+static int amdgpu_dm_set_crtc_irq_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned int crtc_id,
+					enum amdgpu_interrupt_state state)
+{
+	return dm_irq_state(
+		adev,
+		source,
+		crtc_id,
+		state,
+		IRQ_TYPE_VBLANK,
+		__func__);
+}
+
+static int amdgpu_dm_set_vline0_irq_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned int crtc_id,
+					enum amdgpu_interrupt_state state)
+{
+	return dm_irq_state(
+		adev,
+		source,
+		crtc_id,
+		state,
+		IRQ_TYPE_VLINE0,
+		__func__);
+}
+
+static int amdgpu_dm_set_dmub_outbox_irq_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned int crtc_id,
+					enum amdgpu_interrupt_state state)
+{
+	enum dc_irq_source irq_source = DC_IRQ_SOURCE_DMCUB_OUTBOX;
+	bool st = (state == AMDGPU_IRQ_STATE_ENABLE);
+
+	dc_interrupt_set(adev->dm.dc, irq_source, st);
+	return 0;
+}
+
+static int amdgpu_dm_set_vupdate_irq_state(struct amdgpu_device *adev,
+					   struct amdgpu_irq_src *source,
+					   unsigned int crtc_id,
+					   enum amdgpu_interrupt_state state)
+{
+	return dm_irq_state(
+		adev,
+		source,
+		crtc_id,
+		state,
+		IRQ_TYPE_VUPDATE,
+		__func__);
+}
+
+static int amdgpu_dm_set_dmub_trace_irq_state(struct amdgpu_device *adev,
+					   struct amdgpu_irq_src *source,
+					   unsigned int type,
+					   enum amdgpu_interrupt_state state)
+{
+	enum dc_irq_source irq_source = DC_IRQ_SOURCE_DMCUB_OUTBOX0;
+	bool st = (state == AMDGPU_IRQ_STATE_ENABLE);
+
+	dc_interrupt_set(adev->dm.dc, irq_source, st);
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs dm_crtc_irq_funcs = {
+	.set = amdgpu_dm_set_crtc_irq_state,
+	.process = amdgpu_dm_irq_handler,
+};
+
+static const struct amdgpu_irq_src_funcs dm_vline0_irq_funcs = {
+	.set = amdgpu_dm_set_vline0_irq_state,
+	.process = amdgpu_dm_irq_handler,
+};
+
+static const struct amdgpu_irq_src_funcs dm_dmub_outbox_irq_funcs = {
+	.set = amdgpu_dm_set_dmub_outbox_irq_state,
+	.process = amdgpu_dm_irq_handler,
+};
+
+static const struct amdgpu_irq_src_funcs dm_vupdate_irq_funcs = {
+	.set = amdgpu_dm_set_vupdate_irq_state,
+	.process = amdgpu_dm_irq_handler,
+};
+
+static const struct amdgpu_irq_src_funcs dm_dmub_trace_irq_funcs = {
+	.set = amdgpu_dm_set_dmub_trace_irq_state,
+	.process = amdgpu_dm_irq_handler,
+};
+
+static const struct amdgpu_irq_src_funcs dm_pageflip_irq_funcs = {
+	.set = amdgpu_dm_set_pflip_irq_state,
+	.process = amdgpu_dm_irq_handler,
+};
+
+static const struct amdgpu_irq_src_funcs dm_hpd_irq_funcs = {
+	.set = amdgpu_dm_set_hpd_irq_state,
+	.process = amdgpu_dm_irq_handler,
+};
+
+void amdgpu_dm_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->crtc_irq.num_types = adev->mode_info.num_crtc;
+	adev->crtc_irq.funcs = &dm_crtc_irq_funcs;
+
+	adev->vline0_irq.num_types = adev->mode_info.num_crtc;
+	adev->vline0_irq.funcs = &dm_vline0_irq_funcs;
+
+	adev->dmub_outbox_irq.num_types = 1;
+	adev->dmub_outbox_irq.funcs = &dm_dmub_outbox_irq_funcs;
+
+	adev->vupdate_irq.num_types = adev->mode_info.num_crtc;
+	adev->vupdate_irq.funcs = &dm_vupdate_irq_funcs;
+
+	adev->dmub_trace_irq.num_types = 1;
+	adev->dmub_trace_irq.funcs = &dm_dmub_trace_irq_funcs;
+
+	adev->pageflip_irq.num_types = adev->mode_info.num_crtc;
+	adev->pageflip_irq.funcs = &dm_pageflip_irq_funcs;
+
+	adev->hpd_irq.num_types = adev->mode_info.num_hpd;
+	adev->hpd_irq.funcs = &dm_hpd_irq_funcs;
+}
+void amdgpu_dm_outbox_init(struct amdgpu_device *adev)
+{
+	dc_interrupt_set(adev->dm.dc,
+		DC_IRQ_SOURCE_DMCUB_OUTBOX,
+		true);
+}
+
+/**
+ * amdgpu_dm_hpd_init - hpd setup callback.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup the hpd pins used by the card (evergreen+).
+ * Enable the pin, set the polarity, and enable the hpd interrupts.
+ */
+void amdgpu_dm_hpd_init(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev_to_drm(adev);
+	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
+	int irq_type;
+	int i;
+
+	/* First, clear all hpd and hpdrx interrupts */
+	for (i = DC_IRQ_SOURCE_HPD1; i <= DC_IRQ_SOURCE_HPD6RX; i++) {
+		if (!dc_interrupt_set(adev->dm.dc, i, false))
+			drm_err(dev, "Failed to clear hpd(rx) source=%d on init\n",
+				i);
+	}
+
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
+		struct amdgpu_dm_connector *amdgpu_dm_connector;
+		const struct dc_link *dc_link;
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_WRITEBACK)
+			continue;
+
+		amdgpu_dm_connector = to_amdgpu_dm_connector(connector);
+
+		dc_link = amdgpu_dm_connector->dc_link;
+
+		/*
+		 * Get a base driver irq reference for hpd ints for the lifetime
+		 * of dm. Note that only hpd interrupt types are registered with
+		 * base driver; hpd_rx types aren't. IOW, amdgpu_irq_get/put on
+		 * hpd_rx isn't available. DM currently controls hpd_rx
+		 * explicitly with dc_interrupt_set()
+		 */
+		if (dc_link->irq_source_hpd != DC_IRQ_SOURCE_INVALID) {
+			irq_type = dc_link->irq_source_hpd - DC_IRQ_SOURCE_HPD1;
+			/*
+			 * TODO: There's a mismatch between mode_info.num_hpd
+			 * and what bios reports as the # of connectors with hpd
+			 * sources. Since the # of hpd source types registered
+			 * with base driver == mode_info.num_hpd, we have to
+			 * fallback to dc_interrupt_set for the remaining types.
+			 */
+			if (irq_type < adev->mode_info.num_hpd) {
+				if (amdgpu_irq_get(adev, &adev->hpd_irq, irq_type))
+					drm_err(dev, "DM_IRQ: Failed get HPD for source=%d)!\n",
+						dc_link->irq_source_hpd);
+			} else {
+				dc_interrupt_set(adev->dm.dc,
+						 dc_link->irq_source_hpd,
+						 true);
+			}
+		}
+
+		if (dc_link->irq_source_hpd_rx != DC_IRQ_SOURCE_INVALID) {
+			dc_interrupt_set(adev->dm.dc,
+					dc_link->irq_source_hpd_rx,
+					true);
+		}
+	}
+	drm_connector_list_iter_end(&iter);
+}
+
+/**
+ * amdgpu_dm_hpd_fini - hpd tear down callback.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tear down the hpd pins used by the card (evergreen+).
+ * Disable the hpd interrupts.
+ */
+void amdgpu_dm_hpd_fini(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev_to_drm(adev);
+	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
+	int irq_type;
+
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
+		struct amdgpu_dm_connector *amdgpu_dm_connector;
+		const struct dc_link *dc_link;
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_WRITEBACK)
+			continue;
+
+		amdgpu_dm_connector = to_amdgpu_dm_connector(connector);
+		dc_link = amdgpu_dm_connector->dc_link;
+
+		if (dc_link->irq_source_hpd != DC_IRQ_SOURCE_INVALID) {
+			irq_type = dc_link->irq_source_hpd - DC_IRQ_SOURCE_HPD1;
+
+			/* TODO: See same TODO in amdgpu_dm_hpd_init() */
+			if (irq_type < adev->mode_info.num_hpd) {
+				if (amdgpu_irq_put(adev, &adev->hpd_irq, irq_type))
+					drm_err(dev, "DM_IRQ: Failed put HPD for source=%d!\n",
+						dc_link->irq_source_hpd);
+			} else {
+				dc_interrupt_set(adev->dm.dc,
+						 dc_link->irq_source_hpd,
+						 false);
+			}
+		}
+
+		if (dc_link->irq_source_hpd_rx != DC_IRQ_SOURCE_INVALID) {
+			dc_interrupt_set(adev->dm.dc,
+					dc_link->irq_source_hpd_rx,
+					false);
+		}
+	}
+	drm_connector_list_iter_end(&iter);
+}
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.h
new file mode 100644
index 000000000000..4f6b58f4f90d
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.h
@@ -0,0 +1,104 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_DM_IRQ_H__
+#define __AMDGPU_DM_IRQ_H__
+
+#include "irq_types.h" /* DAL irq definitions */
+
+/*
+ * Display Manager IRQ-related interfaces (for use by DAL).
+ */
+
+/**
+ * amdgpu_dm_irq_init - Initialize internal structures of 'amdgpu_dm_irq'.
+ *
+ * This function should be called exactly once - during DM initialization.
+ *
+ * Returns:
+ *	0 - success
+ *	non-zero - error
+ */
+int amdgpu_dm_irq_init(struct amdgpu_device *adev);
+
+/**
+ * amdgpu_dm_irq_fini - deallocate internal structures of 'amdgpu_dm_irq'.
+ *
+ * This function should be called exactly once - during DM destruction.
+ *
+ */
+void amdgpu_dm_irq_fini(struct amdgpu_device *adev);
+
+/**
+ * amdgpu_dm_irq_register_interrupt - register irq handler for Display block.
+ *
+ * @adev: AMD DRM device
+ * @int_params: parameters for the irq
+ * @ih: pointer to the irq hander function
+ * @handler_args: arguments which will be passed to ih
+ *
+ * Returns:
+ * 	IRQ Handler Index on success.
+ * 	NULL on failure.
+ *
+ * Cannot be called from an interrupt handler.
+ */
+void *amdgpu_dm_irq_register_interrupt(struct amdgpu_device *adev,
+				       struct dc_interrupt_params *int_params,
+				       void (*ih)(void *),
+				       void *handler_args);
+
+/**
+ * amdgpu_dm_irq_unregister_interrupt - unregister handler which was registered
+ *	by amdgpu_dm_irq_register_interrupt().
+ *
+ * @adev: AMD DRM device.
+ * @ih_index: irq handler index which was returned by
+ *	amdgpu_dm_irq_register_interrupt
+ */
+void amdgpu_dm_irq_unregister_interrupt(struct amdgpu_device *adev,
+					enum dc_irq_source irq_source,
+					void *ih_index);
+
+void amdgpu_dm_set_irq_funcs(struct amdgpu_device *adev);
+
+void amdgpu_dm_outbox_init(struct amdgpu_device *adev);
+void amdgpu_dm_hpd_init(struct amdgpu_device *adev);
+void amdgpu_dm_hpd_fini(struct amdgpu_device *adev);
+
+/**
+ * amdgpu_dm_irq_suspend - disable ASIC interrupt during suspend.
+ *
+ */
+void amdgpu_dm_irq_suspend(struct amdgpu_device *adev);
+
+/**
+ * amdgpu_dm_irq_resume_early - enable HPDRX ASIC interrupts during resume.
+ * amdgpu_dm_irq_resume - enable ASIC interrupt during resume.
+ *
+ */
+void amdgpu_dm_irq_resume_early(struct amdgpu_device *adev);
+void amdgpu_dm_irq_resume_late(struct amdgpu_device *adev);
+
+#endif /* __AMDGPU_DM_IRQ_H__ */
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq_params.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq_params.h
new file mode 100644
index 000000000000..3c9995275cbd
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq_params.h
@@ -0,0 +1,50 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __AMDGPU_DM_IRQ_PARAMS_H__
+#define __AMDGPU_DM_IRQ_PARAMS_H__
+
+#include "amdgpu_dm_crc.h"
+
+struct dm_irq_params {
+	u32 last_flip_vblank;
+	struct mod_vrr_params vrr_params;
+	struct dc_stream_state *stream;
+	int active_planes;
+	bool allow_sr_entry;
+	struct mod_freesync_config freesync_config;
+
+#ifdef CONFIG_DEBUG_FS
+	enum amdgpu_dm_pipe_crc_source crc_src;
+#ifdef CONFIG_DRM_AMD_SECURE_DISPLAY
+	struct crc_window_param window_param[MAX_CRC_WINDOW_NUM];
+	/* At least one CRC window is activated or not*/
+	bool crc_window_activated;
+#endif
+#endif
+};
+
+#endif /* __AMDGPU_DM_IRQ_PARAMS_H__ */
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c
new file mode 100644
index 000000000000..5e92eaa67aa3
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c
@@ -0,0 +1,1987 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include <linux/vmalloc.h>
+#include <drm/display/drm_dp_helper.h>
+#include <drm/display/drm_dp_mst_helper.h>
+#include <drm/drm_atomic.h>
+#include <drm/drm_atomic_helper.h>
+#include <drm/drm_fixed.h>
+#include <drm/drm_edid.h>
+#include "dm_services.h"
+#include "amdgpu.h"
+#include "amdgpu_dm.h"
+#include "amdgpu_dm_mst_types.h"
+#include "amdgpu_dm_hdcp.h"
+
+#include "dc.h"
+#include "dm_helpers.h"
+
+#include "ddc_service_types.h"
+#include "dpcd_defs.h"
+
+#include "dmub_cmd.h"
+#if defined(CONFIG_DEBUG_FS)
+#include "amdgpu_dm_debugfs.h"
+#endif
+
+#include "dc/resource/dcn20/dcn20_resource.h"
+
+#define PEAK_FACTOR_X1000 1006
+
+/*
+ * This function handles both native AUX and I2C-Over-AUX transactions.
+ */
+static ssize_t dm_dp_aux_transfer(struct drm_dp_aux *aux,
+				  struct drm_dp_aux_msg *msg)
+{
+	ssize_t result = 0;
+	struct aux_payload payload;
+	enum aux_return_code_type operation_result;
+	struct amdgpu_device *adev;
+	struct ddc_service *ddc;
+	uint8_t copy[16];
+
+	if (WARN_ON(msg->size > 16))
+		return -E2BIG;
+
+	payload.address = msg->address;
+	payload.data = msg->buffer;
+	payload.length = msg->size;
+	payload.reply = &msg->reply;
+	payload.i2c_over_aux = (msg->request & DP_AUX_NATIVE_WRITE) == 0;
+	payload.write = (msg->request & DP_AUX_I2C_READ) == 0;
+	payload.mot = (msg->request & DP_AUX_I2C_MOT) != 0;
+	payload.write_status_update =
+			(msg->request & DP_AUX_I2C_WRITE_STATUS_UPDATE) != 0;
+	payload.defer_delay = 0;
+
+	if (payload.write) {
+		memcpy(copy, msg->buffer, msg->size);
+		payload.data = copy;
+	}
+
+	result = dc_link_aux_transfer_raw(TO_DM_AUX(aux)->ddc_service, &payload,
+				      &operation_result);
+
+	/*
+	 * w/a on certain intel platform where hpd is unexpected to pull low during
+	 * 1st sideband message transaction by return AUX_RET_ERROR_HPD_DISCON
+	 * aux transaction is succuess in such case, therefore bypass the error
+	 */
+	ddc = TO_DM_AUX(aux)->ddc_service;
+	adev = ddc->ctx->driver_context;
+	if (adev->dm.aux_hpd_discon_quirk) {
+		if (msg->address == DP_SIDEBAND_MSG_DOWN_REQ_BASE &&
+			operation_result == AUX_RET_ERROR_HPD_DISCON) {
+			result = msg->size;
+			operation_result = AUX_RET_SUCCESS;
+		}
+	}
+
+	/*
+	 * result equals to 0 includes the cases of AUX_DEFER/I2C_DEFER
+	 */
+	if (payload.write && result >= 0) {
+		if (result) {
+			/*one byte indicating partially written bytes*/
+			drm_dbg_dp(adev_to_drm(adev), "AUX partially written\n");
+			result = payload.data[0];
+		} else if (!payload.reply[0])
+			/*I2C_ACK|AUX_ACK*/
+			result = msg->size;
+	}
+
+	if (result < 0) {
+		switch (operation_result) {
+		case AUX_RET_SUCCESS:
+			break;
+		case AUX_RET_ERROR_HPD_DISCON:
+		case AUX_RET_ERROR_UNKNOWN:
+		case AUX_RET_ERROR_INVALID_OPERATION:
+		case AUX_RET_ERROR_PROTOCOL_ERROR:
+			result = -EIO;
+			break;
+		case AUX_RET_ERROR_INVALID_REPLY:
+		case AUX_RET_ERROR_ENGINE_ACQUIRE:
+			result = -EBUSY;
+			break;
+		case AUX_RET_ERROR_TIMEOUT:
+			result = -ETIMEDOUT;
+			break;
+		}
+
+		drm_dbg_dp(adev_to_drm(adev), "DP AUX transfer fail:%d\n", operation_result);
+	}
+
+	if (payload.reply[0])
+		drm_dbg_dp(adev_to_drm(adev), "AUX reply command not ACK: 0x%02x.",
+			payload.reply[0]);
+
+	return result;
+}
+
+static void
+dm_dp_mst_connector_destroy(struct drm_connector *connector)
+{
+	struct amdgpu_dm_connector *aconnector =
+		to_amdgpu_dm_connector(connector);
+
+	if (aconnector->dc_sink) {
+		dc_link_remove_remote_sink(aconnector->dc_link,
+					   aconnector->dc_sink);
+		dc_sink_release(aconnector->dc_sink);
+	}
+
+	drm_edid_free(aconnector->drm_edid);
+
+	drm_connector_cleanup(connector);
+	drm_dp_mst_put_port_malloc(aconnector->mst_output_port);
+	kfree(aconnector);
+}
+
+static int
+amdgpu_dm_mst_connector_late_register(struct drm_connector *connector)
+{
+	struct amdgpu_dm_connector *amdgpu_dm_connector =
+		to_amdgpu_dm_connector(connector);
+	int r;
+
+	r = drm_dp_mst_connector_late_register(connector,
+					       amdgpu_dm_connector->mst_output_port);
+	if (r < 0)
+		return r;
+
+#if defined(CONFIG_DEBUG_FS)
+	connector_debugfs_init(amdgpu_dm_connector);
+#endif
+
+	return 0;
+}
+
+
+static inline void
+amdgpu_dm_mst_reset_mst_connector_setting(struct amdgpu_dm_connector *aconnector)
+{
+	aconnector->drm_edid = NULL;
+	aconnector->dsc_aux = NULL;
+	aconnector->mst_output_port->passthrough_aux = NULL;
+	aconnector->mst_local_bw = 0;
+	aconnector->vc_full_pbn = 0;
+}
+
+static void
+amdgpu_dm_mst_connector_early_unregister(struct drm_connector *connector)
+{
+	struct amdgpu_dm_connector *aconnector =
+		to_amdgpu_dm_connector(connector);
+	struct drm_dp_mst_port *port = aconnector->mst_output_port;
+	struct amdgpu_dm_connector *root = aconnector->mst_root;
+	struct dc_link *dc_link = aconnector->dc_link;
+	struct dc_sink *dc_sink = aconnector->dc_sink;
+
+	drm_dp_mst_connector_early_unregister(connector, port);
+
+	/*
+	 * Release dc_sink for connector which its attached port is
+	 * no longer in the mst topology
+	 */
+	drm_modeset_lock(&root->mst_mgr.base.lock, NULL);
+	if (dc_sink) {
+		if (dc_link->sink_count)
+			dc_link_remove_remote_sink(dc_link, dc_sink);
+
+		drm_dbg_dp(connector->dev,
+			   "DM_MST: remove remote sink 0x%p, %d remaining\n",
+			   dc_sink, dc_link->sink_count);
+
+		dc_sink_release(dc_sink);
+		aconnector->dc_sink = NULL;
+		amdgpu_dm_mst_reset_mst_connector_setting(aconnector);
+	}
+
+	aconnector->mst_status = MST_STATUS_DEFAULT;
+	drm_modeset_unlock(&root->mst_mgr.base.lock);
+}
+
+static const struct drm_connector_funcs dm_dp_mst_connector_funcs = {
+	.fill_modes = drm_helper_probe_single_connector_modes,
+	.destroy = dm_dp_mst_connector_destroy,
+	.reset = amdgpu_dm_connector_funcs_reset,
+	.atomic_duplicate_state = amdgpu_dm_connector_atomic_duplicate_state,
+	.atomic_destroy_state = drm_atomic_helper_connector_destroy_state,
+	.atomic_set_property = amdgpu_dm_connector_atomic_set_property,
+	.atomic_get_property = amdgpu_dm_connector_atomic_get_property,
+	.late_register = amdgpu_dm_mst_connector_late_register,
+	.early_unregister = amdgpu_dm_mst_connector_early_unregister,
+};
+
+bool needs_dsc_aux_workaround(struct dc_link *link)
+{
+	if (link->dpcd_caps.branch_dev_id == DP_BRANCH_DEVICE_ID_90CC24 &&
+	    (link->dpcd_caps.dpcd_rev.raw == DPCD_REV_14 || link->dpcd_caps.dpcd_rev.raw == DPCD_REV_12) &&
+	    link->dpcd_caps.sink_count.bits.SINK_COUNT >= 2)
+		return true;
+
+	return false;
+}
+
+#if defined(CONFIG_DRM_AMD_DC_FP)
+static bool is_synaptics_cascaded_panamera(struct dc_link *link, struct drm_dp_mst_port *port)
+{
+	u8 branch_vendor_data[4] = { 0 }; // Vendor data 0x50C ~ 0x50F
+
+	if (drm_dp_dpcd_read(port->mgr->aux, DP_BRANCH_VENDOR_SPECIFIC_START, &branch_vendor_data, 4) == 4) {
+		if (link->dpcd_caps.branch_dev_id == DP_BRANCH_DEVICE_ID_90CC24 &&
+				IS_SYNAPTICS_CASCADED_PANAMERA(link->dpcd_caps.branch_dev_name, branch_vendor_data)) {
+			DRM_INFO("Synaptics Cascaded MST hub\n");
+			return true;
+		}
+	}
+
+	return false;
+}
+
+static bool validate_dsc_caps_on_connector(struct amdgpu_dm_connector *aconnector)
+{
+	struct dc_sink *dc_sink = aconnector->dc_sink;
+	struct drm_dp_mst_port *port = aconnector->mst_output_port;
+	u8 dsc_caps[16] = { 0 };
+	u8 dsc_branch_dec_caps_raw[3] = { 0 };	// DSC branch decoder caps 0xA0 ~ 0xA2
+	u8 *dsc_branch_dec_caps = NULL;
+
+	aconnector->dsc_aux = drm_dp_mst_dsc_aux_for_port(port);
+
+	/*
+	 * drm_dp_mst_dsc_aux_for_port() will return NULL for certain configs
+	 * because it only check the dsc/fec caps of the "port variable" and not the dock
+	 *
+	 * This case will return NULL: DSC capabe MST dock connected to a non fec/dsc capable display
+	 *
+	 * Workaround: explicitly check the use case above and use the mst dock's aux as dsc_aux
+	 *
+	 */
+	if (!aconnector->dsc_aux && !port->parent->port_parent &&
+	    needs_dsc_aux_workaround(aconnector->dc_link))
+		aconnector->dsc_aux = &aconnector->mst_root->dm_dp_aux.aux;
+
+	/* synaptics cascaded MST hub case */
+	if (is_synaptics_cascaded_panamera(aconnector->dc_link, port))
+		aconnector->dsc_aux = port->mgr->aux;
+
+	if (!aconnector->dsc_aux)
+		return false;
+
+	if (drm_dp_dpcd_read(aconnector->dsc_aux, DP_DSC_SUPPORT, dsc_caps, 16) < 0)
+		return false;
+
+	if (drm_dp_dpcd_read(aconnector->dsc_aux,
+			DP_DSC_BRANCH_OVERALL_THROUGHPUT_0, dsc_branch_dec_caps_raw, 3) == 3)
+		dsc_branch_dec_caps = dsc_branch_dec_caps_raw;
+
+	if (!dc_dsc_parse_dsc_dpcd(aconnector->dc_link->ctx->dc,
+				  dsc_caps, dsc_branch_dec_caps,
+				  &dc_sink->dsc_caps.dsc_dec_caps))
+		return false;
+
+	return true;
+}
+#endif
+
+static bool retrieve_downstream_port_device(struct amdgpu_dm_connector *aconnector)
+{
+	union dp_downstream_port_present ds_port_present;
+
+	if (!aconnector->dsc_aux)
+		return false;
+
+	if (drm_dp_dpcd_read(aconnector->dsc_aux, DP_DOWNSTREAMPORT_PRESENT, &ds_port_present, 1) < 0) {
+		DRM_INFO("Failed to read downstream_port_present 0x05 from DFP of branch device\n");
+		return false;
+	}
+
+	aconnector->mst_downstream_port_present = ds_port_present;
+	DRM_INFO("Downstream port present %d, type %d\n",
+			ds_port_present.fields.PORT_PRESENT, ds_port_present.fields.PORT_TYPE);
+
+	return true;
+}
+
+static bool retrieve_branch_specific_data(struct amdgpu_dm_connector *aconnector)
+{
+	struct drm_connector *connector = &aconnector->base;
+	struct drm_dp_mst_port *port = aconnector->mst_output_port;
+	struct drm_dp_mst_port *port_parent;
+	struct drm_dp_aux *immediate_upstream_aux;
+	struct drm_dp_desc branch_desc;
+
+	if (!port->parent)
+		return false;
+
+	port_parent = port->parent->port_parent;
+
+	immediate_upstream_aux = port_parent ? &port_parent->aux : port->mgr->aux;
+
+	if (drm_dp_read_desc(immediate_upstream_aux, &branch_desc, true))
+		return false;
+
+	aconnector->branch_ieee_oui = (branch_desc.ident.oui[0] << 16) +
+				      (branch_desc.ident.oui[1] << 8) +
+				      (branch_desc.ident.oui[2]);
+
+	drm_dbg_dp(port->aux.drm_dev, "MST branch oui 0x%x detected at %s\n",
+		   aconnector->branch_ieee_oui, connector->name);
+
+	return true;
+}
+
+static int dm_dp_mst_get_modes(struct drm_connector *connector)
+{
+	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector);
+	int ret = 0;
+
+	if (!aconnector)
+		return drm_add_edid_modes(connector, NULL);
+
+	if (!aconnector->drm_edid) {
+		const struct drm_edid *drm_edid;
+
+		drm_edid = drm_dp_mst_edid_read(connector,
+						&aconnector->mst_root->mst_mgr,
+						aconnector->mst_output_port);
+
+		if (!drm_edid) {
+			amdgpu_dm_set_mst_status(&aconnector->mst_status,
+			MST_REMOTE_EDID, false);
+
+			drm_edid_connector_update(
+				&aconnector->base,
+				NULL);
+
+			DRM_DEBUG_KMS("Can't get EDID of %s. Add default remote sink.", connector->name);
+			if (!aconnector->dc_sink) {
+				struct dc_sink *dc_sink;
+				struct dc_sink_init_data init_params = {
+					.link = aconnector->dc_link,
+					.sink_signal = SIGNAL_TYPE_DISPLAY_PORT_MST };
+
+				dc_sink = dc_link_add_remote_sink(
+					aconnector->dc_link,
+					NULL,
+					0,
+					&init_params);
+
+				if (!dc_sink) {
+					DRM_ERROR("Unable to add a remote sink\n");
+					return 0;
+				}
+
+				drm_dbg_dp(connector->dev,
+					   "DM_MST: add remote sink 0x%p, %d remaining\n",
+					   dc_sink,
+					   aconnector->dc_link->sink_count);
+
+				dc_sink->priv = aconnector;
+				aconnector->dc_sink = dc_sink;
+			}
+
+			return ret;
+		}
+
+		aconnector->drm_edid = drm_edid;
+		amdgpu_dm_set_mst_status(&aconnector->mst_status,
+			MST_REMOTE_EDID, true);
+	}
+
+	if (aconnector->dc_sink && aconnector->dc_sink->sink_signal == SIGNAL_TYPE_VIRTUAL) {
+		dc_sink_release(aconnector->dc_sink);
+		aconnector->dc_sink = NULL;
+	}
+
+	if (!aconnector->dc_sink) {
+		struct dc_sink *dc_sink;
+		struct dc_sink_init_data init_params = {
+				.link = aconnector->dc_link,
+				.sink_signal = SIGNAL_TYPE_DISPLAY_PORT_MST };
+		const struct edid *edid;
+
+		edid = drm_edid_raw(aconnector->drm_edid); // FIXME: Get rid of drm_edid_raw()
+		dc_sink = dc_link_add_remote_sink(
+			aconnector->dc_link,
+			(uint8_t *)edid,
+			(edid->extensions + 1) * EDID_LENGTH,
+			&init_params);
+
+		if (!dc_sink) {
+			DRM_ERROR("Unable to add a remote sink\n");
+			return 0;
+		}
+
+		drm_dbg_dp(connector->dev,
+			   "DM_MST: add remote sink 0x%p, %d remaining\n",
+			   dc_sink, aconnector->dc_link->sink_count);
+
+		dc_sink->priv = aconnector;
+		/* dc_link_add_remote_sink returns a new reference */
+		aconnector->dc_sink = dc_sink;
+
+		/* when display is unplugged from mst hub, connctor will be
+		 * destroyed within dm_dp_mst_connector_destroy. connector
+		 * hdcp perperties, like type, undesired, desired, enabled,
+		 * will be lost. So, save hdcp properties into hdcp_work within
+		 * amdgpu_dm_atomic_commit_tail. if the same display is
+		 * plugged back with same display index, its hdcp properties
+		 * will be retrieved from hdcp_work within dm_dp_mst_get_modes
+		 */
+		if (aconnector->dc_sink && connector->state) {
+			struct drm_device *dev = connector->dev;
+			struct amdgpu_device *adev = drm_to_adev(dev);
+
+			if (adev->dm.hdcp_workqueue) {
+				struct hdcp_workqueue *hdcp_work = adev->dm.hdcp_workqueue;
+				struct hdcp_workqueue *hdcp_w =
+					&hdcp_work[aconnector->dc_link->link_index];
+
+				connector->state->hdcp_content_type =
+				hdcp_w->hdcp_content_type[connector->index];
+				connector->state->content_protection =
+				hdcp_w->content_protection[connector->index];
+			}
+		}
+
+		if (aconnector->dc_sink) {
+			amdgpu_dm_update_freesync_caps(
+					connector, aconnector->drm_edid);
+
+#if defined(CONFIG_DRM_AMD_DC_FP)
+			if (!validate_dsc_caps_on_connector(aconnector))
+				memset(&aconnector->dc_sink->dsc_caps,
+				       0, sizeof(aconnector->dc_sink->dsc_caps));
+#endif
+
+			if (!retrieve_downstream_port_device(aconnector))
+				memset(&aconnector->mst_downstream_port_present,
+					0, sizeof(aconnector->mst_downstream_port_present));
+		}
+	}
+
+	drm_edid_connector_update(&aconnector->base, aconnector->drm_edid);
+
+	ret = drm_edid_connector_add_modes(connector);
+
+	return ret;
+}
+
+static struct drm_encoder *
+dm_mst_atomic_best_encoder(struct drm_connector *connector,
+			   struct drm_atomic_state *state)
+{
+	struct drm_connector_state *connector_state = drm_atomic_get_new_connector_state(state,
+											 connector);
+	struct amdgpu_device *adev = drm_to_adev(connector->dev);
+	struct amdgpu_crtc *acrtc = to_amdgpu_crtc(connector_state->crtc);
+
+	return &adev->dm.mst_encoders[acrtc->crtc_id].base;
+}
+
+static int
+dm_dp_mst_detect(struct drm_connector *connector,
+		 struct drm_modeset_acquire_ctx *ctx, bool force)
+{
+	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector);
+	struct amdgpu_dm_connector *master = aconnector->mst_root;
+	struct drm_dp_mst_port *port = aconnector->mst_output_port;
+	int connection_status;
+
+	if (drm_connector_is_unregistered(connector))
+		return connector_status_disconnected;
+
+	connection_status = drm_dp_mst_detect_port(connector, ctx, &master->mst_mgr,
+							aconnector->mst_output_port);
+
+	if (port->pdt != DP_PEER_DEVICE_NONE && !port->dpcd_rev) {
+		uint8_t dpcd_rev;
+		int ret;
+
+		ret = drm_dp_dpcd_readb(&port->aux, DP_DP13_DPCD_REV, &dpcd_rev);
+
+		if (ret == 1) {
+			port->dpcd_rev = dpcd_rev;
+
+			/* Could be DP1.2 DP Rx case*/
+			if (!dpcd_rev) {
+				ret = drm_dp_dpcd_readb(&port->aux, DP_DPCD_REV, &dpcd_rev);
+
+				if (ret == 1)
+					port->dpcd_rev = dpcd_rev;
+			}
+
+			if (!dpcd_rev)
+				DRM_DEBUG_KMS("Can't decide DPCD revision number!");
+		}
+
+		/*
+		 * Could be legacy sink, logical port etc on DP1.2.
+		 * Will get Nack under these cases when issue remote
+		 * DPCD read.
+		 */
+		if (ret != 1)
+			DRM_DEBUG_KMS("Can't access DPCD");
+	} else if (port->pdt == DP_PEER_DEVICE_NONE) {
+		port->dpcd_rev = 0;
+	}
+
+	/*
+	 * Release dc_sink for connector which unplug event is notified by CSN msg
+	 */
+	if (connection_status == connector_status_disconnected && aconnector->dc_sink) {
+		if (aconnector->dc_link->sink_count)
+			dc_link_remove_remote_sink(aconnector->dc_link, aconnector->dc_sink);
+
+		drm_dbg_dp(connector->dev,
+			   "DM_MST: remove remote sink 0x%p, %d remaining\n",
+			   aconnector->dc_link,
+			   aconnector->dc_link->sink_count);
+
+		dc_sink_release(aconnector->dc_sink);
+		aconnector->dc_sink = NULL;
+		amdgpu_dm_mst_reset_mst_connector_setting(aconnector);
+
+		amdgpu_dm_set_mst_status(&aconnector->mst_status,
+			MST_REMOTE_EDID | MST_ALLOCATE_NEW_PAYLOAD | MST_CLEAR_ALLOCATED_PAYLOAD,
+			false);
+	}
+
+	return connection_status;
+}
+
+static int dm_dp_mst_atomic_check(struct drm_connector *connector,
+				  struct drm_atomic_state *state)
+{
+	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector);
+	struct drm_dp_mst_topology_mgr *mst_mgr = &aconnector->mst_root->mst_mgr;
+	struct drm_dp_mst_port *mst_port = aconnector->mst_output_port;
+
+	return drm_dp_atomic_release_time_slots(state, mst_mgr, mst_port);
+}
+
+static const struct drm_connector_helper_funcs dm_dp_mst_connector_helper_funcs = {
+	.get_modes = dm_dp_mst_get_modes,
+	.mode_valid = amdgpu_dm_connector_mode_valid,
+	.atomic_best_encoder = dm_mst_atomic_best_encoder,
+	.detect_ctx = dm_dp_mst_detect,
+	.atomic_check = dm_dp_mst_atomic_check,
+};
+
+static void amdgpu_dm_encoder_destroy(struct drm_encoder *encoder)
+{
+	drm_encoder_cleanup(encoder);
+}
+
+static const struct drm_encoder_funcs amdgpu_dm_encoder_funcs = {
+	.destroy = amdgpu_dm_encoder_destroy,
+};
+
+void
+dm_dp_create_fake_mst_encoders(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev_to_drm(adev);
+	int i;
+
+	for (i = 0; i < adev->dm.display_indexes_num; i++) {
+		struct amdgpu_encoder *amdgpu_encoder = &adev->dm.mst_encoders[i];
+		struct drm_encoder *encoder = &amdgpu_encoder->base;
+
+		encoder->possible_crtcs = amdgpu_dm_get_encoder_crtc_mask(adev);
+
+		drm_encoder_init(
+			dev,
+			&amdgpu_encoder->base,
+			&amdgpu_dm_encoder_funcs,
+			DRM_MODE_ENCODER_DPMST,
+			NULL);
+
+		drm_encoder_helper_add(encoder, &amdgpu_dm_encoder_helper_funcs);
+	}
+}
+
+static struct drm_connector *
+dm_dp_add_mst_connector(struct drm_dp_mst_topology_mgr *mgr,
+			struct drm_dp_mst_port *port,
+			const char *pathprop)
+{
+	struct amdgpu_dm_connector *master = container_of(mgr, struct amdgpu_dm_connector, mst_mgr);
+	struct drm_device *dev = master->base.dev;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	struct amdgpu_dm_connector *aconnector;
+	struct drm_connector *connector;
+	int i;
+
+	aconnector = kzalloc(sizeof(*aconnector), GFP_KERNEL);
+	if (!aconnector)
+		return NULL;
+
+	DRM_DEBUG_DRIVER("%s: Create aconnector 0x%p for port 0x%p\n", __func__, aconnector, port);
+
+	connector = &aconnector->base;
+	aconnector->mst_output_port = port;
+	aconnector->mst_root = master;
+	amdgpu_dm_set_mst_status(&aconnector->mst_status,
+			MST_PROBE, true);
+
+	if (drm_connector_dynamic_init(
+		dev,
+		connector,
+		&dm_dp_mst_connector_funcs,
+		DRM_MODE_CONNECTOR_DisplayPort,
+		NULL)) {
+		kfree(aconnector);
+		return NULL;
+	}
+	drm_connector_helper_add(connector, &dm_dp_mst_connector_helper_funcs);
+
+	amdgpu_dm_connector_init_helper(
+		&adev->dm,
+		aconnector,
+		DRM_MODE_CONNECTOR_DisplayPort,
+		master->dc_link,
+		master->connector_id);
+
+	for (i = 0; i < adev->dm.display_indexes_num; i++) {
+		drm_connector_attach_encoder(&aconnector->base,
+					     &adev->dm.mst_encoders[i].base);
+	}
+
+	connector->max_bpc_property = master->base.max_bpc_property;
+	if (connector->max_bpc_property)
+		drm_connector_attach_max_bpc_property(connector, 8, 16);
+
+	connector->vrr_capable_property = master->base.vrr_capable_property;
+	if (connector->vrr_capable_property)
+		drm_connector_attach_vrr_capable_property(connector);
+
+	drm_object_attach_property(
+		&connector->base,
+		dev->mode_config.path_property,
+		0);
+	drm_object_attach_property(
+		&connector->base,
+		dev->mode_config.tile_property,
+		0);
+	connector->colorspace_property = master->base.colorspace_property;
+	if (connector->colorspace_property)
+		drm_connector_attach_colorspace_property(connector);
+
+	drm_connector_set_path_property(connector, pathprop);
+
+	if (!retrieve_branch_specific_data(aconnector))
+		aconnector->branch_ieee_oui = 0;
+
+	/*
+	 * Initialize connector state before adding the connectror to drm and
+	 * framebuffer lists
+	 */
+	amdgpu_dm_connector_funcs_reset(connector);
+
+	drm_dp_mst_get_port_malloc(port);
+
+	return connector;
+}
+
+void dm_handle_mst_sideband_msg_ready_event(
+	struct drm_dp_mst_topology_mgr *mgr,
+	enum mst_msg_ready_type msg_rdy_type)
+{
+	uint8_t esi[DP_PSR_ERROR_STATUS - DP_SINK_COUNT_ESI] = { 0 };
+	uint8_t dret;
+	bool new_irq_handled = false;
+	int dpcd_addr;
+	uint8_t dpcd_bytes_to_read;
+	const uint8_t max_process_count = 30;
+	uint8_t process_count = 0;
+	u8 retry;
+	struct amdgpu_dm_connector *aconnector =
+			container_of(mgr, struct amdgpu_dm_connector, mst_mgr);
+
+
+	const struct dc_link_status *link_status = dc_link_get_status(aconnector->dc_link);
+
+	if (link_status->dpcd_caps->dpcd_rev.raw < 0x12) {
+		dpcd_bytes_to_read = DP_LANE0_1_STATUS - DP_SINK_COUNT;
+		/* DPCD 0x200 - 0x201 for downstream IRQ */
+		dpcd_addr = DP_SINK_COUNT;
+	} else {
+		dpcd_bytes_to_read = DP_PSR_ERROR_STATUS - DP_SINK_COUNT_ESI;
+		/* DPCD 0x2002 - 0x2005 for downstream IRQ */
+		dpcd_addr = DP_SINK_COUNT_ESI;
+	}
+
+	mutex_lock(&aconnector->handle_mst_msg_ready);
+
+	while (process_count < max_process_count) {
+		u8 ack[DP_PSR_ERROR_STATUS - DP_SINK_COUNT_ESI] = {};
+
+		process_count++;
+
+		dret = drm_dp_dpcd_read(
+			&aconnector->dm_dp_aux.aux,
+			dpcd_addr,
+			esi,
+			dpcd_bytes_to_read);
+
+		if (dret != dpcd_bytes_to_read) {
+			DRM_DEBUG_KMS("DPCD read and acked number is not as expected!");
+			break;
+		}
+
+		DRM_DEBUG_DRIVER("ESI %02x %02x %02x\n", esi[0], esi[1], esi[2]);
+
+		switch (msg_rdy_type) {
+		case DOWN_REP_MSG_RDY_EVENT:
+			/* Only handle DOWN_REP_MSG_RDY case*/
+			esi[1] &= DP_DOWN_REP_MSG_RDY;
+			break;
+		case UP_REQ_MSG_RDY_EVENT:
+			/* Only handle UP_REQ_MSG_RDY case*/
+			esi[1] &= DP_UP_REQ_MSG_RDY;
+			break;
+		default:
+			/* Handle both cases*/
+			esi[1] &= (DP_DOWN_REP_MSG_RDY | DP_UP_REQ_MSG_RDY);
+			break;
+		}
+
+		if (!esi[1])
+			break;
+
+		/* handle MST irq */
+		if (aconnector->mst_mgr.mst_state)
+			drm_dp_mst_hpd_irq_handle_event(&aconnector->mst_mgr,
+						 esi,
+						 ack,
+						 &new_irq_handled);
+
+		if (new_irq_handled) {
+			/* ACK at DPCD to notify down stream */
+			for (retry = 0; retry < 3; retry++) {
+				ssize_t wret;
+
+				wret = drm_dp_dpcd_writeb(&aconnector->dm_dp_aux.aux,
+							  dpcd_addr + 1,
+							  ack[1]);
+				if (wret == 1)
+					break;
+			}
+
+			if (retry == 3) {
+				DRM_ERROR("Failed to ack MST event.\n");
+				break;
+			}
+
+			drm_dp_mst_hpd_irq_send_new_request(&aconnector->mst_mgr);
+
+			new_irq_handled = false;
+		} else {
+			break;
+		}
+	}
+
+	mutex_unlock(&aconnector->handle_mst_msg_ready);
+
+	if (process_count == max_process_count)
+		DRM_DEBUG_DRIVER("Loop exceeded max iterations\n");
+}
+
+static void dm_handle_mst_down_rep_msg_ready(struct drm_dp_mst_topology_mgr *mgr)
+{
+	dm_handle_mst_sideband_msg_ready_event(mgr, DOWN_REP_MSG_RDY_EVENT);
+}
+
+static const struct drm_dp_mst_topology_cbs dm_mst_cbs = {
+	.add_connector = dm_dp_add_mst_connector,
+	.poll_hpd_irq = dm_handle_mst_down_rep_msg_ready,
+};
+
+void amdgpu_dm_initialize_dp_connector(struct amdgpu_display_manager *dm,
+				       struct amdgpu_dm_connector *aconnector,
+				       int link_index)
+{
+	struct dc_link_settings max_link_enc_cap = {0};
+
+	aconnector->dm_dp_aux.aux.name =
+		kasprintf(GFP_KERNEL, "AMDGPU DM aux hw bus %d",
+			  link_index);
+	aconnector->dm_dp_aux.aux.transfer = dm_dp_aux_transfer;
+	aconnector->dm_dp_aux.aux.drm_dev = dm->ddev;
+	aconnector->dm_dp_aux.ddc_service = aconnector->dc_link->ddc;
+
+	drm_dp_aux_init(&aconnector->dm_dp_aux.aux);
+	drm_dp_cec_register_connector(&aconnector->dm_dp_aux.aux,
+				      &aconnector->base);
+	drm_dp_dpcd_set_probe(&aconnector->dm_dp_aux.aux, false);
+
+	if (aconnector->base.connector_type == DRM_MODE_CONNECTOR_eDP)
+		return;
+
+	dc_link_dp_get_max_link_enc_cap(aconnector->dc_link, &max_link_enc_cap);
+	aconnector->mst_mgr.cbs = &dm_mst_cbs;
+	drm_dp_mst_topology_mgr_init(&aconnector->mst_mgr, adev_to_drm(dm->adev),
+				     &aconnector->dm_dp_aux.aux, 16, 4, aconnector->connector_id);
+
+	drm_connector_attach_dp_subconnector_property(&aconnector->base);
+}
+
+uint32_t dm_mst_get_pbn_divider(struct dc_link *link)
+{
+	uint32_t pbn_div_x100;
+	uint64_t dividend, divisor;
+
+	if (!link)
+		return 0;
+
+	dividend = (uint64_t)dc_link_bandwidth_kbps(link, dc_link_get_link_cap(link)) * 100;
+	divisor = 8 * 1000 * 54;
+
+	pbn_div_x100 = div64_u64(dividend, divisor);
+
+	return dfixed_const(pbn_div_x100) / 100;
+}
+
+struct dsc_mst_fairness_params {
+	struct dc_crtc_timing *timing;
+	struct dc_sink *sink;
+	struct dc_dsc_bw_range bw_range;
+	bool compression_possible;
+	struct drm_dp_mst_port *port;
+	enum dsc_clock_force_state clock_force_enable;
+	uint32_t num_slices_h;
+	uint32_t num_slices_v;
+	uint32_t bpp_overwrite;
+	struct amdgpu_dm_connector *aconnector;
+};
+
+#if defined(CONFIG_DRM_AMD_DC_FP)
+static uint16_t get_fec_overhead_multiplier(struct dc_link *dc_link)
+{
+	u8 link_coding_cap;
+	uint16_t fec_overhead_multiplier_x1000 = PBN_FEC_OVERHEAD_MULTIPLIER_8B_10B;
+
+	link_coding_cap = dc_link_dp_mst_decide_link_encoding_format(dc_link);
+	if (link_coding_cap == DP_128b_132b_ENCODING)
+		fec_overhead_multiplier_x1000 = PBN_FEC_OVERHEAD_MULTIPLIER_128B_132B;
+
+	return fec_overhead_multiplier_x1000;
+}
+
+static int kbps_to_peak_pbn(int kbps, uint16_t fec_overhead_multiplier_x1000)
+{
+	u64 peak_kbps = kbps;
+
+	peak_kbps *= 1006;
+	peak_kbps *= fec_overhead_multiplier_x1000;
+	peak_kbps = div_u64(peak_kbps, 1000 * 1000);
+	return (int) DIV64_U64_ROUND_UP(peak_kbps * 64, (54 * 8 * 1000));
+}
+
+static void set_dsc_configs_from_fairness_vars(struct dsc_mst_fairness_params *params,
+		struct dsc_mst_fairness_vars *vars,
+		int count,
+		int k)
+{
+	struct drm_connector *drm_connector;
+	int i;
+	struct dc_dsc_config_options dsc_options = {0};
+
+	for (i = 0; i < count; i++) {
+		drm_connector = &params[i].aconnector->base;
+
+		dc_dsc_get_default_config_option(params[i].sink->ctx->dc, &dsc_options);
+		dsc_options.max_target_bpp_limit_override_x16 = drm_connector->display_info.max_dsc_bpp * 16;
+
+		memset(&params[i].timing->dsc_cfg, 0, sizeof(params[i].timing->dsc_cfg));
+		if (vars[i + k].dsc_enabled && dc_dsc_compute_config(
+					params[i].sink->ctx->dc->res_pool->dscs[0],
+					&params[i].sink->dsc_caps.dsc_dec_caps,
+					&dsc_options,
+					0,
+					params[i].timing,
+					dc_link_get_highest_encoding_format(params[i].aconnector->dc_link),
+					&params[i].timing->dsc_cfg)) {
+			params[i].timing->flags.DSC = 1;
+
+			if (params[i].bpp_overwrite)
+				params[i].timing->dsc_cfg.bits_per_pixel = params[i].bpp_overwrite;
+			else
+				params[i].timing->dsc_cfg.bits_per_pixel = vars[i + k].bpp_x16;
+
+			if (params[i].num_slices_h)
+				params[i].timing->dsc_cfg.num_slices_h = params[i].num_slices_h;
+
+			if (params[i].num_slices_v)
+				params[i].timing->dsc_cfg.num_slices_v = params[i].num_slices_v;
+		} else {
+			params[i].timing->flags.DSC = 0;
+		}
+		params[i].timing->dsc_cfg.mst_pbn = vars[i + k].pbn;
+	}
+
+	for (i = 0; i < count; i++) {
+		if (params[i].sink) {
+			if (params[i].sink->sink_signal != SIGNAL_TYPE_VIRTUAL &&
+				params[i].sink->sink_signal != SIGNAL_TYPE_NONE)
+				DRM_DEBUG_DRIVER("MST_DSC %s i=%d dispname=%s\n", __func__, i,
+					params[i].sink->edid_caps.display_name);
+		}
+
+		DRM_DEBUG_DRIVER("MST_DSC dsc=%d bits_per_pixel=%d pbn=%d\n",
+			params[i].timing->flags.DSC,
+			params[i].timing->dsc_cfg.bits_per_pixel,
+			vars[i + k].pbn);
+	}
+}
+
+static int bpp_x16_from_pbn(struct dsc_mst_fairness_params param, int pbn)
+{
+	struct dc_dsc_config dsc_config;
+	u64 kbps;
+
+	struct drm_connector *drm_connector = &param.aconnector->base;
+	struct dc_dsc_config_options dsc_options = {0};
+
+	dc_dsc_get_default_config_option(param.sink->ctx->dc, &dsc_options);
+	dsc_options.max_target_bpp_limit_override_x16 = drm_connector->display_info.max_dsc_bpp * 16;
+
+	kbps = div_u64((u64)pbn * 994 * 8 * 54, 64);
+	dc_dsc_compute_config(
+			param.sink->ctx->dc->res_pool->dscs[0],
+			&param.sink->dsc_caps.dsc_dec_caps,
+			&dsc_options,
+			(int) kbps, param.timing,
+			dc_link_get_highest_encoding_format(param.aconnector->dc_link),
+			&dsc_config);
+
+	return dsc_config.bits_per_pixel;
+}
+
+static int increase_dsc_bpp(struct drm_atomic_state *state,
+			    struct drm_dp_mst_topology_state *mst_state,
+			    struct dc_link *dc_link,
+			    struct dsc_mst_fairness_params *params,
+			    struct dsc_mst_fairness_vars *vars,
+			    int count,
+			    int k)
+{
+	int i;
+	bool bpp_increased[MAX_PIPES];
+	int initial_slack[MAX_PIPES];
+	int min_initial_slack;
+	int next_index;
+	int remaining_to_increase = 0;
+	int link_timeslots_used;
+	int fair_pbn_alloc;
+	int ret = 0;
+	uint16_t fec_overhead_multiplier_x1000 = get_fec_overhead_multiplier(dc_link);
+
+	for (i = 0; i < count; i++) {
+		if (vars[i + k].dsc_enabled) {
+			initial_slack[i] =
+			kbps_to_peak_pbn(params[i].bw_range.max_kbps, fec_overhead_multiplier_x1000) - vars[i + k].pbn;
+			bpp_increased[i] = false;
+			remaining_to_increase += 1;
+		} else {
+			initial_slack[i] = 0;
+			bpp_increased[i] = true;
+		}
+	}
+
+	while (remaining_to_increase) {
+		next_index = -1;
+		min_initial_slack = -1;
+		for (i = 0; i < count; i++) {
+			if (!bpp_increased[i]) {
+				if (min_initial_slack == -1 || min_initial_slack > initial_slack[i]) {
+					min_initial_slack = initial_slack[i];
+					next_index = i;
+				}
+			}
+		}
+
+		if (next_index == -1)
+			break;
+
+		link_timeslots_used = 0;
+
+		for (i = 0; i < count; i++)
+			link_timeslots_used += DIV_ROUND_UP(vars[i + k].pbn, dfixed_trunc(mst_state->pbn_div));
+
+		fair_pbn_alloc =
+			(63 - link_timeslots_used) / remaining_to_increase * dfixed_trunc(mst_state->pbn_div);
+
+		if (initial_slack[next_index] > fair_pbn_alloc) {
+			vars[next_index].pbn += fair_pbn_alloc;
+			ret = drm_dp_atomic_find_time_slots(state,
+							    params[next_index].port->mgr,
+							    params[next_index].port,
+							    vars[next_index].pbn);
+			if (ret < 0)
+				return ret;
+
+			ret = drm_dp_mst_atomic_check(state);
+			if (ret == 0) {
+				vars[next_index].bpp_x16 = bpp_x16_from_pbn(params[next_index], vars[next_index].pbn);
+			} else {
+				vars[next_index].pbn -= fair_pbn_alloc;
+				ret = drm_dp_atomic_find_time_slots(state,
+								    params[next_index].port->mgr,
+								    params[next_index].port,
+								    vars[next_index].pbn);
+				if (ret < 0)
+					return ret;
+			}
+		} else {
+			vars[next_index].pbn += initial_slack[next_index];
+			ret = drm_dp_atomic_find_time_slots(state,
+							    params[next_index].port->mgr,
+							    params[next_index].port,
+							    vars[next_index].pbn);
+			if (ret < 0)
+				return ret;
+
+			ret = drm_dp_mst_atomic_check(state);
+			if (ret == 0) {
+				vars[next_index].bpp_x16 = params[next_index].bw_range.max_target_bpp_x16;
+			} else {
+				vars[next_index].pbn -= initial_slack[next_index];
+				ret = drm_dp_atomic_find_time_slots(state,
+								    params[next_index].port->mgr,
+								    params[next_index].port,
+								    vars[next_index].pbn);
+				if (ret < 0)
+					return ret;
+			}
+		}
+
+		bpp_increased[next_index] = true;
+		remaining_to_increase--;
+	}
+	return 0;
+}
+
+static int try_disable_dsc(struct drm_atomic_state *state,
+			   struct dc_link *dc_link,
+			   struct dsc_mst_fairness_params *params,
+			   struct dsc_mst_fairness_vars *vars,
+			   int count,
+			   int k)
+{
+	int i;
+	bool tried[MAX_PIPES];
+	int kbps_increase[MAX_PIPES];
+	int max_kbps_increase;
+	int next_index;
+	int remaining_to_try = 0;
+	int ret;
+	uint16_t fec_overhead_multiplier_x1000 = get_fec_overhead_multiplier(dc_link);
+	int var_pbn;
+
+	for (i = 0; i < count; i++) {
+		if (vars[i + k].dsc_enabled
+				&& vars[i + k].bpp_x16 == params[i].bw_range.max_target_bpp_x16
+				&& params[i].clock_force_enable == DSC_CLK_FORCE_DEFAULT) {
+			kbps_increase[i] = params[i].bw_range.stream_kbps - params[i].bw_range.max_kbps;
+			tried[i] = false;
+			remaining_to_try += 1;
+		} else {
+			kbps_increase[i] = 0;
+			tried[i] = true;
+		}
+	}
+
+	while (remaining_to_try) {
+		next_index = -1;
+		max_kbps_increase = -1;
+		for (i = 0; i < count; i++) {
+			if (!tried[i]) {
+				if (max_kbps_increase == -1 || max_kbps_increase < kbps_increase[i]) {
+					max_kbps_increase = kbps_increase[i];
+					next_index = i;
+				}
+			}
+		}
+
+		if (next_index == -1)
+			break;
+
+		DRM_DEBUG_DRIVER("MST_DSC index #%d, try no compression\n", next_index);
+		var_pbn = vars[next_index].pbn;
+		vars[next_index].pbn = kbps_to_peak_pbn(params[next_index].bw_range.stream_kbps, fec_overhead_multiplier_x1000);
+		ret = drm_dp_atomic_find_time_slots(state,
+						    params[next_index].port->mgr,
+						    params[next_index].port,
+						    vars[next_index].pbn);
+		if (ret < 0) {
+			DRM_DEBUG_DRIVER("%s:%d MST_DSC index #%d, failed to set pbn to the state, %d\n",
+						__func__, __LINE__, next_index, ret);
+			vars[next_index].pbn = var_pbn;
+			return ret;
+		}
+
+		ret = drm_dp_mst_atomic_check(state);
+		if (ret == 0) {
+			DRM_DEBUG_DRIVER("MST_DSC index #%d, greedily disable dsc\n", next_index);
+			vars[next_index].dsc_enabled = false;
+			vars[next_index].bpp_x16 = 0;
+		} else {
+			DRM_DEBUG_DRIVER("MST_DSC index #%d, restore optimized pbn value\n", next_index);
+			vars[next_index].pbn = var_pbn;
+			ret = drm_dp_atomic_find_time_slots(state,
+							    params[next_index].port->mgr,
+							    params[next_index].port,
+							    vars[next_index].pbn);
+			if (ret < 0) {
+				DRM_DEBUG_DRIVER("%s:%d MST_DSC index #%d, failed to set pbn to the state, %d\n",
+							__func__, __LINE__, next_index, ret);
+				return ret;
+			}
+		}
+
+		tried[next_index] = true;
+		remaining_to_try--;
+	}
+	return 0;
+}
+
+static void log_dsc_params(int count, struct dsc_mst_fairness_vars *vars, int k)
+{
+	int i;
+
+	for (i = 0; i < count; i++)
+		DRM_DEBUG_DRIVER("MST_DSC DSC params: stream #%d --- dsc_enabled = %d, bpp_x16 = %d, pbn = %d\n",
+				 i, vars[i + k].dsc_enabled, vars[i + k].bpp_x16, vars[i + k].pbn);
+}
+
+static int compute_mst_dsc_configs_for_link(struct drm_atomic_state *state,
+					    struct dc_state *dc_state,
+					    struct dc_link *dc_link,
+					    struct dsc_mst_fairness_vars *vars,
+					    struct drm_dp_mst_topology_mgr *mgr,
+					    int *link_vars_start_index)
+{
+	struct dc_stream_state *stream;
+	struct dsc_mst_fairness_params params[MAX_PIPES];
+	struct amdgpu_dm_connector *aconnector;
+	struct drm_dp_mst_topology_state *mst_state = drm_atomic_get_mst_topology_state(state, mgr);
+	int count = 0;
+	int i, k, ret;
+	bool debugfs_overwrite = false;
+	uint16_t fec_overhead_multiplier_x1000 = get_fec_overhead_multiplier(dc_link);
+	struct drm_connector_state *new_conn_state;
+
+	memset(params, 0, sizeof(params));
+
+	if (IS_ERR(mst_state))
+		return PTR_ERR(mst_state);
+
+	/* Set up params */
+	DRM_DEBUG_DRIVER("%s: MST_DSC Try to set up params from %d streams\n", __func__, dc_state->stream_count);
+	for (i = 0; i < dc_state->stream_count; i++) {
+		struct dc_dsc_policy dsc_policy = {0};
+
+		stream = dc_state->streams[i];
+
+		if (stream->link != dc_link)
+			continue;
+
+		aconnector = (struct amdgpu_dm_connector *)stream->dm_stream_context;
+		if (!aconnector)
+			continue;
+
+		if (!aconnector->mst_output_port)
+			continue;
+
+		new_conn_state = drm_atomic_get_new_connector_state(state, &aconnector->base);
+
+		if (!new_conn_state) {
+			DRM_DEBUG_DRIVER("%s:%d MST_DSC Skip the stream 0x%p with invalid new_conn_state\n",
+					__func__, __LINE__, stream);
+			continue;
+		}
+
+		stream->timing.flags.DSC = 0;
+
+		params[count].timing = &stream->timing;
+		params[count].sink = stream->sink;
+		params[count].aconnector = aconnector;
+		params[count].port = aconnector->mst_output_port;
+		params[count].clock_force_enable = aconnector->dsc_settings.dsc_force_enable;
+		if (params[count].clock_force_enable == DSC_CLK_FORCE_ENABLE)
+			debugfs_overwrite = true;
+		params[count].num_slices_h = aconnector->dsc_settings.dsc_num_slices_h;
+		params[count].num_slices_v = aconnector->dsc_settings.dsc_num_slices_v;
+		params[count].bpp_overwrite = aconnector->dsc_settings.dsc_bits_per_pixel;
+		params[count].compression_possible = stream->sink->dsc_caps.dsc_dec_caps.is_dsc_supported;
+		dc_dsc_get_policy_for_timing(params[count].timing, 0, &dsc_policy, dc_link_get_highest_encoding_format(stream->link));
+		if (!dc_dsc_compute_bandwidth_range(
+				stream->sink->ctx->dc->res_pool->dscs[0],
+				stream->sink->ctx->dc->debug.dsc_min_slice_height_override,
+				dsc_policy.min_target_bpp * 16,
+				dsc_policy.max_target_bpp * 16,
+				&stream->sink->dsc_caps.dsc_dec_caps,
+				&stream->timing,
+				dc_link_get_highest_encoding_format(dc_link),
+				&params[count].bw_range))
+			params[count].bw_range.stream_kbps = dc_bandwidth_in_kbps_from_timing(&stream->timing,
+					dc_link_get_highest_encoding_format(dc_link));
+
+		DRM_DEBUG_DRIVER("MST_DSC #%d stream 0x%p - max_kbps = %u, min_kbps = %u, uncompressed_kbps = %u\n",
+			count, stream, params[count].bw_range.max_kbps, params[count].bw_range.min_kbps,
+			params[count].bw_range.stream_kbps);
+		count++;
+	}
+
+	DRM_DEBUG_DRIVER("%s: MST_DSC Params set up for %d streams\n", __func__, count);
+
+	if (count == 0) {
+		ASSERT(0);
+		return 0;
+	}
+
+	/* k is start index of vars for current phy link used by mst hub */
+	k = *link_vars_start_index;
+	/* set vars start index for next mst hub phy link */
+	*link_vars_start_index += count;
+
+	/* Try no compression */
+	DRM_DEBUG_DRIVER("MST_DSC Try no compression\n");
+	for (i = 0; i < count; i++) {
+		vars[i + k].aconnector = params[i].aconnector;
+		vars[i + k].pbn = kbps_to_peak_pbn(params[i].bw_range.stream_kbps, fec_overhead_multiplier_x1000);
+		vars[i + k].dsc_enabled = false;
+		vars[i + k].bpp_x16 = 0;
+		ret = drm_dp_atomic_find_time_slots(state, params[i].port->mgr, params[i].port,
+						    vars[i + k].pbn);
+		if (ret < 0)
+			return ret;
+	}
+	ret = drm_dp_mst_atomic_check(state);
+	if (ret == 0 && !debugfs_overwrite) {
+		set_dsc_configs_from_fairness_vars(params, vars, count, k);
+		return 0;
+	} else if (ret != -ENOSPC) {
+		return ret;
+	}
+
+	log_dsc_params(count, vars, k);
+
+	/* Try max compression */
+	DRM_DEBUG_DRIVER("MST_DSC Try max compression\n");
+	for (i = 0; i < count; i++) {
+		if (params[i].compression_possible && params[i].clock_force_enable != DSC_CLK_FORCE_DISABLE) {
+			vars[i + k].pbn = kbps_to_peak_pbn(params[i].bw_range.min_kbps, fec_overhead_multiplier_x1000);
+			vars[i + k].dsc_enabled = true;
+			vars[i + k].bpp_x16 = params[i].bw_range.min_target_bpp_x16;
+			ret = drm_dp_atomic_find_time_slots(state, params[i].port->mgr,
+							    params[i].port, vars[i + k].pbn);
+			if (ret < 0)
+				return ret;
+		} else {
+			vars[i + k].pbn = kbps_to_peak_pbn(params[i].bw_range.stream_kbps, fec_overhead_multiplier_x1000);
+			vars[i + k].dsc_enabled = false;
+			vars[i + k].bpp_x16 = 0;
+			ret = drm_dp_atomic_find_time_slots(state, params[i].port->mgr,
+							    params[i].port, vars[i + k].pbn);
+			if (ret < 0)
+				return ret;
+		}
+	}
+	ret = drm_dp_mst_atomic_check(state);
+	if (ret != 0)
+		return ret;
+
+	log_dsc_params(count, vars, k);
+
+	/* Optimize degree of compression */
+	DRM_DEBUG_DRIVER("MST_DSC Try optimize compression\n");
+	ret = increase_dsc_bpp(state, mst_state, dc_link, params, vars, count, k);
+	if (ret < 0) {
+		DRM_DEBUG_DRIVER("MST_DSC Failed to optimize compression\n");
+		return ret;
+	}
+
+	log_dsc_params(count, vars, k);
+
+	DRM_DEBUG_DRIVER("MST_DSC Try disable compression\n");
+	ret = try_disable_dsc(state, dc_link, params, vars, count, k);
+	if (ret < 0) {
+		DRM_DEBUG_DRIVER("MST_DSC Failed to disable compression\n");
+		return ret;
+	}
+
+	log_dsc_params(count, vars, k);
+
+	set_dsc_configs_from_fairness_vars(params, vars, count, k);
+
+	return 0;
+}
+
+static bool is_dsc_need_re_compute(
+	struct drm_atomic_state *state,
+	struct dc_state *dc_state,
+	struct dc_link *dc_link)
+{
+	int i, j;
+	bool is_dsc_need_re_compute = false;
+	struct amdgpu_dm_connector *stream_on_link[MAX_PIPES];
+	int new_stream_on_link_num = 0;
+	struct amdgpu_dm_connector *aconnector;
+	struct dc_stream_state *stream;
+	const struct dc *dc = dc_link->dc;
+
+	/* only check phy used by dsc mst branch */
+	if (dc_link->type != dc_connection_mst_branch)
+		goto out;
+
+	/* add a check for older MST DSC with no virtual DPCDs */
+	if (needs_dsc_aux_workaround(dc_link)  &&
+		(!(dc_link->dpcd_caps.dsc_caps.dsc_basic_caps.fields.dsc_support.DSC_SUPPORT ||
+		dc_link->dpcd_caps.dsc_caps.dsc_basic_caps.fields.dsc_support.DSC_PASSTHROUGH_SUPPORT)))
+		goto out;
+
+	for (i = 0; i < MAX_PIPES; i++)
+		stream_on_link[i] = NULL;
+
+	DRM_DEBUG_DRIVER("%s: MST_DSC check on %d streams in new dc_state\n", __func__, dc_state->stream_count);
+
+	/* check if there is mode change in new request */
+	for (i = 0; i < dc_state->stream_count; i++) {
+		struct drm_crtc_state *new_crtc_state;
+		struct drm_connector_state *new_conn_state;
+
+		stream = dc_state->streams[i];
+		if (!stream)
+			continue;
+
+		DRM_DEBUG_DRIVER("%s:%d MST_DSC checking #%d stream 0x%p\n", __func__, __LINE__, i, stream);
+
+		/* check if stream using the same link for mst */
+		if (stream->link != dc_link)
+			continue;
+
+		aconnector = (struct amdgpu_dm_connector *) stream->dm_stream_context;
+		if (!aconnector)
+			continue;
+
+		stream_on_link[new_stream_on_link_num] = aconnector;
+		new_stream_on_link_num++;
+
+		new_conn_state = drm_atomic_get_new_connector_state(state, &aconnector->base);
+		if (!new_conn_state) {
+			DRM_DEBUG_DRIVER("%s:%d MST_DSC no new_conn_state for stream 0x%p, aconnector 0x%p\n",
+					 __func__, __LINE__, stream, aconnector);
+			continue;
+		}
+
+		if (IS_ERR(new_conn_state))
+			continue;
+
+		if (!new_conn_state->crtc)
+			continue;
+
+		new_crtc_state = drm_atomic_get_new_crtc_state(state, new_conn_state->crtc);
+		if (!new_crtc_state) {
+			DRM_DEBUG_DRIVER("%s:%d MST_DSC no new_crtc_state for crtc of stream 0x%p, aconnector 0x%p\n",
+						__func__, __LINE__, stream, aconnector);
+			continue;
+		}
+
+		if (IS_ERR(new_crtc_state))
+			continue;
+
+		if (new_crtc_state->enable && new_crtc_state->active) {
+			if (new_crtc_state->mode_changed || new_crtc_state->active_changed ||
+					new_crtc_state->connectors_changed) {
+				DRM_DEBUG_DRIVER("%s:%d MST_DSC dsc recompute required."
+						 "stream 0x%p in new dc_state\n",
+						 __func__, __LINE__, stream);
+				is_dsc_need_re_compute = true;
+				goto out;
+			}
+		}
+	}
+
+	if (new_stream_on_link_num == 0) {
+		DRM_DEBUG_DRIVER("%s:%d MST_DSC no mode change request for streams in new dc_state\n",
+				 __func__, __LINE__);
+		is_dsc_need_re_compute = false;
+		goto out;
+	}
+
+	DRM_DEBUG_DRIVER("%s: MST_DSC check on %d streams in current dc_state\n",
+			 __func__, dc->current_state->stream_count);
+
+	/* check current_state if there stream on link but it is not in
+	 * new request state
+	 */
+	for (i = 0; i < dc->current_state->stream_count; i++) {
+		stream = dc->current_state->streams[i];
+		/* only check stream on the mst hub */
+		if (stream->link != dc_link)
+			continue;
+
+		aconnector = (struct amdgpu_dm_connector *)stream->dm_stream_context;
+		if (!aconnector)
+			continue;
+
+		for (j = 0; j < new_stream_on_link_num; j++) {
+			if (stream_on_link[j]) {
+				if (aconnector == stream_on_link[j])
+					break;
+			}
+		}
+
+		if (j == new_stream_on_link_num) {
+			/* not in new state */
+			DRM_DEBUG_DRIVER("%s:%d MST_DSC dsc recompute required."
+					 "stream 0x%p in current dc_state but not in new dc_state\n",
+						__func__, __LINE__, stream);
+			is_dsc_need_re_compute = true;
+			break;
+		}
+	}
+
+out:
+	DRM_DEBUG_DRIVER("%s: MST_DSC dsc recompute %s\n",
+			 __func__, is_dsc_need_re_compute ? "required" : "not required");
+
+	return is_dsc_need_re_compute;
+}
+
+int compute_mst_dsc_configs_for_state(struct drm_atomic_state *state,
+				      struct dc_state *dc_state,
+				      struct dsc_mst_fairness_vars *vars)
+{
+	int i, j;
+	struct dc_stream_state *stream;
+	bool computed_streams[MAX_PIPES];
+	struct amdgpu_dm_connector *aconnector;
+	struct drm_dp_mst_topology_mgr *mst_mgr;
+	struct resource_pool *res_pool;
+	int link_vars_start_index = 0;
+	int ret = 0;
+
+	for (i = 0; i < dc_state->stream_count; i++)
+		computed_streams[i] = false;
+
+	for (i = 0; i < dc_state->stream_count; i++) {
+		stream = dc_state->streams[i];
+		res_pool = stream->ctx->dc->res_pool;
+
+		if (stream->signal != SIGNAL_TYPE_DISPLAY_PORT_MST)
+			continue;
+
+		aconnector = (struct amdgpu_dm_connector *)stream->dm_stream_context;
+
+		DRM_DEBUG_DRIVER("%s: MST_DSC compute mst dsc configs for stream 0x%p, aconnector 0x%p\n",
+				__func__, stream, aconnector);
+
+		if (!aconnector || !aconnector->dc_sink || !aconnector->mst_output_port)
+			continue;
+
+		if (!aconnector->dc_sink->dsc_caps.dsc_dec_caps.is_dsc_supported)
+			continue;
+
+		if (computed_streams[i])
+			continue;
+
+		if (res_pool->funcs->remove_stream_from_ctx &&
+		    res_pool->funcs->remove_stream_from_ctx(stream->ctx->dc, dc_state, stream) != DC_OK)
+			return -EINVAL;
+
+		if (!is_dsc_need_re_compute(state, dc_state, stream->link))
+			continue;
+
+		mst_mgr = aconnector->mst_output_port->mgr;
+		ret = compute_mst_dsc_configs_for_link(state, dc_state, stream->link, vars, mst_mgr,
+						       &link_vars_start_index);
+		if (ret != 0)
+			return ret;
+
+		for (j = 0; j < dc_state->stream_count; j++) {
+			if (dc_state->streams[j]->link == stream->link)
+				computed_streams[j] = true;
+		}
+	}
+
+	for (i = 0; i < dc_state->stream_count; i++) {
+		stream = dc_state->streams[i];
+
+		if (stream->timing.flags.DSC == 1)
+			if (dc_stream_add_dsc_to_resource(stream->ctx->dc, dc_state, stream) != DC_OK) {
+				DRM_DEBUG_DRIVER("%s:%d MST_DSC Failed to request dsc hw resource for stream 0x%p\n",
+							__func__, __LINE__, stream);
+				return -EINVAL;
+			}
+	}
+
+	return ret;
+}
+
+static int pre_compute_mst_dsc_configs_for_state(struct drm_atomic_state *state,
+						 struct dc_state *dc_state,
+						 struct dsc_mst_fairness_vars *vars)
+{
+	int i, j;
+	struct dc_stream_state *stream;
+	bool computed_streams[MAX_PIPES];
+	struct amdgpu_dm_connector *aconnector;
+	struct drm_dp_mst_topology_mgr *mst_mgr;
+	int link_vars_start_index = 0;
+	int ret = 0;
+
+	for (i = 0; i < dc_state->stream_count; i++)
+		computed_streams[i] = false;
+
+	for (i = 0; i < dc_state->stream_count; i++) {
+		stream = dc_state->streams[i];
+
+		if (stream->signal != SIGNAL_TYPE_DISPLAY_PORT_MST)
+			continue;
+
+		aconnector = (struct amdgpu_dm_connector *)stream->dm_stream_context;
+
+		DRM_DEBUG_DRIVER("MST_DSC pre compute mst dsc configs for #%d stream 0x%p, aconnector 0x%p\n",
+					i, stream, aconnector);
+
+		if (!aconnector || !aconnector->dc_sink || !aconnector->mst_output_port)
+			continue;
+
+		if (!aconnector->dc_sink->dsc_caps.dsc_dec_caps.is_dsc_supported)
+			continue;
+
+		if (computed_streams[i])
+			continue;
+
+		if (!is_dsc_need_re_compute(state, dc_state, stream->link))
+			continue;
+
+		mst_mgr = aconnector->mst_output_port->mgr;
+		ret = compute_mst_dsc_configs_for_link(state, dc_state, stream->link, vars, mst_mgr,
+						       &link_vars_start_index);
+		if (ret != 0)
+			return ret;
+
+		for (j = 0; j < dc_state->stream_count; j++) {
+			if (dc_state->streams[j]->link == stream->link)
+				computed_streams[j] = true;
+		}
+	}
+
+	return ret;
+}
+
+static int find_crtc_index_in_state_by_stream(struct drm_atomic_state *state,
+					      struct dc_stream_state *stream)
+{
+	int i;
+	struct drm_crtc *crtc;
+	struct drm_crtc_state *new_state, *old_state;
+
+	for_each_oldnew_crtc_in_state(state, crtc, old_state, new_state, i) {
+		struct dm_crtc_state *dm_state = to_dm_crtc_state(new_state);
+
+		if (dm_state->stream == stream)
+			return i;
+	}
+	return -1;
+}
+
+static bool is_link_to_dschub(struct dc_link *dc_link)
+{
+	union dpcd_dsc_basic_capabilities *dsc_caps =
+			&dc_link->dpcd_caps.dsc_caps.dsc_basic_caps;
+
+	/* only check phy used by dsc mst branch */
+	if (dc_link->type != dc_connection_mst_branch)
+		return false;
+
+	if (!(dsc_caps->fields.dsc_support.DSC_SUPPORT ||
+	      dsc_caps->fields.dsc_support.DSC_PASSTHROUGH_SUPPORT))
+		return false;
+	return true;
+}
+
+static bool is_dsc_precompute_needed(struct drm_atomic_state *state)
+{
+	int i;
+	struct drm_crtc *crtc;
+	struct drm_crtc_state *old_crtc_state, *new_crtc_state;
+	bool ret = false;
+
+	for_each_oldnew_crtc_in_state(state, crtc, old_crtc_state, new_crtc_state, i) {
+		struct dm_crtc_state *dm_crtc_state = to_dm_crtc_state(new_crtc_state);
+
+		if (!amdgpu_dm_find_first_crtc_matching_connector(state, crtc)) {
+			ret =  false;
+			break;
+		}
+		if (dm_crtc_state->stream && dm_crtc_state->stream->link)
+			if (is_link_to_dschub(dm_crtc_state->stream->link))
+				ret = true;
+	}
+	return ret;
+}
+
+int pre_validate_dsc(struct drm_atomic_state *state,
+		     struct dm_atomic_state **dm_state_ptr,
+		     struct dsc_mst_fairness_vars *vars)
+{
+	int i;
+	struct dm_atomic_state *dm_state;
+	struct dc_state *local_dc_state = NULL;
+	int ret = 0;
+
+	if (!is_dsc_precompute_needed(state)) {
+		DRM_INFO_ONCE("%s:%d MST_DSC dsc precompute is not needed\n", __func__, __LINE__);
+		return 0;
+	}
+	ret = dm_atomic_get_state(state, dm_state_ptr);
+	if (ret != 0) {
+		DRM_INFO_ONCE("%s:%d MST_DSC dm_atomic_get_state() failed\n", __func__, __LINE__);
+		return ret;
+	}
+	dm_state = *dm_state_ptr;
+
+	/*
+	 * create local vailable for dc_state. copy content of streams of dm_state->context
+	 * to local variable. make sure stream pointer of local variable not the same as stream
+	 * from dm_state->context.
+	 */
+
+	local_dc_state = vmalloc(sizeof(struct dc_state));
+	if (!local_dc_state)
+		return -ENOMEM;
+	memcpy(local_dc_state, dm_state->context, sizeof(struct dc_state));
+
+	for (i = 0; i < local_dc_state->stream_count; i++) {
+		struct dc_stream_state *stream = dm_state->context->streams[i];
+		int ind = find_crtc_index_in_state_by_stream(state, stream);
+
+		if (ind >= 0) {
+			struct drm_connector *connector;
+			struct drm_connector_state *drm_new_conn_state;
+			struct dm_connector_state *dm_new_conn_state;
+			struct dm_crtc_state *dm_old_crtc_state;
+
+			connector =
+				amdgpu_dm_find_first_crtc_matching_connector(state,
+									     state->crtcs[ind].ptr);
+			if (!connector)
+				continue;
+
+			drm_new_conn_state =
+				drm_atomic_get_new_connector_state(state,
+								   connector);
+			dm_new_conn_state = to_dm_connector_state(drm_new_conn_state);
+			dm_old_crtc_state = to_dm_crtc_state(state->crtcs[ind].old_state);
+
+			local_dc_state->streams[i] =
+				create_validate_stream_for_sink(connector,
+								&state->crtcs[ind].new_state->mode,
+								dm_new_conn_state,
+								dm_old_crtc_state->stream);
+			if (local_dc_state->streams[i] == NULL) {
+				ret = -EINVAL;
+				break;
+			}
+		}
+	}
+
+	if (ret != 0)
+		goto clean_exit;
+
+	ret = pre_compute_mst_dsc_configs_for_state(state, local_dc_state, vars);
+	if (ret != 0) {
+		DRM_INFO_ONCE("%s:%d MST_DSC dsc pre_compute_mst_dsc_configs_for_state() failed\n",
+				__func__, __LINE__);
+		ret = -EINVAL;
+		goto clean_exit;
+	}
+
+	/*
+	 * compare local_streams -> timing  with dm_state->context,
+	 * if the same set crtc_state->mode-change = 0;
+	 */
+	for (i = 0; i < local_dc_state->stream_count; i++) {
+		struct dc_stream_state *stream = dm_state->context->streams[i];
+
+		if (local_dc_state->streams[i] &&
+		    dc_is_timing_changed(stream, local_dc_state->streams[i])) {
+			DRM_INFO_ONCE("%s:%d MST_DSC crtc[%d] needs mode_change\n", __func__, __LINE__, i);
+		} else {
+			int ind = find_crtc_index_in_state_by_stream(state, stream);
+
+			if (ind >= 0) {
+				DRM_INFO_ONCE("%s:%d MST_DSC no mode changed for stream 0x%p\n",
+						__func__, __LINE__, stream);
+				state->crtcs[ind].new_state->mode_changed = 0;
+			}
+		}
+	}
+clean_exit:
+	for (i = 0; i < local_dc_state->stream_count; i++) {
+		struct dc_stream_state *stream = dm_state->context->streams[i];
+
+		if (local_dc_state->streams[i] != stream)
+			dc_stream_release(local_dc_state->streams[i]);
+	}
+
+	vfree(local_dc_state);
+
+	return ret;
+}
+
+static uint32_t kbps_from_pbn(unsigned int pbn)
+{
+	uint64_t kbps = (uint64_t)pbn;
+
+	kbps *= (1000000 / PEAK_FACTOR_X1000);
+	kbps *= 8;
+	kbps *= 54;
+	kbps /= 64;
+
+	return (uint32_t)kbps;
+}
+
+static bool is_dsc_common_config_possible(struct dc_stream_state *stream,
+					  struct dc_dsc_bw_range *bw_range)
+{
+	struct dc_dsc_policy dsc_policy = {0};
+	bool is_dsc_possible;
+
+	dc_dsc_get_policy_for_timing(&stream->timing, 0, &dsc_policy, dc_link_get_highest_encoding_format(stream->link));
+	is_dsc_possible = dc_dsc_compute_bandwidth_range(stream->sink->ctx->dc->res_pool->dscs[0],
+							 stream->sink->ctx->dc->debug.dsc_min_slice_height_override,
+							 dsc_policy.min_target_bpp * 16,
+							 dsc_policy.max_target_bpp * 16,
+							 &stream->sink->dsc_caps.dsc_dec_caps,
+							 &stream->timing, dc_link_get_highest_encoding_format(stream->link), bw_range);
+
+	return is_dsc_possible;
+}
+#endif
+
+#if defined(CONFIG_DRM_AMD_DC_FP)
+static bool dp_get_link_current_set_bw(struct drm_dp_aux *aux, uint32_t *cur_link_bw)
+{
+	uint32_t total_data_bw_efficiency_x10000 = 0;
+	uint32_t link_rate_per_lane_kbps = 0;
+	enum dc_link_rate link_rate;
+	union lane_count_set lane_count;
+	u8 dp_link_encoding;
+	u8 link_bw_set = 0;
+	u8 data[16] = {0};
+
+	*cur_link_bw = 0;
+
+	if (drm_dp_dpcd_read(aux, DP_LINK_BW_SET, data, 16) != 16)
+		return false;
+
+	dp_link_encoding = data[DP_MAIN_LINK_CHANNEL_CODING_SET - DP_LINK_BW_SET];
+	link_bw_set = data[DP_LINK_BW_SET - DP_LINK_BW_SET];
+	lane_count.raw = data[DP_LANE_COUNT_SET - DP_LINK_BW_SET];
+
+	drm_dbg_dp(aux->drm_dev, "MST_DSC downlink setting: %d, 0x%x x %d\n",
+		   dp_link_encoding, link_bw_set, lane_count.bits.LANE_COUNT_SET);
+
+	switch (dp_link_encoding) {
+	case DP_8b_10b_ENCODING:
+		link_rate = link_bw_set;
+		link_rate_per_lane_kbps = link_rate * LINK_RATE_REF_FREQ_IN_KHZ * BITS_PER_DP_BYTE;
+		total_data_bw_efficiency_x10000 = DATA_EFFICIENCY_8b_10b_x10000;
+		total_data_bw_efficiency_x10000 /= 100;
+		total_data_bw_efficiency_x10000 *= DATA_EFFICIENCY_8b_10b_FEC_EFFICIENCY_x100;
+		break;
+	case DP_128b_132b_ENCODING:
+		switch (link_bw_set) {
+		case DP_LINK_BW_10:
+			link_rate = LINK_RATE_UHBR10;
+			break;
+		case DP_LINK_BW_13_5:
+			link_rate = LINK_RATE_UHBR13_5;
+			break;
+		case DP_LINK_BW_20:
+			link_rate = LINK_RATE_UHBR20;
+			break;
+		default:
+			return false;
+		}
+
+		link_rate_per_lane_kbps = link_rate * 10000;
+		total_data_bw_efficiency_x10000 = DATA_EFFICIENCY_128b_132b_x10000;
+		break;
+	default:
+		return false;
+	}
+
+	*cur_link_bw = link_rate_per_lane_kbps * lane_count.bits.LANE_COUNT_SET / 10000 * total_data_bw_efficiency_x10000;
+	return true;
+}
+#endif
+
+enum dc_status dm_dp_mst_is_port_support_mode(
+	struct amdgpu_dm_connector *aconnector,
+	struct dc_stream_state *stream)
+{
+#if defined(CONFIG_DRM_AMD_DC_FP)
+	int branch_max_throughput_mps = 0;
+	struct dc_link_settings cur_link_settings;
+	uint32_t end_to_end_bw_in_kbps = 0;
+	uint32_t root_link_bw_in_kbps = 0;
+	uint32_t virtual_channel_bw_in_kbps = 0;
+	struct dc_dsc_bw_range bw_range = {0};
+	struct dc_dsc_config_options dsc_options = {0};
+	uint32_t stream_kbps;
+
+	/* DSC unnecessary case
+	 * Check if timing could be supported within end-to-end BW
+	 */
+	stream_kbps =
+		dc_bandwidth_in_kbps_from_timing(&stream->timing,
+			dc_link_get_highest_encoding_format(stream->link));
+	cur_link_settings = stream->link->verified_link_cap;
+	root_link_bw_in_kbps = dc_link_bandwidth_kbps(aconnector->dc_link, &cur_link_settings);
+	virtual_channel_bw_in_kbps = kbps_from_pbn(aconnector->mst_output_port->full_pbn);
+
+	/* pick the end to end bw bottleneck */
+	end_to_end_bw_in_kbps = min(root_link_bw_in_kbps, virtual_channel_bw_in_kbps);
+
+	if (stream_kbps <= end_to_end_bw_in_kbps) {
+		DRM_DEBUG_DRIVER("MST_DSC no dsc required. End-to-end bw sufficient\n");
+		return DC_OK;
+	}
+
+	/*DSC necessary case*/
+	if (!aconnector->dsc_aux)
+		return DC_FAIL_BANDWIDTH_VALIDATE;
+
+	if (is_dsc_common_config_possible(stream, &bw_range)) {
+
+		/*capable of dsc passthough. dsc bitstream along the entire path*/
+		if (aconnector->mst_output_port->passthrough_aux) {
+			if (bw_range.min_kbps > end_to_end_bw_in_kbps) {
+				DRM_DEBUG_DRIVER("MST_DSC dsc passthrough and decode at endpoint"
+						 "Max dsc compression bw can't fit into end-to-end bw\n");
+				return DC_FAIL_BANDWIDTH_VALIDATE;
+			}
+		} else {
+			/*dsc bitstream decoded at the dp last link*/
+			struct drm_dp_mst_port *immediate_upstream_port = NULL;
+			uint32_t end_link_bw = 0;
+
+			/*Get last DP link BW capability. Mode shall be supported by Legacy peer*/
+			if (aconnector->mst_output_port->pdt != DP_PEER_DEVICE_DP_LEGACY_CONV &&
+				aconnector->mst_output_port->pdt != DP_PEER_DEVICE_NONE) {
+				if (aconnector->vc_full_pbn != aconnector->mst_output_port->full_pbn) {
+					dp_get_link_current_set_bw(&aconnector->mst_output_port->aux, &end_link_bw);
+					aconnector->vc_full_pbn = aconnector->mst_output_port->full_pbn;
+					aconnector->mst_local_bw = end_link_bw;
+				} else {
+					end_link_bw = aconnector->mst_local_bw;
+				}
+
+				if (end_link_bw > 0 &&
+				    stream_kbps > end_link_bw &&
+				    aconnector->branch_ieee_oui != DP_BRANCH_DEVICE_ID_90CC24) {
+					DRM_DEBUG_DRIVER("MST_DSC dsc decode at last link. "
+							 "Mode required bw can't fit into last link\n");
+					return DC_FAIL_BANDWIDTH_VALIDATE;
+				}
+			}
+
+			/*Get virtual channel bandwidth between source and the link before the last link*/
+			if (aconnector->mst_output_port->parent->port_parent)
+				immediate_upstream_port = aconnector->mst_output_port->parent->port_parent;
+
+			if (immediate_upstream_port) {
+				virtual_channel_bw_in_kbps = kbps_from_pbn(immediate_upstream_port->full_pbn);
+				virtual_channel_bw_in_kbps = min(root_link_bw_in_kbps, virtual_channel_bw_in_kbps);
+			} else {
+				/* For topology LCT 1 case - only one mstb*/
+				virtual_channel_bw_in_kbps = root_link_bw_in_kbps;
+			}
+
+			if (bw_range.min_kbps > virtual_channel_bw_in_kbps) {
+				DRM_DEBUG_DRIVER("MST_DSC dsc decode at last link."
+						 "Max dsc compression can't fit into MST available bw\n");
+				return DC_FAIL_BANDWIDTH_VALIDATE;
+			}
+		}
+
+		/*Confirm if we can obtain dsc config*/
+		dc_dsc_get_default_config_option(stream->link->dc, &dsc_options);
+		dsc_options.max_target_bpp_limit_override_x16 = aconnector->base.display_info.max_dsc_bpp * 16;
+		if (dc_dsc_compute_config(stream->sink->ctx->dc->res_pool->dscs[0],
+				&stream->sink->dsc_caps.dsc_dec_caps,
+				&dsc_options,
+				end_to_end_bw_in_kbps,
+				&stream->timing,
+				dc_link_get_highest_encoding_format(stream->link),
+				&stream->timing.dsc_cfg)) {
+			stream->timing.flags.DSC = 1;
+			DRM_DEBUG_DRIVER("MST_DSC require dsc and dsc config found\n");
+		} else {
+			DRM_DEBUG_DRIVER("MST_DSC require dsc but can't find appropriate dsc config\n");
+			return DC_FAIL_BANDWIDTH_VALIDATE;
+		}
+
+		/* check is mst dsc output bandwidth branch_overall_throughput_0_mps */
+		switch (stream->timing.pixel_encoding) {
+		case PIXEL_ENCODING_RGB:
+		case PIXEL_ENCODING_YCBCR444:
+			branch_max_throughput_mps =
+				aconnector->dc_sink->dsc_caps.dsc_dec_caps.branch_overall_throughput_0_mps;
+			break;
+		case PIXEL_ENCODING_YCBCR422:
+		case PIXEL_ENCODING_YCBCR420:
+			branch_max_throughput_mps =
+				aconnector->dc_sink->dsc_caps.dsc_dec_caps.branch_overall_throughput_1_mps;
+			break;
+		default:
+			break;
+		}
+
+		if (branch_max_throughput_mps != 0 &&
+			((stream->timing.pix_clk_100hz / 10) >  branch_max_throughput_mps * 1000)) {
+			DRM_DEBUG_DRIVER("MST_DSC require dsc but max throughput mps fails\n");
+			return DC_FAIL_BANDWIDTH_VALIDATE;
+		}
+	} else {
+		DRM_DEBUG_DRIVER("MST_DSC require dsc but can't find common dsc config\n");
+		return DC_FAIL_BANDWIDTH_VALIDATE;
+	}
+#endif
+	return DC_OK;
+}
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.h
new file mode 100644
index 000000000000..6f7ea684b555
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.h
@@ -0,0 +1,97 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_AMDGPU_DM_MST_TYPES_H__
+#define __DAL_AMDGPU_DM_MST_TYPES_H__
+
+#define DP_BRANCH_DEVICE_ID_90CC24 0x90CC24
+
+#define SYNAPTICS_RC_COMMAND       0x4B2
+#define SYNAPTICS_RC_RESULT        0x4B3
+#define SYNAPTICS_RC_LENGTH        0x4B8
+#define SYNAPTICS_RC_OFFSET        0x4BC
+#define SYNAPTICS_RC_DATA          0x4C0
+
+#define DP_BRANCH_VENDOR_SPECIFIC_START 0x50C
+
+/**
+ * Panamera MST Hub detection
+ * Offset DPCD 050Eh == 0x5A indicates cascaded MST hub case
+ * Check from beginning of branch device vendor specific field (050Ch)
+ */
+#define IS_SYNAPTICS_PANAMERA(branchDevName) (((int)branchDevName[4] & 0xF0) == 0x50 ? 1 : 0)
+#define BRANCH_HW_REVISION_PANAMERA_A2 0x10
+#define SYNAPTICS_CASCADED_HUB_ID  0x5A
+#define IS_SYNAPTICS_CASCADED_PANAMERA(devName, data) ((IS_SYNAPTICS_PANAMERA(devName) && ((int)data[2] == SYNAPTICS_CASCADED_HUB_ID)) ? 1 : 0)
+
+#define PBN_FEC_OVERHEAD_MULTIPLIER_8B_10B     1031
+#define PBN_FEC_OVERHEAD_MULTIPLIER_128B_132B  1000
+
+enum mst_msg_ready_type {
+	NONE_MSG_RDY_EVENT = 0,
+	DOWN_REP_MSG_RDY_EVENT = 1,
+	UP_REQ_MSG_RDY_EVENT = 2,
+	DOWN_OR_UP_MSG_RDY_EVENT = 3
+};
+
+struct amdgpu_display_manager;
+struct amdgpu_dm_connector;
+
+uint32_t dm_mst_get_pbn_divider(struct dc_link *link);
+
+void amdgpu_dm_initialize_dp_connector(struct amdgpu_display_manager *dm,
+				       struct amdgpu_dm_connector *aconnector,
+				       int link_index);
+
+void
+dm_dp_create_fake_mst_encoders(struct amdgpu_device *adev);
+
+void dm_handle_mst_sideband_msg_ready_event(
+	struct drm_dp_mst_topology_mgr *mgr,
+	enum mst_msg_ready_type msg_rdy_type);
+
+struct dsc_mst_fairness_vars {
+	int pbn;
+	bool dsc_enabled;
+	int bpp_x16;
+	struct amdgpu_dm_connector *aconnector;
+};
+
+int compute_mst_dsc_configs_for_state(struct drm_atomic_state *state,
+				      struct dc_state *dc_state,
+				      struct dsc_mst_fairness_vars *vars);
+
+bool needs_dsc_aux_workaround(struct dc_link *link);
+
+int pre_validate_dsc(struct drm_atomic_state *state,
+		     struct dm_atomic_state **dm_state_ptr,
+		     struct dsc_mst_fairness_vars *vars);
+
+enum dc_status dm_dp_mst_is_port_support_mode(
+	struct amdgpu_dm_connector *aconnector,
+	struct dc_stream_state *stream);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c
new file mode 100644
index 000000000000..e027798ece03
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c
@@ -0,0 +1,1911 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include <drm/drm_atomic_helper.h>
+#include <drm/drm_blend.h>
+#include "drm/drm_framebuffer.h"
+#include <drm/drm_gem_atomic_helper.h>
+#include <drm/drm_plane_helper.h>
+#include <drm/drm_gem_framebuffer_helper.h>
+#include <drm/drm_fourcc.h>
+
+#include "amdgpu.h"
+#include "dal_asic_id.h"
+#include "amdgpu_display.h"
+#include "amdgpu_dm_trace.h"
+#include "amdgpu_dm_plane.h"
+#include "gc/gc_11_0_0_offset.h"
+#include "gc/gc_11_0_0_sh_mask.h"
+
+/*
+ * TODO: these are currently initialized to rgb formats only.
+ * For future use cases we should either initialize them dynamically based on
+ * plane capabilities, or initialize this array to all formats, so internal drm
+ * check will succeed, and let DC implement proper check
+ */
+static const uint32_t rgb_formats[] = {
+	DRM_FORMAT_XRGB8888,
+	DRM_FORMAT_ARGB8888,
+	DRM_FORMAT_RGBA8888,
+	DRM_FORMAT_XRGB2101010,
+	DRM_FORMAT_XBGR2101010,
+	DRM_FORMAT_ARGB2101010,
+	DRM_FORMAT_ABGR2101010,
+	DRM_FORMAT_XRGB16161616,
+	DRM_FORMAT_XBGR16161616,
+	DRM_FORMAT_ARGB16161616,
+	DRM_FORMAT_ABGR16161616,
+	DRM_FORMAT_XBGR8888,
+	DRM_FORMAT_ABGR8888,
+	DRM_FORMAT_RGB565,
+};
+
+static const uint32_t overlay_formats[] = {
+	DRM_FORMAT_XRGB8888,
+	DRM_FORMAT_ARGB8888,
+	DRM_FORMAT_RGBA8888,
+	DRM_FORMAT_XBGR8888,
+	DRM_FORMAT_ABGR8888,
+	DRM_FORMAT_RGB565,
+	DRM_FORMAT_NV21,
+	DRM_FORMAT_NV12,
+	DRM_FORMAT_P010
+};
+
+static const uint32_t video_formats[] = {
+	DRM_FORMAT_NV21,
+	DRM_FORMAT_NV12,
+	DRM_FORMAT_P010
+};
+
+static const u32 cursor_formats[] = {
+	DRM_FORMAT_ARGB8888
+};
+
+enum dm_micro_swizzle {
+	MICRO_SWIZZLE_Z = 0,
+	MICRO_SWIZZLE_S = 1,
+	MICRO_SWIZZLE_D = 2,
+	MICRO_SWIZZLE_R = 3
+};
+
+const struct drm_format_info *amdgpu_dm_plane_get_format_info(u32 pixel_format, u64 modifier)
+{
+	return amdgpu_lookup_format_info(pixel_format, modifier);
+}
+
+void amdgpu_dm_plane_fill_blending_from_plane_state(const struct drm_plane_state *plane_state,
+			       bool *per_pixel_alpha, bool *pre_multiplied_alpha,
+			       bool *global_alpha, int *global_alpha_value)
+{
+	*per_pixel_alpha = false;
+	*pre_multiplied_alpha = true;
+	*global_alpha = false;
+	*global_alpha_value = 0xff;
+
+
+	if (plane_state->pixel_blend_mode == DRM_MODE_BLEND_PREMULTI ||
+		plane_state->pixel_blend_mode == DRM_MODE_BLEND_COVERAGE) {
+		static const uint32_t alpha_formats[] = {
+			DRM_FORMAT_ARGB8888,
+			DRM_FORMAT_RGBA8888,
+			DRM_FORMAT_ABGR8888,
+			DRM_FORMAT_ARGB2101010,
+			DRM_FORMAT_ABGR2101010,
+			DRM_FORMAT_ARGB16161616,
+			DRM_FORMAT_ABGR16161616,
+			DRM_FORMAT_ARGB16161616F,
+		};
+		uint32_t format = plane_state->fb->format->format;
+		unsigned int i;
+
+		for (i = 0; i < ARRAY_SIZE(alpha_formats); ++i) {
+			if (format == alpha_formats[i]) {
+				*per_pixel_alpha = true;
+				break;
+			}
+		}
+
+		if (*per_pixel_alpha && plane_state->pixel_blend_mode == DRM_MODE_BLEND_COVERAGE)
+			*pre_multiplied_alpha = false;
+	}
+
+	if (plane_state->alpha < 0xffff) {
+		*global_alpha = true;
+		*global_alpha_value = plane_state->alpha >> 8;
+	}
+}
+
+static void amdgpu_dm_plane_add_modifier(uint64_t **mods, uint64_t *size, uint64_t *cap, uint64_t mod)
+{
+	if (!*mods)
+		return;
+
+	if (*cap - *size < 1) {
+		uint64_t new_cap = *cap * 2;
+		uint64_t *new_mods = kmalloc_array(new_cap, sizeof(uint64_t), GFP_KERNEL);
+
+		if (!new_mods) {
+			kfree(*mods);
+			*mods = NULL;
+			return;
+		}
+
+		memcpy(new_mods, *mods, sizeof(uint64_t) * *size);
+		kfree(*mods);
+		*mods = new_mods;
+		*cap = new_cap;
+	}
+
+	(*mods)[*size] = mod;
+	*size += 1;
+}
+
+static bool amdgpu_dm_plane_modifier_has_dcc(uint64_t modifier)
+{
+	return IS_AMD_FMT_MOD(modifier) && AMD_FMT_MOD_GET(DCC, modifier);
+}
+
+static unsigned int amdgpu_dm_plane_modifier_gfx9_swizzle_mode(uint64_t modifier)
+{
+	if (modifier == DRM_FORMAT_MOD_LINEAR)
+		return 0;
+
+	return AMD_FMT_MOD_GET(TILE, modifier);
+}
+
+static void amdgpu_dm_plane_fill_gfx8_tiling_info_from_flags(struct dc_tiling_info *tiling_info,
+							     uint64_t tiling_flags)
+{
+	/* Fill GFX8 params */
+	if (AMDGPU_TILING_GET(tiling_flags, ARRAY_MODE) == DC_ARRAY_2D_TILED_THIN1) {
+		unsigned int bankw, bankh, mtaspect, tile_split, num_banks;
+
+		bankw = AMDGPU_TILING_GET(tiling_flags, BANK_WIDTH);
+		bankh = AMDGPU_TILING_GET(tiling_flags, BANK_HEIGHT);
+		mtaspect = AMDGPU_TILING_GET(tiling_flags, MACRO_TILE_ASPECT);
+		tile_split = AMDGPU_TILING_GET(tiling_flags, TILE_SPLIT);
+		num_banks = AMDGPU_TILING_GET(tiling_flags, NUM_BANKS);
+
+		tiling_info->gfxversion = DcGfxVersion8;
+		/* XXX fix me for VI */
+		tiling_info->gfx8.num_banks = num_banks;
+		tiling_info->gfx8.array_mode =
+				DC_ARRAY_2D_TILED_THIN1;
+		tiling_info->gfx8.tile_split = tile_split;
+		tiling_info->gfx8.bank_width = bankw;
+		tiling_info->gfx8.bank_height = bankh;
+		tiling_info->gfx8.tile_aspect = mtaspect;
+		tiling_info->gfx8.tile_mode =
+				DC_ADDR_SURF_MICRO_TILING_DISPLAY;
+	} else if (AMDGPU_TILING_GET(tiling_flags, ARRAY_MODE)
+			== DC_ARRAY_1D_TILED_THIN1) {
+		tiling_info->gfx8.array_mode = DC_ARRAY_1D_TILED_THIN1;
+	}
+
+	tiling_info->gfx8.pipe_config =
+			AMDGPU_TILING_GET(tiling_flags, PIPE_CONFIG);
+}
+
+static void amdgpu_dm_plane_fill_gfx9_tiling_info_from_device(const struct amdgpu_device *adev,
+							      struct dc_tiling_info *tiling_info)
+{
+	/* Fill GFX9 params */
+	tiling_info->gfx9.num_pipes =
+		adev->gfx.config.gb_addr_config_fields.num_pipes;
+	tiling_info->gfx9.num_banks =
+		adev->gfx.config.gb_addr_config_fields.num_banks;
+	tiling_info->gfx9.pipe_interleave =
+		adev->gfx.config.gb_addr_config_fields.pipe_interleave_size;
+	tiling_info->gfx9.num_shader_engines =
+		adev->gfx.config.gb_addr_config_fields.num_se;
+	tiling_info->gfx9.max_compressed_frags =
+		adev->gfx.config.gb_addr_config_fields.max_compress_frags;
+	tiling_info->gfx9.num_rb_per_se =
+		adev->gfx.config.gb_addr_config_fields.num_rb_per_se;
+	tiling_info->gfx9.shaderEnable = 1;
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(10, 3, 0))
+		tiling_info->gfx9.num_pkrs = adev->gfx.config.gb_addr_config_fields.num_pkrs;
+}
+
+static void amdgpu_dm_plane_fill_gfx9_tiling_info_from_modifier(const struct amdgpu_device *adev,
+								struct dc_tiling_info *tiling_info,
+								uint64_t modifier)
+{
+	unsigned int mod_bank_xor_bits = AMD_FMT_MOD_GET(BANK_XOR_BITS, modifier);
+	unsigned int mod_pipe_xor_bits = AMD_FMT_MOD_GET(PIPE_XOR_BITS, modifier);
+	unsigned int pkrs_log2 = AMD_FMT_MOD_GET(PACKERS, modifier);
+	unsigned int pipes_log2;
+
+	pipes_log2 = min(5u, mod_pipe_xor_bits);
+
+	amdgpu_dm_plane_fill_gfx9_tiling_info_from_device(adev, tiling_info);
+
+	if (!IS_AMD_FMT_MOD(modifier))
+		return;
+
+	tiling_info->gfx9.num_pipes = 1u << pipes_log2;
+	tiling_info->gfx9.num_shader_engines = 1u << (mod_pipe_xor_bits - pipes_log2);
+
+	if (adev->family >= AMDGPU_FAMILY_NV) {
+		tiling_info->gfx9.num_pkrs = 1u << pkrs_log2;
+	} else {
+		tiling_info->gfx9.num_banks = 1u << mod_bank_xor_bits;
+
+		/* for DCC we know it isn't rb aligned, so rb_per_se doesn't matter. */
+	}
+}
+
+static int amdgpu_dm_plane_validate_dcc(struct amdgpu_device *adev,
+					const enum surface_pixel_format format,
+					const enum dc_rotation_angle rotation,
+					const struct dc_tiling_info *tiling_info,
+					const struct dc_plane_dcc_param *dcc,
+					const struct dc_plane_address *address,
+					const struct plane_size *plane_size)
+{
+	struct dc *dc = adev->dm.dc;
+	struct dc_dcc_surface_param input;
+	struct dc_surface_dcc_cap output;
+
+	memset(&input, 0, sizeof(input));
+	memset(&output, 0, sizeof(output));
+
+	if (!dcc->enable)
+		return 0;
+
+	if (adev->family < AMDGPU_FAMILY_GC_12_0_0 &&
+	    format >= SURFACE_PIXEL_FORMAT_VIDEO_BEGIN)
+		return -EINVAL;
+
+	if (!dc->cap_funcs.get_dcc_compression_cap)
+		return -EINVAL;
+
+	input.format = format;
+	input.surface_size.width = plane_size->surface_size.width;
+	input.surface_size.height = plane_size->surface_size.height;
+	input.swizzle_mode = tiling_info->gfx9.swizzle;
+
+	if (rotation == ROTATION_ANGLE_0 || rotation == ROTATION_ANGLE_180)
+		input.scan = SCAN_DIRECTION_HORIZONTAL;
+	else if (rotation == ROTATION_ANGLE_90 || rotation == ROTATION_ANGLE_270)
+		input.scan = SCAN_DIRECTION_VERTICAL;
+
+	if (!dc->cap_funcs.get_dcc_compression_cap(dc, &input, &output))
+		return -EINVAL;
+
+	if (!output.capable)
+		return -EINVAL;
+
+	if (dcc->independent_64b_blks == 0 &&
+	    output.grph.rgb.independent_64b_blks != 0)
+		return -EINVAL;
+
+	return 0;
+}
+
+static int amdgpu_dm_plane_fill_gfx9_plane_attributes_from_modifiers(struct amdgpu_device *adev,
+								     const struct amdgpu_framebuffer *afb,
+								     const enum surface_pixel_format format,
+								     const enum dc_rotation_angle rotation,
+								     const struct plane_size *plane_size,
+								     struct dc_tiling_info *tiling_info,
+								     struct dc_plane_dcc_param *dcc,
+								     struct dc_plane_address *address)
+{
+	const uint64_t modifier = afb->base.modifier;
+	int ret = 0;
+
+	amdgpu_dm_plane_fill_gfx9_tiling_info_from_modifier(adev, tiling_info, modifier);
+	tiling_info->gfx9.swizzle = amdgpu_dm_plane_modifier_gfx9_swizzle_mode(modifier);
+	tiling_info->gfxversion = DcGfxVersion9;
+
+	if (amdgpu_dm_plane_modifier_has_dcc(modifier)) {
+		uint64_t dcc_address = afb->address + afb->base.offsets[1];
+		bool independent_64b_blks = AMD_FMT_MOD_GET(DCC_INDEPENDENT_64B, modifier);
+		bool independent_128b_blks = AMD_FMT_MOD_GET(DCC_INDEPENDENT_128B, modifier);
+
+		dcc->enable = 1;
+		dcc->meta_pitch = afb->base.pitches[1];
+		dcc->independent_64b_blks = independent_64b_blks;
+		if (AMD_FMT_MOD_GET(TILE_VERSION, modifier) >= AMD_FMT_MOD_TILE_VER_GFX10_RBPLUS) {
+			if (independent_64b_blks && independent_128b_blks)
+				dcc->dcc_ind_blk = hubp_ind_block_64b_no_128bcl;
+			else if (independent_128b_blks)
+				dcc->dcc_ind_blk = hubp_ind_block_128b;
+			else if (independent_64b_blks && !independent_128b_blks)
+				dcc->dcc_ind_blk = hubp_ind_block_64b;
+			else
+				dcc->dcc_ind_blk = hubp_ind_block_unconstrained;
+		} else {
+			if (independent_64b_blks)
+				dcc->dcc_ind_blk = hubp_ind_block_64b;
+			else
+				dcc->dcc_ind_blk = hubp_ind_block_unconstrained;
+		}
+
+		address->grph.meta_addr.low_part = lower_32_bits(dcc_address);
+		address->grph.meta_addr.high_part = upper_32_bits(dcc_address);
+	}
+
+	ret = amdgpu_dm_plane_validate_dcc(adev, format, rotation, tiling_info, dcc, address, plane_size);
+	if (ret)
+		drm_dbg_kms(adev_to_drm(adev), "amdgpu_dm_plane_validate_dcc: returned error: %d\n", ret);
+
+	return ret;
+}
+
+static int amdgpu_dm_plane_fill_gfx12_plane_attributes_from_modifiers(struct amdgpu_device *adev,
+								      const struct amdgpu_framebuffer *afb,
+								      const enum surface_pixel_format format,
+								      const enum dc_rotation_angle rotation,
+								      const struct plane_size *plane_size,
+								      struct dc_tiling_info *tiling_info,
+								      struct dc_plane_dcc_param *dcc,
+								      struct dc_plane_address *address)
+{
+	const uint64_t modifier = afb->base.modifier;
+	int ret = 0;
+
+	/* TODO: Most of this function shouldn't be needed on GFX12. */
+	amdgpu_dm_plane_fill_gfx9_tiling_info_from_device(adev, tiling_info);
+
+	tiling_info->gfx9.swizzle = amdgpu_dm_plane_modifier_gfx9_swizzle_mode(modifier);
+	tiling_info->gfxversion = DcGfxAddr3;
+
+	if (amdgpu_dm_plane_modifier_has_dcc(modifier)) {
+		int max_compressed_block = AMD_FMT_MOD_GET(DCC_MAX_COMPRESSED_BLOCK, modifier);
+
+		dcc->enable = 1;
+		dcc->independent_64b_blks = max_compressed_block == 0;
+
+		if (max_compressed_block == 0)
+			dcc->dcc_ind_blk = hubp_ind_block_64b;
+		else if (max_compressed_block == 1)
+			dcc->dcc_ind_blk = hubp_ind_block_128b;
+		else
+			dcc->dcc_ind_blk = hubp_ind_block_unconstrained;
+	}
+
+	/* TODO: This seems wrong because there is no DCC plane on GFX12. */
+	ret = amdgpu_dm_plane_validate_dcc(adev, format, rotation, tiling_info, dcc, address, plane_size);
+	if (ret)
+		drm_dbg_kms(adev_to_drm(adev), "amdgpu_dm_plane_validate_dcc: returned error: %d\n", ret);
+
+	return ret;
+}
+
+static void amdgpu_dm_plane_add_gfx10_1_modifiers(const struct amdgpu_device *adev,
+						  uint64_t **mods,
+						  uint64_t *size,
+						  uint64_t *capacity)
+{
+	int pipe_xor_bits = ilog2(adev->gfx.config.gb_addr_config_fields.num_pipes);
+
+	amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+				     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_R_X) |
+				     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX10) |
+				     AMD_FMT_MOD_SET(PIPE_XOR_BITS, pipe_xor_bits) |
+				     AMD_FMT_MOD_SET(DCC, 1) |
+				     AMD_FMT_MOD_SET(DCC_CONSTANT_ENCODE, 1) |
+				     AMD_FMT_MOD_SET(DCC_INDEPENDENT_64B, 1) |
+				     AMD_FMT_MOD_SET(DCC_MAX_COMPRESSED_BLOCK, AMD_FMT_MOD_DCC_BLOCK_64B));
+
+	amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+				     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_R_X) |
+				     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX10) |
+				     AMD_FMT_MOD_SET(PIPE_XOR_BITS, pipe_xor_bits) |
+				     AMD_FMT_MOD_SET(DCC, 1) |
+				     AMD_FMT_MOD_SET(DCC_RETILE, 1) |
+				     AMD_FMT_MOD_SET(DCC_CONSTANT_ENCODE, 1) |
+				     AMD_FMT_MOD_SET(DCC_INDEPENDENT_64B, 1) |
+				     AMD_FMT_MOD_SET(DCC_MAX_COMPRESSED_BLOCK, AMD_FMT_MOD_DCC_BLOCK_64B));
+
+	amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+				     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_R_X) |
+				     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX10) |
+				     AMD_FMT_MOD_SET(PIPE_XOR_BITS, pipe_xor_bits));
+
+	amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+				     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_S_X) |
+				     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX10) |
+				     AMD_FMT_MOD_SET(PIPE_XOR_BITS, pipe_xor_bits));
+
+
+	/* Only supported for 64bpp, will be filtered in amdgpu_dm_plane_format_mod_supported */
+	amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+				     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_D) |
+				     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX9));
+
+	amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+				     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_S) |
+				     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX9));
+}
+
+static void amdgpu_dm_plane_add_gfx9_modifiers(const struct amdgpu_device *adev,
+					       uint64_t **mods,
+					       uint64_t *size,
+					       uint64_t *capacity)
+{
+	int pipes = ilog2(adev->gfx.config.gb_addr_config_fields.num_pipes);
+	int pipe_xor_bits = min(8, pipes +
+				ilog2(adev->gfx.config.gb_addr_config_fields.num_se));
+	int bank_xor_bits = min(8 - pipe_xor_bits,
+				ilog2(adev->gfx.config.gb_addr_config_fields.num_banks));
+	int rb = ilog2(adev->gfx.config.gb_addr_config_fields.num_se) +
+		 ilog2(adev->gfx.config.gb_addr_config_fields.num_rb_per_se);
+
+
+	if (adev->family == AMDGPU_FAMILY_RV) {
+		/* Raven2 and later */
+		bool has_constant_encode = adev->asic_type > CHIP_RAVEN || adev->external_rev_id >= 0x81;
+
+		/*
+		 * No _D DCC swizzles yet because we only allow 32bpp, which
+		 * doesn't support _D on DCN
+		 */
+
+		if (has_constant_encode) {
+			amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+						     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_S_X) |
+						     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX9) |
+						     AMD_FMT_MOD_SET(PIPE_XOR_BITS, pipe_xor_bits) |
+						     AMD_FMT_MOD_SET(BANK_XOR_BITS, bank_xor_bits) |
+						     AMD_FMT_MOD_SET(DCC, 1) |
+						     AMD_FMT_MOD_SET(DCC_INDEPENDENT_64B, 1) |
+						     AMD_FMT_MOD_SET(DCC_MAX_COMPRESSED_BLOCK, AMD_FMT_MOD_DCC_BLOCK_64B) |
+						     AMD_FMT_MOD_SET(DCC_CONSTANT_ENCODE, 1));
+		}
+
+		amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+					     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_S_X) |
+					     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX9) |
+					     AMD_FMT_MOD_SET(PIPE_XOR_BITS, pipe_xor_bits) |
+					     AMD_FMT_MOD_SET(BANK_XOR_BITS, bank_xor_bits) |
+					     AMD_FMT_MOD_SET(DCC, 1) |
+					     AMD_FMT_MOD_SET(DCC_INDEPENDENT_64B, 1) |
+					     AMD_FMT_MOD_SET(DCC_MAX_COMPRESSED_BLOCK, AMD_FMT_MOD_DCC_BLOCK_64B) |
+					     AMD_FMT_MOD_SET(DCC_CONSTANT_ENCODE, 0));
+
+		if (has_constant_encode) {
+			amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+						     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_S_X) |
+						     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX9) |
+						     AMD_FMT_MOD_SET(PIPE_XOR_BITS, pipe_xor_bits) |
+						     AMD_FMT_MOD_SET(BANK_XOR_BITS, bank_xor_bits) |
+						     AMD_FMT_MOD_SET(DCC, 1) |
+						     AMD_FMT_MOD_SET(DCC_RETILE, 1) |
+						     AMD_FMT_MOD_SET(DCC_INDEPENDENT_64B, 1) |
+						     AMD_FMT_MOD_SET(DCC_MAX_COMPRESSED_BLOCK, AMD_FMT_MOD_DCC_BLOCK_64B) |
+						     AMD_FMT_MOD_SET(DCC_CONSTANT_ENCODE, 1) |
+						     AMD_FMT_MOD_SET(RB, rb) |
+						     AMD_FMT_MOD_SET(PIPE, pipes));
+		}
+
+		amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+					     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_S_X) |
+					     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX9) |
+					     AMD_FMT_MOD_SET(PIPE_XOR_BITS, pipe_xor_bits) |
+					     AMD_FMT_MOD_SET(BANK_XOR_BITS, bank_xor_bits) |
+					     AMD_FMT_MOD_SET(DCC, 1) |
+					     AMD_FMT_MOD_SET(DCC_RETILE, 1) |
+					     AMD_FMT_MOD_SET(DCC_INDEPENDENT_64B, 1) |
+					     AMD_FMT_MOD_SET(DCC_MAX_COMPRESSED_BLOCK, AMD_FMT_MOD_DCC_BLOCK_64B) |
+					     AMD_FMT_MOD_SET(DCC_CONSTANT_ENCODE, 0) |
+					     AMD_FMT_MOD_SET(RB, rb) |
+					     AMD_FMT_MOD_SET(PIPE, pipes));
+	}
+
+	/*
+	 * Only supported for 64bpp on Raven, will be filtered on format in
+	 * amdgpu_dm_plane_format_mod_supported.
+	 */
+	amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+				     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_D_X) |
+				     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX9) |
+				     AMD_FMT_MOD_SET(PIPE_XOR_BITS, pipe_xor_bits) |
+				     AMD_FMT_MOD_SET(BANK_XOR_BITS, bank_xor_bits));
+
+	if (adev->family == AMDGPU_FAMILY_RV) {
+		amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+					     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_S_X) |
+					     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX9) |
+					     AMD_FMT_MOD_SET(PIPE_XOR_BITS, pipe_xor_bits) |
+					     AMD_FMT_MOD_SET(BANK_XOR_BITS, bank_xor_bits));
+	}
+
+	/*
+	 * Only supported for 64bpp on Raven, will be filtered on format in
+	 * amdgpu_dm_plane_format_mod_supported.
+	 */
+	amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+				     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_D) |
+				     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX9));
+
+	if (adev->family == AMDGPU_FAMILY_RV) {
+		amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+					     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_S) |
+					     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX9));
+	}
+}
+
+static void amdgpu_dm_plane_add_gfx10_3_modifiers(const struct amdgpu_device *adev,
+						  uint64_t **mods,
+						  uint64_t *size,
+						  uint64_t *capacity)
+{
+	int pipe_xor_bits = ilog2(adev->gfx.config.gb_addr_config_fields.num_pipes);
+	int pkrs = ilog2(adev->gfx.config.gb_addr_config_fields.num_pkrs);
+
+	amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+				     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_R_X) |
+				     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX10_RBPLUS) |
+				     AMD_FMT_MOD_SET(PIPE_XOR_BITS, pipe_xor_bits) |
+				     AMD_FMT_MOD_SET(PACKERS, pkrs) |
+				     AMD_FMT_MOD_SET(DCC, 1) |
+				     AMD_FMT_MOD_SET(DCC_CONSTANT_ENCODE, 1) |
+				     AMD_FMT_MOD_SET(DCC_INDEPENDENT_64B, 1) |
+				     AMD_FMT_MOD_SET(DCC_INDEPENDENT_128B, 1) |
+				     AMD_FMT_MOD_SET(DCC_MAX_COMPRESSED_BLOCK, AMD_FMT_MOD_DCC_BLOCK_64B));
+
+	amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+				     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_R_X) |
+				     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX10_RBPLUS) |
+				     AMD_FMT_MOD_SET(PIPE_XOR_BITS, pipe_xor_bits) |
+				     AMD_FMT_MOD_SET(PACKERS, pkrs) |
+				     AMD_FMT_MOD_SET(DCC, 1) |
+				     AMD_FMT_MOD_SET(DCC_CONSTANT_ENCODE, 1) |
+				     AMD_FMT_MOD_SET(DCC_INDEPENDENT_128B, 1) |
+				     AMD_FMT_MOD_SET(DCC_MAX_COMPRESSED_BLOCK, AMD_FMT_MOD_DCC_BLOCK_128B));
+
+	amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+				     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_R_X) |
+				     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX10_RBPLUS) |
+				     AMD_FMT_MOD_SET(PIPE_XOR_BITS, pipe_xor_bits) |
+				     AMD_FMT_MOD_SET(PACKERS, pkrs) |
+				     AMD_FMT_MOD_SET(DCC, 1) |
+				     AMD_FMT_MOD_SET(DCC_RETILE, 1) |
+				     AMD_FMT_MOD_SET(DCC_CONSTANT_ENCODE, 1) |
+				     AMD_FMT_MOD_SET(DCC_INDEPENDENT_64B, 1) |
+				     AMD_FMT_MOD_SET(DCC_INDEPENDENT_128B, 1) |
+				     AMD_FMT_MOD_SET(DCC_MAX_COMPRESSED_BLOCK, AMD_FMT_MOD_DCC_BLOCK_64B));
+
+	amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+				     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_R_X) |
+				     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX10_RBPLUS) |
+				     AMD_FMT_MOD_SET(PIPE_XOR_BITS, pipe_xor_bits) |
+				     AMD_FMT_MOD_SET(PACKERS, pkrs) |
+				     AMD_FMT_MOD_SET(DCC, 1) |
+				     AMD_FMT_MOD_SET(DCC_RETILE, 1) |
+				     AMD_FMT_MOD_SET(DCC_CONSTANT_ENCODE, 1) |
+				     AMD_FMT_MOD_SET(DCC_INDEPENDENT_128B, 1) |
+				     AMD_FMT_MOD_SET(DCC_MAX_COMPRESSED_BLOCK, AMD_FMT_MOD_DCC_BLOCK_128B));
+
+	amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+				     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_R_X) |
+				     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX10_RBPLUS) |
+				     AMD_FMT_MOD_SET(PIPE_XOR_BITS, pipe_xor_bits) |
+				     AMD_FMT_MOD_SET(PACKERS, pkrs));
+
+	amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+				     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_S_X) |
+				     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX10_RBPLUS) |
+				     AMD_FMT_MOD_SET(PIPE_XOR_BITS, pipe_xor_bits) |
+				     AMD_FMT_MOD_SET(PACKERS, pkrs));
+
+	/* Only supported for 64bpp, will be filtered in amdgpu_dm_plane_format_mod_supported */
+	amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+				     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_D) |
+				     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX9));
+
+	amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+				     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_S) |
+				     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX9));
+}
+
+static void amdgpu_dm_plane_add_gfx11_modifiers(struct amdgpu_device *adev,
+		      uint64_t **mods, uint64_t *size, uint64_t *capacity)
+{
+	int num_pipes = 0;
+	int pipe_xor_bits = 0;
+	int num_pkrs = 0;
+	int pkrs = 0;
+	u32 gb_addr_config;
+	u8 i = 0;
+	unsigned int swizzle_r_x;
+	uint64_t modifier_r_x;
+	uint64_t modifier_dcc_best;
+	uint64_t modifier_dcc_4k;
+
+	/* TODO: GFX11 IP HW init hasnt finish and we get zero if we read from
+	 * adev->gfx.config.gb_addr_config_fields.num_{pkrs,pipes}
+	 */
+	gb_addr_config = RREG32_SOC15(GC, 0, regGB_ADDR_CONFIG);
+	ASSERT(gb_addr_config != 0);
+
+	num_pkrs = 1 << REG_GET_FIELD(gb_addr_config, GB_ADDR_CONFIG, NUM_PKRS);
+	pkrs = ilog2(num_pkrs);
+	num_pipes = 1 << REG_GET_FIELD(gb_addr_config, GB_ADDR_CONFIG, NUM_PIPES);
+	pipe_xor_bits = ilog2(num_pipes);
+
+	for (i = 0; i < 2; i++) {
+		/* Insert the best one first. */
+		/* R_X swizzle modes are the best for rendering and DCC requires them. */
+		if (num_pipes > 16)
+			swizzle_r_x = !i ? AMD_FMT_MOD_TILE_GFX11_256K_R_X : AMD_FMT_MOD_TILE_GFX9_64K_R_X;
+		else
+			swizzle_r_x = !i ? AMD_FMT_MOD_TILE_GFX9_64K_R_X : AMD_FMT_MOD_TILE_GFX11_256K_R_X;
+
+		modifier_r_x = AMD_FMT_MOD |
+			       AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX11) |
+			       AMD_FMT_MOD_SET(PIPE_XOR_BITS, pipe_xor_bits) |
+			       AMD_FMT_MOD_SET(TILE, swizzle_r_x) |
+			       AMD_FMT_MOD_SET(PACKERS, pkrs);
+
+		/* DCC_CONSTANT_ENCODE is not set because it can't vary with gfx11 (it's implied to be 1). */
+		modifier_dcc_best = modifier_r_x | AMD_FMT_MOD_SET(DCC, 1) |
+				    AMD_FMT_MOD_SET(DCC_INDEPENDENT_64B, 0) |
+				    AMD_FMT_MOD_SET(DCC_INDEPENDENT_128B, 1) |
+				    AMD_FMT_MOD_SET(DCC_MAX_COMPRESSED_BLOCK, AMD_FMT_MOD_DCC_BLOCK_128B);
+
+		/* DCC settings for 4K and greater resolutions. (required by display hw) */
+		modifier_dcc_4k = modifier_r_x | AMD_FMT_MOD_SET(DCC, 1) |
+				  AMD_FMT_MOD_SET(DCC_INDEPENDENT_64B, 1) |
+				  AMD_FMT_MOD_SET(DCC_INDEPENDENT_128B, 1) |
+				  AMD_FMT_MOD_SET(DCC_MAX_COMPRESSED_BLOCK, AMD_FMT_MOD_DCC_BLOCK_64B);
+
+		amdgpu_dm_plane_add_modifier(mods, size, capacity, modifier_dcc_best);
+		amdgpu_dm_plane_add_modifier(mods, size, capacity, modifier_dcc_4k);
+
+		amdgpu_dm_plane_add_modifier(mods, size, capacity, modifier_dcc_best | AMD_FMT_MOD_SET(DCC_RETILE, 1));
+		amdgpu_dm_plane_add_modifier(mods, size, capacity, modifier_dcc_4k | AMD_FMT_MOD_SET(DCC_RETILE, 1));
+
+		amdgpu_dm_plane_add_modifier(mods, size, capacity, modifier_r_x);
+	}
+
+	amdgpu_dm_plane_add_modifier(mods, size, capacity, AMD_FMT_MOD |
+				     AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX11) |
+				     AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX9_64K_D));
+}
+
+static void amdgpu_dm_plane_add_gfx12_modifiers(struct amdgpu_device *adev,
+		      uint64_t **mods, uint64_t *size, uint64_t *capacity)
+{
+	uint64_t ver = AMD_FMT_MOD | AMD_FMT_MOD_SET(TILE_VERSION, AMD_FMT_MOD_TILE_VER_GFX12);
+	uint64_t mod_256k = ver | AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX12_256K_2D);
+	uint64_t mod_64k = ver | AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX12_64K_2D);
+	uint64_t mod_4k = ver | AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX12_4K_2D);
+	uint64_t mod_256b = ver | AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX12_256B_2D);
+	uint64_t dcc = ver | AMD_FMT_MOD_SET(DCC, 1);
+	uint8_t max_comp_block[] = {2, 1, 0};
+	uint64_t max_comp_block_mod[ARRAY_SIZE(max_comp_block)] = {0};
+	uint8_t i = 0, j = 0;
+	uint64_t gfx12_modifiers[] = {mod_256k, mod_64k, mod_4k, mod_256b, DRM_FORMAT_MOD_LINEAR};
+
+	for (i = 0; i < ARRAY_SIZE(max_comp_block); i++)
+		max_comp_block_mod[i] = AMD_FMT_MOD_SET(DCC_MAX_COMPRESSED_BLOCK, max_comp_block[i]);
+
+	/* With DCC: Best choice should be kept first. Hence, add all 256k modifiers of different
+	 * max compressed blocks first and then move on to the next smaller sized layouts.
+	 * Do not add the linear modifier here, and hence the condition of size-1 for the loop
+	 */
+	for (j = 0; j < ARRAY_SIZE(gfx12_modifiers) - 1; j++)
+		for (i = 0; i < ARRAY_SIZE(max_comp_block); i++)
+			amdgpu_dm_plane_add_modifier(mods, size, capacity,
+						     ver | dcc | max_comp_block_mod[i] | gfx12_modifiers[j]);
+
+	/* Without DCC. Add all modifiers including linear at the end */
+	for (i = 0; i < ARRAY_SIZE(gfx12_modifiers); i++)
+		amdgpu_dm_plane_add_modifier(mods, size, capacity, gfx12_modifiers[i]);
+
+}
+
+static int amdgpu_dm_plane_get_plane_modifiers(struct amdgpu_device *adev, unsigned int plane_type, uint64_t **mods)
+{
+	uint64_t size = 0, capacity = 128;
+	*mods = NULL;
+
+	/* We have not hooked up any pre-GFX9 modifiers. */
+	if (adev->family < AMDGPU_FAMILY_AI)
+		return 0;
+
+	*mods = kmalloc_array(capacity, sizeof(uint64_t), GFP_KERNEL);
+
+	if (plane_type == DRM_PLANE_TYPE_CURSOR) {
+		amdgpu_dm_plane_add_modifier(mods, &size, &capacity, DRM_FORMAT_MOD_LINEAR);
+		amdgpu_dm_plane_add_modifier(mods, &size, &capacity, DRM_FORMAT_MOD_INVALID);
+		return *mods ? 0 : -ENOMEM;
+	}
+
+	switch (adev->family) {
+	case AMDGPU_FAMILY_AI:
+	case AMDGPU_FAMILY_RV:
+		amdgpu_dm_plane_add_gfx9_modifiers(adev, mods, &size, &capacity);
+		break;
+	case AMDGPU_FAMILY_NV:
+	case AMDGPU_FAMILY_VGH:
+	case AMDGPU_FAMILY_YC:
+	case AMDGPU_FAMILY_GC_10_3_6:
+	case AMDGPU_FAMILY_GC_10_3_7:
+		if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(10, 3, 0))
+			amdgpu_dm_plane_add_gfx10_3_modifiers(adev, mods, &size, &capacity);
+		else
+			amdgpu_dm_plane_add_gfx10_1_modifiers(adev, mods, &size, &capacity);
+		break;
+	case AMDGPU_FAMILY_GC_11_0_0:
+	case AMDGPU_FAMILY_GC_11_0_1:
+	case AMDGPU_FAMILY_GC_11_5_0:
+		amdgpu_dm_plane_add_gfx11_modifiers(adev, mods, &size, &capacity);
+		break;
+	case AMDGPU_FAMILY_GC_12_0_0:
+		amdgpu_dm_plane_add_gfx12_modifiers(adev, mods, &size, &capacity);
+		break;
+	}
+
+	amdgpu_dm_plane_add_modifier(mods, &size, &capacity, DRM_FORMAT_MOD_LINEAR);
+
+	/* INVALID marks the end of the list. */
+	amdgpu_dm_plane_add_modifier(mods, &size, &capacity, DRM_FORMAT_MOD_INVALID);
+
+	if (!*mods)
+		return -ENOMEM;
+
+	return 0;
+}
+
+static int amdgpu_dm_plane_get_plane_formats(const struct drm_plane *plane,
+					     const struct dc_plane_cap *plane_cap,
+					     uint32_t *formats, int max_formats)
+{
+	int i, num_formats = 0;
+
+	/*
+	 * TODO: Query support for each group of formats directly from
+	 * DC plane caps. This will require adding more formats to the
+	 * caps list.
+	 */
+
+	if (plane->type == DRM_PLANE_TYPE_PRIMARY ||
+		(plane_cap && plane_cap->type == DC_PLANE_TYPE_DCN_UNIVERSAL && plane->type != DRM_PLANE_TYPE_CURSOR)) {
+		for (i = 0; i < ARRAY_SIZE(rgb_formats); ++i) {
+			if (num_formats >= max_formats)
+				break;
+
+			formats[num_formats++] = rgb_formats[i];
+		}
+
+		if (plane_cap && plane_cap->pixel_format_support.nv12)
+			formats[num_formats++] = DRM_FORMAT_NV12;
+		if (plane_cap && plane_cap->pixel_format_support.p010)
+			formats[num_formats++] = DRM_FORMAT_P010;
+		if (plane_cap && plane_cap->pixel_format_support.fp16) {
+			formats[num_formats++] = DRM_FORMAT_XRGB16161616F;
+			formats[num_formats++] = DRM_FORMAT_ARGB16161616F;
+			formats[num_formats++] = DRM_FORMAT_XBGR16161616F;
+			formats[num_formats++] = DRM_FORMAT_ABGR16161616F;
+		}
+	} else {
+		switch (plane->type) {
+		case DRM_PLANE_TYPE_OVERLAY:
+			for (i = 0; i < ARRAY_SIZE(overlay_formats); ++i) {
+				if (num_formats >= max_formats)
+					break;
+
+				formats[num_formats++] = overlay_formats[i];
+			}
+			break;
+
+		case DRM_PLANE_TYPE_CURSOR:
+			for (i = 0; i < ARRAY_SIZE(cursor_formats); ++i) {
+				if (num_formats >= max_formats)
+					break;
+
+				formats[num_formats++] = cursor_formats[i];
+			}
+			break;
+
+		default:
+			break;
+		}
+	}
+
+	return num_formats;
+}
+
+int amdgpu_dm_plane_fill_plane_buffer_attributes(struct amdgpu_device *adev,
+			     const struct amdgpu_framebuffer *afb,
+			     const enum surface_pixel_format format,
+			     const enum dc_rotation_angle rotation,
+			     const uint64_t tiling_flags,
+			     struct dc_tiling_info *tiling_info,
+			     struct plane_size *plane_size,
+			     struct dc_plane_dcc_param *dcc,
+			     struct dc_plane_address *address,
+			     bool tmz_surface)
+{
+	const struct drm_framebuffer *fb = &afb->base;
+	int ret;
+
+	memset(tiling_info, 0, sizeof(*tiling_info));
+	memset(plane_size, 0, sizeof(*plane_size));
+	memset(dcc, 0, sizeof(*dcc));
+	memset(address, 0, sizeof(*address));
+
+	address->tmz_surface = tmz_surface;
+
+	if (format < SURFACE_PIXEL_FORMAT_VIDEO_BEGIN) {
+		uint64_t addr = afb->address + fb->offsets[0];
+
+		plane_size->surface_size.x = 0;
+		plane_size->surface_size.y = 0;
+		plane_size->surface_size.width = fb->width;
+		plane_size->surface_size.height = fb->height;
+		plane_size->surface_pitch =
+			fb->pitches[0] / fb->format->cpp[0];
+
+		address->type = PLN_ADDR_TYPE_GRAPHICS;
+		address->grph.addr.low_part = lower_32_bits(addr);
+		address->grph.addr.high_part = upper_32_bits(addr);
+	} else if (format < SURFACE_PIXEL_FORMAT_INVALID) {
+		uint64_t luma_addr = afb->address + fb->offsets[0];
+		uint64_t chroma_addr = afb->address + fb->offsets[1];
+
+		plane_size->surface_size.x = 0;
+		plane_size->surface_size.y = 0;
+		plane_size->surface_size.width = fb->width;
+		plane_size->surface_size.height = fb->height;
+		plane_size->surface_pitch =
+			fb->pitches[0] / fb->format->cpp[0];
+
+		plane_size->chroma_size.x = 0;
+		plane_size->chroma_size.y = 0;
+		/* TODO: set these based on surface format */
+		plane_size->chroma_size.width = fb->width / 2;
+		plane_size->chroma_size.height = fb->height / 2;
+
+		plane_size->chroma_pitch =
+			fb->pitches[1] / fb->format->cpp[1];
+
+		address->type = PLN_ADDR_TYPE_VIDEO_PROGRESSIVE;
+		address->video_progressive.luma_addr.low_part =
+			lower_32_bits(luma_addr);
+		address->video_progressive.luma_addr.high_part =
+			upper_32_bits(luma_addr);
+		address->video_progressive.chroma_addr.low_part =
+			lower_32_bits(chroma_addr);
+		address->video_progressive.chroma_addr.high_part =
+			upper_32_bits(chroma_addr);
+	}
+
+	if (adev->family >= AMDGPU_FAMILY_GC_12_0_0) {
+		ret = amdgpu_dm_plane_fill_gfx12_plane_attributes_from_modifiers(adev, afb, format,
+										 rotation, plane_size,
+										 tiling_info, dcc,
+										 address);
+		if (ret)
+			return ret;
+	} else if (adev->family >= AMDGPU_FAMILY_AI) {
+		ret = amdgpu_dm_plane_fill_gfx9_plane_attributes_from_modifiers(adev, afb, format,
+										rotation, plane_size,
+										tiling_info, dcc,
+										address);
+		if (ret)
+			return ret;
+	} else {
+		amdgpu_dm_plane_fill_gfx8_tiling_info_from_flags(tiling_info, tiling_flags);
+	}
+
+	return 0;
+}
+
+static int amdgpu_dm_plane_helper_prepare_fb(struct drm_plane *plane,
+					     struct drm_plane_state *new_state)
+{
+	struct amdgpu_framebuffer *afb;
+	struct drm_gem_object *obj;
+	struct amdgpu_device *adev;
+	struct amdgpu_bo *rbo;
+	struct dm_plane_state *dm_plane_state_new, *dm_plane_state_old;
+	uint32_t domain;
+	int r;
+
+	if (!new_state->fb) {
+		DRM_DEBUG_KMS("No FB bound\n");
+		return 0;
+	}
+
+	afb = to_amdgpu_framebuffer(new_state->fb);
+	obj = drm_gem_fb_get_obj(new_state->fb, 0);
+	if (!obj) {
+		DRM_ERROR("Failed to get obj from framebuffer\n");
+		return -EINVAL;
+	}
+
+	rbo = gem_to_amdgpu_bo(obj);
+	adev = amdgpu_ttm_adev(rbo->tbo.bdev);
+	r = amdgpu_bo_reserve(rbo, true);
+	if (r) {
+		drm_err(adev_to_drm(adev), "fail to reserve bo (%d)\n", r);
+		return r;
+	}
+
+	r = dma_resv_reserve_fences(rbo->tbo.base.resv, 1);
+	if (r) {
+		drm_err(adev_to_drm(adev), "reserving fence slot failed (%d)\n", r);
+		goto error_unlock;
+	}
+
+	if (plane->type != DRM_PLANE_TYPE_CURSOR)
+		domain = amdgpu_display_supported_domains(adev, rbo->flags);
+	else
+		domain = AMDGPU_GEM_DOMAIN_VRAM;
+
+	rbo->flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+	r = amdgpu_bo_pin(rbo, domain);
+	if (unlikely(r != 0)) {
+		if (r != -ERESTARTSYS)
+			DRM_ERROR("Failed to pin framebuffer with error %d\n", r);
+		goto error_unlock;
+	}
+
+	r = amdgpu_ttm_alloc_gart(&rbo->tbo);
+	if (unlikely(r != 0)) {
+		DRM_ERROR("%p bind failed\n", rbo);
+		goto error_unpin;
+	}
+
+	r = drm_gem_plane_helper_prepare_fb(plane, new_state);
+	if (unlikely(r != 0))
+		goto error_unpin;
+
+	amdgpu_bo_unreserve(rbo);
+
+	afb->address = amdgpu_bo_gpu_offset(rbo);
+
+	amdgpu_bo_ref(rbo);
+
+	/**
+	 * We don't do surface updates on planes that have been newly created,
+	 * but we also don't have the afb->address during atomic check.
+	 *
+	 * Fill in buffer attributes depending on the address here, but only on
+	 * newly created planes since they're not being used by DC yet and this
+	 * won't modify global state.
+	 */
+	dm_plane_state_old = to_dm_plane_state(plane->state);
+	dm_plane_state_new = to_dm_plane_state(new_state);
+
+	if (dm_plane_state_new->dc_state &&
+	    dm_plane_state_old->dc_state != dm_plane_state_new->dc_state) {
+		struct dc_plane_state *plane_state =
+			dm_plane_state_new->dc_state;
+
+		amdgpu_dm_plane_fill_plane_buffer_attributes(
+			adev, afb, plane_state->format, plane_state->rotation,
+			afb->tiling_flags,
+			&plane_state->tiling_info, &plane_state->plane_size,
+			&plane_state->dcc, &plane_state->address,
+			afb->tmz_surface);
+	}
+
+	return 0;
+
+error_unpin:
+	amdgpu_bo_unpin(rbo);
+
+error_unlock:
+	amdgpu_bo_unreserve(rbo);
+	return r;
+}
+
+static void amdgpu_dm_plane_helper_cleanup_fb(struct drm_plane *plane,
+					      struct drm_plane_state *old_state)
+{
+	struct amdgpu_bo *rbo;
+	int r;
+
+	if (!old_state->fb)
+		return;
+
+	rbo = gem_to_amdgpu_bo(old_state->fb->obj[0]);
+	r = amdgpu_bo_reserve(rbo, false);
+	if (unlikely(r)) {
+		DRM_ERROR("failed to reserve rbo before unpin\n");
+		return;
+	}
+
+	amdgpu_bo_unpin(rbo);
+	amdgpu_bo_unreserve(rbo);
+	amdgpu_bo_unref(&rbo);
+}
+
+static void amdgpu_dm_plane_get_min_max_dc_plane_scaling(struct drm_device *dev,
+					 struct drm_framebuffer *fb,
+					 int *min_downscale, int *max_upscale)
+{
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	struct dc *dc = adev->dm.dc;
+	/* Caps for all supported planes are the same on DCE and DCN 1 - 3 */
+	struct dc_plane_cap *plane_cap = &dc->caps.planes[0];
+
+	switch (fb->format->format) {
+	case DRM_FORMAT_P010:
+	case DRM_FORMAT_NV12:
+	case DRM_FORMAT_NV21:
+		*max_upscale = plane_cap->max_upscale_factor.nv12;
+		*min_downscale = plane_cap->max_downscale_factor.nv12;
+		break;
+
+	case DRM_FORMAT_XRGB16161616F:
+	case DRM_FORMAT_ARGB16161616F:
+	case DRM_FORMAT_XBGR16161616F:
+	case DRM_FORMAT_ABGR16161616F:
+		*max_upscale = plane_cap->max_upscale_factor.fp16;
+		*min_downscale = plane_cap->max_downscale_factor.fp16;
+		break;
+
+	default:
+		*max_upscale = plane_cap->max_upscale_factor.argb8888;
+		*min_downscale = plane_cap->max_downscale_factor.argb8888;
+		break;
+	}
+
+	/*
+	 * A factor of 1 in the plane_cap means to not allow scaling, ie. use a
+	 * scaling factor of 1.0 == 1000 units.
+	 */
+	if (*max_upscale == 1)
+		*max_upscale = 1000;
+
+	if (*min_downscale == 1)
+		*min_downscale = 1000;
+}
+
+int amdgpu_dm_plane_helper_check_state(struct drm_plane_state *state,
+				       struct drm_crtc_state *new_crtc_state)
+{
+	struct drm_framebuffer *fb = state->fb;
+	int min_downscale, max_upscale;
+	int min_scale = 0;
+	int max_scale = INT_MAX;
+
+	/* Plane enabled? Validate viewport and get scaling factors from plane caps. */
+	if (fb && state->crtc) {
+		/* Validate viewport to cover the case when only the position changes */
+		if (state->plane->type != DRM_PLANE_TYPE_CURSOR) {
+			int viewport_width = state->crtc_w;
+			int viewport_height = state->crtc_h;
+
+			if (state->crtc_x < 0)
+				viewport_width += state->crtc_x;
+			else if (state->crtc_x + state->crtc_w > new_crtc_state->mode.crtc_hdisplay)
+				viewport_width = new_crtc_state->mode.crtc_hdisplay - state->crtc_x;
+
+			if (state->crtc_y < 0)
+				viewport_height += state->crtc_y;
+			else if (state->crtc_y + state->crtc_h > new_crtc_state->mode.crtc_vdisplay)
+				viewport_height = new_crtc_state->mode.crtc_vdisplay - state->crtc_y;
+
+			if (viewport_width < 0 || viewport_height < 0) {
+				DRM_DEBUG_ATOMIC("Plane completely outside of screen\n");
+				return -EINVAL;
+			} else if (viewport_width < MIN_VIEWPORT_SIZE*2) { /* x2 for width is because of pipe-split. */
+				DRM_DEBUG_ATOMIC("Viewport width %d smaller than %d\n", viewport_width, MIN_VIEWPORT_SIZE*2);
+				return -EINVAL;
+			} else if (viewport_height < MIN_VIEWPORT_SIZE) {
+				DRM_DEBUG_ATOMIC("Viewport height %d smaller than %d\n", viewport_height, MIN_VIEWPORT_SIZE);
+				return -EINVAL;
+			}
+
+		}
+
+		/* Get min/max allowed scaling factors from plane caps. */
+		amdgpu_dm_plane_get_min_max_dc_plane_scaling(state->crtc->dev, fb,
+							     &min_downscale, &max_upscale);
+		/*
+		 * Convert to drm convention: 16.16 fixed point, instead of dc's
+		 * 1.0 == 1000. Also drm scaling is src/dst instead of dc's
+		 * dst/src, so min_scale = 1.0 / max_upscale, etc.
+		 */
+		min_scale = (1000 << 16) / max_upscale;
+		max_scale = (1000 << 16) / min_downscale;
+	}
+
+	return drm_atomic_helper_check_plane_state(
+		state, new_crtc_state, min_scale, max_scale, true, true);
+}
+
+int amdgpu_dm_plane_fill_dc_scaling_info(struct amdgpu_device *adev,
+				const struct drm_plane_state *state,
+				struct dc_scaling_info *scaling_info)
+{
+	int scale_w, scale_h, min_downscale, max_upscale;
+
+	memset(scaling_info, 0, sizeof(*scaling_info));
+
+	/* Source is fixed 16.16 but we ignore mantissa for now... */
+	scaling_info->src_rect.x = state->src_x >> 16;
+	scaling_info->src_rect.y = state->src_y >> 16;
+
+	/*
+	 * For reasons we don't (yet) fully understand a non-zero
+	 * src_y coordinate into an NV12 buffer can cause a
+	 * system hang on DCN1x.
+	 * To avoid hangs (and maybe be overly cautious)
+	 * let's reject both non-zero src_x and src_y.
+	 *
+	 * We currently know of only one use-case to reproduce a
+	 * scenario with non-zero src_x and src_y for NV12, which
+	 * is to gesture the YouTube Android app into full screen
+	 * on ChromeOS.
+	 */
+	if (((amdgpu_ip_version(adev, DCE_HWIP, 0) == IP_VERSION(1, 0, 0)) ||
+	    (amdgpu_ip_version(adev, DCE_HWIP, 0) == IP_VERSION(1, 0, 1))) &&
+	    (state->fb && state->fb->format->format == DRM_FORMAT_NV12 &&
+	    (scaling_info->src_rect.x != 0 || scaling_info->src_rect.y != 0)))
+		return -EINVAL;
+
+	scaling_info->src_rect.width = state->src_w >> 16;
+	if (scaling_info->src_rect.width == 0)
+		return -EINVAL;
+
+	scaling_info->src_rect.height = state->src_h >> 16;
+	if (scaling_info->src_rect.height == 0)
+		return -EINVAL;
+
+	scaling_info->dst_rect.x = state->crtc_x;
+	scaling_info->dst_rect.y = state->crtc_y;
+
+	if (state->crtc_w == 0)
+		return -EINVAL;
+
+	scaling_info->dst_rect.width = state->crtc_w;
+
+	if (state->crtc_h == 0)
+		return -EINVAL;
+
+	scaling_info->dst_rect.height = state->crtc_h;
+
+	/* DRM doesn't specify clipping on destination output. */
+	scaling_info->clip_rect = scaling_info->dst_rect;
+
+	/* Validate scaling per-format with DC plane caps */
+	if (state->plane && state->plane->dev && state->fb) {
+		amdgpu_dm_plane_get_min_max_dc_plane_scaling(state->plane->dev, state->fb,
+							     &min_downscale, &max_upscale);
+	} else {
+		min_downscale = 250;
+		max_upscale = 16000;
+	}
+
+	scale_w = scaling_info->dst_rect.width * 1000 /
+		  scaling_info->src_rect.width;
+
+	if (scale_w < min_downscale || scale_w > max_upscale)
+		return -EINVAL;
+
+	scale_h = scaling_info->dst_rect.height * 1000 /
+		  scaling_info->src_rect.height;
+
+	if (scale_h < min_downscale || scale_h > max_upscale)
+		return -EINVAL;
+
+	/*
+	 * The "scaling_quality" can be ignored for now, quality = 0 has DC
+	 * assume reasonable defaults based on the format.
+	 */
+
+	return 0;
+}
+
+static int amdgpu_dm_plane_atomic_check(struct drm_plane *plane,
+					struct drm_atomic_state *state)
+{
+	struct drm_plane_state *new_plane_state = drm_atomic_get_new_plane_state(state,
+										 plane);
+	struct amdgpu_device *adev = drm_to_adev(plane->dev);
+	struct dc *dc = adev->dm.dc;
+	struct dm_plane_state *dm_plane_state;
+	struct dc_scaling_info scaling_info;
+	struct drm_crtc_state *new_crtc_state;
+	int ret;
+
+	trace_amdgpu_dm_plane_atomic_check(new_plane_state);
+
+	dm_plane_state = to_dm_plane_state(new_plane_state);
+
+	if (!dm_plane_state->dc_state)
+		return 0;
+
+	new_crtc_state =
+		drm_atomic_get_new_crtc_state(state,
+					      new_plane_state->crtc);
+	if (!new_crtc_state)
+		return -EINVAL;
+
+	ret = amdgpu_dm_plane_helper_check_state(new_plane_state, new_crtc_state);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_dm_plane_fill_dc_scaling_info(adev, new_plane_state, &scaling_info);
+	if (ret)
+		return ret;
+
+	if (dc_validate_plane(dc, dm_plane_state->dc_state) == DC_OK)
+		return 0;
+
+	return -EINVAL;
+}
+
+static int amdgpu_dm_plane_atomic_async_check(struct drm_plane *plane,
+					      struct drm_atomic_state *state, bool flip)
+{
+	struct drm_crtc_state *new_crtc_state;
+	struct drm_plane_state *new_plane_state;
+	struct dm_crtc_state *dm_new_crtc_state;
+
+	if (flip) {
+		if (plane->type != DRM_PLANE_TYPE_OVERLAY)
+			return -EINVAL;
+	} else if (plane->type != DRM_PLANE_TYPE_CURSOR) {
+		return -EINVAL;
+	}
+
+	new_plane_state = drm_atomic_get_new_plane_state(state, plane);
+	new_crtc_state = drm_atomic_get_new_crtc_state(state, new_plane_state->crtc);
+	dm_new_crtc_state = to_dm_crtc_state(new_crtc_state);
+	/* Reject overlay cursors for now*/
+	if (!flip && dm_new_crtc_state->cursor_mode == DM_CURSOR_OVERLAY_MODE)
+		return -EINVAL;
+
+	return 0;
+}
+
+int amdgpu_dm_plane_get_cursor_position(struct drm_plane *plane, struct drm_crtc *crtc,
+					struct dc_cursor_position *position)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = drm_to_adev(plane->dev);
+	int x, y;
+	int xorigin = 0, yorigin = 0;
+
+	if (!crtc || !plane->state->fb)
+		return 0;
+
+	if ((plane->state->crtc_w > amdgpu_crtc->max_cursor_width) ||
+	    (plane->state->crtc_h > amdgpu_crtc->max_cursor_height)) {
+		DRM_ERROR("%s: bad cursor width or height %d x %d\n",
+			  __func__,
+			  plane->state->crtc_w,
+			  plane->state->crtc_h);
+		return -EINVAL;
+	}
+
+	x = plane->state->crtc_x;
+	y = plane->state->crtc_y;
+
+	if (x <= -amdgpu_crtc->max_cursor_width ||
+	    y <= -amdgpu_crtc->max_cursor_height)
+		return 0;
+
+	if (x < 0) {
+		xorigin = min(-x, amdgpu_crtc->max_cursor_width - 1);
+		x = 0;
+	}
+	if (y < 0) {
+		yorigin = min(-y, amdgpu_crtc->max_cursor_height - 1);
+		y = 0;
+	}
+	position->enable = true;
+	position->x = x;
+	position->y = y;
+	position->x_hotspot = xorigin;
+	position->y_hotspot = yorigin;
+
+	if (amdgpu_ip_version(adev, DCE_HWIP, 0) < IP_VERSION(4, 0, 1))
+		position->translate_by_source = true;
+
+	return 0;
+}
+
+void amdgpu_dm_plane_handle_cursor_update(struct drm_plane *plane,
+				 struct drm_plane_state *old_plane_state)
+{
+	struct amdgpu_device *adev = drm_to_adev(plane->dev);
+	struct amdgpu_framebuffer *afb = to_amdgpu_framebuffer(plane->state->fb);
+	struct drm_crtc *crtc = afb ? plane->state->crtc : old_plane_state->crtc;
+	struct dm_crtc_state *crtc_state = crtc ? to_dm_crtc_state(crtc->state) : NULL;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	uint64_t address = afb ? afb->address : 0;
+	struct dc_cursor_position position = {0};
+	struct dc_cursor_attributes attributes;
+	int ret;
+
+	if (!plane->state->fb && !old_plane_state->fb)
+		return;
+
+	drm_dbg_atomic(plane->dev, "crtc_id=%d with size %d to %d\n",
+		       amdgpu_crtc->crtc_id, plane->state->crtc_w,
+		       plane->state->crtc_h);
+
+	ret = amdgpu_dm_plane_get_cursor_position(plane, crtc, &position);
+	if (ret)
+		return;
+
+	if (!position.enable) {
+		/* turn off cursor */
+		if (crtc_state && crtc_state->stream) {
+			mutex_lock(&adev->dm.dc_lock);
+			dc_stream_program_cursor_position(crtc_state->stream,
+						      &position);
+			mutex_unlock(&adev->dm.dc_lock);
+		}
+		return;
+	}
+
+	amdgpu_crtc->cursor_width = plane->state->crtc_w;
+	amdgpu_crtc->cursor_height = plane->state->crtc_h;
+
+	memset(&attributes, 0, sizeof(attributes));
+	attributes.address.high_part = upper_32_bits(address);
+	attributes.address.low_part  = lower_32_bits(address);
+	attributes.width             = plane->state->crtc_w;
+	attributes.height            = plane->state->crtc_h;
+	attributes.color_format      = CURSOR_MODE_COLOR_PRE_MULTIPLIED_ALPHA;
+	attributes.rotation_angle    = 0;
+	attributes.attribute_flags.value = 0;
+
+	/* Enable cursor degamma ROM on DCN3+ for implicit sRGB degamma in DRM
+	 * legacy gamma setup.
+	 */
+	if (crtc_state->cm_is_degamma_srgb &&
+	    adev->dm.dc->caps.color.dpp.gamma_corr)
+		attributes.attribute_flags.bits.ENABLE_CURSOR_DEGAMMA = 1;
+
+	if (afb)
+		attributes.pitch = afb->base.pitches[0] / afb->base.format->cpp[0];
+
+	if (crtc_state->stream) {
+		mutex_lock(&adev->dm.dc_lock);
+		if (!dc_stream_program_cursor_attributes(crtc_state->stream,
+							 &attributes))
+			DRM_ERROR("DC failed to set cursor attributes\n");
+
+		if (!dc_stream_program_cursor_position(crtc_state->stream,
+						   &position))
+			DRM_ERROR("DC failed to set cursor position\n");
+		mutex_unlock(&adev->dm.dc_lock);
+	}
+}
+
+static void amdgpu_dm_plane_atomic_async_update(struct drm_plane *plane,
+						struct drm_atomic_state *state)
+{
+	struct drm_plane_state *new_state = drm_atomic_get_new_plane_state(state,
+									   plane);
+	struct drm_plane_state *old_state =
+		drm_atomic_get_old_plane_state(state, plane);
+
+	trace_amdgpu_dm_atomic_update_cursor(new_state);
+
+	swap(plane->state->fb, new_state->fb);
+
+	plane->state->src_x = new_state->src_x;
+	plane->state->src_y = new_state->src_y;
+	plane->state->src_w = new_state->src_w;
+	plane->state->src_h = new_state->src_h;
+	plane->state->crtc_x = new_state->crtc_x;
+	plane->state->crtc_y = new_state->crtc_y;
+	plane->state->crtc_w = new_state->crtc_w;
+	plane->state->crtc_h = new_state->crtc_h;
+
+	amdgpu_dm_plane_handle_cursor_update(plane, old_state);
+}
+
+static void amdgpu_dm_plane_panic_flush(struct drm_plane *plane)
+{
+	struct dm_plane_state *dm_plane_state = to_dm_plane_state(plane->state);
+	struct drm_framebuffer *fb = plane->state->fb;
+	struct dc_plane_state *dc_plane_state;
+
+	if (!dm_plane_state || !dm_plane_state->dc_state)
+		return;
+
+	dc_plane_state = dm_plane_state->dc_state;
+
+	dc_plane_force_dcc_and_tiling_disable(dc_plane_state, fb->modifier ? true : false);
+}
+
+static const struct drm_plane_helper_funcs dm_plane_helper_funcs = {
+	.prepare_fb = amdgpu_dm_plane_helper_prepare_fb,
+	.cleanup_fb = amdgpu_dm_plane_helper_cleanup_fb,
+	.atomic_check = amdgpu_dm_plane_atomic_check,
+	.atomic_async_check = amdgpu_dm_plane_atomic_async_check,
+	.atomic_async_update = amdgpu_dm_plane_atomic_async_update
+};
+
+static const struct drm_plane_helper_funcs dm_primary_plane_helper_funcs = {
+	.prepare_fb = amdgpu_dm_plane_helper_prepare_fb,
+	.cleanup_fb = amdgpu_dm_plane_helper_cleanup_fb,
+	.atomic_check = amdgpu_dm_plane_atomic_check,
+	.atomic_async_check = amdgpu_dm_plane_atomic_async_check,
+	.atomic_async_update = amdgpu_dm_plane_atomic_async_update,
+	.get_scanout_buffer = amdgpu_display_get_scanout_buffer,
+	.panic_flush = amdgpu_dm_plane_panic_flush,
+};
+
+static void amdgpu_dm_plane_drm_plane_reset(struct drm_plane *plane)
+{
+	struct dm_plane_state *amdgpu_state = NULL;
+
+	if (plane->state)
+		plane->funcs->atomic_destroy_state(plane, plane->state);
+
+	amdgpu_state = kzalloc(sizeof(*amdgpu_state), GFP_KERNEL);
+	WARN_ON(amdgpu_state == NULL);
+
+	if (!amdgpu_state)
+		return;
+
+	__drm_atomic_helper_plane_reset(plane, &amdgpu_state->base);
+	amdgpu_state->degamma_tf = AMDGPU_TRANSFER_FUNCTION_DEFAULT;
+	amdgpu_state->hdr_mult = AMDGPU_HDR_MULT_DEFAULT;
+	amdgpu_state->shaper_tf = AMDGPU_TRANSFER_FUNCTION_DEFAULT;
+	amdgpu_state->blend_tf = AMDGPU_TRANSFER_FUNCTION_DEFAULT;
+}
+
+static struct drm_plane_state *amdgpu_dm_plane_drm_plane_duplicate_state(struct drm_plane *plane)
+{
+	struct dm_plane_state *dm_plane_state, *old_dm_plane_state;
+
+	old_dm_plane_state = to_dm_plane_state(plane->state);
+	dm_plane_state = kzalloc(sizeof(*dm_plane_state), GFP_KERNEL);
+	if (!dm_plane_state)
+		return NULL;
+
+	__drm_atomic_helper_plane_duplicate_state(plane, &dm_plane_state->base);
+
+	if (old_dm_plane_state->dc_state) {
+		dm_plane_state->dc_state = old_dm_plane_state->dc_state;
+		dc_plane_state_retain(dm_plane_state->dc_state);
+	}
+
+	if (old_dm_plane_state->degamma_lut)
+		dm_plane_state->degamma_lut =
+			drm_property_blob_get(old_dm_plane_state->degamma_lut);
+	if (old_dm_plane_state->ctm)
+		dm_plane_state->ctm =
+			drm_property_blob_get(old_dm_plane_state->ctm);
+	if (old_dm_plane_state->shaper_lut)
+		dm_plane_state->shaper_lut =
+			drm_property_blob_get(old_dm_plane_state->shaper_lut);
+	if (old_dm_plane_state->lut3d)
+		dm_plane_state->lut3d =
+			drm_property_blob_get(old_dm_plane_state->lut3d);
+	if (old_dm_plane_state->blend_lut)
+		dm_plane_state->blend_lut =
+			drm_property_blob_get(old_dm_plane_state->blend_lut);
+
+	dm_plane_state->degamma_tf = old_dm_plane_state->degamma_tf;
+	dm_plane_state->hdr_mult = old_dm_plane_state->hdr_mult;
+	dm_plane_state->shaper_tf = old_dm_plane_state->shaper_tf;
+	dm_plane_state->blend_tf = old_dm_plane_state->blend_tf;
+
+	return &dm_plane_state->base;
+}
+
+static bool amdgpu_dm_plane_format_mod_supported(struct drm_plane *plane,
+						 uint32_t format,
+						 uint64_t modifier)
+{
+	struct amdgpu_device *adev = drm_to_adev(plane->dev);
+	const struct drm_format_info *info = drm_format_info(format);
+	int i;
+
+	if (!info)
+		return false;
+
+	/*
+	 * We always have to allow these modifiers:
+	 * 1. Core DRM checks for LINEAR support if userspace does not provide modifiers.
+	 * 2. Not passing any modifiers is the same as explicitly passing INVALID.
+	 */
+	if (modifier == DRM_FORMAT_MOD_LINEAR ||
+	    modifier == DRM_FORMAT_MOD_INVALID) {
+		return true;
+	}
+
+	/* Check that the modifier is on the list of the plane's supported modifiers. */
+	for (i = 0; i < plane->modifier_count; i++) {
+		if (modifier == plane->modifiers[i])
+			break;
+	}
+	if (i == plane->modifier_count)
+		return false;
+
+	/* GFX12 doesn't have these limitations. */
+	if (AMD_FMT_MOD_GET(TILE_VERSION, modifier) <= AMD_FMT_MOD_TILE_VER_GFX11) {
+		enum dm_micro_swizzle microtile = amdgpu_dm_plane_modifier_gfx9_swizzle_mode(modifier) & 3;
+
+		/*
+		 * For D swizzle the canonical modifier depends on the bpp, so check
+		 * it here.
+		 */
+		if (AMD_FMT_MOD_GET(TILE_VERSION, modifier) == AMD_FMT_MOD_TILE_VER_GFX9 &&
+		    adev->family >= AMDGPU_FAMILY_NV) {
+			if (microtile == MICRO_SWIZZLE_D && info->cpp[0] == 4)
+				return false;
+		}
+
+		if (adev->family >= AMDGPU_FAMILY_RV && microtile == MICRO_SWIZZLE_D &&
+		    info->cpp[0] < 8)
+			return false;
+
+		if (amdgpu_dm_plane_modifier_has_dcc(modifier)) {
+			/* Per radeonsi comments 16/64 bpp are more complicated. */
+			if (info->cpp[0] != 4)
+				return false;
+			/* We support multi-planar formats, but not when combined with
+			 * additional DCC metadata planes.
+			 */
+			if (info->num_planes > 1)
+				return false;
+		}
+	}
+
+	return true;
+}
+
+static void amdgpu_dm_plane_drm_plane_destroy_state(struct drm_plane *plane,
+						    struct drm_plane_state *state)
+{
+	struct dm_plane_state *dm_plane_state = to_dm_plane_state(state);
+
+	if (dm_plane_state->degamma_lut)
+		drm_property_blob_put(dm_plane_state->degamma_lut);
+	if (dm_plane_state->ctm)
+		drm_property_blob_put(dm_plane_state->ctm);
+	if (dm_plane_state->lut3d)
+		drm_property_blob_put(dm_plane_state->lut3d);
+	if (dm_plane_state->shaper_lut)
+		drm_property_blob_put(dm_plane_state->shaper_lut);
+	if (dm_plane_state->blend_lut)
+		drm_property_blob_put(dm_plane_state->blend_lut);
+
+	if (dm_plane_state->dc_state)
+		dc_plane_state_release(dm_plane_state->dc_state);
+
+	drm_atomic_helper_plane_destroy_state(plane, state);
+}
+
+#ifdef AMD_PRIVATE_COLOR
+static void
+dm_atomic_plane_attach_color_mgmt_properties(struct amdgpu_display_manager *dm,
+					     struct drm_plane *plane)
+{
+	struct amdgpu_mode_info mode_info = dm->adev->mode_info;
+	struct dpp_color_caps dpp_color_caps = dm->dc->caps.color.dpp;
+
+	/* Check HW color pipeline capabilities on DPP block (pre-blending)
+	 * before exposing related properties.
+	 */
+	if (dpp_color_caps.dgam_ram || dpp_color_caps.gamma_corr) {
+		drm_object_attach_property(&plane->base,
+					   mode_info.plane_degamma_lut_property,
+					   0);
+		drm_object_attach_property(&plane->base,
+					   mode_info.plane_degamma_lut_size_property,
+					   MAX_COLOR_LUT_ENTRIES);
+		drm_object_attach_property(&plane->base,
+					   dm->adev->mode_info.plane_degamma_tf_property,
+					   AMDGPU_TRANSFER_FUNCTION_DEFAULT);
+	}
+	/* HDR MULT is always available */
+	drm_object_attach_property(&plane->base,
+				   dm->adev->mode_info.plane_hdr_mult_property,
+				   AMDGPU_HDR_MULT_DEFAULT);
+
+	/* Only enable plane CTM if both DPP and MPC gamut remap is available. */
+	if (dm->dc->caps.color.mpc.gamut_remap)
+		drm_object_attach_property(&plane->base,
+					   dm->adev->mode_info.plane_ctm_property, 0);
+
+	if (dpp_color_caps.hw_3d_lut || dm->dc->caps.color.mpc.preblend) {
+		drm_object_attach_property(&plane->base,
+					   mode_info.plane_shaper_lut_property, 0);
+		drm_object_attach_property(&plane->base,
+					   mode_info.plane_shaper_lut_size_property,
+					   MAX_COLOR_LUT_ENTRIES);
+		drm_object_attach_property(&plane->base,
+					   mode_info.plane_shaper_tf_property,
+					   AMDGPU_TRANSFER_FUNCTION_DEFAULT);
+		drm_object_attach_property(&plane->base,
+					   mode_info.plane_lut3d_property, 0);
+		drm_object_attach_property(&plane->base,
+					   mode_info.plane_lut3d_size_property,
+					   MAX_COLOR_3DLUT_SIZE);
+	}
+
+	if (dpp_color_caps.ogam_ram) {
+		drm_object_attach_property(&plane->base,
+					   mode_info.plane_blend_lut_property, 0);
+		drm_object_attach_property(&plane->base,
+					   mode_info.plane_blend_lut_size_property,
+					   MAX_COLOR_LUT_ENTRIES);
+		drm_object_attach_property(&plane->base,
+					   mode_info.plane_blend_tf_property,
+					   AMDGPU_TRANSFER_FUNCTION_DEFAULT);
+	}
+}
+
+static int
+dm_atomic_plane_set_property(struct drm_plane *plane,
+			     struct drm_plane_state *state,
+			     struct drm_property *property,
+			     uint64_t val)
+{
+	struct dm_plane_state *dm_plane_state = to_dm_plane_state(state);
+	struct amdgpu_device *adev = drm_to_adev(plane->dev);
+	bool replaced = false;
+	int ret;
+
+	if (property == adev->mode_info.plane_degamma_lut_property) {
+		ret = drm_property_replace_blob_from_id(plane->dev,
+							&dm_plane_state->degamma_lut,
+							val, -1,
+							sizeof(struct drm_color_lut),
+							&replaced);
+		dm_plane_state->base.color_mgmt_changed |= replaced;
+		return ret;
+	} else if (property == adev->mode_info.plane_degamma_tf_property) {
+		if (dm_plane_state->degamma_tf != val) {
+			dm_plane_state->degamma_tf = val;
+			dm_plane_state->base.color_mgmt_changed = 1;
+		}
+	} else if (property == adev->mode_info.plane_hdr_mult_property) {
+		if (dm_plane_state->hdr_mult != val) {
+			dm_plane_state->hdr_mult = val;
+			dm_plane_state->base.color_mgmt_changed = 1;
+		}
+	} else if (property == adev->mode_info.plane_ctm_property) {
+		ret = drm_property_replace_blob_from_id(plane->dev,
+							&dm_plane_state->ctm,
+							val,
+							sizeof(struct drm_color_ctm_3x4), -1,
+							&replaced);
+		dm_plane_state->base.color_mgmt_changed |= replaced;
+		return ret;
+	} else if (property == adev->mode_info.plane_shaper_lut_property) {
+		ret = drm_property_replace_blob_from_id(plane->dev,
+							&dm_plane_state->shaper_lut,
+							val, -1,
+							sizeof(struct drm_color_lut),
+							&replaced);
+		dm_plane_state->base.color_mgmt_changed |= replaced;
+		return ret;
+	} else if (property == adev->mode_info.plane_shaper_tf_property) {
+		if (dm_plane_state->shaper_tf != val) {
+			dm_plane_state->shaper_tf = val;
+			dm_plane_state->base.color_mgmt_changed = 1;
+		}
+	} else if (property == adev->mode_info.plane_lut3d_property) {
+		ret = drm_property_replace_blob_from_id(plane->dev,
+							&dm_plane_state->lut3d,
+							val, -1,
+							sizeof(struct drm_color_lut),
+							&replaced);
+		dm_plane_state->base.color_mgmt_changed |= replaced;
+		return ret;
+	} else if (property == adev->mode_info.plane_blend_lut_property) {
+		ret = drm_property_replace_blob_from_id(plane->dev,
+							&dm_plane_state->blend_lut,
+							val, -1,
+							sizeof(struct drm_color_lut),
+							&replaced);
+		dm_plane_state->base.color_mgmt_changed |= replaced;
+		return ret;
+	} else if (property == adev->mode_info.plane_blend_tf_property) {
+		if (dm_plane_state->blend_tf != val) {
+			dm_plane_state->blend_tf = val;
+			dm_plane_state->base.color_mgmt_changed = 1;
+		}
+	} else {
+		drm_dbg_atomic(plane->dev,
+			       "[PLANE:%d:%s] unknown property [PROP:%d:%s]]\n",
+			       plane->base.id, plane->name,
+			       property->base.id, property->name);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int
+dm_atomic_plane_get_property(struct drm_plane *plane,
+			     const struct drm_plane_state *state,
+			     struct drm_property *property,
+			     uint64_t *val)
+{
+	struct dm_plane_state *dm_plane_state = to_dm_plane_state(state);
+	struct amdgpu_device *adev = drm_to_adev(plane->dev);
+
+	if (property == adev->mode_info.plane_degamma_lut_property) {
+		*val = (dm_plane_state->degamma_lut) ?
+			dm_plane_state->degamma_lut->base.id : 0;
+	} else if (property == adev->mode_info.plane_degamma_tf_property) {
+		*val = dm_plane_state->degamma_tf;
+	} else if (property == adev->mode_info.plane_hdr_mult_property) {
+		*val = dm_plane_state->hdr_mult;
+	} else if (property == adev->mode_info.plane_ctm_property) {
+		*val = (dm_plane_state->ctm) ?
+			dm_plane_state->ctm->base.id : 0;
+	} else 	if (property == adev->mode_info.plane_shaper_lut_property) {
+		*val = (dm_plane_state->shaper_lut) ?
+			dm_plane_state->shaper_lut->base.id : 0;
+	} else if (property == adev->mode_info.plane_shaper_tf_property) {
+		*val = dm_plane_state->shaper_tf;
+	} else 	if (property == adev->mode_info.plane_lut3d_property) {
+		*val = (dm_plane_state->lut3d) ?
+			dm_plane_state->lut3d->base.id : 0;
+	} else 	if (property == adev->mode_info.plane_blend_lut_property) {
+		*val = (dm_plane_state->blend_lut) ?
+			dm_plane_state->blend_lut->base.id : 0;
+	} else if (property == adev->mode_info.plane_blend_tf_property) {
+		*val = dm_plane_state->blend_tf;
+
+	} else {
+		return -EINVAL;
+	}
+
+	return 0;
+}
+#endif
+
+static const struct drm_plane_funcs dm_plane_funcs = {
+	.update_plane	= drm_atomic_helper_update_plane,
+	.disable_plane	= drm_atomic_helper_disable_plane,
+	.destroy	= drm_plane_helper_destroy,
+	.reset = amdgpu_dm_plane_drm_plane_reset,
+	.atomic_duplicate_state = amdgpu_dm_plane_drm_plane_duplicate_state,
+	.atomic_destroy_state = amdgpu_dm_plane_drm_plane_destroy_state,
+	.format_mod_supported = amdgpu_dm_plane_format_mod_supported,
+#ifdef AMD_PRIVATE_COLOR
+	.atomic_set_property = dm_atomic_plane_set_property,
+	.atomic_get_property = dm_atomic_plane_get_property,
+#endif
+};
+
+int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm,
+				struct drm_plane *plane,
+				unsigned long possible_crtcs,
+				const struct dc_plane_cap *plane_cap)
+{
+	uint32_t formats[32];
+	int num_formats;
+	int res = -EPERM;
+	unsigned int supported_rotations;
+	uint64_t *modifiers = NULL;
+	unsigned int primary_zpos = dm->dc->caps.max_slave_planes;
+
+	num_formats = amdgpu_dm_plane_get_plane_formats(plane, plane_cap, formats,
+							ARRAY_SIZE(formats));
+
+	res = amdgpu_dm_plane_get_plane_modifiers(dm->adev, plane->type, &modifiers);
+	if (res)
+		return res;
+
+	if (modifiers == NULL)
+		adev_to_drm(dm->adev)->mode_config.fb_modifiers_not_supported = true;
+
+	res = drm_universal_plane_init(adev_to_drm(dm->adev), plane, possible_crtcs,
+				       &dm_plane_funcs, formats, num_formats,
+				       modifiers, plane->type, NULL);
+	kfree(modifiers);
+	if (res)
+		return res;
+
+	if (plane->type == DRM_PLANE_TYPE_OVERLAY &&
+	    plane_cap && plane_cap->per_pixel_alpha) {
+		unsigned int blend_caps = BIT(DRM_MODE_BLEND_PIXEL_NONE) |
+					  BIT(DRM_MODE_BLEND_PREMULTI) |
+					  BIT(DRM_MODE_BLEND_COVERAGE);
+
+		drm_plane_create_alpha_property(plane);
+		drm_plane_create_blend_mode_property(plane, blend_caps);
+	}
+
+	if (plane->type == DRM_PLANE_TYPE_PRIMARY) {
+		/*
+		 * Allow OVERLAY planes to be used as underlays by assigning an
+		 * immutable zpos = # of OVERLAY planes to the PRIMARY plane.
+		 */
+		drm_plane_create_zpos_immutable_property(plane, primary_zpos);
+	} else if (plane->type == DRM_PLANE_TYPE_OVERLAY) {
+		/*
+		 * OVERLAY planes can be below or above the PRIMARY, but cannot
+		 * be above the CURSOR plane.
+		 */
+		unsigned int zpos = primary_zpos + 1 + drm_plane_index(plane);
+
+		drm_plane_create_zpos_property(plane, zpos, 0, 254);
+	} else if (plane->type == DRM_PLANE_TYPE_CURSOR) {
+		drm_plane_create_zpos_immutable_property(plane, 255);
+	}
+
+	if (plane->type == DRM_PLANE_TYPE_PRIMARY &&
+	    plane_cap &&
+	    (plane_cap->pixel_format_support.nv12 ||
+	     plane_cap->pixel_format_support.p010)) {
+		/* This only affects YUV formats. */
+		drm_plane_create_color_properties(
+			plane,
+			BIT(DRM_COLOR_YCBCR_BT601) |
+			BIT(DRM_COLOR_YCBCR_BT709) |
+			BIT(DRM_COLOR_YCBCR_BT2020),
+			BIT(DRM_COLOR_YCBCR_LIMITED_RANGE) |
+			BIT(DRM_COLOR_YCBCR_FULL_RANGE),
+			DRM_COLOR_YCBCR_BT709, DRM_COLOR_YCBCR_LIMITED_RANGE);
+	}
+
+	supported_rotations =
+		DRM_MODE_ROTATE_0 | DRM_MODE_ROTATE_90 |
+		DRM_MODE_ROTATE_180 | DRM_MODE_ROTATE_270;
+
+	if (dm->adev->asic_type >= CHIP_BONAIRE &&
+	    plane->type != DRM_PLANE_TYPE_CURSOR)
+		drm_plane_create_rotation_property(plane, DRM_MODE_ROTATE_0,
+						   supported_rotations);
+
+	if (amdgpu_ip_version(dm->adev, DCE_HWIP, 0) > IP_VERSION(3, 0, 1) &&
+	    plane->type != DRM_PLANE_TYPE_CURSOR)
+		drm_plane_enable_fb_damage_clips(plane);
+
+	if (plane->type == DRM_PLANE_TYPE_PRIMARY)
+		drm_plane_helper_add(plane, &dm_primary_plane_helper_funcs);
+	else
+		drm_plane_helper_add(plane, &dm_plane_helper_funcs);
+
+#ifdef AMD_PRIVATE_COLOR
+	dm_atomic_plane_attach_color_mgmt_properties(dm, plane);
+#endif
+	/* Create (reset) the plane state */
+	if (plane->funcs->reset)
+		plane->funcs->reset(plane);
+
+	return 0;
+}
+
+bool amdgpu_dm_plane_is_video_format(uint32_t format)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(video_formats); i++)
+		if (format == video_formats[i])
+			return true;
+
+	return false;
+}
+
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.h
new file mode 100644
index 000000000000..ea2619b507db
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.h
@@ -0,0 +1,68 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __AMDGPU_DM_PLANE_H__
+#define __AMDGPU_DM_PLANE_H__
+
+#include "dc.h"
+
+int amdgpu_dm_plane_get_cursor_position(struct drm_plane *plane, struct drm_crtc *crtc,
+					struct dc_cursor_position *position);
+
+void amdgpu_dm_plane_handle_cursor_update(struct drm_plane *plane,
+			  struct drm_plane_state *old_plane_state);
+
+int amdgpu_dm_plane_fill_dc_scaling_info(struct amdgpu_device *adev,
+			 const struct drm_plane_state *state,
+			 struct dc_scaling_info *scaling_info);
+
+int amdgpu_dm_plane_helper_check_state(struct drm_plane_state *state,
+				struct drm_crtc_state *new_crtc_state);
+
+int amdgpu_dm_plane_fill_plane_buffer_attributes(struct amdgpu_device *adev,
+				 const struct amdgpu_framebuffer *afb,
+				 const enum surface_pixel_format format,
+				 const enum dc_rotation_angle rotation,
+				 const uint64_t tiling_flags,
+				 struct dc_tiling_info *tiling_info,
+				 struct plane_size *plane_size,
+				 struct dc_plane_dcc_param *dcc,
+				 struct dc_plane_address *address,
+				 bool tmz_surface);
+
+int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm,
+			 struct drm_plane *plane,
+			 unsigned long possible_crtcs,
+			 const struct dc_plane_cap *plane_cap);
+
+const struct drm_format_info *amdgpu_dm_plane_get_format_info(u32 pixel_format, u64 modifier);
+
+void amdgpu_dm_plane_fill_blending_from_plane_state(const struct drm_plane_state *plane_state,
+				    bool *per_pixel_alpha, bool *pre_multiplied_alpha,
+				    bool *global_alpha, int *global_alpha_value);
+
+bool amdgpu_dm_plane_is_video_format(uint32_t format);
+#endif
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_pp_smu.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_pp_smu.c
new file mode 100644
index 000000000000..11b2ea6edf95
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_pp_smu.c
@@ -0,0 +1,808 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ */
+#include <linux/string.h>
+#include <linux/acpi.h>
+
+#include <drm/drm_probe_helper.h>
+#include <drm/amdgpu_drm.h>
+#include "dm_services.h"
+#include "amdgpu.h"
+#include "amdgpu_dm.h"
+#include "amdgpu_dm_irq.h"
+#include "amdgpu_pm.h"
+#include "dm_pp_smu.h"
+
+bool dm_pp_apply_display_requirements(
+		const struct dc_context *ctx,
+		const struct dm_pp_display_configuration *pp_display_cfg)
+{
+	struct amdgpu_device *adev = ctx->driver_context;
+	int i;
+
+	if (adev->pm.dpm_enabled) {
+
+		memset(&adev->pm.pm_display_cfg, 0,
+				sizeof(adev->pm.pm_display_cfg));
+
+		adev->pm.pm_display_cfg.cpu_cc6_disable =
+			pp_display_cfg->cpu_cc6_disable;
+
+		adev->pm.pm_display_cfg.cpu_pstate_disable =
+			pp_display_cfg->cpu_pstate_disable;
+
+		adev->pm.pm_display_cfg.cpu_pstate_separation_time =
+			pp_display_cfg->cpu_pstate_separation_time;
+
+		adev->pm.pm_display_cfg.nb_pstate_switch_disable =
+			pp_display_cfg->nb_pstate_switch_disable;
+
+		adev->pm.pm_display_cfg.num_display =
+				pp_display_cfg->display_count;
+		adev->pm.pm_display_cfg.num_path_including_non_display =
+				pp_display_cfg->display_count;
+
+		adev->pm.pm_display_cfg.min_core_set_clock =
+				pp_display_cfg->min_engine_clock_khz/10;
+		adev->pm.pm_display_cfg.min_core_set_clock_in_sr =
+				pp_display_cfg->min_engine_clock_deep_sleep_khz/10;
+		adev->pm.pm_display_cfg.min_mem_set_clock =
+				pp_display_cfg->min_memory_clock_khz/10;
+
+		adev->pm.pm_display_cfg.min_dcef_deep_sleep_set_clk =
+				pp_display_cfg->min_engine_clock_deep_sleep_khz/10;
+		adev->pm.pm_display_cfg.min_dcef_set_clk =
+				pp_display_cfg->min_dcfclock_khz/10;
+
+		adev->pm.pm_display_cfg.multi_monitor_in_sync =
+				pp_display_cfg->all_displays_in_sync;
+		adev->pm.pm_display_cfg.min_vblank_time =
+				pp_display_cfg->avail_mclk_switch_time_us;
+
+		adev->pm.pm_display_cfg.display_clk =
+				pp_display_cfg->disp_clk_khz/10;
+
+		adev->pm.pm_display_cfg.dce_tolerable_mclk_in_active_latency =
+				pp_display_cfg->avail_mclk_switch_time_in_disp_active_us;
+
+		adev->pm.pm_display_cfg.crtc_index = pp_display_cfg->crtc_index;
+		adev->pm.pm_display_cfg.line_time_in_us =
+				pp_display_cfg->line_time_in_us;
+
+		adev->pm.pm_display_cfg.vrefresh = pp_display_cfg->disp_configs[0].v_refresh;
+		adev->pm.pm_display_cfg.crossfire_display_index = -1;
+		adev->pm.pm_display_cfg.min_bus_bandwidth = 0;
+
+		for (i = 0; i < pp_display_cfg->display_count; i++) {
+			const struct dm_pp_single_disp_config *dc_cfg =
+						&pp_display_cfg->disp_configs[i];
+			adev->pm.pm_display_cfg.displays[i].controller_id = dc_cfg->pipe_idx + 1;
+			adev->pm.pm_display_cfg.displays[i].pixel_clock = dc_cfg->pixel_clock;
+		}
+
+		amdgpu_dpm_display_configuration_change(adev, &adev->pm.pm_display_cfg);
+
+		amdgpu_dpm_compute_clocks(adev);
+	}
+
+	return true;
+}
+
+static void get_default_clock_levels(
+		enum dm_pp_clock_type clk_type,
+		struct dm_pp_clock_levels *clks)
+{
+	uint32_t disp_clks_in_khz[6] = {
+			300000, 400000, 496560, 626090, 685720, 757900 };
+	uint32_t sclks_in_khz[6] = {
+			300000, 360000, 423530, 514290, 626090, 720000 };
+	uint32_t mclks_in_khz[2] = { 333000, 800000 };
+
+	switch (clk_type) {
+	case DM_PP_CLOCK_TYPE_DISPLAY_CLK:
+		clks->num_levels = 6;
+		memmove(clks->clocks_in_khz, disp_clks_in_khz,
+				sizeof(disp_clks_in_khz));
+		break;
+	case DM_PP_CLOCK_TYPE_ENGINE_CLK:
+		clks->num_levels = 6;
+		memmove(clks->clocks_in_khz, sclks_in_khz,
+				sizeof(sclks_in_khz));
+		break;
+	case DM_PP_CLOCK_TYPE_MEMORY_CLK:
+		clks->num_levels = 2;
+		memmove(clks->clocks_in_khz, mclks_in_khz,
+				sizeof(mclks_in_khz));
+		break;
+	default:
+		clks->num_levels = 0;
+		break;
+	}
+}
+
+static enum amd_pp_clock_type dc_to_pp_clock_type(
+		enum dm_pp_clock_type dm_pp_clk_type)
+{
+	enum amd_pp_clock_type amd_pp_clk_type = 0;
+
+	switch (dm_pp_clk_type) {
+	case DM_PP_CLOCK_TYPE_DISPLAY_CLK:
+		amd_pp_clk_type = amd_pp_disp_clock;
+		break;
+	case DM_PP_CLOCK_TYPE_ENGINE_CLK:
+		amd_pp_clk_type = amd_pp_sys_clock;
+		break;
+	case DM_PP_CLOCK_TYPE_MEMORY_CLK:
+		amd_pp_clk_type = amd_pp_mem_clock;
+		break;
+	case DM_PP_CLOCK_TYPE_DCEFCLK:
+		amd_pp_clk_type  = amd_pp_dcef_clock;
+		break;
+	case DM_PP_CLOCK_TYPE_DCFCLK:
+		amd_pp_clk_type = amd_pp_dcf_clock;
+		break;
+	case DM_PP_CLOCK_TYPE_PIXELCLK:
+		amd_pp_clk_type = amd_pp_pixel_clock;
+		break;
+	case DM_PP_CLOCK_TYPE_FCLK:
+		amd_pp_clk_type = amd_pp_f_clock;
+		break;
+	case DM_PP_CLOCK_TYPE_DISPLAYPHYCLK:
+		amd_pp_clk_type = amd_pp_phy_clock;
+		break;
+	case DM_PP_CLOCK_TYPE_DPPCLK:
+		amd_pp_clk_type = amd_pp_dpp_clock;
+		break;
+	default:
+		DRM_ERROR("DM_PPLIB: invalid clock type: %d!\n",
+				dm_pp_clk_type);
+		break;
+	}
+
+	return amd_pp_clk_type;
+}
+
+static enum dm_pp_clocks_state pp_to_dc_powerlevel_state(
+			enum PP_DAL_POWERLEVEL max_clocks_state)
+{
+	switch (max_clocks_state) {
+	case PP_DAL_POWERLEVEL_0:
+		return DM_PP_CLOCKS_DPM_STATE_LEVEL_0;
+	case PP_DAL_POWERLEVEL_1:
+		return DM_PP_CLOCKS_DPM_STATE_LEVEL_1;
+	case PP_DAL_POWERLEVEL_2:
+		return DM_PP_CLOCKS_DPM_STATE_LEVEL_2;
+	case PP_DAL_POWERLEVEL_3:
+		return DM_PP_CLOCKS_DPM_STATE_LEVEL_3;
+	case PP_DAL_POWERLEVEL_4:
+		return DM_PP_CLOCKS_DPM_STATE_LEVEL_4;
+	case PP_DAL_POWERLEVEL_5:
+		return DM_PP_CLOCKS_DPM_STATE_LEVEL_5;
+	case PP_DAL_POWERLEVEL_6:
+		return DM_PP_CLOCKS_DPM_STATE_LEVEL_6;
+	case PP_DAL_POWERLEVEL_7:
+		return DM_PP_CLOCKS_DPM_STATE_LEVEL_7;
+	default:
+		DRM_ERROR("DM_PPLIB: invalid powerlevel state: %d!\n",
+				max_clocks_state);
+		return DM_PP_CLOCKS_STATE_INVALID;
+	}
+}
+
+static void pp_to_dc_clock_levels(
+		const struct amd_pp_clocks *pp_clks,
+		struct dm_pp_clock_levels *dc_clks,
+		enum dm_pp_clock_type dc_clk_type)
+{
+	uint32_t i;
+
+	if (pp_clks->count > DM_PP_MAX_CLOCK_LEVELS) {
+		DRM_INFO("DM_PPLIB: Warning: %s clock: number of levels %d exceeds maximum of %d!\n",
+				DC_DECODE_PP_CLOCK_TYPE(dc_clk_type),
+				pp_clks->count,
+				DM_PP_MAX_CLOCK_LEVELS);
+
+		dc_clks->num_levels = DM_PP_MAX_CLOCK_LEVELS;
+	} else
+		dc_clks->num_levels = pp_clks->count;
+
+	DRM_INFO("DM_PPLIB: values for %s clock\n",
+			DC_DECODE_PP_CLOCK_TYPE(dc_clk_type));
+
+	for (i = 0; i < dc_clks->num_levels; i++) {
+		DRM_INFO("DM_PPLIB:\t %d\n", pp_clks->clock[i]);
+		dc_clks->clocks_in_khz[i] = pp_clks->clock[i];
+	}
+}
+
+static void pp_to_dc_clock_levels_with_latency(
+		const struct pp_clock_levels_with_latency *pp_clks,
+		struct dm_pp_clock_levels_with_latency *clk_level_info,
+		enum dm_pp_clock_type dc_clk_type)
+{
+	uint32_t i;
+
+	if (pp_clks->num_levels > DM_PP_MAX_CLOCK_LEVELS) {
+		DRM_INFO("DM_PPLIB: Warning: %s clock: number of levels %d exceeds maximum of %d!\n",
+				DC_DECODE_PP_CLOCK_TYPE(dc_clk_type),
+				pp_clks->num_levels,
+				DM_PP_MAX_CLOCK_LEVELS);
+
+		clk_level_info->num_levels = DM_PP_MAX_CLOCK_LEVELS;
+	} else
+		clk_level_info->num_levels = pp_clks->num_levels;
+
+	DRM_DEBUG("DM_PPLIB: values for %s clock\n",
+			DC_DECODE_PP_CLOCK_TYPE(dc_clk_type));
+
+	for (i = 0; i < clk_level_info->num_levels; i++) {
+		DRM_DEBUG("DM_PPLIB:\t %d in kHz\n", pp_clks->data[i].clocks_in_khz);
+		clk_level_info->data[i].clocks_in_khz = pp_clks->data[i].clocks_in_khz;
+		clk_level_info->data[i].latency_in_us = pp_clks->data[i].latency_in_us;
+	}
+}
+
+static void pp_to_dc_clock_levels_with_voltage(
+		const struct pp_clock_levels_with_voltage *pp_clks,
+		struct dm_pp_clock_levels_with_voltage *clk_level_info,
+		enum dm_pp_clock_type dc_clk_type)
+{
+	uint32_t i;
+
+	if (pp_clks->num_levels > DM_PP_MAX_CLOCK_LEVELS) {
+		DRM_INFO("DM_PPLIB: Warning: %s clock: number of levels %d exceeds maximum of %d!\n",
+				DC_DECODE_PP_CLOCK_TYPE(dc_clk_type),
+				pp_clks->num_levels,
+				DM_PP_MAX_CLOCK_LEVELS);
+
+		clk_level_info->num_levels = DM_PP_MAX_CLOCK_LEVELS;
+	} else
+		clk_level_info->num_levels = pp_clks->num_levels;
+
+	DRM_INFO("DM_PPLIB: values for %s clock\n",
+			DC_DECODE_PP_CLOCK_TYPE(dc_clk_type));
+
+	for (i = 0; i < clk_level_info->num_levels; i++) {
+		DRM_INFO("DM_PPLIB:\t %d in kHz, %d in mV\n", pp_clks->data[i].clocks_in_khz,
+			 pp_clks->data[i].voltage_in_mv);
+		clk_level_info->data[i].clocks_in_khz = pp_clks->data[i].clocks_in_khz;
+		clk_level_info->data[i].voltage_in_mv = pp_clks->data[i].voltage_in_mv;
+	}
+}
+
+bool dm_pp_get_clock_levels_by_type(
+		const struct dc_context *ctx,
+		enum dm_pp_clock_type clk_type,
+		struct dm_pp_clock_levels *dc_clks)
+{
+	struct amdgpu_device *adev = ctx->driver_context;
+	struct amd_pp_clocks pp_clks = { 0 };
+	struct amd_pp_simple_clock_info validation_clks = { 0 };
+	uint32_t i;
+
+	if (amdgpu_dpm_get_clock_by_type(adev,
+		dc_to_pp_clock_type(clk_type), &pp_clks)) {
+		/* Error in pplib. Provide default values. */
+		get_default_clock_levels(clk_type, dc_clks);
+		return true;
+	}
+
+	pp_to_dc_clock_levels(&pp_clks, dc_clks, clk_type);
+
+	if (amdgpu_dpm_get_display_mode_validation_clks(adev, &validation_clks)) {
+		/* Error in pplib. Provide default values. */
+		DRM_INFO("DM_PPLIB: Warning: using default validation clocks!\n");
+		validation_clks.engine_max_clock = 72000;
+		validation_clks.memory_max_clock = 80000;
+		validation_clks.level = 0;
+	}
+
+	DRM_INFO("DM_PPLIB: Validation clocks:\n");
+	DRM_INFO("DM_PPLIB:    engine_max_clock: %d\n",
+			validation_clks.engine_max_clock);
+	DRM_INFO("DM_PPLIB:    memory_max_clock: %d\n",
+			validation_clks.memory_max_clock);
+	DRM_INFO("DM_PPLIB:    level           : %d\n",
+			validation_clks.level);
+
+	/* Translate 10 kHz to kHz. */
+	validation_clks.engine_max_clock *= 10;
+	validation_clks.memory_max_clock *= 10;
+
+	/* Determine the highest non-boosted level from the Validation Clocks */
+	if (clk_type == DM_PP_CLOCK_TYPE_ENGINE_CLK) {
+		for (i = 0; i < dc_clks->num_levels; i++) {
+			if (dc_clks->clocks_in_khz[i] > validation_clks.engine_max_clock) {
+				/* This clock is higher the validation clock.
+				 * Than means the previous one is the highest
+				 * non-boosted one.
+				 */
+				DRM_INFO("DM_PPLIB: reducing engine clock level from %d to %d\n",
+						dc_clks->num_levels, i);
+				dc_clks->num_levels = i > 0 ? i : 1;
+				break;
+			}
+		}
+	} else if (clk_type == DM_PP_CLOCK_TYPE_MEMORY_CLK) {
+		for (i = 0; i < dc_clks->num_levels; i++) {
+			if (dc_clks->clocks_in_khz[i] > validation_clks.memory_max_clock) {
+				DRM_INFO("DM_PPLIB: reducing memory clock level from %d to %d\n",
+						dc_clks->num_levels, i);
+				dc_clks->num_levels = i > 0 ? i : 1;
+				break;
+			}
+		}
+	}
+
+	return true;
+}
+
+bool dm_pp_get_clock_levels_by_type_with_latency(
+	const struct dc_context *ctx,
+	enum dm_pp_clock_type clk_type,
+	struct dm_pp_clock_levels_with_latency *clk_level_info)
+{
+	struct amdgpu_device *adev = ctx->driver_context;
+	struct pp_clock_levels_with_latency pp_clks = { 0 };
+	int ret;
+
+	ret = amdgpu_dpm_get_clock_by_type_with_latency(adev,
+					dc_to_pp_clock_type(clk_type),
+					&pp_clks);
+	if (ret)
+		return false;
+
+	pp_to_dc_clock_levels_with_latency(&pp_clks, clk_level_info, clk_type);
+
+	return true;
+}
+
+bool dm_pp_get_clock_levels_by_type_with_voltage(
+	const struct dc_context *ctx,
+	enum dm_pp_clock_type clk_type,
+	struct dm_pp_clock_levels_with_voltage *clk_level_info)
+{
+	struct amdgpu_device *adev = ctx->driver_context;
+	struct pp_clock_levels_with_voltage pp_clk_info = {0};
+	int ret;
+
+	ret = amdgpu_dpm_get_clock_by_type_with_voltage(adev,
+					dc_to_pp_clock_type(clk_type),
+					&pp_clk_info);
+	if (ret)
+		return false;
+
+	pp_to_dc_clock_levels_with_voltage(&pp_clk_info, clk_level_info, clk_type);
+
+	return true;
+}
+
+bool dm_pp_notify_wm_clock_changes(
+	const struct dc_context *ctx,
+	struct dm_pp_wm_sets_with_clock_ranges *wm_with_clock_ranges)
+{
+	struct amdgpu_device *adev = ctx->driver_context;
+
+	/*
+	 * Limit this watermark setting for Polaris for now
+	 * TODO: expand this to other ASICs
+	 */
+	if ((adev->asic_type >= CHIP_POLARIS10) &&
+	    (adev->asic_type <= CHIP_VEGAM) &&
+	    !amdgpu_dpm_set_watermarks_for_clocks_ranges(adev,
+							 (void *)wm_with_clock_ranges))
+		return true;
+
+	return false;
+}
+
+bool dm_pp_apply_power_level_change_request(
+	const struct dc_context *ctx,
+	struct dm_pp_power_level_change_request *level_change_req)
+{
+	/* TODO: to be implemented */
+	return false;
+}
+
+bool dm_pp_apply_clock_for_voltage_request(
+	const struct dc_context *ctx,
+	struct dm_pp_clock_for_voltage_req *clock_for_voltage_req)
+{
+	struct amdgpu_device *adev = ctx->driver_context;
+	struct pp_display_clock_request pp_clock_request = {0};
+	int ret = 0;
+
+	pp_clock_request.clock_type = dc_to_pp_clock_type(clock_for_voltage_req->clk_type);
+	pp_clock_request.clock_freq_in_khz = clock_for_voltage_req->clocks_in_khz;
+
+	if (!pp_clock_request.clock_type)
+		return false;
+
+	ret = amdgpu_dpm_display_clock_voltage_request(adev, &pp_clock_request);
+	if (ret && (ret != -EOPNOTSUPP))
+		return false;
+
+	return true;
+}
+
+bool dm_pp_get_static_clocks(
+	const struct dc_context *ctx,
+	struct dm_pp_static_clock_info *static_clk_info)
+{
+	struct amdgpu_device *adev = ctx->driver_context;
+	struct amd_pp_clock_info pp_clk_info = {0};
+
+	if (amdgpu_dpm_get_current_clocks(adev, &pp_clk_info))
+		return false;
+
+	static_clk_info->max_clocks_state = pp_to_dc_powerlevel_state(pp_clk_info.max_clocks_state);
+	static_clk_info->max_mclk_khz = pp_clk_info.max_memory_clock * 10;
+	static_clk_info->max_sclk_khz = pp_clk_info.max_engine_clock * 10;
+
+	return true;
+}
+
+static void pp_rv_set_wm_ranges(struct pp_smu *pp,
+		struct pp_smu_wm_range_sets *ranges)
+{
+	const struct dc_context *ctx = pp->dm;
+	struct amdgpu_device *adev = ctx->driver_context;
+	struct dm_pp_wm_sets_with_clock_ranges_soc15 wm_with_clock_ranges;
+	struct dm_pp_clock_range_for_dmif_wm_set_soc15 *wm_dce_clocks = wm_with_clock_ranges.wm_dmif_clocks_ranges;
+	struct dm_pp_clock_range_for_mcif_wm_set_soc15 *wm_soc_clocks = wm_with_clock_ranges.wm_mcif_clocks_ranges;
+	int32_t i;
+
+	wm_with_clock_ranges.num_wm_dmif_sets = ranges->num_reader_wm_sets;
+	wm_with_clock_ranges.num_wm_mcif_sets = ranges->num_writer_wm_sets;
+
+	for (i = 0; i < wm_with_clock_ranges.num_wm_dmif_sets; i++) {
+		if (ranges->reader_wm_sets[i].wm_inst > 3)
+			wm_dce_clocks[i].wm_set_id = WM_SET_A;
+		else
+			wm_dce_clocks[i].wm_set_id =
+					ranges->reader_wm_sets[i].wm_inst;
+		wm_dce_clocks[i].wm_max_dcfclk_clk_in_khz =
+				ranges->reader_wm_sets[i].max_drain_clk_mhz * 1000;
+		wm_dce_clocks[i].wm_min_dcfclk_clk_in_khz =
+				ranges->reader_wm_sets[i].min_drain_clk_mhz * 1000;
+		wm_dce_clocks[i].wm_max_mem_clk_in_khz =
+				ranges->reader_wm_sets[i].max_fill_clk_mhz * 1000;
+		wm_dce_clocks[i].wm_min_mem_clk_in_khz =
+				ranges->reader_wm_sets[i].min_fill_clk_mhz * 1000;
+	}
+
+	for (i = 0; i < wm_with_clock_ranges.num_wm_mcif_sets; i++) {
+		if (ranges->writer_wm_sets[i].wm_inst > 3)
+			wm_soc_clocks[i].wm_set_id = WM_SET_A;
+		else
+			wm_soc_clocks[i].wm_set_id =
+					ranges->writer_wm_sets[i].wm_inst;
+		wm_soc_clocks[i].wm_max_socclk_clk_in_khz =
+				ranges->writer_wm_sets[i].max_fill_clk_mhz * 1000;
+		wm_soc_clocks[i].wm_min_socclk_clk_in_khz =
+				ranges->writer_wm_sets[i].min_fill_clk_mhz * 1000;
+		wm_soc_clocks[i].wm_max_mem_clk_in_khz =
+				ranges->writer_wm_sets[i].max_drain_clk_mhz * 1000;
+		wm_soc_clocks[i].wm_min_mem_clk_in_khz =
+				ranges->writer_wm_sets[i].min_drain_clk_mhz * 1000;
+	}
+
+	amdgpu_dpm_set_watermarks_for_clocks_ranges(adev,
+						    &wm_with_clock_ranges);
+}
+
+static void pp_rv_set_pme_wa_enable(struct pp_smu *pp)
+{
+	const struct dc_context *ctx = pp->dm;
+	struct amdgpu_device *adev = ctx->driver_context;
+
+	amdgpu_dpm_notify_smu_enable_pwe(adev);
+}
+
+static void pp_rv_set_active_display_count(struct pp_smu *pp, int count)
+{
+	const struct dc_context *ctx = pp->dm;
+	struct amdgpu_device *adev = ctx->driver_context;
+
+	amdgpu_dpm_set_active_display_count(adev, count);
+}
+
+static void pp_rv_set_min_deep_sleep_dcfclk(struct pp_smu *pp, int clock)
+{
+	const struct dc_context *ctx = pp->dm;
+	struct amdgpu_device *adev = ctx->driver_context;
+
+	amdgpu_dpm_set_min_deep_sleep_dcefclk(adev, clock);
+}
+
+static void pp_rv_set_hard_min_dcefclk_by_freq(struct pp_smu *pp, int clock)
+{
+	const struct dc_context *ctx = pp->dm;
+	struct amdgpu_device *adev = ctx->driver_context;
+
+	amdgpu_dpm_set_hard_min_dcefclk_by_freq(adev, clock);
+}
+
+static void pp_rv_set_hard_min_fclk_by_freq(struct pp_smu *pp, int mhz)
+{
+	const struct dc_context *ctx = pp->dm;
+	struct amdgpu_device *adev = ctx->driver_context;
+
+	amdgpu_dpm_set_hard_min_fclk_by_freq(adev, mhz);
+}
+
+static enum pp_smu_status pp_nv_set_wm_ranges(struct pp_smu *pp,
+		struct pp_smu_wm_range_sets *ranges)
+{
+	const struct dc_context *ctx = pp->dm;
+	struct amdgpu_device *adev = ctx->driver_context;
+
+	amdgpu_dpm_set_watermarks_for_clocks_ranges(adev, ranges);
+
+	return PP_SMU_RESULT_OK;
+}
+
+static enum pp_smu_status pp_nv_set_display_count(struct pp_smu *pp, int count)
+{
+	const struct dc_context *ctx = pp->dm;
+	struct amdgpu_device *adev = ctx->driver_context;
+	int ret = 0;
+
+	ret = amdgpu_dpm_set_active_display_count(adev, count);
+	if (ret == -EOPNOTSUPP)
+		return PP_SMU_RESULT_UNSUPPORTED;
+	else if (ret)
+		/* 0: successful or smu.ppt_funcs->set_display_count = NULL;  1: fail */
+		return PP_SMU_RESULT_FAIL;
+
+	return PP_SMU_RESULT_OK;
+}
+
+static enum pp_smu_status
+pp_nv_set_min_deep_sleep_dcfclk(struct pp_smu *pp, int mhz)
+{
+	const struct dc_context *ctx = pp->dm;
+	struct amdgpu_device *adev = ctx->driver_context;
+	int ret = 0;
+
+	/* 0: successful or smu.ppt_funcs->set_deep_sleep_dcefclk = NULL;1: fail */
+	ret = amdgpu_dpm_set_min_deep_sleep_dcefclk(adev, mhz);
+	if (ret == -EOPNOTSUPP)
+		return PP_SMU_RESULT_UNSUPPORTED;
+	else if (ret)
+		return PP_SMU_RESULT_FAIL;
+
+	return PP_SMU_RESULT_OK;
+}
+
+static enum pp_smu_status pp_nv_set_hard_min_dcefclk_by_freq(
+		struct pp_smu *pp, int mhz)
+{
+	const struct dc_context *ctx = pp->dm;
+	struct amdgpu_device *adev = ctx->driver_context;
+	struct pp_display_clock_request clock_req;
+	int ret = 0;
+
+	clock_req.clock_type = amd_pp_dcef_clock;
+	clock_req.clock_freq_in_khz = mhz * 1000;
+
+	/* 0: successful or smu.ppt_funcs->display_clock_voltage_request = NULL
+	 * 1: fail
+	 */
+	ret = amdgpu_dpm_display_clock_voltage_request(adev, &clock_req);
+	if (ret == -EOPNOTSUPP)
+		return PP_SMU_RESULT_UNSUPPORTED;
+	else if (ret)
+		return PP_SMU_RESULT_FAIL;
+
+	return PP_SMU_RESULT_OK;
+}
+
+static enum pp_smu_status
+pp_nv_set_hard_min_uclk_by_freq(struct pp_smu *pp, int mhz)
+{
+	const struct dc_context *ctx = pp->dm;
+	struct amdgpu_device *adev = ctx->driver_context;
+	struct pp_display_clock_request clock_req;
+	int ret = 0;
+
+	clock_req.clock_type = amd_pp_mem_clock;
+	clock_req.clock_freq_in_khz = mhz * 1000;
+
+	/* 0: successful or smu.ppt_funcs->display_clock_voltage_request = NULL
+	 * 1: fail
+	 */
+	ret = amdgpu_dpm_display_clock_voltage_request(adev, &clock_req);
+	if (ret == -EOPNOTSUPP)
+		return PP_SMU_RESULT_UNSUPPORTED;
+	else if (ret)
+		return PP_SMU_RESULT_FAIL;
+
+	return PP_SMU_RESULT_OK;
+}
+
+static enum pp_smu_status pp_nv_set_pstate_handshake_support(
+	struct pp_smu *pp, bool pstate_handshake_supported)
+{
+	const struct dc_context *ctx = pp->dm;
+	struct amdgpu_device *adev = ctx->driver_context;
+
+	if (amdgpu_dpm_display_disable_memory_clock_switch(adev,
+							  !pstate_handshake_supported))
+		return PP_SMU_RESULT_FAIL;
+
+	return PP_SMU_RESULT_OK;
+}
+
+static enum pp_smu_status pp_nv_set_voltage_by_freq(struct pp_smu *pp,
+		enum pp_smu_nv_clock_id clock_id, int mhz)
+{
+	const struct dc_context *ctx = pp->dm;
+	struct amdgpu_device *adev = ctx->driver_context;
+	struct pp_display_clock_request clock_req;
+	int ret = 0;
+
+	switch (clock_id) {
+	case PP_SMU_NV_DISPCLK:
+		clock_req.clock_type = amd_pp_disp_clock;
+		break;
+	case PP_SMU_NV_PHYCLK:
+		clock_req.clock_type = amd_pp_phy_clock;
+		break;
+	case PP_SMU_NV_PIXELCLK:
+		clock_req.clock_type = amd_pp_pixel_clock;
+		break;
+	default:
+		break;
+	}
+	clock_req.clock_freq_in_khz = mhz * 1000;
+
+	/* 0: successful or smu.ppt_funcs->display_clock_voltage_request = NULL
+	 * 1: fail
+	 */
+	ret = amdgpu_dpm_display_clock_voltage_request(adev, &clock_req);
+	if (ret == -EOPNOTSUPP)
+		return PP_SMU_RESULT_UNSUPPORTED;
+	else if (ret)
+		return PP_SMU_RESULT_FAIL;
+
+	return PP_SMU_RESULT_OK;
+}
+
+static enum pp_smu_status pp_nv_get_maximum_sustainable_clocks(
+		struct pp_smu *pp, struct pp_smu_nv_clock_table *max_clocks)
+{
+	const struct dc_context *ctx = pp->dm;
+	struct amdgpu_device *adev = ctx->driver_context;
+	int ret = 0;
+
+	ret = amdgpu_dpm_get_max_sustainable_clocks_by_dc(adev,
+							  max_clocks);
+	if (ret == -EOPNOTSUPP)
+		return PP_SMU_RESULT_UNSUPPORTED;
+	else if (ret)
+		return PP_SMU_RESULT_FAIL;
+
+	return PP_SMU_RESULT_OK;
+}
+
+static enum pp_smu_status pp_nv_get_uclk_dpm_states(struct pp_smu *pp,
+		unsigned int *clock_values_in_khz, unsigned int *num_states)
+{
+	const struct dc_context *ctx = pp->dm;
+	struct amdgpu_device *adev = ctx->driver_context;
+	int ret = 0;
+
+	ret = amdgpu_dpm_get_uclk_dpm_states(adev,
+					     clock_values_in_khz,
+					    num_states);
+	if (ret == -EOPNOTSUPP)
+		return PP_SMU_RESULT_UNSUPPORTED;
+	else if (ret)
+		return PP_SMU_RESULT_FAIL;
+
+	return PP_SMU_RESULT_OK;
+}
+
+static enum pp_smu_status pp_rn_get_dpm_clock_table(
+		struct pp_smu *pp, struct dpm_clocks *clock_table)
+{
+	const struct dc_context *ctx = pp->dm;
+	struct amdgpu_device *adev = ctx->driver_context;
+	int ret = 0;
+
+	ret = amdgpu_dpm_get_dpm_clock_table(adev, clock_table);
+	if (ret == -EOPNOTSUPP)
+		return PP_SMU_RESULT_UNSUPPORTED;
+	else if (ret)
+		return PP_SMU_RESULT_FAIL;
+
+	return PP_SMU_RESULT_OK;
+}
+
+static enum pp_smu_status pp_rn_set_wm_ranges(struct pp_smu *pp,
+		struct pp_smu_wm_range_sets *ranges)
+{
+	const struct dc_context *ctx = pp->dm;
+	struct amdgpu_device *adev = ctx->driver_context;
+
+	amdgpu_dpm_set_watermarks_for_clocks_ranges(adev, ranges);
+
+	return PP_SMU_RESULT_OK;
+}
+
+void dm_pp_get_funcs(
+		struct dc_context *ctx,
+		struct pp_smu_funcs *funcs)
+{
+	switch (ctx->dce_version) {
+	case DCN_VERSION_1_0:
+	case DCN_VERSION_1_01:
+		funcs->ctx.ver = PP_SMU_VER_RV;
+		funcs->rv_funcs.pp_smu.dm = ctx;
+		funcs->rv_funcs.set_wm_ranges = pp_rv_set_wm_ranges;
+		funcs->rv_funcs.set_pme_wa_enable = pp_rv_set_pme_wa_enable;
+		funcs->rv_funcs.set_display_count =
+				pp_rv_set_active_display_count;
+		funcs->rv_funcs.set_min_deep_sleep_dcfclk =
+				pp_rv_set_min_deep_sleep_dcfclk;
+		funcs->rv_funcs.set_hard_min_dcfclk_by_freq =
+				pp_rv_set_hard_min_dcefclk_by_freq;
+		funcs->rv_funcs.set_hard_min_fclk_by_freq =
+				pp_rv_set_hard_min_fclk_by_freq;
+		break;
+	case DCN_VERSION_2_0:
+		funcs->ctx.ver = PP_SMU_VER_NV;
+		funcs->nv_funcs.pp_smu.dm = ctx;
+		funcs->nv_funcs.set_display_count = pp_nv_set_display_count;
+		funcs->nv_funcs.set_hard_min_dcfclk_by_freq =
+				pp_nv_set_hard_min_dcefclk_by_freq;
+		funcs->nv_funcs.set_min_deep_sleep_dcfclk =
+				pp_nv_set_min_deep_sleep_dcfclk;
+		funcs->nv_funcs.set_voltage_by_freq =
+				pp_nv_set_voltage_by_freq;
+		funcs->nv_funcs.set_wm_ranges = pp_nv_set_wm_ranges;
+
+		/* todo set_pme_wa_enable cause 4k@6ohz display not light up */
+		funcs->nv_funcs.set_pme_wa_enable = NULL;
+		/* todo debug waring message */
+		funcs->nv_funcs.set_hard_min_uclk_by_freq = pp_nv_set_hard_min_uclk_by_freq;
+		/* todo  compare data with window driver*/
+		funcs->nv_funcs.get_maximum_sustainable_clocks = pp_nv_get_maximum_sustainable_clocks;
+		/*todo  compare data with window driver */
+		funcs->nv_funcs.get_uclk_dpm_states = pp_nv_get_uclk_dpm_states;
+		funcs->nv_funcs.set_pstate_handshake_support = pp_nv_set_pstate_handshake_support;
+		break;
+
+	case DCN_VERSION_2_1:
+		funcs->ctx.ver = PP_SMU_VER_RN;
+		funcs->rn_funcs.pp_smu.dm = ctx;
+		funcs->rn_funcs.set_wm_ranges = pp_rn_set_wm_ranges;
+		funcs->rn_funcs.get_dpm_clock_table = pp_rn_get_dpm_clock_table;
+		break;
+	default:
+		DRM_ERROR("smu version is not supported !\n");
+		break;
+	}
+}
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_psr.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_psr.c
new file mode 100644
index 000000000000..fd491b7a3cd7
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_psr.c
@@ -0,0 +1,277 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "amdgpu_dm_psr.h"
+#include "dc_dmub_srv.h"
+#include "dc.h"
+#include "amdgpu_dm.h"
+#include "modules/power/power_helpers.h"
+
+static bool link_supports_psrsu(struct dc_link *link)
+{
+	struct dc *dc = link->ctx->dc;
+
+	if (!dc->caps.dmcub_support)
+		return false;
+
+	if (dc->ctx->dce_version < DCN_VERSION_3_1)
+		return false;
+
+	if (!is_psr_su_specific_panel(link))
+		return false;
+
+	if (!link->dpcd_caps.alpm_caps.bits.AUX_WAKE_ALPM_CAP ||
+	    !link->dpcd_caps.psr_info.psr_dpcd_caps.bits.Y_COORDINATE_REQUIRED)
+		return false;
+
+	if (link->dpcd_caps.psr_info.psr_dpcd_caps.bits.SU_GRANULARITY_REQUIRED &&
+	    !link->dpcd_caps.psr_info.psr2_su_y_granularity_cap)
+		return false;
+
+	if (amdgpu_dc_debug_mask & DC_DISABLE_PSR_SU)
+		return false;
+
+	/* Temporarily disable PSR-SU to avoid glitches */
+	return false;
+}
+
+/*
+ * amdgpu_dm_set_psr_caps() - set link psr capabilities
+ * @link: link
+ *
+ */
+void amdgpu_dm_set_psr_caps(struct dc_link *link)
+{
+	if (!(link->connector_signal & SIGNAL_TYPE_EDP)) {
+		link->psr_settings.psr_feature_enabled = false;
+		return;
+	}
+
+	if (link->type == dc_connection_none) {
+		link->psr_settings.psr_feature_enabled = false;
+		return;
+	}
+
+	if (link->dpcd_caps.psr_info.psr_version == 0) {
+		link->psr_settings.psr_version = DC_PSR_VERSION_UNSUPPORTED;
+		link->psr_settings.psr_feature_enabled = false;
+
+	} else {
+		if (link_supports_psrsu(link))
+			link->psr_settings.psr_version = DC_PSR_VERSION_SU_1;
+		else
+			link->psr_settings.psr_version = DC_PSR_VERSION_1;
+
+		link->psr_settings.psr_feature_enabled = true;
+	}
+}
+
+/*
+ * amdgpu_dm_link_setup_psr() - configure psr link
+ * @stream: stream state
+ *
+ * Return: true if success
+ */
+bool amdgpu_dm_link_setup_psr(struct dc_stream_state *stream)
+{
+	struct dc_link *link = NULL;
+	struct psr_config psr_config = {0};
+	struct psr_context psr_context = {0};
+	struct dc *dc = NULL;
+	bool ret = false;
+
+	if (stream == NULL)
+		return false;
+
+	link = stream->link;
+	dc = link->ctx->dc;
+
+	if (link->psr_settings.psr_version != DC_PSR_VERSION_UNSUPPORTED) {
+		mod_power_calc_psr_configs(&psr_config, link, stream);
+
+		/* linux DM specific updating for psr config fields */
+		psr_config.allow_smu_optimizations =
+			(amdgpu_dc_feature_mask & DC_PSR_ALLOW_SMU_OPT) &&
+			mod_power_only_edp(dc->current_state, stream);
+		psr_config.allow_multi_disp_optimizations =
+			(amdgpu_dc_feature_mask & DC_PSR_ALLOW_MULTI_DISP_OPT);
+
+		if (link->psr_settings.psr_version == DC_PSR_VERSION_SU_1) {
+			if (!psr_su_set_dsc_slice_height(dc, link, stream, &psr_config))
+				return false;
+		}
+
+		ret = dc_link_setup_psr(link, stream, &psr_config, &psr_context);
+
+	}
+	DRM_DEBUG_DRIVER("PSR link: %d\n",	link->psr_settings.psr_feature_enabled);
+
+	return ret;
+}
+
+/*
+ * amdgpu_dm_psr_enable() - enable psr f/w
+ * @stream: stream state
+ *
+ */
+void amdgpu_dm_psr_enable(struct dc_stream_state *stream)
+{
+	struct dc_link *link = stream->link;
+	unsigned int vsync_rate_hz = 0;
+	struct dc_static_screen_params params = {0};
+	/* Calculate number of static frames before generating interrupt to
+	 * enter PSR.
+	 */
+	// Init fail safe of 2 frames static
+	unsigned int num_frames_static = 2;
+	unsigned int power_opt = 0;
+	bool psr_enable = true;
+
+	DRM_DEBUG_DRIVER("Enabling psr...\n");
+
+	vsync_rate_hz = div64_u64(div64_u64((
+			stream->timing.pix_clk_100hz * (uint64_t)100),
+			stream->timing.v_total),
+			stream->timing.h_total);
+
+	/* Round up
+	 * Calculate number of frames such that at least 30 ms of time has
+	 * passed.
+	 */
+	if (vsync_rate_hz != 0) {
+		unsigned int frame_time_microsec = 1000000 / vsync_rate_hz;
+
+		num_frames_static = (30000 / frame_time_microsec) + 1;
+	}
+
+	params.triggers.cursor_update = true;
+	params.triggers.overlay_update = true;
+	params.triggers.surface_update = true;
+	params.num_frames = num_frames_static;
+
+	dc_stream_set_static_screen_params(link->ctx->dc,
+					   &stream, 1,
+					   &params);
+
+	/*
+	 * Only enable static-screen optimizations for PSR1. For PSR SU, this
+	 * causes vstartup interrupt issues, used by amdgpu_dm to send vblank
+	 * events.
+	 */
+	if (link->psr_settings.psr_version < DC_PSR_VERSION_SU_1)
+		power_opt |= psr_power_opt_z10_static_screen;
+
+	dc_link_set_psr_allow_active(link, &psr_enable, false, false, &power_opt);
+
+	if (link->ctx->dc->caps.ips_support)
+		dc_allow_idle_optimizations(link->ctx->dc, true);
+}
+
+/*
+ * amdgpu_dm_psr_disable() - disable psr f/w
+ * @stream:  stream state
+ *
+ * Return: true if success
+ */
+bool amdgpu_dm_psr_disable(struct dc_stream_state *stream, bool wait)
+{
+	bool psr_enable = false;
+
+	DRM_DEBUG_DRIVER("Disabling psr...\n");
+
+	return dc_link_set_psr_allow_active(stream->link, &psr_enable, wait, false, NULL);
+}
+
+/*
+ * amdgpu_dm_psr_disable_all() - disable psr f/w for all streams
+ * if psr is enabled on any stream
+ *
+ * Return: true if success
+ */
+bool amdgpu_dm_psr_disable_all(struct amdgpu_display_manager *dm)
+{
+	DRM_DEBUG_DRIVER("Disabling psr if psr is enabled on any stream\n");
+	return dc_set_psr_allow_active(dm->dc, false);
+}
+
+/*
+ * amdgpu_dm_psr_is_active_allowed() - check if psr is allowed on any stream
+ * @dm:  pointer to amdgpu_display_manager
+ *
+ * Return: true if allowed
+ */
+
+bool amdgpu_dm_psr_is_active_allowed(struct amdgpu_display_manager *dm)
+{
+	unsigned int i;
+	bool allow_active = false;
+
+	for (i = 0; i < dm->dc->current_state->stream_count ; i++) {
+		struct dc_link *link;
+		struct dc_stream_state *stream = dm->dc->current_state->streams[i];
+
+		link = stream->link;
+		if (!link)
+			continue;
+		if (link->psr_settings.psr_feature_enabled &&
+		    link->psr_settings.psr_allow_active) {
+			allow_active = true;
+			break;
+		}
+	}
+
+	return allow_active;
+}
+
+/**
+ * amdgpu_dm_psr_wait_disable() - Wait for eDP panel to exit PSR
+ * @stream: stream state attached to the eDP link
+ *
+ * Waits for a max of 500ms for the eDP panel to exit PSR.
+ *
+ * Return: true if panel exited PSR, false otherwise.
+ */
+bool amdgpu_dm_psr_wait_disable(struct dc_stream_state *stream)
+{
+	enum dc_psr_state psr_state = PSR_STATE0;
+	struct dc_link *link = stream->link;
+	int retry_count;
+
+	if (link == NULL)
+		return false;
+
+	for (retry_count = 0; retry_count <= 1000; retry_count++) {
+		dc_link_get_psr_state(link, &psr_state);
+		if (psr_state == PSR_STATE0)
+			break;
+		udelay(500);
+	}
+
+	if (retry_count == 1000)
+		return false;
+
+	return true;
+}
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_psr.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_psr.h
new file mode 100644
index 000000000000..4fb8626913cf
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_psr.h
@@ -0,0 +1,43 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef AMDGPU_DM_AMDGPU_DM_PSR_H_
+#define AMDGPU_DM_AMDGPU_DM_PSR_H_
+
+#include "amdgpu.h"
+
+/* the number of pageflips before enabling psr */
+#define AMDGPU_DM_PSR_ENTRY_DELAY 5
+
+void amdgpu_dm_set_psr_caps(struct dc_link *link);
+void amdgpu_dm_psr_enable(struct dc_stream_state *stream);
+bool amdgpu_dm_link_setup_psr(struct dc_stream_state *stream);
+bool amdgpu_dm_psr_disable(struct dc_stream_state *stream, bool wait);
+bool amdgpu_dm_psr_disable_all(struct amdgpu_display_manager *dm);
+bool amdgpu_dm_psr_is_active_allowed(struct amdgpu_display_manager *dm);
+bool amdgpu_dm_psr_wait_disable(struct dc_stream_state *stream);
+
+#endif /* AMDGPU_DM_AMDGPU_DM_PSR_H_ */
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_quirks.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_quirks.c
new file mode 100644
index 000000000000..1da07ebf9217
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_quirks.c
@@ -0,0 +1,178 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2025 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include <linux/dmi.h>
+
+#include "amdgpu.h"
+#include "amdgpu_dm.h"
+
+struct amdgpu_dm_quirks {
+	bool aux_hpd_discon;
+	bool support_edp0_on_dp1;
+};
+
+static struct amdgpu_dm_quirks quirk_entries = {
+	.aux_hpd_discon = false,
+	.support_edp0_on_dp1 = false
+};
+
+static int edp0_on_dp1_callback(const struct dmi_system_id *id)
+{
+	quirk_entries.support_edp0_on_dp1 = true;
+	return 0;
+}
+
+static int aux_hpd_discon_callback(const struct dmi_system_id *id)
+{
+	quirk_entries.aux_hpd_discon = true;
+	return 0;
+}
+
+static const struct dmi_system_id dmi_quirk_table[] = {
+	{
+		.callback = aux_hpd_discon_callback,
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
+			DMI_MATCH(DMI_PRODUCT_NAME, "Precision 3660"),
+		},
+	},
+	{
+		.callback = aux_hpd_discon_callback,
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
+			DMI_MATCH(DMI_PRODUCT_NAME, "Precision 3260"),
+		},
+	},
+	{
+		.callback = aux_hpd_discon_callback,
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
+			DMI_MATCH(DMI_PRODUCT_NAME, "Precision 3460"),
+		},
+	},
+	{
+		.callback = aux_hpd_discon_callback,
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
+			DMI_MATCH(DMI_PRODUCT_NAME, "OptiPlex Tower Plus 7010"),
+		},
+	},
+	{
+		.callback = aux_hpd_discon_callback,
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
+			DMI_MATCH(DMI_PRODUCT_NAME, "OptiPlex Tower 7010"),
+		},
+	},
+	{
+		.callback = aux_hpd_discon_callback,
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
+			DMI_MATCH(DMI_PRODUCT_NAME, "OptiPlex SFF Plus 7010"),
+		},
+	},
+	{
+		.callback = aux_hpd_discon_callback,
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
+			DMI_MATCH(DMI_PRODUCT_NAME, "OptiPlex SFF 7010"),
+		},
+	},
+	{
+		.callback = aux_hpd_discon_callback,
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
+			DMI_MATCH(DMI_PRODUCT_NAME, "OptiPlex Micro Plus 7010"),
+		},
+	},
+	{
+		.callback = aux_hpd_discon_callback,
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
+			DMI_MATCH(DMI_PRODUCT_NAME, "OptiPlex Micro 7010"),
+		},
+	},
+	{
+		.callback = edp0_on_dp1_callback,
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "HP"),
+			DMI_MATCH(DMI_PRODUCT_NAME, "HP Elite mt645 G8 Mobile Thin Client"),
+		},
+	},
+	{
+		.callback = edp0_on_dp1_callback,
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "HP"),
+			DMI_MATCH(DMI_PRODUCT_NAME, "HP EliteBook 645 14 inch G11 Notebook PC"),
+		},
+	},
+	{
+		.callback = edp0_on_dp1_callback,
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "HP"),
+			DMI_MATCH(DMI_PRODUCT_NAME, "HP EliteBook 665 16 inch G11 Notebook PC"),
+		},
+	},
+	{
+		.callback = edp0_on_dp1_callback,
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "HP"),
+			DMI_MATCH(DMI_PRODUCT_NAME, "HP ProBook 445 14 inch G11 Notebook PC"),
+		},
+	},
+	{
+		.callback = edp0_on_dp1_callback,
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "HP"),
+			DMI_MATCH(DMI_PRODUCT_NAME, "HP ProBook 465 16 inch G11 Notebook PC"),
+		},
+	},
+	{}
+	/* TODO: refactor this from a fixed table to a dynamic option */
+};
+
+void retrieve_dmi_info(struct amdgpu_display_manager *dm)
+{
+	struct drm_device *dev = dm->ddev;
+	int dmi_id;
+
+	dm->aux_hpd_discon_quirk = false;
+	dm->edp0_on_dp1_quirk = false;
+
+	dmi_id = dmi_check_system(dmi_quirk_table);
+
+	if (!dmi_id)
+		return;
+
+	if (quirk_entries.aux_hpd_discon) {
+		dm->aux_hpd_discon_quirk = true;
+		drm_info(dev, "aux_hpd_discon_quirk attached\n");
+	}
+	if (quirk_entries.support_edp0_on_dp1) {
+		dm->edp0_on_dp1_quirk = true;
+		drm_info(dev, "support_edp0_on_dp1 attached\n");
+	}
+}
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_replay.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_replay.c
new file mode 100644
index 000000000000..80704d709e44
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_replay.c
@@ -0,0 +1,209 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "amdgpu_dm_replay.h"
+#include "dc_dmub_srv.h"
+#include "dc.h"
+#include "dm_helpers.h"
+#include "amdgpu_dm.h"
+#include "modules/power/power_helpers.h"
+#include "dmub/inc/dmub_cmd.h"
+#include "dc/inc/link_service.h"
+
+/*
+ * amdgpu_dm_link_supports_replay() - check if the link supports replay
+ * @link: link
+ * @aconnector: aconnector
+ *
+ */
+bool amdgpu_dm_link_supports_replay(struct dc_link *link, struct amdgpu_dm_connector *aconnector)
+{
+	struct dm_connector_state *state = to_dm_connector_state(aconnector->base.state);
+	struct dpcd_caps *dpcd_caps = &link->dpcd_caps;
+	struct adaptive_sync_caps *as_caps = &link->dpcd_caps.adaptive_sync_caps;
+
+	if (!state->freesync_capable)
+		return false;
+
+	if (!aconnector->vsdb_info.replay_mode)
+		return false;
+
+	// Check the eDP version
+	if (dpcd_caps->edp_rev < EDP_REVISION_13)
+		return false;
+
+	if (!dpcd_caps->alpm_caps.bits.AUX_WAKE_ALPM_CAP)
+		return false;
+
+	// Check adaptive sync support cap
+	if (!as_caps->dp_adap_sync_caps.bits.ADAPTIVE_SYNC_SDP_SUPPORT)
+		return false;
+
+	// Sink shall populate line deviation information
+	if (dpcd_caps->pr_info.pixel_deviation_per_line == 0 ||
+		dpcd_caps->pr_info.max_deviation_line == 0)
+		return false;
+
+	return true;
+}
+
+/*
+ * amdgpu_dm_set_replay_caps() - setup Replay capabilities
+ * @link: link
+ * @aconnector: aconnector
+ *
+ */
+bool amdgpu_dm_set_replay_caps(struct dc_link *link, struct amdgpu_dm_connector *aconnector)
+{
+	struct replay_config pr_config = { 0 };
+	union replay_debug_flags *debug_flags = NULL;
+	struct dc *dc = link->ctx->dc;
+
+	// If Replay is already set to support, return true to skip checks
+	if (link->replay_settings.config.replay_supported)
+		return true;
+
+	if (!dc_is_embedded_signal(link->connector_signal))
+		return false;
+
+	if (link->panel_config.psr.disallow_replay)
+		return false;
+
+	if (!amdgpu_dm_link_supports_replay(link, aconnector))
+		return false;
+
+	if (!dc->ctx->dmub_srv || !dc->ctx->dmub_srv->dmub ||
+		!dc->ctx->dmub_srv->dmub->feature_caps.replay_supported)
+		return false;
+
+	// Mark Replay is supported in pr_config
+	pr_config.replay_supported = true;
+
+	debug_flags = (union replay_debug_flags *)&pr_config.debug_flags;
+	debug_flags->u32All = 0;
+	debug_flags->bitfields.visual_confirm =
+		link->ctx->dc->debug.visual_confirm == VISUAL_CONFIRM_REPLAY;
+
+	init_replay_config(link, &pr_config);
+
+	return true;
+}
+
+/*
+ * amdgpu_dm_link_setup_replay() - configure replay link
+ * @link: link
+ * @aconnector: aconnector
+ *
+ */
+bool amdgpu_dm_link_setup_replay(struct dc_link *link, struct amdgpu_dm_connector *aconnector)
+{
+	struct replay_config *pr_config;
+
+	if (link == NULL || aconnector == NULL)
+		return false;
+
+	pr_config = &link->replay_settings.config;
+
+	if (!pr_config->replay_supported)
+		return false;
+
+	pr_config->replay_power_opt_supported = 0x11;
+	pr_config->replay_smu_opt_supported = false;
+	pr_config->replay_enable_option |= pr_enable_option_static_screen;
+	pr_config->replay_support_fast_resync_in_ultra_sleep_mode = aconnector->max_vfreq >= 2 * aconnector->min_vfreq;
+	pr_config->replay_timing_sync_supported = false;
+
+	if (!pr_config->replay_timing_sync_supported)
+		pr_config->replay_enable_option &= ~pr_enable_option_general_ui;
+
+	link->replay_settings.replay_feature_enabled = true;
+
+	return true;
+}
+
+/*
+ * amdgpu_dm_replay_enable() - enable replay f/w
+ * @stream: stream state
+ *
+ * Return: true if success
+ */
+bool amdgpu_dm_replay_enable(struct dc_stream_state *stream, bool wait)
+{
+	bool replay_active = true;
+	struct dc_link *link = NULL;
+
+	if (stream == NULL)
+		return false;
+
+	link = stream->link;
+
+	if (link) {
+		link->dc->link_srv->edp_setup_replay(link, stream);
+		link->dc->link_srv->edp_set_coasting_vtotal(link, stream->timing.v_total);
+		DRM_DEBUG_DRIVER("Enabling replay...\n");
+		link->dc->link_srv->edp_set_replay_allow_active(link, &replay_active, wait, false, NULL);
+		return true;
+	}
+
+	return false;
+}
+
+/*
+ * amdgpu_dm_replay_disable() - disable replay f/w
+ * @stream:  stream state
+ *
+ * Return: true if success
+ */
+bool amdgpu_dm_replay_disable(struct dc_stream_state *stream)
+{
+	bool replay_active = false;
+	struct dc_link *link = NULL;
+
+	if (stream == NULL)
+		return false;
+
+	link = stream->link;
+
+	if (link) {
+		DRM_DEBUG_DRIVER("Disabling replay...\n");
+		link->dc->link_srv->edp_set_replay_allow_active(stream->link, &replay_active, true, false, NULL);
+		return true;
+	}
+
+	return false;
+}
+
+/*
+ * amdgpu_dm_replay_disable_all() - disable replay f/w
+ * if replay is enabled on any stream
+ *
+ * Return: true if success
+ */
+bool amdgpu_dm_replay_disable_all(struct amdgpu_display_manager *dm)
+{
+	DRM_DEBUG_DRIVER("Disabling replay if replay is enabled on any stream\n");
+	return dc_set_replay_allow_active(dm->dc, false);
+}
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_replay.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_replay.h
new file mode 100644
index 000000000000..73b6c67ae5e7
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_replay.h
@@ -0,0 +1,49 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef AMDGPU_DM_AMDGPU_DM_REPLAY_H_
+#define AMDGPU_DM_AMDGPU_DM_REPLAY_H_
+
+#include "amdgpu.h"
+
+enum replay_enable_option {
+	pr_enable_option_static_screen = 0x1,
+	pr_enable_option_mpo_video = 0x2,
+	pr_enable_option_full_screen_video = 0x4,
+	pr_enable_option_general_ui = 0x8,
+	pr_enable_option_static_screen_coasting = 0x10000,
+	pr_enable_option_mpo_video_coasting = 0x20000,
+	pr_enable_option_full_screen_video_coasting = 0x40000,
+};
+
+bool amdgpu_dm_link_supports_replay(struct dc_link *link, struct amdgpu_dm_connector *aconnector);
+bool amdgpu_dm_replay_enable(struct dc_stream_state *stream, bool enable);
+bool amdgpu_dm_set_replay_caps(struct dc_link *link, struct amdgpu_dm_connector *aconnector);
+bool amdgpu_dm_link_setup_replay(struct dc_link *link, struct amdgpu_dm_connector *aconnector);
+bool amdgpu_dm_replay_disable(struct dc_stream_state *stream);
+bool amdgpu_dm_replay_disable_all(struct amdgpu_display_manager *dm);
+
+#endif /* AMDGPU_DM_AMDGPU_DM_REPLAY_H_ */
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_services.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_services.c
new file mode 100644
index 000000000000..8550d5e8b753
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_services.c
@@ -0,0 +1,64 @@
+//SPDX-License-Identifier: MIT
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include <linux/string.h>
+#include <linux/acpi.h>
+
+#include <drm/drm_probe_helper.h>
+#include <drm/amdgpu_drm.h>
+#include "dm_services.h"
+#include "amdgpu.h"
+#include "amdgpu_dm.h"
+#include "amdgpu_dm_irq.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_dm_trace.h"
+
+	unsigned long long
+	dm_get_elapse_time_in_ns(struct dc_context *ctx,
+				 unsigned long long current_time_stamp,
+				 unsigned long long last_time_stamp)
+{
+	return current_time_stamp - last_time_stamp;
+}
+
+void dm_perf_trace_timestamp(const char *func_name, unsigned int line, struct dc_context *ctx)
+{
+	trace_amdgpu_dc_performance(ctx->perf_trace->read_count,
+				    ctx->perf_trace->write_count,
+				    &ctx->perf_trace->last_entry_read,
+				    &ctx->perf_trace->last_entry_write,
+				    func_name, line);
+}
+
+void dm_trace_smu_enter(uint32_t msg_id, uint32_t param_in, unsigned int delay, struct dc_context *ctx)
+{
+}
+
+void dm_trace_smu_exit(bool success, uint32_t response, struct dc_context *ctx)
+{
+}
+
+/**** power component interfaces ****/
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_trace.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_trace.h
new file mode 100644
index 000000000000..aa56fd6d56c3
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_trace.h
@@ -0,0 +1,761 @@
+//SPDX-License-Identifier: MIT
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM amdgpu_dm
+
+#if !defined(_AMDGPU_DM_TRACE_H_) || defined(TRACE_HEADER_MULTI_READ)
+#define _AMDGPU_DM_TRACE_H_
+
+#include <linux/tracepoint.h>
+#include <drm/drm_connector.h>
+#include <drm/drm_crtc.h>
+#include <drm/drm_plane.h>
+#include <drm/drm_fourcc.h>
+#include <drm/drm_framebuffer.h>
+#include <drm/drm_encoder.h>
+#include <drm/drm_atomic.h>
+#include "dc/inc/hw/optc.h"
+
+#include "dc/inc/core_types.h"
+
+DECLARE_EVENT_CLASS(amdgpu_dc_reg_template,
+		    TP_PROTO(unsigned long *count, uint32_t reg, uint32_t value),
+		    TP_ARGS(count, reg, value),
+
+		    TP_STRUCT__entry(
+				     __field(uint32_t, reg)
+				     __field(uint32_t, value)
+		    ),
+
+		    TP_fast_assign(
+				   __entry->reg = reg;
+				   __entry->value = value;
+				   *count = *count + 1;
+		    ),
+
+		    TP_printk("reg=0x%08lx, value=0x%08lx",
+			      (unsigned long)__entry->reg,
+			      (unsigned long)__entry->value)
+);
+
+DEFINE_EVENT(amdgpu_dc_reg_template, amdgpu_dc_rreg,
+	     TP_PROTO(unsigned long *count, uint32_t reg, uint32_t value),
+	     TP_ARGS(count, reg, value));
+
+DEFINE_EVENT(amdgpu_dc_reg_template, amdgpu_dc_wreg,
+	     TP_PROTO(unsigned long *count, uint32_t reg, uint32_t value),
+	     TP_ARGS(count, reg, value));
+
+TRACE_EVENT(amdgpu_dc_performance,
+	TP_PROTO(unsigned long read_count, unsigned long write_count,
+		unsigned long *last_read, unsigned long *last_write,
+		const char *func, unsigned int line),
+	TP_ARGS(read_count, write_count, last_read, last_write, func, line),
+	TP_STRUCT__entry(
+			__field(uint32_t, reads)
+			__field(uint32_t, writes)
+			__field(uint32_t, read_delta)
+			__field(uint32_t, write_delta)
+			__string(func, func)
+			__field(uint32_t, line)
+			),
+	TP_fast_assign(
+			__entry->reads = read_count;
+			__entry->writes = write_count;
+			__entry->read_delta = read_count - *last_read;
+			__entry->write_delta = write_count - *last_write;
+			__assign_str(func);
+			__entry->line = line;
+			*last_read = read_count;
+			*last_write = write_count;
+			),
+	TP_printk("%s:%d reads=%08ld (%08ld total), writes=%08ld (%08ld total)",
+			__get_str(func), __entry->line,
+			(unsigned long)__entry->read_delta,
+			(unsigned long)__entry->reads,
+			(unsigned long)__entry->write_delta,
+			(unsigned long)__entry->writes)
+);
+
+TRACE_EVENT(amdgpu_dm_connector_atomic_check,
+	    TP_PROTO(const struct drm_connector_state *state),
+	    TP_ARGS(state),
+
+	    TP_STRUCT__entry(
+			     __field(uint32_t, conn_id)
+			     __field(const struct drm_connector_state *, conn_state)
+			     __field(const struct drm_atomic_state *, state)
+			     __field(const struct drm_crtc_commit *, commit)
+			     __field(uint32_t, crtc_id)
+			     __field(uint32_t, best_encoder_id)
+			     __field(enum drm_link_status, link_status)
+			     __field(bool, self_refresh_aware)
+			     __field(enum hdmi_picture_aspect, picture_aspect_ratio)
+			     __field(unsigned int, content_type)
+			     __field(unsigned int, hdcp_content_type)
+			     __field(unsigned int, content_protection)
+			     __field(unsigned int, scaling_mode)
+			     __field(u32, colorspace)
+			     __field(u8, max_requested_bpc)
+			     __field(u8, max_bpc)
+	    ),
+
+	    TP_fast_assign(
+			   __entry->conn_id = state->connector->base.id;
+			   __entry->conn_state = state;
+			   __entry->state = state->state;
+			   __entry->commit = state->commit;
+			   __entry->crtc_id = state->crtc ? state->crtc->base.id : 0;
+			   __entry->best_encoder_id = state->best_encoder ?
+						      state->best_encoder->base.id : 0;
+			   __entry->link_status = state->link_status;
+			   __entry->self_refresh_aware = state->self_refresh_aware;
+			   __entry->picture_aspect_ratio = state->picture_aspect_ratio;
+			   __entry->content_type = state->content_type;
+			   __entry->hdcp_content_type = state->hdcp_content_type;
+			   __entry->content_protection = state->content_protection;
+			   __entry->scaling_mode = state->scaling_mode;
+			   __entry->colorspace = state->colorspace;
+			   __entry->max_requested_bpc = state->max_requested_bpc;
+			   __entry->max_bpc = state->max_bpc;
+	    ),
+
+	    TP_printk("conn_id=%u conn_state=%p state=%p commit=%p crtc_id=%u "
+		      "best_encoder_id=%u link_status=%d self_refresh_aware=%d "
+		      "picture_aspect_ratio=%d content_type=%u "
+		      "hdcp_content_type=%u content_protection=%u scaling_mode=%u "
+		      "colorspace=%u max_requested_bpc=%u max_bpc=%u",
+		      __entry->conn_id, __entry->conn_state, __entry->state,
+		      __entry->commit, __entry->crtc_id, __entry->best_encoder_id,
+		      __entry->link_status, __entry->self_refresh_aware,
+		      __entry->picture_aspect_ratio, __entry->content_type,
+		      __entry->hdcp_content_type, __entry->content_protection,
+		      __entry->scaling_mode, __entry->colorspace,
+		      __entry->max_requested_bpc, __entry->max_bpc)
+);
+
+TRACE_EVENT(amdgpu_dm_crtc_atomic_check,
+	    TP_PROTO(const struct drm_crtc_state *state),
+	    TP_ARGS(state),
+
+	    TP_STRUCT__entry(
+			     __field(const struct drm_atomic_state *, state)
+			     __field(const struct drm_crtc_state *, crtc_state)
+			     __field(const struct drm_crtc_commit *, commit)
+			     __field(uint32_t, crtc_id)
+			     __field(bool, enable)
+			     __field(bool, active)
+			     __field(bool, planes_changed)
+			     __field(bool, mode_changed)
+			     __field(bool, active_changed)
+			     __field(bool, connectors_changed)
+			     __field(bool, zpos_changed)
+			     __field(bool, color_mgmt_changed)
+			     __field(bool, no_vblank)
+			     __field(bool, async_flip)
+			     __field(bool, vrr_enabled)
+			     __field(bool, self_refresh_active)
+			     __field(u32, plane_mask)
+			     __field(u32, connector_mask)
+			     __field(u32, encoder_mask)
+	    ),
+
+	    TP_fast_assign(
+			   __entry->state = state->state;
+			   __entry->crtc_state = state;
+			   __entry->crtc_id = state->crtc->base.id;
+			   __entry->commit = state->commit;
+			   __entry->enable = state->enable;
+			   __entry->active = state->active;
+			   __entry->planes_changed = state->planes_changed;
+			   __entry->mode_changed = state->mode_changed;
+			   __entry->active_changed = state->active_changed;
+			   __entry->connectors_changed = state->connectors_changed;
+			   __entry->zpos_changed = state->zpos_changed;
+			   __entry->color_mgmt_changed = state->color_mgmt_changed;
+			   __entry->no_vblank = state->no_vblank;
+			   __entry->async_flip = state->async_flip;
+			   __entry->vrr_enabled = state->vrr_enabled;
+			   __entry->self_refresh_active = state->self_refresh_active;
+			   __entry->plane_mask = state->plane_mask;
+			   __entry->connector_mask = state->connector_mask;
+			   __entry->encoder_mask = state->encoder_mask;
+	    ),
+
+	    TP_printk("crtc_id=%u crtc_state=%p state=%p commit=%p changed("
+		      "planes=%d mode=%d active=%d conn=%d zpos=%d color_mgmt=%d) "
+		      "state(enable=%d active=%d async_flip=%d vrr_enabled=%d "
+		      "self_refresh_active=%d no_vblank=%d) mask(plane=%x conn=%x "
+		      "enc=%x)",
+		      __entry->crtc_id, __entry->crtc_state, __entry->state,
+		      __entry->commit, __entry->planes_changed,
+		      __entry->mode_changed, __entry->active_changed,
+		      __entry->connectors_changed, __entry->zpos_changed,
+		      __entry->color_mgmt_changed, __entry->enable, __entry->active,
+		      __entry->async_flip, __entry->vrr_enabled,
+		      __entry->self_refresh_active, __entry->no_vblank,
+		      __entry->plane_mask, __entry->connector_mask,
+		      __entry->encoder_mask)
+);
+
+DECLARE_EVENT_CLASS(amdgpu_dm_plane_state_template,
+	    TP_PROTO(const struct drm_plane_state *state),
+	    TP_ARGS(state),
+	    TP_STRUCT__entry(
+			     __field(uint32_t, plane_id)
+			     __field(enum drm_plane_type, plane_type)
+			     __field(const struct drm_plane_state *, plane_state)
+			     __field(const struct drm_atomic_state *, state)
+			     __field(uint32_t, crtc_id)
+			     __field(uint32_t, fb_id)
+			     __field(uint32_t, fb_format)
+			     __field(uint8_t, fb_planes)
+			     __field(uint64_t, fb_modifier)
+			     __field(const struct dma_fence *, fence)
+			     __field(int32_t, crtc_x)
+			     __field(int32_t, crtc_y)
+			     __field(uint32_t, crtc_w)
+			     __field(uint32_t, crtc_h)
+			     __field(uint32_t, src_x)
+			     __field(uint32_t, src_y)
+			     __field(uint32_t, src_w)
+			     __field(uint32_t, src_h)
+			     __field(u32, alpha)
+			     __field(uint32_t, pixel_blend_mode)
+			     __field(unsigned int, rotation)
+			     __field(unsigned int, zpos)
+			     __field(unsigned int, normalized_zpos)
+			     __field(enum drm_color_encoding, color_encoding)
+			     __field(enum drm_color_range, color_range)
+			     __field(bool, visible)
+	    ),
+
+	    TP_fast_assign(
+			   __entry->plane_id = state->plane->base.id;
+			   __entry->plane_type = state->plane->type;
+			   __entry->plane_state = state;
+			   __entry->state = state->state;
+			   __entry->crtc_id = state->crtc ? state->crtc->base.id : 0;
+			   __entry->fb_id = state->fb ? state->fb->base.id : 0;
+			   __entry->fb_format = state->fb ? state->fb->format->format : 0;
+			   __entry->fb_planes = state->fb ? state->fb->format->num_planes : 0;
+			   __entry->fb_modifier = state->fb ? state->fb->modifier : 0;
+			   __entry->fence = state->fence;
+			   __entry->crtc_x = state->crtc_x;
+			   __entry->crtc_y = state->crtc_y;
+			   __entry->crtc_w = state->crtc_w;
+			   __entry->crtc_h = state->crtc_h;
+			   __entry->src_x = state->src_x >> 16;
+			   __entry->src_y = state->src_y >> 16;
+			   __entry->src_w = state->src_w >> 16;
+			   __entry->src_h = state->src_h >> 16;
+			   __entry->alpha = state->alpha;
+			   __entry->pixel_blend_mode = state->pixel_blend_mode;
+			   __entry->rotation = state->rotation;
+			   __entry->zpos = state->zpos;
+			   __entry->normalized_zpos = state->normalized_zpos;
+			   __entry->color_encoding = state->color_encoding;
+			   __entry->color_range = state->color_range;
+			   __entry->visible = state->visible;
+	    ),
+
+	    TP_printk("plane_id=%u plane_type=%d plane_state=%p state=%p "
+		      "crtc_id=%u fb(id=%u fmt=%c%c%c%c planes=%u mod=%llu) "
+		      "fence=%p crtc_x=%d crtc_y=%d crtc_w=%u crtc_h=%u "
+		      "src_x=%u src_y=%u src_w=%u src_h=%u alpha=%u "
+		      "pixel_blend_mode=%u rotation=%u zpos=%u "
+		      "normalized_zpos=%u color_encoding=%d color_range=%d "
+		      "visible=%d",
+		      __entry->plane_id, __entry->plane_type, __entry->plane_state,
+		      __entry->state, __entry->crtc_id, __entry->fb_id,
+		      (__entry->fb_format & 0xff) ? (__entry->fb_format & 0xff) : 'N',
+		      ((__entry->fb_format >> 8) & 0xff) ? ((__entry->fb_format >> 8) & 0xff) : 'O',
+		      ((__entry->fb_format >> 16) & 0xff) ? ((__entry->fb_format >> 16) & 0xff) : 'N',
+		      ((__entry->fb_format >> 24) & 0x7f) ? ((__entry->fb_format >> 24) & 0x7f) : 'E',
+		      __entry->fb_planes,
+		      __entry->fb_modifier, __entry->fence, __entry->crtc_x,
+		      __entry->crtc_y, __entry->crtc_w, __entry->crtc_h,
+		      __entry->src_x, __entry->src_y, __entry->src_w, __entry->src_h,
+		      __entry->alpha, __entry->pixel_blend_mode, __entry->rotation,
+		      __entry->zpos, __entry->normalized_zpos,
+		      __entry->color_encoding, __entry->color_range,
+		      __entry->visible)
+);
+
+DEFINE_EVENT(amdgpu_dm_plane_state_template, amdgpu_dm_plane_atomic_check,
+	     TP_PROTO(const struct drm_plane_state *state),
+	     TP_ARGS(state));
+
+DEFINE_EVENT(amdgpu_dm_plane_state_template, amdgpu_dm_atomic_update_cursor,
+	     TP_PROTO(const struct drm_plane_state *state),
+	     TP_ARGS(state));
+
+TRACE_EVENT(amdgpu_dm_atomic_state_template,
+	    TP_PROTO(const struct drm_atomic_state *state),
+	    TP_ARGS(state),
+
+	    TP_STRUCT__entry(
+			     __field(const struct drm_atomic_state *, state)
+			     __field(bool, allow_modeset)
+			     __field(bool, legacy_cursor_update)
+			     __field(bool, async_update)
+			     __field(bool, duplicated)
+			     __field(int, num_connector)
+			     __field(int, num_private_objs)
+	    ),
+
+	    TP_fast_assign(
+			   __entry->state = state;
+			   __entry->allow_modeset = state->allow_modeset;
+			   __entry->legacy_cursor_update = state->legacy_cursor_update;
+			   __entry->async_update = state->async_update;
+			   __entry->duplicated = state->duplicated;
+			   __entry->num_connector = state->num_connector;
+			   __entry->num_private_objs = state->num_private_objs;
+	    ),
+
+	    TP_printk("state=%p allow_modeset=%d legacy_cursor_update=%d "
+		      "async_update=%d duplicated=%d num_connector=%d "
+		      "num_private_objs=%d",
+		      __entry->state, __entry->allow_modeset, __entry->legacy_cursor_update,
+		      __entry->async_update, __entry->duplicated, __entry->num_connector,
+		      __entry->num_private_objs)
+);
+
+DEFINE_EVENT(amdgpu_dm_atomic_state_template, amdgpu_dm_atomic_commit_tail_begin,
+	     TP_PROTO(const struct drm_atomic_state *state),
+	     TP_ARGS(state));
+
+DEFINE_EVENT(amdgpu_dm_atomic_state_template, amdgpu_dm_atomic_commit_tail_finish,
+	     TP_PROTO(const struct drm_atomic_state *state),
+	     TP_ARGS(state));
+
+DEFINE_EVENT(amdgpu_dm_atomic_state_template, amdgpu_dm_atomic_check_begin,
+	     TP_PROTO(const struct drm_atomic_state *state),
+	     TP_ARGS(state));
+
+TRACE_EVENT(amdgpu_dm_atomic_check_finish,
+	    TP_PROTO(const struct drm_atomic_state *state, int res),
+	    TP_ARGS(state, res),
+
+	    TP_STRUCT__entry(
+			     __field(const struct drm_atomic_state *, state)
+			     __field(int, res)
+			     __field(bool, async_update)
+			     __field(bool, allow_modeset)
+	    ),
+
+	    TP_fast_assign(
+			   __entry->state = state;
+			   __entry->res = res;
+			   __entry->async_update = state->async_update;
+			   __entry->allow_modeset = state->allow_modeset;
+	    ),
+
+	    TP_printk("state=%p res=%d async_update=%d allow_modeset=%d",
+		      __entry->state, __entry->res,
+		      __entry->async_update, __entry->allow_modeset)
+);
+
+TRACE_EVENT(amdgpu_dm_dc_pipe_state,
+	    TP_PROTO(int pipe_idx, const struct dc_plane_state *plane_state,
+		     const struct dc_stream_state *stream,
+		     const struct plane_resource *plane_res,
+		     int update_flags),
+	    TP_ARGS(pipe_idx, plane_state, stream, plane_res, update_flags),
+
+	    TP_STRUCT__entry(
+			     __field(int, pipe_idx)
+			     __field(const void *, stream)
+			     __field(int, stream_w)
+			     __field(int, stream_h)
+			     __field(int, dst_x)
+			     __field(int, dst_y)
+			     __field(int, dst_w)
+			     __field(int, dst_h)
+			     __field(int, src_x)
+			     __field(int, src_y)
+			     __field(int, src_w)
+			     __field(int, src_h)
+			     __field(int, clip_x)
+			     __field(int, clip_y)
+			     __field(int, clip_w)
+			     __field(int, clip_h)
+			     __field(int, recout_x)
+			     __field(int, recout_y)
+			     __field(int, recout_w)
+			     __field(int, recout_h)
+			     __field(int, viewport_x)
+			     __field(int, viewport_y)
+			     __field(int, viewport_w)
+			     __field(int, viewport_h)
+			     __field(int, flip_immediate)
+			     __field(int, surface_pitch)
+			     __field(int, format)
+			     __field(int, swizzle)
+			     __field(unsigned int, update_flags)
+	),
+
+	TP_fast_assign(
+		       __entry->pipe_idx = pipe_idx;
+		       __entry->stream = stream;
+		       __entry->stream_w = stream->timing.h_addressable;
+		       __entry->stream_h = stream->timing.v_addressable;
+		       __entry->dst_x = plane_state->dst_rect.x;
+		       __entry->dst_y = plane_state->dst_rect.y;
+		       __entry->dst_w = plane_state->dst_rect.width;
+		       __entry->dst_h = plane_state->dst_rect.height;
+		       __entry->src_x = plane_state->src_rect.x;
+		       __entry->src_y = plane_state->src_rect.y;
+		       __entry->src_w = plane_state->src_rect.width;
+		       __entry->src_h = plane_state->src_rect.height;
+		       __entry->clip_x = plane_state->clip_rect.x;
+		       __entry->clip_y = plane_state->clip_rect.y;
+		       __entry->clip_w = plane_state->clip_rect.width;
+		       __entry->clip_h = plane_state->clip_rect.height;
+		       __entry->recout_x = plane_res->scl_data.recout.x;
+		       __entry->recout_y = plane_res->scl_data.recout.y;
+		       __entry->recout_w = plane_res->scl_data.recout.width;
+		       __entry->recout_h = plane_res->scl_data.recout.height;
+		       __entry->viewport_x = plane_res->scl_data.viewport.x;
+		       __entry->viewport_y = plane_res->scl_data.viewport.y;
+		       __entry->viewport_w = plane_res->scl_data.viewport.width;
+		       __entry->viewport_h = plane_res->scl_data.viewport.height;
+		       __entry->flip_immediate = plane_state->flip_immediate;
+		       __entry->surface_pitch = plane_state->plane_size.surface_pitch;
+		       __entry->format = plane_state->format;
+		       __entry->swizzle = plane_state->tiling_info.gfx9.swizzle;
+		       __entry->update_flags = update_flags;
+	),
+	TP_printk("pipe_idx=%d stream=%p rct(%d,%d) dst=(%d,%d,%d,%d) "
+		  "src=(%d,%d,%d,%d) clip=(%d,%d,%d,%d) recout=(%d,%d,%d,%d) "
+		  "viewport=(%d,%d,%d,%d) flip_immediate=%d pitch=%d "
+		  "format=%d swizzle=%d update_flags=%x",
+		  __entry->pipe_idx,
+		  __entry->stream,
+		  __entry->stream_w,
+		  __entry->stream_h,
+		  __entry->dst_x,
+		  __entry->dst_y,
+		  __entry->dst_w,
+		  __entry->dst_h,
+		  __entry->src_x,
+		  __entry->src_y,
+		  __entry->src_w,
+		  __entry->src_h,
+		  __entry->clip_x,
+		  __entry->clip_y,
+		  __entry->clip_w,
+		  __entry->clip_h,
+		  __entry->recout_x,
+		  __entry->recout_y,
+		  __entry->recout_w,
+		  __entry->recout_h,
+		  __entry->viewport_x,
+		  __entry->viewport_y,
+		  __entry->viewport_w,
+		  __entry->viewport_h,
+		  __entry->flip_immediate,
+		  __entry->surface_pitch,
+		  __entry->format,
+		  __entry->swizzle,
+		  __entry->update_flags
+	)
+);
+
+TRACE_EVENT(amdgpu_dm_dc_clocks_state,
+	    TP_PROTO(const struct dc_clocks *clk),
+	    TP_ARGS(clk),
+
+	    TP_STRUCT__entry(
+			     __field(int, dispclk_khz)
+			     __field(int, dppclk_khz)
+			     __field(int, disp_dpp_voltage_level_khz)
+			     __field(int, dcfclk_khz)
+			     __field(int, socclk_khz)
+			     __field(int, dcfclk_deep_sleep_khz)
+			     __field(int, fclk_khz)
+			     __field(int, phyclk_khz)
+			     __field(int, dramclk_khz)
+			     __field(int, p_state_change_support)
+			     __field(int, prev_p_state_change_support)
+			     __field(int, pwr_state)
+			     __field(int, dtm_level)
+			     __field(int, max_supported_dppclk_khz)
+			     __field(int, max_supported_dispclk_khz)
+			     __field(int, bw_dppclk_khz)
+			     __field(int, bw_dispclk_khz)
+	    ),
+	    TP_fast_assign(
+			   __entry->dispclk_khz = clk->dispclk_khz;
+			   __entry->dppclk_khz = clk->dppclk_khz;
+			   __entry->dcfclk_khz = clk->dcfclk_khz;
+			   __entry->socclk_khz = clk->socclk_khz;
+			   __entry->dcfclk_deep_sleep_khz = clk->dcfclk_deep_sleep_khz;
+			   __entry->fclk_khz = clk->fclk_khz;
+			   __entry->phyclk_khz = clk->phyclk_khz;
+			   __entry->dramclk_khz = clk->dramclk_khz;
+			   __entry->p_state_change_support = clk->p_state_change_support;
+			   __entry->prev_p_state_change_support = clk->prev_p_state_change_support;
+			   __entry->pwr_state = clk->pwr_state;
+			   __entry->prev_p_state_change_support = clk->prev_p_state_change_support;
+			   __entry->dtm_level = clk->dtm_level;
+			   __entry->max_supported_dppclk_khz = clk->max_supported_dppclk_khz;
+			   __entry->max_supported_dispclk_khz = clk->max_supported_dispclk_khz;
+			   __entry->bw_dppclk_khz = clk->bw_dppclk_khz;
+			   __entry->bw_dispclk_khz = clk->bw_dispclk_khz;
+	    ),
+	    TP_printk("dispclk_khz=%d dppclk_khz=%d disp_dpp_voltage_level_khz=%d dcfclk_khz=%d socclk_khz=%d "
+		      "dcfclk_deep_sleep_khz=%d fclk_khz=%d phyclk_khz=%d "
+		      "dramclk_khz=%d p_state_change_support=%d "
+		      "prev_p_state_change_support=%d pwr_state=%d prev_p_state_change_support=%d "
+		      "dtm_level=%d max_supported_dppclk_khz=%d max_supported_dispclk_khz=%d "
+		      "bw_dppclk_khz=%d bw_dispclk_khz=%d ",
+		      __entry->dispclk_khz,
+		      __entry->dppclk_khz,
+		      __entry->disp_dpp_voltage_level_khz,
+		      __entry->dcfclk_khz,
+		      __entry->socclk_khz,
+		      __entry->dcfclk_deep_sleep_khz,
+		      __entry->fclk_khz,
+		      __entry->phyclk_khz,
+		      __entry->dramclk_khz,
+		      __entry->p_state_change_support,
+		      __entry->prev_p_state_change_support,
+		      __entry->pwr_state,
+		      __entry->prev_p_state_change_support,
+		      __entry->dtm_level,
+		      __entry->max_supported_dppclk_khz,
+		      __entry->max_supported_dispclk_khz,
+		      __entry->bw_dppclk_khz,
+		      __entry->bw_dispclk_khz
+	    )
+);
+
+TRACE_EVENT(amdgpu_dm_dce_clocks_state,
+	    TP_PROTO(const struct dce_bw_output *clk),
+	    TP_ARGS(clk),
+
+	    TP_STRUCT__entry(
+			     __field(bool, cpuc_state_change_enable)
+			     __field(bool, cpup_state_change_enable)
+			     __field(bool, stutter_mode_enable)
+			     __field(bool, nbp_state_change_enable)
+			     __field(bool, all_displays_in_sync)
+			     __field(int, sclk_khz)
+			     __field(int, sclk_deep_sleep_khz)
+			     __field(int, yclk_khz)
+			     __field(int, dispclk_khz)
+			     __field(int, blackout_recovery_time_us)
+	    ),
+	    TP_fast_assign(
+			   __entry->cpuc_state_change_enable = clk->cpuc_state_change_enable;
+			   __entry->cpup_state_change_enable = clk->cpup_state_change_enable;
+			   __entry->stutter_mode_enable = clk->stutter_mode_enable;
+			   __entry->nbp_state_change_enable = clk->nbp_state_change_enable;
+			   __entry->all_displays_in_sync = clk->all_displays_in_sync;
+			   __entry->sclk_khz = clk->sclk_khz;
+			   __entry->sclk_deep_sleep_khz = clk->sclk_deep_sleep_khz;
+			   __entry->yclk_khz = clk->yclk_khz;
+			   __entry->dispclk_khz = clk->dispclk_khz;
+			   __entry->blackout_recovery_time_us = clk->blackout_recovery_time_us;
+	    ),
+	    TP_printk("cpuc_state_change_enable=%d cpup_state_change_enable=%d stutter_mode_enable=%d "
+		      "nbp_state_change_enable=%d all_displays_in_sync=%d sclk_khz=%d sclk_deep_sleep_khz=%d "
+		      "yclk_khz=%d dispclk_khz=%d blackout_recovery_time_us=%d",
+		      __entry->cpuc_state_change_enable,
+		      __entry->cpup_state_change_enable,
+		      __entry->stutter_mode_enable,
+		      __entry->nbp_state_change_enable,
+		      __entry->all_displays_in_sync,
+		      __entry->sclk_khz,
+		      __entry->sclk_deep_sleep_khz,
+		      __entry->yclk_khz,
+		      __entry->dispclk_khz,
+		      __entry->blackout_recovery_time_us
+	    )
+);
+
+TRACE_EVENT(amdgpu_dmub_trace_high_irq,
+	TP_PROTO(uint32_t trace_code, uint32_t tick_count, uint32_t param0,
+		 uint32_t param1),
+	TP_ARGS(trace_code, tick_count, param0, param1),
+	TP_STRUCT__entry(
+		__field(uint32_t, trace_code)
+		__field(uint32_t, tick_count)
+		__field(uint32_t, param0)
+		__field(uint32_t, param1)
+		),
+	TP_fast_assign(
+		__entry->trace_code = trace_code;
+		__entry->tick_count = tick_count;
+		__entry->param0 = param0;
+		__entry->param1 = param1;
+	),
+	TP_printk("trace_code=%u tick_count=%u param0=%u param1=%u",
+		  __entry->trace_code, __entry->tick_count,
+		  __entry->param0, __entry->param1)
+);
+
+TRACE_EVENT(amdgpu_refresh_rate_track,
+	TP_PROTO(int crtc_index, ktime_t refresh_rate_ns, uint32_t refresh_rate_hz),
+	TP_ARGS(crtc_index, refresh_rate_ns, refresh_rate_hz),
+	TP_STRUCT__entry(
+		__field(int, crtc_index)
+		__field(ktime_t, refresh_rate_ns)
+		__field(uint32_t, refresh_rate_hz)
+		),
+	TP_fast_assign(
+		__entry->crtc_index = crtc_index;
+		__entry->refresh_rate_ns = refresh_rate_ns;
+		__entry->refresh_rate_hz = refresh_rate_hz;
+	),
+	TP_printk("crtc_index=%d refresh_rate=%dHz (%lld)",
+		  __entry->crtc_index,
+		  __entry->refresh_rate_hz,
+		  __entry->refresh_rate_ns)
+);
+
+TRACE_EVENT(dcn_fpu,
+	    TP_PROTO(bool begin, const char *function, const int line, const int recursion_depth),
+	    TP_ARGS(begin, function, line, recursion_depth),
+
+	    TP_STRUCT__entry(
+			     __field(bool, begin)
+			     __field(const char *, function)
+			     __field(int, line)
+			     __field(int, recursion_depth)
+	    ),
+	    TP_fast_assign(
+			   __entry->begin = begin;
+			   __entry->function = function;
+			   __entry->line = line;
+			   __entry->recursion_depth = recursion_depth;
+	    ),
+	    TP_printk("%s: recursion_depth: %d: %s()+%d:",
+		      __entry->begin ? "begin" : "end",
+		      __entry->recursion_depth,
+		      __entry->function,
+		      __entry->line
+	    )
+);
+
+TRACE_EVENT(dcn_optc_lock_unlock_state,
+	    TP_PROTO(const struct optc *optc_state, int instance, bool lock, const char *function, const int line),
+	    TP_ARGS(optc_state, instance, lock, function, line),
+
+	    TP_STRUCT__entry(
+			     __field(const char *, function)
+			     __field(int, instance)
+			     __field(bool, lock)
+			     __field(int, line)
+			     __field(int, opp_count)
+			     __field(int, max_h_total)
+			     __field(int, max_v_total)
+			     __field(int, min_h_blank)
+			     __field(int, min_h_sync_width)
+			     __field(int, min_v_sync_width)
+			     __field(int, min_v_blank)
+			     __field(int, min_v_blank_interlace)
+			     __field(int, vstartup_start)
+			     __field(int, vupdate_offset)
+			     __field(int, vupdate_width)
+			     __field(int, vready_offset)
+	    ),
+	    TP_fast_assign(
+			   __entry->function = function;
+			   __entry->instance = instance;
+			   __entry->lock = lock;
+			   __entry->line = line;
+			   __entry->opp_count = optc_state->opp_count;
+			   __entry->max_h_total = optc_state->max_h_total;
+			   __entry->max_v_total = optc_state->max_v_total;
+			   __entry->min_h_blank = optc_state->min_h_blank;
+			   __entry->min_h_sync_width = optc_state->min_h_sync_width;
+			   __entry->min_v_sync_width = optc_state->min_v_sync_width;
+			   __entry->min_v_blank = optc_state->min_v_blank;
+			   __entry->min_v_blank_interlace = optc_state->min_v_blank_interlace;
+			   __entry->vstartup_start = optc_state->vstartup_start;
+			   __entry->vupdate_offset = optc_state->vupdate_offset;
+			   __entry->vupdate_width = optc_state->vupdate_width;
+			   __entry->vready_offset = optc_state->vupdate_offset;
+	    ),
+	    TP_printk("%s: %s()+%d: optc_instance=%d opp_count=%d max_h_total=%d max_v_total=%d "
+		      "min_h_blank=%d min_h_sync_width=%d min_v_sync_width=%d min_v_blank=%d "
+		      "min_v_blank_interlace=%d vstartup_start=%d vupdate_offset=%d vupdate_width=%d "
+		      "vready_offset=%d",
+		      __entry->lock ? "Lock" : "Unlock",
+		      __entry->function,
+		      __entry->line,
+		      __entry->instance,
+		      __entry->opp_count,
+		      __entry->max_h_total,
+		      __entry->max_v_total,
+		      __entry->min_h_blank,
+		      __entry->min_h_sync_width,
+		      __entry->min_v_sync_width,
+		      __entry->min_v_blank,
+		      __entry->min_v_blank_interlace,
+		      __entry->vstartup_start,
+		      __entry->vupdate_offset,
+		      __entry->vupdate_width,
+		      __entry->vready_offset
+	    )
+);
+
+TRACE_EVENT(amdgpu_dm_brightness,
+	TP_PROTO(void *function, u32 user_brightness, u32 converted_brightness, bool aux, bool ac),
+	TP_ARGS(function, user_brightness, converted_brightness, aux, ac),
+	TP_STRUCT__entry(
+		__field(void *, function)
+		__field(u32, user_brightness)
+		__field(u32, converted_brightness)
+		__field(bool, aux)
+		__field(bool, ac)
+	),
+	TP_fast_assign(
+		__entry->function = function;
+		__entry->user_brightness = user_brightness;
+		__entry->converted_brightness = converted_brightness;
+		__entry->aux = aux;
+		__entry->ac = ac;
+	),
+	TP_printk("%ps: brightness requested=%u converted=%u aux=%s power=%s",
+		  (void *)__entry->function,
+		  (u32)__entry->user_brightness,
+		  (u32)__entry->converted_brightness,
+		  (__entry->aux) ? "true" : "false",
+		  (__entry->ac) ? "AC" : "DC"
+	)
+);
+
+#endif /* _AMDGPU_DM_TRACE_H_ */
+
+#undef TRACE_INCLUDE_PATH
+#define TRACE_INCLUDE_PATH .
+#define TRACE_INCLUDE_FILE amdgpu_dm_trace
+#include <trace/define_trace.h>
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_wb.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_wb.c
new file mode 100644
index 000000000000..d9527c05fc87
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_wb.c
@@ -0,0 +1,215 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services_types.h"
+
+#include "amdgpu.h"
+#include "amdgpu_dm.h"
+#include "amdgpu_dm_wb.h"
+#include "amdgpu_display.h"
+#include "dc.h"
+
+#include <drm/drm_edid.h>
+#include <drm/drm_atomic_state_helper.h>
+#include <drm/drm_modeset_helper_vtables.h>
+
+static const u32 amdgpu_dm_wb_formats[] = {
+	DRM_FORMAT_XRGB2101010,
+};
+
+static int amdgpu_dm_wb_encoder_atomic_check(struct drm_encoder *encoder,
+					struct drm_crtc_state *crtc_state,
+					struct drm_connector_state *conn_state)
+{
+	struct drm_framebuffer *fb;
+	const struct drm_display_mode *mode = &crtc_state->mode;
+	bool found = false;
+	uint8_t i;
+
+	if (!conn_state->writeback_job || !conn_state->writeback_job->fb)
+		return 0;
+
+	fb = conn_state->writeback_job->fb;
+	if (fb->width != mode->hdisplay || fb->height != mode->vdisplay) {
+		DRM_DEBUG_KMS("Invalid framebuffer size %ux%u\n",
+			      fb->width, fb->height);
+		return -EINVAL;
+	}
+
+	for (i = 0; i < sizeof(amdgpu_dm_wb_formats) / sizeof(u32); i++) {
+		if (fb->format->format == amdgpu_dm_wb_formats[i])
+			found = true;
+	}
+
+	if (!found) {
+		DRM_DEBUG_KMS("Invalid pixel format %p4cc\n",
+			      &fb->format->format);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+
+static int amdgpu_dm_wb_connector_get_modes(struct drm_connector *connector)
+{
+	/* Maximum resolution supported by DWB */
+	return drm_add_modes_noedid(connector, 3840, 2160);
+}
+
+static int amdgpu_dm_wb_prepare_job(struct drm_writeback_connector *wb_connector,
+			       struct drm_writeback_job *job)
+{
+	struct amdgpu_framebuffer *afb;
+	struct drm_gem_object *obj;
+	struct amdgpu_device *adev;
+	struct amdgpu_bo *rbo;
+	uint32_t domain;
+	int r;
+
+	if (!job->fb) {
+		DRM_DEBUG_KMS("No FB bound\n");
+		return 0;
+	}
+
+	afb = to_amdgpu_framebuffer(job->fb);
+	obj = job->fb->obj[0];
+	rbo = gem_to_amdgpu_bo(obj);
+	adev = amdgpu_ttm_adev(rbo->tbo.bdev);
+
+	r = amdgpu_bo_reserve(rbo, true);
+	if (r) {
+		drm_err(adev_to_drm(adev), "fail to reserve bo (%d)\n", r);
+		return r;
+	}
+
+	r = dma_resv_reserve_fences(rbo->tbo.base.resv, 1);
+	if (r) {
+		drm_err(adev_to_drm(adev), "reserving fence slot failed (%d)\n", r);
+		goto error_unlock;
+	}
+
+	domain = amdgpu_display_supported_domains(adev, rbo->flags);
+
+	rbo->flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+	r = amdgpu_bo_pin(rbo, domain);
+	if (unlikely(r != 0)) {
+		if (r != -ERESTARTSYS)
+			DRM_ERROR("Failed to pin framebuffer with error %d\n", r);
+		goto error_unlock;
+	}
+
+	r = amdgpu_ttm_alloc_gart(&rbo->tbo);
+	if (unlikely(r != 0)) {
+		DRM_ERROR("%p bind failed\n", rbo);
+		goto error_unpin;
+	}
+
+	amdgpu_bo_unreserve(rbo);
+
+	afb->address = amdgpu_bo_gpu_offset(rbo);
+
+	amdgpu_bo_ref(rbo);
+
+	return 0;
+
+error_unpin:
+	amdgpu_bo_unpin(rbo);
+
+error_unlock:
+	amdgpu_bo_unreserve(rbo);
+	return r;
+}
+
+static void amdgpu_dm_wb_cleanup_job(struct drm_writeback_connector *connector,
+				struct drm_writeback_job *job)
+{
+	struct amdgpu_bo *rbo;
+	int r;
+
+	if (!job->fb)
+		return;
+
+	rbo = gem_to_amdgpu_bo(job->fb->obj[0]);
+	r = amdgpu_bo_reserve(rbo, false);
+	if (unlikely(r)) {
+		DRM_ERROR("failed to reserve rbo before unpin\n");
+		return;
+	}
+
+	amdgpu_bo_unpin(rbo);
+	amdgpu_bo_unreserve(rbo);
+	amdgpu_bo_unref(&rbo);
+}
+
+static const struct drm_encoder_helper_funcs amdgpu_dm_wb_encoder_helper_funcs = {
+	.atomic_check = amdgpu_dm_wb_encoder_atomic_check,
+};
+
+static const struct drm_connector_funcs amdgpu_dm_wb_connector_funcs = {
+	.fill_modes = drm_helper_probe_single_connector_modes,
+	.destroy = drm_connector_cleanup,
+	.reset = amdgpu_dm_connector_funcs_reset,
+	.atomic_duplicate_state = amdgpu_dm_connector_atomic_duplicate_state,
+	.atomic_destroy_state = drm_atomic_helper_connector_destroy_state,
+};
+
+static const struct drm_connector_helper_funcs amdgpu_dm_wb_conn_helper_funcs = {
+	.get_modes = amdgpu_dm_wb_connector_get_modes,
+	.prepare_writeback_job = amdgpu_dm_wb_prepare_job,
+	.cleanup_writeback_job = amdgpu_dm_wb_cleanup_job,
+};
+
+int amdgpu_dm_wb_connector_init(struct amdgpu_display_manager *dm,
+				struct amdgpu_dm_wb_connector *wbcon,
+				uint32_t link_index)
+{
+	struct dc *dc = dm->dc;
+	struct dc_link *link = dc_get_link_at_index(dc, link_index);
+	int res = 0;
+
+	wbcon->link = link;
+
+	drm_connector_helper_add(&wbcon->base.base, &amdgpu_dm_wb_conn_helper_funcs);
+
+	res = drm_writeback_connector_init(&dm->adev->ddev, &wbcon->base,
+					    &amdgpu_dm_wb_connector_funcs,
+					    &amdgpu_dm_wb_encoder_helper_funcs,
+					    amdgpu_dm_wb_formats,
+					    ARRAY_SIZE(amdgpu_dm_wb_formats),
+					    amdgpu_dm_get_encoder_crtc_mask(dm->adev));
+
+	if (res)
+		return res;
+	/*
+	 * Some of the properties below require access to state, like bpc.
+	 * Allocate some default initial connector state with our reset helper.
+	 */
+	if (wbcon->base.base.funcs->reset)
+		wbcon->base.base.funcs->reset(&wbcon->base.base);
+
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_wb.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_wb.h
new file mode 100644
index 000000000000..13d31c857dee
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_wb.h
@@ -0,0 +1,36 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __AMDGPU_DM_WB_H__
+#define __AMDGPU_DM_WB_H__
+
+#include <drm/drm_writeback.h>
+
+int amdgpu_dm_wb_connector_init(struct amdgpu_display_manager *dm,
+				struct amdgpu_dm_wb_connector *dm_wbcon,
+				uint32_t link_index);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/dc_fpu.c b/drivers/gpu/drm/amd/display/amdgpu_dm/dc_fpu.c
new file mode 100644
index 000000000000..e46f8ce41d87
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/dc_fpu.c
@@ -0,0 +1,112 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dc_trace.h"
+
+#include <linux/fpu.h>
+
+/**
+ * DOC: DC FPU manipulation overview
+ *
+ * DC core uses FPU operations in multiple parts of the code, which requires a
+ * more specialized way to manage these areas' entrance. To fulfill this
+ * requirement, we created some wrapper functions that encapsulate
+ * kernel_fpu_begin/end to better fit our need in the display component. In
+ * summary, in this file, you can find functions related to FPU operation
+ * management.
+ */
+
+static DEFINE_PER_CPU(int, fpu_recursion_depth);
+
+/**
+ * dc_assert_fp_enabled - Check if FPU protection is enabled
+ *
+ * This function tells if the code is already under FPU protection or not. A
+ * function that works as an API for a set of FPU operations can use this
+ * function for checking if the caller invoked it after DC_FP_START(). For
+ * example, take a look at dcn20_fpu.c file.
+ */
+inline void dc_assert_fp_enabled(void)
+{
+	int depth;
+
+	depth = __this_cpu_read(fpu_recursion_depth);
+
+	ASSERT(depth >= 1);
+}
+
+/**
+ * dc_fpu_begin - Enables FPU protection
+ * @function_name: A string containing the function name for debug purposes
+ *   (usually __func__)
+ *
+ * @line: A line number where DC_FP_START was invoked for debug purpose
+ *   (usually __LINE__)
+ *
+ * This function is responsible for managing the use of kernel_fpu_begin() with
+ * the advantage of providing an event trace for debugging.
+ *
+ * Note: Do not call this function directly; always use DC_FP_START().
+ */
+void dc_fpu_begin(const char *function_name, const int line)
+{
+	int depth;
+
+	WARN_ON_ONCE(!in_task());
+	preempt_disable();
+	depth = __this_cpu_inc_return(fpu_recursion_depth);
+	if (depth == 1) {
+		BUG_ON(!kernel_fpu_available());
+		kernel_fpu_begin();
+	}
+
+	TRACE_DCN_FPU(true, function_name, line, depth);
+}
+
+/**
+ * dc_fpu_end - Disable FPU protection
+ * @function_name: A string containing the function name for debug purposes
+ * @line: A-line number where DC_FP_END was invoked for debug purpose
+ *
+ * This function is responsible for managing the use of kernel_fpu_end() with
+ * the advantage of providing an event trace for debugging.
+ *
+ * Note: Do not call this function directly; always use DC_FP_END().
+ */
+void dc_fpu_end(const char *function_name, const int line)
+{
+	int depth;
+
+	depth = __this_cpu_dec_return(fpu_recursion_depth);
+	if (depth == 0) {
+		kernel_fpu_end();
+	} else {
+		WARN_ON_ONCE(depth < 0);
+	}
+
+	TRACE_DCN_FPU(false, function_name, line, depth);
+	preempt_enable();
+}
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/dc_fpu.h b/drivers/gpu/drm/amd/display/amdgpu_dm/dc_fpu.h
new file mode 100644
index 000000000000..b8275b397920
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/dc_fpu.h
@@ -0,0 +1,34 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_FPU_H__
+#define __DC_FPU_H__
+
+void dc_assert_fp_enabled(void);
+void dc_fpu_begin(const char *function_name, const int line);
+void dc_fpu_end(const char *function_name, const int line);
+
+#endif /* __DC_FPU_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/Makefile b/drivers/gpu/drm/amd/display/dc/Makefile
new file mode 100644
index 000000000000..dc943abd6dba
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/Makefile
@@ -0,0 +1,83 @@
+#
+# Copyright 2017 Advanced Micro Devices, Inc.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+#
+# Makefile for Display Core (dc) component.
+
+DC_LIBS = basics bios dml clk_mgr dce gpio hwss irq link virtual dsc resource optc dpp hubbub dccg hubp dio dwb hpo mmhubbub mpc opp pg
+
+ifdef CONFIG_DRM_AMD_DC_FP
+
+KCOV_INSTRUMENT := n
+
+DC_LIBS += dcn10
+DC_LIBS += dcn20
+DC_LIBS += dcn21
+DC_LIBS += dcn201
+DC_LIBS += dcn30
+DC_LIBS += dcn301
+DC_LIBS += dcn31
+DC_LIBS += dml
+DC_LIBS += dml2
+DC_LIBS += soc_and_ip_translator
+endif
+
+DC_LIBS += dce120
+
+DC_LIBS += dce112
+DC_LIBS += dce110
+DC_LIBS += dce80
+
+ifdef CONFIG_DRM_AMD_DC_SI
+DC_LIBS += dce60
+endif
+
+DC_LIBS += hdcp
+
+ifdef CONFIG_DRM_AMD_DC_FP
+DC_LIBS += sspl
+AMD_DISPLAY_FILES += $(addprefix $(AMDDALPATH)/dc/, dc_spl_translate.o)
+endif
+
+AMD_DC = $(addsuffix /Makefile, $(addprefix $(FULL_AMD_DISPLAY_PATH)/dc/,$(DC_LIBS)))
+
+include $(AMD_DC)
+
+FILES =
+FILES += dc_dmub_srv.o
+FILES += dc_edid_parser.o
+FILES += dc_fused_io.o
+FILES += dc_helper.o
+FILES += core/dc.o
+FILES += core/dc_debug.o
+FILES += core/dc_hw_sequencer.o
+FILES += core/dc_link_enc_cfg.o
+FILES += core/dc_link_exports.o
+FILES += core/dc_resource.o
+FILES += core/dc_sink.o
+FILES += core/dc_stat.o
+FILES += core/dc_state.o
+FILES += core/dc_stream.o
+FILES += core/dc_surface.o
+FILES += core/dc_vm_helper.o
+
+AMD_DISPLAY_FILES += $(addprefix $(AMDDALPATH)/dc/, $(FILES))
+
diff --git a/drivers/gpu/drm/amd/display/dc/basics/Makefile b/drivers/gpu/drm/amd/display/dc/basics/Makefile
new file mode 100644
index 000000000000..aabcebf69049
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/basics/Makefile
@@ -0,0 +1,38 @@
+#
+# Copyright 2017 Advanced Micro Devices, Inc.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+#
+# Makefile for the 'utils' sub-component of DAL.
+# It provides the general basic services required by other DAL
+# subcomponents.
+
+BASICS := \
+	conversion.o \
+	fixpt31_32.o \
+	vector.o \
+	dc_common.o \
+	dce_calcs.o \
+	custom_float.o \
+	bw_fixed.o
+
+AMD_DAL_BASICS = $(addprefix $(AMDDALPATH)/dc/basics/,$(BASICS))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_BASICS)
diff --git a/drivers/gpu/drm/amd/display/dc/basics/bw_fixed.c b/drivers/gpu/drm/amd/display/dc/basics/bw_fixed.c
new file mode 100644
index 000000000000..c8cb89e0d4d0
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/basics/bw_fixed.c
@@ -0,0 +1,188 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+#include "dm_services.h"
+#include "bw_fixed.h"
+
+#define MAX_I64 \
+	((int64_t)((1ULL << 63) - 1))
+
+#define MIN_I64 \
+	(-MAX_I64 - 1)
+
+#define FRACTIONAL_PART_MASK \
+	((1ULL << BW_FIXED_BITS_PER_FRACTIONAL_PART) - 1)
+
+#define GET_FRACTIONAL_PART(x) \
+	(FRACTIONAL_PART_MASK & (x))
+
+static uint64_t abs_i64(int64_t arg)
+{
+	if (arg >= 0)
+		return (uint64_t)(arg);
+	else
+		return (uint64_t)(-arg);
+}
+
+struct bw_fixed bw_int_to_fixed_nonconst(int64_t value)
+{
+	struct bw_fixed res;
+
+	ASSERT(value < BW_FIXED_MAX_I32 && value > BW_FIXED_MIN_I32);
+	res.value = value << BW_FIXED_BITS_PER_FRACTIONAL_PART;
+	return res;
+}
+
+struct bw_fixed bw_frc_to_fixed(int64_t numerator, int64_t denominator)
+{
+	struct bw_fixed res;
+	bool arg1_negative = numerator < 0;
+	bool arg2_negative = denominator < 0;
+	uint64_t arg1_value;
+	uint64_t arg2_value;
+	uint64_t remainder;
+
+	/* determine integer part */
+	uint64_t res_value;
+
+	ASSERT(denominator != 0);
+
+	arg1_value = abs_i64(numerator);
+	arg2_value = abs_i64(denominator);
+	res_value = div64_u64_rem(arg1_value, arg2_value, &remainder);
+
+	ASSERT(res_value <= BW_FIXED_MAX_I32);
+
+	/* determine fractional part */
+	{
+		uint32_t i = BW_FIXED_BITS_PER_FRACTIONAL_PART;
+
+		do {
+			remainder <<= 1;
+
+			res_value <<= 1;
+
+			if (remainder >= arg2_value) {
+				res_value |= 1;
+				remainder -= arg2_value;
+			}
+		} while (--i != 0);
+	}
+
+	/* round up LSB */
+	{
+		uint64_t summand = (remainder << 1) >= arg2_value;
+
+		ASSERT(res_value <= MAX_I64 - summand);
+
+		res_value += summand;
+	}
+
+	res.value = (int64_t)(res_value);
+
+	if (arg1_negative ^ arg2_negative)
+		res.value = -res.value;
+	return res;
+}
+
+struct bw_fixed bw_floor2(const struct bw_fixed arg,
+			  const struct bw_fixed significance)
+{
+	struct bw_fixed result;
+	int64_t multiplicand;
+
+	multiplicand = div64_s64(arg.value, abs_i64(significance.value));
+	result.value = abs_i64(significance.value) * multiplicand;
+	ASSERT(abs_i64(result.value) <= abs_i64(arg.value));
+	return result;
+}
+
+struct bw_fixed bw_ceil2(const struct bw_fixed arg,
+			 const struct bw_fixed significance)
+{
+	struct bw_fixed result;
+	int64_t multiplicand;
+
+	multiplicand = div64_s64(arg.value, abs_i64(significance.value));
+	result.value = abs_i64(significance.value) * multiplicand;
+	if (abs_i64(result.value) < abs_i64(arg.value)) {
+		if (arg.value < 0)
+			result.value -= abs_i64(significance.value);
+		else
+			result.value += abs_i64(significance.value);
+	}
+	return result;
+}
+
+struct bw_fixed bw_mul(const struct bw_fixed arg1, const struct bw_fixed arg2)
+{
+	struct bw_fixed res;
+
+	bool arg1_negative = arg1.value < 0;
+	bool arg2_negative = arg2.value < 0;
+
+	uint64_t arg1_value = abs_i64(arg1.value);
+	uint64_t arg2_value = abs_i64(arg2.value);
+
+	uint64_t arg1_int = BW_FIXED_GET_INTEGER_PART(arg1_value);
+	uint64_t arg2_int = BW_FIXED_GET_INTEGER_PART(arg2_value);
+
+	uint64_t arg1_fra = GET_FRACTIONAL_PART(arg1_value);
+	uint64_t arg2_fra = GET_FRACTIONAL_PART(arg2_value);
+
+	uint64_t tmp;
+
+	res.value = arg1_int * arg2_int;
+
+	ASSERT(res.value <= BW_FIXED_MAX_I32);
+
+	res.value <<= BW_FIXED_BITS_PER_FRACTIONAL_PART;
+
+	tmp = arg1_int * arg2_fra;
+
+	ASSERT(tmp <= (uint64_t)(MAX_I64 - res.value));
+
+	res.value += tmp;
+
+	tmp = arg2_int * arg1_fra;
+
+	ASSERT(tmp <= (uint64_t)(MAX_I64 - res.value));
+
+	res.value += tmp;
+
+	tmp = arg1_fra * arg2_fra;
+
+	tmp = (tmp >> BW_FIXED_BITS_PER_FRACTIONAL_PART) +
+		(tmp >= (uint64_t)(bw_frc_to_fixed(1, 2).value));
+
+	ASSERT(tmp <= (uint64_t)(MAX_I64 - res.value));
+
+	res.value += tmp;
+
+	if (arg1_negative ^ arg2_negative)
+		res.value = -res.value;
+	return res;
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/basics/calcs_logger.h b/drivers/gpu/drm/amd/display/dc/basics/calcs_logger.h
new file mode 100644
index 000000000000..62435bfc274d
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/basics/calcs_logger.h
@@ -0,0 +1,578 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef _CALCS_CALCS_LOGGER_H_
+#define _CALCS_CALCS_LOGGER_H_
+#define DC_LOGGER ctx->logger
+
+static void print_bw_calcs_dceip(struct dc_context *ctx, const struct bw_calcs_dceip *dceip)
+{
+
+	DC_LOG_BANDWIDTH_CALCS("#####################################################################");
+	DC_LOG_BANDWIDTH_CALCS("struct bw_calcs_dceip");
+	DC_LOG_BANDWIDTH_CALCS("#####################################################################");
+	DC_LOG_BANDWIDTH_CALCS("	[enum]   bw_calcs_version version %d", dceip->version);
+	DC_LOG_BANDWIDTH_CALCS("	[bool] large_cursor: %d", dceip->large_cursor);
+	DC_LOG_BANDWIDTH_CALCS("	[bool] dmif_pipe_en_fbc_chunk_tracker: %d", dceip->dmif_pipe_en_fbc_chunk_tracker);
+	DC_LOG_BANDWIDTH_CALCS("	[bool] display_write_back_supported: %d", dceip->display_write_back_supported);
+	DC_LOG_BANDWIDTH_CALCS("	[bool] argb_compression_support: %d", dceip->argb_compression_support);
+	DC_LOG_BANDWIDTH_CALCS("	[bool] pre_downscaler_enabled: %d", dceip->pre_downscaler_enabled);
+	DC_LOG_BANDWIDTH_CALCS("	[bool] underlay_downscale_prefetch_enabled: %d",
+				dceip->underlay_downscale_prefetch_enabled);
+	DC_LOG_BANDWIDTH_CALCS("	[bool] graphics_lb_nodownscaling_multi_line_prefetching: %d",
+				dceip->graphics_lb_nodownscaling_multi_line_prefetching);
+	DC_LOG_BANDWIDTH_CALCS("	[bool] limit_excessive_outstanding_dmif_requests: %d",
+				dceip->limit_excessive_outstanding_dmif_requests);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] cursor_max_outstanding_group_num: %d",
+				dceip->cursor_max_outstanding_group_num);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] lines_interleaved_into_lb: %d", dceip->lines_interleaved_into_lb);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] low_power_tiling_mode: %d", dceip->low_power_tiling_mode);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] chunk_width: %d", dceip->chunk_width);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] number_of_graphics_pipes: %d", dceip->number_of_graphics_pipes);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] number_of_underlay_pipes: %d", dceip->number_of_underlay_pipes);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] max_dmif_buffer_allocated: %d", dceip->max_dmif_buffer_allocated);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] graphics_dmif_size: %d", dceip->graphics_dmif_size);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] underlay_luma_dmif_size: %d", dceip->underlay_luma_dmif_size);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] underlay_chroma_dmif_size: %d", dceip->underlay_chroma_dmif_size);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] scatter_gather_lines_of_pte_prefetching_in_linear_mode: %d",
+				dceip->scatter_gather_lines_of_pte_prefetching_in_linear_mode);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] display_write_back420_luma_mcifwr_buffer_size: %d",
+				dceip->display_write_back420_luma_mcifwr_buffer_size);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] display_write_back420_chroma_mcifwr_buffer_size: %d",
+				dceip->display_write_back420_chroma_mcifwr_buffer_size);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] scatter_gather_pte_request_rows_in_tiling_mode: %d",
+				dceip->scatter_gather_pte_request_rows_in_tiling_mode);
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] underlay_vscaler_efficiency10_bit_per_component: %d",
+				bw_fixed_to_int(dceip->underlay_vscaler_efficiency10_bit_per_component));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] underlay_vscaler_efficiency12_bit_per_component: %d",
+				bw_fixed_to_int(dceip->underlay_vscaler_efficiency12_bit_per_component));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] graphics_vscaler_efficiency6_bit_per_component: %d",
+				bw_fixed_to_int(dceip->graphics_vscaler_efficiency6_bit_per_component));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] graphics_vscaler_efficiency8_bit_per_component: %d",
+				bw_fixed_to_int(dceip->graphics_vscaler_efficiency8_bit_per_component));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] graphics_vscaler_efficiency10_bit_per_component: %d",
+				bw_fixed_to_int(dceip->graphics_vscaler_efficiency10_bit_per_component));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] graphics_vscaler_efficiency12_bit_per_component: %d",
+				bw_fixed_to_int(dceip->graphics_vscaler_efficiency12_bit_per_component));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] alpha_vscaler_efficiency: %d",
+				bw_fixed_to_int(dceip->alpha_vscaler_efficiency));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] lb_write_pixels_per_dispclk: %d",
+				bw_fixed_to_int(dceip->lb_write_pixels_per_dispclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] lb_size_per_component444: %d",
+				bw_fixed_to_int(dceip->lb_size_per_component444));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] stutter_and_dram_clock_state_change_gated_before_cursor: %d",
+				bw_fixed_to_int(dceip->stutter_and_dram_clock_state_change_gated_before_cursor));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] underlay420_luma_lb_size_per_component: %d",
+				bw_fixed_to_int(dceip->underlay420_luma_lb_size_per_component));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] underlay420_chroma_lb_size_per_component: %d",
+				bw_fixed_to_int(dceip->underlay420_chroma_lb_size_per_component));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] underlay422_lb_size_per_component: %d",
+				bw_fixed_to_int(dceip->underlay422_lb_size_per_component));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] cursor_chunk_width: %d", bw_fixed_to_int(dceip->cursor_chunk_width));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] cursor_dcp_buffer_lines: %d",
+				bw_fixed_to_int(dceip->cursor_dcp_buffer_lines));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] underlay_maximum_width_efficient_for_tiling: %d",
+				bw_fixed_to_int(dceip->underlay_maximum_width_efficient_for_tiling));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] underlay_maximum_height_efficient_for_tiling: %d",
+				bw_fixed_to_int(dceip->underlay_maximum_height_efficient_for_tiling));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] peak_pte_request_to_eviction_ratio_limiting_multiple_displays_or_single_rotated_display: %d",
+				bw_fixed_to_int(dceip->peak_pte_request_to_eviction_ratio_limiting_multiple_displays_or_single_rotated_display));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] peak_pte_request_to_eviction_ratio_limiting_single_display_no_rotation: %d",
+				bw_fixed_to_int(dceip->peak_pte_request_to_eviction_ratio_limiting_single_display_no_rotation));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] minimum_outstanding_pte_request_limit: %d",
+				bw_fixed_to_int(dceip->minimum_outstanding_pte_request_limit));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] maximum_total_outstanding_pte_requests_allowed_by_saw: %d",
+				bw_fixed_to_int(dceip->maximum_total_outstanding_pte_requests_allowed_by_saw));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] linear_mode_line_request_alternation_slice: %d",
+				bw_fixed_to_int(dceip->linear_mode_line_request_alternation_slice));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] request_efficiency: %d", bw_fixed_to_int(dceip->request_efficiency));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dispclk_per_request: %d", bw_fixed_to_int(dceip->dispclk_per_request));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dispclk_ramping_factor: %d",
+				bw_fixed_to_int(dceip->dispclk_ramping_factor));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] display_pipe_throughput_factor: %d",
+				bw_fixed_to_int(dceip->display_pipe_throughput_factor));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] mcifwr_all_surfaces_burst_time: %d",
+				bw_fixed_to_int(dceip->mcifwr_all_surfaces_burst_time));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dmif_request_buffer_size: %d",
+				bw_fixed_to_int(dceip->dmif_request_buffer_size));
+
+
+}
+
+static void print_bw_calcs_vbios(struct dc_context *ctx, const struct bw_calcs_vbios *vbios)
+{
+
+	DC_LOG_BANDWIDTH_CALCS("#####################################################################");
+	DC_LOG_BANDWIDTH_CALCS("struct bw_calcs_vbios vbios");
+	DC_LOG_BANDWIDTH_CALCS("#####################################################################");
+	DC_LOG_BANDWIDTH_CALCS("	[enum] bw_defines memory_type: %d", vbios->memory_type);
+	DC_LOG_BANDWIDTH_CALCS("	[enum] bw_defines memory_type: %d", vbios->memory_type);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] dram_channel_width_in_bits: %d", vbios->dram_channel_width_in_bits);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] number_of_dram_channels: %d", vbios->number_of_dram_channels);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] number_of_dram_banks: %d", vbios->number_of_dram_banks);
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] low_yclk: %d", bw_fixed_to_int(vbios->low_yclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] mid_yclk: %d", bw_fixed_to_int(vbios->mid_yclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] high_yclk: %d", bw_fixed_to_int(vbios->high_yclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] low_sclk: %d", bw_fixed_to_int(vbios->low_sclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] mid1_sclk: %d", bw_fixed_to_int(vbios->mid1_sclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] mid2_sclk: %d", bw_fixed_to_int(vbios->mid2_sclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] mid3_sclk: %d", bw_fixed_to_int(vbios->mid3_sclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] mid4_sclk: %d", bw_fixed_to_int(vbios->mid4_sclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] mid5_sclk: %d", bw_fixed_to_int(vbios->mid5_sclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] mid6_sclk: %d", bw_fixed_to_int(vbios->mid6_sclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] high_sclk: %d", bw_fixed_to_int(vbios->high_sclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] low_voltage_max_dispclk: %d",
+				bw_fixed_to_int(vbios->low_voltage_max_dispclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] mid_voltage_max_dispclk;: %d",
+				bw_fixed_to_int(vbios->mid_voltage_max_dispclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] high_voltage_max_dispclk;: %d",
+				bw_fixed_to_int(vbios->high_voltage_max_dispclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] low_voltage_max_phyclk: %d",
+				bw_fixed_to_int(vbios->low_voltage_max_phyclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] mid_voltage_max_phyclk: %d",
+				bw_fixed_to_int(vbios->mid_voltage_max_phyclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] high_voltage_max_phyclk: %d",
+				bw_fixed_to_int(vbios->high_voltage_max_phyclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] data_return_bus_width: %d", bw_fixed_to_int(vbios->data_return_bus_width));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] trc: %d", bw_fixed_to_int(vbios->trc));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dmifmc_urgent_latency: %d", bw_fixed_to_int(vbios->dmifmc_urgent_latency));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] stutter_self_refresh_exit_latency: %d",
+				bw_fixed_to_int(vbios->stutter_self_refresh_exit_latency));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] stutter_self_refresh_entry_latency: %d",
+				bw_fixed_to_int(vbios->stutter_self_refresh_entry_latency));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] nbp_state_change_latency: %d",
+				bw_fixed_to_int(vbios->nbp_state_change_latency));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] mcifwrmc_urgent_latency: %d",
+				bw_fixed_to_int(vbios->mcifwrmc_urgent_latency));
+	DC_LOG_BANDWIDTH_CALCS("	[bool] scatter_gather_enable: %d", vbios->scatter_gather_enable);
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] down_spread_percentage: %d",
+				bw_fixed_to_int(vbios->down_spread_percentage));
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] cursor_width: %d", vbios->cursor_width);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] average_compression_rate: %d", vbios->average_compression_rate);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] number_of_request_slots_gmc_reserves_for_dmif_per_channel: %d",
+				vbios->number_of_request_slots_gmc_reserves_for_dmif_per_channel);
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] blackout_duration: %d", bw_fixed_to_int(vbios->blackout_duration));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] maximum_blackout_recovery_time: %d",
+				bw_fixed_to_int(vbios->maximum_blackout_recovery_time));
+
+
+}
+
+static void print_bw_calcs_data(struct dc_context *ctx, struct bw_calcs_data *data)
+{
+
+	int i, j, k;
+
+	DC_LOG_BANDWIDTH_CALCS("#####################################################################");
+	DC_LOG_BANDWIDTH_CALCS("struct bw_calcs_data data");
+	DC_LOG_BANDWIDTH_CALCS("#####################################################################");
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] number_of_displays: %d", data->number_of_displays);
+	DC_LOG_BANDWIDTH_CALCS("	[enum] bw_defines underlay_surface_type: %d", data->underlay_surface_type);
+	DC_LOG_BANDWIDTH_CALCS("	[enum] bw_defines panning_and_bezel_adjustment: %d",
+				data->panning_and_bezel_adjustment);
+	DC_LOG_BANDWIDTH_CALCS("	[enum] bw_defines graphics_tiling_mode: %d", data->graphics_tiling_mode);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] graphics_lb_bpc: %d", data->graphics_lb_bpc);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] underlay_lb_bpc: %d", data->underlay_lb_bpc);
+	DC_LOG_BANDWIDTH_CALCS("	[enum] bw_defines underlay_tiling_mode: %d", data->underlay_tiling_mode);
+	DC_LOG_BANDWIDTH_CALCS("	[enum] bw_defines d0_underlay_mode: %d", data->d0_underlay_mode);
+	DC_LOG_BANDWIDTH_CALCS("	[bool] d1_display_write_back_dwb_enable: %d", data->d1_display_write_back_dwb_enable);
+	DC_LOG_BANDWIDTH_CALCS("	[enum] bw_defines d1_underlay_mode: %d", data->d1_underlay_mode);
+	DC_LOG_BANDWIDTH_CALCS("	[bool] cpup_state_change_enable: %d", data->cpup_state_change_enable);
+	DC_LOG_BANDWIDTH_CALCS("	[bool] cpuc_state_change_enable: %d", data->cpuc_state_change_enable);
+	DC_LOG_BANDWIDTH_CALCS("	[bool] nbp_state_change_enable: %d", data->nbp_state_change_enable);
+	DC_LOG_BANDWIDTH_CALCS("	[bool] stutter_mode_enable: %d", data->stutter_mode_enable);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] y_clk_level: %d", data->y_clk_level);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] sclk_level: %d", data->sclk_level);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] number_of_underlay_surfaces: %d", data->number_of_underlay_surfaces);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] number_of_dram_wrchannels: %d", data->number_of_dram_wrchannels);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] chunk_request_delay: %d", data->chunk_request_delay);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] number_of_dram_channels: %d", data->number_of_dram_channels);
+	DC_LOG_BANDWIDTH_CALCS("	[enum] bw_defines underlay_micro_tile_mode: %d", data->underlay_micro_tile_mode);
+	DC_LOG_BANDWIDTH_CALCS("	[enum] bw_defines graphics_micro_tile_mode: %d", data->graphics_micro_tile_mode);
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] max_phyclk: %d", bw_fixed_to_int(data->max_phyclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dram_efficiency: %d", bw_fixed_to_int(data->dram_efficiency));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] src_width_after_surface_type: %d",
+				bw_fixed_to_int(data->src_width_after_surface_type));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] src_height_after_surface_type: %d",
+				bw_fixed_to_int(data->src_height_after_surface_type));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] hsr_after_surface_type: %d",
+				bw_fixed_to_int(data->hsr_after_surface_type));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] vsr_after_surface_type: %d", bw_fixed_to_int(data->vsr_after_surface_type));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] src_width_after_rotation: %d",
+				bw_fixed_to_int(data->src_width_after_rotation));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] src_height_after_rotation: %d",
+				bw_fixed_to_int(data->src_height_after_rotation));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] hsr_after_rotation: %d", bw_fixed_to_int(data->hsr_after_rotation));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] vsr_after_rotation: %d", bw_fixed_to_int(data->vsr_after_rotation));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] source_height_pixels: %d", bw_fixed_to_int(data->source_height_pixels));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] hsr_after_stereo: %d", bw_fixed_to_int(data->hsr_after_stereo));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] vsr_after_stereo: %d", bw_fixed_to_int(data->vsr_after_stereo));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] source_width_in_lb: %d", bw_fixed_to_int(data->source_width_in_lb));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] lb_line_pitch: %d", bw_fixed_to_int(data->lb_line_pitch));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] underlay_maximum_source_efficient_for_tiling: %d",
+				bw_fixed_to_int(data->underlay_maximum_source_efficient_for_tiling));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] num_lines_at_frame_start: %d",
+				bw_fixed_to_int(data->num_lines_at_frame_start));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] min_dmif_size_in_time: %d", bw_fixed_to_int(data->min_dmif_size_in_time));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] min_mcifwr_size_in_time: %d",
+				bw_fixed_to_int(data->min_mcifwr_size_in_time));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] total_requests_for_dmif_size: %d",
+				bw_fixed_to_int(data->total_requests_for_dmif_size));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] peak_pte_request_to_eviction_ratio_limiting: %d",
+				bw_fixed_to_int(data->peak_pte_request_to_eviction_ratio_limiting));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] useful_pte_per_pte_request: %d",
+				bw_fixed_to_int(data->useful_pte_per_pte_request));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] scatter_gather_pte_request_rows: %d",
+				bw_fixed_to_int(data->scatter_gather_pte_request_rows));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] scatter_gather_row_height: %d",
+				bw_fixed_to_int(data->scatter_gather_row_height));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] scatter_gather_pte_requests_in_vblank: %d",
+				bw_fixed_to_int(data->scatter_gather_pte_requests_in_vblank));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] inefficient_linear_pitch_in_bytes: %d",
+				bw_fixed_to_int(data->inefficient_linear_pitch_in_bytes));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] cursor_total_data: %d", bw_fixed_to_int(data->cursor_total_data));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] cursor_total_request_groups: %d",
+				bw_fixed_to_int(data->cursor_total_request_groups));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] scatter_gather_total_pte_requests: %d",
+				bw_fixed_to_int(data->scatter_gather_total_pte_requests));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] scatter_gather_total_pte_request_groups: %d",
+				bw_fixed_to_int(data->scatter_gather_total_pte_request_groups));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] tile_width_in_pixels: %d", bw_fixed_to_int(data->tile_width_in_pixels));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dmif_total_number_of_data_request_page_close_open: %d",
+				bw_fixed_to_int(data->dmif_total_number_of_data_request_page_close_open));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] mcifwr_total_number_of_data_request_page_close_open: %d",
+				bw_fixed_to_int(data->mcifwr_total_number_of_data_request_page_close_open));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] bytes_per_page_close_open: %d",
+				bw_fixed_to_int(data->bytes_per_page_close_open));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] mcifwr_total_page_close_open_time: %d",
+				bw_fixed_to_int(data->mcifwr_total_page_close_open_time));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] total_requests_for_adjusted_dmif_size: %d",
+				bw_fixed_to_int(data->total_requests_for_adjusted_dmif_size));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] total_dmifmc_urgent_trips: %d",
+				bw_fixed_to_int(data->total_dmifmc_urgent_trips));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] total_dmifmc_urgent_latency: %d",
+				bw_fixed_to_int(data->total_dmifmc_urgent_latency));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] total_display_reads_required_data: %d",
+				bw_fixed_to_int(data->total_display_reads_required_data));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] total_display_reads_required_dram_access_data: %d",
+				bw_fixed_to_int(data->total_display_reads_required_dram_access_data));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] total_display_writes_required_data: %d",
+				bw_fixed_to_int(data->total_display_writes_required_data));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] total_display_writes_required_dram_access_data: %d",
+				bw_fixed_to_int(data->total_display_writes_required_dram_access_data));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] display_reads_required_data: %d",
+				bw_fixed_to_int(data->display_reads_required_data));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] display_reads_required_dram_access_data: %d",
+				bw_fixed_to_int(data->display_reads_required_dram_access_data));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dmif_total_page_close_open_time: %d",
+				bw_fixed_to_int(data->dmif_total_page_close_open_time));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] min_cursor_memory_interface_buffer_size_in_time: %d",
+				bw_fixed_to_int(data->min_cursor_memory_interface_buffer_size_in_time));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] min_read_buffer_size_in_time: %d",
+				bw_fixed_to_int(data->min_read_buffer_size_in_time));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] display_reads_time_for_data_transfer: %d",
+				bw_fixed_to_int(data->display_reads_time_for_data_transfer));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] display_writes_time_for_data_transfer: %d",
+				bw_fixed_to_int(data->display_writes_time_for_data_transfer));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dmif_required_dram_bandwidth: %d",
+				bw_fixed_to_int(data->dmif_required_dram_bandwidth));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] mcifwr_required_dram_bandwidth: %d",
+				bw_fixed_to_int(data->mcifwr_required_dram_bandwidth));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] required_dmifmc_urgent_latency_for_page_close_open: %d",
+				bw_fixed_to_int(data->required_dmifmc_urgent_latency_for_page_close_open));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] required_mcifmcwr_urgent_latency: %d",
+				bw_fixed_to_int(data->required_mcifmcwr_urgent_latency));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] required_dram_bandwidth_gbyte_per_second: %d",
+				bw_fixed_to_int(data->required_dram_bandwidth_gbyte_per_second));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dram_bandwidth: %d", bw_fixed_to_int(data->dram_bandwidth));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dmif_required_sclk: %d", bw_fixed_to_int(data->dmif_required_sclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] mcifwr_required_sclk: %d", bw_fixed_to_int(data->mcifwr_required_sclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] required_sclk: %d", bw_fixed_to_int(data->required_sclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] downspread_factor: %d", bw_fixed_to_int(data->downspread_factor));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] v_scaler_efficiency: %d", bw_fixed_to_int(data->v_scaler_efficiency));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] scaler_limits_factor: %d", bw_fixed_to_int(data->scaler_limits_factor));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] display_pipe_pixel_throughput: %d",
+				bw_fixed_to_int(data->display_pipe_pixel_throughput));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] total_dispclk_required_with_ramping: %d",
+				bw_fixed_to_int(data->total_dispclk_required_with_ramping));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] total_dispclk_required_without_ramping: %d",
+				bw_fixed_to_int(data->total_dispclk_required_without_ramping));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] total_read_request_bandwidth: %d",
+				bw_fixed_to_int(data->total_read_request_bandwidth));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] total_write_request_bandwidth: %d",
+				bw_fixed_to_int(data->total_write_request_bandwidth));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dispclk_required_for_total_read_request_bandwidth: %d",
+				bw_fixed_to_int(data->dispclk_required_for_total_read_request_bandwidth));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] total_dispclk_required_with_ramping_with_request_bandwidth: %d",
+				bw_fixed_to_int(data->total_dispclk_required_with_ramping_with_request_bandwidth));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] total_dispclk_required_without_ramping_with_request_bandwidth: %d",
+				bw_fixed_to_int(data->total_dispclk_required_without_ramping_with_request_bandwidth));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dispclk: %d", bw_fixed_to_int(data->dispclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] blackout_recovery_time: %d", bw_fixed_to_int(data->blackout_recovery_time));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] min_pixels_per_data_fifo_entry: %d",
+				bw_fixed_to_int(data->min_pixels_per_data_fifo_entry));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] sclk_deep_sleep: %d", bw_fixed_to_int(data->sclk_deep_sleep));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] chunk_request_time: %d", bw_fixed_to_int(data->chunk_request_time));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] cursor_request_time: %d", bw_fixed_to_int(data->cursor_request_time));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] line_source_pixels_transfer_time: %d",
+				bw_fixed_to_int(data->line_source_pixels_transfer_time));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dmifdram_access_efficiency: %d",
+				bw_fixed_to_int(data->dmifdram_access_efficiency));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] mcifwrdram_access_efficiency: %d",
+				bw_fixed_to_int(data->mcifwrdram_access_efficiency));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] total_average_bandwidth_no_compression: %d",
+				bw_fixed_to_int(data->total_average_bandwidth_no_compression));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] total_average_bandwidth: %d",
+				bw_fixed_to_int(data->total_average_bandwidth));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] total_stutter_cycle_duration: %d",
+				bw_fixed_to_int(data->total_stutter_cycle_duration));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] stutter_burst_time: %d", bw_fixed_to_int(data->stutter_burst_time));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] time_in_self_refresh: %d", bw_fixed_to_int(data->time_in_self_refresh));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] stutter_efficiency: %d", bw_fixed_to_int(data->stutter_efficiency));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] worst_number_of_trips_to_memory: %d",
+				bw_fixed_to_int(data->worst_number_of_trips_to_memory));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] immediate_flip_time: %d", bw_fixed_to_int(data->immediate_flip_time));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] latency_for_non_dmif_clients: %d",
+				bw_fixed_to_int(data->latency_for_non_dmif_clients));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] latency_for_non_mcifwr_clients: %d",
+				bw_fixed_to_int(data->latency_for_non_mcifwr_clients));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dmifmc_urgent_latency_supported_in_high_sclk_and_yclk: %d",
+				bw_fixed_to_int(data->dmifmc_urgent_latency_supported_in_high_sclk_and_yclk));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] nbp_state_dram_speed_change_margin: %d",
+				bw_fixed_to_int(data->nbp_state_dram_speed_change_margin));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] display_reads_time_for_data_transfer_and_urgent_latency: %d",
+				bw_fixed_to_int(data->display_reads_time_for_data_transfer_and_urgent_latency));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dram_speed_change_margin: %d",
+				bw_fixed_to_int(data->dram_speed_change_margin));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] min_vblank_dram_speed_change_margin: %d",
+				bw_fixed_to_int(data->min_vblank_dram_speed_change_margin));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] min_stutter_refresh_duration: %d",
+				bw_fixed_to_int(data->min_stutter_refresh_duration));
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] total_stutter_dmif_buffer_size: %d", data->total_stutter_dmif_buffer_size);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] total_bytes_requested: %d", data->total_bytes_requested);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] min_stutter_dmif_buffer_size: %d", data->min_stutter_dmif_buffer_size);
+	DC_LOG_BANDWIDTH_CALCS("	[uint32_t] num_stutter_bursts: %d", data->num_stutter_bursts);
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] v_blank_nbp_state_dram_speed_change_latency_supported: %d",
+				bw_fixed_to_int(data->v_blank_nbp_state_dram_speed_change_latency_supported));
+	DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] nbp_state_dram_speed_change_latency_supported: %d",
+				bw_fixed_to_int(data->nbp_state_dram_speed_change_latency_supported));
+
+	for (i = 0; i < maximum_number_of_surfaces; i++) {
+		DC_LOG_BANDWIDTH_CALCS("	[bool] fbc_en[%d]:%d\n", i, data->fbc_en[i]);
+		DC_LOG_BANDWIDTH_CALCS("	[bool] lpt_en[%d]:%d", i, data->lpt_en[i]);
+		DC_LOG_BANDWIDTH_CALCS("	[bool] displays_match_flag[%d]:%d", i, data->displays_match_flag[i]);
+		DC_LOG_BANDWIDTH_CALCS("	[bool] use_alpha[%d]:%d", i, data->use_alpha[i]);
+		DC_LOG_BANDWIDTH_CALCS("	[bool] orthogonal_rotation[%d]:%d", i, data->orthogonal_rotation[i]);
+		DC_LOG_BANDWIDTH_CALCS("	[bool] enable[%d]:%d", i, data->enable[i]);
+		DC_LOG_BANDWIDTH_CALCS("	[bool] access_one_channel_only[%d]:%d", i, data->access_one_channel_only[i]);
+		DC_LOG_BANDWIDTH_CALCS("	[bool] scatter_gather_enable_for_pipe[%d]:%d",
+					i, data->scatter_gather_enable_for_pipe[i]);
+		DC_LOG_BANDWIDTH_CALCS("	[bool] interlace_mode[%d]:%d",
+					i, data->interlace_mode[i]);
+		DC_LOG_BANDWIDTH_CALCS("	[bool] display_pstate_change_enable[%d]:%d",
+					i, data->display_pstate_change_enable[i]);
+		DC_LOG_BANDWIDTH_CALCS("	[bool] line_buffer_prefetch[%d]:%d", i, data->line_buffer_prefetch[i]);
+		DC_LOG_BANDWIDTH_CALCS("	[uint32_t] bytes_per_pixel[%d]:%d", i, data->bytes_per_pixel[i]);
+		DC_LOG_BANDWIDTH_CALCS("	[uint32_t] max_chunks_non_fbc_mode[%d]:%d",
+					i, data->max_chunks_non_fbc_mode[i]);
+		DC_LOG_BANDWIDTH_CALCS("	[uint32_t] lb_bpc[%d]:%d", i, data->lb_bpc[i]);
+		DC_LOG_BANDWIDTH_CALCS("	[uint32_t] output_bpphdmi[%d]:%d", i, data->output_bpphdmi[i]);
+		DC_LOG_BANDWIDTH_CALCS("	[uint32_t] output_bppdp4_lane_hbr[%d]:%d", i, data->output_bppdp4_lane_hbr[i]);
+		DC_LOG_BANDWIDTH_CALCS("	[uint32_t] output_bppdp4_lane_hbr2[%d]:%d",
+					i, data->output_bppdp4_lane_hbr2[i]);
+		DC_LOG_BANDWIDTH_CALCS("	[uint32_t] output_bppdp4_lane_hbr3[%d]:%d",
+					i, data->output_bppdp4_lane_hbr3[i]);
+		DC_LOG_BANDWIDTH_CALCS("	[enum] bw_defines stereo_mode[%d]:%d", i, data->stereo_mode[i]);
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dmif_buffer_transfer_time[%d]:%d",
+					i, bw_fixed_to_int(data->dmif_buffer_transfer_time[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] displays_with_same_mode[%d]:%d",
+					i, bw_fixed_to_int(data->displays_with_same_mode[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] stutter_dmif_buffer_size[%d]:%d",
+					i, bw_fixed_to_int(data->stutter_dmif_buffer_size[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] stutter_refresh_duration[%d]:%d",
+					i, bw_fixed_to_int(data->stutter_refresh_duration[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] stutter_exit_watermark[%d]:%d",
+					i, bw_fixed_to_int(data->stutter_exit_watermark[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] stutter_entry_watermark[%d]:%d",
+					i, bw_fixed_to_int(data->stutter_entry_watermark[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] h_total[%d]:%d", i, bw_fixed_to_int(data->h_total[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] v_total[%d]:%d", i, bw_fixed_to_int(data->v_total[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] pixel_rate[%d]:%d", i, bw_fixed_to_int(data->pixel_rate[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] src_width[%d]:%d", i, bw_fixed_to_int(data->src_width[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] pitch_in_pixels[%d]:%d",
+					i, bw_fixed_to_int(data->pitch_in_pixels[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] pitch_in_pixels_after_surface_type[%d]:%d",
+					i, bw_fixed_to_int(data->pitch_in_pixels_after_surface_type[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] src_height[%d]:%d", i, bw_fixed_to_int(data->src_height[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] scale_ratio[%d]:%d", i, bw_fixed_to_int(data->scale_ratio[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] h_taps[%d]:%d", i, bw_fixed_to_int(data->h_taps[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] v_taps[%d]:%d", i, bw_fixed_to_int(data->v_taps[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] h_scale_ratio[%d]:%d", i, bw_fixed_to_int(data->h_scale_ratio[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] v_scale_ratio[%d]:%d", i, bw_fixed_to_int(data->v_scale_ratio[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] rotation_angle[%d]:%d",
+					i, bw_fixed_to_int(data->rotation_angle[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] compression_rate[%d]:%d",
+					i, bw_fixed_to_int(data->compression_rate[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] hsr[%d]:%d", i, bw_fixed_to_int(data->hsr[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] vsr[%d]:%d", i, bw_fixed_to_int(data->vsr[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] source_width_rounded_up_to_chunks[%d]:%d",
+					i, bw_fixed_to_int(data->source_width_rounded_up_to_chunks[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] source_width_pixels[%d]:%d",
+					i, bw_fixed_to_int(data->source_width_pixels[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] source_height_rounded_up_to_chunks[%d]:%d",
+					i, bw_fixed_to_int(data->source_height_rounded_up_to_chunks[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] display_bandwidth[%d]:%d",
+					i, bw_fixed_to_int(data->display_bandwidth[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] request_bandwidth[%d]:%d",
+					i, bw_fixed_to_int(data->request_bandwidth[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] bytes_per_request[%d]:%d",
+					i, bw_fixed_to_int(data->bytes_per_request[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] useful_bytes_per_request[%d]:%d",
+					i, bw_fixed_to_int(data->useful_bytes_per_request[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] lines_interleaved_in_mem_access[%d]:%d",
+					i, bw_fixed_to_int(data->lines_interleaved_in_mem_access[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] latency_hiding_lines[%d]:%d",
+					i, bw_fixed_to_int(data->latency_hiding_lines[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] lb_partitions[%d]:%d",
+					i, bw_fixed_to_int(data->lb_partitions[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] lb_partitions_max[%d]:%d",
+					i, bw_fixed_to_int(data->lb_partitions_max[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dispclk_required_with_ramping[%d]:%d",
+					i, bw_fixed_to_int(data->dispclk_required_with_ramping[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dispclk_required_without_ramping[%d]:%d",
+					i, bw_fixed_to_int(data->dispclk_required_without_ramping[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] data_buffer_size[%d]:%d",
+					i, bw_fixed_to_int(data->data_buffer_size[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] outstanding_chunk_request_limit[%d]:%d",
+					i, bw_fixed_to_int(data->outstanding_chunk_request_limit[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] urgent_watermark[%d]:%d",
+					i, bw_fixed_to_int(data->urgent_watermark[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] nbp_state_change_watermark[%d]:%d",
+					i, bw_fixed_to_int(data->nbp_state_change_watermark[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] v_filter_init[%d]:%d", i, bw_fixed_to_int(data->v_filter_init[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] stutter_cycle_duration[%d]:%d",
+					i, bw_fixed_to_int(data->stutter_cycle_duration[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] average_bandwidth[%d]:%d",
+					i, bw_fixed_to_int(data->average_bandwidth[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] average_bandwidth_no_compression[%d]:%d",
+					i, bw_fixed_to_int(data->average_bandwidth_no_compression[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] scatter_gather_pte_request_limit[%d]:%d",
+					i, bw_fixed_to_int(data->scatter_gather_pte_request_limit[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] lb_size_per_component[%d]:%d",
+					i, bw_fixed_to_int(data->lb_size_per_component[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] memory_chunk_size_in_bytes[%d]:%d",
+					i, bw_fixed_to_int(data->memory_chunk_size_in_bytes[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] pipe_chunk_size_in_bytes[%d]:%d",
+					i, bw_fixed_to_int(data->pipe_chunk_size_in_bytes[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] number_of_trips_to_memory_for_getting_apte_row[%d]:%d",
+					i, bw_fixed_to_int(data->number_of_trips_to_memory_for_getting_apte_row[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] adjusted_data_buffer_size[%d]:%d",
+					i, bw_fixed_to_int(data->adjusted_data_buffer_size[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] adjusted_data_buffer_size_in_memory[%d]:%d",
+					i, bw_fixed_to_int(data->adjusted_data_buffer_size_in_memory[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] pixels_per_data_fifo_entry[%d]:%d",
+					i, bw_fixed_to_int(data->pixels_per_data_fifo_entry[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] scatter_gather_pte_requests_in_row[%d]:%d",
+					i, bw_fixed_to_int(data->scatter_gather_pte_requests_in_row[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] pte_request_per_chunk[%d]:%d",
+					i, bw_fixed_to_int(data->pte_request_per_chunk[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] scatter_gather_page_width[%d]:%d",
+					i, bw_fixed_to_int(data->scatter_gather_page_width[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] scatter_gather_page_height[%d]:%d",
+					i, bw_fixed_to_int(data->scatter_gather_page_height[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] lb_lines_in_per_line_out_in_beginning_of_frame[%d]:%d",
+					i, bw_fixed_to_int(data->lb_lines_in_per_line_out_in_beginning_of_frame[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] lb_lines_in_per_line_out_in_middle_of_frame[%d]:%d",
+					i, bw_fixed_to_int(data->lb_lines_in_per_line_out_in_middle_of_frame[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] cursor_width_pixels[%d]:%d",
+					i, bw_fixed_to_int(data->cursor_width_pixels[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] minimum_latency_hiding[%d]:%d",
+					i, bw_fixed_to_int(data->minimum_latency_hiding[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] maximum_latency_hiding[%d]:%d",
+					i, bw_fixed_to_int(data->maximum_latency_hiding[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] minimum_latency_hiding_with_cursor[%d]:%d",
+					i, bw_fixed_to_int(data->minimum_latency_hiding_with_cursor[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] maximum_latency_hiding_with_cursor[%d]:%d",
+					i, bw_fixed_to_int(data->maximum_latency_hiding_with_cursor[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] src_pixels_for_first_output_pixel[%d]:%d",
+					i, bw_fixed_to_int(data->src_pixels_for_first_output_pixel[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] src_pixels_for_last_output_pixel[%d]:%d",
+					i, bw_fixed_to_int(data->src_pixels_for_last_output_pixel[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] src_data_for_first_output_pixel[%d]:%d",
+					i, bw_fixed_to_int(data->src_data_for_first_output_pixel[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] src_data_for_last_output_pixel[%d]:%d",
+					i, bw_fixed_to_int(data->src_data_for_last_output_pixel[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] active_time[%d]:%d", i, bw_fixed_to_int(data->active_time[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] horizontal_blank_and_chunk_granularity_factor[%d]:%d",
+					i, bw_fixed_to_int(data->horizontal_blank_and_chunk_granularity_factor[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] cursor_latency_hiding[%d]:%d",
+					i, bw_fixed_to_int(data->cursor_latency_hiding[i]));
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] v_blank_dram_speed_change_margin[%d]:%d",
+					i, bw_fixed_to_int(data->v_blank_dram_speed_change_margin[i]));
+		}
+
+	for (i = 0; i < maximum_number_of_surfaces; i++) {
+		for (j = 0; j < 3; j++) {
+			for (k = 0; k < 8; k++) {
+
+				DC_LOG_BANDWIDTH_CALCS("\n	[bw_fixed] line_source_transfer_time[%d][%d][%d]:%d",
+					i, j, k, bw_fixed_to_int(data->line_source_transfer_time[i][j][k]));
+				DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dram_speed_change_line_source_transfer_time[%d][%d][%d]:%d",
+					i, j, k,
+					bw_fixed_to_int(data->dram_speed_change_line_source_transfer_time[i][j][k]));
+			}
+		}
+	}
+
+	for (i = 0; i < 3; i++) {
+		for (j = 0; j < 8; j++) {
+
+			DC_LOG_BANDWIDTH_CALCS("\n	[uint32_t] num_displays_with_margin[%d][%d]:%d",
+					i, j, data->num_displays_with_margin[i][j]);
+			DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dmif_burst_time[%d][%d]:%d",
+					i, j, bw_fixed_to_int(data->dmif_burst_time[i][j]));
+			DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] mcifwr_burst_time[%d][%d]:%d",
+					i, j, bw_fixed_to_int(data->mcifwr_burst_time[i][j]));
+			DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] min_dram_speed_change_margin[%d][%d]:%d",
+					i, j, bw_fixed_to_int(data->min_dram_speed_change_margin[i][j]));
+			DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dispclk_required_for_dram_speed_change[%d][%d]:%d",
+					i, j, bw_fixed_to_int(data->dispclk_required_for_dram_speed_change[i][j]));
+			DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] blackout_duration_margin[%d][%d]:%d",
+					i, j, bw_fixed_to_int(data->blackout_duration_margin[i][j]));
+			DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dispclk_required_for_blackout_duration[%d][%d]:%d",
+					i, j, bw_fixed_to_int(data->dispclk_required_for_blackout_duration[i][j]));
+			DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dispclk_required_for_blackout_recovery[%d][%d]:%d",
+					i, j, bw_fixed_to_int(data->dispclk_required_for_blackout_recovery[i][j]));
+		}
+	}
+
+	for (i = 0; i < 6; i++) {
+		DC_LOG_BANDWIDTH_CALCS("	[bw_fixed] dmif_required_sclk_for_urgent_latency[%d]:%d",
+					i, bw_fixed_to_int(data->dmif_required_sclk_for_urgent_latency[i]));
+	}
+}
+;
+
+#endif /* _CALCS_CALCS_LOGGER_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/basics/conversion.c b/drivers/gpu/drm/amd/display/dc/basics/conversion.c
new file mode 100644
index 000000000000..bd1f60ecaba4
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/basics/conversion.c
@@ -0,0 +1,158 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+#include "basics/conversion.h"
+
+#define DIVIDER 10000
+
+/* S2D13 value in [-3.00...0.9999] */
+#define S2D13_MIN (-3 * DIVIDER)
+#define S2D13_MAX (3 * DIVIDER)
+
+uint16_t fixed_point_to_int_frac(
+	struct fixed31_32 arg,
+	uint8_t integer_bits,
+	uint8_t fractional_bits)
+{
+	int32_t numerator;
+	int32_t divisor = 1 << fractional_bits;
+
+	uint16_t result;
+
+	uint16_t d = (uint16_t)dc_fixpt_floor(
+		dc_fixpt_abs(
+			arg));
+
+	if (d <= (uint16_t)(1 << integer_bits) - (1 / (uint16_t)divisor))
+		numerator = (uint16_t)dc_fixpt_round(
+			dc_fixpt_mul_int(
+				arg,
+				divisor));
+	else {
+		numerator = dc_fixpt_floor(
+			dc_fixpt_sub(
+				dc_fixpt_from_int(
+					1LL << integer_bits),
+				dc_fixpt_recip(
+					dc_fixpt_from_int(
+						divisor))));
+	}
+
+	if (numerator >= 0)
+		result = (uint16_t)numerator;
+	else
+		result = (uint16_t)(
+		(1 << (integer_bits + fractional_bits + 1)) + numerator);
+
+	if ((result != 0) && dc_fixpt_lt(
+		arg, dc_fixpt_zero))
+		result |= 1 << (integer_bits + fractional_bits);
+
+	return result;
+}
+/*
+ * convert_float_matrix - This converts a double into HW register spec defined format S2D13.
+ */
+void convert_float_matrix(
+	uint16_t *matrix,
+	struct fixed31_32 *flt,
+	uint32_t buffer_size)
+{
+	const struct fixed31_32 min_2_13 =
+		dc_fixpt_from_fraction(S2D13_MIN, DIVIDER);
+	const struct fixed31_32 max_2_13 =
+		dc_fixpt_from_fraction(S2D13_MAX, DIVIDER);
+	uint32_t i;
+
+	for (i = 0; i < buffer_size; ++i) {
+		uint32_t reg_value =
+				fixed_point_to_int_frac(
+					dc_fixpt_clamp(
+						flt[i],
+						min_2_13,
+						max_2_13),
+						2,
+						13);
+
+		matrix[i] = (uint16_t)reg_value;
+	}
+}
+
+static struct fixed31_32 int_frac_to_fixed_point(uint16_t arg,
+						 uint8_t integer_bits,
+						 uint8_t fractional_bits)
+{
+	struct fixed31_32 result;
+	uint16_t sign_mask = 1 << (fractional_bits + integer_bits);
+	uint16_t value_mask = sign_mask - 1;
+
+	result.value = (long long)(arg & value_mask) <<
+		       (FIXED31_32_BITS_PER_FRACTIONAL_PART - fractional_bits);
+
+	if (arg & sign_mask)
+		result = dc_fixpt_neg(result);
+
+	return result;
+}
+
+/**
+ * convert_hw_matrix - converts HW values into fixed31_32 matrix.
+ * @matrix: fixed point 31.32 matrix
+ * @reg: array of register values
+ * @buffer_size: size of the array of register values
+ *
+ * Converts HW register spec defined format S2D13 into a fixed-point 31.32
+ * matrix.
+ */
+void convert_hw_matrix(struct fixed31_32 *matrix,
+		       uint16_t *reg,
+		       uint32_t buffer_size)
+{
+	for (int i = 0; i < buffer_size; ++i)
+		matrix[i] = int_frac_to_fixed_point(reg[i], 2, 13);
+}
+
+static uint32_t find_gcd(uint32_t a, uint32_t b)
+{
+	uint32_t remainder;
+
+	while (b != 0) {
+		remainder = a % b;
+		a = b;
+		b = remainder;
+	}
+	return a;
+}
+
+void reduce_fraction(uint32_t num, uint32_t den,
+		uint32_t *out_num, uint32_t *out_den)
+{
+	uint32_t gcd = 0;
+
+	gcd = find_gcd(num, den);
+	*out_num = num / gcd;
+	*out_den = den / gcd;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/basics/conversion.h b/drivers/gpu/drm/amd/display/dc/basics/conversion.h
new file mode 100644
index 000000000000..a433cef78496
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/basics/conversion.h
@@ -0,0 +1,53 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_CONVERSION_H__
+#define __DAL_CONVERSION_H__
+
+#include "include/fixed31_32.h"
+
+uint16_t fixed_point_to_int_frac(
+	struct fixed31_32 arg,
+	uint8_t integer_bits,
+	uint8_t fractional_bits);
+
+void convert_float_matrix(
+	uint16_t *matrix,
+	struct fixed31_32 *flt,
+	uint32_t buffer_size);
+
+void reduce_fraction(uint32_t num, uint32_t den,
+		uint32_t *out_num, uint32_t *out_den);
+
+void convert_hw_matrix(struct fixed31_32 *matrix,
+		       uint16_t *reg,
+		       uint32_t buffer_size);
+
+static inline unsigned int log_2(unsigned int num)
+{
+	return ilog2(num);
+}
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/basics/custom_float.c b/drivers/gpu/drm/amd/display/dc/basics/custom_float.c
new file mode 100644
index 000000000000..ae05ded9a7f3
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/basics/custom_float.c
@@ -0,0 +1,173 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+#include "dm_services.h"
+#include "custom_float.h"
+
+static bool build_custom_float(struct fixed31_32 value,
+			       const struct custom_float_format *format,
+			       bool *negative,
+			       uint32_t *mantissa,
+			       uint32_t *exponenta)
+{
+	uint32_t exp_offset = (1 << (format->exponenta_bits - 1)) - 1;
+
+	const struct fixed31_32 mantissa_constant_plus_max_fraction =
+		dc_fixpt_from_fraction((1LL << (format->mantissa_bits + 1)) - 1,
+				       1LL << format->mantissa_bits);
+
+	struct fixed31_32 mantiss;
+
+	if (dc_fixpt_eq(value, dc_fixpt_zero)) {
+		*negative = false;
+		*mantissa = 0;
+		*exponenta = 0;
+		return true;
+	}
+
+	if (dc_fixpt_lt(value, dc_fixpt_zero)) {
+		*negative = format->sign;
+		value = dc_fixpt_neg(value);
+	} else {
+		*negative = false;
+	}
+
+	if (dc_fixpt_lt(value, dc_fixpt_one)) {
+		uint32_t i = 1;
+
+		do {
+			value = dc_fixpt_shl(value, 1);
+			++i;
+		} while (dc_fixpt_lt(value, dc_fixpt_one));
+
+		--i;
+
+		if (exp_offset <= i) {
+			*mantissa = 0;
+			*exponenta = 0;
+			return true;
+		}
+
+		*exponenta = exp_offset - i;
+	} else if (dc_fixpt_le(mantissa_constant_plus_max_fraction, value)) {
+		uint32_t i = 1;
+
+		do {
+			value = dc_fixpt_shr(value, 1);
+			++i;
+		} while (dc_fixpt_lt(mantissa_constant_plus_max_fraction, value));
+
+		*exponenta = exp_offset + i - 1;
+	} else {
+		*exponenta = exp_offset;
+	}
+
+	mantiss = dc_fixpt_sub(value, dc_fixpt_one);
+
+	if (dc_fixpt_lt(mantiss, dc_fixpt_zero) ||
+	    dc_fixpt_lt(dc_fixpt_one, mantiss))
+		mantiss = dc_fixpt_zero;
+	else
+		mantiss = dc_fixpt_shl(mantiss, format->mantissa_bits);
+
+	*mantissa = dc_fixpt_floor(mantiss);
+
+	return true;
+}
+
+static bool setup_custom_float(const struct custom_float_format *format,
+			       bool negative,
+			       uint32_t mantissa,
+			       uint32_t exponenta,
+			       uint32_t *result)
+{
+	uint32_t i = 0;
+	uint32_t j = 0;
+	uint32_t value = 0;
+
+	/* verification code:
+	 * once calculation is ok we can remove it
+	 */
+
+	const uint32_t mantissa_mask =
+		(1 << (format->mantissa_bits + 1)) - 1;
+
+	const uint32_t exponenta_mask =
+		(1 << (format->exponenta_bits + 1)) - 1;
+
+	if (mantissa & ~mantissa_mask) {
+		BREAK_TO_DEBUGGER();
+		mantissa = mantissa_mask;
+	}
+
+	if (exponenta & ~exponenta_mask) {
+		BREAK_TO_DEBUGGER();
+		exponenta = exponenta_mask;
+	}
+
+	/* end of verification code */
+
+	while (i < format->mantissa_bits) {
+		uint32_t mask = 1 << i;
+
+		if (mantissa & mask)
+			value |= mask;
+
+		++i;
+	}
+
+	while (j < format->exponenta_bits) {
+		uint32_t mask = 1 << j;
+
+		if (exponenta & mask)
+			value |= mask << i;
+
+		++j;
+	}
+
+	if (negative && format->sign)
+		value |= 1 << (i + j);
+
+	*result = value;
+
+	return true;
+}
+
+bool convert_to_custom_float_format(struct fixed31_32 value,
+				    const struct custom_float_format *format,
+				    uint32_t *result)
+{
+	uint32_t mantissa;
+	uint32_t exponenta;
+	bool negative;
+
+	return build_custom_float(value, format, &negative, &mantissa, &exponenta) &&
+				  setup_custom_float(format,
+						     negative,
+						     mantissa,
+						     exponenta,
+						     result);
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/basics/dc_common.c b/drivers/gpu/drm/amd/display/dc/basics/dc_common.c
new file mode 100644
index 000000000000..a51c2701da24
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/basics/dc_common.c
@@ -0,0 +1,102 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "core_types.h"
+#include "dc_common.h"
+#include "basics/conversion.h"
+
+bool is_rgb_cspace(enum dc_color_space output_color_space)
+{
+	switch (output_color_space) {
+	case COLOR_SPACE_SRGB:
+	case COLOR_SPACE_SRGB_LIMITED:
+	case COLOR_SPACE_2020_RGB_FULLRANGE:
+	case COLOR_SPACE_2020_RGB_LIMITEDRANGE:
+	case COLOR_SPACE_ADOBERGB:
+		return true;
+	case COLOR_SPACE_YCBCR601:
+	case COLOR_SPACE_YCBCR709:
+	case COLOR_SPACE_YCBCR601_LIMITED:
+	case COLOR_SPACE_YCBCR709_LIMITED:
+	case COLOR_SPACE_2020_YCBCR_LIMITED:
+	case COLOR_SPACE_2020_YCBCR_FULL:
+		return false;
+	default:
+		/* Add a case to switch */
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+}
+
+bool is_lower_pipe_tree_visible(struct pipe_ctx *pipe_ctx)
+{
+	if (pipe_ctx->plane_state && pipe_ctx->plane_state->visible)
+		return true;
+	if (pipe_ctx->bottom_pipe && is_lower_pipe_tree_visible(pipe_ctx->bottom_pipe))
+		return true;
+	return false;
+}
+
+bool is_upper_pipe_tree_visible(struct pipe_ctx *pipe_ctx)
+{
+	if (pipe_ctx->plane_state && pipe_ctx->plane_state->visible)
+		return true;
+	if (pipe_ctx->top_pipe && is_upper_pipe_tree_visible(pipe_ctx->top_pipe))
+		return true;
+	return false;
+}
+
+bool is_pipe_tree_visible(struct pipe_ctx *pipe_ctx)
+{
+	if (pipe_ctx->plane_state && pipe_ctx->plane_state->visible)
+		return true;
+	if (pipe_ctx->top_pipe && is_upper_pipe_tree_visible(pipe_ctx->top_pipe))
+		return true;
+	if (pipe_ctx->bottom_pipe && is_lower_pipe_tree_visible(pipe_ctx->bottom_pipe))
+		return true;
+	return false;
+}
+
+void build_prescale_params(struct  dc_bias_and_scale *bias_and_scale,
+		const struct dc_plane_state *plane_state)
+{
+	if (plane_state->format >= SURFACE_PIXEL_FORMAT_VIDEO_BEGIN
+			&& plane_state->format != SURFACE_PIXEL_FORMAT_INVALID
+			&& plane_state->input_csc_color_matrix.enable_adjustment
+			&& plane_state->coeff_reduction_factor.value != 0) {
+		bias_and_scale->scale_blue = fixed_point_to_int_frac(
+			dc_fixpt_mul(plane_state->coeff_reduction_factor,
+					dc_fixpt_from_fraction(256, 255)),
+				2,
+				13);
+		bias_and_scale->scale_red = bias_and_scale->scale_blue;
+		bias_and_scale->scale_green = bias_and_scale->scale_blue;
+	} else {
+		bias_and_scale->scale_blue = 0x2000;
+		bias_and_scale->scale_red = 0x2000;
+		bias_and_scale->scale_green = 0x2000;
+	}
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/basics/dc_common.h b/drivers/gpu/drm/amd/display/dc/basics/dc_common.h
new file mode 100644
index 000000000000..7c0cbf47e8ce
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/basics/dc_common.h
@@ -0,0 +1,42 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_DC_COMMON_H__
+#define __DAL_DC_COMMON_H__
+
+#include "core_types.h"
+
+bool is_rgb_cspace(enum dc_color_space output_color_space);
+
+bool is_lower_pipe_tree_visible(struct pipe_ctx *pipe_ctx);
+
+bool is_upper_pipe_tree_visible(struct pipe_ctx *pipe_ctx);
+
+bool is_pipe_tree_visible(struct pipe_ctx *pipe_ctx);
+
+void build_prescale_params(struct  dc_bias_and_scale *bias_and_scale,
+		const struct dc_plane_state *plane_state);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/basics/dce_calcs.c b/drivers/gpu/drm/amd/display/dc/basics/dce_calcs.c
new file mode 100644
index 000000000000..4da5adab799c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/basics/dce_calcs.c
@@ -0,0 +1,3621 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "resource.h"
+#include "dm_services.h"
+#include "dce_calcs.h"
+#include "dc.h"
+#include "core_types.h"
+#include "dal_asic_id.h"
+#include "calcs_logger.h"
+
+/*
+ * NOTE:
+ *   This file is gcc-parseable HW gospel, coming straight from HW engineers.
+ *
+ * It doesn't adhere to Linux kernel style and sometimes will do things in odd
+ * ways. Unless there is something clearly wrong with it the code should
+ * remain as-is as it provides us with a guarantee from HW that it is correct.
+ */
+
+/*******************************************************************************
+ * Private Functions
+ ******************************************************************************/
+
+static enum bw_calcs_version bw_calcs_version_from_asic_id(struct hw_asic_id asic_id)
+{
+	switch (asic_id.chip_family) {
+
+	case FAMILY_CZ:
+		if (ASIC_REV_IS_STONEY(asic_id.hw_internal_rev))
+			return BW_CALCS_VERSION_STONEY;
+		return BW_CALCS_VERSION_CARRIZO;
+
+	case FAMILY_VI:
+		if (ASIC_REV_IS_POLARIS12_V(asic_id.hw_internal_rev))
+			return BW_CALCS_VERSION_POLARIS12;
+		if (ASIC_REV_IS_POLARIS10_P(asic_id.hw_internal_rev))
+			return BW_CALCS_VERSION_POLARIS10;
+		if (ASIC_REV_IS_POLARIS11_M(asic_id.hw_internal_rev))
+			return BW_CALCS_VERSION_POLARIS11;
+		if (ASIC_REV_IS_VEGAM(asic_id.hw_internal_rev))
+			return BW_CALCS_VERSION_VEGAM;
+		return BW_CALCS_VERSION_INVALID;
+
+	case FAMILY_AI:
+		return BW_CALCS_VERSION_VEGA10;
+
+	default:
+		return BW_CALCS_VERSION_INVALID;
+	}
+}
+
+static void calculate_bandwidth(
+	const struct bw_calcs_dceip *dceip,
+	const struct bw_calcs_vbios *vbios,
+	struct bw_calcs_data *data)
+
+{
+	const int32_t pixels_per_chunk = 512;
+	const int32_t high = 2;
+	const int32_t mid = 1;
+	const int32_t low = 0;
+	const uint32_t s_low = 0;
+	const uint32_t s_mid1 = 1;
+	const uint32_t s_mid2 = 2;
+	const uint32_t s_mid3 = 3;
+	const uint32_t s_mid4 = 4;
+	const uint32_t s_mid5 = 5;
+	const uint32_t s_mid6 = 6;
+	const uint32_t s_high = 7;
+	const uint32_t dmif_chunk_buff_margin = 1;
+
+	uint32_t max_chunks_fbc_mode = 0;
+	int32_t num_cursor_lines;
+
+	int32_t i, j, k;
+	struct bw_fixed *yclk;
+	struct bw_fixed *sclk;
+	bool d0_underlay_enable;
+	bool d1_underlay_enable;
+	bool fbc_enabled;
+	bool lpt_enabled;
+	enum bw_defines sclk_message;
+	enum bw_defines yclk_message;
+	enum bw_defines *tiling_mode;
+	enum bw_defines *surface_type;
+	enum bw_defines voltage;
+	enum bw_defines pipe_check;
+	enum bw_defines hsr_check;
+	enum bw_defines vsr_check;
+	enum bw_defines lb_size_check;
+	enum bw_defines fbc_check;
+	enum bw_defines rotation_check;
+	enum bw_defines mode_check;
+	enum bw_defines nbp_state_change_enable_blank;
+	/*initialize variables*/
+	int32_t number_of_displays_enabled = 0;
+	int32_t number_of_displays_enabled_with_margin = 0;
+	int32_t number_of_aligned_displays_with_no_margin = 0;
+
+	yclk = kcalloc(3, sizeof(*yclk), GFP_KERNEL);
+	if (!yclk)
+		return;
+
+	sclk = kcalloc(8, sizeof(*sclk), GFP_KERNEL);
+	if (!sclk)
+		goto free_yclk;
+
+	tiling_mode = kcalloc(maximum_number_of_surfaces, sizeof(*tiling_mode), GFP_KERNEL);
+	if (!tiling_mode)
+		goto free_sclk;
+
+	surface_type = kcalloc(maximum_number_of_surfaces, sizeof(*surface_type), GFP_KERNEL);
+	if (!surface_type)
+		goto free_tiling_mode;
+
+	yclk[low] = vbios->low_yclk;
+	yclk[mid] = vbios->mid_yclk;
+	yclk[high] = vbios->high_yclk;
+	sclk[s_low] = vbios->low_sclk;
+	sclk[s_mid1] = vbios->mid1_sclk;
+	sclk[s_mid2] = vbios->mid2_sclk;
+	sclk[s_mid3] = vbios->mid3_sclk;
+	sclk[s_mid4] = vbios->mid4_sclk;
+	sclk[s_mid5] = vbios->mid5_sclk;
+	sclk[s_mid6] = vbios->mid6_sclk;
+	sclk[s_high] = vbios->high_sclk;
+	/*''''''''''''''''''*/
+	/* surface assignment:*/
+	/* 0: d0 underlay or underlay luma*/
+	/* 1: d0 underlay chroma*/
+	/* 2: d1 underlay or underlay luma*/
+	/* 3: d1 underlay chroma*/
+	/* 4: d0 graphics*/
+	/* 5: d1 graphics*/
+	/* 6: d2 graphics*/
+	/* 7: d3 graphics, same mode as d2*/
+	/* 8: d4 graphics, same mode as d2*/
+	/* 9: d5 graphics, same mode as d2*/
+	/* ...*/
+	/* maximum_number_of_surfaces-2: d1 display_write_back420 luma*/
+	/* maximum_number_of_surfaces-1: d1 display_write_back420 chroma*/
+	/* underlay luma and chroma surface parameters from spreadsheet*/
+
+
+
+
+	if (data->d0_underlay_mode == bw_def_none)
+		d0_underlay_enable = false;
+	else
+		d0_underlay_enable = true;
+	if (data->d1_underlay_mode == bw_def_none)
+		d1_underlay_enable = false;
+	else
+		d1_underlay_enable = true;
+	data->number_of_underlay_surfaces = d0_underlay_enable + d1_underlay_enable;
+	switch (data->underlay_surface_type) {
+	case bw_def_420:
+		surface_type[0] = bw_def_underlay420_luma;
+		surface_type[2] = bw_def_underlay420_luma;
+		data->bytes_per_pixel[0] = 1;
+		data->bytes_per_pixel[2] = 1;
+		surface_type[1] = bw_def_underlay420_chroma;
+		surface_type[3] = bw_def_underlay420_chroma;
+		data->bytes_per_pixel[1] = 2;
+		data->bytes_per_pixel[3] = 2;
+		data->lb_size_per_component[0] = dceip->underlay420_luma_lb_size_per_component;
+		data->lb_size_per_component[1] = dceip->underlay420_chroma_lb_size_per_component;
+		data->lb_size_per_component[2] = dceip->underlay420_luma_lb_size_per_component;
+		data->lb_size_per_component[3] = dceip->underlay420_chroma_lb_size_per_component;
+		break;
+	case bw_def_422:
+		surface_type[0] = bw_def_underlay422;
+		surface_type[2] = bw_def_underlay422;
+		data->bytes_per_pixel[0] = 2;
+		data->bytes_per_pixel[2] = 2;
+		data->lb_size_per_component[0] = dceip->underlay422_lb_size_per_component;
+		data->lb_size_per_component[2] = dceip->underlay422_lb_size_per_component;
+		break;
+	default:
+		surface_type[0] = bw_def_underlay444;
+		surface_type[2] = bw_def_underlay444;
+		data->bytes_per_pixel[0] = 4;
+		data->bytes_per_pixel[2] = 4;
+		data->lb_size_per_component[0] = dceip->lb_size_per_component444;
+		data->lb_size_per_component[2] = dceip->lb_size_per_component444;
+		break;
+	}
+	if (d0_underlay_enable) {
+		switch (data->underlay_surface_type) {
+		case bw_def_420:
+			data->enable[0] = 1;
+			data->enable[1] = 1;
+			break;
+		default:
+			data->enable[0] = 1;
+			data->enable[1] = 0;
+			break;
+		}
+	}
+	else {
+		data->enable[0] = 0;
+		data->enable[1] = 0;
+	}
+	if (d1_underlay_enable) {
+		switch (data->underlay_surface_type) {
+		case bw_def_420:
+			data->enable[2] = 1;
+			data->enable[3] = 1;
+			break;
+		default:
+			data->enable[2] = 1;
+			data->enable[3] = 0;
+			break;
+		}
+	}
+	else {
+		data->enable[2] = 0;
+		data->enable[3] = 0;
+	}
+	data->use_alpha[0] = 0;
+	data->use_alpha[1] = 0;
+	data->use_alpha[2] = 0;
+	data->use_alpha[3] = 0;
+	data->scatter_gather_enable_for_pipe[0] = vbios->scatter_gather_enable;
+	data->scatter_gather_enable_for_pipe[1] = vbios->scatter_gather_enable;
+	data->scatter_gather_enable_for_pipe[2] = vbios->scatter_gather_enable;
+	data->scatter_gather_enable_for_pipe[3] = vbios->scatter_gather_enable;
+	/*underlay0 same and graphics display pipe0*/
+	data->interlace_mode[0] = data->interlace_mode[4];
+	data->interlace_mode[1] = data->interlace_mode[4];
+	/*underlay1 same and graphics display pipe1*/
+	data->interlace_mode[2] = data->interlace_mode[5];
+	data->interlace_mode[3] = data->interlace_mode[5];
+	/*underlay0 same and graphics display pipe0*/
+	data->h_total[0] = data->h_total[4];
+	data->v_total[0] = data->v_total[4];
+	data->h_total[1] = data->h_total[4];
+	data->v_total[1] = data->v_total[4];
+	/*underlay1 same and graphics display pipe1*/
+	data->h_total[2] = data->h_total[5];
+	data->v_total[2] = data->v_total[5];
+	data->h_total[3] = data->h_total[5];
+	data->v_total[3] = data->v_total[5];
+	/*underlay0 same and graphics display pipe0*/
+	data->pixel_rate[0] = data->pixel_rate[4];
+	data->pixel_rate[1] = data->pixel_rate[4];
+	/*underlay1 same and graphics display pipe1*/
+	data->pixel_rate[2] = data->pixel_rate[5];
+	data->pixel_rate[3] = data->pixel_rate[5];
+	if ((data->underlay_tiling_mode == bw_def_array_linear_general || data->underlay_tiling_mode == bw_def_array_linear_aligned)) {
+		tiling_mode[0] = bw_def_linear;
+		tiling_mode[1] = bw_def_linear;
+		tiling_mode[2] = bw_def_linear;
+		tiling_mode[3] = bw_def_linear;
+	}
+	else {
+		tiling_mode[0] = bw_def_landscape;
+		tiling_mode[1] = bw_def_landscape;
+		tiling_mode[2] = bw_def_landscape;
+		tiling_mode[3] = bw_def_landscape;
+	}
+	data->lb_bpc[0] = data->underlay_lb_bpc;
+	data->lb_bpc[1] = data->underlay_lb_bpc;
+	data->lb_bpc[2] = data->underlay_lb_bpc;
+	data->lb_bpc[3] = data->underlay_lb_bpc;
+	data->compression_rate[0] = bw_int_to_fixed(1);
+	data->compression_rate[1] = bw_int_to_fixed(1);
+	data->compression_rate[2] = bw_int_to_fixed(1);
+	data->compression_rate[3] = bw_int_to_fixed(1);
+	data->access_one_channel_only[0] = 0;
+	data->access_one_channel_only[1] = 0;
+	data->access_one_channel_only[2] = 0;
+	data->access_one_channel_only[3] = 0;
+	data->cursor_width_pixels[0] = bw_int_to_fixed(0);
+	data->cursor_width_pixels[1] = bw_int_to_fixed(0);
+	data->cursor_width_pixels[2] = bw_int_to_fixed(0);
+	data->cursor_width_pixels[3] = bw_int_to_fixed(0);
+	/* graphics surface parameters from spreadsheet*/
+	fbc_enabled = false;
+	lpt_enabled = false;
+	for (i = 4; i <= maximum_number_of_surfaces - 3; i++) {
+		if (i < data->number_of_displays + 4) {
+			if (i == 4 && data->d0_underlay_mode == bw_def_underlay_only) {
+				data->enable[i] = 0;
+				data->use_alpha[i] = 0;
+			}
+			else if (i == 4 && data->d0_underlay_mode == bw_def_blend) {
+				data->enable[i] = 1;
+				data->use_alpha[i] = 1;
+			}
+			else if (i == 4) {
+				data->enable[i] = 1;
+				data->use_alpha[i] = 0;
+			}
+			else if (i == 5 && data->d1_underlay_mode == bw_def_underlay_only) {
+				data->enable[i] = 0;
+				data->use_alpha[i] = 0;
+			}
+			else if (i == 5 && data->d1_underlay_mode == bw_def_blend) {
+				data->enable[i] = 1;
+				data->use_alpha[i] = 1;
+			}
+			else {
+				data->enable[i] = 1;
+				data->use_alpha[i] = 0;
+			}
+		}
+		else {
+			data->enable[i] = 0;
+			data->use_alpha[i] = 0;
+		}
+		data->scatter_gather_enable_for_pipe[i] = vbios->scatter_gather_enable;
+		surface_type[i] = bw_def_graphics;
+		data->lb_size_per_component[i] = dceip->lb_size_per_component444;
+		if (data->graphics_tiling_mode == bw_def_array_linear_general || data->graphics_tiling_mode == bw_def_array_linear_aligned) {
+			tiling_mode[i] = bw_def_linear;
+		}
+		else {
+			tiling_mode[i] = bw_def_tiled;
+		}
+		data->lb_bpc[i] = data->graphics_lb_bpc;
+		if ((data->fbc_en[i] == 1 && (dceip->argb_compression_support || data->d0_underlay_mode != bw_def_blended))) {
+			data->compression_rate[i] = bw_int_to_fixed(vbios->average_compression_rate);
+			data->access_one_channel_only[i] = data->lpt_en[i];
+		}
+		else {
+			data->compression_rate[i] = bw_int_to_fixed(1);
+			data->access_one_channel_only[i] = 0;
+		}
+		if (data->fbc_en[i] == 1) {
+			fbc_enabled = true;
+			if (data->lpt_en[i] == 1) {
+				lpt_enabled = true;
+			}
+		}
+		data->cursor_width_pixels[i] = bw_int_to_fixed(vbios->cursor_width);
+	}
+	/* display_write_back420*/
+	data->scatter_gather_enable_for_pipe[maximum_number_of_surfaces - 2] = 0;
+	data->scatter_gather_enable_for_pipe[maximum_number_of_surfaces - 1] = 0;
+	if (data->d1_display_write_back_dwb_enable == 1) {
+		data->enable[maximum_number_of_surfaces - 2] = 1;
+		data->enable[maximum_number_of_surfaces - 1] = 1;
+	}
+	else {
+		data->enable[maximum_number_of_surfaces - 2] = 0;
+		data->enable[maximum_number_of_surfaces - 1] = 0;
+	}
+	surface_type[maximum_number_of_surfaces - 2] = bw_def_display_write_back420_luma;
+	surface_type[maximum_number_of_surfaces - 1] = bw_def_display_write_back420_chroma;
+	data->lb_size_per_component[maximum_number_of_surfaces - 2] = dceip->underlay420_luma_lb_size_per_component;
+	data->lb_size_per_component[maximum_number_of_surfaces - 1] = dceip->underlay420_chroma_lb_size_per_component;
+	data->bytes_per_pixel[maximum_number_of_surfaces - 2] = 1;
+	data->bytes_per_pixel[maximum_number_of_surfaces - 1] = 2;
+	data->interlace_mode[maximum_number_of_surfaces - 2] = data->interlace_mode[5];
+	data->interlace_mode[maximum_number_of_surfaces - 1] = data->interlace_mode[5];
+	data->h_taps[maximum_number_of_surfaces - 2] = bw_int_to_fixed(1);
+	data->h_taps[maximum_number_of_surfaces - 1] = bw_int_to_fixed(1);
+	data->v_taps[maximum_number_of_surfaces - 2] = bw_int_to_fixed(1);
+	data->v_taps[maximum_number_of_surfaces - 1] = bw_int_to_fixed(1);
+	data->rotation_angle[maximum_number_of_surfaces - 2] = bw_int_to_fixed(0);
+	data->rotation_angle[maximum_number_of_surfaces - 1] = bw_int_to_fixed(0);
+	tiling_mode[maximum_number_of_surfaces - 2] = bw_def_linear;
+	tiling_mode[maximum_number_of_surfaces - 1] = bw_def_linear;
+	data->lb_bpc[maximum_number_of_surfaces - 2] = 8;
+	data->lb_bpc[maximum_number_of_surfaces - 1] = 8;
+	data->compression_rate[maximum_number_of_surfaces - 2] = bw_int_to_fixed(1);
+	data->compression_rate[maximum_number_of_surfaces - 1] = bw_int_to_fixed(1);
+	data->access_one_channel_only[maximum_number_of_surfaces - 2] = 0;
+	data->access_one_channel_only[maximum_number_of_surfaces - 1] = 0;
+	/*assume display pipe1 has dwb enabled*/
+	data->h_total[maximum_number_of_surfaces - 2] = data->h_total[5];
+	data->h_total[maximum_number_of_surfaces - 1] = data->h_total[5];
+	data->v_total[maximum_number_of_surfaces - 2] = data->v_total[5];
+	data->v_total[maximum_number_of_surfaces - 1] = data->v_total[5];
+	data->pixel_rate[maximum_number_of_surfaces - 2] = data->pixel_rate[5];
+	data->pixel_rate[maximum_number_of_surfaces - 1] = data->pixel_rate[5];
+	data->src_width[maximum_number_of_surfaces - 2] = data->src_width[5];
+	data->src_width[maximum_number_of_surfaces - 1] = data->src_width[5];
+	data->src_height[maximum_number_of_surfaces - 2] = data->src_height[5];
+	data->src_height[maximum_number_of_surfaces - 1] = data->src_height[5];
+	data->pitch_in_pixels[maximum_number_of_surfaces - 2] = data->src_width[5];
+	data->pitch_in_pixels[maximum_number_of_surfaces - 1] = data->src_width[5];
+	data->h_scale_ratio[maximum_number_of_surfaces - 2] = bw_int_to_fixed(1);
+	data->h_scale_ratio[maximum_number_of_surfaces - 1] = bw_int_to_fixed(1);
+	data->v_scale_ratio[maximum_number_of_surfaces - 2] = bw_int_to_fixed(1);
+	data->v_scale_ratio[maximum_number_of_surfaces - 1] = bw_int_to_fixed(1);
+	data->stereo_mode[maximum_number_of_surfaces - 2] = bw_def_mono;
+	data->stereo_mode[maximum_number_of_surfaces - 1] = bw_def_mono;
+	data->cursor_width_pixels[maximum_number_of_surfaces - 2] = bw_int_to_fixed(0);
+	data->cursor_width_pixels[maximum_number_of_surfaces - 1] = bw_int_to_fixed(0);
+	data->use_alpha[maximum_number_of_surfaces - 2] = 0;
+	data->use_alpha[maximum_number_of_surfaces - 1] = 0;
+	/*mode check calculations:*/
+	/* mode within dce ip capabilities*/
+	/* fbc*/
+	/* hsr*/
+	/* vsr*/
+	/* lb size*/
+	/*effective scaling source and ratios:*/
+	/*for graphics, non-stereo, non-interlace surfaces when the size of the source and destination are the same, only one tap is used*/
+	/*420 chroma has half the width, height, horizontal and vertical scaling ratios than luma*/
+	/*rotating a graphic or underlay surface swaps the width, height, horizontal and vertical scaling ratios*/
+	/*in top-bottom stereo mode there is 2:1 vertical downscaling for each eye*/
+	/*in side-by-side stereo mode there is 2:1 horizontal downscaling for each eye*/
+	/*in interlace mode there is 2:1 vertical downscaling for each field*/
+	/*in panning or bezel adjustment mode the source width has an extra 128 pixels*/
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if (bw_equ(data->h_scale_ratio[i], bw_int_to_fixed(1)) && bw_equ(data->v_scale_ratio[i], bw_int_to_fixed(1)) && surface_type[i] == bw_def_graphics && data->stereo_mode[i] == bw_def_mono && data->interlace_mode[i] == 0) {
+				data->h_taps[i] = bw_int_to_fixed(1);
+				data->v_taps[i] = bw_int_to_fixed(1);
+			}
+			if (surface_type[i] == bw_def_display_write_back420_chroma || surface_type[i] == bw_def_underlay420_chroma) {
+				data->pitch_in_pixels_after_surface_type[i] = bw_div(data->pitch_in_pixels[i], bw_int_to_fixed(2));
+				data->src_width_after_surface_type = bw_div(data->src_width[i], bw_int_to_fixed(2));
+				data->src_height_after_surface_type = bw_div(data->src_height[i], bw_int_to_fixed(2));
+				data->hsr_after_surface_type = bw_div(data->h_scale_ratio[i], bw_int_to_fixed(2));
+				data->vsr_after_surface_type = bw_div(data->v_scale_ratio[i], bw_int_to_fixed(2));
+			}
+			else {
+				data->pitch_in_pixels_after_surface_type[i] = data->pitch_in_pixels[i];
+				data->src_width_after_surface_type = data->src_width[i];
+				data->src_height_after_surface_type = data->src_height[i];
+				data->hsr_after_surface_type = data->h_scale_ratio[i];
+				data->vsr_after_surface_type = data->v_scale_ratio[i];
+			}
+			if ((bw_equ(data->rotation_angle[i], bw_int_to_fixed(90)) || bw_equ(data->rotation_angle[i], bw_int_to_fixed(270))) && surface_type[i] != bw_def_display_write_back420_luma && surface_type[i] != bw_def_display_write_back420_chroma) {
+				data->src_width_after_rotation = data->src_height_after_surface_type;
+				data->src_height_after_rotation = data->src_width_after_surface_type;
+				data->hsr_after_rotation = data->vsr_after_surface_type;
+				data->vsr_after_rotation = data->hsr_after_surface_type;
+			}
+			else {
+				data->src_width_after_rotation = data->src_width_after_surface_type;
+				data->src_height_after_rotation = data->src_height_after_surface_type;
+				data->hsr_after_rotation = data->hsr_after_surface_type;
+				data->vsr_after_rotation = data->vsr_after_surface_type;
+			}
+			switch (data->stereo_mode[i]) {
+			case bw_def_top_bottom:
+				data->source_width_pixels[i] = data->src_width_after_rotation;
+				data->source_height_pixels = bw_mul(bw_int_to_fixed(2), data->src_height_after_rotation);
+				data->hsr_after_stereo = data->hsr_after_rotation;
+				data->vsr_after_stereo = bw_mul(bw_int_to_fixed(1), data->vsr_after_rotation);
+				break;
+			case bw_def_side_by_side:
+				data->source_width_pixels[i] = bw_mul(bw_int_to_fixed(2), data->src_width_after_rotation);
+				data->source_height_pixels = data->src_height_after_rotation;
+				data->hsr_after_stereo = bw_mul(bw_int_to_fixed(1), data->hsr_after_rotation);
+				data->vsr_after_stereo = data->vsr_after_rotation;
+				break;
+			default:
+				data->source_width_pixels[i] = data->src_width_after_rotation;
+				data->source_height_pixels = data->src_height_after_rotation;
+				data->hsr_after_stereo = data->hsr_after_rotation;
+				data->vsr_after_stereo = data->vsr_after_rotation;
+				break;
+			}
+			data->hsr[i] = data->hsr_after_stereo;
+			if (data->interlace_mode[i]) {
+				data->vsr[i] = bw_mul(data->vsr_after_stereo, bw_int_to_fixed(2));
+			}
+			else {
+				data->vsr[i] = data->vsr_after_stereo;
+			}
+			if (data->panning_and_bezel_adjustment != bw_def_none) {
+				data->source_width_rounded_up_to_chunks[i] = bw_add(bw_floor2(bw_sub(data->source_width_pixels[i], bw_int_to_fixed(1)), bw_int_to_fixed(128)), bw_int_to_fixed(256));
+			}
+			else {
+				data->source_width_rounded_up_to_chunks[i] = bw_ceil2(data->source_width_pixels[i], bw_int_to_fixed(128));
+			}
+			data->source_height_rounded_up_to_chunks[i] = data->source_height_pixels;
+		}
+	}
+	/*mode support checks:*/
+	/*the number of graphics and underlay pipes is limited by the ip support*/
+	/*maximum horizontal and vertical scale ratio is 4, and should not exceed the number of taps*/
+	/*for downscaling with the pre-downscaler, the horizontal scale ratio must be more than the ceiling of one quarter of the number of taps*/
+	/*the pre-downscaler reduces the line buffer source by the horizontal scale ratio*/
+	/*the number of lines in the line buffer has to exceed the number of vertical taps*/
+	/*the size of the line in the line buffer is the product of the source width and the bits per component, rounded up to a multiple of 48*/
+	/*the size of the line in the line buffer in the case of 10 bit per component is the product of the source width rounded up to multiple of 8 and 30.023438 / 3, rounded up to a multiple of 48*/
+	/*the size of the line in the line buffer in the case of 8 bit per component is the product of the source width rounded up to multiple of 8 and 30.023438 / 3, rounded up to a multiple of 48*/
+	/*frame buffer compression is not supported with stereo mode, rotation, or non- 888 formats*/
+	/*rotation is not supported with linear of stereo modes*/
+	if (dceip->number_of_graphics_pipes >= data->number_of_displays && dceip->number_of_underlay_pipes >= data->number_of_underlay_surfaces && !(dceip->display_write_back_supported == 0 && data->d1_display_write_back_dwb_enable == 1)) {
+		pipe_check = bw_def_ok;
+	}
+	else {
+		pipe_check = bw_def_notok;
+	}
+	hsr_check = bw_def_ok;
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if (bw_neq(data->hsr[i], bw_int_to_fixed(1))) {
+				if (bw_mtn(data->hsr[i], bw_int_to_fixed(4))) {
+					hsr_check = bw_def_hsr_mtn_4;
+				}
+				else {
+					if (bw_mtn(data->hsr[i], data->h_taps[i])) {
+						hsr_check = bw_def_hsr_mtn_h_taps;
+					}
+					else {
+						if (dceip->pre_downscaler_enabled == 1 && bw_mtn(data->hsr[i], bw_int_to_fixed(1)) && bw_leq(data->hsr[i], bw_ceil2(bw_div(data->h_taps[i], bw_int_to_fixed(4)), bw_int_to_fixed(1)))) {
+							hsr_check = bw_def_ceiling__h_taps_div_4___meq_hsr;
+						}
+					}
+				}
+			}
+		}
+	}
+	vsr_check = bw_def_ok;
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if (bw_neq(data->vsr[i], bw_int_to_fixed(1))) {
+				if (bw_mtn(data->vsr[i], bw_int_to_fixed(4))) {
+					vsr_check = bw_def_vsr_mtn_4;
+				}
+				else {
+					if (bw_mtn(data->vsr[i], data->v_taps[i])) {
+						vsr_check = bw_def_vsr_mtn_v_taps;
+					}
+				}
+			}
+		}
+	}
+	lb_size_check = bw_def_ok;
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if ((dceip->pre_downscaler_enabled && bw_mtn(data->hsr[i], bw_int_to_fixed(1)))) {
+				data->source_width_in_lb = bw_div(data->source_width_pixels[i], data->hsr[i]);
+			}
+			else {
+				data->source_width_in_lb = data->source_width_pixels[i];
+			}
+			switch (data->lb_bpc[i]) {
+			case 8:
+				data->lb_line_pitch = bw_ceil2(bw_mul(bw_div(bw_frc_to_fixed(2401171875ul, 100000000), bw_int_to_fixed(3)), bw_ceil2(data->source_width_in_lb, bw_int_to_fixed(8))), bw_int_to_fixed(48));
+				break;
+			case 10:
+				data->lb_line_pitch = bw_ceil2(bw_mul(bw_div(bw_frc_to_fixed(300234375, 10000000), bw_int_to_fixed(3)), bw_ceil2(data->source_width_in_lb, bw_int_to_fixed(8))), bw_int_to_fixed(48));
+				break;
+			default:
+				data->lb_line_pitch = bw_ceil2(bw_mul(bw_int_to_fixed(data->lb_bpc[i]), data->source_width_in_lb), bw_int_to_fixed(48));
+				break;
+			}
+			data->lb_partitions[i] = bw_floor2(bw_div(data->lb_size_per_component[i], data->lb_line_pitch), bw_int_to_fixed(1));
+			/* clamp the partitions to the maximum number supported by the lb */
+			if ((surface_type[i] != bw_def_graphics || dceip->graphics_lb_nodownscaling_multi_line_prefetching == 1)) {
+				data->lb_partitions_max[i] = bw_int_to_fixed(10);
+			}
+			else {
+				data->lb_partitions_max[i] = bw_int_to_fixed(7);
+			}
+			data->lb_partitions[i] = bw_min2(data->lb_partitions_max[i], data->lb_partitions[i]);
+			if (bw_mtn(bw_add(data->v_taps[i], bw_int_to_fixed(1)), data->lb_partitions[i])) {
+				lb_size_check = bw_def_notok;
+			}
+		}
+	}
+	fbc_check = bw_def_ok;
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i] && data->fbc_en[i] == 1 && (bw_equ(data->rotation_angle[i], bw_int_to_fixed(90)) || bw_equ(data->rotation_angle[i], bw_int_to_fixed(270)) || data->stereo_mode[i] != bw_def_mono || data->bytes_per_pixel[i] != 4)) {
+			fbc_check = bw_def_invalid_rotation_or_bpp_or_stereo;
+		}
+	}
+	rotation_check = bw_def_ok;
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if ((bw_equ(data->rotation_angle[i], bw_int_to_fixed(90)) || bw_equ(data->rotation_angle[i], bw_int_to_fixed(270))) && (tiling_mode[i] == bw_def_linear || data->stereo_mode[i] != bw_def_mono)) {
+				rotation_check = bw_def_invalid_linear_or_stereo_mode;
+			}
+		}
+	}
+	if (pipe_check == bw_def_ok && hsr_check == bw_def_ok && vsr_check == bw_def_ok && lb_size_check == bw_def_ok && fbc_check == bw_def_ok && rotation_check == bw_def_ok) {
+		mode_check = bw_def_ok;
+	}
+	else {
+		mode_check = bw_def_notok;
+	}
+	/*number of memory channels for write-back client*/
+	data->number_of_dram_wrchannels = vbios->number_of_dram_channels;
+	data->number_of_dram_channels = vbios->number_of_dram_channels;
+	/*modify number of memory channels if lpt mode is enabled*/
+	/* low power tiling mode register*/
+	/* 0 = use channel 0*/
+	/* 1 = use channel 0 and 1*/
+	/* 2 = use channel 0,1,2,3*/
+	if ((fbc_enabled == 1 && lpt_enabled == 1)) {
+		if (vbios->memory_type == bw_def_hbm)
+			data->dram_efficiency = bw_frc_to_fixed(5, 10);
+		else
+			data->dram_efficiency = bw_int_to_fixed(1);
+
+
+		if (dceip->low_power_tiling_mode == 0) {
+			data->number_of_dram_channels = 1;
+		}
+		else if (dceip->low_power_tiling_mode == 1) {
+			data->number_of_dram_channels = 2;
+		}
+		else if (dceip->low_power_tiling_mode == 2) {
+			data->number_of_dram_channels = 4;
+		}
+		else {
+			data->number_of_dram_channels = 1;
+		}
+	}
+	else {
+		if (vbios->memory_type == bw_def_hbm)
+			data->dram_efficiency = bw_frc_to_fixed(5, 10);
+		else
+			data->dram_efficiency = bw_frc_to_fixed(8, 10);
+	}
+	/*memory request size and latency hiding:*/
+	/*request size is normally 64 byte, 2-line interleaved, with full latency hiding*/
+	/*the display write-back requests are single line*/
+	/*for tiled graphics surfaces, or undelay surfaces with width higher than the maximum size for full efficiency, request size is 32 byte in 8 and 16 bpp or if the rotation is orthogonal to the tiling grain. only half is useful of the bytes in the request size in 8 bpp or in 32 bpp if the rotation is orthogonal to the tiling grain.*/
+	/*for undelay surfaces with width lower than the maximum size for full efficiency, requests are 4-line interleaved in 16bpp if the rotation is parallel to the tiling grain, and 8-line interleaved with 4-line latency hiding in 8bpp or if the rotation is orthogonal to the tiling grain.*/
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if ((bw_equ(data->rotation_angle[i], bw_int_to_fixed(90)) || bw_equ(data->rotation_angle[i], bw_int_to_fixed(270)))) {
+				if ((i < 4)) {
+					/*underlay portrait tiling mode is not supported*/
+					data->orthogonal_rotation[i] = 1;
+				}
+				else {
+					/*graphics portrait tiling mode*/
+					if (data->graphics_micro_tile_mode == bw_def_rotated_micro_tiling) {
+						data->orthogonal_rotation[i] = 0;
+					}
+					else {
+						data->orthogonal_rotation[i] = 1;
+					}
+				}
+			}
+			else {
+				if ((i < 4)) {
+					/*underlay landscape tiling mode is only supported*/
+					if (data->underlay_micro_tile_mode == bw_def_display_micro_tiling) {
+						data->orthogonal_rotation[i] = 0;
+					}
+					else {
+						data->orthogonal_rotation[i] = 1;
+					}
+				}
+				else {
+					/*graphics landscape tiling mode*/
+					if (data->graphics_micro_tile_mode == bw_def_display_micro_tiling) {
+						data->orthogonal_rotation[i] = 0;
+					}
+					else {
+						data->orthogonal_rotation[i] = 1;
+					}
+				}
+			}
+			if (bw_equ(data->rotation_angle[i], bw_int_to_fixed(90)) || bw_equ(data->rotation_angle[i], bw_int_to_fixed(270))) {
+				data->underlay_maximum_source_efficient_for_tiling = dceip->underlay_maximum_height_efficient_for_tiling;
+			}
+			else {
+				data->underlay_maximum_source_efficient_for_tiling = dceip->underlay_maximum_width_efficient_for_tiling;
+			}
+			if (surface_type[i] == bw_def_display_write_back420_luma || surface_type[i] == bw_def_display_write_back420_chroma) {
+				data->bytes_per_request[i] = bw_int_to_fixed(64);
+				data->useful_bytes_per_request[i] = bw_int_to_fixed(64);
+				data->lines_interleaved_in_mem_access[i] = bw_int_to_fixed(1);
+				data->latency_hiding_lines[i] = bw_int_to_fixed(1);
+			}
+			else if (tiling_mode[i] == bw_def_linear) {
+				data->bytes_per_request[i] = bw_int_to_fixed(64);
+				data->useful_bytes_per_request[i] = bw_int_to_fixed(64);
+				data->lines_interleaved_in_mem_access[i] = bw_int_to_fixed(2);
+				data->latency_hiding_lines[i] = bw_int_to_fixed(2);
+			}
+			else {
+				if (surface_type[i] == bw_def_graphics || (bw_mtn(data->source_width_rounded_up_to_chunks[i], bw_ceil2(data->underlay_maximum_source_efficient_for_tiling, bw_int_to_fixed(256))))) {
+					switch (data->bytes_per_pixel[i]) {
+					case 8:
+						data->lines_interleaved_in_mem_access[i] = bw_int_to_fixed(2);
+						data->latency_hiding_lines[i] = bw_int_to_fixed(2);
+						if (data->orthogonal_rotation[i]) {
+							data->bytes_per_request[i] = bw_int_to_fixed(32);
+							data->useful_bytes_per_request[i] = bw_int_to_fixed(32);
+						}
+						else {
+							data->bytes_per_request[i] = bw_int_to_fixed(64);
+							data->useful_bytes_per_request[i] = bw_int_to_fixed(64);
+						}
+						break;
+					case 4:
+						if (data->orthogonal_rotation[i]) {
+							data->lines_interleaved_in_mem_access[i] = bw_int_to_fixed(2);
+							data->latency_hiding_lines[i] = bw_int_to_fixed(2);
+							data->bytes_per_request[i] = bw_int_to_fixed(32);
+							data->useful_bytes_per_request[i] = bw_int_to_fixed(16);
+						}
+						else {
+							data->lines_interleaved_in_mem_access[i] = bw_int_to_fixed(2);
+							data->latency_hiding_lines[i] = bw_int_to_fixed(2);
+							data->bytes_per_request[i] = bw_int_to_fixed(64);
+							data->useful_bytes_per_request[i] = bw_int_to_fixed(64);
+						}
+						break;
+					case 2:
+						data->lines_interleaved_in_mem_access[i] = bw_int_to_fixed(2);
+						data->latency_hiding_lines[i] = bw_int_to_fixed(2);
+						data->bytes_per_request[i] = bw_int_to_fixed(32);
+						data->useful_bytes_per_request[i] = bw_int_to_fixed(32);
+						break;
+					default:
+						data->lines_interleaved_in_mem_access[i] = bw_int_to_fixed(2);
+						data->latency_hiding_lines[i] = bw_int_to_fixed(2);
+						data->bytes_per_request[i] = bw_int_to_fixed(32);
+						data->useful_bytes_per_request[i] = bw_int_to_fixed(16);
+						break;
+					}
+				}
+				else {
+					data->bytes_per_request[i] = bw_int_to_fixed(64);
+					data->useful_bytes_per_request[i] = bw_int_to_fixed(64);
+					if (data->orthogonal_rotation[i]) {
+						data->lines_interleaved_in_mem_access[i] = bw_int_to_fixed(8);
+						data->latency_hiding_lines[i] = bw_int_to_fixed(4);
+					}
+					else {
+						switch (data->bytes_per_pixel[i]) {
+						case 4:
+							data->lines_interleaved_in_mem_access[i] = bw_int_to_fixed(2);
+							data->latency_hiding_lines[i] = bw_int_to_fixed(2);
+							break;
+						case 2:
+							data->lines_interleaved_in_mem_access[i] = bw_int_to_fixed(4);
+							data->latency_hiding_lines[i] = bw_int_to_fixed(4);
+							break;
+						default:
+							data->lines_interleaved_in_mem_access[i] = bw_int_to_fixed(8);
+							data->latency_hiding_lines[i] = bw_int_to_fixed(4);
+							break;
+						}
+					}
+				}
+			}
+		}
+	}
+	/*requested peak bandwidth:*/
+	/*the peak request-per-second bandwidth is the product of the maximum source lines in per line out in the beginning*/
+	/*and in the middle of the frame, the ratio of the source width to the line time, the ratio of line interleaving*/
+	/*in memory to lines of latency hiding, and the ratio of bytes per pixel to useful bytes per request.*/
+	/**/
+	/*if the dmif data buffer size holds more than vta_ps worth of source lines, then only vsr is used.*/
+	/*the peak bandwidth is the peak request-per-second bandwidth times the request size.*/
+	/**/
+	/*the line buffer lines in per line out in the beginning of the frame is the vertical filter initialization value*/
+	/*rounded up to even and divided by the line times for initialization, which is normally three.*/
+	/*the line buffer lines in per line out in the middle of the frame is at least one, or the vertical scale ratio,*/
+	/*rounded up to line pairs if not doing line buffer prefetching.*/
+	/**/
+	/*the non-prefetching rounding up of the vertical scale ratio can also be done up to 1 (for a 0,2 pattern), 4/3 (for a 0,2,2 pattern),*/
+	/*6/4 (for a 0,2,2,2 pattern), or 3 (for a 2,4 pattern).*/
+	/**/
+	/*the scaler vertical filter initialization value is calculated by the hardware as the floor of the average of the*/
+	/*vertical scale ratio and the number of vertical taps increased by one.  add one more for possible odd line*/
+	/*panning/bezel adjustment mode.*/
+	/**/
+	/*for the bottom interlace field an extra 50% of the vertical scale ratio is considered for this calculation.*/
+	/*in top-bottom stereo mode software has to set the filter initialization value manually and explicitly limit it to 4.*/
+	/*furthermore, there is only one line time for initialization.*/
+	/**/
+	/*line buffer prefetching is done when the number of lines in the line buffer exceeds the number of taps plus*/
+	/*the ceiling of the vertical scale ratio.*/
+	/**/
+	/*multi-line buffer prefetching is only done in the graphics pipe when the scaler is disabled or when upscaling and the vsr <= 0.8.'*/
+	/**/
+	/*the horizontal blank and chunk granularity factor is indirectly used indicate the interval of time required to transfer the source pixels.*/
+	/*the denominator of this term represents the total number of destination output pixels required for the input source pixels.*/
+	/*it applies when the lines in per line out is not 2 or 4.  it does not apply when there is a line buffer between the scl and blnd.*/
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			data->v_filter_init[i] = bw_floor2(bw_div((bw_add(bw_add(bw_add(bw_int_to_fixed(1), data->v_taps[i]), data->vsr[i]), bw_mul(bw_mul(bw_int_to_fixed(data->interlace_mode[i]), bw_frc_to_fixed(5, 10)), data->vsr[i]))), bw_int_to_fixed(2)), bw_int_to_fixed(1));
+			if (data->panning_and_bezel_adjustment == bw_def_any_lines) {
+				data->v_filter_init[i] = bw_add(data->v_filter_init[i], bw_int_to_fixed(1));
+			}
+			if (data->stereo_mode[i] == bw_def_top_bottom) {
+				data->v_filter_init[i] = bw_min2(data->v_filter_init[i], bw_int_to_fixed(4));
+			}
+			if (data->stereo_mode[i] == bw_def_top_bottom) {
+				data->num_lines_at_frame_start = bw_int_to_fixed(1);
+			}
+			else {
+				data->num_lines_at_frame_start = bw_int_to_fixed(3);
+			}
+			if ((bw_mtn(data->vsr[i], bw_int_to_fixed(1)) && surface_type[i] == bw_def_graphics) || data->panning_and_bezel_adjustment == bw_def_any_lines) {
+				data->line_buffer_prefetch[i] = 0;
+			}
+			else if ((((dceip->underlay_downscale_prefetch_enabled == 1 && surface_type[i] != bw_def_graphics) || surface_type[i] == bw_def_graphics) && (bw_mtn(data->lb_partitions[i], bw_add(data->v_taps[i], bw_ceil2(data->vsr[i], bw_int_to_fixed(1))))))) {
+				data->line_buffer_prefetch[i] = 1;
+			}
+			else {
+				data->line_buffer_prefetch[i] = 0;
+			}
+			data->lb_lines_in_per_line_out_in_beginning_of_frame[i] = bw_div(bw_ceil2(data->v_filter_init[i], bw_int_to_fixed(dceip->lines_interleaved_into_lb)), data->num_lines_at_frame_start);
+			if (data->line_buffer_prefetch[i] == 1) {
+				data->lb_lines_in_per_line_out_in_middle_of_frame[i] = bw_max2(bw_int_to_fixed(1), data->vsr[i]);
+			}
+			else if (bw_leq(data->vsr[i], bw_int_to_fixed(1))) {
+				data->lb_lines_in_per_line_out_in_middle_of_frame[i] = bw_int_to_fixed(1);
+			} else if (bw_leq(data->vsr[i],
+					bw_frc_to_fixed(4, 3))) {
+				data->lb_lines_in_per_line_out_in_middle_of_frame[i] = bw_div(bw_int_to_fixed(4), bw_int_to_fixed(3));
+			} else if (bw_leq(data->vsr[i],
+					bw_frc_to_fixed(6, 4))) {
+				data->lb_lines_in_per_line_out_in_middle_of_frame[i] = bw_div(bw_int_to_fixed(6), bw_int_to_fixed(4));
+			}
+			else if (bw_leq(data->vsr[i], bw_int_to_fixed(2))) {
+				data->lb_lines_in_per_line_out_in_middle_of_frame[i] = bw_int_to_fixed(2);
+			}
+			else if (bw_leq(data->vsr[i], bw_int_to_fixed(3))) {
+				data->lb_lines_in_per_line_out_in_middle_of_frame[i] = bw_int_to_fixed(3);
+			}
+			else {
+				data->lb_lines_in_per_line_out_in_middle_of_frame[i] = bw_int_to_fixed(4);
+			}
+			if (data->line_buffer_prefetch[i] == 1 || bw_equ(data->lb_lines_in_per_line_out_in_middle_of_frame[i], bw_int_to_fixed(2)) || bw_equ(data->lb_lines_in_per_line_out_in_middle_of_frame[i], bw_int_to_fixed(4))) {
+				data->horizontal_blank_and_chunk_granularity_factor[i] = bw_int_to_fixed(1);
+			}
+			else {
+				data->horizontal_blank_and_chunk_granularity_factor[i] = bw_div(data->h_total[i], (bw_div((bw_add(data->h_total[i], bw_div((bw_sub(data->source_width_pixels[i], bw_int_to_fixed(dceip->chunk_width))), data->hsr[i]))), bw_int_to_fixed(2))));
+			}
+			data->request_bandwidth[i] = bw_div(bw_mul(bw_div(bw_mul(bw_div(bw_mul(bw_max2(data->lb_lines_in_per_line_out_in_beginning_of_frame[i], data->lb_lines_in_per_line_out_in_middle_of_frame[i]), data->source_width_rounded_up_to_chunks[i]), (bw_div(data->h_total[i], data->pixel_rate[i]))), bw_int_to_fixed(data->bytes_per_pixel[i])), data->useful_bytes_per_request[i]), data->lines_interleaved_in_mem_access[i]), data->latency_hiding_lines[i]);
+			data->display_bandwidth[i] = bw_mul(data->request_bandwidth[i], data->bytes_per_request[i]);
+		}
+	}
+	/*outstanding chunk request limit*/
+	/*if underlay buffer sharing is enabled, the data buffer size for underlay in 422 or 444 is the sum of the luma and chroma data buffer sizes.*/
+	/*underlay buffer sharing mode is only permitted in orthogonal rotation modes.*/
+	/**/
+	/*if there is only one display enabled, the dmif data buffer size for the graphics surface is increased by concatenating the adjacent buffers.*/
+	/**/
+	/*the memory chunk size in bytes is 1024 for the writeback, and 256 times the memory line interleaving and the bytes per pixel for graphics*/
+	/*and underlay.*/
+	/**/
+	/*the pipe chunk size uses 2 for line interleaving, except for the write back, in which case it is 1.*/
+	/*graphics and underlay data buffer size is adjusted (limited) using the outstanding chunk request limit if there is more than one*/
+	/*display enabled or if the dmif request buffer is not large enough for the total data buffer size.*/
+	/*the outstanding chunk request limit is the ceiling of the adjusted data buffer size divided by the chunk size in bytes*/
+	/*the adjusted data buffer size is the product of the display bandwidth and the minimum effective data buffer size in terms of time,*/
+	/*rounded up to the chunk size in bytes, but should not exceed the original data buffer size*/
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if ((dceip->dmif_pipe_en_fbc_chunk_tracker + 3 == i && fbc_enabled == 0 && tiling_mode[i] != bw_def_linear)) {
+				data->max_chunks_non_fbc_mode[i] = 128 - dmif_chunk_buff_margin;
+			}
+			else {
+				data->max_chunks_non_fbc_mode[i] = 16 - dmif_chunk_buff_margin;
+			}
+		}
+		if (data->fbc_en[i] == 1) {
+			max_chunks_fbc_mode = 128 - dmif_chunk_buff_margin;
+		}
+	}
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			switch (surface_type[i]) {
+			case bw_def_display_write_back420_luma:
+				data->data_buffer_size[i] = bw_int_to_fixed(dceip->display_write_back420_luma_mcifwr_buffer_size);
+				break;
+			case bw_def_display_write_back420_chroma:
+				data->data_buffer_size[i] = bw_int_to_fixed(dceip->display_write_back420_chroma_mcifwr_buffer_size);
+				break;
+			case bw_def_underlay420_luma:
+				data->data_buffer_size[i] = bw_int_to_fixed(dceip->underlay_luma_dmif_size);
+				break;
+			case bw_def_underlay420_chroma:
+				data->data_buffer_size[i] = bw_div(bw_int_to_fixed(dceip->underlay_chroma_dmif_size), bw_int_to_fixed(2));
+				break;
+			case bw_def_underlay422:case bw_def_underlay444:
+				if (data->orthogonal_rotation[i] == 0) {
+					data->data_buffer_size[i] = bw_int_to_fixed(dceip->underlay_luma_dmif_size);
+				}
+				else {
+					data->data_buffer_size[i] = bw_add(bw_int_to_fixed(dceip->underlay_luma_dmif_size), bw_int_to_fixed(dceip->underlay_chroma_dmif_size));
+				}
+				break;
+			default:
+				if (data->fbc_en[i] == 1) {
+					/*data_buffer_size(i) = max_dmif_buffer_allocated * graphics_dmif_size*/
+					if (data->number_of_displays == 1) {
+						data->data_buffer_size[i] = bw_min2(bw_mul(bw_mul(bw_int_to_fixed(max_chunks_fbc_mode), bw_int_to_fixed(pixels_per_chunk)), bw_int_to_fixed(data->bytes_per_pixel[i])), bw_mul(bw_int_to_fixed(dceip->max_dmif_buffer_allocated), bw_int_to_fixed(dceip->graphics_dmif_size)));
+					}
+					else {
+						data->data_buffer_size[i] = bw_min2(bw_mul(bw_mul(bw_int_to_fixed(max_chunks_fbc_mode), bw_int_to_fixed(pixels_per_chunk)), bw_int_to_fixed(data->bytes_per_pixel[i])), bw_int_to_fixed(dceip->graphics_dmif_size));
+					}
+				}
+				else {
+					/*the effective dmif buffer size in non-fbc mode is limited by the 16 entry chunk tracker*/
+					if (data->number_of_displays == 1) {
+						data->data_buffer_size[i] = bw_min2(bw_mul(bw_mul(bw_int_to_fixed(data->max_chunks_non_fbc_mode[i]), bw_int_to_fixed(pixels_per_chunk)), bw_int_to_fixed(data->bytes_per_pixel[i])), bw_mul(bw_int_to_fixed(dceip->max_dmif_buffer_allocated), bw_int_to_fixed(dceip->graphics_dmif_size)));
+					}
+					else {
+						data->data_buffer_size[i] = bw_min2(bw_mul(bw_mul(bw_int_to_fixed(data->max_chunks_non_fbc_mode[i]), bw_int_to_fixed(pixels_per_chunk)), bw_int_to_fixed(data->bytes_per_pixel[i])), bw_int_to_fixed(dceip->graphics_dmif_size));
+					}
+				}
+				break;
+			}
+			if (surface_type[i] == bw_def_display_write_back420_luma || surface_type[i] == bw_def_display_write_back420_chroma) {
+				data->memory_chunk_size_in_bytes[i] = bw_int_to_fixed(1024);
+				data->pipe_chunk_size_in_bytes[i] = bw_int_to_fixed(1024);
+			}
+			else {
+				data->memory_chunk_size_in_bytes[i] = bw_mul(bw_mul(bw_int_to_fixed(dceip->chunk_width), data->lines_interleaved_in_mem_access[i]), bw_int_to_fixed(data->bytes_per_pixel[i]));
+				data->pipe_chunk_size_in_bytes[i] = bw_mul(bw_mul(bw_int_to_fixed(dceip->chunk_width), bw_int_to_fixed(dceip->lines_interleaved_into_lb)), bw_int_to_fixed(data->bytes_per_pixel[i]));
+			}
+		}
+	}
+	data->min_dmif_size_in_time = bw_int_to_fixed(9999);
+	data->min_mcifwr_size_in_time = bw_int_to_fixed(9999);
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if (surface_type[i] != bw_def_display_write_back420_luma && surface_type[i] != bw_def_display_write_back420_chroma) {
+				if (bw_ltn(bw_div(bw_div(bw_mul(data->data_buffer_size[i], data->bytes_per_request[i]), data->useful_bytes_per_request[i]), data->display_bandwidth[i]), data->min_dmif_size_in_time)) {
+					data->min_dmif_size_in_time = bw_div(bw_div(bw_mul(data->data_buffer_size[i], data->bytes_per_request[i]), data->useful_bytes_per_request[i]), data->display_bandwidth[i]);
+				}
+			}
+			else {
+				if (bw_ltn(bw_div(bw_div(bw_mul(data->data_buffer_size[i], data->bytes_per_request[i]), data->useful_bytes_per_request[i]), data->display_bandwidth[i]), data->min_mcifwr_size_in_time)) {
+					data->min_mcifwr_size_in_time = bw_div(bw_div(bw_mul(data->data_buffer_size[i], data->bytes_per_request[i]), data->useful_bytes_per_request[i]), data->display_bandwidth[i]);
+				}
+			}
+		}
+	}
+	data->total_requests_for_dmif_size = bw_int_to_fixed(0);
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i] && surface_type[i] != bw_def_display_write_back420_luma && surface_type[i] != bw_def_display_write_back420_chroma) {
+			data->total_requests_for_dmif_size = bw_add(data->total_requests_for_dmif_size, bw_div(data->data_buffer_size[i], data->useful_bytes_per_request[i]));
+		}
+	}
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if (surface_type[i] != bw_def_display_write_back420_luma && surface_type[i] != bw_def_display_write_back420_chroma && dceip->limit_excessive_outstanding_dmif_requests && (data->number_of_displays > 1 || bw_mtn(data->total_requests_for_dmif_size, dceip->dmif_request_buffer_size))) {
+				data->adjusted_data_buffer_size[i] = bw_min2(data->data_buffer_size[i], bw_ceil2(bw_mul(data->min_dmif_size_in_time, data->display_bandwidth[i]), data->memory_chunk_size_in_bytes[i]));
+			}
+			else {
+				data->adjusted_data_buffer_size[i] = data->data_buffer_size[i];
+			}
+		}
+	}
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if (data->number_of_displays == 1 && data->number_of_underlay_surfaces == 0) {
+				/*set maximum chunk limit if only one graphic pipe is enabled*/
+				data->outstanding_chunk_request_limit[i] = bw_int_to_fixed(127);
+			}
+			else {
+				data->outstanding_chunk_request_limit[i] = bw_ceil2(bw_div(data->adjusted_data_buffer_size[i], data->pipe_chunk_size_in_bytes[i]), bw_int_to_fixed(1));
+				/*clamp maximum chunk limit in the graphic display pipe*/
+				if (i >= 4) {
+					data->outstanding_chunk_request_limit[i] = bw_max2(bw_int_to_fixed(127), data->outstanding_chunk_request_limit[i]);
+				}
+			}
+		}
+	}
+	/*outstanding pte request limit*/
+	/*in tiling mode with no rotation the sg pte requests are 8 useful pt_es, the sg row height is the page height and the sg page width x height is 64x64 for 8bpp, 64x32 for 16 bpp, 32x32 for 32 bpp*/
+	/*in tiling mode with rotation the sg pte requests are only one useful pte, and the sg row height is also the page height, but the sg page width and height are swapped*/
+	/*in linear mode the pte requests are 8 useful pt_es, the sg page width is 4096 divided by the bytes per pixel, the sg page height is 1, but there is just one row whose height is the lines of pte prefetching*/
+	/*the outstanding pte request limit is obtained by multiplying the outstanding chunk request limit by the peak pte request to eviction limiting ratio, rounding up to integer, multiplying by the pte requests per chunk, and rounding up to integer again*/
+	/*if not using peak pte request to eviction limiting, the outstanding pte request limit is the pte requests in the vblank*/
+	/*the pte requests in the vblank is the product of the number of pte request rows times the number of pte requests in a row*/
+	/*the number of pte requests in a row is the quotient of the source width divided by 256, multiplied by the pte requests per chunk, rounded up to even, multiplied by the scatter-gather row height and divided by the scatter-gather page height*/
+	/*the pte requests per chunk is 256 divided by the scatter-gather page width and the useful pt_es per pte request*/
+	if (data->number_of_displays > 1 || (bw_neq(data->rotation_angle[4], bw_int_to_fixed(0)) && bw_neq(data->rotation_angle[4], bw_int_to_fixed(180)))) {
+		data->peak_pte_request_to_eviction_ratio_limiting = dceip->peak_pte_request_to_eviction_ratio_limiting_multiple_displays_or_single_rotated_display;
+	}
+	else {
+		data->peak_pte_request_to_eviction_ratio_limiting = dceip->peak_pte_request_to_eviction_ratio_limiting_single_display_no_rotation;
+	}
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i] && data->scatter_gather_enable_for_pipe[i] == 1) {
+			if (tiling_mode[i] == bw_def_linear) {
+				data->useful_pte_per_pte_request = bw_int_to_fixed(8);
+				data->scatter_gather_page_width[i] = bw_div(bw_int_to_fixed(4096), bw_int_to_fixed(data->bytes_per_pixel[i]));
+				data->scatter_gather_page_height[i] = bw_int_to_fixed(1);
+				data->scatter_gather_pte_request_rows = bw_int_to_fixed(1);
+				data->scatter_gather_row_height = bw_int_to_fixed(dceip->scatter_gather_lines_of_pte_prefetching_in_linear_mode);
+			}
+			else if (bw_equ(data->rotation_angle[i], bw_int_to_fixed(0)) || bw_equ(data->rotation_angle[i], bw_int_to_fixed(180))) {
+				data->useful_pte_per_pte_request = bw_int_to_fixed(8);
+				switch (data->bytes_per_pixel[i]) {
+				case 4:
+					data->scatter_gather_page_width[i] = bw_int_to_fixed(32);
+					data->scatter_gather_page_height[i] = bw_int_to_fixed(32);
+					break;
+				case 2:
+					data->scatter_gather_page_width[i] = bw_int_to_fixed(64);
+					data->scatter_gather_page_height[i] = bw_int_to_fixed(32);
+					break;
+				default:
+					data->scatter_gather_page_width[i] = bw_int_to_fixed(64);
+					data->scatter_gather_page_height[i] = bw_int_to_fixed(64);
+					break;
+				}
+				data->scatter_gather_pte_request_rows = bw_int_to_fixed(dceip->scatter_gather_pte_request_rows_in_tiling_mode);
+				data->scatter_gather_row_height = data->scatter_gather_page_height[i];
+			}
+			else {
+				data->useful_pte_per_pte_request = bw_int_to_fixed(1);
+				switch (data->bytes_per_pixel[i]) {
+				case 4:
+					data->scatter_gather_page_width[i] = bw_int_to_fixed(32);
+					data->scatter_gather_page_height[i] = bw_int_to_fixed(32);
+					break;
+				case 2:
+					data->scatter_gather_page_width[i] = bw_int_to_fixed(32);
+					data->scatter_gather_page_height[i] = bw_int_to_fixed(64);
+					break;
+				default:
+					data->scatter_gather_page_width[i] = bw_int_to_fixed(64);
+					data->scatter_gather_page_height[i] = bw_int_to_fixed(64);
+					break;
+				}
+				data->scatter_gather_pte_request_rows = bw_int_to_fixed(dceip->scatter_gather_pte_request_rows_in_tiling_mode);
+				data->scatter_gather_row_height = data->scatter_gather_page_height[i];
+			}
+			data->pte_request_per_chunk[i] = bw_div(bw_div(bw_int_to_fixed(dceip->chunk_width), data->scatter_gather_page_width[i]), data->useful_pte_per_pte_request);
+			data->scatter_gather_pte_requests_in_row[i] = bw_div(bw_mul(bw_ceil2(bw_mul(bw_div(data->source_width_rounded_up_to_chunks[i], bw_int_to_fixed(dceip->chunk_width)), data->pte_request_per_chunk[i]), bw_int_to_fixed(1)), data->scatter_gather_row_height), data->scatter_gather_page_height[i]);
+			data->scatter_gather_pte_requests_in_vblank = bw_mul(data->scatter_gather_pte_request_rows, data->scatter_gather_pte_requests_in_row[i]);
+			if (bw_equ(data->peak_pte_request_to_eviction_ratio_limiting, bw_int_to_fixed(0))) {
+				data->scatter_gather_pte_request_limit[i] = data->scatter_gather_pte_requests_in_vblank;
+			}
+			else {
+				data->scatter_gather_pte_request_limit[i] = bw_max2(dceip->minimum_outstanding_pte_request_limit, bw_min2(data->scatter_gather_pte_requests_in_vblank, bw_ceil2(bw_mul(bw_mul(bw_div(bw_ceil2(data->adjusted_data_buffer_size[i], data->memory_chunk_size_in_bytes[i]), data->memory_chunk_size_in_bytes[i]), data->pte_request_per_chunk[i]), data->peak_pte_request_to_eviction_ratio_limiting), bw_int_to_fixed(1))));
+			}
+		}
+	}
+	/*pitch padding recommended for efficiency in linear mode*/
+	/*in linear mode graphics or underlay with scatter gather, a pitch that is a multiple of the channel interleave (256 bytes) times the channel-bank rotation is not efficient*/
+	/*if that is the case it is recommended to pad the pitch by at least 256 pixels*/
+	data->inefficient_linear_pitch_in_bytes = bw_mul(bw_mul(bw_int_to_fixed(256), bw_int_to_fixed(vbios->number_of_dram_banks)), bw_int_to_fixed(data->number_of_dram_channels));
+
+	/*pixel transfer time*/
+	/*the dmif and mcifwr yclk(pclk) required is the one that allows the transfer of all pipe's data buffer size in memory in the time for data transfer*/
+	/*for dmif, pte and cursor requests have to be included.*/
+	/*the dram data requirement is doubled when the data request size in bytes is less than the dram channel width times the burst size (8)*/
+	/*the dram data requirement is also multiplied by the number of channels in the case of low power tiling*/
+	/*the page close-open time is determined by trc and the number of page close-opens*/
+	/*in tiled mode graphics or underlay with scatter-gather enabled the bytes per page close-open is the product of the memory line interleave times the maximum of the scatter-gather page width and the product of the tile width (8 pixels) times the number of channels times the number of banks.*/
+	/*in linear mode graphics or underlay with scatter-gather enabled and inefficient pitch, the bytes per page close-open is the line request alternation slice, because different lines are in completely different 4k address bases.*/
+	/*otherwise, the bytes page close-open is the chunk size because that is the arbitration slice.*/
+	/*pte requests are grouped by pte requests per chunk if that is more than 1. each group costs a page close-open time for dmif reads*/
+	/*cursor requests outstanding are limited to a group of two source lines. each group costs a page close-open time for dmif reads*/
+	/*the display reads and writes time for data transfer is the minimum data or cursor buffer size in time minus the mc urgent latency*/
+	/*the mc urgent latency is experienced more than one time if the number of dmif requests in the data buffer exceeds the request buffer size plus the request slots reserved for dmif in the dram channel arbiter queues*/
+	/*the dispclk required is the maximum for all surfaces of the maximum of the source pixels for first output pixel times the throughput factor, divided by the pixels per dispclk, and divided by the minimum latency hiding minus the dram speed/p-state change latency minus the burst time, and the source pixels for last output pixel, times the throughput factor, divided by the pixels per dispclk, and divided by the minimum latency hiding minus the dram speed/p-state change latency minus the burst time, plus the active time.*/
+	/*the data burst time is the maximum of the total page close-open time, total dmif/mcifwr buffer size in memory divided by the dram bandwidth, and the total dmif/mcifwr buffer size in memory divided by the 32 byte sclk data bus bandwidth, each multiplied by its efficiency.*/
+	/*the source line transfer time is the maximum for all surfaces of the maximum of the burst time plus the urgent latency times the floor of the data required divided by the buffer size for the fist pixel, and the burst time plus the urgent latency times the floor of the data required divided by the buffer size for the last pixel plus the active time.*/
+	/*the source pixels for the first output pixel is 512 if the scaler vertical filter initialization value is greater than 2, and it is 4 times the source width if it is greater than 4.*/
+	/*the source pixels for the last output pixel is the source width times the scaler vertical filter initialization value rounded up to even*/
+	/*the source data for these pixels is the number of pixels times the bytes per pixel times the bytes per request divided by the useful bytes per request.*/
+	data->cursor_total_data = bw_int_to_fixed(0);
+	data->cursor_total_request_groups = bw_int_to_fixed(0);
+	data->scatter_gather_total_pte_requests = bw_int_to_fixed(0);
+	data->scatter_gather_total_pte_request_groups = bw_int_to_fixed(0);
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			data->cursor_total_data = bw_add(data->cursor_total_data, bw_mul(bw_mul(bw_int_to_fixed(2), data->cursor_width_pixels[i]), bw_int_to_fixed(4)));
+			if (dceip->large_cursor == 1) {
+				data->cursor_total_request_groups = bw_add(data->cursor_total_request_groups, bw_int_to_fixed((dceip->cursor_max_outstanding_group_num + 1)));
+			}
+			else {
+				data->cursor_total_request_groups = bw_add(data->cursor_total_request_groups, bw_ceil2(bw_div(data->cursor_width_pixels[i], dceip->cursor_chunk_width), bw_int_to_fixed(1)));
+			}
+			if (data->scatter_gather_enable_for_pipe[i]) {
+				data->scatter_gather_total_pte_requests = bw_add(data->scatter_gather_total_pte_requests, data->scatter_gather_pte_request_limit[i]);
+				data->scatter_gather_total_pte_request_groups = bw_add(data->scatter_gather_total_pte_request_groups, bw_ceil2(bw_div(data->scatter_gather_pte_request_limit[i], bw_ceil2(data->pte_request_per_chunk[i], bw_int_to_fixed(1))), bw_int_to_fixed(1)));
+			}
+		}
+	}
+	data->tile_width_in_pixels = bw_int_to_fixed(8);
+	data->dmif_total_number_of_data_request_page_close_open = bw_int_to_fixed(0);
+	data->mcifwr_total_number_of_data_request_page_close_open = bw_int_to_fixed(0);
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if (data->scatter_gather_enable_for_pipe[i] == 1 && tiling_mode[i] != bw_def_linear) {
+				data->bytes_per_page_close_open = bw_mul(data->lines_interleaved_in_mem_access[i], bw_max2(bw_mul(bw_mul(bw_mul(bw_int_to_fixed(data->bytes_per_pixel[i]), data->tile_width_in_pixels), bw_int_to_fixed(vbios->number_of_dram_banks)), bw_int_to_fixed(data->number_of_dram_channels)), bw_mul(bw_int_to_fixed(data->bytes_per_pixel[i]), data->scatter_gather_page_width[i])));
+			}
+			else if (data->scatter_gather_enable_for_pipe[i] == 1 && tiling_mode[i] == bw_def_linear && bw_equ(bw_mod((bw_mul(data->pitch_in_pixels_after_surface_type[i], bw_int_to_fixed(data->bytes_per_pixel[i]))), data->inefficient_linear_pitch_in_bytes), bw_int_to_fixed(0))) {
+				data->bytes_per_page_close_open = dceip->linear_mode_line_request_alternation_slice;
+			}
+			else {
+				data->bytes_per_page_close_open = data->memory_chunk_size_in_bytes[i];
+			}
+			if (surface_type[i] != bw_def_display_write_back420_luma && surface_type[i] != bw_def_display_write_back420_chroma) {
+				data->dmif_total_number_of_data_request_page_close_open = bw_add(data->dmif_total_number_of_data_request_page_close_open, bw_div(bw_ceil2(data->adjusted_data_buffer_size[i], data->memory_chunk_size_in_bytes[i]), data->bytes_per_page_close_open));
+			}
+			else {
+				data->mcifwr_total_number_of_data_request_page_close_open = bw_add(data->mcifwr_total_number_of_data_request_page_close_open, bw_div(bw_ceil2(data->adjusted_data_buffer_size[i], data->memory_chunk_size_in_bytes[i]), data->bytes_per_page_close_open));
+			}
+		}
+	}
+	data->dmif_total_page_close_open_time = bw_div(bw_mul((bw_add(bw_add(data->dmif_total_number_of_data_request_page_close_open, data->scatter_gather_total_pte_request_groups), data->cursor_total_request_groups)), vbios->trc), bw_int_to_fixed(1000));
+	data->mcifwr_total_page_close_open_time = bw_div(bw_mul(data->mcifwr_total_number_of_data_request_page_close_open, vbios->trc), bw_int_to_fixed(1000));
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			data->adjusted_data_buffer_size_in_memory[i] = bw_div(bw_mul(data->adjusted_data_buffer_size[i], data->bytes_per_request[i]), data->useful_bytes_per_request[i]);
+		}
+	}
+	data->total_requests_for_adjusted_dmif_size = bw_int_to_fixed(0);
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if (surface_type[i] != bw_def_display_write_back420_luma && surface_type[i] != bw_def_display_write_back420_chroma) {
+				data->total_requests_for_adjusted_dmif_size = bw_add(data->total_requests_for_adjusted_dmif_size, bw_div(data->adjusted_data_buffer_size[i], data->useful_bytes_per_request[i]));
+			}
+		}
+	}
+	data->total_dmifmc_urgent_trips = bw_ceil2(bw_div(data->total_requests_for_adjusted_dmif_size, (bw_add(dceip->dmif_request_buffer_size, bw_int_to_fixed((uint64_t)vbios->number_of_request_slots_gmc_reserves_for_dmif_per_channel * data->number_of_dram_channels)))), bw_int_to_fixed(1));
+	data->total_dmifmc_urgent_latency = bw_mul(vbios->dmifmc_urgent_latency, data->total_dmifmc_urgent_trips);
+	data->total_display_reads_required_data = bw_int_to_fixed(0);
+	data->total_display_reads_required_dram_access_data = bw_int_to_fixed(0);
+	data->total_display_writes_required_data = bw_int_to_fixed(0);
+	data->total_display_writes_required_dram_access_data = bw_int_to_fixed(0);
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if (surface_type[i] != bw_def_display_write_back420_luma && surface_type[i] != bw_def_display_write_back420_chroma) {
+				data->display_reads_required_data = data->adjusted_data_buffer_size_in_memory[i];
+				/*for hbm memories, each channel is split into 2 pseudo-channels that are each 64 bits in width.  each*/
+				/*pseudo-channel may be read independently of one another.*/
+				/*the read burst length (bl) for hbm memories is 4, so each read command will access 32 bytes of data.*/
+				/*the 64 or 32 byte sized data is stored in one pseudo-channel.*/
+				/*it will take 4 memclk cycles or 8 yclk cycles to fetch 64 bytes of data from the hbm memory (2 read commands).*/
+				/*it will take 2 memclk cycles or 4 yclk cycles to fetch 32 bytes of data from the hbm memory (1 read command).*/
+				/*for gddr5/ddr4 memories, there is additional overhead if the size of the request is smaller than 64 bytes.*/
+				/*the read burst length (bl) for gddr5/ddr4 memories is 8, regardless of the size of the data request.*/
+				/*therefore it will require 8 cycles to fetch 64 or 32 bytes of data from the memory.*/
+				/*the memory efficiency will be 50% for the 32 byte sized data.*/
+				if (vbios->memory_type == bw_def_hbm) {
+					data->display_reads_required_dram_access_data = data->adjusted_data_buffer_size_in_memory[i];
+				}
+				else {
+					data->display_reads_required_dram_access_data = bw_mul(data->adjusted_data_buffer_size_in_memory[i], bw_ceil2(bw_div(bw_int_to_fixed((8 * vbios->dram_channel_width_in_bits / 8)), data->bytes_per_request[i]), bw_int_to_fixed(1)));
+				}
+				data->total_display_reads_required_data = bw_add(data->total_display_reads_required_data, data->display_reads_required_data);
+				data->total_display_reads_required_dram_access_data = bw_add(data->total_display_reads_required_dram_access_data, data->display_reads_required_dram_access_data);
+			}
+			else {
+				data->total_display_writes_required_data = bw_add(data->total_display_writes_required_data, data->adjusted_data_buffer_size_in_memory[i]);
+				data->total_display_writes_required_dram_access_data = bw_add(data->total_display_writes_required_dram_access_data, bw_mul(data->adjusted_data_buffer_size_in_memory[i], bw_ceil2(bw_div(bw_int_to_fixed(vbios->dram_channel_width_in_bits), data->bytes_per_request[i]), bw_int_to_fixed(1))));
+			}
+		}
+	}
+	data->total_display_reads_required_data = bw_add(bw_add(data->total_display_reads_required_data, data->cursor_total_data), bw_mul(data->scatter_gather_total_pte_requests, bw_int_to_fixed(64)));
+	data->total_display_reads_required_dram_access_data = bw_add(bw_add(data->total_display_reads_required_dram_access_data, data->cursor_total_data), bw_mul(data->scatter_gather_total_pte_requests, bw_int_to_fixed(64)));
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if (bw_mtn(data->v_filter_init[i], bw_int_to_fixed(4))) {
+				data->src_pixels_for_first_output_pixel[i] = bw_mul(bw_int_to_fixed(4), data->source_width_rounded_up_to_chunks[i]);
+			}
+			else {
+				if (bw_mtn(data->v_filter_init[i], bw_int_to_fixed(2))) {
+					data->src_pixels_for_first_output_pixel[i] = bw_int_to_fixed(512);
+				}
+				else {
+					data->src_pixels_for_first_output_pixel[i] = bw_int_to_fixed(0);
+				}
+			}
+			data->src_data_for_first_output_pixel[i] = bw_div(bw_mul(bw_mul(data->src_pixels_for_first_output_pixel[i], bw_int_to_fixed(data->bytes_per_pixel[i])), data->bytes_per_request[i]), data->useful_bytes_per_request[i]);
+			data->src_pixels_for_last_output_pixel[i] = bw_mul(data->source_width_rounded_up_to_chunks[i], bw_max2(bw_ceil2(data->v_filter_init[i], bw_int_to_fixed(dceip->lines_interleaved_into_lb)), bw_mul(bw_ceil2(data->vsr[i], bw_int_to_fixed(dceip->lines_interleaved_into_lb)), data->horizontal_blank_and_chunk_granularity_factor[i])));
+			data->src_data_for_last_output_pixel[i] = bw_div(bw_mul(bw_mul(bw_mul(data->source_width_rounded_up_to_chunks[i], bw_max2(bw_ceil2(data->v_filter_init[i], bw_int_to_fixed(dceip->lines_interleaved_into_lb)), data->lines_interleaved_in_mem_access[i])), bw_int_to_fixed(data->bytes_per_pixel[i])), data->bytes_per_request[i]), data->useful_bytes_per_request[i]);
+			data->active_time[i] = bw_div(bw_div(data->source_width_rounded_up_to_chunks[i], data->hsr[i]), data->pixel_rate[i]);
+		}
+	}
+	for (i = 0; i <= 2; i++) {
+		for (j = 0; j <= 7; j++) {
+			data->dmif_burst_time[i][j] = bw_max3(data->dmif_total_page_close_open_time, bw_div(data->total_display_reads_required_dram_access_data, (bw_mul(bw_div(bw_mul(bw_mul(data->dram_efficiency, yclk[i]), bw_int_to_fixed(vbios->dram_channel_width_in_bits)), bw_int_to_fixed(8)), bw_int_to_fixed(data->number_of_dram_channels)))), bw_div(data->total_display_reads_required_data, (bw_mul(bw_mul(sclk[j], vbios->data_return_bus_width), bw_frc_to_fixed(dceip->percent_of_ideal_port_bw_received_after_urgent_latency, 100)))));
+			if (data->d1_display_write_back_dwb_enable == 1) {
+				data->mcifwr_burst_time[i][j] = bw_max3(data->mcifwr_total_page_close_open_time, bw_div(data->total_display_writes_required_dram_access_data, (bw_mul(bw_div(bw_mul(bw_mul(data->dram_efficiency, yclk[i]), bw_int_to_fixed(vbios->dram_channel_width_in_bits)), bw_int_to_fixed(8)), bw_int_to_fixed(data->number_of_dram_wrchannels)))), bw_div(data->total_display_writes_required_data, (bw_mul(sclk[j], vbios->data_return_bus_width))));
+			}
+		}
+	}
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		for (j = 0; j <= 2; j++) {
+			for (k = 0; k <= 7; k++) {
+				if (data->enable[i]) {
+					if (surface_type[i] != bw_def_display_write_back420_luma && surface_type[i] != bw_def_display_write_back420_chroma) {
+						/*time to transfer data from the dmif buffer to the lb.  since the mc to dmif transfer time overlaps*/
+						/*with the dmif to lb transfer time, only time to transfer the last chunk  is considered.*/
+						data->dmif_buffer_transfer_time[i] = bw_mul(data->source_width_rounded_up_to_chunks[i], (bw_div(dceip->lb_write_pixels_per_dispclk, (bw_div(vbios->low_voltage_max_dispclk, dceip->display_pipe_throughput_factor)))));
+						data->line_source_transfer_time[i][j][k] = bw_max2(bw_mul((bw_add(data->total_dmifmc_urgent_latency, data->dmif_burst_time[j][k])), bw_floor2(bw_div(data->src_data_for_first_output_pixel[i], data->adjusted_data_buffer_size_in_memory[i]), bw_int_to_fixed(1))), bw_sub(bw_add(bw_mul((bw_add(data->total_dmifmc_urgent_latency, data->dmif_burst_time[j][k])), bw_floor2(bw_div(data->src_data_for_last_output_pixel[i], data->adjusted_data_buffer_size_in_memory[i]), bw_int_to_fixed(1))), data->dmif_buffer_transfer_time[i]), data->active_time[i]));
+						/*during an mclk switch the requests from the dce ip are stored in the gmc/arb.  these requests should be serviced immediately*/
+						/*after the mclk switch sequence and not incur an urgent latency penalty.  it is assumed that the gmc/arb can hold up to 256 requests*/
+						/*per memory channel.  if the dce ip is urgent after the mclk switch sequence, all pending requests and subsequent requests should be*/
+						/*immediately serviced without a gap in the urgent requests.*/
+						/*the latency incurred would be the time to issue the requests and return the data for the first or last output pixel.*/
+						if (surface_type[i] == bw_def_graphics) {
+							switch (data->lb_bpc[i]) {
+							case 6:
+								data->v_scaler_efficiency = dceip->graphics_vscaler_efficiency6_bit_per_component;
+								break;
+							case 8:
+								data->v_scaler_efficiency = dceip->graphics_vscaler_efficiency8_bit_per_component;
+								break;
+							case 10:
+								data->v_scaler_efficiency = dceip->graphics_vscaler_efficiency10_bit_per_component;
+								break;
+							default:
+								data->v_scaler_efficiency = dceip->graphics_vscaler_efficiency12_bit_per_component;
+								break;
+							}
+							if (data->use_alpha[i] == 1) {
+								data->v_scaler_efficiency = bw_min2(data->v_scaler_efficiency, dceip->alpha_vscaler_efficiency);
+							}
+						}
+						else {
+							switch (data->lb_bpc[i]) {
+							case 6:
+								data->v_scaler_efficiency = dceip->underlay_vscaler_efficiency6_bit_per_component;
+								break;
+							case 8:
+								data->v_scaler_efficiency = dceip->underlay_vscaler_efficiency8_bit_per_component;
+								break;
+							case 10:
+								data->v_scaler_efficiency = dceip->underlay_vscaler_efficiency10_bit_per_component;
+								break;
+							default:
+								data->v_scaler_efficiency = bw_int_to_fixed(3);
+								break;
+							}
+						}
+						if (dceip->pre_downscaler_enabled && bw_mtn(data->hsr[i], bw_int_to_fixed(1))) {
+							data->scaler_limits_factor = bw_max2(bw_div(data->v_taps[i], data->v_scaler_efficiency), bw_div(data->source_width_rounded_up_to_chunks[i], data->h_total[i]));
+						}
+						else {
+							data->scaler_limits_factor = bw_max3(bw_int_to_fixed(1), bw_ceil2(bw_div(data->h_taps[i], bw_int_to_fixed(4)), bw_int_to_fixed(1)), bw_mul(data->hsr[i], bw_max2(bw_div(data->v_taps[i], data->v_scaler_efficiency), bw_int_to_fixed(1))));
+						}
+						data->dram_speed_change_line_source_transfer_time[i][j][k] = bw_mul(bw_int_to_fixed(2), bw_max2((bw_add((bw_div(data->src_data_for_first_output_pixel[i], bw_min2(bw_mul(data->bytes_per_request[i], sclk[k]), bw_div(bw_mul(bw_mul(data->bytes_per_request[i], data->pixel_rate[i]), data->scaler_limits_factor), bw_int_to_fixed(2))))), (bw_mul(data->dmif_burst_time[j][k], bw_floor2(bw_div(data->src_data_for_first_output_pixel[i], data->adjusted_data_buffer_size_in_memory[i]), bw_int_to_fixed(1)))))), (bw_add((bw_div(data->src_data_for_last_output_pixel[i], bw_min2(bw_mul(data->bytes_per_request[i], sclk[k]), bw_div(bw_mul(bw_mul(data->bytes_per_request[i], data->pixel_rate[i]), data->scaler_limits_factor), bw_int_to_fixed(2))))), (bw_sub(bw_mul(data->dmif_burst_time[j][k], bw_floor2(bw_div(data->src_data_for_last_output_pixel[i], data->adjusted_data_buffer_size_in_memory[i]), bw_int_to_fixed(1))), data->active_time[i]))))));
+					}
+					else {
+						data->line_source_transfer_time[i][j][k] = bw_max2(bw_mul((bw_add(vbios->mcifwrmc_urgent_latency, data->mcifwr_burst_time[j][k])), bw_floor2(bw_div(data->src_data_for_first_output_pixel[i], data->adjusted_data_buffer_size_in_memory[i]), bw_int_to_fixed(1))), bw_sub(bw_mul((bw_add(vbios->mcifwrmc_urgent_latency, data->mcifwr_burst_time[j][k])), bw_floor2(bw_div(data->src_data_for_last_output_pixel[i], data->adjusted_data_buffer_size_in_memory[i]), bw_int_to_fixed(1))), data->active_time[i]));
+						/*during an mclk switch the requests from the dce ip are stored in the gmc/arb.  these requests should be serviced immediately*/
+						/*after the mclk switch sequence and not incur an urgent latency penalty.  it is assumed that the gmc/arb can hold up to 256 requests*/
+						/*per memory channel.  if the dce ip is urgent after the mclk switch sequence, all pending requests and subsequent requests should be*/
+						/*immediately serviced without a gap in the urgent requests.*/
+						/*the latency incurred would be the time to issue the requests and return the data for the first or last output pixel.*/
+						data->dram_speed_change_line_source_transfer_time[i][j][k] = bw_max2((bw_add((bw_div(data->src_data_for_first_output_pixel[i], bw_min2(bw_mul(data->bytes_per_request[i], sclk[k]), bw_div(bw_mul(data->bytes_per_request[i], vbios->low_voltage_max_dispclk), bw_int_to_fixed(2))))), (bw_mul(data->mcifwr_burst_time[j][k], bw_floor2(bw_div(data->src_data_for_first_output_pixel[i], data->adjusted_data_buffer_size_in_memory[i]), bw_int_to_fixed(1)))))), (bw_add((bw_div(data->src_data_for_last_output_pixel[i], bw_min2(bw_mul(data->bytes_per_request[i], sclk[k]), bw_div(bw_mul(data->bytes_per_request[i], vbios->low_voltage_max_dispclk), bw_int_to_fixed(2))))), (bw_sub(bw_mul(data->mcifwr_burst_time[j][k], bw_floor2(bw_div(data->src_data_for_last_output_pixel[i], data->adjusted_data_buffer_size_in_memory[i]), bw_int_to_fixed(1))), data->active_time[i])))));
+					}
+				}
+			}
+		}
+	}
+	/*cpu c-state and p-state change enable*/
+	/*for cpu p-state change to be possible for a yclk(pclk) and sclk level the dispclk required has to be enough for the blackout duration*/
+	/*for cpu c-state change to be possible for a yclk(pclk) and sclk level the dispclk required has to be enough for the blackout duration and recovery*/
+	/*condition for the blackout duration:*/
+	/* minimum latency hiding > blackout duration + dmif burst time + line source transfer time*/
+	/*condition for the blackout recovery:*/
+	/* recovery time >  dmif burst time + 2 * urgent latency*/
+	/* recovery time > (display bw * blackout duration  + (2 * urgent latency + dmif burst time)*dispclk - dmif size )*/
+	/*                  / (dispclk - display bw)*/
+	/*the minimum latency hiding is the minimum for all pipes of one screen line time, plus one more line time if doing lb prefetch, plus the dmif data buffer size equivalent in time, minus the urgent latency.*/
+	/*the minimum latency hiding is  further limited by the cursor.  the cursor latency hiding is the number of lines of the cursor buffer, minus one if the downscaling is less than two, or minus three if it is more*/
+
+	/*initialize variables*/
+	number_of_displays_enabled = 0;
+	number_of_displays_enabled_with_margin = 0;
+	for (k = 0; k <= maximum_number_of_surfaces - 1; k++) {
+		if (data->enable[k]) {
+			number_of_displays_enabled = number_of_displays_enabled + 1;
+		}
+		data->display_pstate_change_enable[k] = 0;
+	}
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if ((bw_equ(dceip->stutter_and_dram_clock_state_change_gated_before_cursor, bw_int_to_fixed(0)) && bw_mtn(data->cursor_width_pixels[i], bw_int_to_fixed(0)))) {
+				if (bw_ltn(data->vsr[i], bw_int_to_fixed(2))) {
+					data->cursor_latency_hiding[i] = bw_div(bw_div(bw_mul((bw_sub(dceip->cursor_dcp_buffer_lines, bw_int_to_fixed(1))), data->h_total[i]), data->vsr[i]), data->pixel_rate[i]);
+				}
+				else {
+					data->cursor_latency_hiding[i] = bw_div(bw_div(bw_mul((bw_sub(dceip->cursor_dcp_buffer_lines, bw_int_to_fixed(3))), data->h_total[i]), data->vsr[i]), data->pixel_rate[i]);
+				}
+			}
+			else {
+				data->cursor_latency_hiding[i] = bw_int_to_fixed(9999);
+			}
+		}
+	}
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if (dceip->graphics_lb_nodownscaling_multi_line_prefetching == 1 && (bw_equ(data->vsr[i], bw_int_to_fixed(1)) || (bw_leq(data->vsr[i], bw_frc_to_fixed(8, 10)) && bw_leq(data->v_taps[i], bw_int_to_fixed(2)) && data->lb_bpc[i] == 8)) && surface_type[i] == bw_def_graphics) {
+				if (number_of_displays_enabled > 2)
+					data->minimum_latency_hiding[i] = bw_sub(bw_div(bw_mul((bw_div((bw_add(bw_sub(data->lb_partitions[i], bw_int_to_fixed(2)), bw_div(bw_div(data->data_buffer_size[i], bw_int_to_fixed(data->bytes_per_pixel[i])), data->source_width_pixels[i]))), data->vsr[i])), data->h_total[i]), data->pixel_rate[i]), data->total_dmifmc_urgent_latency);
+				else
+					data->minimum_latency_hiding[i] = bw_sub(bw_div(bw_mul((bw_div((bw_add(bw_sub(data->lb_partitions[i], bw_int_to_fixed(1)), bw_div(bw_div(data->data_buffer_size[i], bw_int_to_fixed(data->bytes_per_pixel[i])), data->source_width_pixels[i]))), data->vsr[i])), data->h_total[i]), data->pixel_rate[i]), data->total_dmifmc_urgent_latency);
+			}
+			else {
+				data->minimum_latency_hiding[i] = bw_sub(bw_div(bw_mul((bw_div((bw_add(bw_int_to_fixed(1 + data->line_buffer_prefetch[i]), bw_div(bw_div(data->data_buffer_size[i], bw_int_to_fixed(data->bytes_per_pixel[i])), data->source_width_pixels[i]))), data->vsr[i])), data->h_total[i]), data->pixel_rate[i]), data->total_dmifmc_urgent_latency);
+			}
+			data->minimum_latency_hiding_with_cursor[i] = bw_min2(data->minimum_latency_hiding[i], data->cursor_latency_hiding[i]);
+		}
+	}
+	for (i = 0; i <= 2; i++) {
+		for (j = 0; j <= 7; j++) {
+			data->blackout_duration_margin[i][j] = bw_int_to_fixed(9999);
+			data->dispclk_required_for_blackout_duration[i][j] = bw_int_to_fixed(0);
+			data->dispclk_required_for_blackout_recovery[i][j] = bw_int_to_fixed(0);
+			for (k = 0; k <= maximum_number_of_surfaces - 1; k++) {
+				if (data->enable[k] && bw_mtn(vbios->blackout_duration, bw_int_to_fixed(0))) {
+					if (surface_type[k] != bw_def_display_write_back420_luma && surface_type[k] != bw_def_display_write_back420_chroma) {
+						data->blackout_duration_margin[i][j] = bw_min2(data->blackout_duration_margin[i][j], bw_sub(bw_sub(bw_sub(data->minimum_latency_hiding_with_cursor[k], vbios->blackout_duration), data->dmif_burst_time[i][j]), data->line_source_transfer_time[k][i][j]));
+						data->dispclk_required_for_blackout_duration[i][j] = bw_max3(data->dispclk_required_for_blackout_duration[i][j], bw_div(bw_div(bw_mul(data->src_pixels_for_first_output_pixel[k], dceip->display_pipe_throughput_factor), dceip->lb_write_pixels_per_dispclk), (bw_sub(bw_sub(data->minimum_latency_hiding_with_cursor[k], vbios->blackout_duration), data->dmif_burst_time[i][j]))), bw_div(bw_div(bw_mul(data->src_pixels_for_last_output_pixel[k], dceip->display_pipe_throughput_factor), dceip->lb_write_pixels_per_dispclk), (bw_add(bw_sub(bw_sub(data->minimum_latency_hiding_with_cursor[k], vbios->blackout_duration), data->dmif_burst_time[i][j]), data->active_time[k]))));
+						if (bw_leq(vbios->maximum_blackout_recovery_time, bw_add(bw_mul(bw_int_to_fixed(2), data->total_dmifmc_urgent_latency), data->dmif_burst_time[i][j]))) {
+							data->dispclk_required_for_blackout_recovery[i][j] = bw_int_to_fixed(9999);
+						}
+						else if (bw_ltn(data->adjusted_data_buffer_size[k], bw_mul(bw_div(bw_mul(data->display_bandwidth[k], data->useful_bytes_per_request[k]), data->bytes_per_request[k]), (bw_add(vbios->blackout_duration, bw_add(bw_mul(bw_int_to_fixed(2), data->total_dmifmc_urgent_latency), data->dmif_burst_time[i][j])))))) {
+							data->dispclk_required_for_blackout_recovery[i][j] = bw_max2(data->dispclk_required_for_blackout_recovery[i][j], bw_div(bw_mul(bw_div(bw_div((bw_sub(bw_mul(bw_div(bw_mul(data->display_bandwidth[k], data->useful_bytes_per_request[k]), data->bytes_per_request[k]), (bw_add(vbios->blackout_duration, vbios->maximum_blackout_recovery_time))), data->adjusted_data_buffer_size[k])), bw_int_to_fixed(data->bytes_per_pixel[k])), (bw_sub(vbios->maximum_blackout_recovery_time, bw_sub(bw_mul(bw_int_to_fixed(2), data->total_dmifmc_urgent_latency), data->dmif_burst_time[i][j])))), data->latency_hiding_lines[k]), data->lines_interleaved_in_mem_access[k]));
+						}
+					}
+					else {
+						data->blackout_duration_margin[i][j] = bw_min2(data->blackout_duration_margin[i][j], bw_sub(bw_sub(bw_sub(bw_sub(data->minimum_latency_hiding_with_cursor[k], vbios->blackout_duration), data->dmif_burst_time[i][j]), data->mcifwr_burst_time[i][j]), data->line_source_transfer_time[k][i][j]));
+						data->dispclk_required_for_blackout_duration[i][j] = bw_max3(data->dispclk_required_for_blackout_duration[i][j], bw_div(bw_div(bw_mul(data->src_pixels_for_first_output_pixel[k], dceip->display_pipe_throughput_factor), dceip->lb_write_pixels_per_dispclk), (bw_sub(bw_sub(bw_sub(data->minimum_latency_hiding_with_cursor[k], vbios->blackout_duration), data->dmif_burst_time[i][j]), data->mcifwr_burst_time[i][j]))), bw_div(bw_div(bw_mul(data->src_pixels_for_last_output_pixel[k], dceip->display_pipe_throughput_factor), dceip->lb_write_pixels_per_dispclk), (bw_add(bw_sub(bw_sub(bw_sub(data->minimum_latency_hiding_with_cursor[k], vbios->blackout_duration), data->dmif_burst_time[i][j]), data->mcifwr_burst_time[i][j]), data->active_time[k]))));
+						if (bw_ltn(vbios->maximum_blackout_recovery_time, bw_add(bw_add(bw_mul(bw_int_to_fixed(2), vbios->mcifwrmc_urgent_latency), data->dmif_burst_time[i][j]), data->mcifwr_burst_time[i][j]))) {
+							data->dispclk_required_for_blackout_recovery[i][j] = bw_int_to_fixed(9999);
+						}
+						else if (bw_ltn(data->adjusted_data_buffer_size[k], bw_mul(bw_div(bw_mul(data->display_bandwidth[k], data->useful_bytes_per_request[k]), data->bytes_per_request[k]), (bw_add(vbios->blackout_duration, bw_add(bw_mul(bw_int_to_fixed(2), data->total_dmifmc_urgent_latency), data->dmif_burst_time[i][j])))))) {
+							data->dispclk_required_for_blackout_recovery[i][j] = bw_max2(data->dispclk_required_for_blackout_recovery[i][j], bw_div(bw_mul(bw_div(bw_div((bw_sub(bw_mul(bw_div(bw_mul(data->display_bandwidth[k], data->useful_bytes_per_request[k]), data->bytes_per_request[k]), (bw_add(vbios->blackout_duration, vbios->maximum_blackout_recovery_time))), data->adjusted_data_buffer_size[k])), bw_int_to_fixed(data->bytes_per_pixel[k])), (bw_sub(vbios->maximum_blackout_recovery_time, (bw_add(bw_mul(bw_int_to_fixed(2), data->total_dmifmc_urgent_latency), data->dmif_burst_time[i][j]))))), data->latency_hiding_lines[k]), data->lines_interleaved_in_mem_access[k]));
+						}
+					}
+				}
+			}
+		}
+	}
+	if (bw_mtn(data->blackout_duration_margin[high][s_high], bw_int_to_fixed(0)) && bw_ltn(data->dispclk_required_for_blackout_duration[high][s_high], vbios->high_voltage_max_dispclk)) {
+		data->cpup_state_change_enable = bw_def_yes;
+		if (bw_ltn(data->dispclk_required_for_blackout_recovery[high][s_high], vbios->high_voltage_max_dispclk)) {
+			data->cpuc_state_change_enable = bw_def_yes;
+		}
+		else {
+			data->cpuc_state_change_enable = bw_def_no;
+		}
+	}
+	else {
+		data->cpup_state_change_enable = bw_def_no;
+		data->cpuc_state_change_enable = bw_def_no;
+	}
+	/*nb p-state change enable*/
+	/*for dram speed/p-state change to be possible for a yclk(pclk) and sclk level there has to be positive margin and the dispclk required has to be*/
+	/*below the maximum.*/
+	/*the dram speed/p-state change margin is the minimum for all surfaces of the maximum latency hiding minus the dram speed/p-state change latency,*/
+	/*minus the dmif burst time, minus the source line transfer time*/
+	/*the maximum latency hiding is the minimum latency hiding plus one source line used for de-tiling in the line buffer, plus half the urgent latency*/
+	/*if stutter and dram clock state change are gated before cursor then the cursor latency hiding does not limit stutter or dram clock state change*/
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			/*maximum_latency_hiding(i) = minimum_latency_hiding(i) + 1 / vsr(i) **/
+			/*      h_total(i) / pixel_rate(i) + 0.5 * total_dmifmc_urgent_latency*/
+			data->maximum_latency_hiding[i] = bw_add(data->minimum_latency_hiding[i],
+				bw_mul(bw_frc_to_fixed(5, 10), data->total_dmifmc_urgent_latency));
+			data->maximum_latency_hiding_with_cursor[i] = bw_min2(data->maximum_latency_hiding[i], data->cursor_latency_hiding[i]);
+		}
+	}
+	for (i = 0; i <= 2; i++) {
+		for (j = 0; j <= 7; j++) {
+			data->min_dram_speed_change_margin[i][j] = bw_int_to_fixed(9999);
+			data->dram_speed_change_margin = bw_int_to_fixed(9999);
+			data->dispclk_required_for_dram_speed_change[i][j] = bw_int_to_fixed(0);
+			data->num_displays_with_margin[i][j] = 0;
+			for (k = 0; k <= maximum_number_of_surfaces - 1; k++) {
+				if (data->enable[k]) {
+					if (surface_type[k] != bw_def_display_write_back420_luma && surface_type[k] != bw_def_display_write_back420_chroma) {
+						data->dram_speed_change_margin = bw_sub(bw_sub(bw_sub(data->maximum_latency_hiding_with_cursor[k], vbios->nbp_state_change_latency), data->dmif_burst_time[i][j]), data->dram_speed_change_line_source_transfer_time[k][i][j]);
+						if ((bw_mtn(data->dram_speed_change_margin, bw_int_to_fixed(0)) && bw_ltn(data->dram_speed_change_margin, bw_int_to_fixed(9999)))) {
+							/*determine the minimum dram clock change margin for each set of clock frequencies*/
+							data->min_dram_speed_change_margin[i][j] = bw_min2(data->min_dram_speed_change_margin[i][j], data->dram_speed_change_margin);
+							/*compute the maximum clock frequency required for the dram clock change at each set of clock frequencies*/
+							data->dispclk_required_for_dram_speed_change_pipe[i][j] = bw_max2(bw_div(bw_div(bw_mul(data->src_pixels_for_first_output_pixel[k], dceip->display_pipe_throughput_factor), dceip->lb_write_pixels_per_dispclk), (bw_sub(bw_sub(bw_sub(data->maximum_latency_hiding_with_cursor[k], vbios->nbp_state_change_latency), data->dmif_burst_time[i][j]), data->dram_speed_change_line_source_transfer_time[k][i][j]))), bw_div(bw_div(bw_mul(data->src_pixels_for_last_output_pixel[k], dceip->display_pipe_throughput_factor), dceip->lb_write_pixels_per_dispclk), (bw_add(bw_sub(bw_sub(bw_sub(data->maximum_latency_hiding_with_cursor[k], vbios->nbp_state_change_latency), data->dmif_burst_time[i][j]), data->dram_speed_change_line_source_transfer_time[k][i][j]), data->active_time[k]))));
+							if ((bw_ltn(data->dispclk_required_for_dram_speed_change_pipe[i][j], vbios->high_voltage_max_dispclk))) {
+								data->display_pstate_change_enable[k] = 1;
+								data->num_displays_with_margin[i][j] = data->num_displays_with_margin[i][j] + 1;
+								data->dispclk_required_for_dram_speed_change[i][j] = bw_max2(data->dispclk_required_for_dram_speed_change[i][j], data->dispclk_required_for_dram_speed_change_pipe[i][j]);
+							}
+						}
+					}
+					else {
+						data->dram_speed_change_margin = bw_sub(bw_sub(bw_sub(bw_sub(data->maximum_latency_hiding_with_cursor[k], vbios->nbp_state_change_latency), data->dmif_burst_time[i][j]), data->mcifwr_burst_time[i][j]), data->dram_speed_change_line_source_transfer_time[k][i][j]);
+						if ((bw_mtn(data->dram_speed_change_margin, bw_int_to_fixed(0)) && bw_ltn(data->dram_speed_change_margin, bw_int_to_fixed(9999)))) {
+							/*determine the minimum dram clock change margin for each display pipe*/
+							data->min_dram_speed_change_margin[i][j] = bw_min2(data->min_dram_speed_change_margin[i][j], data->dram_speed_change_margin);
+							/*compute the maximum clock frequency required for the dram clock change at each set of clock frequencies*/
+							data->dispclk_required_for_dram_speed_change_pipe[i][j] = bw_max2(bw_div(bw_div(bw_mul(data->src_pixels_for_first_output_pixel[k], dceip->display_pipe_throughput_factor), dceip->lb_write_pixels_per_dispclk), (bw_sub(bw_sub(bw_sub(bw_sub(data->maximum_latency_hiding_with_cursor[k], vbios->nbp_state_change_latency), data->dmif_burst_time[i][j]), data->dram_speed_change_line_source_transfer_time[k][i][j]), data->mcifwr_burst_time[i][j]))), bw_div(bw_div(bw_mul(data->src_pixels_for_last_output_pixel[k], dceip->display_pipe_throughput_factor), dceip->lb_write_pixels_per_dispclk), (bw_add(bw_sub(bw_sub(bw_sub(bw_sub(data->maximum_latency_hiding_with_cursor[k], vbios->nbp_state_change_latency), data->dmif_burst_time[i][j]), data->dram_speed_change_line_source_transfer_time[k][i][j]), data->mcifwr_burst_time[i][j]), data->active_time[k]))));
+							if ((bw_ltn(data->dispclk_required_for_dram_speed_change_pipe[i][j], vbios->high_voltage_max_dispclk))) {
+								data->display_pstate_change_enable[k] = 1;
+								data->num_displays_with_margin[i][j] = data->num_displays_with_margin[i][j] + 1;
+								data->dispclk_required_for_dram_speed_change[i][j] = bw_max2(data->dispclk_required_for_dram_speed_change[i][j], data->dispclk_required_for_dram_speed_change_pipe[i][j]);
+							}
+						}
+					}
+				}
+			}
+		}
+	}
+	/*determine the number of displays with margin to switch in the v_active region*/
+	for (k = 0; k <= maximum_number_of_surfaces - 1; k++) {
+		if (data->enable[k] == 1 && data->display_pstate_change_enable[k] == 1) {
+			number_of_displays_enabled_with_margin = number_of_displays_enabled_with_margin + 1;
+		}
+	}
+	/*determine the number of displays that don't have any dram clock change margin, but*/
+	/*have the same resolution.  these displays can switch in a common vblank region if*/
+	/*their frames are aligned.*/
+	data->min_vblank_dram_speed_change_margin = bw_int_to_fixed(9999);
+	for (k = 0; k <= maximum_number_of_surfaces - 1; k++) {
+		if (data->enable[k]) {
+			if (surface_type[k] != bw_def_display_write_back420_luma && surface_type[k] != bw_def_display_write_back420_chroma) {
+				data->v_blank_dram_speed_change_margin[k] = bw_sub(bw_sub(bw_sub(bw_div(bw_mul((bw_sub(data->v_total[k], bw_sub(bw_div(data->src_height[k], data->v_scale_ratio[k]), bw_int_to_fixed(4)))), data->h_total[k]), data->pixel_rate[k]), vbios->nbp_state_change_latency), data->dmif_burst_time[low][s_low]), data->dram_speed_change_line_source_transfer_time[k][low][s_low]);
+				data->min_vblank_dram_speed_change_margin = bw_min2(data->min_vblank_dram_speed_change_margin, data->v_blank_dram_speed_change_margin[k]);
+			}
+			else {
+				data->v_blank_dram_speed_change_margin[k] = bw_sub(bw_sub(bw_sub(bw_sub(bw_div(bw_mul((bw_sub(data->v_total[k], bw_sub(bw_div(data->src_height[k], data->v_scale_ratio[k]), bw_int_to_fixed(4)))), data->h_total[k]), data->pixel_rate[k]), vbios->nbp_state_change_latency), data->dmif_burst_time[low][s_low]), data->mcifwr_burst_time[low][s_low]), data->dram_speed_change_line_source_transfer_time[k][low][s_low]);
+				data->min_vblank_dram_speed_change_margin = bw_min2(data->min_vblank_dram_speed_change_margin, data->v_blank_dram_speed_change_margin[k]);
+			}
+		}
+	}
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		data->displays_with_same_mode[i] = bw_int_to_fixed(0);
+		if (data->enable[i] == 1 && data->display_pstate_change_enable[i] == 0 && bw_mtn(data->v_blank_dram_speed_change_margin[i], bw_int_to_fixed(0))) {
+			for (j = 0; j <= maximum_number_of_surfaces - 1; j++) {
+				if ((i == j || data->display_synchronization_enabled) && (data->enable[j] == 1 && bw_equ(data->source_width_rounded_up_to_chunks[i], data->source_width_rounded_up_to_chunks[j]) && bw_equ(data->source_height_rounded_up_to_chunks[i], data->source_height_rounded_up_to_chunks[j]) && bw_equ(data->vsr[i], data->vsr[j]) && bw_equ(data->hsr[i], data->hsr[j]) && bw_equ(data->pixel_rate[i], data->pixel_rate[j]))) {
+					data->displays_with_same_mode[i] = bw_add(data->displays_with_same_mode[i], bw_int_to_fixed(1));
+				}
+			}
+		}
+	}
+	/*compute the maximum number of aligned displays with no margin*/
+	number_of_aligned_displays_with_no_margin = 0;
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		number_of_aligned_displays_with_no_margin = bw_fixed_to_int(bw_max2(bw_int_to_fixed(number_of_aligned_displays_with_no_margin), data->displays_with_same_mode[i]));
+	}
+	/*dram clock change is possible, if all displays have positive margin except for one display or a group of*/
+	/*aligned displays with the same timing.*/
+	/*the display(s) with the negative margin can be switched in the v_blank region while the other*/
+	/*displays are in v_blank or v_active.*/
+	if (number_of_displays_enabled_with_margin > 0 && (number_of_displays_enabled_with_margin + number_of_aligned_displays_with_no_margin) == number_of_displays_enabled && bw_mtn(data->min_dram_speed_change_margin[high][s_high], bw_int_to_fixed(0)) && bw_ltn(data->min_dram_speed_change_margin[high][s_high], bw_int_to_fixed(9999)) && bw_ltn(data->dispclk_required_for_dram_speed_change[high][s_high], vbios->high_voltage_max_dispclk)) {
+		data->nbp_state_change_enable = bw_def_yes;
+	}
+	else {
+		data->nbp_state_change_enable = bw_def_no;
+	}
+	/*dram clock change is possible only in vblank if all displays are aligned and have no margin*/
+	if (number_of_aligned_displays_with_no_margin == number_of_displays_enabled) {
+		nbp_state_change_enable_blank = bw_def_yes;
+	}
+	else {
+		nbp_state_change_enable_blank = bw_def_no;
+	}
+
+	/*average bandwidth*/
+	/*the average bandwidth with no compression is the vertical active time is the source width times the bytes per pixel divided by the line time, multiplied by the vertical scale ratio and the ratio of bytes per request divided by the useful bytes per request.*/
+	/*the average bandwidth with compression is the same, divided by the compression ratio*/
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			data->average_bandwidth_no_compression[i] = bw_div(bw_mul(bw_mul(bw_div(bw_mul(data->source_width_rounded_up_to_chunks[i], bw_int_to_fixed(data->bytes_per_pixel[i])), (bw_div(data->h_total[i], data->pixel_rate[i]))), data->vsr[i]), data->bytes_per_request[i]), data->useful_bytes_per_request[i]);
+			data->average_bandwidth[i] = bw_div(data->average_bandwidth_no_compression[i], data->compression_rate[i]);
+		}
+	}
+	data->total_average_bandwidth_no_compression = bw_int_to_fixed(0);
+	data->total_average_bandwidth = bw_int_to_fixed(0);
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			data->total_average_bandwidth_no_compression = bw_add(data->total_average_bandwidth_no_compression, data->average_bandwidth_no_compression[i]);
+			data->total_average_bandwidth = bw_add(data->total_average_bandwidth, data->average_bandwidth[i]);
+		}
+	}
+
+	/*required yclk(pclk)*/
+	/*yclk requirement only makes sense if the dmif and mcifwr data total page close-open time is less than the time for data transfer and the total pte requests fit in the scatter-gather saw queque size*/
+	/*if that is the case, the yclk requirement is the maximum of the ones required by dmif and mcifwr, and the high/low yclk(pclk) is chosen accordingly*/
+	/*high yclk(pclk) has to be selected when dram speed/p-state change is not possible.*/
+	data->min_cursor_memory_interface_buffer_size_in_time = bw_int_to_fixed(9999);
+	/* number of cursor lines stored in the cursor data return buffer*/
+	num_cursor_lines = 0;
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if (bw_mtn(data->cursor_width_pixels[i], bw_int_to_fixed(0))) {
+				/*compute number of cursor lines stored in data return buffer*/
+				if (bw_leq(data->cursor_width_pixels[i], bw_int_to_fixed(64)) && dceip->large_cursor == 1) {
+					num_cursor_lines = 4;
+				}
+				else {
+					num_cursor_lines = 2;
+				}
+				data->min_cursor_memory_interface_buffer_size_in_time = bw_min2(data->min_cursor_memory_interface_buffer_size_in_time, bw_div(bw_mul(bw_div(bw_int_to_fixed(num_cursor_lines), data->vsr[i]), data->h_total[i]), data->pixel_rate[i]));
+			}
+		}
+	}
+	/*compute minimum time to read one chunk from the dmif buffer*/
+	if (number_of_displays_enabled > 2) {
+		data->chunk_request_delay = 0;
+	}
+	else {
+		data->chunk_request_delay = bw_fixed_to_int(bw_div(bw_int_to_fixed(512), vbios->high_voltage_max_dispclk));
+	}
+	data->min_read_buffer_size_in_time = bw_min2(data->min_cursor_memory_interface_buffer_size_in_time, data->min_dmif_size_in_time);
+	data->display_reads_time_for_data_transfer = bw_sub(bw_sub(data->min_read_buffer_size_in_time, data->total_dmifmc_urgent_latency), bw_int_to_fixed(data->chunk_request_delay));
+	data->display_writes_time_for_data_transfer = bw_sub(data->min_mcifwr_size_in_time, vbios->mcifwrmc_urgent_latency);
+	data->dmif_required_dram_bandwidth = bw_div(data->total_display_reads_required_dram_access_data, data->display_reads_time_for_data_transfer);
+	data->mcifwr_required_dram_bandwidth = bw_div(data->total_display_writes_required_dram_access_data, data->display_writes_time_for_data_transfer);
+	data->required_dmifmc_urgent_latency_for_page_close_open = bw_div((bw_sub(data->min_read_buffer_size_in_time, data->dmif_total_page_close_open_time)), data->total_dmifmc_urgent_trips);
+	data->required_mcifmcwr_urgent_latency = bw_sub(data->min_mcifwr_size_in_time, data->mcifwr_total_page_close_open_time);
+	if (bw_mtn(data->scatter_gather_total_pte_requests, dceip->maximum_total_outstanding_pte_requests_allowed_by_saw)) {
+		data->required_dram_bandwidth_gbyte_per_second = bw_int_to_fixed(9999);
+		yclk_message = bw_def_exceeded_allowed_outstanding_pte_req_queue_size;
+		data->y_clk_level = high;
+		data->dram_bandwidth = bw_mul(bw_div(bw_mul(bw_mul(data->dram_efficiency, yclk[high]), bw_int_to_fixed(vbios->dram_channel_width_in_bits)), bw_int_to_fixed(8)), bw_int_to_fixed(data->number_of_dram_channels));
+	}
+	else if (bw_mtn(vbios->dmifmc_urgent_latency, data->required_dmifmc_urgent_latency_for_page_close_open) || bw_mtn(vbios->mcifwrmc_urgent_latency, data->required_mcifmcwr_urgent_latency)) {
+		data->required_dram_bandwidth_gbyte_per_second = bw_int_to_fixed(9999);
+		yclk_message = bw_def_exceeded_allowed_page_close_open;
+		data->y_clk_level = high;
+		data->dram_bandwidth = bw_mul(bw_div(bw_mul(bw_mul(data->dram_efficiency, yclk[high]), bw_int_to_fixed(vbios->dram_channel_width_in_bits)), bw_int_to_fixed(8)), bw_int_to_fixed(data->number_of_dram_channels));
+	}
+	else {
+		data->required_dram_bandwidth_gbyte_per_second = bw_div(bw_max2(data->dmif_required_dram_bandwidth, data->mcifwr_required_dram_bandwidth), bw_int_to_fixed(1000));
+		if (bw_ltn(data->total_average_bandwidth_no_compression, bw_mul(bw_mul(bw_mul(bw_frc_to_fixed(dceip->max_average_percent_of_ideal_drambw_display_can_use_in_normal_system_operation, 100),yclk[low]),bw_div(bw_int_to_fixed(vbios->dram_channel_width_in_bits),bw_int_to_fixed(8))),bw_int_to_fixed(vbios->number_of_dram_channels)))
+				&& bw_ltn(bw_mul(data->required_dram_bandwidth_gbyte_per_second, bw_int_to_fixed(1000)), bw_mul(bw_div(bw_mul(bw_mul(data->dram_efficiency, yclk[low]), bw_int_to_fixed(vbios->dram_channel_width_in_bits)), bw_int_to_fixed(8)), bw_int_to_fixed(data->number_of_dram_channels))) && (data->cpup_state_change_enable == bw_def_no || (bw_mtn(data->blackout_duration_margin[low][s_high], bw_int_to_fixed(0)) && bw_ltn(data->dispclk_required_for_blackout_duration[low][s_high], vbios->high_voltage_max_dispclk))) && (data->cpuc_state_change_enable == bw_def_no || (bw_mtn(data->blackout_duration_margin[low][s_high], bw_int_to_fixed(0)) && bw_ltn(data->dispclk_required_for_blackout_duration[low][s_high], vbios->high_voltage_max_dispclk) && bw_ltn(data->dispclk_required_for_blackout_recovery[low][s_high], vbios->high_voltage_max_dispclk))) && (!data->increase_voltage_to_support_mclk_switch || data->nbp_state_change_enable == bw_def_no || (bw_mtn(data->min_dram_speed_change_margin[low][s_high], bw_int_to_fixed(0)) && bw_ltn(data->min_dram_speed_change_margin[low][s_high], bw_int_to_fixed(9999)) && bw_leq(data->dispclk_required_for_dram_speed_change[low][s_high], vbios->high_voltage_max_dispclk) && data->num_displays_with_margin[low][s_high] == number_of_displays_enabled_with_margin))) {
+			yclk_message = bw_fixed_to_int(vbios->low_yclk);
+			data->y_clk_level = low;
+			data->dram_bandwidth = bw_mul(bw_div(bw_mul(bw_mul(data->dram_efficiency, yclk[low]), bw_int_to_fixed(vbios->dram_channel_width_in_bits)), bw_int_to_fixed(8)), bw_int_to_fixed(data->number_of_dram_channels));
+		}
+		else if (bw_ltn(data->total_average_bandwidth_no_compression, bw_mul(bw_mul(bw_mul(bw_frc_to_fixed(dceip->max_average_percent_of_ideal_drambw_display_can_use_in_normal_system_operation, 100),yclk[mid]),bw_div(bw_int_to_fixed(vbios->dram_channel_width_in_bits),bw_int_to_fixed(8))),bw_int_to_fixed(vbios->number_of_dram_channels)))
+				&& bw_ltn(bw_mul(data->required_dram_bandwidth_gbyte_per_second, bw_int_to_fixed(1000)), bw_mul(bw_div(bw_mul(bw_mul(data->dram_efficiency, yclk[mid]), bw_int_to_fixed(vbios->dram_channel_width_in_bits)), bw_int_to_fixed(8)), bw_int_to_fixed(data->number_of_dram_channels))) && (data->cpup_state_change_enable == bw_def_no || (bw_mtn(data->blackout_duration_margin[mid][s_high], bw_int_to_fixed(0)) && bw_ltn(data->dispclk_required_for_blackout_duration[mid][s_high], vbios->high_voltage_max_dispclk))) && (data->cpuc_state_change_enable == bw_def_no || (bw_mtn(data->blackout_duration_margin[mid][s_high], bw_int_to_fixed(0)) && bw_ltn(data->dispclk_required_for_blackout_duration[mid][s_high], vbios->high_voltage_max_dispclk) && bw_ltn(data->dispclk_required_for_blackout_recovery[mid][s_high], vbios->high_voltage_max_dispclk))) && (!data->increase_voltage_to_support_mclk_switch || data->nbp_state_change_enable == bw_def_no || (bw_mtn(data->min_dram_speed_change_margin[mid][s_high], bw_int_to_fixed(0)) && bw_ltn(data->min_dram_speed_change_margin[mid][s_high], bw_int_to_fixed(9999)) && bw_leq(data->dispclk_required_for_dram_speed_change[mid][s_high], vbios->high_voltage_max_dispclk) && data->num_displays_with_margin[mid][s_high] == number_of_displays_enabled_with_margin))) {
+			yclk_message = bw_fixed_to_int(vbios->mid_yclk);
+			data->y_clk_level = mid;
+			data->dram_bandwidth = bw_mul(bw_div(bw_mul(bw_mul(data->dram_efficiency, yclk[mid]), bw_int_to_fixed(vbios->dram_channel_width_in_bits)), bw_int_to_fixed(8)), bw_int_to_fixed(data->number_of_dram_channels));
+		}
+		else if (bw_ltn(data->total_average_bandwidth_no_compression, bw_mul(bw_mul(bw_mul(bw_frc_to_fixed(dceip->max_average_percent_of_ideal_drambw_display_can_use_in_normal_system_operation, 100),yclk[high]),bw_div(bw_int_to_fixed(vbios->dram_channel_width_in_bits),bw_int_to_fixed(8))),bw_int_to_fixed(vbios->number_of_dram_channels)))
+				&& bw_ltn(bw_mul(data->required_dram_bandwidth_gbyte_per_second, bw_int_to_fixed(1000)), bw_mul(bw_div(bw_mul(bw_mul(data->dram_efficiency, yclk[high]), bw_int_to_fixed(vbios->dram_channel_width_in_bits)), bw_int_to_fixed(8)), bw_int_to_fixed(data->number_of_dram_channels)))) {
+			yclk_message = bw_fixed_to_int(vbios->high_yclk);
+			data->y_clk_level = high;
+			data->dram_bandwidth = bw_mul(bw_div(bw_mul(bw_mul(data->dram_efficiency, yclk[high]), bw_int_to_fixed(vbios->dram_channel_width_in_bits)), bw_int_to_fixed(8)), bw_int_to_fixed(data->number_of_dram_channels));
+		}
+		else {
+			yclk_message = bw_def_exceeded_allowed_maximum_bw;
+			data->y_clk_level = high;
+			data->dram_bandwidth = bw_mul(bw_div(bw_mul(bw_mul(data->dram_efficiency, yclk[high]), bw_int_to_fixed(vbios->dram_channel_width_in_bits)), bw_int_to_fixed(8)), bw_int_to_fixed(data->number_of_dram_channels));
+		}
+	}
+	/*required sclk*/
+	/*sclk requirement only makes sense if the total pte requests fit in the scatter-gather saw queque size*/
+	/*if that is the case, the sclk requirement is the maximum of the ones required by dmif and mcifwr, and the high/mid/low sclk is chosen accordingly, unless that choice results in foresaking dram speed/nb p-state change.*/
+	/*the dmif and mcifwr sclk required is the one that allows the transfer of all pipe's data buffer size through the sclk bus in the time for data transfer*/
+	/*for dmif, pte and cursor requests have to be included.*/
+	data->dmif_required_sclk = bw_div(bw_div(data->total_display_reads_required_data, data->display_reads_time_for_data_transfer), (bw_mul(vbios->data_return_bus_width, bw_frc_to_fixed(dceip->percent_of_ideal_port_bw_received_after_urgent_latency, 100))));
+	data->mcifwr_required_sclk = bw_div(bw_div(data->total_display_writes_required_data, data->display_writes_time_for_data_transfer), vbios->data_return_bus_width);
+	if (bw_mtn(data->scatter_gather_total_pte_requests, dceip->maximum_total_outstanding_pte_requests_allowed_by_saw)) {
+		data->required_sclk = bw_int_to_fixed(9999);
+		sclk_message = bw_def_exceeded_allowed_outstanding_pte_req_queue_size;
+		data->sclk_level = s_high;
+	}
+	else if (bw_mtn(vbios->dmifmc_urgent_latency, data->required_dmifmc_urgent_latency_for_page_close_open) || bw_mtn(vbios->mcifwrmc_urgent_latency, data->required_mcifmcwr_urgent_latency)) {
+		data->required_sclk = bw_int_to_fixed(9999);
+		sclk_message = bw_def_exceeded_allowed_page_close_open;
+		data->sclk_level = s_high;
+	}
+	else {
+		data->required_sclk = bw_max2(data->dmif_required_sclk, data->mcifwr_required_sclk);
+		if (bw_ltn(data->total_average_bandwidth_no_compression, bw_mul(bw_mul(bw_frc_to_fixed(dceip->max_average_percent_of_ideal_port_bw_display_can_use_in_normal_system_operation, 100),sclk[low]),vbios->data_return_bus_width))
+				&& bw_ltn(data->required_sclk, sclk[s_low]) && (data->cpup_state_change_enable == bw_def_no || (bw_mtn(data->blackout_duration_margin[data->y_clk_level][s_low], bw_int_to_fixed(0)) && bw_ltn(data->dispclk_required_for_blackout_duration[data->y_clk_level][s_low], vbios->high_voltage_max_dispclk))) && (data->cpuc_state_change_enable == bw_def_no || (bw_mtn(data->blackout_duration_margin[data->y_clk_level][s_low], bw_int_to_fixed(0)) && bw_ltn(data->dispclk_required_for_blackout_duration[data->y_clk_level][s_low], vbios->high_voltage_max_dispclk) && bw_ltn(data->dispclk_required_for_blackout_recovery[data->y_clk_level][s_low], vbios->high_voltage_max_dispclk))) && (!data->increase_voltage_to_support_mclk_switch || data->nbp_state_change_enable == bw_def_no || (bw_mtn(data->min_dram_speed_change_margin[data->y_clk_level][s_low], bw_int_to_fixed(0)) && bw_ltn(data->min_dram_speed_change_margin[data->y_clk_level][s_low], bw_int_to_fixed(9999)) && bw_leq(data->dispclk_required_for_dram_speed_change[data->y_clk_level][s_low], vbios->low_voltage_max_dispclk) && data->num_displays_with_margin[data->y_clk_level][s_low] == number_of_displays_enabled_with_margin))) {
+			sclk_message = bw_def_low;
+			data->sclk_level = s_low;
+			data->required_sclk = vbios->low_sclk;
+		}
+		else if (bw_ltn(data->total_average_bandwidth_no_compression, bw_mul(bw_mul(bw_frc_to_fixed(dceip->max_average_percent_of_ideal_port_bw_display_can_use_in_normal_system_operation, 100),sclk[mid]),vbios->data_return_bus_width))
+				&& bw_ltn(data->required_sclk, sclk[s_mid1]) && (data->cpup_state_change_enable == bw_def_no || (bw_mtn(data->blackout_duration_margin[data->y_clk_level][s_mid1], bw_int_to_fixed(0)) && bw_ltn(data->dispclk_required_for_blackout_duration[data->y_clk_level][s_mid1], vbios->high_voltage_max_dispclk))) && (data->cpuc_state_change_enable == bw_def_no || (bw_mtn(data->blackout_duration_margin[data->y_clk_level][s_mid1], bw_int_to_fixed(0)) && bw_ltn(data->dispclk_required_for_blackout_duration[data->y_clk_level][s_mid1], vbios->high_voltage_max_dispclk) && bw_ltn(data->dispclk_required_for_blackout_recovery[data->y_clk_level][s_mid1], vbios->high_voltage_max_dispclk))) && (!data->increase_voltage_to_support_mclk_switch || data->nbp_state_change_enable == bw_def_no || (bw_mtn(data->min_dram_speed_change_margin[data->y_clk_level][s_mid1], bw_int_to_fixed(0)) && bw_ltn(data->min_dram_speed_change_margin[data->y_clk_level][s_mid1], bw_int_to_fixed(9999)) && bw_leq(data->dispclk_required_for_dram_speed_change[data->y_clk_level][s_mid1], vbios->mid_voltage_max_dispclk) && data->num_displays_with_margin[data->y_clk_level][s_mid1] == number_of_displays_enabled_with_margin))) {
+			sclk_message = bw_def_mid;
+			data->sclk_level = s_mid1;
+			data->required_sclk = vbios->mid1_sclk;
+		}
+		else if (bw_ltn(data->total_average_bandwidth_no_compression, bw_mul(bw_mul(bw_frc_to_fixed(dceip->max_average_percent_of_ideal_port_bw_display_can_use_in_normal_system_operation, 100),sclk[s_mid2]),vbios->data_return_bus_width))
+				&& bw_ltn(data->required_sclk, sclk[s_mid2]) && (data->cpup_state_change_enable == bw_def_no || (bw_mtn(data->blackout_duration_margin[data->y_clk_level][s_mid2], bw_int_to_fixed(0)) && bw_ltn(data->dispclk_required_for_blackout_duration[data->y_clk_level][s_mid2], vbios->high_voltage_max_dispclk))) && (data->cpuc_state_change_enable == bw_def_no || (bw_mtn(data->blackout_duration_margin[data->y_clk_level][s_mid2], bw_int_to_fixed(0)) && bw_ltn(data->dispclk_required_for_blackout_duration[data->y_clk_level][s_mid2], vbios->high_voltage_max_dispclk) && bw_ltn(data->dispclk_required_for_blackout_recovery[data->y_clk_level][s_mid2], vbios->high_voltage_max_dispclk))) && (!data->increase_voltage_to_support_mclk_switch || data->nbp_state_change_enable == bw_def_no || (bw_mtn(data->min_dram_speed_change_margin[data->y_clk_level][s_mid2], bw_int_to_fixed(0)) && bw_ltn(data->min_dram_speed_change_margin[data->y_clk_level][s_mid2], bw_int_to_fixed(9999)) && bw_leq(data->dispclk_required_for_dram_speed_change[data->y_clk_level][s_mid2], vbios->mid_voltage_max_dispclk) && data->num_displays_with_margin[data->y_clk_level][s_mid2] == number_of_displays_enabled_with_margin))) {
+			sclk_message = bw_def_mid;
+			data->sclk_level = s_mid2;
+			data->required_sclk = vbios->mid2_sclk;
+		}
+		else if (bw_ltn(data->total_average_bandwidth_no_compression, bw_mul(bw_mul(bw_frc_to_fixed(dceip->max_average_percent_of_ideal_port_bw_display_can_use_in_normal_system_operation, 100),sclk[s_mid3]),vbios->data_return_bus_width))
+				&& bw_ltn(data->required_sclk, sclk[s_mid3]) && (data->cpup_state_change_enable == bw_def_no || (bw_mtn(data->blackout_duration_margin[data->y_clk_level][s_mid3], bw_int_to_fixed(0)) && bw_ltn(data->dispclk_required_for_blackout_duration[data->y_clk_level][s_mid3], vbios->high_voltage_max_dispclk))) && (data->cpuc_state_change_enable == bw_def_no || (bw_mtn(data->blackout_duration_margin[data->y_clk_level][s_mid3], bw_int_to_fixed(0)) && bw_ltn(data->dispclk_required_for_blackout_duration[data->y_clk_level][s_mid3], vbios->high_voltage_max_dispclk) && bw_ltn(data->dispclk_required_for_blackout_recovery[data->y_clk_level][s_mid3], vbios->high_voltage_max_dispclk))) && (!data->increase_voltage_to_support_mclk_switch || data->nbp_state_change_enable == bw_def_no || (bw_mtn(data->min_dram_speed_change_margin[data->y_clk_level][s_mid3], bw_int_to_fixed(0)) && bw_ltn(data->min_dram_speed_change_margin[data->y_clk_level][s_mid3], bw_int_to_fixed(9999)) && bw_leq(data->dispclk_required_for_dram_speed_change[data->y_clk_level][s_mid3], vbios->mid_voltage_max_dispclk) && data->num_displays_with_margin[data->y_clk_level][s_mid3] == number_of_displays_enabled_with_margin))) {
+			sclk_message = bw_def_mid;
+			data->sclk_level = s_mid3;
+			data->required_sclk = vbios->mid3_sclk;
+		}
+		else if (bw_ltn(data->total_average_bandwidth_no_compression, bw_mul(bw_mul(bw_frc_to_fixed(dceip->max_average_percent_of_ideal_port_bw_display_can_use_in_normal_system_operation, 100),sclk[s_mid4]),vbios->data_return_bus_width))
+				&& bw_ltn(data->required_sclk, sclk[s_mid4]) && (data->cpup_state_change_enable == bw_def_no || (bw_mtn(data->blackout_duration_margin[data->y_clk_level][s_mid4], bw_int_to_fixed(0)) && bw_ltn(data->dispclk_required_for_blackout_duration[data->y_clk_level][s_mid4], vbios->high_voltage_max_dispclk))) && (data->cpuc_state_change_enable == bw_def_no || (bw_mtn(data->blackout_duration_margin[data->y_clk_level][s_mid4], bw_int_to_fixed(0)) && bw_ltn(data->dispclk_required_for_blackout_duration[data->y_clk_level][s_mid4], vbios->high_voltage_max_dispclk) && bw_ltn(data->dispclk_required_for_blackout_recovery[data->y_clk_level][s_mid4], vbios->high_voltage_max_dispclk))) && (!data->increase_voltage_to_support_mclk_switch || data->nbp_state_change_enable == bw_def_no || (bw_mtn(data->min_dram_speed_change_margin[data->y_clk_level][s_mid4], bw_int_to_fixed(0)) && bw_ltn(data->min_dram_speed_change_margin[data->y_clk_level][s_mid4], bw_int_to_fixed(9999)) && bw_leq(data->dispclk_required_for_dram_speed_change[data->y_clk_level][s_mid4], vbios->mid_voltage_max_dispclk) && data->num_displays_with_margin[data->y_clk_level][s_mid4] == number_of_displays_enabled_with_margin))) {
+			sclk_message = bw_def_mid;
+			data->sclk_level = s_mid4;
+			data->required_sclk = vbios->mid4_sclk;
+		}
+		else if (bw_ltn(data->total_average_bandwidth_no_compression, bw_mul(bw_mul(bw_frc_to_fixed(dceip->max_average_percent_of_ideal_port_bw_display_can_use_in_normal_system_operation, 100),sclk[s_mid5]),vbios->data_return_bus_width))
+				&& bw_ltn(data->required_sclk, sclk[s_mid5]) && (data->cpup_state_change_enable == bw_def_no || (bw_mtn(data->blackout_duration_margin[data->y_clk_level][s_mid5], bw_int_to_fixed(0)) && bw_ltn(data->dispclk_required_for_blackout_duration[data->y_clk_level][s_mid5], vbios->high_voltage_max_dispclk))) && (data->cpuc_state_change_enable == bw_def_no || (bw_mtn(data->blackout_duration_margin[data->y_clk_level][s_mid5], bw_int_to_fixed(0)) && bw_ltn(data->dispclk_required_for_blackout_duration[data->y_clk_level][s_mid5], vbios->high_voltage_max_dispclk) && bw_ltn(data->dispclk_required_for_blackout_recovery[data->y_clk_level][s_mid5], vbios->high_voltage_max_dispclk))) && (!data->increase_voltage_to_support_mclk_switch || data->nbp_state_change_enable == bw_def_no || (bw_mtn(data->min_dram_speed_change_margin[data->y_clk_level][s_mid5], bw_int_to_fixed(0)) && bw_ltn(data->min_dram_speed_change_margin[data->y_clk_level][s_mid5], bw_int_to_fixed(9999)) && bw_leq(data->dispclk_required_for_dram_speed_change[data->y_clk_level][s_mid5], vbios->mid_voltage_max_dispclk) && data->num_displays_with_margin[data->y_clk_level][s_mid5] == number_of_displays_enabled_with_margin))) {
+			sclk_message = bw_def_mid;
+			data->sclk_level = s_mid5;
+			data->required_sclk = vbios->mid5_sclk;
+		}
+		else if (bw_ltn(data->total_average_bandwidth_no_compression, bw_mul(bw_mul(bw_frc_to_fixed(dceip->max_average_percent_of_ideal_port_bw_display_can_use_in_normal_system_operation, 100),sclk[s_mid6]),vbios->data_return_bus_width))
+				&& bw_ltn(data->required_sclk, sclk[s_mid6]) && (data->cpup_state_change_enable == bw_def_no || (bw_mtn(data->blackout_duration_margin[data->y_clk_level][s_mid6], bw_int_to_fixed(0)) && bw_ltn(data->dispclk_required_for_blackout_duration[data->y_clk_level][s_mid6], vbios->high_voltage_max_dispclk))) && (data->cpuc_state_change_enable == bw_def_no || (bw_mtn(data->blackout_duration_margin[data->y_clk_level][s_mid6], bw_int_to_fixed(0)) && bw_ltn(data->dispclk_required_for_blackout_duration[data->y_clk_level][s_mid6], vbios->high_voltage_max_dispclk) && bw_ltn(data->dispclk_required_for_blackout_recovery[data->y_clk_level][s_mid6], vbios->high_voltage_max_dispclk))) && (!data->increase_voltage_to_support_mclk_switch || data->nbp_state_change_enable == bw_def_no || (bw_mtn(data->min_dram_speed_change_margin[data->y_clk_level][s_mid6], bw_int_to_fixed(0)) && bw_ltn(data->min_dram_speed_change_margin[data->y_clk_level][s_mid6], bw_int_to_fixed(9999)) && bw_leq(data->dispclk_required_for_dram_speed_change[data->y_clk_level][s_mid6], vbios->high_voltage_max_dispclk) && data->num_displays_with_margin[data->y_clk_level][s_mid6] == number_of_displays_enabled_with_margin))) {
+			sclk_message = bw_def_mid;
+			data->sclk_level = s_mid6;
+			data->required_sclk = vbios->mid6_sclk;
+		}
+		else if (bw_ltn(data->total_average_bandwidth_no_compression, bw_mul(bw_mul(bw_frc_to_fixed(dceip->max_average_percent_of_ideal_port_bw_display_can_use_in_normal_system_operation, 100),sclk[s_high]),vbios->data_return_bus_width))
+				&& bw_ltn(data->required_sclk, sclk[s_high])) {
+			sclk_message = bw_def_high;
+			data->sclk_level = s_high;
+			data->required_sclk = vbios->high_sclk;
+		}
+		else if (bw_meq(data->total_average_bandwidth_no_compression, bw_mul(bw_mul(bw_frc_to_fixed(dceip->max_average_percent_of_ideal_port_bw_display_can_use_in_normal_system_operation, 100),sclk[s_high]),vbios->data_return_bus_width))
+				&& bw_ltn(data->required_sclk, sclk[s_high])) {
+			sclk_message = bw_def_high;
+			data->sclk_level = s_high;
+			data->required_sclk = vbios->high_sclk;
+		}
+		else {
+			sclk_message = bw_def_exceeded_allowed_maximum_sclk;
+			data->sclk_level = s_high;
+			/*required_sclk = high_sclk*/
+		}
+	}
+	/*dispclk*/
+	/*if dispclk is set to the maximum, ramping is not required.  dispclk required without ramping is less than the dispclk required with ramping.*/
+	/*if dispclk required without ramping is more than the maximum dispclk, that is the dispclk required, and the mode is not supported*/
+	/*if that does not happen, but dispclk required with ramping is more than the maximum dispclk, dispclk required is just the maximum dispclk*/
+	/*if that does not happen either, dispclk required is the dispclk required with ramping.*/
+	/*dispclk required without ramping is the maximum of the one required for display pipe pixel throughput, for scaler throughput, for total read request thrrougput and for dram/np p-state change if enabled.*/
+	/*the display pipe pixel throughput is the maximum of lines in per line out in the beginning of the frame and lines in per line out in the middle of the frame multiplied by the horizontal blank and chunk granularity factor, altogether multiplied by the ratio of the source width to the line time, divided by the line buffer pixels per dispclk throughput, and multiplied by the display pipe throughput factor.*/
+	/*the horizontal blank and chunk granularity factor is the ratio of the line time divided by the line time minus half the horizontal blank and chunk time.  it applies when the lines in per line out is not 2 or 4.*/
+	/*the dispclk required for scaler throughput is the product of the pixel rate and the scaling limits factor.*/
+	/*the dispclk required for total read request throughput is the product of the peak request-per-second bandwidth and the dispclk cycles per request, divided by the request efficiency.*/
+	/*for the dispclk required with ramping, instead of multiplying just the pipe throughput by the display pipe throughput factor, we multiply the scaler and pipe throughput by the ramping factor.*/
+	/*the scaling limits factor is the product of the horizontal scale ratio, and the ratio of the vertical taps divided by the scaler efficiency clamped to at least 1.*/
+	/*the scaling limits factor itself it also clamped to at least 1*/
+	/*if doing downscaling with the pre-downscaler enabled, the horizontal scale ratio should not be considered above (use "1")*/
+	data->downspread_factor = bw_add(bw_int_to_fixed(1), bw_div(vbios->down_spread_percentage, bw_int_to_fixed(100)));
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if (surface_type[i] == bw_def_graphics) {
+				switch (data->lb_bpc[i]) {
+				case 6:
+					data->v_scaler_efficiency = dceip->graphics_vscaler_efficiency6_bit_per_component;
+					break;
+				case 8:
+					data->v_scaler_efficiency = dceip->graphics_vscaler_efficiency8_bit_per_component;
+					break;
+				case 10:
+					data->v_scaler_efficiency = dceip->graphics_vscaler_efficiency10_bit_per_component;
+					break;
+				default:
+					data->v_scaler_efficiency = dceip->graphics_vscaler_efficiency12_bit_per_component;
+					break;
+				}
+				if (data->use_alpha[i] == 1) {
+					data->v_scaler_efficiency = bw_min2(data->v_scaler_efficiency, dceip->alpha_vscaler_efficiency);
+				}
+			}
+			else {
+				switch (data->lb_bpc[i]) {
+				case 6:
+					data->v_scaler_efficiency = dceip->underlay_vscaler_efficiency6_bit_per_component;
+					break;
+				case 8:
+					data->v_scaler_efficiency = dceip->underlay_vscaler_efficiency8_bit_per_component;
+					break;
+				case 10:
+					data->v_scaler_efficiency = dceip->underlay_vscaler_efficiency10_bit_per_component;
+					break;
+				default:
+					data->v_scaler_efficiency = dceip->underlay_vscaler_efficiency12_bit_per_component;
+					break;
+				}
+			}
+			if (dceip->pre_downscaler_enabled && bw_mtn(data->hsr[i], bw_int_to_fixed(1))) {
+				data->scaler_limits_factor = bw_max2(bw_div(data->v_taps[i], data->v_scaler_efficiency), bw_div(data->source_width_rounded_up_to_chunks[i], data->h_total[i]));
+			}
+			else {
+				data->scaler_limits_factor = bw_max3(bw_int_to_fixed(1), bw_ceil2(bw_div(data->h_taps[i], bw_int_to_fixed(4)), bw_int_to_fixed(1)), bw_mul(data->hsr[i], bw_max2(bw_div(data->v_taps[i], data->v_scaler_efficiency), bw_int_to_fixed(1))));
+			}
+			data->display_pipe_pixel_throughput = bw_div(bw_div(bw_mul(bw_max2(data->lb_lines_in_per_line_out_in_beginning_of_frame[i], bw_mul(data->lb_lines_in_per_line_out_in_middle_of_frame[i], data->horizontal_blank_and_chunk_granularity_factor[i])), data->source_width_rounded_up_to_chunks[i]), (bw_div(data->h_total[i], data->pixel_rate[i]))), dceip->lb_write_pixels_per_dispclk);
+			data->dispclk_required_without_ramping[i] = bw_mul(data->downspread_factor, bw_max2(bw_mul(data->pixel_rate[i], data->scaler_limits_factor), bw_mul(dceip->display_pipe_throughput_factor, data->display_pipe_pixel_throughput)));
+			data->dispclk_required_with_ramping[i] = bw_mul(dceip->dispclk_ramping_factor, bw_max2(bw_mul(data->pixel_rate[i], data->scaler_limits_factor), data->display_pipe_pixel_throughput));
+		}
+	}
+	data->total_dispclk_required_with_ramping = bw_int_to_fixed(0);
+	data->total_dispclk_required_without_ramping = bw_int_to_fixed(0);
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if (bw_ltn(data->total_dispclk_required_with_ramping, data->dispclk_required_with_ramping[i])) {
+				data->total_dispclk_required_with_ramping = data->dispclk_required_with_ramping[i];
+			}
+			if (bw_ltn(data->total_dispclk_required_without_ramping, data->dispclk_required_without_ramping[i])) {
+				data->total_dispclk_required_without_ramping = data->dispclk_required_without_ramping[i];
+			}
+		}
+	}
+	data->total_read_request_bandwidth = bw_int_to_fixed(0);
+	data->total_write_request_bandwidth = bw_int_to_fixed(0);
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if (surface_type[i] != bw_def_display_write_back420_luma && surface_type[i] != bw_def_display_write_back420_chroma) {
+				data->total_read_request_bandwidth = bw_add(data->total_read_request_bandwidth, data->request_bandwidth[i]);
+			}
+			else {
+				data->total_write_request_bandwidth = bw_add(data->total_write_request_bandwidth, data->request_bandwidth[i]);
+			}
+		}
+	}
+	data->dispclk_required_for_total_read_request_bandwidth = bw_div(bw_mul(data->total_read_request_bandwidth, dceip->dispclk_per_request), dceip->request_efficiency);
+	data->total_dispclk_required_with_ramping_with_request_bandwidth = bw_max2(data->total_dispclk_required_with_ramping, data->dispclk_required_for_total_read_request_bandwidth);
+	data->total_dispclk_required_without_ramping_with_request_bandwidth = bw_max2(data->total_dispclk_required_without_ramping, data->dispclk_required_for_total_read_request_bandwidth);
+	if (data->cpuc_state_change_enable == bw_def_yes) {
+		data->total_dispclk_required_with_ramping_with_request_bandwidth = bw_max3(data->total_dispclk_required_with_ramping_with_request_bandwidth, data->dispclk_required_for_blackout_duration[data->y_clk_level][data->sclk_level], data->dispclk_required_for_blackout_recovery[data->y_clk_level][data->sclk_level]);
+		data->total_dispclk_required_without_ramping_with_request_bandwidth = bw_max3(data->total_dispclk_required_without_ramping_with_request_bandwidth, data->dispclk_required_for_blackout_duration[data->y_clk_level][data->sclk_level], data->dispclk_required_for_blackout_recovery[data->y_clk_level][data->sclk_level]);
+	}
+	if (data->cpup_state_change_enable == bw_def_yes) {
+		data->total_dispclk_required_with_ramping_with_request_bandwidth = bw_max2(data->total_dispclk_required_with_ramping_with_request_bandwidth, data->dispclk_required_for_blackout_duration[data->y_clk_level][data->sclk_level]);
+		data->total_dispclk_required_without_ramping_with_request_bandwidth = bw_max2(data->total_dispclk_required_without_ramping_with_request_bandwidth, data->dispclk_required_for_blackout_duration[data->y_clk_level][data->sclk_level]);
+	}
+	if (data->nbp_state_change_enable == bw_def_yes && data->increase_voltage_to_support_mclk_switch) {
+		data->total_dispclk_required_with_ramping_with_request_bandwidth = bw_max2(data->total_dispclk_required_with_ramping_with_request_bandwidth, data->dispclk_required_for_dram_speed_change[data->y_clk_level][data->sclk_level]);
+		data->total_dispclk_required_without_ramping_with_request_bandwidth = bw_max2(data->total_dispclk_required_without_ramping_with_request_bandwidth, data->dispclk_required_for_dram_speed_change[data->y_clk_level][data->sclk_level]);
+	}
+	if (bw_ltn(data->total_dispclk_required_with_ramping_with_request_bandwidth, vbios->high_voltage_max_dispclk)) {
+		data->dispclk = data->total_dispclk_required_with_ramping_with_request_bandwidth;
+	}
+	else if (bw_ltn(data->total_dispclk_required_without_ramping_with_request_bandwidth, vbios->high_voltage_max_dispclk)) {
+		data->dispclk = vbios->high_voltage_max_dispclk;
+	}
+	else {
+		data->dispclk = data->total_dispclk_required_without_ramping_with_request_bandwidth;
+	}
+	/* required core voltage*/
+	/* the core voltage required is low if sclk, yclk(pclk)and dispclk are within the low limits*/
+	/* otherwise, the core voltage required is medium if yclk (pclk) is within the low limit and sclk and dispclk are within the medium limit*/
+	/* otherwise, the core voltage required is high if the three clocks are within the high limits*/
+	/* otherwise, or if the mode is not supported, core voltage requirement is not applicable*/
+	if (pipe_check == bw_def_notok) {
+		voltage = bw_def_na;
+	}
+	else if (mode_check == bw_def_notok) {
+		voltage = bw_def_notok;
+	}
+	else if (bw_equ(bw_int_to_fixed(yclk_message), vbios->low_yclk) && sclk_message == bw_def_low && bw_ltn(data->dispclk, vbios->low_voltage_max_dispclk)) {
+		voltage = bw_def_0_72;
+	}
+	else if ((bw_equ(bw_int_to_fixed(yclk_message), vbios->low_yclk) || bw_equ(bw_int_to_fixed(yclk_message), vbios->mid_yclk)) && (sclk_message == bw_def_low || sclk_message == bw_def_mid) && bw_ltn(data->dispclk, vbios->mid_voltage_max_dispclk)) {
+		voltage = bw_def_0_8;
+	}
+	else if ((bw_equ(bw_int_to_fixed(yclk_message), vbios->low_yclk) || bw_equ(bw_int_to_fixed(yclk_message), vbios->mid_yclk) || bw_equ(bw_int_to_fixed(yclk_message), vbios->high_yclk)) && (sclk_message == bw_def_low || sclk_message == bw_def_mid || sclk_message == bw_def_high) && bw_leq(data->dispclk, vbios->high_voltage_max_dispclk)) {
+		if ((data->nbp_state_change_enable == bw_def_no && nbp_state_change_enable_blank == bw_def_no)) {
+			voltage = bw_def_high_no_nbp_state_change;
+		}
+		else {
+			voltage = bw_def_0_9;
+		}
+	}
+	else {
+		voltage = bw_def_notok;
+	}
+	if (voltage == bw_def_0_72) {
+		data->max_phyclk = vbios->low_voltage_max_phyclk;
+	}
+	else if (voltage == bw_def_0_8) {
+		data->max_phyclk = vbios->mid_voltage_max_phyclk;
+	}
+	else {
+		data->max_phyclk = vbios->high_voltage_max_phyclk;
+	}
+	/*required blackout recovery time*/
+	data->blackout_recovery_time = bw_int_to_fixed(0);
+	for (k = 0; k <= maximum_number_of_surfaces - 1; k++) {
+		if (data->enable[k] && bw_mtn(vbios->blackout_duration, bw_int_to_fixed(0)) && data->cpup_state_change_enable == bw_def_yes) {
+			if (surface_type[k] != bw_def_display_write_back420_luma && surface_type[k] != bw_def_display_write_back420_chroma) {
+				data->blackout_recovery_time = bw_max2(data->blackout_recovery_time, bw_add(bw_mul(bw_int_to_fixed(2), data->total_dmifmc_urgent_latency), data->dmif_burst_time[data->y_clk_level][data->sclk_level]));
+				if (bw_ltn(data->adjusted_data_buffer_size[k], bw_mul(bw_div(bw_mul(data->display_bandwidth[k], data->useful_bytes_per_request[k]), data->bytes_per_request[k]), (bw_add(vbios->blackout_duration, bw_add(bw_mul(bw_int_to_fixed(2), data->total_dmifmc_urgent_latency), data->dmif_burst_time[data->y_clk_level][data->sclk_level])))))) {
+					data->blackout_recovery_time = bw_max2(data->blackout_recovery_time, bw_div((bw_add(bw_mul(bw_div(bw_mul(data->display_bandwidth[k], data->useful_bytes_per_request[k]), data->bytes_per_request[k]), vbios->blackout_duration), bw_sub(bw_div(bw_mul(bw_mul(bw_mul((bw_add(bw_mul(bw_int_to_fixed(2), data->total_dmifmc_urgent_latency), data->dmif_burst_time[data->y_clk_level][data->sclk_level])), data->dispclk), bw_int_to_fixed(data->bytes_per_pixel[k])), data->lines_interleaved_in_mem_access[k]), data->latency_hiding_lines[k]), data->adjusted_data_buffer_size[k]))), (bw_sub(bw_div(bw_mul(bw_mul(data->dispclk, bw_int_to_fixed(data->bytes_per_pixel[k])), data->lines_interleaved_in_mem_access[k]), data->latency_hiding_lines[k]), bw_div(bw_mul(data->display_bandwidth[k], data->useful_bytes_per_request[k]), data->bytes_per_request[k])))));
+				}
+			}
+			else {
+				data->blackout_recovery_time = bw_max2(data->blackout_recovery_time, bw_add(bw_mul(bw_int_to_fixed(2), vbios->mcifwrmc_urgent_latency), data->mcifwr_burst_time[data->y_clk_level][data->sclk_level]));
+				if (bw_ltn(data->adjusted_data_buffer_size[k], bw_mul(bw_div(bw_mul(data->display_bandwidth[k], data->useful_bytes_per_request[k]), data->bytes_per_request[k]), (bw_add(vbios->blackout_duration, bw_add(bw_mul(bw_int_to_fixed(2), vbios->mcifwrmc_urgent_latency), data->mcifwr_burst_time[data->y_clk_level][data->sclk_level])))))) {
+					data->blackout_recovery_time = bw_max2(data->blackout_recovery_time, bw_div((bw_add(bw_mul(bw_div(bw_mul(data->display_bandwidth[k], data->useful_bytes_per_request[k]), data->bytes_per_request[k]), vbios->blackout_duration), bw_sub(bw_div(bw_mul(bw_mul(bw_mul((bw_add(bw_add(bw_mul(bw_int_to_fixed(2), vbios->mcifwrmc_urgent_latency), data->dmif_burst_time[data->y_clk_level][data->sclk_level]), data->mcifwr_burst_time[data->y_clk_level][data->sclk_level])), data->dispclk), bw_int_to_fixed(data->bytes_per_pixel[k])), data->lines_interleaved_in_mem_access[k]), data->latency_hiding_lines[k]), data->adjusted_data_buffer_size[k]))), (bw_sub(bw_div(bw_mul(bw_mul(data->dispclk, bw_int_to_fixed(data->bytes_per_pixel[k])), data->lines_interleaved_in_mem_access[k]), data->latency_hiding_lines[k]), bw_div(bw_mul(data->display_bandwidth[k], data->useful_bytes_per_request[k]), data->bytes_per_request[k])))));
+				}
+			}
+		}
+	}
+	/*sclk deep sleep*/
+	/*during self-refresh, sclk can be reduced to dispclk divided by the minimum pixels in the data fifo entry, with 15% margin, but shoudl not be set to less than the request bandwidth.*/
+	/*the data fifo entry is 16 pixels for the writeback, 64 bytes/bytes_per_pixel for the graphics, 16 pixels for the parallel rotation underlay,*/
+	/*and 16 bytes/bytes_per_pixel for the orthogonal rotation underlay.*/
+	/*in parallel mode (underlay pipe), the data read from the dmifv buffer is variable and based on the pixel depth (8bbp - 16 bytes, 16 bpp - 32 bytes, 32 bpp - 64 bytes)*/
+	/*in orthogonal mode (underlay pipe), the data read from the dmifv buffer is fixed at 16 bytes.*/
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if (surface_type[i] == bw_def_display_write_back420_luma || surface_type[i] == bw_def_display_write_back420_chroma) {
+				data->pixels_per_data_fifo_entry[i] = bw_int_to_fixed(16);
+			}
+			else if (surface_type[i] == bw_def_graphics) {
+				data->pixels_per_data_fifo_entry[i] = bw_div(bw_int_to_fixed(64), bw_int_to_fixed(data->bytes_per_pixel[i]));
+			}
+			else if (data->orthogonal_rotation[i] == 0) {
+				data->pixels_per_data_fifo_entry[i] = bw_int_to_fixed(16);
+			}
+			else {
+				data->pixels_per_data_fifo_entry[i] = bw_div(bw_int_to_fixed(16), bw_int_to_fixed(data->bytes_per_pixel[i]));
+			}
+		}
+	}
+	data->min_pixels_per_data_fifo_entry = bw_int_to_fixed(9999);
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if (bw_mtn(data->min_pixels_per_data_fifo_entry, data->pixels_per_data_fifo_entry[i])) {
+				data->min_pixels_per_data_fifo_entry = data->pixels_per_data_fifo_entry[i];
+			}
+		}
+	}
+	data->sclk_deep_sleep = bw_max2(bw_div(bw_mul(data->dispclk, bw_frc_to_fixed(115, 100)), data->min_pixels_per_data_fifo_entry), data->total_read_request_bandwidth);
+	/*urgent, stutter and nb-p_state watermark*/
+	/*the urgent watermark is the maximum of the urgent trip time plus the pixel transfer time, the urgent trip times to get data for the first pixel, and the urgent trip times to get data for the last pixel.*/
+	/*the stutter exit watermark is the self refresh exit time plus the maximum of the data burst time plus the pixel transfer time, the data burst times to get data for the first pixel, and the data burst times to get data for the last pixel.  it does not apply to the writeback.*/
+	/*the nb p-state change watermark is the dram speed/p-state change time plus the maximum of the data burst time plus the pixel transfer time, the data burst times to get data for the first pixel, and the data burst times to get data for the last pixel.*/
+	/*the pixel transfer time is the maximum of the time to transfer the source pixels required for the first output pixel, and the time to transfer the pixels for the last output pixel minus the active line time.*/
+	/*blackout_duration is added to the urgent watermark*/
+	data->chunk_request_time = bw_int_to_fixed(0);
+	data->cursor_request_time = bw_int_to_fixed(0);
+	/*compute total time to request one chunk from each active display pipe*/
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			data->chunk_request_time = bw_add(data->chunk_request_time, (bw_div((bw_div(bw_int_to_fixed(pixels_per_chunk * (int64_t)data->bytes_per_pixel[i]), data->useful_bytes_per_request[i])), bw_min2(sclk[data->sclk_level], bw_div(data->dispclk, bw_int_to_fixed(2))))));
+		}
+	}
+	/*compute total time to request cursor data*/
+	data->cursor_request_time = (bw_div(data->cursor_total_data, (bw_mul(bw_int_to_fixed(32), sclk[data->sclk_level]))));
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			data->line_source_pixels_transfer_time = bw_max2(bw_div(bw_div(data->src_pixels_for_first_output_pixel[i], dceip->lb_write_pixels_per_dispclk), (bw_div(data->dispclk, dceip->display_pipe_throughput_factor))), bw_sub(bw_div(bw_div(data->src_pixels_for_last_output_pixel[i], dceip->lb_write_pixels_per_dispclk), (bw_div(data->dispclk, dceip->display_pipe_throughput_factor))), data->active_time[i]));
+			if (surface_type[i] != bw_def_display_write_back420_luma && surface_type[i] != bw_def_display_write_back420_chroma) {
+				data->urgent_watermark[i] = bw_add(bw_add(bw_add(bw_add(bw_add(data->total_dmifmc_urgent_latency, data->dmif_burst_time[data->y_clk_level][data->sclk_level]), bw_max2(data->line_source_pixels_transfer_time, data->line_source_transfer_time[i][data->y_clk_level][data->sclk_level])), vbios->blackout_duration), data->chunk_request_time), data->cursor_request_time);
+				data->stutter_exit_watermark[i] = bw_add(bw_sub(vbios->stutter_self_refresh_exit_latency, data->total_dmifmc_urgent_latency), data->urgent_watermark[i]);
+				data->stutter_entry_watermark[i] = bw_add(bw_sub(bw_add(vbios->stutter_self_refresh_exit_latency, vbios->stutter_self_refresh_entry_latency), data->total_dmifmc_urgent_latency), data->urgent_watermark[i]);
+				/*unconditionally remove black out time from the nb p_state watermark*/
+				if (data->display_pstate_change_enable[i] == 1) {
+					data->nbp_state_change_watermark[i] = bw_add(bw_add(vbios->nbp_state_change_latency, data->dmif_burst_time[data->y_clk_level][data->sclk_level]), bw_max2(data->line_source_pixels_transfer_time, data->dram_speed_change_line_source_transfer_time[i][data->y_clk_level][data->sclk_level]));
+				}
+				else {
+					/*maximize the watermark to force the switch in the vb_lank region of the frame*/
+					data->nbp_state_change_watermark[i] = bw_int_to_fixed(131000);
+				}
+			}
+			else {
+				data->urgent_watermark[i] = bw_add(bw_add(bw_add(bw_add(bw_add(vbios->mcifwrmc_urgent_latency, data->mcifwr_burst_time[data->y_clk_level][data->sclk_level]), bw_max2(data->line_source_pixels_transfer_time, data->line_source_transfer_time[i][data->y_clk_level][data->sclk_level])), vbios->blackout_duration), data->chunk_request_time), data->cursor_request_time);
+				data->stutter_exit_watermark[i] = bw_int_to_fixed(0);
+				data->stutter_entry_watermark[i] = bw_int_to_fixed(0);
+				if (data->display_pstate_change_enable[i] == 1) {
+					data->nbp_state_change_watermark[i] = bw_add(bw_add(vbios->nbp_state_change_latency, data->mcifwr_burst_time[data->y_clk_level][data->sclk_level]), bw_max2(data->line_source_pixels_transfer_time, data->dram_speed_change_line_source_transfer_time[i][data->y_clk_level][data->sclk_level]));
+				}
+				else {
+					/*maximize the watermark to force the switch in the vb_lank region of the frame*/
+					data->nbp_state_change_watermark[i] = bw_int_to_fixed(131000);
+				}
+			}
+		}
+	}
+	/*stutter mode enable*/
+	/*in the multi-display case the stutter exit or entry watermark cannot exceed the minimum latency hiding capabilities of the*/
+	/*display pipe.*/
+	data->stutter_mode_enable = data->cpuc_state_change_enable;
+	if (data->number_of_displays > 1) {
+		for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+			if (data->enable[i]) {
+				if ((bw_mtn(data->stutter_exit_watermark[i], data->minimum_latency_hiding[i]) || bw_mtn(data->stutter_entry_watermark[i], data->minimum_latency_hiding[i]))) {
+					data->stutter_mode_enable = bw_def_no;
+				}
+			}
+		}
+	}
+	/*performance metrics*/
+	/* display read access efficiency (%)*/
+	/* display write back access efficiency (%)*/
+	/* stutter efficiency (%)*/
+	/* extra underlay pitch recommended for efficiency (pixels)*/
+	/* immediate flip time (us)*/
+	/* latency for other clients due to urgent display read (us)*/
+	/* latency for other clients due to urgent display write (us)*/
+	/* average bandwidth consumed by display (no compression) (gb/s)*/
+	/* required dram  bandwidth (gb/s)*/
+	/* required sclk (m_hz)*/
+	/* required rd urgent latency (us)*/
+	/* nb p-state change margin (us)*/
+	/*dmif and mcifwr dram access efficiency*/
+	/*is the ratio between the ideal dram access time (which is the data buffer size in memory divided by the dram bandwidth), and the actual time which is the total page close-open time.  but it cannot exceed the dram efficiency provided by the memory subsystem*/
+	data->dmifdram_access_efficiency = bw_min2(bw_div(bw_div(data->total_display_reads_required_dram_access_data, data->dram_bandwidth), data->dmif_total_page_close_open_time), bw_int_to_fixed(1));
+	if (bw_mtn(data->total_display_writes_required_dram_access_data, bw_int_to_fixed(0))) {
+		data->mcifwrdram_access_efficiency = bw_min2(bw_div(bw_div(data->total_display_writes_required_dram_access_data, data->dram_bandwidth), data->mcifwr_total_page_close_open_time), bw_int_to_fixed(1));
+	}
+	else {
+		data->mcifwrdram_access_efficiency = bw_int_to_fixed(0);
+	}
+	/*stutter efficiency*/
+	/*the stutter efficiency is the frame-average time in self-refresh divided by the frame-average stutter cycle duration.  only applies if the display write-back is not enabled.*/
+	/*the frame-average stutter cycle used is the minimum for all pipes of the frame-average data buffer size in time, times the compression rate*/
+	/*the frame-average time in self-refresh is the stutter cycle minus the self refresh exit latency and the burst time*/
+	/*the stutter cycle is the dmif buffer size reduced by the excess of the stutter exit watermark over the lb size in time.*/
+	/*the burst time is the data needed during the stutter cycle divided by the available bandwidth*/
+	/*compute the time read all the data from the dmif buffer to the lb (dram refresh period)*/
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			data->stutter_refresh_duration[i] = bw_sub(bw_mul(bw_div(bw_div(bw_mul(bw_div(bw_div(data->adjusted_data_buffer_size[i], bw_int_to_fixed(data->bytes_per_pixel[i])), data->source_width_rounded_up_to_chunks[i]), data->h_total[i]), data->vsr[i]), data->pixel_rate[i]), data->compression_rate[i]), bw_max2(bw_int_to_fixed(0), bw_sub(data->stutter_exit_watermark[i], bw_div(bw_mul((bw_sub(data->lb_partitions[i], bw_int_to_fixed(1))), data->h_total[i]), data->pixel_rate[i]))));
+			data->stutter_dmif_buffer_size[i] = bw_div(bw_mul(bw_mul(bw_div(bw_mul(bw_mul(data->stutter_refresh_duration[i], bw_int_to_fixed(data->bytes_per_pixel[i])), data->source_width_rounded_up_to_chunks[i]), data->h_total[i]), data->vsr[i]), data->pixel_rate[i]), data->compression_rate[i]);
+		}
+	}
+	data->min_stutter_refresh_duration = bw_int_to_fixed(9999);
+	data->total_stutter_dmif_buffer_size = 0;
+	data->total_bytes_requested = 0;
+	data->min_stutter_dmif_buffer_size = 9999;
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			if (bw_mtn(data->min_stutter_refresh_duration, data->stutter_refresh_duration[i])) {
+				data->min_stutter_refresh_duration = data->stutter_refresh_duration[i];
+				data->total_bytes_requested = bw_fixed_to_int(bw_add(bw_int_to_fixed(data->total_bytes_requested), (bw_mul(bw_mul(data->source_height_rounded_up_to_chunks[i], data->source_width_rounded_up_to_chunks[i]), bw_int_to_fixed(data->bytes_per_pixel[i])))));
+				data->min_stutter_dmif_buffer_size = bw_fixed_to_int(data->stutter_dmif_buffer_size[i]);
+			}
+			data->total_stutter_dmif_buffer_size = bw_fixed_to_int(bw_add(data->stutter_dmif_buffer_size[i], bw_int_to_fixed(data->total_stutter_dmif_buffer_size)));
+		}
+	}
+	data->stutter_burst_time = bw_div(bw_int_to_fixed(data->total_stutter_dmif_buffer_size), bw_mul(sclk[data->sclk_level], vbios->data_return_bus_width));
+	data->num_stutter_bursts = data->total_bytes_requested / data->min_stutter_dmif_buffer_size;
+	data->total_stutter_cycle_duration = bw_add(bw_add(data->min_stutter_refresh_duration, vbios->stutter_self_refresh_exit_latency), data->stutter_burst_time);
+	data->time_in_self_refresh = data->min_stutter_refresh_duration;
+	if (data->d1_display_write_back_dwb_enable == 1) {
+		data->stutter_efficiency = bw_int_to_fixed(0);
+	}
+	else if (bw_ltn(data->time_in_self_refresh, bw_int_to_fixed(0))) {
+		data->stutter_efficiency = bw_int_to_fixed(0);
+	}
+	else {
+		/*compute stutter efficiency assuming 60 hz refresh rate*/
+		data->stutter_efficiency = bw_max2(bw_int_to_fixed(0), bw_mul((bw_sub(bw_int_to_fixed(1), (bw_div(bw_mul((bw_add(vbios->stutter_self_refresh_exit_latency, data->stutter_burst_time)), bw_int_to_fixed(data->num_stutter_bursts)), bw_frc_to_fixed(166666667, 10000))))), bw_int_to_fixed(100)));
+	}
+	/*immediate flip time*/
+	/*if scatter gather is enabled, the immediate flip takes a number of urgent memory trips equivalent to the pte requests in a row divided by the pte request limit.*/
+	/*otherwise, it may take just one urgenr memory trip*/
+	data->worst_number_of_trips_to_memory = bw_int_to_fixed(1);
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i] && data->scatter_gather_enable_for_pipe[i] == 1) {
+			data->number_of_trips_to_memory_for_getting_apte_row[i] = bw_ceil2(bw_div(data->scatter_gather_pte_requests_in_row[i], data->scatter_gather_pte_request_limit[i]), bw_int_to_fixed(1));
+			if (bw_ltn(data->worst_number_of_trips_to_memory, data->number_of_trips_to_memory_for_getting_apte_row[i])) {
+				data->worst_number_of_trips_to_memory = data->number_of_trips_to_memory_for_getting_apte_row[i];
+			}
+		}
+	}
+	data->immediate_flip_time = bw_mul(data->worst_number_of_trips_to_memory, data->total_dmifmc_urgent_latency);
+	/*worst latency for other clients*/
+	/*it is the urgent latency plus the urgent burst time*/
+	data->latency_for_non_dmif_clients = bw_add(data->total_dmifmc_urgent_latency, data->dmif_burst_time[data->y_clk_level][data->sclk_level]);
+	if (data->d1_display_write_back_dwb_enable == 1) {
+		data->latency_for_non_mcifwr_clients = bw_add(vbios->mcifwrmc_urgent_latency, dceip->mcifwr_all_surfaces_burst_time);
+	}
+	else {
+		data->latency_for_non_mcifwr_clients = bw_int_to_fixed(0);
+	}
+	/*dmif mc urgent latency supported in high sclk and yclk*/
+	data->dmifmc_urgent_latency_supported_in_high_sclk_and_yclk = bw_div((bw_sub(data->min_read_buffer_size_in_time, data->dmif_burst_time[high][s_high])), data->total_dmifmc_urgent_trips);
+	/*dram speed/p-state change margin*/
+	/*in the multi-display case the nb p-state change watermark cannot exceed the average lb size plus the dmif size or the cursor dcp buffer size*/
+	data->v_blank_nbp_state_dram_speed_change_latency_supported = bw_int_to_fixed(99999);
+	data->nbp_state_dram_speed_change_latency_supported = bw_int_to_fixed(99999);
+	for (i = 0; i <= maximum_number_of_surfaces - 1; i++) {
+		if (data->enable[i]) {
+			data->nbp_state_dram_speed_change_latency_supported = bw_min2(data->nbp_state_dram_speed_change_latency_supported, bw_add(bw_sub(data->maximum_latency_hiding_with_cursor[i], data->nbp_state_change_watermark[i]), vbios->nbp_state_change_latency));
+			data->v_blank_nbp_state_dram_speed_change_latency_supported = bw_min2(data->v_blank_nbp_state_dram_speed_change_latency_supported, bw_add(bw_sub(bw_div(bw_mul((bw_sub(data->v_total[i], bw_sub(bw_div(data->src_height[i], data->v_scale_ratio[i]), bw_int_to_fixed(4)))), data->h_total[i]), data->pixel_rate[i]), data->nbp_state_change_watermark[i]), vbios->nbp_state_change_latency));
+		}
+	}
+	/*sclk required vs urgent latency*/
+	for (i = 1; i <= 5; i++) {
+		data->display_reads_time_for_data_transfer_and_urgent_latency = bw_sub(data->min_read_buffer_size_in_time, bw_mul(data->total_dmifmc_urgent_trips, bw_int_to_fixed(i)));
+		if (pipe_check == bw_def_ok && (bw_mtn(data->display_reads_time_for_data_transfer_and_urgent_latency, data->dmif_total_page_close_open_time))) {
+			data->dmif_required_sclk_for_urgent_latency[i] = bw_div(bw_div(data->total_display_reads_required_data, data->display_reads_time_for_data_transfer_and_urgent_latency), (bw_mul(vbios->data_return_bus_width, bw_frc_to_fixed(dceip->percent_of_ideal_port_bw_received_after_urgent_latency, 100))));
+		}
+		else {
+			data->dmif_required_sclk_for_urgent_latency[i] = bw_int_to_fixed(bw_def_na);
+		}
+	}
+	/*output link bit per pixel supported*/
+	for (k = 0; k <= maximum_number_of_surfaces - 1; k++) {
+		data->output_bpphdmi[k] = bw_def_na;
+		data->output_bppdp4_lane_hbr[k] = bw_def_na;
+		data->output_bppdp4_lane_hbr2[k] = bw_def_na;
+		data->output_bppdp4_lane_hbr3[k] = bw_def_na;
+		if (data->enable[k]) {
+			data->output_bpphdmi[k] = bw_fixed_to_int(bw_mul(bw_div(bw_min2(bw_int_to_fixed(600), data->max_phyclk), data->pixel_rate[k]), bw_int_to_fixed(24)));
+			if (bw_meq(data->max_phyclk, bw_int_to_fixed(270))) {
+				data->output_bppdp4_lane_hbr[k] = bw_fixed_to_int(bw_mul(bw_div(bw_mul(bw_int_to_fixed(270), bw_int_to_fixed(4)), data->pixel_rate[k]), bw_int_to_fixed(8)));
+			}
+			if (bw_meq(data->max_phyclk, bw_int_to_fixed(540))) {
+				data->output_bppdp4_lane_hbr2[k] = bw_fixed_to_int(bw_mul(bw_div(bw_mul(bw_int_to_fixed(540), bw_int_to_fixed(4)), data->pixel_rate[k]), bw_int_to_fixed(8)));
+			}
+			if (bw_meq(data->max_phyclk, bw_int_to_fixed(810))) {
+				data->output_bppdp4_lane_hbr3[k] = bw_fixed_to_int(bw_mul(bw_div(bw_mul(bw_int_to_fixed(810), bw_int_to_fixed(4)), data->pixel_rate[k]), bw_int_to_fixed(8)));
+			}
+		}
+	}
+
+	kfree(surface_type);
+free_tiling_mode:
+	kfree(tiling_mode);
+free_sclk:
+	kfree(sclk);
+free_yclk:
+	kfree(yclk);
+}
+
+/*******************************************************************************
+ * Public functions
+ ******************************************************************************/
+void bw_calcs_init(struct bw_calcs_dceip *bw_dceip,
+	struct bw_calcs_vbios *bw_vbios,
+	struct hw_asic_id asic_id)
+{
+	struct bw_calcs_dceip *dceip;
+	struct bw_calcs_vbios *vbios;
+
+	enum bw_calcs_version version = bw_calcs_version_from_asic_id(asic_id);
+
+	dceip = kzalloc(sizeof(*dceip), GFP_KERNEL);
+	if (!dceip)
+		return;
+
+	vbios = kzalloc(sizeof(*vbios), GFP_KERNEL);
+	if (!vbios) {
+		kfree(dceip);
+		return;
+	}
+
+	dceip->version = version;
+
+	switch (version) {
+	case BW_CALCS_VERSION_CARRIZO:
+		vbios->memory_type = bw_def_gddr5;
+		vbios->dram_channel_width_in_bits = 64;
+		vbios->number_of_dram_channels = asic_id.vram_width / vbios->dram_channel_width_in_bits;
+		vbios->number_of_dram_banks = 8;
+		vbios->high_yclk = bw_int_to_fixed(1600);
+		vbios->mid_yclk = bw_int_to_fixed(1600);
+		vbios->low_yclk = bw_frc_to_fixed(66666, 100);
+		vbios->low_sclk = bw_int_to_fixed(200);
+		vbios->mid1_sclk = bw_int_to_fixed(300);
+		vbios->mid2_sclk = bw_int_to_fixed(300);
+		vbios->mid3_sclk = bw_int_to_fixed(300);
+		vbios->mid4_sclk = bw_int_to_fixed(300);
+		vbios->mid5_sclk = bw_int_to_fixed(300);
+		vbios->mid6_sclk = bw_int_to_fixed(300);
+		vbios->high_sclk = bw_frc_to_fixed(62609, 100);
+		vbios->low_voltage_max_dispclk = bw_int_to_fixed(352);
+		vbios->mid_voltage_max_dispclk = bw_int_to_fixed(467);
+		vbios->high_voltage_max_dispclk = bw_int_to_fixed(643);
+		vbios->low_voltage_max_phyclk = bw_int_to_fixed(540);
+		vbios->mid_voltage_max_phyclk = bw_int_to_fixed(810);
+		vbios->high_voltage_max_phyclk = bw_int_to_fixed(810);
+		vbios->data_return_bus_width = bw_int_to_fixed(32);
+		vbios->trc = bw_int_to_fixed(50);
+		vbios->dmifmc_urgent_latency = bw_int_to_fixed(4);
+		vbios->stutter_self_refresh_exit_latency = bw_frc_to_fixed(153, 10);
+		vbios->stutter_self_refresh_entry_latency = bw_int_to_fixed(0);
+		vbios->nbp_state_change_latency = bw_frc_to_fixed(19649, 1000);
+		vbios->mcifwrmc_urgent_latency = bw_int_to_fixed(10);
+		vbios->scatter_gather_enable = true;
+		vbios->down_spread_percentage = bw_frc_to_fixed(5, 10);
+		vbios->cursor_width = 32;
+		vbios->average_compression_rate = 4;
+		vbios->number_of_request_slots_gmc_reserves_for_dmif_per_channel = 256;
+		vbios->blackout_duration = bw_int_to_fixed(0); /* us */
+		vbios->maximum_blackout_recovery_time = bw_int_to_fixed(0);
+
+		dceip->max_average_percent_of_ideal_port_bw_display_can_use_in_normal_system_operation = 100;
+		dceip->max_average_percent_of_ideal_drambw_display_can_use_in_normal_system_operation = 100;
+		dceip->percent_of_ideal_port_bw_received_after_urgent_latency = 100;
+		dceip->large_cursor = false;
+		dceip->dmif_request_buffer_size = bw_int_to_fixed(768);
+		dceip->dmif_pipe_en_fbc_chunk_tracker = false;
+		dceip->cursor_max_outstanding_group_num = 1;
+		dceip->lines_interleaved_into_lb = 2;
+		dceip->chunk_width = 256;
+		dceip->number_of_graphics_pipes = 3;
+		dceip->number_of_underlay_pipes = 1;
+		dceip->low_power_tiling_mode = 0;
+		dceip->display_write_back_supported = false;
+		dceip->argb_compression_support = false;
+		dceip->underlay_vscaler_efficiency6_bit_per_component =
+			bw_frc_to_fixed(35556, 10000);
+		dceip->underlay_vscaler_efficiency8_bit_per_component =
+			bw_frc_to_fixed(34286, 10000);
+		dceip->underlay_vscaler_efficiency10_bit_per_component =
+			bw_frc_to_fixed(32, 10);
+		dceip->underlay_vscaler_efficiency12_bit_per_component =
+			bw_int_to_fixed(3);
+		dceip->graphics_vscaler_efficiency6_bit_per_component =
+			bw_frc_to_fixed(35, 10);
+		dceip->graphics_vscaler_efficiency8_bit_per_component =
+			bw_frc_to_fixed(34286, 10000);
+		dceip->graphics_vscaler_efficiency10_bit_per_component =
+			bw_frc_to_fixed(32, 10);
+		dceip->graphics_vscaler_efficiency12_bit_per_component =
+			bw_int_to_fixed(3);
+		dceip->alpha_vscaler_efficiency = bw_int_to_fixed(3);
+		dceip->max_dmif_buffer_allocated = 2;
+		dceip->graphics_dmif_size = 12288;
+		dceip->underlay_luma_dmif_size = 19456;
+		dceip->underlay_chroma_dmif_size = 23552;
+		dceip->pre_downscaler_enabled = true;
+		dceip->underlay_downscale_prefetch_enabled = true;
+		dceip->lb_write_pixels_per_dispclk = bw_int_to_fixed(1);
+		dceip->lb_size_per_component444 = bw_int_to_fixed(82176);
+		dceip->graphics_lb_nodownscaling_multi_line_prefetching = false;
+		dceip->stutter_and_dram_clock_state_change_gated_before_cursor =
+			bw_int_to_fixed(0);
+		dceip->underlay420_luma_lb_size_per_component = bw_int_to_fixed(
+			82176);
+		dceip->underlay420_chroma_lb_size_per_component =
+			bw_int_to_fixed(164352);
+		dceip->underlay422_lb_size_per_component = bw_int_to_fixed(
+			82176);
+		dceip->cursor_chunk_width = bw_int_to_fixed(64);
+		dceip->cursor_dcp_buffer_lines = bw_int_to_fixed(4);
+		dceip->underlay_maximum_width_efficient_for_tiling =
+			bw_int_to_fixed(1920);
+		dceip->underlay_maximum_height_efficient_for_tiling =
+			bw_int_to_fixed(1080);
+		dceip->peak_pte_request_to_eviction_ratio_limiting_multiple_displays_or_single_rotated_display =
+			bw_frc_to_fixed(3, 10);
+		dceip->peak_pte_request_to_eviction_ratio_limiting_single_display_no_rotation =
+			bw_int_to_fixed(25);
+		dceip->minimum_outstanding_pte_request_limit = bw_int_to_fixed(
+			2);
+		dceip->maximum_total_outstanding_pte_requests_allowed_by_saw =
+			bw_int_to_fixed(128);
+		dceip->limit_excessive_outstanding_dmif_requests = true;
+		dceip->linear_mode_line_request_alternation_slice =
+			bw_int_to_fixed(64);
+		dceip->scatter_gather_lines_of_pte_prefetching_in_linear_mode =
+			32;
+		dceip->display_write_back420_luma_mcifwr_buffer_size = 12288;
+		dceip->display_write_back420_chroma_mcifwr_buffer_size = 8192;
+		dceip->request_efficiency = bw_frc_to_fixed(8, 10);
+		dceip->dispclk_per_request = bw_int_to_fixed(2);
+		dceip->dispclk_ramping_factor = bw_frc_to_fixed(105, 100);
+		dceip->display_pipe_throughput_factor = bw_frc_to_fixed(105, 100);
+		dceip->scatter_gather_pte_request_rows_in_tiling_mode = 2;
+		dceip->mcifwr_all_surfaces_burst_time = bw_int_to_fixed(0); /* todo: this is a bug*/
+		break;
+	case BW_CALCS_VERSION_POLARIS10:
+		/* TODO: Treat VEGAM the same as P10 for now
+		 * Need to tune the para for VEGAM if needed */
+	case BW_CALCS_VERSION_VEGAM:
+		vbios->memory_type = bw_def_gddr5;
+		vbios->dram_channel_width_in_bits = 32;
+		vbios->number_of_dram_channels = asic_id.vram_width / vbios->dram_channel_width_in_bits;
+		vbios->number_of_dram_banks = 8;
+		vbios->high_yclk = bw_int_to_fixed(6000);
+		vbios->mid_yclk = bw_int_to_fixed(3200);
+		vbios->low_yclk = bw_int_to_fixed(1000);
+		vbios->low_sclk = bw_int_to_fixed(300);
+		vbios->mid1_sclk = bw_int_to_fixed(400);
+		vbios->mid2_sclk = bw_int_to_fixed(500);
+		vbios->mid3_sclk = bw_int_to_fixed(600);
+		vbios->mid4_sclk = bw_int_to_fixed(700);
+		vbios->mid5_sclk = bw_int_to_fixed(800);
+		vbios->mid6_sclk = bw_int_to_fixed(974);
+		vbios->high_sclk = bw_int_to_fixed(1154);
+		vbios->low_voltage_max_dispclk = bw_int_to_fixed(459);
+		vbios->mid_voltage_max_dispclk = bw_int_to_fixed(654);
+		vbios->high_voltage_max_dispclk = bw_int_to_fixed(1108);
+		vbios->low_voltage_max_phyclk = bw_int_to_fixed(540);
+		vbios->mid_voltage_max_phyclk = bw_int_to_fixed(810);
+		vbios->high_voltage_max_phyclk = bw_int_to_fixed(810);
+		vbios->data_return_bus_width = bw_int_to_fixed(32);
+		vbios->trc = bw_int_to_fixed(48);
+		vbios->dmifmc_urgent_latency = bw_int_to_fixed(3);
+		vbios->stutter_self_refresh_exit_latency = bw_int_to_fixed(5);
+		vbios->stutter_self_refresh_entry_latency = bw_int_to_fixed(0);
+		vbios->nbp_state_change_latency = bw_int_to_fixed(45);
+		vbios->mcifwrmc_urgent_latency = bw_int_to_fixed(10);
+		vbios->scatter_gather_enable = true;
+		vbios->down_spread_percentage = bw_frc_to_fixed(5, 10);
+		vbios->cursor_width = 32;
+		vbios->average_compression_rate = 4;
+		vbios->number_of_request_slots_gmc_reserves_for_dmif_per_channel = 256;
+		vbios->blackout_duration = bw_int_to_fixed(0); /* us */
+		vbios->maximum_blackout_recovery_time = bw_int_to_fixed(0);
+
+		dceip->max_average_percent_of_ideal_port_bw_display_can_use_in_normal_system_operation = 100;
+		dceip->max_average_percent_of_ideal_drambw_display_can_use_in_normal_system_operation = 100;
+		dceip->percent_of_ideal_port_bw_received_after_urgent_latency = 100;
+		dceip->large_cursor = false;
+		dceip->dmif_request_buffer_size = bw_int_to_fixed(768);
+		dceip->dmif_pipe_en_fbc_chunk_tracker = false;
+		dceip->cursor_max_outstanding_group_num = 1;
+		dceip->lines_interleaved_into_lb = 2;
+		dceip->chunk_width = 256;
+		dceip->number_of_graphics_pipes = 6;
+		dceip->number_of_underlay_pipes = 0;
+		dceip->low_power_tiling_mode = 0;
+		dceip->display_write_back_supported = false;
+		dceip->argb_compression_support = true;
+		dceip->underlay_vscaler_efficiency6_bit_per_component =
+			bw_frc_to_fixed(35556, 10000);
+		dceip->underlay_vscaler_efficiency8_bit_per_component =
+			bw_frc_to_fixed(34286, 10000);
+		dceip->underlay_vscaler_efficiency10_bit_per_component =
+			bw_frc_to_fixed(32, 10);
+		dceip->underlay_vscaler_efficiency12_bit_per_component =
+			bw_int_to_fixed(3);
+		dceip->graphics_vscaler_efficiency6_bit_per_component =
+			bw_frc_to_fixed(35, 10);
+		dceip->graphics_vscaler_efficiency8_bit_per_component =
+			bw_frc_to_fixed(34286, 10000);
+		dceip->graphics_vscaler_efficiency10_bit_per_component =
+			bw_frc_to_fixed(32, 10);
+		dceip->graphics_vscaler_efficiency12_bit_per_component =
+			bw_int_to_fixed(3);
+		dceip->alpha_vscaler_efficiency = bw_int_to_fixed(3);
+		dceip->max_dmif_buffer_allocated = 4;
+		dceip->graphics_dmif_size = 12288;
+		dceip->underlay_luma_dmif_size = 19456;
+		dceip->underlay_chroma_dmif_size = 23552;
+		dceip->pre_downscaler_enabled = true;
+		dceip->underlay_downscale_prefetch_enabled = true;
+		dceip->lb_write_pixels_per_dispclk = bw_int_to_fixed(1);
+		dceip->lb_size_per_component444 = bw_int_to_fixed(245952);
+		dceip->graphics_lb_nodownscaling_multi_line_prefetching = true;
+		dceip->stutter_and_dram_clock_state_change_gated_before_cursor =
+			bw_int_to_fixed(1);
+		dceip->underlay420_luma_lb_size_per_component = bw_int_to_fixed(
+			82176);
+		dceip->underlay420_chroma_lb_size_per_component =
+			bw_int_to_fixed(164352);
+		dceip->underlay422_lb_size_per_component = bw_int_to_fixed(
+			82176);
+		dceip->cursor_chunk_width = bw_int_to_fixed(64);
+		dceip->cursor_dcp_buffer_lines = bw_int_to_fixed(4);
+		dceip->underlay_maximum_width_efficient_for_tiling =
+			bw_int_to_fixed(1920);
+		dceip->underlay_maximum_height_efficient_for_tiling =
+			bw_int_to_fixed(1080);
+		dceip->peak_pte_request_to_eviction_ratio_limiting_multiple_displays_or_single_rotated_display =
+			bw_frc_to_fixed(3, 10);
+		dceip->peak_pte_request_to_eviction_ratio_limiting_single_display_no_rotation =
+			bw_int_to_fixed(25);
+		dceip->minimum_outstanding_pte_request_limit = bw_int_to_fixed(
+			2);
+		dceip->maximum_total_outstanding_pte_requests_allowed_by_saw =
+			bw_int_to_fixed(128);
+		dceip->limit_excessive_outstanding_dmif_requests = true;
+		dceip->linear_mode_line_request_alternation_slice =
+			bw_int_to_fixed(64);
+		dceip->scatter_gather_lines_of_pte_prefetching_in_linear_mode =
+			32;
+		dceip->display_write_back420_luma_mcifwr_buffer_size = 12288;
+		dceip->display_write_back420_chroma_mcifwr_buffer_size = 8192;
+		dceip->request_efficiency = bw_frc_to_fixed(8, 10);
+		dceip->dispclk_per_request = bw_int_to_fixed(2);
+		dceip->dispclk_ramping_factor = bw_frc_to_fixed(105, 100);
+		dceip->display_pipe_throughput_factor = bw_frc_to_fixed(105, 100);
+		dceip->scatter_gather_pte_request_rows_in_tiling_mode = 2;
+		dceip->mcifwr_all_surfaces_burst_time = bw_int_to_fixed(0);
+		break;
+	case BW_CALCS_VERSION_POLARIS11:
+		vbios->memory_type = bw_def_gddr5;
+		vbios->dram_channel_width_in_bits = 32;
+		vbios->number_of_dram_channels = asic_id.vram_width / vbios->dram_channel_width_in_bits;
+		vbios->number_of_dram_banks = 8;
+		vbios->high_yclk = bw_int_to_fixed(6000);
+		vbios->mid_yclk = bw_int_to_fixed(3200);
+		vbios->low_yclk = bw_int_to_fixed(1000);
+		vbios->low_sclk = bw_int_to_fixed(300);
+		vbios->mid1_sclk = bw_int_to_fixed(400);
+		vbios->mid2_sclk = bw_int_to_fixed(500);
+		vbios->mid3_sclk = bw_int_to_fixed(600);
+		vbios->mid4_sclk = bw_int_to_fixed(700);
+		vbios->mid5_sclk = bw_int_to_fixed(800);
+		vbios->mid6_sclk = bw_int_to_fixed(974);
+		vbios->high_sclk = bw_int_to_fixed(1154);
+		vbios->low_voltage_max_dispclk = bw_int_to_fixed(459);
+		vbios->mid_voltage_max_dispclk = bw_int_to_fixed(654);
+		vbios->high_voltage_max_dispclk = bw_int_to_fixed(1108);
+		vbios->low_voltage_max_phyclk = bw_int_to_fixed(540);
+		vbios->mid_voltage_max_phyclk = bw_int_to_fixed(810);
+		vbios->high_voltage_max_phyclk = bw_int_to_fixed(810);
+		vbios->data_return_bus_width = bw_int_to_fixed(32);
+		vbios->trc = bw_int_to_fixed(48);
+		if (vbios->number_of_dram_channels == 2) // 64-bit
+			vbios->dmifmc_urgent_latency = bw_int_to_fixed(4);
+		else
+			vbios->dmifmc_urgent_latency = bw_int_to_fixed(3);
+		vbios->stutter_self_refresh_exit_latency = bw_int_to_fixed(5);
+		vbios->stutter_self_refresh_entry_latency = bw_int_to_fixed(0);
+		vbios->nbp_state_change_latency = bw_int_to_fixed(45);
+		vbios->mcifwrmc_urgent_latency = bw_int_to_fixed(10);
+		vbios->scatter_gather_enable = true;
+		vbios->down_spread_percentage = bw_frc_to_fixed(5, 10);
+		vbios->cursor_width = 32;
+		vbios->average_compression_rate = 4;
+		vbios->number_of_request_slots_gmc_reserves_for_dmif_per_channel = 256;
+		vbios->blackout_duration = bw_int_to_fixed(0); /* us */
+		vbios->maximum_blackout_recovery_time = bw_int_to_fixed(0);
+
+		dceip->max_average_percent_of_ideal_port_bw_display_can_use_in_normal_system_operation = 100;
+		dceip->max_average_percent_of_ideal_drambw_display_can_use_in_normal_system_operation = 100;
+		dceip->percent_of_ideal_port_bw_received_after_urgent_latency = 100;
+		dceip->large_cursor = false;
+		dceip->dmif_request_buffer_size = bw_int_to_fixed(768);
+		dceip->dmif_pipe_en_fbc_chunk_tracker = false;
+		dceip->cursor_max_outstanding_group_num = 1;
+		dceip->lines_interleaved_into_lb = 2;
+		dceip->chunk_width = 256;
+		dceip->number_of_graphics_pipes = 5;
+		dceip->number_of_underlay_pipes = 0;
+		dceip->low_power_tiling_mode = 0;
+		dceip->display_write_back_supported = false;
+		dceip->argb_compression_support = true;
+		dceip->underlay_vscaler_efficiency6_bit_per_component =
+			bw_frc_to_fixed(35556, 10000);
+		dceip->underlay_vscaler_efficiency8_bit_per_component =
+			bw_frc_to_fixed(34286, 10000);
+		dceip->underlay_vscaler_efficiency10_bit_per_component =
+			bw_frc_to_fixed(32, 10);
+		dceip->underlay_vscaler_efficiency12_bit_per_component =
+			bw_int_to_fixed(3);
+		dceip->graphics_vscaler_efficiency6_bit_per_component =
+			bw_frc_to_fixed(35, 10);
+		dceip->graphics_vscaler_efficiency8_bit_per_component =
+			bw_frc_to_fixed(34286, 10000);
+		dceip->graphics_vscaler_efficiency10_bit_per_component =
+			bw_frc_to_fixed(32, 10);
+		dceip->graphics_vscaler_efficiency12_bit_per_component =
+			bw_int_to_fixed(3);
+		dceip->alpha_vscaler_efficiency = bw_int_to_fixed(3);
+		dceip->max_dmif_buffer_allocated = 4;
+		dceip->graphics_dmif_size = 12288;
+		dceip->underlay_luma_dmif_size = 19456;
+		dceip->underlay_chroma_dmif_size = 23552;
+		dceip->pre_downscaler_enabled = true;
+		dceip->underlay_downscale_prefetch_enabled = true;
+		dceip->lb_write_pixels_per_dispclk = bw_int_to_fixed(1);
+		dceip->lb_size_per_component444 = bw_int_to_fixed(245952);
+		dceip->graphics_lb_nodownscaling_multi_line_prefetching = true;
+		dceip->stutter_and_dram_clock_state_change_gated_before_cursor =
+			bw_int_to_fixed(1);
+		dceip->underlay420_luma_lb_size_per_component = bw_int_to_fixed(
+			82176);
+		dceip->underlay420_chroma_lb_size_per_component =
+			bw_int_to_fixed(164352);
+		dceip->underlay422_lb_size_per_component = bw_int_to_fixed(
+			82176);
+		dceip->cursor_chunk_width = bw_int_to_fixed(64);
+		dceip->cursor_dcp_buffer_lines = bw_int_to_fixed(4);
+		dceip->underlay_maximum_width_efficient_for_tiling =
+			bw_int_to_fixed(1920);
+		dceip->underlay_maximum_height_efficient_for_tiling =
+			bw_int_to_fixed(1080);
+		dceip->peak_pte_request_to_eviction_ratio_limiting_multiple_displays_or_single_rotated_display =
+			bw_frc_to_fixed(3, 10);
+		dceip->peak_pte_request_to_eviction_ratio_limiting_single_display_no_rotation =
+			bw_int_to_fixed(25);
+		dceip->minimum_outstanding_pte_request_limit = bw_int_to_fixed(
+			2);
+		dceip->maximum_total_outstanding_pte_requests_allowed_by_saw =
+			bw_int_to_fixed(128);
+		dceip->limit_excessive_outstanding_dmif_requests = true;
+		dceip->linear_mode_line_request_alternation_slice =
+			bw_int_to_fixed(64);
+		dceip->scatter_gather_lines_of_pte_prefetching_in_linear_mode =
+			32;
+		dceip->display_write_back420_luma_mcifwr_buffer_size = 12288;
+		dceip->display_write_back420_chroma_mcifwr_buffer_size = 8192;
+		dceip->request_efficiency = bw_frc_to_fixed(8, 10);
+		dceip->dispclk_per_request = bw_int_to_fixed(2);
+		dceip->dispclk_ramping_factor = bw_frc_to_fixed(105, 100);
+		dceip->display_pipe_throughput_factor = bw_frc_to_fixed(105, 100);
+		dceip->scatter_gather_pte_request_rows_in_tiling_mode = 2;
+		dceip->mcifwr_all_surfaces_burst_time = bw_int_to_fixed(0);
+		break;
+	case BW_CALCS_VERSION_POLARIS12:
+		vbios->memory_type = bw_def_gddr5;
+		vbios->dram_channel_width_in_bits = 32;
+		vbios->number_of_dram_channels = asic_id.vram_width / vbios->dram_channel_width_in_bits;
+		vbios->number_of_dram_banks = 8;
+		vbios->high_yclk = bw_int_to_fixed(6000);
+		vbios->mid_yclk = bw_int_to_fixed(3200);
+		vbios->low_yclk = bw_int_to_fixed(1000);
+		vbios->low_sclk = bw_int_to_fixed(678);
+		vbios->mid1_sclk = bw_int_to_fixed(864);
+		vbios->mid2_sclk = bw_int_to_fixed(900);
+		vbios->mid3_sclk = bw_int_to_fixed(920);
+		vbios->mid4_sclk = bw_int_to_fixed(940);
+		vbios->mid5_sclk = bw_int_to_fixed(960);
+		vbios->mid6_sclk = bw_int_to_fixed(980);
+		vbios->high_sclk = bw_int_to_fixed(1049);
+		vbios->low_voltage_max_dispclk = bw_int_to_fixed(459);
+		vbios->mid_voltage_max_dispclk = bw_int_to_fixed(654);
+		vbios->high_voltage_max_dispclk = bw_int_to_fixed(1108);
+		vbios->low_voltage_max_phyclk = bw_int_to_fixed(540);
+		vbios->mid_voltage_max_phyclk = bw_int_to_fixed(810);
+		vbios->high_voltage_max_phyclk = bw_int_to_fixed(810);
+		vbios->data_return_bus_width = bw_int_to_fixed(32);
+		vbios->trc = bw_int_to_fixed(48);
+		if (vbios->number_of_dram_channels == 2) // 64-bit
+			vbios->dmifmc_urgent_latency = bw_int_to_fixed(4);
+		else
+			vbios->dmifmc_urgent_latency = bw_int_to_fixed(3);
+		vbios->stutter_self_refresh_exit_latency = bw_int_to_fixed(5);
+		vbios->stutter_self_refresh_entry_latency = bw_int_to_fixed(0);
+		vbios->nbp_state_change_latency = bw_int_to_fixed(250);
+		vbios->mcifwrmc_urgent_latency = bw_int_to_fixed(10);
+		vbios->scatter_gather_enable = false;
+		vbios->down_spread_percentage = bw_frc_to_fixed(5, 10);
+		vbios->cursor_width = 32;
+		vbios->average_compression_rate = 4;
+		vbios->number_of_request_slots_gmc_reserves_for_dmif_per_channel = 256;
+		vbios->blackout_duration = bw_int_to_fixed(0); /* us */
+		vbios->maximum_blackout_recovery_time = bw_int_to_fixed(0);
+
+		dceip->max_average_percent_of_ideal_port_bw_display_can_use_in_normal_system_operation = 100;
+		dceip->max_average_percent_of_ideal_drambw_display_can_use_in_normal_system_operation = 100;
+		dceip->percent_of_ideal_port_bw_received_after_urgent_latency = 100;
+		dceip->large_cursor = false;
+		dceip->dmif_request_buffer_size = bw_int_to_fixed(768);
+		dceip->dmif_pipe_en_fbc_chunk_tracker = false;
+		dceip->cursor_max_outstanding_group_num = 1;
+		dceip->lines_interleaved_into_lb = 2;
+		dceip->chunk_width = 256;
+		dceip->number_of_graphics_pipes = 5;
+		dceip->number_of_underlay_pipes = 0;
+		dceip->low_power_tiling_mode = 0;
+		dceip->display_write_back_supported = true;
+		dceip->argb_compression_support = true;
+		dceip->underlay_vscaler_efficiency6_bit_per_component =
+			bw_frc_to_fixed(35556, 10000);
+		dceip->underlay_vscaler_efficiency8_bit_per_component =
+			bw_frc_to_fixed(34286, 10000);
+		dceip->underlay_vscaler_efficiency10_bit_per_component =
+			bw_frc_to_fixed(32, 10);
+		dceip->underlay_vscaler_efficiency12_bit_per_component =
+			bw_int_to_fixed(3);
+		dceip->graphics_vscaler_efficiency6_bit_per_component =
+			bw_frc_to_fixed(35, 10);
+		dceip->graphics_vscaler_efficiency8_bit_per_component =
+			bw_frc_to_fixed(34286, 10000);
+		dceip->graphics_vscaler_efficiency10_bit_per_component =
+			bw_frc_to_fixed(32, 10);
+		dceip->graphics_vscaler_efficiency12_bit_per_component =
+			bw_int_to_fixed(3);
+		dceip->alpha_vscaler_efficiency = bw_int_to_fixed(3);
+		dceip->max_dmif_buffer_allocated = 4;
+		dceip->graphics_dmif_size = 12288;
+		dceip->underlay_luma_dmif_size = 19456;
+		dceip->underlay_chroma_dmif_size = 23552;
+		dceip->pre_downscaler_enabled = true;
+		dceip->underlay_downscale_prefetch_enabled = true;
+		dceip->lb_write_pixels_per_dispclk = bw_int_to_fixed(1);
+		dceip->lb_size_per_component444 = bw_int_to_fixed(245952);
+		dceip->graphics_lb_nodownscaling_multi_line_prefetching = true;
+		dceip->stutter_and_dram_clock_state_change_gated_before_cursor =
+			bw_int_to_fixed(1);
+		dceip->underlay420_luma_lb_size_per_component = bw_int_to_fixed(
+			82176);
+		dceip->underlay420_chroma_lb_size_per_component =
+			bw_int_to_fixed(164352);
+		dceip->underlay422_lb_size_per_component = bw_int_to_fixed(
+			82176);
+		dceip->cursor_chunk_width = bw_int_to_fixed(64);
+		dceip->cursor_dcp_buffer_lines = bw_int_to_fixed(4);
+		dceip->underlay_maximum_width_efficient_for_tiling =
+			bw_int_to_fixed(1920);
+		dceip->underlay_maximum_height_efficient_for_tiling =
+			bw_int_to_fixed(1080);
+		dceip->peak_pte_request_to_eviction_ratio_limiting_multiple_displays_or_single_rotated_display =
+			bw_frc_to_fixed(3, 10);
+		dceip->peak_pte_request_to_eviction_ratio_limiting_single_display_no_rotation =
+			bw_int_to_fixed(25);
+		dceip->minimum_outstanding_pte_request_limit = bw_int_to_fixed(
+			2);
+		dceip->maximum_total_outstanding_pte_requests_allowed_by_saw =
+			bw_int_to_fixed(128);
+		dceip->limit_excessive_outstanding_dmif_requests = true;
+		dceip->linear_mode_line_request_alternation_slice =
+			bw_int_to_fixed(64);
+		dceip->scatter_gather_lines_of_pte_prefetching_in_linear_mode =
+			32;
+		dceip->display_write_back420_luma_mcifwr_buffer_size = 12288;
+		dceip->display_write_back420_chroma_mcifwr_buffer_size = 8192;
+		dceip->request_efficiency = bw_frc_to_fixed(8, 10);
+		dceip->dispclk_per_request = bw_int_to_fixed(2);
+		dceip->dispclk_ramping_factor = bw_frc_to_fixed(105, 100);
+		dceip->display_pipe_throughput_factor = bw_frc_to_fixed(105, 100);
+		dceip->scatter_gather_pte_request_rows_in_tiling_mode = 2;
+		dceip->mcifwr_all_surfaces_burst_time = bw_int_to_fixed(0);
+		break;
+	case BW_CALCS_VERSION_STONEY:
+		vbios->memory_type = bw_def_gddr5;
+		vbios->dram_channel_width_in_bits = 64;
+		vbios->number_of_dram_channels = asic_id.vram_width / vbios->dram_channel_width_in_bits;
+		vbios->number_of_dram_banks = 8;
+		vbios->high_yclk = bw_int_to_fixed(1866);
+		vbios->mid_yclk = bw_int_to_fixed(1866);
+		vbios->low_yclk = bw_int_to_fixed(1333);
+		vbios->low_sclk = bw_int_to_fixed(200);
+		vbios->mid1_sclk = bw_int_to_fixed(600);
+		vbios->mid2_sclk = bw_int_to_fixed(600);
+		vbios->mid3_sclk = bw_int_to_fixed(600);
+		vbios->mid4_sclk = bw_int_to_fixed(600);
+		vbios->mid5_sclk = bw_int_to_fixed(600);
+		vbios->mid6_sclk = bw_int_to_fixed(600);
+		vbios->high_sclk = bw_int_to_fixed(800);
+		vbios->low_voltage_max_dispclk = bw_int_to_fixed(352);
+		vbios->mid_voltage_max_dispclk = bw_int_to_fixed(467);
+		vbios->high_voltage_max_dispclk = bw_int_to_fixed(643);
+		vbios->low_voltage_max_phyclk = bw_int_to_fixed(540);
+		vbios->mid_voltage_max_phyclk = bw_int_to_fixed(810);
+		vbios->high_voltage_max_phyclk = bw_int_to_fixed(810);
+		vbios->data_return_bus_width = bw_int_to_fixed(32);
+		vbios->trc = bw_int_to_fixed(50);
+		vbios->dmifmc_urgent_latency = bw_int_to_fixed(4);
+		vbios->stutter_self_refresh_exit_latency = bw_frc_to_fixed(158, 10);
+		vbios->stutter_self_refresh_entry_latency = bw_int_to_fixed(0);
+		vbios->nbp_state_change_latency = bw_frc_to_fixed(2008, 100);
+		vbios->mcifwrmc_urgent_latency = bw_int_to_fixed(10);
+		vbios->scatter_gather_enable = true;
+		vbios->down_spread_percentage = bw_frc_to_fixed(5, 10);
+		vbios->cursor_width = 32;
+		vbios->average_compression_rate = 4;
+		vbios->number_of_request_slots_gmc_reserves_for_dmif_per_channel = 256;
+		vbios->blackout_duration = bw_int_to_fixed(0); /* us */
+		vbios->maximum_blackout_recovery_time = bw_int_to_fixed(0);
+
+		dceip->max_average_percent_of_ideal_port_bw_display_can_use_in_normal_system_operation = 100;
+		dceip->max_average_percent_of_ideal_drambw_display_can_use_in_normal_system_operation = 100;
+		dceip->percent_of_ideal_port_bw_received_after_urgent_latency = 100;
+		dceip->large_cursor = false;
+		dceip->dmif_request_buffer_size = bw_int_to_fixed(768);
+		dceip->dmif_pipe_en_fbc_chunk_tracker = false;
+		dceip->cursor_max_outstanding_group_num = 1;
+		dceip->lines_interleaved_into_lb = 2;
+		dceip->chunk_width = 256;
+		dceip->number_of_graphics_pipes = 2;
+		dceip->number_of_underlay_pipes = 1;
+		dceip->low_power_tiling_mode = 0;
+		dceip->display_write_back_supported = false;
+		dceip->argb_compression_support = true;
+		dceip->underlay_vscaler_efficiency6_bit_per_component =
+			bw_frc_to_fixed(35556, 10000);
+		dceip->underlay_vscaler_efficiency8_bit_per_component =
+			bw_frc_to_fixed(34286, 10000);
+		dceip->underlay_vscaler_efficiency10_bit_per_component =
+			bw_frc_to_fixed(32, 10);
+		dceip->underlay_vscaler_efficiency12_bit_per_component =
+			bw_int_to_fixed(3);
+		dceip->graphics_vscaler_efficiency6_bit_per_component =
+			bw_frc_to_fixed(35, 10);
+		dceip->graphics_vscaler_efficiency8_bit_per_component =
+			bw_frc_to_fixed(34286, 10000);
+		dceip->graphics_vscaler_efficiency10_bit_per_component =
+			bw_frc_to_fixed(32, 10);
+		dceip->graphics_vscaler_efficiency12_bit_per_component =
+			bw_int_to_fixed(3);
+		dceip->alpha_vscaler_efficiency = bw_int_to_fixed(3);
+		dceip->max_dmif_buffer_allocated = 2;
+		dceip->graphics_dmif_size = 12288;
+		dceip->underlay_luma_dmif_size = 19456;
+		dceip->underlay_chroma_dmif_size = 23552;
+		dceip->pre_downscaler_enabled = true;
+		dceip->underlay_downscale_prefetch_enabled = true;
+		dceip->lb_write_pixels_per_dispclk = bw_int_to_fixed(1);
+		dceip->lb_size_per_component444 = bw_int_to_fixed(82176);
+		dceip->graphics_lb_nodownscaling_multi_line_prefetching = false;
+		dceip->stutter_and_dram_clock_state_change_gated_before_cursor =
+			bw_int_to_fixed(0);
+		dceip->underlay420_luma_lb_size_per_component = bw_int_to_fixed(
+			82176);
+		dceip->underlay420_chroma_lb_size_per_component =
+			bw_int_to_fixed(164352);
+		dceip->underlay422_lb_size_per_component = bw_int_to_fixed(
+			82176);
+		dceip->cursor_chunk_width = bw_int_to_fixed(64);
+		dceip->cursor_dcp_buffer_lines = bw_int_to_fixed(4);
+		dceip->underlay_maximum_width_efficient_for_tiling =
+			bw_int_to_fixed(1920);
+		dceip->underlay_maximum_height_efficient_for_tiling =
+			bw_int_to_fixed(1080);
+		dceip->peak_pte_request_to_eviction_ratio_limiting_multiple_displays_or_single_rotated_display =
+			bw_frc_to_fixed(3, 10);
+		dceip->peak_pte_request_to_eviction_ratio_limiting_single_display_no_rotation =
+			bw_int_to_fixed(25);
+		dceip->minimum_outstanding_pte_request_limit = bw_int_to_fixed(
+			2);
+		dceip->maximum_total_outstanding_pte_requests_allowed_by_saw =
+			bw_int_to_fixed(128);
+		dceip->limit_excessive_outstanding_dmif_requests = true;
+		dceip->linear_mode_line_request_alternation_slice =
+			bw_int_to_fixed(64);
+		dceip->scatter_gather_lines_of_pte_prefetching_in_linear_mode =
+			32;
+		dceip->display_write_back420_luma_mcifwr_buffer_size = 12288;
+		dceip->display_write_back420_chroma_mcifwr_buffer_size = 8192;
+		dceip->request_efficiency = bw_frc_to_fixed(8, 10);
+		dceip->dispclk_per_request = bw_int_to_fixed(2);
+		dceip->dispclk_ramping_factor = bw_frc_to_fixed(105, 100);
+		dceip->display_pipe_throughput_factor = bw_frc_to_fixed(105, 100);
+		dceip->scatter_gather_pte_request_rows_in_tiling_mode = 2;
+		dceip->mcifwr_all_surfaces_burst_time = bw_int_to_fixed(0);
+		break;
+	case BW_CALCS_VERSION_VEGA10:
+		vbios->memory_type = bw_def_hbm;
+		vbios->dram_channel_width_in_bits = 128;
+		vbios->number_of_dram_channels = asic_id.vram_width / vbios->dram_channel_width_in_bits;
+		vbios->number_of_dram_banks = 16;
+		vbios->high_yclk = bw_int_to_fixed(2400);
+		vbios->mid_yclk = bw_int_to_fixed(1700);
+		vbios->low_yclk = bw_int_to_fixed(1000);
+		vbios->low_sclk = bw_int_to_fixed(300);
+		vbios->mid1_sclk = bw_int_to_fixed(350);
+		vbios->mid2_sclk = bw_int_to_fixed(400);
+		vbios->mid3_sclk = bw_int_to_fixed(500);
+		vbios->mid4_sclk = bw_int_to_fixed(600);
+		vbios->mid5_sclk = bw_int_to_fixed(700);
+		vbios->mid6_sclk = bw_int_to_fixed(760);
+		vbios->high_sclk = bw_int_to_fixed(776);
+		vbios->low_voltage_max_dispclk = bw_int_to_fixed(460);
+		vbios->mid_voltage_max_dispclk = bw_int_to_fixed(670);
+		vbios->high_voltage_max_dispclk = bw_int_to_fixed(1133);
+		vbios->low_voltage_max_phyclk = bw_int_to_fixed(540);
+		vbios->mid_voltage_max_phyclk = bw_int_to_fixed(810);
+		vbios->high_voltage_max_phyclk = bw_int_to_fixed(810);
+		vbios->data_return_bus_width = bw_int_to_fixed(32);
+		vbios->trc = bw_int_to_fixed(48);
+		vbios->dmifmc_urgent_latency = bw_int_to_fixed(3);
+		vbios->stutter_self_refresh_exit_latency = bw_frc_to_fixed(75, 10);
+		vbios->stutter_self_refresh_entry_latency = bw_frc_to_fixed(19, 10);
+		vbios->nbp_state_change_latency = bw_int_to_fixed(39);
+		vbios->mcifwrmc_urgent_latency = bw_int_to_fixed(10);
+		vbios->scatter_gather_enable = false;
+		vbios->down_spread_percentage = bw_frc_to_fixed(5, 10);
+		vbios->cursor_width = 32;
+		vbios->average_compression_rate = 4;
+		vbios->number_of_request_slots_gmc_reserves_for_dmif_per_channel = 8;
+		vbios->blackout_duration = bw_int_to_fixed(0); /* us */
+		vbios->maximum_blackout_recovery_time = bw_int_to_fixed(0);
+
+		dceip->max_average_percent_of_ideal_port_bw_display_can_use_in_normal_system_operation = 100;
+		dceip->max_average_percent_of_ideal_drambw_display_can_use_in_normal_system_operation = 100;
+		dceip->percent_of_ideal_port_bw_received_after_urgent_latency = 100;
+		dceip->large_cursor = false;
+		dceip->dmif_request_buffer_size = bw_int_to_fixed(2304);
+		dceip->dmif_pipe_en_fbc_chunk_tracker = true;
+		dceip->cursor_max_outstanding_group_num = 1;
+		dceip->lines_interleaved_into_lb = 2;
+		dceip->chunk_width = 256;
+		dceip->number_of_graphics_pipes = 6;
+		dceip->number_of_underlay_pipes = 0;
+		dceip->low_power_tiling_mode = 0;
+		dceip->display_write_back_supported = true;
+		dceip->argb_compression_support = true;
+		dceip->underlay_vscaler_efficiency6_bit_per_component =
+			bw_frc_to_fixed(35556, 10000);
+		dceip->underlay_vscaler_efficiency8_bit_per_component =
+			bw_frc_to_fixed(34286, 10000);
+		dceip->underlay_vscaler_efficiency10_bit_per_component =
+			bw_frc_to_fixed(32, 10);
+		dceip->underlay_vscaler_efficiency12_bit_per_component =
+			bw_int_to_fixed(3);
+		dceip->graphics_vscaler_efficiency6_bit_per_component =
+			bw_frc_to_fixed(35, 10);
+		dceip->graphics_vscaler_efficiency8_bit_per_component =
+			bw_frc_to_fixed(34286, 10000);
+		dceip->graphics_vscaler_efficiency10_bit_per_component =
+			bw_frc_to_fixed(32, 10);
+		dceip->graphics_vscaler_efficiency12_bit_per_component =
+			bw_int_to_fixed(3);
+		dceip->alpha_vscaler_efficiency = bw_int_to_fixed(3);
+		dceip->max_dmif_buffer_allocated = 4;
+		dceip->graphics_dmif_size = 24576;
+		dceip->underlay_luma_dmif_size = 19456;
+		dceip->underlay_chroma_dmif_size = 23552;
+		dceip->pre_downscaler_enabled = true;
+		dceip->underlay_downscale_prefetch_enabled = false;
+		dceip->lb_write_pixels_per_dispclk = bw_int_to_fixed(1);
+		dceip->lb_size_per_component444 = bw_int_to_fixed(245952);
+		dceip->graphics_lb_nodownscaling_multi_line_prefetching = true;
+		dceip->stutter_and_dram_clock_state_change_gated_before_cursor =
+			bw_int_to_fixed(1);
+		dceip->underlay420_luma_lb_size_per_component = bw_int_to_fixed(
+			82176);
+		dceip->underlay420_chroma_lb_size_per_component =
+			bw_int_to_fixed(164352);
+		dceip->underlay422_lb_size_per_component = bw_int_to_fixed(
+			82176);
+		dceip->cursor_chunk_width = bw_int_to_fixed(64);
+		dceip->cursor_dcp_buffer_lines = bw_int_to_fixed(4);
+		dceip->underlay_maximum_width_efficient_for_tiling =
+			bw_int_to_fixed(1920);
+		dceip->underlay_maximum_height_efficient_for_tiling =
+			bw_int_to_fixed(1080);
+		dceip->peak_pte_request_to_eviction_ratio_limiting_multiple_displays_or_single_rotated_display =
+			bw_frc_to_fixed(3, 10);
+		dceip->peak_pte_request_to_eviction_ratio_limiting_single_display_no_rotation =
+			bw_int_to_fixed(25);
+		dceip->minimum_outstanding_pte_request_limit = bw_int_to_fixed(
+			2);
+		dceip->maximum_total_outstanding_pte_requests_allowed_by_saw =
+			bw_int_to_fixed(128);
+		dceip->limit_excessive_outstanding_dmif_requests = true;
+		dceip->linear_mode_line_request_alternation_slice =
+			bw_int_to_fixed(64);
+		dceip->scatter_gather_lines_of_pte_prefetching_in_linear_mode =
+			32;
+		dceip->display_write_back420_luma_mcifwr_buffer_size = 12288;
+		dceip->display_write_back420_chroma_mcifwr_buffer_size = 8192;
+		dceip->request_efficiency = bw_frc_to_fixed(8, 10);
+		dceip->dispclk_per_request = bw_int_to_fixed(2);
+		dceip->dispclk_ramping_factor = bw_frc_to_fixed(105, 100);
+		dceip->display_pipe_throughput_factor = bw_frc_to_fixed(105, 100);
+		dceip->scatter_gather_pte_request_rows_in_tiling_mode = 2;
+		dceip->mcifwr_all_surfaces_burst_time = bw_int_to_fixed(0);
+		break;
+	default:
+		break;
+	}
+	*bw_dceip = *dceip;
+	*bw_vbios = *vbios;
+
+	kfree(dceip);
+	kfree(vbios);
+}
+
+/*
+ * Compare calculated (required) clocks against the clocks available at
+ * maximum voltage (max Performance Level).
+ */
+static bool is_display_configuration_supported(
+	const struct bw_calcs_vbios *vbios,
+	const struct dce_bw_output *calcs_output)
+{
+	uint32_t int_max_clk;
+
+	int_max_clk = bw_fixed_to_int(vbios->high_voltage_max_dispclk);
+	int_max_clk *= 1000; /* MHz to kHz */
+	if (calcs_output->dispclk_khz > int_max_clk)
+		return false;
+
+	int_max_clk = bw_fixed_to_int(vbios->high_sclk);
+	int_max_clk *= 1000; /* MHz to kHz */
+	if (calcs_output->sclk_khz > int_max_clk)
+		return false;
+
+	return true;
+}
+
+static void populate_initial_data(
+	const struct pipe_ctx pipe[], int pipe_count, struct bw_calcs_data *data)
+{
+	int i, j;
+	int num_displays = 0;
+
+	data->underlay_surface_type = bw_def_420;
+	data->panning_and_bezel_adjustment = bw_def_none;
+	data->graphics_lb_bpc = 10;
+	data->underlay_lb_bpc = 8;
+	data->underlay_tiling_mode = bw_def_tiled;
+	data->graphics_tiling_mode = bw_def_tiled;
+	data->underlay_micro_tile_mode = bw_def_display_micro_tiling;
+	data->graphics_micro_tile_mode = bw_def_display_micro_tiling;
+	data->increase_voltage_to_support_mclk_switch = true;
+
+	/* Pipes with underlay first */
+	for (i = 0; i < pipe_count; i++) {
+		if (!pipe[i].stream || !pipe[i].bottom_pipe)
+			continue;
+
+		ASSERT(pipe[i].plane_state);
+
+		if (num_displays == 0) {
+			if (!pipe[i].plane_state->visible)
+				data->d0_underlay_mode = bw_def_underlay_only;
+			else
+				data->d0_underlay_mode = bw_def_blend;
+		} else {
+			if (!pipe[i].plane_state->visible)
+				data->d1_underlay_mode = bw_def_underlay_only;
+			else
+				data->d1_underlay_mode = bw_def_blend;
+		}
+
+		data->fbc_en[num_displays + 4] = false;
+		data->lpt_en[num_displays + 4] = false;
+		data->h_total[num_displays + 4] = bw_int_to_fixed(pipe[i].stream->timing.h_total);
+		data->v_total[num_displays + 4] = bw_int_to_fixed(pipe[i].stream->timing.v_total);
+		data->pixel_rate[num_displays + 4] = bw_frc_to_fixed(pipe[i].stream->timing.pix_clk_100hz, 10000);
+		data->src_width[num_displays + 4] = bw_int_to_fixed(pipe[i].plane_res.scl_data.viewport.width);
+		data->pitch_in_pixels[num_displays + 4] = data->src_width[num_displays + 4];
+		data->src_height[num_displays + 4] = bw_int_to_fixed(pipe[i].plane_res.scl_data.viewport.height);
+		data->h_taps[num_displays + 4] = bw_int_to_fixed(pipe[i].plane_res.scl_data.taps.h_taps);
+		data->v_taps[num_displays + 4] = bw_int_to_fixed(pipe[i].plane_res.scl_data.taps.v_taps);
+		data->h_scale_ratio[num_displays + 4] = fixed31_32_to_bw_fixed(pipe[i].plane_res.scl_data.ratios.horz.value);
+		data->v_scale_ratio[num_displays + 4] = fixed31_32_to_bw_fixed(pipe[i].plane_res.scl_data.ratios.vert.value);
+		switch (pipe[i].plane_state->rotation) {
+		case ROTATION_ANGLE_0:
+			data->rotation_angle[num_displays + 4] = bw_int_to_fixed(0);
+			break;
+		case ROTATION_ANGLE_90:
+			data->rotation_angle[num_displays + 4] = bw_int_to_fixed(90);
+			break;
+		case ROTATION_ANGLE_180:
+			data->rotation_angle[num_displays + 4] = bw_int_to_fixed(180);
+			break;
+		case ROTATION_ANGLE_270:
+			data->rotation_angle[num_displays + 4] = bw_int_to_fixed(270);
+			break;
+		default:
+			break;
+		}
+		switch (pipe[i].plane_state->format) {
+		case SURFACE_PIXEL_FORMAT_VIDEO_420_YCbCr:
+		case SURFACE_PIXEL_FORMAT_GRPH_ARGB1555:
+		case SURFACE_PIXEL_FORMAT_GRPH_RGB565:
+			data->bytes_per_pixel[num_displays + 4] = 2;
+			break;
+		case SURFACE_PIXEL_FORMAT_GRPH_ARGB8888:
+		case SURFACE_PIXEL_FORMAT_GRPH_ABGR8888:
+		case SURFACE_PIXEL_FORMAT_GRPH_ARGB2101010:
+		case SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010:
+		case SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010_XR_BIAS:
+		case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCbCr:
+		case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCrCb:
+			data->bytes_per_pixel[num_displays + 4] = 4;
+			break;
+		case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616:
+		case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616:
+		case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F:
+			data->bytes_per_pixel[num_displays + 4] = 8;
+			break;
+		default:
+			data->bytes_per_pixel[num_displays + 4] = 4;
+			break;
+		}
+		data->interlace_mode[num_displays + 4] = false;
+		data->stereo_mode[num_displays + 4] = bw_def_mono;
+
+
+		for (j = 0; j < 2; j++) {
+			data->fbc_en[num_displays * 2 + j] = false;
+			data->lpt_en[num_displays * 2 + j] = false;
+
+			data->src_height[num_displays * 2 + j] = bw_int_to_fixed(pipe[i].bottom_pipe->plane_res.scl_data.viewport.height);
+			data->src_width[num_displays * 2 + j] = bw_int_to_fixed(pipe[i].bottom_pipe->plane_res.scl_data.viewport.width);
+			data->pitch_in_pixels[num_displays * 2 + j] = bw_int_to_fixed(
+					pipe[i].bottom_pipe->plane_state->plane_size.surface_pitch);
+			data->h_taps[num_displays * 2 + j] = bw_int_to_fixed(pipe[i].bottom_pipe->plane_res.scl_data.taps.h_taps);
+			data->v_taps[num_displays * 2 + j] = bw_int_to_fixed(pipe[i].bottom_pipe->plane_res.scl_data.taps.v_taps);
+			data->h_scale_ratio[num_displays * 2 + j] = fixed31_32_to_bw_fixed(
+					pipe[i].bottom_pipe->plane_res.scl_data.ratios.horz.value);
+			data->v_scale_ratio[num_displays * 2 + j] = fixed31_32_to_bw_fixed(
+					pipe[i].bottom_pipe->plane_res.scl_data.ratios.vert.value);
+			switch (pipe[i].bottom_pipe->plane_state->rotation) {
+			case ROTATION_ANGLE_0:
+				data->rotation_angle[num_displays * 2 + j] = bw_int_to_fixed(0);
+				break;
+			case ROTATION_ANGLE_90:
+				data->rotation_angle[num_displays * 2 + j] = bw_int_to_fixed(90);
+				break;
+			case ROTATION_ANGLE_180:
+				data->rotation_angle[num_displays * 2 + j] = bw_int_to_fixed(180);
+				break;
+			case ROTATION_ANGLE_270:
+				data->rotation_angle[num_displays * 2 + j] = bw_int_to_fixed(270);
+				break;
+			default:
+				break;
+			}
+			data->stereo_mode[num_displays * 2 + j] = bw_def_mono;
+		}
+
+		num_displays++;
+	}
+
+	/* Pipes without underlay after */
+	for (i = 0; i < pipe_count; i++) {
+		unsigned int pixel_clock_100hz;
+		if (!pipe[i].stream || pipe[i].bottom_pipe)
+			continue;
+
+
+		data->fbc_en[num_displays + 4] = false;
+		data->lpt_en[num_displays + 4] = false;
+		data->h_total[num_displays + 4] = bw_int_to_fixed(pipe[i].stream->timing.h_total);
+		data->v_total[num_displays + 4] = bw_int_to_fixed(pipe[i].stream->timing.v_total);
+		pixel_clock_100hz = pipe[i].stream->timing.pix_clk_100hz;
+		if (pipe[i].stream->timing.timing_3d_format == TIMING_3D_FORMAT_HW_FRAME_PACKING)
+			pixel_clock_100hz *= 2;
+		data->pixel_rate[num_displays + 4] = bw_frc_to_fixed(pixel_clock_100hz, 10000);
+		if (pipe[i].plane_state) {
+			data->src_width[num_displays + 4] = bw_int_to_fixed(pipe[i].plane_res.scl_data.viewport.width);
+			data->pitch_in_pixels[num_displays + 4] = data->src_width[num_displays + 4];
+			data->src_height[num_displays + 4] = bw_int_to_fixed(pipe[i].plane_res.scl_data.viewport.height);
+			data->h_taps[num_displays + 4] = bw_int_to_fixed(pipe[i].plane_res.scl_data.taps.h_taps);
+			data->v_taps[num_displays + 4] = bw_int_to_fixed(pipe[i].plane_res.scl_data.taps.v_taps);
+			data->h_scale_ratio[num_displays + 4] = fixed31_32_to_bw_fixed(pipe[i].plane_res.scl_data.ratios.horz.value);
+			data->v_scale_ratio[num_displays + 4] = fixed31_32_to_bw_fixed(pipe[i].plane_res.scl_data.ratios.vert.value);
+			switch (pipe[i].plane_state->rotation) {
+			case ROTATION_ANGLE_0:
+				data->rotation_angle[num_displays + 4] = bw_int_to_fixed(0);
+				break;
+			case ROTATION_ANGLE_90:
+				data->rotation_angle[num_displays + 4] = bw_int_to_fixed(90);
+				break;
+			case ROTATION_ANGLE_180:
+				data->rotation_angle[num_displays + 4] = bw_int_to_fixed(180);
+				break;
+			case ROTATION_ANGLE_270:
+				data->rotation_angle[num_displays + 4] = bw_int_to_fixed(270);
+				break;
+			default:
+				break;
+			}
+			switch (pipe[i].plane_state->format) {
+			case SURFACE_PIXEL_FORMAT_VIDEO_420_YCbCr:
+			case SURFACE_PIXEL_FORMAT_VIDEO_420_YCrCb:
+			case SURFACE_PIXEL_FORMAT_GRPH_ARGB1555:
+			case SURFACE_PIXEL_FORMAT_GRPH_RGB565:
+				data->bytes_per_pixel[num_displays + 4] = 2;
+				break;
+			case SURFACE_PIXEL_FORMAT_GRPH_ARGB8888:
+			case SURFACE_PIXEL_FORMAT_GRPH_ABGR8888:
+			case SURFACE_PIXEL_FORMAT_GRPH_ARGB2101010:
+			case SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010:
+			case SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010_XR_BIAS:
+			case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCbCr:
+			case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCrCb:
+				data->bytes_per_pixel[num_displays + 4] = 4;
+				break;
+			case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616:
+			case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616:
+			case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F:
+				data->bytes_per_pixel[num_displays + 4] = 8;
+				break;
+			default:
+				data->bytes_per_pixel[num_displays + 4] = 4;
+				break;
+			}
+		} else if (pipe[i].stream->dst.width != 0 &&
+					pipe[i].stream->dst.height != 0 &&
+					pipe[i].stream->src.width != 0 &&
+					pipe[i].stream->src.height != 0) {
+			data->src_width[num_displays + 4] = bw_int_to_fixed(pipe[i].stream->src.width);
+			data->pitch_in_pixels[num_displays + 4] = data->src_width[num_displays + 4];
+			data->src_height[num_displays + 4] = bw_int_to_fixed(pipe[i].stream->src.height);
+			data->h_taps[num_displays + 4] = pipe[i].stream->src.width == pipe[i].stream->dst.width ? bw_int_to_fixed(1) : bw_int_to_fixed(2);
+			data->v_taps[num_displays + 4] = pipe[i].stream->src.height == pipe[i].stream->dst.height ? bw_int_to_fixed(1) : bw_int_to_fixed(2);
+			data->h_scale_ratio[num_displays + 4] = bw_frc_to_fixed(pipe[i].stream->src.width, pipe[i].stream->dst.width);
+			data->v_scale_ratio[num_displays + 4] = bw_frc_to_fixed(pipe[i].stream->src.height, pipe[i].stream->dst.height);
+			data->rotation_angle[num_displays + 4] = bw_int_to_fixed(0);
+			data->bytes_per_pixel[num_displays + 4] = 4;
+		} else {
+			data->src_width[num_displays + 4] = bw_int_to_fixed(pipe[i].stream->timing.h_addressable);
+			data->pitch_in_pixels[num_displays + 4] = data->src_width[num_displays + 4];
+			data->src_height[num_displays + 4] = bw_int_to_fixed(pipe[i].stream->timing.v_addressable);
+			data->h_taps[num_displays + 4] = bw_int_to_fixed(1);
+			data->v_taps[num_displays + 4] = bw_int_to_fixed(1);
+			data->h_scale_ratio[num_displays + 4] = bw_int_to_fixed(1);
+			data->v_scale_ratio[num_displays + 4] = bw_int_to_fixed(1);
+			data->rotation_angle[num_displays + 4] = bw_int_to_fixed(0);
+			data->bytes_per_pixel[num_displays + 4] = 4;
+		}
+
+		data->interlace_mode[num_displays + 4] = false;
+		data->stereo_mode[num_displays + 4] = bw_def_mono;
+		num_displays++;
+	}
+
+	data->number_of_displays = num_displays;
+}
+
+static bool all_displays_in_sync(const struct pipe_ctx pipe[],
+				 int pipe_count)
+{
+	const struct pipe_ctx *active_pipes[MAX_PIPES];
+	int i, num_active_pipes = 0;
+
+	for (i = 0; i < pipe_count; i++) {
+		if (!resource_is_pipe_type(&pipe[i], OPP_HEAD))
+			continue;
+
+		active_pipes[num_active_pipes++] = &pipe[i];
+	}
+
+	if (!num_active_pipes)
+		return false;
+
+	for (i = 1; i < num_active_pipes; ++i) {
+		if (!resource_are_streams_timing_synchronizable(
+			    active_pipes[0]->stream, active_pipes[i]->stream)) {
+			return false;
+		}
+	}
+
+	return true;
+}
+
+/*
+ * Return:
+ *	true -	Display(s) configuration supported.
+ *		In this case 'calcs_output' contains data for HW programming
+ *	false - Display(s) configuration not supported (not enough bandwidth).
+ */
+bool bw_calcs(struct dc_context *ctx,
+	const struct bw_calcs_dceip *dceip,
+	const struct bw_calcs_vbios *vbios,
+	const struct pipe_ctx pipe[],
+	int pipe_count,
+	struct dce_bw_output *calcs_output)
+{
+	struct bw_calcs_data *data = kzalloc(sizeof(struct bw_calcs_data),
+					     GFP_KERNEL);
+	if (!data)
+		return false;
+
+	populate_initial_data(pipe, pipe_count, data);
+
+	if (ctx->dc->config.multi_mon_pp_mclk_switch)
+		calcs_output->all_displays_in_sync = all_displays_in_sync(pipe, pipe_count);
+	else
+		calcs_output->all_displays_in_sync = false;
+
+	if (data->number_of_displays != 0) {
+		uint8_t yclk_lvl;
+		struct bw_fixed high_sclk = vbios->high_sclk;
+		struct bw_fixed mid1_sclk = vbios->mid1_sclk;
+		struct bw_fixed mid2_sclk = vbios->mid2_sclk;
+		struct bw_fixed mid3_sclk = vbios->mid3_sclk;
+		struct bw_fixed mid4_sclk = vbios->mid4_sclk;
+		struct bw_fixed mid5_sclk = vbios->mid5_sclk;
+		struct bw_fixed mid6_sclk = vbios->mid6_sclk;
+		struct bw_fixed low_sclk = vbios->low_sclk;
+		struct bw_fixed high_yclk = vbios->high_yclk;
+		struct bw_fixed mid_yclk = vbios->mid_yclk;
+		struct bw_fixed low_yclk = vbios->low_yclk;
+
+		if (ctx->dc->debug.bandwidth_calcs_trace) {
+			print_bw_calcs_dceip(ctx, dceip);
+			print_bw_calcs_vbios(ctx, vbios);
+			print_bw_calcs_data(ctx, data);
+		}
+		calculate_bandwidth(dceip, vbios, data);
+
+		yclk_lvl = data->y_clk_level;
+
+		calcs_output->nbp_state_change_enable =
+			data->nbp_state_change_enable;
+		calcs_output->cpuc_state_change_enable =
+				data->cpuc_state_change_enable;
+		calcs_output->cpup_state_change_enable =
+				data->cpup_state_change_enable;
+		calcs_output->stutter_mode_enable =
+				data->stutter_mode_enable;
+		calcs_output->dispclk_khz =
+			bw_fixed_to_int(bw_mul(data->dispclk,
+					bw_int_to_fixed(1000)));
+		calcs_output->blackout_recovery_time_us =
+			bw_fixed_to_int(data->blackout_recovery_time);
+		calcs_output->sclk_khz =
+			bw_fixed_to_int(bw_mul(data->required_sclk,
+					bw_int_to_fixed(1000)));
+		calcs_output->sclk_deep_sleep_khz =
+			bw_fixed_to_int(bw_mul(data->sclk_deep_sleep,
+					bw_int_to_fixed(1000)));
+		if (yclk_lvl == 0)
+			calcs_output->yclk_khz = bw_fixed_to_int(
+				bw_mul(low_yclk, bw_int_to_fixed(1000)));
+		else if (yclk_lvl == 1)
+			calcs_output->yclk_khz = bw_fixed_to_int(
+				bw_mul(mid_yclk, bw_int_to_fixed(1000)));
+		else
+			calcs_output->yclk_khz = bw_fixed_to_int(
+				bw_mul(high_yclk, bw_int_to_fixed(1000)));
+
+		/* units: nanosecond, 16bit storage. */
+
+		calcs_output->nbp_state_change_wm_ns[0].a_mark =
+			bw_fixed_to_int(bw_mul(data->
+				nbp_state_change_watermark[4], bw_int_to_fixed(1000)));
+		calcs_output->nbp_state_change_wm_ns[1].a_mark =
+			bw_fixed_to_int(bw_mul(data->
+				nbp_state_change_watermark[5], bw_int_to_fixed(1000)));
+		calcs_output->nbp_state_change_wm_ns[2].a_mark =
+			bw_fixed_to_int(bw_mul(data->
+				nbp_state_change_watermark[6], bw_int_to_fixed(1000)));
+
+		if (ctx->dc->caps.max_slave_planes) {
+			calcs_output->nbp_state_change_wm_ns[3].a_mark =
+				bw_fixed_to_int(bw_mul(data->
+					nbp_state_change_watermark[0], bw_int_to_fixed(1000)));
+			calcs_output->nbp_state_change_wm_ns[4].a_mark =
+				bw_fixed_to_int(bw_mul(data->
+							nbp_state_change_watermark[1], bw_int_to_fixed(1000)));
+		} else {
+			calcs_output->nbp_state_change_wm_ns[3].a_mark =
+				bw_fixed_to_int(bw_mul(data->
+					nbp_state_change_watermark[7], bw_int_to_fixed(1000)));
+			calcs_output->nbp_state_change_wm_ns[4].a_mark =
+				bw_fixed_to_int(bw_mul(data->
+					nbp_state_change_watermark[8], bw_int_to_fixed(1000)));
+		}
+		calcs_output->nbp_state_change_wm_ns[5].a_mark =
+			bw_fixed_to_int(bw_mul(data->
+				nbp_state_change_watermark[9], bw_int_to_fixed(1000)));
+
+
+
+		calcs_output->stutter_exit_wm_ns[0].a_mark =
+			bw_fixed_to_int(bw_mul(data->
+				stutter_exit_watermark[4], bw_int_to_fixed(1000)));
+		calcs_output->stutter_exit_wm_ns[1].a_mark =
+			bw_fixed_to_int(bw_mul(data->
+				stutter_exit_watermark[5], bw_int_to_fixed(1000)));
+		calcs_output->stutter_exit_wm_ns[2].a_mark =
+			bw_fixed_to_int(bw_mul(data->
+				stutter_exit_watermark[6], bw_int_to_fixed(1000)));
+		if (ctx->dc->caps.max_slave_planes) {
+			calcs_output->stutter_exit_wm_ns[3].a_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_exit_watermark[0], bw_int_to_fixed(1000)));
+			calcs_output->stutter_exit_wm_ns[4].a_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_exit_watermark[1], bw_int_to_fixed(1000)));
+		} else {
+			calcs_output->stutter_exit_wm_ns[3].a_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_exit_watermark[7], bw_int_to_fixed(1000)));
+			calcs_output->stutter_exit_wm_ns[4].a_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_exit_watermark[8], bw_int_to_fixed(1000)));
+		}
+		calcs_output->stutter_exit_wm_ns[5].a_mark =
+			bw_fixed_to_int(bw_mul(data->
+				stutter_exit_watermark[9], bw_int_to_fixed(1000)));
+
+		calcs_output->stutter_entry_wm_ns[0].a_mark =
+			bw_fixed_to_int(bw_mul(data->
+				stutter_entry_watermark[4], bw_int_to_fixed(1000)));
+		calcs_output->stutter_entry_wm_ns[1].a_mark =
+			bw_fixed_to_int(bw_mul(data->
+				stutter_entry_watermark[5], bw_int_to_fixed(1000)));
+		calcs_output->stutter_entry_wm_ns[2].a_mark =
+			bw_fixed_to_int(bw_mul(data->
+				stutter_entry_watermark[6], bw_int_to_fixed(1000)));
+		if (ctx->dc->caps.max_slave_planes) {
+			calcs_output->stutter_entry_wm_ns[3].a_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_entry_watermark[0], bw_int_to_fixed(1000)));
+			calcs_output->stutter_entry_wm_ns[4].a_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_entry_watermark[1], bw_int_to_fixed(1000)));
+		} else {
+			calcs_output->stutter_entry_wm_ns[3].a_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_entry_watermark[7], bw_int_to_fixed(1000)));
+			calcs_output->stutter_entry_wm_ns[4].a_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_entry_watermark[8], bw_int_to_fixed(1000)));
+		}
+		calcs_output->stutter_entry_wm_ns[5].a_mark =
+			bw_fixed_to_int(bw_mul(data->
+				stutter_entry_watermark[9], bw_int_to_fixed(1000)));
+
+		calcs_output->urgent_wm_ns[0].a_mark =
+			bw_fixed_to_int(bw_mul(data->
+				urgent_watermark[4], bw_int_to_fixed(1000)));
+		calcs_output->urgent_wm_ns[1].a_mark =
+			bw_fixed_to_int(bw_mul(data->
+				urgent_watermark[5], bw_int_to_fixed(1000)));
+		calcs_output->urgent_wm_ns[2].a_mark =
+			bw_fixed_to_int(bw_mul(data->
+				urgent_watermark[6], bw_int_to_fixed(1000)));
+		if (ctx->dc->caps.max_slave_planes) {
+			calcs_output->urgent_wm_ns[3].a_mark =
+				bw_fixed_to_int(bw_mul(data->
+					urgent_watermark[0], bw_int_to_fixed(1000)));
+			calcs_output->urgent_wm_ns[4].a_mark =
+				bw_fixed_to_int(bw_mul(data->
+					urgent_watermark[1], bw_int_to_fixed(1000)));
+		} else {
+			calcs_output->urgent_wm_ns[3].a_mark =
+				bw_fixed_to_int(bw_mul(data->
+					urgent_watermark[7], bw_int_to_fixed(1000)));
+			calcs_output->urgent_wm_ns[4].a_mark =
+				bw_fixed_to_int(bw_mul(data->
+					urgent_watermark[8], bw_int_to_fixed(1000)));
+		}
+		calcs_output->urgent_wm_ns[5].a_mark =
+			bw_fixed_to_int(bw_mul(data->
+				urgent_watermark[9], bw_int_to_fixed(1000)));
+
+		if (dceip->version != BW_CALCS_VERSION_CARRIZO) {
+			((struct bw_calcs_vbios *)vbios)->low_sclk = mid3_sclk;
+			((struct bw_calcs_vbios *)vbios)->mid1_sclk = mid3_sclk;
+			((struct bw_calcs_vbios *)vbios)->mid2_sclk = mid3_sclk;
+			calculate_bandwidth(dceip, vbios, data);
+
+			calcs_output->nbp_state_change_wm_ns[0].b_mark =
+				bw_fixed_to_int(bw_mul(data->
+					nbp_state_change_watermark[4],bw_int_to_fixed(1000)));
+			calcs_output->nbp_state_change_wm_ns[1].b_mark =
+				bw_fixed_to_int(bw_mul(data->
+					nbp_state_change_watermark[5], bw_int_to_fixed(1000)));
+			calcs_output->nbp_state_change_wm_ns[2].b_mark =
+				bw_fixed_to_int(bw_mul(data->
+					nbp_state_change_watermark[6], bw_int_to_fixed(1000)));
+
+			if (ctx->dc->caps.max_slave_planes) {
+				calcs_output->nbp_state_change_wm_ns[3].b_mark =
+					bw_fixed_to_int(bw_mul(data->
+						nbp_state_change_watermark[0], bw_int_to_fixed(1000)));
+				calcs_output->nbp_state_change_wm_ns[4].b_mark =
+					bw_fixed_to_int(bw_mul(data->
+						nbp_state_change_watermark[1], bw_int_to_fixed(1000)));
+			} else {
+				calcs_output->nbp_state_change_wm_ns[3].b_mark =
+					bw_fixed_to_int(bw_mul(data->
+						nbp_state_change_watermark[7], bw_int_to_fixed(1000)));
+				calcs_output->nbp_state_change_wm_ns[4].b_mark =
+					bw_fixed_to_int(bw_mul(data->
+						nbp_state_change_watermark[8], bw_int_to_fixed(1000)));
+			}
+			calcs_output->nbp_state_change_wm_ns[5].b_mark =
+				bw_fixed_to_int(bw_mul(data->
+					nbp_state_change_watermark[9], bw_int_to_fixed(1000)));
+
+
+
+			calcs_output->stutter_exit_wm_ns[0].b_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_exit_watermark[4], bw_int_to_fixed(1000)));
+			calcs_output->stutter_exit_wm_ns[1].b_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_exit_watermark[5], bw_int_to_fixed(1000)));
+			calcs_output->stutter_exit_wm_ns[2].b_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_exit_watermark[6], bw_int_to_fixed(1000)));
+			if (ctx->dc->caps.max_slave_planes) {
+				calcs_output->stutter_exit_wm_ns[3].b_mark =
+					bw_fixed_to_int(bw_mul(data->
+						stutter_exit_watermark[0], bw_int_to_fixed(1000)));
+				calcs_output->stutter_exit_wm_ns[4].b_mark =
+					bw_fixed_to_int(bw_mul(data->
+						stutter_exit_watermark[1], bw_int_to_fixed(1000)));
+			} else {
+				calcs_output->stutter_exit_wm_ns[3].b_mark =
+					bw_fixed_to_int(bw_mul(data->
+						stutter_exit_watermark[7], bw_int_to_fixed(1000)));
+				calcs_output->stutter_exit_wm_ns[4].b_mark =
+					bw_fixed_to_int(bw_mul(data->
+						stutter_exit_watermark[8], bw_int_to_fixed(1000)));
+			}
+			calcs_output->stutter_exit_wm_ns[5].b_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_exit_watermark[9], bw_int_to_fixed(1000)));
+
+			calcs_output->stutter_entry_wm_ns[0].b_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_entry_watermark[4], bw_int_to_fixed(1000)));
+			calcs_output->stutter_entry_wm_ns[1].b_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_entry_watermark[5], bw_int_to_fixed(1000)));
+			calcs_output->stutter_entry_wm_ns[2].b_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_entry_watermark[6], bw_int_to_fixed(1000)));
+			if (ctx->dc->caps.max_slave_planes) {
+				calcs_output->stutter_entry_wm_ns[3].b_mark =
+					bw_fixed_to_int(bw_mul(data->
+						stutter_entry_watermark[0], bw_int_to_fixed(1000)));
+				calcs_output->stutter_entry_wm_ns[4].b_mark =
+					bw_fixed_to_int(bw_mul(data->
+						stutter_entry_watermark[1], bw_int_to_fixed(1000)));
+			} else {
+				calcs_output->stutter_entry_wm_ns[3].b_mark =
+					bw_fixed_to_int(bw_mul(data->
+						stutter_entry_watermark[7], bw_int_to_fixed(1000)));
+				calcs_output->stutter_entry_wm_ns[4].b_mark =
+					bw_fixed_to_int(bw_mul(data->
+						stutter_entry_watermark[8], bw_int_to_fixed(1000)));
+			}
+			calcs_output->stutter_entry_wm_ns[5].b_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_entry_watermark[9], bw_int_to_fixed(1000)));
+
+			calcs_output->urgent_wm_ns[0].b_mark =
+				bw_fixed_to_int(bw_mul(data->
+					urgent_watermark[4], bw_int_to_fixed(1000)));
+			calcs_output->urgent_wm_ns[1].b_mark =
+				bw_fixed_to_int(bw_mul(data->
+					urgent_watermark[5], bw_int_to_fixed(1000)));
+			calcs_output->urgent_wm_ns[2].b_mark =
+				bw_fixed_to_int(bw_mul(data->
+					urgent_watermark[6], bw_int_to_fixed(1000)));
+			if (ctx->dc->caps.max_slave_planes) {
+				calcs_output->urgent_wm_ns[3].b_mark =
+					bw_fixed_to_int(bw_mul(data->
+						urgent_watermark[0], bw_int_to_fixed(1000)));
+				calcs_output->urgent_wm_ns[4].b_mark =
+					bw_fixed_to_int(bw_mul(data->
+						urgent_watermark[1], bw_int_to_fixed(1000)));
+			} else {
+				calcs_output->urgent_wm_ns[3].b_mark =
+					bw_fixed_to_int(bw_mul(data->
+						urgent_watermark[7], bw_int_to_fixed(1000)));
+				calcs_output->urgent_wm_ns[4].b_mark =
+					bw_fixed_to_int(bw_mul(data->
+						urgent_watermark[8], bw_int_to_fixed(1000)));
+			}
+			calcs_output->urgent_wm_ns[5].b_mark =
+				bw_fixed_to_int(bw_mul(data->
+					urgent_watermark[9], bw_int_to_fixed(1000)));
+
+			((struct bw_calcs_vbios *)vbios)->low_sclk = low_sclk;
+			((struct bw_calcs_vbios *)vbios)->mid1_sclk = mid1_sclk;
+			((struct bw_calcs_vbios *)vbios)->mid2_sclk = mid2_sclk;
+			((struct bw_calcs_vbios *)vbios)->low_yclk = mid_yclk;
+			calculate_bandwidth(dceip, vbios, data);
+
+			calcs_output->nbp_state_change_wm_ns[0].c_mark =
+				bw_fixed_to_int(bw_mul(data->
+					nbp_state_change_watermark[4], bw_int_to_fixed(1000)));
+			calcs_output->nbp_state_change_wm_ns[1].c_mark =
+				bw_fixed_to_int(bw_mul(data->
+					nbp_state_change_watermark[5], bw_int_to_fixed(1000)));
+			calcs_output->nbp_state_change_wm_ns[2].c_mark =
+				bw_fixed_to_int(bw_mul(data->
+					nbp_state_change_watermark[6], bw_int_to_fixed(1000)));
+			if (ctx->dc->caps.max_slave_planes) {
+				calcs_output->nbp_state_change_wm_ns[3].c_mark =
+					bw_fixed_to_int(bw_mul(data->
+						nbp_state_change_watermark[0], bw_int_to_fixed(1000)));
+				calcs_output->nbp_state_change_wm_ns[4].c_mark =
+					bw_fixed_to_int(bw_mul(data->
+						nbp_state_change_watermark[1], bw_int_to_fixed(1000)));
+			} else {
+				calcs_output->nbp_state_change_wm_ns[3].c_mark =
+					bw_fixed_to_int(bw_mul(data->
+						nbp_state_change_watermark[7], bw_int_to_fixed(1000)));
+				calcs_output->nbp_state_change_wm_ns[4].c_mark =
+					bw_fixed_to_int(bw_mul(data->
+						nbp_state_change_watermark[8], bw_int_to_fixed(1000)));
+			}
+			calcs_output->nbp_state_change_wm_ns[5].c_mark =
+				bw_fixed_to_int(bw_mul(data->
+					nbp_state_change_watermark[9], bw_int_to_fixed(1000)));
+
+
+			calcs_output->stutter_exit_wm_ns[0].c_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_exit_watermark[4], bw_int_to_fixed(1000)));
+			calcs_output->stutter_exit_wm_ns[1].c_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_exit_watermark[5], bw_int_to_fixed(1000)));
+			calcs_output->stutter_exit_wm_ns[2].c_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_exit_watermark[6], bw_int_to_fixed(1000)));
+			if (ctx->dc->caps.max_slave_planes) {
+				calcs_output->stutter_exit_wm_ns[3].c_mark =
+					bw_fixed_to_int(bw_mul(data->
+						stutter_exit_watermark[0], bw_int_to_fixed(1000)));
+				calcs_output->stutter_exit_wm_ns[4].c_mark =
+					bw_fixed_to_int(bw_mul(data->
+						stutter_exit_watermark[1], bw_int_to_fixed(1000)));
+			} else {
+				calcs_output->stutter_exit_wm_ns[3].c_mark =
+					bw_fixed_to_int(bw_mul(data->
+						stutter_exit_watermark[7], bw_int_to_fixed(1000)));
+				calcs_output->stutter_exit_wm_ns[4].c_mark =
+					bw_fixed_to_int(bw_mul(data->
+						stutter_exit_watermark[8], bw_int_to_fixed(1000)));
+			}
+			calcs_output->stutter_exit_wm_ns[5].c_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_exit_watermark[9], bw_int_to_fixed(1000)));
+			calcs_output->stutter_entry_wm_ns[0].c_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_entry_watermark[4], bw_int_to_fixed(1000)));
+			calcs_output->stutter_entry_wm_ns[1].c_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_entry_watermark[5], bw_int_to_fixed(1000)));
+			calcs_output->stutter_entry_wm_ns[2].c_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_entry_watermark[6], bw_int_to_fixed(1000)));
+			if (ctx->dc->caps.max_slave_planes) {
+				calcs_output->stutter_entry_wm_ns[3].c_mark =
+					bw_fixed_to_int(bw_mul(data->stutter_entry_watermark[0],
+						bw_int_to_fixed(1000)));
+				calcs_output->stutter_entry_wm_ns[4].c_mark =
+					bw_fixed_to_int(bw_mul(data->stutter_entry_watermark[1],
+						bw_int_to_fixed(1000)));
+			} else {
+				calcs_output->stutter_entry_wm_ns[3].c_mark =
+					bw_fixed_to_int(bw_mul(data->stutter_entry_watermark[7],
+						bw_int_to_fixed(1000)));
+				calcs_output->stutter_entry_wm_ns[4].c_mark =
+					bw_fixed_to_int(bw_mul(data->stutter_entry_watermark[8],
+						bw_int_to_fixed(1000)));
+			}
+			calcs_output->stutter_entry_wm_ns[5].c_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_entry_watermark[9], bw_int_to_fixed(1000)));
+			calcs_output->urgent_wm_ns[0].c_mark =
+				bw_fixed_to_int(bw_mul(data->
+					urgent_watermark[4], bw_int_to_fixed(1000)));
+			calcs_output->urgent_wm_ns[1].c_mark =
+				bw_fixed_to_int(bw_mul(data->
+					urgent_watermark[5], bw_int_to_fixed(1000)));
+			calcs_output->urgent_wm_ns[2].c_mark =
+				bw_fixed_to_int(bw_mul(data->
+					urgent_watermark[6], bw_int_to_fixed(1000)));
+			if (ctx->dc->caps.max_slave_planes) {
+				calcs_output->urgent_wm_ns[3].c_mark =
+					bw_fixed_to_int(bw_mul(data->
+						urgent_watermark[0], bw_int_to_fixed(1000)));
+				calcs_output->urgent_wm_ns[4].c_mark =
+					bw_fixed_to_int(bw_mul(data->
+						urgent_watermark[1], bw_int_to_fixed(1000)));
+			} else {
+				calcs_output->urgent_wm_ns[3].c_mark =
+					bw_fixed_to_int(bw_mul(data->
+						urgent_watermark[7], bw_int_to_fixed(1000)));
+				calcs_output->urgent_wm_ns[4].c_mark =
+					bw_fixed_to_int(bw_mul(data->
+						urgent_watermark[8], bw_int_to_fixed(1000)));
+			}
+			calcs_output->urgent_wm_ns[5].c_mark =
+				bw_fixed_to_int(bw_mul(data->
+					urgent_watermark[9], bw_int_to_fixed(1000)));
+		}
+
+		if (dceip->version == BW_CALCS_VERSION_CARRIZO) {
+			((struct bw_calcs_vbios *)vbios)->low_yclk = high_yclk;
+			((struct bw_calcs_vbios *)vbios)->mid_yclk = high_yclk;
+			((struct bw_calcs_vbios *)vbios)->low_sclk = high_sclk;
+			((struct bw_calcs_vbios *)vbios)->mid1_sclk = high_sclk;
+			((struct bw_calcs_vbios *)vbios)->mid2_sclk = high_sclk;
+			((struct bw_calcs_vbios *)vbios)->mid3_sclk = high_sclk;
+			((struct bw_calcs_vbios *)vbios)->mid4_sclk = high_sclk;
+			((struct bw_calcs_vbios *)vbios)->mid5_sclk = high_sclk;
+			((struct bw_calcs_vbios *)vbios)->mid6_sclk = high_sclk;
+		} else {
+			((struct bw_calcs_vbios *)vbios)->low_yclk = mid_yclk;
+			((struct bw_calcs_vbios *)vbios)->low_sclk = mid3_sclk;
+			((struct bw_calcs_vbios *)vbios)->mid1_sclk = mid3_sclk;
+			((struct bw_calcs_vbios *)vbios)->mid2_sclk = mid3_sclk;
+		}
+
+		calculate_bandwidth(dceip, vbios, data);
+
+		calcs_output->nbp_state_change_wm_ns[0].d_mark =
+			bw_fixed_to_int(bw_mul(data->
+				nbp_state_change_watermark[4], bw_int_to_fixed(1000)));
+		calcs_output->nbp_state_change_wm_ns[1].d_mark =
+			bw_fixed_to_int(bw_mul(data->
+				nbp_state_change_watermark[5], bw_int_to_fixed(1000)));
+		calcs_output->nbp_state_change_wm_ns[2].d_mark =
+			bw_fixed_to_int(bw_mul(data->
+				nbp_state_change_watermark[6], bw_int_to_fixed(1000)));
+		if (ctx->dc->caps.max_slave_planes) {
+			calcs_output->nbp_state_change_wm_ns[3].d_mark =
+				bw_fixed_to_int(bw_mul(data->
+					nbp_state_change_watermark[0], bw_int_to_fixed(1000)));
+			calcs_output->nbp_state_change_wm_ns[4].d_mark =
+				bw_fixed_to_int(bw_mul(data->
+					nbp_state_change_watermark[1], bw_int_to_fixed(1000)));
+		} else {
+			calcs_output->nbp_state_change_wm_ns[3].d_mark =
+				bw_fixed_to_int(bw_mul(data->
+					nbp_state_change_watermark[7], bw_int_to_fixed(1000)));
+			calcs_output->nbp_state_change_wm_ns[4].d_mark =
+				bw_fixed_to_int(bw_mul(data->
+					nbp_state_change_watermark[8], bw_int_to_fixed(1000)));
+		}
+		calcs_output->nbp_state_change_wm_ns[5].d_mark =
+			bw_fixed_to_int(bw_mul(data->
+				nbp_state_change_watermark[9], bw_int_to_fixed(1000)));
+
+		calcs_output->stutter_exit_wm_ns[0].d_mark =
+			bw_fixed_to_int(bw_mul(data->
+				stutter_exit_watermark[4], bw_int_to_fixed(1000)));
+		calcs_output->stutter_exit_wm_ns[1].d_mark =
+			bw_fixed_to_int(bw_mul(data->
+				stutter_exit_watermark[5], bw_int_to_fixed(1000)));
+		calcs_output->stutter_exit_wm_ns[2].d_mark =
+			bw_fixed_to_int(bw_mul(data->
+				stutter_exit_watermark[6], bw_int_to_fixed(1000)));
+		if (ctx->dc->caps.max_slave_planes) {
+			calcs_output->stutter_exit_wm_ns[3].d_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_exit_watermark[0], bw_int_to_fixed(1000)));
+			calcs_output->stutter_exit_wm_ns[4].d_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_exit_watermark[1], bw_int_to_fixed(1000)));
+		} else {
+			calcs_output->stutter_exit_wm_ns[3].d_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_exit_watermark[7], bw_int_to_fixed(1000)));
+			calcs_output->stutter_exit_wm_ns[4].d_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_exit_watermark[8], bw_int_to_fixed(1000)));
+		}
+		calcs_output->stutter_exit_wm_ns[5].d_mark =
+			bw_fixed_to_int(bw_mul(data->
+				stutter_exit_watermark[9], bw_int_to_fixed(1000)));
+
+		calcs_output->stutter_entry_wm_ns[0].d_mark =
+			bw_fixed_to_int(bw_mul(data->
+				stutter_entry_watermark[4], bw_int_to_fixed(1000)));
+		calcs_output->stutter_entry_wm_ns[1].d_mark =
+			bw_fixed_to_int(bw_mul(data->
+				stutter_entry_watermark[5], bw_int_to_fixed(1000)));
+		calcs_output->stutter_entry_wm_ns[2].d_mark =
+			bw_fixed_to_int(bw_mul(data->
+				stutter_entry_watermark[6], bw_int_to_fixed(1000)));
+		if (ctx->dc->caps.max_slave_planes) {
+			calcs_output->stutter_entry_wm_ns[3].d_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_entry_watermark[0], bw_int_to_fixed(1000)));
+			calcs_output->stutter_entry_wm_ns[4].d_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_entry_watermark[1], bw_int_to_fixed(1000)));
+		} else {
+			calcs_output->stutter_entry_wm_ns[3].d_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_entry_watermark[7], bw_int_to_fixed(1000)));
+			calcs_output->stutter_entry_wm_ns[4].d_mark =
+				bw_fixed_to_int(bw_mul(data->
+					stutter_entry_watermark[8], bw_int_to_fixed(1000)));
+		}
+		calcs_output->stutter_entry_wm_ns[5].d_mark =
+			bw_fixed_to_int(bw_mul(data->
+				stutter_entry_watermark[9], bw_int_to_fixed(1000)));
+
+		calcs_output->urgent_wm_ns[0].d_mark =
+			bw_fixed_to_int(bw_mul(data->
+				urgent_watermark[4], bw_int_to_fixed(1000)));
+		calcs_output->urgent_wm_ns[1].d_mark =
+			bw_fixed_to_int(bw_mul(data->
+				urgent_watermark[5], bw_int_to_fixed(1000)));
+		calcs_output->urgent_wm_ns[2].d_mark =
+			bw_fixed_to_int(bw_mul(data->
+				urgent_watermark[6], bw_int_to_fixed(1000)));
+		if (ctx->dc->caps.max_slave_planes) {
+			calcs_output->urgent_wm_ns[3].d_mark =
+				bw_fixed_to_int(bw_mul(data->
+					urgent_watermark[0], bw_int_to_fixed(1000)));
+			calcs_output->urgent_wm_ns[4].d_mark =
+				bw_fixed_to_int(bw_mul(data->
+					urgent_watermark[1], bw_int_to_fixed(1000)));
+		} else {
+			calcs_output->urgent_wm_ns[3].d_mark =
+				bw_fixed_to_int(bw_mul(data->
+					urgent_watermark[7], bw_int_to_fixed(1000)));
+			calcs_output->urgent_wm_ns[4].d_mark =
+				bw_fixed_to_int(bw_mul(data->
+					urgent_watermark[8], bw_int_to_fixed(1000)));
+		}
+		calcs_output->urgent_wm_ns[5].d_mark =
+			bw_fixed_to_int(bw_mul(data->
+				urgent_watermark[9], bw_int_to_fixed(1000)));
+
+		((struct bw_calcs_vbios *)vbios)->low_yclk = low_yclk;
+		((struct bw_calcs_vbios *)vbios)->mid_yclk = mid_yclk;
+		((struct bw_calcs_vbios *)vbios)->low_sclk = low_sclk;
+		((struct bw_calcs_vbios *)vbios)->mid1_sclk = mid1_sclk;
+		((struct bw_calcs_vbios *)vbios)->mid2_sclk = mid2_sclk;
+		((struct bw_calcs_vbios *)vbios)->mid3_sclk = mid3_sclk;
+		((struct bw_calcs_vbios *)vbios)->mid4_sclk = mid4_sclk;
+		((struct bw_calcs_vbios *)vbios)->mid5_sclk = mid5_sclk;
+		((struct bw_calcs_vbios *)vbios)->mid6_sclk = mid6_sclk;
+		((struct bw_calcs_vbios *)vbios)->high_sclk = high_sclk;
+	} else {
+		calcs_output->nbp_state_change_enable = true;
+		calcs_output->cpuc_state_change_enable = true;
+		calcs_output->cpup_state_change_enable = true;
+		calcs_output->stutter_mode_enable = true;
+		calcs_output->dispclk_khz = 0;
+		calcs_output->sclk_khz = 0;
+	}
+
+	kfree(data);
+
+	return is_display_configuration_supported(vbios, calcs_output);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/basics/fixpt31_32.c b/drivers/gpu/drm/amd/display/dc/basics/fixpt31_32.c
new file mode 100644
index 000000000000..6073cadde76c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/basics/fixpt31_32.c
@@ -0,0 +1,510 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+#include "include/fixed31_32.h"
+
+static const struct fixed31_32 dc_fixpt_two_pi = { 26986075409LL };
+static const struct fixed31_32 dc_fixpt_ln2 = { 2977044471LL };
+static const struct fixed31_32 dc_fixpt_ln2_div_2 = { 1488522236LL };
+
+static inline unsigned long long abs_i64(
+	long long arg)
+{
+	if (arg > 0)
+		return (unsigned long long)arg;
+	else
+		return (unsigned long long)(-arg);
+}
+
+/*
+ * @brief
+ * result = dividend / divisor
+ * *remainder = dividend % divisor
+ */
+static inline unsigned long long complete_integer_division_u64(
+	unsigned long long dividend,
+	unsigned long long divisor,
+	unsigned long long *remainder)
+{
+	unsigned long long result;
+
+	result = div64_u64_rem(dividend, divisor, remainder);
+
+	return result;
+}
+
+
+#define FRACTIONAL_PART_MASK \
+	((1ULL << FIXED31_32_BITS_PER_FRACTIONAL_PART) - 1)
+
+#define GET_INTEGER_PART(x) \
+	((x) >> FIXED31_32_BITS_PER_FRACTIONAL_PART)
+
+#define GET_FRACTIONAL_PART(x) \
+	(FRACTIONAL_PART_MASK & (x))
+
+struct fixed31_32 dc_fixpt_from_fraction(long long numerator, long long denominator)
+{
+	struct fixed31_32 res;
+
+	bool arg1_negative = numerator < 0;
+	bool arg2_negative = denominator < 0;
+
+	unsigned long long arg1_value = arg1_negative ? -numerator : numerator;
+	unsigned long long arg2_value = arg2_negative ? -denominator : denominator;
+
+	unsigned long long remainder;
+
+	/* determine integer part */
+
+	unsigned long long res_value = complete_integer_division_u64(
+		arg1_value, arg2_value, &remainder);
+
+	ASSERT(res_value <= LONG_MAX);
+
+	/* determine fractional part */
+	{
+		unsigned int i = FIXED31_32_BITS_PER_FRACTIONAL_PART;
+
+		do {
+			remainder <<= 1;
+
+			res_value <<= 1;
+
+			if (remainder >= arg2_value) {
+				res_value |= 1;
+				remainder -= arg2_value;
+			}
+		} while (--i != 0);
+	}
+
+	/* round up LSB */
+	{
+		unsigned long long summand = (remainder << 1) >= arg2_value;
+
+		ASSERT(res_value <= LLONG_MAX - summand);
+
+		res_value += summand;
+	}
+
+	res.value = (long long)res_value;
+
+	if (arg1_negative ^ arg2_negative)
+		res.value = -res.value;
+
+	return res;
+}
+
+struct fixed31_32 dc_fixpt_mul(struct fixed31_32 arg1, struct fixed31_32 arg2)
+{
+	struct fixed31_32 res;
+
+	bool arg1_negative = arg1.value < 0;
+	bool arg2_negative = arg2.value < 0;
+
+	unsigned long long arg1_value = arg1_negative ? -arg1.value : arg1.value;
+	unsigned long long arg2_value = arg2_negative ? -arg2.value : arg2.value;
+
+	unsigned long long arg1_int = GET_INTEGER_PART(arg1_value);
+	unsigned long long arg2_int = GET_INTEGER_PART(arg2_value);
+
+	unsigned long long arg1_fra = GET_FRACTIONAL_PART(arg1_value);
+	unsigned long long arg2_fra = GET_FRACTIONAL_PART(arg2_value);
+
+	unsigned long long tmp;
+
+	res.value = arg1_int * arg2_int;
+
+	res.value <<= FIXED31_32_BITS_PER_FRACTIONAL_PART;
+
+	tmp = arg1_int * arg2_fra;
+
+	ASSERT(tmp <= (unsigned long long)(LLONG_MAX - res.value));
+
+	res.value += tmp;
+
+	tmp = arg2_int * arg1_fra;
+
+	ASSERT(tmp <= (unsigned long long)(LLONG_MAX - res.value));
+
+	res.value += tmp;
+
+	tmp = arg1_fra * arg2_fra;
+
+	tmp = (tmp >> FIXED31_32_BITS_PER_FRACTIONAL_PART) +
+		(tmp >= (unsigned long long)dc_fixpt_half.value);
+
+	ASSERT(tmp <= (unsigned long long)(LLONG_MAX - res.value));
+
+	res.value += tmp;
+
+	if (arg1_negative ^ arg2_negative)
+		res.value = -res.value;
+
+	return res;
+}
+
+struct fixed31_32 dc_fixpt_sqr(struct fixed31_32 arg)
+{
+	struct fixed31_32 res;
+
+	unsigned long long arg_value = abs_i64(arg.value);
+
+	unsigned long long arg_int = GET_INTEGER_PART(arg_value);
+
+	unsigned long long arg_fra = GET_FRACTIONAL_PART(arg_value);
+
+	unsigned long long tmp;
+
+	res.value = arg_int * arg_int;
+
+	res.value <<= FIXED31_32_BITS_PER_FRACTIONAL_PART;
+
+	tmp = arg_int * arg_fra;
+
+	ASSERT(tmp <= (unsigned long long)(LLONG_MAX - res.value));
+
+	res.value += tmp;
+
+	ASSERT(tmp <= (unsigned long long)(LLONG_MAX - res.value));
+
+	res.value += tmp;
+
+	tmp = arg_fra * arg_fra;
+
+	tmp = (tmp >> FIXED31_32_BITS_PER_FRACTIONAL_PART) +
+		(tmp >= (unsigned long long)dc_fixpt_half.value);
+
+	ASSERT(tmp <= (unsigned long long)(LLONG_MAX - res.value));
+
+	res.value += tmp;
+
+	return res;
+}
+
+struct fixed31_32 dc_fixpt_recip(struct fixed31_32 arg)
+{
+	/*
+	 * @note
+	 * Good idea to use Newton's method
+	 */
+	return dc_fixpt_from_fraction(
+		dc_fixpt_one.value,
+		arg.value);
+}
+
+struct fixed31_32 dc_fixpt_sinc(struct fixed31_32 arg)
+{
+	struct fixed31_32 square;
+
+	struct fixed31_32 res = dc_fixpt_one;
+
+	int n = 27;
+
+	struct fixed31_32 arg_norm = arg;
+
+	if (dc_fixpt_le(
+		dc_fixpt_two_pi,
+		dc_fixpt_abs(arg))) {
+		arg_norm = dc_fixpt_sub(
+			arg_norm,
+			dc_fixpt_mul_int(
+				dc_fixpt_two_pi,
+				(int)div64_s64(
+					arg_norm.value,
+					dc_fixpt_two_pi.value)));
+	}
+
+	square = dc_fixpt_sqr(arg_norm);
+
+	do {
+		res = dc_fixpt_sub(
+			dc_fixpt_one,
+			dc_fixpt_div_int(
+				dc_fixpt_mul(
+					square,
+					res),
+				n * (n - 1)));
+
+		n -= 2;
+	} while (n > 2);
+
+	if (arg.value != arg_norm.value)
+		res = dc_fixpt_div(
+			dc_fixpt_mul(res, arg_norm),
+			arg);
+
+	return res;
+}
+
+struct fixed31_32 dc_fixpt_sin(struct fixed31_32 arg)
+{
+	return dc_fixpt_mul(
+		arg,
+		dc_fixpt_sinc(arg));
+}
+
+struct fixed31_32 dc_fixpt_cos(struct fixed31_32 arg)
+{
+	/* TODO implement argument normalization */
+
+	const struct fixed31_32 square = dc_fixpt_sqr(arg);
+
+	struct fixed31_32 res = dc_fixpt_one;
+
+	int n = 26;
+
+	do {
+		res = dc_fixpt_sub(
+			dc_fixpt_one,
+			dc_fixpt_div_int(
+				dc_fixpt_mul(
+					square,
+					res),
+				(long long)n * (n - 1)));
+
+		n -= 2;
+	} while (n != 0);
+
+	return res;
+}
+
+/*
+ * @brief
+ * result = exp(arg),
+ * where abs(arg) < 1
+ *
+ * Calculated as Taylor series.
+ */
+static struct fixed31_32 fixed31_32_exp_from_taylor_series(struct fixed31_32 arg)
+{
+	unsigned int n = 9;
+
+	struct fixed31_32 res = dc_fixpt_from_fraction(
+		n + 2,
+		n + 1);
+	/* TODO find correct res */
+
+	ASSERT(dc_fixpt_lt(arg, dc_fixpt_one));
+
+	do
+		res = dc_fixpt_add(
+			dc_fixpt_one,
+			dc_fixpt_div_int(
+				dc_fixpt_mul(
+					arg,
+					res),
+				n));
+	while (--n != 1);
+
+	return dc_fixpt_add(
+		dc_fixpt_one,
+		dc_fixpt_mul(
+			arg,
+			res));
+}
+
+struct fixed31_32 dc_fixpt_exp(struct fixed31_32 arg)
+{
+	/*
+	 * @brief
+	 * Main equation is:
+	 * exp(x) = exp(r + m * ln(2)) = (1 << m) * exp(r),
+	 * where m = round(x / ln(2)), r = x - m * ln(2)
+	 */
+
+	if (dc_fixpt_le(
+		dc_fixpt_ln2_div_2,
+		dc_fixpt_abs(arg))) {
+		int m = dc_fixpt_round(
+			dc_fixpt_div(
+				arg,
+				dc_fixpt_ln2));
+
+		struct fixed31_32 r = dc_fixpt_sub(
+			arg,
+			dc_fixpt_mul_int(
+				dc_fixpt_ln2,
+				m));
+
+		ASSERT(m != 0);
+
+		ASSERT(dc_fixpt_lt(
+			dc_fixpt_abs(r),
+			dc_fixpt_one));
+
+		if (m > 0)
+			return dc_fixpt_shl(
+				fixed31_32_exp_from_taylor_series(r),
+				(unsigned char)m);
+		else
+			return dc_fixpt_div_int(
+				fixed31_32_exp_from_taylor_series(r),
+				1LL << -m);
+	} else if (arg.value != 0)
+		return fixed31_32_exp_from_taylor_series(arg);
+	else
+		return dc_fixpt_one;
+}
+
+struct fixed31_32 dc_fixpt_log(struct fixed31_32 arg)
+{
+	struct fixed31_32 res = dc_fixpt_neg(dc_fixpt_one);
+	/* TODO improve 1st estimation */
+
+	struct fixed31_32 error;
+
+	ASSERT(arg.value > 0);
+	/* TODO if arg is negative, return NaN */
+	/* TODO if arg is zero, return -INF */
+
+	do {
+		struct fixed31_32 res1 = dc_fixpt_add(
+			dc_fixpt_sub(
+				res,
+				dc_fixpt_one),
+			dc_fixpt_div(
+				arg,
+				dc_fixpt_exp(res)));
+
+		error = dc_fixpt_sub(
+			res,
+			res1);
+
+		res = res1;
+		/* TODO determine max_allowed_error based on quality of exp() */
+	} while (abs_i64(error.value) > 100ULL);
+
+	return res;
+}
+
+
+/* this function is a generic helper to translate fixed point value to
+ * specified integer format that will consist of integer_bits integer part and
+ * fractional_bits fractional part. For example it is used in
+ * dc_fixpt_u2d19 to receive 2 bits integer part and 19 bits fractional
+ * part in 32 bits. It is used in hw programming (scaler)
+ */
+
+static inline unsigned int ux_dy(
+	long long value,
+	unsigned int integer_bits,
+	unsigned int fractional_bits)
+{
+	/* 1. create mask of integer part */
+	unsigned int result = (1 << integer_bits) - 1;
+	/* 2. mask out fractional part */
+	unsigned int fractional_part = FRACTIONAL_PART_MASK & value;
+	/* 3. shrink fixed point integer part to be of integer_bits width*/
+	result &= GET_INTEGER_PART(value);
+	/* 4. make space for fractional part to be filled in after integer */
+	result <<= fractional_bits;
+	/* 5. shrink fixed point fractional part to of fractional_bits width*/
+	fractional_part >>= FIXED31_32_BITS_PER_FRACTIONAL_PART - fractional_bits;
+	/* 6. merge the result */
+	return result | fractional_part;
+}
+
+static inline unsigned int clamp_ux_dy(
+	long long value,
+	unsigned int integer_bits,
+	unsigned int fractional_bits,
+	unsigned int min_clamp)
+{
+	unsigned int truncated_val = ux_dy(value, integer_bits, fractional_bits);
+
+	if (value >= (1LL << (integer_bits + FIXED31_32_BITS_PER_FRACTIONAL_PART)))
+		return (1 << (integer_bits + fractional_bits)) - 1;
+	else if (truncated_val > min_clamp)
+		return truncated_val;
+	else
+		return min_clamp;
+}
+
+unsigned int dc_fixpt_u4d19(struct fixed31_32 arg)
+{
+	return ux_dy(arg.value, 4, 19);
+}
+
+unsigned int dc_fixpt_u3d19(struct fixed31_32 arg)
+{
+	return ux_dy(arg.value, 3, 19);
+}
+
+unsigned int dc_fixpt_u2d19(struct fixed31_32 arg)
+{
+	return ux_dy(arg.value, 2, 19);
+}
+
+unsigned int dc_fixpt_u0d19(struct fixed31_32 arg)
+{
+	return ux_dy(arg.value, 0, 19);
+}
+
+unsigned int dc_fixpt_clamp_u0d14(struct fixed31_32 arg)
+{
+	return clamp_ux_dy(arg.value, 0, 14, 1);
+}
+
+unsigned int dc_fixpt_clamp_u0d10(struct fixed31_32 arg)
+{
+	return clamp_ux_dy(arg.value, 0, 10, 1);
+}
+
+int dc_fixpt_s4d19(struct fixed31_32 arg)
+{
+	if (arg.value < 0)
+		return -(int)ux_dy(dc_fixpt_abs(arg).value, 4, 19);
+	else
+		return ux_dy(arg.value, 4, 19);
+}
+
+struct fixed31_32 dc_fixpt_from_ux_dy(unsigned int value,
+	unsigned int integer_bits,
+	unsigned int fractional_bits)
+{
+	struct fixed31_32 fixpt_value = dc_fixpt_zero;
+	struct fixed31_32 fixpt_int_value = dc_fixpt_zero;
+	long long frac_mask = ((long long)1 << (long long)integer_bits) - 1;
+
+	fixpt_value.value = (long long)value << (FIXED31_32_BITS_PER_FRACTIONAL_PART - fractional_bits);
+	frac_mask = frac_mask << fractional_bits;
+	fixpt_int_value.value = value & frac_mask;
+	fixpt_int_value.value <<= (FIXED31_32_BITS_PER_FRACTIONAL_PART - fractional_bits);
+	fixpt_value.value |= fixpt_int_value.value;
+	return fixpt_value;
+}
+
+struct fixed31_32 dc_fixpt_from_int_dy(unsigned int int_value,
+	unsigned int frac_value,
+	unsigned int integer_bits,
+	unsigned int fractional_bits)
+{
+	struct fixed31_32 fixpt_value = dc_fixpt_from_int(int_value);
+
+	fixpt_value.value |= (long long)frac_value << (FIXED31_32_BITS_PER_FRACTIONAL_PART - fractional_bits);
+	return fixpt_value;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/basics/logger.h b/drivers/gpu/drm/amd/display/dc/basics/logger.h
new file mode 100644
index 000000000000..09722f0f8aa3
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/basics/logger.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_LOGGER_H__
+#define __DAL_LOGGER_H__
+
+
+#endif /* __DAL_LOGGER_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/basics/vector.c b/drivers/gpu/drm/amd/display/dc/basics/vector.c
new file mode 100644
index 000000000000..b413a672c2c0
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/basics/vector.c
@@ -0,0 +1,306 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+#include "include/vector.h"
+
+bool dal_vector_construct(
+	struct vector *vector,
+	struct dc_context *ctx,
+	uint32_t capacity,
+	uint32_t struct_size)
+{
+	vector->container = NULL;
+
+	if (!struct_size || !capacity) {
+		/* Container must be non-zero size*/
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+
+	vector->container = kcalloc(capacity, struct_size, GFP_KERNEL);
+	if (vector->container == NULL)
+		return false;
+	vector->capacity = capacity;
+	vector->struct_size = struct_size;
+	vector->count = 0;
+	vector->ctx = ctx;
+	return true;
+}
+
+static bool dal_vector_presized_costruct(struct vector *vector,
+					 struct dc_context *ctx,
+					 uint32_t count,
+					 void *initial_value,
+					 uint32_t struct_size)
+{
+	uint32_t i;
+
+	vector->container = NULL;
+
+	if (!struct_size || !count) {
+		/* Container must be non-zero size*/
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+
+	vector->container = kcalloc(count, struct_size, GFP_KERNEL);
+
+	if (vector->container == NULL)
+		return false;
+
+	/* If caller didn't supply initial value then the default
+	 * of all zeros is expected, which is exactly what dal_alloc()
+	 * initialises the memory to. */
+	if (NULL != initial_value) {
+		for (i = 0; i < count; ++i)
+			memmove(
+				vector->container + i * struct_size,
+				initial_value,
+				struct_size);
+	}
+
+	vector->capacity = count;
+	vector->struct_size = struct_size;
+	vector->count = count;
+	return true;
+}
+
+struct vector *dal_vector_presized_create(
+	struct dc_context *ctx,
+	uint32_t size,
+	void *initial_value,
+	uint32_t struct_size)
+{
+	struct vector *vector = kzalloc(sizeof(struct vector), GFP_KERNEL);
+
+	if (vector == NULL)
+		return NULL;
+
+	if (dal_vector_presized_costruct(
+		vector, ctx, size, initial_value, struct_size))
+		return vector;
+
+	BREAK_TO_DEBUGGER();
+	kfree(vector);
+	return NULL;
+}
+
+struct vector *dal_vector_create(
+	struct dc_context *ctx,
+	uint32_t capacity,
+	uint32_t struct_size)
+{
+	struct vector *vector = kzalloc(sizeof(struct vector), GFP_KERNEL);
+
+	if (vector == NULL)
+		return NULL;
+
+	if (dal_vector_construct(vector, ctx, capacity, struct_size))
+		return vector;
+
+	BREAK_TO_DEBUGGER();
+	kfree(vector);
+	return NULL;
+}
+
+void dal_vector_destruct(
+	struct vector *vector)
+{
+	kfree(vector->container);
+	vector->count = 0;
+	vector->capacity = 0;
+}
+
+void dal_vector_destroy(
+	struct vector **vector)
+{
+	if (vector == NULL || *vector == NULL)
+		return;
+	dal_vector_destruct(*vector);
+	kfree(*vector);
+	*vector = NULL;
+}
+
+uint32_t dal_vector_get_count(
+	const struct vector *vector)
+{
+	return vector->count;
+}
+
+void *dal_vector_at_index(
+	const struct vector *vector,
+	uint32_t index)
+{
+	if (vector->container == NULL || index >= vector->count)
+		return NULL;
+	return vector->container + (index * vector->struct_size);
+}
+
+bool dal_vector_remove_at_index(
+	struct vector *vector,
+	uint32_t index)
+{
+	if (index >= vector->count)
+		return false;
+
+	if (index != vector->count - 1)
+		memmove(
+			vector->container + (index * vector->struct_size),
+			vector->container + ((index + 1) * vector->struct_size),
+			(size_t)(vector->count - index - 1) * vector->struct_size);
+	vector->count -= 1;
+
+	return true;
+}
+
+void dal_vector_set_at_index(
+	const struct vector *vector,
+	const void *what,
+	uint32_t index)
+{
+	void *where = dal_vector_at_index(vector, index);
+
+	if (!where) {
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+	memmove(
+		where,
+		what,
+		vector->struct_size);
+}
+
+static inline uint32_t calc_increased_capacity(
+	uint32_t old_capacity)
+{
+	return old_capacity * 2;
+}
+
+bool dal_vector_insert_at(
+	struct vector *vector,
+	const void *what,
+	uint32_t position)
+{
+	uint8_t *insert_address;
+
+	if (vector->count == vector->capacity) {
+		if (!dal_vector_reserve(
+			vector,
+			calc_increased_capacity(vector->capacity)))
+			return false;
+	}
+
+	insert_address = vector->container + (vector->struct_size * position);
+
+	if (vector->count && position < vector->count)
+		memmove(
+			insert_address + vector->struct_size,
+			insert_address,
+			(size_t)vector->struct_size * (vector->count - position));
+
+	memmove(
+		insert_address,
+		what,
+		vector->struct_size);
+
+	vector->count++;
+
+	return true;
+}
+
+bool dal_vector_append(
+	struct vector *vector,
+	const void *item)
+{
+	return dal_vector_insert_at(vector, item, vector->count);
+}
+
+struct vector *dal_vector_clone(
+	const struct vector *vector)
+{
+	struct vector *vec_cloned;
+	uint32_t count;
+
+	/* create new vector */
+	count = dal_vector_get_count(vector);
+
+	if (count == 0)
+		/* when count is 0 we still want to create clone of the vector
+		 */
+		vec_cloned = dal_vector_create(
+			vector->ctx,
+			vector->capacity,
+			vector->struct_size);
+	else
+		/* Call "presized create" version, independently of how the
+		 * original vector was created.
+		 * The owner of original vector must know how to treat the new
+		 * vector - as "presized" or as "regular".
+		 * But from vector point of view it doesn't matter. */
+		vec_cloned = dal_vector_presized_create(vector->ctx, count,
+			NULL,/* no initial value */
+			vector->struct_size);
+
+	if (NULL == vec_cloned) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	/* copy vector's data */
+	memmove(vec_cloned->container, vector->container,
+			(size_t)vec_cloned->struct_size * vec_cloned->capacity);
+
+	return vec_cloned;
+}
+
+uint32_t dal_vector_capacity(const struct vector *vector)
+{
+	return vector->capacity;
+}
+
+bool dal_vector_reserve(struct vector *vector, uint32_t capacity)
+{
+	void *new_container;
+
+	if (capacity <= vector->capacity)
+		return true;
+
+	new_container = krealloc(vector->container,
+				 capacity * vector->struct_size, GFP_KERNEL);
+
+	if (new_container) {
+		vector->container = new_container;
+		vector->capacity = capacity;
+		return true;
+	}
+
+	return false;
+}
+
+void dal_vector_clear(struct vector *vector)
+{
+	vector->count = 0;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/bios/Makefile b/drivers/gpu/drm/amd/display/dc/bios/Makefile
new file mode 100644
index 000000000000..ed6b5e9763f6
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/Makefile
@@ -0,0 +1,57 @@
+#
+# Copyright 2017 Advanced Micro Devices, Inc.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+#
+# Makefile for the 'bios' sub-component of DAL.
+# It provides the parsing and executing controls for atom bios image.
+
+BIOS = bios_parser.o bios_parser_interface.o  bios_parser_helper.o command_table.o command_table_helper.o bios_parser_common.o
+
+BIOS += command_table2.o command_table_helper2.o bios_parser2.o
+
+AMD_DAL_BIOS = $(addprefix $(AMDDALPATH)/dc/bios/,$(BIOS))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_BIOS)
+
+###############################################################################
+# DCE 6x
+###############################################################################
+# All DCE6.x are derived from DCE6.0, so 6.0 MUST be defined if ANY of
+# DCE6.x is compiled.
+ifdef CONFIG_DRM_AMD_DC_SI
+AMD_DISPLAY_FILES += $(AMDDALPATH)/dc/bios/dce60/command_table_helper_dce60.o
+endif
+
+###############################################################################
+# DCE 8x
+###############################################################################
+# All DCE8.x are derived from DCE8.0, so 8.0 MUST be defined if ANY of
+# DCE8.x is compiled.
+AMD_DISPLAY_FILES += $(AMDDALPATH)/dc/bios/dce80/command_table_helper_dce80.o
+
+###############################################################################
+# DCE 11x
+###############################################################################
+AMD_DISPLAY_FILES += $(AMDDALPATH)/dc/bios/dce110/command_table_helper_dce110.o
+
+AMD_DISPLAY_FILES += $(AMDDALPATH)/dc/bios/dce112/command_table_helper_dce112.o
+
+AMD_DISPLAY_FILES += $(AMDDALPATH)/dc/bios/dce112/command_table_helper2_dce112.o
diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c b/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c
new file mode 100644
index 000000000000..154fd2c18e88
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c
@@ -0,0 +1,2945 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include <linux/slab.h>
+
+#include "dm_services.h"
+
+#include "atom.h"
+
+#include "dc_bios_types.h"
+#include "include/gpio_service_interface.h"
+#include "include/grph_object_ctrl_defs.h"
+#include "include/bios_parser_interface.h"
+#include "include/logger_interface.h"
+
+#include "command_table.h"
+#include "bios_parser_helper.h"
+#include "command_table_helper.h"
+#include "bios_parser.h"
+#include "bios_parser_types_internal.h"
+#include "bios_parser_interface.h"
+
+#include "bios_parser_common.h"
+
+#define THREE_PERCENT_OF_10000 300
+
+#define LAST_RECORD_TYPE 0xff
+
+#define DC_LOGGER \
+	bp->base.ctx->logger
+
+#define DATA_TABLES(table) (bp->master_data_tbl->ListOfDataTables.table)
+
+static void get_atom_data_table_revision(
+	ATOM_COMMON_TABLE_HEADER *atom_data_tbl,
+	struct atom_data_revision *tbl_revision);
+static uint32_t get_src_obj_list(struct bios_parser *bp, ATOM_OBJECT *object,
+	uint16_t **id_list);
+static ATOM_OBJECT *get_bios_object(struct bios_parser *bp,
+	struct graphics_object_id id);
+static enum bp_result get_gpio_i2c_info(struct bios_parser *bp,
+	ATOM_I2C_RECORD *record,
+	struct graphics_object_i2c_info *info);
+static ATOM_HPD_INT_RECORD *get_hpd_record(struct bios_parser *bp,
+	ATOM_OBJECT *object);
+static struct device_id device_type_from_device_id(uint16_t device_id);
+static uint32_t signal_to_ss_id(enum as_signal_type signal);
+static uint32_t get_support_mask_for_device_id(struct device_id device_id);
+static ATOM_ENCODER_CAP_RECORD_V2 *get_encoder_cap_record(
+	struct bios_parser *bp,
+	ATOM_OBJECT *object);
+
+#define BIOS_IMAGE_SIZE_OFFSET 2
+#define BIOS_IMAGE_SIZE_UNIT 512
+
+/*****************************************************************************/
+static bool bios_parser_construct(
+	struct bios_parser *bp,
+	struct bp_init_data *init,
+	enum dce_version dce_version);
+
+static uint8_t bios_parser_get_connectors_number(
+	struct dc_bios *dcb);
+
+static enum bp_result bios_parser_get_embedded_panel_info(
+	struct dc_bios *dcb,
+	struct embedded_panel_info *info);
+
+/*****************************************************************************/
+
+struct dc_bios *bios_parser_create(
+	struct bp_init_data *init,
+	enum dce_version dce_version)
+{
+	struct bios_parser *bp;
+
+	bp = kzalloc(sizeof(struct bios_parser), GFP_KERNEL);
+	if (!bp)
+		return NULL;
+
+	if (bios_parser_construct(bp, init, dce_version))
+		return &bp->base;
+
+	kfree(bp);
+	BREAK_TO_DEBUGGER();
+	return NULL;
+}
+
+static void bios_parser_destruct(struct bios_parser *bp)
+{
+	kfree(bp->base.bios_local_image);
+	kfree(bp->base.integrated_info);
+}
+
+static void bios_parser_destroy(struct dc_bios **dcb)
+{
+	struct bios_parser *bp = BP_FROM_DCB(*dcb);
+
+	if (!bp) {
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+	bios_parser_destruct(bp);
+
+	kfree(bp);
+	*dcb = NULL;
+}
+
+static uint8_t get_number_of_objects(struct bios_parser *bp, uint32_t offset)
+{
+	ATOM_OBJECT_TABLE *table;
+
+	uint32_t object_table_offset = bp->object_info_tbl_offset + offset;
+
+	table = ((ATOM_OBJECT_TABLE *) bios_get_image(&bp->base,
+				object_table_offset,
+				struct_size(table, asObjects, 1)));
+
+	if (!table)
+		return 0;
+	else
+		return table->ucNumberOfObjects;
+}
+
+static uint8_t bios_parser_get_connectors_number(struct dc_bios *dcb)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	return get_number_of_objects(bp,
+		le16_to_cpu(bp->object_info_tbl.v1_1->usConnectorObjectTableOffset));
+}
+
+static struct graphics_object_id bios_parser_get_connector_id(
+	struct dc_bios *dcb,
+	uint8_t i)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	struct graphics_object_id object_id = dal_graphics_object_id_init(
+		0, ENUM_ID_UNKNOWN, OBJECT_TYPE_UNKNOWN);
+	uint16_t id;
+
+	uint32_t connector_table_offset = bp->object_info_tbl_offset
+		+ le16_to_cpu(bp->object_info_tbl.v1_1->usConnectorObjectTableOffset);
+
+	ATOM_OBJECT_TABLE *tbl = ((ATOM_OBJECT_TABLE *) bios_get_image(&bp->base,
+				connector_table_offset,
+				struct_size(tbl, asObjects, 1)));
+
+	if (!tbl) {
+		dm_error("Can't get connector table from atom bios.\n");
+		return object_id;
+	}
+
+	if (tbl->ucNumberOfObjects <= i)
+		return object_id;
+
+	id = le16_to_cpu(tbl->asObjects[i].usObjectID);
+	object_id = object_id_from_bios_object_id(id);
+	return object_id;
+}
+
+static enum bp_result bios_parser_get_src_obj(struct dc_bios *dcb,
+	struct graphics_object_id object_id, uint32_t index,
+	struct graphics_object_id *src_object_id)
+{
+	uint32_t number;
+	uint16_t *id;
+	ATOM_OBJECT *object;
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	if (!src_object_id)
+		return BP_RESULT_BADINPUT;
+
+	object = get_bios_object(bp, object_id);
+
+	if (!object) {
+		BREAK_TO_DEBUGGER(); /* Invalid object id */
+		return BP_RESULT_BADINPUT;
+	}
+
+	number = get_src_obj_list(bp, object, &id);
+
+	if (number <= index)
+		return BP_RESULT_BADINPUT;
+
+	*src_object_id = object_id_from_bios_object_id(id[index]);
+
+	return BP_RESULT_OK;
+}
+
+static enum bp_result bios_parser_get_i2c_info(struct dc_bios *dcb,
+	struct graphics_object_id id,
+	struct graphics_object_i2c_info *info)
+{
+	uint32_t offset;
+	ATOM_OBJECT *object;
+	ATOM_COMMON_RECORD_HEADER *header;
+	ATOM_I2C_RECORD *record;
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	if (!info)
+		return BP_RESULT_BADINPUT;
+
+	object = get_bios_object(bp, id);
+
+	if (!object)
+		return BP_RESULT_BADINPUT;
+
+	offset = le16_to_cpu(object->usRecordOffset)
+			+ bp->object_info_tbl_offset;
+
+	for (;;) {
+		header = GET_IMAGE(ATOM_COMMON_RECORD_HEADER, offset);
+
+		if (!header)
+			return BP_RESULT_BADBIOSTABLE;
+
+		if (LAST_RECORD_TYPE == header->ucRecordType ||
+			!header->ucRecordSize)
+			break;
+
+		if (ATOM_I2C_RECORD_TYPE == header->ucRecordType
+			&& sizeof(ATOM_I2C_RECORD) <= header->ucRecordSize) {
+			/* get the I2C info */
+			record = (ATOM_I2C_RECORD *) header;
+
+			if (get_gpio_i2c_info(bp, record, info) == BP_RESULT_OK)
+				return BP_RESULT_OK;
+		}
+
+		offset += header->ucRecordSize;
+	}
+
+	return BP_RESULT_NORECORD;
+}
+
+static enum bp_result bios_parser_get_hpd_info(struct dc_bios *dcb,
+	struct graphics_object_id id,
+	struct graphics_object_hpd_info *info)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	ATOM_OBJECT *object;
+	ATOM_HPD_INT_RECORD *record = NULL;
+
+	if (!info)
+		return BP_RESULT_BADINPUT;
+
+	object = get_bios_object(bp, id);
+
+	if (!object)
+		return BP_RESULT_BADINPUT;
+
+	record = get_hpd_record(bp, object);
+
+	if (record != NULL) {
+		info->hpd_int_gpio_uid = record->ucHPDIntGPIOID;
+		info->hpd_active = record->ucPlugged_PinState;
+		return BP_RESULT_OK;
+	}
+
+	return BP_RESULT_NORECORD;
+}
+
+static enum bp_result bios_parser_get_device_tag_record(
+	struct bios_parser *bp,
+	ATOM_OBJECT *object,
+	ATOM_CONNECTOR_DEVICE_TAG_RECORD **record)
+{
+	ATOM_COMMON_RECORD_HEADER *header;
+	uint32_t offset;
+
+	offset = le16_to_cpu(object->usRecordOffset)
+			+ bp->object_info_tbl_offset;
+
+	for (;;) {
+		header = GET_IMAGE(ATOM_COMMON_RECORD_HEADER, offset);
+
+		if (!header)
+			return BP_RESULT_BADBIOSTABLE;
+
+		offset += header->ucRecordSize;
+
+		if (LAST_RECORD_TYPE == header->ucRecordType ||
+			!header->ucRecordSize)
+			break;
+
+		if (ATOM_CONNECTOR_DEVICE_TAG_RECORD_TYPE !=
+			header->ucRecordType)
+			continue;
+
+		if (sizeof(ATOM_CONNECTOR_DEVICE_TAG) > header->ucRecordSize)
+			continue;
+
+		*record = (ATOM_CONNECTOR_DEVICE_TAG_RECORD *) header;
+		return BP_RESULT_OK;
+	}
+
+	return BP_RESULT_NORECORD;
+}
+
+static enum bp_result bios_parser_get_device_tag(
+	struct dc_bios *dcb,
+	struct graphics_object_id connector_object_id,
+	uint32_t device_tag_index,
+	struct connector_device_tag_info *info)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	ATOM_OBJECT *object;
+	ATOM_CONNECTOR_DEVICE_TAG_RECORD *record = NULL;
+	ATOM_CONNECTOR_DEVICE_TAG *device_tag;
+
+	if (!info)
+		return BP_RESULT_BADINPUT;
+
+	/* getBiosObject will return MXM object */
+	object = get_bios_object(bp, connector_object_id);
+
+	if (!object) {
+		BREAK_TO_DEBUGGER(); /* Invalid object id */
+		return BP_RESULT_BADINPUT;
+	}
+
+	if (bios_parser_get_device_tag_record(bp, object, &record)
+		!= BP_RESULT_OK)
+		return BP_RESULT_NORECORD;
+
+	if (device_tag_index >= record->ucNumberOfDevice)
+		return BP_RESULT_NORECORD;
+
+	device_tag = &record->asDeviceTag[device_tag_index];
+
+	info->acpi_device = le32_to_cpu(device_tag->ulACPIDeviceEnum);
+	info->dev_id =
+		device_type_from_device_id(le16_to_cpu(device_tag->usDeviceID));
+
+	return BP_RESULT_OK;
+}
+
+static enum bp_result get_firmware_info_v1_4(
+	struct bios_parser *bp,
+	struct dc_firmware_info *info);
+static enum bp_result get_firmware_info_v2_1(
+	struct bios_parser *bp,
+	struct dc_firmware_info *info);
+static enum bp_result get_firmware_info_v2_2(
+	struct bios_parser *bp,
+	struct dc_firmware_info *info);
+
+static enum bp_result bios_parser_get_firmware_info(
+	struct dc_bios *dcb,
+	struct dc_firmware_info *info)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	enum bp_result result = BP_RESULT_BADBIOSTABLE;
+	ATOM_COMMON_TABLE_HEADER *header;
+	struct atom_data_revision revision;
+
+	if (info && DATA_TABLES(FirmwareInfo)) {
+		header = GET_IMAGE(ATOM_COMMON_TABLE_HEADER,
+			DATA_TABLES(FirmwareInfo));
+		get_atom_data_table_revision(header, &revision);
+		switch (revision.major) {
+		case 1:
+			switch (revision.minor) {
+			case 4:
+				result = get_firmware_info_v1_4(bp, info);
+				break;
+			default:
+				break;
+			}
+			break;
+
+		case 2:
+			switch (revision.minor) {
+			case 1:
+				result = get_firmware_info_v2_1(bp, info);
+				break;
+			case 2:
+				result = get_firmware_info_v2_2(bp, info);
+				break;
+			default:
+				break;
+			}
+			break;
+		default:
+			break;
+		}
+	}
+
+	return result;
+}
+
+static enum bp_result get_firmware_info_v1_4(
+	struct bios_parser *bp,
+	struct dc_firmware_info *info)
+{
+	ATOM_FIRMWARE_INFO_V1_4 *firmware_info =
+		GET_IMAGE(ATOM_FIRMWARE_INFO_V1_4,
+			DATA_TABLES(FirmwareInfo));
+
+	if (!info)
+		return BP_RESULT_BADINPUT;
+
+	if (!firmware_info)
+		return BP_RESULT_BADBIOSTABLE;
+
+	memset(info, 0, sizeof(*info));
+
+	/* Pixel clock pll information. We need to convert from 10KHz units into
+	 * KHz units */
+	info->pll_info.crystal_frequency =
+		le16_to_cpu(firmware_info->usReferenceClock) * 10;
+	info->pll_info.min_input_pxl_clk_pll_frequency =
+		le16_to_cpu(firmware_info->usMinPixelClockPLL_Input) * 10;
+	info->pll_info.max_input_pxl_clk_pll_frequency =
+		le16_to_cpu(firmware_info->usMaxPixelClockPLL_Input) * 10;
+	info->pll_info.min_output_pxl_clk_pll_frequency =
+		le32_to_cpu(firmware_info->ulMinPixelClockPLL_Output) * 10;
+	info->pll_info.max_output_pxl_clk_pll_frequency =
+		le32_to_cpu(firmware_info->ulMaxPixelClockPLL_Output) * 10;
+
+	if (firmware_info->usFirmwareCapability.sbfAccess.MemoryClockSS_Support)
+		/* Since there is no information on the SS, report conservative
+		 * value 3% for bandwidth calculation */
+		/* unit of 0.01% */
+		info->feature.memory_clk_ss_percentage = THREE_PERCENT_OF_10000;
+
+	if (firmware_info->usFirmwareCapability.sbfAccess.EngineClockSS_Support)
+		/* Since there is no information on the SS,report conservative
+		 * value 3% for bandwidth calculation */
+		/* unit of 0.01% */
+		info->feature.engine_clk_ss_percentage = THREE_PERCENT_OF_10000;
+
+	return BP_RESULT_OK;
+}
+
+static enum bp_result get_ss_info_v3_1(
+	struct bios_parser *bp,
+	uint32_t id,
+	uint32_t index,
+	struct spread_spectrum_info *ss_info);
+
+static enum bp_result get_firmware_info_v2_1(
+	struct bios_parser *bp,
+	struct dc_firmware_info *info)
+{
+	ATOM_FIRMWARE_INFO_V2_1 *firmwareInfo =
+		GET_IMAGE(ATOM_FIRMWARE_INFO_V2_1, DATA_TABLES(FirmwareInfo));
+	struct spread_spectrum_info internalSS;
+	uint32_t index;
+
+	if (!info)
+		return BP_RESULT_BADINPUT;
+
+	if (!firmwareInfo)
+		return BP_RESULT_BADBIOSTABLE;
+
+	memset(info, 0, sizeof(*info));
+
+	/* Pixel clock pll information. We need to convert from 10KHz units into
+	 * KHz units */
+	info->pll_info.crystal_frequency =
+		le16_to_cpu(firmwareInfo->usCoreReferenceClock) * 10;
+	info->pll_info.min_input_pxl_clk_pll_frequency =
+		le16_to_cpu(firmwareInfo->usMinPixelClockPLL_Input) * 10;
+	info->pll_info.max_input_pxl_clk_pll_frequency =
+		le16_to_cpu(firmwareInfo->usMaxPixelClockPLL_Input) * 10;
+	info->pll_info.min_output_pxl_clk_pll_frequency =
+		le32_to_cpu(firmwareInfo->ulMinPixelClockPLL_Output) * 10;
+	info->pll_info.max_output_pxl_clk_pll_frequency =
+		le32_to_cpu(firmwareInfo->ulMaxPixelClockPLL_Output) * 10;
+	info->default_display_engine_pll_frequency =
+		le32_to_cpu(firmwareInfo->ulDefaultDispEngineClkFreq) * 10;
+	info->external_clock_source_frequency_for_dp =
+		le16_to_cpu(firmwareInfo->usUniphyDPModeExtClkFreq) * 10;
+	info->min_allowed_bl_level = firmwareInfo->ucMinAllowedBL_Level;
+
+	/* There should be only one entry in the SS info table for Memory Clock
+	 */
+	index = 0;
+	if (firmwareInfo->usFirmwareCapability.sbfAccess.MemoryClockSS_Support)
+		/* Since there is no information for external SS, report
+		 *  conservative value 3% for bandwidth calculation */
+		/* unit of 0.01% */
+		info->feature.memory_clk_ss_percentage = THREE_PERCENT_OF_10000;
+	else if (get_ss_info_v3_1(bp,
+		ASIC_INTERNAL_MEMORY_SS, index, &internalSS) == BP_RESULT_OK) {
+		if (internalSS.spread_spectrum_percentage) {
+			info->feature.memory_clk_ss_percentage =
+				internalSS.spread_spectrum_percentage;
+			if (internalSS.type.CENTER_MODE) {
+				/* if it is centermode, the exact SS Percentage
+				 * will be round up of half of the percentage
+				 * reported in the SS table */
+				++info->feature.memory_clk_ss_percentage;
+				info->feature.memory_clk_ss_percentage /= 2;
+			}
+		}
+	}
+
+	/* There should be only one entry in the SS info table for Engine Clock
+	 */
+	index = 1;
+	if (firmwareInfo->usFirmwareCapability.sbfAccess.EngineClockSS_Support)
+		/* Since there is no information for external SS, report
+		 * conservative value 3% for bandwidth calculation */
+		/* unit of 0.01% */
+		info->feature.engine_clk_ss_percentage = THREE_PERCENT_OF_10000;
+	else if (get_ss_info_v3_1(bp,
+		ASIC_INTERNAL_ENGINE_SS, index, &internalSS) == BP_RESULT_OK) {
+		if (internalSS.spread_spectrum_percentage) {
+			info->feature.engine_clk_ss_percentage =
+				internalSS.spread_spectrum_percentage;
+			if (internalSS.type.CENTER_MODE) {
+				/* if it is centermode, the exact SS Percentage
+				 * will be round up of half of the percentage
+				 * reported in the SS table */
+				++info->feature.engine_clk_ss_percentage;
+				info->feature.engine_clk_ss_percentage /= 2;
+			}
+		}
+	}
+
+	return BP_RESULT_OK;
+}
+
+static enum bp_result get_firmware_info_v2_2(
+	struct bios_parser *bp,
+	struct dc_firmware_info *info)
+{
+	ATOM_FIRMWARE_INFO_V2_2 *firmware_info;
+	struct spread_spectrum_info internal_ss;
+	uint32_t index;
+
+	if (!info)
+		return BP_RESULT_BADINPUT;
+
+	firmware_info = GET_IMAGE(ATOM_FIRMWARE_INFO_V2_2,
+		DATA_TABLES(FirmwareInfo));
+
+	if (!firmware_info)
+		return BP_RESULT_BADBIOSTABLE;
+
+	memset(info, 0, sizeof(*info));
+
+	/* Pixel clock pll information. We need to convert from 10KHz units into
+	 * KHz units */
+	info->pll_info.crystal_frequency =
+		le16_to_cpu(firmware_info->usCoreReferenceClock) * 10;
+	info->pll_info.min_input_pxl_clk_pll_frequency =
+		le16_to_cpu(firmware_info->usMinPixelClockPLL_Input) * 10;
+	info->pll_info.max_input_pxl_clk_pll_frequency =
+		le16_to_cpu(firmware_info->usMaxPixelClockPLL_Input) * 10;
+	info->pll_info.min_output_pxl_clk_pll_frequency =
+		le32_to_cpu(firmware_info->ulMinPixelClockPLL_Output) * 10;
+	info->pll_info.max_output_pxl_clk_pll_frequency =
+		le32_to_cpu(firmware_info->ulMaxPixelClockPLL_Output) * 10;
+	info->default_display_engine_pll_frequency =
+		le32_to_cpu(firmware_info->ulDefaultDispEngineClkFreq) * 10;
+	info->external_clock_source_frequency_for_dp =
+		le16_to_cpu(firmware_info->usUniphyDPModeExtClkFreq) * 10;
+
+	/* There should be only one entry in the SS info table for Memory Clock
+	 */
+	index = 0;
+	if (firmware_info->usFirmwareCapability.sbfAccess.MemoryClockSS_Support)
+		/* Since there is no information for external SS, report
+		 *  conservative value 3% for bandwidth calculation */
+		/* unit of 0.01% */
+		info->feature.memory_clk_ss_percentage = THREE_PERCENT_OF_10000;
+	else if (get_ss_info_v3_1(bp,
+			ASIC_INTERNAL_MEMORY_SS, index, &internal_ss) == BP_RESULT_OK) {
+		if (internal_ss.spread_spectrum_percentage) {
+			info->feature.memory_clk_ss_percentage =
+					internal_ss.spread_spectrum_percentage;
+			if (internal_ss.type.CENTER_MODE) {
+				/* if it is centermode, the exact SS Percentage
+				 * will be round up of half of the percentage
+				 * reported in the SS table */
+				++info->feature.memory_clk_ss_percentage;
+				info->feature.memory_clk_ss_percentage /= 2;
+			}
+		}
+	}
+
+	/* There should be only one entry in the SS info table for Engine Clock
+	 */
+	index = 1;
+	if (firmware_info->usFirmwareCapability.sbfAccess.EngineClockSS_Support)
+		/* Since there is no information for external SS, report
+		 * conservative value 3% for bandwidth calculation */
+		/* unit of 0.01% */
+		info->feature.engine_clk_ss_percentage = THREE_PERCENT_OF_10000;
+	else if (get_ss_info_v3_1(bp,
+			ASIC_INTERNAL_ENGINE_SS, index, &internal_ss) == BP_RESULT_OK) {
+		if (internal_ss.spread_spectrum_percentage) {
+			info->feature.engine_clk_ss_percentage =
+					internal_ss.spread_spectrum_percentage;
+			if (internal_ss.type.CENTER_MODE) {
+				/* if it is centermode, the exact SS Percentage
+				 * will be round up of half of the percentage
+				 * reported in the SS table */
+				++info->feature.engine_clk_ss_percentage;
+				info->feature.engine_clk_ss_percentage /= 2;
+			}
+		}
+	}
+
+	/* Remote Display */
+	info->remote_display_config = firmware_info->ucRemoteDisplayConfig;
+
+	/* Is allowed minimum BL level */
+	info->min_allowed_bl_level = firmware_info->ucMinAllowedBL_Level;
+	/* Used starting from CI */
+	info->smu_gpu_pll_output_freq =
+			(uint32_t) (le32_to_cpu(firmware_info->ulGPUPLL_OutputFreq) * 10);
+
+	return BP_RESULT_OK;
+}
+
+static enum bp_result get_ss_info_v3_1(
+	struct bios_parser *bp,
+	uint32_t id,
+	uint32_t index,
+	struct spread_spectrum_info *ss_info)
+{
+	ATOM_ASIC_INTERNAL_SS_INFO_V3 *ss_table_header_include;
+	ATOM_ASIC_SS_ASSIGNMENT_V3 *tbl;
+	uint32_t table_size;
+	uint32_t i;
+	uint32_t table_index = 0;
+
+	if (!ss_info)
+		return BP_RESULT_BADINPUT;
+
+	if (!DATA_TABLES(ASIC_InternalSS_Info))
+		return BP_RESULT_UNSUPPORTED;
+
+	ss_table_header_include = ((ATOM_ASIC_INTERNAL_SS_INFO_V3 *) bios_get_image(&bp->base,
+				DATA_TABLES(ASIC_InternalSS_Info),
+				struct_size(ss_table_header_include, asSpreadSpectrum, 1)));
+	if (!ss_table_header_include)
+		return BP_RESULT_UNSUPPORTED;
+
+	table_size =
+		(le16_to_cpu(ss_table_header_include->sHeader.usStructureSize)
+				- sizeof(ATOM_COMMON_TABLE_HEADER))
+				/ sizeof(ATOM_ASIC_SS_ASSIGNMENT_V3);
+
+	tbl = (ATOM_ASIC_SS_ASSIGNMENT_V3 *)
+				&ss_table_header_include->asSpreadSpectrum[0];
+
+	memset(ss_info, 0, sizeof(struct spread_spectrum_info));
+
+	for (i = 0; i < table_size; i++) {
+		if (tbl[i].ucClockIndication != (uint8_t) id)
+			continue;
+
+		if (table_index != index) {
+			table_index++;
+			continue;
+		}
+		/* VBIOS introduced new defines for Version 3, same values as
+		 *  before, so now use these new ones for Version 3.
+		 * Shouldn't affect field VBIOS's V3 as define values are still
+		 *  same.
+		 * #define SS_MODE_V3_CENTRE_SPREAD_MASK                0x01
+		 * #define SS_MODE_V3_EXTERNAL_SS_MASK                  0x02
+
+		 * Old VBIOS defines:
+		 * #define ATOM_SS_CENTRE_SPREAD_MODE_MASK        0x00000001
+		 * #define ATOM_EXTERNAL_SS_MASK                  0x00000002
+		 */
+
+		if (SS_MODE_V3_EXTERNAL_SS_MASK & tbl[i].ucSpreadSpectrumMode)
+			ss_info->type.EXTERNAL = true;
+
+		if (SS_MODE_V3_CENTRE_SPREAD_MASK & tbl[i].ucSpreadSpectrumMode)
+			ss_info->type.CENTER_MODE = true;
+
+		/* Older VBIOS (in field) always provides SS percentage in 0.01%
+		 * units set Divider to 100 */
+		ss_info->spread_percentage_divider = 100;
+
+		/* #define SS_MODE_V3_PERCENTAGE_DIV_BY_1000_MASK 0x10 */
+		if (SS_MODE_V3_PERCENTAGE_DIV_BY_1000_MASK
+				& tbl[i].ucSpreadSpectrumMode)
+			ss_info->spread_percentage_divider = 1000;
+
+		ss_info->type.STEP_AND_DELAY_INFO = false;
+		/* convert [10KHz] into [KHz] */
+		ss_info->target_clock_range =
+				le32_to_cpu(tbl[i].ulTargetClockRange) * 10;
+		ss_info->spread_spectrum_percentage =
+				(uint32_t)le16_to_cpu(tbl[i].usSpreadSpectrumPercentage);
+		ss_info->spread_spectrum_range =
+				(uint32_t)(le16_to_cpu(tbl[i].usSpreadRateIn10Hz) * 10);
+
+		return BP_RESULT_OK;
+	}
+	return BP_RESULT_NORECORD;
+}
+
+static enum bp_result bios_parser_transmitter_control(
+	struct dc_bios *dcb,
+	struct bp_transmitter_control *cntl)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	if (!bp->cmd_tbl.transmitter_control)
+		return BP_RESULT_FAILURE;
+
+	return bp->cmd_tbl.transmitter_control(bp, cntl);
+}
+
+static enum bp_result bios_parser_encoder_control(
+	struct dc_bios *dcb,
+	struct bp_encoder_control *cntl)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	if (!bp->cmd_tbl.dig_encoder_control)
+		return BP_RESULT_FAILURE;
+
+	return bp->cmd_tbl.dig_encoder_control(bp, cntl);
+}
+
+static enum bp_result bios_parser_adjust_pixel_clock(
+	struct dc_bios *dcb,
+	struct bp_adjust_pixel_clock_parameters *bp_params)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	if (!bp->cmd_tbl.adjust_display_pll)
+		return BP_RESULT_FAILURE;
+
+	return bp->cmd_tbl.adjust_display_pll(bp, bp_params);
+}
+
+static enum bp_result bios_parser_set_pixel_clock(
+	struct dc_bios *dcb,
+	struct bp_pixel_clock_parameters *bp_params)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	if (!bp->cmd_tbl.set_pixel_clock)
+		return BP_RESULT_FAILURE;
+
+	return bp->cmd_tbl.set_pixel_clock(bp, bp_params);
+}
+
+static enum bp_result bios_parser_set_dce_clock(
+	struct dc_bios *dcb,
+	struct bp_set_dce_clock_parameters *bp_params)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	if (!bp->cmd_tbl.set_dce_clock)
+		return BP_RESULT_FAILURE;
+
+	return bp->cmd_tbl.set_dce_clock(bp, bp_params);
+}
+
+static enum bp_result bios_parser_enable_spread_spectrum_on_ppll(
+	struct dc_bios *dcb,
+	struct bp_spread_spectrum_parameters *bp_params,
+	bool enable)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	if (!bp->cmd_tbl.enable_spread_spectrum_on_ppll)
+		return BP_RESULT_FAILURE;
+
+	return bp->cmd_tbl.enable_spread_spectrum_on_ppll(
+			bp, bp_params, enable);
+
+}
+
+static enum bp_result bios_parser_program_crtc_timing(
+	struct dc_bios *dcb,
+	struct bp_hw_crtc_timing_parameters *bp_params)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	if (!bp->cmd_tbl.set_crtc_timing)
+		return BP_RESULT_FAILURE;
+
+	return bp->cmd_tbl.set_crtc_timing(bp, bp_params);
+}
+
+static enum bp_result bios_parser_program_display_engine_pll(
+	struct dc_bios *dcb,
+	struct bp_pixel_clock_parameters *bp_params)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	if (!bp->cmd_tbl.program_clock)
+		return BP_RESULT_FAILURE;
+
+	return bp->cmd_tbl.program_clock(bp, bp_params);
+
+}
+
+
+static enum bp_result bios_parser_enable_crtc(
+	struct dc_bios *dcb,
+	enum controller_id id,
+	bool enable)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	if (!bp->cmd_tbl.enable_crtc)
+		return BP_RESULT_FAILURE;
+
+	return bp->cmd_tbl.enable_crtc(bp, id, enable);
+}
+
+static enum bp_result bios_parser_enable_disp_power_gating(
+	struct dc_bios *dcb,
+	enum controller_id controller_id,
+	enum bp_pipe_control_action action)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	if (!bp->cmd_tbl.enable_disp_power_gating)
+		return BP_RESULT_FAILURE;
+
+	return bp->cmd_tbl.enable_disp_power_gating(bp, controller_id,
+		action);
+}
+
+static bool bios_parser_is_device_id_supported(
+	struct dc_bios *dcb,
+	struct device_id id)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	uint32_t mask = get_support_mask_for_device_id(id);
+
+	return (le16_to_cpu(bp->object_info_tbl.v1_1->usDeviceSupport) & mask) != 0;
+}
+
+static ATOM_HPD_INT_RECORD *get_hpd_record(struct bios_parser *bp,
+	ATOM_OBJECT *object)
+{
+	ATOM_COMMON_RECORD_HEADER *header;
+	uint32_t offset;
+
+	if (!object) {
+		BREAK_TO_DEBUGGER(); /* Invalid object */
+		return NULL;
+	}
+
+	offset = le16_to_cpu(object->usRecordOffset)
+			+ bp->object_info_tbl_offset;
+
+	for (;;) {
+		header = GET_IMAGE(ATOM_COMMON_RECORD_HEADER, offset);
+
+		if (!header)
+			return NULL;
+
+		if (LAST_RECORD_TYPE == header->ucRecordType ||
+			!header->ucRecordSize)
+			break;
+
+		if (ATOM_HPD_INT_RECORD_TYPE == header->ucRecordType
+			&& sizeof(ATOM_HPD_INT_RECORD) <= header->ucRecordSize)
+			return (ATOM_HPD_INT_RECORD *) header;
+
+		offset += header->ucRecordSize;
+	}
+
+	return NULL;
+}
+
+static enum bp_result get_ss_info_from_ss_info_table(
+	struct bios_parser *bp,
+	uint32_t id,
+	struct spread_spectrum_info *ss_info);
+static enum bp_result get_ss_info_from_tbl(
+	struct bios_parser *bp,
+	uint32_t id,
+	struct spread_spectrum_info *ss_info);
+/**
+ * bios_parser_get_spread_spectrum_info
+ * Get spread spectrum information from the ASIC_InternalSS_Info(ver 2.1 or
+ * ver 3.1) or SS_Info table from the VBIOS. Currently ASIC_InternalSS_Info
+ * ver 2.1 can co-exist with SS_Info table. Expect ASIC_InternalSS_Info ver 3.1,
+ * there is only one entry for each signal /ss id.  However, there is
+ * no planning of supporting multiple spread Sprectum entry for EverGreen
+ * @dcb:     pointer to the DC BIOS
+ * @signal:  ASSignalType to be converted to info index
+ * @index:   number of entries that match the converted info index
+ * @ss_info: sprectrum information structure,
+ * return:   Bios parser result code
+ */
+static enum bp_result bios_parser_get_spread_spectrum_info(
+	struct dc_bios *dcb,
+	enum as_signal_type signal,
+	uint32_t index,
+	struct spread_spectrum_info *ss_info)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	enum bp_result result = BP_RESULT_UNSUPPORTED;
+	uint32_t clk_id_ss = 0;
+	ATOM_COMMON_TABLE_HEADER *header;
+	struct atom_data_revision tbl_revision;
+
+	if (!ss_info) /* check for bad input */
+		return BP_RESULT_BADINPUT;
+	/* signal translation */
+	clk_id_ss = signal_to_ss_id(signal);
+
+	if (!DATA_TABLES(ASIC_InternalSS_Info))
+		if (!index)
+			return get_ss_info_from_ss_info_table(bp, clk_id_ss,
+				ss_info);
+
+	header = GET_IMAGE(ATOM_COMMON_TABLE_HEADER,
+		DATA_TABLES(ASIC_InternalSS_Info));
+	get_atom_data_table_revision(header, &tbl_revision);
+
+	switch (tbl_revision.major) {
+	case 2:
+		switch (tbl_revision.minor) {
+		case 1:
+			/* there can not be more then one entry for Internal
+			 * SS Info table version 2.1 */
+			if (!index)
+				return get_ss_info_from_tbl(bp, clk_id_ss,
+						ss_info);
+			break;
+		default:
+			break;
+		}
+		break;
+
+	case 3:
+		switch (tbl_revision.minor) {
+		case 1:
+			return get_ss_info_v3_1(bp, clk_id_ss, index, ss_info);
+		default:
+			break;
+		}
+		break;
+	default:
+		break;
+	}
+	/* there can not be more then one entry for SS Info table */
+	return result;
+}
+
+static enum bp_result get_ss_info_from_internal_ss_info_tbl_V2_1(
+	struct bios_parser *bp,
+	uint32_t id,
+	struct spread_spectrum_info *info);
+
+/**
+ * get_ss_info_from_tbl
+ * Get spread sprectrum information from the ASIC_InternalSS_Info Ver 2.1 or
+ * SS_Info table from the VBIOS
+ * There can not be more than 1 entry for  ASIC_InternalSS_Info Ver 2.1 or
+ * SS_Info.
+ *
+ * @bp:      pointer to the BIOS parser
+ * @id:      spread sprectrum info index
+ * @ss_info: sprectrum information structure,
+ * return:   BIOS parser result code
+ */
+static enum bp_result get_ss_info_from_tbl(
+	struct bios_parser *bp,
+	uint32_t id,
+	struct spread_spectrum_info *ss_info)
+{
+	if (!ss_info) /* check for bad input, if ss_info is not NULL */
+		return BP_RESULT_BADINPUT;
+	/* for SS_Info table only support DP and LVDS */
+	if (id == ASIC_INTERNAL_SS_ON_DP || id == ASIC_INTERNAL_SS_ON_LVDS)
+		return get_ss_info_from_ss_info_table(bp, id, ss_info);
+	else
+		return get_ss_info_from_internal_ss_info_tbl_V2_1(bp, id,
+			ss_info);
+}
+
+/**
+ * get_ss_info_from_internal_ss_info_tbl_V2_1
+ * Get spread sprectrum information from the ASIC_InternalSS_Info table Ver 2.1
+ * from the VBIOS
+ * There will not be multiple entry for Ver 2.1
+ *
+ * @bp:    pointer to the Bios parser
+ * @id:    spread sprectrum info index
+ * @info:  sprectrum information structure,
+ * return: Bios parser result code
+ */
+static enum bp_result get_ss_info_from_internal_ss_info_tbl_V2_1(
+	struct bios_parser *bp,
+	uint32_t id,
+	struct spread_spectrum_info *info)
+{
+	enum bp_result result = BP_RESULT_UNSUPPORTED;
+	ATOM_ASIC_INTERNAL_SS_INFO_V2 *header;
+	ATOM_ASIC_SS_ASSIGNMENT_V2 *tbl;
+	uint32_t tbl_size, i;
+
+	if (!DATA_TABLES(ASIC_InternalSS_Info))
+		return result;
+
+	header = ((ATOM_ASIC_INTERNAL_SS_INFO_V2 *) bios_get_image(
+				&bp->base,
+				DATA_TABLES(ASIC_InternalSS_Info),
+				struct_size(header, asSpreadSpectrum, 1)));
+	if (!header)
+		return result;
+
+	memset(info, 0, sizeof(struct spread_spectrum_info));
+
+	tbl_size = (le16_to_cpu(header->sHeader.usStructureSize)
+			- sizeof(ATOM_COMMON_TABLE_HEADER))
+					/ sizeof(ATOM_ASIC_SS_ASSIGNMENT_V2);
+
+	tbl = (ATOM_ASIC_SS_ASSIGNMENT_V2 *)
+					&(header->asSpreadSpectrum[0]);
+	for (i = 0; i < tbl_size; i++) {
+		result = BP_RESULT_NORECORD;
+
+		if (tbl[i].ucClockIndication != (uint8_t)id)
+			continue;
+
+		if (ATOM_EXTERNAL_SS_MASK
+			& tbl[i].ucSpreadSpectrumMode) {
+			info->type.EXTERNAL = true;
+		}
+		if (ATOM_SS_CENTRE_SPREAD_MODE_MASK
+			& tbl[i].ucSpreadSpectrumMode) {
+			info->type.CENTER_MODE = true;
+		}
+		info->type.STEP_AND_DELAY_INFO = false;
+		/* convert [10KHz] into [KHz] */
+		info->target_clock_range =
+			le32_to_cpu(tbl[i].ulTargetClockRange) * 10;
+		info->spread_spectrum_percentage =
+			(uint32_t)le16_to_cpu(tbl[i].usSpreadSpectrumPercentage);
+		info->spread_spectrum_range =
+			(uint32_t)(le16_to_cpu(tbl[i].usSpreadRateIn10Hz) * 10);
+		result = BP_RESULT_OK;
+		break;
+	}
+
+	return result;
+
+}
+
+/**
+ * get_ss_info_from_ss_info_table
+ * Get spread sprectrum information from the SS_Info table from the VBIOS
+ * if the pointer to info is NULL, indicate the caller what to know the number
+ * of entries that matches the id
+ * for, the SS_Info table, there should not be more than 1 entry match.
+ *
+ * @bp:      pointer to the Bios parser
+ * @id:      spread sprectrum id
+ * @ss_info: sprectrum information structure,
+ * return:   Bios parser result code
+ */
+static enum bp_result get_ss_info_from_ss_info_table(
+	struct bios_parser *bp,
+	uint32_t id,
+	struct spread_spectrum_info *ss_info)
+{
+	enum bp_result result = BP_RESULT_UNSUPPORTED;
+	ATOM_SPREAD_SPECTRUM_INFO *tbl;
+	ATOM_COMMON_TABLE_HEADER *header;
+	uint32_t table_size;
+	uint32_t i;
+	uint32_t id_local = SS_ID_UNKNOWN;
+	struct atom_data_revision revision;
+
+	/* exist of the SS_Info table */
+	/* check for bad input, pSSinfo can not be NULL */
+	if (!DATA_TABLES(SS_Info) || !ss_info)
+		return result;
+
+	header = GET_IMAGE(ATOM_COMMON_TABLE_HEADER, DATA_TABLES(SS_Info));
+	get_atom_data_table_revision(header, &revision);
+
+	tbl = GET_IMAGE(ATOM_SPREAD_SPECTRUM_INFO, DATA_TABLES(SS_Info));
+	if (!tbl)
+		return result;
+
+	if (1 != revision.major || 2 > revision.minor)
+		return result;
+
+	/* have to convert from Internal_SS format to SS_Info format */
+	switch (id) {
+	case ASIC_INTERNAL_SS_ON_DP:
+		id_local = SS_ID_DP1;
+		break;
+	case ASIC_INTERNAL_SS_ON_LVDS:
+	{
+		struct embedded_panel_info panel_info;
+
+		if (bios_parser_get_embedded_panel_info(&bp->base, &panel_info)
+				== BP_RESULT_OK)
+			id_local = panel_info.ss_id;
+		break;
+	}
+	default:
+		break;
+	}
+
+	if (id_local == SS_ID_UNKNOWN)
+		return result;
+
+	table_size = (le16_to_cpu(tbl->sHeader.usStructureSize) -
+			sizeof(ATOM_COMMON_TABLE_HEADER)) /
+					sizeof(ATOM_SPREAD_SPECTRUM_ASSIGNMENT);
+
+	for (i = 0; i < table_size; i++) {
+		if (id_local != (uint32_t)tbl->asSS_Info[i].ucSS_Id)
+			continue;
+
+		memset(ss_info, 0, sizeof(struct spread_spectrum_info));
+
+		if (ATOM_EXTERNAL_SS_MASK &
+				tbl->asSS_Info[i].ucSpreadSpectrumType)
+			ss_info->type.EXTERNAL = true;
+
+		if (ATOM_SS_CENTRE_SPREAD_MODE_MASK &
+				tbl->asSS_Info[i].ucSpreadSpectrumType)
+			ss_info->type.CENTER_MODE = true;
+
+		ss_info->type.STEP_AND_DELAY_INFO = true;
+		ss_info->spread_spectrum_percentage =
+			(uint32_t)le16_to_cpu(tbl->asSS_Info[i].usSpreadSpectrumPercentage);
+		ss_info->step_and_delay_info.step = tbl->asSS_Info[i].ucSS_Step;
+		ss_info->step_and_delay_info.delay =
+			tbl->asSS_Info[i].ucSS_Delay;
+		ss_info->step_and_delay_info.recommended_ref_div =
+			tbl->asSS_Info[i].ucRecommendedRef_Div;
+		ss_info->spread_spectrum_range =
+			(uint32_t)tbl->asSS_Info[i].ucSS_Range * 10000;
+
+		/* there will be only one entry for each display type in SS_info
+		 * table */
+		result = BP_RESULT_OK;
+		break;
+	}
+
+	return result;
+}
+static enum bp_result get_embedded_panel_info_v1_2(
+	struct bios_parser *bp,
+	struct embedded_panel_info *info);
+static enum bp_result get_embedded_panel_info_v1_3(
+	struct bios_parser *bp,
+	struct embedded_panel_info *info);
+
+static enum bp_result bios_parser_get_embedded_panel_info(
+	struct dc_bios *dcb,
+	struct embedded_panel_info *info)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	ATOM_COMMON_TABLE_HEADER *hdr;
+
+	if (!DATA_TABLES(LCD_Info))
+		return BP_RESULT_FAILURE;
+
+	hdr = GET_IMAGE(ATOM_COMMON_TABLE_HEADER, DATA_TABLES(LCD_Info));
+
+	if (!hdr)
+		return BP_RESULT_BADBIOSTABLE;
+
+	switch (hdr->ucTableFormatRevision) {
+	case 1:
+		switch (hdr->ucTableContentRevision) {
+		case 0:
+		case 1:
+		case 2:
+			return get_embedded_panel_info_v1_2(bp, info);
+		case 3:
+			return get_embedded_panel_info_v1_3(bp, info);
+		default:
+			break;
+		}
+		break;
+	default:
+		break;
+	}
+
+	return BP_RESULT_FAILURE;
+}
+
+static enum bp_result get_embedded_panel_info_v1_2(
+	struct bios_parser *bp,
+	struct embedded_panel_info *info)
+{
+	ATOM_LVDS_INFO_V12 *lvds;
+
+	if (!info)
+		return BP_RESULT_BADINPUT;
+
+	if (!DATA_TABLES(LVDS_Info))
+		return BP_RESULT_UNSUPPORTED;
+
+	lvds =
+		GET_IMAGE(ATOM_LVDS_INFO_V12, DATA_TABLES(LVDS_Info));
+
+	if (!lvds)
+		return BP_RESULT_BADBIOSTABLE;
+
+	if (1 != lvds->sHeader.ucTableFormatRevision
+		|| 2 > lvds->sHeader.ucTableContentRevision)
+		return BP_RESULT_UNSUPPORTED;
+
+	memset(info, 0, sizeof(struct embedded_panel_info));
+
+	/* We need to convert from 10KHz units into KHz units*/
+	info->lcd_timing.pixel_clk =
+		le16_to_cpu(lvds->sLCDTiming.usPixClk) * 10;
+	/* usHActive does not include borders, according to VBIOS team*/
+	info->lcd_timing.horizontal_addressable =
+		le16_to_cpu(lvds->sLCDTiming.usHActive);
+	/* usHBlanking_Time includes borders, so we should really be subtracting
+	 * borders duing this translation, but LVDS generally*/
+	/* doesn't have borders, so we should be okay leaving this as is for
+	 * now.  May need to revisit if we ever have LVDS with borders*/
+	info->lcd_timing.horizontal_blanking_time =
+			le16_to_cpu(lvds->sLCDTiming.usHBlanking_Time);
+	/* usVActive does not include borders, according to VBIOS team*/
+	info->lcd_timing.vertical_addressable =
+			le16_to_cpu(lvds->sLCDTiming.usVActive);
+	/* usVBlanking_Time includes borders, so we should really be subtracting
+	 * borders duing this translation, but LVDS generally*/
+	/* doesn't have borders, so we should be okay leaving this as is for
+	 * now. May need to revisit if we ever have LVDS with borders*/
+	info->lcd_timing.vertical_blanking_time =
+		le16_to_cpu(lvds->sLCDTiming.usVBlanking_Time);
+	info->lcd_timing.horizontal_sync_offset =
+		le16_to_cpu(lvds->sLCDTiming.usHSyncOffset);
+	info->lcd_timing.horizontal_sync_width =
+		le16_to_cpu(lvds->sLCDTiming.usHSyncWidth);
+	info->lcd_timing.vertical_sync_offset =
+		le16_to_cpu(lvds->sLCDTiming.usVSyncOffset);
+	info->lcd_timing.vertical_sync_width =
+		le16_to_cpu(lvds->sLCDTiming.usVSyncWidth);
+	info->lcd_timing.horizontal_border = lvds->sLCDTiming.ucHBorder;
+	info->lcd_timing.vertical_border = lvds->sLCDTiming.ucVBorder;
+	info->lcd_timing.misc_info.HORIZONTAL_CUT_OFF =
+		lvds->sLCDTiming.susModeMiscInfo.sbfAccess.HorizontalCutOff;
+	info->lcd_timing.misc_info.H_SYNC_POLARITY =
+		~(uint32_t)
+		lvds->sLCDTiming.susModeMiscInfo.sbfAccess.HSyncPolarity;
+	info->lcd_timing.misc_info.V_SYNC_POLARITY =
+		~(uint32_t)
+		lvds->sLCDTiming.susModeMiscInfo.sbfAccess.VSyncPolarity;
+	info->lcd_timing.misc_info.VERTICAL_CUT_OFF =
+		lvds->sLCDTiming.susModeMiscInfo.sbfAccess.VerticalCutOff;
+	info->lcd_timing.misc_info.H_REPLICATION_BY2 =
+		lvds->sLCDTiming.susModeMiscInfo.sbfAccess.H_ReplicationBy2;
+	info->lcd_timing.misc_info.V_REPLICATION_BY2 =
+		lvds->sLCDTiming.susModeMiscInfo.sbfAccess.V_ReplicationBy2;
+	info->lcd_timing.misc_info.COMPOSITE_SYNC =
+		lvds->sLCDTiming.susModeMiscInfo.sbfAccess.CompositeSync;
+	info->lcd_timing.misc_info.INTERLACE =
+		lvds->sLCDTiming.susModeMiscInfo.sbfAccess.Interlace;
+	info->lcd_timing.misc_info.DOUBLE_CLOCK =
+		lvds->sLCDTiming.susModeMiscInfo.sbfAccess.DoubleClock;
+	info->ss_id = lvds->ucSS_Id;
+
+	{
+		uint8_t rr = le16_to_cpu(lvds->usSupportedRefreshRate);
+		/* Get minimum supported refresh rate*/
+		if (SUPPORTED_LCD_REFRESHRATE_30Hz & rr)
+			info->supported_rr.REFRESH_RATE_30HZ = 1;
+		else if (SUPPORTED_LCD_REFRESHRATE_40Hz & rr)
+			info->supported_rr.REFRESH_RATE_40HZ = 1;
+		else if (SUPPORTED_LCD_REFRESHRATE_48Hz & rr)
+			info->supported_rr.REFRESH_RATE_48HZ = 1;
+		else if (SUPPORTED_LCD_REFRESHRATE_50Hz & rr)
+			info->supported_rr.REFRESH_RATE_50HZ = 1;
+		else if (SUPPORTED_LCD_REFRESHRATE_60Hz & rr)
+			info->supported_rr.REFRESH_RATE_60HZ = 1;
+	}
+
+	/*Drr panel support can be reported by VBIOS*/
+	if (LCDPANEL_CAP_DRR_SUPPORTED
+			& lvds->ucLCDPanel_SpecialHandlingCap)
+		info->drr_enabled = 1;
+
+	if (ATOM_PANEL_MISC_DUAL & lvds->ucLVDS_Misc)
+		info->lcd_timing.misc_info.DOUBLE_CLOCK = true;
+
+	if (ATOM_PANEL_MISC_888RGB & lvds->ucLVDS_Misc)
+		info->lcd_timing.misc_info.RGB888 = true;
+
+	info->lcd_timing.misc_info.GREY_LEVEL =
+		(uint32_t) (ATOM_PANEL_MISC_GREY_LEVEL &
+			lvds->ucLVDS_Misc) >> ATOM_PANEL_MISC_GREY_LEVEL_SHIFT;
+
+	if (ATOM_PANEL_MISC_SPATIAL & lvds->ucLVDS_Misc)
+		info->lcd_timing.misc_info.SPATIAL = true;
+
+	if (ATOM_PANEL_MISC_TEMPORAL & lvds->ucLVDS_Misc)
+		info->lcd_timing.misc_info.TEMPORAL = true;
+
+	if (ATOM_PANEL_MISC_API_ENABLED & lvds->ucLVDS_Misc)
+		info->lcd_timing.misc_info.API_ENABLED = true;
+
+	return BP_RESULT_OK;
+}
+
+static enum bp_result get_embedded_panel_info_v1_3(
+	struct bios_parser *bp,
+	struct embedded_panel_info *info)
+{
+	ATOM_LCD_INFO_V13 *lvds;
+
+	if (!info)
+		return BP_RESULT_BADINPUT;
+
+	if (!DATA_TABLES(LCD_Info))
+		return BP_RESULT_UNSUPPORTED;
+
+	lvds = GET_IMAGE(ATOM_LCD_INFO_V13, DATA_TABLES(LCD_Info));
+
+	if (!lvds)
+		return BP_RESULT_BADBIOSTABLE;
+
+	if (!((1 == lvds->sHeader.ucTableFormatRevision)
+			&& (3 <= lvds->sHeader.ucTableContentRevision)))
+		return BP_RESULT_UNSUPPORTED;
+
+	memset(info, 0, sizeof(struct embedded_panel_info));
+
+	/* We need to convert from 10KHz units into KHz units */
+	info->lcd_timing.pixel_clk =
+			le16_to_cpu(lvds->sLCDTiming.usPixClk) * 10;
+	/* usHActive does not include borders, according to VBIOS team */
+	info->lcd_timing.horizontal_addressable =
+			le16_to_cpu(lvds->sLCDTiming.usHActive);
+	/* usHBlanking_Time includes borders, so we should really be subtracting
+	 * borders duing this translation, but LVDS generally*/
+	/* doesn't have borders, so we should be okay leaving this as is for
+	 * now.  May need to revisit if we ever have LVDS with borders*/
+	info->lcd_timing.horizontal_blanking_time =
+		le16_to_cpu(lvds->sLCDTiming.usHBlanking_Time);
+	/* usVActive does not include borders, according to VBIOS team*/
+	info->lcd_timing.vertical_addressable =
+		le16_to_cpu(lvds->sLCDTiming.usVActive);
+	/* usVBlanking_Time includes borders, so we should really be subtracting
+	 * borders duing this translation, but LVDS generally*/
+	/* doesn't have borders, so we should be okay leaving this as is for
+	 * now. May need to revisit if we ever have LVDS with borders*/
+	info->lcd_timing.vertical_blanking_time =
+		le16_to_cpu(lvds->sLCDTiming.usVBlanking_Time);
+	info->lcd_timing.horizontal_sync_offset =
+		le16_to_cpu(lvds->sLCDTiming.usHSyncOffset);
+	info->lcd_timing.horizontal_sync_width =
+		le16_to_cpu(lvds->sLCDTiming.usHSyncWidth);
+	info->lcd_timing.vertical_sync_offset =
+		le16_to_cpu(lvds->sLCDTiming.usVSyncOffset);
+	info->lcd_timing.vertical_sync_width =
+		le16_to_cpu(lvds->sLCDTiming.usVSyncWidth);
+	info->lcd_timing.horizontal_border = lvds->sLCDTiming.ucHBorder;
+	info->lcd_timing.vertical_border = lvds->sLCDTiming.ucVBorder;
+	info->lcd_timing.misc_info.HORIZONTAL_CUT_OFF =
+		lvds->sLCDTiming.susModeMiscInfo.sbfAccess.HorizontalCutOff;
+	info->lcd_timing.misc_info.H_SYNC_POLARITY =
+		~(uint32_t)
+		lvds->sLCDTiming.susModeMiscInfo.sbfAccess.HSyncPolarity;
+	info->lcd_timing.misc_info.V_SYNC_POLARITY =
+		~(uint32_t)
+		lvds->sLCDTiming.susModeMiscInfo.sbfAccess.VSyncPolarity;
+	info->lcd_timing.misc_info.VERTICAL_CUT_OFF =
+		lvds->sLCDTiming.susModeMiscInfo.sbfAccess.VerticalCutOff;
+	info->lcd_timing.misc_info.H_REPLICATION_BY2 =
+		lvds->sLCDTiming.susModeMiscInfo.sbfAccess.H_ReplicationBy2;
+	info->lcd_timing.misc_info.V_REPLICATION_BY2 =
+		lvds->sLCDTiming.susModeMiscInfo.sbfAccess.V_ReplicationBy2;
+	info->lcd_timing.misc_info.COMPOSITE_SYNC =
+		lvds->sLCDTiming.susModeMiscInfo.sbfAccess.CompositeSync;
+	info->lcd_timing.misc_info.INTERLACE =
+		lvds->sLCDTiming.susModeMiscInfo.sbfAccess.Interlace;
+	info->lcd_timing.misc_info.DOUBLE_CLOCK =
+		lvds->sLCDTiming.susModeMiscInfo.sbfAccess.DoubleClock;
+	info->ss_id = lvds->ucSS_Id;
+
+	/* Drr panel support can be reported by VBIOS*/
+	if (LCDPANEL_CAP_V13_DRR_SUPPORTED
+			& lvds->ucLCDPanel_SpecialHandlingCap)
+		info->drr_enabled = 1;
+
+	/* Get supported refresh rate*/
+	if (info->drr_enabled == 1) {
+		uint8_t min_rr =
+				lvds->sRefreshRateSupport.ucMinRefreshRateForDRR;
+		uint8_t rr = lvds->sRefreshRateSupport.ucSupportedRefreshRate;
+
+		if (min_rr != 0) {
+			if (SUPPORTED_LCD_REFRESHRATE_30Hz & min_rr)
+				info->supported_rr.REFRESH_RATE_30HZ = 1;
+			else if (SUPPORTED_LCD_REFRESHRATE_40Hz & min_rr)
+				info->supported_rr.REFRESH_RATE_40HZ = 1;
+			else if (SUPPORTED_LCD_REFRESHRATE_48Hz & min_rr)
+				info->supported_rr.REFRESH_RATE_48HZ = 1;
+			else if (SUPPORTED_LCD_REFRESHRATE_50Hz & min_rr)
+				info->supported_rr.REFRESH_RATE_50HZ = 1;
+			else if (SUPPORTED_LCD_REFRESHRATE_60Hz & min_rr)
+				info->supported_rr.REFRESH_RATE_60HZ = 1;
+		} else {
+			if (SUPPORTED_LCD_REFRESHRATE_30Hz & rr)
+				info->supported_rr.REFRESH_RATE_30HZ = 1;
+			else if (SUPPORTED_LCD_REFRESHRATE_40Hz & rr)
+				info->supported_rr.REFRESH_RATE_40HZ = 1;
+			else if (SUPPORTED_LCD_REFRESHRATE_48Hz & rr)
+				info->supported_rr.REFRESH_RATE_48HZ = 1;
+			else if (SUPPORTED_LCD_REFRESHRATE_50Hz & rr)
+				info->supported_rr.REFRESH_RATE_50HZ = 1;
+			else if (SUPPORTED_LCD_REFRESHRATE_60Hz & rr)
+				info->supported_rr.REFRESH_RATE_60HZ = 1;
+		}
+	}
+
+	if (ATOM_PANEL_MISC_V13_DUAL & lvds->ucLCD_Misc)
+		info->lcd_timing.misc_info.DOUBLE_CLOCK = true;
+
+	if (ATOM_PANEL_MISC_V13_8BIT_PER_COLOR & lvds->ucLCD_Misc)
+		info->lcd_timing.misc_info.RGB888 = true;
+
+	info->lcd_timing.misc_info.GREY_LEVEL =
+			(uint32_t) (ATOM_PANEL_MISC_V13_GREY_LEVEL &
+				lvds->ucLCD_Misc) >> ATOM_PANEL_MISC_V13_GREY_LEVEL_SHIFT;
+
+	return BP_RESULT_OK;
+}
+
+/**
+ * bios_parser_get_encoder_cap_info - get encoder capability
+ *                                    information of input object id
+ *
+ * @dcb:       pointer to the DC BIOS
+ * @object_id: object id
+ * @info:      encoder cap information structure
+ *
+ * return: Bios parser result code
+ */
+static enum bp_result bios_parser_get_encoder_cap_info(
+	struct dc_bios *dcb,
+	struct graphics_object_id object_id,
+	struct bp_encoder_cap_info *info)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	ATOM_OBJECT *object;
+	ATOM_ENCODER_CAP_RECORD_V2 *record = NULL;
+
+	if (!info)
+		return BP_RESULT_BADINPUT;
+
+	object = get_bios_object(bp, object_id);
+
+	if (!object)
+		return BP_RESULT_BADINPUT;
+
+	record = get_encoder_cap_record(bp, object);
+	if (!record)
+		return BP_RESULT_NORECORD;
+
+	info->DP_HBR2_EN = record->usHBR2En;
+	info->DP_HBR3_EN = record->usHBR3En;
+	info->HDMI_6GB_EN = record->usHDMI6GEn;
+	return BP_RESULT_OK;
+}
+
+/**
+ * get_encoder_cap_record - Get encoder cap record for the object
+ *
+ * @bp:      pointer to the BIOS parser
+ * @object:  ATOM object
+ * return:   atom encoder cap record
+ * note:     search all records to find the ATOM_ENCODER_CAP_RECORD_V2 record
+ */
+static ATOM_ENCODER_CAP_RECORD_V2 *get_encoder_cap_record(
+	struct bios_parser *bp,
+	ATOM_OBJECT *object)
+{
+	ATOM_COMMON_RECORD_HEADER *header;
+	uint32_t offset;
+
+	if (!object) {
+		BREAK_TO_DEBUGGER(); /* Invalid object */
+		return NULL;
+	}
+
+	offset = le16_to_cpu(object->usRecordOffset)
+					+ bp->object_info_tbl_offset;
+
+	for (;;) {
+		header = GET_IMAGE(ATOM_COMMON_RECORD_HEADER, offset);
+
+		if (!header)
+			return NULL;
+
+		offset += header->ucRecordSize;
+
+		if (LAST_RECORD_TYPE == header->ucRecordType ||
+				!header->ucRecordSize)
+			break;
+
+		if (ATOM_ENCODER_CAP_RECORD_TYPE != header->ucRecordType)
+			continue;
+
+		if (sizeof(ATOM_ENCODER_CAP_RECORD_V2) <= header->ucRecordSize)
+			return (ATOM_ENCODER_CAP_RECORD_V2 *)header;
+	}
+
+	return NULL;
+}
+
+static uint32_t get_ss_entry_number(
+	struct bios_parser *bp,
+	uint32_t id);
+static uint32_t get_ss_entry_number_from_internal_ss_info_tbl_v2_1(
+	struct bios_parser *bp,
+	uint32_t id);
+static uint32_t get_ss_entry_number_from_internal_ss_info_tbl_V3_1(
+	struct bios_parser *bp,
+	uint32_t id);
+static uint32_t get_ss_entry_number_from_ss_info_tbl(
+	struct bios_parser *bp,
+	uint32_t id);
+
+/**
+ * bios_parser_get_ss_entry_number
+ * Get Number of SpreadSpectrum Entry from the ASIC_InternalSS_Info table from
+ * the VBIOS that match the SSid (to be converted from signal)
+ *
+ * @dcb:    pointer to the DC BIOS
+ * @signal: ASSignalType to be converted to SSid
+ * return: number of SS Entry that match the signal
+ */
+static uint32_t bios_parser_get_ss_entry_number(
+	struct dc_bios *dcb,
+	enum as_signal_type signal)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	uint32_t ss_id = 0;
+	ATOM_COMMON_TABLE_HEADER *header;
+	struct atom_data_revision revision;
+
+	ss_id = signal_to_ss_id(signal);
+
+	if (!DATA_TABLES(ASIC_InternalSS_Info))
+		return get_ss_entry_number_from_ss_info_tbl(bp, ss_id);
+
+	header = GET_IMAGE(ATOM_COMMON_TABLE_HEADER,
+			DATA_TABLES(ASIC_InternalSS_Info));
+	get_atom_data_table_revision(header, &revision);
+
+	switch (revision.major) {
+	case 2:
+		switch (revision.minor) {
+		case 1:
+			return get_ss_entry_number(bp, ss_id);
+		default:
+			break;
+		}
+		break;
+	case 3:
+		switch (revision.minor) {
+		case 1:
+			return
+				get_ss_entry_number_from_internal_ss_info_tbl_V3_1(
+						bp, ss_id);
+		default:
+			break;
+		}
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+/**
+ * get_ss_entry_number_from_ss_info_tbl
+ * Get Number of spread spectrum entry from the SS_Info table from the VBIOS.
+ *
+ * @bp:  pointer to the BIOS parser
+ * @id:  spread spectrum id
+ * return: number of SS Entry that match the id
+ * note: There can only be one entry for each id for SS_Info Table
+ */
+static uint32_t get_ss_entry_number_from_ss_info_tbl(
+	struct bios_parser *bp,
+	uint32_t id)
+{
+	ATOM_SPREAD_SPECTRUM_INFO *tbl;
+	ATOM_COMMON_TABLE_HEADER *header;
+	uint32_t table_size;
+	uint32_t i;
+	uint32_t number = 0;
+	uint32_t id_local = SS_ID_UNKNOWN;
+	struct atom_data_revision revision;
+
+	/* SS_Info table exist */
+	if (!DATA_TABLES(SS_Info))
+		return number;
+
+	header = GET_IMAGE(ATOM_COMMON_TABLE_HEADER,
+			DATA_TABLES(SS_Info));
+	get_atom_data_table_revision(header, &revision);
+
+	tbl = GET_IMAGE(ATOM_SPREAD_SPECTRUM_INFO,
+			DATA_TABLES(SS_Info));
+	if (!tbl)
+		return number;
+
+	if (1 != revision.major || 2 > revision.minor)
+		return number;
+
+	/* have to convert from Internal_SS format to SS_Info format */
+	switch (id) {
+	case ASIC_INTERNAL_SS_ON_DP:
+		id_local = SS_ID_DP1;
+		break;
+	case ASIC_INTERNAL_SS_ON_LVDS: {
+		struct embedded_panel_info panel_info;
+
+		if (bios_parser_get_embedded_panel_info(&bp->base, &panel_info)
+				== BP_RESULT_OK)
+			id_local = panel_info.ss_id;
+		break;
+	}
+	default:
+		break;
+	}
+
+	if (id_local == SS_ID_UNKNOWN)
+		return number;
+
+	table_size = (le16_to_cpu(tbl->sHeader.usStructureSize) -
+			sizeof(ATOM_COMMON_TABLE_HEADER)) /
+					sizeof(ATOM_SPREAD_SPECTRUM_ASSIGNMENT);
+
+	for (i = 0; i < table_size; i++)
+		if (id_local == (uint32_t)tbl->asSS_Info[i].ucSS_Id) {
+			number = 1;
+			break;
+		}
+
+	return number;
+}
+
+/**
+ * get_ss_entry_number
+ * Get spread sprectrum information from the ASIC_InternalSS_Info Ver 2.1 or
+ * SS_Info table from the VBIOS
+ * There can not be more than 1 entry for  ASIC_InternalSS_Info Ver 2.1 or
+ * SS_Info.
+ *
+ * @bp:    pointer to the BIOS parser
+ * @id:    spread sprectrum info index
+ * return: Bios parser result code
+ */
+static uint32_t get_ss_entry_number(struct bios_parser *bp, uint32_t id)
+{
+	if (id == ASIC_INTERNAL_SS_ON_DP || id == ASIC_INTERNAL_SS_ON_LVDS)
+		return get_ss_entry_number_from_ss_info_tbl(bp, id);
+
+	return get_ss_entry_number_from_internal_ss_info_tbl_v2_1(bp, id);
+}
+
+/**
+ * get_ss_entry_number_from_internal_ss_info_tbl_v2_1
+ * Get NUmber of spread sprectrum entry from the ASIC_InternalSS_Info table
+ * Ver 2.1 from the VBIOS
+ * There will not be multiple entry for Ver 2.1
+ *
+ * @bp:    pointer to the BIOS parser
+ * @id:    spread sprectrum info index
+ * return: number of SS Entry that match the id
+ */
+static uint32_t get_ss_entry_number_from_internal_ss_info_tbl_v2_1(
+	struct bios_parser *bp,
+	uint32_t id)
+{
+	ATOM_ASIC_INTERNAL_SS_INFO_V2 *header_include;
+	ATOM_ASIC_SS_ASSIGNMENT_V2 *tbl;
+	uint32_t size;
+	uint32_t i;
+
+	if (!DATA_TABLES(ASIC_InternalSS_Info))
+		return 0;
+
+	header_include = ((ATOM_ASIC_INTERNAL_SS_INFO_V2 *) bios_get_image(
+				&bp->base,
+				DATA_TABLES(ASIC_InternalSS_Info),
+				struct_size(header_include, asSpreadSpectrum, 1)));
+	if (!header_include)
+		return 0;
+
+	size = (le16_to_cpu(header_include->sHeader.usStructureSize)
+			- sizeof(ATOM_COMMON_TABLE_HEADER))
+						/ sizeof(ATOM_ASIC_SS_ASSIGNMENT_V2);
+
+	tbl = (ATOM_ASIC_SS_ASSIGNMENT_V2 *)
+				&header_include->asSpreadSpectrum[0];
+	for (i = 0; i < size; i++)
+		if (tbl[i].ucClockIndication == (uint8_t)id)
+			return 1;
+
+	return 0;
+}
+
+/**
+ * get_ss_entry_number_from_internal_ss_info_tbl_V3_1
+ * Get Number of SpreadSpectrum Entry from the ASIC_InternalSS_Info table of
+ * the VBIOS that matches id
+ *
+ * @bp:    pointer to the BIOS parser
+ * @id:    spread sprectrum id
+ * return: number of SS Entry that match the id
+ */
+static uint32_t get_ss_entry_number_from_internal_ss_info_tbl_V3_1(
+	struct bios_parser *bp,
+	uint32_t id)
+{
+	uint32_t number = 0;
+	ATOM_ASIC_INTERNAL_SS_INFO_V3 *header_include;
+	ATOM_ASIC_SS_ASSIGNMENT_V3 *tbl;
+	uint32_t size;
+	uint32_t i;
+
+	if (!DATA_TABLES(ASIC_InternalSS_Info))
+		return number;
+
+	header_include = ((ATOM_ASIC_INTERNAL_SS_INFO_V3 *) bios_get_image(&bp->base,
+				DATA_TABLES(ASIC_InternalSS_Info),
+				struct_size(header_include, asSpreadSpectrum, 1)));
+	if (!header_include)
+		return number;
+
+	size = (le16_to_cpu(header_include->sHeader.usStructureSize) -
+			sizeof(ATOM_COMMON_TABLE_HEADER)) /
+					sizeof(ATOM_ASIC_SS_ASSIGNMENT_V3);
+
+	tbl = (ATOM_ASIC_SS_ASSIGNMENT_V3 *)
+				&header_include->asSpreadSpectrum[0];
+
+	for (i = 0; i < size; i++)
+		if (tbl[i].ucClockIndication == (uint8_t)id)
+			number++;
+
+	return number;
+}
+
+/**
+ * bios_parser_get_gpio_pin_info
+ * Get GpioPin information of input gpio id
+ *
+ * @dcb:     pointer to the DC BIOS
+ * @gpio_id: GPIO ID
+ * @info:    GpioPin information structure
+ * return:   Bios parser result code
+ * note:
+ *  to get the GPIO PIN INFO, we need:
+ *  1. get the GPIO_ID from other object table, see GetHPDInfo()
+ *  2. in DATA_TABLE.GPIO_Pin_LUT, search all records, to get the registerA
+ *  offset/mask
+ */
+static enum bp_result bios_parser_get_gpio_pin_info(
+	struct dc_bios *dcb,
+	uint32_t gpio_id,
+	struct gpio_pin_info *info)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	ATOM_GPIO_PIN_LUT *header;
+	uint32_t count = 0;
+	uint32_t i = 0;
+
+	if (!DATA_TABLES(GPIO_Pin_LUT))
+		return BP_RESULT_BADBIOSTABLE;
+
+	header = ((ATOM_GPIO_PIN_LUT *) bios_get_image(&bp->base,
+				DATA_TABLES(GPIO_Pin_LUT),
+				struct_size(header, asGPIO_Pin, 1)));
+	if (!header)
+		return BP_RESULT_BADBIOSTABLE;
+
+	if (sizeof(ATOM_COMMON_TABLE_HEADER) + struct_size(header, asGPIO_Pin, 1)
+			> le16_to_cpu(header->sHeader.usStructureSize))
+		return BP_RESULT_BADBIOSTABLE;
+
+	if (1 != header->sHeader.ucTableContentRevision)
+		return BP_RESULT_UNSUPPORTED;
+
+	count = (le16_to_cpu(header->sHeader.usStructureSize)
+			- sizeof(ATOM_COMMON_TABLE_HEADER))
+				/ sizeof(ATOM_GPIO_PIN_ASSIGNMENT);
+	for (i = 0; i < count; ++i) {
+		if (header->asGPIO_Pin[i].ucGPIO_ID != gpio_id)
+			continue;
+
+		info->offset =
+			(uint32_t) le16_to_cpu(header->asGPIO_Pin[i].usGpioPin_AIndex);
+		info->offset_y = info->offset + 2;
+		info->offset_en = info->offset + 1;
+		info->offset_mask = info->offset - 1;
+
+		info->mask = (uint32_t) (1 <<
+			header->asGPIO_Pin[i].ucGpioPinBitShift);
+		info->mask_y = info->mask + 2;
+		info->mask_en = info->mask + 1;
+		info->mask_mask = info->mask - 1;
+
+		return BP_RESULT_OK;
+	}
+
+	return BP_RESULT_NORECORD;
+}
+
+static enum bp_result get_gpio_i2c_info(struct bios_parser *bp,
+	ATOM_I2C_RECORD *record,
+	struct graphics_object_i2c_info *info)
+{
+	ATOM_GPIO_I2C_INFO *header;
+	uint32_t count = 0;
+
+	if (!info)
+		return BP_RESULT_BADINPUT;
+
+	/* get the GPIO_I2C info */
+	if (!DATA_TABLES(GPIO_I2C_Info))
+		return BP_RESULT_BADBIOSTABLE;
+
+	header = GET_IMAGE(ATOM_GPIO_I2C_INFO, DATA_TABLES(GPIO_I2C_Info));
+	if (!header)
+		return BP_RESULT_BADBIOSTABLE;
+
+	if (sizeof(ATOM_COMMON_TABLE_HEADER) + sizeof(ATOM_GPIO_I2C_ASSIGMENT)
+			> le16_to_cpu(header->sHeader.usStructureSize))
+		return BP_RESULT_BADBIOSTABLE;
+
+	if (1 != header->sHeader.ucTableContentRevision)
+		return BP_RESULT_UNSUPPORTED;
+
+	/* get data count */
+	count = (le16_to_cpu(header->sHeader.usStructureSize)
+			- sizeof(ATOM_COMMON_TABLE_HEADER))
+				/ sizeof(ATOM_GPIO_I2C_ASSIGMENT);
+	if (count < record->sucI2cId.bfI2C_LineMux)
+		return BP_RESULT_BADBIOSTABLE;
+
+	/* get the GPIO_I2C_INFO */
+	info->i2c_hw_assist = record->sucI2cId.bfHW_Capable;
+	info->i2c_line = record->sucI2cId.bfI2C_LineMux;
+	info->i2c_engine_id = record->sucI2cId.bfHW_EngineID;
+	info->i2c_slave_address = record->ucI2CAddr;
+
+	info->gpio_info.clk_mask_register_index =
+			le16_to_cpu(header->asGPIO_Info[info->i2c_line].usClkMaskRegisterIndex);
+	info->gpio_info.clk_en_register_index =
+			le16_to_cpu(header->asGPIO_Info[info->i2c_line].usClkEnRegisterIndex);
+	info->gpio_info.clk_y_register_index =
+			le16_to_cpu(header->asGPIO_Info[info->i2c_line].usClkY_RegisterIndex);
+	info->gpio_info.clk_a_register_index =
+			le16_to_cpu(header->asGPIO_Info[info->i2c_line].usClkA_RegisterIndex);
+	info->gpio_info.data_mask_register_index =
+			le16_to_cpu(header->asGPIO_Info[info->i2c_line].usDataMaskRegisterIndex);
+	info->gpio_info.data_en_register_index =
+			le16_to_cpu(header->asGPIO_Info[info->i2c_line].usDataEnRegisterIndex);
+	info->gpio_info.data_y_register_index =
+			le16_to_cpu(header->asGPIO_Info[info->i2c_line].usDataY_RegisterIndex);
+	info->gpio_info.data_a_register_index =
+			le16_to_cpu(header->asGPIO_Info[info->i2c_line].usDataA_RegisterIndex);
+
+	info->gpio_info.clk_mask_shift =
+			header->asGPIO_Info[info->i2c_line].ucClkMaskShift;
+	info->gpio_info.clk_en_shift =
+			header->asGPIO_Info[info->i2c_line].ucClkEnShift;
+	info->gpio_info.clk_y_shift =
+			header->asGPIO_Info[info->i2c_line].ucClkY_Shift;
+	info->gpio_info.clk_a_shift =
+			header->asGPIO_Info[info->i2c_line].ucClkA_Shift;
+	info->gpio_info.data_mask_shift =
+			header->asGPIO_Info[info->i2c_line].ucDataMaskShift;
+	info->gpio_info.data_en_shift =
+			header->asGPIO_Info[info->i2c_line].ucDataEnShift;
+	info->gpio_info.data_y_shift =
+			header->asGPIO_Info[info->i2c_line].ucDataY_Shift;
+	info->gpio_info.data_a_shift =
+			header->asGPIO_Info[info->i2c_line].ucDataA_Shift;
+
+	return BP_RESULT_OK;
+}
+
+static bool dal_graphics_object_id_is_valid(struct graphics_object_id id)
+{
+	bool rc = true;
+
+	switch (id.type) {
+	case OBJECT_TYPE_UNKNOWN:
+		rc = false;
+		break;
+	case OBJECT_TYPE_GPU:
+	case OBJECT_TYPE_ENGINE:
+		/* do NOT check for id.id == 0 */
+		if (id.enum_id == ENUM_ID_UNKNOWN)
+			rc = false;
+		break;
+	default:
+		if (id.id == 0 || id.enum_id == ENUM_ID_UNKNOWN)
+			rc = false;
+		break;
+	}
+
+	return rc;
+}
+
+static bool dal_graphics_object_id_is_equal(
+	struct graphics_object_id id1,
+	struct graphics_object_id id2)
+{
+	if (false == dal_graphics_object_id_is_valid(id1)) {
+		dm_output_to_console(
+		"%s: Warning: comparing invalid object 'id1'!\n", __func__);
+		return false;
+	}
+
+	if (false == dal_graphics_object_id_is_valid(id2)) {
+		dm_output_to_console(
+		"%s: Warning: comparing invalid object 'id2'!\n", __func__);
+		return false;
+	}
+
+	if (id1.id == id2.id && id1.enum_id == id2.enum_id
+		&& id1.type == id2.type)
+		return true;
+
+	return false;
+}
+
+static ATOM_OBJECT *get_bios_object(struct bios_parser *bp,
+	struct graphics_object_id id)
+{
+	uint32_t offset;
+	ATOM_OBJECT_TABLE *tbl;
+	uint32_t i;
+
+	switch (id.type) {
+	case OBJECT_TYPE_ENCODER:
+		offset = le16_to_cpu(bp->object_info_tbl.v1_1->usEncoderObjectTableOffset);
+		break;
+
+	case OBJECT_TYPE_CONNECTOR:
+		offset = le16_to_cpu(bp->object_info_tbl.v1_1->usConnectorObjectTableOffset);
+		break;
+
+	case OBJECT_TYPE_ROUTER:
+		offset = le16_to_cpu(bp->object_info_tbl.v1_1->usRouterObjectTableOffset);
+		break;
+
+	case OBJECT_TYPE_GENERIC:
+		if (bp->object_info_tbl.revision.minor < 3)
+			return NULL;
+		offset = le16_to_cpu(bp->object_info_tbl.v1_3->usMiscObjectTableOffset);
+		break;
+
+	default:
+		return NULL;
+	}
+
+	offset += bp->object_info_tbl_offset;
+
+	tbl = ((ATOM_OBJECT_TABLE *) bios_get_image(&bp->base, offset,
+				struct_size(tbl, asObjects, 1)));
+	if (!tbl)
+		return NULL;
+
+	for (i = 0; i < tbl->ucNumberOfObjects; i++)
+		if (dal_graphics_object_id_is_equal(id,
+				object_id_from_bios_object_id(
+						le16_to_cpu(tbl->asObjects[i].usObjectID))))
+			return &tbl->asObjects[i];
+
+	return NULL;
+}
+
+static uint32_t get_src_obj_list(struct bios_parser *bp, ATOM_OBJECT *object,
+	uint16_t **id_list)
+{
+	uint32_t offset;
+	uint8_t *number;
+
+	if (!object) {
+		BREAK_TO_DEBUGGER(); /* Invalid object id */
+		return 0;
+	}
+
+	offset = le16_to_cpu(object->usSrcDstTableOffset)
+					+ bp->object_info_tbl_offset;
+
+	number = GET_IMAGE(uint8_t, offset);
+	if (!number)
+		return 0;
+
+	offset += sizeof(uint8_t);
+	*id_list = (uint16_t *)bios_get_image(&bp->base, offset, *number * sizeof(uint16_t));
+
+	if (!*id_list)
+		return 0;
+
+	return *number;
+}
+
+static struct device_id device_type_from_device_id(uint16_t device_id)
+{
+
+	struct device_id result_device_id = {0};
+
+	switch (device_id) {
+	case ATOM_DEVICE_LCD1_SUPPORT:
+		result_device_id.device_type = DEVICE_TYPE_LCD;
+		result_device_id.enum_id = 1;
+		break;
+
+	case ATOM_DEVICE_LCD2_SUPPORT:
+		result_device_id.device_type = DEVICE_TYPE_LCD;
+		result_device_id.enum_id = 2;
+		break;
+
+	case ATOM_DEVICE_CRT1_SUPPORT:
+		result_device_id.device_type = DEVICE_TYPE_CRT;
+		result_device_id.enum_id = 1;
+		break;
+
+	case ATOM_DEVICE_CRT2_SUPPORT:
+		result_device_id.device_type = DEVICE_TYPE_CRT;
+		result_device_id.enum_id = 2;
+		break;
+
+	case ATOM_DEVICE_DFP1_SUPPORT:
+		result_device_id.device_type = DEVICE_TYPE_DFP;
+		result_device_id.enum_id = 1;
+		break;
+
+	case ATOM_DEVICE_DFP2_SUPPORT:
+		result_device_id.device_type = DEVICE_TYPE_DFP;
+		result_device_id.enum_id = 2;
+		break;
+
+	case ATOM_DEVICE_DFP3_SUPPORT:
+		result_device_id.device_type = DEVICE_TYPE_DFP;
+		result_device_id.enum_id = 3;
+		break;
+
+	case ATOM_DEVICE_DFP4_SUPPORT:
+		result_device_id.device_type = DEVICE_TYPE_DFP;
+		result_device_id.enum_id = 4;
+		break;
+
+	case ATOM_DEVICE_DFP5_SUPPORT:
+		result_device_id.device_type = DEVICE_TYPE_DFP;
+		result_device_id.enum_id = 5;
+		break;
+
+	case ATOM_DEVICE_DFP6_SUPPORT:
+		result_device_id.device_type = DEVICE_TYPE_DFP;
+		result_device_id.enum_id = 6;
+		break;
+
+	default:
+		BREAK_TO_DEBUGGER(); /* Invalid device Id */
+		result_device_id.device_type = DEVICE_TYPE_UNKNOWN;
+		result_device_id.enum_id = 0;
+	}
+	return result_device_id;
+}
+
+static void get_atom_data_table_revision(
+	ATOM_COMMON_TABLE_HEADER *atom_data_tbl,
+	struct atom_data_revision *tbl_revision)
+{
+	if (!tbl_revision)
+		return;
+
+	/* initialize the revision to 0 which is invalid revision */
+	tbl_revision->major = 0;
+	tbl_revision->minor = 0;
+
+	if (!atom_data_tbl)
+		return;
+
+	tbl_revision->major =
+			(uint32_t) GET_DATA_TABLE_MAJOR_REVISION(atom_data_tbl);
+	tbl_revision->minor =
+			(uint32_t) GET_DATA_TABLE_MINOR_REVISION(atom_data_tbl);
+}
+
+static uint32_t signal_to_ss_id(enum as_signal_type signal)
+{
+	uint32_t clk_id_ss = 0;
+
+	switch (signal) {
+	case AS_SIGNAL_TYPE_DVI:
+		clk_id_ss = ASIC_INTERNAL_SS_ON_TMDS;
+		break;
+	case AS_SIGNAL_TYPE_HDMI:
+		clk_id_ss = ASIC_INTERNAL_SS_ON_HDMI;
+		break;
+	case AS_SIGNAL_TYPE_LVDS:
+		clk_id_ss = ASIC_INTERNAL_SS_ON_LVDS;
+		break;
+	case AS_SIGNAL_TYPE_DISPLAY_PORT:
+		clk_id_ss = ASIC_INTERNAL_SS_ON_DP;
+		break;
+	case AS_SIGNAL_TYPE_GPU_PLL:
+		clk_id_ss = ASIC_INTERNAL_GPUPLL_SS;
+		break;
+	default:
+		break;
+	}
+	return clk_id_ss;
+}
+
+static uint32_t get_support_mask_for_device_id(struct device_id device_id)
+{
+	enum dal_device_type device_type = device_id.device_type;
+	uint32_t enum_id = device_id.enum_id;
+
+	switch (device_type) {
+	case DEVICE_TYPE_LCD:
+		switch (enum_id) {
+		case 1:
+			return ATOM_DEVICE_LCD1_SUPPORT;
+		case 2:
+			return ATOM_DEVICE_LCD2_SUPPORT;
+		default:
+			break;
+		}
+		break;
+	case DEVICE_TYPE_CRT:
+		switch (enum_id) {
+		case 1:
+			return ATOM_DEVICE_CRT1_SUPPORT;
+		case 2:
+			return ATOM_DEVICE_CRT2_SUPPORT;
+		default:
+			break;
+		}
+		break;
+	case DEVICE_TYPE_DFP:
+		switch (enum_id) {
+		case 1:
+			return ATOM_DEVICE_DFP1_SUPPORT;
+		case 2:
+			return ATOM_DEVICE_DFP2_SUPPORT;
+		case 3:
+			return ATOM_DEVICE_DFP3_SUPPORT;
+		case 4:
+			return ATOM_DEVICE_DFP4_SUPPORT;
+		case 5:
+			return ATOM_DEVICE_DFP5_SUPPORT;
+		case 6:
+			return ATOM_DEVICE_DFP6_SUPPORT;
+		default:
+			break;
+		}
+		break;
+	case DEVICE_TYPE_CV:
+		switch (enum_id) {
+		case 1:
+			return ATOM_DEVICE_CV_SUPPORT;
+		default:
+			break;
+		}
+		break;
+	case DEVICE_TYPE_TV:
+		switch (enum_id) {
+		case 1:
+			return ATOM_DEVICE_TV1_SUPPORT;
+		default:
+			break;
+		}
+		break;
+	default:
+		break;
+	}
+
+	/* Unidentified device ID, return empty support mask. */
+	return 0;
+}
+
+/**
+ * bios_parser_set_scratch_critical_state - update critical state
+ *                                          bit in VBIOS scratch register
+ * @dcb:    pointer to the DC BIOS
+ * @state:  set or reset state
+ */
+static void bios_parser_set_scratch_critical_state(
+	struct dc_bios *dcb,
+	bool state)
+{
+	bios_set_scratch_critical_state(dcb, state);
+}
+
+/*
+ * get_integrated_info_v8
+ *
+ * @brief
+ * Get V8 integrated BIOS information
+ *
+ * @param
+ * bios_parser *bp - [in]BIOS parser handler to get master data table
+ * integrated_info *info - [out] store and output integrated info
+ *
+ * return:
+ * enum bp_result - BP_RESULT_OK if information is available,
+ *                  BP_RESULT_BADBIOSTABLE otherwise.
+ */
+static enum bp_result get_integrated_info_v8(
+	struct bios_parser *bp,
+	struct integrated_info *info)
+{
+	ATOM_INTEGRATED_SYSTEM_INFO_V1_8 *info_v8;
+	uint32_t i;
+
+	info_v8 = GET_IMAGE(ATOM_INTEGRATED_SYSTEM_INFO_V1_8,
+			bp->master_data_tbl->ListOfDataTables.IntegratedSystemInfo);
+
+	if (info_v8 == NULL)
+		return BP_RESULT_BADBIOSTABLE;
+	info->boot_up_engine_clock = le32_to_cpu(info_v8->ulBootUpEngineClock) * 10;
+	info->dentist_vco_freq = le32_to_cpu(info_v8->ulDentistVCOFreq) * 10;
+	info->boot_up_uma_clock = le32_to_cpu(info_v8->ulBootUpUMAClock) * 10;
+
+	for (i = 0; i < NUMBER_OF_DISP_CLK_VOLTAGE; ++i) {
+		/* Convert [10KHz] into [KHz] */
+		info->disp_clk_voltage[i].max_supported_clk =
+			le32_to_cpu(info_v8->sDISPCLK_Voltage[i].
+				    ulMaximumSupportedCLK) * 10;
+		info->disp_clk_voltage[i].voltage_index =
+			le32_to_cpu(info_v8->sDISPCLK_Voltage[i].ulVoltageIndex);
+	}
+
+	info->boot_up_req_display_vector =
+		le32_to_cpu(info_v8->ulBootUpReqDisplayVector);
+	info->gpu_cap_info =
+		le32_to_cpu(info_v8->ulGPUCapInfo);
+
+	/*
+	 * system_config: Bit[0] = 0 : PCIE power gating disabled
+	 *                       = 1 : PCIE power gating enabled
+	 *                Bit[1] = 0 : DDR-PLL shut down disabled
+	 *                       = 1 : DDR-PLL shut down enabled
+	 *                Bit[2] = 0 : DDR-PLL power down disabled
+	 *                       = 1 : DDR-PLL power down enabled
+	 */
+	info->system_config = le32_to_cpu(info_v8->ulSystemConfig);
+	info->cpu_cap_info = le32_to_cpu(info_v8->ulCPUCapInfo);
+	info->boot_up_nb_voltage =
+		le16_to_cpu(info_v8->usBootUpNBVoltage);
+	info->ext_disp_conn_info_offset =
+		le16_to_cpu(info_v8->usExtDispConnInfoOffset);
+	info->memory_type = info_v8->ucMemoryType;
+	info->ma_channel_number = info_v8->ucUMAChannelNumber;
+	info->gmc_restore_reset_time =
+		le32_to_cpu(info_v8->ulGMCRestoreResetTime);
+
+	info->minimum_n_clk =
+		le32_to_cpu(info_v8->ulNbpStateNClkFreq[0]);
+	for (i = 1; i < 4; ++i)
+		info->minimum_n_clk =
+			info->minimum_n_clk < le32_to_cpu(info_v8->ulNbpStateNClkFreq[i]) ?
+			info->minimum_n_clk : le32_to_cpu(info_v8->ulNbpStateNClkFreq[i]);
+
+	info->idle_n_clk = le32_to_cpu(info_v8->ulIdleNClk);
+	info->ddr_dll_power_up_time =
+		le32_to_cpu(info_v8->ulDDR_DLL_PowerUpTime);
+	info->ddr_pll_power_up_time =
+		le32_to_cpu(info_v8->ulDDR_PLL_PowerUpTime);
+	info->pcie_clk_ss_type = le16_to_cpu(info_v8->usPCIEClkSSType);
+	info->lvds_ss_percentage =
+		le16_to_cpu(info_v8->usLvdsSSPercentage);
+	info->lvds_sspread_rate_in_10hz =
+		le16_to_cpu(info_v8->usLvdsSSpreadRateIn10Hz);
+	info->hdmi_ss_percentage =
+		le16_to_cpu(info_v8->usHDMISSPercentage);
+	info->hdmi_sspread_rate_in_10hz =
+		le16_to_cpu(info_v8->usHDMISSpreadRateIn10Hz);
+	info->dvi_ss_percentage =
+		le16_to_cpu(info_v8->usDVISSPercentage);
+	info->dvi_sspread_rate_in_10_hz =
+		le16_to_cpu(info_v8->usDVISSpreadRateIn10Hz);
+
+	info->max_lvds_pclk_freq_in_single_link =
+		le16_to_cpu(info_v8->usMaxLVDSPclkFreqInSingleLink);
+	info->lvds_misc = info_v8->ucLvdsMisc;
+	info->lvds_pwr_on_seq_dig_on_to_de_in_4ms =
+		info_v8->ucLVDSPwrOnSeqDIGONtoDE_in4Ms;
+	info->lvds_pwr_on_seq_de_to_vary_bl_in_4ms =
+		info_v8->ucLVDSPwrOnSeqDEtoVARY_BL_in4Ms;
+	info->lvds_pwr_on_seq_vary_bl_to_blon_in_4ms =
+		info_v8->ucLVDSPwrOnSeqVARY_BLtoBLON_in4Ms;
+	info->lvds_pwr_off_seq_vary_bl_to_de_in4ms =
+		info_v8->ucLVDSPwrOffSeqVARY_BLtoDE_in4Ms;
+	info->lvds_pwr_off_seq_de_to_dig_on_in4ms =
+		info_v8->ucLVDSPwrOffSeqDEtoDIGON_in4Ms;
+	info->lvds_pwr_off_seq_blon_to_vary_bl_in_4ms =
+		info_v8->ucLVDSPwrOffSeqBLONtoVARY_BL_in4Ms;
+	info->lvds_off_to_on_delay_in_4ms =
+		info_v8->ucLVDSOffToOnDelay_in4Ms;
+	info->lvds_bit_depth_control_val =
+		le32_to_cpu(info_v8->ulLCDBitDepthControlVal);
+
+	for (i = 0; i < NUMBER_OF_AVAILABLE_SCLK; ++i) {
+		/* Convert [10KHz] into [KHz] */
+		info->avail_s_clk[i].supported_s_clk =
+			le32_to_cpu(info_v8->sAvail_SCLK[i].ulSupportedSCLK) * 10;
+		info->avail_s_clk[i].voltage_index =
+			le16_to_cpu(info_v8->sAvail_SCLK[i].usVoltageIndex);
+		info->avail_s_clk[i].voltage_id =
+			le16_to_cpu(info_v8->sAvail_SCLK[i].usVoltageID);
+	}
+
+	for (i = 0; i < NUMBER_OF_UCHAR_FOR_GUID; ++i) {
+		info->ext_disp_conn_info.gu_id[i] =
+			info_v8->sExtDispConnInfo.ucGuid[i];
+	}
+
+	for (i = 0; i < MAX_NUMBER_OF_EXT_DISPLAY_PATH; ++i) {
+		info->ext_disp_conn_info.path[i].device_connector_id =
+			object_id_from_bios_object_id(
+				le16_to_cpu(info_v8->sExtDispConnInfo.sPath[i].usDeviceConnector));
+
+		info->ext_disp_conn_info.path[i].ext_encoder_obj_id =
+			object_id_from_bios_object_id(
+				le16_to_cpu(info_v8->sExtDispConnInfo.sPath[i].usExtEncoderObjId));
+
+		info->ext_disp_conn_info.path[i].device_tag =
+			le16_to_cpu(info_v8->sExtDispConnInfo.sPath[i].usDeviceTag);
+		info->ext_disp_conn_info.path[i].device_acpi_enum =
+			le16_to_cpu(info_v8->sExtDispConnInfo.sPath[i].usDeviceACPIEnum);
+		info->ext_disp_conn_info.path[i].ext_aux_ddc_lut_index =
+			info_v8->sExtDispConnInfo.sPath[i].ucExtAUXDDCLutIndex;
+		info->ext_disp_conn_info.path[i].ext_hpd_pin_lut_index =
+			info_v8->sExtDispConnInfo.sPath[i].ucExtHPDPINLutIndex;
+		info->ext_disp_conn_info.path[i].channel_mapping.raw =
+			info_v8->sExtDispConnInfo.sPath[i].ucChannelMapping;
+	}
+	info->ext_disp_conn_info.checksum =
+		info_v8->sExtDispConnInfo.ucChecksum;
+
+	return BP_RESULT_OK;
+}
+
+/*
+ * get_integrated_info_v9
+ *
+ * @brief
+ * Get V9 integrated BIOS information
+ *
+ * @param
+ * bios_parser *bp - [in]BIOS parser handler to get master data table
+ * integrated_info *info - [out] store and output integrated info
+ *
+ * return:
+ * enum bp_result - BP_RESULT_OK if information is available,
+ *                  BP_RESULT_BADBIOSTABLE otherwise.
+ */
+static enum bp_result get_integrated_info_v9(
+	struct bios_parser *bp,
+	struct integrated_info *info)
+{
+	ATOM_INTEGRATED_SYSTEM_INFO_V1_9 *info_v9;
+	uint32_t i;
+
+	info_v9 = GET_IMAGE(ATOM_INTEGRATED_SYSTEM_INFO_V1_9,
+			bp->master_data_tbl->ListOfDataTables.IntegratedSystemInfo);
+
+	if (!info_v9)
+		return BP_RESULT_BADBIOSTABLE;
+
+	info->boot_up_engine_clock = le32_to_cpu(info_v9->ulBootUpEngineClock) * 10;
+	info->dentist_vco_freq = le32_to_cpu(info_v9->ulDentistVCOFreq) * 10;
+	info->boot_up_uma_clock = le32_to_cpu(info_v9->ulBootUpUMAClock) * 10;
+
+	for (i = 0; i < NUMBER_OF_DISP_CLK_VOLTAGE; ++i) {
+		/* Convert [10KHz] into [KHz] */
+		info->disp_clk_voltage[i].max_supported_clk =
+			le32_to_cpu(info_v9->sDISPCLK_Voltage[i].ulMaximumSupportedCLK) * 10;
+		info->disp_clk_voltage[i].voltage_index =
+			le32_to_cpu(info_v9->sDISPCLK_Voltage[i].ulVoltageIndex);
+	}
+
+	info->boot_up_req_display_vector =
+		le32_to_cpu(info_v9->ulBootUpReqDisplayVector);
+	info->gpu_cap_info = le32_to_cpu(info_v9->ulGPUCapInfo);
+
+	/*
+	 * system_config: Bit[0] = 0 : PCIE power gating disabled
+	 *                       = 1 : PCIE power gating enabled
+	 *                Bit[1] = 0 : DDR-PLL shut down disabled
+	 *                       = 1 : DDR-PLL shut down enabled
+	 *                Bit[2] = 0 : DDR-PLL power down disabled
+	 *                       = 1 : DDR-PLL power down enabled
+	 */
+	info->system_config = le32_to_cpu(info_v9->ulSystemConfig);
+	info->cpu_cap_info = le32_to_cpu(info_v9->ulCPUCapInfo);
+	info->boot_up_nb_voltage = le16_to_cpu(info_v9->usBootUpNBVoltage);
+	info->ext_disp_conn_info_offset = le16_to_cpu(info_v9->usExtDispConnInfoOffset);
+	info->memory_type = info_v9->ucMemoryType;
+	info->ma_channel_number = info_v9->ucUMAChannelNumber;
+	info->gmc_restore_reset_time = le32_to_cpu(info_v9->ulGMCRestoreResetTime);
+
+	info->minimum_n_clk = le32_to_cpu(info_v9->ulNbpStateNClkFreq[0]);
+	for (i = 1; i < 4; ++i)
+		info->minimum_n_clk =
+			info->minimum_n_clk < le32_to_cpu(info_v9->ulNbpStateNClkFreq[i]) ?
+			info->minimum_n_clk : le32_to_cpu(info_v9->ulNbpStateNClkFreq[i]);
+
+	info->idle_n_clk = le32_to_cpu(info_v9->ulIdleNClk);
+	info->ddr_dll_power_up_time = le32_to_cpu(info_v9->ulDDR_DLL_PowerUpTime);
+	info->ddr_pll_power_up_time = le32_to_cpu(info_v9->ulDDR_PLL_PowerUpTime);
+	info->pcie_clk_ss_type = le16_to_cpu(info_v9->usPCIEClkSSType);
+	info->lvds_ss_percentage = le16_to_cpu(info_v9->usLvdsSSPercentage);
+	info->lvds_sspread_rate_in_10hz = le16_to_cpu(info_v9->usLvdsSSpreadRateIn10Hz);
+	info->hdmi_ss_percentage = le16_to_cpu(info_v9->usHDMISSPercentage);
+	info->hdmi_sspread_rate_in_10hz = le16_to_cpu(info_v9->usHDMISSpreadRateIn10Hz);
+	info->dvi_ss_percentage = le16_to_cpu(info_v9->usDVISSPercentage);
+	info->dvi_sspread_rate_in_10_hz = le16_to_cpu(info_v9->usDVISSpreadRateIn10Hz);
+
+	info->max_lvds_pclk_freq_in_single_link =
+		le16_to_cpu(info_v9->usMaxLVDSPclkFreqInSingleLink);
+	info->lvds_misc = info_v9->ucLvdsMisc;
+	info->lvds_pwr_on_seq_dig_on_to_de_in_4ms =
+		info_v9->ucLVDSPwrOnSeqDIGONtoDE_in4Ms;
+	info->lvds_pwr_on_seq_de_to_vary_bl_in_4ms =
+		info_v9->ucLVDSPwrOnSeqDEtoVARY_BL_in4Ms;
+	info->lvds_pwr_on_seq_vary_bl_to_blon_in_4ms =
+		info_v9->ucLVDSPwrOnSeqVARY_BLtoBLON_in4Ms;
+	info->lvds_pwr_off_seq_vary_bl_to_de_in4ms =
+		info_v9->ucLVDSPwrOffSeqVARY_BLtoDE_in4Ms;
+	info->lvds_pwr_off_seq_de_to_dig_on_in4ms =
+		info_v9->ucLVDSPwrOffSeqDEtoDIGON_in4Ms;
+	info->lvds_pwr_off_seq_blon_to_vary_bl_in_4ms =
+		info_v9->ucLVDSPwrOffSeqBLONtoVARY_BL_in4Ms;
+	info->lvds_off_to_on_delay_in_4ms =
+		info_v9->ucLVDSOffToOnDelay_in4Ms;
+	info->lvds_bit_depth_control_val =
+		le32_to_cpu(info_v9->ulLCDBitDepthControlVal);
+
+	for (i = 0; i < NUMBER_OF_AVAILABLE_SCLK; ++i) {
+		/* Convert [10KHz] into [KHz] */
+		info->avail_s_clk[i].supported_s_clk =
+			le32_to_cpu(info_v9->sAvail_SCLK[i].ulSupportedSCLK) * 10;
+		info->avail_s_clk[i].voltage_index =
+			le16_to_cpu(info_v9->sAvail_SCLK[i].usVoltageIndex);
+		info->avail_s_clk[i].voltage_id =
+			le16_to_cpu(info_v9->sAvail_SCLK[i].usVoltageID);
+	}
+
+	for (i = 0; i < NUMBER_OF_UCHAR_FOR_GUID; ++i) {
+		info->ext_disp_conn_info.gu_id[i] =
+			info_v9->sExtDispConnInfo.ucGuid[i];
+	}
+
+	for (i = 0; i < MAX_NUMBER_OF_EXT_DISPLAY_PATH; ++i) {
+		info->ext_disp_conn_info.path[i].device_connector_id =
+			object_id_from_bios_object_id(
+				le16_to_cpu(info_v9->sExtDispConnInfo.sPath[i].usDeviceConnector));
+
+		info->ext_disp_conn_info.path[i].ext_encoder_obj_id =
+			object_id_from_bios_object_id(
+				le16_to_cpu(info_v9->sExtDispConnInfo.sPath[i].usExtEncoderObjId));
+
+		info->ext_disp_conn_info.path[i].device_tag =
+			le16_to_cpu(info_v9->sExtDispConnInfo.sPath[i].usDeviceTag);
+		info->ext_disp_conn_info.path[i].device_acpi_enum =
+			le16_to_cpu(info_v9->sExtDispConnInfo.sPath[i].usDeviceACPIEnum);
+		info->ext_disp_conn_info.path[i].ext_aux_ddc_lut_index =
+			info_v9->sExtDispConnInfo.sPath[i].ucExtAUXDDCLutIndex;
+		info->ext_disp_conn_info.path[i].ext_hpd_pin_lut_index =
+			info_v9->sExtDispConnInfo.sPath[i].ucExtHPDPINLutIndex;
+		info->ext_disp_conn_info.path[i].channel_mapping.raw =
+			info_v9->sExtDispConnInfo.sPath[i].ucChannelMapping;
+	}
+	info->ext_disp_conn_info.checksum =
+		info_v9->sExtDispConnInfo.ucChecksum;
+
+	return BP_RESULT_OK;
+}
+
+/*
+ * construct_integrated_info
+ *
+ * @brief
+ * Get integrated BIOS information based on table revision
+ *
+ * @param
+ * bios_parser *bp - [in]BIOS parser handler to get master data table
+ * integrated_info *info - [out] store and output integrated info
+ *
+ * return:
+ * enum bp_result - BP_RESULT_OK if information is available,
+ *                  BP_RESULT_BADBIOSTABLE otherwise.
+ */
+static enum bp_result construct_integrated_info(
+	struct bios_parser *bp,
+	struct integrated_info *info)
+{
+	enum bp_result result = BP_RESULT_BADBIOSTABLE;
+
+	ATOM_COMMON_TABLE_HEADER *header;
+	struct atom_data_revision revision;
+
+	if (bp->master_data_tbl->ListOfDataTables.IntegratedSystemInfo) {
+		header = GET_IMAGE(ATOM_COMMON_TABLE_HEADER,
+				bp->master_data_tbl->ListOfDataTables.IntegratedSystemInfo);
+
+		get_atom_data_table_revision(header, &revision);
+
+		/* Don't need to check major revision as they are all 1 */
+		switch (revision.minor) {
+		case 8:
+			result = get_integrated_info_v8(bp, info);
+			break;
+		case 9:
+			result = get_integrated_info_v9(bp, info);
+			break;
+		default:
+			return result;
+
+		}
+	}
+
+	/* Sort voltage table from low to high*/
+	if (result == BP_RESULT_OK) {
+		int32_t i;
+		int32_t j;
+
+		for (i = 1; i < NUMBER_OF_DISP_CLK_VOLTAGE; ++i) {
+			for (j = i; j > 0; --j) {
+				if (
+						info->disp_clk_voltage[j].max_supported_clk <
+						info->disp_clk_voltage[j-1].max_supported_clk) {
+					/* swap j and j - 1*/
+					swap(info->disp_clk_voltage[j - 1],
+					     info->disp_clk_voltage[j]);
+				}
+			}
+		}
+
+	}
+
+	return result;
+}
+
+static struct integrated_info *bios_parser_create_integrated_info(
+	struct dc_bios *dcb)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	struct integrated_info *info;
+
+	info = kzalloc(sizeof(struct integrated_info), GFP_KERNEL);
+
+	if (info == NULL) {
+		ASSERT_CRITICAL(0);
+		return NULL;
+	}
+
+	if (construct_integrated_info(bp, info) == BP_RESULT_OK)
+		return info;
+
+	kfree(info);
+
+	return NULL;
+}
+
+static enum bp_result update_slot_layout_info(struct dc_bios *dcb,
+					      unsigned int i,
+					      struct slot_layout_info *slot_layout_info,
+					      unsigned int record_offset)
+{
+	unsigned int j;
+	struct bios_parser *bp;
+	ATOM_BRACKET_LAYOUT_RECORD *record;
+	ATOM_COMMON_RECORD_HEADER *record_header;
+	enum bp_result result = BP_RESULT_NORECORD;
+
+	bp = BP_FROM_DCB(dcb);
+	record = NULL;
+	record_header = NULL;
+
+	for (;;) {
+
+		record_header = GET_IMAGE(ATOM_COMMON_RECORD_HEADER, record_offset);
+		if (record_header == NULL) {
+			result = BP_RESULT_BADBIOSTABLE;
+			break;
+		}
+
+		/* the end of the list */
+		if (record_header->ucRecordType == 0xff ||
+			record_header->ucRecordSize == 0)	{
+			break;
+		}
+
+		if (record_header->ucRecordType ==
+			ATOM_BRACKET_LAYOUT_RECORD_TYPE &&
+			struct_size(record, asConnInfo, 1)
+			<= record_header->ucRecordSize) {
+			record = (ATOM_BRACKET_LAYOUT_RECORD *)
+				(record_header);
+			result = BP_RESULT_OK;
+			break;
+		}
+
+		record_offset += record_header->ucRecordSize;
+	}
+
+	/* return if the record not found */
+	if (result != BP_RESULT_OK)
+		return result;
+
+	/* get slot sizes */
+	slot_layout_info->length = record->ucLength;
+	slot_layout_info->width = record->ucWidth;
+
+	/* get info for each connector in the slot */
+	slot_layout_info->num_of_connectors = record->ucConnNum;
+	for (j = 0; j < slot_layout_info->num_of_connectors; ++j) {
+		slot_layout_info->connectors[j].connector_type =
+			(enum connector_layout_type)
+			(record->asConnInfo[j].ucConnectorType);
+		switch (record->asConnInfo[j].ucConnectorType) {
+		case CONNECTOR_TYPE_DVI_D:
+			slot_layout_info->connectors[j].connector_type =
+				CONNECTOR_LAYOUT_TYPE_DVI_D;
+			slot_layout_info->connectors[j].length =
+				CONNECTOR_SIZE_DVI;
+			break;
+
+		case CONNECTOR_TYPE_HDMI:
+			slot_layout_info->connectors[j].connector_type =
+				CONNECTOR_LAYOUT_TYPE_HDMI;
+			slot_layout_info->connectors[j].length =
+				CONNECTOR_SIZE_HDMI;
+			break;
+
+		case CONNECTOR_TYPE_DISPLAY_PORT:
+			slot_layout_info->connectors[j].connector_type =
+				CONNECTOR_LAYOUT_TYPE_DP;
+			slot_layout_info->connectors[j].length =
+				CONNECTOR_SIZE_DP;
+			break;
+
+		case CONNECTOR_TYPE_MINI_DISPLAY_PORT:
+			slot_layout_info->connectors[j].connector_type =
+				CONNECTOR_LAYOUT_TYPE_MINI_DP;
+			slot_layout_info->connectors[j].length =
+				CONNECTOR_SIZE_MINI_DP;
+			break;
+
+		default:
+			slot_layout_info->connectors[j].connector_type =
+				CONNECTOR_LAYOUT_TYPE_UNKNOWN;
+			slot_layout_info->connectors[j].length =
+				CONNECTOR_SIZE_UNKNOWN;
+		}
+
+		slot_layout_info->connectors[j].position =
+			record->asConnInfo[j].ucPosition;
+		slot_layout_info->connectors[j].connector_id =
+			object_id_from_bios_object_id(
+				record->asConnInfo[j].usConnectorObjectId);
+	}
+	return result;
+}
+
+
+static enum bp_result get_bracket_layout_record(struct dc_bios *dcb,
+						unsigned int bracket_layout_id,
+						struct slot_layout_info *slot_layout_info)
+{
+	unsigned int i;
+	unsigned int record_offset;
+	struct bios_parser *bp;
+	enum bp_result result;
+	ATOM_OBJECT *object;
+	ATOM_OBJECT_TABLE *object_table;
+	unsigned int genericTableOffset;
+
+	bp = BP_FROM_DCB(dcb);
+	object = NULL;
+	if (slot_layout_info == NULL) {
+		DC_LOG_DETECTION_EDID_PARSER("Invalid slot_layout_info\n");
+		return BP_RESULT_BADINPUT;
+	}
+
+
+	genericTableOffset = bp->object_info_tbl_offset +
+		bp->object_info_tbl.v1_3->usMiscObjectTableOffset;
+	object_table = ((ATOM_OBJECT_TABLE *) bios_get_image(&bp->base,
+				genericTableOffset,
+				struct_size(object_table, asObjects, 1)));
+	if (!object_table)
+		return BP_RESULT_FAILURE;
+
+	result = BP_RESULT_NORECORD;
+	for (i = 0; i < object_table->ucNumberOfObjects; ++i) {
+
+		if (bracket_layout_id ==
+			object_table->asObjects[i].usObjectID) {
+
+			object = &object_table->asObjects[i];
+			record_offset = object->usRecordOffset +
+				bp->object_info_tbl_offset;
+
+			result = update_slot_layout_info(dcb, i,
+				slot_layout_info, record_offset);
+			break;
+		}
+	}
+	return result;
+}
+
+static enum bp_result bios_get_board_layout_info(
+	struct dc_bios *dcb,
+	struct board_layout_info *board_layout_info)
+{
+	unsigned int i;
+	struct bios_parser *bp;
+	enum bp_result record_result;
+
+	const unsigned int slot_index_to_vbios_id[MAX_BOARD_SLOTS] = {
+		GENERICOBJECT_BRACKET_LAYOUT_ENUM_ID1,
+		GENERICOBJECT_BRACKET_LAYOUT_ENUM_ID2,
+		0, 0
+	};
+
+	bp = BP_FROM_DCB(dcb);
+
+	if (board_layout_info == NULL) {
+		DC_LOG_DETECTION_EDID_PARSER("Invalid board_layout_info\n");
+		return BP_RESULT_BADINPUT;
+	}
+
+	board_layout_info->num_of_slots = 0;
+
+	for (i = 0; i < MAX_BOARD_SLOTS; ++i) {
+		record_result = get_bracket_layout_record(dcb,
+			slot_index_to_vbios_id[i],
+			&board_layout_info->slots[i]);
+
+		if (record_result == BP_RESULT_NORECORD && i > 0)
+			break; /* no more slots present in bios */
+		else if (record_result != BP_RESULT_OK)
+			return record_result;  /* fail */
+
+		++board_layout_info->num_of_slots;
+	}
+
+	/* all data is valid */
+	board_layout_info->is_number_of_slots_valid = 1;
+	board_layout_info->is_slots_size_valid = 1;
+	board_layout_info->is_connector_offsets_valid = 1;
+	board_layout_info->is_connector_lengths_valid = 1;
+
+	return BP_RESULT_OK;
+}
+
+/******************************************************************************/
+
+static const struct dc_vbios_funcs vbios_funcs = {
+	.get_connectors_number = bios_parser_get_connectors_number,
+
+	.get_connector_id = bios_parser_get_connector_id,
+
+	.get_src_obj = bios_parser_get_src_obj,
+
+	.get_i2c_info = bios_parser_get_i2c_info,
+
+	.get_hpd_info = bios_parser_get_hpd_info,
+
+	.get_device_tag = bios_parser_get_device_tag,
+
+	.get_spread_spectrum_info = bios_parser_get_spread_spectrum_info,
+
+	.get_ss_entry_number = bios_parser_get_ss_entry_number,
+
+	.get_embedded_panel_info = bios_parser_get_embedded_panel_info,
+
+	.get_gpio_pin_info = bios_parser_get_gpio_pin_info,
+
+	.get_encoder_cap_info = bios_parser_get_encoder_cap_info,
+
+	/* bios scratch register communication */
+	.is_accelerated_mode = bios_is_accelerated_mode,
+
+	.set_scratch_critical_state = bios_parser_set_scratch_critical_state,
+
+	.is_device_id_supported = bios_parser_is_device_id_supported,
+
+	/* COMMANDS */
+	.encoder_control = bios_parser_encoder_control,
+
+	.transmitter_control = bios_parser_transmitter_control,
+
+	.enable_crtc = bios_parser_enable_crtc,
+
+	.adjust_pixel_clock = bios_parser_adjust_pixel_clock,
+
+	.set_pixel_clock = bios_parser_set_pixel_clock,
+
+	.set_dce_clock = bios_parser_set_dce_clock,
+
+	.enable_spread_spectrum_on_ppll = bios_parser_enable_spread_spectrum_on_ppll,
+
+	.program_crtc_timing = bios_parser_program_crtc_timing, /* still use.  should probably retire and program directly */
+
+	.program_display_engine_pll = bios_parser_program_display_engine_pll,
+
+	.enable_disp_power_gating = bios_parser_enable_disp_power_gating,
+
+	/* SW init and patch */
+
+	.bios_parser_destroy = bios_parser_destroy,
+
+	.get_board_layout_info = bios_get_board_layout_info,
+
+	.get_atom_dc_golden_table = NULL
+};
+
+static bool bios_parser_construct(
+	struct bios_parser *bp,
+	struct bp_init_data *init,
+	enum dce_version dce_version)
+{
+	uint16_t *rom_header_offset = NULL;
+	ATOM_ROM_HEADER *rom_header = NULL;
+	ATOM_OBJECT_HEADER *object_info_tbl;
+	struct atom_data_revision tbl_rev = {0};
+
+	if (!init)
+		return false;
+
+	if (!init->bios)
+		return false;
+
+	bp->base.funcs = &vbios_funcs;
+	bp->base.bios = init->bios;
+	bp->base.bios_size = bp->base.bios[BIOS_IMAGE_SIZE_OFFSET] * BIOS_IMAGE_SIZE_UNIT;
+
+	bp->base.ctx = init->ctx;
+	bp->base.bios_local_image = NULL;
+
+	rom_header_offset =
+	GET_IMAGE(uint16_t, OFFSET_TO_POINTER_TO_ATOM_ROM_HEADER);
+
+	if (!rom_header_offset)
+		return false;
+
+	rom_header = GET_IMAGE(ATOM_ROM_HEADER, *rom_header_offset);
+
+	if (!rom_header)
+		return false;
+
+	get_atom_data_table_revision(&rom_header->sHeader, &tbl_rev);
+	if (tbl_rev.major >= 2 && tbl_rev.minor >= 2)
+		return false;
+
+	bp->master_data_tbl =
+	GET_IMAGE(ATOM_MASTER_DATA_TABLE,
+		rom_header->usMasterDataTableOffset);
+
+	if (!bp->master_data_tbl)
+		return false;
+
+	bp->object_info_tbl_offset = DATA_TABLES(Object_Header);
+
+	if (!bp->object_info_tbl_offset)
+		return false;
+
+	object_info_tbl =
+	GET_IMAGE(ATOM_OBJECT_HEADER, bp->object_info_tbl_offset);
+
+	if (!object_info_tbl)
+		return false;
+
+	get_atom_data_table_revision(&object_info_tbl->sHeader,
+		&bp->object_info_tbl.revision);
+
+	if (bp->object_info_tbl.revision.major == 1
+		&& bp->object_info_tbl.revision.minor >= 3) {
+		ATOM_OBJECT_HEADER_V3 *tbl_v3;
+
+		tbl_v3 = GET_IMAGE(ATOM_OBJECT_HEADER_V3,
+			bp->object_info_tbl_offset);
+		if (!tbl_v3)
+			return false;
+
+		bp->object_info_tbl.v1_3 = tbl_v3;
+	} else if (bp->object_info_tbl.revision.major == 1
+		&& bp->object_info_tbl.revision.minor >= 1)
+		bp->object_info_tbl.v1_1 = object_info_tbl;
+	else
+		return false;
+
+	dal_bios_parser_init_cmd_tbl(bp);
+	dal_bios_parser_init_cmd_tbl_helper(&bp->cmd_helper, dce_version);
+
+	bp->base.integrated_info = bios_parser_create_integrated_info(&bp->base);
+	bp->base.fw_info_valid = bios_parser_get_firmware_info(&bp->base, &bp->base.fw_info) == BP_RESULT_OK;
+
+	return true;
+}
+
+/******************************************************************************/
diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser.h b/drivers/gpu/drm/amd/display/dc/bios/bios_parser.h
new file mode 100644
index 000000000000..d6f16275048f
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_BIOS_PARSER_H__
+#define __DAL_BIOS_PARSER_H__
+
+struct dc_bios *bios_parser_create(
+	struct bp_init_data *init,
+	enum dce_version dce_version);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c b/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c
new file mode 100644
index 000000000000..04eb647acc4e
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c
@@ -0,0 +1,3807 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+#include "core_types.h"
+
+#include "ObjectID.h"
+#include "atomfirmware.h"
+
+#include "dc_bios_types.h"
+#include "include/grph_object_ctrl_defs.h"
+#include "include/bios_parser_interface.h"
+#include "include/logger_interface.h"
+
+#include "command_table2.h"
+
+#include "bios_parser_helper.h"
+#include "command_table_helper2.h"
+#include "bios_parser2.h"
+#include "bios_parser_types_internal2.h"
+#include "bios_parser_interface.h"
+
+#include "bios_parser_common.h"
+
+#define DC_LOGGER \
+	bp->base.ctx->logger
+
+#define LAST_RECORD_TYPE 0xff
+#define SMU9_SYSPLL0_ID  0
+
+static enum bp_result get_gpio_i2c_info(struct bios_parser *bp,
+	struct atom_i2c_record *record,
+	struct graphics_object_i2c_info *info);
+
+static enum bp_result bios_parser_get_firmware_info(
+	struct dc_bios *dcb,
+	struct dc_firmware_info *info);
+
+static enum bp_result bios_parser_get_encoder_cap_info(
+	struct dc_bios *dcb,
+	struct graphics_object_id object_id,
+	struct bp_encoder_cap_info *info);
+
+static enum bp_result get_firmware_info_v3_1(
+	struct bios_parser *bp,
+	struct dc_firmware_info *info);
+
+static enum bp_result get_firmware_info_v3_2(
+	struct bios_parser *bp,
+	struct dc_firmware_info *info);
+
+static enum bp_result get_firmware_info_v3_4(
+	struct bios_parser *bp,
+	struct dc_firmware_info *info);
+
+static enum bp_result get_firmware_info_v3_5(
+	struct bios_parser *bp,
+	struct dc_firmware_info *info);
+
+static struct atom_hpd_int_record *get_hpd_record(struct bios_parser *bp,
+		struct atom_display_object_path_v2 *object);
+
+static struct atom_encoder_caps_record *get_encoder_cap_record(
+	struct bios_parser *bp,
+	struct atom_display_object_path_v2 *object);
+
+#define BIOS_IMAGE_SIZE_OFFSET 2
+#define BIOS_IMAGE_SIZE_UNIT 512
+
+#define DATA_TABLES(table) (bp->master_data_tbl->listOfdatatables.table)
+
+static void bios_parser2_destruct(struct bios_parser *bp)
+{
+	kfree(bp->base.bios_local_image);
+	kfree(bp->base.integrated_info);
+}
+
+static void firmware_parser_destroy(struct dc_bios **dcb)
+{
+	struct bios_parser *bp = BP_FROM_DCB(*dcb);
+
+	if (!bp) {
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+	bios_parser2_destruct(bp);
+
+	kfree(bp);
+	*dcb = NULL;
+}
+
+static void get_atom_data_table_revision(
+	struct atom_common_table_header *atom_data_tbl,
+	struct atom_data_revision *tbl_revision)
+{
+	if (!tbl_revision)
+		return;
+
+	/* initialize the revision to 0 which is invalid revision */
+	tbl_revision->major = 0;
+	tbl_revision->minor = 0;
+
+	if (!atom_data_tbl)
+		return;
+
+	tbl_revision->major =
+			(uint32_t) atom_data_tbl->format_revision & 0x3f;
+	tbl_revision->minor =
+			(uint32_t) atom_data_tbl->content_revision & 0x3f;
+}
+
+/* BIOS oject table displaypath is per connector.
+ * There is extra path not for connector. BIOS fill its encoderid as 0
+ */
+static uint8_t bios_parser_get_connectors_number(struct dc_bios *dcb)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	unsigned int count = 0;
+	unsigned int i;
+
+	switch (bp->object_info_tbl.revision.minor) {
+	default:
+	case 4:
+		for (i = 0; i < bp->object_info_tbl.v1_4->number_of_path; i++)
+			if (bp->object_info_tbl.v1_4->display_path[i].encoderobjid != 0)
+				count++;
+
+		break;
+
+	case 5:
+		for (i = 0; i < bp->object_info_tbl.v1_5->number_of_path; i++)
+			if (bp->object_info_tbl.v1_5->display_path[i].encoderobjid != 0)
+				count++;
+
+		break;
+	}
+	return count;
+}
+
+static struct graphics_object_id bios_parser_get_connector_id(
+	struct dc_bios *dcb,
+	uint8_t i)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	struct graphics_object_id object_id = dal_graphics_object_id_init(
+		0, ENUM_ID_UNKNOWN, OBJECT_TYPE_UNKNOWN);
+	struct object_info_table *tbl = &bp->object_info_tbl;
+	struct display_object_info_table_v1_4 *v1_4 = tbl->v1_4;
+
+	struct display_object_info_table_v1_5 *v1_5 = tbl->v1_5;
+
+	switch (bp->object_info_tbl.revision.minor) {
+	default:
+	case 4:
+		if (v1_4->number_of_path > i) {
+			/* If display_objid is generic object id,  the encoderObj
+			 * /extencoderobjId should be 0
+			 */
+			if (v1_4->display_path[i].encoderobjid != 0 &&
+			    v1_4->display_path[i].display_objid != 0)
+				object_id = object_id_from_bios_object_id(
+					v1_4->display_path[i].display_objid);
+		}
+		break;
+
+	case 5:
+		if (v1_5->number_of_path > i) {
+			/* If display_objid is generic object id,  the encoderObjId
+		 * should be 0
+		 */
+			if (v1_5->display_path[i].encoderobjid != 0 &&
+			    v1_5->display_path[i].display_objid != 0)
+				object_id = object_id_from_bios_object_id(
+					v1_5->display_path[i].display_objid);
+		}
+		break;
+	}
+	return object_id;
+}
+
+static enum bp_result bios_parser_get_src_obj(struct dc_bios *dcb,
+	struct graphics_object_id object_id, uint32_t index,
+	struct graphics_object_id *src_object_id)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	unsigned int i;
+	enum bp_result bp_result = BP_RESULT_BADINPUT;
+	struct graphics_object_id obj_id = { 0 };
+	struct object_info_table *tbl = &bp->object_info_tbl;
+
+	if (!src_object_id)
+		return bp_result;
+
+	switch (object_id.type) {
+	/* Encoder's Source is GPU.  BIOS does not provide GPU, since all
+	 * displaypaths point to same GPU (0x1100).  Hardcode GPU object type
+	 */
+	case OBJECT_TYPE_ENCODER:
+		/* TODO: since num of src must be less than 2.
+		 * If found in for loop, should break.
+		 * DAL2 implementation may be changed too
+		 */
+		switch (bp->object_info_tbl.revision.minor) {
+		default:
+		case 4:
+			for (i = 0; i < tbl->v1_4->number_of_path; i++) {
+				obj_id = object_id_from_bios_object_id(
+					tbl->v1_4->display_path[i].encoderobjid);
+				if (object_id.type == obj_id.type &&
+				    object_id.id == obj_id.id &&
+				    object_id.enum_id == obj_id.enum_id) {
+					*src_object_id =
+						object_id_from_bios_object_id(
+							0x1100);
+					/* break; */
+				}
+			}
+			bp_result = BP_RESULT_OK;
+			break;
+
+		case 5:
+			for (i = 0; i < tbl->v1_5->number_of_path; i++) {
+				obj_id = object_id_from_bios_object_id(
+					tbl->v1_5->display_path[i].encoderobjid);
+				if (object_id.type == obj_id.type &&
+				    object_id.id == obj_id.id &&
+				    object_id.enum_id == obj_id.enum_id) {
+					*src_object_id =
+						object_id_from_bios_object_id(
+							0x1100);
+					/* break; */
+				}
+			}
+			bp_result = BP_RESULT_OK;
+			break;
+		}
+		break;
+	case OBJECT_TYPE_CONNECTOR:
+		switch (bp->object_info_tbl.revision.minor) {
+		default:
+		case 4:
+			for (i = 0; i < tbl->v1_4->number_of_path; i++) {
+				obj_id = object_id_from_bios_object_id(
+					tbl->v1_4->display_path[i]
+						.display_objid);
+
+				if (object_id.type == obj_id.type &&
+				    object_id.id == obj_id.id &&
+				    object_id.enum_id == obj_id.enum_id) {
+					*src_object_id =
+						object_id_from_bios_object_id(
+							tbl->v1_4
+								->display_path[i]
+								.encoderobjid);
+					/* break; */
+				}
+			}
+			bp_result = BP_RESULT_OK;
+			break;
+		}
+		bp_result = BP_RESULT_OK;
+		break;
+		case 5:
+			for (i = 0; i < tbl->v1_5->number_of_path; i++) {
+				obj_id = object_id_from_bios_object_id(
+								       tbl->v1_5->display_path[i].display_objid);
+
+				if (object_id.type == obj_id.type &&
+				    object_id.id == obj_id.id &&
+				    object_id.enum_id == obj_id.enum_id) {
+					*src_object_id = object_id_from_bios_object_id(
+										       tbl->v1_5->display_path[i].encoderobjid);
+					/* break; */
+				}
+			}
+		bp_result = BP_RESULT_OK;
+		break;
+
+	default:
+		bp_result = BP_RESULT_OK;
+		break;
+	}
+
+	return bp_result;
+}
+
+/* from graphics_object_id, find display path which includes the object_id */
+static struct atom_display_object_path_v2 *get_bios_object(
+		struct bios_parser *bp,
+		struct graphics_object_id id)
+{
+	unsigned int i;
+	struct graphics_object_id obj_id = {0};
+
+	switch (id.type) {
+	case OBJECT_TYPE_ENCODER:
+		for (i = 0; i < bp->object_info_tbl.v1_4->number_of_path; i++) {
+			obj_id = object_id_from_bios_object_id(
+					bp->object_info_tbl.v1_4->display_path[i].encoderobjid);
+			if (id.type == obj_id.type && id.id == obj_id.id
+					&& id.enum_id == obj_id.enum_id)
+				return &bp->object_info_tbl.v1_4->display_path[i];
+		}
+		fallthrough;
+	case OBJECT_TYPE_CONNECTOR:
+	case OBJECT_TYPE_GENERIC:
+		/* Both Generic and Connector Object ID
+		 * will be stored on display_objid
+		 */
+		for (i = 0; i < bp->object_info_tbl.v1_4->number_of_path; i++) {
+			obj_id = object_id_from_bios_object_id(
+					bp->object_info_tbl.v1_4->display_path[i].display_objid);
+			if (id.type == obj_id.type && id.id == obj_id.id
+					&& id.enum_id == obj_id.enum_id)
+				return &bp->object_info_tbl.v1_4->display_path[i];
+		}
+		fallthrough;
+	default:
+		return NULL;
+	}
+}
+
+/* from graphics_object_id, find display path which includes the object_id */
+static struct atom_display_object_path_v3 *get_bios_object_from_path_v3(struct bios_parser *bp,
+									struct graphics_object_id id)
+{
+	unsigned int i;
+	struct graphics_object_id obj_id = {0};
+
+	switch (id.type) {
+	case OBJECT_TYPE_ENCODER:
+		for (i = 0; i < bp->object_info_tbl.v1_5->number_of_path; i++) {
+			obj_id = object_id_from_bios_object_id(
+					bp->object_info_tbl.v1_5->display_path[i].encoderobjid);
+			if (id.type == obj_id.type && id.id == obj_id.id
+					&& id.enum_id == obj_id.enum_id)
+				return &bp->object_info_tbl.v1_5->display_path[i];
+		}
+	break;
+
+	case OBJECT_TYPE_CONNECTOR:
+	case OBJECT_TYPE_GENERIC:
+		/* Both Generic and Connector Object ID
+		 * will be stored on display_objid
+		 */
+		for (i = 0; i < bp->object_info_tbl.v1_5->number_of_path; i++) {
+			obj_id = object_id_from_bios_object_id(
+					bp->object_info_tbl.v1_5->display_path[i].display_objid);
+			if (id.type == obj_id.type && id.id == obj_id.id
+					&& id.enum_id == obj_id.enum_id)
+				return &bp->object_info_tbl.v1_5->display_path[i];
+		}
+	break;
+
+	default:
+		return NULL;
+	}
+
+	return NULL;
+}
+
+static enum bp_result bios_parser_get_i2c_info(struct dc_bios *dcb,
+	struct graphics_object_id id,
+	struct graphics_object_i2c_info *info)
+{
+	uint32_t offset;
+	struct atom_display_object_path_v2 *object;
+
+	struct atom_display_object_path_v3 *object_path_v3;
+
+	struct atom_common_record_header *header;
+	struct atom_i2c_record *record;
+	struct atom_i2c_record dummy_record = {0};
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	if (!info)
+		return BP_RESULT_BADINPUT;
+
+	if (id.type == OBJECT_TYPE_GENERIC) {
+		dummy_record.i2c_id = id.id;
+
+		if (get_gpio_i2c_info(bp, &dummy_record, info) == BP_RESULT_OK)
+			return BP_RESULT_OK;
+		else
+			return BP_RESULT_NORECORD;
+	}
+
+	switch (bp->object_info_tbl.revision.minor) {
+	case 4:
+	default:
+		object = get_bios_object(bp, id);
+
+		if (!object)
+			return BP_RESULT_BADINPUT;
+
+		offset = object->disp_recordoffset + bp->object_info_tbl_offset;
+		break;
+	case 5:
+		object_path_v3 = get_bios_object_from_path_v3(bp, id);
+
+		if (!object_path_v3)
+			return BP_RESULT_BADINPUT;
+
+		offset = object_path_v3->disp_recordoffset + bp->object_info_tbl_offset;
+		break;
+	}
+
+	for (;;) {
+		header = GET_IMAGE(struct atom_common_record_header, offset);
+
+		if (!header)
+			return BP_RESULT_BADBIOSTABLE;
+
+		if (header->record_type == LAST_RECORD_TYPE ||
+			!header->record_size)
+			break;
+
+		if (header->record_type == ATOM_I2C_RECORD_TYPE
+			&& sizeof(struct atom_i2c_record) <=
+							header->record_size) {
+			/* get the I2C info */
+			record = (struct atom_i2c_record *) header;
+
+			if (get_gpio_i2c_info(bp, record, info) ==
+								BP_RESULT_OK)
+				return BP_RESULT_OK;
+		}
+
+		offset += header->record_size;
+	}
+
+	return BP_RESULT_NORECORD;
+}
+
+static enum bp_result get_gpio_i2c_info(
+	struct bios_parser *bp,
+	struct atom_i2c_record *record,
+	struct graphics_object_i2c_info *info)
+{
+	struct atom_gpio_pin_lut_v2_1 *header;
+	uint32_t count = 0;
+	unsigned int table_index = 0;
+	bool find_valid = false;
+	struct atom_gpio_pin_assignment *pin;
+
+	if (!info)
+		return BP_RESULT_BADINPUT;
+
+	/* get the GPIO_I2C info */
+	if (!DATA_TABLES(gpio_pin_lut))
+		return BP_RESULT_BADBIOSTABLE;
+
+	header = GET_IMAGE(struct atom_gpio_pin_lut_v2_1,
+					DATA_TABLES(gpio_pin_lut));
+	if (!header)
+		return BP_RESULT_BADBIOSTABLE;
+
+	if (sizeof(struct atom_common_table_header) +
+			sizeof(struct atom_gpio_pin_assignment)	>
+			le16_to_cpu(header->table_header.structuresize))
+		return BP_RESULT_BADBIOSTABLE;
+
+	/* TODO: is version change? */
+	if (header->table_header.content_revision != 1)
+		return BP_RESULT_UNSUPPORTED;
+
+	/* get data count */
+	count = (le16_to_cpu(header->table_header.structuresize)
+			- sizeof(struct atom_common_table_header))
+				/ sizeof(struct atom_gpio_pin_assignment);
+
+	pin = (struct atom_gpio_pin_assignment *) header->gpio_pin;
+
+	for (table_index = 0; table_index < count; table_index++) {
+		if (((record->i2c_id & I2C_HW_CAP) 				== (pin->gpio_id & I2C_HW_CAP)) &&
+		    ((record->i2c_id & I2C_HW_ENGINE_ID_MASK)	== (pin->gpio_id & I2C_HW_ENGINE_ID_MASK)) &&
+		    ((record->i2c_id & I2C_HW_LANE_MUX) 		== (pin->gpio_id & I2C_HW_LANE_MUX))) {
+			/* still valid */
+			find_valid = true;
+			break;
+		}
+		pin = (struct atom_gpio_pin_assignment *)((uint8_t *)pin + sizeof(struct atom_gpio_pin_assignment));
+	}
+
+	/* If we don't find the entry that we are looking for then
+	 *  we will return BP_Result_BadBiosTable.
+	 */
+	if (find_valid == false)
+		return BP_RESULT_BADBIOSTABLE;
+
+	/* get the GPIO_I2C_INFO */
+	info->i2c_hw_assist = (record->i2c_id & I2C_HW_CAP) ? true : false;
+	info->i2c_line = record->i2c_id & I2C_HW_LANE_MUX;
+	info->i2c_engine_id = (record->i2c_id & I2C_HW_ENGINE_ID_MASK) >> 4;
+	info->i2c_slave_address = record->i2c_slave_addr;
+
+	/* TODO: check how to get register offset for en, Y, etc. */
+	info->gpio_info.clk_a_register_index = le16_to_cpu(pin->data_a_reg_index);
+	info->gpio_info.clk_a_shift = pin->gpio_bitshift;
+
+	return BP_RESULT_OK;
+}
+
+static struct atom_hpd_int_record *get_hpd_record_for_path_v3(struct bios_parser *bp,
+							      struct atom_display_object_path_v3 *object)
+{
+	struct atom_common_record_header *header;
+	uint32_t offset;
+
+	if (!object) {
+		BREAK_TO_DEBUGGER(); /* Invalid object */
+		return NULL;
+	}
+
+	offset = object->disp_recordoffset + bp->object_info_tbl_offset;
+
+	for (;;) {
+		header = GET_IMAGE(struct atom_common_record_header, offset);
+
+		if (!header)
+			return NULL;
+
+		if (header->record_type == ATOM_RECORD_END_TYPE ||
+			!header->record_size)
+			break;
+
+		if (header->record_type == ATOM_HPD_INT_RECORD_TYPE
+			&& sizeof(struct atom_hpd_int_record) <=
+							header->record_size)
+			return (struct atom_hpd_int_record *) header;
+
+		offset += header->record_size;
+	}
+
+	return NULL;
+}
+
+static enum bp_result bios_parser_get_hpd_info(
+	struct dc_bios *dcb,
+	struct graphics_object_id id,
+	struct graphics_object_hpd_info *info)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	struct atom_display_object_path_v2 *object;
+	struct atom_display_object_path_v3 *object_path_v3;
+	struct atom_hpd_int_record *record = NULL;
+
+	if (!info)
+		return BP_RESULT_BADINPUT;
+
+	switch (bp->object_info_tbl.revision.minor) {
+	case 4:
+	default:
+		object = get_bios_object(bp, id);
+
+		if (!object)
+			return BP_RESULT_BADINPUT;
+
+		record = get_hpd_record(bp, object);
+		break;
+	case 5:
+		object_path_v3 = get_bios_object_from_path_v3(bp, id);
+
+		if (!object_path_v3)
+			return BP_RESULT_BADINPUT;
+
+		record = get_hpd_record_for_path_v3(bp, object_path_v3);
+		break;
+	}
+
+	if (record != NULL) {
+		info->hpd_int_gpio_uid = record->pin_id;
+		info->hpd_active = record->plugin_pin_state;
+		return BP_RESULT_OK;
+	}
+
+	return BP_RESULT_NORECORD;
+}
+
+static struct atom_hpd_int_record *get_hpd_record(
+	struct bios_parser *bp,
+	struct atom_display_object_path_v2 *object)
+{
+	struct atom_common_record_header *header;
+	uint32_t offset;
+
+	if (!object) {
+		BREAK_TO_DEBUGGER(); /* Invalid object */
+		return NULL;
+	}
+
+	offset = le16_to_cpu(object->disp_recordoffset)
+			+ bp->object_info_tbl_offset;
+
+	for (;;) {
+		header = GET_IMAGE(struct atom_common_record_header, offset);
+
+		if (!header)
+			return NULL;
+
+		if (header->record_type == LAST_RECORD_TYPE ||
+			!header->record_size)
+			break;
+
+		if (header->record_type == ATOM_HPD_INT_RECORD_TYPE
+			&& sizeof(struct atom_hpd_int_record) <=
+							header->record_size)
+			return (struct atom_hpd_int_record *) header;
+
+		offset += header->record_size;
+	}
+
+	return NULL;
+}
+
+/**
+ * bios_parser_get_gpio_pin_info
+ * Get GpioPin information of input gpio id
+ *
+ * @dcb:     pointer to the DC BIOS
+ * @gpio_id: GPIO ID
+ * @info:    GpioPin information structure
+ * return: Bios parser result code
+ * note:
+ *  to get the GPIO PIN INFO, we need:
+ *  1. get the GPIO_ID from other object table, see GetHPDInfo()
+ *  2. in DATA_TABLE.GPIO_Pin_LUT, search all records,
+ *	to get the registerA  offset/mask
+ */
+static enum bp_result bios_parser_get_gpio_pin_info(
+	struct dc_bios *dcb,
+	uint32_t gpio_id,
+	struct gpio_pin_info *info)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	struct atom_gpio_pin_lut_v2_1 *header;
+	uint32_t count = 0;
+	uint32_t i = 0;
+
+	if (!DATA_TABLES(gpio_pin_lut))
+		return BP_RESULT_BADBIOSTABLE;
+
+	header = GET_IMAGE(struct atom_gpio_pin_lut_v2_1,
+						DATA_TABLES(gpio_pin_lut));
+	if (!header)
+		return BP_RESULT_BADBIOSTABLE;
+
+	if (sizeof(struct atom_common_table_header) +
+			sizeof(struct atom_gpio_pin_assignment)
+			> le16_to_cpu(header->table_header.structuresize))
+		return BP_RESULT_BADBIOSTABLE;
+
+	if (header->table_header.content_revision != 1)
+		return BP_RESULT_UNSUPPORTED;
+
+	/* Temporary hard code gpio pin info */
+	count = (le16_to_cpu(header->table_header.structuresize)
+			- sizeof(struct atom_common_table_header))
+				/ sizeof(struct atom_gpio_pin_assignment);
+	for (i = 0; i < count; ++i) {
+		if (header->gpio_pin[i].gpio_id != gpio_id)
+			continue;
+
+		info->offset =
+			(uint32_t) le16_to_cpu(
+					header->gpio_pin[i].data_a_reg_index);
+		info->offset_y = info->offset + 2;
+		info->offset_en = info->offset + 1;
+		info->offset_mask = info->offset - 1;
+
+		info->mask = (uint32_t) (1 <<
+			header->gpio_pin[i].gpio_bitshift);
+		info->mask_y = info->mask + 2;
+		info->mask_en = info->mask + 1;
+		info->mask_mask = info->mask - 1;
+
+		return BP_RESULT_OK;
+	}
+
+	return BP_RESULT_NORECORD;
+}
+
+static struct device_id device_type_from_device_id(uint16_t device_id)
+{
+
+	struct device_id result_device_id;
+
+	result_device_id.raw_device_tag = device_id;
+
+	switch (device_id) {
+	case ATOM_DISPLAY_LCD1_SUPPORT:
+		result_device_id.device_type = DEVICE_TYPE_LCD;
+		result_device_id.enum_id = 1;
+		break;
+
+	case ATOM_DISPLAY_LCD2_SUPPORT:
+		result_device_id.device_type = DEVICE_TYPE_LCD;
+		result_device_id.enum_id = 2;
+		break;
+
+	case ATOM_DISPLAY_DFP1_SUPPORT:
+		result_device_id.device_type = DEVICE_TYPE_DFP;
+		result_device_id.enum_id = 1;
+		break;
+
+	case ATOM_DISPLAY_DFP2_SUPPORT:
+		result_device_id.device_type = DEVICE_TYPE_DFP;
+		result_device_id.enum_id = 2;
+		break;
+
+	case ATOM_DISPLAY_DFP3_SUPPORT:
+		result_device_id.device_type = DEVICE_TYPE_DFP;
+		result_device_id.enum_id = 3;
+		break;
+
+	case ATOM_DISPLAY_DFP4_SUPPORT:
+		result_device_id.device_type = DEVICE_TYPE_DFP;
+		result_device_id.enum_id = 4;
+		break;
+
+	case ATOM_DISPLAY_DFP5_SUPPORT:
+		result_device_id.device_type = DEVICE_TYPE_DFP;
+		result_device_id.enum_id = 5;
+		break;
+
+	case ATOM_DISPLAY_DFP6_SUPPORT:
+		result_device_id.device_type = DEVICE_TYPE_DFP;
+		result_device_id.enum_id = 6;
+		break;
+
+	default:
+		BREAK_TO_DEBUGGER(); /* Invalid device Id */
+		result_device_id.device_type = DEVICE_TYPE_UNKNOWN;
+		result_device_id.enum_id = 0;
+	}
+	return result_device_id;
+}
+
+static enum bp_result bios_parser_get_device_tag(
+	struct dc_bios *dcb,
+	struct graphics_object_id connector_object_id,
+	uint32_t device_tag_index,
+	struct connector_device_tag_info *info)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	struct atom_display_object_path_v2 *object;
+
+	struct atom_display_object_path_v3 *object_path_v3;
+
+
+	if (!info)
+		return BP_RESULT_BADINPUT;
+
+	switch (bp->object_info_tbl.revision.minor) {
+	case 4:
+	default:
+	        /* getBiosObject will return MXM object */
+		object = get_bios_object(bp, connector_object_id);
+
+		if (!object) {
+			BREAK_TO_DEBUGGER(); /* Invalid object id */
+			return BP_RESULT_BADINPUT;
+		}
+
+		info->acpi_device = 0; /* BIOS no longer provides this */
+		info->dev_id = device_type_from_device_id(object->device_tag);
+		break;
+	case 5:
+		object_path_v3 = get_bios_object_from_path_v3(bp, connector_object_id);
+
+		if (!object_path_v3) {
+			BREAK_TO_DEBUGGER(); /* Invalid object id */
+			return BP_RESULT_BADINPUT;
+		}
+		info->acpi_device = 0; /* BIOS no longer provides this */
+		info->dev_id = device_type_from_device_id(object_path_v3->device_tag);
+		break;
+	}
+
+	return BP_RESULT_OK;
+}
+
+static enum bp_result get_ss_info_v4_1(
+	struct bios_parser *bp,
+	uint32_t id,
+	uint32_t index,
+	struct spread_spectrum_info *ss_info)
+{
+	enum bp_result result = BP_RESULT_OK;
+	struct atom_display_controller_info_v4_1 *disp_cntl_tbl = NULL;
+	struct atom_smu_info_v3_3 *smu_info = NULL;
+
+	if (!ss_info)
+		return BP_RESULT_BADINPUT;
+
+	if (!DATA_TABLES(dce_info))
+		return BP_RESULT_BADBIOSTABLE;
+
+	disp_cntl_tbl =  GET_IMAGE(struct atom_display_controller_info_v4_1,
+							DATA_TABLES(dce_info));
+	if (!disp_cntl_tbl)
+		return BP_RESULT_BADBIOSTABLE;
+
+
+	ss_info->type.STEP_AND_DELAY_INFO = false;
+	ss_info->spread_percentage_divider = 1000;
+	/* BIOS no longer uses target clock.  Always enable for now */
+	ss_info->target_clock_range = 0xffffffff;
+
+	switch (id) {
+	case AS_SIGNAL_TYPE_DVI:
+		ss_info->spread_spectrum_percentage =
+				disp_cntl_tbl->dvi_ss_percentage;
+		ss_info->spread_spectrum_range =
+				disp_cntl_tbl->dvi_ss_rate_10hz * 10;
+		if (disp_cntl_tbl->dvi_ss_mode & ATOM_SS_CENTRE_SPREAD_MODE)
+			ss_info->type.CENTER_MODE = true;
+
+		DC_LOG_BIOS("AS_SIGNAL_TYPE_DVI ss_percentage: %d\n", ss_info->spread_spectrum_percentage);
+		break;
+	case AS_SIGNAL_TYPE_HDMI:
+		ss_info->spread_spectrum_percentage =
+				disp_cntl_tbl->hdmi_ss_percentage;
+		ss_info->spread_spectrum_range =
+				disp_cntl_tbl->hdmi_ss_rate_10hz * 10;
+		if (disp_cntl_tbl->hdmi_ss_mode & ATOM_SS_CENTRE_SPREAD_MODE)
+			ss_info->type.CENTER_MODE = true;
+
+		DC_LOG_BIOS("AS_SIGNAL_TYPE_HDMI ss_percentage: %d\n", ss_info->spread_spectrum_percentage);
+		break;
+	/* TODO LVDS not support anymore? */
+	case AS_SIGNAL_TYPE_DISPLAY_PORT:
+		ss_info->spread_spectrum_percentage =
+				disp_cntl_tbl->dp_ss_percentage;
+		ss_info->spread_spectrum_range =
+				disp_cntl_tbl->dp_ss_rate_10hz * 10;
+		if (disp_cntl_tbl->dp_ss_mode & ATOM_SS_CENTRE_SPREAD_MODE)
+			ss_info->type.CENTER_MODE = true;
+
+		DC_LOG_BIOS("AS_SIGNAL_TYPE_DISPLAY_PORT ss_percentage: %d\n", ss_info->spread_spectrum_percentage);
+		break;
+	case AS_SIGNAL_TYPE_GPU_PLL:
+		/* atom_firmware: DAL only get data from dce_info table.
+		 * if data within smu_info is needed for DAL, VBIOS should
+		 * copy it into dce_info
+		 */
+		result = BP_RESULT_UNSUPPORTED;
+		break;
+	case AS_SIGNAL_TYPE_XGMI:
+		smu_info =  GET_IMAGE(struct atom_smu_info_v3_3,
+				      DATA_TABLES(smu_info));
+		if (!smu_info)
+			return BP_RESULT_BADBIOSTABLE;
+		DC_LOG_BIOS("gpuclk_ss_percentage (unit of 0.001 percent): %d\n", smu_info->gpuclk_ss_percentage);
+		ss_info->spread_spectrum_percentage =
+				smu_info->waflclk_ss_percentage;
+		ss_info->spread_spectrum_range =
+				smu_info->gpuclk_ss_rate_10hz * 10;
+		if (smu_info->waflclk_ss_mode & ATOM_SS_CENTRE_SPREAD_MODE)
+			ss_info->type.CENTER_MODE = true;
+
+		DC_LOG_BIOS("AS_SIGNAL_TYPE_XGMI ss_percentage: %d\n", ss_info->spread_spectrum_percentage);
+		break;
+	default:
+		result = BP_RESULT_UNSUPPORTED;
+	}
+
+	return result;
+}
+
+static enum bp_result get_ss_info_v4_2(
+	struct bios_parser *bp,
+	uint32_t id,
+	uint32_t index,
+	struct spread_spectrum_info *ss_info)
+{
+	enum bp_result result = BP_RESULT_OK;
+	struct atom_display_controller_info_v4_2 *disp_cntl_tbl = NULL;
+	struct atom_smu_info_v3_1 *smu_info = NULL;
+
+	if (!ss_info)
+		return BP_RESULT_BADINPUT;
+
+	if (!DATA_TABLES(dce_info))
+		return BP_RESULT_BADBIOSTABLE;
+
+	if (!DATA_TABLES(smu_info))
+		return BP_RESULT_BADBIOSTABLE;
+
+	disp_cntl_tbl =  GET_IMAGE(struct atom_display_controller_info_v4_2,
+							DATA_TABLES(dce_info));
+	if (!disp_cntl_tbl)
+		return BP_RESULT_BADBIOSTABLE;
+
+	smu_info =  GET_IMAGE(struct atom_smu_info_v3_1, DATA_TABLES(smu_info));
+	if (!smu_info)
+		return BP_RESULT_BADBIOSTABLE;
+
+	DC_LOG_BIOS("gpuclk_ss_percentage (unit of 0.001 percent): %d\n", smu_info->gpuclk_ss_percentage);
+	ss_info->type.STEP_AND_DELAY_INFO = false;
+	ss_info->spread_percentage_divider = 1000;
+	/* BIOS no longer uses target clock.  Always enable for now */
+	ss_info->target_clock_range = 0xffffffff;
+
+	switch (id) {
+	case AS_SIGNAL_TYPE_DVI:
+		ss_info->spread_spectrum_percentage =
+				disp_cntl_tbl->dvi_ss_percentage;
+		ss_info->spread_spectrum_range =
+				disp_cntl_tbl->dvi_ss_rate_10hz * 10;
+		if (disp_cntl_tbl->dvi_ss_mode & ATOM_SS_CENTRE_SPREAD_MODE)
+			ss_info->type.CENTER_MODE = true;
+
+		DC_LOG_BIOS("AS_SIGNAL_TYPE_DVI ss_percentage: %d\n", ss_info->spread_spectrum_percentage);
+		break;
+	case AS_SIGNAL_TYPE_HDMI:
+		ss_info->spread_spectrum_percentage =
+				disp_cntl_tbl->hdmi_ss_percentage;
+		ss_info->spread_spectrum_range =
+				disp_cntl_tbl->hdmi_ss_rate_10hz * 10;
+		if (disp_cntl_tbl->hdmi_ss_mode & ATOM_SS_CENTRE_SPREAD_MODE)
+			ss_info->type.CENTER_MODE = true;
+
+		DC_LOG_BIOS("AS_SIGNAL_TYPE_HDMI ss_percentage: %d\n", ss_info->spread_spectrum_percentage);
+		break;
+	/* TODO LVDS not support anymore? */
+	case AS_SIGNAL_TYPE_DISPLAY_PORT:
+		ss_info->spread_spectrum_percentage =
+				smu_info->gpuclk_ss_percentage;
+		ss_info->spread_spectrum_range =
+				smu_info->gpuclk_ss_rate_10hz * 10;
+		if (smu_info->gpuclk_ss_mode & ATOM_SS_CENTRE_SPREAD_MODE)
+			ss_info->type.CENTER_MODE = true;
+
+		DC_LOG_BIOS("AS_SIGNAL_TYPE_DISPLAY_PORT ss_percentage: %d\n", ss_info->spread_spectrum_percentage);
+		break;
+	case AS_SIGNAL_TYPE_GPU_PLL:
+		/* atom_firmware: DAL only get data from dce_info table.
+		 * if data within smu_info is needed for DAL, VBIOS should
+		 * copy it into dce_info
+		 */
+		result = BP_RESULT_UNSUPPORTED;
+		break;
+	default:
+		result = BP_RESULT_UNSUPPORTED;
+	}
+
+	return result;
+}
+
+static enum bp_result get_ss_info_v4_5(
+	struct bios_parser *bp,
+	uint32_t id,
+	uint32_t index,
+	struct spread_spectrum_info *ss_info)
+{
+	enum bp_result result = BP_RESULT_OK;
+	struct atom_display_controller_info_v4_5 *disp_cntl_tbl = NULL;
+
+	if (!ss_info)
+		return BP_RESULT_BADINPUT;
+
+	if (!DATA_TABLES(dce_info))
+		return BP_RESULT_BADBIOSTABLE;
+
+	disp_cntl_tbl =  GET_IMAGE(struct atom_display_controller_info_v4_5,
+							DATA_TABLES(dce_info));
+	if (!disp_cntl_tbl)
+		return BP_RESULT_BADBIOSTABLE;
+
+	ss_info->type.STEP_AND_DELAY_INFO = false;
+	ss_info->spread_percentage_divider = 1000;
+	/* BIOS no longer uses target clock.  Always enable for now */
+	ss_info->target_clock_range = 0xffffffff;
+
+	switch (id) {
+	case AS_SIGNAL_TYPE_DVI:
+		ss_info->spread_spectrum_percentage =
+				disp_cntl_tbl->dvi_ss_percentage;
+		ss_info->spread_spectrum_range =
+				disp_cntl_tbl->dvi_ss_rate_10hz * 10;
+		if (disp_cntl_tbl->dvi_ss_mode & ATOM_SS_CENTRE_SPREAD_MODE)
+			ss_info->type.CENTER_MODE = true;
+
+		DC_LOG_BIOS("AS_SIGNAL_TYPE_DVI ss_percentage: %d\n", ss_info->spread_spectrum_percentage);
+		break;
+	case AS_SIGNAL_TYPE_HDMI:
+		ss_info->spread_spectrum_percentage =
+				disp_cntl_tbl->hdmi_ss_percentage;
+		ss_info->spread_spectrum_range =
+				disp_cntl_tbl->hdmi_ss_rate_10hz * 10;
+		if (disp_cntl_tbl->hdmi_ss_mode & ATOM_SS_CENTRE_SPREAD_MODE)
+			ss_info->type.CENTER_MODE = true;
+
+		DC_LOG_BIOS("AS_SIGNAL_TYPE_HDMI ss_percentage: %d\n", ss_info->spread_spectrum_percentage);
+		break;
+	case AS_SIGNAL_TYPE_DISPLAY_PORT:
+		if (bp->base.integrated_info) {
+			DC_LOG_BIOS("gpuclk_ss_percentage (unit of 0.001 percent): %d\n", bp->base.integrated_info->gpuclk_ss_percentage);
+			ss_info->spread_spectrum_percentage =
+					bp->base.integrated_info->gpuclk_ss_percentage;
+			ss_info->type.CENTER_MODE =
+					bp->base.integrated_info->gpuclk_ss_type;
+		} else {
+			ss_info->spread_spectrum_percentage =
+				disp_cntl_tbl->dp_ss_percentage;
+			ss_info->spread_spectrum_range =
+				disp_cntl_tbl->dp_ss_rate_10hz * 10;
+			if (disp_cntl_tbl->dp_ss_mode & ATOM_SS_CENTRE_SPREAD_MODE)
+				ss_info->type.CENTER_MODE = true;
+		}
+		DC_LOG_BIOS("AS_SIGNAL_TYPE_DISPLAY_PORT ss_percentage: %d\n", ss_info->spread_spectrum_percentage);
+		break;
+	case AS_SIGNAL_TYPE_GPU_PLL:
+		/* atom_smu_info_v4_0 does not have fields for SS for SMU Display PLL anymore.
+		 * SMU Display PLL supposed to be without spread.
+		 * Better place for it would be in atom_display_controller_info_v4_5 table.
+		 */
+		result = BP_RESULT_UNSUPPORTED;
+		break;
+	default:
+		result = BP_RESULT_UNSUPPORTED;
+		break;
+	}
+
+	return result;
+}
+
+/**
+ * bios_parser_get_spread_spectrum_info
+ * Get spread spectrum information from the ASIC_InternalSS_Info(ver 2.1 or
+ * ver 3.1) or SS_Info table from the VBIOS. Currently ASIC_InternalSS_Info
+ * ver 2.1 can co-exist with SS_Info table. Expect ASIC_InternalSS_Info
+ * ver 3.1,
+ * there is only one entry for each signal /ss id.  However, there is
+ * no planning of supporting multiple spread Sprectum entry for EverGreen
+ * @dcb:     pointer to the DC BIOS
+ * @signal:  ASSignalType to be converted to info index
+ * @index:   number of entries that match the converted info index
+ * @ss_info: sprectrum information structure,
+ * return: Bios parser result code
+ */
+static enum bp_result bios_parser_get_spread_spectrum_info(
+	struct dc_bios *dcb,
+	enum as_signal_type signal,
+	uint32_t index,
+	struct spread_spectrum_info *ss_info)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	enum bp_result result = BP_RESULT_UNSUPPORTED;
+	struct atom_common_table_header *header;
+	struct atom_data_revision tbl_revision;
+
+	if (!ss_info) /* check for bad input */
+		return BP_RESULT_BADINPUT;
+
+	if (!DATA_TABLES(dce_info))
+		return BP_RESULT_UNSUPPORTED;
+
+	header = GET_IMAGE(struct atom_common_table_header,
+						DATA_TABLES(dce_info));
+	get_atom_data_table_revision(header, &tbl_revision);
+
+	switch (tbl_revision.major) {
+	case 4:
+		switch (tbl_revision.minor) {
+		case 1:
+			return get_ss_info_v4_1(bp, signal, index, ss_info);
+		case 2:
+		case 3:
+		case 4:
+			return get_ss_info_v4_2(bp, signal, index, ss_info);
+		case 5:
+			return get_ss_info_v4_5(bp, signal, index, ss_info);
+
+		default:
+			ASSERT(0);
+			break;
+		}
+		break;
+	default:
+		break;
+	}
+	/* there can not be more then one entry for SS Info table */
+	return result;
+}
+
+static enum bp_result get_soc_bb_info_v4_4(
+	struct bios_parser *bp,
+	struct bp_soc_bb_info *soc_bb_info)
+{
+	enum bp_result result = BP_RESULT_OK;
+	struct atom_display_controller_info_v4_4 *disp_cntl_tbl = NULL;
+
+	if (!soc_bb_info)
+		return BP_RESULT_BADINPUT;
+
+	if (!DATA_TABLES(dce_info))
+		return BP_RESULT_BADBIOSTABLE;
+
+	if (!DATA_TABLES(smu_info))
+		return BP_RESULT_BADBIOSTABLE;
+
+	disp_cntl_tbl =  GET_IMAGE(struct atom_display_controller_info_v4_4,
+							DATA_TABLES(dce_info));
+	if (!disp_cntl_tbl)
+		return BP_RESULT_BADBIOSTABLE;
+
+	soc_bb_info->dram_clock_change_latency_100ns = disp_cntl_tbl->max_mclk_chg_lat;
+	soc_bb_info->dram_sr_enter_exit_latency_100ns = disp_cntl_tbl->max_sr_enter_exit_lat;
+	soc_bb_info->dram_sr_exit_latency_100ns = disp_cntl_tbl->max_sr_exit_lat;
+
+	return result;
+}
+
+static enum bp_result get_soc_bb_info_v4_5(
+	struct bios_parser *bp,
+	struct bp_soc_bb_info *soc_bb_info)
+{
+	enum bp_result result = BP_RESULT_OK;
+	struct atom_display_controller_info_v4_5 *disp_cntl_tbl = NULL;
+
+	if (!soc_bb_info)
+		return BP_RESULT_BADINPUT;
+
+	if (!DATA_TABLES(dce_info))
+		return BP_RESULT_BADBIOSTABLE;
+
+	disp_cntl_tbl =  GET_IMAGE(struct atom_display_controller_info_v4_5,
+							DATA_TABLES(dce_info));
+	if (!disp_cntl_tbl)
+		return BP_RESULT_BADBIOSTABLE;
+
+	soc_bb_info->dram_clock_change_latency_100ns = disp_cntl_tbl->max_mclk_chg_lat;
+	soc_bb_info->dram_sr_enter_exit_latency_100ns = disp_cntl_tbl->max_sr_enter_exit_lat;
+	soc_bb_info->dram_sr_exit_latency_100ns = disp_cntl_tbl->max_sr_exit_lat;
+
+	return result;
+}
+
+static enum bp_result bios_parser_get_soc_bb_info(
+	struct dc_bios *dcb,
+	struct bp_soc_bb_info *soc_bb_info)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	enum bp_result result = BP_RESULT_UNSUPPORTED;
+	struct atom_common_table_header *header;
+	struct atom_data_revision tbl_revision;
+
+	if (!soc_bb_info) /* check for bad input */
+		return BP_RESULT_BADINPUT;
+
+	if (!DATA_TABLES(dce_info))
+		return BP_RESULT_UNSUPPORTED;
+
+	header = GET_IMAGE(struct atom_common_table_header,
+						DATA_TABLES(dce_info));
+	get_atom_data_table_revision(header, &tbl_revision);
+
+	switch (tbl_revision.major) {
+	case 4:
+		switch (tbl_revision.minor) {
+		case 1:
+		case 2:
+		case 3:
+			break;
+		case 4:
+			result = get_soc_bb_info_v4_4(bp, soc_bb_info);
+			break;
+		case 5:
+			result = get_soc_bb_info_v4_5(bp, soc_bb_info);
+			break;
+		default:
+			break;
+		}
+		break;
+	default:
+		break;
+	}
+
+	return result;
+}
+
+static enum bp_result get_disp_caps_v4_1(
+	struct bios_parser *bp,
+	uint8_t *dce_caps)
+{
+	enum bp_result result = BP_RESULT_OK;
+	struct atom_display_controller_info_v4_1 *disp_cntl_tbl = NULL;
+
+	if (!dce_caps)
+		return BP_RESULT_BADINPUT;
+
+	if (!DATA_TABLES(dce_info))
+		return BP_RESULT_BADBIOSTABLE;
+
+	disp_cntl_tbl = GET_IMAGE(struct atom_display_controller_info_v4_1,
+							DATA_TABLES(dce_info));
+
+	if (!disp_cntl_tbl)
+		return BP_RESULT_BADBIOSTABLE;
+
+	*dce_caps = disp_cntl_tbl->display_caps;
+
+	return result;
+}
+
+static enum bp_result get_disp_caps_v4_2(
+	struct bios_parser *bp,
+	uint8_t *dce_caps)
+{
+	enum bp_result result = BP_RESULT_OK;
+	struct atom_display_controller_info_v4_2 *disp_cntl_tbl = NULL;
+
+	if (!dce_caps)
+		return BP_RESULT_BADINPUT;
+
+	if (!DATA_TABLES(dce_info))
+		return BP_RESULT_BADBIOSTABLE;
+
+	disp_cntl_tbl = GET_IMAGE(struct atom_display_controller_info_v4_2,
+							DATA_TABLES(dce_info));
+
+	if (!disp_cntl_tbl)
+		return BP_RESULT_BADBIOSTABLE;
+
+	*dce_caps = disp_cntl_tbl->display_caps;
+
+	return result;
+}
+
+static enum bp_result get_disp_caps_v4_3(
+	struct bios_parser *bp,
+	uint8_t *dce_caps)
+{
+	enum bp_result result = BP_RESULT_OK;
+	struct atom_display_controller_info_v4_3 *disp_cntl_tbl = NULL;
+
+	if (!dce_caps)
+		return BP_RESULT_BADINPUT;
+
+	if (!DATA_TABLES(dce_info))
+		return BP_RESULT_BADBIOSTABLE;
+
+	disp_cntl_tbl = GET_IMAGE(struct atom_display_controller_info_v4_3,
+							DATA_TABLES(dce_info));
+
+	if (!disp_cntl_tbl)
+		return BP_RESULT_BADBIOSTABLE;
+
+	*dce_caps = disp_cntl_tbl->display_caps;
+
+	return result;
+}
+
+static enum bp_result get_disp_caps_v4_4(
+	struct bios_parser *bp,
+	uint8_t *dce_caps)
+{
+	enum bp_result result = BP_RESULT_OK;
+	struct atom_display_controller_info_v4_4 *disp_cntl_tbl = NULL;
+
+	if (!dce_caps)
+		return BP_RESULT_BADINPUT;
+
+	if (!DATA_TABLES(dce_info))
+		return BP_RESULT_BADBIOSTABLE;
+
+	disp_cntl_tbl = GET_IMAGE(struct atom_display_controller_info_v4_4,
+							DATA_TABLES(dce_info));
+
+	if (!disp_cntl_tbl)
+		return BP_RESULT_BADBIOSTABLE;
+
+	*dce_caps = disp_cntl_tbl->display_caps;
+
+	return result;
+}
+
+static enum bp_result get_disp_caps_v4_5(
+	struct bios_parser *bp,
+	uint8_t *dce_caps)
+{
+	enum bp_result result = BP_RESULT_OK;
+	struct atom_display_controller_info_v4_5 *disp_cntl_tbl = NULL;
+
+	if (!dce_caps)
+		return BP_RESULT_BADINPUT;
+
+	if (!DATA_TABLES(dce_info))
+		return BP_RESULT_BADBIOSTABLE;
+
+	disp_cntl_tbl = GET_IMAGE(struct atom_display_controller_info_v4_5,
+							DATA_TABLES(dce_info));
+
+	if (!disp_cntl_tbl)
+		return BP_RESULT_BADBIOSTABLE;
+
+	*dce_caps = disp_cntl_tbl->display_caps;
+
+	return result;
+}
+
+static enum bp_result bios_parser_get_lttpr_interop(
+	struct dc_bios *dcb,
+	uint8_t *dce_caps)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	enum bp_result result = BP_RESULT_UNSUPPORTED;
+	struct atom_common_table_header *header;
+	struct atom_data_revision tbl_revision;
+
+	if (!DATA_TABLES(dce_info))
+		return BP_RESULT_UNSUPPORTED;
+
+	header = GET_IMAGE(struct atom_common_table_header,
+						DATA_TABLES(dce_info));
+	get_atom_data_table_revision(header, &tbl_revision);
+	switch (tbl_revision.major) {
+	case 4:
+		switch (tbl_revision.minor) {
+		case 1:
+			result = get_disp_caps_v4_1(bp, dce_caps);
+			*dce_caps = !!(*dce_caps & DCE_INFO_CAPS_VBIOS_LTTPR_TRANSPARENT_ENABLE);
+			break;
+		case 2:
+			result = get_disp_caps_v4_2(bp, dce_caps);
+			*dce_caps = !!(*dce_caps & DCE_INFO_CAPS_VBIOS_LTTPR_TRANSPARENT_ENABLE);
+			break;
+		case 3:
+			result = get_disp_caps_v4_3(bp, dce_caps);
+			*dce_caps = !!(*dce_caps & DCE_INFO_CAPS_VBIOS_LTTPR_TRANSPARENT_ENABLE);
+			break;
+		case 4:
+			result = get_disp_caps_v4_4(bp, dce_caps);
+			*dce_caps = !!(*dce_caps & DCE_INFO_CAPS_VBIOS_LTTPR_TRANSPARENT_ENABLE);
+			break;
+		case 5:
+			result = get_disp_caps_v4_5(bp, dce_caps);
+			*dce_caps = !!(*dce_caps & DCE_INFO_CAPS_VBIOS_LTTPR_TRANSPARENT_ENABLE);
+			break;
+
+		default:
+			break;
+		}
+		break;
+	default:
+		break;
+	}
+	DC_LOG_BIOS("DCE_INFO_CAPS_VBIOS_LTTPR_TRANSPARENT_ENABLE: %d tbl_revision.major = %d tbl_revision.minor = %d\n", *dce_caps, tbl_revision.major, tbl_revision.minor);
+	return result;
+}
+
+static enum bp_result bios_parser_get_lttpr_caps(
+	struct dc_bios *dcb,
+	uint8_t *dce_caps)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	enum bp_result result = BP_RESULT_UNSUPPORTED;
+	struct atom_common_table_header *header;
+	struct atom_data_revision tbl_revision;
+
+	if (!DATA_TABLES(dce_info))
+		return BP_RESULT_UNSUPPORTED;
+
+	*dce_caps  = 0;
+	header = GET_IMAGE(struct atom_common_table_header,
+						DATA_TABLES(dce_info));
+	get_atom_data_table_revision(header, &tbl_revision);
+	switch (tbl_revision.major) {
+	case 4:
+		switch (tbl_revision.minor) {
+		case 1:
+			result = get_disp_caps_v4_1(bp, dce_caps);
+			*dce_caps = !!(*dce_caps & DCE_INFO_CAPS_LTTPR_SUPPORT_ENABLE);
+			break;
+		case 2:
+			result = get_disp_caps_v4_2(bp, dce_caps);
+			*dce_caps = !!(*dce_caps & DCE_INFO_CAPS_LTTPR_SUPPORT_ENABLE);
+			break;
+		case 3:
+			result = get_disp_caps_v4_3(bp, dce_caps);
+			*dce_caps = !!(*dce_caps & DCE_INFO_CAPS_LTTPR_SUPPORT_ENABLE);
+			break;
+		case 4:
+			result = get_disp_caps_v4_4(bp, dce_caps);
+			*dce_caps = !!(*dce_caps & DCE_INFO_CAPS_LTTPR_SUPPORT_ENABLE);
+			break;
+		case 5:
+			result = get_disp_caps_v4_5(bp, dce_caps);
+			*dce_caps = !!(*dce_caps & DCE_INFO_CAPS_LTTPR_SUPPORT_ENABLE);
+			break;
+		default:
+			break;
+		}
+		break;
+	default:
+		break;
+	}
+	DC_LOG_BIOS("DCE_INFO_CAPS_LTTPR_SUPPORT_ENABLE: %d tbl_revision.major = %d tbl_revision.minor = %d\n", *dce_caps, tbl_revision.major, tbl_revision.minor);
+	if (dcb->ctx->dc->config.force_bios_enable_lttpr && *dce_caps == 0) {
+		*dce_caps = 1;
+		DC_LOG_BIOS("DCE_INFO_CAPS_VBIOS_LTTPR_TRANSPARENT_ENABLE: forced enabled");
+	}
+	return result;
+}
+
+static enum bp_result get_embedded_panel_info_v2_1(
+		struct bios_parser *bp,
+		struct embedded_panel_info *info)
+{
+	struct lcd_info_v2_1 *lvds;
+
+	if (!info)
+		return BP_RESULT_BADINPUT;
+
+	if (!DATA_TABLES(lcd_info))
+		return BP_RESULT_UNSUPPORTED;
+
+	lvds = GET_IMAGE(struct lcd_info_v2_1, DATA_TABLES(lcd_info));
+
+	if (!lvds)
+		return BP_RESULT_BADBIOSTABLE;
+
+	/* TODO: previous vv1_3, should v2_1 */
+	if (!((lvds->table_header.format_revision == 2)
+			&& (lvds->table_header.content_revision >= 1)))
+		return BP_RESULT_UNSUPPORTED;
+
+	memset(info, 0, sizeof(struct embedded_panel_info));
+
+	/* We need to convert from 10KHz units into KHz units */
+	info->lcd_timing.pixel_clk = le16_to_cpu(lvds->lcd_timing.pixclk) * 10;
+	/* usHActive does not include borders, according to VBIOS team */
+	info->lcd_timing.horizontal_addressable = le16_to_cpu(lvds->lcd_timing.h_active);
+	/* usHBlanking_Time includes borders, so we should really be
+	 * subtractingborders duing this translation, but LVDS generally
+	 * doesn't have borders, so we should be okay leaving this as is for
+	 * now.  May need to revisit if we ever have LVDS with borders
+	 */
+	info->lcd_timing.horizontal_blanking_time = le16_to_cpu(lvds->lcd_timing.h_blanking_time);
+	/* usVActive does not include borders, according to VBIOS team*/
+	info->lcd_timing.vertical_addressable = le16_to_cpu(lvds->lcd_timing.v_active);
+	/* usVBlanking_Time includes borders, so we should really be
+	 * subtracting borders duing this translation, but LVDS generally
+	 * doesn't have borders, so we should be okay leaving this as is for
+	 * now. May need to revisit if we ever have LVDS with borders
+	 */
+	info->lcd_timing.vertical_blanking_time = le16_to_cpu(lvds->lcd_timing.v_blanking_time);
+	info->lcd_timing.horizontal_sync_offset = le16_to_cpu(lvds->lcd_timing.h_sync_offset);
+	info->lcd_timing.horizontal_sync_width = le16_to_cpu(lvds->lcd_timing.h_sync_width);
+	info->lcd_timing.vertical_sync_offset = le16_to_cpu(lvds->lcd_timing.v_sync_offset);
+	info->lcd_timing.vertical_sync_width = le16_to_cpu(lvds->lcd_timing.v_syncwidth);
+	info->lcd_timing.horizontal_border = lvds->lcd_timing.h_border;
+	info->lcd_timing.vertical_border = lvds->lcd_timing.v_border;
+
+	/* not provided by VBIOS */
+	info->lcd_timing.misc_info.HORIZONTAL_CUT_OFF = 0;
+
+	info->lcd_timing.misc_info.H_SYNC_POLARITY = ~(uint32_t) (lvds->lcd_timing.miscinfo
+			& ATOM_HSYNC_POLARITY);
+	info->lcd_timing.misc_info.V_SYNC_POLARITY = ~(uint32_t) (lvds->lcd_timing.miscinfo
+			& ATOM_VSYNC_POLARITY);
+
+	/* not provided by VBIOS */
+	info->lcd_timing.misc_info.VERTICAL_CUT_OFF = 0;
+
+	info->lcd_timing.misc_info.H_REPLICATION_BY2 = !!(lvds->lcd_timing.miscinfo
+			& ATOM_H_REPLICATIONBY2);
+	info->lcd_timing.misc_info.V_REPLICATION_BY2 = !!(lvds->lcd_timing.miscinfo
+			& ATOM_V_REPLICATIONBY2);
+	info->lcd_timing.misc_info.COMPOSITE_SYNC = !!(lvds->lcd_timing.miscinfo
+			& ATOM_COMPOSITESYNC);
+	info->lcd_timing.misc_info.INTERLACE = !!(lvds->lcd_timing.miscinfo & ATOM_INTERLACE);
+
+	/* not provided by VBIOS*/
+	info->lcd_timing.misc_info.DOUBLE_CLOCK = 0;
+	/* not provided by VBIOS*/
+	info->ss_id = 0;
+
+	info->realtek_eDPToLVDS = !!(lvds->dplvdsrxid == eDP_TO_LVDS_REALTEK_ID);
+
+	return BP_RESULT_OK;
+}
+
+static enum bp_result bios_parser_get_embedded_panel_info(
+		struct dc_bios *dcb,
+		struct embedded_panel_info *info)
+{
+	struct bios_parser
+	*bp = BP_FROM_DCB(dcb);
+	struct atom_common_table_header *header;
+	struct atom_data_revision tbl_revision;
+
+	if (!DATA_TABLES(lcd_info))
+		return BP_RESULT_FAILURE;
+
+	header = GET_IMAGE(struct atom_common_table_header, DATA_TABLES(lcd_info));
+
+	if (!header)
+		return BP_RESULT_BADBIOSTABLE;
+
+	get_atom_data_table_revision(header, &tbl_revision);
+
+	switch (tbl_revision.major) {
+	case 2:
+		switch (tbl_revision.minor) {
+		case 1:
+			return get_embedded_panel_info_v2_1(bp, info);
+		default:
+			break;
+		}
+		break;
+	default:
+		break;
+	}
+
+	return BP_RESULT_FAILURE;
+}
+
+static uint32_t get_support_mask_for_device_id(struct device_id device_id)
+{
+	enum dal_device_type device_type = device_id.device_type;
+	uint32_t enum_id = device_id.enum_id;
+
+	switch (device_type) {
+	case DEVICE_TYPE_LCD:
+		switch (enum_id) {
+		case 1:
+			return ATOM_DISPLAY_LCD1_SUPPORT;
+		default:
+			break;
+		}
+		break;
+	case DEVICE_TYPE_DFP:
+		switch (enum_id) {
+		case 1:
+			return ATOM_DISPLAY_DFP1_SUPPORT;
+		case 2:
+			return ATOM_DISPLAY_DFP2_SUPPORT;
+		case 3:
+			return ATOM_DISPLAY_DFP3_SUPPORT;
+		case 4:
+			return ATOM_DISPLAY_DFP4_SUPPORT;
+		case 5:
+			return ATOM_DISPLAY_DFP5_SUPPORT;
+		case 6:
+			return ATOM_DISPLAY_DFP6_SUPPORT;
+		default:
+			break;
+		}
+		break;
+	default:
+		break;
+	}
+
+	/* Unidentified device ID, return empty support mask. */
+	return 0;
+}
+
+static bool bios_parser_is_device_id_supported(
+	struct dc_bios *dcb,
+	struct device_id id)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	uint32_t mask = get_support_mask_for_device_id(id);
+
+	switch (bp->object_info_tbl.revision.minor) {
+	case 4:
+	default:
+		return (le16_to_cpu(bp->object_info_tbl.v1_4->supporteddevices) & mask) != 0;
+		break;
+	case 5:
+		return (le16_to_cpu(bp->object_info_tbl.v1_5->supporteddevices) & mask) != 0;
+		break;
+	}
+}
+
+static uint32_t bios_parser_get_ss_entry_number(
+	struct dc_bios *dcb,
+	enum as_signal_type signal)
+{
+	/* TODO: DAL2 atomfirmware implementation does not need this.
+	 * why DAL3 need this?
+	 */
+	return 1;
+}
+
+static enum bp_result bios_parser_transmitter_control(
+	struct dc_bios *dcb,
+	struct bp_transmitter_control *cntl)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	if (!bp->cmd_tbl.transmitter_control)
+		return BP_RESULT_FAILURE;
+
+	return bp->cmd_tbl.transmitter_control(bp, cntl);
+}
+
+static enum bp_result bios_parser_encoder_control(
+	struct dc_bios *dcb,
+	struct bp_encoder_control *cntl)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	if (!bp->cmd_tbl.dig_encoder_control)
+		return BP_RESULT_FAILURE;
+
+	return bp->cmd_tbl.dig_encoder_control(bp, cntl);
+}
+
+static enum bp_result bios_parser_set_pixel_clock(
+	struct dc_bios *dcb,
+	struct bp_pixel_clock_parameters *bp_params)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	if (!bp->cmd_tbl.set_pixel_clock)
+		return BP_RESULT_FAILURE;
+
+	return bp->cmd_tbl.set_pixel_clock(bp, bp_params);
+}
+
+static enum bp_result bios_parser_set_dce_clock(
+	struct dc_bios *dcb,
+	struct bp_set_dce_clock_parameters *bp_params)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	if (!bp->cmd_tbl.set_dce_clock)
+		return BP_RESULT_FAILURE;
+
+	return bp->cmd_tbl.set_dce_clock(bp, bp_params);
+}
+
+static enum bp_result bios_parser_program_crtc_timing(
+	struct dc_bios *dcb,
+	struct bp_hw_crtc_timing_parameters *bp_params)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	if (!bp->cmd_tbl.set_crtc_timing)
+		return BP_RESULT_FAILURE;
+
+	return bp->cmd_tbl.set_crtc_timing(bp, bp_params);
+}
+
+static enum bp_result bios_parser_enable_crtc(
+	struct dc_bios *dcb,
+	enum controller_id id,
+	bool enable)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	if (!bp->cmd_tbl.enable_crtc)
+		return BP_RESULT_FAILURE;
+
+	return bp->cmd_tbl.enable_crtc(bp, id, enable);
+}
+
+static enum bp_result bios_parser_enable_disp_power_gating(
+	struct dc_bios *dcb,
+	enum controller_id controller_id,
+	enum bp_pipe_control_action action)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	if (!bp->cmd_tbl.enable_disp_power_gating)
+		return BP_RESULT_FAILURE;
+
+	return bp->cmd_tbl.enable_disp_power_gating(bp, controller_id,
+		action);
+}
+
+static enum bp_result bios_parser_enable_lvtma_control(
+	struct dc_bios *dcb,
+	uint8_t uc_pwr_on,
+	uint8_t pwrseq_instance,
+	uint8_t bypass_panel_control_wait)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+
+	if (!bp->cmd_tbl.enable_lvtma_control)
+		return BP_RESULT_FAILURE;
+
+	return bp->cmd_tbl.enable_lvtma_control(bp, uc_pwr_on, pwrseq_instance, bypass_panel_control_wait);
+}
+
+static bool bios_parser_is_accelerated_mode(
+	struct dc_bios *dcb)
+{
+	return bios_is_accelerated_mode(dcb);
+}
+
+/**
+ * bios_parser_set_scratch_critical_state - update critical state bit
+ *                                          in VBIOS scratch register
+ *
+ * @dcb:   pointer to the DC BIO
+ * @state: set or reset state
+ */
+static void bios_parser_set_scratch_critical_state(
+	struct dc_bios *dcb,
+	bool state)
+{
+	bios_set_scratch_critical_state(dcb, state);
+}
+
+static enum bp_result bios_parser_get_firmware_info(
+	struct dc_bios *dcb,
+	struct dc_firmware_info *info)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	static enum bp_result result = BP_RESULT_BADBIOSTABLE;
+	struct atom_common_table_header *header;
+
+	struct atom_data_revision revision;
+
+	if (info && DATA_TABLES(firmwareinfo)) {
+		header = GET_IMAGE(struct atom_common_table_header,
+				DATA_TABLES(firmwareinfo));
+		get_atom_data_table_revision(header, &revision);
+		switch (revision.major) {
+		case 3:
+			switch (revision.minor) {
+			case 1:
+				result = get_firmware_info_v3_1(bp, info);
+				break;
+			case 2:
+			case 3:
+				result = get_firmware_info_v3_2(bp, info);
+				break;
+			case 4:
+				result = get_firmware_info_v3_4(bp, info);
+				break;
+			case 5:
+				result = get_firmware_info_v3_5(bp, info);
+				break;
+			default:
+				break;
+			}
+			break;
+		default:
+			break;
+		}
+	}
+
+	return result;
+}
+
+static enum bp_result get_firmware_info_v3_1(
+	struct bios_parser *bp,
+	struct dc_firmware_info *info)
+{
+	struct atom_firmware_info_v3_1 *firmware_info;
+	struct atom_firmware_info_v3_2 *firmware_info32;
+	struct atom_display_controller_info_v4_1 *dce_info = NULL;
+
+	if (!info)
+		return BP_RESULT_BADINPUT;
+
+	firmware_info = GET_IMAGE(struct atom_firmware_info_v3_1,
+			DATA_TABLES(firmwareinfo));
+	firmware_info32 = GET_IMAGE(struct atom_firmware_info_v3_2,
+			DATA_TABLES(firmwareinfo));
+
+	dce_info = GET_IMAGE(struct atom_display_controller_info_v4_1,
+			DATA_TABLES(dce_info));
+
+	if (!firmware_info || !firmware_info32 || !dce_info)
+		return BP_RESULT_BADBIOSTABLE;
+
+	memset(info, 0, sizeof(*info));
+
+	/* Pixel clock pll information. */
+	 /* We need to convert from 10KHz units into KHz units */
+	info->default_memory_clk = firmware_info->bootup_mclk_in10khz * 10;
+	info->default_engine_clk = firmware_info->bootup_sclk_in10khz * 10;
+
+	 /* 27MHz for Vega10: */
+	info->pll_info.crystal_frequency = dce_info->dce_refclk_10khz * 10;
+
+	/* Hardcode frequency if BIOS gives no DCE Ref Clk */
+	if (info->pll_info.crystal_frequency == 0)
+		info->pll_info.crystal_frequency = 27000;
+	/*dp_phy_ref_clk is not correct for atom_display_controller_info_v4_2, but we don't use it*/
+	info->dp_phy_ref_clk     = dce_info->dpphy_refclk_10khz * 10;
+	info->i2c_engine_ref_clk = dce_info->i2c_engine_refclk_10khz * 10;
+
+	/* Get GPU PLL VCO Clock */
+
+	if (bp->cmd_tbl.get_smu_clock_info != NULL) {
+		/* VBIOS gives in 10KHz */
+		info->smu_gpu_pll_output_freq =
+				bp->cmd_tbl.get_smu_clock_info(bp, SMU9_SYSPLL0_ID) * 10;
+	}
+
+	/* These fields are marked as reserved in v3_1, but they appear to be populated
+	 * properly.
+	 */
+	if (firmware_info32 && firmware_info32->board_i2c_feature_id == 0x2) {
+		info->oem_i2c_present = true;
+		info->oem_i2c_obj_id = firmware_info32->board_i2c_feature_gpio_id;
+	} else {
+		info->oem_i2c_present = false;
+	}
+
+	return BP_RESULT_OK;
+}
+
+static enum bp_result get_firmware_info_v3_2(
+	struct bios_parser *bp,
+	struct dc_firmware_info *info)
+{
+	struct atom_firmware_info_v3_2 *firmware_info;
+	struct atom_display_controller_info_v4_1 *dce_info = NULL;
+	struct atom_common_table_header *header;
+	struct atom_data_revision revision;
+	struct atom_smu_info_v3_2 *smu_info_v3_2 = NULL;
+	struct atom_smu_info_v3_3 *smu_info_v3_3 = NULL;
+
+	if (!info)
+		return BP_RESULT_BADINPUT;
+
+	firmware_info = GET_IMAGE(struct atom_firmware_info_v3_2,
+			DATA_TABLES(firmwareinfo));
+
+	dce_info = GET_IMAGE(struct atom_display_controller_info_v4_1,
+			DATA_TABLES(dce_info));
+
+	if (!firmware_info || !dce_info)
+		return BP_RESULT_BADBIOSTABLE;
+
+	memset(info, 0, sizeof(*info));
+
+	header = GET_IMAGE(struct atom_common_table_header,
+					DATA_TABLES(smu_info));
+	get_atom_data_table_revision(header, &revision);
+
+	if (revision.minor == 2) {
+		/* Vega12 */
+		smu_info_v3_2 = GET_IMAGE(struct atom_smu_info_v3_2,
+							DATA_TABLES(smu_info));
+		if (!smu_info_v3_2)
+			return BP_RESULT_BADBIOSTABLE;
+
+		DC_LOG_BIOS("gpuclk_ss_percentage (unit of 0.001 percent): %d\n", smu_info_v3_2->gpuclk_ss_percentage);
+
+		info->default_engine_clk = smu_info_v3_2->bootup_dcefclk_10khz * 10;
+	} else if (revision.minor == 3) {
+		/* Vega20 */
+		smu_info_v3_3 = GET_IMAGE(struct atom_smu_info_v3_3,
+							DATA_TABLES(smu_info));
+		if (!smu_info_v3_3)
+			return BP_RESULT_BADBIOSTABLE;
+
+		DC_LOG_BIOS("gpuclk_ss_percentage (unit of 0.001 percent): %d\n", smu_info_v3_3->gpuclk_ss_percentage);
+
+		info->default_engine_clk = smu_info_v3_3->bootup_dcefclk_10khz * 10;
+	}
+
+	 // We need to convert from 10KHz units into KHz units.
+	info->default_memory_clk = firmware_info->bootup_mclk_in10khz * 10;
+
+	 /* 27MHz for Vega10 & Vega12; 100MHz for Vega20 */
+	info->pll_info.crystal_frequency = dce_info->dce_refclk_10khz * 10;
+	/* Hardcode frequency if BIOS gives no DCE Ref Clk */
+	if (info->pll_info.crystal_frequency == 0) {
+		if (revision.minor == 2)
+			info->pll_info.crystal_frequency = 27000;
+		else if (revision.minor == 3)
+			info->pll_info.crystal_frequency = 100000;
+	}
+	/*dp_phy_ref_clk is not correct for atom_display_controller_info_v4_2, but we don't use it*/
+	info->dp_phy_ref_clk     = dce_info->dpphy_refclk_10khz * 10;
+	info->i2c_engine_ref_clk = dce_info->i2c_engine_refclk_10khz * 10;
+
+	/* Get GPU PLL VCO Clock */
+	if (bp->cmd_tbl.get_smu_clock_info != NULL) {
+		if (revision.minor == 2)
+			info->smu_gpu_pll_output_freq =
+					bp->cmd_tbl.get_smu_clock_info(bp, SMU9_SYSPLL0_ID) * 10;
+		else if (revision.minor == 3)
+			info->smu_gpu_pll_output_freq =
+					bp->cmd_tbl.get_smu_clock_info(bp, SMU11_SYSPLL3_0_ID) * 10;
+	}
+
+	if (firmware_info->board_i2c_feature_id == 0x2) {
+		info->oem_i2c_present = true;
+		info->oem_i2c_obj_id = firmware_info->board_i2c_feature_gpio_id;
+	} else {
+		info->oem_i2c_present = false;
+	}
+
+	return BP_RESULT_OK;
+}
+
+static enum bp_result get_firmware_info_v3_4(
+	struct bios_parser *bp,
+	struct dc_firmware_info *info)
+{
+	struct atom_firmware_info_v3_4 *firmware_info;
+	struct atom_common_table_header *header;
+	struct atom_data_revision revision;
+	struct atom_display_controller_info_v4_1 *dce_info_v4_1 = NULL;
+	struct atom_display_controller_info_v4_4 *dce_info_v4_4 = NULL;
+
+	struct atom_smu_info_v3_5 *smu_info_v3_5 = NULL;
+	struct atom_display_controller_info_v4_5 *dce_info_v4_5 = NULL;
+	struct atom_smu_info_v4_0 *smu_info_v4_0 = NULL;
+
+	if (!info)
+		return BP_RESULT_BADINPUT;
+
+	firmware_info = GET_IMAGE(struct atom_firmware_info_v3_4,
+			DATA_TABLES(firmwareinfo));
+
+	if (!firmware_info)
+		return BP_RESULT_BADBIOSTABLE;
+
+	memset(info, 0, sizeof(*info));
+
+	header = GET_IMAGE(struct atom_common_table_header,
+					DATA_TABLES(dce_info));
+
+	get_atom_data_table_revision(header, &revision);
+
+	switch (revision.major) {
+	case 4:
+		switch (revision.minor) {
+		case 5:
+			dce_info_v4_5 = GET_IMAGE(struct atom_display_controller_info_v4_5,
+							DATA_TABLES(dce_info));
+
+			if (!dce_info_v4_5)
+				return BP_RESULT_BADBIOSTABLE;
+
+			 /* 100MHz expected */
+			info->pll_info.crystal_frequency = dce_info_v4_5->dce_refclk_10khz * 10;
+			info->dp_phy_ref_clk             = dce_info_v4_5->dpphy_refclk_10khz * 10;
+			 /* 50MHz expected */
+			info->i2c_engine_ref_clk         = dce_info_v4_5->i2c_engine_refclk_10khz * 10;
+
+			/* For DCN32/321 Display PLL VCO Frequency from dce_info_v4_5 may not be reliable */
+			break;
+
+		case 4:
+			dce_info_v4_4 = GET_IMAGE(struct atom_display_controller_info_v4_4,
+							DATA_TABLES(dce_info));
+
+			if (!dce_info_v4_4)
+				return BP_RESULT_BADBIOSTABLE;
+
+			/* 100MHz expected */
+			info->pll_info.crystal_frequency = dce_info_v4_4->dce_refclk_10khz * 10;
+			info->dp_phy_ref_clk             = dce_info_v4_4->dpphy_refclk_10khz * 10;
+			/* 50MHz expected */
+			info->i2c_engine_ref_clk         = dce_info_v4_4->i2c_engine_refclk_10khz * 10;
+
+			/* Get SMU Display PLL VCO Frequency in KHz*/
+			info->smu_gpu_pll_output_freq =	dce_info_v4_4->dispclk_pll_vco_freq * 10;
+			break;
+
+		default:
+			/* should not come here, keep as backup, as was before */
+			dce_info_v4_1 = GET_IMAGE(struct atom_display_controller_info_v4_1,
+							DATA_TABLES(dce_info));
+
+			if (!dce_info_v4_1)
+				return BP_RESULT_BADBIOSTABLE;
+
+			info->pll_info.crystal_frequency = dce_info_v4_1->dce_refclk_10khz * 10;
+			info->dp_phy_ref_clk             = dce_info_v4_1->dpphy_refclk_10khz * 10;
+			info->i2c_engine_ref_clk         = dce_info_v4_1->i2c_engine_refclk_10khz * 10;
+			break;
+		}
+		break;
+
+	default:
+		ASSERT(0);
+		break;
+	}
+
+	header = GET_IMAGE(struct atom_common_table_header,
+					DATA_TABLES(smu_info));
+	get_atom_data_table_revision(header, &revision);
+
+	switch (revision.major) {
+	case 3:
+		switch (revision.minor) {
+		case 5:
+			smu_info_v3_5 = GET_IMAGE(struct atom_smu_info_v3_5,
+							DATA_TABLES(smu_info));
+
+			if (!smu_info_v3_5)
+				return BP_RESULT_BADBIOSTABLE;
+			DC_LOG_BIOS("gpuclk_ss_percentage (unit of 0.001 percent): %d\n", smu_info_v3_5->gpuclk_ss_percentage);
+			info->default_engine_clk = smu_info_v3_5->bootup_dcefclk_10khz * 10;
+			break;
+
+		default:
+			break;
+		}
+		break;
+
+	case 4:
+		switch (revision.minor) {
+		case 0:
+			smu_info_v4_0 = GET_IMAGE(struct atom_smu_info_v4_0,
+							DATA_TABLES(smu_info));
+
+			if (!smu_info_v4_0)
+				return BP_RESULT_BADBIOSTABLE;
+
+			/* For DCN32/321 bootup DCFCLK from smu_info_v4_0 may not be reliable */
+			break;
+
+		default:
+			break;
+		}
+		break;
+
+	default:
+		break;
+	}
+
+	 // We need to convert from 10KHz units into KHz units.
+	info->default_memory_clk = firmware_info->bootup_mclk_in10khz * 10;
+
+	if (firmware_info->board_i2c_feature_id == 0x2) {
+		info->oem_i2c_present = true;
+		info->oem_i2c_obj_id = firmware_info->board_i2c_feature_gpio_id;
+	} else {
+		info->oem_i2c_present = false;
+	}
+
+	return BP_RESULT_OK;
+}
+
+static enum bp_result get_firmware_info_v3_5(
+	struct bios_parser *bp,
+	struct dc_firmware_info *info)
+{
+	struct atom_firmware_info_v3_5 *firmware_info;
+	struct atom_common_table_header *header;
+	struct atom_data_revision revision;
+	struct atom_display_controller_info_v4_5 *dce_info_v4_5 = NULL;
+
+	if (!info)
+		return BP_RESULT_BADINPUT;
+
+	firmware_info = GET_IMAGE(struct atom_firmware_info_v3_5,
+			DATA_TABLES(firmwareinfo));
+
+	if (!firmware_info)
+		return BP_RESULT_BADBIOSTABLE;
+
+	memset(info, 0, sizeof(*info));
+
+	if (firmware_info->board_i2c_feature_id == 0x2) {
+		info->oem_i2c_present = true;
+		info->oem_i2c_obj_id = firmware_info->board_i2c_feature_gpio_id;
+	} else {
+		info->oem_i2c_present = false;
+	}
+
+	header = GET_IMAGE(struct atom_common_table_header,
+					DATA_TABLES(dce_info));
+
+	get_atom_data_table_revision(header, &revision);
+
+	switch (revision.major) {
+	case 4:
+		switch (revision.minor) {
+		case 5:
+			dce_info_v4_5 = GET_IMAGE(struct atom_display_controller_info_v4_5,
+							DATA_TABLES(dce_info));
+
+			if (!dce_info_v4_5)
+				return BP_RESULT_BADBIOSTABLE;
+
+			 /* 100MHz expected */
+			info->pll_info.crystal_frequency = dce_info_v4_5->dce_refclk_10khz * 10;
+			break;
+		default:
+			break;
+		}
+		break;
+	default:
+		break;
+	}
+
+
+	return BP_RESULT_OK;
+}
+
+static enum bp_result bios_parser_get_encoder_cap_info(
+	struct dc_bios *dcb,
+	struct graphics_object_id object_id,
+	struct bp_encoder_cap_info *info)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	struct atom_display_object_path_v2 *object;
+	struct atom_encoder_caps_record *record = NULL;
+
+	if (!info)
+		return BP_RESULT_BADINPUT;
+
+#if defined(CONFIG_DRM_AMD_DC_FP)
+	/* encoder cap record not available in v1_5 */
+	if (bp->object_info_tbl.revision.minor == 5)
+		return BP_RESULT_NORECORD;
+#endif
+
+	object = get_bios_object(bp, object_id);
+
+	if (!object)
+		return BP_RESULT_BADINPUT;
+
+	record = get_encoder_cap_record(bp, object);
+	if (!record)
+		return BP_RESULT_NORECORD;
+	DC_LOG_BIOS("record->encodercaps 0x%x for object_id 0x%x", record->encodercaps, object_id.id);
+
+	info->DP_HBR2_CAP = (record->encodercaps &
+			ATOM_ENCODER_CAP_RECORD_HBR2) ? 1 : 0;
+	info->DP_HBR2_EN = (record->encodercaps &
+			ATOM_ENCODER_CAP_RECORD_HBR2_EN) ? 1 : 0;
+	info->DP_HBR3_EN = (record->encodercaps &
+			ATOM_ENCODER_CAP_RECORD_HBR3_EN) ? 1 : 0;
+	info->HDMI_6GB_EN = (record->encodercaps &
+			ATOM_ENCODER_CAP_RECORD_HDMI6Gbps_EN) ? 1 : 0;
+	info->IS_DP2_CAPABLE = (record->encodercaps &
+			ATOM_ENCODER_CAP_RECORD_DP2) ? 1 : 0;
+	info->DP_UHBR10_EN = (record->encodercaps &
+			ATOM_ENCODER_CAP_RECORD_UHBR10_EN) ? 1 : 0;
+	info->DP_UHBR13_5_EN = (record->encodercaps &
+			ATOM_ENCODER_CAP_RECORD_UHBR13_5_EN) ? 1 : 0;
+	info->DP_UHBR20_EN = (record->encodercaps &
+			ATOM_ENCODER_CAP_RECORD_UHBR20_EN) ? 1 : 0;
+	info->DP_IS_USB_C = (record->encodercaps &
+			ATOM_ENCODER_CAP_RECORD_USB_C_TYPE) ? 1 : 0;
+	DC_LOG_BIOS("\t info->DP_IS_USB_C %d", info->DP_IS_USB_C);
+
+	return BP_RESULT_OK;
+}
+
+
+static struct atom_encoder_caps_record *get_encoder_cap_record(
+	struct bios_parser *bp,
+	struct atom_display_object_path_v2 *object)
+{
+	struct atom_common_record_header *header;
+	uint32_t offset;
+
+	if (!object) {
+		BREAK_TO_DEBUGGER(); /* Invalid object */
+		return NULL;
+	}
+
+	offset = object->encoder_recordoffset + bp->object_info_tbl_offset;
+
+	for (;;) {
+		header = GET_IMAGE(struct atom_common_record_header, offset);
+
+		if (!header)
+			return NULL;
+
+		offset += header->record_size;
+
+		if (header->record_type == LAST_RECORD_TYPE ||
+				!header->record_size)
+			break;
+
+		if (header->record_type != ATOM_ENCODER_CAP_RECORD_TYPE)
+			continue;
+
+		if (sizeof(struct atom_encoder_caps_record) <=
+							header->record_size)
+			return (struct atom_encoder_caps_record *)header;
+	}
+
+	return NULL;
+}
+
+static struct atom_disp_connector_caps_record *get_disp_connector_caps_record(
+	struct bios_parser *bp,
+	struct atom_display_object_path_v2 *object)
+{
+	struct atom_common_record_header *header;
+	uint32_t offset;
+
+	if (!object) {
+		BREAK_TO_DEBUGGER(); /* Invalid object */
+		return NULL;
+	}
+
+	offset = object->disp_recordoffset + bp->object_info_tbl_offset;
+
+	for (;;) {
+		header = GET_IMAGE(struct atom_common_record_header, offset);
+
+		if (!header)
+			return NULL;
+
+		offset += header->record_size;
+
+		if (header->record_type == LAST_RECORD_TYPE ||
+				!header->record_size)
+			break;
+
+		if (header->record_type != ATOM_DISP_CONNECTOR_CAPS_RECORD_TYPE)
+			continue;
+
+		if (sizeof(struct atom_disp_connector_caps_record) <=
+							header->record_size)
+			return (struct atom_disp_connector_caps_record *)header;
+	}
+
+	return NULL;
+}
+
+static struct atom_connector_caps_record *get_connector_caps_record(struct bios_parser *bp,
+								    struct atom_display_object_path_v3 *object)
+{
+	struct atom_common_record_header *header;
+	uint32_t offset;
+
+	if (!object) {
+		BREAK_TO_DEBUGGER(); /* Invalid object */
+		return NULL;
+	}
+
+	offset = object->disp_recordoffset + bp->object_info_tbl_offset;
+
+	for (;;) {
+		header = GET_IMAGE(struct atom_common_record_header, offset);
+
+		if (!header)
+			return NULL;
+
+		offset += header->record_size;
+
+		if (header->record_type == ATOM_RECORD_END_TYPE ||
+				!header->record_size)
+			break;
+
+		if (header->record_type != ATOM_CONNECTOR_CAP_RECORD_TYPE)
+			continue;
+
+		if (sizeof(struct atom_connector_caps_record) <= header->record_size)
+			return (struct atom_connector_caps_record *)header;
+	}
+
+	return NULL;
+}
+
+static enum bp_result bios_parser_get_disp_connector_caps_info(
+	struct dc_bios *dcb,
+	struct graphics_object_id object_id,
+	struct bp_disp_connector_caps_info *info)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	struct atom_display_object_path_v2 *object;
+	struct atom_display_object_path_v3 *object_path_v3;
+	struct atom_connector_caps_record *record_path_v3;
+	struct atom_disp_connector_caps_record *record = NULL;
+
+	if (!info)
+		return BP_RESULT_BADINPUT;
+
+	switch (bp->object_info_tbl.revision.minor) {
+	case 4:
+		default:
+			object = get_bios_object(bp, object_id);
+
+			if (!object)
+				return BP_RESULT_BADINPUT;
+
+			record = get_disp_connector_caps_record(bp, object);
+			if (!record)
+				return BP_RESULT_NORECORD;
+
+			info->INTERNAL_DISPLAY =
+				(record->connectcaps & ATOM_CONNECTOR_CAP_INTERNAL_DISPLAY) ? 1 : 0;
+			info->INTERNAL_DISPLAY_BL =
+				(record->connectcaps & ATOM_CONNECTOR_CAP_INTERNAL_DISPLAY_BL) ? 1 : 0;
+			break;
+	case 5:
+		object_path_v3 = get_bios_object_from_path_v3(bp, object_id);
+
+		if (!object_path_v3)
+			return BP_RESULT_BADINPUT;
+
+		record_path_v3 = get_connector_caps_record(bp, object_path_v3);
+		if (!record_path_v3)
+			return BP_RESULT_NORECORD;
+
+		info->INTERNAL_DISPLAY = (record_path_v3->connector_caps & ATOM_CONNECTOR_CAP_INTERNAL_DISPLAY)
+									? 1 : 0;
+		info->INTERNAL_DISPLAY_BL = (record_path_v3->connector_caps & ATOM_CONNECTOR_CAP_INTERNAL_DISPLAY_BL)
+										? 1 : 0;
+		break;
+	}
+
+	return BP_RESULT_OK;
+}
+
+static struct atom_connector_speed_record *get_connector_speed_cap_record(struct bios_parser *bp,
+									  struct atom_display_object_path_v3 *object)
+{
+	struct atom_common_record_header *header;
+	uint32_t offset;
+
+	if (!object) {
+		BREAK_TO_DEBUGGER(); /* Invalid object */
+		return NULL;
+	}
+
+	offset = object->disp_recordoffset + bp->object_info_tbl_offset;
+
+	for (;;) {
+		header = GET_IMAGE(struct atom_common_record_header, offset);
+
+		if (!header)
+			return NULL;
+
+		offset += header->record_size;
+
+		if (header->record_type == ATOM_RECORD_END_TYPE ||
+				!header->record_size)
+			break;
+
+		if (header->record_type != ATOM_CONNECTOR_SPEED_UPTO)
+			continue;
+
+		if (sizeof(struct atom_connector_speed_record) <= header->record_size)
+			return (struct atom_connector_speed_record *)header;
+	}
+
+	return NULL;
+}
+
+static enum bp_result bios_parser_get_connector_speed_cap_info(
+	struct dc_bios *dcb,
+	struct graphics_object_id object_id,
+	struct bp_connector_speed_cap_info *info)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	struct atom_display_object_path_v3 *object_path_v3;
+	//struct atom_connector_speed_record *record = NULL;
+	struct atom_connector_speed_record *record;
+
+	if (!info)
+		return BP_RESULT_BADINPUT;
+
+	object_path_v3 = get_bios_object_from_path_v3(bp, object_id);
+
+	if (!object_path_v3)
+		return BP_RESULT_BADINPUT;
+
+	record = get_connector_speed_cap_record(bp, object_path_v3);
+	if (!record)
+		return BP_RESULT_NORECORD;
+
+	info->DP_HBR2_EN = (record->connector_max_speed >= 5400) ? 1 : 0;
+	info->DP_HBR3_EN = (record->connector_max_speed >= 8100) ? 1 : 0;
+	info->HDMI_6GB_EN = (record->connector_max_speed >= 5940) ? 1 : 0;
+	info->DP_UHBR10_EN = (record->connector_max_speed >= 10000) ? 1 : 0;
+	info->DP_UHBR13_5_EN = (record->connector_max_speed >= 13500) ? 1 : 0;
+	info->DP_UHBR20_EN = (record->connector_max_speed >= 20000) ? 1 : 0;
+	return BP_RESULT_OK;
+}
+
+static enum bp_result get_vram_info_v23(
+	struct bios_parser *bp,
+	struct dc_vram_info *info)
+{
+	struct atom_vram_info_header_v2_3 *info_v23;
+	static enum bp_result result = BP_RESULT_OK;
+
+	info_v23 = GET_IMAGE(struct atom_vram_info_header_v2_3,
+						DATA_TABLES(vram_info));
+
+	if (info_v23 == NULL)
+		return BP_RESULT_BADBIOSTABLE;
+
+	info->num_chans = info_v23->vram_module[0].channel_num;
+	info->dram_channel_width_bytes = (1 << info_v23->vram_module[0].channel_width) / 8;
+
+	return result;
+}
+
+static enum bp_result get_vram_info_v24(
+	struct bios_parser *bp,
+	struct dc_vram_info *info)
+{
+	struct atom_vram_info_header_v2_4 *info_v24;
+	static enum bp_result result = BP_RESULT_OK;
+
+	info_v24 = GET_IMAGE(struct atom_vram_info_header_v2_4,
+						DATA_TABLES(vram_info));
+
+	if (info_v24 == NULL)
+		return BP_RESULT_BADBIOSTABLE;
+
+	info->num_chans = info_v24->vram_module[0].channel_num;
+	info->dram_channel_width_bytes = (1 << info_v24->vram_module[0].channel_width) / 8;
+
+	return result;
+}
+
+static enum bp_result get_vram_info_v25(
+	struct bios_parser *bp,
+	struct dc_vram_info *info)
+{
+	struct atom_vram_info_header_v2_5 *info_v25;
+	static enum bp_result result = BP_RESULT_OK;
+
+	info_v25 = GET_IMAGE(struct atom_vram_info_header_v2_5,
+						DATA_TABLES(vram_info));
+
+	if (info_v25 == NULL)
+		return BP_RESULT_BADBIOSTABLE;
+
+	info->num_chans = info_v25->vram_module[0].channel_num;
+	info->dram_channel_width_bytes = (1 << info_v25->vram_module[0].channel_width) / 8;
+
+	return result;
+}
+
+static enum bp_result get_vram_info_v30(
+	struct bios_parser *bp,
+	struct dc_vram_info *info)
+{
+	struct atom_vram_info_header_v3_0 *info_v30;
+	enum bp_result result = BP_RESULT_OK;
+
+	info_v30 = GET_IMAGE(struct atom_vram_info_header_v3_0,
+						DATA_TABLES(vram_info));
+
+	if (info_v30 == NULL)
+		return BP_RESULT_BADBIOSTABLE;
+
+	info->num_chans = info_v30->channel_num;
+	info->dram_channel_width_bytes = (1 << info_v30->channel_width) / 8;
+
+	return result;
+}
+
+static enum bp_result get_vram_info_from_umc_info_v40(
+		struct bios_parser *bp,
+		struct dc_vram_info *info)
+{
+	struct atom_umc_info_v4_0 *info_v40;
+	enum bp_result result = BP_RESULT_OK;
+
+	info_v40 = GET_IMAGE(struct atom_umc_info_v4_0,
+						DATA_TABLES(umc_info));
+
+	if (info_v40 == NULL)
+		return BP_RESULT_BADBIOSTABLE;
+
+	info->num_chans = info_v40->channel_num;
+	info->dram_channel_width_bytes = (1 << info_v40->channel_width) / 8;
+
+	return result;
+}
+
+/*
+ * get_integrated_info_v11
+ *
+ * @brief
+ * Get V8 integrated BIOS information
+ *
+ * @param
+ * bios_parser *bp - [in]BIOS parser handler to get master data table
+ * integrated_info *info - [out] store and output integrated info
+ *
+ * @return
+ * static enum bp_result - BP_RESULT_OK if information is available,
+ *                  BP_RESULT_BADBIOSTABLE otherwise.
+ */
+static enum bp_result get_integrated_info_v11(
+	struct bios_parser *bp,
+	struct integrated_info *info)
+{
+	struct atom_integrated_system_info_v1_11 *info_v11;
+	uint32_t i;
+
+	info_v11 = GET_IMAGE(struct atom_integrated_system_info_v1_11,
+					DATA_TABLES(integratedsysteminfo));
+
+	if (info_v11 == NULL)
+		return BP_RESULT_BADBIOSTABLE;
+
+	DC_LOG_BIOS("gpuclk_ss_percentage (unit of 0.001 percent): %d\n", info_v11->gpuclk_ss_percentage);
+
+	info->gpu_cap_info =
+	le32_to_cpu(info_v11->gpucapinfo);
+	/*
+	* system_config: Bit[0] = 0 : PCIE power gating disabled
+	*                       = 1 : PCIE power gating enabled
+	*                Bit[1] = 0 : DDR-PLL shut down disabled
+	*                       = 1 : DDR-PLL shut down enabled
+	*                Bit[2] = 0 : DDR-PLL power down disabled
+	*                       = 1 : DDR-PLL power down enabled
+	*/
+	info->system_config = le32_to_cpu(info_v11->system_config);
+	info->cpu_cap_info = le32_to_cpu(info_v11->cpucapinfo);
+	info->memory_type = info_v11->memorytype;
+	info->ma_channel_number = info_v11->umachannelnumber;
+	info->lvds_ss_percentage =
+	le16_to_cpu(info_v11->lvds_ss_percentage);
+	info->dp_ss_control =
+	le16_to_cpu(info_v11->reserved1);
+	info->lvds_sspread_rate_in_10hz =
+	le16_to_cpu(info_v11->lvds_ss_rate_10hz);
+	info->hdmi_ss_percentage =
+	le16_to_cpu(info_v11->hdmi_ss_percentage);
+	info->hdmi_sspread_rate_in_10hz =
+	le16_to_cpu(info_v11->hdmi_ss_rate_10hz);
+	info->dvi_ss_percentage =
+	le16_to_cpu(info_v11->dvi_ss_percentage);
+	info->dvi_sspread_rate_in_10_hz =
+	le16_to_cpu(info_v11->dvi_ss_rate_10hz);
+	info->lvds_misc = info_v11->lvds_misc;
+	for (i = 0; i < NUMBER_OF_UCHAR_FOR_GUID; ++i) {
+		info->ext_disp_conn_info.gu_id[i] =
+				info_v11->extdispconninfo.guid[i];
+	}
+
+	for (i = 0; i < MAX_NUMBER_OF_EXT_DISPLAY_PATH; ++i) {
+		info->ext_disp_conn_info.path[i].device_connector_id =
+		object_id_from_bios_object_id(
+		le16_to_cpu(info_v11->extdispconninfo.path[i].connectorobjid));
+
+		info->ext_disp_conn_info.path[i].ext_encoder_obj_id =
+		object_id_from_bios_object_id(
+			le16_to_cpu(
+			info_v11->extdispconninfo.path[i].ext_encoder_objid));
+
+		info->ext_disp_conn_info.path[i].device_tag =
+			le16_to_cpu(
+				info_v11->extdispconninfo.path[i].device_tag);
+		info->ext_disp_conn_info.path[i].device_acpi_enum =
+		le16_to_cpu(
+			info_v11->extdispconninfo.path[i].device_acpi_enum);
+		info->ext_disp_conn_info.path[i].ext_aux_ddc_lut_index =
+			info_v11->extdispconninfo.path[i].auxddclut_index;
+		info->ext_disp_conn_info.path[i].ext_hpd_pin_lut_index =
+			info_v11->extdispconninfo.path[i].hpdlut_index;
+		info->ext_disp_conn_info.path[i].channel_mapping.raw =
+			info_v11->extdispconninfo.path[i].channelmapping;
+		info->ext_disp_conn_info.path[i].caps =
+				le16_to_cpu(info_v11->extdispconninfo.path[i].caps);
+	}
+	info->ext_disp_conn_info.checksum =
+	info_v11->extdispconninfo.checksum;
+
+	info->dp0_ext_hdmi_slv_addr = info_v11->dp0_retimer_set.HdmiSlvAddr;
+	info->dp0_ext_hdmi_reg_num = info_v11->dp0_retimer_set.HdmiRegNum;
+	for (i = 0; i < info->dp0_ext_hdmi_reg_num; i++) {
+		info->dp0_ext_hdmi_reg_settings[i].i2c_reg_index =
+				info_v11->dp0_retimer_set.HdmiRegSetting[i].ucI2cRegIndex;
+		info->dp0_ext_hdmi_reg_settings[i].i2c_reg_val =
+				info_v11->dp0_retimer_set.HdmiRegSetting[i].ucI2cRegVal;
+	}
+	info->dp0_ext_hdmi_6g_reg_num = info_v11->dp0_retimer_set.Hdmi6GRegNum;
+	for (i = 0; i < info->dp0_ext_hdmi_6g_reg_num; i++) {
+		info->dp0_ext_hdmi_6g_reg_settings[i].i2c_reg_index =
+				info_v11->dp0_retimer_set.Hdmi6GhzRegSetting[i].ucI2cRegIndex;
+		info->dp0_ext_hdmi_6g_reg_settings[i].i2c_reg_val =
+				info_v11->dp0_retimer_set.Hdmi6GhzRegSetting[i].ucI2cRegVal;
+	}
+
+	info->dp1_ext_hdmi_slv_addr = info_v11->dp1_retimer_set.HdmiSlvAddr;
+	info->dp1_ext_hdmi_reg_num = info_v11->dp1_retimer_set.HdmiRegNum;
+	for (i = 0; i < info->dp1_ext_hdmi_reg_num; i++) {
+		info->dp1_ext_hdmi_reg_settings[i].i2c_reg_index =
+				info_v11->dp1_retimer_set.HdmiRegSetting[i].ucI2cRegIndex;
+		info->dp1_ext_hdmi_reg_settings[i].i2c_reg_val =
+				info_v11->dp1_retimer_set.HdmiRegSetting[i].ucI2cRegVal;
+	}
+	info->dp1_ext_hdmi_6g_reg_num = info_v11->dp1_retimer_set.Hdmi6GRegNum;
+	for (i = 0; i < info->dp1_ext_hdmi_6g_reg_num; i++) {
+		info->dp1_ext_hdmi_6g_reg_settings[i].i2c_reg_index =
+				info_v11->dp1_retimer_set.Hdmi6GhzRegSetting[i].ucI2cRegIndex;
+		info->dp1_ext_hdmi_6g_reg_settings[i].i2c_reg_val =
+				info_v11->dp1_retimer_set.Hdmi6GhzRegSetting[i].ucI2cRegVal;
+	}
+
+	info->dp2_ext_hdmi_slv_addr = info_v11->dp2_retimer_set.HdmiSlvAddr;
+	info->dp2_ext_hdmi_reg_num = info_v11->dp2_retimer_set.HdmiRegNum;
+	for (i = 0; i < info->dp2_ext_hdmi_reg_num; i++) {
+		info->dp2_ext_hdmi_reg_settings[i].i2c_reg_index =
+				info_v11->dp2_retimer_set.HdmiRegSetting[i].ucI2cRegIndex;
+		info->dp2_ext_hdmi_reg_settings[i].i2c_reg_val =
+				info_v11->dp2_retimer_set.HdmiRegSetting[i].ucI2cRegVal;
+	}
+	info->dp2_ext_hdmi_6g_reg_num = info_v11->dp2_retimer_set.Hdmi6GRegNum;
+	for (i = 0; i < info->dp2_ext_hdmi_6g_reg_num; i++) {
+		info->dp2_ext_hdmi_6g_reg_settings[i].i2c_reg_index =
+				info_v11->dp2_retimer_set.Hdmi6GhzRegSetting[i].ucI2cRegIndex;
+		info->dp2_ext_hdmi_6g_reg_settings[i].i2c_reg_val =
+				info_v11->dp2_retimer_set.Hdmi6GhzRegSetting[i].ucI2cRegVal;
+	}
+
+	info->dp3_ext_hdmi_slv_addr = info_v11->dp3_retimer_set.HdmiSlvAddr;
+	info->dp3_ext_hdmi_reg_num = info_v11->dp3_retimer_set.HdmiRegNum;
+	for (i = 0; i < info->dp3_ext_hdmi_reg_num; i++) {
+		info->dp3_ext_hdmi_reg_settings[i].i2c_reg_index =
+				info_v11->dp3_retimer_set.HdmiRegSetting[i].ucI2cRegIndex;
+		info->dp3_ext_hdmi_reg_settings[i].i2c_reg_val =
+				info_v11->dp3_retimer_set.HdmiRegSetting[i].ucI2cRegVal;
+	}
+	info->dp3_ext_hdmi_6g_reg_num = info_v11->dp3_retimer_set.Hdmi6GRegNum;
+	for (i = 0; i < info->dp3_ext_hdmi_6g_reg_num; i++) {
+		info->dp3_ext_hdmi_6g_reg_settings[i].i2c_reg_index =
+				info_v11->dp3_retimer_set.Hdmi6GhzRegSetting[i].ucI2cRegIndex;
+		info->dp3_ext_hdmi_6g_reg_settings[i].i2c_reg_val =
+				info_v11->dp3_retimer_set.Hdmi6GhzRegSetting[i].ucI2cRegVal;
+	}
+
+
+	/** TODO - review **/
+	#if 0
+	info->boot_up_engine_clock = le32_to_cpu(info_v11->ulBootUpEngineClock)
+									* 10;
+	info->dentist_vco_freq = le32_to_cpu(info_v11->ulDentistVCOFreq) * 10;
+	info->boot_up_uma_clock = le32_to_cpu(info_v8->ulBootUpUMAClock) * 10;
+
+	for (i = 0; i < NUMBER_OF_DISP_CLK_VOLTAGE; ++i) {
+		/* Convert [10KHz] into [KHz] */
+		info->disp_clk_voltage[i].max_supported_clk =
+		le32_to_cpu(info_v11->sDISPCLK_Voltage[i].
+			ulMaximumSupportedCLK) * 10;
+		info->disp_clk_voltage[i].voltage_index =
+		le32_to_cpu(info_v11->sDISPCLK_Voltage[i].ulVoltageIndex);
+	}
+
+	info->boot_up_req_display_vector =
+			le32_to_cpu(info_v11->ulBootUpReqDisplayVector);
+	info->boot_up_nb_voltage =
+			le16_to_cpu(info_v11->usBootUpNBVoltage);
+	info->ext_disp_conn_info_offset =
+			le16_to_cpu(info_v11->usExtDispConnInfoOffset);
+	info->gmc_restore_reset_time =
+			le32_to_cpu(info_v11->ulGMCRestoreResetTime);
+	info->minimum_n_clk =
+			le32_to_cpu(info_v11->ulNbpStateNClkFreq[0]);
+	for (i = 1; i < 4; ++i)
+		info->minimum_n_clk =
+				info->minimum_n_clk <
+				le32_to_cpu(info_v11->ulNbpStateNClkFreq[i]) ?
+				info->minimum_n_clk : le32_to_cpu(
+					info_v11->ulNbpStateNClkFreq[i]);
+
+	info->idle_n_clk = le32_to_cpu(info_v11->ulIdleNClk);
+	info->ddr_dll_power_up_time =
+	    le32_to_cpu(info_v11->ulDDR_DLL_PowerUpTime);
+	info->ddr_pll_power_up_time =
+		le32_to_cpu(info_v11->ulDDR_PLL_PowerUpTime);
+	info->pcie_clk_ss_type = le16_to_cpu(info_v11->usPCIEClkSSType);
+	info->max_lvds_pclk_freq_in_single_link =
+		le16_to_cpu(info_v11->usMaxLVDSPclkFreqInSingleLink);
+	info->max_lvds_pclk_freq_in_single_link =
+		le16_to_cpu(info_v11->usMaxLVDSPclkFreqInSingleLink);
+	info->lvds_pwr_on_seq_dig_on_to_de_in_4ms =
+		info_v11->ucLVDSPwrOnSeqDIGONtoDE_in4Ms;
+	info->lvds_pwr_on_seq_de_to_vary_bl_in_4ms =
+		info_v11->ucLVDSPwrOnSeqDEtoVARY_BL_in4Ms;
+	info->lvds_pwr_on_seq_vary_bl_to_blon_in_4ms =
+		info_v11->ucLVDSPwrOnSeqVARY_BLtoBLON_in4Ms;
+	info->lvds_pwr_off_seq_vary_bl_to_de_in4ms =
+		info_v11->ucLVDSPwrOffSeqVARY_BLtoDE_in4Ms;
+	info->lvds_pwr_off_seq_de_to_dig_on_in4ms =
+		info_v11->ucLVDSPwrOffSeqDEtoDIGON_in4Ms;
+	info->lvds_pwr_off_seq_blon_to_vary_bl_in_4ms =
+		info_v11->ucLVDSPwrOffSeqBLONtoVARY_BL_in4Ms;
+	info->lvds_off_to_on_delay_in_4ms =
+		info_v11->ucLVDSOffToOnDelay_in4Ms;
+	info->lvds_bit_depth_control_val =
+		le32_to_cpu(info_v11->ulLCDBitDepthControlVal);
+
+	for (i = 0; i < NUMBER_OF_AVAILABLE_SCLK; ++i) {
+		/* Convert [10KHz] into [KHz] */
+		info->avail_s_clk[i].supported_s_clk =
+			le32_to_cpu(info_v11->sAvail_SCLK[i].ulSupportedSCLK)
+									* 10;
+		info->avail_s_clk[i].voltage_index =
+			le16_to_cpu(info_v11->sAvail_SCLK[i].usVoltageIndex);
+		info->avail_s_clk[i].voltage_id =
+			le16_to_cpu(info_v11->sAvail_SCLK[i].usVoltageID);
+	}
+	#endif /* TODO*/
+
+	return BP_RESULT_OK;
+}
+
+static enum bp_result get_integrated_info_v2_1(
+	struct bios_parser *bp,
+	struct integrated_info *info)
+{
+	struct atom_integrated_system_info_v2_1 *info_v2_1;
+	uint32_t i;
+
+	info_v2_1 = GET_IMAGE(struct atom_integrated_system_info_v2_1,
+					DATA_TABLES(integratedsysteminfo));
+
+	if (info_v2_1 == NULL)
+		return BP_RESULT_BADBIOSTABLE;
+
+	DC_LOG_BIOS("gpuclk_ss_percentage (unit of 0.001 percent): %d\n", info_v2_1->gpuclk_ss_percentage);
+
+	info->gpu_cap_info =
+	le32_to_cpu(info_v2_1->gpucapinfo);
+	/*
+	* system_config: Bit[0] = 0 : PCIE power gating disabled
+	*                       = 1 : PCIE power gating enabled
+	*                Bit[1] = 0 : DDR-PLL shut down disabled
+	*                       = 1 : DDR-PLL shut down enabled
+	*                Bit[2] = 0 : DDR-PLL power down disabled
+	*                       = 1 : DDR-PLL power down enabled
+	*/
+	info->system_config = le32_to_cpu(info_v2_1->system_config);
+	info->cpu_cap_info = le32_to_cpu(info_v2_1->cpucapinfo);
+	info->memory_type = info_v2_1->memorytype;
+	info->ma_channel_number = info_v2_1->umachannelnumber;
+	info->dp_ss_control =
+		le16_to_cpu(info_v2_1->reserved1);
+
+	for (i = 0; i < NUMBER_OF_UCHAR_FOR_GUID; ++i) {
+		info->ext_disp_conn_info.gu_id[i] =
+				info_v2_1->extdispconninfo.guid[i];
+	}
+
+	for (i = 0; i < MAX_NUMBER_OF_EXT_DISPLAY_PATH; ++i) {
+		info->ext_disp_conn_info.path[i].device_connector_id =
+		object_id_from_bios_object_id(
+		le16_to_cpu(info_v2_1->extdispconninfo.path[i].connectorobjid));
+
+		info->ext_disp_conn_info.path[i].ext_encoder_obj_id =
+		object_id_from_bios_object_id(
+			le16_to_cpu(
+			info_v2_1->extdispconninfo.path[i].ext_encoder_objid));
+
+		info->ext_disp_conn_info.path[i].device_tag =
+			le16_to_cpu(
+				info_v2_1->extdispconninfo.path[i].device_tag);
+		info->ext_disp_conn_info.path[i].device_acpi_enum =
+		le16_to_cpu(
+			info_v2_1->extdispconninfo.path[i].device_acpi_enum);
+		info->ext_disp_conn_info.path[i].ext_aux_ddc_lut_index =
+			info_v2_1->extdispconninfo.path[i].auxddclut_index;
+		info->ext_disp_conn_info.path[i].ext_hpd_pin_lut_index =
+			info_v2_1->extdispconninfo.path[i].hpdlut_index;
+		info->ext_disp_conn_info.path[i].channel_mapping.raw =
+			info_v2_1->extdispconninfo.path[i].channelmapping;
+		info->ext_disp_conn_info.path[i].caps =
+				le16_to_cpu(info_v2_1->extdispconninfo.path[i].caps);
+	}
+
+	info->ext_disp_conn_info.checksum =
+		info_v2_1->extdispconninfo.checksum;
+	info->dp0_ext_hdmi_slv_addr = info_v2_1->dp0_retimer_set.HdmiSlvAddr;
+	info->dp0_ext_hdmi_reg_num = info_v2_1->dp0_retimer_set.HdmiRegNum;
+	for (i = 0; i < info->dp0_ext_hdmi_reg_num; i++) {
+		info->dp0_ext_hdmi_reg_settings[i].i2c_reg_index =
+				info_v2_1->dp0_retimer_set.HdmiRegSetting[i].ucI2cRegIndex;
+		info->dp0_ext_hdmi_reg_settings[i].i2c_reg_val =
+				info_v2_1->dp0_retimer_set.HdmiRegSetting[i].ucI2cRegVal;
+	}
+	info->dp0_ext_hdmi_6g_reg_num = info_v2_1->dp0_retimer_set.Hdmi6GRegNum;
+	for (i = 0; i < info->dp0_ext_hdmi_6g_reg_num; i++) {
+		info->dp0_ext_hdmi_6g_reg_settings[i].i2c_reg_index =
+				info_v2_1->dp0_retimer_set.Hdmi6GhzRegSetting[i].ucI2cRegIndex;
+		info->dp0_ext_hdmi_6g_reg_settings[i].i2c_reg_val =
+				info_v2_1->dp0_retimer_set.Hdmi6GhzRegSetting[i].ucI2cRegVal;
+	}
+	info->dp1_ext_hdmi_slv_addr = info_v2_1->dp1_retimer_set.HdmiSlvAddr;
+	info->dp1_ext_hdmi_reg_num = info_v2_1->dp1_retimer_set.HdmiRegNum;
+	for (i = 0; i < info->dp1_ext_hdmi_reg_num; i++) {
+		info->dp1_ext_hdmi_reg_settings[i].i2c_reg_index =
+				info_v2_1->dp1_retimer_set.HdmiRegSetting[i].ucI2cRegIndex;
+		info->dp1_ext_hdmi_reg_settings[i].i2c_reg_val =
+				info_v2_1->dp1_retimer_set.HdmiRegSetting[i].ucI2cRegVal;
+	}
+	info->dp1_ext_hdmi_6g_reg_num = info_v2_1->dp1_retimer_set.Hdmi6GRegNum;
+	for (i = 0; i < info->dp1_ext_hdmi_6g_reg_num; i++) {
+		info->dp1_ext_hdmi_6g_reg_settings[i].i2c_reg_index =
+				info_v2_1->dp1_retimer_set.Hdmi6GhzRegSetting[i].ucI2cRegIndex;
+		info->dp1_ext_hdmi_6g_reg_settings[i].i2c_reg_val =
+				info_v2_1->dp1_retimer_set.Hdmi6GhzRegSetting[i].ucI2cRegVal;
+	}
+	info->dp2_ext_hdmi_slv_addr = info_v2_1->dp2_retimer_set.HdmiSlvAddr;
+	info->dp2_ext_hdmi_reg_num = info_v2_1->dp2_retimer_set.HdmiRegNum;
+	for (i = 0; i < info->dp2_ext_hdmi_reg_num; i++) {
+		info->dp2_ext_hdmi_reg_settings[i].i2c_reg_index =
+				info_v2_1->dp2_retimer_set.HdmiRegSetting[i].ucI2cRegIndex;
+		info->dp2_ext_hdmi_reg_settings[i].i2c_reg_val =
+				info_v2_1->dp2_retimer_set.HdmiRegSetting[i].ucI2cRegVal;
+	}
+	info->dp2_ext_hdmi_6g_reg_num = info_v2_1->dp2_retimer_set.Hdmi6GRegNum;
+	for (i = 0; i < info->dp2_ext_hdmi_6g_reg_num; i++) {
+		info->dp2_ext_hdmi_6g_reg_settings[i].i2c_reg_index =
+				info_v2_1->dp2_retimer_set.Hdmi6GhzRegSetting[i].ucI2cRegIndex;
+		info->dp2_ext_hdmi_6g_reg_settings[i].i2c_reg_val =
+				info_v2_1->dp2_retimer_set.Hdmi6GhzRegSetting[i].ucI2cRegVal;
+	}
+	info->dp3_ext_hdmi_slv_addr = info_v2_1->dp3_retimer_set.HdmiSlvAddr;
+	info->dp3_ext_hdmi_reg_num = info_v2_1->dp3_retimer_set.HdmiRegNum;
+	for (i = 0; i < info->dp3_ext_hdmi_reg_num; i++) {
+		info->dp3_ext_hdmi_reg_settings[i].i2c_reg_index =
+				info_v2_1->dp3_retimer_set.HdmiRegSetting[i].ucI2cRegIndex;
+		info->dp3_ext_hdmi_reg_settings[i].i2c_reg_val =
+				info_v2_1->dp3_retimer_set.HdmiRegSetting[i].ucI2cRegVal;
+	}
+	info->dp3_ext_hdmi_6g_reg_num = info_v2_1->dp3_retimer_set.Hdmi6GRegNum;
+	for (i = 0; i < info->dp3_ext_hdmi_6g_reg_num; i++) {
+		info->dp3_ext_hdmi_6g_reg_settings[i].i2c_reg_index =
+				info_v2_1->dp3_retimer_set.Hdmi6GhzRegSetting[i].ucI2cRegIndex;
+		info->dp3_ext_hdmi_6g_reg_settings[i].i2c_reg_val =
+				info_v2_1->dp3_retimer_set.Hdmi6GhzRegSetting[i].ucI2cRegVal;
+	}
+
+	info->edp1_info.edp_backlight_pwm_hz =
+	le16_to_cpu(info_v2_1->edp1_info.edp_backlight_pwm_hz);
+	info->edp1_info.edp_ss_percentage =
+	le16_to_cpu(info_v2_1->edp1_info.edp_ss_percentage);
+	info->edp1_info.edp_ss_rate_10hz =
+	le16_to_cpu(info_v2_1->edp1_info.edp_ss_rate_10hz);
+	info->edp1_info.edp_pwr_on_off_delay =
+		info_v2_1->edp1_info.edp_pwr_on_off_delay;
+	info->edp1_info.edp_pwr_on_vary_bl_to_blon =
+		info_v2_1->edp1_info.edp_pwr_on_vary_bl_to_blon;
+	info->edp1_info.edp_pwr_down_bloff_to_vary_bloff =
+		info_v2_1->edp1_info.edp_pwr_down_bloff_to_vary_bloff;
+	info->edp1_info.edp_panel_bpc =
+		info_v2_1->edp1_info.edp_panel_bpc;
+	info->edp1_info.edp_bootup_bl_level = info_v2_1->edp1_info.edp_bootup_bl_level;
+
+	info->edp2_info.edp_backlight_pwm_hz =
+	le16_to_cpu(info_v2_1->edp2_info.edp_backlight_pwm_hz);
+	info->edp2_info.edp_ss_percentage =
+	le16_to_cpu(info_v2_1->edp2_info.edp_ss_percentage);
+	info->edp2_info.edp_ss_rate_10hz =
+	le16_to_cpu(info_v2_1->edp2_info.edp_ss_rate_10hz);
+	info->edp2_info.edp_pwr_on_off_delay =
+		info_v2_1->edp2_info.edp_pwr_on_off_delay;
+	info->edp2_info.edp_pwr_on_vary_bl_to_blon =
+		info_v2_1->edp2_info.edp_pwr_on_vary_bl_to_blon;
+	info->edp2_info.edp_pwr_down_bloff_to_vary_bloff =
+		info_v2_1->edp2_info.edp_pwr_down_bloff_to_vary_bloff;
+	info->edp2_info.edp_panel_bpc =
+		info_v2_1->edp2_info.edp_panel_bpc;
+	info->edp2_info.edp_bootup_bl_level =
+		info_v2_1->edp2_info.edp_bootup_bl_level;
+
+	return BP_RESULT_OK;
+}
+
+static enum bp_result get_integrated_info_v2_2(
+	struct bios_parser *bp,
+	struct integrated_info *info)
+{
+	struct atom_integrated_system_info_v2_2 *info_v2_2;
+	uint32_t i;
+
+	info_v2_2 = GET_IMAGE(struct atom_integrated_system_info_v2_2,
+					DATA_TABLES(integratedsysteminfo));
+
+	if (info_v2_2 == NULL)
+		return BP_RESULT_BADBIOSTABLE;
+
+	DC_LOG_BIOS("gpuclk_ss_percentage (unit of 0.001 percent): %d\n", info_v2_2->gpuclk_ss_percentage);
+
+	info->gpu_cap_info =
+	le32_to_cpu(info_v2_2->gpucapinfo);
+	/*
+	* system_config: Bit[0] = 0 : PCIE power gating disabled
+	*                       = 1 : PCIE power gating enabled
+	*                Bit[1] = 0 : DDR-PLL shut down disabled
+	*                       = 1 : DDR-PLL shut down enabled
+	*                Bit[2] = 0 : DDR-PLL power down disabled
+	*                       = 1 : DDR-PLL power down enabled
+	*/
+	info->system_config = le32_to_cpu(info_v2_2->system_config);
+	info->cpu_cap_info = le32_to_cpu(info_v2_2->cpucapinfo);
+	info->memory_type = info_v2_2->memorytype;
+	info->ma_channel_number = info_v2_2->umachannelnumber;
+	info->dp_ss_control =
+		le16_to_cpu(info_v2_2->reserved1);
+	info->gpuclk_ss_percentage = info_v2_2->gpuclk_ss_percentage;
+	info->gpuclk_ss_type = info_v2_2->gpuclk_ss_type;
+
+	for (i = 0; i < NUMBER_OF_UCHAR_FOR_GUID; ++i) {
+		info->ext_disp_conn_info.gu_id[i] =
+				info_v2_2->extdispconninfo.guid[i];
+	}
+
+	for (i = 0; i < MAX_NUMBER_OF_EXT_DISPLAY_PATH; ++i) {
+		info->ext_disp_conn_info.path[i].device_connector_id =
+		object_id_from_bios_object_id(
+		le16_to_cpu(info_v2_2->extdispconninfo.path[i].connectorobjid));
+
+		info->ext_disp_conn_info.path[i].ext_encoder_obj_id =
+		object_id_from_bios_object_id(
+			le16_to_cpu(
+			info_v2_2->extdispconninfo.path[i].ext_encoder_objid));
+
+		info->ext_disp_conn_info.path[i].device_tag =
+			le16_to_cpu(
+				info_v2_2->extdispconninfo.path[i].device_tag);
+		info->ext_disp_conn_info.path[i].device_acpi_enum =
+		le16_to_cpu(
+			info_v2_2->extdispconninfo.path[i].device_acpi_enum);
+		info->ext_disp_conn_info.path[i].ext_aux_ddc_lut_index =
+			info_v2_2->extdispconninfo.path[i].auxddclut_index;
+		info->ext_disp_conn_info.path[i].ext_hpd_pin_lut_index =
+			info_v2_2->extdispconninfo.path[i].hpdlut_index;
+		info->ext_disp_conn_info.path[i].channel_mapping.raw =
+			info_v2_2->extdispconninfo.path[i].channelmapping;
+		info->ext_disp_conn_info.path[i].caps =
+				le16_to_cpu(info_v2_2->extdispconninfo.path[i].caps);
+	}
+
+	info->ext_disp_conn_info.checksum =
+		info_v2_2->extdispconninfo.checksum;
+	info->ext_disp_conn_info.fixdpvoltageswing =
+		info_v2_2->extdispconninfo.fixdpvoltageswing;
+
+	info->edp1_info.edp_backlight_pwm_hz =
+	le16_to_cpu(info_v2_2->edp1_info.edp_backlight_pwm_hz);
+	info->edp1_info.edp_ss_percentage =
+	le16_to_cpu(info_v2_2->edp1_info.edp_ss_percentage);
+	info->edp1_info.edp_ss_rate_10hz =
+	le16_to_cpu(info_v2_2->edp1_info.edp_ss_rate_10hz);
+	info->edp1_info.edp_pwr_on_off_delay =
+		info_v2_2->edp1_info.edp_pwr_on_off_delay;
+	info->edp1_info.edp_pwr_on_vary_bl_to_blon =
+		info_v2_2->edp1_info.edp_pwr_on_vary_bl_to_blon;
+	info->edp1_info.edp_pwr_down_bloff_to_vary_bloff =
+		info_v2_2->edp1_info.edp_pwr_down_bloff_to_vary_bloff;
+	info->edp1_info.edp_panel_bpc =
+		info_v2_2->edp1_info.edp_panel_bpc;
+	info->edp1_info.edp_bootup_bl_level =
+
+	info->edp2_info.edp_backlight_pwm_hz =
+	le16_to_cpu(info_v2_2->edp2_info.edp_backlight_pwm_hz);
+	info->edp2_info.edp_ss_percentage =
+	le16_to_cpu(info_v2_2->edp2_info.edp_ss_percentage);
+	info->edp2_info.edp_ss_rate_10hz =
+	le16_to_cpu(info_v2_2->edp2_info.edp_ss_rate_10hz);
+	info->edp2_info.edp_pwr_on_off_delay =
+		info_v2_2->edp2_info.edp_pwr_on_off_delay;
+	info->edp2_info.edp_pwr_on_vary_bl_to_blon =
+		info_v2_2->edp2_info.edp_pwr_on_vary_bl_to_blon;
+	info->edp2_info.edp_pwr_down_bloff_to_vary_bloff =
+		info_v2_2->edp2_info.edp_pwr_down_bloff_to_vary_bloff;
+	info->edp2_info.edp_panel_bpc =
+		info_v2_2->edp2_info.edp_panel_bpc;
+	info->edp2_info.edp_bootup_bl_level =
+		info_v2_2->edp2_info.edp_bootup_bl_level;
+
+	return BP_RESULT_OK;
+}
+
+/*
+ * construct_integrated_info
+ *
+ * @brief
+ * Get integrated BIOS information based on table revision
+ *
+ * @param
+ * bios_parser *bp - [in]BIOS parser handler to get master data table
+ * integrated_info *info - [out] store and output integrated info
+ *
+ * @return
+ * static enum bp_result - BP_RESULT_OK if information is available,
+ *                  BP_RESULT_BADBIOSTABLE otherwise.
+ */
+static enum bp_result construct_integrated_info(
+	struct bios_parser *bp,
+	struct integrated_info *info)
+{
+	static enum bp_result result = BP_RESULT_BADBIOSTABLE;
+
+	struct atom_common_table_header *header;
+	struct atom_data_revision revision;
+
+	int32_t i;
+	int32_t j;
+
+	if (!info)
+		return result;
+
+	if (info && DATA_TABLES(integratedsysteminfo)) {
+		header = GET_IMAGE(struct atom_common_table_header,
+					DATA_TABLES(integratedsysteminfo));
+
+		get_atom_data_table_revision(header, &revision);
+
+		switch (revision.major) {
+		case 1:
+			switch (revision.minor) {
+			case 11:
+			case 12:
+				result = get_integrated_info_v11(bp, info);
+				break;
+			default:
+				return result;
+			}
+			break;
+		case 2:
+			switch (revision.minor) {
+			case 1:
+				result = get_integrated_info_v2_1(bp, info);
+				break;
+			case 2:
+			case 3:
+				result = get_integrated_info_v2_2(bp, info);
+				break;
+			default:
+				return result;
+			}
+			break;
+		default:
+			return result;
+		}
+		if (result == BP_RESULT_OK) {
+
+			DC_LOG_BIOS("edp1:\n"
+						"\tedp_pwr_on_off_delay = %d\n"
+						"\tedp_pwr_on_vary_bl_to_blon = %d\n"
+						"\tedp_pwr_down_bloff_to_vary_bloff = %d\n"
+						"\tedp_bootup_bl_level = %d\n",
+						info->edp1_info.edp_pwr_on_off_delay,
+						info->edp1_info.edp_pwr_on_vary_bl_to_blon,
+						info->edp1_info.edp_pwr_down_bloff_to_vary_bloff,
+						info->edp1_info.edp_bootup_bl_level);
+			DC_LOG_BIOS("edp2:\n"
+						"\tedp_pwr_on_off_delayv = %d\n"
+						"\tedp_pwr_on_vary_bl_to_blon = %d\n"
+						"\tedp_pwr_down_bloff_to_vary_bloff = %d\n"
+						"\tedp_bootup_bl_level = %d\n",
+						info->edp2_info.edp_pwr_on_off_delay,
+						info->edp2_info.edp_pwr_on_vary_bl_to_blon,
+						info->edp2_info.edp_pwr_down_bloff_to_vary_bloff,
+						info->edp2_info.edp_bootup_bl_level);
+		}
+	}
+
+	if (result != BP_RESULT_OK)
+		return result;
+	else {
+		// Log each external path
+		for (i = 0; i < MAX_NUMBER_OF_EXT_DISPLAY_PATH; i++) {
+			if (info->ext_disp_conn_info.path[i].device_tag != 0)
+				DC_LOG_BIOS("integrated_info:For EXTERNAL DISPLAY PATH %d --------------\n"
+						"DEVICE_TAG: 0x%x\n"
+						"DEVICE_ACPI_ENUM: 0x%x\n"
+						"DEVICE_CONNECTOR_ID: 0x%x\n"
+						"EXT_AUX_DDC_LUT_INDEX: %d\n"
+						"EXT_HPD_PIN_LUT_INDEX: %d\n"
+						"EXT_ENCODER_OBJ_ID: 0x%x\n"
+						"Encoder CAPS: 0x%x\n",
+						i,
+						info->ext_disp_conn_info.path[i].device_tag,
+						info->ext_disp_conn_info.path[i].device_acpi_enum,
+						info->ext_disp_conn_info.path[i].device_connector_id.id,
+						info->ext_disp_conn_info.path[i].ext_aux_ddc_lut_index,
+						info->ext_disp_conn_info.path[i].ext_hpd_pin_lut_index,
+						info->ext_disp_conn_info.path[i].ext_encoder_obj_id.id,
+						info->ext_disp_conn_info.path[i].caps
+						);
+			if ((info->ext_disp_conn_info.path[i].caps & AMD_EXT_DISPLAY_PATH_CAPS__EXT_CHIP_MASK) == AMD_EXT_DISPLAY_PATH_CAPS__DP_FIXED_VS_EN)
+				DC_LOG_BIOS("BIOS AMD_EXT_DISPLAY_PATH_CAPS__DP_FIXED_VS_EN on path %d\n", i);
+			else if (bp->base.ctx->dc->config.force_bios_fixed_vs) {
+				info->ext_disp_conn_info.path[i].caps &= ~AMD_EXT_DISPLAY_PATH_CAPS__EXT_CHIP_MASK;
+				info->ext_disp_conn_info.path[i].caps |= AMD_EXT_DISPLAY_PATH_CAPS__DP_FIXED_VS_EN;
+				DC_LOG_BIOS("driver forced AMD_EXT_DISPLAY_PATH_CAPS__DP_FIXED_VS_EN on path %d\n", i);
+			}
+		}
+		// Log the Checksum and Voltage Swing
+		DC_LOG_BIOS("Integrated info table CHECKSUM: %d\n"
+					"Integrated info table FIX_DP_VOLTAGE_SWING: %d\n",
+					info->ext_disp_conn_info.checksum,
+					info->ext_disp_conn_info.fixdpvoltageswing);
+		if (bp->base.ctx->dc->config.force_bios_fixed_vs && info->ext_disp_conn_info.fixdpvoltageswing == 0) {
+			info->ext_disp_conn_info.fixdpvoltageswing = bp->base.ctx->dc->config.force_bios_fixed_vs & 0xF;
+			DC_LOG_BIOS("driver forced fixdpvoltageswing = %d\n", info->ext_disp_conn_info.fixdpvoltageswing);
+		}
+	}
+	/* Sort voltage table from low to high*/
+	for (i = 1; i < NUMBER_OF_DISP_CLK_VOLTAGE; ++i) {
+		for (j = i; j > 0; --j) {
+			if (info->disp_clk_voltage[j].max_supported_clk <
+			    info->disp_clk_voltage[j-1].max_supported_clk)
+				swap(info->disp_clk_voltage[j-1], info->disp_clk_voltage[j]);
+		}
+	}
+
+	return result;
+}
+
+static enum bp_result bios_parser_get_vram_info(
+		struct dc_bios *dcb,
+		struct dc_vram_info *info)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	enum bp_result result = BP_RESULT_BADBIOSTABLE;
+	struct atom_common_table_header *header;
+	struct atom_data_revision revision;
+
+	// vram info moved to umc_info for DCN4x
+	if (info && DATA_TABLES(umc_info)) {
+		header = GET_IMAGE(struct atom_common_table_header,
+					DATA_TABLES(umc_info));
+
+		get_atom_data_table_revision(header, &revision);
+
+		switch (revision.major) {
+		case 4:
+			switch (revision.minor) {
+			case 0:
+				result = get_vram_info_from_umc_info_v40(bp, info);
+				break;
+			default:
+				break;
+			}
+			break;
+		default:
+			break;
+		}
+	}
+
+	if (result != BP_RESULT_OK && info && DATA_TABLES(vram_info)) {
+		header = GET_IMAGE(struct atom_common_table_header,
+					DATA_TABLES(vram_info));
+
+		get_atom_data_table_revision(header, &revision);
+
+		switch (revision.major) {
+		case 2:
+			switch (revision.minor) {
+			case 3:
+				result = get_vram_info_v23(bp, info);
+				break;
+			case 4:
+				result = get_vram_info_v24(bp, info);
+				break;
+			case 5:
+				result = get_vram_info_v25(bp, info);
+				break;
+			default:
+				break;
+			}
+			break;
+
+		case 3:
+			switch (revision.minor) {
+			case 0:
+				result = get_vram_info_v30(bp, info);
+				break;
+			default:
+				break;
+			}
+			break;
+
+		default:
+			return result;
+		}
+
+	}
+	return result;
+}
+
+static struct integrated_info *bios_parser_create_integrated_info(
+	struct dc_bios *dcb)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	struct integrated_info *info;
+
+	info = kzalloc(sizeof(struct integrated_info), GFP_KERNEL);
+
+	if (info == NULL) {
+		ASSERT_CRITICAL(0);
+		return NULL;
+	}
+
+	if (construct_integrated_info(bp, info) == BP_RESULT_OK)
+		return info;
+
+	kfree(info);
+
+	return NULL;
+}
+
+static enum bp_result update_slot_layout_info(
+	struct dc_bios *dcb,
+	unsigned int i,
+	struct slot_layout_info *slot_layout_info)
+{
+	unsigned int record_offset;
+	unsigned int j;
+	struct atom_display_object_path_v2 *object;
+	struct atom_bracket_layout_record *record;
+	struct atom_common_record_header *record_header;
+	static enum bp_result result;
+	struct bios_parser *bp;
+	struct object_info_table *tbl;
+	struct display_object_info_table_v1_4 *v1_4;
+
+	record = NULL;
+	record_header = NULL;
+	result = BP_RESULT_NORECORD;
+
+	bp = BP_FROM_DCB(dcb);
+	tbl = &bp->object_info_tbl;
+	v1_4 = tbl->v1_4;
+
+	object = &v1_4->display_path[i];
+	record_offset = (unsigned int)
+		(object->disp_recordoffset) +
+		(unsigned int)(bp->object_info_tbl_offset);
+
+	for (;;) {
+
+		record_header = (struct atom_common_record_header *)
+			GET_IMAGE(struct atom_common_record_header,
+			record_offset);
+		if (record_header == NULL) {
+			result = BP_RESULT_BADBIOSTABLE;
+			break;
+		}
+
+		/* the end of the list */
+		if (record_header->record_type == 0xff ||
+			record_header->record_size == 0)	{
+			break;
+		}
+
+		if (record_header->record_type ==
+			ATOM_BRACKET_LAYOUT_RECORD_TYPE &&
+			sizeof(struct atom_bracket_layout_record)
+			<= record_header->record_size) {
+			record = (struct atom_bracket_layout_record *)
+				(record_header);
+			result = BP_RESULT_OK;
+			break;
+		}
+
+		record_offset += record_header->record_size;
+	}
+
+	/* return if the record not found */
+	if (result != BP_RESULT_OK)
+		return result;
+
+	/* get slot sizes */
+	slot_layout_info->length = record->bracketlen;
+	slot_layout_info->width = record->bracketwidth;
+
+	/* get info for each connector in the slot */
+	slot_layout_info->num_of_connectors = record->conn_num;
+	for (j = 0; j < slot_layout_info->num_of_connectors; ++j) {
+		slot_layout_info->connectors[j].connector_type =
+			(enum connector_layout_type)
+			(record->conn_info[j].connector_type);
+		switch (record->conn_info[j].connector_type) {
+		case CONNECTOR_TYPE_DVI_D:
+			slot_layout_info->connectors[j].connector_type =
+				CONNECTOR_LAYOUT_TYPE_DVI_D;
+			slot_layout_info->connectors[j].length =
+				CONNECTOR_SIZE_DVI;
+			break;
+
+		case CONNECTOR_TYPE_HDMI:
+			slot_layout_info->connectors[j].connector_type =
+				CONNECTOR_LAYOUT_TYPE_HDMI;
+			slot_layout_info->connectors[j].length =
+				CONNECTOR_SIZE_HDMI;
+			break;
+
+		case CONNECTOR_TYPE_DISPLAY_PORT:
+			slot_layout_info->connectors[j].connector_type =
+				CONNECTOR_LAYOUT_TYPE_DP;
+			slot_layout_info->connectors[j].length =
+				CONNECTOR_SIZE_DP;
+			break;
+
+		case CONNECTOR_TYPE_MINI_DISPLAY_PORT:
+			slot_layout_info->connectors[j].connector_type =
+				CONNECTOR_LAYOUT_TYPE_MINI_DP;
+			slot_layout_info->connectors[j].length =
+				CONNECTOR_SIZE_MINI_DP;
+			break;
+
+		default:
+			slot_layout_info->connectors[j].connector_type =
+				CONNECTOR_LAYOUT_TYPE_UNKNOWN;
+			slot_layout_info->connectors[j].length =
+				CONNECTOR_SIZE_UNKNOWN;
+		}
+
+		slot_layout_info->connectors[j].position =
+			record->conn_info[j].position;
+		slot_layout_info->connectors[j].connector_id =
+			object_id_from_bios_object_id(
+				record->conn_info[j].connectorobjid);
+	}
+	return result;
+}
+
+static enum bp_result update_slot_layout_info_v2(
+	struct dc_bios *dcb,
+	unsigned int i,
+	struct slot_layout_info *slot_layout_info)
+{
+	unsigned int record_offset;
+	struct atom_display_object_path_v3 *object;
+	struct atom_bracket_layout_record_v2 *record;
+	struct atom_common_record_header *record_header;
+	static enum bp_result result;
+	struct bios_parser *bp;
+	struct object_info_table *tbl;
+	struct display_object_info_table_v1_5 *v1_5;
+	struct graphics_object_id connector_id;
+
+	record = NULL;
+	record_header = NULL;
+	result = BP_RESULT_NORECORD;
+
+	bp = BP_FROM_DCB(dcb);
+	tbl = &bp->object_info_tbl;
+	v1_5 = tbl->v1_5;
+
+	object = &v1_5->display_path[i];
+	record_offset = (unsigned int)
+		(object->disp_recordoffset) +
+		(unsigned int)(bp->object_info_tbl_offset);
+
+	for (;;) {
+
+		record_header = (struct atom_common_record_header *)
+			GET_IMAGE(struct atom_common_record_header,
+			record_offset);
+		if (record_header == NULL) {
+			result = BP_RESULT_BADBIOSTABLE;
+			break;
+		}
+
+		/* the end of the list */
+		if (record_header->record_type == ATOM_RECORD_END_TYPE ||
+			record_header->record_size == 0)	{
+			break;
+		}
+
+		if (record_header->record_type ==
+			ATOM_BRACKET_LAYOUT_V2_RECORD_TYPE &&
+			sizeof(struct atom_bracket_layout_record_v2)
+			<= record_header->record_size) {
+			record = (struct atom_bracket_layout_record_v2 *)
+				(record_header);
+			result = BP_RESULT_OK;
+			break;
+		}
+
+		record_offset += record_header->record_size;
+	}
+
+	/* return if the record not found */
+	if (result != BP_RESULT_OK)
+		return result;
+
+	/* get slot sizes */
+	connector_id = object_id_from_bios_object_id(object->display_objid);
+
+	slot_layout_info->length = record->bracketlen;
+	slot_layout_info->width = record->bracketwidth;
+	slot_layout_info->num_of_connectors = v1_5->number_of_path;
+	slot_layout_info->connectors[i].position = record->conn_num;
+	slot_layout_info->connectors[i].connector_id = connector_id;
+
+	switch (connector_id.id) {
+	case CONNECTOR_ID_SINGLE_LINK_DVID:
+	case CONNECTOR_ID_DUAL_LINK_DVID:
+		slot_layout_info->connectors[i].connector_type = CONNECTOR_LAYOUT_TYPE_DVI_D;
+		slot_layout_info->connectors[i].length = CONNECTOR_SIZE_DVI;
+		break;
+
+	case CONNECTOR_ID_HDMI_TYPE_A:
+		slot_layout_info->connectors[i].connector_type = CONNECTOR_LAYOUT_TYPE_HDMI;
+		slot_layout_info->connectors[i].length = CONNECTOR_SIZE_HDMI;
+		break;
+
+	case CONNECTOR_ID_DISPLAY_PORT:
+	case CONNECTOR_ID_USBC:
+		if (record->mini_type == MINI_TYPE_NORMAL) {
+			slot_layout_info->connectors[i].connector_type = CONNECTOR_LAYOUT_TYPE_DP;
+			slot_layout_info->connectors[i].length = CONNECTOR_SIZE_DP;
+		} else {
+			slot_layout_info->connectors[i].connector_type = CONNECTOR_LAYOUT_TYPE_MINI_DP;
+			slot_layout_info->connectors[i].length = CONNECTOR_SIZE_MINI_DP;
+		}
+		break;
+
+	default:
+		slot_layout_info->connectors[i].connector_type = CONNECTOR_LAYOUT_TYPE_UNKNOWN;
+		slot_layout_info->connectors[i].length = CONNECTOR_SIZE_UNKNOWN;
+	}
+	return result;
+}
+
+static enum bp_result get_bracket_layout_record(
+	struct dc_bios *dcb,
+	unsigned int bracket_layout_id,
+	struct slot_layout_info *slot_layout_info)
+{
+	unsigned int i;
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	static enum bp_result result;
+	struct object_info_table *tbl;
+	struct display_object_info_table_v1_4 *v1_4;
+	struct display_object_info_table_v1_5 *v1_5;
+
+	if (slot_layout_info == NULL) {
+		DC_LOG_DETECTION_EDID_PARSER("Invalid slot_layout_info\n");
+		return BP_RESULT_BADINPUT;
+	}
+
+	tbl = &bp->object_info_tbl;
+	v1_4 = tbl->v1_4;
+	v1_5 = tbl->v1_5;
+
+	result = BP_RESULT_NORECORD;
+	switch (bp->object_info_tbl.revision.minor) {
+	case 4:
+	default:
+		for (i = 0; i < v1_4->number_of_path; ++i) {
+			if (bracket_layout_id == v1_4->display_path[i].display_objid) {
+				result = update_slot_layout_info(dcb, i, slot_layout_info);
+				break;
+			}
+		}
+		break;
+	case 5:
+		for (i = 0; i < v1_5->number_of_path; ++i)
+			result = update_slot_layout_info_v2(dcb, i, slot_layout_info);
+		break;
+	}
+
+	return result;
+}
+
+static enum bp_result bios_get_board_layout_info(
+	struct dc_bios *dcb,
+	struct board_layout_info *board_layout_info)
+{
+	unsigned int i;
+	struct bios_parser *bp;
+	static enum bp_result record_result;
+	unsigned int max_slots;
+
+	const unsigned int slot_index_to_vbios_id[MAX_BOARD_SLOTS] = {
+		GENERICOBJECT_BRACKET_LAYOUT_ENUM_ID1,
+		GENERICOBJECT_BRACKET_LAYOUT_ENUM_ID2,
+		0, 0
+	};
+
+	bp = BP_FROM_DCB(dcb);
+
+	if (board_layout_info == NULL) {
+		DC_LOG_DETECTION_EDID_PARSER("Invalid board_layout_info\n");
+		return BP_RESULT_BADINPUT;
+	}
+
+	board_layout_info->num_of_slots = 0;
+	max_slots = MAX_BOARD_SLOTS;
+
+	// Assume single slot on v1_5
+	if (bp->object_info_tbl.revision.minor == 5) {
+		max_slots = 1;
+	}
+
+	for (i = 0; i < max_slots; ++i) {
+		record_result = get_bracket_layout_record(dcb,
+			slot_index_to_vbios_id[i],
+			&board_layout_info->slots[i]);
+
+		if (record_result == BP_RESULT_NORECORD && i > 0)
+			break; /* no more slots present in bios */
+		else if (record_result != BP_RESULT_OK)
+			return record_result;  /* fail */
+
+		++board_layout_info->num_of_slots;
+	}
+
+	/* all data is valid */
+	board_layout_info->is_number_of_slots_valid = 1;
+	board_layout_info->is_slots_size_valid = 1;
+	board_layout_info->is_connector_offsets_valid = 1;
+	board_layout_info->is_connector_lengths_valid = 1;
+
+	return BP_RESULT_OK;
+}
+
+
+static uint16_t bios_parser_pack_data_tables(
+	struct dc_bios *dcb,
+	void *dst)
+{
+	// TODO: There is data bytes alignment issue, disable it for now.
+	return 0;
+}
+
+static struct atom_dc_golden_table_v1 *bios_get_golden_table(
+		struct bios_parser *bp,
+		uint32_t rev_major,
+		uint32_t rev_minor,
+		uint16_t *dc_golden_table_ver)
+{
+	struct atom_display_controller_info_v4_4 *disp_cntl_tbl_4_4 = NULL;
+	uint32_t dc_golden_offset = 0;
+	*dc_golden_table_ver = 0;
+
+	if (!DATA_TABLES(dce_info))
+		return NULL;
+
+	/* ver.4.4 or higher */
+	switch (rev_major) {
+	case 4:
+		switch (rev_minor) {
+		case 4:
+			disp_cntl_tbl_4_4 = GET_IMAGE(struct atom_display_controller_info_v4_4,
+									DATA_TABLES(dce_info));
+			if (!disp_cntl_tbl_4_4)
+				return NULL;
+			dc_golden_offset = DATA_TABLES(dce_info) + disp_cntl_tbl_4_4->dc_golden_table_offset;
+			*dc_golden_table_ver = disp_cntl_tbl_4_4->dc_golden_table_ver;
+			break;
+		case 5:
+		default:
+			/* For atom_display_controller_info_v4_5 there is no need to get golden table from
+			 * dc_golden_table_offset as all these fields previously in golden table used for AUX
+			 * pre-charge settings are now available directly in atom_display_controller_info_v4_5.
+			 */
+			break;
+		}
+		break;
+	}
+
+	if (!dc_golden_offset)
+		return NULL;
+
+	if (*dc_golden_table_ver != 1)
+		return NULL;
+
+	return GET_IMAGE(struct atom_dc_golden_table_v1,
+			dc_golden_offset);
+}
+
+static enum bp_result bios_get_atom_dc_golden_table(
+	struct dc_bios *dcb)
+{
+	struct bios_parser *bp = BP_FROM_DCB(dcb);
+	enum bp_result result = BP_RESULT_OK;
+	struct atom_dc_golden_table_v1 *atom_dc_golden_table = NULL;
+	struct atom_common_table_header *header;
+	struct atom_data_revision tbl_revision;
+	uint16_t dc_golden_table_ver = 0;
+
+	header = GET_IMAGE(struct atom_common_table_header,
+							DATA_TABLES(dce_info));
+	if (!header)
+		return BP_RESULT_UNSUPPORTED;
+
+	get_atom_data_table_revision(header, &tbl_revision);
+
+	atom_dc_golden_table = bios_get_golden_table(bp,
+			tbl_revision.major,
+			tbl_revision.minor,
+			&dc_golden_table_ver);
+
+	if (!atom_dc_golden_table)
+		return BP_RESULT_UNSUPPORTED;
+
+	dcb->golden_table.dc_golden_table_ver = dc_golden_table_ver;
+	dcb->golden_table.aux_dphy_rx_control0_val = atom_dc_golden_table->aux_dphy_rx_control0_val;
+	dcb->golden_table.aux_dphy_rx_control1_val = atom_dc_golden_table->aux_dphy_rx_control1_val;
+	dcb->golden_table.aux_dphy_tx_control_val = atom_dc_golden_table->aux_dphy_tx_control_val;
+	dcb->golden_table.dc_gpio_aux_ctrl_0_val = atom_dc_golden_table->dc_gpio_aux_ctrl_0_val;
+	dcb->golden_table.dc_gpio_aux_ctrl_1_val = atom_dc_golden_table->dc_gpio_aux_ctrl_1_val;
+	dcb->golden_table.dc_gpio_aux_ctrl_2_val = atom_dc_golden_table->dc_gpio_aux_ctrl_2_val;
+	dcb->golden_table.dc_gpio_aux_ctrl_3_val = atom_dc_golden_table->dc_gpio_aux_ctrl_3_val;
+	dcb->golden_table.dc_gpio_aux_ctrl_4_val = atom_dc_golden_table->dc_gpio_aux_ctrl_4_val;
+	dcb->golden_table.dc_gpio_aux_ctrl_5_val = atom_dc_golden_table->dc_gpio_aux_ctrl_5_val;
+
+	return result;
+}
+
+
+static const struct dc_vbios_funcs vbios_funcs = {
+	.get_connectors_number = bios_parser_get_connectors_number,
+
+	.get_connector_id = bios_parser_get_connector_id,
+
+	.get_src_obj = bios_parser_get_src_obj,
+
+	.get_i2c_info = bios_parser_get_i2c_info,
+
+	.get_hpd_info = bios_parser_get_hpd_info,
+
+	.get_device_tag = bios_parser_get_device_tag,
+
+	.get_spread_spectrum_info = bios_parser_get_spread_spectrum_info,
+
+	.get_ss_entry_number = bios_parser_get_ss_entry_number,
+
+	.get_embedded_panel_info = bios_parser_get_embedded_panel_info,
+
+	.get_gpio_pin_info = bios_parser_get_gpio_pin_info,
+
+	.get_encoder_cap_info = bios_parser_get_encoder_cap_info,
+
+	.is_device_id_supported = bios_parser_is_device_id_supported,
+
+	.is_accelerated_mode = bios_parser_is_accelerated_mode,
+
+	.set_scratch_critical_state = bios_parser_set_scratch_critical_state,
+
+
+/*	 COMMANDS */
+	.encoder_control = bios_parser_encoder_control,
+
+	.transmitter_control = bios_parser_transmitter_control,
+
+	.enable_crtc = bios_parser_enable_crtc,
+
+	.set_pixel_clock = bios_parser_set_pixel_clock,
+
+	.set_dce_clock = bios_parser_set_dce_clock,
+
+	.program_crtc_timing = bios_parser_program_crtc_timing,
+
+	.enable_disp_power_gating = bios_parser_enable_disp_power_gating,
+
+	.bios_parser_destroy = firmware_parser_destroy,
+
+	.get_board_layout_info = bios_get_board_layout_info,
+	.pack_data_tables = bios_parser_pack_data_tables,
+
+	.get_atom_dc_golden_table = bios_get_atom_dc_golden_table,
+
+	.enable_lvtma_control = bios_parser_enable_lvtma_control,
+
+	.get_soc_bb_info = bios_parser_get_soc_bb_info,
+
+	.get_disp_connector_caps_info = bios_parser_get_disp_connector_caps_info,
+
+	.get_lttpr_caps = bios_parser_get_lttpr_caps,
+
+	.get_lttpr_interop = bios_parser_get_lttpr_interop,
+
+	.get_connector_speed_cap_info = bios_parser_get_connector_speed_cap_info,
+};
+
+static bool bios_parser2_construct(
+	struct bios_parser *bp,
+	struct bp_init_data *init,
+	enum dce_version dce_version)
+{
+	uint16_t *rom_header_offset = NULL;
+	struct atom_rom_header_v2_2 *rom_header = NULL;
+	struct display_object_info_table_v1_4 *object_info_tbl;
+	struct atom_data_revision tbl_rev = {0};
+
+	if (!init)
+		return false;
+
+	if (!init->bios)
+		return false;
+
+	bp->base.funcs = &vbios_funcs;
+	bp->base.bios = init->bios;
+	bp->base.bios_size = bp->base.bios[OFFSET_TO_ATOM_ROM_IMAGE_SIZE] * BIOS_IMAGE_SIZE_UNIT;
+
+	bp->base.ctx = init->ctx;
+
+	bp->base.bios_local_image = NULL;
+
+	rom_header_offset =
+			GET_IMAGE(uint16_t, OFFSET_TO_ATOM_ROM_HEADER_POINTER);
+
+	if (!rom_header_offset)
+		return false;
+
+	rom_header = GET_IMAGE(struct atom_rom_header_v2_2, *rom_header_offset);
+
+	if (!rom_header)
+		return false;
+
+	get_atom_data_table_revision(&rom_header->table_header, &tbl_rev);
+	if (!(tbl_rev.major >= 2 && tbl_rev.minor >= 2))
+		return false;
+
+	bp->master_data_tbl =
+		GET_IMAGE(struct atom_master_data_table_v2_1,
+				rom_header->masterdatatable_offset);
+
+	if (!bp->master_data_tbl)
+		return false;
+
+	bp->object_info_tbl_offset = DATA_TABLES(displayobjectinfo);
+
+	if (!bp->object_info_tbl_offset)
+		return false;
+
+	object_info_tbl =
+			GET_IMAGE(struct display_object_info_table_v1_4,
+						bp->object_info_tbl_offset);
+
+	if (!object_info_tbl)
+		return false;
+
+	get_atom_data_table_revision(&object_info_tbl->table_header,
+		&bp->object_info_tbl.revision);
+
+	if (bp->object_info_tbl.revision.major == 1
+		&& bp->object_info_tbl.revision.minor == 4) {
+		struct display_object_info_table_v1_4 *tbl_v1_4;
+
+		tbl_v1_4 = GET_IMAGE(struct display_object_info_table_v1_4,
+			bp->object_info_tbl_offset);
+		if (!tbl_v1_4)
+			return false;
+
+		bp->object_info_tbl.v1_4 = tbl_v1_4;
+	} else if (bp->object_info_tbl.revision.major == 1
+		&& bp->object_info_tbl.revision.minor == 5) {
+		struct display_object_info_table_v1_5 *tbl_v1_5;
+
+		tbl_v1_5 = GET_IMAGE(struct display_object_info_table_v1_5,
+			bp->object_info_tbl_offset);
+		if (!tbl_v1_5)
+			return false;
+
+		bp->object_info_tbl.v1_5 = tbl_v1_5;
+	} else {
+		ASSERT(0);
+		return false;
+	}
+
+	dal_firmware_parser_init_cmd_tbl(bp);
+	dal_bios_parser_init_cmd_tbl_helper2(&bp->cmd_helper, dce_version);
+
+	bp->base.integrated_info = bios_parser_create_integrated_info(&bp->base);
+	bp->base.fw_info_valid = bios_parser_get_firmware_info(&bp->base, &bp->base.fw_info) == BP_RESULT_OK;
+	bios_parser_get_vram_info(&bp->base, &bp->base.vram_info);
+	bios_parser_get_soc_bb_info(&bp->base, &bp->base.bb_info);
+	return true;
+}
+
+struct dc_bios *firmware_parser_create(
+	struct bp_init_data *init,
+	enum dce_version dce_version)
+{
+	struct bios_parser *bp;
+
+	bp = kzalloc(sizeof(struct bios_parser), GFP_KERNEL);
+	if (!bp)
+		return NULL;
+
+	if (bios_parser2_construct(bp, init, dce_version))
+		return &bp->base;
+
+	kfree(bp);
+	return NULL;
+}
+
+
diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.h b/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.h
new file mode 100644
index 000000000000..cb40546cdafe
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_BIOS_PARSER2_H__
+#define __DAL_BIOS_PARSER2_H__
+
+struct dc_bios *firmware_parser_create(
+	struct bp_init_data *init,
+	enum dce_version dce_version);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser_common.c b/drivers/gpu/drm/amd/display/dc/bios/bios_parser_common.c
new file mode 100644
index 000000000000..34e3a64f556e
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser_common.c
@@ -0,0 +1,291 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "bios_parser_common.h"
+#include "include/grph_object_ctrl_defs.h"
+
+static enum object_type object_type_from_bios_object_id(uint32_t bios_object_id)
+{
+	uint32_t bios_object_type = (bios_object_id & OBJECT_TYPE_MASK)
+				>> OBJECT_TYPE_SHIFT;
+	enum object_type object_type;
+
+	switch (bios_object_type) {
+	case GRAPH_OBJECT_TYPE_GPU:
+		object_type = OBJECT_TYPE_GPU;
+		break;
+	case GRAPH_OBJECT_TYPE_ENCODER:
+		object_type = OBJECT_TYPE_ENCODER;
+		break;
+	case GRAPH_OBJECT_TYPE_CONNECTOR:
+		object_type = OBJECT_TYPE_CONNECTOR;
+		break;
+	case GRAPH_OBJECT_TYPE_ROUTER:
+		object_type = OBJECT_TYPE_ROUTER;
+		break;
+	case GRAPH_OBJECT_TYPE_GENERIC:
+		object_type = OBJECT_TYPE_GENERIC;
+		break;
+	default:
+		object_type = OBJECT_TYPE_UNKNOWN;
+		break;
+	}
+
+	return object_type;
+}
+
+static enum object_enum_id enum_id_from_bios_object_id(uint32_t bios_object_id)
+{
+	uint32_t bios_enum_id =
+			(bios_object_id & ENUM_ID_MASK) >> ENUM_ID_SHIFT;
+	enum object_enum_id id;
+
+	switch (bios_enum_id) {
+	case GRAPH_OBJECT_ENUM_ID1:
+		id = ENUM_ID_1;
+		break;
+	case GRAPH_OBJECT_ENUM_ID2:
+		id = ENUM_ID_2;
+		break;
+	case GRAPH_OBJECT_ENUM_ID3:
+		id = ENUM_ID_3;
+		break;
+	case GRAPH_OBJECT_ENUM_ID4:
+		id = ENUM_ID_4;
+		break;
+	case GRAPH_OBJECT_ENUM_ID5:
+		id = ENUM_ID_5;
+		break;
+	case GRAPH_OBJECT_ENUM_ID6:
+		id = ENUM_ID_6;
+		break;
+	case GRAPH_OBJECT_ENUM_ID7:
+		id = ENUM_ID_7;
+		break;
+	default:
+		id = ENUM_ID_UNKNOWN;
+		break;
+	}
+
+	return id;
+}
+
+static uint32_t gpu_id_from_bios_object_id(uint32_t bios_object_id)
+{
+	return (bios_object_id & OBJECT_ID_MASK) >> OBJECT_ID_SHIFT;
+}
+
+static enum encoder_id encoder_id_from_bios_object_id(uint32_t bios_object_id)
+{
+	uint32_t bios_encoder_id = gpu_id_from_bios_object_id(bios_object_id);
+	enum encoder_id id;
+
+	switch (bios_encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_LVDS:
+		id = ENCODER_ID_INTERNAL_LVDS;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_TMDS1:
+		id = ENCODER_ID_INTERNAL_TMDS1;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_TMDS2:
+		id = ENCODER_ID_INTERNAL_TMDS2;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_DAC1:
+		id = ENCODER_ID_INTERNAL_DAC1;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_DAC2:
+		id = ENCODER_ID_INTERNAL_DAC2;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_LVTM1:
+		id = ENCODER_ID_INTERNAL_LVTM1;
+		break;
+	case ENCODER_OBJECT_ID_HDMI_INTERNAL:
+		id = ENCODER_ID_INTERNAL_HDMI;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1:
+		id = ENCODER_ID_INTERNAL_KLDSCP_TMDS1;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
+		id = ENCODER_ID_INTERNAL_KLDSCP_DAC1;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2:
+		id = ENCODER_ID_INTERNAL_KLDSCP_DAC2;
+		break;
+	case ENCODER_OBJECT_ID_MVPU_FPGA:
+		id = ENCODER_ID_EXTERNAL_MVPU_FPGA;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_DDI:
+		id = ENCODER_ID_INTERNAL_DDI;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+		id = ENCODER_ID_INTERNAL_UNIPHY;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA:
+		id = ENCODER_ID_INTERNAL_KLDSCP_LVTMA;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+		id = ENCODER_ID_INTERNAL_UNIPHY1;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+		id = ENCODER_ID_INTERNAL_UNIPHY2;
+		break;
+	case ENCODER_OBJECT_ID_ALMOND: /* ENCODER_OBJECT_ID_NUTMEG */
+		id = ENCODER_ID_EXTERNAL_NUTMEG;
+		break;
+	case ENCODER_OBJECT_ID_TRAVIS:
+		id = ENCODER_ID_EXTERNAL_TRAVIS;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+		id = ENCODER_ID_INTERNAL_UNIPHY3;
+		break;
+	default:
+		id = ENCODER_ID_UNKNOWN;
+		ASSERT(0);
+		break;
+	}
+
+	return id;
+}
+
+static enum connector_id connector_id_from_bios_object_id(
+	uint32_t bios_object_id)
+{
+	uint32_t bios_connector_id = gpu_id_from_bios_object_id(bios_object_id);
+
+	enum connector_id id;
+
+	switch (bios_connector_id) {
+	case CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I:
+		id = CONNECTOR_ID_SINGLE_LINK_DVII;
+		break;
+	case CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I:
+		id = CONNECTOR_ID_DUAL_LINK_DVII;
+		break;
+	case CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D:
+		id = CONNECTOR_ID_SINGLE_LINK_DVID;
+		break;
+	case CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D:
+		id = CONNECTOR_ID_DUAL_LINK_DVID;
+		break;
+	case CONNECTOR_OBJECT_ID_VGA:
+		id = CONNECTOR_ID_VGA;
+		break;
+	case CONNECTOR_OBJECT_ID_HDMI_TYPE_A:
+		id = CONNECTOR_ID_HDMI_TYPE_A;
+		break;
+	case CONNECTOR_OBJECT_ID_LVDS:
+		id = CONNECTOR_ID_LVDS;
+		break;
+	case CONNECTOR_OBJECT_ID_PCIE_CONNECTOR:
+		id = CONNECTOR_ID_PCIE;
+		break;
+	case CONNECTOR_OBJECT_ID_HARDCODE_DVI:
+		id = CONNECTOR_ID_HARDCODE_DVI;
+		break;
+	case CONNECTOR_OBJECT_ID_DISPLAYPORT:
+		id = CONNECTOR_ID_DISPLAY_PORT;
+		break;
+	case CONNECTOR_OBJECT_ID_eDP:
+		id = CONNECTOR_ID_EDP;
+		break;
+	case CONNECTOR_OBJECT_ID_MXM:
+		id = CONNECTOR_ID_MXM;
+		break;
+	case CONNECTOR_OBJECT_ID_USBC:
+		id = CONNECTOR_ID_USBC;
+		break;
+	default:
+		id = CONNECTOR_ID_UNKNOWN;
+		break;
+	}
+
+	return id;
+}
+
+static enum generic_id generic_id_from_bios_object_id(uint32_t bios_object_id)
+{
+	uint32_t bios_generic_id = gpu_id_from_bios_object_id(bios_object_id);
+
+	enum generic_id id;
+
+	switch (bios_generic_id) {
+	case GENERIC_OBJECT_ID_MXM_OPM:
+		id = GENERIC_ID_MXM_OPM;
+		break;
+	case GENERIC_OBJECT_ID_GLSYNC:
+		id = GENERIC_ID_GLSYNC;
+		break;
+	case GENERIC_OBJECT_ID_STEREO_PIN:
+		id = GENERIC_ID_STEREO;
+		break;
+	default:
+		id = GENERIC_ID_UNKNOWN;
+		break;
+	}
+
+	return id;
+}
+
+static uint32_t id_from_bios_object_id(enum object_type type,
+	uint32_t bios_object_id)
+{
+	switch (type) {
+	case OBJECT_TYPE_GPU:
+		return gpu_id_from_bios_object_id(bios_object_id);
+	case OBJECT_TYPE_ENCODER:
+		return (uint32_t)encoder_id_from_bios_object_id(bios_object_id);
+	case OBJECT_TYPE_CONNECTOR:
+		return (uint32_t)connector_id_from_bios_object_id(
+				bios_object_id);
+	case OBJECT_TYPE_GENERIC:
+		return generic_id_from_bios_object_id(bios_object_id);
+	default:
+		return 0;
+	}
+}
+
+struct graphics_object_id object_id_from_bios_object_id(uint32_t bios_object_id)
+{
+	enum object_type type;
+	enum object_enum_id enum_id;
+	struct graphics_object_id go_id = { 0 };
+
+	type = object_type_from_bios_object_id(bios_object_id);
+
+	if (OBJECT_TYPE_UNKNOWN == type)
+		return go_id;
+
+	enum_id = enum_id_from_bios_object_id(bios_object_id);
+
+	if (ENUM_ID_UNKNOWN == enum_id)
+		return go_id;
+
+	go_id = dal_graphics_object_id_init(
+			id_from_bios_object_id(type, bios_object_id), enum_id, type);
+
+	return go_id;
+}
+
+
diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser_common.h b/drivers/gpu/drm/amd/display/dc/bios/bios_parser_common.h
new file mode 100644
index 000000000000..a076c61dfae4
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser_common.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __BIOS_PARSER_COMMON_H__
+#define __BIOS_PARSER_COMMON_H__
+
+#include "dm_services.h"
+#include "ObjectID.h"
+
+struct graphics_object_id object_id_from_bios_object_id(uint32_t bios_object_id);
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser_helper.c b/drivers/gpu/drm/amd/display/dc/bios/bios_parser_helper.c
new file mode 100644
index 000000000000..8d2cf95ae739
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser_helper.c
@@ -0,0 +1,80 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+
+#include "atom.h"
+
+#include "include/bios_parser_types.h"
+#include "bios_parser_helper.h"
+#include "command_table_helper.h"
+#include "command_table.h"
+#include "bios_parser_types_internal.h"
+
+uint8_t *bios_get_image(struct dc_bios *bp,
+	uint32_t offset,
+	uint32_t size)
+{
+	if (bp->bios && offset + size < bp->bios_size)
+		return bp->bios + offset;
+	else
+		return NULL;
+}
+
+#include "reg_helper.h"
+
+#define CTX \
+	bios->ctx
+#define REG(reg)\
+	(bios->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+		ATOM_ ## field_name ## _SHIFT, ATOM_ ## field_name
+
+bool bios_is_accelerated_mode(
+	struct dc_bios *bios)
+{
+	uint32_t acc_mode;
+	REG_GET(BIOS_SCRATCH_6, S6_ACC_MODE, &acc_mode);
+	return (acc_mode == 1);
+}
+
+
+void bios_set_scratch_acc_mode_change(
+	struct dc_bios *bios,
+	uint32_t state)
+{
+	REG_UPDATE(BIOS_SCRATCH_6, S6_ACC_MODE, state);
+}
+
+
+void bios_set_scratch_critical_state(
+	struct dc_bios *bios,
+	bool state)
+{
+	uint32_t critial_state = state ? 1 : 0;
+	REG_UPDATE(BIOS_SCRATCH_6, S6_CRITICAL_STATE, critial_state);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser_helper.h b/drivers/gpu/drm/amd/display/dc/bios/bios_parser_helper.h
new file mode 100644
index 000000000000..ab162f2fe577
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser_helper.h
@@ -0,0 +1,40 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_BIOS_PARSER_HELPER_H__
+#define __DAL_BIOS_PARSER_HELPER_H__
+
+struct bios_parser;
+
+uint8_t *bios_get_image(struct dc_bios *bp, uint32_t offset,
+	uint32_t size);
+
+bool bios_is_accelerated_mode(struct dc_bios *bios);
+void bios_set_scratch_acc_mode_change(struct dc_bios *bios, uint32_t state);
+void bios_set_scratch_critical_state(struct dc_bios *bios, bool state);
+
+#define GET_IMAGE(type, offset) ((type *) bios_get_image(&bp->base, offset, sizeof(type)))
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser_interface.c b/drivers/gpu/drm/amd/display/dc/bios/bios_parser_interface.c
new file mode 100644
index 000000000000..0079a1e26efd
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser_interface.c
@@ -0,0 +1,56 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+#include "include/logger_interface.h"
+
+#include "bios_parser_interface.h"
+#include "bios_parser.h"
+
+#include "bios_parser2.h"
+
+
+struct dc_bios *dal_bios_parser_create(
+	struct bp_init_data *init,
+	enum dce_version dce_version)
+{
+	struct dc_bios *bios = NULL;
+
+	bios = firmware_parser_create(init, dce_version);
+
+	/* Fall back to old bios parser for older asics */
+	if (bios == NULL)
+		bios = bios_parser_create(init, dce_version);
+
+	return bios;
+}
+
+void dal_bios_parser_destroy(struct dc_bios **dcb)
+{
+	struct dc_bios *bios = *dcb;
+
+	bios->funcs->bios_parser_destroy(dcb);
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser_types_internal.h b/drivers/gpu/drm/amd/display/dc/bios/bios_parser_types_internal.h
new file mode 100644
index 000000000000..5918923bfb93
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser_types_internal.h
@@ -0,0 +1,72 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_BIOS_PARSER_TYPES_BIOS_H__
+#define __DAL_BIOS_PARSER_TYPES_BIOS_H__
+
+#include "dc_bios_types.h"
+#include "bios_parser_helper.h"
+
+struct atom_data_revision {
+	uint32_t major;
+	uint32_t minor;
+};
+
+struct object_info_table {
+	struct atom_data_revision revision;
+	union {
+		ATOM_OBJECT_HEADER *v1_1;
+		ATOM_OBJECT_HEADER_V3 *v1_3;
+	};
+};
+
+enum spread_spectrum_id {
+	SS_ID_UNKNOWN = 0,
+	SS_ID_DP1 = 0xf1,
+	SS_ID_DP2 = 0xf2,
+	SS_ID_LVLINK_2700MHZ = 0xf3,
+	SS_ID_LVLINK_1620MHZ = 0xf4
+};
+
+struct bios_parser {
+	struct dc_bios base;
+
+	struct object_info_table object_info_tbl;
+	uint32_t object_info_tbl_offset;
+	ATOM_MASTER_DATA_TABLE *master_data_tbl;
+
+	const struct bios_parser_helper *bios_helper;
+
+	const struct command_table_helper *cmd_helper;
+	struct cmd_tbl cmd_tbl;
+
+	bool remap_device_tags;
+};
+
+/* Bios Parser from DC Bios */
+#define BP_FROM_DCB(dc_bios) \
+	container_of(dc_bios, struct bios_parser, base)
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser_types_internal2.h b/drivers/gpu/drm/amd/display/dc/bios/bios_parser_types_internal2.h
new file mode 100644
index 000000000000..41d02d473082
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser_types_internal2.h
@@ -0,0 +1,75 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_BIOS_PARSER_TYPES_BIOS2_H__
+#define __DAL_BIOS_PARSER_TYPES_BIOS2_H__
+
+#include "dc_bios_types.h"
+#include "bios_parser_helper.h"
+
+/* use atomfirmware_bringup.h only. Not atombios.h anymore */
+
+struct atom_data_revision {
+	uint32_t major;
+	uint32_t minor;
+};
+
+struct object_info_table {
+	struct atom_data_revision revision;
+	union {
+		struct display_object_info_table_v1_4 *v1_4;
+		struct display_object_info_table_v1_5 *v1_5;
+	};
+};
+
+enum spread_spectrum_id {
+	SS_ID_UNKNOWN = 0,
+	SS_ID_DP1 = 0xf1,
+	SS_ID_DP2 = 0xf2,
+	SS_ID_LVLINK_2700MHZ = 0xf3,
+	SS_ID_LVLINK_1620MHZ = 0xf4
+};
+
+struct bios_parser {
+	struct dc_bios base;
+
+	struct object_info_table object_info_tbl;
+	uint32_t object_info_tbl_offset;
+	struct atom_master_data_table_v2_1 *master_data_tbl;
+
+
+	const struct bios_parser_helper *bios_helper;
+
+	const struct command_table_helper *cmd_helper;
+	struct cmd_tbl cmd_tbl;
+
+	bool remap_device_tags;
+};
+
+/* Bios Parser from DC Bios */
+#define BP_FROM_DCB(dc_bios) \
+	container_of(dc_bios, struct bios_parser, base)
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/bios/command_table.c b/drivers/gpu/drm/amd/display/dc/bios/command_table.c
new file mode 100644
index 000000000000..58e88778da7f
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/command_table.c
@@ -0,0 +1,2432 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+#include "amdgpu.h"
+#include "atom.h"
+
+#include "include/bios_parser_interface.h"
+
+#include "command_table.h"
+#include "command_table_helper.h"
+#include "bios_parser_helper.h"
+#include "bios_parser_types_internal.h"
+
+#define EXEC_BIOS_CMD_TABLE(command, params)\
+	(amdgpu_atom_execute_table(((struct amdgpu_device *)bp->base.ctx->driver_context)->mode_info.atom_context, \
+		GetIndexIntoMasterTable(COMMAND, command), \
+		(uint32_t *)&params, sizeof(params)) == 0)
+
+#define BIOS_CMD_TABLE_REVISION(command, frev, crev)\
+	amdgpu_atom_parse_cmd_header(((struct amdgpu_device *)bp->base.ctx->driver_context)->mode_info.atom_context, \
+		GetIndexIntoMasterTable(COMMAND, command), &frev, &crev)
+
+#define BIOS_CMD_TABLE_PARA_REVISION(command)\
+	bios_cmd_table_para_revision(bp->base.ctx->driver_context, \
+		GetIndexIntoMasterTable(COMMAND, command))
+
+static void init_dig_encoder_control(struct bios_parser *bp);
+static void init_transmitter_control(struct bios_parser *bp);
+static void init_set_pixel_clock(struct bios_parser *bp);
+static void init_enable_spread_spectrum_on_ppll(struct bios_parser *bp);
+static void init_adjust_display_pll(struct bios_parser *bp);
+static void init_dac_encoder_control(struct bios_parser *bp);
+static void init_dac_output_control(struct bios_parser *bp);
+static void init_set_crtc_timing(struct bios_parser *bp);
+static void init_enable_crtc(struct bios_parser *bp);
+static void init_enable_crtc_mem_req(struct bios_parser *bp);
+static void init_external_encoder_control(struct bios_parser *bp);
+static void init_enable_disp_power_gating(struct bios_parser *bp);
+static void init_program_clock(struct bios_parser *bp);
+static void init_set_dce_clock(struct bios_parser *bp);
+
+void dal_bios_parser_init_cmd_tbl(struct bios_parser *bp)
+{
+	init_dig_encoder_control(bp);
+	init_transmitter_control(bp);
+	init_set_pixel_clock(bp);
+	init_enable_spread_spectrum_on_ppll(bp);
+	init_adjust_display_pll(bp);
+	init_dac_encoder_control(bp);
+	init_dac_output_control(bp);
+	init_set_crtc_timing(bp);
+	init_enable_crtc(bp);
+	init_enable_crtc_mem_req(bp);
+	init_program_clock(bp);
+	init_external_encoder_control(bp);
+	init_enable_disp_power_gating(bp);
+	init_set_dce_clock(bp);
+}
+
+static uint32_t bios_cmd_table_para_revision(void *dev,
+					     uint32_t index)
+{
+	struct amdgpu_device *adev = dev;
+	uint8_t frev, crev;
+
+	if (amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context,
+					index,
+					&frev, &crev))
+		return crev;
+	else
+		return 0;
+}
+
+/*******************************************************************************
+ ********************************************************************************
+ **
+ **                  D I G E N C O D E R C O N T R O L
+ **
+ ********************************************************************************
+ *******************************************************************************/
+static enum bp_result encoder_control_digx_v3(
+	struct bios_parser *bp,
+	struct bp_encoder_control *cntl);
+
+static enum bp_result encoder_control_digx_v4(
+	struct bios_parser *bp,
+	struct bp_encoder_control *cntl);
+
+static enum bp_result encoder_control_digx_v5(
+	struct bios_parser *bp,
+	struct bp_encoder_control *cntl);
+
+static void init_encoder_control_dig_v1(struct bios_parser *bp);
+
+static void init_dig_encoder_control(struct bios_parser *bp)
+{
+	uint32_t version =
+		BIOS_CMD_TABLE_PARA_REVISION(DIGxEncoderControl);
+
+	switch (version) {
+	case 2:
+		bp->cmd_tbl.dig_encoder_control = encoder_control_digx_v3;
+		break;
+	case 4:
+		bp->cmd_tbl.dig_encoder_control = encoder_control_digx_v4;
+		break;
+
+	case 5:
+		bp->cmd_tbl.dig_encoder_control = encoder_control_digx_v5;
+		break;
+
+	default:
+		init_encoder_control_dig_v1(bp);
+		break;
+	}
+}
+
+static enum bp_result encoder_control_dig_v1(
+	struct bios_parser *bp,
+	struct bp_encoder_control *cntl);
+static enum bp_result encoder_control_dig1_v1(
+	struct bios_parser *bp,
+	struct bp_encoder_control *cntl);
+static enum bp_result encoder_control_dig2_v1(
+	struct bios_parser *bp,
+	struct bp_encoder_control *cntl);
+
+static void init_encoder_control_dig_v1(struct bios_parser *bp)
+{
+	struct cmd_tbl *cmd_tbl = &bp->cmd_tbl;
+
+	if (1 == BIOS_CMD_TABLE_PARA_REVISION(DIG1EncoderControl))
+		cmd_tbl->encoder_control_dig1 = encoder_control_dig1_v1;
+	else
+		cmd_tbl->encoder_control_dig1 = NULL;
+
+	if (1 == BIOS_CMD_TABLE_PARA_REVISION(DIG2EncoderControl))
+		cmd_tbl->encoder_control_dig2 = encoder_control_dig2_v1;
+	else
+		cmd_tbl->encoder_control_dig2 = NULL;
+
+	cmd_tbl->dig_encoder_control = encoder_control_dig_v1;
+}
+
+static enum bp_result encoder_control_dig_v1(
+	struct bios_parser *bp,
+	struct bp_encoder_control *cntl)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	struct cmd_tbl *cmd_tbl = &bp->cmd_tbl;
+
+	if (cntl != NULL)
+		switch (cntl->engine_id) {
+		case ENGINE_ID_DIGA:
+			if (cmd_tbl->encoder_control_dig1 != NULL)
+				result =
+					cmd_tbl->encoder_control_dig1(bp, cntl);
+			break;
+		case ENGINE_ID_DIGB:
+			if (cmd_tbl->encoder_control_dig2 != NULL)
+				result =
+					cmd_tbl->encoder_control_dig2(bp, cntl);
+			break;
+
+		default:
+			break;
+		}
+
+	return result;
+}
+
+static enum bp_result encoder_control_dig1_v1(
+	struct bios_parser *bp,
+	struct bp_encoder_control *cntl)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	DIG_ENCODER_CONTROL_PARAMETERS_V2 params = {0};
+
+	bp->cmd_helper->assign_control_parameter(bp->cmd_helper, cntl, &params);
+
+	if (EXEC_BIOS_CMD_TABLE(DIG1EncoderControl, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+static enum bp_result encoder_control_dig2_v1(
+	struct bios_parser *bp,
+	struct bp_encoder_control *cntl)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	DIG_ENCODER_CONTROL_PARAMETERS_V2 params = {0};
+
+	bp->cmd_helper->assign_control_parameter(bp->cmd_helper, cntl, &params);
+
+	if (EXEC_BIOS_CMD_TABLE(DIG2EncoderControl, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+static enum bp_result encoder_control_digx_v3(
+	struct bios_parser *bp,
+	struct bp_encoder_control *cntl)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	DIG_ENCODER_CONTROL_PARAMETERS_V3 params = {0};
+
+	if (LANE_COUNT_FOUR < cntl->lanes_number)
+		params.acConfig.ucDPLinkRate = 1; /* dual link 2.7GHz */
+	else
+		params.acConfig.ucDPLinkRate = 0; /* single link 1.62GHz */
+
+	params.acConfig.ucDigSel = (uint8_t)(cntl->engine_id);
+
+	/* We need to convert from KHz units into 10KHz units */
+	params.ucAction = bp->cmd_helper->encoder_action_to_atom(cntl->action);
+	params.usPixelClock = cpu_to_le16((uint16_t)(cntl->pixel_clock / 10));
+	params.ucEncoderMode =
+			(uint8_t)bp->cmd_helper->encoder_mode_bp_to_atom(
+					cntl->signal,
+					cntl->enable_dp_audio);
+	params.ucLaneNum = (uint8_t)(cntl->lanes_number);
+
+	switch (cntl->color_depth) {
+	case COLOR_DEPTH_888:
+		params.ucBitPerColor = PANEL_8BIT_PER_COLOR;
+		break;
+	case COLOR_DEPTH_101010:
+		params.ucBitPerColor = PANEL_10BIT_PER_COLOR;
+		break;
+	case COLOR_DEPTH_121212:
+		params.ucBitPerColor = PANEL_12BIT_PER_COLOR;
+		break;
+	case COLOR_DEPTH_161616:
+		params.ucBitPerColor = PANEL_16BIT_PER_COLOR;
+		break;
+	default:
+		break;
+	}
+
+	if (EXEC_BIOS_CMD_TABLE(DIGxEncoderControl, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+static enum bp_result encoder_control_digx_v4(
+	struct bios_parser *bp,
+	struct bp_encoder_control *cntl)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	DIG_ENCODER_CONTROL_PARAMETERS_V4 params = {0};
+
+	if (LANE_COUNT_FOUR < cntl->lanes_number)
+		params.acConfig.ucDPLinkRate = 1; /* dual link 2.7GHz */
+	else
+		params.acConfig.ucDPLinkRate = 0; /* single link 1.62GHz */
+
+	params.acConfig.ucDigSel = (uint8_t)(cntl->engine_id);
+
+	/* We need to convert from KHz units into 10KHz units */
+	params.ucAction = bp->cmd_helper->encoder_action_to_atom(cntl->action);
+	params.usPixelClock = cpu_to_le16((uint16_t)(cntl->pixel_clock / 10));
+	params.ucEncoderMode =
+			(uint8_t)(bp->cmd_helper->encoder_mode_bp_to_atom(
+					cntl->signal,
+					cntl->enable_dp_audio));
+	params.ucLaneNum = (uint8_t)(cntl->lanes_number);
+
+	switch (cntl->color_depth) {
+	case COLOR_DEPTH_888:
+		params.ucBitPerColor = PANEL_8BIT_PER_COLOR;
+		break;
+	case COLOR_DEPTH_101010:
+		params.ucBitPerColor = PANEL_10BIT_PER_COLOR;
+		break;
+	case COLOR_DEPTH_121212:
+		params.ucBitPerColor = PANEL_12BIT_PER_COLOR;
+		break;
+	case COLOR_DEPTH_161616:
+		params.ucBitPerColor = PANEL_16BIT_PER_COLOR;
+		break;
+	default:
+		break;
+	}
+
+	if (EXEC_BIOS_CMD_TABLE(DIGxEncoderControl, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+static enum bp_result encoder_control_digx_v5(
+	struct bios_parser *bp,
+	struct bp_encoder_control *cntl)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	ENCODER_STREAM_SETUP_PARAMETERS_V5 params = {0};
+
+	params.ucDigId = (uint8_t)(cntl->engine_id);
+	params.ucAction = bp->cmd_helper->encoder_action_to_atom(cntl->action);
+
+	params.ulPixelClock = cntl->pixel_clock / 10;
+	params.ucDigMode =
+			(uint8_t)(bp->cmd_helper->encoder_mode_bp_to_atom(
+					cntl->signal,
+					cntl->enable_dp_audio));
+	params.ucLaneNum = (uint8_t)(cntl->lanes_number);
+
+	switch (cntl->color_depth) {
+	case COLOR_DEPTH_888:
+		params.ucBitPerColor = PANEL_8BIT_PER_COLOR;
+		break;
+	case COLOR_DEPTH_101010:
+		params.ucBitPerColor = PANEL_10BIT_PER_COLOR;
+		break;
+	case COLOR_DEPTH_121212:
+		params.ucBitPerColor = PANEL_12BIT_PER_COLOR;
+		break;
+	case COLOR_DEPTH_161616:
+		params.ucBitPerColor = PANEL_16BIT_PER_COLOR;
+		break;
+	default:
+		break;
+	}
+
+	if (cntl->signal == SIGNAL_TYPE_HDMI_TYPE_A)
+		switch (cntl->color_depth) {
+		case COLOR_DEPTH_101010:
+			params.ulPixelClock =
+				(params.ulPixelClock * 30) / 24;
+			break;
+		case COLOR_DEPTH_121212:
+			params.ulPixelClock =
+				(params.ulPixelClock * 36) / 24;
+			break;
+		case COLOR_DEPTH_161616:
+			params.ulPixelClock =
+				(params.ulPixelClock * 48) / 24;
+			break;
+		default:
+			break;
+		}
+
+	if (EXEC_BIOS_CMD_TABLE(DIGxEncoderControl, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+/*******************************************************************************
+ ********************************************************************************
+ **
+ **                  TRANSMITTER CONTROL
+ **
+ ********************************************************************************
+ *******************************************************************************/
+
+static enum bp_result transmitter_control_v2(
+	struct bios_parser *bp,
+	struct bp_transmitter_control *cntl);
+static enum bp_result transmitter_control_v3(
+	struct bios_parser *bp,
+	struct bp_transmitter_control *cntl);
+static enum bp_result transmitter_control_v4(
+	struct bios_parser *bp,
+	struct bp_transmitter_control *cntl);
+static enum bp_result transmitter_control_v1_5(
+	struct bios_parser *bp,
+	struct bp_transmitter_control *cntl);
+static enum bp_result transmitter_control_v1_6(
+	struct bios_parser *bp,
+	struct bp_transmitter_control *cntl);
+
+static void init_transmitter_control(struct bios_parser *bp)
+{
+	uint8_t frev;
+	uint8_t crev = 0;
+
+	if (BIOS_CMD_TABLE_REVISION(UNIPHYTransmitterControl,
+			frev, crev) == false)
+		BREAK_TO_DEBUGGER();
+	switch (crev) {
+	case 2:
+		bp->cmd_tbl.transmitter_control = transmitter_control_v2;
+		break;
+	case 3:
+		bp->cmd_tbl.transmitter_control = transmitter_control_v3;
+		break;
+	case 4:
+		bp->cmd_tbl.transmitter_control = transmitter_control_v4;
+		break;
+	case 5:
+		bp->cmd_tbl.transmitter_control = transmitter_control_v1_5;
+		break;
+	case 6:
+		bp->cmd_tbl.transmitter_control = transmitter_control_v1_6;
+		break;
+	default:
+		dm_output_to_console("Don't have transmitter_control for v%d\n", crev);
+		bp->cmd_tbl.transmitter_control = NULL;
+		break;
+	}
+}
+
+static enum bp_result transmitter_control_v2(
+	struct bios_parser *bp,
+	struct bp_transmitter_control *cntl)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	DIG_TRANSMITTER_CONTROL_PARAMETERS_V2 params;
+	enum connector_id connector_id =
+		dal_graphics_object_id_get_connector_id(cntl->connector_obj_id);
+
+	memset(&params, 0, sizeof(params));
+
+	switch (cntl->transmitter) {
+	case TRANSMITTER_UNIPHY_A:
+	case TRANSMITTER_UNIPHY_B:
+	case TRANSMITTER_UNIPHY_C:
+	case TRANSMITTER_UNIPHY_D:
+	case TRANSMITTER_UNIPHY_E:
+	case TRANSMITTER_UNIPHY_F:
+	case TRANSMITTER_TRAVIS_LCD:
+		break;
+	default:
+		return BP_RESULT_BADINPUT;
+	}
+
+	switch (cntl->action) {
+	case TRANSMITTER_CONTROL_INIT:
+		if ((CONNECTOR_ID_DUAL_LINK_DVII == connector_id) ||
+				(CONNECTOR_ID_DUAL_LINK_DVID == connector_id))
+			/* on INIT this bit should be set according to the
+			 * physical connector
+			 * Bit0: dual link connector flag
+			 * =0 connector is single link connector
+			 * =1 connector is dual link connector
+			 */
+			params.acConfig.fDualLinkConnector = 1;
+
+		/* connector object id */
+		params.usInitInfo =
+				cpu_to_le16((uint8_t)cntl->connector_obj_id.id);
+		break;
+	case TRANSMITTER_CONTROL_SET_VOLTAGE_AND_PREEMPASIS:
+		/* voltage swing and pre-emphsis */
+		params.asMode.ucLaneSel = (uint8_t)cntl->lane_select;
+		params.asMode.ucLaneSet = (uint8_t)cntl->lane_settings;
+		break;
+	default:
+		/* if dual-link */
+		if (LANE_COUNT_FOUR < cntl->lanes_number) {
+			/* on ENABLE/DISABLE this bit should be set according to
+			 * actual timing (number of lanes)
+			 * Bit0: dual link connector flag
+			 * =0 connector is single link connector
+			 * =1 connector is dual link connector
+			 */
+			params.acConfig.fDualLinkConnector = 1;
+
+			/* link rate, half for dual link
+			 * We need to convert from KHz units into 20KHz units
+			 */
+			params.usPixelClock =
+					cpu_to_le16((uint16_t)(cntl->pixel_clock / 20));
+		} else
+			/* link rate, half for dual link
+			 * We need to convert from KHz units into 10KHz units
+			 */
+			params.usPixelClock =
+					cpu_to_le16((uint16_t)(cntl->pixel_clock / 10));
+		break;
+	}
+
+	/* 00 - coherent mode
+	 * 01 - incoherent mode
+	 */
+
+	params.acConfig.fCoherentMode = cntl->coherent;
+
+	if ((TRANSMITTER_UNIPHY_B == cntl->transmitter)
+			|| (TRANSMITTER_UNIPHY_D == cntl->transmitter)
+			|| (TRANSMITTER_UNIPHY_F == cntl->transmitter))
+		/* Bit2: Transmitter Link selection
+		 * =0 when bit0=0, single link A/C/E, when bit0=1,
+		 * master link A/C/E
+		 * =1 when bit0=0, single link B/D/F, when bit0=1,
+		 * master link B/D/F
+		 */
+		params.acConfig.ucLinkSel = 1;
+
+	if (ENGINE_ID_DIGB == cntl->engine_id)
+		/* Bit3: Transmitter data source selection
+		 * =0 DIGA is data source.
+		 * =1 DIGB is data source.
+		 * This bit is only useful when ucAction= ATOM_ENABLE
+		 */
+		params.acConfig.ucEncoderSel = 1;
+
+	if (CONNECTOR_ID_DISPLAY_PORT == connector_id ||
+	    CONNECTOR_ID_USBC == connector_id)
+		/* Bit4: DP connector flag
+		 * =0 connector is none-DP connector
+		 * =1 connector is DP connector
+		 */
+		params.acConfig.fDPConnector = 1;
+
+	/* Bit[7:6]: Transmitter selection
+	 * =0 UNIPHY_ENCODER: UNIPHYA/B
+	 * =1 UNIPHY1_ENCODER: UNIPHYC/D
+	 * =2 UNIPHY2_ENCODER: UNIPHYE/F
+	 * =3 reserved
+	 */
+	params.acConfig.ucTransmitterSel =
+			(uint8_t)bp->cmd_helper->transmitter_bp_to_atom(
+					cntl->transmitter);
+
+	params.ucAction = (uint8_t)cntl->action;
+
+	if (EXEC_BIOS_CMD_TABLE(UNIPHYTransmitterControl, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+static enum bp_result transmitter_control_v3(
+	struct bios_parser *bp,
+	struct bp_transmitter_control *cntl)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	DIG_TRANSMITTER_CONTROL_PARAMETERS_V3 params;
+	uint32_t pll_id;
+	enum connector_id conn_id =
+			dal_graphics_object_id_get_connector_id(cntl->connector_obj_id);
+	const struct command_table_helper *cmd = bp->cmd_helper;
+	bool dual_link_conn = (CONNECTOR_ID_DUAL_LINK_DVII == conn_id)
+					|| (CONNECTOR_ID_DUAL_LINK_DVID == conn_id);
+
+	memset(&params, 0, sizeof(params));
+
+	switch (cntl->transmitter) {
+	case TRANSMITTER_UNIPHY_A:
+	case TRANSMITTER_UNIPHY_B:
+	case TRANSMITTER_UNIPHY_C:
+	case TRANSMITTER_UNIPHY_D:
+	case TRANSMITTER_UNIPHY_E:
+	case TRANSMITTER_UNIPHY_F:
+	case TRANSMITTER_TRAVIS_LCD:
+		break;
+	default:
+		return BP_RESULT_BADINPUT;
+	}
+
+	if (!cmd->clock_source_id_to_atom(cntl->pll_id, &pll_id))
+		return BP_RESULT_BADINPUT;
+
+	/* fill information based on the action */
+	switch (cntl->action) {
+	case TRANSMITTER_CONTROL_INIT:
+		if (dual_link_conn) {
+			/* on INIT this bit should be set according to the
+			 * phisycal connector
+			 * Bit0: dual link connector flag
+			 * =0 connector is single link connector
+			 * =1 connector is dual link connector
+			 */
+			params.acConfig.fDualLinkConnector = 1;
+		}
+
+		/* connector object id */
+		params.usInitInfo =
+				cpu_to_le16((uint8_t)(cntl->connector_obj_id.id));
+		break;
+	case TRANSMITTER_CONTROL_SET_VOLTAGE_AND_PREEMPASIS:
+		/* votage swing and pre-emphsis */
+		params.asMode.ucLaneSel = (uint8_t)cntl->lane_select;
+		params.asMode.ucLaneSet = (uint8_t)cntl->lane_settings;
+		break;
+	default:
+		if (dual_link_conn && cntl->multi_path)
+			/* on ENABLE/DISABLE this bit should be set according to
+			 * actual timing (number of lanes)
+			 * Bit0: dual link connector flag
+			 * =0 connector is single link connector
+			 * =1 connector is dual link connector
+			 */
+			params.acConfig.fDualLinkConnector = 1;
+
+		/* if dual-link */
+		if (LANE_COUNT_FOUR < cntl->lanes_number) {
+			/* on ENABLE/DISABLE this bit should be set according to
+			 * actual timing (number of lanes)
+			 * Bit0: dual link connector flag
+			 * =0 connector is single link connector
+			 * =1 connector is dual link connector
+			 */
+			params.acConfig.fDualLinkConnector = 1;
+
+			/* link rate, half for dual link
+			 * We need to convert from KHz units into 20KHz units
+			 */
+			params.usPixelClock =
+					cpu_to_le16((uint16_t)(cntl->pixel_clock / 20));
+		} else {
+			/* link rate, half for dual link
+			 * We need to convert from KHz units into 10KHz units
+			 */
+			params.usPixelClock =
+					cpu_to_le16((uint16_t)(cntl->pixel_clock / 10));
+		}
+		break;
+	}
+
+	/* 00 - coherent mode
+	 * 01 - incoherent mode
+	 */
+
+	params.acConfig.fCoherentMode = cntl->coherent;
+
+	if ((TRANSMITTER_UNIPHY_B == cntl->transmitter)
+		|| (TRANSMITTER_UNIPHY_D == cntl->transmitter)
+		|| (TRANSMITTER_UNIPHY_F == cntl->transmitter))
+		/* Bit2: Transmitter Link selection
+		 * =0 when bit0=0, single link A/C/E, when bit0=1,
+		 * master link A/C/E
+		 * =1 when bit0=0, single link B/D/F, when bit0=1,
+		 * master link B/D/F
+		 */
+		params.acConfig.ucLinkSel = 1;
+
+	if (ENGINE_ID_DIGB == cntl->engine_id)
+		/* Bit3: Transmitter data source selection
+		 * =0 DIGA is data source.
+		 * =1 DIGB is data source.
+		 * This bit is only useful when ucAction= ATOM_ENABLE
+		 */
+		params.acConfig.ucEncoderSel = 1;
+
+	/* Bit[7:6]: Transmitter selection
+	 * =0 UNIPHY_ENCODER: UNIPHYA/B
+	 * =1 UNIPHY1_ENCODER: UNIPHYC/D
+	 * =2 UNIPHY2_ENCODER: UNIPHYE/F
+	 * =3 reserved
+	 */
+	params.acConfig.ucTransmitterSel =
+			(uint8_t)cmd->transmitter_bp_to_atom(cntl->transmitter);
+
+	params.ucLaneNum = (uint8_t)cntl->lanes_number;
+
+	params.acConfig.ucRefClkSource = (uint8_t)pll_id;
+
+	params.ucAction = (uint8_t)cntl->action;
+
+	if (EXEC_BIOS_CMD_TABLE(UNIPHYTransmitterControl, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+static enum bp_result transmitter_control_v4(
+	struct bios_parser *bp,
+	struct bp_transmitter_control *cntl)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	DIG_TRANSMITTER_CONTROL_PARAMETERS_V4 params;
+	uint32_t ref_clk_src_id;
+	enum connector_id conn_id =
+			dal_graphics_object_id_get_connector_id(cntl->connector_obj_id);
+	const struct command_table_helper *cmd = bp->cmd_helper;
+
+	memset(&params, 0, sizeof(params));
+
+	switch (cntl->transmitter) {
+	case TRANSMITTER_UNIPHY_A:
+	case TRANSMITTER_UNIPHY_B:
+	case TRANSMITTER_UNIPHY_C:
+	case TRANSMITTER_UNIPHY_D:
+	case TRANSMITTER_UNIPHY_E:
+	case TRANSMITTER_UNIPHY_F:
+	case TRANSMITTER_TRAVIS_LCD:
+		break;
+	default:
+		return BP_RESULT_BADINPUT;
+	}
+
+	if (!cmd->clock_source_id_to_ref_clk_src(cntl->pll_id, &ref_clk_src_id))
+		return BP_RESULT_BADINPUT;
+
+	switch (cntl->action) {
+	case TRANSMITTER_CONTROL_INIT:
+	{
+		if ((CONNECTOR_ID_DUAL_LINK_DVII == conn_id) ||
+				(CONNECTOR_ID_DUAL_LINK_DVID == conn_id))
+			/* on INIT this bit should be set according to the
+			 * phisycal connector
+			 * Bit0: dual link connector flag
+			 * =0 connector is single link connector
+			 * =1 connector is dual link connector
+			 */
+			params.acConfig.fDualLinkConnector = 1;
+
+		/* connector object id */
+		params.usInitInfo =
+				cpu_to_le16((uint8_t)(cntl->connector_obj_id.id));
+	}
+	break;
+	case TRANSMITTER_CONTROL_SET_VOLTAGE_AND_PREEMPASIS:
+		/* votage swing and pre-emphsis */
+		params.asMode.ucLaneSel = (uint8_t)(cntl->lane_select);
+		params.asMode.ucLaneSet = (uint8_t)(cntl->lane_settings);
+		break;
+	default:
+		if ((CONNECTOR_ID_DUAL_LINK_DVII == conn_id) ||
+				(CONNECTOR_ID_DUAL_LINK_DVID == conn_id))
+			/* on ENABLE/DISABLE this bit should be set according to
+			 * actual timing (number of lanes)
+			 * Bit0: dual link connector flag
+			 * =0 connector is single link connector
+			 * =1 connector is dual link connector
+			 */
+			params.acConfig.fDualLinkConnector = 1;
+
+		/* if dual-link */
+		if (LANE_COUNT_FOUR < cntl->lanes_number)
+			/* link rate, half for dual link
+			 * We need to convert from KHz units into 20KHz units
+			 */
+			params.usPixelClock =
+					cpu_to_le16((uint16_t)(cntl->pixel_clock / 20));
+		else {
+			/* link rate, half for dual link
+			 * We need to convert from KHz units into 10KHz units
+			 */
+			params.usPixelClock =
+					cpu_to_le16((uint16_t)(cntl->pixel_clock / 10));
+		}
+		break;
+	}
+
+	/* 00 - coherent mode
+	 * 01 - incoherent mode
+	 */
+
+	params.acConfig.fCoherentMode = cntl->coherent;
+
+	if ((TRANSMITTER_UNIPHY_B == cntl->transmitter)
+		|| (TRANSMITTER_UNIPHY_D == cntl->transmitter)
+		|| (TRANSMITTER_UNIPHY_F == cntl->transmitter))
+		/* Bit2: Transmitter Link selection
+		 * =0 when bit0=0, single link A/C/E, when bit0=1,
+		 * master link A/C/E
+		 * =1 when bit0=0, single link B/D/F, when bit0=1,
+		 * master link B/D/F
+		 */
+		params.acConfig.ucLinkSel = 1;
+
+	if (ENGINE_ID_DIGB == cntl->engine_id)
+		/* Bit3: Transmitter data source selection
+		 * =0 DIGA is data source.
+		 * =1 DIGB is data source.
+		 * This bit is only useful when ucAction= ATOM_ENABLE
+		 */
+		params.acConfig.ucEncoderSel = 1;
+
+	/* Bit[7:6]: Transmitter selection
+	 * =0 UNIPHY_ENCODER: UNIPHYA/B
+	 * =1 UNIPHY1_ENCODER: UNIPHYC/D
+	 * =2 UNIPHY2_ENCODER: UNIPHYE/F
+	 * =3 reserved
+	 */
+	params.acConfig.ucTransmitterSel =
+		(uint8_t)(cmd->transmitter_bp_to_atom(cntl->transmitter));
+	params.ucLaneNum = (uint8_t)(cntl->lanes_number);
+	params.acConfig.ucRefClkSource = (uint8_t)(ref_clk_src_id);
+	params.ucAction = (uint8_t)(cntl->action);
+
+	if (EXEC_BIOS_CMD_TABLE(UNIPHYTransmitterControl, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+static enum bp_result transmitter_control_v1_5(
+	struct bios_parser *bp,
+	struct bp_transmitter_control *cntl)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	const struct command_table_helper *cmd = bp->cmd_helper;
+	DIG_TRANSMITTER_CONTROL_PARAMETERS_V1_5 params;
+
+	memset(&params, 0, sizeof(params));
+	params.ucPhyId = cmd->phy_id_to_atom(cntl->transmitter);
+	params.ucAction = (uint8_t)cntl->action;
+	params.ucLaneNum = (uint8_t)cntl->lanes_number;
+	params.ucConnObjId = (uint8_t)cntl->connector_obj_id.id;
+
+	params.ucDigMode =
+		cmd->signal_type_to_atom_dig_mode(cntl->signal);
+	params.asConfig.ucPhyClkSrcId =
+		cmd->clock_source_id_to_atom_phy_clk_src_id(cntl->pll_id);
+	/* 00 - coherent mode */
+	params.asConfig.ucCoherentMode = cntl->coherent;
+	params.asConfig.ucHPDSel =
+		cmd->hpd_sel_to_atom(cntl->hpd_sel);
+	params.ucDigEncoderSel =
+		cmd->dig_encoder_sel_to_atom(cntl->engine_id);
+	params.ucDPLaneSet = (uint8_t) cntl->lane_settings;
+	params.usSymClock = cpu_to_le16((uint16_t) (cntl->pixel_clock / 10));
+	/*
+	 * In SI/TN case, caller have to set usPixelClock as following:
+	 * DP mode: usPixelClock = DP_LINK_CLOCK/10
+	 * (DP_LINK_CLOCK = 1.62GHz, 2.7GHz, 5.4GHz)
+	 * DVI single link mode: usPixelClock = pixel clock
+	 * DVI dual link mode: usPixelClock = pixel clock
+	 * HDMI mode: usPixelClock = pixel clock * deep_color_ratio
+	 * (=1: 8bpp, =1.25: 10bpp, =1.5:12bpp, =2: 16bpp)
+	 * LVDS mode: usPixelClock = pixel clock
+	 */
+	if  (cntl->signal == SIGNAL_TYPE_HDMI_TYPE_A) {
+		switch (cntl->color_depth) {
+		case COLOR_DEPTH_101010:
+			params.usSymClock =
+				cpu_to_le16((le16_to_cpu(params.usSymClock) * 30) / 24);
+			break;
+		case COLOR_DEPTH_121212:
+			params.usSymClock =
+				cpu_to_le16((le16_to_cpu(params.usSymClock) * 36) / 24);
+			break;
+		case COLOR_DEPTH_161616:
+			params.usSymClock =
+				cpu_to_le16((le16_to_cpu(params.usSymClock) * 48) / 24);
+			break;
+		default:
+			break;
+		}
+	}
+
+	if (EXEC_BIOS_CMD_TABLE(UNIPHYTransmitterControl, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+static enum bp_result transmitter_control_v1_6(
+	struct bios_parser *bp,
+	struct bp_transmitter_control *cntl)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	const struct command_table_helper *cmd = bp->cmd_helper;
+	DIG_TRANSMITTER_CONTROL_PARAMETERS_V1_6 params;
+
+	memset(&params, 0, sizeof(params));
+	params.ucPhyId = cmd->phy_id_to_atom(cntl->transmitter);
+	params.ucAction = (uint8_t)cntl->action;
+
+	if (cntl->action == TRANSMITTER_CONTROL_SET_VOLTAGE_AND_PREEMPASIS)
+		params.ucDPLaneSet = (uint8_t)cntl->lane_settings;
+	else
+		params.ucDigMode = cmd->signal_type_to_atom_dig_mode(cntl->signal);
+
+	params.ucLaneNum = (uint8_t)cntl->lanes_number;
+	params.ucHPDSel = cmd->hpd_sel_to_atom(cntl->hpd_sel);
+	params.ucDigEncoderSel = cmd->dig_encoder_sel_to_atom(cntl->engine_id);
+	params.ucConnObjId = (uint8_t)cntl->connector_obj_id.id;
+	params.ulSymClock = cntl->pixel_clock/10;
+
+	/*
+	 * In SI/TN case, caller have to set usPixelClock as following:
+	 * DP mode: usPixelClock = DP_LINK_CLOCK/10
+	 * (DP_LINK_CLOCK = 1.62GHz, 2.7GHz, 5.4GHz)
+	 * DVI single link mode: usPixelClock = pixel clock
+	 * DVI dual link mode: usPixelClock = pixel clock
+	 * HDMI mode: usPixelClock = pixel clock * deep_color_ratio
+	 * (=1: 8bpp, =1.25: 10bpp, =1.5:12bpp, =2: 16bpp)
+	 * LVDS mode: usPixelClock = pixel clock
+	 */
+	switch (cntl->signal) {
+	case SIGNAL_TYPE_HDMI_TYPE_A:
+		switch (cntl->color_depth) {
+		case COLOR_DEPTH_101010:
+			params.ulSymClock =
+				cpu_to_le16((le16_to_cpu(params.ulSymClock) * 30) / 24);
+			break;
+		case COLOR_DEPTH_121212:
+			params.ulSymClock =
+				cpu_to_le16((le16_to_cpu(params.ulSymClock) * 36) / 24);
+			break;
+		case COLOR_DEPTH_161616:
+			params.ulSymClock =
+				cpu_to_le16((le16_to_cpu(params.ulSymClock) * 48) / 24);
+			break;
+		default:
+			break;
+		}
+		break;
+		default:
+			break;
+	}
+
+	if (EXEC_BIOS_CMD_TABLE(UNIPHYTransmitterControl, params))
+		result = BP_RESULT_OK;
+	return result;
+}
+
+/*******************************************************************************
+ ********************************************************************************
+ **
+ **                  SET PIXEL CLOCK
+ **
+ ********************************************************************************
+ *******************************************************************************/
+
+static enum bp_result set_pixel_clock_v3(
+	struct bios_parser *bp,
+	struct bp_pixel_clock_parameters *bp_params);
+static enum bp_result set_pixel_clock_v5(
+	struct bios_parser *bp,
+	struct bp_pixel_clock_parameters *bp_params);
+static enum bp_result set_pixel_clock_v6(
+	struct bios_parser *bp,
+	struct bp_pixel_clock_parameters *bp_params);
+static enum bp_result set_pixel_clock_v7(
+	struct bios_parser *bp,
+	struct bp_pixel_clock_parameters *bp_params);
+
+static void init_set_pixel_clock(struct bios_parser *bp)
+{
+	switch (BIOS_CMD_TABLE_PARA_REVISION(SetPixelClock)) {
+	case 3:
+		bp->cmd_tbl.set_pixel_clock = set_pixel_clock_v3;
+		break;
+	case 5:
+		bp->cmd_tbl.set_pixel_clock = set_pixel_clock_v5;
+		break;
+	case 6:
+		bp->cmd_tbl.set_pixel_clock = set_pixel_clock_v6;
+		break;
+	case 7:
+		bp->cmd_tbl.set_pixel_clock = set_pixel_clock_v7;
+		break;
+	default:
+		dm_output_to_console("Don't have set_pixel_clock for v%d\n",
+			 BIOS_CMD_TABLE_PARA_REVISION(SetPixelClock));
+		bp->cmd_tbl.set_pixel_clock = NULL;
+		break;
+	}
+}
+
+static enum bp_result set_pixel_clock_v3(
+	struct bios_parser *bp,
+	struct bp_pixel_clock_parameters *bp_params)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	PIXEL_CLOCK_PARAMETERS_V3 *params;
+	SET_PIXEL_CLOCK_PS_ALLOCATION allocation;
+
+	memset(&allocation, 0, sizeof(allocation));
+
+	if (CLOCK_SOURCE_ID_PLL1 == bp_params->pll_id)
+		allocation.sPCLKInput.ucPpll = ATOM_PPLL1;
+	else if (CLOCK_SOURCE_ID_PLL2 == bp_params->pll_id)
+		allocation.sPCLKInput.ucPpll = ATOM_PPLL2;
+	else
+		return BP_RESULT_BADINPUT;
+
+	allocation.sPCLKInput.usRefDiv =
+			cpu_to_le16((uint16_t)bp_params->reference_divider);
+	allocation.sPCLKInput.usFbDiv =
+			cpu_to_le16((uint16_t)bp_params->feedback_divider);
+	allocation.sPCLKInput.ucFracFbDiv =
+			(uint8_t)(bp_params->fractional_feedback_divider / 100000);
+	allocation.sPCLKInput.ucPostDiv =
+			(uint8_t)bp_params->pixel_clock_post_divider;
+
+	/* We need to convert from 100Hz units into 10KHz units */
+	allocation.sPCLKInput.usPixelClock =
+			cpu_to_le16((uint16_t)(bp_params->target_pixel_clock_100hz / 100));
+
+	params = (PIXEL_CLOCK_PARAMETERS_V3 *)&allocation.sPCLKInput;
+	params->ucTransmitterId =
+			bp->cmd_helper->encoder_id_to_atom(
+					dal_graphics_object_id_get_encoder_id(
+							bp_params->encoder_object_id));
+	params->ucEncoderMode =
+			(uint8_t)(bp->cmd_helper->encoder_mode_bp_to_atom(
+					bp_params->signal_type, false));
+
+	if (bp_params->flags.FORCE_PROGRAMMING_OF_PLL)
+		params->ucMiscInfo |= PIXEL_CLOCK_MISC_FORCE_PROG_PPLL;
+
+	if (bp_params->flags.USE_E_CLOCK_AS_SOURCE_FOR_D_CLOCK)
+		params->ucMiscInfo |= PIXEL_CLOCK_MISC_USE_ENGINE_FOR_DISPCLK;
+
+	if (CONTROLLER_ID_D1 != bp_params->controller_id)
+		params->ucMiscInfo |= PIXEL_CLOCK_MISC_CRTC_SEL_CRTC2;
+
+	if (EXEC_BIOS_CMD_TABLE(SetPixelClock, allocation))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+#ifndef SET_PIXEL_CLOCK_PS_ALLOCATION_V5
+/* video bios did not define this: */
+typedef struct _SET_PIXEL_CLOCK_PS_ALLOCATION_V5 {
+	PIXEL_CLOCK_PARAMETERS_V5 sPCLKInput;
+	/* Caller doesn't need to init this portion */
+	ENABLE_SPREAD_SPECTRUM_ON_PPLL sReserved;
+} SET_PIXEL_CLOCK_PS_ALLOCATION_V5;
+#endif
+
+#ifndef SET_PIXEL_CLOCK_PS_ALLOCATION_V6
+/* video bios did not define this: */
+typedef struct _SET_PIXEL_CLOCK_PS_ALLOCATION_V6 {
+	PIXEL_CLOCK_PARAMETERS_V6 sPCLKInput;
+	/* Caller doesn't need to init this portion */
+	ENABLE_SPREAD_SPECTRUM_ON_PPLL sReserved;
+} SET_PIXEL_CLOCK_PS_ALLOCATION_V6;
+#endif
+
+static enum bp_result set_pixel_clock_v5(
+	struct bios_parser *bp,
+	struct bp_pixel_clock_parameters *bp_params)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	SET_PIXEL_CLOCK_PS_ALLOCATION_V5 clk;
+	uint8_t controller_id;
+	uint32_t pll_id;
+
+	memset(&clk, 0, sizeof(clk));
+
+	if (bp->cmd_helper->clock_source_id_to_atom(bp_params->pll_id, &pll_id)
+			&& bp->cmd_helper->controller_id_to_atom(
+					bp_params->controller_id, &controller_id)) {
+		clk.sPCLKInput.ucCRTC = controller_id;
+		clk.sPCLKInput.ucPpll = (uint8_t)pll_id;
+		clk.sPCLKInput.ucRefDiv =
+				(uint8_t)(bp_params->reference_divider);
+		clk.sPCLKInput.usFbDiv =
+				cpu_to_le16((uint16_t)(bp_params->feedback_divider));
+		clk.sPCLKInput.ulFbDivDecFrac =
+				cpu_to_le32(bp_params->fractional_feedback_divider);
+		clk.sPCLKInput.ucPostDiv =
+				(uint8_t)(bp_params->pixel_clock_post_divider);
+		clk.sPCLKInput.ucTransmitterID =
+				bp->cmd_helper->encoder_id_to_atom(
+						dal_graphics_object_id_get_encoder_id(
+								bp_params->encoder_object_id));
+		clk.sPCLKInput.ucEncoderMode =
+				(uint8_t)bp->cmd_helper->encoder_mode_bp_to_atom(
+						bp_params->signal_type, false);
+
+		/* We need to convert from 100Hz units into 10KHz units */
+		clk.sPCLKInput.usPixelClock =
+				cpu_to_le16((uint16_t)(bp_params->target_pixel_clock_100hz / 100));
+
+		if (bp_params->flags.FORCE_PROGRAMMING_OF_PLL)
+			clk.sPCLKInput.ucMiscInfo |=
+					PIXEL_CLOCK_MISC_FORCE_PROG_PPLL;
+
+		if (bp_params->flags.SET_EXTERNAL_REF_DIV_SRC)
+			clk.sPCLKInput.ucMiscInfo |=
+					PIXEL_CLOCK_MISC_REF_DIV_SRC;
+
+		/* clkV5.ucMiscInfo bit[3:2]= HDMI panel bit depth: =0: 24bpp
+		 * =1:30bpp, =2:32bpp
+		 * driver choose program it itself, i.e. here we program it
+		 * to 888 by default.
+		 */
+		if (bp_params->signal_type == SIGNAL_TYPE_HDMI_TYPE_A)
+			switch (bp_params->color_depth) {
+			case TRANSMITTER_COLOR_DEPTH_30:
+				/* yes this is correct, the atom define is wrong */
+				clk.sPCLKInput.ucMiscInfo |= PIXEL_CLOCK_V5_MISC_HDMI_32BPP;
+				break;
+			case TRANSMITTER_COLOR_DEPTH_36:
+				/* yes this is correct, the atom define is wrong */
+				clk.sPCLKInput.ucMiscInfo |= PIXEL_CLOCK_V5_MISC_HDMI_30BPP;
+				break;
+			default:
+				break;
+			}
+
+		if (EXEC_BIOS_CMD_TABLE(SetPixelClock, clk))
+			result = BP_RESULT_OK;
+	}
+
+	return result;
+}
+
+static enum bp_result set_pixel_clock_v6(
+	struct bios_parser *bp,
+	struct bp_pixel_clock_parameters *bp_params)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	SET_PIXEL_CLOCK_PS_ALLOCATION_V6 clk;
+	uint8_t controller_id;
+	uint32_t pll_id;
+
+	memset(&clk, 0, sizeof(clk));
+
+	if (bp->cmd_helper->clock_source_id_to_atom(bp_params->pll_id, &pll_id)
+			&& bp->cmd_helper->controller_id_to_atom(
+					bp_params->controller_id, &controller_id)) {
+		/* Note: VBIOS still wants to use ucCRTC name which is now
+		 * 1 byte in ULONG
+		 *typedef struct _CRTC_PIXEL_CLOCK_FREQ
+		 *{
+		 * target the pixel clock to drive the CRTC timing.
+		 * ULONG ulPixelClock:24;
+		 * 0 means disable PPLL/DCPLL. Expanded to 24 bits comparing to
+		 * previous version.
+		 * ATOM_CRTC1~6, indicate the CRTC controller to
+		 * ULONG ucCRTC:8;
+		 * drive the pixel clock. not used for DCPLL case.
+		 *}CRTC_PIXEL_CLOCK_FREQ;
+		 *union
+		 *{
+		 * pixel clock and CRTC id frequency
+		 * CRTC_PIXEL_CLOCK_FREQ ulCrtcPclkFreq;
+		 * ULONG ulDispEngClkFreq; dispclk frequency
+		 *};
+		 */
+		clk.sPCLKInput.ulCrtcPclkFreq.ucCRTC = controller_id;
+		clk.sPCLKInput.ucPpll = (uint8_t) pll_id;
+		clk.sPCLKInput.ucRefDiv =
+				(uint8_t) bp_params->reference_divider;
+		clk.sPCLKInput.usFbDiv =
+				cpu_to_le16((uint16_t) bp_params->feedback_divider);
+		clk.sPCLKInput.ulFbDivDecFrac =
+				cpu_to_le32(bp_params->fractional_feedback_divider);
+		clk.sPCLKInput.ucPostDiv =
+				(uint8_t) bp_params->pixel_clock_post_divider;
+		clk.sPCLKInput.ucTransmitterID =
+				bp->cmd_helper->encoder_id_to_atom(
+						dal_graphics_object_id_get_encoder_id(
+								bp_params->encoder_object_id));
+		clk.sPCLKInput.ucEncoderMode =
+				(uint8_t) bp->cmd_helper->encoder_mode_bp_to_atom(
+						bp_params->signal_type, false);
+
+		/* We need to convert from 100 Hz units into 10KHz units */
+		clk.sPCLKInput.ulCrtcPclkFreq.ulPixelClock =
+				cpu_to_le32(bp_params->target_pixel_clock_100hz / 100);
+
+		if (bp_params->flags.FORCE_PROGRAMMING_OF_PLL) {
+			clk.sPCLKInput.ucMiscInfo |=
+					PIXEL_CLOCK_V6_MISC_FORCE_PROG_PPLL;
+		}
+
+		if (bp_params->flags.SET_EXTERNAL_REF_DIV_SRC) {
+			clk.sPCLKInput.ucMiscInfo |=
+					PIXEL_CLOCK_V6_MISC_REF_DIV_SRC;
+		}
+
+		/* clkV6.ucMiscInfo bit[3:2]= HDMI panel bit depth: =0:
+		 * 24bpp =1:30bpp, =2:32bpp
+		 * driver choose program it itself, i.e. here we pass required
+		 * target rate that includes deep color.
+		 */
+		if (bp_params->signal_type == SIGNAL_TYPE_HDMI_TYPE_A)
+			switch (bp_params->color_depth) {
+			case TRANSMITTER_COLOR_DEPTH_30:
+				clk.sPCLKInput.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_HDMI_30BPP_V6;
+				break;
+			case TRANSMITTER_COLOR_DEPTH_36:
+				clk.sPCLKInput.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_HDMI_36BPP_V6;
+				break;
+			case TRANSMITTER_COLOR_DEPTH_48:
+				clk.sPCLKInput.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_HDMI_48BPP;
+				break;
+			default:
+				break;
+			}
+
+		if (EXEC_BIOS_CMD_TABLE(SetPixelClock, clk))
+			result = BP_RESULT_OK;
+	}
+
+	return result;
+}
+
+static enum bp_result set_pixel_clock_v7(
+	struct bios_parser *bp,
+	struct bp_pixel_clock_parameters *bp_params)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	PIXEL_CLOCK_PARAMETERS_V7 clk;
+	uint8_t controller_id;
+	uint32_t pll_id;
+
+	memset(&clk, 0, sizeof(clk));
+
+	if (bp->cmd_helper->clock_source_id_to_atom(bp_params->pll_id, &pll_id)
+			&& bp->cmd_helper->controller_id_to_atom(bp_params->controller_id, &controller_id)) {
+		/* Note: VBIOS still wants to use ucCRTC name which is now
+		 * 1 byte in ULONG
+		 *typedef struct _CRTC_PIXEL_CLOCK_FREQ
+		 *{
+		 * target the pixel clock to drive the CRTC timing.
+		 * ULONG ulPixelClock:24;
+		 * 0 means disable PPLL/DCPLL. Expanded to 24 bits comparing to
+		 * previous version.
+		 * ATOM_CRTC1~6, indicate the CRTC controller to
+		 * ULONG ucCRTC:8;
+		 * drive the pixel clock. not used for DCPLL case.
+		 *}CRTC_PIXEL_CLOCK_FREQ;
+		 *union
+		 *{
+		 * pixel clock and CRTC id frequency
+		 * CRTC_PIXEL_CLOCK_FREQ ulCrtcPclkFreq;
+		 * ULONG ulDispEngClkFreq; dispclk frequency
+		 *};
+		 */
+		clk.ucCRTC = controller_id;
+		clk.ucPpll = (uint8_t) pll_id;
+		clk.ucTransmitterID = bp->cmd_helper->encoder_id_to_atom(dal_graphics_object_id_get_encoder_id(bp_params->encoder_object_id));
+		clk.ucEncoderMode = (uint8_t) bp->cmd_helper->encoder_mode_bp_to_atom(bp_params->signal_type, false);
+
+		clk.ulPixelClock = cpu_to_le32(bp_params->target_pixel_clock_100hz);
+
+		clk.ucDeepColorRatio = (uint8_t) bp->cmd_helper->transmitter_color_depth_to_atom(bp_params->color_depth);
+
+		if (bp_params->flags.FORCE_PROGRAMMING_OF_PLL)
+			clk.ucMiscInfo |= PIXEL_CLOCK_V7_MISC_FORCE_PROG_PPLL;
+
+		if (bp_params->flags.SET_EXTERNAL_REF_DIV_SRC)
+			clk.ucMiscInfo |= PIXEL_CLOCK_V7_MISC_REF_DIV_SRC;
+
+		if (bp_params->flags.PROGRAM_PHY_PLL_ONLY)
+			clk.ucMiscInfo |= PIXEL_CLOCK_V7_MISC_PROG_PHYPLL;
+
+		if (bp_params->flags.SUPPORT_YUV_420)
+			clk.ucMiscInfo |= PIXEL_CLOCK_V7_MISC_YUV420_MODE;
+
+		if (bp_params->flags.SET_XTALIN_REF_SRC)
+			clk.ucMiscInfo |= PIXEL_CLOCK_V7_MISC_REF_DIV_SRC_XTALIN;
+
+		if (bp_params->flags.SET_GENLOCK_REF_DIV_SRC)
+			clk.ucMiscInfo |= PIXEL_CLOCK_V7_MISC_REF_DIV_SRC_GENLK;
+
+		if (bp_params->signal_type == SIGNAL_TYPE_DVI_DUAL_LINK)
+			clk.ucMiscInfo |= PIXEL_CLOCK_V7_MISC_DVI_DUALLINK_EN;
+
+		if (EXEC_BIOS_CMD_TABLE(SetPixelClock, clk))
+			result = BP_RESULT_OK;
+	}
+	return result;
+}
+
+/*******************************************************************************
+ ********************************************************************************
+ **
+ **                  ENABLE PIXEL CLOCK SS
+ **
+ ********************************************************************************
+ *******************************************************************************/
+static enum bp_result enable_spread_spectrum_on_ppll_v1(
+	struct bios_parser *bp,
+	struct bp_spread_spectrum_parameters *bp_params,
+	bool enable);
+static enum bp_result enable_spread_spectrum_on_ppll_v2(
+	struct bios_parser *bp,
+	struct bp_spread_spectrum_parameters *bp_params,
+	bool enable);
+static enum bp_result enable_spread_spectrum_on_ppll_v3(
+	struct bios_parser *bp,
+	struct bp_spread_spectrum_parameters *bp_params,
+	bool enable);
+
+static void init_enable_spread_spectrum_on_ppll(struct bios_parser *bp)
+{
+	switch (BIOS_CMD_TABLE_PARA_REVISION(EnableSpreadSpectrumOnPPLL)) {
+	case 1:
+		bp->cmd_tbl.enable_spread_spectrum_on_ppll =
+				enable_spread_spectrum_on_ppll_v1;
+		break;
+	case 2:
+		bp->cmd_tbl.enable_spread_spectrum_on_ppll =
+				enable_spread_spectrum_on_ppll_v2;
+		break;
+	case 3:
+		bp->cmd_tbl.enable_spread_spectrum_on_ppll =
+				enable_spread_spectrum_on_ppll_v3;
+		break;
+	default:
+		dm_output_to_console("Don't have enable_spread_spectrum_on_ppll for v%d\n",
+			 BIOS_CMD_TABLE_PARA_REVISION(EnableSpreadSpectrumOnPPLL));
+		bp->cmd_tbl.enable_spread_spectrum_on_ppll = NULL;
+		break;
+	}
+}
+
+static enum bp_result enable_spread_spectrum_on_ppll_v1(
+	struct bios_parser *bp,
+	struct bp_spread_spectrum_parameters *bp_params,
+	bool enable)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	ENABLE_SPREAD_SPECTRUM_ON_PPLL params;
+
+	memset(&params, 0, sizeof(params));
+
+	if ((enable == true) && (bp_params->percentage > 0))
+		params.ucEnable = ATOM_ENABLE;
+	else
+		params.ucEnable = ATOM_DISABLE;
+
+	params.usSpreadSpectrumPercentage =
+			cpu_to_le16((uint16_t)bp_params->percentage);
+	params.ucSpreadSpectrumStep =
+			(uint8_t)bp_params->ver1.step;
+	params.ucSpreadSpectrumDelay =
+			(uint8_t)bp_params->ver1.delay;
+	/* convert back to unit of 10KHz */
+	params.ucSpreadSpectrumRange =
+			(uint8_t)(bp_params->ver1.range / 10000);
+
+	if (bp_params->flags.EXTERNAL_SS)
+		params.ucSpreadSpectrumType |= ATOM_EXTERNAL_SS_MASK;
+
+	if (bp_params->flags.CENTER_SPREAD)
+		params.ucSpreadSpectrumType |= ATOM_SS_CENTRE_SPREAD_MODE;
+
+	if (bp_params->pll_id == CLOCK_SOURCE_ID_PLL1)
+		params.ucPpll = ATOM_PPLL1;
+	else if (bp_params->pll_id == CLOCK_SOURCE_ID_PLL2)
+		params.ucPpll = ATOM_PPLL2;
+	else
+		BREAK_TO_DEBUGGER(); /* Unexpected PLL value!! */
+
+	if (EXEC_BIOS_CMD_TABLE(EnableSpreadSpectrumOnPPLL, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+static enum bp_result enable_spread_spectrum_on_ppll_v2(
+	struct bios_parser *bp,
+	struct bp_spread_spectrum_parameters *bp_params,
+	bool enable)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	ENABLE_SPREAD_SPECTRUM_ON_PPLL_V2 params;
+
+	memset(&params, 0, sizeof(params));
+
+	if (bp_params->pll_id == CLOCK_SOURCE_ID_PLL1)
+		params.ucSpreadSpectrumType = ATOM_PPLL_SS_TYPE_V2_P1PLL;
+	else if (bp_params->pll_id == CLOCK_SOURCE_ID_PLL2)
+		params.ucSpreadSpectrumType = ATOM_PPLL_SS_TYPE_V2_P2PLL;
+	else
+		BREAK_TO_DEBUGGER(); /* Unexpected PLL value!! */
+
+	if ((enable == true) && (bp_params->percentage > 0)) {
+		params.ucEnable = ATOM_ENABLE;
+
+		params.usSpreadSpectrumPercentage =
+				cpu_to_le16((uint16_t)(bp_params->percentage));
+		params.usSpreadSpectrumStep =
+				cpu_to_le16((uint16_t)(bp_params->ds.ds_frac_size));
+
+		if (bp_params->flags.EXTERNAL_SS)
+			params.ucSpreadSpectrumType |=
+					ATOM_PPLL_SS_TYPE_V2_EXT_SPREAD;
+
+		if (bp_params->flags.CENTER_SPREAD)
+			params.ucSpreadSpectrumType |=
+					ATOM_PPLL_SS_TYPE_V2_CENTRE_SPREAD;
+
+		/* Both amounts need to be left shifted first before bit
+		 * comparison. Otherwise, the result will always be zero here
+		 */
+		params.usSpreadSpectrumAmount = cpu_to_le16((uint16_t)(
+				((bp_params->ds.feedback_amount <<
+						ATOM_PPLL_SS_AMOUNT_V2_FBDIV_SHIFT) &
+						ATOM_PPLL_SS_AMOUNT_V2_FBDIV_MASK) |
+						((bp_params->ds.nfrac_amount <<
+								ATOM_PPLL_SS_AMOUNT_V2_NFRAC_SHIFT) &
+								ATOM_PPLL_SS_AMOUNT_V2_NFRAC_MASK)));
+	} else
+		params.ucEnable = ATOM_DISABLE;
+
+	if (EXEC_BIOS_CMD_TABLE(EnableSpreadSpectrumOnPPLL, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+static enum bp_result enable_spread_spectrum_on_ppll_v3(
+	struct bios_parser *bp,
+	struct bp_spread_spectrum_parameters *bp_params,
+	bool enable)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	ENABLE_SPREAD_SPECTRUM_ON_PPLL_V3 params;
+
+	memset(&params, 0, sizeof(params));
+
+	switch (bp_params->pll_id) {
+	case CLOCK_SOURCE_ID_PLL0:
+		/* ATOM_PPLL_SS_TYPE_V3_P0PLL; this is pixel clock only,
+		 * not for SI display clock.
+		 */
+		params.ucSpreadSpectrumType = ATOM_PPLL_SS_TYPE_V3_DCPLL;
+		break;
+	case CLOCK_SOURCE_ID_PLL1:
+		params.ucSpreadSpectrumType = ATOM_PPLL_SS_TYPE_V3_P1PLL;
+		break;
+
+	case CLOCK_SOURCE_ID_PLL2:
+		params.ucSpreadSpectrumType = ATOM_PPLL_SS_TYPE_V3_P2PLL;
+		break;
+
+	case CLOCK_SOURCE_ID_DCPLL:
+		params.ucSpreadSpectrumType = ATOM_PPLL_SS_TYPE_V3_DCPLL;
+		break;
+
+	default:
+		BREAK_TO_DEBUGGER();
+		/* Unexpected PLL value!! */
+		return result;
+	}
+
+	if (enable == true) {
+		params.ucEnable = ATOM_ENABLE;
+
+		params.usSpreadSpectrumAmountFrac =
+				cpu_to_le16((uint16_t)(bp_params->ds_frac_amount));
+		params.usSpreadSpectrumStep =
+				cpu_to_le16((uint16_t)(bp_params->ds.ds_frac_size));
+
+		if (bp_params->flags.EXTERNAL_SS)
+			params.ucSpreadSpectrumType |=
+					ATOM_PPLL_SS_TYPE_V3_EXT_SPREAD;
+		if (bp_params->flags.CENTER_SPREAD)
+			params.ucSpreadSpectrumType |=
+					ATOM_PPLL_SS_TYPE_V3_CENTRE_SPREAD;
+
+		/* Both amounts need to be left shifted first before bit
+		 * comparison. Otherwise, the result will always be zero here
+		 */
+		params.usSpreadSpectrumAmount = cpu_to_le16((uint16_t)(
+				((bp_params->ds.feedback_amount <<
+						ATOM_PPLL_SS_AMOUNT_V3_FBDIV_SHIFT) &
+						ATOM_PPLL_SS_AMOUNT_V3_FBDIV_MASK) |
+						((bp_params->ds.nfrac_amount <<
+								ATOM_PPLL_SS_AMOUNT_V3_NFRAC_SHIFT) &
+								ATOM_PPLL_SS_AMOUNT_V3_NFRAC_MASK)));
+	} else
+		params.ucEnable = ATOM_DISABLE;
+
+	if (EXEC_BIOS_CMD_TABLE(EnableSpreadSpectrumOnPPLL, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+/*******************************************************************************
+ ********************************************************************************
+ **
+ **                  ADJUST DISPLAY PLL
+ **
+ ********************************************************************************
+ *******************************************************************************/
+
+static enum bp_result adjust_display_pll_v2(
+	struct bios_parser *bp,
+	struct bp_adjust_pixel_clock_parameters *bp_params);
+static enum bp_result adjust_display_pll_v3(
+	struct bios_parser *bp,
+	struct bp_adjust_pixel_clock_parameters *bp_params);
+
+static void init_adjust_display_pll(struct bios_parser *bp)
+{
+	switch (BIOS_CMD_TABLE_PARA_REVISION(AdjustDisplayPll)) {
+	case 2:
+		bp->cmd_tbl.adjust_display_pll = adjust_display_pll_v2;
+		break;
+	case 3:
+		bp->cmd_tbl.adjust_display_pll = adjust_display_pll_v3;
+		break;
+	default:
+		dm_output_to_console("Don't have adjust_display_pll for v%d\n",
+			 BIOS_CMD_TABLE_PARA_REVISION(AdjustDisplayPll));
+		bp->cmd_tbl.adjust_display_pll = NULL;
+		break;
+	}
+}
+
+static enum bp_result adjust_display_pll_v2(
+	struct bios_parser *bp,
+	struct bp_adjust_pixel_clock_parameters *bp_params)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	ADJUST_DISPLAY_PLL_PS_ALLOCATION params = { 0 };
+
+	/* We need to convert from KHz units into 10KHz units and then convert
+	 * output pixel clock back 10KHz-->KHz */
+	uint32_t pixel_clock_10KHz_in = bp_params->pixel_clock / 10;
+
+	params.usPixelClock = cpu_to_le16((uint16_t)(pixel_clock_10KHz_in));
+	params.ucTransmitterID =
+			bp->cmd_helper->encoder_id_to_atom(
+					dal_graphics_object_id_get_encoder_id(
+							bp_params->encoder_object_id));
+	params.ucEncodeMode =
+			(uint8_t)bp->cmd_helper->encoder_mode_bp_to_atom(
+					bp_params->signal_type, false);
+
+	if (EXEC_BIOS_CMD_TABLE(AdjustDisplayPll, params)) {
+		/* Convert output pixel clock back 10KHz-->KHz: multiply
+		 * original pixel clock in KHz by ratio
+		 * [output pxlClk/input pxlClk] */
+		uint64_t pixel_clk_10_khz_out =
+				(uint64_t)le16_to_cpu(params.usPixelClock);
+		uint64_t pixel_clk = (uint64_t)bp_params->pixel_clock;
+
+		if (pixel_clock_10KHz_in != 0) {
+			bp_params->adjusted_pixel_clock =
+					div_u64(pixel_clk * pixel_clk_10_khz_out,
+							pixel_clock_10KHz_in);
+		} else {
+			bp_params->adjusted_pixel_clock = 0;
+			BREAK_TO_DEBUGGER();
+		}
+
+		result = BP_RESULT_OK;
+	}
+
+	return result;
+}
+
+static enum bp_result adjust_display_pll_v3(
+	struct bios_parser *bp,
+	struct bp_adjust_pixel_clock_parameters *bp_params)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	ADJUST_DISPLAY_PLL_PS_ALLOCATION_V3 params;
+	uint32_t pixel_clk_10_kHz_in = bp_params->pixel_clock / 10;
+
+	memset(&params, 0, sizeof(params));
+
+	/* We need to convert from KHz units into 10KHz units and then convert
+	 * output pixel clock back 10KHz-->KHz */
+	params.sInput.usPixelClock = cpu_to_le16((uint16_t)pixel_clk_10_kHz_in);
+	params.sInput.ucTransmitterID =
+			bp->cmd_helper->encoder_id_to_atom(
+					dal_graphics_object_id_get_encoder_id(
+							bp_params->encoder_object_id));
+	params.sInput.ucEncodeMode =
+			(uint8_t)bp->cmd_helper->encoder_mode_bp_to_atom(
+					bp_params->signal_type, false);
+
+	if (bp_params->ss_enable == true)
+		params.sInput.ucDispPllConfig |= DISPPLL_CONFIG_SS_ENABLE;
+
+	if (bp_params->signal_type == SIGNAL_TYPE_DVI_DUAL_LINK)
+		params.sInput.ucDispPllConfig |= DISPPLL_CONFIG_DUAL_LINK;
+
+	if (EXEC_BIOS_CMD_TABLE(AdjustDisplayPll, params)) {
+		/* Convert output pixel clock back 10KHz-->KHz: multiply
+		 * original pixel clock in KHz by ratio
+		 * [output pxlClk/input pxlClk] */
+		uint64_t pixel_clk_10_khz_out =
+				(uint64_t)le32_to_cpu(params.sOutput.ulDispPllFreq);
+		uint64_t pixel_clk = (uint64_t)bp_params->pixel_clock;
+
+		if (pixel_clk_10_kHz_in != 0) {
+			bp_params->adjusted_pixel_clock =
+					div_u64(pixel_clk * pixel_clk_10_khz_out,
+							pixel_clk_10_kHz_in);
+		} else {
+			bp_params->adjusted_pixel_clock = 0;
+			BREAK_TO_DEBUGGER();
+		}
+
+		bp_params->reference_divider = params.sOutput.ucRefDiv;
+		bp_params->pixel_clock_post_divider = params.sOutput.ucPostDiv;
+
+		result = BP_RESULT_OK;
+	}
+
+	return result;
+}
+
+/*******************************************************************************
+ ********************************************************************************
+ **
+ **                  DAC ENCODER CONTROL
+ **
+ ********************************************************************************
+ *******************************************************************************/
+
+static enum bp_result dac1_encoder_control_v1(
+	struct bios_parser *bp,
+	bool enable,
+	uint32_t pixel_clock,
+	uint8_t dac_standard);
+static enum bp_result dac2_encoder_control_v1(
+	struct bios_parser *bp,
+	bool enable,
+	uint32_t pixel_clock,
+	uint8_t dac_standard);
+
+static void init_dac_encoder_control(struct bios_parser *bp)
+{
+	switch (BIOS_CMD_TABLE_PARA_REVISION(DAC1EncoderControl)) {
+	case 1:
+		bp->cmd_tbl.dac1_encoder_control = dac1_encoder_control_v1;
+		break;
+	default:
+		bp->cmd_tbl.dac1_encoder_control = NULL;
+		break;
+	}
+	switch (BIOS_CMD_TABLE_PARA_REVISION(DAC2EncoderControl)) {
+	case 1:
+		bp->cmd_tbl.dac2_encoder_control = dac2_encoder_control_v1;
+		break;
+	default:
+		bp->cmd_tbl.dac2_encoder_control = NULL;
+		break;
+	}
+}
+
+static void dac_encoder_control_prepare_params(
+	DAC_ENCODER_CONTROL_PS_ALLOCATION *params,
+	bool enable,
+	uint32_t pixel_clock,
+	uint8_t dac_standard)
+{
+	params->ucDacStandard = dac_standard;
+	if (enable)
+		params->ucAction = ATOM_ENABLE;
+	else
+		params->ucAction = ATOM_DISABLE;
+
+	/* We need to convert from KHz units into 10KHz units
+	 * it looks as if the TvControl do not care about pixel clock
+	 */
+	params->usPixelClock = cpu_to_le16((uint16_t)(pixel_clock / 10));
+}
+
+static enum bp_result dac1_encoder_control_v1(
+	struct bios_parser *bp,
+	bool enable,
+	uint32_t pixel_clock,
+	uint8_t dac_standard)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	DAC_ENCODER_CONTROL_PS_ALLOCATION params;
+
+	dac_encoder_control_prepare_params(
+		&params,
+		enable,
+		pixel_clock,
+		dac_standard);
+
+	if (EXEC_BIOS_CMD_TABLE(DAC1EncoderControl, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+static enum bp_result dac2_encoder_control_v1(
+	struct bios_parser *bp,
+	bool enable,
+	uint32_t pixel_clock,
+	uint8_t dac_standard)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	DAC_ENCODER_CONTROL_PS_ALLOCATION params;
+
+	dac_encoder_control_prepare_params(
+		&params,
+		enable,
+		pixel_clock,
+		dac_standard);
+
+	if (EXEC_BIOS_CMD_TABLE(DAC2EncoderControl, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+/*******************************************************************************
+ ********************************************************************************
+ **
+ **                  DAC OUTPUT CONTROL
+ **
+ ********************************************************************************
+ *******************************************************************************/
+static enum bp_result dac1_output_control_v1(
+	struct bios_parser *bp,
+	bool enable);
+static enum bp_result dac2_output_control_v1(
+	struct bios_parser *bp,
+	bool enable);
+
+static void init_dac_output_control(struct bios_parser *bp)
+{
+	switch (BIOS_CMD_TABLE_PARA_REVISION(DAC1OutputControl)) {
+	case 1:
+		bp->cmd_tbl.dac1_output_control = dac1_output_control_v1;
+		break;
+	default:
+		bp->cmd_tbl.dac1_output_control = NULL;
+		break;
+	}
+	switch (BIOS_CMD_TABLE_PARA_REVISION(DAC2OutputControl)) {
+	case 1:
+		bp->cmd_tbl.dac2_output_control = dac2_output_control_v1;
+		break;
+	default:
+		bp->cmd_tbl.dac2_output_control = NULL;
+		break;
+	}
+}
+
+static enum bp_result dac1_output_control_v1(
+	struct bios_parser *bp, bool enable)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	DISPLAY_DEVICE_OUTPUT_CONTROL_PS_ALLOCATION params;
+
+	if (enable)
+		params.ucAction = ATOM_ENABLE;
+	else
+		params.ucAction = ATOM_DISABLE;
+
+	if (EXEC_BIOS_CMD_TABLE(DAC1OutputControl, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+static enum bp_result dac2_output_control_v1(
+	struct bios_parser *bp, bool enable)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	DISPLAY_DEVICE_OUTPUT_CONTROL_PS_ALLOCATION params;
+
+	if (enable)
+		params.ucAction = ATOM_ENABLE;
+	else
+		params.ucAction = ATOM_DISABLE;
+
+	if (EXEC_BIOS_CMD_TABLE(DAC2OutputControl, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+/*******************************************************************************
+ ********************************************************************************
+ **
+ **                  SET CRTC TIMING
+ **
+ ********************************************************************************
+ *******************************************************************************/
+
+static enum bp_result set_crtc_using_dtd_timing_v3(
+	struct bios_parser *bp,
+	struct bp_hw_crtc_timing_parameters *bp_params);
+static enum bp_result set_crtc_timing_v1(
+	struct bios_parser *bp,
+	struct bp_hw_crtc_timing_parameters *bp_params);
+
+static void init_set_crtc_timing(struct bios_parser *bp)
+{
+	uint32_t dtd_version =
+			BIOS_CMD_TABLE_PARA_REVISION(SetCRTC_UsingDTDTiming);
+	if (dtd_version > 2)
+		switch (dtd_version) {
+		case 3:
+			bp->cmd_tbl.set_crtc_timing =
+					set_crtc_using_dtd_timing_v3;
+			break;
+		default:
+			dm_output_to_console("Don't have set_crtc_timing for dtd v%d\n",
+				 dtd_version);
+			bp->cmd_tbl.set_crtc_timing = NULL;
+			break;
+		}
+	else
+		switch (BIOS_CMD_TABLE_PARA_REVISION(SetCRTC_Timing)) {
+		case 1:
+			bp->cmd_tbl.set_crtc_timing = set_crtc_timing_v1;
+			break;
+		default:
+			dm_output_to_console("Don't have set_crtc_timing for v%d\n",
+				 BIOS_CMD_TABLE_PARA_REVISION(SetCRTC_Timing));
+			bp->cmd_tbl.set_crtc_timing = NULL;
+			break;
+		}
+}
+
+static enum bp_result set_crtc_timing_v1(
+	struct bios_parser *bp,
+	struct bp_hw_crtc_timing_parameters *bp_params)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	SET_CRTC_TIMING_PARAMETERS_PS_ALLOCATION params = {0};
+	uint8_t atom_controller_id;
+
+	if (bp->cmd_helper->controller_id_to_atom(
+			bp_params->controller_id, &atom_controller_id))
+		params.ucCRTC = atom_controller_id;
+
+	params.usH_Total = cpu_to_le16((uint16_t)(bp_params->h_total));
+	params.usH_Disp = cpu_to_le16((uint16_t)(bp_params->h_addressable));
+	params.usH_SyncStart = cpu_to_le16((uint16_t)(bp_params->h_sync_start));
+	params.usH_SyncWidth = cpu_to_le16((uint16_t)(bp_params->h_sync_width));
+	params.usV_Total = cpu_to_le16((uint16_t)(bp_params->v_total));
+	params.usV_Disp = cpu_to_le16((uint16_t)(bp_params->v_addressable));
+	params.usV_SyncStart =
+			cpu_to_le16((uint16_t)(bp_params->v_sync_start));
+	params.usV_SyncWidth =
+			cpu_to_le16((uint16_t)(bp_params->v_sync_width));
+
+	/* VBIOS does not expect any value except zero into this call, for
+	 * underscan use another entry ProgramOverscan call but when mode
+	 * 1776x1000 with the overscan 72x44 .e.i. 1920x1080 @30 DAL2 is ok,
+	 * but when same ,but 60 Hz there is corruption
+	 * DAL1 does not allow the mode 1776x1000@60
+	 */
+	params.ucOverscanRight = (uint8_t)bp_params->h_overscan_right;
+	params.ucOverscanLeft = (uint8_t)bp_params->h_overscan_left;
+	params.ucOverscanBottom = (uint8_t)bp_params->v_overscan_bottom;
+	params.ucOverscanTop = (uint8_t)bp_params->v_overscan_top;
+
+	if (0 == bp_params->flags.HSYNC_POSITIVE_POLARITY)
+		params.susModeMiscInfo.usAccess =
+				cpu_to_le16(le16_to_cpu(params.susModeMiscInfo.usAccess) | ATOM_HSYNC_POLARITY);
+
+	if (0 == bp_params->flags.VSYNC_POSITIVE_POLARITY)
+		params.susModeMiscInfo.usAccess =
+				cpu_to_le16(le16_to_cpu(params.susModeMiscInfo.usAccess) | ATOM_VSYNC_POLARITY);
+
+	if (bp_params->flags.INTERLACE) {
+		params.susModeMiscInfo.usAccess =
+				cpu_to_le16(le16_to_cpu(params.susModeMiscInfo.usAccess) | ATOM_INTERLACE);
+
+		/* original DAL code has this condition to apply tis for
+		 * non-TV/CV only due to complex MV testing for possible
+		 * impact
+		 * if (pACParameters->signal != SignalType_YPbPr &&
+		 *  pACParameters->signal != SignalType_Composite &&
+		 *  pACParameters->signal != SignalType_SVideo)
+		 */
+		/* HW will deduct 0.5 line from 2nd feild.
+		 * i.e. for 1080i, it is 2 lines for 1st field, 2.5
+		 * lines for the 2nd feild. we need input as 5 instead
+		 * of 4, but it is 4 either from Edid data
+		 * (spec CEA 861) or CEA timing table.
+		 */
+		params.usV_SyncStart =
+				cpu_to_le16((uint16_t)(bp_params->v_sync_start + 1));
+	}
+
+	if (bp_params->flags.HORZ_COUNT_BY_TWO)
+		params.susModeMiscInfo.usAccess =
+				cpu_to_le16(le16_to_cpu(params.susModeMiscInfo.usAccess) | ATOM_DOUBLE_CLOCK_MODE);
+
+	if (EXEC_BIOS_CMD_TABLE(SetCRTC_Timing, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+static enum bp_result set_crtc_using_dtd_timing_v3(
+	struct bios_parser *bp,
+	struct bp_hw_crtc_timing_parameters *bp_params)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	SET_CRTC_USING_DTD_TIMING_PARAMETERS params = {0};
+	uint8_t atom_controller_id;
+
+	if (bp->cmd_helper->controller_id_to_atom(
+			bp_params->controller_id, &atom_controller_id))
+		params.ucCRTC = atom_controller_id;
+
+	/* bios usH_Size wants h addressable size */
+	params.usH_Size = cpu_to_le16((uint16_t)bp_params->h_addressable);
+	/* bios usH_Blanking_Time wants borders included in blanking */
+	params.usH_Blanking_Time =
+			cpu_to_le16((uint16_t)(bp_params->h_total - bp_params->h_addressable));
+	/* bios usV_Size wants v addressable size */
+	params.usV_Size = cpu_to_le16((uint16_t)bp_params->v_addressable);
+	/* bios usV_Blanking_Time wants borders included in blanking */
+	params.usV_Blanking_Time =
+			cpu_to_le16((uint16_t)(bp_params->v_total - bp_params->v_addressable));
+	/* bios usHSyncOffset is the offset from the end of h addressable,
+	 * our horizontalSyncStart is the offset from the beginning
+	 * of h addressable */
+	params.usH_SyncOffset =
+			cpu_to_le16((uint16_t)(bp_params->h_sync_start - bp_params->h_addressable));
+	params.usH_SyncWidth = cpu_to_le16((uint16_t)bp_params->h_sync_width);
+	/* bios usHSyncOffset is the offset from the end of v addressable,
+	 * our verticalSyncStart is the offset from the beginning of
+	 * v addressable */
+	params.usV_SyncOffset =
+			cpu_to_le16((uint16_t)(bp_params->v_sync_start - bp_params->v_addressable));
+	params.usV_SyncWidth = cpu_to_le16((uint16_t)bp_params->v_sync_width);
+
+	/* we assume that overscan from original timing does not get bigger
+	 * than 255
+	 * we will program all the borders in the Set CRTC Overscan call below
+	 */
+
+	if (0 == bp_params->flags.HSYNC_POSITIVE_POLARITY)
+		params.susModeMiscInfo.usAccess =
+				cpu_to_le16(le16_to_cpu(params.susModeMiscInfo.usAccess) | ATOM_HSYNC_POLARITY);
+
+	if (0 == bp_params->flags.VSYNC_POSITIVE_POLARITY)
+		params.susModeMiscInfo.usAccess =
+				cpu_to_le16(le16_to_cpu(params.susModeMiscInfo.usAccess) | ATOM_VSYNC_POLARITY);
+
+	if (bp_params->flags.INTERLACE)	{
+		params.susModeMiscInfo.usAccess =
+				cpu_to_le16(le16_to_cpu(params.susModeMiscInfo.usAccess) | ATOM_INTERLACE);
+
+		/* original DAL code has this condition to apply this
+		 * for non-TV/CV only
+		 * due to complex MV testing for possible impact
+		 * if ( pACParameters->signal != SignalType_YPbPr &&
+		 *  pACParameters->signal != SignalType_Composite &&
+		 *  pACParameters->signal != SignalType_SVideo)
+		 */
+		{
+			/* HW will deduct 0.5 line from 2nd feild.
+			 * i.e. for 1080i, it is 2 lines for 1st field,
+			 * 2.5 lines for the 2nd feild. we need input as 5
+			 * instead of 4.
+			 * but it is 4 either from Edid data (spec CEA 861)
+			 * or CEA timing table.
+			 */
+			le16_add_cpu(&params.usV_SyncOffset, 1);
+		}
+	}
+
+	if (bp_params->flags.HORZ_COUNT_BY_TWO)
+		params.susModeMiscInfo.usAccess =
+				cpu_to_le16(le16_to_cpu(params.susModeMiscInfo.usAccess) | ATOM_DOUBLE_CLOCK_MODE);
+
+	if (EXEC_BIOS_CMD_TABLE(SetCRTC_UsingDTDTiming, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+/*******************************************************************************
+ ********************************************************************************
+ **
+ **                  ENABLE CRTC
+ **
+ ********************************************************************************
+ *******************************************************************************/
+
+static enum bp_result enable_crtc_v1(
+	struct bios_parser *bp,
+	enum controller_id controller_id,
+	bool enable);
+
+static void init_enable_crtc(struct bios_parser *bp)
+{
+	switch (BIOS_CMD_TABLE_PARA_REVISION(EnableCRTC)) {
+	case 1:
+		bp->cmd_tbl.enable_crtc = enable_crtc_v1;
+		break;
+	default:
+		dm_output_to_console("Don't have enable_crtc for v%d\n",
+			 BIOS_CMD_TABLE_PARA_REVISION(EnableCRTC));
+		bp->cmd_tbl.enable_crtc = NULL;
+		break;
+	}
+}
+
+static enum bp_result enable_crtc_v1(
+	struct bios_parser *bp,
+	enum controller_id controller_id,
+	bool enable)
+{
+	bool result = BP_RESULT_FAILURE;
+	ENABLE_CRTC_PARAMETERS params = {0};
+	uint8_t id;
+
+	if (bp->cmd_helper->controller_id_to_atom(controller_id, &id))
+		params.ucCRTC = id;
+	else
+		return BP_RESULT_BADINPUT;
+
+	if (enable)
+		params.ucEnable = ATOM_ENABLE;
+	else
+		params.ucEnable = ATOM_DISABLE;
+
+	if (EXEC_BIOS_CMD_TABLE(EnableCRTC, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+/*******************************************************************************
+ ********************************************************************************
+ **
+ **                  ENABLE CRTC MEM REQ
+ **
+ ********************************************************************************
+ *******************************************************************************/
+
+static enum bp_result enable_crtc_mem_req_v1(
+	struct bios_parser *bp,
+	enum controller_id controller_id,
+	bool enable);
+
+static void init_enable_crtc_mem_req(struct bios_parser *bp)
+{
+	switch (BIOS_CMD_TABLE_PARA_REVISION(EnableCRTCMemReq)) {
+	case 1:
+		bp->cmd_tbl.enable_crtc_mem_req = enable_crtc_mem_req_v1;
+		break;
+	default:
+		bp->cmd_tbl.enable_crtc_mem_req = NULL;
+		break;
+	}
+}
+
+static enum bp_result enable_crtc_mem_req_v1(
+	struct bios_parser *bp,
+	enum controller_id controller_id,
+	bool enable)
+{
+	bool result = BP_RESULT_BADINPUT;
+	ENABLE_CRTC_PARAMETERS params = {0};
+	uint8_t id;
+
+	if (bp->cmd_helper->controller_id_to_atom(controller_id, &id)) {
+		params.ucCRTC = id;
+
+		if (enable)
+			params.ucEnable = ATOM_ENABLE;
+		else
+			params.ucEnable = ATOM_DISABLE;
+
+		if (EXEC_BIOS_CMD_TABLE(EnableCRTCMemReq, params))
+			result = BP_RESULT_OK;
+		else
+			result = BP_RESULT_FAILURE;
+	}
+
+	return result;
+}
+
+/*******************************************************************************
+ ********************************************************************************
+ **
+ **                  DISPLAY PLL
+ **
+ ********************************************************************************
+ *******************************************************************************/
+
+static enum bp_result program_clock_v5(
+	struct bios_parser *bp,
+	struct bp_pixel_clock_parameters *bp_params);
+static enum bp_result program_clock_v6(
+	struct bios_parser *bp,
+	struct bp_pixel_clock_parameters *bp_params);
+
+static void init_program_clock(struct bios_parser *bp)
+{
+	switch (BIOS_CMD_TABLE_PARA_REVISION(SetPixelClock)) {
+	case 5:
+		bp->cmd_tbl.program_clock = program_clock_v5;
+		break;
+	case 6:
+		bp->cmd_tbl.program_clock = program_clock_v6;
+		break;
+	default:
+		dm_output_to_console("Don't have program_clock for v%d\n",
+			 BIOS_CMD_TABLE_PARA_REVISION(SetPixelClock));
+		bp->cmd_tbl.program_clock = NULL;
+		break;
+	}
+}
+
+static enum bp_result program_clock_v5(
+	struct bios_parser *bp,
+	struct bp_pixel_clock_parameters *bp_params)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+
+	SET_PIXEL_CLOCK_PS_ALLOCATION_V5 params;
+	uint32_t atom_pll_id;
+
+	memset(&params, 0, sizeof(params));
+	if (!bp->cmd_helper->clock_source_id_to_atom(
+			bp_params->pll_id, &atom_pll_id)) {
+		BREAK_TO_DEBUGGER(); /* Invalid Input!! */
+		return BP_RESULT_BADINPUT;
+	}
+
+	/* We need to convert from KHz units into 10KHz units */
+	params.sPCLKInput.ucPpll = (uint8_t) atom_pll_id;
+	params.sPCLKInput.usPixelClock =
+			cpu_to_le16((uint16_t) (bp_params->target_pixel_clock_100hz / 100));
+	params.sPCLKInput.ucCRTC = (uint8_t) ATOM_CRTC_INVALID;
+
+	if (bp_params->flags.SET_EXTERNAL_REF_DIV_SRC)
+		params.sPCLKInput.ucMiscInfo |= PIXEL_CLOCK_MISC_REF_DIV_SRC;
+
+	if (EXEC_BIOS_CMD_TABLE(SetPixelClock, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+static enum bp_result program_clock_v6(
+	struct bios_parser *bp,
+	struct bp_pixel_clock_parameters *bp_params)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+
+	SET_PIXEL_CLOCK_PS_ALLOCATION_V6 params;
+	uint32_t atom_pll_id;
+
+	memset(&params, 0, sizeof(params));
+
+	if (!bp->cmd_helper->clock_source_id_to_atom(
+			bp_params->pll_id, &atom_pll_id)) {
+		BREAK_TO_DEBUGGER(); /*Invalid Input!!*/
+		return BP_RESULT_BADINPUT;
+	}
+
+	/* We need to convert from KHz units into 10KHz units */
+	params.sPCLKInput.ucPpll = (uint8_t)atom_pll_id;
+	params.sPCLKInput.ulDispEngClkFreq =
+			cpu_to_le32(bp_params->target_pixel_clock_100hz / 100);
+
+	if (bp_params->flags.SET_EXTERNAL_REF_DIV_SRC)
+		params.sPCLKInput.ucMiscInfo |= PIXEL_CLOCK_MISC_REF_DIV_SRC;
+
+	if (bp_params->flags.SET_DISPCLK_DFS_BYPASS)
+		params.sPCLKInput.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_DPREFCLK_BYPASS;
+
+	if (EXEC_BIOS_CMD_TABLE(SetPixelClock, params)) {
+		/* True display clock is returned by VBIOS if DFS bypass
+		 * is enabled. */
+		bp_params->dfs_bypass_display_clock =
+				(uint32_t)(le32_to_cpu(params.sPCLKInput.ulDispEngClkFreq) * 10);
+		result = BP_RESULT_OK;
+	}
+
+	return result;
+}
+
+/*******************************************************************************
+ ********************************************************************************
+ **
+ **                  EXTERNAL ENCODER CONTROL
+ **
+ ********************************************************************************
+ *******************************************************************************/
+
+static enum bp_result external_encoder_control_v3(
+	struct bios_parser *bp,
+	struct bp_external_encoder_control *cntl);
+
+static void init_external_encoder_control(
+	struct bios_parser *bp)
+{
+	switch (BIOS_CMD_TABLE_PARA_REVISION(ExternalEncoderControl)) {
+	case 3:
+		bp->cmd_tbl.external_encoder_control =
+				external_encoder_control_v3;
+		break;
+	default:
+		bp->cmd_tbl.external_encoder_control = NULL;
+		break;
+	}
+}
+
+static enum bp_result external_encoder_control_v3(
+	struct bios_parser *bp,
+	struct bp_external_encoder_control *cntl)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+
+	/* we need use _PS_Alloc struct */
+	EXTERNAL_ENCODER_CONTROL_PS_ALLOCATION_V3 params;
+	EXTERNAL_ENCODER_CONTROL_PARAMETERS_V3 *cntl_params;
+	struct graphics_object_id encoder;
+	bool is_input_signal_dp = false;
+
+	memset(&params, 0, sizeof(params));
+
+	cntl_params = &params.sExtEncoder;
+
+	encoder = cntl->encoder_id;
+
+	/* check if encoder supports external encoder control table */
+	switch (dal_graphics_object_id_get_encoder_id(encoder)) {
+	case ENCODER_ID_EXTERNAL_NUTMEG:
+	case ENCODER_ID_EXTERNAL_TRAVIS:
+		is_input_signal_dp = true;
+		break;
+
+	default:
+		BREAK_TO_DEBUGGER();
+		return BP_RESULT_BADINPUT;
+	}
+
+	/* Fill information based on the action
+	 *
+	 * Bit[6:4]: indicate external encoder, applied to all functions.
+	 * =0: external encoder1, mapped to external encoder enum id1
+	 * =1: external encoder2, mapped to external encoder enum id2
+	 *
+	 * enum ObjectEnumId
+	 * {
+	 *  EnumId_Unknown = 0,
+	 *  EnumId_1,
+	 *  EnumId_2,
+	 * };
+	 */
+	cntl_params->ucConfig = (uint8_t)((encoder.enum_id - 1) << 4);
+
+	switch (cntl->action) {
+	case EXTERNAL_ENCODER_CONTROL_INIT:
+		/* output display connector type. Only valid in encoder
+		 * initialization */
+		cntl_params->usConnectorId =
+				cpu_to_le16((uint16_t)cntl->connector_obj_id.id);
+		break;
+	case EXTERNAL_ENCODER_CONTROL_SETUP:
+		/* EXTERNAL_ENCODER_CONTROL_PARAMETERS_V3 pixel clock unit in
+		 * 10KHz
+		 * output display device pixel clock frequency in unit of 10KHz.
+		 * Only valid in setup and enableoutput
+		 */
+		cntl_params->usPixelClock =
+				cpu_to_le16((uint16_t)(cntl->pixel_clock / 10));
+		/* Indicate display output signal type drive by external
+		 * encoder, only valid in setup and enableoutput */
+		cntl_params->ucEncoderMode =
+				(uint8_t)bp->cmd_helper->encoder_mode_bp_to_atom(
+						cntl->signal, false);
+
+		if (is_input_signal_dp) {
+			/* Bit[0]: indicate link rate, =1: 2.7Ghz, =0: 1.62Ghz,
+			 * only valid in encoder setup with DP mode. */
+			if (LINK_RATE_HIGH == cntl->link_rate)
+				cntl_params->ucConfig |= 1;
+			/* output color depth Indicate encoder data bpc format
+			 * in DP mode, only valid in encoder setup in DP mode.
+			 */
+			cntl_params->ucBitPerColor =
+					(uint8_t)(cntl->color_depth);
+		}
+		/* Indicate how many lanes used by external encoder, only valid
+		 * in encoder setup and enableoutput. */
+		cntl_params->ucLaneNum = (uint8_t)(cntl->lanes_number);
+		break;
+	case EXTERNAL_ENCODER_CONTROL_ENABLE:
+		cntl_params->usPixelClock =
+				cpu_to_le16((uint16_t)(cntl->pixel_clock / 10));
+		cntl_params->ucEncoderMode =
+				(uint8_t)bp->cmd_helper->encoder_mode_bp_to_atom(
+						cntl->signal, false);
+		cntl_params->ucLaneNum = (uint8_t)cntl->lanes_number;
+		break;
+	default:
+		break;
+	}
+
+	cntl_params->ucAction = (uint8_t)cntl->action;
+
+	if (EXEC_BIOS_CMD_TABLE(ExternalEncoderControl, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+/*******************************************************************************
+ ********************************************************************************
+ **
+ **                  ENABLE DISPLAY POWER GATING
+ **
+ ********************************************************************************
+ *******************************************************************************/
+
+static enum bp_result enable_disp_power_gating_v2_1(
+	struct bios_parser *bp,
+	enum controller_id crtc_id,
+	enum bp_pipe_control_action action);
+
+static void init_enable_disp_power_gating(
+	struct bios_parser *bp)
+{
+	switch (BIOS_CMD_TABLE_PARA_REVISION(EnableDispPowerGating)) {
+	case 1:
+		bp->cmd_tbl.enable_disp_power_gating =
+				enable_disp_power_gating_v2_1;
+		break;
+	default:
+		dm_output_to_console("Don't enable_disp_power_gating enable_crtc for v%d\n",
+			 BIOS_CMD_TABLE_PARA_REVISION(EnableDispPowerGating));
+		bp->cmd_tbl.enable_disp_power_gating = NULL;
+		break;
+	}
+}
+
+static enum bp_result enable_disp_power_gating_v2_1(
+	struct bios_parser *bp,
+	enum controller_id crtc_id,
+	enum bp_pipe_control_action action)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+
+	ENABLE_DISP_POWER_GATING_PS_ALLOCATION params = {0};
+	uint8_t atom_crtc_id;
+
+	if (bp->cmd_helper->controller_id_to_atom(crtc_id, &atom_crtc_id))
+		params.ucDispPipeId = atom_crtc_id;
+	else
+		return BP_RESULT_BADINPUT;
+
+	params.ucEnable =
+			bp->cmd_helper->disp_power_gating_action_to_atom(action);
+
+	if (EXEC_BIOS_CMD_TABLE(EnableDispPowerGating, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+/*******************************************************************************
+ ********************************************************************************
+ **
+ **                  SET DCE CLOCK
+ **
+ ********************************************************************************
+ *******************************************************************************/
+static enum bp_result set_dce_clock_v2_1(
+	struct bios_parser *bp,
+	struct bp_set_dce_clock_parameters *bp_params);
+
+static void init_set_dce_clock(struct bios_parser *bp)
+{
+	switch (BIOS_CMD_TABLE_PARA_REVISION(SetDCEClock)) {
+	case 1:
+		bp->cmd_tbl.set_dce_clock = set_dce_clock_v2_1;
+		break;
+	default:
+		dm_output_to_console("Don't have set_dce_clock for v%d\n",
+			 BIOS_CMD_TABLE_PARA_REVISION(SetDCEClock));
+		bp->cmd_tbl.set_dce_clock = NULL;
+		break;
+	}
+}
+
+static enum bp_result set_dce_clock_v2_1(
+	struct bios_parser *bp,
+	struct bp_set_dce_clock_parameters *bp_params)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+
+	SET_DCE_CLOCK_PS_ALLOCATION_V2_1 params;
+	uint32_t atom_pll_id;
+	uint32_t atom_clock_type;
+	const struct command_table_helper *cmd = bp->cmd_helper;
+
+	memset(&params, 0, sizeof(params));
+
+	if (!cmd->clock_source_id_to_atom(bp_params->pll_id, &atom_pll_id) ||
+			!cmd->dc_clock_type_to_atom(bp_params->clock_type, &atom_clock_type))
+		return BP_RESULT_BADINPUT;
+
+	params.asParam.ucDCEClkSrc  = atom_pll_id;
+	params.asParam.ucDCEClkType = atom_clock_type;
+
+	if (bp_params->clock_type == DCECLOCK_TYPE_DPREFCLK) {
+		if (bp_params->flags.USE_GENLOCK_AS_SOURCE_FOR_DPREFCLK)
+			params.asParam.ucDCEClkFlag |= DCE_CLOCK_FLAG_PLL_REFCLK_SRC_GENLK;
+
+		if (bp_params->flags.USE_PCIE_AS_SOURCE_FOR_DPREFCLK)
+			params.asParam.ucDCEClkFlag |= DCE_CLOCK_FLAG_PLL_REFCLK_SRC_PCIE;
+
+		if (bp_params->flags.USE_XTALIN_AS_SOURCE_FOR_DPREFCLK)
+			params.asParam.ucDCEClkFlag |= DCE_CLOCK_FLAG_PLL_REFCLK_SRC_XTALIN;
+
+		if (bp_params->flags.USE_GENERICA_AS_SOURCE_FOR_DPREFCLK)
+			params.asParam.ucDCEClkFlag |= DCE_CLOCK_FLAG_PLL_REFCLK_SRC_GENERICA;
+	}
+	else
+		/* only program clock frequency if display clock is used; VBIOS will program DPREFCLK */
+		/* We need to convert from KHz units into 10KHz units */
+		params.asParam.ulDCEClkFreq = cpu_to_le32(bp_params->target_clock_frequency / 10);
+
+	if (EXEC_BIOS_CMD_TABLE(SetDCEClock, params)) {
+		/* Convert from 10KHz units back to KHz */
+		bp_params->target_clock_frequency = le32_to_cpu(params.asParam.ulDCEClkFreq) * 10;
+		result = BP_RESULT_OK;
+	}
+
+	return result;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/bios/command_table.h b/drivers/gpu/drm/amd/display/dc/bios/command_table.h
new file mode 100644
index 000000000000..ad533775e724
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/command_table.h
@@ -0,0 +1,99 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_COMMAND_TABLE_H__
+#define __DAL_COMMAND_TABLE_H__
+
+struct bios_parser;
+struct bp_encoder_control;
+
+struct cmd_tbl {
+	enum bp_result (*dig_encoder_control)(
+		struct bios_parser *bp,
+		struct bp_encoder_control *control);
+	enum bp_result (*encoder_control_dig1)(
+		struct bios_parser *bp,
+		struct bp_encoder_control *control);
+	enum bp_result (*encoder_control_dig2)(
+		struct bios_parser *bp,
+		struct bp_encoder_control *control);
+	enum bp_result (*transmitter_control)(
+		struct bios_parser *bp,
+		struct bp_transmitter_control *control);
+	enum bp_result (*set_pixel_clock)(
+		struct bios_parser *bp,
+		struct bp_pixel_clock_parameters *bp_params);
+	enum bp_result (*enable_spread_spectrum_on_ppll)(
+		struct bios_parser *bp,
+		struct bp_spread_spectrum_parameters *bp_params,
+		bool enable);
+	enum bp_result (*adjust_display_pll)(
+		struct bios_parser *bp,
+		struct bp_adjust_pixel_clock_parameters *bp_params);
+	enum bp_result (*dac1_encoder_control)(
+		struct bios_parser *bp,
+		bool enable,
+		uint32_t pixel_clock,
+		uint8_t dac_standard);
+	enum bp_result (*dac2_encoder_control)(
+		struct bios_parser *bp,
+		bool enable,
+		uint32_t pixel_clock,
+		uint8_t dac_standard);
+	enum bp_result (*dac1_output_control)(
+		struct bios_parser *bp,
+		bool enable);
+	enum bp_result (*dac2_output_control)(
+		struct bios_parser *bp,
+		bool enable);
+	enum bp_result (*set_crtc_timing)(
+		struct bios_parser *bp,
+		struct bp_hw_crtc_timing_parameters *bp_params);
+	enum bp_result (*enable_crtc)(
+		struct bios_parser *bp,
+		enum controller_id controller_id,
+		bool enable);
+	enum bp_result (*enable_crtc_mem_req)(
+		struct bios_parser *bp,
+		enum controller_id controller_id,
+		bool enable);
+	enum bp_result (*program_clock)(
+		struct bios_parser *bp,
+		struct bp_pixel_clock_parameters *bp_params);
+	enum bp_result (*external_encoder_control)(
+			struct bios_parser *bp,
+			struct bp_external_encoder_control *cntl);
+	enum bp_result (*enable_disp_power_gating)(
+		struct bios_parser *bp,
+		enum controller_id crtc_id,
+		enum bp_pipe_control_action action);
+	enum bp_result (*set_dce_clock)(
+		struct bios_parser *bp,
+		struct bp_set_dce_clock_parameters *bp_params);
+};
+
+void dal_bios_parser_init_cmd_tbl(struct bios_parser *bp);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/bios/command_table2.c b/drivers/gpu/drm/amd/display/dc/bios/command_table2.c
new file mode 100644
index 000000000000..f2b1720a6a66
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/command_table2.c
@@ -0,0 +1,1087 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+
+#include "ObjectID.h"
+
+#include "atomfirmware.h"
+#include "atom.h"
+#include "include/bios_parser_interface.h"
+
+#include "command_table2.h"
+#include "command_table_helper2.h"
+#include "bios_parser_helper.h"
+#include "bios_parser_types_internal2.h"
+#include "amdgpu.h"
+
+#include "dc_dmub_srv.h"
+#include "dc.h"
+
+#define DC_LOGGER \
+	bp->base.ctx->logger
+
+#define GET_INDEX_INTO_MASTER_TABLE(MasterOrData, FieldName)\
+	(offsetof(struct atom_master_list_of_##MasterOrData##_functions_v2_1, FieldName) / sizeof(uint16_t))
+
+#define EXEC_BIOS_CMD_TABLE(fname, params)\
+	(amdgpu_atom_execute_table(((struct amdgpu_device *)bp->base.ctx->driver_context)->mode_info.atom_context, \
+		GET_INDEX_INTO_MASTER_TABLE(command, fname), \
+		(uint32_t *)&params, sizeof(params)) == 0)
+
+#define BIOS_CMD_TABLE_REVISION(fname, frev, crev)\
+	amdgpu_atom_parse_cmd_header(((struct amdgpu_device *)bp->base.ctx->driver_context)->mode_info.atom_context, \
+		GET_INDEX_INTO_MASTER_TABLE(command, fname), &frev, &crev)
+
+#define BIOS_CMD_TABLE_PARA_REVISION(fname)\
+	bios_cmd_table_para_revision(bp->base.ctx->driver_context, \
+			GET_INDEX_INTO_MASTER_TABLE(command, fname))
+
+
+
+static uint32_t bios_cmd_table_para_revision(void *dev,
+					     uint32_t index)
+{
+	struct amdgpu_device *adev = dev;
+	uint8_t frev, crev;
+
+	if (amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context,
+					index,
+					&frev, &crev))
+		return crev;
+	else
+		return 0;
+}
+
+/******************************************************************************
+ ******************************************************************************
+ **
+ **                  D I G E N C O D E R C O N T R O L
+ **
+ ******************************************************************************
+ *****************************************************************************/
+
+static enum bp_result encoder_control_digx_v1_5(
+	struct bios_parser *bp,
+	struct bp_encoder_control *cntl);
+
+static enum bp_result encoder_control_fallback(
+	struct bios_parser *bp,
+	struct bp_encoder_control *cntl);
+
+static void init_dig_encoder_control(struct bios_parser *bp)
+{
+	uint32_t version =
+		BIOS_CMD_TABLE_PARA_REVISION(digxencodercontrol);
+
+	switch (version) {
+	case 5:
+		bp->cmd_tbl.dig_encoder_control = encoder_control_digx_v1_5;
+		break;
+	default:
+		bp->cmd_tbl.dig_encoder_control = encoder_control_fallback;
+		break;
+	}
+}
+
+static void encoder_control_dmcub(
+		struct dc_dmub_srv *dmcub,
+		struct dig_encoder_stream_setup_parameters_v1_5 *dig)
+{
+	union dmub_rb_cmd cmd;
+
+	memset(&cmd, 0, sizeof(cmd));
+
+	cmd.digx_encoder_control.header.type = DMUB_CMD__VBIOS;
+	cmd.digx_encoder_control.header.sub_type =
+		DMUB_CMD__VBIOS_DIGX_ENCODER_CONTROL;
+	cmd.digx_encoder_control.header.payload_bytes =
+		sizeof(cmd.digx_encoder_control) -
+		sizeof(cmd.digx_encoder_control.header);
+	cmd.digx_encoder_control.encoder_control.dig.stream_param = *dig;
+
+	dc_wake_and_execute_dmub_cmd(dmcub->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+static enum bp_result encoder_control_digx_v1_5(
+	struct bios_parser *bp,
+	struct bp_encoder_control *cntl)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	struct dig_encoder_stream_setup_parameters_v1_5 params = {0};
+
+	params.digid = (uint8_t)(cntl->engine_id);
+	params.action = bp->cmd_helper->encoder_action_to_atom(cntl->action);
+
+	params.pclk_10khz = cntl->pixel_clock / 10;
+	params.digmode =
+			(uint8_t)(bp->cmd_helper->encoder_mode_bp_to_atom(
+					cntl->signal,
+					cntl->enable_dp_audio));
+	params.lanenum = (uint8_t)(cntl->lanes_number);
+
+	switch (cntl->color_depth) {
+	case COLOR_DEPTH_888:
+		params.bitpercolor = PANEL_8BIT_PER_COLOR;
+		break;
+	case COLOR_DEPTH_101010:
+		params.bitpercolor = PANEL_10BIT_PER_COLOR;
+		break;
+	case COLOR_DEPTH_121212:
+		params.bitpercolor = PANEL_12BIT_PER_COLOR;
+		break;
+	case COLOR_DEPTH_161616:
+		params.bitpercolor = PANEL_16BIT_PER_COLOR;
+		break;
+	default:
+		break;
+	}
+
+	if (cntl->signal == SIGNAL_TYPE_HDMI_TYPE_A)
+		switch (cntl->color_depth) {
+		case COLOR_DEPTH_101010:
+			params.pclk_10khz =
+				(params.pclk_10khz * 30) / 24;
+			break;
+		case COLOR_DEPTH_121212:
+			params.pclk_10khz =
+				(params.pclk_10khz * 36) / 24;
+			break;
+		case COLOR_DEPTH_161616:
+			params.pclk_10khz =
+				(params.pclk_10khz * 48) / 24;
+			break;
+		default:
+			break;
+		}
+
+	if (bp->base.ctx->dc->ctx->dmub_srv &&
+	    bp->base.ctx->dc->debug.dmub_command_table) {
+		encoder_control_dmcub(bp->base.ctx->dmub_srv, &params);
+		return BP_RESULT_OK;
+	}
+
+	if (EXEC_BIOS_CMD_TABLE(digxencodercontrol, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+static enum bp_result encoder_control_fallback(
+	struct bios_parser *bp,
+	struct bp_encoder_control *cntl)
+{
+	if (bp->base.ctx->dc->ctx->dmub_srv &&
+	    bp->base.ctx->dc->debug.dmub_command_table) {
+		return encoder_control_digx_v1_5(bp, cntl);
+	}
+
+	return BP_RESULT_FAILURE;
+}
+
+/*****************************************************************************
+ ******************************************************************************
+ **
+ **                  TRANSMITTER CONTROL
+ **
+ ******************************************************************************
+ *****************************************************************************/
+
+
+static enum bp_result transmitter_control_v1_6(
+	struct bios_parser *bp,
+	struct bp_transmitter_control *cntl);
+
+static enum bp_result transmitter_control_v1_7(
+	struct bios_parser *bp,
+	struct bp_transmitter_control *cntl);
+
+static enum bp_result transmitter_control_fallback(
+	struct bios_parser *bp,
+	struct bp_transmitter_control *cntl);
+
+static void init_transmitter_control(struct bios_parser *bp)
+{
+	uint8_t frev;
+	uint8_t crev = 0;
+
+	if (!BIOS_CMD_TABLE_REVISION(dig1transmittercontrol, frev, crev) && (bp->base.ctx->dc->ctx->dce_version <= DCN_VERSION_2_0))
+		BREAK_TO_DEBUGGER();
+
+	switch (crev) {
+	case 6:
+		bp->cmd_tbl.transmitter_control = transmitter_control_v1_6;
+		break;
+	case 7:
+		bp->cmd_tbl.transmitter_control = transmitter_control_v1_7;
+		break;
+	default:
+		bp->cmd_tbl.transmitter_control = transmitter_control_fallback;
+		break;
+	}
+}
+
+static void transmitter_control_dmcub(
+		struct dc_dmub_srv *dmcub,
+		struct dig_transmitter_control_parameters_v1_6 *dig)
+{
+	union dmub_rb_cmd cmd;
+
+	memset(&cmd, 0, sizeof(cmd));
+
+	cmd.dig1_transmitter_control.header.type = DMUB_CMD__VBIOS;
+	cmd.dig1_transmitter_control.header.sub_type =
+		DMUB_CMD__VBIOS_DIG1_TRANSMITTER_CONTROL;
+	cmd.dig1_transmitter_control.header.payload_bytes =
+		sizeof(cmd.dig1_transmitter_control) -
+		sizeof(cmd.dig1_transmitter_control.header);
+	cmd.dig1_transmitter_control.transmitter_control.dig = *dig;
+
+	dc_wake_and_execute_dmub_cmd(dmcub->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+static enum bp_result transmitter_control_v1_6(
+	struct bios_parser *bp,
+	struct bp_transmitter_control *cntl)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	const struct command_table_helper *cmd = bp->cmd_helper;
+	struct dig_transmitter_control_ps_allocation_v1_6 ps = { { 0 } };
+
+	ps.param.phyid = cmd->phy_id_to_atom(cntl->transmitter);
+	ps.param.action = (uint8_t)cntl->action;
+
+	if (cntl->action == TRANSMITTER_CONTROL_SET_VOLTAGE_AND_PREEMPASIS)
+		ps.param.mode_laneset.dplaneset = (uint8_t)cntl->lane_settings;
+	else
+		ps.param.mode_laneset.digmode =
+				cmd->signal_type_to_atom_dig_mode(cntl->signal);
+
+	ps.param.lanenum = (uint8_t)cntl->lanes_number;
+	ps.param.hpdsel = cmd->hpd_sel_to_atom(cntl->hpd_sel);
+	ps.param.digfe_sel = cmd->dig_encoder_sel_to_atom(cntl->engine_id);
+	ps.param.connobj_id = (uint8_t)cntl->connector_obj_id.id;
+	ps.param.symclk_10khz = cntl->pixel_clock/10;
+
+
+	if (cntl->action == TRANSMITTER_CONTROL_ENABLE ||
+		cntl->action == TRANSMITTER_CONTROL_ACTIAVATE ||
+		cntl->action == TRANSMITTER_CONTROL_DEACTIVATE) {
+		DC_LOG_BIOS("%s:ps.param.symclk_10khz = %d\n",\
+		__func__, ps.param.symclk_10khz);
+	}
+
+	if (bp->base.ctx->dc->ctx->dmub_srv &&
+	    bp->base.ctx->dc->debug.dmub_command_table) {
+		transmitter_control_dmcub(bp->base.ctx->dmub_srv, &ps.param);
+		return BP_RESULT_OK;
+	}
+
+/*color_depth not used any more, driver has deep color factor in the Phyclk*/
+	if (EXEC_BIOS_CMD_TABLE(dig1transmittercontrol, ps))
+		result = BP_RESULT_OK;
+	return result;
+}
+
+static void transmitter_control_dmcub_v1_7(
+		struct dc_dmub_srv *dmcub,
+		struct dmub_dig_transmitter_control_data_v1_7 *dig)
+{
+	union dmub_rb_cmd cmd;
+
+	memset(&cmd, 0, sizeof(cmd));
+
+	cmd.dig1_transmitter_control.header.type = DMUB_CMD__VBIOS;
+	cmd.dig1_transmitter_control.header.sub_type =
+		DMUB_CMD__VBIOS_DIG1_TRANSMITTER_CONTROL;
+	cmd.dig1_transmitter_control.header.payload_bytes =
+		sizeof(cmd.dig1_transmitter_control) -
+		sizeof(cmd.dig1_transmitter_control.header);
+	cmd.dig1_transmitter_control.transmitter_control.dig_v1_7 = *dig;
+
+	dc_wake_and_execute_dmub_cmd(dmcub->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+static struct dc_link *get_link_by_phy_id(struct dc *p_dc, uint32_t phy_id)
+{
+	struct dc_link *link = NULL;
+
+	// Get Transition Bitmask from dc_link structure associated with PHY
+	for (uint8_t link_id = 0; link_id < MAX_LINKS; link_id++) {
+		if (phy_id == p_dc->links[link_id]->link_enc->transmitter) {
+			link = p_dc->links[link_id];
+			break;
+		}
+	}
+
+	return link;
+}
+
+static enum bp_result transmitter_control_v1_7(
+	struct bios_parser *bp,
+	struct bp_transmitter_control *cntl)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	const struct command_table_helper *cmd = bp->cmd_helper;
+	struct dmub_dig_transmitter_control_data_v1_7 dig_v1_7 = {0};
+
+	uint8_t hpo_instance = (uint8_t)cntl->hpo_engine_id - ENGINE_ID_HPO_0;
+
+	if (dc_is_dp_signal(cntl->signal))
+		hpo_instance = (uint8_t)cntl->hpo_engine_id - ENGINE_ID_HPO_DP_0;
+
+	dig_v1_7.phyid = cmd->phy_id_to_atom(cntl->transmitter);
+	dig_v1_7.action = (uint8_t)cntl->action;
+
+	if (cntl->action == TRANSMITTER_CONTROL_SET_VOLTAGE_AND_PREEMPASIS)
+		dig_v1_7.mode_laneset.dplaneset = (uint8_t)cntl->lane_settings;
+	else
+		dig_v1_7.mode_laneset.digmode =
+				cmd->signal_type_to_atom_dig_mode(cntl->signal);
+
+	dig_v1_7.lanenum = (uint8_t)cntl->lanes_number;
+	dig_v1_7.hpdsel = cmd->hpd_sel_to_atom(cntl->hpd_sel);
+	dig_v1_7.digfe_sel = cmd->dig_encoder_sel_to_atom(cntl->engine_id);
+	dig_v1_7.connobj_id = (uint8_t)cntl->connector_obj_id.id;
+	dig_v1_7.HPO_instance = hpo_instance;
+	dig_v1_7.symclk_units.symclk_10khz = cntl->pixel_clock/10;
+
+	if (cntl->action == TRANSMITTER_CONTROL_ENABLE ||
+		cntl->action == TRANSMITTER_CONTROL_ACTIAVATE ||
+		cntl->action == TRANSMITTER_CONTROL_DEACTIVATE) {
+			DC_LOG_BIOS("%s:dig_v1_7.symclk_units.symclk_10khz = %d\n",
+			__func__, dig_v1_7.symclk_units.symclk_10khz);
+	}
+
+	if (bp->base.ctx->dc->ctx->dmub_srv &&
+		bp->base.ctx->dc->debug.dmub_command_table) {
+		struct dm_process_phy_transition_init_params process_phy_transition_init_params = {0};
+		struct dc_link *link = get_link_by_phy_id(bp->base.ctx->dc, dig_v1_7.phyid);
+		bool is_phy_transition_interlock_allowed = false;
+		uint8_t action = dig_v1_7.action;
+
+		if (link) {
+			if (link->phy_transition_bitmask &&
+				(action == TRANSMITTER_CONTROL_ENABLE || action == TRANSMITTER_CONTROL_DISABLE)) {
+				is_phy_transition_interlock_allowed = true;
+
+				// Prepare input parameters for processing ACPI retimers
+				process_phy_transition_init_params.action                   = action;
+				process_phy_transition_init_params.display_port_lanes_count = cntl->lanes_number;
+				process_phy_transition_init_params.phy_id                   = dig_v1_7.phyid;
+				process_phy_transition_init_params.signal                   = cntl->signal;
+				process_phy_transition_init_params.sym_clock_10khz          = dig_v1_7.symclk_units.symclk_10khz;
+				process_phy_transition_init_params.display_port_link_rate   = link->cur_link_settings.link_rate;
+				process_phy_transition_init_params.transition_bitmask       = link->phy_transition_bitmask;
+			}
+			dig_v1_7.skip_phy_ssc_reduction = link->wa_flags.skip_phy_ssc_reduction;
+		}
+
+		// Handle PRE_OFF_TO_ON: Process ACPI PHY Transition Interlock
+		if (is_phy_transition_interlock_allowed && action == TRANSMITTER_CONTROL_ENABLE)
+			dm_acpi_process_phy_transition_interlock(bp->base.ctx, process_phy_transition_init_params);
+
+		transmitter_control_dmcub_v1_7(bp->base.ctx->dmub_srv, &dig_v1_7);
+
+		// Handle POST_ON_TO_OFF: Process ACPI PHY Transition Interlock
+		if (is_phy_transition_interlock_allowed && action == TRANSMITTER_CONTROL_DISABLE)
+			dm_acpi_process_phy_transition_interlock(bp->base.ctx, process_phy_transition_init_params);
+
+		return BP_RESULT_OK;
+	}
+
+/*color_depth not used any more, driver has deep color factor in the Phyclk*/
+	if (EXEC_BIOS_CMD_TABLE(dig1transmittercontrol, dig_v1_7))
+		result = BP_RESULT_OK;
+	return result;
+}
+
+static enum bp_result transmitter_control_fallback(
+	struct bios_parser *bp,
+	struct bp_transmitter_control *cntl)
+{
+	if (bp->base.ctx->dc->ctx->dmub_srv &&
+	    bp->base.ctx->dc->debug.dmub_command_table) {
+		return transmitter_control_v1_7(bp, cntl);
+	}
+
+	return BP_RESULT_FAILURE;
+}
+
+/******************************************************************************
+ ******************************************************************************
+ **
+ **                  SET PIXEL CLOCK
+ **
+ ******************************************************************************
+ *****************************************************************************/
+
+static enum bp_result set_pixel_clock_v7(
+	struct bios_parser *bp,
+	struct bp_pixel_clock_parameters *bp_params);
+
+static enum bp_result set_pixel_clock_fallback(
+	struct bios_parser *bp,
+	struct bp_pixel_clock_parameters *bp_params);
+
+static void init_set_pixel_clock(struct bios_parser *bp)
+{
+	switch (BIOS_CMD_TABLE_PARA_REVISION(setpixelclock)) {
+	case 7:
+		bp->cmd_tbl.set_pixel_clock = set_pixel_clock_v7;
+		break;
+	default:
+		bp->cmd_tbl.set_pixel_clock = set_pixel_clock_fallback;
+		break;
+	}
+}
+
+static void set_pixel_clock_dmcub(
+		struct dc_dmub_srv *dmcub,
+		struct set_pixel_clock_parameter_v1_7 *clk)
+{
+	union dmub_rb_cmd cmd;
+
+	memset(&cmd, 0, sizeof(cmd));
+
+	cmd.set_pixel_clock.header.type = DMUB_CMD__VBIOS;
+	cmd.set_pixel_clock.header.sub_type = DMUB_CMD__VBIOS_SET_PIXEL_CLOCK;
+	cmd.set_pixel_clock.header.payload_bytes =
+		sizeof(cmd.set_pixel_clock) -
+		sizeof(cmd.set_pixel_clock.header);
+	cmd.set_pixel_clock.pixel_clock.clk = *clk;
+
+	dc_wake_and_execute_dmub_cmd(dmcub->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+static enum bp_result set_pixel_clock_v7(
+	struct bios_parser *bp,
+	struct bp_pixel_clock_parameters *bp_params)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	struct set_pixel_clock_parameter_v1_7 clk;
+	uint8_t controller_id;
+	uint32_t pll_id;
+
+	memset(&clk, 0, sizeof(clk));
+
+	if (bp->cmd_helper->clock_source_id_to_atom(bp_params->pll_id, &pll_id)
+			&& bp->cmd_helper->controller_id_to_atom(bp_params->
+					controller_id, &controller_id)) {
+		/* Note: VBIOS still wants to use ucCRTC name which is now
+		 * 1 byte in ULONG
+		 *typedef struct _CRTC_PIXEL_CLOCK_FREQ
+		 *{
+		 * target the pixel clock to drive the CRTC timing.
+		 * ULONG ulPixelClock:24;
+		 * 0 means disable PPLL/DCPLL. Expanded to 24 bits comparing to
+		 * previous version.
+		 * ATOM_CRTC1~6, indicate the CRTC controller to
+		 * ULONG ucCRTC:8;
+		 * drive the pixel clock. not used for DCPLL case.
+		 *}CRTC_PIXEL_CLOCK_FREQ;
+		 *union
+		 *{
+		 * pixel clock and CRTC id frequency
+		 * CRTC_PIXEL_CLOCK_FREQ ulCrtcPclkFreq;
+		 * ULONG ulDispEngClkFreq; dispclk frequency
+		 *};
+		 */
+		clk.crtc_id = controller_id;
+		clk.pll_id = (uint8_t) pll_id;
+		clk.encoderobjid =
+			bp->cmd_helper->encoder_id_to_atom(
+				dal_graphics_object_id_get_encoder_id(
+					bp_params->encoder_object_id));
+
+		clk.encoder_mode = (uint8_t) bp->
+			cmd_helper->encoder_mode_bp_to_atom(
+				bp_params->signal_type, false);
+
+		clk.pixclk_100hz = cpu_to_le32(bp_params->target_pixel_clock_100hz);
+
+		clk.deep_color_ratio =
+			(uint8_t) bp->cmd_helper->
+				transmitter_color_depth_to_atom(
+					bp_params->color_depth);
+
+		DC_LOG_BIOS("%s:program display clock = %d, tg = %d, pll = %d, "\
+				"colorDepth = %d\n", __func__,
+				bp_params->target_pixel_clock_100hz, (int)controller_id,
+				pll_id, bp_params->color_depth);
+
+		if (bp_params->flags.FORCE_PROGRAMMING_OF_PLL)
+			clk.miscinfo |= PIXEL_CLOCK_V7_MISC_FORCE_PROG_PPLL;
+
+		if (bp_params->flags.PROGRAM_PHY_PLL_ONLY)
+			clk.miscinfo |= PIXEL_CLOCK_V7_MISC_PROG_PHYPLL;
+
+		if (bp_params->flags.SUPPORT_YUV_420)
+			clk.miscinfo |= PIXEL_CLOCK_V7_MISC_YUV420_MODE;
+
+		if (bp_params->flags.SET_XTALIN_REF_SRC)
+			clk.miscinfo |= PIXEL_CLOCK_V7_MISC_REF_DIV_SRC_XTALIN;
+
+		if (bp_params->flags.SET_GENLOCK_REF_DIV_SRC)
+			clk.miscinfo |= PIXEL_CLOCK_V7_MISC_REF_DIV_SRC_GENLK;
+
+		if (bp_params->signal_type == SIGNAL_TYPE_DVI_DUAL_LINK)
+			clk.miscinfo |= PIXEL_CLOCK_V7_MISC_DVI_DUALLINK_EN;
+
+		if (bp->base.ctx->dc->ctx->dmub_srv &&
+		    bp->base.ctx->dc->debug.dmub_command_table) {
+			set_pixel_clock_dmcub(bp->base.ctx->dmub_srv, &clk);
+			return BP_RESULT_OK;
+		}
+
+		if (EXEC_BIOS_CMD_TABLE(setpixelclock, clk))
+			result = BP_RESULT_OK;
+	}
+	return result;
+}
+
+static enum bp_result set_pixel_clock_fallback(
+	struct bios_parser *bp,
+	struct bp_pixel_clock_parameters *bp_params)
+{
+	if (bp->base.ctx->dc->ctx->dmub_srv &&
+	    bp->base.ctx->dc->debug.dmub_command_table) {
+		return set_pixel_clock_v7(bp, bp_params);
+	}
+
+	return BP_RESULT_FAILURE;
+}
+
+/******************************************************************************
+ ******************************************************************************
+ **
+ **                  SET CRTC TIMING
+ **
+ ******************************************************************************
+ *****************************************************************************/
+
+static enum bp_result set_crtc_using_dtd_timing_v3(
+	struct bios_parser *bp,
+	struct bp_hw_crtc_timing_parameters *bp_params);
+
+static void init_set_crtc_timing(struct bios_parser *bp)
+{
+	uint32_t dtd_version =
+			BIOS_CMD_TABLE_PARA_REVISION(setcrtc_usingdtdtiming);
+
+	switch (dtd_version) {
+	case 3:
+		bp->cmd_tbl.set_crtc_timing =
+			set_crtc_using_dtd_timing_v3;
+		break;
+	default:
+		bp->cmd_tbl.set_crtc_timing = NULL;
+		break;
+	}
+}
+
+static enum bp_result set_crtc_using_dtd_timing_v3(
+	struct bios_parser *bp,
+	struct bp_hw_crtc_timing_parameters *bp_params)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+	struct set_crtc_using_dtd_timing_parameters params = {0};
+	uint8_t atom_controller_id;
+
+	if (bp->cmd_helper->controller_id_to_atom(
+			bp_params->controller_id, &atom_controller_id))
+		params.crtc_id = atom_controller_id;
+
+	/* bios usH_Size wants h addressable size */
+	params.h_size = cpu_to_le16((uint16_t)bp_params->h_addressable);
+	/* bios usH_Blanking_Time wants borders included in blanking */
+	params.h_blanking_time =
+			cpu_to_le16((uint16_t)(bp_params->h_total -
+					bp_params->h_addressable));
+	/* bios usV_Size wants v addressable size */
+	params.v_size = cpu_to_le16((uint16_t)bp_params->v_addressable);
+	/* bios usV_Blanking_Time wants borders included in blanking */
+	params.v_blanking_time =
+			cpu_to_le16((uint16_t)(bp_params->v_total -
+					bp_params->v_addressable));
+	/* bios usHSyncOffset is the offset from the end of h addressable,
+	 * our horizontalSyncStart is the offset from the beginning
+	 * of h addressable
+	 */
+	params.h_syncoffset =
+			cpu_to_le16((uint16_t)(bp_params->h_sync_start -
+					bp_params->h_addressable));
+	params.h_syncwidth = cpu_to_le16((uint16_t)bp_params->h_sync_width);
+	/* bios usHSyncOffset is the offset from the end of v addressable,
+	 * our verticalSyncStart is the offset from the beginning of
+	 * v addressable
+	 */
+	params.v_syncoffset =
+			cpu_to_le16((uint16_t)(bp_params->v_sync_start -
+					bp_params->v_addressable));
+	params.v_syncwidth = cpu_to_le16((uint16_t)bp_params->v_sync_width);
+
+	/* we assume that overscan from original timing does not get bigger
+	 * than 255
+	 * we will program all the borders in the Set CRTC Overscan call below
+	 */
+
+	if (bp_params->flags.HSYNC_POSITIVE_POLARITY == 0)
+		params.modemiscinfo =
+				cpu_to_le16(le16_to_cpu(params.modemiscinfo) |
+						ATOM_HSYNC_POLARITY);
+
+	if (bp_params->flags.VSYNC_POSITIVE_POLARITY == 0)
+		params.modemiscinfo =
+				cpu_to_le16(le16_to_cpu(params.modemiscinfo) |
+						ATOM_VSYNC_POLARITY);
+
+	if (bp_params->flags.INTERLACE)	{
+		params.modemiscinfo =
+				cpu_to_le16(le16_to_cpu(params.modemiscinfo) |
+						ATOM_INTERLACE);
+
+		/* original DAL code has this condition to apply this
+		 * for non-TV/CV only
+		 * due to complex MV testing for possible impact
+		 * if ( pACParameters->signal != SignalType_YPbPr &&
+		 *  pACParameters->signal != SignalType_Composite &&
+		 *  pACParameters->signal != SignalType_SVideo)
+		 */
+		{
+			/* HW will deduct 0.5 line from 2nd feild.
+			 * i.e. for 1080i, it is 2 lines for 1st field,
+			 * 2.5 lines for the 2nd feild. we need input as 5
+			 * instead of 4.
+			 * but it is 4 either from Edid data (spec CEA 861)
+			 * or CEA timing table.
+			 */
+			le16_add_cpu(&params.v_syncoffset, 1);
+		}
+	}
+
+	if (bp_params->flags.HORZ_COUNT_BY_TWO)
+		params.modemiscinfo =
+			cpu_to_le16(le16_to_cpu(params.modemiscinfo) |
+					0x100); /* ATOM_DOUBLE_CLOCK_MODE */
+
+	if (EXEC_BIOS_CMD_TABLE(setcrtc_usingdtdtiming, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+/******************************************************************************
+ ******************************************************************************
+ **
+ **                  ENABLE CRTC
+ **
+ ******************************************************************************
+ *****************************************************************************/
+
+static enum bp_result enable_crtc_v1(
+	struct bios_parser *bp,
+	enum controller_id controller_id,
+	bool enable);
+
+static void init_enable_crtc(struct bios_parser *bp)
+{
+	switch (BIOS_CMD_TABLE_PARA_REVISION(enablecrtc)) {
+	case 1:
+		bp->cmd_tbl.enable_crtc = enable_crtc_v1;
+		break;
+	default:
+		bp->cmd_tbl.enable_crtc = NULL;
+		break;
+	}
+}
+
+static enum bp_result enable_crtc_v1(
+	struct bios_parser *bp,
+	enum controller_id controller_id,
+	bool enable)
+{
+	bool result = BP_RESULT_FAILURE;
+	struct enable_crtc_parameters params = {0};
+	uint8_t id;
+
+	if (bp->cmd_helper->controller_id_to_atom(controller_id, &id))
+		params.crtc_id = id;
+	else
+		return BP_RESULT_BADINPUT;
+
+	if (enable)
+		params.enable = ATOM_ENABLE;
+	else
+		params.enable = ATOM_DISABLE;
+
+	if (EXEC_BIOS_CMD_TABLE(enablecrtc, params))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+/******************************************************************************
+ ******************************************************************************
+ **
+ **                  DISPLAY PLL
+ **
+ ******************************************************************************
+ *****************************************************************************/
+
+
+
+/******************************************************************************
+ ******************************************************************************
+ **
+ **                  EXTERNAL ENCODER CONTROL
+ **
+ ******************************************************************************
+ *****************************************************************************/
+
+static enum bp_result external_encoder_control_v3(
+	struct bios_parser *bp,
+	struct bp_external_encoder_control *cntl);
+
+static void init_external_encoder_control(
+	struct bios_parser *bp)
+{
+	switch (BIOS_CMD_TABLE_PARA_REVISION(externalencodercontrol)) {
+	case 3:
+		bp->cmd_tbl.external_encoder_control =
+				external_encoder_control_v3;
+		break;
+	default:
+		bp->cmd_tbl.external_encoder_control = NULL;
+		break;
+	}
+}
+
+static enum bp_result external_encoder_control_v3(
+	struct bios_parser *bp,
+	struct bp_external_encoder_control *cntl)
+{
+	/* TODO */
+	return BP_RESULT_OK;
+}
+
+/******************************************************************************
+ ******************************************************************************
+ **
+ **                  ENABLE DISPLAY POWER GATING
+ **
+ ******************************************************************************
+ *****************************************************************************/
+
+static enum bp_result enable_disp_power_gating_v2_1(
+	struct bios_parser *bp,
+	enum controller_id crtc_id,
+	enum bp_pipe_control_action action);
+
+static enum bp_result enable_disp_power_gating_fallback(
+	struct bios_parser *bp,
+	enum controller_id crtc_id,
+	enum bp_pipe_control_action action);
+
+static void init_enable_disp_power_gating(
+	struct bios_parser *bp)
+{
+	switch (BIOS_CMD_TABLE_PARA_REVISION(enabledisppowergating)) {
+	case 1:
+		bp->cmd_tbl.enable_disp_power_gating =
+				enable_disp_power_gating_v2_1;
+		break;
+	default:
+		dm_output_to_console("Don't enable_disp_power_gating enable_crtc for v%d\n",
+			 BIOS_CMD_TABLE_PARA_REVISION(enabledisppowergating));
+		bp->cmd_tbl.enable_disp_power_gating = enable_disp_power_gating_fallback;
+		break;
+	}
+}
+
+static void enable_disp_power_gating_dmcub(
+	struct dc_dmub_srv *dmcub,
+	struct enable_disp_power_gating_parameters_v2_1 *pwr)
+{
+	union dmub_rb_cmd cmd;
+
+	memset(&cmd, 0, sizeof(cmd));
+
+	cmd.enable_disp_power_gating.header.type = DMUB_CMD__VBIOS;
+	cmd.enable_disp_power_gating.header.sub_type =
+		DMUB_CMD__VBIOS_ENABLE_DISP_POWER_GATING;
+	cmd.enable_disp_power_gating.header.payload_bytes =
+		sizeof(cmd.enable_disp_power_gating) -
+		sizeof(cmd.enable_disp_power_gating.header);
+	cmd.enable_disp_power_gating.power_gating.pwr = *pwr;
+
+	dc_wake_and_execute_dmub_cmd(dmcub->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+static enum bp_result enable_disp_power_gating_v2_1(
+	struct bios_parser *bp,
+	enum controller_id crtc_id,
+	enum bp_pipe_control_action action)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+
+
+	struct enable_disp_power_gating_ps_allocation ps = { { 0 } };
+	uint8_t atom_crtc_id;
+
+	if (bp->cmd_helper->controller_id_to_atom(crtc_id, &atom_crtc_id))
+		ps.param.disp_pipe_id = atom_crtc_id;
+	else
+		return BP_RESULT_BADINPUT;
+
+	ps.param.enable =
+		bp->cmd_helper->disp_power_gating_action_to_atom(action);
+
+	if (bp->base.ctx->dc->ctx->dmub_srv &&
+	    bp->base.ctx->dc->debug.dmub_command_table) {
+		enable_disp_power_gating_dmcub(bp->base.ctx->dmub_srv,
+					       &ps.param);
+		return BP_RESULT_OK;
+	}
+
+	if (EXEC_BIOS_CMD_TABLE(enabledisppowergating, ps.param))
+		result = BP_RESULT_OK;
+
+	return result;
+}
+
+static enum bp_result enable_disp_power_gating_fallback(
+	struct bios_parser *bp,
+	enum controller_id crtc_id,
+	enum bp_pipe_control_action action)
+{
+	if (bp->base.ctx->dc->ctx->dmub_srv &&
+	    bp->base.ctx->dc->debug.dmub_command_table) {
+		return enable_disp_power_gating_v2_1(bp, crtc_id, action);
+	}
+
+	return BP_RESULT_FAILURE;
+}
+
+/******************************************************************************
+*******************************************************************************
+ **
+ **                  SET DCE CLOCK
+ **
+*******************************************************************************
+*******************************************************************************/
+
+static enum bp_result set_dce_clock_v2_1(
+	struct bios_parser *bp,
+	struct bp_set_dce_clock_parameters *bp_params);
+
+static void init_set_dce_clock(struct bios_parser *bp)
+{
+	switch (BIOS_CMD_TABLE_PARA_REVISION(setdceclock)) {
+	case 1:
+		bp->cmd_tbl.set_dce_clock = set_dce_clock_v2_1;
+		break;
+	default:
+		bp->cmd_tbl.set_dce_clock = NULL;
+		break;
+	}
+}
+
+static enum bp_result set_dce_clock_v2_1(
+	struct bios_parser *bp,
+	struct bp_set_dce_clock_parameters *bp_params)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+
+	struct set_dce_clock_ps_allocation_v2_1 params;
+	uint32_t atom_pll_id;
+	uint32_t atom_clock_type;
+	const struct command_table_helper *cmd = bp->cmd_helper;
+
+	memset(&params, 0, sizeof(params));
+
+	if (!cmd->clock_source_id_to_atom(bp_params->pll_id, &atom_pll_id) ||
+			!cmd->dc_clock_type_to_atom(bp_params->clock_type,
+					&atom_clock_type))
+		return BP_RESULT_BADINPUT;
+
+	params.param.dceclksrc  = atom_pll_id;
+	params.param.dceclktype = atom_clock_type;
+
+	if (bp_params->clock_type == DCECLOCK_TYPE_DPREFCLK) {
+		if (bp_params->flags.USE_GENLOCK_AS_SOURCE_FOR_DPREFCLK)
+			params.param.dceclkflag |=
+					DCE_CLOCK_FLAG_PLL_REFCLK_SRC_GENLK;
+
+		if (bp_params->flags.USE_PCIE_AS_SOURCE_FOR_DPREFCLK)
+			params.param.dceclkflag |=
+					DCE_CLOCK_FLAG_PLL_REFCLK_SRC_PCIE;
+
+		if (bp_params->flags.USE_XTALIN_AS_SOURCE_FOR_DPREFCLK)
+			params.param.dceclkflag |=
+					DCE_CLOCK_FLAG_PLL_REFCLK_SRC_XTALIN;
+
+		if (bp_params->flags.USE_GENERICA_AS_SOURCE_FOR_DPREFCLK)
+			params.param.dceclkflag |=
+					DCE_CLOCK_FLAG_PLL_REFCLK_SRC_GENERICA;
+	} else
+		/* only program clock frequency if display clock is used;
+		 * VBIOS will program DPREFCLK
+		 * We need to convert from KHz units into 10KHz units
+		 */
+		params.param.dceclk_10khz = cpu_to_le32(
+				bp_params->target_clock_frequency / 10);
+	DC_LOG_BIOS("%s:target_clock_frequency = %d"\
+			"clock_type = %d \n", __func__,\
+			bp_params->target_clock_frequency,\
+			bp_params->clock_type);
+
+	if (EXEC_BIOS_CMD_TABLE(setdceclock, params)) {
+		/* Convert from 10KHz units back to KHz */
+		bp_params->target_clock_frequency = le32_to_cpu(
+				params.param.dceclk_10khz) * 10;
+		result = BP_RESULT_OK;
+	}
+
+	return result;
+}
+
+
+/******************************************************************************
+ ******************************************************************************
+ **
+ **                  GET SMU CLOCK INFO
+ **
+ ******************************************************************************
+ *****************************************************************************/
+
+static unsigned int get_smu_clock_info_v3_1(struct bios_parser *bp, uint8_t id);
+
+static void init_get_smu_clock_info(struct bios_parser *bp)
+{
+	/* TODO add switch for table vrsion */
+	bp->cmd_tbl.get_smu_clock_info = get_smu_clock_info_v3_1;
+
+}
+
+static unsigned int get_smu_clock_info_v3_1(struct bios_parser *bp, uint8_t id)
+{
+	struct atom_get_smu_clock_info_parameters_v3_1 smu_input = {0};
+	struct atom_get_smu_clock_info_output_parameters_v3_1 smu_output;
+
+	smu_input.command = GET_SMU_CLOCK_INFO_V3_1_GET_PLLVCO_FREQ;
+	smu_input.syspll_id = id;
+
+	/* Get Specific Clock */
+	if (EXEC_BIOS_CMD_TABLE(getsmuclockinfo, smu_input)) {
+		memmove(&smu_output, &smu_input, sizeof(
+			struct atom_get_smu_clock_info_parameters_v3_1));
+		return smu_output.atom_smu_outputclkfreq.syspllvcofreq_10khz;
+	}
+
+	return 0;
+}
+
+/******************************************************************************
+ ******************************************************************************
+ **
+ **                  LVTMA CONTROL
+ **
+ ******************************************************************************
+ *****************************************************************************/
+
+static enum bp_result enable_lvtma_control(
+	struct bios_parser *bp,
+	uint8_t uc_pwr_on,
+	uint8_t pwrseq_instance,
+	uint8_t bypass_panel_control_wait);
+
+static void init_enable_lvtma_control(struct bios_parser *bp)
+{
+	/* TODO add switch for table vrsion */
+	bp->cmd_tbl.enable_lvtma_control = enable_lvtma_control;
+
+}
+
+static void enable_lvtma_control_dmcub(
+	struct dc_dmub_srv *dmcub,
+	uint8_t uc_pwr_on,
+	uint8_t pwrseq_instance,
+	uint8_t bypass_panel_control_wait)
+{
+
+	union dmub_rb_cmd cmd;
+
+	memset(&cmd, 0, sizeof(cmd));
+
+	cmd.lvtma_control.header.type = DMUB_CMD__VBIOS;
+	cmd.lvtma_control.header.sub_type =
+			DMUB_CMD__VBIOS_LVTMA_CONTROL;
+	cmd.lvtma_control.data.uc_pwr_action =
+			uc_pwr_on;
+	cmd.lvtma_control.data.pwrseq_inst =
+			pwrseq_instance;
+	cmd.lvtma_control.data.bypass_panel_control_wait =
+			bypass_panel_control_wait;
+	dc_wake_and_execute_dmub_cmd(dmcub->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+static enum bp_result enable_lvtma_control(
+	struct bios_parser *bp,
+	uint8_t uc_pwr_on,
+	uint8_t pwrseq_instance,
+	uint8_t bypass_panel_control_wait)
+{
+	enum bp_result result = BP_RESULT_FAILURE;
+
+	if (bp->base.ctx->dc->ctx->dmub_srv &&
+	    bp->base.ctx->dc->debug.dmub_command_table) {
+		enable_lvtma_control_dmcub(bp->base.ctx->dmub_srv,
+				uc_pwr_on,
+				pwrseq_instance,
+				bypass_panel_control_wait);
+		return BP_RESULT_OK;
+	}
+	return result;
+}
+
+void dal_firmware_parser_init_cmd_tbl(struct bios_parser *bp)
+{
+	init_dig_encoder_control(bp);
+	init_transmitter_control(bp);
+	init_set_pixel_clock(bp);
+
+	init_set_crtc_timing(bp);
+
+	init_enable_crtc(bp);
+
+	init_external_encoder_control(bp);
+	init_enable_disp_power_gating(bp);
+	init_set_dce_clock(bp);
+	init_get_smu_clock_info(bp);
+
+	init_enable_lvtma_control(bp);
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/bios/command_table2.h b/drivers/gpu/drm/amd/display/dc/bios/command_table2.h
new file mode 100644
index 000000000000..41c8c014397f
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/command_table2.h
@@ -0,0 +1,105 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_COMMAND_TABLE2_H__
+#define __DAL_COMMAND_TABLE2_H__
+
+struct bios_parser;
+struct bp_encoder_control;
+
+struct cmd_tbl {
+	enum bp_result (*dig_encoder_control)(
+		struct bios_parser *bp,
+		struct bp_encoder_control *control);
+	enum bp_result (*encoder_control_dig1)(
+		struct bios_parser *bp,
+		struct bp_encoder_control *control);
+	enum bp_result (*encoder_control_dig2)(
+		struct bios_parser *bp,
+		struct bp_encoder_control *control);
+	enum bp_result (*transmitter_control)(
+		struct bios_parser *bp,
+		struct bp_transmitter_control *control);
+	enum bp_result (*set_pixel_clock)(
+		struct bios_parser *bp,
+		struct bp_pixel_clock_parameters *bp_params);
+	enum bp_result (*enable_spread_spectrum_on_ppll)(
+		struct bios_parser *bp,
+		struct bp_spread_spectrum_parameters *bp_params,
+		bool enable);
+	enum bp_result (*adjust_display_pll)(
+		struct bios_parser *bp,
+		struct bp_adjust_pixel_clock_parameters *bp_params);
+	enum bp_result (*dac1_encoder_control)(
+		struct bios_parser *bp,
+		bool enable,
+		uint32_t pixel_clock,
+		uint8_t dac_standard);
+	enum bp_result (*dac2_encoder_control)(
+		struct bios_parser *bp,
+		bool enable,
+		uint32_t pixel_clock,
+		uint8_t dac_standard);
+	enum bp_result (*dac1_output_control)(
+		struct bios_parser *bp,
+		bool enable);
+	enum bp_result (*dac2_output_control)(
+		struct bios_parser *bp,
+		bool enable);
+	enum bp_result (*set_crtc_timing)(
+		struct bios_parser *bp,
+		struct bp_hw_crtc_timing_parameters *bp_params);
+	enum bp_result (*enable_crtc)(
+		struct bios_parser *bp,
+		enum controller_id controller_id,
+		bool enable);
+	enum bp_result (*enable_crtc_mem_req)(
+		struct bios_parser *bp,
+		enum controller_id controller_id,
+		bool enable);
+	enum bp_result (*program_clock)(
+		struct bios_parser *bp,
+		struct bp_pixel_clock_parameters *bp_params);
+	enum bp_result (*external_encoder_control)(
+			struct bios_parser *bp,
+			struct bp_external_encoder_control *cntl);
+	enum bp_result (*enable_disp_power_gating)(
+		struct bios_parser *bp,
+		enum controller_id crtc_id,
+		enum bp_pipe_control_action action);
+	enum bp_result (*set_dce_clock)(
+		struct bios_parser *bp,
+		struct bp_set_dce_clock_parameters *bp_params);
+	unsigned int (*get_smu_clock_info)(
+			struct bios_parser *bp, uint8_t id);
+	enum bp_result (*enable_lvtma_control)(struct bios_parser *bp,
+			uint8_t uc_pwr_on,
+			uint8_t pwrseq_instance,
+			uint8_t bypass_panel_control_wait);
+};
+
+void dal_firmware_parser_init_cmd_tbl(struct bios_parser *bp);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/bios/command_table_helper.c b/drivers/gpu/drm/amd/display/dc/bios/command_table_helper.c
new file mode 100644
index 000000000000..91bc8a06e2cf
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/command_table_helper.c
@@ -0,0 +1,399 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+
+#include "atom.h"
+
+#include "include/bios_parser_types.h"
+
+#include "command_table_helper.h"
+
+bool dal_bios_parser_init_cmd_tbl_helper(
+	const struct command_table_helper **h,
+	enum dce_version dce)
+{
+	switch (dce) {
+#if defined(CONFIG_DRM_AMD_DC_SI)
+	case DCE_VERSION_6_0:
+	case DCE_VERSION_6_1:
+	case DCE_VERSION_6_4:
+		*h = dal_cmd_tbl_helper_dce60_get_table();
+		return true;
+#endif
+
+	case DCE_VERSION_8_0:
+	case DCE_VERSION_8_1:
+	case DCE_VERSION_8_3:
+		*h = dal_cmd_tbl_helper_dce80_get_table();
+		return true;
+
+	case DCE_VERSION_10_0:
+		*h = dal_cmd_tbl_helper_dce110_get_table();
+		return true;
+
+	case DCE_VERSION_11_0:
+		*h = dal_cmd_tbl_helper_dce110_get_table();
+		return true;
+
+	case DCE_VERSION_11_2:
+	case DCE_VERSION_11_22:
+		*h = dal_cmd_tbl_helper_dce112_get_table();
+		return true;
+
+	default:
+		/* Unsupported DCE */
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+}
+
+/* real implementations */
+
+bool dal_cmd_table_helper_controller_id_to_atom(
+	enum controller_id id,
+	uint8_t *atom_id)
+{
+	if (atom_id == NULL) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+
+	switch (id) {
+	case CONTROLLER_ID_D0:
+		*atom_id = ATOM_CRTC1;
+		return true;
+	case CONTROLLER_ID_D1:
+		*atom_id = ATOM_CRTC2;
+		return true;
+	case CONTROLLER_ID_D2:
+		*atom_id = ATOM_CRTC3;
+		return true;
+	case CONTROLLER_ID_D3:
+		*atom_id = ATOM_CRTC4;
+		return true;
+	case CONTROLLER_ID_D4:
+		*atom_id = ATOM_CRTC5;
+		return true;
+	case CONTROLLER_ID_D5:
+		*atom_id = ATOM_CRTC6;
+		return true;
+	case CONTROLLER_ID_UNDERLAY0:
+		*atom_id = ATOM_UNDERLAY_PIPE0;
+		return true;
+	case CONTROLLER_ID_UNDEFINED:
+		*atom_id = ATOM_CRTC_INVALID;
+		return true;
+	default:
+		/* Wrong controller id */
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+}
+
+/**
+ * dal_cmd_table_helper_transmitter_bp_to_atom - Translate the Transmitter to the
+ *                                    corresponding ATOM BIOS value
+ * @t: transmitter
+ * returns: output digitalTransmitter
+ *    // =00: Digital Transmitter1 ( UNIPHY linkAB )
+ *    // =01: Digital Transmitter2 ( UNIPHY linkCD )
+ *    // =02: Digital Transmitter3 ( UNIPHY linkEF )
+ */
+uint8_t dal_cmd_table_helper_transmitter_bp_to_atom(
+	enum transmitter t)
+{
+	switch (t) {
+	case TRANSMITTER_UNIPHY_A:
+	case TRANSMITTER_UNIPHY_B:
+	case TRANSMITTER_TRAVIS_LCD:
+		return 0;
+	case TRANSMITTER_UNIPHY_C:
+	case TRANSMITTER_UNIPHY_D:
+		return 1;
+	case TRANSMITTER_UNIPHY_E:
+	case TRANSMITTER_UNIPHY_F:
+		return 2;
+	default:
+		/* Invalid Transmitter Type! */
+		BREAK_TO_DEBUGGER();
+		return 0;
+	}
+}
+
+uint32_t dal_cmd_table_helper_encoder_mode_bp_to_atom(
+	enum signal_type s,
+	bool enable_dp_audio)
+{
+	switch (s) {
+	case SIGNAL_TYPE_DVI_SINGLE_LINK:
+	case SIGNAL_TYPE_DVI_DUAL_LINK:
+		return ATOM_ENCODER_MODE_DVI;
+	case SIGNAL_TYPE_HDMI_TYPE_A:
+		return ATOM_ENCODER_MODE_HDMI;
+	case SIGNAL_TYPE_LVDS:
+		return ATOM_ENCODER_MODE_LVDS;
+	case SIGNAL_TYPE_EDP:
+	case SIGNAL_TYPE_DISPLAY_PORT_MST:
+	case SIGNAL_TYPE_DISPLAY_PORT:
+	case SIGNAL_TYPE_VIRTUAL:
+		if (enable_dp_audio)
+			return ATOM_ENCODER_MODE_DP_AUDIO;
+		else
+			return ATOM_ENCODER_MODE_DP;
+	case SIGNAL_TYPE_RGB:
+		return ATOM_ENCODER_MODE_CRT;
+	default:
+		return ATOM_ENCODER_MODE_CRT;
+	}
+}
+
+void dal_cmd_table_helper_assign_control_parameter(
+	const struct command_table_helper *h,
+	struct bp_encoder_control *control,
+	DIG_ENCODER_CONTROL_PARAMETERS_V2 *ctrl_param)
+{
+	/* there are three transmitter blocks, each one has two links 4-lanes
+	 * each, A+B, C+D, E+F, Uniphy A, C and E are enumerated as link 0 in
+	 * each transmitter block B, D and F as link 1, third transmitter block
+	 * has non splitable links (UniphyE and UniphyF can not be configured
+	 * separately to drive two different streams)
+	 */
+	if ((control->transmitter == TRANSMITTER_UNIPHY_B) ||
+		(control->transmitter == TRANSMITTER_UNIPHY_D) ||
+		(control->transmitter == TRANSMITTER_UNIPHY_F)) {
+		/* Bit2: Link Select
+		 * =0: PHY linkA/C/E
+		 * =1: PHY linkB/D/F
+		 */
+		ctrl_param->acConfig.ucLinkSel = 1;
+	}
+
+	/* Bit[4:3]: Transmitter Selection
+	 * =00: Digital Transmitter1 ( UNIPHY linkAB )
+	 * =01: Digital Transmitter2 ( UNIPHY linkCD )
+	 * =02: Digital Transmitter3 ( UNIPHY linkEF )
+	 * =03: Reserved
+	 */
+	ctrl_param->acConfig.ucTransmitterSel =
+		(uint8_t)(h->transmitter_bp_to_atom(control->transmitter));
+
+	/* We need to convert from KHz units into 10KHz units */
+	ctrl_param->ucAction = h->encoder_action_to_atom(control->action);
+	ctrl_param->usPixelClock = cpu_to_le16((uint16_t)(control->pixel_clock / 10));
+	ctrl_param->ucEncoderMode =
+		(uint8_t)(h->encoder_mode_bp_to_atom(
+			control->signal, control->enable_dp_audio));
+	ctrl_param->ucLaneNum = (uint8_t)(control->lanes_number);
+}
+
+bool dal_cmd_table_helper_clock_source_id_to_ref_clk_src(
+	enum clock_source_id id,
+	uint32_t *ref_clk_src_id)
+{
+	if (ref_clk_src_id == NULL) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+
+	switch (id) {
+	case CLOCK_SOURCE_ID_PLL1:
+		*ref_clk_src_id = ENCODER_REFCLK_SRC_P1PLL;
+		return true;
+	case CLOCK_SOURCE_ID_PLL2:
+		*ref_clk_src_id = ENCODER_REFCLK_SRC_P2PLL;
+		return true;
+	case CLOCK_SOURCE_ID_DCPLL:
+		*ref_clk_src_id = ENCODER_REFCLK_SRC_DCPLL;
+		return true;
+	case CLOCK_SOURCE_ID_EXTERNAL:
+		*ref_clk_src_id = ENCODER_REFCLK_SRC_EXTCLK;
+		return true;
+	case CLOCK_SOURCE_ID_UNDEFINED:
+		*ref_clk_src_id = ENCODER_REFCLK_SRC_INVALID;
+		return true;
+	default:
+		/* Unsupported clock source id */
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+}
+
+uint8_t dal_cmd_table_helper_encoder_id_to_atom(
+	enum encoder_id id)
+{
+	switch (id) {
+	case ENCODER_ID_INTERNAL_LVDS:
+		return ENCODER_OBJECT_ID_INTERNAL_LVDS;
+	case ENCODER_ID_INTERNAL_TMDS1:
+		return ENCODER_OBJECT_ID_INTERNAL_TMDS1;
+	case ENCODER_ID_INTERNAL_TMDS2:
+		return ENCODER_OBJECT_ID_INTERNAL_TMDS2;
+	case ENCODER_ID_INTERNAL_DAC1:
+		return ENCODER_OBJECT_ID_INTERNAL_DAC1;
+	case ENCODER_ID_INTERNAL_DAC2:
+		return ENCODER_OBJECT_ID_INTERNAL_DAC2;
+	case ENCODER_ID_INTERNAL_LVTM1:
+		return ENCODER_OBJECT_ID_INTERNAL_LVTM1;
+	case ENCODER_ID_INTERNAL_HDMI:
+		return ENCODER_OBJECT_ID_HDMI_INTERNAL;
+	case ENCODER_ID_EXTERNAL_TRAVIS:
+		return ENCODER_OBJECT_ID_TRAVIS;
+	case ENCODER_ID_EXTERNAL_NUTMEG:
+		return ENCODER_OBJECT_ID_NUTMEG;
+	case ENCODER_ID_INTERNAL_KLDSCP_TMDS1:
+		return ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1;
+	case ENCODER_ID_INTERNAL_KLDSCP_DAC1:
+		return ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1;
+	case ENCODER_ID_INTERNAL_KLDSCP_DAC2:
+		return ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2;
+	case ENCODER_ID_EXTERNAL_MVPU_FPGA:
+		return ENCODER_OBJECT_ID_MVPU_FPGA;
+	case ENCODER_ID_INTERNAL_DDI:
+		return ENCODER_OBJECT_ID_INTERNAL_DDI;
+	case ENCODER_ID_INTERNAL_UNIPHY:
+		return ENCODER_OBJECT_ID_INTERNAL_UNIPHY;
+	case ENCODER_ID_INTERNAL_KLDSCP_LVTMA:
+		return ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA;
+	case ENCODER_ID_INTERNAL_UNIPHY1:
+		return ENCODER_OBJECT_ID_INTERNAL_UNIPHY1;
+	case ENCODER_ID_INTERNAL_UNIPHY2:
+		return ENCODER_OBJECT_ID_INTERNAL_UNIPHY2;
+	case ENCODER_ID_INTERNAL_UNIPHY3:
+		return ENCODER_OBJECT_ID_INTERNAL_UNIPHY3;
+	case ENCODER_ID_INTERNAL_WIRELESS:
+		return ENCODER_OBJECT_ID_INTERNAL_VCE;
+	case ENCODER_ID_UNKNOWN:
+		return ENCODER_OBJECT_ID_NONE;
+	default:
+		/* Invalid encoder id */
+		BREAK_TO_DEBUGGER();
+		return ENCODER_OBJECT_ID_NONE;
+	}
+}
+
+uint8_t phy_id_to_atom(enum transmitter t)
+{
+	uint8_t atom_phy_id;
+
+	switch (t) {
+	case TRANSMITTER_UNIPHY_A:
+		atom_phy_id = ATOM_PHY_ID_UNIPHYA;
+		break;
+	case TRANSMITTER_UNIPHY_B:
+		atom_phy_id = ATOM_PHY_ID_UNIPHYB;
+		break;
+	case TRANSMITTER_UNIPHY_C:
+		atom_phy_id = ATOM_PHY_ID_UNIPHYC;
+		break;
+	case TRANSMITTER_UNIPHY_D:
+		atom_phy_id = ATOM_PHY_ID_UNIPHYD;
+		break;
+	case TRANSMITTER_UNIPHY_E:
+		atom_phy_id = ATOM_PHY_ID_UNIPHYE;
+		break;
+	case TRANSMITTER_UNIPHY_F:
+		atom_phy_id = ATOM_PHY_ID_UNIPHYF;
+		break;
+	case TRANSMITTER_UNIPHY_G:
+		atom_phy_id = ATOM_PHY_ID_UNIPHYG;
+		break;
+	default:
+		atom_phy_id = ATOM_PHY_ID_UNIPHYA;
+		break;
+	}
+	return atom_phy_id;
+}
+
+uint8_t clock_source_id_to_atom_phy_clk_src_id(
+		enum clock_source_id id)
+{
+	uint8_t atom_phy_clk_src_id = 0;
+
+	switch (id) {
+	case CLOCK_SOURCE_ID_PLL0:
+		atom_phy_clk_src_id = ATOM_TRANSMITTER_CONFIG_V5_P0PLL;
+		break;
+	case CLOCK_SOURCE_ID_PLL1:
+		atom_phy_clk_src_id = ATOM_TRANSMITTER_CONFIG_V5_P1PLL;
+		break;
+	case CLOCK_SOURCE_ID_PLL2:
+		atom_phy_clk_src_id = ATOM_TRANSMITTER_CONFIG_V5_P2PLL;
+		break;
+	case CLOCK_SOURCE_ID_EXTERNAL:
+		atom_phy_clk_src_id = ATOM_TRANSMITTER_CONFIG_V5_REFCLK_SRC_EXT;
+		break;
+	default:
+		atom_phy_clk_src_id = ATOM_TRANSMITTER_CONFIG_V5_P1PLL;
+		break;
+	}
+
+	return atom_phy_clk_src_id >> 2;
+}
+
+bool engine_bp_to_atom(enum engine_id id, uint32_t *atom_engine_id)
+{
+	bool result = false;
+
+	if (atom_engine_id != NULL)
+		switch (id) {
+		case ENGINE_ID_DIGA:
+			*atom_engine_id = ASIC_INT_DIG1_ENCODER_ID;
+			result = true;
+			break;
+		case ENGINE_ID_DIGB:
+			*atom_engine_id = ASIC_INT_DIG2_ENCODER_ID;
+			result = true;
+			break;
+		case ENGINE_ID_DIGC:
+			*atom_engine_id = ASIC_INT_DIG3_ENCODER_ID;
+			result = true;
+			break;
+		case ENGINE_ID_DIGD:
+			*atom_engine_id = ASIC_INT_DIG4_ENCODER_ID;
+			result = true;
+			break;
+		case ENGINE_ID_DIGE:
+			*atom_engine_id = ASIC_INT_DIG5_ENCODER_ID;
+			result = true;
+			break;
+		case ENGINE_ID_DIGF:
+			*atom_engine_id = ASIC_INT_DIG6_ENCODER_ID;
+			result = true;
+			break;
+		case ENGINE_ID_DIGG:
+			*atom_engine_id = ASIC_INT_DIG7_ENCODER_ID;
+			result = true;
+			break;
+		case ENGINE_ID_DACA:
+			*atom_engine_id = ASIC_INT_DAC1_ENCODER_ID;
+			result = true;
+			break;
+		default:
+			break;
+		}
+
+	return result;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/bios/command_table_helper.h b/drivers/gpu/drm/amd/display/dc/bios/command_table_helper.h
new file mode 100644
index 000000000000..547700e119a6
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/command_table_helper.h
@@ -0,0 +1,70 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_COMMAND_TABLE_HELPER_H__
+#define __DAL_COMMAND_TABLE_HELPER_H__
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+#include "dce60/command_table_helper_dce60.h"
+#endif
+#include "dce80/command_table_helper_dce80.h"
+#include "dce110/command_table_helper_dce110.h"
+#include "dce112/command_table_helper_dce112.h"
+#include "command_table_helper_struct.h"
+
+bool dal_bios_parser_init_cmd_tbl_helper(const struct command_table_helper **h,
+	enum dce_version dce);
+
+bool dal_cmd_table_helper_controller_id_to_atom(
+	enum controller_id id,
+	uint8_t *atom_id);
+
+uint32_t dal_cmd_table_helper_encoder_mode_bp_to_atom(
+	enum signal_type s,
+	bool enable_dp_audio);
+
+void dal_cmd_table_helper_assign_control_parameter(
+	const struct command_table_helper *h,
+	struct bp_encoder_control *control,
+DIG_ENCODER_CONTROL_PARAMETERS_V2 *ctrl_param);
+
+bool dal_cmd_table_helper_clock_source_id_to_ref_clk_src(
+	enum clock_source_id id,
+	uint32_t *ref_clk_src_id);
+
+uint8_t dal_cmd_table_helper_transmitter_bp_to_atom(
+	enum transmitter t);
+
+uint8_t dal_cmd_table_helper_encoder_id_to_atom(
+	enum encoder_id id);
+
+uint8_t phy_id_to_atom(enum transmitter t);
+
+uint8_t clock_source_id_to_atom_phy_clk_src_id(
+		enum clock_source_id id);
+
+bool engine_bp_to_atom(enum engine_id id, uint32_t *atom_engine_id);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/bios/command_table_helper2.c b/drivers/gpu/drm/amd/display/dc/bios/command_table_helper2.c
new file mode 100644
index 000000000000..268e2414b34f
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/command_table_helper2.c
@@ -0,0 +1,283 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+
+#include "ObjectID.h"
+#include "atomfirmware.h"
+
+#include "include/bios_parser_types.h"
+
+#include "command_table_helper2.h"
+
+bool dal_bios_parser_init_cmd_tbl_helper2(
+	const struct command_table_helper **h,
+	enum dce_version dce)
+{
+	switch (dce) {
+#if defined(CONFIG_DRM_AMD_DC_SI)
+	case DCE_VERSION_6_0:
+	case DCE_VERSION_6_1:
+	case DCE_VERSION_6_4:
+		*h = dal_cmd_tbl_helper_dce60_get_table();
+		return true;
+#endif
+
+	case DCE_VERSION_8_0:
+	case DCE_VERSION_8_1:
+	case DCE_VERSION_8_3:
+		*h = dal_cmd_tbl_helper_dce80_get_table();
+		return true;
+
+	case DCE_VERSION_10_0:
+		*h = dal_cmd_tbl_helper_dce110_get_table();
+		return true;
+
+	case DCE_VERSION_11_0:
+		*h = dal_cmd_tbl_helper_dce110_get_table();
+		return true;
+
+	case DCE_VERSION_11_2:
+	case DCE_VERSION_11_22:
+	case DCE_VERSION_12_0:
+	case DCE_VERSION_12_1:
+		*h = dal_cmd_tbl_helper_dce112_get_table2();
+		return true;
+	case DCN_VERSION_1_0:
+	case DCN_VERSION_1_01:
+	case DCN_VERSION_2_0:
+	case DCN_VERSION_2_1:
+	case DCN_VERSION_2_01:
+	case DCN_VERSION_3_0:
+	case DCN_VERSION_3_01:
+	case DCN_VERSION_3_02:
+	case DCN_VERSION_3_03:
+	case DCN_VERSION_3_1:
+	case DCN_VERSION_3_14:
+	case DCN_VERSION_3_15:
+	case DCN_VERSION_3_16:
+	case DCN_VERSION_3_2:
+	case DCN_VERSION_3_21:
+	case DCN_VERSION_3_5:
+	case DCN_VERSION_3_51:
+	case DCN_VERSION_3_6:
+	case DCN_VERSION_4_01:
+		*h = dal_cmd_tbl_helper_dce112_get_table2();
+		return true;
+
+	default:
+		*h = dal_cmd_tbl_helper_dce112_get_table2();
+		return false;
+	}
+}
+
+/* real implementations */
+
+bool dal_cmd_table_helper_controller_id_to_atom2(
+	enum controller_id id,
+	uint8_t *atom_id)
+{
+	if (atom_id == NULL) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+
+	switch (id) {
+	case CONTROLLER_ID_D0:
+		*atom_id = ATOM_CRTC1;
+		return true;
+	case CONTROLLER_ID_D1:
+		*atom_id = ATOM_CRTC2;
+		return true;
+	case CONTROLLER_ID_D2:
+		*atom_id = ATOM_CRTC3;
+		return true;
+	case CONTROLLER_ID_D3:
+		*atom_id = ATOM_CRTC4;
+		return true;
+	case CONTROLLER_ID_D4:
+		*atom_id = ATOM_CRTC5;
+		return true;
+	case CONTROLLER_ID_D5:
+		*atom_id = ATOM_CRTC6;
+		return true;
+	/* TODO :case CONTROLLER_ID_UNDERLAY0:
+		*atom_id = ATOM_UNDERLAY_PIPE0;
+		return true;
+	*/
+	case CONTROLLER_ID_UNDEFINED:
+		*atom_id = ATOM_CRTC_INVALID;
+		return true;
+	default:
+		/* Wrong controller id */
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+}
+
+/**
+ * dal_cmd_table_helper_transmitter_bp_to_atom2 - Translate the Transmitter to the
+ *                                     corresponding ATOM BIOS value
+ *  @t: transmitter
+ *  returns: digitalTransmitter
+ *    // =00: Digital Transmitter1 ( UNIPHY linkAB )
+ *    // =01: Digital Transmitter2 ( UNIPHY linkCD )
+ *    // =02: Digital Transmitter3 ( UNIPHY linkEF )
+ */
+uint8_t dal_cmd_table_helper_transmitter_bp_to_atom2(
+	enum transmitter t)
+{
+	switch (t) {
+	case TRANSMITTER_UNIPHY_A:
+	case TRANSMITTER_UNIPHY_B:
+	case TRANSMITTER_TRAVIS_LCD:
+		return 0;
+	case TRANSMITTER_UNIPHY_C:
+	case TRANSMITTER_UNIPHY_D:
+		return 1;
+	case TRANSMITTER_UNIPHY_E:
+	case TRANSMITTER_UNIPHY_F:
+		return 2;
+	default:
+		/* Invalid Transmitter Type! */
+		BREAK_TO_DEBUGGER();
+		return 0;
+	}
+}
+
+uint32_t dal_cmd_table_helper_encoder_mode_bp_to_atom2(
+	enum signal_type s,
+	bool enable_dp_audio)
+{
+	switch (s) {
+	case SIGNAL_TYPE_DVI_SINGLE_LINK:
+	case SIGNAL_TYPE_DVI_DUAL_LINK:
+		return ATOM_ENCODER_MODE_DVI;
+	case SIGNAL_TYPE_HDMI_TYPE_A:
+		return ATOM_ENCODER_MODE_HDMI;
+	case SIGNAL_TYPE_LVDS:
+		return ATOM_ENCODER_MODE_LVDS;
+	case SIGNAL_TYPE_EDP:
+	case SIGNAL_TYPE_DISPLAY_PORT_MST:
+	case SIGNAL_TYPE_DISPLAY_PORT:
+	case SIGNAL_TYPE_VIRTUAL:
+		if (enable_dp_audio)
+			return ATOM_ENCODER_MODE_DP_AUDIO;
+		else
+			return ATOM_ENCODER_MODE_DP;
+	case SIGNAL_TYPE_RGB:
+		return ATOM_ENCODER_MODE_CRT;
+	default:
+		return ATOM_ENCODER_MODE_CRT;
+	}
+}
+
+bool dal_cmd_table_helper_clock_source_id_to_ref_clk_src2(
+	enum clock_source_id id,
+	uint32_t *ref_clk_src_id)
+{
+	if (ref_clk_src_id == NULL) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+
+	switch (id) {
+	case CLOCK_SOURCE_ID_PLL1:
+		*ref_clk_src_id = ENCODER_REFCLK_SRC_P1PLL;
+		return true;
+	case CLOCK_SOURCE_ID_PLL2:
+		*ref_clk_src_id = ENCODER_REFCLK_SRC_P2PLL;
+		return true;
+	/*TODO:case CLOCK_SOURCE_ID_DCPLL:
+		*ref_clk_src_id = ENCODER_REFCLK_SRC_DCPLL;
+		return true;
+	*/
+	case CLOCK_SOURCE_ID_EXTERNAL:
+		*ref_clk_src_id = ENCODER_REFCLK_SRC_EXTCLK;
+		return true;
+	case CLOCK_SOURCE_ID_UNDEFINED:
+		*ref_clk_src_id = ENCODER_REFCLK_SRC_INVALID;
+		return true;
+	default:
+		/* Unsupported clock source id */
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+}
+
+uint8_t dal_cmd_table_helper_encoder_id_to_atom2(
+	enum encoder_id id)
+{
+	switch (id) {
+	case ENCODER_ID_INTERNAL_LVDS:
+		return ENCODER_OBJECT_ID_INTERNAL_LVDS;
+	case ENCODER_ID_INTERNAL_TMDS1:
+		return ENCODER_OBJECT_ID_INTERNAL_TMDS1;
+	case ENCODER_ID_INTERNAL_TMDS2:
+		return ENCODER_OBJECT_ID_INTERNAL_TMDS2;
+	case ENCODER_ID_INTERNAL_DAC1:
+		return ENCODER_OBJECT_ID_INTERNAL_DAC1;
+	case ENCODER_ID_INTERNAL_DAC2:
+		return ENCODER_OBJECT_ID_INTERNAL_DAC2;
+	case ENCODER_ID_INTERNAL_LVTM1:
+		return ENCODER_OBJECT_ID_INTERNAL_LVTM1;
+	case ENCODER_ID_INTERNAL_HDMI:
+		return ENCODER_OBJECT_ID_HDMI_INTERNAL;
+	case ENCODER_ID_EXTERNAL_TRAVIS:
+		return ENCODER_OBJECT_ID_TRAVIS;
+	case ENCODER_ID_EXTERNAL_NUTMEG:
+		return ENCODER_OBJECT_ID_NUTMEG;
+	case ENCODER_ID_INTERNAL_KLDSCP_TMDS1:
+		return ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1;
+	case ENCODER_ID_INTERNAL_KLDSCP_DAC1:
+		return ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1;
+	case ENCODER_ID_INTERNAL_KLDSCP_DAC2:
+		return ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2;
+	case ENCODER_ID_EXTERNAL_MVPU_FPGA:
+		return ENCODER_OBJECT_ID_MVPU_FPGA;
+	case ENCODER_ID_INTERNAL_DDI:
+		return ENCODER_OBJECT_ID_INTERNAL_DDI;
+	case ENCODER_ID_INTERNAL_UNIPHY:
+		return ENCODER_OBJECT_ID_INTERNAL_UNIPHY;
+	case ENCODER_ID_INTERNAL_KLDSCP_LVTMA:
+		return ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA;
+	case ENCODER_ID_INTERNAL_UNIPHY1:
+		return ENCODER_OBJECT_ID_INTERNAL_UNIPHY1;
+	case ENCODER_ID_INTERNAL_UNIPHY2:
+		return ENCODER_OBJECT_ID_INTERNAL_UNIPHY2;
+	case ENCODER_ID_INTERNAL_UNIPHY3:
+		return ENCODER_OBJECT_ID_INTERNAL_UNIPHY3;
+	case ENCODER_ID_INTERNAL_WIRELESS:
+		return ENCODER_OBJECT_ID_INTERNAL_VCE;
+	case ENCODER_ID_INTERNAL_VIRTUAL:
+		return ENCODER_OBJECT_ID_NONE;
+	case ENCODER_ID_UNKNOWN:
+		return ENCODER_OBJECT_ID_NONE;
+	default:
+		/* Invalid encoder id */
+		BREAK_TO_DEBUGGER();
+		return ENCODER_OBJECT_ID_NONE;
+	}
+}
diff --git a/drivers/gpu/drm/amd/display/dc/bios/command_table_helper2.h b/drivers/gpu/drm/amd/display/dc/bios/command_table_helper2.h
new file mode 100644
index 000000000000..66e0a3e73768
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/command_table_helper2.h
@@ -0,0 +1,57 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_COMMAND_TABLE_HELPER2_H__
+#define __DAL_COMMAND_TABLE_HELPER2_H__
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+#include "dce60/command_table_helper_dce60.h"
+#endif
+#include "dce80/command_table_helper_dce80.h"
+#include "dce110/command_table_helper_dce110.h"
+#include "dce112/command_table_helper2_dce112.h"
+#include "command_table_helper_struct.h"
+
+bool dal_bios_parser_init_cmd_tbl_helper2(const struct command_table_helper **h,
+	enum dce_version dce);
+
+bool dal_cmd_table_helper_controller_id_to_atom2(
+	enum controller_id id,
+	uint8_t *atom_id);
+
+uint32_t dal_cmd_table_helper_encoder_mode_bp_to_atom2(
+	enum signal_type s,
+	bool enable_dp_audio);
+
+bool dal_cmd_table_helper_clock_source_id_to_ref_clk_src2(
+	enum clock_source_id id,
+	uint32_t *ref_clk_src_id);
+
+uint8_t dal_cmd_table_helper_transmitter_bp_to_atom2(
+	enum transmitter t);
+
+uint8_t dal_cmd_table_helper_encoder_id_to_atom2(
+	enum encoder_id id);
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/bios/command_table_helper_struct.h b/drivers/gpu/drm/amd/display/dc/bios/command_table_helper_struct.h
new file mode 100644
index 000000000000..1f2c0a3f06f9
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/command_table_helper_struct.h
@@ -0,0 +1,66 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_COMMAND_TABLE_HELPER_STRUCT_H__
+#define __DAL_COMMAND_TABLE_HELPER_STRUCT_H__
+
+#include "dce80/command_table_helper_dce80.h"
+#include "dce110/command_table_helper_dce110.h"
+#include "dce112/command_table_helper_dce112.h"
+
+struct _DIG_ENCODER_CONTROL_PARAMETERS_V2;
+struct command_table_helper {
+	bool (*controller_id_to_atom)(enum controller_id id, uint8_t *atom_id);
+	uint8_t (*encoder_action_to_atom)(
+			enum bp_encoder_control_action action);
+	uint32_t (*encoder_mode_bp_to_atom)(enum signal_type s,
+			bool enable_dp_audio);
+	bool (*engine_bp_to_atom)(enum engine_id engine_id,
+			uint32_t *atom_engine_id);
+	void (*assign_control_parameter)(
+			const struct command_table_helper *h,
+			struct bp_encoder_control *control,
+			struct _DIG_ENCODER_CONTROL_PARAMETERS_V2 *ctrl_param);
+	bool (*clock_source_id_to_atom)(enum clock_source_id id,
+			uint32_t *atom_pll_id);
+	bool (*clock_source_id_to_ref_clk_src)(
+			enum clock_source_id id,
+			uint32_t *ref_clk_src_id);
+	uint8_t (*transmitter_bp_to_atom)(enum transmitter t);
+	uint8_t (*encoder_id_to_atom)(enum encoder_id id);
+	uint8_t (*clock_source_id_to_atom_phy_clk_src_id)(
+			enum clock_source_id id);
+	uint8_t (*signal_type_to_atom_dig_mode)(enum signal_type s);
+	uint8_t (*hpd_sel_to_atom)(enum hpd_source_id id);
+	uint8_t (*dig_encoder_sel_to_atom)(enum engine_id engine_id);
+	uint8_t (*phy_id_to_atom)(enum transmitter t);
+	uint8_t (*disp_power_gating_action_to_atom)(
+			enum bp_pipe_control_action action);
+	bool (*dc_clock_type_to_atom)(enum bp_dce_clock_type id,
+			uint32_t *atom_clock_type);
+	uint8_t (*transmitter_color_depth_to_atom)(enum transmitter_color_depth id);
+};
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/bios/dce110/command_table_helper_dce110.c b/drivers/gpu/drm/amd/display/dc/bios/dce110/command_table_helper_dce110.c
new file mode 100644
index 000000000000..3099128223df
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/dce110/command_table_helper_dce110.c
@@ -0,0 +1,232 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+
+#include "atom.h"
+
+#include "include/bios_parser_types.h"
+
+#include "../command_table_helper.h"
+
+static uint8_t signal_type_to_atom_dig_mode(enum signal_type s)
+{
+	uint8_t atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V5_DP;
+
+	switch (s) {
+	case SIGNAL_TYPE_DISPLAY_PORT:
+	case SIGNAL_TYPE_EDP:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V5_DP;
+		break;
+	case SIGNAL_TYPE_LVDS:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V5_LVDS;
+		break;
+	case SIGNAL_TYPE_DVI_SINGLE_LINK:
+	case SIGNAL_TYPE_DVI_DUAL_LINK:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V5_DVI;
+		break;
+	case SIGNAL_TYPE_HDMI_TYPE_A:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V5_HDMI;
+		break;
+	case SIGNAL_TYPE_DISPLAY_PORT_MST:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V5_DP_MST;
+		break;
+	default:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V5_DVI;
+		break;
+	}
+
+	return atom_dig_mode;
+}
+
+static uint8_t hpd_sel_to_atom(enum hpd_source_id id)
+{
+	uint8_t atom_hpd_sel = 0;
+
+	switch (id) {
+	case HPD_SOURCEID1:
+		atom_hpd_sel = ATOM_TRANSMITTER_CONFIG_V5_HPD1_SEL;
+		break;
+	case HPD_SOURCEID2:
+		atom_hpd_sel = ATOM_TRANSMITTER_CONFIG_V5_HPD2_SEL;
+		break;
+	case HPD_SOURCEID3:
+		atom_hpd_sel = ATOM_TRANSMITTER_CONFIG_V5_HPD3_SEL;
+		break;
+	case HPD_SOURCEID4:
+		atom_hpd_sel = ATOM_TRANSMITTER_CONFIG_V5_HPD4_SEL;
+		break;
+	case HPD_SOURCEID5:
+		atom_hpd_sel = ATOM_TRANSMITTER_CONFIG_V5_HPD5_SEL;
+		break;
+	case HPD_SOURCEID6:
+		atom_hpd_sel = ATOM_TRANSMITTER_CONFIG_V5_HPD6_SEL;
+		break;
+	case HPD_SOURCEID_UNKNOWN:
+	default:
+		atom_hpd_sel = 0;
+		break;
+	}
+	return atom_hpd_sel >> 4;
+}
+
+static uint8_t dig_encoder_sel_to_atom(enum engine_id id)
+{
+	/* On any ASIC after DCE80, we manually program the DIG_FE
+	 * selection (see connect_dig_be_to_fe function of the link
+	 * encoder), so translation should always return 0 (no FE).
+	 */
+
+	return 0;
+}
+
+static bool clock_source_id_to_atom(
+	enum clock_source_id id,
+	uint32_t *atom_pll_id)
+{
+	bool result = true;
+
+	if (atom_pll_id != NULL)
+		switch (id) {
+		case CLOCK_SOURCE_ID_PLL0:
+			*atom_pll_id = ATOM_PPLL0;
+			break;
+		case CLOCK_SOURCE_ID_PLL1:
+			*atom_pll_id = ATOM_PPLL1;
+			break;
+		case CLOCK_SOURCE_ID_PLL2:
+			*atom_pll_id = ATOM_PPLL2;
+			break;
+		case CLOCK_SOURCE_ID_EXTERNAL:
+			*atom_pll_id = ATOM_PPLL_INVALID;
+			break;
+		case CLOCK_SOURCE_ID_DFS:
+			*atom_pll_id = ATOM_EXT_PLL1;
+			break;
+		case CLOCK_SOURCE_ID_VCE:
+			/* for VCE encoding,
+			 * we need to pass in ATOM_PPLL_INVALID
+			 */
+			*atom_pll_id = ATOM_PPLL_INVALID;
+			break;
+		case CLOCK_SOURCE_ID_DP_DTO:
+			/* When programming DP DTO PLL ID should be invalid */
+			*atom_pll_id = ATOM_PPLL_INVALID;
+			break;
+		case CLOCK_SOURCE_ID_UNDEFINED:
+			/* Should not happen */
+			*atom_pll_id = ATOM_PPLL_INVALID;
+			result = false;
+			break;
+		default:
+			result = false;
+			break;
+		}
+
+	return result;
+}
+
+static uint8_t encoder_action_to_atom(enum bp_encoder_control_action action)
+{
+	uint8_t atom_action = 0;
+
+	switch (action) {
+	case ENCODER_CONTROL_ENABLE:
+		atom_action = ATOM_ENABLE;
+		break;
+	case ENCODER_CONTROL_DISABLE:
+		atom_action = ATOM_DISABLE;
+		break;
+	case ENCODER_CONTROL_SETUP:
+		atom_action = ATOM_ENCODER_CMD_SETUP;
+		break;
+	case ENCODER_CONTROL_INIT:
+		atom_action = ATOM_ENCODER_INIT;
+		break;
+	default:
+		BREAK_TO_DEBUGGER(); /* Unhandle action in driver.!! */
+		break;
+	}
+
+	return atom_action;
+}
+
+static uint8_t disp_power_gating_action_to_atom(
+	enum bp_pipe_control_action action)
+{
+	uint8_t atom_pipe_action = 0;
+
+	switch (action) {
+	case ASIC_PIPE_DISABLE:
+		atom_pipe_action = ATOM_DISABLE;
+		break;
+	case ASIC_PIPE_ENABLE:
+		atom_pipe_action = ATOM_ENABLE;
+		break;
+	case ASIC_PIPE_INIT:
+		atom_pipe_action = ATOM_INIT;
+		break;
+	default:
+		ASSERT_CRITICAL(false); /* Unhandle action in driver! */
+		break;
+	}
+
+	return atom_pipe_action;
+}
+
+/* function table */
+static const struct command_table_helper command_table_helper_funcs = {
+	.controller_id_to_atom = dal_cmd_table_helper_controller_id_to_atom,
+	.encoder_action_to_atom = encoder_action_to_atom,
+	.engine_bp_to_atom = engine_bp_to_atom,
+	.clock_source_id_to_atom = clock_source_id_to_atom,
+	.clock_source_id_to_atom_phy_clk_src_id =
+			clock_source_id_to_atom_phy_clk_src_id,
+	.signal_type_to_atom_dig_mode = signal_type_to_atom_dig_mode,
+	.hpd_sel_to_atom = hpd_sel_to_atom,
+	.dig_encoder_sel_to_atom = dig_encoder_sel_to_atom,
+	.phy_id_to_atom = phy_id_to_atom,
+	.disp_power_gating_action_to_atom = disp_power_gating_action_to_atom,
+	.assign_control_parameter = NULL,
+	.clock_source_id_to_ref_clk_src = NULL,
+	.transmitter_bp_to_atom = NULL,
+	.encoder_id_to_atom = dal_cmd_table_helper_encoder_id_to_atom,
+	.encoder_mode_bp_to_atom = dal_cmd_table_helper_encoder_mode_bp_to_atom,
+};
+
+/*
+ * dal_cmd_tbl_helper_dce110_get_table
+ *
+ * @brief
+ * Initialize command table helper functions
+ *
+ * @param
+ * const struct command_table_helper **h - [out] struct of functions
+ *
+ */
+const struct command_table_helper *dal_cmd_tbl_helper_dce110_get_table(void)
+{
+	return &command_table_helper_funcs;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/bios/dce110/command_table_helper_dce110.h b/drivers/gpu/drm/amd/display/dc/bios/dce110/command_table_helper_dce110.h
new file mode 100644
index 000000000000..eb60c2ead992
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/dce110/command_table_helper_dce110.h
@@ -0,0 +1,34 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_COMMAND_TABLE_HELPER_DCE110_H__
+#define __DAL_COMMAND_TABLE_HELPER_DCE110_H__
+
+struct command_table_helper;
+
+/* Initialize command table helper functions */
+const struct command_table_helper *dal_cmd_tbl_helper_dce110_get_table(void);
+
+#endif /* __DAL_COMMAND_TABLE_HELPER_DCE110_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/bios/dce112/command_table_helper2_dce112.c b/drivers/gpu/drm/amd/display/dc/bios/dce112/command_table_helper2_dce112.c
new file mode 100644
index 000000000000..349f0e5d5856
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/dce112/command_table_helper2_dce112.c
@@ -0,0 +1,288 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+
+#include "atom.h"
+
+#include "include/bios_parser_types.h"
+
+#include "../command_table_helper.h"
+
+#include "../command_table_helper2.h"
+
+static uint8_t signal_type_to_atom_dig_mode(enum signal_type s)
+{
+	uint8_t atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V6_DP;
+
+	switch (s) {
+	case SIGNAL_TYPE_DISPLAY_PORT:
+	case SIGNAL_TYPE_EDP:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V6_DP;
+		break;
+	case SIGNAL_TYPE_DVI_SINGLE_LINK:
+	case SIGNAL_TYPE_DVI_DUAL_LINK:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V6_DVI;
+		break;
+	case SIGNAL_TYPE_HDMI_TYPE_A:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V6_HDMI;
+		break;
+	case SIGNAL_TYPE_DISPLAY_PORT_MST:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V6_DP_MST;
+		break;
+	default:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V6_DVI;
+		break;
+	}
+
+	return atom_dig_mode;
+}
+
+static uint8_t hpd_sel_to_atom(enum hpd_source_id id)
+{
+	uint8_t atom_hpd_sel = 0;
+
+	switch (id) {
+	case HPD_SOURCEID1:
+		atom_hpd_sel = ATOM_TRANSMITTER_V6_HPD1_SEL;
+		break;
+	case HPD_SOURCEID2:
+		atom_hpd_sel = ATOM_TRANSMITTER_V6_HPD2_SEL;
+		break;
+	case HPD_SOURCEID3:
+		atom_hpd_sel = ATOM_TRANSMITTER_V6_HPD3_SEL;
+		break;
+	case HPD_SOURCEID4:
+		atom_hpd_sel = ATOM_TRANSMITTER_V6_HPD4_SEL;
+		break;
+	case HPD_SOURCEID5:
+		atom_hpd_sel = ATOM_TRANSMITTER_V6_HPD5_SEL;
+		break;
+	case HPD_SOURCEID6:
+		atom_hpd_sel = ATOM_TRANSMITTER_V6_HPD6_SEL;
+		break;
+	case HPD_SOURCEID_UNKNOWN:
+	default:
+		atom_hpd_sel = 0;
+		break;
+	}
+	return atom_hpd_sel;
+}
+
+static uint8_t dig_encoder_sel_to_atom(enum engine_id id)
+{
+	/* On any ASIC after DCE80, we manually program the DIG_FE
+	 * selection (see connect_dig_be_to_fe function of the link
+	 * encoder), so translation should always return 0 (no FE).
+	 */
+
+	return 0;
+}
+
+static bool clock_source_id_to_atom(
+	enum clock_source_id id,
+	uint32_t *atom_pll_id)
+{
+	bool result = true;
+
+	if (atom_pll_id != NULL)
+		switch (id) {
+		case CLOCK_SOURCE_COMBO_PHY_PLL0:
+			*atom_pll_id = ATOM_COMBOPHY_PLL0;
+			break;
+		case CLOCK_SOURCE_COMBO_PHY_PLL1:
+			*atom_pll_id = ATOM_COMBOPHY_PLL1;
+			break;
+		case CLOCK_SOURCE_COMBO_PHY_PLL2:
+			*atom_pll_id = ATOM_COMBOPHY_PLL2;
+			break;
+		case CLOCK_SOURCE_COMBO_PHY_PLL3:
+			*atom_pll_id = ATOM_COMBOPHY_PLL3;
+			break;
+		case CLOCK_SOURCE_COMBO_PHY_PLL4:
+			*atom_pll_id = ATOM_COMBOPHY_PLL4;
+			break;
+		case CLOCK_SOURCE_COMBO_PHY_PLL5:
+			*atom_pll_id = ATOM_COMBOPHY_PLL5;
+			break;
+		case CLOCK_SOURCE_COMBO_DISPLAY_PLL0:
+			*atom_pll_id = ATOM_PPLL0;
+			break;
+		case CLOCK_SOURCE_ID_DFS:
+			*atom_pll_id = ATOM_GCK_DFS;
+			break;
+		case CLOCK_SOURCE_ID_VCE:
+			*atom_pll_id = ATOM_DP_DTO;
+			break;
+		case CLOCK_SOURCE_ID_DP_DTO:
+			*atom_pll_id = ATOM_DP_DTO;
+			break;
+		case CLOCK_SOURCE_ID_UNDEFINED:
+			/* Should not happen */
+			*atom_pll_id = ATOM_PPLL_INVALID;
+			result = false;
+			break;
+		default:
+			result = false;
+			break;
+		}
+
+	return result;
+}
+
+static uint8_t encoder_action_to_atom(enum bp_encoder_control_action action)
+{
+	uint8_t atom_action = 0;
+
+	switch (action) {
+	case ENCODER_CONTROL_ENABLE:
+		atom_action = ATOM_ENABLE;
+		break;
+	case ENCODER_CONTROL_DISABLE:
+		atom_action = ATOM_DISABLE;
+		break;
+	case ENCODER_CONTROL_SETUP:
+		atom_action = ATOM_ENCODER_CMD_STREAM_SETUP;
+		break;
+	case ENCODER_CONTROL_INIT:
+		atom_action = ATOM_ENCODER_INIT;
+		break;
+	default:
+		BREAK_TO_DEBUGGER(); /* Unhandle action in driver.!! */
+		break;
+	}
+
+	return atom_action;
+}
+
+static uint8_t disp_power_gating_action_to_atom(
+	enum bp_pipe_control_action action)
+{
+	uint8_t atom_pipe_action = 0;
+
+	switch (action) {
+	case ASIC_PIPE_DISABLE:
+		atom_pipe_action = ATOM_DISABLE;
+		break;
+	case ASIC_PIPE_ENABLE:
+		atom_pipe_action = ATOM_ENABLE;
+		break;
+	case ASIC_PIPE_INIT:
+		atom_pipe_action = ATOM_INIT;
+		break;
+	default:
+		ASSERT_CRITICAL(false); /* Unhandle action in driver! */
+		break;
+	}
+
+	return atom_pipe_action;
+}
+
+static bool dc_clock_type_to_atom(
+		enum bp_dce_clock_type id,
+		uint32_t *atom_clock_type)
+{
+	bool retCode = true;
+
+	if (atom_clock_type != NULL) {
+		switch (id) {
+		case DCECLOCK_TYPE_DISPLAY_CLOCK:
+			*atom_clock_type = DCE_CLOCK_TYPE_DISPCLK;
+			break;
+
+		case DCECLOCK_TYPE_DPREFCLK:
+			*atom_clock_type = DCE_CLOCK_TYPE_DPREFCLK;
+			break;
+
+		default:
+			ASSERT_CRITICAL(false); /* Unhandle action in driver! */
+			break;
+		}
+	}
+
+	return retCode;
+}
+
+static uint8_t transmitter_color_depth_to_atom(enum transmitter_color_depth id)
+{
+	uint8_t atomColorDepth = 0;
+
+	switch (id) {
+	case TRANSMITTER_COLOR_DEPTH_24:
+		atomColorDepth = PIXEL_CLOCK_V7_DEEPCOLOR_RATIO_DIS;
+		break;
+	case TRANSMITTER_COLOR_DEPTH_30:
+		atomColorDepth = PIXEL_CLOCK_V7_DEEPCOLOR_RATIO_5_4;
+		break;
+	case TRANSMITTER_COLOR_DEPTH_36:
+		atomColorDepth = PIXEL_CLOCK_V7_DEEPCOLOR_RATIO_3_2;
+		break;
+	case TRANSMITTER_COLOR_DEPTH_48:
+		atomColorDepth = PIXEL_CLOCK_V7_DEEPCOLOR_RATIO_2_1;
+		break;
+	default:
+		ASSERT_CRITICAL(false); /* Unhandle action in driver! */
+		break;
+	}
+
+	return atomColorDepth;
+}
+
+/* function table */
+static const struct command_table_helper command_table_helper_funcs = {
+	.controller_id_to_atom = dal_cmd_table_helper_controller_id_to_atom2,
+	.encoder_action_to_atom = encoder_action_to_atom,
+	.engine_bp_to_atom = engine_bp_to_atom,
+	.clock_source_id_to_atom = clock_source_id_to_atom,
+	.clock_source_id_to_atom_phy_clk_src_id =
+			clock_source_id_to_atom_phy_clk_src_id,
+	.signal_type_to_atom_dig_mode = signal_type_to_atom_dig_mode,
+	.hpd_sel_to_atom = hpd_sel_to_atom,
+	.dig_encoder_sel_to_atom = dig_encoder_sel_to_atom,
+	.phy_id_to_atom = phy_id_to_atom,
+	.disp_power_gating_action_to_atom = disp_power_gating_action_to_atom,
+	.clock_source_id_to_ref_clk_src = NULL,
+	.transmitter_bp_to_atom = NULL,
+	.encoder_id_to_atom = dal_cmd_table_helper_encoder_id_to_atom2,
+	.encoder_mode_bp_to_atom =
+			dal_cmd_table_helper_encoder_mode_bp_to_atom2,
+	.dc_clock_type_to_atom = dc_clock_type_to_atom,
+	.transmitter_color_depth_to_atom = transmitter_color_depth_to_atom,
+};
+
+/*
+ * dal_cmd_tbl_helper_dce110_get_table
+ *
+ * @brief
+ * Initialize command table helper functions
+ *
+ * @param
+ * const struct command_table_helper **h - [out] struct of functions
+ *
+ */
+const struct command_table_helper *dal_cmd_tbl_helper_dce112_get_table2(void)
+{
+	return &command_table_helper_funcs;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/bios/dce112/command_table_helper2_dce112.h b/drivers/gpu/drm/amd/display/dc/bios/dce112/command_table_helper2_dce112.h
new file mode 100644
index 000000000000..abf28a06f5bc
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/dce112/command_table_helper2_dce112.h
@@ -0,0 +1,34 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_COMMAND_TABLE_HELPER2_DCE112_H__
+#define __DAL_COMMAND_TABLE_HELPER2_DCE112_H__
+
+struct command_table_helper;
+
+/* Initialize command table helper functions */
+const struct command_table_helper *dal_cmd_tbl_helper_dce112_get_table2(void);
+
+#endif /* __DAL_COMMAND_TABLE_HELPER_DCE110_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/bios/dce112/command_table_helper_dce112.c b/drivers/gpu/drm/amd/display/dc/bios/dce112/command_table_helper_dce112.c
new file mode 100644
index 000000000000..1a5fefcde8af
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/dce112/command_table_helper_dce112.c
@@ -0,0 +1,286 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+
+#include "atom.h"
+
+#include "include/bios_parser_types.h"
+
+#include "../command_table_helper.h"
+
+static uint8_t signal_type_to_atom_dig_mode(enum signal_type s)
+{
+	uint8_t atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V6_DP;
+
+	switch (s) {
+	case SIGNAL_TYPE_DISPLAY_PORT:
+	case SIGNAL_TYPE_EDP:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V6_DP;
+		break;
+	case SIGNAL_TYPE_DVI_SINGLE_LINK:
+	case SIGNAL_TYPE_DVI_DUAL_LINK:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V6_DVI;
+		break;
+	case SIGNAL_TYPE_HDMI_TYPE_A:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V6_HDMI;
+		break;
+	case SIGNAL_TYPE_DISPLAY_PORT_MST:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V6_DP_MST;
+		break;
+	default:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V6_DVI;
+		break;
+	}
+
+	return atom_dig_mode;
+}
+
+static uint8_t hpd_sel_to_atom(enum hpd_source_id id)
+{
+	uint8_t atom_hpd_sel = 0;
+
+	switch (id) {
+	case HPD_SOURCEID1:
+		atom_hpd_sel = ATOM_TRANSMITTER_V6_HPD1_SEL;
+		break;
+	case HPD_SOURCEID2:
+		atom_hpd_sel = ATOM_TRANSMITTER_V6_HPD2_SEL;
+		break;
+	case HPD_SOURCEID3:
+		atom_hpd_sel = ATOM_TRANSMITTER_V6_HPD3_SEL;
+		break;
+	case HPD_SOURCEID4:
+		atom_hpd_sel = ATOM_TRANSMITTER_V6_HPD4_SEL;
+		break;
+	case HPD_SOURCEID5:
+		atom_hpd_sel = ATOM_TRANSMITTER_V6_HPD5_SEL;
+		break;
+	case HPD_SOURCEID6:
+		atom_hpd_sel = ATOM_TRANSMITTER_V6_HPD6_SEL;
+		break;
+	case HPD_SOURCEID_UNKNOWN:
+	default:
+		atom_hpd_sel = 0;
+		break;
+	}
+	return atom_hpd_sel;
+}
+
+static uint8_t dig_encoder_sel_to_atom(enum engine_id id)
+{
+	/* On any ASIC after DCE80, we manually program the DIG_FE
+	 * selection (see connect_dig_be_to_fe function of the link
+	 * encoder), so translation should always return 0 (no FE).
+	 */
+
+	return 0;
+}
+
+static bool clock_source_id_to_atom(
+	enum clock_source_id id,
+	uint32_t *atom_pll_id)
+{
+	bool result = true;
+
+	if (atom_pll_id != NULL)
+		switch (id) {
+		case CLOCK_SOURCE_COMBO_PHY_PLL0:
+			*atom_pll_id = ATOM_COMBOPHY_PLL0;
+			break;
+		case CLOCK_SOURCE_COMBO_PHY_PLL1:
+			*atom_pll_id = ATOM_COMBOPHY_PLL1;
+			break;
+		case CLOCK_SOURCE_COMBO_PHY_PLL2:
+			*atom_pll_id = ATOM_COMBOPHY_PLL2;
+			break;
+		case CLOCK_SOURCE_COMBO_PHY_PLL3:
+			*atom_pll_id = ATOM_COMBOPHY_PLL3;
+			break;
+		case CLOCK_SOURCE_COMBO_PHY_PLL4:
+			*atom_pll_id = ATOM_COMBOPHY_PLL4;
+			break;
+		case CLOCK_SOURCE_COMBO_PHY_PLL5:
+			*atom_pll_id = ATOM_COMBOPHY_PLL5;
+			break;
+		case CLOCK_SOURCE_COMBO_DISPLAY_PLL0:
+			*atom_pll_id = ATOM_PPLL0;
+			break;
+		case CLOCK_SOURCE_ID_DFS:
+			*atom_pll_id = ATOM_GCK_DFS;
+			break;
+		case CLOCK_SOURCE_ID_VCE:
+			*atom_pll_id = ATOM_DP_DTO;
+			break;
+		case CLOCK_SOURCE_ID_DP_DTO:
+			*atom_pll_id = ATOM_DP_DTO;
+			break;
+		case CLOCK_SOURCE_ID_UNDEFINED:
+			/* Should not happen */
+			*atom_pll_id = ATOM_PPLL_INVALID;
+			result = false;
+			break;
+		default:
+			result = false;
+			break;
+		}
+
+	return result;
+}
+
+static uint8_t encoder_action_to_atom(enum bp_encoder_control_action action)
+{
+	uint8_t atom_action = 0;
+
+	switch (action) {
+	case ENCODER_CONTROL_ENABLE:
+		atom_action = ATOM_ENABLE;
+		break;
+	case ENCODER_CONTROL_DISABLE:
+		atom_action = ATOM_DISABLE;
+		break;
+	case ENCODER_CONTROL_SETUP:
+		atom_action = ATOM_ENCODER_CMD_STREAM_SETUP;
+		break;
+	case ENCODER_CONTROL_INIT:
+		atom_action = ATOM_ENCODER_INIT;
+		break;
+	default:
+		BREAK_TO_DEBUGGER(); /* Unhandle action in driver.!! */
+		break;
+	}
+
+	return atom_action;
+}
+
+static uint8_t disp_power_gating_action_to_atom(
+	enum bp_pipe_control_action action)
+{
+	uint8_t atom_pipe_action = 0;
+
+	switch (action) {
+	case ASIC_PIPE_DISABLE:
+		atom_pipe_action = ATOM_DISABLE;
+		break;
+	case ASIC_PIPE_ENABLE:
+		atom_pipe_action = ATOM_ENABLE;
+		break;
+	case ASIC_PIPE_INIT:
+		atom_pipe_action = ATOM_INIT;
+		break;
+	default:
+		ASSERT_CRITICAL(false); /* Unhandle action in driver! */
+		break;
+	}
+
+	return atom_pipe_action;
+}
+
+static bool dc_clock_type_to_atom(
+		enum bp_dce_clock_type id,
+		uint32_t *atom_clock_type)
+{
+	bool retCode = true;
+
+	if (atom_clock_type != NULL) {
+		switch (id) {
+		case DCECLOCK_TYPE_DISPLAY_CLOCK:
+			*atom_clock_type = DCE_CLOCK_TYPE_DISPCLK;
+			break;
+
+		case DCECLOCK_TYPE_DPREFCLK:
+			*atom_clock_type = DCE_CLOCK_TYPE_DPREFCLK;
+			break;
+
+		default:
+			ASSERT_CRITICAL(false); /* Unhandle action in driver! */
+			break;
+		}
+	}
+
+	return retCode;
+}
+
+static uint8_t transmitter_color_depth_to_atom(enum transmitter_color_depth id)
+{
+	uint8_t atomColorDepth = 0;
+
+	switch (id) {
+	case TRANSMITTER_COLOR_DEPTH_24:
+		atomColorDepth = PIXEL_CLOCK_V7_DEEPCOLOR_RATIO_DIS;
+		break;
+	case TRANSMITTER_COLOR_DEPTH_30:
+		atomColorDepth = PIXEL_CLOCK_V7_DEEPCOLOR_RATIO_5_4;
+		break;
+	case TRANSMITTER_COLOR_DEPTH_36:
+		atomColorDepth = PIXEL_CLOCK_V7_DEEPCOLOR_RATIO_3_2;
+		break;
+	case TRANSMITTER_COLOR_DEPTH_48:
+		atomColorDepth = PIXEL_CLOCK_V7_DEEPCOLOR_RATIO_2_1;
+		break;
+	default:
+		ASSERT_CRITICAL(false); /* Unhandle action in driver! */
+		break;
+	}
+
+	return atomColorDepth;
+}
+
+/* function table */
+static const struct command_table_helper command_table_helper_funcs = {
+	.controller_id_to_atom = dal_cmd_table_helper_controller_id_to_atom,
+	.encoder_action_to_atom = encoder_action_to_atom,
+	.engine_bp_to_atom = engine_bp_to_atom,
+	.clock_source_id_to_atom = clock_source_id_to_atom,
+	.clock_source_id_to_atom_phy_clk_src_id =
+			clock_source_id_to_atom_phy_clk_src_id,
+	.signal_type_to_atom_dig_mode = signal_type_to_atom_dig_mode,
+	.hpd_sel_to_atom = hpd_sel_to_atom,
+	.dig_encoder_sel_to_atom = dig_encoder_sel_to_atom,
+	.phy_id_to_atom = phy_id_to_atom,
+	.disp_power_gating_action_to_atom = disp_power_gating_action_to_atom,
+	.assign_control_parameter = NULL,
+	.clock_source_id_to_ref_clk_src = NULL,
+	.transmitter_bp_to_atom = NULL,
+	.encoder_id_to_atom = dal_cmd_table_helper_encoder_id_to_atom,
+	.encoder_mode_bp_to_atom = dal_cmd_table_helper_encoder_mode_bp_to_atom,
+	.dc_clock_type_to_atom = dc_clock_type_to_atom,
+	.transmitter_color_depth_to_atom = transmitter_color_depth_to_atom,
+};
+
+/*
+ * dal_cmd_tbl_helper_dce110_get_table
+ *
+ * @brief
+ * Initialize command table helper functions
+ *
+ * @param
+ * const struct command_table_helper **h - [out] struct of functions
+ *
+ */
+const struct command_table_helper *dal_cmd_tbl_helper_dce112_get_table(void)
+{
+	return &command_table_helper_funcs;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/bios/dce112/command_table_helper_dce112.h b/drivers/gpu/drm/amd/display/dc/bios/dce112/command_table_helper_dce112.h
new file mode 100644
index 000000000000..dc3660951355
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/dce112/command_table_helper_dce112.h
@@ -0,0 +1,34 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_COMMAND_TABLE_HELPER_DCE112_H__
+#define __DAL_COMMAND_TABLE_HELPER_DCE112_H__
+
+struct command_table_helper;
+
+/* Initialize command table helper functions */
+const struct command_table_helper *dal_cmd_tbl_helper_dce112_get_table(void);
+
+#endif /* __DAL_COMMAND_TABLE_HELPER_DCE110_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/bios/dce60/command_table_helper_dce60.c b/drivers/gpu/drm/amd/display/dc/bios/dce60/command_table_helper_dce60.c
new file mode 100644
index 000000000000..01ccc803040c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/dce60/command_table_helper_dce60.c
@@ -0,0 +1,250 @@
+/*
+ * Copyright 2020 Mauro Rossi <issor.oruam@gmail.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+
+#include "atom.h"
+
+#include "include/grph_object_id.h"
+#include "include/grph_object_defs.h"
+#include "include/bios_parser_types.h"
+
+#include "../command_table_helper.h"
+
+static uint8_t encoder_action_to_atom(enum bp_encoder_control_action action)
+{
+	uint8_t atom_action = 0;
+
+	switch (action) {
+	case ENCODER_CONTROL_ENABLE:
+		atom_action = ATOM_ENABLE;
+		break;
+	case ENCODER_CONTROL_DISABLE:
+		atom_action = ATOM_DISABLE;
+		break;
+	case ENCODER_CONTROL_SETUP:
+		atom_action = ATOM_ENCODER_CMD_SETUP;
+		break;
+	case ENCODER_CONTROL_INIT:
+		atom_action = ATOM_ENCODER_INIT;
+		break;
+	default:
+		BREAK_TO_DEBUGGER(); /* Unhandle action in driver.!! */
+		break;
+	}
+
+	return atom_action;
+}
+
+static bool clock_source_id_to_atom(
+	enum clock_source_id id,
+	uint32_t *atom_pll_id)
+{
+	bool result = true;
+
+	if (atom_pll_id != NULL)
+		switch (id) {
+		case CLOCK_SOURCE_ID_PLL0:
+			*atom_pll_id = ATOM_PPLL0;
+			break;
+		case CLOCK_SOURCE_ID_PLL1:
+			*atom_pll_id = ATOM_PPLL1;
+			break;
+		case CLOCK_SOURCE_ID_PLL2:
+			*atom_pll_id = ATOM_PPLL2;
+			break;
+		case CLOCK_SOURCE_ID_EXTERNAL:
+			*atom_pll_id = ATOM_PPLL_INVALID;
+			break;
+		case CLOCK_SOURCE_ID_DFS:
+			*atom_pll_id = ATOM_EXT_PLL1;
+			break;
+		case CLOCK_SOURCE_ID_VCE:
+			/* for VCE encoding,
+			 * we need to pass in ATOM_PPLL_INVALID
+			 */
+			*atom_pll_id = ATOM_PPLL_INVALID;
+			break;
+		case CLOCK_SOURCE_ID_DP_DTO:
+			/* When programming DP DTO PLL ID should be invalid */
+			*atom_pll_id = ATOM_PPLL_INVALID;
+			break;
+		case CLOCK_SOURCE_ID_UNDEFINED:
+			BREAK_TO_DEBUGGER(); /* check when this will happen! */
+			*atom_pll_id = ATOM_PPLL_INVALID;
+			result = false;
+			break;
+		default:
+			result = false;
+			break;
+		}
+
+	return result;
+}
+
+static uint8_t signal_type_to_atom_dig_mode(enum signal_type s)
+{
+	uint8_t atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V5_DP;
+
+	switch (s) {
+	case SIGNAL_TYPE_DISPLAY_PORT:
+	case SIGNAL_TYPE_EDP:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V5_DP;
+		break;
+	case SIGNAL_TYPE_LVDS:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V5_LVDS;
+		break;
+	case SIGNAL_TYPE_DVI_SINGLE_LINK:
+	case SIGNAL_TYPE_DVI_DUAL_LINK:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V5_DVI;
+		break;
+	case SIGNAL_TYPE_HDMI_TYPE_A:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V5_HDMI;
+		break;
+	case SIGNAL_TYPE_DISPLAY_PORT_MST:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V5_DP_MST;
+		break;
+	default:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V5_DVI;
+		break;
+	}
+
+	return atom_dig_mode;
+}
+
+static uint8_t hpd_sel_to_atom(enum hpd_source_id id)
+{
+	uint8_t atom_hpd_sel = 0;
+
+	switch (id) {
+	case HPD_SOURCEID1:
+		atom_hpd_sel = ATOM_TRANSMITTER_CONFIG_V5_HPD1_SEL;
+		break;
+	case HPD_SOURCEID2:
+		atom_hpd_sel = ATOM_TRANSMITTER_CONFIG_V5_HPD2_SEL;
+		break;
+	case HPD_SOURCEID3:
+		atom_hpd_sel = ATOM_TRANSMITTER_CONFIG_V5_HPD3_SEL;
+		break;
+	case HPD_SOURCEID4:
+		atom_hpd_sel = ATOM_TRANSMITTER_CONFIG_V5_HPD4_SEL;
+		break;
+	case HPD_SOURCEID5:
+		atom_hpd_sel = ATOM_TRANSMITTER_CONFIG_V5_HPD5_SEL;
+		break;
+	case HPD_SOURCEID6:
+		atom_hpd_sel = ATOM_TRANSMITTER_CONFIG_V5_HPD6_SEL;
+		break;
+	case HPD_SOURCEID_UNKNOWN:
+	default:
+		atom_hpd_sel = 0;
+		break;
+	}
+	return atom_hpd_sel >> 4;
+}
+
+static uint8_t dig_encoder_sel_to_atom(enum engine_id id)
+{
+	uint8_t atom_dig_encoder_sel = 0;
+
+	switch (id) {
+	case ENGINE_ID_DIGA:
+		atom_dig_encoder_sel = ATOM_TRANMSITTER_V5__DIGA_SEL;
+		break;
+	case ENGINE_ID_DIGB:
+		atom_dig_encoder_sel = ATOM_TRANMSITTER_V5__DIGB_SEL;
+		break;
+	case ENGINE_ID_DIGC:
+		atom_dig_encoder_sel = ATOM_TRANMSITTER_V5__DIGC_SEL;
+		break;
+	case ENGINE_ID_DIGD:
+		atom_dig_encoder_sel = ATOM_TRANMSITTER_V5__DIGD_SEL;
+		break;
+	case ENGINE_ID_DIGE:
+		atom_dig_encoder_sel = ATOM_TRANMSITTER_V5__DIGE_SEL;
+		break;
+	case ENGINE_ID_DIGF:
+		atom_dig_encoder_sel = ATOM_TRANMSITTER_V5__DIGF_SEL;
+		break;
+	case ENGINE_ID_DIGG:
+		atom_dig_encoder_sel = ATOM_TRANMSITTER_V5__DIGG_SEL;
+		break;
+	default:
+		atom_dig_encoder_sel = ATOM_TRANMSITTER_V5__DIGA_SEL;
+		break;
+	}
+
+	return atom_dig_encoder_sel;
+}
+
+static uint8_t disp_power_gating_action_to_atom(
+	enum bp_pipe_control_action action)
+{
+	uint8_t atom_pipe_action = 0;
+
+	switch (action) {
+	case ASIC_PIPE_DISABLE:
+		atom_pipe_action = ATOM_DISABLE;
+		break;
+	case ASIC_PIPE_ENABLE:
+		atom_pipe_action = ATOM_ENABLE;
+		break;
+	case ASIC_PIPE_INIT:
+		atom_pipe_action = ATOM_INIT;
+		break;
+	default:
+		BREAK_TO_DEBUGGER(); /* Unhandle action in driver! */
+		break;
+	}
+
+	return atom_pipe_action;
+}
+
+static const struct command_table_helper command_table_helper_funcs = {
+	.controller_id_to_atom = dal_cmd_table_helper_controller_id_to_atom,
+	.encoder_action_to_atom = encoder_action_to_atom,
+	.engine_bp_to_atom = engine_bp_to_atom,
+	.clock_source_id_to_atom = clock_source_id_to_atom,
+	.clock_source_id_to_atom_phy_clk_src_id =
+		clock_source_id_to_atom_phy_clk_src_id,
+	.signal_type_to_atom_dig_mode = signal_type_to_atom_dig_mode,
+	.hpd_sel_to_atom = hpd_sel_to_atom,
+	.dig_encoder_sel_to_atom = dig_encoder_sel_to_atom,
+	.phy_id_to_atom = phy_id_to_atom,
+	.disp_power_gating_action_to_atom = disp_power_gating_action_to_atom,
+	.assign_control_parameter =
+		dal_cmd_table_helper_assign_control_parameter,
+	.clock_source_id_to_ref_clk_src =
+		dal_cmd_table_helper_clock_source_id_to_ref_clk_src,
+	.transmitter_bp_to_atom = dal_cmd_table_helper_transmitter_bp_to_atom,
+	.encoder_id_to_atom = dal_cmd_table_helper_encoder_id_to_atom,
+	.encoder_mode_bp_to_atom =
+		dal_cmd_table_helper_encoder_mode_bp_to_atom,
+};
+
+const struct command_table_helper *dal_cmd_tbl_helper_dce60_get_table(void)
+{
+	return &command_table_helper_funcs;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/bios/dce60/command_table_helper_dce60.h b/drivers/gpu/drm/amd/display/dc/bios/dce60/command_table_helper_dce60.h
new file mode 100644
index 000000000000..f733be553d5a
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/dce60/command_table_helper_dce60.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2020 Mauro Rossi <issor.oruam@gmail.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_COMMAND_TABLE_HELPER_DCE60_H__
+#define __DAL_COMMAND_TABLE_HELPER_DCE60_H__
+
+struct command_table_helper;
+
+const struct command_table_helper *dal_cmd_tbl_helper_dce60_get_table(void);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/bios/dce80/command_table_helper_dce80.c b/drivers/gpu/drm/amd/display/dc/bios/dce80/command_table_helper_dce80.c
new file mode 100644
index 000000000000..2ec5264536c7
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/dce80/command_table_helper_dce80.c
@@ -0,0 +1,250 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+
+#include "atom.h"
+
+#include "include/grph_object_id.h"
+#include "include/grph_object_defs.h"
+#include "include/bios_parser_types.h"
+
+#include "../command_table_helper.h"
+
+static uint8_t encoder_action_to_atom(enum bp_encoder_control_action action)
+{
+	uint8_t atom_action = 0;
+
+	switch (action) {
+	case ENCODER_CONTROL_ENABLE:
+		atom_action = ATOM_ENABLE;
+		break;
+	case ENCODER_CONTROL_DISABLE:
+		atom_action = ATOM_DISABLE;
+		break;
+	case ENCODER_CONTROL_SETUP:
+		atom_action = ATOM_ENCODER_CMD_SETUP;
+		break;
+	case ENCODER_CONTROL_INIT:
+		atom_action = ATOM_ENCODER_INIT;
+		break;
+	default:
+		BREAK_TO_DEBUGGER(); /* Unhandle action in driver.!! */
+		break;
+	}
+
+	return atom_action;
+}
+
+static bool clock_source_id_to_atom(
+	enum clock_source_id id,
+	uint32_t *atom_pll_id)
+{
+	bool result = true;
+
+	if (atom_pll_id != NULL)
+		switch (id) {
+		case CLOCK_SOURCE_ID_PLL0:
+			*atom_pll_id = ATOM_PPLL0;
+			break;
+		case CLOCK_SOURCE_ID_PLL1:
+			*atom_pll_id = ATOM_PPLL1;
+			break;
+		case CLOCK_SOURCE_ID_PLL2:
+			*atom_pll_id = ATOM_PPLL2;
+			break;
+		case CLOCK_SOURCE_ID_EXTERNAL:
+			*atom_pll_id = ATOM_PPLL_INVALID;
+			break;
+		case CLOCK_SOURCE_ID_DFS:
+			*atom_pll_id = ATOM_EXT_PLL1;
+			break;
+		case CLOCK_SOURCE_ID_VCE:
+			/* for VCE encoding,
+			 * we need to pass in ATOM_PPLL_INVALID
+			 */
+			*atom_pll_id = ATOM_PPLL_INVALID;
+			break;
+		case CLOCK_SOURCE_ID_DP_DTO:
+			/* When programming DP DTO PLL ID should be invalid */
+			*atom_pll_id = ATOM_PPLL_INVALID;
+			break;
+		case CLOCK_SOURCE_ID_UNDEFINED:
+			BREAK_TO_DEBUGGER(); /* check when this will happen! */
+			*atom_pll_id = ATOM_PPLL_INVALID;
+			result = false;
+			break;
+		default:
+			result = false;
+			break;
+		}
+
+	return result;
+}
+
+static uint8_t signal_type_to_atom_dig_mode(enum signal_type s)
+{
+	uint8_t atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V5_DP;
+
+	switch (s) {
+	case SIGNAL_TYPE_DISPLAY_PORT:
+	case SIGNAL_TYPE_EDP:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V5_DP;
+		break;
+	case SIGNAL_TYPE_LVDS:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V5_LVDS;
+		break;
+	case SIGNAL_TYPE_DVI_SINGLE_LINK:
+	case SIGNAL_TYPE_DVI_DUAL_LINK:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V5_DVI;
+		break;
+	case SIGNAL_TYPE_HDMI_TYPE_A:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V5_HDMI;
+		break;
+	case SIGNAL_TYPE_DISPLAY_PORT_MST:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V5_DP_MST;
+		break;
+	default:
+		atom_dig_mode = ATOM_TRANSMITTER_DIGMODE_V5_DVI;
+		break;
+	}
+
+	return atom_dig_mode;
+}
+
+static uint8_t hpd_sel_to_atom(enum hpd_source_id id)
+{
+	uint8_t atom_hpd_sel = 0;
+
+	switch (id) {
+	case HPD_SOURCEID1:
+		atom_hpd_sel = ATOM_TRANSMITTER_CONFIG_V5_HPD1_SEL;
+		break;
+	case HPD_SOURCEID2:
+		atom_hpd_sel = ATOM_TRANSMITTER_CONFIG_V5_HPD2_SEL;
+		break;
+	case HPD_SOURCEID3:
+		atom_hpd_sel = ATOM_TRANSMITTER_CONFIG_V5_HPD3_SEL;
+		break;
+	case HPD_SOURCEID4:
+		atom_hpd_sel = ATOM_TRANSMITTER_CONFIG_V5_HPD4_SEL;
+		break;
+	case HPD_SOURCEID5:
+		atom_hpd_sel = ATOM_TRANSMITTER_CONFIG_V5_HPD5_SEL;
+		break;
+	case HPD_SOURCEID6:
+		atom_hpd_sel = ATOM_TRANSMITTER_CONFIG_V5_HPD6_SEL;
+		break;
+	case HPD_SOURCEID_UNKNOWN:
+	default:
+		atom_hpd_sel = 0;
+		break;
+	}
+	return atom_hpd_sel >> 4;
+}
+
+static uint8_t dig_encoder_sel_to_atom(enum engine_id id)
+{
+	uint8_t atom_dig_encoder_sel = 0;
+
+	switch (id) {
+	case ENGINE_ID_DIGA:
+		atom_dig_encoder_sel = ATOM_TRANMSITTER_V5__DIGA_SEL;
+		break;
+	case ENGINE_ID_DIGB:
+		atom_dig_encoder_sel = ATOM_TRANMSITTER_V5__DIGB_SEL;
+		break;
+	case ENGINE_ID_DIGC:
+		atom_dig_encoder_sel = ATOM_TRANMSITTER_V5__DIGC_SEL;
+		break;
+	case ENGINE_ID_DIGD:
+		atom_dig_encoder_sel = ATOM_TRANMSITTER_V5__DIGD_SEL;
+		break;
+	case ENGINE_ID_DIGE:
+		atom_dig_encoder_sel = ATOM_TRANMSITTER_V5__DIGE_SEL;
+		break;
+	case ENGINE_ID_DIGF:
+		atom_dig_encoder_sel = ATOM_TRANMSITTER_V5__DIGF_SEL;
+		break;
+	case ENGINE_ID_DIGG:
+		atom_dig_encoder_sel = ATOM_TRANMSITTER_V5__DIGG_SEL;
+		break;
+	default:
+		atom_dig_encoder_sel = ATOM_TRANMSITTER_V5__DIGA_SEL;
+		break;
+	}
+
+	return atom_dig_encoder_sel;
+}
+
+static uint8_t disp_power_gating_action_to_atom(
+	enum bp_pipe_control_action action)
+{
+	uint8_t atom_pipe_action = 0;
+
+	switch (action) {
+	case ASIC_PIPE_DISABLE:
+		atom_pipe_action = ATOM_DISABLE;
+		break;
+	case ASIC_PIPE_ENABLE:
+		atom_pipe_action = ATOM_ENABLE;
+		break;
+	case ASIC_PIPE_INIT:
+		atom_pipe_action = ATOM_INIT;
+		break;
+	default:
+		BREAK_TO_DEBUGGER(); /* Unhandle action in driver! */
+		break;
+	}
+
+	return atom_pipe_action;
+}
+
+static const struct command_table_helper command_table_helper_funcs = {
+	.controller_id_to_atom = dal_cmd_table_helper_controller_id_to_atom,
+	.encoder_action_to_atom = encoder_action_to_atom,
+	.engine_bp_to_atom = engine_bp_to_atom,
+	.clock_source_id_to_atom = clock_source_id_to_atom,
+	.clock_source_id_to_atom_phy_clk_src_id =
+		clock_source_id_to_atom_phy_clk_src_id,
+	.signal_type_to_atom_dig_mode = signal_type_to_atom_dig_mode,
+	.hpd_sel_to_atom = hpd_sel_to_atom,
+	.dig_encoder_sel_to_atom = dig_encoder_sel_to_atom,
+	.phy_id_to_atom = phy_id_to_atom,
+	.disp_power_gating_action_to_atom = disp_power_gating_action_to_atom,
+	.assign_control_parameter =
+		dal_cmd_table_helper_assign_control_parameter,
+	.clock_source_id_to_ref_clk_src =
+		dal_cmd_table_helper_clock_source_id_to_ref_clk_src,
+	.transmitter_bp_to_atom = dal_cmd_table_helper_transmitter_bp_to_atom,
+	.encoder_id_to_atom = dal_cmd_table_helper_encoder_id_to_atom,
+	.encoder_mode_bp_to_atom =
+		dal_cmd_table_helper_encoder_mode_bp_to_atom,
+};
+
+const struct command_table_helper *dal_cmd_tbl_helper_dce80_get_table(void)
+{
+	return &command_table_helper_funcs;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/bios/dce80/command_table_helper_dce80.h b/drivers/gpu/drm/amd/display/dc/bios/dce80/command_table_helper_dce80.h
new file mode 100644
index 000000000000..e675c359e306
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/bios/dce80/command_table_helper_dce80.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_COMMAND_TABLE_HELPER_DCE80_H__
+#define __DAL_COMMAND_TABLE_HELPER_DCE80_H__
+
+struct command_table_helper;
+
+const struct command_table_helper *dal_cmd_tbl_helper_dce80_get_table(void);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/Makefile b/drivers/gpu/drm/amd/display/dc/clk_mgr/Makefile
new file mode 100644
index 000000000000..60021671b386
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/Makefile
@@ -0,0 +1,191 @@
+#
+# Copyright 2017 Advanced Micro Devices, Inc.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+#
+# Makefile for the 'clk_mgr' sub-component of DAL.
+# It provides the control and status of HW CLK_MGR pins.
+
+CLK_MGR = clk_mgr.o
+
+AMD_DAL_CLK_MGR = $(addprefix $(AMDDALPATH)/dc/clk_mgr/,$(CLK_MGR))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_CLK_MGR)
+
+
+ifdef CONFIG_DRM_AMD_DC_SI
+###############################################################################
+# DCE 60
+###############################################################################
+CLK_MGR_DCE60 = dce60_clk_mgr.o
+
+AMD_DAL_CLK_MGR_DCE60 = $(addprefix $(AMDDALPATH)/dc/clk_mgr/dce60/,$(CLK_MGR_DCE60))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_CLK_MGR_DCE60)
+endif
+
+###############################################################################
+# DCE 100 and DCE8x
+###############################################################################
+CLK_MGR_DCE100 = dce_clk_mgr.o
+
+AMD_DAL_CLK_MGR_DCE100 = $(addprefix $(AMDDALPATH)/dc/clk_mgr/dce100/,$(CLK_MGR_DCE100))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_CLK_MGR_DCE100)
+
+###############################################################################
+# DCE 100 and DCE8x
+###############################################################################
+CLK_MGR_DCE110 = dce110_clk_mgr.o
+
+AMD_DAL_CLK_MGR_DCE110 = $(addprefix $(AMDDALPATH)/dc/clk_mgr/dce110/,$(CLK_MGR_DCE110))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_CLK_MGR_DCE110)
+###############################################################################
+# DCE 112
+###############################################################################
+CLK_MGR_DCE112 = dce112_clk_mgr.o
+
+AMD_DAL_CLK_MGR_DCE112 = $(addprefix $(AMDDALPATH)/dc/clk_mgr/dce112/,$(CLK_MGR_DCE112))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_CLK_MGR_DCE112)
+###############################################################################
+# DCE 120
+###############################################################################
+CLK_MGR_DCE120 = dce120_clk_mgr.o
+
+AMD_DAL_CLK_MGR_DCE120 = $(addprefix $(AMDDALPATH)/dc/clk_mgr/dce120/,$(CLK_MGR_DCE120))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_CLK_MGR_DCE120)
+ifdef CONFIG_DRM_AMD_DC_FP
+###############################################################################
+# DCN10
+###############################################################################
+CLK_MGR_DCN10 = rv1_clk_mgr.o rv1_clk_mgr_vbios_smu.o rv2_clk_mgr.o
+
+AMD_DAL_CLK_MGR_DCN10 = $(addprefix $(AMDDALPATH)/dc/clk_mgr/dcn10/,$(CLK_MGR_DCN10))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_CLK_MGR_DCN10)
+###############################################################################
+# DCN20
+###############################################################################
+CLK_MGR_DCN20 = dcn20_clk_mgr.o
+
+AMD_DAL_CLK_MGR_DCN20 = $(addprefix $(AMDDALPATH)/dc/clk_mgr/dcn20/,$(CLK_MGR_DCN20))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_CLK_MGR_DCN20)
+
+###############################################################################
+# DCN201
+###############################################################################
+CLK_MGR_DCN201 = dcn201_clk_mgr.o
+
+AMD_DAL_CLK_MGR_DCN201 = $(addprefix $(AMDDALPATH)/dc/clk_mgr/dcn201/,$(CLK_MGR_DCN201))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_CLK_MGR_DCN201)
+
+###############################################################################
+# DCN21
+###############################################################################
+CLK_MGR_DCN21 = rn_clk_mgr.o rn_clk_mgr_vbios_smu.o
+
+AMD_DAL_CLK_MGR_DCN21 = $(addprefix $(AMDDALPATH)/dc/clk_mgr/dcn21/,$(CLK_MGR_DCN21))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_CLK_MGR_DCN21)
+###############################################################################
+# DCN30
+###############################################################################
+CLK_MGR_DCN30 = dcn30_clk_mgr.o dcn30_clk_mgr_smu_msg.o dcn30m_clk_mgr.o dcn30m_clk_mgr_smu_msg.o
+
+AMD_DAL_CLK_MGR_DCN30 = $(addprefix $(AMDDALPATH)/dc/clk_mgr/dcn30/,$(CLK_MGR_DCN30))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_CLK_MGR_DCN30)
+###############################################################################
+# DCN301
+###############################################################################
+CLK_MGR_DCN301 = vg_clk_mgr.o dcn301_smu.o
+
+AMD_DAL_CLK_MGR_DCN301 = $(addprefix $(AMDDALPATH)/dc/clk_mgr/dcn301/,$(CLK_MGR_DCN301))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_CLK_MGR_DCN301)
+
+###############################################################################
+# DCN31
+###############################################################################
+CLK_MGR_DCN31 = dcn31_smu.o dcn31_clk_mgr.o
+
+AMD_DAL_CLK_MGR_DCN31 = $(addprefix $(AMDDALPATH)/dc/clk_mgr/dcn31/,$(CLK_MGR_DCN31))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_CLK_MGR_DCN31)
+
+###############################################################################
+# DCN314
+###############################################################################
+CLK_MGR_DCN314 = dcn314_smu.o dcn314_clk_mgr.o
+
+AMD_DAL_CLK_MGR_DCN314 = $(addprefix $(AMDDALPATH)/dc/clk_mgr/dcn314/,$(CLK_MGR_DCN314))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_CLK_MGR_DCN314)
+
+###############################################################################
+# DCN315
+###############################################################################
+CLK_MGR_DCN315 = dcn315_smu.o dcn315_clk_mgr.o
+
+AMD_DAL_CLK_MGR_DCN315 = $(addprefix $(AMDDALPATH)/dc/clk_mgr/dcn315/,$(CLK_MGR_DCN315))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_CLK_MGR_DCN315)
+
+###############################################################################
+# DCN316
+###############################################################################
+CLK_MGR_DCN316 = dcn316_smu.o dcn316_clk_mgr.o
+
+AMD_DAL_CLK_MGR_DCN316 = $(addprefix $(AMDDALPATH)/dc/clk_mgr/dcn316/,$(CLK_MGR_DCN316))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_CLK_MGR_DCN316)
+
+###############################################################################
+# DCN32
+###############################################################################
+CLK_MGR_DCN32 = dcn32_clk_mgr.o dcn32_clk_mgr_smu_msg.o
+
+AMD_DAL_CLK_MGR_DCN32 = $(addprefix $(AMDDALPATH)/dc/clk_mgr/dcn32/,$(CLK_MGR_DCN32))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_CLK_MGR_DCN32)
+
+###############################################################################
+# DCN35
+###############################################################################
+CLK_MGR_DCN35 = dcn35_smu.o dcn351_clk_mgr.o dcn35_clk_mgr.o
+
+AMD_DAL_CLK_MGR_DCN35 = $(addprefix $(AMDDALPATH)/dc/clk_mgr/dcn35/,$(CLK_MGR_DCN35))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_CLK_MGR_DCN35)
+
+###############################################################################
+# DCN401
+###############################################################################
+CLK_MGR_DCN401 = dcn401_clk_mgr.o dcn401_clk_mgr_smu_msg.o
+
+AMD_DAL_CLK_MGR_DCN401 = $(addprefix $(AMDDALPATH)/dc/clk_mgr/dcn401/,$(CLK_MGR_DCN401))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_CLK_MGR_DCN401)
+endif
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/clk_mgr.c
new file mode 100644
index 000000000000..15cf13ec5302
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/clk_mgr.c
@@ -0,0 +1,442 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dal_asic_id.h"
+#include "dc_types.h"
+#include "dccg.h"
+#include "clk_mgr_internal.h"
+#include "dc_state_priv.h"
+#include "link_service.h"
+
+#include "dce100/dce_clk_mgr.h"
+#include "dce110/dce110_clk_mgr.h"
+#include "dce112/dce112_clk_mgr.h"
+#include "dce120/dce120_clk_mgr.h"
+#include "dce60/dce60_clk_mgr.h"
+#include "dcn10/rv1_clk_mgr.h"
+#include "dcn10/rv2_clk_mgr.h"
+#include "dcn20/dcn20_clk_mgr.h"
+#include "dcn21/rn_clk_mgr.h"
+#include "dcn201/dcn201_clk_mgr.h"
+#include "dcn30/dcn30_clk_mgr.h"
+#include "dcn301/vg_clk_mgr.h"
+#include "dcn31/dcn31_clk_mgr.h"
+#include "dcn314/dcn314_clk_mgr.h"
+#include "dcn315/dcn315_clk_mgr.h"
+#include "dcn316/dcn316_clk_mgr.h"
+#include "dcn32/dcn32_clk_mgr.h"
+#include "dcn35/dcn35_clk_mgr.h"
+#include "dcn401/dcn401_clk_mgr.h"
+
+int clk_mgr_helper_get_active_display_cnt(
+		struct dc *dc,
+		struct dc_state *context)
+{
+	int i, display_count;
+
+	display_count = 0;
+	for (i = 0; i < context->stream_count; i++) {
+		const struct dc_stream_state *stream = context->streams[i];
+		const struct dc_stream_status *stream_status = &context->stream_status[i];
+
+		/* Don't count SubVP phantom pipes as part of active
+		 * display count
+		 */
+		if (dc_state_get_stream_subvp_type(context, stream) == SUBVP_PHANTOM)
+			continue;
+
+		if (!stream->dpms_off || dc->is_switch_in_progress_dest || (stream_status && stream_status->plane_count))
+			display_count++;
+	}
+
+	return display_count;
+}
+
+int clk_mgr_helper_get_active_plane_cnt(
+		struct dc *dc,
+		struct dc_state *context)
+{
+	int i, total_plane_count;
+
+	total_plane_count = 0;
+	for (i = 0; i < context->stream_count; i++) {
+		const struct dc_stream_status stream_status = context->stream_status[i];
+
+		/*
+		 * Sum up plane_count for all streams ( active and virtual ).
+		 */
+		total_plane_count += stream_status.plane_count;
+	}
+
+	return total_plane_count;
+}
+
+void clk_mgr_exit_optimized_pwr_state(const struct dc *dc, struct clk_mgr *clk_mgr)
+{
+	struct dc_link *edp_links[MAX_NUM_EDP];
+	struct dc_link *edp_link = NULL;
+	int edp_num;
+	unsigned int panel_inst;
+
+	dc_get_edp_links(dc, edp_links, &edp_num);
+	if (dc->hwss.exit_optimized_pwr_state)
+		dc->hwss.exit_optimized_pwr_state(dc, dc->current_state);
+
+	if (edp_num) {
+		for (panel_inst = 0; panel_inst < edp_num; panel_inst++) {
+			bool allow_active = false;
+
+			edp_link = edp_links[panel_inst];
+			if (!edp_link->psr_settings.psr_feature_enabled)
+				continue;
+			clk_mgr->psr_allow_active_cache = edp_link->psr_settings.psr_allow_active;
+			dc->link_srv->edp_set_psr_allow_active(edp_link, &allow_active, false, false, NULL);
+			dc->link_srv->edp_set_replay_allow_active(edp_link, &allow_active, false, false, NULL);
+		}
+	}
+
+}
+
+void clk_mgr_optimize_pwr_state(const struct dc *dc, struct clk_mgr *clk_mgr)
+{
+	struct dc_link *edp_links[MAX_NUM_EDP];
+	struct dc_link *edp_link = NULL;
+	int edp_num;
+	unsigned int panel_inst;
+
+	dc_get_edp_links(dc, edp_links, &edp_num);
+	if (edp_num) {
+		for (panel_inst = 0; panel_inst < edp_num; panel_inst++) {
+			edp_link = edp_links[panel_inst];
+			if (!edp_link->psr_settings.psr_feature_enabled)
+				continue;
+			dc->link_srv->edp_set_psr_allow_active(edp_link,
+					&clk_mgr->psr_allow_active_cache, false, false, NULL);
+			dc->link_srv->edp_set_replay_allow_active(edp_link,
+					&clk_mgr->psr_allow_active_cache, false, false, NULL);
+		}
+	}
+
+	if (dc->hwss.optimize_pwr_state)
+		dc->hwss.optimize_pwr_state(dc, dc->current_state);
+
+}
+
+struct clk_mgr *dc_clk_mgr_create(struct dc_context *ctx, struct pp_smu_funcs *pp_smu, struct dccg *dccg)
+{
+	struct hw_asic_id asic_id = ctx->asic_id;
+
+	switch (asic_id.chip_family) {
+#if defined(CONFIG_DRM_AMD_DC_SI)
+	case FAMILY_SI: {
+		struct clk_mgr_internal *clk_mgr = kzalloc(sizeof(*clk_mgr), GFP_KERNEL);
+
+		if (clk_mgr == NULL) {
+			BREAK_TO_DEBUGGER();
+			return NULL;
+		}
+		dce60_clk_mgr_construct(ctx, clk_mgr);
+		return &clk_mgr->base;
+	}
+#endif
+	case FAMILY_CI:
+	case FAMILY_KV: {
+		struct clk_mgr_internal *clk_mgr = kzalloc(sizeof(*clk_mgr), GFP_KERNEL);
+
+		if (clk_mgr == NULL) {
+			BREAK_TO_DEBUGGER();
+			return NULL;
+		}
+		dce_clk_mgr_construct(ctx, clk_mgr);
+		return &clk_mgr->base;
+	}
+	case FAMILY_CZ: {
+		struct clk_mgr_internal *clk_mgr = kzalloc(sizeof(*clk_mgr), GFP_KERNEL);
+
+		if (clk_mgr == NULL) {
+			BREAK_TO_DEBUGGER();
+			return NULL;
+		}
+		dce110_clk_mgr_construct(ctx, clk_mgr);
+		return &clk_mgr->base;
+	}
+	case FAMILY_VI: {
+		struct clk_mgr_internal *clk_mgr = kzalloc(sizeof(*clk_mgr), GFP_KERNEL);
+
+		if (clk_mgr == NULL) {
+			BREAK_TO_DEBUGGER();
+			return NULL;
+		}
+		if (ASIC_REV_IS_TONGA_P(asic_id.hw_internal_rev) ||
+				ASIC_REV_IS_FIJI_P(asic_id.hw_internal_rev)) {
+			dce_clk_mgr_construct(ctx, clk_mgr);
+			return &clk_mgr->base;
+		}
+		if (ASIC_REV_IS_POLARIS10_P(asic_id.hw_internal_rev) ||
+				ASIC_REV_IS_POLARIS11_M(asic_id.hw_internal_rev) ||
+				ASIC_REV_IS_POLARIS12_V(asic_id.hw_internal_rev)) {
+			dce112_clk_mgr_construct(ctx, clk_mgr);
+			return &clk_mgr->base;
+		}
+		if (ASIC_REV_IS_VEGAM(asic_id.hw_internal_rev)) {
+			dce112_clk_mgr_construct(ctx, clk_mgr);
+			return &clk_mgr->base;
+		}
+		return &clk_mgr->base;
+	}
+	case FAMILY_AI: {
+		struct clk_mgr_internal *clk_mgr = kzalloc(sizeof(*clk_mgr), GFP_KERNEL);
+
+		if (clk_mgr == NULL) {
+			BREAK_TO_DEBUGGER();
+			return NULL;
+		}
+		if (ASICREV_IS_VEGA20_P(asic_id.hw_internal_rev))
+			dce121_clk_mgr_construct(ctx, clk_mgr);
+		else
+			dce120_clk_mgr_construct(ctx, clk_mgr);
+		return &clk_mgr->base;
+	}
+#if defined(CONFIG_DRM_AMD_DC_FP)
+	case FAMILY_RV: {
+		struct clk_mgr_internal *clk_mgr = kzalloc(sizeof(*clk_mgr), GFP_KERNEL);
+
+		if (clk_mgr == NULL) {
+			BREAK_TO_DEBUGGER();
+			return NULL;
+		}
+
+		if (ASICREV_IS_RENOIR(asic_id.hw_internal_rev)) {
+			rn_clk_mgr_construct(ctx, clk_mgr, pp_smu, dccg);
+			return &clk_mgr->base;
+		}
+
+		if (ASICREV_IS_GREEN_SARDINE(asic_id.hw_internal_rev)) {
+			rn_clk_mgr_construct(ctx, clk_mgr, pp_smu, dccg);
+			return &clk_mgr->base;
+		}
+		if (ASICREV_IS_RAVEN2(asic_id.hw_internal_rev)) {
+			rv2_clk_mgr_construct(ctx, clk_mgr, pp_smu);
+			return &clk_mgr->base;
+		}
+		if (ASICREV_IS_RAVEN(asic_id.hw_internal_rev) ||
+				ASICREV_IS_PICASSO(asic_id.hw_internal_rev)) {
+			rv1_clk_mgr_construct(ctx, clk_mgr, pp_smu);
+			return &clk_mgr->base;
+		}
+		return &clk_mgr->base;
+	}
+	case FAMILY_NV: {
+		struct clk_mgr_internal *clk_mgr = kzalloc(sizeof(*clk_mgr), GFP_KERNEL);
+
+		if (clk_mgr == NULL) {
+			BREAK_TO_DEBUGGER();
+			return NULL;
+		}
+		if (ASICREV_IS_SIENNA_CICHLID_P(asic_id.hw_internal_rev)) {
+			dcn3_clk_mgr_construct(ctx, clk_mgr, pp_smu, dccg);
+			return &clk_mgr->base;
+		}
+		if (ASICREV_IS_DIMGREY_CAVEFISH_P(asic_id.hw_internal_rev)) {
+			dcn3_clk_mgr_construct(ctx, clk_mgr, pp_smu, dccg);
+			return &clk_mgr->base;
+		}
+		if (ASICREV_IS_BEIGE_GOBY_P(asic_id.hw_internal_rev)) {
+			dcn3_clk_mgr_construct(ctx, clk_mgr, pp_smu, dccg);
+			return &clk_mgr->base;
+		}
+		if (ctx->dce_version == DCN_VERSION_2_01) {
+			dcn201_clk_mgr_construct(ctx, clk_mgr, pp_smu, dccg);
+			return &clk_mgr->base;
+		}
+		dcn20_clk_mgr_construct(ctx, clk_mgr, pp_smu, dccg);
+		return &clk_mgr->base;
+	}
+	case FAMILY_VGH:
+		if (ASICREV_IS_VANGOGH(asic_id.hw_internal_rev)) {
+			struct clk_mgr_vgh *clk_mgr = kzalloc(sizeof(*clk_mgr), GFP_KERNEL);
+
+			if (clk_mgr == NULL) {
+				BREAK_TO_DEBUGGER();
+				return NULL;
+			}
+			vg_clk_mgr_construct(ctx, clk_mgr, pp_smu, dccg);
+			return &clk_mgr->base.base;
+		}
+		break;
+
+	case FAMILY_YELLOW_CARP: {
+		struct clk_mgr_dcn31 *clk_mgr = kzalloc(sizeof(*clk_mgr), GFP_KERNEL);
+
+		if (clk_mgr == NULL) {
+			BREAK_TO_DEBUGGER();
+			return NULL;
+		}
+
+		dcn31_clk_mgr_construct(ctx, clk_mgr, pp_smu, dccg);
+		return &clk_mgr->base.base;
+	}
+		break;
+	case AMDGPU_FAMILY_GC_10_3_6: {
+		struct clk_mgr_dcn315 *clk_mgr = kzalloc(sizeof(*clk_mgr), GFP_KERNEL);
+
+		if (clk_mgr == NULL) {
+			BREAK_TO_DEBUGGER();
+			return NULL;
+		}
+
+		dcn315_clk_mgr_construct(ctx, clk_mgr, pp_smu, dccg);
+		return &clk_mgr->base.base;
+	}
+		break;
+	case AMDGPU_FAMILY_GC_10_3_7: {
+		struct clk_mgr_dcn316 *clk_mgr = kzalloc(sizeof(*clk_mgr), GFP_KERNEL);
+
+		if (clk_mgr == NULL) {
+			BREAK_TO_DEBUGGER();
+			return NULL;
+		}
+
+		dcn316_clk_mgr_construct(ctx, clk_mgr, pp_smu, dccg);
+		return &clk_mgr->base.base;
+	}
+		break;
+	case AMDGPU_FAMILY_GC_11_0_0: {
+		struct clk_mgr_internal *clk_mgr = kzalloc(sizeof(*clk_mgr), GFP_KERNEL);
+
+		if (clk_mgr == NULL) {
+			BREAK_TO_DEBUGGER();
+			return NULL;
+		}
+		dcn32_clk_mgr_construct(ctx, clk_mgr, pp_smu, dccg);
+		return &clk_mgr->base;
+	}
+
+	case AMDGPU_FAMILY_GC_11_0_1: {
+		struct clk_mgr_dcn314 *clk_mgr = kzalloc(sizeof(*clk_mgr), GFP_KERNEL);
+
+		if (clk_mgr == NULL) {
+			BREAK_TO_DEBUGGER();
+			return NULL;
+		}
+
+		dcn314_clk_mgr_construct(ctx, clk_mgr, pp_smu, dccg);
+		return &clk_mgr->base.base;
+	}
+	break;
+
+	case AMDGPU_FAMILY_GC_11_5_0: {
+		struct clk_mgr_dcn35 *clk_mgr = kzalloc(sizeof(*clk_mgr), GFP_KERNEL);
+
+		if (clk_mgr == NULL) {
+			BREAK_TO_DEBUGGER();
+			return NULL;
+		}
+		if (ctx->dce_version == DCN_VERSION_3_51)
+			dcn351_clk_mgr_construct(ctx, clk_mgr, pp_smu, dccg);
+		else
+			dcn35_clk_mgr_construct(ctx, clk_mgr, pp_smu, dccg);
+
+		return &clk_mgr->base.base;
+	}
+	break;
+
+	case AMDGPU_FAMILY_GC_12_0_0: {
+		struct clk_mgr_internal *clk_mgr = dcn401_clk_mgr_construct(ctx, dccg);
+
+		if (clk_mgr == NULL) {
+			BREAK_TO_DEBUGGER();
+			return NULL;
+		}
+
+		return &clk_mgr->base;
+	}
+	break;
+#endif	/* CONFIG_DRM_AMD_DC_FP */
+	default:
+		ASSERT(0); /* Unknown Asic */
+		break;
+	}
+
+	return NULL;
+}
+
+void dc_destroy_clk_mgr(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+#ifdef CONFIG_DRM_AMD_DC_FP
+	switch (clk_mgr_base->ctx->asic_id.chip_family) {
+	case FAMILY_NV:
+		if (ASICREV_IS_SIENNA_CICHLID_P(clk_mgr_base->ctx->asic_id.hw_internal_rev)) {
+			dcn3_clk_mgr_destroy(clk_mgr);
+		} else if (ASICREV_IS_DIMGREY_CAVEFISH_P(clk_mgr_base->ctx->asic_id.hw_internal_rev)) {
+			dcn3_clk_mgr_destroy(clk_mgr);
+		}
+		if (ASICREV_IS_BEIGE_GOBY_P(clk_mgr_base->ctx->asic_id.hw_internal_rev)) {
+			dcn3_clk_mgr_destroy(clk_mgr);
+		}
+		break;
+
+	case FAMILY_VGH:
+		if (ASICREV_IS_VANGOGH(clk_mgr_base->ctx->asic_id.hw_internal_rev))
+			vg_clk_mgr_destroy(clk_mgr);
+		break;
+
+	case FAMILY_YELLOW_CARP:
+		dcn31_clk_mgr_destroy(clk_mgr);
+		break;
+
+	case AMDGPU_FAMILY_GC_10_3_6:
+		dcn315_clk_mgr_destroy(clk_mgr);
+		break;
+
+	case AMDGPU_FAMILY_GC_10_3_7:
+		dcn316_clk_mgr_destroy(clk_mgr);
+		break;
+
+	case AMDGPU_FAMILY_GC_11_0_0:
+		dcn32_clk_mgr_destroy(clk_mgr);
+		break;
+
+	case AMDGPU_FAMILY_GC_11_0_1:
+		dcn314_clk_mgr_destroy(clk_mgr);
+		break;
+
+	case AMDGPU_FAMILY_GC_11_5_0:
+		dcn35_clk_mgr_destroy(clk_mgr);
+		break;
+	case AMDGPU_FAMILY_GC_12_0_0:
+		dcn401_clk_mgr_destroy(clk_mgr);
+		break;
+
+	default:
+		break;
+	}
+#endif /* CONFIG_DRM_AMD_DC_FP */
+
+	kfree(clk_mgr);
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dce100/dce_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dce100/dce_clk_mgr.c
new file mode 100644
index 000000000000..6131ede2db7a
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dce100/dce_clk_mgr.c
@@ -0,0 +1,472 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+#include "dccg.h"
+#include "clk_mgr_internal.h"
+#include "dce_clk_mgr.h"
+#include "dce110/dce110_clk_mgr.h"
+#include "dce112/dce112_clk_mgr.h"
+#include "reg_helper.h"
+#include "dmcu.h"
+#include "core_types.h"
+#include "dal_asic_id.h"
+
+/*
+ * Currently the register shifts and masks in this file are used for dce100 and dce80
+ * which has identical definitions.
+ * TODO: remove this when DPREFCLK_CNTL and dpref DENTIST_DISPCLK_CNTL
+ * is moved to dccg, where it belongs
+ */
+#include "dce/dce_8_0_d.h"
+#include "dce/dce_8_0_sh_mask.h"
+
+#define REG(reg) \
+	(clk_mgr->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	clk_mgr->clk_mgr_shift->field_name, clk_mgr->clk_mgr_mask->field_name
+
+static const struct clk_mgr_registers disp_clk_regs = {
+		CLK_COMMON_REG_LIST_DCE_BASE()
+};
+
+static const struct clk_mgr_shift disp_clk_shift = {
+		CLK_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(__SHIFT)
+};
+
+static const struct clk_mgr_mask disp_clk_mask = {
+		CLK_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(_MASK)
+};
+
+
+/* Max clock values for each state indexed by "enum clocks_state": */
+static const struct state_dependent_clocks dce80_max_clks_by_state[] = {
+/* ClocksStateInvalid - should not be used */
+{ .display_clk_khz = 0, .pixel_clk_khz = 0 },
+/* ClocksStateUltraLow - not expected to be used for DCE 8.0 */
+{ .display_clk_khz = 0, .pixel_clk_khz = 0 },
+/* ClocksStateLow */
+{ .display_clk_khz = 352000, .pixel_clk_khz = 330000},
+/* ClocksStateNominal */
+{ .display_clk_khz = 625000, .pixel_clk_khz = 400000 },
+/* ClocksStatePerformance */
+{ .display_clk_khz = 625000, .pixel_clk_khz = 400000 } };
+
+int dentist_get_divider_from_did(int did)
+{
+	if (did < DENTIST_BASE_DID_1)
+		did = DENTIST_BASE_DID_1;
+	if (did > DENTIST_MAX_DID)
+		did = DENTIST_MAX_DID;
+
+	if (did < DENTIST_BASE_DID_2) {
+		return DENTIST_DIVIDER_RANGE_1_START + DENTIST_DIVIDER_RANGE_1_STEP
+							* (did - DENTIST_BASE_DID_1);
+	} else if (did < DENTIST_BASE_DID_3) {
+		return DENTIST_DIVIDER_RANGE_2_START + DENTIST_DIVIDER_RANGE_2_STEP
+							* (did - DENTIST_BASE_DID_2);
+	} else if (did < DENTIST_BASE_DID_4) {
+		return DENTIST_DIVIDER_RANGE_3_START + DENTIST_DIVIDER_RANGE_3_STEP
+							* (did - DENTIST_BASE_DID_3);
+	} else {
+		return DENTIST_DIVIDER_RANGE_4_START + DENTIST_DIVIDER_RANGE_4_STEP
+							* (did - DENTIST_BASE_DID_4);
+	}
+}
+
+/* SW will adjust DP REF Clock average value for all purposes
+ * (DP DTO / DP Audio DTO and DP GTC)
+ if clock is spread for all cases:
+ -if SS enabled on DP Ref clock and HW de-spreading enabled with SW
+ calculations for DS_INCR/DS_MODULO (this is planned to be default case)
+ -if SS enabled on DP Ref clock and HW de-spreading enabled with HW
+ calculations (not planned to be used, but average clock should still
+ be valid)
+ -if SS enabled on DP Ref clock and HW de-spreading disabled
+ (should not be case with CIK) then SW should program all rates
+ generated according to average value (case as with previous ASICs)
+  */
+
+int dce_adjust_dp_ref_freq_for_ss(struct clk_mgr_internal *clk_mgr_dce, int dp_ref_clk_khz)
+{
+	if (clk_mgr_dce->ss_on_dprefclk && clk_mgr_dce->dprefclk_ss_divider != 0) {
+		struct fixed31_32 ss_percentage = dc_fixpt_div_int(
+				dc_fixpt_from_fraction(clk_mgr_dce->dprefclk_ss_percentage,
+							clk_mgr_dce->dprefclk_ss_divider), 200);
+		struct fixed31_32 adj_dp_ref_clk_khz;
+
+		ss_percentage = dc_fixpt_sub(dc_fixpt_one, ss_percentage);
+		adj_dp_ref_clk_khz = dc_fixpt_mul_int(ss_percentage, dp_ref_clk_khz);
+		dp_ref_clk_khz = dc_fixpt_floor(adj_dp_ref_clk_khz);
+	}
+	return dp_ref_clk_khz;
+}
+
+int dce_get_dp_ref_freq_khz(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	int dprefclk_wdivider;
+	int dprefclk_src_sel;
+	int dp_ref_clk_khz;
+	int target_div;
+
+	/* ASSERT DP Reference Clock source is from DFS*/
+	REG_GET(DPREFCLK_CNTL, DPREFCLK_SRC_SEL, &dprefclk_src_sel);
+	ASSERT(dprefclk_src_sel == 0);
+
+	/* Read the mmDENTIST_DISPCLK_CNTL to get the currently
+	 * programmed DID DENTIST_DPREFCLK_WDIVIDER*/
+	REG_GET(DENTIST_DISPCLK_CNTL, DENTIST_DPREFCLK_WDIVIDER, &dprefclk_wdivider);
+
+	/* Convert DENTIST_DPREFCLK_WDIVIDERto actual divider*/
+	target_div = dentist_get_divider_from_did(dprefclk_wdivider);
+
+	/* Calculate the current DFS clock, in kHz.*/
+	dp_ref_clk_khz = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+		* clk_mgr->base.dentist_vco_freq_khz) / target_div;
+
+	return dce_adjust_dp_ref_freq_for_ss(clk_mgr, dp_ref_clk_khz);
+}
+
+int dce12_get_dp_ref_freq_khz(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr_dce = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	return dce_adjust_dp_ref_freq_for_ss(clk_mgr_dce, clk_mgr_base->dprefclk_khz);
+}
+
+/* unit: in_khz before mode set, get pixel clock from context. ASIC register
+ * may not be programmed yet
+ */
+uint32_t dce_get_max_pixel_clock_for_all_paths(struct dc_state *context)
+{
+	uint32_t max_pix_clk = 0;
+	int i;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i];
+
+		if (pipe_ctx->stream == NULL)
+			continue;
+
+		/* do not check under lay */
+		if (pipe_ctx->top_pipe)
+			continue;
+
+		if (pipe_ctx->stream_res.pix_clk_params.requested_pix_clk_100hz / 10 > max_pix_clk)
+			max_pix_clk = pipe_ctx->stream_res.pix_clk_params.requested_pix_clk_100hz / 10;
+
+		/* raise clock state for HBR3/2 if required. Confirmed with HW DCE/DPCS
+		 * logic for HBR3 still needs Nominal (0.8V) on VDDC rail
+		 */
+		if (dc_is_dp_signal(pipe_ctx->stream->signal) &&
+				pipe_ctx->stream_res.pix_clk_params.requested_sym_clk > max_pix_clk)
+			max_pix_clk = pipe_ctx->stream_res.pix_clk_params.requested_sym_clk;
+	}
+
+	return max_pix_clk;
+}
+
+enum dm_pp_clocks_state dce_get_required_clocks_state(
+	struct clk_mgr *clk_mgr_base,
+	struct dc_state *context)
+{
+	struct clk_mgr_internal *clk_mgr_dce = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	int i;
+	enum dm_pp_clocks_state low_req_clk;
+	int max_pix_clk = dce_get_max_pixel_clock_for_all_paths(context);
+
+	/* Iterate from highest supported to lowest valid state, and update
+	 * lowest RequiredState with the lowest state that satisfies
+	 * all required clocks
+	 */
+	for (i = clk_mgr_dce->max_clks_state; i >= DM_PP_CLOCKS_STATE_ULTRA_LOW; i--)
+		if (context->bw_ctx.bw.dce.dispclk_khz >
+				clk_mgr_dce->max_clks_by_state[i].display_clk_khz
+			|| max_pix_clk >
+				clk_mgr_dce->max_clks_by_state[i].pixel_clk_khz)
+			break;
+
+	low_req_clk = i + 1;
+	if (low_req_clk > clk_mgr_dce->max_clks_state) {
+		/* set max clock state for high phyclock, invalid on exceeding display clock */
+		if (clk_mgr_dce->max_clks_by_state[clk_mgr_dce->max_clks_state].display_clk_khz
+				< context->bw_ctx.bw.dce.dispclk_khz)
+			low_req_clk = DM_PP_CLOCKS_STATE_INVALID;
+		else
+			low_req_clk = clk_mgr_dce->max_clks_state;
+	}
+
+	return low_req_clk;
+}
+
+
+/* TODO: remove use the two broken down functions */
+int dce_set_clock(
+	struct clk_mgr *clk_mgr_base,
+	int requested_clk_khz)
+{
+	struct clk_mgr_internal *clk_mgr_dce = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct bp_pixel_clock_parameters pxl_clk_params = { 0 };
+	struct dc_bios *bp = clk_mgr_base->ctx->dc_bios;
+	int actual_clock = requested_clk_khz;
+	struct dmcu *dmcu = clk_mgr_dce->base.ctx->dc->res_pool->dmcu;
+
+	/* Make sure requested clock isn't lower than minimum threshold*/
+	if (requested_clk_khz > 0)
+		requested_clk_khz = max(requested_clk_khz,
+				clk_mgr_dce->base.dentist_vco_freq_khz / 64);
+
+	/* Prepare to program display clock*/
+	pxl_clk_params.target_pixel_clock_100hz = requested_clk_khz * 10;
+	pxl_clk_params.pll_id = CLOCK_SOURCE_ID_DFS;
+
+	/* DCE 6.0, DCE 6.4: engine clock is the same as PLL0 */
+	if (clk_mgr_base->ctx->dce_version == DCE_VERSION_6_0 ||
+	    clk_mgr_base->ctx->dce_version == DCE_VERSION_6_4)
+		pxl_clk_params.pll_id = CLOCK_SOURCE_ID_PLL0;
+
+	if (clk_mgr_dce->dfs_bypass_active)
+		pxl_clk_params.flags.SET_DISPCLK_DFS_BYPASS = true;
+
+	bp->funcs->program_display_engine_pll(bp, &pxl_clk_params);
+
+	if (clk_mgr_dce->dfs_bypass_active) {
+		/* Cache the fixed display clock*/
+		clk_mgr_dce->dfs_bypass_disp_clk =
+			pxl_clk_params.dfs_bypass_display_clock;
+		actual_clock = pxl_clk_params.dfs_bypass_display_clock;
+	}
+
+	/* from power down, we need mark the clock state as ClocksStateNominal
+	 * from HWReset, so when resume we will call pplib voltage regulator.*/
+	if (requested_clk_khz == 0)
+		clk_mgr_dce->cur_min_clks_state = DM_PP_CLOCKS_STATE_NOMINAL;
+
+	if (dmcu && dmcu->funcs->is_dmcu_initialized(dmcu))
+		dmcu->funcs->set_psr_wait_loop(dmcu, actual_clock / 1000 / 7);
+
+	return actual_clock;
+}
+
+
+static void dce_clock_read_integrated_info(struct clk_mgr_internal *clk_mgr_dce)
+{
+	struct dc_debug_options *debug = &clk_mgr_dce->base.ctx->dc->debug;
+	struct dc_bios *bp = clk_mgr_dce->base.ctx->dc_bios;
+	int i;
+
+	if (bp->integrated_info)
+		clk_mgr_dce->base.dentist_vco_freq_khz = bp->integrated_info->dentist_vco_freq;
+	if (clk_mgr_dce->base.dentist_vco_freq_khz == 0) {
+		clk_mgr_dce->base.dentist_vco_freq_khz = bp->fw_info.smu_gpu_pll_output_freq;
+		if (clk_mgr_dce->base.dentist_vco_freq_khz == 0)
+			clk_mgr_dce->base.dentist_vco_freq_khz = 3600000;
+	}
+
+	/*update the maximum display clock for each power state*/
+	for (i = 0; i < NUMBER_OF_DISP_CLK_VOLTAGE; ++i) {
+		enum dm_pp_clocks_state clk_state = DM_PP_CLOCKS_STATE_INVALID;
+
+		switch (i) {
+		case 0:
+			clk_state = DM_PP_CLOCKS_STATE_ULTRA_LOW;
+			break;
+
+		case 1:
+			clk_state = DM_PP_CLOCKS_STATE_LOW;
+			break;
+
+		case 2:
+			clk_state = DM_PP_CLOCKS_STATE_NOMINAL;
+			break;
+
+		case 3:
+			clk_state = DM_PP_CLOCKS_STATE_PERFORMANCE;
+			break;
+
+		default:
+			clk_state = DM_PP_CLOCKS_STATE_INVALID;
+			break;
+		}
+
+		/*Do not allow bad VBIOS/SBIOS to override with invalid values,
+		 * check for > 100MHz*/
+		if (bp->integrated_info)
+			if (bp->integrated_info->disp_clk_voltage[i].max_supported_clk >= 100000)
+				clk_mgr_dce->max_clks_by_state[clk_state].display_clk_khz =
+					bp->integrated_info->disp_clk_voltage[i].max_supported_clk;
+	}
+
+	if (!debug->disable_dfs_bypass && bp->integrated_info)
+		if (bp->integrated_info->gpu_cap_info & DFS_BYPASS_ENABLE)
+			clk_mgr_dce->dfs_bypass_enabled = true;
+}
+
+void dce_clock_read_ss_info(struct clk_mgr_internal *clk_mgr_dce)
+{
+	struct dc_bios *bp = clk_mgr_dce->base.ctx->dc_bios;
+	int ss_info_num = bp->funcs->get_ss_entry_number(
+			bp, AS_SIGNAL_TYPE_GPU_PLL);
+
+	if (ss_info_num) {
+		struct spread_spectrum_info info = { { 0 } };
+		enum bp_result result = bp->funcs->get_spread_spectrum_info(
+				bp, AS_SIGNAL_TYPE_GPU_PLL, 0, &info);
+
+		/* Based on VBIOS, VBIOS will keep entry for GPU PLL SS
+		 * even if SS not enabled and in that case
+		 * SSInfo.spreadSpectrumPercentage !=0 would be sign
+		 * that SS is enabled
+		 */
+		if (result == BP_RESULT_OK &&
+				info.spread_spectrum_percentage != 0) {
+			clk_mgr_dce->ss_on_dprefclk = true;
+			clk_mgr_dce->dprefclk_ss_divider = info.spread_percentage_divider;
+
+			if (info.type.CENTER_MODE == 0) {
+				/* TODO: Currently for DP Reference clock we
+				 * need only SS percentage for
+				 * downspread */
+				clk_mgr_dce->dprefclk_ss_percentage =
+						info.spread_spectrum_percentage;
+			}
+
+			return;
+		}
+
+		result = bp->funcs->get_spread_spectrum_info(
+				bp, AS_SIGNAL_TYPE_DISPLAY_PORT, 0, &info);
+
+		/* Based on VBIOS, VBIOS will keep entry for DPREFCLK SS
+		 * even if SS not enabled and in that case
+		 * SSInfo.spreadSpectrumPercentage !=0 would be sign
+		 * that SS is enabled
+		 */
+		if (result == BP_RESULT_OK &&
+				info.spread_spectrum_percentage != 0) {
+			clk_mgr_dce->ss_on_dprefclk = true;
+			clk_mgr_dce->dprefclk_ss_divider = info.spread_percentage_divider;
+
+			if (info.type.CENTER_MODE == 0) {
+				/* Currently for DP Reference clock we
+				 * need only SS percentage for
+				 * downspread */
+				clk_mgr_dce->dprefclk_ss_percentage =
+						info.spread_spectrum_percentage;
+			}
+			if (clk_mgr_dce->base.ctx->dc->config.ignore_dpref_ss)
+				clk_mgr_dce->dprefclk_ss_percentage = 0;
+		}
+	}
+}
+
+static void dce_pplib_apply_display_requirements(
+	struct dc *dc,
+	struct dc_state *context)
+{
+	struct dm_pp_display_configuration *pp_display_cfg = &context->pp_display_cfg;
+
+	dce110_fill_display_configs(context, pp_display_cfg);
+
+	if (memcmp(&dc->current_state->pp_display_cfg, pp_display_cfg, sizeof(*pp_display_cfg)) !=  0)
+		dm_pp_apply_display_requirements(dc->ctx, pp_display_cfg);
+}
+
+static void dce_update_clocks(struct clk_mgr *clk_mgr_base,
+			struct dc_state *context,
+			bool safe_to_lower)
+{
+	struct clk_mgr_internal *clk_mgr_dce = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dm_pp_power_level_change_request level_change_req;
+	const int max_disp_clk =
+		clk_mgr_dce->max_clks_by_state[DM_PP_CLOCKS_STATE_PERFORMANCE].display_clk_khz;
+	int patched_disp_clk = MIN(max_disp_clk, context->bw_ctx.bw.dce.dispclk_khz);
+
+	level_change_req.power_level = dce_get_required_clocks_state(clk_mgr_base, context);
+	/* get max clock state from PPLIB */
+	if ((level_change_req.power_level < clk_mgr_dce->cur_min_clks_state && safe_to_lower)
+			|| level_change_req.power_level > clk_mgr_dce->cur_min_clks_state) {
+		if (dm_pp_apply_power_level_change_request(clk_mgr_base->ctx, &level_change_req))
+			clk_mgr_dce->cur_min_clks_state = level_change_req.power_level;
+	}
+
+	if (should_set_clock(safe_to_lower, patched_disp_clk, clk_mgr_base->clks.dispclk_khz)) {
+		patched_disp_clk = dce_set_clock(clk_mgr_base, patched_disp_clk);
+		clk_mgr_base->clks.dispclk_khz = patched_disp_clk;
+	}
+	dce_pplib_apply_display_requirements(clk_mgr_base->ctx->dc, context);
+}
+
+
+
+
+
+
+
+
+static struct clk_mgr_funcs dce_funcs = {
+	.get_dp_ref_clk_frequency = dce_get_dp_ref_freq_khz,
+	.update_clocks = dce_update_clocks
+};
+
+void dce_clk_mgr_construct(
+		struct dc_context *ctx,
+		struct clk_mgr_internal *clk_mgr)
+{
+	struct clk_mgr *base = &clk_mgr->base;
+	struct dm_pp_static_clock_info static_clk_info = {0};
+
+	memcpy(clk_mgr->max_clks_by_state,
+		dce80_max_clks_by_state,
+		sizeof(dce80_max_clks_by_state));
+
+	base->ctx = ctx;
+	base->funcs = &dce_funcs;
+
+	clk_mgr->regs = &disp_clk_regs;
+	clk_mgr->clk_mgr_shift = &disp_clk_shift;
+	clk_mgr->clk_mgr_mask = &disp_clk_mask;
+	clk_mgr->dfs_bypass_disp_clk = 0;
+
+	clk_mgr->dprefclk_ss_percentage = 0;
+	clk_mgr->dprefclk_ss_divider = 1000;
+	clk_mgr->ss_on_dprefclk = false;
+
+	if (dm_pp_get_static_clocks(ctx, &static_clk_info))
+		clk_mgr->max_clks_state = static_clk_info.max_clocks_state;
+	else
+		clk_mgr->max_clks_state = DM_PP_CLOCKS_STATE_NOMINAL;
+	clk_mgr->cur_min_clks_state = DM_PP_CLOCKS_STATE_INVALID;
+
+	base->clks.max_supported_dispclk_khz =
+		clk_mgr->max_clks_by_state[DM_PP_CLOCKS_STATE_PERFORMANCE].display_clk_khz;
+
+	dce_clock_read_integrated_info(clk_mgr);
+	dce_clock_read_ss_info(clk_mgr);
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dce100/dce_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dce100/dce_clk_mgr.h
new file mode 100644
index 000000000000..f6622f58f62e
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dce100/dce_clk_mgr.h
@@ -0,0 +1,59 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+#ifndef _DCE_CLK_MGR_H_
+#define _DCE_CLK_MGR_H_
+
+#include "dc.h"
+
+/* functions shared by other dce clk mgrs */
+int dce_adjust_dp_ref_freq_for_ss(struct clk_mgr_internal *clk_mgr_dce, int dp_ref_clk_khz);
+int dce_get_dp_ref_freq_khz(struct clk_mgr *clk_mgr_base);
+enum dm_pp_clocks_state dce_get_required_clocks_state(
+	struct clk_mgr *clk_mgr_base,
+	struct dc_state *context);
+
+uint32_t dce_get_max_pixel_clock_for_all_paths(struct dc_state *context);
+
+
+void dce_clk_mgr_construct(
+		struct dc_context *ctx,
+		struct clk_mgr_internal *clk_mgr_dce);
+
+void dce_clock_read_ss_info(struct clk_mgr_internal *dccg_dce);
+
+int dce12_get_dp_ref_freq_khz(struct clk_mgr *dccg);
+
+int dce_set_clock(
+	struct clk_mgr *clk_mgr_base,
+	int requested_clk_khz);
+
+
+void dce_clk_mgr_destroy(struct clk_mgr **clk_mgr);
+
+int dentist_get_divider_from_did(int did);
+
+#endif /* _DCE_CLK_MGR_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dce110/dce110_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dce110/dce110_clk_mgr.c
new file mode 100644
index 000000000000..d50b9440210e
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dce110/dce110_clk_mgr.c
@@ -0,0 +1,302 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "core_types.h"
+#include "clk_mgr_internal.h"
+
+#include "dce/dce_11_0_d.h"
+#include "dce/dce_11_0_sh_mask.h"
+#include "dce110_clk_mgr.h"
+#include "../clk_mgr/dce100/dce_clk_mgr.h"
+
+/* set register offset */
+#define SR(reg_name)\
+	.reg_name = mm ## reg_name
+
+/* set register offset with instance */
+#define SRI(reg_name, block, id)\
+	.reg_name = mm ## block ## id ## _ ## reg_name
+
+static const struct clk_mgr_registers disp_clk_regs = {
+		CLK_COMMON_REG_LIST_DCE_BASE()
+};
+
+static const struct clk_mgr_shift disp_clk_shift = {
+		CLK_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(__SHIFT)
+};
+
+static const struct clk_mgr_mask disp_clk_mask = {
+		CLK_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(_MASK)
+};
+
+static const struct state_dependent_clocks dce110_max_clks_by_state[] = {
+/*ClocksStateInvalid - should not be used*/
+{ .display_clk_khz = 0, .pixel_clk_khz = 0 },
+/*ClocksStateUltraLow - currently by HW design team not supposed to be used*/
+{ .display_clk_khz = 352000, .pixel_clk_khz = 330000 },
+/*ClocksStateLow*/
+{ .display_clk_khz = 352000, .pixel_clk_khz = 330000 },
+/*ClocksStateNominal*/
+{ .display_clk_khz = 467000, .pixel_clk_khz = 400000 },
+/*ClocksStatePerformance*/
+{ .display_clk_khz = 643000, .pixel_clk_khz = 400000 } };
+
+static int determine_sclk_from_bounding_box(
+		const struct dc *dc,
+		int required_sclk)
+{
+	int i;
+
+	/*
+	 * Some asics do not give us sclk levels, so we just report the actual
+	 * required sclk
+	 */
+	if (dc->sclk_lvls.num_levels == 0)
+		return required_sclk;
+
+	for (i = 0; i < dc->sclk_lvls.num_levels; i++) {
+		if (dc->sclk_lvls.clocks_in_khz[i] >= required_sclk)
+			return dc->sclk_lvls.clocks_in_khz[i];
+	}
+	/*
+	 * even maximum level could not satisfy requirement, this
+	 * is unexpected at this stage, should have been caught at
+	 * validation time
+	 */
+	ASSERT(0);
+	return dc->sclk_lvls.clocks_in_khz[dc->sclk_lvls.num_levels - 1];
+}
+
+uint32_t dce110_get_min_vblank_time_us(const struct dc_state *context)
+{
+	uint8_t j;
+	uint32_t min_vertical_blank_time = -1;
+
+	for (j = 0; j < context->stream_count; j++) {
+		struct dc_stream_state *stream = context->streams[j];
+		uint32_t vertical_blank_in_pixels = 0;
+		uint32_t vertical_blank_time = 0;
+		uint32_t vertical_total_min = stream->timing.v_total;
+		struct dc_crtc_timing_adjust adjust = stream->adjust;
+		if (adjust.v_total_max != adjust.v_total_min)
+			vertical_total_min = adjust.v_total_min;
+
+		vertical_blank_in_pixels = stream->timing.h_total *
+			(vertical_total_min
+			 - stream->timing.v_addressable);
+		vertical_blank_time = vertical_blank_in_pixels
+			* 10000 / stream->timing.pix_clk_100hz;
+
+		if (min_vertical_blank_time > vertical_blank_time)
+			min_vertical_blank_time = vertical_blank_time;
+	}
+
+	return min_vertical_blank_time;
+}
+
+void dce110_fill_display_configs(
+	const struct dc_state *context,
+	struct dm_pp_display_configuration *pp_display_cfg)
+{
+	struct dc *dc = context->clk_mgr->ctx->dc;
+	int j;
+	int num_cfgs = 0;
+
+	pp_display_cfg->avail_mclk_switch_time_us = dce110_get_min_vblank_time_us(context);
+	pp_display_cfg->disp_clk_khz = dc->clk_mgr->clks.dispclk_khz;
+	pp_display_cfg->avail_mclk_switch_time_in_disp_active_us = 0;
+	pp_display_cfg->crtc_index = dc->res_pool->res_cap->num_timing_generator;
+
+	for (j = 0; j < context->stream_count; j++) {
+		int k;
+
+		const struct dc_stream_state *stream = context->streams[j];
+		struct dm_pp_single_disp_config *cfg =
+			&pp_display_cfg->disp_configs[num_cfgs];
+		const struct pipe_ctx *pipe_ctx = NULL;
+
+		for (k = 0; k < MAX_PIPES; k++)
+			if (stream == context->res_ctx.pipe_ctx[k].stream) {
+				pipe_ctx = &context->res_ctx.pipe_ctx[k];
+				break;
+			}
+
+		ASSERT(pipe_ctx != NULL);
+
+		/* only notify active stream */
+		if (stream->dpms_off)
+			continue;
+
+		num_cfgs++;
+		cfg->signal = pipe_ctx->stream->signal;
+		cfg->pipe_idx = pipe_ctx->stream_res.tg->inst;
+		cfg->src_height = stream->src.height;
+		cfg->src_width = stream->src.width;
+		cfg->ddi_channel_mapping =
+			stream->link->ddi_channel_mapping.raw;
+		cfg->transmitter =
+			stream->link->link_enc->transmitter;
+		cfg->link_settings.lane_count =
+			stream->link->cur_link_settings.lane_count;
+		cfg->link_settings.link_rate =
+			stream->link->cur_link_settings.link_rate;
+		cfg->link_settings.link_spread =
+			stream->link->cur_link_settings.link_spread;
+		cfg->pixel_clock = stream->phy_pix_clk;
+		/* Round v_refresh*/
+		cfg->v_refresh = stream->timing.pix_clk_100hz * 100;
+		cfg->v_refresh /= stream->timing.h_total;
+		cfg->v_refresh = (cfg->v_refresh + stream->timing.v_total / 2)
+							/ stream->timing.v_total;
+
+		/* Find first CRTC index and calculate its line time.
+		 * This is necessary for DPM on SI GPUs.
+		 */
+		if (cfg->pipe_idx < pp_display_cfg->crtc_index) {
+			const struct dc_crtc_timing *timing =
+				&context->streams[0]->timing;
+
+			pp_display_cfg->crtc_index = cfg->pipe_idx;
+			pp_display_cfg->line_time_in_us =
+				timing->h_total * 10000 / timing->pix_clk_100hz;
+		}
+	}
+
+	if (!num_cfgs) {
+		pp_display_cfg->crtc_index = 0;
+		pp_display_cfg->line_time_in_us = 0;
+	}
+
+	pp_display_cfg->display_count = num_cfgs;
+}
+
+void dce11_pplib_apply_display_requirements(
+	struct dc *dc,
+	struct dc_state *context)
+{
+	struct dm_pp_display_configuration *pp_display_cfg = &context->pp_display_cfg;
+	int memory_type_multiplier = MEMORY_TYPE_MULTIPLIER_CZ;
+
+	if (dc->bw_vbios && dc->bw_vbios->memory_type == bw_def_hbm)
+		memory_type_multiplier = MEMORY_TYPE_HBM;
+
+	pp_display_cfg->all_displays_in_sync =
+		context->bw_ctx.bw.dce.all_displays_in_sync;
+	pp_display_cfg->nb_pstate_switch_disable =
+			context->bw_ctx.bw.dce.nbp_state_change_enable == false;
+	pp_display_cfg->cpu_cc6_disable =
+			context->bw_ctx.bw.dce.cpuc_state_change_enable == false;
+	pp_display_cfg->cpu_pstate_disable =
+			context->bw_ctx.bw.dce.cpup_state_change_enable == false;
+	pp_display_cfg->cpu_pstate_separation_time =
+			context->bw_ctx.bw.dce.blackout_recovery_time_us;
+
+	/*
+	 * TODO: determine whether the bandwidth has reached memory's limitation
+	 * , then change minimum memory clock based on real-time bandwidth
+	 * limitation.
+	 */
+	if (dc->bw_vbios && (dc->ctx->asic_id.chip_family == FAMILY_AI) &&
+	     ASICREV_IS_VEGA20_P(dc->ctx->asic_id.hw_internal_rev) && (context->stream_count >= 2)) {
+		pp_display_cfg->min_memory_clock_khz = max(pp_display_cfg->min_memory_clock_khz,
+							   (uint32_t) div64_s64(
+								   div64_s64(dc->bw_vbios->high_yclk.value,
+									     memory_type_multiplier), 10000));
+	} else {
+		pp_display_cfg->min_memory_clock_khz = context->bw_ctx.bw.dce.yclk_khz
+			/ memory_type_multiplier;
+	}
+
+	pp_display_cfg->min_engine_clock_khz = determine_sclk_from_bounding_box(
+			dc,
+			context->bw_ctx.bw.dce.sclk_khz);
+
+	/*
+	 * As workaround for >4x4K lightup set dcfclock to min_engine_clock value.
+	 * This is not required for less than 5 displays,
+	 * thus don't request decfclk in dc to avoid impact
+	 * on power saving.
+	 *
+	 */
+	pp_display_cfg->min_dcfclock_khz = (context->stream_count > 4) ?
+			pp_display_cfg->min_engine_clock_khz : 0;
+
+	pp_display_cfg->min_engine_clock_deep_sleep_khz
+			= context->bw_ctx.bw.dce.sclk_deep_sleep_khz;
+
+	dce110_fill_display_configs(context, pp_display_cfg);
+
+	if (memcmp(&dc->current_state->pp_display_cfg, pp_display_cfg, sizeof(*pp_display_cfg)) !=  0)
+		dm_pp_apply_display_requirements(dc->ctx, pp_display_cfg);
+}
+
+static void dce11_update_clocks(struct clk_mgr *clk_mgr_base,
+			struct dc_state *context,
+			bool safe_to_lower)
+{
+	struct clk_mgr_internal *clk_mgr_dce = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dm_pp_power_level_change_request level_change_req;
+	int patched_disp_clk = context->bw_ctx.bw.dce.dispclk_khz;
+
+	/*TODO: W/A for dal3 linux, investigate why this works */
+	if (!clk_mgr_dce->dfs_bypass_active)
+		patched_disp_clk = patched_disp_clk * 115 / 100;
+
+	level_change_req.power_level = dce_get_required_clocks_state(clk_mgr_base, context);
+	/* get max clock state from PPLIB */
+	if ((level_change_req.power_level < clk_mgr_dce->cur_min_clks_state && safe_to_lower)
+			|| level_change_req.power_level > clk_mgr_dce->cur_min_clks_state) {
+		if (dm_pp_apply_power_level_change_request(clk_mgr_base->ctx, &level_change_req))
+			clk_mgr_dce->cur_min_clks_state = level_change_req.power_level;
+	}
+
+	if (should_set_clock(safe_to_lower, patched_disp_clk, clk_mgr_base->clks.dispclk_khz)) {
+		context->bw_ctx.bw.dce.dispclk_khz = dce_set_clock(clk_mgr_base, patched_disp_clk);
+		clk_mgr_base->clks.dispclk_khz = patched_disp_clk;
+	}
+	dce11_pplib_apply_display_requirements(clk_mgr_base->ctx->dc, context);
+}
+
+static struct clk_mgr_funcs dce110_funcs = {
+	.get_dp_ref_clk_frequency = dce_get_dp_ref_freq_khz,
+	.update_clocks = dce11_update_clocks
+};
+
+void dce110_clk_mgr_construct(
+		struct dc_context *ctx,
+		struct clk_mgr_internal *clk_mgr)
+{
+	dce_clk_mgr_construct(ctx, clk_mgr);
+
+	memcpy(clk_mgr->max_clks_by_state,
+		dce110_max_clks_by_state,
+		sizeof(dce110_max_clks_by_state));
+
+	clk_mgr->regs = &disp_clk_regs;
+	clk_mgr->clk_mgr_shift = &disp_clk_shift;
+	clk_mgr->clk_mgr_mask = &disp_clk_mask;
+	clk_mgr->base.funcs = &dce110_funcs;
+
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dce110/dce110_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dce110/dce110_clk_mgr.h
new file mode 100644
index 000000000000..c0eb2ea6fb3a
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dce110/dce110_clk_mgr.h
@@ -0,0 +1,44 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DAL_DC_DCE_DCE110_CLK_MGR_H_
+#define DAL_DC_DCE_DCE110_CLK_MGR_H_
+
+void dce110_clk_mgr_construct(
+		struct dc_context *ctx,
+		struct clk_mgr_internal *clk_mgr);
+
+void dce110_fill_display_configs(
+	const struct dc_state *context,
+	struct dm_pp_display_configuration *pp_display_cfg);
+
+/* functions shared with other clk mgr*/
+void dce11_pplib_apply_display_requirements(
+	struct dc *dc,
+	struct dc_state *context);
+
+uint32_t dce110_get_min_vblank_time_us(const struct dc_state *context);
+
+#endif /* DAL_DC_DCE_DCE110_CLK_MGR_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dce112/dce112_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dce112/dce112_clk_mgr.c
new file mode 100644
index 000000000000..1f36ad8a7de4
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dce112/dce112_clk_mgr.c
@@ -0,0 +1,237 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "core_types.h"
+#include "clk_mgr_internal.h"
+
+#include "dce/dce_11_2_d.h"
+#include "dce/dce_11_2_sh_mask.h"
+#include "dce100/dce_clk_mgr.h"
+#include "dce110/dce110_clk_mgr.h"
+#include "dce112_clk_mgr.h"
+#include "dal_asic_id.h"
+
+/* set register offset */
+#define SR(reg_name)\
+	.reg_name = mm ## reg_name
+
+/* set register offset with instance */
+#define SRI(reg_name, block, id)\
+	.reg_name = mm ## block ## id ## _ ## reg_name
+
+static const struct clk_mgr_registers disp_clk_regs = {
+		CLK_COMMON_REG_LIST_DCE_BASE()
+};
+
+static const struct clk_mgr_shift disp_clk_shift = {
+		CLK_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(__SHIFT)
+};
+
+static const struct clk_mgr_mask disp_clk_mask = {
+		CLK_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(_MASK)
+};
+
+static const struct state_dependent_clocks dce112_max_clks_by_state[] = {
+/*ClocksStateInvalid - should not be used*/
+{ .display_clk_khz = 0, .pixel_clk_khz = 0 },
+/*ClocksStateUltraLow - currently by HW design team not supposed to be used*/
+{ .display_clk_khz = 389189, .pixel_clk_khz = 346672 },
+/*ClocksStateLow*/
+{ .display_clk_khz = 459000, .pixel_clk_khz = 400000 },
+/*ClocksStateNominal*/
+{ .display_clk_khz = 667000, .pixel_clk_khz = 600000 },
+/*ClocksStatePerformance*/
+{ .display_clk_khz = 1132000, .pixel_clk_khz = 600000 } };
+
+
+//TODO: remove use the two broken down functions
+int dce112_set_clock(struct clk_mgr *clk_mgr_base, int requested_clk_khz)
+{
+	struct clk_mgr_internal *clk_mgr_dce = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct bp_set_dce_clock_parameters dce_clk_params;
+	struct dc_bios *bp = clk_mgr_base->ctx->dc_bios;
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	struct dmcu *dmcu = dc->res_pool->dmcu;
+	int actual_clock = requested_clk_khz;
+	/* Prepare to program display clock*/
+	memset(&dce_clk_params, 0, sizeof(dce_clk_params));
+
+	/* Make sure requested clock isn't lower than minimum threshold*/
+	requested_clk_khz = max(requested_clk_khz,
+				clk_mgr_dce->base.dentist_vco_freq_khz / 62);
+
+	dce_clk_params.target_clock_frequency = requested_clk_khz;
+	dce_clk_params.pll_id = CLOCK_SOURCE_ID_DFS;
+	dce_clk_params.clock_type = DCECLOCK_TYPE_DISPLAY_CLOCK;
+
+	bp->funcs->set_dce_clock(bp, &dce_clk_params);
+	actual_clock = dce_clk_params.target_clock_frequency;
+
+	/*
+	 * from power down, we need mark the clock state as ClocksStateNominal
+	 * from HWReset, so when resume we will call pplib voltage regulator.
+	 */
+	if (requested_clk_khz == 0)
+		clk_mgr_dce->cur_min_clks_state = DM_PP_CLOCKS_STATE_NOMINAL;
+
+	/*Program DP ref Clock*/
+	/*VBIOS will determine DPREFCLK frequency, so we don't set it*/
+	dce_clk_params.target_clock_frequency = 0;
+	dce_clk_params.clock_type = DCECLOCK_TYPE_DPREFCLK;
+
+	if (!((clk_mgr_base->ctx->asic_id.chip_family == FAMILY_AI) &&
+	       ASICREV_IS_VEGA20_P(clk_mgr_base->ctx->asic_id.hw_internal_rev)))
+		dce_clk_params.flags.USE_GENLOCK_AS_SOURCE_FOR_DPREFCLK =
+			(dce_clk_params.pll_id ==
+					CLOCK_SOURCE_COMBO_DISPLAY_PLL0);
+	else
+		dce_clk_params.flags.USE_GENLOCK_AS_SOURCE_FOR_DPREFCLK = false;
+
+	bp->funcs->set_dce_clock(bp, &dce_clk_params);
+
+	if (dmcu && dmcu->funcs->is_dmcu_initialized(dmcu)) {
+		if (clk_mgr_dce->dfs_bypass_disp_clk != actual_clock)
+			dmcu->funcs->set_psr_wait_loop(dmcu,
+					actual_clock / 1000 / 7);
+	}
+
+	clk_mgr_dce->dfs_bypass_disp_clk = actual_clock;
+	return actual_clock;
+}
+
+int dce112_set_dispclk(struct clk_mgr_internal *clk_mgr, int requested_clk_khz)
+{
+	struct bp_set_dce_clock_parameters dce_clk_params;
+	struct dc_bios *bp = clk_mgr->base.ctx->dc_bios;
+	struct dc *dc = clk_mgr->base.ctx->dc;
+	struct dmcu *dmcu = dc->res_pool->dmcu;
+	int actual_clock = requested_clk_khz;
+	/* Prepare to program display clock*/
+	memset(&dce_clk_params, 0, sizeof(dce_clk_params));
+
+	/* Make sure requested clock isn't lower than minimum threshold*/
+	if (requested_clk_khz > 0)
+		requested_clk_khz = max(requested_clk_khz,
+				clk_mgr->base.dentist_vco_freq_khz / 62);
+
+	dce_clk_params.target_clock_frequency = requested_clk_khz;
+	dce_clk_params.pll_id = CLOCK_SOURCE_ID_DFS;
+	dce_clk_params.clock_type = DCECLOCK_TYPE_DISPLAY_CLOCK;
+
+	bp->funcs->set_dce_clock(bp, &dce_clk_params);
+	actual_clock = dce_clk_params.target_clock_frequency;
+
+	/*
+	 * from power down, we need mark the clock state as ClocksStateNominal
+	 * from HWReset, so when resume we will call pplib voltage regulator.
+	 */
+	if (requested_clk_khz == 0)
+		clk_mgr->cur_min_clks_state = DM_PP_CLOCKS_STATE_NOMINAL;
+
+
+	if (dmcu && dmcu->funcs->is_dmcu_initialized(dmcu)) {
+		if (clk_mgr->dfs_bypass_disp_clk != actual_clock)
+			dmcu->funcs->set_psr_wait_loop(dmcu,
+					actual_clock / 1000 / 7);
+	}
+
+	clk_mgr->dfs_bypass_disp_clk = actual_clock;
+	return actual_clock;
+
+}
+
+int dce112_set_dprefclk(struct clk_mgr_internal *clk_mgr)
+{
+	struct bp_set_dce_clock_parameters dce_clk_params;
+	struct dc_bios *bp = clk_mgr->base.ctx->dc_bios;
+
+	memset(&dce_clk_params, 0, sizeof(dce_clk_params));
+
+	/*Program DP ref Clock*/
+	/*VBIOS will determine DPREFCLK frequency, so we don't set it*/
+	dce_clk_params.target_clock_frequency = 0;
+	dce_clk_params.pll_id = CLOCK_SOURCE_ID_DFS;
+	dce_clk_params.clock_type = DCECLOCK_TYPE_DPREFCLK;
+	if (!((clk_mgr->base.ctx->asic_id.chip_family == FAMILY_AI) &&
+	       ASICREV_IS_VEGA20_P(clk_mgr->base.ctx->asic_id.hw_internal_rev)))
+		dce_clk_params.flags.USE_GENLOCK_AS_SOURCE_FOR_DPREFCLK =
+			(dce_clk_params.pll_id ==
+					CLOCK_SOURCE_COMBO_DISPLAY_PLL0);
+	else
+		dce_clk_params.flags.USE_GENLOCK_AS_SOURCE_FOR_DPREFCLK = false;
+
+	bp->funcs->set_dce_clock(bp, &dce_clk_params);
+
+	/* Returns the dp_refclk that was set */
+	return dce_clk_params.target_clock_frequency;
+}
+
+static void dce112_update_clocks(struct clk_mgr *clk_mgr_base,
+			struct dc_state *context,
+			bool safe_to_lower)
+{
+	struct clk_mgr_internal *clk_mgr_dce = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dm_pp_power_level_change_request level_change_req;
+	int patched_disp_clk = context->bw_ctx.bw.dce.dispclk_khz;
+
+	/*TODO: W/A for dal3 linux, investigate why this works */
+	if (!clk_mgr_dce->dfs_bypass_active)
+		patched_disp_clk = patched_disp_clk * 115 / 100;
+
+	level_change_req.power_level = dce_get_required_clocks_state(clk_mgr_base, context);
+	/* get max clock state from PPLIB */
+	if ((level_change_req.power_level < clk_mgr_dce->cur_min_clks_state && safe_to_lower)
+			|| level_change_req.power_level > clk_mgr_dce->cur_min_clks_state) {
+		if (dm_pp_apply_power_level_change_request(clk_mgr_base->ctx, &level_change_req))
+			clk_mgr_dce->cur_min_clks_state = level_change_req.power_level;
+	}
+
+	if (should_set_clock(safe_to_lower, patched_disp_clk, clk_mgr_base->clks.dispclk_khz)) {
+		patched_disp_clk = dce112_set_clock(clk_mgr_base, patched_disp_clk);
+		clk_mgr_base->clks.dispclk_khz = patched_disp_clk;
+	}
+	dce11_pplib_apply_display_requirements(clk_mgr_base->ctx->dc, context);
+}
+
+static struct clk_mgr_funcs dce112_funcs = {
+	.get_dp_ref_clk_frequency = dce_get_dp_ref_freq_khz,
+	.update_clocks = dce112_update_clocks
+};
+
+void dce112_clk_mgr_construct(
+		struct dc_context *ctx,
+		struct clk_mgr_internal *clk_mgr)
+{
+	dce_clk_mgr_construct(ctx, clk_mgr);
+
+	memcpy(clk_mgr->max_clks_by_state,
+		dce112_max_clks_by_state,
+		sizeof(dce112_max_clks_by_state));
+
+	clk_mgr->regs = &disp_clk_regs;
+	clk_mgr->clk_mgr_shift = &disp_clk_shift;
+	clk_mgr->clk_mgr_mask = &disp_clk_mask;
+	clk_mgr->base.funcs = &dce112_funcs;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dce112/dce112_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dce112/dce112_clk_mgr.h
new file mode 100644
index 000000000000..dfb06db118e1
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dce112/dce112_clk_mgr.h
@@ -0,0 +1,39 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DAL_DC_DCE_DCE112_CLK_MGR_H_
+#define DAL_DC_DCE_DCE112_CLK_MGR_H_
+
+
+void dce112_clk_mgr_construct(
+		struct dc_context *ctx,
+		struct clk_mgr_internal *clk_mgr);
+
+/* functions shared with other clk mgr */
+int dce112_set_clock(struct clk_mgr *clk_mgr_base, int requested_clk_khz);
+int dce112_set_dispclk(struct clk_mgr_internal *clk_mgr, int requested_clk_khz);
+int dce112_set_dprefclk(struct clk_mgr_internal *clk_mgr);
+
+#endif /* DAL_DC_DCE_DCE112_CLK_MGR_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dce120/dce120_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dce120/dce120_clk_mgr.c
new file mode 100644
index 000000000000..c9ba7b3fd2c3
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dce120/dce120_clk_mgr.c
@@ -0,0 +1,153 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "core_types.h"
+#include "clk_mgr_internal.h"
+
+#include "dce112/dce112_clk_mgr.h"
+#include "dce110/dce110_clk_mgr.h"
+#include "dce120_clk_mgr.h"
+#include "dce100/dce_clk_mgr.h"
+#include "dce120/dce120_hwseq.h"
+
+static const struct state_dependent_clocks dce120_max_clks_by_state[] = {
+/*ClocksStateInvalid - should not be used*/
+{ .display_clk_khz = 0, .pixel_clk_khz = 0 },
+/*ClocksStateUltraLow - currently by HW design team not supposed to be used*/
+{ .display_clk_khz = 0, .pixel_clk_khz = 0 },
+/*ClocksStateLow*/
+{ .display_clk_khz = 460000, .pixel_clk_khz = 400000 },
+/*ClocksStateNominal*/
+{ .display_clk_khz = 670000, .pixel_clk_khz = 600000 },
+/*ClocksStatePerformance*/
+{ .display_clk_khz = 1133000, .pixel_clk_khz = 600000 } };
+
+/**
+ * dce121_clock_patch_xgmi_ss_info() - Save XGMI spread spectrum info
+ * @clk_mgr_dce: clock manager internal structure
+ *
+ * Reads from VBIOS the XGMI spread spectrum info and saves it within
+ * the dce clock manager. This operation will overwrite the existing dprefclk
+ * SS values if the vBIOS query succeeds. Otherwise, it does nothing. It also
+ * sets the ->xgmi_enabled flag.
+ */
+static void dce121_clock_patch_xgmi_ss_info(struct clk_mgr_internal *clk_mgr_dce)
+{
+	enum bp_result result;
+	struct spread_spectrum_info info = { { 0 } };
+	struct dc_bios *bp = clk_mgr_dce->base.ctx->dc_bios;
+
+	clk_mgr_dce->xgmi_enabled = false;
+
+	result = bp->funcs->get_spread_spectrum_info(bp, AS_SIGNAL_TYPE_XGMI,
+						     0, &info);
+	if (result == BP_RESULT_OK && info.spread_spectrum_percentage != 0) {
+		clk_mgr_dce->xgmi_enabled = true;
+		clk_mgr_dce->ss_on_dprefclk = true;
+		clk_mgr_dce->dprefclk_ss_divider =
+				info.spread_percentage_divider;
+
+		if (info.type.CENTER_MODE == 0) {
+			/*
+			 * Currently for DP Reference clock we
+			 * need only SS percentage for
+			 * downspread
+			 */
+			clk_mgr_dce->dprefclk_ss_percentage =
+					info.spread_spectrum_percentage;
+		}
+	}
+}
+
+static void dce12_update_clocks(struct clk_mgr *clk_mgr_base,
+			struct dc_state *context,
+			bool safe_to_lower)
+{
+	struct clk_mgr_internal *clk_mgr_dce = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dm_pp_clock_for_voltage_req clock_voltage_req = {0};
+	int max_pix_clk = dce_get_max_pixel_clock_for_all_paths(context);
+	int patched_disp_clk = context->bw_ctx.bw.dce.dispclk_khz;
+
+	/*TODO: W/A for dal3 linux, investigate why this works */
+	if (!clk_mgr_dce->dfs_bypass_active)
+		patched_disp_clk = patched_disp_clk * 115 / 100;
+
+	if (should_set_clock(safe_to_lower, patched_disp_clk, clk_mgr_base->clks.dispclk_khz)) {
+		clock_voltage_req.clk_type = DM_PP_CLOCK_TYPE_DISPLAY_CLK;
+		/*
+		 * When xGMI is enabled, the display clk needs to be adjusted
+		 * with the WAFL link's SS percentage.
+		 */
+		if (clk_mgr_dce->xgmi_enabled)
+			patched_disp_clk = dce_adjust_dp_ref_freq_for_ss(
+					clk_mgr_dce, patched_disp_clk);
+		clock_voltage_req.clocks_in_khz = patched_disp_clk;
+		clk_mgr_base->clks.dispclk_khz = dce112_set_clock(clk_mgr_base, patched_disp_clk);
+
+		dm_pp_apply_clock_for_voltage_request(clk_mgr_base->ctx, &clock_voltage_req);
+	}
+
+	if (should_set_clock(safe_to_lower, max_pix_clk, clk_mgr_base->clks.phyclk_khz)) {
+		clock_voltage_req.clk_type = DM_PP_CLOCK_TYPE_DISPLAYPHYCLK;
+		clock_voltage_req.clocks_in_khz = max_pix_clk;
+		clk_mgr_base->clks.phyclk_khz = max_pix_clk;
+
+		dm_pp_apply_clock_for_voltage_request(clk_mgr_base->ctx, &clock_voltage_req);
+	}
+	dce11_pplib_apply_display_requirements(clk_mgr_base->ctx->dc, context);
+}
+
+
+static struct clk_mgr_funcs dce120_funcs = {
+	.get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz,
+	.update_clocks = dce12_update_clocks
+};
+
+void dce120_clk_mgr_construct(struct dc_context *ctx, struct clk_mgr_internal *clk_mgr)
+{
+	dce_clk_mgr_construct(ctx, clk_mgr);
+
+	memcpy(clk_mgr->max_clks_by_state,
+		dce120_max_clks_by_state,
+		sizeof(dce120_max_clks_by_state));
+
+	clk_mgr->base.dprefclk_khz = 600000;
+	clk_mgr->base.funcs = &dce120_funcs;
+}
+
+void dce121_clk_mgr_construct(struct dc_context *ctx, struct clk_mgr_internal *clk_mgr)
+{
+	dce120_clk_mgr_construct(ctx, clk_mgr);
+	clk_mgr->base.dprefclk_khz = 625000;
+
+	/*
+	 * The xGMI enabled info is used to determine if audio and display
+	 * clocks need to be adjusted with the WAFL link's SS info.
+	 */
+	if (dce121_xgmi_enabled(ctx->dc->hwseq))
+		dce121_clock_patch_xgmi_ss_info(clk_mgr);
+
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dce120/dce120_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dce120/dce120_clk_mgr.h
new file mode 100644
index 000000000000..d12d6fcb167d
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dce120/dce120_clk_mgr.h
@@ -0,0 +1,34 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DAL_DC_DCE_DCE120_CLK_MGR_H_
+#define DAL_DC_DCE_DCE120_CLK_MGR_H_
+
+void dce120_clk_mgr_construct(struct dc_context *ctx, struct clk_mgr_internal *clk_mgr);
+void dce121_clk_mgr_construct(struct dc_context *ctx, struct clk_mgr_internal *clk_mgr);
+
+
+
+#endif /* DAL_DC_DCE_DCE120_CLK_MGR_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dce60/dce60_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dce60/dce60_clk_mgr.c
new file mode 100644
index 000000000000..69dd80d9f738
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dce60/dce60_clk_mgr.c
@@ -0,0 +1,166 @@
+/*
+ * Copyright 2020 Mauro Rossi <issor.oruam@gmail.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+#include "dccg.h"
+#include "clk_mgr_internal.h"
+#include "dce100/dce_clk_mgr.h"
+#include "dce110/dce110_clk_mgr.h"
+#include "dce60_clk_mgr.h"
+#include "reg_helper.h"
+#include "dmcu.h"
+#include "core_types.h"
+#include "dal_asic_id.h"
+
+/*
+ * Currently the register shifts and masks in this file are used for dce60
+ * which has no DPREFCLK_CNTL register
+ * TODO: remove this when DENTIST_DISPCLK_CNTL
+ * is moved to dccg, where it belongs
+ */
+#include "dce/dce_6_0_d.h"
+#include "dce/dce_6_0_sh_mask.h"
+
+#define REG(reg) \
+	(clk_mgr->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	clk_mgr->clk_mgr_shift->field_name, clk_mgr->clk_mgr_mask->field_name
+
+/* set register offset */
+#define SR(reg_name)\
+	.reg_name = mm ## reg_name
+
+static const struct clk_mgr_registers disp_clk_regs = {
+		CLK_COMMON_REG_LIST_DCE60_BASE()
+};
+
+static const struct clk_mgr_shift disp_clk_shift = {
+		CLK_COMMON_MASK_SH_LIST_DCE60_COMMON_BASE(__SHIFT)
+};
+
+static const struct clk_mgr_mask disp_clk_mask = {
+		CLK_COMMON_MASK_SH_LIST_DCE60_COMMON_BASE(_MASK)
+};
+
+
+/* Max clock values for each state indexed by "enum clocks_state": */
+static const struct state_dependent_clocks dce60_max_clks_by_state[] = {
+/* ClocksStateInvalid - should not be used */
+{ .display_clk_khz = 0, .pixel_clk_khz = 0 },
+/* ClocksStateUltraLow - not expected to be used for DCE 6.0 */
+{ .display_clk_khz = 0, .pixel_clk_khz = 0 },
+/* ClocksStateLow */
+{ .display_clk_khz = 352000, .pixel_clk_khz = 330000},
+/* ClocksStateNominal */
+{ .display_clk_khz = 600000, .pixel_clk_khz = 400000 },
+/* ClocksStatePerformance */
+{ .display_clk_khz = 600000, .pixel_clk_khz = 400000 } };
+
+static int dce60_get_dp_ref_freq_khz(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dc_context *ctx = clk_mgr_base->ctx;
+	int dp_ref_clk_khz = 0;
+
+	if (ASIC_REV_IS_TAHITI_P(ctx->asic_id.hw_internal_rev))
+		dp_ref_clk_khz = ctx->dc_bios->fw_info.default_display_engine_pll_frequency;
+	else
+		dp_ref_clk_khz = clk_mgr_base->clks.dispclk_khz;
+
+	return dce_adjust_dp_ref_freq_for_ss(clk_mgr, dp_ref_clk_khz);
+}
+
+static void dce60_pplib_apply_display_requirements(
+	struct dc *dc,
+	struct dc_state *context)
+{
+	struct dm_pp_display_configuration *pp_display_cfg = &context->pp_display_cfg;
+
+	dce110_fill_display_configs(context, pp_display_cfg);
+
+	if (memcmp(&dc->current_state->pp_display_cfg, pp_display_cfg, sizeof(*pp_display_cfg)) !=  0)
+		dm_pp_apply_display_requirements(dc->ctx, pp_display_cfg);
+}
+
+static void dce60_update_clocks(struct clk_mgr *clk_mgr_base,
+			struct dc_state *context,
+			bool safe_to_lower)
+{
+	struct clk_mgr_internal *clk_mgr_dce = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dm_pp_power_level_change_request level_change_req;
+	const int max_disp_clk =
+		clk_mgr_dce->max_clks_by_state[DM_PP_CLOCKS_STATE_PERFORMANCE].display_clk_khz;
+	int patched_disp_clk = MIN(max_disp_clk, context->bw_ctx.bw.dce.dispclk_khz);
+
+	level_change_req.power_level = dce_get_required_clocks_state(clk_mgr_base, context);
+	/* get max clock state from PPLIB */
+	if ((level_change_req.power_level < clk_mgr_dce->cur_min_clks_state && safe_to_lower)
+			|| level_change_req.power_level > clk_mgr_dce->cur_min_clks_state) {
+		if (dm_pp_apply_power_level_change_request(clk_mgr_base->ctx, &level_change_req))
+			clk_mgr_dce->cur_min_clks_state = level_change_req.power_level;
+	}
+
+	if (should_set_clock(safe_to_lower, patched_disp_clk, clk_mgr_base->clks.dispclk_khz)) {
+		patched_disp_clk = dce_set_clock(clk_mgr_base, patched_disp_clk);
+		clk_mgr_base->clks.dispclk_khz = patched_disp_clk;
+	}
+	dce60_pplib_apply_display_requirements(clk_mgr_base->ctx->dc, context);
+}
+
+
+
+
+
+
+
+
+static struct clk_mgr_funcs dce60_funcs = {
+	.get_dp_ref_clk_frequency = dce60_get_dp_ref_freq_khz,
+	.update_clocks = dce60_update_clocks
+};
+
+void dce60_clk_mgr_construct(
+		struct dc_context *ctx,
+		struct clk_mgr_internal *clk_mgr)
+{
+	struct clk_mgr *base = &clk_mgr->base;
+
+	dce_clk_mgr_construct(ctx, clk_mgr);
+
+	memcpy(clk_mgr->max_clks_by_state,
+		dce60_max_clks_by_state,
+		sizeof(dce60_max_clks_by_state));
+
+	clk_mgr->regs = &disp_clk_regs;
+	clk_mgr->clk_mgr_shift = &disp_clk_shift;
+	clk_mgr->clk_mgr_mask = &disp_clk_mask;
+	clk_mgr->base.funcs = &dce60_funcs;
+
+	base->clks.max_supported_dispclk_khz =
+		clk_mgr->max_clks_by_state[DM_PP_CLOCKS_STATE_PERFORMANCE].display_clk_khz;
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dce60/dce60_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dce60/dce60_clk_mgr.h
new file mode 100644
index 000000000000..eca3e5168089
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dce60/dce60_clk_mgr.h
@@ -0,0 +1,36 @@
+/*
+ * Copyright 2020 Mauro Rossi <issor.oruam@gmail.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+#ifndef DAL_DC_DCE_DCE60_CLK_MGR_H_
+#define DAL_DC_DCE_DCE60_CLK_MGR_H_
+
+#include "dc.h"
+
+void dce60_clk_mgr_construct(
+		struct dc_context *ctx,
+		struct clk_mgr_internal *clk_mgr_dce);
+
+#endif /* DAL_DC_DCE_DCE60_CLK_MGR_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn10/rv1_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn10/rv1_clk_mgr.c
new file mode 100644
index 000000000000..369421e46c52
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn10/rv1_clk_mgr.c
@@ -0,0 +1,348 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "core_types.h"
+#include "clk_mgr_internal.h"
+#include "rv1_clk_mgr.h"
+#include "dce100/dce_clk_mgr.h"
+#include "dce112/dce112_clk_mgr.h"
+#include "rv1_clk_mgr_vbios_smu.h"
+#include "rv1_clk_mgr_clk.h"
+
+static void rv1_init_clocks(struct clk_mgr *clk_mgr)
+{
+	memset(&(clk_mgr->clks), 0, sizeof(struct dc_clocks));
+}
+
+static int rv1_determine_dppclk_threshold(struct clk_mgr_internal *clk_mgr, struct dc_clocks *new_clocks)
+{
+	bool request_dpp_div = new_clocks->dispclk_khz > new_clocks->dppclk_khz;
+	bool dispclk_increase = new_clocks->dispclk_khz > clk_mgr->base.clks.dispclk_khz;
+	int disp_clk_threshold = new_clocks->max_supported_dppclk_khz;
+	bool cur_dpp_div = clk_mgr->base.clks.dispclk_khz > clk_mgr->base.clks.dppclk_khz;
+
+	/* increase clock, looking for div is 0 for current, request div is 1*/
+	if (dispclk_increase) {
+		/* already divided by 2, no need to reach target clk with 2 steps*/
+		if (cur_dpp_div)
+			return new_clocks->dispclk_khz;
+
+		/* request disp clk is lower than maximum supported dpp clk,
+		 * no need to reach target clk with two steps.
+		 */
+		if (new_clocks->dispclk_khz <= disp_clk_threshold)
+			return new_clocks->dispclk_khz;
+
+		/* target dpp clk not request divided by 2, still within threshold */
+		if (!request_dpp_div)
+			return new_clocks->dispclk_khz;
+
+	} else {
+		/* decrease clock, looking for current dppclk divided by 2,
+		 * request dppclk not divided by 2.
+		 */
+
+		/* current dpp clk not divided by 2, no need to ramp*/
+		if (!cur_dpp_div)
+			return new_clocks->dispclk_khz;
+
+		/* current disp clk is lower than current maximum dpp clk,
+		 * no need to ramp
+		 */
+		if (clk_mgr->base.clks.dispclk_khz <= disp_clk_threshold)
+			return new_clocks->dispclk_khz;
+
+		/* request dpp clk need to be divided by 2 */
+		if (request_dpp_div)
+			return new_clocks->dispclk_khz;
+	}
+
+	return disp_clk_threshold;
+}
+
+static void ramp_up_dispclk_with_dpp(
+		struct clk_mgr_internal *clk_mgr,
+		struct dc *dc,
+		struct dc_clocks *new_clocks,
+		bool safe_to_lower)
+{
+	int i;
+	int dispclk_to_dpp_threshold = rv1_determine_dppclk_threshold(clk_mgr, new_clocks);
+	bool request_dpp_div = new_clocks->dispclk_khz > new_clocks->dppclk_khz;
+
+	/* this function is to change dispclk, dppclk and dprefclk according to
+	 * bandwidth requirement. Its call stack is rv1_update_clocks -->
+	 * update_clocks --> dcn10_prepare_bandwidth / dcn10_optimize_bandwidth
+	 * --> prepare_bandwidth / optimize_bandwidth. before change dcn hw,
+	 * prepare_bandwidth will be called first to allow enough clock,
+	 * watermark for change, after end of dcn hw change, optimize_bandwidth
+	 * is executed to lower clock to save power for new dcn hw settings.
+	 *
+	 * below is sequence of commit_planes_for_stream:
+	 *
+	 * step 1: prepare_bandwidth - raise clock to have enough bandwidth
+	 * step 2: lock_doublebuffer_enable
+	 * step 3: pipe_control_lock(true) - make dchubp register change will
+	 * not take effect right way
+	 * step 4: apply_ctx_for_surface - program dchubp
+	 * step 5: pipe_control_lock(false) - dchubp register change take effect
+	 * step 6: optimize_bandwidth --> dc_post_update_surfaces_to_stream
+	 * for full_date, optimize clock to save power
+	 *
+	 * at end of step 1, dcn clocks (dprefclk, dispclk, dppclk) may be
+	 * changed for new dchubp configuration. but real dcn hub dchubps are
+	 * still running with old configuration until end of step 5. this need
+	 * clocks settings at step 1 should not less than that before step 1.
+	 * this is checked by two conditions: 1. if (should_set_clock(safe_to_lower
+	 * , new_clocks->dispclk_khz, clk_mgr_base->clks.dispclk_khz) ||
+	 * new_clocks->dispclk_khz == clk_mgr_base->clks.dispclk_khz)
+	 * 2. request_dpp_div = new_clocks->dispclk_khz > new_clocks->dppclk_khz
+	 *
+	 * the second condition is based on new dchubp configuration. dppclk
+	 * for new dchubp may be different from dppclk before step 1.
+	 * for example, before step 1, dchubps are as below:
+	 * pipe 0: recout=(0,40,1920,980) viewport=(0,0,1920,979)
+	 * pipe 1: recout=(0,0,1920,1080) viewport=(0,0,1920,1080)
+	 * for dppclk for pipe0 need dppclk = dispclk
+	 *
+	 * new dchubp pipe split configuration:
+	 * pipe 0: recout=(0,0,960,1080) viewport=(0,0,960,1080)
+	 * pipe 1: recout=(960,0,960,1080) viewport=(960,0,960,1080)
+	 * dppclk only needs dppclk = dispclk /2.
+	 *
+	 * dispclk, dppclk are not lock by otg master lock. they take effect
+	 * after step 1. during this transition, dispclk are the same, but
+	 * dppclk is changed to half of previous clock for old dchubp
+	 * configuration between step 1 and step 6. This may cause p-state
+	 * warning intermittently.
+	 *
+	 * for new_clocks->dispclk_khz == clk_mgr_base->clks.dispclk_khz, we
+	 * need make sure dppclk are not changed to less between step 1 and 6.
+	 * for new_clocks->dispclk_khz > clk_mgr_base->clks.dispclk_khz,
+	 * new display clock is raised, but we do not know ratio of
+	 * new_clocks->dispclk_khz and clk_mgr_base->clks.dispclk_khz,
+	 * new_clocks->dispclk_khz /2 does not guarantee equal or higher than
+	 * old dppclk. we could ignore power saving different between
+	 * dppclk = displck and dppclk = dispclk / 2 between step 1 and step 6.
+	 * as long as safe_to_lower = false, set dpclk = dispclk to simplify
+	 * condition check.
+	 * todo: review this change for other asic.
+	 **/
+	if (!safe_to_lower)
+		request_dpp_div = false;
+
+	/* set disp clk to dpp clk threshold */
+
+	clk_mgr->funcs->set_dispclk(clk_mgr, dispclk_to_dpp_threshold);
+	clk_mgr->funcs->set_dprefclk(clk_mgr);
+
+
+	/* update request dpp clk division option */
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *pipe_ctx = &dc->current_state->res_ctx.pipe_ctx[i];
+
+		if (!pipe_ctx->plane_state)
+			continue;
+
+		pipe_ctx->plane_res.dpp->funcs->dpp_dppclk_control(
+				pipe_ctx->plane_res.dpp,
+				request_dpp_div,
+				true);
+	}
+
+	/* If target clk not same as dppclk threshold, set to target clock */
+	if (dispclk_to_dpp_threshold != new_clocks->dispclk_khz) {
+		clk_mgr->funcs->set_dispclk(clk_mgr, new_clocks->dispclk_khz);
+		clk_mgr->funcs->set_dprefclk(clk_mgr);
+	}
+
+
+	clk_mgr->base.clks.dispclk_khz = new_clocks->dispclk_khz;
+	clk_mgr->base.clks.dppclk_khz = new_clocks->dppclk_khz;
+	clk_mgr->base.clks.max_supported_dppclk_khz = new_clocks->max_supported_dppclk_khz;
+}
+
+static void rv1_update_clocks(struct clk_mgr *clk_mgr_base,
+			struct dc_state *context,
+			bool safe_to_lower)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	struct dc_debug_options *debug = &dc->debug;
+	struct dc_clocks *new_clocks = &context->bw_ctx.bw.dcn.clk;
+	struct pp_smu_funcs_rv *pp_smu = NULL;
+	bool send_request_to_increase = false;
+	bool send_request_to_lower = false;
+	int display_count;
+
+	bool enter_display_off = false;
+
+	ASSERT(clk_mgr->pp_smu);
+
+	if (dc->work_arounds.skip_clock_update)
+		return;
+
+	pp_smu = &clk_mgr->pp_smu->rv_funcs;
+
+	display_count = clk_mgr_helper_get_active_display_cnt(dc, context);
+
+	if (display_count == 0)
+		enter_display_off = true;
+
+	if (enter_display_off == safe_to_lower) {
+		/*
+		 * Notify SMU active displays
+		 * if function pointer not set up, this message is
+		 * sent as part of pplib_apply_display_requirements.
+		 */
+		if (pp_smu->set_display_count)
+			pp_smu->set_display_count(&pp_smu->pp_smu, display_count);
+	}
+
+	if (new_clocks->dispclk_khz > clk_mgr_base->clks.dispclk_khz
+			|| new_clocks->phyclk_khz > clk_mgr_base->clks.phyclk_khz
+			|| new_clocks->fclk_khz > clk_mgr_base->clks.fclk_khz
+			|| new_clocks->dcfclk_khz > clk_mgr_base->clks.dcfclk_khz)
+		send_request_to_increase = true;
+
+	if (should_set_clock(safe_to_lower, new_clocks->phyclk_khz, clk_mgr_base->clks.phyclk_khz)) {
+		clk_mgr_base->clks.phyclk_khz = new_clocks->phyclk_khz;
+		send_request_to_lower = true;
+	}
+
+	// F Clock
+	if (debug->force_fclk_khz != 0)
+		new_clocks->fclk_khz = debug->force_fclk_khz;
+
+	if (should_set_clock(safe_to_lower, new_clocks->fclk_khz, clk_mgr_base->clks.fclk_khz)) {
+		clk_mgr_base->clks.fclk_khz = new_clocks->fclk_khz;
+		send_request_to_lower = true;
+	}
+
+	//DCF Clock
+	if (should_set_clock(safe_to_lower, new_clocks->dcfclk_khz, clk_mgr_base->clks.dcfclk_khz)) {
+		clk_mgr_base->clks.dcfclk_khz = new_clocks->dcfclk_khz;
+		send_request_to_lower = true;
+	}
+
+	if (should_set_clock(safe_to_lower,
+			new_clocks->dcfclk_deep_sleep_khz, clk_mgr_base->clks.dcfclk_deep_sleep_khz)) {
+		clk_mgr_base->clks.dcfclk_deep_sleep_khz = new_clocks->dcfclk_deep_sleep_khz;
+		send_request_to_lower = true;
+	}
+
+	/* make sure dcf clk is before dpp clk to
+	 * make sure we have enough voltage to run dpp clk
+	 */
+	if (send_request_to_increase) {
+		/*use dcfclk to request voltage*/
+		if (pp_smu->set_hard_min_fclk_by_freq &&
+				pp_smu->set_hard_min_dcfclk_by_freq &&
+				pp_smu->set_min_deep_sleep_dcfclk) {
+			pp_smu->set_hard_min_fclk_by_freq(&pp_smu->pp_smu, khz_to_mhz_ceil(new_clocks->fclk_khz));
+			pp_smu->set_hard_min_dcfclk_by_freq(&pp_smu->pp_smu, khz_to_mhz_ceil(new_clocks->dcfclk_khz));
+			pp_smu->set_min_deep_sleep_dcfclk(&pp_smu->pp_smu, khz_to_mhz_ceil(new_clocks->dcfclk_deep_sleep_khz));
+		}
+	}
+
+	/* dcn1 dppclk is tied to dispclk */
+	/* program dispclk on = as a w/a for sleep resume clock ramping issues */
+	if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, clk_mgr_base->clks.dispclk_khz)
+			|| new_clocks->dispclk_khz == clk_mgr_base->clks.dispclk_khz) {
+		ramp_up_dispclk_with_dpp(clk_mgr, dc, new_clocks, safe_to_lower);
+		clk_mgr_base->clks.dispclk_khz = new_clocks->dispclk_khz;
+		send_request_to_lower = true;
+	}
+
+	if (!send_request_to_increase && send_request_to_lower) {
+		/*use dcfclk to request voltage*/
+		if (pp_smu->set_hard_min_fclk_by_freq &&
+				pp_smu->set_hard_min_dcfclk_by_freq &&
+				pp_smu->set_min_deep_sleep_dcfclk) {
+			pp_smu->set_hard_min_fclk_by_freq(&pp_smu->pp_smu, khz_to_mhz_ceil(new_clocks->fclk_khz));
+			pp_smu->set_hard_min_dcfclk_by_freq(&pp_smu->pp_smu, khz_to_mhz_ceil(new_clocks->dcfclk_khz));
+			pp_smu->set_min_deep_sleep_dcfclk(&pp_smu->pp_smu, khz_to_mhz_ceil(new_clocks->dcfclk_deep_sleep_khz));
+		}
+	}
+}
+
+static void rv1_enable_pme_wa(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct pp_smu_funcs_rv *pp_smu = NULL;
+
+	if (clk_mgr->pp_smu) {
+		pp_smu = &clk_mgr->pp_smu->rv_funcs;
+
+		if (pp_smu->set_pme_wa_enable)
+			pp_smu->set_pme_wa_enable(&pp_smu->pp_smu);
+	}
+}
+
+static struct clk_mgr_funcs rv1_clk_funcs = {
+	.init_clocks = rv1_init_clocks,
+	.get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz,
+	.update_clocks = rv1_update_clocks,
+	.enable_pme_wa = rv1_enable_pme_wa,
+};
+
+static struct clk_mgr_internal_funcs rv1_clk_internal_funcs = {
+	.set_dispclk = rv1_vbios_smu_set_dispclk,
+	.set_dprefclk = dce112_set_dprefclk
+};
+
+void rv1_clk_mgr_construct(struct dc_context *ctx, struct clk_mgr_internal *clk_mgr, struct pp_smu_funcs *pp_smu)
+{
+	struct dc_debug_options *debug = &ctx->dc->debug;
+	struct dc_bios *bp = ctx->dc_bios;
+
+	clk_mgr->base.ctx = ctx;
+	clk_mgr->pp_smu = pp_smu;
+	clk_mgr->base.funcs = &rv1_clk_funcs;
+	clk_mgr->funcs = &rv1_clk_internal_funcs;
+
+	clk_mgr->dfs_bypass_disp_clk = 0;
+
+	clk_mgr->dprefclk_ss_percentage = 0;
+	clk_mgr->dprefclk_ss_divider = 1000;
+	clk_mgr->ss_on_dprefclk = false;
+	clk_mgr->base.dprefclk_khz = 600000;
+
+	if (bp->integrated_info)
+		clk_mgr->base.dentist_vco_freq_khz = bp->integrated_info->dentist_vco_freq;
+	if (bp->fw_info_valid && clk_mgr->base.dentist_vco_freq_khz == 0) {
+		clk_mgr->base.dentist_vco_freq_khz = bp->fw_info.smu_gpu_pll_output_freq;
+		if (clk_mgr->base.dentist_vco_freq_khz == 0)
+			clk_mgr->base.dentist_vco_freq_khz = 3600000;
+	}
+
+	if (!debug->disable_dfs_bypass && bp->integrated_info)
+		if (bp->integrated_info->gpu_cap_info & DFS_BYPASS_ENABLE)
+			clk_mgr->dfs_bypass_enabled = true;
+
+	dce_clock_read_ss_info(clk_mgr);
+}
+
+
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn10/rv1_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn10/rv1_clk_mgr.h
new file mode 100644
index 000000000000..0807478c8212
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn10/rv1_clk_mgr.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __RV1_CLK_MGR_H__
+#define __RV1_CLK_MGR_H__
+
+void rv1_clk_mgr_construct(struct dc_context *ctx, struct clk_mgr_internal *clk_mgr, struct pp_smu_funcs *pp_smu);
+
+#endif //__DCN10_CLK_MGR_H__
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn10/rv1_clk_mgr_clk.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn10/rv1_clk_mgr_clk.h
new file mode 100644
index 000000000000..b68e3452efb9
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn10/rv1_clk_mgr_clk.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DAL_DC_DCN10_RV1_CLK_MGR_CLK_H_
+#define DAL_DC_DCN10_RV1_CLK_MGR_CLK_H_
+
+#endif /* DAL_DC_DCN10_RV1_CLK_MGR_CLK_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn10/rv1_clk_mgr_vbios_smu.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn10/rv1_clk_mgr_vbios_smu.c
new file mode 100644
index 000000000000..d82a52319088
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn10/rv1_clk_mgr_vbios_smu.c
@@ -0,0 +1,144 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "core_types.h"
+#include "clk_mgr_internal.h"
+#include "reg_helper.h"
+
+#include "rv1_clk_mgr_vbios_smu.h"
+
+#define MAX_INSTANCE	5
+#define MAX_SEGMENT		5
+
+struct IP_BASE_INSTANCE {
+	unsigned int segment[MAX_SEGMENT];
+};
+
+struct IP_BASE {
+	struct IP_BASE_INSTANCE instance[MAX_INSTANCE];
+};
+
+
+static const struct IP_BASE MP1_BASE  = { { { { 0x00016000, 0, 0, 0, 0 } },
+											 { { 0, 0, 0, 0, 0 } },
+											 { { 0, 0, 0, 0, 0 } },
+											 { { 0, 0, 0, 0, 0 } },
+											 { { 0, 0, 0, 0, 0 } } } };
+
+#define mmMP1_SMN_C2PMSG_91            0x29B
+#define mmMP1_SMN_C2PMSG_83            0x293
+#define mmMP1_SMN_C2PMSG_67            0x283
+#define mmMP1_SMN_C2PMSG_91_BASE_IDX   0
+#define mmMP1_SMN_C2PMSG_83_BASE_IDX   0
+#define mmMP1_SMN_C2PMSG_67_BASE_IDX   0
+
+#define MP1_SMN_C2PMSG_91__CONTENT_MASK                    0xffffffffL
+#define MP1_SMN_C2PMSG_83__CONTENT_MASK                    0xffffffffL
+#define MP1_SMN_C2PMSG_67__CONTENT_MASK                    0xffffffffL
+#define MP1_SMN_C2PMSG_91__CONTENT__SHIFT                  0x00000000
+#define MP1_SMN_C2PMSG_83__CONTENT__SHIFT                  0x00000000
+#define MP1_SMN_C2PMSG_67__CONTENT__SHIFT                  0x00000000
+
+#define REG(reg_name) \
+	(MP1_BASE.instance[0].segment[mm ## reg_name ## _BASE_IDX] + mm ## reg_name)
+
+#define FN(reg_name, field) \
+	FD(reg_name##__##field)
+
+#define VBIOSSMC_MSG_SetDispclkFreq           0x4
+#define VBIOSSMC_MSG_SetDprefclkFreq          0x5
+
+#define VBIOSSMC_Status_BUSY                      0x0
+#define VBIOSSMC_Result_OK                        0x1
+#define VBIOSSMC_Result_Failed                    0xFF
+#define VBIOSSMC_Result_UnknownCmd                0xFE
+#define VBIOSSMC_Result_CmdRejectedPrereq         0xFD
+#define VBIOSSMC_Result_CmdRejectedBusy           0xFC
+
+/*
+ * Function to be used instead of REG_WAIT macro because the wait ends when
+ * the register is NOT EQUAL to zero, and because the translation in msg_if.h
+ * won't work with REG_WAIT.
+ */
+static uint32_t rv1_smu_wait_for_response(struct clk_mgr_internal *clk_mgr, unsigned int delay_us, unsigned int max_retries)
+{
+	uint32_t res_val = VBIOSSMC_Status_BUSY;
+
+	do {
+		res_val = REG_READ(MP1_SMN_C2PMSG_91);
+		if (res_val != VBIOSSMC_Status_BUSY)
+			break;
+
+		if (delay_us >= 1000)
+			msleep(delay_us/1000);
+		else if (delay_us > 0)
+			udelay(delay_us);
+	} while (max_retries--);
+
+	return res_val;
+}
+
+static int rv1_vbios_smu_send_msg_with_param(struct clk_mgr_internal *clk_mgr,
+		unsigned int msg_id, unsigned int param)
+{
+	uint32_t result;
+
+	/* First clear response register */
+	REG_WRITE(MP1_SMN_C2PMSG_91, VBIOSSMC_Status_BUSY);
+
+	/* Set the parameter register for the SMU message, unit is Mhz */
+	REG_WRITE(MP1_SMN_C2PMSG_83, param);
+
+	/* Trigger the message transaction by writing the message ID */
+	REG_WRITE(MP1_SMN_C2PMSG_67, msg_id);
+
+	result = rv1_smu_wait_for_response(clk_mgr, 10, 1000);
+
+	ASSERT(result == VBIOSSMC_Result_OK);
+
+	/* Actual dispclk set is returned in the parameter register */
+	return REG_READ(MP1_SMN_C2PMSG_83);
+}
+
+int rv1_vbios_smu_set_dispclk(struct clk_mgr_internal *clk_mgr, int requested_dispclk_khz)
+{
+	int actual_dispclk_set_mhz = -1;
+	struct dc *dc = clk_mgr->base.ctx->dc;
+	struct dmcu *dmcu = dc->res_pool->dmcu;
+
+	/*  Unit of SMU msg parameter is Mhz */
+	actual_dispclk_set_mhz = rv1_vbios_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDispclkFreq,
+			khz_to_mhz_ceil(requested_dispclk_khz));
+
+	if (dmcu && dmcu->funcs->is_dmcu_initialized(dmcu)) {
+		if (clk_mgr->dfs_bypass_disp_clk != actual_dispclk_set_mhz)
+			dmcu->funcs->set_psr_wait_loop(dmcu,
+					actual_dispclk_set_mhz / 7);
+	}
+
+	return actual_dispclk_set_mhz * 1000;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn10/rv1_clk_mgr_vbios_smu.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn10/rv1_clk_mgr_vbios_smu.h
new file mode 100644
index 000000000000..81d7c912549c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn10/rv1_clk_mgr_vbios_smu.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DAL_DC_DCN10_RV1_CLK_MGR_VBIOS_SMU_H_
+#define DAL_DC_DCN10_RV1_CLK_MGR_VBIOS_SMU_H_
+
+int rv1_vbios_smu_set_dispclk(struct clk_mgr_internal *clk_mgr, int requested_dispclk_khz);
+
+#endif /* DAL_DC_DCN10_RV1_CLK_MGR_VBIOS_SMU_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn10/rv2_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn10/rv2_clk_mgr.c
new file mode 100644
index 000000000000..b9ba6dbc2b46
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn10/rv2_clk_mgr.c
@@ -0,0 +1,43 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "core_types.h"
+#include "clk_mgr_internal.h"
+#include "rv1_clk_mgr.h"
+#include "rv2_clk_mgr.h"
+#include "dce112/dce112_clk_mgr.h"
+
+static struct clk_mgr_internal_funcs rv2_clk_internal_funcs = {
+	.set_dispclk = dce112_set_dispclk,
+	.set_dprefclk = dce112_set_dprefclk
+};
+
+void rv2_clk_mgr_construct(struct dc_context *ctx, struct clk_mgr_internal *clk_mgr, struct pp_smu_funcs *pp_smu)
+
+{
+	rv1_clk_mgr_construct(ctx, clk_mgr, pp_smu);
+
+	clk_mgr->funcs = &rv2_clk_internal_funcs;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn10/rv2_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn10/rv2_clk_mgr.h
new file mode 100644
index 000000000000..0c1f26ca563b
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn10/rv2_clk_mgr.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __RV2_CLK_MGR_H__
+#define __RV2_CLK_MGR_H__
+
+void rv2_clk_mgr_construct(struct dc_context *ctx, struct clk_mgr_internal *clk_mgr, struct pp_smu_funcs *pp_smu);
+
+
+#endif //__DCN10_CLK_MGR_H__
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn20/dcn20_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn20/dcn20_clk_mgr.c
new file mode 100644
index 000000000000..bb4f3bd7532e
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn20/dcn20_clk_mgr.c
@@ -0,0 +1,589 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dccg.h"
+#include "clk_mgr_internal.h"
+
+#include "dce100/dce_clk_mgr.h"
+#include "dcn20_clk_mgr.h"
+#include "reg_helper.h"
+#include "core_types.h"
+#include "dm_helpers.h"
+
+#include "navi10_ip_offset.h"
+#include "dcn/dcn_2_0_0_offset.h"
+#include "dcn/dcn_2_0_0_sh_mask.h"
+#include "clk/clk_11_0_0_offset.h"
+#include "clk/clk_11_0_0_sh_mask.h"
+
+
+#undef FN
+#define FN(reg_name, field_name) \
+	clk_mgr->clk_mgr_shift->field_name, clk_mgr->clk_mgr_mask->field_name
+
+#define REG(reg) \
+	(clk_mgr->regs->reg)
+
+#define BASE_INNER(seg) DCN_BASE__INST0_SEG ## seg
+
+#define BASE(seg) BASE_INNER(seg)
+
+#define SR(reg_name)\
+		.reg_name = BASE(mm ## reg_name ## _BASE_IDX) +  \
+					mm ## reg_name
+
+#define CLK_BASE_INNER(seg) \
+	CLK_BASE__INST0_SEG ## seg
+
+
+static const struct clk_mgr_registers clk_mgr_regs = {
+	CLK_REG_LIST_NV10()
+};
+
+static const struct clk_mgr_shift clk_mgr_shift = {
+	CLK_MASK_SH_LIST_NV10(__SHIFT)
+};
+
+static const struct clk_mgr_mask clk_mgr_mask = {
+	CLK_MASK_SH_LIST_NV10(_MASK)
+};
+
+uint32_t dentist_get_did_from_divider(int divider)
+{
+	uint32_t divider_id;
+
+	/* we want to floor here to get higher clock than required rather than lower */
+	if (divider < DENTIST_DIVIDER_RANGE_2_START) {
+		if (divider < DENTIST_DIVIDER_RANGE_1_START)
+			divider_id = DENTIST_BASE_DID_1;
+		else
+			divider_id = DENTIST_BASE_DID_1
+				+ (divider - DENTIST_DIVIDER_RANGE_1_START)
+					/ DENTIST_DIVIDER_RANGE_1_STEP;
+	} else if (divider < DENTIST_DIVIDER_RANGE_3_START) {
+		divider_id = DENTIST_BASE_DID_2
+				+ (divider - DENTIST_DIVIDER_RANGE_2_START)
+					/ DENTIST_DIVIDER_RANGE_2_STEP;
+	} else if (divider < DENTIST_DIVIDER_RANGE_4_START) {
+		divider_id = DENTIST_BASE_DID_3
+				+ (divider - DENTIST_DIVIDER_RANGE_3_START)
+					/ DENTIST_DIVIDER_RANGE_3_STEP;
+	} else {
+		divider_id = DENTIST_BASE_DID_4
+				+ (divider - DENTIST_DIVIDER_RANGE_4_START)
+					/ DENTIST_DIVIDER_RANGE_4_STEP;
+		if (divider_id > DENTIST_MAX_DID)
+			divider_id = DENTIST_MAX_DID;
+	}
+
+	return divider_id;
+}
+
+void dcn20_update_clocks_update_dpp_dto(struct clk_mgr_internal *clk_mgr,
+		struct dc_state *context, bool safe_to_lower)
+{
+	int i;
+
+	clk_mgr->dccg->ref_dppclk = clk_mgr->base.clks.dppclk_khz;
+	for (i = 0; i < clk_mgr->base.ctx->dc->res_pool->pipe_count; i++) {
+		int dpp_inst, dppclk_khz, prev_dppclk_khz;
+
+		/* Loop index will match dpp->inst if resource exists,
+		 * and we want to avoid dependency on dpp object
+		 */
+		dpp_inst = i;
+		dppclk_khz = context->res_ctx.pipe_ctx[i].plane_res.bw.dppclk_khz;
+
+		prev_dppclk_khz = clk_mgr->dccg->pipe_dppclk_khz[i];
+
+		if (safe_to_lower || prev_dppclk_khz < dppclk_khz)
+			clk_mgr->dccg->funcs->update_dpp_dto(
+							clk_mgr->dccg, dpp_inst, dppclk_khz);
+	}
+}
+
+void dcn20_update_clocks_update_dentist(struct clk_mgr_internal *clk_mgr, struct dc_state *context)
+{
+	int dpp_divider = 0;
+	int disp_divider = 0;
+	uint32_t dppclk_wdivider = 0;
+	uint32_t dispclk_wdivider = 0;
+	uint32_t current_dispclk_wdivider;
+	uint32_t i;
+
+	if (clk_mgr->base.clks.dppclk_khz == 0 || clk_mgr->base.clks.dispclk_khz == 0)
+		return;
+
+	dpp_divider = DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+		* clk_mgr->base.dentist_vco_freq_khz / clk_mgr->base.clks.dppclk_khz;
+	disp_divider = DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+		* clk_mgr->base.dentist_vco_freq_khz / clk_mgr->base.clks.dispclk_khz;
+
+	dppclk_wdivider = dentist_get_did_from_divider(dpp_divider);
+	dispclk_wdivider = dentist_get_did_from_divider(disp_divider);
+
+	REG_GET(DENTIST_DISPCLK_CNTL,
+			DENTIST_DISPCLK_WDIVIDER, &current_dispclk_wdivider);
+
+	/* When changing divider to or from 127, some extra programming is required to prevent corruption */
+	if (current_dispclk_wdivider == 127 && dispclk_wdivider != 127) {
+		for (i = 0; i < clk_mgr->base.ctx->dc->res_pool->pipe_count; i++) {
+			struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i];
+			uint32_t fifo_level;
+			struct dccg *dccg = clk_mgr->base.ctx->dc->res_pool->dccg;
+			struct stream_encoder *stream_enc = pipe_ctx->stream_res.stream_enc;
+			int32_t N;
+			int32_t j;
+
+			if (!resource_is_pipe_type(pipe_ctx, OTG_MASTER))
+				continue;
+			/* Virtual encoders don't have this function */
+			if (!stream_enc->funcs->get_fifo_cal_average_level)
+				continue;
+			fifo_level = stream_enc->funcs->get_fifo_cal_average_level(
+					stream_enc);
+			N = fifo_level / 4;
+			dccg->funcs->set_fifo_errdet_ovr_en(
+					dccg,
+					true);
+			for (j = 0; j < N - 4; j++)
+				dccg->funcs->otg_drop_pixel(
+						dccg,
+						pipe_ctx->stream_res.tg->inst);
+			dccg->funcs->set_fifo_errdet_ovr_en(
+					dccg,
+					false);
+		}
+	} else if (dispclk_wdivider == 127 && current_dispclk_wdivider != 127) {
+		REG_UPDATE(DENTIST_DISPCLK_CNTL,
+				DENTIST_DISPCLK_WDIVIDER, 126);
+		REG_WAIT(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_CHG_DONE, 1, 50, 2000);
+		for (i = 0; i < clk_mgr->base.ctx->dc->res_pool->pipe_count; i++) {
+			struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i];
+			struct dccg *dccg = clk_mgr->base.ctx->dc->res_pool->dccg;
+			struct stream_encoder *stream_enc = pipe_ctx->stream_res.stream_enc;
+			uint32_t fifo_level;
+			int32_t N;
+			int32_t j;
+
+			if (!resource_is_pipe_type(pipe_ctx, OTG_MASTER))
+				continue;
+			/* Virtual encoders don't have this function */
+			if (!stream_enc->funcs->get_fifo_cal_average_level)
+				continue;
+			fifo_level = stream_enc->funcs->get_fifo_cal_average_level(
+					stream_enc);
+			N = fifo_level / 4;
+			dccg->funcs->set_fifo_errdet_ovr_en(dccg, true);
+			for (j = 0; j < 12 - N; j++)
+				dccg->funcs->otg_add_pixel(dccg,
+						pipe_ctx->stream_res.tg->inst);
+			dccg->funcs->set_fifo_errdet_ovr_en(dccg, false);
+		}
+	}
+
+	REG_UPDATE(DENTIST_DISPCLK_CNTL,
+			DENTIST_DISPCLK_WDIVIDER, dispclk_wdivider);
+	REG_WAIT(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_CHG_DONE, 1, 50, 2000);
+	REG_UPDATE(DENTIST_DISPCLK_CNTL,
+			DENTIST_DPPCLK_WDIVIDER, dppclk_wdivider);
+	REG_WAIT(DENTIST_DISPCLK_CNTL, DENTIST_DPPCLK_CHG_DONE, 1, 5, 100);
+}
+
+
+void dcn2_update_clocks(struct clk_mgr *clk_mgr_base,
+			struct dc_state *context,
+			bool safe_to_lower)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dc_clocks *new_clocks = &context->bw_ctx.bw.dcn.clk;
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	struct pp_smu_funcs_nv *pp_smu = NULL;
+	int display_count;
+	bool update_dppclk = false;
+	bool update_dispclk = false;
+	bool enter_display_off = false;
+	bool dpp_clock_lowered = false;
+	struct dmcu *dmcu = clk_mgr_base->ctx->dc->res_pool->dmcu;
+	bool force_reset = false;
+	bool p_state_change_support;
+	int total_plane_count;
+
+	if (dc->work_arounds.skip_clock_update)
+		return;
+
+	if (clk_mgr_base->clks.dispclk_khz == 0 ||
+		dc->debug.force_clock_mode & 0x1) {
+		//this is from resume or boot up, if forced_clock cfg option used, we bypass program dispclk and DPPCLK, but need set them for S3.
+		force_reset = true;
+
+		dcn2_read_clocks_from_hw_dentist(clk_mgr_base);
+
+		//force_clock_mode 0x1:  force reset the clock even it is the same clock as long as it is in Passive level.
+	}
+	display_count = clk_mgr_helper_get_active_display_cnt(dc, context);
+	if (dc->res_pool->pp_smu)
+		pp_smu = &dc->res_pool->pp_smu->nv_funcs;
+
+	if (display_count == 0)
+		enter_display_off = true;
+
+	if (enter_display_off == safe_to_lower) {
+		if (pp_smu && pp_smu->set_display_count)
+			pp_smu->set_display_count(&pp_smu->pp_smu, display_count);
+	}
+
+	if (dc->debug.force_min_dcfclk_mhz > 0)
+		new_clocks->dcfclk_khz = (new_clocks->dcfclk_khz > (dc->debug.force_min_dcfclk_mhz * 1000)) ?
+				new_clocks->dcfclk_khz : (dc->debug.force_min_dcfclk_mhz * 1000);
+
+	if (should_set_clock(safe_to_lower, new_clocks->dcfclk_khz, clk_mgr_base->clks.dcfclk_khz)) {
+		clk_mgr_base->clks.dcfclk_khz = new_clocks->dcfclk_khz;
+		if (pp_smu && pp_smu->set_hard_min_dcfclk_by_freq)
+			pp_smu->set_hard_min_dcfclk_by_freq(&pp_smu->pp_smu, khz_to_mhz_ceil(clk_mgr_base->clks.dcfclk_khz));
+	}
+
+	if (should_set_clock(safe_to_lower,
+			new_clocks->dcfclk_deep_sleep_khz, clk_mgr_base->clks.dcfclk_deep_sleep_khz)) {
+		clk_mgr_base->clks.dcfclk_deep_sleep_khz = new_clocks->dcfclk_deep_sleep_khz;
+		if (pp_smu && pp_smu->set_min_deep_sleep_dcfclk)
+			pp_smu->set_min_deep_sleep_dcfclk(&pp_smu->pp_smu, khz_to_mhz_ceil(clk_mgr_base->clks.dcfclk_deep_sleep_khz));
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->socclk_khz, clk_mgr_base->clks.socclk_khz)) {
+		clk_mgr_base->clks.socclk_khz = new_clocks->socclk_khz;
+		if (pp_smu && pp_smu->set_hard_min_socclk_by_freq)
+			pp_smu->set_hard_min_socclk_by_freq(&pp_smu->pp_smu, khz_to_mhz_ceil(clk_mgr_base->clks.socclk_khz));
+	}
+
+	total_plane_count = clk_mgr_helper_get_active_plane_cnt(dc, context);
+	p_state_change_support = new_clocks->p_state_change_support || (total_plane_count == 0);
+	if (should_update_pstate_support(safe_to_lower, p_state_change_support, clk_mgr_base->clks.p_state_change_support)) {
+		clk_mgr_base->clks.prev_p_state_change_support = clk_mgr_base->clks.p_state_change_support;
+		clk_mgr_base->clks.p_state_change_support = p_state_change_support;
+		if (pp_smu && pp_smu->set_pstate_handshake_support)
+			pp_smu->set_pstate_handshake_support(&pp_smu->pp_smu, clk_mgr_base->clks.p_state_change_support);
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dramclk_khz, clk_mgr_base->clks.dramclk_khz)) {
+		clk_mgr_base->clks.dramclk_khz = new_clocks->dramclk_khz;
+		if (pp_smu && pp_smu->set_hard_min_uclk_by_freq)
+			pp_smu->set_hard_min_uclk_by_freq(&pp_smu->pp_smu, khz_to_mhz_ceil(clk_mgr_base->clks.dramclk_khz));
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dppclk_khz, clk_mgr->base.clks.dppclk_khz)) {
+		if (clk_mgr->base.clks.dppclk_khz > new_clocks->dppclk_khz)
+			dpp_clock_lowered = true;
+		clk_mgr->base.clks.dppclk_khz = new_clocks->dppclk_khz;
+
+		update_dppclk = true;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, clk_mgr_base->clks.dispclk_khz)) {
+		clk_mgr_base->clks.dispclk_khz = new_clocks->dispclk_khz;
+
+		update_dispclk = true;
+	}
+
+	if (update_dppclk || update_dispclk) {
+		new_clocks->disp_dpp_voltage_level_khz = new_clocks->dppclk_khz;
+
+		if (update_dispclk)
+			new_clocks->disp_dpp_voltage_level_khz = new_clocks->dispclk_khz > new_clocks->dppclk_khz ? new_clocks->dispclk_khz : new_clocks->dppclk_khz;
+
+		clk_mgr_base->clks.disp_dpp_voltage_level_khz = new_clocks->disp_dpp_voltage_level_khz;
+		if (pp_smu && pp_smu->set_voltage_by_freq)
+			pp_smu->set_voltage_by_freq(&pp_smu->pp_smu, PP_SMU_NV_DISPCLK, khz_to_mhz_ceil(clk_mgr_base->clks.disp_dpp_voltage_level_khz));
+	}
+
+	if (dc->config.forced_clocks == false || (force_reset && safe_to_lower)) {
+		if (dpp_clock_lowered) {
+			// if clock is being lowered, increase DTO before lowering refclk
+			dcn20_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
+			dcn20_update_clocks_update_dentist(clk_mgr, context);
+		} else {
+			// if clock is being raised, increase refclk before lowering DTO
+			if (update_dppclk || update_dispclk)
+				dcn20_update_clocks_update_dentist(clk_mgr, context);
+			// always update dtos unless clock is lowered and not safe to lower
+			dcn20_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
+		}
+	}
+
+	if (update_dispclk &&
+			dmcu && dmcu->funcs->is_dmcu_initialized(dmcu)) {
+		/*update dmcu for wait_loop count*/
+		dmcu->funcs->set_psr_wait_loop(dmcu,
+			clk_mgr_base->clks.dispclk_khz / 1000 / 7);
+	}
+}
+
+void dcn2_update_clocks_fpga(struct clk_mgr *clk_mgr,
+		struct dc_state *context,
+		bool safe_to_lower)
+{
+	struct clk_mgr_internal *clk_mgr_int = TO_CLK_MGR_INTERNAL(clk_mgr);
+
+	struct dc_clocks *new_clocks = &context->bw_ctx.bw.dcn.clk;
+	/* Min fclk = 1.2GHz since all the extra scemi logic seems to run off of it */
+	int fclk_adj = new_clocks->fclk_khz > 1200000 ? new_clocks->fclk_khz : 1200000;
+
+	if (should_set_clock(safe_to_lower, new_clocks->phyclk_khz, clk_mgr->clks.phyclk_khz)) {
+		clk_mgr->clks.phyclk_khz = new_clocks->phyclk_khz;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dcfclk_khz, clk_mgr->clks.dcfclk_khz)) {
+		clk_mgr->clks.dcfclk_khz = new_clocks->dcfclk_khz;
+	}
+
+	if (should_set_clock(safe_to_lower,
+			new_clocks->dcfclk_deep_sleep_khz, clk_mgr->clks.dcfclk_deep_sleep_khz)) {
+		clk_mgr->clks.dcfclk_deep_sleep_khz = new_clocks->dcfclk_deep_sleep_khz;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->socclk_khz, clk_mgr->clks.socclk_khz)) {
+		clk_mgr->clks.socclk_khz = new_clocks->socclk_khz;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dramclk_khz, clk_mgr->clks.dramclk_khz)) {
+		clk_mgr->clks.dramclk_khz = new_clocks->dramclk_khz;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dppclk_khz, clk_mgr->clks.dppclk_khz)) {
+		clk_mgr->clks.dppclk_khz = new_clocks->dppclk_khz;
+	}
+
+	if (should_set_clock(safe_to_lower, fclk_adj, clk_mgr->clks.fclk_khz)) {
+		clk_mgr->clks.fclk_khz = fclk_adj;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, clk_mgr->clks.dispclk_khz)) {
+		clk_mgr->clks.dispclk_khz = new_clocks->dispclk_khz;
+	}
+
+	/* Both fclk and ref_dppclk run on the same scemi clock.
+	 * So take the higher value since the DPP DTO is typically programmed
+	 * such that max dppclk is 1:1 with ref_dppclk.
+	 */
+	if (clk_mgr->clks.fclk_khz > clk_mgr->clks.dppclk_khz)
+		clk_mgr->clks.dppclk_khz = clk_mgr->clks.fclk_khz;
+	if (clk_mgr->clks.dppclk_khz > clk_mgr->clks.fclk_khz)
+		clk_mgr->clks.fclk_khz = clk_mgr->clks.dppclk_khz;
+
+	// Both fclk and ref_dppclk run on the same scemi clock.
+	clk_mgr_int->dccg->ref_dppclk = clk_mgr->clks.fclk_khz;
+
+	/* TODO: set dtbclk in correct place */
+	clk_mgr->clks.dtbclk_en = false;
+	dm_set_dcn_clocks(clk_mgr->ctx, &clk_mgr->clks);
+}
+
+void dcn2_init_clocks(struct clk_mgr *clk_mgr)
+{
+	memset(&(clk_mgr->clks), 0, sizeof(struct dc_clocks));
+	// Assumption is that boot state always supports pstate
+	clk_mgr->clks.p_state_change_support = true;
+	clk_mgr->clks.prev_p_state_change_support = true;
+}
+
+static void dcn2_enable_pme_wa(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct pp_smu_funcs_nv *pp_smu = NULL;
+
+	if (clk_mgr->pp_smu) {
+		pp_smu = &clk_mgr->pp_smu->nv_funcs;
+
+		if (pp_smu->set_pme_wa_enable)
+			pp_smu->set_pme_wa_enable(&pp_smu->pp_smu);
+	}
+}
+
+
+void dcn2_read_clocks_from_hw_dentist(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	uint32_t dispclk_wdivider;
+	uint32_t dppclk_wdivider;
+	int disp_divider;
+	int dpp_divider;
+
+	REG_GET(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_WDIVIDER, &dispclk_wdivider);
+	REG_GET(DENTIST_DISPCLK_CNTL, DENTIST_DPPCLK_WDIVIDER, &dppclk_wdivider);
+
+	disp_divider = dentist_get_divider_from_did(dispclk_wdivider);
+	dpp_divider = dentist_get_divider_from_did(dppclk_wdivider);
+
+	if (disp_divider && dpp_divider) {
+		/* Calculate the current DFS clock, in kHz.*/
+		clk_mgr_base->clks.dispclk_khz = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+			* clk_mgr->base.dentist_vco_freq_khz) / disp_divider;
+
+		clk_mgr_base->clks.dppclk_khz = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+				* clk_mgr->base.dentist_vco_freq_khz) / dpp_divider;
+	}
+}
+
+void dcn2_get_clock(struct clk_mgr *clk_mgr,
+		struct dc_state *context,
+			enum dc_clock_type clock_type,
+			struct dc_clock_config *clock_cfg)
+{
+
+	if (clock_type == DC_CLOCK_TYPE_DISPCLK) {
+		clock_cfg->max_clock_khz = context->bw_ctx.bw.dcn.clk.max_supported_dispclk_khz;
+		clock_cfg->min_clock_khz = DCN_MINIMUM_DISPCLK_Khz;
+		clock_cfg->current_clock_khz = clk_mgr->clks.dispclk_khz;
+		clock_cfg->bw_requirequired_clock_khz = context->bw_ctx.bw.dcn.clk.bw_dispclk_khz;
+	}
+	if (clock_type == DC_CLOCK_TYPE_DPPCLK) {
+		clock_cfg->max_clock_khz = context->bw_ctx.bw.dcn.clk.max_supported_dppclk_khz;
+		clock_cfg->min_clock_khz = DCN_MINIMUM_DPPCLK_Khz;
+		clock_cfg->current_clock_khz = clk_mgr->clks.dppclk_khz;
+		clock_cfg->bw_requirequired_clock_khz = context->bw_ctx.bw.dcn.clk.bw_dppclk_khz;
+	}
+}
+
+static bool dcn2_are_clock_states_equal(struct dc_clocks *a,
+		struct dc_clocks *b)
+{
+	if (a->dispclk_khz != b->dispclk_khz)
+		return false;
+	else if (a->dppclk_khz != b->dppclk_khz)
+		return false;
+	else if (a->disp_dpp_voltage_level_khz != b->disp_dpp_voltage_level_khz)
+		return false;
+	else if (a->dcfclk_khz != b->dcfclk_khz)
+		return false;
+	else if (a->socclk_khz != b->socclk_khz)
+		return false;
+	else if (a->dcfclk_deep_sleep_khz != b->dcfclk_deep_sleep_khz)
+		return false;
+	else if (a->dramclk_khz != b->dramclk_khz)
+		return false;
+	else if (a->p_state_change_support != b->p_state_change_support)
+		return false;
+
+	return true;
+}
+
+/* Notify clk_mgr of a change in link rate, update phyclk frequency if necessary */
+static void dcn2_notify_link_rate_change(struct clk_mgr *clk_mgr_base, struct dc_link *link)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	unsigned int i, max_phyclk_req = 0;
+	struct pp_smu_funcs_nv *pp_smu = NULL;
+
+	if (!clk_mgr->pp_smu || !clk_mgr->pp_smu->nv_funcs.set_voltage_by_freq)
+		return;
+
+	pp_smu = &clk_mgr->pp_smu->nv_funcs;
+
+	clk_mgr->cur_phyclk_req_table[link->link_index] = link->cur_link_settings.link_rate * LINK_RATE_REF_FREQ_IN_KHZ;
+
+	for (i = 0; i < MAX_LINKS; i++) {
+		if (clk_mgr->cur_phyclk_req_table[i] > max_phyclk_req)
+			max_phyclk_req = clk_mgr->cur_phyclk_req_table[i];
+	}
+
+	if (max_phyclk_req != clk_mgr_base->clks.phyclk_khz) {
+		clk_mgr_base->clks.phyclk_khz = max_phyclk_req;
+		pp_smu->set_voltage_by_freq(&pp_smu->pp_smu, PP_SMU_NV_PHYCLK, khz_to_mhz_ceil(clk_mgr_base->clks.phyclk_khz));
+	}
+}
+
+static struct clk_mgr_funcs dcn2_funcs = {
+	.get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz,
+	.update_clocks = dcn2_update_clocks,
+	.init_clocks = dcn2_init_clocks,
+	.enable_pme_wa = dcn2_enable_pme_wa,
+	.get_clock = dcn2_get_clock,
+	.are_clock_states_equal = dcn2_are_clock_states_equal,
+	.notify_link_rate_change = dcn2_notify_link_rate_change,
+};
+
+
+void dcn20_clk_mgr_construct(
+		struct dc_context *ctx,
+		struct clk_mgr_internal *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg)
+{
+	int dprefclk_did;
+	int target_div;
+	uint32_t pll_req_reg;
+	struct fixed31_32 pll_req;
+
+	clk_mgr->base.ctx = ctx;
+	clk_mgr->pp_smu = pp_smu;
+	clk_mgr->base.funcs = &dcn2_funcs;
+	clk_mgr->regs = &clk_mgr_regs;
+	clk_mgr->clk_mgr_shift = &clk_mgr_shift;
+	clk_mgr->clk_mgr_mask = &clk_mgr_mask;
+
+	clk_mgr->dccg = dccg;
+	clk_mgr->dfs_bypass_disp_clk = 0;
+
+	clk_mgr->dprefclk_ss_percentage = 0;
+	clk_mgr->dprefclk_ss_divider = 1000;
+	clk_mgr->ss_on_dprefclk = false;
+
+	clk_mgr->base.dprefclk_khz = 700000; // 700 MHz planned if VCO is 3.85 GHz, will be retrieved
+
+	/* DFS Slice 2 should be used for DPREFCLK */
+	dprefclk_did = REG_READ(CLK3_CLK2_DFS_CNTL);
+	/* Convert DPREFCLK DFS Slice DID to actual divider */
+	target_div = dentist_get_divider_from_did(dprefclk_did);
+	/* get FbMult value */
+	pll_req_reg = REG_READ(CLK3_CLK_PLL_REQ);
+
+	/* set up a fixed-point number
+	 * this works because the int part is on the right edge of the register
+	 * and the frac part is on the left edge
+	 */
+
+	pll_req = dc_fixpt_from_int(pll_req_reg & clk_mgr->clk_mgr_mask->FbMult_int);
+	pll_req.value |= pll_req_reg & clk_mgr->clk_mgr_mask->FbMult_frac;
+
+	/* multiply by REFCLK period */
+	pll_req = dc_fixpt_mul_int(pll_req, 100000);
+
+	/* integer part is now VCO frequency in kHz */
+	clk_mgr->base.dentist_vco_freq_khz = dc_fixpt_floor(pll_req);
+
+	/* in case we don't get a value from the register, use default */
+	if (clk_mgr->base.dentist_vco_freq_khz == 0)
+		clk_mgr->base.dentist_vco_freq_khz = 3850000;
+
+	/* Calculate the DPREFCLK in kHz.*/
+	clk_mgr->base.dprefclk_khz = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+		* clk_mgr->base.dentist_vco_freq_khz) / target_div;
+	//Integrated_info table does not exist on dGPU projects so should not be referenced
+	//anywhere in code for dGPUs.
+	//Also there is no plan for now that DFS BYPASS will be used on NV10/12/14.
+	clk_mgr->dfs_bypass_enabled = false;
+
+	dce_clock_read_ss_info(clk_mgr);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn20/dcn20_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn20/dcn20_clk_mgr.h
new file mode 100644
index 000000000000..d254d0b6fba1
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn20/dcn20_clk_mgr.h
@@ -0,0 +1,59 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCN20_CLK_MGR_H__
+#define __DCN20_CLK_MGR_H__
+
+void dcn2_update_clocks(struct clk_mgr *dccg,
+			struct dc_state *context,
+			bool safe_to_lower);
+
+void dcn2_update_clocks_fpga(struct clk_mgr *clk_mgr,
+			struct dc_state *context,
+			bool safe_to_lower);
+void dcn20_update_clocks_update_dpp_dto(struct clk_mgr_internal *clk_mgr,
+		struct dc_state *context, bool safe_to_lower);
+
+void dcn2_init_clocks(struct clk_mgr *clk_mgr);
+
+void dcn20_clk_mgr_construct(struct dc_context *ctx,
+		struct clk_mgr_internal *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg);
+
+uint32_t dentist_get_did_from_divider(int divider);
+
+void dcn2_get_clock(struct clk_mgr *clk_mgr,
+		struct dc_state *context,
+			enum dc_clock_type clock_type,
+			struct dc_clock_config *clock_cfg);
+
+void dcn20_update_clocks_update_dentist(struct clk_mgr_internal *clk_mgr,
+					struct dc_state *context);
+
+void dcn2_read_clocks_from_hw_dentist(struct clk_mgr *clk_mgr_base);
+
+
+#endif //__DCN20_CLK_MGR_H__
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn201/dcn201_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn201/dcn201_clk_mgr.c
new file mode 100644
index 000000000000..76c612ecfe3c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn201/dcn201_clk_mgr.c
@@ -0,0 +1,217 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+#include "core_types.h"
+#include "dccg.h"
+#include "clk_mgr_internal.h"
+#include "dcn201_clk_mgr.h"
+#include "dcn20/dcn20_clk_mgr.h"
+#include "dce100/dce_clk_mgr.h"
+#include "dm_helpers.h"
+#include "dm_services.h"
+
+#include "cyan_skillfish_ip_offset.h"
+#include "dcn/dcn_2_0_1_offset.h"
+#include "dcn/dcn_2_0_1_sh_mask.h"
+#include "clk/clk_11_0_1_offset.h"
+#include "clk/clk_11_0_1_sh_mask.h"
+
+#define REG(reg) \
+	(clk_mgr->regs->reg)
+
+#define BASE_INNER(seg) DMU_BASE__INST0_SEG ## seg
+
+#define BASE(seg) BASE_INNER(seg)
+
+#define SR(reg_name)\
+		.reg_name = BASE(mm ## reg_name ## _BASE_IDX) +  \
+					mm ## reg_name
+
+#define CLK_BASE_INNER(seg) \
+	CLK_BASE__INST0_SEG ## seg
+
+#undef FN
+#define FN(reg_name, field_name) \
+	clk_mgr->clk_mgr_shift->field_name, clk_mgr->clk_mgr_mask->field_name
+
+#define CTX \
+	clk_mgr->base.ctx
+
+static const struct clk_mgr_registers clk_mgr_regs = {
+		CLK_COMMON_REG_LIST_DCN_201()
+};
+
+static const struct clk_mgr_shift clk_mgr_shift = {
+	CLK_COMMON_MASK_SH_LIST_DCN201_BASE(__SHIFT)
+};
+
+static const struct clk_mgr_mask clk_mgr_mask = {
+	CLK_COMMON_MASK_SH_LIST_DCN201_BASE(_MASK)
+};
+
+static void dcn201_init_clocks(struct clk_mgr *clk_mgr)
+{
+	memset(&(clk_mgr->clks), 0, sizeof(struct dc_clocks));
+	clk_mgr->clks.p_state_change_support = true;
+	clk_mgr->clks.prev_p_state_change_support = true;
+	clk_mgr->clks.max_supported_dppclk_khz = 1200000;
+	clk_mgr->clks.max_supported_dispclk_khz = 1200000;
+}
+
+static void dcn201_update_clocks(struct clk_mgr *clk_mgr_base,
+	struct dc_state *context,
+	bool safe_to_lower)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dc_clocks *new_clocks = &context->bw_ctx.bw.dcn.clk;
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	bool update_dppclk = false;
+	bool update_dispclk = false;
+	bool dpp_clock_lowered = false;
+	bool force_reset = false;
+	bool p_state_change_support;
+	int total_plane_count;
+
+	if (dc->work_arounds.skip_clock_update)
+		return;
+
+	if (clk_mgr_base->clks.dispclk_khz == 0 ||
+	    dc->debug.force_clock_mode & 0x1) {
+		/* this is from resume or boot up, if forced_clock cfg option
+		 * used, we bypass program dispclk and DPPCLK, but need set them
+		 * for S3.
+		 */
+
+		force_reset = true;
+		/* force_clock_mode 0x1:  force reset the clock even it is the
+		 * same clock as long as it is in Passive level.
+		 */
+
+		dcn2_read_clocks_from_hw_dentist(clk_mgr_base);
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->phyclk_khz, clk_mgr_base->clks.phyclk_khz))
+		clk_mgr_base->clks.phyclk_khz = new_clocks->phyclk_khz;
+
+	if (dc->debug.force_min_dcfclk_mhz > 0)
+		new_clocks->dcfclk_khz = (new_clocks->dcfclk_khz > (dc->debug.force_min_dcfclk_mhz * 1000)) ?
+		new_clocks->dcfclk_khz : (dc->debug.force_min_dcfclk_mhz * 1000);
+
+	if (should_set_clock(safe_to_lower, new_clocks->dcfclk_khz, clk_mgr_base->clks.dcfclk_khz))
+		clk_mgr_base->clks.dcfclk_khz = new_clocks->dcfclk_khz;
+
+	if (should_set_clock(safe_to_lower,
+		new_clocks->dcfclk_deep_sleep_khz, clk_mgr_base->clks.dcfclk_deep_sleep_khz))
+		clk_mgr_base->clks.dcfclk_deep_sleep_khz = new_clocks->dcfclk_deep_sleep_khz;
+
+	if (should_set_clock(safe_to_lower, new_clocks->socclk_khz, clk_mgr_base->clks.socclk_khz))
+		clk_mgr_base->clks.socclk_khz = new_clocks->socclk_khz;
+
+	total_plane_count = clk_mgr_helper_get_active_plane_cnt(dc, context);
+	p_state_change_support = new_clocks->p_state_change_support || (total_plane_count == 0);
+	if (should_update_pstate_support(safe_to_lower, p_state_change_support, clk_mgr_base->clks.p_state_change_support)) {
+		clk_mgr_base->clks.prev_p_state_change_support = clk_mgr_base->clks.p_state_change_support;
+		clk_mgr_base->clks.p_state_change_support = p_state_change_support;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dramclk_khz, clk_mgr_base->clks.dramclk_khz))
+		clk_mgr_base->clks.dramclk_khz = new_clocks->dramclk_khz;
+
+	if (should_set_clock(safe_to_lower, new_clocks->dppclk_khz, clk_mgr->base.clks.dppclk_khz)) {
+		if (clk_mgr->base.clks.dppclk_khz > new_clocks->dppclk_khz)
+			dpp_clock_lowered = true;
+		clk_mgr->base.clks.dppclk_khz = new_clocks->dppclk_khz;
+
+		update_dppclk = true;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, clk_mgr_base->clks.dispclk_khz)) {
+		clk_mgr_base->clks.dispclk_khz = new_clocks->dispclk_khz;
+
+		update_dispclk = true;
+	}
+
+	if (dc->config.forced_clocks == false || (force_reset && safe_to_lower)) {
+		if (dpp_clock_lowered) {
+			// if clock is being lowered, increase DTO before lowering refclk
+			dcn20_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
+			dcn20_update_clocks_update_dentist(clk_mgr, context);
+		} else {
+			// if clock is being raised, increase refclk before lowering DTO
+			if (update_dppclk || update_dispclk)
+				dcn20_update_clocks_update_dentist(clk_mgr, context);
+			// always update dtos unless clock is lowered and not safe to lower
+			dcn20_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
+		}
+	}
+}
+
+static struct clk_mgr_funcs dcn201_funcs = {
+	.get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz,
+	.update_clocks = dcn201_update_clocks,
+	.init_clocks = dcn201_init_clocks,
+	.get_clock = dcn2_get_clock,
+};
+
+void dcn201_clk_mgr_construct(struct dc_context *ctx,
+		struct clk_mgr_internal *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg)
+{
+	struct dc_debug_options *debug = &ctx->dc->debug;
+	struct dc_bios *bp = ctx->dc_bios;
+	clk_mgr->base.ctx = ctx;
+	clk_mgr->base.funcs = &dcn201_funcs;
+	clk_mgr->regs = &clk_mgr_regs;
+	clk_mgr->clk_mgr_shift = &clk_mgr_shift;
+	clk_mgr->clk_mgr_mask = &clk_mgr_mask;
+
+	clk_mgr->dccg = dccg;
+
+	clk_mgr->dfs_bypass_disp_clk = 0;
+
+	clk_mgr->dprefclk_ss_percentage = 0;
+	clk_mgr->dprefclk_ss_divider = 1000;
+	clk_mgr->ss_on_dprefclk = false;
+
+	clk_mgr->base.dprefclk_khz = REG_READ(CLK4_CLK2_CURRENT_CNT);
+	clk_mgr->base.dprefclk_khz *= 100;
+
+	if (clk_mgr->base.dprefclk_khz == 0)
+		clk_mgr->base.dprefclk_khz = 600000;
+
+	REG_GET(CLK4_CLK_PLL_REQ, FbMult_int, &clk_mgr->base.dentist_vco_freq_khz);
+	clk_mgr->base.dentist_vco_freq_khz *= 100000;
+
+	if (clk_mgr->base.dentist_vco_freq_khz == 0)
+		clk_mgr->base.dentist_vco_freq_khz = 3000000;
+
+	if (!debug->disable_dfs_bypass && bp->integrated_info)
+		if (bp->integrated_info->gpu_cap_info & DFS_BYPASS_ENABLE)
+			clk_mgr->dfs_bypass_enabled = true;
+
+	dce_clock_read_ss_info(clk_mgr);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn201/dcn201_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn201/dcn201_clk_mgr.h
new file mode 100644
index 000000000000..ae463baaff47
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn201/dcn201_clk_mgr.h
@@ -0,0 +1,34 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCN201_CLK_MGR_H__
+#define __DCN201_CLK_MGR_H__
+
+void dcn201_clk_mgr_construct(struct dc_context *ctx,
+		struct clk_mgr_internal *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg);
+
+#endif //__DCN201_CLK_MGR_H__
+\ No newline at end of file
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c
new file mode 100644
index 000000000000..e18097f82091
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c
@@ -0,0 +1,787 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dccg.h"
+#include "rn_clk_mgr.h"
+
+#include "dcn20/dcn20_clk_mgr.h"
+#include "dml/dcn20/dcn20_fpu.h"
+
+#include "dce100/dce_clk_mgr.h"
+#include "rn_clk_mgr_vbios_smu.h"
+#include "reg_helper.h"
+#include "core_types.h"
+#include "dm_helpers.h"
+
+#include "atomfirmware.h"
+#include "clk/clk_10_0_2_offset.h"
+#include "clk/clk_10_0_2_sh_mask.h"
+#include "renoir_ip_offset.h"
+
+
+/* Constants */
+
+#define SMU_VER_55_51_0 0x373300 /* SMU Version that is able to set DISPCLK below 100MHz */
+
+/* Macros */
+
+#define REG(reg_name) \
+	(CLK_BASE.instance[0].segment[mm ## reg_name ## _BASE_IDX] + mm ## reg_name)
+
+
+/* TODO: evaluate how to lower or disable all dcn clocks in screen off case */
+static int rn_get_active_display_cnt_wa(struct dc *dc, struct dc_state *context)
+{
+	int i, display_count;
+	bool tmds_present = false;
+
+	display_count = 0;
+	for (i = 0; i < context->stream_count; i++) {
+		const struct dc_stream_state *stream = context->streams[i];
+
+		if (stream->signal == SIGNAL_TYPE_HDMI_TYPE_A ||
+				stream->signal == SIGNAL_TYPE_DVI_SINGLE_LINK ||
+				stream->signal == SIGNAL_TYPE_DVI_DUAL_LINK)
+			tmds_present = true;
+	}
+
+	for (i = 0; i < dc->link_count; i++) {
+		const struct dc_link *link = dc->links[i];
+
+		/* abusing the fact that the dig and phy are coupled to see if the phy is enabled */
+		if (link->link_enc->funcs->is_dig_enabled &&
+		    link->link_enc->funcs->is_dig_enabled(link->link_enc))
+			display_count++;
+	}
+
+	/* WA for hang on HDMI after display off back back on*/
+	if (display_count == 0 && tmds_present)
+		display_count = 1;
+
+	return display_count;
+}
+
+static void rn_set_low_power_state(struct clk_mgr *clk_mgr_base)
+{
+	int display_count;
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	struct dc_state *context = dc->current_state;
+
+	if (clk_mgr_base->clks.pwr_state != DCN_PWR_STATE_LOW_POWER) {
+
+		display_count = rn_get_active_display_cnt_wa(dc, context);
+
+		/* if we can go lower, go lower */
+		if (display_count == 0) {
+			rn_vbios_smu_set_dcn_low_power_state(clk_mgr, DCN_PWR_STATE_LOW_POWER);
+			/* update power state */
+			clk_mgr_base->clks.pwr_state = DCN_PWR_STATE_LOW_POWER;
+		}
+	}
+}
+
+static void rn_update_clocks_update_dpp_dto(struct clk_mgr_internal *clk_mgr,
+		struct dc_state *context, int ref_dpp_clk, bool safe_to_lower)
+{
+	int i;
+
+	clk_mgr->dccg->ref_dppclk = ref_dpp_clk;
+
+	for (i = 0; i < clk_mgr->base.ctx->dc->res_pool->pipe_count; i++) {
+		int dpp_inst, dppclk_khz, prev_dppclk_khz;
+
+		/* Loop index may not match dpp->inst if some pipes disabled,
+		 * so select correct inst from res_pool
+		 */
+		dpp_inst = clk_mgr->base.ctx->dc->res_pool->dpps[i]->inst;
+		dppclk_khz = context->res_ctx.pipe_ctx[i].plane_res.bw.dppclk_khz;
+
+		prev_dppclk_khz = clk_mgr->dccg->pipe_dppclk_khz[dpp_inst];
+
+		if (safe_to_lower || prev_dppclk_khz < dppclk_khz)
+			clk_mgr->dccg->funcs->update_dpp_dto(
+							clk_mgr->dccg, dpp_inst, dppclk_khz);
+	}
+}
+
+
+static void rn_update_clocks(struct clk_mgr *clk_mgr_base,
+			struct dc_state *context,
+			bool safe_to_lower)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dc_clocks *new_clocks = &context->bw_ctx.bw.dcn.clk;
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	int display_count;
+	bool update_dppclk = false;
+	bool update_dispclk = false;
+	bool dpp_clock_lowered = false;
+
+	struct dmcu *dmcu = clk_mgr_base->ctx->dc->res_pool->dmcu;
+
+	if (dc->work_arounds.skip_clock_update)
+		return;
+
+	/*
+	 * if it is safe to lower, but we are already in the lower state, we don't have to do anything
+	 * also if safe to lower is false, we just go in the higher state
+	 */
+	if (safe_to_lower && !dc->debug.disable_48mhz_pwrdwn) {
+		/* check that we're not already in lower */
+		if (clk_mgr_base->clks.pwr_state != DCN_PWR_STATE_LOW_POWER) {
+
+			display_count = rn_get_active_display_cnt_wa(dc, context);
+
+			/* if we can go lower, go lower */
+			if (display_count == 0) {
+				rn_vbios_smu_set_dcn_low_power_state(clk_mgr, DCN_PWR_STATE_LOW_POWER);
+				/* update power state */
+				clk_mgr_base->clks.pwr_state = DCN_PWR_STATE_LOW_POWER;
+			}
+		}
+	} else {
+		/* check that we're not already in D0 */
+		if (clk_mgr_base->clks.pwr_state != DCN_PWR_STATE_MISSION_MODE) {
+			rn_vbios_smu_set_dcn_low_power_state(clk_mgr, DCN_PWR_STATE_MISSION_MODE);
+			/* update power state */
+			clk_mgr_base->clks.pwr_state = DCN_PWR_STATE_MISSION_MODE;
+		}
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dcfclk_khz, clk_mgr_base->clks.dcfclk_khz)) {
+		clk_mgr_base->clks.dcfclk_khz = new_clocks->dcfclk_khz;
+		rn_vbios_smu_set_hard_min_dcfclk(clk_mgr, clk_mgr_base->clks.dcfclk_khz);
+	}
+
+	if (should_set_clock(safe_to_lower,
+			new_clocks->dcfclk_deep_sleep_khz, clk_mgr_base->clks.dcfclk_deep_sleep_khz)) {
+		clk_mgr_base->clks.dcfclk_deep_sleep_khz = new_clocks->dcfclk_deep_sleep_khz;
+		rn_vbios_smu_set_min_deep_sleep_dcfclk(clk_mgr, clk_mgr_base->clks.dcfclk_deep_sleep_khz);
+	}
+
+	// workaround: Limit dppclk to 100Mhz to avoid lower eDP panel switch to plus 4K monitor underflow.
+	// Do not adjust dppclk if dppclk is 0 to avoid unexpected result
+	if (new_clocks->dppclk_khz < 100000 && new_clocks->dppclk_khz > 0)
+		new_clocks->dppclk_khz = 100000;
+
+	/*
+	 * Temporally ignore thew 0 cases for disp and dpp clks.
+	 * We may have a new feature that requires 0 clks in the future.
+	 */
+	if (new_clocks->dppclk_khz == 0 || new_clocks->dispclk_khz == 0) {
+		new_clocks->dppclk_khz = clk_mgr_base->clks.dppclk_khz;
+		new_clocks->dispclk_khz = clk_mgr_base->clks.dispclk_khz;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dppclk_khz, clk_mgr_base->clks.dppclk_khz)) {
+		if (clk_mgr_base->clks.dppclk_khz > new_clocks->dppclk_khz)
+			dpp_clock_lowered = true;
+		clk_mgr_base->clks.dppclk_khz = new_clocks->dppclk_khz;
+		update_dppclk = true;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, clk_mgr_base->clks.dispclk_khz)) {
+		clk_mgr_base->clks.dispclk_khz = new_clocks->dispclk_khz;
+		clk_mgr_base->clks.actual_dispclk_khz = rn_vbios_smu_set_dispclk(clk_mgr, clk_mgr_base->clks.dispclk_khz);
+
+		update_dispclk = true;
+	}
+
+	if (dpp_clock_lowered) {
+		// increase per DPP DTO before lowering global dppclk with requested dppclk
+		rn_update_clocks_update_dpp_dto(
+				clk_mgr,
+				context,
+				clk_mgr_base->clks.dppclk_khz,
+				safe_to_lower);
+
+		clk_mgr_base->clks.actual_dppclk_khz =
+				rn_vbios_smu_set_dppclk(clk_mgr, clk_mgr_base->clks.dppclk_khz);
+
+		//update dpp dto with actual dpp clk.
+		rn_update_clocks_update_dpp_dto(
+				clk_mgr,
+				context,
+				clk_mgr_base->clks.actual_dppclk_khz,
+				safe_to_lower);
+
+	} else {
+		// increase global DPPCLK before lowering per DPP DTO
+		if (update_dppclk || update_dispclk)
+			clk_mgr_base->clks.actual_dppclk_khz =
+					rn_vbios_smu_set_dppclk(clk_mgr, clk_mgr_base->clks.dppclk_khz);
+
+		// always update dtos unless clock is lowered and not safe to lower
+		rn_update_clocks_update_dpp_dto(
+				clk_mgr,
+				context,
+				clk_mgr_base->clks.actual_dppclk_khz,
+				safe_to_lower);
+	}
+
+	if (update_dispclk &&
+			dmcu && dmcu->funcs->is_dmcu_initialized(dmcu)) {
+		/*update dmcu for wait_loop count*/
+		dmcu->funcs->set_psr_wait_loop(dmcu,
+			clk_mgr_base->clks.dispclk_khz / 1000 / 7);
+	}
+}
+
+static int get_vco_frequency_from_reg(struct clk_mgr_internal *clk_mgr)
+{
+	/* get FbMult value */
+	struct fixed31_32 pll_req;
+	unsigned int fbmult_frac_val = 0;
+	unsigned int fbmult_int_val = 0;
+
+
+	/*
+	 * Register value of fbmult is in 8.16 format, we are converting to 31.32
+	 * to leverage the fix point operations available in driver
+	 */
+
+	REG_GET(CLK1_CLK_PLL_REQ, FbMult_frac, &fbmult_frac_val); /* 16 bit fractional part*/
+	REG_GET(CLK1_CLK_PLL_REQ, FbMult_int, &fbmult_int_val); /* 8 bit integer part */
+
+	pll_req = dc_fixpt_from_int(fbmult_int_val);
+
+	/*
+	 * since fractional part is only 16 bit in register definition but is 32 bit
+	 * in our fix point definiton, need to shift left by 16 to obtain correct value
+	 */
+	pll_req.value |= fbmult_frac_val << 16;
+
+	/* multiply by REFCLK period */
+	pll_req = dc_fixpt_mul_int(pll_req, clk_mgr->dfs_ref_freq_khz);
+
+	/* integer part is now VCO frequency in kHz */
+	return dc_fixpt_floor(pll_req);
+}
+
+static void rn_dump_clk_registers_internal(struct rn_clk_internal *internal, struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	internal->CLK1_CLK3_CURRENT_CNT = REG_READ(CLK1_CLK3_CURRENT_CNT);
+	internal->CLK1_CLK3_BYPASS_CNTL = REG_READ(CLK1_CLK3_BYPASS_CNTL);
+
+	internal->CLK1_CLK3_DS_CNTL = REG_READ(CLK1_CLK3_DS_CNTL);	//dcf deep sleep divider
+	internal->CLK1_CLK3_ALLOW_DS = REG_READ(CLK1_CLK3_ALLOW_DS);
+
+	internal->CLK1_CLK1_CURRENT_CNT = REG_READ(CLK1_CLK1_CURRENT_CNT);
+	internal->CLK1_CLK1_BYPASS_CNTL = REG_READ(CLK1_CLK1_BYPASS_CNTL);
+
+	internal->CLK1_CLK2_CURRENT_CNT = REG_READ(CLK1_CLK2_CURRENT_CNT);
+	internal->CLK1_CLK2_BYPASS_CNTL = REG_READ(CLK1_CLK2_BYPASS_CNTL);
+
+	internal->CLK1_CLK0_CURRENT_CNT = REG_READ(CLK1_CLK0_CURRENT_CNT);
+	internal->CLK1_CLK0_BYPASS_CNTL = REG_READ(CLK1_CLK0_BYPASS_CNTL);
+}
+
+/* This function collect raw clk register values */
+static void rn_dump_clk_registers(struct clk_state_registers_and_bypass *regs_and_bypass,
+		struct clk_mgr *clk_mgr_base, struct clk_log_info *log_info)
+{
+	struct rn_clk_internal internal = {0};
+	char *bypass_clks[5] = {"0x0 DFS", "0x1 REFCLK", "0x2 ERROR", "0x3 400 FCH", "0x4 600 FCH"};
+	unsigned int chars_printed = 0;
+	unsigned int remaining_buffer = log_info->bufSize;
+
+	rn_dump_clk_registers_internal(&internal, clk_mgr_base);
+
+	regs_and_bypass->dcfclk = internal.CLK1_CLK3_CURRENT_CNT / 10;
+	regs_and_bypass->dcf_deep_sleep_divider = internal.CLK1_CLK3_DS_CNTL / 10;
+	regs_and_bypass->dcf_deep_sleep_allow = internal.CLK1_CLK3_ALLOW_DS;
+	regs_and_bypass->dprefclk = internal.CLK1_CLK2_CURRENT_CNT / 10;
+	regs_and_bypass->dispclk = internal.CLK1_CLK0_CURRENT_CNT / 10;
+	regs_and_bypass->dppclk = internal.CLK1_CLK1_CURRENT_CNT / 10;
+
+	regs_and_bypass->dppclk_bypass = internal.CLK1_CLK1_BYPASS_CNTL & 0x0007;
+	if (regs_and_bypass->dppclk_bypass > 4)
+		regs_and_bypass->dppclk_bypass = 0;
+	regs_and_bypass->dcfclk_bypass = internal.CLK1_CLK3_BYPASS_CNTL & 0x0007;
+	if (regs_and_bypass->dcfclk_bypass > 4)
+		regs_and_bypass->dcfclk_bypass = 0;
+	regs_and_bypass->dispclk_bypass = internal.CLK1_CLK0_BYPASS_CNTL & 0x0007;
+	if (regs_and_bypass->dispclk_bypass > 4)
+		regs_and_bypass->dispclk_bypass = 0;
+	regs_and_bypass->dprefclk_bypass = internal.CLK1_CLK2_BYPASS_CNTL & 0x0007;
+	if (regs_and_bypass->dprefclk_bypass > 4)
+		regs_and_bypass->dprefclk_bypass = 0;
+
+	if (log_info->enabled) {
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "clk_type,clk_value,deepsleep_cntl,deepsleep_allow,bypass\n");
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "dcfclk,%d,%d,%d,%s\n",
+			regs_and_bypass->dcfclk,
+			regs_and_bypass->dcf_deep_sleep_divider,
+			regs_and_bypass->dcf_deep_sleep_allow,
+			bypass_clks[(int) regs_and_bypass->dcfclk_bypass]);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "dprefclk,%d,N/A,N/A,%s\n",
+			regs_and_bypass->dprefclk,
+			bypass_clks[(int) regs_and_bypass->dprefclk_bypass]);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "dispclk,%d,N/A,N/A,%s\n",
+			regs_and_bypass->dispclk,
+			bypass_clks[(int) regs_and_bypass->dispclk_bypass]);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		//split
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "SPLIT\n");
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		// REGISTER VALUES
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "reg_name,value,clk_type\n");
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "CLK1_CLK3_CURRENT_CNT,%d,dcfclk\n",
+				internal.CLK1_CLK3_CURRENT_CNT);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "CLK1_CLK3_DS_CNTL,%d,dcf_deep_sleep_divider\n",
+					internal.CLK1_CLK3_DS_CNTL);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "CLK1_CLK3_ALLOW_DS,%d,dcf_deep_sleep_allow\n",
+					internal.CLK1_CLK3_ALLOW_DS);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "CLK1_CLK2_CURRENT_CNT,%d,dprefclk\n",
+					internal.CLK1_CLK2_CURRENT_CNT);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "CLK1_CLK0_CURRENT_CNT,%d,dispclk\n",
+					internal.CLK1_CLK0_CURRENT_CNT);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "CLK1_CLK1_CURRENT_CNT,%d,dppclk\n",
+					internal.CLK1_CLK1_CURRENT_CNT);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "CLK1_CLK3_BYPASS_CNTL,%d,dcfclk_bypass\n",
+					internal.CLK1_CLK3_BYPASS_CNTL);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "CLK1_CLK2_BYPASS_CNTL,%d,dprefclk_bypass\n",
+					internal.CLK1_CLK2_BYPASS_CNTL);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "CLK1_CLK0_BYPASS_CNTL,%d,dispclk_bypass\n",
+					internal.CLK1_CLK0_BYPASS_CNTL);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "CLK1_CLK1_BYPASS_CNTL,%d,dppclk_bypass\n",
+					internal.CLK1_CLK1_BYPASS_CNTL);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+	}
+}
+
+static void rn_enable_pme_wa(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	rn_vbios_smu_enable_pme_wa(clk_mgr);
+}
+
+static void rn_init_clocks(struct clk_mgr *clk_mgr)
+{
+	memset(&(clk_mgr->clks), 0, sizeof(struct dc_clocks));
+	// Assumption is that boot state always supports pstate
+	clk_mgr->clks.p_state_change_support = true;
+	clk_mgr->clks.prev_p_state_change_support = true;
+	clk_mgr->clks.pwr_state = DCN_PWR_STATE_UNKNOWN;
+}
+
+static void build_watermark_ranges(struct clk_bw_params *bw_params, struct pp_smu_wm_range_sets *ranges)
+{
+	int i, num_valid_sets;
+
+	num_valid_sets = 0;
+
+	for (i = 0; i < WM_SET_COUNT; i++) {
+		/* skip empty entries, the smu array has no holes*/
+		if (!bw_params->wm_table.entries[i].valid)
+			continue;
+
+		ranges->reader_wm_sets[num_valid_sets].wm_inst = bw_params->wm_table.entries[i].wm_inst;
+		ranges->reader_wm_sets[num_valid_sets].wm_type = bw_params->wm_table.entries[i].wm_type;
+		/* We will not select WM based on fclk, so leave it as unconstrained */
+		ranges->reader_wm_sets[num_valid_sets].min_fill_clk_mhz = PP_SMU_WM_SET_RANGE_CLK_UNCONSTRAINED_MIN;
+		ranges->reader_wm_sets[num_valid_sets].max_fill_clk_mhz = PP_SMU_WM_SET_RANGE_CLK_UNCONSTRAINED_MAX;
+		/* dcfclk wil be used to select WM*/
+
+		if (ranges->reader_wm_sets[num_valid_sets].wm_type == WM_TYPE_PSTATE_CHG) {
+			if (i == 0)
+				ranges->reader_wm_sets[num_valid_sets].min_drain_clk_mhz = 0;
+			else {
+				/* add 1 to make it non-overlapping with next lvl */
+				ranges->reader_wm_sets[num_valid_sets].min_drain_clk_mhz = bw_params->clk_table.entries[i - 1].dcfclk_mhz + 1;
+			}
+			ranges->reader_wm_sets[num_valid_sets].max_drain_clk_mhz = bw_params->clk_table.entries[i].dcfclk_mhz;
+
+		} else {
+			/* unconstrained for memory retraining */
+			ranges->reader_wm_sets[num_valid_sets].min_fill_clk_mhz = PP_SMU_WM_SET_RANGE_CLK_UNCONSTRAINED_MIN;
+			ranges->reader_wm_sets[num_valid_sets].max_fill_clk_mhz = PP_SMU_WM_SET_RANGE_CLK_UNCONSTRAINED_MAX;
+
+			/* Modify previous watermark range to cover up to max */
+			if (num_valid_sets > 0)
+				ranges->reader_wm_sets[num_valid_sets - 1].max_fill_clk_mhz = PP_SMU_WM_SET_RANGE_CLK_UNCONSTRAINED_MAX;
+		}
+		num_valid_sets++;
+	}
+
+	ASSERT(num_valid_sets != 0); /* Must have at least one set of valid watermarks */
+	ranges->num_reader_wm_sets = num_valid_sets;
+
+	/* modify the min and max to make sure we cover the whole range*/
+	ranges->reader_wm_sets[0].min_drain_clk_mhz = PP_SMU_WM_SET_RANGE_CLK_UNCONSTRAINED_MIN;
+	ranges->reader_wm_sets[0].min_fill_clk_mhz = PP_SMU_WM_SET_RANGE_CLK_UNCONSTRAINED_MIN;
+	ranges->reader_wm_sets[ranges->num_reader_wm_sets - 1].max_drain_clk_mhz = PP_SMU_WM_SET_RANGE_CLK_UNCONSTRAINED_MAX;
+	ranges->reader_wm_sets[ranges->num_reader_wm_sets - 1].max_fill_clk_mhz = PP_SMU_WM_SET_RANGE_CLK_UNCONSTRAINED_MAX;
+
+	/* This is for writeback only, does not matter currently as no writeback support*/
+	ranges->num_writer_wm_sets = 1;
+	ranges->writer_wm_sets[0].wm_inst = WM_A;
+	ranges->writer_wm_sets[0].min_fill_clk_mhz = PP_SMU_WM_SET_RANGE_CLK_UNCONSTRAINED_MIN;
+	ranges->writer_wm_sets[0].max_fill_clk_mhz = PP_SMU_WM_SET_RANGE_CLK_UNCONSTRAINED_MAX;
+	ranges->writer_wm_sets[0].min_drain_clk_mhz = PP_SMU_WM_SET_RANGE_CLK_UNCONSTRAINED_MIN;
+	ranges->writer_wm_sets[0].max_drain_clk_mhz = PP_SMU_WM_SET_RANGE_CLK_UNCONSTRAINED_MAX;
+
+}
+
+static void rn_notify_wm_ranges(struct clk_mgr *clk_mgr_base)
+{
+	struct dc_debug_options *debug = &clk_mgr_base->ctx->dc->debug;
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct pp_smu_funcs *pp_smu = clk_mgr->pp_smu;
+
+	if (!debug->disable_pplib_wm_range) {
+		build_watermark_ranges(clk_mgr_base->bw_params, &clk_mgr_base->ranges);
+
+		/* Notify PP Lib/SMU which Watermarks to use for which clock ranges */
+		if (pp_smu && pp_smu->rn_funcs.set_wm_ranges)
+			pp_smu->rn_funcs.set_wm_ranges(&pp_smu->rn_funcs.pp_smu, &clk_mgr_base->ranges);
+	}
+
+}
+
+static bool rn_are_clock_states_equal(struct dc_clocks *a,
+		struct dc_clocks *b)
+{
+	if (a->dispclk_khz != b->dispclk_khz)
+		return false;
+	else if (a->dppclk_khz != b->dppclk_khz)
+		return false;
+	else if (a->dcfclk_khz != b->dcfclk_khz)
+		return false;
+	else if (a->dcfclk_deep_sleep_khz != b->dcfclk_deep_sleep_khz)
+		return false;
+
+	return true;
+}
+
+
+/* Notify clk_mgr of a change in link rate, update phyclk frequency if necessary */
+static void rn_notify_link_rate_change(struct clk_mgr *clk_mgr_base, struct dc_link *link)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	unsigned int i, max_phyclk_req = 0;
+
+	clk_mgr->cur_phyclk_req_table[link->link_index] = link->cur_link_settings.link_rate * LINK_RATE_REF_FREQ_IN_KHZ;
+
+	for (i = 0; i < MAX_LINKS; i++) {
+		if (clk_mgr->cur_phyclk_req_table[i] > max_phyclk_req)
+			max_phyclk_req = clk_mgr->cur_phyclk_req_table[i];
+	}
+
+	if (max_phyclk_req != clk_mgr_base->clks.phyclk_khz) {
+		clk_mgr_base->clks.phyclk_khz = max_phyclk_req;
+		rn_vbios_smu_set_phyclk(clk_mgr, clk_mgr_base->clks.phyclk_khz);
+	}
+}
+
+static struct clk_mgr_funcs dcn21_funcs = {
+	.get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz,
+	.update_clocks = rn_update_clocks,
+	.init_clocks = rn_init_clocks,
+	.enable_pme_wa = rn_enable_pme_wa,
+	.are_clock_states_equal = rn_are_clock_states_equal,
+	.set_low_power_state = rn_set_low_power_state,
+	.notify_wm_ranges = rn_notify_wm_ranges,
+	.notify_link_rate_change = rn_notify_link_rate_change,
+};
+
+static struct clk_bw_params rn_bw_params = {
+	.vram_type = Ddr4MemType,
+	.num_channels = 1,
+	.clk_table = {
+		.entries = {
+			{
+				.voltage = 0,
+				.dcfclk_mhz = 400,
+				.fclk_mhz = 400,
+				.memclk_mhz = 800,
+				.socclk_mhz = 0,
+			},
+			{
+				.voltage = 0,
+				.dcfclk_mhz = 483,
+				.fclk_mhz = 800,
+				.memclk_mhz = 1600,
+				.socclk_mhz = 0,
+			},
+			{
+				.voltage = 0,
+				.dcfclk_mhz = 602,
+				.fclk_mhz = 1067,
+				.memclk_mhz = 1067,
+				.socclk_mhz = 0,
+			},
+			{
+				.voltage = 0,
+				.dcfclk_mhz = 738,
+				.fclk_mhz = 1333,
+				.memclk_mhz = 1600,
+				.socclk_mhz = 0,
+			},
+		},
+
+		.num_entries = 4,
+	},
+
+};
+
+static unsigned int find_socclk_for_voltage(struct dpm_clocks *clock_table, unsigned int voltage)
+{
+	int i;
+
+	for (i = 0; i < PP_SMU_NUM_SOCCLK_DPM_LEVELS; i++) {
+		if (clock_table->SocClocks[i].Vol == voltage)
+			return clock_table->SocClocks[i].Freq;
+	}
+
+	ASSERT(0);
+	return 0;
+}
+
+static unsigned int find_dcfclk_for_voltage(struct dpm_clocks *clock_table, unsigned int voltage)
+{
+	int i;
+
+	for (i = 0; i < PP_SMU_NUM_DCFCLK_DPM_LEVELS; i++) {
+		if (clock_table->DcfClocks[i].Vol == voltage)
+			return clock_table->DcfClocks[i].Freq;
+	}
+
+	ASSERT(0);
+	return 0;
+}
+
+static void rn_clk_mgr_helper_populate_bw_params(struct clk_bw_params *bw_params, struct dpm_clocks *clock_table, struct integrated_info *bios_info)
+{
+	int i, j = 0;
+
+	j = -1;
+
+	static_assert(PP_SMU_NUM_FCLK_DPM_LEVELS <= MAX_NUM_DPM_LVL,
+		"number of reported FCLK DPM levels exceed maximum");
+
+	/* Find lowest DPM, FCLK is filled in reverse order*/
+
+	for (i = PP_SMU_NUM_FCLK_DPM_LEVELS - 1; i >= 0; i--) {
+		if (clock_table->FClocks[i].Freq != 0 && clock_table->FClocks[i].Vol != 0) {
+			j = i;
+			break;
+		}
+	}
+
+	if (j == -1) {
+		/* clock table is all 0s, just use our own hardcode */
+		ASSERT(0);
+		return;
+	}
+
+	bw_params->clk_table.num_entries = j + 1;
+
+	for (i = 0; i < bw_params->clk_table.num_entries; i++, j--) {
+		bw_params->clk_table.entries[i].fclk_mhz = clock_table->FClocks[j].Freq;
+		bw_params->clk_table.entries[i].memclk_mhz = clock_table->MemClocks[j].Freq;
+		bw_params->clk_table.entries[i].voltage = clock_table->FClocks[j].Vol;
+		bw_params->clk_table.entries[i].dcfclk_mhz = find_dcfclk_for_voltage(clock_table, clock_table->FClocks[j].Vol);
+		bw_params->clk_table.entries[i].socclk_mhz = find_socclk_for_voltage(clock_table,
+									bw_params->clk_table.entries[i].voltage);
+	}
+
+	bw_params->vram_type = bios_info->memory_type;
+	bw_params->num_channels = bios_info->ma_channel_number;
+
+	for (i = 0; i < WM_SET_COUNT; i++) {
+		bw_params->wm_table.entries[i].wm_inst = i;
+
+		if (i >= bw_params->clk_table.num_entries) {
+			bw_params->wm_table.entries[i].valid = false;
+			continue;
+		}
+
+		bw_params->wm_table.entries[i].wm_type = WM_TYPE_PSTATE_CHG;
+		bw_params->wm_table.entries[i].valid = true;
+	}
+
+	if (bw_params->vram_type == LpDdr4MemType) {
+		/*
+		 * WM set D will be re-purposed for memory retraining
+		 */
+		DC_FP_START();
+		dcn21_clk_mgr_set_bw_params_wm_table(bw_params);
+		DC_FP_END();
+	}
+}
+
+void rn_clk_mgr_construct(
+		struct dc_context *ctx,
+		struct clk_mgr_internal *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg)
+{
+	struct dc_debug_options *debug = &ctx->dc->debug;
+	struct dpm_clocks clock_table = { 0 };
+	enum pp_smu_status status = 0;
+	int is_green_sardine = 0;
+	struct clk_log_info log_info = {0};
+
+	is_green_sardine = ASICREV_IS_GREEN_SARDINE(ctx->asic_id.hw_internal_rev);
+
+	clk_mgr->base.ctx = ctx;
+	clk_mgr->base.funcs = &dcn21_funcs;
+
+	clk_mgr->pp_smu = pp_smu;
+
+	clk_mgr->dccg = dccg;
+	clk_mgr->dfs_bypass_disp_clk = 0;
+
+	clk_mgr->dprefclk_ss_percentage = 0;
+	clk_mgr->dprefclk_ss_divider = 1000;
+	clk_mgr->ss_on_dprefclk = false;
+	clk_mgr->dfs_ref_freq_khz = 48000;
+
+	clk_mgr->smu_ver = rn_vbios_smu_get_smu_version(clk_mgr);
+
+	clk_mgr->periodic_retraining_disabled = rn_vbios_smu_is_periodic_retraining_disabled(clk_mgr);
+
+	/* SMU Version 55.51.0 and up no longer have an issue
+	 * that needs to limit minimum dispclk */
+	if (clk_mgr->smu_ver >= SMU_VER_55_51_0)
+		debug->min_disp_clk_khz = 0;
+
+	/* TODO: Check we get what we expect during bringup */
+	clk_mgr->base.dentist_vco_freq_khz = get_vco_frequency_from_reg(clk_mgr);
+
+	/* in case we don't get a value from the register, use default */
+	if (clk_mgr->base.dentist_vco_freq_khz == 0)
+		clk_mgr->base.dentist_vco_freq_khz = 3600000;
+
+	if (ctx->dc_bios->integrated_info->memory_type == LpDdr4MemType) {
+		if (clk_mgr->periodic_retraining_disabled) {
+			rn_bw_params.wm_table = lpddr4_wm_table_with_disabled_ppt;
+		} else {
+			if (is_green_sardine)
+				rn_bw_params.wm_table = lpddr4_wm_table_gs;
+			else
+				rn_bw_params.wm_table = lpddr4_wm_table_rn;
+		}
+	} else {
+		if (is_green_sardine)
+			rn_bw_params.wm_table = ddr4_wm_table_gs;
+		else {
+			if (ctx->dc->config.is_single_rank_dimm)
+				rn_bw_params.wm_table = ddr4_1R_wm_table_rn;
+			else
+				rn_bw_params.wm_table = ddr4_wm_table_rn;
+		}
+	}
+	/* Saved clocks configured at boot for debug purposes */
+	rn_dump_clk_registers(&clk_mgr->base.boot_snapshot, &clk_mgr->base, &log_info);
+
+	clk_mgr->base.dprefclk_khz = 600000;
+	dce_clock_read_ss_info(clk_mgr);
+
+
+	clk_mgr->base.bw_params = &rn_bw_params;
+
+	if (pp_smu && pp_smu->rn_funcs.get_dpm_clock_table) {
+		status = pp_smu->rn_funcs.get_dpm_clock_table(&pp_smu->rn_funcs.pp_smu, &clock_table);
+
+		if (status == PP_SMU_RESULT_OK &&
+		    ctx->dc_bios->integrated_info) {
+			rn_clk_mgr_helper_populate_bw_params (clk_mgr->base.bw_params, &clock_table, ctx->dc_bios->integrated_info);
+			/* treat memory config as single channel if memory is asymmetrics. */
+			if (ctx->dc->config.is_asymmetric_memory)
+				clk_mgr->base.bw_params->num_channels = 1;
+		}
+	}
+
+	/* enable powerfeatures when displaycount goes to 0 */
+	if (clk_mgr->smu_ver >= 0x00371500)
+		rn_vbios_smu_enable_48mhz_tmdp_refclk_pwrdwn(clk_mgr, !debug->disable_48mhz_pwrdwn);
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.h
new file mode 100644
index 000000000000..f1319957e400
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.h
@@ -0,0 +1,49 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __RN_CLK_MGR_H__
+#define __RN_CLK_MGR_H__
+
+#include "clk_mgr.h"
+#include "dm_pp_smu.h"
+#include "clk_mgr_internal.h"
+
+extern struct wm_table ddr4_wm_table_gs;
+extern struct wm_table lpddr4_wm_table_gs;
+extern struct wm_table lpddr4_wm_table_with_disabled_ppt;
+extern struct wm_table ddr4_wm_table_rn;
+extern struct wm_table ddr4_1R_wm_table_rn;
+extern struct wm_table lpddr4_wm_table_rn;
+
+struct rn_clk_registers {
+	uint32_t CLK1_CLK0_CURRENT_CNT; /* DPREFCLK */
+};
+
+void rn_clk_mgr_construct(struct dc_context *ctx,
+		struct clk_mgr_internal *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg);
+
+#endif //__RN_CLK_MGR_H__
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr_vbios_smu.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr_vbios_smu.c
new file mode 100644
index 000000000000..5a633333dbb5
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr_vbios_smu.c
@@ -0,0 +1,257 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "core_types.h"
+#include "clk_mgr_internal.h"
+#include "reg_helper.h"
+#include "dm_helpers.h"
+
+#include "rn_clk_mgr_vbios_smu.h"
+
+#include <linux/delay.h>
+
+#include "renoir_ip_offset.h"
+
+#include "mp/mp_12_0_0_offset.h"
+#include "mp/mp_12_0_0_sh_mask.h"
+
+#define REG(reg_name) \
+	(MP0_BASE.instance[0].segment[mm ## reg_name ## _BASE_IDX] + mm ## reg_name)
+
+#define FN(reg_name, field) \
+	FD(reg_name##__##field)
+
+#include "logger_types.h"
+#undef DC_LOGGER
+#define DC_LOGGER \
+	CTX->logger
+#define smu_print(str, ...) {DC_LOG_SMU(str, ##__VA_ARGS__); }
+
+#define VBIOSSMC_MSG_TestMessage                  0x1
+#define VBIOSSMC_MSG_GetSmuVersion                0x2
+#define VBIOSSMC_MSG_PowerUpGfx                   0x3
+#define VBIOSSMC_MSG_SetDispclkFreq               0x4
+#define VBIOSSMC_MSG_SetDprefclkFreq              0x5
+#define VBIOSSMC_MSG_PowerDownGfx                 0x6
+#define VBIOSSMC_MSG_SetDppclkFreq                0x7
+#define VBIOSSMC_MSG_SetHardMinDcfclkByFreq       0x8
+#define VBIOSSMC_MSG_SetMinDeepSleepDcfclk        0x9
+#define VBIOSSMC_MSG_SetPhyclkVoltageByFreq       0xA
+#define VBIOSSMC_MSG_GetFclkFrequency             0xB
+#define VBIOSSMC_MSG_SetDisplayCount              0xC
+#define VBIOSSMC_MSG_EnableTmdp48MHzRefclkPwrDown 0xD
+#define VBIOSSMC_MSG_UpdatePmeRestore             0xE
+#define VBIOSSMC_MSG_IsPeriodicRetrainingDisabled 0xF
+
+#define VBIOSSMC_Status_BUSY                      0x0
+#define VBIOSSMC_Result_OK                        0x1
+#define VBIOSSMC_Result_Failed                    0xFF
+#define VBIOSSMC_Result_UnknownCmd                0xFE
+#define VBIOSSMC_Result_CmdRejectedPrereq         0xFD
+#define VBIOSSMC_Result_CmdRejectedBusy           0xFC
+
+/*
+ * Function to be used instead of REG_WAIT macro because the wait ends when
+ * the register is NOT EQUAL to zero, and because the translation in msg_if.h
+ * won't work with REG_WAIT.
+ */
+static uint32_t rn_smu_wait_for_response(struct clk_mgr_internal *clk_mgr, unsigned int delay_us, unsigned int max_retries)
+{
+	uint32_t res_val = VBIOSSMC_Status_BUSY;
+
+	do {
+		res_val = REG_READ(MP1_SMN_C2PMSG_91);
+		if (res_val != VBIOSSMC_Status_BUSY)
+			break;
+
+		if (delay_us >= 1000)
+			msleep(delay_us/1000);
+		else if (delay_us > 0)
+			udelay(delay_us);
+	} while (max_retries--);
+
+	return res_val;
+}
+
+
+static int rn_vbios_smu_send_msg_with_param(struct clk_mgr_internal *clk_mgr,
+					    unsigned int msg_id,
+					    unsigned int param)
+{
+	uint32_t result;
+
+	result = rn_smu_wait_for_response(clk_mgr, 10, 200000);
+
+	if (result != VBIOSSMC_Result_OK)
+		smu_print("SMU Response was not OK. SMU response after wait received is: %d\n", result);
+
+	if (result == VBIOSSMC_Status_BUSY) {
+		return -1;
+	}
+
+	/* First clear response register */
+	REG_WRITE(MP1_SMN_C2PMSG_91, VBIOSSMC_Status_BUSY);
+
+	/* Set the parameter register for the SMU message, unit is Mhz */
+	REG_WRITE(MP1_SMN_C2PMSG_83, param);
+
+	/* Trigger the message transaction by writing the message ID */
+	REG_WRITE(MP1_SMN_C2PMSG_67, msg_id);
+
+	result = rn_smu_wait_for_response(clk_mgr, 10, 200000);
+
+	if (IS_SMU_TIMEOUT(result)) {
+		ASSERT(0);
+		dm_helpers_smu_timeout(CTX, msg_id, param, 10 * 200000);
+	}
+
+	/* Actual dispclk set is returned in the parameter register */
+	return REG_READ(MP1_SMN_C2PMSG_83);
+}
+
+int rn_vbios_smu_get_smu_version(struct clk_mgr_internal *clk_mgr)
+{
+	return rn_vbios_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_GetSmuVersion,
+			0);
+}
+
+
+int rn_vbios_smu_set_dispclk(struct clk_mgr_internal *clk_mgr, int requested_dispclk_khz)
+{
+	int actual_dispclk_set_mhz = -1;
+	struct dc *dc = clk_mgr->base.ctx->dc;
+	struct dmcu *dmcu = dc->res_pool->dmcu;
+
+	/*  Unit of SMU msg parameter is Mhz */
+	actual_dispclk_set_mhz = rn_vbios_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDispclkFreq,
+			khz_to_mhz_ceil(requested_dispclk_khz));
+
+	if (dmcu && dmcu->funcs->is_dmcu_initialized(dmcu)) {
+		if (clk_mgr->dfs_bypass_disp_clk != actual_dispclk_set_mhz)
+			dmcu->funcs->set_psr_wait_loop(dmcu,
+					actual_dispclk_set_mhz / 7);
+	}
+
+	// pmfw always set clock more than or equal requested clock
+	ASSERT(actual_dispclk_set_mhz >= khz_to_mhz_ceil(requested_dispclk_khz));
+
+	return actual_dispclk_set_mhz * 1000;
+}
+
+int rn_vbios_smu_set_hard_min_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_dcfclk_khz)
+{
+	int actual_dcfclk_set_mhz = -1;
+
+	if (clk_mgr->smu_ver < 0x370c00)
+		return actual_dcfclk_set_mhz;
+
+	actual_dcfclk_set_mhz = rn_vbios_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetHardMinDcfclkByFreq,
+			khz_to_mhz_ceil(requested_dcfclk_khz));
+
+	return actual_dcfclk_set_mhz * 1000;
+}
+
+int rn_vbios_smu_set_min_deep_sleep_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_min_ds_dcfclk_khz)
+{
+	int actual_min_ds_dcfclk_mhz = -1;
+
+	if (clk_mgr->smu_ver < 0x370c00)
+		return actual_min_ds_dcfclk_mhz;
+
+	actual_min_ds_dcfclk_mhz = rn_vbios_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetMinDeepSleepDcfclk,
+			khz_to_mhz_ceil(requested_min_ds_dcfclk_khz));
+
+	return actual_min_ds_dcfclk_mhz * 1000;
+}
+
+void rn_vbios_smu_set_phyclk(struct clk_mgr_internal *clk_mgr, int requested_phyclk_khz)
+{
+	rn_vbios_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetPhyclkVoltageByFreq,
+			khz_to_mhz_ceil(requested_phyclk_khz));
+}
+
+int rn_vbios_smu_set_dppclk(struct clk_mgr_internal *clk_mgr, int requested_dpp_khz)
+{
+	int actual_dppclk_set_mhz = -1;
+
+	actual_dppclk_set_mhz = rn_vbios_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDppclkFreq,
+			khz_to_mhz_ceil(requested_dpp_khz));
+
+	ASSERT(actual_dppclk_set_mhz >= khz_to_mhz_ceil(requested_dpp_khz));
+
+	return actual_dppclk_set_mhz * 1000;
+}
+
+void rn_vbios_smu_set_dcn_low_power_state(struct clk_mgr_internal *clk_mgr, enum dcn_pwr_state state)
+{
+	int disp_count;
+
+	if (state == DCN_PWR_STATE_LOW_POWER)
+		disp_count = 0;
+	else
+		disp_count = 1;
+
+	rn_vbios_smu_send_msg_with_param(
+		clk_mgr,
+		VBIOSSMC_MSG_SetDisplayCount,
+		disp_count);
+}
+
+void rn_vbios_smu_enable_48mhz_tmdp_refclk_pwrdwn(struct clk_mgr_internal *clk_mgr, bool enable)
+{
+	rn_vbios_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_EnableTmdp48MHzRefclkPwrDown,
+			enable);
+}
+
+void rn_vbios_smu_enable_pme_wa(struct clk_mgr_internal *clk_mgr)
+{
+	rn_vbios_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_UpdatePmeRestore,
+			0);
+}
+
+int rn_vbios_smu_is_periodic_retraining_disabled(struct clk_mgr_internal *clk_mgr)
+{
+	return rn_vbios_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_IsPeriodicRetrainingDisabled,
+			1);	// if PMFW doesn't support this message, assume retraining is disabled
+				// so we only use most optimal watermark if we know retraining is enabled.
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr_vbios_smu.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr_vbios_smu.h
new file mode 100644
index 000000000000..f76fad87f0e1
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr_vbios_smu.h
@@ -0,0 +1,42 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DAL_DC_RN_CLK_MGR_VBIOS_SMU_H_
+#define DAL_DC_RN_CLK_MGR_VBIOS_SMU_H_
+
+enum dcn_pwr_state;
+
+int rn_vbios_smu_get_smu_version(struct clk_mgr_internal *clk_mgr);
+int rn_vbios_smu_set_dispclk(struct clk_mgr_internal *clk_mgr, int requested_dispclk_khz);
+int rn_vbios_smu_set_hard_min_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_dcfclk_khz);
+int rn_vbios_smu_set_min_deep_sleep_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_min_ds_dcfclk_khz);
+void rn_vbios_smu_set_phyclk(struct clk_mgr_internal *clk_mgr, int requested_phyclk_khz);
+int rn_vbios_smu_set_dppclk(struct clk_mgr_internal *clk_mgr, int requested_dpp_khz);
+void rn_vbios_smu_set_dcn_low_power_state(struct clk_mgr_internal *clk_mgr, enum dcn_pwr_state);
+void rn_vbios_smu_enable_48mhz_tmdp_refclk_pwrdwn(struct clk_mgr_internal *clk_mgr, bool enable);
+void rn_vbios_smu_enable_pme_wa(struct clk_mgr_internal *clk_mgr);
+int rn_vbios_smu_is_periodic_retraining_disabled(struct clk_mgr_internal *clk_mgr);
+
+#endif /* DAL_DC_DCN10_RV1_CLK_MGR_VBIOS_SMU_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dalsmc.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dalsmc.h
new file mode 100644
index 000000000000..06da34676965
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dalsmc.h
@@ -0,0 +1,62 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+// TEMPORARY until this exists in the proper location
+#ifndef DALSMC_H
+#define DALSMC_H
+
+#define DALSMC_VERSION 0x1
+
+// SMU Response Codes:
+#define DALSMC_Result_OK                   0x1
+#define DALSMC_Result_Failed               0xFF
+#define DALSMC_Result_UnknownCmd           0xFE
+#define DALSMC_Result_CmdRejectedPrereq    0xFD
+#define DALSMC_Result_CmdRejectedBusy      0xFC
+
+
+
+// Message Definitions:
+#define DALSMC_MSG_TestMessage                    0x1
+#define DALSMC_MSG_GetSmuVersion                  0x2
+#define DALSMC_MSG_GetDriverIfVersion             0x3
+#define DALSMC_MSG_GetMsgHeaderVersion            0x4
+#define DALSMC_MSG_SetDalDramAddrHigh             0x5
+#define DALSMC_MSG_SetDalDramAddrLow              0x6
+#define DALSMC_MSG_TransferTableSmu2Dram          0x7
+#define DALSMC_MSG_TransferTableDram2Smu          0x8
+#define DALSMC_MSG_SetHardMinByFreq               0x9
+#define DALSMC_MSG_SetHardMaxByFreq               0xA
+#define DALSMC_MSG_GetDpmFreqByIndex              0xB
+#define DALSMC_MSG_GetDcModeMaxDpmFreq            0xC
+#define DALSMC_MSG_SetMinDeepSleepDcefclk         0xD
+#define DALSMC_MSG_NumOfDisplays                  0xE
+#define DALSMC_MSG_SetDisplayRefreshFromMall      0xF
+#define DALSMC_MSG_SetExternalClientDfCstateAllow 0x10
+#define DALSMC_MSG_BacoAudioD3PME                 0x11
+#define DALSMC_MSG_SmartAccess                    0x12
+#define DALSMC_Message_Count                      0x13
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30_clk_mgr.c
new file mode 100644
index 000000000000..ef77fcd164ed
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30_clk_mgr.c
@@ -0,0 +1,587 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dccg.h"
+#include "clk_mgr_internal.h"
+#include "dcn30_clk_mgr_smu_msg.h"
+#include "dcn20/dcn20_clk_mgr.h"
+#include "dce100/dce_clk_mgr.h"
+#include "dcn30/dcn30_clk_mgr.h"
+#include "dml/dcn30/dcn30_fpu.h"
+#include "dcn30/dcn30m_clk_mgr.h"
+#include "reg_helper.h"
+#include "core_types.h"
+#include "dm_helpers.h"
+#include "atomfirmware.h"
+#include "sienna_cichlid_ip_offset.h"
+#include "dcn/dcn_3_0_0_offset.h"
+#include "dcn/dcn_3_0_0_sh_mask.h"
+#include "nbio/nbio_7_4_offset.h"
+#include "dpcs/dpcs_3_0_0_offset.h"
+#include "dpcs/dpcs_3_0_0_sh_mask.h"
+#include "mmhub/mmhub_2_0_0_offset.h"
+#include "mmhub/mmhub_2_0_0_sh_mask.h"
+#include "dcn30_smu11_driver_if.h"
+
+#undef FN
+#define FN(reg_name, field_name) \
+	clk_mgr->clk_mgr_shift->field_name, clk_mgr->clk_mgr_mask->field_name
+
+#define REG(reg) \
+	(clk_mgr->regs->reg)
+
+#define BASE_INNER(seg) DCN_BASE__INST0_SEG ## seg
+
+#define BASE(seg) BASE_INNER(seg)
+
+#define SR(reg_name)\
+		.reg_name = BASE(mm ## reg_name ## _BASE_IDX) +  \
+					mm ## reg_name
+
+#undef CLK_SRI
+#define CLK_SRI(reg_name, block, inst)\
+	.reg_name = mm ## block ## _ ## reg_name
+
+static const struct clk_mgr_registers clk_mgr_regs = {
+	CLK_REG_LIST_DCN3()
+};
+
+static const struct clk_mgr_shift clk_mgr_shift = {
+	CLK_COMMON_MASK_SH_LIST_DCN20_BASE(__SHIFT)
+};
+
+static const struct clk_mgr_mask clk_mgr_mask = {
+	CLK_COMMON_MASK_SH_LIST_DCN20_BASE(_MASK)
+};
+
+
+/* Query SMU for all clock states for a particular clock */
+static void dcn3_init_single_clock(struct clk_mgr_internal *clk_mgr, uint32_t clk, unsigned int *entry_0, unsigned int *num_levels)
+{
+	unsigned int i;
+	char *entry_i = (char *)entry_0;
+	uint32_t ret = dcn30_smu_get_dpm_freq_by_index(clk_mgr, clk, 0xFF);
+
+	if (ret & (1 << 31))
+		/* fine-grained, only min and max */
+		*num_levels = 2;
+	else
+		/* discrete, a number of fixed states */
+		/* will set num_levels to 0 on failure */
+		*num_levels = ret & 0xFF;
+
+	/* if the initial message failed, num_levels will be 0 */
+	for (i = 0; i < *num_levels; i++) {
+		*((unsigned int *)entry_i) = (dcn30_smu_get_dpm_freq_by_index(clk_mgr, clk, i) & 0xFFFF);
+		entry_i += sizeof(clk_mgr->base.bw_params->clk_table.entries[0]);
+	}
+}
+
+static void dcn3_build_wm_range_table(struct clk_mgr_internal *clk_mgr)
+{
+	DC_FP_START();
+	dcn3_fpu_build_wm_range_table(&clk_mgr->base);
+	DC_FP_END();
+}
+
+void dcn3_init_clocks(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	unsigned int num_levels;
+
+	memset(&(clk_mgr_base->clks), 0, sizeof(struct dc_clocks));
+	clk_mgr_base->clks.p_state_change_support = true;
+	clk_mgr_base->clks.prev_p_state_change_support = true;
+	clk_mgr->smu_present = false;
+
+	if (!clk_mgr_base->bw_params)
+		return;
+
+	if (!clk_mgr_base->force_smu_not_present && dcn30_smu_get_smu_version(clk_mgr, &clk_mgr->smu_ver))
+		clk_mgr->smu_present = true;
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	// do we fail if these fail? if so, how? do we not care to check?
+	dcn30_smu_check_driver_if_version(clk_mgr);
+	dcn30_smu_check_msg_header_version(clk_mgr);
+
+	/* DCFCLK */
+	dcn3_init_single_clock(clk_mgr, PPCLK_DCEFCLK,
+			&clk_mgr_base->bw_params->clk_table.entries[0].dcfclk_mhz,
+			&num_levels);
+	dcn30_smu_set_min_deep_sleep_dcef_clk(clk_mgr, 0);
+
+	/* DTBCLK */
+	dcn3_init_single_clock(clk_mgr, PPCLK_DTBCLK,
+			&clk_mgr_base->bw_params->clk_table.entries[0].dtbclk_mhz,
+			&num_levels);
+
+	/* SOCCLK */
+	dcn3_init_single_clock(clk_mgr, PPCLK_SOCCLK,
+					&clk_mgr_base->bw_params->clk_table.entries[0].socclk_mhz,
+					&num_levels);
+	// DPREFCLK ???
+
+	/* DISPCLK */
+	dcn3_init_single_clock(clk_mgr, PPCLK_DISPCLK,
+			&clk_mgr_base->bw_params->clk_table.entries[0].dispclk_mhz,
+			&num_levels);
+
+	/* DPPCLK */
+	dcn3_init_single_clock(clk_mgr, PPCLK_PIXCLK,
+			&clk_mgr_base->bw_params->clk_table.entries[0].dppclk_mhz,
+			&num_levels);
+
+	/* PHYCLK */
+	dcn3_init_single_clock(clk_mgr, PPCLK_PHYCLK,
+			&clk_mgr_base->bw_params->clk_table.entries[0].phyclk_mhz,
+			&num_levels);
+
+	/* Get UCLK, update bounding box */
+	clk_mgr_base->funcs->get_memclk_states_from_smu(clk_mgr_base);
+
+	/* WM range table */
+	DC_FP_START();
+	dcn3_build_wm_range_table(clk_mgr);
+	DC_FP_END();
+}
+
+static int dcn30_get_vco_frequency_from_reg(struct clk_mgr_internal *clk_mgr)
+{
+	/* get FbMult value */
+	struct fixed31_32 pll_req;
+	/* get FbMult value */
+	uint32_t pll_req_reg = REG_READ(CLK0_CLK_PLL_REQ);
+
+	/* set up a fixed-point number
+	 * this works because the int part is on the right edge of the register
+	 * and the frac part is on the left edge
+	 */
+	pll_req = dc_fixpt_from_int(pll_req_reg & clk_mgr->clk_mgr_mask->FbMult_int);
+	pll_req.value |= pll_req_reg & clk_mgr->clk_mgr_mask->FbMult_frac;
+
+	/* multiply by REFCLK period */
+	pll_req = dc_fixpt_mul_int(pll_req, clk_mgr->dfs_ref_freq_khz);
+
+	return dc_fixpt_floor(pll_req);
+}
+
+static void dcn3_update_clocks(struct clk_mgr *clk_mgr_base,
+			struct dc_state *context,
+			bool safe_to_lower)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dc_clocks *new_clocks = &context->bw_ctx.bw.dcn.clk;
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	int display_count;
+	bool update_dppclk = false;
+	bool update_dispclk = false;
+	bool enter_display_off = false;
+	bool dpp_clock_lowered = false;
+	bool update_pstate_unsupported_clk = false;
+	struct dmcu *dmcu = clk_mgr_base->ctx->dc->res_pool->dmcu;
+	bool force_reset = false;
+	bool update_uclk = false;
+	bool p_state_change_support;
+
+	if (dc->work_arounds.skip_clock_update || !clk_mgr->smu_present)
+		return;
+
+	if (clk_mgr_base->clks.dispclk_khz == 0 ||
+			(dc->debug.force_clock_mode & 0x1)) {
+		/* this is from resume or boot up, if forced_clock cfg option used, we bypass program dispclk and DPPCLK, but need set them for S3. */
+		force_reset = true;
+
+		dcn2_read_clocks_from_hw_dentist(clk_mgr_base);
+
+		/* force_clock_mode 0x1:  force reset the clock even it is the same clock as long as it is in Passive level. */
+	}
+	display_count = clk_mgr_helper_get_active_display_cnt(dc, context);
+
+	if (display_count == 0)
+		enter_display_off = true;
+
+	if (enter_display_off == safe_to_lower)
+		dcn30_smu_set_num_of_displays(clk_mgr, display_count);
+
+	if (dc->debug.force_min_dcfclk_mhz > 0)
+		new_clocks->dcfclk_khz = (new_clocks->dcfclk_khz > (dc->debug.force_min_dcfclk_mhz * 1000)) ?
+				new_clocks->dcfclk_khz : (dc->debug.force_min_dcfclk_mhz * 1000);
+
+	if (should_set_clock(safe_to_lower, new_clocks->dcfclk_khz, clk_mgr_base->clks.dcfclk_khz)) {
+		clk_mgr_base->clks.dcfclk_khz = new_clocks->dcfclk_khz;
+		dcn30_smu_set_hard_min_by_freq(clk_mgr, PPCLK_DCEFCLK, khz_to_mhz_ceil(clk_mgr_base->clks.dcfclk_khz));
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dcfclk_deep_sleep_khz, clk_mgr_base->clks.dcfclk_deep_sleep_khz)) {
+		clk_mgr_base->clks.dcfclk_deep_sleep_khz = new_clocks->dcfclk_deep_sleep_khz;
+		dcn30_smu_set_min_deep_sleep_dcef_clk(clk_mgr, khz_to_mhz_ceil(clk_mgr_base->clks.dcfclk_deep_sleep_khz));
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->socclk_khz, clk_mgr_base->clks.socclk_khz))
+		/* We don't actually care about socclk, don't notify SMU of hard min */
+		clk_mgr_base->clks.socclk_khz = new_clocks->socclk_khz;
+
+	clk_mgr_base->clks.prev_p_state_change_support = clk_mgr_base->clks.p_state_change_support;
+	p_state_change_support = new_clocks->p_state_change_support;
+
+	// invalidate the current P-State forced min in certain dc_mode_softmax situations
+	if (dc->clk_mgr->dc_mode_softmax_enabled && safe_to_lower && !p_state_change_support) {
+		if ((new_clocks->dramclk_khz <= dc->clk_mgr->bw_params->dc_mode_softmax_memclk * 1000) !=
+				(clk_mgr_base->clks.dramclk_khz <= dc->clk_mgr->bw_params->dc_mode_softmax_memclk * 1000))
+			update_pstate_unsupported_clk = true;
+	}
+
+	if (should_update_pstate_support(safe_to_lower, p_state_change_support, clk_mgr_base->clks.p_state_change_support) ||
+			update_pstate_unsupported_clk) {
+		clk_mgr_base->clks.p_state_change_support = p_state_change_support;
+
+		/* to disable P-State switching, set UCLK min = max */
+		if (!clk_mgr_base->clks.p_state_change_support) {
+			if (dc->clk_mgr->dc_mode_softmax_enabled &&
+				new_clocks->dramclk_khz <= dc->clk_mgr->bw_params->dc_mode_softmax_memclk * 1000)
+				dcn30_smu_set_hard_min_by_freq(clk_mgr, PPCLK_UCLK,
+					dc->clk_mgr->bw_params->dc_mode_softmax_memclk);
+			else
+				dcn30_smu_set_hard_min_by_freq(clk_mgr, PPCLK_UCLK,
+					clk_mgr_base->bw_params->clk_table.entries[clk_mgr_base->bw_params->clk_table.num_entries - 1].memclk_mhz);
+		}
+	}
+
+	/* Always update saved value, even if new value not set due to P-State switching unsupported */
+	if (should_set_clock(safe_to_lower, new_clocks->dramclk_khz, clk_mgr_base->clks.dramclk_khz)) {
+		clk_mgr_base->clks.dramclk_khz = new_clocks->dramclk_khz;
+		update_uclk = true;
+	}
+
+	/* set UCLK to requested value if P-State switching is supported, or to re-enable P-State switching */
+	if (clk_mgr_base->clks.p_state_change_support &&
+			(update_uclk || !clk_mgr_base->clks.prev_p_state_change_support))
+		dcn30_smu_set_hard_min_by_freq(clk_mgr, PPCLK_UCLK, khz_to_mhz_ceil(clk_mgr_base->clks.dramclk_khz));
+
+	if (should_set_clock(safe_to_lower, new_clocks->dppclk_khz, clk_mgr_base->clks.dppclk_khz)) {
+		if (clk_mgr_base->clks.dppclk_khz > new_clocks->dppclk_khz)
+			dpp_clock_lowered = true;
+
+		clk_mgr_base->clks.dppclk_khz = new_clocks->dppclk_khz;
+		dcn30_smu_set_hard_min_by_freq(clk_mgr, PPCLK_PIXCLK, khz_to_mhz_ceil(clk_mgr_base->clks.dppclk_khz));
+		update_dppclk = true;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, clk_mgr_base->clks.dispclk_khz)) {
+		clk_mgr_base->clks.dispclk_khz = new_clocks->dispclk_khz;
+		dcn30_smu_set_hard_min_by_freq(clk_mgr, PPCLK_DISPCLK, khz_to_mhz_ceil(clk_mgr_base->clks.dispclk_khz));
+		update_dispclk = true;
+	}
+
+	if (dc->config.forced_clocks == false || (force_reset && safe_to_lower)) {
+		if (dpp_clock_lowered) {
+			/* if clock is being lowered, increase DTO before lowering refclk */
+			dcn20_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
+			dcn20_update_clocks_update_dentist(clk_mgr, context);
+		} else {
+			/* if clock is being raised, increase refclk before lowering DTO */
+			if (update_dppclk || update_dispclk)
+				dcn20_update_clocks_update_dentist(clk_mgr, context);
+			/* There is a check inside dcn20_update_clocks_update_dpp_dto which ensures
+			 * that we do not lower dto when it is not safe to lower. We do not need to
+			 * compare the current and new dppclk before calling this function.*/
+			dcn20_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
+		}
+	}
+
+	if (update_dispclk && dmcu && dmcu->funcs->is_dmcu_initialized(dmcu))
+		/*update dmcu for wait_loop count*/
+		dmcu->funcs->set_psr_wait_loop(dmcu,
+				clk_mgr_base->clks.dispclk_khz / 1000 / 7);
+}
+
+
+static void dcn3_notify_wm_ranges(struct clk_mgr *clk_mgr_base)
+{
+	unsigned int i;
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	WatermarksExternal_t *table = (WatermarksExternal_t *) clk_mgr->wm_range_table;
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	if (!table)
+		// should log failure
+		return;
+
+	memset(table, 0, sizeof(*table));
+
+	/* collect valid ranges, place in pmfw table */
+	for (i = 0; i < WM_SET_COUNT; i++)
+		if (clk_mgr->base.bw_params->wm_table.nv_entries[i].valid) {
+			table->Watermarks.WatermarkRow[WM_DCEFCLK][i].MinClock = clk_mgr->base.bw_params->wm_table.nv_entries[i].pmfw_breakdown.min_dcfclk;
+			table->Watermarks.WatermarkRow[WM_DCEFCLK][i].MaxClock = clk_mgr->base.bw_params->wm_table.nv_entries[i].pmfw_breakdown.max_dcfclk;
+			table->Watermarks.WatermarkRow[WM_DCEFCLK][i].MinUclk = clk_mgr->base.bw_params->wm_table.nv_entries[i].pmfw_breakdown.min_uclk;
+			table->Watermarks.WatermarkRow[WM_DCEFCLK][i].MaxUclk = clk_mgr->base.bw_params->wm_table.nv_entries[i].pmfw_breakdown.max_uclk;
+			table->Watermarks.WatermarkRow[WM_DCEFCLK][i].WmSetting = i;
+			table->Watermarks.WatermarkRow[WM_DCEFCLK][i].Flags = clk_mgr->base.bw_params->wm_table.nv_entries[i].pmfw_breakdown.wm_type;
+		}
+
+	dcn30_smu_set_dram_addr_high(clk_mgr, clk_mgr->wm_range_table_addr >> 32);
+	dcn30_smu_set_dram_addr_low(clk_mgr, clk_mgr->wm_range_table_addr & 0xFFFFFFFF);
+	dcn30_smu_transfer_wm_table_dram_2_smu(clk_mgr);
+}
+
+/* Set min memclk to minimum, either constrained by the current mode or DPM0 */
+static void dcn3_set_hard_min_memclk(struct clk_mgr *clk_mgr_base, bool current_mode)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	if (current_mode) {
+		if (clk_mgr_base->clks.p_state_change_support)
+			dcn30_smu_set_hard_min_by_freq(clk_mgr, PPCLK_UCLK,
+					khz_to_mhz_ceil(clk_mgr_base->clks.dramclk_khz));
+		else
+			dcn30_smu_set_hard_min_by_freq(clk_mgr, PPCLK_UCLK,
+					clk_mgr_base->bw_params->clk_table.entries[clk_mgr_base->bw_params->clk_table.num_entries - 1].memclk_mhz);
+	} else {
+		dcn30_smu_set_hard_min_by_freq(clk_mgr, PPCLK_UCLK,
+				clk_mgr_base->bw_params->clk_table.entries[0].memclk_mhz);
+	}
+}
+
+/* Set max memclk to highest DPM value */
+static void dcn3_set_hard_max_memclk(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn30_smu_set_hard_max_by_freq(clk_mgr, PPCLK_UCLK,
+			clk_mgr_base->bw_params->clk_table.entries[clk_mgr_base->bw_params->clk_table.num_entries - 1].memclk_mhz);
+}
+
+static void dcn3_set_max_memclk(struct clk_mgr *clk_mgr_base, unsigned int memclk_mhz)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn30_smu_set_hard_max_by_freq(clk_mgr, PPCLK_UCLK, memclk_mhz);
+}
+static void dcn3_set_min_memclk(struct clk_mgr *clk_mgr_base, unsigned int memclk_mhz)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	if (!clk_mgr->smu_present)
+		return;
+	dcn30_smu_set_hard_min_by_freq(clk_mgr, PPCLK_UCLK, memclk_mhz);
+}
+
+/* Get current memclk states, update bounding box */
+static void dcn3_get_memclk_states_from_smu(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	unsigned int num_levels;
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	/* Refresh memclk states */
+	dcn3_init_single_clock(clk_mgr, PPCLK_UCLK,
+			&clk_mgr_base->bw_params->clk_table.entries[0].memclk_mhz,
+			&num_levels);
+	clk_mgr_base->bw_params->clk_table.num_entries = num_levels ? num_levels : 1;
+
+	clk_mgr_base->bw_params->dc_mode_softmax_memclk = dcn30_smu_get_dc_mode_max_dpm_freq(clk_mgr, PPCLK_UCLK);
+
+	/* Refresh bounding box */
+	DC_FP_START();
+	clk_mgr_base->ctx->dc->res_pool->funcs->update_bw_bounding_box(
+			clk_mgr->base.ctx->dc, clk_mgr_base->bw_params);
+	DC_FP_END();
+}
+
+static bool dcn3_is_smu_present(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	return clk_mgr->smu_present;
+}
+
+static bool dcn3_are_clock_states_equal(struct dc_clocks *a,
+					struct dc_clocks *b)
+{
+	if (a->dispclk_khz != b->dispclk_khz)
+		return false;
+	else if (a->dppclk_khz != b->dppclk_khz)
+		return false;
+	else if (a->dcfclk_khz != b->dcfclk_khz)
+		return false;
+	else if (a->dcfclk_deep_sleep_khz != b->dcfclk_deep_sleep_khz)
+		return false;
+	else if (a->dramclk_khz != b->dramclk_khz)
+		return false;
+	else if (a->p_state_change_support != b->p_state_change_support)
+		return false;
+
+	return true;
+}
+
+static void dcn3_enable_pme_wa(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn30_smu_set_pme_workaround(clk_mgr);
+}
+
+/* Notify clk_mgr of a change in link rate, update phyclk frequency if necessary */
+static void dcn30_notify_link_rate_change(struct clk_mgr *clk_mgr_base, struct dc_link *link)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	unsigned int i, max_phyclk_req = clk_mgr_base->bw_params->clk_table.entries[0].phyclk_mhz * 1000;
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	/* TODO - DP2.0 HW: calculate link 128b/132 link rate in clock manager with new formula */
+
+	clk_mgr->cur_phyclk_req_table[link->link_index] = link->cur_link_settings.link_rate * LINK_RATE_REF_FREQ_IN_KHZ;
+
+	for (i = 0; i < MAX_LINKS; i++) {
+		if (clk_mgr->cur_phyclk_req_table[i] > max_phyclk_req)
+			max_phyclk_req = clk_mgr->cur_phyclk_req_table[i];
+	}
+
+	if (max_phyclk_req != clk_mgr_base->clks.phyclk_khz) {
+		clk_mgr_base->clks.phyclk_khz = max_phyclk_req;
+		dcn30_smu_set_hard_min_by_freq(clk_mgr, PPCLK_PHYCLK, khz_to_mhz_ceil(clk_mgr_base->clks.phyclk_khz));
+	}
+}
+
+static struct clk_mgr_funcs dcn3_funcs = {
+		.get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz,
+		.update_clocks = dcn3_update_clocks,
+		.init_clocks = dcn3_init_clocks,
+		.notify_wm_ranges = dcn3_notify_wm_ranges,
+		.set_hard_min_memclk = dcn3_set_hard_min_memclk,
+		.set_hard_max_memclk = dcn3_set_hard_max_memclk,
+		.set_max_memclk = dcn3_set_max_memclk,
+		.set_min_memclk = dcn3_set_min_memclk,
+		.get_memclk_states_from_smu = dcn3_get_memclk_states_from_smu,
+		.are_clock_states_equal = dcn3_are_clock_states_equal,
+		.enable_pme_wa = dcn3_enable_pme_wa,
+		.notify_link_rate_change = dcn30_notify_link_rate_change,
+		.is_smu_present = dcn3_is_smu_present,
+		.set_smartmux_switch = dcn30m_set_smartmux_switch
+};
+
+static void dcn3_init_clocks_fpga(struct clk_mgr *clk_mgr)
+{
+	dcn2_init_clocks(clk_mgr);
+
+/* TODO: Implement the functions and remove the ifndef guard */
+}
+
+struct clk_mgr_funcs dcn3_fpga_funcs = {
+	.get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz,
+	.update_clocks = dcn2_update_clocks_fpga,
+	.init_clocks = dcn3_init_clocks_fpga,
+};
+
+/*todo for dcn30 for clk register offset*/
+void dcn3_clk_mgr_construct(
+		struct dc_context *ctx,
+		struct clk_mgr_internal *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg)
+{
+	struct clk_state_registers_and_bypass s = { 0 };
+
+	clk_mgr->base.ctx = ctx;
+	clk_mgr->base.funcs = &dcn3_funcs;
+	clk_mgr->regs = &clk_mgr_regs;
+	clk_mgr->clk_mgr_shift = &clk_mgr_shift;
+	clk_mgr->clk_mgr_mask = &clk_mgr_mask;
+
+	clk_mgr->dccg = dccg;
+	clk_mgr->dfs_bypass_disp_clk = 0;
+
+	clk_mgr->dprefclk_ss_percentage = 0;
+	clk_mgr->dprefclk_ss_divider = 1000;
+	clk_mgr->ss_on_dprefclk = false;
+	clk_mgr->dfs_ref_freq_khz = 100000;
+
+	clk_mgr->base.dprefclk_khz = 730000; // 700 MHz planned if VCO is 3.85 GHz, will be retrieved
+
+	/* integer part is now VCO frequency in kHz */
+	clk_mgr->base.dentist_vco_freq_khz = dcn30_get_vco_frequency_from_reg(clk_mgr);
+
+	/* in case we don't get a value from the register, use default */
+	if (clk_mgr->base.dentist_vco_freq_khz == 0)
+		clk_mgr->base.dentist_vco_freq_khz = 3650000;
+	/* Convert dprefclk units from MHz to KHz */
+	/* Value already divided by 10, some resolution lost */
+
+	/*TODO: uncomment assert once dcn3_dump_clk_registers is implemented */
+	//ASSERT(s.dprefclk != 0);
+	if (s.dprefclk != 0)
+		clk_mgr->base.dprefclk_khz = s.dprefclk * 1000;
+
+	clk_mgr->dfs_bypass_enabled = false;
+
+	clk_mgr->smu_present = false;
+
+	dce_clock_read_ss_info(clk_mgr);
+
+	clk_mgr->base.bw_params = kzalloc(sizeof(*clk_mgr->base.bw_params), GFP_KERNEL);
+	if (!clk_mgr->base.bw_params) {
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+	/* need physical address of table to give to PMFW */
+	clk_mgr->wm_range_table = dm_helpers_allocate_gpu_mem(clk_mgr->base.ctx,
+			DC_MEM_ALLOC_TYPE_GART, sizeof(WatermarksExternal_t),
+			&clk_mgr->wm_range_table_addr);
+	if (!clk_mgr->wm_range_table) {
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+void dcn3_clk_mgr_destroy(struct clk_mgr_internal *clk_mgr)
+{
+	kfree(clk_mgr->base.bw_params);
+
+	if (clk_mgr->wm_range_table)
+		dm_helpers_free_gpu_mem(clk_mgr->base.ctx, DC_MEM_ALLOC_TYPE_GART,
+				clk_mgr->wm_range_table);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30_clk_mgr.h
new file mode 100644
index 000000000000..2cd95ec38266
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30_clk_mgr.h
@@ -0,0 +1,98 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCN30_CLK_MGR_H__
+#define __DCN30_CLK_MGR_H__
+
+//CLK1_CLK_PLL_REQ
+#ifndef CLK11_CLK1_CLK_PLL_REQ__FbMult_int__SHIFT
+#define CLK11_CLK1_CLK_PLL_REQ__FbMult_int__SHIFT                                                                   0x0
+#define CLK11_CLK1_CLK_PLL_REQ__PllSpineDiv__SHIFT                                                                  0xc
+#define CLK11_CLK1_CLK_PLL_REQ__FbMult_frac__SHIFT                                                                  0x10
+#define CLK11_CLK1_CLK_PLL_REQ__FbMult_int_MASK                                                                     0x000001FFL
+#define CLK11_CLK1_CLK_PLL_REQ__PllSpineDiv_MASK                                                                    0x0000F000L
+#define CLK11_CLK1_CLK_PLL_REQ__FbMult_frac_MASK                                                                    0xFFFF0000L
+//CLK1_CLK0_DFS_CNTL
+#define CLK11_CLK1_CLK0_DFS_CNTL__CLK0_DIVIDER__SHIFT                                                               0x0
+#define CLK11_CLK1_CLK0_DFS_CNTL__CLK0_DIVIDER_MASK                                                                 0x0000007FL
+/*DPREF clock related*/
+#define CLK0_CLK3_DFS_CNTL__CLK3_DIVIDER__SHIFT                                                               0x0
+#define CLK0_CLK3_DFS_CNTL__CLK3_DIVIDER_MASK                                                                 0x0000007FL
+#define CLK1_CLK3_DFS_CNTL__CLK3_DIVIDER__SHIFT                                                               0x0
+#define CLK1_CLK3_DFS_CNTL__CLK3_DIVIDER_MASK                                                                 0x0000007FL
+#define CLK2_CLK3_DFS_CNTL__CLK3_DIVIDER__SHIFT                                                               0x0
+#define CLK2_CLK3_DFS_CNTL__CLK3_DIVIDER_MASK                                                                 0x0000007FL
+#define CLK3_CLK3_DFS_CNTL__CLK3_DIVIDER__SHIFT                                                               0x0
+#define CLK3_CLK3_DFS_CNTL__CLK3_DIVIDER_MASK                                                                 0x0000007FL
+
+//CLK3_0_CLK3_CLK_PLL_REQ
+#define CLK3_0_CLK3_CLK_PLL_REQ__FbMult_int__SHIFT                                                            0x0
+#define CLK3_0_CLK3_CLK_PLL_REQ__PllSpineDiv__SHIFT                                                           0xc
+#define CLK3_0_CLK3_CLK_PLL_REQ__FbMult_frac__SHIFT                                                           0x10
+#define CLK3_0_CLK3_CLK_PLL_REQ__FbMult_int_MASK                                                              0x000001FFL
+#define CLK3_0_CLK3_CLK_PLL_REQ__PllSpineDiv_MASK                                                             0x0000F000L
+#define CLK3_0_CLK3_CLK_PLL_REQ__FbMult_frac_MASK                                                             0xFFFF0000L
+
+#define mmCLK0_CLK2_DFS_CNTL                            0x16C55
+#define mmCLK00_CLK0_CLK2_DFS_CNTL                      0x16C55
+#define mmCLK01_CLK0_CLK2_DFS_CNTL                      0x16E55
+#define mmCLK02_CLK0_CLK2_DFS_CNTL                      0x17055
+
+#define mmCLK0_CLK3_DFS_CNTL                            0x16C60
+#define mmCLK00_CLK0_CLK3_DFS_CNTL                      0x16C60
+#define mmCLK01_CLK0_CLK3_DFS_CNTL                      0x16E60
+#define mmCLK02_CLK0_CLK3_DFS_CNTL                      0x17060
+#define mmCLK03_CLK0_CLK3_DFS_CNTL                      0x17260
+
+#define mmCLK0_CLK_PLL_REQ                              0x16C10
+#define mmCLK00_CLK0_CLK_PLL_REQ                        0x16C10
+#define mmCLK01_CLK0_CLK_PLL_REQ                        0x16E10
+#define mmCLK02_CLK0_CLK_PLL_REQ                        0x17010
+#define mmCLK03_CLK0_CLK_PLL_REQ                        0x17210
+
+#define mmCLK1_CLK_PLL_REQ                              0x1B00D
+#define mmCLK10_CLK1_CLK_PLL_REQ                        0x1B00D
+#define mmCLK11_CLK1_CLK_PLL_REQ                        0x1B20D
+#define mmCLK12_CLK1_CLK_PLL_REQ                        0x1B40D
+#define mmCLK13_CLK1_CLK_PLL_REQ                        0x1B60D
+
+#define mmCLK2_CLK_PLL_REQ                              0x17E0D
+
+/*AMCLK*/
+
+#define mmCLK11_CLK1_CLK0_DFS_CNTL                      0x1B23F
+#define mmCLK11_CLK1_CLK_PLL_REQ                        0x1B20D
+
+#endif
+void dcn3_init_clocks(struct clk_mgr *clk_mgr_base);
+
+void dcn3_clk_mgr_construct(struct dc_context *ctx,
+		struct clk_mgr_internal *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg);
+
+void dcn3_clk_mgr_destroy(struct clk_mgr_internal *clk_mgr);
+
+#endif //__DCN30_CLK_MGR_H__
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30_clk_mgr_smu_msg.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30_clk_mgr_smu_msg.c
new file mode 100644
index 000000000000..827bc2431d5d
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30_clk_mgr_smu_msg.c
@@ -0,0 +1,338 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dcn30_clk_mgr_smu_msg.h"
+
+#include "clk_mgr_internal.h"
+#include "reg_helper.h"
+#include "dm_helpers.h"
+
+#include "dalsmc.h"
+#include "dcn30_smu11_driver_if.h"
+
+#define mmDAL_MSG_REG  0x1628A
+#define mmDAL_ARG_REG  0x16273
+#define mmDAL_RESP_REG 0x16274
+
+#define REG(reg_name) \
+	mm ## reg_name
+
+#include "logger_types.h"
+#undef DC_LOGGER
+#define DC_LOGGER \
+	CTX->logger
+#define smu_print(str, ...) {DC_LOG_SMU(str, ##__VA_ARGS__); }
+
+
+/*
+ * Function to be used instead of REG_WAIT macro because the wait ends when
+ * the register is NOT EQUAL to zero, and because the translation in msg_if.h
+ * won't work with REG_WAIT.
+ */
+static uint32_t dcn30_smu_wait_for_response(struct clk_mgr_internal *clk_mgr, unsigned int delay_us, unsigned int max_retries)
+{
+	const uint32_t initial_max_retries = max_retries;
+	uint32_t reg = 0;
+
+	do {
+		reg = REG_READ(DAL_RESP_REG);
+		if (reg)
+			break;
+
+		if (delay_us >= 1000)
+			msleep(delay_us/1000);
+		else if (delay_us > 0)
+			udelay(delay_us);
+	} while (max_retries--);
+
+	/* handle DALSMC_Result_CmdRejectedBusy? */
+
+	TRACE_SMU_MSG_DELAY(0, 0, delay_us * (initial_max_retries - max_retries), clk_mgr->base.ctx);
+
+	return reg;
+}
+
+static bool dcn30_smu_send_msg_with_param(struct clk_mgr_internal *clk_mgr, uint32_t msg_id, uint32_t param_in, uint32_t *param_out)
+{
+	uint32_t result;
+	/* Wait for response register to be ready */
+	dcn30_smu_wait_for_response(clk_mgr, 10, 200000);
+
+	/* Clear response register */
+	REG_WRITE(DAL_RESP_REG, 0);
+
+	/* Set the parameter register for the SMU message */
+	REG_WRITE(DAL_ARG_REG, param_in);
+
+	/* Trigger the message transaction by writing the message ID */
+	REG_WRITE(DAL_MSG_REG, msg_id);
+
+	TRACE_SMU_MSG(msg_id, param_in, clk_mgr->base.ctx);
+
+	result = dcn30_smu_wait_for_response(clk_mgr, 10, 200000);
+
+	if (IS_SMU_TIMEOUT(result)) {
+		dm_helpers_smu_timeout(CTX, msg_id, param_in, 10 * 200000);
+	}
+
+	/* Wait for response */
+	if (result == DALSMC_Result_OK) {
+		if (param_out)
+			*param_out = REG_READ(DAL_ARG_REG);
+
+		return true;
+	}
+
+	return false;
+}
+
+/* Test message should return input + 1 */
+bool dcn30_smu_test_message(struct clk_mgr_internal *clk_mgr, uint32_t input)
+{
+	uint32_t response = 0;
+
+	smu_print("SMU Test message: %d\n", input);
+
+	if (dcn30_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_TestMessage, input, &response))
+		if (response == input + 1)
+			return true;
+
+	return false;
+}
+
+bool dcn30_smu_get_smu_version(struct clk_mgr_internal *clk_mgr, unsigned int *version)
+{
+	smu_print("SMU Get SMU version\n");
+
+	if (dcn30_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_GetSmuVersion, 0, version)) {
+
+		smu_print("SMU version: %d\n", *version);
+
+		return true;
+	}
+
+	return false;
+}
+
+/* Message output should match SMU11_DRIVER_IF_VERSION in smu11_driver_if.h */
+bool dcn30_smu_check_driver_if_version(struct clk_mgr_internal *clk_mgr)
+{
+	uint32_t response = 0;
+
+	smu_print("SMU Check driver if version\n");
+
+	if (dcn30_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_GetDriverIfVersion, 0, &response)) {
+
+		smu_print("SMU driver if version: %d\n", response);
+
+		if (response == SMU11_DRIVER_IF_VERSION)
+			return true;
+	}
+
+	return false;
+}
+
+/* Message output should match DALSMC_VERSION in dalsmc.h */
+bool dcn30_smu_check_msg_header_version(struct clk_mgr_internal *clk_mgr)
+{
+	uint32_t response = 0;
+
+	smu_print("SMU Check msg header version\n");
+
+	if (dcn30_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_GetMsgHeaderVersion, 0, &response)) {
+
+		smu_print("SMU msg header version: %d\n", response);
+
+		if (response == DALSMC_VERSION)
+			return true;
+	}
+
+	return false;
+}
+
+void dcn30_smu_set_dram_addr_high(struct clk_mgr_internal *clk_mgr, uint32_t addr_high)
+{
+	smu_print("SMU Set DRAM addr high: %d\n", addr_high);
+
+	dcn30_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_SetDalDramAddrHigh, addr_high, NULL);
+}
+
+void dcn30_smu_set_dram_addr_low(struct clk_mgr_internal *clk_mgr, uint32_t addr_low)
+{
+	smu_print("SMU Set DRAM addr low: %d\n", addr_low);
+
+	dcn30_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_SetDalDramAddrLow, addr_low, NULL);
+}
+
+void dcn30_smu_transfer_wm_table_smu_2_dram(struct clk_mgr_internal *clk_mgr)
+{
+	smu_print("SMU Transfer WM table SMU 2 DRAM\n");
+
+	dcn30_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_TransferTableSmu2Dram, TABLE_WATERMARKS, NULL);
+}
+
+void dcn30_smu_transfer_wm_table_dram_2_smu(struct clk_mgr_internal *clk_mgr)
+{
+	smu_print("SMU Transfer WM table DRAM 2 SMU\n");
+
+	dcn30_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_TransferTableDram2Smu, TABLE_WATERMARKS, NULL);
+}
+
+/* Returns the actual frequency that was set in MHz, 0 on failure */
+unsigned int dcn30_smu_set_hard_min_by_freq(struct clk_mgr_internal *clk_mgr, uint32_t clk, uint16_t freq_mhz)
+{
+	uint32_t response = 0;
+
+	/* bits 23:16 for clock type, lower 16 bits for frequency in MHz */
+	uint32_t param = (clk << 16) | freq_mhz;
+
+	smu_print("SMU Set hard min by freq: clk = %d, freq_mhz = %d MHz\n", clk, freq_mhz);
+
+	dcn30_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_SetHardMinByFreq, param, &response);
+
+	smu_print("SMU Frequency set = %d MHz\n", response);
+
+	return response;
+}
+
+/* Returns the actual frequency that was set in MHz, 0 on failure */
+unsigned int dcn30_smu_set_hard_max_by_freq(struct clk_mgr_internal *clk_mgr, uint32_t clk, uint16_t freq_mhz)
+{
+	uint32_t response = 0;
+
+	/* bits 23:16 for clock type, lower 16 bits for frequency in MHz */
+	uint32_t param = (clk << 16) | freq_mhz;
+
+	smu_print("SMU Set hard max by freq: clk = %d, freq_mhz = %d MHz\n", clk, freq_mhz);
+
+	dcn30_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_SetHardMaxByFreq, param, &response);
+
+	smu_print("SMU Frequency set = %d MHz\n", response);
+
+	return response;
+}
+
+/*
+ * Frequency in MHz returned in lower 16 bits for valid DPM level
+ *
+ * Call with dpm_level = 0xFF to query features, return value will be:
+ *     Bits 7:0 - number of DPM levels
+ *     Bit   28 - 1 = auto DPM on
+ *     Bit   29 - 1 = sweep DPM on
+ *     Bit   30 - 1 = forced DPM on
+ *     Bit   31 - 0 = discrete, 1 = fine-grained
+ *
+ * With fine-grained DPM, only min and max frequencies will be reported
+ *
+ * Returns 0 on failure
+ */
+unsigned int dcn30_smu_get_dpm_freq_by_index(struct clk_mgr_internal *clk_mgr, uint32_t clk, uint8_t dpm_level)
+{
+	uint32_t response = 0;
+
+	/* bits 23:16 for clock type, lower 8 bits for DPM level */
+	uint32_t param = (clk << 16) | dpm_level;
+
+	smu_print("SMU Get dpm freq by index: clk = %d, dpm_level = %d\n", clk, dpm_level);
+
+	dcn30_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_GetDpmFreqByIndex, param, &response);
+
+	smu_print("SMU dpm freq: %d MHz\n", response);
+
+	return response;
+}
+
+/* Returns the max DPM frequency in DC mode in MHz, 0 on failure */
+unsigned int dcn30_smu_get_dc_mode_max_dpm_freq(struct clk_mgr_internal *clk_mgr, uint32_t clk)
+{
+	uint32_t response = 0;
+
+	/* bits 23:16 for clock type */
+	uint32_t param = clk << 16;
+
+	smu_print("SMU Get DC mode max DPM freq: clk = %d\n", clk);
+
+	dcn30_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_GetDcModeMaxDpmFreq, param, &response);
+
+	smu_print("SMU DC mode max DMP freq: %d MHz\n", response);
+
+	return response;
+}
+
+void dcn30_smu_set_min_deep_sleep_dcef_clk(struct clk_mgr_internal *clk_mgr, uint32_t freq_mhz)
+{
+	smu_print("SMU Set min deep sleep dcef clk: freq_mhz = %d MHz\n", freq_mhz);
+
+	dcn30_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_SetMinDeepSleepDcefclk, freq_mhz, NULL);
+}
+
+void dcn30_smu_set_num_of_displays(struct clk_mgr_internal *clk_mgr, uint32_t num_displays)
+{
+	smu_print("SMU Set num of displays: num_displays = %d\n", num_displays);
+
+	dcn30_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_NumOfDisplays, num_displays, NULL);
+}
+
+void dcn30_smu_set_display_refresh_from_mall(struct clk_mgr_internal *clk_mgr, bool enable, uint8_t cache_timer_delay, uint8_t cache_timer_scale)
+{
+	/* bits 8:7 for cache timer scale, bits 6:1 for cache timer delay, bit 0 = 1 for enable, = 0 for disable */
+	uint32_t param = (cache_timer_scale << 7) | (cache_timer_delay << 1) | (enable ? 1 : 0);
+
+	smu_print("SMU Set display refresh from mall: enable = %d, cache_timer_delay = %d, cache_timer_scale = %d\n",
+		enable, cache_timer_delay, cache_timer_scale);
+
+	dcn30_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_SetDisplayRefreshFromMall, param, NULL);
+}
+
+void dcn30_smu_set_external_client_df_cstate_allow(struct clk_mgr_internal *clk_mgr, bool enable)
+{
+	smu_print("SMU Set external client df cstate allow: enable = %d\n", enable);
+
+	dcn30_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_SetExternalClientDfCstateAllow, enable ? 1 : 0, NULL);
+}
+
+void dcn30_smu_set_pme_workaround(struct clk_mgr_internal *clk_mgr)
+{
+	smu_print("SMU Set PME workaround\n");
+
+	dcn30_smu_send_msg_with_param(clk_mgr,
+	DALSMC_MSG_BacoAudioD3PME, 0, NULL);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30_clk_mgr_smu_msg.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30_clk_mgr_smu_msg.h
new file mode 100644
index 000000000000..ca9f5296be94
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30_clk_mgr_smu_msg.h
@@ -0,0 +1,51 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DAL_DC_DCN30_CLK_MGR_SMU_MSG_H_
+#define DAL_DC_DCN30_CLK_MGR_SMU_MSG_H_
+
+#include "core_types.h"
+
+struct clk_mgr_internal;
+
+bool         dcn30_smu_test_message(struct clk_mgr_internal *clk_mgr, uint32_t input);
+bool         dcn30_smu_get_smu_version(struct clk_mgr_internal *clk_mgr, unsigned int *version);
+bool         dcn30_smu_check_driver_if_version(struct clk_mgr_internal *clk_mgr);
+bool         dcn30_smu_check_msg_header_version(struct clk_mgr_internal *clk_mgr);
+void         dcn30_smu_set_dram_addr_high(struct clk_mgr_internal *clk_mgr, uint32_t addr_high);
+void         dcn30_smu_set_dram_addr_low(struct clk_mgr_internal *clk_mgr, uint32_t addr_low);
+void         dcn30_smu_transfer_wm_table_smu_2_dram(struct clk_mgr_internal *clk_mgr);
+void         dcn30_smu_transfer_wm_table_dram_2_smu(struct clk_mgr_internal *clk_mgr);
+unsigned int dcn30_smu_set_hard_min_by_freq(struct clk_mgr_internal *clk_mgr, uint32_t clk, uint16_t freq_mhz);
+unsigned int dcn30_smu_set_hard_max_by_freq(struct clk_mgr_internal *clk_mgr, uint32_t clk, uint16_t freq_mhz);
+unsigned int dcn30_smu_get_dpm_freq_by_index(struct clk_mgr_internal *clk_mgr, uint32_t clk, uint8_t dpm_level);
+unsigned int dcn30_smu_get_dc_mode_max_dpm_freq(struct clk_mgr_internal *clk_mgr, uint32_t clk);
+void         dcn30_smu_set_min_deep_sleep_dcef_clk(struct clk_mgr_internal *clk_mgr, uint32_t freq_mhz);
+void         dcn30_smu_set_num_of_displays(struct clk_mgr_internal *clk_mgr, uint32_t num_displays);
+void         dcn30_smu_set_display_refresh_from_mall(struct clk_mgr_internal *clk_mgr, bool enable, uint8_t cache_timer_delay, uint8_t cache_timer_scale);
+void         dcn30_smu_set_external_client_df_cstate_allow(struct clk_mgr_internal *clk_mgr, bool enable);
+void         dcn30_smu_set_pme_workaround(struct clk_mgr_internal *clk_mgr);
+
+#endif /* DAL_DC_DCN30_CLK_MGR_SMU_MSG_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30_smu11_driver_if.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30_smu11_driver_if.h
new file mode 100644
index 000000000000..1bfd6f66f035
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30_smu11_driver_if.h
@@ -0,0 +1,75 @@
+// SPDX-License-Identifier: MIT
+/* Copyright © 2022-2024 Advanced Micro Devices, Inc. All rights reserved. */
+
+#define SMU11_DRIVER_IF_VERSION 0x40
+
+//Only Clks that have DPM descriptors are listed here
+typedef enum {
+	PPCLK_GFXCLK = 0,
+	PPCLK_SOCCLK,
+	PPCLK_UCLK,
+	PPCLK_FCLK,
+	PPCLK_DCLK_0,
+	PPCLK_VCLK_0,
+	PPCLK_DCLK_1,
+	PPCLK_VCLK_1,
+	PPCLK_DCEFCLK,
+	PPCLK_DISPCLK,
+	PPCLK_PIXCLK,
+	PPCLK_PHYCLK,
+	PPCLK_DTBCLK,
+	PPCLK_COUNT,
+} PPCLK_e;
+
+typedef struct {
+	uint16_t MinClock; // This is either DCEFCLK or SOCCLK (in MHz)
+	uint16_t MaxClock; // This is either DCEFCLK or SOCCLK (in MHz)
+	uint16_t MinUclk;
+	uint16_t MaxUclk;
+
+	uint8_t  WmSetting;
+	uint8_t  Flags;
+	uint8_t  Padding[2];
+
+} WatermarkRowGeneric_t;
+
+#define NUM_WM_RANGES 4
+
+typedef enum {
+	WM_SOCCLK = 0,
+	WM_DCEFCLK,
+	WM_COUNT,
+} WM_CLOCK_e;
+
+typedef enum {
+	WATERMARKS_CLOCK_RANGE = 0,
+	WATERMARKS_DUMMY_PSTATE,
+	WATERMARKS_MALL,
+	WATERMARKS_COUNT,
+} WATERMARKS_FLAGS_e;
+
+typedef struct {
+	// Watermarks
+	WatermarkRowGeneric_t WatermarkRow[WM_COUNT][NUM_WM_RANGES];
+} Watermarks_t;
+
+typedef struct {
+	Watermarks_t Watermarks;
+
+	uint32_t     MmHubPadding[8]; // SMU internal use
+} WatermarksExternal_t;
+
+// Table types
+#define TABLE_PPTABLE                 0
+#define TABLE_WATERMARKS              1
+#define TABLE_AVFS_PSM_DEBUG          2
+#define TABLE_AVFS_FUSE_OVERRIDE      3
+#define TABLE_PMSTATUSLOG             4
+#define TABLE_SMU_METRICS             5
+#define TABLE_DRIVER_SMU_CONFIG       6
+#define TABLE_ACTIVITY_MONITOR_COEFF  7
+#define TABLE_OVERDRIVE               8
+#define TABLE_I2C_COMMANDS            9
+#define TABLE_PACE                   10
+#define TABLE_ECCINFO                11
+#define TABLE_COUNT                  12
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30m_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30m_clk_mgr.c
new file mode 100644
index 000000000000..8e8a11c7437e
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30m_clk_mgr.c
@@ -0,0 +1,36 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "clk_mgr_internal.h"
+#include "dcn30/dcn30m_clk_mgr.h"
+#include "dcn30m_clk_mgr_smu_msg.h"
+
+
+uint32_t dcn30m_set_smartmux_switch(struct clk_mgr *clk_mgr_base, uint32_t pins_to_set)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	return dcn30m_smu_set_smart_mux_switch(clk_mgr, pins_to_set);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30m_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30m_clk_mgr.h
new file mode 100644
index 000000000000..757985b2eadc
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30m_clk_mgr.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCN30M_CLK_MGR_H__
+#define __DCN30M_CLK_MGR_H__
+
+uint32_t dcn30m_set_smartmux_switch(struct clk_mgr *clk_mgr_base, uint32_t pins_to_set);
+
+#endif //__DCN30M_CLK_MGR_H__
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30m_clk_mgr_smu_msg.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30m_clk_mgr_smu_msg.c
new file mode 100644
index 000000000000..0dd0583ff21e
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30m_clk_mgr_smu_msg.c
@@ -0,0 +1,118 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dcn30m_clk_mgr_smu_msg.h"
+
+#include "clk_mgr_internal.h"
+#include "reg_helper.h"
+#include "dm_helpers.h"
+
+#include "dalsmc.h"
+
+#define mmDAL_MSG_REG  0x1628A
+#define mmDAL_ARG_REG  0x16273
+#define mmDAL_RESP_REG 0x16274
+
+#define REG(reg_name) \
+	mm ## reg_name
+
+#include "logger_types.h"
+#undef DC_LOGGER
+#define DC_LOGGER \
+	CTX->logger
+#define smu_print(str, ...) {DC_LOG_SMU(str, ##__VA_ARGS__); }
+
+
+/*
+ * Function to be used instead of REG_WAIT macro because the wait ends when
+ * the register is NOT EQUAL to zero, and because the translation in msg_if.h
+ * won't work with REG_WAIT.
+ */
+static uint32_t dcn30m_smu_wait_for_response(struct clk_mgr_internal *clk_mgr,
+	unsigned int delay_us, unsigned int max_retries)
+{
+	uint32_t reg = 0;
+
+	do {
+		reg = REG_READ(DAL_RESP_REG);
+		if (reg)
+			break;
+
+		if (delay_us >= 1000)
+			msleep(delay_us/1000);
+		else if (delay_us > 0)
+			udelay(delay_us);
+	} while (max_retries--);
+
+	/* handle DALSMC_Result_CmdRejectedBusy? */
+
+	/* Log? */
+
+	return reg;
+}
+
+static bool dcn30m_smu_send_msg_with_param(struct clk_mgr_internal *clk_mgr,
+	uint32_t msg_id, uint32_t param_in, uint32_t *param_out)
+{
+	uint32_t result;
+	/* Wait for response register to be ready */
+	dcn30m_smu_wait_for_response(clk_mgr, 10, 200000);
+
+	/* Clear response register */
+	REG_WRITE(DAL_RESP_REG, 0);
+
+	/* Set the parameter register for the SMU message */
+	REG_WRITE(DAL_ARG_REG, param_in);
+
+	/* Trigger the message transaction by writing the message ID */
+	REG_WRITE(DAL_MSG_REG, msg_id);
+
+	result = dcn30m_smu_wait_for_response(clk_mgr, 10, 200000);
+
+	if (IS_SMU_TIMEOUT(result))
+		dm_helpers_smu_timeout(CTX, msg_id, param_in, 10 * 200000);
+
+	/* Wait for response */
+	if (result == DALSMC_Result_OK) {
+		if (param_out)
+			*param_out = REG_READ(DAL_ARG_REG);
+
+		return true;
+	}
+
+	return false;
+}
+
+uint32_t dcn30m_smu_set_smart_mux_switch(struct clk_mgr_internal *clk_mgr, uint32_t pins_to_set)
+{
+	uint32_t response = 0;
+
+	smu_print("SMU Set SmartMux Switch: switch_dgpu = %d\n", pins_to_set);
+
+	dcn30m_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_SmartAccess, pins_to_set, &response);
+
+	return response;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30m_clk_mgr_smu_msg.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30m_clk_mgr_smu_msg.h
new file mode 100644
index 000000000000..8a59a473fc5e
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn30/dcn30m_clk_mgr_smu_msg.h
@@ -0,0 +1,34 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DAL_DC_DCN30M_CLK_MGR_SMU_MSG_H_
+#define DAL_DC_DCN30M_CLK_MGR_SMU_MSG_H_
+
+#include "core_types.h"
+
+struct clk_mgr_internal;
+
+uint32_t     dcn30m_smu_set_smart_mux_switch(struct clk_mgr_internal *clk_mgr, uint32_t pins_to_set);
+#endif /* DAL_DC_DCN30M_CLK_MGR_SMU_MSG_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn301/dcn301_smu.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn301/dcn301_smu.c
new file mode 100644
index 000000000000..b4fb17b7a096
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn301/dcn301_smu.c
@@ -0,0 +1,281 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "core_types.h"
+#include "clk_mgr_internal.h"
+#include "reg_helper.h"
+#include <linux/delay.h>
+
+#include "dcn301_smu.h"
+#include "dm_helpers.h"
+
+#include "vangogh_ip_offset.h"
+
+#include "mp/mp_11_5_0_offset.h"
+#include "mp/mp_11_5_0_sh_mask.h"
+
+#define REG(reg_name) \
+	(MP0_BASE.instance[0].segment[mm ## reg_name ## _BASE_IDX] + mm ## reg_name)
+
+#define FN(reg_name, field) \
+	FD(reg_name##__##field)
+
+#include "logger_types.h"
+#undef DC_LOGGER
+#define DC_LOGGER \
+	CTX->logger
+#define smu_print(str, ...) {DC_LOG_SMU(str, ##__VA_ARGS__); }
+
+#define VBIOSSMC_MSG_GetSmuVersion                0x2
+#define VBIOSSMC_MSG_SetDispclkFreq               0x4
+#define VBIOSSMC_MSG_SetDprefclkFreq              0x5
+#define VBIOSSMC_MSG_SetDppclkFreq                0x6
+#define VBIOSSMC_MSG_SetHardMinDcfclkByFreq       0x7
+#define VBIOSSMC_MSG_SetMinDeepSleepDcfclk        0x8
+//#define VBIOSSMC_MSG_SetPhyclkVoltageByFreq       0xA
+#define VBIOSSMC_MSG_GetFclkFrequency             0xA
+//#define VBIOSSMC_MSG_SetDisplayCount              0xC
+//#define VBIOSSMC_MSG_EnableTmdp48MHzRefclkPwrDown 0xD
+#define VBIOSSMC_MSG_UpdatePmeRestore			  0xD
+#define VBIOSSMC_MSG_SetVbiosDramAddrHigh         0xE   //Used for WM table txfr
+#define VBIOSSMC_MSG_SetVbiosDramAddrLow          0xF
+#define VBIOSSMC_MSG_TransferTableSmu2Dram        0x10
+#define VBIOSSMC_MSG_TransferTableDram2Smu        0x11
+#define VBIOSSMC_MSG_SetDisplayIdleOptimizations  0x12
+
+#define VBIOSSMC_Status_BUSY                      0x0
+#define VBIOSSMC_Result_OK                        0x1
+#define VBIOSSMC_Result_Failed                    0xFF
+#define VBIOSSMC_Result_UnknownCmd                0xFE
+#define VBIOSSMC_Result_CmdRejectedPrereq         0xFD
+#define VBIOSSMC_Result_CmdRejectedBusy           0xFC
+
+/*
+ * Function to be used instead of REG_WAIT macro because the wait ends when
+ * the register is NOT EQUAL to zero, and because the translation in msg_if.h
+ * won't work with REG_WAIT.
+ */
+static uint32_t dcn301_smu_wait_for_response(struct clk_mgr_internal *clk_mgr, unsigned int delay_us, unsigned int max_retries)
+{
+	uint32_t res_val = VBIOSSMC_Status_BUSY;
+
+	do {
+		res_val = REG_READ(MP1_SMN_C2PMSG_91);
+		if (res_val != VBIOSSMC_Status_BUSY)
+			break;
+
+		if (delay_us >= 1000)
+			msleep(delay_us/1000);
+		else if (delay_us > 0)
+			udelay(delay_us);
+	} while (max_retries--);
+
+	return res_val;
+}
+
+static int dcn301_smu_send_msg_with_param(struct clk_mgr_internal *clk_mgr,
+					  unsigned int msg_id,
+					  unsigned int param)
+{
+	uint32_t result;
+
+	result = dcn301_smu_wait_for_response(clk_mgr, 10, 200000);
+
+	if (result != VBIOSSMC_Result_OK)
+		smu_print("SMU Response was not OK. SMU response after wait received is: %d\n", result);
+
+	if (result == VBIOSSMC_Status_BUSY) {
+		return -1;
+	}
+
+	/* First clear response register */
+	REG_WRITE(MP1_SMN_C2PMSG_91, VBIOSSMC_Status_BUSY);
+
+	/* Set the parameter register for the SMU message, unit is Mhz */
+	REG_WRITE(MP1_SMN_C2PMSG_83, param);
+
+	/* Trigger the message transaction by writing the message ID */
+	REG_WRITE(MP1_SMN_C2PMSG_67, msg_id);
+
+	result = dcn301_smu_wait_for_response(clk_mgr, 10, 200000);
+
+	if (IS_SMU_TIMEOUT(result)) {
+		ASSERT(0);
+		dm_helpers_smu_timeout(CTX, msg_id, param, 10 * 200000);
+	}
+
+	/* Actual dispclk set is returned in the parameter register */
+	return REG_READ(MP1_SMN_C2PMSG_83);
+}
+
+int dcn301_smu_get_smu_version(struct clk_mgr_internal *clk_mgr)
+{
+	int smu_version = dcn301_smu_send_msg_with_param(clk_mgr,
+							 VBIOSSMC_MSG_GetSmuVersion,
+							 0);
+
+	DC_LOG_DEBUG("%s %x\n", __func__, smu_version);
+
+	return smu_version;
+}
+
+
+int dcn301_smu_set_dispclk(struct clk_mgr_internal *clk_mgr, int requested_dispclk_khz)
+{
+	int actual_dispclk_set_mhz = -1;
+
+	DC_LOG_DEBUG("%s(%d)\n", __func__, requested_dispclk_khz);
+
+	/*  Unit of SMU msg parameter is Mhz */
+	actual_dispclk_set_mhz = dcn301_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDispclkFreq,
+			khz_to_mhz_ceil(requested_dispclk_khz));
+
+	return actual_dispclk_set_mhz * 1000;
+}
+
+int dcn301_smu_set_dprefclk(struct clk_mgr_internal *clk_mgr)
+{
+	int actual_dprefclk_set_mhz = -1;
+
+	DC_LOG_DEBUG("%s %d\n", __func__, clk_mgr->base.dprefclk_khz / 1000);
+
+	actual_dprefclk_set_mhz = dcn301_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDprefclkFreq,
+			khz_to_mhz_ceil(clk_mgr->base.dprefclk_khz));
+
+	/* TODO: add code for programing DP DTO, currently this is down by command table */
+
+	return actual_dprefclk_set_mhz * 1000;
+}
+
+int dcn301_smu_set_hard_min_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_dcfclk_khz)
+{
+	int actual_dcfclk_set_mhz = -1;
+
+	DC_LOG_DEBUG("%s(%d)\n", __func__, requested_dcfclk_khz);
+
+	actual_dcfclk_set_mhz = dcn301_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetHardMinDcfclkByFreq,
+			khz_to_mhz_ceil(requested_dcfclk_khz));
+
+	return actual_dcfclk_set_mhz * 1000;
+}
+
+int dcn301_smu_set_min_deep_sleep_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_min_ds_dcfclk_khz)
+{
+	int actual_min_ds_dcfclk_mhz = -1;
+
+	DC_LOG_DEBUG("%s(%d)\n", __func__, requested_min_ds_dcfclk_khz);
+
+	actual_min_ds_dcfclk_mhz = dcn301_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetMinDeepSleepDcfclk,
+			khz_to_mhz_ceil(requested_min_ds_dcfclk_khz));
+
+	return actual_min_ds_dcfclk_mhz * 1000;
+}
+
+int dcn301_smu_set_dppclk(struct clk_mgr_internal *clk_mgr, int requested_dpp_khz)
+{
+	int actual_dppclk_set_mhz = -1;
+
+	DC_LOG_DEBUG("%s(%d)\n", __func__, requested_dpp_khz);
+
+	actual_dppclk_set_mhz = dcn301_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDppclkFreq,
+			khz_to_mhz_ceil(requested_dpp_khz));
+
+	return actual_dppclk_set_mhz * 1000;
+}
+
+void dcn301_smu_set_display_idle_optimization(struct clk_mgr_internal *clk_mgr, uint32_t idle_info)
+{
+	//TODO: Work with smu team to define optimization options.
+
+	DC_LOG_DEBUG("%s(%x)\n", __func__, idle_info);
+
+	dcn301_smu_send_msg_with_param(
+		clk_mgr,
+		VBIOSSMC_MSG_SetDisplayIdleOptimizations,
+		idle_info);
+}
+
+void dcn301_smu_enable_phy_refclk_pwrdwn(struct clk_mgr_internal *clk_mgr, bool enable)
+{
+	union display_idle_optimization_u idle_info = { 0 };
+
+	if (enable) {
+		idle_info.idle_info.df_request_disabled = 1;
+		idle_info.idle_info.phy_ref_clk_off = 1;
+	}
+
+	DC_LOG_DEBUG("%s(%d)\n", __func__, enable);
+
+	dcn301_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDisplayIdleOptimizations,
+			idle_info.data);
+}
+
+void dcn301_smu_enable_pme_wa(struct clk_mgr_internal *clk_mgr)
+{
+	dcn301_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_UpdatePmeRestore,
+			0);
+}
+
+void dcn301_smu_set_dram_addr_high(struct clk_mgr_internal *clk_mgr, uint32_t addr_high)
+{
+	DC_LOG_DEBUG("%s(%x)\n", __func__, addr_high);
+
+	dcn301_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_SetVbiosDramAddrHigh, addr_high);
+}
+
+void dcn301_smu_set_dram_addr_low(struct clk_mgr_internal *clk_mgr, uint32_t addr_low)
+{
+	DC_LOG_DEBUG("%s(%x)\n", __func__, addr_low);
+
+	dcn301_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_SetVbiosDramAddrLow, addr_low);
+}
+
+void dcn301_smu_transfer_dpm_table_smu_2_dram(struct clk_mgr_internal *clk_mgr)
+{
+	dcn301_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_TransferTableSmu2Dram, TABLE_DPMCLOCKS);
+}
+
+void dcn301_smu_transfer_wm_table_dram_2_smu(struct clk_mgr_internal *clk_mgr)
+{
+	dcn301_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_TransferTableDram2Smu, TABLE_WATERMARKS);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn301/dcn301_smu.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn301/dcn301_smu.h
new file mode 100644
index 000000000000..b640df85a17f
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn301/dcn301_smu.h
@@ -0,0 +1,164 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DAL_DC_301_SMU_H_
+#define DAL_DC_301_SMU_H_
+
+#define SMU13_DRIVER_IF_VERSION 2
+
+typedef struct {
+	uint32_t fclk;
+	uint32_t memclk;
+	uint32_t voltage;
+} df_pstate_t;
+
+typedef struct {
+	uint32_t vclk;
+	uint32_t dclk;
+} vcn_clk_t;
+
+typedef enum {
+	DSPCLK_DCFCLK = 0,
+	DSPCLK_DISPCLK,
+	DSPCLK_PIXCLK,
+	DSPCLK_PHYCLK,
+	DSPCLK_COUNT,
+} DSPCLK_e;
+
+typedef struct {
+	uint16_t Freq; // in MHz
+	uint16_t Vid;  // min voltage in SVI2 VID
+} DisplayClockTable_t;
+
+typedef struct {
+	uint16_t MinClock; // This is either DCFCLK or SOCCLK (in MHz)
+	uint16_t MaxClock; // This is either DCFCLK or SOCCLK (in MHz)
+	uint16_t MinMclk;
+	uint16_t MaxMclk;
+
+	uint8_t  WmSetting;
+	uint8_t  WmType;  // Used for normal pstate change or memory retraining
+	uint8_t  Padding[2];
+} WatermarkRowGeneric_t;
+
+
+#define NUM_WM_RANGES 4
+
+typedef enum {
+	WM_SOCCLK = 0,
+	WM_DCFCLK,
+	WM_COUNT,
+} WM_CLOCK_e;
+
+typedef struct {
+  // Watermarks
+	WatermarkRowGeneric_t WatermarkRow[WM_COUNT][NUM_WM_RANGES];
+
+	uint32_t     MmHubPadding[7]; // SMU internal use
+} Watermarks_t;
+
+
+#define TABLE_WATERMARKS         1
+#define TABLE_DPMCLOCKS          4 // Called by Driver
+
+
+#define VG_NUM_DCFCLK_DPM_LEVELS   7
+#define VG_NUM_DISPCLK_DPM_LEVELS  7
+#define VG_NUM_DPPCLK_DPM_LEVELS   7
+#define VG_NUM_SOCCLK_DPM_LEVELS   7
+#define VG_NUM_ISPICLK_DPM_LEVELS  7
+#define VG_NUM_ISPXCLK_DPM_LEVELS  7
+#define VG_NUM_VCN_DPM_LEVELS      5
+#define VG_NUM_FCLK_DPM_LEVELS     4
+#define VG_NUM_SOC_VOLTAGE_LEVELS  8
+
+// copy from vgh/vangogh/pmfw_driver_if.h
+struct vg_dpm_clocks {
+	uint32_t DcfClocks[VG_NUM_DCFCLK_DPM_LEVELS];
+	uint32_t DispClocks[VG_NUM_DISPCLK_DPM_LEVELS];
+	uint32_t DppClocks[VG_NUM_DPPCLK_DPM_LEVELS];
+	uint32_t SocClocks[VG_NUM_SOCCLK_DPM_LEVELS];
+	uint32_t IspiClocks[VG_NUM_ISPICLK_DPM_LEVELS];
+	uint32_t IspxClocks[VG_NUM_ISPXCLK_DPM_LEVELS];
+	vcn_clk_t VcnClocks[VG_NUM_VCN_DPM_LEVELS];
+
+	uint32_t SocVoltage[VG_NUM_SOC_VOLTAGE_LEVELS];
+
+	df_pstate_t DfPstateTable[VG_NUM_FCLK_DPM_LEVELS];
+
+	uint32_t MinGfxClk;
+	uint32_t MaxGfxClk;
+
+	uint8_t NumDfPstatesEnabled;
+	uint8_t NumDcfclkLevelsEnabled;
+	uint8_t NumDispClkLevelsEnabled;  //applies to both dispclk and dppclk
+	uint8_t NumSocClkLevelsEnabled;
+
+	uint8_t IspClkLevelsEnabled;  //applies to both ispiclk and ispxclk
+	uint8_t VcnClkLevelsEnabled;  //applies to both vclk/dclk
+	uint8_t spare[2];
+};
+
+struct smu_dpm_clks {
+	struct vg_dpm_clocks *dpm_clks;
+	union large_integer mc_address;
+};
+
+struct watermarks {
+  // Watermarks
+	WatermarkRowGeneric_t WatermarkRow[WM_COUNT][NUM_WM_RANGES];
+
+	uint32_t     MmHubPadding[7]; // SMU internal use
+};
+
+
+struct display_idle_optimization {
+	unsigned int df_request_disabled : 1;
+	unsigned int phy_ref_clk_off     : 1;
+	unsigned int s0i2_rdy            : 1;
+	unsigned int reserved            : 29;
+};
+
+union display_idle_optimization_u {
+	struct display_idle_optimization idle_info;
+	uint32_t data;
+};
+
+
+int dcn301_smu_get_smu_version(struct clk_mgr_internal *clk_mgr);
+int dcn301_smu_set_dispclk(struct clk_mgr_internal *clk_mgr, int requested_dispclk_khz);
+int dcn301_smu_set_dprefclk(struct clk_mgr_internal *clk_mgr);
+int dcn301_smu_set_hard_min_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_dcfclk_khz);
+int dcn301_smu_set_min_deep_sleep_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_min_ds_dcfclk_khz);
+int dcn301_smu_set_dppclk(struct clk_mgr_internal *clk_mgr, int requested_dpp_khz);
+void dcn301_smu_set_display_idle_optimization(struct clk_mgr_internal *clk_mgr, uint32_t idle_info);
+void dcn301_smu_enable_phy_refclk_pwrdwn(struct clk_mgr_internal *clk_mgr, bool enable);
+void dcn301_smu_enable_pme_wa(struct clk_mgr_internal *clk_mgr);
+void dcn301_smu_set_dram_addr_high(struct clk_mgr_internal *clk_mgr, uint32_t addr_high);
+void dcn301_smu_set_dram_addr_low(struct clk_mgr_internal *clk_mgr, uint32_t addr_low);
+void dcn301_smu_transfer_dpm_table_smu_2_dram(struct clk_mgr_internal *clk_mgr);
+void dcn301_smu_transfer_wm_table_dram_2_smu(struct clk_mgr_internal *clk_mgr);
+
+#endif /* DAL_DC_301_SMU_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn301/vg_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn301/vg_clk_mgr.c
new file mode 100644
index 000000000000..7aee02d56292
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn301/vg_clk_mgr.c
@@ -0,0 +1,769 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dccg.h"
+#include "clk_mgr_internal.h"
+
+// For dce12_get_dp_ref_freq_khz
+#include "dce100/dce_clk_mgr.h"
+
+// For dcn20_update_clocks_update_dpp_dto
+#include "dcn20/dcn20_clk_mgr.h"
+
+// For DML FPU code
+#include "dml/dcn20/dcn20_fpu.h"
+
+#include "vg_clk_mgr.h"
+#include "dcn301_smu.h"
+#include "reg_helper.h"
+#include "core_types.h"
+#include "dm_helpers.h"
+
+#include "atomfirmware.h"
+#include "vangogh_ip_offset.h"
+#include "clk/clk_11_5_0_offset.h"
+#include "clk/clk_11_5_0_sh_mask.h"
+
+/* Constants */
+
+#define LPDDR_MEM_RETRAIN_LATENCY 4.977 /* Number obtained from LPDDR4 Training Counter Requirement doc */
+
+/* Macros */
+
+#define TO_CLK_MGR_VGH(clk_mgr)\
+	container_of(clk_mgr, struct clk_mgr_vgh, base)
+
+#define REG(reg_name) \
+	(CLK_BASE.instance[0].segment[mm ## reg_name ## _BASE_IDX] + mm ## reg_name)
+
+/* TODO: evaluate how to lower or disable all dcn clocks in screen off case */
+static int vg_get_active_display_cnt_wa(
+		struct dc *dc,
+		struct dc_state *context)
+{
+	int i, display_count;
+	bool tmds_present = false;
+
+	display_count = 0;
+	for (i = 0; i < context->stream_count; i++) {
+		const struct dc_stream_state *stream = context->streams[i];
+
+		if (stream->signal == SIGNAL_TYPE_HDMI_TYPE_A ||
+				stream->signal == SIGNAL_TYPE_DVI_SINGLE_LINK ||
+				stream->signal == SIGNAL_TYPE_DVI_DUAL_LINK)
+			tmds_present = true;
+	}
+
+	for (i = 0; i < dc->link_count; i++) {
+		const struct dc_link *link = dc->links[i];
+
+		/* abusing the fact that the dig and phy are coupled to see if the phy is enabled */
+		if (link->link_enc->funcs->is_dig_enabled &&
+				link->link_enc->funcs->is_dig_enabled(link->link_enc))
+			display_count++;
+	}
+
+	/* WA for hang on HDMI after display off back back on*/
+	if (display_count == 0 && tmds_present)
+		display_count = 1;
+
+	return display_count;
+}
+
+static void vg_update_clocks(struct clk_mgr *clk_mgr_base,
+			     struct dc_state *context,
+			     bool safe_to_lower)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dc_clocks *new_clocks = &context->bw_ctx.bw.dcn.clk;
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	int display_count;
+	bool update_dppclk = false;
+	bool update_dispclk = false;
+	bool dpp_clock_lowered = false;
+
+	if (dc->work_arounds.skip_clock_update)
+		return;
+
+	/*
+	 * if it is safe to lower, but we are already in the lower state, we don't have to do anything
+	 * also if safe to lower is false, we just go in the higher state
+	 */
+	if (safe_to_lower) {
+		/* check that we're not already in lower */
+		if (clk_mgr_base->clks.pwr_state != DCN_PWR_STATE_LOW_POWER) {
+
+			display_count = vg_get_active_display_cnt_wa(dc, context);
+			/* if we can go lower, go lower */
+			if (display_count == 0) {
+				union display_idle_optimization_u idle_info = { 0 };
+
+				idle_info.idle_info.df_request_disabled = 1;
+				idle_info.idle_info.phy_ref_clk_off = 1;
+
+				dcn301_smu_set_display_idle_optimization(clk_mgr, idle_info.data);
+				/* update power state */
+				clk_mgr_base->clks.pwr_state = DCN_PWR_STATE_LOW_POWER;
+			}
+		}
+	} else {
+		/* check that we're not already in D0 */
+		if (clk_mgr_base->clks.pwr_state != DCN_PWR_STATE_MISSION_MODE) {
+			union display_idle_optimization_u idle_info = { 0 };
+
+			dcn301_smu_set_display_idle_optimization(clk_mgr, idle_info.data);
+			/* update power state */
+			clk_mgr_base->clks.pwr_state = DCN_PWR_STATE_MISSION_MODE;
+		}
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dcfclk_khz, clk_mgr_base->clks.dcfclk_khz) && !dc->debug.disable_min_fclk) {
+		clk_mgr_base->clks.dcfclk_khz = new_clocks->dcfclk_khz;
+		dcn301_smu_set_hard_min_dcfclk(clk_mgr, clk_mgr_base->clks.dcfclk_khz);
+	}
+
+	if (should_set_clock(safe_to_lower,
+			new_clocks->dcfclk_deep_sleep_khz, clk_mgr_base->clks.dcfclk_deep_sleep_khz) && !dc->debug.disable_min_fclk) {
+		clk_mgr_base->clks.dcfclk_deep_sleep_khz = new_clocks->dcfclk_deep_sleep_khz;
+		dcn301_smu_set_min_deep_sleep_dcfclk(clk_mgr, clk_mgr_base->clks.dcfclk_deep_sleep_khz);
+	}
+
+	// workaround: Limit dppclk to 100Mhz to avoid lower eDP panel switch to plus 4K monitor underflow.
+	if (new_clocks->dppclk_khz < 100000)
+		new_clocks->dppclk_khz = 100000;
+
+	if (should_set_clock(safe_to_lower, new_clocks->dppclk_khz, clk_mgr->base.clks.dppclk_khz)) {
+		if (clk_mgr->base.clks.dppclk_khz > new_clocks->dppclk_khz)
+			dpp_clock_lowered = true;
+		clk_mgr_base->clks.dppclk_khz = new_clocks->dppclk_khz;
+		update_dppclk = true;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, clk_mgr_base->clks.dispclk_khz)) {
+		clk_mgr_base->clks.dispclk_khz = new_clocks->dispclk_khz;
+		dcn301_smu_set_dispclk(clk_mgr, clk_mgr_base->clks.dispclk_khz);
+
+		update_dispclk = true;
+	}
+
+	if (dpp_clock_lowered) {
+		// increase per DPP DTO before lowering global dppclk
+		dcn20_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
+		dcn301_smu_set_dppclk(clk_mgr, clk_mgr_base->clks.dppclk_khz);
+	} else {
+		// increase global DPPCLK before lowering per DPP DTO
+		if (update_dppclk || update_dispclk)
+			dcn301_smu_set_dppclk(clk_mgr, clk_mgr_base->clks.dppclk_khz);
+		// always update dtos unless clock is lowered and not safe to lower
+		dcn20_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
+	}
+}
+
+
+static int get_vco_frequency_from_reg(struct clk_mgr_internal *clk_mgr)
+{
+	/* get FbMult value */
+	struct fixed31_32 pll_req;
+	unsigned int fbmult_frac_val = 0;
+	unsigned int fbmult_int_val = 0;
+
+
+	/*
+	 * Register value of fbmult is in 8.16 format, we are converting to 31.32
+	 * to leverage the fix point operations available in driver
+	 */
+
+	REG_GET(CLK1_0_CLK1_CLK_PLL_REQ, FbMult_frac, &fbmult_frac_val); /* 16 bit fractional part*/
+	REG_GET(CLK1_0_CLK1_CLK_PLL_REQ, FbMult_int, &fbmult_int_val); /* 8 bit integer part */
+
+	pll_req = dc_fixpt_from_int(fbmult_int_val);
+
+	/*
+	 * since fractional part is only 16 bit in register definition but is 32 bit
+	 * in our fix point definiton, need to shift left by 16 to obtain correct value
+	 */
+	pll_req.value |= fbmult_frac_val << 16;
+
+	/* multiply by REFCLK period */
+	pll_req = dc_fixpt_mul_int(pll_req, clk_mgr->dfs_ref_freq_khz);
+
+	/* integer part is now VCO frequency in kHz */
+	return dc_fixpt_floor(pll_req);
+}
+
+static void vg_dump_clk_registers_internal(struct dcn301_clk_internal *internal, struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	internal->CLK1_CLK3_CURRENT_CNT = REG_READ(CLK1_0_CLK1_CLK3_CURRENT_CNT);
+	internal->CLK1_CLK3_BYPASS_CNTL = REG_READ(CLK1_0_CLK1_CLK3_BYPASS_CNTL);
+
+	internal->CLK1_CLK3_DS_CNTL = REG_READ(CLK1_0_CLK1_CLK3_DS_CNTL);	//dcf deep sleep divider
+	internal->CLK1_CLK3_ALLOW_DS = REG_READ(CLK1_0_CLK1_CLK3_ALLOW_DS);
+
+	internal->CLK1_CLK1_CURRENT_CNT = REG_READ(CLK1_0_CLK1_CLK1_CURRENT_CNT);
+	internal->CLK1_CLK1_BYPASS_CNTL = REG_READ(CLK1_0_CLK1_CLK1_BYPASS_CNTL);
+
+	internal->CLK1_CLK2_CURRENT_CNT = REG_READ(CLK1_0_CLK1_CLK2_CURRENT_CNT);
+	internal->CLK1_CLK2_BYPASS_CNTL = REG_READ(CLK1_0_CLK1_CLK2_BYPASS_CNTL);
+
+	internal->CLK1_CLK0_CURRENT_CNT = REG_READ(CLK1_0_CLK1_CLK0_CURRENT_CNT);
+	internal->CLK1_CLK0_BYPASS_CNTL = REG_READ(CLK1_0_CLK1_CLK0_BYPASS_CNTL);
+}
+
+/* This function collect raw clk register values */
+static void vg_dump_clk_registers(struct clk_state_registers_and_bypass *regs_and_bypass,
+		struct clk_mgr *clk_mgr_base, struct clk_log_info *log_info)
+{
+	struct dcn301_clk_internal internal = {0};
+	char *bypass_clks[5] = {"0x0 DFS", "0x1 REFCLK", "0x2 ERROR", "0x3 400 FCH", "0x4 600 FCH"};
+	unsigned int chars_printed = 0;
+	unsigned int remaining_buffer = log_info->bufSize;
+
+	vg_dump_clk_registers_internal(&internal, clk_mgr_base);
+
+	regs_and_bypass->dcfclk = internal.CLK1_CLK3_CURRENT_CNT / 10;
+	regs_and_bypass->dcf_deep_sleep_divider = internal.CLK1_CLK3_DS_CNTL / 10;
+	regs_and_bypass->dcf_deep_sleep_allow = internal.CLK1_CLK3_ALLOW_DS;
+	regs_and_bypass->dprefclk = internal.CLK1_CLK2_CURRENT_CNT / 10;
+	regs_and_bypass->dispclk = internal.CLK1_CLK0_CURRENT_CNT / 10;
+	regs_and_bypass->dppclk = internal.CLK1_CLK1_CURRENT_CNT / 10;
+
+	regs_and_bypass->dppclk_bypass = internal.CLK1_CLK1_BYPASS_CNTL & 0x0007;
+	if (regs_and_bypass->dppclk_bypass > 4)
+		regs_and_bypass->dppclk_bypass = 0;
+	regs_and_bypass->dcfclk_bypass = internal.CLK1_CLK3_BYPASS_CNTL & 0x0007;
+	if (regs_and_bypass->dcfclk_bypass > 4)
+		regs_and_bypass->dcfclk_bypass = 0;
+	regs_and_bypass->dispclk_bypass = internal.CLK1_CLK0_BYPASS_CNTL & 0x0007;
+	if (regs_and_bypass->dispclk_bypass > 4)
+		regs_and_bypass->dispclk_bypass = 0;
+	regs_and_bypass->dprefclk_bypass = internal.CLK1_CLK2_BYPASS_CNTL & 0x0007;
+	if (regs_and_bypass->dprefclk_bypass > 4)
+		regs_and_bypass->dprefclk_bypass = 0;
+
+	if (log_info->enabled) {
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "clk_type,clk_value,deepsleep_cntl,deepsleep_allow,bypass\n");
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "dcfclk,%d,%d,%d,%s\n",
+			regs_and_bypass->dcfclk,
+			regs_and_bypass->dcf_deep_sleep_divider,
+			regs_and_bypass->dcf_deep_sleep_allow,
+			bypass_clks[(int) regs_and_bypass->dcfclk_bypass]);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "dprefclk,%d,N/A,N/A,%s\n",
+			regs_and_bypass->dprefclk,
+			bypass_clks[(int) regs_and_bypass->dprefclk_bypass]);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "dispclk,%d,N/A,N/A,%s\n",
+			regs_and_bypass->dispclk,
+			bypass_clks[(int) regs_and_bypass->dispclk_bypass]);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		//split
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "SPLIT\n");
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		// REGISTER VALUES
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "reg_name,value,clk_type\n");
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "CLK1_CLK3_CURRENT_CNT,%d,dcfclk\n",
+				internal.CLK1_CLK3_CURRENT_CNT);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "CLK1_CLK3_DS_CNTL,%d,dcf_deep_sleep_divider\n",
+					internal.CLK1_CLK3_DS_CNTL);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "CLK1_CLK3_ALLOW_DS,%d,dcf_deep_sleep_allow\n",
+					internal.CLK1_CLK3_ALLOW_DS);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "CLK1_CLK2_CURRENT_CNT,%d,dprefclk\n",
+					internal.CLK1_CLK2_CURRENT_CNT);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "CLK1_CLK0_CURRENT_CNT,%d,dispclk\n",
+					internal.CLK1_CLK0_CURRENT_CNT);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "CLK1_CLK1_CURRENT_CNT,%d,dppclk\n",
+					internal.CLK1_CLK1_CURRENT_CNT);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "CLK1_CLK3_BYPASS_CNTL,%d,dcfclk_bypass\n",
+					internal.CLK1_CLK3_BYPASS_CNTL);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "CLK1_CLK2_BYPASS_CNTL,%d,dprefclk_bypass\n",
+					internal.CLK1_CLK2_BYPASS_CNTL);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "CLK1_CLK0_BYPASS_CNTL,%d,dispclk_bypass\n",
+					internal.CLK1_CLK0_BYPASS_CNTL);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+
+		chars_printed = snprintf_count(log_info->pBuf, remaining_buffer, "CLK1_CLK1_BYPASS_CNTL,%d,dppclk_bypass\n",
+					internal.CLK1_CLK1_BYPASS_CNTL);
+		remaining_buffer -= chars_printed;
+		*log_info->sum_chars_printed += chars_printed;
+		log_info->pBuf += chars_printed;
+	}
+}
+
+static void vg_enable_pme_wa(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	dcn301_smu_enable_pme_wa(clk_mgr);
+}
+
+static void vg_init_clocks(struct clk_mgr *clk_mgr)
+{
+	memset(&(clk_mgr->clks), 0, sizeof(struct dc_clocks));
+	// Assumption is that boot state always supports pstate
+	clk_mgr->clks.p_state_change_support = true;
+	clk_mgr->clks.prev_p_state_change_support = true;
+	clk_mgr->clks.pwr_state = DCN_PWR_STATE_UNKNOWN;
+}
+
+static void vg_build_watermark_ranges(struct clk_bw_params *bw_params, struct watermarks *table)
+{
+	int i, num_valid_sets;
+
+	num_valid_sets = 0;
+
+	for (i = 0; i < WM_SET_COUNT; i++) {
+		/* skip empty entries, the smu array has no holes*/
+		if (!bw_params->wm_table.entries[i].valid)
+			continue;
+
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].WmSetting = bw_params->wm_table.entries[i].wm_inst;
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].WmType = bw_params->wm_table.entries[i].wm_type;
+		/* We will not select WM based on fclk, so leave it as unconstrained */
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinClock = 0;
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].MaxClock = 0xFFFF;
+
+		if (table->WatermarkRow[WM_DCFCLK][num_valid_sets].WmType == WM_TYPE_PSTATE_CHG) {
+			if (i == 0)
+				table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinMclk = 0;
+			else {
+				/* add 1 to make it non-overlapping with next lvl */
+				table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinMclk =
+						bw_params->clk_table.entries[i - 1].dcfclk_mhz + 1;
+			}
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets].MaxMclk =
+					bw_params->clk_table.entries[i].dcfclk_mhz;
+
+		} else {
+			/* unconstrained for memory retraining */
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinClock = 0;
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets].MaxClock = 0xFFFF;
+
+			/* Modify previous watermark range to cover up to max */
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets - 1].MaxClock = 0xFFFF;
+		}
+		num_valid_sets++;
+	}
+
+	ASSERT(num_valid_sets != 0); /* Must have at least one set of valid watermarks */
+
+	/* modify the min and max to make sure we cover the whole range*/
+	table->WatermarkRow[WM_DCFCLK][0].MinMclk = 0;
+	table->WatermarkRow[WM_DCFCLK][0].MinClock = 0;
+	table->WatermarkRow[WM_DCFCLK][num_valid_sets - 1].MaxMclk = 0xFFFF;
+	table->WatermarkRow[WM_DCFCLK][num_valid_sets - 1].MaxClock = 0xFFFF;
+
+	/* This is for writeback only, does not matter currently as no writeback support*/
+	table->WatermarkRow[WM_SOCCLK][0].WmSetting = WM_A;
+	table->WatermarkRow[WM_SOCCLK][0].MinClock = 0;
+	table->WatermarkRow[WM_SOCCLK][0].MaxClock = 0xFFFF;
+	table->WatermarkRow[WM_SOCCLK][0].MinMclk = 0;
+	table->WatermarkRow[WM_SOCCLK][0].MaxMclk = 0xFFFF;
+}
+
+
+static void vg_notify_wm_ranges(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct clk_mgr_vgh *clk_mgr_vgh = TO_CLK_MGR_VGH(clk_mgr);
+	struct watermarks *table = clk_mgr_vgh->smu_wm_set.wm_set;
+
+	if (!clk_mgr->smu_ver)
+		return;
+
+	if (!table || clk_mgr_vgh->smu_wm_set.mc_address.quad_part == 0)
+		return;
+
+	memset(table, 0, sizeof(*table));
+
+	vg_build_watermark_ranges(clk_mgr_base->bw_params, table);
+
+	dcn301_smu_set_dram_addr_high(clk_mgr,
+			clk_mgr_vgh->smu_wm_set.mc_address.high_part);
+	dcn301_smu_set_dram_addr_low(clk_mgr,
+			clk_mgr_vgh->smu_wm_set.mc_address.low_part);
+	dcn301_smu_transfer_wm_table_dram_2_smu(clk_mgr);
+}
+
+static bool vg_are_clock_states_equal(struct dc_clocks *a,
+		struct dc_clocks *b)
+{
+	if (a->dispclk_khz != b->dispclk_khz)
+		return false;
+	else if (a->dppclk_khz != b->dppclk_khz)
+		return false;
+	else if (a->dcfclk_khz != b->dcfclk_khz)
+		return false;
+	else if (a->dcfclk_deep_sleep_khz != b->dcfclk_deep_sleep_khz)
+		return false;
+
+	return true;
+}
+
+
+static struct clk_mgr_funcs vg_funcs = {
+	.get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz,
+	.update_clocks = vg_update_clocks,
+	.init_clocks = vg_init_clocks,
+	.enable_pme_wa = vg_enable_pme_wa,
+	.are_clock_states_equal = vg_are_clock_states_equal,
+	.notify_wm_ranges = vg_notify_wm_ranges
+};
+
+static struct clk_bw_params vg_bw_params = {
+	.vram_type = Ddr4MemType,
+	.num_channels = 1,
+	.clk_table = {
+		.entries = {
+			{
+				.voltage = 0,
+				.dcfclk_mhz = 400,
+				.fclk_mhz = 400,
+				.memclk_mhz = 800,
+				.socclk_mhz = 0,
+			},
+			{
+				.voltage = 0,
+				.dcfclk_mhz = 483,
+				.fclk_mhz = 800,
+				.memclk_mhz = 1600,
+				.socclk_mhz = 0,
+			},
+			{
+				.voltage = 0,
+				.dcfclk_mhz = 602,
+				.fclk_mhz = 1067,
+				.memclk_mhz = 1067,
+				.socclk_mhz = 0,
+			},
+			{
+				.voltage = 0,
+				.dcfclk_mhz = 738,
+				.fclk_mhz = 1333,
+				.memclk_mhz = 1600,
+				.socclk_mhz = 0,
+			},
+		},
+
+		.num_entries = 4,
+	},
+
+};
+
+static uint32_t find_max_clk_value(const uint32_t clocks[], uint32_t num_clocks)
+{
+	uint32_t max = 0;
+	int i;
+
+	for (i = 0; i < num_clocks; ++i) {
+		if (clocks[i] > max)
+			max = clocks[i];
+	}
+
+	return max;
+}
+
+static unsigned int find_dcfclk_for_voltage(const struct vg_dpm_clocks *clock_table,
+		unsigned int voltage)
+{
+	int i;
+
+	for (i = 0; i < VG_NUM_SOC_VOLTAGE_LEVELS; i++) {
+		if (i >= VG_NUM_DCFCLK_DPM_LEVELS)
+			break;
+		if (clock_table->SocVoltage[i] == voltage)
+			return clock_table->DcfClocks[i];
+	}
+
+	ASSERT(0);
+	return 0;
+}
+
+static void vg_clk_mgr_helper_populate_bw_params(
+		struct clk_mgr_internal *clk_mgr,
+		struct integrated_info *bios_info,
+		const struct vg_dpm_clocks *clock_table)
+{
+	int i, j;
+	struct clk_bw_params *bw_params = clk_mgr->base.bw_params;
+	uint32_t max_dispclk = 0, max_dppclk = 0;
+
+	j = -1;
+
+	static_assert(VG_NUM_FCLK_DPM_LEVELS <= MAX_NUM_DPM_LVL,
+		"number of reported FCLK DPM levels exceeds maximum");
+
+	/* Find lowest DPM, FCLK is filled in reverse order*/
+
+	for (i = VG_NUM_FCLK_DPM_LEVELS - 1; i >= 0; i--) {
+		if (clock_table->DfPstateTable[i].fclk != 0) {
+			j = i;
+			break;
+		}
+	}
+
+	if (j == -1) {
+		/* clock table is all 0s, just use our own hardcode */
+		ASSERT(0);
+		return;
+	}
+
+	/* dispclk and dppclk can be max at any voltage, same number of levels for both */
+	if (clock_table->NumDispClkLevelsEnabled <= VG_NUM_DISPCLK_DPM_LEVELS &&
+	    clock_table->NumDispClkLevelsEnabled <= VG_NUM_DPPCLK_DPM_LEVELS) {
+		max_dispclk = find_max_clk_value(clock_table->DispClocks, clock_table->NumDispClkLevelsEnabled);
+		max_dppclk = find_max_clk_value(clock_table->DppClocks, clock_table->NumDispClkLevelsEnabled);
+	} else {
+		ASSERT(0);
+	}
+
+	bw_params->clk_table.num_entries = j + 1;
+
+	for (i = 0; i < bw_params->clk_table.num_entries - 1; i++, j--) {
+		bw_params->clk_table.entries[i].fclk_mhz = clock_table->DfPstateTable[j].fclk;
+		bw_params->clk_table.entries[i].memclk_mhz = clock_table->DfPstateTable[j].memclk;
+		bw_params->clk_table.entries[i].voltage = clock_table->DfPstateTable[j].voltage;
+		bw_params->clk_table.entries[i].dcfclk_mhz = find_dcfclk_for_voltage(clock_table, clock_table->DfPstateTable[j].voltage);
+
+		/* Now update clocks we do read */
+		bw_params->clk_table.entries[i].dispclk_mhz = max_dispclk;
+		bw_params->clk_table.entries[i].dppclk_mhz = max_dppclk;
+	}
+	bw_params->clk_table.entries[i].fclk_mhz = clock_table->DfPstateTable[j].fclk;
+	bw_params->clk_table.entries[i].memclk_mhz = clock_table->DfPstateTable[j].memclk;
+	bw_params->clk_table.entries[i].voltage = clock_table->DfPstateTable[j].voltage;
+	bw_params->clk_table.entries[i].dcfclk_mhz = find_max_clk_value(clock_table->DcfClocks, VG_NUM_DCFCLK_DPM_LEVELS);
+	bw_params->clk_table.entries[i].dispclk_mhz = find_max_clk_value(clock_table->DispClocks, VG_NUM_DISPCLK_DPM_LEVELS);
+	bw_params->clk_table.entries[i].dppclk_mhz = find_max_clk_value(clock_table->DppClocks, VG_NUM_DPPCLK_DPM_LEVELS);
+
+	bw_params->vram_type = bios_info->memory_type;
+	bw_params->num_channels = bios_info->ma_channel_number;
+
+	for (i = 0; i < WM_SET_COUNT; i++) {
+		bw_params->wm_table.entries[i].wm_inst = i;
+
+		if (i >= bw_params->clk_table.num_entries) {
+			bw_params->wm_table.entries[i].valid = false;
+			continue;
+		}
+
+		bw_params->wm_table.entries[i].wm_type = WM_TYPE_PSTATE_CHG;
+		bw_params->wm_table.entries[i].valid = true;
+	}
+
+	if (bw_params->vram_type == LpDdr4MemType) {
+		/*
+		 * WM set D will be re-purposed for memory retraining
+		 */
+		DC_FP_START();
+		dcn21_clk_mgr_set_bw_params_wm_table(bw_params);
+		DC_FP_END();
+	}
+
+}
+
+/* Temporary Place holder until we can get them from fuse */
+static struct vg_dpm_clocks dummy_clocks = {
+		.DcfClocks = { 201, 403, 403, 403, 403, 403, 403 },
+		.SocClocks = { 400, 600, 600, 600, 600, 600, 600 },
+		.SocVoltage = { 2800, 2860, 2860, 2860, 2860, 2860, 2860, 2860 },
+		.DfPstateTable = {
+				{ .fclk = 400,  .memclk = 400, .voltage = 2800 },
+				{ .fclk = 400,  .memclk = 400, .voltage = 2800 },
+				{ .fclk = 400,  .memclk = 400, .voltage = 2800 },
+				{ .fclk = 400,  .memclk = 400, .voltage = 2800 }
+		}
+};
+
+static struct watermarks dummy_wms = { 0 };
+
+static void vg_get_dpm_table_from_smu(struct clk_mgr_internal *clk_mgr,
+		struct smu_dpm_clks *smu_dpm_clks)
+{
+	struct vg_dpm_clocks *table = smu_dpm_clks->dpm_clks;
+
+	if (!clk_mgr->smu_ver)
+		return;
+
+	if (!table || smu_dpm_clks->mc_address.quad_part == 0)
+		return;
+
+	memset(table, 0, sizeof(*table));
+
+	dcn301_smu_set_dram_addr_high(clk_mgr,
+			smu_dpm_clks->mc_address.high_part);
+	dcn301_smu_set_dram_addr_low(clk_mgr,
+			smu_dpm_clks->mc_address.low_part);
+	dcn301_smu_transfer_dpm_table_smu_2_dram(clk_mgr);
+}
+
+void vg_clk_mgr_construct(
+		struct dc_context *ctx,
+		struct clk_mgr_vgh *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg)
+{
+	struct smu_dpm_clks smu_dpm_clks = { 0 };
+	struct clk_log_info log_info = {0};
+
+	clk_mgr->base.base.ctx = ctx;
+	clk_mgr->base.base.funcs = &vg_funcs;
+
+	clk_mgr->base.pp_smu = pp_smu;
+
+	clk_mgr->base.dccg = dccg;
+	clk_mgr->base.dfs_bypass_disp_clk = 0;
+
+	clk_mgr->base.dprefclk_ss_percentage = 0;
+	clk_mgr->base.dprefclk_ss_divider = 1000;
+	clk_mgr->base.ss_on_dprefclk = false;
+	clk_mgr->base.dfs_ref_freq_khz = 48000;
+
+	clk_mgr->smu_wm_set.wm_set = (struct watermarks *)dm_helpers_allocate_gpu_mem(
+				clk_mgr->base.base.ctx,
+				DC_MEM_ALLOC_TYPE_FRAME_BUFFER,
+				sizeof(struct watermarks),
+				&clk_mgr->smu_wm_set.mc_address.quad_part);
+
+	if (!clk_mgr->smu_wm_set.wm_set) {
+		clk_mgr->smu_wm_set.wm_set = &dummy_wms;
+		clk_mgr->smu_wm_set.mc_address.quad_part = 0;
+	}
+	ASSERT(clk_mgr->smu_wm_set.wm_set);
+
+	smu_dpm_clks.dpm_clks = (struct vg_dpm_clocks *)dm_helpers_allocate_gpu_mem(
+				clk_mgr->base.base.ctx,
+				DC_MEM_ALLOC_TYPE_FRAME_BUFFER,
+				sizeof(struct vg_dpm_clocks),
+				&smu_dpm_clks.mc_address.quad_part);
+
+	if (smu_dpm_clks.dpm_clks == NULL) {
+		smu_dpm_clks.dpm_clks = &dummy_clocks;
+		smu_dpm_clks.mc_address.quad_part = 0;
+	}
+
+	ASSERT(smu_dpm_clks.dpm_clks);
+
+	clk_mgr->base.smu_ver = dcn301_smu_get_smu_version(&clk_mgr->base);
+
+	if (clk_mgr->base.smu_ver)
+		clk_mgr->base.smu_present = true;
+
+	/* TODO: Check we get what we expect during bringup */
+	clk_mgr->base.base.dentist_vco_freq_khz = get_vco_frequency_from_reg(&clk_mgr->base);
+
+	/* in case we don't get a value from the register, use default */
+	if (clk_mgr->base.base.dentist_vco_freq_khz == 0)
+		clk_mgr->base.base.dentist_vco_freq_khz = 3600000;
+
+	if (ctx->dc_bios->integrated_info->memory_type == LpDdr5MemType) {
+		vg_bw_params.wm_table = lpddr5_wm_table;
+	} else {
+		vg_bw_params.wm_table = ddr4_wm_table;
+	}
+	/* Saved clocks configured at boot for debug purposes */
+	vg_dump_clk_registers(&clk_mgr->base.base.boot_snapshot, &clk_mgr->base.base, &log_info);
+
+	clk_mgr->base.base.dprefclk_khz = 600000;
+	dce_clock_read_ss_info(&clk_mgr->base);
+
+	clk_mgr->base.base.bw_params = &vg_bw_params;
+
+	vg_get_dpm_table_from_smu(&clk_mgr->base, &smu_dpm_clks);
+	if (ctx->dc_bios->integrated_info) {
+		vg_clk_mgr_helper_populate_bw_params(
+				&clk_mgr->base,
+				ctx->dc_bios->integrated_info,
+				smu_dpm_clks.dpm_clks);
+	}
+
+	if (smu_dpm_clks.dpm_clks && smu_dpm_clks.mc_address.quad_part != 0)
+		dm_helpers_free_gpu_mem(clk_mgr->base.base.ctx, DC_MEM_ALLOC_TYPE_FRAME_BUFFER,
+				smu_dpm_clks.dpm_clks);
+}
+
+void vg_clk_mgr_destroy(struct clk_mgr_internal *clk_mgr_int)
+{
+	struct clk_mgr_vgh *clk_mgr = TO_CLK_MGR_VGH(clk_mgr_int);
+
+	if (clk_mgr->smu_wm_set.wm_set && clk_mgr->smu_wm_set.mc_address.quad_part != 0)
+		dm_helpers_free_gpu_mem(clk_mgr_int->base.ctx, DC_MEM_ALLOC_TYPE_FRAME_BUFFER,
+				clk_mgr->smu_wm_set.wm_set);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn301/vg_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn301/vg_clk_mgr.h
new file mode 100644
index 000000000000..75884f572989
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn301/vg_clk_mgr.h
@@ -0,0 +1,52 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __VG_CLK_MGR_H__
+#define __VG_CLK_MGR_H__
+#include "clk_mgr_internal.h"
+
+struct watermarks;
+
+extern struct wm_table ddr4_wm_table;
+extern struct wm_table lpddr5_wm_table;
+
+struct smu_watermark_set {
+	struct watermarks *wm_set;
+	union large_integer mc_address;
+};
+
+struct clk_mgr_vgh {
+	struct clk_mgr_internal base;
+	struct smu_watermark_set smu_wm_set;
+};
+
+void vg_clk_mgr_construct(struct dc_context *ctx,
+		struct clk_mgr_vgh *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg);
+
+void vg_clk_mgr_destroy(struct clk_mgr_internal *clk_mgr);
+
+#endif //__VG_CLK_MGR_H__
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn31/dcn31_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn31/dcn31_clk_mgr.c
new file mode 100644
index 000000000000..051052bd10c9
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn31/dcn31_clk_mgr.c
@@ -0,0 +1,809 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+
+#include "dccg.h"
+#include "clk_mgr_internal.h"
+
+// For dce12_get_dp_ref_freq_khz
+#include "dce100/dce_clk_mgr.h"
+
+// For dcn20_update_clocks_update_dpp_dto
+#include "dcn20/dcn20_clk_mgr.h"
+
+
+
+#include "dcn31_clk_mgr.h"
+
+#include "reg_helper.h"
+#include "core_types.h"
+#include "dcn31_smu.h"
+#include "dm_helpers.h"
+
+/* TODO: remove this include once we ported over remaining clk mgr functions*/
+#include "dcn30/dcn30_clk_mgr.h"
+
+#include "dc_dmub_srv.h"
+#include "link_service.h"
+
+#include "logger_types.h"
+
+
+#include "yellow_carp_offset.h"
+#undef DC_LOGGER
+#define DC_LOGGER \
+	clk_mgr->base.base.ctx->logger
+
+#define regCLK1_CLK_PLL_REQ			0x0237
+#define regCLK1_CLK_PLL_REQ_BASE_IDX		0
+
+#define CLK1_CLK_PLL_REQ__FbMult_int__SHIFT	0x0
+#define CLK1_CLK_PLL_REQ__PllSpineDiv__SHIFT	0xc
+#define CLK1_CLK_PLL_REQ__FbMult_frac__SHIFT	0x10
+#define CLK1_CLK_PLL_REQ__FbMult_int_MASK	0x000001FFL
+#define CLK1_CLK_PLL_REQ__PllSpineDiv_MASK	0x0000F000L
+#define CLK1_CLK_PLL_REQ__FbMult_frac_MASK	0xFFFF0000L
+
+#define REG(reg_name) \
+	(CLK_BASE.instance[0].segment[reg ## reg_name ## _BASE_IDX] + reg ## reg_name)
+
+#define TO_CLK_MGR_DCN31(clk_mgr)\
+	container_of(clk_mgr, struct clk_mgr_dcn31, base)
+
+static int dcn31_get_active_display_cnt_wa(
+		struct dc *dc,
+		struct dc_state *context)
+{
+	int i, display_count;
+	bool tmds_present = false;
+
+	display_count = 0;
+	for (i = 0; i < context->stream_count; i++) {
+		const struct dc_stream_state *stream = context->streams[i];
+
+		if (stream->signal == SIGNAL_TYPE_HDMI_TYPE_A ||
+				stream->signal == SIGNAL_TYPE_DVI_SINGLE_LINK ||
+				stream->signal == SIGNAL_TYPE_DVI_DUAL_LINK)
+			tmds_present = true;
+
+		/* Checking stream / link detection ensuring that PHY is active*/
+		if (dc_is_dp_signal(stream->signal) && !stream->dpms_off)
+			display_count++;
+
+	}
+
+	for (i = 0; i < dc->link_count; i++) {
+		const struct dc_link *link = dc->links[i];
+
+		/* abusing the fact that the dig and phy are coupled to see if the phy is enabled */
+		if (link->link_enc && link->link_enc->funcs->is_dig_enabled &&
+				link->link_enc->funcs->is_dig_enabled(link->link_enc))
+			display_count++;
+	}
+
+	/* WA for hang on HDMI after display off back back on*/
+	if (display_count == 0 && tmds_present)
+		display_count = 1;
+
+	return display_count;
+}
+
+static void dcn31_disable_otg_wa(struct clk_mgr *clk_mgr_base, struct dc_state *context, bool disable)
+{
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	int i;
+
+	for (i = 0; i < dc->res_pool->pipe_count; ++i) {
+		struct pipe_ctx *pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+
+		if (pipe->top_pipe || pipe->prev_odm_pipe)
+			continue;
+		if (pipe->stream && (pipe->stream->dpms_off || dc_is_virtual_signal(pipe->stream->signal))) {
+			if (disable) {
+				pipe->stream_res.tg->funcs->immediate_disable_crtc(pipe->stream_res.tg);
+				reset_sync_context_for_pipe(dc, context, i);
+			} else
+				pipe->stream_res.tg->funcs->enable_crtc(pipe->stream_res.tg);
+		}
+	}
+}
+
+void dcn31_update_clocks(struct clk_mgr *clk_mgr_base,
+			struct dc_state *context,
+			bool safe_to_lower)
+{
+	union dmub_rb_cmd cmd;
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dc_clocks *new_clocks = &context->bw_ctx.bw.dcn.clk;
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	int display_count;
+	bool update_dppclk = false;
+	bool update_dispclk = false;
+	bool dpp_clock_lowered = false;
+
+	if (dc->work_arounds.skip_clock_update)
+		return;
+
+	/*
+	 * if it is safe to lower, but we are already in the lower state, we don't have to do anything
+	 * also if safe to lower is false, we just go in the higher state
+	 */
+	if (safe_to_lower) {
+		if (new_clocks->zstate_support != DCN_ZSTATE_SUPPORT_DISALLOW &&
+				new_clocks->zstate_support != clk_mgr_base->clks.zstate_support) {
+			dcn31_smu_set_zstate_support(clk_mgr, new_clocks->zstate_support);
+			dm_helpers_enable_periodic_detection(clk_mgr_base->ctx, true);
+			clk_mgr_base->clks.zstate_support = new_clocks->zstate_support;
+		}
+
+		if (clk_mgr_base->clks.dtbclk_en && !new_clocks->dtbclk_en) {
+			dcn31_smu_set_dtbclk(clk_mgr, false);
+			clk_mgr_base->clks.dtbclk_en = new_clocks->dtbclk_en;
+		}
+		/* check that we're not already in lower */
+		if (clk_mgr_base->clks.pwr_state != DCN_PWR_STATE_LOW_POWER) {
+			display_count = dcn31_get_active_display_cnt_wa(dc, context);
+			/* if we can go lower, go lower */
+			if (display_count == 0) {
+				union display_idle_optimization_u idle_info = { 0 };
+				idle_info.idle_info.df_request_disabled = 1;
+				idle_info.idle_info.phy_ref_clk_off = 1;
+				idle_info.idle_info.s0i2_rdy = 1;
+				dcn31_smu_set_display_idle_optimization(clk_mgr, idle_info.data);
+				/* update power state */
+				clk_mgr_base->clks.pwr_state = DCN_PWR_STATE_LOW_POWER;
+			}
+		}
+	} else {
+		if (new_clocks->zstate_support == DCN_ZSTATE_SUPPORT_DISALLOW &&
+				new_clocks->zstate_support != clk_mgr_base->clks.zstate_support) {
+			dcn31_smu_set_zstate_support(clk_mgr, DCN_ZSTATE_SUPPORT_DISALLOW);
+			dm_helpers_enable_periodic_detection(clk_mgr_base->ctx, false);
+			clk_mgr_base->clks.zstate_support = new_clocks->zstate_support;
+		}
+
+		if (!clk_mgr_base->clks.dtbclk_en && new_clocks->dtbclk_en) {
+			dcn31_smu_set_dtbclk(clk_mgr, true);
+			clk_mgr_base->clks.dtbclk_en = new_clocks->dtbclk_en;
+		}
+
+		/* check that we're not already in D0 */
+		if (clk_mgr_base->clks.pwr_state != DCN_PWR_STATE_MISSION_MODE) {
+			union display_idle_optimization_u idle_info = { 0 };
+			dcn31_smu_set_display_idle_optimization(clk_mgr, idle_info.data);
+			/* update power state */
+			clk_mgr_base->clks.pwr_state = DCN_PWR_STATE_MISSION_MODE;
+		}
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dcfclk_khz, clk_mgr_base->clks.dcfclk_khz)) {
+		clk_mgr_base->clks.dcfclk_khz = new_clocks->dcfclk_khz;
+		dcn31_smu_set_hard_min_dcfclk(clk_mgr, clk_mgr_base->clks.dcfclk_khz);
+	}
+
+	if (should_set_clock(safe_to_lower,
+			new_clocks->dcfclk_deep_sleep_khz, clk_mgr_base->clks.dcfclk_deep_sleep_khz)) {
+		clk_mgr_base->clks.dcfclk_deep_sleep_khz = new_clocks->dcfclk_deep_sleep_khz;
+		dcn31_smu_set_min_deep_sleep_dcfclk(clk_mgr, clk_mgr_base->clks.dcfclk_deep_sleep_khz);
+	}
+
+	// workaround: Limit dppclk to 100Mhz to avoid lower eDP panel switch to plus 4K monitor underflow.
+	if (new_clocks->dppclk_khz < 100000)
+		new_clocks->dppclk_khz = 100000;
+
+	if (should_set_clock(safe_to_lower, new_clocks->dppclk_khz, clk_mgr->base.clks.dppclk_khz)) {
+		if (clk_mgr->base.clks.dppclk_khz > new_clocks->dppclk_khz)
+			dpp_clock_lowered = true;
+		clk_mgr_base->clks.dppclk_khz = new_clocks->dppclk_khz;
+		update_dppclk = true;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, clk_mgr_base->clks.dispclk_khz)) {
+		dcn31_disable_otg_wa(clk_mgr_base, context, true);
+
+		clk_mgr_base->clks.dispclk_khz = new_clocks->dispclk_khz;
+		dcn31_smu_set_dispclk(clk_mgr, clk_mgr_base->clks.dispclk_khz);
+		dcn31_disable_otg_wa(clk_mgr_base, context, false);
+
+		update_dispclk = true;
+	}
+
+	if (dpp_clock_lowered) {
+		// increase per DPP DTO before lowering global dppclk
+		dcn20_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
+		dcn31_smu_set_dppclk(clk_mgr, clk_mgr_base->clks.dppclk_khz);
+	} else {
+		// increase global DPPCLK before lowering per DPP DTO
+		if (update_dppclk || update_dispclk)
+			dcn31_smu_set_dppclk(clk_mgr, clk_mgr_base->clks.dppclk_khz);
+		// always update dtos unless clock is lowered and not safe to lower
+		if (new_clocks->dppclk_khz >= dc->current_state->bw_ctx.bw.dcn.clk.dppclk_khz)
+			dcn20_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
+	}
+
+	// notify DMCUB of latest clocks
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.notify_clocks.header.type = DMUB_CMD__CLK_MGR;
+	cmd.notify_clocks.header.sub_type = DMUB_CMD__CLK_MGR_NOTIFY_CLOCKS;
+	cmd.notify_clocks.clocks.dcfclk_khz = clk_mgr_base->clks.dcfclk_khz;
+	cmd.notify_clocks.clocks.dcfclk_deep_sleep_khz =
+		clk_mgr_base->clks.dcfclk_deep_sleep_khz;
+	cmd.notify_clocks.clocks.dispclk_khz = clk_mgr_base->clks.dispclk_khz;
+	cmd.notify_clocks.clocks.dppclk_khz = clk_mgr_base->clks.dppclk_khz;
+
+	dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+static int get_vco_frequency_from_reg(struct clk_mgr_internal *clk_mgr)
+{
+	/* get FbMult value */
+	struct fixed31_32 pll_req;
+	unsigned int fbmult_frac_val = 0;
+	unsigned int fbmult_int_val = 0;
+
+	/*
+	 * Register value of fbmult is in 8.16 format, we are converting to 31.32
+	 * to leverage the fix point operations available in driver
+	 */
+
+	REG_GET(CLK1_CLK_PLL_REQ, FbMult_frac, &fbmult_frac_val); /* 16 bit fractional part*/
+	REG_GET(CLK1_CLK_PLL_REQ, FbMult_int, &fbmult_int_val); /* 8 bit integer part */
+
+	pll_req = dc_fixpt_from_int(fbmult_int_val);
+
+	/*
+	 * since fractional part is only 16 bit in register definition but is 32 bit
+	 * in our fix point definiton, need to shift left by 16 to obtain correct value
+	 */
+	pll_req.value |= fbmult_frac_val << 16;
+
+	/* multiply by REFCLK period */
+	pll_req = dc_fixpt_mul_int(pll_req, clk_mgr->dfs_ref_freq_khz);
+
+	/* integer part is now VCO frequency in kHz */
+	return dc_fixpt_floor(pll_req);
+}
+
+static void dcn31_enable_pme_wa(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	dcn31_smu_enable_pme_wa(clk_mgr);
+}
+
+void dcn31_init_clocks(struct clk_mgr *clk_mgr)
+{
+	uint32_t ref_dtbclk = clk_mgr->clks.ref_dtbclk_khz;
+
+	memset(&(clk_mgr->clks), 0, sizeof(struct dc_clocks));
+	// Assumption is that boot state always supports pstate
+	clk_mgr->clks.ref_dtbclk_khz = ref_dtbclk;	// restore ref_dtbclk
+	clk_mgr->clks.p_state_change_support = true;
+	clk_mgr->clks.prev_p_state_change_support = true;
+	clk_mgr->clks.pwr_state = DCN_PWR_STATE_UNKNOWN;
+	clk_mgr->clks.zstate_support = DCN_ZSTATE_SUPPORT_UNKNOWN;
+}
+
+bool dcn31_are_clock_states_equal(struct dc_clocks *a,
+		struct dc_clocks *b)
+{
+	if (a->dispclk_khz != b->dispclk_khz)
+		return false;
+	else if (a->dppclk_khz != b->dppclk_khz)
+		return false;
+	else if (a->dcfclk_khz != b->dcfclk_khz)
+		return false;
+	else if (a->dcfclk_deep_sleep_khz != b->dcfclk_deep_sleep_khz)
+		return false;
+	else if (a->zstate_support != b->zstate_support)
+		return false;
+	else if (a->dtbclk_en != b->dtbclk_en)
+		return false;
+
+	return true;
+}
+
+static void dcn31_dump_clk_registers(struct clk_state_registers_and_bypass *regs_and_bypass,
+		struct clk_mgr *clk_mgr_base, struct clk_log_info *log_info)
+{
+	return;
+}
+
+static struct clk_bw_params dcn31_bw_params = {
+	.vram_type = Ddr4MemType,
+	.num_channels = 1,
+	.clk_table = {
+		.num_entries = 4,
+	},
+
+};
+
+static struct wm_table ddr5_wm_table = {
+	.entries = {
+		{
+			.wm_inst = WM_A,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 9,
+			.sr_enter_plus_exit_time_us = 11,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_B,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 9,
+			.sr_enter_plus_exit_time_us = 11,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_C,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 9,
+			.sr_enter_plus_exit_time_us = 11,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_D,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 9,
+			.sr_enter_plus_exit_time_us = 11,
+			.valid = true,
+		},
+	}
+};
+
+static struct wm_table lpddr5_wm_table = {
+	.entries = {
+		{
+			.wm_inst = WM_A,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 11.5,
+			.sr_enter_plus_exit_time_us = 14.5,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_B,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 11.5,
+			.sr_enter_plus_exit_time_us = 14.5,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_C,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 11.5,
+			.sr_enter_plus_exit_time_us = 14.5,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_D,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 11.5,
+			.sr_enter_plus_exit_time_us = 14.5,
+			.valid = true,
+		},
+	}
+};
+
+static DpmClocks_t dummy_clocks;
+
+static struct dcn31_watermarks dummy_wms = { 0 };
+
+static void dcn31_build_watermark_ranges(struct clk_bw_params *bw_params, struct dcn31_watermarks *table)
+{
+	int i, num_valid_sets;
+
+	num_valid_sets = 0;
+
+	for (i = 0; i < WM_SET_COUNT; i++) {
+		/* skip empty entries, the smu array has no holes*/
+		if (!bw_params->wm_table.entries[i].valid)
+			continue;
+
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].WmSetting = bw_params->wm_table.entries[i].wm_inst;
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].WmType = bw_params->wm_table.entries[i].wm_type;
+		/* We will not select WM based on fclk, so leave it as unconstrained */
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinClock = 0;
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].MaxClock = 0xFFFF;
+
+		if (table->WatermarkRow[WM_DCFCLK][num_valid_sets].WmType == WM_TYPE_PSTATE_CHG) {
+			if (i == 0)
+				table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinMclk = 0;
+			else {
+				/* add 1 to make it non-overlapping with next lvl */
+				table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinMclk =
+						bw_params->clk_table.entries[i - 1].dcfclk_mhz + 1;
+			}
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets].MaxMclk =
+					bw_params->clk_table.entries[i].dcfclk_mhz;
+
+		} else {
+			/* unconstrained for memory retraining */
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinClock = 0;
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets].MaxClock = 0xFFFF;
+
+			/* Modify previous watermark range to cover up to max */
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets - 1].MaxClock = 0xFFFF;
+		}
+		num_valid_sets++;
+	}
+
+	ASSERT(num_valid_sets != 0); /* Must have at least one set of valid watermarks */
+
+	/* modify the min and max to make sure we cover the whole range*/
+	table->WatermarkRow[WM_DCFCLK][0].MinMclk = 0;
+	table->WatermarkRow[WM_DCFCLK][0].MinClock = 0;
+	table->WatermarkRow[WM_DCFCLK][num_valid_sets - 1].MaxMclk = 0xFFFF;
+	table->WatermarkRow[WM_DCFCLK][num_valid_sets - 1].MaxClock = 0xFFFF;
+
+	/* This is for writeback only, does not matter currently as no writeback support*/
+	table->WatermarkRow[WM_SOCCLK][0].WmSetting = WM_A;
+	table->WatermarkRow[WM_SOCCLK][0].MinClock = 0;
+	table->WatermarkRow[WM_SOCCLK][0].MaxClock = 0xFFFF;
+	table->WatermarkRow[WM_SOCCLK][0].MinMclk = 0;
+	table->WatermarkRow[WM_SOCCLK][0].MaxMclk = 0xFFFF;
+}
+
+static void dcn31_notify_wm_ranges(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct clk_mgr_dcn31 *clk_mgr_dcn31 = TO_CLK_MGR_DCN31(clk_mgr);
+	struct dcn31_watermarks *table = clk_mgr_dcn31->smu_wm_set.wm_set;
+
+	if (!clk_mgr->smu_ver)
+		return;
+
+	if (!table || clk_mgr_dcn31->smu_wm_set.mc_address.quad_part == 0)
+		return;
+
+	memset(table, 0, sizeof(*table));
+
+	dcn31_build_watermark_ranges(clk_mgr_base->bw_params, table);
+
+	dcn31_smu_set_dram_addr_high(clk_mgr,
+			clk_mgr_dcn31->smu_wm_set.mc_address.high_part);
+	dcn31_smu_set_dram_addr_low(clk_mgr,
+			clk_mgr_dcn31->smu_wm_set.mc_address.low_part);
+	dcn31_smu_transfer_wm_table_dram_2_smu(clk_mgr);
+}
+
+static void dcn31_get_dpm_table_from_smu(struct clk_mgr_internal *clk_mgr,
+		struct dcn31_smu_dpm_clks *smu_dpm_clks)
+{
+	DpmClocks_t *table = smu_dpm_clks->dpm_clks;
+
+	if (!clk_mgr->smu_ver)
+		return;
+
+	if (!table || smu_dpm_clks->mc_address.quad_part == 0)
+		return;
+
+	memset(table, 0, sizeof(*table));
+
+	dcn31_smu_set_dram_addr_high(clk_mgr,
+			smu_dpm_clks->mc_address.high_part);
+	dcn31_smu_set_dram_addr_low(clk_mgr,
+			smu_dpm_clks->mc_address.low_part);
+	dcn31_smu_transfer_dpm_table_smu_2_dram(clk_mgr);
+}
+
+static uint32_t find_max_clk_value(const uint32_t clocks[], uint32_t num_clocks)
+{
+	uint32_t max = 0;
+	int i;
+
+	for (i = 0; i < num_clocks; ++i) {
+		if (clocks[i] > max)
+			max = clocks[i];
+	}
+
+	return max;
+}
+
+static unsigned int find_clk_for_voltage(
+		const DpmClocks_t *clock_table,
+		const uint32_t clocks[],
+		unsigned int voltage)
+{
+	int i;
+	int max_voltage = 0;
+	int clock = 0;
+
+	for (i = 0; i < NUM_SOC_VOLTAGE_LEVELS; i++) {
+		if (clock_table->SocVoltage[i] == voltage) {
+			return clocks[i];
+		} else if (clock_table->SocVoltage[i] >= max_voltage &&
+				clock_table->SocVoltage[i] < voltage) {
+			max_voltage = clock_table->SocVoltage[i];
+			clock = clocks[i];
+		}
+	}
+
+	ASSERT(clock);
+	return clock;
+}
+
+static void dcn31_clk_mgr_helper_populate_bw_params(struct clk_mgr_internal *clk_mgr,
+						    struct integrated_info *bios_info,
+						    const DpmClocks_t *clock_table)
+{
+	int i, j;
+	struct clk_bw_params *bw_params = clk_mgr->base.bw_params;
+	uint32_t max_dispclk = 0, max_dppclk = 0;
+
+	j = -1;
+
+	static_assert(NUM_DF_PSTATE_LEVELS <= MAX_NUM_DPM_LVL,
+		"number of reported pstate levels exceeds maximum");
+
+	/* Find lowest DPM, FCLK is filled in reverse order*/
+
+	for (i = NUM_DF_PSTATE_LEVELS - 1; i >= 0; i--) {
+		if (clock_table->DfPstateTable[i].FClk != 0) {
+			j = i;
+			break;
+		}
+	}
+
+	if (j == -1) {
+		/* clock table is all 0s, just use our own hardcode */
+		ASSERT(0);
+		return;
+	}
+
+	bw_params->clk_table.num_entries = j + 1;
+
+	/* dispclk and dppclk can be max at any voltage, same number of levels for both */
+	if (clock_table->NumDispClkLevelsEnabled <= NUM_DISPCLK_DPM_LEVELS &&
+	    clock_table->NumDispClkLevelsEnabled <= NUM_DPPCLK_DPM_LEVELS) {
+		max_dispclk = find_max_clk_value(clock_table->DispClocks, clock_table->NumDispClkLevelsEnabled);
+		max_dppclk = find_max_clk_value(clock_table->DppClocks, clock_table->NumDispClkLevelsEnabled);
+	} else {
+		ASSERT(0);
+	}
+
+	for (i = 0; i < bw_params->clk_table.num_entries; i++, j--) {
+		bw_params->clk_table.entries[i].fclk_mhz = clock_table->DfPstateTable[j].FClk;
+		bw_params->clk_table.entries[i].memclk_mhz = clock_table->DfPstateTable[j].MemClk;
+		bw_params->clk_table.entries[i].voltage = clock_table->DfPstateTable[j].Voltage;
+		switch (clock_table->DfPstateTable[j].WckRatio) {
+		case WCK_RATIO_1_2:
+			bw_params->clk_table.entries[i].wck_ratio = 2;
+			break;
+		case WCK_RATIO_1_4:
+			bw_params->clk_table.entries[i].wck_ratio = 4;
+			break;
+		default:
+			bw_params->clk_table.entries[i].wck_ratio = 1;
+		}
+		bw_params->clk_table.entries[i].dcfclk_mhz = find_clk_for_voltage(clock_table, clock_table->DcfClocks, clock_table->DfPstateTable[j].Voltage);
+		bw_params->clk_table.entries[i].socclk_mhz = find_clk_for_voltage(clock_table, clock_table->SocClocks, clock_table->DfPstateTable[j].Voltage);
+		bw_params->clk_table.entries[i].dispclk_mhz = max_dispclk;
+		bw_params->clk_table.entries[i].dppclk_mhz = max_dppclk;
+	}
+
+	bw_params->vram_type = bios_info->memory_type;
+
+	bw_params->dram_channel_width_bytes = bios_info->memory_type == 0x22 ? 8 : 4;
+	//bw_params->dram_channel_width_bytes = dc->ctx->asic_id.vram_width;
+	bw_params->num_channels = bios_info->ma_channel_number ? bios_info->ma_channel_number : 4;
+	for (i = 0; i < WM_SET_COUNT; i++) {
+		bw_params->wm_table.entries[i].wm_inst = i;
+
+		if (i >= bw_params->clk_table.num_entries) {
+			bw_params->wm_table.entries[i].valid = false;
+			continue;
+		}
+
+		bw_params->wm_table.entries[i].wm_type = WM_TYPE_PSTATE_CHG;
+		bw_params->wm_table.entries[i].valid = true;
+	}
+}
+
+static void dcn31_set_low_power_state(struct clk_mgr *clk_mgr_base)
+{
+	int display_count;
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	struct dc_state *context = dc->current_state;
+
+	if (clk_mgr_base->clks.pwr_state != DCN_PWR_STATE_LOW_POWER) {
+		display_count = dcn31_get_active_display_cnt_wa(dc, context);
+		/* if we can go lower, go lower */
+		if (display_count == 0) {
+			union display_idle_optimization_u idle_info = { 0 };
+
+			idle_info.idle_info.df_request_disabled = 1;
+			idle_info.idle_info.phy_ref_clk_off = 1;
+			idle_info.idle_info.s0i2_rdy = 1;
+			dcn31_smu_set_display_idle_optimization(clk_mgr, idle_info.data);
+			/* update power state */
+			clk_mgr_base->clks.pwr_state = DCN_PWR_STATE_LOW_POWER;
+		}
+	}
+}
+
+int dcn31_get_dtb_ref_freq_khz(struct clk_mgr *clk_mgr_base)
+{
+	return clk_mgr_base->clks.ref_dtbclk_khz;
+}
+
+static struct clk_mgr_funcs dcn31_funcs = {
+	.get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz,
+	.get_dtb_ref_clk_frequency = dcn31_get_dtb_ref_freq_khz,
+	.update_clocks = dcn31_update_clocks,
+	.init_clocks = dcn31_init_clocks,
+	.enable_pme_wa = dcn31_enable_pme_wa,
+	.are_clock_states_equal = dcn31_are_clock_states_equal,
+	.notify_wm_ranges = dcn31_notify_wm_ranges,
+	.set_low_power_state = dcn31_set_low_power_state
+};
+extern struct clk_mgr_funcs dcn3_fpga_funcs;
+
+void dcn31_clk_mgr_construct(
+		struct dc_context *ctx,
+		struct clk_mgr_dcn31 *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg)
+{
+	struct dcn31_smu_dpm_clks smu_dpm_clks = { 0 };
+	struct clk_log_info log_info = {0};
+
+	clk_mgr->base.base.ctx = ctx;
+	clk_mgr->base.base.funcs = &dcn31_funcs;
+
+	clk_mgr->base.pp_smu = pp_smu;
+
+	clk_mgr->base.dccg = dccg;
+	clk_mgr->base.dfs_bypass_disp_clk = 0;
+
+	clk_mgr->base.dprefclk_ss_percentage = 0;
+	clk_mgr->base.dprefclk_ss_divider = 1000;
+	clk_mgr->base.ss_on_dprefclk = false;
+	clk_mgr->base.dfs_ref_freq_khz = 48000;
+
+	clk_mgr->smu_wm_set.wm_set = (struct dcn31_watermarks *)dm_helpers_allocate_gpu_mem(
+				clk_mgr->base.base.ctx,
+				DC_MEM_ALLOC_TYPE_FRAME_BUFFER,
+				sizeof(struct dcn31_watermarks),
+				&clk_mgr->smu_wm_set.mc_address.quad_part);
+
+	if (!clk_mgr->smu_wm_set.wm_set) {
+		clk_mgr->smu_wm_set.wm_set = &dummy_wms;
+		clk_mgr->smu_wm_set.mc_address.quad_part = 0;
+	}
+	ASSERT(clk_mgr->smu_wm_set.wm_set);
+
+	smu_dpm_clks.dpm_clks = (DpmClocks_t *)dm_helpers_allocate_gpu_mem(
+				clk_mgr->base.base.ctx,
+				DC_MEM_ALLOC_TYPE_FRAME_BUFFER,
+				sizeof(DpmClocks_t),
+				&smu_dpm_clks.mc_address.quad_part);
+
+	if (smu_dpm_clks.dpm_clks == NULL) {
+		smu_dpm_clks.dpm_clks = &dummy_clocks;
+		smu_dpm_clks.mc_address.quad_part = 0;
+	}
+
+	ASSERT(smu_dpm_clks.dpm_clks);
+
+	clk_mgr->base.smu_ver = dcn31_smu_get_smu_version(&clk_mgr->base);
+
+	if (clk_mgr->base.smu_ver)
+		clk_mgr->base.smu_present = true;
+
+	/* TODO: Check we get what we expect during bringup */
+	clk_mgr->base.base.dentist_vco_freq_khz = get_vco_frequency_from_reg(&clk_mgr->base);
+
+	if (ctx->dc_bios->integrated_info->memory_type == LpDdr5MemType) {
+		dcn31_bw_params.wm_table = lpddr5_wm_table;
+	} else {
+		dcn31_bw_params.wm_table = ddr5_wm_table;
+	}
+	/* Saved clocks configured at boot for debug purposes */
+	dcn31_dump_clk_registers(&clk_mgr->base.base.boot_snapshot,
+				 &clk_mgr->base.base, &log_info);
+
+	clk_mgr->base.base.dprefclk_khz = 600000;
+	clk_mgr->base.base.clks.ref_dtbclk_khz = 600000;
+	dce_clock_read_ss_info(&clk_mgr->base);
+	/*if bios enabled SS, driver needs to adjust dtb clock, only enable with correct bios*/
+	//clk_mgr->base.dccg->ref_dtbclk_khz = dce_adjust_dp_ref_freq_for_ss(clk_mgr_internal, clk_mgr->base.base.dprefclk_khz);
+
+	clk_mgr->base.base.bw_params = &dcn31_bw_params;
+
+	if (clk_mgr->base.base.ctx->dc->debug.pstate_enabled) {
+		int i;
+
+		dcn31_get_dpm_table_from_smu(&clk_mgr->base, &smu_dpm_clks);
+
+		DC_LOG_SMU("NumDcfClkLevelsEnabled: %d\n"
+				   "NumDispClkLevelsEnabled: %d\n"
+				   "NumSocClkLevelsEnabled: %d\n"
+				   "VcnClkLevelsEnabled: %d\n"
+				   "NumDfPst atesEnabled: %d\n"
+				   "MinGfxClk: %d\n"
+				   "MaxGfxClk: %d\n",
+				   smu_dpm_clks.dpm_clks->NumDcfClkLevelsEnabled,
+				   smu_dpm_clks.dpm_clks->NumDispClkLevelsEnabled,
+				   smu_dpm_clks.dpm_clks->NumSocClkLevelsEnabled,
+				   smu_dpm_clks.dpm_clks->VcnClkLevelsEnabled,
+				   smu_dpm_clks.dpm_clks->NumDfPstatesEnabled,
+				   smu_dpm_clks.dpm_clks->MinGfxClk,
+				   smu_dpm_clks.dpm_clks->MaxGfxClk);
+		for (i = 0; i < smu_dpm_clks.dpm_clks->NumDcfClkLevelsEnabled; i++) {
+			DC_LOG_SMU("smu_dpm_clks.dpm_clks->DcfClocks[%d] = %d\n",
+					   i,
+					   smu_dpm_clks.dpm_clks->DcfClocks[i]);
+		}
+		for (i = 0; i < smu_dpm_clks.dpm_clks->NumDispClkLevelsEnabled; i++) {
+			DC_LOG_SMU("smu_dpm_clks.dpm_clks->DispClocks[%d] = %d\n",
+					   i, smu_dpm_clks.dpm_clks->DispClocks[i]);
+		}
+		for (i = 0; i < smu_dpm_clks.dpm_clks->NumSocClkLevelsEnabled; i++) {
+			DC_LOG_SMU("smu_dpm_clks.dpm_clks->SocClocks[%d] = %d\n",
+					   i, smu_dpm_clks.dpm_clks->SocClocks[i]);
+		}
+		for (i = 0; i < NUM_SOC_VOLTAGE_LEVELS; i++)
+			DC_LOG_SMU("smu_dpm_clks.dpm_clks->SocVoltage[%d] = %d\n",
+					   i, smu_dpm_clks.dpm_clks->SocVoltage[i]);
+
+		for (i = 0; i < NUM_DF_PSTATE_LEVELS; i++) {
+			DC_LOG_SMU("smu_dpm_clks.dpm_clks.DfPstateTable[%d].FClk = %d\n"
+					   "smu_dpm_clks.dpm_clks->DfPstateTable[%d].MemClk= %d\n"
+					   "smu_dpm_clks.dpm_clks->DfPstateTable[%d].Voltage = %d\n",
+					   i, smu_dpm_clks.dpm_clks->DfPstateTable[i].FClk,
+					   i, smu_dpm_clks.dpm_clks->DfPstateTable[i].MemClk,
+					   i, smu_dpm_clks.dpm_clks->DfPstateTable[i].Voltage);
+		}
+		if (ctx->dc_bios->integrated_info) {
+			dcn31_clk_mgr_helper_populate_bw_params(
+					&clk_mgr->base,
+					ctx->dc_bios->integrated_info,
+					smu_dpm_clks.dpm_clks);
+		}
+	}
+
+	if (smu_dpm_clks.dpm_clks && smu_dpm_clks.mc_address.quad_part != 0)
+		dm_helpers_free_gpu_mem(clk_mgr->base.base.ctx, DC_MEM_ALLOC_TYPE_FRAME_BUFFER,
+				smu_dpm_clks.dpm_clks);
+}
+
+void dcn31_clk_mgr_destroy(struct clk_mgr_internal *clk_mgr_int)
+{
+	struct clk_mgr_dcn31 *clk_mgr = TO_CLK_MGR_DCN31(clk_mgr_int);
+
+	if (clk_mgr->smu_wm_set.wm_set && clk_mgr->smu_wm_set.mc_address.quad_part != 0)
+		dm_helpers_free_gpu_mem(clk_mgr_int->base.ctx, DC_MEM_ALLOC_TYPE_FRAME_BUFFER,
+				clk_mgr->smu_wm_set.wm_set);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn31/dcn31_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn31/dcn31_clk_mgr.h
new file mode 100644
index 000000000000..be06fdbd0c22
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn31/dcn31_clk_mgr.h
@@ -0,0 +1,58 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCN31_CLK_MGR_H__
+#define __DCN31_CLK_MGR_H__
+#include "clk_mgr_internal.h"
+
+struct dcn31_watermarks;
+
+struct dcn31_smu_watermark_set {
+	struct dcn31_watermarks *wm_set;
+	union large_integer mc_address;
+};
+
+struct clk_mgr_dcn31 {
+	struct clk_mgr_internal base;
+	struct dcn31_smu_watermark_set smu_wm_set;
+};
+
+bool dcn31_are_clock_states_equal(struct dc_clocks *a,
+		struct dc_clocks *b);
+void dcn31_init_clocks(struct clk_mgr *clk_mgr);
+void dcn31_update_clocks(struct clk_mgr *clk_mgr_base,
+			struct dc_state *context,
+			bool safe_to_lower);
+
+void dcn31_clk_mgr_construct(struct dc_context *ctx,
+		struct clk_mgr_dcn31 *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg);
+
+int dcn31_get_dtb_ref_freq_khz(struct clk_mgr *clk_mgr_base);
+
+void dcn31_clk_mgr_destroy(struct clk_mgr_internal *clk_mgr_int);
+
+#endif //__DCN31_CLK_MGR_H__
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn31/dcn31_smu.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn31/dcn31_smu.c
new file mode 100644
index 000000000000..f201628e4e98
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn31/dcn31_smu.c
@@ -0,0 +1,359 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "core_types.h"
+#include "clk_mgr_internal.h"
+#include "reg_helper.h"
+#include "dm_helpers.h"
+#include "dcn31_smu.h"
+
+#include "yellow_carp_offset.h"
+#include "mp/mp_13_0_2_offset.h"
+#include "mp/mp_13_0_2_sh_mask.h"
+
+#define REG(reg_name) \
+	(MP0_BASE.instance[0].segment[reg ## reg_name ## _BASE_IDX] + reg ## reg_name)
+
+#define FN(reg_name, field) \
+	FD(reg_name##__##field)
+
+#include "logger_types.h"
+#undef DC_LOGGER
+#define DC_LOGGER \
+	CTX->logger
+#define smu_print(str, ...) {DC_LOG_SMU(str, ##__VA_ARGS__); }
+
+#define VBIOSSMC_MSG_TestMessage                  0x1
+#define VBIOSSMC_MSG_GetSmuVersion                0x2
+#define VBIOSSMC_MSG_PowerUpGfx                   0x3
+#define VBIOSSMC_MSG_SetDispclkFreq               0x4
+#define VBIOSSMC_MSG_SetDprefclkFreq              0x5   //Not used. DPRef is constant
+#define VBIOSSMC_MSG_SetDppclkFreq                0x6
+#define VBIOSSMC_MSG_SetHardMinDcfclkByFreq       0x7
+#define VBIOSSMC_MSG_SetMinDeepSleepDcfclk        0x8
+#define VBIOSSMC_MSG_SetPhyclkVoltageByFreq       0x9	//Keep it in case VMIN dees not support phy clk
+#define VBIOSSMC_MSG_GetFclkFrequency             0xA
+#define VBIOSSMC_MSG_SetDisplayCount              0xB   //Not used anymore
+#define VBIOSSMC_MSG_EnableTmdp48MHzRefclkPwrDown 0xC   //Not used anymore
+#define VBIOSSMC_MSG_UpdatePmeRestore             0xD
+#define VBIOSSMC_MSG_SetVbiosDramAddrHigh         0xE   //Used for WM table txfr
+#define VBIOSSMC_MSG_SetVbiosDramAddrLow          0xF
+#define VBIOSSMC_MSG_TransferTableSmu2Dram        0x10
+#define VBIOSSMC_MSG_TransferTableDram2Smu        0x11
+#define VBIOSSMC_MSG_SetDisplayIdleOptimizations  0x12
+#define VBIOSSMC_MSG_GetDprefclkFreq              0x13
+#define VBIOSSMC_MSG_GetDtbclkFreq                0x14
+#define VBIOSSMC_MSG_AllowZstatesEntry            0x15
+#define VBIOSSMC_MSG_DisallowZstatesEntry     	  0x16
+#define VBIOSSMC_MSG_SetDtbClk                    0x17
+#define VBIOSSMC_Message_Count                    0x18
+
+#define VBIOSSMC_Status_BUSY                      0x0
+#define VBIOSSMC_Result_OK                        0x1
+#define VBIOSSMC_Result_Failed                    0xFF
+#define VBIOSSMC_Result_UnknownCmd                0xFE
+#define VBIOSSMC_Result_CmdRejectedPrereq         0xFD
+#define VBIOSSMC_Result_CmdRejectedBusy           0xFC
+
+/*
+ * Function to be used instead of REG_WAIT macro because the wait ends when
+ * the register is NOT EQUAL to zero, and because the translation in msg_if.h
+ * won't work with REG_WAIT.
+ */
+static uint32_t dcn31_smu_wait_for_response(struct clk_mgr_internal *clk_mgr, unsigned int delay_us, unsigned int max_retries)
+{
+	uint32_t res_val = VBIOSSMC_Status_BUSY;
+
+	do {
+		res_val = REG_READ(MP1_SMN_C2PMSG_91);
+		if (res_val != VBIOSSMC_Status_BUSY)
+			break;
+
+		if (delay_us >= 1000)
+			msleep(delay_us/1000);
+		else if (delay_us > 0)
+			udelay(delay_us);
+	} while (max_retries--);
+
+	return res_val;
+}
+
+static int dcn31_smu_send_msg_with_param(struct clk_mgr_internal *clk_mgr,
+					 unsigned int msg_id,
+					 unsigned int param)
+{
+	uint32_t result;
+
+	result = dcn31_smu_wait_for_response(clk_mgr, 10, 200000);
+
+	if (result != VBIOSSMC_Result_OK)
+		smu_print("SMU Response was not OK. SMU response after wait received is: %d\n", result);
+
+	if (result == VBIOSSMC_Status_BUSY) {
+		return -1;
+	}
+
+	/* First clear response register */
+	REG_WRITE(MP1_SMN_C2PMSG_91, VBIOSSMC_Status_BUSY);
+
+	/* Set the parameter register for the SMU message, unit is Mhz */
+	REG_WRITE(MP1_SMN_C2PMSG_83, param);
+
+	/* Trigger the message transaction by writing the message ID */
+	REG_WRITE(MP1_SMN_C2PMSG_67, msg_id);
+
+	result = dcn31_smu_wait_for_response(clk_mgr, 10, 200000);
+
+	if (result == VBIOSSMC_Result_Failed) {
+		if (msg_id == VBIOSSMC_MSG_TransferTableDram2Smu &&
+		    param == TABLE_WATERMARKS)
+			DC_LOG_DEBUG("Watermarks table not configured properly by SMU");
+		else
+			ASSERT(0);
+		REG_WRITE(MP1_SMN_C2PMSG_91, VBIOSSMC_Result_OK);
+		return -1;
+	}
+
+	if (IS_SMU_TIMEOUT(result)) {
+		ASSERT(0);
+		dm_helpers_smu_timeout(CTX, msg_id, param, 10 * 200000);
+	}
+
+	return REG_READ(MP1_SMN_C2PMSG_83);
+}
+
+int dcn31_smu_get_smu_version(struct clk_mgr_internal *clk_mgr)
+{
+	return dcn31_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_GetSmuVersion,
+			0);
+}
+
+
+int dcn31_smu_set_dispclk(struct clk_mgr_internal *clk_mgr, int requested_dispclk_khz)
+{
+	int actual_dispclk_set_mhz = -1;
+
+	if (!clk_mgr->smu_present)
+		return requested_dispclk_khz;
+
+	/*  Unit of SMU msg parameter is Mhz */
+	actual_dispclk_set_mhz = dcn31_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDispclkFreq,
+			khz_to_mhz_ceil(requested_dispclk_khz));
+
+	return actual_dispclk_set_mhz * 1000;
+}
+
+int dcn31_smu_set_dprefclk(struct clk_mgr_internal *clk_mgr)
+{
+	int actual_dprefclk_set_mhz = -1;
+
+	if (!clk_mgr->smu_present)
+		return clk_mgr->base.dprefclk_khz;
+
+	actual_dprefclk_set_mhz = dcn31_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDprefclkFreq,
+			khz_to_mhz_ceil(clk_mgr->base.dprefclk_khz));
+
+	/* TODO: add code for programing DP DTO, currently this is down by command table */
+
+	return actual_dprefclk_set_mhz * 1000;
+}
+
+int dcn31_smu_set_hard_min_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_dcfclk_khz)
+{
+	int actual_dcfclk_set_mhz = -1;
+
+	if (!clk_mgr->base.ctx->dc->debug.pstate_enabled)
+		return -1;
+
+	if (!clk_mgr->smu_present)
+		return requested_dcfclk_khz;
+
+	actual_dcfclk_set_mhz = dcn31_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetHardMinDcfclkByFreq,
+			khz_to_mhz_ceil(requested_dcfclk_khz));
+
+	return actual_dcfclk_set_mhz * 1000;
+}
+
+int dcn31_smu_set_min_deep_sleep_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_min_ds_dcfclk_khz)
+{
+	int actual_min_ds_dcfclk_mhz = -1;
+
+	if (!clk_mgr->base.ctx->dc->debug.pstate_enabled)
+		return -1;
+
+	if (!clk_mgr->smu_present)
+		return requested_min_ds_dcfclk_khz;
+
+	actual_min_ds_dcfclk_mhz = dcn31_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetMinDeepSleepDcfclk,
+			khz_to_mhz_ceil(requested_min_ds_dcfclk_khz));
+
+	return actual_min_ds_dcfclk_mhz * 1000;
+}
+
+int dcn31_smu_set_dppclk(struct clk_mgr_internal *clk_mgr, int requested_dpp_khz)
+{
+	int actual_dppclk_set_mhz = -1;
+
+	if (!clk_mgr->smu_present)
+		return requested_dpp_khz;
+
+	actual_dppclk_set_mhz = dcn31_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDppclkFreq,
+			khz_to_mhz_ceil(requested_dpp_khz));
+
+	return actual_dppclk_set_mhz * 1000;
+}
+
+void dcn31_smu_set_display_idle_optimization(struct clk_mgr_internal *clk_mgr, uint32_t idle_info)
+{
+	if (!clk_mgr->base.ctx->dc->debug.pstate_enabled)
+		return;
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	//TODO: Work with smu team to define optimization options.
+	dcn31_smu_send_msg_with_param(
+		clk_mgr,
+		VBIOSSMC_MSG_SetDisplayIdleOptimizations,
+		idle_info);
+}
+
+void dcn31_smu_enable_phy_refclk_pwrdwn(struct clk_mgr_internal *clk_mgr, bool enable)
+{
+	union display_idle_optimization_u idle_info = { 0 };
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	if (enable) {
+		idle_info.idle_info.df_request_disabled = 1;
+		idle_info.idle_info.phy_ref_clk_off = 1;
+	}
+
+	dcn31_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDisplayIdleOptimizations,
+			idle_info.data);
+}
+
+void dcn31_smu_enable_pme_wa(struct clk_mgr_internal *clk_mgr)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn31_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_UpdatePmeRestore,
+			0);
+}
+
+void dcn31_smu_set_dram_addr_high(struct clk_mgr_internal *clk_mgr, uint32_t addr_high)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn31_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_SetVbiosDramAddrHigh, addr_high);
+}
+
+void dcn31_smu_set_dram_addr_low(struct clk_mgr_internal *clk_mgr, uint32_t addr_low)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn31_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_SetVbiosDramAddrLow, addr_low);
+}
+
+void dcn31_smu_transfer_dpm_table_smu_2_dram(struct clk_mgr_internal *clk_mgr)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn31_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_TransferTableSmu2Dram, TABLE_DPMCLOCKS);
+}
+
+void dcn31_smu_transfer_wm_table_dram_2_smu(struct clk_mgr_internal *clk_mgr)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn31_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_TransferTableDram2Smu, TABLE_WATERMARKS);
+}
+
+void dcn31_smu_set_zstate_support(struct clk_mgr_internal *clk_mgr, enum dcn_zstate_support_state support)
+{
+	unsigned int msg_id, param;
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	if (!clk_mgr->base.ctx->dc->debug.enable_z9_disable_interface &&
+			(support == DCN_ZSTATE_SUPPORT_ALLOW_Z10_ONLY))
+		support = DCN_ZSTATE_SUPPORT_DISALLOW;
+
+	if (support == DCN_ZSTATE_SUPPORT_ALLOW_Z10_ONLY ||
+	    support == DCN_ZSTATE_SUPPORT_ALLOW_Z8_Z10_ONLY)
+		param = 1;
+	else
+		param = 0;
+
+	if (support == DCN_ZSTATE_SUPPORT_DISALLOW)
+		msg_id = VBIOSSMC_MSG_DisallowZstatesEntry;
+	else
+		msg_id = VBIOSSMC_MSG_AllowZstatesEntry;
+
+	dcn31_smu_send_msg_with_param(
+		clk_mgr,
+		msg_id,
+		param);
+
+}
+
+/* Arg = 1: Turn DTB on; 0: Turn DTB CLK OFF. when it is on, it is 600MHZ */
+void dcn31_smu_set_dtbclk(struct clk_mgr_internal *clk_mgr, bool enable)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn31_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDtbClk,
+			enable);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn31/dcn31_smu.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn31/dcn31_smu.h
new file mode 100644
index 000000000000..dfa25a76a6d1
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn31/dcn31_smu.h
@@ -0,0 +1,271 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DAL_DC_31_SMU_H_
+#define DAL_DC_31_SMU_H_
+
+#ifndef PMFW_DRIVER_IF_H
+#define PMFW_DRIVER_IF_H
+#define PMFW_DRIVER_IF_VERSION 4
+
+typedef struct {
+  int32_t value;
+  uint32_t numFractionalBits;
+} FloatInIntFormat_t;
+
+typedef enum {
+  DSPCLK_DCFCLK = 0,
+  DSPCLK_DISPCLK,
+  DSPCLK_PIXCLK,
+  DSPCLK_PHYCLK,
+  DSPCLK_COUNT,
+} DSPCLK_e;
+
+typedef struct {
+  uint16_t Freq; // in MHz
+  uint16_t Vid;  // min voltage in SVI3 VID
+} DisplayClockTable_t;
+
+typedef struct {
+  uint16_t MinClock; // This is either DCFCLK or SOCCLK (in MHz)
+  uint16_t MaxClock; // This is either DCFCLK or SOCCLK (in MHz)
+  uint16_t MinMclk;
+  uint16_t MaxMclk;
+
+  uint8_t  WmSetting;
+  uint8_t  WmType;  // Used for normal pstate change or memory retraining
+  uint8_t  Padding[2];
+} WatermarkRowGeneric_t;
+
+#define NUM_WM_RANGES 4
+#define WM_PSTATE_CHG 0
+#define WM_RETRAINING 1
+
+typedef enum {
+  WM_SOCCLK = 0,
+  WM_DCFCLK,
+  WM_COUNT,
+} WM_CLOCK_e;
+
+typedef struct {
+  // Watermarks
+  WatermarkRowGeneric_t WatermarkRow[WM_COUNT][NUM_WM_RANGES];
+
+  uint32_t MmHubPadding[7]; // SMU internal use
+} Watermarks_t;
+
+typedef enum {
+  CUSTOM_DPM_SETTING_GFXCLK,
+  CUSTOM_DPM_SETTING_CCLK,
+  CUSTOM_DPM_SETTING_FCLK_CCX,
+  CUSTOM_DPM_SETTING_FCLK_GFX,
+  CUSTOM_DPM_SETTING_FCLK_STALLS,
+  CUSTOM_DPM_SETTING_LCLK,
+  CUSTOM_DPM_SETTING_COUNT,
+} CUSTOM_DPM_SETTING_e;
+
+typedef struct {
+  uint8_t             ActiveHystLimit;
+  uint8_t             IdleHystLimit;
+  uint8_t             FPS;
+  uint8_t             MinActiveFreqType;
+  FloatInIntFormat_t  MinActiveFreq;
+  FloatInIntFormat_t  PD_Data_limit;
+  FloatInIntFormat_t  PD_Data_time_constant;
+  FloatInIntFormat_t  PD_Data_error_coeff;
+  FloatInIntFormat_t  PD_Data_error_rate_coeff;
+} DpmActivityMonitorCoeffExt_t;
+
+typedef struct {
+  DpmActivityMonitorCoeffExt_t DpmActivityMonitorCoeff[CUSTOM_DPM_SETTING_COUNT];
+} CustomDpmSettings_t;
+
+#define NUM_DCFCLK_DPM_LEVELS   8
+#define NUM_DISPCLK_DPM_LEVELS  8
+#define NUM_DPPCLK_DPM_LEVELS   8
+#define NUM_SOCCLK_DPM_LEVELS   8
+#define NUM_VCN_DPM_LEVELS      8
+#define NUM_SOC_VOLTAGE_LEVELS  8
+#define NUM_DF_PSTATE_LEVELS    4
+
+typedef enum{
+  WCK_RATIO_1_1 = 0,  // DDR5, Wck:ck is always 1:1;
+  WCK_RATIO_1_2,
+  WCK_RATIO_1_4,
+  WCK_RATIO_MAX
+} WCK_RATIO_e;
+
+typedef struct {
+  uint32_t FClk;
+  uint32_t MemClk;
+  uint32_t Voltage;
+  uint8_t  WckRatio;
+  uint8_t  Spare[3];
+} DfPstateTable_t;
+
+//Freq in MHz
+//Voltage in milli volts with 2 fractional bits
+typedef struct {
+  uint32_t DcfClocks[NUM_DCFCLK_DPM_LEVELS];
+  uint32_t DispClocks[NUM_DISPCLK_DPM_LEVELS];
+  uint32_t DppClocks[NUM_DPPCLK_DPM_LEVELS];
+  uint32_t SocClocks[NUM_SOCCLK_DPM_LEVELS];
+  uint32_t VClocks[NUM_VCN_DPM_LEVELS];
+  uint32_t DClocks[NUM_VCN_DPM_LEVELS];
+  uint32_t SocVoltage[NUM_SOC_VOLTAGE_LEVELS];
+  DfPstateTable_t DfPstateTable[NUM_DF_PSTATE_LEVELS];
+
+  uint8_t  NumDcfClkLevelsEnabled;
+  uint8_t  NumDispClkLevelsEnabled; //Applies to both Dispclk and Dppclk
+  uint8_t  NumSocClkLevelsEnabled;
+  uint8_t  VcnClkLevelsEnabled;     //Applies to both Vclk and Dclk
+  uint8_t  NumDfPstatesEnabled;
+  uint8_t  spare[3];
+
+  uint32_t MinGfxClk;
+  uint32_t MaxGfxClk;
+} DpmClocks_t;
+
+
+// Throttler Status Bitmask
+#define THROTTLER_STATUS_BIT_SPL            0
+#define THROTTLER_STATUS_BIT_FPPT           1
+#define THROTTLER_STATUS_BIT_SPPT           2
+#define THROTTLER_STATUS_BIT_SPPT_APU       3
+#define THROTTLER_STATUS_BIT_THM_CORE       4
+#define THROTTLER_STATUS_BIT_THM_GFX        5
+#define THROTTLER_STATUS_BIT_THM_SOC        6
+#define THROTTLER_STATUS_BIT_TDC_VDD        7
+#define THROTTLER_STATUS_BIT_TDC_SOC        8
+#define THROTTLER_STATUS_BIT_PROCHOT_CPU    9
+#define THROTTLER_STATUS_BIT_PROCHOT_GFX   10
+#define THROTTLER_STATUS_BIT_EDC_CPU       11
+#define THROTTLER_STATUS_BIT_EDC_GFX       12
+
+typedef struct {
+  uint16_t GfxclkFrequency;             //[MHz]
+  uint16_t SocclkFrequency;             //[MHz]
+  uint16_t VclkFrequency;               //[MHz]
+  uint16_t DclkFrequency;               //[MHz]
+  uint16_t MemclkFrequency;             //[MHz]
+  uint16_t spare;
+
+  uint16_t GfxActivity;                 //[centi]
+  uint16_t UvdActivity;                 //[centi]
+
+  uint16_t Voltage[2];                  //[mV] indices: VDDCR_VDD, VDDCR_SOC
+  uint16_t Current[2];                  //[mA] indices: VDDCR_VDD, VDDCR_SOC
+  uint16_t Power[2];                    //[mW] indices: VDDCR_VDD, VDDCR_SOC
+
+  //3rd party tools in Windows need this info in the case of APUs
+  uint16_t CoreFrequency[8];            //[MHz]
+  uint16_t CorePower[8];                //[mW]
+  uint16_t CoreTemperature[8];          //[centi-Celsius]
+  uint16_t L3Frequency;                 //[MHz]
+  uint16_t L3Temperature;               //[centi-Celsius]
+
+  uint16_t GfxTemperature;              //[centi-Celsius]
+  uint16_t SocTemperature;              //[centi-Celsius]
+  uint16_t ThrottlerStatus;
+
+  uint16_t CurrentSocketPower;          //[mW]
+  uint16_t StapmOriginalLimit;          //[W]
+  uint16_t StapmCurrentLimit;           //[W]
+  uint16_t ApuPower;                    //[W]
+  uint16_t dGpuPower;                   //[W]
+
+  uint16_t VddTdcValue;                 //[mA]
+  uint16_t SocTdcValue;                 //[mA]
+  uint16_t VddEdcValue;                 //[mA]
+  uint16_t SocEdcValue;                 //[mA]
+
+  uint16_t InfrastructureCpuMaxFreq;    //[MHz]
+  uint16_t InfrastructureGfxMaxFreq;    //[MHz]
+} SmuMetrics_t;
+
+
+// Workload bits
+#define WORKLOAD_PPLIB_FULL_SCREEN_3D_BIT 0
+#define WORKLOAD_PPLIB_VIDEO_BIT          2
+#define WORKLOAD_PPLIB_VR_BIT             3
+#define WORKLOAD_PPLIB_COMPUTE_BIT        4
+#define WORKLOAD_PPLIB_CUSTOM_BIT         5
+#define WORKLOAD_PPLIB_COUNT              6
+
+#define TABLE_BIOS_IF            0 // Called by BIOS
+#define TABLE_WATERMARKS         1 // Called by DAL through VBIOS
+#define TABLE_CUSTOM_DPM         2 // Called by Driver
+#define TABLE_SPARE1             3
+#define TABLE_DPMCLOCKS          4 // Called by Driver
+#define TABLE_MOMENTARY_PM       5 // Called by Tools
+#define TABLE_MODERN_STDBY       6 // Called by Tools for Modern Standby Log
+#define TABLE_SMU_METRICS        7 // Called by Driver
+#define TABLE_COUNT              8
+
+#endif
+
+struct dcn31_watermarks {
+  // Watermarks
+  WatermarkRowGeneric_t WatermarkRow[WM_COUNT][NUM_WM_RANGES];
+
+  uint32_t MmHubPadding[7]; // SMU internal use
+};
+
+struct dcn31_smu_dpm_clks {
+	DpmClocks_t *dpm_clks;
+	union large_integer mc_address;
+};
+
+/* TODO: taken from vgh, may not be correct */
+struct display_idle_optimization {
+	unsigned int df_request_disabled : 1;
+	unsigned int phy_ref_clk_off     : 1;
+	unsigned int s0i2_rdy            : 1;
+	unsigned int reserved            : 29;
+};
+
+union display_idle_optimization_u {
+	struct display_idle_optimization idle_info;
+	uint32_t data;
+};
+
+int dcn31_smu_get_smu_version(struct clk_mgr_internal *clk_mgr);
+int dcn31_smu_set_dispclk(struct clk_mgr_internal *clk_mgr, int requested_dispclk_khz);
+int dcn31_smu_set_dprefclk(struct clk_mgr_internal *clk_mgr);
+int dcn31_smu_set_hard_min_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_dcfclk_khz);
+int dcn31_smu_set_min_deep_sleep_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_min_ds_dcfclk_khz);
+int dcn31_smu_set_dppclk(struct clk_mgr_internal *clk_mgr, int requested_dpp_khz);
+void dcn31_smu_set_display_idle_optimization(struct clk_mgr_internal *clk_mgr, uint32_t idle_info);
+void dcn31_smu_enable_phy_refclk_pwrdwn(struct clk_mgr_internal *clk_mgr, bool enable);
+void dcn31_smu_enable_pme_wa(struct clk_mgr_internal *clk_mgr);
+void dcn31_smu_set_dram_addr_high(struct clk_mgr_internal *clk_mgr, uint32_t addr_high);
+void dcn31_smu_set_dram_addr_low(struct clk_mgr_internal *clk_mgr, uint32_t addr_low);
+void dcn31_smu_transfer_dpm_table_smu_2_dram(struct clk_mgr_internal *clk_mgr);
+void dcn31_smu_transfer_wm_table_dram_2_smu(struct clk_mgr_internal *clk_mgr);
+
+void dcn31_smu_set_zstate_support(struct clk_mgr_internal *clk_mgr, enum dcn_zstate_support_state support);
+void dcn31_smu_set_dtbclk(struct clk_mgr_internal *clk_mgr, bool enable);
+
+#endif /* DAL_DC_31_SMU_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_clk_mgr.c
new file mode 100644
index 000000000000..9e63fa72101c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_clk_mgr.c
@@ -0,0 +1,1051 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+
+#include "dcn314_clk_mgr.h"
+
+#include "dccg.h"
+#include "clk_mgr_internal.h"
+
+// For dce12_get_dp_ref_freq_khz
+#include "dce100/dce_clk_mgr.h"
+
+// For dcn20_update_clocks_update_dpp_dto
+#include "dcn20/dcn20_clk_mgr.h"
+
+
+
+#include "reg_helper.h"
+#include "core_types.h"
+#include "dm_helpers.h"
+
+/* TODO: remove this include once we ported over remaining clk mgr functions*/
+#include "dcn30/dcn30_clk_mgr.h"
+#include "dcn31/dcn31_clk_mgr.h"
+
+#include "dc_dmub_srv.h"
+#include "link_service.h"
+#include "dcn314_smu.h"
+
+
+#include "logger_types.h"
+
+
+#define MAX_INSTANCE                                        7
+#define MAX_SEGMENT                                         8
+
+struct IP_BASE_INSTANCE {
+	unsigned int segment[MAX_SEGMENT];
+};
+
+struct IP_BASE {
+	struct IP_BASE_INSTANCE instance[MAX_INSTANCE];
+};
+
+static const struct IP_BASE CLK_BASE = { { { { 0x00016C00, 0x02401800, 0, 0, 0, 0, 0, 0 } },
+					{ { 0x00016E00, 0x02401C00, 0, 0, 0, 0, 0, 0 } },
+					{ { 0x00017000, 0x02402000, 0, 0, 0, 0, 0, 0 } },
+					{ { 0x00017200, 0x02402400, 0, 0, 0, 0, 0, 0 } },
+					{ { 0x0001B000, 0x0242D800, 0, 0, 0, 0, 0, 0 } },
+					{ { 0x0001B200, 0x0242DC00, 0, 0, 0, 0, 0, 0 } },
+					{ { 0x0001B400, 0x0242E000, 0, 0, 0, 0, 0, 0 } } } };
+
+#undef DC_LOGGER
+#define DC_LOGGER \
+	clk_mgr->base.base.ctx->logger
+
+#define regCLK1_CLK_PLL_REQ			0x0237
+#define regCLK1_CLK_PLL_REQ_BASE_IDX		0
+
+#define CLK1_CLK_PLL_REQ__FbMult_int__SHIFT	0x0
+#define CLK1_CLK_PLL_REQ__PllSpineDiv__SHIFT	0xc
+#define CLK1_CLK_PLL_REQ__FbMult_frac__SHIFT	0x10
+#define CLK1_CLK_PLL_REQ__FbMult_int_MASK	0x000001FFL
+#define CLK1_CLK_PLL_REQ__PllSpineDiv_MASK	0x0000F000L
+#define CLK1_CLK_PLL_REQ__FbMult_frac_MASK	0xFFFF0000L
+
+#define regCLK1_CLK0_DFS_CNTL				0x0269
+#define regCLK1_CLK0_DFS_CNTL_BASE_IDX		0
+#define regCLK1_CLK1_DFS_CNTL				0x026c
+#define regCLK1_CLK1_DFS_CNTL_BASE_IDX		0
+#define regCLK1_CLK2_DFS_CNTL				0x026f
+#define regCLK1_CLK2_DFS_CNTL_BASE_IDX		0
+#define regCLK1_CLK3_DFS_CNTL				0x0272
+#define regCLK1_CLK3_DFS_CNTL_BASE_IDX		0
+#define regCLK1_CLK4_DFS_CNTL				0x0275
+#define regCLK1_CLK4_DFS_CNTL_BASE_IDX		0
+#define regCLK1_CLK5_DFS_CNTL				0x0278
+#define regCLK1_CLK5_DFS_CNTL_BASE_IDX		0
+
+#define regCLK1_CLK0_CURRENT_CNT			0x02fb
+#define regCLK1_CLK0_CURRENT_CNT_BASE_IDX	0
+#define regCLK1_CLK1_CURRENT_CNT			0x02fc
+#define regCLK1_CLK1_CURRENT_CNT_BASE_IDX	0
+#define regCLK1_CLK2_CURRENT_CNT			0x02fd
+#define regCLK1_CLK2_CURRENT_CNT_BASE_IDX	0
+#define regCLK1_CLK3_CURRENT_CNT			0x02fe
+#define regCLK1_CLK3_CURRENT_CNT_BASE_IDX	0
+#define regCLK1_CLK4_CURRENT_CNT			0x02ff
+#define regCLK1_CLK4_CURRENT_CNT_BASE_IDX	0
+#define regCLK1_CLK5_CURRENT_CNT			0x0300
+#define regCLK1_CLK5_CURRENT_CNT_BASE_IDX	0
+
+#define regCLK1_CLK0_BYPASS_CNTL			0x028a
+#define regCLK1_CLK0_BYPASS_CNTL_BASE_IDX	0
+#define regCLK1_CLK1_BYPASS_CNTL			0x0293
+#define regCLK1_CLK1_BYPASS_CNTL_BASE_IDX	0
+#define regCLK1_CLK2_BYPASS_CNTL			0x029c
+#define regCLK1_CLK2_BYPASS_CNTL_BASE_IDX	0
+#define regCLK1_CLK3_BYPASS_CNTL			0x02a5
+#define regCLK1_CLK3_BYPASS_CNTL_BASE_IDX	0
+#define regCLK1_CLK4_BYPASS_CNTL			0x02ae
+#define regCLK1_CLK4_BYPASS_CNTL_BASE_IDX	0
+#define regCLK1_CLK5_BYPASS_CNTL			0x02b7
+#define regCLK1_CLK5_BYPASS_CNTL_BASE_IDX	0
+
+#define regCLK1_CLK0_DS_CNTL				0x0283
+#define regCLK1_CLK0_DS_CNTL_BASE_IDX		0
+#define regCLK1_CLK1_DS_CNTL				0x028c
+#define regCLK1_CLK1_DS_CNTL_BASE_IDX		0
+#define regCLK1_CLK2_DS_CNTL				0x0295
+#define regCLK1_CLK2_DS_CNTL_BASE_IDX		0
+#define regCLK1_CLK3_DS_CNTL				0x029e
+#define regCLK1_CLK3_DS_CNTL_BASE_IDX		0
+#define regCLK1_CLK4_DS_CNTL				0x02a7
+#define regCLK1_CLK4_DS_CNTL_BASE_IDX		0
+#define regCLK1_CLK5_DS_CNTL				0x02b0
+#define regCLK1_CLK5_DS_CNTL_BASE_IDX		0
+
+#define regCLK1_CLK0_ALLOW_DS				0x0284
+#define regCLK1_CLK0_ALLOW_DS_BASE_IDX		0
+#define regCLK1_CLK1_ALLOW_DS				0x028d
+#define regCLK1_CLK1_ALLOW_DS_BASE_IDX		0
+#define regCLK1_CLK2_ALLOW_DS				0x0296
+#define regCLK1_CLK2_ALLOW_DS_BASE_IDX		0
+#define regCLK1_CLK3_ALLOW_DS				0x029f
+#define regCLK1_CLK3_ALLOW_DS_BASE_IDX		0
+#define regCLK1_CLK4_ALLOW_DS				0x02a8
+#define regCLK1_CLK4_ALLOW_DS_BASE_IDX		0
+#define regCLK1_CLK5_ALLOW_DS				0x02b1
+#define regCLK1_CLK5_ALLOW_DS_BASE_IDX		0
+
+#define CLK1_CLK2_BYPASS_CNTL__CLK2_BYPASS_SEL__SHIFT	0x0
+#define CLK1_CLK2_BYPASS_CNTL__CLK2_BYPASS_DIV__SHIFT	0x10
+#define CLK1_CLK2_BYPASS_CNTL__CLK2_BYPASS_SEL_MASK		0x00000007L
+#define CLK1_CLK2_BYPASS_CNTL__CLK2_BYPASS_DIV_MASK		0x000F0000L
+
+#define regCLK6_0_CLK6_spll_field_8				0x464b
+#define regCLK6_0_CLK6_spll_field_8_BASE_IDX	0
+
+#define CLK6_0_CLK6_spll_field_8__spll_ssc_en__SHIFT	0xd
+#define CLK6_0_CLK6_spll_field_8__spll_ssc_en_MASK		0x00002000L
+
+#define REG(reg_name) \
+	(CLK_BASE.instance[0].segment[reg ## reg_name ## _BASE_IDX] + reg ## reg_name)
+
+#define TO_CLK_MGR_DCN314(clk_mgr)\
+	container_of(clk_mgr, struct clk_mgr_dcn314, base)
+
+static int dcn314_get_active_display_cnt_wa(
+		struct dc *dc,
+		struct dc_state *context)
+{
+	int i, display_count;
+	bool tmds_present = false;
+
+	display_count = 0;
+	for (i = 0; i < context->stream_count; i++) {
+		const struct dc_stream_state *stream = context->streams[i];
+
+		if (stream->signal == SIGNAL_TYPE_HDMI_TYPE_A ||
+				stream->signal == SIGNAL_TYPE_DVI_SINGLE_LINK ||
+				stream->signal == SIGNAL_TYPE_DVI_DUAL_LINK)
+			tmds_present = true;
+
+		/* Checking stream / link detection ensuring that PHY is active*/
+		if (dc_is_dp_signal(stream->signal) && !stream->dpms_off)
+			display_count++;
+
+	}
+
+	for (i = 0; i < dc->link_count; i++) {
+		const struct dc_link *link = dc->links[i];
+
+		/* abusing the fact that the dig and phy are coupled to see if the phy is enabled */
+		if (link->link_enc && link->link_enc->funcs->is_dig_enabled &&
+				link->link_enc->funcs->is_dig_enabled(link->link_enc))
+			display_count++;
+	}
+
+	/* WA for hang on HDMI after display off back on*/
+	if (display_count == 0 && tmds_present)
+		display_count = 1;
+
+	return display_count;
+}
+
+static void dcn314_disable_otg_wa(struct clk_mgr *clk_mgr_base, struct dc_state *context,
+				  bool safe_to_lower, bool disable)
+{
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	int i;
+
+	for (i = 0; i < dc->res_pool->pipe_count; ++i) {
+		struct pipe_ctx *pipe = safe_to_lower
+			? &context->res_ctx.pipe_ctx[i]
+			: &dc->current_state->res_ctx.pipe_ctx[i];
+
+		if (pipe->top_pipe || pipe->prev_odm_pipe)
+			continue;
+		if (pipe->stream && (pipe->stream->dpms_off || dc_is_virtual_signal(pipe->stream->signal))) {
+			if (disable) {
+				if (pipe->stream_res.tg && pipe->stream_res.tg->funcs->immediate_disable_crtc)
+					pipe->stream_res.tg->funcs->immediate_disable_crtc(pipe->stream_res.tg);
+
+				reset_sync_context_for_pipe(dc, context, i);
+			} else {
+				pipe->stream_res.tg->funcs->enable_crtc(pipe->stream_res.tg);
+			}
+		}
+	}
+}
+
+bool dcn314_is_spll_ssc_enabled(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	uint32_t ssc_enable;
+
+	REG_GET(CLK6_0_CLK6_spll_field_8, spll_ssc_en, &ssc_enable);
+
+	return ssc_enable == 1;
+}
+
+void dcn314_init_clocks(struct clk_mgr *clk_mgr)
+{
+	struct clk_mgr_internal *clk_mgr_int = TO_CLK_MGR_INTERNAL(clk_mgr);
+	uint32_t ref_dtbclk = clk_mgr->clks.ref_dtbclk_khz;
+	struct clk_mgr_dcn314 *clk_mgr_dcn314 = TO_CLK_MGR_DCN314(clk_mgr_int);
+	struct clk_log_info log_info = {0};
+
+	memset(&(clk_mgr->clks), 0, sizeof(struct dc_clocks));
+	// Assumption is that boot state always supports pstate
+	clk_mgr->clks.ref_dtbclk_khz = ref_dtbclk;	// restore ref_dtbclk
+	clk_mgr->clks.p_state_change_support = true;
+	clk_mgr->clks.prev_p_state_change_support = true;
+	clk_mgr->clks.pwr_state = DCN_PWR_STATE_UNKNOWN;
+	clk_mgr->clks.zstate_support = DCN_ZSTATE_SUPPORT_UNKNOWN;
+
+	// to adjust dp_dto reference clock if ssc is enable otherwise to apply dprefclk
+	if (dcn314_is_spll_ssc_enabled(clk_mgr))
+		clk_mgr->dp_dto_source_clock_in_khz =
+			dce_adjust_dp_ref_freq_for_ss(clk_mgr_int, clk_mgr->dprefclk_khz);
+	else
+		clk_mgr->dp_dto_source_clock_in_khz = clk_mgr->dprefclk_khz;
+
+	dcn314_dump_clk_registers(&clk_mgr->boot_snapshot, &clk_mgr_dcn314->base.base, &log_info);
+	clk_mgr->clks.dispclk_khz =  clk_mgr->boot_snapshot.dispclk * 1000;
+}
+
+void dcn314_update_clocks(struct clk_mgr *clk_mgr_base,
+			struct dc_state *context,
+			bool safe_to_lower)
+{
+	union dmub_rb_cmd cmd;
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dc_clocks *new_clocks = &context->bw_ctx.bw.dcn.clk;
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	int display_count;
+	bool update_dppclk = false;
+	bool update_dispclk = false;
+	bool dpp_clock_lowered = false;
+
+	if (dc->work_arounds.skip_clock_update)
+		return;
+
+	display_count = dcn314_get_active_display_cnt_wa(dc, context);
+
+	/*
+	 * if it is safe to lower, but we are already in the lower state, we don't have to do anything
+	 * also if safe to lower is false, we just go in the higher state
+	 */
+	if (safe_to_lower) {
+		if (new_clocks->zstate_support != DCN_ZSTATE_SUPPORT_DISALLOW &&
+				new_clocks->zstate_support != clk_mgr_base->clks.zstate_support) {
+			dcn314_smu_set_zstate_support(clk_mgr, new_clocks->zstate_support);
+			dm_helpers_enable_periodic_detection(clk_mgr_base->ctx, true);
+			clk_mgr_base->clks.zstate_support = new_clocks->zstate_support;
+		}
+
+		if (clk_mgr_base->clks.dtbclk_en && !new_clocks->dtbclk_en) {
+			dcn314_smu_set_dtbclk(clk_mgr, false);
+			clk_mgr_base->clks.dtbclk_en = new_clocks->dtbclk_en;
+		}
+		/* check that we're not already in lower */
+		if (clk_mgr_base->clks.pwr_state != DCN_PWR_STATE_LOW_POWER) {
+			/* if we can go lower, go lower */
+			if (display_count == 0) {
+				union display_idle_optimization_u idle_info = { 0 };
+				idle_info.idle_info.df_request_disabled = 1;
+				idle_info.idle_info.phy_ref_clk_off = 1;
+				idle_info.idle_info.s0i2_rdy = 1;
+				dcn314_smu_set_display_idle_optimization(clk_mgr, idle_info.data);
+				/* update power state */
+				clk_mgr_base->clks.pwr_state = DCN_PWR_STATE_LOW_POWER;
+			}
+		}
+	} else {
+		if (new_clocks->zstate_support == DCN_ZSTATE_SUPPORT_DISALLOW &&
+				new_clocks->zstate_support != clk_mgr_base->clks.zstate_support) {
+			dcn314_smu_set_zstate_support(clk_mgr, DCN_ZSTATE_SUPPORT_DISALLOW);
+			dm_helpers_enable_periodic_detection(clk_mgr_base->ctx, false);
+			clk_mgr_base->clks.zstate_support = new_clocks->zstate_support;
+		}
+
+		if (!clk_mgr_base->clks.dtbclk_en && new_clocks->dtbclk_en) {
+			dcn314_smu_set_dtbclk(clk_mgr, true);
+			clk_mgr_base->clks.dtbclk_en = new_clocks->dtbclk_en;
+		}
+
+		/* check that we're not already in D0 */
+		if (clk_mgr_base->clks.pwr_state != DCN_PWR_STATE_MISSION_MODE) {
+			union display_idle_optimization_u idle_info = { 0 };
+
+			dcn314_smu_set_display_idle_optimization(clk_mgr, idle_info.data);
+			/* update power state */
+			clk_mgr_base->clks.pwr_state = DCN_PWR_STATE_MISSION_MODE;
+		}
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dcfclk_khz, clk_mgr_base->clks.dcfclk_khz)) {
+		clk_mgr_base->clks.dcfclk_khz = new_clocks->dcfclk_khz;
+		dcn314_smu_set_hard_min_dcfclk(clk_mgr, clk_mgr_base->clks.dcfclk_khz);
+	}
+
+	if (should_set_clock(safe_to_lower,
+			new_clocks->dcfclk_deep_sleep_khz, clk_mgr_base->clks.dcfclk_deep_sleep_khz)) {
+		clk_mgr_base->clks.dcfclk_deep_sleep_khz = new_clocks->dcfclk_deep_sleep_khz;
+		dcn314_smu_set_min_deep_sleep_dcfclk(clk_mgr, clk_mgr_base->clks.dcfclk_deep_sleep_khz);
+	}
+
+	// workaround: Limit dppclk to 100Mhz to avoid lower eDP panel switch to plus 4K monitor underflow.
+	if (new_clocks->dppclk_khz < 100000)
+		new_clocks->dppclk_khz = 100000;
+
+	if (should_set_clock(safe_to_lower, new_clocks->dppclk_khz, clk_mgr->base.clks.dppclk_khz)) {
+		if (clk_mgr->base.clks.dppclk_khz > new_clocks->dppclk_khz)
+			dpp_clock_lowered = true;
+		clk_mgr_base->clks.dppclk_khz = new_clocks->dppclk_khz;
+		update_dppclk = true;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, clk_mgr_base->clks.dispclk_khz) &&
+	    (new_clocks->dispclk_khz > 0 || (safe_to_lower && display_count == 0))) {
+		int requested_dispclk_khz = new_clocks->dispclk_khz;
+
+		dcn314_disable_otg_wa(clk_mgr_base, context, safe_to_lower, true);
+
+		/* Clamp the requested clock to PMFW based on their limit. */
+		if (dc->debug.min_disp_clk_khz > 0 && requested_dispclk_khz < dc->debug.min_disp_clk_khz)
+			requested_dispclk_khz = dc->debug.min_disp_clk_khz;
+
+		dcn314_smu_set_dispclk(clk_mgr, requested_dispclk_khz);
+		clk_mgr_base->clks.dispclk_khz = new_clocks->dispclk_khz;
+
+		dcn314_disable_otg_wa(clk_mgr_base, context, safe_to_lower, false);
+
+		update_dispclk = true;
+	}
+
+	if (dpp_clock_lowered) {
+		// increase per DPP DTO before lowering global dppclk
+		dcn20_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
+		dcn314_smu_set_dppclk(clk_mgr, clk_mgr_base->clks.dppclk_khz);
+	} else {
+		// increase global DPPCLK before lowering per DPP DTO
+		if (update_dppclk || update_dispclk)
+			dcn314_smu_set_dppclk(clk_mgr, clk_mgr_base->clks.dppclk_khz);
+		// always update dtos unless clock is lowered and not safe to lower
+		if (new_clocks->dppclk_khz >= dc->current_state->bw_ctx.bw.dcn.clk.dppclk_khz)
+			dcn20_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
+	}
+
+	// notify DMCUB of latest clocks
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.notify_clocks.header.type = DMUB_CMD__CLK_MGR;
+	cmd.notify_clocks.header.sub_type = DMUB_CMD__CLK_MGR_NOTIFY_CLOCKS;
+	cmd.notify_clocks.clocks.dcfclk_khz = clk_mgr_base->clks.dcfclk_khz;
+	cmd.notify_clocks.clocks.dcfclk_deep_sleep_khz =
+		clk_mgr_base->clks.dcfclk_deep_sleep_khz;
+	cmd.notify_clocks.clocks.dispclk_khz = clk_mgr_base->clks.dispclk_khz;
+	cmd.notify_clocks.clocks.dppclk_khz = clk_mgr_base->clks.dppclk_khz;
+
+	dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+static int get_vco_frequency_from_reg(struct clk_mgr_internal *clk_mgr)
+{
+	/* get FbMult value */
+	struct fixed31_32 pll_req;
+	unsigned int fbmult_frac_val = 0;
+	unsigned int fbmult_int_val = 0;
+
+	/*
+	 * Register value of fbmult is in 8.16 format, we are converting to 314.32
+	 * to leverage the fix point operations available in driver
+	 */
+
+	REG_GET(CLK1_CLK_PLL_REQ, FbMult_frac, &fbmult_frac_val); /* 16 bit fractional part*/
+	REG_GET(CLK1_CLK_PLL_REQ, FbMult_int, &fbmult_int_val); /* 8 bit integer part */
+
+	pll_req = dc_fixpt_from_int(fbmult_int_val);
+
+	/*
+	 * since fractional part is only 16 bit in register definition but is 32 bit
+	 * in our fix point definiton, need to shift left by 16 to obtain correct value
+	 */
+	pll_req.value |= fbmult_frac_val << 16;
+
+	/* multiply by REFCLK period */
+	pll_req = dc_fixpt_mul_int(pll_req, clk_mgr->dfs_ref_freq_khz);
+
+	/* integer part is now VCO frequency in kHz */
+	return dc_fixpt_floor(pll_req);
+}
+
+static void dcn314_enable_pme_wa(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	dcn314_smu_enable_pme_wa(clk_mgr);
+}
+
+bool dcn314_are_clock_states_equal(struct dc_clocks *a,
+		struct dc_clocks *b)
+{
+	if (a->dispclk_khz != b->dispclk_khz)
+		return false;
+	else if (a->dppclk_khz != b->dppclk_khz)
+		return false;
+	else if (a->dcfclk_khz != b->dcfclk_khz)
+		return false;
+	else if (a->dcfclk_deep_sleep_khz != b->dcfclk_deep_sleep_khz)
+		return false;
+	else if (a->zstate_support != b->zstate_support)
+		return false;
+	else if (a->dtbclk_en != b->dtbclk_en)
+		return false;
+
+	return true;
+}
+
+
+static void dcn314_dump_clk_registers_internal(struct dcn35_clk_internal *internal, struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	// read dtbclk
+	internal->CLK1_CLK4_CURRENT_CNT = REG_READ(CLK1_CLK4_CURRENT_CNT);
+	internal->CLK1_CLK4_BYPASS_CNTL = REG_READ(CLK1_CLK4_BYPASS_CNTL);
+
+	// read dcfclk
+	internal->CLK1_CLK3_CURRENT_CNT = REG_READ(CLK1_CLK3_CURRENT_CNT);
+	internal->CLK1_CLK3_BYPASS_CNTL = REG_READ(CLK1_CLK3_BYPASS_CNTL);
+
+	// read dcf deep sleep divider
+	internal->CLK1_CLK3_DS_CNTL = REG_READ(CLK1_CLK3_DS_CNTL);
+	internal->CLK1_CLK3_ALLOW_DS = REG_READ(CLK1_CLK3_ALLOW_DS);
+
+	// read dppclk
+	internal->CLK1_CLK1_CURRENT_CNT = REG_READ(CLK1_CLK1_CURRENT_CNT);
+	internal->CLK1_CLK1_BYPASS_CNTL = REG_READ(CLK1_CLK1_BYPASS_CNTL);
+
+	// read dprefclk
+	internal->CLK1_CLK2_CURRENT_CNT = REG_READ(CLK1_CLK2_CURRENT_CNT);
+	internal->CLK1_CLK2_BYPASS_CNTL = REG_READ(CLK1_CLK2_BYPASS_CNTL);
+
+	// read dispclk
+	internal->CLK1_CLK0_CURRENT_CNT = REG_READ(CLK1_CLK0_CURRENT_CNT);
+	internal->CLK1_CLK0_BYPASS_CNTL = REG_READ(CLK1_CLK0_BYPASS_CNTL);
+}
+
+void dcn314_dump_clk_registers(struct clk_state_registers_and_bypass *regs_and_bypass,
+		struct clk_mgr *clk_mgr_base, struct clk_log_info *log_info)
+{
+
+	struct dcn35_clk_internal internal = {0};
+
+	dcn314_dump_clk_registers_internal(&internal, clk_mgr_base);
+
+	regs_and_bypass->dcfclk = internal.CLK1_CLK3_CURRENT_CNT / 10;
+	regs_and_bypass->dcf_deep_sleep_divider = internal.CLK1_CLK3_DS_CNTL / 10;
+	regs_and_bypass->dcf_deep_sleep_allow = internal.CLK1_CLK3_ALLOW_DS;
+	regs_and_bypass->dprefclk = internal.CLK1_CLK2_CURRENT_CNT / 10;
+	regs_and_bypass->dispclk = internal.CLK1_CLK0_CURRENT_CNT / 10;
+	regs_and_bypass->dppclk = internal.CLK1_CLK1_CURRENT_CNT / 10;
+	regs_and_bypass->dtbclk = internal.CLK1_CLK4_CURRENT_CNT / 10;
+
+	regs_and_bypass->dppclk_bypass = internal.CLK1_CLK1_BYPASS_CNTL & 0x0007;
+	if (regs_and_bypass->dppclk_bypass < 0 || regs_and_bypass->dppclk_bypass > 4)
+		regs_and_bypass->dppclk_bypass = 0;
+	regs_and_bypass->dcfclk_bypass = internal.CLK1_CLK3_BYPASS_CNTL & 0x0007;
+	if (regs_and_bypass->dcfclk_bypass < 0 || regs_and_bypass->dcfclk_bypass > 4)
+		regs_and_bypass->dcfclk_bypass = 0;
+	regs_and_bypass->dispclk_bypass = internal.CLK1_CLK0_BYPASS_CNTL & 0x0007;
+	if (regs_and_bypass->dispclk_bypass < 0 || regs_and_bypass->dispclk_bypass > 4)
+		regs_and_bypass->dispclk_bypass = 0;
+	regs_and_bypass->dprefclk_bypass = internal.CLK1_CLK2_BYPASS_CNTL & 0x0007;
+	if (regs_and_bypass->dprefclk_bypass < 0 || regs_and_bypass->dprefclk_bypass > 4)
+		regs_and_bypass->dprefclk_bypass = 0;
+
+}
+
+static struct clk_bw_params dcn314_bw_params = {
+	.vram_type = Ddr4MemType,
+	.num_channels = 1,
+	.clk_table = {
+		.num_entries = 4,
+	},
+
+};
+
+static struct wm_table ddr5_wm_table = {
+	.entries = {
+		{
+			.wm_inst = WM_A,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 12.5,
+			.sr_enter_plus_exit_time_us = 14.5,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_B,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 12.5,
+			.sr_enter_plus_exit_time_us = 14.5,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_C,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 12.5,
+			.sr_enter_plus_exit_time_us = 14.5,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_D,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 12.5,
+			.sr_enter_plus_exit_time_us = 14.5,
+			.valid = true,
+		},
+	}
+};
+
+static struct wm_table lpddr5_wm_table = {
+	.entries = {
+		{
+			.wm_inst = WM_A,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 30.0,
+			.sr_enter_plus_exit_time_us = 32.0,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_B,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 30.0,
+			.sr_enter_plus_exit_time_us = 32.0,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_C,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 30.0,
+			.sr_enter_plus_exit_time_us = 32.0,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_D,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 30.0,
+			.sr_enter_plus_exit_time_us = 32.0,
+			.valid = true,
+		},
+	}
+};
+
+static DpmClocks314_t dummy_clocks;
+
+static struct dcn314_watermarks dummy_wms = { 0 };
+
+static struct dcn314_ss_info_table ss_info_table = {
+	.ss_divider = 1000,
+	.ss_percentage = {0, 0, 375, 375, 375}
+};
+
+static void dcn314_build_watermark_ranges(struct clk_bw_params *bw_params, struct dcn314_watermarks *table)
+{
+	int i, num_valid_sets;
+
+	num_valid_sets = 0;
+
+	for (i = 0; i < WM_SET_COUNT; i++) {
+		/* skip empty entries, the smu array has no holes*/
+		if (!bw_params->wm_table.entries[i].valid)
+			continue;
+
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].WmSetting = bw_params->wm_table.entries[i].wm_inst;
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].WmType = bw_params->wm_table.entries[i].wm_type;
+		/* We will not select WM based on fclk, so leave it as unconstrained */
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinClock = 0;
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].MaxClock = 0xFFFF;
+
+		if (table->WatermarkRow[WM_DCFCLK][num_valid_sets].WmType == WM_TYPE_PSTATE_CHG) {
+			if (i == 0)
+				table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinMclk = 0;
+			else {
+				/* add 1 to make it non-overlapping with next lvl */
+				table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinMclk =
+						bw_params->clk_table.entries[i - 1].dcfclk_mhz + 1;
+			}
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets].MaxMclk =
+					bw_params->clk_table.entries[i].dcfclk_mhz;
+
+		} else {
+			/* unconstrained for memory retraining */
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinClock = 0;
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets].MaxClock = 0xFFFF;
+
+			/* Modify previous watermark range to cover up to max */
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets - 1].MaxClock = 0xFFFF;
+		}
+		num_valid_sets++;
+	}
+
+	ASSERT(num_valid_sets != 0); /* Must have at least one set of valid watermarks */
+
+	/* modify the min and max to make sure we cover the whole range*/
+	table->WatermarkRow[WM_DCFCLK][0].MinMclk = 0;
+	table->WatermarkRow[WM_DCFCLK][0].MinClock = 0;
+	table->WatermarkRow[WM_DCFCLK][num_valid_sets - 1].MaxMclk = 0xFFFF;
+	table->WatermarkRow[WM_DCFCLK][num_valid_sets - 1].MaxClock = 0xFFFF;
+
+	/* This is for writeback only, does not matter currently as no writeback support*/
+	table->WatermarkRow[WM_SOCCLK][0].WmSetting = WM_A;
+	table->WatermarkRow[WM_SOCCLK][0].MinClock = 0;
+	table->WatermarkRow[WM_SOCCLK][0].MaxClock = 0xFFFF;
+	table->WatermarkRow[WM_SOCCLK][0].MinMclk = 0;
+	table->WatermarkRow[WM_SOCCLK][0].MaxMclk = 0xFFFF;
+}
+
+static void dcn314_notify_wm_ranges(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct clk_mgr_dcn314 *clk_mgr_dcn314 = TO_CLK_MGR_DCN314(clk_mgr);
+	struct dcn314_watermarks *table = clk_mgr_dcn314->smu_wm_set.wm_set;
+
+	if (!clk_mgr->smu_ver)
+		return;
+
+	if (!table || clk_mgr_dcn314->smu_wm_set.mc_address.quad_part == 0)
+		return;
+
+	memset(table, 0, sizeof(*table));
+
+	dcn314_build_watermark_ranges(clk_mgr_base->bw_params, table);
+
+	dcn314_smu_set_dram_addr_high(clk_mgr,
+			clk_mgr_dcn314->smu_wm_set.mc_address.high_part);
+	dcn314_smu_set_dram_addr_low(clk_mgr,
+			clk_mgr_dcn314->smu_wm_set.mc_address.low_part);
+	dcn314_smu_transfer_wm_table_dram_2_smu(clk_mgr);
+}
+
+static void dcn314_get_dpm_table_from_smu(struct clk_mgr_internal *clk_mgr,
+		struct dcn314_smu_dpm_clks *smu_dpm_clks)
+{
+	DpmClocks314_t *table = smu_dpm_clks->dpm_clks;
+
+	if (!clk_mgr->smu_ver)
+		return;
+
+	if (!table || smu_dpm_clks->mc_address.quad_part == 0)
+		return;
+
+	memset(table, 0, sizeof(*table));
+
+	dcn314_smu_set_dram_addr_high(clk_mgr,
+			smu_dpm_clks->mc_address.high_part);
+	dcn314_smu_set_dram_addr_low(clk_mgr,
+			smu_dpm_clks->mc_address.low_part);
+	dcn314_smu_transfer_dpm_table_smu_2_dram(clk_mgr);
+}
+
+static inline bool is_valid_clock_value(uint32_t clock_value)
+{
+	return clock_value > 1 && clock_value < 100000;
+}
+
+static unsigned int convert_wck_ratio(uint8_t wck_ratio)
+{
+	switch (wck_ratio) {
+	case WCK_RATIO_1_2:
+		return 2;
+
+	case WCK_RATIO_1_4:
+		return 4;
+
+	default:
+		break;
+	}
+	return 1;
+}
+
+static uint32_t find_max_clk_value(const uint32_t clocks[], uint32_t num_clocks)
+{
+	uint32_t max = 0;
+	int i;
+
+	for (i = 0; i < num_clocks; ++i) {
+		if (clocks[i] > max)
+			max = clocks[i];
+	}
+
+	return max;
+}
+
+static void dcn314_clk_mgr_helper_populate_bw_params(struct clk_mgr_internal *clk_mgr,
+						    struct integrated_info *bios_info,
+						    const DpmClocks314_t *clock_table)
+{
+	struct clk_bw_params *bw_params = clk_mgr->base.bw_params;
+	struct clk_limit_table_entry def_max = bw_params->clk_table.entries[bw_params->clk_table.num_entries - 1];
+	uint32_t max_pstate = 0,  max_fclk = 0,  min_pstate = 0, max_dispclk = 0, max_dppclk = 0;
+	int i;
+
+	/* Find highest valid fclk pstate */
+	for (i = 0; i < clock_table->NumDfPstatesEnabled; i++) {
+		if (is_valid_clock_value(clock_table->DfPstateTable[i].FClk) &&
+		    clock_table->DfPstateTable[i].FClk > max_fclk) {
+			max_fclk = clock_table->DfPstateTable[i].FClk;
+			max_pstate = i;
+		}
+	}
+
+	/* We expect the table to contain at least one valid fclk entry. */
+	ASSERT(is_valid_clock_value(max_fclk));
+
+	/* Dispclk and dppclk can be max at any voltage, same number of levels for both */
+	if (clock_table->NumDispClkLevelsEnabled <= NUM_DISPCLK_DPM_LEVELS &&
+	    clock_table->NumDispClkLevelsEnabled <= NUM_DPPCLK_DPM_LEVELS) {
+		max_dispclk = find_max_clk_value(clock_table->DispClocks, clock_table->NumDispClkLevelsEnabled);
+		max_dppclk = find_max_clk_value(clock_table->DppClocks, clock_table->NumDispClkLevelsEnabled);
+	} else {
+		/* Invalid number of entries in the table from PMFW. */
+		ASSERT(0);
+	}
+
+	/* Base the clock table on dcfclk, need at least one entry regardless of pmfw table */
+	for (i = 0; i < clock_table->NumDcfClkLevelsEnabled; i++) {
+		uint32_t min_fclk = clock_table->DfPstateTable[0].FClk;
+		int j;
+
+		for (j = 1; j < clock_table->NumDfPstatesEnabled; j++) {
+			if (is_valid_clock_value(clock_table->DfPstateTable[j].FClk) &&
+			    clock_table->DfPstateTable[j].FClk < min_fclk &&
+			    clock_table->DfPstateTable[j].Voltage <= clock_table->SocVoltage[i]) {
+				min_fclk = clock_table->DfPstateTable[j].FClk;
+				min_pstate = j;
+			}
+		}
+
+		/* First search defaults for the clocks we don't read using closest lower or equal default dcfclk */
+		for (j = bw_params->clk_table.num_entries - 1; j > 0; j--)
+			if (bw_params->clk_table.entries[j].dcfclk_mhz <= clock_table->DcfClocks[i])
+				break;
+
+		bw_params->clk_table.entries[i].phyclk_mhz = bw_params->clk_table.entries[j].phyclk_mhz;
+		bw_params->clk_table.entries[i].phyclk_d18_mhz = bw_params->clk_table.entries[j].phyclk_d18_mhz;
+		bw_params->clk_table.entries[i].dtbclk_mhz = bw_params->clk_table.entries[j].dtbclk_mhz;
+
+		/* Now update clocks we do read */
+		bw_params->clk_table.entries[i].fclk_mhz = min_fclk;
+		bw_params->clk_table.entries[i].memclk_mhz = clock_table->DfPstateTable[min_pstate].MemClk;
+		bw_params->clk_table.entries[i].voltage = clock_table->DfPstateTable[min_pstate].Voltage;
+		bw_params->clk_table.entries[i].dcfclk_mhz = clock_table->DcfClocks[i];
+		bw_params->clk_table.entries[i].socclk_mhz = clock_table->SocClocks[i];
+		bw_params->clk_table.entries[i].dispclk_mhz = max_dispclk;
+		bw_params->clk_table.entries[i].dppclk_mhz = max_dppclk;
+		bw_params->clk_table.entries[i].wck_ratio = convert_wck_ratio(
+			clock_table->DfPstateTable[min_pstate].WckRatio);
+	}
+
+	/* Make sure to include at least one entry at highest pstate */
+	if (max_pstate != min_pstate || i == 0) {
+		if (i > MAX_NUM_DPM_LVL - 1)
+			i = MAX_NUM_DPM_LVL - 1;
+
+		bw_params->clk_table.entries[i].fclk_mhz = max_fclk;
+		bw_params->clk_table.entries[i].memclk_mhz = clock_table->DfPstateTable[max_pstate].MemClk;
+		bw_params->clk_table.entries[i].voltage = clock_table->DfPstateTable[max_pstate].Voltage;
+		bw_params->clk_table.entries[i].dcfclk_mhz = find_max_clk_value(clock_table->DcfClocks, NUM_DCFCLK_DPM_LEVELS);
+		bw_params->clk_table.entries[i].socclk_mhz = find_max_clk_value(clock_table->SocClocks, NUM_SOCCLK_DPM_LEVELS);
+		bw_params->clk_table.entries[i].dispclk_mhz = max_dispclk;
+		bw_params->clk_table.entries[i].dppclk_mhz = max_dppclk;
+		bw_params->clk_table.entries[i].wck_ratio = convert_wck_ratio(
+			clock_table->DfPstateTable[max_pstate].WckRatio);
+		i++;
+	}
+	bw_params->clk_table.num_entries = i--;
+
+	/* Make sure all highest clocks are included*/
+	bw_params->clk_table.entries[i].socclk_mhz = find_max_clk_value(clock_table->SocClocks, NUM_SOCCLK_DPM_LEVELS);
+	bw_params->clk_table.entries[i].dispclk_mhz = find_max_clk_value(clock_table->DispClocks, NUM_DISPCLK_DPM_LEVELS);
+	bw_params->clk_table.entries[i].dppclk_mhz = find_max_clk_value(clock_table->DppClocks, NUM_DPPCLK_DPM_LEVELS);
+	ASSERT(clock_table->DcfClocks[i] == find_max_clk_value(clock_table->DcfClocks, NUM_DCFCLK_DPM_LEVELS));
+	bw_params->clk_table.entries[i].phyclk_mhz = def_max.phyclk_mhz;
+	bw_params->clk_table.entries[i].phyclk_d18_mhz = def_max.phyclk_d18_mhz;
+	bw_params->clk_table.entries[i].dtbclk_mhz = def_max.dtbclk_mhz;
+
+	/*
+	 * Set any 0 clocks to max default setting. Not an issue for
+	 * power since we aren't doing switching in such case anyway
+	 */
+	for (i = 0; i < bw_params->clk_table.num_entries; i++) {
+		if (!bw_params->clk_table.entries[i].fclk_mhz) {
+			bw_params->clk_table.entries[i].fclk_mhz = def_max.fclk_mhz;
+			bw_params->clk_table.entries[i].memclk_mhz = def_max.memclk_mhz;
+			bw_params->clk_table.entries[i].voltage = def_max.voltage;
+		}
+		if (!bw_params->clk_table.entries[i].dcfclk_mhz)
+			bw_params->clk_table.entries[i].dcfclk_mhz = def_max.dcfclk_mhz;
+		if (!bw_params->clk_table.entries[i].socclk_mhz)
+			bw_params->clk_table.entries[i].socclk_mhz = def_max.socclk_mhz;
+		if (!bw_params->clk_table.entries[i].dispclk_mhz)
+			bw_params->clk_table.entries[i].dispclk_mhz = def_max.dispclk_mhz;
+		if (!bw_params->clk_table.entries[i].dppclk_mhz)
+			bw_params->clk_table.entries[i].dppclk_mhz = def_max.dppclk_mhz;
+		if (!bw_params->clk_table.entries[i].phyclk_mhz)
+			bw_params->clk_table.entries[i].phyclk_mhz = def_max.phyclk_mhz;
+		if (!bw_params->clk_table.entries[i].phyclk_d18_mhz)
+			bw_params->clk_table.entries[i].phyclk_d18_mhz = def_max.phyclk_d18_mhz;
+		if (!bw_params->clk_table.entries[i].dtbclk_mhz)
+			bw_params->clk_table.entries[i].dtbclk_mhz = def_max.dtbclk_mhz;
+	}
+	ASSERT(bw_params->clk_table.entries[i-1].dcfclk_mhz);
+	bw_params->vram_type = bios_info->memory_type;
+
+	bw_params->dram_channel_width_bytes = bios_info->memory_type == 0x22 ? 8 : 4;
+	bw_params->num_channels = bios_info->ma_channel_number ? bios_info->ma_channel_number : 4;
+
+	for (i = 0; i < WM_SET_COUNT; i++) {
+		bw_params->wm_table.entries[i].wm_inst = i;
+
+		if (i >= bw_params->clk_table.num_entries) {
+			bw_params->wm_table.entries[i].valid = false;
+			continue;
+		}
+
+		bw_params->wm_table.entries[i].wm_type = WM_TYPE_PSTATE_CHG;
+		bw_params->wm_table.entries[i].valid = true;
+	}
+}
+
+static struct clk_mgr_funcs dcn314_funcs = {
+	.get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz,
+	.get_dtb_ref_clk_frequency = dcn31_get_dtb_ref_freq_khz,
+	.update_clocks = dcn314_update_clocks,
+	.init_clocks = dcn314_init_clocks,
+	.enable_pme_wa = dcn314_enable_pme_wa,
+	.are_clock_states_equal = dcn314_are_clock_states_equal,
+	.notify_wm_ranges = dcn314_notify_wm_ranges
+};
+extern struct clk_mgr_funcs dcn3_fpga_funcs;
+
+static void dcn314_read_ss_info_from_lut(struct clk_mgr_internal *clk_mgr)
+{
+	uint32_t clock_source;
+	//uint32_t ssc_enable;
+
+	REG_GET(CLK1_CLK2_BYPASS_CNTL, CLK2_BYPASS_SEL, &clock_source);
+	//REG_GET(CLK6_0_CLK6_spll_field_8, spll_ssc_en, &ssc_enable);
+
+	if (dcn314_is_spll_ssc_enabled(&clk_mgr->base) && (clock_source < ARRAY_SIZE(ss_info_table.ss_percentage))) {
+		clk_mgr->dprefclk_ss_percentage = ss_info_table.ss_percentage[clock_source];
+
+		if (clk_mgr->dprefclk_ss_percentage != 0) {
+			clk_mgr->ss_on_dprefclk = true;
+			clk_mgr->dprefclk_ss_divider = ss_info_table.ss_divider;
+		}
+	}
+}
+
+void dcn314_clk_mgr_construct(
+		struct dc_context *ctx,
+		struct clk_mgr_dcn314 *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg)
+{
+	struct dcn314_smu_dpm_clks smu_dpm_clks = { 0 };
+	struct clk_log_info log_info = {0};
+
+	clk_mgr->base.base.ctx = ctx;
+	clk_mgr->base.base.funcs = &dcn314_funcs;
+
+	clk_mgr->base.pp_smu = pp_smu;
+
+	clk_mgr->base.dccg = dccg;
+	clk_mgr->base.dfs_bypass_disp_clk = 0;
+
+	clk_mgr->base.dprefclk_ss_percentage = 0;
+	clk_mgr->base.dprefclk_ss_divider = 1000;
+	clk_mgr->base.ss_on_dprefclk = false;
+	clk_mgr->base.dfs_ref_freq_khz = 48000;
+
+	clk_mgr->smu_wm_set.wm_set = (struct dcn314_watermarks *)dm_helpers_allocate_gpu_mem(
+				clk_mgr->base.base.ctx,
+				DC_MEM_ALLOC_TYPE_FRAME_BUFFER,
+				sizeof(struct dcn314_watermarks),
+				&clk_mgr->smu_wm_set.mc_address.quad_part);
+
+	if (!clk_mgr->smu_wm_set.wm_set) {
+		clk_mgr->smu_wm_set.wm_set = &dummy_wms;
+		clk_mgr->smu_wm_set.mc_address.quad_part = 0;
+	}
+	ASSERT(clk_mgr->smu_wm_set.wm_set);
+
+	smu_dpm_clks.dpm_clks = (DpmClocks314_t *)dm_helpers_allocate_gpu_mem(
+				clk_mgr->base.base.ctx,
+				DC_MEM_ALLOC_TYPE_FRAME_BUFFER,
+				sizeof(DpmClocks314_t),
+				&smu_dpm_clks.mc_address.quad_part);
+
+	if (smu_dpm_clks.dpm_clks == NULL) {
+		smu_dpm_clks.dpm_clks = &dummy_clocks;
+		smu_dpm_clks.mc_address.quad_part = 0;
+	}
+
+	ASSERT(smu_dpm_clks.dpm_clks);
+
+	clk_mgr->base.smu_ver = dcn314_smu_get_smu_version(&clk_mgr->base);
+
+	if (clk_mgr->base.smu_ver)
+		clk_mgr->base.smu_present = true;
+
+	/* TODO: Check we get what we expect during bringup */
+	clk_mgr->base.base.dentist_vco_freq_khz = get_vco_frequency_from_reg(&clk_mgr->base);
+
+	if (ctx->dc_bios->integrated_info->memory_type == LpDdr5MemType)
+		dcn314_bw_params.wm_table = lpddr5_wm_table;
+	else
+		dcn314_bw_params.wm_table = ddr5_wm_table;
+
+	/* Saved clocks configured at boot for debug purposes */
+	dcn314_dump_clk_registers(&clk_mgr->base.base.boot_snapshot,
+				  &clk_mgr->base.base, &log_info);
+
+	clk_mgr->base.base.dprefclk_khz = 600000;
+	clk_mgr->base.base.clks.ref_dtbclk_khz = 600000;
+	dce_clock_read_ss_info(&clk_mgr->base);
+	dcn314_read_ss_info_from_lut(&clk_mgr->base);
+	/*if bios enabled SS, driver needs to adjust dtb clock, only enable with correct bios*/
+
+	clk_mgr->base.base.bw_params = &dcn314_bw_params;
+
+	if (clk_mgr->base.base.ctx->dc->debug.pstate_enabled) {
+		int i;
+
+		dcn314_get_dpm_table_from_smu(&clk_mgr->base, &smu_dpm_clks);
+		DC_LOG_SMU("NumDcfClkLevelsEnabled: %d\n"
+				   "NumDispClkLevelsEnabled: %d\n"
+				   "NumSocClkLevelsEnabled: %d\n"
+				   "VcnClkLevelsEnabled: %d\n"
+				   "NumDfPst atesEnabled: %d\n"
+				   "MinGfxClk: %d\n"
+				   "MaxGfxClk: %d\n",
+				   smu_dpm_clks.dpm_clks->NumDcfClkLevelsEnabled,
+				   smu_dpm_clks.dpm_clks->NumDispClkLevelsEnabled,
+				   smu_dpm_clks.dpm_clks->NumSocClkLevelsEnabled,
+				   smu_dpm_clks.dpm_clks->VcnClkLevelsEnabled,
+				   smu_dpm_clks.dpm_clks->NumDfPstatesEnabled,
+				   smu_dpm_clks.dpm_clks->MinGfxClk,
+				   smu_dpm_clks.dpm_clks->MaxGfxClk);
+		for (i = 0; i < smu_dpm_clks.dpm_clks->NumDcfClkLevelsEnabled; i++) {
+			DC_LOG_SMU("smu_dpm_clks.dpm_clks->DcfClocks[%d] = %d\n",
+					   i,
+					   smu_dpm_clks.dpm_clks->DcfClocks[i]);
+		}
+		for (i = 0; i < smu_dpm_clks.dpm_clks->NumDispClkLevelsEnabled; i++) {
+			DC_LOG_SMU("smu_dpm_clks.dpm_clks->DispClocks[%d] = %d\n",
+					   i, smu_dpm_clks.dpm_clks->DispClocks[i]);
+		}
+		for (i = 0; i < smu_dpm_clks.dpm_clks->NumSocClkLevelsEnabled; i++) {
+			DC_LOG_SMU("smu_dpm_clks.dpm_clks->SocClocks[%d] = %d\n",
+					   i, smu_dpm_clks.dpm_clks->SocClocks[i]);
+		}
+		for (i = 0; i < NUM_SOC_VOLTAGE_LEVELS; i++)
+			DC_LOG_SMU("smu_dpm_clks.dpm_clks->SocVoltage[%d] = %d\n",
+					   i, smu_dpm_clks.dpm_clks->SocVoltage[i]);
+
+		for (i = 0; i < NUM_DF_PSTATE_LEVELS; i++) {
+			DC_LOG_SMU("smu_dpm_clks.dpm_clks.DfPstateTable[%d].FClk = %d\n"
+					   "smu_dpm_clks.dpm_clks->DfPstateTable[%d].MemClk= %d\n"
+					   "smu_dpm_clks.dpm_clks->DfPstateTable[%d].Voltage = %d\n",
+					   i, smu_dpm_clks.dpm_clks->DfPstateTable[i].FClk,
+					   i, smu_dpm_clks.dpm_clks->DfPstateTable[i].MemClk,
+					   i, smu_dpm_clks.dpm_clks->DfPstateTable[i].Voltage);
+		}
+
+		if (ctx->dc_bios->integrated_info && ctx->dc->config.use_default_clock_table == false) {
+			dcn314_clk_mgr_helper_populate_bw_params(
+					&clk_mgr->base,
+					ctx->dc_bios->integrated_info,
+					smu_dpm_clks.dpm_clks);
+		}
+	}
+
+	if (smu_dpm_clks.dpm_clks && smu_dpm_clks.mc_address.quad_part != 0)
+		dm_helpers_free_gpu_mem(clk_mgr->base.base.ctx, DC_MEM_ALLOC_TYPE_FRAME_BUFFER,
+				smu_dpm_clks.dpm_clks);
+}
+
+void dcn314_clk_mgr_destroy(struct clk_mgr_internal *clk_mgr_int)
+{
+	struct clk_mgr_dcn314 *clk_mgr = TO_CLK_MGR_DCN314(clk_mgr_int);
+
+	if (clk_mgr->smu_wm_set.wm_set && clk_mgr->smu_wm_set.mc_address.quad_part != 0)
+		dm_helpers_free_gpu_mem(clk_mgr_int->base.ctx, DC_MEM_ALLOC_TYPE_FRAME_BUFFER,
+				clk_mgr->smu_wm_set.wm_set);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_clk_mgr.h
new file mode 100644
index 000000000000..0577eb527bc3
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_clk_mgr.h
@@ -0,0 +1,73 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCN314_CLK_MGR_H__
+#define __DCN314_CLK_MGR_H__
+#include "clk_mgr_internal.h"
+
+#define DCN314_NUM_CLOCK_SOURCES   5
+
+struct dcn314_watermarks;
+
+struct dcn314_smu_watermark_set {
+	struct dcn314_watermarks *wm_set;
+	union large_integer mc_address;
+};
+
+struct clk_mgr_dcn314 {
+	struct clk_mgr_internal base;
+	struct dcn314_smu_watermark_set smu_wm_set;
+};
+
+struct dcn314_ss_info_table {
+	uint32_t ss_divider;
+	uint32_t ss_percentage[DCN314_NUM_CLOCK_SOURCES];
+};
+
+bool dcn314_are_clock_states_equal(struct dc_clocks *a,
+		struct dc_clocks *b);
+
+bool dcn314_is_spll_ssc_enabled(struct clk_mgr *clk_mgr_base);
+
+void dcn314_init_clocks(struct clk_mgr *clk_mgr);
+
+void dcn314_update_clocks(struct clk_mgr *clk_mgr_base,
+			struct dc_state *context,
+			bool safe_to_lower);
+
+void dcn314_clk_mgr_construct(struct dc_context *ctx,
+		struct clk_mgr_dcn314 *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg);
+
+void dcn314_clk_mgr_destroy(struct clk_mgr_internal *clk_mgr_int);
+
+
+void dcn314_dump_clk_registers(struct clk_state_registers_and_bypass *regs_and_bypass,
+		struct clk_mgr *clk_mgr_base, struct clk_log_info *log_info);
+
+
+#endif //__DCN314_CLK_MGR_H__
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_smu.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_smu.c
new file mode 100644
index 000000000000..c4af406146b7
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_smu.c
@@ -0,0 +1,398 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+
+#include "core_types.h"
+#include "clk_mgr_internal.h"
+#include "reg_helper.h"
+#include "dm_helpers.h"
+#include "dcn314_smu.h"
+
+#include "mp/mp_13_0_5_offset.h"
+
+/* TODO: Use the real headers when they're correct */
+#define MP1_BASE__INST0_SEG0                       0x00016000
+#define MP1_BASE__INST0_SEG1                       0x0243FC00
+#define MP1_BASE__INST0_SEG2                       0x00DC0000
+#define MP1_BASE__INST0_SEG3                       0x00E00000
+#define MP1_BASE__INST0_SEG4                       0x00E40000
+#define MP1_BASE__INST0_SEG5                       0
+
+#ifdef BASE_INNER
+#undef BASE_INNER
+#endif
+
+#define BASE_INNER(seg) MP1_BASE__INST0_SEG ## seg
+
+#define BASE(seg) BASE_INNER(seg)
+
+#define REG(reg_name) (BASE(reg##reg_name##_BASE_IDX) + reg##reg_name)
+
+#define FN(reg_name, field) \
+	FD(reg_name##__##field)
+
+#include "logger_types.h"
+#undef DC_LOGGER
+#define DC_LOGGER \
+	CTX->logger
+#define smu_print(str, ...) {DC_LOG_SMU(str, ##__VA_ARGS__); }
+
+#define VBIOSSMC_MSG_TestMessage                  0x1
+#define VBIOSSMC_MSG_GetSmuVersion                0x2
+#define VBIOSSMC_MSG_PowerUpGfx                   0x3
+#define VBIOSSMC_MSG_SetDispclkFreq               0x4
+#define VBIOSSMC_MSG_SetDprefclkFreq              0x5   //Not used. DPRef is constant
+#define VBIOSSMC_MSG_SetDppclkFreq                0x6
+#define VBIOSSMC_MSG_SetHardMinDcfclkByFreq       0x7
+#define VBIOSSMC_MSG_SetMinDeepSleepDcfclk        0x8
+#define VBIOSSMC_MSG_SetPhyclkVoltageByFreq       0x9	//Keep it in case VMIN dees not support phy clk
+#define VBIOSSMC_MSG_GetFclkFrequency             0xA
+#define VBIOSSMC_MSG_SetDisplayCount              0xB   //Not used anymore
+#define VBIOSSMC_MSG_EnableTmdp48MHzRefclkPwrDown 0xC   //Not used anymore
+#define VBIOSSMC_MSG_UpdatePmeRestore             0xD
+#define VBIOSSMC_MSG_SetVbiosDramAddrHigh         0xE   //Used for WM table txfr
+#define VBIOSSMC_MSG_SetVbiosDramAddrLow          0xF
+#define VBIOSSMC_MSG_TransferTableSmu2Dram        0x10
+#define VBIOSSMC_MSG_TransferTableDram2Smu        0x11
+#define VBIOSSMC_MSG_SetDisplayIdleOptimizations  0x12
+#define VBIOSSMC_MSG_GetDprefclkFreq              0x13
+#define VBIOSSMC_MSG_GetDtbclkFreq                0x14
+#define VBIOSSMC_MSG_AllowZstatesEntry            0x15
+#define VBIOSSMC_MSG_DisallowZstatesEntry	  0x16
+#define VBIOSSMC_MSG_SetDtbClk                    0x17
+#define VBIOSSMC_Message_Count                    0x18
+
+#define VBIOSSMC_Status_BUSY                      0x0
+#define VBIOSSMC_Result_OK                        0x1
+#define VBIOSSMC_Result_Failed                    0xFF
+#define VBIOSSMC_Result_UnknownCmd                0xFE
+#define VBIOSSMC_Result_CmdRejectedPrereq         0xFD
+#define VBIOSSMC_Result_CmdRejectedBusy           0xFC
+
+/*
+ * Function to be used instead of REG_WAIT macro because the wait ends when
+ * the register is NOT EQUAL to zero, and because the translation in msg_if.h
+ * won't work with REG_WAIT.
+ */
+static uint32_t dcn314_smu_wait_for_response(struct clk_mgr_internal *clk_mgr, unsigned int delay_us, unsigned int max_retries)
+{
+	uint32_t res_val = VBIOSSMC_Status_BUSY;
+
+	do {
+		res_val = REG_READ(MP1_SMN_C2PMSG_91);
+		if (res_val != VBIOSSMC_Status_BUSY)
+			break;
+
+		if (delay_us >= 1000)
+			msleep(delay_us/1000);
+		else if (delay_us > 0)
+			udelay(delay_us);
+	} while (max_retries--);
+
+	return res_val;
+}
+
+static int dcn314_smu_send_msg_with_param(struct clk_mgr_internal *clk_mgr,
+					 unsigned int msg_id,
+					 unsigned int param)
+{
+	uint32_t result;
+
+	result = dcn314_smu_wait_for_response(clk_mgr, 10, 200000);
+
+	if (result != VBIOSSMC_Result_OK)
+		smu_print("SMU Response was not OK. SMU response after wait received is: %d\n",
+				result);
+
+	if (result == VBIOSSMC_Status_BUSY)
+		return -1;
+
+	/* First clear response register */
+	REG_WRITE(MP1_SMN_C2PMSG_91, VBIOSSMC_Status_BUSY);
+
+	/* Set the parameter register for the SMU message, unit is Mhz */
+	REG_WRITE(MP1_SMN_C2PMSG_83, param);
+
+	/* Trigger the message transaction by writing the message ID */
+	REG_WRITE(MP1_SMN_C2PMSG_67, msg_id);
+
+	result = dcn314_smu_wait_for_response(clk_mgr, 10, 200000);
+
+	if (result == VBIOSSMC_Result_Failed) {
+		if (msg_id == VBIOSSMC_MSG_TransferTableDram2Smu &&
+		    param == TABLE_WATERMARKS)
+			DC_LOG_DEBUG("Watermarks table not configured properly by SMU");
+		else if (msg_id == VBIOSSMC_MSG_SetHardMinDcfclkByFreq ||
+			 msg_id == VBIOSSMC_MSG_SetMinDeepSleepDcfclk)
+			DC_LOG_WARNING("DCFCLK_DPM is not enabled by BIOS");
+		else
+			ASSERT(0);
+		REG_WRITE(MP1_SMN_C2PMSG_91, VBIOSSMC_Result_OK);
+		return -1;
+	}
+
+	if (IS_SMU_TIMEOUT(result)) {
+		ASSERT(0);
+		dm_helpers_smu_timeout(CTX, msg_id, param, 10 * 200000);
+	}
+
+	return REG_READ(MP1_SMN_C2PMSG_83);
+}
+
+int dcn314_smu_get_smu_version(struct clk_mgr_internal *clk_mgr)
+{
+	return dcn314_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_GetSmuVersion,
+			0);
+}
+
+
+int dcn314_smu_set_dispclk(struct clk_mgr_internal *clk_mgr, int requested_dispclk_khz)
+{
+	int actual_dispclk_set_mhz = -1;
+
+	if (!clk_mgr->smu_present)
+		return requested_dispclk_khz;
+
+	/*  Unit of SMU msg parameter is Mhz */
+	actual_dispclk_set_mhz = dcn314_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDispclkFreq,
+			khz_to_mhz_ceil(requested_dispclk_khz));
+
+	return actual_dispclk_set_mhz * 1000;
+}
+
+int dcn314_smu_set_dprefclk(struct clk_mgr_internal *clk_mgr)
+{
+	int actual_dprefclk_set_mhz = -1;
+
+	if (!clk_mgr->smu_present)
+		return clk_mgr->base.dprefclk_khz;
+
+	actual_dprefclk_set_mhz = dcn314_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDprefclkFreq,
+			khz_to_mhz_ceil(clk_mgr->base.dprefclk_khz));
+
+	/* TODO: add code for programing DP DTO, currently this is down by command table */
+
+	return actual_dprefclk_set_mhz * 1000;
+}
+
+int dcn314_smu_set_hard_min_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_dcfclk_khz)
+{
+	int actual_dcfclk_set_mhz = -1;
+
+	if (!clk_mgr->base.ctx->dc->debug.pstate_enabled)
+		return -1;
+
+	if (!clk_mgr->smu_present)
+		return requested_dcfclk_khz;
+
+	actual_dcfclk_set_mhz = dcn314_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetHardMinDcfclkByFreq,
+			khz_to_mhz_ceil(requested_dcfclk_khz));
+
+	return actual_dcfclk_set_mhz * 1000;
+}
+
+int dcn314_smu_set_min_deep_sleep_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_min_ds_dcfclk_khz)
+{
+	int actual_min_ds_dcfclk_mhz = -1;
+
+	if (!clk_mgr->base.ctx->dc->debug.pstate_enabled)
+		return -1;
+
+	if (!clk_mgr->smu_present)
+		return requested_min_ds_dcfclk_khz;
+
+	actual_min_ds_dcfclk_mhz = dcn314_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetMinDeepSleepDcfclk,
+			khz_to_mhz_ceil(requested_min_ds_dcfclk_khz));
+
+	return actual_min_ds_dcfclk_mhz * 1000;
+}
+
+int dcn314_smu_set_dppclk(struct clk_mgr_internal *clk_mgr, int requested_dpp_khz)
+{
+	int actual_dppclk_set_mhz = -1;
+
+	if (!clk_mgr->smu_present)
+		return requested_dpp_khz;
+
+	actual_dppclk_set_mhz = dcn314_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDppclkFreq,
+			khz_to_mhz_ceil(requested_dpp_khz));
+
+	return actual_dppclk_set_mhz * 1000;
+}
+
+void dcn314_smu_set_display_idle_optimization(struct clk_mgr_internal *clk_mgr, uint32_t idle_info)
+{
+	if (!clk_mgr->base.ctx->dc->debug.pstate_enabled)
+		return;
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	//TODO: Work with smu team to define optimization options.
+	dcn314_smu_send_msg_with_param(
+		clk_mgr,
+		VBIOSSMC_MSG_SetDisplayIdleOptimizations,
+		idle_info);
+}
+
+void dcn314_smu_enable_phy_refclk_pwrdwn(struct clk_mgr_internal *clk_mgr, bool enable)
+{
+	union display_idle_optimization_u idle_info = { 0 };
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	if (enable) {
+		idle_info.idle_info.df_request_disabled = 1;
+		idle_info.idle_info.phy_ref_clk_off = 1;
+	}
+
+	dcn314_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDisplayIdleOptimizations,
+			idle_info.data);
+}
+
+void dcn314_smu_enable_pme_wa(struct clk_mgr_internal *clk_mgr)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn314_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_UpdatePmeRestore,
+			0);
+}
+
+void dcn314_smu_set_dram_addr_high(struct clk_mgr_internal *clk_mgr, uint32_t addr_high)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn314_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_SetVbiosDramAddrHigh, addr_high);
+}
+
+void dcn314_smu_set_dram_addr_low(struct clk_mgr_internal *clk_mgr, uint32_t addr_low)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn314_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_SetVbiosDramAddrLow, addr_low);
+}
+
+void dcn314_smu_transfer_dpm_table_smu_2_dram(struct clk_mgr_internal *clk_mgr)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn314_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_TransferTableSmu2Dram, TABLE_DPMCLOCKS);
+}
+
+void dcn314_smu_transfer_wm_table_dram_2_smu(struct clk_mgr_internal *clk_mgr)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn314_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_TransferTableDram2Smu, TABLE_WATERMARKS);
+}
+
+void dcn314_smu_set_zstate_support(struct clk_mgr_internal *clk_mgr, enum dcn_zstate_support_state support)
+{
+	unsigned int msg_id, param;
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	switch (support) {
+
+	case DCN_ZSTATE_SUPPORT_ALLOW:
+		msg_id = VBIOSSMC_MSG_AllowZstatesEntry;
+		param = (1 << 10) | (1 << 9) | (1 << 8);
+		break;
+
+	case DCN_ZSTATE_SUPPORT_DISALLOW:
+		msg_id = VBIOSSMC_MSG_AllowZstatesEntry;
+		param = 0;
+		break;
+
+
+	case DCN_ZSTATE_SUPPORT_ALLOW_Z10_ONLY:
+		msg_id = VBIOSSMC_MSG_AllowZstatesEntry;
+		param = (1 << 10);
+		break;
+
+	case DCN_ZSTATE_SUPPORT_ALLOW_Z8_Z10_ONLY:
+		msg_id = VBIOSSMC_MSG_AllowZstatesEntry;
+		param = (1 << 10) | (1 << 8);
+		break;
+
+	case DCN_ZSTATE_SUPPORT_ALLOW_Z8_ONLY:
+		msg_id = VBIOSSMC_MSG_AllowZstatesEntry;
+		param = (1 << 8);
+		break;
+
+	default: //DCN_ZSTATE_SUPPORT_UNKNOWN
+		msg_id = VBIOSSMC_MSG_AllowZstatesEntry;
+		param = 0;
+		break;
+	}
+
+
+	dcn314_smu_send_msg_with_param(
+		clk_mgr,
+		msg_id,
+		param);
+
+}
+
+/* Arg = 1: Turn DTB on; 0: Turn DTB CLK OFF. when it is on, it is 600MHZ */
+void dcn314_smu_set_dtbclk(struct clk_mgr_internal *clk_mgr, bool enable)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn314_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDtbClk,
+			enable);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_smu.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_smu.h
new file mode 100644
index 000000000000..78ca1e5c5e9e
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_smu.h
@@ -0,0 +1,110 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DAL_DC_314_SMU_H_
+#define DAL_DC_314_SMU_H_
+
+#include "smu13_driver_if_v13_0_4.h"
+
+typedef enum {
+	WCK_RATIO_1_1 = 0,  // DDR5, Wck:ck is always 1:1;
+	WCK_RATIO_1_2,
+	WCK_RATIO_1_4,
+	WCK_RATIO_MAX
+} WCK_RATIO_e;
+
+typedef struct {
+	uint32_t FClk;
+	uint32_t MemClk;
+	uint32_t Voltage;
+	uint8_t  WckRatio;
+	uint8_t  Spare[3];
+} DfPstateTable314_t;
+
+//Freq in MHz
+//Voltage in milli volts with 2 fractional bits
+typedef struct {
+	uint32_t DcfClocks[NUM_DCFCLK_DPM_LEVELS];
+	uint32_t DispClocks[NUM_DISPCLK_DPM_LEVELS];
+	uint32_t DppClocks[NUM_DPPCLK_DPM_LEVELS];
+	uint32_t SocClocks[NUM_SOCCLK_DPM_LEVELS];
+	uint32_t VClocks[NUM_VCN_DPM_LEVELS];
+	uint32_t DClocks[NUM_VCN_DPM_LEVELS];
+	uint32_t SocVoltage[NUM_SOC_VOLTAGE_LEVELS];
+	DfPstateTable314_t DfPstateTable[NUM_DF_PSTATE_LEVELS];
+
+	uint8_t  NumDcfClkLevelsEnabled;
+	uint8_t  NumDispClkLevelsEnabled; //Applies to both Dispclk and Dppclk
+	uint8_t  NumSocClkLevelsEnabled;
+	uint8_t  VcnClkLevelsEnabled;     //Applies to both Vclk and Dclk
+	uint8_t  NumDfPstatesEnabled;
+	uint8_t  spare[3];
+
+	uint32_t MinGfxClk;
+	uint32_t MaxGfxClk;
+} DpmClocks314_t;
+
+struct dcn314_watermarks {
+	// Watermarks
+	WatermarkRowGeneric_t WatermarkRow[WM_COUNT][NUM_WM_RANGES];
+	uint32_t MmHubPadding[7]; // SMU internal use
+};
+
+struct dcn314_smu_dpm_clks {
+	DpmClocks314_t *dpm_clks;
+	union large_integer mc_address;
+};
+
+struct display_idle_optimization {
+	unsigned int df_request_disabled : 1;
+	unsigned int phy_ref_clk_off     : 1;
+	unsigned int s0i2_rdy            : 1;
+	unsigned int reserved            : 29;
+};
+
+union display_idle_optimization_u {
+	struct display_idle_optimization idle_info;
+	uint32_t data;
+};
+
+int dcn314_smu_get_smu_version(struct clk_mgr_internal *clk_mgr);
+int dcn314_smu_set_dispclk(struct clk_mgr_internal *clk_mgr, int requested_dispclk_khz);
+int dcn314_smu_set_dprefclk(struct clk_mgr_internal *clk_mgr);
+int dcn314_smu_set_hard_min_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_dcfclk_khz);
+int dcn314_smu_set_min_deep_sleep_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_min_ds_dcfclk_khz);
+int dcn314_smu_set_dppclk(struct clk_mgr_internal *clk_mgr, int requested_dpp_khz);
+void dcn314_smu_set_display_idle_optimization(struct clk_mgr_internal *clk_mgr, uint32_t idle_info);
+void dcn314_smu_enable_phy_refclk_pwrdwn(struct clk_mgr_internal *clk_mgr, bool enable);
+void dcn314_smu_enable_pme_wa(struct clk_mgr_internal *clk_mgr);
+void dcn314_smu_set_dram_addr_high(struct clk_mgr_internal *clk_mgr, uint32_t addr_high);
+void dcn314_smu_set_dram_addr_low(struct clk_mgr_internal *clk_mgr, uint32_t addr_low);
+void dcn314_smu_transfer_dpm_table_smu_2_dram(struct clk_mgr_internal *clk_mgr);
+void dcn314_smu_transfer_wm_table_dram_2_smu(struct clk_mgr_internal *clk_mgr);
+
+void dcn314_smu_set_zstate_support(struct clk_mgr_internal *clk_mgr, enum dcn_zstate_support_state support);
+void dcn314_smu_set_dtbclk(struct clk_mgr_internal *clk_mgr, bool enable);
+
+#endif /* DAL_DC_314_SMU_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.c
new file mode 100644
index 000000000000..b315ed91e010
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.c
@@ -0,0 +1,737 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+
+#include "dccg.h"
+#include "clk_mgr_internal.h"
+
+// For dce12_get_dp_ref_freq_khz
+#include "dce100/dce_clk_mgr.h"
+// For dcn20_update_clocks_update_dpp_dto
+#include "dcn20/dcn20_clk_mgr.h"
+#include "dcn31/dcn31_clk_mgr.h"
+#include "dcn315_clk_mgr.h"
+
+#include "core_types.h"
+#include "dcn315_smu.h"
+#include "dm_helpers.h"
+
+#include "dc_dmub_srv.h"
+
+#include "logger_types.h"
+#undef DC_LOGGER
+#define DC_LOGGER \
+	clk_mgr->base.base.ctx->logger
+
+#include "link_service.h"
+
+#define TO_CLK_MGR_DCN315(clk_mgr)\
+	container_of(clk_mgr, struct clk_mgr_dcn315, base)
+
+#define UNSUPPORTED_DCFCLK 10000000
+#define MIN_DPP_DISP_CLK     100000
+
+static int dcn315_get_active_display_cnt_wa(
+		struct dc *dc,
+		struct dc_state *context)
+{
+	int i, display_count;
+	bool tmds_present = false;
+
+	display_count = 0;
+	for (i = 0; i < context->stream_count; i++) {
+		const struct dc_stream_state *stream = context->streams[i];
+
+		if (stream->signal == SIGNAL_TYPE_HDMI_TYPE_A ||
+				stream->signal == SIGNAL_TYPE_DVI_SINGLE_LINK ||
+				stream->signal == SIGNAL_TYPE_DVI_DUAL_LINK)
+			tmds_present = true;
+	}
+
+	for (i = 0; i < dc->link_count; i++) {
+		const struct dc_link *link = dc->links[i];
+
+		/* abusing the fact that the dig and phy are coupled to see if the phy is enabled */
+		if (link->link_enc && link->link_enc->funcs->is_dig_enabled &&
+				link->link_enc->funcs->is_dig_enabled(link->link_enc))
+			display_count++;
+	}
+
+	/* WA for hang on HDMI after display off back back on*/
+	if (display_count == 0 && tmds_present)
+		display_count = 1;
+
+	return display_count;
+}
+
+static bool should_disable_otg(struct pipe_ctx *pipe)
+{
+	bool ret = true;
+
+	if (pipe->stream->link->link_enc && pipe->stream->link->link_enc->funcs->is_dig_enabled &&
+			pipe->stream->link->link_enc->funcs->is_dig_enabled(pipe->stream->link->link_enc))
+		ret = false;
+	return ret;
+}
+
+static void dcn315_disable_otg_wa(struct clk_mgr *clk_mgr_base, struct dc_state *context, bool disable)
+{
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	int i;
+
+	for (i = 0; i < dc->res_pool->pipe_count; ++i) {
+		struct pipe_ctx *pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+
+		if (pipe->top_pipe || pipe->prev_odm_pipe)
+			continue;
+		if (pipe->stream && (pipe->stream->dpms_off || pipe->plane_state == NULL ||
+					dc_is_virtual_signal(pipe->stream->signal))) {
+
+			/* This w/a should not trigger when we have a dig active */
+			if (should_disable_otg(pipe)) {
+				if (disable) {
+					pipe->stream_res.tg->funcs->immediate_disable_crtc(pipe->stream_res.tg);
+					reset_sync_context_for_pipe(dc, context, i);
+				} else
+					pipe->stream_res.tg->funcs->enable_crtc(pipe->stream_res.tg);
+			}
+		}
+	}
+}
+
+static void dcn315_update_clocks(struct clk_mgr *clk_mgr_base,
+			struct dc_state *context,
+			bool safe_to_lower)
+{
+	union dmub_rb_cmd cmd;
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dc_clocks *new_clocks = &context->bw_ctx.bw.dcn.clk;
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	int display_count = 0;
+	bool update_dppclk = false;
+	bool update_dispclk = false;
+	bool dpp_clock_lowered = false;
+
+	if (dc->work_arounds.skip_clock_update)
+		return;
+
+	clk_mgr_base->clks.zstate_support = new_clocks->zstate_support;
+	/*
+	 * if it is safe to lower, but we are already in the lower state, we don't have to do anything
+	 * also if safe to lower is false, we just go in the higher state
+	 */
+	clk_mgr_base->clks.zstate_support = new_clocks->zstate_support;
+	if (safe_to_lower) {
+		if (clk_mgr_base->clks.dtbclk_en && !new_clocks->dtbclk_en) {
+			dcn315_smu_set_dtbclk(clk_mgr, false);
+			clk_mgr_base->clks.dtbclk_en = new_clocks->dtbclk_en;
+		}
+		/* check that we're not already in lower */
+		if (clk_mgr_base->clks.pwr_state != DCN_PWR_STATE_LOW_POWER) {
+			display_count = dcn315_get_active_display_cnt_wa(dc, context);
+			/* if we can go lower, go lower */
+			if (display_count == 0) {
+				union display_idle_optimization_u idle_info = { 0 };
+				idle_info.idle_info.df_request_disabled = 1;
+				idle_info.idle_info.phy_ref_clk_off = 1;
+				idle_info.idle_info.s0i2_rdy = 1;
+				dcn315_smu_set_display_idle_optimization(clk_mgr, idle_info.data);
+				/* update power state */
+				clk_mgr_base->clks.pwr_state = DCN_PWR_STATE_LOW_POWER;
+			}
+		}
+	} else {
+		if (!clk_mgr_base->clks.dtbclk_en && new_clocks->dtbclk_en) {
+			dcn315_smu_set_dtbclk(clk_mgr, true);
+			clk_mgr_base->clks.dtbclk_en = new_clocks->dtbclk_en;
+		}
+		/* check that we're not already in D0 */
+		if (clk_mgr_base->clks.pwr_state != DCN_PWR_STATE_MISSION_MODE) {
+			union display_idle_optimization_u idle_info = { 0 };
+			dcn315_smu_set_display_idle_optimization(clk_mgr, idle_info.data);
+			/* update power state */
+			clk_mgr_base->clks.pwr_state = DCN_PWR_STATE_MISSION_MODE;
+		}
+	}
+
+	/* Lock pstate by requesting unsupported dcfclk if change is unsupported */
+	if (!new_clocks->p_state_change_support)
+		new_clocks->dcfclk_khz = UNSUPPORTED_DCFCLK;
+	if (should_set_clock(safe_to_lower, new_clocks->dcfclk_khz, clk_mgr_base->clks.dcfclk_khz)) {
+		clk_mgr_base->clks.dcfclk_khz = new_clocks->dcfclk_khz;
+		dcn315_smu_set_hard_min_dcfclk(clk_mgr, clk_mgr_base->clks.dcfclk_khz);
+	}
+
+	if (should_set_clock(safe_to_lower,
+			new_clocks->dcfclk_deep_sleep_khz, clk_mgr_base->clks.dcfclk_deep_sleep_khz)) {
+		clk_mgr_base->clks.dcfclk_deep_sleep_khz = new_clocks->dcfclk_deep_sleep_khz;
+		dcn315_smu_set_min_deep_sleep_dcfclk(clk_mgr, clk_mgr_base->clks.dcfclk_deep_sleep_khz);
+	}
+
+	// workaround: Limit dppclk to 100Mhz to avoid lower eDP panel switch to plus 4K monitor underflow.
+	if (new_clocks->dppclk_khz < MIN_DPP_DISP_CLK)
+		new_clocks->dppclk_khz = MIN_DPP_DISP_CLK;
+
+	if (should_set_clock(safe_to_lower, new_clocks->dppclk_khz, clk_mgr->base.clks.dppclk_khz)) {
+		if (clk_mgr->base.clks.dppclk_khz > new_clocks->dppclk_khz)
+			dpp_clock_lowered = true;
+		clk_mgr_base->clks.dppclk_khz = new_clocks->dppclk_khz;
+		update_dppclk = true;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, clk_mgr_base->clks.dispclk_khz) &&
+	    (new_clocks->dispclk_khz > 0 || (safe_to_lower && display_count == 0))) {
+		int requested_dispclk_khz = new_clocks->dispclk_khz;
+
+		dcn315_disable_otg_wa(clk_mgr_base, context, true);
+
+		/* Clamp the requested clock to PMFW based on their limit. */
+		if (dc->debug.min_disp_clk_khz > 0 && requested_dispclk_khz < dc->debug.min_disp_clk_khz)
+			requested_dispclk_khz = dc->debug.min_disp_clk_khz;
+
+		dcn315_smu_set_dispclk(clk_mgr, requested_dispclk_khz);
+		clk_mgr_base->clks.dispclk_khz = new_clocks->dispclk_khz;
+		dcn315_disable_otg_wa(clk_mgr_base, context, false);
+
+		update_dispclk = true;
+	}
+
+	if (dpp_clock_lowered) {
+		// increase per DPP DTO before lowering global dppclk
+		dcn20_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
+		dcn315_smu_set_dppclk(clk_mgr, clk_mgr_base->clks.dppclk_khz);
+	} else {
+		// increase global DPPCLK before lowering per DPP DTO
+		if (update_dppclk || update_dispclk)
+			dcn315_smu_set_dppclk(clk_mgr, clk_mgr_base->clks.dppclk_khz);
+		// always update dtos unless clock is lowered and not safe to lower
+		if (new_clocks->dppclk_khz >= dc->current_state->bw_ctx.bw.dcn.clk.dppclk_khz)
+			dcn20_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
+	}
+
+	// notify DMCUB of latest clocks
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.notify_clocks.header.type = DMUB_CMD__CLK_MGR;
+	cmd.notify_clocks.header.sub_type = DMUB_CMD__CLK_MGR_NOTIFY_CLOCKS;
+	cmd.notify_clocks.clocks.dcfclk_khz = clk_mgr_base->clks.dcfclk_khz;
+	cmd.notify_clocks.clocks.dcfclk_deep_sleep_khz =
+		clk_mgr_base->clks.dcfclk_deep_sleep_khz;
+	cmd.notify_clocks.clocks.dispclk_khz = clk_mgr_base->clks.dispclk_khz;
+	cmd.notify_clocks.clocks.dppclk_khz = clk_mgr_base->clks.dppclk_khz;
+
+	dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+static void dcn315_dump_clk_registers(struct clk_state_registers_and_bypass *regs_and_bypass,
+		struct clk_mgr *clk_mgr_base, struct clk_log_info *log_info)
+{
+	return;
+}
+
+static struct clk_bw_params dcn315_bw_params = {
+	.vram_type = Ddr4MemType,
+	.num_channels = 2,
+	.clk_table = {
+		.entries = {
+			{
+				.voltage = 0,
+				.dispclk_mhz = 640,
+				.dppclk_mhz = 640,
+				.phyclk_mhz = 810,
+				.phyclk_d18_mhz = 667,
+				.dtbclk_mhz = 600,
+			},
+			{
+				.voltage = 1,
+				.dispclk_mhz = 739,
+				.dppclk_mhz = 739,
+				.phyclk_mhz = 810,
+				.phyclk_d18_mhz = 667,
+				.dtbclk_mhz = 600,
+			},
+			{
+				.voltage = 2,
+				.dispclk_mhz = 960,
+				.dppclk_mhz = 960,
+				.phyclk_mhz = 810,
+				.phyclk_d18_mhz = 667,
+				.dtbclk_mhz = 600,
+			},
+			{
+				.voltage = 3,
+				.dispclk_mhz = 1200,
+				.dppclk_mhz = 1200,
+				.phyclk_mhz = 810,
+				.phyclk_d18_mhz = 667,
+				.dtbclk_mhz = 600,
+			},
+			{
+				.voltage = 4,
+				.dispclk_mhz = 1372,
+				.dppclk_mhz = 1372,
+				.phyclk_mhz = 810,
+				.phyclk_d18_mhz = 667,
+				.dtbclk_mhz = 600,
+			},
+		},
+		.num_entries = 5,
+	},
+
+};
+
+static struct wm_table ddr5_wm_table = {
+	.entries = {
+		{
+			.wm_inst = WM_A,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 129.0,
+			.sr_exit_time_us = 11.5,
+			.sr_enter_plus_exit_time_us = 14.5,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_B,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 129.0,
+			.sr_exit_time_us = 11.5,
+			.sr_enter_plus_exit_time_us = 14.5,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_C,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 129.0,
+			.sr_exit_time_us = 11.5,
+			.sr_enter_plus_exit_time_us = 14.5,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_D,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 129.0,
+			.sr_exit_time_us = 11.5,
+			.sr_enter_plus_exit_time_us = 14.5,
+			.valid = true,
+		},
+	}
+};
+
+static struct wm_table lpddr5_wm_table = {
+	.entries = {
+		{
+			.wm_inst = WM_A,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 129.0,
+			.sr_exit_time_us = 11.5,
+			.sr_enter_plus_exit_time_us = 14.5,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_B,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 129.0,
+			.sr_exit_time_us = 11.5,
+			.sr_enter_plus_exit_time_us = 14.5,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_C,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 129.0,
+			.sr_exit_time_us = 11.5,
+			.sr_enter_plus_exit_time_us = 14.5,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_D,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 129.0,
+			.sr_exit_time_us = 11.5,
+			.sr_enter_plus_exit_time_us = 14.5,
+			.valid = true,
+		},
+	}
+};
+
+/* Temporary Place holder until we can get them from fuse */
+static DpmClocks_315_t dummy_clocks = { 0 };
+static struct dcn315_watermarks dummy_wms = { 0 };
+
+static void dcn315_build_watermark_ranges(struct clk_bw_params *bw_params, struct dcn315_watermarks *table)
+{
+	int i, num_valid_sets;
+
+	num_valid_sets = 0;
+
+	for (i = 0; i < WM_SET_COUNT; i++) {
+		/* skip empty entries, the smu array has no holes*/
+		if (!bw_params->wm_table.entries[i].valid)
+			continue;
+
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].WmSetting = bw_params->wm_table.entries[i].wm_inst;
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].WmType = bw_params->wm_table.entries[i].wm_type;
+		/* We will not select WM based on fclk, so leave it as unconstrained */
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinClock = 0;
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].MaxClock = 0xFFFF;
+
+		if (table->WatermarkRow[WM_DCFCLK][num_valid_sets].WmType == WM_TYPE_PSTATE_CHG) {
+			if (i == 0)
+				table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinMclk = 0;
+			else {
+				/* add 1 to make it non-overlapping with next lvl */
+				table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinMclk =
+						bw_params->clk_table.entries[i - 1].dcfclk_mhz + 1;
+			}
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets].MaxMclk =
+					bw_params->clk_table.entries[i].dcfclk_mhz;
+
+		} else {
+			/* unconstrained for memory retraining */
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinClock = 0;
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets].MaxClock = 0xFFFF;
+
+			/* Modify previous watermark range to cover up to max */
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets - 1].MaxClock = 0xFFFF;
+		}
+		num_valid_sets++;
+	}
+
+	ASSERT(num_valid_sets != 0); /* Must have at least one set of valid watermarks */
+
+	/* modify the min and max to make sure we cover the whole range*/
+	table->WatermarkRow[WM_DCFCLK][0].MinMclk = 0;
+	table->WatermarkRow[WM_DCFCLK][0].MinClock = 0;
+	table->WatermarkRow[WM_DCFCLK][num_valid_sets - 1].MaxMclk = 0xFFFF;
+	table->WatermarkRow[WM_DCFCLK][num_valid_sets - 1].MaxClock = 0xFFFF;
+
+	/* This is for writeback only, does not matter currently as no writeback support*/
+	table->WatermarkRow[WM_SOCCLK][0].WmSetting = WM_A;
+	table->WatermarkRow[WM_SOCCLK][0].MinClock = 0;
+	table->WatermarkRow[WM_SOCCLK][0].MaxClock = 0xFFFF;
+	table->WatermarkRow[WM_SOCCLK][0].MinMclk = 0;
+	table->WatermarkRow[WM_SOCCLK][0].MaxMclk = 0xFFFF;
+}
+
+static void dcn315_notify_wm_ranges(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct clk_mgr_dcn315 *clk_mgr_dcn315 = TO_CLK_MGR_DCN315(clk_mgr);
+	struct dcn315_watermarks *table = clk_mgr_dcn315->smu_wm_set.wm_set;
+
+	if (!clk_mgr->smu_ver)
+		return;
+
+	if (!table || clk_mgr_dcn315->smu_wm_set.mc_address.quad_part == 0)
+		return;
+
+	memset(table, 0, sizeof(*table));
+
+	dcn315_build_watermark_ranges(clk_mgr_base->bw_params, table);
+
+	dcn315_smu_set_dram_addr_high(clk_mgr,
+			clk_mgr_dcn315->smu_wm_set.mc_address.high_part);
+	dcn315_smu_set_dram_addr_low(clk_mgr,
+			clk_mgr_dcn315->smu_wm_set.mc_address.low_part);
+	dcn315_smu_transfer_wm_table_dram_2_smu(clk_mgr);
+}
+
+static void dcn315_get_dpm_table_from_smu(struct clk_mgr_internal *clk_mgr,
+		struct dcn315_smu_dpm_clks *smu_dpm_clks)
+{
+	DpmClocks_315_t *table = smu_dpm_clks->dpm_clks;
+
+	if (!clk_mgr->smu_ver)
+		return;
+
+	if (!table || smu_dpm_clks->mc_address.quad_part == 0)
+		return;
+
+	memset(table, 0, sizeof(*table));
+
+	dcn315_smu_set_dram_addr_high(clk_mgr,
+			smu_dpm_clks->mc_address.high_part);
+	dcn315_smu_set_dram_addr_low(clk_mgr,
+			smu_dpm_clks->mc_address.low_part);
+	dcn315_smu_transfer_dpm_table_smu_2_dram(clk_mgr);
+}
+
+static void dcn315_clk_mgr_helper_populate_bw_params(
+		struct clk_mgr_internal *clk_mgr,
+		struct integrated_info *bios_info,
+		const DpmClocks_315_t *clock_table)
+{
+	int i;
+	struct clk_bw_params *bw_params = clk_mgr->base.bw_params;
+	uint32_t max_pstate = clock_table->NumDfPstatesEnabled - 1;
+	struct clk_limit_table_entry def_max = bw_params->clk_table.entries[bw_params->clk_table.num_entries - 1];
+
+	/* For 315 we want to base clock table on dcfclk, need at least one entry regardless of pmfw table */
+	for (i = 0; i < clock_table->NumDcfClkLevelsEnabled; i++) {
+		int j;
+
+		/* DF table is sorted with clocks decreasing */
+		for (j = clock_table->NumDfPstatesEnabled - 2; j >= 0; j--) {
+			if (clock_table->DfPstateTable[j].Voltage <= clock_table->SocVoltage[i])
+				max_pstate = j;
+		}
+		/* Max DCFCLK should match up with max pstate */
+		if (i == clock_table->NumDcfClkLevelsEnabled - 1)
+			max_pstate = 0;
+
+		/* First search defaults for the clocks we don't read using closest lower or equal default dcfclk */
+		for (j = bw_params->clk_table.num_entries - 1; j > 0; j--)
+			if (bw_params->clk_table.entries[j].dcfclk_mhz <= clock_table->DcfClocks[i])
+				break;
+		bw_params->clk_table.entries[i].phyclk_mhz = bw_params->clk_table.entries[j].phyclk_mhz;
+		bw_params->clk_table.entries[i].phyclk_d18_mhz = bw_params->clk_table.entries[j].phyclk_d18_mhz;
+		bw_params->clk_table.entries[i].dtbclk_mhz = bw_params->clk_table.entries[j].dtbclk_mhz;
+
+		/* Now update clocks we do read */
+		bw_params->clk_table.entries[i].fclk_mhz = clock_table->DfPstateTable[max_pstate].FClk;
+		bw_params->clk_table.entries[i].memclk_mhz = clock_table->DfPstateTable[max_pstate].MemClk;
+		bw_params->clk_table.entries[i].voltage = clock_table->SocVoltage[i];
+		bw_params->clk_table.entries[i].dcfclk_mhz = clock_table->DcfClocks[i];
+		bw_params->clk_table.entries[i].socclk_mhz = clock_table->SocClocks[i];
+		bw_params->clk_table.entries[i].dispclk_mhz = clock_table->DispClocks[i];
+		bw_params->clk_table.entries[i].dppclk_mhz = clock_table->DppClocks[i];
+		bw_params->clk_table.entries[i].wck_ratio = 1;
+	}
+
+	/* Make sure to include at least one entry */
+	if (i == 0) {
+		bw_params->clk_table.entries[i].fclk_mhz = clock_table->DfPstateTable[0].FClk;
+		bw_params->clk_table.entries[i].memclk_mhz = clock_table->DfPstateTable[0].MemClk;
+		bw_params->clk_table.entries[i].voltage = clock_table->DfPstateTable[0].Voltage;
+		bw_params->clk_table.entries[i].dcfclk_mhz = clock_table->DcfClocks[0];
+		bw_params->clk_table.entries[i].wck_ratio = 1;
+		i++;
+	} else if (clock_table->NumDcfClkLevelsEnabled != clock_table->NumSocClkLevelsEnabled) {
+		bw_params->clk_table.entries[i-1].voltage = clock_table->SocVoltage[clock_table->NumSocClkLevelsEnabled - 1];
+		bw_params->clk_table.entries[i-1].socclk_mhz = clock_table->SocClocks[clock_table->NumSocClkLevelsEnabled - 1];
+		bw_params->clk_table.entries[i-1].dispclk_mhz = clock_table->DispClocks[clock_table->NumDispClkLevelsEnabled - 1];
+		bw_params->clk_table.entries[i-1].dppclk_mhz = clock_table->DppClocks[clock_table->NumDispClkLevelsEnabled - 1];
+	}
+	bw_params->clk_table.num_entries = i;
+
+	/* Set any 0 clocks to max default setting. Not an issue for
+	 * power since we aren't doing switching in such case anyway
+	 */
+	for (i = 0; i < bw_params->clk_table.num_entries; i++) {
+		if (!bw_params->clk_table.entries[i].fclk_mhz) {
+			bw_params->clk_table.entries[i].fclk_mhz = def_max.fclk_mhz;
+			bw_params->clk_table.entries[i].memclk_mhz = def_max.memclk_mhz;
+			bw_params->clk_table.entries[i].voltage = def_max.voltage;
+		}
+		if (!bw_params->clk_table.entries[i].dcfclk_mhz)
+			bw_params->clk_table.entries[i].dcfclk_mhz = def_max.dcfclk_mhz;
+		if (!bw_params->clk_table.entries[i].socclk_mhz)
+			bw_params->clk_table.entries[i].socclk_mhz = def_max.socclk_mhz;
+		if (!bw_params->clk_table.entries[i].dispclk_mhz)
+			bw_params->clk_table.entries[i].dispclk_mhz = def_max.dispclk_mhz;
+		if (!bw_params->clk_table.entries[i].dppclk_mhz)
+			bw_params->clk_table.entries[i].dppclk_mhz = def_max.dppclk_mhz;
+		if (!bw_params->clk_table.entries[i].phyclk_mhz)
+			bw_params->clk_table.entries[i].phyclk_mhz = def_max.phyclk_mhz;
+		if (!bw_params->clk_table.entries[i].phyclk_d18_mhz)
+			bw_params->clk_table.entries[i].phyclk_d18_mhz = def_max.phyclk_d18_mhz;
+		if (!bw_params->clk_table.entries[i].dtbclk_mhz)
+			bw_params->clk_table.entries[i].dtbclk_mhz = def_max.dtbclk_mhz;
+	}
+
+	/* Make sure all highest default clocks are included*/
+	ASSERT(bw_params->clk_table.entries[i-1].phyclk_mhz == def_max.phyclk_mhz);
+	ASSERT(bw_params->clk_table.entries[i-1].phyclk_d18_mhz == def_max.phyclk_d18_mhz);
+	ASSERT(bw_params->clk_table.entries[i-1].dtbclk_mhz == def_max.dtbclk_mhz);
+	ASSERT(bw_params->clk_table.entries[i-1].dcfclk_mhz);
+	bw_params->vram_type = bios_info->memory_type;
+	bw_params->num_channels = bios_info->ma_channel_number;
+	bw_params->dram_channel_width_bytes = bios_info->memory_type == 0x22 ? 8 : 4;
+
+	for (i = 0; i < WM_SET_COUNT; i++) {
+		bw_params->wm_table.entries[i].wm_inst = i;
+
+		if (i >= bw_params->clk_table.num_entries) {
+			bw_params->wm_table.entries[i].valid = false;
+			continue;
+		}
+
+		bw_params->wm_table.entries[i].wm_type = WM_TYPE_PSTATE_CHG;
+		bw_params->wm_table.entries[i].valid = true;
+	}
+}
+
+static void dcn315_enable_pme_wa(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	dcn315_smu_enable_pme_wa(clk_mgr);
+}
+
+static struct clk_mgr_funcs dcn315_funcs = {
+	.get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz,
+	.get_dtb_ref_clk_frequency = dcn31_get_dtb_ref_freq_khz,
+	.update_clocks = dcn315_update_clocks,
+	.init_clocks = dcn31_init_clocks,
+	.enable_pme_wa = dcn315_enable_pme_wa,
+	.are_clock_states_equal = dcn31_are_clock_states_equal,
+	.notify_wm_ranges = dcn315_notify_wm_ranges
+};
+extern struct clk_mgr_funcs dcn3_fpga_funcs;
+
+void dcn315_clk_mgr_construct(
+		struct dc_context *ctx,
+		struct clk_mgr_dcn315 *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg)
+{
+	struct dcn315_smu_dpm_clks smu_dpm_clks = { 0 };
+	struct clk_log_info log_info = {0};
+
+	clk_mgr->base.base.ctx = ctx;
+	clk_mgr->base.base.funcs = &dcn315_funcs;
+
+	clk_mgr->base.pp_smu = pp_smu;
+
+	clk_mgr->base.dccg = dccg;
+	clk_mgr->base.dfs_bypass_disp_clk = 0;
+
+	clk_mgr->base.dprefclk_ss_percentage = 0;
+	clk_mgr->base.dprefclk_ss_divider = 1000;
+	clk_mgr->base.ss_on_dprefclk = false;
+	clk_mgr->base.dfs_ref_freq_khz = 48000;
+
+	clk_mgr->smu_wm_set.wm_set = (struct dcn315_watermarks *)dm_helpers_allocate_gpu_mem(
+				clk_mgr->base.base.ctx,
+				DC_MEM_ALLOC_TYPE_FRAME_BUFFER,
+				sizeof(struct dcn315_watermarks),
+				&clk_mgr->smu_wm_set.mc_address.quad_part);
+
+	if (!clk_mgr->smu_wm_set.wm_set) {
+		clk_mgr->smu_wm_set.wm_set = &dummy_wms;
+		clk_mgr->smu_wm_set.mc_address.quad_part = 0;
+	}
+	ASSERT(clk_mgr->smu_wm_set.wm_set);
+
+	smu_dpm_clks.dpm_clks = (DpmClocks_315_t *)dm_helpers_allocate_gpu_mem(
+				clk_mgr->base.base.ctx,
+				DC_MEM_ALLOC_TYPE_FRAME_BUFFER,
+				sizeof(DpmClocks_315_t),
+				&smu_dpm_clks.mc_address.quad_part);
+
+	if (smu_dpm_clks.dpm_clks == NULL) {
+		smu_dpm_clks.dpm_clks = &dummy_clocks;
+		smu_dpm_clks.mc_address.quad_part = 0;
+	}
+
+	ASSERT(smu_dpm_clks.dpm_clks);
+
+	clk_mgr->base.smu_ver = dcn315_smu_get_smu_version(&clk_mgr->base);
+
+	if (clk_mgr->base.smu_ver > 0)
+		clk_mgr->base.smu_present = true;
+
+	if (ctx->dc_bios->integrated_info->memory_type == LpDdr5MemType) {
+		dcn315_bw_params.wm_table = lpddr5_wm_table;
+	} else {
+		dcn315_bw_params.wm_table = ddr5_wm_table;
+	}
+	/* Saved clocks configured at boot for debug purposes */
+	dcn315_dump_clk_registers(&clk_mgr->base.base.boot_snapshot,
+				  &clk_mgr->base.base, &log_info);
+
+	clk_mgr->base.base.dprefclk_khz = 600000;
+	clk_mgr->base.base.dprefclk_khz = dcn315_smu_get_dpref_clk(&clk_mgr->base);
+	clk_mgr->base.base.clks.ref_dtbclk_khz = clk_mgr->base.base.dprefclk_khz;
+	dce_clock_read_ss_info(&clk_mgr->base);
+	clk_mgr->base.base.clks.ref_dtbclk_khz = dce_adjust_dp_ref_freq_for_ss(&clk_mgr->base, clk_mgr->base.base.dprefclk_khz);
+
+	clk_mgr->base.base.bw_params = &dcn315_bw_params;
+
+	if (clk_mgr->base.base.ctx->dc->debug.pstate_enabled) {
+		int i;
+
+		dcn315_get_dpm_table_from_smu(&clk_mgr->base, &smu_dpm_clks);
+		DC_LOG_SMU("NumDcfClkLevelsEnabled: %d\n"
+				   "NumDispClkLevelsEnabled: %d\n"
+				   "NumSocClkLevelsEnabled: %d\n"
+				   "VcnClkLevelsEnabled: %d\n"
+				   "NumDfPst atesEnabled: %d\n"
+				   "MinGfxClk: %d\n"
+				   "MaxGfxClk: %d\n",
+				   smu_dpm_clks.dpm_clks->NumDcfClkLevelsEnabled,
+				   smu_dpm_clks.dpm_clks->NumDispClkLevelsEnabled,
+				   smu_dpm_clks.dpm_clks->NumSocClkLevelsEnabled,
+				   smu_dpm_clks.dpm_clks->VcnClkLevelsEnabled,
+				   smu_dpm_clks.dpm_clks->NumDfPstatesEnabled,
+				   smu_dpm_clks.dpm_clks->MinGfxClk,
+				   smu_dpm_clks.dpm_clks->MaxGfxClk);
+		for (i = 0; i < smu_dpm_clks.dpm_clks->NumDcfClkLevelsEnabled; i++) {
+			DC_LOG_SMU("smu_dpm_clks.dpm_clks->DcfClocks[%d] = %d\n",
+					   i,
+					   smu_dpm_clks.dpm_clks->DcfClocks[i]);
+		}
+		for (i = 0; i < smu_dpm_clks.dpm_clks->NumDispClkLevelsEnabled; i++) {
+			DC_LOG_SMU("smu_dpm_clks.dpm_clks->DispClocks[%d] = %d\n",
+					   i, smu_dpm_clks.dpm_clks->DispClocks[i]);
+		}
+		for (i = 0; i < smu_dpm_clks.dpm_clks->NumSocClkLevelsEnabled; i++) {
+			DC_LOG_SMU("smu_dpm_clks.dpm_clks->SocClocks[%d] = %d\n",
+					   i, smu_dpm_clks.dpm_clks->SocClocks[i]);
+		}
+		for (i = 0; i < NUM_SOC_VOLTAGE_LEVELS; i++)
+			DC_LOG_SMU("smu_dpm_clks.dpm_clks->SocVoltage[%d] = %d\n",
+					   i, smu_dpm_clks.dpm_clks->SocVoltage[i]);
+
+		for (i = 0; i < NUM_DF_PSTATE_LEVELS; i++) {
+			DC_LOG_SMU("smu_dpm_clks.dpm_clks.DfPstateTable[%d].FClk = %d\n"
+					   "smu_dpm_clks.dpm_clks->DfPstateTable[%d].MemClk= %d\n"
+					   "smu_dpm_clks.dpm_clks->DfPstateTable[%d].Voltage = %d\n",
+					   i, smu_dpm_clks.dpm_clks->DfPstateTable[i].FClk,
+					   i, smu_dpm_clks.dpm_clks->DfPstateTable[i].MemClk,
+					   i, smu_dpm_clks.dpm_clks->DfPstateTable[i].Voltage);
+		}
+
+		if (ctx->dc_bios->integrated_info) {
+			dcn315_clk_mgr_helper_populate_bw_params(
+					&clk_mgr->base,
+					ctx->dc_bios->integrated_info,
+					smu_dpm_clks.dpm_clks);
+		}
+	}
+
+	if (smu_dpm_clks.dpm_clks && smu_dpm_clks.mc_address.quad_part != 0)
+		dm_helpers_free_gpu_mem(clk_mgr->base.base.ctx, DC_MEM_ALLOC_TYPE_FRAME_BUFFER,
+				smu_dpm_clks.dpm_clks);
+}
+
+void dcn315_clk_mgr_destroy(struct clk_mgr_internal *clk_mgr_int)
+{
+	struct clk_mgr_dcn315 *clk_mgr = TO_CLK_MGR_DCN315(clk_mgr_int);
+
+	if (clk_mgr->smu_wm_set.wm_set && clk_mgr->smu_wm_set.mc_address.quad_part != 0)
+		dm_helpers_free_gpu_mem(clk_mgr_int->base.ctx, DC_MEM_ALLOC_TYPE_FRAME_BUFFER,
+				clk_mgr->smu_wm_set.wm_set);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.h
new file mode 100644
index 000000000000..ac36ddf5dd1a
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.h
@@ -0,0 +1,49 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCN315_CLK_MGR_H__
+#define __DCN315_CLK_MGR_H__
+#include "clk_mgr_internal.h"
+
+struct dcn315_watermarks;
+
+struct dcn315_smu_watermark_set {
+	struct dcn315_watermarks *wm_set;
+	union large_integer mc_address;
+};
+
+struct clk_mgr_dcn315 {
+	struct clk_mgr_internal base;
+	struct dcn315_smu_watermark_set smu_wm_set;
+};
+
+void dcn315_clk_mgr_construct(struct dc_context *ctx,
+		struct clk_mgr_dcn315 *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg);
+
+void dcn315_clk_mgr_destroy(struct clk_mgr_internal *clk_mgr_int);
+
+#endif //__DCN315_CLK_MGR_H__
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_smu.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_smu.c
new file mode 100644
index 000000000000..478b4d6a3544
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_smu.c
@@ -0,0 +1,366 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "core_types.h"
+#include "clk_mgr_internal.h"
+#include "reg_helper.h"
+#include "dm_helpers.h"
+#include "dcn315_smu.h"
+#include "mp/mp_13_0_5_offset.h"
+#include "logger_types.h"
+
+#define MAX_INSTANCE                                        6
+#define MAX_SEGMENT                                         6
+#define SMU_REGISTER_WRITE_RETRY_COUNT                      5
+
+struct IP_BASE_INSTANCE {
+    unsigned int segment[MAX_SEGMENT];
+};
+
+struct IP_BASE {
+    struct IP_BASE_INSTANCE instance[MAX_INSTANCE];
+};
+
+static const struct IP_BASE MP0_BASE = { { { { 0x00016000, 0x00DC0000, 0x00E00000, 0x00E40000, 0x0243FC00, 0 } },
+					{ { 0, 0, 0, 0, 0, 0 } },
+					{ { 0, 0, 0, 0, 0, 0 } },
+					{ { 0, 0, 0, 0, 0, 0 } },
+					{ { 0, 0, 0, 0, 0, 0 } },
+					{ { 0, 0, 0, 0, 0, 0 } } } };
+
+#define CTX clk_mgr->base.ctx
+#define IND_REG(offset)	offset
+
+#define regBIF_BX_PF2_RSMU_INDEX                                                                        0x0000
+#define regBIF_BX_PF2_RSMU_INDEX_BASE_IDX                                                               1
+#define regBIF_BX_PF2_RSMU_DATA                                                                         0x0001
+#define regBIF_BX_PF2_RSMU_DATA_BASE_IDX                                                                1
+
+#define REG(reg_name) \
+	(MP0_BASE.instance[0].segment[reg ## reg_name ## _BASE_IDX] + reg ## reg_name)
+
+#define FN(reg_name, field) \
+	FD(reg_name##__##field)
+
+#undef DC_LOGGER
+#define DC_LOGGER \
+	CTX->logger
+#define smu_print(str, ...) {DC_LOG_SMU(str, ##__VA_ARGS__); }
+
+#define mmMP1_C2PMSG_3                            0x3B1050C
+
+#define reg__MP1_C2PMSG_3_MASK					(0xFFFFFFFF)
+#define reg__MP1_C2PMSG_3__SHIFT					(0)
+
+
+#define data_reg_name__MP1_C2PMSG_3_MASK		(0xFFFFFFFF)
+#define data_reg_name__MP1_C2PMSG_3__SHIFT		(0)
+
+#define VBIOSSMC_MSG_TestMessage                  0x01 ///< To check if PMFW is alive and responding. Requirement specified by PMFW team
+#define VBIOSSMC_MSG_GetPmfwVersion               0x02 ///< Get PMFW version
+#define VBIOSSMC_MSG_Spare0                       0x03 ///< Spare0
+#define VBIOSSMC_MSG_SetDispclkFreq               0x04 ///< Set display clock frequency in MHZ
+#define VBIOSSMC_MSG_Spare1                       0x05 ///< Spare1
+#define VBIOSSMC_MSG_SetDppclkFreq                0x06 ///< Set DPP clock frequency in MHZ
+#define VBIOSSMC_MSG_SetHardMinDcfclkByFreq       0x07 ///< Set DCF clock frequency hard min in MHZ
+#define VBIOSSMC_MSG_SetMinDeepSleepDcfclk        0x08 ///< Set DCF clock minimum frequency in deep sleep in MHZ
+#define VBIOSSMC_MSG_GetDtbclkFreq                0x09 ///< Get display dtb clock frequency in MHZ in case VMIN does not support phy frequency
+#define VBIOSSMC_MSG_SetDtbClk                    0x0A ///< Set dtb clock frequency, return frequemcy in MHZ
+#define VBIOSSMC_MSG_SetDisplayCount              0x0B ///< Inform PMFW of number of display connected
+#define VBIOSSMC_MSG_EnableTmdp48MHzRefclkPwrDown 0x0C ///< To ask PMFW turn off TMDP 48MHz refclk during display off to save power
+#define VBIOSSMC_MSG_UpdatePmeRestore             0x0D ///< To ask PMFW to write into Azalia for PME wake up event
+#define VBIOSSMC_MSG_SetVbiosDramAddrHigh         0x0E ///< Set DRAM address high 32 bits for WM table transfer
+#define VBIOSSMC_MSG_SetVbiosDramAddrLow          0x0F ///< Set DRAM address low 32 bits for WM table transfer
+#define VBIOSSMC_MSG_TransferTableSmu2Dram        0x10 ///< Transfer table from PMFW SRAM to system DRAM
+#define VBIOSSMC_MSG_TransferTableDram2Smu        0x11 ///< Transfer table from system DRAM to PMFW
+#define VBIOSSMC_MSG_SetDisplayIdleOptimizations  0x12 ///< Set Idle state optimization for display off
+#define VBIOSSMC_MSG_GetDprefclkFreq              0x13 ///< Get DPREF clock frequency. Return in MHZ
+#define VBIOSSMC_Message_Count                    0x14 ///< Total number of VBIS and DAL messages
+
+#define VBIOSSMC_Status_BUSY                      0x0
+#define VBIOSSMC_Result_OK                        0x01 ///< Message Response OK
+#define VBIOSSMC_Result_Failed                    0xFF ///< Message Response Failed
+#define VBIOSSMC_Result_UnknownCmd                0xFE ///< Message Response Unknown Command
+#define VBIOSSMC_Result_CmdRejectedPrereq         0xFD ///< Message Response Command Failed Prerequisite
+#define VBIOSSMC_Result_CmdRejectedBusy           0xFC ///< Message Response Command Rejected due to PMFW is busy. Sender should retry sending this message
+
+/*
+ * Function to be used instead of REG_WAIT macro because the wait ends when
+ * the register is NOT EQUAL to zero, and because the translation in msg_if.h
+ * won't work with REG_WAIT.
+ */
+static uint32_t dcn315_smu_wait_for_response(struct clk_mgr_internal *clk_mgr, unsigned int delay_us, unsigned int max_retries)
+{
+	uint32_t res_val = VBIOSSMC_Status_BUSY;
+
+	do {
+		res_val = REG_READ(MP1_SMN_C2PMSG_38);
+		if (res_val != VBIOSSMC_Status_BUSY)
+			break;
+
+		if (delay_us >= 1000)
+			msleep(delay_us/1000);
+		else if (delay_us > 0)
+			udelay(delay_us);
+	} while (max_retries--);
+
+	return res_val;
+}
+
+static int dcn315_smu_send_msg_with_param(
+		struct clk_mgr_internal *clk_mgr,
+		unsigned int msg_id, unsigned int param)
+{
+	uint32_t result;
+	uint32_t i = 0;
+	uint32_t read_back_data;
+
+	result = dcn315_smu_wait_for_response(clk_mgr, 10, 200000);
+
+	if (result != VBIOSSMC_Result_OK)
+		smu_print("SMU Response was not OK. SMU response after wait received is: %d\n", result);
+
+	if (result == VBIOSSMC_Status_BUSY) {
+		return -1;
+	}
+
+	/* First clear response register */
+	REG_WRITE(MP1_SMN_C2PMSG_38, VBIOSSMC_Status_BUSY);
+
+	/* Set the parameter register for the SMU message, unit is Mhz */
+	REG_WRITE(MP1_SMN_C2PMSG_37, param);
+
+	for (i = 0; i < SMU_REGISTER_WRITE_RETRY_COUNT; i++) {
+		/* Trigger the message transaction by writing the message ID */
+		IX_REG_SET_SYNC(mmMP1_C2PMSG_3, 0,
+			MP1_C2PMSG_3, msg_id);
+		IX_REG_GET_SYNC(mmMP1_C2PMSG_3,
+			MP1_C2PMSG_3, &read_back_data);
+		if (read_back_data == msg_id)
+			break;
+		udelay(2);
+		smu_print("SMU msg id write fail %x times. \n", i + 1);
+	}
+
+	result = dcn315_smu_wait_for_response(clk_mgr, 10, 200000);
+
+	if (result == VBIOSSMC_Status_BUSY) {
+		ASSERT(0);
+		dm_helpers_smu_timeout(CTX, msg_id, param, 10 * 200000);
+	}
+
+	return REG_READ(MP1_SMN_C2PMSG_37);
+}
+
+int dcn315_smu_get_smu_version(struct clk_mgr_internal *clk_mgr)
+{
+	return dcn315_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_GetPmfwVersion,
+			0);
+}
+
+
+int dcn315_smu_set_dispclk(struct clk_mgr_internal *clk_mgr, int requested_dispclk_khz)
+{
+	int actual_dispclk_set_mhz = -1;
+
+	if (!clk_mgr->smu_present)
+		return requested_dispclk_khz;
+
+	/*  Unit of SMU msg parameter is Mhz */
+	actual_dispclk_set_mhz = dcn315_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDispclkFreq,
+			khz_to_mhz_ceil(requested_dispclk_khz));
+
+	return actual_dispclk_set_mhz * 1000;
+}
+
+int dcn315_smu_set_hard_min_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_dcfclk_khz)
+{
+	int actual_dcfclk_set_mhz = -1;
+
+	if (!clk_mgr->base.ctx->dc->debug.pstate_enabled)
+		return -1;
+
+	if (!clk_mgr->smu_present)
+		return requested_dcfclk_khz;
+
+	actual_dcfclk_set_mhz = dcn315_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetHardMinDcfclkByFreq,
+			khz_to_mhz_ceil(requested_dcfclk_khz));
+
+	return actual_dcfclk_set_mhz * 1000;
+}
+
+int dcn315_smu_set_min_deep_sleep_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_min_ds_dcfclk_khz)
+{
+	int actual_min_ds_dcfclk_mhz = -1;
+
+	if (!clk_mgr->base.ctx->dc->debug.pstate_enabled)
+		return -1;
+
+	if (!clk_mgr->smu_present)
+		return requested_min_ds_dcfclk_khz;
+
+	actual_min_ds_dcfclk_mhz = dcn315_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetMinDeepSleepDcfclk,
+			khz_to_mhz_ceil(requested_min_ds_dcfclk_khz));
+
+	return actual_min_ds_dcfclk_mhz * 1000;
+}
+
+int dcn315_smu_set_dppclk(struct clk_mgr_internal *clk_mgr, int requested_dpp_khz)
+{
+	int actual_dppclk_set_mhz = -1;
+
+	if (!clk_mgr->smu_present)
+		return requested_dpp_khz;
+
+	actual_dppclk_set_mhz = dcn315_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDppclkFreq,
+			khz_to_mhz_ceil(requested_dpp_khz));
+
+	return actual_dppclk_set_mhz * 1000;
+}
+
+void dcn315_smu_set_display_idle_optimization(struct clk_mgr_internal *clk_mgr, uint32_t idle_info)
+{
+	if (!clk_mgr->base.ctx->dc->debug.pstate_enabled)
+		return;
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	//TODO: Work with smu team to define optimization options.
+	dcn315_smu_send_msg_with_param(
+		clk_mgr,
+		VBIOSSMC_MSG_SetDisplayIdleOptimizations,
+		idle_info);
+}
+
+void dcn315_smu_enable_phy_refclk_pwrdwn(struct clk_mgr_internal *clk_mgr, bool enable)
+{
+	union display_idle_optimization_u idle_info = { 0 };
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	if (enable) {
+		idle_info.idle_info.df_request_disabled = 1;
+		idle_info.idle_info.phy_ref_clk_off = 1;
+	}
+
+	dcn315_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDisplayIdleOptimizations,
+			idle_info.data);
+}
+
+void dcn315_smu_enable_pme_wa(struct clk_mgr_internal *clk_mgr)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn315_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_UpdatePmeRestore,
+			0);
+}
+void dcn315_smu_set_dram_addr_high(struct clk_mgr_internal *clk_mgr, uint32_t addr_high)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn315_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_SetVbiosDramAddrHigh, addr_high);
+}
+
+void dcn315_smu_set_dram_addr_low(struct clk_mgr_internal *clk_mgr, uint32_t addr_low)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn315_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_SetVbiosDramAddrLow, addr_low);
+}
+
+void dcn315_smu_transfer_dpm_table_smu_2_dram(struct clk_mgr_internal *clk_mgr)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn315_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_TransferTableSmu2Dram, TABLE_DPMCLOCKS);
+}
+
+void dcn315_smu_transfer_wm_table_dram_2_smu(struct clk_mgr_internal *clk_mgr)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn315_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_TransferTableDram2Smu, TABLE_WATERMARKS);
+}
+
+int dcn315_smu_get_dpref_clk(struct clk_mgr_internal *clk_mgr)
+{
+	int dprefclk_get_mhz = -1;
+	if (clk_mgr->smu_present) {
+		dprefclk_get_mhz = dcn315_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_GetDprefclkFreq,
+			0);
+	}
+	return (dprefclk_get_mhz * 1000);
+}
+
+int dcn315_smu_get_dtbclk(struct clk_mgr_internal *clk_mgr)
+{
+	int fclk_get_mhz = -1;
+
+	if (clk_mgr->smu_present) {
+		fclk_get_mhz = dcn315_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_GetDtbclkFreq,
+			0);
+	}
+	return (fclk_get_mhz * 1000);
+}
+
+void dcn315_smu_set_dtbclk(struct clk_mgr_internal *clk_mgr, bool enable)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn315_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDtbClk,
+			enable);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_smu.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_smu.h
new file mode 100644
index 000000000000..5aa3275ac7d8
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_smu.h
@@ -0,0 +1,130 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DAL_DC_315_SMU_H_
+#define DAL_DC_315_SMU_H_
+#include "os_types.h"
+
+#define PMFW_DRIVER_IF_VERSION 4
+
+#define NUM_DCFCLK_DPM_LEVELS   4
+#define NUM_DISPCLK_DPM_LEVELS  4
+#define NUM_DPPCLK_DPM_LEVELS   4
+#define NUM_SOCCLK_DPM_LEVELS   4
+#define NUM_VCN_DPM_LEVELS      4
+#define NUM_SOC_VOLTAGE_LEVELS  4
+#define NUM_DF_PSTATE_LEVELS    4
+
+
+typedef struct {
+  uint16_t MinClock; // This is either DCFCLK or SOCCLK (in MHz)
+  uint16_t MaxClock; // This is either DCFCLK or SOCCLK (in MHz)
+  uint16_t MinMclk;
+  uint16_t MaxMclk;
+  uint8_t  WmSetting;
+  uint8_t  WmType;  // Used for normal pstate change or memory retraining
+  uint8_t  Padding[2];
+} WatermarkRowGeneric_t;
+
+#define NUM_WM_RANGES 4
+#define WM_PSTATE_CHG 0
+#define WM_RETRAINING 1
+
+typedef enum {
+  WM_SOCCLK = 0,
+  WM_DCFCLK,
+  WM_COUNT,
+} WM_CLOCK_e;
+
+typedef struct {
+  uint32_t FClk;
+  uint32_t MemClk;
+  uint32_t Voltage;
+} DfPstateTable_t;
+
+//Freq in MHz
+//Voltage in milli volts with 2 fractional bits
+typedef struct {
+  uint32_t DcfClocks[NUM_DCFCLK_DPM_LEVELS];
+  uint32_t DispClocks[NUM_DISPCLK_DPM_LEVELS];
+  uint32_t DppClocks[NUM_DPPCLK_DPM_LEVELS];
+  uint32_t SocClocks[NUM_SOCCLK_DPM_LEVELS];
+  uint32_t VClocks[NUM_VCN_DPM_LEVELS];
+  uint32_t DClocks[NUM_VCN_DPM_LEVELS];
+  uint32_t SocVoltage[NUM_SOC_VOLTAGE_LEVELS];
+  DfPstateTable_t DfPstateTable[NUM_DF_PSTATE_LEVELS];
+  uint8_t  NumDcfClkLevelsEnabled;
+  uint8_t  NumDispClkLevelsEnabled; //Applies to both Dispclk and Dppclk
+  uint8_t  NumSocClkLevelsEnabled;
+  uint8_t  VcnClkLevelsEnabled;     //Applies to both Vclk and Dclk
+  uint8_t  NumDfPstatesEnabled;
+  uint8_t  spare[3];
+  uint32_t MinGfxClk;
+  uint32_t MaxGfxClk;
+} DpmClocks_315_t;
+
+struct dcn315_watermarks {
+  // Watermarks
+  WatermarkRowGeneric_t WatermarkRow[WM_COUNT][NUM_WM_RANGES];
+  uint32_t MmHubPadding[7]; // SMU internal use
+};
+
+struct dcn315_smu_dpm_clks {
+	DpmClocks_315_t *dpm_clks;
+	union large_integer mc_address;
+};
+
+#define TABLE_WATERMARKS         1 // Called by DAL through VBIOS
+#define TABLE_DPMCLOCKS          4 // Called by Driver and VBIOS
+
+struct display_idle_optimization {
+	unsigned int df_request_disabled : 1;
+	unsigned int phy_ref_clk_off     : 1;
+	unsigned int s0i2_rdy            : 1;
+	unsigned int reserved            : 29;
+};
+
+union display_idle_optimization_u {
+	struct display_idle_optimization idle_info;
+	uint32_t data;
+};
+
+int dcn315_smu_get_smu_version(struct clk_mgr_internal *clk_mgr);
+int dcn315_smu_set_dispclk(struct clk_mgr_internal *clk_mgr, int requested_dispclk_khz);
+int dcn315_smu_set_hard_min_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_dcfclk_khz);
+int dcn315_smu_set_min_deep_sleep_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_min_ds_dcfclk_khz);
+int dcn315_smu_set_dppclk(struct clk_mgr_internal *clk_mgr, int requested_dpp_khz);
+void dcn315_smu_set_display_idle_optimization(struct clk_mgr_internal *clk_mgr, uint32_t idle_info);
+void dcn315_smu_enable_phy_refclk_pwrdwn(struct clk_mgr_internal *clk_mgr, bool enable);
+void dcn315_smu_set_dram_addr_high(struct clk_mgr_internal *clk_mgr, uint32_t addr_high);
+void dcn315_smu_set_dram_addr_low(struct clk_mgr_internal *clk_mgr, uint32_t addr_low);
+void dcn315_smu_transfer_dpm_table_smu_2_dram(struct clk_mgr_internal *clk_mgr);
+void dcn315_smu_transfer_wm_table_dram_2_smu(struct clk_mgr_internal *clk_mgr);
+void dcn315_smu_request_voltage_via_phyclk(struct clk_mgr_internal *clk_mgr, int requested_phyclk_khz);
+void dcn315_smu_enable_pme_wa(struct clk_mgr_internal *clk_mgr);
+int dcn315_smu_get_dpref_clk(struct clk_mgr_internal *clk_mgr);
+int dcn315_smu_get_dtbclk(struct clk_mgr_internal *clk_mgr);
+void dcn315_smu_set_dtbclk(struct clk_mgr_internal *clk_mgr, bool enable);
+#endif /* DAL_DC_315_SMU_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_clk_mgr.c
new file mode 100644
index 000000000000..1769b1f26e75
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_clk_mgr.c
@@ -0,0 +1,680 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+
+#include "dccg.h"
+#include "clk_mgr_internal.h"
+
+// For dce12_get_dp_ref_freq_khz
+#include "dce100/dce_clk_mgr.h"
+// For dcn20_update_clocks_update_dpp_dto
+#include "dcn20/dcn20_clk_mgr.h"
+#include "dcn31/dcn31_clk_mgr.h"
+#include "dcn316_clk_mgr.h"
+#include "reg_helper.h"
+#include "core_types.h"
+#include "dcn316_smu.h"
+#include "dm_helpers.h"
+#include "dc_dmub_srv.h"
+#include "link_service.h"
+
+// DCN316 this is CLK1 instance
+#define MAX_INSTANCE                                        7
+#define MAX_SEGMENT                                         6
+
+struct IP_BASE_INSTANCE {
+    unsigned int segment[MAX_SEGMENT];
+};
+
+struct IP_BASE {
+    struct IP_BASE_INSTANCE instance[MAX_INSTANCE];
+};
+
+#define regCLK1_CLK_PLL_REQ						0x0237
+#define regCLK1_CLK_PLL_REQ_BASE_IDX			0
+
+#define CLK1_CLK_PLL_REQ__FbMult_int__SHIFT		0x0
+#define CLK1_CLK_PLL_REQ__PllSpineDiv__SHIFT	0xc
+#define CLK1_CLK_PLL_REQ__FbMult_frac__SHIFT	0x10
+#define CLK1_CLK_PLL_REQ__FbMult_int_MASK		0x000001FFL
+#define CLK1_CLK_PLL_REQ__PllSpineDiv_MASK		0x0000F000L
+#define CLK1_CLK_PLL_REQ__FbMult_frac_MASK		0xFFFF0000L
+
+#define TO_CLK_MGR_DCN316(clk_mgr)\
+	container_of(clk_mgr, struct clk_mgr_dcn316, base)
+
+static int dcn316_get_active_display_cnt_wa(
+		struct dc *dc,
+		struct dc_state *context)
+{
+	int i, display_count;
+	bool tmds_present = false;
+
+	display_count = 0;
+	for (i = 0; i < context->stream_count; i++) {
+		const struct dc_stream_state *stream = context->streams[i];
+
+		if (stream->signal == SIGNAL_TYPE_HDMI_TYPE_A ||
+				stream->signal == SIGNAL_TYPE_DVI_SINGLE_LINK ||
+				stream->signal == SIGNAL_TYPE_DVI_DUAL_LINK)
+			tmds_present = true;
+	}
+
+	for (i = 0; i < dc->link_count; i++) {
+		const struct dc_link *link = dc->links[i];
+
+		/* abusing the fact that the dig and phy are coupled to see if the phy is enabled */
+		if (link->link_enc && link->link_enc->funcs->is_dig_enabled &&
+				link->link_enc->funcs->is_dig_enabled(link->link_enc))
+			display_count++;
+	}
+
+	/* WA for hang on HDMI after display off back back on*/
+	if (display_count == 0 && tmds_present)
+		display_count = 1;
+
+	return display_count;
+}
+
+static void dcn316_disable_otg_wa(struct clk_mgr *clk_mgr_base, struct dc_state *context,
+		bool safe_to_lower, bool disable)
+{
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	int i;
+
+	for (i = 0; i < dc->res_pool->pipe_count; ++i) {
+		struct pipe_ctx *pipe = safe_to_lower
+			? &context->res_ctx.pipe_ctx[i]
+			: &dc->current_state->res_ctx.pipe_ctx[i];
+
+		if (pipe->top_pipe || pipe->prev_odm_pipe)
+			continue;
+		if (pipe->stream && (pipe->stream->dpms_off || dc_is_virtual_signal(pipe->stream->signal) ||
+				     !pipe->stream->link_enc)) {
+			if (disable) {
+				if (pipe->stream_res.tg && pipe->stream_res.tg->funcs->immediate_disable_crtc)
+					pipe->stream_res.tg->funcs->immediate_disable_crtc(pipe->stream_res.tg);
+
+				reset_sync_context_for_pipe(dc, context, i);
+			} else
+				pipe->stream_res.tg->funcs->enable_crtc(pipe->stream_res.tg);
+		}
+	}
+}
+
+static void dcn316_enable_pme_wa(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	dcn316_smu_enable_pme_wa(clk_mgr);
+}
+
+static void dcn316_update_clocks(struct clk_mgr *clk_mgr_base,
+			struct dc_state *context,
+			bool safe_to_lower)
+{
+	union dmub_rb_cmd cmd;
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dc_clocks *new_clocks = &context->bw_ctx.bw.dcn.clk;
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	int display_count = 0;
+	bool update_dppclk = false;
+	bool update_dispclk = false;
+	bool dpp_clock_lowered = false;
+
+	if (dc->work_arounds.skip_clock_update)
+		return;
+
+	/*
+	 * if it is safe to lower, but we are already in the lower state, we don't have to do anything
+	 * also if safe to lower is false, we just go in the higher state
+	 */
+	clk_mgr_base->clks.zstate_support = new_clocks->zstate_support;
+	if (safe_to_lower) {
+		if (clk_mgr_base->clks.dtbclk_en && !new_clocks->dtbclk_en) {
+			dcn316_smu_set_dtbclk(clk_mgr, false);
+			clk_mgr_base->clks.dtbclk_en = new_clocks->dtbclk_en;
+		}
+		/* check that we're not already in lower */
+		if (clk_mgr_base->clks.pwr_state != DCN_PWR_STATE_LOW_POWER) {
+			display_count = dcn316_get_active_display_cnt_wa(dc, context);
+			/* if we can go lower, go lower */
+			if (display_count == 0) {
+				union display_idle_optimization_u idle_info = { 0 };
+				idle_info.idle_info.df_request_disabled = 1;
+				idle_info.idle_info.phy_ref_clk_off = 1;
+				idle_info.idle_info.s0i2_rdy = 1;
+				dcn316_smu_set_display_idle_optimization(clk_mgr, idle_info.data);
+				/* update power state */
+				clk_mgr_base->clks.pwr_state = DCN_PWR_STATE_LOW_POWER;
+			}
+		}
+	} else {
+		if (!clk_mgr_base->clks.dtbclk_en && new_clocks->dtbclk_en) {
+			dcn316_smu_set_dtbclk(clk_mgr, true);
+			clk_mgr_base->clks.dtbclk_en = new_clocks->dtbclk_en;
+		}
+
+		/* check that we're not already in D0 */
+		if (clk_mgr_base->clks.pwr_state != DCN_PWR_STATE_MISSION_MODE) {
+			union display_idle_optimization_u idle_info = { 0 };
+			dcn316_smu_set_display_idle_optimization(clk_mgr, idle_info.data);
+			/* update power state */
+			clk_mgr_base->clks.pwr_state = DCN_PWR_STATE_MISSION_MODE;
+		}
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dcfclk_khz, clk_mgr_base->clks.dcfclk_khz)) {
+		clk_mgr_base->clks.dcfclk_khz = new_clocks->dcfclk_khz;
+		dcn316_smu_set_hard_min_dcfclk(clk_mgr, clk_mgr_base->clks.dcfclk_khz);
+	}
+
+	if (should_set_clock(safe_to_lower,
+			new_clocks->dcfclk_deep_sleep_khz, clk_mgr_base->clks.dcfclk_deep_sleep_khz)) {
+		clk_mgr_base->clks.dcfclk_deep_sleep_khz = new_clocks->dcfclk_deep_sleep_khz;
+		dcn316_smu_set_min_deep_sleep_dcfclk(clk_mgr, clk_mgr_base->clks.dcfclk_deep_sleep_khz);
+	}
+
+	// workaround: Limit dppclk to 100Mhz to avoid lower eDP panel switch to plus 4K monitor underflow.
+	if (new_clocks->dppclk_khz < 100000)
+		new_clocks->dppclk_khz = 100000;
+
+	if (should_set_clock(safe_to_lower, new_clocks->dppclk_khz, clk_mgr->base.clks.dppclk_khz)) {
+		if (clk_mgr->base.clks.dppclk_khz > new_clocks->dppclk_khz)
+			dpp_clock_lowered = true;
+		clk_mgr_base->clks.dppclk_khz = new_clocks->dppclk_khz;
+		update_dppclk = true;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, clk_mgr_base->clks.dispclk_khz) &&
+	    (new_clocks->dispclk_khz > 0 || (safe_to_lower && display_count == 0))) {
+		int requested_dispclk_khz = new_clocks->dispclk_khz;
+
+		dcn316_disable_otg_wa(clk_mgr_base, context, safe_to_lower, true);
+
+		/* Clamp the requested clock to PMFW based on their limit. */
+		if (dc->debug.min_disp_clk_khz > 0 && requested_dispclk_khz < dc->debug.min_disp_clk_khz)
+			requested_dispclk_khz = dc->debug.min_disp_clk_khz;
+
+		dcn316_smu_set_dispclk(clk_mgr, requested_dispclk_khz);
+		clk_mgr_base->clks.dispclk_khz = new_clocks->dispclk_khz;
+		dcn316_disable_otg_wa(clk_mgr_base, context, safe_to_lower, false);
+
+		update_dispclk = true;
+	}
+
+	if (dpp_clock_lowered) {
+		// increase per DPP DTO before lowering global dppclk
+		dcn20_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
+		dcn316_smu_set_dppclk(clk_mgr, clk_mgr_base->clks.dppclk_khz);
+	} else {
+		// increase global DPPCLK before lowering per DPP DTO
+		if (update_dppclk || update_dispclk)
+			dcn316_smu_set_dppclk(clk_mgr, clk_mgr_base->clks.dppclk_khz);
+		// always update dtos unless clock is lowered and not safe to lower
+		if (new_clocks->dppclk_khz >= dc->current_state->bw_ctx.bw.dcn.clk.dppclk_khz)
+			dcn20_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
+	}
+
+	// notify DMCUB of latest clocks
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.notify_clocks.header.type = DMUB_CMD__CLK_MGR;
+	cmd.notify_clocks.header.sub_type = DMUB_CMD__CLK_MGR_NOTIFY_CLOCKS;
+	cmd.notify_clocks.clocks.dcfclk_khz = clk_mgr_base->clks.dcfclk_khz;
+	cmd.notify_clocks.clocks.dcfclk_deep_sleep_khz =
+		clk_mgr_base->clks.dcfclk_deep_sleep_khz;
+	cmd.notify_clocks.clocks.dispclk_khz = clk_mgr_base->clks.dispclk_khz;
+	cmd.notify_clocks.clocks.dppclk_khz = clk_mgr_base->clks.dppclk_khz;
+
+	dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+static void dcn316_dump_clk_registers(struct clk_state_registers_and_bypass *regs_and_bypass,
+		struct clk_mgr *clk_mgr_base, struct clk_log_info *log_info)
+{
+	return;
+}
+
+static struct clk_bw_params dcn316_bw_params = {
+	.vram_type = Ddr4MemType,
+	.num_channels = 1,
+	.clk_table = {
+		.num_entries = 5,
+	},
+
+};
+
+static struct wm_table ddr4_wm_table = {
+	.entries = {
+		{
+			.wm_inst = WM_A,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 6.09,
+			.sr_enter_plus_exit_time_us = 7.14,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_B,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 10.12,
+			.sr_enter_plus_exit_time_us = 11.48,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_C,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 10.12,
+			.sr_enter_plus_exit_time_us = 11.48,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_D,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 10.12,
+			.sr_enter_plus_exit_time_us = 11.48,
+			.valid = true,
+		},
+	}
+};
+
+static struct wm_table lpddr5_wm_table = {
+	.entries = {
+		{
+			.wm_inst = WM_A,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 11.5,
+			.sr_enter_plus_exit_time_us = 14.5,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_B,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 11.5,
+			.sr_enter_plus_exit_time_us = 14.5,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_C,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 11.5,
+			.sr_enter_plus_exit_time_us = 14.5,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_D,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 11.5,
+			.sr_enter_plus_exit_time_us = 14.5,
+			.valid = true,
+		},
+	}
+};
+
+static DpmClocks_316_t dummy_clocks;
+
+static struct dcn316_watermarks dummy_wms = { 0 };
+
+static void dcn316_build_watermark_ranges(struct clk_bw_params *bw_params, struct dcn316_watermarks *table)
+{
+	int i, num_valid_sets;
+
+	num_valid_sets = 0;
+
+	for (i = 0; i < WM_SET_COUNT; i++) {
+		/* skip empty entries, the smu array has no holes*/
+		if (!bw_params->wm_table.entries[i].valid)
+			continue;
+
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].WmSetting = bw_params->wm_table.entries[i].wm_inst;
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].WmType = bw_params->wm_table.entries[i].wm_type;
+		/* We will not select WM based on fclk, so leave it as unconstrained */
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinClock = 0;
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].MaxClock = 0xFFFF;
+
+		if (table->WatermarkRow[WM_DCFCLK][num_valid_sets].WmType == WM_TYPE_PSTATE_CHG) {
+			if (i == 0)
+				table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinMclk = 0;
+			else {
+				/* add 1 to make it non-overlapping with next lvl */
+				table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinMclk =
+						bw_params->clk_table.entries[i - 1].dcfclk_mhz + 1;
+			}
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets].MaxMclk =
+					bw_params->clk_table.entries[i].dcfclk_mhz;
+
+		} else {
+			/* unconstrained for memory retraining */
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinClock = 0;
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets].MaxClock = 0xFFFF;
+
+			/* Modify previous watermark range to cover up to max */
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets - 1].MaxClock = 0xFFFF;
+		}
+		num_valid_sets++;
+	}
+
+	ASSERT(num_valid_sets != 0); /* Must have at least one set of valid watermarks */
+
+	/* modify the min and max to make sure we cover the whole range*/
+	table->WatermarkRow[WM_DCFCLK][0].MinMclk = 0;
+	table->WatermarkRow[WM_DCFCLK][0].MinClock = 0;
+	table->WatermarkRow[WM_DCFCLK][num_valid_sets - 1].MaxMclk = 0xFFFF;
+	table->WatermarkRow[WM_DCFCLK][num_valid_sets - 1].MaxClock = 0xFFFF;
+
+	/* This is for writeback only, does not matter currently as no writeback support*/
+	table->WatermarkRow[WM_SOCCLK][0].WmSetting = WM_A;
+	table->WatermarkRow[WM_SOCCLK][0].MinClock = 0;
+	table->WatermarkRow[WM_SOCCLK][0].MaxClock = 0xFFFF;
+	table->WatermarkRow[WM_SOCCLK][0].MinMclk = 0;
+	table->WatermarkRow[WM_SOCCLK][0].MaxMclk = 0xFFFF;
+}
+
+static void dcn316_notify_wm_ranges(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct clk_mgr_dcn316 *clk_mgr_dcn316 = TO_CLK_MGR_DCN316(clk_mgr);
+	struct dcn316_watermarks *table = clk_mgr_dcn316->smu_wm_set.wm_set;
+
+	if (!clk_mgr->smu_ver)
+		return;
+
+	if (!table || clk_mgr_dcn316->smu_wm_set.mc_address.quad_part == 0)
+		return;
+
+	memset(table, 0, sizeof(*table));
+
+	dcn316_build_watermark_ranges(clk_mgr_base->bw_params, table);
+
+	dcn316_smu_set_dram_addr_high(clk_mgr,
+			clk_mgr_dcn316->smu_wm_set.mc_address.high_part);
+	dcn316_smu_set_dram_addr_low(clk_mgr,
+			clk_mgr_dcn316->smu_wm_set.mc_address.low_part);
+	dcn316_smu_transfer_wm_table_dram_2_smu(clk_mgr);
+}
+
+static void dcn316_get_dpm_table_from_smu(struct clk_mgr_internal *clk_mgr,
+		struct dcn316_smu_dpm_clks *smu_dpm_clks)
+{
+	DpmClocks_316_t *table = smu_dpm_clks->dpm_clks;
+
+	if (!clk_mgr->smu_ver)
+		return;
+
+	if (!table || smu_dpm_clks->mc_address.quad_part == 0)
+		return;
+
+	memset(table, 0, sizeof(*table));
+
+	dcn316_smu_set_dram_addr_high(clk_mgr,
+			smu_dpm_clks->mc_address.high_part);
+	dcn316_smu_set_dram_addr_low(clk_mgr,
+			smu_dpm_clks->mc_address.low_part);
+	dcn316_smu_transfer_dpm_table_smu_2_dram(clk_mgr);
+}
+
+static uint32_t find_max_clk_value(const uint32_t clocks[], uint32_t num_clocks)
+{
+	uint32_t max = 0;
+	int i;
+
+	for (i = 0; i < num_clocks; ++i) {
+		if (clocks[i] > max)
+			max = clocks[i];
+	}
+
+	return max;
+}
+
+static unsigned int find_clk_for_voltage(
+		const DpmClocks_316_t *clock_table,
+		const uint32_t clocks[],
+		unsigned int voltage)
+{
+	int i;
+	int max_voltage = 0;
+	int clock = 0;
+
+	for (i = 0; i < NUM_SOC_VOLTAGE_LEVELS; i++) {
+		if (clock_table->SocVoltage[i] == voltage) {
+			return clocks[i];
+		} else if (clock_table->SocVoltage[i] >= max_voltage &&
+				clock_table->SocVoltage[i] < voltage) {
+			max_voltage = clock_table->SocVoltage[i];
+			clock = clocks[i];
+		}
+	}
+
+	ASSERT(clock);
+	return clock;
+}
+
+static void dcn316_clk_mgr_helper_populate_bw_params(
+		struct clk_mgr_internal *clk_mgr,
+		struct integrated_info *bios_info,
+		const DpmClocks_316_t *clock_table)
+{
+	int i, j;
+	struct clk_bw_params *bw_params = clk_mgr->base.bw_params;
+	uint32_t max_dispclk = 0, max_dppclk = 0;
+
+	j = -1;
+
+	static_assert(NUM_DF_PSTATE_LEVELS <= MAX_NUM_DPM_LVL,
+		"number of reported pstate levels exceeds maximum");
+
+	/* Find lowest DPM, FCLK is filled in reverse order*/
+
+	for (i = NUM_DF_PSTATE_LEVELS - 1; i >= 0; i--) {
+		if (clock_table->DfPstateTable[i].FClk != 0) {
+			j = i;
+			break;
+		}
+	}
+
+	if (j == -1) {
+		/* clock table is all 0s, just use our own hardcode */
+		ASSERT(0);
+		return;
+	}
+
+	bw_params->clk_table.num_entries = j + 1;
+
+	/* dispclk and dppclk can be max at any voltage, same number of levels for both */
+	if (clock_table->NumDispClkLevelsEnabled <= NUM_DISPCLK_DPM_LEVELS &&
+	    clock_table->NumDispClkLevelsEnabled <= NUM_DPPCLK_DPM_LEVELS) {
+		max_dispclk = find_max_clk_value(clock_table->DispClocks, clock_table->NumDispClkLevelsEnabled);
+		max_dppclk = find_max_clk_value(clock_table->DppClocks, clock_table->NumDispClkLevelsEnabled);
+	} else {
+		ASSERT(0);
+	}
+
+	for (i = 0; i < bw_params->clk_table.num_entries; i++, j--) {
+		int temp;
+
+		bw_params->clk_table.entries[i].fclk_mhz = clock_table->DfPstateTable[j].FClk;
+		bw_params->clk_table.entries[i].memclk_mhz = clock_table->DfPstateTable[j].MemClk;
+		bw_params->clk_table.entries[i].voltage = clock_table->DfPstateTable[j].Voltage;
+		switch (clock_table->DfPstateTable[j].WckRatio) {
+		case WCK_RATIO_1_2:
+			bw_params->clk_table.entries[i].wck_ratio = 2;
+			break;
+		case WCK_RATIO_1_4:
+			bw_params->clk_table.entries[i].wck_ratio = 4;
+			break;
+		default:
+			bw_params->clk_table.entries[i].wck_ratio = 1;
+		}
+		temp = find_clk_for_voltage(clock_table, clock_table->DcfClocks, clock_table->DfPstateTable[j].Voltage);
+		if (temp)
+			bw_params->clk_table.entries[i].dcfclk_mhz = temp;
+		temp = find_clk_for_voltage(clock_table, clock_table->SocClocks, clock_table->DfPstateTable[j].Voltage);
+		if (temp)
+			bw_params->clk_table.entries[i].socclk_mhz = temp;
+		bw_params->clk_table.entries[i].dispclk_mhz = max_dispclk;
+		bw_params->clk_table.entries[i].dppclk_mhz = max_dppclk;
+	}
+
+	bw_params->vram_type = bios_info->memory_type;
+	bw_params->num_channels = bios_info->ma_channel_number;
+	bw_params->dram_channel_width_bytes = bios_info->memory_type == 0x22 ? 8 : 4;
+
+	for (i = 0; i < WM_SET_COUNT; i++) {
+		bw_params->wm_table.entries[i].wm_inst = i;
+
+		if (i >= bw_params->clk_table.num_entries) {
+			bw_params->wm_table.entries[i].valid = false;
+			continue;
+		}
+
+		bw_params->wm_table.entries[i].wm_type = WM_TYPE_PSTATE_CHG;
+		bw_params->wm_table.entries[i].valid = true;
+	}
+}
+
+
+
+static struct clk_mgr_funcs dcn316_funcs = {
+	.enable_pme_wa = dcn316_enable_pme_wa,
+	.get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz,
+	.get_dtb_ref_clk_frequency = dcn31_get_dtb_ref_freq_khz,
+	.update_clocks = dcn316_update_clocks,
+	.init_clocks = dcn31_init_clocks,
+	.are_clock_states_equal = dcn31_are_clock_states_equal,
+	.notify_wm_ranges = dcn316_notify_wm_ranges
+};
+extern struct clk_mgr_funcs dcn3_fpga_funcs;
+
+void dcn316_clk_mgr_construct(
+		struct dc_context *ctx,
+		struct clk_mgr_dcn316 *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg)
+{
+	struct dcn316_smu_dpm_clks smu_dpm_clks = { 0 };
+	struct clk_log_info log_info = {0};
+
+	clk_mgr->base.base.ctx = ctx;
+	clk_mgr->base.base.funcs = &dcn316_funcs;
+
+	clk_mgr->base.pp_smu = pp_smu;
+
+	clk_mgr->base.dccg = dccg;
+	clk_mgr->base.dfs_bypass_disp_clk = 0;
+
+	clk_mgr->base.dprefclk_ss_percentage = 0;
+	clk_mgr->base.dprefclk_ss_divider = 1000;
+	clk_mgr->base.ss_on_dprefclk = false;
+	clk_mgr->base.dfs_ref_freq_khz = 48000;
+
+	clk_mgr->smu_wm_set.wm_set = (struct dcn316_watermarks *)dm_helpers_allocate_gpu_mem(
+				clk_mgr->base.base.ctx,
+				DC_MEM_ALLOC_TYPE_FRAME_BUFFER,
+				sizeof(struct dcn316_watermarks),
+				&clk_mgr->smu_wm_set.mc_address.quad_part);
+
+	if (!clk_mgr->smu_wm_set.wm_set) {
+		clk_mgr->smu_wm_set.wm_set = &dummy_wms;
+		clk_mgr->smu_wm_set.mc_address.quad_part = 0;
+	}
+	ASSERT(clk_mgr->smu_wm_set.wm_set);
+
+	smu_dpm_clks.dpm_clks = (DpmClocks_316_t *)dm_helpers_allocate_gpu_mem(
+				clk_mgr->base.base.ctx,
+				DC_MEM_ALLOC_TYPE_FRAME_BUFFER,
+				sizeof(DpmClocks_316_t),
+				&smu_dpm_clks.mc_address.quad_part);
+
+	if (smu_dpm_clks.dpm_clks == NULL) {
+		smu_dpm_clks.dpm_clks = &dummy_clocks;
+		smu_dpm_clks.mc_address.quad_part = 0;
+	}
+
+	ASSERT(smu_dpm_clks.dpm_clks);
+
+	clk_mgr->base.smu_ver = dcn316_smu_get_smu_version(&clk_mgr->base);
+
+	if (clk_mgr->base.smu_ver > 0)
+		clk_mgr->base.smu_present = true;
+
+	// Skip this for now as it did not work on DCN315, renable during bring up
+	//clk_mgr->base.base.dentist_vco_freq_khz = get_vco_frequency_from_reg(&clk_mgr->base);
+	clk_mgr->base.base.dentist_vco_freq_khz = 2500000;
+
+	/* in case we don't get a value from the register, use default */
+	if (clk_mgr->base.base.dentist_vco_freq_khz == 0)
+		clk_mgr->base.base.dentist_vco_freq_khz = 2500000; /* 2400MHz */
+
+
+	if (ctx->dc_bios->integrated_info->memory_type == LpDdr5MemType) {
+		dcn316_bw_params.wm_table = lpddr5_wm_table;
+	} else {
+		dcn316_bw_params.wm_table = ddr4_wm_table;
+	}
+	/* Saved clocks configured at boot for debug purposes */
+	dcn316_dump_clk_registers(&clk_mgr->base.base.boot_snapshot,
+				  &clk_mgr->base.base, &log_info);
+
+	clk_mgr->base.base.dprefclk_khz = 600000;
+	clk_mgr->base.base.dprefclk_khz = dcn316_smu_get_dpref_clk(&clk_mgr->base);
+	clk_mgr->base.base.clks.ref_dtbclk_khz = clk_mgr->base.base.dprefclk_khz;
+	dce_clock_read_ss_info(&clk_mgr->base);
+	/*clk_mgr->base.dccg->ref_dtbclk_khz =
+	dce_adjust_dp_ref_freq_for_ss(&clk_mgr->base, clk_mgr->base.base.dprefclk_khz);*/
+
+	clk_mgr->base.base.bw_params = &dcn316_bw_params;
+
+	if (clk_mgr->base.base.ctx->dc->debug.pstate_enabled) {
+		dcn316_get_dpm_table_from_smu(&clk_mgr->base, &smu_dpm_clks);
+
+		if (ctx->dc_bios->integrated_info) {
+			dcn316_clk_mgr_helper_populate_bw_params(
+					&clk_mgr->base,
+					ctx->dc_bios->integrated_info,
+					smu_dpm_clks.dpm_clks);
+		}
+	}
+
+	if (smu_dpm_clks.dpm_clks && smu_dpm_clks.mc_address.quad_part != 0)
+		dm_helpers_free_gpu_mem(clk_mgr->base.base.ctx, DC_MEM_ALLOC_TYPE_FRAME_BUFFER,
+				smu_dpm_clks.dpm_clks);
+}
+
+void dcn316_clk_mgr_destroy(struct clk_mgr_internal *clk_mgr_int)
+{
+	struct clk_mgr_dcn316 *clk_mgr = TO_CLK_MGR_DCN316(clk_mgr_int);
+
+	if (clk_mgr->smu_wm_set.wm_set && clk_mgr->smu_wm_set.mc_address.quad_part != 0)
+		dm_helpers_free_gpu_mem(clk_mgr_int->base.ctx, DC_MEM_ALLOC_TYPE_FRAME_BUFFER,
+				clk_mgr->smu_wm_set.wm_set);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_clk_mgr.h
new file mode 100644
index 000000000000..864d1f6cef26
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_clk_mgr.h
@@ -0,0 +1,49 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCN316_CLK_MGR_H__
+#define __DCN316_CLK_MGR_H__
+#include "clk_mgr_internal.h"
+
+struct dcn316_watermarks;
+
+struct dcn316_smu_watermark_set {
+	struct dcn316_watermarks *wm_set;
+	union large_integer mc_address;
+};
+
+struct clk_mgr_dcn316 {
+	struct clk_mgr_internal base;
+	struct dcn316_smu_watermark_set smu_wm_set;
+};
+
+void dcn316_clk_mgr_construct(struct dc_context *ctx,
+		struct clk_mgr_dcn316 *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg);
+
+void dcn316_clk_mgr_destroy(struct clk_mgr_internal *clk_mgr_int);
+
+#endif //__DCN316_CLK_MGR_H__
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_smu.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_smu.c
new file mode 100644
index 000000000000..8b82092b91cd
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_smu.c
@@ -0,0 +1,344 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "core_types.h"
+#include "clk_mgr_internal.h"
+#include "reg_helper.h"
+#include "dm_helpers.h"
+#include "dcn316_smu.h"
+#include "mp/mp_13_0_8_offset.h"
+#include "mp/mp_13_0_8_sh_mask.h"
+
+#define MAX_INSTANCE                                        7
+#define MAX_SEGMENT                                         6
+
+struct IP_BASE_INSTANCE {
+    unsigned int segment[MAX_SEGMENT];
+};
+
+struct IP_BASE {
+    struct IP_BASE_INSTANCE instance[MAX_INSTANCE];
+};
+
+static const struct IP_BASE MP0_BASE = { { { { 0x00016000, 0x00DC0000, 0x00E00000, 0x00E40000, 0x0243FC00, 0 } },
+					{ { 0, 0, 0, 0, 0, 0 } },
+					{ { 0, 0, 0, 0, 0, 0 } },
+					{ { 0, 0, 0, 0, 0, 0 } },
+					{ { 0, 0, 0, 0, 0, 0 } },
+					{ { 0, 0, 0, 0, 0, 0 } },
+					{ { 0, 0, 0, 0, 0, 0 } } } };
+
+#define REG(reg_name) \
+	(MP0_BASE.instance[0].segment[reg ## reg_name ## _BASE_IDX] + reg ## reg_name)
+
+#define FN(reg_name, field) \
+	FD(reg_name##__##field)
+
+#include "logger_types.h"
+#undef DC_LOGGER
+#define DC_LOGGER \
+	CTX->logger
+#define smu_print(str, ...) {DC_LOG_SMU(str, ##__VA_ARGS__); }
+
+#define VBIOSSMC_MSG_TestMessage                  0x01 ///< To check if PMFW is alive and responding. Requirement specified by PMFW team
+#define VBIOSSMC_MSG_GetPmfwVersion               0x02 ///< Get PMFW version
+#define VBIOSSMC_MSG_Spare0                       0x03 ///< Spare0
+#define VBIOSSMC_MSG_SetDispclkFreq               0x04 ///< Set display clock frequency in MHZ
+#define VBIOSSMC_MSG_Spare1                       0x05 ///< Spare1
+#define VBIOSSMC_MSG_SetDppclkFreq                0x06 ///< Set DPP clock frequency in MHZ
+#define VBIOSSMC_MSG_SetHardMinDcfclkByFreq       0x07 ///< Set DCF clock frequency hard min in MHZ
+#define VBIOSSMC_MSG_SetMinDeepSleepDcfclk        0x08 ///< Set DCF clock minimum frequency in deep sleep in MHZ
+#define VBIOSSMC_MSG_SetPhyclkVoltageByFreq       0x09 ///< Set display phy clock frequency in MHZ in case VMIN does not support phy frequency
+#define VBIOSSMC_MSG_GetFclkFrequency             0x0A ///< Get FCLK frequency, return frequemcy in MHZ
+#define VBIOSSMC_MSG_SetDisplayCount              0x0B ///< Inform PMFW of number of display connected
+#define VBIOSSMC_MSG_SPARE						  0x0C ///< SPARE
+#define VBIOSSMC_MSG_UpdatePmeRestore             0x0D ///< To ask PMFW to write into Azalia for PME wake up event
+#define VBIOSSMC_MSG_SetVbiosDramAddrHigh         0x0E ///< Set DRAM address high 32 bits for WM table transfer
+#define VBIOSSMC_MSG_SetVbiosDramAddrLow          0x0F ///< Set DRAM address low 32 bits for WM table transfer
+#define VBIOSSMC_MSG_TransferTableSmu2Dram        0x10 ///< Transfer table from PMFW SRAM to system DRAM
+#define VBIOSSMC_MSG_TransferTableDram2Smu        0x11 ///< Transfer table from system DRAM to PMFW
+#define VBIOSSMC_MSG_SetDisplayIdleOptimizations  0x12 ///< Set Idle state optimization for display off
+#define VBIOSSMC_MSG_GetDprefclkFreq              0x13 ///< Get DPREF clock frequency. Return in MHZ
+#define VBIOSSMC_MSG_GetDtbclkFreq                0x14 ///< Get DPREF clock frequency. Return in MHZ
+#define VBIOSSMC_MSG_SetDtbclkFreq                0x15 ///< Inform PMFW to turn on/off DTB clock arg = 1, turn DTB clock on 600MHz/ arg = 0 turn DTB clock off
+#define VBIOSSMC_Message_Count                    0x16 ///< Total number of VBIS and DAL messages
+
+#define VBIOSSMC_Status_BUSY                      0x0
+#define VBIOSSMC_Result_OK                        0x01 ///< Message Response OK
+#define VBIOSSMC_Result_Failed                    0xFF ///< Message Response Failed
+#define VBIOSSMC_Result_UnknownCmd                0xFE ///< Message Response Unknown Command
+#define VBIOSSMC_Result_CmdRejectedPrereq         0xFD ///< Message Response Command Failed Prerequisite
+#define VBIOSSMC_Result_CmdRejectedBusy           0xFC ///< Message Response Command Rejected due to PMFW is busy. Sender should retry sending this message
+
+/*
+ * Function to be used instead of REG_WAIT macro because the wait ends when
+ * the register is NOT EQUAL to zero, and because the translation in msg_if.h
+ * won't work with REG_WAIT.
+ */
+static uint32_t dcn316_smu_wait_for_response(struct clk_mgr_internal *clk_mgr, unsigned int delay_us, unsigned int max_retries)
+{
+	uint32_t res_val = VBIOSSMC_Status_BUSY;
+
+	do {
+		res_val = REG_READ(MP1_SMN_C2PMSG_91);
+		if (res_val != VBIOSSMC_Status_BUSY)
+			break;
+
+		if (delay_us >= 1000)
+			msleep(delay_us/1000);
+		else if (delay_us > 0)
+			udelay(delay_us);
+	} while (max_retries--);
+
+	return res_val;
+}
+
+static int dcn316_smu_send_msg_with_param(
+		struct clk_mgr_internal *clk_mgr,
+		unsigned int msg_id, unsigned int param)
+{
+	uint32_t result;
+
+	result = dcn316_smu_wait_for_response(clk_mgr, 10, 200000);
+
+	if (result != VBIOSSMC_Result_OK)
+		smu_print("SMU Response was not OK. SMU response after wait received is: %d\n", result);
+
+	if (result == VBIOSSMC_Status_BUSY) {
+		return -1;
+	}
+
+	/* First clear response register */
+	REG_WRITE(MP1_SMN_C2PMSG_91, VBIOSSMC_Status_BUSY);
+
+	/* Set the parameter register for the SMU message, unit is Mhz */
+	REG_WRITE(MP1_SMN_C2PMSG_83, param);
+
+	/* Trigger the message transaction by writing the message ID */
+	REG_WRITE(MP1_SMN_C2PMSG_67, msg_id);
+
+	result = dcn316_smu_wait_for_response(clk_mgr, 10, 200000);
+
+	if (result == VBIOSSMC_Status_BUSY) {
+		ASSERT(0);
+		dm_helpers_smu_timeout(CTX, msg_id, param, 10 * 200000);
+	}
+
+	return REG_READ(MP1_SMN_C2PMSG_83);
+}
+
+int dcn316_smu_get_smu_version(struct clk_mgr_internal *clk_mgr)
+{
+	return dcn316_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_GetPmfwVersion,
+			0);
+}
+
+
+int dcn316_smu_set_dispclk(struct clk_mgr_internal *clk_mgr, int requested_dispclk_khz)
+{
+	int actual_dispclk_set_mhz = -1;
+
+	if (!clk_mgr->smu_present)
+		return requested_dispclk_khz;
+
+	/*  Unit of SMU msg parameter is Mhz */
+	actual_dispclk_set_mhz = dcn316_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDispclkFreq,
+			khz_to_mhz_ceil(requested_dispclk_khz));
+
+	return actual_dispclk_set_mhz * 1000;
+}
+
+int dcn316_smu_set_hard_min_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_dcfclk_khz)
+{
+	int actual_dcfclk_set_mhz = -1;
+
+	if (!clk_mgr->base.ctx->dc->debug.pstate_enabled)
+		return -1;
+
+	if (!clk_mgr->smu_present)
+		return requested_dcfclk_khz;
+
+	actual_dcfclk_set_mhz = dcn316_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetHardMinDcfclkByFreq,
+			khz_to_mhz_ceil(requested_dcfclk_khz));
+
+	return actual_dcfclk_set_mhz * 1000;
+}
+
+int dcn316_smu_set_min_deep_sleep_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_min_ds_dcfclk_khz)
+{
+	int actual_min_ds_dcfclk_mhz = -1;
+
+	if (!clk_mgr->base.ctx->dc->debug.pstate_enabled)
+		return -1;
+
+	if (!clk_mgr->smu_present)
+		return requested_min_ds_dcfclk_khz;
+
+	actual_min_ds_dcfclk_mhz = dcn316_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetMinDeepSleepDcfclk,
+			khz_to_mhz_ceil(requested_min_ds_dcfclk_khz));
+
+	return actual_min_ds_dcfclk_mhz * 1000;
+}
+
+int dcn316_smu_set_dppclk(struct clk_mgr_internal *clk_mgr, int requested_dpp_khz)
+{
+	int actual_dppclk_set_mhz = -1;
+
+	if (!clk_mgr->smu_present)
+		return requested_dpp_khz;
+
+	actual_dppclk_set_mhz = dcn316_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDppclkFreq,
+			khz_to_mhz_ceil(requested_dpp_khz));
+
+	return actual_dppclk_set_mhz * 1000;
+}
+
+void dcn316_smu_set_display_idle_optimization(struct clk_mgr_internal *clk_mgr, uint32_t idle_info)
+{
+	if (!clk_mgr->base.ctx->dc->debug.pstate_enabled)
+		return;
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	//TODO: Work with smu team to define optimization options.
+	dcn316_smu_send_msg_with_param(
+		clk_mgr,
+		VBIOSSMC_MSG_SetDisplayIdleOptimizations,
+		idle_info);
+}
+
+void dcn316_smu_enable_phy_refclk_pwrdwn(struct clk_mgr_internal *clk_mgr, bool enable)
+{
+	union display_idle_optimization_u idle_info = { 0 };
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	if (enable) {
+		idle_info.idle_info.df_request_disabled = 1;
+		idle_info.idle_info.phy_ref_clk_off = 1;
+	}
+
+	dcn316_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDisplayIdleOptimizations,
+			idle_info.data);
+}
+
+void dcn316_smu_set_dram_addr_high(struct clk_mgr_internal *clk_mgr, uint32_t addr_high)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn316_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_SetVbiosDramAddrHigh, addr_high);
+}
+
+void dcn316_smu_set_dram_addr_low(struct clk_mgr_internal *clk_mgr, uint32_t addr_low)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn316_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_SetVbiosDramAddrLow, addr_low);
+}
+
+void dcn316_smu_transfer_dpm_table_smu_2_dram(struct clk_mgr_internal *clk_mgr)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn316_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_TransferTableSmu2Dram, TABLE_DPMCLOCKS);
+}
+
+void dcn316_smu_transfer_wm_table_dram_2_smu(struct clk_mgr_internal *clk_mgr)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn316_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_TransferTableDram2Smu, TABLE_WATERMARKS);
+}
+
+void dcn316_smu_enable_pme_wa(struct clk_mgr_internal *clk_mgr)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn316_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_UpdatePmeRestore,
+			0);
+}
+
+/* Arg = 1: Turn DTB on; 0: Turn DTB CLK OFF. when it is on, it is 600MHZ */
+void dcn316_smu_set_dtbclk(struct clk_mgr_internal *clk_mgr, bool enable)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn316_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDtbclkFreq,
+			enable);
+}
+
+int dcn316_smu_get_dpref_clk(struct clk_mgr_internal *clk_mgr)
+{
+	int dprefclk_get_mhz = -1;
+
+	if (clk_mgr->smu_present) {
+		dprefclk_get_mhz = dcn316_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_GetDprefclkFreq,
+			0);
+	}
+	return (dprefclk_get_mhz * 1000);
+}
+
+int dcn316_smu_get_smu_fclk(struct clk_mgr_internal *clk_mgr)
+{
+	int fclk_get_mhz = -1;
+
+	if (clk_mgr->smu_present) {
+		fclk_get_mhz = dcn316_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_GetFclkFrequency,
+			0);
+	}
+	return (fclk_get_mhz * 1000);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_smu.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_smu.h
new file mode 100644
index 000000000000..2a7293f66515
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_smu.h
@@ -0,0 +1,139 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DAL_DC_316_SMU_H_
+#define DAL_DC_316_SMU_H_
+#include "os_types.h"
+
+#define PMFW_DRIVER_IF_VERSION 4
+
+#define NUM_DCFCLK_DPM_LEVELS   8
+#define NUM_DISPCLK_DPM_LEVELS  8
+#define NUM_DPPCLK_DPM_LEVELS   8
+#define NUM_SOCCLK_DPM_LEVELS   8
+#define NUM_VCN_DPM_LEVELS      8
+#define NUM_SOC_VOLTAGE_LEVELS  8
+#define NUM_DF_PSTATE_LEVELS    4
+
+typedef struct {
+  uint16_t MinClock; // This is either DCFCLK or SOCCLK (in MHz)
+  uint16_t MaxClock; // This is either DCFCLK or SOCCLK (in MHz)
+  uint16_t MinMclk;
+  uint16_t MaxMclk;
+  uint8_t  WmSetting;
+  uint8_t  WmType;  // Used for normal pstate change or memory retraining
+  uint8_t  Padding[2];
+} WatermarkRowGeneric_t;
+
+#define NUM_WM_RANGES 4
+#define WM_PSTATE_CHG 0
+#define WM_RETRAINING 1
+
+typedef enum {
+  WM_SOCCLK = 0,
+  WM_DCFCLK,
+  WM_COUNT,
+} WM_CLOCK_e;
+
+typedef enum{
+  WCK_RATIO_1_1 = 0,  // DDR5, Wck:ck is always 1:1;
+  WCK_RATIO_1_2,
+  WCK_RATIO_1_4,
+  WCK_RATIO_MAX
+} WCK_RATIO_e;
+
+typedef struct {
+  uint32_t FClk;
+  uint32_t MemClk;
+  uint32_t Voltage;
+  uint8_t  WckRatio;
+  uint8_t  Spare[3];
+} DfPstateTable_t;
+
+//Freq in MHz
+//Voltage in milli volts with 2 fractional bits
+typedef struct {
+  uint32_t DcfClocks[NUM_DCFCLK_DPM_LEVELS];
+  uint32_t DispClocks[NUM_DISPCLK_DPM_LEVELS];
+  uint32_t DppClocks[NUM_DPPCLK_DPM_LEVELS];
+  uint32_t SocClocks[NUM_SOCCLK_DPM_LEVELS];
+  uint32_t VClocks[NUM_VCN_DPM_LEVELS];
+  uint32_t DClocks[NUM_VCN_DPM_LEVELS];
+  uint32_t SocVoltage[NUM_SOC_VOLTAGE_LEVELS];
+  DfPstateTable_t DfPstateTable[NUM_DF_PSTATE_LEVELS];
+  uint8_t  NumDcfClkLevelsEnabled;
+  uint8_t  NumDispClkLevelsEnabled; //Applies to both Dispclk and Dppclk
+  uint8_t  NumSocClkLevelsEnabled;
+  uint8_t  VcnClkLevelsEnabled;     //Applies to both Vclk and Dclk
+  uint8_t  NumDfPstatesEnabled;
+  uint8_t  spare[3];
+  uint32_t MinGfxClk;
+  uint32_t MaxGfxClk;
+} DpmClocks_316_t;
+
+struct dcn316_watermarks {
+  // Watermarks
+  WatermarkRowGeneric_t WatermarkRow[WM_COUNT][NUM_WM_RANGES];
+  uint32_t MmHubPadding[7]; // SMU internal use
+};
+
+struct dcn316_smu_dpm_clks {
+	DpmClocks_316_t *dpm_clks;
+	union large_integer mc_address;
+};
+
+#define TABLE_WATERMARKS         1 // Called by DAL through VBIOS
+#define TABLE_DPMCLOCKS          4 // Called by Driver and VBIOS
+
+struct display_idle_optimization {
+	unsigned int df_request_disabled : 1;
+	unsigned int phy_ref_clk_off     : 1;
+	unsigned int s0i2_rdy            : 1;
+	unsigned int reserved            : 29;
+};
+
+union display_idle_optimization_u {
+	struct display_idle_optimization idle_info;
+	uint32_t data;
+};
+
+int dcn316_smu_get_smu_version(struct clk_mgr_internal *clk_mgr);
+int dcn316_smu_set_dispclk(struct clk_mgr_internal *clk_mgr, int requested_dispclk_khz);
+int dcn316_smu_set_hard_min_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_dcfclk_khz);
+int dcn316_smu_set_min_deep_sleep_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_min_ds_dcfclk_khz);
+int dcn316_smu_set_dppclk(struct clk_mgr_internal *clk_mgr, int requested_dpp_khz);
+void dcn316_smu_set_display_idle_optimization(struct clk_mgr_internal *clk_mgr, uint32_t idle_info);
+void dcn316_smu_enable_phy_refclk_pwrdwn(struct clk_mgr_internal *clk_mgr, bool enable);
+void dcn316_smu_set_dram_addr_high(struct clk_mgr_internal *clk_mgr, uint32_t addr_high);
+void dcn316_smu_set_dram_addr_low(struct clk_mgr_internal *clk_mgr, uint32_t addr_low);
+void dcn316_smu_transfer_dpm_table_smu_2_dram(struct clk_mgr_internal *clk_mgr);
+void dcn316_smu_transfer_wm_table_dram_2_smu(struct clk_mgr_internal *clk_mgr);
+void dcn316_smu_request_voltage_via_phyclk(struct clk_mgr_internal *clk_mgr, int requested_phyclk_khz);
+void dcn316_smu_enable_pme_wa(struct clk_mgr_internal *clk_mgr);
+void dcn316_smu_set_dtbclk(struct clk_mgr_internal *clk_mgr, bool enable);
+int dcn316_smu_get_dpref_clk(struct clk_mgr_internal *clk_mgr);
+int dcn316_smu_get_smu_fclk(struct clk_mgr_internal *clk_mgr);
+
+#endif /* DAL_DC_316_SMU_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dalsmc.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dalsmc.h
new file mode 100644
index 000000000000..724a508b0adb
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dalsmc.h
@@ -0,0 +1,74 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+#ifndef DALSMC_H
+#define DALSMC_H
+
+#define DALSMC_VERSION 0x1
+
+// SMU Response Codes:
+#define DALSMC_Result_OK                   0x1
+#define DALSMC_Result_Failed               0xFF
+#define DALSMC_Result_UnknownCmd           0xFE
+#define DALSMC_Result_CmdRejectedPrereq    0xFD
+#define DALSMC_Result_CmdRejectedBusy      0xFC
+
+// Message Definitions:
+#define DALSMC_MSG_TestMessage                    0x1
+#define DALSMC_MSG_GetSmuVersion                  0x2
+#define DALSMC_MSG_GetDriverIfVersion             0x3
+#define DALSMC_MSG_GetMsgHeaderVersion            0x4
+#define DALSMC_MSG_SetDalDramAddrHigh             0x5
+#define DALSMC_MSG_SetDalDramAddrLow              0x6
+#define DALSMC_MSG_TransferTableSmu2Dram          0x7
+#define DALSMC_MSG_TransferTableDram2Smu          0x8
+#define DALSMC_MSG_SetHardMinByFreq               0x9
+#define DALSMC_MSG_SetHardMaxByFreq               0xA
+#define DALSMC_MSG_GetDpmFreqByIndex              0xB
+#define DALSMC_MSG_GetDcModeMaxDpmFreq            0xC
+#define DALSMC_MSG_SetMinDeepSleepDcfclk          0xD
+#define DALSMC_MSG_NumOfDisplays                  0xE
+#define DALSMC_MSG_SetExternalClientDfCstateAllow 0xF
+#define DALSMC_MSG_BacoAudioD3PME                 0x10
+#define DALSMC_MSG_SetFclkSwitchAllow             0x11
+#define DALSMC_MSG_SetCabForUclkPstate            0x12
+#define DALSMC_MSG_SetWorstCaseUclkLatency        0x13
+#define DALSMC_MSG_SetAlwaysWaitDmcubResp         0x14
+#define DALSMC_MSG_ReturnHardMinStatus            0x15
+#define DALSMC_Message_Count                      0x16
+
+#define CHECK_HARD_MIN_CLK_DISPCLK                0x1
+#define CHECK_HARD_MIN_CLK_DPPCLK                 0x2
+#define CHECK_HARD_MIN_CLK_DPREFCLK               0x4
+#define CHECK_HARD_MIN_CLK_DCFCLK                 0x8
+#define CHECK_HARD_MIN_CLK_DTBCLK                 0x10
+#define CHECK_HARD_MIN_CLK_UCLK                   0x20
+
+typedef enum {
+	FCLK_SWITCH_DISALLOW,
+	FCLK_SWITCH_ALLOW,
+} FclkSwitchAllow_e;
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_clk_mgr.c
new file mode 100644
index 000000000000..7da7b41bd092
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_clk_mgr.c
@@ -0,0 +1,1233 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dccg.h"
+#include "clk_mgr_internal.h"
+#include "dcn32/dcn32_clk_mgr_smu_msg.h"
+#include "dcn20/dcn20_clk_mgr.h"
+#include "dce100/dce_clk_mgr.h"
+#include "dcn31/dcn31_clk_mgr.h"
+#include "dcn32/dcn32_clk_mgr.h"
+#include "reg_helper.h"
+#include "core_types.h"
+#include "dm_helpers.h"
+#include "link_service.h"
+#include "dc_state_priv.h"
+#include "atomfirmware.h"
+#include "dcn32_smu13_driver_if.h"
+
+#include "dcn/dcn_3_2_0_offset.h"
+#include "dcn/dcn_3_2_0_sh_mask.h"
+
+#include "dml/dcn32/dcn32_fpu.h"
+
+#define DCN_BASE__INST0_SEG1                       0x000000C0
+
+#define mmCLK1_CLK_PLL_REQ                              0x16E37
+#define mmCLK1_CLK0_DFS_CNTL                            0x16E69
+#define mmCLK1_CLK1_DFS_CNTL                            0x16E6C
+#define mmCLK1_CLK2_DFS_CNTL                            0x16E6F
+#define mmCLK1_CLK3_DFS_CNTL                            0x16E72
+#define mmCLK1_CLK4_DFS_CNTL                            0x16E75
+
+#define mmCLK1_CLK0_CURRENT_CNT                         0x16EE7
+#define mmCLK1_CLK1_CURRENT_CNT                         0x16EE8
+#define mmCLK1_CLK2_CURRENT_CNT                         0x16EE9
+#define mmCLK1_CLK3_CURRENT_CNT                         0x16EEA
+#define mmCLK1_CLK4_CURRENT_CNT                         0x16EEB
+
+#define mmCLK4_CLK0_CURRENT_CNT                         0x1B0C9
+
+#define CLK1_CLK_PLL_REQ__FbMult_int_MASK               0x000001ffUL
+#define CLK1_CLK_PLL_REQ__PllSpineDiv_MASK              0x0000f000UL
+#define CLK1_CLK_PLL_REQ__FbMult_frac_MASK              0xffff0000UL
+#define CLK1_CLK_PLL_REQ__FbMult_int__SHIFT             0x00000000
+#define CLK1_CLK_PLL_REQ__PllSpineDiv__SHIFT            0x0000000c
+#define CLK1_CLK_PLL_REQ__FbMult_frac__SHIFT            0x00000010
+
+#define mmCLK01_CLK0_CLK_PLL_REQ                        0x16E37
+#define mmCLK01_CLK0_CLK0_DFS_CNTL                      0x16E64
+#define mmCLK01_CLK0_CLK1_DFS_CNTL                      0x16E67
+#define mmCLK01_CLK0_CLK2_DFS_CNTL                      0x16E6A
+#define mmCLK01_CLK0_CLK3_DFS_CNTL                      0x16E6D
+#define mmCLK01_CLK0_CLK4_DFS_CNTL                      0x16E70
+
+#define CLK0_CLK_PLL_REQ__FbMult_int_MASK               0x000001ffL
+#define CLK0_CLK_PLL_REQ__PllSpineDiv_MASK              0x0000f000L
+#define CLK0_CLK_PLL_REQ__FbMult_frac_MASK              0xffff0000L
+#define CLK0_CLK_PLL_REQ__FbMult_int__SHIFT             0x00000000
+#define CLK0_CLK_PLL_REQ__PllSpineDiv__SHIFT            0x0000000c
+#define CLK0_CLK_PLL_REQ__FbMult_frac__SHIFT            0x00000010
+
+#undef FN
+#define FN(reg_name, field_name) \
+	clk_mgr->clk_mgr_shift->field_name, clk_mgr->clk_mgr_mask->field_name
+
+#define REG(reg) \
+	(clk_mgr->regs->reg)
+
+#define BASE_INNER(seg) DCN_BASE__INST0_SEG ## seg
+
+#define BASE(seg) BASE_INNER(seg)
+
+#define SR(reg_name)\
+		.reg_name = BASE(reg ## reg_name ## _BASE_IDX) +  \
+					reg ## reg_name
+
+#define CLK_SR_DCN32(reg_name)\
+	.reg_name = mm ## reg_name
+
+static const struct clk_mgr_registers clk_mgr_regs_dcn32 = {
+	CLK_REG_LIST_DCN32()
+};
+
+static const struct clk_mgr_shift clk_mgr_shift_dcn32 = {
+	CLK_COMMON_MASK_SH_LIST_DCN32(__SHIFT)
+};
+
+static const struct clk_mgr_mask clk_mgr_mask_dcn32 = {
+	CLK_COMMON_MASK_SH_LIST_DCN32(_MASK)
+};
+
+
+#define CLK_SR_DCN321(reg_name, block, inst)\
+	.reg_name = mm ## block ## _ ## reg_name
+
+static const struct clk_mgr_registers clk_mgr_regs_dcn321 = {
+	CLK_REG_LIST_DCN321()
+};
+
+static const struct clk_mgr_shift clk_mgr_shift_dcn321 = {
+	CLK_COMMON_MASK_SH_LIST_DCN321(__SHIFT)
+};
+
+static const struct clk_mgr_mask clk_mgr_mask_dcn321 = {
+	CLK_COMMON_MASK_SH_LIST_DCN321(_MASK)
+};
+
+
+/* Query SMU for all clock states for a particular clock */
+static void dcn32_init_single_clock(struct clk_mgr_internal *clk_mgr, PPCLK_e clk, unsigned int *entry_0,
+		unsigned int *num_levels)
+{
+	unsigned int i;
+	char *entry_i = (char *)entry_0;
+
+	uint32_t ret = dcn30_smu_get_dpm_freq_by_index(clk_mgr, clk, 0xFF);
+
+	if (ret & (1 << 31))
+		/* fine-grained, only min and max */
+		*num_levels = 2;
+	else
+		/* discrete, a number of fixed states */
+		/* will set num_levels to 0 on failure */
+		*num_levels = ret & 0xFF;
+
+	/* if the initial message failed, num_levels will be 0 */
+	for (i = 0; i < *num_levels; i++) {
+		*((unsigned int *)entry_i) = (dcn30_smu_get_dpm_freq_by_index(clk_mgr, clk, i) & 0xFFFF);
+		entry_i += sizeof(clk_mgr->base.bw_params->clk_table.entries[0]);
+	}
+}
+
+static void dcn32_build_wm_range_table(struct clk_mgr_internal *clk_mgr)
+{
+	DC_FP_START();
+	dcn32_build_wm_range_table_fpu(clk_mgr);
+	DC_FP_END();
+}
+
+void dcn32_init_clocks(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	unsigned int num_levels;
+	struct clk_limit_num_entries *num_entries_per_clk;
+	unsigned int i;
+
+	if (!clk_mgr_base->bw_params)
+		return;
+
+	num_entries_per_clk = &clk_mgr_base->bw_params->clk_table.num_entries_per_clk;
+
+	memset(&(clk_mgr_base->clks), 0, sizeof(struct dc_clocks));
+	clk_mgr_base->clks.p_state_change_support = true;
+	clk_mgr_base->clks.prev_p_state_change_support = true;
+	clk_mgr_base->clks.fclk_prev_p_state_change_support = true;
+	clk_mgr->smu_present = false;
+	clk_mgr->dpm_present = false;
+
+	if (!clk_mgr_base->force_smu_not_present && dcn30_smu_get_smu_version(clk_mgr, &clk_mgr->smu_ver))
+		clk_mgr->smu_present = true;
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn30_smu_check_driver_if_version(clk_mgr);
+	dcn30_smu_check_msg_header_version(clk_mgr);
+
+	/* DCFCLK */
+	dcn32_init_single_clock(clk_mgr, PPCLK_DCFCLK,
+			&clk_mgr_base->bw_params->clk_table.entries[0].dcfclk_mhz,
+			&num_entries_per_clk->num_dcfclk_levels);
+	clk_mgr_base->bw_params->dc_mode_limit.dcfclk_mhz = dcn30_smu_get_dc_mode_max_dpm_freq(clk_mgr, PPCLK_DCFCLK);
+
+	/* SOCCLK */
+	dcn32_init_single_clock(clk_mgr, PPCLK_SOCCLK,
+					&clk_mgr_base->bw_params->clk_table.entries[0].socclk_mhz,
+					&num_entries_per_clk->num_socclk_levels);
+	clk_mgr_base->bw_params->dc_mode_limit.socclk_mhz = dcn30_smu_get_dc_mode_max_dpm_freq(clk_mgr, PPCLK_SOCCLK);
+
+	/* DTBCLK */
+	if (!clk_mgr->base.ctx->dc->debug.disable_dtb_ref_clk_switch) {
+		dcn32_init_single_clock(clk_mgr, PPCLK_DTBCLK,
+				&clk_mgr_base->bw_params->clk_table.entries[0].dtbclk_mhz,
+				&num_entries_per_clk->num_dtbclk_levels);
+		clk_mgr_base->bw_params->dc_mode_limit.dtbclk_mhz =
+				dcn30_smu_get_dc_mode_max_dpm_freq(clk_mgr, PPCLK_DTBCLK);
+	}
+
+	/* DISPCLK */
+	dcn32_init_single_clock(clk_mgr, PPCLK_DISPCLK,
+			&clk_mgr_base->bw_params->clk_table.entries[0].dispclk_mhz,
+			&num_entries_per_clk->num_dispclk_levels);
+	num_levels = num_entries_per_clk->num_dispclk_levels;
+	clk_mgr_base->bw_params->dc_mode_limit.dispclk_mhz = dcn30_smu_get_dc_mode_max_dpm_freq(clk_mgr, PPCLK_DISPCLK);
+	//HW recommends limit of 1950 MHz in display clock for all DCN3.2.x
+	if (clk_mgr_base->bw_params->dc_mode_limit.dispclk_mhz > 1950)
+		clk_mgr_base->bw_params->dc_mode_limit.dispclk_mhz = 1950;
+
+	/* DPPCLK */
+	dcn32_init_single_clock(clk_mgr, PPCLK_DPPCLK,
+			&clk_mgr_base->bw_params->clk_table.entries[0].dppclk_mhz,
+			&num_entries_per_clk->num_dppclk_levels);
+	num_levels = num_entries_per_clk->num_dppclk_levels;
+	clk_mgr_base->bw_params->dc_mode_limit.dppclk_mhz = dcn30_smu_get_dc_mode_max_dpm_freq(clk_mgr, PPCLK_DPPCLK);
+	//HW recommends limit of 1950 MHz in display clock for all DCN3.2.x
+	if (clk_mgr_base->bw_params->dc_mode_limit.dppclk_mhz > 1950)
+		clk_mgr_base->bw_params->dc_mode_limit.dppclk_mhz = 1950;
+
+	if (num_entries_per_clk->num_dcfclk_levels &&
+			num_entries_per_clk->num_dtbclk_levels &&
+			num_entries_per_clk->num_dispclk_levels)
+		clk_mgr->dpm_present = true;
+
+	if (clk_mgr_base->ctx->dc->debug.min_disp_clk_khz) {
+		for (i = 0; i < num_levels; i++)
+			if (clk_mgr_base->bw_params->clk_table.entries[i].dispclk_mhz
+					< khz_to_mhz_ceil(clk_mgr_base->ctx->dc->debug.min_disp_clk_khz))
+				clk_mgr_base->bw_params->clk_table.entries[i].dispclk_mhz
+					= khz_to_mhz_ceil(clk_mgr_base->ctx->dc->debug.min_disp_clk_khz);
+	}
+	for (i = 0; i < num_levels; i++)
+		if (clk_mgr_base->bw_params->clk_table.entries[i].dispclk_mhz > 1950)
+			clk_mgr_base->bw_params->clk_table.entries[i].dispclk_mhz = 1950;
+
+	if (clk_mgr_base->ctx->dc->debug.min_dpp_clk_khz) {
+		for (i = 0; i < num_levels; i++)
+			if (clk_mgr_base->bw_params->clk_table.entries[i].dppclk_mhz
+					< khz_to_mhz_ceil(clk_mgr_base->ctx->dc->debug.min_dpp_clk_khz))
+				clk_mgr_base->bw_params->clk_table.entries[i].dppclk_mhz
+					= khz_to_mhz_ceil(clk_mgr_base->ctx->dc->debug.min_dpp_clk_khz);
+	}
+
+	for (i = 0; i < num_levels; i++)
+		if (clk_mgr_base->bw_params->clk_table.entries[i].dppclk_mhz > 1950)
+			clk_mgr_base->bw_params->clk_table.entries[i].dppclk_mhz = 1950;
+
+	/* Get UCLK, update bounding box */
+	clk_mgr_base->funcs->get_memclk_states_from_smu(clk_mgr_base);
+
+	/* WM range table */
+	dcn32_build_wm_range_table(clk_mgr);
+}
+
+static void dcn32_update_clocks_update_dtb_dto(struct clk_mgr_internal *clk_mgr,
+			struct dc_state *context,
+			int ref_dtbclk_khz)
+{
+	struct dccg *dccg = clk_mgr->dccg;
+	uint32_t tg_mask = 0;
+	int i;
+
+	for (i = 0; i < clk_mgr->base.ctx->dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i];
+		struct dtbclk_dto_params dto_params = {0};
+
+		/* use mask to program DTO once per tg */
+		if (pipe_ctx->stream_res.tg &&
+				!(tg_mask & (1 << pipe_ctx->stream_res.tg->inst))) {
+			tg_mask |= (1 << pipe_ctx->stream_res.tg->inst);
+
+			dto_params.otg_inst = pipe_ctx->stream_res.tg->inst;
+			dto_params.ref_dtbclk_khz = ref_dtbclk_khz;
+
+			dccg->funcs->set_dtbclk_dto(clk_mgr->dccg, &dto_params);
+			//dccg->funcs->set_audio_dtbclk_dto(clk_mgr->dccg, &dto_params);
+		}
+	}
+}
+
+/* Since DPPCLK request to PMFW needs to be exact (due to DPP DTO programming),
+ * update DPPCLK to be the exact frequency that will be set after the DPPCLK
+ * divider is updated. This will prevent rounding issues that could cause DPP
+ * refclk and DPP DTO to not match up.
+ */
+static void dcn32_update_dppclk_dispclk_freq(struct clk_mgr_internal *clk_mgr, struct dc_clocks *new_clocks)
+{
+	int dpp_divider = 0;
+	int disp_divider = 0;
+
+	if (new_clocks->dppclk_khz) {
+		dpp_divider = DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+				* clk_mgr->base.dentist_vco_freq_khz / new_clocks->dppclk_khz;
+		new_clocks->dppclk_khz = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR * clk_mgr->base.dentist_vco_freq_khz) / dpp_divider;
+	}
+	if (new_clocks->dispclk_khz > 0) {
+		disp_divider = DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+				* clk_mgr->base.dentist_vco_freq_khz / new_clocks->dispclk_khz;
+		new_clocks->dispclk_khz = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR * clk_mgr->base.dentist_vco_freq_khz) / disp_divider;
+	}
+}
+
+void dcn32_update_clocks_update_dpp_dto(struct clk_mgr_internal *clk_mgr,
+		struct dc_state *context, bool safe_to_lower)
+{
+	int i;
+
+	clk_mgr->dccg->ref_dppclk = clk_mgr->base.clks.dppclk_khz;
+	for (i = 0; i < clk_mgr->base.ctx->dc->res_pool->pipe_count; i++) {
+		int dpp_inst = 0, dppclk_khz, prev_dppclk_khz;
+
+		dppclk_khz = context->res_ctx.pipe_ctx[i].plane_res.bw.dppclk_khz;
+
+		if (context->res_ctx.pipe_ctx[i].plane_res.dpp)
+			dpp_inst = context->res_ctx.pipe_ctx[i].plane_res.dpp->inst;
+		else if (!context->res_ctx.pipe_ctx[i].plane_res.dpp && dppclk_khz == 0) {
+			/* dpp == NULL && dppclk_khz == 0 is valid because of pipe harvesting.
+			 * In this case just continue in loop
+			 */
+			continue;
+		} else if (!context->res_ctx.pipe_ctx[i].plane_res.dpp && dppclk_khz > 0) {
+			/* The software state is not valid if dpp resource is NULL and
+			 * dppclk_khz > 0.
+			 */
+			ASSERT(false);
+			continue;
+		}
+
+		prev_dppclk_khz = clk_mgr->dccg->pipe_dppclk_khz[i];
+
+		if (safe_to_lower || prev_dppclk_khz < dppclk_khz)
+			clk_mgr->dccg->funcs->update_dpp_dto(
+							clk_mgr->dccg, dpp_inst, dppclk_khz);
+	}
+}
+
+static void dcn32_update_clocks_update_dentist(
+		struct clk_mgr_internal *clk_mgr,
+		struct dc_state *context)
+{
+	uint32_t new_disp_divider = 0;
+	uint32_t new_dispclk_wdivider = 0;
+	uint32_t old_dispclk_wdivider = 0;
+	uint32_t i;
+	uint32_t dentist_dispclk_wdivider_readback = 0;
+	struct dc *dc = clk_mgr->base.ctx->dc;
+
+	if (clk_mgr->base.clks.dispclk_khz == 0)
+		return;
+
+	new_disp_divider = DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+			* clk_mgr->base.dentist_vco_freq_khz / clk_mgr->base.clks.dispclk_khz;
+
+	new_dispclk_wdivider = dentist_get_did_from_divider(new_disp_divider);
+	REG_GET(DENTIST_DISPCLK_CNTL,
+			DENTIST_DISPCLK_WDIVIDER, &old_dispclk_wdivider);
+
+	/* When changing divider to or from 127, some extra programming is required to prevent corruption */
+	if (old_dispclk_wdivider == 127 && new_dispclk_wdivider != 127) {
+		for (i = 0; i < clk_mgr->base.ctx->dc->res_pool->pipe_count; i++) {
+			struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i];
+			uint32_t fifo_level;
+			struct dccg *dccg = clk_mgr->base.ctx->dc->res_pool->dccg;
+			struct stream_encoder *stream_enc = pipe_ctx->stream_res.stream_enc;
+			int32_t N;
+			int32_t j;
+
+			if (!resource_is_pipe_type(pipe_ctx, OTG_MASTER))
+				continue;
+			/* Virtual encoders don't have this function */
+			if (!stream_enc->funcs->get_fifo_cal_average_level)
+				continue;
+			fifo_level = stream_enc->funcs->get_fifo_cal_average_level(
+					stream_enc);
+			N = fifo_level / 4;
+			dccg->funcs->set_fifo_errdet_ovr_en(
+					dccg,
+					true);
+			for (j = 0; j < N - 4; j++)
+				dccg->funcs->otg_drop_pixel(
+						dccg,
+						pipe_ctx->stream_res.tg->inst);
+			dccg->funcs->set_fifo_errdet_ovr_en(
+					dccg,
+					false);
+		}
+	} else if (new_dispclk_wdivider == 127 && old_dispclk_wdivider != 127) {
+		/* request clock with 126 divider first */
+		uint32_t temp_disp_divider = dentist_get_divider_from_did(126);
+		uint32_t temp_dispclk_khz = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR * clk_mgr->base.dentist_vco_freq_khz) / temp_disp_divider;
+
+		if (clk_mgr->smu_present)
+			/*
+			 * SMU uses discrete dispclk presets. We applied
+			 * the same formula to increase our dppclk_khz
+			 * to the next matching discrete value. By
+			 * contract, we should use the preset dispclk
+			 * floored in Mhz to describe the intended clock.
+			 */
+			dcn32_smu_set_hard_min_by_freq(clk_mgr, PPCLK_DISPCLK,
+					khz_to_mhz_floor(temp_dispclk_khz));
+
+		if (dc->debug.override_dispclk_programming) {
+			REG_GET(DENTIST_DISPCLK_CNTL,
+					DENTIST_DISPCLK_WDIVIDER, &dentist_dispclk_wdivider_readback);
+
+			if (dentist_dispclk_wdivider_readback != 126) {
+				REG_UPDATE(DENTIST_DISPCLK_CNTL,
+						DENTIST_DISPCLK_WDIVIDER, 126);
+				REG_WAIT(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_CHG_DONE, 1, 50, 2000);
+			}
+		}
+
+		for (i = 0; i < clk_mgr->base.ctx->dc->res_pool->pipe_count; i++) {
+			struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i];
+			struct dccg *dccg = clk_mgr->base.ctx->dc->res_pool->dccg;
+			struct stream_encoder *stream_enc = pipe_ctx->stream_res.stream_enc;
+			uint32_t fifo_level;
+			int32_t N;
+			int32_t j;
+
+			if (!resource_is_pipe_type(pipe_ctx, OTG_MASTER))
+				continue;
+			/* Virtual encoders don't have this function */
+			if (!stream_enc->funcs->get_fifo_cal_average_level)
+				continue;
+			fifo_level = stream_enc->funcs->get_fifo_cal_average_level(
+					stream_enc);
+			N = fifo_level / 4;
+			dccg->funcs->set_fifo_errdet_ovr_en(dccg, true);
+			for (j = 0; j < 12 - N; j++)
+				dccg->funcs->otg_add_pixel(dccg,
+						pipe_ctx->stream_res.tg->inst);
+			dccg->funcs->set_fifo_errdet_ovr_en(dccg, false);
+		}
+	}
+
+	/* do requested DISPCLK updates*/
+	if (clk_mgr->smu_present)
+		/*
+		 * SMU uses discrete dispclk presets. We applied
+		 * the same formula to increase our dppclk_khz
+		 * to the next matching discrete value. By
+		 * contract, we should use the preset dispclk
+		 * floored in Mhz to describe the intended clock.
+		 */
+		dcn32_smu_set_hard_min_by_freq(clk_mgr, PPCLK_DISPCLK,
+				khz_to_mhz_floor(clk_mgr->base.clks.dispclk_khz));
+
+	if (dc->debug.override_dispclk_programming) {
+		REG_GET(DENTIST_DISPCLK_CNTL,
+				DENTIST_DISPCLK_WDIVIDER, &dentist_dispclk_wdivider_readback);
+
+		if (dentist_dispclk_wdivider_readback > new_dispclk_wdivider) {
+			REG_UPDATE(DENTIST_DISPCLK_CNTL,
+					DENTIST_DISPCLK_WDIVIDER, new_dispclk_wdivider);
+			REG_WAIT(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_CHG_DONE, 1, 50, 2000);
+		}
+	}
+
+}
+
+static int dcn32_get_dispclk_from_dentist(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	uint32_t dispclk_wdivider;
+	int disp_divider;
+
+	REG_GET(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_WDIVIDER, &dispclk_wdivider);
+	disp_divider = dentist_get_divider_from_did(dispclk_wdivider);
+
+	/* Return DISPCLK freq in Khz */
+	if (disp_divider)
+		return (DENTIST_DIVIDER_RANGE_SCALE_FACTOR * clk_mgr->base.dentist_vco_freq_khz) / disp_divider;
+
+	return 0;
+}
+
+static bool dcn32_check_native_scaling(struct pipe_ctx *pipe)
+{
+	bool is_native_scaling = false;
+	int width = pipe->plane_state->src_rect.width;
+	int height = pipe->plane_state->src_rect.height;
+
+	if (pipe->stream->timing.h_addressable == width &&
+			pipe->stream->timing.v_addressable == height &&
+			pipe->plane_state->dst_rect.width == width &&
+			pipe->plane_state->dst_rect.height == height)
+		is_native_scaling = true;
+
+	return is_native_scaling;
+}
+
+static void dcn32_auto_dpm_test_log(
+		struct dc_clocks *new_clocks,
+		struct clk_mgr_internal *clk_mgr,
+		struct dc_state *context)
+{
+	unsigned int dispclk_khz_reg, dppclk_khz_reg, dprefclk_khz_reg, dcfclk_khz_reg, dtbclk_khz_reg,
+				 fclk_khz_reg, mall_ss_size_bytes;
+	int dramclk_khz_override, fclk_khz_override, num_fclk_levels;
+
+	struct pipe_ctx *pipe_ctx_list[MAX_PIPES];
+	int active_pipe_count = 0;
+
+	for (int i = 0; i < MAX_PIPES; i++) {
+		struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i];
+
+		if (pipe_ctx->stream && dc_state_get_pipe_subvp_type(context, pipe_ctx) != SUBVP_PHANTOM) {
+			pipe_ctx_list[active_pipe_count] = pipe_ctx;
+			active_pipe_count++;
+		}
+	}
+
+	msleep(5);
+
+	mall_ss_size_bytes = context->bw_ctx.bw.dcn.mall_ss_size_bytes;
+
+    dispclk_khz_reg    = REG_READ(CLK1_CLK0_CURRENT_CNT); // DISPCLK
+    dppclk_khz_reg     = REG_READ(CLK1_CLK1_CURRENT_CNT); // DPPCLK
+    dprefclk_khz_reg   = REG_READ(CLK1_CLK2_CURRENT_CNT); // DPREFCLK
+    dcfclk_khz_reg     = REG_READ(CLK1_CLK3_CURRENT_CNT); // DCFCLK
+    dtbclk_khz_reg     = REG_READ(CLK1_CLK4_CURRENT_CNT); // DTBCLK
+    fclk_khz_reg       = REG_READ(CLK4_CLK0_CURRENT_CNT); // FCLK
+
+    // Overrides for these clocks in case there is no p_state change support
+    dramclk_khz_override = new_clocks->dramclk_khz;
+    fclk_khz_override = new_clocks->fclk_khz;
+
+    num_fclk_levels = clk_mgr->base.bw_params->clk_table.num_entries_per_clk.num_fclk_levels - 1;
+
+    if (!new_clocks->p_state_change_support) {
+	    dramclk_khz_override = clk_mgr->base.bw_params->max_memclk_mhz * 1000;
+    }
+    if (!new_clocks->fclk_p_state_change_support) {
+	    fclk_khz_override = clk_mgr->base.bw_params->clk_table.entries[num_fclk_levels].fclk_mhz * 1000;
+    }
+
+	////////////////////////////////////////////////////////////////////////////
+	//	IMPORTANT: 	When adding more clocks to these logs, do NOT put a newline
+	//	 			anywhere other than at the very end of the string.
+	//
+	//	Formatting example (make sure to have " - " between each entry):
+	//
+	//				AutoDPMTest: clk1:%d - clk2:%d - clk3:%d - clk4:%d\n"
+	////////////////////////////////////////////////////////////////////////////
+	if (active_pipe_count > 0 &&
+		new_clocks->dramclk_khz > 0 &&
+		new_clocks->fclk_khz > 0 &&
+		new_clocks->dcfclk_khz > 0 &&
+		new_clocks->dppclk_khz > 0) {
+
+		uint32_t pix_clk_list[MAX_PIPES] = {0};
+		int p_state_list[MAX_PIPES] = {0};
+		int disp_src_width_list[MAX_PIPES] = {0};
+		int disp_src_height_list[MAX_PIPES] = {0};
+		uint64_t disp_src_refresh_list[MAX_PIPES] = {0};
+		bool is_scaled_list[MAX_PIPES] = {0};
+
+		for (int i = 0; i < active_pipe_count; i++) {
+			struct pipe_ctx *curr_pipe_ctx = pipe_ctx_list[i];
+			uint64_t refresh_rate;
+
+			pix_clk_list[i] = curr_pipe_ctx->stream->timing.pix_clk_100hz;
+			p_state_list[i] = curr_pipe_ctx->p_state_type;
+
+			refresh_rate = (curr_pipe_ctx->stream->timing.pix_clk_100hz * (uint64_t)100 +
+				curr_pipe_ctx->stream->timing.v_total * (uint64_t)curr_pipe_ctx->stream->timing.h_total - (uint64_t)1);
+			refresh_rate = div_u64(refresh_rate, curr_pipe_ctx->stream->timing.v_total);
+			refresh_rate = div_u64(refresh_rate, curr_pipe_ctx->stream->timing.h_total);
+			disp_src_refresh_list[i] = refresh_rate;
+
+			if (curr_pipe_ctx->plane_state) {
+				is_scaled_list[i] = !(dcn32_check_native_scaling(curr_pipe_ctx));
+				disp_src_width_list[i] = curr_pipe_ctx->plane_state->src_rect.width;
+				disp_src_height_list[i] = curr_pipe_ctx->plane_state->src_rect.height;
+			}
+		}
+
+		DC_LOG_AUTO_DPM_TEST("AutoDPMTest: dramclk:%d - fclk:%d - "
+			"dcfclk:%d - dppclk:%d - dispclk_hw:%d - "
+			"dppclk_hw:%d - dprefclk_hw:%d - dcfclk_hw:%d - "
+			"dtbclk_hw:%d - fclk_hw:%d - pix_clk_0:%d - pix_clk_1:%d - "
+			"pix_clk_2:%d - pix_clk_3:%d - mall_ss_size:%d - p_state_type_0:%d - "
+			"p_state_type_1:%d - p_state_type_2:%d - p_state_type_3:%d - "
+			"pix_width_0:%d - pix_height_0:%d - refresh_rate_0:%lld - is_scaled_0:%d - "
+			"pix_width_1:%d - pix_height_1:%d - refresh_rate_1:%lld - is_scaled_1:%d - "
+			"pix_width_2:%d - pix_height_2:%d - refresh_rate_2:%lld - is_scaled_2:%d - "
+			"pix_width_3:%d - pix_height_3:%d - refresh_rate_3:%lld - is_scaled_3:%d - LOG_END\n",
+			dramclk_khz_override,
+			fclk_khz_override,
+			new_clocks->dcfclk_khz,
+			new_clocks->dppclk_khz,
+			dispclk_khz_reg,
+			dppclk_khz_reg,
+			dprefclk_khz_reg,
+			dcfclk_khz_reg,
+			dtbclk_khz_reg,
+			fclk_khz_reg,
+			pix_clk_list[0], pix_clk_list[1], pix_clk_list[3], pix_clk_list[2],
+			mall_ss_size_bytes,
+			p_state_list[0], p_state_list[1], p_state_list[2], p_state_list[3],
+			disp_src_width_list[0], disp_src_height_list[0], disp_src_refresh_list[0], is_scaled_list[0],
+			disp_src_width_list[1], disp_src_height_list[1], disp_src_refresh_list[1], is_scaled_list[1],
+			disp_src_width_list[2], disp_src_height_list[2], disp_src_refresh_list[2], is_scaled_list[2],
+			disp_src_width_list[3], disp_src_height_list[3], disp_src_refresh_list[3], is_scaled_list[3]);
+	}
+}
+
+static void dcn32_update_clocks(struct clk_mgr *clk_mgr_base,
+			struct dc_state *context,
+			bool safe_to_lower)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dc_clocks *new_clocks = &context->bw_ctx.bw.dcn.clk;
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	int display_count;
+	bool update_dppclk = false;
+	bool update_dispclk = false;
+	bool enter_display_off = false;
+	bool dpp_clock_lowered = false;
+	struct dmcu *dmcu = clk_mgr_base->ctx->dc->res_pool->dmcu;
+	bool force_reset = false;
+	bool update_uclk = false, update_fclk = false;
+	bool p_state_change_support;
+	bool fclk_p_state_change_support;
+
+	if (clk_mgr_base->clks.dispclk_khz == 0 ||
+			(dc->debug.force_clock_mode & 0x1)) {
+		/* This is from resume or boot up, if forced_clock cfg option used,
+		 * we bypass program dispclk and DPPCLK, but need set them for S3.
+		 */
+		force_reset = true;
+
+		dcn2_read_clocks_from_hw_dentist(clk_mgr_base);
+
+		/* Force_clock_mode 0x1:  force reset the clock even it is the same clock
+		 * as long as it is in Passive level.
+		 */
+	}
+	display_count = clk_mgr_helper_get_active_display_cnt(dc, context);
+
+	if (display_count == 0)
+		enter_display_off = true;
+
+	if (clk_mgr->smu_present) {
+		if (enter_display_off == safe_to_lower)
+			dcn30_smu_set_num_of_displays(clk_mgr, display_count);
+
+		clk_mgr_base->clks.fclk_prev_p_state_change_support = clk_mgr_base->clks.fclk_p_state_change_support;
+
+		fclk_p_state_change_support = new_clocks->fclk_p_state_change_support;
+
+		if (should_update_pstate_support(safe_to_lower, fclk_p_state_change_support, clk_mgr_base->clks.fclk_p_state_change_support) &&
+				!dc->work_arounds.clock_update_disable_mask.fclk) {
+			clk_mgr_base->clks.fclk_p_state_change_support = fclk_p_state_change_support;
+
+			/* To enable FCLK P-state switching, send FCLK_PSTATE_SUPPORTED message to PMFW */
+			if (clk_mgr_base->ctx->dce_version != DCN_VERSION_3_21 && clk_mgr_base->clks.fclk_p_state_change_support) {
+				/* Handle the code for sending a message to PMFW that FCLK P-state change is supported */
+				dcn32_smu_send_fclk_pstate_message(clk_mgr, FCLK_PSTATE_SUPPORTED);
+			}
+		}
+
+		if (dc->debug.force_min_dcfclk_mhz > 0)
+			new_clocks->dcfclk_khz = (new_clocks->dcfclk_khz > (dc->debug.force_min_dcfclk_mhz * 1000)) ?
+					new_clocks->dcfclk_khz : (dc->debug.force_min_dcfclk_mhz * 1000);
+
+		if (should_set_clock(safe_to_lower, new_clocks->dcfclk_khz, clk_mgr_base->clks.dcfclk_khz) &&
+				!dc->work_arounds.clock_update_disable_mask.dcfclk) {
+			clk_mgr_base->clks.dcfclk_khz = new_clocks->dcfclk_khz;
+			dcn32_smu_set_hard_min_by_freq(clk_mgr, PPCLK_DCFCLK, khz_to_mhz_ceil(clk_mgr_base->clks.dcfclk_khz));
+		}
+
+		if (should_set_clock(safe_to_lower, new_clocks->dcfclk_deep_sleep_khz, clk_mgr_base->clks.dcfclk_deep_sleep_khz) &&
+				!dc->work_arounds.clock_update_disable_mask.dcfclk_ds) {
+			clk_mgr_base->clks.dcfclk_deep_sleep_khz = new_clocks->dcfclk_deep_sleep_khz;
+			dcn30_smu_set_min_deep_sleep_dcef_clk(clk_mgr, khz_to_mhz_ceil(clk_mgr_base->clks.dcfclk_deep_sleep_khz));
+		}
+
+		if (should_set_clock(safe_to_lower, new_clocks->socclk_khz, clk_mgr_base->clks.socclk_khz))
+			/* We don't actually care about socclk, don't notify SMU of hard min */
+			clk_mgr_base->clks.socclk_khz = new_clocks->socclk_khz;
+
+		clk_mgr_base->clks.prev_p_state_change_support = clk_mgr_base->clks.p_state_change_support;
+		clk_mgr_base->clks.prev_num_ways = clk_mgr_base->clks.num_ways;
+
+		if (clk_mgr_base->clks.num_ways != new_clocks->num_ways &&
+				clk_mgr_base->clks.num_ways < new_clocks->num_ways) {
+			clk_mgr_base->clks.num_ways = new_clocks->num_ways;
+			dcn32_smu_send_cab_for_uclk_message(clk_mgr, clk_mgr_base->clks.num_ways);
+		}
+
+		p_state_change_support = new_clocks->p_state_change_support;
+		if (should_update_pstate_support(safe_to_lower, p_state_change_support, clk_mgr_base->clks.p_state_change_support) &&
+				!dc->work_arounds.clock_update_disable_mask.uclk) {
+			clk_mgr_base->clks.p_state_change_support = p_state_change_support;
+
+			/* to disable P-State switching, set UCLK min = max */
+			if (!clk_mgr_base->clks.p_state_change_support) {
+				if (dc->clk_mgr->dc_mode_softmax_enabled) {
+					/* On DCN32x we will never have the functional UCLK min above the softmax
+					 * since we calculate mode support based on softmax being the max UCLK
+					 * frequency.
+					 */
+					if (dc->debug.disable_dc_mode_overwrite) {
+						dcn30_smu_set_hard_max_by_freq(clk_mgr, PPCLK_UCLK, dc->clk_mgr->bw_params->max_memclk_mhz);
+						dcn32_smu_set_hard_min_by_freq(clk_mgr, PPCLK_UCLK, dc->clk_mgr->bw_params->max_memclk_mhz);
+					} else
+						dcn32_smu_set_hard_min_by_freq(clk_mgr, PPCLK_UCLK,
+								dc->clk_mgr->bw_params->dc_mode_softmax_memclk);
+				} else {
+					dcn32_smu_set_hard_min_by_freq(clk_mgr, PPCLK_UCLK, dc->clk_mgr->bw_params->max_memclk_mhz);
+				}
+			}
+		}
+
+		if (context->bw_ctx.bw.dcn.clk.fw_based_mclk_switching)
+			dcn32_smu_wait_for_dmub_ack_mclk(clk_mgr, true);
+		else
+			dcn32_smu_wait_for_dmub_ack_mclk(clk_mgr, false);
+
+		/* Always update saved value, even if new value not set due to P-State switching unsupported. Also check safe_to_lower for FCLK */
+		if (safe_to_lower && (clk_mgr_base->clks.fclk_p_state_change_support != clk_mgr_base->clks.fclk_prev_p_state_change_support)) {
+			update_fclk = true;
+		}
+
+		if (clk_mgr_base->ctx->dce_version != DCN_VERSION_3_21 && !clk_mgr_base->clks.fclk_p_state_change_support && update_fclk &&
+				!dc->work_arounds.clock_update_disable_mask.fclk) {
+			/* Handle code for sending a message to PMFW that FCLK P-state change is not supported */
+			dcn32_smu_send_fclk_pstate_message(clk_mgr, FCLK_PSTATE_NOTSUPPORTED);
+		}
+
+		/* Always update saved value, even if new value not set due to P-State switching unsupported */
+		if (should_set_clock(safe_to_lower, new_clocks->dramclk_khz, clk_mgr_base->clks.dramclk_khz) &&
+				!dc->work_arounds.clock_update_disable_mask.uclk) {
+			clk_mgr_base->clks.dramclk_khz = new_clocks->dramclk_khz;
+			update_uclk = true;
+		}
+
+		/* set UCLK to requested value if P-State switching is supported, or to re-enable P-State switching */
+		if (clk_mgr_base->clks.p_state_change_support &&
+				(update_uclk || !clk_mgr_base->clks.prev_p_state_change_support) &&
+				!dc->work_arounds.clock_update_disable_mask.uclk) {
+			if (dc->clk_mgr->dc_mode_softmax_enabled && dc->debug.disable_dc_mode_overwrite)
+				dcn30_smu_set_hard_max_by_freq(clk_mgr, PPCLK_UCLK,
+						max((int)dc->clk_mgr->bw_params->dc_mode_softmax_memclk, khz_to_mhz_ceil(clk_mgr_base->clks.dramclk_khz)));
+
+			dcn32_smu_set_hard_min_by_freq(clk_mgr, PPCLK_UCLK, khz_to_mhz_ceil(clk_mgr_base->clks.dramclk_khz));
+		}
+
+		if (clk_mgr_base->clks.num_ways != new_clocks->num_ways &&
+				clk_mgr_base->clks.num_ways > new_clocks->num_ways) {
+			clk_mgr_base->clks.num_ways = new_clocks->num_ways;
+			dcn32_smu_send_cab_for_uclk_message(clk_mgr, clk_mgr_base->clks.num_ways);
+		}
+	}
+
+	dcn32_update_dppclk_dispclk_freq(clk_mgr, new_clocks);
+	if (should_set_clock(safe_to_lower, new_clocks->dppclk_khz, clk_mgr_base->clks.dppclk_khz)) {
+		if (clk_mgr_base->clks.dppclk_khz > new_clocks->dppclk_khz)
+			dpp_clock_lowered = true;
+
+		clk_mgr_base->clks.dppclk_khz = new_clocks->dppclk_khz;
+
+		if (clk_mgr->smu_present && !dpp_clock_lowered)
+			/*
+			 * SMU uses discrete dppclk presets. We applied
+			 * the same formula to increase our dppclk_khz
+			 * to the next matching discrete value. By
+			 * contract, we should use the preset dppclk
+			 * floored in Mhz to describe the intended clock.
+			 */
+			dcn32_smu_set_hard_min_by_freq(clk_mgr, PPCLK_DPPCLK,
+					khz_to_mhz_floor(clk_mgr_base->clks.dppclk_khz));
+
+		update_dppclk = true;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, clk_mgr_base->clks.dispclk_khz)) {
+		clk_mgr_base->clks.dispclk_khz = new_clocks->dispclk_khz;
+
+		update_dispclk = true;
+	}
+
+	if (!new_clocks->dtbclk_en) {
+		new_clocks->ref_dtbclk_khz = clk_mgr_base->bw_params->clk_table.entries[0].dtbclk_mhz * 1000;
+	}
+
+	/* clock limits are received with MHz precision, divide by 1000 to prevent setting clocks at every call */
+	if (!dc->debug.disable_dtb_ref_clk_switch &&
+			should_set_clock(safe_to_lower, new_clocks->ref_dtbclk_khz / 1000, clk_mgr_base->clks.ref_dtbclk_khz / 1000)) {
+		/* DCCG requires KHz precision for DTBCLK */
+		clk_mgr_base->clks.ref_dtbclk_khz =
+				dcn32_smu_set_hard_min_by_freq(clk_mgr, PPCLK_DTBCLK, khz_to_mhz_ceil(new_clocks->ref_dtbclk_khz));
+
+		dcn32_update_clocks_update_dtb_dto(clk_mgr, context, clk_mgr_base->clks.ref_dtbclk_khz);
+	}
+
+	if (dc->config.forced_clocks == false || (force_reset && safe_to_lower)) {
+		if (dpp_clock_lowered) {
+			/* if clock is being lowered, increase DTO before lowering refclk */
+			dcn32_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
+			dcn32_update_clocks_update_dentist(clk_mgr, context);
+			if (clk_mgr->smu_present)
+				/*
+				 * SMU uses discrete dppclk presets. We applied
+				 * the same formula to increase our dppclk_khz
+				 * to the next matching discrete value. By
+				 * contract, we should use the preset dppclk
+				 * floored in Mhz to describe the intended clock.
+				 */
+				dcn32_smu_set_hard_min_by_freq(clk_mgr, PPCLK_DPPCLK,
+						khz_to_mhz_floor(clk_mgr_base->clks.dppclk_khz));
+		} else {
+			/* if clock is being raised, increase refclk before lowering DTO */
+			if (update_dppclk || update_dispclk)
+				dcn32_update_clocks_update_dentist(clk_mgr, context);
+			/* There is a check inside dcn20_update_clocks_update_dpp_dto which ensures
+			 * that we do not lower dto when it is not safe to lower. We do not need to
+			 * compare the current and new dppclk before calling this function.
+			 */
+			dcn32_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
+		}
+	}
+
+	if (update_dispclk && dmcu && dmcu->funcs->is_dmcu_initialized(dmcu))
+		/*update dmcu for wait_loop count*/
+		dmcu->funcs->set_psr_wait_loop(dmcu,
+				clk_mgr_base->clks.dispclk_khz / 1000 / 7);
+
+	if (dc->config.enable_auto_dpm_test_logs) {
+	    dcn32_auto_dpm_test_log(new_clocks, clk_mgr, context);
+	}
+}
+
+static uint32_t dcn32_get_vco_frequency_from_reg(struct clk_mgr_internal *clk_mgr)
+{
+		struct fixed31_32 pll_req;
+		uint32_t pll_req_reg = 0;
+
+		/* get FbMult value */
+		if (ASICREV_IS_GC_11_0_2(clk_mgr->base.ctx->asic_id.hw_internal_rev))
+			pll_req_reg = REG_READ(CLK0_CLK_PLL_REQ);
+		else
+			pll_req_reg = REG_READ(CLK1_CLK_PLL_REQ);
+
+		/* set up a fixed-point number
+		 * this works because the int part is on the right edge of the register
+		 * and the frac part is on the left edge
+		 */
+		pll_req = dc_fixpt_from_int(pll_req_reg & clk_mgr->clk_mgr_mask->FbMult_int);
+		pll_req.value |= pll_req_reg & clk_mgr->clk_mgr_mask->FbMult_frac;
+
+		/* multiply by REFCLK period */
+		pll_req = dc_fixpt_mul_int(pll_req, clk_mgr->dfs_ref_freq_khz);
+
+		return dc_fixpt_floor(pll_req);
+}
+
+static void dcn32_dump_clk_registers(struct clk_state_registers_and_bypass *regs_and_bypass,
+		struct clk_mgr *clk_mgr_base, struct clk_log_info *log_info)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	uint32_t dprefclk_did = 0;
+	uint32_t dcfclk_did = 0;
+	uint32_t dtbclk_did = 0;
+	uint32_t dispclk_did = 0;
+	uint32_t dppclk_did = 0;
+	uint32_t target_div = 0;
+
+	if (ASICREV_IS_GC_11_0_2(clk_mgr->base.ctx->asic_id.hw_internal_rev)) {
+		/* DFS Slice 0 is used for DISPCLK */
+		dispclk_did = REG_READ(CLK0_CLK0_DFS_CNTL);
+		/* DFS Slice 1 is used for DPPCLK */
+		dppclk_did = REG_READ(CLK0_CLK1_DFS_CNTL);
+		/* DFS Slice 2 is used for DPREFCLK */
+		dprefclk_did = REG_READ(CLK0_CLK2_DFS_CNTL);
+		/* DFS Slice 3 is used for DCFCLK */
+		dcfclk_did = REG_READ(CLK0_CLK3_DFS_CNTL);
+		/* DFS Slice 4 is used for DTBCLK */
+		dtbclk_did = REG_READ(CLK0_CLK4_DFS_CNTL);
+	} else {
+		/* DFS Slice 0 is used for DISPCLK */
+		dispclk_did = REG_READ(CLK1_CLK0_DFS_CNTL);
+		/* DFS Slice 1 is used for DPPCLK */
+		dppclk_did = REG_READ(CLK1_CLK1_DFS_CNTL);
+		/* DFS Slice 2 is used for DPREFCLK */
+		dprefclk_did = REG_READ(CLK1_CLK2_DFS_CNTL);
+		/* DFS Slice 3 is used for DCFCLK */
+		dcfclk_did = REG_READ(CLK1_CLK3_DFS_CNTL);
+		/* DFS Slice 4 is used for DTBCLK */
+		dtbclk_did = REG_READ(CLK1_CLK4_DFS_CNTL);
+	}
+
+	/* Convert DISPCLK DFS Slice DID to divider*/
+	target_div = dentist_get_divider_from_did(dispclk_did);
+	//Get dispclk in khz
+	regs_and_bypass->dispclk = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+			* clk_mgr->base.dentist_vco_freq_khz) / target_div;
+
+	/* Convert DISPCLK DFS Slice DID to divider*/
+	target_div = dentist_get_divider_from_did(dppclk_did);
+	//Get dppclk in khz
+	regs_and_bypass->dppclk = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+			* clk_mgr->base.dentist_vco_freq_khz) / target_div;
+
+	/* Convert DPREFCLK DFS Slice DID to divider*/
+	target_div = dentist_get_divider_from_did(dprefclk_did);
+	//Get dprefclk in khz
+	regs_and_bypass->dprefclk = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+			* clk_mgr->base.dentist_vco_freq_khz) / target_div;
+
+	/* Convert DCFCLK DFS Slice DID to divider*/
+	target_div = dentist_get_divider_from_did(dcfclk_did);
+	//Get dcfclk in khz
+	regs_and_bypass->dcfclk = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+			* clk_mgr->base.dentist_vco_freq_khz) / target_div;
+
+	/* Convert DTBCLK DFS Slice DID to divider*/
+	target_div = dentist_get_divider_from_did(dtbclk_did);
+	//Get dtbclk in khz
+	regs_and_bypass->dtbclk = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+			* clk_mgr->base.dentist_vco_freq_khz) / target_div;
+}
+
+static void dcn32_clock_read_ss_info(struct clk_mgr_internal *clk_mgr)
+{
+	struct dc_bios *bp = clk_mgr->base.ctx->dc_bios;
+	int ss_info_num = bp->funcs->get_ss_entry_number(
+			bp, AS_SIGNAL_TYPE_GPU_PLL);
+
+	if (ss_info_num) {
+		struct spread_spectrum_info info = { { 0 } };
+		enum bp_result result = bp->funcs->get_spread_spectrum_info(
+				bp, AS_SIGNAL_TYPE_GPU_PLL, 0, &info);
+
+		/* SSInfo.spreadSpectrumPercentage !=0 would be sign
+		 * that SS is enabled
+		 */
+		if (result == BP_RESULT_OK &&
+				info.spread_spectrum_percentage != 0) {
+			clk_mgr->ss_on_dprefclk = true;
+			clk_mgr->dprefclk_ss_divider = info.spread_percentage_divider;
+
+			if (info.type.CENTER_MODE == 0) {
+				/* Currently for DP Reference clock we
+				 * need only SS percentage for
+				 * downspread
+				 */
+				clk_mgr->dprefclk_ss_percentage =
+						info.spread_spectrum_percentage;
+			}
+		}
+	}
+}
+static void dcn32_notify_wm_ranges(struct clk_mgr *clk_mgr_base)
+{
+	unsigned int i;
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	WatermarksExternal_t *table = (WatermarksExternal_t *) clk_mgr->wm_range_table;
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	if (!table)
+		return;
+
+	memset(table, 0, sizeof(*table));
+
+	/* collect valid ranges, place in pmfw table */
+	for (i = 0; i < WM_SET_COUNT; i++)
+		if (clk_mgr->base.bw_params->wm_table.nv_entries[i].valid) {
+			table->Watermarks.WatermarkRow[i].WmSetting = i;
+			table->Watermarks.WatermarkRow[i].Flags = clk_mgr->base.bw_params->wm_table.nv_entries[i].pmfw_breakdown.wm_type;
+		}
+	dcn30_smu_set_dram_addr_high(clk_mgr, clk_mgr->wm_range_table_addr >> 32);
+	dcn30_smu_set_dram_addr_low(clk_mgr, clk_mgr->wm_range_table_addr & 0xFFFFFFFF);
+	dcn32_smu_transfer_wm_table_dram_2_smu(clk_mgr);
+}
+
+/* Set min memclk to minimum, either constrained by the current mode or DPM0 */
+static void dcn32_set_hard_min_memclk(struct clk_mgr *clk_mgr_base, bool current_mode)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	if (current_mode) {
+		if (clk_mgr_base->clks.p_state_change_support)
+			dcn32_smu_set_hard_min_by_freq(clk_mgr, PPCLK_UCLK,
+					khz_to_mhz_ceil(clk_mgr_base->clks.dramclk_khz));
+		else
+			dcn32_smu_set_hard_min_by_freq(clk_mgr, PPCLK_UCLK,
+					clk_mgr_base->bw_params->max_memclk_mhz);
+	} else {
+		dcn32_smu_set_hard_min_by_freq(clk_mgr, PPCLK_UCLK,
+				clk_mgr_base->bw_params->clk_table.entries[0].memclk_mhz);
+	}
+}
+
+/* Set max memclk to highest DPM value */
+static void dcn32_set_hard_max_memclk(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn30_smu_set_hard_max_by_freq(clk_mgr, PPCLK_UCLK, clk_mgr_base->bw_params->max_memclk_mhz);
+}
+
+/* Get current memclk states, update bounding box */
+static void dcn32_get_memclk_states_from_smu(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct clk_limit_num_entries *num_entries_per_clk = &clk_mgr_base->bw_params->clk_table.num_entries_per_clk;
+	unsigned int num_levels;
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	/* Refresh memclk and fclk states */
+	dcn32_init_single_clock(clk_mgr, PPCLK_UCLK,
+			&clk_mgr_base->bw_params->clk_table.entries[0].memclk_mhz,
+			&num_entries_per_clk->num_memclk_levels);
+	clk_mgr_base->bw_params->dc_mode_limit.memclk_mhz = dcn30_smu_get_dc_mode_max_dpm_freq(clk_mgr, PPCLK_UCLK);
+	clk_mgr_base->bw_params->dc_mode_softmax_memclk = clk_mgr_base->bw_params->dc_mode_limit.memclk_mhz;
+
+	/* memclk must have at least one level */
+	num_entries_per_clk->num_memclk_levels = num_entries_per_clk->num_memclk_levels ? num_entries_per_clk->num_memclk_levels : 1;
+
+	dcn32_init_single_clock(clk_mgr, PPCLK_FCLK,
+			&clk_mgr_base->bw_params->clk_table.entries[0].fclk_mhz,
+			&num_entries_per_clk->num_fclk_levels);
+	clk_mgr_base->bw_params->dc_mode_limit.fclk_mhz = dcn30_smu_get_dc_mode_max_dpm_freq(clk_mgr, PPCLK_FCLK);
+
+	num_levels = max(num_entries_per_clk->num_memclk_levels, num_entries_per_clk->num_fclk_levels);
+
+	clk_mgr_base->bw_params->max_memclk_mhz =
+			clk_mgr_base->bw_params->clk_table.entries[num_entries_per_clk->num_memclk_levels - 1].memclk_mhz;
+	clk_mgr_base->bw_params->clk_table.num_entries = num_levels ? num_levels : 1;
+
+	if (clk_mgr->dpm_present && !num_levels)
+		clk_mgr->dpm_present = false;
+
+	if (!clk_mgr->dpm_present)
+		dcn32_patch_dpm_table(clk_mgr_base->bw_params);
+
+	DC_FP_START();
+	/* Refresh bounding box */
+	clk_mgr_base->ctx->dc->res_pool->funcs->update_bw_bounding_box(
+			clk_mgr->base.ctx->dc, clk_mgr_base->bw_params);
+	DC_FP_END();
+}
+
+static bool dcn32_are_clock_states_equal(struct dc_clocks *a,
+					struct dc_clocks *b)
+{
+	if (a->dispclk_khz != b->dispclk_khz)
+		return false;
+	else if (a->dppclk_khz != b->dppclk_khz)
+		return false;
+	else if (a->dcfclk_khz != b->dcfclk_khz)
+		return false;
+	else if (a->dcfclk_deep_sleep_khz != b->dcfclk_deep_sleep_khz)
+		return false;
+	else if (a->dramclk_khz != b->dramclk_khz)
+		return false;
+	else if (a->p_state_change_support != b->p_state_change_support)
+		return false;
+	else if (a->fclk_p_state_change_support != b->fclk_p_state_change_support)
+		return false;
+
+	return true;
+}
+
+static void dcn32_enable_pme_wa(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn32_smu_set_pme_workaround(clk_mgr);
+}
+
+static bool dcn32_is_smu_present(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	return clk_mgr->smu_present;
+}
+
+static void dcn32_set_max_memclk(struct clk_mgr *clk_mgr_base, unsigned int memclk_mhz)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn30_smu_set_hard_max_by_freq(clk_mgr, PPCLK_UCLK, memclk_mhz);
+}
+
+static void dcn32_set_min_memclk(struct clk_mgr *clk_mgr_base, unsigned int memclk_mhz)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn32_smu_set_hard_min_by_freq(clk_mgr, PPCLK_UCLK, memclk_mhz);
+}
+
+static struct clk_mgr_funcs dcn32_funcs = {
+		.get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz,
+		.get_dtb_ref_clk_frequency = dcn31_get_dtb_ref_freq_khz,
+		.update_clocks = dcn32_update_clocks,
+		.dump_clk_registers = dcn32_dump_clk_registers,
+		.init_clocks = dcn32_init_clocks,
+		.notify_wm_ranges = dcn32_notify_wm_ranges,
+		.set_hard_min_memclk = dcn32_set_hard_min_memclk,
+		.set_hard_max_memclk = dcn32_set_hard_max_memclk,
+		.set_max_memclk = dcn32_set_max_memclk,
+		.set_min_memclk = dcn32_set_min_memclk,
+		.get_memclk_states_from_smu = dcn32_get_memclk_states_from_smu,
+		.are_clock_states_equal = dcn32_are_clock_states_equal,
+		.enable_pme_wa = dcn32_enable_pme_wa,
+		.is_smu_present = dcn32_is_smu_present,
+		.get_dispclk_from_dentist = dcn32_get_dispclk_from_dentist,
+};
+
+void dcn32_clk_mgr_construct(
+		struct dc_context *ctx,
+		struct clk_mgr_internal *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg)
+{
+	struct clk_log_info log_info = {0};
+
+	clk_mgr->base.ctx = ctx;
+	clk_mgr->base.funcs = &dcn32_funcs;
+	if (ASICREV_IS_GC_11_0_2(clk_mgr->base.ctx->asic_id.hw_internal_rev)) {
+		clk_mgr->regs = &clk_mgr_regs_dcn321;
+		clk_mgr->clk_mgr_shift = &clk_mgr_shift_dcn321;
+		clk_mgr->clk_mgr_mask = &clk_mgr_mask_dcn321;
+	} else {
+		clk_mgr->regs = &clk_mgr_regs_dcn32;
+		clk_mgr->clk_mgr_shift = &clk_mgr_shift_dcn32;
+		clk_mgr->clk_mgr_mask = &clk_mgr_mask_dcn32;
+	}
+
+	clk_mgr->dccg = dccg;
+	clk_mgr->dfs_bypass_disp_clk = 0;
+
+	clk_mgr->dprefclk_ss_percentage = 0;
+	clk_mgr->dprefclk_ss_divider = 1000;
+	clk_mgr->ss_on_dprefclk = false;
+	clk_mgr->dfs_ref_freq_khz = 100000;
+
+	/* Changed from DCN3.2_clock_frequency doc to match
+	 * dcn32_dump_clk_registers from 4 * dentist_vco_freq_khz /
+	 * dprefclk DID divider
+	 */
+	clk_mgr->base.dprefclk_khz = 716666;
+	if (ctx->dc->debug.disable_dtb_ref_clk_switch) {
+		//initialize DTB ref clock value if DPM disabled
+		if (ctx->dce_version == DCN_VERSION_3_21)
+			clk_mgr->base.clks.ref_dtbclk_khz = 477800;
+		else
+			clk_mgr->base.clks.ref_dtbclk_khz = 268750;
+	}
+
+
+	/* integer part is now VCO frequency in kHz */
+	clk_mgr->base.dentist_vco_freq_khz = dcn32_get_vco_frequency_from_reg(clk_mgr);
+
+	/* in case we don't get a value from the register, use default */
+	if (clk_mgr->base.dentist_vco_freq_khz == 0)
+		clk_mgr->base.dentist_vco_freq_khz = 4300000; /* Updated as per HW docs */
+
+	dcn32_dump_clk_registers(&clk_mgr->base.boot_snapshot, &clk_mgr->base, &log_info);
+
+	if (ctx->dc->debug.disable_dtb_ref_clk_switch &&
+			clk_mgr->base.clks.ref_dtbclk_khz != clk_mgr->base.boot_snapshot.dtbclk) {
+		clk_mgr->base.clks.ref_dtbclk_khz = clk_mgr->base.boot_snapshot.dtbclk;
+	}
+
+	if (clk_mgr->base.boot_snapshot.dprefclk != 0) {
+		clk_mgr->base.dprefclk_khz = clk_mgr->base.boot_snapshot.dprefclk;
+	}
+	dcn32_clock_read_ss_info(clk_mgr);
+
+	clk_mgr->dfs_bypass_enabled = false;
+
+	clk_mgr->smu_present = false;
+
+	clk_mgr->base.bw_params = kzalloc(sizeof(*clk_mgr->base.bw_params), GFP_KERNEL);
+	if (!clk_mgr->base.bw_params) {
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+	/* need physical address of table to give to PMFW */
+	clk_mgr->wm_range_table = dm_helpers_allocate_gpu_mem(clk_mgr->base.ctx,
+			DC_MEM_ALLOC_TYPE_GART, sizeof(WatermarksExternal_t),
+			&clk_mgr->wm_range_table_addr);
+	if (!clk_mgr->wm_range_table) {
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+void dcn32_clk_mgr_destroy(struct clk_mgr_internal *clk_mgr)
+{
+	kfree(clk_mgr->base.bw_params);
+
+	if (clk_mgr->wm_range_table)
+		dm_helpers_free_gpu_mem(clk_mgr->base.ctx, DC_MEM_ALLOC_TYPE_GART,
+				clk_mgr->wm_range_table);
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_clk_mgr.h
new file mode 100644
index 000000000000..186daada7b03
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_clk_mgr.h
@@ -0,0 +1,42 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+#ifndef __DCN32_CLK_MGR_H_
+#define __DCN32_CLK_MGR_H_
+
+void dcn32_init_clocks(struct clk_mgr *clk_mgr_base);
+
+void dcn32_clk_mgr_construct(struct dc_context *ctx,
+		struct clk_mgr_internal *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg);
+
+void dcn32_update_clocks_update_dpp_dto(struct clk_mgr_internal *clk_mgr,
+		struct dc_state *context, bool safe_to_lower);
+
+void dcn32_clk_mgr_destroy(struct clk_mgr_internal *clk_mgr);
+
+
+
+#endif /* __DCN32_CLK_MGR_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_clk_mgr_smu_msg.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_clk_mgr_smu_msg.c
new file mode 100644
index 000000000000..5d80fdf63ffc
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_clk_mgr_smu_msg.c
@@ -0,0 +1,308 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dcn32_clk_mgr_smu_msg.h"
+
+#include "clk_mgr_internal.h"
+#include "reg_helper.h"
+#include "dalsmc.h"
+#include "dcn32_smu13_driver_if.h"
+
+#define mmDAL_MSG_REG  0x1628A
+#define mmDAL_ARG_REG  0x16273
+#define mmDAL_RESP_REG 0x16274
+
+#define REG(reg_name) \
+	mm ## reg_name
+
+#include "logger_types.h"
+
+#define smu_print(str, ...) {DC_LOG_SMU(str, ##__VA_ARGS__); }
+
+
+/*
+ * Function to be used instead of REG_WAIT macro because the wait ends when
+ * the register is NOT EQUAL to zero, and because the translation in msg_if.h
+ * won't work with REG_WAIT.
+ */
+static uint32_t dcn32_smu_wait_for_response(struct clk_mgr_internal *clk_mgr, unsigned int delay_us, unsigned int max_retries)
+{
+	const uint32_t initial_max_retries = max_retries;
+	uint32_t reg = 0;
+
+	do {
+		reg = REG_READ(DAL_RESP_REG);
+		if (reg)
+			break;
+
+		if (delay_us >= 1000)
+			msleep(delay_us/1000);
+		else if (delay_us > 0)
+			udelay(delay_us);
+	} while (max_retries--);
+
+	TRACE_SMU_MSG_DELAY(0, 0, delay_us * (initial_max_retries - max_retries), clk_mgr->base.ctx);
+
+
+	return reg;
+}
+
+static bool dcn32_smu_send_msg_with_param(struct clk_mgr_internal *clk_mgr, uint32_t msg_id, uint32_t param_in, uint32_t *param_out)
+{
+	/* Wait for response register to be ready */
+	dcn32_smu_wait_for_response(clk_mgr, 10, 200000);
+
+	/* Clear response register */
+	REG_WRITE(DAL_RESP_REG, 0);
+
+	/* Set the parameter register for the SMU message */
+	REG_WRITE(DAL_ARG_REG, param_in);
+
+	/* Trigger the message transaction by writing the message ID */
+	REG_WRITE(DAL_MSG_REG, msg_id);
+
+	TRACE_SMU_MSG(msg_id, param_in, clk_mgr->base.ctx);
+
+	/* Wait for response */
+	if (dcn32_smu_wait_for_response(clk_mgr, 10, 200000) == DALSMC_Result_OK) {
+		if (param_out)
+			*param_out = REG_READ(DAL_ARG_REG);
+
+		return true;
+	}
+
+	return false;
+}
+
+/*
+ * Use these functions to return back delay information so we can aggregate the total
+ *  delay when requesting hardmin clk
+ *
+ * dcn32_smu_wait_for_response_delay
+ * dcn32_smu_send_msg_with_param_delay
+ *
+ */
+static uint32_t dcn32_smu_wait_for_response_delay(struct clk_mgr_internal *clk_mgr, unsigned int delay_us, unsigned int max_retries, unsigned int *total_delay_us)
+{
+	uint32_t reg = 0;
+	*total_delay_us = 0;
+
+	do {
+		reg = REG_READ(DAL_RESP_REG);
+		if (reg)
+			break;
+
+		if (delay_us >= 1000)
+			msleep(delay_us/1000);
+		else if (delay_us > 0)
+			udelay(delay_us);
+		*total_delay_us += delay_us;
+	} while (max_retries--);
+
+	TRACE_SMU_MSG_DELAY(0, 0, *total_delay_us, clk_mgr->base.ctx);
+
+	return reg;
+}
+
+static bool dcn32_smu_send_msg_with_param_delay(struct clk_mgr_internal *clk_mgr, uint32_t msg_id, uint32_t param_in, uint32_t *param_out, unsigned int *total_delay_us)
+{
+	unsigned int delay1_us, delay2_us;
+	*total_delay_us = 0;
+
+	/* Wait for response register to be ready */
+	dcn32_smu_wait_for_response_delay(clk_mgr, 10, 200000, &delay1_us);
+
+	/* Clear response register */
+	REG_WRITE(DAL_RESP_REG, 0);
+
+	/* Set the parameter register for the SMU message */
+	REG_WRITE(DAL_ARG_REG, param_in);
+
+	/* Trigger the message transaction by writing the message ID */
+	REG_WRITE(DAL_MSG_REG, msg_id);
+
+	TRACE_SMU_MSG(msg_id, param_in, clk_mgr->base.ctx);
+
+	/* Wait for response */
+	if (dcn32_smu_wait_for_response_delay(clk_mgr, 10, 200000, &delay2_us) == DALSMC_Result_OK) {
+		if (param_out)
+			*param_out = REG_READ(DAL_ARG_REG);
+
+		*total_delay_us = delay1_us + delay2_us;
+		return true;
+	}
+
+	*total_delay_us = delay1_us + 2000000;
+	return false;
+}
+
+void dcn32_smu_send_fclk_pstate_message(struct clk_mgr_internal *clk_mgr, bool enable)
+{
+	smu_print("FCLK P-state support value is : %d\n", enable);
+
+	dcn32_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_SetFclkSwitchAllow, enable ? FCLK_PSTATE_SUPPORTED : FCLK_PSTATE_NOTSUPPORTED, NULL);
+}
+
+void dcn32_smu_send_cab_for_uclk_message(struct clk_mgr_internal *clk_mgr, unsigned int num_ways)
+{
+	uint32_t param = (num_ways << 1) | (num_ways > 0);
+
+	dcn32_smu_send_msg_with_param(clk_mgr, DALSMC_MSG_SetCabForUclkPstate, param, NULL);
+	smu_print("Numways for SubVP : %d\n", num_ways);
+}
+
+void dcn32_smu_transfer_wm_table_dram_2_smu(struct clk_mgr_internal *clk_mgr)
+{
+	smu_print("SMU Transfer WM table DRAM 2 SMU\n");
+
+	dcn32_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_TransferTableDram2Smu, TABLE_WATERMARKS, NULL);
+}
+
+void dcn32_smu_set_pme_workaround(struct clk_mgr_internal *clk_mgr)
+{
+	smu_print("SMU Set PME workaround\n");
+
+	dcn32_smu_send_msg_with_param(clk_mgr,
+		DALSMC_MSG_BacoAudioD3PME, 0, NULL);
+}
+
+/* Check PMFW version if it supports ReturnHardMinStatus message */
+static bool dcn32_get_hard_min_status_supported(struct clk_mgr_internal *clk_mgr)
+{
+	if (ASICREV_IS_GC_11_0_0(clk_mgr->base.ctx->asic_id.hw_internal_rev)) {
+		if (clk_mgr->smu_ver >= 0x4e6a00)
+			return true;
+	} else if (ASICREV_IS_GC_11_0_2(clk_mgr->base.ctx->asic_id.hw_internal_rev)) {
+		if (clk_mgr->smu_ver >= 0x524e00)
+			return true;
+	} else { /* ASICREV_IS_GC_11_0_3 */
+		if (clk_mgr->smu_ver >= 0x503900)
+			return true;
+	}
+	return false;
+}
+
+/* Returns the clocks which were fulfilled by the DAL hard min arbiter in PMFW */
+static unsigned int dcn32_smu_get_hard_min_status(struct clk_mgr_internal *clk_mgr, bool *no_timeout, unsigned int *total_delay_us)
+{
+	uint32_t response = 0;
+
+	/* bits 23:16 for clock type, lower 16 bits for frequency in MHz */
+	uint32_t param = 0;
+
+	*no_timeout = dcn32_smu_send_msg_with_param_delay(clk_mgr,
+			DALSMC_MSG_ReturnHardMinStatus, param, &response, total_delay_us);
+
+	smu_print("SMU Get hard min status: no_timeout %d delay %d us clk bits %x\n",
+		*no_timeout, *total_delay_us, response);
+
+	return response;
+}
+
+static bool dcn32_smu_wait_get_hard_min_status(struct clk_mgr_internal *clk_mgr,
+	uint32_t clk)
+{
+	int readDalHardMinClkBits, checkDalHardMinClkBits;
+	unsigned int total_delay_us, read_total_delay_us;
+	bool no_timeout, hard_min_done;
+
+	static unsigned int cur_wait_get_hard_min_max_us;
+	static unsigned int cur_wait_get_hard_min_max_timeouts;
+
+	checkDalHardMinClkBits = CHECK_HARD_MIN_CLK_DPREFCLK;
+	if (clk == PPCLK_DISPCLK)
+		checkDalHardMinClkBits |= CHECK_HARD_MIN_CLK_DISPCLK;
+	if (clk == PPCLK_DPPCLK)
+		checkDalHardMinClkBits |= CHECK_HARD_MIN_CLK_DPPCLK;
+	if (clk == PPCLK_DCFCLK)
+		checkDalHardMinClkBits |= CHECK_HARD_MIN_CLK_DCFCLK;
+	if (clk == PPCLK_DTBCLK)
+		checkDalHardMinClkBits |= CHECK_HARD_MIN_CLK_DTBCLK;
+	if (clk == PPCLK_UCLK)
+		checkDalHardMinClkBits |= CHECK_HARD_MIN_CLK_UCLK;
+
+	if (checkDalHardMinClkBits == CHECK_HARD_MIN_CLK_DPREFCLK)
+		return 0;
+
+	total_delay_us = 0;
+	hard_min_done = false;
+	while (1) {
+		readDalHardMinClkBits = dcn32_smu_get_hard_min_status(clk_mgr, &no_timeout, &read_total_delay_us);
+		total_delay_us += read_total_delay_us;
+		if (checkDalHardMinClkBits == (readDalHardMinClkBits & checkDalHardMinClkBits)) {
+			hard_min_done = true;
+			break;
+		}
+
+
+		if (total_delay_us >= 2000000) {
+			cur_wait_get_hard_min_max_timeouts++;
+			smu_print("SMU Wait get hard min status: %d timeouts\n", cur_wait_get_hard_min_max_timeouts);
+			break;
+		}
+		msleep(1);
+		total_delay_us += 1000;
+	}
+
+	if (total_delay_us > cur_wait_get_hard_min_max_us)
+		cur_wait_get_hard_min_max_us = total_delay_us;
+
+	smu_print("SMU Wait get hard min status: no_timeout %d, delay %d us, max %d us, read %x, check %x\n",
+		no_timeout, total_delay_us, cur_wait_get_hard_min_max_us, readDalHardMinClkBits, checkDalHardMinClkBits);
+
+	return hard_min_done;
+}
+
+/* Returns the actual frequency that was set in MHz, 0 on failure */
+unsigned int dcn32_smu_set_hard_min_by_freq(struct clk_mgr_internal *clk_mgr, uint32_t clk, uint16_t freq_mhz)
+{
+	uint32_t response = 0;
+	bool hard_min_done = false;
+
+	/* bits 23:16 for clock type, lower 16 bits for frequency in MHz */
+	uint32_t param = (clk << 16) | freq_mhz;
+
+	smu_print("SMU Set hard min by freq: clk = %d, freq_mhz = %d MHz\n", clk, freq_mhz);
+
+	dcn32_smu_send_msg_with_param(clk_mgr,
+		DALSMC_MSG_SetHardMinByFreq, param, &response);
+
+	if (dcn32_get_hard_min_status_supported(clk_mgr)) {
+		hard_min_done = dcn32_smu_wait_get_hard_min_status(clk_mgr, clk);
+		smu_print("SMU Frequency set = %d KHz hard_min_done %d\n", response, hard_min_done);
+	} else
+		smu_print("SMU Frequency set = %d KHz\n", response);
+
+	return response;
+}
+
+void dcn32_smu_wait_for_dmub_ack_mclk(struct clk_mgr_internal *clk_mgr, bool enable)
+{
+	smu_print("PMFW to wait for DMCUB ack for MCLK : %d\n", enable);
+
+	dcn32_smu_send_msg_with_param(clk_mgr, 0x14, enable ? 1 : 0, NULL);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_clk_mgr_smu_msg.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_clk_mgr_smu_msg.h
new file mode 100644
index 000000000000..5c44ab0e8667
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_clk_mgr_smu_msg.h
@@ -0,0 +1,46 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCN32_CLK_MGR_SMU_MSG_H_
+#define __DCN32_CLK_MGR_SMU_MSG_H_
+
+#include "core_types.h"
+#include "dcn30/dcn30_clk_mgr_smu_msg.h"
+
+#define FCLK_PSTATE_NOTSUPPORTED       0x00
+#define FCLK_PSTATE_SUPPORTED          0x01
+
+/* TODO Remove this MSG ID define after it becomes available in dalsmc */
+#define DALSMC_MSG_SetCabForUclkPstate	0x12
+#define DALSMC_Result_OK				0x1
+
+void dcn32_smu_send_fclk_pstate_message(struct clk_mgr_internal *clk_mgr, bool enable);
+void dcn32_smu_send_cab_for_uclk_message(struct clk_mgr_internal *clk_mgr, unsigned int num_ways);
+void dcn32_smu_transfer_wm_table_dram_2_smu(struct clk_mgr_internal *clk_mgr);
+void dcn32_smu_set_pme_workaround(struct clk_mgr_internal *clk_mgr);
+unsigned int dcn32_smu_set_hard_min_by_freq(struct clk_mgr_internal *clk_mgr, uint32_t clk, uint16_t freq_mhz);
+void dcn32_smu_wait_for_dmub_ack_mclk(struct clk_mgr_internal *clk_mgr, bool enable);
+
+#endif /* __DCN32_CLK_MGR_SMU_MSG_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_smu13_driver_if.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_smu13_driver_if.h
new file mode 100644
index 000000000000..8d54865bbd5d
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_smu13_driver_if.h
@@ -0,0 +1,64 @@
+// SPDX-License-Identifier: MIT
+/* Copyright © 2022-2024 Advanced Micro Devices, Inc. All rights reserved. */
+
+#define SMU13_DRIVER_IF_VERSION  0x18
+
+//Only Clks that have DPM descriptors are listed here
+typedef enum {
+	PPCLK_GFXCLK = 0,
+	PPCLK_SOCCLK,
+	PPCLK_UCLK,
+	PPCLK_FCLK,
+	PPCLK_DCLK_0,
+	PPCLK_VCLK_0,
+	PPCLK_DCLK_1,
+	PPCLK_VCLK_1,
+	PPCLK_DISPCLK,
+	PPCLK_DPPCLK,
+	PPCLK_DPREFCLK,
+	PPCLK_DCFCLK,
+	PPCLK_DTBCLK,
+	PPCLK_COUNT,
+} PPCLK_e;
+
+typedef struct {
+	uint8_t  WmSetting;
+	uint8_t  Flags;
+	uint8_t  Padding[2];
+
+} WatermarkRowGeneric_t;
+
+#define NUM_WM_RANGES 4
+
+typedef enum {
+	WATERMARKS_CLOCK_RANGE = 0,
+	WATERMARKS_DUMMY_PSTATE,
+	WATERMARKS_MALL,
+	WATERMARKS_COUNT,
+} WATERMARKS_FLAGS_e;
+
+typedef struct {
+	// Watermarks
+	WatermarkRowGeneric_t WatermarkRow[NUM_WM_RANGES];
+} Watermarks_t;
+
+typedef struct {
+	Watermarks_t Watermarks;
+	uint32_t  Spare[16];
+
+	uint32_t     MmHubPadding[8]; // SMU internal use
+} WatermarksExternal_t;
+
+// Table types
+#define TABLE_PMFW_PPTABLE            0
+#define TABLE_COMBO_PPTABLE           1
+#define TABLE_WATERMARKS              2
+#define TABLE_AVFS_PSM_DEBUG          3
+#define TABLE_PMSTATUSLOG             4
+#define TABLE_SMU_METRICS             5
+#define TABLE_DRIVER_SMU_CONFIG       6
+#define TABLE_ACTIVITY_MONITOR_COEFF  7
+#define TABLE_OVERDRIVE               8
+#define TABLE_I2C_COMMANDS            9
+#define TABLE_DRIVER_INFO             10
+#define TABLE_COUNT                   11
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn351_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn351_clk_mgr.c
new file mode 100644
index 000000000000..4607eff07253
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn351_clk_mgr.c
@@ -0,0 +1,141 @@
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "core_types.h"
+#include "dcn35_clk_mgr.h"
+
+#define DCN_BASE__INST0_SEG1 0x000000C0
+#define mmCLK1_CLK_PLL_REQ 0x16E37
+
+#define mmCLK1_CLK0_DFS_CNTL 0x16E69
+#define mmCLK1_CLK1_DFS_CNTL 0x16E6C
+#define mmCLK1_CLK2_DFS_CNTL 0x16E6F
+#define mmCLK1_CLK3_DFS_CNTL 0x16E72
+#define mmCLK1_CLK4_DFS_CNTL 0x16E75
+#define mmCLK1_CLK5_DFS_CNTL 0x16E78
+
+#define mmCLK1_CLK0_CURRENT_CNT 0x16EFC
+#define mmCLK1_CLK1_CURRENT_CNT 0x16EFD
+#define mmCLK1_CLK2_CURRENT_CNT 0x16EFE
+#define mmCLK1_CLK3_CURRENT_CNT 0x16EFF
+#define mmCLK1_CLK4_CURRENT_CNT 0x16F00
+#define mmCLK1_CLK5_CURRENT_CNT 0x16F01
+
+#define mmCLK1_CLK0_BYPASS_CNTL 0x16E8A
+#define mmCLK1_CLK1_BYPASS_CNTL 0x16E93
+#define mmCLK1_CLK2_BYPASS_CNTL 0x16E9C
+#define mmCLK1_CLK3_BYPASS_CNTL 0x16EA5
+#define mmCLK1_CLK4_BYPASS_CNTL 0x16EAE
+#define mmCLK1_CLK5_BYPASS_CNTL 0x16EB7
+
+#define mmCLK1_CLK0_DS_CNTL 0x16E83
+#define mmCLK1_CLK1_DS_CNTL 0x16E8C
+#define mmCLK1_CLK2_DS_CNTL 0x16E95
+#define mmCLK1_CLK3_DS_CNTL 0x16E9E
+#define mmCLK1_CLK4_DS_CNTL 0x16EA7
+#define mmCLK1_CLK5_DS_CNTL 0x16EB0
+
+#define mmCLK1_CLK0_ALLOW_DS 0x16E84
+#define mmCLK1_CLK1_ALLOW_DS 0x16E8D
+#define mmCLK1_CLK2_ALLOW_DS 0x16E96
+#define mmCLK1_CLK3_ALLOW_DS 0x16E9F
+#define mmCLK1_CLK4_ALLOW_DS 0x16EA8
+#define mmCLK1_CLK5_ALLOW_DS 0x16EB1
+
+#define mmCLK5_spll_field_8 0x1B04B
+#define mmCLK6_spll_field_8 0x1B24B
+#define mmDENTIST_DISPCLK_CNTL 0x0124
+#define regDENTIST_DISPCLK_CNTL 0x0064
+#define regDENTIST_DISPCLK_CNTL_BASE_IDX 1
+
+#define CLK1_CLK_PLL_REQ__FbMult_int__SHIFT 0x0
+#define CLK1_CLK_PLL_REQ__PllSpineDiv__SHIFT 0xc
+#define CLK1_CLK_PLL_REQ__FbMult_frac__SHIFT 0x10
+#define CLK1_CLK_PLL_REQ__FbMult_int_MASK 0x000001FFL
+#define CLK1_CLK_PLL_REQ__PllSpineDiv_MASK 0x0000F000L
+#define CLK1_CLK_PLL_REQ__FbMult_frac_MASK 0xFFFF0000L
+
+#define CLK1_CLK2_BYPASS_CNTL__CLK2_BYPASS_SEL_MASK 0x00000007L
+
+// DENTIST_DISPCLK_CNTL
+#define DENTIST_DISPCLK_CNTL__DENTIST_DISPCLK_WDIVIDER__SHIFT 0x0
+#define DENTIST_DISPCLK_CNTL__DENTIST_DISPCLK_RDIVIDER__SHIFT 0x8
+#define DENTIST_DISPCLK_CNTL__DENTIST_DISPCLK_CHG_DONE__SHIFT 0x13
+#define DENTIST_DISPCLK_CNTL__DENTIST_DPPCLK_CHG_DONE__SHIFT 0x14
+#define DENTIST_DISPCLK_CNTL__DENTIST_DPPCLK_WDIVIDER__SHIFT 0x18
+#define DENTIST_DISPCLK_CNTL__DENTIST_DISPCLK_WDIVIDER_MASK 0x0000007FL
+#define DENTIST_DISPCLK_CNTL__DENTIST_DISPCLK_RDIVIDER_MASK 0x00007F00L
+#define DENTIST_DISPCLK_CNTL__DENTIST_DISPCLK_CHG_DONE_MASK 0x00080000L
+#define DENTIST_DISPCLK_CNTL__DENTIST_DPPCLK_CHG_DONE_MASK 0x00100000L
+#define DENTIST_DISPCLK_CNTL__DENTIST_DPPCLK_WDIVIDER_MASK 0x7F000000L
+
+#define CLK5_spll_field_8__spll_ssc_en_MASK 0x00002000L
+
+#define REG(reg) \
+	(clk_mgr->regs->reg)
+
+#define BASE_INNER(seg) DCN_BASE__INST0_SEG ## seg
+
+#define BASE(seg) BASE_INNER(seg)
+
+#define SR(reg_name)\
+		.reg_name = BASE(reg ## reg_name ## _BASE_IDX) +  \
+					reg ## reg_name
+
+#define CLK_SR_DCN35(reg_name)\
+	.reg_name = mm ## reg_name
+
+static const struct clk_mgr_registers clk_mgr_regs_dcn351 = {
+	CLK_REG_LIST_DCN35()
+};
+
+static const struct clk_mgr_shift clk_mgr_shift_dcn351 = {
+	CLK_COMMON_MASK_SH_LIST_DCN32(__SHIFT)
+};
+
+static const struct clk_mgr_mask clk_mgr_mask_dcn351 = {
+	CLK_COMMON_MASK_SH_LIST_DCN32(_MASK)
+};
+
+#define TO_CLK_MGR_DCN35(clk_mgr)\
+	container_of(clk_mgr, struct clk_mgr_dcn35, base)
+
+
+void dcn351_clk_mgr_construct(
+		struct dc_context *ctx,
+		struct clk_mgr_dcn35 *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg)
+{
+	/*register offset changed*/
+	clk_mgr->base.regs = &clk_mgr_regs_dcn351;
+	clk_mgr->base.clk_mgr_shift = &clk_mgr_shift_dcn351;
+	clk_mgr->base.clk_mgr_mask = &clk_mgr_mask_dcn351;
+
+	dcn35_clk_mgr_construct(ctx,  clk_mgr, pp_smu, dccg);
+
+}
+
+
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn35_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn35_clk_mgr.c
new file mode 100644
index 000000000000..b11383fba35f
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn35_clk_mgr.c
@@ -0,0 +1,1556 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+#include "dcn35_clk_mgr.h"
+
+#include "dccg.h"
+#include "clk_mgr_internal.h"
+
+// For dce12_get_dp_ref_freq_khz
+#include "dce100/dce_clk_mgr.h"
+
+// For dcn20_update_clocks_update_dpp_dto
+#include "dcn20/dcn20_clk_mgr.h"
+
+
+#include "reg_helper.h"
+#include "core_types.h"
+#include "dcn35_smu.h"
+#include "dm_helpers.h"
+
+#include "dcn31/dcn31_clk_mgr.h"
+
+#include "dc_dmub_srv.h"
+#include "link_service.h"
+#include "logger_types.h"
+
+#undef DC_LOGGER
+#define DC_LOGGER \
+	clk_mgr->base.base.ctx->logger
+
+#define DCN_BASE__INST0_SEG1 0x000000C0
+#define mmCLK1_CLK_PLL_REQ 0x16E37
+
+#define mmCLK1_CLK0_DFS_CNTL 0x16E69
+#define mmCLK1_CLK1_DFS_CNTL 0x16E6C
+#define mmCLK1_CLK2_DFS_CNTL 0x16E6F
+#define mmCLK1_CLK3_DFS_CNTL 0x16E72
+#define mmCLK1_CLK4_DFS_CNTL 0x16E75
+#define mmCLK1_CLK5_DFS_CNTL 0x16E78
+
+#define mmCLK1_CLK0_CURRENT_CNT 0x16EFB
+#define mmCLK1_CLK1_CURRENT_CNT 0x16EFC
+#define mmCLK1_CLK2_CURRENT_CNT 0x16EFD
+#define mmCLK1_CLK3_CURRENT_CNT 0x16EFE
+#define mmCLK1_CLK4_CURRENT_CNT 0x16EFF
+#define mmCLK1_CLK5_CURRENT_CNT 0x16F00
+
+#define mmCLK1_CLK0_BYPASS_CNTL 0x16E8A
+#define mmCLK1_CLK1_BYPASS_CNTL 0x16E93
+#define mmCLK1_CLK2_BYPASS_CNTL 0x16E9C
+#define mmCLK1_CLK3_BYPASS_CNTL 0x16EA5
+#define mmCLK1_CLK4_BYPASS_CNTL 0x16EAE
+#define mmCLK1_CLK5_BYPASS_CNTL 0x16EB7
+
+#define mmCLK1_CLK0_DS_CNTL 0x16E83
+#define mmCLK1_CLK1_DS_CNTL 0x16E8C
+#define mmCLK1_CLK2_DS_CNTL 0x16E95
+#define mmCLK1_CLK3_DS_CNTL 0x16E9E
+#define mmCLK1_CLK4_DS_CNTL 0x16EA7
+#define mmCLK1_CLK5_DS_CNTL 0x16EB0
+
+#define mmCLK1_CLK0_ALLOW_DS 0x16E84
+#define mmCLK1_CLK1_ALLOW_DS 0x16E8D
+#define mmCLK1_CLK2_ALLOW_DS 0x16E96
+#define mmCLK1_CLK3_ALLOW_DS 0x16E9F
+#define mmCLK1_CLK4_ALLOW_DS 0x16EA8
+#define mmCLK1_CLK5_ALLOW_DS 0x16EB1
+
+#define mmCLK5_spll_field_8 0x1B24B
+#define mmCLK6_spll_field_8 0x1B24B
+#define mmDENTIST_DISPCLK_CNTL 0x0124
+#define regDENTIST_DISPCLK_CNTL 0x0064
+#define regDENTIST_DISPCLK_CNTL_BASE_IDX 1
+
+#define CLK1_CLK_PLL_REQ__FbMult_int__SHIFT 0x0
+#define CLK1_CLK_PLL_REQ__PllSpineDiv__SHIFT 0xc
+#define CLK1_CLK_PLL_REQ__FbMult_frac__SHIFT 0x10
+#define CLK1_CLK_PLL_REQ__FbMult_int_MASK 0x000001FFL
+#define CLK1_CLK_PLL_REQ__PllSpineDiv_MASK 0x0000F000L
+#define CLK1_CLK_PLL_REQ__FbMult_frac_MASK 0xFFFF0000L
+
+#define CLK1_CLK2_BYPASS_CNTL__CLK2_BYPASS_SEL_MASK 0x00000007L
+#define CLK1_CLK2_BYPASS_CNTL__CLK2_BYPASS_DIV_MASK 0x000F0000L
+// DENTIST_DISPCLK_CNTL
+#define DENTIST_DISPCLK_CNTL__DENTIST_DISPCLK_WDIVIDER__SHIFT 0x0
+#define DENTIST_DISPCLK_CNTL__DENTIST_DISPCLK_RDIVIDER__SHIFT 0x8
+#define DENTIST_DISPCLK_CNTL__DENTIST_DISPCLK_CHG_DONE__SHIFT 0x13
+#define DENTIST_DISPCLK_CNTL__DENTIST_DPPCLK_CHG_DONE__SHIFT 0x14
+#define DENTIST_DISPCLK_CNTL__DENTIST_DPPCLK_WDIVIDER__SHIFT 0x18
+#define DENTIST_DISPCLK_CNTL__DENTIST_DISPCLK_WDIVIDER_MASK 0x0000007FL
+#define DENTIST_DISPCLK_CNTL__DENTIST_DISPCLK_RDIVIDER_MASK 0x00007F00L
+#define DENTIST_DISPCLK_CNTL__DENTIST_DISPCLK_CHG_DONE_MASK 0x00080000L
+#define DENTIST_DISPCLK_CNTL__DENTIST_DPPCLK_CHG_DONE_MASK 0x00100000L
+#define DENTIST_DISPCLK_CNTL__DENTIST_DPPCLK_WDIVIDER_MASK 0x7F000000L
+
+#define CLK5_spll_field_8__spll_ssc_en_MASK 0x00002000L
+#define CLK6_spll_field_8__spll_ssc_en_MASK 0x00002000L
+
+#define SMU_VER_THRESHOLD 0x5D4A00 //93.74.0
+#undef FN
+#define FN(reg_name, field_name) \
+	clk_mgr->clk_mgr_shift->field_name, clk_mgr->clk_mgr_mask->field_name
+
+#define REG(reg) \
+	(clk_mgr->regs->reg)
+
+#define BASE_INNER(seg) DCN_BASE__INST0_SEG ## seg
+
+#define BASE(seg) BASE_INNER(seg)
+
+#define SR(reg_name)\
+		.reg_name = BASE(reg ## reg_name ## _BASE_IDX) +  \
+					reg ## reg_name
+
+#define CLK_SR_DCN35(reg_name)\
+	.reg_name = mm ## reg_name
+
+static const struct clk_mgr_registers clk_mgr_regs_dcn35 = {
+	CLK_REG_LIST_DCN35()
+};
+
+static const struct clk_mgr_shift clk_mgr_shift_dcn35 = {
+	CLK_COMMON_MASK_SH_LIST_DCN32(__SHIFT)
+};
+
+static const struct clk_mgr_mask clk_mgr_mask_dcn35 = {
+	CLK_COMMON_MASK_SH_LIST_DCN32(_MASK)
+};
+
+#define TO_CLK_MGR_DCN35(clk_mgr)\
+	container_of(clk_mgr, struct clk_mgr_dcn35, base)
+
+static int dcn35_get_active_display_cnt_wa(
+		struct dc *dc,
+		struct dc_state *context,
+		int *all_active_disps)
+{
+	int i, display_count = 0;
+	bool tmds_present = false;
+
+	for (i = 0; i < context->stream_count; i++) {
+		const struct dc_stream_state *stream = context->streams[i];
+
+		if (stream->signal == SIGNAL_TYPE_HDMI_TYPE_A ||
+				stream->signal == SIGNAL_TYPE_DVI_SINGLE_LINK ||
+				stream->signal == SIGNAL_TYPE_DVI_DUAL_LINK)
+			tmds_present = true;
+	}
+
+	for (i = 0; i < dc->link_count; i++) {
+		const struct dc_link *link = dc->links[i];
+
+		/* abusing the fact that the dig and phy are coupled to see if the phy is enabled */
+		if (link->link_enc && link->link_enc->funcs->is_dig_enabled &&
+				link->link_enc->funcs->is_dig_enabled(link->link_enc))
+			display_count++;
+	}
+	if (all_active_disps != NULL)
+		*all_active_disps = display_count;
+	/* WA for hang on HDMI after display off back on*/
+	if (display_count == 0 && tmds_present)
+		display_count = 1;
+
+	return display_count;
+}
+static void dcn35_disable_otg_wa(struct clk_mgr *clk_mgr_base, struct dc_state *context,
+		bool safe_to_lower, bool disable)
+{
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	int i;
+
+	if (dc->ctx->dce_environment == DCE_ENV_DIAG)
+		return;
+
+	for (i = 0; i < dc->res_pool->pipe_count; ++i) {
+		struct pipe_ctx *old_pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+		struct pipe_ctx *new_pipe = &context->res_ctx.pipe_ctx[i];
+		struct clk_mgr_internal *clk_mgr_internal = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+		struct dccg *dccg = clk_mgr_internal->dccg;
+		struct pipe_ctx *pipe = safe_to_lower
+			? &context->res_ctx.pipe_ctx[i]
+			: &dc->current_state->res_ctx.pipe_ctx[i];
+		struct link_encoder *new_pipe_link_enc = new_pipe->link_res.dio_link_enc;
+		struct link_encoder *pipe_link_enc = pipe->link_res.dio_link_enc;
+		bool stream_changed_otg_dig_on = false;
+		bool has_active_hpo = false;
+
+		if (pipe->top_pipe || pipe->prev_odm_pipe)
+			continue;
+
+		if (!dc->config.unify_link_enc_assignment) {
+			if (new_pipe->stream)
+				new_pipe_link_enc = new_pipe->stream->link_enc;
+			if (pipe->stream)
+				pipe_link_enc = pipe->stream->link_enc;
+		}
+
+		stream_changed_otg_dig_on = old_pipe->stream && new_pipe->stream &&
+		old_pipe->stream != new_pipe->stream &&
+		old_pipe->stream_res.tg == new_pipe->stream_res.tg &&
+		new_pipe_link_enc && !new_pipe->stream->dpms_off &&
+		new_pipe_link_enc->funcs->is_dig_enabled &&
+		new_pipe_link_enc->funcs->is_dig_enabled(
+		new_pipe_link_enc) &&
+		new_pipe->stream_res.stream_enc &&
+		new_pipe->stream_res.stream_enc->funcs->is_fifo_enabled &&
+		new_pipe->stream_res.stream_enc->funcs->is_fifo_enabled(new_pipe->stream_res.stream_enc);
+
+		if (old_pipe->stream && new_pipe->stream && old_pipe->stream == new_pipe->stream) {
+			has_active_hpo =  dccg->ctx->dc->link_srv->dp_is_128b_132b_signal(old_pipe) &&
+			dccg->ctx->dc->link_srv->dp_is_128b_132b_signal(new_pipe);
+
+		}
+
+		if (!has_active_hpo && !stream_changed_otg_dig_on && pipe->stream &&
+		    (pipe->stream->dpms_off || dc_is_virtual_signal(pipe->stream->signal) || !pipe_link_enc) &&
+		    !dccg->ctx->dc->link_srv->dp_is_128b_132b_signal(pipe)) {
+			/* This w/a should not trigger when we have a dig active */
+			if (disable) {
+				if (pipe->stream_res.tg && pipe->stream_res.tg->funcs->immediate_disable_crtc)
+					pipe->stream_res.tg->funcs->immediate_disable_crtc(pipe->stream_res.tg);
+
+				reset_sync_context_for_pipe(dc, context, i);
+			} else {
+				pipe->stream_res.tg->funcs->enable_crtc(pipe->stream_res.tg);
+			}
+		}
+	}
+}
+
+static void dcn35_update_clocks_update_dtb_dto(struct clk_mgr_internal *clk_mgr,
+			struct dc_state *context,
+			int ref_dtbclk_khz)
+{
+	struct dccg *dccg = clk_mgr->dccg;
+	uint32_t tg_mask = 0;
+	int i;
+
+	for (i = 0; i < clk_mgr->base.ctx->dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i];
+		struct dtbclk_dto_params dto_params = {0};
+
+		/* use mask to program DTO once per tg */
+		if (pipe_ctx->stream_res.tg &&
+				!(tg_mask & (1 << pipe_ctx->stream_res.tg->inst))) {
+			tg_mask |= (1 << pipe_ctx->stream_res.tg->inst);
+
+			dto_params.otg_inst = pipe_ctx->stream_res.tg->inst;
+			dto_params.ref_dtbclk_khz = ref_dtbclk_khz;
+
+			dccg->funcs->set_dtbclk_dto(clk_mgr->dccg, &dto_params);
+			//dccg->funcs->set_audio_dtbclk_dto(clk_mgr->dccg, &dto_params);
+		}
+	}
+}
+
+static void dcn35_update_clocks_update_dpp_dto(struct clk_mgr_internal *clk_mgr,
+		struct dc_state *context, bool safe_to_lower)
+{
+	int i;
+	bool dppclk_active[MAX_PIPES] = {0};
+
+
+	clk_mgr->dccg->ref_dppclk = clk_mgr->base.clks.dppclk_khz;
+	for (i = 0; i < clk_mgr->base.ctx->dc->res_pool->pipe_count; i++) {
+		int dpp_inst = 0, dppclk_khz, prev_dppclk_khz;
+
+		dppclk_khz = context->res_ctx.pipe_ctx[i].plane_res.bw.dppclk_khz;
+
+		if (context->res_ctx.pipe_ctx[i].plane_res.dpp)
+			dpp_inst = context->res_ctx.pipe_ctx[i].plane_res.dpp->inst;
+		else if (!context->res_ctx.pipe_ctx[i].plane_res.dpp && dppclk_khz == 0) {
+			/* dpp == NULL && dppclk_khz == 0 is valid because of pipe harvesting.
+			 * In this case just continue in loop
+			 */
+			continue;
+		} else if (!context->res_ctx.pipe_ctx[i].plane_res.dpp && dppclk_khz > 0) {
+			/* The software state is not valid if dpp resource is NULL and
+			 * dppclk_khz > 0.
+			 */
+			ASSERT(false);
+			continue;
+		}
+
+		prev_dppclk_khz = clk_mgr->dccg->pipe_dppclk_khz[i];
+
+		if (safe_to_lower || prev_dppclk_khz < dppclk_khz)
+			clk_mgr->dccg->funcs->update_dpp_dto(
+							clk_mgr->dccg, dpp_inst, dppclk_khz);
+		dppclk_active[dpp_inst] = true;
+	}
+	if (safe_to_lower)
+		for (i = 0; i < clk_mgr->base.ctx->dc->res_pool->pipe_count; i++) {
+			struct dpp *old_dpp = clk_mgr->base.ctx->dc->current_state->res_ctx.pipe_ctx[i].plane_res.dpp;
+
+			if (old_dpp && !dppclk_active[old_dpp->inst])
+				clk_mgr->dccg->funcs->update_dpp_dto(clk_mgr->dccg, old_dpp->inst, 0);
+		}
+}
+
+static uint8_t get_lowest_dpia_index(const struct dc_link *link)
+{
+	const struct dc *dc_struct = link->dc;
+	uint8_t idx = 0xFF;
+	int i;
+
+	for (i = 0; i < MAX_PIPES * 2; ++i) {
+		if (!dc_struct->links[i] || dc_struct->links[i]->ep_type != DISPLAY_ENDPOINT_USB4_DPIA)
+			continue;
+
+		if (idx > dc_struct->links[i]->link_index)
+			idx = dc_struct->links[i]->link_index;
+	}
+
+	return idx;
+}
+
+static void dcn35_notify_host_router_bw(struct clk_mgr *clk_mgr_base, struct dc_state *context,
+					bool safe_to_lower)
+{
+	struct dc_clocks *new_clocks = &context->bw_ctx.bw.dcn.clk;
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	uint32_t host_router_bw_kbps[MAX_HOST_ROUTERS_NUM] = { 0 };
+	int i;
+	for (i = 0; i < context->stream_count; ++i) {
+		const struct dc_stream_state *stream = context->streams[i];
+		const struct dc_link *link = stream->link;
+		uint8_t lowest_dpia_index = 0;
+		unsigned int hr_index = 0;
+
+		if (!link)
+			continue;
+
+		lowest_dpia_index = get_lowest_dpia_index(link);
+		if (link->link_index < lowest_dpia_index)
+			continue;
+
+		hr_index = (link->link_index - lowest_dpia_index) / 2;
+		if (hr_index >= MAX_HOST_ROUTERS_NUM)
+			continue;
+		host_router_bw_kbps[hr_index] += dc_bandwidth_in_kbps_from_timing(
+			&stream->timing, dc_link_get_highest_encoding_format(link));
+	}
+
+	for (i = 0; i < MAX_HOST_ROUTERS_NUM; ++i) {
+		new_clocks->host_router_bw_kbps[i] = host_router_bw_kbps[i];
+		if (should_set_clock(safe_to_lower, new_clocks->host_router_bw_kbps[i], clk_mgr_base->clks.host_router_bw_kbps[i])) {
+			clk_mgr_base->clks.host_router_bw_kbps[i] = new_clocks->host_router_bw_kbps[i];
+			dcn35_smu_notify_host_router_bw(clk_mgr, i, new_clocks->host_router_bw_kbps[i]);
+		}
+	}
+}
+
+void dcn35_update_clocks(struct clk_mgr *clk_mgr_base,
+			struct dc_state *context,
+			bool safe_to_lower)
+{
+	union dmub_rb_cmd cmd;
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dc_clocks *new_clocks = &context->bw_ctx.bw.dcn.clk;
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	int display_count = 0;
+	bool update_dppclk = false;
+	bool update_dispclk = false;
+	bool dpp_clock_lowered = false;
+	int all_active_disps = 0;
+
+	if (dc->work_arounds.skip_clock_update)
+		return;
+
+	display_count = dcn35_get_active_display_cnt_wa(dc, context, &all_active_disps);
+	if (new_clocks->dtbclk_en && !new_clocks->ref_dtbclk_khz)
+		new_clocks->ref_dtbclk_khz = 600000;
+
+	/*
+	 * if it is safe to lower, but we are already in the lower state, we don't have to do anything
+	 * also if safe to lower is false, we just go in the higher state
+	 */
+	if (safe_to_lower) {
+		if (new_clocks->zstate_support != DCN_ZSTATE_SUPPORT_DISALLOW &&
+				new_clocks->zstate_support != clk_mgr_base->clks.zstate_support) {
+			dcn35_smu_set_zstate_support(clk_mgr, new_clocks->zstate_support);
+			dm_helpers_enable_periodic_detection(clk_mgr_base->ctx, true);
+			clk_mgr_base->clks.zstate_support = new_clocks->zstate_support;
+		}
+
+		if (clk_mgr_base->clks.dtbclk_en && !new_clocks->dtbclk_en) {
+			if (clk_mgr->base.ctx->dc->config.allow_0_dtb_clk)
+				dcn35_smu_set_dtbclk(clk_mgr, false);
+
+			clk_mgr_base->clks.dtbclk_en = new_clocks->dtbclk_en;
+		}
+		/* check that we're not already in lower */
+		if (clk_mgr_base->clks.pwr_state != DCN_PWR_STATE_LOW_POWER) {
+			/* if we can go lower, go lower */
+			if (display_count == 0)
+				clk_mgr_base->clks.pwr_state = DCN_PWR_STATE_LOW_POWER;
+		}
+	} else {
+		if (new_clocks->zstate_support == DCN_ZSTATE_SUPPORT_DISALLOW &&
+				new_clocks->zstate_support != clk_mgr_base->clks.zstate_support) {
+			dcn35_smu_set_zstate_support(clk_mgr, DCN_ZSTATE_SUPPORT_DISALLOW);
+			dm_helpers_enable_periodic_detection(clk_mgr_base->ctx, false);
+			clk_mgr_base->clks.zstate_support = new_clocks->zstate_support;
+		}
+
+		if (!clk_mgr_base->clks.dtbclk_en && new_clocks->dtbclk_en) {
+			int actual_dtbclk = 0;
+
+			dcn35_update_clocks_update_dtb_dto(clk_mgr, context, new_clocks->ref_dtbclk_khz);
+			dcn35_smu_set_dtbclk(clk_mgr, true);
+
+			actual_dtbclk = REG_READ(CLK1_CLK4_CURRENT_CNT);
+
+			if (actual_dtbclk) {
+				clk_mgr_base->clks.ref_dtbclk_khz = new_clocks->ref_dtbclk_khz;
+				clk_mgr_base->clks.dtbclk_en = new_clocks->dtbclk_en;
+			}
+		}
+
+		/* check that we're not already in D0 */
+		if (clk_mgr_base->clks.pwr_state != DCN_PWR_STATE_MISSION_MODE) {
+			union display_idle_optimization_u idle_info = { 0 };
+
+			dcn35_smu_set_display_idle_optimization(clk_mgr, idle_info.data);
+			/* update power state */
+			clk_mgr_base->clks.pwr_state = DCN_PWR_STATE_MISSION_MODE;
+		}
+	}
+	if (dc->debug.force_min_dcfclk_mhz > 0)
+		new_clocks->dcfclk_khz = (new_clocks->dcfclk_khz > (dc->debug.force_min_dcfclk_mhz * 1000)) ?
+				new_clocks->dcfclk_khz : (dc->debug.force_min_dcfclk_mhz * 1000);
+
+	if (should_set_clock(safe_to_lower, new_clocks->dcfclk_khz, clk_mgr_base->clks.dcfclk_khz)) {
+		clk_mgr_base->clks.dcfclk_khz = new_clocks->dcfclk_khz;
+		dcn35_smu_set_hard_min_dcfclk(clk_mgr, clk_mgr_base->clks.dcfclk_khz);
+	}
+
+	if (should_set_clock(safe_to_lower,
+			new_clocks->dcfclk_deep_sleep_khz, clk_mgr_base->clks.dcfclk_deep_sleep_khz)) {
+		clk_mgr_base->clks.dcfclk_deep_sleep_khz = new_clocks->dcfclk_deep_sleep_khz;
+		dcn35_smu_set_min_deep_sleep_dcfclk(clk_mgr, clk_mgr_base->clks.dcfclk_deep_sleep_khz);
+	}
+
+	// workaround: Limit dppclk to 100Mhz to avoid lower eDP panel switch to plus 4K monitor underflow.
+	if (new_clocks->dppclk_khz < 100000)
+		new_clocks->dppclk_khz = 100000;
+
+	if (should_set_clock(safe_to_lower, new_clocks->dppclk_khz, clk_mgr->base.clks.dppclk_khz)) {
+		if (clk_mgr->base.clks.dppclk_khz > new_clocks->dppclk_khz)
+			dpp_clock_lowered = true;
+		clk_mgr_base->clks.dppclk_khz = new_clocks->dppclk_khz;
+		update_dppclk = true;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, clk_mgr_base->clks.dispclk_khz) &&
+	    (new_clocks->dispclk_khz > 0 || (safe_to_lower && display_count == 0))) {
+		int requested_dispclk_khz = new_clocks->dispclk_khz;
+
+		dcn35_disable_otg_wa(clk_mgr_base, context, safe_to_lower, true);
+
+		/* Clamp the requested clock to PMFW based on their limit. */
+		if (dc->debug.min_disp_clk_khz > 0 && requested_dispclk_khz < dc->debug.min_disp_clk_khz)
+			requested_dispclk_khz = dc->debug.min_disp_clk_khz;
+
+		dcn35_smu_set_dispclk(clk_mgr, requested_dispclk_khz);
+		clk_mgr_base->clks.dispclk_khz = new_clocks->dispclk_khz;
+
+		dcn35_disable_otg_wa(clk_mgr_base, context, safe_to_lower, false);
+
+		update_dispclk = true;
+	}
+
+	/* clock limits are received with MHz precision, divide by 1000 to prevent setting clocks at every call */
+	if (!dc->debug.disable_dtb_ref_clk_switch &&
+	    should_set_clock(safe_to_lower, new_clocks->ref_dtbclk_khz / 1000,
+			     clk_mgr_base->clks.ref_dtbclk_khz / 1000)) {
+		dcn35_update_clocks_update_dtb_dto(clk_mgr, context, new_clocks->ref_dtbclk_khz);
+		clk_mgr_base->clks.ref_dtbclk_khz = new_clocks->ref_dtbclk_khz;
+	}
+
+	if (dpp_clock_lowered) {
+		// increase per DPP DTO before lowering global dppclk
+		dcn35_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
+		dcn35_smu_set_dppclk(clk_mgr, clk_mgr_base->clks.dppclk_khz);
+	} else {
+		// increase global DPPCLK before lowering per DPP DTO
+		if (update_dppclk || update_dispclk)
+			dcn35_smu_set_dppclk(clk_mgr, clk_mgr_base->clks.dppclk_khz);
+		dcn35_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
+	}
+
+	// notify PMFW of bandwidth per DPIA tunnel
+	if (dc->debug.notify_dpia_hr_bw)
+		dcn35_notify_host_router_bw(clk_mgr_base, context, safe_to_lower);
+
+	// notify DMCUB of latest clocks
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.notify_clocks.header.type = DMUB_CMD__CLK_MGR;
+	cmd.notify_clocks.header.sub_type = DMUB_CMD__CLK_MGR_NOTIFY_CLOCKS;
+	cmd.notify_clocks.clocks.dcfclk_khz = clk_mgr_base->clks.dcfclk_khz;
+	cmd.notify_clocks.clocks.dcfclk_deep_sleep_khz =
+		clk_mgr_base->clks.dcfclk_deep_sleep_khz;
+	cmd.notify_clocks.clocks.dispclk_khz = clk_mgr_base->clks.dispclk_khz;
+	cmd.notify_clocks.clocks.dppclk_khz = clk_mgr_base->clks.dppclk_khz;
+
+	dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+static int get_vco_frequency_from_reg(struct clk_mgr_internal *clk_mgr)
+{
+	/* get FbMult value */
+	struct fixed31_32 pll_req;
+	unsigned int fbmult_frac_val = 0;
+	unsigned int fbmult_int_val = 0;
+
+	/*
+	 * Register value of fbmult is in 8.16 format, we are converting to 314.32
+	 * to leverage the fix point operations available in driver
+	 */
+
+	REG_GET(CLK1_CLK_PLL_REQ, FbMult_frac, &fbmult_frac_val); /* 16 bit fractional part*/
+	REG_GET(CLK1_CLK_PLL_REQ, FbMult_int, &fbmult_int_val); /* 8 bit integer part */
+
+	pll_req = dc_fixpt_from_int(fbmult_int_val);
+
+	/*
+	 * since fractional part is only 16 bit in register definition but is 32 bit
+	 * in our fix point definiton, need to shift left by 16 to obtain correct value
+	 */
+	pll_req.value |= fbmult_frac_val << 16;
+
+	/* multiply by REFCLK period */
+	pll_req = dc_fixpt_mul_int(pll_req, clk_mgr->dfs_ref_freq_khz);
+
+	/* integer part is now VCO frequency in kHz */
+	return dc_fixpt_floor(pll_req);
+}
+
+static void dcn35_enable_pme_wa(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	dcn35_smu_enable_pme_wa(clk_mgr);
+}
+
+
+bool dcn35_are_clock_states_equal(struct dc_clocks *a,
+		struct dc_clocks *b)
+{
+	if (a->dispclk_khz != b->dispclk_khz)
+		return false;
+	else if (a->dppclk_khz != b->dppclk_khz)
+		return false;
+	else if (a->dcfclk_khz != b->dcfclk_khz)
+		return false;
+	else if (a->dcfclk_deep_sleep_khz != b->dcfclk_deep_sleep_khz)
+		return false;
+	else if (a->zstate_support != b->zstate_support)
+		return false;
+	else if (a->dtbclk_en != b->dtbclk_en)
+		return false;
+
+	return true;
+}
+
+static void dcn35_save_clk_registers_internal(struct dcn35_clk_internal *internal, struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	// read dtbclk
+	internal->CLK1_CLK4_CURRENT_CNT = REG_READ(CLK1_CLK4_CURRENT_CNT);
+	internal->CLK1_CLK4_BYPASS_CNTL = REG_READ(CLK1_CLK4_BYPASS_CNTL);
+
+	// read dcfclk
+	internal->CLK1_CLK3_CURRENT_CNT = REG_READ(CLK1_CLK3_CURRENT_CNT);
+	internal->CLK1_CLK3_BYPASS_CNTL = REG_READ(CLK1_CLK3_BYPASS_CNTL);
+
+	// read dcf deep sleep divider
+	internal->CLK1_CLK3_DS_CNTL = REG_READ(CLK1_CLK3_DS_CNTL);
+	internal->CLK1_CLK3_ALLOW_DS = REG_READ(CLK1_CLK3_ALLOW_DS);
+
+	// read dppclk
+	internal->CLK1_CLK1_CURRENT_CNT = REG_READ(CLK1_CLK1_CURRENT_CNT);
+	internal->CLK1_CLK1_BYPASS_CNTL = REG_READ(CLK1_CLK1_BYPASS_CNTL);
+
+	// read dprefclk
+	internal->CLK1_CLK2_CURRENT_CNT = REG_READ(CLK1_CLK2_CURRENT_CNT);
+	internal->CLK1_CLK2_BYPASS_CNTL = REG_READ(CLK1_CLK2_BYPASS_CNTL);
+
+	// read dispclk
+	internal->CLK1_CLK0_CURRENT_CNT = REG_READ(CLK1_CLK0_CURRENT_CNT);
+	internal->CLK1_CLK0_BYPASS_CNTL = REG_READ(CLK1_CLK0_BYPASS_CNTL);
+}
+
+static void dcn35_save_clk_registers(struct clk_state_registers_and_bypass *regs_and_bypass,
+		struct clk_mgr_dcn35 *clk_mgr)
+{
+	struct dcn35_clk_internal internal = {0};
+	char *bypass_clks[5] = {"0x0 DFS", "0x1 REFCLK", "0x2 ERROR", "0x3 400 FCH", "0x4 600 FCH"};
+
+	dcn35_save_clk_registers_internal(&internal, &clk_mgr->base.base);
+
+	regs_and_bypass->dcfclk = internal.CLK1_CLK3_CURRENT_CNT / 10;
+	regs_and_bypass->dcf_deep_sleep_divider = internal.CLK1_CLK3_DS_CNTL / 10;
+	regs_and_bypass->dcf_deep_sleep_allow = internal.CLK1_CLK3_ALLOW_DS;
+	regs_and_bypass->dprefclk = internal.CLK1_CLK2_CURRENT_CNT / 10;
+	regs_and_bypass->dispclk = internal.CLK1_CLK0_CURRENT_CNT / 10;
+	regs_and_bypass->dppclk = internal.CLK1_CLK1_CURRENT_CNT / 10;
+	regs_and_bypass->dtbclk = internal.CLK1_CLK4_CURRENT_CNT / 10;
+
+	regs_and_bypass->dppclk_bypass = internal.CLK1_CLK1_BYPASS_CNTL & 0x0007;
+	if (regs_and_bypass->dppclk_bypass < 0 || regs_and_bypass->dppclk_bypass > 4)
+		regs_and_bypass->dppclk_bypass = 0;
+	regs_and_bypass->dcfclk_bypass = internal.CLK1_CLK3_BYPASS_CNTL & 0x0007;
+	if (regs_and_bypass->dcfclk_bypass < 0 || regs_and_bypass->dcfclk_bypass > 4)
+		regs_and_bypass->dcfclk_bypass = 0;
+	regs_and_bypass->dispclk_bypass = internal.CLK1_CLK0_BYPASS_CNTL & 0x0007;
+	if (regs_and_bypass->dispclk_bypass < 0 || regs_and_bypass->dispclk_bypass > 4)
+		regs_and_bypass->dispclk_bypass = 0;
+	regs_and_bypass->dprefclk_bypass = internal.CLK1_CLK2_BYPASS_CNTL & 0x0007;
+	if (regs_and_bypass->dprefclk_bypass < 0 || regs_and_bypass->dprefclk_bypass > 4)
+		regs_and_bypass->dprefclk_bypass = 0;
+
+	if (clk_mgr->base.base.ctx->dc->debug.pstate_enabled) {
+		DC_LOG_SMU("clk_type,clk_value,deepsleep_cntl,deepsleep_allow,bypass\n");
+
+		DC_LOG_SMU("dcfclk,%d,%d,%d,%s\n",
+				   regs_and_bypass->dcfclk,
+				   regs_and_bypass->dcf_deep_sleep_divider,
+				   regs_and_bypass->dcf_deep_sleep_allow,
+				   bypass_clks[(int) regs_and_bypass->dcfclk_bypass]);
+
+		DC_LOG_SMU("dprefclk,%d,N/A,N/A,%s\n",
+			regs_and_bypass->dprefclk,
+			bypass_clks[(int) regs_and_bypass->dprefclk_bypass]);
+
+		DC_LOG_SMU("dispclk,%d,N/A,N/A,%s\n",
+			regs_and_bypass->dispclk,
+			bypass_clks[(int) regs_and_bypass->dispclk_bypass]);
+
+		// REGISTER VALUES
+		DC_LOG_SMU("reg_name,value,clk_type");
+
+		DC_LOG_SMU("CLK1_CLK3_CURRENT_CNT,%d,dcfclk",
+				internal.CLK1_CLK3_CURRENT_CNT);
+
+		DC_LOG_SMU("CLK1_CLK4_CURRENT_CNT,%d,dtbclk",
+					internal.CLK1_CLK4_CURRENT_CNT);
+
+		DC_LOG_SMU("CLK1_CLK3_DS_CNTL,%d,dcf_deep_sleep_divider",
+					internal.CLK1_CLK3_DS_CNTL);
+
+		DC_LOG_SMU("CLK1_CLK3_ALLOW_DS,%d,dcf_deep_sleep_allow",
+					internal.CLK1_CLK3_ALLOW_DS);
+
+		DC_LOG_SMU("CLK1_CLK2_CURRENT_CNT,%d,dprefclk",
+					internal.CLK1_CLK2_CURRENT_CNT);
+
+		DC_LOG_SMU("CLK1_CLK0_CURRENT_CNT,%d,dispclk",
+					internal.CLK1_CLK0_CURRENT_CNT);
+
+		DC_LOG_SMU("CLK1_CLK1_CURRENT_CNT,%d,dppclk",
+					internal.CLK1_CLK1_CURRENT_CNT);
+
+		DC_LOG_SMU("CLK1_CLK3_BYPASS_CNTL,%d,dcfclk_bypass",
+					internal.CLK1_CLK3_BYPASS_CNTL);
+
+		DC_LOG_SMU("CLK1_CLK2_BYPASS_CNTL,%d,dprefclk_bypass",
+					internal.CLK1_CLK2_BYPASS_CNTL);
+
+		DC_LOG_SMU("CLK1_CLK0_BYPASS_CNTL,%d,dispclk_bypass",
+					internal.CLK1_CLK0_BYPASS_CNTL);
+
+		DC_LOG_SMU("CLK1_CLK1_BYPASS_CNTL,%d,dppclk_bypass",
+					internal.CLK1_CLK1_BYPASS_CNTL);
+
+	}
+}
+
+static bool dcn35_is_spll_ssc_enabled(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	uint32_t ssc_enable;
+
+	if (clk_mgr_base->ctx->dce_version == DCN_VERSION_3_51) {
+		ssc_enable = REG_READ(CLK6_spll_field_8) & CLK6_spll_field_8__spll_ssc_en_MASK;
+	} else {
+		ssc_enable = REG_READ(CLK5_spll_field_8) & CLK5_spll_field_8__spll_ssc_en_MASK;
+	}
+
+	return ssc_enable != 0;
+}
+
+static void init_clk_states(struct clk_mgr *clk_mgr)
+{
+	uint32_t ref_dtbclk = clk_mgr->clks.ref_dtbclk_khz;
+
+	memset(&(clk_mgr->clks), 0, sizeof(struct dc_clocks));
+
+	clk_mgr->clks.ref_dtbclk_khz = ref_dtbclk;	// restore ref_dtbclk
+	clk_mgr->clks.p_state_change_support = true;
+	clk_mgr->clks.prev_p_state_change_support = true;
+	clk_mgr->clks.pwr_state = DCN_PWR_STATE_UNKNOWN;
+	clk_mgr->clks.zstate_support = DCN_ZSTATE_SUPPORT_UNKNOWN;
+}
+
+void dcn35_init_clocks(struct clk_mgr *clk_mgr)
+{
+	struct clk_mgr_internal *clk_mgr_int = TO_CLK_MGR_INTERNAL(clk_mgr);
+	struct clk_mgr_dcn35 *clk_mgr_dcn35 = TO_CLK_MGR_DCN35(clk_mgr_int);
+
+	init_clk_states(clk_mgr);
+
+	// to adjust dp_dto reference clock if ssc is enable otherwise to apply dprefclk
+	if (dcn35_is_spll_ssc_enabled(clk_mgr))
+		clk_mgr->dp_dto_source_clock_in_khz =
+			dce_adjust_dp_ref_freq_for_ss(clk_mgr_int, clk_mgr->dprefclk_khz);
+	else
+		clk_mgr->dp_dto_source_clock_in_khz = clk_mgr->dprefclk_khz;
+
+	dcn35_save_clk_registers(&clk_mgr->boot_snapshot, clk_mgr_dcn35);
+
+	clk_mgr->clks.ref_dtbclk_khz =  clk_mgr->boot_snapshot.dtbclk * 10;
+	if (clk_mgr->boot_snapshot.dtbclk > 59000) {
+		/*dtbclk enabled based on */
+		clk_mgr->clks.dtbclk_en = true;
+	}
+}
+static struct clk_bw_params dcn35_bw_params = {
+	.vram_type = Ddr4MemType,
+	.num_channels = 1,
+	.clk_table = {
+		.num_entries = 4,
+	},
+
+};
+
+static struct wm_table ddr5_wm_table = {
+	.entries = {
+		{
+			.wm_inst = WM_A,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 28.0,
+			.sr_enter_plus_exit_time_us = 30.0,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_B,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 28.0,
+			.sr_enter_plus_exit_time_us = 30.0,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_C,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 28.0,
+			.sr_enter_plus_exit_time_us = 30.0,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_D,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 28.0,
+			.sr_enter_plus_exit_time_us = 30.0,
+			.valid = true,
+		},
+	}
+};
+
+static struct wm_table lpddr5_wm_table = {
+	.entries = {
+		{
+			.wm_inst = WM_A,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 28.0,
+			.sr_enter_plus_exit_time_us = 30.0,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_B,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 28.0,
+			.sr_enter_plus_exit_time_us = 30.0,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_C,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 28.0,
+			.sr_enter_plus_exit_time_us = 30.0,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_D,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 28.0,
+			.sr_enter_plus_exit_time_us = 30.0,
+			.valid = true,
+		},
+	}
+};
+
+static DpmClocks_t_dcn35 dummy_clocks;
+static DpmClocks_t_dcn351 dummy_clocks_dcn351;
+
+static struct dcn35_watermarks dummy_wms = { 0 };
+
+static struct dcn35_ss_info_table ss_info_table = {
+	.ss_divider = 1000,
+	.ss_percentage = {0, 0, 375, 375, 375}
+};
+
+static void dcn35_read_ss_info_from_lut(struct clk_mgr_internal *clk_mgr)
+{
+	uint32_t clock_source = 0;
+
+	clock_source = REG_READ(CLK1_CLK2_BYPASS_CNTL) & CLK1_CLK2_BYPASS_CNTL__CLK2_BYPASS_SEL_MASK;
+
+	// If it's DFS mode, clock_source is 0.
+	if (dcn35_is_spll_ssc_enabled(&clk_mgr->base) && (clock_source < ARRAY_SIZE(ss_info_table.ss_percentage))) {
+		clk_mgr->dprefclk_ss_percentage = ss_info_table.ss_percentage[clock_source];
+
+		if (clk_mgr->dprefclk_ss_percentage != 0) {
+			clk_mgr->ss_on_dprefclk = true;
+			clk_mgr->dprefclk_ss_divider = ss_info_table.ss_divider;
+		}
+	}
+}
+
+static void dcn35_build_watermark_ranges(struct clk_bw_params *bw_params, struct dcn35_watermarks *table)
+{
+	int i, num_valid_sets;
+
+	num_valid_sets = 0;
+
+	for (i = 0; i < WM_SET_COUNT; i++) {
+		/* skip empty entries, the smu array has no holes*/
+		if (!bw_params->wm_table.entries[i].valid)
+			continue;
+
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].WmSetting = bw_params->wm_table.entries[i].wm_inst;
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].WmType = bw_params->wm_table.entries[i].wm_type;
+		/* We will not select WM based on fclk, so leave it as unconstrained */
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinClock = 0;
+		table->WatermarkRow[WM_DCFCLK][num_valid_sets].MaxClock = 0xFFFF;
+
+		if (table->WatermarkRow[WM_DCFCLK][num_valid_sets].WmType == WM_TYPE_PSTATE_CHG) {
+			if (i == 0)
+				table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinMclk = 0;
+			else {
+				/* add 1 to make it non-overlapping with next lvl */
+				table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinMclk =
+						bw_params->clk_table.entries[i - 1].dcfclk_mhz + 1;
+			}
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets].MaxMclk =
+					bw_params->clk_table.entries[i].dcfclk_mhz;
+
+		} else {
+			/* unconstrained for memory retraining */
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets].MinClock = 0;
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets].MaxClock = 0xFFFF;
+
+			/* Modify previous watermark range to cover up to max */
+			table->WatermarkRow[WM_DCFCLK][num_valid_sets - 1].MaxClock = 0xFFFF;
+		}
+		num_valid_sets++;
+	}
+
+	ASSERT(num_valid_sets != 0); /* Must have at least one set of valid watermarks */
+
+	/* modify the min and max to make sure we cover the whole range*/
+	table->WatermarkRow[WM_DCFCLK][0].MinMclk = 0;
+	table->WatermarkRow[WM_DCFCLK][0].MinClock = 0;
+	table->WatermarkRow[WM_DCFCLK][num_valid_sets - 1].MaxMclk = 0xFFFF;
+	table->WatermarkRow[WM_DCFCLK][num_valid_sets - 1].MaxClock = 0xFFFF;
+
+	/* This is for writeback only, does not matter currently as no writeback support*/
+	table->WatermarkRow[WM_SOCCLK][0].WmSetting = WM_A;
+	table->WatermarkRow[WM_SOCCLK][0].MinClock = 0;
+	table->WatermarkRow[WM_SOCCLK][0].MaxClock = 0xFFFF;
+	table->WatermarkRow[WM_SOCCLK][0].MinMclk = 0;
+	table->WatermarkRow[WM_SOCCLK][0].MaxMclk = 0xFFFF;
+}
+
+static void dcn35_notify_wm_ranges(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct clk_mgr_dcn35 *clk_mgr_dcn35 = TO_CLK_MGR_DCN35(clk_mgr);
+	struct dcn35_watermarks *table = clk_mgr_dcn35->smu_wm_set.wm_set;
+
+	if (!clk_mgr->smu_ver)
+		return;
+
+	if (!table || clk_mgr_dcn35->smu_wm_set.mc_address.quad_part == 0)
+		return;
+
+	memset(table, 0, sizeof(*table));
+
+	dcn35_build_watermark_ranges(clk_mgr_base->bw_params, table);
+
+	dcn35_smu_set_dram_addr_high(clk_mgr,
+			clk_mgr_dcn35->smu_wm_set.mc_address.high_part);
+	dcn35_smu_set_dram_addr_low(clk_mgr,
+			clk_mgr_dcn35->smu_wm_set.mc_address.low_part);
+	dcn35_smu_transfer_wm_table_dram_2_smu(clk_mgr);
+}
+
+static void dcn35_get_dpm_table_from_smu(struct clk_mgr_internal *clk_mgr,
+		struct dcn35_smu_dpm_clks *smu_dpm_clks)
+{
+	DpmClocks_t_dcn35 *table = smu_dpm_clks->dpm_clks;
+
+	if (!clk_mgr->smu_ver)
+		return;
+
+	if (!table || smu_dpm_clks->mc_address.quad_part == 0)
+		return;
+
+	memset(table, 0, sizeof(*table));
+
+	dcn35_smu_set_dram_addr_high(clk_mgr,
+			smu_dpm_clks->mc_address.high_part);
+	dcn35_smu_set_dram_addr_low(clk_mgr,
+			smu_dpm_clks->mc_address.low_part);
+	dcn35_smu_transfer_dpm_table_smu_2_dram(clk_mgr);
+}
+
+static void dcn351_get_dpm_table_from_smu(struct clk_mgr_internal *clk_mgr,
+		struct dcn351_smu_dpm_clks *smu_dpm_clks)
+{
+	DpmClocks_t_dcn351 *table = smu_dpm_clks->dpm_clks;
+
+	if (!clk_mgr->smu_ver)
+		return;
+	if (!table || smu_dpm_clks->mc_address.quad_part == 0)
+		return;
+	memset(table, 0, sizeof(*table));
+	dcn35_smu_set_dram_addr_high(clk_mgr,
+			smu_dpm_clks->mc_address.high_part);
+	dcn35_smu_set_dram_addr_low(clk_mgr,
+			smu_dpm_clks->mc_address.low_part);
+	dcn35_smu_transfer_dpm_table_smu_2_dram(clk_mgr);
+}
+static uint32_t find_max_clk_value(const uint32_t clocks[], uint32_t num_clocks)
+{
+	uint32_t max = 0;
+	int i;
+
+	for (i = 0; i < num_clocks; ++i) {
+		if (clocks[i] > max)
+			max = clocks[i];
+	}
+
+	return max;
+}
+
+static inline bool is_valid_clock_value(uint32_t clock_value)
+{
+	return clock_value > 1 && clock_value < 100000;
+}
+
+static unsigned int convert_wck_ratio(uint8_t wck_ratio)
+{
+	switch (wck_ratio) {
+	case WCK_RATIO_1_2:
+		return 2;
+
+	case WCK_RATIO_1_4:
+		return 4;
+	/* Find lowest DPM, FCLK is filled in reverse order*/
+
+	default:
+			break;
+	}
+
+	return 1;
+}
+
+static inline uint32_t calc_dram_speed_mts(const MemPstateTable_t *entry)
+{
+	return entry->UClk * convert_wck_ratio(entry->WckRatio) * 2;
+}
+
+static void dcn35_clk_mgr_helper_populate_bw_params(struct clk_mgr_internal *clk_mgr,
+						    struct integrated_info *bios_info,
+						    DpmClocks_t_dcn35 *clock_table)
+{
+	struct clk_bw_params *bw_params = clk_mgr->base.bw_params;
+	struct clk_limit_table_entry def_max = bw_params->clk_table.entries[bw_params->clk_table.num_entries - 1];
+	uint32_t max_fclk = 0, min_pstate = 0, max_dispclk = 0, max_dppclk = 0;
+	uint32_t max_pstate = 0, max_dram_speed_mts = 0, min_dram_speed_mts = 0;
+	uint32_t num_memps, num_fclk, num_dcfclk;
+	int i;
+
+	/* Determine min/max p-state values. */
+	num_memps = (clock_table->NumMemPstatesEnabled > NUM_MEM_PSTATE_LEVELS) ? NUM_MEM_PSTATE_LEVELS :
+		clock_table->NumMemPstatesEnabled;
+	for (i = 0; i < num_memps; i++) {
+		uint32_t dram_speed_mts = calc_dram_speed_mts(&clock_table->MemPstateTable[i]);
+
+		if (is_valid_clock_value(dram_speed_mts) && dram_speed_mts > max_dram_speed_mts) {
+			max_dram_speed_mts = dram_speed_mts;
+			max_pstate = i;
+		}
+	}
+
+	min_dram_speed_mts = max_dram_speed_mts;
+	min_pstate = max_pstate;
+
+	for (i = 0; i < num_memps; i++) {
+		uint32_t dram_speed_mts = calc_dram_speed_mts(&clock_table->MemPstateTable[i]);
+
+		if (is_valid_clock_value(dram_speed_mts) && dram_speed_mts < min_dram_speed_mts) {
+			min_dram_speed_mts = dram_speed_mts;
+			min_pstate = i;
+		}
+	}
+
+	/* We expect the table to contain at least one valid P-state entry. */
+	ASSERT(clock_table->NumMemPstatesEnabled &&
+	       is_valid_clock_value(max_dram_speed_mts) &&
+	       is_valid_clock_value(min_dram_speed_mts));
+
+	/* dispclk and dppclk can be max at any voltage, same number of levels for both */
+	if (clock_table->NumDispClkLevelsEnabled <= NUM_DISPCLK_DPM_LEVELS &&
+	    clock_table->NumDispClkLevelsEnabled <= NUM_DPPCLK_DPM_LEVELS) {
+		max_dispclk = find_max_clk_value(clock_table->DispClocks,
+			clock_table->NumDispClkLevelsEnabled);
+		max_dppclk = find_max_clk_value(clock_table->DppClocks,
+			clock_table->NumDispClkLevelsEnabled);
+	} else {
+		/* Invalid number of entries in the table from PMFW. */
+		ASSERT(0);
+	}
+
+	/* Base the clock table on dcfclk, need at least one entry regardless of pmfw table */
+	ASSERT(clock_table->NumDcfClkLevelsEnabled > 0);
+
+	num_fclk = (clock_table->NumFclkLevelsEnabled > NUM_FCLK_DPM_LEVELS) ? NUM_FCLK_DPM_LEVELS :
+		clock_table->NumFclkLevelsEnabled;
+	max_fclk = find_max_clk_value(clock_table->FclkClocks_Freq, num_fclk);
+
+	num_dcfclk = (clock_table->NumDcfClkLevelsEnabled > NUM_DCFCLK_DPM_LEVELS) ? NUM_DCFCLK_DPM_LEVELS :
+		clock_table->NumDcfClkLevelsEnabled;
+	for (i = 0; i < num_dcfclk; i++) {
+		int j;
+
+		/* First search defaults for the clocks we don't read using closest lower or equal default dcfclk */
+		for (j = bw_params->clk_table.num_entries - 1; j > 0; j--)
+			if (bw_params->clk_table.entries[j].dcfclk_mhz <= clock_table->DcfClocks[i])
+				break;
+
+		bw_params->clk_table.entries[i].phyclk_mhz = bw_params->clk_table.entries[j].phyclk_mhz;
+		bw_params->clk_table.entries[i].phyclk_d18_mhz = bw_params->clk_table.entries[j].phyclk_d18_mhz;
+		bw_params->clk_table.entries[i].dtbclk_mhz = bw_params->clk_table.entries[j].dtbclk_mhz;
+
+		/* Now update clocks we do read */
+		bw_params->clk_table.entries[i].memclk_mhz = clock_table->MemPstateTable[min_pstate].MemClk;
+		bw_params->clk_table.entries[i].voltage = clock_table->MemPstateTable[min_pstate].Voltage;
+		bw_params->clk_table.entries[i].dcfclk_mhz = clock_table->DcfClocks[i];
+		bw_params->clk_table.entries[i].socclk_mhz = clock_table->SocClocks[i];
+		bw_params->clk_table.entries[i].dispclk_mhz = max_dispclk;
+		bw_params->clk_table.entries[i].dppclk_mhz = max_dppclk;
+		bw_params->clk_table.entries[i].wck_ratio =
+			convert_wck_ratio(clock_table->MemPstateTable[min_pstate].WckRatio);
+
+		/* Dcfclk and Fclk are tied, but at a different ratio */
+		bw_params->clk_table.entries[i].fclk_mhz = min(max_fclk, 2 * clock_table->DcfClocks[i]);
+	}
+
+	/* Make sure to include at least one entry at highest pstate */
+	if (max_pstate != min_pstate || i == 0) {
+		if (i > MAX_NUM_DPM_LVL - 1)
+			i = MAX_NUM_DPM_LVL - 1;
+
+		bw_params->clk_table.entries[i].fclk_mhz = max_fclk;
+		bw_params->clk_table.entries[i].memclk_mhz = clock_table->MemPstateTable[max_pstate].MemClk;
+		bw_params->clk_table.entries[i].voltage = clock_table->MemPstateTable[max_pstate].Voltage;
+		bw_params->clk_table.entries[i].dcfclk_mhz =
+			find_max_clk_value(clock_table->DcfClocks, NUM_DCFCLK_DPM_LEVELS);
+		bw_params->clk_table.entries[i].socclk_mhz =
+			find_max_clk_value(clock_table->SocClocks, NUM_SOCCLK_DPM_LEVELS);
+		bw_params->clk_table.entries[i].dispclk_mhz = max_dispclk;
+		bw_params->clk_table.entries[i].dppclk_mhz = max_dppclk;
+		bw_params->clk_table.entries[i].wck_ratio = convert_wck_ratio(
+			clock_table->MemPstateTable[max_pstate].WckRatio);
+		i++;
+	}
+	bw_params->clk_table.num_entries = i--;
+
+	/* Make sure all highest clocks are included*/
+	bw_params->clk_table.entries[i].socclk_mhz =
+		find_max_clk_value(clock_table->SocClocks, NUM_SOCCLK_DPM_LEVELS);
+	bw_params->clk_table.entries[i].dispclk_mhz =
+		find_max_clk_value(clock_table->DispClocks, NUM_DISPCLK_DPM_LEVELS);
+	bw_params->clk_table.entries[i].dppclk_mhz =
+		find_max_clk_value(clock_table->DppClocks, NUM_DPPCLK_DPM_LEVELS);
+	bw_params->clk_table.entries[i].fclk_mhz =
+		find_max_clk_value(clock_table->FclkClocks_Freq, NUM_FCLK_DPM_LEVELS);
+	ASSERT(clock_table->DcfClocks[i] == find_max_clk_value(clock_table->DcfClocks, NUM_DCFCLK_DPM_LEVELS));
+	bw_params->clk_table.entries[i].phyclk_mhz = def_max.phyclk_mhz;
+	bw_params->clk_table.entries[i].phyclk_d18_mhz = def_max.phyclk_d18_mhz;
+	bw_params->clk_table.entries[i].dtbclk_mhz = def_max.dtbclk_mhz;
+	bw_params->clk_table.num_entries_per_clk.num_dcfclk_levels = clock_table->NumDcfClkLevelsEnabled;
+	bw_params->clk_table.num_entries_per_clk.num_dispclk_levels = clock_table->NumDispClkLevelsEnabled;
+	bw_params->clk_table.num_entries_per_clk.num_dppclk_levels = clock_table->NumDispClkLevelsEnabled;
+	bw_params->clk_table.num_entries_per_clk.num_fclk_levels = clock_table->NumFclkLevelsEnabled;
+	bw_params->clk_table.num_entries_per_clk.num_memclk_levels = clock_table->NumMemPstatesEnabled;
+	bw_params->clk_table.num_entries_per_clk.num_socclk_levels = clock_table->NumSocClkLevelsEnabled;
+
+	/*
+	 * Set any 0 clocks to max default setting. Not an issue for
+	 * power since we aren't doing switching in such case anyway
+	 */
+	for (i = 0; i < bw_params->clk_table.num_entries; i++) {
+		if (!bw_params->clk_table.entries[i].fclk_mhz) {
+			bw_params->clk_table.entries[i].fclk_mhz = def_max.fclk_mhz;
+			bw_params->clk_table.entries[i].memclk_mhz = def_max.memclk_mhz;
+			bw_params->clk_table.entries[i].voltage = def_max.voltage;
+		}
+		if (!bw_params->clk_table.entries[i].dcfclk_mhz)
+			bw_params->clk_table.entries[i].dcfclk_mhz = def_max.dcfclk_mhz;
+		if (!bw_params->clk_table.entries[i].socclk_mhz)
+			bw_params->clk_table.entries[i].socclk_mhz = def_max.socclk_mhz;
+		if (!bw_params->clk_table.entries[i].dispclk_mhz)
+			bw_params->clk_table.entries[i].dispclk_mhz = def_max.dispclk_mhz;
+		if (!bw_params->clk_table.entries[i].dppclk_mhz)
+			bw_params->clk_table.entries[i].dppclk_mhz = def_max.dppclk_mhz;
+		if (!bw_params->clk_table.entries[i].fclk_mhz)
+			bw_params->clk_table.entries[i].fclk_mhz = def_max.fclk_mhz;
+		if (!bw_params->clk_table.entries[i].phyclk_mhz)
+			bw_params->clk_table.entries[i].phyclk_mhz = def_max.phyclk_mhz;
+		if (!bw_params->clk_table.entries[i].phyclk_d18_mhz)
+			bw_params->clk_table.entries[i].phyclk_d18_mhz = def_max.phyclk_d18_mhz;
+		if (!bw_params->clk_table.entries[i].dtbclk_mhz)
+			bw_params->clk_table.entries[i].dtbclk_mhz = def_max.dtbclk_mhz;
+	}
+	ASSERT(bw_params->clk_table.entries[i-1].dcfclk_mhz);
+	bw_params->vram_type = bios_info->memory_type;
+	bw_params->dram_channel_width_bytes = bios_info->memory_type == 0x22 ? 8 : 4;
+	bw_params->num_channels = bios_info->ma_channel_number ? bios_info->ma_channel_number : 4;
+
+	for (i = 0; i < WM_SET_COUNT; i++) {
+		bw_params->wm_table.entries[i].wm_inst = i;
+
+		if (i >= bw_params->clk_table.num_entries) {
+			bw_params->wm_table.entries[i].valid = false;
+			continue;
+		}
+
+		bw_params->wm_table.entries[i].wm_type = WM_TYPE_PSTATE_CHG;
+		bw_params->wm_table.entries[i].valid = true;
+	}
+}
+
+static void dcn35_set_low_power_state(struct clk_mgr *clk_mgr_base)
+{
+	int display_count;
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	struct dc_state *context = dc->current_state;
+
+	if (clk_mgr_base->clks.pwr_state != DCN_PWR_STATE_LOW_POWER) {
+		display_count = dcn35_get_active_display_cnt_wa(dc, context, NULL);
+		/* if we can go lower, go lower */
+		if (display_count == 0)
+			clk_mgr_base->clks.pwr_state = DCN_PWR_STATE_LOW_POWER;
+	}
+}
+
+static void dcn35_exit_low_power_state(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	//SMU optimization is performed part of low power state exit.
+	dcn35_smu_exit_low_power_state(clk_mgr);
+
+}
+
+static bool dcn35_is_ips_supported(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	return dcn35_smu_get_ips_supported(clk_mgr) ? true : false;
+}
+
+static void dcn35_init_clocks_fpga(struct clk_mgr *clk_mgr)
+{
+	init_clk_states(clk_mgr);
+
+/* TODO: Implement the functions and remove the ifndef guard */
+}
+
+static void dcn35_update_clocks_fpga(struct clk_mgr *clk_mgr,
+		struct dc_state *context,
+		bool safe_to_lower)
+{
+	struct clk_mgr_internal *clk_mgr_int = TO_CLK_MGR_INTERNAL(clk_mgr);
+	struct dc_clocks *new_clocks = &context->bw_ctx.bw.dcn.clk;
+	int fclk_adj = new_clocks->fclk_khz;
+
+	/* TODO: remove this after correctly set by DML */
+	new_clocks->dcfclk_khz = 400000;
+	new_clocks->socclk_khz = 400000;
+
+	/* Min fclk = 1.2GHz since all the extra scemi logic seems to run off of it */
+	//int fclk_adj = new_clocks->fclk_khz > 1200000 ? new_clocks->fclk_khz : 1200000;
+	new_clocks->fclk_khz = 4320000;
+
+	if (should_set_clock(safe_to_lower, new_clocks->phyclk_khz, clk_mgr->clks.phyclk_khz)) {
+		clk_mgr->clks.phyclk_khz = new_clocks->phyclk_khz;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dcfclk_khz, clk_mgr->clks.dcfclk_khz)) {
+		clk_mgr->clks.dcfclk_khz = new_clocks->dcfclk_khz;
+	}
+
+	if (should_set_clock(safe_to_lower,
+			new_clocks->dcfclk_deep_sleep_khz, clk_mgr->clks.dcfclk_deep_sleep_khz)) {
+		clk_mgr->clks.dcfclk_deep_sleep_khz = new_clocks->dcfclk_deep_sleep_khz;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->socclk_khz, clk_mgr->clks.socclk_khz)) {
+		clk_mgr->clks.socclk_khz = new_clocks->socclk_khz;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dramclk_khz, clk_mgr->clks.dramclk_khz)) {
+		clk_mgr->clks.dramclk_khz = new_clocks->dramclk_khz;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dppclk_khz, clk_mgr->clks.dppclk_khz)) {
+		clk_mgr->clks.dppclk_khz = new_clocks->dppclk_khz;
+	}
+
+	if (should_set_clock(safe_to_lower, fclk_adj, clk_mgr->clks.fclk_khz)) {
+		clk_mgr->clks.fclk_khz = fclk_adj;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, clk_mgr->clks.dispclk_khz)) {
+		clk_mgr->clks.dispclk_khz = new_clocks->dispclk_khz;
+	}
+
+	/* Both fclk and ref_dppclk run on the same scemi clock.
+	 * So take the higher value since the DPP DTO is typically programmed
+	 * such that max dppclk is 1:1 with ref_dppclk.
+	 */
+	if (clk_mgr->clks.fclk_khz > clk_mgr->clks.dppclk_khz)
+		clk_mgr->clks.dppclk_khz = clk_mgr->clks.fclk_khz;
+	if (clk_mgr->clks.dppclk_khz > clk_mgr->clks.fclk_khz)
+		clk_mgr->clks.fclk_khz = clk_mgr->clks.dppclk_khz;
+
+	// Both fclk and ref_dppclk run on the same scemi clock.
+	clk_mgr_int->dccg->ref_dppclk = clk_mgr->clks.fclk_khz;
+
+	/* TODO: set dtbclk in correct place */
+	clk_mgr->clks.dtbclk_en = true;
+	dm_set_dcn_clocks(clk_mgr->ctx, &clk_mgr->clks);
+	dcn35_update_clocks_update_dpp_dto(clk_mgr_int, context, safe_to_lower);
+
+	dcn35_update_clocks_update_dtb_dto(clk_mgr_int, context, clk_mgr->clks.ref_dtbclk_khz);
+}
+
+static struct clk_mgr_funcs dcn35_funcs = {
+	.get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz,
+	.get_dtb_ref_clk_frequency = dcn31_get_dtb_ref_freq_khz,
+	.update_clocks = dcn35_update_clocks,
+	.init_clocks = dcn35_init_clocks,
+	.enable_pme_wa = dcn35_enable_pme_wa,
+	.are_clock_states_equal = dcn35_are_clock_states_equal,
+	.notify_wm_ranges = dcn35_notify_wm_ranges,
+	.set_low_power_state = dcn35_set_low_power_state,
+	.exit_low_power_state = dcn35_exit_low_power_state,
+	.is_ips_supported = dcn35_is_ips_supported,
+};
+
+struct clk_mgr_funcs dcn35_fpga_funcs = {
+	.get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz,
+	.update_clocks = dcn35_update_clocks_fpga,
+	.init_clocks = dcn35_init_clocks_fpga,
+	.get_dtb_ref_clk_frequency = dcn31_get_dtb_ref_freq_khz,
+};
+
+static void translate_to_DpmClocks_t_dcn35(struct dcn351_smu_dpm_clks *smu_dpm_clks_a,
+		struct dcn35_smu_dpm_clks *smu_dpm_clks_b)
+{
+	/*translate two structures and only take need clock tables*/
+	uint8_t i;
+
+	if (smu_dpm_clks_a == NULL || smu_dpm_clks_b == NULL ||
+		smu_dpm_clks_a->dpm_clks == NULL || smu_dpm_clks_b->dpm_clks == NULL)
+		return;
+
+	for (i = 0; i < NUM_DCFCLK_DPM_LEVELS; i++)
+		smu_dpm_clks_b->dpm_clks->DcfClocks[i] = smu_dpm_clks_a->dpm_clks->DcfClocks[i];
+
+	for (i = 0; i < NUM_DISPCLK_DPM_LEVELS; i++)
+		smu_dpm_clks_b->dpm_clks->DispClocks[i] = smu_dpm_clks_a->dpm_clks->DispClocks[i];
+
+	for (i = 0; i < NUM_DPPCLK_DPM_LEVELS; i++)
+		smu_dpm_clks_b->dpm_clks->DppClocks[i] = smu_dpm_clks_a->dpm_clks->DppClocks[i];
+
+	for (i = 0; i < NUM_FCLK_DPM_LEVELS; i++) {
+		smu_dpm_clks_b->dpm_clks->FclkClocks_Freq[i] = smu_dpm_clks_a->dpm_clks->FclkClocks_Freq[i];
+		smu_dpm_clks_b->dpm_clks->FclkClocks_Voltage[i] = smu_dpm_clks_a->dpm_clks->FclkClocks_Voltage[i];
+	}
+	for (i = 0; i < NUM_MEM_PSTATE_LEVELS; i++) {
+		smu_dpm_clks_b->dpm_clks->MemPstateTable[i].MemClk =
+			smu_dpm_clks_a->dpm_clks->MemPstateTable[i].MemClk;
+		smu_dpm_clks_b->dpm_clks->MemPstateTable[i].UClk =
+			smu_dpm_clks_a->dpm_clks->MemPstateTable[i].UClk;
+		smu_dpm_clks_b->dpm_clks->MemPstateTable[i].Voltage =
+			smu_dpm_clks_a->dpm_clks->MemPstateTable[i].Voltage;
+		smu_dpm_clks_b->dpm_clks->MemPstateTable[i].WckRatio =
+			smu_dpm_clks_a->dpm_clks->MemPstateTable[i].WckRatio;
+	}
+	smu_dpm_clks_b->dpm_clks->MaxGfxClk = smu_dpm_clks_a->dpm_clks->MaxGfxClk;
+	smu_dpm_clks_b->dpm_clks->MinGfxClk = smu_dpm_clks_a->dpm_clks->MinGfxClk;
+	smu_dpm_clks_b->dpm_clks->NumDcfClkLevelsEnabled =
+		smu_dpm_clks_a->dpm_clks->NumDcfClkLevelsEnabled;
+	smu_dpm_clks_b->dpm_clks->NumDispClkLevelsEnabled =
+		smu_dpm_clks_a->dpm_clks->NumDispClkLevelsEnabled;
+	smu_dpm_clks_b->dpm_clks->NumFclkLevelsEnabled =
+		smu_dpm_clks_a->dpm_clks->NumFclkLevelsEnabled;
+	smu_dpm_clks_b->dpm_clks->NumMemPstatesEnabled =
+		smu_dpm_clks_a->dpm_clks->NumMemPstatesEnabled;
+	smu_dpm_clks_b->dpm_clks->NumSocClkLevelsEnabled =
+		smu_dpm_clks_a->dpm_clks->NumSocClkLevelsEnabled;
+
+	for (i = 0; i < NUM_SOC_VOLTAGE_LEVELS; i++) {
+		smu_dpm_clks_b->dpm_clks->SocClocks[i] = smu_dpm_clks_a->dpm_clks->SocClocks[i];
+		smu_dpm_clks_b->dpm_clks->SocVoltage[i] = smu_dpm_clks_a->dpm_clks->SocVoltage[i];
+	}
+}
+void dcn35_clk_mgr_construct(
+		struct dc_context *ctx,
+		struct clk_mgr_dcn35 *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg)
+{
+	struct dcn35_smu_dpm_clks smu_dpm_clks = { 0 };
+	struct dcn351_smu_dpm_clks smu_dpm_clks_dcn351 = { 0 };
+	clk_mgr->base.base.ctx = ctx;
+	clk_mgr->base.base.funcs = &dcn35_funcs;
+
+	clk_mgr->base.pp_smu = pp_smu;
+
+	clk_mgr->base.dccg = dccg;
+	clk_mgr->base.dfs_bypass_disp_clk = 0;
+
+	clk_mgr->base.dprefclk_ss_percentage = 0;
+	clk_mgr->base.dprefclk_ss_divider = 1000;
+	clk_mgr->base.ss_on_dprefclk = false;
+	clk_mgr->base.dfs_ref_freq_khz = 48000;
+	if (ctx->dce_version != DCN_VERSION_3_51) {
+		clk_mgr->base.regs = &clk_mgr_regs_dcn35;
+		clk_mgr->base.clk_mgr_shift = &clk_mgr_shift_dcn35;
+		clk_mgr->base.clk_mgr_mask = &clk_mgr_mask_dcn35;
+	}
+
+
+	clk_mgr->smu_wm_set.wm_set = (struct dcn35_watermarks *)dm_helpers_allocate_gpu_mem(
+				clk_mgr->base.base.ctx,
+				DC_MEM_ALLOC_TYPE_GART,
+				sizeof(struct dcn35_watermarks),
+				&clk_mgr->smu_wm_set.mc_address.quad_part);
+
+	if (!clk_mgr->smu_wm_set.wm_set) {
+		clk_mgr->smu_wm_set.wm_set = &dummy_wms;
+		clk_mgr->smu_wm_set.mc_address.quad_part = 0;
+	}
+	ASSERT(clk_mgr->smu_wm_set.wm_set);
+
+	smu_dpm_clks.dpm_clks = (DpmClocks_t_dcn35 *)dm_helpers_allocate_gpu_mem(
+				clk_mgr->base.base.ctx,
+				DC_MEM_ALLOC_TYPE_GART,
+				sizeof(DpmClocks_t_dcn35),
+				&smu_dpm_clks.mc_address.quad_part);
+	if (smu_dpm_clks.dpm_clks == NULL) {
+		smu_dpm_clks.dpm_clks = &dummy_clocks;
+		smu_dpm_clks.mc_address.quad_part = 0;
+	}
+	ASSERT(smu_dpm_clks.dpm_clks);
+
+	if (ctx->dce_version == DCN_VERSION_3_51) {
+		smu_dpm_clks_dcn351.dpm_clks = (DpmClocks_t_dcn351 *)dm_helpers_allocate_gpu_mem(
+				clk_mgr->base.base.ctx,
+				DC_MEM_ALLOC_TYPE_GART,
+				sizeof(DpmClocks_t_dcn351),
+				&smu_dpm_clks_dcn351.mc_address.quad_part);
+		if (smu_dpm_clks_dcn351.dpm_clks == NULL) {
+			smu_dpm_clks_dcn351.dpm_clks = &dummy_clocks_dcn351;
+			smu_dpm_clks_dcn351.mc_address.quad_part = 0;
+		}
+	}
+
+	clk_mgr->base.smu_ver = dcn35_smu_get_smu_version(&clk_mgr->base);
+
+	if (clk_mgr->base.smu_ver)
+		clk_mgr->base.smu_present = true;
+
+	/* TODO: Check we get what we expect during bringup */
+	clk_mgr->base.base.dentist_vco_freq_khz = get_vco_frequency_from_reg(&clk_mgr->base);
+
+	if (ctx->dc_bios->integrated_info->memory_type == LpDdr5MemType) {
+		dcn35_bw_params.wm_table = lpddr5_wm_table;
+	} else {
+		dcn35_bw_params.wm_table = ddr5_wm_table;
+	}
+	/* Saved clocks configured at boot for debug purposes */
+	dcn35_save_clk_registers(&clk_mgr->base.base.boot_snapshot, clk_mgr);
+
+	clk_mgr->base.base.dprefclk_khz = dcn35_smu_get_dprefclk(&clk_mgr->base);
+	clk_mgr->base.base.clks.ref_dtbclk_khz = 600000;
+
+	dce_clock_read_ss_info(&clk_mgr->base);
+	/*when clk src is from FCH, it could have ss, same clock src as DPREF clk*/
+
+	dcn35_read_ss_info_from_lut(&clk_mgr->base);
+
+	clk_mgr->base.base.bw_params = &dcn35_bw_params;
+
+	if (clk_mgr->base.base.ctx->dc->debug.pstate_enabled) {
+		int i;
+		if (ctx->dce_version == DCN_VERSION_3_51) {
+			dcn351_get_dpm_table_from_smu(&clk_mgr->base, &smu_dpm_clks_dcn351);
+			translate_to_DpmClocks_t_dcn35(&smu_dpm_clks_dcn351, &smu_dpm_clks);
+		} else
+			dcn35_get_dpm_table_from_smu(&clk_mgr->base, &smu_dpm_clks);
+		DC_LOG_SMU("NumDcfClkLevelsEnabled: %d\n"
+				   "NumDispClkLevelsEnabled: %d\n"
+				   "NumSocClkLevelsEnabled: %d\n"
+				   "VcnClkLevelsEnabled: %d\n"
+				   "FClkLevelsEnabled: %d\n"
+				   "NumMemPstatesEnabled: %d\n"
+				   "MinGfxClk: %d\n"
+				   "MaxGfxClk: %d\n",
+				   smu_dpm_clks.dpm_clks->NumDcfClkLevelsEnabled,
+				   smu_dpm_clks.dpm_clks->NumDispClkLevelsEnabled,
+				   smu_dpm_clks.dpm_clks->NumSocClkLevelsEnabled,
+				   smu_dpm_clks.dpm_clks->VcnClkLevelsEnabled,
+				   smu_dpm_clks.dpm_clks->NumFclkLevelsEnabled,
+				   smu_dpm_clks.dpm_clks->NumMemPstatesEnabled,
+				   smu_dpm_clks.dpm_clks->MinGfxClk,
+				   smu_dpm_clks.dpm_clks->MaxGfxClk);
+		for (i = 0; i < smu_dpm_clks.dpm_clks->NumDcfClkLevelsEnabled; i++) {
+			DC_LOG_SMU("smu_dpm_clks.dpm_clks->DcfClocks[%d] = %d\n",
+					   i,
+					   smu_dpm_clks.dpm_clks->DcfClocks[i]);
+		}
+		for (i = 0; i < smu_dpm_clks.dpm_clks->NumDispClkLevelsEnabled; i++) {
+			DC_LOG_SMU("smu_dpm_clks.dpm_clks->DispClocks[%d] = %d\n",
+					   i, smu_dpm_clks.dpm_clks->DispClocks[i]);
+		}
+		for (i = 0; i < smu_dpm_clks.dpm_clks->NumSocClkLevelsEnabled; i++) {
+			DC_LOG_SMU("smu_dpm_clks.dpm_clks->SocClocks[%d] = %d\n",
+					   i, smu_dpm_clks.dpm_clks->SocClocks[i]);
+		}
+		for (i = 0; i < smu_dpm_clks.dpm_clks->NumFclkLevelsEnabled; i++) {
+			DC_LOG_SMU("smu_dpm_clks.dpm_clks->FclkClocks_Freq[%d] = %d\n",
+					   i, smu_dpm_clks.dpm_clks->FclkClocks_Freq[i]);
+			DC_LOG_SMU("smu_dpm_clks.dpm_clks->FclkClocks_Voltage[%d] = %d\n",
+					   i, smu_dpm_clks.dpm_clks->FclkClocks_Voltage[i]);
+		}
+		for (i = 0; i < smu_dpm_clks.dpm_clks->NumSocClkLevelsEnabled; i++)
+			DC_LOG_SMU("smu_dpm_clks.dpm_clks->SocVoltage[%d] = %d\n",
+					   i, smu_dpm_clks.dpm_clks->SocVoltage[i]);
+
+		for (i = 0; i < smu_dpm_clks.dpm_clks->NumMemPstatesEnabled; i++) {
+			DC_LOG_SMU("smu_dpm_clks.dpm_clks.MemPstateTable[%d].UClk = %d\n"
+					   "smu_dpm_clks.dpm_clks->MemPstateTable[%d].MemClk= %d\n"
+					   "smu_dpm_clks.dpm_clks->MemPstateTable[%d].Voltage = %d\n",
+					   i, smu_dpm_clks.dpm_clks->MemPstateTable[i].UClk,
+					   i, smu_dpm_clks.dpm_clks->MemPstateTable[i].MemClk,
+					   i, smu_dpm_clks.dpm_clks->MemPstateTable[i].Voltage);
+		}
+
+		if (ctx->dc_bios->integrated_info && ctx->dc->config.use_default_clock_table == false) {
+			dcn35_clk_mgr_helper_populate_bw_params(
+					&clk_mgr->base,
+					ctx->dc_bios->integrated_info,
+					smu_dpm_clks.dpm_clks);
+		}
+	}
+
+	if (smu_dpm_clks.dpm_clks && smu_dpm_clks.mc_address.quad_part != 0)
+		dm_helpers_free_gpu_mem(clk_mgr->base.base.ctx, DC_MEM_ALLOC_TYPE_GART,
+				smu_dpm_clks.dpm_clks);
+
+	if (smu_dpm_clks_dcn351.dpm_clks && smu_dpm_clks_dcn351.mc_address.quad_part != 0)
+		dm_helpers_free_gpu_mem(clk_mgr->base.base.ctx, DC_MEM_ALLOC_TYPE_GART,
+				smu_dpm_clks_dcn351.dpm_clks);
+
+	if (ctx->dc->config.disable_ips != DMUB_IPS_DISABLE_ALL) {
+		bool ips_support = false;
+
+		/*avoid call pmfw at init*/
+		ips_support = dcn35_smu_get_ips_supported(&clk_mgr->base);
+		if (ips_support) {
+			ctx->dc->debug.ignore_pg = false;
+			ctx->dc->debug.disable_dpp_power_gate = false;
+			ctx->dc->debug.disable_hubp_power_gate = false;
+			ctx->dc->debug.disable_dsc_power_gate = false;
+
+			/* Disable dynamic IPS2 in older PMFW (93.12) for Z8 interop. */
+			if (ctx->dc->config.disable_ips == DMUB_IPS_ENABLE &&
+			    ctx->dce_version != DCN_VERSION_3_51 &&
+			    ((clk_mgr->base.smu_ver & 0x00FFFFFF) <= 0x005d0c00))
+				ctx->dc->config.disable_ips = DMUB_IPS_RCG_IN_ACTIVE_IPS2_IN_OFF;
+		} else {
+			/*let's reset the config control flag*/
+			ctx->dc->config.disable_ips = DMUB_IPS_DISABLE_ALL; /*pmfw not support it, disable it all*/
+		}
+	}
+}
+
+void dcn35_clk_mgr_destroy(struct clk_mgr_internal *clk_mgr_int)
+{
+	struct clk_mgr_dcn35 *clk_mgr = TO_CLK_MGR_DCN35(clk_mgr_int);
+
+	if (clk_mgr->smu_wm_set.wm_set && clk_mgr->smu_wm_set.mc_address.quad_part != 0)
+		dm_helpers_free_gpu_mem(clk_mgr_int->base.ctx, DC_MEM_ALLOC_TYPE_FRAME_BUFFER,
+				clk_mgr->smu_wm_set.wm_set);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn35_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn35_clk_mgr.h
new file mode 100644
index 000000000000..a12a9bf90806
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn35_clk_mgr.h
@@ -0,0 +1,67 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCN35_CLK_MGR_H__
+#define __DCN35_CLK_MGR_H__
+#include "clk_mgr_internal.h"
+
+#define NUM_CLOCK_SOURCES   5
+
+struct dcn35_watermarks;
+
+struct dcn35_smu_watermark_set {
+	struct dcn35_watermarks *wm_set;
+	union large_integer mc_address;
+};
+
+struct dcn35_ss_info_table {
+	uint32_t ss_divider;
+	uint32_t ss_percentage[NUM_CLOCK_SOURCES];
+};
+
+struct clk_mgr_dcn35 {
+	struct clk_mgr_internal base;
+	struct dcn35_smu_watermark_set smu_wm_set;
+};
+
+bool dcn35_are_clock_states_equal(struct dc_clocks *a,
+		struct dc_clocks *b);
+void dcn35_init_clocks(struct clk_mgr *clk_mgr);
+void dcn35_update_clocks(struct clk_mgr *clk_mgr_base,
+			struct dc_state *context,
+			bool safe_to_lower);
+
+void dcn35_clk_mgr_construct(struct dc_context *ctx,
+		struct clk_mgr_dcn35 *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg);
+
+void dcn35_clk_mgr_destroy(struct clk_mgr_internal *clk_mgr_int);
+
+void dcn351_clk_mgr_construct(struct dc_context *ctx,
+		struct clk_mgr_dcn35 *clk_mgr,
+		struct pp_smu_funcs *pp_smu,
+		struct dccg *dccg);
+#endif //__DCN35_CLK_MGR_H__
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn35_smu.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn35_smu.c
new file mode 100644
index 000000000000..604d256cb47a
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn35_smu.c
@@ -0,0 +1,508 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+
+#include "core_types.h"
+#include "clk_mgr_internal.h"
+#include "reg_helper.h"
+#include "dm_helpers.h"
+#include "dcn35_smu.h"
+
+#include "mp/mp_14_0_0_offset.h"
+#include "mp/mp_14_0_0_sh_mask.h"
+
+/* TODO: Use the real headers when they're correct */
+#define MP1_BASE__INST0_SEG0                       0x00016000
+#define MP1_BASE__INST0_SEG1                       0x0243FC00
+#define MP1_BASE__INST0_SEG2                       0x00DC0000
+#define MP1_BASE__INST0_SEG3                       0x00E00000
+#define MP1_BASE__INST0_SEG4                       0x00E40000
+#define MP1_BASE__INST0_SEG5                       0
+
+#ifdef BASE_INNER
+#undef BASE_INNER
+#endif
+
+#define BASE_INNER(seg) MP1_BASE__INST0_SEG ## seg
+
+#define BASE(seg) BASE_INNER(seg)
+
+#define REG(reg_name) (BASE(reg##reg_name##_BASE_IDX) + reg##reg_name)
+
+#define FN(reg_name, field) \
+	FD(reg_name##__##field)
+
+#include "logger_types.h"
+#undef DC_LOGGER
+#define DC_LOGGER \
+	CTX->logger
+#define smu_print(str, ...) {DC_LOG_SMU(str, ##__VA_ARGS__); }
+
+#define VBIOSSMC_MSG_TestMessage                  0x1
+#define VBIOSSMC_MSG_GetSmuVersion                0x2
+#define VBIOSSMC_MSG_PowerUpGfx                   0x3
+#define VBIOSSMC_MSG_SetDispclkFreq               0x4
+#define VBIOSSMC_MSG_SetDprefclkFreq              0x5   //Not used. DPRef is constant
+#define VBIOSSMC_MSG_SetDppclkFreq                0x6
+#define VBIOSSMC_MSG_SetHardMinDcfclkByFreq       0x7
+#define VBIOSSMC_MSG_SetMinDeepSleepDcfclk        0x8
+#define VBIOSSMC_MSG_SetPhyclkVoltageByFreq       0x9	//Keep it in case VMIN dees not support phy clk
+#define VBIOSSMC_MSG_GetFclkFrequency             0xA
+#define VBIOSSMC_MSG_SetDisplayCount              0xB   //Not used anymore
+#define VBIOSSMC_MSG_EnableTmdp48MHzRefclkPwrDown 0xC   //To ask PMFW turn off TMDP 48MHz refclk during display off to save power
+#define VBIOSSMC_MSG_UpdatePmeRestore             0xD
+#define VBIOSSMC_MSG_SetVbiosDramAddrHigh         0xE   //Used for WM table txfr
+#define VBIOSSMC_MSG_SetVbiosDramAddrLow          0xF
+#define VBIOSSMC_MSG_TransferTableSmu2Dram        0x10
+#define VBIOSSMC_MSG_TransferTableDram2Smu        0x11
+#define VBIOSSMC_MSG_SetDisplayIdleOptimizations  0x12
+#define VBIOSSMC_MSG_GetDprefclkFreq              0x13
+#define VBIOSSMC_MSG_GetDtbclkFreq                0x14
+#define VBIOSSMC_MSG_AllowZstatesEntry            0x15
+#define VBIOSSMC_MSG_DisallowZstatesEntry     	  0x16
+#define VBIOSSMC_MSG_SetDtbClk                    0x17
+#define VBIOSSMC_MSG_DispIPS2Entry                0x18 ///< Display IPS2 entry, DMU
+#define VBIOSSMC_MSG_DispIPS2Exit                 0x19 ///< Display IPS2 exit, DMU
+#define VBIOSSMC_MSG_DisableLSdma                 0x1A ///< Disable LSDMA; only sent by VBIOS
+#define VBIOSSMC_MSG_DpControllerPhyStatus        0x1B ///< Inform PMFW about the pre conditions for turning SLDO2 on/off . bit[0]==1 precondition is met, bit[1-2] are for DPPHY number
+#define VBIOSSMC_MSG_QueryIPS2Support             0x1C ///< Return 1: support; else not supported
+#define VBIOSSMC_MSG_NotifyHostRouterBW           0x1D
+#define VBIOSSMC_Message_Count                    0x1E
+
+#define VBIOSSMC_Status_BUSY                      0x0
+#define VBIOSSMC_Result_OK                        0x1
+#define VBIOSSMC_Result_Failed                    0xFF
+#define VBIOSSMC_Result_UnknownCmd                0xFE
+#define VBIOSSMC_Result_CmdRejectedPrereq         0xFD
+#define VBIOSSMC_Result_CmdRejectedBusy           0xFC
+
+union dcn35_dpia_host_router_bw {
+	struct {
+		uint32_t hr_id : 16;
+		uint32_t bw_mbps : 16;
+	} bits;
+	uint32_t all;
+};
+
+/*
+ * Function to be used instead of REG_WAIT macro because the wait ends when
+ * the register is NOT EQUAL to zero, and because `the translation in msg_if.h
+ * won't work with REG_WAIT.
+ */
+static uint32_t dcn35_smu_wait_for_response(struct clk_mgr_internal *clk_mgr, unsigned int delay_us, unsigned int max_retries)
+{
+	uint32_t res_val = VBIOSSMC_Status_BUSY;
+
+	do {
+		res_val = REG_READ(MP1_SMN_C2PMSG_91);
+		if (res_val != VBIOSSMC_Status_BUSY)
+			break;
+
+		if (delay_us >= 1000)
+			msleep(delay_us/1000);
+		else if (delay_us > 0)
+			udelay(delay_us);
+
+		if (clk_mgr->base.ctx->dc->debug.disable_timeout)
+			max_retries++;
+	} while (max_retries--);
+
+	return res_val;
+}
+
+static int dcn35_smu_send_msg_with_param(struct clk_mgr_internal *clk_mgr,
+					 unsigned int msg_id,
+					 unsigned int param)
+{
+	uint32_t result;
+
+	result = dcn35_smu_wait_for_response(clk_mgr, 10, 2000000);
+	ASSERT(result == VBIOSSMC_Result_OK);
+
+	if (result != VBIOSSMC_Result_OK) {
+		DC_LOG_WARNING("SMU response after wait: %d, msg id = %d\n", result, msg_id);
+
+		if (result == VBIOSSMC_Status_BUSY)
+			return -1;
+	}
+
+	/* First clear response register */
+	REG_WRITE(MP1_SMN_C2PMSG_91, VBIOSSMC_Status_BUSY);
+
+	/* Set the parameter register for the SMU message, unit is Mhz */
+	REG_WRITE(MP1_SMN_C2PMSG_83, param);
+
+	/* Trigger the message transaction by writing the message ID */
+	REG_WRITE(MP1_SMN_C2PMSG_67, msg_id);
+
+	result = dcn35_smu_wait_for_response(clk_mgr, 10, 2000000);
+
+	if (result == VBIOSSMC_Result_Failed) {
+		if (msg_id == VBIOSSMC_MSG_TransferTableDram2Smu &&
+		    param == TABLE_WATERMARKS)
+			DC_LOG_WARNING("Watermarks table not configured properly by SMU");
+		else
+			ASSERT(0);
+		REG_WRITE(MP1_SMN_C2PMSG_91, VBIOSSMC_Result_OK);
+		DC_LOG_WARNING("SMU response after wait: %d, msg id = %d\n", result, msg_id);
+		return -1;
+	}
+
+	if (IS_SMU_TIMEOUT(result)) {
+		ASSERT(0);
+		result = dcn35_smu_wait_for_response(clk_mgr, 10, 2000000);
+		//dm_helpers_smu_timeout(CTX, msg_id, param, 10 * 200000);
+		DC_LOG_WARNING("SMU response after wait: %d, msg id = %d\n", result, msg_id);
+	}
+
+	return REG_READ(MP1_SMN_C2PMSG_83);
+}
+
+int dcn35_smu_get_smu_version(struct clk_mgr_internal *clk_mgr)
+{
+	return dcn35_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_GetSmuVersion,
+			0);
+}
+
+
+int dcn35_smu_set_dispclk(struct clk_mgr_internal *clk_mgr, int requested_dispclk_khz)
+{
+	int actual_dispclk_set_mhz = -1;
+
+	if (!clk_mgr->smu_present)
+		return requested_dispclk_khz;
+
+	/*  Unit of SMU msg parameter is Mhz */
+	actual_dispclk_set_mhz = dcn35_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDispclkFreq,
+			khz_to_mhz_ceil(requested_dispclk_khz));
+
+	smu_print("requested_dispclk_khz = %d, actual_dispclk_set_mhz: %d\n", requested_dispclk_khz, actual_dispclk_set_mhz);
+	return actual_dispclk_set_mhz * 1000;
+}
+
+int dcn35_smu_set_dprefclk(struct clk_mgr_internal *clk_mgr)
+{
+	int actual_dprefclk_set_mhz = -1;
+
+	if (!clk_mgr->smu_present)
+		return clk_mgr->base.dprefclk_khz;
+
+	actual_dprefclk_set_mhz = dcn35_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDprefclkFreq,
+			khz_to_mhz_ceil(clk_mgr->base.dprefclk_khz));
+
+	/* TODO: add code for programing DP DTO, currently this is down by command table */
+
+	return actual_dprefclk_set_mhz * 1000;
+}
+
+int dcn35_smu_set_hard_min_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_dcfclk_khz)
+{
+	int actual_dcfclk_set_mhz = -1;
+
+	if (!clk_mgr->smu_present)
+		return requested_dcfclk_khz;
+
+	actual_dcfclk_set_mhz = dcn35_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetHardMinDcfclkByFreq,
+			khz_to_mhz_ceil(requested_dcfclk_khz));
+
+	smu_print("requested_dcfclk_khz = %d, actual_dcfclk_set_mhz: %d\n", requested_dcfclk_khz, actual_dcfclk_set_mhz);
+
+	return actual_dcfclk_set_mhz * 1000;
+}
+
+int dcn35_smu_set_min_deep_sleep_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_min_ds_dcfclk_khz)
+{
+	int actual_min_ds_dcfclk_mhz = -1;
+
+	if (!clk_mgr->smu_present)
+		return requested_min_ds_dcfclk_khz;
+
+	actual_min_ds_dcfclk_mhz = dcn35_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetMinDeepSleepDcfclk,
+			khz_to_mhz_ceil(requested_min_ds_dcfclk_khz));
+
+	smu_print("requested_min_ds_dcfclk_khz = %d, actual_min_ds_dcfclk_mhz: %d\n", requested_min_ds_dcfclk_khz, actual_min_ds_dcfclk_mhz);
+
+	return actual_min_ds_dcfclk_mhz * 1000;
+}
+
+int dcn35_smu_set_dppclk(struct clk_mgr_internal *clk_mgr, int requested_dpp_khz)
+{
+	int actual_dppclk_set_mhz = -1;
+
+	if (!clk_mgr->smu_present)
+		return requested_dpp_khz;
+
+	actual_dppclk_set_mhz = dcn35_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDppclkFreq,
+			khz_to_mhz_ceil(requested_dpp_khz));
+
+	smu_print("requested_dpp_khz = %d, actual_dppclk_set_mhz: %d\n", requested_dpp_khz, actual_dppclk_set_mhz);
+
+	return actual_dppclk_set_mhz * 1000;
+}
+
+void dcn35_smu_set_display_idle_optimization(struct clk_mgr_internal *clk_mgr, uint32_t idle_info)
+{
+	if (!clk_mgr->base.ctx->dc->debug.pstate_enabled)
+		return;
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	//TODO: Work with smu team to define optimization options.
+	dcn35_smu_send_msg_with_param(
+		clk_mgr,
+		VBIOSSMC_MSG_SetDisplayIdleOptimizations,
+		idle_info);
+	smu_print("%s: VBIOSSMC_MSG_SetDisplayIdleOptimizations idle_info  = %x\n", __func__, idle_info);
+}
+
+void dcn35_smu_enable_phy_refclk_pwrdwn(struct clk_mgr_internal *clk_mgr, bool enable)
+{
+	union display_idle_optimization_u idle_info = { 0 };
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	if (enable) {
+		idle_info.idle_info.df_request_disabled = 1;
+		idle_info.idle_info.phy_ref_clk_off = 1;
+	}
+
+	dcn35_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDisplayIdleOptimizations,
+			idle_info.data);
+	smu_print("%s smu_enable_phy_refclk_pwrdwn  = %d\n", __func__, enable ? 1 : 0);
+}
+
+void dcn35_smu_enable_pme_wa(struct clk_mgr_internal *clk_mgr)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn35_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_UpdatePmeRestore,
+			0);
+	smu_print("%s: SMC_MSG_UpdatePmeRestore\n", __func__);
+}
+
+void dcn35_smu_set_dram_addr_high(struct clk_mgr_internal *clk_mgr, uint32_t addr_high)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn35_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_SetVbiosDramAddrHigh, addr_high);
+}
+
+void dcn35_smu_set_dram_addr_low(struct clk_mgr_internal *clk_mgr, uint32_t addr_low)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn35_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_SetVbiosDramAddrLow, addr_low);
+}
+
+void dcn35_smu_transfer_dpm_table_smu_2_dram(struct clk_mgr_internal *clk_mgr)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn35_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_TransferTableSmu2Dram, TABLE_DPMCLOCKS);
+}
+
+void dcn35_smu_transfer_wm_table_dram_2_smu(struct clk_mgr_internal *clk_mgr)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn35_smu_send_msg_with_param(clk_mgr,
+			VBIOSSMC_MSG_TransferTableDram2Smu, TABLE_WATERMARKS);
+}
+
+void dcn35_smu_set_zstate_support(struct clk_mgr_internal *clk_mgr, enum dcn_zstate_support_state support)
+{
+	unsigned int msg_id, param, retv;
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	switch (support) {
+
+	case DCN_ZSTATE_SUPPORT_ALLOW:
+		msg_id = VBIOSSMC_MSG_AllowZstatesEntry;
+		param = (1 << 10) | (1 << 9) | (1 << 8);
+		smu_print("%s: SMC_MSG_AllowZstatesEntry msg = ALLOW, param = 0x%x\n", __func__, param);
+		break;
+
+	case DCN_ZSTATE_SUPPORT_DISALLOW:
+		msg_id = VBIOSSMC_MSG_AllowZstatesEntry;
+		param = 0;
+		smu_print("%s: SMC_MSG_AllowZstatesEntry msg_id = DISALLOW, param = 0x%x\n",  __func__, param);
+		break;
+
+
+	case DCN_ZSTATE_SUPPORT_ALLOW_Z10_ONLY:
+		msg_id = VBIOSSMC_MSG_AllowZstatesEntry;
+		param = (1 << 10);
+		smu_print("%s: SMC_MSG_AllowZstatesEntry msg = ALLOW_Z10_ONLY, param = 0x%x\n", __func__, param);
+		break;
+
+	case DCN_ZSTATE_SUPPORT_ALLOW_Z8_Z10_ONLY:
+		msg_id = VBIOSSMC_MSG_AllowZstatesEntry;
+		param = (1 << 10) | (1 << 8);
+		smu_print("%s: SMC_MSG_AllowZstatesEntry msg = ALLOW_Z8_Z10_ONLY, param = 0x%x\n", __func__, param);
+		break;
+
+	case DCN_ZSTATE_SUPPORT_ALLOW_Z8_ONLY:
+		msg_id = VBIOSSMC_MSG_AllowZstatesEntry;
+		param = (1 << 8);
+		smu_print("%s: SMC_MSG_AllowZstatesEntry msg = ALLOW_Z8_ONLY, param = 0x%x\n", __func__, param);
+		break;
+
+	default: //DCN_ZSTATE_SUPPORT_UNKNOWN
+		msg_id = VBIOSSMC_MSG_AllowZstatesEntry;
+		param = 0;
+		break;
+	}
+
+
+	retv = dcn35_smu_send_msg_with_param(
+		clk_mgr,
+		msg_id,
+		param);
+	smu_print("%s:  msg_id = %d, param = 0x%x, return = 0x%x\n", __func__, msg_id, param, retv);
+}
+
+int dcn35_smu_get_dprefclk(struct clk_mgr_internal *clk_mgr)
+{
+	int dprefclk;
+
+	if (!clk_mgr->smu_present)
+		return 0;
+
+	dprefclk = dcn35_smu_send_msg_with_param(clk_mgr,
+						 VBIOSSMC_MSG_GetDprefclkFreq,
+						 0);
+
+	smu_print("%s:  SMU DPREF clk  = %d mhz\n",  __func__, dprefclk);
+	return dprefclk * 1000;
+}
+
+int dcn35_smu_get_dtbclk(struct clk_mgr_internal *clk_mgr)
+{
+	int dtbclk;
+
+	if (!clk_mgr->smu_present)
+		return 0;
+
+	dtbclk = dcn35_smu_send_msg_with_param(clk_mgr,
+					       VBIOSSMC_MSG_GetDtbclkFreq,
+					       0);
+
+	smu_print("%s: get_dtbclk  = %dmhz\n", __func__, dtbclk);
+	return dtbclk * 1000;
+}
+/* Arg = 1: Turn DTB on; 0: Turn DTB CLK OFF. when it is on, it is 600MHZ */
+void dcn35_smu_set_dtbclk(struct clk_mgr_internal *clk_mgr, bool enable)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn35_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_SetDtbClk,
+			enable);
+	smu_print("%s: smu_set_dtbclk = %d\n", __func__, enable ? 1 : 0);
+}
+
+void dcn35_vbios_smu_enable_48mhz_tmdp_refclk_pwrdwn(struct clk_mgr_internal *clk_mgr, bool enable)
+{
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn35_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_EnableTmdp48MHzRefclkPwrDown,
+			enable);
+	smu_print("%s: smu_enable_48mhz_tmdp_refclk_pwrdwn = %d\n", __func__, enable ? 1 : 0);
+}
+
+int dcn35_smu_exit_low_power_state(struct clk_mgr_internal *clk_mgr)
+{
+	int retv;
+
+	if (!clk_mgr->smu_present)
+		return 0;
+
+	retv = dcn35_smu_send_msg_with_param(
+		clk_mgr,
+		VBIOSSMC_MSG_DispIPS2Exit,
+		0);
+	smu_print("%s: smu_exit_low_power_state return = %d\n", __func__, retv);
+	return retv;
+}
+
+int dcn35_smu_get_ips_supported(struct clk_mgr_internal *clk_mgr)
+{
+	int retv;
+
+	if (!clk_mgr->smu_present)
+		return 0;
+
+	retv = dcn35_smu_send_msg_with_param(
+			clk_mgr,
+			VBIOSSMC_MSG_QueryIPS2Support,
+			0);
+
+	//smu_print("%s: VBIOSSMC_MSG_QueryIPS2Support return = %x\n", __func__, retv);
+	return retv;
+}
+
+void dcn35_smu_notify_host_router_bw(struct clk_mgr_internal *clk_mgr, uint32_t hr_id, uint32_t bw_kbps)
+{
+	union dcn35_dpia_host_router_bw msg_data = { 0 };
+
+	msg_data.bits.hr_id = hr_id;
+	msg_data.bits.bw_mbps = bw_kbps / 1000;
+
+	dcn35_smu_send_msg_with_param(clk_mgr, VBIOSSMC_MSG_NotifyHostRouterBW, msg_data.all);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn35_smu.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn35_smu.h
new file mode 100644
index 000000000000..ab9d21ba0c43
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn35_smu.h
@@ -0,0 +1,220 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DAL_DC_35_SMU_H_
+#define DAL_DC_35_SMU_H_
+
+#include "os_types.h"
+
+#ifndef PMFW_DRIVER_IF_H
+#define PMFW_DRIVER_IF_H
+#define PMFW_DRIVER_IF_VERSION 4
+
+typedef enum {
+  DSPCLK_DCFCLK = 0,
+  DSPCLK_DISPCLK,
+  DSPCLK_PIXCLK,
+  DSPCLK_PHYCLK,
+  DSPCLK_COUNT,
+} DSPCLK_e;
+
+typedef struct {
+  uint16_t Freq; // in MHz
+  uint16_t Vid;  // min voltage in SVI3 VID
+} DisplayClockTable_t;
+
+typedef struct {
+  uint16_t MinClock; // This is either DCFCLK or SOCCLK (in MHz)
+  uint16_t MaxClock; // This is either DCFCLK or SOCCLK (in MHz)
+  uint16_t MinMclk;
+  uint16_t MaxMclk;
+
+  uint8_t  WmSetting;
+  uint8_t  WmType;  // Used for normal pstate change or memory retraining
+  uint8_t  Padding[2];
+} WatermarkRowGeneric_t;
+
+#define NUM_WM_RANGES 4
+#define WM_PSTATE_CHG 0
+#define WM_RETRAINING 1
+
+typedef enum {
+  WM_SOCCLK = 0,
+  WM_DCFCLK,
+  WM_COUNT,
+} WM_CLOCK_e;
+
+typedef struct {
+  // Watermarks
+  WatermarkRowGeneric_t WatermarkRow[WM_COUNT][NUM_WM_RANGES];
+
+  uint32_t MmHubPadding[7]; // SMU internal use
+} Watermarks_t;
+
+#define NUM_DCFCLK_DPM_LEVELS   8
+#define NUM_DISPCLK_DPM_LEVELS  8
+#define NUM_DPPCLK_DPM_LEVELS   8
+#define NUM_SOCCLK_DPM_LEVELS   8
+#define NUM_VCN_DPM_LEVELS      8
+#define NUM_SOC_VOLTAGE_LEVELS  8
+#define NUM_VPE_DPM_LEVELS        8
+#define NUM_FCLK_DPM_LEVELS       8
+#define NUM_MEM_PSTATE_LEVELS     4
+
+typedef enum{
+  WCK_RATIO_1_1 = 0,  // DDR5, Wck:ck is always 1:1;
+  WCK_RATIO_1_2,
+  WCK_RATIO_1_4,
+  WCK_RATIO_MAX
+} WCK_RATIO_e;
+
+typedef struct {
+  uint32_t UClk;
+  uint32_t MemClk;
+  uint32_t Voltage;
+  uint8_t  WckRatio;
+  uint8_t  Spare[3];
+} MemPstateTable_t;
+
+//Freq in MHz
+//Voltage in milli volts with 2 fractional bits
+typedef struct {
+  uint32_t DcfClocks[NUM_DCFCLK_DPM_LEVELS];
+  uint32_t DispClocks[NUM_DISPCLK_DPM_LEVELS];
+  uint32_t DppClocks[NUM_DPPCLK_DPM_LEVELS];
+  uint32_t SocClocks[NUM_SOCCLK_DPM_LEVELS];
+  uint32_t VClocks[NUM_VCN_DPM_LEVELS];
+  uint32_t DClocks[NUM_VCN_DPM_LEVELS];
+  uint32_t VPEClocks[NUM_VPE_DPM_LEVELS];
+  uint32_t FclkClocks_Freq[NUM_FCLK_DPM_LEVELS];
+  uint32_t FclkClocks_Voltage[NUM_FCLK_DPM_LEVELS];
+  uint32_t SocVoltage[NUM_SOC_VOLTAGE_LEVELS];
+  MemPstateTable_t MemPstateTable[NUM_MEM_PSTATE_LEVELS];
+
+  uint8_t  NumDcfClkLevelsEnabled;
+  uint8_t  NumDispClkLevelsEnabled; //Applies to both Dispclk and Dppclk
+  uint8_t  NumSocClkLevelsEnabled;
+  uint8_t  VcnClkLevelsEnabled;     //Applies to both Vclk and Dclk
+  uint8_t  VpeClkLevelsEnabled;
+  uint8_t  NumMemPstatesEnabled;
+  uint8_t  NumFclkLevelsEnabled;
+  uint8_t  spare[2];
+
+  uint32_t MinGfxClk;
+  uint32_t MaxGfxClk;
+} DpmClocks_t_dcn35;
+
+typedef struct {
+	uint32_t DcfClocks[NUM_DCFCLK_DPM_LEVELS];
+	uint32_t DispClocks[NUM_DISPCLK_DPM_LEVELS];
+	uint32_t DppClocks[NUM_DPPCLK_DPM_LEVELS];
+	uint32_t SocClocks[NUM_SOCCLK_DPM_LEVELS];
+	uint32_t VClocks0[NUM_VCN_DPM_LEVELS];
+	uint32_t VClocks1[NUM_VCN_DPM_LEVELS];
+	uint32_t DClocks0[NUM_VCN_DPM_LEVELS];
+	uint32_t DClocks1[NUM_VCN_DPM_LEVELS];
+	uint32_t VPEClocks[NUM_VPE_DPM_LEVELS];
+	uint32_t FclkClocks_Freq[NUM_FCLK_DPM_LEVELS];
+	uint32_t FclkClocks_Voltage[NUM_FCLK_DPM_LEVELS];
+	uint32_t SocVoltage[NUM_SOC_VOLTAGE_LEVELS];
+	MemPstateTable_t MemPstateTable[NUM_MEM_PSTATE_LEVELS];
+	uint8_t NumDcfClkLevelsEnabled;
+	uint8_t NumDispClkLevelsEnabled; // Applies to both Dispclk and Dppclk
+	uint8_t NumSocClkLevelsEnabled;
+	uint8_t Vcn0ClkLevelsEnabled; // Applies to both Vclk0 and Dclk0
+	uint8_t Vcn1ClkLevelsEnabled; // Applies to both Vclk1 and Dclk1
+	uint8_t VpeClkLevelsEnabled;
+	uint8_t NumMemPstatesEnabled;
+	uint8_t NumFclkLevelsEnabled;
+	uint32_t MinGfxClk;
+	uint32_t MaxGfxClk;
+} DpmClocks_t_dcn351;
+
+#define TABLE_BIOS_IF            0 // Called by BIOS
+#define TABLE_WATERMARKS         1 // Called by DAL through VBIOS
+#define TABLE_CUSTOM_DPM         2 // Called by Driver
+#define TABLE_SPARE1             3
+#define TABLE_DPMCLOCKS          4 // Called by Driver
+#define TABLE_MOMENTARY_PM       5 // Called by Tools
+#define TABLE_MODERN_STDBY       6 // Called by Tools for Modern Standby Log
+#define TABLE_SMU_METRICS        7 // Called by Driver
+#define TABLE_COUNT              8
+
+#endif
+
+struct dcn35_watermarks {
+  // Watermarks
+  WatermarkRowGeneric_t WatermarkRow[WM_COUNT][NUM_WM_RANGES];
+
+  uint32_t MmHubPadding[7]; // SMU internal use
+};
+
+struct dcn35_smu_dpm_clks {
+	DpmClocks_t_dcn35 *dpm_clks;
+	union large_integer mc_address;
+};
+
+struct dcn351_smu_dpm_clks {
+	DpmClocks_t_dcn351 *dpm_clks;
+	union large_integer mc_address;
+};
+/* TODO: taken from vgh, may not be correct */
+struct display_idle_optimization {
+	unsigned int df_request_disabled : 1;
+	unsigned int phy_ref_clk_off     : 1;
+	unsigned int s0i2_rdy            : 1;
+	unsigned int reserved            : 29;
+};
+
+union display_idle_optimization_u {
+	struct display_idle_optimization idle_info;
+	uint32_t data;
+};
+
+int dcn35_smu_get_smu_version(struct clk_mgr_internal *clk_mgr);
+int dcn35_smu_set_dispclk(struct clk_mgr_internal *clk_mgr, int requested_dispclk_khz);
+int dcn35_smu_set_dprefclk(struct clk_mgr_internal *clk_mgr);
+int dcn35_smu_set_hard_min_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_dcfclk_khz);
+int dcn35_smu_set_min_deep_sleep_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_min_ds_dcfclk_khz);
+int dcn35_smu_set_dppclk(struct clk_mgr_internal *clk_mgr, int requested_dpp_khz);
+void dcn35_smu_set_display_idle_optimization(struct clk_mgr_internal *clk_mgr, uint32_t idle_info);
+void dcn35_smu_enable_phy_refclk_pwrdwn(struct clk_mgr_internal *clk_mgr, bool enable);
+void dcn35_smu_enable_pme_wa(struct clk_mgr_internal *clk_mgr);
+void dcn35_smu_set_dram_addr_high(struct clk_mgr_internal *clk_mgr, uint32_t addr_high);
+void dcn35_smu_set_dram_addr_low(struct clk_mgr_internal *clk_mgr, uint32_t addr_low);
+void dcn35_smu_transfer_dpm_table_smu_2_dram(struct clk_mgr_internal *clk_mgr);
+void dcn35_smu_transfer_wm_table_dram_2_smu(struct clk_mgr_internal *clk_mgr);
+
+void dcn35_smu_set_zstate_support(struct clk_mgr_internal *clk_mgr, enum dcn_zstate_support_state support);
+void dcn35_smu_set_dtbclk(struct clk_mgr_internal *clk_mgr, bool enable);
+void dcn35_vbios_smu_enable_48mhz_tmdp_refclk_pwrdwn(struct clk_mgr_internal *clk_mgr, bool enable);
+
+int dcn35_smu_exit_low_power_state(struct clk_mgr_internal *clk_mgr);
+int dcn35_smu_get_ips_supported(struct clk_mgr_internal *clk_mgr);
+int dcn35_smu_get_dtbclk(struct clk_mgr_internal *clk_mgr);
+int dcn35_smu_get_dprefclk(struct clk_mgr_internal *clk_mgr);
+void dcn35_smu_notify_host_router_bw(struct clk_mgr_internal *clk_mgr, uint32_t hr_id, uint32_t bw_kbps);
+
+#endif /* DAL_DC_35_SMU_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dalsmc.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dalsmc.h
new file mode 100644
index 000000000000..2e0d34fd7512
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dalsmc.h
@@ -0,0 +1,55 @@
+// SPDX-License-Identifier: MIT
+//
+// Copyright 2024 Advanced Micro Devices, Inc.
+
+#ifndef DALSMC_H
+#define DALSMC_H
+
+#define DALSMC_VERSION 0x1
+
+// SMU Response Codes:
+#define DALSMC_Result_OK                   0x1
+#define DALSMC_Result_Failed               0xFF
+#define DALSMC_Result_UnknownCmd           0xFE
+#define DALSMC_Result_CmdRejectedPrereq    0xFD
+#define DALSMC_Result_CmdRejectedBusy      0xFC
+
+
+
+// Message Definitions:
+#define DALSMC_MSG_TestMessage                    0x1
+#define DALSMC_MSG_GetSmuVersion                  0x2
+#define DALSMC_MSG_GetDriverIfVersion             0x3
+#define DALSMC_MSG_GetMsgHeaderVersion            0x4
+#define DALSMC_MSG_SetDalDramAddrHigh             0x5
+#define DALSMC_MSG_SetDalDramAddrLow              0x6
+#define DALSMC_MSG_TransferTableSmu2Dram          0x7
+#define DALSMC_MSG_TransferTableDram2Smu          0x8
+#define DALSMC_MSG_SetHardMinByFreq               0x9
+#define DALSMC_MSG_SetHardMaxByFreq               0xA
+#define DALSMC_MSG_GetDpmFreqByIndex              0xB
+#define DALSMC_MSG_GetDcModeMaxDpmFreq            0xC
+#define DALSMC_MSG_SetMinDeepSleepDcfclk          0xD
+#define DALSMC_MSG_NumOfDisplays                  0xE
+#define DALSMC_MSG_SetExternalClientDfCstateAllow 0xF
+#define DALSMC_MSG_BacoAudioD3PME                 0x10
+#define DALSMC_MSG_SetFclkSwitchAllow             0x11
+#define DALSMC_MSG_SetCabForUclkPstate            0x12
+#define DALSMC_MSG_SetWorstCaseUclkLatency        0x13
+#define DALSMC_MSG_DcnExitReset                   0x14
+#define DALSMC_MSG_ReturnHardMinStatus            0x15
+#define DALSMC_MSG_SetAlwaysWaitDmcubResp         0x16
+#define DALSMC_MSG_IndicateDrrStatus              0x17  // PMFW 15811
+#define DALSMC_MSG_ActiveUclkFclk                 0x18
+#define DALSMC_MSG_IdleUclkFclk                   0x19
+#define DALSMC_MSG_SetUclkPstateAllow             0x1A
+#define DALSMC_MSG_SubvpUclkFclk                  0x1B
+#define DALSMC_MSG_GetNumUmcChannels              0x1C
+#define DALSMC_Message_Count                      0x1D
+
+typedef enum {
+  FCLK_SWITCH_DISALLOW,
+  FCLK_SWITCH_ALLOW,
+} FclkSwitchAllow_e;
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr.c
new file mode 100644
index 000000000000..306016c1f109
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr.c
@@ -0,0 +1,1631 @@
+// SPDX-License-Identifier: MIT
+//
+// Copyright 2024 Advanced Micro Devices, Inc.
+
+#include "dccg.h"
+#include "clk_mgr_internal.h"
+#include "dcn401/dcn401_clk_mgr_smu_msg.h"
+#include "dcn20/dcn20_clk_mgr.h"
+#include "dce100/dce_clk_mgr.h"
+#include "dcn31/dcn31_clk_mgr.h"
+#include "dcn32/dcn32_clk_mgr.h"
+#include "dcn401/dcn401_clk_mgr.h"
+#include "reg_helper.h"
+#include "core_types.h"
+#include "dm_helpers.h"
+#include "link_service.h"
+#include "dc_state_priv.h"
+#include "atomfirmware.h"
+
+#include "dcn401_smu14_driver_if.h"
+
+#include "dcn/dcn_4_1_0_offset.h"
+#include "dcn/dcn_4_1_0_sh_mask.h"
+
+#define DCN_BASE__INST0_SEG1                       0x000000C0
+
+#define mmCLK01_CLK0_CLK_PLL_REQ                        0x16E37
+#define mmCLK01_CLK0_CLK0_DFS_CNTL                      0x16E69
+#define mmCLK01_CLK0_CLK1_DFS_CNTL                      0x16E6C
+#define mmCLK01_CLK0_CLK2_DFS_CNTL                      0x16E6F
+#define mmCLK01_CLK0_CLK3_DFS_CNTL                      0x16E72
+#define mmCLK01_CLK0_CLK4_DFS_CNTL                      0x16E75
+#define mmCLK20_CLK2_CLK2_DFS_CNTL                      0x1B051
+
+#define CLK0_CLK_PLL_REQ__FbMult_int_MASK                  0x000001ffUL
+#define CLK0_CLK_PLL_REQ__PllSpineDiv_MASK                 0x0000f000UL
+#define CLK0_CLK_PLL_REQ__FbMult_frac_MASK                 0xffff0000UL
+#define CLK0_CLK_PLL_REQ__FbMult_int__SHIFT                0x00000000
+#define CLK0_CLK_PLL_REQ__PllSpineDiv__SHIFT               0x0000000c
+#define CLK0_CLK_PLL_REQ__FbMult_frac__SHIFT               0x00000010
+
+#undef FN
+#define FN(reg_name, field_name) \
+	clk_mgr->clk_mgr_shift->field_name, clk_mgr->clk_mgr_mask->field_name
+
+#define REG(reg) \
+	(clk_mgr->regs->reg)
+
+#define BASE_INNER(seg) DCN_BASE__INST0_SEG ## seg
+
+#define BASE(seg) BASE_INNER(seg)
+
+#define SR(reg_name)\
+		.reg_name = BASE(reg ## reg_name ## _BASE_IDX) +  \
+					reg ## reg_name
+
+#define CLK_SR_DCN401(reg_name, block, inst)\
+	.reg_name = mm ## block ## _ ## reg_name
+
+static const struct clk_mgr_registers clk_mgr_regs_dcn401 = {
+	CLK_REG_LIST_DCN401()
+};
+
+static const struct clk_mgr_shift clk_mgr_shift_dcn401 = {
+	CLK_COMMON_MASK_SH_LIST_DCN401(__SHIFT)
+};
+
+static const struct clk_mgr_mask clk_mgr_mask_dcn401 = {
+	CLK_COMMON_MASK_SH_LIST_DCN401(_MASK)
+};
+
+#define TO_DCN401_CLK_MGR(clk_mgr)\
+	container_of(clk_mgr, struct dcn401_clk_mgr, base)
+
+static bool dcn401_is_ppclk_dpm_enabled(struct clk_mgr_internal *clk_mgr, PPCLK_e clk)
+{
+	bool ppclk_dpm_enabled = false;
+
+	switch (clk) {
+	case PPCLK_SOCCLK:
+		ppclk_dpm_enabled =
+				clk_mgr->base.bw_params->clk_table.num_entries_per_clk.num_socclk_levels > 1;
+		break;
+	case PPCLK_UCLK:
+		ppclk_dpm_enabled =
+				clk_mgr->base.bw_params->clk_table.num_entries_per_clk.num_memclk_levels > 1;
+		break;
+	case PPCLK_FCLK:
+		ppclk_dpm_enabled =
+				clk_mgr->base.bw_params->clk_table.num_entries_per_clk.num_fclk_levels > 1;
+		break;
+	case PPCLK_DISPCLK:
+		ppclk_dpm_enabled =
+				clk_mgr->base.bw_params->clk_table.num_entries_per_clk.num_dispclk_levels > 1;
+		break;
+	case PPCLK_DPPCLK:
+		ppclk_dpm_enabled =
+				clk_mgr->base.bw_params->clk_table.num_entries_per_clk.num_dppclk_levels > 1;
+		break;
+	case PPCLK_DPREFCLK:
+		ppclk_dpm_enabled = false;
+		break;
+	case PPCLK_DCFCLK:
+		ppclk_dpm_enabled =
+				clk_mgr->base.bw_params->clk_table.num_entries_per_clk.num_dcfclk_levels > 1;
+		break;
+	case PPCLK_DTBCLK:
+		ppclk_dpm_enabled =
+				clk_mgr->base.bw_params->clk_table.num_entries_per_clk.num_dtbclk_levels > 1;
+		break;
+	default:
+		ppclk_dpm_enabled = false;
+	}
+
+	ppclk_dpm_enabled &= clk_mgr->smu_present;
+
+	return ppclk_dpm_enabled;
+}
+
+static bool dcn401_is_ppclk_idle_dpm_enabled(struct clk_mgr_internal *clk_mgr, PPCLK_e clk)
+{
+	bool ppclk_idle_dpm_enabled = false;
+
+	switch (clk) {
+	case PPCLK_UCLK:
+	case PPCLK_FCLK:
+		if (ASICREV_IS_GC_12_0_0_A0(clk_mgr->base.ctx->asic_id.hw_internal_rev) &&
+				clk_mgr->smu_ver >= 0x681800) {
+			ppclk_idle_dpm_enabled = true;
+		} else if (ASICREV_IS_GC_12_0_1_A0(clk_mgr->base.ctx->asic_id.hw_internal_rev) &&
+				clk_mgr->smu_ver >= 0x661300) {
+			ppclk_idle_dpm_enabled = true;
+		}
+		break;
+	default:
+		ppclk_idle_dpm_enabled = false;
+	}
+
+	ppclk_idle_dpm_enabled &= clk_mgr->smu_present;
+
+	return ppclk_idle_dpm_enabled;
+}
+
+static bool dcn401_is_df_throttle_opt_enabled(struct clk_mgr_internal *clk_mgr)
+{
+	bool is_df_throttle_opt_enabled = false;
+
+	if (ASICREV_IS_GC_12_0_1_A0(clk_mgr->base.ctx->asic_id.hw_internal_rev) &&
+			clk_mgr->smu_ver >= 0x663500) {
+		is_df_throttle_opt_enabled = !clk_mgr->base.ctx->dc->debug.force_subvp_df_throttle;
+	}
+
+	is_df_throttle_opt_enabled &= clk_mgr->smu_present;
+
+	return is_df_throttle_opt_enabled;
+}
+
+/* Query SMU for all clock states for a particular clock */
+static void dcn401_init_single_clock(struct clk_mgr_internal *clk_mgr, PPCLK_e clk, unsigned int *entry_0,
+		unsigned int *num_levels)
+{
+	unsigned int i;
+	char *entry_i = (char *)entry_0;
+
+	uint32_t ret = dcn401_smu_get_dpm_freq_by_index(clk_mgr, clk, 0xFF);
+
+	if (ret & (1 << 31))
+		/* fine-grained, only min and max */
+		*num_levels = 2;
+	else
+		/* discrete, a number of fixed states */
+		/* will set num_levels to 0 on failure */
+		*num_levels = ret & 0xFF;
+
+	/* if the initial message failed, num_levels will be 0 */
+	for (i = 0; i < *num_levels && i < ARRAY_SIZE(clk_mgr->base.bw_params->clk_table.entries); i++) {
+		*((unsigned int *)entry_i) = (dcn401_smu_get_dpm_freq_by_index(clk_mgr, clk, i) & 0xFFFF);
+		entry_i += sizeof(clk_mgr->base.bw_params->clk_table.entries[0]);
+	}
+}
+
+static void dcn401_build_wm_range_table(struct clk_mgr *clk_mgr)
+{
+	/* For min clocks use as reported by PM FW and report those as min */
+	uint16_t min_uclk_mhz = clk_mgr->bw_params->clk_table.entries[0].memclk_mhz;
+	uint16_t min_dcfclk_mhz	= clk_mgr->bw_params->clk_table.entries[0].dcfclk_mhz;
+
+	/* Set A - Normal - default values */
+	clk_mgr->bw_params->wm_table.nv_entries[WM_A].valid = true;
+	clk_mgr->bw_params->wm_table.nv_entries[WM_A].pmfw_breakdown.wm_type = WATERMARKS_CLOCK_RANGE;
+	clk_mgr->bw_params->wm_table.nv_entries[WM_A].pmfw_breakdown.min_dcfclk = min_dcfclk_mhz;
+	clk_mgr->bw_params->wm_table.nv_entries[WM_A].pmfw_breakdown.max_dcfclk = 0xFFFF;
+	clk_mgr->bw_params->wm_table.nv_entries[WM_A].pmfw_breakdown.min_uclk = min_uclk_mhz;
+	clk_mgr->bw_params->wm_table.nv_entries[WM_A].pmfw_breakdown.max_uclk = 0xFFFF;
+
+	/* Set B - Unused on dcn4 */
+	clk_mgr->bw_params->wm_table.nv_entries[WM_B].valid = false;
+
+	/* Set 1A - Dummy P-State - P-State latency set to "dummy p-state" value */
+	/* 'DalDummyClockChangeLatencyNs' registry key option set to 0x7FFFFFFF can be used to disable Set C for dummy p-state */
+	if (clk_mgr->ctx->dc->bb_overrides.dummy_clock_change_latency_ns != 0x7FFFFFFF) {
+		clk_mgr->bw_params->wm_table.nv_entries[WM_1A].valid = true;
+		clk_mgr->bw_params->wm_table.nv_entries[WM_1A].pmfw_breakdown.wm_type = WATERMARKS_DUMMY_PSTATE;
+		clk_mgr->bw_params->wm_table.nv_entries[WM_1A].pmfw_breakdown.min_dcfclk = min_dcfclk_mhz;
+		clk_mgr->bw_params->wm_table.nv_entries[WM_1A].pmfw_breakdown.max_dcfclk = 0xFFFF;
+		clk_mgr->bw_params->wm_table.nv_entries[WM_1A].pmfw_breakdown.min_uclk = min_uclk_mhz;
+		clk_mgr->bw_params->wm_table.nv_entries[WM_1A].pmfw_breakdown.max_uclk = 0xFFFF;
+	} else {
+		clk_mgr->bw_params->wm_table.nv_entries[WM_1A].valid = false;
+	}
+
+	/* Set 1B - Unused on dcn4 */
+	clk_mgr->bw_params->wm_table.nv_entries[WM_1B].valid = false;
+}
+
+void dcn401_init_clocks(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct clk_limit_num_entries *num_entries_per_clk;
+	unsigned int i;
+
+	if (!clk_mgr_base->bw_params)
+		return;
+
+	num_entries_per_clk = &clk_mgr_base->bw_params->clk_table.num_entries_per_clk;
+
+	memset(&(clk_mgr_base->clks), 0, sizeof(struct dc_clocks));
+	clk_mgr_base->clks.p_state_change_support = true;
+	clk_mgr_base->clks.prev_p_state_change_support = true;
+	clk_mgr_base->clks.fclk_prev_p_state_change_support = true;
+	clk_mgr->smu_present = false;
+	clk_mgr->dpm_present = false;
+
+	if (!clk_mgr_base->force_smu_not_present && dcn401_smu_get_smu_version(clk_mgr, &clk_mgr->smu_ver))
+		clk_mgr->smu_present = true;
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn401_smu_check_driver_if_version(clk_mgr);
+	dcn401_smu_check_msg_header_version(clk_mgr);
+
+	/* DCFCLK */
+	dcn401_init_single_clock(clk_mgr, PPCLK_DCFCLK,
+			&clk_mgr_base->bw_params->clk_table.entries[0].dcfclk_mhz,
+			&num_entries_per_clk->num_dcfclk_levels);
+	clk_mgr_base->bw_params->dc_mode_limit.dcfclk_mhz = dcn401_smu_get_dc_mode_max_dpm_freq(clk_mgr, PPCLK_DCFCLK);
+	if (num_entries_per_clk->num_dcfclk_levels && clk_mgr_base->bw_params->dc_mode_limit.dcfclk_mhz ==
+			clk_mgr_base->bw_params->clk_table.entries[num_entries_per_clk->num_dcfclk_levels - 1].dcfclk_mhz)
+		clk_mgr_base->bw_params->dc_mode_limit.dcfclk_mhz = 0;
+
+	/* SOCCLK */
+	dcn401_init_single_clock(clk_mgr, PPCLK_SOCCLK,
+					&clk_mgr_base->bw_params->clk_table.entries[0].socclk_mhz,
+					&num_entries_per_clk->num_socclk_levels);
+	clk_mgr_base->bw_params->dc_mode_limit.socclk_mhz = dcn401_smu_get_dc_mode_max_dpm_freq(clk_mgr, PPCLK_SOCCLK);
+	if (num_entries_per_clk->num_socclk_levels && clk_mgr_base->bw_params->dc_mode_limit.socclk_mhz ==
+			clk_mgr_base->bw_params->clk_table.entries[num_entries_per_clk->num_socclk_levels - 1].socclk_mhz)
+		clk_mgr_base->bw_params->dc_mode_limit.socclk_mhz = 0;
+
+	/* DTBCLK */
+	if (!clk_mgr->base.ctx->dc->debug.disable_dtb_ref_clk_switch) {
+		dcn401_init_single_clock(clk_mgr, PPCLK_DTBCLK,
+				&clk_mgr_base->bw_params->clk_table.entries[0].dtbclk_mhz,
+				&num_entries_per_clk->num_dtbclk_levels);
+		clk_mgr_base->bw_params->dc_mode_limit.dtbclk_mhz = dcn401_smu_get_dc_mode_max_dpm_freq(clk_mgr, PPCLK_DTBCLK);
+		if (num_entries_per_clk->num_dtbclk_levels && clk_mgr_base->bw_params->dc_mode_limit.dtbclk_mhz ==
+				clk_mgr_base->bw_params->clk_table.entries[num_entries_per_clk->num_dtbclk_levels - 1].dtbclk_mhz)
+			clk_mgr_base->bw_params->dc_mode_limit.dtbclk_mhz = 0;
+	}
+
+	/* DISPCLK */
+	dcn401_init_single_clock(clk_mgr, PPCLK_DISPCLK,
+			&clk_mgr_base->bw_params->clk_table.entries[0].dispclk_mhz,
+			&num_entries_per_clk->num_dispclk_levels);
+	clk_mgr_base->bw_params->dc_mode_limit.dispclk_mhz = dcn401_smu_get_dc_mode_max_dpm_freq(clk_mgr, PPCLK_DISPCLK);
+	if (num_entries_per_clk->num_dispclk_levels && clk_mgr_base->bw_params->dc_mode_limit.dispclk_mhz ==
+			clk_mgr_base->bw_params->clk_table.entries[num_entries_per_clk->num_dispclk_levels - 1].dispclk_mhz)
+		clk_mgr_base->bw_params->dc_mode_limit.dispclk_mhz = 0;
+
+	/* DPPCLK */
+	dcn401_init_single_clock(clk_mgr, PPCLK_DPPCLK,
+			&clk_mgr_base->bw_params->clk_table.entries[0].dppclk_mhz,
+			&num_entries_per_clk->num_dppclk_levels);
+
+	if (num_entries_per_clk->num_dcfclk_levels &&
+			num_entries_per_clk->num_dtbclk_levels &&
+			num_entries_per_clk->num_dispclk_levels)
+		clk_mgr->dpm_present = true;
+
+	if (clk_mgr_base->ctx->dc->debug.min_disp_clk_khz) {
+		for (i = 0; i < num_entries_per_clk->num_dispclk_levels; i++)
+			if (clk_mgr_base->bw_params->clk_table.entries[i].dispclk_mhz
+					< khz_to_mhz_ceil(clk_mgr_base->ctx->dc->debug.min_disp_clk_khz))
+				clk_mgr_base->bw_params->clk_table.entries[i].dispclk_mhz
+					= khz_to_mhz_ceil(clk_mgr_base->ctx->dc->debug.min_disp_clk_khz);
+	}
+
+	if (clk_mgr_base->ctx->dc->debug.min_dpp_clk_khz) {
+		for (i = 0; i < num_entries_per_clk->num_dppclk_levels; i++)
+			if (clk_mgr_base->bw_params->clk_table.entries[i].dppclk_mhz
+					< khz_to_mhz_ceil(clk_mgr_base->ctx->dc->debug.min_dpp_clk_khz))
+				clk_mgr_base->bw_params->clk_table.entries[i].dppclk_mhz
+					= khz_to_mhz_ceil(clk_mgr_base->ctx->dc->debug.min_dpp_clk_khz);
+	}
+
+	/* Get UCLK, update bounding box */
+	clk_mgr_base->funcs->get_memclk_states_from_smu(clk_mgr_base);
+
+	/* WM range table */
+	dcn401_build_wm_range_table(clk_mgr_base);
+}
+
+bool dcn401_is_dc_mode_present(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	return clk_mgr->smu_present && clk_mgr->dpm_present &&
+			((clk_mgr_base->bw_params->clk_table.num_entries_per_clk.num_dcfclk_levels &&
+			clk_mgr_base->bw_params->dc_mode_limit.dcfclk_mhz) ||
+			(clk_mgr_base->bw_params->clk_table.num_entries_per_clk.num_dispclk_levels &&
+			clk_mgr_base->bw_params->dc_mode_limit.dispclk_mhz) ||
+			(clk_mgr_base->bw_params->clk_table.num_entries_per_clk.num_dtbclk_levels &&
+			clk_mgr_base->bw_params->dc_mode_limit.dtbclk_mhz) ||
+			(clk_mgr_base->bw_params->clk_table.num_entries_per_clk.num_fclk_levels &&
+			clk_mgr_base->bw_params->dc_mode_limit.fclk_mhz) ||
+			(clk_mgr_base->bw_params->clk_table.num_entries_per_clk.num_memclk_levels &&
+			clk_mgr_base->bw_params->dc_mode_limit.memclk_mhz) ||
+			(clk_mgr_base->bw_params->clk_table.num_entries_per_clk.num_socclk_levels &&
+			clk_mgr_base->bw_params->dc_mode_limit.socclk_mhz));
+}
+
+static void dcn401_dump_clk_registers(struct clk_state_registers_and_bypass *regs_and_bypass,
+		struct clk_mgr *clk_mgr_base, struct clk_log_info *log_info)
+{
+		struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+		uint32_t dprefclk_did = 0;
+		uint32_t dcfclk_did = 0;
+		uint32_t dtbclk_did = 0;
+		uint32_t dispclk_did = 0;
+		uint32_t dppclk_did = 0;
+		uint32_t fclk_did = 0;
+		uint32_t target_div = 0;
+
+		/* DFS Slice 0 is used for DISPCLK */
+		dispclk_did = REG_READ(CLK0_CLK0_DFS_CNTL);
+		/* DFS Slice 1 is used for DPPCLK */
+		dppclk_did = REG_READ(CLK0_CLK1_DFS_CNTL);
+		/* DFS Slice 2 is used for DPREFCLK */
+		dprefclk_did = REG_READ(CLK0_CLK2_DFS_CNTL);
+		/* DFS Slice 3 is used for DCFCLK */
+		dcfclk_did = REG_READ(CLK0_CLK3_DFS_CNTL);
+		/* DFS Slice 4 is used for DTBCLK */
+		dtbclk_did = REG_READ(CLK0_CLK4_DFS_CNTL);
+		/* DFS Slice _ is used for FCLK */
+		fclk_did = REG_READ(CLK2_CLK2_DFS_CNTL);
+
+		/* Convert DISPCLK DFS Slice DID to divider*/
+		target_div = dentist_get_divider_from_did(dispclk_did);
+		//Get dispclk in khz
+		regs_and_bypass->dispclk = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+				* clk_mgr->base.dentist_vco_freq_khz) / target_div;
+
+		/* Convert DISPCLK DFS Slice DID to divider*/
+		target_div = dentist_get_divider_from_did(dppclk_did);
+		//Get dppclk in khz
+		regs_and_bypass->dppclk = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+				* clk_mgr->base.dentist_vco_freq_khz) / target_div;
+
+		/* Convert DPREFCLK DFS Slice DID to divider*/
+		target_div = dentist_get_divider_from_did(dprefclk_did);
+		//Get dprefclk in khz
+		regs_and_bypass->dprefclk = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+				* clk_mgr->base.dentist_vco_freq_khz) / target_div;
+
+		/* Convert DCFCLK DFS Slice DID to divider*/
+		target_div = dentist_get_divider_from_did(dcfclk_did);
+		//Get dcfclk in khz
+		regs_and_bypass->dcfclk = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+				* clk_mgr->base.dentist_vco_freq_khz) / target_div;
+
+		/* Convert DTBCLK DFS Slice DID to divider*/
+		target_div = dentist_get_divider_from_did(dtbclk_did);
+		//Get dtbclk in khz
+		regs_and_bypass->dtbclk = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+				* clk_mgr->base.dentist_vco_freq_khz) / target_div;
+
+		/* Convert DTBCLK DFS Slice DID to divider*/
+		target_div = dentist_get_divider_from_did(fclk_did);
+		//Get fclk in khz
+		regs_and_bypass->fclk = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+				* clk_mgr->base.dentist_vco_freq_khz) / target_div;
+}
+
+static bool dcn401_check_native_scaling(struct pipe_ctx *pipe)
+{
+	bool is_native_scaling = false;
+	int width = pipe->plane_state->src_rect.width;
+	int height = pipe->plane_state->src_rect.height;
+
+	if (pipe->stream->timing.h_addressable == width &&
+			pipe->stream->timing.v_addressable == height &&
+			pipe->plane_state->dst_rect.width == width &&
+			pipe->plane_state->dst_rect.height == height)
+		is_native_scaling = true;
+
+	return is_native_scaling;
+}
+
+static void dcn401_auto_dpm_test_log(
+		struct dc_clocks *new_clocks,
+		struct clk_mgr_internal *clk_mgr,
+		struct dc_state *context)
+{
+	unsigned int mall_ss_size_bytes;
+	int dramclk_khz_override, fclk_khz_override, num_fclk_levels;
+
+	struct pipe_ctx *pipe_ctx_list[MAX_PIPES];
+	int active_pipe_count = 0;
+
+	for (int i = 0; i < MAX_PIPES; i++) {
+		struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i];
+
+		if (pipe_ctx->stream && dc_state_get_pipe_subvp_type(context, pipe_ctx) != SUBVP_PHANTOM) {
+			pipe_ctx_list[active_pipe_count] = pipe_ctx;
+			active_pipe_count++;
+		}
+	}
+
+	msleep(5);
+
+	mall_ss_size_bytes = context->bw_ctx.bw.dcn.mall_ss_size_bytes;
+
+	struct clk_log_info log_info = {0};
+	struct clk_state_registers_and_bypass clk_register_dump;
+
+	dcn401_dump_clk_registers(&clk_register_dump, &clk_mgr->base, &log_info);
+
+	// Overrides for these clocks in case there is no p_state change support
+	dramclk_khz_override = new_clocks->dramclk_khz;
+	fclk_khz_override = new_clocks->fclk_khz;
+
+	num_fclk_levels = clk_mgr->base.bw_params->clk_table.num_entries_per_clk.num_fclk_levels - 1;
+
+	if (!new_clocks->p_state_change_support)
+		dramclk_khz_override = clk_mgr->base.bw_params->max_memclk_mhz * 1000;
+
+	if (!new_clocks->fclk_p_state_change_support)
+		fclk_khz_override = clk_mgr->base.bw_params->clk_table.entries[num_fclk_levels].fclk_mhz * 1000;
+
+
+	////////////////////////////////////////////////////////////////////////////
+	//	IMPORTANT: 	When adding more clocks to these logs, do NOT put a newline
+	//	 			anywhere other than at the very end of the string.
+	//
+	//	Formatting example (make sure to have " - " between each entry):
+	//
+	//				AutoDPMTest: clk1:%d - clk2:%d - clk3:%d - clk4:%d\n"
+	////////////////////////////////////////////////////////////////////////////
+	if (active_pipe_count > 0 &&
+		new_clocks->dramclk_khz > 0 &&
+		new_clocks->fclk_khz > 0 &&
+		new_clocks->dcfclk_khz > 0 &&
+		new_clocks->dppclk_khz > 0) {
+
+		uint32_t pix_clk_list[MAX_PIPES] = {0};
+		int p_state_list[MAX_PIPES] = {0};
+		int disp_src_width_list[MAX_PIPES] = {0};
+		int disp_src_height_list[MAX_PIPES] = {0};
+		uint64_t disp_src_refresh_list[MAX_PIPES] = {0};
+		bool is_scaled_list[MAX_PIPES] = {0};
+
+		for (int i = 0; i < active_pipe_count; i++) {
+			struct pipe_ctx *curr_pipe_ctx = pipe_ctx_list[i];
+			uint64_t refresh_rate;
+
+			pix_clk_list[i] = curr_pipe_ctx->stream->timing.pix_clk_100hz;
+			p_state_list[i] = curr_pipe_ctx->p_state_type;
+
+			refresh_rate = (curr_pipe_ctx->stream->timing.pix_clk_100hz * (uint64_t)100 +
+				curr_pipe_ctx->stream->timing.v_total
+				* (uint64_t) curr_pipe_ctx->stream->timing.h_total - (uint64_t)1);
+			refresh_rate = div_u64(refresh_rate, curr_pipe_ctx->stream->timing.v_total);
+			refresh_rate = div_u64(refresh_rate, curr_pipe_ctx->stream->timing.h_total);
+			disp_src_refresh_list[i] = refresh_rate;
+
+			if (curr_pipe_ctx->plane_state) {
+				is_scaled_list[i] = !(dcn401_check_native_scaling(curr_pipe_ctx));
+				disp_src_width_list[i] = curr_pipe_ctx->plane_state->src_rect.width;
+				disp_src_height_list[i] = curr_pipe_ctx->plane_state->src_rect.height;
+			}
+		}
+
+		DC_LOG_AUTO_DPM_TEST("AutoDPMTest: dramclk:%d - fclk:%d - "
+			"dcfclk:%d - dppclk:%d - dispclk_hw:%d - "
+			"dppclk_hw:%d - dprefclk_hw:%d - dcfclk_hw:%d - "
+			"dtbclk_hw:%d - fclk_hw:%d - pix_clk_0:%d - pix_clk_1:%d - "
+			"pix_clk_2:%d - pix_clk_3:%d - mall_ss_size:%d - p_state_type_0:%d - "
+			"p_state_type_1:%d - p_state_type_2:%d - p_state_type_3:%d - "
+			"pix_width_0:%d - pix_height_0:%d - refresh_rate_0:%lld - is_scaled_0:%d - "
+			"pix_width_1:%d - pix_height_1:%d - refresh_rate_1:%lld - is_scaled_1:%d - "
+			"pix_width_2:%d - pix_height_2:%d - refresh_rate_2:%lld - is_scaled_2:%d - "
+			"pix_width_3:%d - pix_height_3:%d - refresh_rate_3:%lld - is_scaled_3:%d - LOG_END\n",
+			dramclk_khz_override,
+			fclk_khz_override,
+			new_clocks->dcfclk_khz,
+			new_clocks->dppclk_khz,
+			clk_register_dump.dispclk,
+			clk_register_dump.dppclk,
+			clk_register_dump.dprefclk,
+			clk_register_dump.dcfclk,
+			clk_register_dump.dtbclk,
+			clk_register_dump.fclk,
+			pix_clk_list[0], pix_clk_list[1], pix_clk_list[3], pix_clk_list[2],
+			mall_ss_size_bytes,
+			p_state_list[0], p_state_list[1], p_state_list[2], p_state_list[3],
+			disp_src_width_list[0], disp_src_height_list[0], disp_src_refresh_list[0], is_scaled_list[0],
+			disp_src_width_list[1], disp_src_height_list[1], disp_src_refresh_list[1], is_scaled_list[1],
+			disp_src_width_list[2], disp_src_height_list[2], disp_src_refresh_list[2], is_scaled_list[2],
+			disp_src_width_list[3], disp_src_height_list[3], disp_src_refresh_list[3], is_scaled_list[3]);
+	}
+}
+
+static void dcn401_update_clocks_update_dtb_dto(struct clk_mgr_internal *clk_mgr,
+			struct dc_state *context,
+			int ref_dtbclk_khz)
+{
+	int i;
+	struct dccg *dccg = clk_mgr->dccg;
+	struct pipe_ctx *otg_master;
+	bool use_hpo_encoder;
+
+
+	for (i = 0; i < context->stream_count; i++) {
+		otg_master = resource_get_otg_master_for_stream(
+				&context->res_ctx, context->streams[i]);
+		ASSERT(otg_master);
+		ASSERT(otg_master->clock_source);
+		ASSERT(otg_master->clock_source->funcs->program_pix_clk);
+		ASSERT(otg_master->stream_res.pix_clk_params.controller_id >= CONTROLLER_ID_D0);
+
+		use_hpo_encoder = dccg->ctx->dc->link_srv->dp_is_128b_132b_signal(otg_master);
+		if (!use_hpo_encoder)
+			continue;
+
+		if (otg_master->stream_res.pix_clk_params.controller_id > CONTROLLER_ID_UNDEFINED)
+			otg_master->clock_source->funcs->program_pix_clk(
+				otg_master->clock_source,
+				&otg_master->stream_res.pix_clk_params,
+				dccg->ctx->dc->link_srv->dp_get_encoding_format(
+					&otg_master->link_config.dp_link_settings),
+				&otg_master->pll_settings);
+	}
+}
+
+static void dcn401_update_clocks_update_dpp_dto(struct clk_mgr_internal *clk_mgr,
+		struct dc_state *context, bool safe_to_lower, int ref_dppclk_khz)
+{
+	int i;
+
+	clk_mgr->dccg->ref_dppclk = ref_dppclk_khz;
+	for (i = 0; i < clk_mgr->base.ctx->dc->res_pool->pipe_count; i++) {
+		int dpp_inst = 0, dppclk_khz, prev_dppclk_khz;
+
+		dppclk_khz = context->res_ctx.pipe_ctx[i].plane_res.bw.dppclk_khz;
+
+		if (context->res_ctx.pipe_ctx[i].plane_res.dpp)
+			dpp_inst = context->res_ctx.pipe_ctx[i].plane_res.dpp->inst;
+		else if (!context->res_ctx.pipe_ctx[i].plane_res.dpp && dppclk_khz == 0) {
+			/* dpp == NULL && dppclk_khz == 0 is valid because of pipe harvesting.
+			 * In this case just continue in loop
+			 */
+			continue;
+		} else if (!context->res_ctx.pipe_ctx[i].plane_res.dpp && dppclk_khz > 0) {
+			/* The software state is not valid if dpp resource is NULL and
+			 * dppclk_khz > 0.
+			 */
+			ASSERT(false);
+			continue;
+		}
+
+		prev_dppclk_khz = clk_mgr->dccg->pipe_dppclk_khz[i];
+
+		if (safe_to_lower || prev_dppclk_khz < dppclk_khz)
+			clk_mgr->dccg->funcs->update_dpp_dto(
+							clk_mgr->dccg, dpp_inst, dppclk_khz);
+	}
+}
+
+static int dcn401_set_hard_min_by_freq_optimized(struct clk_mgr_internal *clk_mgr, PPCLK_e clk, int requested_clk_khz)
+{
+	if (!clk_mgr->smu_present || !dcn401_is_ppclk_dpm_enabled(clk_mgr, clk))
+		return 0;
+
+	/*
+	 * SMU set hard min interface takes requested clock in mhz and return
+	 * actual clock configured in khz. If we floor requested clk to mhz,
+	 * there is a chance that the actual clock configured in khz is less
+	 * than requested. If we ceil it to mhz, there is a chance that it
+	 * unnecessarily dumps up to a higher dpm level, which burns more power.
+	 * The solution is to set by flooring it to mhz first. If the actual
+	 * clock returned is less than requested, then we will ceil the
+	 * requested value to mhz and call it again.
+	 */
+	int actual_clk_khz = dcn401_smu_set_hard_min_by_freq(clk_mgr, clk, khz_to_mhz_floor(requested_clk_khz));
+
+	if (actual_clk_khz < requested_clk_khz)
+		actual_clk_khz = dcn401_smu_set_hard_min_by_freq(clk_mgr, clk, khz_to_mhz_ceil(requested_clk_khz));
+
+	return actual_clk_khz;
+}
+
+static void dcn401_update_clocks_update_dentist(
+		struct clk_mgr_internal *clk_mgr,
+		struct dc_state *context)
+{
+	uint32_t new_disp_divider = 0;
+	uint32_t new_dispclk_wdivider = 0;
+	uint32_t dentist_dispclk_wdivider_readback = 0;
+	struct dc *dc = clk_mgr->base.ctx->dc;
+
+	if (clk_mgr->base.clks.dispclk_khz == 0)
+		return;
+
+	new_disp_divider = DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+			* clk_mgr->base.dentist_vco_freq_khz / clk_mgr->base.clks.dispclk_khz;
+
+	new_dispclk_wdivider = dentist_get_did_from_divider(new_disp_divider);
+
+	if (dc->debug.override_dispclk_programming) {
+		REG_GET(DENTIST_DISPCLK_CNTL,
+				DENTIST_DISPCLK_WDIVIDER, &dentist_dispclk_wdivider_readback);
+
+		if (dentist_dispclk_wdivider_readback > new_dispclk_wdivider) {
+			REG_UPDATE(DENTIST_DISPCLK_CNTL,
+					DENTIST_DISPCLK_WDIVIDER, new_dispclk_wdivider);
+			REG_WAIT(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_CHG_DONE, 1, 50, 2000);
+		}
+	}
+
+}
+
+static void dcn401_execute_block_sequence(struct clk_mgr *clk_mgr_base, unsigned int num_steps)
+{
+	struct clk_mgr_internal *clk_mgr_internal = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dcn401_clk_mgr *clk_mgr401 = TO_DCN401_CLK_MGR(clk_mgr_internal);
+
+	unsigned int i;
+	union dcn401_clk_mgr_block_sequence_params *params;
+
+	/* execute sequence */
+	for (i = 0; i < num_steps; i++) {
+		params = &clk_mgr401->block_sequence[i].params;
+
+		switch (clk_mgr401->block_sequence[i].func) {
+		case CLK_MGR401_READ_CLOCKS_FROM_DENTIST:
+			dcn2_read_clocks_from_hw_dentist(clk_mgr_base);
+			break;
+		case CLK_MGR401_UPDATE_NUM_DISPLAYS:
+			dcn401_smu_set_num_of_displays(clk_mgr_internal,
+					params->update_num_displays_params.num_displays);
+			break;
+		case CLK_MGR401_UPDATE_HARDMIN_PPCLK:
+			if (params->update_hardmin_params.response)
+				*params->update_hardmin_params.response = dcn401_smu_set_hard_min_by_freq(
+						clk_mgr_internal,
+						params->update_hardmin_params.ppclk,
+						params->update_hardmin_params.freq_mhz);
+			else
+				dcn401_smu_set_hard_min_by_freq(clk_mgr_internal,
+						params->update_hardmin_params.ppclk,
+						params->update_hardmin_params.freq_mhz);
+			break;
+		case CLK_MGR401_UPDATE_HARDMIN_PPCLK_OPTIMIZED:
+			if (params->update_hardmin_optimized_params.response)
+				*params->update_hardmin_optimized_params.response = dcn401_set_hard_min_by_freq_optimized(
+						clk_mgr_internal,
+						params->update_hardmin_optimized_params.ppclk,
+						params->update_hardmin_optimized_params.freq_khz);
+			else
+				dcn401_set_hard_min_by_freq_optimized(clk_mgr_internal,
+						params->update_hardmin_optimized_params.ppclk,
+						params->update_hardmin_optimized_params.freq_khz);
+			break;
+		case CLK_MGR401_UPDATE_ACTIVE_HARDMINS:
+			dcn401_smu_set_active_uclk_fclk_hardmin(
+					clk_mgr_internal,
+					params->update_idle_hardmin_params.uclk_mhz,
+					params->update_idle_hardmin_params.fclk_mhz);
+			break;
+		case CLK_MGR401_UPDATE_IDLE_HARDMINS:
+			dcn401_smu_set_idle_uclk_fclk_hardmin(
+					clk_mgr_internal,
+					params->update_idle_hardmin_params.uclk_mhz,
+					params->update_idle_hardmin_params.fclk_mhz);
+			break;
+		case CLK_MGR401_UPDATE_SUBVP_HARDMINS:
+			dcn401_smu_set_subvp_uclk_fclk_hardmin(
+					clk_mgr_internal,
+					params->update_idle_hardmin_params.uclk_mhz,
+					params->update_idle_hardmin_params.fclk_mhz);
+			break;
+		case CLK_MGR401_UPDATE_DEEP_SLEEP_DCFCLK:
+			dcn401_smu_set_min_deep_sleep_dcef_clk(
+					clk_mgr_internal,
+					params->update_deep_sleep_dcfclk_params.freq_mhz);
+			break;
+		case CLK_MGR401_UPDATE_FCLK_PSTATE_SUPPORT:
+			dcn401_smu_send_fclk_pstate_message(
+					clk_mgr_internal,
+					params->update_pstate_support_params.support);
+			break;
+		case CLK_MGR401_UPDATE_UCLK_PSTATE_SUPPORT:
+			dcn401_smu_send_uclk_pstate_message(
+					clk_mgr_internal,
+					params->update_pstate_support_params.support);
+			break;
+		case CLK_MGR401_UPDATE_CAB_FOR_UCLK:
+			dcn401_smu_send_cab_for_uclk_message(
+				clk_mgr_internal,
+				params->update_cab_for_uclk_params.num_ways);
+			break;
+		case CLK_MGR401_UPDATE_WAIT_FOR_DMUB_ACK:
+			dcn401_smu_wait_for_dmub_ack_mclk(
+					clk_mgr_internal,
+					params->update_wait_for_dmub_ack_params.enable);
+			break;
+		case CLK_MGR401_INDICATE_DRR_STATUS:
+			dcn401_smu_indicate_drr_status(
+					clk_mgr_internal,
+					params->indicate_drr_status_params.mod_drr_for_pstate);
+			break;
+		case CLK_MGR401_UPDATE_DPPCLK_DTO:
+			dcn401_update_clocks_update_dpp_dto(
+					clk_mgr_internal,
+					params->update_dppclk_dto_params.context,
+					params->update_dppclk_dto_params.safe_to_lower,
+					*params->update_dppclk_dto_params.ref_dppclk_khz);
+			break;
+		case CLK_MGR401_UPDATE_DTBCLK_DTO:
+			dcn401_update_clocks_update_dtb_dto(
+					clk_mgr_internal,
+					params->update_dtbclk_dto_params.context,
+					*params->update_dtbclk_dto_params.ref_dtbclk_khz);
+			break;
+		case CLK_MGR401_UPDATE_DENTIST:
+			dcn401_update_clocks_update_dentist(
+					clk_mgr_internal,
+					params->update_dentist_params.context);
+			break;
+		case CLK_MGR401_UPDATE_PSR_WAIT_LOOP:
+			params->update_psr_wait_loop_params.dmcu->funcs->set_psr_wait_loop(
+					params->update_psr_wait_loop_params.dmcu,
+					params->update_psr_wait_loop_params.wait);
+			break;
+		default:
+			/* this should never happen */
+			BREAK_TO_DEBUGGER();
+			break;
+		}
+	}
+}
+
+static unsigned int dcn401_build_update_bandwidth_clocks_sequence(
+		struct clk_mgr *clk_mgr_base,
+		struct dc_state *context,
+		struct dc_clocks *new_clocks,
+		bool safe_to_lower)
+{
+	struct clk_mgr_internal *clk_mgr_internal = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dcn401_clk_mgr *clk_mgr401 = TO_DCN401_CLK_MGR(clk_mgr_internal);
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	struct dcn401_clk_mgr_block_sequence *block_sequence = clk_mgr401->block_sequence;
+	bool enter_display_off = false;
+	bool update_active_fclk = false;
+	bool update_active_uclk = false;
+	bool update_idle_fclk = false;
+	bool update_idle_uclk = false;
+	bool update_subvp_prefetch_dramclk = false;
+	bool update_subvp_prefetch_fclk = false;
+	bool is_idle_dpm_enabled = dcn401_is_ppclk_dpm_enabled(clk_mgr_internal, PPCLK_UCLK) &&
+			dcn401_is_ppclk_dpm_enabled(clk_mgr_internal, PPCLK_FCLK) &&
+			dcn401_is_ppclk_idle_dpm_enabled(clk_mgr_internal, PPCLK_UCLK) &&
+			dcn401_is_ppclk_idle_dpm_enabled(clk_mgr_internal, PPCLK_FCLK);
+	bool is_df_throttle_opt_enabled = is_idle_dpm_enabled &&
+		dcn401_is_df_throttle_opt_enabled(clk_mgr_internal);
+	int total_plane_count = clk_mgr_helper_get_active_plane_cnt(dc, context);
+	int active_uclk_mhz = khz_to_mhz_ceil(clk_mgr_base->clks.dramclk_khz);
+	int active_fclk_mhz = khz_to_mhz_ceil(clk_mgr_base->clks.fclk_khz);
+	int idle_uclk_mhz = khz_to_mhz_ceil(clk_mgr_base->clks.idle_dramclk_khz);
+	int idle_fclk_mhz = khz_to_mhz_ceil(clk_mgr_base->clks.idle_fclk_khz);
+	int subvp_prefetch_dramclk_mhz = khz_to_mhz_ceil(clk_mgr_base->clks.subvp_prefetch_dramclk_khz);
+	int subvp_prefetch_fclk_mhz = khz_to_mhz_ceil(clk_mgr_base->clks.subvp_prefetch_fclk_khz);
+
+	unsigned int num_steps = 0;
+
+	int display_count;
+	bool fclk_p_state_change_support, uclk_p_state_change_support;
+
+	/* CLK_MGR401_UPDATE_NUM_DISPLAYS */
+	if (clk_mgr_internal->smu_present) {
+		display_count = clk_mgr_helper_get_active_display_cnt(dc, context);
+
+		if (display_count == 0)
+			enter_display_off = true;
+
+		if (enter_display_off == safe_to_lower) {
+			block_sequence[num_steps].params.update_num_displays_params.num_displays = display_count;
+			block_sequence[num_steps].func = CLK_MGR401_UPDATE_NUM_DISPLAYS;
+			num_steps++;
+		}
+	}
+
+	/* CLK_MGR401_UPDATE_FCLK_PSTATE_SUPPORT */
+	clk_mgr_base->clks.fclk_prev_p_state_change_support = clk_mgr_base->clks.fclk_p_state_change_support;
+	fclk_p_state_change_support = new_clocks->fclk_p_state_change_support || (total_plane_count == 0);
+	if (should_update_pstate_support(safe_to_lower, fclk_p_state_change_support, clk_mgr_base->clks.fclk_prev_p_state_change_support)) {
+		clk_mgr_base->clks.fclk_p_state_change_support = fclk_p_state_change_support;
+		update_active_fclk = true;
+		update_idle_fclk = true;
+
+		/* To enable FCLK P-state switching, send PSTATE_SUPPORTED message to PMFW (message not supported on DCN401)*/
+		// if (clk_mgr_base->clks.fclk_p_state_change_support) {
+		// 	/* Handle the code for sending a message to PMFW that FCLK P-state change is supported */
+		// 	if (dcn401_is_ppclk_dpm_enabled(clk_mgr_internal, PPCLK_FCLK)) {
+		// 		block_sequence[num_steps].params.update_pstate_support_params.support = true;
+		// 		block_sequence[num_steps].func = CLK_MGR401_UPDATE_FCLK_PSTATE_SUPPORT;
+		// 		num_steps++;
+		// 	}
+		// }
+	}
+
+	if (!clk_mgr_base->clks.fclk_p_state_change_support && dcn401_is_ppclk_dpm_enabled(clk_mgr_internal, PPCLK_FCLK)) {
+		/* when P-State switching disabled, set UCLK min = max */
+		idle_fclk_mhz =
+				clk_mgr_base->bw_params->clk_table.entries[clk_mgr_base->bw_params->clk_table.num_entries_per_clk.num_fclk_levels - 1].fclk_mhz;
+		active_fclk_mhz = idle_fclk_mhz;
+	}
+
+	/* UPDATE DCFCLK */
+	if (dc->debug.force_min_dcfclk_mhz > 0)
+		new_clocks->dcfclk_khz = (new_clocks->dcfclk_khz > (dc->debug.force_min_dcfclk_mhz * 1000)) ?
+				new_clocks->dcfclk_khz : (dc->debug.force_min_dcfclk_mhz * 1000);
+
+	if (should_set_clock(safe_to_lower, new_clocks->dcfclk_khz, clk_mgr_base->clks.dcfclk_khz)) {
+		clk_mgr_base->clks.dcfclk_khz = new_clocks->dcfclk_khz;
+		if (dcn401_is_ppclk_dpm_enabled(clk_mgr_internal, PPCLK_DCFCLK)) {
+			block_sequence[num_steps].params.update_hardmin_params.ppclk = PPCLK_DCFCLK;
+			block_sequence[num_steps].params.update_hardmin_params.freq_mhz = khz_to_mhz_ceil(clk_mgr_base->clks.dcfclk_khz);
+			block_sequence[num_steps].params.update_hardmin_params.response = NULL;
+			block_sequence[num_steps].func = CLK_MGR401_UPDATE_HARDMIN_PPCLK;
+			num_steps++;
+		}
+	}
+
+	/* CLK_MGR401_UPDATE_DEEP_SLEEP_DCFCLK */
+	if (should_set_clock(safe_to_lower, new_clocks->dcfclk_deep_sleep_khz, clk_mgr_base->clks.dcfclk_deep_sleep_khz)) {
+		clk_mgr_base->clks.dcfclk_deep_sleep_khz = new_clocks->dcfclk_deep_sleep_khz;
+		if (dcn401_is_ppclk_dpm_enabled(clk_mgr_internal, PPCLK_DCFCLK)) {
+			block_sequence[num_steps].params.update_deep_sleep_dcfclk_params.freq_mhz = khz_to_mhz_ceil(clk_mgr_base->clks.dcfclk_deep_sleep_khz);
+			block_sequence[num_steps].func = CLK_MGR401_UPDATE_DEEP_SLEEP_DCFCLK;
+			num_steps++;
+		}
+	}
+
+	/* SOCCLK */
+	if (should_set_clock(safe_to_lower, new_clocks->socclk_khz, clk_mgr_base->clks.socclk_khz))
+		/* We don't actually care about socclk, don't notify SMU of hard min */
+		clk_mgr_base->clks.socclk_khz = new_clocks->socclk_khz;
+
+	/* UCLK */
+	if (new_clocks->fw_based_mclk_switching != clk_mgr_base->clks.fw_based_mclk_switching &&
+			new_clocks->fw_based_mclk_switching) {
+		/* enable FAMS features */
+		clk_mgr_base->clks.fw_based_mclk_switching = new_clocks->fw_based_mclk_switching;
+
+		block_sequence[num_steps].params.update_wait_for_dmub_ack_params.enable = clk_mgr_base->clks.fw_based_mclk_switching;
+		block_sequence[num_steps].func = CLK_MGR401_UPDATE_WAIT_FOR_DMUB_ACK;
+		num_steps++;
+
+		block_sequence[num_steps].params.indicate_drr_status_params.mod_drr_for_pstate = clk_mgr_base->clks.fw_based_mclk_switching;
+		block_sequence[num_steps].func = CLK_MGR401_INDICATE_DRR_STATUS;
+		num_steps++;
+	}
+
+	/* CLK_MGR401_UPDATE_CAB_FOR_UCLK */
+	clk_mgr_base->clks.prev_num_ways = clk_mgr_base->clks.num_ways;
+	if (clk_mgr_base->clks.num_ways != new_clocks->num_ways &&
+			clk_mgr_base->clks.num_ways < new_clocks->num_ways) {
+		/* increase num ways for subvp */
+		clk_mgr_base->clks.num_ways = new_clocks->num_ways;
+		if (dcn401_is_ppclk_dpm_enabled(clk_mgr_internal, PPCLK_UCLK)) {
+			block_sequence[num_steps].params.update_cab_for_uclk_params.num_ways = clk_mgr_base->clks.num_ways;
+			block_sequence[num_steps].func = CLK_MGR401_UPDATE_CAB_FOR_UCLK;
+			num_steps++;
+		}
+	}
+
+	clk_mgr_base->clks.prev_p_state_change_support = clk_mgr_base->clks.p_state_change_support;
+	uclk_p_state_change_support = new_clocks->p_state_change_support || (total_plane_count == 0);
+	if (should_update_pstate_support(safe_to_lower, uclk_p_state_change_support, clk_mgr_base->clks.prev_p_state_change_support)) {
+		clk_mgr_base->clks.p_state_change_support = uclk_p_state_change_support;
+		update_active_uclk = true;
+		update_idle_uclk = true;
+
+		if (clk_mgr_base->clks.p_state_change_support) {
+			/* enable UCLK switching  */
+			if (dcn401_is_ppclk_dpm_enabled(clk_mgr_internal, PPCLK_UCLK)) {
+				block_sequence[num_steps].params.update_pstate_support_params.support = true;
+				block_sequence[num_steps].func = CLK_MGR401_UPDATE_UCLK_PSTATE_SUPPORT;
+				num_steps++;
+			}
+		}
+	}
+
+	if (!clk_mgr_base->clks.p_state_change_support && dcn401_is_ppclk_dpm_enabled(clk_mgr_internal, PPCLK_UCLK)) {
+		/* when P-State switching disabled, set UCLK min = max */
+		if (dc->clk_mgr->dc_mode_softmax_enabled) {
+			/* will never have the functional UCLK min above the softmax
+			* since we calculate mode support based on softmax being the max UCLK
+			* frequency.
+			*/
+			active_uclk_mhz = clk_mgr_base->bw_params->dc_mode_softmax_memclk;
+		} else {
+			active_uclk_mhz = clk_mgr_base->bw_params->max_memclk_mhz;
+		}
+		idle_uclk_mhz = active_uclk_mhz;
+	}
+
+	/* Always update saved value, even if new value not set due to P-State switching unsupported */
+	if (should_set_clock(safe_to_lower, new_clocks->dramclk_khz, clk_mgr_base->clks.dramclk_khz)) {
+		clk_mgr_base->clks.dramclk_khz = new_clocks->dramclk_khz;
+
+		if (clk_mgr_base->clks.p_state_change_support) {
+			update_active_uclk = true;
+			active_uclk_mhz = khz_to_mhz_ceil(clk_mgr_base->clks.dramclk_khz);
+		}
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->idle_dramclk_khz, clk_mgr_base->clks.idle_dramclk_khz)) {
+		clk_mgr_base->clks.idle_dramclk_khz = new_clocks->idle_dramclk_khz;
+
+		if (clk_mgr_base->clks.p_state_change_support) {
+			update_idle_uclk = true;
+			idle_uclk_mhz = khz_to_mhz_ceil(clk_mgr_base->clks.idle_dramclk_khz);
+		}
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->subvp_prefetch_dramclk_khz, clk_mgr_base->clks.subvp_prefetch_dramclk_khz)) {
+		clk_mgr_base->clks.subvp_prefetch_dramclk_khz = new_clocks->subvp_prefetch_dramclk_khz;
+		update_subvp_prefetch_dramclk = true;
+		subvp_prefetch_dramclk_mhz = khz_to_mhz_ceil(clk_mgr_base->clks.subvp_prefetch_dramclk_khz);
+	}
+
+	/* FCLK */
+	/* Always update saved value, even if new value not set due to P-State switching unsupported */
+	if (should_set_clock(safe_to_lower, new_clocks->fclk_khz, clk_mgr_base->clks.fclk_khz)) {
+		clk_mgr_base->clks.fclk_khz = new_clocks->fclk_khz;
+
+		if (clk_mgr_base->clks.fclk_p_state_change_support) {
+			update_active_fclk = true;
+			active_fclk_mhz = khz_to_mhz_ceil(clk_mgr_base->clks.fclk_khz);
+		}
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->idle_fclk_khz, clk_mgr_base->clks.idle_fclk_khz)) {
+		clk_mgr_base->clks.idle_fclk_khz = new_clocks->idle_fclk_khz;
+
+		if (clk_mgr_base->clks.fclk_p_state_change_support) {
+			update_idle_fclk = true;
+			idle_fclk_mhz = khz_to_mhz_ceil(clk_mgr_base->clks.idle_fclk_khz);
+		}
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->subvp_prefetch_fclk_khz, clk_mgr_base->clks.subvp_prefetch_fclk_khz)) {
+		clk_mgr_base->clks.subvp_prefetch_fclk_khz = new_clocks->subvp_prefetch_fclk_khz;
+		update_subvp_prefetch_fclk = true;
+		subvp_prefetch_fclk_mhz = khz_to_mhz_ceil(clk_mgr_base->clks.subvp_prefetch_fclk_khz);
+	}
+
+	/* When idle DPM is enabled, need to send active and idle hardmins separately */
+	/* CLK_MGR401_UPDATE_ACTIVE_HARDMINS */
+	if ((update_active_uclk || update_active_fclk) && is_idle_dpm_enabled) {
+		block_sequence[num_steps].params.update_idle_hardmin_params.uclk_mhz = active_uclk_mhz;
+		block_sequence[num_steps].params.update_idle_hardmin_params.fclk_mhz = active_fclk_mhz;
+		block_sequence[num_steps].func = CLK_MGR401_UPDATE_ACTIVE_HARDMINS;
+		num_steps++;
+	}
+
+	/* CLK_MGR401_UPDATE_IDLE_HARDMINS */
+	if ((update_idle_uclk || update_idle_fclk) && is_idle_dpm_enabled) {
+		block_sequence[num_steps].params.update_idle_hardmin_params.uclk_mhz = idle_uclk_mhz;
+		block_sequence[num_steps].params.update_idle_hardmin_params.fclk_mhz = idle_fclk_mhz;
+		block_sequence[num_steps].func = CLK_MGR401_UPDATE_IDLE_HARDMINS;
+		num_steps++;
+	}
+
+	/* CLK_MGR401_UPDATE_SUBVP_HARDMINS */
+	if ((update_subvp_prefetch_dramclk || update_subvp_prefetch_fclk) && is_df_throttle_opt_enabled) {
+		block_sequence[num_steps].params.update_idle_hardmin_params.uclk_mhz = subvp_prefetch_dramclk_mhz;
+		block_sequence[num_steps].params.update_idle_hardmin_params.fclk_mhz = subvp_prefetch_fclk_mhz;
+		block_sequence[num_steps].func = CLK_MGR401_UPDATE_SUBVP_HARDMINS;
+		num_steps++;
+	}
+
+	/* set UCLK to requested value if P-State switching is supported, or to re-enable P-State switching */
+	if (update_active_uclk || update_idle_uclk) {
+		if (!is_idle_dpm_enabled) {
+			block_sequence[num_steps].params.update_hardmin_params.ppclk = PPCLK_UCLK;
+			block_sequence[num_steps].params.update_hardmin_params.freq_mhz = active_uclk_mhz;
+			block_sequence[num_steps].params.update_hardmin_params.response = NULL;
+			block_sequence[num_steps].func = CLK_MGR401_UPDATE_HARDMIN_PPCLK;
+			num_steps++;
+		}
+
+		/* disable UCLK P-State support if needed */
+		if (!uclk_p_state_change_support &&
+				should_update_pstate_support(safe_to_lower, uclk_p_state_change_support, clk_mgr_base->clks.prev_p_state_change_support) &&
+				dcn401_is_ppclk_dpm_enabled(clk_mgr_internal, PPCLK_UCLK)) {
+			block_sequence[num_steps].params.update_pstate_support_params.support = false;
+			block_sequence[num_steps].func = CLK_MGR401_UPDATE_UCLK_PSTATE_SUPPORT;
+			num_steps++;
+		}
+	}
+
+	/* set FCLK to requested value if P-State switching is supported, or to re-enable P-State switching */
+	if (update_active_fclk || update_idle_fclk) {
+		/* No need to send active FCLK hardmin, automatically set based on DCFCLK */
+		// if (!is_idle_dpm_enabled) {
+		// 	block_sequence[*num_steps].update_hardmin_params.clk_mgr = clk_mgr;
+		// 	block_sequence[*num_steps].update_hardmin_params.ppclk = PPCLK_FCLK;
+		// 	block_sequence[*num_steps].update_hardmin_params.freq_mhz = active_fclk_mhz;
+		// 	block_sequence[*num_steps].update_hardmin_params.response = NULL;
+		// 	block_sequence[*num_steps].func = CLK_MGR401_UPDATE_HARDMIN_PPCLK;
+		// 	(*num_steps)++;
+		// }
+
+		/* disable FCLK P-State support if needed (message not supported on DCN401)*/
+		// if (!fclk_p_state_change_support &&
+		// 		should_update_pstate_support(safe_to_lower, fclk_p_state_change_support, clk_mgr_base->clks.fclk_prev_p_state_change_support) &&
+		// 		dcn401_is_ppclk_dpm_enabled(clk_mgr_internal, PPCLK_FCLK)) {
+		// 	block_sequence[num_steps].params.update_pstate_support_params.support = false;
+		// 	block_sequence[num_steps].func = CLK_MGR401_UPDATE_FCLK_PSTATE_SUPPORT;
+		// 	num_steps++;
+		// }
+	}
+
+	if (new_clocks->fw_based_mclk_switching != clk_mgr_base->clks.fw_based_mclk_switching &&
+			safe_to_lower && !new_clocks->fw_based_mclk_switching) {
+		/* disable FAMS features */
+		clk_mgr_base->clks.fw_based_mclk_switching = new_clocks->fw_based_mclk_switching;
+
+		block_sequence[num_steps].params.update_wait_for_dmub_ack_params.enable = clk_mgr_base->clks.fw_based_mclk_switching;
+		block_sequence[num_steps].func = CLK_MGR401_UPDATE_WAIT_FOR_DMUB_ACK;
+		num_steps++;
+
+		block_sequence[num_steps].params.indicate_drr_status_params.mod_drr_for_pstate = clk_mgr_base->clks.fw_based_mclk_switching;
+		block_sequence[num_steps].func = CLK_MGR401_INDICATE_DRR_STATUS;
+		num_steps++;
+	}
+
+	/* CLK_MGR401_UPDATE_CAB_FOR_UCLK */
+	if (clk_mgr_base->clks.num_ways != new_clocks->num_ways &&
+			safe_to_lower && clk_mgr_base->clks.num_ways > new_clocks->num_ways) {
+		/* decrease num ways for subvp */
+		clk_mgr_base->clks.num_ways = new_clocks->num_ways;
+		if (dcn401_is_ppclk_dpm_enabled(clk_mgr_internal, PPCLK_UCLK)) {
+			block_sequence[num_steps].params.update_cab_for_uclk_params.num_ways = clk_mgr_base->clks.num_ways;
+			block_sequence[num_steps].func = CLK_MGR401_UPDATE_CAB_FOR_UCLK;
+			num_steps++;
+		}
+	}
+
+	return num_steps;
+}
+
+static unsigned int dcn401_build_update_display_clocks_sequence(
+		struct clk_mgr *clk_mgr_base,
+		struct dc_state *context,
+		struct dc_clocks *new_clocks,
+		bool safe_to_lower)
+{
+	struct clk_mgr_internal *clk_mgr_internal = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct dcn401_clk_mgr *clk_mgr401 = TO_DCN401_CLK_MGR(clk_mgr_internal);
+	struct dc *dc = clk_mgr_base->ctx->dc;
+	struct dmcu *dmcu = clk_mgr_base->ctx->dc->res_pool->dmcu;
+	struct dcn401_clk_mgr_block_sequence *block_sequence = clk_mgr401->block_sequence;
+	bool force_reset = false;
+	bool update_dispclk = false;
+	bool update_dppclk = false;
+	bool dppclk_lowered = false;
+
+	unsigned int num_steps = 0;
+
+	/* CLK_MGR401_READ_CLOCKS_FROM_DENTIST */
+	if (clk_mgr_base->clks.dispclk_khz == 0 ||
+			(dc->debug.force_clock_mode & 0x1)) {
+		/* This is from resume or boot up, if forced_clock cfg option used,
+		 * we bypass program dispclk and DPPCLK, but need set them for S3.
+		 * Force_clock_mode 0x1:  force reset the clock even it is the same clock
+		 * as long as it is in Passive level.
+		 */
+		force_reset = true;
+
+		clk_mgr_base->clks.dispclk_khz = clk_mgr_base->boot_snapshot.dispclk;
+		clk_mgr_base->clks.actual_dispclk_khz = clk_mgr_base->clks.dispclk_khz;
+
+		clk_mgr_base->clks.dppclk_khz = clk_mgr_base->boot_snapshot.dppclk;
+		clk_mgr_base->clks.actual_dppclk_khz = clk_mgr_base->clks.dppclk_khz;
+	}
+
+	/* DTBCLK */
+	if (!new_clocks->dtbclk_en && dcn401_is_ppclk_dpm_enabled(clk_mgr_internal, PPCLK_DTBCLK)) {
+		new_clocks->ref_dtbclk_khz = clk_mgr_base->bw_params->clk_table.entries[0].dtbclk_mhz * 1000;
+	}
+
+	/* clock limits are received with MHz precision, divide by 1000 to prevent setting clocks at every call */
+	if (!dc->debug.disable_dtb_ref_clk_switch &&
+			should_set_clock(safe_to_lower, new_clocks->ref_dtbclk_khz / 1000, clk_mgr_base->clks.ref_dtbclk_khz / 1000) && //TODO these should be ceiled
+			dcn401_is_ppclk_dpm_enabled(clk_mgr_internal, PPCLK_DTBCLK)) {
+		/* DCCG requires KHz precision for DTBCLK */
+		block_sequence[num_steps].params.update_hardmin_params.ppclk = PPCLK_DTBCLK;
+		block_sequence[num_steps].params.update_hardmin_params.freq_mhz = khz_to_mhz_ceil(new_clocks->ref_dtbclk_khz);
+		block_sequence[num_steps].params.update_hardmin_params.response = &clk_mgr_base->clks.ref_dtbclk_khz;
+		block_sequence[num_steps].func = CLK_MGR401_UPDATE_HARDMIN_PPCLK;
+		num_steps++;
+
+		/* Update DTO in DCCG */
+		block_sequence[num_steps].params.update_dtbclk_dto_params.context = context;
+		block_sequence[num_steps].params.update_dtbclk_dto_params.ref_dtbclk_khz = &clk_mgr_base->clks.ref_dtbclk_khz;
+		block_sequence[num_steps].func = CLK_MGR401_UPDATE_DTBCLK_DTO;
+		num_steps++;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dppclk_khz, clk_mgr_base->clks.dppclk_khz)) {
+		if (clk_mgr_base->clks.dppclk_khz > new_clocks->dppclk_khz)
+			dppclk_lowered = true;
+
+		clk_mgr_base->clks.dppclk_khz = new_clocks->dppclk_khz;
+		clk_mgr_base->clks.actual_dppclk_khz = new_clocks->dppclk_khz;
+
+		update_dppclk = true;
+	}
+
+	if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, clk_mgr_base->clks.dispclk_khz)) {
+		clk_mgr_base->clks.dispclk_khz = new_clocks->dispclk_khz;
+
+		block_sequence[num_steps].params.update_hardmin_optimized_params.ppclk = PPCLK_DISPCLK;
+		block_sequence[num_steps].params.update_hardmin_optimized_params.freq_khz = clk_mgr_base->clks.dispclk_khz;
+		block_sequence[num_steps].params.update_hardmin_optimized_params.response = &clk_mgr_base->clks.actual_dispclk_khz;
+		block_sequence[num_steps].func = CLK_MGR401_UPDATE_HARDMIN_PPCLK_OPTIMIZED;
+		num_steps++;
+
+		update_dispclk = true;
+	}
+
+	if (dc->config.forced_clocks == false || (force_reset && safe_to_lower)) {
+		if (dppclk_lowered) {
+			/* if clock is being lowered, increase DTO before lowering refclk */
+			block_sequence[num_steps].params.update_dppclk_dto_params.context = context;
+			block_sequence[num_steps].params.update_dppclk_dto_params.ref_dppclk_khz = &clk_mgr_base->clks.dppclk_khz;
+			block_sequence[num_steps].params.update_dppclk_dto_params.safe_to_lower = safe_to_lower;
+			block_sequence[num_steps].func = CLK_MGR401_UPDATE_DPPCLK_DTO;
+			num_steps++;
+
+			block_sequence[num_steps].params.update_dentist_params.context = context;
+			block_sequence[num_steps].func = CLK_MGR401_UPDATE_DENTIST;
+			num_steps++;
+
+			if (dcn401_is_ppclk_dpm_enabled(clk_mgr_internal, PPCLK_DPPCLK)) {
+				block_sequence[num_steps].params.update_hardmin_optimized_params.ppclk = PPCLK_DPPCLK;
+				block_sequence[num_steps].params.update_hardmin_optimized_params.freq_khz = clk_mgr_base->clks.dppclk_khz;
+				block_sequence[num_steps].params.update_hardmin_optimized_params.response = &clk_mgr_base->clks.actual_dppclk_khz;
+				block_sequence[num_steps].func = CLK_MGR401_UPDATE_HARDMIN_PPCLK_OPTIMIZED;
+				num_steps++;
+
+				block_sequence[num_steps].params.update_dppclk_dto_params.context = context;
+				block_sequence[num_steps].params.update_dppclk_dto_params.ref_dppclk_khz = &clk_mgr_base->clks.actual_dppclk_khz;
+				block_sequence[num_steps].params.update_dppclk_dto_params.safe_to_lower = safe_to_lower;
+				block_sequence[num_steps].func = CLK_MGR401_UPDATE_DPPCLK_DTO;
+				num_steps++;
+			}
+		} else {
+			/* if clock is being raised, increase refclk before lowering DTO */
+			if (update_dppclk && dcn401_is_ppclk_dpm_enabled(clk_mgr_internal, PPCLK_DPPCLK)) {
+				block_sequence[num_steps].params.update_hardmin_optimized_params.ppclk = PPCLK_DPPCLK;
+				block_sequence[num_steps].params.update_hardmin_optimized_params.freq_khz = clk_mgr_base->clks.dppclk_khz;
+				block_sequence[num_steps].params.update_hardmin_optimized_params.response = &clk_mgr_base->clks.actual_dppclk_khz;
+				block_sequence[num_steps].func = CLK_MGR401_UPDATE_HARDMIN_PPCLK_OPTIMIZED;
+				num_steps++;
+			}
+
+			if (update_dppclk || update_dispclk) {
+				block_sequence[num_steps].params.update_dentist_params.context = context;
+				block_sequence[num_steps].func = CLK_MGR401_UPDATE_DENTIST;
+				num_steps++;
+			}
+
+			block_sequence[num_steps].params.update_dppclk_dto_params.context = context;
+			block_sequence[num_steps].params.update_dppclk_dto_params.ref_dppclk_khz = &clk_mgr_base->clks.actual_dppclk_khz;
+			block_sequence[num_steps].params.update_dppclk_dto_params.safe_to_lower = safe_to_lower;
+			block_sequence[num_steps].func = CLK_MGR401_UPDATE_DPPCLK_DTO;
+			num_steps++;
+		}
+	}
+
+	if (update_dispclk && dmcu && dmcu->funcs->is_dmcu_initialized(dmcu)) {
+		/*update dmcu for wait_loop count*/
+		block_sequence[num_steps].params.update_psr_wait_loop_params.dmcu = dmcu;
+		block_sequence[num_steps].params.update_psr_wait_loop_params.wait = clk_mgr_base->clks.dispclk_khz / 1000 / 7;
+		block_sequence[num_steps].func = CLK_MGR401_UPDATE_PSR_WAIT_LOOP;
+		num_steps++;
+	}
+
+	return num_steps;
+}
+
+static void dcn401_update_clocks(struct clk_mgr *clk_mgr_base,
+		struct dc_state *context,
+		bool safe_to_lower)
+{
+	struct dc *dc = clk_mgr_base->ctx->dc;
+
+	unsigned int num_steps = 0;
+
+	/* build bandwidth related clocks update sequence */
+	num_steps = dcn401_build_update_bandwidth_clocks_sequence(clk_mgr_base,
+			context,
+			&context->bw_ctx.bw.dcn.clk,
+			safe_to_lower);
+
+	/* execute sequence */
+	dcn401_execute_block_sequence(clk_mgr_base,	num_steps);
+
+	/* build display related clocks update sequence */
+	num_steps = dcn401_build_update_display_clocks_sequence(clk_mgr_base,
+			context,
+			&context->bw_ctx.bw.dcn.clk,
+			safe_to_lower);
+
+	/* execute sequence */
+	dcn401_execute_block_sequence(clk_mgr_base,	num_steps);
+
+	if (dc->config.enable_auto_dpm_test_logs)
+		dcn401_auto_dpm_test_log(&context->bw_ctx.bw.dcn.clk, TO_CLK_MGR_INTERNAL(clk_mgr_base), context);
+
+}
+
+
+static uint32_t dcn401_get_vco_frequency_from_reg(struct clk_mgr_internal *clk_mgr)
+{
+		struct fixed31_32 pll_req;
+		uint32_t pll_req_reg = 0;
+
+		/* get FbMult value */
+		pll_req_reg = REG_READ(CLK0_CLK_PLL_REQ);
+
+		/* set up a fixed-point number
+		 * this works because the int part is on the right edge of the register
+		 * and the frac part is on the left edge
+		 */
+		pll_req = dc_fixpt_from_int(pll_req_reg & clk_mgr->clk_mgr_mask->FbMult_int);
+		pll_req.value |= pll_req_reg & clk_mgr->clk_mgr_mask->FbMult_frac;
+
+		/* multiply by REFCLK period */
+		pll_req = dc_fixpt_mul_int(pll_req, clk_mgr->dfs_ref_freq_khz);
+
+		return dc_fixpt_floor(pll_req);
+}
+
+static void dcn401_clock_read_ss_info(struct clk_mgr_internal *clk_mgr)
+{
+	struct dc_bios *bp = clk_mgr->base.ctx->dc_bios;
+	int ss_info_num = bp->funcs->get_ss_entry_number(
+			bp, AS_SIGNAL_TYPE_GPU_PLL);
+
+	if (ss_info_num) {
+		struct spread_spectrum_info info = { { 0 } };
+		enum bp_result result = bp->funcs->get_spread_spectrum_info(
+				bp, AS_SIGNAL_TYPE_GPU_PLL, 0, &info);
+
+		/* SSInfo.spreadSpectrumPercentage !=0 would be sign
+		 * that SS is enabled
+		 */
+		if (result == BP_RESULT_OK &&
+				info.spread_spectrum_percentage != 0) {
+			clk_mgr->ss_on_dprefclk = true;
+			clk_mgr->dprefclk_ss_divider = info.spread_percentage_divider;
+
+			if (info.type.CENTER_MODE == 0) {
+				/* Currently for DP Reference clock we
+				 * need only SS percentage for
+				 * downspread
+				 */
+				clk_mgr->dprefclk_ss_percentage =
+						info.spread_spectrum_percentage;
+			}
+		}
+	}
+}
+static void dcn401_notify_wm_ranges(struct clk_mgr *clk_mgr_base)
+{
+	unsigned int i;
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	WatermarksExternal_t *table = (WatermarksExternal_t *) clk_mgr->wm_range_table;
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	if (!table)
+		return;
+
+	memset(table, 0, sizeof(*table));
+
+	/* collect valid ranges, place in pmfw table */
+	for (i = 0; i < WM_SET_COUNT; i++)
+		if (clk_mgr->base.bw_params->wm_table.nv_entries[i].valid) {
+			table->Watermarks.WatermarkRow[i].WmSetting = i;
+			table->Watermarks.WatermarkRow[i].Flags = clk_mgr->base.bw_params->wm_table.nv_entries[i].pmfw_breakdown.wm_type;
+		}
+	dcn401_smu_set_dram_addr_high(clk_mgr, clk_mgr->wm_range_table_addr >> 32);
+	dcn401_smu_set_dram_addr_low(clk_mgr, clk_mgr->wm_range_table_addr & 0xFFFFFFFF);
+	dcn401_smu_transfer_wm_table_dram_2_smu(clk_mgr);
+}
+
+/* Set min memclk to minimum, either constrained by the current mode or DPM0 */
+static void dcn401_set_hard_min_memclk(struct clk_mgr *clk_mgr_base, bool current_mode)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	const struct dc *dc = clk_mgr->base.ctx->dc;
+	struct dc_state *context = dc->current_state;
+	struct dc_clocks new_clocks;
+	int num_steps;
+
+	if (!clk_mgr->smu_present || !dcn401_is_ppclk_dpm_enabled(clk_mgr, PPCLK_UCLK))
+		return;
+
+	/* build clock update */
+	memcpy(&new_clocks, &clk_mgr_base->clks, sizeof(struct dc_clocks));
+
+	if (current_mode) {
+		new_clocks.dramclk_khz = context->bw_ctx.bw.dcn.clk.dramclk_khz;
+		new_clocks.idle_dramclk_khz = context->bw_ctx.bw.dcn.clk.idle_dramclk_khz;
+		new_clocks.p_state_change_support = context->bw_ctx.bw.dcn.clk.p_state_change_support;
+	} else {
+		new_clocks.dramclk_khz = clk_mgr_base->bw_params->clk_table.entries[0].memclk_mhz * 1000;
+		new_clocks.idle_dramclk_khz = new_clocks.dramclk_khz;
+		new_clocks.p_state_change_support = true;
+	}
+
+	num_steps = dcn401_build_update_bandwidth_clocks_sequence(clk_mgr_base,
+			context,
+			&new_clocks,
+			true);
+
+	/* execute sequence */
+	dcn401_execute_block_sequence(clk_mgr_base,	num_steps);
+}
+
+static int dcn401_get_hard_min_memclk(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	return clk_mgr->base.ctx->dc->current_state->bw_ctx.bw.dcn.clk.dramclk_khz;
+}
+
+static int dcn401_get_hard_min_fclk(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	return clk_mgr->base.ctx->dc->current_state->bw_ctx.bw.dcn.clk.fclk_khz;
+}
+
+/* Get current memclk states, update bounding box */
+static void dcn401_get_memclk_states_from_smu(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	struct clk_limit_num_entries *num_entries_per_clk = &clk_mgr_base->bw_params->clk_table.num_entries_per_clk;
+	unsigned int num_levels;
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	/* Refresh memclk and fclk states */
+	dcn401_init_single_clock(clk_mgr, PPCLK_UCLK,
+			&clk_mgr_base->bw_params->clk_table.entries[0].memclk_mhz,
+			&num_entries_per_clk->num_memclk_levels);
+	if (num_entries_per_clk->num_memclk_levels) {
+		clk_mgr_base->bw_params->max_memclk_mhz =
+				clk_mgr_base->bw_params->clk_table.entries[num_entries_per_clk->num_memclk_levels - 1].memclk_mhz;
+	}
+
+	clk_mgr_base->bw_params->dc_mode_limit.memclk_mhz = dcn401_smu_get_dc_mode_max_dpm_freq(clk_mgr, PPCLK_UCLK);
+	if (num_entries_per_clk->num_memclk_levels && clk_mgr_base->bw_params->dc_mode_limit.memclk_mhz ==
+			clk_mgr_base->bw_params->clk_table.entries[num_entries_per_clk->num_memclk_levels - 1].memclk_mhz)
+		clk_mgr_base->bw_params->dc_mode_limit.memclk_mhz = 0;
+	clk_mgr_base->bw_params->dc_mode_softmax_memclk = clk_mgr_base->bw_params->dc_mode_limit.memclk_mhz;
+
+	dcn401_init_single_clock(clk_mgr, PPCLK_FCLK,
+			&clk_mgr_base->bw_params->clk_table.entries[0].fclk_mhz,
+			&num_entries_per_clk->num_fclk_levels);
+	clk_mgr_base->bw_params->dc_mode_limit.fclk_mhz = dcn401_smu_get_dc_mode_max_dpm_freq(clk_mgr, PPCLK_FCLK);
+	if (num_entries_per_clk->num_fclk_levels && clk_mgr_base->bw_params->dc_mode_limit.fclk_mhz ==
+			clk_mgr_base->bw_params->clk_table.entries[num_entries_per_clk->num_fclk_levels - 1].fclk_mhz)
+		clk_mgr_base->bw_params->dc_mode_limit.fclk_mhz = 0;
+
+	num_levels = max(num_entries_per_clk->num_memclk_levels, num_entries_per_clk->num_fclk_levels);
+
+	clk_mgr_base->bw_params->clk_table.num_entries = num_levels ? num_levels : 1;
+
+	if (clk_mgr->dpm_present && !num_levels)
+		clk_mgr->dpm_present = false;
+
+	clk_mgr_base->bw_params->num_channels = dcn401_smu_get_num_of_umc_channels(clk_mgr);
+	if (clk_mgr_base->ctx->dc_bios) {
+		/* use BIOS values if none provided by PMFW */
+		if (clk_mgr_base->bw_params->num_channels == 0) {
+			clk_mgr_base->bw_params->num_channels = clk_mgr_base->ctx->dc_bios->vram_info.num_chans;
+		}
+		clk_mgr_base->bw_params->dram_channel_width_bytes = clk_mgr_base->ctx->dc_bios->vram_info.dram_channel_width_bytes;
+	}
+
+	/* Refresh bounding box */
+	clk_mgr_base->ctx->dc->res_pool->funcs->update_bw_bounding_box(
+			clk_mgr->base.ctx->dc, clk_mgr_base->bw_params);
+}
+
+static bool dcn401_are_clock_states_equal(struct dc_clocks *a,
+					struct dc_clocks *b)
+{
+	if (a->dispclk_khz != b->dispclk_khz)
+		return false;
+	else if (a->dppclk_khz != b->dppclk_khz)
+		return false;
+	else if (a->dcfclk_khz != b->dcfclk_khz)
+		return false;
+	else if (a->dcfclk_deep_sleep_khz != b->dcfclk_deep_sleep_khz)
+		return false;
+	else if (a->dramclk_khz != b->dramclk_khz)
+		return false;
+	else if (a->p_state_change_support != b->p_state_change_support)
+		return false;
+	else if (a->fclk_p_state_change_support != b->fclk_p_state_change_support)
+		return false;
+
+	return true;
+}
+
+static void dcn401_enable_pme_wa(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	if (!clk_mgr->smu_present)
+		return;
+
+	dcn401_smu_set_pme_workaround(clk_mgr);
+}
+
+static bool dcn401_is_smu_present(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	return clk_mgr->smu_present;
+}
+
+
+static int dcn401_get_dtb_ref_freq_khz(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	int dtb_ref_clk_khz = 0;
+
+	if (clk_mgr->smu_present && dcn401_is_ppclk_dpm_enabled(clk_mgr, PPCLK_DTBCLK)) {
+		/* DPM enabled, use currently set value */
+		dtb_ref_clk_khz = clk_mgr_base->clks.ref_dtbclk_khz;
+	} else {
+		/* DPM disabled, so use boot snapshot */
+		dtb_ref_clk_khz = clk_mgr_base->boot_snapshot.dtbclk;
+	}
+
+	return dtb_ref_clk_khz;
+}
+
+static int dcn401_get_dispclk_from_dentist(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	uint32_t dispclk_wdivider;
+	int disp_divider;
+
+	REG_GET(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_WDIVIDER, &dispclk_wdivider);
+	disp_divider = dentist_get_divider_from_did(dispclk_wdivider);
+
+	/* Return DISPCLK freq in Khz */
+	if (disp_divider)
+		return (DENTIST_DIVIDER_RANGE_SCALE_FACTOR * clk_mgr->base.dentist_vco_freq_khz) / disp_divider;
+
+	return 0;
+}
+
+unsigned int dcn401_get_max_clock_khz(struct clk_mgr *clk_mgr_base, enum clk_type clk_type)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+
+	unsigned int num_clk_levels;
+
+	switch (clk_type) {
+	case CLK_TYPE_DISPCLK:
+		num_clk_levels = clk_mgr->base.bw_params->clk_table.num_entries_per_clk.num_dispclk_levels;
+		return dcn401_is_ppclk_dpm_enabled(clk_mgr, PPCLK_DISPCLK) ?
+				clk_mgr->base.bw_params->clk_table.entries[num_clk_levels - 1].dispclk_mhz * 1000 :
+				clk_mgr->base.boot_snapshot.dispclk;
+	case CLK_TYPE_DPPCLK:
+		num_clk_levels = clk_mgr->base.bw_params->clk_table.num_entries_per_clk.num_dppclk_levels;
+		return dcn401_is_ppclk_dpm_enabled(clk_mgr, PPCLK_DPPCLK) ?
+				clk_mgr->base.bw_params->clk_table.entries[num_clk_levels - 1].dppclk_mhz * 1000 :
+				clk_mgr->base.boot_snapshot.dppclk;
+	case CLK_TYPE_DSCCLK:
+		num_clk_levels = clk_mgr->base.bw_params->clk_table.num_entries_per_clk.num_dispclk_levels;
+		return dcn401_is_ppclk_dpm_enabled(clk_mgr, PPCLK_DISPCLK) ?
+				clk_mgr->base.bw_params->clk_table.entries[num_clk_levels - 1].dispclk_mhz * 1000 / 3 :
+				clk_mgr->base.boot_snapshot.dispclk / 3;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static struct clk_mgr_funcs dcn401_funcs = {
+		.get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz,
+		.get_dtb_ref_clk_frequency = dcn401_get_dtb_ref_freq_khz,
+		.update_clocks = dcn401_update_clocks,
+		.dump_clk_registers = dcn401_dump_clk_registers,
+		.init_clocks = dcn401_init_clocks,
+		.notify_wm_ranges = dcn401_notify_wm_ranges,
+		.set_hard_min_memclk = dcn401_set_hard_min_memclk,
+		.get_memclk_states_from_smu = dcn401_get_memclk_states_from_smu,
+		.are_clock_states_equal = dcn401_are_clock_states_equal,
+		.enable_pme_wa = dcn401_enable_pme_wa,
+		.is_smu_present = dcn401_is_smu_present,
+		.get_dispclk_from_dentist = dcn401_get_dispclk_from_dentist,
+		.get_hard_min_memclk = dcn401_get_hard_min_memclk,
+		.get_hard_min_fclk = dcn401_get_hard_min_fclk,
+		.is_dc_mode_present = dcn401_is_dc_mode_present,
+		.get_max_clock_khz = dcn401_get_max_clock_khz,
+};
+
+struct clk_mgr_internal *dcn401_clk_mgr_construct(
+		struct dc_context *ctx,
+		struct dccg *dccg)
+{
+	struct clk_log_info log_info = {0};
+	struct dcn401_clk_mgr *clk_mgr401 = kzalloc(sizeof(struct dcn401_clk_mgr), GFP_KERNEL);
+	struct clk_mgr_internal *clk_mgr;
+
+	if (!clk_mgr401)
+		return NULL;
+
+	clk_mgr = &clk_mgr401->base;
+	clk_mgr->base.ctx = ctx;
+	clk_mgr->base.funcs = &dcn401_funcs;
+	clk_mgr->regs = &clk_mgr_regs_dcn401;
+	clk_mgr->clk_mgr_shift = &clk_mgr_shift_dcn401;
+	clk_mgr->clk_mgr_mask = &clk_mgr_mask_dcn401;
+
+	clk_mgr->dccg = dccg;
+	clk_mgr->dfs_bypass_disp_clk = 0;
+
+	clk_mgr->dprefclk_ss_percentage = 0;
+	clk_mgr->dprefclk_ss_divider = 1000;
+	clk_mgr->ss_on_dprefclk = false;
+	clk_mgr->dfs_ref_freq_khz = 100000;
+
+	/* Changed from DCN3.2_clock_frequency doc to match
+	 * dcn401_dump_clk_registers from 4 * dentist_vco_freq_khz /
+	 * dprefclk DID divider
+	 */
+	clk_mgr->base.dprefclk_khz = 720000; //TODO update from VBIOS
+
+	/* integer part is now VCO frequency in kHz */
+	clk_mgr->base.dentist_vco_freq_khz = dcn401_get_vco_frequency_from_reg(clk_mgr);
+
+	/* in case we don't get a value from the register, use default */
+	if (clk_mgr->base.dentist_vco_freq_khz == 0)
+		clk_mgr->base.dentist_vco_freq_khz = 4500000; //TODO Update from VBIOS
+
+	dcn401_dump_clk_registers(&clk_mgr->base.boot_snapshot, &clk_mgr->base, &log_info);
+
+	if (ctx->dc->debug.disable_dtb_ref_clk_switch &&
+			clk_mgr->base.clks.ref_dtbclk_khz != clk_mgr->base.boot_snapshot.dtbclk) {
+		clk_mgr->base.clks.ref_dtbclk_khz = clk_mgr->base.boot_snapshot.dtbclk;
+	}
+
+	if (clk_mgr->base.boot_snapshot.dprefclk != 0) {
+		clk_mgr->base.dprefclk_khz = clk_mgr->base.boot_snapshot.dprefclk;
+	}
+	dcn401_clock_read_ss_info(clk_mgr);
+
+	clk_mgr->dfs_bypass_enabled = false;
+
+	clk_mgr->smu_present = false;
+
+	clk_mgr->base.bw_params = kzalloc(sizeof(*clk_mgr->base.bw_params), GFP_KERNEL);
+	if (!clk_mgr->base.bw_params) {
+		BREAK_TO_DEBUGGER();
+		kfree(clk_mgr401);
+		return NULL;
+	}
+
+	/* need physical address of table to give to PMFW */
+	clk_mgr->wm_range_table = dm_helpers_allocate_gpu_mem(clk_mgr->base.ctx,
+			DC_MEM_ALLOC_TYPE_GART, sizeof(WatermarksExternal_t),
+			&clk_mgr->wm_range_table_addr);
+	if (!clk_mgr->wm_range_table) {
+		BREAK_TO_DEBUGGER();
+		kfree(clk_mgr->base.bw_params);
+		kfree(clk_mgr401);
+		return NULL;
+	}
+
+	return &clk_mgr401->base;
+}
+
+void dcn401_clk_mgr_destroy(struct clk_mgr_internal *clk_mgr)
+{
+	kfree(clk_mgr->base.bw_params);
+
+	if (clk_mgr->wm_range_table)
+		dm_helpers_free_gpu_mem(clk_mgr->base.ctx, DC_MEM_ALLOC_TYPE_GART,
+				clk_mgr->wm_range_table);
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr.h
new file mode 100644
index 000000000000..97a1ce1e8a9e
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr.h
@@ -0,0 +1,117 @@
+// SPDX-License-Identifier: MIT
+//
+// Copyright 2024 Advanced Micro Devices, Inc.
+
+#ifndef __DCN401_CLK_MGR_H_
+#define __DCN401_CLK_MGR_H_
+
+#define DCN401_CLK_MGR_MAX_SEQUENCE_SIZE 30
+
+union dcn401_clk_mgr_block_sequence_params {
+	struct {
+		/* inputs */
+		uint32_t num_displays;
+	} update_num_displays_params;
+	struct {
+		/* inputs */
+		uint32_t ppclk;
+		uint16_t freq_mhz;
+		/* outputs */
+		uint32_t *response;
+	} update_hardmin_params;
+	struct {
+		/* inputs */
+		uint32_t ppclk;
+		int freq_khz;
+		/* outputs */
+		uint32_t *response;
+	} update_hardmin_optimized_params;
+	struct {
+		/* inputs */
+		uint16_t uclk_mhz;
+		uint16_t fclk_mhz;
+	} update_idle_hardmin_params;
+	struct {
+		/* inputs */
+		uint16_t freq_mhz;
+	} update_deep_sleep_dcfclk_params;
+	struct {
+		/* inputs */
+		bool support;
+	} update_pstate_support_params;
+	struct {
+		/* inputs */
+		unsigned int num_ways;
+	} update_cab_for_uclk_params;
+	struct {
+		/* inputs */
+		bool enable;
+	} update_wait_for_dmub_ack_params;
+	struct {
+		/* inputs */
+		bool mod_drr_for_pstate;
+	} indicate_drr_status_params;
+	struct {
+		/* inputs */
+		struct dc_state *context;
+		int *ref_dppclk_khz;
+		bool safe_to_lower;
+	} update_dppclk_dto_params;
+	struct {
+		/* inputs */
+		struct dc_state *context;
+		int *ref_dtbclk_khz;
+	} update_dtbclk_dto_params;
+	struct {
+		/* inputs */
+		struct dc_state *context;
+	} update_dentist_params;
+	struct {
+		/* inputs */
+		struct dmcu *dmcu;
+		unsigned int wait;
+	} update_psr_wait_loop_params;
+};
+
+enum dcn401_clk_mgr_block_sequence_func {
+	CLK_MGR401_READ_CLOCKS_FROM_DENTIST,
+	CLK_MGR401_UPDATE_NUM_DISPLAYS,
+	CLK_MGR401_UPDATE_HARDMIN_PPCLK,
+	CLK_MGR401_UPDATE_HARDMIN_PPCLK_OPTIMIZED,
+	CLK_MGR401_UPDATE_ACTIVE_HARDMINS,
+	CLK_MGR401_UPDATE_IDLE_HARDMINS,
+	CLK_MGR401_UPDATE_DEEP_SLEEP_DCFCLK,
+	CLK_MGR401_UPDATE_FCLK_PSTATE_SUPPORT,
+	CLK_MGR401_UPDATE_UCLK_PSTATE_SUPPORT,
+	CLK_MGR401_UPDATE_CAB_FOR_UCLK,
+	CLK_MGR401_UPDATE_WAIT_FOR_DMUB_ACK,
+	CLK_MGR401_INDICATE_DRR_STATUS,
+	CLK_MGR401_UPDATE_DPPCLK_DTO,
+	CLK_MGR401_UPDATE_DTBCLK_DTO,
+	CLK_MGR401_UPDATE_DENTIST,
+	CLK_MGR401_UPDATE_PSR_WAIT_LOOP,
+	CLK_MGR401_UPDATE_SUBVP_HARDMINS,
+};
+
+struct dcn401_clk_mgr_block_sequence {
+	union dcn401_clk_mgr_block_sequence_params params;
+	enum dcn401_clk_mgr_block_sequence_func func;
+};
+
+struct dcn401_clk_mgr {
+	struct clk_mgr_internal base;
+
+	struct dcn401_clk_mgr_block_sequence block_sequence[DCN401_CLK_MGR_MAX_SEQUENCE_SIZE];
+};
+
+void dcn401_init_clocks(struct clk_mgr *clk_mgr_base);
+bool dcn401_is_dc_mode_present(struct clk_mgr *clk_mgr_base);
+
+struct clk_mgr_internal *dcn401_clk_mgr_construct(struct dc_context *ctx,
+		struct dccg *dccg);
+
+void dcn401_clk_mgr_destroy(struct clk_mgr_internal *clk_mgr);
+
+unsigned int dcn401_get_max_clock_khz(struct clk_mgr *clk_mgr_base, enum clk_type clk_type);
+
+#endif /* __DCN401_CLK_MGR_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr_smu_msg.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr_smu_msg.c
new file mode 100644
index 000000000000..3a263840893e
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr_smu_msg.c
@@ -0,0 +1,472 @@
+// SPDX-License-Identifier: MIT
+//
+// Copyright 2024 Advanced Micro Devices, Inc.
+
+#include "dcn401_clk_mgr_smu_msg.h"
+
+#include "clk_mgr_internal.h"
+#include "reg_helper.h"
+
+#include "dalsmc.h"
+#include "dcn401_smu14_driver_if.h"
+
+#define mmDAL_MSG_REG  0x1628A
+#define mmDAL_ARG_REG  0x16273
+#define mmDAL_RESP_REG 0x16274
+
+#define REG(reg_name) \
+	mm ## reg_name
+
+#include "logger_types.h"
+
+#define smu_print(str, ...) {DC_LOG_SMU(str, ##__VA_ARGS__); }
+
+/* temporary define */
+#ifndef DALSMC_MSG_SubvpUclkFclk
+#define DALSMC_MSG_SubvpUclkFclk 0x1B
+#endif
+#ifndef DALSMC_MSG_GetNumUmcChannels
+#define DALSMC_MSG_GetNumUmcChannels 0x1C
+#endif
+
+/*
+ * Function to be used instead of REG_WAIT macro because the wait ends when
+ * the register is NOT EQUAL to zero, and because the translation in msg_if.h
+ * won't work with REG_WAIT.
+ */
+static uint32_t dcn401_smu_wait_for_response(struct clk_mgr_internal *clk_mgr, unsigned int delay_us, unsigned int max_retries)
+{
+	uint32_t reg = 0;
+
+	do {
+		reg = REG_READ(DAL_RESP_REG);
+		if (reg)
+			break;
+
+		if (delay_us >= 1000)
+			msleep(delay_us/1000);
+		else if (delay_us > 0)
+			udelay(delay_us);
+	} while (max_retries--);
+
+	return reg;
+}
+
+static bool dcn401_smu_send_msg_with_param(struct clk_mgr_internal *clk_mgr, uint32_t msg_id, uint32_t param_in, uint32_t *param_out)
+{
+	/* Wait for response register to be ready */
+	dcn401_smu_wait_for_response(clk_mgr, 10, 200000);
+
+	TRACE_SMU_MSG_ENTER(msg_id, param_in, clk_mgr->base.ctx);
+
+	/* Clear response register */
+	REG_WRITE(DAL_RESP_REG, 0);
+
+	/* Set the parameter register for the SMU message */
+	REG_WRITE(DAL_ARG_REG, param_in);
+
+	/* Trigger the message transaction by writing the message ID */
+	REG_WRITE(DAL_MSG_REG, msg_id);
+
+	/* Wait for response */
+	if (dcn401_smu_wait_for_response(clk_mgr, 10, 200000) == DALSMC_Result_OK) {
+		if (param_out)
+			*param_out = REG_READ(DAL_ARG_REG);
+
+		TRACE_SMU_MSG_EXIT(true, param_out ? *param_out : 0, clk_mgr->base.ctx);
+		return true;
+	}
+
+	TRACE_SMU_MSG_EXIT(false, 0, clk_mgr->base.ctx);
+	return false;
+}
+
+/*
+ * Use these functions to return back delay information so we can aggregate the total
+ *  delay when requesting hardmin clk
+ *
+ * dcn401_smu_wait_for_response_delay
+ * dcn401_smu_send_msg_with_param_delay
+ *
+ */
+static uint32_t dcn401_smu_wait_for_response_delay(struct clk_mgr_internal *clk_mgr, unsigned int delay_us, unsigned int max_retries, unsigned int *total_delay_us)
+{
+	uint32_t reg = 0;
+	*total_delay_us = 0;
+
+	do {
+		reg = REG_READ(DAL_RESP_REG);
+		if (reg)
+			break;
+
+		if (delay_us >= 1000)
+			msleep(delay_us/1000);
+		else if (delay_us > 0)
+			udelay(delay_us);
+		*total_delay_us += delay_us;
+	} while (max_retries--);
+
+	return reg;
+}
+
+static bool dcn401_smu_send_msg_with_param_delay(struct clk_mgr_internal *clk_mgr, uint32_t msg_id, uint32_t param_in, uint32_t *param_out, unsigned int *total_delay_us)
+{
+	unsigned int delay1_us, delay2_us;
+	*total_delay_us = 0;
+
+	/* Wait for response register to be ready */
+	dcn401_smu_wait_for_response_delay(clk_mgr, 10, 200000, &delay1_us);
+
+	TRACE_SMU_MSG_ENTER(msg_id, param_in, clk_mgr->base.ctx);
+
+	/* Clear response register */
+	REG_WRITE(DAL_RESP_REG, 0);
+
+	/* Set the parameter register for the SMU message */
+	REG_WRITE(DAL_ARG_REG, param_in);
+
+	/* Trigger the message transaction by writing the message ID */
+	REG_WRITE(DAL_MSG_REG, msg_id);
+
+	/* Wait for response */
+	if (dcn401_smu_wait_for_response_delay(clk_mgr, 10, 200000, &delay2_us) == DALSMC_Result_OK) {
+		if (param_out)
+			*param_out = REG_READ(DAL_ARG_REG);
+
+		*total_delay_us = delay1_us + delay2_us;
+		TRACE_SMU_MSG_EXIT(true, param_out ? *param_out : 0, clk_mgr->base.ctx);
+		return true;
+	}
+
+	*total_delay_us = delay1_us + 2000000;
+	TRACE_SMU_MSG_EXIT(false, 0, clk_mgr->base.ctx);
+	return false;
+}
+
+bool dcn401_smu_get_smu_version(struct clk_mgr_internal *clk_mgr, unsigned int *version)
+{
+	smu_print("SMU Get SMU version\n");
+
+	if (dcn401_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_GetSmuVersion, 0, version)) {
+
+		smu_print("SMU version: %d\n", *version);
+
+		return true;
+	}
+
+	return false;
+}
+
+/* Message output should match SMU11_DRIVER_IF_VERSION in smu11_driver_if.h */
+bool dcn401_smu_check_driver_if_version(struct clk_mgr_internal *clk_mgr)
+{
+	uint32_t response = 0;
+
+	smu_print("SMU Check driver if version\n");
+
+	if (dcn401_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_GetDriverIfVersion, 0, &response)) {
+
+		smu_print("SMU driver if version: %d\n", response);
+
+		if (response == SMU14_DRIVER_IF_VERSION)
+			return true;
+	}
+
+	return false;
+}
+
+/* Message output should match DALSMC_VERSION in dalsmc.h */
+bool dcn401_smu_check_msg_header_version(struct clk_mgr_internal *clk_mgr)
+{
+	uint32_t response = 0;
+
+	smu_print("SMU Check msg header version\n");
+
+	if (dcn401_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_GetMsgHeaderVersion, 0, &response)) {
+
+		smu_print("SMU msg header version: %d\n", response);
+
+		if (response == DALSMC_VERSION)
+			return true;
+	}
+
+	return false;
+}
+
+void dcn401_smu_send_fclk_pstate_message(struct clk_mgr_internal *clk_mgr, bool support)
+{
+	smu_print("FCLK P-state support value is : %d\n", support);
+
+	dcn401_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_SetFclkSwitchAllow, support, NULL);
+}
+
+void dcn401_smu_send_uclk_pstate_message(struct clk_mgr_internal *clk_mgr, bool support)
+{
+	smu_print("UCLK P-state support value is : %d\n", support);
+
+	dcn401_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_SetUclkPstateAllow, support, NULL);
+}
+
+void dcn401_smu_send_cab_for_uclk_message(struct clk_mgr_internal *clk_mgr, unsigned int num_ways)
+{
+	uint32_t param = (num_ways << 1) | (num_ways > 0);
+
+	dcn401_smu_send_msg_with_param(clk_mgr, DALSMC_MSG_SetCabForUclkPstate, param, NULL);
+	smu_print("Numways for SubVP : %d\n", num_ways);
+}
+
+void dcn401_smu_set_dram_addr_high(struct clk_mgr_internal *clk_mgr, uint32_t addr_high)
+{
+	smu_print("SMU Set DRAM addr high: %d\n", addr_high);
+
+	dcn401_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_SetDalDramAddrHigh, addr_high, NULL);
+}
+
+void dcn401_smu_set_dram_addr_low(struct clk_mgr_internal *clk_mgr, uint32_t addr_low)
+{
+	smu_print("SMU Set DRAM addr low: %d\n", addr_low);
+
+	dcn401_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_SetDalDramAddrLow, addr_low, NULL);
+}
+
+void dcn401_smu_transfer_wm_table_dram_2_smu(struct clk_mgr_internal *clk_mgr)
+{
+	smu_print("SMU Transfer WM table DRAM 2 SMU\n");
+
+	dcn401_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_TransferTableDram2Smu, TABLE_WATERMARKS, NULL);
+}
+
+void dcn401_smu_set_pme_workaround(struct clk_mgr_internal *clk_mgr)
+{
+	smu_print("SMU Set PME workaround\n");
+
+	dcn401_smu_send_msg_with_param(clk_mgr,
+		DALSMC_MSG_BacoAudioD3PME, 0, NULL);
+}
+
+static unsigned int dcn401_smu_get_hard_min_status(struct clk_mgr_internal *clk_mgr, bool *no_timeout, unsigned int *total_delay_us)
+{
+	uint32_t response = 0;
+
+	/* bits 23:16 for clock type, lower 16 bits for frequency in MHz */
+	uint32_t param = 0;
+
+	*no_timeout = dcn401_smu_send_msg_with_param_delay(clk_mgr,
+			DALSMC_MSG_ReturnHardMinStatus, param, &response, total_delay_us);
+
+	smu_print("SMU Get hard min status: no_timeout %d delay %d us clk bits %x\n",
+		*no_timeout, *total_delay_us, response);
+
+	return response;
+}
+
+static bool dcn401_smu_wait_hard_min_status(struct clk_mgr_internal *clk_mgr, uint32_t ppclk)
+{
+	const unsigned int max_delay_us = 1000000;
+
+	unsigned int hardmin_status_mask = (1 << ppclk);
+	unsigned int total_delay_us = 0;
+	bool hardmin_done = false;
+
+	while (!hardmin_done && total_delay_us < max_delay_us) {
+		unsigned int hardmin_status;
+		unsigned int read_total_delay_us;
+		bool no_timeout;
+
+		if (!hardmin_done && total_delay_us > 0) {
+			/* hardmin not yet fulfilled, wait 500us and retry*/
+			udelay(500);
+			total_delay_us += 500;
+
+			smu_print("SMU Wait hard min status for %d us\n", total_delay_us);
+		}
+
+		hardmin_status = dcn401_smu_get_hard_min_status(clk_mgr, &no_timeout, &read_total_delay_us);
+		total_delay_us += read_total_delay_us;
+		hardmin_done = hardmin_status & hardmin_status_mask;
+	}
+
+	return hardmin_done;
+}
+
+/* Returns the actual frequency that was set in MHz, 0 on failure */
+unsigned int dcn401_smu_set_hard_min_by_freq(struct clk_mgr_internal *clk_mgr, uint32_t clk, uint16_t freq_mhz)
+{
+	uint32_t response = 0;
+	bool hard_min_done = false;
+
+	/* bits 23:16 for clock type, lower 16 bits for frequency in MHz */
+	uint32_t param = (clk << 16) | freq_mhz;
+
+	smu_print("SMU Set hard min by freq: clk = %d, freq_mhz = %d MHz\n", clk, freq_mhz);
+
+	dcn401_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_SetHardMinByFreq, param, &response);
+
+	/* wait until hardmin acknowledged */
+	hard_min_done = dcn401_smu_wait_hard_min_status(clk_mgr, clk);
+	smu_print("SMU Frequency set = %d KHz hard_min_done %d\n", response, hard_min_done);
+
+	return response;
+}
+
+void dcn401_smu_wait_for_dmub_ack_mclk(struct clk_mgr_internal *clk_mgr, bool enable)
+{
+	smu_print("SMU to wait for DMCUB ack for MCLK : %d\n", enable);
+
+	dcn401_smu_send_msg_with_param(clk_mgr, DALSMC_MSG_SetAlwaysWaitDmcubResp, enable ? 1 : 0, NULL);
+}
+
+void dcn401_smu_indicate_drr_status(struct clk_mgr_internal *clk_mgr, bool mod_drr_for_pstate)
+{
+	smu_print("SMU Set indicate drr status = %d\n", mod_drr_for_pstate);
+
+	dcn401_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_IndicateDrrStatus, mod_drr_for_pstate ? 1 : 0, NULL);
+}
+
+bool dcn401_smu_set_idle_uclk_fclk_hardmin(struct clk_mgr_internal *clk_mgr,
+		uint16_t uclk_freq_mhz,
+		uint16_t fclk_freq_mhz)
+{
+	uint32_t response = 0;
+	bool success;
+
+	/* 15:0 for uclk, 32:16 for fclk */
+	uint32_t param = (fclk_freq_mhz << 16) | uclk_freq_mhz;
+
+	smu_print("SMU Set idle hardmin by freq: uclk_freq_mhz = %d MHz, fclk_freq_mhz = %d MHz\n", uclk_freq_mhz, fclk_freq_mhz);
+
+	success = dcn401_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_IdleUclkFclk, param, &response);
+
+	/* wait until hardmin acknowledged */
+	success &= dcn401_smu_wait_hard_min_status(clk_mgr, PPCLK_UCLK);
+	smu_print("SMU hard_min_done %d\n", success);
+
+	return success;
+}
+
+bool dcn401_smu_set_active_uclk_fclk_hardmin(struct clk_mgr_internal *clk_mgr,
+		uint16_t uclk_freq_mhz,
+		uint16_t fclk_freq_mhz)
+{
+	uint32_t response = 0;
+	bool success;
+
+	/* 15:0 for uclk, 32:16 for fclk */
+	uint32_t param = (fclk_freq_mhz << 16) | uclk_freq_mhz;
+
+	smu_print("SMU Set active hardmin by freq: uclk_freq_mhz = %d MHz, fclk_freq_mhz = %d MHz\n", uclk_freq_mhz, fclk_freq_mhz);
+
+	success = dcn401_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_ActiveUclkFclk, param, &response);
+
+	/* wait until hardmin acknowledged */
+	success &= dcn401_smu_wait_hard_min_status(clk_mgr, PPCLK_UCLK);
+	smu_print("SMU hard_min_done %d\n", success);
+
+	return success;
+}
+
+bool dcn401_smu_set_subvp_uclk_fclk_hardmin(struct clk_mgr_internal *clk_mgr,
+		uint16_t uclk_freq_mhz,
+		uint16_t fclk_freq_mhz)
+{
+	uint32_t response = 0;
+	bool success;
+
+	/* 15:0 for uclk, 32:16 for fclk */
+	uint32_t param = (fclk_freq_mhz << 16) | uclk_freq_mhz;
+
+	smu_print("SMU Set active hardmin by freq: uclk_freq_mhz = %d MHz, fclk_freq_mhz = %d MHz\n", uclk_freq_mhz, fclk_freq_mhz);
+
+	success = dcn401_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_SubvpUclkFclk, param, &response);
+
+	return success;
+}
+
+void dcn401_smu_set_min_deep_sleep_dcef_clk(struct clk_mgr_internal *clk_mgr, uint32_t freq_mhz)
+{
+	smu_print("SMU Set min deep sleep dcef clk: freq_mhz = %d MHz\n", freq_mhz);
+
+	dcn401_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_SetMinDeepSleepDcfclk, freq_mhz, NULL);
+}
+
+void dcn401_smu_set_num_of_displays(struct clk_mgr_internal *clk_mgr, uint32_t num_displays)
+{
+	smu_print("SMU Set num of displays: num_displays = %d\n", num_displays);
+
+	dcn401_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_NumOfDisplays, num_displays, NULL);
+}
+
+unsigned int dcn401_smu_get_num_of_umc_channels(struct clk_mgr_internal *clk_mgr)
+{
+	unsigned int response = 0;
+
+	dcn401_smu_send_msg_with_param(clk_mgr, DALSMC_MSG_GetNumUmcChannels, 0, &response);
+
+	smu_print("SMU Get Num UMC Channels: num_umc_channels = %d\n", response);
+
+	return response;
+}
+
+/*
+ * Frequency in MHz returned in lower 16 bits for valid DPM level
+ *
+ * Call with dpm_level = 0xFF to query features, return value will be:
+ *     Bits 7:0 - number of DPM levels
+ *     Bit   28 - 1 = auto DPM on
+ *     Bit   29 - 1 = sweep DPM on
+ *     Bit   30 - 1 = forced DPM on
+ *     Bit   31 - 0 = discrete, 1 = fine-grained
+ *
+ * With fine-grained DPM, only min and max frequencies will be reported
+ *
+ * Returns 0 on failure
+ */
+unsigned int dcn401_smu_get_dpm_freq_by_index(struct clk_mgr_internal *clk_mgr, uint32_t clk, uint8_t dpm_level)
+{
+	uint32_t response = 0;
+
+	/* bits 23:16 for clock type, lower 8 bits for DPM level */
+	uint32_t param = (clk << 16) | dpm_level;
+
+	smu_print("SMU Get dpm freq by index: clk = %d, dpm_level = %d\n", clk, dpm_level);
+
+	dcn401_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_GetDpmFreqByIndex, param, &response);
+
+	smu_print("SMU dpm freq: %d MHz\n", response);
+
+	return response;
+}
+
+/* Returns the max DPM frequency in DC mode in MHz, 0 on failure */
+unsigned int dcn401_smu_get_dc_mode_max_dpm_freq(struct clk_mgr_internal *clk_mgr, uint32_t clk)
+{
+	uint32_t response = 0;
+
+	/* bits 23:16 for clock type */
+	uint32_t param = clk << 16;
+
+	smu_print("SMU Get DC mode max DPM freq: clk = %d\n", clk);
+
+	dcn401_smu_send_msg_with_param(clk_mgr,
+			DALSMC_MSG_GetDcModeMaxDpmFreq, param, &response);
+
+	smu_print("SMU DC mode max DMP freq: %d MHz\n", response);
+
+	return response;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr_smu_msg.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr_smu_msg.h
new file mode 100644
index 000000000000..4f5ac603e822
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr_smu_msg.h
@@ -0,0 +1,41 @@
+// SPDX-License-Identifier: MIT
+//
+// Copyright 2024 Advanced Micro Devices, Inc.
+
+#ifndef __DCN401_CLK_MGR_SMU_MSG_H_
+#define __DCN401_CLK_MGR_SMU_MSG_H_
+
+#include "os_types.h"
+#include "core_types.h"
+
+struct clk_mgr_internal;
+
+bool dcn401_smu_get_smu_version(struct clk_mgr_internal *clk_mgr, unsigned int *version);
+bool dcn401_smu_check_driver_if_version(struct clk_mgr_internal *clk_mgr);
+bool dcn401_smu_check_msg_header_version(struct clk_mgr_internal *clk_mgr);
+void dcn401_smu_send_fclk_pstate_message(struct clk_mgr_internal *clk_mgr, bool support);
+void dcn401_smu_send_uclk_pstate_message(struct clk_mgr_internal *clk_mgr, bool support);
+void dcn401_smu_send_cab_for_uclk_message(struct clk_mgr_internal *clk_mgr, unsigned int num_ways);
+void dcn401_smu_set_dram_addr_high(struct clk_mgr_internal *clk_mgr, uint32_t addr_high);
+void dcn401_smu_set_dram_addr_low(struct clk_mgr_internal *clk_mgr, uint32_t addr_low);
+void dcn401_smu_transfer_wm_table_dram_2_smu(struct clk_mgr_internal *clk_mgr);
+void dcn401_smu_set_pme_workaround(struct clk_mgr_internal *clk_mgr);
+unsigned int dcn401_smu_set_hard_min_by_freq(struct clk_mgr_internal *clk_mgr, uint32_t clk, uint16_t freq_mhz);
+void dcn401_smu_wait_for_dmub_ack_mclk(struct clk_mgr_internal *clk_mgr, bool enable);
+void dcn401_smu_indicate_drr_status(struct clk_mgr_internal *clk_mgr, bool mod_drr_for_pstate);
+bool dcn401_smu_set_idle_uclk_fclk_hardmin(struct clk_mgr_internal *clk_mgr,
+		uint16_t uclk_freq_mhz,
+		uint16_t fclk_freq_mhz);
+bool dcn401_smu_set_active_uclk_fclk_hardmin(struct clk_mgr_internal *clk_mgr,
+		uint16_t uclk_freq_mhz,
+		uint16_t fclk_freq_mhz);
+bool dcn401_smu_set_subvp_uclk_fclk_hardmin(struct clk_mgr_internal *clk_mgr,
+		uint16_t uclk_freq_mhz,
+		uint16_t fclk_freq_mhz);
+void dcn401_smu_set_min_deep_sleep_dcef_clk(struct clk_mgr_internal *clk_mgr, uint32_t freq_mhz);
+void dcn401_smu_set_num_of_displays(struct clk_mgr_internal *clk_mgr, uint32_t num_displays);
+unsigned int dcn401_smu_get_num_of_umc_channels(struct clk_mgr_internal *clk_mgr);
+unsigned int dcn401_smu_get_dc_mode_max_dpm_freq(struct clk_mgr_internal *clk_mgr, uint32_t clk);
+unsigned int dcn401_smu_get_dpm_freq_by_index(struct clk_mgr_internal *clk_mgr, uint32_t clk, uint8_t dpm_level);
+
+#endif /* __DCN401_CLK_MGR_SMU_MSG_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_smu14_driver_if.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_smu14_driver_if.h
new file mode 100644
index 000000000000..36034b32870c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_smu14_driver_if.h
@@ -0,0 +1,66 @@
+// SPDX-License-Identifier: MIT
+//
+// Copyright 2024 Advanced Micro Devices, Inc.
+//
+// This is a stripped-down version of the smu13_driver_if.h file for the relevant DAL interfaces.
+
+#define SMU14_DRIVER_IF_VERSION  0x1
+
+//Only Clks that have DPM descriptors are listed here
+typedef enum {
+  PPCLK_GFXCLK = 0,
+  PPCLK_SOCCLK,
+  PPCLK_UCLK,
+  PPCLK_FCLK,
+  PPCLK_DCLK_0,
+  PPCLK_VCLK_0,
+  PPCLK_DISPCLK,
+  PPCLK_DPPCLK,
+  PPCLK_DPREFCLK,
+  PPCLK_DCFCLK,
+  PPCLK_DTBCLK,
+  PPCLK_COUNT,
+} PPCLK_e;
+
+typedef struct {
+  uint8_t  WmSetting;
+  uint8_t  Flags;
+  uint8_t  Padding[2];
+
+} WatermarkRowGeneric_t;
+
+#define NUM_WM_RANGES 4
+
+typedef enum {
+  WATERMARKS_CLOCK_RANGE = 0,
+  WATERMARKS_DUMMY_PSTATE,
+  WATERMARKS_MALL,
+  WATERMARKS_COUNT,
+} WATERMARKS_FLAGS_e;
+
+typedef struct {
+  // Watermarks
+  WatermarkRowGeneric_t WatermarkRow[NUM_WM_RANGES];
+} Watermarks_t;
+
+typedef struct {
+  Watermarks_t Watermarks;
+  uint32_t  Spare[16];
+
+  uint32_t     MmHubPadding[8]; // SMU internal use
+} WatermarksExternal_t;
+
+// Table types
+#define TABLE_PMFW_PPTABLE            0
+#define TABLE_COMBO_PPTABLE           1
+#define TABLE_WATERMARKS              2
+#define TABLE_AVFS_PSM_DEBUG          3
+#define TABLE_PMSTATUSLOG             4
+#define TABLE_SMU_METRICS             5
+#define TABLE_DRIVER_SMU_CONFIG       6
+#define TABLE_ACTIVITY_MONITOR_COEFF  7
+#define TABLE_OVERDRIVE               8
+#define TABLE_I2C_COMMANDS            9
+#define TABLE_DRIVER_INFO             10
+#define TABLE_ECCINFO                 11
+#define TABLE_COUNT                   12
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c
new file mode 100644
index 000000000000..5f2d5638c819
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/core/dc.c
@@ -0,0 +1,6380 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ */
+
+#include "dm_services.h"
+
+#include "amdgpu.h"
+
+#include "dc.h"
+
+#include "core_status.h"
+#include "core_types.h"
+#include "hw_sequencer.h"
+#include "dce/dce_hwseq.h"
+
+#include "resource.h"
+#include "dc_state.h"
+#include "dc_state_priv.h"
+#include "dc_plane.h"
+#include "dc_plane_priv.h"
+#include "dc_stream_priv.h"
+
+#include "gpio_service_interface.h"
+#include "clk_mgr.h"
+#include "clock_source.h"
+#include "dc_bios_types.h"
+
+#include "bios_parser_interface.h"
+#include "bios/bios_parser_helper.h"
+#include "include/irq_service_interface.h"
+#include "transform.h"
+#include "dmcu.h"
+#include "dpp.h"
+#include "timing_generator.h"
+#include "abm.h"
+#include "virtual/virtual_link_encoder.h"
+#include "hubp.h"
+
+#include "link_hwss.h"
+#include "link_encoder.h"
+#include "link_enc_cfg.h"
+
+#include "link_service.h"
+#include "dm_helpers.h"
+#include "mem_input.h"
+
+#include "dc_dmub_srv.h"
+
+#include "dsc.h"
+
+#include "vm_helper.h"
+
+#include "dce/dce_i2c.h"
+
+#include "dmub/dmub_srv.h"
+
+#include "dce/dmub_psr.h"
+
+#include "dce/dmub_hw_lock_mgr.h"
+
+#include "dc_trace.h"
+
+#include "hw_sequencer_private.h"
+
+#if defined(CONFIG_DRM_AMD_DC_FP)
+#include "dml2/dml2_internal_types.h"
+#include "soc_and_ip_translator.h"
+#endif
+
+#include "dce/dmub_outbox.h"
+
+#define CTX \
+	dc->ctx
+
+#define DC_LOGGER \
+	dc->ctx->logger
+
+static const char DC_BUILD_ID[] = "production-build";
+
+/**
+ * DOC: Overview
+ *
+ * DC is the OS-agnostic component of the amdgpu DC driver.
+ *
+ * DC maintains and validates a set of structs representing the state of the
+ * driver and writes that state to AMD hardware
+ *
+ * Main DC HW structs:
+ *
+ * struct dc - The central struct.  One per driver.  Created on driver load,
+ * destroyed on driver unload.
+ *
+ * struct dc_context - One per driver.
+ * Used as a backpointer by most other structs in dc.
+ *
+ * struct dc_link - One per connector (the physical DP, HDMI, miniDP, or eDP
+ * plugpoints).  Created on driver load, destroyed on driver unload.
+ *
+ * struct dc_sink - One per display.  Created on boot or hotplug.
+ * Destroyed on shutdown or hotunplug.  A dc_link can have a local sink
+ * (the display directly attached).  It may also have one or more remote
+ * sinks (in the Multi-Stream Transport case)
+ *
+ * struct resource_pool - One per driver.  Represents the hw blocks not in the
+ * main pipeline.  Not directly accessible by dm.
+ *
+ * Main dc state structs:
+ *
+ * These structs can be created and destroyed as needed.  There is a full set of
+ * these structs in dc->current_state representing the currently programmed state.
+ *
+ * struct dc_state - The global DC state to track global state information,
+ * such as bandwidth values.
+ *
+ * struct dc_stream_state - Represents the hw configuration for the pipeline from
+ * a framebuffer to a display.  Maps one-to-one with dc_sink.
+ *
+ * struct dc_plane_state - Represents a framebuffer.  Each stream has at least one,
+ * and may have more in the Multi-Plane Overlay case.
+ *
+ * struct resource_context - Represents the programmable state of everything in
+ * the resource_pool.  Not directly accessible by dm.
+ *
+ * struct pipe_ctx - A member of struct resource_context.  Represents the
+ * internal hardware pipeline components.  Each dc_plane_state has either
+ * one or two (in the pipe-split case).
+ */
+
+/* Private functions */
+
+static inline void elevate_update_type(enum surface_update_type *original, enum surface_update_type new)
+{
+	if (new > *original)
+		*original = new;
+}
+
+static void destroy_links(struct dc *dc)
+{
+	uint32_t i;
+
+	for (i = 0; i < dc->link_count; i++) {
+		if (NULL != dc->links[i])
+			dc->link_srv->destroy_link(&dc->links[i]);
+	}
+}
+
+static uint32_t get_num_of_internal_disp(struct dc_link **links, uint32_t num_links)
+{
+	int i;
+	uint32_t count = 0;
+
+	for (i = 0; i < num_links; i++) {
+		if (links[i]->connector_signal == SIGNAL_TYPE_EDP ||
+				links[i]->is_internal_display)
+			count++;
+	}
+
+	return count;
+}
+
+static int get_seamless_boot_stream_count(struct dc_state *ctx)
+{
+	uint8_t i;
+	uint8_t seamless_boot_stream_count = 0;
+
+	for (i = 0; i < ctx->stream_count; i++)
+		if (ctx->streams[i]->apply_seamless_boot_optimization)
+			seamless_boot_stream_count++;
+
+	return seamless_boot_stream_count;
+}
+
+static bool create_links(
+		struct dc *dc,
+		uint32_t num_virtual_links)
+{
+	int i;
+	int connectors_num;
+	struct dc_bios *bios = dc->ctx->dc_bios;
+
+	dc->link_count = 0;
+
+	connectors_num = bios->funcs->get_connectors_number(bios);
+
+	DC_LOG_DC("BIOS object table - number of connectors: %d", connectors_num);
+
+	if (connectors_num > ENUM_ID_COUNT) {
+		dm_error(
+			"DC: Number of connectors %d exceeds maximum of %d!\n",
+			connectors_num,
+			ENUM_ID_COUNT);
+		return false;
+	}
+
+	dm_output_to_console(
+		"DC: %s: connectors_num: physical:%d, virtual:%d\n",
+		__func__,
+		connectors_num,
+		num_virtual_links);
+
+	/* When getting the number of connectors, the VBIOS reports the number of valid indices,
+	 * but it doesn't say which indices are valid, and not every index has an actual connector.
+	 * So, if we don't find a connector on an index, that is not an error.
+	 *
+	 * - There is no guarantee that the first N indices will be valid
+	 * - VBIOS may report a higher amount of valid indices than there are actual connectors
+	 * - Some VBIOS have valid configurations for more connectors than there actually are
+	 *   on the card. This may be because the manufacturer used the same VBIOS for different
+	 *   variants of the same card.
+	 */
+	for (i = 0; dc->link_count < connectors_num && i < MAX_LINKS; i++) {
+		struct graphics_object_id connector_id = bios->funcs->get_connector_id(bios, i);
+		struct link_init_data link_init_params = {0};
+		struct dc_link *link;
+
+		if (connector_id.id == CONNECTOR_ID_UNKNOWN)
+			continue;
+
+		DC_LOG_DC("BIOS object table - printing link object info for connector number: %d, link_index: %d", i, dc->link_count);
+
+		link_init_params.ctx = dc->ctx;
+		/* next BIOS object table connector */
+		link_init_params.connector_index = i;
+		link_init_params.link_index = dc->link_count;
+		link_init_params.dc = dc;
+		link = dc->link_srv->create_link(&link_init_params);
+
+		if (link) {
+			dc->links[dc->link_count] = link;
+			link->dc = dc;
+			++dc->link_count;
+		}
+	}
+
+	DC_LOG_DC("BIOS object table - end");
+
+	/* Create a link for each usb4 dpia port */
+	dc->lowest_dpia_link_index = MAX_LINKS;
+	for (i = 0; i < dc->res_pool->usb4_dpia_count; i++) {
+		struct link_init_data link_init_params = {0};
+		struct dc_link *link;
+
+		link_init_params.ctx = dc->ctx;
+		link_init_params.connector_index = i;
+		link_init_params.link_index = dc->link_count;
+		link_init_params.dc = dc;
+		link_init_params.is_dpia_link = true;
+
+		link = dc->link_srv->create_link(&link_init_params);
+		if (link) {
+			if (dc->lowest_dpia_link_index > dc->link_count)
+				dc->lowest_dpia_link_index = dc->link_count;
+
+			dc->links[dc->link_count] = link;
+			link->dc = dc;
+			++dc->link_count;
+		}
+	}
+
+	for (i = 0; i < num_virtual_links; i++) {
+		struct dc_link *link = kzalloc(sizeof(*link), GFP_KERNEL);
+		struct encoder_init_data enc_init = {0};
+
+		if (link == NULL) {
+			BREAK_TO_DEBUGGER();
+			goto failed_alloc;
+		}
+
+		link->link_index = dc->link_count;
+		dc->links[dc->link_count] = link;
+		dc->link_count++;
+
+		link->ctx = dc->ctx;
+		link->dc = dc;
+		link->connector_signal = SIGNAL_TYPE_VIRTUAL;
+		link->link_id.type = OBJECT_TYPE_CONNECTOR;
+		link->link_id.id = CONNECTOR_ID_VIRTUAL;
+		link->link_id.enum_id = ENUM_ID_1;
+		link->psr_settings.psr_version = DC_PSR_VERSION_UNSUPPORTED;
+		link->link_enc = kzalloc(sizeof(*link->link_enc), GFP_KERNEL);
+
+		if (!link->link_enc) {
+			BREAK_TO_DEBUGGER();
+			goto failed_alloc;
+		}
+
+		link->link_status.dpcd_caps = &link->dpcd_caps;
+
+		enc_init.ctx = dc->ctx;
+		enc_init.channel = CHANNEL_ID_UNKNOWN;
+		enc_init.hpd_source = HPD_SOURCEID_UNKNOWN;
+		enc_init.transmitter = TRANSMITTER_UNKNOWN;
+		enc_init.connector = link->link_id;
+		enc_init.encoder.type = OBJECT_TYPE_ENCODER;
+		enc_init.encoder.id = ENCODER_ID_INTERNAL_VIRTUAL;
+		enc_init.encoder.enum_id = ENUM_ID_1;
+		virtual_link_encoder_construct(link->link_enc, &enc_init);
+	}
+
+	dc->caps.num_of_internal_disp = get_num_of_internal_disp(dc->links, dc->link_count);
+
+	return true;
+
+failed_alloc:
+	return false;
+}
+
+/* Create additional DIG link encoder objects if fewer than the platform
+ * supports were created during link construction. This can happen if the
+ * number of physical connectors is less than the number of DIGs.
+ */
+static bool create_link_encoders(struct dc *dc)
+{
+	bool res = true;
+	unsigned int num_usb4_dpia = dc->res_pool->res_cap->num_usb4_dpia;
+	unsigned int num_dig_link_enc = dc->res_pool->res_cap->num_dig_link_enc;
+	int i;
+
+	/* A platform without USB4 DPIA endpoints has a fixed mapping between DIG
+	 * link encoders and physical display endpoints and does not require
+	 * additional link encoder objects.
+	 */
+	if (num_usb4_dpia == 0)
+		return res;
+
+	/* Create as many link encoder objects as the platform supports. DPIA
+	 * endpoints can be programmably mapped to any DIG.
+	 */
+	if (num_dig_link_enc > dc->res_pool->dig_link_enc_count) {
+		for (i = 0; i < num_dig_link_enc; i++) {
+			struct link_encoder *link_enc = dc->res_pool->link_encoders[i];
+
+			if (!link_enc && dc->res_pool->funcs->link_enc_create_minimal) {
+				link_enc = dc->res_pool->funcs->link_enc_create_minimal(dc->ctx,
+						(enum engine_id)(ENGINE_ID_DIGA + i));
+				if (link_enc) {
+					dc->res_pool->link_encoders[i] = link_enc;
+					dc->res_pool->dig_link_enc_count++;
+				} else {
+					res = false;
+				}
+			}
+		}
+	}
+
+	return res;
+}
+
+/* Destroy any additional DIG link encoder objects created by
+ * create_link_encoders().
+ * NB: Must only be called after destroy_links().
+ */
+static void destroy_link_encoders(struct dc *dc)
+{
+	unsigned int num_usb4_dpia;
+	unsigned int num_dig_link_enc;
+	int i;
+
+	if (!dc->res_pool)
+		return;
+
+	num_usb4_dpia = dc->res_pool->res_cap->num_usb4_dpia;
+	num_dig_link_enc = dc->res_pool->res_cap->num_dig_link_enc;
+
+	/* A platform without USB4 DPIA endpoints has a fixed mapping between DIG
+	 * link encoders and physical display endpoints and does not require
+	 * additional link encoder objects.
+	 */
+	if (num_usb4_dpia == 0)
+		return;
+
+	for (i = 0; i < num_dig_link_enc; i++) {
+		struct link_encoder *link_enc = dc->res_pool->link_encoders[i];
+
+		if (link_enc) {
+			link_enc->funcs->destroy(&link_enc);
+			dc->res_pool->link_encoders[i] = NULL;
+			dc->res_pool->dig_link_enc_count--;
+		}
+	}
+}
+
+static struct dc_perf_trace *dc_perf_trace_create(void)
+{
+	return kzalloc(sizeof(struct dc_perf_trace), GFP_KERNEL);
+}
+
+static void dc_perf_trace_destroy(struct dc_perf_trace **perf_trace)
+{
+	kfree(*perf_trace);
+	*perf_trace = NULL;
+}
+
+static bool set_long_vtotal(struct dc *dc, struct dc_stream_state *stream, struct dc_crtc_timing_adjust *adjust)
+{
+	if (!dc || !stream || !adjust)
+		return false;
+
+	if (!dc->current_state)
+		return false;
+
+	int i;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct pipe_ctx *pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+
+		if (pipe->stream == stream && pipe->stream_res.tg) {
+			if (dc->hwss.set_long_vtotal)
+				dc->hwss.set_long_vtotal(&pipe, 1, adjust->v_total_min, adjust->v_total_max);
+
+			return true;
+		}
+	}
+
+	return false;
+}
+
+/**
+ *  dc_stream_adjust_vmin_vmax - look up pipe context & update parts of DRR
+ *  @dc:     dc reference
+ *  @stream: Initial dc stream state
+ *  @adjust: Updated parameters for vertical_total_min and vertical_total_max
+ *
+ *  Looks up the pipe context of dc_stream_state and updates the
+ *  vertical_total_min and vertical_total_max of the DRR, Dynamic Refresh
+ *  Rate, which is a power-saving feature that targets reducing panel
+ *  refresh rate while the screen is static
+ *
+ *  Return: %true if the pipe context is found and adjusted;
+ *          %false if the pipe context is not found.
+ */
+bool dc_stream_adjust_vmin_vmax(struct dc *dc,
+		struct dc_stream_state *stream,
+		struct dc_crtc_timing_adjust *adjust)
+{
+	int i;
+
+	/*
+	 * Don't adjust DRR while there's bandwidth optimizations pending to
+	 * avoid conflicting with firmware updates.
+	 */
+	if (dc->ctx->dce_version > DCE_VERSION_MAX) {
+		if (dc->optimized_required &&
+			(stream->adjust.v_total_max != adjust->v_total_max ||
+			stream->adjust.v_total_min != adjust->v_total_min)) {
+			stream->adjust.timing_adjust_pending = true;
+			return false;
+		}
+	}
+
+	dc_exit_ips_for_hw_access(dc);
+
+	stream->adjust.v_total_max = adjust->v_total_max;
+	stream->adjust.v_total_mid = adjust->v_total_mid;
+	stream->adjust.v_total_mid_frame_num = adjust->v_total_mid_frame_num;
+	stream->adjust.v_total_min = adjust->v_total_min;
+	stream->adjust.allow_otg_v_count_halt = adjust->allow_otg_v_count_halt;
+
+	if (dc->caps.max_v_total != 0 &&
+		(adjust->v_total_max > dc->caps.max_v_total || adjust->v_total_min > dc->caps.max_v_total)) {
+		stream->adjust.timing_adjust_pending = false;
+		if (adjust->allow_otg_v_count_halt)
+			return set_long_vtotal(dc, stream, adjust);
+		else
+			return false;
+	}
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct pipe_ctx *pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+
+		if (pipe->stream == stream && pipe->stream_res.tg) {
+			dc->hwss.set_drr(&pipe,
+					1,
+					*adjust);
+			stream->adjust.timing_adjust_pending = false;
+			return true;
+		}
+	}
+	return false;
+}
+
+/**
+ * dc_stream_get_last_used_drr_vtotal - Looks up the pipe context of
+ * dc_stream_state and gets the last VTOTAL used by DRR (Dynamic Refresh Rate)
+ *
+ * @dc: [in] dc reference
+ * @stream: [in] Initial dc stream state
+ * @refresh_rate: [in] new refresh_rate
+ *
+ * Return: %true if the pipe context is found and there is an associated
+ *         timing_generator for the DC;
+ *         %false if the pipe context is not found or there is no
+ *         timing_generator for the DC.
+ */
+bool dc_stream_get_last_used_drr_vtotal(struct dc *dc,
+		struct dc_stream_state *stream,
+		uint32_t *refresh_rate)
+{
+	bool status = false;
+
+	int i = 0;
+
+	dc_exit_ips_for_hw_access(dc);
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct pipe_ctx *pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+
+		if (pipe->stream == stream && pipe->stream_res.tg) {
+			/* Only execute if a function pointer has been defined for
+			 * the DC version in question
+			 */
+			if (pipe->stream_res.tg->funcs->get_last_used_drr_vtotal) {
+				pipe->stream_res.tg->funcs->get_last_used_drr_vtotal(pipe->stream_res.tg, refresh_rate);
+
+				status = true;
+
+				break;
+			}
+		}
+	}
+
+	return status;
+}
+
+#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY)
+static inline void
+dc_stream_forward_dmub_crc_window(struct dc_dmub_srv *dmub_srv,
+		struct rect *rect, struct otg_phy_mux *mux_mapping, bool is_stop)
+{
+	union dmub_rb_cmd cmd = {0};
+
+	cmd.secure_display.roi_info.phy_id = mux_mapping->phy_output_num;
+	cmd.secure_display.roi_info.otg_id = mux_mapping->otg_output_num;
+
+	if (is_stop) {
+		cmd.secure_display.header.type = DMUB_CMD__SECURE_DISPLAY;
+		cmd.secure_display.header.sub_type = DMUB_CMD__SECURE_DISPLAY_CRC_STOP_UPDATE;
+	} else {
+		cmd.secure_display.header.type = DMUB_CMD__SECURE_DISPLAY;
+		cmd.secure_display.header.sub_type = DMUB_CMD__SECURE_DISPLAY_CRC_WIN_NOTIFY;
+		cmd.secure_display.roi_info.x_start = rect->x;
+		cmd.secure_display.roi_info.y_start = rect->y;
+		cmd.secure_display.roi_info.x_end = rect->x + rect->width;
+		cmd.secure_display.roi_info.y_end = rect->y + rect->height;
+	}
+
+	dc_wake_and_execute_dmub_cmd(dmub_srv->ctx, &cmd, DM_DMUB_WAIT_TYPE_NO_WAIT);
+}
+
+static inline void
+dc_stream_forward_dmcu_crc_window(struct dmcu *dmcu,
+		struct rect *rect, struct otg_phy_mux *mux_mapping, bool is_stop)
+{
+	if (is_stop)
+		dmcu->funcs->stop_crc_win_update(dmcu, mux_mapping);
+	else
+		dmcu->funcs->forward_crc_window(dmcu, rect, mux_mapping);
+}
+
+bool
+dc_stream_forward_crc_window(struct dc_stream_state *stream,
+		struct rect *rect, uint8_t phy_id, bool is_stop)
+{
+	struct dmcu *dmcu;
+	struct dc_dmub_srv *dmub_srv;
+	struct otg_phy_mux mux_mapping;
+	struct pipe_ctx *pipe;
+	int i;
+	struct dc *dc = stream->ctx->dc;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+		if (pipe->stream == stream && !pipe->top_pipe && !pipe->prev_odm_pipe)
+			break;
+	}
+
+	/* Stream not found */
+	if (i == MAX_PIPES)
+		return false;
+
+	mux_mapping.phy_output_num = phy_id;
+	mux_mapping.otg_output_num = pipe->stream_res.tg->inst;
+
+	dmcu = dc->res_pool->dmcu;
+	dmub_srv = dc->ctx->dmub_srv;
+
+	/* forward to dmub */
+	if (dmub_srv)
+		dc_stream_forward_dmub_crc_window(dmub_srv, rect, &mux_mapping, is_stop);
+	/* forward to dmcu */
+	else if (dmcu && dmcu->funcs->is_dmcu_initialized(dmcu))
+		dc_stream_forward_dmcu_crc_window(dmcu, rect, &mux_mapping, is_stop);
+	else
+		return false;
+
+	return true;
+}
+
+static void
+dc_stream_forward_dmub_multiple_crc_window(struct dc_dmub_srv *dmub_srv,
+		struct crc_window *window, struct otg_phy_mux *mux_mapping, bool stop)
+{
+	int i;
+	union dmub_rb_cmd cmd = {0};
+
+	cmd.secure_display.mul_roi_ctl.phy_id = mux_mapping->phy_output_num;
+	cmd.secure_display.mul_roi_ctl.otg_id = mux_mapping->otg_output_num;
+
+	cmd.secure_display.header.type = DMUB_CMD__SECURE_DISPLAY;
+
+	if (stop) {
+		cmd.secure_display.header.sub_type = DMUB_CMD__SECURE_DISPLAY_MULTIPLE_CRC_STOP_UPDATE;
+	} else {
+		cmd.secure_display.header.sub_type = DMUB_CMD__SECURE_DISPLAY_MULTIPLE_CRC_WIN_NOTIFY;
+		for (i = 0; i < MAX_CRC_WINDOW_NUM; i++) {
+			cmd.secure_display.mul_roi_ctl.roi_ctl[i].x_start = window[i].rect.x;
+			cmd.secure_display.mul_roi_ctl.roi_ctl[i].y_start = window[i].rect.y;
+			cmd.secure_display.mul_roi_ctl.roi_ctl[i].x_end = window[i].rect.x + window[i].rect.width;
+			cmd.secure_display.mul_roi_ctl.roi_ctl[i].y_end = window[i].rect.y + window[i].rect.height;
+			cmd.secure_display.mul_roi_ctl.roi_ctl[i].enable = window[i].enable;
+		}
+	}
+
+	dc_wake_and_execute_dmub_cmd(dmub_srv->ctx, &cmd, DM_DMUB_WAIT_TYPE_NO_WAIT);
+}
+
+bool
+dc_stream_forward_multiple_crc_window(struct dc_stream_state *stream,
+		struct crc_window *window, uint8_t phy_id, bool stop)
+{
+	struct dc_dmub_srv *dmub_srv;
+	struct otg_phy_mux mux_mapping;
+	struct pipe_ctx *pipe;
+	int i;
+	struct dc *dc = stream->ctx->dc;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+		if (pipe->stream == stream && !pipe->top_pipe && !pipe->prev_odm_pipe)
+			break;
+	}
+
+	/* Stream not found */
+	if (i == MAX_PIPES)
+		return false;
+
+	mux_mapping.phy_output_num = phy_id;
+	mux_mapping.otg_output_num = pipe->stream_res.tg->inst;
+
+	dmub_srv = dc->ctx->dmub_srv;
+
+	/* forward to dmub only. no dmcu support*/
+	if (dmub_srv)
+		dc_stream_forward_dmub_multiple_crc_window(dmub_srv, window, &mux_mapping, stop);
+	else
+		return false;
+
+	return true;
+}
+#endif /* CONFIG_DRM_AMD_SECURE_DISPLAY */
+
+/**
+ * dc_stream_configure_crc() - Configure CRC capture for the given stream.
+ * @dc: DC Object
+ * @stream: The stream to configure CRC on.
+ * @crc_window: CRC window (x/y start/end) information
+ * @enable: Enable CRC if true, disable otherwise.
+ * @continuous: Capture CRC on every frame if true. Otherwise, only capture
+ *              once.
+ * @idx: Capture CRC on which CRC engine instance
+ * @reset: Reset CRC engine before the configuration
+ *
+ * By default, the entire frame is used to calculate the CRC.
+ *
+ * Return: %false if the stream is not found or CRC capture is not supported;
+ *         %true if the stream has been configured.
+ */
+bool dc_stream_configure_crc(struct dc *dc, struct dc_stream_state *stream,
+			     struct crc_params *crc_window, bool enable, bool continuous,
+			     uint8_t idx, bool reset)
+{
+	struct pipe_ctx *pipe;
+	struct crc_params param;
+	struct timing_generator *tg;
+
+	pipe = resource_get_otg_master_for_stream(
+			&dc->current_state->res_ctx, stream);
+
+	/* Stream not found */
+	if (pipe == NULL)
+		return false;
+
+	dc_exit_ips_for_hw_access(dc);
+
+	/* By default, capture the full frame */
+	param.windowa_x_start = 0;
+	param.windowa_y_start = 0;
+	param.windowa_x_end = pipe->stream->timing.h_addressable;
+	param.windowa_y_end = pipe->stream->timing.v_addressable;
+	param.windowb_x_start = 0;
+	param.windowb_y_start = 0;
+	param.windowb_x_end = pipe->stream->timing.h_addressable;
+	param.windowb_y_end = pipe->stream->timing.v_addressable;
+
+	if (crc_window) {
+		param.windowa_x_start = crc_window->windowa_x_start;
+		param.windowa_y_start = crc_window->windowa_y_start;
+		param.windowa_x_end = crc_window->windowa_x_end;
+		param.windowa_y_end = crc_window->windowa_y_end;
+		param.windowb_x_start = crc_window->windowb_x_start;
+		param.windowb_y_start = crc_window->windowb_y_start;
+		param.windowb_x_end = crc_window->windowb_x_end;
+		param.windowb_y_end = crc_window->windowb_y_end;
+	}
+
+	param.dsc_mode = pipe->stream->timing.flags.DSC ? 1:0;
+	param.odm_mode = pipe->next_odm_pipe ? 1:0;
+
+	/* Default to the union of both windows */
+	param.selection = UNION_WINDOW_A_B;
+	param.continuous_mode = continuous;
+	param.enable = enable;
+
+	param.crc_eng_inst = idx;
+	param.reset = reset;
+
+	tg = pipe->stream_res.tg;
+
+	/* Only call if supported */
+	if (tg->funcs->configure_crc)
+		return tg->funcs->configure_crc(tg, &param);
+	DC_LOG_WARNING("CRC capture not supported.");
+	return false;
+}
+
+/**
+ * dc_stream_get_crc() - Get CRC values for the given stream.
+ *
+ * @dc: DC object.
+ * @stream: The DC stream state of the stream to get CRCs from.
+ * @idx: index of crc engine to get CRC from
+ * @r_cr: CRC value for the red component.
+ * @g_y:  CRC value for the green component.
+ * @b_cb: CRC value for the blue component.
+ *
+ * dc_stream_configure_crc needs to be called beforehand to enable CRCs.
+ *
+ * Return:
+ * %false if stream is not found, or if CRCs are not enabled.
+ */
+bool dc_stream_get_crc(struct dc *dc, struct dc_stream_state *stream, uint8_t idx,
+		       uint32_t *r_cr, uint32_t *g_y, uint32_t *b_cb)
+{
+	int i;
+	struct pipe_ctx *pipe;
+	struct timing_generator *tg;
+
+	dc_exit_ips_for_hw_access(dc);
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+		if (pipe->stream == stream)
+			break;
+	}
+	/* Stream not found */
+	if (i == MAX_PIPES)
+		return false;
+
+	tg = pipe->stream_res.tg;
+
+	if (tg->funcs->get_crc)
+		return tg->funcs->get_crc(tg, idx, r_cr, g_y, b_cb);
+	DC_LOG_WARNING("CRC capture not supported.");
+	return false;
+}
+
+void dc_stream_set_dyn_expansion(struct dc *dc, struct dc_stream_state *stream,
+		enum dc_dynamic_expansion option)
+{
+	/* OPP FMT dyn expansion updates*/
+	int i;
+	struct pipe_ctx *pipe_ctx;
+
+	dc_exit_ips_for_hw_access(dc);
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		if (dc->current_state->res_ctx.pipe_ctx[i].stream
+				== stream) {
+			pipe_ctx = &dc->current_state->res_ctx.pipe_ctx[i];
+			pipe_ctx->stream_res.opp->dyn_expansion = option;
+			pipe_ctx->stream_res.opp->funcs->opp_set_dyn_expansion(
+					pipe_ctx->stream_res.opp,
+					COLOR_SPACE_YCBCR601,
+					stream->timing.display_color_depth,
+					stream->signal);
+		}
+	}
+}
+
+void dc_stream_set_dither_option(struct dc_stream_state *stream,
+		enum dc_dither_option option)
+{
+	struct bit_depth_reduction_params params;
+	struct dc_link *link = stream->link;
+	struct pipe_ctx *pipes = NULL;
+	int i;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		if (link->dc->current_state->res_ctx.pipe_ctx[i].stream ==
+				stream) {
+			pipes = &link->dc->current_state->res_ctx.pipe_ctx[i];
+			break;
+		}
+	}
+
+	if (!pipes)
+		return;
+	if (option > DITHER_OPTION_MAX)
+		return;
+
+	dc_exit_ips_for_hw_access(stream->ctx->dc);
+
+	stream->dither_option = option;
+
+	memset(&params, 0, sizeof(params));
+	resource_build_bit_depth_reduction_params(stream, &params);
+	stream->bit_depth_params = params;
+
+	if (pipes->plane_res.xfm &&
+	    pipes->plane_res.xfm->funcs->transform_set_pixel_storage_depth) {
+		pipes->plane_res.xfm->funcs->transform_set_pixel_storage_depth(
+			pipes->plane_res.xfm,
+			pipes->plane_res.scl_data.lb_params.depth,
+			&stream->bit_depth_params);
+	}
+
+	pipes->stream_res.opp->funcs->
+		opp_program_bit_depth_reduction(pipes->stream_res.opp, &params);
+}
+
+bool dc_stream_set_gamut_remap(struct dc *dc, const struct dc_stream_state *stream)
+{
+	int i;
+	bool ret = false;
+	struct pipe_ctx *pipes;
+
+	dc_exit_ips_for_hw_access(dc);
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		if (dc->current_state->res_ctx.pipe_ctx[i].stream == stream) {
+			pipes = &dc->current_state->res_ctx.pipe_ctx[i];
+			dc->hwss.program_gamut_remap(pipes);
+			ret = true;
+		}
+	}
+
+	return ret;
+}
+
+bool dc_stream_program_csc_matrix(struct dc *dc, struct dc_stream_state *stream)
+{
+	int i;
+	bool ret = false;
+	struct pipe_ctx *pipes;
+
+	dc_exit_ips_for_hw_access(dc);
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		if (dc->current_state->res_ctx.pipe_ctx[i].stream
+				== stream) {
+
+			pipes = &dc->current_state->res_ctx.pipe_ctx[i];
+			dc->hwss.program_output_csc(dc,
+					pipes,
+					stream->output_color_space,
+					stream->csc_color_matrix.matrix,
+					pipes->stream_res.opp->inst);
+			ret = true;
+		}
+	}
+
+	return ret;
+}
+
+void dc_stream_set_static_screen_params(struct dc *dc,
+		struct dc_stream_state **streams,
+		int num_streams,
+		const struct dc_static_screen_params *params)
+{
+	int i, j;
+	struct pipe_ctx *pipes_affected[MAX_PIPES];
+	int num_pipes_affected = 0;
+
+	dc_exit_ips_for_hw_access(dc);
+
+	for (i = 0; i < num_streams; i++) {
+		struct dc_stream_state *stream = streams[i];
+
+		for (j = 0; j < MAX_PIPES; j++) {
+			if (dc->current_state->res_ctx.pipe_ctx[j].stream
+					== stream) {
+				pipes_affected[num_pipes_affected++] =
+						&dc->current_state->res_ctx.pipe_ctx[j];
+			}
+		}
+	}
+
+	dc->hwss.set_static_screen_control(pipes_affected, num_pipes_affected, params);
+}
+
+static void dc_destruct(struct dc *dc)
+{
+	// reset link encoder assignment table on destruct
+	if (dc->res_pool && dc->res_pool->funcs->link_encs_assign &&
+			!dc->config.unify_link_enc_assignment)
+		link_enc_cfg_init(dc, dc->current_state);
+
+	if (dc->current_state) {
+		dc_state_release(dc->current_state);
+		dc->current_state = NULL;
+	}
+
+	destroy_links(dc);
+
+	destroy_link_encoders(dc);
+
+	if (dc->clk_mgr) {
+		dc_destroy_clk_mgr(dc->clk_mgr);
+		dc->clk_mgr = NULL;
+	}
+
+	dc_destroy_resource_pool(dc);
+#ifdef CONFIG_DRM_AMD_DC_FP
+	dc_destroy_soc_and_ip_translator(&dc->soc_and_ip_translator);
+#endif
+	if (dc->link_srv)
+		link_destroy_link_service(&dc->link_srv);
+
+	if (dc->ctx) {
+		if (dc->ctx->gpio_service)
+			dal_gpio_service_destroy(&dc->ctx->gpio_service);
+
+		if (dc->ctx->created_bios)
+			dal_bios_parser_destroy(&dc->ctx->dc_bios);
+		kfree(dc->ctx->logger);
+		dc_perf_trace_destroy(&dc->ctx->perf_trace);
+
+		kfree(dc->ctx);
+		dc->ctx = NULL;
+	}
+
+	kfree(dc->bw_vbios);
+	dc->bw_vbios = NULL;
+
+	kfree(dc->bw_dceip);
+	dc->bw_dceip = NULL;
+
+	kfree(dc->dcn_soc);
+	dc->dcn_soc = NULL;
+
+	kfree(dc->dcn_ip);
+	dc->dcn_ip = NULL;
+
+	kfree(dc->vm_helper);
+	dc->vm_helper = NULL;
+
+}
+
+static bool dc_construct_ctx(struct dc *dc,
+		const struct dc_init_data *init_params)
+{
+	struct dc_context *dc_ctx;
+
+	dc_ctx = kzalloc(sizeof(*dc_ctx), GFP_KERNEL);
+	if (!dc_ctx)
+		return false;
+
+	dc_stream_init_rmcm_3dlut(dc);
+
+	dc_ctx->cgs_device = init_params->cgs_device;
+	dc_ctx->driver_context = init_params->driver;
+	dc_ctx->dc = dc;
+	dc_ctx->asic_id = init_params->asic_id;
+	dc_ctx->dc_sink_id_count = 0;
+	dc_ctx->dc_stream_id_count = 0;
+	dc_ctx->dce_environment = init_params->dce_environment;
+	dc_ctx->dcn_reg_offsets = init_params->dcn_reg_offsets;
+	dc_ctx->nbio_reg_offsets = init_params->nbio_reg_offsets;
+	dc_ctx->clk_reg_offsets = init_params->clk_reg_offsets;
+
+	/* Create logger */
+	dc_ctx->logger = kmalloc(sizeof(*dc_ctx->logger), GFP_KERNEL);
+
+	if (!dc_ctx->logger) {
+		kfree(dc_ctx);
+		return false;
+	}
+
+	dc_ctx->logger->dev = adev_to_drm(init_params->driver);
+	dc->dml.logger = dc_ctx->logger;
+
+	dc_ctx->dce_version = resource_parse_asic_id(init_params->asic_id);
+
+	dc_ctx->perf_trace = dc_perf_trace_create();
+	if (!dc_ctx->perf_trace) {
+		kfree(dc_ctx);
+		ASSERT_CRITICAL(false);
+		return false;
+	}
+
+	dc->ctx = dc_ctx;
+
+	dc->link_srv = link_create_link_service();
+	if (!dc->link_srv)
+		return false;
+
+	return true;
+}
+
+static bool dc_construct(struct dc *dc,
+		const struct dc_init_data *init_params)
+{
+	struct dc_context *dc_ctx;
+	struct bw_calcs_dceip *dc_dceip;
+	struct bw_calcs_vbios *dc_vbios;
+	struct dcn_soc_bounding_box *dcn_soc;
+	struct dcn_ip_params *dcn_ip;
+
+	dc->config = init_params->flags;
+
+	// Allocate memory for the vm_helper
+	dc->vm_helper = kzalloc(sizeof(struct vm_helper), GFP_KERNEL);
+	if (!dc->vm_helper) {
+		dm_error("%s: failed to create dc->vm_helper\n", __func__);
+		goto fail;
+	}
+
+	memcpy(&dc->bb_overrides, &init_params->bb_overrides, sizeof(dc->bb_overrides));
+
+	dc_dceip = kzalloc(sizeof(*dc_dceip), GFP_KERNEL);
+	if (!dc_dceip) {
+		dm_error("%s: failed to create dceip\n", __func__);
+		goto fail;
+	}
+
+	dc->bw_dceip = dc_dceip;
+
+	dc_vbios = kzalloc(sizeof(*dc_vbios), GFP_KERNEL);
+	if (!dc_vbios) {
+		dm_error("%s: failed to create vbios\n", __func__);
+		goto fail;
+	}
+
+	dc->bw_vbios = dc_vbios;
+	dcn_soc = kzalloc(sizeof(*dcn_soc), GFP_KERNEL);
+	if (!dcn_soc) {
+		dm_error("%s: failed to create dcn_soc\n", __func__);
+		goto fail;
+	}
+
+	dc->dcn_soc = dcn_soc;
+
+	dcn_ip = kzalloc(sizeof(*dcn_ip), GFP_KERNEL);
+	if (!dcn_ip) {
+		dm_error("%s: failed to create dcn_ip\n", __func__);
+		goto fail;
+	}
+
+	dc->dcn_ip = dcn_ip;
+
+	if (init_params->bb_from_dmub)
+		dc->dml2_options.bb_from_dmub = init_params->bb_from_dmub;
+	else
+		dc->dml2_options.bb_from_dmub = NULL;
+
+	if (!dc_construct_ctx(dc, init_params)) {
+		dm_error("%s: failed to create ctx\n", __func__);
+		goto fail;
+	}
+
+	dc_ctx = dc->ctx;
+
+	/* Resource should construct all asic specific resources.
+	 * This should be the only place where we need to parse the asic id
+	 */
+	if (init_params->vbios_override)
+		dc_ctx->dc_bios = init_params->vbios_override;
+	else {
+		/* Create BIOS parser */
+		struct bp_init_data bp_init_data;
+
+		bp_init_data.ctx = dc_ctx;
+		bp_init_data.bios = init_params->asic_id.atombios_base_address;
+
+		dc_ctx->dc_bios = dal_bios_parser_create(
+				&bp_init_data, dc_ctx->dce_version);
+
+		if (!dc_ctx->dc_bios) {
+			ASSERT_CRITICAL(false);
+			goto fail;
+		}
+
+		dc_ctx->created_bios = true;
+	}
+
+	dc->vendor_signature = init_params->vendor_signature;
+
+	/* Create GPIO service */
+	dc_ctx->gpio_service = dal_gpio_service_create(
+			dc_ctx->dce_version,
+			dc_ctx->dce_environment,
+			dc_ctx);
+
+	if (!dc_ctx->gpio_service) {
+		ASSERT_CRITICAL(false);
+		goto fail;
+	}
+
+	dc->res_pool = dc_create_resource_pool(dc, init_params, dc_ctx->dce_version);
+	if (!dc->res_pool)
+		goto fail;
+
+	/* set i2c speed if not done by the respective dcnxxx__resource.c */
+	if (dc->caps.i2c_speed_in_khz_hdcp == 0)
+		dc->caps.i2c_speed_in_khz_hdcp = dc->caps.i2c_speed_in_khz;
+	if (dc->caps.max_optimizable_video_width == 0)
+		dc->caps.max_optimizable_video_width = 5120;
+	dc->clk_mgr = dc_clk_mgr_create(dc->ctx, dc->res_pool->pp_smu, dc->res_pool->dccg);
+	if (!dc->clk_mgr)
+		goto fail;
+#ifdef CONFIG_DRM_AMD_DC_FP
+	dc->clk_mgr->force_smu_not_present = init_params->force_smu_not_present;
+
+	if (dc->res_pool->funcs->update_bw_bounding_box) {
+		DC_FP_START();
+		dc->res_pool->funcs->update_bw_bounding_box(dc, dc->clk_mgr->bw_params);
+		DC_FP_END();
+	}
+	dc->soc_and_ip_translator = dc_create_soc_and_ip_translator(dc_ctx->dce_version);
+	if (!dc->soc_and_ip_translator)
+		goto fail;
+#endif
+
+	if (!create_links(dc, init_params->num_virtual_links))
+		goto fail;
+
+	/* Create additional DIG link encoder objects if fewer than the platform
+	 * supports were created during link construction.
+	 */
+	if (!create_link_encoders(dc))
+		goto fail;
+
+	/* Creation of current_state must occur after dc->dml
+	 * is initialized in dc_create_resource_pool because
+	 * on creation it copies the contents of dc->dml
+	 */
+	dc->current_state = dc_state_create(dc, NULL);
+
+	if (!dc->current_state) {
+		dm_error("%s: failed to create validate ctx\n", __func__);
+		goto fail;
+	}
+
+	return true;
+
+fail:
+	return false;
+}
+
+static void disable_all_writeback_pipes_for_stream(
+		const struct dc *dc,
+		struct dc_stream_state *stream,
+		struct dc_state *context)
+{
+	int i;
+
+	for (i = 0; i < stream->num_wb_info; i++)
+		stream->writeback_info[i].wb_enabled = false;
+}
+
+static void apply_ctx_interdependent_lock(struct dc *dc,
+					  struct dc_state *context,
+					  struct dc_stream_state *stream,
+					  bool lock)
+{
+	int i;
+
+	/* Checks if interdependent update function pointer is NULL or not, takes care of DCE110 case */
+	if (dc->hwss.interdependent_update_lock)
+		dc->hwss.interdependent_update_lock(dc, context, lock);
+	else {
+		for (i = 0; i < dc->res_pool->pipe_count; i++) {
+			struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i];
+			struct pipe_ctx *old_pipe_ctx = &dc->current_state->res_ctx.pipe_ctx[i];
+
+			// Copied conditions that were previously in dce110_apply_ctx_for_surface
+			if (stream == pipe_ctx->stream) {
+				if (resource_is_pipe_type(pipe_ctx, OPP_HEAD) &&
+					(pipe_ctx->plane_state || old_pipe_ctx->plane_state))
+					dc->hwss.pipe_control_lock(dc, pipe_ctx, lock);
+			}
+		}
+	}
+}
+
+static void dc_update_visual_confirm_color(struct dc *dc, struct dc_state *context, struct pipe_ctx *pipe_ctx)
+{
+	if (dc->debug.visual_confirm & VISUAL_CONFIRM_EXPLICIT) {
+		memcpy(&pipe_ctx->visual_confirm_color, &pipe_ctx->plane_state->visual_confirm_color,
+		sizeof(pipe_ctx->visual_confirm_color));
+		return;
+	}
+
+	if (dc->ctx->dce_version >= DCN_VERSION_1_0) {
+		memset(&pipe_ctx->visual_confirm_color, 0, sizeof(struct tg_color));
+
+		if (dc->debug.visual_confirm == VISUAL_CONFIRM_HDR)
+			get_hdr_visual_confirm_color(pipe_ctx, &(pipe_ctx->visual_confirm_color));
+		else if (dc->debug.visual_confirm == VISUAL_CONFIRM_SURFACE)
+			get_surface_visual_confirm_color(pipe_ctx, &(pipe_ctx->visual_confirm_color));
+		else if (dc->debug.visual_confirm == VISUAL_CONFIRM_SWIZZLE)
+			get_surface_tile_visual_confirm_color(pipe_ctx, &(pipe_ctx->visual_confirm_color));
+		else if (dc->debug.visual_confirm == VISUAL_CONFIRM_HW_CURSOR)
+			get_cursor_visual_confirm_color(pipe_ctx, &(pipe_ctx->visual_confirm_color));
+		else if (dc->debug.visual_confirm == VISUAL_CONFIRM_DCC)
+			get_dcc_visual_confirm_color(dc, pipe_ctx, &(pipe_ctx->visual_confirm_color));
+		else {
+			if (dc->ctx->dce_version < DCN_VERSION_2_0)
+				color_space_to_black_color(
+					dc, pipe_ctx->stream->output_color_space, &(pipe_ctx->visual_confirm_color));
+		}
+		if (dc->ctx->dce_version >= DCN_VERSION_2_0) {
+			if (dc->debug.visual_confirm == VISUAL_CONFIRM_MPCTREE)
+				get_mpctree_visual_confirm_color(pipe_ctx, &(pipe_ctx->visual_confirm_color));
+			else if (dc->debug.visual_confirm == VISUAL_CONFIRM_SUBVP)
+				get_subvp_visual_confirm_color(pipe_ctx, &(pipe_ctx->visual_confirm_color));
+			else if (dc->debug.visual_confirm == VISUAL_CONFIRM_MCLK_SWITCH)
+				get_mclk_switch_visual_confirm_color(pipe_ctx, &(pipe_ctx->visual_confirm_color));
+			else if (dc->debug.visual_confirm == VISUAL_CONFIRM_FAMS2)
+				get_fams2_visual_confirm_color(dc, context, pipe_ctx, &(pipe_ctx->visual_confirm_color));
+			else if (dc->debug.visual_confirm == VISUAL_CONFIRM_VABC)
+				get_vabc_visual_confirm_color(pipe_ctx, &(pipe_ctx->visual_confirm_color));
+		}
+	}
+}
+
+void dc_get_visual_confirm_for_stream(
+	struct dc *dc,
+	struct dc_stream_state *stream_state,
+	struct tg_color *color)
+{
+	struct dc_stream_status *stream_status = dc_stream_get_status(stream_state);
+	struct pipe_ctx *pipe_ctx;
+	int i;
+	struct dc_plane_state *plane_state = NULL;
+
+	if (!stream_status)
+		return;
+
+	switch (dc->debug.visual_confirm) {
+	case VISUAL_CONFIRM_DISABLE:
+		return;
+	case VISUAL_CONFIRM_PSR:
+	case VISUAL_CONFIRM_FAMS:
+		pipe_ctx = dc_stream_get_pipe_ctx(stream_state);
+		if (!pipe_ctx)
+			return;
+		dc_dmub_srv_get_visual_confirm_color_cmd(dc, pipe_ctx);
+		memcpy(color, &dc->ctx->dmub_srv->dmub->visual_confirm_color, sizeof(struct tg_color));
+		return;
+
+	default:
+		/* find plane with highest layer_index */
+		for (i = 0; i < stream_status->plane_count; i++) {
+			if (stream_status->plane_states[i]->visible)
+				plane_state = stream_status->plane_states[i];
+		}
+		if (!plane_state)
+			return;
+		/* find pipe that contains plane with highest layer index */
+		for (i = 0; i < MAX_PIPES; i++) {
+			struct pipe_ctx *pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+
+			if (pipe->plane_state == plane_state) {
+				memcpy(color, &pipe->visual_confirm_color, sizeof(struct tg_color));
+				return;
+			}
+		}
+	}
+}
+
+static void disable_dangling_plane(struct dc *dc, struct dc_state *context)
+{
+	int i, j;
+	struct dc_state *dangling_context = dc_state_create_current_copy(dc);
+	struct dc_state *current_ctx;
+	struct pipe_ctx *pipe;
+	struct timing_generator *tg;
+
+	if (dangling_context == NULL)
+		return;
+
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		struct dc_stream_state *old_stream =
+				dc->current_state->res_ctx.pipe_ctx[i].stream;
+		bool should_disable = true;
+		bool pipe_split_change = false;
+
+		if ((context->res_ctx.pipe_ctx[i].top_pipe) &&
+			(dc->current_state->res_ctx.pipe_ctx[i].top_pipe))
+			pipe_split_change = context->res_ctx.pipe_ctx[i].top_pipe->pipe_idx !=
+				dc->current_state->res_ctx.pipe_ctx[i].top_pipe->pipe_idx;
+		else
+			pipe_split_change = context->res_ctx.pipe_ctx[i].top_pipe !=
+				dc->current_state->res_ctx.pipe_ctx[i].top_pipe;
+
+		for (j = 0; j < context->stream_count; j++) {
+			if (old_stream == context->streams[j]) {
+				should_disable = false;
+				break;
+			}
+		}
+		if (!should_disable && pipe_split_change &&
+				dc->current_state->stream_count != context->stream_count)
+			should_disable = true;
+
+		if (old_stream && !dc->current_state->res_ctx.pipe_ctx[i].top_pipe &&
+				!dc->current_state->res_ctx.pipe_ctx[i].prev_odm_pipe) {
+			struct pipe_ctx *old_pipe, *new_pipe;
+
+			old_pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+			new_pipe = &context->res_ctx.pipe_ctx[i];
+
+			if (old_pipe->plane_state && !new_pipe->plane_state)
+				should_disable = true;
+		}
+
+		if (should_disable && old_stream) {
+			bool is_phantom = dc_state_get_stream_subvp_type(dc->current_state, old_stream) == SUBVP_PHANTOM;
+			pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+			tg = pipe->stream_res.tg;
+			/* When disabling plane for a phantom pipe, we must turn on the
+			 * phantom OTG so the disable programming gets the double buffer
+			 * update. Otherwise the pipe will be left in a partially disabled
+			 * state that can result in underflow or hang when enabling it
+			 * again for different use.
+			 */
+			if (is_phantom) {
+				if (tg->funcs->enable_crtc) {
+					if (dc->hwseq->funcs.blank_pixel_data)
+						dc->hwseq->funcs.blank_pixel_data(dc, pipe, true);
+					tg->funcs->enable_crtc(tg);
+				}
+			}
+
+			if (is_phantom)
+				dc_state_rem_all_phantom_planes_for_stream(dc, old_stream, dangling_context, true);
+			else
+				dc_state_rem_all_planes_for_stream(dc, old_stream, dangling_context);
+			disable_all_writeback_pipes_for_stream(dc, old_stream, dangling_context);
+
+			if (pipe->stream && pipe->plane_state) {
+				if (!dc->debug.using_dml2)
+					set_p_state_switch_method(dc, context, pipe);
+				dc_update_visual_confirm_color(dc, context, pipe);
+			}
+
+			if (dc->hwss.apply_ctx_for_surface) {
+				apply_ctx_interdependent_lock(dc, dc->current_state, old_stream, true);
+				dc->hwss.apply_ctx_for_surface(dc, old_stream, 0, dangling_context);
+				apply_ctx_interdependent_lock(dc, dc->current_state, old_stream, false);
+				dc->hwss.post_unlock_program_front_end(dc, dangling_context);
+			}
+
+			if (dc->res_pool->funcs->prepare_mcache_programming)
+				dc->res_pool->funcs->prepare_mcache_programming(dc, dangling_context);
+			if (dc->hwss.program_front_end_for_ctx) {
+				dc->hwss.interdependent_update_lock(dc, dc->current_state, true);
+				dc->hwss.program_front_end_for_ctx(dc, dangling_context);
+				dc->hwss.interdependent_update_lock(dc, dc->current_state, false);
+				dc->hwss.post_unlock_program_front_end(dc, dangling_context);
+			}
+			/* We need to put the phantom OTG back into it's default (disabled) state or we
+			 * can get corruption when transition from one SubVP config to a different one.
+			 * The OTG is set to disable on falling edge of VUPDATE so the plane disable
+			 * will still get it's double buffer update.
+			 */
+			if (is_phantom) {
+				if (tg->funcs->disable_phantom_crtc)
+					tg->funcs->disable_phantom_crtc(tg);
+			}
+		}
+	}
+
+	current_ctx = dc->current_state;
+	dc->current_state = dangling_context;
+	dc_state_release(current_ctx);
+}
+
+static void disable_vbios_mode_if_required(
+		struct dc *dc,
+		struct dc_state *context)
+{
+	unsigned int i, j;
+
+	/* check if timing_changed, disable stream*/
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		struct dc_stream_state *stream = NULL;
+		struct dc_link *link = NULL;
+		struct pipe_ctx *pipe = NULL;
+
+		pipe = &context->res_ctx.pipe_ctx[i];
+		stream = pipe->stream;
+		if (stream == NULL)
+			continue;
+
+		if (stream->apply_seamless_boot_optimization)
+			continue;
+
+		// only looking for first odm pipe
+		if (pipe->prev_odm_pipe)
+			continue;
+
+		if (stream->link->local_sink &&
+			stream->link->local_sink->sink_signal == SIGNAL_TYPE_EDP) {
+			link = stream->link;
+		}
+
+		if (link != NULL && link->link_enc->funcs->is_dig_enabled(link->link_enc)) {
+			unsigned int enc_inst, tg_inst = 0;
+			unsigned int pix_clk_100hz = 0;
+
+			enc_inst = link->link_enc->funcs->get_dig_frontend(link->link_enc);
+			if (enc_inst != ENGINE_ID_UNKNOWN) {
+				for (j = 0; j < dc->res_pool->stream_enc_count; j++) {
+					if (dc->res_pool->stream_enc[j]->id == enc_inst) {
+						tg_inst = dc->res_pool->stream_enc[j]->funcs->dig_source_otg(
+							dc->res_pool->stream_enc[j]);
+						break;
+					}
+				}
+
+				dc->res_pool->dp_clock_source->funcs->get_pixel_clk_frequency_100hz(
+					dc->res_pool->dp_clock_source,
+					tg_inst, &pix_clk_100hz);
+
+				if (link->link_status.link_active) {
+					uint32_t requested_pix_clk_100hz =
+						pipe->stream_res.pix_clk_params.requested_pix_clk_100hz;
+
+					if (pix_clk_100hz != requested_pix_clk_100hz) {
+						dc->link_srv->set_dpms_off(pipe);
+						pipe->stream->dpms_off = false;
+					}
+				}
+			}
+		}
+	}
+}
+
+/* Public functions */
+
+struct dc *dc_create(const struct dc_init_data *init_params)
+{
+	struct dc *dc = kzalloc(sizeof(*dc), GFP_KERNEL);
+	unsigned int full_pipe_count;
+
+	if (!dc)
+		return NULL;
+
+	if (init_params->dce_environment == DCE_ENV_VIRTUAL_HW) {
+		dc->caps.linear_pitch_alignment = 64;
+		if (!dc_construct_ctx(dc, init_params))
+			goto destruct_dc;
+	} else {
+		if (!dc_construct(dc, init_params))
+			goto destruct_dc;
+
+		full_pipe_count = dc->res_pool->pipe_count;
+		if (dc->res_pool->underlay_pipe_index != NO_UNDERLAY_PIPE)
+			full_pipe_count--;
+		dc->caps.max_streams = min(
+				full_pipe_count,
+				dc->res_pool->stream_enc_count);
+
+		dc->caps.max_links = dc->link_count;
+		dc->caps.max_audios = dc->res_pool->audio_count;
+		dc->caps.linear_pitch_alignment = 64;
+
+		dc->caps.max_dp_protocol_version = DP_VERSION_1_4;
+
+		dc->caps.max_otg_num = dc->res_pool->res_cap->num_timing_generator;
+
+		if (dc->res_pool->dmcu != NULL)
+			dc->versions.dmcu_version = dc->res_pool->dmcu->dmcu_version;
+	}
+
+	dc->dcn_reg_offsets = init_params->dcn_reg_offsets;
+	dc->nbio_reg_offsets = init_params->nbio_reg_offsets;
+	dc->clk_reg_offsets = init_params->clk_reg_offsets;
+
+	/* Populate versioning information */
+	dc->versions.dc_ver = DC_VER;
+
+	dc->build_id = DC_BUILD_ID;
+
+	DC_LOG_DC("Display Core initialized\n");
+
+	return dc;
+
+destruct_dc:
+	dc_destruct(dc);
+	kfree(dc);
+	return NULL;
+}
+
+static void detect_edp_presence(struct dc *dc)
+{
+	struct dc_link *edp_links[MAX_NUM_EDP];
+	struct dc_link *edp_link = NULL;
+	enum dc_connection_type type;
+	int i;
+	int edp_num;
+
+	dc_get_edp_links(dc, edp_links, &edp_num);
+	if (!edp_num)
+		return;
+
+	for (i = 0; i < edp_num; i++) {
+		edp_link = edp_links[i];
+		if (dc->config.edp_not_connected) {
+			edp_link->edp_sink_present = false;
+		} else {
+			dc_link_detect_connection_type(edp_link, &type);
+			edp_link->edp_sink_present = (type != dc_connection_none);
+		}
+	}
+}
+
+void dc_hardware_init(struct dc *dc)
+{
+
+	detect_edp_presence(dc);
+	if (dc->ctx->dce_environment != DCE_ENV_VIRTUAL_HW)
+		dc->hwss.init_hw(dc);
+	dc_dmub_srv_notify_fw_dc_power_state(dc->ctx->dmub_srv, DC_ACPI_CM_POWER_STATE_D0);
+}
+
+void dc_init_callbacks(struct dc *dc,
+		const struct dc_callback_init *init_params)
+{
+	dc->ctx->cp_psp = init_params->cp_psp;
+}
+
+void dc_deinit_callbacks(struct dc *dc)
+{
+	memset(&dc->ctx->cp_psp, 0, sizeof(dc->ctx->cp_psp));
+}
+
+void dc_destroy(struct dc **dc)
+{
+	dc_destruct(*dc);
+	kfree(*dc);
+	*dc = NULL;
+}
+
+static void enable_timing_multisync(
+		struct dc *dc,
+		struct dc_state *ctx)
+{
+	int i, multisync_count = 0;
+	int pipe_count = dc->res_pool->pipe_count;
+	struct pipe_ctx *multisync_pipes[MAX_PIPES] = { NULL };
+
+	for (i = 0; i < pipe_count; i++) {
+		if (!ctx->res_ctx.pipe_ctx[i].stream ||
+				!ctx->res_ctx.pipe_ctx[i].stream->triggered_crtc_reset.enabled)
+			continue;
+		if (ctx->res_ctx.pipe_ctx[i].stream == ctx->res_ctx.pipe_ctx[i].stream->triggered_crtc_reset.event_source)
+			continue;
+		multisync_pipes[multisync_count] = &ctx->res_ctx.pipe_ctx[i];
+		multisync_count++;
+	}
+
+	if (multisync_count > 0) {
+		dc->hwss.enable_per_frame_crtc_position_reset(
+			dc, multisync_count, multisync_pipes);
+	}
+}
+
+static void program_timing_sync(
+		struct dc *dc,
+		struct dc_state *ctx)
+{
+	int i, j, k;
+	int group_index = 0;
+	int num_group = 0;
+	int pipe_count = dc->res_pool->pipe_count;
+	struct pipe_ctx *unsynced_pipes[MAX_PIPES] = { NULL };
+
+	for (i = 0; i < pipe_count; i++) {
+		if (!ctx->res_ctx.pipe_ctx[i].stream
+				|| ctx->res_ctx.pipe_ctx[i].top_pipe
+				|| ctx->res_ctx.pipe_ctx[i].prev_odm_pipe)
+			continue;
+
+		unsynced_pipes[i] = &ctx->res_ctx.pipe_ctx[i];
+	}
+
+	for (i = 0; i < pipe_count; i++) {
+		int group_size = 1;
+		enum timing_synchronization_type sync_type = NOT_SYNCHRONIZABLE;
+		struct pipe_ctx *pipe_set[MAX_PIPES];
+
+		if (!unsynced_pipes[i])
+			continue;
+
+		pipe_set[0] = unsynced_pipes[i];
+		unsynced_pipes[i] = NULL;
+
+		/* Add tg to the set, search rest of the tg's for ones with
+		 * same timing, add all tgs with same timing to the group
+		 */
+		for (j = i + 1; j < pipe_count; j++) {
+			if (!unsynced_pipes[j])
+				continue;
+			if (sync_type != TIMING_SYNCHRONIZABLE &&
+				dc->hwss.enable_vblanks_synchronization &&
+				unsynced_pipes[j]->stream_res.tg->funcs->align_vblanks &&
+				resource_are_vblanks_synchronizable(
+					unsynced_pipes[j]->stream,
+					pipe_set[0]->stream)) {
+				sync_type = VBLANK_SYNCHRONIZABLE;
+				pipe_set[group_size] = unsynced_pipes[j];
+				unsynced_pipes[j] = NULL;
+				group_size++;
+			} else
+			if (sync_type != VBLANK_SYNCHRONIZABLE &&
+				resource_are_streams_timing_synchronizable(
+					unsynced_pipes[j]->stream,
+					pipe_set[0]->stream)) {
+				sync_type = TIMING_SYNCHRONIZABLE;
+				pipe_set[group_size] = unsynced_pipes[j];
+				unsynced_pipes[j] = NULL;
+				group_size++;
+			}
+		}
+
+		/* set first unblanked pipe as master */
+		for (j = 0; j < group_size; j++) {
+			bool is_blanked;
+
+			if (pipe_set[j]->stream_res.opp->funcs->dpg_is_blanked)
+				is_blanked =
+					pipe_set[j]->stream_res.opp->funcs->dpg_is_blanked(pipe_set[j]->stream_res.opp);
+			else
+				is_blanked =
+					pipe_set[j]->stream_res.tg->funcs->is_blanked(pipe_set[j]->stream_res.tg);
+			if (!is_blanked) {
+				if (j == 0)
+					break;
+
+				swap(pipe_set[0], pipe_set[j]);
+				break;
+			}
+		}
+
+		for (k = 0; k < group_size; k++) {
+			struct dc_stream_status *status = dc_state_get_stream_status(ctx, pipe_set[k]->stream);
+
+			if (!status)
+				continue;
+
+			status->timing_sync_info.group_id = num_group;
+			status->timing_sync_info.group_size = group_size;
+			if (k == 0)
+				status->timing_sync_info.master = true;
+			else
+				status->timing_sync_info.master = false;
+
+		}
+
+		/* remove any other unblanked pipes as they have already been synced */
+		if (dc->config.use_pipe_ctx_sync_logic) {
+			/* check pipe's syncd to decide which pipe to be removed */
+			for (j = 1; j < group_size; j++) {
+				if (pipe_set[j]->pipe_idx_syncd == pipe_set[0]->pipe_idx_syncd) {
+					group_size--;
+					pipe_set[j] = pipe_set[group_size];
+					j--;
+				} else
+					/* link slave pipe's syncd with master pipe */
+					pipe_set[j]->pipe_idx_syncd = pipe_set[0]->pipe_idx_syncd;
+			}
+		} else {
+			/* remove any other pipes by checking valid plane */
+			for (j = j + 1; j < group_size; j++) {
+				bool is_blanked;
+
+				if (pipe_set[j]->stream_res.opp->funcs->dpg_is_blanked)
+					is_blanked =
+						pipe_set[j]->stream_res.opp->funcs->dpg_is_blanked(pipe_set[j]->stream_res.opp);
+				else
+					is_blanked =
+						pipe_set[j]->stream_res.tg->funcs->is_blanked(pipe_set[j]->stream_res.tg);
+				if (!is_blanked) {
+					group_size--;
+					pipe_set[j] = pipe_set[group_size];
+					j--;
+				}
+			}
+		}
+
+		if (group_size > 1) {
+			if (sync_type == TIMING_SYNCHRONIZABLE) {
+				dc->hwss.enable_timing_synchronization(
+					dc, ctx, group_index, group_size, pipe_set);
+			} else
+				if (sync_type == VBLANK_SYNCHRONIZABLE) {
+				dc->hwss.enable_vblanks_synchronization(
+					dc, group_index, group_size, pipe_set);
+				}
+			group_index++;
+		}
+		num_group++;
+	}
+}
+
+static bool streams_changed(struct dc *dc,
+			    struct dc_stream_state *streams[],
+			    uint8_t stream_count)
+{
+	uint8_t i;
+
+	if (stream_count != dc->current_state->stream_count)
+		return true;
+
+	for (i = 0; i < dc->current_state->stream_count; i++) {
+		if (dc->current_state->streams[i] != streams[i])
+			return true;
+		if (!streams[i]->link->link_state_valid)
+			return true;
+	}
+
+	return false;
+}
+
+bool dc_validate_boot_timing(const struct dc *dc,
+				const struct dc_sink *sink,
+				struct dc_crtc_timing *crtc_timing)
+{
+	struct timing_generator *tg;
+	struct stream_encoder *se = NULL;
+
+	struct dc_crtc_timing hw_crtc_timing = {0};
+
+	struct dc_link *link = sink->link;
+	unsigned int i, enc_inst, tg_inst = 0;
+
+	/* Support seamless boot on EDP displays only */
+	if (sink->sink_signal != SIGNAL_TYPE_EDP) {
+		return false;
+	}
+
+	if (dc->debug.force_odm_combine) {
+		DC_LOG_DEBUG("boot timing validation failed due to force_odm_combine\n");
+		return false;
+	}
+
+	/* Check for enabled DIG to identify enabled display */
+	if (!link->link_enc->funcs->is_dig_enabled(link->link_enc)) {
+		DC_LOG_DEBUG("boot timing validation failed due to disabled DIG\n");
+		return false;
+	}
+
+	enc_inst = link->link_enc->funcs->get_dig_frontend(link->link_enc);
+
+	if (enc_inst == ENGINE_ID_UNKNOWN) {
+		DC_LOG_DEBUG("boot timing validation failed due to unknown DIG engine ID\n");
+		return false;
+	}
+
+	for (i = 0; i < dc->res_pool->stream_enc_count; i++) {
+		if (dc->res_pool->stream_enc[i]->id == enc_inst) {
+
+			se = dc->res_pool->stream_enc[i];
+
+			tg_inst = dc->res_pool->stream_enc[i]->funcs->dig_source_otg(
+				dc->res_pool->stream_enc[i]);
+			break;
+		}
+	}
+
+	// tg_inst not found
+	if (i == dc->res_pool->stream_enc_count) {
+		DC_LOG_DEBUG("boot timing validation failed due to timing generator instance not found\n");
+		return false;
+	}
+
+	if (tg_inst >= dc->res_pool->timing_generator_count) {
+		DC_LOG_DEBUG("boot timing validation failed due to invalid timing generator count\n");
+		return false;
+	}
+
+	if (tg_inst != link->link_enc->preferred_engine) {
+		DC_LOG_DEBUG("boot timing validation failed due to non-preferred timing generator\n");
+		return false;
+	}
+
+	tg = dc->res_pool->timing_generators[tg_inst];
+
+	if (!tg->funcs->get_hw_timing) {
+		DC_LOG_DEBUG("boot timing validation failed due to missing get_hw_timing callback\n");
+		return false;
+	}
+
+	if (!tg->funcs->get_hw_timing(tg, &hw_crtc_timing)) {
+		DC_LOG_DEBUG("boot timing validation failed due to failed get_hw_timing return\n");
+		return false;
+	}
+
+	if (crtc_timing->h_total != hw_crtc_timing.h_total) {
+		DC_LOG_DEBUG("boot timing validation failed due to h_total mismatch\n");
+		return false;
+	}
+
+	if (crtc_timing->h_border_left != hw_crtc_timing.h_border_left) {
+		DC_LOG_DEBUG("boot timing validation failed due to h_border_left mismatch\n");
+		return false;
+	}
+
+	if (crtc_timing->h_addressable != hw_crtc_timing.h_addressable) {
+		DC_LOG_DEBUG("boot timing validation failed due to h_addressable mismatch\n");
+		return false;
+	}
+
+	if (crtc_timing->h_border_right != hw_crtc_timing.h_border_right) {
+		DC_LOG_DEBUG("boot timing validation failed due to h_border_right mismatch\n");
+		return false;
+	}
+
+	if (crtc_timing->h_front_porch != hw_crtc_timing.h_front_porch) {
+		DC_LOG_DEBUG("boot timing validation failed due to h_front_porch mismatch\n");
+		return false;
+	}
+
+	if (crtc_timing->h_sync_width != hw_crtc_timing.h_sync_width) {
+		DC_LOG_DEBUG("boot timing validation failed due to h_sync_width mismatch\n");
+		return false;
+	}
+
+	if (crtc_timing->v_total != hw_crtc_timing.v_total) {
+		DC_LOG_DEBUG("boot timing validation failed due to v_total mismatch\n");
+		return false;
+	}
+
+	if (crtc_timing->v_border_top != hw_crtc_timing.v_border_top) {
+		DC_LOG_DEBUG("boot timing validation failed due to v_border_top mismatch\n");
+		return false;
+	}
+
+	if (crtc_timing->v_addressable != hw_crtc_timing.v_addressable) {
+		DC_LOG_DEBUG("boot timing validation failed due to v_addressable mismatch\n");
+		return false;
+	}
+
+	if (crtc_timing->v_border_bottom != hw_crtc_timing.v_border_bottom) {
+		DC_LOG_DEBUG("boot timing validation failed due to v_border_bottom mismatch\n");
+		return false;
+	}
+
+	if (crtc_timing->v_front_porch != hw_crtc_timing.v_front_porch) {
+		DC_LOG_DEBUG("boot timing validation failed due to v_front_porch mismatch\n");
+		return false;
+	}
+
+	if (crtc_timing->v_sync_width != hw_crtc_timing.v_sync_width) {
+		DC_LOG_DEBUG("boot timing validation failed due to v_sync_width mismatch\n");
+		return false;
+	}
+
+	/* block DSC for now, as VBIOS does not currently support DSC timings */
+	if (crtc_timing->flags.DSC) {
+		DC_LOG_DEBUG("boot timing validation failed due to DSC\n");
+		return false;
+	}
+
+	if (dc_is_dp_signal(link->connector_signal)) {
+		unsigned int pix_clk_100hz = 0;
+		uint32_t numOdmPipes = 1;
+		uint32_t id_src[4] = {0};
+
+		dc->res_pool->dp_clock_source->funcs->get_pixel_clk_frequency_100hz(
+			dc->res_pool->dp_clock_source,
+			tg_inst, &pix_clk_100hz);
+
+		if (tg->funcs->get_optc_source)
+			tg->funcs->get_optc_source(tg,
+						&numOdmPipes, &id_src[0], &id_src[1]);
+
+		if (numOdmPipes == 2) {
+			pix_clk_100hz *= 2;
+		} else if (numOdmPipes == 4) {
+			pix_clk_100hz *= 4;
+		} else if (se && se->funcs->get_pixels_per_cycle) {
+			uint32_t pixels_per_cycle = se->funcs->get_pixels_per_cycle(se);
+
+			if (pixels_per_cycle != 1 && !dc->debug.enable_dp_dig_pixel_rate_div_policy) {
+				DC_LOG_DEBUG("boot timing validation failed due to pixels_per_cycle\n");
+				return false;
+			}
+
+			pix_clk_100hz *= pixels_per_cycle;
+		}
+
+		// Note: In rare cases, HW pixclk may differ from crtc's pixclk
+		// slightly due to rounding issues in 10 kHz units.
+		if (crtc_timing->pix_clk_100hz != pix_clk_100hz) {
+			DC_LOG_DEBUG("boot timing validation failed due to pix_clk_100hz mismatch\n");
+			return false;
+		}
+
+		if (!se || !se->funcs->dp_get_pixel_format) {
+			DC_LOG_DEBUG("boot timing validation failed due to missing dp_get_pixel_format\n");
+			return false;
+		}
+
+		if (!se->funcs->dp_get_pixel_format(
+			se,
+			&hw_crtc_timing.pixel_encoding,
+			&hw_crtc_timing.display_color_depth)) {
+			DC_LOG_DEBUG("boot timing validation failed due to dp_get_pixel_format failure\n");
+			return false;
+		}
+
+		if (hw_crtc_timing.display_color_depth != crtc_timing->display_color_depth) {
+			DC_LOG_DEBUG("boot timing validation failed due to display_color_depth mismatch\n");
+			return false;
+		}
+
+		if (hw_crtc_timing.pixel_encoding != crtc_timing->pixel_encoding) {
+			DC_LOG_DEBUG("boot timing validation failed due to pixel_encoding mismatch\n");
+			return false;
+		}
+	}
+
+
+	if (link->dpcd_caps.dprx_feature.bits.VSC_SDP_COLORIMETRY_SUPPORTED) {
+		DC_LOG_DEBUG("boot timing validation failed due to VSC SDP colorimetry\n");
+		return false;
+	}
+
+	if (link->dpcd_caps.channel_coding_cap.bits.DP_128b_132b_SUPPORTED) {
+		DC_LOG_DEBUG("boot timing validation failed due to DP 128b/132b\n");
+		return false;
+	}
+
+	if (dc->link_srv->edp_is_ilr_optimization_required(link, crtc_timing)) {
+		DC_LOG_EVENT_LINK_TRAINING("Seamless boot disabled to optimize eDP link rate\n");
+		return false;
+	}
+
+	return true;
+}
+
+static inline bool should_update_pipe_for_stream(
+		struct dc_state *context,
+		struct pipe_ctx *pipe_ctx,
+		struct dc_stream_state *stream)
+{
+	return (pipe_ctx->stream && pipe_ctx->stream == stream);
+}
+
+static inline bool should_update_pipe_for_plane(
+		struct dc_state *context,
+		struct pipe_ctx *pipe_ctx,
+		struct dc_plane_state *plane_state)
+{
+	return (pipe_ctx->plane_state == plane_state);
+}
+
+void dc_enable_stereo(
+	struct dc *dc,
+	struct dc_state *context,
+	struct dc_stream_state *streams[],
+	uint8_t stream_count)
+{
+	int i, j;
+	struct pipe_ctx *pipe;
+
+	dc_exit_ips_for_hw_access(dc);
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		if (context != NULL) {
+			pipe = &context->res_ctx.pipe_ctx[i];
+		} else {
+			context = dc->current_state;
+			pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+		}
+
+		for (j = 0; pipe && j < stream_count; j++)  {
+			if (should_update_pipe_for_stream(context, pipe, streams[j]) &&
+				dc->hwss.setup_stereo)
+				dc->hwss.setup_stereo(pipe, dc);
+		}
+	}
+}
+
+void dc_trigger_sync(struct dc *dc, struct dc_state *context)
+{
+	if (context->stream_count > 1 && !dc->debug.disable_timing_sync) {
+		dc_exit_ips_for_hw_access(dc);
+
+		enable_timing_multisync(dc, context);
+		program_timing_sync(dc, context);
+	}
+}
+
+static uint8_t get_stream_mask(struct dc *dc, struct dc_state *context)
+{
+	int i;
+	unsigned int stream_mask = 0;
+
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		if (context->res_ctx.pipe_ctx[i].stream)
+			stream_mask |= 1 << i;
+	}
+
+	return stream_mask;
+}
+
+void dc_z10_restore(const struct dc *dc)
+{
+	if (dc->hwss.z10_restore)
+		dc->hwss.z10_restore(dc);
+}
+
+void dc_z10_save_init(struct dc *dc)
+{
+	if (dc->hwss.z10_save_init)
+		dc->hwss.z10_save_init(dc);
+}
+
+/* Set a pipe unlock order based on the change in DET allocation and stores it in dc scratch memory
+ * Prevents over allocation of DET during unlock process
+ * e.g. 2 pipe config with different streams with a max of 20 DET segments
+ *	Before:								After:
+ *		- Pipe0: 10 DET segments			- Pipe0: 12 DET segments
+ *		- Pipe1: 10 DET segments			- Pipe1: 8 DET segments
+ * If Pipe0 gets updated first, 22 DET segments will be allocated
+ */
+static void determine_pipe_unlock_order(struct dc *dc, struct dc_state *context)
+{
+	unsigned int i = 0;
+	struct pipe_ctx *pipe = NULL;
+	struct timing_generator *tg = NULL;
+
+	if (!dc->config.set_pipe_unlock_order)
+		return;
+
+	memset(dc->scratch.pipes_to_unlock_first, 0, sizeof(dc->scratch.pipes_to_unlock_first));
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		pipe = &context->res_ctx.pipe_ctx[i];
+		tg = pipe->stream_res.tg;
+
+		if (!resource_is_pipe_type(pipe, OTG_MASTER) ||
+				!tg->funcs->is_tg_enabled(tg) ||
+				dc_state_get_pipe_subvp_type(context, pipe) == SUBVP_PHANTOM) {
+			continue;
+		}
+
+		if (resource_calculate_det_for_stream(context, pipe) <
+				resource_calculate_det_for_stream(dc->current_state, &dc->current_state->res_ctx.pipe_ctx[i])) {
+			dc->scratch.pipes_to_unlock_first[i] = true;
+		}
+	}
+}
+
+/**
+ * dc_commit_state_no_check - Apply context to the hardware
+ *
+ * @dc: DC object with the current status to be updated
+ * @context: New state that will become the current status at the end of this function
+ *
+ * Applies given context to the hardware and copy it into current context.
+ * It's up to the user to release the src context afterwards.
+ *
+ * Return: an enum dc_status result code for the operation
+ */
+static enum dc_status dc_commit_state_no_check(struct dc *dc, struct dc_state *context)
+{
+	struct dc_bios *dcb = dc->ctx->dc_bios;
+	enum dc_status result = DC_ERROR_UNEXPECTED;
+	struct pipe_ctx *pipe;
+	int i, k, l;
+	struct dc_stream_state *dc_streams[MAX_STREAMS] = {0};
+	struct dc_state *old_state;
+	bool subvp_prev_use = false;
+
+	dc_z10_restore(dc);
+	dc_allow_idle_optimizations(dc, false);
+
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *old_pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+
+		/* Check old context for SubVP */
+		subvp_prev_use |= (dc_state_get_pipe_subvp_type(dc->current_state, old_pipe) == SUBVP_PHANTOM);
+		if (subvp_prev_use)
+			break;
+	}
+
+	for (i = 0; i < context->stream_count; i++)
+		dc_streams[i] =  context->streams[i];
+
+	if (!dcb->funcs->is_accelerated_mode(dcb)) {
+		disable_vbios_mode_if_required(dc, context);
+		dc->hwss.enable_accelerated_mode(dc, context);
+	}
+
+	if (dc->hwseq->funcs.wait_for_pipe_update_if_needed) {
+		for (i = 0; i < dc->res_pool->pipe_count; i++) {
+			pipe = &context->res_ctx.pipe_ctx[i];
+			//Only delay otg master for a given config
+			if (resource_is_pipe_type(pipe, OTG_MASTER)) {
+				//dc_commit_state_no_check is always a full update
+				dc->hwseq->funcs.wait_for_pipe_update_if_needed(dc, pipe, false);
+				break;
+			}
+		}
+	}
+
+	if (context->stream_count > get_seamless_boot_stream_count(context) ||
+		context->stream_count == 0)
+		dc->hwss.prepare_bandwidth(dc, context);
+
+	/* When SubVP is active, all HW programming must be done while
+	 * SubVP lock is acquired
+	 */
+	if (dc->hwss.subvp_pipe_control_lock)
+		dc->hwss.subvp_pipe_control_lock(dc, context, true, true, NULL, subvp_prev_use);
+	if (dc->hwss.fams2_global_control_lock)
+		dc->hwss.fams2_global_control_lock(dc, context, true);
+
+	if (dc->hwss.update_dsc_pg)
+		dc->hwss.update_dsc_pg(dc, context, false);
+
+	disable_dangling_plane(dc, context);
+	/* re-program planes for existing stream, in case we need to
+	 * free up plane resource for later use
+	 */
+	if (dc->hwss.apply_ctx_for_surface) {
+		for (i = 0; i < context->stream_count; i++) {
+			if (context->streams[i]->mode_changed)
+				continue;
+			apply_ctx_interdependent_lock(dc, context, context->streams[i], true);
+			dc->hwss.apply_ctx_for_surface(
+				dc, context->streams[i],
+				context->stream_status[i].plane_count,
+				context); /* use new pipe config in new context */
+			apply_ctx_interdependent_lock(dc, context, context->streams[i], false);
+			dc->hwss.post_unlock_program_front_end(dc, context);
+		}
+	}
+
+	/* Program hardware */
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		pipe = &context->res_ctx.pipe_ctx[i];
+		dc->hwss.wait_for_mpcc_disconnect(dc, dc->res_pool, pipe);
+	}
+
+	result = dc->hwss.apply_ctx_to_hw(dc, context);
+
+	if (result != DC_OK) {
+		/* Application of dc_state to hardware stopped. */
+		dc->current_state->res_ctx.link_enc_cfg_ctx.mode = LINK_ENC_CFG_STEADY;
+		return result;
+	}
+
+	dc_trigger_sync(dc, context);
+
+	/* Full update should unconditionally be triggered when dc_commit_state_no_check is called */
+	for (i = 0; i < context->stream_count; i++) {
+		uint32_t prev_dsc_changed = context->streams[i]->update_flags.bits.dsc_changed;
+
+		context->streams[i]->update_flags.raw = 0xFFFFFFFF;
+		context->streams[i]->update_flags.bits.dsc_changed = prev_dsc_changed;
+	}
+
+	determine_pipe_unlock_order(dc, context);
+	/* Program all planes within new context*/
+	if (dc->res_pool->funcs->prepare_mcache_programming)
+		dc->res_pool->funcs->prepare_mcache_programming(dc, context);
+	if (dc->hwss.program_front_end_for_ctx) {
+		dc->hwss.interdependent_update_lock(dc, context, true);
+		dc->hwss.program_front_end_for_ctx(dc, context);
+
+		if (dc->hwseq->funcs.set_wait_for_update_needed_for_pipe) {
+			for (i = 0; i < dc->res_pool->pipe_count; i++) {
+				pipe = &context->res_ctx.pipe_ctx[i];
+				dc->hwseq->funcs.set_wait_for_update_needed_for_pipe(dc, pipe);
+			}
+		}
+
+		dc->hwss.interdependent_update_lock(dc, context, false);
+		dc->hwss.post_unlock_program_front_end(dc, context);
+	}
+
+	if (dc->hwss.commit_subvp_config)
+		dc->hwss.commit_subvp_config(dc, context);
+	if (dc->hwss.subvp_pipe_control_lock)
+		dc->hwss.subvp_pipe_control_lock(dc, context, false, true, NULL, subvp_prev_use);
+	if (dc->hwss.fams2_global_control_lock)
+		dc->hwss.fams2_global_control_lock(dc, context, false);
+
+	for (i = 0; i < context->stream_count; i++) {
+		const struct dc_link *link = context->streams[i]->link;
+
+		if (!context->streams[i]->mode_changed)
+			continue;
+
+		if (dc->hwss.apply_ctx_for_surface) {
+			apply_ctx_interdependent_lock(dc, context, context->streams[i], true);
+			dc->hwss.apply_ctx_for_surface(
+					dc, context->streams[i],
+					context->stream_status[i].plane_count,
+					context);
+			apply_ctx_interdependent_lock(dc, context, context->streams[i], false);
+			dc->hwss.post_unlock_program_front_end(dc, context);
+		}
+
+		/*
+		 * enable stereo
+		 * TODO rework dc_enable_stereo call to work with validation sets?
+		 */
+		for (k = 0; k < MAX_PIPES; k++) {
+			pipe = &context->res_ctx.pipe_ctx[k];
+
+			for (l = 0 ; pipe && l < context->stream_count; l++)  {
+				if (context->streams[l] &&
+					context->streams[l] == pipe->stream &&
+					dc->hwss.setup_stereo)
+					dc->hwss.setup_stereo(pipe, dc);
+			}
+		}
+
+		CONN_MSG_MODE(link, "{%dx%d, %dx%d@%dKhz}",
+				context->streams[i]->timing.h_addressable,
+				context->streams[i]->timing.v_addressable,
+				context->streams[i]->timing.h_total,
+				context->streams[i]->timing.v_total,
+				context->streams[i]->timing.pix_clk_100hz / 10);
+	}
+
+	dc_enable_stereo(dc, context, dc_streams, context->stream_count);
+
+	if (get_seamless_boot_stream_count(context) == 0 ||
+		context->stream_count == 0) {
+		/* Must wait for no flips to be pending before doing optimize bw */
+		hwss_wait_for_no_pipes_pending(dc, context);
+		/*
+		 * optimized dispclk depends on ODM setup. Need to wait for ODM
+		 * update pending complete before optimizing bandwidth.
+		 */
+		hwss_wait_for_odm_update_pending_complete(dc, context);
+		/* pplib is notified if disp_num changed */
+		dc->hwss.optimize_bandwidth(dc, context);
+		/* Need to do otg sync again as otg could be out of sync due to otg
+		 * workaround applied during clock update
+		 */
+		dc_trigger_sync(dc, context);
+	}
+
+	if (dc->hwss.update_dsc_pg)
+		dc->hwss.update_dsc_pg(dc, context, true);
+
+	if (dc->ctx->dce_version >= DCE_VERSION_MAX)
+		TRACE_DCN_CLOCK_STATE(&context->bw_ctx.bw.dcn.clk);
+	else
+		TRACE_DCE_CLOCK_STATE(&context->bw_ctx.bw.dce);
+
+	context->stream_mask = get_stream_mask(dc, context);
+
+	if (context->stream_mask != dc->current_state->stream_mask)
+		dc_dmub_srv_notify_stream_mask(dc->ctx->dmub_srv, context->stream_mask);
+
+	for (i = 0; i < context->stream_count; i++)
+		context->streams[i]->mode_changed = false;
+
+	/* Clear update flags that were set earlier to avoid redundant programming */
+	for (i = 0; i < context->stream_count; i++) {
+		context->streams[i]->update_flags.raw = 0x0;
+	}
+
+	old_state = dc->current_state;
+	dc->current_state = context;
+
+	dc_state_release(old_state);
+
+	dc_state_retain(dc->current_state);
+
+	return result;
+}
+
+static bool commit_minimal_transition_state(struct dc *dc,
+		struct dc_state *transition_base_context);
+
+/**
+ * dc_commit_streams - Commit current stream state
+ *
+ * @dc: DC object with the commit state to be configured in the hardware
+ * @params: Parameters for the commit, including the streams to be committed
+ *
+ * Function responsible for commit streams change to the hardware.
+ *
+ * Return:
+ * Return DC_OK if everything work as expected, otherwise, return a dc_status
+ * code.
+ */
+enum dc_status dc_commit_streams(struct dc *dc, struct dc_commit_streams_params *params)
+{
+	int i, j;
+	struct dc_state *context;
+	enum dc_status res = DC_OK;
+	struct dc_validation_set set[MAX_STREAMS] = {0};
+	struct pipe_ctx *pipe;
+	bool handle_exit_odm2to1 = false;
+
+	if (!params)
+		return DC_ERROR_UNEXPECTED;
+
+	if (dc->ctx->dce_environment == DCE_ENV_VIRTUAL_HW)
+		return res;
+
+	if (!streams_changed(dc, params->streams, params->stream_count) &&
+			dc->current_state->power_source == params->power_source)
+		return res;
+
+	dc_exit_ips_for_hw_access(dc);
+
+	DC_LOG_DC("%s: %d streams\n", __func__, params->stream_count);
+
+	for (i = 0; i < params->stream_count; i++) {
+		struct dc_stream_state *stream = params->streams[i];
+		struct dc_stream_status *status = dc_stream_get_status(stream);
+		struct dc_sink *sink = stream->sink;
+
+		/* revalidate streams */
+		if (!dc_is_virtual_signal(sink->sink_signal)) {
+			res = dc_validate_stream(dc, stream);
+			if (res != DC_OK)
+				return res;
+		}
+
+
+		dc_stream_log(dc, stream);
+
+		set[i].stream = stream;
+
+		if (status) {
+			set[i].plane_count = status->plane_count;
+			for (j = 0; j < status->plane_count; j++)
+				set[i].plane_states[j] = status->plane_states[j];
+		}
+	}
+
+	/* ODM Combine 2:1 power optimization is only applied for single stream
+	 * scenario, it uses extra pipes than needed to reduce power consumption
+	 * We need to switch off this feature to make room for new streams.
+	 */
+	if (params->stream_count > dc->current_state->stream_count &&
+			dc->current_state->stream_count == 1) {
+		for (i = 0; i < dc->res_pool->pipe_count; i++) {
+			pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+			if (pipe->next_odm_pipe)
+				handle_exit_odm2to1 = true;
+		}
+	}
+
+	if (handle_exit_odm2to1)
+		res = commit_minimal_transition_state(dc, dc->current_state);
+
+	context = dc_state_create_current_copy(dc);
+	if (!context)
+		goto context_alloc_fail;
+
+	context->power_source = params->power_source;
+
+	res = dc_validate_with_context(dc, set, params->stream_count, context, DC_VALIDATE_MODE_AND_PROGRAMMING);
+
+	/*
+	 * Only update link encoder to stream assignment after bandwidth validation passed.
+	 */
+	if (res == DC_OK && dc->res_pool->funcs->link_encs_assign && !dc->config.unify_link_enc_assignment)
+		dc->res_pool->funcs->link_encs_assign(
+			dc, context, context->streams, context->stream_count);
+
+	if (res != DC_OK) {
+		BREAK_TO_DEBUGGER();
+		goto fail;
+	}
+
+	/*
+	 * If not already seamless, make transition seamless by inserting intermediate minimal transition
+	 */
+	if (dc->hwss.is_pipe_topology_transition_seamless &&
+			!dc->hwss.is_pipe_topology_transition_seamless(dc, dc->current_state, context)) {
+		res = commit_minimal_transition_state(dc, context);
+		if (res != DC_OK) {
+			BREAK_TO_DEBUGGER();
+			goto fail;
+		}
+	}
+
+	res = dc_commit_state_no_check(dc, context);
+
+	for (i = 0; i < params->stream_count; i++) {
+		for (j = 0; j < context->stream_count; j++) {
+			if (params->streams[i]->stream_id == context->streams[j]->stream_id)
+				params->streams[i]->out.otg_offset = context->stream_status[j].primary_otg_inst;
+
+			if (dc_is_embedded_signal(params->streams[i]->signal)) {
+				struct dc_stream_status *status = dc_state_get_stream_status(context, params->streams[i]);
+
+				if (!status)
+					continue;
+
+				if (dc->hwss.is_abm_supported)
+					status->is_abm_supported = dc->hwss.is_abm_supported(dc, context, params->streams[i]);
+				else
+					status->is_abm_supported = true;
+			}
+		}
+	}
+
+fail:
+	dc_state_release(context);
+
+context_alloc_fail:
+
+	DC_LOG_DC("%s Finished.\n", __func__);
+
+	return res;
+}
+
+bool dc_acquire_release_mpc_3dlut(
+		struct dc *dc, bool acquire,
+		struct dc_stream_state *stream,
+		struct dc_3dlut **lut,
+		struct dc_transfer_func **shaper)
+{
+	int pipe_idx;
+	bool ret = false;
+	bool found_pipe_idx = false;
+	const struct resource_pool *pool = dc->res_pool;
+	struct resource_context *res_ctx = &dc->current_state->res_ctx;
+	int mpcc_id = 0;
+
+	if (pool && res_ctx) {
+		if (acquire) {
+			/*find pipe idx for the given stream*/
+			for (pipe_idx = 0; pipe_idx < pool->pipe_count; pipe_idx++) {
+				if (res_ctx->pipe_ctx[pipe_idx].stream == stream) {
+					found_pipe_idx = true;
+					mpcc_id = res_ctx->pipe_ctx[pipe_idx].plane_res.hubp->inst;
+					break;
+				}
+			}
+		} else
+			found_pipe_idx = true;/*for release pipe_idx is not required*/
+
+		if (found_pipe_idx) {
+			if (acquire && pool->funcs->acquire_post_bldn_3dlut)
+				ret = pool->funcs->acquire_post_bldn_3dlut(res_ctx, pool, mpcc_id, lut, shaper);
+			else if (!acquire && pool->funcs->release_post_bldn_3dlut)
+				ret = pool->funcs->release_post_bldn_3dlut(res_ctx, pool, lut, shaper);
+		}
+	}
+	return ret;
+}
+
+static bool is_flip_pending_in_pipes(struct dc *dc, struct dc_state *context)
+{
+	int i;
+	struct pipe_ctx *pipe;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe = &context->res_ctx.pipe_ctx[i];
+
+		// Don't check flip pending on phantom pipes
+		if (!pipe->plane_state || (dc_state_get_pipe_subvp_type(context, pipe) == SUBVP_PHANTOM))
+			continue;
+
+		/* Must set to false to start with, due to OR in update function */
+		pipe->plane_state->status.is_flip_pending = false;
+		dc->hwss.update_pending_status(pipe);
+		if (pipe->plane_state->status.is_flip_pending)
+			return true;
+	}
+	return false;
+}
+
+/* Perform updates here which need to be deferred until next vupdate
+ *
+ * i.e. blnd lut, 3dlut, and shaper lut bypass regs are double buffered
+ * but forcing lut memory to shutdown state is immediate. This causes
+ * single frame corruption as lut gets disabled mid-frame unless shutdown
+ * is deferred until after entering bypass.
+ */
+static void process_deferred_updates(struct dc *dc)
+{
+	int i = 0;
+
+	if (dc->debug.enable_mem_low_power.bits.cm) {
+		ASSERT(dc->dcn_ip->max_num_dpp);
+		for (i = 0; i < dc->dcn_ip->max_num_dpp; i++)
+			if (dc->res_pool->dpps[i]->funcs->dpp_deferred_update)
+				dc->res_pool->dpps[i]->funcs->dpp_deferred_update(dc->res_pool->dpps[i]);
+	}
+}
+
+void dc_post_update_surfaces_to_stream(struct dc *dc)
+{
+	int i;
+	struct dc_state *context = dc->current_state;
+
+	if ((!dc->optimized_required) || get_seamless_boot_stream_count(context) > 0)
+		return;
+
+	post_surface_trace(dc);
+
+	/*
+	 * Only relevant for DCN behavior where we can guarantee the optimization
+	 * is safe to apply - retain the legacy behavior for DCE.
+	 */
+
+	if (dc->ctx->dce_version < DCE_VERSION_MAX)
+		TRACE_DCE_CLOCK_STATE(&context->bw_ctx.bw.dce);
+	else {
+		TRACE_DCN_CLOCK_STATE(&context->bw_ctx.bw.dcn.clk);
+
+		if (is_flip_pending_in_pipes(dc, context))
+			return;
+
+		for (i = 0; i < dc->res_pool->pipe_count; i++)
+			if (context->res_ctx.pipe_ctx[i].stream == NULL ||
+					context->res_ctx.pipe_ctx[i].plane_state == NULL) {
+				context->res_ctx.pipe_ctx[i].pipe_idx = i;
+				dc->hwss.disable_plane(dc, context, &context->res_ctx.pipe_ctx[i]);
+			}
+
+		process_deferred_updates(dc);
+
+		dc->hwss.optimize_bandwidth(dc, context);
+
+		if (dc->hwss.update_dsc_pg)
+			dc->hwss.update_dsc_pg(dc, context, true);
+	}
+
+	dc->optimized_required = false;
+}
+
+bool dc_set_generic_gpio_for_stereo(bool enable,
+		struct gpio_service *gpio_service)
+{
+	enum gpio_result gpio_result = GPIO_RESULT_NON_SPECIFIC_ERROR;
+	struct gpio_pin_info pin_info;
+	struct gpio *generic;
+	struct gpio_generic_mux_config *config = kzalloc(sizeof(struct gpio_generic_mux_config),
+			   GFP_KERNEL);
+
+	if (!config)
+		return false;
+	pin_info = dal_gpio_get_generic_pin_info(gpio_service, GPIO_ID_GENERIC, 0);
+
+	if (pin_info.mask == 0xFFFFFFFF || pin_info.offset == 0xFFFFFFFF) {
+		kfree(config);
+		return false;
+	} else {
+		generic = dal_gpio_service_create_generic_mux(
+			gpio_service,
+			pin_info.offset,
+			pin_info.mask);
+	}
+
+	if (!generic) {
+		kfree(config);
+		return false;
+	}
+
+	gpio_result = dal_gpio_open(generic, GPIO_MODE_OUTPUT);
+
+	config->enable_output_from_mux = enable;
+	config->mux_select = GPIO_SIGNAL_SOURCE_PASS_THROUGH_STEREO_SYNC;
+
+	if (gpio_result == GPIO_RESULT_OK)
+		gpio_result = dal_mux_setup_config(generic, config);
+
+	if (gpio_result == GPIO_RESULT_OK) {
+		dal_gpio_close(generic);
+		dal_gpio_destroy_generic_mux(&generic);
+		kfree(config);
+		return true;
+	} else {
+		dal_gpio_close(generic);
+		dal_gpio_destroy_generic_mux(&generic);
+		kfree(config);
+		return false;
+	}
+}
+
+static bool is_surface_in_context(
+		const struct dc_state *context,
+		const struct dc_plane_state *plane_state)
+{
+	int j;
+
+	for (j = 0; j < MAX_PIPES; j++) {
+		const struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[j];
+
+		if (plane_state == pipe_ctx->plane_state) {
+			return true;
+		}
+	}
+
+	return false;
+}
+
+static enum surface_update_type get_plane_info_update_type(const struct dc *dc, const struct dc_surface_update *u)
+{
+	union surface_update_flags *update_flags = &u->surface->update_flags;
+	enum surface_update_type update_type = UPDATE_TYPE_FAST;
+
+	if (!u->plane_info)
+		return UPDATE_TYPE_FAST;
+
+	if (u->plane_info->color_space != u->surface->color_space) {
+		update_flags->bits.color_space_change = 1;
+		elevate_update_type(&update_type, UPDATE_TYPE_MED);
+	}
+
+	if (u->plane_info->horizontal_mirror != u->surface->horizontal_mirror) {
+		update_flags->bits.horizontal_mirror_change = 1;
+		elevate_update_type(&update_type, UPDATE_TYPE_MED);
+	}
+
+	if (u->plane_info->rotation != u->surface->rotation) {
+		update_flags->bits.rotation_change = 1;
+		elevate_update_type(&update_type, UPDATE_TYPE_FULL);
+	}
+
+	if (u->plane_info->format != u->surface->format) {
+		update_flags->bits.pixel_format_change = 1;
+		elevate_update_type(&update_type, UPDATE_TYPE_FULL);
+	}
+
+	if (u->plane_info->stereo_format != u->surface->stereo_format) {
+		update_flags->bits.stereo_format_change = 1;
+		elevate_update_type(&update_type, UPDATE_TYPE_FULL);
+	}
+
+	if (u->plane_info->per_pixel_alpha != u->surface->per_pixel_alpha) {
+		update_flags->bits.per_pixel_alpha_change = 1;
+		elevate_update_type(&update_type, UPDATE_TYPE_MED);
+	}
+
+	if (u->plane_info->global_alpha_value != u->surface->global_alpha_value) {
+		update_flags->bits.global_alpha_change = 1;
+		elevate_update_type(&update_type, UPDATE_TYPE_MED);
+	}
+
+	if (u->plane_info->dcc.enable != u->surface->dcc.enable
+			|| u->plane_info->dcc.dcc_ind_blk != u->surface->dcc.dcc_ind_blk
+			|| u->plane_info->dcc.meta_pitch != u->surface->dcc.meta_pitch) {
+		/* During DCC on/off, stutter period is calculated before
+		 * DCC has fully transitioned. This results in incorrect
+		 * stutter period calculation. Triggering a full update will
+		 * recalculate stutter period.
+		 */
+		update_flags->bits.dcc_change = 1;
+		elevate_update_type(&update_type, UPDATE_TYPE_FULL);
+	}
+
+	if (resource_pixel_format_to_bpp(u->plane_info->format) !=
+			resource_pixel_format_to_bpp(u->surface->format)) {
+		/* different bytes per element will require full bandwidth
+		 * and DML calculation
+		 */
+		update_flags->bits.bpp_change = 1;
+		elevate_update_type(&update_type, UPDATE_TYPE_FULL);
+	}
+
+	if (u->plane_info->plane_size.surface_pitch != u->surface->plane_size.surface_pitch
+			|| u->plane_info->plane_size.chroma_pitch != u->surface->plane_size.chroma_pitch) {
+		update_flags->bits.plane_size_change = 1;
+		elevate_update_type(&update_type, UPDATE_TYPE_MED);
+	}
+
+
+	if (memcmp(&u->plane_info->tiling_info, &u->surface->tiling_info,
+			sizeof(struct dc_tiling_info)) != 0) {
+		update_flags->bits.swizzle_change = 1;
+		elevate_update_type(&update_type, UPDATE_TYPE_MED);
+
+		/* todo: below are HW dependent, we should add a hook to
+		 * DCE/N resource and validated there.
+		 */
+		if (!dc->debug.skip_full_updated_if_possible) {
+			/* swizzled mode requires RQ to be setup properly,
+			 * thus need to run DML to calculate RQ settings
+			 */
+			update_flags->bits.bandwidth_change = 1;
+			elevate_update_type(&update_type, UPDATE_TYPE_FULL);
+		}
+	}
+
+	/* This should be UPDATE_TYPE_FAST if nothing has changed. */
+	return update_type;
+}
+
+static enum surface_update_type get_scaling_info_update_type(
+		const struct dc *dc,
+		const struct dc_surface_update *u)
+{
+	union surface_update_flags *update_flags = &u->surface->update_flags;
+
+	if (!u->scaling_info)
+		return UPDATE_TYPE_FAST;
+
+	if (u->scaling_info->src_rect.width != u->surface->src_rect.width
+			|| u->scaling_info->src_rect.height != u->surface->src_rect.height
+			|| u->scaling_info->dst_rect.width != u->surface->dst_rect.width
+			|| u->scaling_info->dst_rect.height != u->surface->dst_rect.height
+			|| u->scaling_info->clip_rect.width != u->surface->clip_rect.width
+			|| u->scaling_info->clip_rect.height != u->surface->clip_rect.height
+			|| u->scaling_info->scaling_quality.integer_scaling !=
+					u->surface->scaling_quality.integer_scaling) {
+		update_flags->bits.scaling_change = 1;
+
+		if (u->scaling_info->src_rect.width > u->surface->src_rect.width
+				|| u->scaling_info->src_rect.height > u->surface->src_rect.height)
+			/* Making src rect bigger requires a bandwidth change */
+			update_flags->bits.clock_change = 1;
+
+		if ((u->scaling_info->dst_rect.width < u->surface->dst_rect.width
+			|| u->scaling_info->dst_rect.height < u->surface->dst_rect.height)
+				&& (u->scaling_info->dst_rect.width < u->surface->src_rect.width
+					|| u->scaling_info->dst_rect.height < u->surface->src_rect.height))
+			/* Making dst rect smaller requires a bandwidth change */
+			update_flags->bits.bandwidth_change = 1;
+
+		if (u->scaling_info->src_rect.width > dc->caps.max_optimizable_video_width &&
+			(u->scaling_info->clip_rect.width > u->surface->clip_rect.width ||
+			 u->scaling_info->clip_rect.height > u->surface->clip_rect.height))
+			 /* Changing clip size of a large surface may result in MPC slice count change */
+			update_flags->bits.bandwidth_change = 1;
+	}
+
+	if (u->scaling_info->src_rect.x != u->surface->src_rect.x
+			|| u->scaling_info->src_rect.y != u->surface->src_rect.y
+			|| u->scaling_info->clip_rect.x != u->surface->clip_rect.x
+			|| u->scaling_info->clip_rect.y != u->surface->clip_rect.y
+			|| u->scaling_info->dst_rect.x != u->surface->dst_rect.x
+			|| u->scaling_info->dst_rect.y != u->surface->dst_rect.y)
+		update_flags->bits.position_change = 1;
+
+	/* process every update flag before returning */
+	if (update_flags->bits.clock_change
+			|| update_flags->bits.bandwidth_change
+			|| update_flags->bits.scaling_change)
+		return UPDATE_TYPE_FULL;
+
+	if (update_flags->bits.position_change)
+		return UPDATE_TYPE_MED;
+
+	return UPDATE_TYPE_FAST;
+}
+
+static enum surface_update_type det_surface_update(const struct dc *dc,
+		const struct dc_surface_update *u)
+{
+	const struct dc_state *context = dc->current_state;
+	enum surface_update_type type;
+	enum surface_update_type overall_type = UPDATE_TYPE_FAST;
+	union surface_update_flags *update_flags = &u->surface->update_flags;
+
+	if (!is_surface_in_context(context, u->surface) || u->surface->force_full_update) {
+		update_flags->raw = 0xFFFFFFFF;
+		return UPDATE_TYPE_FULL;
+	}
+
+	update_flags->raw = 0; // Reset all flags
+
+	type = get_plane_info_update_type(dc, u);
+	elevate_update_type(&overall_type, type);
+
+	type = get_scaling_info_update_type(dc, u);
+	elevate_update_type(&overall_type, type);
+
+	if (u->flip_addr) {
+		update_flags->bits.addr_update = 1;
+		if (u->flip_addr->address.tmz_surface != u->surface->address.tmz_surface) {
+			update_flags->bits.tmz_changed = 1;
+			elevate_update_type(&overall_type, UPDATE_TYPE_FULL);
+		}
+	}
+	if (u->in_transfer_func)
+		update_flags->bits.in_transfer_func_change = 1;
+
+	if (u->input_csc_color_matrix)
+		update_flags->bits.input_csc_change = 1;
+
+	if (u->coeff_reduction_factor)
+		update_flags->bits.coeff_reduction_change = 1;
+
+	if (u->gamut_remap_matrix)
+		update_flags->bits.gamut_remap_change = 1;
+
+	if (u->blend_tf)
+		update_flags->bits.gamma_change = 1;
+
+	if (u->gamma) {
+		enum surface_pixel_format format = SURFACE_PIXEL_FORMAT_GRPH_BEGIN;
+
+		if (u->plane_info)
+			format = u->plane_info->format;
+		else
+			format = u->surface->format;
+
+		if (dce_use_lut(format))
+			update_flags->bits.gamma_change = 1;
+	}
+
+	if (u->lut3d_func || u->func_shaper)
+		update_flags->bits.lut_3d = 1;
+
+	if (u->hdr_mult.value)
+		if (u->hdr_mult.value != u->surface->hdr_mult.value) {
+			update_flags->bits.hdr_mult = 1;
+			elevate_update_type(&overall_type, UPDATE_TYPE_MED);
+		}
+
+	if (u->sdr_white_level_nits)
+		if (u->sdr_white_level_nits != u->surface->sdr_white_level_nits) {
+			update_flags->bits.sdr_white_level_nits = 1;
+			elevate_update_type(&overall_type, UPDATE_TYPE_FULL);
+		}
+
+	if (u->cm2_params) {
+		if ((u->cm2_params->component_settings.shaper_3dlut_setting
+					!= u->surface->mcm_shaper_3dlut_setting)
+				|| (u->cm2_params->component_settings.lut1d_enable
+					!= u->surface->mcm_lut1d_enable))
+			update_flags->bits.mcm_transfer_function_enable_change = 1;
+		if (u->cm2_params->cm2_luts.lut3d_data.lut3d_src
+				!= u->surface->mcm_luts.lut3d_data.lut3d_src)
+			update_flags->bits.mcm_transfer_function_enable_change = 1;
+	}
+	if (update_flags->bits.in_transfer_func_change) {
+		type = UPDATE_TYPE_MED;
+		elevate_update_type(&overall_type, type);
+	}
+
+	if (update_flags->bits.lut_3d &&
+			u->surface->mcm_luts.lut3d_data.lut3d_src != DC_CM2_TRANSFER_FUNC_SOURCE_VIDMEM) {
+		type = UPDATE_TYPE_FULL;
+		elevate_update_type(&overall_type, type);
+	}
+	if (update_flags->bits.mcm_transfer_function_enable_change) {
+		type = UPDATE_TYPE_FULL;
+		elevate_update_type(&overall_type, type);
+	}
+
+	if (dc->debug.enable_legacy_fast_update &&
+			(update_flags->bits.gamma_change ||
+			update_flags->bits.gamut_remap_change ||
+			update_flags->bits.input_csc_change ||
+			update_flags->bits.coeff_reduction_change)) {
+		type = UPDATE_TYPE_FULL;
+		elevate_update_type(&overall_type, type);
+	}
+	return overall_type;
+}
+
+/* May need to flip the desktop plane in cases where MPO plane receives a flip but desktop plane doesn't
+ * while both planes are flip_immediate
+ */
+static void force_immediate_gsl_plane_flip(struct dc *dc, struct dc_surface_update *updates, int surface_count)
+{
+	bool has_flip_immediate_plane = false;
+	int i;
+
+	for (i = 0; i < surface_count; i++) {
+		if (updates[i].surface->flip_immediate) {
+			has_flip_immediate_plane = true;
+			break;
+		}
+	}
+
+	if (has_flip_immediate_plane && surface_count > 1) {
+		for (i = 0; i < surface_count; i++) {
+			if (updates[i].surface->flip_immediate)
+				updates[i].surface->update_flags.bits.addr_update = 1;
+		}
+	}
+}
+
+static enum surface_update_type check_update_surfaces_for_stream(
+		struct dc *dc,
+		struct dc_surface_update *updates,
+		int surface_count,
+		struct dc_stream_update *stream_update,
+		const struct dc_stream_status *stream_status)
+{
+	int i;
+	enum surface_update_type overall_type = UPDATE_TYPE_FAST;
+
+	if (dc->idle_optimizations_allowed || dc_can_clear_cursor_limit(dc))
+		overall_type = UPDATE_TYPE_FULL;
+
+	if (stream_status == NULL || stream_status->plane_count != surface_count)
+		overall_type = UPDATE_TYPE_FULL;
+
+	if (stream_update && stream_update->pending_test_pattern) {
+		overall_type = UPDATE_TYPE_FULL;
+	}
+
+	if (stream_update && stream_update->hw_cursor_req) {
+		overall_type = UPDATE_TYPE_FULL;
+	}
+
+	/* some stream updates require passive update */
+	if (stream_update) {
+		union stream_update_flags *su_flags = &stream_update->stream->update_flags;
+
+		if ((stream_update->src.height != 0 && stream_update->src.width != 0) ||
+			(stream_update->dst.height != 0 && stream_update->dst.width != 0) ||
+			stream_update->integer_scaling_update)
+			su_flags->bits.scaling = 1;
+
+		if (dc->debug.enable_legacy_fast_update && stream_update->out_transfer_func)
+			su_flags->bits.out_tf = 1;
+
+		if (stream_update->abm_level)
+			su_flags->bits.abm_level = 1;
+
+		if (stream_update->dpms_off)
+			su_flags->bits.dpms_off = 1;
+
+		if (stream_update->gamut_remap)
+			su_flags->bits.gamut_remap = 1;
+
+		if (stream_update->wb_update)
+			su_flags->bits.wb_update = 1;
+
+		if (stream_update->dsc_config)
+			su_flags->bits.dsc_changed = 1;
+
+		if (stream_update->mst_bw_update)
+			su_flags->bits.mst_bw = 1;
+
+		if (stream_update->stream->freesync_on_desktop &&
+			(stream_update->vrr_infopacket || stream_update->allow_freesync ||
+				stream_update->vrr_active_variable || stream_update->vrr_active_fixed))
+			su_flags->bits.fams_changed = 1;
+
+		if (stream_update->scaler_sharpener_update)
+			su_flags->bits.scaler_sharpener = 1;
+
+		if (stream_update->sharpening_required)
+			su_flags->bits.sharpening_required = 1;
+
+		if (stream_update->output_color_space)
+			su_flags->bits.out_csc = 1;
+
+		if (su_flags->raw != 0)
+			overall_type = UPDATE_TYPE_FULL;
+
+		if (stream_update->output_csc_transform)
+			su_flags->bits.out_csc = 1;
+
+		/* Output transfer function changes do not require bandwidth recalculation,
+		 * so don't trigger a full update
+		 */
+		if (!dc->debug.enable_legacy_fast_update && stream_update->out_transfer_func)
+			su_flags->bits.out_tf = 1;
+	}
+
+	for (i = 0 ; i < surface_count; i++) {
+		enum surface_update_type type =
+				det_surface_update(dc, &updates[i]);
+
+		elevate_update_type(&overall_type, type);
+	}
+
+	return overall_type;
+}
+
+/*
+ * dc_check_update_surfaces_for_stream() - Determine update type (fast, med, or full)
+ *
+ * See :c:type:`enum surface_update_type <surface_update_type>` for explanation of update types
+ */
+enum surface_update_type dc_check_update_surfaces_for_stream(
+		struct dc *dc,
+		struct dc_surface_update *updates,
+		int surface_count,
+		struct dc_stream_update *stream_update,
+		const struct dc_stream_status *stream_status)
+{
+	int i;
+	enum surface_update_type type;
+
+	if (stream_update)
+		stream_update->stream->update_flags.raw = 0;
+	for (i = 0; i < surface_count; i++)
+		updates[i].surface->update_flags.raw = 0;
+
+	type = check_update_surfaces_for_stream(dc, updates, surface_count, stream_update, stream_status);
+	if (type == UPDATE_TYPE_FULL) {
+		if (stream_update) {
+			uint32_t dsc_changed = stream_update->stream->update_flags.bits.dsc_changed;
+			stream_update->stream->update_flags.raw = 0xFFFFFFFF;
+			stream_update->stream->update_flags.bits.dsc_changed = dsc_changed;
+		}
+		for (i = 0; i < surface_count; i++)
+			updates[i].surface->update_flags.raw = 0xFFFFFFFF;
+	}
+
+	if (type == UPDATE_TYPE_FAST) {
+		// If there's an available clock comparator, we use that.
+		if (dc->clk_mgr->funcs->are_clock_states_equal) {
+			if (!dc->clk_mgr->funcs->are_clock_states_equal(&dc->clk_mgr->clks, &dc->current_state->bw_ctx.bw.dcn.clk))
+				dc->optimized_required = true;
+		// Else we fallback to mem compare.
+		} else if (memcmp(&dc->current_state->bw_ctx.bw.dcn.clk, &dc->clk_mgr->clks, offsetof(struct dc_clocks, prev_p_state_change_support)) != 0) {
+			dc->optimized_required = true;
+		}
+	}
+
+	return type;
+}
+
+static struct dc_stream_status *stream_get_status(
+	struct dc_state *ctx,
+	struct dc_stream_state *stream)
+{
+	uint8_t i;
+
+	for (i = 0; i < ctx->stream_count; i++) {
+		if (stream == ctx->streams[i]) {
+			return &ctx->stream_status[i];
+		}
+	}
+
+	return NULL;
+}
+
+static const enum surface_update_type update_surface_trace_level = UPDATE_TYPE_FULL;
+
+static void copy_surface_update_to_plane(
+		struct dc_plane_state *surface,
+		struct dc_surface_update *srf_update)
+{
+	if (srf_update->flip_addr) {
+		surface->address = srf_update->flip_addr->address;
+		surface->flip_immediate =
+			srf_update->flip_addr->flip_immediate;
+		surface->time.time_elapsed_in_us[surface->time.index] =
+			srf_update->flip_addr->flip_timestamp_in_us -
+				surface->time.prev_update_time_in_us;
+		surface->time.prev_update_time_in_us =
+			srf_update->flip_addr->flip_timestamp_in_us;
+		surface->time.index++;
+		if (surface->time.index >= DC_PLANE_UPDATE_TIMES_MAX)
+			surface->time.index = 0;
+
+		surface->triplebuffer_flips = srf_update->flip_addr->triplebuffer_flips;
+	}
+
+	if (srf_update->scaling_info) {
+		surface->scaling_quality =
+				srf_update->scaling_info->scaling_quality;
+		surface->dst_rect =
+				srf_update->scaling_info->dst_rect;
+		surface->src_rect =
+				srf_update->scaling_info->src_rect;
+		surface->clip_rect =
+				srf_update->scaling_info->clip_rect;
+	}
+
+	if (srf_update->plane_info) {
+		surface->color_space =
+				srf_update->plane_info->color_space;
+		surface->format =
+				srf_update->plane_info->format;
+		surface->plane_size =
+				srf_update->plane_info->plane_size;
+		surface->rotation =
+				srf_update->plane_info->rotation;
+		surface->horizontal_mirror =
+				srf_update->plane_info->horizontal_mirror;
+		surface->stereo_format =
+				srf_update->plane_info->stereo_format;
+		surface->tiling_info =
+				srf_update->plane_info->tiling_info;
+		surface->visible =
+				srf_update->plane_info->visible;
+		surface->per_pixel_alpha =
+				srf_update->plane_info->per_pixel_alpha;
+		surface->global_alpha =
+				srf_update->plane_info->global_alpha;
+		surface->global_alpha_value =
+				srf_update->plane_info->global_alpha_value;
+		surface->dcc =
+				srf_update->plane_info->dcc;
+		surface->layer_index =
+				srf_update->plane_info->layer_index;
+	}
+
+	if (srf_update->gamma) {
+		memcpy(&surface->gamma_correction.entries,
+			&srf_update->gamma->entries,
+			sizeof(struct dc_gamma_entries));
+		surface->gamma_correction.is_identity =
+			srf_update->gamma->is_identity;
+		surface->gamma_correction.num_entries =
+			srf_update->gamma->num_entries;
+		surface->gamma_correction.type =
+			srf_update->gamma->type;
+	}
+
+	if (srf_update->in_transfer_func) {
+		surface->in_transfer_func.sdr_ref_white_level =
+			srf_update->in_transfer_func->sdr_ref_white_level;
+		surface->in_transfer_func.tf =
+			srf_update->in_transfer_func->tf;
+		surface->in_transfer_func.type =
+			srf_update->in_transfer_func->type;
+		memcpy(&surface->in_transfer_func.tf_pts,
+			&srf_update->in_transfer_func->tf_pts,
+			sizeof(struct dc_transfer_func_distributed_points));
+	}
+
+	if (srf_update->cm2_params) {
+		surface->mcm_shaper_3dlut_setting = srf_update->cm2_params->component_settings.shaper_3dlut_setting;
+		surface->mcm_lut1d_enable = srf_update->cm2_params->component_settings.lut1d_enable;
+		surface->mcm_luts = srf_update->cm2_params->cm2_luts;
+	}
+
+	if (srf_update->func_shaper) {
+		memcpy(&surface->in_shaper_func, srf_update->func_shaper,
+		sizeof(surface->in_shaper_func));
+
+		if (surface->mcm_shaper_3dlut_setting >= DC_CM2_SHAPER_3DLUT_SETTING_ENABLE_SHAPER)
+			surface->mcm_luts.shaper = &surface->in_shaper_func;
+	}
+
+	if (srf_update->lut3d_func)
+		memcpy(&surface->lut3d_func, srf_update->lut3d_func,
+		sizeof(surface->lut3d_func));
+
+	if (srf_update->hdr_mult.value)
+		surface->hdr_mult =
+				srf_update->hdr_mult;
+
+	if (srf_update->sdr_white_level_nits)
+		surface->sdr_white_level_nits =
+				srf_update->sdr_white_level_nits;
+
+	if (srf_update->blend_tf) {
+		memcpy(&surface->blend_tf, srf_update->blend_tf,
+		sizeof(surface->blend_tf));
+
+		if (surface->mcm_lut1d_enable)
+			surface->mcm_luts.lut1d_func = &surface->blend_tf;
+	}
+
+	if (srf_update->cm2_params || srf_update->blend_tf)
+		surface->lut_bank_a = !surface->lut_bank_a;
+
+	if (srf_update->input_csc_color_matrix)
+		surface->input_csc_color_matrix =
+			*srf_update->input_csc_color_matrix;
+
+	if (srf_update->coeff_reduction_factor)
+		surface->coeff_reduction_factor =
+			*srf_update->coeff_reduction_factor;
+
+	if (srf_update->gamut_remap_matrix)
+		surface->gamut_remap_matrix =
+			*srf_update->gamut_remap_matrix;
+
+	if (srf_update->cursor_csc_color_matrix)
+		surface->cursor_csc_color_matrix =
+			*srf_update->cursor_csc_color_matrix;
+
+	if (srf_update->bias_and_scale.bias_and_scale_valid)
+			surface->bias_and_scale =
+					srf_update->bias_and_scale;
+}
+
+static void copy_stream_update_to_stream(struct dc *dc,
+					 struct dc_state *context,
+					 struct dc_stream_state *stream,
+					 struct dc_stream_update *update)
+{
+	struct dc_context *dc_ctx = dc->ctx;
+
+	if (update == NULL || stream == NULL)
+		return;
+
+	if (update->src.height && update->src.width)
+		stream->src = update->src;
+
+	if (update->dst.height && update->dst.width)
+		stream->dst = update->dst;
+
+	if (update->out_transfer_func) {
+		stream->out_transfer_func.sdr_ref_white_level =
+			update->out_transfer_func->sdr_ref_white_level;
+		stream->out_transfer_func.tf = update->out_transfer_func->tf;
+		stream->out_transfer_func.type =
+			update->out_transfer_func->type;
+		memcpy(&stream->out_transfer_func.tf_pts,
+		       &update->out_transfer_func->tf_pts,
+		       sizeof(struct dc_transfer_func_distributed_points));
+	}
+
+	if (update->hdr_static_metadata)
+		stream->hdr_static_metadata = *update->hdr_static_metadata;
+
+	if (update->abm_level)
+		stream->abm_level = *update->abm_level;
+
+	if (update->periodic_interrupt)
+		stream->periodic_interrupt = *update->periodic_interrupt;
+
+	if (update->gamut_remap)
+		stream->gamut_remap_matrix = *update->gamut_remap;
+
+	/* Note: this being updated after mode set is currently not a use case
+	 * however if it arises OCSC would need to be reprogrammed at the
+	 * minimum
+	 */
+	if (update->output_color_space)
+		stream->output_color_space = *update->output_color_space;
+
+	if (update->output_csc_transform)
+		stream->csc_color_matrix = *update->output_csc_transform;
+
+	if (update->vrr_infopacket)
+		stream->vrr_infopacket = *update->vrr_infopacket;
+
+	if (update->hw_cursor_req)
+		stream->hw_cursor_req = *update->hw_cursor_req;
+
+	if (update->allow_freesync)
+		stream->allow_freesync = *update->allow_freesync;
+
+	if (update->vrr_active_variable)
+		stream->vrr_active_variable = *update->vrr_active_variable;
+
+	if (update->vrr_active_fixed)
+		stream->vrr_active_fixed = *update->vrr_active_fixed;
+
+	if (update->crtc_timing_adjust) {
+		if (stream->adjust.v_total_min != update->crtc_timing_adjust->v_total_min ||
+			stream->adjust.v_total_max != update->crtc_timing_adjust->v_total_max ||
+			stream->adjust.timing_adjust_pending)
+			update->crtc_timing_adjust->timing_adjust_pending = true;
+		stream->adjust = *update->crtc_timing_adjust;
+		update->crtc_timing_adjust->timing_adjust_pending = false;
+	}
+
+	if (update->dpms_off)
+		stream->dpms_off = *update->dpms_off;
+
+	if (update->hfvsif_infopacket)
+		stream->hfvsif_infopacket = *update->hfvsif_infopacket;
+
+	if (update->vtem_infopacket)
+		stream->vtem_infopacket = *update->vtem_infopacket;
+
+	if (update->vsc_infopacket)
+		stream->vsc_infopacket = *update->vsc_infopacket;
+
+	if (update->vsp_infopacket)
+		stream->vsp_infopacket = *update->vsp_infopacket;
+
+	if (update->adaptive_sync_infopacket)
+		stream->adaptive_sync_infopacket = *update->adaptive_sync_infopacket;
+
+	if (update->avi_infopacket)
+		stream->avi_infopacket = *update->avi_infopacket;
+
+	if (update->dither_option)
+		stream->dither_option = *update->dither_option;
+
+	if (update->pending_test_pattern)
+		stream->test_pattern = *update->pending_test_pattern;
+	/* update current stream with writeback info */
+	if (update->wb_update) {
+		int i;
+
+		stream->num_wb_info = update->wb_update->num_wb_info;
+		ASSERT(stream->num_wb_info <= MAX_DWB_PIPES);
+		for (i = 0; i < stream->num_wb_info; i++)
+			stream->writeback_info[i] =
+				update->wb_update->writeback_info[i];
+	}
+	if (update->dsc_config) {
+		struct dc_dsc_config old_dsc_cfg = stream->timing.dsc_cfg;
+		uint32_t old_dsc_enabled = stream->timing.flags.DSC;
+		uint32_t enable_dsc = (update->dsc_config->num_slices_h != 0 &&
+				       update->dsc_config->num_slices_v != 0);
+
+		/* Use temporarry context for validating new DSC config */
+		struct dc_state *dsc_validate_context = dc_state_create_copy(dc->current_state);
+
+		if (dsc_validate_context) {
+			stream->timing.dsc_cfg = *update->dsc_config;
+			stream->timing.flags.DSC = enable_dsc;
+			if (dc->res_pool->funcs->validate_bandwidth(dc, dsc_validate_context,
+				DC_VALIDATE_MODE_ONLY) != DC_OK) {
+				stream->timing.dsc_cfg = old_dsc_cfg;
+				stream->timing.flags.DSC = old_dsc_enabled;
+				update->dsc_config = NULL;
+			}
+
+			dc_state_release(dsc_validate_context);
+		} else {
+			DC_ERROR("Failed to allocate new validate context for DSC change\n");
+			update->dsc_config = NULL;
+		}
+	}
+	if (update->scaler_sharpener_update)
+		stream->scaler_sharpener_update = *update->scaler_sharpener_update;
+	if (update->sharpening_required)
+		stream->sharpening_required = *update->sharpening_required;
+}
+
+static void backup_planes_and_stream_state(
+		struct dc_scratch_space *scratch,
+		struct dc_stream_state *stream)
+{
+	int i;
+	struct dc_stream_status *status = dc_stream_get_status(stream);
+
+	if (!status)
+		return;
+
+	for (i = 0; i < status->plane_count; i++) {
+		dc_plane_copy_config(&scratch->plane_states[i], status->plane_states[i]);
+	}
+	scratch->stream_state = *stream;
+}
+
+static void restore_planes_and_stream_state(
+		struct dc_scratch_space *scratch,
+		struct dc_stream_state *stream)
+{
+	int i;
+	struct dc_stream_status *status = dc_stream_get_status(stream);
+
+	if (!status)
+		return;
+
+	for (i = 0; i < status->plane_count; i++) {
+		dc_plane_copy_config(status->plane_states[i], &scratch->plane_states[i]);
+	}
+	*stream = scratch->stream_state;
+}
+
+/**
+ * update_seamless_boot_flags() - Helper function for updating seamless boot flags
+ *
+ * @dc: Current DC state
+ * @context: New DC state to be programmed
+ * @surface_count: Number of surfaces that have an updated
+ * @stream: Corresponding stream to be updated in the current flip
+ *
+ * Updating seamless boot flags do not need to be part of the commit sequence. This
+ * helper function will update the seamless boot flags on each flip (if required)
+ * outside of the HW commit sequence (fast or slow).
+ *
+ * Return: void
+ */
+static void update_seamless_boot_flags(struct dc *dc,
+		struct dc_state *context,
+		int surface_count,
+		struct dc_stream_state *stream)
+{
+	if (get_seamless_boot_stream_count(context) > 0 && (surface_count > 0 || stream->dpms_off)) {
+		/* Optimize seamless boot flag keeps clocks and watermarks high until
+		 * first flip. After first flip, optimization is required to lower
+		 * bandwidth. Important to note that it is expected UEFI will
+		 * only light up a single display on POST, therefore we only expect
+		 * one stream with seamless boot flag set.
+		 */
+		if (stream->apply_seamless_boot_optimization) {
+			stream->apply_seamless_boot_optimization = false;
+
+			if (get_seamless_boot_stream_count(context) == 0)
+				dc->optimized_required = true;
+		}
+	}
+}
+
+/**
+ * update_planes_and_stream_state() - The function takes planes and stream
+ * updates as inputs and determines the appropriate update type. If update type
+ * is FULL, the function allocates a new context, populates and validates it.
+ * Otherwise, it updates current dc context. The function will return both
+ * new_context and new_update_type back to the caller. The function also backs
+ * up both current and new contexts into corresponding dc state scratch memory.
+ * TODO: The function does too many things, and even conditionally allocates dc
+ * context memory implicitly. We should consider to break it down.
+ *
+ * @dc: Current DC state
+ * @srf_updates: an array of surface updates
+ * @surface_count: surface update count
+ * @stream: Corresponding stream to be updated
+ * @stream_update: stream update
+ * @new_update_type: [out] determined update type by the function
+ * @new_context: [out] new context allocated and validated if update type is
+ * FULL, reference to current context if update type is less than FULL.
+ *
+ * Return: true if a valid update is populated into new_context, false
+ * otherwise.
+ */
+static bool update_planes_and_stream_state(struct dc *dc,
+		struct dc_surface_update *srf_updates, int surface_count,
+		struct dc_stream_state *stream,
+		struct dc_stream_update *stream_update,
+		enum surface_update_type *new_update_type,
+		struct dc_state **new_context)
+{
+	struct dc_state *context;
+	int i, j;
+	enum surface_update_type update_type;
+	const struct dc_stream_status *stream_status;
+	struct dc_context *dc_ctx = dc->ctx;
+
+	stream_status = dc_stream_get_status(stream);
+
+	if (!stream_status) {
+		if (surface_count) /* Only an error condition if surf_count non-zero*/
+			ASSERT(false);
+
+		return false; /* Cannot commit surface to stream that is not committed */
+	}
+
+	context = dc->current_state;
+	update_type = dc_check_update_surfaces_for_stream(
+			dc, srf_updates, surface_count, stream_update, stream_status);
+	/* It is possible to receive a flip for one plane while there are multiple flip_immediate planes in the same stream.
+	 * E.g. Desktop and MPO plane are flip_immediate but only the MPO plane received a flip
+	 * Force the other flip_immediate planes to flip so GSL doesn't wait for a flip that won't come.
+	 */
+	force_immediate_gsl_plane_flip(dc, srf_updates, surface_count);
+	if (update_type == UPDATE_TYPE_FULL)
+		backup_planes_and_stream_state(&dc->scratch.current_state, stream);
+
+	/* update current stream with the new updates */
+	copy_stream_update_to_stream(dc, context, stream, stream_update);
+
+	/* do not perform surface update if surface has invalid dimensions
+	 * (all zero) and no scaling_info is provided
+	 */
+	if (surface_count > 0) {
+		for (i = 0; i < surface_count; i++) {
+			if ((srf_updates[i].surface->src_rect.width == 0 ||
+				 srf_updates[i].surface->src_rect.height == 0 ||
+				 srf_updates[i].surface->dst_rect.width == 0 ||
+				 srf_updates[i].surface->dst_rect.height == 0) &&
+				(!srf_updates[i].scaling_info ||
+				  srf_updates[i].scaling_info->src_rect.width == 0 ||
+				  srf_updates[i].scaling_info->src_rect.height == 0 ||
+				  srf_updates[i].scaling_info->dst_rect.width == 0 ||
+				  srf_updates[i].scaling_info->dst_rect.height == 0)) {
+				DC_ERROR("Invalid src/dst rects in surface update!\n");
+				return false;
+			}
+		}
+	}
+
+	if (update_type >= update_surface_trace_level)
+		update_surface_trace(dc, srf_updates, surface_count);
+
+	for (i = 0; i < surface_count; i++)
+		copy_surface_update_to_plane(srf_updates[i].surface, &srf_updates[i]);
+
+	if (update_type >= UPDATE_TYPE_FULL) {
+		struct dc_plane_state *new_planes[MAX_SURFACES] = {0};
+
+		for (i = 0; i < surface_count; i++)
+			new_planes[i] = srf_updates[i].surface;
+
+		/* initialize scratch memory for building context */
+		context = dc_state_create_copy(dc->current_state);
+		if (context == NULL) {
+			DC_ERROR("Failed to allocate new validate context!\n");
+			return false;
+		}
+
+		/* For each full update, remove all existing phantom pipes first.
+		 * Ensures that we have enough pipes for newly added MPO planes
+		 */
+		dc_state_remove_phantom_streams_and_planes(dc, context);
+		dc_state_release_phantom_streams_and_planes(dc, context);
+
+		/*remove old surfaces from context */
+		if (!dc_state_rem_all_planes_for_stream(dc, stream, context)) {
+
+			BREAK_TO_DEBUGGER();
+			goto fail;
+		}
+
+		/* add surface to context */
+		if (!dc_state_add_all_planes_for_stream(dc, stream, new_planes, surface_count, context)) {
+
+			BREAK_TO_DEBUGGER();
+			goto fail;
+		}
+	}
+
+	/* save update parameters into surface */
+	for (i = 0; i < surface_count; i++) {
+		struct dc_plane_state *surface = srf_updates[i].surface;
+
+		if (update_type != UPDATE_TYPE_MED)
+			continue;
+		if (surface->update_flags.bits.position_change) {
+			for (j = 0; j < dc->res_pool->pipe_count; j++) {
+				struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[j];
+
+				if (pipe_ctx->plane_state != surface)
+					continue;
+
+				resource_build_scaling_params(pipe_ctx);
+			}
+		}
+	}
+
+	if (update_type == UPDATE_TYPE_FULL) {
+		if (dc->res_pool->funcs->validate_bandwidth(dc, context, DC_VALIDATE_MODE_AND_PROGRAMMING) != DC_OK) {
+			BREAK_TO_DEBUGGER();
+			goto fail;
+		}
+	}
+	update_seamless_boot_flags(dc, context, surface_count, stream);
+
+	*new_context = context;
+	*new_update_type = update_type;
+	if (update_type == UPDATE_TYPE_FULL)
+		backup_planes_and_stream_state(&dc->scratch.new_state, stream);
+
+	return true;
+
+fail:
+	dc_state_release(context);
+
+	return false;
+
+}
+
+static void commit_planes_do_stream_update(struct dc *dc,
+		struct dc_stream_state *stream,
+		struct dc_stream_update *stream_update,
+		enum surface_update_type update_type,
+		struct dc_state *context)
+{
+	int j;
+
+	// Stream updates
+	for (j = 0; j < dc->res_pool->pipe_count; j++) {
+		struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[j];
+
+		if (resource_is_pipe_type(pipe_ctx, OTG_MASTER) && pipe_ctx->stream == stream) {
+
+			if (stream_update->periodic_interrupt && dc->hwss.setup_periodic_interrupt)
+				dc->hwss.setup_periodic_interrupt(dc, pipe_ctx);
+
+			if ((stream_update->hdr_static_metadata && !stream->use_dynamic_meta) ||
+					stream_update->vrr_infopacket ||
+					stream_update->vsc_infopacket ||
+					stream_update->vsp_infopacket ||
+					stream_update->hfvsif_infopacket ||
+					stream_update->adaptive_sync_infopacket ||
+					stream_update->vtem_infopacket ||
+					stream_update->avi_infopacket) {
+				resource_build_info_frame(pipe_ctx);
+				dc->hwss.update_info_frame(pipe_ctx);
+
+				if (dc_is_dp_signal(pipe_ctx->stream->signal))
+					dc->link_srv->dp_trace_source_sequence(
+							pipe_ctx->stream->link,
+							DPCD_SOURCE_SEQ_AFTER_UPDATE_INFO_FRAME);
+			}
+
+			if (stream_update->hdr_static_metadata &&
+					stream->use_dynamic_meta &&
+					dc->hwss.set_dmdata_attributes &&
+					pipe_ctx->stream->dmdata_address.quad_part != 0)
+				dc->hwss.set_dmdata_attributes(pipe_ctx);
+
+			if (stream_update->gamut_remap)
+				dc_stream_set_gamut_remap(dc, stream);
+
+			if (stream_update->output_csc_transform)
+				dc_stream_program_csc_matrix(dc, stream);
+
+			if (stream_update->dither_option) {
+				struct pipe_ctx *odm_pipe = pipe_ctx->next_odm_pipe;
+				resource_build_bit_depth_reduction_params(pipe_ctx->stream,
+									&pipe_ctx->stream->bit_depth_params);
+				pipe_ctx->stream_res.opp->funcs->opp_program_fmt(pipe_ctx->stream_res.opp,
+						&stream->bit_depth_params,
+						&stream->clamping);
+				while (odm_pipe) {
+					odm_pipe->stream_res.opp->funcs->opp_program_fmt(odm_pipe->stream_res.opp,
+							&stream->bit_depth_params,
+							&stream->clamping);
+					odm_pipe = odm_pipe->next_odm_pipe;
+				}
+			}
+
+			if (stream_update->cursor_attributes)
+				program_cursor_attributes(dc, stream);
+
+			if (stream_update->cursor_position)
+				program_cursor_position(dc, stream);
+
+			/* Full fe update*/
+			if (update_type == UPDATE_TYPE_FAST)
+				continue;
+
+			if (stream_update->dsc_config)
+				dc->link_srv->update_dsc_config(pipe_ctx);
+
+			if (stream_update->mst_bw_update) {
+				if (stream_update->mst_bw_update->is_increase)
+					dc->link_srv->increase_mst_payload(pipe_ctx,
+							stream_update->mst_bw_update->mst_stream_bw);
+				else
+					dc->link_srv->reduce_mst_payload(pipe_ctx,
+							stream_update->mst_bw_update->mst_stream_bw);
+			}
+
+			if (stream_update->pending_test_pattern) {
+				/*
+				 * test pattern params depends on ODM topology
+				 * changes that we could be applying to front
+				 * end. Since at the current stage front end
+				 * changes are not yet applied. We can only
+				 * apply test pattern in hw based on current
+				 * state and populate the final test pattern
+				 * params in new state. If current and new test
+				 * pattern params are different as result of
+				 * different ODM topology being used, it will be
+				 * detected and handle during front end
+				 * programming update.
+				 */
+				dc->link_srv->dp_set_test_pattern(stream->link,
+					stream->test_pattern.type,
+					stream->test_pattern.color_space,
+					stream->test_pattern.p_link_settings,
+					stream->test_pattern.p_custom_pattern,
+					stream->test_pattern.cust_pattern_size);
+				resource_build_test_pattern_params(&context->res_ctx, pipe_ctx);
+			}
+
+			if (stream_update->dpms_off) {
+				if (*stream_update->dpms_off) {
+					dc->link_srv->set_dpms_off(pipe_ctx);
+					/* for dpms, keep acquired resources*/
+					if (pipe_ctx->stream_res.audio && !dc->debug.az_endpoint_mute_only)
+						pipe_ctx->stream_res.audio->funcs->az_disable(pipe_ctx->stream_res.audio);
+
+					dc->optimized_required = true;
+
+				} else {
+					if (get_seamless_boot_stream_count(context) == 0)
+						dc->hwss.prepare_bandwidth(dc, dc->current_state);
+					dc->link_srv->set_dpms_on(dc->current_state, pipe_ctx);
+				}
+			} else if (pipe_ctx->stream->link->wa_flags.blank_stream_on_ocs_change && stream_update->output_color_space
+					&& !stream->dpms_off && dc_is_dp_signal(pipe_ctx->stream->signal)) {
+				/*
+				 * Workaround for firmware issue in some receivers where they don't pick up
+				 * correct output color space unless DP link is disabled/re-enabled
+				 */
+				dc->link_srv->set_dpms_on(dc->current_state, pipe_ctx);
+			}
+
+			if (stream_update->abm_level && pipe_ctx->stream_res.abm) {
+				bool should_program_abm = true;
+
+				// if otg funcs defined check if blanked before programming
+				if (pipe_ctx->stream_res.tg->funcs->is_blanked)
+					if (pipe_ctx->stream_res.tg->funcs->is_blanked(pipe_ctx->stream_res.tg))
+						should_program_abm = false;
+
+				if (should_program_abm) {
+					if (*stream_update->abm_level == ABM_LEVEL_IMMEDIATE_DISABLE) {
+						dc->hwss.set_abm_immediate_disable(pipe_ctx);
+					} else {
+						pipe_ctx->stream_res.abm->funcs->set_abm_level(
+							pipe_ctx->stream_res.abm, stream->abm_level);
+					}
+				}
+			}
+		}
+	}
+}
+
+static bool dc_dmub_should_send_dirty_rect_cmd(struct dc *dc, struct dc_stream_state *stream)
+{
+	if ((stream->link->psr_settings.psr_version == DC_PSR_VERSION_SU_1
+			|| stream->link->psr_settings.psr_version == DC_PSR_VERSION_1)
+			&& stream->ctx->dce_version >= DCN_VERSION_3_1)
+		return true;
+
+	if (stream->link->replay_settings.config.replay_supported)
+		return true;
+
+	if (stream->ctx->dce_version >= DCN_VERSION_3_5 && stream->abm_level)
+		return true;
+
+	return false;
+}
+
+void dc_dmub_update_dirty_rect(struct dc *dc,
+			       int surface_count,
+			       struct dc_stream_state *stream,
+			       struct dc_surface_update *srf_updates,
+			       struct dc_state *context)
+{
+	union dmub_rb_cmd cmd;
+	struct dmub_cmd_update_dirty_rect_data *update_dirty_rect;
+	unsigned int i, j;
+	unsigned int panel_inst = 0;
+
+	if (!dc_dmub_should_send_dirty_rect_cmd(dc, stream))
+		return;
+
+	if (!dc_get_edp_link_panel_inst(dc, stream->link, &panel_inst))
+		return;
+
+	memset(&cmd, 0x0, sizeof(cmd));
+	cmd.update_dirty_rect.header.type = DMUB_CMD__UPDATE_DIRTY_RECT;
+	cmd.update_dirty_rect.header.sub_type = 0;
+	cmd.update_dirty_rect.header.payload_bytes =
+		sizeof(cmd.update_dirty_rect) -
+		sizeof(cmd.update_dirty_rect.header);
+	update_dirty_rect = &cmd.update_dirty_rect.update_dirty_rect_data;
+	for (i = 0; i < surface_count; i++) {
+		struct dc_plane_state *plane_state = srf_updates[i].surface;
+		const struct dc_flip_addrs *flip_addr = srf_updates[i].flip_addr;
+
+		if (!srf_updates[i].surface || !flip_addr)
+			continue;
+		/* Do not send in immediate flip mode */
+		if (srf_updates[i].surface->flip_immediate)
+			continue;
+
+		update_dirty_rect->cmd_version = DMUB_CMD_PSR_CONTROL_VERSION_1;
+		update_dirty_rect->dirty_rect_count = flip_addr->dirty_rect_count;
+		memcpy(update_dirty_rect->src_dirty_rects, flip_addr->dirty_rects,
+				sizeof(flip_addr->dirty_rects));
+		for (j = 0; j < dc->res_pool->pipe_count; j++) {
+			struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[j];
+
+			if (pipe_ctx->stream != stream)
+				continue;
+			if (pipe_ctx->plane_state != plane_state)
+				continue;
+
+			update_dirty_rect->panel_inst = panel_inst;
+			update_dirty_rect->pipe_idx = j;
+			dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_NO_WAIT);
+		}
+	}
+}
+
+static void build_dmub_update_dirty_rect(
+		struct dc *dc,
+		int surface_count,
+		struct dc_stream_state *stream,
+		struct dc_surface_update *srf_updates,
+		struct dc_state *context,
+		struct dc_dmub_cmd dc_dmub_cmd[],
+		unsigned int *dmub_cmd_count)
+{
+	union dmub_rb_cmd cmd;
+	struct dmub_cmd_update_dirty_rect_data *update_dirty_rect;
+	unsigned int i, j;
+	unsigned int panel_inst = 0;
+
+	if (!dc_dmub_should_send_dirty_rect_cmd(dc, stream))
+		return;
+
+	if (!dc_get_edp_link_panel_inst(dc, stream->link, &panel_inst))
+		return;
+
+	memset(&cmd, 0x0, sizeof(cmd));
+	cmd.update_dirty_rect.header.type = DMUB_CMD__UPDATE_DIRTY_RECT;
+	cmd.update_dirty_rect.header.sub_type = 0;
+	cmd.update_dirty_rect.header.payload_bytes =
+		sizeof(cmd.update_dirty_rect) -
+		sizeof(cmd.update_dirty_rect.header);
+	update_dirty_rect = &cmd.update_dirty_rect.update_dirty_rect_data;
+	for (i = 0; i < surface_count; i++) {
+		struct dc_plane_state *plane_state = srf_updates[i].surface;
+		const struct dc_flip_addrs *flip_addr = srf_updates[i].flip_addr;
+
+		if (!srf_updates[i].surface || !flip_addr)
+			continue;
+		/* Do not send in immediate flip mode */
+		if (srf_updates[i].surface->flip_immediate)
+			continue;
+		update_dirty_rect->cmd_version = DMUB_CMD_PSR_CONTROL_VERSION_1;
+		update_dirty_rect->dirty_rect_count = flip_addr->dirty_rect_count;
+		memcpy(update_dirty_rect->src_dirty_rects, flip_addr->dirty_rects,
+				sizeof(flip_addr->dirty_rects));
+		for (j = 0; j < dc->res_pool->pipe_count; j++) {
+			struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[j];
+
+			if (pipe_ctx->stream != stream)
+				continue;
+			if (pipe_ctx->plane_state != plane_state)
+				continue;
+			update_dirty_rect->panel_inst = panel_inst;
+			update_dirty_rect->pipe_idx = j;
+			dc_dmub_cmd[*dmub_cmd_count].dmub_cmd = cmd;
+			dc_dmub_cmd[*dmub_cmd_count].wait_type = DM_DMUB_WAIT_TYPE_NO_WAIT;
+			(*dmub_cmd_count)++;
+		}
+	}
+}
+
+static bool check_address_only_update(union surface_update_flags update_flags)
+{
+	union surface_update_flags addr_only_update_flags;
+	addr_only_update_flags.raw = 0;
+	addr_only_update_flags.bits.addr_update = 1;
+
+	return update_flags.bits.addr_update &&
+			!(update_flags.raw & ~addr_only_update_flags.raw);
+}
+
+/**
+ * build_dmub_cmd_list() - Build an array of DMCUB commands to be sent to DMCUB
+ *
+ * @dc: Current DC state
+ * @srf_updates: Array of surface updates
+ * @surface_count: Number of surfaces that have an updated
+ * @stream: Corresponding stream to be updated in the current flip
+ * @context: New DC state to be programmed
+ *
+ * @dc_dmub_cmd: Array of DMCUB commands to be sent to DMCUB
+ * @dmub_cmd_count: Count indicating the number of DMCUB commands in dc_dmub_cmd array
+ *
+ * This function builds an array of DMCUB commands to be sent to DMCUB. This function is required
+ * to build an array of commands and have them sent while the OTG lock is acquired.
+ *
+ * Return: void
+ */
+static void build_dmub_cmd_list(struct dc *dc,
+		struct dc_surface_update *srf_updates,
+		int surface_count,
+		struct dc_stream_state *stream,
+		struct dc_state *context,
+		struct dc_dmub_cmd dc_dmub_cmd[],
+		unsigned int *dmub_cmd_count)
+{
+	// Initialize cmd count to 0
+	*dmub_cmd_count = 0;
+	build_dmub_update_dirty_rect(dc, surface_count, stream, srf_updates, context, dc_dmub_cmd, dmub_cmd_count);
+}
+
+static void commit_plane_for_stream_offload_fams2_flip(struct dc *dc,
+		struct dc_surface_update *srf_updates,
+		int surface_count,
+		struct dc_stream_state *stream,
+		struct dc_state *context)
+{
+	int i, j;
+
+	/* update dirty rect for PSR */
+	dc_dmub_update_dirty_rect(dc, surface_count, stream,
+			srf_updates, context);
+
+	/* Perform requested Updates */
+	for (i = 0; i < surface_count; i++) {
+		struct dc_plane_state *plane_state = srf_updates[i].surface;
+
+		for (j = 0; j < dc->res_pool->pipe_count; j++) {
+			struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[j];
+
+			if (!should_update_pipe_for_stream(context, pipe_ctx, stream))
+				continue;
+
+			if (!should_update_pipe_for_plane(context, pipe_ctx, plane_state))
+				continue;
+
+			/* update pipe context for plane */
+			if (pipe_ctx->plane_state->update_flags.bits.addr_update)
+				dc->hwss.update_plane_addr(dc, pipe_ctx);
+		}
+	}
+
+	/* Send commands to DMCUB */
+	dc_dmub_srv_fams2_passthrough_flip(dc,
+				context,
+				stream,
+				srf_updates,
+				surface_count);
+}
+
+static void commit_planes_for_stream_fast(struct dc *dc,
+		struct dc_surface_update *srf_updates,
+		int surface_count,
+		struct dc_stream_state *stream,
+		struct dc_stream_update *stream_update,
+		enum surface_update_type update_type,
+		struct dc_state *context)
+{
+	int i, j;
+	struct pipe_ctx *top_pipe_to_program = NULL;
+	struct dc_stream_status *stream_status = NULL;
+	bool should_offload_fams2_flip = false;
+	bool should_lock_all_pipes = (update_type != UPDATE_TYPE_FAST);
+
+	if (should_lock_all_pipes)
+		determine_pipe_unlock_order(dc, context);
+
+	if (dc->debug.fams2_config.bits.enable &&
+			dc->debug.fams2_config.bits.enable_offload_flip &&
+			dc_state_is_fams2_in_use(dc, context)) {
+		/* if not offloading to HWFQ, offload to FAMS2 if needed */
+		should_offload_fams2_flip = true;
+		for (i = 0; i < surface_count; i++) {
+			if (srf_updates[i].surface &&
+					srf_updates[i].surface->update_flags.raw &&
+					!check_address_only_update(srf_updates[i].surface->update_flags)) {
+				/* more than address update, need to acquire FAMS2 lock */
+				should_offload_fams2_flip = false;
+				break;
+			}
+		}
+		if (stream_update) {
+			/* more than address update, need to acquire FAMS2 lock */
+			should_offload_fams2_flip = false;
+		}
+	}
+
+	dc_exit_ips_for_hw_access(dc);
+
+	dc_z10_restore(dc);
+
+	top_pipe_to_program = resource_get_otg_master_for_stream(
+			&context->res_ctx,
+			stream);
+
+	if (!top_pipe_to_program)
+		return;
+
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
+
+		if (pipe->stream && pipe->plane_state) {
+			if (!dc->debug.using_dml2)
+				set_p_state_switch_method(dc, context, pipe);
+
+			if (dc->debug.visual_confirm)
+				dc_update_visual_confirm_color(dc, context, pipe);
+		}
+	}
+
+	for (i = 0; i < surface_count; i++) {
+		struct dc_plane_state *plane_state = srf_updates[i].surface;
+		/*set logical flag for lock/unlock use*/
+		for (j = 0; j < dc->res_pool->pipe_count; j++) {
+			struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[j];
+
+			if (!pipe_ctx->plane_state)
+				continue;
+			if (!should_update_pipe_for_plane(context, pipe_ctx, plane_state))
+				continue;
+
+			pipe_ctx->plane_state->triplebuffer_flips = false;
+			if (update_type == UPDATE_TYPE_FAST &&
+					dc->hwss.program_triplebuffer != NULL &&
+					!pipe_ctx->plane_state->flip_immediate && dc->debug.enable_tri_buf) {
+				/*triple buffer for VUpdate only*/
+				pipe_ctx->plane_state->triplebuffer_flips = true;
+			}
+		}
+	}
+
+	stream_status = dc_state_get_stream_status(context, stream);
+
+	if (should_offload_fams2_flip) {
+		commit_plane_for_stream_offload_fams2_flip(dc,
+				srf_updates,
+				surface_count,
+				stream,
+				context);
+	} else if (stream_status) {
+		build_dmub_cmd_list(dc,
+				srf_updates,
+				surface_count,
+				stream,
+				context,
+				context->dc_dmub_cmd,
+				&(context->dmub_cmd_count));
+		hwss_build_fast_sequence(dc,
+				context->dc_dmub_cmd,
+				context->dmub_cmd_count,
+				context->block_sequence,
+				&(context->block_sequence_steps),
+				top_pipe_to_program,
+				stream_status,
+				context);
+		hwss_execute_sequence(dc,
+				context->block_sequence,
+				context->block_sequence_steps);
+	}
+
+	/* Clear update flags so next flip doesn't have redundant programming
+	 * (if there's no stream update, the update flags are not cleared).
+	 * Surface updates are cleared unconditionally at the beginning of each flip,
+	 * so no need to clear here.
+	 */
+	if (top_pipe_to_program->stream)
+		top_pipe_to_program->stream->update_flags.raw = 0;
+}
+
+static void commit_planes_for_stream(struct dc *dc,
+		struct dc_surface_update *srf_updates,
+		int surface_count,
+		struct dc_stream_state *stream,
+		struct dc_stream_update *stream_update,
+		enum surface_update_type update_type,
+		struct dc_state *context)
+{
+	int i, j;
+	struct pipe_ctx *top_pipe_to_program = NULL;
+	bool should_lock_all_pipes = (update_type != UPDATE_TYPE_FAST);
+	bool subvp_prev_use = false;
+	bool subvp_curr_use = false;
+	uint8_t current_stream_mask = 0;
+
+	if (should_lock_all_pipes)
+		determine_pipe_unlock_order(dc, context);
+	// Once we apply the new subvp context to hardware it won't be in the
+	// dc->current_state anymore, so we have to cache it before we apply
+	// the new SubVP context
+	subvp_prev_use = false;
+	dc_exit_ips_for_hw_access(dc);
+
+	dc_z10_restore(dc);
+	if (update_type == UPDATE_TYPE_FULL && dc->optimized_required)
+		hwss_process_outstanding_hw_updates(dc, dc->current_state);
+
+	if (update_type != UPDATE_TYPE_FAST && dc->res_pool->funcs->prepare_mcache_programming)
+		dc->res_pool->funcs->prepare_mcache_programming(dc, context);
+
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
+
+		if (pipe->stream && pipe->plane_state) {
+			if (!dc->debug.using_dml2)
+				set_p_state_switch_method(dc, context, pipe);
+
+			if (dc->debug.visual_confirm)
+				dc_update_visual_confirm_color(dc, context, pipe);
+		}
+	}
+
+	if (update_type == UPDATE_TYPE_FULL) {
+		dc_allow_idle_optimizations(dc, false);
+
+		if (get_seamless_boot_stream_count(context) == 0)
+			dc->hwss.prepare_bandwidth(dc, context);
+
+		if (dc->hwss.update_dsc_pg)
+			dc->hwss.update_dsc_pg(dc, context, false);
+
+		context_clock_trace(dc, context);
+	}
+
+	if (update_type == UPDATE_TYPE_FULL)
+		hwss_wait_for_outstanding_hw_updates(dc, dc->current_state);
+
+	top_pipe_to_program = resource_get_otg_master_for_stream(
+				&context->res_ctx,
+				stream);
+	ASSERT(top_pipe_to_program != NULL);
+
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *old_pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+
+		// Check old context for SubVP
+		subvp_prev_use |= (dc_state_get_pipe_subvp_type(dc->current_state, old_pipe) == SUBVP_PHANTOM);
+		if (subvp_prev_use)
+			break;
+	}
+
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
+
+		if (dc_state_get_pipe_subvp_type(context, pipe) == SUBVP_PHANTOM) {
+			subvp_curr_use = true;
+			break;
+		}
+	}
+
+	if (stream->test_pattern.type != DP_TEST_PATTERN_VIDEO_MODE) {
+		struct pipe_ctx *mpcc_pipe;
+		struct pipe_ctx *odm_pipe;
+
+		for (mpcc_pipe = top_pipe_to_program; mpcc_pipe; mpcc_pipe = mpcc_pipe->bottom_pipe)
+			for (odm_pipe = mpcc_pipe; odm_pipe; odm_pipe = odm_pipe->next_odm_pipe)
+				odm_pipe->ttu_regs.min_ttu_vblank = MAX_TTU;
+	}
+
+	if ((update_type != UPDATE_TYPE_FAST) && stream->update_flags.bits.dsc_changed)
+		if (top_pipe_to_program &&
+			top_pipe_to_program->stream_res.tg->funcs->lock_doublebuffer_enable) {
+			if (should_use_dmub_lock(stream->link)) {
+				union dmub_hw_lock_flags hw_locks = { 0 };
+				struct dmub_hw_lock_inst_flags inst_flags = { 0 };
+
+				hw_locks.bits.lock_dig = 1;
+				inst_flags.dig_inst = top_pipe_to_program->stream_res.tg->inst;
+
+				dmub_hw_lock_mgr_cmd(dc->ctx->dmub_srv,
+							true,
+							&hw_locks,
+							&inst_flags);
+			} else
+				top_pipe_to_program->stream_res.tg->funcs->lock_doublebuffer_enable(
+						top_pipe_to_program->stream_res.tg);
+		}
+
+	if (dc->hwss.wait_for_dcc_meta_propagation) {
+		dc->hwss.wait_for_dcc_meta_propagation(dc, top_pipe_to_program);
+	}
+
+	if (dc->hwseq->funcs.wait_for_pipe_update_if_needed)
+		dc->hwseq->funcs.wait_for_pipe_update_if_needed(dc, top_pipe_to_program, update_type < UPDATE_TYPE_FULL);
+
+	if (should_lock_all_pipes && dc->hwss.interdependent_update_lock) {
+		if (dc->hwss.subvp_pipe_control_lock)
+			dc->hwss.subvp_pipe_control_lock(dc, context, true, should_lock_all_pipes, NULL, subvp_prev_use);
+
+		if (dc->hwss.fams2_global_control_lock)
+			dc->hwss.fams2_global_control_lock(dc, context, true);
+
+		dc->hwss.interdependent_update_lock(dc, context, true);
+	} else {
+		if (dc->hwss.subvp_pipe_control_lock)
+			dc->hwss.subvp_pipe_control_lock(dc, context, true, should_lock_all_pipes, top_pipe_to_program, subvp_prev_use);
+
+		if (dc->hwss.fams2_global_control_lock)
+			dc->hwss.fams2_global_control_lock(dc, context, true);
+
+		/* Lock the top pipe while updating plane addrs, since freesync requires
+		 *  plane addr update event triggers to be synchronized.
+		 *  top_pipe_to_program is expected to never be NULL
+		 */
+		dc->hwss.pipe_control_lock(dc, top_pipe_to_program, true);
+	}
+
+	dc_dmub_update_dirty_rect(dc, surface_count, stream, srf_updates, context);
+
+	// Stream updates
+	if (stream_update)
+		commit_planes_do_stream_update(dc, stream, stream_update, update_type, context);
+
+	if (surface_count == 0) {
+		/*
+		 * In case of turning off screen, no need to program front end a second time.
+		 * just return after program blank.
+		 */
+		if (dc->hwss.apply_ctx_for_surface)
+			dc->hwss.apply_ctx_for_surface(dc, stream, 0, context);
+		if (dc->hwss.program_front_end_for_ctx)
+			dc->hwss.program_front_end_for_ctx(dc, context);
+
+		if (should_lock_all_pipes && dc->hwss.interdependent_update_lock) {
+			dc->hwss.interdependent_update_lock(dc, context, false);
+		} else {
+			dc->hwss.pipe_control_lock(dc, top_pipe_to_program, false);
+		}
+		dc->hwss.post_unlock_program_front_end(dc, context);
+
+		if (update_type != UPDATE_TYPE_FAST)
+			if (dc->hwss.commit_subvp_config)
+				dc->hwss.commit_subvp_config(dc, context);
+
+		/* Since phantom pipe programming is moved to post_unlock_program_front_end,
+		 * move the SubVP lock to after the phantom pipes have been setup
+		 */
+		if (dc->hwss.subvp_pipe_control_lock)
+			dc->hwss.subvp_pipe_control_lock(dc, context, false, should_lock_all_pipes,
+							 NULL, subvp_prev_use);
+
+		if (dc->hwss.fams2_global_control_lock)
+			dc->hwss.fams2_global_control_lock(dc, context, false);
+
+		return;
+	}
+
+	if (update_type != UPDATE_TYPE_FAST) {
+		for (j = 0; j < dc->res_pool->pipe_count; j++) {
+			struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[j];
+
+			if ((dc->debug.visual_confirm == VISUAL_CONFIRM_SUBVP ||
+				dc->debug.visual_confirm == VISUAL_CONFIRM_MCLK_SWITCH) &&
+				pipe_ctx->stream && pipe_ctx->plane_state) {
+				/* Only update visual confirm for SUBVP and Mclk switching here.
+				 * The bar appears on all pipes, so we need to update the bar on all displays,
+				 * so the information doesn't get stale.
+				 */
+				dc->hwss.update_visual_confirm_color(dc, pipe_ctx,
+						pipe_ctx->plane_res.hubp->inst);
+			}
+		}
+	}
+
+	for (i = 0; i < surface_count; i++) {
+		struct dc_plane_state *plane_state = srf_updates[i].surface;
+
+		/*set logical flag for lock/unlock use*/
+		for (j = 0; j < dc->res_pool->pipe_count; j++) {
+			struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[j];
+			if (!pipe_ctx->plane_state)
+				continue;
+			if (!should_update_pipe_for_plane(context, pipe_ctx, plane_state))
+				continue;
+			pipe_ctx->plane_state->triplebuffer_flips = false;
+			if (update_type == UPDATE_TYPE_FAST &&
+					dc->hwss.program_triplebuffer != NULL &&
+					!pipe_ctx->plane_state->flip_immediate && dc->debug.enable_tri_buf) {
+				/*triple buffer for VUpdate only*/
+				pipe_ctx->plane_state->triplebuffer_flips = true;
+			}
+		}
+		if (update_type == UPDATE_TYPE_FULL) {
+			/* force vsync flip when reconfiguring pipes to prevent underflow */
+			plane_state->flip_immediate = false;
+			plane_state->triplebuffer_flips = false;
+		}
+	}
+
+	// Update Type FULL, Surface updates
+	for (j = 0; j < dc->res_pool->pipe_count; j++) {
+		struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[j];
+
+		if (!pipe_ctx->top_pipe &&
+			!pipe_ctx->prev_odm_pipe &&
+			should_update_pipe_for_stream(context, pipe_ctx, stream)) {
+			struct dc_stream_status *stream_status = NULL;
+
+			if (!pipe_ctx->plane_state)
+				continue;
+
+			/* Full fe update*/
+			if (update_type == UPDATE_TYPE_FAST)
+				continue;
+
+			stream_status =
+				stream_get_status(context, pipe_ctx->stream);
+
+			if (dc->hwss.apply_ctx_for_surface && stream_status)
+				dc->hwss.apply_ctx_for_surface(
+					dc, pipe_ctx->stream, stream_status->plane_count, context);
+		}
+	}
+
+	for (j = 0; j < dc->res_pool->pipe_count; j++) {
+		struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[j];
+
+		if (!pipe_ctx->plane_state)
+			continue;
+
+		/* Full fe update*/
+		if (update_type == UPDATE_TYPE_FAST)
+			continue;
+
+		ASSERT(!pipe_ctx->plane_state->triplebuffer_flips);
+		if (dc->hwss.program_triplebuffer != NULL && dc->debug.enable_tri_buf) {
+			/*turn off triple buffer for full update*/
+			dc->hwss.program_triplebuffer(
+				dc, pipe_ctx, pipe_ctx->plane_state->triplebuffer_flips);
+		}
+	}
+
+	if (dc->hwss.program_front_end_for_ctx && update_type != UPDATE_TYPE_FAST) {
+		dc->hwss.program_front_end_for_ctx(dc, context);
+
+		//Pipe busy until some frame and line #
+		if (dc->hwseq->funcs.set_wait_for_update_needed_for_pipe && update_type == UPDATE_TYPE_FULL) {
+			for (j = 0; j < dc->res_pool->pipe_count; j++) {
+				struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[j];
+
+				dc->hwseq->funcs.set_wait_for_update_needed_for_pipe(dc, pipe_ctx);
+			}
+		}
+
+		if (dc->debug.validate_dml_output) {
+			for (i = 0; i < dc->res_pool->pipe_count; i++) {
+				struct pipe_ctx *cur_pipe = &context->res_ctx.pipe_ctx[i];
+				if (cur_pipe->stream == NULL)
+					continue;
+
+				cur_pipe->plane_res.hubp->funcs->validate_dml_output(
+						cur_pipe->plane_res.hubp, dc->ctx,
+						&context->res_ctx.pipe_ctx[i].rq_regs,
+						&context->res_ctx.pipe_ctx[i].dlg_regs,
+						&context->res_ctx.pipe_ctx[i].ttu_regs);
+			}
+		}
+	}
+
+	// Update Type FAST, Surface updates
+	if (update_type == UPDATE_TYPE_FAST) {
+		if (dc->hwss.set_flip_control_gsl)
+			for (i = 0; i < surface_count; i++) {
+				struct dc_plane_state *plane_state = srf_updates[i].surface;
+
+				for (j = 0; j < dc->res_pool->pipe_count; j++) {
+					struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[j];
+
+					if (!should_update_pipe_for_stream(context, pipe_ctx, stream))
+						continue;
+
+					if (!should_update_pipe_for_plane(context, pipe_ctx, plane_state))
+						continue;
+
+					// GSL has to be used for flip immediate
+					dc->hwss.set_flip_control_gsl(pipe_ctx,
+							pipe_ctx->plane_state->flip_immediate);
+				}
+			}
+
+		/* Perform requested Updates */
+		for (i = 0; i < surface_count; i++) {
+			struct dc_plane_state *plane_state = srf_updates[i].surface;
+
+			for (j = 0; j < dc->res_pool->pipe_count; j++) {
+				struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[j];
+
+				if (!should_update_pipe_for_stream(context, pipe_ctx, stream))
+					continue;
+
+				if (!should_update_pipe_for_plane(context, pipe_ctx, plane_state))
+					continue;
+
+				if (srf_updates[i].cm2_params &&
+						srf_updates[i].cm2_params->cm2_luts.lut3d_data.lut3d_src ==
+								DC_CM2_TRANSFER_FUNC_SOURCE_VIDMEM &&
+						srf_updates[i].cm2_params->component_settings.shaper_3dlut_setting ==
+								DC_CM2_SHAPER_3DLUT_SETTING_ENABLE_SHAPER_3DLUT &&
+						dc->hwss.trigger_3dlut_dma_load)
+					dc->hwss.trigger_3dlut_dma_load(dc, pipe_ctx);
+
+				/*program triple buffer after lock based on flip type*/
+				if (dc->hwss.program_triplebuffer != NULL && dc->debug.enable_tri_buf) {
+					/*only enable triplebuffer for fast_update*/
+					dc->hwss.program_triplebuffer(
+						dc, pipe_ctx, pipe_ctx->plane_state->triplebuffer_flips);
+				}
+				if (pipe_ctx->plane_state->update_flags.bits.addr_update)
+					dc->hwss.update_plane_addr(dc, pipe_ctx);
+			}
+		}
+	}
+
+	if (should_lock_all_pipes && dc->hwss.interdependent_update_lock) {
+		dc->hwss.interdependent_update_lock(dc, context, false);
+	} else {
+		dc->hwss.pipe_control_lock(dc, top_pipe_to_program, false);
+	}
+
+	if ((update_type != UPDATE_TYPE_FAST) && stream->update_flags.bits.dsc_changed)
+		if (top_pipe_to_program &&
+		    top_pipe_to_program->stream_res.tg->funcs->lock_doublebuffer_enable) {
+			top_pipe_to_program->stream_res.tg->funcs->wait_for_state(
+				top_pipe_to_program->stream_res.tg,
+				CRTC_STATE_VACTIVE);
+			top_pipe_to_program->stream_res.tg->funcs->wait_for_state(
+				top_pipe_to_program->stream_res.tg,
+				CRTC_STATE_VBLANK);
+			top_pipe_to_program->stream_res.tg->funcs->wait_for_state(
+				top_pipe_to_program->stream_res.tg,
+				CRTC_STATE_VACTIVE);
+
+			if (should_use_dmub_lock(stream->link)) {
+				union dmub_hw_lock_flags hw_locks = { 0 };
+				struct dmub_hw_lock_inst_flags inst_flags = { 0 };
+
+				hw_locks.bits.lock_dig = 1;
+				inst_flags.dig_inst = top_pipe_to_program->stream_res.tg->inst;
+
+				dmub_hw_lock_mgr_cmd(dc->ctx->dmub_srv,
+							false,
+							&hw_locks,
+							&inst_flags);
+			} else
+				top_pipe_to_program->stream_res.tg->funcs->lock_doublebuffer_disable(
+					top_pipe_to_program->stream_res.tg);
+		}
+
+	if (subvp_curr_use) {
+		/* If enabling subvp or transitioning from subvp->subvp, enable the
+		 * phantom streams before we program front end for the phantom pipes.
+		 */
+		if (update_type != UPDATE_TYPE_FAST) {
+			if (dc->hwss.enable_phantom_streams)
+				dc->hwss.enable_phantom_streams(dc, context);
+		}
+	}
+
+	if (update_type != UPDATE_TYPE_FAST)
+		dc->hwss.post_unlock_program_front_end(dc, context);
+
+	if (subvp_prev_use && !subvp_curr_use) {
+		/* If disabling subvp, disable phantom streams after front end
+		 * programming has completed (we turn on phantom OTG in order
+		 * to complete the plane disable for phantom pipes).
+		 */
+
+		if (dc->hwss.disable_phantom_streams)
+			dc->hwss.disable_phantom_streams(dc, context);
+	}
+
+	if (update_type != UPDATE_TYPE_FAST)
+		if (dc->hwss.commit_subvp_config)
+			dc->hwss.commit_subvp_config(dc, context);
+	/* Since phantom pipe programming is moved to post_unlock_program_front_end,
+	 * move the SubVP lock to after the phantom pipes have been setup
+	 */
+	if (should_lock_all_pipes && dc->hwss.interdependent_update_lock) {
+		if (dc->hwss.subvp_pipe_control_lock)
+			dc->hwss.subvp_pipe_control_lock(dc, context, false, should_lock_all_pipes, NULL, subvp_prev_use);
+		if (dc->hwss.fams2_global_control_lock)
+			dc->hwss.fams2_global_control_lock(dc, context, false);
+	} else {
+		if (dc->hwss.subvp_pipe_control_lock)
+			dc->hwss.subvp_pipe_control_lock(dc, context, false, should_lock_all_pipes, top_pipe_to_program, subvp_prev_use);
+		if (dc->hwss.fams2_global_control_lock)
+			dc->hwss.fams2_global_control_lock(dc, context, false);
+	}
+
+	// Fire manual trigger only when bottom plane is flipped
+	for (j = 0; j < dc->res_pool->pipe_count; j++) {
+		struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[j];
+
+		if (!pipe_ctx->plane_state)
+			continue;
+
+		if (pipe_ctx->bottom_pipe || pipe_ctx->next_odm_pipe ||
+				!pipe_ctx->stream || !should_update_pipe_for_stream(context, pipe_ctx, stream) ||
+				!pipe_ctx->plane_state->update_flags.bits.addr_update ||
+				pipe_ctx->plane_state->skip_manual_trigger)
+			continue;
+
+		if (pipe_ctx->stream_res.tg->funcs->program_manual_trigger)
+			pipe_ctx->stream_res.tg->funcs->program_manual_trigger(pipe_ctx->stream_res.tg);
+	}
+
+	current_stream_mask = get_stream_mask(dc, context);
+	if (current_stream_mask != context->stream_mask) {
+		context->stream_mask = current_stream_mask;
+		dc_dmub_srv_notify_stream_mask(dc->ctx->dmub_srv, current_stream_mask);
+	}
+}
+
+/**
+ * could_mpcc_tree_change_for_active_pipes - Check if an OPP associated with MPCC might change
+ *
+ * @dc: Used to get the current state status
+ * @stream: Target stream, which we want to remove the attached planes
+ * @srf_updates: Array of surface updates
+ * @surface_count: Number of surface update
+ * @is_plane_addition: [in] Fill out with true if it is a plane addition case
+ *
+ * DCN32x and newer support a feature named Dynamic ODM which can conflict with
+ * the MPO if used simultaneously in some specific configurations (e.g.,
+ * 4k@144). This function checks if the incoming context requires applying a
+ * transition state with unnecessary pipe splitting and ODM disabled to
+ * circumvent our hardware limitations to prevent this edge case. If the OPP
+ * associated with an MPCC might change due to plane additions, this function
+ * returns true.
+ *
+ * Return:
+ * Return true if OPP and MPCC might change, otherwise, return false.
+ */
+static bool could_mpcc_tree_change_for_active_pipes(struct dc *dc,
+		struct dc_stream_state *stream,
+		struct dc_surface_update *srf_updates,
+		int surface_count,
+		bool *is_plane_addition)
+{
+
+	struct dc_stream_status *cur_stream_status = stream_get_status(dc->current_state, stream);
+	bool force_minimal_pipe_splitting = false;
+	bool subvp_active = false;
+	uint32_t i;
+
+	*is_plane_addition = false;
+
+	if (cur_stream_status &&
+			dc->current_state->stream_count > 0 &&
+			dc->debug.pipe_split_policy != MPC_SPLIT_AVOID) {
+		/* determine if minimal transition is required due to MPC*/
+		if (surface_count > 0) {
+			if (cur_stream_status->plane_count > surface_count) {
+				force_minimal_pipe_splitting = true;
+			} else if (cur_stream_status->plane_count < surface_count) {
+				force_minimal_pipe_splitting = true;
+				*is_plane_addition = true;
+			}
+		}
+	}
+
+	if (cur_stream_status &&
+			dc->current_state->stream_count == 1 &&
+			dc->debug.enable_single_display_2to1_odm_policy) {
+		/* determine if minimal transition is required due to dynamic ODM*/
+		if (surface_count > 0) {
+			if (cur_stream_status->plane_count > 2 && cur_stream_status->plane_count > surface_count) {
+				force_minimal_pipe_splitting = true;
+			} else if (surface_count > 2 && cur_stream_status->plane_count < surface_count) {
+				force_minimal_pipe_splitting = true;
+				*is_plane_addition = true;
+			}
+		}
+	}
+
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+
+		if (dc_state_get_pipe_subvp_type(dc->current_state, pipe) != SUBVP_NONE) {
+			subvp_active = true;
+			break;
+		}
+	}
+
+	/* For SubVP when adding or removing planes we need to add a minimal transition
+	 * (even when disabling all planes). Whenever disabling a phantom pipe, we
+	 * must use the minimal transition path to disable the pipe correctly.
+	 *
+	 * We want to use the minimal transition whenever subvp is active, not only if
+	 * a plane is being added / removed from a subvp stream (MPO plane can be added
+	 * to a DRR pipe of SubVP + DRR config, in which case we still want to run through
+	 * a min transition to disable subvp.
+	 */
+	if (cur_stream_status && subvp_active) {
+		/* determine if minimal transition is required due to SubVP*/
+		if (cur_stream_status->plane_count > surface_count) {
+			force_minimal_pipe_splitting = true;
+		} else if (cur_stream_status->plane_count < surface_count) {
+			force_minimal_pipe_splitting = true;
+			*is_plane_addition = true;
+		}
+	}
+
+	return force_minimal_pipe_splitting;
+}
+
+struct pipe_split_policy_backup {
+	bool dynamic_odm_policy;
+	bool subvp_policy;
+	enum pipe_split_policy mpc_policy;
+	char force_odm[MAX_PIPES];
+};
+
+static void backup_and_set_minimal_pipe_split_policy(struct dc *dc,
+		struct dc_state *context,
+		struct pipe_split_policy_backup *policy)
+{
+	int i;
+
+	if (!dc->config.is_vmin_only_asic) {
+		policy->mpc_policy = dc->debug.pipe_split_policy;
+		dc->debug.pipe_split_policy = MPC_SPLIT_AVOID;
+	}
+	policy->dynamic_odm_policy = dc->debug.enable_single_display_2to1_odm_policy;
+	dc->debug.enable_single_display_2to1_odm_policy = false;
+	policy->subvp_policy = dc->debug.force_disable_subvp;
+	dc->debug.force_disable_subvp = true;
+	for (i = 0; i < context->stream_count; i++) {
+		policy->force_odm[i] = context->streams[i]->debug.force_odm_combine_segments;
+		if (context->streams[i]->debug.allow_transition_for_forced_odm)
+			context->streams[i]->debug.force_odm_combine_segments = 0;
+	}
+}
+
+static void restore_minimal_pipe_split_policy(struct dc *dc,
+		struct dc_state *context,
+		struct pipe_split_policy_backup *policy)
+{
+	uint8_t i;
+
+	if (!dc->config.is_vmin_only_asic)
+		dc->debug.pipe_split_policy = policy->mpc_policy;
+	dc->debug.enable_single_display_2to1_odm_policy =
+			policy->dynamic_odm_policy;
+	dc->debug.force_disable_subvp = policy->subvp_policy;
+	for (i = 0; i < context->stream_count; i++)
+		context->streams[i]->debug.force_odm_combine_segments = policy->force_odm[i];
+}
+
+static void release_minimal_transition_state(struct dc *dc,
+		struct dc_state *minimal_transition_context,
+		struct dc_state *base_context,
+		struct pipe_split_policy_backup *policy)
+{
+	restore_minimal_pipe_split_policy(dc, base_context, policy);
+	dc_state_release(minimal_transition_context);
+}
+
+static void force_vsync_flip_in_minimal_transition_context(struct dc_state *context)
+{
+	uint8_t i;
+	int j;
+	struct dc_stream_status *stream_status;
+
+	for (i = 0; i < context->stream_count; i++) {
+		stream_status = &context->stream_status[i];
+
+		for (j = 0; j < stream_status->plane_count; j++)
+			stream_status->plane_states[j]->flip_immediate = false;
+	}
+}
+
+static struct dc_state *create_minimal_transition_state(struct dc *dc,
+		struct dc_state *base_context, struct pipe_split_policy_backup *policy)
+{
+	struct dc_state *minimal_transition_context = NULL;
+
+	minimal_transition_context = dc_state_create_copy(base_context);
+	if (!minimal_transition_context)
+		return NULL;
+
+	backup_and_set_minimal_pipe_split_policy(dc, base_context, policy);
+	/* commit minimal state */
+	if (dc->res_pool->funcs->validate_bandwidth(dc, minimal_transition_context,
+		DC_VALIDATE_MODE_AND_PROGRAMMING) == DC_OK) {
+		/* prevent underflow and corruption when reconfiguring pipes */
+		force_vsync_flip_in_minimal_transition_context(minimal_transition_context);
+	} else {
+		/*
+		 * This should never happen, minimal transition state should
+		 * always be validated first before adding pipe split features.
+		 */
+		release_minimal_transition_state(dc, minimal_transition_context, base_context, policy);
+		BREAK_TO_DEBUGGER();
+		minimal_transition_context = NULL;
+	}
+	return minimal_transition_context;
+}
+
+static bool is_pipe_topology_transition_seamless_with_intermediate_step(
+		struct dc *dc,
+		struct dc_state *initial_state,
+		struct dc_state *intermediate_state,
+		struct dc_state *final_state)
+{
+	return dc->hwss.is_pipe_topology_transition_seamless(dc, initial_state,
+			intermediate_state) &&
+			dc->hwss.is_pipe_topology_transition_seamless(dc,
+					intermediate_state, final_state);
+}
+
+static void swap_and_release_current_context(struct dc *dc,
+		struct dc_state *new_context, struct dc_stream_state *stream)
+{
+
+	int i;
+	struct dc_state *old = dc->current_state;
+	struct pipe_ctx *pipe_ctx;
+
+	/* Since memory free requires elevated IRQ, an interrupt
+	 * request is generated by mem free. If this happens
+	 * between freeing and reassigning the context, our vsync
+	 * interrupt will call into dc and cause a memory
+	 * corruption. Hence, we first reassign the context,
+	 * then free the old context.
+	 */
+	dc->current_state = new_context;
+	dc_state_release(old);
+
+	// clear any forced full updates
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		pipe_ctx = &new_context->res_ctx.pipe_ctx[i];
+
+		if (pipe_ctx->plane_state && pipe_ctx->stream == stream)
+			pipe_ctx->plane_state->force_full_update = false;
+	}
+}
+
+static int initialize_empty_surface_updates(
+		struct dc_stream_state *stream,
+		struct dc_surface_update *srf_updates)
+{
+	struct dc_stream_status *status = dc_stream_get_status(stream);
+	int i;
+
+	if (!status)
+		return 0;
+
+	for (i = 0; i < status->plane_count; i++)
+		srf_updates[i].surface = status->plane_states[i];
+
+	return status->plane_count;
+}
+
+static bool commit_minimal_transition_based_on_new_context(struct dc *dc,
+		struct dc_state *new_context,
+		struct dc_stream_state *stream,
+		struct dc_surface_update *srf_updates,
+		int surface_count)
+{
+	bool success = false;
+	struct pipe_split_policy_backup policy;
+	struct dc_state *intermediate_context =
+			create_minimal_transition_state(dc, new_context,
+					&policy);
+
+	if (intermediate_context) {
+		if (is_pipe_topology_transition_seamless_with_intermediate_step(
+				dc,
+				dc->current_state,
+				intermediate_context,
+				new_context)) {
+			DC_LOG_DC("commit minimal transition state: base = new state\n");
+			commit_planes_for_stream(dc, srf_updates,
+					surface_count, stream, NULL,
+					UPDATE_TYPE_FULL, intermediate_context);
+			swap_and_release_current_context(
+					dc, intermediate_context, stream);
+			dc_state_retain(dc->current_state);
+			success = true;
+		}
+		release_minimal_transition_state(
+				dc, intermediate_context, new_context, &policy);
+	}
+	return success;
+}
+
+static bool commit_minimal_transition_based_on_current_context(struct dc *dc,
+		struct dc_state *new_context, struct dc_stream_state *stream)
+{
+	bool success = false;
+	struct pipe_split_policy_backup policy;
+	struct dc_state *intermediate_context;
+	struct dc_state *old_current_state = dc->current_state;
+	struct dc_surface_update srf_updates[MAX_SURFACES] = {0};
+	int surface_count;
+
+	/*
+	 * Both current and new contexts share the same stream and plane state
+	 * pointers. When new context is validated, stream and planes get
+	 * populated with new updates such as new plane addresses. This makes
+	 * the current context no longer valid because stream and planes are
+	 * modified from the original. We backup current stream and plane states
+	 * into scratch space whenever we are populating new context. So we can
+	 * restore the original values back by calling the restore function now.
+	 * This restores back the original stream and plane states associated
+	 * with the current state.
+	 */
+	restore_planes_and_stream_state(&dc->scratch.current_state, stream);
+	dc_state_retain(old_current_state);
+	intermediate_context = create_minimal_transition_state(dc,
+			old_current_state, &policy);
+
+	if (intermediate_context) {
+		if (is_pipe_topology_transition_seamless_with_intermediate_step(
+				dc,
+				dc->current_state,
+				intermediate_context,
+				new_context)) {
+			DC_LOG_DC("commit minimal transition state: base = current state\n");
+			surface_count = initialize_empty_surface_updates(
+					stream, srf_updates);
+			commit_planes_for_stream(dc, srf_updates,
+					surface_count, stream, NULL,
+					UPDATE_TYPE_FULL, intermediate_context);
+			swap_and_release_current_context(
+					dc, intermediate_context, stream);
+			dc_state_retain(dc->current_state);
+			success = true;
+		}
+		release_minimal_transition_state(dc, intermediate_context,
+				old_current_state, &policy);
+	}
+	dc_state_release(old_current_state);
+	/*
+	 * Restore stream and plane states back to the values associated with
+	 * new context.
+	 */
+	restore_planes_and_stream_state(&dc->scratch.new_state, stream);
+	return success;
+}
+
+/**
+ * commit_minimal_transition_state_in_dc_update - Commit a minimal state based
+ * on current or new context
+ *
+ * @dc: DC structure, used to get the current state
+ * @new_context: New context
+ * @stream: Stream getting the update for the flip
+ * @srf_updates: Surface updates
+ * @surface_count: Number of surfaces
+ *
+ * The function takes in current state and new state and determine a minimal
+ * transition state as the intermediate step which could make the transition
+ * between current and new states seamless. If found, it will commit the minimal
+ * transition state and update current state to this minimal transition state
+ * and return true, if not, it will return false.
+ *
+ * Return:
+ * Return True if the minimal transition succeeded, false otherwise
+ */
+static bool commit_minimal_transition_state_in_dc_update(struct dc *dc,
+		struct dc_state *new_context,
+		struct dc_stream_state *stream,
+		struct dc_surface_update *srf_updates,
+		int surface_count)
+{
+	bool success = commit_minimal_transition_based_on_new_context(
+				dc, new_context, stream, srf_updates,
+				surface_count);
+	if (!success)
+		success = commit_minimal_transition_based_on_current_context(dc,
+				new_context, stream);
+	if (!success)
+		DC_LOG_ERROR("Fail to commit a seamless minimal transition state between current and new states.\nThis pipe topology update is non-seamless!\n");
+	return success;
+}
+
+/**
+ * commit_minimal_transition_state - Create a transition pipe split state
+ *
+ * @dc: Used to get the current state status
+ * @transition_base_context: New transition state
+ *
+ * In some specific configurations, such as pipe split on multi-display with
+ * MPO and/or Dynamic ODM, removing a plane may cause unsupported pipe
+ * programming when moving to new planes. To mitigate those types of problems,
+ * this function adds a transition state that minimizes pipe usage before
+ * programming the new configuration. When adding a new plane, the current
+ * state requires the least pipes, so it is applied without splitting. When
+ * removing a plane, the new state requires the least pipes, so it is applied
+ * without splitting.
+ *
+ * Return:
+ * Return false if something is wrong in the transition state.
+ */
+static bool commit_minimal_transition_state(struct dc *dc,
+		struct dc_state *transition_base_context)
+{
+	struct dc_state *transition_context;
+	struct pipe_split_policy_backup policy;
+	enum dc_status ret = DC_ERROR_UNEXPECTED;
+	unsigned int i, j;
+	unsigned int pipe_in_use = 0;
+	bool subvp_in_use = false;
+	bool odm_in_use = false;
+
+	/* check current pipes in use*/
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *pipe = &transition_base_context->res_ctx.pipe_ctx[i];
+
+		if (pipe->plane_state)
+			pipe_in_use++;
+	}
+
+	/* If SubVP is enabled and we are adding or removing planes from any main subvp
+	 * pipe, we must use the minimal transition.
+	 */
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+
+		if (pipe->stream && dc_state_get_pipe_subvp_type(dc->current_state, pipe) == SUBVP_PHANTOM) {
+			subvp_in_use = true;
+			break;
+		}
+	}
+
+	/* If ODM is enabled and we are adding or removing planes from any ODM
+	 * pipe, we must use the minimal transition.
+	 */
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *pipe = &transition_base_context->res_ctx.pipe_ctx[i];
+
+		if (resource_is_pipe_type(pipe, OTG_MASTER)) {
+			odm_in_use = resource_get_odm_slice_count(pipe) > 1;
+			break;
+		}
+	}
+
+	/* When the OS add a new surface if we have been used all of pipes with odm combine
+	 * and mpc split feature, it need use commit_minimal_transition_state to transition safely.
+	 * After OS exit MPO, it will back to use odm and mpc split with all of pipes, we need
+	 * call it again. Otherwise return true to skip.
+	 *
+	 * Reduce the scenarios to use dc_commit_state_no_check in the stage of flip. Especially
+	 * enter/exit MPO when DCN still have enough resources.
+	 */
+	if (pipe_in_use != dc->res_pool->pipe_count && !subvp_in_use && !odm_in_use)
+		return true;
+
+	DC_LOG_DC("%s base = %s state, reason = %s\n", __func__,
+			dc->current_state == transition_base_context ? "current" : "new",
+			subvp_in_use ? "Subvp In Use" :
+			odm_in_use ? "ODM in Use" :
+			dc->debug.pipe_split_policy != MPC_SPLIT_AVOID ? "MPC in Use" :
+			"Unknown");
+
+	dc_state_retain(transition_base_context);
+	transition_context = create_minimal_transition_state(dc,
+			transition_base_context, &policy);
+	if (transition_context) {
+		ret = dc_commit_state_no_check(dc, transition_context);
+		release_minimal_transition_state(dc, transition_context, transition_base_context, &policy);
+	}
+	dc_state_release(transition_base_context);
+
+	if (ret != DC_OK) {
+		/* this should never happen */
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+
+	/* force full surface update */
+	for (i = 0; i < dc->current_state->stream_count; i++) {
+		for (j = 0; j < dc->current_state->stream_status[i].plane_count; j++) {
+			dc->current_state->stream_status[i].plane_states[j]->update_flags.raw = 0xFFFFFFFF;
+		}
+	}
+
+	return true;
+}
+
+void populate_fast_updates(struct dc_fast_update *fast_update,
+		struct dc_surface_update *srf_updates,
+		int surface_count,
+		struct dc_stream_update *stream_update)
+{
+	int i = 0;
+
+	if (stream_update) {
+		fast_update[0].out_transfer_func = stream_update->out_transfer_func;
+		fast_update[0].output_csc_transform = stream_update->output_csc_transform;
+	} else {
+		fast_update[0].out_transfer_func = NULL;
+		fast_update[0].output_csc_transform = NULL;
+	}
+
+	for (i = 0; i < surface_count; i++) {
+		fast_update[i].flip_addr = srf_updates[i].flip_addr;
+		fast_update[i].gamma = srf_updates[i].gamma;
+		fast_update[i].gamut_remap_matrix = srf_updates[i].gamut_remap_matrix;
+		fast_update[i].input_csc_color_matrix = srf_updates[i].input_csc_color_matrix;
+		fast_update[i].coeff_reduction_factor = srf_updates[i].coeff_reduction_factor;
+		fast_update[i].cursor_csc_color_matrix = srf_updates[i].cursor_csc_color_matrix;
+	}
+}
+
+static bool fast_updates_exist(struct dc_fast_update *fast_update, int surface_count)
+{
+	int i;
+
+	if (fast_update[0].out_transfer_func ||
+		fast_update[0].output_csc_transform)
+		return true;
+
+	for (i = 0; i < surface_count; i++) {
+		if (fast_update[i].flip_addr ||
+				fast_update[i].gamma ||
+				fast_update[i].gamut_remap_matrix ||
+				fast_update[i].input_csc_color_matrix ||
+				fast_update[i].cursor_csc_color_matrix ||
+				fast_update[i].coeff_reduction_factor)
+			return true;
+	}
+
+	return false;
+}
+
+bool fast_nonaddr_updates_exist(struct dc_fast_update *fast_update, int surface_count)
+{
+	int i;
+
+	if (fast_update[0].out_transfer_func ||
+		fast_update[0].output_csc_transform)
+		return true;
+
+	for (i = 0; i < surface_count; i++) {
+		if (fast_update[i].input_csc_color_matrix ||
+				fast_update[i].gamma ||
+				fast_update[i].gamut_remap_matrix ||
+				fast_update[i].coeff_reduction_factor ||
+				fast_update[i].cursor_csc_color_matrix)
+			return true;
+	}
+
+	return false;
+}
+
+static bool full_update_required(struct dc *dc,
+		struct dc_surface_update *srf_updates,
+		int surface_count,
+		struct dc_stream_update *stream_update,
+		struct dc_stream_state *stream)
+{
+
+	int i;
+	struct dc_stream_status *stream_status;
+	const struct dc_state *context = dc->current_state;
+
+	for (i = 0; i < surface_count; i++) {
+		if (srf_updates &&
+				(srf_updates[i].plane_info ||
+				srf_updates[i].scaling_info ||
+				(srf_updates[i].hdr_mult.value &&
+				srf_updates[i].hdr_mult.value != srf_updates->surface->hdr_mult.value) ||
+				(srf_updates[i].sdr_white_level_nits &&
+				srf_updates[i].sdr_white_level_nits != srf_updates->surface->sdr_white_level_nits) ||
+				srf_updates[i].in_transfer_func ||
+				srf_updates[i].func_shaper ||
+				srf_updates[i].lut3d_func ||
+				srf_updates[i].surface->force_full_update ||
+				(srf_updates[i].flip_addr &&
+				srf_updates[i].flip_addr->address.tmz_surface != srf_updates[i].surface->address.tmz_surface) ||
+				(srf_updates[i].cm2_params &&
+				 (srf_updates[i].cm2_params->component_settings.shaper_3dlut_setting != srf_updates[i].surface->mcm_shaper_3dlut_setting ||
+				  srf_updates[i].cm2_params->component_settings.lut1d_enable != srf_updates[i].surface->mcm_lut1d_enable)) ||
+				!is_surface_in_context(context, srf_updates[i].surface)))
+			return true;
+	}
+
+	if (stream_update &&
+			(((stream_update->src.height != 0 && stream_update->src.width != 0) ||
+			(stream_update->dst.height != 0 && stream_update->dst.width != 0) ||
+			stream_update->integer_scaling_update) ||
+			stream_update->hdr_static_metadata ||
+			stream_update->abm_level ||
+			stream_update->periodic_interrupt ||
+			stream_update->vrr_infopacket ||
+			stream_update->vsc_infopacket ||
+			stream_update->vsp_infopacket ||
+			stream_update->hfvsif_infopacket ||
+			stream_update->vtem_infopacket ||
+			stream_update->adaptive_sync_infopacket ||
+			stream_update->avi_infopacket ||
+			stream_update->dpms_off ||
+			stream_update->allow_freesync ||
+			stream_update->vrr_active_variable ||
+			stream_update->vrr_active_fixed ||
+			stream_update->gamut_remap ||
+			stream_update->output_color_space ||
+			stream_update->dither_option ||
+			stream_update->wb_update ||
+			stream_update->dsc_config ||
+			stream_update->mst_bw_update ||
+			stream_update->func_shaper ||
+			stream_update->lut3d_func ||
+			stream_update->pending_test_pattern ||
+			stream_update->crtc_timing_adjust ||
+			stream_update->scaler_sharpener_update ||
+			stream_update->hw_cursor_req))
+		return true;
+
+	if (stream) {
+		stream_status = dc_stream_get_status(stream);
+		if (stream_status == NULL || stream_status->plane_count != surface_count)
+			return true;
+	}
+	if (dc->idle_optimizations_allowed)
+		return true;
+
+	if (dc_can_clear_cursor_limit(dc))
+		return true;
+
+	return false;
+}
+
+static bool fast_update_only(struct dc *dc,
+		struct dc_fast_update *fast_update,
+		struct dc_surface_update *srf_updates,
+		int surface_count,
+		struct dc_stream_update *stream_update,
+		struct dc_stream_state *stream)
+{
+	return fast_updates_exist(fast_update, surface_count)
+			&& !full_update_required(dc, srf_updates, surface_count, stream_update, stream);
+}
+
+static bool update_planes_and_stream_v2(struct dc *dc,
+		struct dc_surface_update *srf_updates, int surface_count,
+		struct dc_stream_state *stream,
+		struct dc_stream_update *stream_update)
+{
+	struct dc_state *context;
+	enum surface_update_type update_type;
+	struct dc_fast_update fast_update[MAX_SURFACES] = {0};
+
+	/* In cases where MPO and split or ODM are used transitions can
+	 * cause underflow. Apply stream configuration with minimal pipe
+	 * split first to avoid unsupported transitions for active pipes.
+	 */
+	bool force_minimal_pipe_splitting = 0;
+	bool is_plane_addition = 0;
+	bool is_fast_update_only;
+
+	populate_fast_updates(fast_update, srf_updates, surface_count, stream_update);
+	is_fast_update_only = fast_update_only(dc, fast_update, srf_updates,
+			surface_count, stream_update, stream);
+	force_minimal_pipe_splitting = could_mpcc_tree_change_for_active_pipes(
+			dc,
+			stream,
+			srf_updates,
+			surface_count,
+			&is_plane_addition);
+
+	/* on plane addition, minimal state is the current one */
+	if (force_minimal_pipe_splitting && is_plane_addition &&
+		!commit_minimal_transition_state(dc, dc->current_state))
+		return false;
+
+	if (!update_planes_and_stream_state(
+			dc,
+			srf_updates,
+			surface_count,
+			stream,
+			stream_update,
+			&update_type,
+			&context))
+		return false;
+
+	/* on plane removal, minimal state is the new one */
+	if (force_minimal_pipe_splitting && !is_plane_addition) {
+		if (!commit_minimal_transition_state(dc, context)) {
+			dc_state_release(context);
+			return false;
+		}
+		update_type = UPDATE_TYPE_FULL;
+	}
+
+	if (dc->hwss.is_pipe_topology_transition_seamless &&
+			!dc->hwss.is_pipe_topology_transition_seamless(
+					dc, dc->current_state, context))
+		commit_minimal_transition_state_in_dc_update(dc, context, stream,
+				srf_updates, surface_count);
+
+	if (is_fast_update_only && !dc->debug.enable_legacy_fast_update) {
+		commit_planes_for_stream_fast(dc,
+				srf_updates,
+				surface_count,
+				stream,
+				stream_update,
+				update_type,
+				context);
+	} else {
+		if (!stream_update &&
+				dc->hwss.is_pipe_topology_transition_seamless &&
+				!dc->hwss.is_pipe_topology_transition_seamless(
+						dc, dc->current_state, context)) {
+			DC_LOG_ERROR("performing non-seamless pipe topology transition with surface only update!\n");
+			BREAK_TO_DEBUGGER();
+		}
+		commit_planes_for_stream(
+				dc,
+				srf_updates,
+				surface_count,
+				stream,
+				stream_update,
+				update_type,
+				context);
+	}
+	if (dc->current_state != context)
+		swap_and_release_current_context(dc, context, stream);
+	return true;
+}
+
+static void commit_planes_and_stream_update_on_current_context(struct dc *dc,
+		struct dc_surface_update *srf_updates, int surface_count,
+		struct dc_stream_state *stream,
+		struct dc_stream_update *stream_update,
+		enum surface_update_type update_type)
+{
+	struct dc_fast_update fast_update[MAX_SURFACES] = {0};
+
+	ASSERT(update_type < UPDATE_TYPE_FULL);
+	populate_fast_updates(fast_update, srf_updates, surface_count,
+			stream_update);
+	if (fast_update_only(dc, fast_update, srf_updates, surface_count,
+			stream_update, stream) &&
+			!dc->debug.enable_legacy_fast_update)
+		commit_planes_for_stream_fast(dc,
+				srf_updates,
+				surface_count,
+				stream,
+				stream_update,
+				update_type,
+				dc->current_state);
+	else
+		commit_planes_for_stream(
+				dc,
+				srf_updates,
+				surface_count,
+				stream,
+				stream_update,
+				update_type,
+				dc->current_state);
+}
+
+static void commit_planes_and_stream_update_with_new_context(struct dc *dc,
+		struct dc_surface_update *srf_updates, int surface_count,
+		struct dc_stream_state *stream,
+		struct dc_stream_update *stream_update,
+		enum surface_update_type update_type,
+		struct dc_state *new_context)
+{
+	ASSERT(update_type >= UPDATE_TYPE_FULL);
+	if (!dc->hwss.is_pipe_topology_transition_seamless(dc,
+			dc->current_state, new_context))
+		/*
+		 * It is required by the feature design that all pipe topologies
+		 * using extra free pipes for power saving purposes such as
+		 * dynamic ODM or SubVp shall only be enabled when it can be
+		 * transitioned seamlessly to AND from its minimal transition
+		 * state. A minimal transition state is defined as the same dc
+		 * state but with all power saving features disabled. So it uses
+		 * the minimum pipe topology. When we can't seamlessly
+		 * transition from state A to state B, we will insert the
+		 * minimal transition state A' or B' in between so seamless
+		 * transition between A and B can be made possible.
+		 */
+		commit_minimal_transition_state_in_dc_update(dc, new_context,
+				stream, srf_updates, surface_count);
+
+	commit_planes_for_stream(
+			dc,
+			srf_updates,
+			surface_count,
+			stream,
+			stream_update,
+			update_type,
+			new_context);
+}
+
+static bool update_planes_and_stream_v3(struct dc *dc,
+		struct dc_surface_update *srf_updates, int surface_count,
+		struct dc_stream_state *stream,
+		struct dc_stream_update *stream_update)
+{
+	struct dc_state *new_context;
+	enum surface_update_type update_type;
+
+	/*
+	 * When this function returns true and new_context is not equal to
+	 * current state, the function allocates and validates a new dc state
+	 * and assigns it to new_context. The function expects that the caller
+	 * is responsible to free this memory when new_context is no longer
+	 * used. We swap current with new context and free current instead. So
+	 * new_context's memory will live until the next full update after it is
+	 * replaced by a newer context. Refer to the use of
+	 * swap_and_free_current_context below.
+	 */
+	if (!update_planes_and_stream_state(dc, srf_updates, surface_count,
+				stream, stream_update, &update_type,
+				&new_context))
+		return false;
+
+	if (new_context == dc->current_state) {
+		commit_planes_and_stream_update_on_current_context(dc,
+				srf_updates, surface_count, stream,
+				stream_update, update_type);
+	} else {
+		commit_planes_and_stream_update_with_new_context(dc,
+				srf_updates, surface_count, stream,
+				stream_update, update_type, new_context);
+		swap_and_release_current_context(dc, new_context, stream);
+	}
+
+	return true;
+}
+
+static void clear_update_flags(struct dc_surface_update *srf_updates,
+	int surface_count, struct dc_stream_state *stream)
+{
+	int i;
+
+	if (stream)
+		stream->update_flags.raw = 0;
+
+	for (i = 0; i < surface_count; i++)
+		if (srf_updates[i].surface)
+			srf_updates[i].surface->update_flags.raw = 0;
+}
+
+bool dc_update_planes_and_stream(struct dc *dc,
+		struct dc_surface_update *srf_updates, int surface_count,
+		struct dc_stream_state *stream,
+		struct dc_stream_update *stream_update)
+{
+	bool ret = false;
+
+	dc_exit_ips_for_hw_access(dc);
+	/*
+	 * update planes and stream version 3 separates FULL and FAST updates
+	 * to their own sequences. It aims to clean up frequent checks for
+	 * update type resulting unnecessary branching in logic flow. It also
+	 * adds a new commit minimal transition sequence, which detects the need
+	 * for minimal transition based on the actual comparison of current and
+	 * new states instead of "predicting" it based on per feature software
+	 * policy.i.e could_mpcc_tree_change_for_active_pipes. The new commit
+	 * minimal transition sequence is made universal to any power saving
+	 * features that would use extra free pipes such as Dynamic ODM/MPC
+	 * Combine, MPO or SubVp. Therefore there is no longer a need to
+	 * specially handle compatibility problems with transitions among those
+	 * features as they are now transparent to the new sequence.
+	 */
+	if (dc->ctx->dce_version >= DCN_VERSION_4_01)
+		ret = update_planes_and_stream_v3(dc, srf_updates,
+				surface_count, stream, stream_update);
+	else
+		ret = update_planes_and_stream_v2(dc, srf_updates,
+			surface_count, stream, stream_update);
+	if (ret && (dc->ctx->dce_version >= DCN_VERSION_3_2 ||
+		dc->ctx->dce_version == DCN_VERSION_3_01))
+		clear_update_flags(srf_updates, surface_count, stream);
+
+	return ret;
+}
+
+void dc_commit_updates_for_stream(struct dc *dc,
+		struct dc_surface_update *srf_updates,
+		int surface_count,
+		struct dc_stream_state *stream,
+		struct dc_stream_update *stream_update,
+		struct dc_state *state)
+{
+	bool ret = false;
+
+	dc_exit_ips_for_hw_access(dc);
+	/* TODO: Since change commit sequence can have a huge impact,
+	 * we decided to only enable it for DCN3x. However, as soon as
+	 * we get more confident about this change we'll need to enable
+	 * the new sequence for all ASICs.
+	 */
+	if (dc->ctx->dce_version >= DCN_VERSION_4_01) {
+		ret = update_planes_and_stream_v3(dc, srf_updates, surface_count,
+				stream, stream_update);
+	} else {
+		ret = update_planes_and_stream_v2(dc, srf_updates, surface_count,
+				stream, stream_update);
+	}
+
+	if (ret && dc->ctx->dce_version >= DCN_VERSION_3_2)
+		clear_update_flags(srf_updates, surface_count, stream);
+}
+
+uint8_t dc_get_current_stream_count(struct dc *dc)
+{
+	return dc->current_state->stream_count;
+}
+
+struct dc_stream_state *dc_get_stream_at_index(struct dc *dc, uint8_t i)
+{
+	if (i < dc->current_state->stream_count)
+		return dc->current_state->streams[i];
+	return NULL;
+}
+
+enum dc_irq_source dc_interrupt_to_irq_source(
+		struct dc *dc,
+		uint32_t src_id,
+		uint32_t ext_id)
+{
+	return dal_irq_service_to_irq_source(dc->res_pool->irqs, src_id, ext_id);
+}
+
+/*
+ * dc_interrupt_set() - Enable/disable an AMD hw interrupt source
+ */
+bool dc_interrupt_set(struct dc *dc, enum dc_irq_source src, bool enable)
+{
+
+	if (dc == NULL)
+		return false;
+
+	return dal_irq_service_set(dc->res_pool->irqs, src, enable);
+}
+
+void dc_interrupt_ack(struct dc *dc, enum dc_irq_source src)
+{
+	dal_irq_service_ack(dc->res_pool->irqs, src);
+}
+
+void dc_power_down_on_boot(struct dc *dc)
+{
+	if (dc->ctx->dce_environment != DCE_ENV_VIRTUAL_HW &&
+	    dc->hwss.power_down_on_boot) {
+		if (dc->caps.ips_support)
+			dc_exit_ips_for_hw_access(dc);
+		dc->hwss.power_down_on_boot(dc);
+	}
+}
+
+void dc_set_power_state(struct dc *dc, enum dc_acpi_cm_power_state power_state)
+{
+	if (!dc->current_state)
+		return;
+
+	switch (power_state) {
+	case DC_ACPI_CM_POWER_STATE_D0:
+		dc_state_construct(dc, dc->current_state);
+
+		dc_exit_ips_for_hw_access(dc);
+
+		dc_z10_restore(dc);
+
+		dc_dmub_srv_notify_fw_dc_power_state(dc->ctx->dmub_srv, power_state);
+
+		dc->hwss.init_hw(dc);
+
+		if (dc->hwss.init_sys_ctx != NULL &&
+			dc->vm_pa_config.valid) {
+			dc->hwss.init_sys_ctx(dc->hwseq, dc, &dc->vm_pa_config);
+		}
+		break;
+	case DC_ACPI_CM_POWER_STATE_D3:
+		if (dc->caps.ips_support)
+			dc_dmub_srv_notify_fw_dc_power_state(dc->ctx->dmub_srv, DC_ACPI_CM_POWER_STATE_D3);
+
+		if (dc->caps.ips_v2_support) {
+			if (dc->clk_mgr->funcs->set_low_power_state)
+				dc->clk_mgr->funcs->set_low_power_state(dc->clk_mgr);
+		}
+		break;
+	default:
+		ASSERT(dc->current_state->stream_count == 0);
+		dc_dmub_srv_notify_fw_dc_power_state(dc->ctx->dmub_srv, power_state);
+
+		dc_state_destruct(dc->current_state);
+
+		break;
+	}
+}
+
+void dc_resume(struct dc *dc)
+{
+	uint32_t i;
+
+	for (i = 0; i < dc->link_count; i++)
+		dc->link_srv->resume(dc->links[i]);
+}
+
+bool dc_is_dmcu_initialized(struct dc *dc)
+{
+	struct dmcu *dmcu = dc->res_pool->dmcu;
+
+	if (dmcu)
+		return dmcu->funcs->is_dmcu_initialized(dmcu);
+	return false;
+}
+
+enum dc_status dc_set_clock(struct dc *dc, enum dc_clock_type clock_type, uint32_t clk_khz, uint32_t stepping)
+{
+	if (dc->hwss.set_clock)
+		return dc->hwss.set_clock(dc, clock_type, clk_khz, stepping);
+	return DC_ERROR_UNEXPECTED;
+}
+void dc_get_clock(struct dc *dc, enum dc_clock_type clock_type, struct dc_clock_config *clock_cfg)
+{
+	if (dc->hwss.get_clock)
+		dc->hwss.get_clock(dc, clock_type, clock_cfg);
+}
+
+/* enable/disable eDP PSR without specify stream for eDP */
+bool dc_set_psr_allow_active(struct dc *dc, bool enable)
+{
+	int i;
+	bool allow_active;
+
+	for (i = 0; i < dc->current_state->stream_count ; i++) {
+		struct dc_link *link;
+		struct dc_stream_state *stream = dc->current_state->streams[i];
+
+		link = stream->link;
+		if (!link)
+			continue;
+
+		if (link->psr_settings.psr_feature_enabled) {
+			if (enable && !link->psr_settings.psr_allow_active) {
+				allow_active = true;
+				if (!dc_link_set_psr_allow_active(link, &allow_active, false, false, NULL))
+					return false;
+			} else if (!enable && link->psr_settings.psr_allow_active) {
+				allow_active = false;
+				if (!dc_link_set_psr_allow_active(link, &allow_active, true, false, NULL))
+					return false;
+			}
+		}
+	}
+
+	return true;
+}
+
+/* enable/disable eDP Replay without specify stream for eDP */
+bool dc_set_replay_allow_active(struct dc *dc, bool active)
+{
+	int i;
+	bool allow_active;
+
+	for (i = 0; i < dc->current_state->stream_count; i++) {
+		struct dc_link *link;
+		struct dc_stream_state *stream = dc->current_state->streams[i];
+
+		link = stream->link;
+		if (!link)
+			continue;
+
+		if (link->replay_settings.replay_feature_enabled) {
+			if (active && !link->replay_settings.replay_allow_active) {
+				allow_active = true;
+				if (!dc_link_set_replay_allow_active(link, &allow_active,
+					false, false, NULL))
+					return false;
+			} else if (!active && link->replay_settings.replay_allow_active) {
+				allow_active = false;
+				if (!dc_link_set_replay_allow_active(link, &allow_active,
+					true, false, NULL))
+					return false;
+			}
+		}
+	}
+
+	return true;
+}
+
+/* set IPS disable state */
+bool dc_set_ips_disable(struct dc *dc, unsigned int disable_ips)
+{
+	dc_exit_ips_for_hw_access(dc);
+
+	dc->config.disable_ips = disable_ips;
+
+	return true;
+}
+
+void dc_allow_idle_optimizations_internal(struct dc *dc, bool allow, char const *caller_name)
+{
+	int idle_fclk_khz = 0, idle_dramclk_khz = 0, i = 0;
+	enum mall_stream_type subvp_pipe_type[MAX_PIPES] = {0};
+	struct pipe_ctx *pipe = NULL;
+	struct dc_state *context = dc->current_state;
+
+	if (dc->debug.disable_idle_power_optimizations) {
+		DC_LOG_DEBUG("%s: disabled\n", __func__);
+		return;
+	}
+
+	if (allow != dc->idle_optimizations_allowed)
+		DC_LOG_IPS("%s: allow_idle old=%d new=%d (caller=%s)\n", __func__,
+			   dc->idle_optimizations_allowed, allow, caller_name);
+
+	if (dc->caps.ips_support && (dc->config.disable_ips == DMUB_IPS_DISABLE_ALL))
+		return;
+
+	if (dc->clk_mgr != NULL && dc->clk_mgr->funcs->is_smu_present)
+		if (!dc->clk_mgr->funcs->is_smu_present(dc->clk_mgr))
+			return;
+
+	if (allow == dc->idle_optimizations_allowed)
+		return;
+
+	if (dc->hwss.apply_idle_power_optimizations && dc->clk_mgr != NULL &&
+	    dc->hwss.apply_idle_power_optimizations(dc, allow)) {
+		dc->idle_optimizations_allowed = allow;
+		DC_LOG_DEBUG("%s: %s\n", __func__, allow ? "enabled" : "disabled");
+	}
+
+	// log idle clocks and sub vp pipe types at idle optimization time
+	if (dc->clk_mgr != NULL && dc->clk_mgr->funcs->get_hard_min_fclk)
+		idle_fclk_khz = dc->clk_mgr->funcs->get_hard_min_fclk(dc->clk_mgr);
+
+	if (dc->clk_mgr != NULL && dc->clk_mgr->funcs->get_hard_min_memclk)
+		idle_dramclk_khz = dc->clk_mgr->funcs->get_hard_min_memclk(dc->clk_mgr);
+
+	if (dc->res_pool && context) {
+		for (i = 0; i < dc->res_pool->pipe_count; i++) {
+			pipe = &context->res_ctx.pipe_ctx[i];
+			subvp_pipe_type[i] = dc_state_get_pipe_subvp_type(context, pipe);
+		}
+	}
+	if (!dc->caps.is_apu)
+		DC_LOG_DC("%s: allow_idle=%d\n HardMinUClk_Khz=%d HardMinDramclk_Khz=%d\n Pipe_0=%d Pipe_1=%d Pipe_2=%d Pipe_3=%d Pipe_4=%d Pipe_5=%d (caller=%s)\n",
+			__func__, allow, idle_fclk_khz, idle_dramclk_khz, subvp_pipe_type[0], subvp_pipe_type[1], subvp_pipe_type[2],
+			subvp_pipe_type[3], subvp_pipe_type[4], subvp_pipe_type[5], caller_name);
+
+}
+
+void dc_exit_ips_for_hw_access_internal(struct dc *dc, const char *caller_name)
+{
+	if (dc->caps.ips_support)
+		dc_allow_idle_optimizations_internal(dc, false, caller_name);
+}
+
+bool dc_dmub_is_ips_idle_state(struct dc *dc)
+{
+	if (dc->debug.disable_idle_power_optimizations)
+		return false;
+
+	if (!dc->caps.ips_support || (dc->config.disable_ips == DMUB_IPS_DISABLE_ALL))
+		return false;
+
+	if (!dc->ctx->dmub_srv)
+		return false;
+
+	return dc->ctx->dmub_srv->idle_allowed;
+}
+
+/* set min and max memory clock to lowest and highest DPM level, respectively */
+void dc_unlock_memory_clock_frequency(struct dc *dc)
+{
+	if (dc->clk_mgr->funcs->set_hard_min_memclk)
+		dc->clk_mgr->funcs->set_hard_min_memclk(dc->clk_mgr, false);
+
+	if (dc->clk_mgr->funcs->set_hard_max_memclk)
+		dc->clk_mgr->funcs->set_hard_max_memclk(dc->clk_mgr);
+}
+
+/* set min memory clock to the min required for current mode, max to maxDPM */
+void dc_lock_memory_clock_frequency(struct dc *dc)
+{
+	if (dc->clk_mgr->funcs->get_memclk_states_from_smu)
+		dc->clk_mgr->funcs->get_memclk_states_from_smu(dc->clk_mgr);
+
+	if (dc->clk_mgr->funcs->set_hard_min_memclk)
+		dc->clk_mgr->funcs->set_hard_min_memclk(dc->clk_mgr, true);
+
+	if (dc->clk_mgr->funcs->set_hard_max_memclk)
+		dc->clk_mgr->funcs->set_hard_max_memclk(dc->clk_mgr);
+}
+
+static void blank_and_force_memclk(struct dc *dc, bool apply, unsigned int memclk_mhz)
+{
+	struct dc_state *context = dc->current_state;
+	struct hubp *hubp;
+	struct pipe_ctx *pipe;
+	int i;
+
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		pipe = &context->res_ctx.pipe_ctx[i];
+
+		if (pipe->stream != NULL) {
+			dc->hwss.disable_pixel_data(dc, pipe, true);
+
+			// wait for double buffer
+			pipe->stream_res.tg->funcs->wait_for_state(pipe->stream_res.tg, CRTC_STATE_VACTIVE);
+			pipe->stream_res.tg->funcs->wait_for_state(pipe->stream_res.tg, CRTC_STATE_VBLANK);
+			pipe->stream_res.tg->funcs->wait_for_state(pipe->stream_res.tg, CRTC_STATE_VACTIVE);
+
+			hubp = pipe->plane_res.hubp;
+			hubp->funcs->set_blank_regs(hubp, true);
+		}
+	}
+	if (dc->clk_mgr->funcs->set_max_memclk)
+		dc->clk_mgr->funcs->set_max_memclk(dc->clk_mgr, memclk_mhz);
+	if (dc->clk_mgr->funcs->set_min_memclk)
+		dc->clk_mgr->funcs->set_min_memclk(dc->clk_mgr, memclk_mhz);
+
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		pipe = &context->res_ctx.pipe_ctx[i];
+
+		if (pipe->stream != NULL) {
+			dc->hwss.disable_pixel_data(dc, pipe, false);
+
+			hubp = pipe->plane_res.hubp;
+			hubp->funcs->set_blank_regs(hubp, false);
+		}
+	}
+}
+
+
+/**
+ * dc_enable_dcmode_clk_limit() - lower clocks in dc (battery) mode
+ * @dc: pointer to dc of the dm calling this
+ * @enable: True = transition to DC mode, false = transition back to AC mode
+ *
+ * Some SoCs define additional clock limits when in DC mode, DM should
+ * invoke this function when the platform undergoes a power source transition
+ * so DC can apply/unapply the limit. This interface may be disruptive to
+ * the onscreen content.
+ *
+ * Context: Triggered by OS through DM interface, or manually by escape calls.
+ * Need to hold a dclock when doing so.
+ *
+ * Return: none (void function)
+ *
+ */
+void dc_enable_dcmode_clk_limit(struct dc *dc, bool enable)
+{
+	unsigned int softMax = 0, maxDPM = 0, funcMin = 0, i;
+	bool p_state_change_support;
+
+	if (!dc->config.dc_mode_clk_limit_support)
+		return;
+
+	softMax = dc->clk_mgr->bw_params->dc_mode_softmax_memclk;
+	for (i = 0; i < dc->clk_mgr->bw_params->clk_table.num_entries; i++) {
+		if (dc->clk_mgr->bw_params->clk_table.entries[i].memclk_mhz > maxDPM)
+			maxDPM = dc->clk_mgr->bw_params->clk_table.entries[i].memclk_mhz;
+	}
+	funcMin = (dc->clk_mgr->clks.dramclk_khz + 999) / 1000;
+	p_state_change_support = dc->clk_mgr->clks.p_state_change_support;
+
+	if (enable && !dc->clk_mgr->dc_mode_softmax_enabled) {
+		if (p_state_change_support) {
+			if (funcMin <= softMax && dc->clk_mgr->funcs->set_max_memclk)
+				dc->clk_mgr->funcs->set_max_memclk(dc->clk_mgr, softMax);
+			// else: No-Op
+		} else {
+			if (funcMin <= softMax)
+				blank_and_force_memclk(dc, true, softMax);
+			// else: No-Op
+		}
+	} else if (!enable && dc->clk_mgr->dc_mode_softmax_enabled) {
+		if (p_state_change_support) {
+			if (funcMin <= softMax && dc->clk_mgr->funcs->set_max_memclk)
+				dc->clk_mgr->funcs->set_max_memclk(dc->clk_mgr, maxDPM);
+			// else: No-Op
+		} else {
+			if (funcMin <= softMax)
+				blank_and_force_memclk(dc, true, maxDPM);
+			// else: No-Op
+		}
+	}
+	dc->clk_mgr->dc_mode_softmax_enabled = enable;
+}
+bool dc_is_plane_eligible_for_idle_optimizations(struct dc *dc,
+		unsigned int pitch,
+		unsigned int height,
+		enum surface_pixel_format format,
+		struct dc_cursor_attributes *cursor_attr)
+{
+	if (dc->hwss.does_plane_fit_in_mall && dc->hwss.does_plane_fit_in_mall(dc, pitch, height, format, cursor_attr))
+		return true;
+	return false;
+}
+
+/* cleanup on driver unload */
+void dc_hardware_release(struct dc *dc)
+{
+	dc_mclk_switch_using_fw_based_vblank_stretch_shut_down(dc);
+
+	if (dc->hwss.hardware_release)
+		dc->hwss.hardware_release(dc);
+}
+
+void dc_mclk_switch_using_fw_based_vblank_stretch_shut_down(struct dc *dc)
+{
+	if (dc->current_state)
+		dc->current_state->bw_ctx.bw.dcn.clk.fw_based_mclk_switching_shut_down = true;
+}
+
+/**
+ * dc_is_dmub_outbox_supported - Check if DMUB firmware support outbox notification
+ *
+ * @dc: [in] dc structure
+ *
+ * Checks whether DMUB FW supports outbox notifications, if supported DM
+ * should register outbox interrupt prior to actually enabling interrupts
+ * via dc_enable_dmub_outbox
+ *
+ * Return:
+ * True if DMUB FW supports outbox notifications, False otherwise
+ */
+bool dc_is_dmub_outbox_supported(struct dc *dc)
+{
+	if (!dc->caps.dmcub_support)
+		return false;
+
+	switch (dc->ctx->asic_id.chip_family) {
+
+	case FAMILY_YELLOW_CARP:
+		/* DCN31 B0 USB4 DPIA needs dmub notifications for interrupts */
+		if (dc->ctx->asic_id.hw_internal_rev == YELLOW_CARP_B0 &&
+		    !dc->debug.dpia_debug.bits.disable_dpia)
+			return true;
+	break;
+
+	case AMDGPU_FAMILY_GC_11_0_1:
+	case AMDGPU_FAMILY_GC_11_5_0:
+		if (!dc->debug.dpia_debug.bits.disable_dpia)
+			return true;
+	break;
+
+	default:
+		break;
+	}
+
+	/* dmub aux needs dmub notifications to be enabled */
+	return dc->debug.enable_dmub_aux_for_legacy_ddc;
+
+}
+
+/**
+ * dc_enable_dmub_notifications - Check if dmub fw supports outbox
+ *
+ * @dc: [in] dc structure
+ *
+ * Calls dc_is_dmub_outbox_supported to check if dmub fw supports outbox
+ * notifications. All DMs shall switch to dc_is_dmub_outbox_supported.  This
+ * API shall be removed after switching.
+ *
+ * Return:
+ * True if DMUB FW supports outbox notifications, False otherwise
+ */
+bool dc_enable_dmub_notifications(struct dc *dc)
+{
+	return dc_is_dmub_outbox_supported(dc);
+}
+
+/**
+ * dc_enable_dmub_outbox - Enables DMUB unsolicited notification
+ *
+ * @dc: [in] dc structure
+ *
+ * Enables DMUB unsolicited notifications to x86 via outbox.
+ */
+void dc_enable_dmub_outbox(struct dc *dc)
+{
+	struct dc_context *dc_ctx = dc->ctx;
+
+	dmub_enable_outbox_notification(dc_ctx->dmub_srv);
+	DC_LOG_DC("%s: dmub outbox notifications enabled\n", __func__);
+}
+
+/**
+ * dc_process_dmub_aux_transfer_async - Submits aux command to dmub via inbox message
+ *                                      Sets port index appropriately for legacy DDC
+ * @dc: dc structure
+ * @link_index: link index
+ * @payload: aux payload
+ *
+ * Returns: True if successful, False if failure
+ */
+bool dc_process_dmub_aux_transfer_async(struct dc *dc,
+				uint32_t link_index,
+				struct aux_payload *payload)
+{
+	uint8_t action;
+	union dmub_rb_cmd cmd = {0};
+
+	ASSERT(payload->length <= 16);
+
+	cmd.dp_aux_access.header.type = DMUB_CMD__DP_AUX_ACCESS;
+	cmd.dp_aux_access.header.payload_bytes = 0;
+	/* For dpia, ddc_pin is set to NULL */
+	if (!dc->links[link_index]->ddc->ddc_pin)
+		cmd.dp_aux_access.aux_control.type = AUX_CHANNEL_DPIA;
+	else
+		cmd.dp_aux_access.aux_control.type = AUX_CHANNEL_LEGACY_DDC;
+
+	cmd.dp_aux_access.aux_control.instance = dc->links[link_index]->ddc_hw_inst;
+	cmd.dp_aux_access.aux_control.sw_crc_enabled = 0;
+	cmd.dp_aux_access.aux_control.timeout = 0;
+	cmd.dp_aux_access.aux_control.dpaux.address = payload->address;
+	cmd.dp_aux_access.aux_control.dpaux.is_i2c_over_aux = payload->i2c_over_aux;
+	cmd.dp_aux_access.aux_control.dpaux.length = payload->length;
+
+	/* set aux action */
+	if (payload->i2c_over_aux) {
+		if (payload->write) {
+			if (payload->mot)
+				action = DP_AUX_REQ_ACTION_I2C_WRITE_MOT;
+			else
+				action = DP_AUX_REQ_ACTION_I2C_WRITE;
+		} else {
+			if (payload->mot)
+				action = DP_AUX_REQ_ACTION_I2C_READ_MOT;
+			else
+				action = DP_AUX_REQ_ACTION_I2C_READ;
+			}
+	} else {
+		if (payload->write)
+			action = DP_AUX_REQ_ACTION_DPCD_WRITE;
+		else
+			action = DP_AUX_REQ_ACTION_DPCD_READ;
+	}
+
+	cmd.dp_aux_access.aux_control.dpaux.action = action;
+
+	if (payload->length && payload->write) {
+		memcpy(cmd.dp_aux_access.aux_control.dpaux.data,
+			payload->data,
+			payload->length
+			);
+	}
+
+	dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+
+	return true;
+}
+
+uint8_t get_link_index_from_dpia_port_index(const struct dc *dc,
+					    uint8_t dpia_port_index)
+{
+	uint8_t index, link_index = 0xFF;
+
+	for (index = 0; index < dc->link_count; index++) {
+		/* ddc_hw_inst has dpia port index for dpia links
+		 * and ddc instance for legacy links
+		 */
+		if (!dc->links[index]->ddc->ddc_pin) {
+			if (dc->links[index]->ddc_hw_inst == dpia_port_index) {
+				link_index = index;
+				break;
+			}
+		}
+	}
+	ASSERT(link_index != 0xFF);
+	return link_index;
+}
+
+/**
+ * dc_process_dmub_set_config_async - Submits set_config command
+ *
+ * @dc: [in] dc structure
+ * @link_index: [in] link_index: link index
+ * @payload: [in] aux payload
+ * @notify: [out] set_config immediate reply
+ *
+ * Submits set_config command to dmub via inbox message.
+ *
+ * Return:
+ * True if successful, False if failure
+ */
+bool dc_process_dmub_set_config_async(struct dc *dc,
+				uint32_t link_index,
+				struct set_config_cmd_payload *payload,
+				struct dmub_notification *notify)
+{
+	union dmub_rb_cmd cmd = {0};
+	bool is_cmd_complete = true;
+
+	/* prepare SET_CONFIG command */
+	cmd.set_config_access.header.type = DMUB_CMD__DPIA;
+	cmd.set_config_access.header.sub_type = DMUB_CMD__DPIA_SET_CONFIG_ACCESS;
+
+	cmd.set_config_access.set_config_control.instance = dc->links[link_index]->ddc_hw_inst;
+	cmd.set_config_access.set_config_control.cmd_pkt.msg_type = payload->msg_type;
+	cmd.set_config_access.set_config_control.cmd_pkt.msg_data = payload->msg_data;
+
+	if (!dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY)) {
+		/* command is not processed by dmub */
+		notify->sc_status = SET_CONFIG_UNKNOWN_ERROR;
+		return is_cmd_complete;
+	}
+
+	/* command processed by dmub, if ret_status is 1, it is completed instantly */
+	if (cmd.set_config_access.header.ret_status == 1)
+		notify->sc_status = cmd.set_config_access.set_config_control.immed_status;
+	else
+		/* cmd pending, will receive notification via outbox */
+		is_cmd_complete = false;
+
+	return is_cmd_complete;
+}
+
+/**
+ * dc_process_dmub_set_mst_slots - Submits MST solt allocation
+ *
+ * @dc: [in] dc structure
+ * @link_index: [in] link index
+ * @mst_alloc_slots: [in] mst slots to be allotted
+ * @mst_slots_in_use: [out] mst slots in use returned in failure case
+ *
+ * Submits mst slot allocation command to dmub via inbox message
+ *
+ * Return:
+ * DC_OK if successful, DC_ERROR if failure
+ */
+enum dc_status dc_process_dmub_set_mst_slots(const struct dc *dc,
+				uint32_t link_index,
+				uint8_t mst_alloc_slots,
+				uint8_t *mst_slots_in_use)
+{
+	union dmub_rb_cmd cmd = {0};
+
+	/* prepare MST_ALLOC_SLOTS command */
+	cmd.set_mst_alloc_slots.header.type = DMUB_CMD__DPIA;
+	cmd.set_mst_alloc_slots.header.sub_type = DMUB_CMD__DPIA_MST_ALLOC_SLOTS;
+
+	cmd.set_mst_alloc_slots.mst_slots_control.instance = dc->links[link_index]->ddc_hw_inst;
+	cmd.set_mst_alloc_slots.mst_slots_control.mst_alloc_slots = mst_alloc_slots;
+
+	if (!dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY))
+		/* command is not processed by dmub */
+		return DC_ERROR_UNEXPECTED;
+
+	/* command processed by dmub, if ret_status is 1 */
+	if (cmd.set_config_access.header.ret_status != 1)
+		/* command processing error */
+		return DC_ERROR_UNEXPECTED;
+
+	/* command processed and we have a status of 2, mst not enabled in dpia */
+	if (cmd.set_mst_alloc_slots.mst_slots_control.immed_status == 2)
+		return DC_FAIL_UNSUPPORTED_1;
+
+	/* previously configured mst alloc and used slots did not match */
+	if (cmd.set_mst_alloc_slots.mst_slots_control.immed_status == 3) {
+		*mst_slots_in_use = cmd.set_mst_alloc_slots.mst_slots_control.mst_slots_in_use;
+		return DC_NOT_SUPPORTED;
+	}
+
+	return DC_OK;
+}
+
+/**
+ * dc_process_dmub_dpia_set_tps_notification - Submits tps notification
+ *
+ * @dc: [in] dc structure
+ * @link_index: [in] link index
+ * @tps: [in] request tps
+ *
+ * Submits set_tps_notification command to dmub via inbox message
+ */
+void dc_process_dmub_dpia_set_tps_notification(const struct dc *dc, uint32_t link_index, uint8_t tps)
+{
+	union dmub_rb_cmd cmd = {0};
+
+	cmd.set_tps_notification.header.type = DMUB_CMD__DPIA;
+	cmd.set_tps_notification.header.sub_type = DMUB_CMD__DPIA_SET_TPS_NOTIFICATION;
+	cmd.set_tps_notification.tps_notification.instance = dc->links[link_index]->ddc_hw_inst;
+	cmd.set_tps_notification.tps_notification.tps = tps;
+
+	dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+/**
+ * dc_process_dmub_dpia_hpd_int_enable - Submits DPIA DPD interruption
+ *
+ * @dc: [in] dc structure
+ * @hpd_int_enable: [in] 1 for hpd int enable, 0 to disable
+ *
+ * Submits dpia hpd int enable command to dmub via inbox message
+ */
+void dc_process_dmub_dpia_hpd_int_enable(const struct dc *dc,
+				uint32_t hpd_int_enable)
+{
+	union dmub_rb_cmd cmd = {0};
+
+	cmd.dpia_hpd_int_enable.header.type = DMUB_CMD__DPIA_HPD_INT_ENABLE;
+	cmd.dpia_hpd_int_enable.enable = hpd_int_enable;
+
+	dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+
+	DC_LOG_DEBUG("%s: hpd_int_enable(%d)\n", __func__, hpd_int_enable);
+}
+
+/**
+ * dc_print_dmub_diagnostic_data - Print DMUB diagnostic data for debugging
+ *
+ * @dc: [in] dc structure
+ *
+ *
+ */
+void dc_print_dmub_diagnostic_data(const struct dc *dc)
+{
+	dc_dmub_srv_log_diagnostic_data(dc->ctx->dmub_srv);
+}
+
+/**
+ * dc_disable_accelerated_mode - disable accelerated mode
+ * @dc: dc structure
+ */
+void dc_disable_accelerated_mode(struct dc *dc)
+{
+	bios_set_scratch_acc_mode_change(dc->ctx->dc_bios, 0);
+}
+
+
+/**
+ *  dc_notify_vsync_int_state - notifies vsync enable/disable state
+ *  @dc: dc structure
+ *  @stream: stream where vsync int state changed
+ *  @enable: whether vsync is enabled or disabled
+ *
+ *  Called when vsync is enabled/disabled Will notify DMUB to start/stop ABM
+ *  interrupts after steady state is reached.
+ */
+void dc_notify_vsync_int_state(struct dc *dc, struct dc_stream_state *stream, bool enable)
+{
+	int i;
+	int edp_num;
+	struct pipe_ctx *pipe = NULL;
+	struct dc_link *link = stream->sink->link;
+	struct dc_link *edp_links[MAX_NUM_EDP];
+
+
+	if (link->psr_settings.psr_feature_enabled)
+		return;
+
+	if (link->replay_settings.replay_feature_enabled)
+		return;
+
+	/*find primary pipe associated with stream*/
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+
+		if (pipe->stream == stream && pipe->stream_res.tg)
+			break;
+	}
+
+	if (i == MAX_PIPES) {
+		ASSERT(0);
+		return;
+	}
+
+	dc_get_edp_links(dc, edp_links, &edp_num);
+
+	/* Determine panel inst */
+	for (i = 0; i < edp_num; i++) {
+		if (edp_links[i] == link)
+			break;
+	}
+
+	if (i == edp_num) {
+		return;
+	}
+
+	if (pipe->stream_res.abm && pipe->stream_res.abm->funcs->set_abm_pause)
+		pipe->stream_res.abm->funcs->set_abm_pause(pipe->stream_res.abm, !enable, i, pipe->stream_res.tg->inst);
+}
+
+/*****************************************************************************
+ *  dc_abm_save_restore() - Interface to DC for save+pause and restore+un-pause
+ *                          ABM
+ *  @dc: dc structure
+ *	@stream: stream where vsync int state changed
+ *  @pData: abm hw states
+ *
+ ****************************************************************************/
+bool dc_abm_save_restore(
+		struct dc *dc,
+		struct dc_stream_state *stream,
+		struct abm_save_restore *pData)
+{
+	int i;
+	int edp_num;
+	struct pipe_ctx *pipe = NULL;
+	struct dc_link *link = stream->sink->link;
+	struct dc_link *edp_links[MAX_NUM_EDP];
+
+	if (link->replay_settings.replay_feature_enabled)
+		return false;
+
+	/*find primary pipe associated with stream*/
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+
+		if (pipe->stream == stream && pipe->stream_res.tg)
+			break;
+	}
+
+	if (i == MAX_PIPES) {
+		ASSERT(0);
+		return false;
+	}
+
+	dc_get_edp_links(dc, edp_links, &edp_num);
+
+	/* Determine panel inst */
+	for (i = 0; i < edp_num; i++)
+		if (edp_links[i] == link)
+			break;
+
+	if (i == edp_num)
+		return false;
+
+	if (pipe->stream_res.abm &&
+		pipe->stream_res.abm->funcs->save_restore)
+		return pipe->stream_res.abm->funcs->save_restore(
+				pipe->stream_res.abm,
+				i,
+				pData);
+	return false;
+}
+
+void dc_query_current_properties(struct dc *dc, struct dc_current_properties *properties)
+{
+	unsigned int i;
+	unsigned int max_cursor_size = dc->caps.max_cursor_size;
+	unsigned int stream_cursor_size;
+
+	if (dc->debug.allow_sw_cursor_fallback && dc->res_pool->funcs->get_max_hw_cursor_size) {
+		for (i = 0; i < dc->current_state->stream_count; i++) {
+			stream_cursor_size = dc->res_pool->funcs->get_max_hw_cursor_size(dc,
+					dc->current_state,
+					dc->current_state->streams[i]);
+
+			if (stream_cursor_size < max_cursor_size) {
+				max_cursor_size = stream_cursor_size;
+			}
+		}
+	}
+
+	properties->cursor_size_limit = max_cursor_size;
+}
+
+/**
+ * dc_set_edp_power() - DM controls eDP power to be ON/OFF
+ *
+ * Called when DM wants to power on/off eDP.
+ *     Only work on links with flag skip_implict_edp_power_control is set.
+ *
+ * @dc: Current DC state
+ * @edp_link: a link with eDP connector signal type
+ * @powerOn: power on/off eDP
+ *
+ * Return: void
+ */
+void dc_set_edp_power(const struct dc *dc, struct dc_link *edp_link,
+				 bool powerOn)
+{
+	if (edp_link->connector_signal != SIGNAL_TYPE_EDP)
+		return;
+
+	if (edp_link->skip_implict_edp_power_control == false)
+		return;
+
+	edp_link->dc->link_srv->edp_set_panel_power(edp_link, powerOn);
+}
+
+/**
+ * dc_get_power_profile_for_dc_state() - extracts power profile from dc state
+ *
+ * Called when DM wants to make power policy decisions based on dc_state
+ *
+ * @context: Pointer to the dc_state from which the power profile is extracted.
+ *
+ * Return: The power profile structure containing the power level information.
+ */
+struct dc_power_profile dc_get_power_profile_for_dc_state(const struct dc_state *context)
+{
+	struct dc_power_profile profile = { 0 };
+
+	profile.power_level = !context->bw_ctx.bw.dcn.clk.p_state_change_support;
+	if (!context->clk_mgr || !context->clk_mgr->ctx || !context->clk_mgr->ctx->dc)
+		return profile;
+	struct dc *dc = context->clk_mgr->ctx->dc;
+
+	if (dc->res_pool->funcs->get_power_profile)
+		profile.power_level = dc->res_pool->funcs->get_power_profile(context);
+	return profile;
+}
+
+/**
+ * dc_get_det_buffer_size_from_state() - extracts detile buffer size from dc state
+ *
+ * This function is called to log the detile buffer size from the dc_state.
+ *
+ * @context: a pointer to the dc_state from which the detile buffer size is extracted.
+ *
+ * Return: the size of the detile buffer, or 0 if not available.
+ */
+unsigned int dc_get_det_buffer_size_from_state(const struct dc_state *context)
+{
+	struct dc *dc = context->clk_mgr->ctx->dc;
+
+	if (dc->res_pool->funcs->get_det_buffer_size)
+		return dc->res_pool->funcs->get_det_buffer_size(context);
+	else
+		return 0;
+}
+
+/**
+ * dc_get_host_router_index: Get index of host router from a dpia link
+ *
+ * This function return a host router index of the target link. If the target link is dpia link.
+ *
+ * @link: Pointer to the target link (input)
+ * @host_router_index: Pointer to store the host router index of the target link (output).
+ *
+ * Return: true if the host router index is found and valid.
+ *
+ */
+bool dc_get_host_router_index(const struct dc_link *link, unsigned int *host_router_index)
+{
+	struct dc *dc;
+
+	if (!link || !host_router_index || link->ep_type != DISPLAY_ENDPOINT_USB4_DPIA)
+		return false;
+
+	dc = link->ctx->dc;
+
+	if (link->link_index < dc->lowest_dpia_link_index)
+		return false;
+
+	*host_router_index = (link->link_index - dc->lowest_dpia_link_index) / dc->caps.num_of_dpias_per_host_router;
+	if (*host_router_index < dc->caps.num_of_host_routers)
+		return true;
+	else
+		return false;
+}
+
+bool dc_is_cursor_limit_pending(struct dc *dc)
+{
+	uint32_t i;
+
+	for (i = 0; i < dc->current_state->stream_count; i++) {
+		if (dc_stream_is_cursor_limit_pending(dc, dc->current_state->streams[i]))
+			return true;
+	}
+
+	return false;
+}
+
+bool dc_can_clear_cursor_limit(struct dc *dc)
+{
+	uint32_t i;
+
+	for (i = 0; i < dc->current_state->stream_count; i++) {
+		if (dc_state_can_clear_stream_cursor_subvp_limit(dc->current_state->streams[i], dc->current_state))
+			return true;
+	}
+
+	return false;
+}
+
+void dc_get_underflow_debug_data_for_otg(struct dc *dc, int primary_otg_inst,
+				struct dc_underflow_debug_data *out_data)
+{
+	struct timing_generator *tg = NULL;
+
+	for (int i = 0; i < MAX_PIPES; i++) {
+		if (dc->res_pool->timing_generators[i] &&
+			dc->res_pool->timing_generators[i]->inst == primary_otg_inst) {
+				tg = dc->res_pool->timing_generators[i];
+				break;
+		}
+	}
+
+	dc_exit_ips_for_hw_access(dc);
+	if (dc->hwss.get_underflow_debug_data)
+		dc->hwss.get_underflow_debug_data(dc, tg, out_data);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_debug.c b/drivers/gpu/drm/amd/display/dc/core/dc_debug.c
new file mode 100644
index 000000000000..bbce751b485f
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_debug.c
@@ -0,0 +1,318 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+/*
+ * dc_debug.c
+ *
+ *  Created on: Nov 3, 2016
+ *      Author: yonsun
+ */
+
+#include "dm_services.h"
+
+#include "dc.h"
+
+#include "core_status.h"
+#include "core_types.h"
+
+#include "resource.h"
+
+#define DC_LOGGER \
+	dc->ctx->logger
+#define DC_LOGGER_INIT(logger)
+
+
+#define SURFACE_TRACE(...) do {\
+		if (dc->debug.surface_trace) \
+			DC_LOG_IF_TRACE(__VA_ARGS__); \
+} while (0)
+
+#define CLOCK_TRACE(...) do {\
+	if (dc->debug.clock_trace) \
+		DC_LOG_BANDWIDTH_CALCS(__VA_ARGS__); \
+} while (0)
+
+void update_surface_trace(
+		struct dc *dc,
+		const struct dc_surface_update *updates,
+		int surface_count)
+{
+	int i;
+	DC_LOGGER_INIT(dc->ctx->logger);
+
+	for (i = 0; i < surface_count; i++) {
+		const struct dc_surface_update *update = &updates[i];
+
+		SURFACE_TRACE("Update %d\n", i);
+		if (update->flip_addr) {
+			SURFACE_TRACE("flip_addr->address.type = %d;\n"
+					"flip_addr->address.grph.addr.quad_part = 0x%llX;\n"
+					"flip_addr->address.grph.meta_addr.quad_part = 0x%llX;\n"
+					"flip_addr->flip_immediate = %d;\n",
+					update->flip_addr->address.type,
+					update->flip_addr->address.grph.addr.quad_part,
+					update->flip_addr->address.grph.meta_addr.quad_part,
+					update->flip_addr->flip_immediate);
+		}
+
+		if (update->plane_info) {
+			SURFACE_TRACE(
+					"plane_info->color_space = %d;\n"
+					"plane_info->format = %d;\n"
+					"plane_info->plane_size.surface_pitch = %d;\n"
+					"plane_info->plane_size.surface_size.height = %d;\n"
+					"plane_info->plane_size.surface_size.width = %d;\n"
+					"plane_info->plane_size.surface_size.x = %d;\n"
+					"plane_info->plane_size.surface_size.y = %d;\n"
+					"plane_info->rotation = %d;\n"
+					"plane_info->stereo_format = %d;\n",
+					update->plane_info->color_space,
+					update->plane_info->format,
+					update->plane_info->plane_size.surface_pitch,
+					update->plane_info->plane_size.surface_size.height,
+					update->plane_info->plane_size.surface_size.width,
+					update->plane_info->plane_size.surface_size.x,
+					update->plane_info->plane_size.surface_size.y,
+					update->plane_info->rotation,
+					update->plane_info->stereo_format);
+
+			SURFACE_TRACE(
+					"plane_info->tiling_info.gfx8.num_banks = %d;\n"
+					"plane_info->tiling_info.gfx8.bank_width = %d;\n"
+					"plane_info->tiling_info.gfx8.bank_width_c = %d;\n"
+					"plane_info->tiling_info.gfx8.bank_height = %d;\n"
+					"plane_info->tiling_info.gfx8.bank_height_c = %d;\n"
+					"plane_info->tiling_info.gfx8.tile_aspect = %d;\n"
+					"plane_info->tiling_info.gfx8.tile_aspect_c = %d;\n"
+					"plane_info->tiling_info.gfx8.tile_split = %d;\n"
+					"plane_info->tiling_info.gfx8.tile_split_c = %d;\n"
+					"plane_info->tiling_info.gfx8.tile_mode = %d;\n"
+					"plane_info->tiling_info.gfx8.tile_mode_c = %d;\n",
+					update->plane_info->tiling_info.gfx8.num_banks,
+					update->plane_info->tiling_info.gfx8.bank_width,
+					update->plane_info->tiling_info.gfx8.bank_width_c,
+					update->plane_info->tiling_info.gfx8.bank_height,
+					update->plane_info->tiling_info.gfx8.bank_height_c,
+					update->plane_info->tiling_info.gfx8.tile_aspect,
+					update->plane_info->tiling_info.gfx8.tile_aspect_c,
+					update->plane_info->tiling_info.gfx8.tile_split,
+					update->plane_info->tiling_info.gfx8.tile_split_c,
+					update->plane_info->tiling_info.gfx8.tile_mode,
+					update->plane_info->tiling_info.gfx8.tile_mode_c);
+
+			SURFACE_TRACE(
+					"plane_info->tiling_info.gfx8.pipe_config = %d;\n"
+					"plane_info->tiling_info.gfx8.array_mode = %d;\n"
+					"plane_info->visible = %d;\n"
+					"plane_info->per_pixel_alpha = %d;\n",
+					update->plane_info->tiling_info.gfx8.pipe_config,
+					update->plane_info->tiling_info.gfx8.array_mode,
+					update->plane_info->visible,
+					update->plane_info->per_pixel_alpha);
+
+			SURFACE_TRACE("surface->tiling_info.gfx9.swizzle = %d;\n",
+					update->plane_info->tiling_info.gfx9.swizzle);
+		}
+
+		if (update->scaling_info) {
+			SURFACE_TRACE(
+					"scaling_info->src_rect.x = %d;\n"
+					"scaling_info->src_rect.y = %d;\n"
+					"scaling_info->src_rect.width = %d;\n"
+					"scaling_info->src_rect.height = %d;\n"
+					"scaling_info->dst_rect.x = %d;\n"
+					"scaling_info->dst_rect.y = %d;\n"
+					"scaling_info->dst_rect.width = %d;\n"
+					"scaling_info->dst_rect.height = %d;\n"
+					"scaling_info->clip_rect.x = %d;\n"
+					"scaling_info->clip_rect.y = %d;\n"
+					"scaling_info->clip_rect.width = %d;\n"
+					"scaling_info->clip_rect.height = %d;\n"
+					"scaling_info->scaling_quality.h_taps = %d;\n"
+					"scaling_info->scaling_quality.v_taps = %d;\n"
+					"scaling_info->scaling_quality.h_taps_c = %d;\n"
+					"scaling_info->scaling_quality.v_taps_c = %d;\n",
+					update->scaling_info->src_rect.x,
+					update->scaling_info->src_rect.y,
+					update->scaling_info->src_rect.width,
+					update->scaling_info->src_rect.height,
+					update->scaling_info->dst_rect.x,
+					update->scaling_info->dst_rect.y,
+					update->scaling_info->dst_rect.width,
+					update->scaling_info->dst_rect.height,
+					update->scaling_info->clip_rect.x,
+					update->scaling_info->clip_rect.y,
+					update->scaling_info->clip_rect.width,
+					update->scaling_info->clip_rect.height,
+					update->scaling_info->scaling_quality.h_taps,
+					update->scaling_info->scaling_quality.v_taps,
+					update->scaling_info->scaling_quality.h_taps_c,
+					update->scaling_info->scaling_quality.v_taps_c);
+		}
+		SURFACE_TRACE("\n");
+	}
+	SURFACE_TRACE("\n");
+}
+
+void post_surface_trace(struct dc *dc)
+{
+	DC_LOGGER_INIT(dc->ctx->logger);
+
+	SURFACE_TRACE("post surface process.\n");
+
+}
+
+void context_clock_trace(
+		struct dc *dc,
+		struct dc_state *context)
+{
+	DC_LOGGER_INIT(dc->ctx->logger);
+	CLOCK_TRACE("Current: dispclk_khz:%d  max_dppclk_khz:%d  dcfclk_khz:%d\n"
+			"dcfclk_deep_sleep_khz:%d  fclk_khz:%d  socclk_khz:%d\n",
+			context->bw_ctx.bw.dcn.clk.dispclk_khz,
+			context->bw_ctx.bw.dcn.clk.dppclk_khz,
+			context->bw_ctx.bw.dcn.clk.dcfclk_khz,
+			context->bw_ctx.bw.dcn.clk.dcfclk_deep_sleep_khz,
+			context->bw_ctx.bw.dcn.clk.fclk_khz,
+			context->bw_ctx.bw.dcn.clk.socclk_khz);
+	CLOCK_TRACE("Calculated: dispclk_khz:%d  max_dppclk_khz:%d  dcfclk_khz:%d\n"
+			"dcfclk_deep_sleep_khz:%d  fclk_khz:%d  socclk_khz:%d\n",
+			context->bw_ctx.bw.dcn.clk.dispclk_khz,
+			context->bw_ctx.bw.dcn.clk.dppclk_khz,
+			context->bw_ctx.bw.dcn.clk.dcfclk_khz,
+			context->bw_ctx.bw.dcn.clk.dcfclk_deep_sleep_khz,
+			context->bw_ctx.bw.dcn.clk.fclk_khz,
+			context->bw_ctx.bw.dcn.clk.socclk_khz);
+}
+
+/**
+ * dc_status_to_str - convert dc_status to a human readable string
+ * @status: dc_status to be converted
+ *
+ * Return:
+ * A string describing the DC status.
+ */
+char *dc_status_to_str(enum dc_status status)
+{
+	switch (status) {
+	case DC_OK:
+		return "DC OK";
+	case DC_NO_CONTROLLER_RESOURCE:
+		return "No controller resource";
+	case DC_NO_STREAM_ENC_RESOURCE:
+		return "No stream encoder";
+	case DC_NO_CLOCK_SOURCE_RESOURCE:
+		return "No clock source";
+	case DC_FAIL_CONTROLLER_VALIDATE:
+		return "Controller validation failure";
+	case DC_FAIL_ENC_VALIDATE:
+		return "Encoder validation failure";
+	case DC_FAIL_ATTACH_SURFACES:
+		return "Surfaces attachment failure";
+	case DC_FAIL_DETACH_SURFACES:
+		return "Surfaces detachment failure";
+	case DC_FAIL_SURFACE_VALIDATE:
+		return "Surface validation failure";
+	case DC_NO_DP_LINK_BANDWIDTH:
+		return "No DP link bandwidth";
+	case DC_EXCEED_DONGLE_CAP:
+		return "Exceed dongle capability";
+	case DC_SURFACE_PIXEL_FORMAT_UNSUPPORTED:
+		return "Unsupported pixel format";
+	case DC_FAIL_BANDWIDTH_VALIDATE:
+		return "Bandwidth validation failure (BW and Watermark)";
+	case DC_FAIL_SCALING:
+		return "Scaling failure";
+	case DC_FAIL_DP_LINK_TRAINING:
+		return "DP link training failure";
+	case DC_FAIL_DSC_VALIDATE:
+		return "DSC validation failure";
+	case DC_NO_DSC_RESOURCE:
+		return "No DSC resource";
+	case DC_FAIL_UNSUPPORTED_1:
+		return "Unsupported";
+	case DC_FAIL_CLK_EXCEED_MAX:
+		return "Clk exceed max failure";
+	case DC_FAIL_CLK_BELOW_MIN:
+		return "Fail clk below minimum";
+	case DC_FAIL_CLK_BELOW_CFG_REQUIRED:
+		return "Fail clk below required CFG (hard_min in PPLIB)";
+	case DC_NOT_SUPPORTED:
+		return "The operation is not supported.";
+	case DC_UNSUPPORTED_VALUE:
+		return "The value specified is not supported.";
+	case DC_NO_LINK_ENC_RESOURCE:
+		return "No link encoder resource";
+	case DC_FAIL_DP_PAYLOAD_ALLOCATION:
+		return "Fail dp payload allocation";
+	case DC_FAIL_DP_LINK_BANDWIDTH:
+		return "Insufficient DP link bandwidth";
+	case DC_FAIL_HW_CURSOR_SUPPORT:
+		return "HW Cursor not supported";
+	case DC_FAIL_DP_TUNNEL_BW_VALIDATE:
+		return "Fail DP Tunnel BW validation";
+	case DC_ERROR_UNEXPECTED:
+		return "Unexpected error";
+	}
+
+	return "Unexpected status error";
+}
+
+char *dc_pixel_encoding_to_str(enum dc_pixel_encoding pixel_encoding)
+{
+	switch (pixel_encoding) {
+	case PIXEL_ENCODING_RGB:
+		return "RGB";
+	case PIXEL_ENCODING_YCBCR422:
+		return "YUV422";
+	case PIXEL_ENCODING_YCBCR444:
+		return "YUV444";
+	case PIXEL_ENCODING_YCBCR420:
+		return "YUV420";
+	default:
+		return "Unknown";
+	}
+}
+
+char *dc_color_depth_to_str(enum dc_color_depth color_depth)
+{
+	switch (color_depth) {
+	case COLOR_DEPTH_666:
+		return "6-bpc";
+	case COLOR_DEPTH_888:
+		return "8-bpc";
+	case COLOR_DEPTH_101010:
+		return "10-bpc";
+	case COLOR_DEPTH_121212:
+		return "12-bpc";
+	case COLOR_DEPTH_141414:
+		return "14-bpc";
+	case COLOR_DEPTH_161616:
+		return "16-bpc";
+	case COLOR_DEPTH_999:
+		return "9-bpc";
+	case COLOR_DEPTH_111111:
+		return "11-bpc";
+	default:
+		return "Unknown";
+	}
+}
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c
new file mode 100644
index 000000000000..d82b1cb467f4
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c
@@ -0,0 +1,1266 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+#include "core_types.h"
+#include "timing_generator.h"
+#include "hw_sequencer.h"
+#include "hw_sequencer_private.h"
+#include "basics/dc_common.h"
+#include "resource.h"
+#include "dc_dmub_srv.h"
+#include "dc_state_priv.h"
+
+#define NUM_ELEMENTS(a) (sizeof(a) / sizeof((a)[0]))
+#define MAX_NUM_MCACHE 8
+
+/* used as index in array of black_color_format */
+enum black_color_format {
+	BLACK_COLOR_FORMAT_RGB_FULLRANGE = 0,
+	BLACK_COLOR_FORMAT_RGB_LIMITED,
+	BLACK_COLOR_FORMAT_YUV_TV,
+	BLACK_COLOR_FORMAT_YUV_CV,
+	BLACK_COLOR_FORMAT_YUV_SUPER_AA,
+	BLACK_COLOR_FORMAT_DEBUG,
+};
+
+enum dc_color_space_type {
+	COLOR_SPACE_RGB_TYPE,
+	COLOR_SPACE_RGB_LIMITED_TYPE,
+	COLOR_SPACE_YCBCR601_TYPE,
+	COLOR_SPACE_YCBCR709_TYPE,
+	COLOR_SPACE_YCBCR2020_TYPE,
+	COLOR_SPACE_YCBCR601_LIMITED_TYPE,
+	COLOR_SPACE_YCBCR709_LIMITED_TYPE,
+	COLOR_SPACE_YCBCR709_BLACK_TYPE,
+};
+
+static const struct tg_color black_color_format[] = {
+	/* BlackColorFormat_RGB_FullRange */
+	{0, 0, 0},
+	/* BlackColorFormat_RGB_Limited */
+	{0x40, 0x40, 0x40},
+	/* BlackColorFormat_YUV_TV */
+	{0x200, 0x40, 0x200},
+	/* BlackColorFormat_YUV_CV */
+	{0x1f4, 0x40, 0x1f4},
+	/* BlackColorFormat_YUV_SuperAA */
+	{0x1a2, 0x20, 0x1a2},
+	/* visual confirm debug */
+	{0xff, 0xff, 0},
+};
+
+struct out_csc_color_matrix_type {
+	enum dc_color_space_type color_space_type;
+	uint16_t regval[12];
+};
+
+static const struct out_csc_color_matrix_type output_csc_matrix[] = {
+	{ COLOR_SPACE_RGB_TYPE,
+		{ 0x2000, 0,      0,      0,
+		  0,      0x2000, 0,      0,
+		  0,      0,      0x2000, 0} },
+	{ COLOR_SPACE_RGB_LIMITED_TYPE,
+		{ 0x1B67, 0,      0,      0x201,
+		  0,      0x1B67, 0,      0x201,
+		  0,      0,      0x1B67, 0x201} },
+	{ COLOR_SPACE_YCBCR601_TYPE,
+		{ 0xE04,  0xF444, 0xFDB9, 0x1004,
+		  0x831,  0x1016, 0x320,  0x201,
+		  0xFB45, 0xF6B7, 0xE04,  0x1004} },
+	{ COLOR_SPACE_YCBCR709_TYPE,
+		{ 0xE04,  0xF345, 0xFEB7, 0x1004,
+		  0x5D3,  0x1399, 0x1FA,  0x201,
+		  0xFCCA, 0xF533, 0xE04,  0x1004} },
+	/* TODO: correct values below */
+	{ COLOR_SPACE_YCBCR601_LIMITED_TYPE,
+		{ 0xE00,  0xF447, 0xFDB9, 0x1000,
+		  0x991,  0x12C9, 0x3A6,  0x200,
+		  0xFB47, 0xF6B9, 0xE00,  0x1000} },
+	{ COLOR_SPACE_YCBCR709_LIMITED_TYPE,
+		{ 0xE00, 0xF349, 0xFEB7, 0x1000,
+		  0x6CE, 0x16E3, 0x24F,  0x200,
+		  0xFCCB, 0xF535, 0xE00, 0x1000} },
+	{ COLOR_SPACE_YCBCR2020_TYPE,
+		{ 0x1000, 0xF149, 0xFEB7, 0x1004,
+		  0x0868, 0x15B2, 0x01E6, 0x201,
+		  0xFB88, 0xF478, 0x1000, 0x1004} },
+	{ COLOR_SPACE_YCBCR709_BLACK_TYPE,
+		{ 0x0000, 0x0000, 0x0000, 0x1000,
+		  0x0000, 0x0000, 0x0000, 0x0200,
+		  0x0000, 0x0000, 0x0000, 0x1000} },
+};
+
+static bool is_rgb_type(
+		enum dc_color_space color_space)
+{
+	bool ret = false;
+
+	if (color_space == COLOR_SPACE_SRGB			||
+		color_space == COLOR_SPACE_XR_RGB		||
+		color_space == COLOR_SPACE_MSREF_SCRGB		||
+		color_space == COLOR_SPACE_2020_RGB_FULLRANGE	||
+		color_space == COLOR_SPACE_ADOBERGB		||
+		color_space == COLOR_SPACE_DCIP3	||
+		color_space == COLOR_SPACE_DOLBYVISION)
+		ret = true;
+	return ret;
+}
+
+static bool is_rgb_limited_type(
+		enum dc_color_space color_space)
+{
+	bool ret = false;
+
+	if (color_space == COLOR_SPACE_SRGB_LIMITED		||
+		color_space == COLOR_SPACE_2020_RGB_LIMITEDRANGE)
+		ret = true;
+	return ret;
+}
+
+static bool is_ycbcr601_type(
+		enum dc_color_space color_space)
+{
+	bool ret = false;
+
+	if (color_space == COLOR_SPACE_YCBCR601	||
+		color_space == COLOR_SPACE_XV_YCC_601)
+		ret = true;
+	return ret;
+}
+
+static bool is_ycbcr601_limited_type(
+		enum dc_color_space color_space)
+{
+	bool ret = false;
+
+	if (color_space == COLOR_SPACE_YCBCR601_LIMITED)
+		ret = true;
+	return ret;
+}
+
+static bool is_ycbcr709_type(
+		enum dc_color_space color_space)
+{
+	bool ret = false;
+
+	if (color_space == COLOR_SPACE_YCBCR709	||
+		color_space == COLOR_SPACE_XV_YCC_709)
+		ret = true;
+	return ret;
+}
+
+static bool is_ycbcr2020_type(
+	enum dc_color_space color_space)
+{
+	bool ret = false;
+
+	if (color_space == COLOR_SPACE_2020_YCBCR_LIMITED || color_space == COLOR_SPACE_2020_YCBCR_FULL)
+		ret = true;
+	return ret;
+}
+
+static bool is_ycbcr709_limited_type(
+		enum dc_color_space color_space)
+{
+	bool ret = false;
+
+	if (color_space == COLOR_SPACE_YCBCR709_LIMITED)
+		ret = true;
+	return ret;
+}
+
+static enum dc_color_space_type get_color_space_type(enum dc_color_space color_space)
+{
+	enum dc_color_space_type type = COLOR_SPACE_RGB_TYPE;
+
+	if (is_rgb_type(color_space))
+		type = COLOR_SPACE_RGB_TYPE;
+	else if (is_rgb_limited_type(color_space))
+		type = COLOR_SPACE_RGB_LIMITED_TYPE;
+	else if (is_ycbcr601_type(color_space))
+		type = COLOR_SPACE_YCBCR601_TYPE;
+	else if (is_ycbcr709_type(color_space))
+		type = COLOR_SPACE_YCBCR709_TYPE;
+	else if (is_ycbcr601_limited_type(color_space))
+		type = COLOR_SPACE_YCBCR601_LIMITED_TYPE;
+	else if (is_ycbcr709_limited_type(color_space))
+		type = COLOR_SPACE_YCBCR709_LIMITED_TYPE;
+	else if (is_ycbcr2020_type(color_space))
+		type = COLOR_SPACE_YCBCR2020_TYPE;
+	else if (color_space == COLOR_SPACE_YCBCR709)
+		type = COLOR_SPACE_YCBCR709_BLACK_TYPE;
+	else if (color_space == COLOR_SPACE_YCBCR709_BLACK)
+		type = COLOR_SPACE_YCBCR709_BLACK_TYPE;
+	return type;
+}
+
+const uint16_t *find_color_matrix(enum dc_color_space color_space,
+							uint32_t *array_size)
+{
+	int i;
+	enum dc_color_space_type type;
+	const uint16_t *val = NULL;
+	int arr_size = NUM_ELEMENTS(output_csc_matrix);
+
+	type = get_color_space_type(color_space);
+	for (i = 0; i < arr_size; i++)
+		if (output_csc_matrix[i].color_space_type == type) {
+			val = output_csc_matrix[i].regval;
+			*array_size = 12;
+			break;
+		}
+
+	return val;
+}
+
+
+void color_space_to_black_color(
+	const struct dc *dc,
+	enum dc_color_space colorspace,
+	struct tg_color *black_color)
+{
+	switch (colorspace) {
+	case COLOR_SPACE_YCBCR601:
+	case COLOR_SPACE_YCBCR709:
+	case COLOR_SPACE_YCBCR709_BLACK:
+	case COLOR_SPACE_YCBCR601_LIMITED:
+	case COLOR_SPACE_YCBCR709_LIMITED:
+	case COLOR_SPACE_2020_YCBCR_LIMITED:
+	case COLOR_SPACE_2020_YCBCR_FULL:
+		*black_color = black_color_format[BLACK_COLOR_FORMAT_YUV_CV];
+		break;
+
+	case COLOR_SPACE_SRGB_LIMITED:
+		*black_color =
+			black_color_format[BLACK_COLOR_FORMAT_RGB_LIMITED];
+		break;
+
+	/**
+	 * Remove default and add case for all color space
+	 * so when we forget to add new color space
+	 * compiler will give a warning
+	 */
+	case COLOR_SPACE_UNKNOWN:
+	case COLOR_SPACE_SRGB:
+	case COLOR_SPACE_XR_RGB:
+	case COLOR_SPACE_MSREF_SCRGB:
+	case COLOR_SPACE_XV_YCC_709:
+	case COLOR_SPACE_XV_YCC_601:
+	case COLOR_SPACE_2020_RGB_FULLRANGE:
+	case COLOR_SPACE_2020_RGB_LIMITEDRANGE:
+	case COLOR_SPACE_ADOBERGB:
+	case COLOR_SPACE_DCIP3:
+	case COLOR_SPACE_DISPLAYNATIVE:
+	case COLOR_SPACE_DOLBYVISION:
+	case COLOR_SPACE_APPCTRL:
+	case COLOR_SPACE_CUSTOMPOINTS:
+		/* fefault is sRGB black (full range). */
+		*black_color =
+			black_color_format[BLACK_COLOR_FORMAT_RGB_FULLRANGE];
+		/* default is sRGB black 0. */
+		break;
+	}
+}
+
+bool hwss_wait_for_blank_complete(
+		struct timing_generator *tg)
+{
+	int counter;
+
+	/* Not applicable if the pipe is not primary, save 300ms of boot time */
+	if (!tg->funcs->is_blanked)
+		return true;
+	for (counter = 0; counter < 100; counter++) {
+		if (tg->funcs->is_blanked(tg))
+			break;
+
+		msleep(1);
+	}
+
+	if (counter == 100) {
+		dm_error("DC: failed to blank crtc!\n");
+		return false;
+	}
+
+	return true;
+}
+
+void get_mpctree_visual_confirm_color(
+		struct pipe_ctx *pipe_ctx,
+		struct tg_color *color)
+{
+	const struct tg_color pipe_colors[6] = {
+			{MAX_TG_COLOR_VALUE, 0, 0}, /* red */
+			{MAX_TG_COLOR_VALUE, MAX_TG_COLOR_VALUE, 0}, /* yellow */
+			{0, MAX_TG_COLOR_VALUE, 0}, /* green */
+			{0, MAX_TG_COLOR_VALUE, MAX_TG_COLOR_VALUE}, /* cyan */
+			{0, 0, MAX_TG_COLOR_VALUE}, /* blue */
+			{MAX_TG_COLOR_VALUE, 0, MAX_TG_COLOR_VALUE}, /* magenta */
+	};
+
+	struct pipe_ctx *top_pipe = pipe_ctx;
+
+	while (top_pipe->top_pipe)
+		top_pipe = top_pipe->top_pipe;
+
+	*color = pipe_colors[top_pipe->pipe_idx];
+}
+
+void get_surface_visual_confirm_color(
+		const struct pipe_ctx *pipe_ctx,
+		struct tg_color *color)
+{
+	uint32_t color_value = MAX_TG_COLOR_VALUE;
+
+	switch (pipe_ctx->plane_res.scl_data.format) {
+	case PIXEL_FORMAT_ARGB8888:
+		/* set border color to red */
+		color->color_r_cr = color_value;
+		if (pipe_ctx->plane_state->layer_index > 0) {
+			/* set border color to pink */
+			color->color_b_cb = color_value;
+			color->color_g_y = color_value * 0.5;
+		}
+		break;
+
+	case PIXEL_FORMAT_ARGB2101010:
+		/* set border color to blue */
+		color->color_b_cb = color_value;
+		if (pipe_ctx->plane_state->layer_index > 0) {
+			/* set border color to cyan */
+			color->color_g_y = color_value;
+		}
+		break;
+	case PIXEL_FORMAT_420BPP8:
+		/* set border color to green */
+		color->color_g_y = color_value;
+		break;
+	case PIXEL_FORMAT_420BPP10:
+		/* set border color to yellow */
+		color->color_g_y = color_value;
+		color->color_r_cr = color_value;
+		break;
+	case PIXEL_FORMAT_FP16:
+		/* set border color to white */
+		color->color_r_cr = color_value;
+		color->color_b_cb = color_value;
+		color->color_g_y = color_value;
+		if (pipe_ctx->plane_state->layer_index > 0) {
+			/* set border color to orange */
+			color->color_g_y = 0.22 * color_value;
+			color->color_b_cb = 0;
+		}
+		break;
+	default:
+		break;
+	}
+}
+
+void get_hdr_visual_confirm_color(
+		struct pipe_ctx *pipe_ctx,
+		struct tg_color *color)
+{
+	uint32_t color_value = MAX_TG_COLOR_VALUE;
+	bool is_sdr = false;
+
+	/* Determine the overscan color based on the top-most (desktop) plane's context */
+	struct pipe_ctx *top_pipe_ctx  = pipe_ctx;
+
+	while (top_pipe_ctx->top_pipe != NULL)
+		top_pipe_ctx = top_pipe_ctx->top_pipe;
+
+	switch (top_pipe_ctx->plane_res.scl_data.format) {
+	case PIXEL_FORMAT_ARGB2101010:
+		if (top_pipe_ctx->stream->out_transfer_func.tf == TRANSFER_FUNCTION_PQ) {
+			/* HDR10, ARGB2101010 - set border color to red */
+			color->color_r_cr = color_value;
+		} else if (top_pipe_ctx->stream->out_transfer_func.tf == TRANSFER_FUNCTION_GAMMA22) {
+			/* FreeSync 2 ARGB2101010 - set border color to pink */
+			color->color_r_cr = color_value;
+			color->color_b_cb = color_value;
+		} else
+			is_sdr = true;
+		break;
+	case PIXEL_FORMAT_FP16:
+		if (top_pipe_ctx->stream->out_transfer_func.tf == TRANSFER_FUNCTION_PQ) {
+			/* HDR10, FP16 - set border color to blue */
+			color->color_b_cb = color_value;
+		} else if (top_pipe_ctx->stream->out_transfer_func.tf == TRANSFER_FUNCTION_GAMMA22) {
+			/* FreeSync 2 HDR - set border color to green */
+			color->color_g_y = color_value;
+		} else
+			is_sdr = true;
+		break;
+	default:
+		is_sdr = true;
+		break;
+	}
+
+	if (is_sdr) {
+		/* SDR - set border color to Gray */
+		color->color_r_cr = color_value/2;
+		color->color_b_cb = color_value/2;
+		color->color_g_y = color_value/2;
+	}
+}
+
+/* Visual Confirm color definition for Smart Mux */
+void get_smartmux_visual_confirm_color(
+	struct dc *dc,
+	struct tg_color *color)
+{
+	uint32_t color_value = MAX_TG_COLOR_VALUE;
+
+	const struct tg_color sm_ver_colors[5] = {
+			{0, 0, 0},					/* SMUX_MUXCONTROL_UNSUPPORTED - Black */
+			{0, MAX_TG_COLOR_VALUE, 0},			/* SMUX_MUXCONTROL_v10 - Green */
+			{0, MAX_TG_COLOR_VALUE, MAX_TG_COLOR_VALUE},	/* SMUX_MUXCONTROL_v15 - Cyan */
+			{MAX_TG_COLOR_VALUE, MAX_TG_COLOR_VALUE, 0}, 	/* SMUX_MUXCONTROL_MDM - Yellow */
+			{MAX_TG_COLOR_VALUE, 0, MAX_TG_COLOR_VALUE}, 	/* SMUX_MUXCONTROL_vUNKNOWN - Magenta*/
+	};
+
+	if (dc->caps.is_apu) {
+		/* APU driving the eDP */
+		*color = sm_ver_colors[dc->config.smart_mux_version];
+	} else {
+		/* dGPU driving the eDP - red */
+		color->color_r_cr = color_value;
+		color->color_g_y = 0;
+		color->color_b_cb = 0;
+	}
+}
+
+/* Visual Confirm color definition for VABC */
+void get_vabc_visual_confirm_color(
+	struct pipe_ctx *pipe_ctx,
+	struct tg_color *color)
+{
+	uint32_t color_value = MAX_TG_COLOR_VALUE;
+	struct dc_link *edp_link = NULL;
+
+	if (pipe_ctx && pipe_ctx->stream && pipe_ctx->stream->link) {
+		if (pipe_ctx->stream->link->connector_signal == SIGNAL_TYPE_EDP)
+			edp_link = pipe_ctx->stream->link;
+	}
+
+	if (edp_link) {
+		switch (edp_link->backlight_control_type) {
+		case BACKLIGHT_CONTROL_PWM:
+			color->color_r_cr = color_value;
+			color->color_g_y = 0;
+			color->color_b_cb = 0;
+			break;
+		case BACKLIGHT_CONTROL_AMD_AUX:
+			color->color_r_cr = 0;
+			color->color_g_y = color_value;
+			color->color_b_cb = 0;
+			break;
+		case BACKLIGHT_CONTROL_VESA_AUX:
+			color->color_r_cr = 0;
+			color->color_g_y = 0;
+			color->color_b_cb = color_value;
+			break;
+		}
+	} else {
+		color->color_r_cr = 0;
+		color->color_g_y = 0;
+		color->color_b_cb = 0;
+	}
+}
+
+void get_subvp_visual_confirm_color(
+		struct pipe_ctx *pipe_ctx,
+		struct tg_color *color)
+{
+	uint32_t color_value = MAX_TG_COLOR_VALUE;
+	if (pipe_ctx) {
+		switch (pipe_ctx->p_state_type) {
+		case P_STATE_SUB_VP:
+			color->color_r_cr = color_value;
+			color->color_g_y  = 0;
+			color->color_b_cb = 0;
+			break;
+		case P_STATE_DRR_SUB_VP:
+			color->color_r_cr = 0;
+			color->color_g_y  = color_value;
+			color->color_b_cb = 0;
+			break;
+		case P_STATE_V_BLANK_SUB_VP:
+			color->color_r_cr = 0;
+			color->color_g_y  = 0;
+			color->color_b_cb = color_value;
+			break;
+		default:
+			break;
+		}
+	}
+}
+
+void get_mclk_switch_visual_confirm_color(
+		struct pipe_ctx *pipe_ctx,
+		struct tg_color *color)
+{
+	uint32_t color_value = MAX_TG_COLOR_VALUE;
+
+	if (pipe_ctx) {
+		switch (pipe_ctx->p_state_type) {
+		case P_STATE_V_BLANK:
+			color->color_r_cr = color_value;
+			color->color_g_y = color_value;
+			color->color_b_cb = 0;
+			break;
+		case P_STATE_FPO:
+			color->color_r_cr = 0;
+			color->color_g_y  = color_value;
+			color->color_b_cb = color_value;
+			break;
+		case P_STATE_V_ACTIVE:
+			color->color_r_cr = color_value;
+			color->color_g_y  = 0;
+			color->color_b_cb = color_value;
+			break;
+		case P_STATE_SUB_VP:
+			color->color_r_cr = color_value;
+			color->color_g_y  = 0;
+			color->color_b_cb = 0;
+			break;
+		case P_STATE_DRR_SUB_VP:
+			color->color_r_cr = 0;
+			color->color_g_y  = color_value;
+			color->color_b_cb = 0;
+			break;
+		case P_STATE_V_BLANK_SUB_VP:
+			color->color_r_cr = 0;
+			color->color_g_y  = 0;
+			color->color_b_cb = color_value;
+			break;
+		default:
+			break;
+		}
+	}
+}
+
+void get_cursor_visual_confirm_color(
+		struct pipe_ctx *pipe_ctx,
+		struct tg_color *color)
+{
+	uint32_t color_value = MAX_TG_COLOR_VALUE;
+
+	if (pipe_ctx->stream && pipe_ctx->stream->cursor_position.enable) {
+		color->color_r_cr = color_value;
+		color->color_g_y = 0;
+		color->color_b_cb = 0;
+	} else {
+		color->color_r_cr = 0;
+		color->color_g_y = 0;
+		color->color_b_cb = color_value;
+	}
+}
+
+void get_dcc_visual_confirm_color(
+	struct dc *dc,
+	struct pipe_ctx *pipe_ctx,
+	struct tg_color *color)
+{
+	const uint32_t MCACHE_ID_UNASSIGNED = 0xF;
+
+	if (!pipe_ctx->plane_state->dcc.enable) {
+		color->color_r_cr = 0; /* black - DCC disabled */
+		color->color_g_y = 0;
+		color->color_b_cb = 0;
+		return;
+	}
+
+	if (dc->ctx->dce_version < DCN_VERSION_4_01) {
+		color->color_r_cr = MAX_TG_COLOR_VALUE; /* red - DCC enabled */
+		color->color_g_y = 0;
+		color->color_b_cb = 0;
+		return;
+	}
+
+	uint32_t first_id = pipe_ctx->mcache_regs.main.p0.mcache_id_first;
+	uint32_t second_id = pipe_ctx->mcache_regs.main.p0.mcache_id_second;
+
+	if (first_id != MCACHE_ID_UNASSIGNED && second_id != MCACHE_ID_UNASSIGNED && first_id != second_id) {
+		color->color_r_cr = MAX_TG_COLOR_VALUE/2; /* grey - 2 mcache */
+		color->color_g_y = MAX_TG_COLOR_VALUE/2;
+		color->color_b_cb = MAX_TG_COLOR_VALUE/2;
+	}
+
+	else if (first_id != MCACHE_ID_UNASSIGNED || second_id != MCACHE_ID_UNASSIGNED) {
+		const struct tg_color id_colors[MAX_NUM_MCACHE] = {
+		{0, MAX_TG_COLOR_VALUE, 0}, /* green */
+		{0, 0, MAX_TG_COLOR_VALUE}, /* blue */
+		{MAX_TG_COLOR_VALUE, MAX_TG_COLOR_VALUE, 0}, /* yellow */
+		{MAX_TG_COLOR_VALUE, 0, MAX_TG_COLOR_VALUE}, /* magenta */
+		{0, MAX_TG_COLOR_VALUE, MAX_TG_COLOR_VALUE}, /* cyan */
+		{MAX_TG_COLOR_VALUE, MAX_TG_COLOR_VALUE, MAX_TG_COLOR_VALUE}, /* white */
+		{MAX_TG_COLOR_VALUE/2, 0, 0}, /* dark red */
+		{0, MAX_TG_COLOR_VALUE/2, 0}, /* dark green */
+		};
+
+		uint32_t assigned_id = (first_id != MCACHE_ID_UNASSIGNED) ? first_id : second_id;
+		*color = id_colors[assigned_id];
+	}
+}
+
+void set_p_state_switch_method(
+		struct dc *dc,
+		struct dc_state *context,
+		struct pipe_ctx *pipe_ctx)
+{
+	struct vba_vars_st *vba = &context->bw_ctx.dml.vba;
+	bool enable_subvp;
+
+	if (!dc->ctx || !dc->ctx->dmub_srv || !pipe_ctx || !vba)
+		return;
+
+	pipe_ctx->p_state_type = P_STATE_UNKNOWN;
+	if (vba->DRAMClockChangeSupport[vba->VoltageLevel][vba->maxMpcComb] !=
+			dm_dram_clock_change_unsupported) {
+		/* MCLK switching is supported */
+		if (!pipe_ctx->has_vactive_margin) {
+			/* In Vblank - yellow */
+			pipe_ctx->p_state_type = P_STATE_V_BLANK;
+
+			if (context->bw_ctx.bw.dcn.clk.fw_based_mclk_switching) {
+				/* FPO + Vblank - cyan */
+				pipe_ctx->p_state_type = P_STATE_FPO;
+			}
+		} else {
+			/* In Vactive - pink */
+			pipe_ctx->p_state_type = P_STATE_V_ACTIVE;
+		}
+
+		/* SubVP */
+		enable_subvp = false;
+
+		for (int i = 0; i < dc->res_pool->pipe_count; i++) {
+			struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
+
+			if (pipe->stream && dc_state_get_paired_subvp_stream(context, pipe->stream) &&
+					dc_state_get_pipe_subvp_type(context, pipe) == SUBVP_MAIN) {
+				/* SubVP enable - red */
+				pipe_ctx->p_state_type = P_STATE_SUB_VP;
+				enable_subvp = true;
+
+				if (pipe_ctx->stream == pipe->stream)
+					return;
+				break;
+			}
+		}
+
+		if (enable_subvp && dc_state_get_pipe_subvp_type(context, pipe_ctx) == SUBVP_NONE) {
+			if (pipe_ctx->stream->allow_freesync == 1) {
+				/* SubVP enable and DRR on - green */
+				pipe_ctx->p_state_type = P_STATE_DRR_SUB_VP;
+			} else {
+				/* SubVP enable and No DRR - blue */
+				pipe_ctx->p_state_type = P_STATE_V_BLANK_SUB_VP;
+			}
+		}
+	}
+}
+
+void set_drr_and_clear_adjust_pending(
+		struct pipe_ctx *pipe_ctx,
+		struct dc_stream_state *stream,
+		struct drr_params *params)
+{
+	/* params can be null.*/
+	if (pipe_ctx && pipe_ctx->stream_res.tg &&
+			pipe_ctx->stream_res.tg->funcs->set_drr)
+		pipe_ctx->stream_res.tg->funcs->set_drr(
+				pipe_ctx->stream_res.tg, params);
+
+	if (stream)
+		stream->adjust.timing_adjust_pending = false;
+}
+
+void get_fams2_visual_confirm_color(
+		struct dc *dc,
+		struct dc_state *context,
+		struct pipe_ctx *pipe_ctx,
+		struct tg_color *color)
+{
+	uint32_t color_value = MAX_TG_COLOR_VALUE;
+
+	if (!dc->ctx || !dc->ctx->dmub_srv || !pipe_ctx || !context || !dc->debug.fams2_config.bits.enable)
+		return;
+
+	/* driver only handles visual confirm when FAMS2 is disabled */
+	if (!dc_state_is_fams2_in_use(dc, context)) {
+		/* when FAMS2 is disabled, all pipes are grey */
+		color->color_g_y = color_value / 2;
+		color->color_b_cb = color_value / 2;
+		color->color_r_cr = color_value / 2;
+	}
+}
+
+void hwss_build_fast_sequence(struct dc *dc,
+		struct dc_dmub_cmd *dc_dmub_cmd,
+		unsigned int dmub_cmd_count,
+		struct block_sequence block_sequence[MAX_HWSS_BLOCK_SEQUENCE_SIZE],
+		unsigned int *num_steps,
+		struct pipe_ctx *pipe_ctx,
+		struct dc_stream_status *stream_status,
+		struct dc_state *context)
+{
+	struct dc_plane_state *plane = pipe_ctx->plane_state;
+	struct dc_stream_state *stream = pipe_ctx->stream;
+	struct dce_hwseq *hws = dc->hwseq;
+	struct pipe_ctx *current_pipe = NULL;
+	struct pipe_ctx *current_mpc_pipe = NULL;
+	unsigned int i = 0;
+
+	*num_steps = 0; // Initialize to 0
+
+	if (!plane || !stream)
+		return;
+
+	if (dc->hwss.wait_for_dcc_meta_propagation) {
+		block_sequence[*num_steps].params.wait_for_dcc_meta_propagation_params.dc = dc;
+		block_sequence[*num_steps].params.wait_for_dcc_meta_propagation_params.top_pipe_to_program = pipe_ctx;
+		block_sequence[*num_steps].func = HUBP_WAIT_FOR_DCC_META_PROP;
+		(*num_steps)++;
+	}
+	if (dc->hwss.subvp_pipe_control_lock_fast) {
+		block_sequence[*num_steps].params.subvp_pipe_control_lock_fast_params.dc = dc;
+		block_sequence[*num_steps].params.subvp_pipe_control_lock_fast_params.lock = true;
+		block_sequence[*num_steps].params.subvp_pipe_control_lock_fast_params.subvp_immediate_flip =
+				plane->flip_immediate && stream_status->mall_stream_config.type == SUBVP_MAIN;
+		block_sequence[*num_steps].func = DMUB_SUBVP_PIPE_CONTROL_LOCK_FAST;
+		(*num_steps)++;
+	}
+	if (dc->hwss.fams2_global_control_lock_fast) {
+		block_sequence[*num_steps].params.fams2_global_control_lock_fast_params.dc = dc;
+		block_sequence[*num_steps].params.fams2_global_control_lock_fast_params.lock = true;
+		block_sequence[*num_steps].params.fams2_global_control_lock_fast_params.is_required = dc_state_is_fams2_in_use(dc, context);
+		block_sequence[*num_steps].func = DMUB_FAMS2_GLOBAL_CONTROL_LOCK_FAST;
+		(*num_steps)++;
+	}
+	if (dc->hwss.pipe_control_lock) {
+		block_sequence[*num_steps].params.pipe_control_lock_params.dc = dc;
+		block_sequence[*num_steps].params.pipe_control_lock_params.lock = true;
+		block_sequence[*num_steps].params.pipe_control_lock_params.pipe_ctx = pipe_ctx;
+		block_sequence[*num_steps].func = OPTC_PIPE_CONTROL_LOCK;
+		(*num_steps)++;
+	}
+
+	for (i = 0; i < dmub_cmd_count; i++) {
+		block_sequence[*num_steps].params.send_dmcub_cmd_params.ctx = dc->ctx;
+		block_sequence[*num_steps].params.send_dmcub_cmd_params.cmd = &(dc_dmub_cmd[i].dmub_cmd);
+		block_sequence[*num_steps].params.send_dmcub_cmd_params.wait_type = dc_dmub_cmd[i].wait_type;
+		block_sequence[*num_steps].func = DMUB_SEND_DMCUB_CMD;
+		(*num_steps)++;
+	}
+
+	current_pipe = pipe_ctx;
+	while (current_pipe) {
+		current_mpc_pipe = current_pipe;
+		while (current_mpc_pipe) {
+			if (current_mpc_pipe->plane_state) {
+				if (dc->hwss.set_flip_control_gsl && current_mpc_pipe->plane_state->update_flags.raw) {
+					block_sequence[*num_steps].params.set_flip_control_gsl_params.pipe_ctx = current_mpc_pipe;
+					block_sequence[*num_steps].params.set_flip_control_gsl_params.flip_immediate = current_mpc_pipe->plane_state->flip_immediate;
+					block_sequence[*num_steps].func = HUBP_SET_FLIP_CONTROL_GSL;
+					(*num_steps)++;
+				}
+				if (dc->hwss.program_triplebuffer && dc->debug.enable_tri_buf && current_mpc_pipe->plane_state->update_flags.raw) {
+					block_sequence[*num_steps].params.program_triplebuffer_params.dc = dc;
+					block_sequence[*num_steps].params.program_triplebuffer_params.pipe_ctx = current_mpc_pipe;
+					block_sequence[*num_steps].params.program_triplebuffer_params.enableTripleBuffer = current_mpc_pipe->plane_state->triplebuffer_flips;
+					block_sequence[*num_steps].func = HUBP_PROGRAM_TRIPLEBUFFER;
+					(*num_steps)++;
+				}
+				if (dc->hwss.update_plane_addr && current_mpc_pipe->plane_state->update_flags.bits.addr_update) {
+					if (resource_is_pipe_type(current_mpc_pipe, OTG_MASTER) &&
+							stream_status->mall_stream_config.type == SUBVP_MAIN) {
+						block_sequence[*num_steps].params.subvp_save_surf_addr.dc_dmub_srv = dc->ctx->dmub_srv;
+						block_sequence[*num_steps].params.subvp_save_surf_addr.addr = &current_mpc_pipe->plane_state->address;
+						block_sequence[*num_steps].params.subvp_save_surf_addr.subvp_index = current_mpc_pipe->subvp_index;
+						block_sequence[*num_steps].func = DMUB_SUBVP_SAVE_SURF_ADDR;
+						(*num_steps)++;
+					}
+
+					block_sequence[*num_steps].params.update_plane_addr_params.dc = dc;
+					block_sequence[*num_steps].params.update_plane_addr_params.pipe_ctx = current_mpc_pipe;
+					block_sequence[*num_steps].func = HUBP_UPDATE_PLANE_ADDR;
+					(*num_steps)++;
+				}
+
+				if (hws->funcs.set_input_transfer_func && current_mpc_pipe->plane_state->update_flags.bits.gamma_change) {
+					block_sequence[*num_steps].params.set_input_transfer_func_params.dc = dc;
+					block_sequence[*num_steps].params.set_input_transfer_func_params.pipe_ctx = current_mpc_pipe;
+					block_sequence[*num_steps].params.set_input_transfer_func_params.plane_state = current_mpc_pipe->plane_state;
+					block_sequence[*num_steps].func = DPP_SET_INPUT_TRANSFER_FUNC;
+					(*num_steps)++;
+				}
+
+				if (dc->hwss.program_gamut_remap && current_mpc_pipe->plane_state->update_flags.bits.gamut_remap_change) {
+					block_sequence[*num_steps].params.program_gamut_remap_params.pipe_ctx = current_mpc_pipe;
+					block_sequence[*num_steps].func = DPP_PROGRAM_GAMUT_REMAP;
+					(*num_steps)++;
+				}
+				if (current_mpc_pipe->plane_state->update_flags.bits.input_csc_change) {
+					block_sequence[*num_steps].params.setup_dpp_params.pipe_ctx = current_mpc_pipe;
+					block_sequence[*num_steps].func = DPP_SETUP_DPP;
+					(*num_steps)++;
+				}
+				if (current_mpc_pipe->plane_state->update_flags.bits.coeff_reduction_change) {
+					block_sequence[*num_steps].params.program_bias_and_scale_params.pipe_ctx = current_mpc_pipe;
+					block_sequence[*num_steps].func = DPP_PROGRAM_BIAS_AND_SCALE;
+					(*num_steps)++;
+				}
+			}
+			if (hws->funcs.set_output_transfer_func && current_mpc_pipe->stream->update_flags.bits.out_tf) {
+				block_sequence[*num_steps].params.set_output_transfer_func_params.dc = dc;
+				block_sequence[*num_steps].params.set_output_transfer_func_params.pipe_ctx = current_mpc_pipe;
+				block_sequence[*num_steps].params.set_output_transfer_func_params.stream = current_mpc_pipe->stream;
+				block_sequence[*num_steps].func = DPP_SET_OUTPUT_TRANSFER_FUNC;
+				(*num_steps)++;
+			}
+			if (dc->debug.visual_confirm != VISUAL_CONFIRM_DISABLE &&
+				dc->hwss.update_visual_confirm_color) {
+				block_sequence[*num_steps].params.update_visual_confirm_params.dc = dc;
+				block_sequence[*num_steps].params.update_visual_confirm_params.pipe_ctx = current_mpc_pipe;
+				block_sequence[*num_steps].params.update_visual_confirm_params.mpcc_id = current_mpc_pipe->plane_res.hubp->inst;
+				block_sequence[*num_steps].func = MPC_UPDATE_VISUAL_CONFIRM;
+				(*num_steps)++;
+			}
+			if (current_mpc_pipe->stream->update_flags.bits.out_csc) {
+				block_sequence[*num_steps].params.power_on_mpc_mem_pwr_params.mpc = dc->res_pool->mpc;
+				block_sequence[*num_steps].params.power_on_mpc_mem_pwr_params.mpcc_id = current_mpc_pipe->plane_res.hubp->inst;
+				block_sequence[*num_steps].params.power_on_mpc_mem_pwr_params.power_on = true;
+				block_sequence[*num_steps].func = MPC_POWER_ON_MPC_MEM_PWR;
+				(*num_steps)++;
+
+				if (current_mpc_pipe->stream->csc_color_matrix.enable_adjustment == true) {
+					block_sequence[*num_steps].params.set_output_csc_params.mpc = dc->res_pool->mpc;
+					block_sequence[*num_steps].params.set_output_csc_params.opp_id = current_mpc_pipe->stream_res.opp->inst;
+					block_sequence[*num_steps].params.set_output_csc_params.regval = current_mpc_pipe->stream->csc_color_matrix.matrix;
+					block_sequence[*num_steps].params.set_output_csc_params.ocsc_mode = MPC_OUTPUT_CSC_COEF_A;
+					block_sequence[*num_steps].func = MPC_SET_OUTPUT_CSC;
+					(*num_steps)++;
+				} else {
+					block_sequence[*num_steps].params.set_ocsc_default_params.mpc = dc->res_pool->mpc;
+					block_sequence[*num_steps].params.set_ocsc_default_params.opp_id = current_mpc_pipe->stream_res.opp->inst;
+					block_sequence[*num_steps].params.set_ocsc_default_params.color_space = current_mpc_pipe->stream->output_color_space;
+					block_sequence[*num_steps].params.set_ocsc_default_params.ocsc_mode = MPC_OUTPUT_CSC_COEF_A;
+					block_sequence[*num_steps].func = MPC_SET_OCSC_DEFAULT;
+					(*num_steps)++;
+				}
+			}
+			current_mpc_pipe = current_mpc_pipe->bottom_pipe;
+		}
+		current_pipe = current_pipe->next_odm_pipe;
+	}
+
+	if (dc->hwss.pipe_control_lock) {
+		block_sequence[*num_steps].params.pipe_control_lock_params.dc = dc;
+		block_sequence[*num_steps].params.pipe_control_lock_params.lock = false;
+		block_sequence[*num_steps].params.pipe_control_lock_params.pipe_ctx = pipe_ctx;
+		block_sequence[*num_steps].func = OPTC_PIPE_CONTROL_LOCK;
+		(*num_steps)++;
+	}
+	if (dc->hwss.subvp_pipe_control_lock_fast) {
+		block_sequence[*num_steps].params.subvp_pipe_control_lock_fast_params.dc = dc;
+		block_sequence[*num_steps].params.subvp_pipe_control_lock_fast_params.lock = false;
+		block_sequence[*num_steps].params.subvp_pipe_control_lock_fast_params.subvp_immediate_flip =
+				plane->flip_immediate && stream_status->mall_stream_config.type == SUBVP_MAIN;
+		block_sequence[*num_steps].func = DMUB_SUBVP_PIPE_CONTROL_LOCK_FAST;
+		(*num_steps)++;
+	}
+	if (dc->hwss.fams2_global_control_lock_fast) {
+		block_sequence[*num_steps].params.fams2_global_control_lock_fast_params.dc = dc;
+		block_sequence[*num_steps].params.fams2_global_control_lock_fast_params.lock = false;
+		block_sequence[*num_steps].params.fams2_global_control_lock_fast_params.is_required = dc_state_is_fams2_in_use(dc, context);
+		block_sequence[*num_steps].func = DMUB_FAMS2_GLOBAL_CONTROL_LOCK_FAST;
+		(*num_steps)++;
+	}
+
+	current_pipe = pipe_ctx;
+	while (current_pipe) {
+		current_mpc_pipe = current_pipe;
+
+		while (current_mpc_pipe) {
+			if (!current_mpc_pipe->bottom_pipe && !current_mpc_pipe->next_odm_pipe &&
+					current_mpc_pipe->stream && current_mpc_pipe->plane_state &&
+					current_mpc_pipe->plane_state->update_flags.bits.addr_update &&
+					!current_mpc_pipe->plane_state->skip_manual_trigger) {
+				block_sequence[*num_steps].params.program_manual_trigger_params.pipe_ctx = current_mpc_pipe;
+				block_sequence[*num_steps].func = OPTC_PROGRAM_MANUAL_TRIGGER;
+				(*num_steps)++;
+			}
+			current_mpc_pipe = current_mpc_pipe->bottom_pipe;
+		}
+		current_pipe = current_pipe->next_odm_pipe;
+	}
+}
+
+void hwss_execute_sequence(struct dc *dc,
+		struct block_sequence block_sequence[MAX_HWSS_BLOCK_SEQUENCE_SIZE],
+		int num_steps)
+{
+	unsigned int i;
+	union block_sequence_params *params;
+	struct dce_hwseq *hws = dc->hwseq;
+
+	for (i = 0; i < num_steps; i++) {
+		params = &(block_sequence[i].params);
+		switch (block_sequence[i].func) {
+
+		case DMUB_SUBVP_PIPE_CONTROL_LOCK_FAST:
+			dc->hwss.subvp_pipe_control_lock_fast(params);
+			break;
+		case OPTC_PIPE_CONTROL_LOCK:
+			dc->hwss.pipe_control_lock(params->pipe_control_lock_params.dc,
+					params->pipe_control_lock_params.pipe_ctx,
+					params->pipe_control_lock_params.lock);
+			break;
+		case HUBP_SET_FLIP_CONTROL_GSL:
+			dc->hwss.set_flip_control_gsl(params->set_flip_control_gsl_params.pipe_ctx,
+					params->set_flip_control_gsl_params.flip_immediate);
+			break;
+		case HUBP_PROGRAM_TRIPLEBUFFER:
+			dc->hwss.program_triplebuffer(params->program_triplebuffer_params.dc,
+					params->program_triplebuffer_params.pipe_ctx,
+					params->program_triplebuffer_params.enableTripleBuffer);
+			break;
+		case HUBP_UPDATE_PLANE_ADDR:
+			dc->hwss.update_plane_addr(params->update_plane_addr_params.dc,
+					params->update_plane_addr_params.pipe_ctx);
+			break;
+		case DPP_SET_INPUT_TRANSFER_FUNC:
+			hws->funcs.set_input_transfer_func(params->set_input_transfer_func_params.dc,
+					params->set_input_transfer_func_params.pipe_ctx,
+					params->set_input_transfer_func_params.plane_state);
+			break;
+		case DPP_PROGRAM_GAMUT_REMAP:
+			dc->hwss.program_gamut_remap(params->program_gamut_remap_params.pipe_ctx);
+			break;
+		case DPP_SETUP_DPP:
+			hwss_setup_dpp(params);
+			break;
+		case DPP_PROGRAM_BIAS_AND_SCALE:
+			hwss_program_bias_and_scale(params);
+			break;
+		case OPTC_PROGRAM_MANUAL_TRIGGER:
+			hwss_program_manual_trigger(params);
+			break;
+		case DPP_SET_OUTPUT_TRANSFER_FUNC:
+			hws->funcs.set_output_transfer_func(params->set_output_transfer_func_params.dc,
+					params->set_output_transfer_func_params.pipe_ctx,
+					params->set_output_transfer_func_params.stream);
+			break;
+		case MPC_UPDATE_VISUAL_CONFIRM:
+			dc->hwss.update_visual_confirm_color(params->update_visual_confirm_params.dc,
+					params->update_visual_confirm_params.pipe_ctx,
+					params->update_visual_confirm_params.mpcc_id);
+			break;
+		case MPC_POWER_ON_MPC_MEM_PWR:
+			hwss_power_on_mpc_mem_pwr(params);
+			break;
+		case MPC_SET_OUTPUT_CSC:
+			hwss_set_output_csc(params);
+			break;
+		case MPC_SET_OCSC_DEFAULT:
+			hwss_set_ocsc_default(params);
+			break;
+		case DMUB_SEND_DMCUB_CMD:
+			hwss_send_dmcub_cmd(params);
+			break;
+		case DMUB_SUBVP_SAVE_SURF_ADDR:
+			hwss_subvp_save_surf_addr(params);
+			break;
+		case HUBP_WAIT_FOR_DCC_META_PROP:
+			dc->hwss.wait_for_dcc_meta_propagation(
+					params->wait_for_dcc_meta_propagation_params.dc,
+					params->wait_for_dcc_meta_propagation_params.top_pipe_to_program);
+			break;
+		case DMUB_FAMS2_GLOBAL_CONTROL_LOCK_FAST:
+			dc->hwss.fams2_global_control_lock_fast(params);
+			break;
+		default:
+			ASSERT(false);
+			break;
+		}
+	}
+}
+
+void hwss_send_dmcub_cmd(union block_sequence_params *params)
+{
+	struct dc_context *ctx = params->send_dmcub_cmd_params.ctx;
+	union dmub_rb_cmd *cmd = params->send_dmcub_cmd_params.cmd;
+	enum dm_dmub_wait_type wait_type = params->send_dmcub_cmd_params.wait_type;
+
+	dc_wake_and_execute_dmub_cmd(ctx, cmd, wait_type);
+}
+
+void hwss_program_manual_trigger(union block_sequence_params *params)
+{
+	struct pipe_ctx *pipe_ctx = params->program_manual_trigger_params.pipe_ctx;
+
+	if (pipe_ctx->stream_res.tg->funcs->program_manual_trigger)
+		pipe_ctx->stream_res.tg->funcs->program_manual_trigger(pipe_ctx->stream_res.tg);
+}
+
+void hwss_setup_dpp(union block_sequence_params *params)
+{
+	struct pipe_ctx *pipe_ctx = params->setup_dpp_params.pipe_ctx;
+	struct dpp *dpp = pipe_ctx->plane_res.dpp;
+	struct dc_plane_state *plane_state = pipe_ctx->plane_state;
+
+	if (!plane_state)
+		return;
+
+	if (dpp && dpp->funcs->dpp_setup) {
+		// program the input csc
+		dpp->funcs->dpp_setup(dpp,
+				plane_state->format,
+				EXPANSION_MODE_ZERO,
+				plane_state->input_csc_color_matrix,
+				plane_state->color_space,
+				NULL);
+	}
+
+	if (dpp && dpp->funcs->set_cursor_matrix) {
+		dpp->funcs->set_cursor_matrix(dpp,
+			plane_state->color_space,
+			plane_state->cursor_csc_color_matrix);
+	}
+}
+
+void hwss_program_bias_and_scale(union block_sequence_params *params)
+{
+	struct pipe_ctx *pipe_ctx = params->program_bias_and_scale_params.pipe_ctx;
+	struct dpp *dpp = pipe_ctx->plane_res.dpp;
+	struct dc_plane_state *plane_state = pipe_ctx->plane_state;
+	struct dc_bias_and_scale bns_params = plane_state->bias_and_scale;
+
+	//TODO :for CNVC set scale and bias registers if necessary
+	if (dpp->funcs->dpp_program_bias_and_scale) {
+		dpp->funcs->dpp_program_bias_and_scale(dpp, &bns_params);
+	}
+}
+
+void hwss_power_on_mpc_mem_pwr(union block_sequence_params *params)
+{
+	struct mpc *mpc = params->power_on_mpc_mem_pwr_params.mpc;
+	int mpcc_id = params->power_on_mpc_mem_pwr_params.mpcc_id;
+	bool power_on = params->power_on_mpc_mem_pwr_params.power_on;
+
+	if (mpc->funcs->power_on_mpc_mem_pwr)
+		mpc->funcs->power_on_mpc_mem_pwr(mpc, mpcc_id, power_on);
+}
+
+void hwss_set_output_csc(union block_sequence_params *params)
+{
+	struct mpc *mpc = params->set_output_csc_params.mpc;
+	int opp_id = params->set_output_csc_params.opp_id;
+	const uint16_t *matrix = params->set_output_csc_params.regval;
+	enum mpc_output_csc_mode ocsc_mode = params->set_output_csc_params.ocsc_mode;
+
+	if (mpc->funcs->set_output_csc != NULL)
+		mpc->funcs->set_output_csc(mpc,
+				opp_id,
+				matrix,
+				ocsc_mode);
+}
+
+void hwss_set_ocsc_default(union block_sequence_params *params)
+{
+	struct mpc *mpc = params->set_ocsc_default_params.mpc;
+	int opp_id = params->set_ocsc_default_params.opp_id;
+	enum dc_color_space colorspace = params->set_ocsc_default_params.color_space;
+	enum mpc_output_csc_mode ocsc_mode = params->set_ocsc_default_params.ocsc_mode;
+
+	if (mpc->funcs->set_ocsc_default != NULL)
+		mpc->funcs->set_ocsc_default(mpc,
+				opp_id,
+				colorspace,
+				ocsc_mode);
+}
+
+void hwss_subvp_save_surf_addr(union block_sequence_params *params)
+{
+	struct dc_dmub_srv *dc_dmub_srv = params->subvp_save_surf_addr.dc_dmub_srv;
+	const struct dc_plane_address *addr = params->subvp_save_surf_addr.addr;
+	uint8_t subvp_index = params->subvp_save_surf_addr.subvp_index;
+
+	dc_dmub_srv_subvp_save_surf_addr(dc_dmub_srv, addr, subvp_index);
+}
+
+void get_surface_tile_visual_confirm_color(
+		struct pipe_ctx *pipe_ctx,
+		struct tg_color *color)
+{
+	uint32_t color_value = MAX_TG_COLOR_VALUE;
+	/* Determine the overscan color based on the bottom-most plane's context */
+	struct pipe_ctx *bottom_pipe_ctx  = pipe_ctx;
+
+	while (bottom_pipe_ctx->bottom_pipe != NULL)
+		bottom_pipe_ctx = bottom_pipe_ctx->bottom_pipe;
+
+	switch (bottom_pipe_ctx->plane_state->tiling_info.gfx9.swizzle) {
+	case DC_SW_LINEAR:
+		/* LINEAR Surface - set border color to red */
+		color->color_r_cr = color_value;
+		break;
+	default:
+		break;
+	}
+}
+
+/**
+ * hwss_wait_for_all_blank_complete - wait for all active OPPs to finish pending blank
+ * pattern updates
+ *
+ * @dc: [in] dc reference
+ * @context: [in] hardware context in use
+ */
+void hwss_wait_for_all_blank_complete(struct dc *dc,
+		struct dc_state *context)
+{
+	struct pipe_ctx *opp_head;
+	struct dce_hwseq *hws = dc->hwseq;
+	int i;
+
+	if (!hws->funcs.wait_for_blank_complete)
+		return;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		opp_head = &context->res_ctx.pipe_ctx[i];
+
+		if (!resource_is_pipe_type(opp_head, OPP_HEAD) ||
+				dc_state_get_pipe_subvp_type(context, opp_head) == SUBVP_PHANTOM)
+			continue;
+
+		hws->funcs.wait_for_blank_complete(opp_head->stream_res.opp);
+	}
+}
+
+void hwss_wait_for_odm_update_pending_complete(struct dc *dc, struct dc_state *context)
+{
+	struct pipe_ctx *otg_master;
+	struct timing_generator *tg;
+	int i;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		otg_master = &context->res_ctx.pipe_ctx[i];
+		if (!resource_is_pipe_type(otg_master, OTG_MASTER) ||
+				dc_state_get_pipe_subvp_type(context, otg_master) == SUBVP_PHANTOM)
+			continue;
+		tg = otg_master->stream_res.tg;
+		if (tg->funcs->wait_odm_doublebuffer_pending_clear)
+			tg->funcs->wait_odm_doublebuffer_pending_clear(tg);
+		if (tg->funcs->wait_otg_disable)
+			tg->funcs->wait_otg_disable(tg);
+	}
+
+	/* ODM update may require to reprogram blank pattern for each OPP */
+	hwss_wait_for_all_blank_complete(dc, context);
+}
+
+void hwss_wait_for_no_pipes_pending(struct dc *dc, struct dc_state *context)
+{
+	int i;
+	for (i = 0; i < MAX_PIPES; i++) {
+		int count = 0;
+		struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
+
+		if (!pipe->plane_state || dc_state_get_pipe_subvp_type(context, pipe) == SUBVP_PHANTOM)
+			continue;
+
+		/* Timeout 100 ms */
+		while (count < 100000) {
+			/* Must set to false to start with, due to OR in update function */
+			pipe->plane_state->status.is_flip_pending = false;
+			dc->hwss.update_pending_status(pipe);
+			if (!pipe->plane_state->status.is_flip_pending)
+				break;
+			udelay(1);
+			count++;
+		}
+		ASSERT(!pipe->plane_state->status.is_flip_pending);
+	}
+}
+
+void hwss_wait_for_outstanding_hw_updates(struct dc *dc, struct dc_state *dc_context)
+{
+/*
+ * This function calls HWSS to wait for any potentially double buffered
+ * operations to complete. It should be invoked as a pre-amble prior
+ * to full update programming before asserting any HW locks.
+ */
+	int pipe_idx;
+	int opp_inst;
+	int opp_count = dc->res_pool->res_cap->num_opp;
+	struct hubp *hubp;
+	int mpcc_inst;
+	const struct pipe_ctx *pipe_ctx;
+
+	for (pipe_idx = 0; pipe_idx < dc->res_pool->pipe_count; pipe_idx++) {
+		pipe_ctx = &dc_context->res_ctx.pipe_ctx[pipe_idx];
+
+		if (!pipe_ctx->stream)
+			continue;
+
+		/* For full update we must wait for all double buffer updates, not just DRR updates. This
+		 * is particularly important for minimal transitions. Only check for OTG_MASTER pipes,
+		 * as non-OTG Master pipes share the same OTG as
+		 */
+		if (resource_is_pipe_type(pipe_ctx, OTG_MASTER) && dc->hwss.wait_for_all_pending_updates) {
+			dc->hwss.wait_for_all_pending_updates(pipe_ctx);
+		}
+
+		hubp = pipe_ctx->plane_res.hubp;
+		if (!hubp)
+			continue;
+
+		mpcc_inst = hubp->inst;
+		// MPCC inst is equal to pipe index in practice
+		for (opp_inst = 0; opp_inst < opp_count; opp_inst++) {
+			if ((dc->res_pool->opps[opp_inst] != NULL) &&
+				(dc->res_pool->opps[opp_inst]->mpcc_disconnect_pending[mpcc_inst])) {
+				dc->res_pool->mpc->funcs->wait_for_idle(dc->res_pool->mpc, mpcc_inst);
+				dc->res_pool->opps[opp_inst]->mpcc_disconnect_pending[mpcc_inst] = false;
+				break;
+			}
+		}
+	}
+	hwss_wait_for_odm_update_pending_complete(dc, dc_context);
+}
+
+void hwss_process_outstanding_hw_updates(struct dc *dc, struct dc_state *dc_context)
+{
+	/* wait for outstanding updates */
+	hwss_wait_for_outstanding_hw_updates(dc, dc_context);
+
+	/* perform outstanding post update programming */
+	if (dc->hwss.program_outstanding_updates)
+		dc->hwss.program_outstanding_updates(dc, dc_context);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link_enc_cfg.c b/drivers/gpu/drm/amd/display/dc/core/dc_link_enc_cfg.c
new file mode 100644
index 000000000000..a180f68f711c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_link_enc_cfg.c
@@ -0,0 +1,745 @@
+/* Copyright 2021 Advanced Micro Devices, Inc. All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "link_enc_cfg.h"
+#include "resource.h"
+#include "link_service.h"
+
+#define DC_LOGGER dc->ctx->logger
+
+/* Check whether stream is supported by DIG link encoders. */
+static bool is_dig_link_enc_stream(struct dc_stream_state *stream)
+{
+	bool is_dig_stream = false;
+	struct link_encoder *link_enc = NULL;
+	int i;
+
+	/* Loop over created link encoder objects. */
+	if (stream) {
+		for (i = 0; i < stream->ctx->dc->res_pool->res_cap->num_dig_link_enc; i++) {
+			link_enc = stream->ctx->dc->res_pool->link_encoders[i];
+
+			/* Need to check link signal type rather than stream signal type which may not
+			 * yet match.
+			 */
+			if (link_enc && ((uint32_t)stream->link->connector_signal & link_enc->output_signals)) {
+				is_dig_stream = true;
+				break;
+			}
+		}
+	}
+	return is_dig_stream;
+}
+
+static struct link_enc_assignment get_assignment(struct dc *dc, int i)
+{
+	struct link_enc_assignment assignment;
+
+	if (dc->current_state->res_ctx.link_enc_cfg_ctx.mode == LINK_ENC_CFG_TRANSIENT)
+		assignment = dc->current_state->res_ctx.link_enc_cfg_ctx.transient_assignments[i];
+	else /* LINK_ENC_CFG_STEADY */
+		assignment = dc->current_state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i];
+
+	return assignment;
+}
+
+/* Return stream using DIG link encoder resource. NULL if unused. */
+static struct dc_stream_state *get_stream_using_link_enc(
+		struct dc_state *state,
+		enum engine_id eng_id)
+{
+	struct dc_stream_state *stream = NULL;
+	int i;
+
+	for (i = 0; i < state->stream_count; i++) {
+		struct link_enc_assignment assignment = state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i];
+
+		if ((assignment.valid == true) && (assignment.eng_id == eng_id)) {
+			stream = state->streams[i];
+			break;
+		}
+	}
+
+	return stream;
+}
+
+static void remove_link_enc_assignment(
+		struct dc_state *state,
+		struct dc_stream_state *stream,
+		enum engine_id eng_id)
+{
+	int eng_idx;
+	int i;
+
+	if (eng_id != ENGINE_ID_UNKNOWN) {
+		eng_idx = eng_id - ENGINE_ID_DIGA;
+
+		/* stream ptr of stream in dc_state used to update correct entry in
+		 * link_enc_assignments table.
+		 */
+		for (i = 0; i < MAX_PIPES; i++) {
+			struct link_enc_assignment assignment = state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i];
+
+			if (assignment.valid && assignment.stream == stream) {
+				state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i].valid = false;
+				/* Only add link encoder back to availability pool if not being
+				 * used by any other stream (i.e. removing SST stream or last MST stream).
+				 */
+				if (get_stream_using_link_enc(state, eng_id) == NULL)
+					state->res_ctx.link_enc_cfg_ctx.link_enc_avail[eng_idx] = eng_id;
+
+				stream->link_enc = NULL;
+				state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i].eng_id = ENGINE_ID_UNKNOWN;
+				state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i].stream = NULL;
+				dc_stream_release(stream);
+				break;
+			}
+		}
+	}
+}
+
+static void add_link_enc_assignment(
+		struct dc_state *state,
+		struct dc_stream_state *stream,
+		enum engine_id eng_id)
+{
+	int eng_idx;
+	int i;
+
+	if (eng_id != ENGINE_ID_UNKNOWN) {
+		eng_idx = eng_id - ENGINE_ID_DIGA;
+
+		/* stream ptr of stream in dc_state used to update correct entry in
+		 * link_enc_assignments table.
+		 */
+		for (i = 0; i < state->stream_count; i++) {
+			if (stream == state->streams[i]) {
+				state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i] = (struct link_enc_assignment){
+					.valid = true,
+					.ep_id = (struct display_endpoint_id) {
+						.link_id = stream->link->link_id,
+						.ep_type = stream->link->ep_type},
+					.eng_id = eng_id,
+					.stream = stream};
+				dc_stream_retain(stream);
+				state->res_ctx.link_enc_cfg_ctx.link_enc_avail[eng_idx] = ENGINE_ID_UNKNOWN;
+				stream->link_enc = stream->ctx->dc->res_pool->link_encoders[eng_idx];
+				break;
+			}
+		}
+
+		/* Attempted to add an encoder assignment for a stream not in dc_state. */
+		ASSERT(i != state->stream_count);
+	}
+}
+
+/* Return first available DIG link encoder. */
+static enum engine_id find_first_avail_link_enc(
+		const struct dc_context *ctx,
+		const struct dc_state *state,
+		enum engine_id eng_id_requested)
+{
+	enum engine_id eng_id = ENGINE_ID_UNKNOWN;
+	int i;
+
+	if (eng_id_requested != ENGINE_ID_UNKNOWN) {
+
+		for (i = 0; i < ctx->dc->res_pool->res_cap->num_dig_link_enc; i++) {
+			eng_id = state->res_ctx.link_enc_cfg_ctx.link_enc_avail[i];
+			if (eng_id == eng_id_requested)
+				return eng_id;
+		}
+	}
+
+	eng_id = ENGINE_ID_UNKNOWN;
+
+	for (i = 0; i < ctx->dc->res_pool->res_cap->num_dig_link_enc; i++) {
+		eng_id = state->res_ctx.link_enc_cfg_ctx.link_enc_avail[i];
+		if (eng_id != ENGINE_ID_UNKNOWN)
+			break;
+	}
+
+	return eng_id;
+}
+
+/* Check for availability of link encoder eng_id. */
+static bool is_avail_link_enc(struct dc_state *state, enum engine_id eng_id, struct dc_stream_state *stream)
+{
+	bool is_avail = false;
+	int eng_idx = eng_id - ENGINE_ID_DIGA;
+
+	/* An encoder is available if it is still in the availability pool. */
+	if (eng_id != ENGINE_ID_UNKNOWN && state->res_ctx.link_enc_cfg_ctx.link_enc_avail[eng_idx] != ENGINE_ID_UNKNOWN) {
+		is_avail = true;
+	} else {
+		struct dc_stream_state *stream_assigned = NULL;
+
+		/* MST streams share the same link and should share the same encoder.
+		 * If a stream that has already been assigned a link encoder uses as the
+		 * same link as the stream checking for availability, it is an MST stream
+		 * and should use the same link encoder.
+		 */
+		stream_assigned = get_stream_using_link_enc(state, eng_id);
+		if (stream_assigned && stream != stream_assigned && stream->link == stream_assigned->link)
+			is_avail = true;
+	}
+
+	return is_avail;
+}
+
+/* Test for display_endpoint_id equality. */
+static bool are_ep_ids_equal(struct display_endpoint_id *lhs, struct display_endpoint_id *rhs)
+{
+	bool are_equal = false;
+
+	if (lhs->link_id.id == rhs->link_id.id &&
+			lhs->link_id.enum_id == rhs->link_id.enum_id &&
+			lhs->link_id.type == rhs->link_id.type &&
+			lhs->ep_type == rhs->ep_type)
+		are_equal = true;
+
+	return are_equal;
+}
+
+static struct link_encoder *get_link_enc_used_by_link(
+		struct dc_state *state,
+		const struct dc_link *link)
+{
+	struct link_encoder *link_enc = NULL;
+	struct display_endpoint_id ep_id;
+	int i;
+
+	ep_id = (struct display_endpoint_id) {
+		.link_id = link->link_id,
+		.ep_type = link->ep_type};
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct link_enc_assignment assignment = state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i];
+		if (assignment.eng_id == ENGINE_ID_UNKNOWN)
+			continue;
+
+		if (assignment.valid == true && are_ep_ids_equal(&assignment.ep_id, &ep_id))
+			link_enc = link->dc->res_pool->link_encoders[assignment.eng_id - ENGINE_ID_DIGA];
+	}
+
+	return link_enc;
+}
+/* Clear all link encoder assignments. */
+static void clear_enc_assignments(const struct dc *dc, struct dc_state *state)
+{
+	int i;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i].valid = false;
+		state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i].eng_id = ENGINE_ID_UNKNOWN;
+		if (state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i].stream != NULL) {
+			dc_stream_release(state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i].stream);
+			state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i].stream = NULL;
+		}
+	}
+
+	for (i = 0; i < dc->res_pool->res_cap->num_dig_link_enc; i++) {
+		if (dc->res_pool->link_encoders[i])
+			state->res_ctx.link_enc_cfg_ctx.link_enc_avail[i] = (enum engine_id) i;
+		else
+			state->res_ctx.link_enc_cfg_ctx.link_enc_avail[i] = ENGINE_ID_UNKNOWN;
+	}
+}
+
+void link_enc_cfg_init(
+		const struct dc *dc,
+		struct dc_state *state)
+{
+	clear_enc_assignments(dc, state);
+
+	state->res_ctx.link_enc_cfg_ctx.mode = LINK_ENC_CFG_STEADY;
+}
+
+void link_enc_cfg_copy(const struct dc_state *src_ctx, struct dc_state *dst_ctx)
+{
+	memcpy(&dst_ctx->res_ctx.link_enc_cfg_ctx,
+	       &src_ctx->res_ctx.link_enc_cfg_ctx,
+	       sizeof(dst_ctx->res_ctx.link_enc_cfg_ctx));
+}
+
+void link_enc_cfg_link_encs_assign(
+		struct dc *dc,
+		struct dc_state *state,
+		struct dc_stream_state *streams[],
+		uint8_t stream_count)
+{
+	enum engine_id eng_id = ENGINE_ID_UNKNOWN, eng_id_req = ENGINE_ID_UNKNOWN;
+	int i;
+	int j;
+
+	ASSERT(state->stream_count == stream_count);
+	ASSERT(dc->current_state->res_ctx.link_enc_cfg_ctx.mode == LINK_ENC_CFG_STEADY);
+
+	/* Release DIG link encoder resources before running assignment algorithm. */
+	for (i = 0; i < dc->current_state->stream_count; i++)
+		dc->res_pool->funcs->link_enc_unassign(state, dc->current_state->streams[i]);
+
+	for (i = 0; i < MAX_PIPES; i++)
+		ASSERT(state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i].valid == false);
+
+	/* (a) Assign DIG link encoders to physical (unmappable) endpoints first. */
+	for (i = 0; i < stream_count; i++) {
+		struct dc_stream_state *stream = streams[i];
+
+		/* skip it if the link is mappable endpoint. */
+		if (stream->link->is_dig_mapping_flexible)
+			continue;
+
+		/* Skip stream if not supported by DIG link encoder. */
+		if (!is_dig_link_enc_stream(stream))
+			continue;
+
+		/* Physical endpoints have a fixed mapping to DIG link encoders. */
+		eng_id = stream->link->eng_id;
+		add_link_enc_assignment(state, stream, eng_id);
+	}
+
+	/* (b) Retain previous assignments for mappable endpoints if encoders still available. */
+	eng_id = ENGINE_ID_UNKNOWN;
+
+	if (state != dc->current_state) {
+		struct dc_state *prev_state = dc->current_state;
+
+		for (i = 0; i < stream_count; i++) {
+			struct dc_stream_state *stream = state->streams[i];
+
+			/* Skip it if the link is NOT mappable endpoint. */
+			if (!stream->link->is_dig_mapping_flexible)
+				continue;
+
+			/* Skip stream if not supported by DIG link encoder. */
+			if (!is_dig_link_enc_stream(stream))
+				continue;
+
+			for (j = 0; j < prev_state->stream_count; j++) {
+				struct dc_stream_state *prev_stream = prev_state->streams[j];
+
+				if (stream == prev_stream && stream->link == prev_stream->link &&
+						prev_state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[j].valid) {
+					eng_id = prev_state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[j].eng_id;
+
+					if (is_avail_link_enc(state, eng_id, stream))
+						add_link_enc_assignment(state, stream, eng_id);
+				}
+			}
+		}
+	}
+
+	/* (c) Then assign encoders to remaining mappable endpoints. */
+	eng_id = ENGINE_ID_UNKNOWN;
+
+	for (i = 0; i < stream_count; i++) {
+		struct dc_stream_state *stream = streams[i];
+		struct link_encoder *link_enc = NULL;
+
+		/* Skip it if the link is NOT mappable endpoint. */
+		if (!stream->link->is_dig_mapping_flexible)
+			continue;
+
+		/* Skip if encoder assignment retained in step (b) above. */
+		if (stream->link_enc)
+			continue;
+
+		/* Skip stream if not supported by DIG link encoder. */
+		if (!is_dig_link_enc_stream(stream)) {
+			ASSERT(stream->link->is_dig_mapping_flexible != true);
+			continue;
+		}
+
+		/* Mappable endpoints have a flexible mapping to DIG link encoders. */
+
+		/* For MST, multiple streams will share the same link / display
+		 * endpoint. These streams should use the same link encoder
+		 * assigned to that endpoint.
+		 */
+		link_enc = get_link_enc_used_by_link(state, stream->link);
+		if (link_enc == NULL) {
+
+			if (stream->link->ep_type == DISPLAY_ENDPOINT_USB4_DPIA &&
+					stream->link->dpia_preferred_eng_id != ENGINE_ID_UNKNOWN)
+				eng_id_req = stream->link->dpia_preferred_eng_id;
+
+			eng_id = find_first_avail_link_enc(stream->ctx, state, eng_id_req);
+		}
+		else
+			eng_id =  link_enc->preferred_engine;
+
+		add_link_enc_assignment(state, stream, eng_id);
+	}
+
+	link_enc_cfg_validate(dc, state);
+
+	/* Update transient assignments. */
+	for (i = 0; i < MAX_PIPES; i++) {
+		dc->current_state->res_ctx.link_enc_cfg_ctx.transient_assignments[i] =
+			state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i];
+	}
+
+	/* Log encoder assignments. */
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct link_enc_assignment assignment =
+				dc->current_state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i];
+
+		if (assignment.valid)
+			DC_LOG_DEBUG("%s: CUR %s(%d) - enc_id(%d)\n",
+					__func__,
+					assignment.ep_id.ep_type == DISPLAY_ENDPOINT_PHY ? "PHY" : "DPIA",
+					assignment.ep_id.ep_type == DISPLAY_ENDPOINT_PHY ?
+							assignment.ep_id.link_id.enum_id :
+							assignment.ep_id.link_id.enum_id - 1,
+					assignment.eng_id);
+	}
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct link_enc_assignment assignment =
+				state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i];
+
+		if (assignment.valid)
+			DC_LOG_DEBUG("%s: NEW %s(%d) - enc_id(%d)\n",
+					__func__,
+					assignment.ep_id.ep_type == DISPLAY_ENDPOINT_PHY ? "PHY" : "DPIA",
+					assignment.ep_id.ep_type == DISPLAY_ENDPOINT_PHY ?
+							assignment.ep_id.link_id.enum_id :
+							assignment.ep_id.link_id.enum_id - 1,
+					assignment.eng_id);
+	}
+
+	/* Current state mode will be set to steady once this state committed. */
+	state->res_ctx.link_enc_cfg_ctx.mode = LINK_ENC_CFG_STEADY;
+}
+
+void link_enc_cfg_link_enc_unassign(
+		struct dc_state *state,
+		struct dc_stream_state *stream)
+{
+	enum engine_id eng_id = ENGINE_ID_UNKNOWN;
+
+	if (stream->link_enc)
+		eng_id = stream->link_enc->preferred_engine;
+
+	remove_link_enc_assignment(state, stream, eng_id);
+}
+
+bool link_enc_cfg_is_transmitter_mappable(
+		struct dc *dc,
+		struct link_encoder *link_enc)
+{
+	bool is_mappable = false;
+	enum engine_id eng_id = link_enc->preferred_engine;
+	struct dc_stream_state *stream = link_enc_cfg_get_stream_using_link_enc(dc, eng_id);
+
+	if (stream)
+		is_mappable = stream->link->is_dig_mapping_flexible;
+
+	return is_mappable;
+}
+
+struct dc_stream_state *link_enc_cfg_get_stream_using_link_enc(
+		struct dc *dc,
+		enum engine_id eng_id)
+{
+	struct dc_stream_state *stream = NULL;
+	int i;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct link_enc_assignment assignment = get_assignment(dc, i);
+
+		if ((assignment.valid == true) && (assignment.eng_id == eng_id)) {
+			stream = assignment.stream;
+			break;
+		}
+	}
+
+	return stream;
+}
+
+struct dc_link *link_enc_cfg_get_link_using_link_enc(
+		struct dc *dc,
+		enum engine_id eng_id)
+{
+	struct dc_link *link = NULL;
+	struct dc_stream_state *stream = NULL;
+
+	stream = link_enc_cfg_get_stream_using_link_enc(dc, eng_id);
+
+	if (stream)
+		link = stream->link;
+
+	return link;
+}
+
+struct link_encoder *link_enc_cfg_get_link_enc_used_by_link(
+		struct dc *dc,
+		const struct dc_link *link)
+{
+	struct link_encoder *link_enc = NULL;
+	struct display_endpoint_id ep_id;
+	int i;
+
+	ep_id = (struct display_endpoint_id) {
+		.link_id = link->link_id,
+		.ep_type = link->ep_type};
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct link_enc_assignment assignment = get_assignment(dc, i);
+		if (assignment.eng_id == ENGINE_ID_UNKNOWN)
+			continue;
+
+		if (assignment.valid == true && are_ep_ids_equal(&assignment.ep_id, &ep_id)) {
+			link_enc = link->dc->res_pool->link_encoders[assignment.eng_id - ENGINE_ID_DIGA];
+			break;
+		}
+	}
+
+	return link_enc;
+}
+
+struct link_encoder *link_enc_cfg_get_next_avail_link_enc(struct dc *dc)
+{
+	struct link_encoder *link_enc = NULL;
+	enum engine_id encs_assigned[MAX_DIG_LINK_ENCODERS];
+	int i;
+
+	for (i = 0; i < MAX_DIG_LINK_ENCODERS; i++)
+		encs_assigned[i] = ENGINE_ID_UNKNOWN;
+
+	/* Add assigned encoders to list. */
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct link_enc_assignment assignment = get_assignment(dc, i);
+
+		if (assignment.valid && assignment.eng_id != ENGINE_ID_UNKNOWN)
+			encs_assigned[assignment.eng_id - ENGINE_ID_DIGA] = assignment.eng_id;
+	}
+
+	for (i = 0; i < dc->res_pool->res_cap->num_dig_link_enc; i++) {
+		if (encs_assigned[i] == ENGINE_ID_UNKNOWN &&
+				dc->res_pool->link_encoders[i] != NULL) {
+			link_enc = dc->res_pool->link_encoders[i];
+			break;
+		}
+	}
+
+	return link_enc;
+}
+
+struct link_encoder *link_enc_cfg_get_link_enc(
+		const struct dc_link *link)
+{
+	struct link_encoder *link_enc = NULL;
+
+	/* Links supporting dynamically assigned link encoder will be assigned next
+	 * available encoder if one not already assigned.
+	 */
+	if (link->is_dig_mapping_flexible &&
+	    link->dc->res_pool->funcs->link_encs_assign) {
+		link_enc = link_enc_cfg_get_link_enc_used_by_link(link->ctx->dc, link);
+		if (link_enc == NULL)
+			link_enc = link_enc_cfg_get_next_avail_link_enc(
+				link->ctx->dc);
+	} else
+		link_enc = link->link_enc;
+
+	return link_enc;
+}
+
+struct link_encoder *link_enc_cfg_get_link_enc_used_by_stream_current(
+		struct dc *dc,
+		const struct dc_stream_state *stream)
+{
+	struct link_encoder *link_enc = NULL;
+	struct display_endpoint_id ep_id;
+	int i;
+
+	ep_id = (struct display_endpoint_id) {
+		.link_id = stream->link->link_id,
+		.ep_type = stream->link->ep_type};
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct link_enc_assignment assignment =
+			dc->current_state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i];
+
+		if (assignment.eng_id == ENGINE_ID_UNKNOWN)
+			continue;
+
+		if (assignment.valid == true && are_ep_ids_equal(&assignment.ep_id, &ep_id)) {
+			link_enc = stream->link->dc->res_pool->link_encoders[assignment.eng_id - ENGINE_ID_DIGA];
+			break;
+		}
+	}
+
+	return link_enc;
+}
+
+bool link_enc_cfg_is_link_enc_avail(struct dc *dc, enum engine_id eng_id, struct dc_link *link)
+{
+	bool is_avail = true;
+	int i;
+
+	/* An encoder is not available if it has already been assigned to a different endpoint. */
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct link_enc_assignment assignment = get_assignment(dc, i);
+		struct display_endpoint_id ep_id = (struct display_endpoint_id) {
+				.link_id = link->link_id,
+				.ep_type = link->ep_type};
+
+		if (assignment.valid && assignment.eng_id == eng_id && !are_ep_ids_equal(&ep_id, &assignment.ep_id)) {
+			is_avail = false;
+			break;
+		}
+	}
+
+	return is_avail;
+}
+
+bool link_enc_cfg_validate(struct dc *dc, struct dc_state *state)
+{
+	bool is_valid = false;
+	bool valid_entries = true;
+	bool valid_stream_ptrs = true;
+	bool valid_uniqueness = true;
+	bool valid_avail = true;
+	bool valid_streams = true;
+	int i, j;
+	uint8_t valid_count = 0;
+	uint8_t dig_stream_count = 0;
+	int eng_ids_per_ep_id[MAX_PIPES] = {0};
+	int ep_ids_per_eng_id[MAX_PIPES] = {0};
+	int valid_bitmap = 0;
+
+	/* (1) No. valid entries same as stream count. */
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct link_enc_assignment assignment = state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i];
+
+		if (assignment.valid)
+			valid_count++;
+
+		if (is_dig_link_enc_stream(state->streams[i]))
+			dig_stream_count++;
+	}
+	if (valid_count != dig_stream_count)
+		valid_entries = false;
+
+	/* (2) Matching stream ptrs. */
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct link_enc_assignment assignment = state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i];
+
+		if (assignment.valid) {
+			if (assignment.stream != state->streams[i])
+				valid_stream_ptrs = false;
+		}
+	}
+
+	/* (3) Each endpoint assigned unique encoder. */
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct link_enc_assignment assignment_i = state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i];
+
+		if (assignment_i.valid) {
+			struct display_endpoint_id ep_id_i = assignment_i.ep_id;
+
+			eng_ids_per_ep_id[i]++;
+			ep_ids_per_eng_id[i]++;
+			for (j = 0; j < MAX_PIPES; j++) {
+				struct link_enc_assignment assignment_j =
+					state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[j];
+
+				if (j == i)
+					continue;
+
+				if (assignment_j.valid) {
+					struct display_endpoint_id ep_id_j = assignment_j.ep_id;
+
+					if (are_ep_ids_equal(&ep_id_i, &ep_id_j) &&
+							assignment_i.eng_id != assignment_j.eng_id) {
+						valid_uniqueness = false;
+						eng_ids_per_ep_id[i]++;
+					} else if (!are_ep_ids_equal(&ep_id_i, &ep_id_j) &&
+							assignment_i.eng_id == assignment_j.eng_id) {
+						valid_uniqueness = false;
+						ep_ids_per_eng_id[i]++;
+					}
+				}
+			}
+		}
+	}
+
+	/* (4) Assigned encoders not in available pool. */
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct link_enc_assignment assignment = state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i];
+
+		if (assignment.valid) {
+			for (j = 0; j < dc->res_pool->res_cap->num_dig_link_enc; j++) {
+				if (state->res_ctx.link_enc_cfg_ctx.link_enc_avail[j] == assignment.eng_id) {
+					valid_avail = false;
+					break;
+				}
+			}
+		}
+	}
+
+	/* (5) All streams have valid link encoders. */
+	for (i = 0; i < state->stream_count; i++) {
+		struct dc_stream_state *stream = state->streams[i];
+
+		if (is_dig_link_enc_stream(stream) && stream->link_enc == NULL) {
+			valid_streams = false;
+			break;
+		}
+	}
+
+	is_valid = valid_entries && valid_stream_ptrs && valid_uniqueness && valid_avail && valid_streams;
+	ASSERT(is_valid);
+
+	if (is_valid == false) {
+		valid_bitmap =
+			(valid_entries & 0x1) |
+			((valid_stream_ptrs & 0x1) << 1) |
+			((valid_uniqueness & 0x1) << 2) |
+			((valid_avail & 0x1) << 3) |
+			((valid_streams & 0x1) << 4);
+		DC_LOG_ERROR("%s: Invalid link encoder assignments - 0x%x\n", __func__, valid_bitmap);
+	}
+
+	return is_valid;
+}
+
+void link_enc_cfg_set_transient_mode(struct dc *dc, struct dc_state *current_state, struct dc_state *new_state)
+{
+	int i = 0;
+	int num_transient_assignments = 0;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		if (current_state->res_ctx.link_enc_cfg_ctx.transient_assignments[i].valid)
+			num_transient_assignments++;
+	}
+
+	/* Only enter transient mode if the new encoder assignments are valid. */
+	if (new_state->stream_count == num_transient_assignments) {
+		current_state->res_ctx.link_enc_cfg_ctx.mode = LINK_ENC_CFG_TRANSIENT;
+		DC_LOG_DEBUG("%s: current_state(%p) mode(%d)\n", __func__, current_state, LINK_ENC_CFG_TRANSIENT);
+	}
+}
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link_exports.c b/drivers/gpu/drm/amd/display/dc/core/dc_link_exports.c
new file mode 100644
index 000000000000..9acd30019717
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_link_exports.c
@@ -0,0 +1,530 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+/* FILE POLICY AND INTENDED USAGE:
+ * This file provides single entrance to link functionality declared in dc
+ * public headers. The file is intended to be used as a thin translation layer
+ * that directly calls link internal functions without adding new functional
+ * behavior.
+ *
+ * When exporting a new link related dc function, add function declaration in
+ * dc.h with detail interface documentation, then add function implementation
+ * in this file which calls link functions.
+ */
+#include "link_service.h"
+#include "dce/dce_i2c.h"
+
+struct dc_link *dc_get_link_at_index(struct dc *dc, uint32_t link_index)
+{
+	if (link_index >= MAX_LINKS)
+		return NULL;
+
+	return dc->links[link_index];
+}
+
+void dc_get_edp_links(const struct dc *dc,
+		struct dc_link **edp_links,
+		int *edp_num)
+{
+	int i;
+
+	*edp_num = 0;
+	for (i = 0; i < dc->link_count; i++) {
+		// report any eDP links, even unconnected DDI's
+		if (!dc->links[i])
+			continue;
+		if (dc->links[i]->connector_signal == SIGNAL_TYPE_EDP) {
+			edp_links[*edp_num] = dc->links[i];
+			if (++(*edp_num) == MAX_NUM_EDP)
+				return;
+		}
+	}
+}
+
+bool dc_get_edp_link_panel_inst(const struct dc *dc,
+		const struct dc_link *link,
+		unsigned int *inst_out)
+{
+	struct dc_link *edp_links[MAX_NUM_EDP];
+	int edp_num, i;
+
+	*inst_out = 0;
+	if (link->connector_signal != SIGNAL_TYPE_EDP)
+		return false;
+	dc_get_edp_links(dc, edp_links, &edp_num);
+	for (i = 0; i < edp_num; i++) {
+		if (link == edp_links[i])
+			break;
+		(*inst_out)++;
+	}
+	return true;
+}
+
+bool dc_link_detect(struct dc_link *link, enum dc_detect_reason reason)
+{
+	return link->dc->link_srv->detect_link(link, reason);
+}
+
+bool dc_link_detect_connection_type(struct dc_link *link,
+		enum dc_connection_type *type)
+{
+	return link->dc->link_srv->detect_connection_type(link, type);
+}
+
+const struct dc_link_status *dc_link_get_status(const struct dc_link *link)
+{
+	return link->dc->link_srv->get_status(link);
+}
+
+/* return true if the connected receiver supports the hdcp version */
+bool dc_link_is_hdcp14(struct dc_link *link, enum signal_type signal)
+{
+	return link->dc->link_srv->is_hdcp1x_supported(link, signal);
+}
+
+bool dc_link_is_hdcp22(struct dc_link *link, enum signal_type signal)
+{
+	return link->dc->link_srv->is_hdcp2x_supported(link, signal);
+}
+
+void dc_link_clear_dprx_states(struct dc_link *link)
+{
+	link->dc->link_srv->clear_dprx_states(link);
+}
+
+bool dc_link_reset_cur_dp_mst_topology(struct dc_link *link)
+{
+	return link->dc->link_srv->reset_cur_dp_mst_topology(link);
+}
+
+uint32_t dc_link_bandwidth_kbps(
+	const struct dc_link *link,
+	const struct dc_link_settings *link_settings)
+{
+	return link->dc->link_srv->dp_link_bandwidth_kbps(link, link_settings);
+}
+
+uint32_t dc_link_required_hblank_size_bytes(
+	const struct dc_link *link,
+	struct dp_audio_bandwidth_params *audio_params)
+{
+	return link->dc->link_srv->dp_required_hblank_size_bytes(link,
+			audio_params);
+}
+
+void dc_get_cur_link_res_map(const struct dc *dc, uint32_t *map)
+{
+	dc->link_srv->get_cur_res_map(dc, map);
+}
+
+void dc_restore_link_res_map(const struct dc *dc, uint32_t *map)
+{
+	dc->link_srv->restore_res_map(dc, map);
+}
+
+bool dc_link_update_dsc_config(struct pipe_ctx *pipe_ctx)
+{
+	struct dc_link *link = pipe_ctx->stream->link;
+
+	return link->dc->link_srv->update_dsc_config(pipe_ctx);
+}
+
+struct ddc_service *
+dc_get_oem_i2c_device(struct dc *dc)
+{
+	return dc->res_pool->oem_device;
+}
+
+bool dc_is_oem_i2c_device_present(
+	struct dc *dc,
+	size_t slave_address)
+{
+	if (dc->res_pool->oem_device)
+		return dce_i2c_oem_device_present(
+			dc->res_pool,
+			dc->res_pool->oem_device,
+			slave_address);
+
+	return false;
+}
+
+bool dc_submit_i2c(
+		struct dc *dc,
+		uint32_t link_index,
+		struct i2c_command *cmd)
+{
+
+	struct dc_link *link = dc->links[link_index];
+	struct ddc_service *ddc = link->ddc;
+
+	return dce_i2c_submit_command(
+		dc->res_pool,
+		ddc->ddc_pin,
+		cmd);
+}
+
+bool dc_submit_i2c_oem(
+		struct dc *dc,
+		struct i2c_command *cmd)
+{
+	struct ddc_service *ddc = dc->res_pool->oem_device;
+
+	if (ddc)
+		return dce_i2c_submit_command(
+			dc->res_pool,
+			ddc->ddc_pin,
+			cmd);
+
+	return false;
+}
+
+void dc_link_dp_handle_automated_test(struct dc_link *link)
+{
+	link->dc->link_srv->dp_handle_automated_test(link);
+}
+
+bool dc_link_dp_set_test_pattern(
+	struct dc_link *link,
+	enum dp_test_pattern test_pattern,
+	enum dp_test_pattern_color_space test_pattern_color_space,
+	const struct link_training_settings *p_link_settings,
+	const unsigned char *p_custom_pattern,
+	unsigned int cust_pattern_size)
+{
+	return link->dc->link_srv->dp_set_test_pattern(link, test_pattern,
+			test_pattern_color_space, p_link_settings,
+			p_custom_pattern, cust_pattern_size);
+}
+
+void dc_link_set_drive_settings(struct dc *dc,
+				struct link_training_settings *lt_settings,
+				struct dc_link *link)
+{
+	struct link_resource link_res;
+
+	dc->link_srv->get_cur_link_res(link, &link_res);
+	dc->link_srv->dp_set_drive_settings(link, &link_res, lt_settings);
+}
+
+void dc_link_set_preferred_link_settings(struct dc *dc,
+					 struct dc_link_settings *link_setting,
+					 struct dc_link *link)
+{
+	dc->link_srv->dp_set_preferred_link_settings(dc, link_setting, link);
+}
+
+void dc_link_set_preferred_training_settings(struct dc *dc,
+		struct dc_link_settings *link_setting,
+		struct dc_link_training_overrides *lt_overrides,
+		struct dc_link *link,
+		bool skip_immediate_retrain)
+{
+	dc->link_srv->dp_set_preferred_training_settings(dc, link_setting,
+			lt_overrides, link, skip_immediate_retrain);
+}
+
+bool dc_dp_trace_is_initialized(struct dc_link *link)
+{
+	return link->dc->link_srv->dp_trace_is_initialized(link);
+}
+
+void dc_dp_trace_set_is_logged_flag(struct dc_link *link,
+		bool in_detection,
+		bool is_logged)
+{
+	link->dc->link_srv->dp_trace_set_is_logged_flag(link, in_detection, is_logged);
+}
+
+bool dc_dp_trace_is_logged(struct dc_link *link, bool in_detection)
+{
+	return link->dc->link_srv->dp_trace_is_logged(link, in_detection);
+}
+
+unsigned long long dc_dp_trace_get_lt_end_timestamp(struct dc_link *link,
+		bool in_detection)
+{
+	return link->dc->link_srv->dp_trace_get_lt_end_timestamp(link, in_detection);
+}
+
+const struct dp_trace_lt_counts *dc_dp_trace_get_lt_counts(struct dc_link *link,
+		bool in_detection)
+{
+	return link->dc->link_srv->dp_trace_get_lt_counts(link, in_detection);
+}
+
+unsigned int dc_dp_trace_get_link_loss_count(struct dc_link *link)
+{
+	return link->dc->link_srv->dp_trace_get_link_loss_count(link);
+}
+
+struct dc_sink *dc_link_add_remote_sink(
+		struct dc_link *link,
+		const uint8_t *edid,
+		int len,
+		struct dc_sink_init_data *init_data)
+{
+	return link->dc->link_srv->add_remote_sink(link, edid, len, init_data);
+}
+
+void dc_link_remove_remote_sink(struct dc_link *link, struct dc_sink *sink)
+{
+	link->dc->link_srv->remove_remote_sink(link, sink);
+}
+
+int dc_link_aux_transfer_raw(struct ddc_service *ddc,
+		struct aux_payload *payload,
+		enum aux_return_code_type *operation_result)
+{
+	const struct dc *dc = ddc->link->dc;
+
+	return dc->link_srv->aux_transfer_raw(
+			ddc, payload, operation_result);
+}
+
+uint32_t dc_link_bw_kbps_from_raw_frl_link_rate_data(const struct dc *dc, uint8_t bw)
+{
+	return dc->link_srv->bw_kbps_from_raw_frl_link_rate_data(bw);
+}
+
+bool dc_link_decide_edp_link_settings(struct dc_link *link,
+		struct dc_link_settings *link_setting, uint32_t req_bw)
+{
+	return link->dc->link_srv->edp_decide_link_settings(link, link_setting, req_bw);
+}
+
+
+bool dc_link_dp_get_max_link_enc_cap(const struct dc_link *link,
+		struct dc_link_settings *max_link_enc_cap)
+{
+	return link->dc->link_srv->dp_get_max_link_enc_cap(link, max_link_enc_cap);
+}
+
+enum dp_link_encoding dc_link_dp_mst_decide_link_encoding_format(
+		const struct dc_link *link)
+{
+	return link->dc->link_srv->mst_decide_link_encoding_format(link);
+}
+
+const struct dc_link_settings *dc_link_get_link_cap(const struct dc_link *link)
+{
+	return link->dc->link_srv->dp_get_verified_link_cap(link);
+}
+
+enum dc_link_encoding_format dc_link_get_highest_encoding_format(const struct dc_link *link)
+{
+	if (dc_is_dp_signal(link->connector_signal)) {
+		if (link->dpcd_caps.dongle_type >= DISPLAY_DONGLE_DP_DVI_DONGLE &&
+				link->dpcd_caps.dongle_type <= DISPLAY_DONGLE_DP_HDMI_MISMATCHED_DONGLE)
+			return DC_LINK_ENCODING_HDMI_TMDS;
+		else if (link->dc->link_srv->dp_get_encoding_format(&link->verified_link_cap) ==
+				DP_8b_10b_ENCODING)
+			return DC_LINK_ENCODING_DP_8b_10b;
+		else if (link->dc->link_srv->dp_get_encoding_format(&link->verified_link_cap) ==
+				DP_128b_132b_ENCODING)
+			return DC_LINK_ENCODING_DP_128b_132b;
+	} else if (dc_is_hdmi_signal(link->connector_signal)) {
+	}
+
+	return DC_LINK_ENCODING_UNSPECIFIED;
+}
+
+bool dc_link_is_dp_sink_present(struct dc_link *link)
+{
+	return link->dc->link_srv->dp_is_sink_present(link);
+}
+
+bool dc_link_is_fec_supported(const struct dc_link *link)
+{
+	return link->dc->link_srv->dp_is_fec_supported(link);
+}
+
+void dc_link_overwrite_extended_receiver_cap(
+		struct dc_link *link)
+{
+	link->dc->link_srv->dp_overwrite_extended_receiver_cap(link);
+}
+
+bool dc_link_should_enable_fec(const struct dc_link *link)
+{
+	return link->dc->link_srv->dp_should_enable_fec(link);
+}
+
+void dc_link_dp_dpia_handle_usb4_bandwidth_allocation_for_link(
+		struct dc_link *link, int peak_bw)
+{
+	link->dc->link_srv->dpia_handle_usb4_bandwidth_allocation_for_link(link, peak_bw);
+}
+
+bool dc_link_check_link_loss_status(
+	struct dc_link *link,
+	union hpd_irq_data *hpd_irq_dpcd_data)
+{
+	return link->dc->link_srv->dp_parse_link_loss_status(link, hpd_irq_dpcd_data);
+}
+
+bool dc_link_dp_allow_hpd_rx_irq(const struct dc_link *link)
+{
+	return link->dc->link_srv->dp_should_allow_hpd_rx_irq(link);
+}
+
+void dc_link_dp_handle_link_loss(struct dc_link *link)
+{
+	link->dc->link_srv->dp_handle_link_loss(link);
+}
+
+enum dc_status dc_link_dp_read_hpd_rx_irq_data(
+	struct dc_link *link,
+	union hpd_irq_data *irq_data)
+{
+	return link->dc->link_srv->dp_read_hpd_rx_irq_data(link, irq_data);
+}
+
+bool dc_link_handle_hpd_rx_irq(struct dc_link *link,
+		union hpd_irq_data *out_hpd_irq_dpcd_data, bool *out_link_loss,
+		bool defer_handling, bool *has_left_work)
+{
+	return link->dc->link_srv->dp_handle_hpd_rx_irq(link, out_hpd_irq_dpcd_data,
+			out_link_loss, defer_handling, has_left_work);
+}
+
+void dc_link_dp_receiver_power_ctrl(struct dc_link *link, bool on)
+{
+	link->dc->link_srv->dpcd_write_rx_power_ctrl(link, on);
+}
+
+enum lttpr_mode dc_link_decide_lttpr_mode(struct dc_link *link,
+		struct dc_link_settings *link_setting)
+{
+	return link->dc->link_srv->dp_decide_lttpr_mode(link, link_setting);
+}
+
+void dc_link_edp_panel_backlight_power_on(struct dc_link *link, bool wait_for_hpd)
+{
+	link->dc->link_srv->edp_panel_backlight_power_on(link, wait_for_hpd);
+}
+
+int dc_link_get_backlight_level(const struct dc_link *link)
+{
+	return link->dc->link_srv->edp_get_backlight_level(link);
+}
+
+bool dc_link_get_backlight_level_nits(struct dc_link *link,
+		uint32_t *backlight_millinits_avg,
+		uint32_t *backlight_millinits_peak)
+{
+	return link->dc->link_srv->edp_get_backlight_level_nits(link,
+			backlight_millinits_avg,
+			backlight_millinits_peak);
+}
+
+bool dc_link_set_backlight_level(const struct dc_link *link,
+		struct set_backlight_level_params *backlight_level_params)
+{
+	return link->dc->link_srv->edp_set_backlight_level(link,
+			backlight_level_params);
+}
+
+bool dc_link_set_backlight_level_nits(struct dc_link *link,
+		bool isHDR,
+		uint32_t backlight_millinits,
+		uint32_t transition_time_in_ms)
+{
+	return link->dc->link_srv->edp_set_backlight_level_nits(link, isHDR,
+			backlight_millinits, transition_time_in_ms);
+}
+
+int dc_link_get_target_backlight_pwm(const struct dc_link *link)
+{
+	return link->dc->link_srv->edp_get_target_backlight_pwm(link);
+}
+
+bool dc_link_get_psr_state(const struct dc_link *link, enum dc_psr_state *state)
+{
+	return link->dc->link_srv->edp_get_psr_state(link, state);
+}
+
+bool dc_link_set_psr_allow_active(struct dc_link *link, const bool *allow_active,
+		bool wait, bool force_static, const unsigned int *power_opts)
+{
+	return link->dc->link_srv->edp_set_psr_allow_active(link, allow_active, wait,
+			force_static, power_opts);
+}
+
+bool dc_link_setup_psr(struct dc_link *link,
+		const struct dc_stream_state *stream, struct psr_config *psr_config,
+		struct psr_context *psr_context)
+{
+	return link->dc->link_srv->edp_setup_psr(link, stream, psr_config, psr_context);
+}
+
+bool dc_link_set_replay_allow_active(struct dc_link *link, const bool *allow_active,
+		bool wait, bool force_static, const unsigned int *power_opts)
+{
+	return link->dc->link_srv->edp_set_replay_allow_active(link, allow_active, wait,
+			force_static, power_opts);
+}
+
+bool dc_link_get_replay_state(const struct dc_link *link, uint64_t *state)
+{
+	return link->dc->link_srv->edp_get_replay_state(link, state);
+}
+
+bool dc_link_wait_for_t12(struct dc_link *link)
+{
+	return link->dc->link_srv->edp_wait_for_t12(link);
+}
+
+bool dc_link_get_hpd_state(struct dc_link *link)
+{
+	return link->dc->link_srv->get_hpd_state(link);
+}
+
+void dc_link_enable_hpd(const struct dc_link *link)
+{
+	link->dc->link_srv->enable_hpd(link);
+}
+
+void dc_link_disable_hpd(const struct dc_link *link)
+{
+	link->dc->link_srv->disable_hpd(link);
+}
+
+void dc_link_enable_hpd_filter(struct dc_link *link, bool enable)
+{
+	link->dc->link_srv->enable_hpd_filter(link, enable);
+}
+
+enum dc_status dc_link_validate_dp_tunneling_bandwidth(const struct dc *dc, const struct dc_state *new_ctx)
+{
+	return dc->link_srv->validate_dp_tunnel_bandwidth(dc, new_ctx);
+}
+
+void dc_link_get_alpm_support(struct dc_link *link,
+	bool *auxless_support,
+	bool *auxwake_support)
+{
+	link->dc->link_srv->edp_get_alpm_support(link, auxless_support, auxwake_support);
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c
new file mode 100644
index 000000000000..bc5dedf5f60c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c
@@ -0,0 +1,5620 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+
+#include "resource.h"
+#include "include/irq_service_interface.h"
+#include "link_encoder.h"
+#include "stream_encoder.h"
+#include "opp.h"
+#include "timing_generator.h"
+#include "transform.h"
+#include "dccg.h"
+#include "dchubbub.h"
+#include "dpp.h"
+#include "core_types.h"
+#include "set_mode_types.h"
+#include "virtual/virtual_stream_encoder.h"
+#include "dpcd_defs.h"
+#include "link_enc_cfg.h"
+#include "link_service.h"
+#include "clk_mgr.h"
+#include "dc_state_priv.h"
+#include "dc_stream_priv.h"
+
+#include "virtual/virtual_link_hwss.h"
+#include "link/hwss/link_hwss_dio.h"
+#include "link/hwss/link_hwss_dpia.h"
+#include "link/hwss/link_hwss_hpo_dp.h"
+#include "link/hwss/link_hwss_dio_fixed_vs_pe_retimer.h"
+#include "link/hwss/link_hwss_hpo_fixed_vs_pe_retimer_dp.h"
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+#include "dce60/dce60_resource.h"
+#endif
+#include "dce80/dce80_resource.h"
+#include "dce100/dce100_resource.h"
+#include "dce110/dce110_resource.h"
+#include "dce112/dce112_resource.h"
+#include "dce120/dce120_resource.h"
+#include "dcn10/dcn10_resource.h"
+#include "dcn20/dcn20_resource.h"
+#include "dcn21/dcn21_resource.h"
+#include "dcn201/dcn201_resource.h"
+#include "dcn30/dcn30_resource.h"
+#include "dcn301/dcn301_resource.h"
+#include "dcn302/dcn302_resource.h"
+#include "dcn303/dcn303_resource.h"
+#include "dcn31/dcn31_resource.h"
+#include "dcn314/dcn314_resource.h"
+#include "dcn315/dcn315_resource.h"
+#include "dcn316/dcn316_resource.h"
+#include "dcn32/dcn32_resource.h"
+#include "dcn321/dcn321_resource.h"
+#include "dcn35/dcn35_resource.h"
+#include "dcn351/dcn351_resource.h"
+#include "dcn36/dcn36_resource.h"
+#include "dcn401/dcn401_resource.h"
+#if defined(CONFIG_DRM_AMD_DC_FP)
+#include "dc_spl_translate.h"
+#endif
+
+#define VISUAL_CONFIRM_BASE_DEFAULT 3
+#define VISUAL_CONFIRM_BASE_MIN 1
+#define VISUAL_CONFIRM_BASE_MAX 10
+/* we choose 240 because it is a common denominator of common v addressable
+ * such as 2160, 1440, 1200, 960. So we take 1/240 portion of v addressable as
+ * the visual confirm dpp offset height. So visual confirm height can stay
+ * relatively the same independent from timing used.
+ */
+#define VISUAL_CONFIRM_DPP_OFFSET_DENO 240
+
+#define DC_LOGGER \
+	dc->ctx->logger
+#define DC_LOGGER_INIT(logger)
+#include "dml2/dml2_wrapper.h"
+
+#define UNABLE_TO_SPLIT -1
+
+enum dce_version resource_parse_asic_id(struct hw_asic_id asic_id)
+{
+	enum dce_version dc_version = DCE_VERSION_UNKNOWN;
+
+	switch (asic_id.chip_family) {
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+	case FAMILY_SI:
+		if (ASIC_REV_IS_TAHITI_P(asic_id.hw_internal_rev) ||
+		    ASIC_REV_IS_PITCAIRN_PM(asic_id.hw_internal_rev) ||
+		    ASIC_REV_IS_CAPEVERDE_M(asic_id.hw_internal_rev))
+			dc_version = DCE_VERSION_6_0;
+		else if (ASIC_REV_IS_OLAND_M(asic_id.hw_internal_rev))
+			dc_version = DCE_VERSION_6_4;
+		else
+			dc_version = DCE_VERSION_6_1;
+		break;
+#endif
+	case FAMILY_CI:
+		dc_version = DCE_VERSION_8_0;
+		break;
+	case FAMILY_KV:
+		if (ASIC_REV_IS_KALINDI(asic_id.hw_internal_rev) ||
+		    ASIC_REV_IS_BHAVANI(asic_id.hw_internal_rev) ||
+		    ASIC_REV_IS_GODAVARI(asic_id.hw_internal_rev))
+			dc_version = DCE_VERSION_8_3;
+		else
+			dc_version = DCE_VERSION_8_1;
+		break;
+	case FAMILY_CZ:
+		dc_version = DCE_VERSION_11_0;
+		break;
+
+	case FAMILY_VI:
+		if (ASIC_REV_IS_TONGA_P(asic_id.hw_internal_rev) ||
+				ASIC_REV_IS_FIJI_P(asic_id.hw_internal_rev)) {
+			dc_version = DCE_VERSION_10_0;
+			break;
+		}
+		if (ASIC_REV_IS_POLARIS10_P(asic_id.hw_internal_rev) ||
+				ASIC_REV_IS_POLARIS11_M(asic_id.hw_internal_rev) ||
+				ASIC_REV_IS_POLARIS12_V(asic_id.hw_internal_rev)) {
+			dc_version = DCE_VERSION_11_2;
+		}
+		if (ASIC_REV_IS_VEGAM(asic_id.hw_internal_rev))
+			dc_version = DCE_VERSION_11_22;
+		break;
+	case FAMILY_AI:
+		if (ASICREV_IS_VEGA20_P(asic_id.hw_internal_rev))
+			dc_version = DCE_VERSION_12_1;
+		else
+			dc_version = DCE_VERSION_12_0;
+		break;
+	case FAMILY_RV:
+		dc_version = DCN_VERSION_1_0;
+		if (ASICREV_IS_RAVEN2(asic_id.hw_internal_rev))
+			dc_version = DCN_VERSION_1_01;
+		if (ASICREV_IS_RENOIR(asic_id.hw_internal_rev))
+			dc_version = DCN_VERSION_2_1;
+		if (ASICREV_IS_GREEN_SARDINE(asic_id.hw_internal_rev))
+			dc_version = DCN_VERSION_2_1;
+		break;
+
+	case FAMILY_NV:
+		dc_version = DCN_VERSION_2_0;
+		if (asic_id.chip_id == DEVICE_ID_NV_13FE ||
+		    asic_id.chip_id == DEVICE_ID_NV_143F ||
+		    asic_id.chip_id == DEVICE_ID_NV_13F9 ||
+		    asic_id.chip_id == DEVICE_ID_NV_13FA ||
+		    asic_id.chip_id == DEVICE_ID_NV_13FB ||
+		    asic_id.chip_id == DEVICE_ID_NV_13FC ||
+		    asic_id.chip_id == DEVICE_ID_NV_13DB) {
+			dc_version = DCN_VERSION_2_01;
+			break;
+		}
+		if (ASICREV_IS_SIENNA_CICHLID_P(asic_id.hw_internal_rev))
+			dc_version = DCN_VERSION_3_0;
+		if (ASICREV_IS_DIMGREY_CAVEFISH_P(asic_id.hw_internal_rev))
+			dc_version = DCN_VERSION_3_02;
+		if (ASICREV_IS_BEIGE_GOBY_P(asic_id.hw_internal_rev))
+			dc_version = DCN_VERSION_3_03;
+		break;
+
+	case FAMILY_VGH:
+		dc_version = DCN_VERSION_3_01;
+		break;
+
+	case FAMILY_YELLOW_CARP:
+		if (ASICREV_IS_YELLOW_CARP(asic_id.hw_internal_rev))
+			dc_version = DCN_VERSION_3_1;
+		break;
+	case AMDGPU_FAMILY_GC_10_3_6:
+		if (ASICREV_IS_GC_10_3_6(asic_id.hw_internal_rev))
+			dc_version = DCN_VERSION_3_15;
+		break;
+	case AMDGPU_FAMILY_GC_10_3_7:
+		if (ASICREV_IS_GC_10_3_7(asic_id.hw_internal_rev))
+			dc_version = DCN_VERSION_3_16;
+		break;
+	case AMDGPU_FAMILY_GC_11_0_0:
+		dc_version = DCN_VERSION_3_2;
+		if (ASICREV_IS_GC_11_0_2(asic_id.hw_internal_rev))
+			dc_version = DCN_VERSION_3_21;
+		break;
+	case AMDGPU_FAMILY_GC_11_0_1:
+		dc_version = DCN_VERSION_3_14;
+		break;
+	case AMDGPU_FAMILY_GC_11_5_0:
+		dc_version = DCN_VERSION_3_5;
+		if (ASICREV_IS_GC_11_0_4(asic_id.hw_internal_rev))
+			dc_version = DCN_VERSION_3_51;
+		if (ASICREV_IS_DCN36(asic_id.hw_internal_rev))
+			dc_version = DCN_VERSION_3_6;
+		break;
+	case AMDGPU_FAMILY_GC_12_0_0:
+		if (ASICREV_IS_GC_12_0_1_A0(asic_id.hw_internal_rev) ||
+			ASICREV_IS_GC_12_0_0_A0(asic_id.hw_internal_rev))
+			dc_version = DCN_VERSION_4_01;
+		break;
+	default:
+		dc_version = DCE_VERSION_UNKNOWN;
+		break;
+	}
+	return dc_version;
+}
+
+struct resource_pool *dc_create_resource_pool(struct dc  *dc,
+					      const struct dc_init_data *init_data,
+					      enum dce_version dc_version)
+{
+	struct resource_pool *res_pool = NULL;
+
+	switch (dc_version) {
+#if defined(CONFIG_DRM_AMD_DC_SI)
+	case DCE_VERSION_6_0:
+		res_pool = dce60_create_resource_pool(
+			init_data->num_virtual_links, dc);
+		break;
+	case DCE_VERSION_6_1:
+		res_pool = dce61_create_resource_pool(
+			init_data->num_virtual_links, dc);
+		break;
+	case DCE_VERSION_6_4:
+		res_pool = dce64_create_resource_pool(
+			init_data->num_virtual_links, dc);
+		break;
+#endif
+	case DCE_VERSION_8_0:
+		res_pool = dce80_create_resource_pool(
+				init_data->num_virtual_links, dc);
+		break;
+	case DCE_VERSION_8_1:
+		res_pool = dce81_create_resource_pool(
+				init_data->num_virtual_links, dc);
+		break;
+	case DCE_VERSION_8_3:
+		res_pool = dce83_create_resource_pool(
+				init_data->num_virtual_links, dc);
+		break;
+	case DCE_VERSION_10_0:
+		res_pool = dce100_create_resource_pool(
+				init_data->num_virtual_links, dc);
+		break;
+	case DCE_VERSION_11_0:
+		res_pool = dce110_create_resource_pool(
+				init_data->num_virtual_links, dc,
+				init_data->asic_id);
+		break;
+	case DCE_VERSION_11_2:
+	case DCE_VERSION_11_22:
+		res_pool = dce112_create_resource_pool(
+				init_data->num_virtual_links, dc);
+		break;
+	case DCE_VERSION_12_0:
+	case DCE_VERSION_12_1:
+		res_pool = dce120_create_resource_pool(
+				init_data->num_virtual_links, dc);
+		break;
+
+#if defined(CONFIG_DRM_AMD_DC_FP)
+	case DCN_VERSION_1_0:
+	case DCN_VERSION_1_01:
+		res_pool = dcn10_create_resource_pool(init_data, dc);
+		break;
+	case DCN_VERSION_2_0:
+		res_pool = dcn20_create_resource_pool(init_data, dc);
+		break;
+	case DCN_VERSION_2_1:
+		res_pool = dcn21_create_resource_pool(init_data, dc);
+		break;
+	case DCN_VERSION_2_01:
+		res_pool = dcn201_create_resource_pool(init_data, dc);
+		break;
+	case DCN_VERSION_3_0:
+		res_pool = dcn30_create_resource_pool(init_data, dc);
+		break;
+	case DCN_VERSION_3_01:
+		res_pool = dcn301_create_resource_pool(init_data, dc);
+		break;
+	case DCN_VERSION_3_02:
+		res_pool = dcn302_create_resource_pool(init_data, dc);
+		break;
+	case DCN_VERSION_3_03:
+		res_pool = dcn303_create_resource_pool(init_data, dc);
+		break;
+	case DCN_VERSION_3_1:
+		res_pool = dcn31_create_resource_pool(init_data, dc);
+		break;
+	case DCN_VERSION_3_14:
+		res_pool = dcn314_create_resource_pool(init_data, dc);
+		break;
+	case DCN_VERSION_3_15:
+		res_pool = dcn315_create_resource_pool(init_data, dc);
+		break;
+	case DCN_VERSION_3_16:
+		res_pool = dcn316_create_resource_pool(init_data, dc);
+		break;
+	case DCN_VERSION_3_2:
+		res_pool = dcn32_create_resource_pool(init_data, dc);
+		break;
+	case DCN_VERSION_3_21:
+		res_pool = dcn321_create_resource_pool(init_data, dc);
+		break;
+	case DCN_VERSION_3_5:
+		res_pool = dcn35_create_resource_pool(init_data, dc);
+		break;
+	case DCN_VERSION_3_51:
+		res_pool = dcn351_create_resource_pool(init_data, dc);
+		break;
+	case DCN_VERSION_3_6:
+		res_pool = dcn36_create_resource_pool(init_data, dc);
+		break;
+	case DCN_VERSION_4_01:
+		res_pool = dcn401_create_resource_pool(init_data, dc);
+		break;
+#endif /* CONFIG_DRM_AMD_DC_FP */
+	default:
+		break;
+	}
+
+	if (res_pool != NULL) {
+		if (dc->ctx->dc_bios->fw_info_valid) {
+			res_pool->ref_clocks.xtalin_clock_inKhz =
+				dc->ctx->dc_bios->fw_info.pll_info.crystal_frequency;
+			/* initialize with firmware data first, no all
+			 * ASIC have DCCG SW component. FPGA or
+			 * simulation need initialization of
+			 * dccg_ref_clock_inKhz, dchub_ref_clock_inKhz
+			 * with xtalin_clock_inKhz
+			 */
+			res_pool->ref_clocks.dccg_ref_clock_inKhz =
+				res_pool->ref_clocks.xtalin_clock_inKhz;
+			res_pool->ref_clocks.dchub_ref_clock_inKhz =
+				res_pool->ref_clocks.xtalin_clock_inKhz;
+		} else
+			ASSERT_CRITICAL(false);
+	}
+
+	return res_pool;
+}
+
+void dc_destroy_resource_pool(struct dc *dc)
+{
+	if (dc) {
+		if (dc->res_pool)
+			dc->res_pool->funcs->destroy(&dc->res_pool);
+
+		kfree(dc->hwseq);
+	}
+}
+
+static void update_num_audio(
+	const struct resource_straps *straps,
+	unsigned int *num_audio,
+	struct audio_support *aud_support)
+{
+	aud_support->dp_audio = true;
+	aud_support->hdmi_audio_native = false;
+	aud_support->hdmi_audio_on_dongle = false;
+
+	if (straps->hdmi_disable == 0) {
+		if (straps->dc_pinstraps_audio & 0x2) {
+			aud_support->hdmi_audio_on_dongle = true;
+			aud_support->hdmi_audio_native = true;
+		}
+	}
+
+	switch (straps->audio_stream_number) {
+	case 0: /* multi streams supported */
+		break;
+	case 1: /* multi streams not supported */
+		*num_audio = 1;
+		break;
+	default:
+		DC_ERR("DC: unexpected audio fuse!\n");
+	}
+}
+
+bool resource_construct(
+	unsigned int num_virtual_links,
+	struct dc  *dc,
+	struct resource_pool *pool,
+	const struct resource_create_funcs *create_funcs)
+{
+	struct dc_context *ctx = dc->ctx;
+	const struct resource_caps *caps = pool->res_cap;
+	int i;
+	unsigned int num_audio = caps->num_audio;
+	struct resource_straps straps = {0};
+
+	if (create_funcs->read_dce_straps)
+		create_funcs->read_dce_straps(dc->ctx, &straps);
+
+	pool->audio_count = 0;
+	if (create_funcs->create_audio) {
+		/* find the total number of streams available via the
+		 * AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT
+		 * registers (one for each pin) starting from pin 1
+		 * up to the max number of audio pins.
+		 * We stop on the first pin where
+		 * PORT_CONNECTIVITY == 1 (as instructed by HW team).
+		 */
+		update_num_audio(&straps, &num_audio, &pool->audio_support);
+		for (i = 0; i < caps->num_audio; i++) {
+			struct audio *aud = create_funcs->create_audio(ctx, i);
+
+			if (aud == NULL) {
+				DC_ERR("DC: failed to create audio!\n");
+				return false;
+			}
+			if (!aud->funcs->endpoint_valid(aud)) {
+				aud->funcs->destroy(&aud);
+				break;
+			}
+			pool->audios[i] = aud;
+			pool->audio_count++;
+		}
+	}
+
+	pool->stream_enc_count = 0;
+	if (create_funcs->create_stream_encoder) {
+		for (i = 0; i < caps->num_stream_encoder; i++) {
+			pool->stream_enc[i] = create_funcs->create_stream_encoder(i, ctx);
+			if (pool->stream_enc[i] == NULL)
+				DC_ERR("DC: failed to create stream_encoder!\n");
+			pool->stream_enc_count++;
+		}
+	}
+
+	pool->hpo_dp_stream_enc_count = 0;
+	if (create_funcs->create_hpo_dp_stream_encoder) {
+		for (i = 0; i < caps->num_hpo_dp_stream_encoder; i++) {
+			pool->hpo_dp_stream_enc[i] = create_funcs->create_hpo_dp_stream_encoder(i+ENGINE_ID_HPO_DP_0, ctx);
+			if (pool->hpo_dp_stream_enc[i] == NULL)
+				DC_ERR("DC: failed to create HPO DP stream encoder!\n");
+			pool->hpo_dp_stream_enc_count++;
+
+		}
+	}
+
+	pool->hpo_dp_link_enc_count = 0;
+	if (create_funcs->create_hpo_dp_link_encoder) {
+		for (i = 0; i < caps->num_hpo_dp_link_encoder; i++) {
+			pool->hpo_dp_link_enc[i] = create_funcs->create_hpo_dp_link_encoder(i, ctx);
+			if (pool->hpo_dp_link_enc[i] == NULL)
+				DC_ERR("DC: failed to create HPO DP link encoder!\n");
+			pool->hpo_dp_link_enc_count++;
+		}
+	}
+
+	for (i = 0; i < caps->num_mpc_3dlut; i++) {
+		pool->mpc_lut[i] = dc_create_3dlut_func();
+		if (pool->mpc_lut[i] == NULL)
+			DC_ERR("DC: failed to create MPC 3dlut!\n");
+		pool->mpc_shaper[i] = dc_create_transfer_func();
+		if (pool->mpc_shaper[i] == NULL)
+			DC_ERR("DC: failed to create MPC shaper!\n");
+	}
+
+	dc->caps.dynamic_audio = false;
+	if (pool->audio_count < pool->stream_enc_count) {
+		dc->caps.dynamic_audio = true;
+	}
+	for (i = 0; i < num_virtual_links; i++) {
+		pool->stream_enc[pool->stream_enc_count] =
+			virtual_stream_encoder_create(
+					ctx, ctx->dc_bios);
+		if (pool->stream_enc[pool->stream_enc_count] == NULL) {
+			DC_ERR("DC: failed to create stream_encoder!\n");
+			return false;
+		}
+		pool->stream_enc_count++;
+	}
+
+	dc->hwseq = create_funcs->create_hwseq(ctx);
+
+	return true;
+}
+static int find_matching_clock_source(
+		const struct resource_pool *pool,
+		struct clock_source *clock_source)
+{
+
+	int i;
+
+	for (i = 0; i < pool->clk_src_count; i++) {
+		if (pool->clock_sources[i] == clock_source)
+			return i;
+	}
+	return -1;
+}
+
+void resource_unreference_clock_source(
+		struct resource_context *res_ctx,
+		const struct resource_pool *pool,
+		struct clock_source *clock_source)
+{
+	int i = find_matching_clock_source(pool, clock_source);
+
+	if (i > -1)
+		res_ctx->clock_source_ref_count[i]--;
+
+	if (pool->dp_clock_source == clock_source)
+		res_ctx->dp_clock_source_ref_count--;
+}
+
+void resource_reference_clock_source(
+		struct resource_context *res_ctx,
+		const struct resource_pool *pool,
+		struct clock_source *clock_source)
+{
+	int i = find_matching_clock_source(pool, clock_source);
+
+	if (i > -1)
+		res_ctx->clock_source_ref_count[i]++;
+
+	if (pool->dp_clock_source == clock_source)
+		res_ctx->dp_clock_source_ref_count++;
+}
+
+int resource_get_clock_source_reference(
+		struct resource_context *res_ctx,
+		const struct resource_pool *pool,
+		struct clock_source *clock_source)
+{
+	int i = find_matching_clock_source(pool, clock_source);
+
+	if (i > -1)
+		return res_ctx->clock_source_ref_count[i];
+
+	if (pool->dp_clock_source == clock_source)
+		return res_ctx->dp_clock_source_ref_count;
+
+	return -1;
+}
+
+bool resource_are_vblanks_synchronizable(
+	struct dc_stream_state *stream1,
+	struct dc_stream_state *stream2)
+{
+	uint32_t base60_refresh_rates[] = {10, 20, 5};
+	uint8_t i;
+	uint8_t rr_count = ARRAY_SIZE(base60_refresh_rates);
+	uint64_t frame_time_diff;
+
+	if (stream1->ctx->dc->config.vblank_alignment_dto_params &&
+		stream1->ctx->dc->config.vblank_alignment_max_frame_time_diff > 0 &&
+		dc_is_dp_signal(stream1->signal) &&
+		dc_is_dp_signal(stream2->signal) &&
+		false == stream1->has_non_synchronizable_pclk &&
+		false == stream2->has_non_synchronizable_pclk &&
+		stream1->timing.flags.VBLANK_SYNCHRONIZABLE &&
+		stream2->timing.flags.VBLANK_SYNCHRONIZABLE) {
+		/* disable refresh rates higher than 60Hz for now */
+		if (stream1->timing.pix_clk_100hz*100/stream1->timing.h_total/
+				stream1->timing.v_total > 60)
+			return false;
+		if (stream2->timing.pix_clk_100hz*100/stream2->timing.h_total/
+				stream2->timing.v_total > 60)
+			return false;
+		frame_time_diff = (uint64_t)10000 *
+			stream1->timing.h_total *
+			stream1->timing.v_total *
+			stream2->timing.pix_clk_100hz;
+		frame_time_diff = div_u64(frame_time_diff, stream1->timing.pix_clk_100hz);
+		frame_time_diff = div_u64(frame_time_diff, stream2->timing.h_total);
+		frame_time_diff = div_u64(frame_time_diff, stream2->timing.v_total);
+		for (i = 0; i < rr_count; i++) {
+			int64_t diff = (int64_t)div_u64(frame_time_diff * base60_refresh_rates[i], 10) - 10000;
+
+			if (diff < 0)
+				diff = -diff;
+			if (diff < stream1->ctx->dc->config.vblank_alignment_max_frame_time_diff)
+				return true;
+		}
+	}
+	return false;
+}
+
+bool resource_are_streams_timing_synchronizable(
+	struct dc_stream_state *stream1,
+	struct dc_stream_state *stream2)
+{
+	if (stream1->timing.h_total != stream2->timing.h_total)
+		return false;
+
+	if (stream1->timing.v_total != stream2->timing.v_total)
+		return false;
+
+	if (stream1->timing.h_addressable
+				!= stream2->timing.h_addressable)
+		return false;
+
+	if (stream1->timing.v_addressable
+				!= stream2->timing.v_addressable)
+		return false;
+
+	if (stream1->timing.v_front_porch
+				!= stream2->timing.v_front_porch)
+		return false;
+
+	if (stream1->timing.pix_clk_100hz
+				!= stream2->timing.pix_clk_100hz)
+		return false;
+
+	if (stream1->clamping.c_depth != stream2->clamping.c_depth)
+		return false;
+
+	if (stream1->phy_pix_clk != stream2->phy_pix_clk
+			&& (!dc_is_dp_signal(stream1->signal)
+			|| !dc_is_dp_signal(stream2->signal)))
+		return false;
+
+	if (stream1->view_format != stream2->view_format)
+		return false;
+
+	if (stream1->ignore_msa_timing_param || stream2->ignore_msa_timing_param)
+		return false;
+
+	return true;
+}
+static bool is_dp_and_hdmi_sharable(
+		struct dc_stream_state *stream1,
+		struct dc_stream_state *stream2)
+{
+	if (stream1->ctx->dc->caps.disable_dp_clk_share)
+		return false;
+
+	if (stream1->clamping.c_depth != COLOR_DEPTH_888 ||
+		stream2->clamping.c_depth != COLOR_DEPTH_888)
+		return false;
+
+	return true;
+
+}
+
+static bool is_sharable_clk_src(
+	const struct pipe_ctx *pipe_with_clk_src,
+	const struct pipe_ctx *pipe)
+{
+	if (pipe_with_clk_src->clock_source == NULL)
+		return false;
+
+	if (pipe_with_clk_src->stream->signal == SIGNAL_TYPE_VIRTUAL)
+		return false;
+
+	if (dc_is_dp_signal(pipe_with_clk_src->stream->signal) ||
+		(dc_is_dp_signal(pipe->stream->signal) &&
+		!is_dp_and_hdmi_sharable(pipe_with_clk_src->stream,
+				     pipe->stream)))
+		return false;
+
+	if (dc_is_hdmi_signal(pipe_with_clk_src->stream->signal)
+			&& dc_is_dual_link_signal(pipe->stream->signal))
+		return false;
+
+	if (dc_is_hdmi_signal(pipe->stream->signal)
+			&& dc_is_dual_link_signal(pipe_with_clk_src->stream->signal))
+		return false;
+
+	if (!resource_are_streams_timing_synchronizable(
+			pipe_with_clk_src->stream, pipe->stream))
+		return false;
+
+	return true;
+}
+
+struct clock_source *resource_find_used_clk_src_for_sharing(
+					struct resource_context *res_ctx,
+					struct pipe_ctx *pipe_ctx)
+{
+	int i;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		if (is_sharable_clk_src(&res_ctx->pipe_ctx[i], pipe_ctx))
+			return res_ctx->pipe_ctx[i].clock_source;
+	}
+
+	return NULL;
+}
+
+static enum pixel_format convert_pixel_format_to_dalsurface(
+		enum surface_pixel_format surface_pixel_format)
+{
+	enum pixel_format dal_pixel_format = PIXEL_FORMAT_UNKNOWN;
+
+	switch (surface_pixel_format) {
+	case SURFACE_PIXEL_FORMAT_GRPH_PALETA_256_COLORS:
+		dal_pixel_format = PIXEL_FORMAT_INDEX8;
+		break;
+	case SURFACE_PIXEL_FORMAT_GRPH_ARGB1555:
+		dal_pixel_format = PIXEL_FORMAT_RGB565;
+		break;
+	case SURFACE_PIXEL_FORMAT_GRPH_RGB565:
+		dal_pixel_format = PIXEL_FORMAT_RGB565;
+		break;
+	case SURFACE_PIXEL_FORMAT_GRPH_ARGB8888:
+		dal_pixel_format = PIXEL_FORMAT_ARGB8888;
+		break;
+	case SURFACE_PIXEL_FORMAT_GRPH_ABGR8888:
+		dal_pixel_format = PIXEL_FORMAT_ARGB8888;
+		break;
+	case SURFACE_PIXEL_FORMAT_GRPH_ARGB2101010:
+		dal_pixel_format = PIXEL_FORMAT_ARGB2101010;
+		break;
+	case SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010:
+		dal_pixel_format = PIXEL_FORMAT_ARGB2101010;
+		break;
+	case SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010_XR_BIAS:
+		dal_pixel_format = PIXEL_FORMAT_ARGB2101010_XRBIAS;
+		break;
+	case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F:
+	case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616F:
+		dal_pixel_format = PIXEL_FORMAT_FP16;
+		break;
+	case SURFACE_PIXEL_FORMAT_VIDEO_420_YCbCr:
+	case SURFACE_PIXEL_FORMAT_VIDEO_420_YCrCb:
+		dal_pixel_format = PIXEL_FORMAT_420BPP8;
+		break;
+	case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCbCr:
+	case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCrCb:
+		dal_pixel_format = PIXEL_FORMAT_420BPP10;
+		break;
+	case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616:
+	case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616:
+	default:
+		dal_pixel_format = PIXEL_FORMAT_UNKNOWN;
+		break;
+	}
+	return dal_pixel_format;
+}
+
+static inline void get_vp_scan_direction(
+	enum dc_rotation_angle rotation,
+	bool horizontal_mirror,
+	bool *orthogonal_rotation,
+	bool *flip_vert_scan_dir,
+	bool *flip_horz_scan_dir)
+{
+	*orthogonal_rotation = false;
+	*flip_vert_scan_dir = false;
+	*flip_horz_scan_dir = false;
+	if (rotation == ROTATION_ANGLE_180) {
+		*flip_vert_scan_dir = true;
+		*flip_horz_scan_dir = true;
+	} else if (rotation == ROTATION_ANGLE_90) {
+		*orthogonal_rotation = true;
+		*flip_horz_scan_dir = true;
+	} else if (rotation == ROTATION_ANGLE_270) {
+		*orthogonal_rotation = true;
+		*flip_vert_scan_dir = true;
+	}
+
+	if (horizontal_mirror)
+		*flip_horz_scan_dir = !*flip_horz_scan_dir;
+}
+
+static struct rect intersect_rec(const struct rect *r0, const struct rect *r1)
+{
+	struct rect rec;
+	int r0_x_end = r0->x + r0->width;
+	int r1_x_end = r1->x + r1->width;
+	int r0_y_end = r0->y + r0->height;
+	int r1_y_end = r1->y + r1->height;
+
+	rec.x = r0->x > r1->x ? r0->x : r1->x;
+	rec.width = r0_x_end > r1_x_end ? r1_x_end - rec.x : r0_x_end - rec.x;
+	rec.y = r0->y > r1->y ? r0->y : r1->y;
+	rec.height = r0_y_end > r1_y_end ? r1_y_end - rec.y : r0_y_end - rec.y;
+
+	/* in case that there is no intersection */
+	if (rec.width < 0 || rec.height < 0)
+		memset(&rec, 0, sizeof(rec));
+
+	return rec;
+}
+
+static struct rect shift_rec(const struct rect *rec_in, int x, int y)
+{
+	struct rect rec_out = *rec_in;
+
+	rec_out.x += x;
+	rec_out.y += y;
+
+	return rec_out;
+}
+
+static struct rect calculate_plane_rec_in_timing_active(
+		struct pipe_ctx *pipe_ctx,
+		const struct rect *rec_in)
+{
+	/*
+	 * The following diagram shows an example where we map a 1920x1200
+	 * desktop to a 2560x1440 timing with a plane rect in the middle
+	 * of the screen. To map a plane rect from Stream Source to Timing
+	 * Active space, we first multiply stream scaling ratios (i.e 2304/1920
+	 * horizontal and 1440/1200 vertical) to the plane's x and y, then
+	 * we add stream destination offsets (i.e 128 horizontal, 0 vertical).
+	 * This will give us a plane rect's position in Timing Active. However
+	 * we have to remove the fractional. The rule is that we find left/right
+	 * and top/bottom positions and round the value to the adjacent integer.
+	 *
+	 * Stream Source Space
+	 * ------------
+	 *        __________________________________________________
+	 *       |Stream Source (1920 x 1200) ^                     |
+	 *       |                            y                     |
+	 *       |         <------- w --------|>                    |
+	 *       |          __________________V                     |
+	 *       |<-- x -->|Plane//////////////| ^                  |
+	 *       |         |(pre scale)////////| |                  |
+	 *       |         |///////////////////| |                  |
+	 *       |         |///////////////////| h                  |
+	 *       |         |///////////////////| |                  |
+	 *       |         |///////////////////| |                  |
+	 *       |         |///////////////////| V                  |
+	 *       |                                                  |
+	 *       |                                                  |
+	 *       |__________________________________________________|
+	 *
+	 *
+	 * Timing Active Space
+	 * ---------------------------------
+	 *
+	 *       Timing Active (2560 x 1440)
+	 *        __________________________________________________
+	 *       |*****|  Stteam Destination (2304 x 1440)    |*****|
+	 *       |*****|                                      |*****|
+	 *       |<128>|                                      |*****|
+	 *       |*****|     __________________               |*****|
+	 *       |*****|    |Plane/////////////|              |*****|
+	 *       |*****|    |(post scale)//////|              |*****|
+	 *       |*****|    |//////////////////|              |*****|
+	 *       |*****|    |//////////////////|              |*****|
+	 *       |*****|    |//////////////////|              |*****|
+	 *       |*****|    |//////////////////|              |*****|
+	 *       |*****|                                      |*****|
+	 *       |*****|                                      |*****|
+	 *       |*****|                                      |*****|
+	 *       |*****|______________________________________|*****|
+	 *
+	 * So the resulting formulas are shown below:
+	 *
+	 * recout_x = 128 + round(plane_x * 2304 / 1920)
+	 * recout_w = 128 + round((plane_x + plane_w) * 2304 / 1920) - recout_x
+	 * recout_y = 0 + round(plane_y * 1440 / 1280)
+	 * recout_h = 0 + round((plane_y + plane_h) * 1440 / 1200) - recout_y
+	 *
+	 * NOTE: fixed point division is not error free. To reduce errors
+	 * introduced by fixed point division, we divide only after
+	 * multiplication is complete.
+	 */
+	const struct dc_stream_state *stream = pipe_ctx->stream;
+	struct rect rec_out = {0};
+	struct fixed31_32 temp;
+
+	temp = dc_fixpt_from_fraction(rec_in->x * (long long)stream->dst.width,
+			stream->src.width);
+	rec_out.x = stream->dst.x + dc_fixpt_round(temp);
+
+	temp = dc_fixpt_from_fraction(
+			(rec_in->x + rec_in->width) * (long long)stream->dst.width,
+			stream->src.width);
+	rec_out.width = stream->dst.x + dc_fixpt_round(temp) - rec_out.x;
+
+	temp = dc_fixpt_from_fraction(rec_in->y * (long long)stream->dst.height,
+			stream->src.height);
+	rec_out.y = stream->dst.y + dc_fixpt_round(temp);
+
+	temp = dc_fixpt_from_fraction(
+			(rec_in->y + rec_in->height) * (long long)stream->dst.height,
+			stream->src.height);
+	rec_out.height = stream->dst.y + dc_fixpt_round(temp) - rec_out.y;
+
+	return rec_out;
+}
+
+static struct rect calculate_mpc_slice_in_timing_active(
+		struct pipe_ctx *pipe_ctx,
+		struct rect *plane_clip_rec)
+{
+	const struct dc_stream_state *stream = pipe_ctx->stream;
+	int mpc_slice_count = resource_get_mpc_slice_count(pipe_ctx);
+	int mpc_slice_idx = resource_get_mpc_slice_index(pipe_ctx);
+	int epimo = mpc_slice_count - plane_clip_rec->width % mpc_slice_count - 1;
+	struct rect mpc_rec;
+
+	mpc_rec.width = plane_clip_rec->width / mpc_slice_count;
+	mpc_rec.x = plane_clip_rec->x + mpc_rec.width * mpc_slice_idx;
+	mpc_rec.height = plane_clip_rec->height;
+	mpc_rec.y = plane_clip_rec->y;
+	ASSERT(mpc_slice_count == 1 ||
+			stream->view_format != VIEW_3D_FORMAT_SIDE_BY_SIDE ||
+			mpc_rec.width % 2 == 0);
+
+	if (stream->view_format == VIEW_3D_FORMAT_SIDE_BY_SIDE)
+		mpc_rec.x -= (mpc_rec.width * mpc_slice_idx);
+
+	/* extra pixels in the division remainder need to go to pipes after
+	 * the extra pixel index minus one(epimo) defined here as:
+	 */
+	if (mpc_slice_idx > epimo) {
+		mpc_rec.x += mpc_slice_idx - epimo - 1;
+		mpc_rec.width += 1;
+	}
+
+	if (stream->view_format == VIEW_3D_FORMAT_TOP_AND_BOTTOM) {
+		ASSERT(mpc_rec.height % 2 == 0);
+		mpc_rec.height /= 2;
+	}
+	return mpc_rec;
+}
+
+static void calculate_adjust_recout_for_visual_confirm(struct pipe_ctx *pipe_ctx,
+	int *base_offset, int *dpp_offset)
+{
+	struct dc *dc = pipe_ctx->stream->ctx->dc;
+	*base_offset = 0;
+	*dpp_offset = 0;
+
+	if (dc->debug.visual_confirm == VISUAL_CONFIRM_DISABLE || !pipe_ctx->plane_res.dpp)
+		return;
+
+	*dpp_offset = pipe_ctx->stream->timing.v_addressable / VISUAL_CONFIRM_DPP_OFFSET_DENO;
+	*dpp_offset *= pipe_ctx->plane_res.dpp->inst;
+
+	if ((dc->debug.visual_confirm_rect_height >= VISUAL_CONFIRM_BASE_MIN) &&
+			dc->debug.visual_confirm_rect_height <= VISUAL_CONFIRM_BASE_MAX)
+		*base_offset = dc->debug.visual_confirm_rect_height;
+	else
+		*base_offset = VISUAL_CONFIRM_BASE_DEFAULT;
+}
+
+static void reverse_adjust_recout_for_visual_confirm(struct rect *recout,
+		struct pipe_ctx *pipe_ctx)
+{
+	int dpp_offset, base_offset;
+
+	calculate_adjust_recout_for_visual_confirm(pipe_ctx, &base_offset,
+		&dpp_offset);
+	recout->height += base_offset;
+	recout->height += dpp_offset;
+}
+
+static void adjust_recout_for_visual_confirm(struct rect *recout,
+		struct pipe_ctx *pipe_ctx)
+{
+	int dpp_offset, base_offset;
+
+	calculate_adjust_recout_for_visual_confirm(pipe_ctx, &base_offset,
+		&dpp_offset);
+	recout->height -= base_offset;
+	recout->height -= dpp_offset;
+}
+
+/*
+ * The function maps a plane clip from Stream Source Space to ODM Slice Space
+ * and calculates the rec of the overlapping area of MPC slice of the plane
+ * clip, ODM slice associated with the pipe context and stream destination rec.
+ */
+static void calculate_recout(struct pipe_ctx *pipe_ctx)
+{
+	/*
+	 * A plane clip represents the desired plane size and position in Stream
+	 * Source Space. Stream Source is the destination where all planes are
+	 * blended (i.e. positioned, scaled and overlaid). It is a canvas where
+	 * all planes associated with the current stream are drawn together.
+	 * After Stream Source is completed, we will further scale and
+	 * reposition the entire canvas of the stream source to Stream
+	 * Destination in Timing Active Space. This could be due to display
+	 * overscan adjustment where we will need to rescale and reposition all
+	 * the planes so they can fit into a TV with overscan or downscale
+	 * upscale features such as GPU scaling or VSR.
+	 *
+	 * This two step blending is a virtual procedure in software. In
+	 * hardware there is no such thing as Stream Source. all planes are
+	 * blended once in Timing Active Space. Software virtualizes a Stream
+	 * Source space to decouple the math complicity so scaling param
+	 * calculation focuses on one step at a time.
+	 *
+	 * In the following two diagrams, user applied 10% overscan adjustment
+	 * so the Stream Source needs to be scaled down a little before mapping
+	 * to Timing Active Space. As a result the Plane Clip is also scaled
+	 * down by the same ratio, Plane Clip position (i.e. x and y) with
+	 * respect to Stream Source is also scaled down. To map it in Timing
+	 * Active Space additional x and y offsets from Stream Destination are
+	 * added to Plane Clip as well.
+	 *
+	 * Stream Source Space
+	 * ------------
+	 *        __________________________________________________
+	 *       |Stream Source (3840 x 2160) ^                     |
+	 *       |                            y                     |
+	 *       |                            |                     |
+	 *       |          __________________V                     |
+	 *       |<-- x -->|Plane Clip/////////|                    |
+	 *       |         |(pre scale)////////|                    |
+	 *       |         |///////////////////|                    |
+	 *       |         |///////////////////|                    |
+	 *       |         |///////////////////|                    |
+	 *       |         |///////////////////|                    |
+	 *       |         |///////////////////|                    |
+	 *       |                                                  |
+	 *       |                                                  |
+	 *       |__________________________________________________|
+	 *
+	 *
+	 * Timing Active Space (3840 x 2160)
+	 * ---------------------------------
+	 *
+	 *       Timing Active
+	 *        __________________________________________________
+	 *       | y_____________________________________________   |
+	 *       |x |Stream Destination (3456 x 1944)            |  |
+	 *       |  |                                            |  |
+	 *       |  |        __________________                  |  |
+	 *       |  |       |Plane Clip////////|                 |  |
+	 *       |  |       |(post scale)//////|                 |  |
+	 *       |  |       |//////////////////|                 |  |
+	 *       |  |       |//////////////////|                 |  |
+	 *       |  |       |//////////////////|                 |  |
+	 *       |  |       |//////////////////|                 |  |
+	 *       |  |                                            |  |
+	 *       |  |                                            |  |
+	 *       |  |____________________________________________|  |
+	 *       |__________________________________________________|
+	 *
+	 *
+	 * In Timing Active Space a plane clip could be further sliced into
+	 * pieces called MPC slices. Each Pipe Context is responsible for
+	 * processing only one MPC slice so the plane processing workload can be
+	 * distributed to multiple DPP Pipes. MPC slices could be blended
+	 * together to a single ODM slice. Each ODM slice is responsible for
+	 * processing a portion of Timing Active divided horizontally so the
+	 * output pixel processing workload can be distributed to multiple OPP
+	 * pipes. All ODM slices are mapped together in ODM block so all MPC
+	 * slices belong to different ODM slices could be pieced together to
+	 * form a single image in Timing Active. MPC slices must belong to
+	 * single ODM slice. If an MPC slice goes across ODM slice boundary, it
+	 * needs to be divided into two MPC slices one for each ODM slice.
+	 *
+	 * In the following diagram the output pixel processing workload is
+	 * divided horizontally into two ODM slices one for each OPP blend tree.
+	 * OPP0 blend tree is responsible for processing left half of Timing
+	 * Active, while OPP2 blend tree is responsible for processing right
+	 * half.
+	 *
+	 * The plane has two MPC slices. However since the right MPC slice goes
+	 * across ODM boundary, two DPP pipes are needed one for each OPP blend
+	 * tree. (i.e. DPP1 for OPP0 blend tree and DPP2 for OPP2 blend tree).
+	 *
+	 * Assuming that we have a Pipe Context associated with OPP0 and DPP1
+	 * working on processing the plane in the diagram. We want to know the
+	 * width and height of the shaded rectangle and its relative position
+	 * with respect to the ODM slice0. This is called the recout of the pipe
+	 * context.
+	 *
+	 * Planes can be at arbitrary size and position and there could be an
+	 * arbitrary number of MPC and ODM slices. The algorithm needs to take
+	 * all scenarios into account.
+	 *
+	 * Timing Active Space (3840 x 2160)
+	 * ---------------------------------
+	 *
+	 *       Timing Active
+	 *        __________________________________________________
+	 *       |OPP0(ODM slice0)^        |OPP2(ODM slice1)        |
+	 *       |                y        |                        |
+	 *       |                |  <- w ->                        |
+	 *       |           _____V________|____                    |
+	 *       |          |DPP0 ^  |DPP1 |DPP2|                   |
+	 *       |<------ x |-----|->|/////|    |                   |
+	 *       |          |     |  |/////|    |                   |
+	 *       |          |     h  |/////|    |                   |
+	 *       |          |     |  |/////|    |                   |
+	 *       |          |_____V__|/////|____|                   |
+	 *       |                         |                        |
+	 *       |                         |                        |
+	 *       |                         |                        |
+	 *       |_________________________|________________________|
+	 *
+	 *
+	 */
+	struct rect plane_clip;
+	struct rect mpc_slice_of_plane_clip;
+	struct rect odm_slice_src;
+	struct rect overlapping_area;
+
+	plane_clip = calculate_plane_rec_in_timing_active(pipe_ctx,
+			&pipe_ctx->plane_state->clip_rect);
+	/* guard plane clip from drawing beyond stream dst here */
+	plane_clip = intersect_rec(&plane_clip,
+				&pipe_ctx->stream->dst);
+	mpc_slice_of_plane_clip = calculate_mpc_slice_in_timing_active(
+			pipe_ctx, &plane_clip);
+	odm_slice_src = resource_get_odm_slice_src_rect(pipe_ctx);
+	overlapping_area = intersect_rec(&mpc_slice_of_plane_clip, &odm_slice_src);
+	if (overlapping_area.height > 0 &&
+			overlapping_area.width > 0) {
+		/* shift the overlapping area so it is with respect to current
+		 * ODM slice source's position
+		 */
+		pipe_ctx->plane_res.scl_data.recout = shift_rec(
+				&overlapping_area,
+				-odm_slice_src.x, -odm_slice_src.y);
+		adjust_recout_for_visual_confirm(
+				&pipe_ctx->plane_res.scl_data.recout,
+				pipe_ctx);
+	} else {
+		/* if there is no overlap, zero recout */
+		memset(&pipe_ctx->plane_res.scl_data.recout, 0,
+				sizeof(struct rect));
+	}
+
+}
+
+static void calculate_scaling_ratios(struct pipe_ctx *pipe_ctx)
+{
+	const struct dc_plane_state *plane_state = pipe_ctx->plane_state;
+	const struct dc_stream_state *stream = pipe_ctx->stream;
+	struct rect surf_src = plane_state->src_rect;
+	const int in_w = stream->src.width;
+	const int in_h = stream->src.height;
+	const int out_w = stream->dst.width;
+	const int out_h = stream->dst.height;
+
+	/*Swap surf_src height and width since scaling ratios are in recout rotation*/
+	if (pipe_ctx->plane_state->rotation == ROTATION_ANGLE_90 ||
+			pipe_ctx->plane_state->rotation == ROTATION_ANGLE_270)
+		swap(surf_src.height, surf_src.width);
+
+	pipe_ctx->plane_res.scl_data.ratios.horz = dc_fixpt_from_fraction(
+					surf_src.width,
+					plane_state->dst_rect.width);
+	pipe_ctx->plane_res.scl_data.ratios.vert = dc_fixpt_from_fraction(
+					surf_src.height,
+					plane_state->dst_rect.height);
+
+	if (stream->view_format == VIEW_3D_FORMAT_SIDE_BY_SIDE)
+		pipe_ctx->plane_res.scl_data.ratios.horz.value *= 2;
+	else if (stream->view_format == VIEW_3D_FORMAT_TOP_AND_BOTTOM)
+		pipe_ctx->plane_res.scl_data.ratios.vert.value *= 2;
+
+	pipe_ctx->plane_res.scl_data.ratios.vert.value = div64_s64(
+		pipe_ctx->plane_res.scl_data.ratios.vert.value * in_h, out_h);
+	pipe_ctx->plane_res.scl_data.ratios.horz.value = div64_s64(
+		pipe_ctx->plane_res.scl_data.ratios.horz.value * in_w, out_w);
+
+	pipe_ctx->plane_res.scl_data.ratios.horz_c = pipe_ctx->plane_res.scl_data.ratios.horz;
+	pipe_ctx->plane_res.scl_data.ratios.vert_c = pipe_ctx->plane_res.scl_data.ratios.vert;
+
+	if (pipe_ctx->plane_res.scl_data.format == PIXEL_FORMAT_420BPP8
+			|| pipe_ctx->plane_res.scl_data.format == PIXEL_FORMAT_420BPP10) {
+		pipe_ctx->plane_res.scl_data.ratios.horz_c.value /= 2;
+		pipe_ctx->plane_res.scl_data.ratios.vert_c.value /= 2;
+	}
+	pipe_ctx->plane_res.scl_data.ratios.horz = dc_fixpt_truncate(
+			pipe_ctx->plane_res.scl_data.ratios.horz, 19);
+	pipe_ctx->plane_res.scl_data.ratios.vert = dc_fixpt_truncate(
+			pipe_ctx->plane_res.scl_data.ratios.vert, 19);
+	pipe_ctx->plane_res.scl_data.ratios.horz_c = dc_fixpt_truncate(
+			pipe_ctx->plane_res.scl_data.ratios.horz_c, 19);
+	pipe_ctx->plane_res.scl_data.ratios.vert_c = dc_fixpt_truncate(
+			pipe_ctx->plane_res.scl_data.ratios.vert_c, 19);
+}
+
+
+/*
+ * We completely calculate vp offset, size and inits here based entirely on scaling
+ * ratios and recout for pixel perfect pipe combine.
+ */
+static void calculate_init_and_vp(
+		bool flip_scan_dir,
+		int recout_offset_within_recout_full,
+		int recout_size,
+		int src_size,
+		int taps,
+		struct fixed31_32 ratio,
+		struct fixed31_32 *init,
+		int *vp_offset,
+		int *vp_size)
+{
+	struct fixed31_32 temp;
+	int int_part;
+
+	/*
+	 * First of the taps starts sampling pixel number <init_int_part> corresponding to recout
+	 * pixel 1. Next recout pixel samples int part of <init + scaling ratio> and so on.
+	 * All following calculations are based on this logic.
+	 *
+	 * Init calculated according to formula:
+	 * 	init = (scaling_ratio + number_of_taps + 1) / 2
+	 * 	init_bot = init + scaling_ratio
+	 * 	to get pixel perfect combine add the fraction from calculating vp offset
+	 */
+	temp = dc_fixpt_mul_int(ratio, recout_offset_within_recout_full);
+	*vp_offset = dc_fixpt_floor(temp);
+	temp.value &= 0xffffffff;
+	*init = dc_fixpt_truncate(dc_fixpt_add(dc_fixpt_div_int(
+			dc_fixpt_add_int(ratio, taps + 1), 2), temp), 19);
+	/*
+	 * If viewport has non 0 offset and there are more taps than covered by init then
+	 * we should decrease the offset and increase init so we are never sampling
+	 * outside of viewport.
+	 */
+	int_part = dc_fixpt_floor(*init);
+	if (int_part < taps) {
+		int_part = taps - int_part;
+		if (int_part > *vp_offset)
+			int_part = *vp_offset;
+		*vp_offset -= int_part;
+		*init = dc_fixpt_add_int(*init, int_part);
+	}
+	/*
+	 * If taps are sampling outside of viewport at end of recout and there are more pixels
+	 * available in the surface we should increase the viewport size, regardless set vp to
+	 * only what is used.
+	 */
+	temp = dc_fixpt_add(*init, dc_fixpt_mul_int(ratio, recout_size - 1));
+	*vp_size = dc_fixpt_floor(temp);
+	if (*vp_size + *vp_offset > src_size)
+		*vp_size = src_size - *vp_offset;
+
+	/* We did all the math assuming we are scanning same direction as display does,
+	 * however mirror/rotation changes how vp scans vs how it is offset. If scan direction
+	 * is flipped we simply need to calculate offset from the other side of plane.
+	 * Note that outside of viewport all scaling hardware works in recout space.
+	 */
+	if (flip_scan_dir)
+		*vp_offset = src_size - *vp_offset - *vp_size;
+}
+
+static void calculate_inits_and_viewports(struct pipe_ctx *pipe_ctx)
+{
+	const struct dc_plane_state *plane_state = pipe_ctx->plane_state;
+	struct scaler_data *data = &pipe_ctx->plane_res.scl_data;
+	struct rect src = plane_state->src_rect;
+	struct rect recout_dst_in_active_timing;
+	struct rect recout_clip_in_active_timing;
+	struct rect recout_clip_in_recout_dst;
+	struct rect overlap_in_active_timing;
+	struct rect odm_slice_src = resource_get_odm_slice_src_rect(pipe_ctx);
+	int vpc_div = (data->format == PIXEL_FORMAT_420BPP8
+				|| data->format == PIXEL_FORMAT_420BPP10) ? 2 : 1;
+	bool orthogonal_rotation, flip_vert_scan_dir, flip_horz_scan_dir;
+
+	recout_clip_in_active_timing = shift_rec(
+			&data->recout, odm_slice_src.x, odm_slice_src.y);
+	recout_dst_in_active_timing = calculate_plane_rec_in_timing_active(
+			pipe_ctx, &plane_state->dst_rect);
+	overlap_in_active_timing = intersect_rec(&recout_clip_in_active_timing,
+			&recout_dst_in_active_timing);
+	if (overlap_in_active_timing.width > 0 &&
+			overlap_in_active_timing.height > 0)
+		recout_clip_in_recout_dst = shift_rec(&overlap_in_active_timing,
+				-recout_dst_in_active_timing.x,
+				-recout_dst_in_active_timing.y);
+	else
+		memset(&recout_clip_in_recout_dst, 0, sizeof(struct rect));
+
+	/*
+	 * Work in recout rotation since that requires less transformations
+	 */
+	get_vp_scan_direction(
+			plane_state->rotation,
+			plane_state->horizontal_mirror,
+			&orthogonal_rotation,
+			&flip_vert_scan_dir,
+			&flip_horz_scan_dir);
+
+	if (orthogonal_rotation) {
+		swap(src.width, src.height);
+		swap(flip_vert_scan_dir, flip_horz_scan_dir);
+	}
+
+	calculate_init_and_vp(
+			flip_horz_scan_dir,
+			recout_clip_in_recout_dst.x,
+			data->recout.width,
+			src.width,
+			data->taps.h_taps,
+			data->ratios.horz,
+			&data->inits.h,
+			&data->viewport.x,
+			&data->viewport.width);
+	calculate_init_and_vp(
+			flip_horz_scan_dir,
+			recout_clip_in_recout_dst.x,
+			data->recout.width,
+			src.width / vpc_div,
+			data->taps.h_taps_c,
+			data->ratios.horz_c,
+			&data->inits.h_c,
+			&data->viewport_c.x,
+			&data->viewport_c.width);
+	calculate_init_and_vp(
+			flip_vert_scan_dir,
+			recout_clip_in_recout_dst.y,
+			data->recout.height,
+			src.height,
+			data->taps.v_taps,
+			data->ratios.vert,
+			&data->inits.v,
+			&data->viewport.y,
+			&data->viewport.height);
+	calculate_init_and_vp(
+			flip_vert_scan_dir,
+			recout_clip_in_recout_dst.y,
+			data->recout.height,
+			src.height / vpc_div,
+			data->taps.v_taps_c,
+			data->ratios.vert_c,
+			&data->inits.v_c,
+			&data->viewport_c.y,
+			&data->viewport_c.height);
+	if (orthogonal_rotation) {
+		swap(data->viewport.x, data->viewport.y);
+		swap(data->viewport.width, data->viewport.height);
+		swap(data->viewport_c.x, data->viewport_c.y);
+		swap(data->viewport_c.width, data->viewport_c.height);
+	}
+	data->viewport.x += src.x;
+	data->viewport.y += src.y;
+	ASSERT(src.x % vpc_div == 0 && src.y % vpc_div == 0);
+	data->viewport_c.x += src.x / vpc_div;
+	data->viewport_c.y += src.y / vpc_div;
+}
+
+static enum controller_dp_test_pattern convert_dp_to_controller_test_pattern(
+				enum dp_test_pattern test_pattern)
+{
+	enum controller_dp_test_pattern controller_test_pattern;
+
+	switch (test_pattern) {
+	case DP_TEST_PATTERN_COLOR_SQUARES:
+		controller_test_pattern =
+				CONTROLLER_DP_TEST_PATTERN_COLORSQUARES;
+	break;
+	case DP_TEST_PATTERN_COLOR_SQUARES_CEA:
+		controller_test_pattern =
+				CONTROLLER_DP_TEST_PATTERN_COLORSQUARES_CEA;
+	break;
+	case DP_TEST_PATTERN_VERTICAL_BARS:
+		controller_test_pattern =
+				CONTROLLER_DP_TEST_PATTERN_VERTICALBARS;
+	break;
+	case DP_TEST_PATTERN_HORIZONTAL_BARS:
+		controller_test_pattern =
+				CONTROLLER_DP_TEST_PATTERN_HORIZONTALBARS;
+	break;
+	case DP_TEST_PATTERN_COLOR_RAMP:
+		controller_test_pattern =
+				CONTROLLER_DP_TEST_PATTERN_COLORRAMP;
+	break;
+	default:
+		controller_test_pattern =
+				CONTROLLER_DP_TEST_PATTERN_VIDEOMODE;
+	break;
+	}
+
+	return controller_test_pattern;
+}
+
+static enum controller_dp_color_space convert_dp_to_controller_color_space(
+		enum dp_test_pattern_color_space color_space)
+{
+	enum controller_dp_color_space controller_color_space;
+
+	switch (color_space) {
+	case DP_TEST_PATTERN_COLOR_SPACE_RGB:
+		controller_color_space = CONTROLLER_DP_COLOR_SPACE_RGB;
+		break;
+	case DP_TEST_PATTERN_COLOR_SPACE_YCBCR601:
+		controller_color_space = CONTROLLER_DP_COLOR_SPACE_YCBCR601;
+		break;
+	case DP_TEST_PATTERN_COLOR_SPACE_YCBCR709:
+		controller_color_space = CONTROLLER_DP_COLOR_SPACE_YCBCR709;
+		break;
+	case DP_TEST_PATTERN_COLOR_SPACE_UNDEFINED:
+	default:
+		controller_color_space = CONTROLLER_DP_COLOR_SPACE_UDEFINED;
+		break;
+	}
+
+	return controller_color_space;
+}
+
+void resource_build_test_pattern_params(struct resource_context *res_ctx,
+				struct pipe_ctx *otg_master)
+{
+	struct pipe_ctx *opp_heads[MAX_PIPES];
+	struct test_pattern_params *params;
+	int odm_cnt;
+	enum controller_dp_test_pattern controller_test_pattern;
+	enum controller_dp_color_space controller_color_space;
+	enum dc_color_depth color_depth = otg_master->stream->timing.display_color_depth;
+	struct rect odm_slice_src;
+	int i;
+
+	controller_test_pattern = convert_dp_to_controller_test_pattern(
+			otg_master->stream->test_pattern.type);
+	controller_color_space = convert_dp_to_controller_color_space(
+			otg_master->stream->test_pattern.color_space);
+
+	if (controller_test_pattern == CONTROLLER_DP_TEST_PATTERN_VIDEOMODE)
+		return;
+
+	odm_cnt = resource_get_opp_heads_for_otg_master(otg_master, res_ctx, opp_heads);
+
+	for (i = 0; i < odm_cnt; i++) {
+		odm_slice_src = resource_get_odm_slice_src_rect(opp_heads[i]);
+		params = &opp_heads[i]->stream_res.test_pattern_params;
+		params->test_pattern = controller_test_pattern;
+		params->color_space = controller_color_space;
+		params->color_depth = color_depth;
+		params->height = odm_slice_src.height;
+		params->offset = odm_slice_src.x;
+		params->width = odm_slice_src.width;
+	}
+}
+
+bool resource_build_scaling_params(struct pipe_ctx *pipe_ctx)
+{
+	const struct dc_plane_state *plane_state = pipe_ctx->plane_state;
+	struct dc_crtc_timing *timing = &pipe_ctx->stream->timing;
+	const struct rect odm_slice_src = resource_get_odm_slice_src_rect(pipe_ctx);
+	struct scaling_taps temp = {0};
+	bool res = false;
+
+	DC_LOGGER_INIT(pipe_ctx->stream->ctx->logger);
+
+	/* Invalid input */
+	if (!plane_state ||
+			!plane_state->dst_rect.width ||
+			!plane_state->dst_rect.height ||
+			!plane_state->src_rect.width ||
+			!plane_state->src_rect.height) {
+		ASSERT(0);
+		return false;
+	}
+
+	/* Timing borders are part of vactive that we are also supposed to skip in addition
+	 * to any stream dst offset. Since dm logic assumes dst is in addressable
+	 * space we need to add the left and top borders to dst offsets temporarily.
+	 * TODO: fix in DM, stream dst is supposed to be in vactive
+	 */
+	pipe_ctx->stream->dst.x += timing->h_border_left;
+	pipe_ctx->stream->dst.y += timing->v_border_top;
+
+	/* Calculate H and V active size */
+	pipe_ctx->plane_res.scl_data.h_active = odm_slice_src.width;
+	pipe_ctx->plane_res.scl_data.v_active = odm_slice_src.height;
+	pipe_ctx->plane_res.scl_data.format = convert_pixel_format_to_dalsurface(
+			pipe_ctx->plane_state->format);
+
+#if defined(CONFIG_DRM_AMD_DC_FP)
+	if ((pipe_ctx->stream->ctx->dc->config.use_spl)	&& (!pipe_ctx->stream->ctx->dc->debug.disable_spl)) {
+		struct spl_in *spl_in = &pipe_ctx->plane_res.spl_in;
+		struct spl_out *spl_out = &pipe_ctx->plane_res.spl_out;
+
+		if (plane_state->ctx->dce_version > DCE_VERSION_MAX)
+			pipe_ctx->plane_res.scl_data.lb_params.depth = LB_PIXEL_DEPTH_36BPP;
+		else
+			pipe_ctx->plane_res.scl_data.lb_params.depth = LB_PIXEL_DEPTH_30BPP;
+
+		pipe_ctx->plane_res.scl_data.lb_params.alpha_en = plane_state->per_pixel_alpha;
+
+		// Convert pipe_ctx to respective input params for SPL
+		translate_SPL_in_params_from_pipe_ctx(pipe_ctx, spl_in);
+		/* Pass visual confirm debug information */
+		calculate_adjust_recout_for_visual_confirm(pipe_ctx,
+			&spl_in->debug.visual_confirm_base_offset,
+			&spl_in->debug.visual_confirm_dpp_offset);
+		// Set SPL output parameters to dscl_prog_data to be used for hw registers
+		spl_out->dscl_prog_data = resource_get_dscl_prog_data(pipe_ctx);
+		// Calculate scaler parameters from SPL
+		res = spl_calculate_scaler_params(spl_in, spl_out);
+		// Convert respective out params from SPL to scaler data
+		translate_SPL_out_params_to_pipe_ctx(pipe_ctx, spl_out);
+
+		/* Ignore scaler failure if pipe context plane is phantom plane */
+		if (!res && plane_state->is_phantom)
+			res = true;
+	} else {
+#endif
+	/* depends on h_active */
+	calculate_recout(pipe_ctx);
+	/* depends on pixel format */
+	calculate_scaling_ratios(pipe_ctx);
+
+	/*
+	 * LB calculations depend on vp size, h/v_active and scaling ratios
+	 * Setting line buffer pixel depth to 24bpp yields banding
+	 * on certain displays, such as the Sharp 4k. 36bpp is needed
+	 * to support SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616 and
+	 * SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616 with actual > 10 bpc
+	 * precision on DCN display engines, but apparently not for DCE, as
+	 * far as testing on DCE-11.2 and DCE-8 showed. Various DCE parts have
+	 * problems: Carrizo with DCE_VERSION_11_0 does not like 36 bpp lb depth,
+	 * neither do DCE-8 at 4k resolution, or DCE-11.2 (broken identify pixel
+	 * passthrough). Therefore only use 36 bpp on DCN where it is actually needed.
+	 */
+	if (plane_state->ctx->dce_version > DCE_VERSION_MAX)
+		pipe_ctx->plane_res.scl_data.lb_params.depth = LB_PIXEL_DEPTH_36BPP;
+	else
+		pipe_ctx->plane_res.scl_data.lb_params.depth = LB_PIXEL_DEPTH_30BPP;
+
+	pipe_ctx->plane_res.scl_data.lb_params.alpha_en = plane_state->per_pixel_alpha;
+
+	// get TAP value with 100x100 dummy data for max scaling qualify, override
+	// if a new scaling quality required
+	pipe_ctx->plane_res.scl_data.viewport.width = 100;
+	pipe_ctx->plane_res.scl_data.viewport.height = 100;
+	pipe_ctx->plane_res.scl_data.viewport_c.width = 100;
+	pipe_ctx->plane_res.scl_data.viewport_c.height = 100;
+	if (pipe_ctx->plane_res.xfm != NULL)
+		res = pipe_ctx->plane_res.xfm->funcs->transform_get_optimal_number_of_taps(
+				pipe_ctx->plane_res.xfm, &pipe_ctx->plane_res.scl_data, &plane_state->scaling_quality);
+
+	if (pipe_ctx->plane_res.dpp != NULL)
+		res = pipe_ctx->plane_res.dpp->funcs->dpp_get_optimal_number_of_taps(
+				pipe_ctx->plane_res.dpp, &pipe_ctx->plane_res.scl_data, &plane_state->scaling_quality);
+
+	temp = pipe_ctx->plane_res.scl_data.taps;
+
+	calculate_inits_and_viewports(pipe_ctx);
+
+	if (pipe_ctx->plane_res.xfm != NULL)
+		res = pipe_ctx->plane_res.xfm->funcs->transform_get_optimal_number_of_taps(
+				pipe_ctx->plane_res.xfm, &pipe_ctx->plane_res.scl_data, &plane_state->scaling_quality);
+
+	if (pipe_ctx->plane_res.dpp != NULL)
+		res = pipe_ctx->plane_res.dpp->funcs->dpp_get_optimal_number_of_taps(
+				pipe_ctx->plane_res.dpp, &pipe_ctx->plane_res.scl_data, &plane_state->scaling_quality);
+
+
+	if (!res) {
+		/* Try 24 bpp linebuffer */
+		pipe_ctx->plane_res.scl_data.lb_params.depth = LB_PIXEL_DEPTH_24BPP;
+
+		if (pipe_ctx->plane_res.xfm != NULL)
+			res = pipe_ctx->plane_res.xfm->funcs->transform_get_optimal_number_of_taps(
+					pipe_ctx->plane_res.xfm,
+					&pipe_ctx->plane_res.scl_data,
+					&plane_state->scaling_quality);
+
+		if (pipe_ctx->plane_res.dpp != NULL)
+			res = pipe_ctx->plane_res.dpp->funcs->dpp_get_optimal_number_of_taps(
+					pipe_ctx->plane_res.dpp,
+					&pipe_ctx->plane_res.scl_data,
+					&plane_state->scaling_quality);
+	}
+
+	/* Ignore scaler failure if pipe context plane is phantom plane */
+	if (!res && plane_state->is_phantom)
+		res = true;
+
+	if (res && (pipe_ctx->plane_res.scl_data.taps.v_taps != temp.v_taps ||
+		pipe_ctx->plane_res.scl_data.taps.h_taps != temp.h_taps ||
+		pipe_ctx->plane_res.scl_data.taps.v_taps_c != temp.v_taps_c ||
+		pipe_ctx->plane_res.scl_data.taps.h_taps_c != temp.h_taps_c))
+		calculate_inits_and_viewports(pipe_ctx);
+
+	/*
+	 * Handle side by side and top bottom 3d recout offsets after vp calculation
+	 * since 3d is special and needs to calculate vp as if there is no recout offset
+	 * This may break with rotation, good thing we aren't mixing hw rotation and 3d
+	 */
+	if (pipe_ctx->top_pipe && pipe_ctx->top_pipe->plane_state == plane_state) {
+		ASSERT(plane_state->rotation == ROTATION_ANGLE_0 ||
+			(pipe_ctx->stream->view_format != VIEW_3D_FORMAT_TOP_AND_BOTTOM &&
+				pipe_ctx->stream->view_format != VIEW_3D_FORMAT_SIDE_BY_SIDE));
+		if (pipe_ctx->stream->view_format == VIEW_3D_FORMAT_TOP_AND_BOTTOM)
+			pipe_ctx->plane_res.scl_data.recout.y += pipe_ctx->plane_res.scl_data.recout.height;
+		else if (pipe_ctx->stream->view_format == VIEW_3D_FORMAT_SIDE_BY_SIDE)
+			pipe_ctx->plane_res.scl_data.recout.x += pipe_ctx->plane_res.scl_data.recout.width;
+	}
+
+	/* Clamp minimum viewport size */
+	if (pipe_ctx->plane_res.scl_data.viewport.height < MIN_VIEWPORT_SIZE)
+		pipe_ctx->plane_res.scl_data.viewport.height = MIN_VIEWPORT_SIZE;
+	if (pipe_ctx->plane_res.scl_data.viewport.width < MIN_VIEWPORT_SIZE)
+		pipe_ctx->plane_res.scl_data.viewport.width = MIN_VIEWPORT_SIZE;
+#ifdef CONFIG_DRM_AMD_DC_FP
+	}
+#endif
+	DC_LOG_SCALER("%s pipe %d:\nViewport: height:%d width:%d x:%d y:%d  Recout: height:%d width:%d x:%d y:%d  HACTIVE:%d VACTIVE:%d\n"
+			"src_rect: height:%d width:%d x:%d y:%d  dst_rect: height:%d width:%d x:%d y:%d  clip_rect: height:%d width:%d x:%d y:%d\n",
+			__func__,
+			pipe_ctx->pipe_idx,
+			pipe_ctx->plane_res.scl_data.viewport.height,
+			pipe_ctx->plane_res.scl_data.viewport.width,
+			pipe_ctx->plane_res.scl_data.viewport.x,
+			pipe_ctx->plane_res.scl_data.viewport.y,
+			pipe_ctx->plane_res.scl_data.recout.height,
+			pipe_ctx->plane_res.scl_data.recout.width,
+			pipe_ctx->plane_res.scl_data.recout.x,
+			pipe_ctx->plane_res.scl_data.recout.y,
+			pipe_ctx->plane_res.scl_data.h_active,
+			pipe_ctx->plane_res.scl_data.v_active,
+			plane_state->src_rect.height,
+			plane_state->src_rect.width,
+			plane_state->src_rect.x,
+			plane_state->src_rect.y,
+			plane_state->dst_rect.height,
+			plane_state->dst_rect.width,
+			plane_state->dst_rect.x,
+			plane_state->dst_rect.y,
+			plane_state->clip_rect.height,
+			plane_state->clip_rect.width,
+			plane_state->clip_rect.x,
+			plane_state->clip_rect.y);
+
+	pipe_ctx->stream->dst.x -= timing->h_border_left;
+	pipe_ctx->stream->dst.y -= timing->v_border_top;
+
+	return res;
+}
+
+bool resource_can_pipe_disable_cursor(struct pipe_ctx *pipe_ctx)
+{
+	struct pipe_ctx *test_pipe, *split_pipe;
+	struct rect r1 = pipe_ctx->plane_res.scl_data.recout;
+	int r1_right, r1_bottom;
+	int cur_layer = pipe_ctx->plane_state->layer_index;
+
+	reverse_adjust_recout_for_visual_confirm(&r1, pipe_ctx);
+	r1_right = r1.x + r1.width;
+	r1_bottom = r1.y + r1.height;
+
+	/**
+	 * Disable the cursor if there's another pipe above this with a
+	 * plane that contains this pipe's viewport to prevent double cursor
+	 * and incorrect scaling artifacts.
+	 */
+	for (test_pipe = pipe_ctx->top_pipe; test_pipe;
+	     test_pipe = test_pipe->top_pipe) {
+		struct rect r2;
+		int r2_right, r2_bottom;
+		// Skip invisible layer and pipe-split plane on same layer
+		if (!test_pipe->plane_state ||
+		    !test_pipe->plane_state->visible ||
+		    test_pipe->plane_state->layer_index == cur_layer)
+			continue;
+
+		r2 = test_pipe->plane_res.scl_data.recout;
+		reverse_adjust_recout_for_visual_confirm(&r2, test_pipe);
+		r2_right = r2.x + r2.width;
+		r2_bottom = r2.y + r2.height;
+
+		/**
+		 * There is another half plane on same layer because of
+		 * pipe-split, merge together per same height.
+		 */
+		for (split_pipe = pipe_ctx->top_pipe; split_pipe;
+		     split_pipe = split_pipe->top_pipe)
+			if (split_pipe->plane_state->layer_index == test_pipe->plane_state->layer_index) {
+				struct rect r2_half;
+
+				r2_half = split_pipe->plane_res.scl_data.recout;
+				reverse_adjust_recout_for_visual_confirm(&r2_half, split_pipe);
+				r2.x = min(r2_half.x, r2.x);
+				r2.width = r2.width + r2_half.width;
+				r2_right = r2.x + r2.width;
+				r2_bottom = min(r2_bottom, r2_half.y + r2_half.height);
+				break;
+			}
+
+		if (r1.x >= r2.x && r1.y >= r2.y && r1_right <= r2_right && r1_bottom <= r2_bottom)
+			return true;
+	}
+
+	return false;
+}
+
+
+enum dc_status resource_build_scaling_params_for_context(
+	const struct dc  *dc,
+	struct dc_state *context)
+{
+	int i;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		if (context->res_ctx.pipe_ctx[i].plane_state != NULL &&
+				context->res_ctx.pipe_ctx[i].stream != NULL)
+			if (!resource_build_scaling_params(&context->res_ctx.pipe_ctx[i]))
+				return DC_FAIL_SCALING;
+	}
+
+	return DC_OK;
+}
+
+struct pipe_ctx *resource_find_free_secondary_pipe_legacy(
+		struct resource_context *res_ctx,
+		const struct resource_pool *pool,
+		const struct pipe_ctx *primary_pipe)
+{
+	int i;
+	struct pipe_ctx *secondary_pipe = NULL;
+
+	/*
+	 * We add a preferred pipe mapping to avoid the chance that
+	 * MPCCs already in use will need to be reassigned to other trees.
+	 * For example, if we went with the strict, assign backwards logic:
+	 *
+	 * (State 1)
+	 * Display A on, no surface, top pipe = 0
+	 * Display B on, no surface, top pipe = 1
+	 *
+	 * (State 2)
+	 * Display A on, no surface, top pipe = 0
+	 * Display B on, surface enable, top pipe = 1, bottom pipe = 5
+	 *
+	 * (State 3)
+	 * Display A on, surface enable, top pipe = 0, bottom pipe = 5
+	 * Display B on, surface enable, top pipe = 1, bottom pipe = 4
+	 *
+	 * The state 2->3 transition requires remapping MPCC 5 from display B
+	 * to display A.
+	 *
+	 * However, with the preferred pipe logic, state 2 would look like:
+	 *
+	 * (State 2)
+	 * Display A on, no surface, top pipe = 0
+	 * Display B on, surface enable, top pipe = 1, bottom pipe = 4
+	 *
+	 * This would then cause 2->3 to not require remapping any MPCCs.
+	 */
+	if (primary_pipe) {
+		int preferred_pipe_idx = (pool->pipe_count - 1) - primary_pipe->pipe_idx;
+		if (res_ctx->pipe_ctx[preferred_pipe_idx].stream == NULL) {
+			secondary_pipe = &res_ctx->pipe_ctx[preferred_pipe_idx];
+			secondary_pipe->pipe_idx = preferred_pipe_idx;
+		}
+	}
+
+	/*
+	 * search backwards for the second pipe to keep pipe
+	 * assignment more consistent
+	 */
+	if (!secondary_pipe)
+		for (i = pool->pipe_count - 1; i >= 0; i--) {
+			if (res_ctx->pipe_ctx[i].stream == NULL) {
+				secondary_pipe = &res_ctx->pipe_ctx[i];
+				secondary_pipe->pipe_idx = i;
+				break;
+			}
+		}
+
+	return secondary_pipe;
+}
+
+int resource_find_free_pipe_used_as_sec_opp_head_by_cur_otg_master(
+		const struct resource_context *cur_res_ctx,
+		struct resource_context *new_res_ctx,
+		const struct pipe_ctx *cur_otg_master)
+{
+	const struct pipe_ctx *cur_sec_opp_head = cur_otg_master->next_odm_pipe;
+	struct pipe_ctx *new_pipe;
+	int free_pipe_idx = FREE_PIPE_INDEX_NOT_FOUND;
+
+	while (cur_sec_opp_head) {
+		new_pipe = &new_res_ctx->pipe_ctx[cur_sec_opp_head->pipe_idx];
+		if (resource_is_pipe_type(new_pipe, FREE_PIPE)) {
+			free_pipe_idx = cur_sec_opp_head->pipe_idx;
+			break;
+		}
+		cur_sec_opp_head = cur_sec_opp_head->next_odm_pipe;
+	}
+
+	return free_pipe_idx;
+}
+
+int resource_find_free_pipe_used_in_cur_mpc_blending_tree(
+		const struct resource_context *cur_res_ctx,
+		struct resource_context *new_res_ctx,
+		const struct pipe_ctx *cur_opp_head)
+{
+	const struct pipe_ctx *cur_sec_dpp = cur_opp_head->bottom_pipe;
+	struct pipe_ctx *new_pipe;
+	int free_pipe_idx = FREE_PIPE_INDEX_NOT_FOUND;
+
+	while (cur_sec_dpp) {
+		/* find a free pipe used in current opp blend tree,
+		 * this is to avoid MPO pipe switching to different opp blending
+		 * tree
+		 */
+		new_pipe = &new_res_ctx->pipe_ctx[cur_sec_dpp->pipe_idx];
+		if (resource_is_pipe_type(new_pipe, FREE_PIPE)) {
+			free_pipe_idx = cur_sec_dpp->pipe_idx;
+			break;
+		}
+		cur_sec_dpp = cur_sec_dpp->bottom_pipe;
+	}
+
+	return free_pipe_idx;
+}
+
+int recource_find_free_pipe_not_used_in_cur_res_ctx(
+		const struct resource_context *cur_res_ctx,
+		struct resource_context *new_res_ctx,
+		const struct resource_pool *pool)
+{
+	int free_pipe_idx = FREE_PIPE_INDEX_NOT_FOUND;
+	const struct pipe_ctx *new_pipe, *cur_pipe;
+	int i;
+
+	for (i = 0; i < pool->pipe_count; i++) {
+		cur_pipe = &cur_res_ctx->pipe_ctx[i];
+		new_pipe = &new_res_ctx->pipe_ctx[i];
+
+		if (resource_is_pipe_type(cur_pipe, FREE_PIPE) &&
+				resource_is_pipe_type(new_pipe, FREE_PIPE)) {
+			free_pipe_idx = i;
+			break;
+		}
+	}
+
+	return free_pipe_idx;
+}
+
+int recource_find_free_pipe_used_as_otg_master_in_cur_res_ctx(
+		const struct resource_context *cur_res_ctx,
+		struct resource_context *new_res_ctx,
+		const struct resource_pool *pool)
+{
+	int free_pipe_idx = FREE_PIPE_INDEX_NOT_FOUND;
+	const struct pipe_ctx *new_pipe, *cur_pipe;
+	int i;
+
+	for (i = 0; i < pool->pipe_count; i++) {
+		cur_pipe = &cur_res_ctx->pipe_ctx[i];
+		new_pipe = &new_res_ctx->pipe_ctx[i];
+
+		if (resource_is_pipe_type(cur_pipe, OTG_MASTER) &&
+				resource_is_pipe_type(new_pipe, FREE_PIPE)) {
+			free_pipe_idx = i;
+			break;
+		}
+	}
+
+	return free_pipe_idx;
+}
+
+int resource_find_free_pipe_used_as_cur_sec_dpp(
+		const struct resource_context *cur_res_ctx,
+		struct resource_context *new_res_ctx,
+		const struct resource_pool *pool)
+{
+	int free_pipe_idx = FREE_PIPE_INDEX_NOT_FOUND;
+	const struct pipe_ctx *new_pipe, *cur_pipe;
+	int i;
+
+	for (i = 0; i < pool->pipe_count; i++) {
+		cur_pipe = &cur_res_ctx->pipe_ctx[i];
+		new_pipe = &new_res_ctx->pipe_ctx[i];
+
+		if (resource_is_pipe_type(cur_pipe, DPP_PIPE) &&
+				!resource_is_pipe_type(cur_pipe, OPP_HEAD) &&
+				resource_is_pipe_type(new_pipe, FREE_PIPE)) {
+			free_pipe_idx = i;
+			break;
+		}
+	}
+
+	return free_pipe_idx;
+}
+
+int resource_find_free_pipe_used_as_cur_sec_dpp_in_mpcc_combine(
+		const struct resource_context *cur_res_ctx,
+		struct resource_context *new_res_ctx,
+		const struct resource_pool *pool)
+{
+	int free_pipe_idx = FREE_PIPE_INDEX_NOT_FOUND;
+	const struct pipe_ctx *new_pipe, *cur_pipe;
+	int i;
+
+	for (i = 0; i < pool->pipe_count; i++) {
+		cur_pipe = &cur_res_ctx->pipe_ctx[i];
+		new_pipe = &new_res_ctx->pipe_ctx[i];
+
+		if (resource_is_pipe_type(cur_pipe, DPP_PIPE) &&
+				!resource_is_pipe_type(cur_pipe, OPP_HEAD) &&
+				resource_get_mpc_slice_index(cur_pipe) > 0 &&
+				resource_is_pipe_type(new_pipe, FREE_PIPE)) {
+			free_pipe_idx = i;
+			break;
+		}
+	}
+
+	return free_pipe_idx;
+}
+
+int resource_find_any_free_pipe(struct resource_context *new_res_ctx,
+		const struct resource_pool *pool)
+{
+	int free_pipe_idx = FREE_PIPE_INDEX_NOT_FOUND;
+	const struct pipe_ctx *new_pipe;
+	int i;
+
+	for (i = 0; i < pool->pipe_count; i++) {
+		new_pipe = &new_res_ctx->pipe_ctx[i];
+
+		if (resource_is_pipe_type(new_pipe, FREE_PIPE)) {
+			free_pipe_idx = i;
+			break;
+		}
+	}
+
+	return free_pipe_idx;
+}
+
+bool resource_is_pipe_type(const struct pipe_ctx *pipe_ctx, enum pipe_type type)
+{
+	switch (type) {
+	case OTG_MASTER:
+		return !pipe_ctx->prev_odm_pipe &&
+				!pipe_ctx->top_pipe &&
+				pipe_ctx->stream;
+	case OPP_HEAD:
+		return !pipe_ctx->top_pipe && pipe_ctx->stream;
+	case DPP_PIPE:
+		return pipe_ctx->plane_state && pipe_ctx->stream;
+	case FREE_PIPE:
+		return !pipe_ctx->plane_state && !pipe_ctx->stream;
+	default:
+		return false;
+	}
+}
+
+struct pipe_ctx *resource_get_otg_master_for_stream(
+		struct resource_context *res_ctx,
+		const struct dc_stream_state *stream)
+{
+	int i;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		if (res_ctx->pipe_ctx[i].stream == stream &&
+				resource_is_pipe_type(&res_ctx->pipe_ctx[i], OTG_MASTER))
+			return &res_ctx->pipe_ctx[i];
+	}
+	return NULL;
+}
+
+int resource_get_opp_heads_for_otg_master(const struct pipe_ctx *otg_master,
+		struct resource_context *res_ctx,
+		struct pipe_ctx *opp_heads[MAX_PIPES])
+{
+	struct pipe_ctx *opp_head = &res_ctx->pipe_ctx[otg_master->pipe_idx];
+	struct dc *dc = otg_master->stream->ctx->dc;
+	int i = 0;
+
+	DC_LOGGER_INIT(dc->ctx->logger);
+
+	if (!resource_is_pipe_type(otg_master, OTG_MASTER)) {
+		DC_LOG_WARNING("%s called from a non OTG master, something "
+			       "is wrong in the pipe configuration",
+			       __func__);
+		ASSERT(0);
+		return 0;
+	}
+	while (opp_head) {
+		ASSERT(i < MAX_PIPES);
+		opp_heads[i++] = opp_head;
+		opp_head = opp_head->next_odm_pipe;
+	}
+	return i;
+}
+
+int resource_get_dpp_pipes_for_opp_head(const struct pipe_ctx *opp_head,
+		struct resource_context *res_ctx,
+		struct pipe_ctx *dpp_pipes[MAX_PIPES])
+{
+	struct pipe_ctx *pipe = &res_ctx->pipe_ctx[opp_head->pipe_idx];
+	int i = 0;
+
+	if (!resource_is_pipe_type(opp_head, OPP_HEAD)) {
+		ASSERT(0);
+		return 0;
+	}
+	while (pipe && resource_is_pipe_type(pipe, DPP_PIPE)) {
+		ASSERT(i < MAX_PIPES);
+		dpp_pipes[i++] = pipe;
+		pipe = pipe->bottom_pipe;
+	}
+	return i;
+}
+
+int resource_get_dpp_pipes_for_plane(const struct dc_plane_state *plane,
+		struct resource_context *res_ctx,
+		struct pipe_ctx *dpp_pipes[MAX_PIPES])
+{
+	int i = 0, j;
+	struct pipe_ctx *pipe;
+
+	for (j = 0; j < MAX_PIPES; j++) {
+		pipe = &res_ctx->pipe_ctx[j];
+		if (pipe->plane_state == plane && pipe->prev_odm_pipe == NULL) {
+			if (resource_is_pipe_type(pipe, OPP_HEAD) ||
+					pipe->top_pipe->plane_state != plane)
+				break;
+		}
+	}
+
+	if (j < MAX_PIPES) {
+		if (pipe->next_odm_pipe)
+			while (pipe) {
+				dpp_pipes[i++] = pipe;
+				pipe = pipe->next_odm_pipe;
+			}
+		else
+			while (pipe && pipe->plane_state == plane) {
+				dpp_pipes[i++] = pipe;
+				pipe = pipe->bottom_pipe;
+			}
+	}
+	return i;
+}
+
+struct pipe_ctx *resource_get_otg_master(const struct pipe_ctx *pipe_ctx)
+{
+	struct pipe_ctx *otg_master = resource_get_opp_head(pipe_ctx);
+
+	while (otg_master->prev_odm_pipe)
+		otg_master = otg_master->prev_odm_pipe;
+	return otg_master;
+}
+
+struct pipe_ctx *resource_get_opp_head(const struct pipe_ctx *pipe_ctx)
+{
+	struct pipe_ctx *opp_head = (struct pipe_ctx *) pipe_ctx;
+
+	ASSERT(!resource_is_pipe_type(opp_head, FREE_PIPE));
+	while (opp_head->top_pipe)
+		opp_head = opp_head->top_pipe;
+	return opp_head;
+}
+
+struct pipe_ctx *resource_get_primary_dpp_pipe(const struct pipe_ctx *dpp_pipe)
+{
+	struct pipe_ctx *pri_dpp_pipe = (struct pipe_ctx *) dpp_pipe;
+
+	ASSERT(resource_is_pipe_type(dpp_pipe, DPP_PIPE));
+	while (pri_dpp_pipe->prev_odm_pipe)
+		pri_dpp_pipe = pri_dpp_pipe->prev_odm_pipe;
+	while (pri_dpp_pipe->top_pipe &&
+			pri_dpp_pipe->top_pipe->plane_state == pri_dpp_pipe->plane_state)
+		pri_dpp_pipe = pri_dpp_pipe->top_pipe;
+	return pri_dpp_pipe;
+}
+
+
+int resource_get_mpc_slice_index(const struct pipe_ctx *pipe_ctx)
+{
+	struct pipe_ctx *split_pipe = pipe_ctx->top_pipe;
+	int index = 0;
+
+	while (split_pipe && split_pipe->plane_state == pipe_ctx->plane_state) {
+		index++;
+		split_pipe = split_pipe->top_pipe;
+	}
+
+	return index;
+}
+
+int resource_get_mpc_slice_count(const struct pipe_ctx *pipe)
+{
+	int mpc_split_count = 1;
+	const struct pipe_ctx *other_pipe = pipe->bottom_pipe;
+
+	while (other_pipe && other_pipe->plane_state == pipe->plane_state) {
+		mpc_split_count++;
+		other_pipe = other_pipe->bottom_pipe;
+	}
+	other_pipe = pipe->top_pipe;
+	while (other_pipe && other_pipe->plane_state == pipe->plane_state) {
+		mpc_split_count++;
+		other_pipe = other_pipe->top_pipe;
+	}
+
+	return mpc_split_count;
+}
+
+int resource_get_odm_slice_count(const struct pipe_ctx *pipe)
+{
+	int odm_split_count = 1;
+
+	pipe = resource_get_otg_master(pipe);
+
+	while (pipe->next_odm_pipe) {
+		odm_split_count++;
+		pipe = pipe->next_odm_pipe;
+	}
+	return odm_split_count;
+}
+
+int resource_get_odm_slice_index(const struct pipe_ctx *pipe_ctx)
+{
+	int index = 0;
+
+	pipe_ctx = resource_get_opp_head(pipe_ctx);
+	if (!pipe_ctx)
+		return 0;
+
+	while (pipe_ctx->prev_odm_pipe) {
+		index++;
+		pipe_ctx = pipe_ctx->prev_odm_pipe;
+	}
+
+	return index;
+}
+
+int resource_get_odm_slice_dst_width(struct pipe_ctx *otg_master,
+		bool is_last_segment)
+{
+	const struct dc_crtc_timing *timing;
+	int count;
+	int h_active;
+	int width;
+	bool two_pixel_alignment_required = false;
+
+	if (!otg_master || !otg_master->stream)
+		return 0;
+
+	timing = &otg_master->stream->timing;
+	count = resource_get_odm_slice_count(otg_master);
+	h_active = timing->h_addressable +
+			timing->h_border_left +
+			timing->h_border_right +
+			otg_master->dsc_padding_params.dsc_hactive_padding;
+	width = h_active / count;
+
+	if (otg_master->stream_res.tg)
+		two_pixel_alignment_required =
+				otg_master->stream_res.tg->funcs->is_two_pixels_per_container(timing) ||
+				/*
+				 * 422 is sub-sampled horizontally. 1 set of chromas
+				 * (Cb/Cr) is shared for 2 lumas (i.e 2 Y values).
+				 * Therefore even if 422 is still 1 pixel per container,
+				 * ODM segment width still needs to be 2 pixel aligned.
+				 */
+				timing->pixel_encoding == PIXEL_ENCODING_YCBCR422;
+	if ((width % 2) && two_pixel_alignment_required)
+		width++;
+
+	return is_last_segment ?
+			h_active - width * (count - 1) :
+			width;
+}
+
+struct rect resource_get_odm_slice_dst_rect(struct pipe_ctx *pipe_ctx)
+{
+	const struct dc_stream_state *stream = pipe_ctx->stream;
+	bool is_last_odm_slice = pipe_ctx->next_odm_pipe == NULL;
+	struct pipe_ctx *otg_master = resource_get_otg_master(pipe_ctx);
+	int odm_slice_idx = resource_get_odm_slice_index(pipe_ctx);
+	int odm_segment_offset = resource_get_odm_slice_dst_width(otg_master, false);
+	struct rect odm_slice_dst;
+
+	odm_slice_dst.x = odm_segment_offset * odm_slice_idx;
+	odm_slice_dst.width = resource_get_odm_slice_dst_width(otg_master, is_last_odm_slice);
+	odm_slice_dst.y = 0;
+	odm_slice_dst.height = stream->timing.v_addressable +
+			stream->timing.v_border_bottom +
+			stream->timing.v_border_top;
+
+	return odm_slice_dst;
+}
+
+struct rect resource_get_odm_slice_src_rect(struct pipe_ctx *pipe_ctx)
+{
+	struct rect odm_slice_dst;
+	struct rect odm_slice_src;
+	struct pipe_ctx *opp_head = resource_get_opp_head(pipe_ctx);
+	struct output_pixel_processor *opp = opp_head->stream_res.opp;
+	uint32_t left_edge_extra_pixel_count;
+
+	odm_slice_dst = resource_get_odm_slice_dst_rect(opp_head);
+	odm_slice_src = odm_slice_dst;
+
+	if (opp && opp->funcs->opp_get_left_edge_extra_pixel_count)
+		left_edge_extra_pixel_count =
+				opp->funcs->opp_get_left_edge_extra_pixel_count(
+						opp, pipe_ctx->stream->timing.pixel_encoding,
+						resource_is_pipe_type(opp_head, OTG_MASTER));
+	else
+		left_edge_extra_pixel_count = 0;
+
+	odm_slice_src.x -= left_edge_extra_pixel_count;
+	odm_slice_src.width += left_edge_extra_pixel_count;
+
+	return odm_slice_src;
+}
+
+bool resource_is_pipe_topology_changed(const struct dc_state *state_a,
+		const struct dc_state *state_b)
+{
+	int i;
+	const struct pipe_ctx *pipe_a, *pipe_b;
+
+	if (state_a->stream_count != state_b->stream_count)
+		return true;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe_a = &state_a->res_ctx.pipe_ctx[i];
+		pipe_b = &state_b->res_ctx.pipe_ctx[i];
+
+		if (pipe_a->stream && !pipe_b->stream)
+			return true;
+		else if (!pipe_a->stream && pipe_b->stream)
+			return true;
+
+		if (pipe_a->plane_state && !pipe_b->plane_state)
+			return true;
+		else if (!pipe_a->plane_state && pipe_b->plane_state)
+			return true;
+
+		if (pipe_a->bottom_pipe && pipe_b->bottom_pipe) {
+			if (pipe_a->bottom_pipe->pipe_idx != pipe_b->bottom_pipe->pipe_idx)
+				return true;
+			if ((pipe_a->bottom_pipe->plane_state == pipe_a->plane_state) &&
+					(pipe_b->bottom_pipe->plane_state != pipe_b->plane_state))
+				return true;
+			else if ((pipe_a->bottom_pipe->plane_state != pipe_a->plane_state) &&
+					(pipe_b->bottom_pipe->plane_state == pipe_b->plane_state))
+				return true;
+		} else if (pipe_a->bottom_pipe || pipe_b->bottom_pipe) {
+			return true;
+		}
+
+		if (pipe_a->next_odm_pipe && pipe_b->next_odm_pipe) {
+			if (pipe_a->next_odm_pipe->pipe_idx != pipe_b->next_odm_pipe->pipe_idx)
+				return true;
+		} else if (pipe_a->next_odm_pipe || pipe_b->next_odm_pipe) {
+			return true;
+		}
+	}
+	return false;
+}
+
+bool resource_is_odm_topology_changed(const struct pipe_ctx *otg_master_a,
+		const struct pipe_ctx *otg_master_b)
+{
+	const struct pipe_ctx *opp_head_a = otg_master_a;
+	const struct pipe_ctx *opp_head_b = otg_master_b;
+
+	if (!resource_is_pipe_type(otg_master_a, OTG_MASTER) ||
+			!resource_is_pipe_type(otg_master_b, OTG_MASTER))
+		return true;
+
+	while (opp_head_a && opp_head_b) {
+		if (opp_head_a->stream_res.opp != opp_head_b->stream_res.opp)
+			return true;
+		if ((opp_head_a->next_odm_pipe && !opp_head_b->next_odm_pipe) ||
+				(!opp_head_a->next_odm_pipe && opp_head_b->next_odm_pipe))
+			return true;
+		opp_head_a = opp_head_a->next_odm_pipe;
+		opp_head_b = opp_head_b->next_odm_pipe;
+	}
+
+	return false;
+}
+
+/*
+ * Sample log:
+ *    pipe topology update
+ *  ________________________
+ * | plane0  slice0  stream0|
+ * |DPP0----OPP0----OTG0----| <--- case 0 (OTG master pipe with plane)
+ * | plane1 |       |       |
+ * |DPP1----|       |       | <--- case 5 (DPP pipe not in last slice)
+ * | plane0  slice1 |       |
+ * |DPP2----OPP2----|       | <--- case 2 (OPP head pipe with plane)
+ * | plane1 |               |
+ * |DPP3----|               | <--- case 4 (DPP pipe in last slice)
+ * |         slice0  stream1|
+ * |DPG4----OPP4----OTG4----| <--- case 1 (OTG master pipe without plane)
+ * |         slice1 |       |
+ * |DPG5----OPP5----|       | <--- case 3 (OPP head pipe without plane)
+ * |________________________|
+ */
+
+static void resource_log_pipe(struct dc *dc, struct pipe_ctx *pipe,
+		int stream_idx, int slice_idx, int plane_idx, int slice_count,
+		bool is_primary)
+{
+	DC_LOGGER_INIT(dc->ctx->logger);
+
+	if (slice_idx == 0 && plane_idx == 0 && is_primary) {
+		/* case 0 (OTG master pipe with plane) */
+		DC_LOG_DC(" | plane%d  slice%d  stream%d|",
+				plane_idx, slice_idx, stream_idx);
+		DC_LOG_DC(" |DPP%d----OPP%d----OTG%d----|",
+				pipe->plane_res.dpp->inst,
+				pipe->stream_res.opp->inst,
+				pipe->stream_res.tg->inst);
+	} else if (slice_idx == 0 && plane_idx == -1) {
+		/* case 1 (OTG master pipe without plane) */
+		DC_LOG_DC(" |         slice%d  stream%d|",
+				slice_idx, stream_idx);
+		DC_LOG_DC(" |DPG%d----OPP%d----OTG%d----|",
+				pipe->stream_res.opp->inst,
+				pipe->stream_res.opp->inst,
+				pipe->stream_res.tg->inst);
+	} else if (slice_idx != 0 && plane_idx == 0 && is_primary) {
+		/* case 2 (OPP head pipe with plane) */
+		DC_LOG_DC(" | plane%d  slice%d |       |",
+				plane_idx, slice_idx);
+		DC_LOG_DC(" |DPP%d----OPP%d----|       |",
+				pipe->plane_res.dpp->inst,
+				pipe->stream_res.opp->inst);
+	} else if (slice_idx != 0 && plane_idx == -1) {
+		/* case 3 (OPP head pipe without plane) */
+		DC_LOG_DC(" |         slice%d |       |", slice_idx);
+		DC_LOG_DC(" |DPG%d----OPP%d----|       |",
+				pipe->plane_res.dpp->inst,
+				pipe->stream_res.opp->inst);
+	} else if (slice_idx == slice_count - 1) {
+		/* case 4 (DPP pipe in last slice) */
+		DC_LOG_DC(" | plane%d |               |", plane_idx);
+		DC_LOG_DC(" |DPP%d----|               |",
+				pipe->plane_res.dpp->inst);
+	} else {
+		/* case 5 (DPP pipe not in last slice) */
+		DC_LOG_DC(" | plane%d |       |       |", plane_idx);
+		DC_LOG_DC(" |DPP%d----|       |       |",
+				pipe->plane_res.dpp->inst);
+	}
+}
+
+static void resource_log_pipe_for_stream(struct dc *dc, struct dc_state *state,
+		struct pipe_ctx *otg_master, int stream_idx)
+{
+	struct pipe_ctx *opp_heads[MAX_PIPES];
+	struct pipe_ctx *dpp_pipes[MAX_PIPES];
+
+	int slice_idx, dpp_idx, plane_idx, slice_count, dpp_count;
+	bool is_primary;
+	DC_LOGGER_INIT(dc->ctx->logger);
+
+	slice_count = resource_get_opp_heads_for_otg_master(otg_master,
+			&state->res_ctx, opp_heads);
+	for (slice_idx = 0; slice_idx < slice_count; slice_idx++) {
+		plane_idx = -1;
+		if (opp_heads[slice_idx]->plane_state) {
+			dpp_count = resource_get_dpp_pipes_for_opp_head(
+					opp_heads[slice_idx],
+					&state->res_ctx,
+					dpp_pipes);
+			for (dpp_idx = 0; dpp_idx < dpp_count; dpp_idx++) {
+				is_primary = !dpp_pipes[dpp_idx]->top_pipe ||
+						dpp_pipes[dpp_idx]->top_pipe->plane_state != dpp_pipes[dpp_idx]->plane_state;
+				if (is_primary)
+					plane_idx++;
+				resource_log_pipe(dc, dpp_pipes[dpp_idx],
+						stream_idx, slice_idx,
+						plane_idx, slice_count,
+						is_primary);
+			}
+		} else {
+			resource_log_pipe(dc, opp_heads[slice_idx],
+					stream_idx, slice_idx, plane_idx,
+					slice_count, true);
+		}
+
+	}
+}
+
+static int resource_stream_to_stream_idx(struct dc_state *state,
+		struct dc_stream_state *stream)
+{
+	int i, stream_idx = -1;
+
+	for (i = 0; i < state->stream_count; i++)
+		if (state->streams[i] == stream) {
+			stream_idx = i;
+			break;
+		}
+
+	/* never return negative array index */
+	if (stream_idx == -1) {
+		ASSERT(0);
+		return 0;
+	}
+
+	return stream_idx;
+}
+
+void resource_log_pipe_topology_update(struct dc *dc, struct dc_state *state)
+{
+	struct pipe_ctx *otg_master;
+	int stream_idx, phantom_stream_idx;
+	DC_LOGGER_INIT(dc->ctx->logger);
+
+	DC_LOG_DC("    pipe topology update");
+	DC_LOG_DC("  ________________________");
+	for (stream_idx = 0; stream_idx < state->stream_count; stream_idx++) {
+		if (state->streams[stream_idx]->is_phantom)
+			continue;
+
+		otg_master = resource_get_otg_master_for_stream(
+				&state->res_ctx, state->streams[stream_idx]);
+
+		if (!otg_master)
+			continue;
+
+		resource_log_pipe_for_stream(dc, state, otg_master, stream_idx);
+	}
+	if (state->phantom_stream_count > 0) {
+		DC_LOG_DC(" |    (phantom pipes)     |");
+		for (stream_idx = 0; stream_idx < state->stream_count; stream_idx++) {
+			if (state->stream_status[stream_idx].mall_stream_config.type != SUBVP_MAIN)
+				continue;
+
+			phantom_stream_idx = resource_stream_to_stream_idx(state,
+					state->stream_status[stream_idx].mall_stream_config.paired_stream);
+			otg_master = resource_get_otg_master_for_stream(
+					&state->res_ctx, state->streams[phantom_stream_idx]);
+			if (!otg_master)
+				continue;
+
+			resource_log_pipe_for_stream(dc, state, otg_master, stream_idx);
+		}
+	}
+	DC_LOG_DC(" |________________________|\n");
+}
+
+static struct pipe_ctx *get_tail_pipe(
+		struct pipe_ctx *head_pipe)
+{
+	struct pipe_ctx *tail_pipe = head_pipe->bottom_pipe;
+
+	while (tail_pipe) {
+		head_pipe = tail_pipe;
+		tail_pipe = tail_pipe->bottom_pipe;
+	}
+
+	return head_pipe;
+}
+
+static struct pipe_ctx *get_last_opp_head(
+		struct pipe_ctx *opp_head)
+{
+	ASSERT(resource_is_pipe_type(opp_head, OPP_HEAD));
+	while (opp_head->next_odm_pipe)
+		opp_head = opp_head->next_odm_pipe;
+	return opp_head;
+}
+
+static struct pipe_ctx *get_last_dpp_pipe_in_mpcc_combine(
+		struct pipe_ctx *dpp_pipe)
+{
+	ASSERT(resource_is_pipe_type(dpp_pipe, DPP_PIPE));
+	while (dpp_pipe->bottom_pipe &&
+			dpp_pipe->plane_state == dpp_pipe->bottom_pipe->plane_state)
+		dpp_pipe = dpp_pipe->bottom_pipe;
+	return dpp_pipe;
+}
+
+static bool update_pipe_params_after_odm_slice_count_change(
+		struct pipe_ctx *otg_master,
+		struct dc_state *context,
+		const struct resource_pool *pool)
+{
+	int i;
+	struct pipe_ctx *pipe;
+	bool result = true;
+
+	for (i = 0; i < pool->pipe_count && result; i++) {
+		pipe = &context->res_ctx.pipe_ctx[i];
+		if (pipe->stream == otg_master->stream && pipe->plane_state)
+			result = resource_build_scaling_params(pipe);
+	}
+
+	if (pool->funcs->build_pipe_pix_clk_params)
+		pool->funcs->build_pipe_pix_clk_params(otg_master);
+
+	resource_build_test_pattern_params(&context->res_ctx, otg_master);
+
+	return result;
+}
+
+static bool update_pipe_params_after_mpc_slice_count_change(
+		const struct dc_plane_state *plane,
+		struct dc_state *context,
+		const struct resource_pool *pool)
+{
+	int i;
+	struct pipe_ctx *pipe;
+	bool result = true;
+
+	for (i = 0; i < pool->pipe_count && result; i++) {
+		pipe = &context->res_ctx.pipe_ctx[i];
+		if (pipe->plane_state == plane)
+			result = resource_build_scaling_params(pipe);
+	}
+	return result;
+}
+
+static int acquire_first_split_pipe(
+		struct resource_context *res_ctx,
+		const struct resource_pool *pool,
+		struct dc_stream_state *stream)
+{
+	int i;
+
+	for (i = 0; i < pool->pipe_count; i++) {
+		struct pipe_ctx *split_pipe = &res_ctx->pipe_ctx[i];
+
+		if (split_pipe->top_pipe &&
+				split_pipe->top_pipe->plane_state == split_pipe->plane_state) {
+			split_pipe->top_pipe->bottom_pipe = split_pipe->bottom_pipe;
+			if (split_pipe->bottom_pipe)
+				split_pipe->bottom_pipe->top_pipe = split_pipe->top_pipe;
+
+			if (split_pipe->top_pipe->plane_state)
+				resource_build_scaling_params(split_pipe->top_pipe);
+
+			memset(split_pipe, 0, sizeof(*split_pipe));
+			split_pipe->stream_res.tg = pool->timing_generators[i];
+			split_pipe->plane_res.hubp = pool->hubps[i];
+			split_pipe->plane_res.ipp = pool->ipps[i];
+			split_pipe->plane_res.dpp = pool->dpps[i];
+			split_pipe->stream_res.opp = pool->opps[i];
+			split_pipe->plane_res.mpcc_inst = pool->dpps[i]->inst;
+			split_pipe->pipe_idx = i;
+
+			split_pipe->stream = stream;
+			return i;
+		}
+	}
+	return FREE_PIPE_INDEX_NOT_FOUND;
+}
+
+static void update_stream_engine_usage(
+		struct resource_context *res_ctx,
+		const struct resource_pool *pool,
+		struct stream_encoder *stream_enc,
+		bool acquired)
+{
+	int i;
+
+	for (i = 0; i < pool->stream_enc_count; i++) {
+		if (pool->stream_enc[i] == stream_enc)
+			res_ctx->is_stream_enc_acquired[i] = acquired;
+	}
+}
+
+static void update_hpo_dp_stream_engine_usage(
+		struct resource_context *res_ctx,
+		const struct resource_pool *pool,
+		struct hpo_dp_stream_encoder *hpo_dp_stream_enc,
+		bool acquired)
+{
+	int i;
+
+	for (i = 0; i < pool->hpo_dp_stream_enc_count; i++) {
+		if (pool->hpo_dp_stream_enc[i] == hpo_dp_stream_enc)
+			res_ctx->is_hpo_dp_stream_enc_acquired[i] = acquired;
+	}
+}
+
+static inline int find_acquired_hpo_dp_link_enc_for_link(
+		const struct resource_context *res_ctx,
+		const struct dc_link *link)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(res_ctx->hpo_dp_link_enc_to_link_idx); i++)
+		if (res_ctx->hpo_dp_link_enc_ref_cnts[i] > 0 &&
+				res_ctx->hpo_dp_link_enc_to_link_idx[i] == link->link_index)
+			return i;
+
+	return -1;
+}
+
+static inline int find_free_hpo_dp_link_enc(const struct resource_context *res_ctx,
+		const struct resource_pool *pool)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(res_ctx->hpo_dp_link_enc_ref_cnts); i++)
+		if (res_ctx->hpo_dp_link_enc_ref_cnts[i] == 0)
+			break;
+
+	return (i < ARRAY_SIZE(res_ctx->hpo_dp_link_enc_ref_cnts) &&
+			i < pool->hpo_dp_link_enc_count) ? i : -1;
+}
+
+static inline void acquire_hpo_dp_link_enc(
+		struct resource_context *res_ctx,
+		unsigned int link_index,
+		int enc_index)
+{
+	res_ctx->hpo_dp_link_enc_to_link_idx[enc_index] = link_index;
+	res_ctx->hpo_dp_link_enc_ref_cnts[enc_index] = 1;
+}
+
+static inline void retain_hpo_dp_link_enc(
+		struct resource_context *res_ctx,
+		int enc_index)
+{
+	res_ctx->hpo_dp_link_enc_ref_cnts[enc_index]++;
+}
+
+static inline void release_hpo_dp_link_enc(
+		struct resource_context *res_ctx,
+		int enc_index)
+{
+	ASSERT(res_ctx->hpo_dp_link_enc_ref_cnts[enc_index] > 0);
+	res_ctx->hpo_dp_link_enc_ref_cnts[enc_index]--;
+}
+
+static bool add_hpo_dp_link_enc_to_ctx(struct resource_context *res_ctx,
+		const struct resource_pool *pool,
+		struct pipe_ctx *pipe_ctx,
+		struct dc_stream_state *stream)
+{
+	int enc_index;
+
+	enc_index = find_acquired_hpo_dp_link_enc_for_link(res_ctx, stream->link);
+
+	if (enc_index >= 0) {
+		retain_hpo_dp_link_enc(res_ctx, enc_index);
+	} else {
+		enc_index = find_free_hpo_dp_link_enc(res_ctx, pool);
+		if (enc_index >= 0)
+			acquire_hpo_dp_link_enc(res_ctx, stream->link->link_index, enc_index);
+	}
+
+	if (enc_index >= 0)
+		pipe_ctx->link_res.hpo_dp_link_enc = pool->hpo_dp_link_enc[enc_index];
+
+	return pipe_ctx->link_res.hpo_dp_link_enc != NULL;
+}
+
+static void remove_hpo_dp_link_enc_from_ctx(struct resource_context *res_ctx,
+		struct pipe_ctx *pipe_ctx,
+		struct dc_stream_state *stream)
+{
+	int enc_index;
+
+	enc_index = find_acquired_hpo_dp_link_enc_for_link(res_ctx, stream->link);
+
+	if (enc_index >= 0) {
+		release_hpo_dp_link_enc(res_ctx, enc_index);
+		pipe_ctx->link_res.hpo_dp_link_enc = NULL;
+	}
+}
+
+static inline int find_acquired_dio_link_enc_for_link(
+		const struct resource_context *res_ctx,
+		const struct dc_link *link)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(res_ctx->dio_link_enc_ref_cnts); i++)
+		if (res_ctx->dio_link_enc_ref_cnts[i] > 0 &&
+				res_ctx->dio_link_enc_to_link_idx[i] == link->link_index)
+			return i;
+
+	return -1;
+}
+
+static inline int find_fixed_dio_link_enc(const struct dc_link *link)
+{
+	/* the 8b10b dp phy can only use fixed link encoder */
+	return link->eng_id;
+}
+
+static inline int find_free_dio_link_enc(const struct resource_context *res_ctx,
+		const struct dc_link *link, const struct resource_pool *pool)
+{
+	int i;
+	int enc_count = pool->dig_link_enc_count;
+
+	/* for dpia, check preferred encoder first and then the next one */
+	for (i = 0; i < enc_count; i++)
+		if (res_ctx->dio_link_enc_ref_cnts[(link->dpia_preferred_eng_id + i) % enc_count] == 0)
+			break;
+
+	return (i >= 0 && i < enc_count) ? (link->dpia_preferred_eng_id + i) % enc_count : -1;
+}
+
+static inline void acquire_dio_link_enc(
+		struct resource_context *res_ctx,
+		unsigned int link_index,
+		int enc_index)
+{
+	res_ctx->dio_link_enc_to_link_idx[enc_index] = link_index;
+	res_ctx->dio_link_enc_ref_cnts[enc_index] = 1;
+}
+
+static inline void retain_dio_link_enc(
+		struct resource_context *res_ctx,
+		int enc_index)
+{
+	res_ctx->dio_link_enc_ref_cnts[enc_index]++;
+}
+
+static inline void release_dio_link_enc(
+		struct resource_context *res_ctx,
+		int enc_index)
+{
+	ASSERT(res_ctx->dio_link_enc_ref_cnts[enc_index] > 0);
+	res_ctx->dio_link_enc_ref_cnts[enc_index]--;
+}
+
+static bool is_dio_enc_acquired_by_other_link(const struct dc_link *link,
+		int enc_index,
+		int *link_index)
+{
+	const struct dc *dc  = link->dc;
+	const struct resource_context *res_ctx = &dc->current_state->res_ctx;
+
+	/* pass the link_index that acquired the enc_index */
+	if (res_ctx->dio_link_enc_ref_cnts[enc_index] > 0 &&
+			res_ctx->dio_link_enc_to_link_idx[enc_index] != link->link_index) {
+		*link_index = res_ctx->dio_link_enc_to_link_idx[enc_index];
+		return true;
+	}
+
+	return false;
+}
+
+static void swap_dio_link_enc_to_muxable_ctx(struct dc_state *context,
+		const struct resource_pool *pool,
+		int new_encoder,
+		int old_encoder)
+{
+	struct resource_context *res_ctx = &context->res_ctx;
+	int stream_count = context->stream_count;
+	int i = 0;
+
+	res_ctx->dio_link_enc_ref_cnts[new_encoder] = res_ctx->dio_link_enc_ref_cnts[old_encoder];
+	res_ctx->dio_link_enc_to_link_idx[new_encoder] = res_ctx->dio_link_enc_to_link_idx[old_encoder];
+	res_ctx->dio_link_enc_ref_cnts[old_encoder] = 0;
+
+	for (i = 0; i < stream_count; i++) {
+		struct dc_stream_state *stream = context->streams[i];
+		struct pipe_ctx *pipe_ctx = resource_get_otg_master_for_stream(&context->res_ctx, stream);
+
+		if (pipe_ctx && pipe_ctx->link_res.dio_link_enc == pool->link_encoders[old_encoder])
+			pipe_ctx->link_res.dio_link_enc = pool->link_encoders[new_encoder];
+	}
+}
+
+static bool add_dio_link_enc_to_ctx(const struct dc *dc,
+		struct dc_state *context,
+		const struct resource_pool *pool,
+		struct pipe_ctx *pipe_ctx,
+		struct dc_stream_state *stream)
+{
+	struct resource_context *res_ctx = &context->res_ctx;
+	int enc_index;
+
+	enc_index = find_acquired_dio_link_enc_for_link(res_ctx, stream->link);
+
+	if (enc_index >= 0) {
+		retain_dio_link_enc(res_ctx, enc_index);
+	} else {
+		if (stream->link->is_dig_mapping_flexible)
+			enc_index = find_free_dio_link_enc(res_ctx, stream->link, pool);
+		else {
+			int link_index = 0;
+
+			enc_index = find_fixed_dio_link_enc(stream->link);
+			/* Fixed mapping link can only use its fixed link encoder.
+			 * If the encoder is acquired by other link then get a new free encoder and swap the new
+			 * one into the acquiring link.
+			 */
+			if (enc_index >= 0 && is_dio_enc_acquired_by_other_link(stream->link, enc_index, &link_index)) {
+				int new_enc_index = find_free_dio_link_enc(res_ctx, dc->links[link_index], pool);
+
+				if (new_enc_index >= 0)
+					swap_dio_link_enc_to_muxable_ctx(context, pool, new_enc_index, enc_index);
+				else
+					return false;
+			}
+		}
+
+		if (enc_index >= 0)
+			acquire_dio_link_enc(res_ctx, stream->link->link_index, enc_index);
+	}
+
+	if (enc_index >= 0)
+		pipe_ctx->link_res.dio_link_enc = pool->link_encoders[enc_index];
+
+	return pipe_ctx->link_res.dio_link_enc != NULL;
+}
+
+static void remove_dio_link_enc_from_ctx(struct resource_context *res_ctx,
+		struct pipe_ctx *pipe_ctx,
+		struct dc_stream_state *stream)
+{
+	int enc_index = -1;
+
+	if (stream->link)
+		enc_index = find_acquired_dio_link_enc_for_link(res_ctx, stream->link);
+
+	if (enc_index >= 0) {
+		release_dio_link_enc(res_ctx, enc_index);
+		pipe_ctx->link_res.dio_link_enc = NULL;
+	}
+}
+
+static int get_num_of_free_pipes(const struct resource_pool *pool, const struct dc_state *context)
+{
+	int i;
+	int count = 0;
+
+	for (i = 0; i < pool->pipe_count; i++)
+		if (resource_is_pipe_type(&context->res_ctx.pipe_ctx[i], FREE_PIPE))
+			count++;
+	return count;
+}
+
+enum dc_status resource_add_otg_master_for_stream_output(struct dc_state *new_ctx,
+		const struct resource_pool *pool,
+		struct dc_stream_state *stream)
+{
+	struct dc *dc = stream->ctx->dc;
+
+	return dc->res_pool->funcs->add_stream_to_ctx(dc, new_ctx, stream);
+}
+
+void resource_remove_otg_master_for_stream_output(struct dc_state *context,
+		const struct resource_pool *pool,
+		struct dc_stream_state *stream)
+{
+	struct pipe_ctx *otg_master = resource_get_otg_master_for_stream(
+			&context->res_ctx, stream);
+
+	if (!otg_master)
+		return;
+
+	ASSERT(resource_get_odm_slice_count(otg_master) == 1);
+	ASSERT(otg_master->plane_state == NULL);
+	ASSERT(otg_master->stream_res.stream_enc);
+	update_stream_engine_usage(
+			&context->res_ctx,
+			pool,
+			otg_master->stream_res.stream_enc,
+			false);
+
+	if (stream->ctx->dc->link_srv->dp_is_128b_132b_signal(otg_master)) {
+		update_hpo_dp_stream_engine_usage(
+			&context->res_ctx, pool,
+			otg_master->stream_res.hpo_dp_stream_enc,
+			false);
+		remove_hpo_dp_link_enc_from_ctx(
+				&context->res_ctx, otg_master, stream);
+	}
+
+	if (stream->ctx->dc->config.unify_link_enc_assignment)
+		remove_dio_link_enc_from_ctx(&context->res_ctx, otg_master, stream);
+
+	if (otg_master->stream_res.audio)
+		update_audio_usage(
+			&context->res_ctx,
+			pool,
+			otg_master->stream_res.audio,
+			false);
+
+	resource_unreference_clock_source(&context->res_ctx,
+					  pool,
+					  otg_master->clock_source);
+
+	if (pool->funcs->remove_stream_from_ctx)
+		pool->funcs->remove_stream_from_ctx(
+				stream->ctx->dc, context, stream);
+
+	memset(otg_master, 0, sizeof(*otg_master));
+}
+
+/* For each OPP head of an OTG master, add top plane at plane index 0.
+ *
+ * In the following example, the stream has 2 ODM slices without a top plane.
+ * By adding a plane 0 to OPP heads, we are configuring our hardware to render
+ * plane 0 by using each OPP head's DPP.
+ *
+ *       Inter-pipe Relation (Before Adding Plane)
+ *        __________________________________________________
+ *       |PIPE IDX|   DPP PIPES   | OPP HEADS | OTG MASTER  |
+ *       |        |               | slice 0   |             |
+ *       |   0    |               |blank ----ODM----------- |
+ *       |        |               | slice 1 | |             |
+ *       |   1    |               |blank ---- |             |
+ *       |________|_______________|___________|_____________|
+ *
+ *       Inter-pipe Relation (After Adding Plane)
+ *        __________________________________________________
+ *       |PIPE IDX|   DPP PIPES   | OPP HEADS | OTG MASTER  |
+ *       |        |  plane 0      | slice 0   |             |
+ *       |   0    | -------------------------ODM----------- |
+ *       |        |  plane 0      | slice 1 | |             |
+ *       |   1    | ------------------------- |             |
+ *       |________|_______________|___________|_____________|
+ */
+static bool add_plane_to_opp_head_pipes(struct pipe_ctx *otg_master_pipe,
+		struct dc_plane_state *plane_state,
+		struct dc_state *context)
+{
+	struct pipe_ctx *opp_head_pipe = otg_master_pipe;
+
+	while (opp_head_pipe) {
+		if (opp_head_pipe->plane_state) {
+			ASSERT(0);
+			return false;
+		}
+		opp_head_pipe->plane_state = plane_state;
+		opp_head_pipe = opp_head_pipe->next_odm_pipe;
+	}
+
+	return true;
+}
+
+/* For each OPP head of an OTG master, acquire a secondary DPP pipe and add
+ * the plane. So the plane is added to all ODM slices associated with the OTG
+ * master pipe in the bottom layer.
+ *
+ * In the following example, the stream has 2 ODM slices and a top plane 0.
+ * By acquiring secondary DPP pipes and adding a plane 1, we are configuring our
+ * hardware to render the plane 1 by acquiring a new pipe for each ODM slice and
+ * render plane 1 using new pipes' DPP in the Z axis below plane 0.
+ *
+ *       Inter-pipe Relation (Before Adding Plane)
+ *        __________________________________________________
+ *       |PIPE IDX|   DPP PIPES   | OPP HEADS | OTG MASTER  |
+ *       |        |  plane 0      | slice 0   |             |
+ *       |   0    | -------------------------ODM----------- |
+ *       |        |  plane 0      | slice 1 | |             |
+ *       |   1    | ------------------------- |             |
+ *       |________|_______________|___________|_____________|
+ *
+ *       Inter-pipe Relation (After Acquiring and Adding Plane)
+ *        __________________________________________________
+ *       |PIPE IDX|   DPP PIPES   | OPP HEADS | OTG MASTER  |
+ *       |        |  plane 0      | slice 0   |             |
+ *       |   0    | -------------MPC---------ODM----------- |
+ *       |        |  plane 1    | |         | |             |
+ *       |   2    | ------------- |         | |             |
+ *       |        |  plane 0      | slice 1 | |             |
+ *       |   1    | -------------MPC--------- |             |
+ *       |        |  plane 1    | |           |             |
+ *       |   3    | ------------- |           |             |
+ *       |________|_______________|___________|_____________|
+ */
+static bool acquire_secondary_dpp_pipes_and_add_plane(
+		struct pipe_ctx *otg_master_pipe,
+		struct dc_plane_state *plane_state,
+		struct dc_state *new_ctx,
+		struct dc_state *cur_ctx,
+		struct resource_pool *pool)
+{
+	struct pipe_ctx *sec_pipe, *tail_pipe;
+	struct pipe_ctx *opp_heads[MAX_PIPES];
+	int opp_head_count;
+	int i;
+
+	if (!pool->funcs->acquire_free_pipe_as_secondary_dpp_pipe) {
+		ASSERT(0);
+		return false;
+	}
+
+	opp_head_count = resource_get_opp_heads_for_otg_master(otg_master_pipe,
+			&new_ctx->res_ctx, opp_heads);
+	if (get_num_of_free_pipes(pool, new_ctx) < opp_head_count)
+		/* not enough free pipes */
+		return false;
+
+	for (i = 0; i < opp_head_count; i++) {
+		sec_pipe = pool->funcs->acquire_free_pipe_as_secondary_dpp_pipe(
+				cur_ctx,
+				new_ctx,
+				pool,
+				opp_heads[i]);
+		ASSERT(sec_pipe);
+		sec_pipe->plane_state = plane_state;
+
+		/* establish pipe relationship */
+		tail_pipe = get_tail_pipe(opp_heads[i]);
+		tail_pipe->bottom_pipe = sec_pipe;
+		sec_pipe->top_pipe = tail_pipe;
+		sec_pipe->bottom_pipe = NULL;
+		if (tail_pipe->prev_odm_pipe) {
+			ASSERT(tail_pipe->prev_odm_pipe->bottom_pipe);
+			sec_pipe->prev_odm_pipe = tail_pipe->prev_odm_pipe->bottom_pipe;
+			tail_pipe->prev_odm_pipe->bottom_pipe->next_odm_pipe = sec_pipe;
+		} else {
+			sec_pipe->prev_odm_pipe = NULL;
+		}
+	}
+	return true;
+}
+
+bool resource_append_dpp_pipes_for_plane_composition(
+		struct dc_state *new_ctx,
+		struct dc_state *cur_ctx,
+		struct resource_pool *pool,
+		struct pipe_ctx *otg_master_pipe,
+		struct dc_plane_state *plane_state)
+{
+	bool success;
+
+	if (otg_master_pipe->plane_state == NULL)
+		success = add_plane_to_opp_head_pipes(otg_master_pipe,
+				plane_state, new_ctx);
+	else
+		success = acquire_secondary_dpp_pipes_and_add_plane(
+				otg_master_pipe, plane_state, new_ctx,
+				cur_ctx, pool);
+	if (success) {
+		/* when appending a plane mpc slice count changes from 0 to 1 */
+		success = update_pipe_params_after_mpc_slice_count_change(
+				plane_state, new_ctx, pool);
+		if (!success)
+			resource_remove_dpp_pipes_for_plane_composition(new_ctx,
+					pool, plane_state);
+	}
+
+	return success;
+}
+
+void resource_remove_dpp_pipes_for_plane_composition(
+		struct dc_state *context,
+		const struct resource_pool *pool,
+		const struct dc_plane_state *plane_state)
+{
+	int i;
+
+	for (i = pool->pipe_count - 1; i >= 0; i--) {
+		struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i];
+
+		if (pipe_ctx->plane_state == plane_state) {
+			if (pipe_ctx->top_pipe)
+				pipe_ctx->top_pipe->bottom_pipe = pipe_ctx->bottom_pipe;
+
+			/* Second condition is to avoid setting NULL to top pipe
+			 * of tail pipe making it look like head pipe in subsequent
+			 * deletes
+			 */
+			if (pipe_ctx->bottom_pipe && pipe_ctx->top_pipe)
+				pipe_ctx->bottom_pipe->top_pipe = pipe_ctx->top_pipe;
+
+			/*
+			 * For head pipe detach surfaces from pipe for tail
+			 * pipe just zero it out
+			 */
+			if (!pipe_ctx->top_pipe)
+				pipe_ctx->plane_state = NULL;
+			else
+				memset(pipe_ctx, 0, sizeof(*pipe_ctx));
+		}
+	}
+}
+
+/*
+ * Increase ODM slice count by 1 by acquiring pipes and adding a new ODM slice
+ * at the last index.
+ * return - true if a new ODM slice is added and required pipes are acquired.
+ * false if new_ctx is no longer a valid state after new ODM slice is added.
+ *
+ * This is achieved by duplicating MPC blending tree from previous ODM slice.
+ * In the following example, we have a single MPC tree and 1 ODM slice 0. We
+ * want to add a new odm slice by duplicating the MPC blending tree and add
+ * ODM slice 1.
+ *
+ *       Inter-pipe Relation (Before Acquiring and Adding ODM Slice)
+ *        __________________________________________________
+ *       |PIPE IDX|   DPP PIPES   | OPP HEADS | OTG MASTER  |
+ *       |        |  plane 0      | slice 0   |             |
+ *       |   0    | -------------MPC---------ODM----------- |
+ *       |        |  plane 1    | |           |             |
+ *       |   1    | ------------- |           |             |
+ *       |________|_______________|___________|_____________|
+ *
+ *       Inter-pipe Relation (After Acquiring and Adding ODM Slice)
+ *        __________________________________________________
+ *       |PIPE IDX|   DPP PIPES   | OPP HEADS | OTG MASTER  |
+ *       |        |  plane 0      | slice 0   |             |
+ *       |   0    | -------------MPC---------ODM----------- |
+ *       |        |  plane 1    | |         | |             |
+ *       |   1    | ------------- |         | |             |
+ *       |        |  plane 0      | slice 1 | |             |
+ *       |   2    | -------------MPC--------- |             |
+ *       |        |  plane 1    | |           |             |
+ *       |   3    | ------------- |           |             |
+ *       |________|_______________|___________|_____________|
+ */
+static bool acquire_pipes_and_add_odm_slice(
+		struct pipe_ctx *otg_master_pipe,
+		struct dc_state *new_ctx,
+		const struct dc_state *cur_ctx,
+		const struct resource_pool *pool)
+{
+	struct pipe_ctx *last_opp_head = get_last_opp_head(otg_master_pipe);
+	struct pipe_ctx *new_opp_head;
+	struct pipe_ctx *last_top_dpp_pipe, *last_bottom_dpp_pipe,
+			*new_top_dpp_pipe, *new_bottom_dpp_pipe;
+
+	if (!pool->funcs->acquire_free_pipe_as_secondary_opp_head) {
+		ASSERT(0);
+		return false;
+	}
+	new_opp_head = pool->funcs->acquire_free_pipe_as_secondary_opp_head(
+					cur_ctx, new_ctx, pool,
+					otg_master_pipe);
+	if (!new_opp_head)
+		return false;
+
+	last_opp_head->next_odm_pipe = new_opp_head;
+	new_opp_head->prev_odm_pipe = last_opp_head;
+	new_opp_head->next_odm_pipe = NULL;
+	new_opp_head->plane_state = last_opp_head->plane_state;
+	last_top_dpp_pipe = last_opp_head;
+	new_top_dpp_pipe = new_opp_head;
+
+	while (last_top_dpp_pipe->bottom_pipe) {
+		last_bottom_dpp_pipe = last_top_dpp_pipe->bottom_pipe;
+		new_bottom_dpp_pipe = pool->funcs->acquire_free_pipe_as_secondary_dpp_pipe(
+				cur_ctx, new_ctx, pool,
+				new_opp_head);
+		if (!new_bottom_dpp_pipe)
+			return false;
+
+		new_bottom_dpp_pipe->plane_state = last_bottom_dpp_pipe->plane_state;
+		new_top_dpp_pipe->bottom_pipe = new_bottom_dpp_pipe;
+		new_bottom_dpp_pipe->top_pipe = new_top_dpp_pipe;
+		last_bottom_dpp_pipe->next_odm_pipe = new_bottom_dpp_pipe;
+		new_bottom_dpp_pipe->prev_odm_pipe = last_bottom_dpp_pipe;
+		new_bottom_dpp_pipe->next_odm_pipe = NULL;
+		last_top_dpp_pipe = last_bottom_dpp_pipe;
+	}
+
+	return true;
+}
+
+/*
+ * Decrease ODM slice count by 1 by releasing pipes and removing the ODM slice
+ * at the last index.
+ * return - true if the last ODM slice is removed and related pipes are
+ * released. false if there is no removable ODM slice.
+ *
+ * In the following example, we have 2 MPC trees and ODM slice 0 and slice 1.
+ * We want to remove the last ODM i.e slice 1. We are releasing secondary DPP
+ * pipe 3 and OPP head pipe 2.
+ *
+ *       Inter-pipe Relation (Before Releasing and Removing ODM Slice)
+ *        __________________________________________________
+ *       |PIPE IDX|   DPP PIPES   | OPP HEADS | OTG MASTER  |
+ *       |        |  plane 0      | slice 0   |             |
+ *       |   0    | -------------MPC---------ODM----------- |
+ *       |        |  plane 1    | |         | |             |
+ *       |   1    | ------------- |         | |             |
+ *       |        |  plane 0      | slice 1 | |             |
+ *       |   2    | -------------MPC--------- |             |
+ *       |        |  plane 1    | |           |             |
+ *       |   3    | ------------- |           |             |
+ *       |________|_______________|___________|_____________|
+ *
+ *       Inter-pipe Relation (After Releasing and Removing ODM Slice)
+ *        __________________________________________________
+ *       |PIPE IDX|   DPP PIPES   | OPP HEADS | OTG MASTER  |
+ *       |        |  plane 0      | slice 0   |             |
+ *       |   0    | -------------MPC---------ODM----------- |
+ *       |        |  plane 1    | |           |             |
+ *       |   1    | ------------- |           |             |
+ *       |________|_______________|___________|_____________|
+ */
+static bool release_pipes_and_remove_odm_slice(
+		struct pipe_ctx *otg_master_pipe,
+		struct dc_state *context,
+		const struct resource_pool *pool)
+{
+	struct pipe_ctx *last_opp_head = get_last_opp_head(otg_master_pipe);
+	struct pipe_ctx *tail_pipe = get_tail_pipe(last_opp_head);
+
+	if (!pool->funcs->release_pipe) {
+		ASSERT(0);
+		return false;
+	}
+
+	if (resource_is_pipe_type(last_opp_head, OTG_MASTER))
+		return false;
+
+	while (tail_pipe->top_pipe) {
+		tail_pipe->prev_odm_pipe->next_odm_pipe = NULL;
+		tail_pipe = tail_pipe->top_pipe;
+		pool->funcs->release_pipe(context, tail_pipe->bottom_pipe, pool);
+		tail_pipe->bottom_pipe = NULL;
+	}
+	last_opp_head->prev_odm_pipe->next_odm_pipe = NULL;
+	pool->funcs->release_pipe(context, last_opp_head, pool);
+
+	return true;
+}
+
+/*
+ * Increase MPC slice count by 1 by acquiring a new DPP pipe and add it as the
+ * last MPC slice of the plane associated with dpp_pipe.
+ *
+ * return - true if a new MPC slice is added and required pipes are acquired.
+ * false if new_ctx is no longer a valid state after new MPC slice is added.
+ *
+ * In the following example, we add a new MPC slice for plane 0 into the
+ * new_ctx. To do so we pass pipe 0 as dpp_pipe. The function acquires a new DPP
+ * pipe 2 for plane 0 as the bottom most pipe for plane 0.
+ *
+ *       Inter-pipe Relation (Before Acquiring and Adding MPC Slice)
+ *        __________________________________________________
+ *       |PIPE IDX|   DPP PIPES   | OPP HEADS | OTG MASTER  |
+ *       |        |  plane 0      |           |             |
+ *       |   0    | -------------MPC----------------------- |
+ *       |        |  plane 1    | |           |             |
+ *       |   1    | ------------- |           |             |
+ *       |________|_______________|___________|_____________|
+ *
+ *       Inter-pipe Relation (After Acquiring and Adding MPC Slice)
+ *        __________________________________________________
+ *       |PIPE IDX|   DPP PIPES   | OPP HEADS | OTG MASTER  |
+ *       |        |  plane 0      |           |             |
+ *       |   0    | -------------MPC----------------------- |
+ *       |        |  plane 0    | |           |             |
+ *       |   2    | ------------- |           |             |
+ *       |        |  plane 1    | |           |             |
+ *       |   1    | ------------- |           |             |
+ *       |________|_______________|___________|_____________|
+ */
+static bool acquire_dpp_pipe_and_add_mpc_slice(
+		struct pipe_ctx *dpp_pipe,
+		struct dc_state *new_ctx,
+		const struct dc_state *cur_ctx,
+		const struct resource_pool *pool)
+{
+	struct pipe_ctx *last_dpp_pipe =
+			get_last_dpp_pipe_in_mpcc_combine(dpp_pipe);
+	struct pipe_ctx *opp_head = resource_get_opp_head(dpp_pipe);
+	struct pipe_ctx *new_dpp_pipe;
+
+	if (!pool->funcs->acquire_free_pipe_as_secondary_dpp_pipe) {
+		ASSERT(0);
+		return false;
+	}
+	new_dpp_pipe = pool->funcs->acquire_free_pipe_as_secondary_dpp_pipe(
+			cur_ctx, new_ctx, pool, opp_head);
+	if (!new_dpp_pipe || resource_get_odm_slice_count(dpp_pipe) > 1)
+		return false;
+
+	new_dpp_pipe->bottom_pipe = last_dpp_pipe->bottom_pipe;
+	if (new_dpp_pipe->bottom_pipe)
+		new_dpp_pipe->bottom_pipe->top_pipe = new_dpp_pipe;
+	new_dpp_pipe->top_pipe = last_dpp_pipe;
+	last_dpp_pipe->bottom_pipe = new_dpp_pipe;
+	new_dpp_pipe->plane_state = last_dpp_pipe->plane_state;
+
+	return true;
+}
+
+/*
+ * Reduce MPC slice count by 1 by releasing the bottom DPP pipe in MPCC combine
+ * with dpp_pipe and removing last MPC slice of the plane associated with
+ * dpp_pipe.
+ *
+ * return - true if the last MPC slice of the plane associated with dpp_pipe is
+ * removed and last DPP pipe in MPCC combine with dpp_pipe is released.
+ * false if there is no removable MPC slice.
+ *
+ * In the following example, we remove an MPC slice for plane 0 from the
+ * context. To do so we pass pipe 0 as dpp_pipe. The function releases pipe 1 as
+ * it is the last pipe for plane 0.
+ *
+ *       Inter-pipe Relation (Before Releasing and Removing MPC Slice)
+ *        __________________________________________________
+ *       |PIPE IDX|   DPP PIPES   | OPP HEADS | OTG MASTER  |
+ *       |        |  plane 0      |           |             |
+ *       |   0    | -------------MPC----------------------- |
+ *       |        |  plane 0    | |           |             |
+ *       |   1    | ------------- |           |             |
+ *       |        |  plane 1    | |           |             |
+ *       |   2    | ------------- |           |             |
+ *       |________|_______________|___________|_____________|
+ *
+ *       Inter-pipe Relation (After Releasing and Removing MPC Slice)
+ *        __________________________________________________
+ *       |PIPE IDX|   DPP PIPES   | OPP HEADS | OTG MASTER  |
+ *       |        |  plane 0      |           |             |
+ *       |   0    | -------------MPC----------------------- |
+ *       |        |  plane 1    | |           |             |
+ *       |   2    | ------------- |           |             |
+ *       |________|_______________|___________|_____________|
+ */
+static bool release_dpp_pipe_and_remove_mpc_slice(
+		struct pipe_ctx *dpp_pipe,
+		struct dc_state *context,
+		const struct resource_pool *pool)
+{
+	struct pipe_ctx *last_dpp_pipe =
+			get_last_dpp_pipe_in_mpcc_combine(dpp_pipe);
+
+	if (!pool->funcs->release_pipe) {
+		ASSERT(0);
+		return false;
+	}
+
+	if (resource_is_pipe_type(last_dpp_pipe, OPP_HEAD) ||
+			resource_get_odm_slice_count(dpp_pipe) > 1)
+		return false;
+
+	last_dpp_pipe->top_pipe->bottom_pipe = last_dpp_pipe->bottom_pipe;
+	if (last_dpp_pipe->bottom_pipe)
+		last_dpp_pipe->bottom_pipe->top_pipe = last_dpp_pipe->top_pipe;
+	pool->funcs->release_pipe(context, last_dpp_pipe, pool);
+
+	return true;
+}
+
+bool resource_update_pipes_for_stream_with_slice_count(
+		struct dc_state *new_ctx,
+		const struct dc_state *cur_ctx,
+		const struct resource_pool *pool,
+		const struct dc_stream_state *stream,
+		int new_slice_count)
+{
+	int i;
+	struct pipe_ctx *otg_master = resource_get_otg_master_for_stream(
+			&new_ctx->res_ctx, stream);
+	int cur_slice_count;
+	bool result = true;
+
+	if (!otg_master)
+		return false;
+
+	cur_slice_count = resource_get_odm_slice_count(otg_master);
+
+	if (new_slice_count == cur_slice_count)
+		return result;
+
+	if (new_slice_count > cur_slice_count)
+		for (i = 0; i < new_slice_count - cur_slice_count && result; i++)
+			result = acquire_pipes_and_add_odm_slice(
+					otg_master, new_ctx, cur_ctx, pool);
+	else
+		for (i = 0; i < cur_slice_count - new_slice_count && result; i++)
+			result = release_pipes_and_remove_odm_slice(
+					otg_master, new_ctx, pool);
+	if (result)
+		result = update_pipe_params_after_odm_slice_count_change(
+				otg_master, new_ctx, pool);
+	return result;
+}
+
+bool resource_update_pipes_for_plane_with_slice_count(
+		struct dc_state *new_ctx,
+		const struct dc_state *cur_ctx,
+		const struct resource_pool *pool,
+		const struct dc_plane_state *plane,
+		int new_slice_count)
+{
+	int i;
+	int dpp_pipe_count;
+	int cur_slice_count;
+	struct pipe_ctx *dpp_pipes[MAX_PIPES] = {0};
+	bool result = true;
+
+	dpp_pipe_count = resource_get_dpp_pipes_for_plane(plane,
+			&new_ctx->res_ctx, dpp_pipes);
+	ASSERT(dpp_pipe_count > 0);
+	cur_slice_count = resource_get_mpc_slice_count(dpp_pipes[0]);
+
+	if (new_slice_count == cur_slice_count)
+		return result;
+
+	if (new_slice_count > cur_slice_count)
+		for (i = 0; i < new_slice_count - cur_slice_count && result; i++)
+			result = acquire_dpp_pipe_and_add_mpc_slice(
+					dpp_pipes[0], new_ctx, cur_ctx, pool);
+	else
+		for (i = 0; i < cur_slice_count - new_slice_count && result; i++)
+			result = release_dpp_pipe_and_remove_mpc_slice(
+					dpp_pipes[0], new_ctx, pool);
+	if (result)
+		result = update_pipe_params_after_mpc_slice_count_change(
+				dpp_pipes[0]->plane_state, new_ctx, pool);
+	return result;
+}
+
+bool dc_is_timing_changed(struct dc_stream_state *cur_stream,
+		       struct dc_stream_state *new_stream)
+{
+	if (cur_stream == NULL)
+		return true;
+
+	/* If output color space is changed, need to reprogram info frames */
+	if (cur_stream->output_color_space != new_stream->output_color_space)
+		return true;
+
+	return memcmp(
+		&cur_stream->timing,
+		&new_stream->timing,
+		sizeof(struct dc_crtc_timing)) != 0;
+}
+
+static bool are_stream_backends_same(
+	struct dc_stream_state *stream_a, struct dc_stream_state *stream_b)
+{
+	if (stream_a == stream_b)
+		return true;
+
+	if (stream_a == NULL || stream_b == NULL)
+		return false;
+
+	if (dc_is_timing_changed(stream_a, stream_b))
+		return false;
+
+	if (stream_a->signal != stream_b->signal)
+		return false;
+
+	if (stream_a->dpms_off != stream_b->dpms_off)
+		return false;
+
+	return true;
+}
+
+/*
+ * dc_is_stream_unchanged() - Compare two stream states for equivalence.
+ *
+ * Checks if there a difference between the two states
+ * that would require a mode change.
+ *
+ * Does not compare cursor position or attributes.
+ */
+bool dc_is_stream_unchanged(
+	struct dc_stream_state *old_stream, struct dc_stream_state *stream)
+{
+	if (!old_stream || !stream)
+		return false;
+
+	if (!are_stream_backends_same(old_stream, stream))
+		return false;
+
+	if (old_stream->ignore_msa_timing_param != stream->ignore_msa_timing_param)
+		return false;
+
+	/*compare audio info*/
+	if (memcmp(&old_stream->audio_info, &stream->audio_info, sizeof(stream->audio_info)) != 0)
+		return false;
+
+	return true;
+}
+
+/*
+ * dc_is_stream_scaling_unchanged() - Compare scaling rectangles of two streams.
+ */
+bool dc_is_stream_scaling_unchanged(struct dc_stream_state *old_stream,
+				    struct dc_stream_state *stream)
+{
+	if (old_stream == stream)
+		return true;
+
+	if (old_stream == NULL || stream == NULL)
+		return false;
+
+	if (memcmp(&old_stream->src,
+			&stream->src,
+			sizeof(struct rect)) != 0)
+		return false;
+
+	if (memcmp(&old_stream->dst,
+			&stream->dst,
+			sizeof(struct rect)) != 0)
+		return false;
+
+	return true;
+}
+
+/* TODO: release audio object */
+void update_audio_usage(
+		struct resource_context *res_ctx,
+		const struct resource_pool *pool,
+		struct audio *audio,
+		bool acquired)
+{
+	int i;
+	for (i = 0; i < pool->audio_count; i++) {
+		if (pool->audios[i] == audio)
+			res_ctx->is_audio_acquired[i] = acquired;
+	}
+}
+
+static struct hpo_dp_stream_encoder *find_first_free_match_hpo_dp_stream_enc_for_link(
+		struct resource_context *res_ctx,
+		const struct resource_pool *pool,
+		struct dc_stream_state *stream)
+{
+	int i;
+
+	for (i = 0; i < pool->hpo_dp_stream_enc_count; i++) {
+		if (!res_ctx->is_hpo_dp_stream_enc_acquired[i] &&
+				pool->hpo_dp_stream_enc[i]) {
+
+			return pool->hpo_dp_stream_enc[i];
+		}
+	}
+
+	return NULL;
+}
+
+static struct audio *find_first_free_audio(
+		struct resource_context *res_ctx,
+		const struct resource_pool *pool,
+		enum engine_id id,
+		enum dce_version dc_version)
+{
+	int i, available_audio_count;
+
+	if (id == ENGINE_ID_UNKNOWN)
+		return NULL;
+
+	available_audio_count = pool->audio_count;
+
+	for (i = 0; i < available_audio_count; i++) {
+		if ((res_ctx->is_audio_acquired[i] == false) && (res_ctx->is_stream_enc_acquired[i] == true)) {
+			/*we have enough audio endpoint, find the matching inst*/
+			if (id != i)
+				continue;
+			return pool->audios[i];
+		}
+	}
+
+	/* use engine id to find free audio */
+	if ((id < available_audio_count) && (res_ctx->is_audio_acquired[id] == false)) {
+		return pool->audios[id];
+	}
+	/*not found the matching one, first come first serve*/
+	for (i = 0; i < available_audio_count; i++) {
+		if (res_ctx->is_audio_acquired[i] == false) {
+			return pool->audios[i];
+		}
+	}
+	return NULL;
+}
+
+static struct dc_stream_state *find_pll_sharable_stream(
+		struct dc_stream_state *stream_needs_pll,
+		struct dc_state *context)
+{
+	int i;
+
+	for (i = 0; i < context->stream_count; i++) {
+		struct dc_stream_state *stream_has_pll = context->streams[i];
+
+		/* We are looking for non dp, non virtual stream */
+		if (resource_are_streams_timing_synchronizable(
+			stream_needs_pll, stream_has_pll)
+			&& !dc_is_dp_signal(stream_has_pll->signal)
+			&& stream_has_pll->link->connector_signal
+			!= SIGNAL_TYPE_VIRTUAL)
+			return stream_has_pll;
+
+	}
+
+	return NULL;
+}
+
+static int get_norm_pix_clk(const struct dc_crtc_timing *timing)
+{
+	uint32_t pix_clk = timing->pix_clk_100hz;
+	uint32_t normalized_pix_clk = pix_clk;
+
+	if (timing->pixel_encoding == PIXEL_ENCODING_YCBCR420)
+		pix_clk /= 2;
+	if (timing->pixel_encoding != PIXEL_ENCODING_YCBCR422) {
+		switch (timing->display_color_depth) {
+		case COLOR_DEPTH_666:
+		case COLOR_DEPTH_888:
+			normalized_pix_clk = pix_clk;
+			break;
+		case COLOR_DEPTH_101010:
+			normalized_pix_clk = (pix_clk * 30) / 24;
+			break;
+		case COLOR_DEPTH_121212:
+			normalized_pix_clk = (pix_clk * 36) / 24;
+			break;
+		case COLOR_DEPTH_141414:
+			normalized_pix_clk = (pix_clk * 42) / 24;
+			break;
+		case COLOR_DEPTH_161616:
+			normalized_pix_clk = (pix_clk * 48) / 24;
+			break;
+		default:
+			ASSERT(0);
+		break;
+		}
+	}
+	return normalized_pix_clk;
+}
+
+static void calculate_phy_pix_clks(struct dc_stream_state *stream)
+{
+	/* update actual pixel clock on all streams */
+	if (dc_is_hdmi_signal(stream->signal))
+		stream->phy_pix_clk = get_norm_pix_clk(
+			&stream->timing) / 10;
+	else
+		stream->phy_pix_clk =
+			stream->timing.pix_clk_100hz / 10;
+
+	if (stream->timing.timing_3d_format == TIMING_3D_FORMAT_HW_FRAME_PACKING)
+		stream->phy_pix_clk *= 2;
+}
+
+static int acquire_resource_from_hw_enabled_state(
+		struct resource_context *res_ctx,
+		const struct resource_pool *pool,
+		struct dc_stream_state *stream)
+{
+	struct dc_link *link = stream->link;
+	unsigned int i, inst, tg_inst = 0;
+	uint32_t numPipes = 1;
+	uint32_t id_src[4] = {0};
+
+	/* Check for enabled DIG to identify enabled display */
+	if (!link->link_enc->funcs->is_dig_enabled(link->link_enc))
+		return -1;
+
+	inst = link->link_enc->funcs->get_dig_frontend(link->link_enc);
+
+	if (inst == ENGINE_ID_UNKNOWN)
+		return -1;
+
+	for (i = 0; i < pool->stream_enc_count; i++) {
+		if (pool->stream_enc[i]->id == inst) {
+			tg_inst = pool->stream_enc[i]->funcs->dig_source_otg(
+				pool->stream_enc[i]);
+			break;
+		}
+	}
+
+	// tg_inst not found
+	if (i == pool->stream_enc_count)
+		return -1;
+
+	if (tg_inst >= pool->timing_generator_count)
+		return -1;
+
+	if (!res_ctx->pipe_ctx[tg_inst].stream) {
+		struct pipe_ctx *pipe_ctx = &res_ctx->pipe_ctx[tg_inst];
+
+		pipe_ctx->stream_res.tg = pool->timing_generators[tg_inst];
+		id_src[0] = tg_inst;
+
+		if (pipe_ctx->stream_res.tg->funcs->get_optc_source)
+			pipe_ctx->stream_res.tg->funcs->get_optc_source(pipe_ctx->stream_res.tg,
+						&numPipes, &id_src[0], &id_src[1]);
+
+		if (id_src[0] == 0xf && id_src[1] == 0xf) {
+			id_src[0] = tg_inst;
+			numPipes = 1;
+		}
+
+		for (i = 0; i < numPipes; i++) {
+			//Check if src id invalid
+			if (id_src[i] == 0xf)
+				return -1;
+
+			pipe_ctx = &res_ctx->pipe_ctx[id_src[i]];
+
+			pipe_ctx->stream_res.tg = pool->timing_generators[tg_inst];
+			pipe_ctx->plane_res.mi = pool->mis[id_src[i]];
+			pipe_ctx->plane_res.hubp = pool->hubps[id_src[i]];
+			pipe_ctx->plane_res.ipp = pool->ipps[id_src[i]];
+			pipe_ctx->plane_res.xfm = pool->transforms[id_src[i]];
+			pipe_ctx->plane_res.dpp = pool->dpps[id_src[i]];
+			pipe_ctx->stream_res.opp = pool->opps[id_src[i]];
+
+			if (pool->dpps[id_src[i]]) {
+				pipe_ctx->plane_res.mpcc_inst = pool->dpps[id_src[i]]->inst;
+
+				if (pool->mpc->funcs->read_mpcc_state) {
+					struct mpcc_state s = {0};
+
+					pool->mpc->funcs->read_mpcc_state(pool->mpc, pipe_ctx->plane_res.mpcc_inst, &s);
+
+					if (s.dpp_id < MAX_MPCC)
+						pool->mpc->mpcc_array[pipe_ctx->plane_res.mpcc_inst].dpp_id =
+								s.dpp_id;
+
+					if (s.bot_mpcc_id < MAX_MPCC)
+						pool->mpc->mpcc_array[pipe_ctx->plane_res.mpcc_inst].mpcc_bot =
+								&pool->mpc->mpcc_array[s.bot_mpcc_id];
+
+					if (s.opp_id < MAX_OPP)
+						pipe_ctx->stream_res.opp->mpc_tree_params.opp_id = s.opp_id;
+				}
+			}
+			pipe_ctx->pipe_idx = id_src[i];
+
+			if (id_src[i] >= pool->timing_generator_count) {
+				id_src[i] = pool->timing_generator_count - 1;
+
+				pipe_ctx->stream_res.tg = pool->timing_generators[id_src[i]];
+				pipe_ctx->stream_res.opp = pool->opps[id_src[i]];
+			}
+
+			pipe_ctx->stream = stream;
+		}
+
+		if (numPipes == 2) {
+			stream->apply_boot_odm_mode = dm_odm_combine_policy_2to1;
+			res_ctx->pipe_ctx[id_src[0]].next_odm_pipe = &res_ctx->pipe_ctx[id_src[1]];
+			res_ctx->pipe_ctx[id_src[0]].prev_odm_pipe = NULL;
+			res_ctx->pipe_ctx[id_src[1]].next_odm_pipe = NULL;
+			res_ctx->pipe_ctx[id_src[1]].prev_odm_pipe = &res_ctx->pipe_ctx[id_src[0]];
+		} else
+			stream->apply_boot_odm_mode = dm_odm_combine_mode_disabled;
+
+		return id_src[0];
+	}
+
+	return -1;
+}
+
+static void mark_seamless_boot_stream(const struct dc  *dc,
+				      struct dc_stream_state *stream)
+{
+	struct dc_bios *dcb = dc->ctx->dc_bios;
+
+	DC_LOGGER_INIT(dc->ctx->logger);
+
+	if (stream->apply_seamless_boot_optimization)
+		return;
+	if (!dc->config.allow_seamless_boot_optimization)
+		return;
+	if (dcb->funcs->is_accelerated_mode(dcb))
+		return;
+	if (dc_validate_boot_timing(dc, stream->sink, &stream->timing)) {
+		stream->apply_seamless_boot_optimization = true;
+		DC_LOG_DC("Marked stream for seamless boot optimization\n");
+	}
+}
+
+/*
+ * Acquire a pipe as OTG master and assign to the stream in new dc context.
+ * return - true if OTG master pipe is acquired and new dc context is updated.
+ * false if it fails to acquire an OTG master pipe for this stream.
+ *
+ * In the example below, we acquired pipe 0 as OTG master pipe for the stream.
+ * After the function its Inter-pipe Relation is represented by the diagram
+ * below.
+ *
+ *       Inter-pipe Relation
+ *        __________________________________________________
+ *       |PIPE IDX|   DPP PIPES   | OPP HEADS | OTG MASTER  |
+ *       |        |               |           |             |
+ *       |   0    |               |blank ------------------ |
+ *       |________|_______________|___________|_____________|
+ */
+static bool acquire_otg_master_pipe_for_stream(
+		const struct dc_state *cur_ctx,
+		struct dc_state *new_ctx,
+		const struct resource_pool *pool,
+		struct dc_stream_state *stream)
+{
+	/* TODO: Move this function to DCN specific resource file and acquire
+	 * DSC resource here. The reason is that the function should have the
+	 * same level of responsibility as when we acquire secondary OPP head.
+	 * We acquire DSC when we acquire secondary OPP head, so we should
+	 * acquire DSC when we acquire OTG master.
+	 */
+	int pipe_idx;
+	struct pipe_ctx *pipe_ctx = NULL;
+
+	/*
+	 * Upper level code is responsible to optimize unnecessary addition and
+	 * removal for unchanged streams. So unchanged stream will keep the same
+	 * OTG master instance allocated. When current stream is removed and a
+	 * new stream is added, we want to reuse the OTG instance made available
+	 * by the removed stream first. If not found, we try to avoid of using
+	 * any free pipes already used in current context as this could tear
+	 * down exiting ODM/MPC/MPO configuration unnecessarily.
+	 */
+
+	/*
+	 * Try to acquire the same OTG master already in use. This is not
+	 * optimal because resetting an enabled OTG master pipe for a new stream
+	 * requires an extra frame of wait. However there are test automation
+	 * and eDP assumptions that rely on reusing the same OTG master pipe
+	 * during mode change. We have to keep this logic as is for now.
+	 */
+	pipe_idx = recource_find_free_pipe_used_as_otg_master_in_cur_res_ctx(
+			&cur_ctx->res_ctx, &new_ctx->res_ctx, pool);
+	/*
+	 * Try to acquire a pipe not used in current resource context to avoid
+	 * pipe swapping.
+	 */
+	if (pipe_idx == FREE_PIPE_INDEX_NOT_FOUND)
+		pipe_idx = recource_find_free_pipe_not_used_in_cur_res_ctx(
+				&cur_ctx->res_ctx, &new_ctx->res_ctx, pool);
+	/*
+	 * If pipe swapping is unavoidable, try to acquire pipe used as
+	 * secondary DPP pipe in current state as we prioritize to support more
+	 * streams over supporting MPO planes.
+	 */
+	if (pipe_idx == FREE_PIPE_INDEX_NOT_FOUND)
+		pipe_idx = resource_find_free_pipe_used_as_cur_sec_dpp(
+				&cur_ctx->res_ctx, &new_ctx->res_ctx, pool);
+	if (pipe_idx == FREE_PIPE_INDEX_NOT_FOUND)
+		pipe_idx = resource_find_any_free_pipe(&new_ctx->res_ctx, pool);
+	if (pipe_idx != FREE_PIPE_INDEX_NOT_FOUND) {
+		pipe_ctx = &new_ctx->res_ctx.pipe_ctx[pipe_idx];
+		memset(pipe_ctx, 0, sizeof(*pipe_ctx));
+		pipe_ctx->pipe_idx = pipe_idx;
+		pipe_ctx->stream_res.tg = pool->timing_generators[pipe_idx];
+		pipe_ctx->plane_res.mi = pool->mis[pipe_idx];
+		pipe_ctx->plane_res.hubp = pool->hubps[pipe_idx];
+		pipe_ctx->plane_res.ipp = pool->ipps[pipe_idx];
+		pipe_ctx->plane_res.xfm = pool->transforms[pipe_idx];
+		pipe_ctx->plane_res.dpp = pool->dpps[pipe_idx];
+		pipe_ctx->stream_res.opp = pool->opps[pipe_idx];
+		if (pool->dpps[pipe_idx])
+			pipe_ctx->plane_res.mpcc_inst = pool->dpps[pipe_idx]->inst;
+
+		if (pipe_idx >= pool->timing_generator_count && pool->timing_generator_count != 0) {
+			int tg_inst = pool->timing_generator_count - 1;
+
+			pipe_ctx->stream_res.tg = pool->timing_generators[tg_inst];
+			pipe_ctx->stream_res.opp = pool->opps[tg_inst];
+		}
+
+		pipe_ctx->stream = stream;
+	} else {
+		pipe_idx = acquire_first_split_pipe(&new_ctx->res_ctx, pool, stream);
+	}
+
+	return pipe_idx != FREE_PIPE_INDEX_NOT_FOUND;
+}
+
+enum dc_status resource_map_pool_resources(
+		const struct dc  *dc,
+		struct dc_state *context,
+		struct dc_stream_state *stream)
+{
+	const struct resource_pool *pool = dc->res_pool;
+	int i;
+	struct dc_context *dc_ctx = dc->ctx;
+	struct pipe_ctx *pipe_ctx = NULL;
+	int pipe_idx = -1;
+	bool acquired = false;
+	bool is_dio_encoder = true;
+
+	calculate_phy_pix_clks(stream);
+
+	mark_seamless_boot_stream(dc, stream);
+
+	if (stream->apply_seamless_boot_optimization) {
+		pipe_idx = acquire_resource_from_hw_enabled_state(
+				&context->res_ctx,
+				pool,
+				stream);
+		if (pipe_idx < 0)
+			/* hw resource was assigned to other stream */
+			stream->apply_seamless_boot_optimization = false;
+		else
+			acquired = true;
+	}
+
+	if (!acquired)
+		/* acquire new resources */
+		acquired = acquire_otg_master_pipe_for_stream(dc->current_state,
+				context, pool, stream);
+
+	pipe_ctx = resource_get_otg_master_for_stream(&context->res_ctx, stream);
+
+	if (!pipe_ctx || pipe_ctx->stream_res.tg == NULL)
+		return DC_NO_CONTROLLER_RESOURCE;
+
+	pipe_ctx->stream_res.stream_enc =
+		dc->res_pool->funcs->find_first_free_match_stream_enc_for_link(
+			&context->res_ctx, pool, stream);
+
+	if (!pipe_ctx->stream_res.stream_enc)
+		return DC_NO_STREAM_ENC_RESOURCE;
+
+	update_stream_engine_usage(
+		&context->res_ctx, pool,
+		pipe_ctx->stream_res.stream_enc,
+		true);
+
+	/* Allocate DP HPO Stream Encoder based on signal, hw capabilities
+	 * and link settings
+	 */
+	if (dc_is_dp_signal(stream->signal) ||
+			dc_is_virtual_signal(stream->signal)) {
+		if (!dc->link_srv->dp_decide_link_settings(stream,
+				&pipe_ctx->link_config.dp_link_settings))
+			return DC_FAIL_DP_LINK_BANDWIDTH;
+
+		dc->link_srv->dp_decide_tunnel_settings(stream,
+				&pipe_ctx->link_config.dp_tunnel_settings);
+
+		if (dc->link_srv->dp_get_encoding_format(
+				&pipe_ctx->link_config.dp_link_settings) == DP_128b_132b_ENCODING) {
+			pipe_ctx->stream_res.hpo_dp_stream_enc =
+					find_first_free_match_hpo_dp_stream_enc_for_link(
+							&context->res_ctx, pool, stream);
+
+			if (!pipe_ctx->stream_res.hpo_dp_stream_enc)
+				return DC_NO_STREAM_ENC_RESOURCE;
+
+			update_hpo_dp_stream_engine_usage(
+					&context->res_ctx, pool,
+					pipe_ctx->stream_res.hpo_dp_stream_enc,
+					true);
+			if (!add_hpo_dp_link_enc_to_ctx(&context->res_ctx, pool, pipe_ctx, stream))
+				return DC_NO_LINK_ENC_RESOURCE;
+		}
+	}
+
+	if (dc->config.unify_link_enc_assignment && is_dio_encoder)
+		if (!add_dio_link_enc_to_ctx(dc, context, pool, pipe_ctx, stream))
+			return DC_NO_LINK_ENC_RESOURCE;
+
+	/* TODO: Add check if ASIC support and EDID audio */
+	if (!stream->converter_disable_audio &&
+	    dc_is_audio_capable_signal(pipe_ctx->stream->signal) &&
+	    stream->audio_info.mode_count &&
+		(stream->audio_info.flags.all ||
+		(stream->sink && stream->sink->edid_caps.panel_patch.skip_audio_sab_check))) {
+		pipe_ctx->stream_res.audio = find_first_free_audio(
+		&context->res_ctx, pool, pipe_ctx->stream_res.stream_enc->id, dc_ctx->dce_version);
+
+		/*
+		 * Audio assigned in order first come first get.
+		 * There are asics which has number of audio
+		 * resources less then number of pipes
+		 */
+		if (pipe_ctx->stream_res.audio)
+			update_audio_usage(&context->res_ctx, pool,
+					   pipe_ctx->stream_res.audio, true);
+	}
+
+	/* Add ABM to the resource if on EDP */
+	if (pipe_ctx->stream && dc_is_embedded_signal(pipe_ctx->stream->signal)) {
+		if (pool->abm)
+			pipe_ctx->stream_res.abm = pool->abm;
+		else
+			pipe_ctx->stream_res.abm = pool->multiple_abms[pipe_ctx->stream_res.tg->inst];
+	}
+
+	for (i = 0; i < context->stream_count; i++)
+		if (context->streams[i] == stream) {
+			context->stream_status[i].primary_otg_inst = pipe_ctx->stream_res.tg->inst;
+			context->stream_status[i].stream_enc_inst = pipe_ctx->stream_res.stream_enc->stream_enc_inst;
+			context->stream_status[i].audio_inst =
+				pipe_ctx->stream_res.audio ? pipe_ctx->stream_res.audio->inst : -1;
+
+			return DC_OK;
+		}
+
+	DC_ERROR("Stream %p not found in new ctx!\n", stream);
+	return DC_ERROR_UNEXPECTED;
+}
+
+bool dc_resource_is_dsc_encoding_supported(const struct dc *dc)
+{
+	if (dc->res_pool == NULL)
+		return false;
+
+	return dc->res_pool->res_cap->num_dsc > 0;
+}
+
+static bool planes_changed_for_existing_stream(struct dc_state *context,
+					       struct dc_stream_state *stream,
+					       const struct dc_validation_set set[],
+					       int set_count)
+{
+	int i, j;
+	struct dc_stream_status *stream_status = NULL;
+
+	for (i = 0; i < context->stream_count; i++) {
+		if (context->streams[i] == stream) {
+			stream_status = &context->stream_status[i];
+			break;
+		}
+	}
+
+	if (!stream_status) {
+		ASSERT(0);
+		return false;
+	}
+
+	for (i = 0; i < set_count; i++)
+		if (set[i].stream == stream)
+			break;
+
+	if (i == set_count)
+		ASSERT(0);
+
+	if (set[i].plane_count != stream_status->plane_count)
+		return true;
+
+	for (j = 0; j < set[i].plane_count; j++)
+		if (set[i].plane_states[j] != stream_status->plane_states[j])
+			return true;
+
+	return false;
+}
+
+static bool add_all_planes_for_stream(
+		const struct dc *dc,
+		struct dc_stream_state *stream,
+		const struct dc_validation_set set[],
+		int set_count,
+		struct dc_state *state)
+{
+	int i, j;
+
+	for (i = 0; i < set_count; i++)
+		if (set[i].stream == stream)
+			break;
+
+	if (i == set_count) {
+		dm_error("Stream %p not found in set!\n", stream);
+		return false;
+	}
+
+	for (j = 0; j < set[i].plane_count; j++)
+		if (!dc_state_add_plane(dc, stream, set[i].plane_states[j], state))
+			return false;
+
+	return true;
+}
+
+/**
+ * dc_validate_with_context - Validate and update the potential new stream in the context object
+ *
+ * @dc: Used to get the current state status
+ * @set: An array of dc_validation_set with all the current streams reference
+ * @set_count: Total of streams
+ * @context: New context
+ * @validate_mode: identify the validation mode
+ *
+ * This function updates the potential new stream in the context object. It
+ * creates multiple lists for the add, remove, and unchanged streams. In
+ * particular, if the unchanged streams have a plane that changed, it is
+ * necessary to remove all planes from the unchanged streams. In summary, this
+ * function is responsible for validating the new context.
+ *
+ * Return:
+ * In case of success, return DC_OK (1), otherwise, return a DC error.
+ */
+enum dc_status dc_validate_with_context(struct dc *dc,
+					const struct dc_validation_set set[],
+					int set_count,
+					struct dc_state *context,
+					enum dc_validate_mode validate_mode)
+{
+	struct dc_stream_state *unchanged_streams[MAX_PIPES] = { 0 };
+	struct dc_stream_state *del_streams[MAX_PIPES] = { 0 };
+	struct dc_stream_state *add_streams[MAX_PIPES] = { 0 };
+	int old_stream_count = context->stream_count;
+	enum dc_status res = DC_ERROR_UNEXPECTED;
+	int unchanged_streams_count = 0;
+	int del_streams_count = 0;
+	int add_streams_count = 0;
+	bool found = false;
+	int i, j, k;
+
+	DC_LOGGER_INIT(dc->ctx->logger);
+
+	/* First build a list of streams to be remove from current context */
+	for (i = 0; i < old_stream_count; i++) {
+		struct dc_stream_state *stream = context->streams[i];
+
+		for (j = 0; j < set_count; j++) {
+			if (stream == set[j].stream) {
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			del_streams[del_streams_count++] = stream;
+
+		found = false;
+	}
+
+	/* Second, build a list of new streams */
+	for (i = 0; i < set_count; i++) {
+		struct dc_stream_state *stream = set[i].stream;
+
+		for (j = 0; j < old_stream_count; j++) {
+			if (stream == context->streams[j]) {
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			add_streams[add_streams_count++] = stream;
+
+		found = false;
+	}
+
+	/* Build a list of unchanged streams which is necessary for handling
+	 * planes change such as added, removed, and updated.
+	 */
+	for (i = 0; i < set_count; i++) {
+		/* Check if stream is part of the delete list */
+		for (j = 0; j < del_streams_count; j++) {
+			if (set[i].stream == del_streams[j]) {
+				found = true;
+				break;
+			}
+		}
+
+		if (!found) {
+			/* Check if stream is part of the add list */
+			for (j = 0; j < add_streams_count; j++) {
+				if (set[i].stream == add_streams[j]) {
+					found = true;
+					break;
+				}
+			}
+		}
+
+		if (!found)
+			unchanged_streams[unchanged_streams_count++] = set[i].stream;
+
+		found = false;
+	}
+
+	/* Remove all planes for unchanged streams if planes changed */
+	for (i = 0; i < unchanged_streams_count; i++) {
+		if (planes_changed_for_existing_stream(context,
+						       unchanged_streams[i],
+						       set,
+						       set_count)) {
+
+			if (!dc_state_rem_all_planes_for_stream(dc,
+							  unchanged_streams[i],
+							  context)) {
+				res = DC_FAIL_DETACH_SURFACES;
+				goto fail;
+			}
+		}
+	}
+
+	/* Remove all planes for removed streams and then remove the streams */
+	for (i = 0; i < del_streams_count; i++) {
+		/* Need to cpy the dwb data from the old stream in order to efc to work */
+		if (del_streams[i]->num_wb_info > 0) {
+			for (j = 0; j < add_streams_count; j++) {
+				if (del_streams[i]->sink == add_streams[j]->sink) {
+					add_streams[j]->num_wb_info = del_streams[i]->num_wb_info;
+					for (k = 0; k < del_streams[i]->num_wb_info; k++)
+						add_streams[j]->writeback_info[k] = del_streams[i]->writeback_info[k];
+				}
+			}
+		}
+
+		if (dc_state_get_stream_subvp_type(context, del_streams[i]) == SUBVP_PHANTOM) {
+			/* remove phantoms specifically */
+			if (!dc_state_rem_all_phantom_planes_for_stream(dc, del_streams[i], context, true)) {
+				res = DC_FAIL_DETACH_SURFACES;
+				goto fail;
+			}
+
+			res = dc_state_remove_phantom_stream(dc, context, del_streams[i]);
+			dc_state_release_phantom_stream(dc, context, del_streams[i]);
+		} else {
+			if (!dc_state_rem_all_planes_for_stream(dc, del_streams[i], context)) {
+				res = DC_FAIL_DETACH_SURFACES;
+				goto fail;
+			}
+
+			res = dc_state_remove_stream(dc, context, del_streams[i]);
+		}
+
+		if (res != DC_OK)
+			goto fail;
+	}
+
+	/* Swap seamless boot stream to pipe 0 (if needed) to ensure pipe_ctx
+	 * matches. This may change in the future if seamless_boot_stream can be
+	 * multiple.
+	 */
+	for (i = 0; i < add_streams_count; i++) {
+		mark_seamless_boot_stream(dc, add_streams[i]);
+		if (add_streams[i]->apply_seamless_boot_optimization && i != 0) {
+			struct dc_stream_state *temp = add_streams[0];
+
+			add_streams[0] = add_streams[i];
+			add_streams[i] = temp;
+			break;
+		}
+	}
+
+	/* Add new streams and then add all planes for the new stream */
+	for (i = 0; i < add_streams_count; i++) {
+		calculate_phy_pix_clks(add_streams[i]);
+		res = dc_state_add_stream(dc, context, add_streams[i]);
+		if (res != DC_OK)
+			goto fail;
+
+		if (!add_all_planes_for_stream(dc, add_streams[i], set, set_count, context)) {
+			res = DC_FAIL_ATTACH_SURFACES;
+			goto fail;
+		}
+	}
+
+	/* Add all planes for unchanged streams if planes changed */
+	for (i = 0; i < unchanged_streams_count; i++) {
+		if (planes_changed_for_existing_stream(context,
+						       unchanged_streams[i],
+						       set,
+						       set_count)) {
+			if (!add_all_planes_for_stream(dc, unchanged_streams[i], set, set_count, context)) {
+				res = DC_FAIL_ATTACH_SURFACES;
+				goto fail;
+			}
+		}
+	}
+
+	/* clear subvp cursor limitations */
+	for (i = 0; i < context->stream_count; i++) {
+		dc_state_set_stream_subvp_cursor_limit(context->streams[i], context, false);
+	}
+
+	res = dc_validate_global_state(dc, context, validate_mode);
+
+	/* calculate pixel rate divider after deciding pxiel clock & odm combine  */
+	if ((dc->hwss.calculate_pix_rate_divider) && (res == DC_OK)) {
+		for (i = 0; i < add_streams_count; i++)
+			dc->hwss.calculate_pix_rate_divider(dc, context, add_streams[i]);
+	}
+
+fail:
+	if (res != DC_OK)
+		DC_LOG_WARNING("%s:resource validation failed, dc_status:%d\n",
+			       __func__,
+			       res);
+
+	return res;
+}
+
+#if defined(CONFIG_DRM_AMD_DC_FP)
+#endif /* CONFIG_DRM_AMD_DC_FP */
+
+/**
+ * calculate_timing_params_for_dsc_with_padding - Calculates timing parameters for DSC with padding.
+ * @pipe_ctx: Pointer to the pipe context structure.
+ *
+ * This function calculates the timing parameters for a given pipe context based on the
+ * display stream compression (DSC) configuration. If the horizontal active pixels (hactive) are less
+ * than the total width of the DSC slices, it sets the dsc_hactive_padding value to the difference. If the
+ * total horizontal timing minus the dsc_hactive_padding value is less than 32, it resets the dsc_hactive_padding
+ * value to 0.
+ */
+static void calculate_timing_params_for_dsc_with_padding(struct pipe_ctx *pipe_ctx)
+{
+	struct dc_stream_state *stream = NULL;
+
+	if (!pipe_ctx)
+		return;
+
+	stream = pipe_ctx->stream;
+	pipe_ctx->dsc_padding_params.dsc_hactive_padding = 0;
+	pipe_ctx->dsc_padding_params.dsc_htotal_padding = 0;
+
+	if (stream)
+		pipe_ctx->dsc_padding_params.dsc_pix_clk_100hz = stream->timing.pix_clk_100hz;
+
+}
+
+/**
+ * dc_validate_global_state() - Determine if hardware can support a given state
+ *
+ * @dc: dc struct for this driver
+ * @new_ctx: state to be validated
+ * @validate_mode: identify the validation mode
+ *
+ * Checks hardware resource availability and bandwidth requirement.
+ *
+ * Return:
+ * DC_OK if the result can be programmed. Otherwise, an error code.
+ */
+enum dc_status dc_validate_global_state(
+		struct dc *dc,
+		struct dc_state *new_ctx,
+		enum dc_validate_mode validate_mode)
+{
+	enum dc_status result = DC_ERROR_UNEXPECTED;
+	int i, j;
+
+	if (!new_ctx)
+		return DC_ERROR_UNEXPECTED;
+
+	if (dc->res_pool->funcs->validate_global) {
+		result = dc->res_pool->funcs->validate_global(dc, new_ctx);
+		if (result != DC_OK)
+			return result;
+	}
+
+	for (i = 0; i < new_ctx->stream_count; i++) {
+		struct dc_stream_state *stream = new_ctx->streams[i];
+
+		for (j = 0; j < dc->res_pool->pipe_count; j++) {
+			struct pipe_ctx *pipe_ctx = &new_ctx->res_ctx.pipe_ctx[j];
+
+			if (pipe_ctx->stream != stream)
+				continue;
+
+			/* Decide whether hblank borrow is needed and save it in pipe_ctx */
+			if (dc->debug.enable_hblank_borrow)
+				calculate_timing_params_for_dsc_with_padding(pipe_ctx);
+
+			if (dc->res_pool->funcs->patch_unknown_plane_state &&
+					pipe_ctx->plane_state &&
+					pipe_ctx->plane_state->tiling_info.gfx9.swizzle == DC_SW_UNKNOWN) {
+				result = dc->res_pool->funcs->patch_unknown_plane_state(pipe_ctx->plane_state);
+				if (result != DC_OK)
+					return result;
+			}
+
+			/* Switch to dp clock source only if there is
+			 * no non dp stream that shares the same timing
+			 * with the dp stream.
+			 */
+			if (dc_is_dp_signal(pipe_ctx->stream->signal) &&
+				!find_pll_sharable_stream(stream, new_ctx)) {
+
+				resource_unreference_clock_source(
+						&new_ctx->res_ctx,
+						dc->res_pool,
+						pipe_ctx->clock_source);
+
+				pipe_ctx->clock_source = dc->res_pool->dp_clock_source;
+				resource_reference_clock_source(
+						&new_ctx->res_ctx,
+						dc->res_pool,
+						 pipe_ctx->clock_source);
+			}
+		}
+	}
+
+	result = resource_build_scaling_params_for_context(dc, new_ctx);
+
+	if (result == DC_OK)
+		result = dc->res_pool->funcs->validate_bandwidth(dc, new_ctx, validate_mode);
+
+	return result;
+}
+
+static void patch_gamut_packet_checksum(
+		struct dc_info_packet *gamut_packet)
+{
+	/* For gamut we recalc checksum */
+	if (gamut_packet->valid) {
+		uint8_t chk_sum = 0;
+		uint8_t *ptr;
+		uint8_t i;
+
+		/*start of the Gamut data. */
+		ptr = &gamut_packet->sb[3];
+
+		for (i = 0; i <= gamut_packet->sb[1]; i++)
+			chk_sum += ptr[i];
+
+		gamut_packet->sb[2] = (uint8_t) (0x100 - chk_sum);
+	}
+}
+
+static void set_avi_info_frame(
+		struct dc_info_packet *info_packet,
+		struct pipe_ctx *pipe_ctx)
+{
+	struct dc_stream_state *stream = pipe_ctx->stream;
+	enum dc_color_space color_space = COLOR_SPACE_UNKNOWN;
+	uint32_t pixel_encoding = 0;
+	enum scanning_type scan_type = SCANNING_TYPE_NODATA;
+	enum dc_aspect_ratio aspect = ASPECT_RATIO_NO_DATA;
+	uint8_t *check_sum = NULL;
+	uint8_t byte_index = 0;
+	union hdmi_info_packet hdmi_info;
+	unsigned int vic = pipe_ctx->stream->timing.vic;
+	unsigned int rid = pipe_ctx->stream->timing.rid;
+	unsigned int fr_ind = pipe_ctx->stream->timing.fr_index;
+	enum dc_timing_3d_format format;
+
+	if (stream->avi_infopacket.valid) {
+		*info_packet = stream->avi_infopacket;
+		return;
+	}
+
+	memset(&hdmi_info, 0, sizeof(union hdmi_info_packet));
+
+
+	color_space = pipe_ctx->stream->output_color_space;
+	if (color_space == COLOR_SPACE_UNKNOWN)
+		color_space = (stream->timing.pixel_encoding == PIXEL_ENCODING_RGB) ?
+			COLOR_SPACE_SRGB:COLOR_SPACE_YCBCR709;
+
+	/* Initialize header */
+	hdmi_info.bits.header.info_frame_type = HDMI_INFOFRAME_TYPE_AVI;
+	/* InfoFrameVersion_3 is defined by CEA861F (Section 6.4), but shall
+	* not be used in HDMI 2.0 (Section 10.1) */
+	hdmi_info.bits.header.version = 2;
+	hdmi_info.bits.header.length = HDMI_AVI_INFOFRAME_SIZE;
+
+	/*
+	 * IDO-defined (Y2,Y1,Y0 = 1,1,1) shall not be used by devices built
+	 * according to HDMI 2.0 spec (Section 10.1)
+	 */
+
+	switch (stream->timing.pixel_encoding) {
+	case PIXEL_ENCODING_YCBCR422:
+		pixel_encoding = 1;
+		break;
+
+	case PIXEL_ENCODING_YCBCR444:
+		pixel_encoding = 2;
+		break;
+	case PIXEL_ENCODING_YCBCR420:
+		pixel_encoding = 3;
+		break;
+
+	case PIXEL_ENCODING_RGB:
+	default:
+		pixel_encoding = 0;
+	}
+
+	/* Y0_Y1_Y2 : The pixel encoding */
+	/* H14b AVI InfoFrame has extension on Y-field from 2 bits to 3 bits */
+	hdmi_info.bits.Y0_Y1_Y2 = pixel_encoding;
+
+	/* A0 = 1 Active Format Information valid */
+	hdmi_info.bits.A0 = ACTIVE_FORMAT_VALID;
+
+	/* B0, B1 = 3; Bar info data is valid */
+	hdmi_info.bits.B0_B1 = BAR_INFO_BOTH_VALID;
+
+	hdmi_info.bits.SC0_SC1 = PICTURE_SCALING_UNIFORM;
+
+	/* S0, S1 : Underscan / Overscan */
+	/* TODO: un-hardcode scan type */
+	scan_type = SCANNING_TYPE_UNDERSCAN;
+	hdmi_info.bits.S0_S1 = scan_type;
+
+	/* C0, C1 : Colorimetry */
+	switch (color_space) {
+	case COLOR_SPACE_YCBCR709:
+	case COLOR_SPACE_YCBCR709_LIMITED:
+		hdmi_info.bits.C0_C1 = COLORIMETRY_ITU709;
+		break;
+	case COLOR_SPACE_YCBCR601:
+	case COLOR_SPACE_YCBCR601_LIMITED:
+		hdmi_info.bits.C0_C1 = COLORIMETRY_ITU601;
+		break;
+	case COLOR_SPACE_2020_RGB_FULLRANGE:
+	case COLOR_SPACE_2020_RGB_LIMITEDRANGE:
+	case COLOR_SPACE_2020_YCBCR_LIMITED:
+		hdmi_info.bits.EC0_EC2 = COLORIMETRYEX_BT2020RGBYCBCR;
+		hdmi_info.bits.C0_C1   = COLORIMETRY_EXTENDED;
+		break;
+	case COLOR_SPACE_ADOBERGB:
+		hdmi_info.bits.EC0_EC2 = COLORIMETRYEX_ADOBERGB;
+		hdmi_info.bits.C0_C1   = COLORIMETRY_EXTENDED;
+		break;
+	case COLOR_SPACE_SRGB:
+	default:
+		hdmi_info.bits.C0_C1 = COLORIMETRY_NO_DATA;
+		break;
+	}
+
+	if (pixel_encoding && color_space == COLOR_SPACE_2020_YCBCR_LIMITED &&
+			stream->out_transfer_func.tf == TRANSFER_FUNCTION_GAMMA22) {
+		hdmi_info.bits.EC0_EC2 = 0;
+		hdmi_info.bits.C0_C1 = COLORIMETRY_ITU709;
+	}
+
+	/* TODO: un-hardcode aspect ratio */
+	aspect = stream->timing.aspect_ratio;
+
+	switch (aspect) {
+	case ASPECT_RATIO_4_3:
+	case ASPECT_RATIO_16_9:
+		hdmi_info.bits.M0_M1 = aspect;
+		break;
+
+	case ASPECT_RATIO_NO_DATA:
+	case ASPECT_RATIO_64_27:
+	case ASPECT_RATIO_256_135:
+	default:
+		hdmi_info.bits.M0_M1 = 0;
+	}
+
+	/* Active Format Aspect ratio - same as Picture Aspect Ratio. */
+	hdmi_info.bits.R0_R3 = ACTIVE_FORMAT_ASPECT_RATIO_SAME_AS_PICTURE;
+
+	switch (stream->content_type) {
+	case DISPLAY_CONTENT_TYPE_NO_DATA:
+		hdmi_info.bits.CN0_CN1 = 0;
+		hdmi_info.bits.ITC = 1;
+		break;
+	case DISPLAY_CONTENT_TYPE_GRAPHICS:
+		hdmi_info.bits.CN0_CN1 = 0;
+		hdmi_info.bits.ITC = 1;
+		break;
+	case DISPLAY_CONTENT_TYPE_PHOTO:
+		hdmi_info.bits.CN0_CN1 = 1;
+		hdmi_info.bits.ITC = 1;
+		break;
+	case DISPLAY_CONTENT_TYPE_CINEMA:
+		hdmi_info.bits.CN0_CN1 = 2;
+		hdmi_info.bits.ITC = 1;
+		break;
+	case DISPLAY_CONTENT_TYPE_GAME:
+		hdmi_info.bits.CN0_CN1 = 3;
+		hdmi_info.bits.ITC = 1;
+		break;
+	}
+
+	if (stream->qs_bit == 1) {
+		if (color_space == COLOR_SPACE_SRGB ||
+			color_space == COLOR_SPACE_2020_RGB_FULLRANGE)
+			hdmi_info.bits.Q0_Q1   = RGB_QUANTIZATION_FULL_RANGE;
+		else if (color_space == COLOR_SPACE_SRGB_LIMITED ||
+					color_space == COLOR_SPACE_2020_RGB_LIMITEDRANGE)
+			hdmi_info.bits.Q0_Q1   = RGB_QUANTIZATION_LIMITED_RANGE;
+		else
+			hdmi_info.bits.Q0_Q1   = RGB_QUANTIZATION_DEFAULT_RANGE;
+	} else
+		hdmi_info.bits.Q0_Q1   = RGB_QUANTIZATION_DEFAULT_RANGE;
+
+	/* TODO : We should handle YCC quantization */
+	/* but we do not have matrix calculation */
+	hdmi_info.bits.YQ0_YQ1 = YYC_QUANTIZATION_LIMITED_RANGE;
+
+	///VIC
+	if (pipe_ctx->stream->timing.hdmi_vic != 0)
+		vic = 0;
+	format = stream->timing.timing_3d_format;
+	/*todo, add 3DStereo support*/
+	if (format != TIMING_3D_FORMAT_NONE) {
+		// Based on HDMI specs hdmi vic needs to be converted to cea vic when 3D is enabled
+		switch (pipe_ctx->stream->timing.hdmi_vic) {
+		case 1:
+			vic = 95;
+			break;
+		case 2:
+			vic = 94;
+			break;
+		case 3:
+			vic = 93;
+			break;
+		case 4:
+			vic = 98;
+			break;
+		default:
+			break;
+		}
+	}
+	/* If VIC >= 128, the Source shall use AVI InfoFrame Version 3*/
+	hdmi_info.bits.VIC0_VIC7 = vic;
+	if (vic >= 128)
+		hdmi_info.bits.header.version = 3;
+	/* If (C1, C0)=(1, 1) and (EC2, EC1, EC0)=(1, 1, 1),
+	 * the Source shall use 20 AVI InfoFrame Version 4
+	 */
+	if (hdmi_info.bits.C0_C1 == COLORIMETRY_EXTENDED &&
+			hdmi_info.bits.EC0_EC2 == COLORIMETRYEX_RESERVED) {
+		hdmi_info.bits.header.version = 4;
+		hdmi_info.bits.header.length = 14;
+	}
+
+	if (rid != 0 && fr_ind != 0) {
+		hdmi_info.bits.header.version = 4;
+		hdmi_info.bits.header.length = 15;
+
+		hdmi_info.bits.FR0_FR3 = fr_ind & 0xF;
+		hdmi_info.bits.FR4 = (fr_ind >> 4) & 0x1;
+		hdmi_info.bits.RID0_RID5 = rid;
+	}
+
+	/* pixel repetition
+	 * PR0 - PR3 start from 0 whereas pHwPathMode->mode.timing.flags.pixel
+	 * repetition start from 1 */
+	hdmi_info.bits.PR0_PR3 = 0;
+
+	/* Bar Info
+	 * barTop:    Line Number of End of Top Bar.
+	 * barBottom: Line Number of Start of Bottom Bar.
+	 * barLeft:   Pixel Number of End of Left Bar.
+	 * barRight:  Pixel Number of Start of Right Bar. */
+	hdmi_info.bits.bar_top = stream->timing.v_border_top;
+	hdmi_info.bits.bar_bottom = (stream->timing.v_total
+			- stream->timing.v_border_bottom + 1);
+	hdmi_info.bits.bar_left  = stream->timing.h_border_left;
+	hdmi_info.bits.bar_right = (stream->timing.h_total
+			- stream->timing.h_border_right + 1);
+
+    /* Additional Colorimetry Extension
+     * Used in conduction with C0-C1 and EC0-EC2
+     * 0 = DCI-P3 RGB (D65)
+     * 1 = DCI-P3 RGB (theater)
+     */
+	hdmi_info.bits.ACE0_ACE3 = 0;
+
+	/* check_sum - Calculate AFMT_AVI_INFO0 ~ AFMT_AVI_INFO3 */
+	check_sum = &hdmi_info.packet_raw_data.sb[0];
+
+	*check_sum = HDMI_INFOFRAME_TYPE_AVI + hdmi_info.bits.header.length + hdmi_info.bits.header.version;
+
+	for (byte_index = 1; byte_index <= hdmi_info.bits.header.length; byte_index++)
+		*check_sum += hdmi_info.packet_raw_data.sb[byte_index];
+
+	/* one byte complement */
+	*check_sum = (uint8_t) (0x100 - *check_sum);
+
+	/* Store in hw_path_mode */
+	info_packet->hb0 = hdmi_info.packet_raw_data.hb0;
+	info_packet->hb1 = hdmi_info.packet_raw_data.hb1;
+	info_packet->hb2 = hdmi_info.packet_raw_data.hb2;
+
+	for (byte_index = 0; byte_index < sizeof(hdmi_info.packet_raw_data.sb); byte_index++)
+		info_packet->sb[byte_index] = hdmi_info.packet_raw_data.sb[byte_index];
+
+	info_packet->valid = true;
+}
+
+static void set_vendor_info_packet(
+		struct dc_info_packet *info_packet,
+		struct dc_stream_state *stream)
+{
+	/* SPD info packet for FreeSync */
+
+	/* Check if Freesync is supported. Return if false. If true,
+	 * set the corresponding bit in the info packet
+	 */
+	if (!stream->vsp_infopacket.valid)
+		return;
+
+	*info_packet = stream->vsp_infopacket;
+}
+
+static void set_spd_info_packet(
+		struct dc_info_packet *info_packet,
+		struct dc_stream_state *stream)
+{
+	/* SPD info packet for FreeSync */
+
+	/* Check if Freesync is supported. Return if false. If true,
+	 * set the corresponding bit in the info packet
+	 */
+	if (!stream->vrr_infopacket.valid)
+		return;
+
+	*info_packet = stream->vrr_infopacket;
+}
+
+static void set_hdr_static_info_packet(
+		struct dc_info_packet *info_packet,
+		struct dc_stream_state *stream)
+{
+	/* HDR Static Metadata info packet for HDR10 */
+
+	if (!stream->hdr_static_metadata.valid ||
+			stream->use_dynamic_meta)
+		return;
+
+	*info_packet = stream->hdr_static_metadata;
+}
+
+static void set_vsc_info_packet(
+		struct dc_info_packet *info_packet,
+		struct dc_stream_state *stream)
+{
+	if (!stream->vsc_infopacket.valid)
+		return;
+
+	*info_packet = stream->vsc_infopacket;
+}
+static void set_hfvs_info_packet(
+		struct dc_info_packet *info_packet,
+		struct dc_stream_state *stream)
+{
+	if (!stream->hfvsif_infopacket.valid)
+		return;
+
+	*info_packet = stream->hfvsif_infopacket;
+}
+
+static void adaptive_sync_override_dp_info_packets_sdp_line_num(
+		const struct dc_crtc_timing *timing,
+		struct enc_sdp_line_num *sdp_line_num,
+		unsigned int vstartup_start)
+{
+	uint32_t asic_blank_start = 0;
+	uint32_t asic_blank_end   = 0;
+	uint32_t v_update = 0;
+
+	const struct dc_crtc_timing *tg = timing;
+
+	/* blank_start = frame end - front porch */
+	asic_blank_start = tg->v_total - tg->v_front_porch;
+
+	/* blank_end = blank_start - active */
+	asic_blank_end = (asic_blank_start - tg->v_border_bottom -
+						tg->v_addressable - tg->v_border_top);
+
+	if (vstartup_start > asic_blank_end) {
+		v_update = (tg->v_total - (vstartup_start - asic_blank_end));
+		sdp_line_num->adaptive_sync_line_num_valid = true;
+		sdp_line_num->adaptive_sync_line_num = (tg->v_total - v_update - 1);
+	} else {
+		sdp_line_num->adaptive_sync_line_num_valid = false;
+		sdp_line_num->adaptive_sync_line_num = 0;
+	}
+}
+
+static void set_adaptive_sync_info_packet(
+		struct dc_info_packet *info_packet,
+		const struct dc_stream_state *stream,
+		struct encoder_info_frame *info_frame,
+		unsigned int vstartup_start)
+{
+	if (!stream->adaptive_sync_infopacket.valid)
+		return;
+
+	adaptive_sync_override_dp_info_packets_sdp_line_num(
+			&stream->timing,
+			&info_frame->sdp_line_num,
+			vstartup_start);
+
+	*info_packet = stream->adaptive_sync_infopacket;
+}
+
+static void set_vtem_info_packet(
+		struct dc_info_packet *info_packet,
+		struct dc_stream_state *stream)
+{
+	if (!stream->vtem_infopacket.valid)
+		return;
+
+	*info_packet = stream->vtem_infopacket;
+}
+
+struct clock_source *dc_resource_find_first_free_pll(
+		struct resource_context *res_ctx,
+		const struct resource_pool *pool)
+{
+	int i;
+
+	for (i = 0; i < pool->clk_src_count; ++i) {
+		if (res_ctx->clock_source_ref_count[i] == 0)
+			return pool->clock_sources[i];
+	}
+
+	return NULL;
+}
+
+void resource_build_info_frame(struct pipe_ctx *pipe_ctx)
+{
+	enum signal_type signal = SIGNAL_TYPE_NONE;
+	struct encoder_info_frame *info = &pipe_ctx->stream_res.encoder_info_frame;
+	unsigned int vstartup_start = 0;
+
+	/* default all packets to invalid */
+	info->avi.valid = false;
+	info->gamut.valid = false;
+	info->vendor.valid = false;
+	info->spd.valid = false;
+	info->hdrsmd.valid = false;
+	info->vsc.valid = false;
+	info->hfvsif.valid = false;
+	info->vtem.valid = false;
+	info->adaptive_sync.valid = false;
+	signal = pipe_ctx->stream->signal;
+
+	if (pipe_ctx->stream->ctx->dc->res_pool->funcs->get_vstartup_for_pipe)
+		vstartup_start = pipe_ctx->stream->ctx->dc->res_pool->funcs->get_vstartup_for_pipe(pipe_ctx);
+
+	/* HDMi and DP have different info packets*/
+	if (dc_is_hdmi_signal(signal)) {
+		set_avi_info_frame(&info->avi, pipe_ctx);
+
+		set_vendor_info_packet(&info->vendor, pipe_ctx->stream);
+		set_hfvs_info_packet(&info->hfvsif, pipe_ctx->stream);
+		set_vtem_info_packet(&info->vtem, pipe_ctx->stream);
+
+		set_spd_info_packet(&info->spd, pipe_ctx->stream);
+
+		set_hdr_static_info_packet(&info->hdrsmd, pipe_ctx->stream);
+
+	} else if (dc_is_dp_signal(signal)) {
+		set_vsc_info_packet(&info->vsc, pipe_ctx->stream);
+
+		set_spd_info_packet(&info->spd, pipe_ctx->stream);
+
+		set_hdr_static_info_packet(&info->hdrsmd, pipe_ctx->stream);
+		set_adaptive_sync_info_packet(&info->adaptive_sync,
+										pipe_ctx->stream,
+										info,
+										vstartup_start);
+	}
+
+	patch_gamut_packet_checksum(&info->gamut);
+}
+
+enum dc_status resource_map_clock_resources(
+		const struct dc  *dc,
+		struct dc_state *context,
+		struct dc_stream_state *stream)
+{
+	/* acquire new resources */
+	const struct resource_pool *pool = dc->res_pool;
+	struct pipe_ctx *pipe_ctx = resource_get_otg_master_for_stream(
+				&context->res_ctx, stream);
+
+	if (!pipe_ctx)
+		return DC_ERROR_UNEXPECTED;
+
+	if (dc_is_dp_signal(pipe_ctx->stream->signal)
+		|| pipe_ctx->stream->signal == SIGNAL_TYPE_VIRTUAL)
+		pipe_ctx->clock_source = pool->dp_clock_source;
+	else {
+		pipe_ctx->clock_source = NULL;
+
+		if (!dc->config.disable_disp_pll_sharing)
+			pipe_ctx->clock_source = resource_find_used_clk_src_for_sharing(
+				&context->res_ctx,
+				pipe_ctx);
+
+		if (pipe_ctx->clock_source == NULL)
+			pipe_ctx->clock_source =
+				dc_resource_find_first_free_pll(
+					&context->res_ctx,
+					pool);
+	}
+
+	if (pipe_ctx->clock_source == NULL)
+		return DC_NO_CLOCK_SOURCE_RESOURCE;
+
+	resource_reference_clock_source(
+		&context->res_ctx, pool,
+		pipe_ctx->clock_source);
+
+	return DC_OK;
+}
+
+/*
+ * Note: We need to disable output if clock sources change,
+ * since bios does optimization and doesn't apply if changing
+ * PHY when not already disabled.
+ */
+bool pipe_need_reprogram(
+		struct pipe_ctx *pipe_ctx_old,
+		struct pipe_ctx *pipe_ctx)
+{
+	if (!pipe_ctx_old->stream)
+		return false;
+
+	if (pipe_ctx_old->stream->sink != pipe_ctx->stream->sink)
+		return true;
+
+	if (pipe_ctx_old->stream->signal != pipe_ctx->stream->signal)
+		return true;
+
+	if (pipe_ctx_old->stream_res.audio != pipe_ctx->stream_res.audio)
+		return true;
+
+	if (pipe_ctx_old->clock_source != pipe_ctx->clock_source
+			&& pipe_ctx_old->stream != pipe_ctx->stream)
+		return true;
+
+	if (pipe_ctx_old->stream_res.stream_enc != pipe_ctx->stream_res.stream_enc)
+		return true;
+
+	if (dc_is_timing_changed(pipe_ctx_old->stream, pipe_ctx->stream))
+		return true;
+
+	if (pipe_ctx_old->stream->dpms_off != pipe_ctx->stream->dpms_off)
+		return true;
+
+	if (false == pipe_ctx_old->stream->link->link_state_valid &&
+		false == pipe_ctx_old->stream->dpms_off)
+		return true;
+
+	if (pipe_ctx_old->stream_res.dsc != pipe_ctx->stream_res.dsc)
+		return true;
+
+	if (pipe_ctx_old->stream_res.hpo_dp_stream_enc != pipe_ctx->stream_res.hpo_dp_stream_enc)
+		return true;
+	if (pipe_ctx_old->link_res.hpo_dp_link_enc != pipe_ctx->link_res.hpo_dp_link_enc)
+		return true;
+
+	/* DIG link encoder resource assignment for stream changed. */
+	if (pipe_ctx_old->stream->ctx->dc->config.unify_link_enc_assignment) {
+		if (pipe_ctx_old->link_res.dio_link_enc != pipe_ctx->link_res.dio_link_enc)
+			return true;
+	} else if (pipe_ctx_old->stream->ctx->dc->res_pool->funcs->link_encs_assign) {
+		bool need_reprogram = false;
+		struct dc *dc = pipe_ctx_old->stream->ctx->dc;
+		struct link_encoder *link_enc_prev =
+			link_enc_cfg_get_link_enc_used_by_stream_current(dc, pipe_ctx_old->stream);
+
+		if (link_enc_prev != pipe_ctx->stream->link_enc)
+			need_reprogram = true;
+
+		return need_reprogram;
+	}
+
+	return false;
+}
+
+void resource_build_bit_depth_reduction_params(struct dc_stream_state *stream,
+		struct bit_depth_reduction_params *fmt_bit_depth)
+{
+	enum dc_dither_option option = stream->dither_option;
+	enum dc_pixel_encoding pixel_encoding =
+			stream->timing.pixel_encoding;
+
+	memset(fmt_bit_depth, 0, sizeof(*fmt_bit_depth));
+
+	if (option == DITHER_OPTION_DEFAULT) {
+		switch (stream->timing.display_color_depth) {
+		case COLOR_DEPTH_666:
+			option = DITHER_OPTION_SPATIAL6;
+			break;
+		case COLOR_DEPTH_888:
+			option = DITHER_OPTION_SPATIAL8;
+			break;
+		case COLOR_DEPTH_101010:
+			option = DITHER_OPTION_TRUN10;
+			break;
+		default:
+			option = DITHER_OPTION_DISABLE;
+		}
+	}
+
+	if (option == DITHER_OPTION_DISABLE)
+		return;
+
+	if (option == DITHER_OPTION_TRUN6) {
+		fmt_bit_depth->flags.TRUNCATE_ENABLED = 1;
+		fmt_bit_depth->flags.TRUNCATE_DEPTH = 0;
+	} else if (option == DITHER_OPTION_TRUN8 ||
+			option == DITHER_OPTION_TRUN8_SPATIAL6 ||
+			option == DITHER_OPTION_TRUN8_FM6) {
+		fmt_bit_depth->flags.TRUNCATE_ENABLED = 1;
+		fmt_bit_depth->flags.TRUNCATE_DEPTH = 1;
+	} else if (option == DITHER_OPTION_TRUN10        ||
+			option == DITHER_OPTION_TRUN10_SPATIAL6   ||
+			option == DITHER_OPTION_TRUN10_SPATIAL8   ||
+			option == DITHER_OPTION_TRUN10_FM8     ||
+			option == DITHER_OPTION_TRUN10_FM6     ||
+			option == DITHER_OPTION_TRUN10_SPATIAL8_FM6) {
+		fmt_bit_depth->flags.TRUNCATE_ENABLED = 1;
+		fmt_bit_depth->flags.TRUNCATE_DEPTH = 2;
+		if (option == DITHER_OPTION_TRUN10)
+			fmt_bit_depth->flags.TRUNCATE_MODE = 1;
+	}
+
+	/* special case - Formatter can only reduce by 4 bits at most.
+	 * When reducing from 12 to 6 bits,
+	 * HW recommends we use trunc with round mode
+	 * (if we did nothing, trunc to 10 bits would be used)
+	 * note that any 12->10 bit reduction is ignored prior to DCE8,
+	 * as the input was 10 bits.
+	 */
+	if (option == DITHER_OPTION_SPATIAL6_FRAME_RANDOM ||
+			option == DITHER_OPTION_SPATIAL6 ||
+			option == DITHER_OPTION_FM6) {
+		fmt_bit_depth->flags.TRUNCATE_ENABLED = 1;
+		fmt_bit_depth->flags.TRUNCATE_DEPTH = 2;
+		fmt_bit_depth->flags.TRUNCATE_MODE = 1;
+	}
+
+	/* spatial dither
+	 * note that spatial modes 1-3 are never used
+	 */
+	if (option == DITHER_OPTION_SPATIAL6_FRAME_RANDOM            ||
+			option == DITHER_OPTION_SPATIAL6 ||
+			option == DITHER_OPTION_TRUN10_SPATIAL6      ||
+			option == DITHER_OPTION_TRUN8_SPATIAL6) {
+		fmt_bit_depth->flags.SPATIAL_DITHER_ENABLED = 1;
+		fmt_bit_depth->flags.SPATIAL_DITHER_DEPTH = 0;
+		fmt_bit_depth->flags.HIGHPASS_RANDOM = 1;
+		fmt_bit_depth->flags.RGB_RANDOM =
+				(pixel_encoding == PIXEL_ENCODING_RGB) ? 1 : 0;
+	} else if (option == DITHER_OPTION_SPATIAL8_FRAME_RANDOM            ||
+			option == DITHER_OPTION_SPATIAL8 ||
+			option == DITHER_OPTION_SPATIAL8_FM6        ||
+			option == DITHER_OPTION_TRUN10_SPATIAL8      ||
+			option == DITHER_OPTION_TRUN10_SPATIAL8_FM6) {
+		fmt_bit_depth->flags.SPATIAL_DITHER_ENABLED = 1;
+		fmt_bit_depth->flags.SPATIAL_DITHER_DEPTH = 1;
+		fmt_bit_depth->flags.HIGHPASS_RANDOM = 1;
+		fmt_bit_depth->flags.RGB_RANDOM =
+				(pixel_encoding == PIXEL_ENCODING_RGB) ? 1 : 0;
+	} else if (option == DITHER_OPTION_SPATIAL10_FRAME_RANDOM ||
+			option == DITHER_OPTION_SPATIAL10 ||
+			option == DITHER_OPTION_SPATIAL10_FM8 ||
+			option == DITHER_OPTION_SPATIAL10_FM6) {
+		fmt_bit_depth->flags.SPATIAL_DITHER_ENABLED = 1;
+		fmt_bit_depth->flags.SPATIAL_DITHER_DEPTH = 2;
+		fmt_bit_depth->flags.HIGHPASS_RANDOM = 1;
+		fmt_bit_depth->flags.RGB_RANDOM =
+				(pixel_encoding == PIXEL_ENCODING_RGB) ? 1 : 0;
+	}
+
+	if (option == DITHER_OPTION_SPATIAL6 ||
+			option == DITHER_OPTION_SPATIAL8 ||
+			option == DITHER_OPTION_SPATIAL10) {
+		fmt_bit_depth->flags.FRAME_RANDOM = 0;
+	} else {
+		fmt_bit_depth->flags.FRAME_RANDOM = 1;
+	}
+
+	//////////////////////
+	//// temporal dither
+	//////////////////////
+	if (option == DITHER_OPTION_FM6           ||
+			option == DITHER_OPTION_SPATIAL8_FM6     ||
+			option == DITHER_OPTION_SPATIAL10_FM6     ||
+			option == DITHER_OPTION_TRUN10_FM6     ||
+			option == DITHER_OPTION_TRUN8_FM6      ||
+			option == DITHER_OPTION_TRUN10_SPATIAL8_FM6) {
+		fmt_bit_depth->flags.FRAME_MODULATION_ENABLED = 1;
+		fmt_bit_depth->flags.FRAME_MODULATION_DEPTH = 0;
+	} else if (option == DITHER_OPTION_FM8        ||
+			option == DITHER_OPTION_SPATIAL10_FM8  ||
+			option == DITHER_OPTION_TRUN10_FM8) {
+		fmt_bit_depth->flags.FRAME_MODULATION_ENABLED = 1;
+		fmt_bit_depth->flags.FRAME_MODULATION_DEPTH = 1;
+	} else if (option == DITHER_OPTION_FM10) {
+		fmt_bit_depth->flags.FRAME_MODULATION_ENABLED = 1;
+		fmt_bit_depth->flags.FRAME_MODULATION_DEPTH = 2;
+	}
+
+	fmt_bit_depth->pixel_encoding = pixel_encoding;
+}
+
+enum dc_status dc_validate_stream(struct dc *dc, struct dc_stream_state *stream)
+{
+	if (dc == NULL || stream == NULL)
+		return DC_ERROR_UNEXPECTED;
+
+	struct dc_link *link = stream->link;
+	struct timing_generator *tg = dc->res_pool->timing_generators[0];
+	enum dc_status res = DC_OK;
+
+	calculate_phy_pix_clks(stream);
+
+	if (!tg->funcs->validate_timing(tg, &stream->timing))
+		res = DC_FAIL_CONTROLLER_VALIDATE;
+
+	if (res == DC_OK) {
+		if (link->ep_type == DISPLAY_ENDPOINT_PHY &&
+				!link->link_enc->funcs->validate_output_with_stream(
+						link->link_enc, stream))
+			res = DC_FAIL_ENC_VALIDATE;
+	}
+
+	/* TODO: validate audio ASIC caps, encoder */
+
+	if (res == DC_OK)
+		res = dc->link_srv->validate_mode_timing(stream,
+		      link,
+		      &stream->timing);
+
+	return res;
+}
+
+enum dc_status dc_validate_plane(struct dc *dc, const struct dc_plane_state *plane_state)
+{
+	enum dc_status res = DC_OK;
+
+	/* check if surface has invalid dimensions */
+	if (plane_state->src_rect.width == 0 || plane_state->src_rect.height == 0 ||
+		plane_state->dst_rect.width == 0 || plane_state->dst_rect.height == 0)
+		return DC_FAIL_SURFACE_VALIDATE;
+
+	/* TODO For now validates pixel format only */
+	if (dc->res_pool->funcs->validate_plane)
+		return dc->res_pool->funcs->validate_plane(plane_state, &dc->caps);
+
+	return res;
+}
+
+unsigned int resource_pixel_format_to_bpp(enum surface_pixel_format format)
+{
+	switch (format) {
+	case SURFACE_PIXEL_FORMAT_GRPH_PALETA_256_COLORS:
+		return 8;
+	case SURFACE_PIXEL_FORMAT_VIDEO_420_YCbCr:
+	case SURFACE_PIXEL_FORMAT_VIDEO_420_YCrCb:
+		return 12;
+	case SURFACE_PIXEL_FORMAT_GRPH_ARGB1555:
+	case SURFACE_PIXEL_FORMAT_GRPH_RGB565:
+	case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCbCr:
+	case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCrCb:
+		return 16;
+	case SURFACE_PIXEL_FORMAT_GRPH_ARGB8888:
+	case SURFACE_PIXEL_FORMAT_GRPH_ABGR8888:
+	case SURFACE_PIXEL_FORMAT_GRPH_ARGB2101010:
+	case SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010:
+	case SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010_XR_BIAS:
+	case SURFACE_PIXEL_FORMAT_GRPH_RGBE:
+	case SURFACE_PIXEL_FORMAT_GRPH_RGBE_ALPHA:
+		return 32;
+	case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616:
+	case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616:
+	case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616F:
+	case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F:
+		return 64;
+	default:
+		ASSERT_CRITICAL(false);
+		return -1;
+	}
+}
+static unsigned int get_max_audio_sample_rate(struct audio_mode *modes)
+{
+	if (modes) {
+		if (modes->sample_rates.rate.RATE_192)
+			return 192000;
+		if (modes->sample_rates.rate.RATE_176_4)
+			return 176400;
+		if (modes->sample_rates.rate.RATE_96)
+			return 96000;
+		if (modes->sample_rates.rate.RATE_88_2)
+			return 88200;
+		if (modes->sample_rates.rate.RATE_48)
+			return 48000;
+		if (modes->sample_rates.rate.RATE_44_1)
+			return 44100;
+		if (modes->sample_rates.rate.RATE_32)
+			return 32000;
+	}
+	/*original logic when no audio info*/
+	return 441000;
+}
+
+void get_audio_check(struct audio_info *aud_modes,
+	struct audio_check *audio_chk)
+{
+	unsigned int i;
+	unsigned int max_sample_rate = 0;
+
+	if (aud_modes) {
+		audio_chk->audio_packet_type = 0x2;/*audio sample packet AP = .25 for layout0, 1 for layout1*/
+
+		audio_chk->max_audiosample_rate = 0;
+		for (i = 0; i < aud_modes->mode_count; i++) {
+			max_sample_rate = get_max_audio_sample_rate(&aud_modes->modes[i]);
+			if (audio_chk->max_audiosample_rate < max_sample_rate)
+				audio_chk->max_audiosample_rate = max_sample_rate;
+			/*dts takes the same as type 2: AP = 0.25*/
+		}
+		/*check which one take more bandwidth*/
+		if (audio_chk->max_audiosample_rate > 192000)
+			audio_chk->audio_packet_type = 0x9;/*AP =1*/
+		audio_chk->acat = 0;/*not support*/
+	}
+}
+
+struct link_encoder *get_temp_dio_link_enc(
+		const struct resource_context *res_ctx,
+		const struct resource_pool *const pool,
+		const struct dc_link *link)
+{
+	struct link_encoder *link_enc = NULL;
+	int enc_index;
+
+	if (link->is_dig_mapping_flexible)
+		enc_index = find_acquired_dio_link_enc_for_link(res_ctx, link);
+	else
+		enc_index = link->eng_id;
+
+	if (enc_index < 0)
+		enc_index = find_free_dio_link_enc(res_ctx, link, pool);
+
+	if (enc_index >= 0)
+		link_enc = pool->link_encoders[enc_index];
+
+	return link_enc;
+}
+
+static struct hpo_dp_link_encoder *get_temp_hpo_dp_link_enc(
+		const struct resource_context *res_ctx,
+		const struct resource_pool *const pool,
+		const struct dc_link *link)
+{
+	struct hpo_dp_link_encoder *hpo_dp_link_enc = NULL;
+	int enc_index;
+
+	enc_index = find_acquired_hpo_dp_link_enc_for_link(res_ctx, link);
+
+	if (enc_index < 0)
+		enc_index = find_free_hpo_dp_link_enc(res_ctx, pool);
+
+	if (enc_index >= 0)
+		hpo_dp_link_enc = pool->hpo_dp_link_enc[enc_index];
+
+	return hpo_dp_link_enc;
+}
+
+bool get_temp_dp_link_res(struct dc_link *link,
+		struct link_resource *link_res,
+		struct dc_link_settings *link_settings)
+{
+	const struct dc *dc  = link->dc;
+	const struct resource_context *res_ctx = &dc->current_state->res_ctx;
+
+	memset(link_res, 0, sizeof(*link_res));
+
+	if (dc->link_srv->dp_get_encoding_format(link_settings) == DP_128b_132b_ENCODING) {
+		link_res->hpo_dp_link_enc = get_temp_hpo_dp_link_enc(res_ctx, dc->res_pool, link);
+		if (!link_res->hpo_dp_link_enc)
+			return false;
+	} else if (dc->link_srv->dp_get_encoding_format(link_settings) == DP_8b_10b_ENCODING &&
+				dc->config.unify_link_enc_assignment) {
+		link_res->dio_link_enc = get_temp_dio_link_enc(res_ctx,
+				dc->res_pool, link);
+		if (!link_res->dio_link_enc)
+			return false;
+	}
+
+	return true;
+}
+
+void reset_syncd_pipes_from_disabled_pipes(struct dc *dc,
+		struct dc_state *context)
+{
+	int i, j;
+	struct pipe_ctx *pipe_ctx_old, *pipe_ctx, *pipe_ctx_syncd;
+
+	/* If pipe backend is reset, need to reset pipe syncd status */
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		pipe_ctx_old =	&dc->current_state->res_ctx.pipe_ctx[i];
+		pipe_ctx = &context->res_ctx.pipe_ctx[i];
+
+		if (!resource_is_pipe_type(pipe_ctx_old, OTG_MASTER))
+			continue;
+
+		if (!pipe_ctx->stream ||
+				pipe_need_reprogram(pipe_ctx_old, pipe_ctx)) {
+
+			/* Reset all the syncd pipes from the disabled pipe */
+			for (j = 0; j < dc->res_pool->pipe_count; j++) {
+				pipe_ctx_syncd = &context->res_ctx.pipe_ctx[j];
+				if ((GET_PIPE_SYNCD_FROM_PIPE(pipe_ctx_syncd) == pipe_ctx_old->pipe_idx) ||
+					!IS_PIPE_SYNCD_VALID(pipe_ctx_syncd))
+					SET_PIPE_SYNCD_TO_PIPE(pipe_ctx_syncd, j);
+			}
+		}
+	}
+}
+
+void check_syncd_pipes_for_disabled_master_pipe(struct dc *dc,
+	struct dc_state *context,
+	uint8_t disabled_master_pipe_idx)
+{
+	int i;
+	struct pipe_ctx *pipe_ctx, *pipe_ctx_check;
+
+	pipe_ctx = &context->res_ctx.pipe_ctx[disabled_master_pipe_idx];
+	if ((GET_PIPE_SYNCD_FROM_PIPE(pipe_ctx) != disabled_master_pipe_idx) ||
+		!IS_PIPE_SYNCD_VALID(pipe_ctx))
+		SET_PIPE_SYNCD_TO_PIPE(pipe_ctx, disabled_master_pipe_idx);
+
+	/* for the pipe disabled, check if any slave pipe exists and assert */
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		pipe_ctx_check = &context->res_ctx.pipe_ctx[i];
+
+		if ((GET_PIPE_SYNCD_FROM_PIPE(pipe_ctx_check) == disabled_master_pipe_idx) &&
+		    IS_PIPE_SYNCD_VALID(pipe_ctx_check) && (i != disabled_master_pipe_idx)) {
+			struct pipe_ctx *first_pipe = pipe_ctx_check;
+
+			while (first_pipe->prev_odm_pipe)
+				first_pipe = first_pipe->prev_odm_pipe;
+			/* When ODM combine is enabled, this case is expected. If the disabled pipe
+			 * is part of the ODM tree, then we should not print an error.
+			 * */
+			if (first_pipe->pipe_idx == disabled_master_pipe_idx)
+				continue;
+
+			DC_ERR("DC: Failure: pipe_idx[%d] syncd with disabled master pipe_idx[%d]\n",
+				   i, disabled_master_pipe_idx);
+		}
+	}
+}
+
+void reset_sync_context_for_pipe(const struct dc *dc,
+	struct dc_state *context,
+	uint8_t pipe_idx)
+{
+	int i;
+	struct pipe_ctx *pipe_ctx_reset;
+
+	/* reset the otg sync context for the pipe and its slave pipes if any */
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		pipe_ctx_reset = &context->res_ctx.pipe_ctx[i];
+
+		if (((GET_PIPE_SYNCD_FROM_PIPE(pipe_ctx_reset) == pipe_idx) &&
+			IS_PIPE_SYNCD_VALID(pipe_ctx_reset)) || (i == pipe_idx))
+			SET_PIPE_SYNCD_TO_PIPE(pipe_ctx_reset, i);
+	}
+}
+
+uint8_t resource_transmitter_to_phy_idx(const struct dc *dc, enum transmitter transmitter)
+{
+	/* TODO - get transmitter to phy idx mapping from DMUB */
+	uint8_t phy_idx = transmitter - TRANSMITTER_UNIPHY_A;
+
+	if (dc->ctx->dce_version == DCN_VERSION_3_1 &&
+			dc->ctx->asic_id.hw_internal_rev == YELLOW_CARP_B0) {
+		switch (transmitter) {
+		case TRANSMITTER_UNIPHY_A:
+			phy_idx = 0;
+			break;
+		case TRANSMITTER_UNIPHY_B:
+			phy_idx = 1;
+			break;
+		case TRANSMITTER_UNIPHY_C:
+			phy_idx = 5;
+			break;
+		case TRANSMITTER_UNIPHY_D:
+			phy_idx = 6;
+			break;
+		case TRANSMITTER_UNIPHY_E:
+			phy_idx = 4;
+			break;
+		default:
+			phy_idx = 0;
+			break;
+		}
+	}
+
+	return phy_idx;
+}
+
+const struct link_hwss *get_link_hwss(const struct dc_link *link,
+		const struct link_resource *link_res)
+{
+	/* Link_hwss is only accessible by getter function instead of accessing
+	 * by pointers in dc with the intent to protect against breaking polymorphism.
+	 */
+	if (can_use_hpo_dp_link_hwss(link, link_res))
+		/* TODO: some assumes that if decided link settings is 128b/132b
+		 * channel coding format hpo_dp_link_enc should be used.
+		 * Others believe that if hpo_dp_link_enc is available in link
+		 * resource then hpo_dp_link_enc must be used. This bound between
+		 * hpo_dp_link_enc != NULL and decided link settings is loosely coupled
+		 * with a premise that both hpo_dp_link_enc pointer and decided link
+		 * settings are determined based on single policy function like
+		 * "decide_link_settings" from upper layer. This "convention"
+		 * cannot be maintained and enforced at current level.
+		 * Therefore a refactor is due so we can enforce a strong bound
+		 * between those two parameters at this level.
+		 *
+		 * To put it simple, we want to make enforcement at low level so that
+		 * we will not return link hwss if caller plans to do 8b/10b
+		 * with an hpo encoder. Or we can return a very dummy one that doesn't
+		 * do work for all functions
+		 */
+		return (requires_fixed_vs_pe_retimer_hpo_link_hwss(link) ?
+				get_hpo_fixed_vs_pe_retimer_dp_link_hwss() : get_hpo_dp_link_hwss());
+	else if (can_use_dpia_link_hwss(link, link_res))
+		return get_dpia_link_hwss();
+	else if (can_use_dio_link_hwss(link, link_res))
+		return (requires_fixed_vs_pe_retimer_dio_link_hwss(link)) ?
+				get_dio_fixed_vs_pe_retimer_link_hwss() : get_dio_link_hwss();
+	else
+		return get_virtual_link_hwss();
+}
+
+bool is_h_timing_divisible_by_2(struct dc_stream_state *stream)
+{
+	bool divisible = false;
+	uint16_t h_blank_start = 0;
+	uint16_t h_blank_end = 0;
+
+	if (stream) {
+		h_blank_start = stream->timing.h_total - stream->timing.h_front_porch;
+		h_blank_end = h_blank_start - stream->timing.h_addressable;
+
+		/* HTOTAL, Hblank start/end, and Hsync start/end all must be
+		 * divisible by 2 in order for the horizontal timing params
+		 * to be considered divisible by 2. Hsync start is always 0.
+		 */
+		divisible = (stream->timing.h_total % 2 == 0) &&
+				(h_blank_start % 2 == 0) &&
+				(h_blank_end % 2 == 0) &&
+				(stream->timing.h_sync_width % 2 == 0);
+	}
+	return divisible;
+}
+
+/* This interface is deprecated for new DCNs. It is replaced by the following
+ * new interfaces. These two interfaces encapsulate pipe selection priority
+ * with DCN specific minimum hardware transition optimization algorithm. With
+ * the new interfaces caller no longer needs to know the implementation detail
+ * of a pipe topology.
+ *
+ * resource_update_pipes_with_odm_slice_count
+ * resource_update_pipes_with_mpc_slice_count
+ *
+ */
+bool dc_resource_acquire_secondary_pipe_for_mpc_odm_legacy(
+		const struct dc *dc,
+		struct dc_state *state,
+		struct pipe_ctx *pri_pipe,
+		struct pipe_ctx *sec_pipe,
+		bool odm)
+{
+	int pipe_idx = sec_pipe->pipe_idx;
+	struct pipe_ctx *sec_top, *sec_bottom, *sec_next, *sec_prev;
+	const struct resource_pool *pool = dc->res_pool;
+
+	sec_top = sec_pipe->top_pipe;
+	sec_bottom = sec_pipe->bottom_pipe;
+	sec_next = sec_pipe->next_odm_pipe;
+	sec_prev = sec_pipe->prev_odm_pipe;
+
+	if (pri_pipe == NULL)
+		return false;
+
+	*sec_pipe = *pri_pipe;
+
+	sec_pipe->top_pipe = sec_top;
+	sec_pipe->bottom_pipe = sec_bottom;
+	sec_pipe->next_odm_pipe = sec_next;
+	sec_pipe->prev_odm_pipe = sec_prev;
+
+	sec_pipe->pipe_idx = pipe_idx;
+	sec_pipe->plane_res.mi = pool->mis[pipe_idx];
+	sec_pipe->plane_res.hubp = pool->hubps[pipe_idx];
+	sec_pipe->plane_res.ipp = pool->ipps[pipe_idx];
+	sec_pipe->plane_res.xfm = pool->transforms[pipe_idx];
+	sec_pipe->plane_res.dpp = pool->dpps[pipe_idx];
+	sec_pipe->plane_res.mpcc_inst = pool->dpps[pipe_idx]->inst;
+	sec_pipe->stream_res.dsc = NULL;
+	if (odm) {
+		if (!sec_pipe->top_pipe)
+			sec_pipe->stream_res.opp = pool->opps[pipe_idx];
+		else
+			sec_pipe->stream_res.opp = sec_pipe->top_pipe->stream_res.opp;
+		if (sec_pipe->stream->timing.flags.DSC == 1) {
+#if defined(CONFIG_DRM_AMD_DC_FP)
+			dcn20_acquire_dsc(dc, &state->res_ctx, &sec_pipe->stream_res.dsc, sec_pipe->stream_res.opp->inst);
+#endif
+			ASSERT(sec_pipe->stream_res.dsc);
+			if (sec_pipe->stream_res.dsc == NULL)
+				return false;
+		}
+#if defined(CONFIG_DRM_AMD_DC_FP)
+		dcn20_build_mapped_resource(dc, state, sec_pipe->stream);
+#endif
+	}
+
+	return true;
+}
+
+enum dc_status update_dp_encoder_resources_for_test_harness(const struct dc *dc,
+		struct dc_state *context,
+		struct pipe_ctx *pipe_ctx)
+{
+	if (dc->link_srv->dp_get_encoding_format(&pipe_ctx->link_config.dp_link_settings) == DP_128b_132b_ENCODING) {
+		if (pipe_ctx->stream_res.hpo_dp_stream_enc == NULL) {
+			pipe_ctx->stream_res.hpo_dp_stream_enc =
+					find_first_free_match_hpo_dp_stream_enc_for_link(
+							&context->res_ctx, dc->res_pool, pipe_ctx->stream);
+
+			if (!pipe_ctx->stream_res.hpo_dp_stream_enc)
+				return DC_NO_STREAM_ENC_RESOURCE;
+
+			update_hpo_dp_stream_engine_usage(
+					&context->res_ctx, dc->res_pool,
+					pipe_ctx->stream_res.hpo_dp_stream_enc,
+					true);
+		}
+
+		if (pipe_ctx->link_res.hpo_dp_link_enc == NULL) {
+			if (!add_hpo_dp_link_enc_to_ctx(&context->res_ctx, dc->res_pool, pipe_ctx, pipe_ctx->stream))
+				return DC_NO_LINK_ENC_RESOURCE;
+		}
+	} else {
+		if (pipe_ctx->stream_res.hpo_dp_stream_enc) {
+			update_hpo_dp_stream_engine_usage(
+					&context->res_ctx, dc->res_pool,
+					pipe_ctx->stream_res.hpo_dp_stream_enc,
+					false);
+			pipe_ctx->stream_res.hpo_dp_stream_enc = NULL;
+		}
+		if (pipe_ctx->link_res.hpo_dp_link_enc)
+			remove_hpo_dp_link_enc_from_ctx(&context->res_ctx, pipe_ctx, pipe_ctx->stream);
+	}
+
+	if (pipe_ctx->link_res.dio_link_enc == NULL && dc->config.unify_link_enc_assignment)
+		if (!add_dio_link_enc_to_ctx(dc, context, dc->res_pool, pipe_ctx, pipe_ctx->stream))
+			return DC_NO_LINK_ENC_RESOURCE;
+
+	return DC_OK;
+}
+
+struct dscl_prog_data *resource_get_dscl_prog_data(struct pipe_ctx *pipe_ctx)
+{
+	return &pipe_ctx->plane_res.scl_data.dscl_prog_data;
+}
+
+static bool resource_allocate_mcache(struct dc_state *context, const struct  dc_mcache_params *mcache_params)
+{
+	if (context->clk_mgr->ctx->dc->res_pool->funcs->program_mcache_pipe_config)
+		context->clk_mgr->ctx->dc->res_pool->funcs->program_mcache_pipe_config(context, mcache_params);
+
+	return true;
+}
+
+void resource_init_common_dml2_callbacks(struct dc *dc, struct dml2_configuration_options *dml2_options)
+{
+	dml2_options->callbacks.dc = dc;
+	dml2_options->callbacks.build_scaling_params = &resource_build_scaling_params;
+	dml2_options->callbacks.build_test_pattern_params = &resource_build_test_pattern_params;
+	dml2_options->callbacks.acquire_secondary_pipe_for_mpc_odm = &dc_resource_acquire_secondary_pipe_for_mpc_odm_legacy;
+	dml2_options->callbacks.update_pipes_for_stream_with_slice_count = &resource_update_pipes_for_stream_with_slice_count;
+	dml2_options->callbacks.update_pipes_for_plane_with_slice_count = &resource_update_pipes_for_plane_with_slice_count;
+	dml2_options->callbacks.get_mpc_slice_index = &resource_get_mpc_slice_index;
+	dml2_options->callbacks.get_mpc_slice_count = &resource_get_mpc_slice_count;
+	dml2_options->callbacks.get_odm_slice_index = &resource_get_odm_slice_index;
+	dml2_options->callbacks.get_odm_slice_count = &resource_get_odm_slice_count;
+	dml2_options->callbacks.get_opp_head = &resource_get_opp_head;
+	dml2_options->callbacks.get_otg_master_for_stream = &resource_get_otg_master_for_stream;
+	dml2_options->callbacks.get_opp_heads_for_otg_master = &resource_get_opp_heads_for_otg_master;
+	dml2_options->callbacks.get_dpp_pipes_for_plane = &resource_get_dpp_pipes_for_plane;
+	dml2_options->callbacks.get_stream_status = &dc_state_get_stream_status;
+	dml2_options->callbacks.get_stream_from_id = &dc_state_get_stream_from_id;
+	dml2_options->callbacks.get_max_flickerless_instant_vtotal_increase = &dc_stream_get_max_flickerless_instant_vtotal_increase;
+	dml2_options->callbacks.allocate_mcache = &resource_allocate_mcache;
+
+	dml2_options->svp_pstate.callbacks.dc = dc;
+	dml2_options->svp_pstate.callbacks.add_phantom_plane = &dc_state_add_phantom_plane;
+	dml2_options->svp_pstate.callbacks.add_phantom_stream = &dc_state_add_phantom_stream;
+	dml2_options->svp_pstate.callbacks.build_scaling_params = &resource_build_scaling_params;
+	dml2_options->svp_pstate.callbacks.create_phantom_plane = &dc_state_create_phantom_plane;
+	dml2_options->svp_pstate.callbacks.remove_phantom_plane = &dc_state_remove_phantom_plane;
+	dml2_options->svp_pstate.callbacks.remove_phantom_stream = &dc_state_remove_phantom_stream;
+	dml2_options->svp_pstate.callbacks.create_phantom_stream = &dc_state_create_phantom_stream;
+	dml2_options->svp_pstate.callbacks.release_phantom_plane = &dc_state_release_phantom_plane;
+	dml2_options->svp_pstate.callbacks.release_phantom_stream = &dc_state_release_phantom_stream;
+	dml2_options->svp_pstate.callbacks.get_pipe_subvp_type = &dc_state_get_pipe_subvp_type;
+	dml2_options->svp_pstate.callbacks.get_stream_subvp_type = &dc_state_get_stream_subvp_type;
+	dml2_options->svp_pstate.callbacks.get_paired_subvp_stream = &dc_state_get_paired_subvp_stream;
+	dml2_options->svp_pstate.callbacks.remove_phantom_streams_and_planes = &dc_state_remove_phantom_streams_and_planes;
+	dml2_options->svp_pstate.callbacks.release_phantom_streams_and_planes = &dc_state_release_phantom_streams_and_planes;
+}
+
+/* Returns number of DET segments allocated for a given OTG_MASTER pipe */
+int resource_calculate_det_for_stream(struct dc_state *state, struct pipe_ctx *otg_master)
+{
+	struct pipe_ctx *opp_heads[MAX_PIPES];
+	struct pipe_ctx *dpp_pipes[MAX_PIPES];
+
+	int dpp_count = 0;
+	int det_segments = 0;
+
+	if (!otg_master->stream)
+		return 0;
+
+	int slice_count = resource_get_opp_heads_for_otg_master(otg_master,
+			&state->res_ctx, opp_heads);
+
+	for (int slice_idx = 0; slice_idx < slice_count; slice_idx++) {
+		if (opp_heads[slice_idx]->plane_state) {
+			dpp_count = resource_get_dpp_pipes_for_opp_head(
+					opp_heads[slice_idx],
+					&state->res_ctx,
+					dpp_pipes);
+			for (int dpp_idx = 0; dpp_idx < dpp_count; dpp_idx++)
+				det_segments += dpp_pipes[dpp_idx]->hubp_regs.det_size;
+		}
+	}
+	return det_segments;
+}
+
+bool resource_is_hpo_acquired(struct dc_state *context)
+{
+	int i;
+
+	for (i = 0; i < MAX_HPO_DP2_ENCODERS; i++) {
+		if (context->res_ctx.is_hpo_dp_stream_enc_acquired[i]) {
+			return true;
+		}
+	}
+
+	return false;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_sink.c b/drivers/gpu/drm/amd/display/dc/core/dc_sink.c
new file mode 100644
index 000000000000..455fa5dd1420
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_sink.c
@@ -0,0 +1,100 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+#include "dm_helpers.h"
+#include "core_types.h"
+
+/*******************************************************************************
+ * Private functions
+ ******************************************************************************/
+
+static bool dc_sink_construct(struct dc_sink *sink, const struct dc_sink_init_data *init_params)
+{
+
+	struct dc_link *link = init_params->link;
+
+	if (!link)
+		return false;
+
+	sink->sink_signal = init_params->sink_signal;
+	sink->link = link;
+	sink->ctx = link->ctx;
+	sink->dongle_max_pix_clk = init_params->dongle_max_pix_clk;
+	sink->converter_disable_audio = init_params->converter_disable_audio;
+	sink->dc_container_id = NULL;
+	sink->sink_id = init_params->link->ctx->dc_sink_id_count;
+	// increment dc_sink_id_count because we don't want two sinks with same ID
+	// unless they are actually the same
+	init_params->link->ctx->dc_sink_id_count++;
+
+	return true;
+}
+
+/*******************************************************************************
+ * Public functions
+ ******************************************************************************/
+
+void dc_sink_retain(struct dc_sink *sink)
+{
+	kref_get(&sink->refcount);
+}
+
+static void dc_sink_free(struct kref *kref)
+{
+	struct dc_sink *sink = container_of(kref, struct dc_sink, refcount);
+	kfree(sink->dc_container_id);
+	kfree(sink);
+}
+
+void dc_sink_release(struct dc_sink *sink)
+{
+	kref_put(&sink->refcount, dc_sink_free);
+}
+
+struct dc_sink *dc_sink_create(const struct dc_sink_init_data *init_params)
+{
+	struct dc_sink *sink = kzalloc(sizeof(*sink), GFP_KERNEL);
+
+	if (NULL == sink)
+		goto alloc_fail;
+
+	if (false == dc_sink_construct(sink, init_params))
+		goto construct_fail;
+
+	kref_init(&sink->refcount);
+
+	return sink;
+
+construct_fail:
+	kfree(sink);
+
+alloc_fail:
+	return NULL;
+}
+
+/*******************************************************************************
+ * Protected functions - visible only inside of DC (not visible in DM)
+ ******************************************************************************/
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_stat.c b/drivers/gpu/drm/amd/display/dc/core/dc_stat.c
new file mode 100644
index 000000000000..f976ffd6d466
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_stat.c
@@ -0,0 +1,90 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ */
+
+#include "dc/dc_stat.h"
+#include "dmub/dmub_srv_stat.h"
+#include "dc_dmub_srv.h"
+
+/**
+ * DOC: DC STAT Interface
+ *
+ * These interfaces are called without acquiring DAL and DC locks.
+ * Hence, there is limitations on whese interfaces can access. Only
+ * variables exclusively defined for these interfaces can be modified.
+ */
+
+/**
+ *  dc_stat_get_dmub_notification
+ *
+ * Calls dmub layer to retrieve dmub notification
+ *
+ * @dc: dc structure
+ * @notify: dmub notification structure
+ *
+ * Returns
+ *     None
+ */
+void dc_stat_get_dmub_notification(const struct dc *dc, struct dmub_notification *notify)
+{
+	/**
+	 * This function is called without dal and dc locks, so
+	 * we shall not modify any dc, dc_dmub_srv or dmub variables
+	 * except variables exclusively accessed by this function
+	 */
+	struct dmub_srv *dmub = dc->ctx->dmub_srv->dmub;
+	enum dmub_status status;
+
+	status = dmub_srv_stat_get_notification(dmub, notify);
+	ASSERT(status == DMUB_STATUS_OK);
+
+	/* For HPD/HPD RX, convert dpia port index into link index */
+	if (notify->type == DMUB_NOTIFICATION_HPD ||
+	    notify->type == DMUB_NOTIFICATION_HPD_IRQ ||
+	    notify->type == DMUB_NOTIFICATION_AUX_REPLY ||
+	    notify->type == DMUB_NOTIFICATION_DPIA_NOTIFICATION ||
+	    notify->type == DMUB_NOTIFICATION_SET_CONFIG_REPLY) {
+		notify->link_index =
+			get_link_index_from_dpia_port_index(dc, notify->instance);
+	}
+}
+
+/**
+ * dc_stat_get_dmub_dataout
+ *
+ * Calls dmub layer to retrieve dmub gpint dataout
+ *
+ * @dc: dc structure
+ * @dataout: dmub gpint dataout
+ *
+ * Returns
+ *     None
+ */
+void dc_stat_get_dmub_dataout(const struct dc *dc, uint32_t *dataout)
+{
+	struct dmub_srv *dmub = dc->ctx->dmub_srv->dmub;
+	enum dmub_status status;
+
+	status = dmub_srv_get_gpint_dataout(dmub, dataout);
+	ASSERT(status == DMUB_STATUS_OK);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_state.c b/drivers/gpu/drm/amd/display/dc/core/dc_state.c
new file mode 100644
index 000000000000..c61300a7cb1c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_state.c
@@ -0,0 +1,1077 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+#include "dc_types.h"
+#include "core_types.h"
+#include "core_status.h"
+#include "dc_state.h"
+#include "dc_state_priv.h"
+#include "dc_stream_priv.h"
+#include "dc_plane_priv.h"
+
+#include "dm_services.h"
+#include "resource.h"
+#include "link_enc_cfg.h"
+
+#if defined(CONFIG_DRM_AMD_DC_FP)
+#include "dml2/dml2_wrapper.h"
+#include "dml2/dml2_internal_types.h"
+#endif
+
+#define DC_LOGGER \
+	dc->ctx->logger
+#define DC_LOGGER_INIT(logger)
+
+/* Private dc_state helper functions */
+static bool dc_state_track_phantom_stream(struct dc_state *state,
+		struct dc_stream_state *phantom_stream)
+{
+	if (state->phantom_stream_count >= MAX_PHANTOM_PIPES)
+		return false;
+
+	state->phantom_streams[state->phantom_stream_count++] = phantom_stream;
+
+	return true;
+}
+
+static bool dc_state_untrack_phantom_stream(struct dc_state *state, struct dc_stream_state *phantom_stream)
+{
+	bool res = false;
+	int i;
+
+	/* first find phantom stream in the dc_state */
+	for (i = 0; i < state->phantom_stream_count; i++) {
+		if (state->phantom_streams[i] == phantom_stream) {
+			state->phantom_streams[i] = NULL;
+			res = true;
+			break;
+		}
+	}
+
+	/* failed to find stream in state */
+	if (!res)
+		return res;
+
+	/* trim back phantom streams */
+	state->phantom_stream_count--;
+	for (; i < state->phantom_stream_count; i++)
+		state->phantom_streams[i] = state->phantom_streams[i + 1];
+
+	return res;
+}
+
+static bool dc_state_is_phantom_stream_tracked(struct dc_state *state, struct dc_stream_state *phantom_stream)
+{
+	int i;
+
+	for (i = 0; i < state->phantom_stream_count; i++) {
+		if (state->phantom_streams[i] == phantom_stream)
+			return true;
+	}
+
+	return false;
+}
+
+static bool dc_state_track_phantom_plane(struct dc_state *state,
+		struct dc_plane_state *phantom_plane)
+{
+	if (state->phantom_plane_count >= MAX_PHANTOM_PIPES)
+		return false;
+
+	state->phantom_planes[state->phantom_plane_count++] = phantom_plane;
+
+	return true;
+}
+
+static bool dc_state_untrack_phantom_plane(struct dc_state *state, struct dc_plane_state *phantom_plane)
+{
+	bool res = false;
+	int i;
+
+	/* first find phantom plane in the dc_state */
+	for (i = 0; i < state->phantom_plane_count; i++) {
+		if (state->phantom_planes[i] == phantom_plane) {
+			state->phantom_planes[i] = NULL;
+			res = true;
+			break;
+		}
+	}
+
+	/* failed to find plane in state */
+	if (!res)
+		return res;
+
+	/* trim back phantom planes */
+	state->phantom_plane_count--;
+	for (; i < state->phantom_plane_count; i++)
+		state->phantom_planes[i] = state->phantom_planes[i + 1];
+
+	return res;
+}
+
+static bool dc_state_is_phantom_plane_tracked(struct dc_state *state, struct dc_plane_state *phantom_plane)
+{
+	int i;
+
+	for (i = 0; i < state->phantom_plane_count; i++) {
+		if (state->phantom_planes[i] == phantom_plane)
+			return true;
+	}
+
+	return false;
+}
+
+static void dc_state_copy_internal(struct dc_state *dst_state, struct dc_state *src_state)
+{
+	int i, j;
+
+	memcpy(dst_state, src_state, sizeof(struct dc_state));
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct pipe_ctx *cur_pipe = &dst_state->res_ctx.pipe_ctx[i];
+
+		if (cur_pipe->top_pipe)
+			cur_pipe->top_pipe =  &dst_state->res_ctx.pipe_ctx[cur_pipe->top_pipe->pipe_idx];
+
+		if (cur_pipe->bottom_pipe)
+			cur_pipe->bottom_pipe = &dst_state->res_ctx.pipe_ctx[cur_pipe->bottom_pipe->pipe_idx];
+
+		if (cur_pipe->prev_odm_pipe)
+			cur_pipe->prev_odm_pipe =  &dst_state->res_ctx.pipe_ctx[cur_pipe->prev_odm_pipe->pipe_idx];
+
+		if (cur_pipe->next_odm_pipe)
+			cur_pipe->next_odm_pipe = &dst_state->res_ctx.pipe_ctx[cur_pipe->next_odm_pipe->pipe_idx];
+	}
+
+	/* retain phantoms */
+	for (i = 0; i < dst_state->phantom_stream_count; i++)
+		dc_stream_retain(dst_state->phantom_streams[i]);
+
+	for (i = 0; i < dst_state->phantom_plane_count; i++)
+		dc_plane_state_retain(dst_state->phantom_planes[i]);
+
+	/* retain streams and planes */
+	for (i = 0; i < dst_state->stream_count; i++) {
+		dc_stream_retain(dst_state->streams[i]);
+		for (j = 0; j < dst_state->stream_status[i].plane_count; j++)
+			dc_plane_state_retain(
+					dst_state->stream_status[i].plane_states[j]);
+	}
+
+}
+
+static void init_state(struct dc *dc, struct dc_state *state)
+{
+	/* Each context must have their own instance of VBA and in order to
+	 * initialize and obtain IP and SOC the base DML instance from DC is
+	 * initially copied into every context
+	 */
+	memcpy(&state->bw_ctx.dml, &dc->dml, sizeof(struct display_mode_lib));
+}
+
+/* Public dc_state functions */
+struct dc_state *dc_state_create(struct dc *dc, struct dc_state_create_params *params)
+{
+	struct dc_state *state;
+
+	state = kvzalloc(sizeof(struct dc_state), GFP_KERNEL);
+
+	if (!state)
+		return NULL;
+
+	init_state(dc, state);
+	dc_state_construct(dc, state);
+	state->power_source = params ? params->power_source : DC_POWER_SOURCE_AC;
+
+#ifdef CONFIG_DRM_AMD_DC_FP
+	if (dc->debug.using_dml2) {
+		if (!dml2_create(dc, &dc->dml2_options, &state->bw_ctx.dml2)) {
+			dc_state_release(state);
+			return NULL;
+		}
+
+		if (dc->caps.dcmode_power_limits_present && !dml2_create(dc, &dc->dml2_dc_power_options, &state->bw_ctx.dml2_dc_power_source)) {
+			dc_state_release(state);
+			return NULL;
+		}
+	}
+#endif
+
+	kref_init(&state->refcount);
+
+	return state;
+}
+
+void dc_state_copy(struct dc_state *dst_state, struct dc_state *src_state)
+{
+	struct kref refcount = dst_state->refcount;
+#ifdef CONFIG_DRM_AMD_DC_FP
+	struct dml2_context *dst_dml2 = dst_state->bw_ctx.dml2;
+	struct dml2_context *dst_dml2_dc_power_source = dst_state->bw_ctx.dml2_dc_power_source;
+#endif
+
+	dc_state_copy_internal(dst_state, src_state);
+
+#ifdef CONFIG_DRM_AMD_DC_FP
+	dst_state->bw_ctx.dml2 = dst_dml2;
+	if (src_state->bw_ctx.dml2)
+		dml2_copy(dst_state->bw_ctx.dml2, src_state->bw_ctx.dml2);
+
+	dst_state->bw_ctx.dml2_dc_power_source = dst_dml2_dc_power_source;
+	if (src_state->bw_ctx.dml2_dc_power_source)
+		dml2_copy(dst_state->bw_ctx.dml2_dc_power_source, src_state->bw_ctx.dml2_dc_power_source);
+#endif
+
+	/* context refcount should not be overridden */
+	dst_state->refcount = refcount;
+}
+
+struct dc_state *dc_state_create_copy(struct dc_state *src_state)
+{
+	struct dc_state *new_state;
+
+	new_state = kvmalloc(sizeof(struct dc_state),
+			GFP_KERNEL);
+	if (!new_state)
+		return NULL;
+
+	dc_state_copy_internal(new_state, src_state);
+
+#ifdef CONFIG_DRM_AMD_DC_FP
+	new_state->bw_ctx.dml2 = NULL;
+	new_state->bw_ctx.dml2_dc_power_source = NULL;
+
+	if (src_state->bw_ctx.dml2 &&
+			!dml2_create_copy(&new_state->bw_ctx.dml2, src_state->bw_ctx.dml2)) {
+		dc_state_release(new_state);
+		return NULL;
+	}
+
+	if (src_state->bw_ctx.dml2_dc_power_source &&
+			!dml2_create_copy(&new_state->bw_ctx.dml2_dc_power_source, src_state->bw_ctx.dml2_dc_power_source)) {
+		dc_state_release(new_state);
+		return NULL;
+	}
+#endif
+
+	kref_init(&new_state->refcount);
+
+	return new_state;
+}
+
+void dc_state_copy_current(struct dc *dc, struct dc_state *dst_state)
+{
+	dc_state_copy(dst_state, dc->current_state);
+}
+
+struct dc_state *dc_state_create_current_copy(struct dc *dc)
+{
+	return dc_state_create_copy(dc->current_state);
+}
+
+void dc_state_construct(struct dc *dc, struct dc_state *state)
+{
+	state->clk_mgr = dc->clk_mgr;
+
+	/* Initialise DIG link encoder resource tracking variables. */
+	if (dc->res_pool)
+		link_enc_cfg_init(dc, state);
+}
+
+void dc_state_destruct(struct dc_state *state)
+{
+	int i, j;
+
+	for (i = 0; i < state->stream_count; i++) {
+		for (j = 0; j < state->stream_status[i].plane_count; j++)
+			dc_plane_state_release(
+					state->stream_status[i].plane_states[j]);
+
+		state->stream_status[i].plane_count = 0;
+		dc_stream_release(state->streams[i]);
+		state->streams[i] = NULL;
+	}
+	state->stream_count = 0;
+
+	/* release tracked phantoms */
+	for (i = 0; i < state->phantom_stream_count; i++) {
+		dc_stream_release(state->phantom_streams[i]);
+		state->phantom_streams[i] = NULL;
+	}
+	state->phantom_stream_count = 0;
+
+	for (i = 0; i < state->phantom_plane_count; i++) {
+		dc_plane_state_release(state->phantom_planes[i]);
+		state->phantom_planes[i] = NULL;
+	}
+	state->phantom_plane_count = 0;
+
+	state->stream_mask = 0;
+	memset(&state->res_ctx, 0, sizeof(state->res_ctx));
+	memset(&state->pp_display_cfg, 0, sizeof(state->pp_display_cfg));
+	memset(&state->dcn_bw_vars, 0, sizeof(state->dcn_bw_vars));
+	state->clk_mgr = NULL;
+	memset(&state->bw_ctx.bw, 0, sizeof(state->bw_ctx.bw));
+	memset(state->block_sequence, 0, sizeof(state->block_sequence));
+	state->block_sequence_steps = 0;
+	memset(state->dc_dmub_cmd, 0, sizeof(state->dc_dmub_cmd));
+	state->dmub_cmd_count = 0;
+	memset(&state->perf_params, 0, sizeof(state->perf_params));
+}
+
+void dc_state_retain(struct dc_state *state)
+{
+	kref_get(&state->refcount);
+}
+
+static void dc_state_free(struct kref *kref)
+{
+	struct dc_state *state = container_of(kref, struct dc_state, refcount);
+
+	dc_state_destruct(state);
+
+#ifdef CONFIG_DRM_AMD_DC_FP
+	dml2_destroy(state->bw_ctx.dml2);
+	state->bw_ctx.dml2 = 0;
+
+	dml2_destroy(state->bw_ctx.dml2_dc_power_source);
+	state->bw_ctx.dml2_dc_power_source = 0;
+#endif
+
+	kvfree(state);
+}
+
+void dc_state_release(struct dc_state *state)
+{
+	if (state != NULL)
+		kref_put(&state->refcount, dc_state_free);
+}
+/*
+ * dc_state_add_stream() - Add a new dc_stream_state to a dc_state.
+ */
+enum dc_status dc_state_add_stream(
+		const struct dc *dc,
+		struct dc_state *state,
+		struct dc_stream_state *stream)
+{
+	enum dc_status res;
+
+	DC_LOGGER_INIT(dc->ctx->logger);
+
+	if (state->stream_count >= dc->res_pool->timing_generator_count) {
+		DC_LOG_WARNING("Max streams reached, can't add stream %p !\n", stream);
+		return DC_ERROR_UNEXPECTED;
+	}
+
+	state->streams[state->stream_count] = stream;
+	dc_stream_retain(stream);
+	state->stream_count++;
+
+	res = resource_add_otg_master_for_stream_output(
+			state, dc->res_pool, stream);
+	if (res != DC_OK)
+		DC_LOG_WARNING("Adding stream %p to context failed with err %d!\n", stream, res);
+
+	return res;
+}
+
+/*
+ * dc_state_remove_stream() - Remove a stream from a dc_state.
+ */
+enum dc_status dc_state_remove_stream(
+		const struct dc *dc,
+		struct dc_state *state,
+		struct dc_stream_state *stream)
+{
+	int i;
+	struct pipe_ctx *del_pipe = resource_get_otg_master_for_stream(
+			&state->res_ctx, stream);
+
+	if (!del_pipe) {
+		dm_error("Pipe not found for stream %p !\n", stream);
+		return DC_ERROR_UNEXPECTED;
+	}
+
+	resource_update_pipes_for_stream_with_slice_count(state,
+			dc->current_state, dc->res_pool, stream, 1);
+	resource_remove_otg_master_for_stream_output(
+			state, dc->res_pool, stream);
+
+	for (i = 0; i < state->stream_count; i++)
+		if (state->streams[i] == stream)
+			break;
+
+	if (state->streams[i] != stream) {
+		dm_error("Context doesn't have stream %p !\n", stream);
+		return DC_ERROR_UNEXPECTED;
+	}
+
+	dc_stream_release_3dlut_for_stream(dc, stream);
+
+	dc_stream_release(state->streams[i]);
+	state->stream_count--;
+
+	/* Trim back arrays */
+	for (; i < state->stream_count; i++) {
+		state->streams[i] = state->streams[i + 1];
+		state->stream_status[i] = state->stream_status[i + 1];
+	}
+
+	state->streams[state->stream_count] = NULL;
+	memset(
+			&state->stream_status[state->stream_count],
+			0,
+			sizeof(state->stream_status[0]));
+
+	return DC_OK;
+}
+
+static void remove_mpc_combine_for_stream(const struct dc *dc,
+		struct dc_state *new_ctx,
+		const struct dc_state *cur_ctx,
+		struct dc_stream_status *status)
+{
+	int i;
+
+	for (i = 0; i < status->plane_count; i++)
+		resource_update_pipes_for_plane_with_slice_count(
+				new_ctx, cur_ctx, dc->res_pool,
+				status->plane_states[i], 1);
+}
+
+bool dc_state_add_plane(
+		const struct dc *dc,
+		struct dc_stream_state *stream,
+		struct dc_plane_state *plane_state,
+		struct dc_state *state)
+{
+	struct resource_pool *pool = dc->res_pool;
+	struct pipe_ctx *otg_master_pipe;
+	struct dc_stream_status *stream_status = NULL;
+	bool added = false;
+	int odm_slice_count;
+	int i;
+
+	stream_status = dc_state_get_stream_status(state, stream);
+	otg_master_pipe = resource_get_otg_master_for_stream(
+			&state->res_ctx, stream);
+	if (stream_status == NULL) {
+		dm_error("Existing stream not found; failed to attach surface!\n");
+		goto out;
+	} else if (stream_status->plane_count == MAX_SURFACES) {
+		dm_error("Surface: can not attach plane_state %p! Maximum is: %d\n",
+				plane_state, MAX_SURFACES);
+		goto out;
+	} else if (!otg_master_pipe) {
+		goto out;
+	}
+
+	added = resource_append_dpp_pipes_for_plane_composition(state,
+			dc->current_state, pool, otg_master_pipe, plane_state);
+
+	if (!added) {
+		/* try to remove MPC combine to free up pipes */
+		for (i = 0; i < state->stream_count; i++)
+			remove_mpc_combine_for_stream(dc, state,
+					dc->current_state,
+					&state->stream_status[i]);
+		added = resource_append_dpp_pipes_for_plane_composition(state,
+					dc->current_state, pool,
+					otg_master_pipe, plane_state);
+	}
+
+	if (!added) {
+		/* try to decrease ODM slice count gradually to free up pipes */
+		odm_slice_count = resource_get_odm_slice_count(otg_master_pipe);
+		for (i = odm_slice_count - 1; i > 0; i--) {
+			resource_update_pipes_for_stream_with_slice_count(state,
+					dc->current_state, dc->res_pool, stream,
+					i);
+			added = resource_append_dpp_pipes_for_plane_composition(
+					state,
+					dc->current_state, pool,
+					otg_master_pipe, plane_state);
+			if (added)
+				break;
+		}
+	}
+
+	if (added) {
+		stream_status->plane_states[stream_status->plane_count] =
+				plane_state;
+		stream_status->plane_count++;
+		dc_plane_state_retain(plane_state);
+	}
+
+out:
+	return added;
+}
+
+bool dc_state_remove_plane(
+		const struct dc *dc,
+		struct dc_stream_state *stream,
+		struct dc_plane_state *plane_state,
+		struct dc_state *state)
+{
+	int i;
+	struct dc_stream_status *stream_status = NULL;
+	struct resource_pool *pool = dc->res_pool;
+
+	if (!plane_state)
+		return true;
+
+	for (i = 0; i < state->stream_count; i++)
+		if (state->streams[i] == stream) {
+			stream_status = &state->stream_status[i];
+			break;
+		}
+
+	if (stream_status == NULL) {
+		dm_error("Existing stream not found; failed to remove plane.\n");
+		return false;
+	}
+
+	resource_remove_dpp_pipes_for_plane_composition(
+			state, pool, plane_state);
+
+	for (i = 0; i < stream_status->plane_count; i++) {
+		if (stream_status->plane_states[i] == plane_state) {
+			dc_plane_state_release(stream_status->plane_states[i]);
+			break;
+		}
+	}
+
+	if (i == stream_status->plane_count) {
+		dm_error("Existing plane_state not found; failed to detach it!\n");
+		return false;
+	}
+
+	stream_status->plane_count--;
+
+	/* Start at the plane we've just released, and move all the planes one index forward to "trim" the array */
+	for (; i < stream_status->plane_count; i++)
+		stream_status->plane_states[i] = stream_status->plane_states[i + 1];
+
+	stream_status->plane_states[stream_status->plane_count] = NULL;
+
+	return true;
+}
+
+/**
+ * dc_state_rem_all_planes_for_stream - Remove planes attached to the target stream.
+ *
+ * @dc: Current dc state.
+ * @stream: Target stream, which we want to remove the attached plans.
+ * @state: context from which the planes are to be removed.
+ *
+ * Return:
+ * Return true if DC was able to remove all planes from the target
+ * stream, otherwise, return false.
+ */
+bool dc_state_rem_all_planes_for_stream(
+		const struct dc *dc,
+		struct dc_stream_state *stream,
+		struct dc_state *state)
+{
+	int i, old_plane_count;
+	struct dc_stream_status *stream_status = NULL;
+	struct dc_plane_state *del_planes[MAX_SURFACES] = { 0 };
+
+	for (i = 0; i < state->stream_count; i++)
+		if (state->streams[i] == stream) {
+			stream_status = &state->stream_status[i];
+			break;
+		}
+
+	if (stream_status == NULL) {
+		dm_error("Existing stream %p not found!\n", stream);
+		return false;
+	}
+
+	old_plane_count = stream_status->plane_count;
+
+	for (i = 0; i < old_plane_count; i++)
+		del_planes[i] = stream_status->plane_states[i];
+
+	for (i = 0; i < old_plane_count; i++)
+		if (!dc_state_remove_plane(dc, stream, del_planes[i], state))
+			return false;
+
+	return true;
+}
+
+bool dc_state_add_all_planes_for_stream(
+		const struct dc *dc,
+		struct dc_stream_state *stream,
+		struct dc_plane_state * const *plane_states,
+		int plane_count,
+		struct dc_state *state)
+{
+	int i;
+	bool result = true;
+
+	for (i = 0; i < plane_count; i++)
+		if (!dc_state_add_plane(dc, stream, plane_states[i], state)) {
+			result = false;
+			break;
+		}
+
+	return result;
+}
+
+/* Private dc_state functions */
+
+/**
+ * dc_state_get_stream_status - Get stream status from given dc state
+ * @state: DC state to find the stream status in
+ * @stream: The stream to get the stream status for
+ *
+ * The given stream is expected to exist in the given dc state. Otherwise, NULL
+ * will be returned.
+ */
+struct dc_stream_status *dc_state_get_stream_status(
+		struct dc_state *state,
+		const struct dc_stream_state *stream)
+{
+	uint8_t i;
+
+	if (state == NULL)
+		return NULL;
+
+	for (i = 0; i < state->stream_count; i++) {
+		if (stream == state->streams[i])
+			return &state->stream_status[i];
+	}
+
+	return NULL;
+}
+
+enum mall_stream_type dc_state_get_pipe_subvp_type(const struct dc_state *state,
+		const struct pipe_ctx *pipe_ctx)
+{
+	return dc_state_get_stream_subvp_type(state, pipe_ctx->stream);
+}
+
+enum mall_stream_type dc_state_get_stream_subvp_type(const struct dc_state *state,
+		const struct dc_stream_state *stream)
+{
+	int i;
+
+	enum mall_stream_type type = SUBVP_NONE;
+
+	for (i = 0; i < state->stream_count; i++) {
+		if (state->streams[i] == stream) {
+			type = state->stream_status[i].mall_stream_config.type;
+			break;
+		}
+	}
+
+	return type;
+}
+
+struct dc_stream_state *dc_state_get_paired_subvp_stream(const struct dc_state *state,
+		const struct dc_stream_state *stream)
+{
+	int i;
+
+	struct dc_stream_state *paired_stream = NULL;
+
+	for (i = 0; i < state->stream_count; i++) {
+		if (state->streams[i] == stream) {
+			paired_stream = state->stream_status[i].mall_stream_config.paired_stream;
+			break;
+		}
+	}
+
+	return paired_stream;
+}
+
+struct dc_stream_state *dc_state_create_phantom_stream(const struct dc *dc,
+		struct dc_state *state,
+		struct dc_stream_state *main_stream)
+{
+	struct dc_stream_state *phantom_stream;
+
+	DC_LOGGER_INIT(dc->ctx->logger);
+
+	phantom_stream = dc_create_stream_for_sink(main_stream->sink);
+
+	if (!phantom_stream) {
+		DC_LOG_ERROR("Failed to allocate phantom stream.\n");
+		return NULL;
+	}
+
+	/* track phantom stream in dc_state */
+	dc_state_track_phantom_stream(state, phantom_stream);
+
+	phantom_stream->is_phantom = true;
+	phantom_stream->signal = SIGNAL_TYPE_VIRTUAL;
+	phantom_stream->dpms_off = true;
+
+	return phantom_stream;
+}
+
+void dc_state_release_phantom_stream(const struct dc *dc,
+		struct dc_state *state,
+		struct dc_stream_state *phantom_stream)
+{
+	DC_LOGGER_INIT(dc->ctx->logger);
+
+	if (!dc_state_untrack_phantom_stream(state, phantom_stream)) {
+		DC_LOG_ERROR("Failed to free phantom stream %p in dc state %p.\n", phantom_stream, state);
+		return;
+	}
+
+	dc_stream_release(phantom_stream);
+}
+
+struct dc_plane_state *dc_state_create_phantom_plane(const struct dc *dc,
+		struct dc_state *state,
+		struct dc_plane_state *main_plane)
+{
+	struct dc_plane_state *phantom_plane = dc_create_plane_state(dc);
+
+	DC_LOGGER_INIT(dc->ctx->logger);
+
+	if (!phantom_plane) {
+		DC_LOG_ERROR("Failed to allocate phantom plane.\n");
+		return NULL;
+	}
+
+	/* track phantom inside dc_state */
+	dc_state_track_phantom_plane(state, phantom_plane);
+
+	phantom_plane->is_phantom = true;
+
+	return phantom_plane;
+}
+
+void dc_state_release_phantom_plane(const struct dc *dc,
+		struct dc_state *state,
+		struct dc_plane_state *phantom_plane)
+{
+	DC_LOGGER_INIT(dc->ctx->logger);
+
+	if (!dc_state_untrack_phantom_plane(state, phantom_plane)) {
+		DC_LOG_ERROR("Failed to free phantom plane %p in dc state %p.\n", phantom_plane, state);
+		return;
+	}
+
+	dc_plane_state_release(phantom_plane);
+}
+
+/* add phantom streams to context and generate correct meta inside dc_state */
+enum dc_status dc_state_add_phantom_stream(const struct dc *dc,
+		struct dc_state *state,
+		struct dc_stream_state *phantom_stream,
+		struct dc_stream_state *main_stream)
+{
+	struct dc_stream_status *main_stream_status;
+	struct dc_stream_status *phantom_stream_status;
+	enum dc_status res = dc_state_add_stream(dc, state, phantom_stream);
+
+	/* check if stream is tracked */
+	if (res == DC_OK && !dc_state_is_phantom_stream_tracked(state, phantom_stream)) {
+		/* stream must be tracked if added to state */
+		dc_state_track_phantom_stream(state, phantom_stream);
+	}
+
+	/* setup subvp meta */
+	main_stream_status = dc_state_get_stream_status(state, main_stream);
+	if (main_stream_status) {
+		main_stream_status->mall_stream_config.type = SUBVP_MAIN;
+		main_stream_status->mall_stream_config.paired_stream = phantom_stream;
+	}
+
+	phantom_stream_status = dc_state_get_stream_status(state, phantom_stream);
+	if (phantom_stream_status) {
+		phantom_stream_status->mall_stream_config.type = SUBVP_PHANTOM;
+		phantom_stream_status->mall_stream_config.paired_stream = main_stream;
+		phantom_stream_status->mall_stream_config.subvp_limit_cursor_size = false;
+		phantom_stream_status->mall_stream_config.cursor_size_limit_subvp = false;
+	}
+
+	dc_state_set_stream_subvp_cursor_limit(main_stream, state, true);
+
+	return res;
+}
+
+enum dc_status dc_state_remove_phantom_stream(const struct dc *dc,
+		struct dc_state *state,
+		struct dc_stream_state *phantom_stream)
+{
+	struct dc_stream_status *main_stream_status = NULL;
+	struct dc_stream_status *phantom_stream_status;
+
+	/* reset subvp meta */
+	phantom_stream_status = dc_state_get_stream_status(state, phantom_stream);
+	if (phantom_stream_status) {
+		main_stream_status = dc_state_get_stream_status(state, phantom_stream_status->mall_stream_config.paired_stream);
+		phantom_stream_status->mall_stream_config.type = SUBVP_NONE;
+		phantom_stream_status->mall_stream_config.paired_stream = NULL;
+	}
+
+	if (main_stream_status) {
+		main_stream_status->mall_stream_config.type = SUBVP_NONE;
+		main_stream_status->mall_stream_config.paired_stream = NULL;
+	}
+
+	/* remove stream from state */
+	return dc_state_remove_stream(dc, state, phantom_stream);
+}
+
+bool dc_state_add_phantom_plane(
+		const struct dc *dc,
+		struct dc_stream_state *phantom_stream,
+		struct dc_plane_state *phantom_plane,
+		struct dc_state *state)
+{
+	bool res = dc_state_add_plane(dc, phantom_stream, phantom_plane, state);
+
+	/* check if stream is tracked */
+	if (res && !dc_state_is_phantom_plane_tracked(state, phantom_plane)) {
+		/* stream must be tracked if added to state */
+		dc_state_track_phantom_plane(state, phantom_plane);
+	}
+
+	return res;
+}
+
+bool dc_state_remove_phantom_plane(
+		const struct dc *dc,
+		struct dc_stream_state *phantom_stream,
+		struct dc_plane_state *phantom_plane,
+		struct dc_state *state)
+{
+	return dc_state_remove_plane(dc, phantom_stream, phantom_plane, state);
+}
+
+bool dc_state_rem_all_phantom_planes_for_stream(
+		const struct dc *dc,
+		struct dc_stream_state *phantom_stream,
+		struct dc_state *state,
+		bool should_release_planes)
+{
+	int i, old_plane_count;
+	struct dc_stream_status *stream_status = NULL;
+	struct dc_plane_state *del_planes[MAX_SURFACES] = { 0 };
+
+	for (i = 0; i < state->stream_count; i++)
+		if (state->streams[i] == phantom_stream) {
+			stream_status = &state->stream_status[i];
+			break;
+		}
+
+	if (stream_status == NULL) {
+		dm_error("Existing stream %p not found!\n", phantom_stream);
+		return false;
+	}
+
+	old_plane_count = stream_status->plane_count;
+
+	for (i = 0; i < old_plane_count; i++)
+		del_planes[i] = stream_status->plane_states[i];
+
+	for (i = 0; i < old_plane_count; i++) {
+		if (!dc_state_remove_plane(dc, phantom_stream, del_planes[i], state))
+			return false;
+		if (should_release_planes)
+			dc_state_release_phantom_plane(dc, state, del_planes[i]);
+	}
+
+	return true;
+}
+
+bool dc_state_add_all_phantom_planes_for_stream(
+		const struct dc *dc,
+		struct dc_stream_state *phantom_stream,
+		struct dc_plane_state * const *phantom_planes,
+		int plane_count,
+		struct dc_state *state)
+{
+	return dc_state_add_all_planes_for_stream(dc, phantom_stream, phantom_planes, plane_count, state);
+}
+
+bool dc_state_remove_phantom_streams_and_planes(
+	const struct dc *dc,
+	struct dc_state *state)
+{
+	int i;
+	bool removed_phantom = false;
+	struct dc_stream_state *phantom_stream = NULL;
+
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *pipe = &state->res_ctx.pipe_ctx[i];
+
+		if (pipe->plane_state && pipe->stream && dc_state_get_pipe_subvp_type(state, pipe) == SUBVP_PHANTOM) {
+			phantom_stream = pipe->stream;
+
+			dc_state_rem_all_phantom_planes_for_stream(dc, phantom_stream, state, false);
+			dc_state_remove_phantom_stream(dc, state, phantom_stream);
+			removed_phantom = true;
+		}
+	}
+	return removed_phantom;
+}
+
+void dc_state_release_phantom_streams_and_planes(
+		const struct dc *dc,
+		struct dc_state *state)
+{
+	unsigned int phantom_count;
+	struct dc_stream_state *phantom_streams[MAX_PHANTOM_PIPES];
+	struct dc_plane_state *phantom_planes[MAX_PHANTOM_PIPES];
+	int i;
+
+	phantom_count = state->phantom_stream_count;
+	memcpy(phantom_streams, state->phantom_streams, sizeof(struct dc_stream_state *) * MAX_PHANTOM_PIPES);
+	for (i = 0; i < phantom_count; i++)
+		dc_state_release_phantom_stream(dc, state, phantom_streams[i]);
+
+	phantom_count = state->phantom_plane_count;
+	memcpy(phantom_planes, state->phantom_planes, sizeof(struct dc_plane_state *) * MAX_PHANTOM_PIPES);
+	for (i = 0; i < phantom_count; i++)
+		dc_state_release_phantom_plane(dc, state, phantom_planes[i]);
+}
+
+struct dc_stream_state *dc_state_get_stream_from_id(const struct dc_state *state, unsigned int id)
+{
+	struct dc_stream_state *stream = NULL;
+	int i;
+
+	for (i = 0; i < state->stream_count; i++) {
+		if (state->streams[i] && state->streams[i]->stream_id == id) {
+			stream = state->streams[i];
+			break;
+		}
+	}
+
+	return stream;
+}
+
+bool dc_state_is_fams2_in_use(
+		const struct dc *dc,
+		const struct dc_state *state)
+{
+	bool is_fams2_in_use = false;
+
+	if (state)
+		is_fams2_in_use |= state->bw_ctx.bw.dcn.fams2_global_config.features.bits.enable;
+
+	if (dc->current_state)
+		is_fams2_in_use |= dc->current_state->bw_ctx.bw.dcn.fams2_global_config.features.bits.enable;
+
+	return is_fams2_in_use;
+}
+
+void dc_state_set_stream_subvp_cursor_limit(const struct dc_stream_state *stream,
+		struct dc_state *state,
+		bool limit)
+{
+	struct dc_stream_status *stream_status;
+
+	stream_status = dc_state_get_stream_status(state, stream);
+
+	if (stream_status) {
+		stream_status->mall_stream_config.subvp_limit_cursor_size = limit;
+	}
+}
+
+bool dc_state_get_stream_subvp_cursor_limit(const struct dc_stream_state *stream,
+		struct dc_state *state)
+{
+	bool limit = false;
+
+	struct dc_stream_status *stream_status;
+
+	stream_status = dc_state_get_stream_status(state, stream);
+
+	if (stream_status) {
+		limit = stream_status->mall_stream_config.subvp_limit_cursor_size;
+	}
+
+	return limit;
+}
+
+void dc_state_set_stream_cursor_subvp_limit(const struct dc_stream_state *stream,
+		struct dc_state *state,
+		bool limit)
+{
+	struct dc_stream_status *stream_status;
+
+	stream_status = dc_state_get_stream_status(state, stream);
+
+	if (stream_status) {
+		stream_status->mall_stream_config.cursor_size_limit_subvp = limit;
+	}
+}
+
+bool dc_state_get_stream_cursor_subvp_limit(const struct dc_stream_state *stream,
+		struct dc_state *state)
+{
+	bool limit = false;
+
+	struct dc_stream_status *stream_status;
+
+	stream_status = dc_state_get_stream_status(state, stream);
+
+	if (stream_status) {
+		limit = stream_status->mall_stream_config.cursor_size_limit_subvp;
+	}
+
+	return limit;
+}
+
+bool dc_state_can_clear_stream_cursor_subvp_limit(const struct dc_stream_state *stream,
+		struct dc_state *state)
+{
+	bool can_clear_limit = false;
+
+	struct dc_stream_status *stream_status;
+
+	stream_status = dc_state_get_stream_status(state, stream);
+
+	if (stream_status) {
+		can_clear_limit = dc_state_get_stream_cursor_subvp_limit(stream, state) &&
+				(stream_status->mall_stream_config.type == SUBVP_PHANTOM ||
+				stream->hw_cursor_req ||
+				!stream_status->mall_stream_config.subvp_limit_cursor_size ||
+				!stream->cursor_position.enable ||
+				dc_stream_check_cursor_attributes(stream, state, &stream->cursor_attributes));
+	}
+
+	return can_clear_limit;
+}
+
+bool dc_state_is_subvp_in_use(struct dc_state *state)
+{
+	uint32_t i;
+
+	for (i = 0; i < state->stream_count; i++) {
+		if (dc_state_get_stream_subvp_type(state, state->streams[i]) != SUBVP_NONE)
+			return true;
+	}
+
+	return false;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_stream.c b/drivers/gpu/drm/amd/display/dc/core/dc_stream.c
new file mode 100644
index 000000000000..9ac2d41f8fca
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_stream.c
@@ -0,0 +1,1232 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+#include "basics/dc_common.h"
+#include "dc.h"
+#include "core_types.h"
+#include "resource.h"
+#include "ipp.h"
+#include "timing_generator.h"
+#include "dc_dmub_srv.h"
+#include "dc_state_priv.h"
+#include "dc_stream_priv.h"
+
+#define DC_LOGGER dc->ctx->logger
+#ifndef MIN
+#define MIN(X, Y) ((X) < (Y) ? (X) : (Y))
+#endif
+#ifndef MAX
+#define MAX(x, y) ((x > y) ? x : y)
+#endif
+
+/*******************************************************************************
+ * Private functions
+ ******************************************************************************/
+void update_stream_signal(struct dc_stream_state *stream, struct dc_sink *sink)
+{
+	if (sink->sink_signal == SIGNAL_TYPE_NONE)
+		stream->signal = stream->link->connector_signal;
+	else
+		stream->signal = sink->sink_signal;
+
+	if (dc_is_dvi_signal(stream->signal)) {
+		if (stream->ctx->dc->caps.dual_link_dvi &&
+			(stream->timing.pix_clk_100hz / 10) > TMDS_MAX_PIXEL_CLOCK &&
+			sink->sink_signal != SIGNAL_TYPE_DVI_SINGLE_LINK)
+			stream->signal = SIGNAL_TYPE_DVI_DUAL_LINK;
+		else
+			stream->signal = SIGNAL_TYPE_DVI_SINGLE_LINK;
+	}
+}
+
+bool dc_stream_construct(struct dc_stream_state *stream,
+	struct dc_sink *dc_sink_data)
+{
+	uint32_t i = 0;
+
+	stream->sink = dc_sink_data;
+	dc_sink_retain(dc_sink_data);
+
+	stream->ctx = dc_sink_data->ctx;
+	stream->link = dc_sink_data->link;
+	stream->sink_patches = dc_sink_data->edid_caps.panel_patch;
+	stream->converter_disable_audio = dc_sink_data->converter_disable_audio;
+	stream->qs_bit = dc_sink_data->edid_caps.qs_bit;
+	stream->qy_bit = dc_sink_data->edid_caps.qy_bit;
+
+	/* Copy audio modes */
+	/* TODO - Remove this translation */
+	for (i = 0; i < (dc_sink_data->edid_caps.audio_mode_count); i++) {
+		stream->audio_info.modes[i].channel_count = dc_sink_data->edid_caps.audio_modes[i].channel_count;
+		stream->audio_info.modes[i].format_code = dc_sink_data->edid_caps.audio_modes[i].format_code;
+		stream->audio_info.modes[i].sample_rates.all = dc_sink_data->edid_caps.audio_modes[i].sample_rate;
+		stream->audio_info.modes[i].sample_size = dc_sink_data->edid_caps.audio_modes[i].sample_size;
+	}
+	stream->audio_info.mode_count = dc_sink_data->edid_caps.audio_mode_count;
+	stream->audio_info.audio_latency = dc_sink_data->edid_caps.audio_latency;
+	stream->audio_info.video_latency = dc_sink_data->edid_caps.video_latency;
+	memmove(
+		stream->audio_info.display_name,
+		dc_sink_data->edid_caps.display_name,
+		AUDIO_INFO_DISPLAY_NAME_SIZE_IN_CHARS);
+	stream->audio_info.manufacture_id = dc_sink_data->edid_caps.manufacturer_id;
+	stream->audio_info.product_id = dc_sink_data->edid_caps.product_id;
+	stream->audio_info.flags.all = dc_sink_data->edid_caps.speaker_flags;
+
+	if (dc_sink_data->dc_container_id != NULL) {
+		struct dc_container_id *dc_container_id = dc_sink_data->dc_container_id;
+
+		stream->audio_info.port_id[0] = dc_container_id->portId[0];
+		stream->audio_info.port_id[1] = dc_container_id->portId[1];
+	} else {
+		/* TODO - WindowDM has implemented,
+		other DMs need Unhardcode port_id */
+		stream->audio_info.port_id[0] = 0x5558859e;
+		stream->audio_info.port_id[1] = 0xd989449;
+	}
+
+	/* EDID CAP translation for HDMI 2.0 */
+	stream->timing.flags.LTE_340MCSC_SCRAMBLE = dc_sink_data->edid_caps.lte_340mcsc_scramble;
+
+	memset(&stream->timing.dsc_cfg, 0, sizeof(stream->timing.dsc_cfg));
+	stream->timing.dsc_cfg.num_slices_h = 0;
+	stream->timing.dsc_cfg.num_slices_v = 0;
+	stream->timing.dsc_cfg.bits_per_pixel = 128;
+	stream->timing.dsc_cfg.block_pred_enable = 1;
+	stream->timing.dsc_cfg.linebuf_depth = 9;
+	stream->timing.dsc_cfg.version_minor = 2;
+	stream->timing.dsc_cfg.ycbcr422_simple = 0;
+
+	update_stream_signal(stream, dc_sink_data);
+
+	stream->out_transfer_func.type = TF_TYPE_BYPASS;
+
+	dc_stream_assign_stream_id(stream);
+
+	return true;
+}
+
+void dc_stream_destruct(struct dc_stream_state *stream)
+{
+	dc_sink_release(stream->sink);
+}
+
+void dc_stream_assign_stream_id(struct dc_stream_state *stream)
+{
+	/* MSB is reserved to indicate phantoms */
+	stream->stream_id = stream->ctx->dc_stream_id_count;
+	stream->ctx->dc_stream_id_count++;
+}
+
+void dc_stream_retain(struct dc_stream_state *stream)
+{
+	kref_get(&stream->refcount);
+}
+
+static void dc_stream_free(struct kref *kref)
+{
+	struct dc_stream_state *stream = container_of(kref, struct dc_stream_state, refcount);
+
+	dc_stream_destruct(stream);
+	kfree(stream);
+}
+
+void dc_stream_release(struct dc_stream_state *stream)
+{
+	if (stream != NULL) {
+		kref_put(&stream->refcount, dc_stream_free);
+	}
+}
+
+struct dc_stream_state *dc_create_stream_for_sink(
+		struct dc_sink *sink)
+{
+	struct dc_stream_state *stream;
+
+	if (sink == NULL)
+		return NULL;
+
+	stream = kzalloc(sizeof(struct dc_stream_state), GFP_KERNEL);
+	if (stream == NULL)
+		goto alloc_fail;
+
+	if (dc_stream_construct(stream, sink) == false)
+		goto construct_fail;
+
+	kref_init(&stream->refcount);
+
+	return stream;
+
+construct_fail:
+	kfree(stream);
+
+alloc_fail:
+	return NULL;
+}
+
+struct dc_stream_state *dc_copy_stream(const struct dc_stream_state *stream)
+{
+	struct dc_stream_state *new_stream;
+
+	new_stream = kmemdup(stream, sizeof(struct dc_stream_state), GFP_KERNEL);
+	if (!new_stream)
+		return NULL;
+
+	if (new_stream->sink)
+		dc_sink_retain(new_stream->sink);
+
+	dc_stream_assign_stream_id(new_stream);
+
+	/* If using dynamic encoder assignment, wait till stream committed to assign encoder. */
+	if (new_stream->ctx->dc->res_pool->funcs->link_encs_assign &&
+			!new_stream->ctx->dc->config.unify_link_enc_assignment)
+		new_stream->link_enc = NULL;
+
+	kref_init(&new_stream->refcount);
+
+	return new_stream;
+}
+
+/**
+ * dc_stream_get_status() - Get current stream status of the given stream state
+ * @stream: The stream to get the stream status for.
+ *
+ * The given stream is expected to exist in dc->current_state. Otherwise, NULL
+ * will be returned.
+ */
+struct dc_stream_status *dc_stream_get_status(
+	struct dc_stream_state *stream)
+{
+	struct dc *dc = stream->ctx->dc;
+	return dc_state_get_stream_status(dc->current_state, stream);
+}
+
+void program_cursor_attributes(
+	struct dc *dc,
+	struct dc_stream_state *stream)
+{
+	int i;
+	struct resource_context *res_ctx;
+	struct pipe_ctx *pipe_to_program = NULL;
+
+	if (!stream)
+		return;
+
+	res_ctx = &dc->current_state->res_ctx;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct pipe_ctx *pipe_ctx = &res_ctx->pipe_ctx[i];
+
+		if (pipe_ctx->stream != stream)
+			continue;
+
+		if (!pipe_to_program) {
+			pipe_to_program = pipe_ctx;
+			dc->hwss.cursor_lock(dc, pipe_to_program, true);
+			if (pipe_to_program->next_odm_pipe)
+				dc->hwss.cursor_lock(dc, pipe_to_program->next_odm_pipe, true);
+		}
+
+		dc->hwss.set_cursor_attribute(pipe_ctx);
+		if (dc->ctx->dmub_srv)
+			dc_send_update_cursor_info_to_dmu(pipe_ctx, i);
+		if (dc->hwss.set_cursor_sdr_white_level)
+			dc->hwss.set_cursor_sdr_white_level(pipe_ctx);
+	}
+
+	if (pipe_to_program) {
+		dc->hwss.cursor_lock(dc, pipe_to_program, false);
+		if (pipe_to_program->next_odm_pipe)
+			dc->hwss.cursor_lock(dc, pipe_to_program->next_odm_pipe, false);
+	}
+}
+
+/*
+ * dc_stream_check_cursor_attributes() - Check validitity of cursor attributes and surface address
+ */
+bool dc_stream_check_cursor_attributes(
+	const struct dc_stream_state *stream,
+	struct dc_state *state,
+	const struct dc_cursor_attributes *attributes)
+{
+	const struct dc *dc;
+
+	unsigned int max_cursor_size;
+
+	if (NULL == stream) {
+		dm_error("DC: dc_stream is NULL!\n");
+		return false;
+	}
+	if (NULL == attributes) {
+		dm_error("DC: attributes is NULL!\n");
+		return false;
+	}
+
+	if (attributes->address.quad_part == 0) {
+		dm_output_to_console("DC: Cursor address is 0!\n");
+		return false;
+	}
+
+	dc = stream->ctx->dc;
+
+	/* SubVP is not compatible with HW cursor larger than what can fit in cursor SRAM.
+	 * Therefore, if cursor is greater than this, fallback to SW cursor.
+	 */
+	if (dc->debug.allow_sw_cursor_fallback && dc->res_pool->funcs->get_max_hw_cursor_size) {
+		max_cursor_size = dc->res_pool->funcs->get_max_hw_cursor_size(dc, state, stream);
+		max_cursor_size = max_cursor_size * max_cursor_size * 4;
+
+		if (attributes->height * attributes->width * 4 > max_cursor_size) {
+			return false;
+		}
+	}
+
+	return true;
+}
+
+/*
+ * dc_stream_set_cursor_attributes() - Update cursor attributes and set cursor surface address
+ */
+bool dc_stream_set_cursor_attributes(
+	struct dc_stream_state *stream,
+	const struct dc_cursor_attributes *attributes)
+{
+	bool result = false;
+
+	if (!stream)
+		return false;
+
+	if (dc_stream_check_cursor_attributes(stream, stream->ctx->dc->current_state, attributes)) {
+		stream->cursor_attributes = *attributes;
+		result = true;
+	}
+
+	return result;
+}
+
+bool dc_stream_program_cursor_attributes(
+	struct dc_stream_state *stream,
+	const struct dc_cursor_attributes *attributes)
+{
+	struct dc  *dc;
+	bool reset_idle_optimizations = false;
+
+	if (!stream)
+		return false;
+
+	dc = stream->ctx->dc;
+
+	if (dc_stream_set_cursor_attributes(stream, attributes)) {
+		dc_z10_restore(dc);
+		/* disable idle optimizations while updating cursor */
+		if (dc->idle_optimizations_allowed) {
+			dc_allow_idle_optimizations(dc, false);
+			reset_idle_optimizations = true;
+		}
+
+		program_cursor_attributes(dc, stream);
+
+		/* re-enable idle optimizations if necessary */
+		if (reset_idle_optimizations && !dc->debug.disable_dmub_reallow_idle)
+			dc_allow_idle_optimizations(dc, true);
+
+		return true;
+	}
+
+	return false;
+}
+
+void program_cursor_position(
+	struct dc *dc,
+	struct dc_stream_state *stream)
+{
+	int i;
+	struct resource_context *res_ctx;
+	struct pipe_ctx *pipe_to_program = NULL;
+
+	if (!stream)
+		return;
+
+	res_ctx = &dc->current_state->res_ctx;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct pipe_ctx *pipe_ctx = &res_ctx->pipe_ctx[i];
+
+		if (pipe_ctx->stream != stream ||
+				(!pipe_ctx->plane_res.mi  && !pipe_ctx->plane_res.hubp) ||
+				!pipe_ctx->plane_state ||
+				(!pipe_ctx->plane_res.xfm && !pipe_ctx->plane_res.dpp) ||
+				(!pipe_ctx->plane_res.ipp && !pipe_ctx->plane_res.dpp))
+			continue;
+
+		if (!pipe_to_program) {
+			pipe_to_program = pipe_ctx;
+			dc->hwss.cursor_lock(dc, pipe_to_program, true);
+		}
+
+		dc->hwss.set_cursor_position(pipe_ctx);
+		if (dc->ctx->dmub_srv)
+			dc_send_update_cursor_info_to_dmu(pipe_ctx, i);
+	}
+
+	if (pipe_to_program)
+		dc->hwss.cursor_lock(dc, pipe_to_program, false);
+}
+
+bool dc_stream_set_cursor_position(
+	struct dc_stream_state *stream,
+	const struct dc_cursor_position *position)
+{
+	if (NULL == stream) {
+		dm_error("DC: dc_stream is NULL!\n");
+		return false;
+	}
+
+	if (NULL == position) {
+		dm_error("DC: cursor position is NULL!\n");
+		return false;
+	}
+
+	stream->cursor_position = *position;
+
+
+	return true;
+}
+
+bool dc_stream_program_cursor_position(
+	struct dc_stream_state *stream,
+	const struct dc_cursor_position *position)
+{
+	struct dc *dc;
+	bool reset_idle_optimizations = false;
+	const struct dc_cursor_position *old_position;
+
+	if (!stream)
+		return false;
+
+	old_position = &stream->cursor_position;
+	dc = stream->ctx->dc;
+
+	if (dc_stream_set_cursor_position(stream, position)) {
+		dc_z10_restore(dc);
+
+		/* disable idle optimizations if enabling cursor */
+		if (dc->idle_optimizations_allowed &&
+		    (!old_position->enable || dc->debug.exit_idle_opt_for_cursor_updates) &&
+		    position->enable) {
+			dc_allow_idle_optimizations(dc, false);
+			reset_idle_optimizations = true;
+		}
+
+		program_cursor_position(dc, stream);
+		/* re-enable idle optimizations if necessary */
+		if (reset_idle_optimizations && !dc->debug.disable_dmub_reallow_idle)
+			dc_allow_idle_optimizations(dc, true);
+
+		/* apply/update visual confirm */
+		if (dc->debug.visual_confirm == VISUAL_CONFIRM_HW_CURSOR) {
+			/* update software state */
+			int i;
+
+			for (i = 0; i < dc->res_pool->pipe_count; i++) {
+				struct pipe_ctx *pipe_ctx = &dc->current_state->res_ctx.pipe_ctx[i];
+
+				/* adjust visual confirm color for all pipes with current stream */
+				if (stream == pipe_ctx->stream) {
+					get_cursor_visual_confirm_color(pipe_ctx, &(pipe_ctx->visual_confirm_color));
+
+					/* programming hardware */
+					if (pipe_ctx->plane_state)
+						dc->hwss.update_visual_confirm_color(dc, pipe_ctx,
+								pipe_ctx->plane_res.hubp->mpcc_id);
+				}
+			}
+		}
+
+		return true;
+	}
+
+	return false;
+}
+
+bool dc_stream_add_writeback(struct dc *dc,
+		struct dc_stream_state *stream,
+		struct dc_writeback_info *wb_info)
+{
+	bool isDrc = false;
+	int i = 0;
+	struct dwbc *dwb;
+
+	if (stream == NULL) {
+		dm_error("DC: dc_stream is NULL!\n");
+		return false;
+	}
+
+	if (wb_info == NULL) {
+		dm_error("DC: dc_writeback_info is NULL!\n");
+		return false;
+	}
+
+	if (wb_info->dwb_pipe_inst >= MAX_DWB_PIPES) {
+		dm_error("DC: writeback pipe is invalid!\n");
+		return false;
+	}
+
+	dc_exit_ips_for_hw_access(dc);
+
+	wb_info->dwb_params.out_transfer_func = &stream->out_transfer_func;
+
+	dwb = dc->res_pool->dwbc[wb_info->dwb_pipe_inst];
+	dwb->dwb_is_drc = false;
+
+	/* recalculate and apply DML parameters */
+
+	for (i = 0; i < stream->num_wb_info; i++) {
+		/*dynamic update*/
+		if (stream->writeback_info[i].wb_enabled &&
+			stream->writeback_info[i].dwb_pipe_inst == wb_info->dwb_pipe_inst) {
+			stream->writeback_info[i] = *wb_info;
+			isDrc = true;
+		}
+	}
+
+	if (!isDrc) {
+		ASSERT(stream->num_wb_info + 1 <= MAX_DWB_PIPES);
+		stream->writeback_info[stream->num_wb_info++] = *wb_info;
+	}
+
+	if (dc->hwss.enable_writeback) {
+		struct dc_stream_status *stream_status = dc_stream_get_status(stream);
+		struct dwbc *dwb = dc->res_pool->dwbc[wb_info->dwb_pipe_inst];
+		if (stream_status)
+			dwb->otg_inst = stream_status->primary_otg_inst;
+	}
+
+	if (!dc->hwss.update_bandwidth(dc, dc->current_state)) {
+		dm_error("DC: update_bandwidth failed!\n");
+		return false;
+	}
+
+	/* enable writeback */
+	if (dc->hwss.enable_writeback) {
+		struct dwbc *dwb = dc->res_pool->dwbc[wb_info->dwb_pipe_inst];
+
+		if (dwb->funcs->is_enabled(dwb)) {
+			/* writeback pipe already enabled, only need to update */
+			dc->hwss.update_writeback(dc, wb_info, dc->current_state);
+		} else {
+			/* Enable writeback pipe from scratch*/
+			dc->hwss.enable_writeback(dc, wb_info, dc->current_state);
+		}
+	}
+
+	return true;
+}
+
+bool dc_stream_fc_disable_writeback(struct dc *dc,
+		struct dc_stream_state *stream,
+		uint32_t dwb_pipe_inst)
+{
+	struct dwbc *dwb = dc->res_pool->dwbc[dwb_pipe_inst];
+
+	if (stream == NULL) {
+		dm_error("DC: dc_stream is NULL!\n");
+		return false;
+	}
+
+	if (dwb_pipe_inst >= MAX_DWB_PIPES) {
+		dm_error("DC: writeback pipe is invalid!\n");
+		return false;
+	}
+
+	if (stream->num_wb_info > MAX_DWB_PIPES) {
+		dm_error("DC: num_wb_info is invalid!\n");
+		return false;
+	}
+
+	dc_exit_ips_for_hw_access(dc);
+
+	if (dwb->funcs->set_fc_enable)
+		dwb->funcs->set_fc_enable(dwb, DWB_FRAME_CAPTURE_DISABLE);
+
+	return true;
+}
+
+/**
+ * dc_stream_remove_writeback() - Disables writeback and removes writeback info.
+ * @dc: Display core control structure.
+ * @stream: Display core stream state.
+ * @dwb_pipe_inst: Display writeback pipe.
+ *
+ * Return: returns true on success, false otherwise.
+ */
+bool dc_stream_remove_writeback(struct dc *dc,
+		struct dc_stream_state *stream,
+		uint32_t dwb_pipe_inst)
+{
+	unsigned int i, j;
+	if (stream == NULL) {
+		dm_error("DC: dc_stream is NULL!\n");
+		return false;
+	}
+
+	if (dwb_pipe_inst >= MAX_DWB_PIPES) {
+		dm_error("DC: writeback pipe is invalid!\n");
+		return false;
+	}
+
+	if (stream->num_wb_info > MAX_DWB_PIPES) {
+		dm_error("DC: num_wb_info is invalid!\n");
+		return false;
+	}
+
+	/* remove writeback info for disabled writeback pipes from stream */
+	for (i = 0, j = 0; i < stream->num_wb_info; i++) {
+		if (stream->writeback_info[i].wb_enabled) {
+
+			if (stream->writeback_info[i].dwb_pipe_inst == dwb_pipe_inst)
+				stream->writeback_info[i].wb_enabled = false;
+
+			/* trim the array */
+			if (j < i) {
+				memcpy(&stream->writeback_info[j], &stream->writeback_info[i],
+						sizeof(struct dc_writeback_info));
+				j++;
+			}
+		}
+	}
+	stream->num_wb_info = j;
+
+	/* recalculate and apply DML parameters */
+	if (!dc->hwss.update_bandwidth(dc, dc->current_state)) {
+		dm_error("DC: update_bandwidth failed!\n");
+		return false;
+	}
+
+	dc_exit_ips_for_hw_access(dc);
+
+	/* disable writeback */
+	if (dc->hwss.disable_writeback) {
+		struct dwbc *dwb = dc->res_pool->dwbc[dwb_pipe_inst];
+
+		if (dwb->funcs->is_enabled(dwb))
+			dc->hwss.disable_writeback(dc, dwb_pipe_inst);
+	}
+
+	return true;
+}
+
+uint32_t dc_stream_get_vblank_counter(const struct dc_stream_state *stream)
+{
+	uint8_t i;
+	struct dc  *dc = stream->ctx->dc;
+	struct resource_context *res_ctx =
+		&dc->current_state->res_ctx;
+
+	dc_exit_ips_for_hw_access(dc);
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct timing_generator *tg = res_ctx->pipe_ctx[i].stream_res.tg;
+
+		if (res_ctx->pipe_ctx[i].stream != stream || !tg)
+			continue;
+
+		return tg->funcs->get_frame_count(tg);
+	}
+
+	return 0;
+}
+
+bool dc_stream_send_dp_sdp(const struct dc_stream_state *stream,
+		const uint8_t *custom_sdp_message,
+		unsigned int sdp_message_size)
+{
+	int i;
+	struct dc  *dc;
+	struct resource_context *res_ctx;
+
+	if (stream == NULL) {
+		dm_error("DC: dc_stream is NULL!\n");
+		return false;
+	}
+
+	dc = stream->ctx->dc;
+	res_ctx = &dc->current_state->res_ctx;
+
+	dc_exit_ips_for_hw_access(dc);
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct pipe_ctx *pipe_ctx = &res_ctx->pipe_ctx[i];
+
+		if (pipe_ctx->stream != stream)
+			continue;
+
+		if (dc->hwss.send_immediate_sdp_message != NULL)
+			dc->hwss.send_immediate_sdp_message(pipe_ctx,
+								custom_sdp_message,
+								sdp_message_size);
+		else
+			DC_LOG_WARNING("%s:send_immediate_sdp_message not implemented on this ASIC\n",
+			__func__);
+
+	}
+
+	return true;
+}
+
+bool dc_stream_get_scanoutpos(const struct dc_stream_state *stream,
+				  uint32_t *v_blank_start,
+				  uint32_t *v_blank_end,
+				  uint32_t *h_position,
+				  uint32_t *v_position)
+{
+	uint8_t i;
+	bool ret = false;
+	struct dc  *dc = stream->ctx->dc;
+	struct resource_context *res_ctx =
+		&dc->current_state->res_ctx;
+
+	dc_exit_ips_for_hw_access(dc);
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct timing_generator *tg = res_ctx->pipe_ctx[i].stream_res.tg;
+
+		if (res_ctx->pipe_ctx[i].stream != stream || !tg)
+			continue;
+
+		tg->funcs->get_scanoutpos(tg,
+					  v_blank_start,
+					  v_blank_end,
+					  h_position,
+					  v_position);
+
+		ret = true;
+		break;
+	}
+
+	return ret;
+}
+
+bool dc_stream_dmdata_status_done(struct dc *dc, struct dc_stream_state *stream)
+{
+	struct pipe_ctx *pipe = NULL;
+	int i;
+
+	if (!dc->hwss.dmdata_status_done)
+		return false;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+		if (pipe->stream == stream)
+			break;
+	}
+	/* Stream not found, by default we'll assume HUBP fetched dm data */
+	if (i == MAX_PIPES)
+		return true;
+
+	dc_exit_ips_for_hw_access(dc);
+
+	return dc->hwss.dmdata_status_done(pipe);
+}
+
+bool dc_stream_set_dynamic_metadata(struct dc *dc,
+		struct dc_stream_state *stream,
+		struct dc_dmdata_attributes *attr)
+{
+	struct pipe_ctx *pipe_ctx = NULL;
+	struct hubp *hubp;
+	int i;
+
+	/* Dynamic metadata is only supported on HDMI or DP */
+	if (!dc_is_hdmi_signal(stream->signal) && !dc_is_dp_signal(stream->signal))
+		return false;
+
+	/* Check hardware support */
+	if (!dc->hwss.program_dmdata_engine)
+		return false;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe_ctx = &dc->current_state->res_ctx.pipe_ctx[i];
+		if (pipe_ctx->stream == stream)
+			break;
+	}
+
+	if (i == MAX_PIPES)
+		return false;
+
+	hubp = pipe_ctx->plane_res.hubp;
+	if (hubp == NULL)
+		return false;
+
+	pipe_ctx->stream->dmdata_address = attr->address;
+
+	dc_exit_ips_for_hw_access(dc);
+
+	dc->hwss.program_dmdata_engine(pipe_ctx);
+
+	if (hubp->funcs->dmdata_set_attributes != NULL &&
+			pipe_ctx->stream->dmdata_address.quad_part != 0) {
+		hubp->funcs->dmdata_set_attributes(hubp, attr);
+	}
+
+	return true;
+}
+
+enum dc_status dc_stream_add_dsc_to_resource(struct dc *dc,
+		struct dc_state *state,
+		struct dc_stream_state *stream)
+{
+	if (dc->res_pool->funcs->add_dsc_to_stream_resource) {
+		return dc->res_pool->funcs->add_dsc_to_stream_resource(dc, state, stream);
+	} else {
+		return DC_NO_DSC_RESOURCE;
+	}
+}
+
+struct pipe_ctx *dc_stream_get_pipe_ctx(struct dc_stream_state *stream)
+{
+	int i = 0;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct pipe_ctx *pipe = &stream->ctx->dc->current_state->res_ctx.pipe_ctx[i];
+
+		if (pipe->stream == stream)
+			return pipe;
+	}
+
+	return NULL;
+}
+
+void dc_stream_log(const struct dc *dc, const struct dc_stream_state *stream)
+{
+	DC_LOG_DC(
+			"core_stream 0x%p: src: %d, %d, %d, %d; dst: %d, %d, %d, %d, colorSpace:%d\n",
+			stream,
+			stream->src.x,
+			stream->src.y,
+			stream->src.width,
+			stream->src.height,
+			stream->dst.x,
+			stream->dst.y,
+			stream->dst.width,
+			stream->dst.height,
+			stream->output_color_space);
+	DC_LOG_DC(
+			"\tpix_clk_khz: %d, h_total: %d, v_total: %d, pixel_encoding:%s, color_depth:%s\n",
+			stream->timing.pix_clk_100hz / 10,
+			stream->timing.h_total,
+			stream->timing.v_total,
+			dc_pixel_encoding_to_str(stream->timing.pixel_encoding),
+			dc_color_depth_to_str(stream->timing.display_color_depth));
+	DC_LOG_DC(
+			"\tlink: %d\n",
+			stream->link->link_index);
+
+	DC_LOG_DC(
+			"\tdsc: %d, mst_pbn: %d\n",
+			stream->timing.flags.DSC,
+			stream->timing.dsc_cfg.mst_pbn);
+
+	if (stream->sink) {
+		if (stream->sink->sink_signal != SIGNAL_TYPE_VIRTUAL &&
+			stream->sink->sink_signal != SIGNAL_TYPE_NONE) {
+
+			DC_LOG_DC(
+					"\tdispname: %s signal: %x\n",
+					stream->sink->edid_caps.display_name,
+					stream->signal);
+		}
+	}
+}
+
+/*
+*	dc_stream_get_3dlut()
+*	Requirements:
+*	1. Is stream already owns an RMCM instance, return it.
+*	2. If it doesn't and we don't need to allocate, return NULL.
+*	3. If there's a free RMCM instance, assign to stream and return it.
+*	4. If no free RMCM instances, return NULL.
+*/
+
+struct dc_rmcm_3dlut *dc_stream_get_3dlut_for_stream(
+	const struct dc *dc,
+	const struct dc_stream_state *stream,
+	bool allocate_one)
+{
+	unsigned int num_rmcm = dc->caps.color.mpc.num_rmcm_3dluts;
+
+	// see if one is allocated for this stream
+	for (int i = 0; i < num_rmcm; i++) {
+		if (dc->res_pool->rmcm_3dlut[i].isInUse &&
+			dc->res_pool->rmcm_3dlut[i].stream == stream)
+			return &dc->res_pool->rmcm_3dlut[i];
+	}
+
+	//case: not found one, and dont need to allocate
+	if (!allocate_one)
+		return NULL;
+
+	//see if there is an unused 3dlut, allocate
+	for (int i = 0; i < num_rmcm; i++) {
+		if (!dc->res_pool->rmcm_3dlut[i].isInUse) {
+			dc->res_pool->rmcm_3dlut[i].isInUse = true;
+			dc->res_pool->rmcm_3dlut[i].stream = stream;
+			return &dc->res_pool->rmcm_3dlut[i];
+		}
+	}
+
+	//dont have a 3dlut
+	return NULL;
+}
+
+
+void dc_stream_release_3dlut_for_stream(
+	const struct dc *dc,
+	const struct dc_stream_state *stream)
+{
+	struct dc_rmcm_3dlut *rmcm_3dlut =
+		dc_stream_get_3dlut_for_stream(dc, stream, false);
+
+	if (rmcm_3dlut) {
+		rmcm_3dlut->isInUse = false;
+		rmcm_3dlut->stream  = NULL;
+		rmcm_3dlut->protection_bits = 0;
+	}
+}
+
+
+void dc_stream_init_rmcm_3dlut(struct dc *dc)
+{
+	unsigned int num_rmcm = dc->caps.color.mpc.num_rmcm_3dluts;
+
+	for (int i = 0; i < num_rmcm; i++) {
+		dc->res_pool->rmcm_3dlut[i].isInUse = false;
+		dc->res_pool->rmcm_3dlut[i].stream = NULL;
+		dc->res_pool->rmcm_3dlut[i].protection_bits = 0;
+	}
+}
+
+/*
+ * Finds the greatest index in refresh_rate_hz that contains a value <= refresh
+ */
+static int dc_stream_get_nearest_smallest_index(struct dc_stream_state *stream, int refresh)
+{
+	for (int i = 0; i < (LUMINANCE_DATA_TABLE_SIZE - 1); ++i) {
+		if ((stream->lumin_data.refresh_rate_hz[i] <= refresh) && (refresh < stream->lumin_data.refresh_rate_hz[i + 1])) {
+			return i;
+		}
+	}
+	return 9;
+}
+
+/*
+ * Finds a corresponding brightness for a given refresh rate between 2 given indices, where index1 < index2
+ */
+static int dc_stream_get_brightness_millinits_linear_interpolation (struct dc_stream_state *stream,
+								     int index1,
+								     int index2,
+								     int refresh_hz)
+{
+	long long slope = 0;
+	if (stream->lumin_data.refresh_rate_hz[index2] != stream->lumin_data.refresh_rate_hz[index1]) {
+		slope = (stream->lumin_data.luminance_millinits[index2] - stream->lumin_data.luminance_millinits[index1]) /
+			    (stream->lumin_data.refresh_rate_hz[index2] - stream->lumin_data.refresh_rate_hz[index1]);
+	}
+
+	int y_intercept = stream->lumin_data.luminance_millinits[index2] - slope * stream->lumin_data.refresh_rate_hz[index2];
+
+	return (y_intercept + refresh_hz * slope);
+}
+
+/*
+ * Finds a corresponding refresh rate for a given brightness between 2 given indices, where index1 < index2
+ */
+static int dc_stream_get_refresh_hz_linear_interpolation (struct dc_stream_state *stream,
+							   int index1,
+							   int index2,
+							   int brightness_millinits)
+{
+	long long slope = 1;
+	if (stream->lumin_data.refresh_rate_hz[index2] != stream->lumin_data.refresh_rate_hz[index1]) {
+		slope = (stream->lumin_data.luminance_millinits[index2] - stream->lumin_data.luminance_millinits[index1]) /
+				(stream->lumin_data.refresh_rate_hz[index2] - stream->lumin_data.refresh_rate_hz[index1]);
+	}
+
+	int y_intercept = stream->lumin_data.luminance_millinits[index2] - slope * stream->lumin_data.refresh_rate_hz[index2];
+
+	return ((int)div64_s64((brightness_millinits - y_intercept), slope));
+}
+
+/*
+ * Finds the current brightness in millinits given a refresh rate
+ */
+static int dc_stream_get_brightness_millinits_from_refresh (struct dc_stream_state *stream, int refresh_hz)
+{
+	int nearest_smallest_index = dc_stream_get_nearest_smallest_index(stream, refresh_hz);
+	int nearest_smallest_value = stream->lumin_data.refresh_rate_hz[nearest_smallest_index];
+
+	if (nearest_smallest_value == refresh_hz)
+		return stream->lumin_data.luminance_millinits[nearest_smallest_index];
+
+	if (nearest_smallest_index >= 9)
+		return dc_stream_get_brightness_millinits_linear_interpolation(stream, nearest_smallest_index - 1, nearest_smallest_index, refresh_hz);
+
+	if (nearest_smallest_value == stream->lumin_data.refresh_rate_hz[nearest_smallest_index + 1])
+		return stream->lumin_data.luminance_millinits[nearest_smallest_index];
+
+	return dc_stream_get_brightness_millinits_linear_interpolation(stream, nearest_smallest_index, nearest_smallest_index + 1, refresh_hz);
+}
+
+/*
+ * Finds the lowest/highest refresh rate (depending on search_for_max_increase)
+ * that can be achieved from starting_refresh_hz while staying
+ * within flicker criteria
+ */
+static int dc_stream_calculate_flickerless_refresh_rate(struct dc_stream_state *stream,
+							 int current_brightness,
+							 int starting_refresh_hz,
+							 bool is_gaming,
+							 bool search_for_max_increase)
+{
+	int nearest_smallest_index = dc_stream_get_nearest_smallest_index(stream, starting_refresh_hz);
+
+	int flicker_criteria_millinits = is_gaming ?
+					 stream->lumin_data.flicker_criteria_milli_nits_GAMING :
+					 stream->lumin_data.flicker_criteria_milli_nits_STATIC;
+
+	int safe_upper_bound = current_brightness + flicker_criteria_millinits;
+	int safe_lower_bound = current_brightness - flicker_criteria_millinits;
+	int lumin_millinits_temp = 0;
+
+	int offset = -1;
+	if (search_for_max_increase) {
+		offset = 1;
+	}
+
+	/*
+	 * Increments up or down by 1 depending on search_for_max_increase
+	 */
+	for (int i = nearest_smallest_index; (i > 0 && !search_for_max_increase) || (i < (LUMINANCE_DATA_TABLE_SIZE - 1) && search_for_max_increase); i += offset) {
+
+		lumin_millinits_temp = stream->lumin_data.luminance_millinits[i + offset];
+
+		if ((lumin_millinits_temp >= safe_upper_bound) || (lumin_millinits_temp <= safe_lower_bound)) {
+
+			if (stream->lumin_data.refresh_rate_hz[i + offset] == stream->lumin_data.refresh_rate_hz[i])
+				return stream->lumin_data.refresh_rate_hz[i];
+
+			int target_brightness = (stream->lumin_data.luminance_millinits[i + offset] >= (current_brightness + flicker_criteria_millinits)) ?
+											current_brightness + flicker_criteria_millinits :
+											current_brightness - flicker_criteria_millinits;
+
+			int refresh = 0;
+
+			/*
+			 * Need the second input to be < third input for dc_stream_get_refresh_hz_linear_interpolation
+			 */
+			if (search_for_max_increase)
+				refresh = dc_stream_get_refresh_hz_linear_interpolation(stream, i, i + offset, target_brightness);
+			else
+				refresh = dc_stream_get_refresh_hz_linear_interpolation(stream, i + offset, i, target_brightness);
+
+			if (refresh == stream->lumin_data.refresh_rate_hz[i + offset])
+				return stream->lumin_data.refresh_rate_hz[i + offset];
+
+			return refresh;
+		}
+	}
+
+	if (search_for_max_increase)
+		return (int)div64_s64((long long)stream->timing.pix_clk_100hz*100, stream->timing.v_total*(long long)stream->timing.h_total);
+	else
+		return stream->lumin_data.refresh_rate_hz[0];
+}
+
+/*
+ * Gets the max delta luminance within a specified refresh range
+ */
+static int dc_stream_get_max_delta_lumin_millinits(struct dc_stream_state *stream, int hz1, int hz2, bool isGaming)
+{
+	int lower_refresh_brightness = dc_stream_get_brightness_millinits_from_refresh (stream, hz1);
+	int higher_refresh_brightness = dc_stream_get_brightness_millinits_from_refresh (stream, hz2);
+
+	int min = lower_refresh_brightness;
+	int max = higher_refresh_brightness;
+
+	/*
+	 * Static screen, therefore no need to scan through array
+	 */
+	if (!isGaming) {
+		if (lower_refresh_brightness >= higher_refresh_brightness) {
+			return lower_refresh_brightness - higher_refresh_brightness;
+		}
+		return higher_refresh_brightness - lower_refresh_brightness;
+	}
+
+	min = MIN(lower_refresh_brightness, higher_refresh_brightness);
+	max = MAX(lower_refresh_brightness, higher_refresh_brightness);
+
+	int nearest_smallest_index = dc_stream_get_nearest_smallest_index(stream, hz1);
+
+	for (; nearest_smallest_index < (LUMINANCE_DATA_TABLE_SIZE - 1) &&
+			stream->lumin_data.refresh_rate_hz[nearest_smallest_index + 1] <= hz2 ; nearest_smallest_index++) {
+		min = MIN(min, stream->lumin_data.luminance_millinits[nearest_smallest_index + 1]);
+		max = MAX(max, stream->lumin_data.luminance_millinits[nearest_smallest_index + 1]);
+	}
+
+	return (max - min);
+}
+
+/*
+ * Determines the max flickerless instant vtotal delta for a stream.
+ * Determines vtotal increase/decrease based on the bool "increase"
+ */
+static unsigned int dc_stream_get_max_flickerless_instant_vtotal_delta(struct dc_stream_state *stream, bool is_gaming, bool increase)
+{
+	if (stream->timing.v_total * stream->timing.h_total == 0)
+		return 0;
+
+	int current_refresh_hz = (int)div64_s64((long long)stream->timing.pix_clk_100hz*100, stream->timing.v_total*(long long)stream->timing.h_total);
+
+	int safe_refresh_hz = dc_stream_calculate_flickerless_refresh_rate(stream,
+							 dc_stream_get_brightness_millinits_from_refresh(stream, current_refresh_hz),
+							 current_refresh_hz,
+							 is_gaming,
+							 increase);
+
+	int safe_refresh_v_total = (int)div64_s64((long long)stream->timing.pix_clk_100hz*100, safe_refresh_hz*(long long)stream->timing.h_total);
+
+	if (increase)
+		return (((int) stream->timing.v_total - safe_refresh_v_total) >= 0) ? (stream->timing.v_total - safe_refresh_v_total) : 0;
+
+	return ((safe_refresh_v_total - (int) stream->timing.v_total) >= 0) ? (safe_refresh_v_total - stream->timing.v_total) : 0;
+}
+
+/*
+ * Finds the highest refresh rate that can be achieved
+ * from starting_refresh_hz while staying within flicker criteria
+ */
+int dc_stream_calculate_max_flickerless_refresh_rate(struct dc_stream_state *stream, int starting_refresh_hz, bool is_gaming)
+{
+	if (!stream->lumin_data.is_valid)
+		return 0;
+
+	int current_brightness = dc_stream_get_brightness_millinits_from_refresh(stream, starting_refresh_hz);
+
+	return dc_stream_calculate_flickerless_refresh_rate(stream,
+							    current_brightness,
+							    starting_refresh_hz,
+							    is_gaming,
+							    true);
+}
+
+/*
+ * Finds the lowest refresh rate that can be achieved
+ * from starting_refresh_hz while staying within flicker criteria
+ */
+int dc_stream_calculate_min_flickerless_refresh_rate(struct dc_stream_state *stream, int starting_refresh_hz, bool is_gaming)
+{
+	if (!stream->lumin_data.is_valid)
+			return 0;
+
+	int current_brightness = dc_stream_get_brightness_millinits_from_refresh(stream, starting_refresh_hz);
+
+	return dc_stream_calculate_flickerless_refresh_rate(stream,
+							    current_brightness,
+							    starting_refresh_hz,
+							    is_gaming,
+							    false);
+}
+
+/*
+ * Determines if there will be a flicker when moving between 2 refresh rates
+ */
+bool dc_stream_is_refresh_rate_range_flickerless(struct dc_stream_state *stream, int hz1, int hz2, bool is_gaming)
+{
+
+	/*
+	 * Assume that we wont flicker if there is invalid data
+	 */
+	if (!stream->lumin_data.is_valid)
+		return false;
+
+	int dl = dc_stream_get_max_delta_lumin_millinits(stream, hz1, hz2, is_gaming);
+
+	int flicker_criteria_millinits = (is_gaming) ?
+					  stream->lumin_data.flicker_criteria_milli_nits_GAMING :
+					  stream->lumin_data.flicker_criteria_milli_nits_STATIC;
+
+	return (dl <= flicker_criteria_millinits);
+}
+
+/*
+ * Determines the max instant vtotal delta increase that can be applied without
+ * flickering for a given stream
+ */
+unsigned int dc_stream_get_max_flickerless_instant_vtotal_decrease(struct dc_stream_state *stream,
+									  bool is_gaming)
+{
+	if (!stream->lumin_data.is_valid)
+		return 0;
+
+	return dc_stream_get_max_flickerless_instant_vtotal_delta(stream, is_gaming, true);
+}
+
+/*
+ * Determines the max instant vtotal delta decrease that can be applied without
+ * flickering for a given stream
+ */
+unsigned int dc_stream_get_max_flickerless_instant_vtotal_increase(struct dc_stream_state *stream,
+									  bool is_gaming)
+{
+	if (!stream->lumin_data.is_valid)
+		return 0;
+
+	return dc_stream_get_max_flickerless_instant_vtotal_delta(stream, is_gaming, false);
+}
+
+bool dc_stream_is_cursor_limit_pending(struct dc *dc, struct dc_stream_state *stream)
+{
+	bool is_limit_pending = false;
+
+	if (dc->current_state)
+		is_limit_pending = dc_state_get_stream_cursor_subvp_limit(stream, dc->current_state);
+
+	return is_limit_pending;
+}
+
+bool dc_stream_can_clear_cursor_limit(struct dc *dc, struct dc_stream_state *stream)
+{
+	bool can_clear_limit = false;
+
+	if (dc->current_state)
+		can_clear_limit = dc_state_get_stream_cursor_subvp_limit(stream, dc->current_state) &&
+				(stream->hw_cursor_req ||
+				!stream->cursor_position.enable ||
+				dc_stream_check_cursor_attributes(stream, dc->current_state, &stream->cursor_attributes));
+
+	return can_clear_limit;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_surface.c b/drivers/gpu/drm/amd/display/dc/core/dc_surface.c
new file mode 100644
index 000000000000..922f23557f5d
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_surface.c
@@ -0,0 +1,312 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+/* DC interface (public) */
+#include "dm_services.h"
+#include "dc.h"
+
+/* DC core (private) */
+#include "core_types.h"
+#include "transform.h"
+#include "dpp.h"
+
+#include "dc_plane_priv.h"
+
+/*******************************************************************************
+ * Private functions
+ ******************************************************************************/
+void dc_plane_construct(struct dc_context *ctx, struct dc_plane_state *plane_state)
+{
+	plane_state->ctx = ctx;
+
+	plane_state->gamma_correction.is_identity = true;
+
+	plane_state->in_transfer_func.type = TF_TYPE_BYPASS;
+
+	plane_state->in_shaper_func.type = TF_TYPE_BYPASS;
+
+	plane_state->lut3d_func.state.raw = 0;
+
+	plane_state->blend_tf.type = TF_TYPE_BYPASS;
+
+	plane_state->pre_multiplied_alpha = true;
+
+}
+
+void dc_plane_destruct(struct dc_plane_state *plane_state)
+{
+	// no more pointers to free within dc_plane_state
+}
+
+
+/* dc_state is passed in separately since it may differ from the current dc state accessible from plane_state e.g.
+ * if the driver is doing an update from an old context to a new one and the caller wants the pipe mask for the new
+ * context rather than the existing one
+ */
+uint8_t  dc_plane_get_pipe_mask(struct dc_state *dc_state, const struct dc_plane_state *plane_state)
+{
+	uint8_t pipe_mask = 0;
+	int i;
+
+	for (i = 0; i < plane_state->ctx->dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *pipe_ctx = &dc_state->res_ctx.pipe_ctx[i];
+
+		if (pipe_ctx->plane_state == plane_state && pipe_ctx->plane_res.hubp)
+			pipe_mask |= 1 << pipe_ctx->plane_res.hubp->inst;
+	}
+
+	return pipe_mask;
+}
+
+/*******************************************************************************
+ * Public functions
+ ******************************************************************************/
+struct dc_plane_state *dc_create_plane_state(const struct dc *dc)
+{
+	struct dc_plane_state *plane_state = kvzalloc(sizeof(*plane_state),
+							GFP_KERNEL);
+
+	if (NULL == plane_state)
+		return NULL;
+
+	kref_init(&plane_state->refcount);
+	dc_plane_construct(dc->ctx, plane_state);
+
+	return plane_state;
+}
+
+/*
+ *****************************************************************************
+ *  Function: dc_plane_get_status
+ *
+ *  @brief
+ *     Looks up the pipe context of plane_state and updates the pending status
+ *     of the pipe context. Then returns plane_state->status
+ *
+ *  @param [in] plane_state: pointer to the plane_state to get the status of
+ *****************************************************************************
+ */
+const struct dc_plane_status *dc_plane_get_status(
+		const struct dc_plane_state *plane_state,
+		union dc_plane_status_update_flags flags)
+{
+	const struct dc_plane_status *plane_status;
+	struct dc  *dc;
+	int i;
+
+	if (!plane_state ||
+		!plane_state->ctx ||
+		!plane_state->ctx->dc) {
+		ASSERT(0);
+		return NULL; /* remove this if above assert never hit */
+	}
+
+	plane_status = &plane_state->status;
+	dc = plane_state->ctx->dc;
+
+	if (dc->current_state == NULL)
+		return NULL;
+
+	/* Find the current plane state and set its pending bit to false */
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *pipe_ctx =
+				&dc->current_state->res_ctx.pipe_ctx[i];
+
+		if (pipe_ctx->plane_state != plane_state)
+			continue;
+
+		if (pipe_ctx->plane_state && flags.bits.address)
+			pipe_ctx->plane_state->status.is_flip_pending = false;
+
+		break;
+	}
+
+	dc_exit_ips_for_hw_access(dc);
+
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *pipe_ctx =
+				&dc->current_state->res_ctx.pipe_ctx[i];
+
+		if (pipe_ctx->plane_state != plane_state)
+			continue;
+
+		if (flags.bits.address)
+			dc->hwss.update_pending_status(pipe_ctx);
+	}
+
+	return plane_status;
+}
+
+void dc_plane_state_retain(struct dc_plane_state *plane_state)
+{
+	kref_get(&plane_state->refcount);
+}
+
+static void dc_plane_state_free(struct kref *kref)
+{
+	struct dc_plane_state *plane_state = container_of(kref, struct dc_plane_state, refcount);
+	dc_plane_destruct(plane_state);
+	kvfree(plane_state);
+}
+
+void dc_plane_state_release(struct dc_plane_state *plane_state)
+{
+	kref_put(&plane_state->refcount, dc_plane_state_free);
+}
+
+void dc_gamma_retain(struct dc_gamma *gamma)
+{
+	kref_get(&gamma->refcount);
+}
+
+static void dc_gamma_free(struct kref *kref)
+{
+	struct dc_gamma *gamma = container_of(kref, struct dc_gamma, refcount);
+	kvfree(gamma);
+}
+
+void dc_gamma_release(struct dc_gamma **gamma)
+{
+	kref_put(&(*gamma)->refcount, dc_gamma_free);
+	*gamma = NULL;
+}
+
+struct dc_gamma *dc_create_gamma(void)
+{
+	struct dc_gamma *gamma = kvzalloc(sizeof(*gamma), GFP_KERNEL);
+
+	if (gamma == NULL)
+		goto alloc_fail;
+
+	kref_init(&gamma->refcount);
+	return gamma;
+
+alloc_fail:
+	return NULL;
+}
+
+void dc_transfer_func_retain(struct dc_transfer_func *tf)
+{
+	kref_get(&tf->refcount);
+}
+
+static void dc_transfer_func_free(struct kref *kref)
+{
+	struct dc_transfer_func *tf = container_of(kref, struct dc_transfer_func, refcount);
+	kvfree(tf);
+}
+
+void dc_transfer_func_release(struct dc_transfer_func *tf)
+{
+	kref_put(&tf->refcount, dc_transfer_func_free);
+}
+
+struct dc_transfer_func *dc_create_transfer_func(void)
+{
+	struct dc_transfer_func *tf = kvzalloc(sizeof(*tf), GFP_KERNEL);
+
+	if (tf == NULL)
+		goto alloc_fail;
+
+	kref_init(&tf->refcount);
+
+	return tf;
+
+alloc_fail:
+	return NULL;
+}
+
+static void dc_3dlut_func_free(struct kref *kref)
+{
+	struct dc_3dlut *lut = container_of(kref, struct dc_3dlut, refcount);
+
+	kvfree(lut);
+}
+
+struct dc_3dlut *dc_create_3dlut_func(void)
+{
+	struct dc_3dlut *lut = kvzalloc(sizeof(*lut), GFP_KERNEL);
+
+	if (lut == NULL)
+		goto alloc_fail;
+
+	kref_init(&lut->refcount);
+	lut->state.raw = 0;
+
+	return lut;
+
+alloc_fail:
+	return NULL;
+
+}
+
+void dc_3dlut_func_release(struct dc_3dlut *lut)
+{
+	kref_put(&lut->refcount, dc_3dlut_func_free);
+}
+
+void dc_3dlut_func_retain(struct dc_3dlut *lut)
+{
+	kref_get(&lut->refcount);
+}
+
+void dc_plane_force_dcc_and_tiling_disable(struct dc_plane_state *plane_state,
+					   bool clear_tiling)
+{
+	struct dc *dc;
+	int i;
+
+	if (!plane_state)
+		return;
+
+	dc = plane_state->ctx->dc;
+
+	if (!dc || !dc->current_state)
+		return;
+
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *pipe_ctx = &dc->current_state->res_ctx.pipe_ctx[i];
+
+		if (!pipe_ctx)
+			continue;
+
+		if (dc->hwss.clear_surface_dcc_and_tiling)
+			dc->hwss.clear_surface_dcc_and_tiling(pipe_ctx, plane_state, clear_tiling);
+	}
+}
+
+void dc_plane_copy_config(struct dc_plane_state *dst, const struct dc_plane_state *src)
+{
+	struct kref temp_refcount;
+
+	/* backup persistent info */
+	memcpy(&temp_refcount, &dst->refcount, sizeof(struct kref));
+
+	/* copy all configuration information */
+	memcpy(dst, src, sizeof(struct dc_plane_state));
+
+	/* restore persistent info */
+	memcpy(&dst->refcount, &temp_refcount, sizeof(struct kref));
+}
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_vm_helper.c b/drivers/gpu/drm/amd/display/dc/core/dc_vm_helper.c
new file mode 100644
index 000000000000..d1e68dc57a2a
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_vm_helper.c
@@ -0,0 +1,78 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "vm_helper.h"
+#include "dc.h"
+
+void vm_helper_mark_vmid_used(struct vm_helper *vm_helper, unsigned int pos, uint8_t hubp_idx)
+{
+	struct vmid_usage vmids = vm_helper->hubp_vmid_usage[hubp_idx];
+
+	vmids.vmid_usage[0] = vmids.vmid_usage[1];
+	vmids.vmid_usage[1] = 1 << pos;
+}
+
+int dc_setup_system_context(struct dc *dc, struct dc_phy_addr_space_config *pa_config)
+{
+	int num_vmids = 0;
+
+	/* Call HWSS to setup HUBBUB for address config */
+	if (dc->hwss.init_sys_ctx) {
+		num_vmids = dc->hwss.init_sys_ctx(dc->hwseq, dc, pa_config);
+
+		/* Pre-init system aperture start/end for all HUBP instances (if not gating?)
+		 * or cache system aperture if using power gating
+		 */
+		memcpy(&dc->vm_pa_config, pa_config, sizeof(struct dc_phy_addr_space_config));
+		dc->vm_pa_config.valid = true;
+		dc->dml2_options.gpuvm_enable = true;
+		dc_z10_save_init(dc);
+	}
+
+	return num_vmids;
+}
+
+void dc_setup_vm_context(struct dc *dc, struct dc_virtual_addr_space_config *va_config, int vmid)
+{
+	dc->hwss.init_vm_ctx(dc->hwseq, dc, va_config, vmid);
+}
+
+int dc_get_vmid_use_vector(struct dc *dc)
+{
+	int i;
+	int in_use = 0;
+
+	for (i = 0; i < MAX_HUBP; i++)
+		in_use |= dc->vm_helper->hubp_vmid_usage[i].vmid_usage[0]
+			| dc->vm_helper->hubp_vmid_usage[i].vmid_usage[1];
+	return in_use;
+}
+
+void vm_helper_init(struct vm_helper *vm_helper, unsigned int num_vmid)
+{
+	vm_helper->num_vmid = num_vmid;
+
+	memset(vm_helper->hubp_vmid_usage, 0, sizeof(vm_helper->hubp_vmid_usage[0]) * MAX_HUBP);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h
new file mode 100644
index 000000000000..98f0b6b3c213
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc.h
@@ -0,0 +1,2754 @@
+/*
+ * Copyright 2012-2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DC_INTERFACE_H_
+#define DC_INTERFACE_H_
+
+#include "dc_types.h"
+#include "dc_state.h"
+#include "dc_plane.h"
+#include "grph_object_defs.h"
+#include "logger_types.h"
+#include "hdcp_msg_types.h"
+#include "gpio_types.h"
+#include "link_service_types.h"
+#include "grph_object_ctrl_defs.h"
+#include <inc/hw/opp.h>
+
+#include "hwss/hw_sequencer.h"
+#include "inc/compressor.h"
+#include "inc/hw/dmcu.h"
+#include "dml/display_mode_lib.h"
+
+#include "dml2/dml2_wrapper.h"
+
+#include "dmub/inc/dmub_cmd.h"
+
+#include "sspl/dc_spl_types.h"
+
+struct abm_save_restore;
+
+/* forward declaration */
+struct aux_payload;
+struct set_config_cmd_payload;
+struct dmub_notification;
+
+#define DC_VER "3.2.351"
+
+/**
+ * MAX_SURFACES - representative of the upper bound of surfaces that can be piped to a single CRTC
+ */
+#define MAX_SURFACES 4
+/**
+ * MAX_PLANES - representative of the upper bound of planes that are supported by the HW
+ */
+#define MAX_PLANES 6
+#define MAX_STREAMS 6
+#define MIN_VIEWPORT_SIZE 12
+#define MAX_NUM_EDP 2
+#define MAX_SUPPORTED_FORMATS 7
+
+#define MAX_HOST_ROUTERS_NUM 3
+#define MAX_DPIA_PER_HOST_ROUTER 3
+#define MAX_DPIA_NUM  (MAX_HOST_ROUTERS_NUM * MAX_DPIA_PER_HOST_ROUTER)
+
+/* Display Core Interfaces */
+struct dc_versions {
+	const char *dc_ver;
+	struct dmcu_version dmcu_version;
+};
+
+enum dp_protocol_version {
+	DP_VERSION_1_4 = 0,
+	DP_VERSION_2_1,
+	DP_VERSION_UNKNOWN,
+};
+
+enum dc_plane_type {
+	DC_PLANE_TYPE_INVALID,
+	DC_PLANE_TYPE_DCE_RGB,
+	DC_PLANE_TYPE_DCE_UNDERLAY,
+	DC_PLANE_TYPE_DCN_UNIVERSAL,
+};
+
+// Sizes defined as multiples of 64KB
+enum det_size {
+	DET_SIZE_DEFAULT = 0,
+	DET_SIZE_192KB = 3,
+	DET_SIZE_256KB = 4,
+	DET_SIZE_320KB = 5,
+	DET_SIZE_384KB = 6
+};
+
+
+struct dc_plane_cap {
+	enum dc_plane_type type;
+	uint32_t per_pixel_alpha : 1;
+	struct {
+		uint32_t argb8888 : 1;
+		uint32_t nv12 : 1;
+		uint32_t fp16 : 1;
+		uint32_t p010 : 1;
+		uint32_t ayuv : 1;
+	} pixel_format_support;
+	// max upscaling factor x1000
+	// upscaling factors are always >= 1
+	// for example, 1080p -> 8K is 4.0, or 4000 raw value
+	struct {
+		uint32_t argb8888;
+		uint32_t nv12;
+		uint32_t fp16;
+	} max_upscale_factor;
+	// max downscale factor x1000
+	// downscale factors are always <= 1
+	// for example, 8K -> 1080p is 0.25, or 250 raw value
+	struct {
+		uint32_t argb8888;
+		uint32_t nv12;
+		uint32_t fp16;
+	} max_downscale_factor;
+	// minimal width/height
+	uint32_t min_width;
+	uint32_t min_height;
+};
+
+/**
+ * DOC: color-management-caps
+ *
+ * **Color management caps (DPP and MPC)**
+ *
+ * Modules/color calculates various color operations which are translated to
+ * abstracted HW. DCE 5-12 had almost no important changes, but starting with
+ * DCN1, every new generation comes with fairly major differences in color
+ * pipeline. Therefore, we abstract color pipe capabilities so modules/DM can
+ * decide mapping to HW block based on logical capabilities.
+ */
+
+/**
+ * struct rom_curve_caps - predefined transfer function caps for degamma and regamma
+ * @srgb: RGB color space transfer func
+ * @bt2020: BT.2020 transfer func
+ * @gamma2_2: standard gamma
+ * @pq: perceptual quantizer transfer function
+ * @hlg: hybrid log–gamma transfer function
+ */
+struct rom_curve_caps {
+	uint16_t srgb : 1;
+	uint16_t bt2020 : 1;
+	uint16_t gamma2_2 : 1;
+	uint16_t pq : 1;
+	uint16_t hlg : 1;
+};
+
+/**
+ * struct dpp_color_caps - color pipeline capabilities for display pipe and
+ * plane blocks
+ *
+ * @dcn_arch: all DCE generations treated the same
+ * @input_lut_shared: shared with DGAM. Input LUT is different than most LUTs,
+ * just plain 256-entry lookup
+ * @icsc: input color space conversion
+ * @dgam_ram: programmable degamma LUT
+ * @post_csc: post color space conversion, before gamut remap
+ * @gamma_corr: degamma correction
+ * @hw_3d_lut: 3D LUT support. It implies a shaper LUT before. It may be shared
+ * with MPC by setting mpc:shared_3d_lut flag
+ * @ogam_ram: programmable out/blend gamma LUT
+ * @ocsc: output color space conversion
+ * @dgam_rom_for_yuv: pre-defined degamma LUT for YUV planes
+ * @dgam_rom_caps: pre-definied curve caps for degamma 1D LUT
+ * @ogam_rom_caps: pre-definied curve caps for regamma 1D LUT
+ *
+ * Note: hdr_mult and gamut remap (CTM) are always available in DPP (in that order)
+ */
+struct dpp_color_caps {
+	uint16_t dcn_arch : 1;
+	uint16_t input_lut_shared : 1;
+	uint16_t icsc : 1;
+	uint16_t dgam_ram : 1;
+	uint16_t post_csc : 1;
+	uint16_t gamma_corr : 1;
+	uint16_t hw_3d_lut : 1;
+	uint16_t ogam_ram : 1;
+	uint16_t ocsc : 1;
+	uint16_t dgam_rom_for_yuv : 1;
+	struct rom_curve_caps dgam_rom_caps;
+	struct rom_curve_caps ogam_rom_caps;
+};
+
+/* Below structure is to describe the HW support for mem layout, extend support
+	range to match what OS could handle in the roadmap */
+struct lut3d_caps {
+	uint32_t dma_3d_lut : 1; /*< DMA mode support for 3D LUT */
+	struct {
+		uint32_t swizzle_3d_rgb : 1;
+		uint32_t swizzle_3d_bgr : 1;
+		uint32_t linear_1d : 1;
+	} mem_layout_support;
+	struct {
+		uint32_t unorm_12msb : 1;
+		uint32_t unorm_12lsb : 1;
+		uint32_t float_fp1_5_10 : 1;
+	} mem_format_support;
+	struct {
+		uint32_t order_rgba : 1;
+		uint32_t order_bgra : 1;
+	} mem_pixel_order_support;
+	/*< size options are 9, 17, 33, 45, 65 */
+	struct {
+		uint32_t dim_9 : 1; /* 3D LUT support for 9x9x9 */
+		uint32_t dim_17 : 1; /* 3D LUT support for 17x17x17 */
+		uint32_t dim_33 : 1; /* 3D LUT support for 33x33x33 */
+		uint32_t dim_45 : 1; /* 3D LUT support for 45x45x45 */
+		uint32_t dim_65 : 1; /* 3D LUT support for 65x65x65 */
+	} lut_dim_caps;
+};
+
+/**
+ * struct mpc_color_caps - color pipeline capabilities for multiple pipe and
+ * plane combined blocks
+ *
+ * @gamut_remap: color transformation matrix
+ * @ogam_ram: programmable out gamma LUT
+ * @ocsc: output color space conversion matrix
+ * @num_3dluts: MPC 3D LUT; always assumes a preceding shaper LUT
+ * @num_rmcm_3dluts: number of RMCM 3D LUTS; always assumes a preceding shaper LUT
+ * @shared_3d_lut: shared 3D LUT flag. Can be either DPP or MPC, but single
+ * instance
+ * @ogam_rom_caps: pre-definied curve caps for regamma 1D LUT
+ * @mcm_3d_lut_caps: HW support cap for MCM LUT memory
+ * @rmcm_3d_lut_caps: HW support cap for RMCM LUT memory
+ * @preblend: whether color manager supports preblend with MPC
+ */
+struct mpc_color_caps {
+	uint16_t gamut_remap : 1;
+	uint16_t ogam_ram : 1;
+	uint16_t ocsc : 1;
+	uint16_t num_3dluts : 3;
+	uint16_t num_rmcm_3dluts : 3;
+	uint16_t shared_3d_lut:1;
+	struct rom_curve_caps ogam_rom_caps;
+	struct lut3d_caps mcm_3d_lut_caps;
+	struct lut3d_caps rmcm_3d_lut_caps;
+	bool preblend;
+};
+
+/**
+ * struct dc_color_caps - color pipes capabilities for DPP and MPC hw blocks
+ * @dpp: color pipes caps for DPP
+ * @mpc: color pipes caps for MPC
+ */
+struct dc_color_caps {
+	struct dpp_color_caps dpp;
+	struct mpc_color_caps mpc;
+};
+
+struct dc_dmub_caps {
+	bool psr;
+	bool mclk_sw;
+	bool subvp_psr;
+	bool gecc_enable;
+	uint8_t fams_ver;
+	bool aux_backlight_support;
+};
+
+struct dc_scl_caps {
+	bool sharpener_support;
+};
+
+struct dc_caps {
+	uint32_t max_streams;
+	uint32_t max_links;
+	uint32_t max_audios;
+	uint32_t max_slave_planes;
+	uint32_t max_slave_yuv_planes;
+	uint32_t max_slave_rgb_planes;
+	uint32_t max_planes;
+	uint32_t max_downscale_ratio;
+	uint32_t i2c_speed_in_khz;
+	uint32_t i2c_speed_in_khz_hdcp;
+	uint32_t dmdata_alloc_size;
+	unsigned int max_cursor_size;
+	unsigned int max_buffered_cursor_size;
+	unsigned int max_video_width;
+	/*
+	 * max video plane width that can be safely assumed to be always
+	 * supported by single DPP pipe.
+	 */
+	unsigned int max_optimizable_video_width;
+	unsigned int min_horizontal_blanking_period;
+	int linear_pitch_alignment;
+	bool dcc_const_color;
+	bool dynamic_audio;
+	bool is_apu;
+	bool dual_link_dvi;
+	bool post_blend_color_processing;
+	bool force_dp_tps4_for_cp2520;
+	bool disable_dp_clk_share;
+	bool psp_setup_panel_mode;
+	bool extended_aux_timeout_support;
+	bool dmcub_support;
+	bool zstate_support;
+	bool ips_support;
+	bool ips_v2_support;
+	uint32_t num_of_internal_disp;
+	enum dp_protocol_version max_dp_protocol_version;
+	unsigned int mall_size_per_mem_channel;
+	unsigned int mall_size_total;
+	unsigned int cursor_cache_size;
+	struct dc_plane_cap planes[MAX_PLANES];
+	struct dc_color_caps color;
+	struct dc_dmub_caps dmub_caps;
+	bool dp_hpo;
+	bool dp_hdmi21_pcon_support;
+	bool edp_dsc_support;
+	bool vbios_lttpr_aware;
+	bool vbios_lttpr_enable;
+	bool fused_io_supported;
+	uint32_t max_otg_num;
+	uint32_t max_cab_allocation_bytes;
+	uint32_t cache_line_size;
+	uint32_t cache_num_ways;
+	uint16_t subvp_fw_processing_delay_us;
+	uint8_t subvp_drr_max_vblank_margin_us;
+	uint16_t subvp_prefetch_end_to_mall_start_us;
+	uint8_t subvp_swath_height_margin_lines; // subvp start line must be aligned to 2 x swath height
+	uint16_t subvp_pstate_allow_width_us;
+	uint16_t subvp_vertical_int_margin_us;
+	bool seamless_odm;
+	uint32_t max_v_total;
+	bool vtotal_limited_by_fp2;
+	uint32_t max_disp_clock_khz_at_vmin;
+	uint8_t subvp_drr_vblank_start_margin_us;
+	bool cursor_not_scaled;
+	bool dcmode_power_limits_present;
+	bool sequential_ono;
+	/* Conservative limit for DCC cases which require ODM4:1 to support*/
+	uint32_t dcc_plane_width_limit;
+	struct dc_scl_caps scl_caps;
+	uint8_t num_of_host_routers;
+	uint8_t num_of_dpias_per_host_router;
+	/* limit of the ODM only, could be limited by other factors (like pipe count)*/
+	uint8_t max_odm_combine_factor;
+};
+
+struct dc_bug_wa {
+	bool no_connect_phy_config;
+	bool dedcn20_305_wa;
+	bool skip_clock_update;
+	bool lt_early_cr_pattern;
+	struct {
+		uint8_t uclk : 1;
+		uint8_t fclk : 1;
+		uint8_t dcfclk : 1;
+		uint8_t dcfclk_ds: 1;
+	} clock_update_disable_mask;
+	bool skip_psr_ips_crtc_disable;
+};
+struct dc_dcc_surface_param {
+	struct dc_size surface_size;
+	enum surface_pixel_format format;
+	unsigned int plane0_pitch;
+	struct dc_size plane1_size;
+	unsigned int plane1_pitch;
+	union {
+		enum swizzle_mode_values swizzle_mode;
+		enum swizzle_mode_addr3_values swizzle_mode_addr3;
+	};
+	enum dc_scan_direction scan;
+};
+
+struct dc_dcc_setting {
+	unsigned int max_compressed_blk_size;
+	unsigned int max_uncompressed_blk_size;
+	bool independent_64b_blks;
+	//These bitfields to be used starting with DCN 3.0
+	struct {
+		uint32_t dcc_256_64_64 : 1;//available in ASICs before DCN 3.0 (the worst compression case)
+		uint32_t dcc_128_128_uncontrained : 1;  //available in ASICs before DCN 3.0
+		uint32_t dcc_256_128_128 : 1;		//available starting with DCN 3.0
+		uint32_t dcc_256_256_unconstrained : 1;  //available in ASICs before DCN 3.0 (the best compression case)
+		uint32_t dcc_256_256 : 1;  //available in ASICs starting with DCN 4.0x (the best compression case)
+		uint32_t dcc_256_128 : 1;  //available in ASICs starting with DCN 4.0x
+		uint32_t dcc_256_64 : 1;   //available in ASICs starting with DCN 4.0x (the worst compression case)
+	} dcc_controls;
+};
+
+struct dc_surface_dcc_cap {
+	union {
+		struct {
+			struct dc_dcc_setting rgb;
+		} grph;
+
+		struct {
+			struct dc_dcc_setting luma;
+			struct dc_dcc_setting chroma;
+		} video;
+	};
+
+	bool capable;
+	bool const_color_support;
+};
+
+struct dc_static_screen_params {
+	struct {
+		bool force_trigger;
+		bool cursor_update;
+		bool surface_update;
+		bool overlay_update;
+	} triggers;
+	unsigned int num_frames;
+};
+
+
+/* Surface update type is used by dc_update_surfaces_and_stream
+ * The update type is determined at the very beginning of the function based
+ * on parameters passed in and decides how much programming (or updating) is
+ * going to be done during the call.
+ *
+ * UPDATE_TYPE_FAST is used for really fast updates that do not require much
+ * logical calculations or hardware register programming. This update MUST be
+ * ISR safe on windows. Currently fast update will only be used to flip surface
+ * address.
+ *
+ * UPDATE_TYPE_MED is used for slower updates which require significant hw
+ * re-programming however do not affect bandwidth consumption or clock
+ * requirements. At present, this is the level at which front end updates
+ * that do not require us to run bw_calcs happen. These are in/out transfer func
+ * updates, viewport offset changes, recout size changes and pixel depth changes.
+ * This update can be done at ISR, but we want to minimize how often this happens.
+ *
+ * UPDATE_TYPE_FULL is slow. Really slow. This requires us to recalculate our
+ * bandwidth and clocks, possibly rearrange some pipes and reprogram anything front
+ * end related. Any time viewport dimensions, recout dimensions, scaling ratios or
+ * gamma need to be adjusted or pipe needs to be turned on (or disconnected) we do
+ * a full update. This cannot be done at ISR level and should be a rare event.
+ * Unless someone is stress testing mpo enter/exit, playing with colour or adjusting
+ * underscan we don't expect to see this call at all.
+ */
+
+enum surface_update_type {
+	UPDATE_TYPE_FAST, /* super fast, safe to execute in isr */
+	UPDATE_TYPE_MED,  /* ISR safe, most of programming needed, no bw/clk change*/
+	UPDATE_TYPE_FULL, /* may need to shuffle resources */
+};
+
+/* Forward declaration*/
+struct dc;
+struct dc_plane_state;
+struct dc_state;
+
+struct dc_cap_funcs {
+	bool (*get_dcc_compression_cap)(const struct dc *dc,
+			const struct dc_dcc_surface_param *input,
+			struct dc_surface_dcc_cap *output);
+	bool (*get_subvp_en)(struct dc *dc, struct dc_state *context);
+};
+
+struct link_training_settings;
+
+union allow_lttpr_non_transparent_mode {
+	struct {
+		bool DP1_4A : 1;
+		bool DP2_0 : 1;
+	} bits;
+	unsigned char raw;
+};
+
+/* Structure to hold configuration flags set by dm at dc creation. */
+struct dc_config {
+	bool gpu_vm_support;
+	bool disable_disp_pll_sharing;
+	bool fbc_support;
+	bool disable_fractional_pwm;
+	bool allow_seamless_boot_optimization;
+	bool seamless_boot_edp_requested;
+	bool edp_not_connected;
+	bool edp_no_power_sequencing;
+	bool force_enum_edp;
+	bool forced_clocks;
+	union allow_lttpr_non_transparent_mode allow_lttpr_non_transparent_mode;
+	bool multi_mon_pp_mclk_switch;
+	bool disable_dmcu;
+	bool enable_4to1MPC;
+	bool enable_windowed_mpo_odm;
+	bool forceHBR2CP2520; // Used for switching between test patterns TPS4 and CP2520
+	uint32_t allow_edp_hotplug_detection;
+	bool skip_riommu_prefetch_wa;
+	bool clamp_min_dcfclk;
+	uint64_t vblank_alignment_dto_params;
+	uint8_t  vblank_alignment_max_frame_time_diff;
+	bool is_asymmetric_memory;
+	bool is_single_rank_dimm;
+	bool is_vmin_only_asic;
+	bool use_spl;
+	bool prefer_easf;
+	bool use_pipe_ctx_sync_logic;
+	int smart_mux_version;
+	bool ignore_dpref_ss;
+	bool enable_mipi_converter_optimization;
+	bool use_default_clock_table;
+	bool force_bios_enable_lttpr;
+	uint8_t force_bios_fixed_vs;
+	int sdpif_request_limit_words_per_umc;
+	bool dc_mode_clk_limit_support;
+	bool EnableMinDispClkODM;
+	bool enable_auto_dpm_test_logs;
+	unsigned int disable_ips;
+	unsigned int disable_ips_rcg;
+	unsigned int disable_ips_in_vpb;
+	bool disable_ips_in_dpms_off;
+	bool usb4_bw_alloc_support;
+	bool allow_0_dtb_clk;
+	bool use_assr_psp_message;
+	bool support_edp0_on_dp1;
+	unsigned int enable_fpo_flicker_detection;
+	bool disable_hbr_audio_dp2;
+	bool consolidated_dpia_dp_lt;
+	bool set_pipe_unlock_order;
+	bool enable_dpia_pre_training;
+	bool unify_link_enc_assignment;
+	struct spl_sharpness_range dcn_sharpness_range;
+	struct spl_sharpness_range dcn_override_sharpness_range;
+};
+
+enum visual_confirm {
+	VISUAL_CONFIRM_DISABLE = 0,
+	VISUAL_CONFIRM_SURFACE = 1,
+	VISUAL_CONFIRM_HDR = 2,
+	VISUAL_CONFIRM_MPCTREE = 4,
+	VISUAL_CONFIRM_PSR = 5,
+	VISUAL_CONFIRM_SWAPCHAIN = 6,
+	VISUAL_CONFIRM_FAMS = 7,
+	VISUAL_CONFIRM_SWIZZLE = 9,
+	VISUAL_CONFIRM_SMARTMUX_DGPU = 10,
+	VISUAL_CONFIRM_REPLAY = 12,
+	VISUAL_CONFIRM_SUBVP = 14,
+	VISUAL_CONFIRM_MCLK_SWITCH = 16,
+	VISUAL_CONFIRM_FAMS2 = 19,
+	VISUAL_CONFIRM_HW_CURSOR = 20,
+	VISUAL_CONFIRM_VABC = 21,
+	VISUAL_CONFIRM_DCC = 22,
+	VISUAL_CONFIRM_EXPLICIT = 0x80000000,
+};
+
+enum dc_psr_power_opts {
+	psr_power_opt_invalid = 0x0,
+	psr_power_opt_smu_opt_static_screen = 0x1,
+	psr_power_opt_z10_static_screen = 0x10,
+	psr_power_opt_ds_disable_allow = 0x100,
+};
+
+enum dml_hostvm_override_opts {
+	DML_HOSTVM_NO_OVERRIDE = 0x0,
+	DML_HOSTVM_OVERRIDE_FALSE = 0x1,
+	DML_HOSTVM_OVERRIDE_TRUE = 0x2,
+};
+
+enum dc_replay_power_opts {
+	replay_power_opt_invalid		= 0x0,
+	replay_power_opt_smu_opt_static_screen	= 0x1,
+	replay_power_opt_z10_static_screen	= 0x10,
+};
+
+enum dcc_option {
+	DCC_ENABLE = 0,
+	DCC_DISABLE = 1,
+	DCC_HALF_REQ_DISALBE = 2,
+};
+
+enum in_game_fams_config {
+	INGAME_FAMS_SINGLE_DISP_ENABLE, // enable in-game fams
+	INGAME_FAMS_DISABLE, // disable in-game fams
+	INGAME_FAMS_MULTI_DISP_ENABLE, //enable in-game fams for multi-display
+	INGAME_FAMS_MULTI_DISP_CLAMPED_ONLY, //enable in-game fams for multi-display only for clamped RR strategies
+};
+
+/**
+ * enum pipe_split_policy - Pipe split strategy supported by DCN
+ *
+ * This enum is used to define the pipe split policy supported by DCN. By
+ * default, DC favors MPC_SPLIT_DYNAMIC.
+ */
+enum pipe_split_policy {
+	/**
+	 * @MPC_SPLIT_DYNAMIC: DC will automatically decide how to split the
+	 * pipe in order to bring the best trade-off between performance and
+	 * power consumption. This is the recommended option.
+	 */
+	MPC_SPLIT_DYNAMIC = 0,
+
+	/**
+	 * @MPC_SPLIT_AVOID: Avoid pipe split, which means that DC will not
+	 * try any sort of split optimization.
+	 */
+	MPC_SPLIT_AVOID = 1,
+
+	/**
+	 * @MPC_SPLIT_AVOID_MULT_DISP: With this option, DC will only try to
+	 * optimize the pipe utilization when using a single display; if the
+	 * user connects to a second display, DC will avoid pipe split.
+	 */
+	MPC_SPLIT_AVOID_MULT_DISP = 2,
+};
+
+enum wm_report_mode {
+	WM_REPORT_DEFAULT = 0,
+	WM_REPORT_OVERRIDE = 1,
+};
+enum dtm_pstate{
+	dtm_level_p0 = 0,/*highest voltage*/
+	dtm_level_p1,
+	dtm_level_p2,
+	dtm_level_p3,
+	dtm_level_p4,/*when active_display_count = 0*/
+};
+
+enum dcn_pwr_state {
+	DCN_PWR_STATE_UNKNOWN = -1,
+	DCN_PWR_STATE_MISSION_MODE = 0,
+	DCN_PWR_STATE_LOW_POWER = 3,
+};
+
+enum dcn_zstate_support_state {
+	DCN_ZSTATE_SUPPORT_UNKNOWN,
+	DCN_ZSTATE_SUPPORT_ALLOW,
+	DCN_ZSTATE_SUPPORT_ALLOW_Z8_ONLY,
+	DCN_ZSTATE_SUPPORT_ALLOW_Z8_Z10_ONLY,
+	DCN_ZSTATE_SUPPORT_ALLOW_Z10_ONLY,
+	DCN_ZSTATE_SUPPORT_DISALLOW,
+};
+
+/*
+ * struct dc_clocks - DC pipe clocks
+ *
+ * For any clocks that may differ per pipe only the max is stored in this
+ * structure
+ */
+struct dc_clocks {
+	int dispclk_khz;
+	int actual_dispclk_khz;
+	int dppclk_khz;
+	int actual_dppclk_khz;
+	int disp_dpp_voltage_level_khz;
+	int dcfclk_khz;
+	int socclk_khz;
+	int dcfclk_deep_sleep_khz;
+	int fclk_khz;
+	int phyclk_khz;
+	int dramclk_khz;
+	bool p_state_change_support;
+	enum dcn_zstate_support_state zstate_support;
+	bool dtbclk_en;
+	int ref_dtbclk_khz;
+	bool fclk_p_state_change_support;
+	enum dcn_pwr_state pwr_state;
+	/*
+	 * Elements below are not compared for the purposes of
+	 * optimization required
+	 */
+	bool prev_p_state_change_support;
+	bool fclk_prev_p_state_change_support;
+	int num_ways;
+	int host_router_bw_kbps[MAX_HOST_ROUTERS_NUM];
+
+	/*
+	 * @fw_based_mclk_switching
+	 *
+	 * DC has a mechanism that leverage the variable refresh rate to switch
+	 * memory clock in cases that we have a large latency to achieve the
+	 * memory clock change and a short vblank window. DC has some
+	 * requirements to enable this feature, and this field describes if the
+	 * system support or not such a feature.
+	 */
+	bool fw_based_mclk_switching;
+	bool fw_based_mclk_switching_shut_down;
+	int prev_num_ways;
+	enum dtm_pstate dtm_level;
+	int max_supported_dppclk_khz;
+	int max_supported_dispclk_khz;
+	int bw_dppclk_khz; /*a copy of dppclk_khz*/
+	int bw_dispclk_khz;
+	int idle_dramclk_khz;
+	int idle_fclk_khz;
+	int subvp_prefetch_dramclk_khz;
+	int subvp_prefetch_fclk_khz;
+
+	/* Stutter efficiency is technically not clock values
+	 * but stored here so the values are part of the update_clocks call similar to num_ways
+	 * Efficiencies are stored as percentage (0-100)
+	 */
+	struct {
+		uint8_t base_efficiency; //LP1
+		uint8_t low_power_efficiency; //LP2
+	} stutter_efficiency;
+};
+
+struct dc_bw_validation_profile {
+	bool enable;
+
+	unsigned long long total_ticks;
+	unsigned long long voltage_level_ticks;
+	unsigned long long watermark_ticks;
+	unsigned long long rq_dlg_ticks;
+
+	unsigned long long total_count;
+	unsigned long long skip_fast_count;
+	unsigned long long skip_pass_count;
+	unsigned long long skip_fail_count;
+};
+
+#define BW_VAL_TRACE_SETUP() \
+		unsigned long long end_tick = 0; \
+		unsigned long long voltage_level_tick = 0; \
+		unsigned long long watermark_tick = 0; \
+		unsigned long long start_tick = dc->debug.bw_val_profile.enable ? \
+				dm_get_timestamp(dc->ctx) : 0
+
+#define BW_VAL_TRACE_COUNT() \
+		if (dc->debug.bw_val_profile.enable) \
+			dc->debug.bw_val_profile.total_count++
+
+#define BW_VAL_TRACE_SKIP(status) \
+		if (dc->debug.bw_val_profile.enable) { \
+			if (!voltage_level_tick) \
+				voltage_level_tick = dm_get_timestamp(dc->ctx); \
+			dc->debug.bw_val_profile.skip_ ## status ## _count++; \
+		}
+
+#define BW_VAL_TRACE_END_VOLTAGE_LEVEL() \
+		if (dc->debug.bw_val_profile.enable) \
+			voltage_level_tick = dm_get_timestamp(dc->ctx)
+
+#define BW_VAL_TRACE_END_WATERMARKS() \
+		if (dc->debug.bw_val_profile.enable) \
+			watermark_tick = dm_get_timestamp(dc->ctx)
+
+#define BW_VAL_TRACE_FINISH() \
+		if (dc->debug.bw_val_profile.enable) { \
+			end_tick = dm_get_timestamp(dc->ctx); \
+			dc->debug.bw_val_profile.total_ticks += end_tick - start_tick; \
+			dc->debug.bw_val_profile.voltage_level_ticks += voltage_level_tick - start_tick; \
+			if (watermark_tick) { \
+				dc->debug.bw_val_profile.watermark_ticks += watermark_tick - voltage_level_tick; \
+				dc->debug.bw_val_profile.rq_dlg_ticks += end_tick - watermark_tick; \
+			} \
+		}
+
+union mem_low_power_enable_options {
+	struct {
+		bool vga: 1;
+		bool i2c: 1;
+		bool dmcu: 1;
+		bool dscl: 1;
+		bool cm: 1;
+		bool mpc: 1;
+		bool optc: 1;
+		bool vpg: 1;
+		bool afmt: 1;
+	} bits;
+	uint32_t u32All;
+};
+
+union root_clock_optimization_options {
+	struct {
+		bool dpp: 1;
+		bool dsc: 1;
+		bool hdmistream: 1;
+		bool hdmichar: 1;
+		bool dpstream: 1;
+		bool symclk32_se: 1;
+		bool symclk32_le: 1;
+		bool symclk_fe: 1;
+		bool physymclk: 1;
+		bool dpiasymclk: 1;
+		uint32_t reserved: 22;
+	} bits;
+	uint32_t u32All;
+};
+
+union fine_grain_clock_gating_enable_options {
+	struct {
+		bool dccg_global_fgcg_rep : 1; /* Global fine grain clock gating of repeaters */
+		bool dchub : 1;	   /* Display controller hub */
+		bool dchubbub : 1;
+		bool dpp : 1;	   /* Display pipes and planes */
+		bool opp : 1;	   /* Output pixel processing */
+		bool optc : 1;	   /* Output pipe timing combiner */
+		bool dio : 1;	   /* Display output */
+		bool dwb : 1;	   /* Display writeback */
+		bool mmhubbub : 1; /* Multimedia hub */
+		bool dmu : 1;	   /* Display core management unit */
+		bool az : 1;	   /* Azalia */
+		bool dchvm : 1;
+		bool dsc : 1;	   /* Display stream compression */
+
+		uint32_t reserved : 19;
+	} bits;
+	uint32_t u32All;
+};
+
+enum pg_hw_pipe_resources {
+	PG_HUBP = 0,
+	PG_DPP,
+	PG_DSC,
+	PG_MPCC,
+	PG_OPP,
+	PG_OPTC,
+	PG_DPSTREAM,
+	PG_HDMISTREAM,
+	PG_PHYSYMCLK,
+	PG_HW_PIPE_RESOURCES_NUM_ELEMENT
+};
+
+enum pg_hw_resources {
+	PG_DCCG = 0,
+	PG_DCIO,
+	PG_DIO,
+	PG_DCHUBBUB,
+	PG_DCHVM,
+	PG_DWB,
+	PG_HPO,
+	PG_DCOH,
+	PG_HW_RESOURCES_NUM_ELEMENT
+};
+
+struct pg_block_update {
+	bool pg_pipe_res_update[PG_HW_PIPE_RESOURCES_NUM_ELEMENT][MAX_PIPES];
+	bool pg_res_update[PG_HW_RESOURCES_NUM_ELEMENT];
+};
+
+union dpia_debug_options {
+	struct {
+		uint32_t disable_dpia:1; /* bit 0 */
+		uint32_t force_non_lttpr:1; /* bit 1 */
+		uint32_t extend_aux_rd_interval:1; /* bit 2 */
+		uint32_t disable_mst_dsc_work_around:1; /* bit 3 */
+		uint32_t enable_force_tbt3_work_around:1; /* bit 4 */
+		uint32_t disable_usb4_pm_support:1; /* bit 5 */
+		uint32_t enable_usb4_bw_zero_alloc_patch:1; /* bit 6 */
+		uint32_t enable_bw_allocation_mode:1; /* bit 7 */
+		uint32_t reserved:24;
+	} bits;
+	uint32_t raw;
+};
+
+/* AUX wake work around options
+ * 0: enable/disable work around
+ * 1: use default timeout LINK_AUX_WAKE_TIMEOUT_MS
+ * 15-2: reserved
+ * 31-16: timeout in ms
+ */
+union aux_wake_wa_options {
+	struct {
+		uint32_t enable_wa : 1;
+		uint32_t use_default_timeout : 1;
+		uint32_t rsvd: 14;
+		uint32_t timeout_ms : 16;
+	} bits;
+	uint32_t raw;
+};
+
+struct dc_debug_data {
+	uint32_t ltFailCount;
+	uint32_t i2cErrorCount;
+	uint32_t auxErrorCount;
+};
+
+struct dc_phy_addr_space_config {
+	struct {
+		uint64_t start_addr;
+		uint64_t end_addr;
+		uint64_t fb_top;
+		uint64_t fb_offset;
+		uint64_t fb_base;
+		uint64_t agp_top;
+		uint64_t agp_bot;
+		uint64_t agp_base;
+	} system_aperture;
+
+	struct {
+		uint64_t page_table_start_addr;
+		uint64_t page_table_end_addr;
+		uint64_t page_table_base_addr;
+		bool base_addr_is_mc_addr;
+	} gart_config;
+
+	bool valid;
+	bool is_hvm_enabled;
+	uint64_t page_table_default_page_addr;
+};
+
+struct dc_virtual_addr_space_config {
+	uint64_t	page_table_base_addr;
+	uint64_t	page_table_start_addr;
+	uint64_t	page_table_end_addr;
+	uint32_t	page_table_block_size_in_bytes;
+	uint8_t		page_table_depth; // 1 = 1 level, 2 = 2 level, etc.  0 = invalid
+};
+
+struct dc_bounding_box_overrides {
+	int sr_exit_time_ns;
+	int sr_enter_plus_exit_time_ns;
+	int sr_exit_z8_time_ns;
+	int sr_enter_plus_exit_z8_time_ns;
+	int urgent_latency_ns;
+	int percent_of_ideal_drambw;
+	int dram_clock_change_latency_ns;
+	int dummy_clock_change_latency_ns;
+	int fclk_clock_change_latency_ns;
+	/* This forces a hard min on the DCFCLK we use
+	 * for DML.  Unlike the debug option for forcing
+	 * DCFCLK, this override affects watermark calculations
+	 */
+	int min_dcfclk_mhz;
+};
+
+struct dc_state;
+struct resource_pool;
+struct dce_hwseq;
+struct link_service;
+
+/*
+ * struct dc_debug_options - DC debug struct
+ *
+ * This struct provides a simple mechanism for developers to change some
+ * configurations, enable/disable features, and activate extra debug options.
+ * This can be very handy to narrow down whether some specific feature is
+ * causing an issue or not.
+ */
+struct dc_debug_options {
+	bool native422_support;
+	bool disable_dsc;
+	enum visual_confirm visual_confirm;
+	int visual_confirm_rect_height;
+
+	bool sanity_checks;
+	bool max_disp_clk;
+	bool surface_trace;
+	bool clock_trace;
+	bool validation_trace;
+	bool bandwidth_calcs_trace;
+	int max_downscale_src_width;
+
+	/* stutter efficiency related */
+	bool disable_stutter;
+	bool use_max_lb;
+	enum dcc_option disable_dcc;
+
+	/*
+	 * @pipe_split_policy: Define which pipe split policy is used by the
+	 * display core.
+	 */
+	enum pipe_split_policy pipe_split_policy;
+	bool force_single_disp_pipe_split;
+	bool voltage_align_fclk;
+	bool disable_min_fclk;
+
+	bool hdcp_lc_force_fw_enable;
+	bool hdcp_lc_enable_sw_fallback;
+
+	bool disable_dfs_bypass;
+	bool disable_dpp_power_gate;
+	bool disable_hubp_power_gate;
+	bool disable_dsc_power_gate;
+	bool disable_optc_power_gate;
+	bool disable_hpo_power_gate;
+	bool disable_io_clk_power_gate;
+	bool disable_mem_power_gate;
+	bool disable_dio_power_gate;
+	int dsc_min_slice_height_override;
+	int dsc_bpp_increment_div;
+	bool disable_pplib_wm_range;
+	enum wm_report_mode pplib_wm_report_mode;
+	unsigned int min_disp_clk_khz;
+	unsigned int min_dpp_clk_khz;
+	unsigned int min_dram_clk_khz;
+	int sr_exit_time_dpm0_ns;
+	int sr_enter_plus_exit_time_dpm0_ns;
+	int sr_exit_time_ns;
+	int sr_enter_plus_exit_time_ns;
+	int sr_exit_z8_time_ns;
+	int sr_enter_plus_exit_z8_time_ns;
+	int urgent_latency_ns;
+	uint32_t underflow_assert_delay_us;
+	int percent_of_ideal_drambw;
+	int dram_clock_change_latency_ns;
+	bool optimized_watermark;
+	int always_scale;
+	bool disable_pplib_clock_request;
+	bool disable_clock_gate;
+	bool disable_mem_low_power;
+	bool pstate_enabled;
+	bool disable_dmcu;
+	bool force_abm_enable;
+	bool disable_stereo_support;
+	bool vsr_support;
+	bool performance_trace;
+	bool az_endpoint_mute_only;
+	bool always_use_regamma;
+	bool recovery_enabled;
+	bool avoid_vbios_exec_table;
+	bool scl_reset_length10;
+	bool hdmi20_disable;
+	bool skip_detection_link_training;
+	uint32_t edid_read_retry_times;
+	unsigned int force_odm_combine; //bit vector based on otg inst
+	unsigned int seamless_boot_odm_combine;
+	unsigned int force_odm_combine_4to1; //bit vector based on otg inst
+	int minimum_z8_residency_time;
+	int minimum_z10_residency_time;
+	bool disable_z9_mpc;
+	unsigned int force_fclk_khz;
+	bool enable_tri_buf;
+	bool ips_disallow_entry;
+	bool dmub_offload_enabled;
+	bool dmcub_emulation;
+	bool disable_idle_power_optimizations;
+	unsigned int mall_size_override;
+	unsigned int mall_additional_timer_percent;
+	bool mall_error_as_fatal;
+	bool dmub_command_table; /* for testing only */
+	struct dc_bw_validation_profile bw_val_profile;
+	bool disable_fec;
+	bool disable_48mhz_pwrdwn;
+	/* This forces a hard min on the DCFCLK requested to SMU/PP
+	 * watermarks are not affected.
+	 */
+	unsigned int force_min_dcfclk_mhz;
+	int dwb_fi_phase;
+	bool disable_timing_sync;
+	bool cm_in_bypass;
+	int force_clock_mode;/*every mode change.*/
+
+	bool disable_dram_clock_change_vactive_support;
+	bool validate_dml_output;
+	bool enable_dmcub_surface_flip;
+	bool usbc_combo_phy_reset_wa;
+	bool enable_dram_clock_change_one_display_vactive;
+	/* TODO - remove once tested */
+	bool legacy_dp2_lt;
+	bool set_mst_en_for_sst;
+	bool disable_uhbr;
+	bool force_dp2_lt_fallback_method;
+	bool ignore_cable_id;
+	union mem_low_power_enable_options enable_mem_low_power;
+	union root_clock_optimization_options root_clock_optimization;
+	union fine_grain_clock_gating_enable_options enable_fine_grain_clock_gating;
+	bool hpo_optimization;
+	bool force_vblank_alignment;
+
+	/* Enable dmub aux for legacy ddc */
+	bool enable_dmub_aux_for_legacy_ddc;
+	bool disable_fams;
+	enum in_game_fams_config disable_fams_gaming;
+	/* FEC/PSR1 sequence enable delay in 100us */
+	uint8_t fec_enable_delay_in100us;
+	bool enable_driver_sequence_debug;
+	enum det_size crb_alloc_policy;
+	int crb_alloc_policy_min_disp_count;
+	bool disable_z10;
+	bool enable_z9_disable_interface;
+	bool psr_skip_crtc_disable;
+	uint32_t ips_skip_crtc_disable_mask;
+	union dpia_debug_options dpia_debug;
+	bool disable_fixed_vs_aux_timeout_wa;
+	uint32_t fixed_vs_aux_delay_config_wa;
+	bool force_disable_subvp;
+	bool force_subvp_mclk_switch;
+	bool allow_sw_cursor_fallback;
+	unsigned int force_subvp_num_ways;
+	unsigned int force_mall_ss_num_ways;
+	bool alloc_extra_way_for_cursor;
+	uint32_t subvp_extra_lines;
+	bool disable_force_pstate_allow_on_hw_release;
+	bool force_usr_allow;
+	/* uses value at boot and disables switch */
+	bool disable_dtb_ref_clk_switch;
+	bool extended_blank_optimization;
+	union aux_wake_wa_options aux_wake_wa;
+	uint32_t mst_start_top_delay;
+	uint8_t psr_power_use_phy_fsm;
+	enum dml_hostvm_override_opts dml_hostvm_override;
+	bool dml_disallow_alternate_prefetch_modes;
+	bool use_legacy_soc_bb_mechanism;
+	bool exit_idle_opt_for_cursor_updates;
+	bool using_dml2;
+	bool enable_single_display_2to1_odm_policy;
+	bool enable_double_buffered_dsc_pg_support;
+	bool enable_dp_dig_pixel_rate_div_policy;
+	bool using_dml21;
+	enum lttpr_mode lttpr_mode_override;
+	unsigned int dsc_delay_factor_wa_x1000;
+	unsigned int min_prefetch_in_strobe_ns;
+	bool disable_unbounded_requesting;
+	bool dig_fifo_off_in_blank;
+	bool override_dispclk_programming;
+	bool otg_crc_db;
+	bool disallow_dispclk_dppclk_ds;
+	bool disable_fpo_optimizations;
+	bool support_eDP1_5;
+	uint32_t fpo_vactive_margin_us;
+	bool disable_fpo_vactive;
+	bool disable_boot_optimizations;
+	bool override_odm_optimization;
+	bool minimize_dispclk_using_odm;
+	bool disable_subvp_high_refresh;
+	bool disable_dp_plus_plus_wa;
+	uint32_t fpo_vactive_min_active_margin_us;
+	uint32_t fpo_vactive_max_blank_us;
+	bool enable_hpo_pg_support;
+	bool enable_legacy_fast_update;
+	bool disable_dc_mode_overwrite;
+	bool replay_skip_crtc_disabled;
+	bool ignore_pg;/*do nothing, let pmfw control it*/
+	bool psp_disabled_wa;
+	unsigned int ips2_eval_delay_us;
+	unsigned int ips2_entry_delay_us;
+	bool optimize_ips_handshake;
+	bool disable_dmub_reallow_idle;
+	bool disable_timeout;
+	bool disable_extblankadj;
+	bool enable_idle_reg_checks;
+	unsigned int static_screen_wait_frames;
+	uint32_t pwm_freq;
+	bool force_chroma_subsampling_1tap;
+	unsigned int dcc_meta_propagation_delay_us;
+	bool disable_422_left_edge_pixel;
+	bool dml21_force_pstate_method;
+	uint32_t dml21_force_pstate_method_values[MAX_PIPES];
+	uint32_t dml21_disable_pstate_method_mask;
+	union fw_assisted_mclk_switch_version fams_version;
+	union dmub_fams2_global_feature_config fams2_config;
+	unsigned int force_cositing;
+	unsigned int disable_spl;
+	unsigned int force_easf;
+	unsigned int force_sharpness;
+	unsigned int force_sharpness_level;
+	unsigned int force_lls;
+	bool notify_dpia_hr_bw;
+	bool enable_ips_visual_confirm;
+	unsigned int sharpen_policy;
+	unsigned int scale_to_sharpness_policy;
+	bool skip_full_updated_if_possible;
+	unsigned int enable_oled_edp_power_up_opt;
+	bool enable_hblank_borrow;
+	bool force_subvp_df_throttle;
+	uint32_t acpi_transition_bitmasks[MAX_PIPES];
+	bool enable_pg_cntl_debug_logs;
+	unsigned int auxless_alpm_lfps_setup_ns;
+	unsigned int auxless_alpm_lfps_period_ns;
+	unsigned int auxless_alpm_lfps_silence_ns;
+	unsigned int auxless_alpm_lfps_t1t2_us;
+	short auxless_alpm_lfps_t1t2_offset_us;
+	bool disable_stutter_for_wm_program;
+};
+
+
+/* Generic structure that can be used to query properties of DC. More fields
+ * can be added as required.
+ */
+struct dc_current_properties {
+	unsigned int cursor_size_limit;
+};
+
+enum frame_buffer_mode {
+	FRAME_BUFFER_MODE_LOCAL_ONLY = 0,
+	FRAME_BUFFER_MODE_ZFB_ONLY,
+	FRAME_BUFFER_MODE_MIXED_ZFB_AND_LOCAL,
+} ;
+
+struct dchub_init_data {
+	int64_t zfb_phys_addr_base;
+	int64_t zfb_mc_base_addr;
+	uint64_t zfb_size_in_byte;
+	enum frame_buffer_mode fb_mode;
+	bool dchub_initialzied;
+	bool dchub_info_valid;
+};
+
+struct dml2_soc_bb;
+
+struct dc_init_data {
+	struct hw_asic_id asic_id;
+	void *driver; /* ctx */
+	struct cgs_device *cgs_device;
+	struct dc_bounding_box_overrides bb_overrides;
+
+	int num_virtual_links;
+	/*
+	 * If 'vbios_override' not NULL, it will be called instead
+	 * of the real VBIOS. Intended use is Diagnostics on FPGA.
+	 */
+	struct dc_bios *vbios_override;
+	enum dce_environment dce_environment;
+
+	struct dmub_offload_funcs *dmub_if;
+	struct dc_reg_helper_state *dmub_offload;
+
+	struct dc_config flags;
+	uint64_t log_mask;
+
+	struct dpcd_vendor_signature vendor_signature;
+	bool force_smu_not_present;
+	/*
+	 * IP offset for run time initializaion of register addresses
+	 *
+	 * DCN3.5+ will fail dc_create() if these fields are null for them. They are
+	 * applicable starting with DCN32/321 and are not used for ASICs upstreamed
+	 * before them.
+	 */
+	uint32_t *dcn_reg_offsets;
+	uint32_t *nbio_reg_offsets;
+	uint32_t *clk_reg_offsets;
+	void *bb_from_dmub;
+};
+
+struct dc_callback_init {
+	struct cp_psp cp_psp;
+};
+
+struct dc *dc_create(const struct dc_init_data *init_params);
+void dc_hardware_init(struct dc *dc);
+
+int dc_get_vmid_use_vector(struct dc *dc);
+void dc_setup_vm_context(struct dc *dc, struct dc_virtual_addr_space_config *va_config, int vmid);
+/* Returns the number of vmids supported */
+int dc_setup_system_context(struct dc *dc, struct dc_phy_addr_space_config *pa_config);
+void dc_init_callbacks(struct dc *dc,
+		const struct dc_callback_init *init_params);
+void dc_deinit_callbacks(struct dc *dc);
+void dc_destroy(struct dc **dc);
+
+/* Surface Interfaces */
+
+enum {
+	TRANSFER_FUNC_POINTS = 1025
+};
+
+struct dc_hdr_static_metadata {
+	/* display chromaticities and white point in units of 0.00001 */
+	unsigned int chromaticity_green_x;
+	unsigned int chromaticity_green_y;
+	unsigned int chromaticity_blue_x;
+	unsigned int chromaticity_blue_y;
+	unsigned int chromaticity_red_x;
+	unsigned int chromaticity_red_y;
+	unsigned int chromaticity_white_point_x;
+	unsigned int chromaticity_white_point_y;
+
+	uint32_t min_luminance;
+	uint32_t max_luminance;
+	uint32_t maximum_content_light_level;
+	uint32_t maximum_frame_average_light_level;
+};
+
+enum dc_transfer_func_type {
+	TF_TYPE_PREDEFINED,
+	TF_TYPE_DISTRIBUTED_POINTS,
+	TF_TYPE_BYPASS,
+	TF_TYPE_HWPWL
+};
+
+struct dc_transfer_func_distributed_points {
+	struct fixed31_32 red[TRANSFER_FUNC_POINTS];
+	struct fixed31_32 green[TRANSFER_FUNC_POINTS];
+	struct fixed31_32 blue[TRANSFER_FUNC_POINTS];
+
+	uint16_t end_exponent;
+	uint16_t x_point_at_y1_red;
+	uint16_t x_point_at_y1_green;
+	uint16_t x_point_at_y1_blue;
+};
+
+enum dc_transfer_func_predefined {
+	TRANSFER_FUNCTION_SRGB,
+	TRANSFER_FUNCTION_BT709,
+	TRANSFER_FUNCTION_PQ,
+	TRANSFER_FUNCTION_LINEAR,
+	TRANSFER_FUNCTION_UNITY,
+	TRANSFER_FUNCTION_HLG,
+	TRANSFER_FUNCTION_HLG12,
+	TRANSFER_FUNCTION_GAMMA22,
+	TRANSFER_FUNCTION_GAMMA24,
+	TRANSFER_FUNCTION_GAMMA26
+};
+
+
+struct dc_transfer_func {
+	struct kref refcount;
+	enum dc_transfer_func_type type;
+	enum dc_transfer_func_predefined tf;
+	/* FP16 1.0 reference level in nits, default is 80 nits, only for PQ*/
+	uint32_t sdr_ref_white_level;
+	union {
+		struct pwl_params pwl;
+		struct dc_transfer_func_distributed_points tf_pts;
+	};
+};
+
+
+union dc_3dlut_state {
+	struct {
+		uint32_t initialized:1;		/*if 3dlut is went through color module for initialization */
+		uint32_t rmu_idx_valid:1;	/*if mux settings are valid*/
+		uint32_t rmu_mux_num:3;		/*index of mux to use*/
+		uint32_t mpc_rmu0_mux:4;	/*select mpcc on mux, one of the following : mpcc0, mpcc1, mpcc2, mpcc3*/
+		uint32_t mpc_rmu1_mux:4;
+		uint32_t mpc_rmu2_mux:4;
+		uint32_t reserved:15;
+	} bits;
+	uint32_t raw;
+};
+
+
+#define MATRIX_9C__DIM_128_ALIGNED_LEN   16 // 9+8 :  9 * 8 +  7 * 8 = 72  + 56  = 128 % 128 = 0
+#define MATRIX_17C__DIM_128_ALIGNED_LEN  32 //17+15:  17 * 8 + 15 * 8 = 136 + 120 = 256 % 128 = 0
+#define MATRIX_33C__DIM_128_ALIGNED_LEN  64 //17+47:  17 * 8 + 47 * 8 = 136 + 376 = 512 % 128 = 0
+
+struct lut_rgb {
+	uint16_t b;
+	uint16_t g;
+	uint16_t r;
+	uint16_t padding;
+};
+
+//this structure maps directly to how the lut will read it from memory
+struct lut_mem_mapping {
+	union {
+		//NATIVE MODE 1, 2
+		//RGB layout          [b][g][r]      //red  is 128 byte aligned
+		//BGR layout          [r][g][b]      //blue is 128 byte aligned
+		struct lut_rgb rgb_17c[17][17][MATRIX_17C__DIM_128_ALIGNED_LEN];
+		struct lut_rgb rgb_33c[33][33][MATRIX_33C__DIM_128_ALIGNED_LEN];
+
+		//TRANSFORMED
+		uint16_t linear_rgb[(33*33*33*4/128+1)*128];
+	};
+	uint16_t size;
+};
+
+struct dc_rmcm_3dlut {
+	bool isInUse;
+	const struct dc_stream_state *stream;
+	uint8_t protection_bits;
+};
+
+struct dc_3dlut {
+	struct kref refcount;
+	struct tetrahedral_params lut_3d;
+	struct fixed31_32 hdr_multiplier;
+	union dc_3dlut_state state;
+};
+/*
+ * This structure is filled in by dc_surface_get_status and contains
+ * the last requested address and the currently active address so the called
+ * can determine if there are any outstanding flips
+ */
+struct dc_plane_status {
+	struct dc_plane_address requested_address;
+	struct dc_plane_address current_address;
+	bool is_flip_pending;
+	bool is_right_eye;
+};
+
+union surface_update_flags {
+
+	struct {
+		uint32_t addr_update:1;
+		/* Medium updates */
+		uint32_t dcc_change:1;
+		uint32_t color_space_change:1;
+		uint32_t horizontal_mirror_change:1;
+		uint32_t per_pixel_alpha_change:1;
+		uint32_t global_alpha_change:1;
+		uint32_t hdr_mult:1;
+		uint32_t rotation_change:1;
+		uint32_t swizzle_change:1;
+		uint32_t scaling_change:1;
+		uint32_t position_change:1;
+		uint32_t in_transfer_func_change:1;
+		uint32_t input_csc_change:1;
+		uint32_t coeff_reduction_change:1;
+		uint32_t pixel_format_change:1;
+		uint32_t plane_size_change:1;
+		uint32_t gamut_remap_change:1;
+
+		/* Full updates */
+		uint32_t new_plane:1;
+		uint32_t bpp_change:1;
+		uint32_t gamma_change:1;
+		uint32_t bandwidth_change:1;
+		uint32_t clock_change:1;
+		uint32_t stereo_format_change:1;
+		uint32_t lut_3d:1;
+		uint32_t tmz_changed:1;
+		uint32_t mcm_transfer_function_enable_change:1; /* disable or enable MCM transfer func */
+		uint32_t full_update:1;
+		uint32_t sdr_white_level_nits:1;
+	} bits;
+
+	uint32_t raw;
+};
+
+#define DC_REMOVE_PLANE_POINTERS 1
+
+struct dc_plane_state {
+	struct dc_plane_address address;
+	struct dc_plane_flip_time time;
+	bool triplebuffer_flips;
+	struct scaling_taps scaling_quality;
+	struct rect src_rect;
+	struct rect dst_rect;
+	struct rect clip_rect;
+
+	struct plane_size plane_size;
+	struct dc_tiling_info tiling_info;
+
+	struct dc_plane_dcc_param dcc;
+
+	struct dc_gamma gamma_correction;
+	struct dc_transfer_func in_transfer_func;
+	struct dc_bias_and_scale bias_and_scale;
+	struct dc_csc_transform input_csc_color_matrix;
+	struct fixed31_32 coeff_reduction_factor;
+	struct fixed31_32 hdr_mult;
+	struct colorspace_transform gamut_remap_matrix;
+
+	// TODO: No longer used, remove
+	struct dc_hdr_static_metadata hdr_static_ctx;
+
+	enum dc_color_space color_space;
+
+	struct dc_3dlut lut3d_func;
+	struct dc_transfer_func in_shaper_func;
+	struct dc_transfer_func blend_tf;
+
+	struct dc_transfer_func *gamcor_tf;
+	enum surface_pixel_format format;
+	enum dc_rotation_angle rotation;
+	enum plane_stereo_format stereo_format;
+
+	bool is_tiling_rotated;
+	bool per_pixel_alpha;
+	bool pre_multiplied_alpha;
+	bool global_alpha;
+	int  global_alpha_value;
+	bool visible;
+	bool flip_immediate;
+	bool horizontal_mirror;
+	int layer_index;
+
+	union surface_update_flags update_flags;
+	bool flip_int_enabled;
+	bool skip_manual_trigger;
+
+	/* private to DC core */
+	struct dc_plane_status status;
+	struct dc_context *ctx;
+
+	/* HACK: Workaround for forcing full reprogramming under some conditions */
+	bool force_full_update;
+
+	bool is_phantom; // TODO: Change mall_stream_config into mall_plane_config instead
+
+	/* private to dc_surface.c */
+	enum dc_irq_source irq_source;
+	struct kref refcount;
+	struct tg_color visual_confirm_color;
+
+	bool is_statically_allocated;
+	enum chroma_cositing cositing;
+	enum dc_cm2_shaper_3dlut_setting mcm_shaper_3dlut_setting;
+	bool mcm_lut1d_enable;
+	struct dc_cm2_func_luts mcm_luts;
+	bool lut_bank_a;
+	enum mpcc_movable_cm_location mcm_location;
+	struct dc_csc_transform cursor_csc_color_matrix;
+	bool adaptive_sharpness_en;
+	int adaptive_sharpness_policy;
+	int sharpness_level;
+	enum linear_light_scaling linear_light_scaling;
+	unsigned int sdr_white_level_nits;
+	struct spl_sharpness_range sharpness_range;
+	enum sharpness_range_source sharpness_source;
+};
+
+struct dc_plane_info {
+	struct plane_size plane_size;
+	struct dc_tiling_info tiling_info;
+	struct dc_plane_dcc_param dcc;
+	enum surface_pixel_format format;
+	enum dc_rotation_angle rotation;
+	enum plane_stereo_format stereo_format;
+	enum dc_color_space color_space;
+	bool horizontal_mirror;
+	bool visible;
+	bool per_pixel_alpha;
+	bool pre_multiplied_alpha;
+	bool global_alpha;
+	int  global_alpha_value;
+	bool input_csc_enabled;
+	int layer_index;
+	enum chroma_cositing cositing;
+};
+
+#include "dc_stream.h"
+
+struct dc_scratch_space {
+	/* used to temporarily backup plane states of a stream during
+	 * dc update. The reason is that plane states are overwritten
+	 * with surface updates in dc update. Once they are overwritten
+	 * current state is no longer valid. We want to temporarily
+	 * store current value in plane states so we can still recover
+	 * a valid current state during dc update.
+	 */
+	struct dc_plane_state plane_states[MAX_SURFACES];
+
+	struct dc_stream_state stream_state;
+};
+
+/*
+ * A link contains one or more sinks and their connected status.
+ * The currently active signal type (HDMI, DP-SST, DP-MST) is also reported.
+ */
+ struct dc_link {
+	struct dc_sink *remote_sinks[MAX_SINKS_PER_LINK];
+	unsigned int sink_count;
+	struct dc_sink *local_sink;
+	unsigned int link_index;
+	enum dc_connection_type type;
+	enum signal_type connector_signal;
+	enum dc_irq_source irq_source_hpd;
+	enum dc_irq_source irq_source_hpd_rx;/* aka DP Short Pulse  */
+	enum dc_irq_source irq_source_read_request;/* Read Request */
+
+	bool is_hpd_filter_disabled;
+	bool dp_ss_off;
+
+	/**
+	 * @link_state_valid:
+	 *
+	 * If there is no link and local sink, this variable should be set to
+	 * false. Otherwise, it should be set to true; usually, the function
+	 * core_link_enable_stream sets this field to true.
+	 */
+	bool link_state_valid;
+	bool aux_access_disabled;
+	bool sync_lt_in_progress;
+	bool skip_stream_reenable;
+	bool is_internal_display;
+	/** @todo Rename. Flag an endpoint as having a programmable mapping to a DIG encoder. */
+	bool is_dig_mapping_flexible;
+	bool hpd_status; /* HPD status of link without physical HPD pin. */
+	bool is_hpd_pending; /* Indicates a new received hpd */
+
+	/* USB4 DPIA links skip verifying link cap, instead performing the fallback method
+	 * for every link training. This is incompatible with DP LL compliance automation,
+	 * which expects the same link settings to be used every retry on a link loss.
+	 * This flag is used to skip the fallback when link loss occurs during automation.
+	 */
+	bool skip_fallback_on_link_loss;
+
+	bool edp_sink_present;
+
+	struct dp_trace dp_trace;
+
+	/* caps is the same as reported_link_cap. link_traing use
+	 * reported_link_cap. Will clean up.  TODO
+	 */
+	struct dc_link_settings reported_link_cap;
+	struct dc_link_settings verified_link_cap;
+	struct dc_link_settings cur_link_settings;
+	struct dc_lane_settings cur_lane_setting[LANE_COUNT_DP_MAX];
+	struct dc_link_settings preferred_link_setting;
+	/* preferred_training_settings are override values that
+	 * come from DM. DM is responsible for the memory
+	 * management of the override pointers.
+	 */
+	struct dc_link_training_overrides preferred_training_settings;
+	struct dp_audio_test_data audio_test_data;
+
+	uint8_t ddc_hw_inst;
+
+	uint8_t hpd_src;
+
+	uint8_t link_enc_hw_inst;
+	/* DIG link encoder ID. Used as index in link encoder resource pool.
+	 * For links with fixed mapping to DIG, this is not changed after dc_link
+	 * object creation.
+	 */
+	enum engine_id eng_id;
+	enum engine_id dpia_preferred_eng_id;
+
+	bool test_pattern_enabled;
+	/* Pending/Current test pattern are only used to perform and track
+	 * FIXED_VS retimer test pattern/lane adjustment override state.
+	 * Pending allows link HWSS to differentiate PHY vs non-PHY pattern,
+	 * to perform specific lane adjust overrides before setting certain
+	 * PHY test patterns. In cases when lane adjust and set test pattern
+	 * calls are not performed atomically (i.e. performing link training),
+	 * pending_test_pattern will be invalid or contain a non-PHY test pattern
+	 * and current_test_pattern will contain required context for any future
+	 * set pattern/set lane adjust to transition between override state(s).
+	 * */
+	enum dp_test_pattern current_test_pattern;
+	enum dp_test_pattern pending_test_pattern;
+
+	union compliance_test_state compliance_test_state;
+
+	void *priv;
+
+	struct ddc_service *ddc;
+
+	enum dp_panel_mode panel_mode;
+	bool aux_mode;
+
+	/* Private to DC core */
+
+	const struct dc *dc;
+
+	struct dc_context *ctx;
+
+	struct panel_cntl *panel_cntl;
+	struct link_encoder *link_enc;
+	struct graphics_object_id link_id;
+	/* Endpoint type distinguishes display endpoints which do not have entries
+	 * in the BIOS connector table from those that do. Helps when tracking link
+	 * encoder to display endpoint assignments.
+	 */
+	enum display_endpoint_type ep_type;
+	union ddi_channel_mapping ddi_channel_mapping;
+	struct connector_device_tag_info device_tag;
+	struct dpcd_caps dpcd_caps;
+	uint32_t dongle_max_pix_clk;
+	unsigned short chip_caps;
+	unsigned int dpcd_sink_count;
+	struct hdcp_caps hdcp_caps;
+	enum edp_revision edp_revision;
+	union dpcd_sink_ext_caps dpcd_sink_ext_caps;
+
+	struct psr_settings psr_settings;
+	struct replay_settings replay_settings;
+
+	/* Drive settings read from integrated info table */
+	struct dc_lane_settings bios_forced_drive_settings;
+
+	/* Vendor specific LTTPR workaround variables */
+	uint8_t vendor_specific_lttpr_link_rate_wa;
+	bool apply_vendor_specific_lttpr_link_rate_wa;
+
+	/* MST record stream using this link */
+	struct link_flags {
+		bool dp_keep_receiver_powered;
+		bool dp_skip_DID2;
+		bool dp_skip_reset_segment;
+		bool dp_skip_fs_144hz;
+		bool dp_mot_reset_segment;
+		/* Some USB4 docks do not handle turning off MST DSC once it has been enabled. */
+		bool dpia_mst_dsc_always_on;
+		/* Forced DPIA into TBT3 compatibility mode. */
+		bool dpia_forced_tbt3_mode;
+		bool dongle_mode_timing_override;
+		bool blank_stream_on_ocs_change;
+		bool read_dpcd204h_on_irq_hpd;
+		bool force_dp_ffe_preset;
+		bool skip_phy_ssc_reduction;
+	} wa_flags;
+	union dc_dp_ffe_preset forced_dp_ffe_preset;
+	struct link_mst_stream_allocation_table mst_stream_alloc_table;
+
+	struct dc_link_status link_status;
+	struct dprx_states dprx_states;
+
+	struct gpio *hpd_gpio;
+	enum dc_link_fec_state fec_state;
+	bool is_dds;
+	bool is_display_mux_present;
+	bool link_powered_externally;	// Used to bypass hardware sequencing delays when panel is powered down forcibly
+
+	struct dc_panel_config panel_config;
+	struct phy_state phy_state;
+	uint32_t phy_transition_bitmask;
+	// BW ALLOCATON USB4 ONLY
+	struct dc_dpia_bw_alloc dpia_bw_alloc_config;
+	bool skip_implict_edp_power_control;
+	enum backlight_control_type backlight_control_type;
+};
+
+struct dc {
+	struct dc_debug_options debug;
+	struct dc_versions versions;
+	struct dc_caps caps;
+	struct dc_cap_funcs cap_funcs;
+	struct dc_config config;
+	struct dc_bounding_box_overrides bb_overrides;
+	struct dc_bug_wa work_arounds;
+	struct dc_context *ctx;
+	struct dc_phy_addr_space_config vm_pa_config;
+
+	uint8_t link_count;
+	struct dc_link *links[MAX_LINKS];
+	uint8_t lowest_dpia_link_index;
+	struct link_service *link_srv;
+
+	struct dc_state *current_state;
+	struct resource_pool *res_pool;
+
+	struct clk_mgr *clk_mgr;
+
+	/* Display Engine Clock levels */
+	struct dm_pp_clock_levels sclk_lvls;
+
+	/* Inputs into BW and WM calculations. */
+	struct bw_calcs_dceip *bw_dceip;
+	struct bw_calcs_vbios *bw_vbios;
+	struct dcn_soc_bounding_box *dcn_soc;
+	struct dcn_ip_params *dcn_ip;
+	struct display_mode_lib dml;
+
+	/* HW functions */
+	struct hw_sequencer_funcs hwss;
+	struct dce_hwseq *hwseq;
+
+	/* Require to optimize clocks and bandwidth for added/removed planes */
+	bool optimized_required;
+	bool idle_optimizations_allowed;
+	bool enable_c20_dtm_b0;
+
+	/* Require to maintain clocks and bandwidth for UEFI enabled HW */
+
+	/* For eDP to know the switching state of SmartMux */
+	bool is_switch_in_progress_orig;
+	bool is_switch_in_progress_dest;
+
+	/* FBC compressor */
+	struct compressor *fbc_compressor;
+
+	struct dc_debug_data debug_data;
+	struct dpcd_vendor_signature vendor_signature;
+
+	const char *build_id;
+	struct vm_helper *vm_helper;
+
+	uint32_t *dcn_reg_offsets;
+	uint32_t *nbio_reg_offsets;
+	uint32_t *clk_reg_offsets;
+
+	/* Scratch memory */
+	struct {
+		struct {
+			/*
+			 * For matching clock_limits table in driver with table
+			 * from PMFW.
+			 */
+			struct _vcs_dpi_voltage_scaling_st clock_limits[DC__VOLTAGE_STATES];
+		} update_bw_bounding_box;
+		struct dc_scratch_space current_state;
+		struct dc_scratch_space new_state;
+		struct dc_stream_state temp_stream; // Used so we don't need to allocate stream on the stack
+		struct dc_link temp_link;
+		bool pipes_to_unlock_first[MAX_PIPES]; /* Any of the pipes indicated here should be unlocked first */
+	} scratch;
+
+	struct dml2_configuration_options dml2_options;
+	struct dml2_configuration_options dml2_dc_power_options;
+	enum dc_acpi_cm_power_state power_state;
+	struct soc_and_ip_translator *soc_and_ip_translator;
+};
+
+struct dc_scaling_info {
+	struct rect src_rect;
+	struct rect dst_rect;
+	struct rect clip_rect;
+	struct scaling_taps scaling_quality;
+};
+
+struct dc_fast_update {
+	const struct dc_flip_addrs *flip_addr;
+	const struct dc_gamma *gamma;
+	const struct colorspace_transform *gamut_remap_matrix;
+	const struct dc_csc_transform *input_csc_color_matrix;
+	const struct fixed31_32 *coeff_reduction_factor;
+	struct dc_transfer_func *out_transfer_func;
+	struct dc_csc_transform *output_csc_transform;
+	const struct dc_csc_transform *cursor_csc_color_matrix;
+};
+
+struct dc_surface_update {
+	struct dc_plane_state *surface;
+
+	/* isr safe update parameters.  null means no updates */
+	const struct dc_flip_addrs *flip_addr;
+	const struct dc_plane_info *plane_info;
+	const struct dc_scaling_info *scaling_info;
+	struct fixed31_32 hdr_mult;
+	/* following updates require alloc/sleep/spin that is not isr safe,
+	 * null means no updates
+	 */
+	const struct dc_gamma *gamma;
+	const struct dc_transfer_func *in_transfer_func;
+
+	const struct dc_csc_transform *input_csc_color_matrix;
+	const struct fixed31_32 *coeff_reduction_factor;
+	const struct dc_transfer_func *func_shaper;
+	const struct dc_3dlut *lut3d_func;
+	const struct dc_transfer_func *blend_tf;
+	const struct colorspace_transform *gamut_remap_matrix;
+	/*
+	 * Color Transformations for pre-blend MCM (Shaper, 3DLUT, 1DLUT)
+	 *
+	 * change cm2_params.component_settings: Full update
+	 * change cm2_params.cm2_luts: Fast update
+	 */
+	const struct dc_cm2_parameters *cm2_params;
+	const struct dc_csc_transform *cursor_csc_color_matrix;
+	unsigned int sdr_white_level_nits;
+	struct dc_bias_and_scale bias_and_scale;
+};
+
+struct dc_underflow_debug_data {
+	uint32_t otg_inst;
+	uint32_t otg_underflow;
+	uint32_t h_position;
+	uint32_t v_position;
+	uint32_t otg_frame_count;
+	struct dc_underflow_per_hubp_debug_data {
+		uint32_t hubp_underflow;
+		uint32_t hubp_in_blank;
+		uint32_t hubp_readline;
+		uint32_t det_config_error;
+	} hubps[MAX_PIPES];
+	uint32_t curr_det_sizes[MAX_PIPES];
+	uint32_t target_det_sizes[MAX_PIPES];
+	uint32_t compbuf_config_error;
+};
+
+/*
+ * Create a new surface with default parameters;
+ */
+void dc_gamma_retain(struct dc_gamma *dc_gamma);
+void dc_gamma_release(struct dc_gamma **dc_gamma);
+struct dc_gamma *dc_create_gamma(void);
+
+void dc_transfer_func_retain(struct dc_transfer_func *dc_tf);
+void dc_transfer_func_release(struct dc_transfer_func *dc_tf);
+struct dc_transfer_func *dc_create_transfer_func(void);
+
+struct dc_3dlut *dc_create_3dlut_func(void);
+void dc_3dlut_func_release(struct dc_3dlut *lut);
+void dc_3dlut_func_retain(struct dc_3dlut *lut);
+
+void dc_post_update_surfaces_to_stream(
+		struct dc *dc);
+
+/**
+ * struct dc_validation_set - Struct to store surface/stream associations for validation
+ */
+struct dc_validation_set {
+	/**
+	 * @stream: Stream state properties
+	 */
+	struct dc_stream_state *stream;
+
+	/**
+	 * @plane_states: Surface state
+	 */
+	struct dc_plane_state *plane_states[MAX_SURFACES];
+
+	/**
+	 * @plane_count: Total of active planes
+	 */
+	uint8_t plane_count;
+};
+
+bool dc_validate_boot_timing(const struct dc *dc,
+				const struct dc_sink *sink,
+				struct dc_crtc_timing *crtc_timing);
+
+enum dc_status dc_validate_plane(struct dc *dc, const struct dc_plane_state *plane_state);
+
+enum dc_status dc_validate_with_context(struct dc *dc,
+					const struct dc_validation_set set[],
+					int set_count,
+					struct dc_state *context,
+					enum dc_validate_mode validate_mode);
+
+bool dc_set_generic_gpio_for_stereo(bool enable,
+		struct gpio_service *gpio_service);
+
+enum dc_status dc_validate_global_state(
+		struct dc *dc,
+		struct dc_state *new_ctx,
+		enum dc_validate_mode validate_mode);
+
+bool dc_acquire_release_mpc_3dlut(
+		struct dc *dc, bool acquire,
+		struct dc_stream_state *stream,
+		struct dc_3dlut **lut,
+		struct dc_transfer_func **shaper);
+
+bool dc_resource_is_dsc_encoding_supported(const struct dc *dc);
+void get_audio_check(struct audio_info *aud_modes,
+	struct audio_check *aud_chk);
+
+bool fast_nonaddr_updates_exist(struct dc_fast_update *fast_update, int surface_count);
+void populate_fast_updates(struct dc_fast_update *fast_update,
+		struct dc_surface_update *srf_updates,
+		int surface_count,
+		struct dc_stream_update *stream_update);
+/*
+ * Set up streams and links associated to drive sinks
+ * The streams parameter is an absolute set of all active streams.
+ *
+ * After this call:
+ *   Phy, Encoder, Timing Generator are programmed and enabled.
+ *   New streams are enabled with blank stream; no memory read.
+ */
+enum dc_status dc_commit_streams(struct dc *dc, struct dc_commit_streams_params *params);
+
+
+struct dc_plane_state *dc_get_surface_for_mpcc(struct dc *dc,
+		struct dc_stream_state *stream,
+		int mpcc_inst);
+
+
+uint32_t dc_get_opp_for_plane(struct dc *dc, struct dc_plane_state *plane);
+
+void dc_set_disable_128b_132b_stream_overhead(bool disable);
+
+/* The function returns minimum bandwidth required to drive a given timing
+ * return - minimum required timing bandwidth in kbps.
+ */
+uint32_t dc_bandwidth_in_kbps_from_timing(
+		const struct dc_crtc_timing *timing,
+		const enum dc_link_encoding_format link_encoding);
+
+/* Link Interfaces */
+/* Return an enumerated dc_link.
+ * dc_link order is constant and determined at
+ * boot time.  They cannot be created or destroyed.
+ * Use dc_get_caps() to get number of links.
+ */
+struct dc_link *dc_get_link_at_index(struct dc *dc, uint32_t link_index);
+
+/* Return instance id of the edp link. Inst 0 is primary edp link. */
+bool dc_get_edp_link_panel_inst(const struct dc *dc,
+		const struct dc_link *link,
+		unsigned int *inst_out);
+
+/* Return an array of link pointers to edp links. */
+void dc_get_edp_links(const struct dc *dc,
+		struct dc_link **edp_links,
+		int *edp_num);
+
+void dc_set_edp_power(const struct dc *dc, struct dc_link *edp_link,
+				 bool powerOn);
+
+/* The function initiates detection handshake over the given link. It first
+ * determines if there are display connections over the link. If so it initiates
+ * detection protocols supported by the connected receiver device. The function
+ * contains protocol specific handshake sequences which are sometimes mandatory
+ * to establish a proper connection between TX and RX. So it is always
+ * recommended to call this function as the first link operation upon HPD event
+ * or power up event. Upon completion, the function will update link structure
+ * in place based on latest RX capabilities. The function may also cause dpms
+ * to be reset to off for all currently enabled streams to the link. It is DM's
+ * responsibility to serialize detection and DPMS updates.
+ *
+ * @reason - Indicate which event triggers this detection. dc may customize
+ * detection flow depending on the triggering events.
+ * return false - if detection is not fully completed. This could happen when
+ * there is an unrecoverable error during detection or detection is partially
+ * completed (detection has been delegated to dm mst manager ie.
+ * link->connection_type == dc_connection_mst_branch when returning false).
+ * return true - detection is completed, link has been fully updated with latest
+ * detection result.
+ */
+bool dc_link_detect(struct dc_link *link, enum dc_detect_reason reason);
+
+struct dc_sink_init_data;
+
+/* When link connection type is dc_connection_mst_branch, remote sink can be
+ * added to the link. The interface creates a remote sink and associates it with
+ * current link. The sink will be retained by link until remove remote sink is
+ * called.
+ *
+ * @dc_link - link the remote sink will be added to.
+ * @edid - byte array of EDID raw data.
+ * @len - size of the edid in byte
+ * @init_data -
+ */
+struct dc_sink *dc_link_add_remote_sink(
+		struct dc_link *dc_link,
+		const uint8_t *edid,
+		int len,
+		struct dc_sink_init_data *init_data);
+
+/* Remove remote sink from a link with dc_connection_mst_branch connection type.
+ * @link - link the sink should be removed from
+ * @sink - sink to be removed.
+ */
+void dc_link_remove_remote_sink(
+	struct dc_link *link,
+	struct dc_sink *sink);
+
+/* Enable HPD interrupt handler for a given link */
+void dc_link_enable_hpd(const struct dc_link *link);
+
+/* Disable HPD interrupt handler for a given link */
+void dc_link_disable_hpd(const struct dc_link *link);
+
+/* determine if there is a sink connected to the link
+ *
+ * @type - dc_connection_single if connected, dc_connection_none otherwise.
+ * return - false if an unexpected error occurs, true otherwise.
+ *
+ * NOTE: This function doesn't detect downstream sink connections i.e
+ * dc_connection_mst_branch, dc_connection_sst_branch. In this case, it will
+ * return dc_connection_single if the branch device is connected despite of
+ * downstream sink's connection status.
+ */
+bool dc_link_detect_connection_type(struct dc_link *link,
+		enum dc_connection_type *type);
+
+/* query current hpd pin value
+ * return - true HPD is asserted (HPD high), false otherwise (HPD low)
+ *
+ */
+bool dc_link_get_hpd_state(struct dc_link *link);
+
+/* Getter for cached link status from given link */
+const struct dc_link_status *dc_link_get_status(const struct dc_link *link);
+
+/* enable/disable hardware HPD filter.
+ *
+ * @link - The link the HPD pin is associated with.
+ * @enable = true - enable hardware HPD filter. HPD event will only queued to irq
+ * handler once after no HPD change has been detected within dc default HPD
+ * filtering interval since last HPD event. i.e if display keeps toggling hpd
+ * pulses within default HPD interval, no HPD event will be received until HPD
+ * toggles have stopped. Then HPD event will be queued to irq handler once after
+ * dc default HPD filtering interval since last HPD event.
+ *
+ * @enable = false - disable hardware HPD filter. HPD event will be queued
+ * immediately to irq handler after no HPD change has been detected within
+ * IRQ_HPD (aka HPD short pulse) interval (i.e 2ms).
+ */
+void dc_link_enable_hpd_filter(struct dc_link *link, bool enable);
+
+/* submit i2c read/write payloads through ddc channel
+ * @link_index - index to a link with ddc in i2c mode
+ * @cmd - i2c command structure
+ * return - true if success, false otherwise.
+ */
+bool dc_submit_i2c(
+		struct dc *dc,
+		uint32_t link_index,
+		struct i2c_command *cmd);
+
+/* submit i2c read/write payloads through oem channel
+ * @link_index - index to a link with ddc in i2c mode
+ * @cmd - i2c command structure
+ * return - true if success, false otherwise.
+ */
+bool dc_submit_i2c_oem(
+		struct dc *dc,
+		struct i2c_command *cmd);
+
+enum aux_return_code_type;
+/* Attempt to transfer the given aux payload. This function does not perform
+ * retries or handle error states. The reply is returned in the payload->reply
+ * and the result through operation_result. Returns the number of bytes
+ * transferred,or -1 on a failure.
+ */
+int dc_link_aux_transfer_raw(struct ddc_service *ddc,
+		struct aux_payload *payload,
+		enum aux_return_code_type *operation_result);
+
+struct ddc_service *
+dc_get_oem_i2c_device(struct dc *dc);
+
+bool dc_is_oem_i2c_device_present(
+	struct dc *dc,
+	size_t slave_address
+);
+
+/* return true if the connected receiver supports the hdcp version */
+bool dc_link_is_hdcp14(struct dc_link *link, enum signal_type signal);
+bool dc_link_is_hdcp22(struct dc_link *link, enum signal_type signal);
+
+/* Notify DC about DP RX Interrupt (aka DP IRQ_HPD).
+ *
+ * TODO - When defer_handling is true the function will have a different purpose.
+ * It no longer does complete hpd rx irq handling. We should create a separate
+ * interface specifically for this case.
+ *
+ * Return:
+ * true - Downstream port status changed. DM should call DC to do the
+ * detection.
+ * false - no change in Downstream port status. No further action required
+ * from DM.
+ */
+bool dc_link_handle_hpd_rx_irq(struct dc_link *dc_link,
+		union hpd_irq_data *hpd_irq_dpcd_data, bool *out_link_loss,
+		bool defer_handling, bool *has_left_work);
+/* handle DP specs define test automation sequence*/
+void dc_link_dp_handle_automated_test(struct dc_link *link);
+
+/* handle DP Link loss sequence and try to recover RX link loss with best
+ * effort
+ */
+void dc_link_dp_handle_link_loss(struct dc_link *link);
+
+/* Determine if hpd rx irq should be handled or ignored
+ * return true - hpd rx irq should be handled.
+ * return false - it is safe to ignore hpd rx irq event
+ */
+bool dc_link_dp_allow_hpd_rx_irq(const struct dc_link *link);
+
+/* Determine if link loss is indicated with a given hpd_irq_dpcd_data.
+ * @link - link the hpd irq data associated with
+ * @hpd_irq_dpcd_data - input hpd irq data
+ * return - true if hpd irq data indicates a link lost
+ */
+bool dc_link_check_link_loss_status(struct dc_link *link,
+		union hpd_irq_data *hpd_irq_dpcd_data);
+
+/* Read hpd rx irq data from a given link
+ * @link - link where the hpd irq data should be read from
+ * @irq_data - output hpd irq data
+ * return - DC_OK if hpd irq data is read successfully, otherwise hpd irq data
+ * read has failed.
+ */
+enum dc_status dc_link_dp_read_hpd_rx_irq_data(
+	struct dc_link *link,
+	union hpd_irq_data *irq_data);
+
+/* The function clears recorded DP RX states in the link. DM should call this
+ * function when it is resuming from S3 power state to previously connected links.
+ *
+ * TODO - in the future we should consider to expand link resume interface to
+ * support clearing previous rx states. So we don't have to rely on dm to call
+ * this interface explicitly.
+ */
+void dc_link_clear_dprx_states(struct dc_link *link);
+
+/* Destruct the mst topology of the link and reset the allocated payload table
+ *
+ * NOTE: this should only be called if DM chooses not to call dc_link_detect but
+ * still wants to reset MST topology on an unplug event */
+bool dc_link_reset_cur_dp_mst_topology(struct dc_link *link);
+
+/* The function calculates effective DP link bandwidth when a given link is
+ * using the given link settings.
+ *
+ * return - total effective link bandwidth in kbps.
+ */
+uint32_t dc_link_bandwidth_kbps(
+	const struct dc_link *link,
+	const struct dc_link_settings *link_setting);
+
+struct dp_audio_bandwidth_params {
+	const struct dc_crtc_timing *crtc_timing;
+	enum dp_link_encoding link_encoding;
+	uint32_t channel_count;
+	uint32_t sample_rate_hz;
+};
+
+/* The function calculates the minimum size of hblank (in bytes) needed to
+ * support the specified channel count and sample rate combination, given the
+ * link encoding and timing to be used. This calculation is not supported
+ * for 8b/10b SST.
+ *
+ * return - min hblank size in bytes, 0 if 8b/10b SST.
+ */
+uint32_t dc_link_required_hblank_size_bytes(
+	const struct dc_link *link,
+	struct dp_audio_bandwidth_params *audio_params);
+
+/* The function takes a snapshot of current link resource allocation state
+ * @dc: pointer to dc of the dm calling this
+ * @map: a dc link resource snapshot defined internally to dc.
+ *
+ * DM needs to capture a snapshot of current link resource allocation mapping
+ * and store it in its persistent storage.
+ *
+ * Some of the link resource is using first come first serve policy.
+ * The allocation mapping depends on original hotplug order. This information
+ * is lost after driver is loaded next time. The snapshot is used in order to
+ * restore link resource to its previous state so user will get consistent
+ * link capability allocation across reboot.
+ *
+ */
+void dc_get_cur_link_res_map(const struct dc *dc, uint32_t *map);
+
+/* This function restores link resource allocation state from a snapshot
+ * @dc: pointer to dc of the dm calling this
+ * @map: a dc link resource snapshot defined internally to dc.
+ *
+ * DM needs to call this function after initial link detection on boot and
+ * before first commit streams to restore link resource allocation state
+ * from previous boot session.
+ *
+ * Some of the link resource is using first come first serve policy.
+ * The allocation mapping depends on original hotplug order. This information
+ * is lost after driver is loaded next time. The snapshot is used in order to
+ * restore link resource to its previous state so user will get consistent
+ * link capability allocation across reboot.
+ *
+ */
+void dc_restore_link_res_map(const struct dc *dc, uint32_t *map);
+
+/* TODO: this is not meant to be exposed to DM. Should switch to stream update
+ * interface i.e stream_update->dsc_config
+ */
+bool dc_link_update_dsc_config(struct pipe_ctx *pipe_ctx);
+
+/* translate a raw link rate data to bandwidth in kbps */
+uint32_t dc_link_bw_kbps_from_raw_frl_link_rate_data(const struct dc *dc, uint8_t bw);
+
+/* determine the optimal bandwidth given link and required bw.
+ * @link - current detected link
+ * @req_bw - requested bandwidth in kbps
+ * @link_settings - returned most optimal link settings that can fit the
+ * requested bandwidth
+ * return - false if link can't support requested bandwidth, true if link
+ * settings is found.
+ */
+bool dc_link_decide_edp_link_settings(struct dc_link *link,
+		struct dc_link_settings *link_settings,
+		uint32_t req_bw);
+
+/* return the max dp link settings can be driven by the link without considering
+ * connected RX device and its capability
+ */
+bool dc_link_dp_get_max_link_enc_cap(const struct dc_link *link,
+		struct dc_link_settings *max_link_enc_cap);
+
+/* determine when the link is driving MST mode, what DP link channel coding
+ * format will be used. The decision will remain unchanged until next HPD event.
+ *
+ * @link -  a link with DP RX connection
+ * return - if stream is committed to this link with MST signal type, type of
+ * channel coding format dc will choose.
+ */
+enum dp_link_encoding dc_link_dp_mst_decide_link_encoding_format(
+		const struct dc_link *link);
+
+/* get max dp link settings the link can enable with all things considered. (i.e
+ * TX/RX/Cable capabilities and dp override policies.
+ *
+ * @link - a link with DP RX connection
+ * return - max dp link settings the link can enable.
+ *
+ */
+const struct dc_link_settings *dc_link_get_link_cap(const struct dc_link *link);
+
+/* Get the highest encoding format that the link supports; highest meaning the
+ * encoding format which supports the maximum bandwidth.
+ *
+ * @link - a link with DP RX connection
+ * return - highest encoding format link supports.
+ */
+enum dc_link_encoding_format dc_link_get_highest_encoding_format(const struct dc_link *link);
+
+/* Check if a RX (ex. DP sink, MST hub, passive or active dongle) is connected
+ * to a link with dp connector signal type.
+ * @link - a link with dp connector signal type
+ * return - true if connected, false otherwise
+ */
+bool dc_link_is_dp_sink_present(struct dc_link *link);
+
+/* Force DP lane settings update to main-link video signal and notify the change
+ * to DP RX via DPCD. This is a debug interface used for video signal integrity
+ * tuning purpose. The interface assumes link has already been enabled with DP
+ * signal.
+ *
+ * @lt_settings - a container structure with desired hw_lane_settings
+ */
+void dc_link_set_drive_settings(struct dc *dc,
+				struct link_training_settings *lt_settings,
+				struct dc_link *link);
+
+/* Enable a test pattern in Link or PHY layer in an active link for compliance
+ * test or debugging purpose. The test pattern will remain until next un-plug.
+ *
+ * @link - active link with DP signal output enabled.
+ * @test_pattern - desired test pattern to output.
+ * NOTE: set to DP_TEST_PATTERN_VIDEO_MODE to disable previous test pattern.
+ * @test_pattern_color_space - for video test pattern choose a desired color
+ * space.
+ * @p_link_settings - For PHY pattern choose a desired link settings
+ * @p_custom_pattern - some test pattern will require a custom input to
+ * customize some pattern details. Otherwise keep it to NULL.
+ * @cust_pattern_size - size of the custom pattern input.
+ *
+ */
+bool dc_link_dp_set_test_pattern(
+	struct dc_link *link,
+	enum dp_test_pattern test_pattern,
+	enum dp_test_pattern_color_space test_pattern_color_space,
+	const struct link_training_settings *p_link_settings,
+	const unsigned char *p_custom_pattern,
+	unsigned int cust_pattern_size);
+
+/* Force DP link settings to always use a specific value until reboot to a
+ * specific link. If link has already been enabled, the interface will also
+ * switch to desired link settings immediately. This is a debug interface to
+ * generic dp issue trouble shooting.
+ */
+void dc_link_set_preferred_link_settings(struct dc *dc,
+		struct dc_link_settings *link_setting,
+		struct dc_link *link);
+
+/* Force DP link to customize a specific link training behavior by overriding to
+ * standard DP specs defined protocol. This is a debug interface to trouble shoot
+ * display specific link training issues or apply some display specific
+ * workaround in link training.
+ *
+ * @link_settings - if not NULL, force preferred link settings to the link.
+ * @lt_override - a set of override pointers. If any pointer is none NULL, dc
+ * will apply this particular override in future link training. If NULL is
+ * passed in, dc resets previous overrides.
+ * NOTE: DM must keep the memory from override pointers until DM resets preferred
+ * training settings.
+ */
+void dc_link_set_preferred_training_settings(struct dc *dc,
+		struct dc_link_settings *link_setting,
+		struct dc_link_training_overrides *lt_overrides,
+		struct dc_link *link,
+		bool skip_immediate_retrain);
+
+/* return - true if FEC is supported with connected DP RX, false otherwise */
+bool dc_link_is_fec_supported(const struct dc_link *link);
+
+/* query FEC enablement policy to determine if FEC will be enabled by dc during
+ * link enablement.
+ * return - true if FEC should be enabled, false otherwise.
+ */
+bool dc_link_should_enable_fec(const struct dc_link *link);
+
+/* determine lttpr mode the current link should be enabled with a specific link
+ * settings.
+ */
+enum lttpr_mode dc_link_decide_lttpr_mode(struct dc_link *link,
+		struct dc_link_settings *link_setting);
+
+/* Force DP RX to update its power state.
+ * NOTE: this interface doesn't update dp main-link. Calling this function will
+ * cause DP TX main-link and DP RX power states out of sync. DM has to restore
+ * RX power state back upon finish DM specific execution requiring DP RX in a
+ * specific power state.
+ * @on - true to set DP RX in D0 power state, false to set DP RX in D3 power
+ * state.
+ */
+void dc_link_dp_receiver_power_ctrl(struct dc_link *link, bool on);
+
+/* Force link to read base dp receiver caps from dpcd 000h - 00Fh and overwrite
+ * current value read from extended receiver cap from 02200h - 0220Fh.
+ * Some DP RX has problems of providing accurate DP receiver caps from extended
+ * field, this interface is a workaround to revert link back to use base caps.
+ */
+void dc_link_overwrite_extended_receiver_cap(
+		struct dc_link *link);
+
+void dc_link_edp_panel_backlight_power_on(struct dc_link *link,
+		bool wait_for_hpd);
+
+/* Set backlight level of an embedded panel (eDP, LVDS).
+ * backlight_pwm_u16_16 is unsigned 32 bit with 16 bit integer
+ * and 16 bit fractional, where 1.0 is max backlight value.
+ */
+bool dc_link_set_backlight_level(const struct dc_link *dc_link,
+		struct set_backlight_level_params *backlight_level_params);
+
+/* Set/get nits-based backlight level of an embedded panel (eDP, LVDS). */
+bool dc_link_set_backlight_level_nits(struct dc_link *link,
+		bool isHDR,
+		uint32_t backlight_millinits,
+		uint32_t transition_time_in_ms);
+
+bool dc_link_get_backlight_level_nits(struct dc_link *link,
+		uint32_t *backlight_millinits,
+		uint32_t *backlight_millinits_peak);
+
+int dc_link_get_backlight_level(const struct dc_link *dc_link);
+
+int dc_link_get_target_backlight_pwm(const struct dc_link *link);
+
+bool dc_link_set_psr_allow_active(struct dc_link *dc_link, const bool *enable,
+		bool wait, bool force_static, const unsigned int *power_opts);
+
+bool dc_link_get_psr_state(const struct dc_link *dc_link, enum dc_psr_state *state);
+
+bool dc_link_setup_psr(struct dc_link *dc_link,
+		const struct dc_stream_state *stream, struct psr_config *psr_config,
+		struct psr_context *psr_context);
+
+/*
+ * Communicate with DMUB to allow or disallow Panel Replay on the specified link:
+ *
+ * @link: pointer to the dc_link struct instance
+ * @enable: enable(active) or disable(inactive) replay
+ * @wait: state transition need to wait the active set completed.
+ * @force_static: force disable(inactive) the replay
+ * @power_opts: set power optimazation parameters to DMUB.
+ *
+ * return: allow Replay active will return true, else will return false.
+ */
+bool dc_link_set_replay_allow_active(struct dc_link *dc_link, const bool *enable,
+		bool wait, bool force_static, const unsigned int *power_opts);
+
+bool dc_link_get_replay_state(const struct dc_link *dc_link, uint64_t *state);
+
+/* On eDP links this function call will stall until T12 has elapsed.
+ * If the panel is not in power off state, this function will return
+ * immediately.
+ */
+bool dc_link_wait_for_t12(struct dc_link *link);
+
+/* Determine if dp trace has been initialized to reflect upto date result *
+ * return - true if trace is initialized and has valid data. False dp trace
+ * doesn't have valid result.
+ */
+bool dc_dp_trace_is_initialized(struct dc_link *link);
+
+/* Query a dp trace flag to indicate if the current dp trace data has been
+ * logged before
+ */
+bool dc_dp_trace_is_logged(struct dc_link *link,
+		bool in_detection);
+
+/* Set dp trace flag to indicate whether DM has already logged the current dp
+ * trace data. DM can set is_logged to true upon logging and check
+ * dc_dp_trace_is_logged before logging to avoid logging the same result twice.
+ */
+void dc_dp_trace_set_is_logged_flag(struct dc_link *link,
+		bool in_detection,
+		bool is_logged);
+
+/* Obtain driver time stamp for last dp link training end. The time stamp is
+ * formatted based on dm_get_timestamp DM function.
+ * @in_detection - true to get link training end time stamp of last link
+ * training in detection sequence. false to get link training end time stamp
+ * of last link training in commit (dpms) sequence
+ */
+unsigned long long dc_dp_trace_get_lt_end_timestamp(struct dc_link *link,
+		bool in_detection);
+
+/* Get how many link training attempts dc has done with latest sequence.
+ * @in_detection - true to get link training count of last link
+ * training in detection sequence. false to get link training count of last link
+ * training in commit (dpms) sequence
+ */
+const struct dp_trace_lt_counts *dc_dp_trace_get_lt_counts(struct dc_link *link,
+		bool in_detection);
+
+/* Get how many link loss has happened since last link training attempts */
+unsigned int dc_dp_trace_get_link_loss_count(struct dc_link *link);
+
+/*
+ *  USB4 DPIA BW ALLOCATION PUBLIC FUNCTIONS
+ */
+/*
+ * Send a request from DP-Tx requesting to allocate BW remotely after
+ * allocating it locally. This will get processed by CM and a CB function
+ * will be called.
+ *
+ * @link: pointer to the dc_link struct instance
+ * @req_bw: The requested bw in Kbyte to allocated
+ *
+ * return: none
+ */
+void dc_link_set_usb4_req_bw_req(struct dc_link *link, int req_bw);
+
+/*
+ * Handle the USB4 BW Allocation related functionality here:
+ * Plug => Try to allocate max bw from timing parameters supported by the sink
+ * Unplug => de-allocate bw
+ *
+ * @link: pointer to the dc_link struct instance
+ * @peak_bw: Peak bw used by the link/sink
+ *
+ */
+void dc_link_dp_dpia_handle_usb4_bandwidth_allocation_for_link(
+		struct dc_link *link, int peak_bw);
+
+/*
+ * Calculates the DP tunneling bandwidth required for the stream timing
+ * and aggregates the stream bandwidth for the respective DP tunneling link
+ *
+ * return: dc_status
+ */
+enum dc_status dc_link_validate_dp_tunneling_bandwidth(const struct dc *dc, const struct dc_state *new_ctx);
+
+/*
+ * Get if ALPM is supported by the link
+ */
+void dc_link_get_alpm_support(struct dc_link *link, bool *auxless_support,
+	bool *auxwake_support);
+
+/* Sink Interfaces - A sink corresponds to a display output device */
+
+struct dc_container_id {
+	// 128bit GUID in binary form
+	unsigned char  guid[16];
+	// 8 byte port ID -> ELD.PortID
+	unsigned int   portId[2];
+	// 128bit GUID in binary formufacturer name -> ELD.ManufacturerName
+	unsigned short manufacturerName;
+	// 2 byte product code -> ELD.ProductCode
+	unsigned short productCode;
+};
+
+
+struct dc_sink_dsc_caps {
+	// 'true' if these are virtual DPCD's DSC caps (immediately upstream of sink in MST topology),
+	// 'false' if they are sink's DSC caps
+	bool is_virtual_dpcd_dsc;
+	// 'true' if MST topology supports DSC passthrough for sink
+	// 'false' if MST topology does not support DSC passthrough
+	bool is_dsc_passthrough_supported;
+	struct dsc_dec_dpcd_caps dsc_dec_caps;
+};
+
+struct dc_sink_hblank_expansion_caps {
+	// 'true' if these are virtual DPCD's HBlank expansion caps (immediately upstream of sink in MST topology),
+	// 'false' if they are sink's HBlank expansion caps
+	bool is_virtual_dpcd_hblank_expansion;
+	struct hblank_expansion_dpcd_caps dpcd_caps;
+};
+
+struct dc_sink_fec_caps {
+	bool is_rx_fec_supported;
+	bool is_topology_fec_supported;
+};
+
+struct scdc_caps {
+	union hdmi_scdc_manufacturer_OUI_data manufacturer_OUI;
+	union hdmi_scdc_device_id_data device_id;
+};
+
+/*
+ * The sink structure contains EDID and other display device properties
+ */
+struct dc_sink {
+	enum signal_type sink_signal;
+	struct dc_edid dc_edid; /* raw edid */
+	struct dc_edid_caps edid_caps; /* parse display caps */
+	struct dc_container_id *dc_container_id;
+	uint32_t dongle_max_pix_clk;
+	void *priv;
+	struct stereo_3d_features features_3d[TIMING_3D_FORMAT_MAX];
+	bool converter_disable_audio;
+
+	struct scdc_caps scdc_caps;
+	struct dc_sink_dsc_caps dsc_caps;
+	struct dc_sink_fec_caps fec_caps;
+	struct dc_sink_hblank_expansion_caps hblank_expansion_caps;
+
+	bool is_vsc_sdp_colorimetry_supported;
+
+	/* private to DC core */
+	struct dc_link *link;
+	struct dc_context *ctx;
+
+	uint32_t sink_id;
+
+	/* private to dc_sink.c */
+	// refcount must be the last member in dc_sink, since we want the
+	// sink structure to be logically cloneable up to (but not including)
+	// refcount
+	struct kref refcount;
+};
+
+void dc_sink_retain(struct dc_sink *sink);
+void dc_sink_release(struct dc_sink *sink);
+
+struct dc_sink_init_data {
+	enum signal_type sink_signal;
+	struct dc_link *link;
+	uint32_t dongle_max_pix_clk;
+	bool converter_disable_audio;
+};
+
+struct dc_sink *dc_sink_create(const struct dc_sink_init_data *init_params);
+
+/* Newer interfaces  */
+struct dc_cursor {
+	struct dc_plane_address address;
+	struct dc_cursor_attributes attributes;
+};
+
+
+/* Interrupt interfaces */
+enum dc_irq_source dc_interrupt_to_irq_source(
+		struct dc *dc,
+		uint32_t src_id,
+		uint32_t ext_id);
+bool dc_interrupt_set(struct dc *dc, enum dc_irq_source src, bool enable);
+void dc_interrupt_ack(struct dc *dc, enum dc_irq_source src);
+enum dc_irq_source dc_get_hpd_irq_source_at_index(
+		struct dc *dc, uint32_t link_index);
+
+void dc_notify_vsync_int_state(struct dc *dc, struct dc_stream_state *stream, bool enable);
+
+/* Power Interfaces */
+
+void dc_set_power_state(
+		struct dc *dc,
+		enum dc_acpi_cm_power_state power_state);
+void dc_resume(struct dc *dc);
+
+void dc_power_down_on_boot(struct dc *dc);
+
+/*
+ * HDCP Interfaces
+ */
+enum hdcp_message_status dc_process_hdcp_msg(
+		enum signal_type signal,
+		struct dc_link *link,
+		struct hdcp_protection_message *message_info);
+bool dc_is_dmcu_initialized(struct dc *dc);
+
+enum dc_status dc_set_clock(struct dc *dc, enum dc_clock_type clock_type, uint32_t clk_khz, uint32_t stepping);
+void dc_get_clock(struct dc *dc, enum dc_clock_type clock_type, struct dc_clock_config *clock_cfg);
+
+bool dc_is_plane_eligible_for_idle_optimizations(struct dc *dc,
+		unsigned int pitch,
+		unsigned int height,
+		enum surface_pixel_format format,
+		struct dc_cursor_attributes *cursor_attr);
+
+#define dc_allow_idle_optimizations(dc, allow) dc_allow_idle_optimizations_internal(dc, allow, __func__)
+#define dc_exit_ips_for_hw_access(dc) dc_exit_ips_for_hw_access_internal(dc, __func__)
+
+void dc_allow_idle_optimizations_internal(struct dc *dc, bool allow, const char *caller_name);
+void dc_exit_ips_for_hw_access_internal(struct dc *dc, const char *caller_name);
+bool dc_dmub_is_ips_idle_state(struct dc *dc);
+
+/* set min and max memory clock to lowest and highest DPM level, respectively */
+void dc_unlock_memory_clock_frequency(struct dc *dc);
+
+/* set min memory clock to the min required for current mode, max to maxDPM */
+void dc_lock_memory_clock_frequency(struct dc *dc);
+
+/* set soft max for memclk, to be used for AC/DC switching clock limitations */
+void dc_enable_dcmode_clk_limit(struct dc *dc, bool enable);
+
+/* cleanup on driver unload */
+void dc_hardware_release(struct dc *dc);
+
+/* disables fw based mclk switch */
+void dc_mclk_switch_using_fw_based_vblank_stretch_shut_down(struct dc *dc);
+
+bool dc_set_psr_allow_active(struct dc *dc, bool enable);
+
+bool dc_set_replay_allow_active(struct dc *dc, bool active);
+
+bool dc_set_ips_disable(struct dc *dc, unsigned int disable_ips);
+
+void dc_z10_restore(const struct dc *dc);
+void dc_z10_save_init(struct dc *dc);
+
+bool dc_is_dmub_outbox_supported(struct dc *dc);
+bool dc_enable_dmub_notifications(struct dc *dc);
+
+bool dc_abm_save_restore(
+		struct dc *dc,
+		struct dc_stream_state *stream,
+		struct abm_save_restore *pData);
+
+void dc_enable_dmub_outbox(struct dc *dc);
+
+bool dc_process_dmub_aux_transfer_async(struct dc *dc,
+				uint32_t link_index,
+				struct aux_payload *payload);
+
+/* Get dc link index from dpia port index */
+uint8_t get_link_index_from_dpia_port_index(const struct dc *dc,
+				uint8_t dpia_port_index);
+
+bool dc_process_dmub_set_config_async(struct dc *dc,
+				uint32_t link_index,
+				struct set_config_cmd_payload *payload,
+				struct dmub_notification *notify);
+
+enum dc_status dc_process_dmub_set_mst_slots(const struct dc *dc,
+				uint32_t link_index,
+				uint8_t mst_alloc_slots,
+				uint8_t *mst_slots_in_use);
+
+void dc_process_dmub_dpia_set_tps_notification(const struct dc *dc, uint32_t link_index, uint8_t tps);
+
+void dc_process_dmub_dpia_hpd_int_enable(const struct dc *dc,
+				uint32_t hpd_int_enable);
+
+void dc_print_dmub_diagnostic_data(const struct dc *dc);
+
+void dc_query_current_properties(struct dc *dc, struct dc_current_properties *properties);
+
+struct dc_power_profile {
+	int power_level; /* Lower is better */
+};
+
+struct dc_power_profile dc_get_power_profile_for_dc_state(const struct dc_state *context);
+
+unsigned int dc_get_det_buffer_size_from_state(const struct dc_state *context);
+
+bool dc_get_host_router_index(const struct dc_link *link, unsigned int *host_router_index);
+
+/* DSC Interfaces */
+#include "dc_dsc.h"
+
+void dc_get_visual_confirm_for_stream(
+	struct dc *dc,
+	struct dc_stream_state *stream_state,
+	struct tg_color *color);
+
+/* Disable acc mode Interfaces */
+void dc_disable_accelerated_mode(struct dc *dc);
+
+bool dc_is_timing_changed(struct dc_stream_state *cur_stream,
+		       struct dc_stream_state *new_stream);
+
+bool dc_is_cursor_limit_pending(struct dc *dc);
+bool dc_can_clear_cursor_limit(struct dc *dc);
+
+/**
+ * dc_get_underflow_debug_data_for_otg() - Retrieve underflow debug data.
+ *
+ * @dc: Pointer to the display core context.
+ * @primary_otg_inst: Instance index of the primary OTG that underflowed.
+ * @out_data: Pointer to a dc_underflow_debug_data struct to be filled with debug information.
+ *
+ * This function collects and logs underflow-related HW states when underflow happens,
+ * including OTG underflow status, current read positions, frame count, and per-HUBP debug data.
+ * The results are stored in the provided out_data structure for further analysis or logging.
+ */
+void dc_get_underflow_debug_data_for_otg(struct dc *dc, int primary_otg_inst, struct dc_underflow_debug_data *out_data);
+
+#endif /* DC_INTERFACE_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/dc_bios_types.h b/drivers/gpu/drm/amd/display/dc/dc_bios_types.h
new file mode 100644
index 000000000000..5fa5e2b63fb7
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_bios_types.h
@@ -0,0 +1,190 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DC_BIOS_TYPES_H
+#define DC_BIOS_TYPES_H
+
+/******************************************************************************
+ * Interface file for VBIOS implementations.
+ *
+ * The default implementation is inside DC.
+ * Display Manager (which instantiates DC) has the option to supply it's own
+ * (external to DC) implementation of VBIOS, which will be called by DC, using
+ * this interface.
+ * (The intended use is Diagnostics, but other uses may appear.)
+ *****************************************************************************/
+
+#include "include/bios_parser_types.h"
+
+struct dc_vbios_funcs {
+	uint8_t (*get_connectors_number)(struct dc_bios *bios);
+
+	struct graphics_object_id (*get_connector_id)(
+		struct dc_bios *bios,
+		uint8_t connector_index);
+	enum bp_result (*get_src_obj)(
+		struct dc_bios *bios,
+		struct graphics_object_id object_id, uint32_t index,
+		struct graphics_object_id *src_object_id);
+	enum bp_result (*get_i2c_info)(
+		struct dc_bios *dcb,
+		struct graphics_object_id id,
+		struct graphics_object_i2c_info *info);
+	enum bp_result (*get_hpd_info)(
+		struct dc_bios *bios,
+		struct graphics_object_id id,
+		struct graphics_object_hpd_info *info);
+	enum bp_result (*get_device_tag)(
+		struct dc_bios *bios,
+		struct graphics_object_id connector_object_id,
+		uint32_t device_tag_index,
+		struct connector_device_tag_info *info);
+	enum bp_result (*get_spread_spectrum_info)(
+		struct dc_bios *bios,
+		enum as_signal_type signal,
+		uint32_t index,
+		struct spread_spectrum_info *ss_info);
+	uint32_t (*get_ss_entry_number)(
+		struct dc_bios *bios,
+		enum as_signal_type signal);
+	enum bp_result (*get_embedded_panel_info)(
+		struct dc_bios *bios,
+		struct embedded_panel_info *info);
+	enum bp_result (*get_gpio_pin_info)(
+		struct dc_bios *bios,
+		uint32_t gpio_id,
+		struct gpio_pin_info *info);
+	enum bp_result (*get_encoder_cap_info)(
+		struct dc_bios *bios,
+		struct graphics_object_id object_id,
+		struct bp_encoder_cap_info *info);
+
+	bool (*is_accelerated_mode)(
+		struct dc_bios *bios);
+	void (*set_scratch_critical_state)(
+		struct dc_bios *bios,
+		bool state);
+	bool (*is_device_id_supported)(
+		struct dc_bios *bios,
+		struct device_id id);
+	/* COMMANDS */
+
+	enum bp_result (*encoder_control)(
+		struct dc_bios *bios,
+		struct bp_encoder_control *cntl);
+	enum bp_result (*transmitter_control)(
+		struct dc_bios *bios,
+		struct bp_transmitter_control *cntl);
+	enum bp_result (*enable_crtc)(
+		struct dc_bios *bios,
+		enum controller_id id,
+		bool enable);
+	enum bp_result (*adjust_pixel_clock)(
+		struct dc_bios *bios,
+		struct bp_adjust_pixel_clock_parameters *bp_params);
+	enum bp_result (*set_pixel_clock)(
+		struct dc_bios *bios,
+		struct bp_pixel_clock_parameters *bp_params);
+	enum bp_result (*set_dce_clock)(
+		struct dc_bios *bios,
+		struct bp_set_dce_clock_parameters *bp_params);
+	enum bp_result (*enable_spread_spectrum_on_ppll)(
+		struct dc_bios *bios,
+		struct bp_spread_spectrum_parameters *bp_params,
+		bool enable);
+	enum bp_result (*program_crtc_timing)(
+		struct dc_bios *bios,
+		struct bp_hw_crtc_timing_parameters *bp_params);
+	enum bp_result (*program_display_engine_pll)(
+		struct dc_bios *bios,
+		struct bp_pixel_clock_parameters *bp_params);
+	enum bp_result (*enable_disp_power_gating)(
+		struct dc_bios *bios,
+		enum controller_id controller_id,
+		enum bp_pipe_control_action action);
+
+	void (*bios_parser_destroy)(struct dc_bios **dcb);
+
+	enum bp_result (*get_board_layout_info)(
+		struct dc_bios *dcb,
+		struct board_layout_info *board_layout_info);
+	uint16_t (*pack_data_tables)(
+		struct dc_bios *dcb,
+		void *dst);
+
+	enum bp_result (*get_atom_dc_golden_table)(
+			struct dc_bios *dcb);
+
+	enum bp_result (*enable_lvtma_control)(
+		struct dc_bios *bios,
+		uint8_t uc_pwr_on,
+		uint8_t pwrseq_instance,
+		uint8_t bypass_panel_control_wait);
+
+	enum bp_result (*get_soc_bb_info)(
+		struct dc_bios *dcb,
+		struct bp_soc_bb_info *soc_bb_info);
+
+	enum bp_result (*get_disp_connector_caps_info)(
+			struct dc_bios *dcb,
+			struct graphics_object_id object_id,
+			struct bp_disp_connector_caps_info *info);
+	enum bp_result (*get_lttpr_caps)(
+			struct dc_bios *dcb,
+			uint8_t *dce_caps);
+	enum bp_result (*get_lttpr_interop)(
+			struct dc_bios *dcb,
+			uint8_t *dce_caps);
+
+	enum bp_result (*get_connector_speed_cap_info)(
+		struct dc_bios *bios,
+		struct graphics_object_id object_id,
+		struct bp_connector_speed_cap_info *info);
+};
+
+struct bios_registers {
+	uint32_t BIOS_SCRATCH_3;
+	uint32_t BIOS_SCRATCH_6;
+};
+
+struct dc_bios {
+	const struct dc_vbios_funcs *funcs;
+
+	uint8_t *bios;
+	uint32_t bios_size;
+
+	uint8_t *bios_local_image;
+
+	struct dc_context *ctx;
+	const struct bios_registers *regs;
+	struct integrated_info *integrated_info;
+	struct dc_firmware_info fw_info;
+	bool fw_info_valid;
+	struct dc_vram_info vram_info;
+	struct bp_soc_bb_info bb_info;
+	struct dc_golden_table golden_table;
+};
+
+#endif /* DC_BIOS_TYPES_H */
diff --git a/drivers/gpu/drm/amd/display/dc/dc_ddc_types.h b/drivers/gpu/drm/amd/display/dc/dc_ddc_types.h
new file mode 100644
index 000000000000..428e3a9ab65a
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_ddc_types.h
@@ -0,0 +1,198 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+#ifndef DC_DDC_TYPES_H_
+#define DC_DDC_TYPES_H_
+
+enum aux_transaction_type {
+	AUX_TRANSACTION_TYPE_DP,
+	AUX_TRANSACTION_TYPE_I2C
+};
+
+
+enum i2caux_transaction_action {
+	I2CAUX_TRANSACTION_ACTION_I2C_WRITE = 0x00,
+	I2CAUX_TRANSACTION_ACTION_I2C_READ = 0x10,
+	I2CAUX_TRANSACTION_ACTION_I2C_STATUS_REQUEST = 0x20,
+
+	I2CAUX_TRANSACTION_ACTION_I2C_WRITE_MOT = 0x40,
+	I2CAUX_TRANSACTION_ACTION_I2C_READ_MOT = 0x50,
+	I2CAUX_TRANSACTION_ACTION_I2C_STATUS_REQUEST_MOT = 0x60,
+
+	I2CAUX_TRANSACTION_ACTION_DP_WRITE = 0x80,
+	I2CAUX_TRANSACTION_ACTION_DP_READ = 0x90
+};
+
+struct aux_request_transaction_data {
+	enum aux_transaction_type type;
+	enum i2caux_transaction_action action;
+	/* 20-bit AUX channel transaction address */
+	uint32_t address;
+	/* delay, in 100-microsecond units */
+	uint8_t delay;
+	uint32_t length;
+	uint8_t *data;
+};
+
+enum aux_transaction_reply {
+	AUX_TRANSACTION_REPLY_AUX_ACK = 0x00,
+	AUX_TRANSACTION_REPLY_AUX_NACK = 0x01,
+	AUX_TRANSACTION_REPLY_AUX_DEFER = 0x02,
+	AUX_TRANSACTION_REPLY_I2C_OVER_AUX_NACK = 0x04,
+	AUX_TRANSACTION_REPLY_I2C_OVER_AUX_DEFER = 0x08,
+
+	AUX_TRANSACTION_REPLY_I2C_ACK = 0x00,
+	AUX_TRANSACTION_REPLY_I2C_NACK = 0x10,
+	AUX_TRANSACTION_REPLY_I2C_DEFER = 0x20,
+
+	AUX_TRANSACTION_REPLY_HPD_DISCON = 0x40,
+
+	AUX_TRANSACTION_REPLY_INVALID = 0xFF
+};
+
+struct aux_reply_transaction_data {
+	enum aux_transaction_reply status;
+	uint32_t length;
+	uint8_t *data;
+};
+
+struct aux_payload {
+	/* set following flag to read/write I2C data,
+	 * reset it to read/write DPCD data */
+	bool i2c_over_aux;
+	/* set following flag to write data,
+	 * reset it to read data */
+	bool write;
+	bool mot;
+	bool write_status_update;
+
+	uint32_t address;
+	uint32_t length;
+	uint8_t *data;
+	/*
+	 * used to return the reply type of the transaction
+	 * ignored if NULL
+	 */
+	uint8_t *reply;
+	/* expressed in milliseconds
+	 * zero means "use default value"
+	 */
+	uint32_t defer_delay;
+
+};
+#define DEFAULT_AUX_MAX_DATA_SIZE 16
+
+struct i2c_payload {
+	bool write;
+	uint8_t address;
+	uint32_t length;
+	uint8_t *data;
+};
+
+enum i2c_command_engine {
+	I2C_COMMAND_ENGINE_DEFAULT,
+	I2C_COMMAND_ENGINE_SW,
+	I2C_COMMAND_ENGINE_HW
+};
+
+#define DDC_I2C_COMMAND_ENGINE I2C_COMMAND_ENGINE_SW
+
+struct i2c_command {
+	struct i2c_payload *payloads;
+	uint8_t number_of_payloads;
+
+	enum i2c_command_engine engine;
+
+	/* expressed in KHz
+	 * zero means "use default value" */
+	uint32_t speed;
+};
+
+struct gpio_ddc_hw_info {
+	bool hw_supported;
+	uint32_t ddc_channel;
+};
+
+struct ddc {
+	struct gpio *pin_data;
+	struct gpio *pin_clock;
+	struct gpio_ddc_hw_info hw_info;
+	struct dc_context *ctx;
+};
+
+union ddc_wa {
+	struct {
+		uint32_t DP_SKIP_POWER_OFF:1;
+		uint32_t DP_AUX_POWER_UP_WA_DELAY:1;
+	} bits;
+	uint32_t raw;
+};
+
+struct ddc_flags {
+	uint8_t EDID_QUERY_DONE_ONCE:1;
+	uint8_t IS_INTERNAL_DISPLAY:1;
+	uint8_t FORCE_READ_REPEATED_START:1;
+	uint8_t EDID_STRESS_READ:1;
+
+};
+
+enum ddc_transaction_type {
+	DDC_TRANSACTION_TYPE_NONE = 0,
+	DDC_TRANSACTION_TYPE_I2C,
+	DDC_TRANSACTION_TYPE_I2C_OVER_AUX,
+	DDC_TRANSACTION_TYPE_I2C_OVER_AUX_WITH_DEFER,
+	DDC_TRANSACTION_TYPE_I2C_OVER_AUX_RETRY_DEFER
+};
+
+enum display_dongle_type {
+	DISPLAY_DONGLE_NONE = 0,
+	/* Active converter types*/
+	DISPLAY_DONGLE_DP_VGA_CONVERTER,
+	DISPLAY_DONGLE_DP_DVI_CONVERTER,
+	DISPLAY_DONGLE_DP_HDMI_CONVERTER,
+	/* DP-HDMI/DVI passive dongles (Type 1 and Type 2)*/
+	DISPLAY_DONGLE_DP_DVI_DONGLE,
+	DISPLAY_DONGLE_DP_HDMI_DONGLE,
+	/* Other types of dongle*/
+	DISPLAY_DONGLE_DP_HDMI_MISMATCHED_DONGLE,
+};
+
+#define DC_MAX_EDID_BUFFER_SIZE 2048
+#define DC_EDID_BLOCK_SIZE 128
+
+struct ddc_service {
+	struct ddc *ddc_pin;
+	struct ddc_flags flags;
+	union ddc_wa wa;
+	enum ddc_transaction_type transaction_type;
+	enum display_dongle_type dongle_type;
+	struct dc_context *ctx;
+	struct dc_link *link;
+
+	uint32_t address;
+	uint32_t edid_buf_len;
+	uint8_t edid_buf[DC_MAX_EDID_BUFFER_SIZE];
+};
+
+#endif /* DC_DDC_TYPES_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c
new file mode 100644
index 000000000000..53a088ebddef
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c
@@ -0,0 +1,2250 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+#include "dc.h"
+#include "dc_dmub_srv.h"
+#include "../dmub/dmub_srv.h"
+#include "dm_helpers.h"
+#include "dc_hw_types.h"
+#include "core_types.h"
+#include "../basics/conversion.h"
+#include "cursor_reg_cache.h"
+#include "resource.h"
+#include "clk_mgr.h"
+#include "dc_state_priv.h"
+#include "dc_plane_priv.h"
+
+#define CTX dc_dmub_srv->ctx
+#define DC_LOGGER CTX->logger
+#define GPINT_RETRY_NUM 20
+
+static void dc_dmub_srv_construct(struct dc_dmub_srv *dc_srv, struct dc *dc,
+				  struct dmub_srv *dmub)
+{
+	dc_srv->dmub = dmub;
+	dc_srv->ctx = dc->ctx;
+}
+
+struct dc_dmub_srv *dc_dmub_srv_create(struct dc *dc, struct dmub_srv *dmub)
+{
+	struct dc_dmub_srv *dc_srv =
+		kzalloc(sizeof(struct dc_dmub_srv), GFP_KERNEL);
+
+	if (dc_srv == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	dc_dmub_srv_construct(dc_srv, dc, dmub);
+
+	return dc_srv;
+}
+
+void dc_dmub_srv_destroy(struct dc_dmub_srv **dmub_srv)
+{
+	if (*dmub_srv) {
+		kfree(*dmub_srv);
+		*dmub_srv = NULL;
+	}
+}
+
+bool dc_dmub_srv_wait_for_pending(struct dc_dmub_srv *dc_dmub_srv)
+{
+	struct dmub_srv *dmub;
+	struct dc_context *dc_ctx;
+	enum dmub_status status;
+
+	if (!dc_dmub_srv || !dc_dmub_srv->dmub)
+		return false;
+
+	dc_ctx = dc_dmub_srv->ctx;
+	dmub = dc_dmub_srv->dmub;
+
+	do {
+		status = dmub_srv_wait_for_pending(dmub, 100000);
+	} while (dc_dmub_srv->ctx->dc->debug.disable_timeout && status != DMUB_STATUS_OK);
+
+	if (status != DMUB_STATUS_OK) {
+		DC_ERROR("Error waiting for DMUB idle: status=%d\n", status);
+		dc_dmub_srv_log_diagnostic_data(dc_dmub_srv);
+	}
+
+	return status == DMUB_STATUS_OK;
+}
+
+void dc_dmub_srv_clear_inbox0_ack(struct dc_dmub_srv *dc_dmub_srv)
+{
+	struct dmub_srv *dmub = dc_dmub_srv->dmub;
+	struct dc_context *dc_ctx = dc_dmub_srv->ctx;
+	enum dmub_status status = DMUB_STATUS_OK;
+
+	status = dmub_srv_clear_inbox0_ack(dmub);
+	if (status != DMUB_STATUS_OK) {
+		DC_ERROR("Error clearing INBOX0 ack: status=%d\n", status);
+		dc_dmub_srv_log_diagnostic_data(dc_dmub_srv);
+	}
+}
+
+void dc_dmub_srv_wait_for_inbox0_ack(struct dc_dmub_srv *dc_dmub_srv)
+{
+	struct dmub_srv *dmub = dc_dmub_srv->dmub;
+	struct dc_context *dc_ctx = dc_dmub_srv->ctx;
+	enum dmub_status status = DMUB_STATUS_OK;
+
+	status = dmub_srv_wait_for_inbox0_ack(dmub, 100000);
+	if (status != DMUB_STATUS_OK) {
+		DC_ERROR("Error waiting for INBOX0 HW Lock Ack\n");
+		dc_dmub_srv_log_diagnostic_data(dc_dmub_srv);
+	}
+}
+
+void dc_dmub_srv_send_inbox0_cmd(struct dc_dmub_srv *dc_dmub_srv,
+				 union dmub_inbox0_data_register data)
+{
+	struct dmub_srv *dmub = dc_dmub_srv->dmub;
+	struct dc_context *dc_ctx = dc_dmub_srv->ctx;
+	enum dmub_status status = DMUB_STATUS_OK;
+
+	status = dmub_srv_send_inbox0_cmd(dmub, data);
+	if (status != DMUB_STATUS_OK) {
+		DC_ERROR("Error sending INBOX0 cmd\n");
+		dc_dmub_srv_log_diagnostic_data(dc_dmub_srv);
+	}
+}
+
+static bool dc_dmub_srv_reg_cmd_list_queue_execute(struct dc_dmub_srv *dc_dmub_srv,
+		unsigned int count,
+		union dmub_rb_cmd *cmd_list)
+{
+	struct dc_context *dc_ctx;
+	struct dmub_srv *dmub;
+	enum dmub_status status = DMUB_STATUS_OK;
+	int i;
+
+	if (!dc_dmub_srv || !dc_dmub_srv->dmub)
+		return false;
+
+	dc_ctx = dc_dmub_srv->ctx;
+	dmub = dc_dmub_srv->dmub;
+
+	for (i = 0 ; i < count; i++) {
+		/* confirm no messages pending */
+		do {
+			status = dmub_srv_wait_for_idle(dmub, 100000);
+		} while (dc_dmub_srv->ctx->dc->debug.disable_timeout && status != DMUB_STATUS_OK);
+
+		/* queue command */
+		if (status == DMUB_STATUS_OK)
+			status = dmub_srv_reg_cmd_execute(dmub, &cmd_list[i]);
+
+		/* check for errors */
+		if (status != DMUB_STATUS_OK) {
+			break;
+		}
+	}
+
+	if (status != DMUB_STATUS_OK) {
+		if (status != DMUB_STATUS_POWER_STATE_D3) {
+			DC_ERROR("Error starting DMUB execution: status=%d\n", status);
+			dc_dmub_srv_log_diagnostic_data(dc_dmub_srv);
+		}
+		return false;
+	}
+
+	return true;
+}
+
+static bool dc_dmub_srv_fb_cmd_list_queue_execute(struct dc_dmub_srv *dc_dmub_srv,
+		unsigned int count,
+		union dmub_rb_cmd *cmd_list)
+{
+	struct dc_context *dc_ctx;
+	struct dmub_srv *dmub;
+	enum dmub_status status;
+	int i;
+
+	if (!dc_dmub_srv || !dc_dmub_srv->dmub)
+		return false;
+
+	dc_ctx = dc_dmub_srv->ctx;
+	dmub = dc_dmub_srv->dmub;
+
+	for (i = 0 ; i < count; i++) {
+		// Queue command
+		if (!cmd_list[i].cmd_common.header.multi_cmd_pending ||
+				dmub_rb_num_free(&dmub->inbox1.rb) >= count - i) {
+			status = dmub_srv_fb_cmd_queue(dmub, &cmd_list[i]);
+		} else {
+			status = DMUB_STATUS_QUEUE_FULL;
+		}
+
+		if (status == DMUB_STATUS_QUEUE_FULL) {
+			/* Execute and wait for queue to become empty again. */
+			status = dmub_srv_fb_cmd_execute(dmub);
+			if (status == DMUB_STATUS_POWER_STATE_D3)
+				return false;
+
+			do {
+					status = dmub_srv_wait_for_inbox_free(dmub, 100000, count - i);
+			} while (dc_dmub_srv->ctx->dc->debug.disable_timeout && status != DMUB_STATUS_OK);
+
+			/* Requeue the command. */
+			status = dmub_srv_fb_cmd_queue(dmub, &cmd_list[i]);
+		}
+
+		if (status != DMUB_STATUS_OK) {
+			if (status != DMUB_STATUS_POWER_STATE_D3) {
+				DC_ERROR("Error queueing DMUB command: status=%d\n", status);
+				dc_dmub_srv_log_diagnostic_data(dc_dmub_srv);
+			}
+			return false;
+		}
+	}
+
+	status = dmub_srv_fb_cmd_execute(dmub);
+	if (status != DMUB_STATUS_OK) {
+		if (status != DMUB_STATUS_POWER_STATE_D3) {
+			DC_ERROR("Error starting DMUB execution: status=%d\n", status);
+			dc_dmub_srv_log_diagnostic_data(dc_dmub_srv);
+		}
+		return false;
+	}
+
+	return true;
+}
+
+bool dc_dmub_srv_cmd_list_queue_execute(struct dc_dmub_srv *dc_dmub_srv,
+		unsigned int count,
+		union dmub_rb_cmd *cmd_list)
+{
+	bool res = false;
+
+	if (dc_dmub_srv && dc_dmub_srv->dmub) {
+		if (dc_dmub_srv->dmub->inbox_type == DMUB_CMD_INTERFACE_REG) {
+			res = dc_dmub_srv_reg_cmd_list_queue_execute(dc_dmub_srv, count, cmd_list);
+		} else {
+			res = dc_dmub_srv_fb_cmd_list_queue_execute(dc_dmub_srv, count, cmd_list);
+		}
+
+		if (res)
+			res = dmub_srv_update_inbox_status(dc_dmub_srv->dmub) == DMUB_STATUS_OK;
+	}
+
+	return res;
+}
+
+bool dc_dmub_srv_wait_for_idle(struct dc_dmub_srv *dc_dmub_srv,
+		enum dm_dmub_wait_type wait_type,
+		union dmub_rb_cmd *cmd_list)
+{
+	struct dmub_srv *dmub;
+	enum dmub_status status;
+
+	if (!dc_dmub_srv || !dc_dmub_srv->dmub)
+		return false;
+
+	dmub = dc_dmub_srv->dmub;
+
+	// Wait for DMUB to process command
+	if (wait_type != DM_DMUB_WAIT_TYPE_NO_WAIT) {
+		do {
+			status = dmub_srv_wait_for_idle(dmub, 100000);
+		} while (dc_dmub_srv->ctx->dc->debug.disable_timeout && status != DMUB_STATUS_OK);
+
+		if (status != DMUB_STATUS_OK) {
+			DC_LOG_DEBUG("No reply for DMUB command: status=%d\n", status);
+			if (!dmub->debug.timeout_info.timeout_occured) {
+				dmub->debug.timeout_info.timeout_occured = true;
+				if (cmd_list)
+					dmub->debug.timeout_info.timeout_cmd = *cmd_list;
+				dmub->debug.timeout_info.timestamp = dm_get_timestamp(dc_dmub_srv->ctx);
+			}
+			dc_dmub_srv_log_diagnostic_data(dc_dmub_srv);
+			return false;
+		}
+
+		// Copy data back from ring buffer into command
+		if (wait_type == DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY && cmd_list) {
+			dmub_srv_cmd_get_response(dc_dmub_srv->dmub, cmd_list);
+		}
+	}
+
+	return true;
+}
+
+bool dc_dmub_srv_cmd_run(struct dc_dmub_srv *dc_dmub_srv, union dmub_rb_cmd *cmd, enum dm_dmub_wait_type wait_type)
+{
+	return dc_dmub_srv_cmd_run_list(dc_dmub_srv, 1, cmd, wait_type);
+}
+
+bool dc_dmub_srv_cmd_run_list(struct dc_dmub_srv *dc_dmub_srv, unsigned int count, union dmub_rb_cmd *cmd_list, enum dm_dmub_wait_type wait_type)
+{
+	if (!dc_dmub_srv_cmd_list_queue_execute(dc_dmub_srv, count, cmd_list))
+		return false;
+
+	return dc_dmub_srv_wait_for_idle(dc_dmub_srv, wait_type, cmd_list);
+}
+
+bool dc_dmub_srv_optimized_init_done(struct dc_dmub_srv *dc_dmub_srv)
+{
+	struct dmub_srv *dmub;
+	struct dc_context *dc_ctx;
+	union dmub_fw_boot_status boot_status;
+	enum dmub_status status;
+
+	if (!dc_dmub_srv || !dc_dmub_srv->dmub)
+		return false;
+
+	dmub = dc_dmub_srv->dmub;
+	dc_ctx = dc_dmub_srv->ctx;
+
+	status = dmub_srv_get_fw_boot_status(dmub, &boot_status);
+	if (status != DMUB_STATUS_OK) {
+		DC_ERROR("Error querying DMUB boot status: error=%d\n", status);
+		return false;
+	}
+
+	return boot_status.bits.optimized_init_done;
+}
+
+bool dc_dmub_srv_notify_stream_mask(struct dc_dmub_srv *dc_dmub_srv,
+				    unsigned int stream_mask)
+{
+	if (!dc_dmub_srv || !dc_dmub_srv->dmub)
+		return false;
+
+	return dc_wake_and_execute_gpint(dc_dmub_srv->ctx, DMUB_GPINT__IDLE_OPT_NOTIFY_STREAM_MASK,
+					 stream_mask, NULL, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+bool dc_dmub_srv_is_restore_required(struct dc_dmub_srv *dc_dmub_srv)
+{
+	struct dmub_srv *dmub;
+	struct dc_context *dc_ctx;
+	union dmub_fw_boot_status boot_status;
+	enum dmub_status status;
+
+	if (!dc_dmub_srv || !dc_dmub_srv->dmub)
+		return false;
+
+	dmub = dc_dmub_srv->dmub;
+	dc_ctx = dc_dmub_srv->ctx;
+
+	status = dmub_srv_get_fw_boot_status(dmub, &boot_status);
+	if (status != DMUB_STATUS_OK) {
+		DC_ERROR("Error querying DMUB boot status: error=%d\n", status);
+		return false;
+	}
+
+	return boot_status.bits.restore_required;
+}
+
+bool dc_dmub_srv_get_dmub_outbox0_msg(const struct dc *dc, struct dmcub_trace_buf_entry *entry)
+{
+	struct dmub_srv *dmub = dc->ctx->dmub_srv->dmub;
+	return dmub_srv_get_outbox0_msg(dmub, entry);
+}
+
+void dc_dmub_trace_event_control(struct dc *dc, bool enable)
+{
+	dm_helpers_dmub_outbox_interrupt_control(dc->ctx, enable);
+}
+
+void dc_dmub_srv_drr_update_cmd(struct dc *dc, uint32_t tg_inst, uint32_t vtotal_min, uint32_t vtotal_max)
+{
+	union dmub_rb_cmd cmd = { 0 };
+
+	cmd.drr_update.header.type = DMUB_CMD__FW_ASSISTED_MCLK_SWITCH;
+	cmd.drr_update.header.sub_type = DMUB_CMD__FAMS_DRR_UPDATE;
+	cmd.drr_update.dmub_optc_state_req.v_total_max = vtotal_max;
+	cmd.drr_update.dmub_optc_state_req.v_total_min = vtotal_min;
+	cmd.drr_update.dmub_optc_state_req.tg_inst = tg_inst;
+
+	cmd.drr_update.header.payload_bytes = sizeof(cmd.drr_update) - sizeof(cmd.drr_update.header);
+
+	// Send the command to the DMCUB.
+	dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+void dc_dmub_srv_set_drr_manual_trigger_cmd(struct dc *dc, uint32_t tg_inst)
+{
+	union dmub_rb_cmd cmd = { 0 };
+
+	cmd.drr_update.header.type = DMUB_CMD__FW_ASSISTED_MCLK_SWITCH;
+	cmd.drr_update.header.sub_type = DMUB_CMD__FAMS_SET_MANUAL_TRIGGER;
+	cmd.drr_update.dmub_optc_state_req.tg_inst = tg_inst;
+
+	cmd.drr_update.header.payload_bytes = sizeof(cmd.drr_update) - sizeof(cmd.drr_update.header);
+
+	// Send the command to the DMCUB.
+	dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+static uint8_t dc_dmub_srv_get_pipes_for_stream(struct dc *dc, struct dc_stream_state *stream)
+{
+	uint8_t pipes = 0;
+	int i = 0;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct pipe_ctx *pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+
+		if (pipe->stream == stream && pipe->stream_res.tg)
+			pipes = i;
+	}
+	return pipes;
+}
+
+static void dc_dmub_srv_populate_fams_pipe_info(struct dc *dc, struct dc_state *context,
+		struct pipe_ctx *head_pipe,
+		struct dmub_cmd_fw_assisted_mclk_switch_pipe_data *fams_pipe_data)
+{
+	int j;
+	int pipe_idx = 0;
+
+	fams_pipe_data->pipe_index[pipe_idx++] = head_pipe->plane_res.hubp->inst;
+	for (j = 0; j < dc->res_pool->pipe_count; j++) {
+		struct pipe_ctx *split_pipe = &context->res_ctx.pipe_ctx[j];
+
+		if (split_pipe->stream == head_pipe->stream && (split_pipe->top_pipe || split_pipe->prev_odm_pipe)) {
+			fams_pipe_data->pipe_index[pipe_idx++] = split_pipe->plane_res.hubp->inst;
+		}
+	}
+	fams_pipe_data->pipe_count = pipe_idx;
+}
+
+bool dc_dmub_srv_p_state_delegate(struct dc *dc, bool should_manage_pstate, struct dc_state *context)
+{
+	union dmub_rb_cmd cmd = { 0 };
+	struct dmub_cmd_fw_assisted_mclk_switch_config *config_data = &cmd.fw_assisted_mclk_switch.config_data;
+	int i = 0, k = 0;
+	int ramp_up_num_steps = 1; // TODO: Ramp is currently disabled. Reenable it.
+	uint8_t visual_confirm_enabled;
+	int pipe_idx = 0;
+	struct dc_stream_status *stream_status = NULL;
+
+	if (dc == NULL)
+		return false;
+
+	visual_confirm_enabled = dc->debug.visual_confirm == VISUAL_CONFIRM_FAMS;
+
+	// Format command.
+	cmd.fw_assisted_mclk_switch.header.type = DMUB_CMD__FW_ASSISTED_MCLK_SWITCH;
+	cmd.fw_assisted_mclk_switch.header.sub_type = DMUB_CMD__FAMS_SETUP_FW_CTRL;
+	cmd.fw_assisted_mclk_switch.config_data.fams_enabled = should_manage_pstate;
+	cmd.fw_assisted_mclk_switch.config_data.visual_confirm_enabled = visual_confirm_enabled;
+
+	if (should_manage_pstate) {
+		for (i = 0, pipe_idx = 0; i < dc->res_pool->pipe_count; i++) {
+			struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
+
+			if (!pipe->stream)
+				continue;
+
+			/* If FAMS is being used to support P-State and there is a stream
+			 * that does not use FAMS, we are in an FPO + VActive scenario.
+			 * Assign vactive stretch margin in this case.
+			 */
+			stream_status = dc_state_get_stream_status(context, pipe->stream);
+			if (stream_status && !stream_status->fpo_in_use) {
+				cmd.fw_assisted_mclk_switch.config_data.vactive_stretch_margin_us = dc->debug.fpo_vactive_margin_us;
+				break;
+			}
+			pipe_idx++;
+		}
+	}
+
+	for (i = 0, k = 0; context && i < dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
+
+		if (!resource_is_pipe_type(pipe, OTG_MASTER))
+			continue;
+
+		stream_status = dc_state_get_stream_status(context, pipe->stream);
+		if (stream_status && stream_status->fpo_in_use) {
+			struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
+			uint8_t min_refresh_in_hz = (pipe->stream->timing.min_refresh_in_uhz + 999999) / 1000000;
+
+			config_data->pipe_data[k].pix_clk_100hz = pipe->stream->timing.pix_clk_100hz;
+			config_data->pipe_data[k].min_refresh_in_hz = min_refresh_in_hz;
+			config_data->pipe_data[k].max_ramp_step = ramp_up_num_steps;
+			config_data->pipe_data[k].pipes = dc_dmub_srv_get_pipes_for_stream(dc, pipe->stream);
+			dc_dmub_srv_populate_fams_pipe_info(dc, context, pipe, &config_data->pipe_data[k]);
+			k++;
+		}
+	}
+	cmd.fw_assisted_mclk_switch.header.payload_bytes =
+		sizeof(cmd.fw_assisted_mclk_switch) - sizeof(cmd.fw_assisted_mclk_switch.header);
+
+	// Send the command to the DMCUB.
+	dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+
+	return true;
+}
+
+void dc_dmub_srv_query_caps_cmd(struct dc_dmub_srv *dc_dmub_srv)
+{
+	union dmub_rb_cmd cmd = { 0 };
+
+	if (dc_dmub_srv->ctx->dc->debug.dmcub_emulation)
+		return;
+
+	memset(&cmd, 0, sizeof(cmd));
+
+	/* Prepare fw command */
+	cmd.query_feature_caps.header.type = DMUB_CMD__QUERY_FEATURE_CAPS;
+	cmd.query_feature_caps.header.sub_type = 0;
+	cmd.query_feature_caps.header.ret_status = 1;
+	cmd.query_feature_caps.header.payload_bytes = sizeof(struct dmub_cmd_query_feature_caps_data);
+
+	/* If command was processed, copy feature caps to dmub srv */
+	if (dc_wake_and_execute_dmub_cmd(dc_dmub_srv->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY) &&
+	    cmd.query_feature_caps.header.ret_status == 0) {
+		memcpy(&dc_dmub_srv->dmub->feature_caps,
+		       &cmd.query_feature_caps.query_feature_caps_data,
+		       sizeof(struct dmub_feature_caps));
+	}
+}
+
+void dc_dmub_srv_get_visual_confirm_color_cmd(struct dc *dc, struct pipe_ctx *pipe_ctx)
+{
+	union dmub_rb_cmd cmd = { 0 };
+	unsigned int panel_inst = 0;
+
+	if (!dc_get_edp_link_panel_inst(dc, pipe_ctx->stream->link, &panel_inst) &&
+			dc->debug.visual_confirm == VISUAL_CONFIRM_DISABLE)
+		return;
+
+	memset(&cmd, 0, sizeof(cmd));
+
+	// Prepare fw command
+	cmd.visual_confirm_color.header.type = DMUB_CMD__GET_VISUAL_CONFIRM_COLOR;
+	cmd.visual_confirm_color.header.sub_type = 0;
+	cmd.visual_confirm_color.header.ret_status = 1;
+	cmd.visual_confirm_color.header.payload_bytes = sizeof(struct dmub_cmd_visual_confirm_color_data);
+	cmd.visual_confirm_color.visual_confirm_color_data.visual_confirm_color.panel_inst = panel_inst;
+
+	// If command was processed, copy feature caps to dmub srv
+	if (dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY) &&
+		cmd.visual_confirm_color.header.ret_status == 0) {
+		memcpy(&dc->ctx->dmub_srv->dmub->visual_confirm_color,
+			&cmd.visual_confirm_color.visual_confirm_color_data,
+			sizeof(struct dmub_visual_confirm_color));
+	}
+}
+
+/**
+ * populate_subvp_cmd_drr_info - Helper to populate DRR pipe info for the DMCUB subvp command
+ *
+ * @dc: [in] pointer to dc object
+ * @subvp_pipe: [in] pipe_ctx for the SubVP pipe
+ * @vblank_pipe: [in] pipe_ctx for the DRR pipe
+ * @pipe_data: [in] Pipe data which stores the VBLANK/DRR info
+ * @context: [in] DC state for access to phantom stream
+ *
+ * Populate the DMCUB SubVP command with DRR pipe info. All the information
+ * required for calculating the SubVP + DRR microschedule is populated here.
+ *
+ * High level algorithm:
+ * 1. Get timing for SubVP pipe, phantom pipe, and DRR pipe
+ * 2. Calculate the min and max vtotal which supports SubVP + DRR microschedule
+ * 3. Populate the drr_info with the min and max supported vtotal values
+ */
+static void populate_subvp_cmd_drr_info(struct dc *dc,
+		struct dc_state *context,
+		struct pipe_ctx *subvp_pipe,
+		struct pipe_ctx *vblank_pipe,
+		struct dmub_cmd_fw_assisted_mclk_switch_pipe_data_v2 *pipe_data)
+{
+	struct dc_stream_state *phantom_stream = dc_state_get_paired_subvp_stream(context, subvp_pipe->stream);
+	struct dc_crtc_timing *main_timing = &subvp_pipe->stream->timing;
+	struct dc_crtc_timing *phantom_timing;
+	struct dc_crtc_timing *drr_timing = &vblank_pipe->stream->timing;
+	uint16_t drr_frame_us = 0;
+	uint16_t min_drr_supported_us = 0;
+	uint16_t max_drr_supported_us = 0;
+	uint16_t max_drr_vblank_us = 0;
+	uint16_t max_drr_mallregion_us = 0;
+	uint16_t mall_region_us = 0;
+	uint16_t prefetch_us = 0;
+	uint16_t subvp_active_us = 0;
+	uint16_t drr_active_us = 0;
+	uint16_t min_vtotal_supported = 0;
+	uint16_t max_vtotal_supported = 0;
+
+	if (!phantom_stream)
+		return;
+
+	phantom_timing = &phantom_stream->timing;
+
+	pipe_data->pipe_config.vblank_data.drr_info.drr_in_use = true;
+	pipe_data->pipe_config.vblank_data.drr_info.use_ramping = false; // for now don't use ramping
+	pipe_data->pipe_config.vblank_data.drr_info.drr_window_size_ms = 4; // hardcode 4ms DRR window for now
+
+	drr_frame_us = div64_u64(((uint64_t)drr_timing->v_total * drr_timing->h_total * 1000000),
+			(((uint64_t)drr_timing->pix_clk_100hz * 100)));
+	// P-State allow width and FW delays already included phantom_timing->v_addressable
+	mall_region_us = div64_u64(((uint64_t)phantom_timing->v_addressable * phantom_timing->h_total * 1000000),
+			(((uint64_t)phantom_timing->pix_clk_100hz * 100)));
+	min_drr_supported_us = drr_frame_us + mall_region_us + SUBVP_DRR_MARGIN_US;
+	min_vtotal_supported = div64_u64(((uint64_t)drr_timing->pix_clk_100hz * 100 * min_drr_supported_us),
+			(((uint64_t)drr_timing->h_total * 1000000)));
+
+	prefetch_us = div64_u64(((uint64_t)(phantom_timing->v_total - phantom_timing->v_front_porch) * phantom_timing->h_total * 1000000),
+			(((uint64_t)phantom_timing->pix_clk_100hz * 100) + dc->caps.subvp_prefetch_end_to_mall_start_us));
+	subvp_active_us = div64_u64(((uint64_t)main_timing->v_addressable * main_timing->h_total * 1000000),
+			(((uint64_t)main_timing->pix_clk_100hz * 100)));
+	drr_active_us = div64_u64(((uint64_t)drr_timing->v_addressable * drr_timing->h_total * 1000000),
+			(((uint64_t)drr_timing->pix_clk_100hz * 100)));
+	max_drr_vblank_us = div64_u64((subvp_active_us - prefetch_us -
+			dc->caps.subvp_fw_processing_delay_us - drr_active_us), 2) + drr_active_us;
+	max_drr_mallregion_us = subvp_active_us - prefetch_us - mall_region_us - dc->caps.subvp_fw_processing_delay_us;
+	max_drr_supported_us = max_drr_vblank_us > max_drr_mallregion_us ? max_drr_vblank_us : max_drr_mallregion_us;
+	max_vtotal_supported = div64_u64(((uint64_t)drr_timing->pix_clk_100hz * 100 * max_drr_supported_us),
+			(((uint64_t)drr_timing->h_total * 1000000)));
+
+	/* When calculating the max vtotal supported for SubVP + DRR cases, add
+	 * margin due to possible rounding errors (being off by 1 line in the
+	 * FW calculation can incorrectly push the P-State switch to wait 1 frame
+	 * longer).
+	 */
+	max_vtotal_supported = max_vtotal_supported - dc->caps.subvp_drr_max_vblank_margin_us;
+
+	pipe_data->pipe_config.vblank_data.drr_info.min_vtotal_supported = min_vtotal_supported;
+	pipe_data->pipe_config.vblank_data.drr_info.max_vtotal_supported = max_vtotal_supported;
+	pipe_data->pipe_config.vblank_data.drr_info.drr_vblank_start_margin = dc->caps.subvp_drr_vblank_start_margin_us;
+}
+
+/**
+ * populate_subvp_cmd_vblank_pipe_info - Helper to populate VBLANK pipe info for the DMUB subvp command
+ *
+ * @dc: [in] current dc state
+ * @context: [in] new dc state
+ * @cmd: [in] DMUB cmd to be populated with SubVP info
+ * @vblank_pipe: [in] pipe_ctx for the VBLANK pipe
+ * @cmd_pipe_index: [in] index for the pipe array in DMCUB SubVP cmd
+ *
+ * Populate the DMCUB SubVP command with VBLANK pipe info. All the information
+ * required to calculate the microschedule for SubVP + VBLANK case is stored in
+ * the pipe_data (subvp_data and vblank_data).  Also check if the VBLANK pipe
+ * is a DRR display -- if it is make a call to populate drr_info.
+ */
+static void populate_subvp_cmd_vblank_pipe_info(struct dc *dc,
+		struct dc_state *context,
+		union dmub_rb_cmd *cmd,
+		struct pipe_ctx *vblank_pipe,
+		uint8_t cmd_pipe_index)
+{
+	uint32_t i;
+	struct pipe_ctx *pipe = NULL;
+	struct dmub_cmd_fw_assisted_mclk_switch_pipe_data_v2 *pipe_data =
+			&cmd->fw_assisted_mclk_switch_v2.config_data.pipe_data[cmd_pipe_index];
+
+	// Find the SubVP pipe
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		pipe = &context->res_ctx.pipe_ctx[i];
+
+		// We check for master pipe, but it shouldn't matter since we only need
+		// the pipe for timing info (stream should be same for any pipe splits)
+		if (!resource_is_pipe_type(pipe, OTG_MASTER) ||
+				!resource_is_pipe_type(pipe, DPP_PIPE))
+			continue;
+
+		// Find the SubVP pipe
+		if (dc_state_get_pipe_subvp_type(context, pipe) == SUBVP_MAIN)
+			break;
+	}
+
+	pipe_data->mode = VBLANK;
+	pipe_data->pipe_config.vblank_data.pix_clk_100hz = vblank_pipe->stream->timing.pix_clk_100hz;
+	pipe_data->pipe_config.vblank_data.vblank_start = vblank_pipe->stream->timing.v_total -
+							vblank_pipe->stream->timing.v_front_porch;
+	pipe_data->pipe_config.vblank_data.vtotal = vblank_pipe->stream->timing.v_total;
+	pipe_data->pipe_config.vblank_data.htotal = vblank_pipe->stream->timing.h_total;
+	pipe_data->pipe_config.vblank_data.vblank_pipe_index = vblank_pipe->pipe_idx;
+	pipe_data->pipe_config.vblank_data.vstartup_start = vblank_pipe->pipe_dlg_param.vstartup_start;
+	pipe_data->pipe_config.vblank_data.vblank_end =
+			vblank_pipe->stream->timing.v_total - vblank_pipe->stream->timing.v_front_porch - vblank_pipe->stream->timing.v_addressable;
+
+	if (vblank_pipe->stream->ignore_msa_timing_param &&
+		(vblank_pipe->stream->allow_freesync || vblank_pipe->stream->vrr_active_variable || vblank_pipe->stream->vrr_active_fixed))
+		populate_subvp_cmd_drr_info(dc, context, pipe, vblank_pipe, pipe_data);
+}
+
+/**
+ * update_subvp_prefetch_end_to_mall_start - Helper for SubVP + SubVP case
+ *
+ * @dc: [in] current dc state
+ * @context: [in] new dc state
+ * @cmd: [in] DMUB cmd to be populated with SubVP info
+ * @subvp_pipes: [in] Array of SubVP pipes (should always be length 2)
+ *
+ * For SubVP + SubVP, we use a single vertical interrupt to start the
+ * microschedule for both SubVP pipes. In order for this to work correctly, the
+ * MALL REGION of both SubVP pipes must start at the same time. This function
+ * lengthens the prefetch end to mall start delay of the SubVP pipe that has
+ * the shorter prefetch so that both MALL REGION's will start at the same time.
+ */
+static void update_subvp_prefetch_end_to_mall_start(struct dc *dc,
+		struct dc_state *context,
+		union dmub_rb_cmd *cmd,
+		struct pipe_ctx *subvp_pipes[])
+{
+	uint32_t subvp0_prefetch_us = 0;
+	uint32_t subvp1_prefetch_us = 0;
+	uint32_t prefetch_delta_us = 0;
+	struct dc_stream_state *phantom_stream0 = NULL;
+	struct dc_stream_state *phantom_stream1 = NULL;
+	struct dc_crtc_timing *phantom_timing0 = NULL;
+	struct dc_crtc_timing *phantom_timing1 = NULL;
+	struct dmub_cmd_fw_assisted_mclk_switch_pipe_data_v2 *pipe_data = NULL;
+
+	phantom_stream0 = dc_state_get_paired_subvp_stream(context, subvp_pipes[0]->stream);
+	if (!phantom_stream0)
+		return;
+
+	phantom_stream1 = dc_state_get_paired_subvp_stream(context, subvp_pipes[1]->stream);
+	if (!phantom_stream1)
+		return;
+
+	phantom_timing0 = &phantom_stream0->timing;
+	phantom_timing1 = &phantom_stream1->timing;
+
+	subvp0_prefetch_us = div64_u64(((uint64_t)(phantom_timing0->v_total - phantom_timing0->v_front_porch) *
+			(uint64_t)phantom_timing0->h_total * 1000000),
+			(((uint64_t)phantom_timing0->pix_clk_100hz * 100) + dc->caps.subvp_prefetch_end_to_mall_start_us));
+	subvp1_prefetch_us = div64_u64(((uint64_t)(phantom_timing1->v_total - phantom_timing1->v_front_porch) *
+			(uint64_t)phantom_timing1->h_total * 1000000),
+			(((uint64_t)phantom_timing1->pix_clk_100hz * 100) + dc->caps.subvp_prefetch_end_to_mall_start_us));
+
+	// Whichever SubVP PIPE has the smaller prefetch (including the prefetch end to mall start time)
+	// should increase it's prefetch time to match the other
+	if (subvp0_prefetch_us > subvp1_prefetch_us) {
+		pipe_data = &cmd->fw_assisted_mclk_switch_v2.config_data.pipe_data[1];
+		prefetch_delta_us = subvp0_prefetch_us - subvp1_prefetch_us;
+		pipe_data->pipe_config.subvp_data.prefetch_to_mall_start_lines =
+				div64_u64(((uint64_t)(dc->caps.subvp_prefetch_end_to_mall_start_us + prefetch_delta_us) *
+					((uint64_t)phantom_timing1->pix_clk_100hz * 100) + ((uint64_t)phantom_timing1->h_total * 1000000 - 1)),
+					((uint64_t)phantom_timing1->h_total * 1000000));
+
+	} else if (subvp1_prefetch_us >  subvp0_prefetch_us) {
+		pipe_data = &cmd->fw_assisted_mclk_switch_v2.config_data.pipe_data[0];
+		prefetch_delta_us = subvp1_prefetch_us - subvp0_prefetch_us;
+		pipe_data->pipe_config.subvp_data.prefetch_to_mall_start_lines =
+				div64_u64(((uint64_t)(dc->caps.subvp_prefetch_end_to_mall_start_us + prefetch_delta_us) *
+					((uint64_t)phantom_timing0->pix_clk_100hz * 100) + ((uint64_t)phantom_timing0->h_total * 1000000 - 1)),
+					((uint64_t)phantom_timing0->h_total * 1000000));
+	}
+}
+
+/**
+ * populate_subvp_cmd_pipe_info - Helper to populate the SubVP pipe info for the DMUB subvp command
+ *
+ * @dc: [in] current dc state
+ * @context: [in] new dc state
+ * @cmd: [in] DMUB cmd to be populated with SubVP info
+ * @subvp_pipe: [in] pipe_ctx for the SubVP pipe
+ * @cmd_pipe_index: [in] index for the pipe array in DMCUB SubVP cmd
+ *
+ * Populate the DMCUB SubVP command with SubVP pipe info. All the information
+ * required to calculate the microschedule for the SubVP pipe is stored in the
+ * pipe_data of the DMCUB SubVP command.
+ */
+static void populate_subvp_cmd_pipe_info(struct dc *dc,
+		struct dc_state *context,
+		union dmub_rb_cmd *cmd,
+		struct pipe_ctx *subvp_pipe,
+		uint8_t cmd_pipe_index)
+{
+	uint32_t j;
+	struct dmub_cmd_fw_assisted_mclk_switch_pipe_data_v2 *pipe_data =
+			&cmd->fw_assisted_mclk_switch_v2.config_data.pipe_data[cmd_pipe_index];
+	struct dc_stream_state *phantom_stream = dc_state_get_paired_subvp_stream(context, subvp_pipe->stream);
+	struct dc_crtc_timing *main_timing = &subvp_pipe->stream->timing;
+	struct dc_crtc_timing *phantom_timing;
+	uint32_t out_num_stream, out_den_stream, out_num_plane, out_den_plane, out_num, out_den;
+
+	if (!phantom_stream)
+		return;
+
+	phantom_timing = &phantom_stream->timing;
+
+	pipe_data->mode = SUBVP;
+	pipe_data->pipe_config.subvp_data.pix_clk_100hz = subvp_pipe->stream->timing.pix_clk_100hz;
+	pipe_data->pipe_config.subvp_data.htotal = subvp_pipe->stream->timing.h_total;
+	pipe_data->pipe_config.subvp_data.vtotal = subvp_pipe->stream->timing.v_total;
+	pipe_data->pipe_config.subvp_data.main_vblank_start =
+			main_timing->v_total - main_timing->v_front_porch;
+	pipe_data->pipe_config.subvp_data.main_vblank_end =
+			main_timing->v_total - main_timing->v_front_porch - main_timing->v_addressable;
+	pipe_data->pipe_config.subvp_data.mall_region_lines = phantom_timing->v_addressable;
+	pipe_data->pipe_config.subvp_data.main_pipe_index = subvp_pipe->stream_res.tg->inst;
+	pipe_data->pipe_config.subvp_data.is_drr = subvp_pipe->stream->ignore_msa_timing_param &&
+		(subvp_pipe->stream->allow_freesync || subvp_pipe->stream->vrr_active_variable || subvp_pipe->stream->vrr_active_fixed);
+
+	/* Calculate the scaling factor from the src and dst height.
+	 * e.g. If 3840x2160 being downscaled to 1920x1080, the scaling factor is 1/2.
+	 * Reduce the fraction 1080/2160 = 1/2 for the "scaling factor"
+	 *
+	 * Make sure to combine stream and plane scaling together.
+	 */
+	reduce_fraction(subvp_pipe->stream->src.height, subvp_pipe->stream->dst.height,
+			&out_num_stream, &out_den_stream);
+	reduce_fraction(subvp_pipe->plane_state->src_rect.height, subvp_pipe->plane_state->dst_rect.height,
+			&out_num_plane, &out_den_plane);
+	reduce_fraction(out_num_stream * out_num_plane, out_den_stream * out_den_plane, &out_num, &out_den);
+	pipe_data->pipe_config.subvp_data.scale_factor_numerator = out_num;
+	pipe_data->pipe_config.subvp_data.scale_factor_denominator = out_den;
+
+	// Prefetch lines is equal to VACTIVE + BP + VSYNC
+	pipe_data->pipe_config.subvp_data.prefetch_lines =
+			phantom_timing->v_total - phantom_timing->v_front_porch;
+
+	// Round up
+	pipe_data->pipe_config.subvp_data.prefetch_to_mall_start_lines =
+			div64_u64(((uint64_t)dc->caps.subvp_prefetch_end_to_mall_start_us * ((uint64_t)phantom_timing->pix_clk_100hz * 100) +
+					((uint64_t)phantom_timing->h_total * 1000000 - 1)), ((uint64_t)phantom_timing->h_total * 1000000));
+	pipe_data->pipe_config.subvp_data.processing_delay_lines =
+			div64_u64(((uint64_t)(dc->caps.subvp_fw_processing_delay_us) * ((uint64_t)phantom_timing->pix_clk_100hz * 100) +
+					((uint64_t)phantom_timing->h_total * 1000000 - 1)), ((uint64_t)phantom_timing->h_total * 1000000));
+
+	if (subvp_pipe->bottom_pipe) {
+		pipe_data->pipe_config.subvp_data.main_split_pipe_index = subvp_pipe->bottom_pipe->pipe_idx;
+	} else if (subvp_pipe->next_odm_pipe) {
+		pipe_data->pipe_config.subvp_data.main_split_pipe_index = subvp_pipe->next_odm_pipe->pipe_idx;
+	} else {
+		pipe_data->pipe_config.subvp_data.main_split_pipe_index = 0xF;
+	}
+
+	// Find phantom pipe index based on phantom stream
+	for (j = 0; j < dc->res_pool->pipe_count; j++) {
+		struct pipe_ctx *phantom_pipe = &context->res_ctx.pipe_ctx[j];
+
+		if (resource_is_pipe_type(phantom_pipe, OTG_MASTER) &&
+				phantom_pipe->stream == dc_state_get_paired_subvp_stream(context, subvp_pipe->stream)) {
+			pipe_data->pipe_config.subvp_data.phantom_pipe_index = phantom_pipe->stream_res.tg->inst;
+			if (phantom_pipe->bottom_pipe) {
+				pipe_data->pipe_config.subvp_data.phantom_split_pipe_index = phantom_pipe->bottom_pipe->plane_res.hubp->inst;
+			} else if (phantom_pipe->next_odm_pipe) {
+				pipe_data->pipe_config.subvp_data.phantom_split_pipe_index = phantom_pipe->next_odm_pipe->plane_res.hubp->inst;
+			} else {
+				pipe_data->pipe_config.subvp_data.phantom_split_pipe_index = 0xF;
+			}
+			break;
+		}
+	}
+}
+
+/**
+ * dc_dmub_setup_subvp_dmub_command - Populate the DMCUB SubVP command
+ *
+ * @dc: [in] current dc state
+ * @context: [in] new dc state
+ * @enable: [in] if true enables the pipes population
+ *
+ * This function loops through each pipe and populates the DMUB SubVP CMD info
+ * based on the pipe (e.g. SubVP, VBLANK).
+ */
+void dc_dmub_setup_subvp_dmub_command(struct dc *dc,
+		struct dc_state *context,
+		bool enable)
+{
+	uint8_t cmd_pipe_index = 0;
+	uint32_t i, pipe_idx;
+	uint8_t subvp_count = 0;
+	union dmub_rb_cmd cmd;
+	struct pipe_ctx *subvp_pipes[2];
+	uint32_t wm_val_refclk = 0;
+	enum mall_stream_type pipe_mall_type;
+
+	memset(&cmd, 0, sizeof(cmd));
+	// FW command for SUBVP
+	cmd.fw_assisted_mclk_switch_v2.header.type = DMUB_CMD__FW_ASSISTED_MCLK_SWITCH;
+	cmd.fw_assisted_mclk_switch_v2.header.sub_type = DMUB_CMD__HANDLE_SUBVP_CMD;
+	cmd.fw_assisted_mclk_switch_v2.header.payload_bytes =
+			sizeof(cmd.fw_assisted_mclk_switch_v2) - sizeof(cmd.fw_assisted_mclk_switch_v2.header);
+
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
+
+		/* For SubVP pipe count, only count the top most (ODM / MPC) pipe
+		 */
+		if (resource_is_pipe_type(pipe, OTG_MASTER) &&
+				resource_is_pipe_type(pipe, DPP_PIPE) &&
+				dc_state_get_pipe_subvp_type(context, pipe) == SUBVP_MAIN)
+			subvp_pipes[subvp_count++] = pipe;
+	}
+
+	if (enable) {
+		// For each pipe that is a "main" SUBVP pipe, fill in pipe data for DMUB SUBVP cmd
+		for (i = 0, pipe_idx = 0; i < dc->res_pool->pipe_count; i++) {
+			struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
+			pipe_mall_type = dc_state_get_pipe_subvp_type(context, pipe);
+
+			if (!pipe->stream)
+				continue;
+
+			/* When populating subvp cmd info, only pass in the top most (ODM / MPC) pipe.
+			 * Any ODM or MPC splits being used in SubVP will be handled internally in
+			 * populate_subvp_cmd_pipe_info
+			 */
+			if (resource_is_pipe_type(pipe, OTG_MASTER) &&
+					resource_is_pipe_type(pipe, DPP_PIPE) &&
+					pipe_mall_type == SUBVP_MAIN) {
+				populate_subvp_cmd_pipe_info(dc, context, &cmd, pipe, cmd_pipe_index++);
+			} else if (resource_is_pipe_type(pipe, OTG_MASTER) &&
+					resource_is_pipe_type(pipe, DPP_PIPE) &&
+					pipe_mall_type == SUBVP_NONE) {
+				// Don't need to check for ActiveDRAMClockChangeMargin < 0, not valid in cases where
+				// we run through DML without calculating "natural" P-state support
+				populate_subvp_cmd_vblank_pipe_info(dc, context, &cmd, pipe, cmd_pipe_index++);
+
+			}
+			pipe_idx++;
+		}
+		if (subvp_count == 2) {
+			update_subvp_prefetch_end_to_mall_start(dc, context, &cmd, subvp_pipes);
+		}
+		cmd.fw_assisted_mclk_switch_v2.config_data.pstate_allow_width_us = dc->caps.subvp_pstate_allow_width_us;
+		cmd.fw_assisted_mclk_switch_v2.config_data.vertical_int_margin_us = dc->caps.subvp_vertical_int_margin_us;
+
+		// Store the original watermark value for this SubVP config so we can lower it when the
+		// MCLK switch starts
+		wm_val_refclk = context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.pstate_change_ns *
+				(dc->res_pool->ref_clocks.dchub_ref_clock_inKhz / 1000) / 1000;
+
+		cmd.fw_assisted_mclk_switch_v2.config_data.watermark_a_cache = wm_val_refclk < 0xFFFF ? wm_val_refclk : 0xFFFF;
+	}
+
+	dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+bool dc_dmub_srv_get_diagnostic_data(struct dc_dmub_srv *dc_dmub_srv)
+{
+	if (!dc_dmub_srv || !dc_dmub_srv->dmub)
+		return false;
+	return dmub_srv_get_diagnostic_data(dc_dmub_srv->dmub);
+}
+
+void dc_dmub_srv_log_diagnostic_data(struct dc_dmub_srv *dc_dmub_srv)
+{
+	uint32_t i;
+
+	if (!dc_dmub_srv || !dc_dmub_srv->dmub) {
+		DC_LOG_ERROR("%s: invalid parameters.", __func__);
+		return;
+	}
+
+	DC_LOG_ERROR("%s: DMCUB error - collecting diagnostic data\n", __func__);
+
+	if (!dc_dmub_srv_get_diagnostic_data(dc_dmub_srv)) {
+		DC_LOG_ERROR("%s: dc_dmub_srv_get_diagnostic_data failed.", __func__);
+		return;
+	}
+
+	DC_LOG_DEBUG("DMCUB STATE:");
+	DC_LOG_DEBUG("    dmcub_version      : %08x", dc_dmub_srv->dmub->debug.dmcub_version);
+	DC_LOG_DEBUG("    scratch  [0]       : %08x", dc_dmub_srv->dmub->debug.scratch[0]);
+	DC_LOG_DEBUG("    scratch  [1]       : %08x", dc_dmub_srv->dmub->debug.scratch[1]);
+	DC_LOG_DEBUG("    scratch  [2]       : %08x", dc_dmub_srv->dmub->debug.scratch[2]);
+	DC_LOG_DEBUG("    scratch  [3]       : %08x", dc_dmub_srv->dmub->debug.scratch[3]);
+	DC_LOG_DEBUG("    scratch  [4]       : %08x", dc_dmub_srv->dmub->debug.scratch[4]);
+	DC_LOG_DEBUG("    scratch  [5]       : %08x", dc_dmub_srv->dmub->debug.scratch[5]);
+	DC_LOG_DEBUG("    scratch  [6]       : %08x", dc_dmub_srv->dmub->debug.scratch[6]);
+	DC_LOG_DEBUG("    scratch  [7]       : %08x", dc_dmub_srv->dmub->debug.scratch[7]);
+	DC_LOG_DEBUG("    scratch  [8]       : %08x", dc_dmub_srv->dmub->debug.scratch[8]);
+	DC_LOG_DEBUG("    scratch  [9]       : %08x", dc_dmub_srv->dmub->debug.scratch[9]);
+	DC_LOG_DEBUG("    scratch [10]       : %08x", dc_dmub_srv->dmub->debug.scratch[10]);
+	DC_LOG_DEBUG("    scratch [11]       : %08x", dc_dmub_srv->dmub->debug.scratch[11]);
+	DC_LOG_DEBUG("    scratch [12]       : %08x", dc_dmub_srv->dmub->debug.scratch[12]);
+	DC_LOG_DEBUG("    scratch [13]       : %08x", dc_dmub_srv->dmub->debug.scratch[13]);
+	DC_LOG_DEBUG("    scratch [14]       : %08x", dc_dmub_srv->dmub->debug.scratch[14]);
+	DC_LOG_DEBUG("    scratch [15]       : %08x", dc_dmub_srv->dmub->debug.scratch[15]);
+	for (i = 0; i < DMUB_PC_SNAPSHOT_COUNT; i++)
+		DC_LOG_DEBUG("    pc[%d]             : %08x", i, dc_dmub_srv->dmub->debug.pc[i]);
+	DC_LOG_DEBUG("    unk_fault_addr     : %08x", dc_dmub_srv->dmub->debug.undefined_address_fault_addr);
+	DC_LOG_DEBUG("    inst_fault_addr    : %08x", dc_dmub_srv->dmub->debug.inst_fetch_fault_addr);
+	DC_LOG_DEBUG("    data_fault_addr    : %08x", dc_dmub_srv->dmub->debug.data_write_fault_addr);
+	DC_LOG_DEBUG("    inbox1_rptr        : %08x", dc_dmub_srv->dmub->debug.inbox1_rptr);
+	DC_LOG_DEBUG("    inbox1_wptr        : %08x", dc_dmub_srv->dmub->debug.inbox1_wptr);
+	DC_LOG_DEBUG("    inbox1_size        : %08x", dc_dmub_srv->dmub->debug.inbox1_size);
+	DC_LOG_DEBUG("    inbox0_rptr        : %08x", dc_dmub_srv->dmub->debug.inbox0_rptr);
+	DC_LOG_DEBUG("    inbox0_wptr        : %08x", dc_dmub_srv->dmub->debug.inbox0_wptr);
+	DC_LOG_DEBUG("    inbox0_size        : %08x", dc_dmub_srv->dmub->debug.inbox0_size);
+	DC_LOG_DEBUG("    outbox1_rptr       : %08x", dc_dmub_srv->dmub->debug.outbox1_rptr);
+	DC_LOG_DEBUG("    outbox1_wptr       : %08x", dc_dmub_srv->dmub->debug.outbox1_wptr);
+	DC_LOG_DEBUG("    outbox1_size       : %08x", dc_dmub_srv->dmub->debug.outbox1_size);
+	DC_LOG_DEBUG("    is_enabled         : %d", dc_dmub_srv->dmub->debug.is_dmcub_enabled);
+	DC_LOG_DEBUG("    is_soft_reset      : %d", dc_dmub_srv->dmub->debug.is_dmcub_soft_reset);
+	DC_LOG_DEBUG("    is_secure_reset    : %d", dc_dmub_srv->dmub->debug.is_dmcub_secure_reset);
+	DC_LOG_DEBUG("    is_traceport_en    : %d", dc_dmub_srv->dmub->debug.is_traceport_en);
+	DC_LOG_DEBUG("    is_cw0_en          : %d", dc_dmub_srv->dmub->debug.is_cw0_enabled);
+	DC_LOG_DEBUG("    is_cw6_en          : %d", dc_dmub_srv->dmub->debug.is_cw6_enabled);
+}
+
+static bool dc_dmub_should_update_cursor_data(struct pipe_ctx *pipe_ctx)
+{
+	if (pipe_ctx->plane_state != NULL) {
+		if (pipe_ctx->plane_state->address.type == PLN_ADDR_TYPE_VIDEO_PROGRESSIVE ||
+			resource_can_pipe_disable_cursor(pipe_ctx))
+			return false;
+	}
+
+	if ((pipe_ctx->stream->link->psr_settings.psr_version == DC_PSR_VERSION_SU_1 ||
+		pipe_ctx->stream->link->psr_settings.psr_version == DC_PSR_VERSION_1) &&
+		pipe_ctx->stream->ctx->dce_version >= DCN_VERSION_3_1)
+		return true;
+
+	if (pipe_ctx->stream->link->replay_settings.config.replay_supported)
+		return true;
+
+	return false;
+}
+
+static void dc_build_cursor_update_payload0(
+		struct pipe_ctx *pipe_ctx, uint8_t p_idx,
+		struct dmub_cmd_update_cursor_payload0 *payload)
+{
+	struct hubp *hubp = pipe_ctx->plane_res.hubp;
+	unsigned int panel_inst = 0;
+
+	if (!dc_get_edp_link_panel_inst(hubp->ctx->dc,
+		pipe_ctx->stream->link, &panel_inst))
+		return;
+
+	/* Payload: Cursor Rect is built from position & attribute
+	 * x & y are obtained from postion
+	 */
+	payload->cursor_rect.x = hubp->cur_rect.x;
+	payload->cursor_rect.y = hubp->cur_rect.y;
+	/* w & h are obtained from attribute */
+	payload->cursor_rect.width  = hubp->cur_rect.w;
+	payload->cursor_rect.height = hubp->cur_rect.h;
+
+	payload->enable      = hubp->pos.cur_ctl.bits.cur_enable;
+	payload->pipe_idx    = p_idx;
+	payload->cmd_version = DMUB_CMD_PSR_CONTROL_VERSION_1;
+	payload->panel_inst  = panel_inst;
+}
+
+static void dc_build_cursor_position_update_payload0(
+		struct dmub_cmd_update_cursor_payload0 *pl, const uint8_t p_idx,
+		const struct hubp *hubp, const struct dpp *dpp)
+{
+	/* Hubp */
+	pl->position_cfg.pHubp.cur_ctl.raw  = hubp->pos.cur_ctl.raw;
+	pl->position_cfg.pHubp.position.raw = hubp->pos.position.raw;
+	pl->position_cfg.pHubp.hot_spot.raw = hubp->pos.hot_spot.raw;
+	pl->position_cfg.pHubp.dst_offset.raw = hubp->pos.dst_offset.raw;
+
+	/* dpp */
+	pl->position_cfg.pDpp.cur0_ctl.raw = dpp->pos.cur0_ctl.raw;
+	pl->position_cfg.pipe_idx = p_idx;
+}
+
+static void dc_build_cursor_attribute_update_payload1(
+		struct dmub_cursor_attributes_cfg *pl_A, const uint8_t p_idx,
+		const struct hubp *hubp, const struct dpp *dpp)
+{
+	/* Hubp */
+	pl_A->aHubp.SURFACE_ADDR_HIGH = hubp->att.SURFACE_ADDR_HIGH;
+	pl_A->aHubp.SURFACE_ADDR = hubp->att.SURFACE_ADDR;
+	pl_A->aHubp.cur_ctl.raw  = hubp->att.cur_ctl.raw;
+	pl_A->aHubp.size.raw     = hubp->att.size.raw;
+	pl_A->aHubp.settings.raw = hubp->att.settings.raw;
+
+	/* dpp */
+	pl_A->aDpp.cur0_ctl.raw = dpp->att.cur0_ctl.raw;
+}
+
+/**
+ * dc_send_update_cursor_info_to_dmu - Populate the DMCUB Cursor update info command
+ *
+ * @pCtx: [in] pipe context
+ * @pipe_idx: [in] pipe index
+ *
+ * This function would store the cursor related information and pass it into
+ * dmub
+ */
+void dc_send_update_cursor_info_to_dmu(
+		struct pipe_ctx *pCtx, uint8_t pipe_idx)
+{
+	union dmub_rb_cmd cmd[2];
+	union dmub_cmd_update_cursor_info_data *update_cursor_info_0 =
+					&cmd[0].update_cursor_info.update_cursor_info_data;
+
+	memset(cmd, 0, sizeof(cmd));
+
+	if (!dc_dmub_should_update_cursor_data(pCtx))
+		return;
+	/*
+	 * Since we use multi_cmd_pending for dmub command, the 2nd command is
+	 * only assigned to store cursor attributes info.
+	 * 1st command can view as 2 parts, 1st is for PSR/Replay data, the other
+	 * is to store cursor position info.
+	 *
+	 * Command heaer type must be the same type if using  multi_cmd_pending.
+	 * Besides, while process 2nd command in DMU, the sub type is useless.
+	 * So it's meanless to pass the sub type header with different type.
+	 */
+
+	{
+		/* Build Payload#0 Header */
+		cmd[0].update_cursor_info.header.type = DMUB_CMD__UPDATE_CURSOR_INFO;
+		cmd[0].update_cursor_info.header.payload_bytes =
+				sizeof(cmd[0].update_cursor_info.update_cursor_info_data);
+		cmd[0].update_cursor_info.header.multi_cmd_pending = 1; //To combine multi dmu cmd, 1st cmd
+
+		/* Prepare Payload */
+		dc_build_cursor_update_payload0(pCtx, pipe_idx, &update_cursor_info_0->payload0);
+
+		dc_build_cursor_position_update_payload0(&update_cursor_info_0->payload0, pipe_idx,
+				pCtx->plane_res.hubp, pCtx->plane_res.dpp);
+		}
+	{
+		/* Build Payload#1 Header */
+		cmd[1].update_cursor_info.header.type = DMUB_CMD__UPDATE_CURSOR_INFO;
+		cmd[1].update_cursor_info.header.payload_bytes = sizeof(struct cursor_attributes_cfg);
+		cmd[1].update_cursor_info.header.multi_cmd_pending = 0; //Indicate it's the last command.
+
+		dc_build_cursor_attribute_update_payload1(
+				&cmd[1].update_cursor_info.update_cursor_info_data.payload1.attribute_cfg,
+				pipe_idx, pCtx->plane_res.hubp, pCtx->plane_res.dpp);
+
+		/* Combine 2nd cmds update_curosr_info to DMU */
+		dc_wake_and_execute_dmub_cmd_list(pCtx->stream->ctx, 2, cmd, DM_DMUB_WAIT_TYPE_WAIT);
+	}
+}
+
+bool dc_dmub_check_min_version(struct dmub_srv *srv)
+{
+	if (!srv->hw_funcs.is_psrsu_supported)
+		return true;
+	return srv->hw_funcs.is_psrsu_supported(srv);
+}
+
+void dc_dmub_srv_enable_dpia_trace(const struct dc *dc)
+{
+	struct dc_dmub_srv *dc_dmub_srv = dc->ctx->dmub_srv;
+
+	if (!dc_dmub_srv || !dc_dmub_srv->dmub) {
+		DC_LOG_ERROR("%s: invalid parameters.", __func__);
+		return;
+	}
+
+	if (!dc_wake_and_execute_gpint(dc->ctx, DMUB_GPINT__SET_TRACE_BUFFER_MASK_WORD1,
+				       0x0010, NULL, DM_DMUB_WAIT_TYPE_WAIT)) {
+		DC_LOG_ERROR("timeout updating trace buffer mask word\n");
+		return;
+	}
+
+	if (!dc_wake_and_execute_gpint(dc->ctx, DMUB_GPINT__UPDATE_TRACE_BUFFER_MASK,
+				       0x0000, NULL, DM_DMUB_WAIT_TYPE_WAIT)) {
+		DC_LOG_ERROR("timeout updating trace buffer mask word\n");
+		return;
+	}
+
+	DC_LOG_DEBUG("Enabled DPIA trace\n");
+}
+
+void dc_dmub_srv_subvp_save_surf_addr(const struct dc_dmub_srv *dc_dmub_srv, const struct dc_plane_address *addr, uint8_t subvp_index)
+{
+	dmub_srv_subvp_save_surf_addr(dc_dmub_srv->dmub, addr, subvp_index);
+}
+
+bool dc_dmub_srv_is_hw_pwr_up(struct dc_dmub_srv *dc_dmub_srv, bool wait)
+{
+	struct dc_context *dc_ctx;
+	enum dmub_status status;
+
+	if (!dc_dmub_srv || !dc_dmub_srv->dmub)
+		return true;
+
+	if (dc_dmub_srv->ctx->dc->debug.dmcub_emulation)
+		return true;
+
+	dc_ctx = dc_dmub_srv->ctx;
+
+	if (wait) {
+		if (dc_dmub_srv->ctx->dc->debug.disable_timeout) {
+			do {
+				status = dmub_srv_wait_for_hw_pwr_up(dc_dmub_srv->dmub, 500000);
+			} while (status != DMUB_STATUS_OK);
+		} else {
+			status = dmub_srv_wait_for_hw_pwr_up(dc_dmub_srv->dmub, 500000);
+			if (status != DMUB_STATUS_OK) {
+				DC_ERROR("Error querying DMUB hw power up status: error=%d\n", status);
+				return false;
+			}
+		}
+	} else
+		return dmub_srv_is_hw_pwr_up(dc_dmub_srv->dmub);
+
+	return true;
+}
+
+static int count_active_streams(const struct dc *dc)
+{
+	int i, count = 0;
+
+	for (i = 0; i < dc->current_state->stream_count; ++i) {
+		struct dc_stream_state *stream = dc->current_state->streams[i];
+
+		if (stream && (!stream->dpms_off || dc->config.disable_ips_in_dpms_off))
+			count += 1;
+	}
+
+	return count;
+}
+
+static void dc_dmub_srv_notify_idle(const struct dc *dc, bool allow_idle)
+{
+	volatile const struct dmub_shared_state_ips_fw *ips_fw;
+	struct dc_dmub_srv *dc_dmub_srv;
+	union dmub_rb_cmd cmd = {0};
+
+	if (dc->debug.dmcub_emulation)
+		return;
+
+	if (!dc->ctx->dmub_srv || !dc->ctx->dmub_srv->dmub)
+		return;
+
+	dc_dmub_srv = dc->ctx->dmub_srv;
+	ips_fw = &dc_dmub_srv->dmub->shared_state[DMUB_SHARED_SHARE_FEATURE__IPS_FW].data.ips_fw;
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.idle_opt_notify_idle.header.type = DMUB_CMD__IDLE_OPT;
+	cmd.idle_opt_notify_idle.header.sub_type = DMUB_CMD__IDLE_OPT_DCN_NOTIFY_IDLE;
+	cmd.idle_opt_notify_idle.header.payload_bytes =
+		sizeof(cmd.idle_opt_notify_idle) -
+		sizeof(cmd.idle_opt_notify_idle.header);
+
+	cmd.idle_opt_notify_idle.cntl_data.driver_idle = allow_idle;
+
+	if (dc->work_arounds.skip_psr_ips_crtc_disable)
+		cmd.idle_opt_notify_idle.cntl_data.skip_otg_disable = true;
+
+	if (allow_idle) {
+		volatile struct dmub_shared_state_ips_driver *ips_driver =
+			&dc_dmub_srv->dmub->shared_state[DMUB_SHARED_SHARE_FEATURE__IPS_DRIVER].data.ips_driver;
+		union dmub_shared_state_ips_driver_signals new_signals;
+
+		DC_LOG_IPS(
+			"%s wait idle (ips1_commit=%u ips2_commit=%u)",
+			__func__,
+			ips_fw->signals.bits.ips1_commit,
+			ips_fw->signals.bits.ips2_commit);
+
+		dc_dmub_srv_wait_for_idle(dc->ctx->dmub_srv, DM_DMUB_WAIT_TYPE_WAIT, NULL);
+
+		memset(&new_signals, 0, sizeof(new_signals));
+
+		new_signals.bits.allow_idle = 1; /* always set */
+
+		if (dc->config.disable_ips == DMUB_IPS_ENABLE ||
+		    dc->config.disable_ips == DMUB_IPS_DISABLE_DYNAMIC) {
+			new_signals.bits.allow_pg = 1;
+			new_signals.bits.allow_ips1 = 1;
+			new_signals.bits.allow_ips2 = 1;
+			new_signals.bits.allow_z10 = 1;
+			// New in IPSv2.0
+			new_signals.bits.allow_ips1z8 = 1;
+		} else if (dc->config.disable_ips == DMUB_IPS_DISABLE_IPS1) {
+			new_signals.bits.allow_ips1 = 1;
+		} else if (dc->config.disable_ips == DMUB_IPS_DISABLE_IPS2) {
+			// IPSv1.0 only
+			new_signals.bits.allow_pg = 1;
+			new_signals.bits.allow_ips1 = 1;
+		} else if (dc->config.disable_ips == DMUB_IPS_DISABLE_IPS2_Z10) {
+			// IPSv1.0 only
+			new_signals.bits.allow_pg = 1;
+			new_signals.bits.allow_ips1 = 1;
+			new_signals.bits.allow_ips2 = 1;
+		} else if (dc->config.disable_ips == DMUB_IPS_RCG_IN_ACTIVE_IPS2_IN_OFF) {
+			/* TODO: Move this logic out to hwseq */
+			if (count_active_streams(dc) == 0) {
+				/* IPS2 - Display off */
+				new_signals.bits.allow_pg = 1;
+				new_signals.bits.allow_ips1 = 1;
+				new_signals.bits.allow_ips2 = 1;
+				new_signals.bits.allow_z10 = 1;
+				// New in IPSv2.0
+				new_signals.bits.allow_ips1z8 = 1;
+			} else {
+				/* RCG only */
+				new_signals.bits.allow_pg = 0;
+				new_signals.bits.allow_ips1 = 1;
+				new_signals.bits.allow_ips2 = 0;
+				new_signals.bits.allow_z10 = 0;
+			}
+		} else if (dc->config.disable_ips == DMUB_IPS_DISABLE_Z8_RETENTION) {
+			new_signals.bits.allow_pg = 1;
+			new_signals.bits.allow_ips1 = 1;
+			new_signals.bits.allow_ips2 = 1;
+			new_signals.bits.allow_z10 = 1;
+		}
+		// Setting RCG allow bits (IPSv2.0)
+		if (dc->config.disable_ips_rcg == DMUB_IPS_RCG_ENABLE) {
+			new_signals.bits.allow_ips0_rcg = 1;
+			new_signals.bits.allow_ips1_rcg = 1;
+		} else if (dc->config.disable_ips_rcg == DMUB_IPS0_RCG_DISABLE) {
+			new_signals.bits.allow_ips1_rcg = 1;
+		} else if (dc->config.disable_ips_rcg == DMUB_IPS1_RCG_DISABLE) {
+			new_signals.bits.allow_ips0_rcg = 1;
+		}
+		// IPS dynamic allow bits (IPSv2 change, vpb use case)
+		if (dc->config.disable_ips_in_vpb == DMUB_IPS_VPB_ENABLE_IPS1_AND_RCG) {
+			new_signals.bits.allow_dynamic_ips1 = 1;
+		} else if (dc->config.disable_ips_in_vpb == DMUB_IPS_VPB_ENABLE_ALL) {
+			new_signals.bits.allow_dynamic_ips1 = 1;
+			new_signals.bits.allow_dynamic_ips1_z8 = 1;
+		}
+		ips_driver->signals = new_signals;
+		dc_dmub_srv->driver_signals = ips_driver->signals;
+	}
+
+	DC_LOG_IPS(
+		"%s send allow_idle=%d (ips1_commit=%u ips2_commit=%u)",
+		__func__,
+		allow_idle,
+		ips_fw->signals.bits.ips1_commit,
+		ips_fw->signals.bits.ips2_commit);
+
+	/* NOTE: This does not use the "wake" interface since this is part of the wake path. */
+	/* We also do not perform a wait since DMCUB could enter idle after the notification. */
+	dm_execute_dmub_cmd(dc->ctx, &cmd, allow_idle ? DM_DMUB_WAIT_TYPE_NO_WAIT : DM_DMUB_WAIT_TYPE_WAIT);
+
+	/* Register access should stop at this point. */
+	if (allow_idle)
+		dc_dmub_srv->needs_idle_wake = true;
+}
+
+static void dc_dmub_srv_exit_low_power_state(const struct dc *dc)
+{
+	struct dc_dmub_srv *dc_dmub_srv;
+	uint32_t rcg_exit_count = 0, ips1_exit_count = 0, ips2_exit_count = 0, ips1z8_exit_count = 0;
+
+	if (dc->debug.dmcub_emulation)
+		return;
+
+	if (!dc->ctx->dmub_srv || !dc->ctx->dmub_srv->dmub)
+		return;
+
+	dc_dmub_srv = dc->ctx->dmub_srv;
+
+	if (dc->clk_mgr->funcs->exit_low_power_state) {
+		volatile const struct dmub_shared_state_ips_fw *ips_fw =
+			&dc_dmub_srv->dmub->shared_state[DMUB_SHARED_SHARE_FEATURE__IPS_FW].data.ips_fw;
+		volatile struct dmub_shared_state_ips_driver *ips_driver =
+			&dc_dmub_srv->dmub->shared_state[DMUB_SHARED_SHARE_FEATURE__IPS_DRIVER].data.ips_driver;
+		union dmub_shared_state_ips_driver_signals prev_driver_signals = ips_driver->signals;
+
+		rcg_exit_count = ips_fw->rcg_exit_count;
+		ips1_exit_count = ips_fw->ips1_exit_count;
+		ips2_exit_count = ips_fw->ips2_exit_count;
+		ips1z8_exit_count = ips_fw->ips1_z8ret_exit_count;
+
+		ips_driver->signals.all = 0;
+		dc_dmub_srv->driver_signals = ips_driver->signals;
+
+		DC_LOG_IPS(
+			"%s (allow ips1=%u ips2=%u) (commit ips1=%u ips2=%u ips1z8=%u) (count rcg=%u ips1=%u ips2=%u ips1_z8=%u)",
+			__func__,
+			ips_driver->signals.bits.allow_ips1,
+			ips_driver->signals.bits.allow_ips2,
+			ips_fw->signals.bits.ips1_commit,
+			ips_fw->signals.bits.ips2_commit,
+			ips_fw->signals.bits.ips1z8_commit,
+			ips_fw->rcg_entry_count,
+			ips_fw->ips1_entry_count,
+			ips_fw->ips2_entry_count,
+			ips_fw->ips1_z8ret_entry_count);
+
+		/* Note: register access has technically not resumed for DCN here, but we
+		 * need to be message PMFW through our standard register interface.
+		 */
+		dc_dmub_srv->needs_idle_wake = false;
+
+		if (!dc->caps.ips_v2_support && ((prev_driver_signals.bits.allow_ips2 || prev_driver_signals.all == 0) &&
+		    (!dc->debug.optimize_ips_handshake ||
+		     ips_fw->signals.bits.ips2_commit || !ips_fw->signals.bits.in_idle))) {
+			DC_LOG_IPS(
+				"wait IPS2 eval (ips1_commit=%u ips2_commit=%u )",
+				ips_fw->signals.bits.ips1_commit,
+				ips_fw->signals.bits.ips2_commit);
+
+			if (!dc->debug.optimize_ips_handshake || !ips_fw->signals.bits.ips2_commit)
+				udelay(dc->debug.ips2_eval_delay_us);
+
+			DC_LOG_IPS(
+				"exit IPS2 #1 (ips1_commit=%u ips2_commit=%u)",
+				ips_fw->signals.bits.ips1_commit,
+				ips_fw->signals.bits.ips2_commit);
+
+			// Tell PMFW to exit low power state
+			dc->clk_mgr->funcs->exit_low_power_state(dc->clk_mgr);
+
+			if (ips_fw->signals.bits.ips2_commit) {
+
+				DC_LOG_IPS(
+					"wait IPS2 entry delay (ips1_commit=%u ips2_commit=%u)",
+					ips_fw->signals.bits.ips1_commit,
+					ips_fw->signals.bits.ips2_commit);
+
+				// Wait for IPS2 entry upper bound
+				udelay(dc->debug.ips2_entry_delay_us);
+
+				DC_LOG_IPS(
+					"exit IPS2 #2 (ips1_commit=%u ips2_commit=%u)",
+					ips_fw->signals.bits.ips1_commit,
+					ips_fw->signals.bits.ips2_commit);
+
+				dc->clk_mgr->funcs->exit_low_power_state(dc->clk_mgr);
+
+				DC_LOG_IPS(
+					"wait IPS2 commit clear (ips1_commit=%u ips2_commit=%u)",
+					ips_fw->signals.bits.ips1_commit,
+					ips_fw->signals.bits.ips2_commit);
+
+				while (ips_fw->signals.bits.ips2_commit)
+					udelay(1);
+
+				DC_LOG_IPS(
+					"wait hw_pwr_up (ips1_commit=%u ips2_commit=%u)",
+					ips_fw->signals.bits.ips1_commit,
+					ips_fw->signals.bits.ips2_commit);
+
+				if (!dc_dmub_srv_is_hw_pwr_up(dc->ctx->dmub_srv, true))
+					ASSERT(0);
+
+				DC_LOG_IPS(
+					"resync inbox1 (ips1_commit=%u ips2_commit=%u)",
+					ips_fw->signals.bits.ips1_commit,
+					ips_fw->signals.bits.ips2_commit);
+
+				dmub_srv_sync_inboxes(dc->ctx->dmub_srv->dmub);
+			}
+		}
+
+		dc_dmub_srv_notify_idle(dc, false);
+		if (prev_driver_signals.bits.allow_ips1 || prev_driver_signals.all == 0) {
+			DC_LOG_IPS(
+				"wait for IPS1 commit clear (ips1_commit=%u ips2_commit=%u ips1z8=%u)",
+				ips_fw->signals.bits.ips1_commit,
+				ips_fw->signals.bits.ips2_commit,
+				ips_fw->signals.bits.ips1z8_commit);
+
+			while (ips_fw->signals.bits.ips1_commit)
+				udelay(1);
+
+			DC_LOG_IPS(
+				"wait for IPS1 commit clear done (ips1_commit=%u ips2_commit=%u ips1z8=%u)",
+				ips_fw->signals.bits.ips1_commit,
+				ips_fw->signals.bits.ips2_commit,
+				ips_fw->signals.bits.ips1z8_commit);
+		}
+	}
+
+	if (!dc_dmub_srv_is_hw_pwr_up(dc->ctx->dmub_srv, true))
+		ASSERT(0);
+
+	DC_LOG_IPS("%s exit (count rcg=%u ips1=%u ips2=%u ips1z8=%u)",
+		__func__,
+		rcg_exit_count,
+		ips1_exit_count,
+		ips2_exit_count,
+		ips1z8_exit_count);
+}
+
+void dc_dmub_srv_set_power_state(struct dc_dmub_srv *dc_dmub_srv, enum dc_acpi_cm_power_state power_state)
+{
+	struct dmub_srv *dmub;
+
+	if (!dc_dmub_srv)
+		return;
+
+	dmub = dc_dmub_srv->dmub;
+
+	if (power_state == DC_ACPI_CM_POWER_STATE_D0)
+		dmub_srv_set_power_state(dmub, DMUB_POWER_STATE_D0);
+	else
+		dmub_srv_set_power_state(dmub, DMUB_POWER_STATE_D3);
+}
+
+void dc_dmub_srv_notify_fw_dc_power_state(struct dc_dmub_srv *dc_dmub_srv,
+					  enum dc_acpi_cm_power_state power_state)
+{
+	union dmub_rb_cmd cmd;
+
+	if (!dc_dmub_srv)
+		return;
+
+	memset(&cmd, 0, sizeof(cmd));
+
+	cmd.idle_opt_set_dc_power_state.header.type = DMUB_CMD__IDLE_OPT;
+	cmd.idle_opt_set_dc_power_state.header.sub_type = DMUB_CMD__IDLE_OPT_SET_DC_POWER_STATE;
+	cmd.idle_opt_set_dc_power_state.header.payload_bytes =
+		sizeof(cmd.idle_opt_set_dc_power_state) - sizeof(cmd.idle_opt_set_dc_power_state.header);
+
+	if (power_state == DC_ACPI_CM_POWER_STATE_D0) {
+		cmd.idle_opt_set_dc_power_state.data.power_state = DMUB_IDLE_OPT_DC_POWER_STATE_D0;
+	} else if (power_state == DC_ACPI_CM_POWER_STATE_D3) {
+		cmd.idle_opt_set_dc_power_state.data.power_state = DMUB_IDLE_OPT_DC_POWER_STATE_D3;
+	} else {
+		cmd.idle_opt_set_dc_power_state.data.power_state = DMUB_IDLE_OPT_DC_POWER_STATE_UNKNOWN;
+	}
+
+	dc_wake_and_execute_dmub_cmd(dc_dmub_srv->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+bool dc_dmub_srv_should_detect(struct dc_dmub_srv *dc_dmub_srv)
+{
+	volatile const struct dmub_shared_state_ips_fw *ips_fw;
+	bool reallow_idle = false, should_detect = false;
+
+	if (!dc_dmub_srv || !dc_dmub_srv->dmub)
+		return false;
+
+	if (dc_dmub_srv->dmub->shared_state &&
+	    dc_dmub_srv->dmub->meta_info.feature_bits.bits.shared_state_link_detection) {
+		ips_fw = &dc_dmub_srv->dmub->shared_state[DMUB_SHARED_SHARE_FEATURE__IPS_FW].data.ips_fw;
+		return ips_fw->signals.bits.detection_required;
+	}
+
+	/* Detection may require reading scratch 0 - exit out of idle prior to the read. */
+	if (dc_dmub_srv->idle_allowed) {
+		dc_dmub_srv_apply_idle_power_optimizations(dc_dmub_srv->ctx->dc, false);
+		reallow_idle = true;
+	}
+
+	should_detect = dmub_srv_should_detect(dc_dmub_srv->dmub);
+
+	/* Re-enter idle if we're not about to immediately redetect links. */
+	if (!should_detect && reallow_idle && dc_dmub_srv->idle_exit_counter == 0 &&
+	    !dc_dmub_srv->ctx->dc->debug.disable_dmub_reallow_idle)
+		dc_dmub_srv_apply_idle_power_optimizations(dc_dmub_srv->ctx->dc, true);
+
+	return should_detect;
+}
+
+void dc_dmub_srv_apply_idle_power_optimizations(const struct dc *dc, bool allow_idle)
+{
+	struct dc_dmub_srv *dc_dmub_srv = dc->ctx->dmub_srv;
+
+	if (!dc_dmub_srv || !dc_dmub_srv->dmub)
+		return;
+
+	allow_idle &= (!dc->debug.ips_disallow_entry);
+
+	if (dc_dmub_srv->idle_allowed == allow_idle)
+		return;
+
+	DC_LOG_IPS("%s state change: old=%d new=%d", __func__, dc_dmub_srv->idle_allowed, allow_idle);
+
+	/*
+	 * Entering a low power state requires a driver notification.
+	 * Powering up the hardware requires notifying PMFW and DMCUB.
+	 * Clearing the driver idle allow requires a DMCUB command.
+	 * DMCUB commands requires the DMCUB to be powered up and restored.
+	 */
+
+	if (!allow_idle) {
+		dc_dmub_srv->idle_exit_counter += 1;
+
+		dc_dmub_srv_exit_low_power_state(dc);
+		/*
+		 * Idle is considered fully exited only after the sequence above
+		 * fully completes. If we have a race of two threads exiting
+		 * at the same time then it's safe to perform the sequence
+		 * twice as long as we're not re-entering.
+		 *
+		 * Infinite command submission is avoided by using the
+		 * dm_execute_dmub_cmd submission instead of the "wake" helpers.
+		 */
+		dc_dmub_srv->idle_allowed = false;
+
+		dc_dmub_srv->idle_exit_counter -= 1;
+		if (dc_dmub_srv->idle_exit_counter < 0) {
+			ASSERT(0);
+			dc_dmub_srv->idle_exit_counter = 0;
+		}
+	} else {
+		/* Consider idle as notified prior to the actual submission to
+		 * prevent multiple entries. */
+		dc_dmub_srv->idle_allowed = true;
+
+		dc_dmub_srv_notify_idle(dc, allow_idle);
+	}
+}
+
+bool dc_wake_and_execute_dmub_cmd(const struct dc_context *ctx, union dmub_rb_cmd *cmd,
+				  enum dm_dmub_wait_type wait_type)
+{
+	return dc_wake_and_execute_dmub_cmd_list(ctx, 1, cmd, wait_type);
+}
+
+bool dc_wake_and_execute_dmub_cmd_list(const struct dc_context *ctx, unsigned int count,
+				       union dmub_rb_cmd *cmd, enum dm_dmub_wait_type wait_type)
+{
+	struct dc_dmub_srv *dc_dmub_srv = ctx->dmub_srv;
+	bool result = false, reallow_idle = false;
+
+	if (!dc_dmub_srv || !dc_dmub_srv->dmub)
+		return false;
+
+	if (count == 0)
+		return true;
+
+	if (dc_dmub_srv->idle_allowed) {
+		dc_dmub_srv_apply_idle_power_optimizations(ctx->dc, false);
+		reallow_idle = true;
+	}
+
+	/*
+	 * These may have different implementations in DM, so ensure
+	 * that we guide it to the expected helper.
+	 */
+	if (count > 1)
+		result = dm_execute_dmub_cmd_list(ctx, count, cmd, wait_type);
+	else
+		result = dm_execute_dmub_cmd(ctx, cmd, wait_type);
+
+	if (result && reallow_idle && dc_dmub_srv->idle_exit_counter == 0 &&
+	    !ctx->dc->debug.disable_dmub_reallow_idle)
+		dc_dmub_srv_apply_idle_power_optimizations(ctx->dc, true);
+
+	return result;
+}
+
+static bool dc_dmub_execute_gpint(const struct dc_context *ctx, enum dmub_gpint_command command_code,
+				  uint16_t param, uint32_t *response, enum dm_dmub_wait_type wait_type)
+{
+	struct dc_dmub_srv *dc_dmub_srv = ctx->dmub_srv;
+	const uint32_t wait_us = wait_type == DM_DMUB_WAIT_TYPE_NO_WAIT ? 0 : 30;
+	enum dmub_status status;
+
+	if (response)
+		*response = 0;
+
+	if (!dc_dmub_srv || !dc_dmub_srv->dmub)
+		return false;
+
+	status = dmub_srv_send_gpint_command(dc_dmub_srv->dmub, command_code, param, wait_us);
+	if (status != DMUB_STATUS_OK) {
+		if (status == DMUB_STATUS_TIMEOUT && wait_type == DM_DMUB_WAIT_TYPE_NO_WAIT)
+			return true;
+
+		return false;
+	}
+
+	if (response && wait_type == DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY)
+		dmub_srv_get_gpint_response(dc_dmub_srv->dmub, response);
+
+	return true;
+}
+
+bool dc_wake_and_execute_gpint(const struct dc_context *ctx, enum dmub_gpint_command command_code,
+			       uint16_t param, uint32_t *response, enum dm_dmub_wait_type wait_type)
+{
+	struct dc_dmub_srv *dc_dmub_srv = ctx->dmub_srv;
+	bool result = false, reallow_idle = false;
+
+	if (!dc_dmub_srv || !dc_dmub_srv->dmub)
+		return false;
+
+	if (dc_dmub_srv->idle_allowed) {
+		dc_dmub_srv_apply_idle_power_optimizations(ctx->dc, false);
+		reallow_idle = true;
+	}
+
+	result = dc_dmub_execute_gpint(ctx, command_code, param, response, wait_type);
+
+	if (result && reallow_idle && dc_dmub_srv->idle_exit_counter == 0 &&
+	    !ctx->dc->debug.disable_dmub_reallow_idle)
+		dc_dmub_srv_apply_idle_power_optimizations(ctx->dc, true);
+
+	return result;
+}
+
+static void dc_dmub_srv_rb_based_fams2_update_config(struct dc *dc,
+		struct dc_state *context,
+		bool enable)
+{
+	uint8_t num_cmds = 1;
+	uint32_t i;
+	union dmub_rb_cmd cmd[2 * MAX_STREAMS + 1];
+	struct dmub_rb_cmd_fams2 *global_cmd = &cmd[0].fams2_config;
+
+	memset(cmd, 0, sizeof(union dmub_rb_cmd) * (2 * MAX_STREAMS + 1));
+	/* fill in generic command header */
+	global_cmd->header.type = DMUB_CMD__FW_ASSISTED_MCLK_SWITCH;
+	global_cmd->header.sub_type = DMUB_CMD__FAMS2_CONFIG;
+	global_cmd->header.payload_bytes =
+			sizeof(struct dmub_rb_cmd_fams2) - sizeof(struct dmub_cmd_header);
+
+	if (enable) {
+		/* send global configuration parameters */
+		memcpy(&global_cmd->config.global, &context->bw_ctx.bw.dcn.fams2_global_config, sizeof(struct dmub_cmd_fams2_global_config));
+
+		/* copy static feature configuration overrides */
+		global_cmd->config.global.features.bits.enable_stall_recovery = dc->debug.fams2_config.bits.enable_stall_recovery;
+		global_cmd->config.global.features.bits.enable_debug = dc->debug.fams2_config.bits.enable_debug;
+		global_cmd->config.global.features.bits.enable_offload_flip = dc->debug.fams2_config.bits.enable_offload_flip;
+
+		/* construct per-stream configs */
+		for (i = 0; i < context->bw_ctx.bw.dcn.fams2_global_config.num_streams; i++) {
+			struct dmub_rb_cmd_fams2 *stream_base_cmd = &cmd[i+1].fams2_config;
+			struct dmub_rb_cmd_fams2 *stream_sub_state_cmd = &cmd[i+1+context->bw_ctx.bw.dcn.fams2_global_config.num_streams].fams2_config;
+
+			/* configure command header */
+			stream_base_cmd->header.type = DMUB_CMD__FW_ASSISTED_MCLK_SWITCH;
+			stream_base_cmd->header.sub_type = DMUB_CMD__FAMS2_CONFIG;
+			stream_base_cmd->header.payload_bytes =
+					sizeof(struct dmub_rb_cmd_fams2) - sizeof(struct dmub_cmd_header);
+			stream_base_cmd->header.multi_cmd_pending = 1;
+			stream_sub_state_cmd->header.type = DMUB_CMD__FW_ASSISTED_MCLK_SWITCH;
+			stream_sub_state_cmd->header.sub_type = DMUB_CMD__FAMS2_CONFIG;
+			stream_sub_state_cmd->header.payload_bytes =
+					sizeof(struct dmub_rb_cmd_fams2) - sizeof(struct dmub_cmd_header);
+			stream_sub_state_cmd->header.multi_cmd_pending = 1;
+			/* copy stream static base state */
+			memcpy(&stream_base_cmd->config,
+					&context->bw_ctx.bw.dcn.fams2_stream_base_params[i],
+					sizeof(union dmub_cmd_fams2_config));
+			/* copy stream static sub state */
+			memcpy(&stream_sub_state_cmd->config,
+					&context->bw_ctx.bw.dcn.fams2_stream_sub_params[i],
+					sizeof(union dmub_cmd_fams2_config));
+		}
+	}
+
+	/* apply feature configuration based on current driver state */
+	global_cmd->config.global.features.bits.enable_visual_confirm = dc->debug.visual_confirm == VISUAL_CONFIRM_FAMS2;
+	global_cmd->config.global.features.bits.enable = enable;
+
+	if (enable && context->bw_ctx.bw.dcn.fams2_global_config.features.bits.enable) {
+		/* set multi pending for global, and unset for last stream cmd */
+		global_cmd->header.multi_cmd_pending = 1;
+		cmd[2 * context->bw_ctx.bw.dcn.fams2_global_config.num_streams].fams2_config.header.multi_cmd_pending = 0;
+		num_cmds += 2 * context->bw_ctx.bw.dcn.fams2_global_config.num_streams;
+	}
+
+	dm_execute_dmub_cmd_list(dc->ctx, num_cmds, cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+static void dc_dmub_srv_ib_based_fams2_update_config(struct dc *dc,
+		struct dc_state *context,
+		bool enable)
+{
+	struct dmub_fams2_config_v2 *config = (struct dmub_fams2_config_v2 *)dc->ctx->dmub_srv->dmub->ib_mem_gart.cpu_addr;
+	union dmub_rb_cmd cmd;
+	uint32_t i;
+
+	memset(config, 0, sizeof(*config));
+	memset(&cmd, 0, sizeof(cmd));
+
+	cmd.ib_fams2_config.header.type = DMUB_CMD__FW_ASSISTED_MCLK_SWITCH;
+	cmd.ib_fams2_config.header.sub_type = DMUB_CMD__FAMS2_IB_CONFIG;
+
+	cmd.ib_fams2_config.ib_data.src.quad_part = dc->ctx->dmub_srv->dmub->ib_mem_gart.gpu_addr;
+	cmd.ib_fams2_config.ib_data.size = sizeof(*config);
+
+	if (enable && context->bw_ctx.bw.dcn.fams2_global_config.features.bits.enable) {
+		/* copy static feature configuration overrides */
+		config->global.features.bits.enable_stall_recovery = dc->debug.fams2_config.bits.enable_stall_recovery;
+		config->global.features.bits.enable_offload_flip = dc->debug.fams2_config.bits.enable_offload_flip;
+		config->global.features.bits.enable_debug = dc->debug.fams2_config.bits.enable_debug;
+
+		/* send global configuration parameters */
+		memcpy(&config->global, &context->bw_ctx.bw.dcn.fams2_global_config,
+			sizeof(struct dmub_cmd_fams2_global_config));
+
+		/* construct per-stream configs */
+		for (i = 0; i < context->bw_ctx.bw.dcn.fams2_global_config.num_streams; i++) {
+			/* copy stream static base state */
+			memcpy(&config->stream_v1[i].base,
+				&context->bw_ctx.bw.dcn.fams2_stream_base_params[i],
+				sizeof(config->stream_v1[i].base));
+
+			/* copy stream static sub-state */
+			memcpy(&config->stream_v1[i].sub_state,
+				&context->bw_ctx.bw.dcn.fams2_stream_sub_params_v2[i],
+				sizeof(config->stream_v1[i].sub_state));
+		}
+	}
+
+	config->global.features.bits.enable_visual_confirm = dc->debug.visual_confirm == VISUAL_CONFIRM_FAMS2;
+	config->global.features.bits.enable = enable;
+
+	dm_execute_dmub_cmd_list(dc->ctx, 1, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+void dc_dmub_srv_fams2_update_config(struct dc *dc,
+		struct dc_state *context,
+		bool enable)
+{
+	if (dc->debug.fams_version.major == 2)
+		dc_dmub_srv_rb_based_fams2_update_config(dc, context, enable);
+	if (dc->debug.fams_version.major == 3)
+		dc_dmub_srv_ib_based_fams2_update_config(dc, context, enable);
+}
+
+void dc_dmub_srv_fams2_drr_update(struct dc *dc,
+		uint32_t tg_inst,
+		uint32_t vtotal_min,
+		uint32_t vtotal_max,
+		uint32_t vtotal_mid,
+		uint32_t vtotal_mid_frame_num,
+		bool program_manual_trigger)
+{
+	union dmub_rb_cmd cmd = { 0 };
+
+	cmd.fams2_drr_update.header.type = DMUB_CMD__FW_ASSISTED_MCLK_SWITCH;
+	cmd.fams2_drr_update.header.sub_type = DMUB_CMD__FAMS2_DRR_UPDATE;
+	cmd.fams2_drr_update.dmub_optc_state_req.tg_inst = tg_inst;
+	cmd.fams2_drr_update.dmub_optc_state_req.v_total_max = vtotal_max;
+	cmd.fams2_drr_update.dmub_optc_state_req.v_total_min = vtotal_min;
+	cmd.fams2_drr_update.dmub_optc_state_req.v_total_mid = vtotal_mid;
+	cmd.fams2_drr_update.dmub_optc_state_req.v_total_mid_frame_num = vtotal_mid_frame_num;
+	cmd.fams2_drr_update.dmub_optc_state_req.program_manual_trigger = program_manual_trigger;
+
+	cmd.fams2_drr_update.header.payload_bytes =
+			sizeof(cmd.fams2_drr_update) - sizeof(cmd.fams2_drr_update.header);
+
+	dm_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+void dc_dmub_srv_fams2_passthrough_flip(
+		struct dc *dc,
+		struct dc_state *state,
+		struct dc_stream_state *stream,
+		struct dc_surface_update *srf_updates,
+		int surface_count)
+{
+	int plane_index;
+	union dmub_rb_cmd cmds[MAX_PLANES];
+	struct dc_plane_address *address;
+	struct dc_plane_state *plane_state;
+	int num_cmds = 0;
+	struct dc_stream_status *stream_status = dc_stream_get_status(stream);
+
+	if (surface_count <= 0 || stream_status == NULL)
+		return;
+
+	memset(cmds, 0, sizeof(union dmub_rb_cmd) * MAX_PLANES);
+
+	/* build command for each surface update */
+	for (plane_index = 0; plane_index < surface_count; plane_index++) {
+		plane_state = srf_updates[plane_index].surface;
+		address = &plane_state->address;
+
+		/* skip if there is no address update for plane */
+		if (!srf_updates[plane_index].flip_addr)
+			continue;
+
+		/* build command header */
+		cmds[num_cmds].fams2_flip.header.type = DMUB_CMD__FW_ASSISTED_MCLK_SWITCH;
+		cmds[num_cmds].fams2_flip.header.sub_type = DMUB_CMD__FAMS2_FLIP;
+		cmds[num_cmds].fams2_flip.header.payload_bytes =
+				sizeof(struct dmub_rb_cmd_fams2_flip) - sizeof(struct dmub_cmd_header);
+
+		/* for chaining multiple commands, all but last command should set to 1 */
+		cmds[num_cmds].fams2_flip.header.multi_cmd_pending = 1;
+
+		/* set topology info */
+		cmds[num_cmds].fams2_flip.flip_info.pipe_mask = dc_plane_get_pipe_mask(state, plane_state);
+		if (stream_status)
+			cmds[num_cmds].fams2_flip.flip_info.otg_inst = stream_status->primary_otg_inst;
+
+		cmds[num_cmds].fams2_flip.flip_info.config.bits.is_immediate = plane_state->flip_immediate;
+
+		/* build address info for command */
+		switch (address->type) {
+		case PLN_ADDR_TYPE_GRAPHICS:
+			if (address->grph.addr.quad_part == 0) {
+				BREAK_TO_DEBUGGER();
+				break;
+			}
+
+			cmds[num_cmds].fams2_flip.flip_info.addr_info.meta_addr_lo =
+					address->grph.meta_addr.low_part;
+			cmds[num_cmds].fams2_flip.flip_info.addr_info.meta_addr_hi =
+					(uint16_t)address->grph.meta_addr.high_part;
+			cmds[num_cmds].fams2_flip.flip_info.addr_info.surf_addr_lo =
+					address->grph.addr.low_part;
+			cmds[num_cmds].fams2_flip.flip_info.addr_info.surf_addr_hi =
+					(uint16_t)address->grph.addr.high_part;
+			break;
+		case PLN_ADDR_TYPE_VIDEO_PROGRESSIVE:
+			if (address->video_progressive.luma_addr.quad_part == 0 ||
+				address->video_progressive.chroma_addr.quad_part == 0) {
+				BREAK_TO_DEBUGGER();
+				break;
+			}
+
+			cmds[num_cmds].fams2_flip.flip_info.addr_info.meta_addr_lo =
+					address->video_progressive.luma_meta_addr.low_part;
+			cmds[num_cmds].fams2_flip.flip_info.addr_info.meta_addr_hi =
+					(uint16_t)address->video_progressive.luma_meta_addr.high_part;
+			cmds[num_cmds].fams2_flip.flip_info.addr_info.meta_addr_c_lo =
+					address->video_progressive.chroma_meta_addr.low_part;
+			cmds[num_cmds].fams2_flip.flip_info.addr_info.meta_addr_c_hi =
+					(uint16_t)address->video_progressive.chroma_meta_addr.high_part;
+			cmds[num_cmds].fams2_flip.flip_info.addr_info.surf_addr_lo =
+					address->video_progressive.luma_addr.low_part;
+			cmds[num_cmds].fams2_flip.flip_info.addr_info.surf_addr_hi =
+					(uint16_t)address->video_progressive.luma_addr.high_part;
+			cmds[num_cmds].fams2_flip.flip_info.addr_info.surf_addr_c_lo =
+					address->video_progressive.chroma_addr.low_part;
+			cmds[num_cmds].fams2_flip.flip_info.addr_info.surf_addr_c_hi =
+					(uint16_t)address->video_progressive.chroma_addr.high_part;
+			break;
+		default:
+			// Should never be hit
+			BREAK_TO_DEBUGGER();
+			break;
+		}
+
+		num_cmds++;
+	}
+
+	if (num_cmds > 0)  {
+		cmds[num_cmds - 1].fams2_flip.header.multi_cmd_pending = 0;
+		dm_execute_dmub_cmd_list(dc->ctx, num_cmds, cmds, DM_DMUB_WAIT_TYPE_WAIT);
+	}
+}
+
+
+bool dc_dmub_srv_ips_residency_cntl(const struct dc_context *ctx, uint8_t panel_inst, bool start_measurement)
+{
+	union dmub_rb_cmd cmd;
+
+	memset(&cmd, 0, sizeof(cmd));
+
+	cmd.ips_residency_cntl.header.type = DMUB_CMD__IPS;
+	cmd.ips_residency_cntl.header.sub_type = DMUB_CMD__IPS_RESIDENCY_CNTL;
+	cmd.ips_residency_cntl.header.payload_bytes = sizeof(struct dmub_cmd_ips_residency_cntl_data);
+
+	// only panel_inst=0 is supported at the moment
+	cmd.ips_residency_cntl.cntl_data.panel_inst = panel_inst;
+	cmd.ips_residency_cntl.cntl_data.start_measurement = start_measurement;
+
+	if (!dc_wake_and_execute_dmub_cmd(ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY))
+		return false;
+
+	return true;
+}
+
+bool dc_dmub_srv_ips_query_residency_info(const struct dc_context *ctx, uint8_t panel_inst, struct dmub_ips_residency_info *driver_info,
+					  enum ips_residency_mode ips_mode)
+{
+	union dmub_rb_cmd cmd;
+	uint32_t bytes = sizeof(struct dmub_ips_residency_info);
+
+	dmub_flush_buffer_mem(&ctx->dmub_srv->dmub->scratch_mem_fb);
+	memset(&cmd, 0, sizeof(cmd));
+
+	cmd.ips_query_residency_info.header.type = DMUB_CMD__IPS;
+	cmd.ips_query_residency_info.header.sub_type = DMUB_CMD__IPS_QUERY_RESIDENCY_INFO;
+	cmd.ips_query_residency_info.header.payload_bytes = sizeof(struct dmub_cmd_ips_query_residency_info_data);
+
+	cmd.ips_query_residency_info.info_data.dest.quad_part = ctx->dmub_srv->dmub->scratch_mem_fb.gpu_addr;
+	cmd.ips_query_residency_info.info_data.size = bytes;
+	cmd.ips_query_residency_info.info_data.panel_inst = panel_inst;
+	cmd.ips_query_residency_info.info_data.ips_mode = (uint32_t)ips_mode;
+
+	if (!dc_wake_and_execute_dmub_cmd(ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY) ||
+					  cmd.ips_query_residency_info.header.ret_status == 0)
+		return false;
+
+	// copy the result to the output since ret_status != 0 means the command returned data
+	memcpy(driver_info, ctx->dmub_srv->dmub->scratch_mem_fb.cpu_addr, bytes);
+
+	return true;
+}
+
+bool dmub_lsdma_init(struct dc_dmub_srv *dc_dmub_srv)
+{
+	struct dc_context *dc_ctx = dc_dmub_srv->ctx;
+	union dmub_rb_cmd cmd;
+	enum dm_dmub_wait_type wait_type;
+	struct dmub_cmd_lsdma_data *lsdma_data = &cmd.lsdma.lsdma_data;
+	bool result;
+
+	memset(&cmd, 0, sizeof(cmd));
+
+	cmd.cmd_common.header.type     = DMUB_CMD__LSDMA;
+	cmd.cmd_common.header.sub_type = DMUB_CMD__LSDMA_INIT_CONFIG;
+	wait_type                      = DM_DMUB_WAIT_TYPE_NO_WAIT;
+
+	lsdma_data->u.init_data.gpu_addr_base.quad_part = dc_ctx->dmub_srv->dmub->lsdma_rb_fb.gpu_addr;
+	lsdma_data->u.init_data.ring_size               = dc_ctx->dmub_srv->dmub->lsdma_rb_fb.size;
+
+	result = dc_wake_and_execute_dmub_cmd(dc_ctx, &cmd, wait_type);
+
+	if (!result)
+		DC_ERROR("LSDMA Init failed in DMUB");
+
+	return result;
+}
+
+bool dmub_lsdma_send_linear_copy_command(
+	struct dc_dmub_srv *dc_dmub_srv,
+	uint64_t src_addr,
+	uint64_t dst_addr,
+	uint32_t count
+)
+{
+	struct dc_context *dc_ctx = dc_dmub_srv->ctx;
+	union dmub_rb_cmd cmd;
+	enum dm_dmub_wait_type wait_type;
+	struct dmub_cmd_lsdma_data *lsdma_data = &cmd.lsdma.lsdma_data;
+	bool result;
+
+	memset(&cmd, 0, sizeof(cmd));
+
+	cmd.cmd_common.header.type     = DMUB_CMD__LSDMA;
+	cmd.cmd_common.header.sub_type = DMUB_CMD__LSDMA_LINEAR_COPY;
+	wait_type                      = DM_DMUB_WAIT_TYPE_NO_WAIT;
+
+	lsdma_data->u.linear_copy_data.count   = count - 1; // LSDMA controller expects bytes to copy -1
+	lsdma_data->u.linear_copy_data.src_lo  = src_addr & 0xFFFFFFFF;
+	lsdma_data->u.linear_copy_data.src_hi  = (src_addr >> 32) & 0xFFFFFFFF;
+	lsdma_data->u.linear_copy_data.dst_lo  = dst_addr & 0xFFFFFFFF;
+	lsdma_data->u.linear_copy_data.dst_hi  = (dst_addr >> 32) & 0xFFFFFFFF;
+
+	result = dc_wake_and_execute_dmub_cmd(dc_ctx, &cmd, wait_type);
+
+	if (!result)
+		DC_ERROR("LSDMA Linear Copy failed in DMUB");
+
+	return result;
+}
+
+bool dmub_lsdma_send_linear_sub_window_copy_command(
+	struct dc_dmub_srv *dc_dmub_srv,
+	struct lsdma_linear_sub_window_copy_params copy_data
+)
+{
+	struct dc_context *dc_ctx = dc_dmub_srv->ctx;
+	union dmub_rb_cmd cmd;
+	enum dm_dmub_wait_type wait_type;
+	struct dmub_cmd_lsdma_data *lsdma_data = &cmd.lsdma.lsdma_data;
+	bool result;
+
+	memset(&cmd, 0, sizeof(cmd));
+
+	cmd.cmd_common.header.type     = DMUB_CMD__LSDMA;
+	cmd.cmd_common.header.sub_type = DMUB_CMD__LSDMA_LINEAR_SUB_WINDOW_COPY;
+	wait_type                      = DM_DMUB_WAIT_TYPE_NO_WAIT;
+
+	lsdma_data->u.linear_sub_window_copy_data.tmz              = copy_data.tmz;
+	lsdma_data->u.linear_sub_window_copy_data.element_size     = copy_data.element_size;
+	lsdma_data->u.linear_sub_window_copy_data.src_lo           = copy_data.src_lo;
+	lsdma_data->u.linear_sub_window_copy_data.src_hi           = copy_data.src_hi;
+	lsdma_data->u.linear_sub_window_copy_data.src_x            = copy_data.src_x;
+	lsdma_data->u.linear_sub_window_copy_data.src_y            = copy_data.src_y;
+	lsdma_data->u.linear_sub_window_copy_data.src_pitch        = copy_data.src_pitch;
+	lsdma_data->u.linear_sub_window_copy_data.src_slice_pitch  = copy_data.src_slice_pitch;
+	lsdma_data->u.linear_sub_window_copy_data.dst_lo           = copy_data.dst_lo;
+	lsdma_data->u.linear_sub_window_copy_data.dst_hi           = copy_data.dst_hi;
+	lsdma_data->u.linear_sub_window_copy_data.dst_x            = copy_data.dst_x;
+	lsdma_data->u.linear_sub_window_copy_data.dst_y            = copy_data.dst_y;
+	lsdma_data->u.linear_sub_window_copy_data.dst_pitch        = copy_data.dst_pitch;
+	lsdma_data->u.linear_sub_window_copy_data.dst_slice_pitch  = copy_data.dst_slice_pitch;
+	lsdma_data->u.linear_sub_window_copy_data.rect_x           = copy_data.rect_x;
+	lsdma_data->u.linear_sub_window_copy_data.rect_y           = copy_data.rect_y;
+	lsdma_data->u.linear_sub_window_copy_data.src_cache_policy = copy_data.src_cache_policy;
+	lsdma_data->u.linear_sub_window_copy_data.dst_cache_policy = copy_data.dst_cache_policy;
+
+	result = dc_wake_and_execute_dmub_cmd(dc_ctx, &cmd, wait_type);
+
+	if (!result)
+		DC_ERROR("LSDMA Linear Sub Window Copy failed in DMUB");
+
+	return result;
+}
+
+bool dmub_lsdma_send_tiled_to_tiled_copy_command(
+	struct dc_dmub_srv *dc_dmub_srv,
+	struct lsdma_send_tiled_to_tiled_copy_command_params params
+)
+{
+	struct dc_context *dc_ctx = dc_dmub_srv->ctx;
+	union dmub_rb_cmd cmd;
+	enum dm_dmub_wait_type wait_type;
+	struct dmub_cmd_lsdma_data *lsdma_data = &cmd.lsdma.lsdma_data;
+	bool result;
+
+	memset(&cmd, 0, sizeof(cmd));
+
+	cmd.cmd_common.header.type     = DMUB_CMD__LSDMA;
+	cmd.cmd_common.header.sub_type = DMUB_CMD__LSDMA_TILED_TO_TILED_COPY;
+	wait_type                      = DM_DMUB_WAIT_TYPE_NO_WAIT;
+
+	lsdma_data->u.tiled_copy_data.src_addr_lo      = params.src_addr & 0xFFFFFFFF;
+	lsdma_data->u.tiled_copy_data.src_addr_hi      = (params.src_addr >> 32) & 0xFFFFFFFF;
+	lsdma_data->u.tiled_copy_data.dst_addr_lo      = params.dst_addr & 0xFFFFFFFF;
+	lsdma_data->u.tiled_copy_data.dst_addr_hi      = (params.dst_addr >> 32) & 0xFFFFFFFF;
+	lsdma_data->u.tiled_copy_data.src_x            = params.src_x;
+	lsdma_data->u.tiled_copy_data.src_y            = params.src_y;
+	lsdma_data->u.tiled_copy_data.dst_x            = params.dst_x;
+	lsdma_data->u.tiled_copy_data.dst_y            = params.dst_y;
+	lsdma_data->u.tiled_copy_data.src_width        = params.src_width;
+	lsdma_data->u.tiled_copy_data.dst_width        = params.dst_width;
+	lsdma_data->u.tiled_copy_data.src_swizzle_mode = params.swizzle_mode;
+	lsdma_data->u.tiled_copy_data.dst_swizzle_mode = params.swizzle_mode;
+	lsdma_data->u.tiled_copy_data.src_element_size = params.element_size;
+	lsdma_data->u.tiled_copy_data.dst_element_size = params.element_size;
+	lsdma_data->u.tiled_copy_data.rect_x           = params.rect_x;
+	lsdma_data->u.tiled_copy_data.rect_y           = params.rect_y;
+	lsdma_data->u.tiled_copy_data.dcc              = params.dcc;
+	lsdma_data->u.tiled_copy_data.tmz              = params.tmz;
+	lsdma_data->u.tiled_copy_data.read_compress    = params.read_compress;
+	lsdma_data->u.tiled_copy_data.write_compress   = params.write_compress;
+	lsdma_data->u.tiled_copy_data.src_height       = params.src_height;
+	lsdma_data->u.tiled_copy_data.dst_height       = params.dst_height;
+	lsdma_data->u.tiled_copy_data.data_format      = params.data_format;
+	lsdma_data->u.tiled_copy_data.max_com          = params.max_com;
+	lsdma_data->u.tiled_copy_data.max_uncom        = params.max_uncom;
+
+	result = dc_wake_and_execute_dmub_cmd(dc_ctx, &cmd, wait_type);
+
+	if (!result)
+		DC_ERROR("LSDMA Tiled to Tiled Copy failed in DMUB");
+
+	return result;
+}
+
+bool dmub_lsdma_send_pio_copy_command(
+	struct dc_dmub_srv *dc_dmub_srv,
+	uint64_t src_addr,
+	uint64_t dst_addr,
+	uint32_t byte_count,
+	uint32_t overlap_disable
+)
+{
+	struct dc_context *dc_ctx = dc_dmub_srv->ctx;
+	union dmub_rb_cmd cmd;
+	enum dm_dmub_wait_type wait_type;
+	struct dmub_cmd_lsdma_data *lsdma_data = &cmd.lsdma.lsdma_data;
+	bool result;
+
+	memset(&cmd, 0, sizeof(cmd));
+
+	cmd.cmd_common.header.type     = DMUB_CMD__LSDMA;
+	cmd.cmd_common.header.sub_type = DMUB_CMD__LSDMA_PIO_COPY;
+	wait_type                      = DM_DMUB_WAIT_TYPE_NO_WAIT;
+
+	lsdma_data->u.pio_copy_data.packet.fields.byte_count      = byte_count;
+	lsdma_data->u.pio_copy_data.packet.fields.overlap_disable = overlap_disable;
+	lsdma_data->u.pio_copy_data.src_lo                        = src_addr & 0xFFFFFFFF;
+	lsdma_data->u.pio_copy_data.src_hi                        = (src_addr >> 32) & 0xFFFFFFFF;
+	lsdma_data->u.pio_copy_data.dst_lo                        = dst_addr & 0xFFFFFFFF;
+	lsdma_data->u.pio_copy_data.dst_hi                        = (dst_addr >> 32) & 0xFFFFFFFF;
+
+	result = dc_wake_and_execute_dmub_cmd(dc_ctx, &cmd, wait_type);
+
+	if (!result)
+		DC_ERROR("LSDMA PIO Copy failed in DMUB");
+
+	return result;
+}
+
+bool dmub_lsdma_send_pio_constfill_command(
+	struct dc_dmub_srv *dc_dmub_srv,
+	uint64_t dst_addr,
+	uint32_t byte_count,
+	uint32_t data
+)
+{
+	struct dc_context *dc_ctx = dc_dmub_srv->ctx;
+	union dmub_rb_cmd cmd;
+	enum dm_dmub_wait_type wait_type;
+	struct dmub_cmd_lsdma_data *lsdma_data = &cmd.lsdma.lsdma_data;
+	bool result;
+
+	memset(&cmd, 0, sizeof(cmd));
+
+	cmd.cmd_common.header.type     = DMUB_CMD__LSDMA;
+	cmd.cmd_common.header.sub_type = DMUB_CMD__LSDMA_PIO_CONSTFILL;
+	wait_type                      = DM_DMUB_WAIT_TYPE_NO_WAIT;
+
+	lsdma_data->u.pio_constfill_data.packet.fields.constant_fill = 1;
+	lsdma_data->u.pio_constfill_data.packet.fields.byte_count    = byte_count;
+	lsdma_data->u.pio_constfill_data.dst_lo                      = dst_addr & 0xFFFFFFFF;
+	lsdma_data->u.pio_constfill_data.dst_hi                      = (dst_addr >> 32) & 0xFFFFFFFF;
+	lsdma_data->u.pio_constfill_data.data                        = data;
+
+	result = dc_wake_and_execute_dmub_cmd(dc_ctx, &cmd, wait_type);
+
+	if (!result)
+		DC_ERROR("LSDMA PIO Constfill failed in DMUB");
+
+	return result;
+}
+
+bool dmub_lsdma_send_poll_reg_write_command(struct dc_dmub_srv *dc_dmub_srv, uint32_t reg_addr, uint32_t reg_data)
+{
+	struct dc_context *dc_ctx = dc_dmub_srv->ctx;
+	union dmub_rb_cmd cmd;
+	enum dm_dmub_wait_type wait_type;
+	struct dmub_cmd_lsdma_data *lsdma_data = &cmd.lsdma.lsdma_data;
+	bool result;
+
+	memset(&cmd, 0, sizeof(cmd));
+
+	cmd.cmd_common.header.type     = DMUB_CMD__LSDMA;
+	cmd.cmd_common.header.sub_type = DMUB_CMD__LSDMA_POLL_REG_WRITE;
+	wait_type                      = DM_DMUB_WAIT_TYPE_NO_WAIT;
+
+	lsdma_data->u.reg_write_data.reg_addr = reg_addr;
+	lsdma_data->u.reg_write_data.reg_data = reg_data;
+
+	result = dc_wake_and_execute_dmub_cmd(dc_ctx, &cmd, wait_type);
+
+	if (!result)
+		DC_ERROR("LSDMA Poll Reg failed in DMUB");
+
+	return result;
+}
+
+void dc_dmub_srv_release_hw(const struct dc *dc)
+{
+	struct dc_dmub_srv *dc_dmub_srv = dc->ctx->dmub_srv;
+	union dmub_rb_cmd cmd = {0};
+
+	if (!dc_dmub_srv || !dc_dmub_srv->dmub)
+		return;
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.idle_opt_notify_idle.header.type = DMUB_CMD__IDLE_OPT;
+	cmd.idle_opt_notify_idle.header.sub_type = DMUB_CMD__IDLE_OPT_RELEASE_HW;
+	cmd.idle_opt_notify_idle.header.payload_bytes =
+		sizeof(cmd.idle_opt_notify_idle) -
+		sizeof(cmd.idle_opt_notify_idle.header);
+
+	dm_execute_dmub_cmd(dc->ctx, &cmd,  DM_DMUB_WAIT_TYPE_WAIT);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.h b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.h
new file mode 100644
index 000000000000..7ef93444ef3c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.h
@@ -0,0 +1,334 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef _DMUB_DC_SRV_H_
+#define _DMUB_DC_SRV_H_
+
+#include "dm_services_types.h"
+#include "dmub/dmub_srv.h"
+
+struct dmub_srv;
+struct dc;
+struct pipe_ctx;
+struct dc_crtc_timing_adjust;
+struct dc_crtc_timing;
+struct dc_state;
+struct dc_surface_update;
+
+struct dc_reg_helper_state {
+	bool gather_in_progress;
+	uint32_t same_addr_count;
+	bool should_burst_write;
+	union dmub_rb_cmd cmd_data;
+	unsigned int reg_seq_count;
+};
+
+struct dc_dmub_srv {
+	struct dmub_srv *dmub;
+	struct dc_reg_helper_state reg_helper_offload;
+
+	struct dc_context *ctx;
+	void *dm;
+
+	int32_t idle_exit_counter;
+	union dmub_shared_state_ips_driver_signals driver_signals;
+	bool idle_allowed;
+	bool needs_idle_wake;
+};
+
+bool dc_dmub_srv_wait_for_pending(struct dc_dmub_srv *dc_dmub_srv);
+
+bool dc_dmub_srv_optimized_init_done(struct dc_dmub_srv *dc_dmub_srv);
+
+bool dc_dmub_srv_cmd_list_queue_execute(struct dc_dmub_srv *dc_dmub_srv,
+		unsigned int count,
+		union dmub_rb_cmd *cmd_list);
+
+bool dc_dmub_srv_wait_for_idle(struct dc_dmub_srv *dc_dmub_srv,
+		enum dm_dmub_wait_type wait_type,
+		union dmub_rb_cmd *cmd_list);
+
+bool dc_dmub_srv_cmd_run(struct dc_dmub_srv *dc_dmub_srv, union dmub_rb_cmd *cmd, enum dm_dmub_wait_type wait_type);
+
+bool dc_dmub_srv_cmd_run_list(struct dc_dmub_srv *dc_dmub_srv, unsigned int count, union dmub_rb_cmd *cmd_list, enum dm_dmub_wait_type wait_type);
+
+bool dc_dmub_srv_notify_stream_mask(struct dc_dmub_srv *dc_dmub_srv,
+				   unsigned int stream_mask);
+
+bool dc_dmub_srv_is_restore_required(struct dc_dmub_srv *dc_dmub_srv);
+
+bool dc_dmub_srv_get_dmub_outbox0_msg(const struct dc *dc, struct dmcub_trace_buf_entry *entry);
+
+void dc_dmub_trace_event_control(struct dc *dc, bool enable);
+
+void dc_dmub_srv_drr_update_cmd(struct dc *dc, uint32_t tg_inst, uint32_t vtotal_min, uint32_t vtotal_max);
+
+void dc_dmub_srv_set_drr_manual_trigger_cmd(struct dc *dc, uint32_t tg_inst);
+bool dc_dmub_srv_p_state_delegate(struct dc *dc, bool enable_pstate, struct dc_state *context);
+
+void dc_dmub_srv_query_caps_cmd(struct dc_dmub_srv *dc_dmub_srv);
+void dc_dmub_srv_get_visual_confirm_color_cmd(struct dc *dc, struct pipe_ctx *pipe_ctx);
+void dc_dmub_srv_clear_inbox0_ack(struct dc_dmub_srv *dmub_srv);
+void dc_dmub_srv_wait_for_inbox0_ack(struct dc_dmub_srv *dmub_srv);
+void dc_dmub_srv_send_inbox0_cmd(struct dc_dmub_srv *dmub_srv, union dmub_inbox0_data_register data);
+
+bool dc_dmub_srv_get_diagnostic_data(struct dc_dmub_srv *dc_dmub_srv);
+
+void dc_dmub_setup_subvp_dmub_command(struct dc *dc, struct dc_state *context, bool enable);
+void dc_dmub_srv_log_diagnostic_data(struct dc_dmub_srv *dc_dmub_srv);
+
+void dc_send_update_cursor_info_to_dmu(struct pipe_ctx *pCtx, uint8_t pipe_idx);
+bool dc_dmub_check_min_version(struct dmub_srv *srv);
+
+void dc_dmub_srv_enable_dpia_trace(const struct dc *dc);
+void dc_dmub_srv_subvp_save_surf_addr(const struct dc_dmub_srv *dc_dmub_srv, const struct dc_plane_address *addr, uint8_t subvp_index);
+
+bool dc_dmub_srv_is_hw_pwr_up(struct dc_dmub_srv *dc_dmub_srv, bool wait);
+
+void dc_dmub_srv_apply_idle_power_optimizations(const struct dc *dc, bool allow_idle);
+
+/**
+ * dc_dmub_srv_set_power_state() - Sets the power state for DMUB service.
+ *
+ * Controls whether messaging the DMCUB or interfacing with it via HW register
+ * interaction is permittable.
+ *
+ * @dc_dmub_srv - The DC DMUB service pointer
+ * @power_state - the DC power state
+ */
+void dc_dmub_srv_set_power_state(struct dc_dmub_srv *dc_dmub_srv, enum dc_acpi_cm_power_state power_state);
+
+/**
+ * dc_dmub_srv_notify_fw_dc_power_state() - Notifies firmware of the DC power state.
+ *
+ * Differs from dc_dmub_srv_set_power_state in that it needs to access HW in order
+ * to message DMCUB of the state transition. Should come after the D0 exit and
+ * before D3 set power state.
+ *
+ * @dc_dmub_srv - The DC DMUB service pointer
+ * @power_state - the DC power state
+ */
+void dc_dmub_srv_notify_fw_dc_power_state(struct dc_dmub_srv *dc_dmub_srv,
+					  enum dc_acpi_cm_power_state power_state);
+
+/**
+ * @dc_dmub_srv_should_detect() - Checks if link detection is required.
+ *
+ * While in idle power states we may need driver to manually redetect in
+ * the case of a missing hotplug. Should be called from a polling timer.
+ *
+ * Return: true if redetection is required.
+ */
+bool dc_dmub_srv_should_detect(struct dc_dmub_srv *dc_dmub_srv);
+
+/**
+ * dc_wake_and_execute_dmub_cmd() - Wrapper for DMUB command execution.
+ *
+ * Refer to dc_wake_and_execute_dmub_cmd_list() for usage and limitations,
+ * This function is a convenience wrapper for a single command execution.
+ *
+ * @ctx: DC context
+ * @cmd: The command to send/receive
+ * @wait_type: The wait behavior for the execution
+ *
+ * Return: true on command submission success, false otherwise
+ */
+bool dc_wake_and_execute_dmub_cmd(const struct dc_context *ctx, union dmub_rb_cmd *cmd,
+				  enum dm_dmub_wait_type wait_type);
+
+/**
+ * dc_wake_and_execute_dmub_cmd_list() - Wrapper for DMUB command list execution.
+ *
+ * If the DMCUB hardware was asleep then it wakes the DMUB before
+ * executing the command and attempts to re-enter if the command
+ * submission was successful.
+ *
+ * This should be the preferred command submission interface provided
+ * the DC lock is acquired.
+ *
+ * Entry/exit out of idle power optimizations would need to be
+ * manually performed otherwise through dc_allow_idle_optimizations().
+ *
+ * @ctx: DC context
+ * @count: Number of commands to send/receive
+ * @cmd: Array of commands to send
+ * @wait_type: The wait behavior for the execution
+ *
+ * Return: true on command submission success, false otherwise
+ */
+bool dc_wake_and_execute_dmub_cmd_list(const struct dc_context *ctx, unsigned int count,
+				       union dmub_rb_cmd *cmd, enum dm_dmub_wait_type wait_type);
+
+/**
+ * dc_wake_and_execute_gpint()
+ *
+ * @ctx: DC context
+ * @command_code: The command ID to send to DMCUB
+ * @param: The parameter to message DMCUB
+ * @response: Optional response out value - may be NULL.
+ * @wait_type: The wait behavior for the execution
+ */
+bool dc_wake_and_execute_gpint(const struct dc_context *ctx, enum dmub_gpint_command command_code,
+			       uint16_t param, uint32_t *response, enum dm_dmub_wait_type wait_type);
+
+void dc_dmub_srv_fams2_update_config(struct dc *dc,
+		struct dc_state *context,
+		bool enable);
+void dc_dmub_srv_fams2_drr_update(struct dc *dc,
+		uint32_t tg_inst,
+		uint32_t vtotal_min,
+		uint32_t vtotal_max,
+		uint32_t vtotal_mid,
+		uint32_t vtotal_mid_frame_num,
+		bool program_manual_trigger);
+void dc_dmub_srv_fams2_passthrough_flip(
+		struct dc *dc,
+		struct dc_state *state,
+		struct dc_stream_state *stream,
+		struct dc_surface_update *srf_updates,
+		int surface_count);
+
+bool dmub_lsdma_init(struct dc_dmub_srv *dc_dmub_srv);
+bool dmub_lsdma_send_linear_copy_command(
+	struct dc_dmub_srv *dc_dmub_srv,
+	uint64_t src_addr,
+	uint64_t dst_addr,
+	uint32_t count);
+
+struct lsdma_linear_sub_window_copy_params {
+	uint32_t src_lo;
+	uint32_t src_hi;
+
+	uint32_t dst_lo;
+	uint32_t dst_hi;
+
+	uint32_t src_x        : 16;
+	uint32_t src_y        : 16;
+
+	uint32_t dst_x        : 16;
+	uint32_t dst_y        : 16;
+
+	uint32_t rect_x       : 16;
+	uint32_t rect_y       : 16;
+
+	uint32_t src_pitch    : 16;
+	uint32_t dst_pitch    : 16;
+
+	uint32_t src_slice_pitch;
+	uint32_t dst_slice_pitch;
+
+	uint32_t tmz              : 1;
+	uint32_t element_size     : 3;
+	uint32_t src_cache_policy : 3;
+	uint32_t dst_cache_policy : 3;
+	uint32_t padding          : 22;
+};
+
+bool dmub_lsdma_send_linear_sub_window_copy_command(
+	struct dc_dmub_srv *dc_dmub_srv,
+	struct lsdma_linear_sub_window_copy_params copy_data
+);
+bool dmub_lsdma_send_pio_copy_command(
+	struct dc_dmub_srv *dc_dmub_srv,
+	uint64_t src_addr,
+	uint64_t dst_addr,
+	uint32_t byte_count,
+	uint32_t overlap_disable);
+bool dmub_lsdma_send_pio_constfill_command(
+	struct dc_dmub_srv *dc_dmub_srv,
+	uint64_t dst_addr,
+	uint32_t byte_count,
+	uint32_t data);
+
+struct lsdma_send_tiled_to_tiled_copy_command_params {
+	uint64_t src_addr;
+	uint64_t dst_addr;
+
+	uint32_t src_x            : 16;
+	uint32_t src_y            : 16;
+
+	uint32_t dst_x            : 16;
+	uint32_t dst_y            : 16;
+
+	uint32_t src_width        : 16;
+	uint32_t dst_width        : 16;
+
+	uint32_t rect_x           : 16;
+	uint32_t rect_y           : 16;
+
+	uint32_t src_height       : 16;
+	uint32_t dst_height       : 16;
+
+	uint32_t data_format      : 6;
+	uint32_t swizzle_mode     : 5;
+	uint32_t element_size     : 3;
+	uint32_t dcc              : 1;
+	uint32_t tmz              : 1;
+	uint32_t read_compress    : 2;
+	uint32_t write_compress   : 2;
+	uint32_t max_com          : 2;
+	uint32_t max_uncom        : 1;
+	uint32_t padding          : 9;
+};
+
+bool dmub_lsdma_send_tiled_to_tiled_copy_command(
+	struct dc_dmub_srv *dc_dmub_srv,
+	struct lsdma_send_tiled_to_tiled_copy_command_params params);
+bool dmub_lsdma_send_poll_reg_write_command(struct dc_dmub_srv *dc_dmub_srv, uint32_t reg_addr, uint32_t reg_data);
+
+/**
+ * struct ips_residency_info - struct containing info from dmub_ips_residency_stats
+ *
+ * @ips_mode: The mode of IPS that the follow stats appertain to
+ * @residency_percent: The percentage of time spent in given IPS mode in millipercent
+ * @entry_counter: The number of entries made in to this IPS state
+ * @total_active_time_us: uint32_t array of length 2 representing time in the given IPS mode
+ *                        in microseconds. Index 0 is lower 32 bits, index 1 is upper 32 bits.
+ * @total_inactive_time_us: uint32_t array of length 2 representing time outside the given IPS mode
+ *                          in microseconds. Index 0 is lower 32 bits, index 1 is upper 32 bits.
+ * @histogram: Histogram of given IPS state durations - bucket definitions in dmub_ips.c
+ */
+struct ips_residency_info {
+	enum ips_residency_mode ips_mode;
+	unsigned int residency_percent;
+	unsigned int entry_counter;
+	unsigned int total_active_time_us[2];
+	unsigned int total_inactive_time_us[2];
+	unsigned int histogram[16];
+};
+
+bool dc_dmub_srv_ips_residency_cntl(const struct dc_context *ctx, uint8_t panel_inst, bool start_measurement);
+
+bool dc_dmub_srv_ips_query_residency_info(const struct dc_context *ctx, uint8_t panel_inst,
+					  struct dmub_ips_residency_info *driver_info,
+					  enum ips_residency_mode ips_mode);
+
+/**
+ * dc_dmub_srv_release_hw() - Notifies DMUB service that HW access is no longer required.
+ *
+ * @dc - pointer to DC object
+ */
+void dc_dmub_srv_release_hw(const struct dc *dc);
+#endif /* _DMUB_DC_SRV_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/dc_dp_types.h b/drivers/gpu/drm/amd/display/dc/dc_dp_types.h
new file mode 100644
index 000000000000..db669ccb1d58
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_dp_types.h
@@ -0,0 +1,1531 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DC_DP_TYPES_H
+#define DC_DP_TYPES_H
+
+#include "os_types.h"
+#include "dc_ddc_types.h"
+
+enum dc_lane_count {
+	LANE_COUNT_UNKNOWN = 0,
+	LANE_COUNT_ONE = 1,
+	LANE_COUNT_TWO = 2,
+	LANE_COUNT_FOUR = 4,
+	LANE_COUNT_EIGHT = 8,
+	LANE_COUNT_DP_MAX = LANE_COUNT_FOUR
+};
+
+/* This is actually a reference clock (27MHz) multiplier
+ * 162MBps bandwidth for 1.62GHz like rate,
+ * 270MBps for 2.70GHz,
+ * 324MBps for 3.24Ghz,
+ * 540MBps for 5.40GHz
+ * 810MBps for 8.10GHz
+ */
+enum dc_link_rate {
+	LINK_RATE_UNKNOWN = 0,
+	LINK_RATE_LOW = 0x06,		// Rate_1 (RBR)  - 1.62 Gbps/Lane
+	LINK_RATE_RATE_2 = 0x08,	// Rate_2        - 2.16 Gbps/Lane
+	LINK_RATE_RATE_3 = 0x09,	// Rate_3        - 2.43 Gbps/Lane
+	LINK_RATE_HIGH = 0x0A,		// Rate_4 (HBR)  - 2.70 Gbps/Lane
+	LINK_RATE_RBR2 = 0x0C,		// Rate_5 (RBR2) - 3.24 Gbps/Lane
+	LINK_RATE_RATE_6 = 0x10,	// Rate_6        - 4.32 Gbps/Lane
+	LINK_RATE_HIGH2 = 0x14,		// Rate_7 (HBR2) - 5.40 Gbps/Lane
+	LINK_RATE_RATE_8 = 0x19,	// Rate_8        - 6.75 Gbps/Lane
+	LINK_RATE_HIGH3 = 0x1E,		// Rate_9 (HBR3) - 8.10 Gbps/Lane
+	/* Starting from DP2.0 link rate enum directly represents actual
+	 * link rate value in unit of 10 mbps
+	 */
+	LINK_RATE_UHBR10 = 1000,	// UHBR10 - 10.0 Gbps/Lane
+	LINK_RATE_UHBR13_5 = 1350,	// UHBR13.5 - 13.5 Gbps/Lane
+	LINK_RATE_UHBR20 = 2000,	// UHBR20 - 20.0 Gbps/Lane
+};
+
+enum dc_link_spread {
+	LINK_SPREAD_DISABLED = 0x00,
+	/* 0.5 % downspread 30 kHz */
+	LINK_SPREAD_05_DOWNSPREAD_30KHZ = 0x10,
+	/* 0.5 % downspread 33 kHz */
+	LINK_SPREAD_05_DOWNSPREAD_33KHZ = 0x11
+};
+
+enum dc_voltage_swing {
+	VOLTAGE_SWING_LEVEL0 = 0,	/* direct HW translation! */
+	VOLTAGE_SWING_LEVEL1,
+	VOLTAGE_SWING_LEVEL2,
+	VOLTAGE_SWING_LEVEL3,
+	VOLTAGE_SWING_MAX_LEVEL = VOLTAGE_SWING_LEVEL3
+};
+
+enum dc_pre_emphasis {
+	PRE_EMPHASIS_DISABLED = 0,	/* direct HW translation! */
+	PRE_EMPHASIS_LEVEL1,
+	PRE_EMPHASIS_LEVEL2,
+	PRE_EMPHASIS_LEVEL3,
+	PRE_EMPHASIS_MAX_LEVEL = PRE_EMPHASIS_LEVEL3
+};
+/* Post Cursor 2 is optional for transmitter
+ * and it applies only to the main link operating at HBR2
+ */
+enum dc_post_cursor2 {
+	POST_CURSOR2_DISABLED = 0,	/* direct HW translation! */
+	POST_CURSOR2_LEVEL1,
+	POST_CURSOR2_LEVEL2,
+	POST_CURSOR2_LEVEL3,
+	POST_CURSOR2_MAX_LEVEL = POST_CURSOR2_LEVEL3,
+};
+
+enum dc_dp_ffe_preset_level {
+	DP_FFE_PRESET_LEVEL0 = 0,
+	DP_FFE_PRESET_LEVEL1,
+	DP_FFE_PRESET_LEVEL2,
+	DP_FFE_PRESET_LEVEL3,
+	DP_FFE_PRESET_LEVEL4,
+	DP_FFE_PRESET_LEVEL5,
+	DP_FFE_PRESET_LEVEL6,
+	DP_FFE_PRESET_LEVEL7,
+	DP_FFE_PRESET_LEVEL8,
+	DP_FFE_PRESET_LEVEL9,
+	DP_FFE_PRESET_LEVEL10,
+	DP_FFE_PRESET_LEVEL11,
+	DP_FFE_PRESET_LEVEL12,
+	DP_FFE_PRESET_LEVEL13,
+	DP_FFE_PRESET_LEVEL14,
+	DP_FFE_PRESET_LEVEL15,
+	DP_FFE_PRESET_MAX_LEVEL = DP_FFE_PRESET_LEVEL15,
+};
+
+enum dc_dp_training_pattern {
+	DP_TRAINING_PATTERN_SEQUENCE_1 = 0,
+	DP_TRAINING_PATTERN_SEQUENCE_2,
+	DP_TRAINING_PATTERN_SEQUENCE_3,
+	DP_TRAINING_PATTERN_SEQUENCE_4,
+	DP_TRAINING_PATTERN_VIDEOIDLE,
+	DP_128b_132b_TPS1,
+	DP_128b_132b_TPS2,
+	DP_128b_132b_TPS2_CDS,
+};
+
+enum dp_link_encoding {
+	DP_UNKNOWN_ENCODING = 0,
+	DP_8b_10b_ENCODING = 1,
+	DP_128b_132b_ENCODING = 2,
+};
+
+enum dp_test_link_rate {
+	DP_TEST_LINK_RATE_RBR		= 0x06,
+	DP_TEST_LINK_RATE_RATE_2    = 0x08,	// Rate_2        - 2.16 Gbps/Lane
+	DP_TEST_LINK_RATE_RATE_3    = 0x09,	// Rate_3        - 2.43 Gbps/Lane
+	DP_TEST_LINK_RATE_HBR		= 0x0A,
+	DP_TEST_LINK_RATE_RBR2      = 0x0C,	// Rate_5 (RBR2) - 3.24 Gbps/Lane
+	DP_TEST_LINK_RATE_RATE_6    = 0x10,	// Rate_6        - 4.32 Gbps/Lane
+	DP_TEST_LINK_RATE_HBR2		= 0x14,
+	DP_TEST_LINK_RATE_RATE_8    = 0x19,	// Rate_8        - 6.75 Gbps/Lane
+	DP_TEST_LINK_RATE_HBR3		= 0x1E,
+	DP_TEST_LINK_RATE_UHBR10	= 0x01,
+	DP_TEST_LINK_RATE_UHBR20	= 0x02,
+	DP_TEST_LINK_RATE_UHBR13_5_LEGACY	= 0x03, /* For backward compatibility*/
+	DP_TEST_LINK_RATE_UHBR13_5	= 0x04,
+};
+
+struct dc_link_settings {
+	enum dc_lane_count lane_count;
+	enum dc_link_rate link_rate;
+	enum dc_link_spread link_spread;
+	bool use_link_rate_set;
+	uint8_t link_rate_set;
+};
+
+struct dc_tunnel_settings {
+	bool should_enable_dp_tunneling;
+	bool should_use_dp_bw_allocation;
+	uint8_t cm_id;
+	uint8_t group_id;
+	uint32_t bw_granularity;
+	uint32_t estimated_bw;
+	uint32_t allocated_bw;
+};
+
+union dc_dp_ffe_preset {
+	struct {
+		uint8_t level		: 4;
+		uint8_t reserved	: 1;
+		uint8_t no_preshoot	: 1;
+		uint8_t no_deemphasis	: 1;
+		uint8_t method2		: 1;
+	} settings;
+	uint8_t raw;
+};
+
+struct dc_lane_settings {
+	enum dc_voltage_swing VOLTAGE_SWING;
+	enum dc_pre_emphasis PRE_EMPHASIS;
+	enum dc_post_cursor2 POST_CURSOR2;
+	union dc_dp_ffe_preset FFE_PRESET;
+};
+
+struct dc_link_training_overrides {
+	enum dc_voltage_swing *voltage_swing;
+	enum dc_pre_emphasis *pre_emphasis;
+	enum dc_post_cursor2 *post_cursor2;
+	union dc_dp_ffe_preset *ffe_preset;
+
+	uint16_t *cr_pattern_time;
+	uint16_t *eq_pattern_time;
+	enum dc_dp_training_pattern *pattern_for_cr;
+	enum dc_dp_training_pattern *pattern_for_eq;
+
+	enum dc_link_spread *downspread;
+	bool *alternate_scrambler_reset;
+	bool *enhanced_framing;
+	bool *mst_enable;
+	bool *fec_enable;
+};
+
+union payload_table_update_status {
+	struct {
+		uint8_t  VC_PAYLOAD_TABLE_UPDATED:1;
+		uint8_t  ACT_HANDLED:1;
+	} bits;
+	uint8_t  raw;
+};
+
+union dpcd_rev {
+	struct {
+		uint8_t MINOR:4;
+		uint8_t MAJOR:4;
+	} bits;
+	uint8_t raw;
+};
+
+union max_lane_count {
+	struct {
+		uint8_t MAX_LANE_COUNT:5;
+		uint8_t POST_LT_ADJ_REQ_SUPPORTED:1;
+		uint8_t TPS3_SUPPORTED:1;
+		uint8_t ENHANCED_FRAME_CAP:1;
+	} bits;
+	uint8_t raw;
+};
+
+union max_down_spread {
+	struct {
+		uint8_t MAX_DOWN_SPREAD:1;
+		uint8_t RESERVED:5;
+		uint8_t NO_AUX_HANDSHAKE_LINK_TRAINING:1;
+		uint8_t TPS4_SUPPORTED:1;
+	} bits;
+	uint8_t raw;
+};
+
+union mstm_cap {
+	struct {
+		uint8_t MST_CAP:1;
+		uint8_t RESERVED:7;
+	} bits;
+	uint8_t raw;
+};
+
+union lane_count_set {
+	struct {
+		uint8_t LANE_COUNT_SET:5;
+		uint8_t POST_LT_ADJ_REQ_GRANTED:1;
+		uint8_t RESERVED:1;
+		uint8_t ENHANCED_FRAMING:1;
+	} bits;
+	uint8_t raw;
+};
+
+union lane_status {
+	struct {
+		uint8_t CR_DONE_0:1;
+		uint8_t CHANNEL_EQ_DONE_0:1;
+		uint8_t SYMBOL_LOCKED_0:1;
+		uint8_t RESERVED0:1;
+		uint8_t CR_DONE_1:1;
+		uint8_t CHANNEL_EQ_DONE_1:1;
+		uint8_t SYMBOL_LOCKED_1:1;
+		uint8_t RESERVED_1:1;
+	} bits;
+	uint8_t raw;
+};
+
+union device_service_irq {
+	struct {
+		uint8_t REMOTE_CONTROL_CMD_PENDING:1;
+		uint8_t AUTOMATED_TEST:1;
+		uint8_t CP_IRQ:1;
+		uint8_t MCCS_IRQ:1;
+		uint8_t DOWN_REP_MSG_RDY:1;
+		uint8_t UP_REQ_MSG_RDY:1;
+		uint8_t SINK_SPECIFIC:1;
+		uint8_t reserved:1;
+	} bits;
+	uint8_t raw;
+};
+
+union sink_count {
+	struct {
+		uint8_t SINK_COUNT:6;
+		uint8_t CPREADY:1;
+		uint8_t RESERVED:1;
+	} bits;
+	uint8_t raw;
+};
+
+union lane_align_status_updated {
+	struct {
+		uint8_t INTERLANE_ALIGN_DONE:1;
+		uint8_t POST_LT_ADJ_REQ_IN_PROGRESS:1;
+		uint8_t EQ_INTERLANE_ALIGN_DONE_128b_132b:1;
+		uint8_t CDS_INTERLANE_ALIGN_DONE_128b_132b:1;
+		uint8_t LT_FAILED_128b_132b:1;
+		uint8_t RESERVED:1;
+		uint8_t DOWNSTREAM_PORT_STATUS_CHANGED:1;
+		uint8_t LINK_STATUS_UPDATED:1;
+	} bits;
+	uint8_t raw;
+};
+
+union link_service_irq_vector_esi0 {
+	struct {
+		uint8_t DP_LINK_RX_CAP_CHANGED:1;
+		uint8_t DP_LINK_STATUS_CHANGED:1;
+		uint8_t DP_LINK_STREAM_STATUS_CHANGED:1;
+		uint8_t DP_LINK_HDMI_LINK_STATUS_CHANGED:1;
+		uint8_t DP_LINK_CONNECTED_OFF_ENTRY_REQUESTED:1;
+		uint8_t DP_LINK_TUNNELING_IRQ:1;
+		uint8_t reserved:2;
+	} bits;
+	uint8_t raw;
+};
+
+union lane_adjust {
+	struct {
+		uint8_t VOLTAGE_SWING_LANE:2;
+		uint8_t PRE_EMPHASIS_LANE:2;
+		uint8_t RESERVED:4;
+	} bits;
+	struct {
+		uint8_t PRESET_VALUE	:4;
+		uint8_t RESERVED	:4;
+	} tx_ffe;
+	uint8_t raw;
+};
+
+union dpcd_training_pattern {
+	struct {
+		uint8_t TRAINING_PATTERN_SET:4;
+		uint8_t RECOVERED_CLOCK_OUT_EN:1;
+		uint8_t SCRAMBLING_DISABLE:1;
+		uint8_t SYMBOL_ERROR_COUNT_SEL:2;
+	} v1_4;
+	struct {
+		uint8_t TRAINING_PATTERN_SET:2;
+		uint8_t LINK_QUAL_PATTERN_SET:2;
+		uint8_t RESERVED:4;
+	} v1_3;
+	uint8_t raw;
+};
+
+/* Training Lane is used to configure downstream DP device's voltage swing
+and pre-emphasis levels*/
+/* The DPCD addresses are from 0x103 to 0x106*/
+union dpcd_training_lane {
+	struct {
+		uint8_t VOLTAGE_SWING_SET:2;
+		uint8_t MAX_SWING_REACHED:1;
+		uint8_t PRE_EMPHASIS_SET:2;
+		uint8_t MAX_PRE_EMPHASIS_REACHED:1;
+		uint8_t RESERVED:2;
+	} bits;
+	struct {
+		uint8_t PRESET_VALUE	:4;
+		uint8_t RESERVED	:4;
+	} tx_ffe;
+	uint8_t raw;
+};
+
+/* TMDS-converter related */
+union dwnstream_port_caps_byte0 {
+	struct {
+		uint8_t DWN_STRM_PORTX_TYPE:3;
+		uint8_t DWN_STRM_PORTX_HPD:1;
+		uint8_t RESERVERD:4;
+	} bits;
+	uint8_t raw;
+};
+
+/* these are the detailed types stored at DWN_STRM_PORTX_CAP (00080h)*/
+enum dpcd_downstream_port_detailed_type {
+	DOWN_STREAM_DETAILED_DP = 0,
+	DOWN_STREAM_DETAILED_VGA,
+	DOWN_STREAM_DETAILED_DVI,
+	DOWN_STREAM_DETAILED_HDMI,
+	DOWN_STREAM_DETAILED_NONDDC,/* has no EDID (TV,CV)*/
+	DOWN_STREAM_DETAILED_DP_PLUS_PLUS
+};
+
+union dwnstream_port_caps_byte2 {
+	struct {
+		uint8_t MAX_BITS_PER_COLOR_COMPONENT:2;
+		uint8_t MAX_ENCODED_LINK_BW_SUPPORT:3;
+		uint8_t SOURCE_CONTROL_MODE_SUPPORT:1;
+		uint8_t CONCURRENT_LINK_BRING_UP_SEQ_SUPPORT:1;
+		uint8_t RESERVED:1;
+	} bits;
+	uint8_t raw;
+};
+
+union dp_downstream_port_present {
+	uint8_t byte;
+	struct {
+		uint8_t PORT_PRESENT:1;
+		uint8_t PORT_TYPE:2;
+		uint8_t FMT_CONVERSION:1;
+		uint8_t DETAILED_CAPS:1;
+		uint8_t RESERVED:3;
+	} fields;
+};
+
+union dwnstream_port_caps_byte3_dvi {
+	struct {
+		uint8_t RESERVED1:1;
+		uint8_t DUAL_LINK:1;
+		uint8_t HIGH_COLOR_DEPTH:1;
+		uint8_t RESERVED2:5;
+	} bits;
+	uint8_t raw;
+};
+
+union dwnstream_port_caps_byte3_hdmi {
+	struct {
+		uint8_t FRAME_SEQ_TO_FRAME_PACK:1;
+		uint8_t YCrCr422_PASS_THROUGH:1;
+		uint8_t YCrCr420_PASS_THROUGH:1;
+		uint8_t YCrCr422_CONVERSION:1;
+		uint8_t YCrCr420_CONVERSION:1;
+		uint8_t RESERVED:3;
+	} bits;
+	uint8_t raw;
+};
+
+union hdmi_encoded_link_bw {
+	struct {
+		uint8_t FRL_MODE:1; // Bit 0
+		uint8_t BW_9Gbps:1;
+		uint8_t BW_18Gbps:1;
+		uint8_t BW_24Gbps:1;
+		uint8_t BW_32Gbps:1;
+		uint8_t BW_40Gbps:1;
+		uint8_t BW_48Gbps:1;
+		uint8_t FRL_LINK_TRAINING_FINISHED:1; // Bit 7
+	} bits;
+	uint8_t raw;
+};
+
+union hdmi_tx_link_status {
+	struct {
+		uint8_t HDMI_TX_LINK_ACTIVE_STATUS:1;
+		uint8_t HDMI_TX_READY_STATUS:1;
+		uint8_t RESERVED:6;
+	} bits;
+	uint8_t raw;
+};
+
+union autonomous_mode_and_frl_link_status {
+	struct {
+		uint8_t FRL_LT_IN_PROGRESS_STATUS:1;
+		uint8_t FRL_LT_LINK_CONFIG_IN_PROGRESS:3;
+		uint8_t RESERVED:1;
+		uint8_t FALLBACK_POLICY:1;
+		uint8_t FALLBACK_POLICY_VALID:1;
+		uint8_t REGULATED_AUTONOMOUS_MODE_SUPPORTED:1;
+	} bits;
+	uint8_t raw;
+};
+
+/*4-byte structure for detailed capabilities of a down-stream port
+(DP-to-TMDS converter).*/
+union dwnstream_portxcaps {
+	struct {
+		union dwnstream_port_caps_byte0 byte0;
+		unsigned char max_TMDS_clock;   //byte1
+		union dwnstream_port_caps_byte2 byte2;
+
+		union {
+			union dwnstream_port_caps_byte3_dvi byteDVI;
+			union dwnstream_port_caps_byte3_hdmi byteHDMI;
+		} byte3;
+	} bytes;
+
+	unsigned char raw[4];
+};
+
+union downstream_port {
+	struct {
+		unsigned char   present:1;
+		unsigned char   type:2;
+		unsigned char   format_conv:1;
+		unsigned char   detailed_caps:1;
+		unsigned char   reserved:3;
+	} bits;
+	unsigned char raw;
+};
+
+
+union sink_status {
+	struct {
+		uint8_t RX_PORT0_STATUS:1;
+		uint8_t RX_PORT1_STATUS:1;
+		uint8_t RESERVED:6;
+	} bits;
+	uint8_t raw;
+};
+
+/* 7-byte structure corresponding to 6 registers (200h-205h)
+ * and LINK_SERVICE_IRQ_ESI0 (2005h) for tunneling IRQ
+ * read during handling of HPD-IRQ
+ */
+union hpd_irq_data {
+	struct {
+		union sink_count sink_cnt;/* 200h */
+		union device_service_irq device_service_irq;/* 201h */
+		union lane_status lane01_status;/* 202h */
+		union lane_status lane23_status;/* 203h */
+		union lane_align_status_updated lane_status_updated;/* 204h */
+		union sink_status sink_status;/* 205h */
+		union link_service_irq_vector_esi0 link_service_irq_esi0;/* 2005h */
+	} bytes;
+	uint8_t raw[7];
+};
+
+union down_stream_port_count {
+	struct {
+		uint8_t DOWN_STR_PORT_COUNT:4;
+		uint8_t RESERVED:2; /*Bits 5:4 = RESERVED. Read all 0s.*/
+		/*Bit 6 = MSA_TIMING_PAR_IGNORED
+		0 = Sink device requires the MSA timing parameters
+		1 = Sink device is capable of rendering incoming video
+		 stream without MSA timing parameters*/
+		uint8_t IGNORE_MSA_TIMING_PARAM:1;
+		/*Bit 7 = OUI Support
+		0 = OUI not supported
+		1 = OUI supported
+		(OUI and Device Identification mandatory for DP 1.2)*/
+		uint8_t OUI_SUPPORT:1;
+	} bits;
+	uint8_t raw;
+};
+
+union down_spread_ctrl {
+	struct {
+		uint8_t RESERVED1:4;/* Bit 3:0 = RESERVED. Read all 0s*/
+	/* Bits 4 = SPREAD_AMP. Spreading amplitude
+	0 = Main link signal is not downspread
+	1 = Main link signal is downspread <= 0.5%
+	with frequency in the range of 30kHz ~ 33kHz*/
+		uint8_t SPREAD_AMP:1;
+		uint8_t RESERVED2:1;/*Bit 5 = RESERVED. Read all 0s*/
+	/* Bit 6 = FIXED_VTOTAL_AS_SDP_EN_IN_PR_ACTIVE.
+	0 = FIXED_VTOTAL_AS_SDP_EN_IN_PR_ACTIVE is not enabled by the Source device (default)
+	1 = FIXED_VTOTAL_AS_SDP_EN_IN_PR_ACTIVE is enabled by Source device */
+		uint8_t FIXED_VTOTAL_AS_SDP_EN_IN_PR_ACTIVE:1;
+	/*Bit 7 = MSA_TIMING_PAR_IGNORE_EN
+	0 = Source device will send valid data for the MSA Timing Params
+	1 = Source device may send invalid data for these MSA Timing Params*/
+		uint8_t IGNORE_MSA_TIMING_PARAM:1;
+	} bits;
+	uint8_t raw;
+};
+
+union dpcd_edp_config {
+	struct {
+		uint8_t PANEL_MODE_EDP:1;
+		uint8_t FRAMING_CHANGE_ENABLE:1;
+		uint8_t RESERVED:5;
+		uint8_t PANEL_SELF_TEST_ENABLE:1;
+	} bits;
+	uint8_t raw;
+};
+
+struct dp_device_vendor_id {
+	uint8_t ieee_oui[3];/*24-bit IEEE OUI*/
+	uint8_t ieee_device_id[6];/*usually 6-byte ASCII name*/
+};
+
+struct dp_sink_hw_fw_revision {
+	uint8_t ieee_hw_rev;
+	uint8_t ieee_fw_rev[2];
+};
+
+struct dpcd_vendor_signature {
+	bool is_valid;
+
+	union dpcd_ieee_vendor_signature {
+		struct {
+			uint8_t ieee_oui[3];/*24-bit IEEE OUI*/
+			uint8_t ieee_device_id[6];/*usually 6-byte ASCII name*/
+			uint8_t ieee_hw_rev;
+			uint8_t ieee_fw_rev[2];
+		};
+		uint8_t raw[12];
+	} data;
+};
+
+struct dpcd_amd_signature {
+	uint8_t AMD_IEEE_TxSignature_byte1;
+	uint8_t AMD_IEEE_TxSignature_byte2;
+	uint8_t AMD_IEEE_TxSignature_byte3;
+};
+
+struct dpcd_amd_device_id {
+	uint8_t device_id_byte1;
+	uint8_t device_id_byte2;
+	uint8_t zero[4];
+	uint8_t dce_version;
+	uint8_t dal_version_byte1;
+	uint8_t dal_version_byte2;
+};
+
+struct target_luminance_value {
+	uint8_t byte0;
+	uint8_t byte1;
+	uint8_t byte2;
+};
+
+struct dpcd_source_backlight_set {
+	struct  {
+		uint8_t byte0;
+		uint8_t byte1;
+		uint8_t byte2;
+		uint8_t byte3;
+	} backlight_level_millinits;
+
+	struct  {
+		uint8_t byte0;
+		uint8_t byte1;
+	} backlight_transition_time_ms;
+};
+
+union dpcd_source_backlight_get {
+	struct {
+		uint32_t backlight_millinits_peak; /* 326h */
+		uint32_t backlight_millinits_avg; /* 32Ah */
+	} bytes;
+	uint8_t raw[8];
+};
+
+/*DPCD register of DP receiver capability field bits-*/
+union edp_configuration_cap {
+	struct {
+		uint8_t ALT_SCRAMBLER_RESET:1;
+		uint8_t FRAMING_CHANGE:1;
+		uint8_t RESERVED:1;
+		uint8_t DPCD_DISPLAY_CONTROL_CAPABLE:1;
+		uint8_t RESERVED2:4;
+	} bits;
+	uint8_t raw;
+};
+
+union dprx_feature {
+	struct {
+		uint8_t GTC_CAP:1;                             // bit 0: DP 1.3+
+		uint8_t SST_SPLIT_SDP_CAP:1;                   // bit 1: DP 1.4
+		uint8_t AV_SYNC_CAP:1;                         // bit 2: DP 1.3+
+		uint8_t VSC_SDP_COLORIMETRY_SUPPORTED:1;       // bit 3: DP 1.3+
+		uint8_t VSC_EXT_VESA_SDP_SUPPORTED:1;          // bit 4: DP 1.4
+		uint8_t VSC_EXT_VESA_SDP_CHAINING_SUPPORTED:1; // bit 5: DP 1.4
+		uint8_t VSC_EXT_CEA_SDP_SUPPORTED:1;           // bit 6: DP 1.4
+		uint8_t VSC_EXT_CEA_SDP_CHAINING_SUPPORTED:1;  // bit 7: DP 1.4
+	} bits;
+	uint8_t raw;
+};
+
+union training_aux_rd_interval {
+	struct {
+		uint8_t TRAINIG_AUX_RD_INTERVAL:7;
+		uint8_t EXT_RECEIVER_CAP_FIELD_PRESENT:1;
+	} bits;
+	uint8_t raw;
+};
+
+/* Automated test structures */
+union test_request {
+	struct {
+	uint8_t LINK_TRAINING                :1;
+	uint8_t LINK_TEST_PATTRN             :1;
+	uint8_t EDID_READ                    :1;
+	uint8_t PHY_TEST_PATTERN             :1;
+	uint8_t PHY_TEST_CHANNEL_CODING_TYPE :2;
+	uint8_t AUDIO_TEST_PATTERN           :1;
+	uint8_t TEST_AUDIO_DISABLED_VIDEO    :1;
+	} bits;
+	uint8_t raw;
+};
+
+union test_response {
+	struct {
+		uint8_t ACK         :1;
+		uint8_t NO_ACK      :1;
+		uint8_t EDID_CHECKSUM_WRITE:1;
+		uint8_t RESERVED    :5;
+	} bits;
+	uint8_t raw;
+};
+
+union phy_test_pattern {
+	struct {
+		/* This field is 7 bits for DP2.0 */
+		uint8_t PATTERN     :7;
+		uint8_t RESERVED    :1;
+	} bits;
+	uint8_t raw;
+};
+
+/* States of Compliance Test Specification (CTS DP1.2). */
+union compliance_test_state {
+	struct {
+		unsigned char STEREO_3D_RUNNING        : 1;
+		unsigned char RESERVED                 : 7;
+	} bits;
+	unsigned char raw;
+};
+
+union link_test_pattern {
+	struct {
+		/* dpcd_link_test_patterns */
+		unsigned char PATTERN :2;
+		unsigned char RESERVED:6;
+	} bits;
+	unsigned char raw;
+};
+
+union test_misc {
+	struct dpcd_test_misc_bits {
+		unsigned char SYNC_CLOCK  :1;
+		/* dpcd_test_color_format */
+		unsigned char CLR_FORMAT  :2;
+		/* dpcd_test_dyn_range */
+		unsigned char DYN_RANGE   :1;
+		unsigned char YCBCR_COEFS :1;
+		/* dpcd_test_bit_depth */
+		unsigned char BPC         :3;
+	} bits;
+	unsigned char raw;
+};
+
+union audio_test_mode {
+	struct {
+		unsigned char sampling_rate   :4;
+		unsigned char channel_count   :4;
+	} bits;
+	unsigned char raw;
+};
+
+union audio_test_pattern_period {
+	struct {
+		unsigned char pattern_period   :4;
+		unsigned char reserved         :4;
+	} bits;
+	unsigned char raw;
+};
+
+struct audio_test_pattern_type {
+	unsigned char value;
+};
+
+struct dp_audio_test_data_flags {
+	uint8_t test_requested  :1;
+	uint8_t disable_video   :1;
+};
+
+struct dp_audio_test_data {
+
+	struct dp_audio_test_data_flags flags;
+	uint8_t sampling_rate;
+	uint8_t channel_count;
+	uint8_t pattern_type;
+	uint8_t pattern_period[8];
+};
+
+/* FEC capability DPCD register field bits-*/
+union dpcd_fec_capability {
+	struct {
+		uint8_t FEC_CAPABLE:1;
+		uint8_t UNCORRECTED_BLOCK_ERROR_COUNT_CAPABLE:1;
+		uint8_t CORRECTED_BLOCK_ERROR_COUNT_CAPABLE:1;
+		uint8_t BIT_ERROR_COUNT_CAPABLE:1;
+		uint8_t PARITY_BLOCK_ERROR_COUNT_CAPABLE:1;
+		uint8_t ARITY_BIT_ERROR_COUNT_CAPABLE:1;
+		uint8_t FEC_RUNNING_INDICATOR_SUPPORTED:1;
+		uint8_t FEC_ERROR_REPORTING_POLICY_SUPPORTED:1;
+	} bits;
+	uint8_t raw;
+};
+
+/* DSC capability DPCD register field bits-*/
+struct dpcd_dsc_support {
+	uint8_t DSC_SUPPORT		:1;
+	uint8_t DSC_PASSTHROUGH_SUPPORT	:1;
+	uint8_t RESERVED		:6;
+};
+
+struct dpcd_dsc_algorithm_revision {
+	uint8_t DSC_VERSION_MAJOR	:4;
+	uint8_t DSC_VERSION_MINOR	:4;
+};
+
+struct dpcd_dsc_rc_buffer_block_size {
+	uint8_t RC_BLOCK_BUFFER_SIZE	:2;
+	uint8_t RESERVED		:6;
+};
+
+struct dpcd_dsc_slice_capability1 {
+	uint8_t ONE_SLICE_PER_DP_DSC_SINK_DEVICE	:1;
+	uint8_t TWO_SLICES_PER_DP_DSC_SINK_DEVICE	:1;
+	uint8_t RESERVED				:1;
+	uint8_t FOUR_SLICES_PER_DP_DSC_SINK_DEVICE	:1;
+	uint8_t SIX_SLICES_PER_DP_DSC_SINK_DEVICE	:1;
+	uint8_t EIGHT_SLICES_PER_DP_DSC_SINK_DEVICE	:1;
+	uint8_t TEN_SLICES_PER_DP_DSC_SINK_DEVICE	:1;
+	uint8_t TWELVE_SLICES_PER_DP_DSC_SINK_DEVICE	:1;
+};
+
+struct dpcd_dsc_line_buffer_bit_depth {
+	uint8_t LINE_BUFFER_BIT_DEPTH	:4;
+	uint8_t RESERVED		:4;
+};
+
+struct dpcd_dsc_block_prediction_support {
+	uint8_t BLOCK_PREDICTION_SUPPORT:1;
+	uint8_t RESERVED		:7;
+};
+
+struct dpcd_maximum_bits_per_pixel_supported_by_the_decompressor {
+	uint8_t MAXIMUM_BITS_PER_PIXEL_SUPPORTED_BY_THE_DECOMPRESSOR_LOW	:7;
+	uint8_t MAXIMUM_BITS_PER_PIXEL_SUPPORTED_BY_THE_DECOMPRESSOR_HIGH	:7;
+	uint8_t RESERVED							:2;
+};
+
+struct dpcd_dsc_decoder_color_format_capabilities {
+	uint8_t RGB_SUPPORT			:1;
+	uint8_t Y_CB_CR_444_SUPPORT		:1;
+	uint8_t Y_CB_CR_SIMPLE_422_SUPPORT	:1;
+	uint8_t Y_CB_CR_NATIVE_422_SUPPORT	:1;
+	uint8_t Y_CB_CR_NATIVE_420_SUPPORT	:1;
+	uint8_t RESERVED			:3;
+};
+
+struct dpcd_dsc_decoder_color_depth_capabilities {
+	uint8_t RESERVED0			:1;
+	uint8_t EIGHT_BITS_PER_COLOR_SUPPORT	:1;
+	uint8_t TEN_BITS_PER_COLOR_SUPPORT	:1;
+	uint8_t TWELVE_BITS_PER_COLOR_SUPPORT	:1;
+	uint8_t RESERVED1			:4;
+};
+
+struct dpcd_peak_dsc_throughput_dsc_sink {
+	uint8_t THROUGHPUT_MODE_0:4;
+	uint8_t THROUGHPUT_MODE_1:4;
+};
+
+struct dpcd_dsc_slice_capabilities_2 {
+	uint8_t SIXTEEN_SLICES_PER_DSC_SINK_DEVICE	:1;
+	uint8_t TWENTY_SLICES_PER_DSC_SINK_DEVICE	:1;
+	uint8_t TWENTYFOUR_SLICES_PER_DSC_SINK_DEVICE	:1;
+	uint8_t RESERVED				:5;
+};
+
+struct dpcd_bits_per_pixel_increment{
+	uint8_t INCREMENT_OF_BITS_PER_PIXEL_SUPPORTED	:3;
+	uint8_t RESERVED				:5;
+};
+union dpcd_dsc_basic_capabilities {
+	struct {
+		struct dpcd_dsc_support dsc_support;
+		struct dpcd_dsc_algorithm_revision dsc_algorithm_revision;
+		struct dpcd_dsc_rc_buffer_block_size dsc_rc_buffer_block_size;
+		uint8_t dsc_rc_buffer_size;
+		struct dpcd_dsc_slice_capability1 dsc_slice_capabilities_1;
+		struct dpcd_dsc_line_buffer_bit_depth dsc_line_buffer_bit_depth;
+		struct dpcd_dsc_block_prediction_support dsc_block_prediction_support;
+		struct dpcd_maximum_bits_per_pixel_supported_by_the_decompressor maximum_bits_per_pixel_supported_by_the_decompressor;
+		struct dpcd_dsc_decoder_color_format_capabilities dsc_decoder_color_format_capabilities;
+		struct dpcd_dsc_decoder_color_depth_capabilities dsc_decoder_color_depth_capabilities;
+		struct dpcd_peak_dsc_throughput_dsc_sink peak_dsc_throughput_dsc_sink;
+		uint8_t dsc_maximum_slice_width;
+		struct dpcd_dsc_slice_capabilities_2 dsc_slice_capabilities_2;
+		uint8_t reserved;
+		struct dpcd_bits_per_pixel_increment bits_per_pixel_increment;
+	} fields;
+	uint8_t raw[16];
+};
+
+union dpcd_dsc_branch_decoder_capabilities {
+	struct {
+		uint8_t BRANCH_OVERALL_THROUGHPUT_0;
+		uint8_t BRANCH_OVERALL_THROUGHPUT_1;
+		uint8_t BRANCH_MAX_LINE_WIDTH;
+	} fields;
+	uint8_t raw[3];
+};
+
+struct dpcd_dsc_capabilities {
+	union dpcd_dsc_basic_capabilities dsc_basic_caps;
+	union dpcd_dsc_branch_decoder_capabilities dsc_branch_decoder_caps;
+};
+
+/* These parameters are from PSR capabilities reported by Sink DPCD */
+struct psr_caps {
+	unsigned char psr_version;
+	unsigned int psr_rfb_setup_time;
+	bool psr_exit_link_training_required;
+	unsigned char edp_revision;
+	unsigned char support_ver;
+	bool su_granularity_required;
+	bool y_coordinate_required;
+	uint8_t su_y_granularity;
+	bool alpm_cap;
+	bool standby_support;
+	uint8_t rate_control_caps;
+	unsigned int psr_power_opt_flag;
+};
+
+union dpcd_dprx_feature_enumeration_list_cont_1 {
+	struct {
+		uint8_t ADAPTIVE_SYNC_SDP_SUPPORT:1;
+		uint8_t AS_SDP_FIRST_HALF_LINE_OR_3840_PIXEL_CYCLE_WINDOW_NOT_SUPPORTED: 1;
+		uint8_t RESERVED0: 2;
+		uint8_t VSC_EXT_SDP_VER1_SUPPORT: 1;
+		uint8_t RESERVED1: 3;
+	} bits;
+	uint8_t raw;
+};
+
+struct adaptive_sync_caps {
+	union dpcd_dprx_feature_enumeration_list_cont_1 dp_adap_sync_caps;
+};
+
+/* Length of router topology ID read from DPCD in bytes. */
+#define DPCD_USB4_TOPOLOGY_ID_LEN 5
+
+/* DPCD[0xE000D] DP_TUNNELING_CAPABILITIES SUPPORT register. */
+union dp_tun_cap_support {
+	struct {
+		uint8_t dp_tunneling :1;
+		uint8_t rsvd :5;
+		uint8_t panel_replay_tun_opt :1;
+		uint8_t dpia_bw_alloc :1;
+	} bits;
+	uint8_t raw;
+};
+
+/* DPCD[0xE000E] DP_IN_ADAPTER_INFO register. */
+union dpia_info {
+	struct {
+		uint8_t dpia_num :5;
+		uint8_t rsvd :3;
+	} bits;
+	uint8_t raw;
+};
+
+/* DPCD[0xE0020] USB4_DRIVER_BW_CAPABILITY register. */
+union usb4_driver_bw_cap {
+	struct {
+		uint8_t rsvd :7;
+		uint8_t driver_bw_alloc_support :1;
+	} bits;
+	uint8_t raw;
+};
+
+/* DPCD[0xE0021] DP_IN_ADAPTER_TUNNEL_INFORMATION register. */
+union dpia_tunnel_info {
+	struct {
+		uint8_t group_id :3;
+		uint8_t rsvd :5;
+	} bits;
+	uint8_t raw;
+};
+
+/* DP Tunneling over USB4 */
+struct dpcd_usb4_dp_tunneling_info {
+	union dp_tun_cap_support dp_tun_cap;
+	union dpia_info dpia_info;
+	union usb4_driver_bw_cap driver_bw_cap;
+	union dpia_tunnel_info dpia_tunnel_info;
+	uint8_t usb4_driver_id;
+	uint8_t usb4_topology_id[DPCD_USB4_TOPOLOGY_ID_LEN];
+};
+
+union dp_main_line_channel_coding_cap {
+	struct {
+		uint8_t DP_8b_10b_SUPPORTED	:1;
+		uint8_t DP_128b_132b_SUPPORTED	:1;
+		uint8_t RESERVED		:6;
+	} bits;
+	uint8_t raw;
+};
+
+union dp_main_link_channel_coding_lttpr_cap {
+	struct {
+		uint8_t DP_128b_132b_SUPPORTED	:1;
+		uint8_t RESERVED		:7;
+	} bits;
+	uint8_t raw;
+};
+
+union dp_128b_132b_supported_link_rates {
+	struct {
+		uint8_t UHBR10	:1;
+		uint8_t UHBR20	:1;
+		uint8_t UHBR13_5:1;
+		uint8_t RESERVED:5;
+	} bits;
+	uint8_t raw;
+};
+
+union dp_128b_132b_supported_lttpr_link_rates {
+	struct {
+		uint8_t UHBR10	:1;
+		uint8_t UHBR20	:1;
+		uint8_t UHBR13_5:1;
+		uint8_t RESERVED:5;
+	} bits;
+	uint8_t raw;
+};
+
+union dp_alpm_lttpr_cap {
+	struct {
+		uint8_t AUX_LESS_ALPM_SUPPORTED	:1;
+		uint8_t ASSR_SUPPORTED			:1;
+		uint8_t RESERVED			:6;
+	} bits;
+	uint8_t raw;
+};
+
+union dp_sink_video_fallback_formats {
+	struct {
+		uint8_t dp_1024x768_60Hz_24bpp_support	:1;
+		uint8_t dp_1280x720_60Hz_24bpp_support	:1;
+		uint8_t dp_1920x1080_60Hz_24bpp_support	:1;
+		uint8_t RESERVED			:5;
+	} bits;
+	uint8_t raw;
+};
+
+union dp_receive_port0_cap {
+	struct {
+		uint8_t RESERVED					:1;
+		uint8_t LOCAL_EDID_PRESENT			:1;
+		uint8_t ASSOCIATED_TO_PRECEDING_PORT:1;
+		uint8_t HBLANK_EXPANSION_CAPABLE	:1;
+		uint8_t BUFFER_SIZE_UNIT			:1;
+		uint8_t BUFFER_SIZE_PER_PORT		:1;
+		uint8_t HBLANK_REDUCTION_CAPABLE	:1;
+		uint8_t RESERVED2:1;
+		uint8_t BUFFER_SIZE:8;
+	} bits;
+	uint8_t raw[2];
+};
+
+union dpcd_max_uncompressed_pixel_rate_cap {
+	struct {
+		uint16_t max_uncompressed_pixel_rate_cap	:15;
+		uint16_t valid			:1;
+	} bits;
+	uint8_t raw[2];
+};
+
+union dp_fec_capability1 {
+	struct {
+		uint8_t AGGREGATED_ERROR_COUNTERS_CAPABLE	:1;
+		uint8_t RESERVED				:7;
+	} bits;
+	uint8_t raw;
+};
+
+union dp_cable_id {
+	struct {
+		uint8_t UHBR10_20_CAPABILITY	:2;
+		uint8_t UHBR13_5_CAPABILITY	:1;
+		uint8_t CABLE_TYPE		:3;
+		uint8_t RESERVED		:2;
+	} bits;
+	uint8_t raw;
+};
+
+struct dp_color_depth_caps {
+	uint8_t support_6bpc	:1;
+	uint8_t support_8bpc	:1;
+	uint8_t support_10bpc	:1;
+	uint8_t support_12bpc	:1;
+	uint8_t support_16bpc	:1;
+	uint8_t RESERVED	:3;
+};
+
+struct dp_encoding_format_caps {
+	uint8_t support_rgb	:1;
+	uint8_t support_ycbcr444:1;
+	uint8_t support_ycbcr422:1;
+	uint8_t support_ycbcr420:1;
+	uint8_t RESERVED	:4;
+};
+
+union dp_dfp_cap_ext {
+	struct {
+		uint8_t supported;
+		uint8_t max_pixel_rate_in_mps[2];
+		uint8_t max_video_h_active_width[2];
+		uint8_t max_video_v_active_height[2];
+		struct dp_encoding_format_caps encoding_format_caps;
+		struct dp_color_depth_caps rgb_color_depth_caps;
+		struct dp_color_depth_caps ycbcr444_color_depth_caps;
+		struct dp_color_depth_caps ycbcr422_color_depth_caps;
+		struct dp_color_depth_caps ycbcr420_color_depth_caps;
+	} fields;
+	uint8_t raw[12];
+};
+
+union dp_128b_132b_training_aux_rd_interval {
+	struct {
+		uint8_t VALUE	:7;
+		uint8_t UNIT	:1;
+	} bits;
+	uint8_t raw;
+};
+
+union edp_alpm_caps {
+	struct {
+		uint8_t AUX_WAKE_ALPM_CAP                               :1;
+		uint8_t PM_STATE_2A_SUPPORT                             :1;
+		uint8_t AUX_LESS_ALPM_CAP                               :1;
+		uint8_t AUX_LESS_ALPM_ML_PHY_SLEEP_STATUS_SUPPORTED     :1;
+		uint8_t RESERVED                                        :4;
+	} bits;
+	uint8_t raw;
+};
+
+union edp_psr_dpcd_caps {
+	struct {
+		uint8_t LINK_TRAINING_ON_EXIT_NOT_REQUIRED      :1;
+		uint8_t PSR_SETUP_TIME  :3;
+		uint8_t Y_COORDINATE_REQUIRED   :1;
+		uint8_t SU_GRANULARITY_REQUIRED :1;
+		uint8_t FRAME_SYNC_IS_NOT_NEEDED_FOR_SU :1;
+		uint8_t RESERVED                :1;
+	} bits;
+	uint8_t raw;
+};
+
+struct edp_psr_info {
+	uint8_t psr_version;
+	union edp_psr_dpcd_caps psr_dpcd_caps;
+	uint8_t psr2_su_y_granularity_cap;
+	uint8_t force_psrsu_cap;
+};
+
+struct replay_info {
+	uint8_t pixel_deviation_per_line;
+	uint8_t max_deviation_line;
+};
+
+struct dprx_states {
+	bool cable_id_written;
+};
+
+enum dpcd_downstream_port_max_bpc {
+	DOWN_STREAM_MAX_8BPC = 0,
+	DOWN_STREAM_MAX_10BPC,
+	DOWN_STREAM_MAX_12BPC,
+	DOWN_STREAM_MAX_16BPC
+};
+
+enum link_training_offset {
+	DPRX                = 0,
+	LTTPR_PHY_REPEATER1 = 1,
+	LTTPR_PHY_REPEATER2 = 2,
+	LTTPR_PHY_REPEATER3 = 3,
+	LTTPR_PHY_REPEATER4 = 4,
+	LTTPR_PHY_REPEATER5 = 5,
+	LTTPR_PHY_REPEATER6 = 6,
+	LTTPR_PHY_REPEATER7 = 7,
+	LTTPR_PHY_REPEATER8 = 8
+};
+
+#define MAX_REPEATER_CNT 8
+
+struct dc_lttpr_caps {
+	union dpcd_rev revision;
+	uint8_t mode;
+	uint8_t max_lane_count;
+	uint8_t max_link_rate;
+	uint8_t phy_repeater_cnt;
+	uint8_t max_ext_timeout;
+	union dp_main_link_channel_coding_lttpr_cap main_link_channel_coding;
+	union dp_128b_132b_supported_lttpr_link_rates supported_128b_132b_rates;
+	union dp_alpm_lttpr_cap alpm;
+	uint8_t aux_rd_interval[MAX_REPEATER_CNT - 1];
+	uint8_t lttpr_ieee_oui[3]; // Always read from closest LTTPR to host
+	uint8_t lttpr_device_id[6]; // Always read from closest LTTPR to host
+};
+
+struct dc_dongle_dfp_cap_ext {
+	bool supported;
+	uint16_t max_pixel_rate_in_mps;
+	uint16_t max_video_h_active_width;
+	uint16_t max_video_v_active_height;
+	struct dp_encoding_format_caps encoding_format_caps;
+	struct dp_color_depth_caps rgb_color_depth_caps;
+	struct dp_color_depth_caps ycbcr444_color_depth_caps;
+	struct dp_color_depth_caps ycbcr422_color_depth_caps;
+	struct dp_color_depth_caps ycbcr420_color_depth_caps;
+};
+
+struct dc_dongle_caps {
+	/* dongle type (DP converter, CV smart dongle) */
+	enum display_dongle_type dongle_type;
+	bool extendedCapValid;
+	/* If dongle_type == DISPLAY_DONGLE_DP_HDMI_CONVERTER,
+	indicates 'Frame Sequential-to-lllFrame Pack' conversion capability.*/
+	bool is_dp_hdmi_s3d_converter;
+	bool is_dp_hdmi_ycbcr422_pass_through;
+	bool is_dp_hdmi_ycbcr420_pass_through;
+	bool is_dp_hdmi_ycbcr422_converter;
+	bool is_dp_hdmi_ycbcr420_converter;
+	uint32_t dp_hdmi_max_bpc;
+	uint32_t dp_hdmi_max_pixel_clk_in_khz;
+	uint32_t dp_hdmi_frl_max_link_bw_in_kbps;
+	uint32_t dp_hdmi_regulated_autonomous_mode_support;
+	struct dc_dongle_dfp_cap_ext dfp_cap_ext;
+};
+
+struct dpcd_caps {
+	union dpcd_rev dpcd_rev;
+	union max_lane_count max_ln_count;
+	union max_down_spread max_down_spread;
+	union dprx_feature dprx_feature;
+
+	/* valid only for eDP v1.4 or higher*/
+	uint8_t edp_supported_link_rates_count;
+	enum dc_link_rate edp_supported_link_rates[8];
+
+	/* dongle type (DP converter, CV smart dongle) */
+	enum display_dongle_type dongle_type;
+	bool is_dongle_type_one;
+	/* branch device or sink device */
+	bool is_branch_dev;
+	/* Dongle's downstream count. */
+	union sink_count sink_count;
+	bool is_mst_capable;
+	/* If dongle_type == DISPLAY_DONGLE_DP_HDMI_CONVERTER,
+	indicates 'Frame Sequential-to-lllFrame Pack' conversion capability.*/
+	struct dc_dongle_caps dongle_caps;
+
+	uint32_t sink_dev_id;
+	int8_t sink_dev_id_str[6];
+	int8_t sink_hw_revision;
+	int8_t sink_fw_revision[2];
+
+	uint32_t branch_dev_id;
+	int8_t branch_dev_name[6];
+	int8_t branch_hw_revision;
+	int8_t branch_fw_revision[2];
+	int8_t branch_vendor_specific_data[4];
+
+	bool allow_invalid_MSA_timing_param;
+	bool panel_mode_edp;
+	bool dpcd_display_control_capable;
+	bool ext_receiver_cap_field_present;
+	bool set_power_state_capable_edp;
+	bool dynamic_backlight_capable_edp;
+	union dpcd_fec_capability fec_cap;
+	struct dpcd_dsc_capabilities dsc_caps;
+	struct dc_lttpr_caps lttpr_caps;
+	struct adaptive_sync_caps adaptive_sync_caps;
+	struct dpcd_usb4_dp_tunneling_info usb4_dp_tun_info;
+	union dpcd_max_uncompressed_pixel_rate_cap max_uncompressed_pixel_rate_cap;
+
+	union dp_128b_132b_supported_link_rates dp_128b_132b_supported_link_rates;
+	union dp_main_line_channel_coding_cap channel_coding_cap;
+	union dp_sink_video_fallback_formats fallback_formats;
+	union dp_fec_capability1 fec_cap1;
+	bool panel_luminance_control;
+	union dp_cable_id cable_id;
+	uint8_t edp_rev;
+	union edp_alpm_caps alpm_caps;
+	struct edp_psr_info psr_info;
+
+	struct replay_info pr_info;
+	uint16_t edp_oled_emission_rate;
+	union dp_receive_port0_cap receive_port0_cap;
+	/* Indicates the number of SST links supported by MSO (Multi-Stream Output) */
+	uint8_t mso_cap_sst_links_supported;
+	uint8_t dp_edp_general_cap_2;
+};
+
+union dpcd_sink_ext_caps {
+	struct {
+		/* 0 - Sink supports backlight adjust via PWM during SDR/HDR mode
+		 * 1 - Sink supports backlight adjust via AUX during SDR/HDR mode.
+		 */
+		uint8_t sdr_aux_backlight_control : 1;
+		uint8_t hdr_aux_backlight_control : 1;
+		uint8_t reserved_1 : 2;
+		uint8_t oled : 1;
+		uint8_t reserved_2 : 1;
+		uint8_t miniled : 1;
+		uint8_t emission_output : 1;
+	} bits;
+	uint8_t raw;
+};
+
+enum dc_link_fec_state {
+	dc_link_fec_not_ready,
+	dc_link_fec_ready,
+	dc_link_fec_enabled
+};
+
+union dpcd_psr_configuration {
+	struct {
+		unsigned char ENABLE                    : 1;
+		unsigned char TRANSMITTER_ACTIVE_IN_PSR : 1;
+		unsigned char CRC_VERIFICATION          : 1;
+		unsigned char FRAME_CAPTURE_INDICATION  : 1;
+		/* For eDP 1.4, PSR v2*/
+		unsigned char LINE_CAPTURE_INDICATION   : 1;
+		/* For eDP 1.4, PSR v2*/
+		unsigned char IRQ_HPD_WITH_CRC_ERROR    : 1;
+		unsigned char ENABLE_PSR2               : 1;
+		unsigned char EARLY_TRANSPORT_ENABLE    : 1;
+	} bits;
+	unsigned char raw;
+};
+
+union replay_enable_and_configuration {
+	struct {
+		unsigned char FREESYNC_PANEL_REPLAY_MODE              :1;
+		unsigned char TIMING_DESYNC_ERROR_VERIFICATION        :1;
+		unsigned char STATE_TRANSITION_ERROR_DETECTION        :1;
+		unsigned char RESERVED                                :5;
+	} bits;
+	unsigned char raw;
+};
+
+union dpcd_replay_configuration {
+	struct {
+		unsigned char STATE_TRANSITION_ERROR_STATUS    : 1;
+		unsigned char DESYNC_ERROR_STATUS              : 1;
+		unsigned char SINK_DEVICE_REPLAY_STATUS        : 3;
+		unsigned char SINK_FRAME_LOCKED                : 2;
+		unsigned char RESERVED                         : 1;
+	} bits;
+	unsigned char raw;
+};
+
+union dpcd_alpm_configuration {
+	struct {
+		unsigned char ENABLE                    : 1;
+		unsigned char IRQ_HPD_ENABLE            : 1;
+		unsigned char ALPM_MODE_SEL             : 1;
+		unsigned char ACDS_PERIOD_DURATION      : 1;
+		unsigned char RESERVED                  : 4;
+	} bits;
+	unsigned char raw;
+};
+
+union dpcd_sink_active_vtotal_control_mode {
+	struct {
+		unsigned char ENABLE                    : 1;
+		unsigned char RESERVED                  : 7;
+	} bits;
+	unsigned char raw;
+};
+
+union psr_error_status {
+	struct {
+		unsigned char LINK_CRC_ERROR        :1;
+		unsigned char RFB_STORAGE_ERROR     :1;
+		unsigned char VSC_SDP_ERROR         :1;
+		unsigned char RESERVED              :5;
+	} bits;
+	unsigned char raw;
+};
+
+union psr_sink_psr_status {
+	struct {
+	unsigned char SINK_SELF_REFRESH_STATUS  :3;
+	unsigned char RESERVED                  :5;
+	} bits;
+	unsigned char raw;
+};
+
+struct edp_trace_power_timestamps {
+	uint64_t poweroff;
+	uint64_t poweron;
+};
+
+struct dp_trace_lt_counts {
+	unsigned int total;
+	unsigned int fail;
+};
+
+enum link_training_result {
+	LINK_TRAINING_SUCCESS,
+	LINK_TRAINING_CR_FAIL_LANE0,
+	LINK_TRAINING_CR_FAIL_LANE1,
+	LINK_TRAINING_CR_FAIL_LANE23,
+	/* CR DONE bit is cleared during EQ step */
+	LINK_TRAINING_EQ_FAIL_CR,
+	/* CR DONE bit is cleared but LANE0_CR_DONE is set during EQ step */
+	LINK_TRAINING_EQ_FAIL_CR_PARTIAL,
+	/* other failure during EQ step */
+	LINK_TRAINING_EQ_FAIL_EQ,
+	LINK_TRAINING_LQA_FAIL,
+	/* one of the CR,EQ or symbol lock is dropped */
+	LINK_TRAINING_LINK_LOSS,
+	/* Abort link training (because sink unplugged) */
+	LINK_TRAINING_ABORT,
+	DP_128b_132b_LT_FAILED,
+	DP_128b_132b_MAX_LOOP_COUNT_REACHED,
+	DP_128b_132b_CHANNEL_EQ_DONE_TIMEOUT,
+	DP_128b_132b_CDS_DONE_TIMEOUT,
+};
+
+struct dp_trace_lt {
+	struct dp_trace_lt_counts counts;
+	struct dp_trace_timestamps {
+		unsigned long long start;
+		unsigned long long end;
+	} timestamps;
+	enum link_training_result result;
+	bool is_logged;
+};
+
+struct dp_trace {
+	struct dp_trace_lt detect_lt_trace;
+	struct dp_trace_lt commit_lt_trace;
+	unsigned int link_loss_count;
+	bool is_initialized;
+	struct edp_trace_power_timestamps edp_trace_power_timestamps;
+};
+
+/* TODO - This is a temporary location for any new DPCD definitions.
+ * We should move these to drm_dp header.
+ */
+#ifndef DP_LINK_SQUARE_PATTERN
+#define DP_LINK_SQUARE_PATTERN				0x10F
+#endif
+#ifndef DP_CABLE_ATTRIBUTES_UPDATED_BY_DPRX
+#define DP_CABLE_ATTRIBUTES_UPDATED_BY_DPRX		0x2217
+#endif
+#ifndef DP_CABLE_ATTRIBUTES_UPDATED_BY_DPTX
+#define DP_CABLE_ATTRIBUTES_UPDATED_BY_DPTX		0x110
+#endif
+#ifndef DPCD_MAX_UNCOMPRESSED_PIXEL_RATE_CAP
+#define DPCD_MAX_UNCOMPRESSED_PIXEL_RATE_CAP    0x221c
+#endif
+#ifndef DP_LTTPR_ALPM_CAPABILITIES
+#define DP_LTTPR_ALPM_CAPABILITIES              0xF0009
+#endif
+#ifndef DP_REGULATED_AUTONOMOUS_MODE_SUPPORTED_AND_HDMI_LINK_TRAINING_STATUS
+#define DP_REGULATED_AUTONOMOUS_MODE_SUPPORTED_AND_HDMI_LINK_TRAINING_STATUS	0x303C
+#endif
+#ifndef DP_REPEATER_CONFIGURATION_AND_STATUS_SIZE
+#define DP_REPEATER_CONFIGURATION_AND_STATUS_SIZE	0x50
+#endif
+#ifndef DP_TUNNELING_IRQ
+#define DP_TUNNELING_IRQ				(1 << 5)
+#endif
+#ifndef DP_BRANCH_VENDOR_SPECIFIC_START
+#define DP_BRANCH_VENDOR_SPECIFIC_START     0x50C
+#endif
+#ifndef DP_LTTPR_IEEE_OUI
+#define DP_LTTPR_IEEE_OUI 0xF003D
+#endif
+#ifndef DP_LTTPR_DEVICE_ID
+#define DP_LTTPR_DEVICE_ID 0xF0040
+#endif
+/** USB4 DPCD BW Allocation Registers Chapter 10.7 **/
+#ifndef DP_TUNNELING_CAPABILITIES
+#define DP_TUNNELING_CAPABILITIES			0xE000D /* 1.4a */
+#endif
+#ifndef USB4_DRIVER_ID
+#define USB4_DRIVER_ID					0xE000F /* 1.4a */
+#endif
+#ifndef USB4_DRIVER_BW_CAPABILITY
+#define USB4_DRIVER_BW_CAPABILITY			0xE0020 /* 1.4a */
+#endif
+#ifndef DP_IN_ADAPTER_TUNNEL_INFO
+#define DP_IN_ADAPTER_TUNNEL_INFO			0xE0021 /* 1.4a */
+#endif
+#ifndef DP_BW_GRANULALITY
+#define DP_BW_GRANULALITY				0xE0022 /* 1.4a */
+#endif
+#ifndef ESTIMATED_BW
+#define ESTIMATED_BW					0xE0023 /* 1.4a */
+#endif
+#ifndef ALLOCATED_BW
+#define ALLOCATED_BW					0xE0024 /* 1.4a */
+#endif
+#ifndef DP_TUNNELING_STATUS
+#define DP_TUNNELING_STATUS				0xE0025 /* 1.4a */
+#endif
+#ifndef DP_TUNNELING_MAX_LINK_RATE
+#define DP_TUNNELING_MAX_LINK_RATE			0xE0028 /* 1.4a */
+#endif
+#ifndef DP_TUNNELING_MAX_LANE_COUNT
+#define DP_TUNNELING_MAX_LANE_COUNT			0xE0029 /* 1.4a */
+#endif
+#ifndef DPTX_BW_ALLOCATION_MODE_CONTROL
+#define DPTX_BW_ALLOCATION_MODE_CONTROL			0xE0030 /* 1.4a */
+#endif
+#ifndef REQUESTED_BW
+#define REQUESTED_BW					0xE0031 /* 1.4a */
+#endif
+# ifndef DP_TUNNELING_BW_ALLOC_BITS_MASK
+# define DP_TUNNELING_BW_ALLOC_BITS_MASK		(0x0F << 0)
+# endif
+# ifndef DP_TUNNELING_BW_REQUEST_FAILED
+# define DP_TUNNELING_BW_REQUEST_FAILED			(1 << 0)
+# endif
+# ifndef DP_TUNNELING_BW_REQUEST_SUCCEEDED
+# define DP_TUNNELING_BW_REQUEST_SUCCEEDED		(1 << 1)
+# endif
+# ifndef DP_TUNNELING_ESTIMATED_BW_CHANGED
+# define DP_TUNNELING_ESTIMATED_BW_CHANGED		(1 << 2)
+# endif
+# ifndef DP_TUNNELING_BW_ALLOC_CAP_CHANGED
+# define DP_TUNNELING_BW_ALLOC_CAP_CHANGED		(1 << 3)
+# endif
+# ifndef DPTX_BW_ALLOC_UNMASK_IRQ
+# define DPTX_BW_ALLOC_UNMASK_IRQ			(1 << 6)
+# endif
+# ifndef DPTX_BW_ALLOC_MODE_ENABLE
+# define DPTX_BW_ALLOC_MODE_ENABLE			(1 << 7)
+# endif
+
+#endif /* DC_DP_TYPES_H */
diff --git a/drivers/gpu/drm/amd/display/dc/dc_dsc.h b/drivers/gpu/drm/amd/display/dc/dc_dsc.h
new file mode 100644
index 000000000000..9d18f1c08079
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_dsc.h
@@ -0,0 +1,120 @@
+#ifndef DC_DSC_H_
+#define DC_DSC_H_
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Author: AMD
+ */
+
+/* put it here temporarily until linux has the new addresses official defined */
+/* DP Extended DSC Capabilities */
+#define DP_DSC_BRANCH_OVERALL_THROUGHPUT_0  0x0a0   /* DP 1.4a SCR */
+#define DP_DSC_BRANCH_OVERALL_THROUGHPUT_1  0x0a1
+#define DP_DSC_BRANCH_MAX_LINE_WIDTH        0x0a2
+#include "dc_types.h"
+
+struct dc_dsc_bw_range {
+	uint32_t min_kbps; /* Bandwidth if min_target_bpp_x16 is used */
+	uint32_t min_target_bpp_x16;
+	uint32_t max_kbps; /* Bandwidth if max_target_bpp_x16 is used */
+	uint32_t max_target_bpp_x16;
+	uint32_t stream_kbps; /* Uncompressed stream bandwidth */
+};
+
+struct display_stream_compressor {
+	const struct dsc_funcs *funcs;
+	struct dc_context *ctx;
+	int inst;
+};
+
+struct dc_dsc_policy {
+	bool use_min_slices_h;
+	int max_slices_h; // Maximum available if 0
+	int min_slice_height; // Must not be less than 8
+	uint32_t max_target_bpp;
+	uint32_t min_target_bpp;
+	bool enable_dsc_when_not_needed;
+};
+
+struct dc_dsc_config_options {
+	uint32_t dsc_min_slice_height_override;
+	uint32_t max_target_bpp_limit_override_x16;
+	uint32_t slice_height_granularity;
+	uint32_t dsc_force_odm_hslice_override;
+	bool force_dsc_when_not_needed;
+};
+
+bool dc_dsc_parse_dsc_dpcd(const struct dc *dc,
+		const uint8_t *dpcd_dsc_basic_data,
+		const uint8_t *dpcd_dsc_ext_data,
+		struct dsc_dec_dpcd_caps *dsc_sink_caps);
+
+bool dc_dsc_compute_bandwidth_range(
+		const struct display_stream_compressor *dsc,
+		uint32_t dsc_min_slice_height_override,
+		uint32_t min_bpp_x16,
+		uint32_t max_bpp_x16,
+		const struct dsc_dec_dpcd_caps *dsc_sink_caps,
+		const struct dc_crtc_timing *timing,
+		const enum dc_link_encoding_format link_encoding,
+		struct dc_dsc_bw_range *range);
+
+bool dc_dsc_compute_config(
+		const struct display_stream_compressor *dsc,
+		const struct dsc_dec_dpcd_caps *dsc_sink_caps,
+		const struct dc_dsc_config_options *options,
+		uint32_t target_bandwidth_kbps,
+		const struct dc_crtc_timing *timing,
+		const enum dc_link_encoding_format link_encoding,
+		struct dc_dsc_config *dsc_cfg);
+
+uint32_t dc_dsc_stream_bandwidth_in_kbps(const struct dc_crtc_timing *timing,
+		uint32_t bpp_x16, uint32_t num_slices_h, bool is_dp);
+
+uint32_t dc_dsc_stream_bandwidth_overhead_in_kbps(
+		const struct dc_crtc_timing *timing,
+		const int num_slices_h,
+		const bool is_dp);
+
+void dc_dsc_dump_decoder_caps(const struct display_stream_compressor *dsc,
+			      const struct dsc_dec_dpcd_caps *dsc_sink_caps);
+void dc_dsc_dump_encoder_caps(const struct display_stream_compressor *dsc,
+			      const struct dc_crtc_timing *timing);
+
+/* TODO - Hardware/specs limitation should be owned by dc dsc and returned to DM,
+ * and DM can choose to OVERRIDE the limitation on CASE BY CASE basis.
+ * Hardware/specs limitation should not be writable by DM.
+ * It should be decoupled from DM specific policy and named differently.
+ */
+void dc_dsc_get_policy_for_timing(const struct dc_crtc_timing *timing,
+		uint32_t max_target_bpp_limit_override_x16,
+		struct dc_dsc_policy *policy,
+		const enum dc_link_encoding_format link_encoding);
+
+void dc_dsc_policy_set_max_target_bpp_limit(uint32_t limit);
+
+void dc_dsc_policy_set_enable_dsc_when_not_needed(bool enable);
+
+void dc_dsc_policy_set_disable_dsc_stream_overhead(bool disable);
+
+void dc_dsc_get_default_config_option(const struct dc *dc, struct dc_dsc_config_options *options);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dc_edid_parser.c b/drivers/gpu/drm/amd/display/dc/dc_edid_parser.c
new file mode 100644
index 000000000000..0db5b49e9d5e
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_edid_parser.c
@@ -0,0 +1,80 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dce/dce_dmcu.h"
+#include "dc_edid_parser.h"
+
+bool dc_edid_parser_send_cea(struct dc *dc,
+		int offset,
+		int total_length,
+		uint8_t *data,
+		int length)
+{
+	struct dmcu *dmcu = dc->res_pool->dmcu;
+
+	if (dmcu &&
+	    dmcu->funcs->is_dmcu_initialized(dmcu) &&
+	    dmcu->funcs->send_edid_cea) {
+		return dmcu->funcs->send_edid_cea(dmcu,
+				offset,
+				total_length,
+				data,
+				length);
+	}
+
+	return false;
+}
+
+bool dc_edid_parser_recv_cea_ack(struct dc *dc, int *offset)
+{
+	struct dmcu *dmcu = dc->res_pool->dmcu;
+
+	if (dmcu &&
+	    dmcu->funcs->is_dmcu_initialized(dmcu) &&
+	    dmcu->funcs->recv_edid_cea_ack) {
+		return dmcu->funcs->recv_edid_cea_ack(dmcu, offset);
+	}
+
+	return false;
+}
+
+bool dc_edid_parser_recv_amd_vsdb(struct dc *dc,
+		int *version,
+		int *min_frame_rate,
+		int *max_frame_rate)
+{
+	struct dmcu *dmcu = dc->res_pool->dmcu;
+
+	if (dmcu &&
+	    dmcu->funcs->is_dmcu_initialized(dmcu) &&
+	    dmcu->funcs->recv_amd_vsdb) {
+		return dmcu->funcs->recv_amd_vsdb(dmcu,
+				version,
+				min_frame_rate,
+				max_frame_rate);
+	}
+
+	return false;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dc_edid_parser.h b/drivers/gpu/drm/amd/display/dc/dc_edid_parser.h
new file mode 100644
index 000000000000..da67ec06f0a2
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_edid_parser.h
@@ -0,0 +1,44 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef _DC_EDID_PARSER_H_
+#define _DC_EDID_PARSER_H_
+
+#include "core_types.h"
+
+bool dc_edid_parser_send_cea(struct dc *dc,
+		int offset,
+		int total_length,
+		uint8_t *data,
+		int length);
+
+bool dc_edid_parser_recv_cea_ack(struct dc *dc, int *offset);
+
+bool dc_edid_parser_recv_amd_vsdb(struct dc *dc,
+		int *version,
+		int *min_frame_rate,
+		int *max_frame_rate);
+
+#endif /* _DC_EDID_PARSER_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/dc_fused_io.c b/drivers/gpu/drm/amd/display/dc/dc_fused_io.c
new file mode 100644
index 000000000000..fee69642fb93
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_fused_io.c
@@ -0,0 +1,148 @@
+// SPDX-License-Identifier: MIT
+//
+// Copyright 2025 Advanced Micro Devices, Inc.
+
+#include "dc_fused_io.h"
+
+#include "dm_helpers.h"
+#include "gpio.h"
+
+static bool op_i2c_convert(
+		union dmub_rb_cmd *cmd,
+		const struct mod_hdcp_atomic_op_i2c *op,
+		enum dmub_cmd_fused_request_type type,
+		uint32_t ddc_line,
+		bool over_aux
+)
+{
+	struct dmub_cmd_fused_request *req = &cmd->fused_io.request;
+	struct dmub_cmd_fused_request_location_i2c *loc = &req->u.i2c;
+
+	if (!op || op->size > sizeof(req->buffer))
+		return false;
+
+	req->type = type;
+	loc->is_aux = false;
+	loc->ddc_line = ddc_line;
+	loc->over_aux = over_aux;
+	loc->address = op->address;
+	loc->offset = op->offset;
+	loc->length = op->size;
+	memcpy(req->buffer, op->data, op->size);
+
+	return true;
+}
+
+static bool op_aux_convert(
+		union dmub_rb_cmd *cmd,
+		const struct mod_hdcp_atomic_op_aux *op,
+		enum dmub_cmd_fused_request_type type,
+		uint32_t ddc_line
+)
+{
+	struct dmub_cmd_fused_request *req = &cmd->fused_io.request;
+	struct dmub_cmd_fused_request_location_aux *loc = &req->u.aux;
+
+	if (!op || op->size > sizeof(req->buffer))
+		return false;
+
+	req->type = type;
+	loc->is_aux = true;
+	loc->ddc_line = ddc_line;
+	loc->address = op->address;
+	loc->length = op->size;
+	memcpy(req->buffer, op->data, op->size);
+
+	return true;
+}
+
+static bool atomic_write_poll_read(
+		struct dc_link *link,
+		union dmub_rb_cmd commands[3],
+		uint32_t poll_timeout_us,
+		uint8_t poll_mask_msb
+)
+{
+	const uint8_t count = 3;
+	const uint32_t timeout_per_request_us = 10000;
+	const uint32_t timeout_per_aux_transaction_us = 10000;
+	uint64_t timeout_us = 0;
+
+	commands[1].fused_io.request.poll_mask_msb = poll_mask_msb;
+	commands[1].fused_io.request.timeout_us = poll_timeout_us;
+
+	for (uint8_t i = 0; i < count; i++) {
+		struct dmub_rb_cmd_fused_io *io = &commands[i].fused_io;
+
+		io->header.type = DMUB_CMD__FUSED_IO;
+		io->header.sub_type = DMUB_CMD__FUSED_IO_EXECUTE;
+		io->header.multi_cmd_pending = i != count - 1;
+		io->header.payload_bytes = sizeof(commands[i].fused_io) - sizeof(io->header);
+
+		timeout_us += timeout_per_request_us + io->request.timeout_us;
+		if (!io->request.timeout_us && io->request.u.aux.is_aux)
+			timeout_us += timeout_per_aux_transaction_us * (io->request.u.aux.length / 16);
+	}
+
+	if (!dm_helpers_execute_fused_io(link->ctx, link, commands, count, timeout_us))
+		return false;
+
+	return commands[0].fused_io.request.status == FUSED_REQUEST_STATUS_SUCCESS;
+}
+
+bool dm_atomic_write_poll_read_i2c(
+		struct dc_link *link,
+		const struct mod_hdcp_atomic_op_i2c *write,
+		const struct mod_hdcp_atomic_op_i2c *poll,
+		struct mod_hdcp_atomic_op_i2c *read,
+		uint32_t poll_timeout_us,
+		uint8_t poll_mask_msb
+)
+{
+	if (!link)
+		return false;
+
+	const bool over_aux = false;
+	const uint32_t ddc_line = link->ddc->ddc_pin->pin_data->en;
+
+	union dmub_rb_cmd commands[3] = { 0 };
+	const bool converted = op_i2c_convert(&commands[0], write, FUSED_REQUEST_WRITE, ddc_line, over_aux)
+			&& op_i2c_convert(&commands[1], poll, FUSED_REQUEST_POLL, ddc_line, over_aux)
+			&& op_i2c_convert(&commands[2], read, FUSED_REQUEST_READ, ddc_line, over_aux);
+
+	if (!converted)
+		return false;
+
+	const bool result = atomic_write_poll_read(link, commands, poll_timeout_us, poll_mask_msb);
+
+	memcpy(read->data, commands[0].fused_io.request.buffer, read->size);
+	return result;
+}
+
+bool dm_atomic_write_poll_read_aux(
+		struct dc_link *link,
+		const struct mod_hdcp_atomic_op_aux *write,
+		const struct mod_hdcp_atomic_op_aux *poll,
+		struct mod_hdcp_atomic_op_aux *read,
+		uint32_t poll_timeout_us,
+		uint8_t poll_mask_msb
+)
+{
+	if (!link)
+		return false;
+
+	const uint32_t ddc_line = link->ddc->ddc_pin->pin_data->en;
+	union dmub_rb_cmd commands[3] = { 0 };
+	const bool converted = op_aux_convert(&commands[0], write, FUSED_REQUEST_WRITE, ddc_line)
+			&& op_aux_convert(&commands[1], poll, FUSED_REQUEST_POLL, ddc_line)
+			&& op_aux_convert(&commands[2], read, FUSED_REQUEST_READ, ddc_line);
+
+	if (!converted)
+		return false;
+
+	const bool result = atomic_write_poll_read(link, commands, poll_timeout_us, poll_mask_msb);
+
+	memcpy(read->data, commands[0].fused_io.request.buffer, read->size);
+	return result;
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/dc_fused_io.h b/drivers/gpu/drm/amd/display/dc/dc_fused_io.h
new file mode 100644
index 000000000000..c74917240985
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_fused_io.h
@@ -0,0 +1,31 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2025 Advanced Micro Devices, Inc.
+ */
+
+#ifndef __DC_FUSED_IO_H__
+#define __DC_FUSED_IO_H__
+
+#include "dc.h"
+#include "mod_hdcp.h"
+
+bool dm_atomic_write_poll_read_i2c(
+		struct dc_link *link,
+		const struct mod_hdcp_atomic_op_i2c *write,
+		const struct mod_hdcp_atomic_op_i2c *poll,
+		struct mod_hdcp_atomic_op_i2c *read,
+		uint32_t poll_timeout_us,
+		uint8_t poll_mask_msb
+);
+
+bool dm_atomic_write_poll_read_aux(
+		struct dc_link *link,
+		const struct mod_hdcp_atomic_op_aux *write,
+		const struct mod_hdcp_atomic_op_aux *poll,
+		struct mod_hdcp_atomic_op_aux *read,
+		uint32_t poll_timeout_us,
+		uint8_t poll_mask_msb
+);
+
+#endif  // __DC_FUSED_IO_H__
+
diff --git a/drivers/gpu/drm/amd/display/dc/dc_hdmi_types.h b/drivers/gpu/drm/amd/display/dc/dc_hdmi_types.h
new file mode 100644
index 000000000000..b015e80672ec
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_hdmi_types.h
@@ -0,0 +1,133 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DC_HDMI_TYPES_H
+#define DC_HDMI_TYPES_H
+
+#include "os_types.h"
+
+/* Address range from 0x00 to 0x1F.*/
+#define DP_ADAPTOR_TYPE2_SIZE 0x20
+#define DP_ADAPTOR_TYPE2_REG_ID 0x10
+#define DP_ADAPTOR_TYPE2_REG_MAX_TMDS_CLK 0x1D
+/* Identifies adaptor as Dual-mode adaptor */
+#define DP_ADAPTOR_TYPE2_ID 0xA0
+/* MHz*/
+#define DP_ADAPTOR_TYPE2_MAX_TMDS_CLK 600
+/* MHz*/
+#define DP_ADAPTOR_TYPE2_MIN_TMDS_CLK 25
+/* kHZ*/
+#define DP_ADAPTOR_DVI_MAX_TMDS_CLK 165000
+/* kHZ*/
+#define DP_ADAPTOR_HDMI_SAFE_MAX_TMDS_CLK 165000
+
+struct dp_hdmi_dongle_signature_data {
+	int8_t id[15];/* "DP-HDMI ADAPTOR"*/
+	uint8_t eot;/* end of transmition '\x4' */
+};
+
+/* DP-HDMI dongle slave address for retrieving dongle signature*/
+#define DP_HDMI_DONGLE_ADDRESS 0x40
+#define DP_HDMI_DONGLE_SIGNATURE_EOT 0x04
+
+
+/* SCDC Address defines (HDMI 2.0)*/
+#define HDMI_SCDC_WRITE_UPDATE_0_ARRAY 3
+#define HDMI_SCDC_ADDRESS  0x54
+#define HDMI_SCDC_SINK_VERSION 0x01
+#define HDMI_SCDC_SOURCE_VERSION 0x02
+#define HDMI_SCDC_UPDATE_0 0x10
+#define HDMI_SCDC_TMDS_CONFIG 0x20
+#define HDMI_SCDC_SCRAMBLER_STATUS 0x21
+#define HDMI_SCDC_CONFIG_0 0x30
+#define HDMI_SCDC_CONFIG_1 0x31
+#define HDMI_SCDC_SOURCE_TEST_REQ 0x35
+#define HDMI_SCDC_STATUS_FLAGS 0x40
+#define HDMI_SCDC_ERR_DETECT 0x50
+#define HDMI_SCDC_TEST_CONFIG 0xC0
+
+#define HDMI_SCDC_MANUFACTURER_OUI 0xD0
+#define HDMI_SCDC_DEVICE_ID 0xDB
+
+union hdmi_scdc_update_read_data {
+	uint8_t byte[2];
+	struct {
+		uint8_t STATUS_UPDATE:1;
+		uint8_t CED_UPDATE:1;
+		uint8_t RR_TEST:1;
+		uint8_t RESERVED:5;
+		uint8_t RESERVED2:8;
+	} fields;
+};
+
+union hdmi_scdc_status_flags_data {
+	uint8_t byte;
+	struct {
+		uint8_t CLOCK_DETECTED:1;
+		uint8_t CH0_LOCKED:1;
+		uint8_t CH1_LOCKED:1;
+		uint8_t CH2_LOCKED:1;
+		uint8_t RESERVED:4;
+	} fields;
+};
+
+union hdmi_scdc_ced_data {
+	uint8_t byte[11];
+	struct {
+		uint8_t CH0_8LOW:8;
+		uint8_t CH0_7HIGH:7;
+		uint8_t CH0_VALID:1;
+		uint8_t CH1_8LOW:8;
+		uint8_t CH1_7HIGH:7;
+		uint8_t CH1_VALID:1;
+		uint8_t CH2_8LOW:8;
+		uint8_t CH2_7HIGH:7;
+		uint8_t CH2_VALID:1;
+		uint8_t CHECKSUM:8;
+		uint8_t RESERVED:8;
+		uint8_t RESERVED2:8;
+		uint8_t RESERVED3:8;
+		uint8_t RESERVED4:4;
+	} fields;
+};
+
+union hdmi_scdc_manufacturer_OUI_data {
+	uint8_t byte[3];
+	struct {
+		uint8_t Manufacturer_OUI_1:8;
+		uint8_t Manufacturer_OUI_2:8;
+		uint8_t Manufacturer_OUI_3:8;
+	} fields;
+};
+
+union hdmi_scdc_device_id_data {
+	uint8_t byte;
+	struct {
+		uint8_t Hardware_Minor_Rev:4;
+		uint8_t Hardware_Major_Rev:4;
+	} fields;
+};
+
+#endif /* DC_HDMI_TYPES_H */
diff --git a/drivers/gpu/drm/amd/display/dc/dc_helper.c b/drivers/gpu/drm/amd/display/dc/dc_helper.c
new file mode 100644
index 000000000000..5a365bd19933
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_helper.c
@@ -0,0 +1,762 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+/*
+ * dc_helper.c
+ *
+ *  Created on: Aug 30, 2016
+ *      Author: agrodzov
+ */
+
+#include <linux/delay.h>
+#include <linux/stdarg.h>
+
+#include "dm_services.h"
+
+#include "dc.h"
+#include "dc_dmub_srv.h"
+#include "reg_helper.h"
+
+#define DC_LOGGER \
+	ctx->logger
+
+static inline void submit_dmub_read_modify_write(
+	struct dc_reg_helper_state *offload,
+	const struct dc_context *ctx)
+{
+	struct dmub_rb_cmd_read_modify_write *cmd_buf = &offload->cmd_data.read_modify_write;
+
+	offload->should_burst_write =
+			(offload->same_addr_count == (DMUB_READ_MODIFY_WRITE_SEQ__MAX - 1));
+	cmd_buf->header.payload_bytes =
+			sizeof(struct dmub_cmd_read_modify_write_sequence) * offload->reg_seq_count;
+
+	dc_wake_and_execute_dmub_cmd(ctx, &offload->cmd_data, DM_DMUB_WAIT_TYPE_NO_WAIT);
+
+	memset(cmd_buf, 0, sizeof(*cmd_buf));
+
+	offload->reg_seq_count = 0;
+	offload->same_addr_count = 0;
+}
+
+static inline void submit_dmub_burst_write(
+	struct dc_reg_helper_state *offload,
+	const struct dc_context *ctx)
+{
+	struct dmub_rb_cmd_burst_write *cmd_buf = &offload->cmd_data.burst_write;
+
+	cmd_buf->header.payload_bytes =
+			sizeof(uint32_t) * offload->reg_seq_count;
+
+	dc_wake_and_execute_dmub_cmd(ctx, &offload->cmd_data, DM_DMUB_WAIT_TYPE_NO_WAIT);
+
+	memset(cmd_buf, 0, sizeof(*cmd_buf));
+
+	offload->reg_seq_count = 0;
+}
+
+static inline void submit_dmub_reg_wait(
+		struct dc_reg_helper_state *offload,
+		const struct dc_context *ctx)
+{
+	struct dmub_rb_cmd_reg_wait *cmd_buf = &offload->cmd_data.reg_wait;
+
+	dc_wake_and_execute_dmub_cmd(ctx, &offload->cmd_data, DM_DMUB_WAIT_TYPE_NO_WAIT);
+
+	memset(cmd_buf, 0, sizeof(*cmd_buf));
+	offload->reg_seq_count = 0;
+}
+
+struct dc_reg_value_masks {
+	uint32_t value;
+	uint32_t mask;
+};
+
+static inline void set_reg_field_value_masks(
+	struct dc_reg_value_masks *field_value_mask,
+	uint32_t value,
+	uint32_t mask,
+	uint8_t shift)
+{
+	ASSERT(mask != 0);
+
+	field_value_mask->value = (field_value_mask->value & ~mask) | (mask & (value << shift));
+	field_value_mask->mask = field_value_mask->mask | mask;
+}
+
+static void set_reg_field_values(struct dc_reg_value_masks *field_value_mask,
+		uint32_t addr, int n,
+		uint8_t shift1, uint32_t mask1, uint32_t field_value1,
+		va_list ap)
+{
+	uint32_t shift, mask, field_value;
+	int i = 1;
+
+	/* gather all bits value/mask getting updated in this register */
+	set_reg_field_value_masks(field_value_mask,
+			field_value1, mask1, shift1);
+
+	while (i < n) {
+		shift = va_arg(ap, uint32_t);
+		mask = va_arg(ap, uint32_t);
+		field_value = va_arg(ap, uint32_t);
+
+		set_reg_field_value_masks(field_value_mask,
+				field_value, mask, shift);
+		i++;
+	}
+}
+
+static void dmub_flush_buffer_execute(
+		struct dc_reg_helper_state *offload,
+		const struct dc_context *ctx)
+{
+	submit_dmub_read_modify_write(offload, ctx);
+}
+
+static void dmub_flush_burst_write_buffer_execute(
+		struct dc_reg_helper_state *offload,
+		const struct dc_context *ctx)
+{
+	submit_dmub_burst_write(offload, ctx);
+}
+
+static bool dmub_reg_value_burst_set_pack(const struct dc_context *ctx, uint32_t addr,
+		uint32_t reg_val)
+{
+	struct dc_reg_helper_state *offload = &ctx->dmub_srv->reg_helper_offload;
+	struct dmub_rb_cmd_burst_write *cmd_buf = &offload->cmd_data.burst_write;
+
+	/* flush command if buffer is full */
+	if (offload->reg_seq_count == DMUB_BURST_WRITE_VALUES__MAX)
+		dmub_flush_burst_write_buffer_execute(offload, ctx);
+
+	if (offload->cmd_data.cmd_common.header.type == DMUB_CMD__REG_SEQ_BURST_WRITE &&
+			addr != cmd_buf->addr) {
+		dmub_flush_burst_write_buffer_execute(offload, ctx);
+		return false;
+	}
+
+	cmd_buf->header.type = DMUB_CMD__REG_SEQ_BURST_WRITE;
+	cmd_buf->header.sub_type = 0;
+	cmd_buf->addr = addr;
+	cmd_buf->write_values[offload->reg_seq_count] = reg_val;
+	offload->reg_seq_count++;
+
+	return true;
+}
+
+static uint32_t dmub_reg_value_pack(const struct dc_context *ctx, uint32_t addr,
+		struct dc_reg_value_masks *field_value_mask)
+{
+	struct dc_reg_helper_state *offload = &ctx->dmub_srv->reg_helper_offload;
+	struct dmub_rb_cmd_read_modify_write *cmd_buf = &offload->cmd_data.read_modify_write;
+	struct dmub_cmd_read_modify_write_sequence *seq;
+
+	/* flush command if buffer is full */
+	if (offload->cmd_data.cmd_common.header.type != DMUB_CMD__REG_SEQ_BURST_WRITE &&
+			offload->reg_seq_count == DMUB_READ_MODIFY_WRITE_SEQ__MAX)
+		dmub_flush_buffer_execute(offload, ctx);
+
+	if (offload->should_burst_write) {
+		if (dmub_reg_value_burst_set_pack(ctx, addr, field_value_mask->value))
+			return field_value_mask->value;
+		else
+			offload->should_burst_write = false;
+	}
+
+	/* pack commands */
+	cmd_buf->header.type = DMUB_CMD__REG_SEQ_READ_MODIFY_WRITE;
+	cmd_buf->header.sub_type = 0;
+	seq = &cmd_buf->seq[offload->reg_seq_count];
+
+	if (offload->reg_seq_count) {
+		if (cmd_buf->seq[offload->reg_seq_count - 1].addr == addr)
+			offload->same_addr_count++;
+		else
+			offload->same_addr_count = 0;
+	}
+
+	seq->addr = addr;
+	seq->modify_mask = field_value_mask->mask;
+	seq->modify_value = field_value_mask->value;
+	offload->reg_seq_count++;
+
+	return field_value_mask->value;
+}
+
+static void dmub_reg_wait_done_pack(const struct dc_context *ctx, uint32_t addr,
+		uint32_t mask, uint32_t shift, uint32_t condition_value, uint32_t time_out_us)
+{
+	struct dc_reg_helper_state *offload = &ctx->dmub_srv->reg_helper_offload;
+	struct dmub_rb_cmd_reg_wait *cmd_buf = &offload->cmd_data.reg_wait;
+
+	cmd_buf->header.type = DMUB_CMD__REG_REG_WAIT;
+	cmd_buf->header.sub_type = 0;
+	cmd_buf->reg_wait.addr = addr;
+	cmd_buf->reg_wait.condition_field_value = mask & (condition_value << shift);
+	cmd_buf->reg_wait.mask = mask;
+	cmd_buf->reg_wait.time_out_us = time_out_us;
+}
+
+uint32_t generic_reg_update_ex(const struct dc_context *ctx,
+		uint32_t addr, int n,
+		uint8_t shift1, uint32_t mask1, uint32_t field_value1,
+		...)
+{
+	struct dc_reg_value_masks field_value_mask = {0};
+	uint32_t reg_val;
+	va_list ap;
+
+	va_start(ap, field_value1);
+
+	set_reg_field_values(&field_value_mask, addr, n, shift1, mask1,
+			field_value1, ap);
+
+	va_end(ap);
+
+	if (ctx->dmub_srv &&
+	    ctx->dmub_srv->reg_helper_offload.gather_in_progress)
+		return dmub_reg_value_pack(ctx, addr, &field_value_mask);
+		/* todo: return void so we can decouple code running in driver from register states */
+
+	/* mmio write directly */
+	reg_val = dm_read_reg(ctx, addr);
+	reg_val = (reg_val & ~field_value_mask.mask) | field_value_mask.value;
+	dm_write_reg(ctx, addr, reg_val);
+	return reg_val;
+}
+
+uint32_t generic_reg_set_ex(const struct dc_context *ctx,
+		uint32_t addr, uint32_t reg_val, int n,
+		uint8_t shift1, uint32_t mask1, uint32_t field_value1,
+		...)
+{
+	struct dc_reg_value_masks field_value_mask = {0};
+	va_list ap;
+
+	va_start(ap, field_value1);
+
+	set_reg_field_values(&field_value_mask, addr, n, shift1, mask1,
+			field_value1, ap);
+
+	va_end(ap);
+
+	/* mmio write directly */
+	reg_val = (reg_val & ~field_value_mask.mask) | field_value_mask.value;
+
+	if (ctx->dmub_srv &&
+	    ctx->dmub_srv->reg_helper_offload.gather_in_progress) {
+		return dmub_reg_value_burst_set_pack(ctx, addr, reg_val);
+		/* todo: return void so we can decouple code running in driver from register states */
+	}
+
+	dm_write_reg(ctx, addr, reg_val);
+	return reg_val;
+}
+
+uint32_t generic_reg_get(const struct dc_context *ctx, uint32_t addr,
+		uint8_t shift, uint32_t mask, uint32_t *field_value)
+{
+	uint32_t reg_val = dm_read_reg(ctx, addr);
+	*field_value = get_reg_field_value_ex(reg_val, mask, shift);
+	return reg_val;
+}
+
+uint32_t generic_reg_get2(const struct dc_context *ctx, uint32_t addr,
+		uint8_t shift1, uint32_t mask1, uint32_t *field_value1,
+		uint8_t shift2, uint32_t mask2, uint32_t *field_value2)
+{
+	uint32_t reg_val = dm_read_reg(ctx, addr);
+	*field_value1 = get_reg_field_value_ex(reg_val, mask1, shift1);
+	*field_value2 = get_reg_field_value_ex(reg_val, mask2, shift2);
+	return reg_val;
+}
+
+uint32_t generic_reg_get3(const struct dc_context *ctx, uint32_t addr,
+		uint8_t shift1, uint32_t mask1, uint32_t *field_value1,
+		uint8_t shift2, uint32_t mask2, uint32_t *field_value2,
+		uint8_t shift3, uint32_t mask3, uint32_t *field_value3)
+{
+	uint32_t reg_val = dm_read_reg(ctx, addr);
+	*field_value1 = get_reg_field_value_ex(reg_val, mask1, shift1);
+	*field_value2 = get_reg_field_value_ex(reg_val, mask2, shift2);
+	*field_value3 = get_reg_field_value_ex(reg_val, mask3, shift3);
+	return reg_val;
+}
+
+uint32_t generic_reg_get4(const struct dc_context *ctx, uint32_t addr,
+		uint8_t shift1, uint32_t mask1, uint32_t *field_value1,
+		uint8_t shift2, uint32_t mask2, uint32_t *field_value2,
+		uint8_t shift3, uint32_t mask3, uint32_t *field_value3,
+		uint8_t shift4, uint32_t mask4, uint32_t *field_value4)
+{
+	uint32_t reg_val = dm_read_reg(ctx, addr);
+	*field_value1 = get_reg_field_value_ex(reg_val, mask1, shift1);
+	*field_value2 = get_reg_field_value_ex(reg_val, mask2, shift2);
+	*field_value3 = get_reg_field_value_ex(reg_val, mask3, shift3);
+	*field_value4 = get_reg_field_value_ex(reg_val, mask4, shift4);
+	return reg_val;
+}
+
+uint32_t generic_reg_get5(const struct dc_context *ctx, uint32_t addr,
+		uint8_t shift1, uint32_t mask1, uint32_t *field_value1,
+		uint8_t shift2, uint32_t mask2, uint32_t *field_value2,
+		uint8_t shift3, uint32_t mask3, uint32_t *field_value3,
+		uint8_t shift4, uint32_t mask4, uint32_t *field_value4,
+		uint8_t shift5, uint32_t mask5, uint32_t *field_value5)
+{
+	uint32_t reg_val = dm_read_reg(ctx, addr);
+	*field_value1 = get_reg_field_value_ex(reg_val, mask1, shift1);
+	*field_value2 = get_reg_field_value_ex(reg_val, mask2, shift2);
+	*field_value3 = get_reg_field_value_ex(reg_val, mask3, shift3);
+	*field_value4 = get_reg_field_value_ex(reg_val, mask4, shift4);
+	*field_value5 = get_reg_field_value_ex(reg_val, mask5, shift5);
+	return reg_val;
+}
+
+uint32_t generic_reg_get6(const struct dc_context *ctx, uint32_t addr,
+		uint8_t shift1, uint32_t mask1, uint32_t *field_value1,
+		uint8_t shift2, uint32_t mask2, uint32_t *field_value2,
+		uint8_t shift3, uint32_t mask3, uint32_t *field_value3,
+		uint8_t shift4, uint32_t mask4, uint32_t *field_value4,
+		uint8_t shift5, uint32_t mask5, uint32_t *field_value5,
+		uint8_t shift6, uint32_t mask6, uint32_t *field_value6)
+{
+	uint32_t reg_val = dm_read_reg(ctx, addr);
+	*field_value1 = get_reg_field_value_ex(reg_val, mask1, shift1);
+	*field_value2 = get_reg_field_value_ex(reg_val, mask2, shift2);
+	*field_value3 = get_reg_field_value_ex(reg_val, mask3, shift3);
+	*field_value4 = get_reg_field_value_ex(reg_val, mask4, shift4);
+	*field_value5 = get_reg_field_value_ex(reg_val, mask5, shift5);
+	*field_value6 = get_reg_field_value_ex(reg_val, mask6, shift6);
+	return reg_val;
+}
+
+uint32_t generic_reg_get7(const struct dc_context *ctx, uint32_t addr,
+		uint8_t shift1, uint32_t mask1, uint32_t *field_value1,
+		uint8_t shift2, uint32_t mask2, uint32_t *field_value2,
+		uint8_t shift3, uint32_t mask3, uint32_t *field_value3,
+		uint8_t shift4, uint32_t mask4, uint32_t *field_value4,
+		uint8_t shift5, uint32_t mask5, uint32_t *field_value5,
+		uint8_t shift6, uint32_t mask6, uint32_t *field_value6,
+		uint8_t shift7, uint32_t mask7, uint32_t *field_value7)
+{
+	uint32_t reg_val = dm_read_reg(ctx, addr);
+	*field_value1 = get_reg_field_value_ex(reg_val, mask1, shift1);
+	*field_value2 = get_reg_field_value_ex(reg_val, mask2, shift2);
+	*field_value3 = get_reg_field_value_ex(reg_val, mask3, shift3);
+	*field_value4 = get_reg_field_value_ex(reg_val, mask4, shift4);
+	*field_value5 = get_reg_field_value_ex(reg_val, mask5, shift5);
+	*field_value6 = get_reg_field_value_ex(reg_val, mask6, shift6);
+	*field_value7 = get_reg_field_value_ex(reg_val, mask7, shift7);
+	return reg_val;
+}
+
+uint32_t generic_reg_get8(const struct dc_context *ctx, uint32_t addr,
+		uint8_t shift1, uint32_t mask1, uint32_t *field_value1,
+		uint8_t shift2, uint32_t mask2, uint32_t *field_value2,
+		uint8_t shift3, uint32_t mask3, uint32_t *field_value3,
+		uint8_t shift4, uint32_t mask4, uint32_t *field_value4,
+		uint8_t shift5, uint32_t mask5, uint32_t *field_value5,
+		uint8_t shift6, uint32_t mask6, uint32_t *field_value6,
+		uint8_t shift7, uint32_t mask7, uint32_t *field_value7,
+		uint8_t shift8, uint32_t mask8, uint32_t *field_value8)
+{
+	uint32_t reg_val = dm_read_reg(ctx, addr);
+	*field_value1 = get_reg_field_value_ex(reg_val, mask1, shift1);
+	*field_value2 = get_reg_field_value_ex(reg_val, mask2, shift2);
+	*field_value3 = get_reg_field_value_ex(reg_val, mask3, shift3);
+	*field_value4 = get_reg_field_value_ex(reg_val, mask4, shift4);
+	*field_value5 = get_reg_field_value_ex(reg_val, mask5, shift5);
+	*field_value6 = get_reg_field_value_ex(reg_val, mask6, shift6);
+	*field_value7 = get_reg_field_value_ex(reg_val, mask7, shift7);
+	*field_value8 = get_reg_field_value_ex(reg_val, mask8, shift8);
+	return reg_val;
+}
+/* note:  va version of this is pretty bad idea, since there is a output parameter pass by pointer
+ * compiler won't be able to check for size match and is prone to stack corruption type of bugs
+
+uint32_t generic_reg_get(const struct dc_context *ctx,
+		uint32_t addr, int n, ...)
+{
+	uint32_t shift, mask;
+	uint32_t *field_value;
+	uint32_t reg_val;
+	int i = 0;
+
+	reg_val = dm_read_reg(ctx, addr);
+
+	va_list ap;
+	va_start(ap, n);
+
+	while (i < n) {
+		shift = va_arg(ap, uint32_t);
+		mask = va_arg(ap, uint32_t);
+		field_value = va_arg(ap, uint32_t *);
+
+		*field_value = get_reg_field_value_ex(reg_val, mask, shift);
+		i++;
+	}
+
+	va_end(ap);
+
+	return reg_val;
+}
+*/
+
+void generic_reg_wait(const struct dc_context *ctx,
+	uint32_t addr, uint32_t shift, uint32_t mask, uint32_t condition_value,
+	unsigned int delay_between_poll_us, unsigned int time_out_num_tries,
+	const char *func_name, int line)
+{
+	uint32_t field_value;
+	uint32_t reg_val;
+	int i;
+
+	if (ctx->dmub_srv &&
+	    ctx->dmub_srv->reg_helper_offload.gather_in_progress) {
+		dmub_reg_wait_done_pack(ctx, addr, mask, shift, condition_value,
+				delay_between_poll_us * time_out_num_tries);
+		return;
+	}
+
+	/*
+	 * Something is terribly wrong if time out is > 3000ms.
+	 * 3000ms is the maximum time needed for SMU to pass values back.
+	 * This value comes from experiments.
+	 *
+	 */
+	ASSERT(delay_between_poll_us * time_out_num_tries <= 3000000);
+
+	for (i = 0; i <= time_out_num_tries; i++) {
+		if (i) {
+			if (delay_between_poll_us >= 1000)
+				msleep(delay_between_poll_us/1000);
+			else if (delay_between_poll_us > 0)
+				udelay(delay_between_poll_us);
+		}
+
+		reg_val = dm_read_reg(ctx, addr);
+
+		field_value = get_reg_field_value_ex(reg_val, mask, shift);
+
+		if (field_value == condition_value) {
+			if (i * delay_between_poll_us > 1000)
+				DC_LOG_DC("REG_WAIT taking a while: %dms in %s line:%d\n",
+						delay_between_poll_us * i / 1000,
+						func_name, line);
+			return;
+		}
+	}
+
+	DC_LOG_WARNING("REG_WAIT timeout %dus * %d tries - %s line:%d\n",
+			delay_between_poll_us, time_out_num_tries,
+			func_name, line);
+
+	BREAK_TO_DEBUGGER();
+}
+
+void generic_write_indirect_reg(const struct dc_context *ctx,
+		uint32_t addr_index, uint32_t addr_data,
+		uint32_t index, uint32_t data)
+{
+	dm_write_reg(ctx, addr_index, index);
+	dm_write_reg(ctx, addr_data, data);
+}
+
+uint32_t generic_read_indirect_reg(const struct dc_context *ctx,
+		uint32_t addr_index, uint32_t addr_data,
+		uint32_t index)
+{
+	uint32_t value = 0;
+
+	// when reg read, there should not be any offload.
+	if (ctx->dmub_srv &&
+	    ctx->dmub_srv->reg_helper_offload.gather_in_progress) {
+		ASSERT(false);
+	}
+
+	dm_write_reg(ctx, addr_index, index);
+	value = dm_read_reg(ctx, addr_data);
+
+	return value;
+}
+
+uint32_t generic_indirect_reg_get(const struct dc_context *ctx,
+		uint32_t addr_index, uint32_t addr_data,
+		uint32_t index, int n,
+		uint8_t shift1, uint32_t mask1, uint32_t *field_value1,
+		...)
+{
+	uint32_t shift, mask, *field_value;
+	uint32_t value = 0;
+	int i = 1;
+
+	va_list ap;
+
+	va_start(ap, field_value1);
+
+	value = generic_read_indirect_reg(ctx, addr_index, addr_data, index);
+	*field_value1 = get_reg_field_value_ex(value, mask1, shift1);
+
+	while (i < n) {
+		shift = va_arg(ap, uint32_t);
+		mask = va_arg(ap, uint32_t);
+		field_value = va_arg(ap, uint32_t *);
+
+		*field_value = get_reg_field_value_ex(value, mask, shift);
+		i++;
+	}
+
+	va_end(ap);
+
+	return value;
+}
+
+uint32_t generic_indirect_reg_update_ex(const struct dc_context *ctx,
+		uint32_t addr_index, uint32_t addr_data,
+		uint32_t index, uint32_t reg_val, int n,
+		uint8_t shift1, uint32_t mask1, uint32_t field_value1,
+		...)
+{
+	uint32_t shift, mask, field_value;
+	int i = 1;
+
+	va_list ap;
+
+	va_start(ap, field_value1);
+
+	reg_val = set_reg_field_value_ex(reg_val, field_value1, mask1, shift1);
+
+	while (i < n) {
+		shift = va_arg(ap, uint32_t);
+		mask = va_arg(ap, uint32_t);
+		field_value = va_arg(ap, uint32_t);
+
+		reg_val = set_reg_field_value_ex(reg_val, field_value, mask, shift);
+		i++;
+	}
+
+	generic_write_indirect_reg(ctx, addr_index, addr_data, index, reg_val);
+	va_end(ap);
+
+	return reg_val;
+}
+
+
+uint32_t generic_indirect_reg_update_ex_sync(const struct dc_context *ctx,
+		uint32_t index, uint32_t reg_val, int n,
+		uint8_t shift1, uint32_t mask1, uint32_t field_value1,
+		...)
+{
+	uint32_t shift, mask, field_value;
+	int i = 1;
+
+	va_list ap;
+
+	va_start(ap, field_value1);
+
+	reg_val = set_reg_field_value_ex(reg_val, field_value1, mask1, shift1);
+
+	while (i < n) {
+		shift = va_arg(ap, uint32_t);
+		mask = va_arg(ap, uint32_t);
+		field_value = va_arg(ap, uint32_t);
+
+		reg_val = set_reg_field_value_ex(reg_val, field_value, mask, shift);
+		i++;
+	}
+
+	dm_write_index_reg(ctx, CGS_IND_REG__PCIE, index, reg_val);
+	va_end(ap);
+
+	return reg_val;
+}
+
+uint32_t generic_indirect_reg_get_sync(const struct dc_context *ctx,
+		uint32_t index, int n,
+		uint8_t shift1, uint32_t mask1, uint32_t *field_value1,
+		...)
+{
+	uint32_t shift, mask, *field_value;
+	uint32_t value = 0;
+	int i = 1;
+
+	va_list ap;
+
+	va_start(ap, field_value1);
+
+	value = dm_read_index_reg(ctx, CGS_IND_REG__PCIE, index);
+	*field_value1 = get_reg_field_value_ex(value, mask1, shift1);
+
+	while (i < n) {
+		shift = va_arg(ap, uint32_t);
+		mask = va_arg(ap, uint32_t);
+		field_value = va_arg(ap, uint32_t *);
+
+		*field_value = get_reg_field_value_ex(value, mask, shift);
+		i++;
+	}
+
+	va_end(ap);
+
+	return value;
+}
+
+void reg_sequence_start_gather(const struct dc_context *ctx)
+{
+	/* if reg sequence is supported and enabled, set flag to
+	 * indicate we want to have REG_SET, REG_UPDATE macro build
+	 * reg sequence command buffer rather than MMIO directly.
+	 */
+
+	if (ctx->dmub_srv && ctx->dc->debug.dmub_offload_enabled) {
+		struct dc_reg_helper_state *offload =
+			&ctx->dmub_srv->reg_helper_offload;
+
+		/* caller sequence mismatch.  need to debug caller.  offload will not work!!! */
+		ASSERT(!offload->gather_in_progress);
+
+		offload->gather_in_progress = true;
+	}
+}
+
+void reg_sequence_start_execute(const struct dc_context *ctx)
+{
+	struct dc_reg_helper_state *offload;
+
+	if (!ctx->dmub_srv)
+		return;
+
+	offload = &ctx->dmub_srv->reg_helper_offload;
+
+	if (offload && offload->gather_in_progress) {
+		offload->gather_in_progress = false;
+		offload->should_burst_write = false;
+		switch (offload->cmd_data.cmd_common.header.type) {
+		case DMUB_CMD__REG_SEQ_READ_MODIFY_WRITE:
+			submit_dmub_read_modify_write(offload, ctx);
+			break;
+		case DMUB_CMD__REG_REG_WAIT:
+			submit_dmub_reg_wait(offload, ctx);
+			break;
+		case DMUB_CMD__REG_SEQ_BURST_WRITE:
+			submit_dmub_burst_write(offload, ctx);
+			break;
+		default:
+			return;
+		}
+	}
+}
+
+void reg_sequence_wait_done(const struct dc_context *ctx)
+{
+	/* callback to DM to poll for last submission done*/
+	struct dc_reg_helper_state *offload;
+
+	if (!ctx->dmub_srv)
+		return;
+
+	offload = &ctx->dmub_srv->reg_helper_offload;
+
+	if (offload &&
+	    ctx->dc->debug.dmub_offload_enabled &&
+	    !ctx->dc->debug.dmcub_emulation) {
+		dc_dmub_srv_wait_for_idle(ctx->dmub_srv, DM_DMUB_WAIT_TYPE_WAIT, NULL);
+	}
+}
+
+char *dce_version_to_string(const int version)
+{
+	switch (version) {
+	case DCE_VERSION_6_0:
+		return "DCE 6.0";
+	case DCE_VERSION_6_1:
+		return "DCE 6.1";
+	case DCE_VERSION_6_4:
+		return "DCE 6.4";
+	case DCE_VERSION_8_0:
+		return "DCE 8.0";
+	case DCE_VERSION_8_1:
+		return "DCE 8.1";
+	case DCE_VERSION_8_3:
+		return "DCE 8.3";
+	case DCE_VERSION_10_0:
+		return "DCE 10.0";
+	case DCE_VERSION_11_0:
+		return "DCE 11.0";
+	case DCE_VERSION_11_2:
+		return "DCE 11.2";
+	case DCE_VERSION_11_22:
+		return "DCE 11.22";
+	case DCE_VERSION_12_0:
+		return "DCE 12.0";
+	case DCE_VERSION_12_1:
+		return "DCE 12.1";
+	case DCN_VERSION_1_0:
+		return "DCN 1.0";
+	case DCN_VERSION_1_01:
+		return "DCN 1.0.1";
+	case DCN_VERSION_2_0:
+		return "DCN 2.0";
+	case DCN_VERSION_2_1:
+		return "DCN 2.1";
+	case DCN_VERSION_2_01:
+		return "DCN 2.0.1";
+	case DCN_VERSION_3_0:
+		return "DCN 3.0";
+	case DCN_VERSION_3_01:
+		return "DCN 3.0.1";
+	case DCN_VERSION_3_02:
+		return "DCN 3.0.2";
+	case DCN_VERSION_3_03:
+		return "DCN 3.0.3";
+	case DCN_VERSION_3_1:
+		return "DCN 3.1.2";
+	case DCN_VERSION_3_14:
+		return "DCN 3.1.4";
+	case DCN_VERSION_3_15:
+		return "DCN 3.1.5";
+	case DCN_VERSION_3_16:
+		return "DCN 3.1.6";
+	case DCN_VERSION_3_2:
+		return "DCN 3.2";
+	case DCN_VERSION_3_21:
+		return "DCN 3.2.1";
+	case DCN_VERSION_3_5:
+		return "DCN 3.5";
+	case DCN_VERSION_3_51:
+		return "DCN 3.5.1";
+	case DCN_VERSION_3_6:
+		return "DCN 3.6";
+	case DCN_VERSION_4_01:
+		return "DCN 4.0.1";
+	default:
+		return "Unknown";
+	}
+}
+
+bool dc_supports_vrr(const enum dce_version v)
+{
+	return v >= DCE_VERSION_8_0;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dc_hw_types.h b/drivers/gpu/drm/amd/display/dc/dc_hw_types.h
new file mode 100644
index 000000000000..667852517246
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_hw_types.h
@@ -0,0 +1,1160 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DC_HW_TYPES_H
+#define DC_HW_TYPES_H
+
+#include "os_types.h"
+#include "fixed31_32.h"
+#include "signal_types.h"
+
+/******************************************************************************
+ * Data types for Virtual HW Layer of DAL3.
+ * (see DAL3 design documents for HW Layer definition)
+ *
+ * The intended uses are:
+ * 1. Generation pseudocode sequences for HW programming.
+ * 2. Implementation of real HW programming by HW Sequencer of DAL3.
+ *
+ * Note: do *not* add any types which are *not* used for HW programming - this
+ * will ensure separation of Logic layer from HW layer.
+ ******************************************************************************/
+
+union large_integer {
+	struct {
+		uint32_t low_part;
+		int32_t high_part;
+	};
+
+	struct {
+		uint32_t low_part;
+		int32_t high_part;
+	} u;
+
+	int64_t quad_part;
+};
+
+#define PHYSICAL_ADDRESS_LOC union large_integer
+
+enum dc_plane_addr_type {
+	PLN_ADDR_TYPE_GRAPHICS = 0,
+	PLN_ADDR_TYPE_3DLUT,
+	PLN_ADDR_TYPE_GRPH_STEREO,
+	PLN_ADDR_TYPE_VIDEO_PROGRESSIVE,
+	PLN_ADDR_TYPE_RGBEA
+};
+
+struct dc_plane_address {
+	enum dc_plane_addr_type type;
+	uint8_t tmz_surface;
+	union {
+		struct{
+			PHYSICAL_ADDRESS_LOC addr;
+			PHYSICAL_ADDRESS_LOC cursor_cache_addr;
+			PHYSICAL_ADDRESS_LOC meta_addr;
+			union large_integer dcc_const_color;
+		} grph;
+
+		struct {
+			PHYSICAL_ADDRESS_LOC addr;
+		} lut3d;
+
+		/*stereo*/
+		struct {
+			PHYSICAL_ADDRESS_LOC left_addr;
+			PHYSICAL_ADDRESS_LOC left_meta_addr;
+			union large_integer left_dcc_const_color;
+
+			PHYSICAL_ADDRESS_LOC right_addr;
+			PHYSICAL_ADDRESS_LOC right_meta_addr;
+			union large_integer right_dcc_const_color;
+
+			PHYSICAL_ADDRESS_LOC left_alpha_addr;
+			PHYSICAL_ADDRESS_LOC left_alpha_meta_addr;
+			union large_integer left_alpha_dcc_const_color;
+
+			PHYSICAL_ADDRESS_LOC right_alpha_addr;
+			PHYSICAL_ADDRESS_LOC right_alpha_meta_addr;
+			union large_integer right_alpha_dcc_const_color;
+		} grph_stereo;
+
+		/*video  progressive*/
+		struct {
+			PHYSICAL_ADDRESS_LOC luma_addr;
+			PHYSICAL_ADDRESS_LOC luma_meta_addr;
+			union large_integer luma_dcc_const_color;
+
+			PHYSICAL_ADDRESS_LOC chroma_addr;
+			PHYSICAL_ADDRESS_LOC chroma_meta_addr;
+			union large_integer chroma_dcc_const_color;
+		} video_progressive;
+
+		struct {
+			PHYSICAL_ADDRESS_LOC addr;
+			PHYSICAL_ADDRESS_LOC meta_addr;
+			union large_integer dcc_const_color;
+
+			PHYSICAL_ADDRESS_LOC alpha_addr;
+			PHYSICAL_ADDRESS_LOC alpha_meta_addr;
+			union large_integer alpha_dcc_const_color;
+		} rgbea;
+	};
+
+	union large_integer page_table_base;
+
+	uint8_t vmid;
+};
+
+struct dc_size {
+	int width;
+	int height;
+};
+
+struct rect {
+	int x;
+	int y;
+	int width;
+	int height;
+};
+
+struct plane_size {
+	/* Graphic surface pitch in pixels.
+	 * In LINEAR_GENERAL mode, pitch
+	 * is 32 pixel aligned.
+	 */
+	int surface_pitch;
+	int chroma_pitch;
+	struct rect surface_size;
+	struct rect chroma_size;
+};
+
+struct dc_plane_dcc_param {
+	bool enable;
+
+	int meta_pitch;
+	bool independent_64b_blks;
+	uint8_t dcc_ind_blk;
+
+	int meta_pitch_c;
+	bool independent_64b_blks_c;
+	uint8_t dcc_ind_blk_c;
+};
+
+/*Displayable pixel format in fb*/
+enum surface_pixel_format {
+	SURFACE_PIXEL_FORMAT_GRPH_BEGIN = 0,
+	/*TOBE REMOVED paletta 256 colors*/
+	SURFACE_PIXEL_FORMAT_GRPH_PALETA_256_COLORS =
+		SURFACE_PIXEL_FORMAT_GRPH_BEGIN,
+	/*16 bpp*/
+	SURFACE_PIXEL_FORMAT_GRPH_ARGB1555,
+	/*16 bpp*/
+	SURFACE_PIXEL_FORMAT_GRPH_RGB565,
+	/*32 bpp*/
+	SURFACE_PIXEL_FORMAT_GRPH_ARGB8888,
+	/*32 bpp swaped*/
+	SURFACE_PIXEL_FORMAT_GRPH_ABGR8888,
+
+	SURFACE_PIXEL_FORMAT_GRPH_ARGB2101010,
+	/*swaped*/
+	SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010,
+	/*TOBE REMOVED swaped, XR_BIAS has no differance
+	 * for pixel layout than previous and we can
+	 * delete this after discusion*/
+	SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010_XR_BIAS,
+	/*64 bpp */
+	SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616,
+	/*swapped*/
+	SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616,
+	/*float*/
+	SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616F,
+	/*swaped & float*/
+	SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F,
+	/*grow graphics here if necessary */
+	SURFACE_PIXEL_FORMAT_GRPH_RGB111110_FIX,
+	SURFACE_PIXEL_FORMAT_GRPH_BGR101111_FIX,
+	SURFACE_PIXEL_FORMAT_GRPH_RGB111110_FLOAT,
+	SURFACE_PIXEL_FORMAT_GRPH_BGR101111_FLOAT,
+	SURFACE_PIXEL_FORMAT_GRPH_RGBE,
+	SURFACE_PIXEL_FORMAT_GRPH_RGBE_ALPHA,
+	SURFACE_PIXEL_FORMAT_VIDEO_BEGIN,
+	SURFACE_PIXEL_FORMAT_VIDEO_420_YCbCr =
+		SURFACE_PIXEL_FORMAT_VIDEO_BEGIN,
+	SURFACE_PIXEL_FORMAT_VIDEO_420_YCrCb,
+	SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCbCr,
+	SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCrCb,
+	SURFACE_PIXEL_FORMAT_SUBSAMPLE_END,
+	SURFACE_PIXEL_FORMAT_VIDEO_ACrYCb2101010 =
+		SURFACE_PIXEL_FORMAT_SUBSAMPLE_END,
+	SURFACE_PIXEL_FORMAT_VIDEO_CrYCbA1010102,
+	SURFACE_PIXEL_FORMAT_VIDEO_AYCrCb8888,
+	SURFACE_PIXEL_FORMAT_INVALID
+
+	/*grow 444 video here if necessary */
+};
+
+
+
+/* Pixel format */
+enum pixel_format {
+	/*graph*/
+	PIXEL_FORMAT_UNINITIALIZED,
+	PIXEL_FORMAT_INDEX8,
+	PIXEL_FORMAT_RGB565,
+	PIXEL_FORMAT_ARGB8888,
+	PIXEL_FORMAT_ARGB2101010,
+	PIXEL_FORMAT_ARGB2101010_XRBIAS,
+	PIXEL_FORMAT_FP16,
+	/*video*/
+	PIXEL_FORMAT_420BPP8,
+	PIXEL_FORMAT_420BPP10,
+	/*end of pixel format definition*/
+	PIXEL_FORMAT_INVALID,
+
+	PIXEL_FORMAT_GRPH_BEGIN = PIXEL_FORMAT_INDEX8,
+	PIXEL_FORMAT_GRPH_END = PIXEL_FORMAT_FP16,
+	PIXEL_FORMAT_VIDEO_BEGIN = PIXEL_FORMAT_420BPP8,
+	PIXEL_FORMAT_VIDEO_END = PIXEL_FORMAT_420BPP10,
+	PIXEL_FORMAT_UNKNOWN
+};
+
+/*
+ * This structure holds a surface address.  There could be multiple addresses
+ * in cases such as Stereo 3D, Planar YUV, etc.  Other per-flip attributes such
+ * as frame durations and DCC format can also be set.
+ */
+#define DC_MAX_DIRTY_RECTS 3
+struct dc_flip_addrs {
+	struct dc_plane_address address;
+	unsigned long long flip_timestamp_in_us;
+	bool flip_immediate;
+	/* TODO: add flip duration for FreeSync */
+	bool triplebuffer_flips;
+	unsigned int dirty_rect_count;
+	struct rect dirty_rects[DC_MAX_DIRTY_RECTS];
+};
+
+enum tile_split_values {
+	DC_DISPLAY_MICRO_TILING = 0x0,
+	DC_THIN_MICRO_TILING = 0x1,
+	DC_DEPTH_MICRO_TILING = 0x2,
+	DC_ROTATED_MICRO_TILING = 0x3,
+};
+
+enum tripleBuffer_enable {
+	DC_TRIPLEBUFFER_DISABLE = 0x0,
+	DC_TRIPLEBUFFER_ENABLE = 0x1,
+};
+enum tile_split_values_new {
+	DC_SURF_TILE_SPLIT_1KB = 0x4,
+};
+
+/* TODO: These values come from hardware spec. We need to readdress this
+ * if they ever change.
+ */
+enum array_mode_values {
+	DC_ARRAY_LINEAR_GENERAL = 0,
+	DC_ARRAY_LINEAR_ALLIGNED,
+	DC_ARRAY_1D_TILED_THIN1,
+	DC_ARRAY_1D_TILED_THICK,
+	DC_ARRAY_2D_TILED_THIN1,
+	DC_ARRAY_PRT_TILED_THIN1,
+	DC_ARRAY_PRT_2D_TILED_THIN1,
+	DC_ARRAY_2D_TILED_THICK,
+	DC_ARRAY_2D_TILED_X_THICK,
+	DC_ARRAY_PRT_TILED_THICK,
+	DC_ARRAY_PRT_2D_TILED_THICK,
+	DC_ARRAY_PRT_3D_TILED_THIN1,
+	DC_ARRAY_3D_TILED_THIN1,
+	DC_ARRAY_3D_TILED_THICK,
+	DC_ARRAY_3D_TILED_X_THICK,
+	DC_ARRAY_PRT_3D_TILED_THICK,
+};
+
+enum tile_mode_values {
+	DC_ADDR_SURF_MICRO_TILING_DISPLAY = 0x0,
+	DC_ADDR_SURF_MICRO_TILING_NON_DISPLAY = 0x1,
+};
+
+enum swizzle_mode_values {
+	DC_SW_LINEAR = 0,
+	DC_SW_256B_S = 1,
+	DC_SW_256_D = 2,
+	DC_SW_256_R = 3,
+	DC_SW_4KB_S = 5,
+	DC_SW_4KB_D = 6,
+	DC_SW_4KB_R = 7,
+	DC_SW_64KB_S = 9,
+	DC_SW_64KB_D = 10,
+	DC_SW_64KB_R = 11,
+	DC_SW_VAR_S = 13,
+	DC_SW_VAR_D = 14,
+	DC_SW_VAR_R = 15,
+	DC_SW_64KB_S_T = 17,
+	DC_SW_64KB_D_T = 18,
+	DC_SW_4KB_S_X = 21,
+	DC_SW_4KB_D_X = 22,
+	DC_SW_4KB_R_X = 23,
+	DC_SW_64KB_S_X = 25,
+	DC_SW_64KB_D_X = 26,
+	DC_SW_64KB_R_X = 27,
+	DC_SW_VAR_S_X = 29,
+	DC_SW_VAR_D_X = 30,
+	DC_SW_VAR_R_X = 31,
+	DC_SW_MAX = 32,
+	DC_SW_UNKNOWN = DC_SW_MAX
+};
+
+// Definition of swizzle modes with addr3 ASICs
+enum swizzle_mode_addr3_values {
+	DC_ADDR3_SW_LINEAR = 0,
+	DC_ADDR3_SW_256B_2D = 1,
+	DC_ADDR3_SW_4KB_2D = 2,
+	DC_ADDR3_SW_64KB_2D = 3,
+	DC_ADDR3_SW_256KB_2D = 4,
+	DC_ADDR3_SW_4KB_3D = 5,
+	DC_ADDR3_SW_64KB_3D = 6,
+	DC_ADDR3_SW_256KB_3D = 7,
+	DC_ADDR3_SW_MAX = 8,
+	DC_ADDR3_SW_UNKNOWN = DC_ADDR3_SW_MAX
+};
+
+enum dc_gfxversion {
+	DcGfxVersion7 = 0,
+	DcGfxVersion8,
+	DcGfxVersion9,
+	DcGfxVersion10,
+	DcGfxVersion11,
+	DcGfxAddr3,
+	DcGfxVersionUnknown
+};
+
+ struct dc_tiling_info {
+	unsigned int gfxversion;     // Specifies which part of the union to use. Must use DalGfxVersion enum
+	union {
+		struct {
+			/* Specifies the number of memory banks for tiling
+			 *	purposes.
+			 * Only applies to 2D and 3D tiling modes.
+			 *	POSSIBLE VALUES: 2,4,8,16
+			 */
+			unsigned int num_banks;
+			/* Specifies the number of tiles in the x direction
+			 *	to be incorporated into the same bank.
+			 * Only applies to 2D and 3D tiling modes.
+			 *	POSSIBLE VALUES: 1,2,4,8
+			 */
+			unsigned int bank_width;
+			unsigned int bank_width_c;
+			/* Specifies the number of tiles in the y direction to
+			 *	be incorporated into the same bank.
+			 * Only applies to 2D and 3D tiling modes.
+			 *	POSSIBLE VALUES: 1,2,4,8
+			 */
+			unsigned int bank_height;
+			unsigned int bank_height_c;
+			/* Specifies the macro tile aspect ratio. Only applies
+			 * to 2D and 3D tiling modes.
+			 */
+			unsigned int tile_aspect;
+			unsigned int tile_aspect_c;
+			/* Specifies the number of bytes that will be stored
+			 *	contiguously for each tile.
+			 * If the tile data requires more storage than this
+			 *	amount, it is split into multiple slices.
+			 * This field must not be larger than
+			 *	GB_ADDR_CONFIG.DRAM_ROW_SIZE.
+			 * Only applies to 2D and 3D tiling modes.
+			 * For color render targets, TILE_SPLIT >= 256B.
+			 */
+			enum tile_split_values tile_split;
+			enum tile_split_values tile_split_c;
+			/* Specifies the addressing within a tile.
+			 *	0x0 - DISPLAY_MICRO_TILING
+			 *	0x1 - THIN_MICRO_TILING
+			 *	0x2 - DEPTH_MICRO_TILING
+			 *	0x3 - ROTATED_MICRO_TILING
+			 */
+			enum tile_mode_values tile_mode;
+			enum tile_mode_values tile_mode_c;
+			/* Specifies the number of pipes and how they are
+			 *	interleaved in the surface.
+			 * Refer to memory addressing document for complete
+			 *	details and constraints.
+			 */
+			unsigned int pipe_config;
+			/* Specifies the tiling mode of the surface.
+			 * THIN tiles use an 8x8x1 tile size.
+			 * THICK tiles use an 8x8x4 tile size.
+			 * 2D tiling modes rotate banks for successive Z slices
+			 * 3D tiling modes rotate pipes and banks for Z slices
+			 * Refer to memory addressing document for complete
+			 *	details and constraints.
+			 */
+			enum array_mode_values array_mode;
+		} gfx8;
+
+		struct {
+			enum swizzle_mode_values swizzle;
+			unsigned int num_pipes;
+			unsigned int max_compressed_frags;
+			unsigned int pipe_interleave;
+
+			unsigned int num_banks;
+			unsigned int num_shader_engines;
+			unsigned int num_rb_per_se;
+			bool shaderEnable;
+
+			bool meta_linear;
+			bool rb_aligned;
+			bool pipe_aligned;
+			unsigned int num_pkrs;
+		} gfx9;/*gfx9, gfx10 and above*/
+		struct {
+			enum swizzle_mode_addr3_values swizzle;
+		} gfx_addr3;/*gfx with addr3 and above*/
+	};
+};
+
+/* Rotation angle */
+enum dc_rotation_angle {
+	ROTATION_ANGLE_0 = 0,
+	ROTATION_ANGLE_90,
+	ROTATION_ANGLE_180,
+	ROTATION_ANGLE_270,
+	ROTATION_ANGLE_COUNT
+};
+
+enum dc_scan_direction {
+	SCAN_DIRECTION_UNKNOWN = 0,
+	SCAN_DIRECTION_HORIZONTAL = 1,  /* 0, 180 rotation */
+	SCAN_DIRECTION_VERTICAL = 2,    /* 90, 270 rotation */
+};
+
+/**
+ * struct dc_cursor_position: Hardware cursor data.
+ *
+ * This struct keeps the action information related to the cursor that will be
+ * sent and received from our DC core.
+ */
+struct dc_cursor_position {
+	/**
+	 * @x: It represents the top left abscissa coordinate of the cursor.
+	 */
+	uint32_t x;
+
+	/**
+	 * @y: It is the top ordinate of the cursor coordinate.
+	 */
+	uint32_t y;
+
+	/**
+	 * @x_hotspot: Define the abscissa point where mouse click happens.
+	 */
+	uint32_t x_hotspot;
+
+	/**
+	 * @y_hotspot: Define the ordinate point where mouse click happens.
+	 */
+	uint32_t y_hotspot;
+
+	/**
+	 * @enable: This parameter indicates whether hardware cursor should be
+	 * enabled.
+	 */
+	bool enable;
+
+	/**
+	 * @translate_by_source: Translate cursor x/y by the source rectangle
+	 * for each plane.
+	 */
+	bool translate_by_source;
+};
+
+struct dc_cursor_mi_param {
+	unsigned int pixel_clk_khz;
+	unsigned int ref_clk_khz;
+	struct rect viewport;
+	struct rect recout;
+	struct fixed31_32 h_scale_ratio;
+	struct fixed31_32 v_scale_ratio;
+	enum dc_rotation_angle rotation;
+	bool mirror;
+	struct dc_stream_state *stream;
+};
+
+/* IPP related types */
+
+enum {
+	GAMMA_RGB_256_ENTRIES = 256,
+	GAMMA_RGB_FLOAT_1024_ENTRIES = 1024,
+	GAMMA_CS_TFM_1D_ENTRIES = 4096,
+	GAMMA_CUSTOM_ENTRIES = 4096,
+	GAMMA_MAX_ENTRIES = 4096
+};
+
+enum dc_gamma_type {
+	GAMMA_RGB_256 = 1,
+	GAMMA_RGB_FLOAT_1024 = 2,
+	GAMMA_CS_TFM_1D = 3,
+	GAMMA_CUSTOM = 4,
+};
+
+struct dc_csc_transform {
+	uint16_t matrix[12];
+	bool enable_adjustment;
+};
+
+struct dc_rgb_fixed {
+	struct fixed31_32 red;
+	struct fixed31_32 green;
+	struct fixed31_32 blue;
+};
+
+struct dc_gamma {
+	struct kref refcount;
+	enum dc_gamma_type type;
+	unsigned int num_entries;
+
+	struct dc_gamma_entries {
+		struct fixed31_32 red[GAMMA_MAX_ENTRIES];
+		struct fixed31_32 green[GAMMA_MAX_ENTRIES];
+		struct fixed31_32 blue[GAMMA_MAX_ENTRIES];
+	} entries;
+
+	/* private to DC core */
+	struct dc_context *ctx;
+
+	/* is_identity is used for RGB256 gamma identity which can also be programmed in INPUT_LUT.
+	 * is_logical_identity indicates the given gamma ramp regardless of type is identity.
+	 */
+	bool is_identity;
+};
+
+/* Used by both ipp amd opp functions*/
+/* TODO: to be consolidated with enum color_space */
+
+/**
+ * enum dc_cursor_color_format - DC cursor programming mode
+ *
+ * This enum is for programming CURSOR_MODE register field. What this register
+ * should be programmed to depends on OS requested cursor shape flags and what
+ * we stored in the cursor surface.
+ */
+enum dc_cursor_color_format {
+	CURSOR_MODE_MONO,
+	CURSOR_MODE_COLOR_1BIT_AND,
+	CURSOR_MODE_COLOR_PRE_MULTIPLIED_ALPHA,
+	CURSOR_MODE_COLOR_UN_PRE_MULTIPLIED_ALPHA,
+	CURSOR_MODE_COLOR_64BIT_FP_PRE_MULTIPLIED,
+	CURSOR_MODE_COLOR_64BIT_FP_UN_PRE_MULTIPLIED
+};
+
+/*
+ * This is all the parameters required by DAL in order to update the cursor
+ * attributes, including the new cursor image surface address, size, hotspot
+ * location, color format, etc.
+ */
+
+union dc_cursor_attribute_flags {
+	struct {
+		uint32_t ENABLE_MAGNIFICATION:1;
+		uint32_t INVERSE_TRANSPARENT_CLAMPING:1;
+		uint32_t HORIZONTAL_MIRROR:1;
+		uint32_t VERTICAL_MIRROR:1;
+		uint32_t INVERT_PIXEL_DATA:1;
+		uint32_t ZERO_EXPANSION:1;
+		uint32_t MIN_MAX_INVERT:1;
+		uint32_t ENABLE_CURSOR_DEGAMMA:1;
+		uint32_t RESERVED:24;
+	} bits;
+	uint32_t value;
+};
+
+struct dc_cursor_attributes {
+	/**
+	 * @address: This field represents the framebuffer address associated
+	 * with the cursor. It is important to highlight that this address is
+	 * divided into a high and low parts.
+	 */
+	PHYSICAL_ADDRESS_LOC address;
+
+	/**
+	 * @pitch: Cursor line stride.
+	 */
+	uint32_t pitch;
+
+	/**
+	 * @width: Width should correspond to cursor surface width.
+	 */
+	uint32_t width;
+	/**
+	 * @heigh: Height should correspond to cursor surface heigh.
+	 */
+	uint32_t height;
+
+	/**
+	 * @color_format: DC cursor programming mode.
+	 */
+	enum dc_cursor_color_format color_format;
+	/**
+	 * @sdr_white_level: Boosting (SDR) cursor in HDR mode.
+	 */
+	uint32_t sdr_white_level;
+
+	/**
+	 * @rotation_angle: In case we support HW Cursor rotation in the future
+	 */
+	enum dc_rotation_angle rotation_angle;
+
+	union dc_cursor_attribute_flags attribute_flags;
+};
+
+struct dpp_cursor_attributes {
+	int bias;
+	int scale;
+};
+
+/* OPP */
+
+enum dc_color_space {
+	COLOR_SPACE_UNKNOWN,
+	COLOR_SPACE_SRGB,
+	COLOR_SPACE_XR_RGB,
+	COLOR_SPACE_SRGB_LIMITED,
+	COLOR_SPACE_MSREF_SCRGB,
+	COLOR_SPACE_YCBCR601,
+	COLOR_SPACE_YCBCR709,
+	COLOR_SPACE_XV_YCC_709,
+	COLOR_SPACE_XV_YCC_601,
+	COLOR_SPACE_YCBCR601_LIMITED,
+	COLOR_SPACE_YCBCR709_LIMITED,
+	COLOR_SPACE_2020_RGB_FULLRANGE,
+	COLOR_SPACE_2020_RGB_LIMITEDRANGE,
+	COLOR_SPACE_2020_YCBCR_LIMITED,
+	COLOR_SPACE_2020_YCBCR_FULL,
+	COLOR_SPACE_ADOBERGB,
+	COLOR_SPACE_DCIP3,
+	COLOR_SPACE_DISPLAYNATIVE,
+	COLOR_SPACE_DOLBYVISION,
+	COLOR_SPACE_APPCTRL,
+	COLOR_SPACE_CUSTOMPOINTS,
+	COLOR_SPACE_YCBCR709_BLACK,
+	COLOR_SPACE_2020_YCBCR = COLOR_SPACE_2020_YCBCR_LIMITED,
+};
+
+enum dc_dither_option {
+	DITHER_OPTION_DEFAULT,
+	DITHER_OPTION_DISABLE,
+	DITHER_OPTION_FM6,
+	DITHER_OPTION_FM8,
+	DITHER_OPTION_FM10,
+	DITHER_OPTION_SPATIAL6_FRAME_RANDOM,
+	DITHER_OPTION_SPATIAL8_FRAME_RANDOM,
+	DITHER_OPTION_SPATIAL10_FRAME_RANDOM,
+	DITHER_OPTION_SPATIAL6,
+	DITHER_OPTION_SPATIAL8,
+	DITHER_OPTION_SPATIAL10,
+	DITHER_OPTION_TRUN6,
+	DITHER_OPTION_TRUN8,
+	DITHER_OPTION_TRUN10,
+	DITHER_OPTION_TRUN10_SPATIAL8,
+	DITHER_OPTION_TRUN10_SPATIAL6,
+	DITHER_OPTION_TRUN10_FM8,
+	DITHER_OPTION_TRUN10_FM6,
+	DITHER_OPTION_TRUN10_SPATIAL8_FM6,
+	DITHER_OPTION_SPATIAL10_FM8,
+	DITHER_OPTION_SPATIAL10_FM6,
+	DITHER_OPTION_TRUN8_SPATIAL6,
+	DITHER_OPTION_TRUN8_FM6,
+	DITHER_OPTION_SPATIAL8_FM6,
+	DITHER_OPTION_MAX = DITHER_OPTION_SPATIAL8_FM6,
+	DITHER_OPTION_INVALID
+};
+
+enum dc_quantization_range {
+	QUANTIZATION_RANGE_UNKNOWN,
+	QUANTIZATION_RANGE_FULL,
+	QUANTIZATION_RANGE_LIMITED
+};
+
+enum dc_dynamic_expansion {
+	DYN_EXPANSION_AUTO,
+	DYN_EXPANSION_DISABLE
+};
+
+/* XFM */
+
+/* used in  struct dc_plane_state */
+struct scaling_taps {
+	uint32_t v_taps;
+	uint32_t h_taps;
+	uint32_t v_taps_c;
+	uint32_t h_taps_c;
+	bool integer_scaling;
+};
+
+enum dc_timing_standard {
+	DC_TIMING_STANDARD_UNDEFINED,
+	DC_TIMING_STANDARD_DMT,
+	DC_TIMING_STANDARD_GTF,
+	DC_TIMING_STANDARD_CVT,
+	DC_TIMING_STANDARD_CVT_RB,
+	DC_TIMING_STANDARD_CEA770,
+	DC_TIMING_STANDARD_CEA861,
+	DC_TIMING_STANDARD_HDMI,
+	DC_TIMING_STANDARD_TV_NTSC,
+	DC_TIMING_STANDARD_TV_NTSC_J,
+	DC_TIMING_STANDARD_TV_PAL,
+	DC_TIMING_STANDARD_TV_PAL_M,
+	DC_TIMING_STANDARD_TV_PAL_CN,
+	DC_TIMING_STANDARD_TV_SECAM,
+	DC_TIMING_STANDARD_EXPLICIT,
+	/*!< For explicit timings from EDID, VBIOS, etc.*/
+	DC_TIMING_STANDARD_USER_OVERRIDE,
+	/*!< For mode timing override by user*/
+	DC_TIMING_STANDARD_MAX
+};
+
+enum dc_color_depth {
+	COLOR_DEPTH_UNDEFINED,
+	COLOR_DEPTH_666,
+	COLOR_DEPTH_888,
+	COLOR_DEPTH_101010,
+	COLOR_DEPTH_121212,
+	COLOR_DEPTH_141414,
+	COLOR_DEPTH_161616,
+	COLOR_DEPTH_999,
+	COLOR_DEPTH_111111,
+	COLOR_DEPTH_COUNT
+};
+
+enum dc_pixel_encoding {
+	PIXEL_ENCODING_UNDEFINED,
+	PIXEL_ENCODING_RGB,
+	PIXEL_ENCODING_YCBCR422,
+	PIXEL_ENCODING_YCBCR444,
+	PIXEL_ENCODING_YCBCR420,
+	PIXEL_ENCODING_COUNT
+};
+
+enum dc_aspect_ratio {
+	ASPECT_RATIO_NO_DATA,
+	ASPECT_RATIO_4_3,
+	ASPECT_RATIO_16_9,
+	ASPECT_RATIO_64_27,
+	ASPECT_RATIO_256_135,
+	ASPECT_RATIO_FUTURE
+};
+
+enum scanning_type {
+	SCANNING_TYPE_NODATA = 0,
+	SCANNING_TYPE_OVERSCAN,
+	SCANNING_TYPE_UNDERSCAN,
+	SCANNING_TYPE_FUTURE,
+	SCANNING_TYPE_UNDEFINED
+};
+
+struct dc_crtc_timing_flags {
+	uint32_t INTERLACE :1;
+	uint32_t HSYNC_POSITIVE_POLARITY :1; /* when set to 1,
+	 it is positive polarity --reversed with dal1 or video bios define*/
+	uint32_t VSYNC_POSITIVE_POLARITY :1; /* when set to 1,
+	 it is positive polarity --reversed with dal1 or video bios define*/
+
+	uint32_t HORZ_COUNT_BY_TWO:1;
+
+	uint32_t EXCLUSIVE_3D :1; /* if this bit set,
+	 timing can be driven in 3D format only
+	 and there is no corresponding 2D timing*/
+	uint32_t RIGHT_EYE_3D_POLARITY :1; /* 1 - means right eye polarity
+	 (right eye = '1', left eye = '0') */
+	uint32_t SUB_SAMPLE_3D :1; /* 1 - means left/right  images subsampled
+	 when mixed into 3D image. 0 - means summation (3D timing is doubled)*/
+	uint32_t USE_IN_3D_VIEW_ONLY :1; /* Do not use this timing in 2D View,
+	 because corresponding 2D timing also present in the list*/
+	uint32_t STEREO_3D_PREFERENCE :1; /* Means this is 2D timing
+	 and we want to match priority of corresponding 3D timing*/
+	uint32_t Y_ONLY :1;
+
+	uint32_t YCBCR420 :1; /* TODO: shouldn't need this flag, should be a separate pixel format */
+	uint32_t DTD_COUNTER :5; /* values 1 to 16 */
+
+	uint32_t FORCE_HDR :1;
+
+	/* HDMI 2.0 - Support scrambling for TMDS character
+	 * rates less than or equal to 340Mcsc */
+	uint32_t LTE_340MCSC_SCRAMBLE:1;
+
+	uint32_t DSC : 1; /* Use DSC with this timing */
+	uint32_t VBLANK_SYNCHRONIZABLE: 1;
+};
+
+enum dc_timing_3d_format {
+	TIMING_3D_FORMAT_NONE,
+	TIMING_3D_FORMAT_FRAME_ALTERNATE, /* No stereosync at all*/
+	TIMING_3D_FORMAT_INBAND_FA, /* Inband Frame Alternate (DVI/DP)*/
+	TIMING_3D_FORMAT_DP_HDMI_INBAND_FA, /* Inband FA to HDMI Frame Pack*/
+	/* for active DP-HDMI dongle*/
+	TIMING_3D_FORMAT_SIDEBAND_FA, /* Sideband Frame Alternate (eDP)*/
+	TIMING_3D_FORMAT_HW_FRAME_PACKING,
+	TIMING_3D_FORMAT_SW_FRAME_PACKING,
+	TIMING_3D_FORMAT_ROW_INTERLEAVE,
+	TIMING_3D_FORMAT_COLUMN_INTERLEAVE,
+	TIMING_3D_FORMAT_PIXEL_INTERLEAVE,
+	TIMING_3D_FORMAT_SIDE_BY_SIDE,
+	TIMING_3D_FORMAT_TOP_AND_BOTTOM,
+	TIMING_3D_FORMAT_SBS_SW_PACKED,
+	/* Side-by-side, packed by application/driver into 2D frame*/
+	TIMING_3D_FORMAT_TB_SW_PACKED,
+	/* Top-and-bottom, packed by application/driver into 2D frame*/
+
+	TIMING_3D_FORMAT_MAX,
+};
+
+#define DC_DSC_QP_SET_SIZE 15
+#define DC_DSC_RC_BUF_THRESH_SIZE 14
+struct dc_dsc_rc_params_override {
+	int32_t rc_model_size;
+	int32_t rc_buf_thresh[DC_DSC_RC_BUF_THRESH_SIZE];
+	int32_t rc_minqp[DC_DSC_QP_SET_SIZE];
+	int32_t rc_maxqp[DC_DSC_QP_SET_SIZE];
+	int32_t rc_offset[DC_DSC_QP_SET_SIZE];
+
+	int32_t rc_tgt_offset_hi;
+	int32_t rc_tgt_offset_lo;
+	int32_t rc_edge_factor;
+	int32_t rc_quant_incr_limit0;
+	int32_t rc_quant_incr_limit1;
+
+	int32_t initial_fullness_offset;
+	int32_t initial_delay;
+
+	int32_t flatness_min_qp;
+	int32_t flatness_max_qp;
+	int32_t flatness_det_thresh;
+};
+
+struct dc_dsc_config {
+	uint32_t num_slices_h; /* Number of DSC slices - horizontal */
+	uint32_t num_slices_v; /* Number of DSC slices - vertical */
+	uint32_t bits_per_pixel; /* DSC target bitrate in 1/16 of bpp (e.g. 128 -> 8bpp) */
+	bool block_pred_enable; /* DSC block prediction enable */
+	uint32_t linebuf_depth; /* DSC line buffer depth */
+	uint32_t version_minor; /* DSC minor version. Full version is formed as 1.version_minor. */
+	bool ycbcr422_simple; /* Tell DSC engine to convert YCbCr 4:2:2 to 'YCbCr 4:2:2 simple'. */
+	int32_t rc_buffer_size; /* DSC RC buffer block size in bytes */
+	bool is_frl; /* indicate if DSC is applied based on HDMI FRL sink's capability */
+	bool is_dp; /* indicate if DSC is applied based on DP's capability */
+	uint32_t mst_pbn; /* pbn of display on dsc mst hub */
+	const struct dc_dsc_rc_params_override *rc_params_ovrd; /* DM owned memory. If not NULL, apply custom dsc rc params */
+};
+
+/**
+ * struct dc_crtc_timing - Timing parameters used to configure DCN blocks
+ *
+ * DCN provides multiple signals and parameters that can be used to adjust
+ * timing parameters, this struct aggregate multiple of these values for easy
+ * access. In this struct, fields prefixed with h_* are related to horizontal
+ * timing, and v_* to vertical timing. Keep in mind that when we talk about
+ * vertical timings, the values, in general, are described in the number of
+ * lines; on the other hand, the horizontal values are in pixels.
+ */
+struct dc_crtc_timing {
+	/**
+	 * @h_total: The total number of pixels from the rising edge of HSync
+	 * until the rising edge of the current HSync.
+	 */
+	uint32_t h_total;
+
+	/**
+	 * @h_border_left: The black pixels related to the left border
+	 */
+	uint32_t h_border_left;
+
+	/**
+	 * @h_addressable: It is the range of pixels displayed horizontally.
+	 * For example, if the display resolution is 3840@2160, the horizontal
+	 * addressable area is 3840.
+	 */
+	uint32_t h_addressable;
+
+	/**
+	 * @h_border_right: The black pixels related to the right border
+	 */
+	uint32_t h_border_right;
+
+	/**
+	 * @h_front_porch: Period (in pixels) between HBlank start and the
+	 * rising edge of HSync.
+	 */
+	uint32_t h_front_porch;
+
+	/**
+	 * @h_sync_width: HSync duration in pixels.
+	 */
+	uint32_t h_sync_width;
+
+	/**
+	 * @v_total: It is the total number of lines from the rising edge of
+	 * the previous VSync until the rising edge of the current VSync.
+	 *
+	 *          |--------------------------|
+	 *          +-+        V_TOTAL         +-+
+	 *          | |                        | |
+	 * VSync ---+ +--------- // -----------+ +---
+	 */
+	uint32_t v_total;
+
+	/**
+	 * @v_border_top: The black border on the top.
+	 */
+	uint32_t v_border_top;
+
+	/**
+	 * @v_addressable: It is the range of the scanout at which the
+	 * framebuffer is displayed. For example, if the display resolution is
+	 * 3840@2160, the addressable area is 2160 lines, or if the resolution
+	 * is 1920x1080, the addressable area is 1080 lines.
+	 */
+	uint32_t v_addressable;
+
+	/**
+	 * @v_border_bottom: The black border on the bottom.
+	 */
+	uint32_t v_border_bottom;
+
+	/**
+	 * @v_front_porch: Period (in lines) between VBlank start and rising
+	 * edge of VSync.
+	 *                  +-+
+	 * VSync            | |
+	 *        ----------+ +--------...
+	 *          +------------------...
+	 * VBlank   |
+	 *        --+
+	 *          |-------|
+	 *        v_front_porch
+	 */
+	uint32_t v_front_porch;
+
+	/**
+	 * @v_sync_width: VSync signal width in lines.
+	 */
+	uint32_t v_sync_width;
+
+	/**
+	 * @pix_clk_100hz: Pipe pixel precision
+	 *
+	 * This field is used to communicate pixel clocks with 100 Hz accuracy
+	 * from dc_crtc_timing to BIOS command table.
+	 */
+	uint32_t pix_clk_100hz;
+
+	uint32_t min_refresh_in_uhz;
+	uint32_t max_refresh_in_uhz;
+
+	uint32_t vic;
+	uint32_t hdmi_vic;
+	uint32_t rid;
+	uint32_t fr_index;
+	uint32_t frl_uncompressed_video_bandwidth_in_kbps;
+	enum dc_timing_3d_format timing_3d_format;
+	enum dc_color_depth display_color_depth;
+	enum dc_pixel_encoding pixel_encoding;
+	enum dc_aspect_ratio aspect_ratio;
+	enum scanning_type scan_type;
+
+	struct dc_crtc_timing_flags flags;
+	uint32_t dsc_fixed_bits_per_pixel_x16; /* DSC target bitrate in 1/16 of bpp (e.g. 128 -> 8bpp) */
+	struct dc_dsc_config dsc_cfg;
+
+	/* The number of pixels that HBlank has been expanded by from the original EDID timing. */
+	uint32_t expanded_hblank;
+};
+
+enum trigger_delay {
+	TRIGGER_DELAY_NEXT_PIXEL = 0,
+	TRIGGER_DELAY_NEXT_LINE,
+};
+
+enum crtc_event {
+	CRTC_EVENT_VSYNC_RISING = 0,
+	CRTC_EVENT_VSYNC_FALLING
+};
+
+struct crtc_trigger_info {
+	bool enabled;
+	struct dc_stream_state *event_source;
+	enum crtc_event event;
+	enum trigger_delay delay;
+};
+
+struct dc_crtc_timing_adjust {
+	uint32_t v_total_min;
+	uint32_t v_total_max;
+	uint32_t v_total_mid;
+	uint32_t v_total_mid_frame_num;
+	uint32_t allow_otg_v_count_halt;
+	uint8_t timing_adjust_pending;
+};
+
+
+/* Passed on init */
+enum vram_type {
+	VIDEO_MEMORY_TYPE_GDDR5  = 2,
+	VIDEO_MEMORY_TYPE_DDR3   = 3,
+	VIDEO_MEMORY_TYPE_DDR4   = 4,
+	VIDEO_MEMORY_TYPE_HBM    = 5,
+	VIDEO_MEMORY_TYPE_GDDR6  = 6,
+};
+
+enum dwb_cnv_out_bpc {
+	DWB_CNV_OUT_BPC_8BPC  = 0,
+	DWB_CNV_OUT_BPC_10BPC = 1,
+};
+
+enum dwb_output_depth {
+	DWB_OUTPUT_PIXEL_DEPTH_8BPC = 0,
+	DWB_OUTPUT_PIXEL_DEPTH_10BPC = 1,
+};
+
+enum dwb_capture_rate {
+	dwb_capture_rate_0 = 0,	/* Every frame is captured. */
+	dwb_capture_rate_1 = 1,	/* Every other frame is captured. */
+	dwb_capture_rate_2 = 2,	/* Every 3rd frame is captured. */
+	dwb_capture_rate_3 = 3,	/* Every 4th frame is captured. */
+};
+
+enum dwb_scaler_mode {
+	dwb_scaler_mode_bypass444 = 0,
+	dwb_scaler_mode_rgb444 = 1,
+	dwb_scaler_mode_yuv444 = 2,
+	dwb_scaler_mode_yuv420 = 3
+};
+
+enum dwb_subsample_position {
+	DWB_INTERSTITIAL_SUBSAMPLING = 0,
+	DWB_COSITED_SUBSAMPLING      = 1
+};
+
+enum dwb_stereo_eye_select {
+	DWB_STEREO_EYE_LEFT  = 1,		/* Capture left eye only */
+	DWB_STEREO_EYE_RIGHT = 2,		/* Capture right eye only */
+};
+
+enum dwb_stereo_type {
+	DWB_STEREO_TYPE_FRAME_PACKING = 0,		/* Frame packing */
+	DWB_STEREO_TYPE_FRAME_SEQUENTIAL = 3,	/* Frame sequential */
+};
+
+enum dwb_out_format {
+	DWB_OUT_FORMAT_32BPP_ARGB = 0,
+	DWB_OUT_FORMAT_32BPP_RGBA = 1,
+	DWB_OUT_FORMAT_64BPP_ARGB = 2,
+	DWB_OUT_FORMAT_64BPP_RGBA = 3
+};
+
+enum dwb_out_denorm {
+	DWB_OUT_DENORM_10BPC = 0,
+	DWB_OUT_DENORM_8BPC = 1,
+	DWB_OUT_DENORM_BYPASS = 2
+};
+
+enum cm_gamut_remap_select {
+	CM_GAMUT_REMAP_MODE_BYPASS = 0,
+	CM_GAMUT_REMAP_MODE_RAMA_COEFF,
+	CM_GAMUT_REMAP_MODE_RAMB_COEFF,
+	CM_GAMUT_REMAP_MODE_RESERVED
+};
+
+enum cm_gamut_coef_format {
+	CM_GAMUT_REMAP_COEF_FORMAT_S2_13 = 0,
+	CM_GAMUT_REMAP_COEF_FORMAT_S3_12 = 1
+};
+
+enum mpcc_gamut_remap_mode_select {
+	MPCC_GAMUT_REMAP_MODE_SELECT_0 = 0,
+	MPCC_GAMUT_REMAP_MODE_SELECT_1,
+	MPCC_GAMUT_REMAP_MODE_SELECT_2
+};
+
+enum mpcc_gamut_remap_id {
+	MPCC_OGAM_GAMUT_REMAP,
+	MPCC_MCM_FIRST_GAMUT_REMAP,
+	MPCC_MCM_SECOND_GAMUT_REMAP,
+	MPCC_RMCM_GAMUT_REMAP,
+};
+
+enum cursor_matrix_mode {
+	CUR_MATRIX_BYPASS = 0,
+	CUR_MATRIX_SET_A,
+	CUR_MATRIX_SET_B
+};
+
+struct mcif_warmup_params {
+	union large_integer	start_address;
+	unsigned int		address_increment;
+	unsigned int		region_size;
+	unsigned int		p_vmid;
+};
+
+#define MCIF_BUF_COUNT	4
+
+struct mcif_buf_params {
+	unsigned long long	luma_address[MCIF_BUF_COUNT];
+	unsigned long long	chroma_address[MCIF_BUF_COUNT];
+	unsigned int		luma_pitch;
+	unsigned int		chroma_pitch;
+	unsigned int		warmup_pitch;
+	unsigned int		swlock;
+	unsigned int		p_vmid;
+};
+
+
+#define MAX_TG_COLOR_VALUE 0x3FF
+struct tg_color {
+	/* Maximum 10 bits color value */
+	uint16_t color_r_cr;
+	uint16_t color_g_y;
+	uint16_t color_b_cb;
+};
+
+enum symclk_state {
+	SYMCLK_OFF_TX_OFF,
+	SYMCLK_ON_TX_ON,
+	SYMCLK_ON_TX_OFF,
+};
+
+struct phy_state {
+	struct {
+		uint8_t otg		: 1;
+		uint8_t reserved	: 7;
+	} symclk_ref_cnts;
+	enum symclk_state symclk_state;
+};
+
+#endif /* DC_HW_TYPES_H */
+
diff --git a/drivers/gpu/drm/amd/display/dc/dc_plane.h b/drivers/gpu/drm/amd/display/dc/dc_plane.h
new file mode 100644
index 000000000000..14feb843e694
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_plane.h
@@ -0,0 +1,51 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef _DC_PLANE_H_
+#define _DC_PLANE_H_
+
+#include "dc_hw_types.h"
+
+union dc_plane_status_update_flags {
+	struct {
+		uint32_t address : 1;
+	} bits;
+	uint32_t raw;
+};
+
+struct dc_plane_state *dc_create_plane_state(const struct dc *dc);
+const struct dc_plane_status *dc_plane_get_status(
+		const struct dc_plane_state *plane_state,
+		union dc_plane_status_update_flags flags);
+void dc_plane_state_retain(struct dc_plane_state *plane_state);
+void dc_plane_state_release(struct dc_plane_state *plane_state);
+
+void dc_plane_force_dcc_and_tiling_disable(struct dc_plane_state *plane_state,
+					   bool clear_tiling);
+
+
+void dc_plane_copy_config(struct dc_plane_state *dst, const struct dc_plane_state *src);
+
+#endif /* _DC_PLANE_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/dc_plane_priv.h b/drivers/gpu/drm/amd/display/dc/dc_plane_priv.h
new file mode 100644
index 000000000000..ab13335f1d01
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_plane_priv.h
@@ -0,0 +1,35 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef _DC_PLANE_PRIV_H_
+#define _DC_PLANE_PRIV_H_
+
+#include "dc_plane.h"
+
+void dc_plane_construct(struct dc_context *ctx, struct dc_plane_state *plane_state);
+void dc_plane_destruct(struct dc_plane_state *plane_state);
+uint8_t dc_plane_get_pipe_mask(struct dc_state *dc_state, const struct dc_plane_state *plane_state);
+
+#endif /* _DC_PLANE_PRIV_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/dc_spl_translate.c b/drivers/gpu/drm/amd/display/dc/dc_spl_translate.c
new file mode 100644
index 000000000000..55704d4457ef
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_spl_translate.c
@@ -0,0 +1,229 @@
+// SPDX-License-Identifier: MIT
+//
+// Copyright 2024 Advanced Micro Devices, Inc.
+
+#include "dc_spl_translate.h"
+#include "dcn20/dcn20_dpp.h"
+#include "dcn32/dcn32_dpp.h"
+#include "dcn401/dcn401_dpp.h"
+
+static struct spl_callbacks dcn2_spl_callbacks = {
+	.spl_calc_lb_num_partitions = dscl2_spl_calc_lb_num_partitions,
+};
+static struct spl_callbacks dcn32_spl_callbacks = {
+	.spl_calc_lb_num_partitions = dscl32_spl_calc_lb_num_partitions,
+};
+static struct spl_callbacks dcn401_spl_callbacks = {
+	.spl_calc_lb_num_partitions = dscl401_spl_calc_lb_num_partitions,
+};
+static void populate_splrect_from_rect(struct spl_rect *spl_rect, const struct rect *rect)
+{
+	spl_rect->x = rect->x;
+	spl_rect->y = rect->y;
+	spl_rect->width = rect->width;
+	spl_rect->height = rect->height;
+}
+static void populate_rect_from_splrect(struct rect *rect, const struct spl_rect *spl_rect)
+{
+	rect->x = spl_rect->x;
+	rect->y = spl_rect->y;
+	rect->width = spl_rect->width;
+	rect->height = spl_rect->height;
+}
+static void populate_spltaps_from_taps(struct spl_taps *spl_scaling_quality,
+		const struct scaling_taps *scaling_quality)
+{
+	spl_scaling_quality->h_taps_c = scaling_quality->h_taps_c;
+	spl_scaling_quality->h_taps = scaling_quality->h_taps;
+	spl_scaling_quality->v_taps_c = scaling_quality->v_taps_c;
+	spl_scaling_quality->v_taps = scaling_quality->v_taps;
+	spl_scaling_quality->integer_scaling = scaling_quality->integer_scaling;
+}
+static void populate_taps_from_spltaps(struct scaling_taps *scaling_quality,
+		const struct spl_taps *spl_scaling_quality)
+{
+	scaling_quality->h_taps_c = spl_scaling_quality->h_taps_c + 1;
+	scaling_quality->h_taps = spl_scaling_quality->h_taps + 1;
+	scaling_quality->v_taps_c = spl_scaling_quality->v_taps_c + 1;
+	scaling_quality->v_taps = spl_scaling_quality->v_taps + 1;
+}
+static void populate_ratios_from_splratios(struct scaling_ratios *ratios,
+		const struct ratio *spl_ratios)
+{
+	ratios->horz = dc_fixpt_from_ux_dy(spl_ratios->h_scale_ratio >> 5, 3, 19);
+	ratios->vert = dc_fixpt_from_ux_dy(spl_ratios->v_scale_ratio >> 5, 3, 19);
+	ratios->horz_c = dc_fixpt_from_ux_dy(spl_ratios->h_scale_ratio_c >> 5, 3, 19);
+	ratios->vert_c = dc_fixpt_from_ux_dy(spl_ratios->v_scale_ratio_c >> 5, 3, 19);
+}
+static void populate_inits_from_splinits(struct scl_inits *inits,
+		const struct init *spl_inits)
+{
+	inits->h = dc_fixpt_from_int_dy(spl_inits->h_filter_init_int, spl_inits->h_filter_init_frac >> 5, 0, 19);
+	inits->v = dc_fixpt_from_int_dy(spl_inits->v_filter_init_int, spl_inits->v_filter_init_frac >> 5, 0, 19);
+	inits->h_c = dc_fixpt_from_int_dy(spl_inits->h_filter_init_int_c, spl_inits->h_filter_init_frac_c >> 5, 0, 19);
+	inits->v_c = dc_fixpt_from_int_dy(spl_inits->v_filter_init_int_c, spl_inits->v_filter_init_frac_c >> 5, 0, 19);
+}
+static void populate_splformat_from_format(enum spl_pixel_format *spl_pixel_format, const enum pixel_format pixel_format)
+{
+	if (pixel_format < PIXEL_FORMAT_INVALID)
+		*spl_pixel_format = (enum spl_pixel_format)pixel_format;
+	else
+		*spl_pixel_format = SPL_PIXEL_FORMAT_INVALID;
+}
+/// @brief Translate SPL input parameters from pipe context
+/// @param pipe_ctx
+/// @param spl_in
+void translate_SPL_in_params_from_pipe_ctx(struct pipe_ctx *pipe_ctx, struct spl_in *spl_in)
+{
+	const struct dc_plane_state *plane_state = pipe_ctx->plane_state;
+	const struct dc_stream_state *stream = pipe_ctx->stream;
+	struct rect odm_slice_src = resource_get_odm_slice_src_rect(pipe_ctx);
+
+	// Assign the function to calculate the number of partitions in the line buffer
+	// This is used to determine the vtap support
+	switch (plane_state->ctx->dce_version)	{
+	case DCN_VERSION_2_0:
+		spl_in->callbacks = dcn2_spl_callbacks;
+		break;
+	case DCN_VERSION_3_2:
+		spl_in->callbacks = dcn32_spl_callbacks;
+		break;
+	case DCN_VERSION_4_01:
+		spl_in->callbacks = dcn401_spl_callbacks;
+		break;
+	default:
+		spl_in->callbacks = dcn2_spl_callbacks;
+	}
+	// Make format field from spl_in point to plane_res scl_data format
+	populate_splformat_from_format(&spl_in->basic_in.format, pipe_ctx->plane_res.scl_data.format);
+	// Make view_format from basic_out point to view_format from stream
+	spl_in->basic_out.view_format = (enum spl_view_3d)stream->view_format;
+	// Populate spl input basic input clip rect from plane state clip rect
+	populate_splrect_from_rect(&spl_in->basic_in.clip_rect, &plane_state->clip_rect);
+	// Populate spl input basic out src rect from stream src rect
+	populate_splrect_from_rect(&spl_in->basic_out.src_rect, &stream->src);
+	// Populate spl input basic out dst rect from stream dst rect
+	populate_splrect_from_rect(&spl_in->basic_out.dst_rect, &stream->dst);
+	// Make spl input basic input info rotation field point to plane state rotation
+	spl_in->basic_in.rotation = (enum spl_rotation_angle)plane_state->rotation;
+	// Populate spl input basic input src rect from plane state src rect
+	populate_splrect_from_rect(&spl_in->basic_in.src_rect, &plane_state->src_rect);
+	// Populate spl input basic input dst rect from plane state dst rect
+	populate_splrect_from_rect(&spl_in->basic_in.dst_rect, &plane_state->dst_rect);
+	// Make spl input basic input info horiz mirror field point to plane state horz mirror
+	spl_in->basic_in.horizontal_mirror = plane_state->horizontal_mirror;
+
+	// Calculate horizontal splits and split index
+	spl_in->basic_in.num_h_slices_recout_width_align.use_recout_width_aligned = false;
+	spl_in->basic_in.num_h_slices_recout_width_align.num_slices_recout_width.mpc_num_h_slices =
+		resource_get_mpc_slice_count(pipe_ctx);
+
+	if (stream->view_format == VIEW_3D_FORMAT_SIDE_BY_SIDE)
+		spl_in->basic_in.mpc_h_slice_index = 0;
+	else
+		spl_in->basic_in.mpc_h_slice_index = resource_get_mpc_slice_index(pipe_ctx);
+
+	populate_splrect_from_rect(&spl_in->basic_out.odm_slice_rect, &odm_slice_src);
+	spl_in->basic_out.odm_combine_factor = 0;
+	spl_in->odm_slice_index = resource_get_odm_slice_index(pipe_ctx);
+	// Make spl input basic out info output_size width point to stream h active
+	spl_in->basic_out.output_size.width =
+		stream->timing.h_addressable + stream->timing.h_border_left + stream->timing.h_border_right + pipe_ctx->dsc_padding_params.dsc_hactive_padding;
+	// Make spl input basic out info output_size height point to v active
+	spl_in->basic_out.output_size.height =
+		stream->timing.v_addressable + stream->timing.v_border_bottom + stream->timing.v_border_top;
+	spl_in->basic_out.max_downscale_src_width =
+			pipe_ctx->stream->ctx->dc->debug.max_downscale_src_width;
+	spl_in->basic_out.always_scale = pipe_ctx->stream->ctx->dc->debug.always_scale;
+	// Make spl input basic output info alpha_en field point to plane res scl_data lb_params alpha_en
+	spl_in->basic_out.alpha_en = pipe_ctx->plane_res.scl_data.lb_params.alpha_en;
+	spl_in->basic_out.use_two_pixels_per_container = pipe_ctx->stream_res.tg->funcs->is_two_pixels_per_container(&stream->timing);
+	// Make spl input basic input info scaling quality field point to plane state scaling_quality
+	populate_spltaps_from_taps(&spl_in->scaling_quality, &plane_state->scaling_quality);
+	// Translate edge adaptive scaler preference
+	spl_in->prefer_easf = pipe_ctx->stream->ctx->dc->config.prefer_easf;
+	spl_in->disable_easf = false;
+	if (pipe_ctx->stream->ctx->dc->debug.force_easf == 1)
+		spl_in->prefer_easf = false;
+	else if (pipe_ctx->stream->ctx->dc->debug.force_easf == 2)
+		spl_in->disable_easf = true;
+	/* Translate adaptive sharpening preference */
+	unsigned int sharpness_setting = pipe_ctx->stream->ctx->dc->debug.force_sharpness;
+	unsigned int force_sharpness_level = pipe_ctx->stream->ctx->dc->debug.force_sharpness_level;
+	if (sharpness_setting == SHARPNESS_HW_OFF)
+		spl_in->adaptive_sharpness.enable = false;
+	else if (sharpness_setting == SHARPNESS_ZERO) {
+		spl_in->adaptive_sharpness.enable = true;
+		spl_in->adaptive_sharpness.sharpness_level = 0;
+	} else if (sharpness_setting == SHARPNESS_CUSTOM) {
+		/* SAT: read harpness_range from dc_plane_state */
+		spl_in->adaptive_sharpness.sharpness_range.sdr_rgb_min = plane_state->sharpness_range.sdr_rgb_min;
+		spl_in->adaptive_sharpness.sharpness_range.sdr_rgb_max = plane_state->sharpness_range.sdr_rgb_max;
+		spl_in->adaptive_sharpness.sharpness_range.sdr_rgb_mid = plane_state->sharpness_range.sdr_rgb_mid;
+		spl_in->adaptive_sharpness.sharpness_range.sdr_yuv_min = plane_state->sharpness_range.sdr_yuv_min;
+		spl_in->adaptive_sharpness.sharpness_range.sdr_yuv_max = plane_state->sharpness_range.sdr_yuv_max;
+		spl_in->adaptive_sharpness.sharpness_range.sdr_yuv_mid = plane_state->sharpness_range.sdr_yuv_mid;
+		spl_in->adaptive_sharpness.sharpness_range.hdr_rgb_min = plane_state->sharpness_range.hdr_rgb_min;
+		spl_in->adaptive_sharpness.sharpness_range.hdr_rgb_max = plane_state->sharpness_range.hdr_rgb_max;
+		spl_in->adaptive_sharpness.sharpness_range.hdr_rgb_mid = plane_state->sharpness_range.hdr_rgb_mid;
+
+		if (force_sharpness_level > 0) {
+			if (force_sharpness_level > 10)
+				force_sharpness_level = 10;
+			spl_in->adaptive_sharpness.enable = true;
+			spl_in->adaptive_sharpness.sharpness_level = force_sharpness_level;
+		} else if (!plane_state->adaptive_sharpness_en) {
+			spl_in->adaptive_sharpness.enable = false;
+			spl_in->adaptive_sharpness.sharpness_level = 0;
+		} else {
+			spl_in->adaptive_sharpness.enable = true;
+			spl_in->adaptive_sharpness.sharpness_level = plane_state->sharpness_level;
+		}
+	}
+	// Translate linear light scaling preference
+	if (pipe_ctx->stream->ctx->dc->debug.force_lls > 0)
+		spl_in->lls_pref = pipe_ctx->stream->ctx->dc->debug.force_lls;
+	else
+		spl_in->lls_pref = plane_state->linear_light_scaling;
+	/* Translate chroma subsampling offset ( cositing ) */
+	if (pipe_ctx->stream->ctx->dc->debug.force_cositing)
+		spl_in->basic_in.cositing = pipe_ctx->stream->ctx->dc->debug.force_cositing - 1;
+	else
+		spl_in->basic_in.cositing = plane_state->cositing;
+	/* Translate transfer function */
+	spl_in->basic_in.tf_type = (enum spl_transfer_func_type) plane_state->in_transfer_func.type;
+	spl_in->basic_in.tf_predefined_type = (enum spl_transfer_func_predefined) plane_state->in_transfer_func.tf;
+
+	spl_in->h_active = pipe_ctx->plane_res.scl_data.h_active;
+	spl_in->v_active = pipe_ctx->plane_res.scl_data.v_active;
+
+	spl_in->sharpen_policy = (enum sharpen_policy)plane_state->adaptive_sharpness_policy;
+	spl_in->debug.scale_to_sharpness_policy =
+		(enum scale_to_sharpness_policy)pipe_ctx->stream->ctx->dc->debug.scale_to_sharpness_policy;
+
+	/* Check if it is stream is in fullscreen and if its HDR.
+	 * Use this to determine sharpness levels
+	 */
+	spl_in->is_fullscreen = pipe_ctx->stream->sharpening_required;
+	spl_in->is_hdr_on = dm_helpers_is_hdr_on(pipe_ctx->stream->ctx, pipe_ctx->stream);
+	spl_in->sdr_white_level_nits = plane_state->sdr_white_level_nits;
+}
+
+/// @brief Translate SPL output parameters to pipe context
+/// @param pipe_ctx
+/// @param spl_out
+void translate_SPL_out_params_to_pipe_ctx(struct pipe_ctx *pipe_ctx, struct spl_out *spl_out)
+{
+	// Make scaler data recout point to spl output field recout
+	populate_rect_from_splrect(&pipe_ctx->plane_res.scl_data.recout, &spl_out->dscl_prog_data->recout);
+	// Make scaler data ratios point to spl output field ratios
+	populate_ratios_from_splratios(&pipe_ctx->plane_res.scl_data.ratios, &spl_out->dscl_prog_data->ratios);
+	// Make scaler data viewport point to spl output field viewport
+	populate_rect_from_splrect(&pipe_ctx->plane_res.scl_data.viewport, &spl_out->dscl_prog_data->viewport);
+	// Make scaler data viewport_c point to spl output field viewport_c
+	populate_rect_from_splrect(&pipe_ctx->plane_res.scl_data.viewport_c, &spl_out->dscl_prog_data->viewport_c);
+	// Make scaler data taps point to spl output field scaling taps
+	populate_taps_from_spltaps(&pipe_ctx->plane_res.scl_data.taps, &spl_out->dscl_prog_data->taps);
+	// Make scaler data init point to spl output field init
+	populate_inits_from_splinits(&pipe_ctx->plane_res.scl_data.inits, &spl_out->dscl_prog_data->init);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dc_spl_translate.h b/drivers/gpu/drm/amd/display/dc/dc_spl_translate.h
new file mode 100644
index 000000000000..eaa5c5373b28
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_spl_translate.h
@@ -0,0 +1,23 @@
+// SPDX-License-Identifier: MIT
+//
+// Copyright 2024 Advanced Micro Devices, Inc.
+
+#ifndef __DC_SPL_TRANSLATE_H__
+#define __DC_SPL_TRANSLATE_H__
+#include "dc.h"
+#include "resource.h"
+#include "dm_helpers.h"
+
+/* Map SPL input parameters to pipe context
+ * @pipe_ctx: pipe context
+ * @spl_in: spl input structure
+ */
+void translate_SPL_in_params_from_pipe_ctx(struct pipe_ctx *pipe_ctx, struct spl_in *spl_in);
+
+/* Map SPL output parameters to pipe context
+ * @pipe_ctx: pipe context
+ * @spl_out: spl output structure
+ */
+void translate_SPL_out_params_to_pipe_ctx(struct pipe_ctx *pipe_ctx, struct spl_out *spl_out);
+
+#endif /* __DC_SPL_TRANSLATE_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dc_stat.h b/drivers/gpu/drm/amd/display/dc/dc_stat.h
new file mode 100644
index 000000000000..aacbfd786c6c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_stat.h
@@ -0,0 +1,43 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef _DC_STAT_H_
+#define _DC_STAT_H_
+
+/**
+ * DOC: DC STAT Interface
+ *
+ * These interfaces are called without acquiring DAL and DC locks.
+ * Hence, there is limitations on whese interfaces can access. Only
+ * variables exclusively defined for these interfaces can be modified.
+ */
+
+#include "dc.h"
+#include "dmub/dmub_srv.h"
+
+void dc_stat_get_dmub_notification(const struct dc *dc, struct dmub_notification *notify);
+void dc_stat_get_dmub_dataout(const struct dc *dc, uint32_t *dataout);
+
+#endif /* _DC_STAT_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/dc_state.h b/drivers/gpu/drm/amd/display/dc/dc_state.h
new file mode 100644
index 000000000000..db1e63a7d460
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_state.h
@@ -0,0 +1,77 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef _DC_STATE_H_
+#define _DC_STATE_H_
+
+#include "inc/core_status.h"
+
+struct dc_state *dc_state_create(struct dc *dc, struct dc_state_create_params *params);
+void dc_state_copy(struct dc_state *dst_state, struct dc_state *src_state);
+struct dc_state *dc_state_create_copy(struct dc_state *src_state);
+void dc_state_copy_current(struct dc *dc, struct dc_state *dst_state);
+struct dc_state *dc_state_create_current_copy(struct dc *dc);
+void dc_state_construct(struct dc *dc, struct dc_state *state);
+void dc_state_destruct(struct dc_state *state);
+void dc_state_retain(struct dc_state *state);
+void dc_state_release(struct dc_state *state);
+
+enum dc_status dc_state_add_stream(const struct dc *dc,
+				    struct dc_state *state,
+				    struct dc_stream_state *stream);
+
+enum dc_status dc_state_remove_stream(
+		const struct dc *dc,
+		struct dc_state *state,
+		struct dc_stream_state *stream);
+
+bool dc_state_add_plane(
+		const struct dc *dc,
+		struct dc_stream_state *stream,
+		struct dc_plane_state *plane_state,
+		struct dc_state *state);
+
+bool dc_state_remove_plane(
+		const struct dc *dc,
+		struct dc_stream_state *stream,
+		struct dc_plane_state *plane_state,
+		struct dc_state *state);
+
+bool dc_state_rem_all_planes_for_stream(
+		const struct dc *dc,
+		struct dc_stream_state *stream,
+		struct dc_state *state);
+
+bool dc_state_add_all_planes_for_stream(
+		const struct dc *dc,
+		struct dc_stream_state *stream,
+		struct dc_plane_state * const *plane_states,
+		int plane_count,
+		struct dc_state *state);
+
+struct dc_stream_status *dc_state_get_stream_status(
+	struct dc_state *state,
+	const struct dc_stream_state *stream);
+#endif /* _DC_STATE_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/dc_state_priv.h b/drivers/gpu/drm/amd/display/dc/dc_state_priv.h
new file mode 100644
index 000000000000..1d9bae56ff6a
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_state_priv.h
@@ -0,0 +1,128 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef _DC_STATE_PRIV_H_
+#define _DC_STATE_PRIV_H_
+
+#include "dc_state.h"
+#include "dc_stream.h"
+
+struct dc_stream_state *dc_state_get_stream_from_id(const struct dc_state *state, unsigned int id);
+
+/* Get the type of the provided resource (none, phantom, main) based on the provided
+ * context. If the context is unavailable, determine only if phantom or not.
+ */
+enum mall_stream_type dc_state_get_pipe_subvp_type(const struct dc_state *state,
+		const struct pipe_ctx *pipe_ctx);
+enum mall_stream_type dc_state_get_stream_subvp_type(const struct dc_state *state,
+		const struct dc_stream_state *stream);
+
+/* Gets the phantom stream if main is provided, gets the main if phantom is provided.*/
+struct dc_stream_state *dc_state_get_paired_subvp_stream(const struct dc_state *state,
+		const struct dc_stream_state *stream);
+
+/* allocate's phantom stream or plane and returns pointer to the object */
+struct dc_stream_state *dc_state_create_phantom_stream(const struct dc *dc,
+		struct dc_state *state,
+		struct dc_stream_state *main_stream);
+struct dc_plane_state *dc_state_create_phantom_plane(const struct dc *dc,
+		struct dc_state *state,
+		struct dc_plane_state *main_plane);
+
+/* deallocate's phantom stream or plane */
+void dc_state_release_phantom_stream(const struct dc *dc,
+		struct dc_state *state,
+		struct dc_stream_state *phantom_stream);
+void dc_state_release_phantom_plane(const struct dc *dc,
+		struct dc_state *state,
+		struct dc_plane_state *phantom_plane);
+
+/* add/remove phantom stream to context and generate subvp meta data */
+enum dc_status dc_state_add_phantom_stream(const struct dc *dc,
+		struct dc_state *state,
+		struct dc_stream_state *phantom_stream,
+		struct dc_stream_state *main_stream);
+enum dc_status dc_state_remove_phantom_stream(const struct dc *dc,
+		struct dc_state *state,
+		struct dc_stream_state *phantom_stream);
+
+bool dc_state_add_phantom_plane(
+		const struct dc *dc,
+		struct dc_stream_state *phantom_stream,
+		struct dc_plane_state *phantom_plane,
+		struct dc_state *state);
+
+bool dc_state_remove_phantom_plane(
+		const struct dc *dc,
+		struct dc_stream_state *phantom_stream,
+		struct dc_plane_state *phantom_plane,
+		struct dc_state *state);
+
+bool dc_state_rem_all_phantom_planes_for_stream(
+		const struct dc *dc,
+		struct dc_stream_state *phantom_stream,
+		struct dc_state *state,
+		bool should_release_planes);
+
+bool dc_state_add_all_phantom_planes_for_stream(
+		const struct dc *dc,
+		struct dc_stream_state *phantom_stream,
+		struct dc_plane_state * const *phantom_planes,
+		int plane_count,
+		struct dc_state *state);
+
+bool dc_state_remove_phantom_streams_and_planes(
+		const struct dc *dc,
+		struct dc_state *state);
+
+void dc_state_release_phantom_streams_and_planes(
+		const struct dc *dc,
+		struct dc_state *state);
+
+bool dc_state_is_fams2_in_use(
+		const struct dc *dc,
+		const struct dc_state *state);
+
+
+void dc_state_set_stream_subvp_cursor_limit(const struct dc_stream_state *stream,
+		struct dc_state *state,
+		bool limit);
+
+bool dc_state_get_stream_subvp_cursor_limit(const struct dc_stream_state *stream,
+		struct dc_state *state);
+
+void dc_state_set_stream_cursor_subvp_limit(const struct dc_stream_state *stream,
+		struct dc_state *state,
+		bool limit);
+
+bool dc_state_get_stream_cursor_subvp_limit(const struct dc_stream_state *stream,
+		struct dc_state *state);
+
+bool dc_state_can_clear_stream_cursor_subvp_limit(const struct dc_stream_state *stream,
+		struct dc_state *state);
+
+bool dc_state_is_subvp_in_use(struct dc_state *state);
+
+#endif /* _DC_STATE_PRIV_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/dc_stream.h b/drivers/gpu/drm/amd/display/dc/dc_stream.h
new file mode 100644
index 000000000000..76cf9fdedab0
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_stream.h
@@ -0,0 +1,607 @@
+/*
+ * Copyright 2012-14 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DC_STREAM_H_
+#define DC_STREAM_H_
+
+#include "dc_types.h"
+#include "grph_object_defs.h"
+
+/*******************************************************************************
+ * Stream Interfaces
+ ******************************************************************************/
+struct timing_sync_info {
+	int group_id;
+	int group_size;
+	bool master;
+};
+
+struct mall_stream_config {
+	/* MALL stream config to indicate if the stream is phantom or not.
+	 * We will use a phantom stream to indicate that the pipe is phantom.
+	 */
+	enum mall_stream_type type;
+	struct dc_stream_state *paired_stream;	// master / slave stream
+	bool subvp_limit_cursor_size; /* stream has/is using subvp limiting hw cursor support */
+	bool cursor_size_limit_subvp; /* stream is using hw cursor config preventing subvp */
+};
+
+struct dc_stream_status {
+	int primary_otg_inst;
+	int stream_enc_inst;
+
+	/**
+	 * @plane_count: Total of planes attached to a single stream
+	 */
+	int plane_count;
+	int audio_inst;
+	struct timing_sync_info timing_sync_info;
+	struct dc_plane_state *plane_states[MAX_SURFACES];
+	bool is_abm_supported;
+	struct mall_stream_config mall_stream_config;
+	bool fpo_in_use;
+};
+
+enum hubp_dmdata_mode {
+	DMDATA_SW_MODE,
+	DMDATA_HW_MODE
+};
+
+struct dc_dmdata_attributes {
+	/* Specifies whether dynamic meta data will be updated by software
+	 * or has to be fetched by hardware (DMA mode)
+	 */
+	enum hubp_dmdata_mode dmdata_mode;
+	/* Specifies if current dynamic meta data is to be used only for the current frame */
+	bool dmdata_repeat;
+	/* Specifies the size of Dynamic Metadata surface in byte.  Size of 0 means no Dynamic metadata is fetched */
+	uint32_t dmdata_size;
+	/* Specifies if a new dynamic meta data should be fetched for an upcoming frame */
+	bool dmdata_updated;
+	/* If hardware mode is used, the base address where DMDATA surface is located */
+	PHYSICAL_ADDRESS_LOC address;
+	/* Specifies whether QOS level will be provided by TTU or it will come from DMDATA_QOS_LEVEL */
+	bool dmdata_qos_mode;
+	/* If qos_mode = 1, this is the QOS value to be used: */
+	uint32_t dmdata_qos_level;
+	/* Specifies the value in unit of REFCLK cycles to be added to the
+	 * current time to produce the Amortized deadline for Dynamic Metadata chunk request
+	 */
+	uint32_t dmdata_dl_delta;
+	/* An unbounded array of uint32s, represents software dmdata to be loaded */
+	uint32_t *dmdata_sw_data;
+};
+
+struct dc_writeback_info {
+	bool wb_enabled;
+	int dwb_pipe_inst;
+	struct dc_dwb_params dwb_params;
+	struct mcif_buf_params mcif_buf_params;
+	struct mcif_warmup_params mcif_warmup_params;
+	/* the plane that is the input to TOP_MUX for MPCC that is the DWB source */
+	struct dc_plane_state *writeback_source_plane;
+	/* source MPCC instance.  for use by internally by dc */
+	int mpcc_inst;
+};
+
+struct dc_writeback_update {
+	unsigned int num_wb_info;
+	struct dc_writeback_info writeback_info[MAX_DWB_PIPES];
+};
+
+enum vertical_interrupt_ref_point {
+	START_V_UPDATE = 0,
+	START_V_SYNC,
+	INVALID_POINT
+
+	//For now, only v_update interrupt is used.
+	//START_V_BLANK,
+	//START_V_ACTIVE
+};
+
+struct periodic_interrupt_config {
+	enum vertical_interrupt_ref_point ref_point;
+	int lines_offset;
+};
+
+struct dc_mst_stream_bw_update {
+	bool is_increase; // is bandwidth reduced or increased
+	uint32_t mst_stream_bw; // new mst bandwidth in kbps
+};
+
+union stream_update_flags {
+	struct {
+		uint32_t scaling:1;
+		uint32_t out_tf:1;
+		uint32_t out_csc:1;
+		uint32_t abm_level:1;
+		uint32_t dpms_off:1;
+		uint32_t gamut_remap:1;
+		uint32_t wb_update:1;
+		uint32_t dsc_changed : 1;
+		uint32_t mst_bw : 1;
+		uint32_t crtc_timing_adjust : 1;
+		uint32_t fams_changed : 1;
+		uint32_t scaler_sharpener : 1;
+		uint32_t sharpening_required : 1;
+	} bits;
+
+	uint32_t raw;
+};
+
+struct test_pattern {
+	enum dp_test_pattern type;
+	enum dp_test_pattern_color_space color_space;
+	struct link_training_settings const *p_link_settings;
+	unsigned char const *p_custom_pattern;
+	unsigned int cust_pattern_size;
+};
+
+#define SUBVP_DRR_MARGIN_US 100 // 100us for DRR margin (SubVP + DRR)
+
+struct dc_stream_debug_options {
+	char force_odm_combine_segments;
+	/*
+	 * When force_odm_combine_segments is non zero, allow dc to
+	 * temporarily transition to ODM bypass when minimal transition state
+	 * is required to prevent visual glitches showing on the screen
+	 */
+	char allow_transition_for_forced_odm;
+};
+
+#define LUMINANCE_DATA_TABLE_SIZE 10
+
+struct luminance_data {
+	bool is_valid;
+	int refresh_rate_hz[LUMINANCE_DATA_TABLE_SIZE];
+	int luminance_millinits[LUMINANCE_DATA_TABLE_SIZE];
+	int flicker_criteria_milli_nits_GAMING;
+	int flicker_criteria_milli_nits_STATIC;
+	int nominal_refresh_rate;
+	int dm_max_decrease_from_nominal;
+};
+
+struct dc_stream_state {
+	// sink is deprecated, new code should not reference
+	// this pointer
+	struct dc_sink *sink;
+
+	struct dc_link *link;
+	/* For dynamic link encoder assignment, update the link encoder assigned to
+	 * a stream via the volatile dc_state rather than the static dc_link.
+	 */
+	struct link_encoder *link_enc;
+	struct dc_stream_debug_options debug;
+	struct dc_panel_patch sink_patches;
+	struct dc_crtc_timing timing;
+	struct dc_crtc_timing_adjust adjust;
+	struct dc_info_packet vrr_infopacket;
+	struct dc_info_packet vsc_infopacket;
+	struct dc_info_packet vsp_infopacket;
+	struct dc_info_packet hfvsif_infopacket;
+	struct dc_info_packet vtem_infopacket;
+	struct dc_info_packet adaptive_sync_infopacket;
+	struct dc_info_packet avi_infopacket;
+	uint8_t dsc_packed_pps[128];
+	struct rect src; /* composition area */
+	struct rect dst; /* stream addressable area */
+
+	struct audio_info audio_info;
+
+	struct dc_info_packet hdr_static_metadata;
+	PHYSICAL_ADDRESS_LOC dmdata_address;
+	bool   use_dynamic_meta;
+
+	struct dc_transfer_func out_transfer_func;
+	struct colorspace_transform gamut_remap_matrix;
+	struct dc_csc_transform csc_color_matrix;
+
+	enum dc_color_space output_color_space;
+	enum display_content_type content_type;
+	enum dc_dither_option dither_option;
+
+	enum view_3d_format view_format;
+
+	bool use_vsc_sdp_for_colorimetry;
+	bool ignore_msa_timing_param;
+
+	/**
+	 * @allow_freesync:
+	 *
+	 * It say if Freesync is enabled or not.
+	 */
+	bool allow_freesync;
+
+	/**
+	 * @vrr_active_variable:
+	 *
+	 * It describes if VRR is in use.
+	 */
+	bool vrr_active_variable;
+	bool freesync_on_desktop;
+	bool vrr_active_fixed;
+
+	bool converter_disable_audio;
+	uint8_t qs_bit;
+	uint8_t qy_bit;
+
+	/* TODO: custom INFO packets */
+	/* TODO: ABM info (DMCU) */
+	/* TODO: CEA VIC */
+
+	/* DMCU info */
+	unsigned int abm_level;
+
+	struct periodic_interrupt_config periodic_interrupt;
+
+	/* from core_stream struct */
+	struct dc_context *ctx;
+
+	/* used by DCP and FMT */
+	struct bit_depth_reduction_params bit_depth_params;
+	struct clamping_and_pixel_encoding_params clamping;
+
+	int phy_pix_clk;
+	enum signal_type signal;
+	bool dpms_off;
+
+	void *dm_stream_context;
+
+	struct dc_cursor_attributes cursor_attributes;
+	struct dc_cursor_position cursor_position;
+	bool hw_cursor_req;
+
+	uint32_t sdr_white_level; // for boosting (SDR) cursor in HDR mode
+
+	/* from stream struct */
+	struct kref refcount;
+
+	struct crtc_trigger_info triggered_crtc_reset;
+
+	/* writeback */
+	unsigned int num_wb_info;
+	struct dc_writeback_info writeback_info[MAX_DWB_PIPES];
+	const struct dc_transfer_func *func_shaper;
+	const struct dc_3dlut *lut3d_func;
+	/* Computed state bits */
+	bool mode_changed : 1;
+
+	/* Output from DC when stream state is committed or altered
+	 * DC may only access these values during:
+	 * dc_commit_state, dc_commit_state_no_check, dc_commit_streams
+	 * values may not change outside of those calls
+	 */
+	struct {
+		// For interrupt management, some hardware instance
+		// offsets need to be exposed to DM
+		uint8_t otg_offset;
+	} out;
+
+	bool apply_edp_fast_boot_optimization;
+	bool apply_seamless_boot_optimization;
+	uint32_t apply_boot_odm_mode;
+
+	uint32_t stream_id;
+
+	struct test_pattern test_pattern;
+	union stream_update_flags update_flags;
+
+	bool has_non_synchronizable_pclk;
+	bool vblank_synchronized;
+	bool is_phantom;
+
+	struct luminance_data lumin_data;
+	bool scaler_sharpener_update;
+	bool sharpening_required;
+};
+
+#define ABM_LEVEL_IMMEDIATE_DISABLE 255
+
+struct dc_stream_update {
+	struct dc_stream_state *stream;
+
+	struct rect src;
+	struct rect dst;
+	struct dc_transfer_func *out_transfer_func;
+	struct dc_info_packet *hdr_static_metadata;
+	unsigned int *abm_level;
+
+	struct periodic_interrupt_config *periodic_interrupt;
+
+	struct dc_info_packet *vrr_infopacket;
+	struct dc_info_packet *vsc_infopacket;
+	struct dc_info_packet *vsp_infopacket;
+	struct dc_info_packet *hfvsif_infopacket;
+	struct dc_info_packet *vtem_infopacket;
+	struct dc_info_packet *adaptive_sync_infopacket;
+	struct dc_info_packet *avi_infopacket;
+
+	bool *dpms_off;
+	bool integer_scaling_update;
+	bool *allow_freesync;
+	bool *vrr_active_variable;
+	bool *vrr_active_fixed;
+
+	struct colorspace_transform *gamut_remap;
+	enum dc_color_space *output_color_space;
+	enum dc_dither_option *dither_option;
+
+	struct dc_csc_transform *output_csc_transform;
+
+	struct dc_writeback_update *wb_update;
+	struct dc_dsc_config *dsc_config;
+	struct dc_mst_stream_bw_update *mst_bw_update;
+	struct dc_transfer_func *func_shaper;
+	struct dc_3dlut *lut3d_func;
+
+	struct test_pattern *pending_test_pattern;
+	struct dc_crtc_timing_adjust *crtc_timing_adjust;
+
+	struct dc_cursor_attributes *cursor_attributes;
+	struct dc_cursor_position *cursor_position;
+	bool *hw_cursor_req;
+	bool *scaler_sharpener_update;
+	bool *sharpening_required;
+};
+
+bool dc_is_stream_unchanged(
+	struct dc_stream_state *old_stream, struct dc_stream_state *stream);
+bool dc_is_stream_scaling_unchanged(
+	struct dc_stream_state *old_stream, struct dc_stream_state *stream);
+
+/*
+ * Setup stream attributes if no stream updates are provided
+ * there will be no impact on the stream parameters
+ *
+ * Set up surface attributes and associate to a stream
+ * The surfaces parameter is an absolute set of all surface active for the stream.
+ * If no surfaces are provided, the stream will be blanked; no memory read.
+ * Any flip related attribute changes must be done through this interface.
+ *
+ * After this call:
+ *   Surfaces attributes are programmed and configured to be composed into stream.
+ *   This does not trigger a flip.  No surface address is programmed.
+ *
+ */
+bool dc_update_planes_and_stream(struct dc *dc,
+		struct dc_surface_update *surface_updates, int surface_count,
+		struct dc_stream_state *dc_stream,
+		struct dc_stream_update *stream_update);
+
+/*
+ * Set up surface attributes and associate to a stream
+ * The surfaces parameter is an absolute set of all surface active for the stream.
+ * If no surfaces are provided, the stream will be blanked; no memory read.
+ * Any flip related attribute changes must be done through this interface.
+ *
+ * After this call:
+ *   Surfaces attributes are programmed and configured to be composed into stream.
+ *   This does not trigger a flip.  No surface address is programmed.
+ */
+void dc_commit_updates_for_stream(struct dc *dc,
+		struct dc_surface_update *srf_updates,
+		int surface_count,
+		struct dc_stream_state *stream,
+		struct dc_stream_update *stream_update,
+		struct dc_state *state);
+/*
+ * Log the current stream state.
+ */
+void dc_stream_log(const struct dc *dc, const struct dc_stream_state *stream);
+
+uint8_t dc_get_current_stream_count(struct dc *dc);
+struct dc_stream_state *dc_get_stream_at_index(struct dc *dc, uint8_t i);
+
+/*
+ * Return the current frame counter.
+ */
+uint32_t dc_stream_get_vblank_counter(const struct dc_stream_state *stream);
+
+/*
+ * Send dp sdp message.
+ */
+bool dc_stream_send_dp_sdp(const struct dc_stream_state *stream,
+		const uint8_t *custom_sdp_message,
+		unsigned int sdp_message_size);
+
+/* TODO: Return parsed values rather than direct register read
+ * This has a dependency on the caller (amdgpu_display_get_crtc_scanoutpos)
+ * being refactored properly to be dce-specific
+ */
+bool dc_stream_get_scanoutpos(const struct dc_stream_state *stream,
+				  uint32_t *v_blank_start,
+				  uint32_t *v_blank_end,
+				  uint32_t *h_position,
+				  uint32_t *v_position);
+
+bool dc_stream_add_writeback(struct dc *dc,
+		struct dc_stream_state *stream,
+		struct dc_writeback_info *wb_info);
+
+bool dc_stream_fc_disable_writeback(struct dc *dc,
+		struct dc_stream_state *stream,
+		uint32_t dwb_pipe_inst);
+
+bool dc_stream_remove_writeback(struct dc *dc,
+		struct dc_stream_state *stream,
+		uint32_t dwb_pipe_inst);
+
+enum dc_status dc_stream_add_dsc_to_resource(struct dc *dc,
+		struct dc_state *state,
+		struct dc_stream_state *stream);
+
+bool dc_stream_dmdata_status_done(struct dc *dc, struct dc_stream_state *stream);
+
+bool dc_stream_set_dynamic_metadata(struct dc *dc,
+		struct dc_stream_state *stream,
+		struct dc_dmdata_attributes *dmdata_attr);
+
+enum dc_status dc_validate_stream(struct dc *dc, struct dc_stream_state *stream);
+
+/*
+ * Enable stereo when commit_streams is not required,
+ * for example, frame alternate.
+ */
+void dc_enable_stereo(
+	struct dc *dc,
+	struct dc_state *context,
+	struct dc_stream_state *streams[],
+	uint8_t stream_count);
+
+/* Triggers multi-stream synchronization. */
+void dc_trigger_sync(struct dc *dc, struct dc_state *context);
+
+enum surface_update_type dc_check_update_surfaces_for_stream(
+		struct dc *dc,
+		struct dc_surface_update *updates,
+		int surface_count,
+		struct dc_stream_update *stream_update,
+		const struct dc_stream_status *stream_status);
+
+/**
+ * Create a new default stream for the requested sink
+ */
+struct dc_stream_state *dc_create_stream_for_sink(struct dc_sink *dc_sink);
+
+struct dc_stream_state *dc_copy_stream(const struct dc_stream_state *stream);
+
+void update_stream_signal(struct dc_stream_state *stream, struct dc_sink *sink);
+
+void dc_stream_retain(struct dc_stream_state *dc_stream);
+void dc_stream_release(struct dc_stream_state *dc_stream);
+
+struct dc_stream_status *dc_stream_get_status(
+	struct dc_stream_state *dc_stream);
+
+/*******************************************************************************
+ * Cursor interfaces - To manages the cursor within a stream
+ ******************************************************************************/
+/* TODO: Deprecated once we switch to dc_set_cursor_position */
+
+void program_cursor_attributes(
+	struct dc *dc,
+	struct dc_stream_state *stream);
+
+void program_cursor_position(
+	struct dc *dc,
+	struct dc_stream_state *stream);
+
+bool dc_stream_check_cursor_attributes(
+	const struct dc_stream_state *stream,
+	struct dc_state *state,
+	const struct dc_cursor_attributes *attributes);
+
+bool dc_stream_set_cursor_attributes(
+	struct dc_stream_state *stream,
+	const struct dc_cursor_attributes *attributes);
+
+bool dc_stream_program_cursor_attributes(
+	struct dc_stream_state *stream,
+	const struct dc_cursor_attributes *attributes);
+
+bool dc_stream_set_cursor_position(
+	struct dc_stream_state *stream,
+	const struct dc_cursor_position *position);
+
+bool dc_stream_program_cursor_position(
+	struct dc_stream_state *stream,
+	const struct dc_cursor_position *position);
+
+
+bool dc_stream_adjust_vmin_vmax(struct dc *dc,
+				struct dc_stream_state *stream,
+				struct dc_crtc_timing_adjust *adjust);
+
+bool dc_stream_get_last_used_drr_vtotal(struct dc *dc,
+		struct dc_stream_state *stream,
+		uint32_t *refresh_rate);
+
+#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY)
+bool dc_stream_forward_crc_window(struct dc_stream_state *stream,
+		struct rect *rect,
+		uint8_t phy_id,
+		bool is_stop);
+
+bool dc_stream_forward_multiple_crc_window(struct dc_stream_state *stream,
+		struct crc_window *window,
+		uint8_t phy_id,
+		bool stop);
+#endif
+
+bool dc_stream_configure_crc(struct dc *dc,
+			     struct dc_stream_state *stream,
+			     struct crc_params *crc_window,
+			     bool enable,
+			     bool continuous,
+			     uint8_t idx,
+			     bool reset);
+
+bool dc_stream_get_crc(struct dc *dc,
+		       struct dc_stream_state *stream,
+		       uint8_t idx,
+		       uint32_t *r_cr,
+		       uint32_t *g_y,
+		       uint32_t *b_cb);
+
+void dc_stream_set_static_screen_params(struct dc *dc,
+					struct dc_stream_state **stream,
+					int num_streams,
+					const struct dc_static_screen_params *params);
+
+void dc_stream_set_dyn_expansion(struct dc *dc, struct dc_stream_state *stream,
+		enum dc_dynamic_expansion option);
+
+void dc_stream_set_dither_option(struct dc_stream_state *stream,
+				 enum dc_dither_option option);
+
+bool dc_stream_set_gamut_remap(struct dc *dc,
+			       const struct dc_stream_state *stream);
+
+bool dc_stream_program_csc_matrix(struct dc *dc,
+				  struct dc_stream_state *stream);
+
+struct dc_rmcm_3dlut *dc_stream_get_3dlut_for_stream(
+	const struct dc *dc,
+	const struct dc_stream_state *stream,
+	bool allocate_one);
+
+void dc_stream_release_3dlut_for_stream(
+	const struct dc *dc,
+	const struct dc_stream_state *stream);
+
+void dc_stream_init_rmcm_3dlut(struct dc *dc);
+
+struct pipe_ctx *dc_stream_get_pipe_ctx(struct dc_stream_state *stream);
+
+void dc_dmub_update_dirty_rect(struct dc *dc,
+			       int surface_count,
+			       struct dc_stream_state *stream,
+			       struct dc_surface_update *srf_updates,
+			       struct dc_state *context);
+
+bool dc_stream_is_cursor_limit_pending(struct dc *dc, struct dc_stream_state *stream);
+bool dc_stream_can_clear_cursor_limit(struct dc *dc, struct dc_stream_state *stream);
+
+#endif /* DC_STREAM_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/dc_stream_priv.h b/drivers/gpu/drm/amd/display/dc/dc_stream_priv.h
new file mode 100644
index 000000000000..ca37eac20986
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_stream_priv.h
@@ -0,0 +1,75 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef _DC_STREAM_PRIV_H_
+#define _DC_STREAM_PRIV_H_
+
+#include "dc_stream.h"
+
+bool dc_stream_construct(struct dc_stream_state *stream,
+	struct dc_sink *dc_sink_data);
+void dc_stream_destruct(struct dc_stream_state *stream);
+
+void dc_stream_assign_stream_id(struct dc_stream_state *stream);
+
+/*
+ * Finds the highest refresh rate that can be achieved
+ * from starting_freq while staying within flicker criteria
+ */
+int dc_stream_calculate_max_flickerless_refresh_rate(struct dc_stream_state *stream,
+						      int starting_refresh_hz,
+						      bool is_gaming);
+
+/*
+ * Finds the lowest refresh rate that can be achieved
+ * from starting_freq while staying within flicker criteria
+ */
+int dc_stream_calculate_min_flickerless_refresh_rate(struct dc_stream_state *stream,
+						      int starting_refresh_hz,
+						      bool is_gaming);
+
+/*
+ * Determines if there will be a flicker when moving between 2 refresh rates
+ */
+bool dc_stream_is_refresh_rate_range_flickerless(struct dc_stream_state *stream,
+						  int hz1,
+						  int hz2,
+						  bool is_gaming);
+
+/*
+ * Determines the max instant vtotal delta increase that can be applied without
+ * flickering for a given stream
+ */
+unsigned int dc_stream_get_max_flickerless_instant_vtotal_decrease(struct dc_stream_state *stream,
+									  bool is_gaming);
+
+/*
+ * Determines the max instant vtotal delta decrease that can be applied without
+ * flickering for a given stream
+ */
+unsigned int dc_stream_get_max_flickerless_instant_vtotal_increase(struct dc_stream_state *stream,
+									  bool is_gaming);
+
+#endif // _DC_STREAM_PRIV_H_
diff --git a/drivers/gpu/drm/amd/display/dc/dc_trace.h b/drivers/gpu/drm/amd/display/dc/dc_trace.h
new file mode 100644
index 000000000000..bbec308a3a5e
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_trace.h
@@ -0,0 +1,44 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu_dm_trace.h"
+
+#define TRACE_DC_PIPE_STATE(pipe_ctx, index, max_pipes) \
+	for (index = 0; index < max_pipes; ++index) { \
+		struct pipe_ctx *pipe_ctx = &dc->current_state->res_ctx.pipe_ctx[index]; \
+		if (pipe_ctx->plane_state) \
+			trace_amdgpu_dm_dc_pipe_state(pipe_ctx->pipe_idx, pipe_ctx->plane_state, \
+						      pipe_ctx->stream, &pipe_ctx->plane_res, \
+						      pipe_ctx->update_flags.raw); \
+	}
+
+#define TRACE_DCE_CLOCK_STATE(dce_clocks) \
+	trace_amdgpu_dm_dce_clocks_state(dce_clocks)
+
+#define TRACE_DCN_CLOCK_STATE(dcn_clocks) \
+	trace_amdgpu_dm_dc_clocks_state(dcn_clocks)
+
+#define TRACE_DCN_FPU(begin, function, line, ref_count) \
+	trace_dcn_fpu(begin, function, line, ref_count)
+#define TRACE_OPTC_LOCK_UNLOCK_STATE(optc, inst, lock) \
+	trace_dcn_optc_lock_unlock_state(optc, inst, lock, __func__, __LINE__)
diff --git a/drivers/gpu/drm/amd/display/dc/dc_types.h b/drivers/gpu/drm/amd/display/dc/dc_types.h
new file mode 100644
index 000000000000..b5aa03a3e39c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dc_types.h
@@ -0,0 +1,1410 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+#ifndef DC_TYPES_H_
+#define DC_TYPES_H_
+
+/* AND EdidUtility only needs a portion
+ * of this file, including the rest only
+ * causes additional issues.
+ */
+#include "os_types.h"
+#include "fixed31_32.h"
+#include "irq_types.h"
+#include "dc_ddc_types.h"
+#include "dc_dp_types.h"
+#include "dc_hdmi_types.h"
+#include "dc_hw_types.h"
+#include "dal_types.h"
+#include "grph_object_defs.h"
+#include "grph_object_ctrl_defs.h"
+
+#include "dm_cp_psp.h"
+
+/* forward declarations */
+struct dc_plane_state;
+struct dc_stream_state;
+struct dc_link;
+struct dc_sink;
+struct dal;
+struct dc_dmub_srv;
+
+/********************************
+ * Environment definitions
+ ********************************/
+enum dce_environment {
+	DCE_ENV_PRODUCTION_DRV = 0,
+	/* Emulation on FPGA, in "Maximus" System.
+	 * This environment enforces that *only* DC registers accessed.
+	 * (access to non-DC registers will hang FPGA) */
+	DCE_ENV_FPGA_MAXIMUS,
+	/* Emulation on real HW or on FPGA. Used by Diagnostics, enforces
+	 * requirements of Diagnostics team. */
+	DCE_ENV_DIAG,
+	/*
+	 * Guest VM system, DC HW may exist but is not virtualized and
+	 * should not be used.  SW support for VDI only.
+	 */
+	DCE_ENV_VIRTUAL_HW
+};
+
+struct dc_perf_trace {
+	unsigned long read_count;
+	unsigned long write_count;
+	unsigned long last_entry_read;
+	unsigned long last_entry_write;
+};
+
+#define NUM_PIXEL_FORMATS 10
+
+enum tiling_mode {
+	TILING_MODE_INVALID,
+	TILING_MODE_LINEAR,
+	TILING_MODE_TILED,
+	TILING_MODE_COUNT
+};
+
+enum view_3d_format {
+	VIEW_3D_FORMAT_NONE = 0,
+	VIEW_3D_FORMAT_FRAME_SEQUENTIAL,
+	VIEW_3D_FORMAT_SIDE_BY_SIDE,
+	VIEW_3D_FORMAT_TOP_AND_BOTTOM,
+	VIEW_3D_FORMAT_COUNT,
+	VIEW_3D_FORMAT_FIRST = VIEW_3D_FORMAT_FRAME_SEQUENTIAL
+};
+
+enum plane_stereo_format {
+	PLANE_STEREO_FORMAT_NONE = 0,
+	PLANE_STEREO_FORMAT_SIDE_BY_SIDE = 1,
+	PLANE_STEREO_FORMAT_TOP_AND_BOTTOM = 2,
+	PLANE_STEREO_FORMAT_FRAME_ALTERNATE = 3,
+	PLANE_STEREO_FORMAT_ROW_INTERLEAVED = 5,
+	PLANE_STEREO_FORMAT_COLUMN_INTERLEAVED = 6,
+	PLANE_STEREO_FORMAT_CHECKER_BOARD = 7
+};
+
+/* TODO: Find way to calculate number of bits
+ *  Please increase if pixel_format enum increases
+ * num  from  PIXEL_FORMAT_INDEX8 to PIXEL_FORMAT_444BPP32
+ */
+
+enum dc_edid_connector_type {
+	DC_EDID_CONNECTOR_UNKNOWN = 0,
+	DC_EDID_CONNECTOR_ANALOG = 1,
+	DC_EDID_CONNECTOR_DIGITAL = 10,
+	DC_EDID_CONNECTOR_DVI = 11,
+	DC_EDID_CONNECTOR_HDMIA = 12,
+	DC_EDID_CONNECTOR_MDDI = 14,
+	DC_EDID_CONNECTOR_DISPLAYPORT = 15
+};
+
+enum dc_edid_status {
+	EDID_OK,
+	EDID_BAD_INPUT,
+	EDID_NO_RESPONSE,
+	EDID_BAD_CHECKSUM,
+	EDID_THE_SAME,
+	EDID_FALL_BACK,
+	EDID_PARTIAL_VALID,
+};
+
+enum act_return_status {
+	ACT_SUCCESS,
+	ACT_LINK_LOST,
+	ACT_FAILED
+};
+
+/* audio capability from EDID*/
+struct dc_cea_audio_mode {
+	uint8_t format_code; /* ucData[0] [6:3]*/
+	uint8_t channel_count; /* ucData[0] [2:0]*/
+	uint8_t sample_rate; /* ucData[1]*/
+	union {
+		uint8_t sample_size; /* for LPCM*/
+		/*  for Audio Formats 2-8 (Max bit rate divided by 8 kHz)*/
+		uint8_t max_bit_rate;
+		uint8_t audio_codec_vendor_specific; /* for Audio Formats 9-15*/
+	};
+};
+
+struct dc_edid {
+	uint32_t length;
+	uint8_t raw_edid[DC_MAX_EDID_BUFFER_SIZE];
+};
+
+/* When speaker location data block is not available, DEFAULT_SPEAKER_LOCATION
+ * is used. In this case we assume speaker location are: front left, front
+ * right and front center. */
+#define DEFAULT_SPEAKER_LOCATION 5
+
+#define DC_MAX_AUDIO_DESC_COUNT 16
+
+#define AUDIO_INFO_DISPLAY_NAME_SIZE_IN_CHARS 20
+
+struct dc_panel_patch {
+	unsigned int dppowerup_delay;
+	unsigned int extra_t12_ms;
+	unsigned int extra_delay_backlight_off;
+	unsigned int extra_t7_ms;
+	unsigned int skip_scdc_overwrite;
+	unsigned int delay_ignore_msa;
+	unsigned int disable_fec;
+	unsigned int extra_t3_ms;
+	unsigned int max_dsc_target_bpp_limit;
+	unsigned int embedded_tiled_slave;
+	unsigned int disable_fams;
+	unsigned int skip_avmute;
+	unsigned int skip_audio_sab_check;
+	unsigned int mst_start_top_delay;
+	unsigned int remove_sink_ext_caps;
+	unsigned int disable_colorimetry;
+	uint8_t blankstream_before_otg_off;
+	bool oled_optimize_display_on;
+	unsigned int force_mst_blocked_discovery;
+	unsigned int wait_after_dpcd_poweroff_ms;
+};
+
+struct dc_edid_caps {
+	/* sink identification */
+	uint16_t manufacturer_id;
+	uint16_t product_id;
+	uint32_t serial_number;
+	uint8_t manufacture_week;
+	uint8_t manufacture_year;
+	uint8_t display_name[AUDIO_INFO_DISPLAY_NAME_SIZE_IN_CHARS];
+
+	/* audio caps */
+	uint8_t speaker_flags;
+	uint32_t audio_mode_count;
+	struct dc_cea_audio_mode audio_modes[DC_MAX_AUDIO_DESC_COUNT];
+	uint32_t audio_latency;
+	uint32_t video_latency;
+
+	uint8_t qs_bit;
+	uint8_t qy_bit;
+
+	uint32_t max_tmds_clk_mhz;
+
+	/*HDMI 2.0 caps*/
+	bool lte_340mcsc_scramble;
+
+	bool edid_hdmi;
+	bool hdr_supported;
+	bool rr_capable;
+
+	struct dc_panel_patch panel_patch;
+};
+
+struct dc_mode_flags {
+	/* note: part of refresh rate flag*/
+	uint32_t INTERLACE :1;
+	/* native display timing*/
+	uint32_t NATIVE :1;
+	/* preferred is the recommended mode, one per display */
+	uint32_t PREFERRED :1;
+	/* true if this mode should use reduced blanking timings
+	 *_not_ related to the Reduced Blanking adjustment*/
+	uint32_t REDUCED_BLANKING :1;
+	/* note: part of refreshrate flag*/
+	uint32_t VIDEO_OPTIMIZED_RATE :1;
+	/* should be reported to upper layers as mode_flags*/
+	uint32_t PACKED_PIXEL_FORMAT :1;
+	/*< preferred view*/
+	uint32_t PREFERRED_VIEW :1;
+	/* this timing should be used only in tiled mode*/
+	uint32_t TILED_MODE :1;
+	uint32_t DSE_MODE :1;
+	/* Refresh rate divider when Miracast sink is using a
+	 different rate than the output display device
+	 Must be zero for wired displays and non-zero for
+	 Miracast displays*/
+	uint32_t MIRACAST_REFRESH_DIVIDER;
+};
+
+
+enum dc_timing_source {
+	TIMING_SOURCE_UNDEFINED,
+
+	/* explicitly specifed by user, most important*/
+	TIMING_SOURCE_USER_FORCED,
+	TIMING_SOURCE_USER_OVERRIDE,
+	TIMING_SOURCE_CUSTOM,
+	TIMING_SOURCE_EXPLICIT,
+
+	/* explicitly specified by the display device, more important*/
+	TIMING_SOURCE_EDID_CEA_SVD_3D,
+	TIMING_SOURCE_EDID_CEA_SVD_PREFERRED,
+	TIMING_SOURCE_EDID_CEA_SVD_420,
+	TIMING_SOURCE_EDID_DETAILED,
+	TIMING_SOURCE_EDID_ESTABLISHED,
+	TIMING_SOURCE_EDID_STANDARD,
+	TIMING_SOURCE_EDID_CEA_SVD,
+	TIMING_SOURCE_EDID_CVT_3BYTE,
+	TIMING_SOURCE_EDID_4BYTE,
+	TIMING_SOURCE_EDID_CEA_DISPLAYID_VTDB,
+	TIMING_SOURCE_EDID_CEA_RID,
+	TIMING_SOURCE_EDID_DISPLAYID_TYPE5,
+	TIMING_SOURCE_VBIOS,
+	TIMING_SOURCE_CV,
+	TIMING_SOURCE_TV,
+	TIMING_SOURCE_HDMI_VIC,
+
+	/* implicitly specified by display device, still safe but less important*/
+	TIMING_SOURCE_DEFAULT,
+
+	/* only used for custom base modes */
+	TIMING_SOURCE_CUSTOM_BASE,
+
+	/* these timing might not work, least important*/
+	TIMING_SOURCE_RANGELIMIT,
+	TIMING_SOURCE_OS_FORCED,
+	TIMING_SOURCE_IMPLICIT,
+
+	/* only used by default mode list*/
+	TIMING_SOURCE_BASICMODE,
+
+	TIMING_SOURCE_COUNT
+};
+
+
+struct stereo_3d_features {
+	bool supported			;
+	bool allTimings			;
+	bool cloneMode			;
+	bool scaling			;
+	bool singleFrameSWPacked;
+};
+
+enum dc_timing_support_method {
+	TIMING_SUPPORT_METHOD_UNDEFINED,
+	TIMING_SUPPORT_METHOD_EXPLICIT,
+	TIMING_SUPPORT_METHOD_IMPLICIT,
+	TIMING_SUPPORT_METHOD_NATIVE
+};
+
+struct dc_mode_info {
+	uint32_t pixel_width;
+	uint32_t pixel_height;
+	uint32_t field_rate;
+	/* Vertical refresh rate for progressive modes.
+	* Field rate for interlaced modes.*/
+
+	enum dc_timing_standard timing_standard;
+	enum dc_timing_source timing_source;
+	struct dc_mode_flags flags;
+};
+
+enum dc_power_state {
+	DC_POWER_STATE_ON = 1,
+	DC_POWER_STATE_STANDBY,
+	DC_POWER_STATE_SUSPEND,
+	DC_POWER_STATE_OFF
+};
+
+/* DC PowerStates */
+enum dc_video_power_state {
+	DC_VIDEO_POWER_UNSPECIFIED = 0,
+	DC_VIDEO_POWER_ON = 1,
+	DC_VIDEO_POWER_STANDBY,
+	DC_VIDEO_POWER_SUSPEND,
+	DC_VIDEO_POWER_OFF,
+	DC_VIDEO_POWER_HIBERNATE,
+	DC_VIDEO_POWER_SHUTDOWN,
+	DC_VIDEO_POWER_ULPS,	/* BACO or Ultra-Light-Power-State */
+	DC_VIDEO_POWER_AFTER_RESET,
+	DC_VIDEO_POWER_MAXIMUM
+};
+
+enum dc_acpi_cm_power_state {
+	DC_ACPI_CM_POWER_STATE_D0 = 1,
+	DC_ACPI_CM_POWER_STATE_D1 = 2,
+	DC_ACPI_CM_POWER_STATE_D2 = 4,
+	DC_ACPI_CM_POWER_STATE_D3 = 8
+};
+
+enum dc_connection_type {
+	dc_connection_none,
+	dc_connection_single,
+	dc_connection_mst_branch,
+	dc_connection_sst_branch
+};
+
+struct dc_csc_adjustments {
+	struct fixed31_32 contrast;
+	struct fixed31_32 saturation;
+	struct fixed31_32 brightness;
+	struct fixed31_32 hue;
+};
+
+/* Scaling format */
+enum scaling_transformation {
+	SCALING_TRANSFORMATION_UNINITIALIZED,
+	SCALING_TRANSFORMATION_IDENTITY = 0x0001,
+	SCALING_TRANSFORMATION_CENTER_TIMING = 0x0002,
+	SCALING_TRANSFORMATION_FULL_SCREEN_SCALE = 0x0004,
+	SCALING_TRANSFORMATION_PRESERVE_ASPECT_RATIO_SCALE = 0x0008,
+	SCALING_TRANSFORMATION_DAL_DECIDE = 0x0010,
+	SCALING_TRANSFORMATION_INVALID = 0x80000000,
+
+	/* Flag the first and last */
+	SCALING_TRANSFORMATION_BEGING = SCALING_TRANSFORMATION_IDENTITY,
+	SCALING_TRANSFORMATION_END =
+		SCALING_TRANSFORMATION_PRESERVE_ASPECT_RATIO_SCALE
+};
+
+enum display_content_type {
+	DISPLAY_CONTENT_TYPE_NO_DATA = 0,
+	DISPLAY_CONTENT_TYPE_GRAPHICS = 1,
+	DISPLAY_CONTENT_TYPE_PHOTO = 2,
+	DISPLAY_CONTENT_TYPE_CINEMA = 4,
+	DISPLAY_CONTENT_TYPE_GAME = 8
+};
+
+enum cm_gamut_adjust_type {
+	CM_GAMUT_ADJUST_TYPE_BYPASS = 0,
+	CM_GAMUT_ADJUST_TYPE_HW, /* without adjustments */
+	CM_GAMUT_ADJUST_TYPE_SW /* use adjustments */
+};
+
+struct cm_grph_csc_adjustment {
+	struct fixed31_32 temperature_matrix[12];
+	enum cm_gamut_adjust_type gamut_adjust_type;
+	enum cm_gamut_coef_format gamut_coef_format;
+};
+
+/* writeback */
+struct dwb_stereo_params {
+	bool				stereo_enabled;		/* false: normal mode, true: 3D stereo */
+	enum dwb_stereo_type		stereo_type;		/* indicates stereo format */
+	bool				stereo_polarity;	/* indicates left eye or right eye comes first in stereo mode */
+	enum dwb_stereo_eye_select	stereo_eye_select;	/* indicate which eye should be captured */
+};
+
+struct dc_dwb_cnv_params {
+	unsigned int		src_width;	/* input active width */
+	unsigned int		src_height;	/* input active height (half-active height in interlaced mode) */
+	unsigned int		crop_width;	/* cropped window width at cnv output */
+	bool			crop_en;	/* window cropping enable in cnv */
+	unsigned int		crop_height;	/* cropped window height at cnv output */
+	unsigned int		crop_x;		/* cropped window start x value at cnv output */
+	unsigned int		crop_y;		/* cropped window start y value at cnv output */
+	enum dwb_cnv_out_bpc cnv_out_bpc;	/* cnv output pixel depth - 8bpc or 10bpc */
+	enum dwb_out_format	fc_out_format;	/* dwb output pixel format - 2101010 or 16161616 and ARGB or RGBA */
+	enum dwb_out_denorm	out_denorm_mode;/* dwb output denormalization mode */
+	unsigned int		out_max_pix_val;/* pixel values greater than out_max_pix_val are clamped to out_max_pix_val */
+	unsigned int		out_min_pix_val;/* pixel values less than out_min_pix_val are clamped to out_min_pix_val */
+};
+
+struct dc_dwb_params {
+	unsigned int			dwbscl_black_color; /* must be in FP1.5.10 */
+	unsigned int			hdr_mult;	/* must be in FP1.6.12 */
+	struct cm_grph_csc_adjustment	csc_params;
+	struct dwb_stereo_params	stereo_params;
+	struct dc_dwb_cnv_params	cnv_params;	/* CNV source size and cropping window parameters */
+	unsigned int			dest_width;	/* Destination width */
+	unsigned int			dest_height;	/* Destination height */
+	enum dwb_scaler_mode		out_format;	/* default = YUV420 - TODO: limit this to 0 and 1 on dcn3 */
+	enum dwb_output_depth		output_depth;	/* output pixel depth - 8bpc or 10bpc */
+	enum dwb_capture_rate		capture_rate;	/* controls the frame capture rate */
+	struct scaling_taps 		scaler_taps;	/* Scaling taps */
+	enum dwb_subsample_position	subsample_position;
+	const struct dc_transfer_func *out_transfer_func;
+};
+
+/* audio*/
+
+union audio_sample_rates {
+	struct sample_rates {
+		uint8_t RATE_32:1;
+		uint8_t RATE_44_1:1;
+		uint8_t RATE_48:1;
+		uint8_t RATE_88_2:1;
+		uint8_t RATE_96:1;
+		uint8_t RATE_176_4:1;
+		uint8_t RATE_192:1;
+	} rate;
+
+	uint8_t all;
+};
+
+struct audio_speaker_flags {
+	uint32_t FL_FR:1;
+	uint32_t LFE:1;
+	uint32_t FC:1;
+	uint32_t RL_RR:1;
+	uint32_t RC:1;
+	uint32_t FLC_FRC:1;
+	uint32_t RLC_RRC:1;
+	uint32_t SUPPORT_AI:1;
+};
+
+struct audio_speaker_info {
+	uint32_t ALLSPEAKERS:7;
+	uint32_t SUPPORT_AI:1;
+};
+
+
+struct audio_info_flags {
+
+	union {
+
+		struct audio_speaker_flags speaker_flags;
+		struct audio_speaker_info   info;
+
+		uint8_t all;
+	};
+};
+
+enum audio_format_code {
+	AUDIO_FORMAT_CODE_FIRST = 1,
+	AUDIO_FORMAT_CODE_LINEARPCM = AUDIO_FORMAT_CODE_FIRST,
+
+	AUDIO_FORMAT_CODE_AC3,
+	/*Layers 1 & 2 */
+	AUDIO_FORMAT_CODE_MPEG1,
+	/*MPEG1 Layer 3 */
+	AUDIO_FORMAT_CODE_MP3,
+	/*multichannel */
+	AUDIO_FORMAT_CODE_MPEG2,
+	AUDIO_FORMAT_CODE_AAC,
+	AUDIO_FORMAT_CODE_DTS,
+	AUDIO_FORMAT_CODE_ATRAC,
+	AUDIO_FORMAT_CODE_1BITAUDIO,
+	AUDIO_FORMAT_CODE_DOLBYDIGITALPLUS,
+	AUDIO_FORMAT_CODE_DTS_HD,
+	AUDIO_FORMAT_CODE_MAT_MLP,
+	AUDIO_FORMAT_CODE_DST,
+	AUDIO_FORMAT_CODE_WMAPRO,
+	AUDIO_FORMAT_CODE_LAST,
+	AUDIO_FORMAT_CODE_COUNT =
+		AUDIO_FORMAT_CODE_LAST - AUDIO_FORMAT_CODE_FIRST
+};
+
+struct audio_mode {
+	 /* ucData[0] [6:3] */
+	enum audio_format_code format_code;
+	/* ucData[0] [2:0] */
+	uint8_t channel_count;
+	/* ucData[1] */
+	union audio_sample_rates sample_rates;
+	union {
+		/* for LPCM */
+		uint8_t sample_size;
+		/* for Audio Formats 2-8 (Max bit rate divided by 8 kHz) */
+		uint8_t max_bit_rate;
+		/* for Audio Formats 9-15 */
+		uint8_t vendor_specific;
+	};
+};
+
+struct audio_info {
+	struct audio_info_flags flags;
+	uint32_t video_latency;
+	uint32_t audio_latency;
+	uint32_t display_index;
+	uint8_t display_name[AUDIO_INFO_DISPLAY_NAME_SIZE_IN_CHARS];
+	uint32_t manufacture_id;
+	uint32_t product_id;
+	/* PortID used for ContainerID when defined */
+	uint32_t port_id[2];
+	uint32_t mode_count;
+	/* this field must be last in this struct */
+	struct audio_mode modes[DC_MAX_AUDIO_DESC_COUNT];
+};
+struct audio_check {
+	unsigned int audio_packet_type;
+	unsigned int max_audiosample_rate;
+	unsigned int acat;
+};
+enum dc_infoframe_type {
+	DC_HDMI_INFOFRAME_TYPE_VENDOR = 0x81,
+	DC_HDMI_INFOFRAME_TYPE_AVI = 0x82,
+	DC_HDMI_INFOFRAME_TYPE_SPD = 0x83,
+	DC_HDMI_INFOFRAME_TYPE_AUDIO = 0x84,
+	DC_DP_INFOFRAME_TYPE_PPS = 0x10,
+};
+
+struct dc_info_packet {
+	bool valid;
+	uint8_t hb0;
+	uint8_t hb1;
+	uint8_t hb2;
+	uint8_t hb3;
+	uint8_t sb[32];
+};
+
+struct dc_info_packet_128 {
+	bool valid;
+	uint8_t hb0;
+	uint8_t hb1;
+	uint8_t hb2;
+	uint8_t hb3;
+	uint8_t sb[128];
+};
+
+struct dc_edid_read_policy {
+	uint32_t max_retry_count;
+	uint32_t delay_time_ms;
+	uint32_t ignore_checksum;
+};
+
+#define DC_PLANE_UPDATE_TIMES_MAX 10
+
+struct dc_plane_flip_time {
+	unsigned int time_elapsed_in_us[DC_PLANE_UPDATE_TIMES_MAX];
+	unsigned int index;
+	unsigned int prev_update_time_in_us;
+};
+
+enum dc_alpm_mode {
+	DC_ALPM_AUXWAKE = 0,
+	DC_ALPM_AUXLESS = 1,
+	DC_ALPM_UNSUPPORTED = 0xF,
+};
+
+enum dc_psr_state {
+	PSR_STATE0 = 0x0,
+	PSR_STATE1,
+	PSR_STATE1a,
+	PSR_STATE2,
+	PSR_STATE2a,
+	PSR_STATE2b,
+	PSR_STATE3,
+	PSR_STATE3Init,
+	PSR_STATE4,
+	PSR_STATE4a,
+	PSR_STATE4b,
+	PSR_STATE4c,
+	PSR_STATE4d,
+	PSR_STATE4_FULL_FRAME,
+	PSR_STATE4a_FULL_FRAME,
+	PSR_STATE4b_FULL_FRAME,
+	PSR_STATE4c_FULL_FRAME,
+	PSR_STATE4_FULL_FRAME_POWERUP,
+	PSR_STATE4_FULL_FRAME_HW_LOCK,
+	PSR_STATE5,
+	PSR_STATE5a,
+	PSR_STATE5b,
+	PSR_STATE5c,
+	PSR_STATE_HWLOCK_MGR,
+	PSR_STATE_POLLVUPDATE,
+	PSR_STATE_RELEASE_HWLOCK_MGR_FULL_FRAME,
+	PSR_STATE_INVALID = 0xFF
+};
+
+struct psr_config {
+	unsigned char psr_version;
+	unsigned int psr_rfb_setup_time;
+	bool psr_exit_link_training_required;
+	bool psr_frame_capture_indication_req;
+	unsigned int psr_sdp_transmit_line_num_deadline;
+	bool allow_smu_optimizations;
+	bool allow_multi_disp_optimizations;
+	/* Panel self refresh 2 selective update granularity required */
+	bool su_granularity_required;
+	/* psr2 selective update y granularity capability */
+	uint8_t su_y_granularity;
+	unsigned int line_time_in_us;
+	uint8_t rate_control_caps;
+	uint16_t dsc_slice_height;
+	bool os_request_force_ffu;
+};
+
+union dmcu_psr_level {
+	struct {
+		unsigned int SKIP_CRC:1;
+		unsigned int SKIP_DP_VID_STREAM_DISABLE:1;
+		unsigned int SKIP_PHY_POWER_DOWN:1;
+		unsigned int SKIP_AUX_ACK_CHECK:1;
+		unsigned int SKIP_CRTC_DISABLE:1;
+		unsigned int SKIP_AUX_RFB_CAPTURE_CHECK:1;
+		unsigned int SKIP_SMU_NOTIFICATION:1;
+		unsigned int SKIP_AUTO_STATE_ADVANCE:1;
+		unsigned int DISABLE_PSR_ENTRY_ABORT:1;
+		unsigned int SKIP_SINGLE_OTG_DISABLE:1;
+		unsigned int DISABLE_ALPM:1;
+		unsigned int ALPM_DEFAULT_PD_MODE:1;
+		unsigned int RESERVED:20;
+	} bits;
+	unsigned int u32all;
+};
+
+enum physical_phy_id {
+	PHYLD_0,
+	PHYLD_1,
+	PHYLD_2,
+	PHYLD_3,
+	PHYLD_4,
+	PHYLD_5,
+	PHYLD_6,
+	PHYLD_7,
+	PHYLD_8,
+	PHYLD_9,
+	PHYLD_COUNT,
+	PHYLD_UNKNOWN = (-1L)
+};
+
+enum phy_type {
+	PHY_TYPE_UNKNOWN  = 1,
+	PHY_TYPE_PCIE_PHY = 2,
+	PHY_TYPE_UNIPHY = 3,
+};
+
+struct psr_context {
+	/* ddc line */
+	enum channel_id channel;
+	/* Transmitter id */
+	enum transmitter transmitterId;
+	/* Engine Id is used for Dig Be source select */
+	enum engine_id engineId;
+	/* Controller Id used for Dig Fe source select */
+	enum controller_id controllerId;
+	/* Pcie or Uniphy */
+	enum phy_type phyType;
+	/* Physical PHY Id used by SMU interpretation */
+	enum physical_phy_id smuPhyId;
+	/* Vertical total pixels from crtc timing.
+	 * This is used for static screen detection.
+	 * ie. If we want to detect half a frame,
+	 * we use this to determine the hyst lines.
+	 */
+	unsigned int crtcTimingVerticalTotal;
+	/* PSR supported from panel capabilities and
+	 * current display configuration
+	 */
+	bool psrSupportedDisplayConfig;
+	/* Whether fast link training is supported by the panel */
+	bool psrExitLinkTrainingRequired;
+	/* If RFB setup time is greater than the total VBLANK time,
+	 * it is not possible for the sink to capture the video frame
+	 * in the same frame the SDP is sent. In this case,
+	 * the frame capture indication bit should be set and an extra
+	 * static frame should be transmitted to the sink.
+	 */
+	bool psrFrameCaptureIndicationReq;
+	/* Set the last possible line SDP may be transmitted without violating
+	 * the RFB setup time or entering the active video frame.
+	 */
+	unsigned int sdpTransmitLineNumDeadline;
+	/* The VSync rate in Hz used to calculate the
+	 * step size for smooth brightness feature
+	 */
+	unsigned int vsync_rate_hz;
+	unsigned int skipPsrWaitForPllLock;
+	unsigned int numberOfControllers;
+	/* Unused, for future use. To indicate that first changed frame from
+	 * state3 shouldn't result in psr_inactive, but rather to perform
+	 * an automatic single frame rfb_update.
+	 */
+	bool rfb_update_auto_en;
+	/* Number of frame before entering static screen */
+	unsigned int timehyst_frames;
+	/* Partial frames before entering static screen */
+	unsigned int hyst_lines;
+	/* # of repeated AUX transaction attempts to make before
+	 * indicating failure to the driver
+	 */
+	unsigned int aux_repeats;
+	/* Controls hw blocks to power down during PSR active state */
+	union dmcu_psr_level psr_level;
+	/* Controls additional delay after remote frame capture before
+	 * continuing powerd own
+	 */
+	unsigned int frame_delay;
+	bool allow_smu_optimizations;
+	bool allow_multi_disp_optimizations;
+	/* Panel self refresh 2 selective update granularity required */
+	bool su_granularity_required;
+	/* psr2 selective update y granularity capability */
+	uint8_t su_y_granularity;
+	unsigned int line_time_in_us;
+	uint8_t rate_control_caps;
+	uint16_t dsc_slice_height;
+	bool os_request_force_ffu;
+};
+
+struct colorspace_transform {
+	struct fixed31_32 matrix[12];
+	bool enable_remap;
+};
+
+enum i2c_mot_mode {
+	I2C_MOT_UNDEF,
+	I2C_MOT_TRUE,
+	I2C_MOT_FALSE
+};
+
+struct AsicStateEx {
+	unsigned int memoryClock;
+	unsigned int displayClock;
+	unsigned int engineClock;
+	unsigned int maxSupportedDppClock;
+	unsigned int dppClock;
+	unsigned int socClock;
+	unsigned int dcfClockDeepSleep;
+	unsigned int fClock;
+	unsigned int phyClock;
+};
+
+
+enum dc_clock_type {
+	DC_CLOCK_TYPE_DISPCLK = 0,
+	DC_CLOCK_TYPE_DPPCLK        = 1,
+};
+
+struct dc_clock_config {
+	uint32_t max_clock_khz;
+	uint32_t min_clock_khz;
+	uint32_t bw_requirequired_clock_khz;
+	uint32_t current_clock_khz;/*current clock in use*/
+};
+
+struct hw_asic_id {
+	uint32_t chip_id;
+	uint32_t chip_family;
+	uint32_t pci_revision_id;
+	uint32_t hw_internal_rev;
+	uint32_t vram_type;
+	uint32_t vram_width;
+	uint32_t feature_flags;
+	uint32_t fake_paths_num;
+	void *atombios_base_address;
+};
+
+struct dc_context {
+	struct dc *dc;
+
+	void *driver_context; /* e.g. amdgpu_device */
+	struct dal_logger *logger;
+	struct dc_perf_trace *perf_trace;
+	void *cgs_device;
+
+	enum dce_environment dce_environment;
+	struct hw_asic_id asic_id;
+
+	/* todo: below should probably move to dc.  to facilitate removal
+	 * of AS we will store these here
+	 */
+	enum dce_version dce_version;
+	struct dc_bios *dc_bios;
+	bool created_bios;
+	struct gpio_service *gpio_service;
+	uint32_t dc_sink_id_count;
+	uint32_t dc_stream_id_count;
+	uint32_t dc_edp_id_count;
+	uint64_t fbc_gpu_addr;
+	struct dc_dmub_srv *dmub_srv;
+	struct cp_psp cp_psp;
+	uint32_t *dcn_reg_offsets;
+	uint32_t *nbio_reg_offsets;
+	uint32_t *clk_reg_offsets;
+};
+
+/* DSC DPCD capabilities */
+union dsc_slice_caps1 {
+	struct {
+		uint8_t NUM_SLICES_1 : 1;
+		uint8_t NUM_SLICES_2 : 1;
+		uint8_t RESERVED : 1;
+		uint8_t NUM_SLICES_4 : 1;
+		uint8_t NUM_SLICES_6 : 1;
+		uint8_t NUM_SLICES_8 : 1;
+		uint8_t NUM_SLICES_10 : 1;
+		uint8_t NUM_SLICES_12 : 1;
+	} bits;
+	uint8_t raw;
+};
+
+union dsc_slice_caps2 {
+	struct {
+		uint8_t NUM_SLICES_16 : 1;
+		uint8_t NUM_SLICES_20 : 1;
+		uint8_t NUM_SLICES_24 : 1;
+		uint8_t RESERVED : 5;
+	} bits;
+	uint8_t raw;
+};
+
+union dsc_color_formats {
+	struct {
+		uint8_t RGB : 1;
+		uint8_t YCBCR_444 : 1;
+		uint8_t YCBCR_SIMPLE_422 : 1;
+		uint8_t YCBCR_NATIVE_422 : 1;
+		uint8_t YCBCR_NATIVE_420 : 1;
+		uint8_t RESERVED : 3;
+	} bits;
+	uint8_t raw;
+};
+
+union dsc_color_depth {
+	struct {
+		uint8_t RESERVED1 : 1;
+		uint8_t COLOR_DEPTH_8_BPC : 1;
+		uint8_t COLOR_DEPTH_10_BPC : 1;
+		uint8_t COLOR_DEPTH_12_BPC : 1;
+		uint8_t RESERVED2 : 3;
+	} bits;
+	uint8_t raw;
+};
+
+struct dsc_dec_dpcd_caps {
+	bool is_dsc_supported;
+	uint8_t dsc_version;
+	int32_t rc_buffer_size; /* DSC RC buffer block size in bytes */
+	union dsc_slice_caps1 slice_caps1;
+	union dsc_slice_caps2 slice_caps2;
+	int32_t lb_bit_depth;
+	bool is_block_pred_supported;
+	int32_t edp_max_bits_per_pixel; /* Valid only in eDP */
+	union dsc_color_formats color_formats;
+	union dsc_color_depth color_depth;
+	int32_t throughput_mode_0_mps; /* In MPs */
+	int32_t throughput_mode_1_mps; /* In MPs */
+	int32_t max_slice_width;
+	uint32_t bpp_increment_div; /* bpp increment divisor, e.g. if 16, it's 1/16th of a bit */
+
+	/* Extended DSC caps */
+	uint32_t branch_overall_throughput_0_mps; /* In MPs */
+	uint32_t branch_overall_throughput_1_mps; /* In MPs */
+	uint32_t branch_max_line_width;
+	bool is_dp; /* Decoded format */
+};
+
+struct hblank_expansion_dpcd_caps {
+	bool expansion_supported;
+	bool reduction_supported;
+	bool buffer_unit_bytes; /* True: buffer size in bytes. False: buffer size in pixels*/
+	bool buffer_per_port; /* True: buffer size per port. False: buffer size per lane*/
+	uint32_t buffer_size; /* Add 1 to value and multiply by 32 */
+};
+
+struct dc_golden_table {
+	uint16_t dc_golden_table_ver;
+	uint32_t aux_dphy_rx_control0_val;
+	uint32_t aux_dphy_tx_control_val;
+	uint32_t aux_dphy_rx_control1_val;
+	uint32_t dc_gpio_aux_ctrl_0_val;
+	uint32_t dc_gpio_aux_ctrl_1_val;
+	uint32_t dc_gpio_aux_ctrl_2_val;
+	uint32_t dc_gpio_aux_ctrl_3_val;
+	uint32_t dc_gpio_aux_ctrl_4_val;
+	uint32_t dc_gpio_aux_ctrl_5_val;
+};
+
+enum dc_gpu_mem_alloc_type {
+	DC_MEM_ALLOC_TYPE_GART,
+	DC_MEM_ALLOC_TYPE_FRAME_BUFFER,
+	DC_MEM_ALLOC_TYPE_INVISIBLE_FRAME_BUFFER,
+	DC_MEM_ALLOC_TYPE_AGP
+};
+
+enum dc_link_encoding_format {
+	DC_LINK_ENCODING_UNSPECIFIED = 0,
+	DC_LINK_ENCODING_DP_8b_10b,
+	DC_LINK_ENCODING_DP_128b_132b,
+	DC_LINK_ENCODING_HDMI_TMDS,
+	DC_LINK_ENCODING_HDMI_FRL
+};
+
+enum dc_psr_version {
+	DC_PSR_VERSION_1			= 0,
+	DC_PSR_VERSION_SU_1			= 1,
+	DC_PSR_VERSION_UNSUPPORTED		= 0xFFFFFFFF,
+};
+
+/* Possible values of display_endpoint_id.endpoint */
+enum display_endpoint_type {
+	DISPLAY_ENDPOINT_PHY = 0, /* Physical connector. */
+	DISPLAY_ENDPOINT_USB4_DPIA, /* USB4 DisplayPort tunnel. */
+	DISPLAY_ENDPOINT_UNKNOWN = -1
+};
+
+/* Extends graphics_object_id with an additional member 'ep_type' for
+ * distinguishing between physical endpoints (with entries in BIOS connector table) and
+ * logical endpoints.
+ */
+struct display_endpoint_id {
+	struct graphics_object_id link_id;
+	enum display_endpoint_type ep_type;
+};
+
+enum backlight_control_type {
+	BACKLIGHT_CONTROL_PWM = 0,
+	BACKLIGHT_CONTROL_VESA_AUX = 1,
+	BACKLIGHT_CONTROL_AMD_AUX = 2,
+};
+
+#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY)
+#define MAX_CRC_WINDOW_NUM	2
+
+struct otg_phy_mux {
+	uint8_t phy_output_num;
+	uint8_t otg_output_num;
+};
+
+struct crc_window {
+	struct rect rect;
+	bool enable;
+};
+#endif
+
+enum dc_detect_reason {
+	DETECT_REASON_BOOT,
+	DETECT_REASON_RESUMEFROMS3S4,
+	DETECT_REASON_HPD,
+	DETECT_REASON_HPDRX,
+	DETECT_REASON_FALLBACK,
+	DETECT_REASON_RETRAIN,
+	DETECT_REASON_TDR,
+};
+
+struct dc_link_status {
+	bool link_active;
+	struct dpcd_caps *dpcd_caps;
+};
+
+union hdcp_rx_caps {
+	struct {
+		uint8_t version;
+		uint8_t reserved;
+		struct {
+			uint8_t repeater	: 1;
+			uint8_t hdcp_capable	: 1;
+			uint8_t reserved	: 6;
+		} byte0;
+	} fields;
+	uint8_t raw[3];
+};
+
+union hdcp_bcaps {
+	struct {
+		uint8_t HDCP_CAPABLE:1;
+		uint8_t REPEATER:1;
+		uint8_t RESERVED:6;
+	} bits;
+	uint8_t raw;
+};
+
+struct hdcp_caps {
+	union hdcp_rx_caps rx_caps;
+	union hdcp_bcaps bcaps;
+};
+
+/* DP MST stream allocation (payload bandwidth number) */
+struct link_mst_stream_allocation {
+	/* DIG front */
+	const struct stream_encoder *stream_enc;
+	/* HPO DP Stream Encoder */
+	const struct hpo_dp_stream_encoder *hpo_dp_stream_enc;
+	/* associate DRM payload table with DC stream encoder */
+	uint8_t vcp_id;
+	/* number of slots required for the DP stream in transport packet */
+	uint8_t slot_count;
+};
+
+#define MAX_CONTROLLER_NUM 6
+
+/* DP MST stream allocation table */
+struct link_mst_stream_allocation_table {
+	/* number of DP video streams */
+	int stream_count;
+	/* array of stream allocations */
+	struct link_mst_stream_allocation stream_allocations[MAX_CONTROLLER_NUM];
+};
+
+/* PSR feature flags */
+struct psr_settings {
+	bool psr_feature_enabled;		// PSR is supported by sink
+	bool psr_allow_active;			// PSR is currently active
+	enum dc_psr_version psr_version;		// Internal PSR version, determined based on DPCD
+	bool psr_vtotal_control_support;	// Vtotal control is supported by sink
+	unsigned long long psr_dirty_rects_change_timestamp_ns;	// for delay of enabling PSR-SU
+
+	/* These parameters are calculated in Driver,
+	 * based on display timing and Sink capabilities.
+	 * If VBLANK region is too small and Sink takes a long time
+	 * to set up RFB, it may take an extra frame to enter PSR state.
+	 */
+	bool psr_frame_capture_indication_req;
+	unsigned int psr_sdp_transmit_line_num_deadline;
+	uint8_t force_ffu_mode;
+	unsigned int psr_power_opt;
+
+	/**
+	 * Some panels cannot handle idle pattern during PSR entry.
+	 * To power down phy before disable stream to avoid sending
+	 * idle pattern.
+	 */
+	uint8_t power_down_phy_before_disable_stream;
+};
+
+enum replay_coasting_vtotal_type {
+	PR_COASTING_TYPE_NOM = 0,
+	PR_COASTING_TYPE_STATIC,
+	PR_COASTING_TYPE_FULL_SCREEN_VIDEO,
+	PR_COASTING_TYPE_TEST_HARNESS,
+	PR_COASTING_TYPE_NUM,
+};
+
+enum replay_link_off_frame_count_level {
+	PR_LINK_OFF_FRAME_COUNT_FAIL = 0x0,
+	PR_LINK_OFF_FRAME_COUNT_GOOD = 0x2,
+	PR_LINK_OFF_FRAME_COUNT_BEST = 0x6,
+};
+
+/*
+ * This is general Interface for Replay to
+ * set an 32 bit variable to dmub
+ * The Message_type indicates which variable
+ * passed to DMUB.
+ */
+enum replay_FW_Message_type {
+	Replay_Msg_Not_Support = -1,
+	Replay_Set_Timing_Sync_Supported,
+	Replay_Set_Residency_Frameupdate_Timer,
+	Replay_Set_Pseudo_VTotal,
+	Replay_Disabled_Adaptive_Sync_SDP,
+	Replay_Set_General_Cmd,
+};
+
+union replay_error_status {
+	struct {
+		unsigned int STATE_TRANSITION_ERROR     :1;
+		unsigned int LINK_CRC_ERROR             :1;
+		unsigned int DESYNC_ERROR               :1;
+		unsigned int RESERVED_3                 :1;
+		unsigned int LOW_RR_INCORRECT_VTOTAL    :1;
+		unsigned int NO_DOUBLED_RR              :1;
+		unsigned int RESERVED_6_7               :2;
+	} bits;
+	unsigned char raw;
+};
+
+union replay_low_refresh_rate_enable_options {
+	struct {
+	//BIT[0-3]: Replay Low Hz Support control
+		unsigned int ENABLE_LOW_RR_SUPPORT          :1;
+		unsigned int SKIP_ASIC_CHECK                :1;
+		unsigned int RESERVED_2_3                   :2;
+	//BIT[4-15]: Replay Low Hz Enable Scenarios
+		unsigned int ENABLE_STATIC_SCREEN           :1;
+		unsigned int ENABLE_FULL_SCREEN_VIDEO       :1;
+		unsigned int ENABLE_GENERAL_UI              :1;
+		unsigned int RESERVED_7_15                  :9;
+	//BIT[16-31]: Replay Low Hz Enable Check
+		unsigned int ENABLE_STATIC_FLICKER_CHECK    :1;
+		unsigned int RESERVED_17_31                 :15;
+	} bits;
+	unsigned int raw;
+};
+
+struct replay_config {
+	/* Replay feature is supported */
+	bool replay_supported;
+	/* Replay caps support DPCD & EDID caps*/
+	bool replay_cap_support;
+	/* Power opt flags that are supported */
+	unsigned int replay_power_opt_supported;
+	/* SMU optimization is supported */
+	bool replay_smu_opt_supported;
+	/* Replay enablement option */
+	unsigned int replay_enable_option;
+	/* Replay debug flags */
+	uint32_t debug_flags;
+	/* Replay sync is supported */
+	bool replay_timing_sync_supported;
+	/* Replay Disable desync error check. */
+	bool force_disable_desync_error_check;
+	/* Replay Received Desync Error HPD. */
+	bool received_desync_error_hpd;
+	/* Replay feature is supported long vblank */
+	bool replay_support_fast_resync_in_ultra_sleep_mode;
+	/* Replay error status */
+	union replay_error_status replay_error_status;
+	/* Replay Low Hz enable Options */
+	union replay_low_refresh_rate_enable_options low_rr_enable_options;
+	/* Replay coasting vtotal is within low refresh rate range. */
+	bool low_rr_activated;
+	/* Replay low refresh rate supported*/
+	bool low_rr_supported;
+	/* Replay Video Conferencing Optimization Enabled */
+	bool replay_video_conferencing_optimization_enabled;
+	/* Replay alpm mode */
+	enum dc_alpm_mode alpm_mode;
+	/* Replay full screen only */
+	bool os_request_force_ffu;
+};
+
+/* Replay feature flags*/
+struct replay_settings {
+	/* Replay configuration */
+	struct replay_config config;
+	/* Replay feature is ready for activating */
+	bool replay_feature_enabled;
+	/* Replay is currently active */
+	bool replay_allow_active;
+	/* Replay is currently active */
+	bool replay_allow_long_vblank;
+	/* Power opt flags that are activated currently */
+	unsigned int replay_power_opt_active;
+	/* SMU optimization is enabled */
+	bool replay_smu_opt_enable;
+	/* Current Coasting vtotal */
+	uint32_t coasting_vtotal;
+	/* Coasting vtotal table */
+	uint32_t coasting_vtotal_table[PR_COASTING_TYPE_NUM];
+	/* Defer Update Coasting vtotal table */
+	uint32_t defer_update_coasting_vtotal_table[PR_COASTING_TYPE_NUM];
+	/* Maximum link off frame count */
+	uint32_t link_off_frame_count;
+	/* Replay pseudo vtotal for low refresh rate*/
+	uint16_t low_rr_full_screen_video_pseudo_vtotal;
+	/* Replay last pseudo vtotal set to DMUB */
+	uint16_t last_pseudo_vtotal;
+	/* Replay desync error */
+	uint32_t replay_desync_error_fail_count;
+};
+
+/* To split out "global" and "per-panel" config settings.
+ * Add a struct dc_panel_config under dc_link
+ */
+struct dc_panel_config {
+	/* extra panel power sequence parameters */
+	struct pps {
+		unsigned int extra_t3_ms;
+		unsigned int extra_t7_ms;
+		unsigned int extra_delay_backlight_off;
+		unsigned int extra_post_t7_ms;
+		unsigned int extra_pre_t11_ms;
+		unsigned int extra_t12_ms;
+		unsigned int extra_post_OUI_ms;
+	} pps;
+	/* nit brightness */
+	struct nits_brightness {
+		unsigned int peak; /* nits */
+		unsigned int max_avg; /* nits */
+		unsigned int min; /* 1/10000 nits */
+		unsigned int max_nonboost_brightness_millinits;
+		unsigned int min_brightness_millinits;
+	} nits_brightness;
+	/* PSR */
+	struct psr {
+		bool disable_psr;
+		bool disallow_psrsu;
+		bool disallow_replay;
+		bool rc_disable;
+		bool rc_allow_static_screen;
+		bool rc_allow_fullscreen_VPB;
+		bool read_psrcap_again;
+		unsigned int replay_enable_option;
+	} psr;
+	/* ABM */
+	struct varib {
+		unsigned int varibright_feature_enable;
+		unsigned int def_varibright_level;
+		unsigned int abm_config_setting;
+	} varib;
+	/* edp DSC */
+	struct dsc {
+		bool disable_dsc_edp;
+		unsigned int force_dsc_edp_policy;
+	} dsc;
+	/* eDP ILR */
+	struct ilr {
+		bool optimize_edp_link_rate; /* eDP ILR */
+	} ilr;
+};
+
+#define MAX_SINKS_PER_LINK 4
+
+/*
+ *  USB4 DPIA BW ALLOCATION STRUCTS
+ */
+struct dc_dpia_bw_alloc {
+	int remote_sink_req_bw[MAX_SINKS_PER_LINK]; // BW requested by remote sinks
+	int link_verified_bw;  // The Verified BW that link can allocated and use that has been verified already
+	int link_max_bw;       // The Max BW that link can require/support
+	int allocated_bw;      // The Actual Allocated BW for this DPIA
+	int estimated_bw;      // The estimated available BW for this DPIA
+	int bw_granularity;    // BW Granularity
+	int dp_overhead;       // DP overhead in dp tunneling
+	bool bw_alloc_enabled; // The BW Alloc Mode Support is turned ON for all 3:  DP-Tx & Dpia & CM
+	uint8_t nrd_max_lane_count; // Non-reduced max lane count
+	uint8_t nrd_max_link_rate; // Non-reduced max link rate
+};
+
+enum dc_hpd_enable_select {
+	HPD_EN_FOR_ALL_EDP = 0,
+	HPD_EN_FOR_PRIMARY_EDP_ONLY,
+	HPD_EN_FOR_SECONDARY_EDP_ONLY,
+};
+
+enum dc_cm2_shaper_3dlut_setting {
+	DC_CM2_SHAPER_3DLUT_SETTING_BYPASS_ALL,
+	DC_CM2_SHAPER_3DLUT_SETTING_ENABLE_SHAPER,
+	/* Bypassing Shaper will always bypass 3DLUT */
+	DC_CM2_SHAPER_3DLUT_SETTING_ENABLE_SHAPER_3DLUT
+};
+
+enum dc_cm2_gpu_mem_layout {
+	DC_CM2_GPU_MEM_LAYOUT_3D_SWIZZLE_LINEAR_RGB,
+	DC_CM2_GPU_MEM_LAYOUT_3D_SWIZZLE_LINEAR_BGR,
+	DC_CM2_GPU_MEM_LAYOUT_1D_PACKED_LINEAR
+};
+
+enum dc_cm2_gpu_mem_pixel_component_order {
+	DC_CM2_GPU_MEM_PIXEL_COMPONENT_ORDER_RGBA,
+};
+
+enum dc_cm2_gpu_mem_format {
+	DC_CM2_GPU_MEM_FORMAT_16161616_UNORM_12MSB,
+	DC_CM2_GPU_MEM_FORMAT_16161616_UNORM_12LSB,
+	DC_CM2_GPU_MEM_FORMAT_16161616_FLOAT_FP1_5_10
+};
+
+struct dc_cm2_gpu_mem_format_parameters {
+	enum dc_cm2_gpu_mem_format format;
+	union {
+		struct {
+			/* bias & scale for float only */
+			uint16_t bias;
+			uint16_t scale;
+		} float_params;
+	};
+};
+
+enum dc_cm2_gpu_mem_size {
+	DC_CM2_GPU_MEM_SIZE_171717,
+	DC_CM2_GPU_MEM_SIZE_TRANSFORMED,
+};
+
+struct dc_cm2_gpu_mem_parameters {
+	struct dc_plane_address addr;
+	enum dc_cm2_gpu_mem_layout layout;
+	struct dc_cm2_gpu_mem_format_parameters format_params;
+	enum dc_cm2_gpu_mem_pixel_component_order component_order;
+	enum dc_cm2_gpu_mem_size  size;
+	uint16_t bit_depth;
+};
+
+enum dc_cm2_transfer_func_source {
+	DC_CM2_TRANSFER_FUNC_SOURCE_SYSMEM,
+	DC_CM2_TRANSFER_FUNC_SOURCE_VIDMEM
+};
+
+struct dc_cm2_component_settings {
+	enum dc_cm2_shaper_3dlut_setting shaper_3dlut_setting;
+	bool lut1d_enable;
+};
+
+/*
+ * All pointers in this struct must remain valid for as long as the 3DLUTs are used
+ */
+struct dc_cm2_func_luts {
+	const struct dc_transfer_func *shaper;
+	struct {
+		enum dc_cm2_transfer_func_source lut3d_src;
+		union {
+			const struct dc_3dlut *lut3d_func;
+			struct dc_cm2_gpu_mem_parameters gpu_mem_params;
+		};
+		bool rmcm_3dlut_shaper_select;
+		bool mpc_3dlut_enable;
+		bool rmcm_3dlut_enable;
+		bool mpc_mcm_post_blend;
+		uint8_t rmcm_tmz;
+	} lut3d_data;
+	const struct dc_transfer_func *lut1d_func;
+};
+
+struct dc_cm2_parameters {
+	struct dc_cm2_component_settings component_settings;
+	struct dc_cm2_func_luts cm2_luts;
+};
+
+enum mall_stream_type {
+	SUBVP_NONE, // subvp not in use
+	SUBVP_MAIN, // subvp in use, this stream is main stream
+	SUBVP_PHANTOM, // subvp in use, this stream is a phantom stream
+};
+
+enum dc_power_source_type {
+	DC_POWER_SOURCE_AC, // wall power
+	DC_POWER_SOURCE_DC, // battery power
+};
+
+struct dc_state_create_params {
+	enum dc_power_source_type power_source;
+};
+
+struct dc_commit_streams_params {
+	struct dc_stream_state **streams;
+	uint8_t stream_count;
+	enum dc_power_source_type power_source;
+};
+
+struct set_backlight_level_params {
+	/* backlight in pwm */
+	uint32_t backlight_pwm_u16_16;
+	/* brightness ramping */
+	uint32_t frame_ramp;
+	/* backlight control type
+	 * 0: PWM backlight control
+	 * 1: VESA AUX backlight control
+	 * 2: AMD AUX backlight control
+	 */
+	enum backlight_control_type control_type;
+	/* backlight in millinits */
+	uint32_t backlight_millinits;
+	/* transition time in ms */
+	uint32_t transition_time_in_ms;
+	/* minimum luminance in nits */
+	uint32_t min_luminance;
+	/* maximum luminance in nits */
+	uint32_t max_luminance;
+	/* minimum backlight in pwm */
+	uint32_t min_backlight_pwm;
+	/* maximum backlight in pwm */
+	uint32_t max_backlight_pwm;
+	/* AUX HW instance */
+	uint8_t aux_inst;
+};
+
+enum dc_validate_mode {
+	/* validate the mode and program HW */
+	DC_VALIDATE_MODE_AND_PROGRAMMING = 0,
+	/* only validate the mode */
+	DC_VALIDATE_MODE_ONLY = 1,
+	/* validate the mode and get the max state (voltage level) */
+	DC_VALIDATE_MODE_AND_STATE_INDEX = 2,
+};
+
+struct dc_validation_dpia_set {
+	const struct dc_link *link;
+	const struct dc_tunnel_settings *tunnel_settings;
+	uint32_t required_bw;
+};
+
+#endif /* DC_TYPES_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/Makefile b/drivers/gpu/drm/amd/display/dc/dccg/Makefile
new file mode 100644
index 000000000000..1d5cf0f8e79d
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/Makefile
@@ -0,0 +1,103 @@
+
+# Copyright 2022 Advanced Micro Devices, Inc.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+# Makefile for the 'dccg' sub-component of DAL.
+#
+ifdef CONFIG_DRM_AMD_DC_FP
+###############################################################################
+# DCN
+###############################################################################
+
+DCCG_DCN20 = dcn20_dccg.o
+
+AMD_DAL_DCCG_DCN20 = $(addprefix $(AMDDALPATH)/dc/dccg/dcn20/,$(DCCG_DCN20))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCCG_DCN20)
+
+###############################################################################
+
+DCCG_DCN201 = dcn201_dccg.o
+
+AMD_DAL_DCCG_DCN201 = $(addprefix $(AMDDALPATH)/dc/dccg/dcn201/,$(DCCG_DCN201))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCCG_DCN201)
+
+###############################################################################
+
+DCCG_DCN21 = dcn21_dccg.o
+
+AMD_DAL_DCCG_DCN21 = $(addprefix $(AMDDALPATH)/dc/dccg/dcn21/,$(DCCG_DCN21))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCCG_DCN21)
+
+###############################################################################
+DCCG_DCN30 = dcn30_dccg.o
+
+AMD_DAL_DCCG_DCN30 = $(addprefix $(AMDDALPATH)/dc/dccg/dcn30/,$(DCCG_DCN30))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCCG_DCN30)
+
+###############################################################################
+DCCG_DCN301 = dcn301_dccg.o
+
+AMD_DAL_DCCG_DCN301 = $(addprefix $(AMDDALPATH)/dc/dccg/dcn301/,$(DCCG_DCN301))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCCG_DCN301)
+
+###############################################################################
+
+DCCG_DCN31 = dcn31_dccg.o
+
+AMD_DAL_DCCG_DCN31 = $(addprefix $(AMDDALPATH)/dc/dccg/dcn31/,$(DCCG_DCN31))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCCG_DCN31)
+
+###############################################################################
+
+DCCG_DCN314 = dcn314_dccg.o
+
+AMD_DAL_DCCG_DCN314 = $(addprefix $(AMDDALPATH)/dc/dccg/dcn314/,$(DCCG_DCN314))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCCG_DCN314)
+
+###############################################################################
+
+DCCG_DCN32 = dcn32_dccg.o
+
+AMD_DAL_DCCG_DCN32 = $(addprefix $(AMDDALPATH)/dc/dccg/dcn32/,$(DCCG_DCN32))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCCG_DCN32)
+
+###############################################################################
+
+DCCG_DCN35 = dcn35_dccg.o
+
+AMD_DAL_DCCG_DCN35 = $(addprefix $(AMDDALPATH)/dc/dccg/dcn35/,$(DCCG_DCN35))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCCG_DCN35)
+
+###############################################################################
+DCCG_DCN401 = dcn401_dccg.o
+
+AMD_DAL_DCCG_DCN401 = $(addprefix $(AMDDALPATH)/dc/dccg/dcn401/,$(DCCG_DCN401))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCCG_DCN401)
+endif
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn20/dcn20_dccg.c b/drivers/gpu/drm/amd/display/dc/dccg/dcn20/dcn20_dccg.c
new file mode 100644
index 000000000000..5999b2da3a01
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn20/dcn20_dccg.c
@@ -0,0 +1,176 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include <linux/slab.h>
+
+#include "reg_helper.h"
+#include "core_types.h"
+#include "dcn20_dccg.h"
+
+#define TO_DCN_DCCG(dccg)\
+	container_of(dccg, struct dcn_dccg, base)
+
+#define REG(reg) \
+	(dccg_dcn->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	dccg_dcn->dccg_shift->field_name, dccg_dcn->dccg_mask->field_name
+
+#define CTX \
+	dccg_dcn->base.ctx
+#define DC_LOGGER \
+	dccg->ctx->logger
+
+void dccg2_update_dpp_dto(struct dccg *dccg, int dpp_inst, int req_dppclk)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (dccg->ref_dppclk && req_dppclk) {
+		int ref_dppclk = dccg->ref_dppclk;
+		int modulo, phase;
+
+		// phase / modulo = dpp pipe clk / dpp global clk
+		modulo = 0xff;   // use FF at the end
+		phase = ((modulo * req_dppclk) + ref_dppclk - 1) / ref_dppclk;
+
+		if (phase > 0xff) {
+			ASSERT(false);
+			phase = 0xff;
+		}
+
+		REG_SET_2(DPPCLK_DTO_PARAM[dpp_inst], 0,
+				DPPCLK0_DTO_PHASE, phase,
+				DPPCLK0_DTO_MODULO, modulo);
+		REG_UPDATE(DPPCLK_DTO_CTRL,
+				DPPCLK_DTO_ENABLE[dpp_inst], 1);
+	} else {
+		REG_UPDATE(DPPCLK_DTO_CTRL,
+				DPPCLK_DTO_ENABLE[dpp_inst], 0);
+	}
+
+	dccg->pipe_dppclk_khz[dpp_inst] = req_dppclk;
+}
+
+void dccg2_get_dccg_ref_freq(struct dccg *dccg,
+		unsigned int xtalin_freq_inKhz,
+		unsigned int *dccg_ref_freq_inKhz)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+	uint32_t clk_en = 0;
+	uint32_t clk_sel = 0;
+
+	REG_GET_2(REFCLK_CNTL, REFCLK_CLOCK_EN, &clk_en, REFCLK_SRC_SEL, &clk_sel);
+
+	if (clk_en != 0) {
+		// DCN20 has never been validated for non-xtalin as reference
+		// frequency.  There's actually no way for DC to determine what
+		// frequency a non-xtalin source is.
+		ASSERT_CRITICAL(false);
+	}
+
+	*dccg_ref_freq_inKhz = xtalin_freq_inKhz;
+
+	return;
+}
+
+void dccg2_set_fifo_errdet_ovr_en(struct dccg *dccg,
+		bool en)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	REG_UPDATE(DISPCLK_FREQ_CHANGE_CNTL,
+			DCCG_FIFO_ERRDET_OVR_EN, en ? 1 : 0);
+}
+
+void dccg2_otg_add_pixel(struct dccg *dccg,
+		uint32_t otg_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	REG_UPDATE_2(OTG_PIXEL_RATE_CNTL[otg_inst],
+			OTG_ADD_PIXEL[otg_inst], 0,
+			OTG_DROP_PIXEL[otg_inst], 0);
+	REG_UPDATE(OTG_PIXEL_RATE_CNTL[otg_inst],
+			OTG_ADD_PIXEL[otg_inst], 1);
+}
+
+void dccg2_otg_drop_pixel(struct dccg *dccg,
+		uint32_t otg_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	REG_UPDATE_2(OTG_PIXEL_RATE_CNTL[otg_inst],
+			OTG_ADD_PIXEL[otg_inst], 0,
+			OTG_DROP_PIXEL[otg_inst], 0);
+	REG_UPDATE(OTG_PIXEL_RATE_CNTL[otg_inst],
+			OTG_DROP_PIXEL[otg_inst], 1);
+}
+
+void dccg2_init(struct dccg *dccg)
+{
+}
+
+static const struct dccg_funcs dccg2_funcs = {
+	.update_dpp_dto = dccg2_update_dpp_dto,
+	.get_dccg_ref_freq = dccg2_get_dccg_ref_freq,
+	.set_fifo_errdet_ovr_en = dccg2_set_fifo_errdet_ovr_en,
+	.otg_add_pixel = dccg2_otg_add_pixel,
+	.otg_drop_pixel = dccg2_otg_drop_pixel,
+	.dccg_init = dccg2_init
+};
+
+struct dccg *dccg2_create(
+	struct dc_context *ctx,
+	const struct dccg_registers *regs,
+	const struct dccg_shift *dccg_shift,
+	const struct dccg_mask *dccg_mask)
+{
+	struct dcn_dccg *dccg_dcn = kzalloc(sizeof(*dccg_dcn), GFP_ATOMIC);
+	struct dccg *base;
+
+	if (dccg_dcn == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	base = &dccg_dcn->base;
+	base->ctx = ctx;
+	base->funcs = &dccg2_funcs;
+
+	dccg_dcn->regs = regs;
+	dccg_dcn->dccg_shift = dccg_shift;
+	dccg_dcn->dccg_mask = dccg_mask;
+
+	return &dccg_dcn->base;
+}
+
+void dcn_dccg_destroy(struct dccg **dccg)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(*dccg);
+
+	kfree(dccg_dcn);
+	*dccg = NULL;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn20/dcn20_dccg.h b/drivers/gpu/drm/amd/display/dc/dccg/dcn20/dcn20_dccg.h
new file mode 100644
index 000000000000..a9b88f5e0c04
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn20/dcn20_dccg.h
@@ -0,0 +1,464 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCN20_DCCG_H__
+#define __DCN20_DCCG_H__
+
+#include "dccg.h"
+
+#define DCCG_COMMON_REG_LIST_DCN_BASE() \
+	SR(DPPCLK_DTO_CTRL),\
+	DCCG_SRII(DTO_PARAM, DPPCLK, 0),\
+	DCCG_SRII(DTO_PARAM, DPPCLK, 1),\
+	DCCG_SRII(DTO_PARAM, DPPCLK, 2),\
+	DCCG_SRII(DTO_PARAM, DPPCLK, 3),\
+	SR(REFCLK_CNTL),\
+	DCCG_SRII(PIXEL_RATE_CNTL, OTG, 0),\
+	DCCG_SRII(PIXEL_RATE_CNTL, OTG, 1),\
+	SR(DISPCLK_FREQ_CHANGE_CNTL)
+
+#define DCCG_REG_LIST_DCN2() \
+	DCCG_COMMON_REG_LIST_DCN_BASE(),\
+	DCCG_SRII(DTO_PARAM, DPPCLK, 4),\
+	DCCG_SRII(DTO_PARAM, DPPCLK, 5),\
+	DCCG_SRII(PIXEL_RATE_CNTL, OTG, 2),\
+	DCCG_SRII(PIXEL_RATE_CNTL, OTG, 3),\
+	DCCG_SRII(PIXEL_RATE_CNTL, OTG, 4),\
+	DCCG_SRII(PIXEL_RATE_CNTL, OTG, 5)
+
+#define DCCG_SF(reg_name, field_name, post_fix)\
+	.field_name = reg_name ## __ ## field_name ## post_fix
+
+#define DCCG_SFI(reg_name, field_name, field_prefix, inst, post_fix)\
+	.field_prefix ## _ ## field_name[inst] = reg_name ## __ ## field_prefix ## inst ## _ ## field_name ## post_fix
+
+#define DCCG_SFII(block, reg_name, field_prefix, field_name, inst, post_fix)\
+	.field_prefix ## _ ## field_name[inst] = block ## inst ## _ ## reg_name ## __ ## field_prefix ## inst ## _ ## field_name ## post_fix
+
+#define DCCG_COMMON_MASK_SH_LIST_DCN_COMMON_BASE(mask_sh) \
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 0, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 0, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 1, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 1, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 2, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 2, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 3, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 3, mask_sh),\
+	DCCG_SF(DPPCLK0_DTO_PARAM, DPPCLK0_DTO_PHASE, mask_sh),\
+	DCCG_SF(DPPCLK0_DTO_PARAM, DPPCLK0_DTO_MODULO, mask_sh),\
+	DCCG_SF(REFCLK_CNTL, REFCLK_CLOCK_EN, mask_sh),\
+	DCCG_SF(REFCLK_CNTL, REFCLK_SRC_SEL, mask_sh),\
+	DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DISPCLK_STEP_DELAY, mask_sh),\
+	DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DISPCLK_STEP_SIZE, mask_sh),\
+	DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DISPCLK_FREQ_RAMP_DONE, mask_sh),\
+	DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DISPCLK_MAX_ERRDET_CYCLES, mask_sh),\
+	DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DCCG_FIFO_ERRDET_RESET, mask_sh),\
+	DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DCCG_FIFO_ERRDET_STATE, mask_sh),\
+	DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DCCG_FIFO_ERRDET_OVR_EN, mask_sh),\
+	DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DISPCLK_CHG_FWD_CORR_DISABLE, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, DROP_PIXEL, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, DROP_PIXEL, 1, mask_sh)
+
+
+
+
+#define DCCG_MASK_SH_LIST_DCN2(mask_sh) \
+	DCCG_COMMON_MASK_SH_LIST_DCN_COMMON_BASE(mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 4, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 4, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 5, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 5, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 3, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 4, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 5, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, DROP_PIXEL, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, DROP_PIXEL, 3, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, DROP_PIXEL, 4, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, DROP_PIXEL, 5, mask_sh)
+
+#define DCCG_MASK_SH_LIST_DCN2_1(mask_sh) \
+	DCCG_COMMON_MASK_SH_LIST_DCN_COMMON_BASE(mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 4, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 4, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 5, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 5, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 3, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, DROP_PIXEL, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, DROP_PIXEL, 3, mask_sh)
+
+
+#define DCCG_REG_FIELD_LIST(type) \
+	type DPPCLK0_DTO_PHASE;\
+	type DPPCLK0_DTO_MODULO;\
+	type DPPCLK_DTO_ENABLE[6];\
+	type DPPCLK_DTO_DB_EN[6];\
+	type REFCLK_CLOCK_EN;\
+	type REFCLK_SRC_SEL;\
+	type DISPCLK_STEP_DELAY;\
+	type DISPCLK_STEP_SIZE;\
+	type DISPCLK_FREQ_RAMP_DONE;\
+	type DISPCLK_MAX_ERRDET_CYCLES;\
+	type DCCG_FIFO_ERRDET_RESET;\
+	type DCCG_FIFO_ERRDET_STATE;\
+	type DCCG_FIFO_ERRDET_OVR_EN;\
+	type DISPCLK_CHG_FWD_CORR_DISABLE;\
+	type DISPCLK_FREQ_CHANGE_CNTL;\
+	type OTG_ADD_PIXEL[MAX_PIPES];\
+	type OTG_DROP_PIXEL[MAX_PIPES];
+
+#define DCCG3_REG_FIELD_LIST(type) \
+	type HDMICHARCLK0_EN;\
+	type HDMICHARCLK0_SRC_SEL;\
+	type PHYASYMCLK_FORCE_EN;\
+	type PHYASYMCLK_FORCE_SRC_SEL;\
+	type PHYBSYMCLK_FORCE_EN;\
+	type PHYBSYMCLK_FORCE_SRC_SEL;\
+	type PHYCSYMCLK_FORCE_EN;\
+	type PHYCSYMCLK_FORCE_SRC_SEL;
+
+#define DCCG31_REG_FIELD_LIST(type) \
+	type PHYDSYMCLK_FORCE_EN;\
+	type PHYDSYMCLK_FORCE_SRC_SEL;\
+	type PHYESYMCLK_FORCE_EN;\
+	type PHYESYMCLK_FORCE_SRC_SEL;\
+	type DPSTREAMCLK_PIPE0_EN;\
+	type DPSTREAMCLK_PIPE1_EN;\
+	type DPSTREAMCLK_PIPE2_EN;\
+	type DPSTREAMCLK_PIPE3_EN;\
+	type HDMISTREAMCLK0_SRC_SEL;\
+	type HDMISTREAMCLK0_DTO_FORCE_DIS;\
+	type SYMCLK32_SE0_SRC_SEL;\
+	type SYMCLK32_SE1_SRC_SEL;\
+	type SYMCLK32_SE2_SRC_SEL;\
+	type SYMCLK32_SE3_SRC_SEL;\
+	type SYMCLK32_SE0_EN;\
+	type SYMCLK32_SE1_EN;\
+	type SYMCLK32_SE2_EN;\
+	type SYMCLK32_SE3_EN;\
+	type SYMCLK32_LE0_SRC_SEL;\
+	type SYMCLK32_LE1_SRC_SEL;\
+	type SYMCLK32_LE0_EN;\
+	type SYMCLK32_LE1_EN;\
+	type DTBCLK_DTO_ENABLE[MAX_PIPES];\
+	type DTBCLKDTO_ENABLE_STATUS[MAX_PIPES];\
+	type PIPE_DTO_SRC_SEL[MAX_PIPES];\
+	type DTBCLK_DTO_DIV[MAX_PIPES];\
+	type DCCG_AUDIO_DTO_SEL;\
+	type DCCG_AUDIO_DTO0_SOURCE_SEL;\
+	type DENTIST_DISPCLK_CHG_MODE;\
+	type DSCCLK0_DTO_PHASE;\
+	type DSCCLK0_DTO_MODULO;\
+	type DSCCLK1_DTO_PHASE;\
+	type DSCCLK1_DTO_MODULO;\
+	type DSCCLK2_DTO_PHASE;\
+	type DSCCLK2_DTO_MODULO;\
+	type DSCCLK0_DTO_ENABLE;\
+	type DSCCLK1_DTO_ENABLE;\
+	type DSCCLK2_DTO_ENABLE;\
+	type SYMCLK32_ROOT_SE0_GATE_DISABLE;\
+	type SYMCLK32_ROOT_SE1_GATE_DISABLE;\
+	type SYMCLK32_ROOT_SE2_GATE_DISABLE;\
+	type SYMCLK32_ROOT_SE3_GATE_DISABLE;\
+	type SYMCLK32_SE0_GATE_DISABLE;\
+	type SYMCLK32_SE1_GATE_DISABLE;\
+	type SYMCLK32_SE2_GATE_DISABLE;\
+	type SYMCLK32_SE3_GATE_DISABLE;\
+	type SYMCLK32_ROOT_LE0_GATE_DISABLE;\
+	type SYMCLK32_ROOT_LE1_GATE_DISABLE;\
+	type SYMCLK32_LE0_GATE_DISABLE;\
+	type SYMCLK32_LE1_GATE_DISABLE;\
+	type DPSTREAMCLK_ROOT_GATE_DISABLE;\
+	type DPSTREAMCLK_GATE_DISABLE;\
+	type HDMISTREAMCLK0_DTO_PHASE;\
+	type HDMISTREAMCLK0_DTO_MODULO;\
+	type HDMICHARCLK0_GATE_DISABLE;\
+	type HDMICHARCLK0_ROOT_GATE_DISABLE; \
+	type PHYASYMCLK_GATE_DISABLE; \
+	type PHYBSYMCLK_GATE_DISABLE; \
+	type PHYCSYMCLK_GATE_DISABLE; \
+	type PHYDSYMCLK_GATE_DISABLE; \
+	type PHYESYMCLK_GATE_DISABLE;
+
+#define DCCG314_REG_FIELD_LIST(type) \
+	type DSCCLK3_DTO_PHASE;\
+	type DSCCLK3_DTO_MODULO;\
+	type DSCCLK3_DTO_ENABLE;\
+	type DENTIST_DISPCLK_RDIVIDER;\
+	type DENTIST_DISPCLK_WDIVIDER;
+
+#define DCCG32_REG_FIELD_LIST(type) \
+	type DPSTREAMCLK0_EN;\
+	type DPSTREAMCLK1_EN;\
+	type DPSTREAMCLK2_EN;\
+	type DPSTREAMCLK3_EN;\
+	type DPSTREAMCLK0_SRC_SEL;\
+	type DPSTREAMCLK1_SRC_SEL;\
+	type DPSTREAMCLK2_SRC_SEL;\
+	type DPSTREAMCLK3_SRC_SEL;\
+	type HDMISTREAMCLK0_EN;\
+	type OTG0_PIXEL_RATE_DIVK1;\
+	type OTG0_PIXEL_RATE_DIVK2;\
+	type OTG1_PIXEL_RATE_DIVK1;\
+	type OTG1_PIXEL_RATE_DIVK2;\
+	type OTG2_PIXEL_RATE_DIVK1;\
+	type OTG2_PIXEL_RATE_DIVK2;\
+	type OTG3_PIXEL_RATE_DIVK1;\
+	type OTG3_PIXEL_RATE_DIVK2;\
+	type DTBCLK_P0_SRC_SEL;\
+	type DTBCLK_P0_EN;\
+	type DTBCLK_P1_SRC_SEL;\
+	type DTBCLK_P1_EN;\
+	type DTBCLK_P2_SRC_SEL;\
+	type DTBCLK_P2_EN;\
+	type DTBCLK_P3_SRC_SEL;\
+	type DTBCLK_P3_EN;\
+	type DENTIST_DISPCLK_CHG_DONE;
+
+#define DCCG35_REG_FIELD_LIST(type) \
+	type DPPCLK0_EN;\
+	type DPPCLK1_EN;\
+	type DPPCLK2_EN;\
+	type DPPCLK3_EN;\
+	type DSCCLK0_EN;\
+	type DSCCLK1_EN;\
+	type DSCCLK2_EN;\
+	type DSCCLK3_EN;\
+	type DISPCLK_DCCG_GATE_DISABLE;\
+	type DCCG_GLOBAL_FGCG_REP_DIS; \
+	type PHYASYMCLK_EN;\
+	type PHYASYMCLK_SRC_SEL;\
+	type PHYBSYMCLK_EN;\
+	type PHYBSYMCLK_SRC_SEL;\
+	type PHYCSYMCLK_EN;\
+	type PHYCSYMCLK_SRC_SEL;\
+	type PHYDSYMCLK_EN;\
+	type PHYDSYMCLK_SRC_SEL;\
+	type PHYESYMCLK_EN;\
+	type PHYESYMCLK_SRC_SEL;\
+	type PHYASYMCLK_ROOT_GATE_DISABLE;\
+	type PHYBSYMCLK_ROOT_GATE_DISABLE;\
+	type PHYCSYMCLK_ROOT_GATE_DISABLE;\
+	type PHYDSYMCLK_ROOT_GATE_DISABLE;\
+	type PHYESYMCLK_ROOT_GATE_DISABLE;\
+	type HDMISTREAMCLK0_GATE_DISABLE;\
+	type HDMISTREAMCLK1_GATE_DISABLE;\
+	type HDMISTREAMCLK2_GATE_DISABLE;\
+	type HDMISTREAMCLK3_GATE_DISABLE;\
+	type HDMISTREAMCLK4_GATE_DISABLE;\
+	type HDMISTREAMCLK5_GATE_DISABLE;\
+	type SYMCLKA_CLOCK_ENABLE;\
+	type SYMCLKB_CLOCK_ENABLE;\
+	type SYMCLKC_CLOCK_ENABLE;\
+	type SYMCLKD_CLOCK_ENABLE;\
+	type SYMCLKE_CLOCK_ENABLE;\
+	type SYMCLKA_FE_EN;\
+	type SYMCLKB_FE_EN;\
+	type SYMCLKC_FE_EN;\
+	type SYMCLKD_FE_EN;\
+	type SYMCLKE_FE_EN;\
+	type SYMCLKA_SRC_SEL;\
+	type SYMCLKB_SRC_SEL;\
+	type SYMCLKC_SRC_SEL;\
+	type SYMCLKD_SRC_SEL;\
+	type SYMCLKE_SRC_SEL;\
+	type SYMCLKA_FE_SRC_SEL;\
+	type SYMCLKB_FE_SRC_SEL;\
+	type SYMCLKC_FE_SRC_SEL;\
+	type SYMCLKD_FE_SRC_SEL;\
+	type SYMCLKE_FE_SRC_SEL;\
+	type DTBCLK_P0_GATE_DISABLE;\
+	type DTBCLK_P1_GATE_DISABLE;\
+	type DTBCLK_P2_GATE_DISABLE;\
+	type DTBCLK_P3_GATE_DISABLE;\
+	type DSCCLK0_ROOT_GATE_DISABLE;\
+	type DSCCLK1_ROOT_GATE_DISABLE;\
+	type DSCCLK2_ROOT_GATE_DISABLE;\
+	type DSCCLK3_ROOT_GATE_DISABLE;\
+	type SYMCLKA_FE_ROOT_GATE_DISABLE;\
+	type SYMCLKB_FE_ROOT_GATE_DISABLE;\
+	type SYMCLKC_FE_ROOT_GATE_DISABLE;\
+	type SYMCLKD_FE_ROOT_GATE_DISABLE;\
+	type SYMCLKE_FE_ROOT_GATE_DISABLE;\
+	type DPPCLK0_ROOT_GATE_DISABLE;\
+	type DPPCLK1_ROOT_GATE_DISABLE;\
+	type DPPCLK2_ROOT_GATE_DISABLE;\
+	type DPPCLK3_ROOT_GATE_DISABLE;\
+	type HDMISTREAMCLK0_ROOT_GATE_DISABLE;\
+	type SYMCLKA_ROOT_GATE_DISABLE;\
+	type SYMCLKB_ROOT_GATE_DISABLE;\
+	type SYMCLKC_ROOT_GATE_DISABLE;\
+	type SYMCLKD_ROOT_GATE_DISABLE;\
+	type SYMCLKE_ROOT_GATE_DISABLE;\
+	type PHYA_REFCLK_ROOT_GATE_DISABLE;\
+	type PHYB_REFCLK_ROOT_GATE_DISABLE;\
+	type PHYC_REFCLK_ROOT_GATE_DISABLE;\
+	type PHYD_REFCLK_ROOT_GATE_DISABLE;\
+	type PHYE_REFCLK_ROOT_GATE_DISABLE;\
+	type DPSTREAMCLK0_ROOT_GATE_DISABLE;\
+	type DPSTREAMCLK1_ROOT_GATE_DISABLE;\
+	type DPSTREAMCLK2_ROOT_GATE_DISABLE;\
+	type DPSTREAMCLK3_ROOT_GATE_DISABLE;\
+	type DPSTREAMCLK0_GATE_DISABLE;\
+	type DPSTREAMCLK1_GATE_DISABLE;\
+	type DPSTREAMCLK2_GATE_DISABLE;\
+	type DPSTREAMCLK3_GATE_DISABLE;\
+	type SYMCLKA_FE_GATE_DISABLE;\
+	type SYMCLKB_FE_GATE_DISABLE;\
+	type SYMCLKC_FE_GATE_DISABLE;\
+	type SYMCLKD_FE_GATE_DISABLE;\
+	type SYMCLKE_FE_GATE_DISABLE;\
+	type SYMCLKA_GATE_DISABLE;\
+	type SYMCLKB_GATE_DISABLE;\
+	type SYMCLKC_GATE_DISABLE;\
+	type SYMCLKD_GATE_DISABLE;\
+	type SYMCLKE_GATE_DISABLE;\
+
+
+#define DCCG401_REG_FIELD_LIST(type) \
+	type OTG0_TMDS_PIXEL_RATE_DIV;\
+	type DPDTO0_INT;\
+	type OTG1_TMDS_PIXEL_RATE_DIV;\
+	type DPDTO1_INT;\
+	type OTG2_TMDS_PIXEL_RATE_DIV;\
+	type DPDTO2_INT;\
+	type OTG3_TMDS_PIXEL_RATE_DIV;\
+	type DPDTO3_INT;\
+	type SYMCLK32_ROOT_LE2_GATE_DISABLE;\
+	type SYMCLK32_ROOT_LE3_GATE_DISABLE;\
+	type SYMCLK32_LE2_GATE_DISABLE;\
+	type SYMCLK32_LE3_GATE_DISABLE;\
+	type SYMCLK32_LE2_SRC_SEL;\
+	type SYMCLK32_LE3_SRC_SEL;\
+	type SYMCLK32_LE2_EN;\
+	type SYMCLK32_LE3_EN;\
+	type DP_DTO_ENABLE[MAX_PIPES];
+
+struct dccg_shift {
+	DCCG_REG_FIELD_LIST(uint8_t)
+	DCCG3_REG_FIELD_LIST(uint8_t)
+	DCCG31_REG_FIELD_LIST(uint8_t)
+	DCCG314_REG_FIELD_LIST(uint8_t)
+	DCCG32_REG_FIELD_LIST(uint8_t)
+	DCCG35_REG_FIELD_LIST(uint8_t)
+	DCCG401_REG_FIELD_LIST(uint8_t)
+};
+
+struct dccg_mask {
+	DCCG_REG_FIELD_LIST(uint32_t)
+	DCCG3_REG_FIELD_LIST(uint32_t)
+	DCCG31_REG_FIELD_LIST(uint32_t)
+	DCCG314_REG_FIELD_LIST(uint32_t)
+	DCCG32_REG_FIELD_LIST(uint32_t)
+	DCCG35_REG_FIELD_LIST(uint32_t)
+	DCCG401_REG_FIELD_LIST(uint32_t)
+};
+
+#define DCCG_REG_VARIABLE_LIST \
+	uint32_t DPPCLK_DTO_CTRL; \
+	uint32_t DPPCLK_DTO_PARAM[6]; \
+	uint32_t REFCLK_CNTL; \
+	uint32_t DISPCLK_FREQ_CHANGE_CNTL; \
+	uint32_t OTG_PIXEL_RATE_CNTL[MAX_PIPES]; \
+	uint32_t HDMICHARCLK_CLOCK_CNTL[6]; \
+	uint32_t PHYASYMCLK_CLOCK_CNTL; \
+	uint32_t PHYBSYMCLK_CLOCK_CNTL; \
+	uint32_t PHYCSYMCLK_CLOCK_CNTL; \
+	uint32_t PHYDSYMCLK_CLOCK_CNTL; \
+	uint32_t PHYESYMCLK_CLOCK_CNTL; \
+	uint32_t DTBCLK_DTO_MODULO[MAX_PIPES]; \
+	uint32_t DTBCLK_DTO_PHASE[MAX_PIPES]; \
+	uint32_t DCCG_AUDIO_DTBCLK_DTO_MODULO; \
+	uint32_t DCCG_AUDIO_DTBCLK_DTO_PHASE; \
+	uint32_t DCCG_AUDIO_DTO_SOURCE; \
+	uint32_t DPSTREAMCLK_CNTL; \
+	uint32_t HDMISTREAMCLK_CNTL; \
+	uint32_t SYMCLK32_SE_CNTL; \
+	uint32_t SYMCLK32_LE_CNTL; \
+	uint32_t DENTIST_DISPCLK_CNTL; \
+	uint32_t DSCCLK_DTO_CTRL; \
+	uint32_t DSCCLK0_DTO_PARAM; \
+	uint32_t DSCCLK1_DTO_PARAM; \
+	uint32_t DSCCLK2_DTO_PARAM; \
+	uint32_t DSCCLK3_DTO_PARAM; \
+	uint32_t DPSTREAMCLK_ROOT_GATE_DISABLE; \
+	uint32_t DPSTREAMCLK_GATE_DISABLE; \
+	uint32_t DCCG_GATE_DISABLE_CNTL; \
+	uint32_t DCCG_GATE_DISABLE_CNTL2; \
+	uint32_t DCCG_GATE_DISABLE_CNTL3; \
+	uint32_t HDMISTREAMCLK0_DTO_PARAM; \
+	uint32_t DCCG_GATE_DISABLE_CNTL4; \
+	uint32_t OTG_PIXEL_RATE_DIV; \
+	uint32_t DTBCLK_P_CNTL; \
+	uint32_t DPPCLK_CTRL; \
+	uint32_t DCCG_GATE_DISABLE_CNTL5; \
+	uint32_t DCCG_GATE_DISABLE_CNTL6; \
+	uint32_t DCCG_GLOBAL_FGCG_REP_CNTL; \
+	uint32_t SYMCLKA_CLOCK_ENABLE; \
+	uint32_t SYMCLKB_CLOCK_ENABLE; \
+	uint32_t SYMCLKC_CLOCK_ENABLE; \
+	uint32_t SYMCLKD_CLOCK_ENABLE; \
+	uint32_t SYMCLKE_CLOCK_ENABLE; \
+	uint32_t DP_DTO_MODULO[MAX_PIPES]; \
+	uint32_t DP_DTO_PHASE[MAX_PIPES]
+struct dccg_registers {
+	DCCG_REG_VARIABLE_LIST;
+};
+
+struct dcn_dccg {
+	struct dccg base;
+	const struct dccg_registers *regs;
+	const struct dccg_shift *dccg_shift;
+	const struct dccg_mask *dccg_mask;
+};
+
+void dccg2_update_dpp_dto(struct dccg *dccg, int dpp_inst, int req_dppclk);
+
+void dccg2_get_dccg_ref_freq(struct dccg *dccg,
+		unsigned int xtalin_freq_inKhz,
+		unsigned int *dccg_ref_freq_inKhz);
+
+void dccg2_set_fifo_errdet_ovr_en(struct dccg *dccg,
+		bool en);
+void dccg2_otg_add_pixel(struct dccg *dccg,
+		uint32_t otg_inst);
+void dccg2_otg_drop_pixel(struct dccg *dccg,
+		uint32_t otg_inst);
+
+
+void dccg2_init(struct dccg *dccg);
+
+struct dccg *dccg2_create(
+	struct dc_context *ctx,
+	const struct dccg_registers *regs,
+	const struct dccg_shift *dccg_shift,
+	const struct dccg_mask *dccg_mask);
+
+void dcn_dccg_destroy(struct dccg **dccg);
+
+#endif //__DCN20_DCCG_H__
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn201/dcn201_dccg.c b/drivers/gpu/drm/amd/display/dc/dccg/dcn201/dcn201_dccg.c
new file mode 100644
index 000000000000..9a3402148fde
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn201/dcn201_dccg.c
@@ -0,0 +1,85 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dcn201_dccg.h"
+
+#include "reg_helper.h"
+#include "core_types.h"
+
+#define TO_DCN_DCCG(dccg)\
+	container_of(dccg, struct dcn_dccg, base)
+
+#define REG(reg) \
+	(dccg_dcn->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	dccg_dcn->dccg_shift->field_name, dccg_dcn->dccg_mask->field_name
+
+#define CTX \
+	dccg_dcn->base.ctx
+
+#define DC_LOGGER \
+	dccg->ctx->logger
+
+static void dccg201_update_dpp_dto(struct dccg *dccg, int dpp_inst,
+				   int req_dppclk)
+{
+	/* vbios handles it */
+}
+
+static const struct dccg_funcs dccg201_funcs = {
+	.update_dpp_dto = dccg201_update_dpp_dto,
+	.get_dccg_ref_freq = dccg2_get_dccg_ref_freq,
+	.set_fifo_errdet_ovr_en = dccg2_set_fifo_errdet_ovr_en,
+	.otg_add_pixel = dccg2_otg_add_pixel,
+	.otg_drop_pixel = dccg2_otg_drop_pixel,
+	.dccg_init = dccg2_init
+};
+
+struct dccg *dccg201_create(
+	struct dc_context *ctx,
+	const struct dccg_registers *regs,
+	const struct dccg_shift *dccg_shift,
+	const struct dccg_mask *dccg_mask)
+{
+	struct dcn_dccg *dccg_dcn = kzalloc(sizeof(*dccg_dcn), GFP_KERNEL);
+	struct dccg *base;
+
+	if (dccg_dcn == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	base = &dccg_dcn->base;
+	base->ctx = ctx;
+	base->funcs = &dccg201_funcs;
+
+	dccg_dcn->regs = regs;
+	dccg_dcn->dccg_shift = dccg_shift;
+	dccg_dcn->dccg_mask = dccg_mask;
+
+	return &dccg_dcn->base;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn201/dcn201_dccg.h b/drivers/gpu/drm/amd/display/dc/dccg/dcn201/dcn201_dccg.h
new file mode 100644
index 000000000000..80888b0484fb
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn201/dcn201_dccg.h
@@ -0,0 +1,37 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCN201_DCCG_H__
+#define __DCN201_DCCG_H__
+
+#include "dcn20/dcn20_dccg.h"
+
+struct dccg *dccg201_create(
+	struct dc_context *ctx,
+	const struct dccg_registers *regs,
+	const struct dccg_shift *dccg_shift,
+	const struct dccg_mask *dccg_mask);
+
+#endif //__DCN201_DCCG_H__
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn21/dcn21_dccg.c b/drivers/gpu/drm/amd/display/dc/dccg/dcn21/dcn21_dccg.c
new file mode 100644
index 000000000000..d07c04458d31
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn21/dcn21_dccg.c
@@ -0,0 +1,132 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+#include "core_types.h"
+#include "dcn20/dcn20_dccg.h"
+#include "dcn21_dccg.h"
+
+#define TO_DCN_DCCG(dccg)\
+	container_of(dccg, struct dcn_dccg, base)
+
+#define REG(reg) \
+	(dccg_dcn->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	dccg_dcn->dccg_shift->field_name, dccg_dcn->dccg_mask->field_name
+
+#define CTX \
+	dccg_dcn->base.ctx
+#define DC_LOGGER \
+	dccg->ctx->logger
+
+static void dccg21_update_dpp_dto(struct dccg *dccg, int dpp_inst, int req_dppclk)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (dccg->ref_dppclk) {
+		int ref_dppclk = dccg->ref_dppclk;
+		int modulo = ref_dppclk / 10000;
+		int phase;
+
+		if (req_dppclk) {
+			/*
+			 * program DPP DTO phase and modulo as below
+			 * phase = ceiling(dpp_pipe_clk_mhz / 10)
+			 * module = trunc(dpp_global_clk_mhz / 10)
+			 *
+			 * storing frequencies in registers allow dmcub fw
+			 * to run time lower clocks when possible for power saving
+			 *
+			 * ceiling phase and truncate modulo guarentees the divided
+			 * down per pipe dpp clock has high enough frequency
+			 */
+			phase = (req_dppclk + 9999) / 10000;
+
+			if (phase > modulo) {
+				/* phase > modulo result in screen corruption
+				 * ie phase = 30, mod = 29 for 4k@60 HDMI
+				 * in these case we don't want pipe clock to be divided
+				 */
+				phase = modulo;
+			}
+		} else {
+			/*
+			 *  set phase to 10 if dpp isn't used to
+			 *  prevent hard hang if access dpp register
+			 *  on unused pipe
+			 *
+			 *  DTO should be on to divide down un-used
+			 *  pipe clock for power saving
+			 */
+			phase = 10;
+		}
+
+		REG_SET_2(DPPCLK_DTO_PARAM[dpp_inst], 0,
+				DPPCLK0_DTO_PHASE, phase,
+				DPPCLK0_DTO_MODULO, modulo);
+
+		REG_UPDATE(DPPCLK_DTO_CTRL,
+				DPPCLK_DTO_ENABLE[dpp_inst], 1);
+	}
+
+	dccg->pipe_dppclk_khz[dpp_inst] = req_dppclk;
+}
+
+
+static const struct dccg_funcs dccg21_funcs = {
+	.update_dpp_dto = dccg21_update_dpp_dto,
+	.get_dccg_ref_freq = dccg2_get_dccg_ref_freq,
+	.set_fifo_errdet_ovr_en = dccg2_set_fifo_errdet_ovr_en,
+	.otg_add_pixel = dccg2_otg_add_pixel,
+	.otg_drop_pixel = dccg2_otg_drop_pixel,
+	.dccg_init = dccg2_init
+};
+
+struct dccg *dccg21_create(
+	struct dc_context *ctx,
+	const struct dccg_registers *regs,
+	const struct dccg_shift *dccg_shift,
+	const struct dccg_mask *dccg_mask)
+{
+	struct dcn_dccg *dccg_dcn = kzalloc(sizeof(*dccg_dcn), GFP_KERNEL);
+	struct dccg *base;
+
+	if (dccg_dcn == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	base = &dccg_dcn->base;
+	base->ctx = ctx;
+	base->funcs = &dccg21_funcs;
+
+	dccg_dcn->regs = regs;
+	dccg_dcn->dccg_shift = dccg_shift;
+	dccg_dcn->dccg_mask = dccg_mask;
+
+	return &dccg_dcn->base;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn21/dcn21_dccg.h b/drivers/gpu/drm/amd/display/dc/dccg/dcn21/dcn21_dccg.h
new file mode 100644
index 000000000000..b7efa777ec73
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn21/dcn21_dccg.h
@@ -0,0 +1,36 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCN21_DCCG_H__
+#define __DCN21_DCCG_H__
+
+struct dccg *dccg21_create(
+	struct dc_context *ctx,
+	const struct dccg_registers *regs,
+	const struct dccg_shift *dccg_shift,
+	const struct dccg_mask *dccg_mask);
+
+
+#endif /* __DCN21_DCCG_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn30/dcn30_dccg.c b/drivers/gpu/drm/amd/display/dc/dccg/dcn30/dcn30_dccg.c
new file mode 100644
index 000000000000..d445dfefc047
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn30/dcn30_dccg.c
@@ -0,0 +1,103 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+#include "core_types.h"
+#include "dcn30_dccg.h"
+
+#define TO_DCN_DCCG(dccg)\
+	container_of(dccg, struct dcn_dccg, base)
+
+#define REG(reg) \
+	(dccg_dcn->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	dccg_dcn->dccg_shift->field_name, dccg_dcn->dccg_mask->field_name
+
+#define CTX \
+	dccg_dcn->base.ctx
+#define DC_LOGGER \
+	dccg->ctx->logger
+
+
+static const struct dccg_funcs dccg3_funcs = {
+	.update_dpp_dto = dccg2_update_dpp_dto,
+	.get_dccg_ref_freq = dccg2_get_dccg_ref_freq,
+	.set_fifo_errdet_ovr_en = dccg2_set_fifo_errdet_ovr_en,
+	.otg_add_pixel = dccg2_otg_add_pixel,
+	.otg_drop_pixel = dccg2_otg_drop_pixel,
+	.dccg_init = dccg2_init
+};
+
+struct dccg *dccg3_create(
+	struct dc_context *ctx,
+	const struct dccg_registers *regs,
+	const struct dccg_shift *dccg_shift,
+	const struct dccg_mask *dccg_mask)
+{
+	struct dcn_dccg *dccg_dcn = kzalloc(sizeof(*dccg_dcn), GFP_KERNEL);
+	struct dccg *base;
+
+	if (dccg_dcn == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	base = &dccg_dcn->base;
+	base->ctx = ctx;
+	base->funcs = &dccg3_funcs;
+
+	dccg_dcn->regs = regs;
+	dccg_dcn->dccg_shift = dccg_shift;
+	dccg_dcn->dccg_mask = dccg_mask;
+
+	return &dccg_dcn->base;
+}
+
+struct dccg *dccg30_create(
+	struct dc_context *ctx,
+	const struct dccg_registers *regs,
+	const struct dccg_shift *dccg_shift,
+	const struct dccg_mask *dccg_mask)
+{
+	struct dcn_dccg *dccg_dcn = kzalloc(sizeof(*dccg_dcn), GFP_KERNEL);
+	struct dccg *base;
+
+	if (dccg_dcn == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	base = &dccg_dcn->base;
+	base->ctx = ctx;
+	base->funcs = &dccg3_funcs;
+
+	dccg_dcn->regs = regs;
+	dccg_dcn->dccg_shift = dccg_shift;
+	dccg_dcn->dccg_mask = dccg_mask;
+
+	return &dccg_dcn->base;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn30/dcn30_dccg.h b/drivers/gpu/drm/amd/display/dc/dccg/dcn30/dcn30_dccg.h
new file mode 100644
index 000000000000..3f1da7f3a91c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn30/dcn30_dccg.h
@@ -0,0 +1,66 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCN30_DCCG_H__
+#define __DCN30_DCCG_H__
+
+#include "dcn20/dcn20_dccg.h"
+
+
+#define DCCG_REG_LIST_DCN30() \
+	DCCG_REG_LIST_DCN2(),\
+	DCCG_SRII(CLOCK_CNTL, HDMICHARCLK, 0),\
+	DCCG_SRII(PIXEL_RATE_CNTL, OTG, 2),\
+	DCCG_SRII(PIXEL_RATE_CNTL, OTG, 3),\
+	DCCG_SRII(PIXEL_RATE_CNTL, OTG, 4),\
+	DCCG_SRII(PIXEL_RATE_CNTL, OTG, 5),\
+	SR(PHYASYMCLK_CLOCK_CNTL),\
+	SR(PHYBSYMCLK_CLOCK_CNTL),\
+	SR(PHYCSYMCLK_CLOCK_CNTL)
+
+#define DCCG_MASK_SH_LIST_DCN3(mask_sh) \
+	DCCG_MASK_SH_LIST_DCN2(mask_sh),\
+	DCCG_SF(HDMICHARCLK0_CLOCK_CNTL, HDMICHARCLK0_EN, mask_sh),\
+	DCCG_SF(HDMICHARCLK0_CLOCK_CNTL, HDMICHARCLK0_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYASYMCLK_CLOCK_CNTL, PHYASYMCLK_FORCE_EN, mask_sh),\
+	DCCG_SF(PHYASYMCLK_CLOCK_CNTL, PHYASYMCLK_FORCE_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYBSYMCLK_CLOCK_CNTL, PHYBSYMCLK_FORCE_EN, mask_sh),\
+	DCCG_SF(PHYBSYMCLK_CLOCK_CNTL, PHYBSYMCLK_FORCE_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYCSYMCLK_CLOCK_CNTL, PHYCSYMCLK_FORCE_EN, mask_sh),\
+	DCCG_SF(PHYCSYMCLK_CLOCK_CNTL, PHYCSYMCLK_FORCE_SRC_SEL, mask_sh),\
+
+struct dccg *dccg3_create(
+	struct dc_context *ctx,
+	const struct dccg_registers *regs,
+	const struct dccg_shift *dccg_shift,
+	const struct dccg_mask *dccg_mask);
+
+struct dccg *dccg30_create(
+	struct dc_context *ctx,
+	const struct dccg_registers *regs,
+	const struct dccg_shift *dccg_shift,
+	const struct dccg_mask *dccg_mask);
+
+#endif //__DCN30_DCCG_H__
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn301/dcn301_dccg.c b/drivers/gpu/drm/amd/display/dc/dccg/dcn301/dcn301_dccg.c
new file mode 100644
index 000000000000..97e9be87afd9
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn301/dcn301_dccg.c
@@ -0,0 +1,78 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+#include "core_types.h"
+#include "dcn301_dccg.h"
+
+#define TO_DCN_DCCG(dccg)\
+	container_of(dccg, struct dcn_dccg, base)
+
+#define REG(reg) \
+	(dccg_dcn->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	dccg_dcn->dccg_shift->field_name, dccg_dcn->dccg_mask->field_name
+
+#define CTX \
+	dccg_dcn->base.ctx
+#define DC_LOGGER \
+	dccg->ctx->logger
+
+static const struct dccg_funcs dccg301_funcs = {
+	.update_dpp_dto = dccg2_update_dpp_dto,
+	.get_dccg_ref_freq = dccg2_get_dccg_ref_freq,
+	.set_fifo_errdet_ovr_en = dccg2_set_fifo_errdet_ovr_en,
+	.otg_add_pixel = dccg2_otg_add_pixel,
+	.otg_drop_pixel = dccg2_otg_drop_pixel,
+	.dccg_init = dccg2_init
+};
+
+struct dccg *dccg301_create(
+	struct dc_context *ctx,
+	const struct dccg_registers *regs,
+	const struct dccg_shift *dccg_shift,
+	const struct dccg_mask *dccg_mask)
+{
+	struct dcn_dccg *dccg_dcn = kzalloc(sizeof(*dccg_dcn), GFP_KERNEL);
+	struct dccg *base;
+
+	if (dccg_dcn == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	base = &dccg_dcn->base;
+	base->ctx = ctx;
+	base->funcs = &dccg301_funcs;
+
+	dccg_dcn->regs = regs;
+	dccg_dcn->dccg_shift = dccg_shift;
+	dccg_dcn->dccg_mask = dccg_mask;
+
+	return &dccg_dcn->base;
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn301/dcn301_dccg.h b/drivers/gpu/drm/amd/display/dc/dccg/dcn301/dcn301_dccg.h
new file mode 100644
index 000000000000..067e49cb238e
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn301/dcn301_dccg.h
@@ -0,0 +1,59 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCN301_DCCG_H__
+#define __DCN301_DCCG_H__
+
+#include "dcn20/dcn20_dccg.h"
+
+#define DCCG_REG_LIST_DCN301() \
+	SR(DPPCLK_DTO_CTRL),\
+	DCCG_SRII(DTO_PARAM, DPPCLK, 0),\
+	DCCG_SRII(DTO_PARAM, DPPCLK, 1),\
+	DCCG_SRII(DTO_PARAM, DPPCLK, 2),\
+	DCCG_SRII(DTO_PARAM, DPPCLK, 3),\
+	SR(REFCLK_CNTL)
+
+#define DCCG_MASK_SH_LIST_DCN301(mask_sh) \
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 0, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 0, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 1, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 1, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 2, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 2, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 3, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 3, mask_sh),\
+	DCCG_SF(DPPCLK0_DTO_PARAM, DPPCLK0_DTO_PHASE, mask_sh),\
+	DCCG_SF(DPPCLK0_DTO_PARAM, DPPCLK0_DTO_MODULO, mask_sh),\
+	DCCG_SF(REFCLK_CNTL, REFCLK_CLOCK_EN, mask_sh),\
+	DCCG_SF(REFCLK_CNTL, REFCLK_SRC_SEL, mask_sh)
+
+struct dccg *dccg301_create(
+	struct dc_context *ctx,
+	const struct dccg_registers *regs,
+	const struct dccg_shift *dccg_shift,
+	const struct dccg_mask *dccg_mask);
+
+#endif //__DCN301_DCCG_H__
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn302/dcn302_dccg.h b/drivers/gpu/drm/amd/display/dc/dccg/dcn302/dcn302_dccg.h
new file mode 100644
index 000000000000..c884dde1bb25
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn302/dcn302_dccg.h
@@ -0,0 +1,41 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCN302_DCCG_H__
+#define __DCN302_DCCG_H__
+
+#include "dcn30/dcn30_dccg.h"
+
+
+#define DCCG_REG_LIST_DCN3_02() \
+	DCCG_COMMON_REG_LIST_DCN_BASE(),\
+	DCCG_SRII(DTO_PARAM, DPPCLK, 4)
+
+#define DCCG_MASK_SH_LIST_DCN3_02(mask_sh) \
+	DCCG_COMMON_MASK_SH_LIST_DCN_COMMON_BASE(mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 4, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 4, mask_sh)
+
+#endif //__DCN302_DCCG_H__
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn303/dcn303_dccg.h b/drivers/gpu/drm/amd/display/dc/dccg/dcn303/dcn303_dccg.h
new file mode 100644
index 000000000000..2e12fb643005
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn303/dcn303_dccg.h
@@ -0,0 +1,64 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright (C) 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ */
+
+#ifndef __DCN303_DCCG_H__
+#define __DCN303_DCCG_H__
+
+#include "dcn30/dcn30_dccg.h"
+
+
+#define DCCG_REG_LIST_DCN3_03() \
+	SR(DPPCLK_DTO_CTRL),\
+	DCCG_SRII(DTO_PARAM, DPPCLK, 0),\
+	DCCG_SRII(DTO_PARAM, DPPCLK, 1),\
+	SR(REFCLK_CNTL),\
+	SR(DISPCLK_FREQ_CHANGE_CNTL),\
+	DCCG_SRII(PIXEL_RATE_CNTL, OTG, 0),\
+	DCCG_SRII(PIXEL_RATE_CNTL, OTG, 1)
+
+
+#define DCCG_MASK_SH_LIST_DCN3_03(mask_sh) \
+		DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 0, mask_sh),\
+		DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 0, mask_sh),\
+		DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 1, mask_sh),\
+		DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 1, mask_sh),\
+		DCCG_SF(DPPCLK0_DTO_PARAM, DPPCLK0_DTO_PHASE, mask_sh),\
+		DCCG_SF(DPPCLK0_DTO_PARAM, DPPCLK0_DTO_MODULO, mask_sh),\
+		DCCG_SF(REFCLK_CNTL, REFCLK_CLOCK_EN, mask_sh),\
+		DCCG_SF(REFCLK_CNTL, REFCLK_SRC_SEL, mask_sh),\
+		DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DISPCLK_STEP_DELAY, mask_sh),\
+		DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DISPCLK_STEP_SIZE, mask_sh),\
+		DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DISPCLK_FREQ_RAMP_DONE, mask_sh),\
+		DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DISPCLK_MAX_ERRDET_CYCLES, mask_sh),\
+		DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DCCG_FIFO_ERRDET_RESET, mask_sh),\
+		DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DCCG_FIFO_ERRDET_STATE, mask_sh),\
+		DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DCCG_FIFO_ERRDET_OVR_EN, mask_sh),\
+		DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DISPCLK_CHG_FWD_CORR_DISABLE, mask_sh),\
+		DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 0, mask_sh),\
+		DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 1, mask_sh),\
+		DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, DROP_PIXEL, 0, mask_sh),\
+		DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, DROP_PIXEL, 1, mask_sh)
+
+#endif //__DCN303_DCCG_H__
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.c b/drivers/gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.c
new file mode 100644
index 000000000000..8664f0c4c9b7
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.c
@@ -0,0 +1,755 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+#include "core_types.h"
+#include "dcn31_dccg.h"
+#include "dal_asic_id.h"
+
+#define TO_DCN_DCCG(dccg)\
+	container_of(dccg, struct dcn_dccg, base)
+
+#define REG(reg) \
+	(dccg_dcn->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	dccg_dcn->dccg_shift->field_name, dccg_dcn->dccg_mask->field_name
+
+#define CTX \
+	dccg_dcn->base.ctx
+#define DC_LOGGER \
+	dccg->ctx->logger
+
+void dccg31_update_dpp_dto(struct dccg *dccg, int dpp_inst, int req_dppclk)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (dccg->dpp_clock_gated[dpp_inst]) {
+		/*
+		 * Do not update the DPPCLK DTO if the clock is stopped.
+		 * It is treated the same as if the pipe itself were in PG.
+		 */
+		return;
+	}
+
+	if (dccg->ref_dppclk && req_dppclk) {
+		int ref_dppclk = dccg->ref_dppclk;
+		int modulo, phase;
+
+		// phase / modulo = dpp pipe clk / dpp global clk
+		modulo = 0xff;   // use FF at the end
+		phase = ((modulo * req_dppclk) + ref_dppclk - 1) / ref_dppclk;
+
+		if (phase > 0xff) {
+			ASSERT(false);
+			phase = 0xff;
+		}
+
+		REG_SET_2(DPPCLK_DTO_PARAM[dpp_inst], 0,
+				DPPCLK0_DTO_PHASE, phase,
+				DPPCLK0_DTO_MODULO, modulo);
+		REG_UPDATE(DPPCLK_DTO_CTRL,
+				DPPCLK_DTO_ENABLE[dpp_inst], 1);
+	} else {
+		REG_UPDATE(DPPCLK_DTO_CTRL,
+				DPPCLK_DTO_ENABLE[dpp_inst], 0);
+	}
+	dccg->pipe_dppclk_khz[dpp_inst] = req_dppclk;
+}
+
+static enum phyd32clk_clock_source get_phy_mux_symclk(
+		struct dcn_dccg *dccg_dcn,
+		enum phyd32clk_clock_source src)
+{
+	if (dccg_dcn->base.ctx->asic_id.chip_family == FAMILY_YELLOW_CARP &&
+			dccg_dcn->base.ctx->asic_id.hw_internal_rev == YELLOW_CARP_B0) {
+		if (src == PHYD32CLKC)
+			src = PHYD32CLKF;
+		if (src == PHYD32CLKD)
+			src = PHYD32CLKG;
+	}
+	return src;
+}
+
+static void dccg31_enable_dpstreamclk(struct dccg *dccg, int otg_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	/* enabled to select one of the DTBCLKs for pipe */
+	switch (otg_inst) {
+	case 0:
+		REG_UPDATE(DPSTREAMCLK_CNTL,
+				DPSTREAMCLK_PIPE0_EN, 1);
+		break;
+	case 1:
+		REG_UPDATE(DPSTREAMCLK_CNTL,
+				DPSTREAMCLK_PIPE1_EN, 1);
+		break;
+	case 2:
+		REG_UPDATE(DPSTREAMCLK_CNTL,
+				DPSTREAMCLK_PIPE2_EN, 1);
+		break;
+	case 3:
+		REG_UPDATE(DPSTREAMCLK_CNTL,
+				DPSTREAMCLK_PIPE3_EN, 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+	if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream)
+		REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+			DPSTREAMCLK_GATE_DISABLE, 1,
+			DPSTREAMCLK_ROOT_GATE_DISABLE, 1);
+}
+
+static void dccg31_disable_dpstreamclk(struct dccg *dccg, int otg_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream)
+		REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+				DPSTREAMCLK_ROOT_GATE_DISABLE, 0,
+				DPSTREAMCLK_GATE_DISABLE, 0);
+
+	switch (otg_inst) {
+	case 0:
+		REG_UPDATE(DPSTREAMCLK_CNTL,
+				DPSTREAMCLK_PIPE0_EN, 0);
+		break;
+	case 1:
+		REG_UPDATE(DPSTREAMCLK_CNTL,
+				DPSTREAMCLK_PIPE1_EN, 0);
+		break;
+	case 2:
+		REG_UPDATE(DPSTREAMCLK_CNTL,
+				DPSTREAMCLK_PIPE2_EN, 0);
+		break;
+	case 3:
+		REG_UPDATE(DPSTREAMCLK_CNTL,
+				DPSTREAMCLK_PIPE3_EN, 0);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+void dccg31_set_dpstreamclk(
+		struct dccg *dccg,
+		enum streamclk_source src,
+		int otg_inst,
+		int dp_hpo_inst)
+{
+	if (src == REFCLK)
+		dccg31_disable_dpstreamclk(dccg, otg_inst);
+	else
+		dccg31_enable_dpstreamclk(dccg, otg_inst);
+}
+
+void dccg31_enable_symclk32_se(
+		struct dccg *dccg,
+		int hpo_se_inst,
+		enum phyd32clk_clock_source phyd32clk)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	phyd32clk = get_phy_mux_symclk(dccg_dcn, phyd32clk);
+
+	/* select one of the PHYD32CLKs as the source for symclk32_se */
+	switch (hpo_se_inst) {
+	case 0:
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_SE0_GATE_DISABLE, 1,
+					SYMCLK32_ROOT_SE0_GATE_DISABLE, 1);
+		REG_UPDATE_2(SYMCLK32_SE_CNTL,
+				SYMCLK32_SE0_SRC_SEL, phyd32clk,
+				SYMCLK32_SE0_EN, 1);
+		break;
+	case 1:
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_SE1_GATE_DISABLE, 1,
+					SYMCLK32_ROOT_SE1_GATE_DISABLE, 1);
+		REG_UPDATE_2(SYMCLK32_SE_CNTL,
+				SYMCLK32_SE1_SRC_SEL, phyd32clk,
+				SYMCLK32_SE1_EN, 1);
+		break;
+	case 2:
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_SE2_GATE_DISABLE, 1,
+					SYMCLK32_ROOT_SE2_GATE_DISABLE, 1);
+		REG_UPDATE_2(SYMCLK32_SE_CNTL,
+				SYMCLK32_SE2_SRC_SEL, phyd32clk,
+				SYMCLK32_SE2_EN, 1);
+		break;
+	case 3:
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_SE3_GATE_DISABLE, 1,
+					SYMCLK32_ROOT_SE3_GATE_DISABLE, 1);
+		REG_UPDATE_2(SYMCLK32_SE_CNTL,
+				SYMCLK32_SE3_SRC_SEL, phyd32clk,
+				SYMCLK32_SE3_EN, 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+void dccg31_disable_symclk32_se(
+		struct dccg *dccg,
+		int hpo_se_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	/* set refclk as the source for symclk32_se */
+	switch (hpo_se_inst) {
+	case 0:
+		REG_UPDATE_2(SYMCLK32_SE_CNTL,
+				SYMCLK32_SE0_SRC_SEL, 0,
+				SYMCLK32_SE0_EN, 0);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_SE0_GATE_DISABLE, 0,
+					SYMCLK32_ROOT_SE0_GATE_DISABLE, 0);
+		break;
+	case 1:
+		REG_UPDATE_2(SYMCLK32_SE_CNTL,
+				SYMCLK32_SE1_SRC_SEL, 0,
+				SYMCLK32_SE1_EN, 0);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_SE1_GATE_DISABLE, 0,
+					SYMCLK32_ROOT_SE1_GATE_DISABLE, 0);
+		break;
+	case 2:
+		REG_UPDATE_2(SYMCLK32_SE_CNTL,
+				SYMCLK32_SE2_SRC_SEL, 0,
+				SYMCLK32_SE2_EN, 0);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_SE2_GATE_DISABLE, 0,
+					SYMCLK32_ROOT_SE2_GATE_DISABLE, 0);
+		break;
+	case 3:
+		REG_UPDATE_2(SYMCLK32_SE_CNTL,
+				SYMCLK32_SE3_SRC_SEL, 0,
+				SYMCLK32_SE3_EN, 0);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_SE3_GATE_DISABLE, 0,
+					SYMCLK32_ROOT_SE3_GATE_DISABLE, 0);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+void dccg31_enable_symclk32_le(
+		struct dccg *dccg,
+		int hpo_le_inst,
+		enum phyd32clk_clock_source phyd32clk)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	phyd32clk = get_phy_mux_symclk(dccg_dcn, phyd32clk);
+
+	/* select one of the PHYD32CLKs as the source for symclk32_le */
+	switch (hpo_le_inst) {
+	case 0:
+		REG_UPDATE_2(SYMCLK32_LE_CNTL,
+				SYMCLK32_LE0_SRC_SEL, phyd32clk,
+				SYMCLK32_LE0_EN, 1);
+		break;
+	case 1:
+		REG_UPDATE_2(SYMCLK32_LE_CNTL,
+				SYMCLK32_LE1_SRC_SEL, phyd32clk,
+				SYMCLK32_LE1_EN, 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+void dccg31_disable_symclk32_le(
+		struct dccg *dccg,
+		int hpo_le_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	/* set refclk as the source for symclk32_le */
+	switch (hpo_le_inst) {
+	case 0:
+		REG_UPDATE_2(SYMCLK32_LE_CNTL,
+				SYMCLK32_LE0_SRC_SEL, 0,
+				SYMCLK32_LE0_EN, 0);
+		break;
+	case 1:
+		REG_UPDATE_2(SYMCLK32_LE_CNTL,
+				SYMCLK32_LE1_SRC_SEL, 0,
+				SYMCLK32_LE1_EN, 0);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+void dccg31_set_symclk32_le_root_clock_gating(
+		struct dccg *dccg,
+		int hpo_le_inst,
+		bool enable)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (!dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_le)
+		return;
+
+	switch (hpo_le_inst) {
+	case 0:
+		REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+				SYMCLK32_LE0_GATE_DISABLE, enable ? 1 : 0,
+				SYMCLK32_ROOT_LE0_GATE_DISABLE, enable ? 1 : 0);
+		break;
+	case 1:
+		REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+				SYMCLK32_LE1_GATE_DISABLE, enable ? 1 : 0,
+				SYMCLK32_ROOT_LE1_GATE_DISABLE, enable ? 1 : 0);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+void dccg31_disable_dscclk(struct dccg *dccg, int inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (!dccg->ctx->dc->debug.root_clock_optimization.bits.dsc)
+		return;
+	//DTO must be enabled to generate a 0 Hz clock output
+	switch (inst) {
+	case 0:
+		REG_UPDATE(DSCCLK_DTO_CTRL,
+				DSCCLK0_DTO_ENABLE, 1);
+		REG_UPDATE_2(DSCCLK0_DTO_PARAM,
+				DSCCLK0_DTO_PHASE, 0,
+				DSCCLK0_DTO_MODULO, 1);
+		break;
+	case 1:
+		REG_UPDATE(DSCCLK_DTO_CTRL,
+				DSCCLK1_DTO_ENABLE, 1);
+		REG_UPDATE_2(DSCCLK1_DTO_PARAM,
+				DSCCLK1_DTO_PHASE, 0,
+				DSCCLK1_DTO_MODULO, 1);
+		break;
+	case 2:
+		REG_UPDATE(DSCCLK_DTO_CTRL,
+				DSCCLK2_DTO_ENABLE, 1);
+		REG_UPDATE_2(DSCCLK2_DTO_PARAM,
+				DSCCLK2_DTO_PHASE, 0,
+				DSCCLK2_DTO_MODULO, 1);
+		break;
+	case 3:
+		if (REG(DSCCLK3_DTO_PARAM)) {
+			REG_UPDATE(DSCCLK_DTO_CTRL,
+					DSCCLK3_DTO_ENABLE, 1);
+			REG_UPDATE_2(DSCCLK3_DTO_PARAM,
+					DSCCLK3_DTO_PHASE, 0,
+					DSCCLK3_DTO_MODULO, 1);
+		}
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+void dccg31_enable_dscclk(struct dccg *dccg, int inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (!dccg->ctx->dc->debug.root_clock_optimization.bits.dsc)
+		return;
+	//Disable DTO
+	switch (inst) {
+	case 0:
+		REG_UPDATE_2(DSCCLK0_DTO_PARAM,
+				DSCCLK0_DTO_PHASE, 0,
+				DSCCLK0_DTO_MODULO, 0);
+		REG_UPDATE(DSCCLK_DTO_CTRL,
+				DSCCLK0_DTO_ENABLE, 0);
+		break;
+	case 1:
+		REG_UPDATE_2(DSCCLK1_DTO_PARAM,
+				DSCCLK1_DTO_PHASE, 0,
+				DSCCLK1_DTO_MODULO, 0);
+		REG_UPDATE(DSCCLK_DTO_CTRL,
+				DSCCLK1_DTO_ENABLE, 0);
+		break;
+	case 2:
+		REG_UPDATE_2(DSCCLK2_DTO_PARAM,
+				DSCCLK2_DTO_PHASE, 0,
+				DSCCLK2_DTO_MODULO, 0);
+		REG_UPDATE(DSCCLK_DTO_CTRL,
+				DSCCLK2_DTO_ENABLE, 0);
+		break;
+	case 3:
+		if (REG(DSCCLK3_DTO_PARAM)) {
+			REG_UPDATE(DSCCLK_DTO_CTRL,
+					DSCCLK3_DTO_ENABLE, 0);
+			REG_UPDATE_2(DSCCLK3_DTO_PARAM,
+					DSCCLK3_DTO_PHASE, 0,
+					DSCCLK3_DTO_MODULO, 0);
+		}
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+void dccg31_set_physymclk(
+		struct dccg *dccg,
+		int phy_inst,
+		enum physymclk_clock_source clk_src,
+		bool force_enable)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	/* Force PHYSYMCLK on and Select phyd32clk as the source of clock which is output to PHY through DCIO */
+	switch (phy_inst) {
+	case 0:
+		if (force_enable) {
+			REG_UPDATE_2(PHYASYMCLK_CLOCK_CNTL,
+					PHYASYMCLK_FORCE_EN, 1,
+					PHYASYMCLK_FORCE_SRC_SEL, clk_src);
+			if (dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk)
+				REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+					PHYASYMCLK_GATE_DISABLE, 1);
+		} else {
+			REG_UPDATE_2(PHYASYMCLK_CLOCK_CNTL,
+					PHYASYMCLK_FORCE_EN, 0,
+					PHYASYMCLK_FORCE_SRC_SEL, 0);
+			if (dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk)
+				REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+					PHYASYMCLK_GATE_DISABLE, 0);
+		}
+		break;
+	case 1:
+		if (force_enable) {
+			REG_UPDATE_2(PHYBSYMCLK_CLOCK_CNTL,
+					PHYBSYMCLK_FORCE_EN, 1,
+					PHYBSYMCLK_FORCE_SRC_SEL, clk_src);
+			if (dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk)
+				REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+					PHYBSYMCLK_GATE_DISABLE, 1);
+		} else {
+			REG_UPDATE_2(PHYBSYMCLK_CLOCK_CNTL,
+					PHYBSYMCLK_FORCE_EN, 0,
+					PHYBSYMCLK_FORCE_SRC_SEL, 0);
+			if (dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk)
+				REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+					PHYBSYMCLK_GATE_DISABLE, 0);
+		}
+		break;
+	case 2:
+		if (force_enable) {
+			REG_UPDATE_2(PHYCSYMCLK_CLOCK_CNTL,
+					PHYCSYMCLK_FORCE_EN, 1,
+					PHYCSYMCLK_FORCE_SRC_SEL, clk_src);
+			if (dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk)
+				REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+					PHYCSYMCLK_GATE_DISABLE, 1);
+		} else {
+			REG_UPDATE_2(PHYCSYMCLK_CLOCK_CNTL,
+					PHYCSYMCLK_FORCE_EN, 0,
+					PHYCSYMCLK_FORCE_SRC_SEL, 0);
+			if (dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk)
+				REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+					PHYCSYMCLK_GATE_DISABLE, 0);
+		}
+		break;
+	case 3:
+		if (force_enable) {
+			REG_UPDATE_2(PHYDSYMCLK_CLOCK_CNTL,
+					PHYDSYMCLK_FORCE_EN, 1,
+					PHYDSYMCLK_FORCE_SRC_SEL, clk_src);
+			if (dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk)
+				REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+					PHYDSYMCLK_GATE_DISABLE, 1);
+		} else {
+			REG_UPDATE_2(PHYDSYMCLK_CLOCK_CNTL,
+					PHYDSYMCLK_FORCE_EN, 0,
+					PHYDSYMCLK_FORCE_SRC_SEL, 0);
+			if (dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk)
+				REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+					PHYDSYMCLK_GATE_DISABLE, 0);
+		}
+		break;
+	case 4:
+		if (force_enable) {
+			REG_UPDATE_2(PHYESYMCLK_CLOCK_CNTL,
+					PHYESYMCLK_FORCE_EN, 1,
+					PHYESYMCLK_FORCE_SRC_SEL, clk_src);
+			if (dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk)
+				REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+					PHYESYMCLK_GATE_DISABLE, 1);
+		} else {
+			REG_UPDATE_2(PHYESYMCLK_CLOCK_CNTL,
+					PHYESYMCLK_FORCE_EN, 0,
+					PHYESYMCLK_FORCE_SRC_SEL, 0);
+			if (dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk)
+				REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+					PHYESYMCLK_GATE_DISABLE, 0);
+		}
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+/* Controls the generation of pixel valid for OTG in (OTG -> HPO case) */
+void dccg31_set_dtbclk_dto(
+		struct dccg *dccg,
+		const struct dtbclk_dto_params *params)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+	int req_dtbclk_khz = params->pixclk_khz;
+	uint32_t dtbdto_div;
+
+	/* Mode	                DTBDTO Rate       DTBCLK_DTO<x>_DIV Register
+	 * ODM 4:1 combine      pixel rate/4      2
+	 * ODM 2:1 combine      pixel rate/2      4
+	 * non-DSC 4:2:0 mode   pixel rate/2      4
+	 * DSC native 4:2:0     pixel rate/2      4
+	 * DSC native 4:2:2     pixel rate/2      4
+	 * Other modes          pixel rate        8
+	 */
+	if (params->num_odm_segments == 4) {
+		dtbdto_div = 2;
+		req_dtbclk_khz = params->pixclk_khz / 4;
+	} else if ((params->num_odm_segments == 2) ||
+			(params->timing->pixel_encoding == PIXEL_ENCODING_YCBCR420) ||
+			(params->timing->flags.DSC && params->timing->pixel_encoding == PIXEL_ENCODING_YCBCR422
+					&& !params->timing->dsc_cfg.ycbcr422_simple)) {
+		dtbdto_div = 4;
+		req_dtbclk_khz = params->pixclk_khz / 2;
+	} else
+		dtbdto_div = 8;
+
+	if (params->ref_dtbclk_khz && req_dtbclk_khz) {
+		uint32_t modulo, phase;
+
+		// phase / modulo = dtbclk / dtbclk ref
+		modulo = params->ref_dtbclk_khz * 1000;
+		phase = div_u64((((unsigned long long)modulo * req_dtbclk_khz) + params->ref_dtbclk_khz - 1),
+				params->ref_dtbclk_khz);
+
+		REG_UPDATE(OTG_PIXEL_RATE_CNTL[params->otg_inst],
+				DTBCLK_DTO_DIV[params->otg_inst], dtbdto_div);
+
+		REG_WRITE(DTBCLK_DTO_MODULO[params->otg_inst], modulo);
+		REG_WRITE(DTBCLK_DTO_PHASE[params->otg_inst], phase);
+
+		REG_UPDATE(OTG_PIXEL_RATE_CNTL[params->otg_inst],
+				DTBCLK_DTO_ENABLE[params->otg_inst], 1);
+
+		REG_WAIT(OTG_PIXEL_RATE_CNTL[params->otg_inst],
+				DTBCLKDTO_ENABLE_STATUS[params->otg_inst], 1,
+				1, 100);
+
+		/* The recommended programming sequence to enable DTBCLK DTO to generate
+		 * valid pixel HPO DPSTREAM ENCODER, specifies that DTO source select should
+		 * be set only after DTO is enabled
+		 */
+		REG_UPDATE(OTG_PIXEL_RATE_CNTL[params->otg_inst],
+				PIPE_DTO_SRC_SEL[params->otg_inst], 1);
+	} else {
+		REG_UPDATE_3(OTG_PIXEL_RATE_CNTL[params->otg_inst],
+				DTBCLK_DTO_ENABLE[params->otg_inst], 0,
+				PIPE_DTO_SRC_SEL[params->otg_inst], 0,
+				DTBCLK_DTO_DIV[params->otg_inst], dtbdto_div);
+
+		REG_WRITE(DTBCLK_DTO_MODULO[params->otg_inst], 0);
+		REG_WRITE(DTBCLK_DTO_PHASE[params->otg_inst], 0);
+	}
+}
+
+void dccg31_set_audio_dtbclk_dto(
+		struct dccg *dccg,
+		const struct dtbclk_dto_params *params)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (params->ref_dtbclk_khz && params->req_audio_dtbclk_khz) {
+		uint32_t modulo, phase;
+
+		// phase / modulo = dtbclk / dtbclk ref
+		modulo = params->ref_dtbclk_khz * 1000;
+		phase = div_u64((((unsigned long long)modulo * params->req_audio_dtbclk_khz) + params->ref_dtbclk_khz - 1),
+			params->ref_dtbclk_khz);
+
+
+		REG_WRITE(DCCG_AUDIO_DTBCLK_DTO_MODULO, modulo);
+		REG_WRITE(DCCG_AUDIO_DTBCLK_DTO_PHASE, phase);
+
+		//REG_UPDATE(DCCG_AUDIO_DTO_SOURCE,
+		//		DCCG_AUDIO_DTBCLK_DTO_USE_512FBR_DTO, 1);
+
+		REG_UPDATE(DCCG_AUDIO_DTO_SOURCE,
+				DCCG_AUDIO_DTO_SEL, 4);  //  04 - DCCG_AUDIO_DTO_SEL_AUDIO_DTO_DTBCLK
+	} else {
+		REG_WRITE(DCCG_AUDIO_DTBCLK_DTO_PHASE, 0);
+		REG_WRITE(DCCG_AUDIO_DTBCLK_DTO_MODULO, 0);
+
+		REG_UPDATE(DCCG_AUDIO_DTO_SOURCE,
+				DCCG_AUDIO_DTO_SEL, 3);  //  03 - DCCG_AUDIO_DTO_SEL_NO_AUDIO_DTO
+	}
+}
+
+void dccg31_get_dccg_ref_freq(struct dccg *dccg,
+		unsigned int xtalin_freq_inKhz,
+		unsigned int *dccg_ref_freq_inKhz)
+{
+	/*
+	 * Assume refclk is sourced from xtalin
+	 * expect 24MHz
+	 */
+	*dccg_ref_freq_inKhz = xtalin_freq_inKhz;
+	return;
+}
+
+void dccg31_set_dispclk_change_mode(
+	struct dccg *dccg,
+	enum dentist_dispclk_change_mode change_mode)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	REG_UPDATE(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_CHG_MODE,
+		   change_mode == DISPCLK_CHANGE_MODE_RAMPING ? 2 : 0);
+}
+
+void dccg31_init(struct dccg *dccg)
+{
+	/* Set HPO stream encoder to use refclk to avoid case where PHY is
+	 * disabled and SYMCLK32 for HPO SE is sourced from PHYD32CLK which
+	 * will cause DCN to hang.
+	 */
+	dccg31_disable_symclk32_se(dccg, 0);
+	dccg31_disable_symclk32_se(dccg, 1);
+	dccg31_disable_symclk32_se(dccg, 2);
+	dccg31_disable_symclk32_se(dccg, 3);
+
+	dccg31_set_symclk32_le_root_clock_gating(dccg, 0, false);
+	dccg31_set_symclk32_le_root_clock_gating(dccg, 1, false);
+
+	if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream) {
+		dccg31_disable_dpstreamclk(dccg, 0);
+		dccg31_disable_dpstreamclk(dccg, 1);
+		dccg31_disable_dpstreamclk(dccg, 2);
+		dccg31_disable_dpstreamclk(dccg, 3);
+	}
+
+	if (dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk) {
+		dccg31_set_physymclk(dccg, 0, PHYSYMCLK_FORCE_SRC_SYMCLK, false);
+		dccg31_set_physymclk(dccg, 1, PHYSYMCLK_FORCE_SRC_SYMCLK, false);
+		dccg31_set_physymclk(dccg, 2, PHYSYMCLK_FORCE_SRC_SYMCLK, false);
+		dccg31_set_physymclk(dccg, 3, PHYSYMCLK_FORCE_SRC_SYMCLK, false);
+		dccg31_set_physymclk(dccg, 4, PHYSYMCLK_FORCE_SRC_SYMCLK, false);
+	}
+}
+
+void dccg31_otg_add_pixel(struct dccg *dccg,
+				 uint32_t otg_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	REG_UPDATE(OTG_PIXEL_RATE_CNTL[otg_inst],
+			OTG_ADD_PIXEL[otg_inst], 1);
+}
+
+void dccg31_otg_drop_pixel(struct dccg *dccg,
+				  uint32_t otg_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	REG_UPDATE(OTG_PIXEL_RATE_CNTL[otg_inst],
+			OTG_DROP_PIXEL[otg_inst], 1);
+}
+
+static const struct dccg_funcs dccg31_funcs = {
+	.update_dpp_dto = dccg31_update_dpp_dto,
+	.get_dccg_ref_freq = dccg31_get_dccg_ref_freq,
+	.dccg_init = dccg31_init,
+	.set_dpstreamclk = dccg31_set_dpstreamclk,
+	.enable_symclk32_se = dccg31_enable_symclk32_se,
+	.disable_symclk32_se = dccg31_disable_symclk32_se,
+	.enable_symclk32_le = dccg31_enable_symclk32_le,
+	.disable_symclk32_le = dccg31_disable_symclk32_le,
+	.set_physymclk = dccg31_set_physymclk,
+	.set_dtbclk_dto = dccg31_set_dtbclk_dto,
+	.set_audio_dtbclk_dto = dccg31_set_audio_dtbclk_dto,
+	.set_fifo_errdet_ovr_en = dccg2_set_fifo_errdet_ovr_en,
+	.otg_add_pixel = dccg31_otg_add_pixel,
+	.otg_drop_pixel = dccg31_otg_drop_pixel,
+	.set_dispclk_change_mode = dccg31_set_dispclk_change_mode,
+	.disable_dsc = dccg31_disable_dscclk,
+	.enable_dsc = dccg31_enable_dscclk,
+};
+
+struct dccg *dccg31_create(
+	struct dc_context *ctx,
+	const struct dccg_registers *regs,
+	const struct dccg_shift *dccg_shift,
+	const struct dccg_mask *dccg_mask)
+{
+	struct dcn_dccg *dccg_dcn = kzalloc(sizeof(*dccg_dcn), GFP_KERNEL);
+	struct dccg *base;
+
+	if (dccg_dcn == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	base = &dccg_dcn->base;
+	base->ctx = ctx;
+	base->funcs = &dccg31_funcs;
+
+	dccg_dcn->regs = regs;
+	dccg_dcn->dccg_shift = dccg_shift;
+	dccg_dcn->dccg_mask = dccg_mask;
+
+	return &dccg_dcn->base;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.h b/drivers/gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.h
new file mode 100644
index 000000000000..cd261051dc2c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.h
@@ -0,0 +1,239 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCN31_DCCG_H__
+#define __DCN31_DCCG_H__
+
+#include "dcn30/dcn30_dccg.h"
+
+#define DCCG_REG_LIST_DCN31() \
+	SR(DPPCLK_DTO_CTRL),\
+	DCCG_SRII(DTO_PARAM, DPPCLK, 0),\
+	DCCG_SRII(DTO_PARAM, DPPCLK, 1),\
+	DCCG_SRII(DTO_PARAM, DPPCLK, 2),\
+	DCCG_SRII(DTO_PARAM, DPPCLK, 3),\
+	DCCG_SRII(CLOCK_CNTL, HDMICHARCLK, 0),\
+	SR(PHYASYMCLK_CLOCK_CNTL),\
+	SR(PHYBSYMCLK_CLOCK_CNTL),\
+	SR(PHYCSYMCLK_CLOCK_CNTL),\
+	SR(PHYDSYMCLK_CLOCK_CNTL),\
+	SR(PHYESYMCLK_CLOCK_CNTL),\
+	SR(DPSTREAMCLK_CNTL),\
+	SR(HDMISTREAMCLK_CNTL),\
+	SR(SYMCLK32_SE_CNTL),\
+	SR(SYMCLK32_LE_CNTL),\
+	DCCG_SRII(PIXEL_RATE_CNTL, OTG, 0),\
+	DCCG_SRII(PIXEL_RATE_CNTL, OTG, 1),\
+	DCCG_SRII(PIXEL_RATE_CNTL, OTG, 2),\
+	DCCG_SRII(PIXEL_RATE_CNTL, OTG, 3),\
+	DCCG_SRII(MODULO, DTBCLK_DTO, 0),\
+	DCCG_SRII(MODULO, DTBCLK_DTO, 1),\
+	DCCG_SRII(MODULO, DTBCLK_DTO, 2),\
+	DCCG_SRII(MODULO, DTBCLK_DTO, 3),\
+	DCCG_SRII(PHASE, DTBCLK_DTO, 0),\
+	DCCG_SRII(PHASE, DTBCLK_DTO, 1),\
+	DCCG_SRII(PHASE, DTBCLK_DTO, 2),\
+	DCCG_SRII(PHASE, DTBCLK_DTO, 3),\
+	SR(DCCG_AUDIO_DTBCLK_DTO_MODULO),\
+	SR(DCCG_AUDIO_DTBCLK_DTO_PHASE),\
+	SR(DCCG_AUDIO_DTO_SOURCE),\
+	SR(DENTIST_DISPCLK_CNTL),\
+	SR(DSCCLK0_DTO_PARAM),\
+	SR(DSCCLK1_DTO_PARAM),\
+	SR(DSCCLK2_DTO_PARAM),\
+	SR(DSCCLK_DTO_CTRL),\
+	SR(DCCG_GATE_DISABLE_CNTL2),\
+	SR(DCCG_GATE_DISABLE_CNTL3),\
+	SR(HDMISTREAMCLK0_DTO_PARAM)
+
+
+#define DCCG_MASK_SH_LIST_DCN31(mask_sh) \
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 0, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 0, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 1, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 1, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 2, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 2, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 3, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 3, mask_sh),\
+	DCCG_SF(DPPCLK0_DTO_PARAM, DPPCLK0_DTO_PHASE, mask_sh),\
+	DCCG_SF(DPPCLK0_DTO_PARAM, DPPCLK0_DTO_MODULO, mask_sh),\
+	DCCG_SF(HDMICHARCLK0_CLOCK_CNTL, HDMICHARCLK0_EN, mask_sh),\
+	DCCG_SF(HDMICHARCLK0_CLOCK_CNTL, HDMICHARCLK0_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYASYMCLK_CLOCK_CNTL, PHYASYMCLK_FORCE_EN, mask_sh),\
+	DCCG_SF(PHYASYMCLK_CLOCK_CNTL, PHYASYMCLK_FORCE_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYBSYMCLK_CLOCK_CNTL, PHYBSYMCLK_FORCE_EN, mask_sh),\
+	DCCG_SF(PHYBSYMCLK_CLOCK_CNTL, PHYBSYMCLK_FORCE_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYCSYMCLK_CLOCK_CNTL, PHYCSYMCLK_FORCE_EN, mask_sh),\
+	DCCG_SF(PHYCSYMCLK_CLOCK_CNTL, PHYCSYMCLK_FORCE_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYDSYMCLK_CLOCK_CNTL, PHYDSYMCLK_FORCE_EN, mask_sh),\
+	DCCG_SF(PHYDSYMCLK_CLOCK_CNTL, PHYDSYMCLK_FORCE_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYESYMCLK_CLOCK_CNTL, PHYESYMCLK_FORCE_EN, mask_sh),\
+	DCCG_SF(PHYESYMCLK_CLOCK_CNTL, PHYESYMCLK_FORCE_SRC_SEL, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK_PIPE0_EN, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK_PIPE1_EN, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK_PIPE2_EN, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK_PIPE3_EN, mask_sh),\
+	DCCG_SF(HDMISTREAMCLK_CNTL, HDMISTREAMCLK0_SRC_SEL, mask_sh),\
+	DCCG_SF(HDMISTREAMCLK_CNTL, HDMISTREAMCLK0_DTO_FORCE_DIS, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE0_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE1_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE2_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE3_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE0_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE1_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE2_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE3_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_LE_CNTL, SYMCLK32_LE0_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_LE_CNTL, SYMCLK32_LE1_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_LE_CNTL, SYMCLK32_LE0_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_LE_CNTL, SYMCLK32_LE1_EN, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLK_DTO, ENABLE, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLK_DTO, ENABLE, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLK_DTO, ENABLE, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLK_DTO, ENABLE, 3, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLKDTO, ENABLE_STATUS, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLKDTO, ENABLE_STATUS, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLKDTO, ENABLE_STATUS, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLKDTO, ENABLE_STATUS, 3, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 3, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLK_DTO, DIV, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLK_DTO, DIV, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLK_DTO, DIV, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLK_DTO, DIV, 3, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 3, mask_sh),\
+	DCCG_SF(DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO_SEL, mask_sh),\
+	DCCG_SF(DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO0_SOURCE_SEL, mask_sh),\
+	DCCG_SF(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_CHG_MODE, mask_sh),\
+	DCCG_SF(DSCCLK0_DTO_PARAM, DSCCLK0_DTO_PHASE, mask_sh),\
+	DCCG_SF(DSCCLK0_DTO_PARAM, DSCCLK0_DTO_MODULO, mask_sh),\
+	DCCG_SF(DSCCLK1_DTO_PARAM, DSCCLK1_DTO_PHASE, mask_sh),\
+	DCCG_SF(DSCCLK1_DTO_PARAM, DSCCLK1_DTO_MODULO, mask_sh),\
+	DCCG_SF(DSCCLK2_DTO_PARAM, DSCCLK2_DTO_PHASE, mask_sh),\
+	DCCG_SF(DSCCLK2_DTO_PARAM, DSCCLK2_DTO_MODULO, mask_sh),\
+	DCCG_SF(DSCCLK_DTO_CTRL, DSCCLK0_DTO_ENABLE, mask_sh),\
+	DCCG_SF(DSCCLK_DTO_CTRL, DSCCLK1_DTO_ENABLE, mask_sh),\
+	DCCG_SF(DSCCLK_DTO_CTRL, DSCCLK2_DTO_ENABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, PHYASYMCLK_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, PHYBSYMCLK_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, PHYCSYMCLK_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, PHYDSYMCLK_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, PHYESYMCLK_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, DPSTREAMCLK_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, DPSTREAMCLK_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_SE0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_SE1_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_SE2_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_SE3_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_LE0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_LE1_GATE_DISABLE, mask_sh),\
+	DCCG_SF(HDMISTREAMCLK0_DTO_PARAM, HDMISTREAMCLK0_DTO_PHASE, mask_sh),\
+	DCCG_SF(HDMISTREAMCLK0_DTO_PARAM, HDMISTREAMCLK0_DTO_MODULO, mask_sh)
+
+
+struct dccg *dccg31_create(
+	struct dc_context *ctx,
+	const struct dccg_registers *regs,
+	const struct dccg_shift *dccg_shift,
+	const struct dccg_mask *dccg_mask);
+
+void dccg31_init(struct dccg *dccg);
+
+void dccg31_enable_symclk32_se(
+		struct dccg *dccg,
+		int hpo_se_inst,
+		enum phyd32clk_clock_source phyd32clk);
+
+void dccg31_disable_symclk32_se(
+		struct dccg *dccg,
+		int hpo_se_inst);
+
+void dccg31_enable_symclk32_le(
+		struct dccg *dccg,
+		int hpo_le_inst,
+		enum phyd32clk_clock_source phyd32clk);
+
+void dccg31_disable_symclk32_le(
+		struct dccg *dccg,
+		int hpo_le_inst);
+
+void dccg31_set_symclk32_le_root_clock_gating(
+		struct dccg *dccg,
+		int hpo_le_inst,
+		bool enable);
+
+void dccg31_set_physymclk(
+		struct dccg *dccg,
+		int phy_inst,
+		enum physymclk_clock_source clk_src,
+		bool force_enable);
+
+void dccg31_set_audio_dtbclk_dto(
+		struct dccg *dccg,
+		const struct dtbclk_dto_params *params);
+
+void dccg31_update_dpp_dto(
+	struct dccg *dccg,
+	int dpp_inst,
+	int req_dppclk);
+
+void dccg31_get_dccg_ref_freq(
+	struct dccg *dccg,
+	unsigned int xtalin_freq_inKhz,
+	unsigned int *dccg_ref_freq_inKhz);
+
+void dccg31_set_dpstreamclk(
+	struct dccg *dccg,
+	enum streamclk_source src,
+	int otg_inst,
+	int dp_hpo_inst);
+
+void dccg31_set_dtbclk_dto(
+		struct dccg *dccg,
+		const struct dtbclk_dto_params *params);
+
+void dccg31_otg_add_pixel(
+	struct dccg *dccg,
+	uint32_t otg_inst);
+
+void dccg31_otg_drop_pixel(
+	struct dccg *dccg,
+	uint32_t otg_inst);
+
+void dccg31_set_dispclk_change_mode(
+	struct dccg *dccg,
+	enum dentist_dispclk_change_mode change_mode);
+
+void dccg31_disable_dscclk(struct dccg *dccg, int inst);
+
+void dccg31_enable_dscclk(struct dccg *dccg, int inst);
+
+#endif //__DCN31_DCCG_H__
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn314/dcn314_dccg.c b/drivers/gpu/drm/amd/display/dc/dccg/dcn314/dcn314_dccg.c
new file mode 100644
index 000000000000..8f6edd8e9beb
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn314/dcn314_dccg.c
@@ -0,0 +1,406 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+#include "core_types.h"
+
+#include "dcn31/dcn31_dccg.h"
+#include "dcn314_dccg.h"
+
+#define TO_DCN_DCCG(dccg)\
+	container_of(dccg, struct dcn_dccg, base)
+
+#define REG(reg) \
+	(dccg_dcn->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	dccg_dcn->dccg_shift->field_name, dccg_dcn->dccg_mask->field_name
+
+#define CTX \
+	dccg_dcn->base.ctx
+#define DC_LOGGER \
+	dccg->ctx->logger
+
+static void dccg314_trigger_dio_fifo_resync(
+	struct dccg *dccg)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+	uint32_t dispclk_rdivider_value = 0;
+
+	REG_GET(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_RDIVIDER, &dispclk_rdivider_value);
+	REG_UPDATE(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_WDIVIDER, dispclk_rdivider_value);
+}
+
+static void dccg314_get_pixel_rate_div(
+		struct dccg *dccg,
+		uint32_t otg_inst,
+		uint32_t *k1,
+		uint32_t *k2)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+	uint32_t val_k1 = PIXEL_RATE_DIV_NA, val_k2 = PIXEL_RATE_DIV_NA;
+
+	*k1 = PIXEL_RATE_DIV_NA;
+	*k2 = PIXEL_RATE_DIV_NA;
+
+	switch (otg_inst) {
+	case 0:
+		REG_GET_2(OTG_PIXEL_RATE_DIV,
+			OTG0_PIXEL_RATE_DIVK1, &val_k1,
+			OTG0_PIXEL_RATE_DIVK2, &val_k2);
+		break;
+	case 1:
+		REG_GET_2(OTG_PIXEL_RATE_DIV,
+			OTG1_PIXEL_RATE_DIVK1, &val_k1,
+			OTG1_PIXEL_RATE_DIVK2, &val_k2);
+		break;
+	case 2:
+		REG_GET_2(OTG_PIXEL_RATE_DIV,
+			OTG2_PIXEL_RATE_DIVK1, &val_k1,
+			OTG2_PIXEL_RATE_DIVK2, &val_k2);
+		break;
+	case 3:
+		REG_GET_2(OTG_PIXEL_RATE_DIV,
+			OTG3_PIXEL_RATE_DIVK1, &val_k1,
+			OTG3_PIXEL_RATE_DIVK2, &val_k2);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+	*k1 = val_k1;
+	*k2 = val_k2;
+}
+
+static void dccg314_set_pixel_rate_div(
+		struct dccg *dccg,
+		uint32_t otg_inst,
+		enum pixel_rate_div k1,
+		enum pixel_rate_div k2)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+	uint32_t cur_k1 = PIXEL_RATE_DIV_NA;
+	uint32_t cur_k2 = PIXEL_RATE_DIV_NA;
+
+	// Don't program 0xF into the register field. Not valid since
+	// K1 / K2 field is only 1 / 2 bits wide
+	if (k1 == PIXEL_RATE_DIV_NA || k2 == PIXEL_RATE_DIV_NA) {
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+	dccg314_get_pixel_rate_div(dccg, otg_inst, &cur_k1, &cur_k2);
+	if (k1 == cur_k1 && k2 == cur_k2)
+		return;
+
+	switch (otg_inst) {
+	case 0:
+		REG_UPDATE_2(OTG_PIXEL_RATE_DIV,
+				OTG0_PIXEL_RATE_DIVK1, k1,
+				OTG0_PIXEL_RATE_DIVK2, k2);
+		break;
+	case 1:
+		REG_UPDATE_2(OTG_PIXEL_RATE_DIV,
+				OTG1_PIXEL_RATE_DIVK1, k1,
+				OTG1_PIXEL_RATE_DIVK2, k2);
+		break;
+	case 2:
+		REG_UPDATE_2(OTG_PIXEL_RATE_DIV,
+				OTG2_PIXEL_RATE_DIVK1, k1,
+				OTG2_PIXEL_RATE_DIVK2, k2);
+		break;
+	case 3:
+		REG_UPDATE_2(OTG_PIXEL_RATE_DIV,
+				OTG3_PIXEL_RATE_DIVK1, k1,
+				OTG3_PIXEL_RATE_DIVK2, k2);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+static void dccg314_set_dtbclk_p_src(
+		struct dccg *dccg,
+		enum streamclk_source src,
+		uint32_t otg_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	uint32_t p_src_sel = 0; /* selects dprefclk */
+
+	if (src == DTBCLK0)
+		p_src_sel = 2;  /* selects dtbclk0 */
+
+	switch (otg_inst) {
+	case 0:
+		if (src == REFCLK)
+			REG_UPDATE(DTBCLK_P_CNTL,
+					DTBCLK_P0_EN, 0);
+		else
+			REG_UPDATE_2(DTBCLK_P_CNTL,
+					DTBCLK_P0_SRC_SEL, p_src_sel,
+					DTBCLK_P0_EN, 1);
+		break;
+	case 1:
+		if (src == REFCLK)
+			REG_UPDATE(DTBCLK_P_CNTL,
+					DTBCLK_P1_EN, 0);
+		else
+			REG_UPDATE_2(DTBCLK_P_CNTL,
+					DTBCLK_P1_SRC_SEL, p_src_sel,
+					DTBCLK_P1_EN, 1);
+		break;
+	case 2:
+		if (src == REFCLK)
+			REG_UPDATE(DTBCLK_P_CNTL,
+					DTBCLK_P2_EN, 0);
+		else
+			REG_UPDATE_2(DTBCLK_P_CNTL,
+					DTBCLK_P2_SRC_SEL, p_src_sel,
+					DTBCLK_P2_EN, 1);
+		break;
+	case 3:
+		if (src == REFCLK)
+			REG_UPDATE(DTBCLK_P_CNTL,
+					DTBCLK_P3_EN, 0);
+		else
+			REG_UPDATE_2(DTBCLK_P_CNTL,
+					DTBCLK_P3_SRC_SEL, p_src_sel,
+					DTBCLK_P3_EN, 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+}
+
+/* Controls the generation of pixel valid for OTG in (OTG -> HPO case) */
+static void dccg314_set_dtbclk_dto(
+		struct dccg *dccg,
+		const struct dtbclk_dto_params *params)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+	/* DTO Output Rate / Pixel Rate = 1/4 */
+	int req_dtbclk_khz = params->pixclk_khz / 4;
+
+	if (params->ref_dtbclk_khz && req_dtbclk_khz) {
+		uint32_t modulo, phase;
+
+		// phase / modulo = dtbclk / dtbclk ref
+		modulo = params->ref_dtbclk_khz * 1000;
+		phase = req_dtbclk_khz * 1000;
+
+		REG_WRITE(DTBCLK_DTO_MODULO[params->otg_inst], modulo);
+		REG_WRITE(DTBCLK_DTO_PHASE[params->otg_inst], phase);
+
+		REG_UPDATE(OTG_PIXEL_RATE_CNTL[params->otg_inst],
+				DTBCLK_DTO_ENABLE[params->otg_inst], 1);
+
+		REG_WAIT(OTG_PIXEL_RATE_CNTL[params->otg_inst],
+				DTBCLKDTO_ENABLE_STATUS[params->otg_inst], 1,
+				1, 100);
+
+		/* program OTG_PIXEL_RATE_DIV for DIVK1 and DIVK2 fields */
+		dccg314_set_pixel_rate_div(dccg, params->otg_inst, PIXEL_RATE_DIV_BY_1, PIXEL_RATE_DIV_BY_1);
+
+		/* The recommended programming sequence to enable DTBCLK DTO to generate
+		 * valid pixel HPO DPSTREAM ENCODER, specifies that DTO source select should
+		 * be set only after DTO is enabled
+		 */
+		REG_UPDATE(OTG_PIXEL_RATE_CNTL[params->otg_inst],
+				PIPE_DTO_SRC_SEL[params->otg_inst], 2);
+	} else {
+		REG_UPDATE_2(OTG_PIXEL_RATE_CNTL[params->otg_inst],
+				DTBCLK_DTO_ENABLE[params->otg_inst], 0,
+				PIPE_DTO_SRC_SEL[params->otg_inst], 1);
+
+		REG_WRITE(DTBCLK_DTO_MODULO[params->otg_inst], 0);
+		REG_WRITE(DTBCLK_DTO_PHASE[params->otg_inst], 0);
+	}
+}
+
+void dccg314_set_dpstreamclk(
+		struct dccg *dccg,
+		enum streamclk_source src,
+		int otg_inst,
+		int dp_hpo_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	/* set the dtbclk_p source */
+	dccg314_set_dtbclk_p_src(dccg, src, otg_inst);
+
+	/* enabled to select one of the DTBCLKs for pipe */
+	switch (dp_hpo_inst) {
+	case 0:
+		REG_UPDATE_2(DPSTREAMCLK_CNTL,
+					DPSTREAMCLK0_EN, (src == REFCLK) ? 0 : 1,
+					DPSTREAMCLK0_SRC_SEL, otg_inst);
+		break;
+	case 1:
+		REG_UPDATE_2(DPSTREAMCLK_CNTL,
+					DPSTREAMCLK1_EN, (src == REFCLK) ? 0 : 1,
+					DPSTREAMCLK1_SRC_SEL, otg_inst);
+		break;
+	case 2:
+		REG_UPDATE_2(DPSTREAMCLK_CNTL,
+					DPSTREAMCLK2_EN, (src == REFCLK) ? 0 : 1,
+					DPSTREAMCLK2_SRC_SEL, otg_inst);
+		break;
+	case 3:
+		REG_UPDATE_2(DPSTREAMCLK_CNTL,
+					DPSTREAMCLK3_EN, (src == REFCLK) ? 0 : 1,
+					DPSTREAMCLK3_SRC_SEL, otg_inst);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+static void dccg314_init(struct dccg *dccg)
+{
+	int otg_inst;
+
+	/* Set HPO stream encoder to use refclk to avoid case where PHY is
+	 * disabled and SYMCLK32 for HPO SE is sourced from PHYD32CLK which
+	 * will cause DCN to hang.
+	 */
+	for (otg_inst = 0; otg_inst < 4; otg_inst++)
+		dccg31_disable_symclk32_se(dccg, otg_inst);
+
+	if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_le)
+		for (otg_inst = 0; otg_inst < 2; otg_inst++)
+			dccg31_disable_symclk32_le(dccg, otg_inst);
+
+	if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream)
+		for (otg_inst = 0; otg_inst < 4; otg_inst++)
+			dccg314_set_dpstreamclk(dccg, REFCLK, otg_inst,
+						otg_inst);
+
+	if (dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk)
+		for (otg_inst = 0; otg_inst < 5; otg_inst++)
+			dccg31_set_physymclk(dccg, otg_inst,
+					     PHYSYMCLK_FORCE_SRC_SYMCLK, false);
+}
+
+static void dccg314_set_valid_pixel_rate(
+		struct dccg *dccg,
+		int ref_dtbclk_khz,
+		int otg_inst,
+		int pixclk_khz)
+{
+	struct dtbclk_dto_params dto_params = {0};
+
+	dto_params.ref_dtbclk_khz = ref_dtbclk_khz;
+	dto_params.otg_inst = otg_inst;
+	dto_params.pixclk_khz = pixclk_khz;
+
+	dccg314_set_dtbclk_dto(dccg, &dto_params);
+}
+
+static void dccg314_dpp_root_clock_control(
+		struct dccg *dccg,
+		unsigned int dpp_inst,
+		bool clock_on)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (dccg->dpp_clock_gated[dpp_inst] != clock_on)
+		return;
+
+	if (clock_on) {
+		/* turn off the DTO and leave phase/modulo at max */
+		REG_UPDATE(DPPCLK_DTO_CTRL, DPPCLK_DTO_ENABLE[dpp_inst], 0);
+		REG_SET_2(DPPCLK_DTO_PARAM[dpp_inst], 0,
+			  DPPCLK0_DTO_PHASE, 0xFF,
+			  DPPCLK0_DTO_MODULO, 0xFF);
+	} else {
+		/* turn on the DTO to generate a 0hz clock */
+		REG_UPDATE(DPPCLK_DTO_CTRL, DPPCLK_DTO_ENABLE[dpp_inst], 1);
+		REG_SET_2(DPPCLK_DTO_PARAM[dpp_inst], 0,
+			  DPPCLK0_DTO_PHASE, 0,
+			  DPPCLK0_DTO_MODULO, 1);
+	}
+
+	dccg->dpp_clock_gated[dpp_inst] = !clock_on;
+}
+
+static const struct dccg_funcs dccg314_funcs = {
+	.update_dpp_dto = dccg31_update_dpp_dto,
+	.dpp_root_clock_control = dccg314_dpp_root_clock_control,
+	.get_dccg_ref_freq = dccg31_get_dccg_ref_freq,
+	.dccg_init = dccg314_init,
+	.set_dpstreamclk = dccg314_set_dpstreamclk,
+	.enable_symclk32_se = dccg31_enable_symclk32_se,
+	.disable_symclk32_se = dccg31_disable_symclk32_se,
+	.enable_symclk32_le = dccg31_enable_symclk32_le,
+	.disable_symclk32_le = dccg31_disable_symclk32_le,
+	.set_symclk32_le_root_clock_gating = dccg31_set_symclk32_le_root_clock_gating,
+	.set_physymclk = dccg31_set_physymclk,
+	.set_dtbclk_dto = dccg314_set_dtbclk_dto,
+	.set_audio_dtbclk_dto = dccg31_set_audio_dtbclk_dto,
+	.set_fifo_errdet_ovr_en = dccg2_set_fifo_errdet_ovr_en,
+	.otg_add_pixel = dccg31_otg_add_pixel,
+	.otg_drop_pixel = dccg31_otg_drop_pixel,
+	.set_dispclk_change_mode = dccg31_set_dispclk_change_mode,
+	.disable_dsc = dccg31_disable_dscclk,
+	.enable_dsc = dccg31_enable_dscclk,
+	.set_pixel_rate_div = dccg314_set_pixel_rate_div,
+	.get_pixel_rate_div = dccg314_get_pixel_rate_div,
+	.trigger_dio_fifo_resync = dccg314_trigger_dio_fifo_resync,
+	.set_valid_pixel_rate = dccg314_set_valid_pixel_rate,
+	.set_dtbclk_p_src = dccg314_set_dtbclk_p_src
+};
+
+struct dccg *dccg314_create(
+	struct dc_context *ctx,
+	const struct dccg_registers *regs,
+	const struct dccg_shift *dccg_shift,
+	const struct dccg_mask *dccg_mask)
+{
+	struct dcn_dccg *dccg_dcn = kzalloc(sizeof(*dccg_dcn), GFP_KERNEL);
+	struct dccg *base;
+
+	if (dccg_dcn == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	base = &dccg_dcn->base;
+	base->ctx = ctx;
+	base->funcs = &dccg314_funcs;
+
+	dccg_dcn->regs = regs;
+	dccg_dcn->dccg_shift = dccg_shift;
+	dccg_dcn->dccg_mask = dccg_mask;
+
+	return &dccg_dcn->base;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn314/dcn314_dccg.h b/drivers/gpu/drm/amd/display/dc/dccg/dcn314/dcn314_dccg.h
new file mode 100644
index 000000000000..60ea1d248deb
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn314/dcn314_dccg.h
@@ -0,0 +1,212 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCN314_DCCG_H__
+#define __DCN314_DCCG_H__
+
+#include "dcn31/dcn31_dccg.h"
+
+#define DCCG_SFII(block, reg_name, field_prefix, field_name, inst, post_fix)\
+	.field_prefix ## _ ## field_name[inst] = block ## inst ## _ ## reg_name ## __ ## field_prefix ## inst ## _ ## field_name ## post_fix
+
+
+#define DCCG_REG_LIST_DCN314() \
+	SR(DPPCLK_DTO_CTRL),\
+	DCCG_SRII(DTO_PARAM, DPPCLK, 0),\
+	DCCG_SRII(DTO_PARAM, DPPCLK, 1),\
+	DCCG_SRII(DTO_PARAM, DPPCLK, 2),\
+	DCCG_SRII(DTO_PARAM, DPPCLK, 3),\
+	DCCG_SRII(CLOCK_CNTL, HDMICHARCLK, 0),\
+	SR(PHYASYMCLK_CLOCK_CNTL),\
+	SR(PHYBSYMCLK_CLOCK_CNTL),\
+	SR(PHYCSYMCLK_CLOCK_CNTL),\
+	SR(PHYDSYMCLK_CLOCK_CNTL),\
+	SR(PHYESYMCLK_CLOCK_CNTL),\
+	SR(DPSTREAMCLK_CNTL),\
+	SR(HDMISTREAMCLK_CNTL),\
+	SR(SYMCLK32_SE_CNTL),\
+	SR(SYMCLK32_LE_CNTL),\
+	DCCG_SRII(PIXEL_RATE_CNTL, OTG, 0),\
+	DCCG_SRII(PIXEL_RATE_CNTL, OTG, 1),\
+	DCCG_SRII(PIXEL_RATE_CNTL, OTG, 2),\
+	DCCG_SRII(PIXEL_RATE_CNTL, OTG, 3),\
+	DCCG_SRII(MODULO, DTBCLK_DTO, 0),\
+	DCCG_SRII(MODULO, DTBCLK_DTO, 1),\
+	DCCG_SRII(MODULO, DTBCLK_DTO, 2),\
+	DCCG_SRII(MODULO, DTBCLK_DTO, 3),\
+	DCCG_SRII(PHASE, DTBCLK_DTO, 0),\
+	DCCG_SRII(PHASE, DTBCLK_DTO, 1),\
+	DCCG_SRII(PHASE, DTBCLK_DTO, 2),\
+	DCCG_SRII(PHASE, DTBCLK_DTO, 3),\
+	SR(DCCG_AUDIO_DTBCLK_DTO_MODULO),\
+	SR(DCCG_AUDIO_DTBCLK_DTO_PHASE),\
+	SR(DCCG_AUDIO_DTO_SOURCE),\
+	SR(DENTIST_DISPCLK_CNTL),\
+	SR(DSCCLK0_DTO_PARAM),\
+	SR(DSCCLK1_DTO_PARAM),\
+	SR(DSCCLK2_DTO_PARAM),\
+	SR(DSCCLK3_DTO_PARAM),\
+	SR(DSCCLK_DTO_CTRL),\
+	SR(DCCG_GATE_DISABLE_CNTL2),\
+	SR(DCCG_GATE_DISABLE_CNTL3),\
+	SR(HDMISTREAMCLK0_DTO_PARAM),\
+	SR(OTG_PIXEL_RATE_DIV),\
+	SR(DTBCLK_P_CNTL),\
+	SR(DCCG_AUDIO_DTO_SOURCE)
+
+#define DCCG_MASK_SH_LIST_DCN314_COMMON(mask_sh) \
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 0, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 1, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 2, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 3, mask_sh),\
+	DCCG_SF(DPPCLK0_DTO_PARAM, DPPCLK0_DTO_PHASE, mask_sh),\
+	DCCG_SF(DPPCLK0_DTO_PARAM, DPPCLK0_DTO_MODULO, mask_sh),\
+	DCCG_SF(HDMICHARCLK0_CLOCK_CNTL, HDMICHARCLK0_EN, mask_sh),\
+	DCCG_SF(HDMICHARCLK0_CLOCK_CNTL, HDMICHARCLK0_SRC_SEL, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK0_EN, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK1_EN, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK2_EN, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK3_EN, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK0_SRC_SEL, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK1_SRC_SEL, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK2_SRC_SEL, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK3_SRC_SEL, mask_sh),\
+	DCCG_SF(HDMISTREAMCLK_CNTL, HDMISTREAMCLK0_EN, mask_sh),\
+	DCCG_SF(HDMISTREAMCLK_CNTL, HDMISTREAMCLK0_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE0_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE1_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE2_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE3_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE0_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE1_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE2_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE3_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_LE_CNTL, SYMCLK32_LE0_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_LE_CNTL, SYMCLK32_LE1_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_LE_CNTL, SYMCLK32_LE0_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_LE_CNTL, SYMCLK32_LE1_EN, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLK_DTO, ENABLE, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLK_DTO, ENABLE, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLK_DTO, ENABLE, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLK_DTO, ENABLE, 3, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLKDTO, ENABLE_STATUS, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLKDTO, ENABLE_STATUS, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLKDTO, ENABLE_STATUS, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLKDTO, ENABLE_STATUS, 3, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 3, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 3, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG0_PIXEL_RATE_DIVK1, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG0_PIXEL_RATE_DIVK2, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG1_PIXEL_RATE_DIVK1, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG1_PIXEL_RATE_DIVK2, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG2_PIXEL_RATE_DIVK1, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG2_PIXEL_RATE_DIVK2, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG3_PIXEL_RATE_DIVK1, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG3_PIXEL_RATE_DIVK2, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG3_PIXEL_RATE_DIVK2, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P0_SRC_SEL, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P0_EN, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P1_SRC_SEL, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P1_EN, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P2_SRC_SEL, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P2_EN, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P3_SRC_SEL, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P3_EN, mask_sh),\
+	DCCG_SF(DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO0_SOURCE_SEL, mask_sh),\
+	DCCG_SF(DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO_SEL, mask_sh),\
+	DCCG_SF(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_CHG_MODE, mask_sh),\
+	DCCG_SF(DSCCLK0_DTO_PARAM, DSCCLK0_DTO_PHASE, mask_sh),\
+	DCCG_SF(DSCCLK0_DTO_PARAM, DSCCLK0_DTO_MODULO, mask_sh),\
+	DCCG_SF(DSCCLK1_DTO_PARAM, DSCCLK1_DTO_PHASE, mask_sh),\
+	DCCG_SF(DSCCLK1_DTO_PARAM, DSCCLK1_DTO_MODULO, mask_sh),\
+	DCCG_SF(DSCCLK2_DTO_PARAM, DSCCLK2_DTO_PHASE, mask_sh),\
+	DCCG_SF(DSCCLK2_DTO_PARAM, DSCCLK2_DTO_MODULO, mask_sh),\
+	DCCG_SF(DSCCLK3_DTO_PARAM, DSCCLK3_DTO_PHASE, mask_sh),\
+	DCCG_SF(DSCCLK3_DTO_PARAM, DSCCLK3_DTO_MODULO, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_SE0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_SE1_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_SE2_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_SE3_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_LE0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_LE1_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_SE0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_SE1_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_SE2_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_SE3_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_LE0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_LE1_GATE_DISABLE, mask_sh),\
+	DCCG_SF(HDMISTREAMCLK0_DTO_PARAM, HDMISTREAMCLK0_DTO_PHASE, mask_sh),\
+	DCCG_SF(HDMISTREAMCLK0_DTO_PARAM, HDMISTREAMCLK0_DTO_MODULO, mask_sh)
+
+#define DCCG_MASK_SH_LIST_DCN314(mask_sh) \
+	DCCG_MASK_SH_LIST_DCN314_COMMON(mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 0, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 1, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 2, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 3, mask_sh),\
+	DCCG_SF(PHYASYMCLK_CLOCK_CNTL, PHYASYMCLK_FORCE_EN, mask_sh),\
+	DCCG_SF(PHYASYMCLK_CLOCK_CNTL, PHYASYMCLK_FORCE_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYBSYMCLK_CLOCK_CNTL, PHYBSYMCLK_FORCE_EN, mask_sh),\
+	DCCG_SF(PHYBSYMCLK_CLOCK_CNTL, PHYBSYMCLK_FORCE_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYCSYMCLK_CLOCK_CNTL, PHYCSYMCLK_FORCE_EN, mask_sh),\
+	DCCG_SF(PHYCSYMCLK_CLOCK_CNTL, PHYCSYMCLK_FORCE_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYDSYMCLK_CLOCK_CNTL, PHYDSYMCLK_FORCE_EN, mask_sh),\
+	DCCG_SF(PHYDSYMCLK_CLOCK_CNTL, PHYDSYMCLK_FORCE_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYESYMCLK_CLOCK_CNTL, PHYESYMCLK_FORCE_EN, mask_sh),\
+	DCCG_SF(PHYESYMCLK_CLOCK_CNTL, PHYESYMCLK_FORCE_SRC_SEL, mask_sh),\
+	DCCG_SF(HDMISTREAMCLK_CNTL, HDMISTREAMCLK0_DTO_FORCE_DIS, mask_sh),\
+	DCCG_SF(DSCCLK_DTO_CTRL, DSCCLK0_DTO_ENABLE, mask_sh),\
+	DCCG_SF(DSCCLK_DTO_CTRL, DSCCLK1_DTO_ENABLE, mask_sh),\
+	DCCG_SF(DSCCLK_DTO_CTRL, DSCCLK2_DTO_ENABLE, mask_sh),\
+	DCCG_SF(DSCCLK_DTO_CTRL, DSCCLK3_DTO_ENABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, PHYASYMCLK_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, PHYBSYMCLK_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, PHYCSYMCLK_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, PHYDSYMCLK_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, PHYESYMCLK_GATE_DISABLE, mask_sh),\
+	DCCG_SF(HDMISTREAMCLK0_DTO_PARAM, HDMISTREAMCLK0_DTO_MODULO, mask_sh),\
+	DCCG_SF(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_RDIVIDER, mask_sh),\
+	DCCG_SF(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_WDIVIDER, mask_sh)
+
+struct dccg *dccg314_create(
+	struct dc_context *ctx,
+	const struct dccg_registers *regs,
+	const struct dccg_shift *dccg_shift,
+	const struct dccg_mask *dccg_mask);
+
+void dccg314_set_dpstreamclk(
+		struct dccg *dccg,
+		enum streamclk_source src,
+		int otg_inst,
+		int dp_hpo_inst);
+
+#endif //__DCN314_DCCG_H__
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn32/dcn32_dccg.c b/drivers/gpu/drm/amd/display/dc/dccg/dcn32/dcn32_dccg.c
new file mode 100644
index 000000000000..21a6ca5ca192
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn32/dcn32_dccg.c
@@ -0,0 +1,375 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+#include "core_types.h"
+#include "dcn32_dccg.h"
+
+#define TO_DCN_DCCG(dccg)\
+	container_of(dccg, struct dcn_dccg, base)
+
+#define REG(reg) \
+	(dccg_dcn->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	dccg_dcn->dccg_shift->field_name, dccg_dcn->dccg_mask->field_name
+
+#define CTX \
+	dccg_dcn->base.ctx
+#define DC_LOGGER \
+	dccg->ctx->logger
+
+static void dccg32_trigger_dio_fifo_resync(
+	struct dccg *dccg)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+	uint32_t dispclk_rdivider_value = 0;
+
+	REG_GET(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_RDIVIDER, &dispclk_rdivider_value);
+
+	/* Not valid for the WDIVIDER to be set to 0 */
+	if (dispclk_rdivider_value != 0)
+		REG_UPDATE(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_WDIVIDER, dispclk_rdivider_value);
+}
+
+static void dccg32_get_pixel_rate_div(
+		struct dccg *dccg,
+		uint32_t otg_inst,
+		uint32_t *k1,
+		uint32_t *k2)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+	uint32_t val_k1 = PIXEL_RATE_DIV_NA, val_k2 = PIXEL_RATE_DIV_NA;
+
+	*k1 = PIXEL_RATE_DIV_NA;
+	*k2 = PIXEL_RATE_DIV_NA;
+
+	switch (otg_inst) {
+	case 0:
+		REG_GET_2(OTG_PIXEL_RATE_DIV,
+			OTG0_PIXEL_RATE_DIVK1, &val_k1,
+			OTG0_PIXEL_RATE_DIVK2, &val_k2);
+		break;
+	case 1:
+		REG_GET_2(OTG_PIXEL_RATE_DIV,
+			OTG1_PIXEL_RATE_DIVK1, &val_k1,
+			OTG1_PIXEL_RATE_DIVK2, &val_k2);
+		break;
+	case 2:
+		REG_GET_2(OTG_PIXEL_RATE_DIV,
+			OTG2_PIXEL_RATE_DIVK1, &val_k1,
+			OTG2_PIXEL_RATE_DIVK2, &val_k2);
+		break;
+	case 3:
+		REG_GET_2(OTG_PIXEL_RATE_DIV,
+			OTG3_PIXEL_RATE_DIVK1, &val_k1,
+			OTG3_PIXEL_RATE_DIVK2, &val_k2);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+	*k1 = val_k1;
+	*k2 = val_k2;
+}
+
+static void dccg32_set_pixel_rate_div(
+		struct dccg *dccg,
+		uint32_t otg_inst,
+		enum pixel_rate_div k1,
+		enum pixel_rate_div k2)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+	uint32_t cur_k1 = PIXEL_RATE_DIV_NA;
+	uint32_t cur_k2 = PIXEL_RATE_DIV_NA;
+
+	// Don't program 0xF into the register field. Not valid since
+	// K1 / K2 field is only 1 / 2 bits wide
+	if (k1 == PIXEL_RATE_DIV_NA || k2 == PIXEL_RATE_DIV_NA) {
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+	dccg32_get_pixel_rate_div(dccg, otg_inst, &cur_k1, &cur_k2);
+	if (k1 == cur_k1 && k2 == cur_k2)
+		return;
+
+	switch (otg_inst) {
+	case 0:
+		REG_UPDATE_2(OTG_PIXEL_RATE_DIV,
+				OTG0_PIXEL_RATE_DIVK1, k1,
+				OTG0_PIXEL_RATE_DIVK2, k2);
+		break;
+	case 1:
+		REG_UPDATE_2(OTG_PIXEL_RATE_DIV,
+				OTG1_PIXEL_RATE_DIVK1, k1,
+				OTG1_PIXEL_RATE_DIVK2, k2);
+		break;
+	case 2:
+		REG_UPDATE_2(OTG_PIXEL_RATE_DIV,
+				OTG2_PIXEL_RATE_DIVK1, k1,
+				OTG2_PIXEL_RATE_DIVK2, k2);
+		break;
+	case 3:
+		REG_UPDATE_2(OTG_PIXEL_RATE_DIV,
+				OTG3_PIXEL_RATE_DIVK1, k1,
+				OTG3_PIXEL_RATE_DIVK2, k2);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+static void dccg32_set_dtbclk_p_src(
+		struct dccg *dccg,
+		enum streamclk_source src,
+		uint32_t otg_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	uint32_t p_src_sel = 0; /* selects dprefclk */
+	if (src == DTBCLK0)
+		p_src_sel = 2;  /* selects dtbclk0 */
+
+	switch (otg_inst) {
+	case 0:
+		if (src == REFCLK)
+			REG_UPDATE(DTBCLK_P_CNTL,
+					DTBCLK_P0_EN, 0);
+		else
+			REG_UPDATE_2(DTBCLK_P_CNTL,
+					DTBCLK_P0_SRC_SEL, p_src_sel,
+					DTBCLK_P0_EN, 1);
+		break;
+	case 1:
+		if (src == REFCLK)
+			REG_UPDATE(DTBCLK_P_CNTL,
+					DTBCLK_P1_EN, 0);
+		else
+			REG_UPDATE_2(DTBCLK_P_CNTL,
+					DTBCLK_P1_SRC_SEL, p_src_sel,
+					DTBCLK_P1_EN, 1);
+		break;
+	case 2:
+		if (src == REFCLK)
+			REG_UPDATE(DTBCLK_P_CNTL,
+					DTBCLK_P2_EN, 0);
+		else
+			REG_UPDATE_2(DTBCLK_P_CNTL,
+					DTBCLK_P2_SRC_SEL, p_src_sel,
+					DTBCLK_P2_EN, 1);
+		break;
+	case 3:
+		if (src == REFCLK)
+			REG_UPDATE(DTBCLK_P_CNTL,
+					DTBCLK_P3_EN, 0);
+		else
+			REG_UPDATE_2(DTBCLK_P_CNTL,
+					DTBCLK_P3_SRC_SEL, p_src_sel,
+					DTBCLK_P3_EN, 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+}
+
+/* Controls the generation of pixel valid for OTG in (OTG -> HPO case) */
+static void dccg32_set_dtbclk_dto(
+		struct dccg *dccg,
+		const struct dtbclk_dto_params *params)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+	/* DTO Output Rate / Pixel Rate = 1/4 */
+	int req_dtbclk_khz = params->pixclk_khz / 4;
+
+	if (params->ref_dtbclk_khz && req_dtbclk_khz) {
+		uint32_t modulo, phase;
+
+		// phase / modulo = dtbclk / dtbclk ref
+		modulo = params->ref_dtbclk_khz * 1000;
+		phase = req_dtbclk_khz * 1000;
+
+		REG_WRITE(DTBCLK_DTO_MODULO[params->otg_inst], modulo);
+		REG_WRITE(DTBCLK_DTO_PHASE[params->otg_inst], phase);
+
+		REG_UPDATE(OTG_PIXEL_RATE_CNTL[params->otg_inst],
+				DTBCLK_DTO_ENABLE[params->otg_inst], 1);
+
+		REG_WAIT(OTG_PIXEL_RATE_CNTL[params->otg_inst],
+				DTBCLKDTO_ENABLE_STATUS[params->otg_inst], 1,
+				1, 100);
+
+		/* program OTG_PIXEL_RATE_DIV for DIVK1 and DIVK2 fields */
+		dccg32_set_pixel_rate_div(dccg, params->otg_inst, PIXEL_RATE_DIV_BY_1, PIXEL_RATE_DIV_BY_1);
+
+		/* The recommended programming sequence to enable DTBCLK DTO to generate
+		 * valid pixel HPO DPSTREAM ENCODER, specifies that DTO source select should
+		 * be set only after DTO is enabled
+		 */
+		REG_UPDATE(OTG_PIXEL_RATE_CNTL[params->otg_inst],
+				PIPE_DTO_SRC_SEL[params->otg_inst], 2);
+	} else {
+		REG_UPDATE_2(OTG_PIXEL_RATE_CNTL[params->otg_inst],
+				DTBCLK_DTO_ENABLE[params->otg_inst], 0,
+				PIPE_DTO_SRC_SEL[params->otg_inst], params->is_hdmi ? 0 : 1);
+		REG_WRITE(DTBCLK_DTO_MODULO[params->otg_inst], 0);
+		REG_WRITE(DTBCLK_DTO_PHASE[params->otg_inst], 0);
+	}
+}
+
+static void dccg32_set_valid_pixel_rate(
+		struct dccg *dccg,
+		int ref_dtbclk_khz,
+		int otg_inst,
+		int pixclk_khz)
+{
+	struct dtbclk_dto_params dto_params = {0};
+
+	dto_params.ref_dtbclk_khz = ref_dtbclk_khz;
+	dto_params.otg_inst = otg_inst;
+	dto_params.pixclk_khz = pixclk_khz;
+	dto_params.is_hdmi = true;
+
+	dccg32_set_dtbclk_dto(dccg, &dto_params);
+}
+
+static void dccg32_get_dccg_ref_freq(struct dccg *dccg,
+		unsigned int xtalin_freq_inKhz,
+		unsigned int *dccg_ref_freq_inKhz)
+{
+	/*
+	 * Assume refclk is sourced from xtalin
+	 * expect 100MHz
+	 */
+	*dccg_ref_freq_inKhz = xtalin_freq_inKhz;
+	return;
+}
+
+static void dccg32_set_dpstreamclk(
+		struct dccg *dccg,
+		enum streamclk_source src,
+		int otg_inst,
+		int dp_hpo_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	/* set the dtbclk_p source */
+	/* always program refclk as DTBCLK. No use-case expected to require DPREFCLK as refclk */
+	dccg32_set_dtbclk_p_src(dccg, DTBCLK0, otg_inst);
+
+	/* enabled to select one of the DTBCLKs for pipe */
+	switch (dp_hpo_inst) {
+	case 0:
+		REG_UPDATE_2(DPSTREAMCLK_CNTL,
+			     DPSTREAMCLK0_EN,
+			     (src == REFCLK) ? 0 : 1, DPSTREAMCLK0_SRC_SEL, otg_inst);
+		break;
+	case 1:
+		REG_UPDATE_2(DPSTREAMCLK_CNTL, DPSTREAMCLK1_EN,
+			     (src == REFCLK) ? 0 : 1, DPSTREAMCLK1_SRC_SEL, otg_inst);
+		break;
+	case 2:
+		REG_UPDATE_2(DPSTREAMCLK_CNTL, DPSTREAMCLK2_EN,
+			     (src == REFCLK) ? 0 : 1, DPSTREAMCLK2_SRC_SEL, otg_inst);
+		break;
+	case 3:
+		REG_UPDATE_2(DPSTREAMCLK_CNTL, DPSTREAMCLK3_EN,
+			     (src == REFCLK) ? 0 : 1, DPSTREAMCLK3_SRC_SEL, otg_inst);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+static void dccg32_otg_add_pixel(struct dccg *dccg,
+		uint32_t otg_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	REG_UPDATE(OTG_PIXEL_RATE_CNTL[otg_inst],
+			OTG_ADD_PIXEL[otg_inst], 1);
+}
+
+static void dccg32_otg_drop_pixel(struct dccg *dccg,
+		uint32_t otg_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	REG_UPDATE(OTG_PIXEL_RATE_CNTL[otg_inst],
+			OTG_DROP_PIXEL[otg_inst], 1);
+}
+
+static const struct dccg_funcs dccg32_funcs = {
+	.update_dpp_dto = dccg2_update_dpp_dto,
+	.get_dccg_ref_freq = dccg32_get_dccg_ref_freq,
+	.dccg_init = dccg31_init,
+	.set_dpstreamclk = dccg32_set_dpstreamclk,
+	.enable_symclk32_se = dccg31_enable_symclk32_se,
+	.disable_symclk32_se = dccg31_disable_symclk32_se,
+	.enable_symclk32_le = dccg31_enable_symclk32_le,
+	.disable_symclk32_le = dccg31_disable_symclk32_le,
+	.set_physymclk = dccg31_set_physymclk,
+	.set_dtbclk_dto = dccg32_set_dtbclk_dto,
+	.set_valid_pixel_rate = dccg32_set_valid_pixel_rate,
+	.set_fifo_errdet_ovr_en = dccg2_set_fifo_errdet_ovr_en,
+	.set_audio_dtbclk_dto = dccg31_set_audio_dtbclk_dto,
+	.otg_add_pixel = dccg32_otg_add_pixel,
+	.otg_drop_pixel = dccg32_otg_drop_pixel,
+	.set_pixel_rate_div = dccg32_set_pixel_rate_div,
+	.get_pixel_rate_div = dccg32_get_pixel_rate_div,
+	.trigger_dio_fifo_resync = dccg32_trigger_dio_fifo_resync,
+	.set_dtbclk_p_src = dccg32_set_dtbclk_p_src,
+};
+
+struct dccg *dccg32_create(
+	struct dc_context *ctx,
+	const struct dccg_registers *regs,
+	const struct dccg_shift *dccg_shift,
+	const struct dccg_mask *dccg_mask)
+{
+	struct dcn_dccg *dccg_dcn = kzalloc(sizeof(*dccg_dcn), GFP_KERNEL);
+	struct dccg *base;
+
+	if (dccg_dcn == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	base = &dccg_dcn->base;
+	base->ctx = ctx;
+	base->funcs = &dccg32_funcs;
+
+	dccg_dcn->regs = regs;
+	dccg_dcn->dccg_shift = dccg_shift;
+	dccg_dcn->dccg_mask = dccg_mask;
+
+	return &dccg_dcn->base;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn32/dcn32_dccg.h b/drivers/gpu/drm/amd/display/dc/dccg/dcn32/dcn32_dccg.h
new file mode 100644
index 000000000000..cf5508718122
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn32/dcn32_dccg.h
@@ -0,0 +1,125 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCN32_DCCG_H__
+#define __DCN32_DCCG_H__
+
+#include "dcn31/dcn31_dccg.h"
+
+#define DCCG_SFII(block, reg_name, field_prefix, field_name, inst, post_fix)\
+	.field_prefix ## _ ## field_name[inst] = block ## inst ## _ ## reg_name ## __ ## field_prefix ## inst ## _ ## field_name ## post_fix
+
+#define DCCG_MASK_SH_LIST_DCN32(mask_sh) \
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 0, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 0, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 1, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 1, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 2, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 2, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_ENABLE, DPPCLK, 3, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 3, mask_sh),\
+	DCCG_SF(DPPCLK0_DTO_PARAM, DPPCLK0_DTO_PHASE, mask_sh),\
+	DCCG_SF(DPPCLK0_DTO_PARAM, DPPCLK0_DTO_MODULO, mask_sh),\
+	DCCG_SF(HDMICHARCLK0_CLOCK_CNTL, HDMICHARCLK0_EN, mask_sh),\
+	DCCG_SF(HDMICHARCLK0_CLOCK_CNTL, HDMICHARCLK0_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYASYMCLK_CLOCK_CNTL, PHYASYMCLK_FORCE_EN, mask_sh),\
+	DCCG_SF(PHYASYMCLK_CLOCK_CNTL, PHYASYMCLK_FORCE_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYBSYMCLK_CLOCK_CNTL, PHYBSYMCLK_FORCE_EN, mask_sh),\
+	DCCG_SF(PHYBSYMCLK_CLOCK_CNTL, PHYBSYMCLK_FORCE_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYCSYMCLK_CLOCK_CNTL, PHYCSYMCLK_FORCE_EN, mask_sh),\
+	DCCG_SF(PHYCSYMCLK_CLOCK_CNTL, PHYCSYMCLK_FORCE_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYDSYMCLK_CLOCK_CNTL, PHYDSYMCLK_FORCE_EN, mask_sh),\
+	DCCG_SF(PHYDSYMCLK_CLOCK_CNTL, PHYDSYMCLK_FORCE_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYESYMCLK_CLOCK_CNTL, PHYESYMCLK_FORCE_EN, mask_sh),\
+	DCCG_SF(PHYESYMCLK_CLOCK_CNTL, PHYESYMCLK_FORCE_SRC_SEL, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK0_EN, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK1_EN, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK2_EN, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK3_EN, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK0_SRC_SEL, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK1_SRC_SEL, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK2_SRC_SEL, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK3_SRC_SEL, mask_sh),\
+	DCCG_SF(HDMISTREAMCLK_CNTL, HDMISTREAMCLK0_EN, mask_sh),\
+	DCCG_SF(HDMISTREAMCLK_CNTL, HDMISTREAMCLK0_DTO_FORCE_DIS, mask_sh),\
+	DCCG_SF(HDMISTREAMCLK_CNTL, HDMISTREAMCLK0_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE0_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE1_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE2_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE3_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE0_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE1_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE2_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE3_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_LE_CNTL, SYMCLK32_LE0_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_LE_CNTL, SYMCLK32_LE1_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_LE_CNTL, SYMCLK32_LE0_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_LE_CNTL, SYMCLK32_LE1_EN, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLK_DTO, ENABLE, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLK_DTO, ENABLE, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLK_DTO, ENABLE, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLK_DTO, ENABLE, 3, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLKDTO, ENABLE_STATUS, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLKDTO, ENABLE_STATUS, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLKDTO, ENABLE_STATUS, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLKDTO, ENABLE_STATUS, 3, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 3, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 3, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG0_PIXEL_RATE_DIVK1, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG0_PIXEL_RATE_DIVK2, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG1_PIXEL_RATE_DIVK1, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG1_PIXEL_RATE_DIVK2, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG2_PIXEL_RATE_DIVK1, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG2_PIXEL_RATE_DIVK2, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG3_PIXEL_RATE_DIVK1, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG3_PIXEL_RATE_DIVK2, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG3_PIXEL_RATE_DIVK2, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P0_SRC_SEL, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P0_EN, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P1_SRC_SEL, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P1_EN, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P2_SRC_SEL, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P2_EN, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P3_SRC_SEL, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P3_EN, mask_sh),\
+	DCCG_SF(DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO_SEL, mask_sh),\
+	DCCG_SF(DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO0_SOURCE_SEL, mask_sh),\
+	DCCG_SF(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_CHG_DONE, mask_sh),\
+	DCCG_SF(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_RDIVIDER, mask_sh),\
+	DCCG_SF(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_WDIVIDER, mask_sh)
+
+struct dccg *dccg32_create(
+	struct dc_context *ctx,
+	const struct dccg_registers *regs,
+	const struct dccg_shift *dccg_shift,
+	const struct dccg_mask *dccg_mask);
+
+#endif //__DCN32_DCCG_H__
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.c b/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.c
new file mode 100644
index 000000000000..de6d62401362
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.c
@@ -0,0 +1,2470 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "reg_helper.h"
+#include "core_types.h"
+#include "resource.h"
+#include "dcn35_dccg.h"
+
+#define TO_DCN_DCCG(dccg)\
+	container_of(dccg, struct dcn_dccg, base)
+
+#define REG(reg) \
+	(dccg_dcn->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	dccg_dcn->dccg_shift->field_name, dccg_dcn->dccg_mask->field_name
+
+#define CTX \
+	dccg_dcn->base.ctx
+#include "logger_types.h"
+#define DC_LOGGER \
+	dccg->ctx->logger
+
+enum symclk_fe_source {
+	SYMCLK_FE_SYMCLK_A = 0,	// Select functional clock from backend symclk A
+	SYMCLK_FE_SYMCLK_B,
+	SYMCLK_FE_SYMCLK_C,
+	SYMCLK_FE_SYMCLK_D,
+	SYMCLK_FE_SYMCLK_E,
+	SYMCLK_FE_REFCLK = 0xFF,	// Arbitrary value to pass refclk selection in software
+};
+
+enum symclk_be_source {
+	SYMCLK_BE_PHYCLK = 0,	// Select phy clk when sym_clk_enable = 1
+	SYMCLK_BE_DPIACLK_810 = 4,
+	SYMCLK_BE_DPIACLK_162 = 5,
+	SYMCLK_BE_DPIACLK_540 = 6,
+	SYMCLK_BE_DPIACLK_270 = 7,
+	SYMCLK_BE_REFCLK = 0xFF,	// Arbitrary value to pass refclk selection in software
+};
+
+enum physymclk_source {
+	PHYSYMCLK_PHYCLK = 0,		// Select symclk as source of clock which is output to PHY through DCIO.
+	PHYSYMCLK_PHYD18CLK,		// Select phyd18clk as the source of clock which is output to PHY through DCIO.
+	PHYSYMCLK_PHYD32CLK,		// Select phyd32clk as the source of clock which is output to PHY through DCIO.
+	PHYSYMCLK_REFCLK = 0xFF,	// Arbitrary value to pass refclk selection in software
+};
+
+enum dtbclk_source {
+	DTBCLK_DPREFCLK = 0,		// Selects source for DTBCLK_P# as DPREFCLK (src sel 0 and 1 are same)
+	DTBCLK_DPREFCLK_0,			// Selects source for DTBCLK_P# as DPREFCLK (src sel 0 and 1 are same)
+	DTBCLK_DTBCLK0,				// Selects source for DTBCLK_P# as DTBCLK0
+	DTBCLK_DTBCLK1,				// Selects source for DTBCLK_P# as DTBCLK0
+	DTBCLK_REFCLK = 0xFF,		// Arbitrary value to pass refclk selection in software
+};
+
+enum dppclk_clock_source {
+	DPP_REFCLK = 0,				// refclk is selected
+	DPP_DCCG_DTO,				// Functional clock selected is DTO tuned DPPCLK
+};
+
+enum dp_stream_clk_source {
+	DP_STREAM_DTBCLK_P0 = 0,	// Selects functional for DP_STREAM_CLK as DTBCLK_P#
+	DP_STREAM_DTBCLK_P1,
+	DP_STREAM_DTBCLK_P2,
+	DP_STREAM_DTBCLK_P3,
+	DP_STREAM_DTBCLK_P4,
+	DP_STREAM_DTBCLK_P5,
+	DP_STREAM_REFCLK = 0xFF,	// Arbitrary value to pass refclk selection in software
+};
+
+enum hdmi_char_clk {
+	HDMI_CHAR_PHYAD18CLK = 0,	// Selects functional for hdmi_char_clk as UNIPHYA PHYD18CLK
+	HDMI_CHAR_PHYBD18CLK,
+	HDMI_CHAR_PHYCD18CLK,
+	HDMI_CHAR_PHYDD18CLK,
+	HDMI_CHAR_PHYED18CLK,
+	HDMI_CHAR_REFCLK = 0xFF,	// Arbitrary value to pass refclk selection in software
+};
+
+enum hdmi_stream_clk_source {
+	HDMI_STREAM_DTBCLK_P0 = 0,	// Selects functional for HDMI_STREAM_CLK as DTBCLK_P#
+	HDMI_STREAM_DTBCLK_P1,
+	HDMI_STREAM_DTBCLK_P2,
+	HDMI_STREAM_DTBCLK_P3,
+	HDMI_STREAM_DTBCLK_P4,
+	HDMI_STREAM_DTBCLK_P5,
+	HDMI_STREAM_REFCLK = 0xFF,	// Arbitrary value to pass refclk selection in software
+};
+
+enum symclk32_se_clk_source {
+	SYMCLK32_SE_PHYAD32CLK = 0,	// Selects functional for SYMCLK32 as UNIPHYA PHYD32CLK
+	SYMCLK32_SE_PHYBD32CLK,
+	SYMCLK32_SE_PHYCD32CLK,
+	SYMCLK32_SE_PHYDD32CLK,
+	SYMCLK32_SE_PHYED32CLK,
+	SYMCLK32_SE_REFCLK = 0xFF,	// Arbitrary value to pass refclk selection in software
+};
+
+enum symclk32_le_clk_source {
+	SYMCLK32_LE_PHYAD32CLK = 0,	// Selects functional for SYMCLK32 as UNIPHYA PHYD32CLK
+	SYMCLK32_LE_PHYBD32CLK,
+	SYMCLK32_LE_PHYCD32CLK,
+	SYMCLK32_LE_PHYDD32CLK,
+	SYMCLK32_LE_PHYED32CLK,
+	SYMCLK32_LE_REFCLK = 0xFF,	// Arbitrary value to pass refclk selection in software
+};
+
+enum dsc_clk_source {
+	DSC_CLK_REF_CLK = 0,			// Ref clock selected for DSC_CLK
+	DSC_DTO_TUNED_CK_GPU_DISCLK_3,	// DTO divided clock selected as functional clock
+};
+
+
+static void dccg35_set_dsc_clk_rcg(struct dccg *dccg, int inst, bool allow_rcg)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (!dccg->ctx->dc->debug.root_clock_optimization.bits.dsc && allow_rcg)
+		return;
+
+	switch (inst) {
+	case 0:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL6, DSCCLK0_ROOT_GATE_DISABLE, allow_rcg ? 0 : 1);
+		break;
+	case 1:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL6, DSCCLK1_ROOT_GATE_DISABLE, allow_rcg ? 0 : 1);
+		break;
+	case 2:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL6, DSCCLK2_ROOT_GATE_DISABLE, allow_rcg ? 0 : 1);
+		break;
+	case 3:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL6, DSCCLK3_ROOT_GATE_DISABLE, allow_rcg ? 0 : 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+	/* Wait for clock to ramp */
+	if (!allow_rcg)
+		udelay(10);
+}
+
+static void dccg35_set_symclk32_se_rcg(
+	struct dccg *dccg,
+	int inst,
+	bool enable)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (!dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se && enable)
+		return;
+
+	/* SYMCLK32_ROOT_SE#_GATE_DISABLE will clock gate in DCCG */
+	/* SYMCLK32_SE#_GATE_DISABLE will clock gate in HPO only */
+	switch (inst) {
+	case 0:
+		REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+				   SYMCLK32_SE0_GATE_DISABLE, enable ? 0 : 1,
+				   SYMCLK32_ROOT_SE0_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 1:
+		REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+				   SYMCLK32_SE1_GATE_DISABLE, enable ? 0 : 1,
+				   SYMCLK32_ROOT_SE1_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 2:
+		REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+				   SYMCLK32_SE2_GATE_DISABLE, enable ? 0 : 1,
+				   SYMCLK32_ROOT_SE2_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 3:
+		REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+				   SYMCLK32_SE3_GATE_DISABLE, enable ? 0 : 1,
+				   SYMCLK32_ROOT_SE3_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+static void dccg35_set_symclk32_le_rcg(
+	struct dccg *dccg,
+	int inst,
+	bool enable)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (!dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_le && enable)
+		return;
+
+	switch (inst) {
+	case 0:
+		REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+				   SYMCLK32_LE0_GATE_DISABLE, enable ? 0 : 1,
+				   SYMCLK32_ROOT_LE0_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 1:
+		REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+				   SYMCLK32_LE1_GATE_DISABLE, enable ? 0 : 1,
+				   SYMCLK32_ROOT_LE1_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+static void dccg35_set_physymclk_rcg(
+	struct dccg *dccg,
+	int inst,
+	bool enable)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (!dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk && enable)
+		return;
+
+	switch (inst) {
+	case 0:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+				PHYASYMCLK_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 1:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+				PHYBSYMCLK_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 2:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+				PHYCSYMCLK_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 3:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+				PHYDSYMCLK_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 4:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+				PHYESYMCLK_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+static void dccg35_set_symclk_fe_rcg(
+	struct dccg *dccg,
+	int inst,
+	bool enable)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (!dccg->ctx->dc->debug.root_clock_optimization.bits.symclk_fe && enable)
+		return;
+
+	switch (inst) {
+	case 0:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+				   SYMCLKA_FE_GATE_DISABLE, enable ? 0 : 1);
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL5,
+				   SYMCLKA_FE_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 1:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+				   SYMCLKB_FE_GATE_DISABLE, enable ? 0 : 1);
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL5,
+				   SYMCLKB_FE_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 2:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+				   SYMCLKC_FE_GATE_DISABLE, enable ? 0 : 1);
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL5,
+				   SYMCLKC_FE_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 3:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+				   SYMCLKD_FE_GATE_DISABLE, enable ? 0 : 1);
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL5,
+				   SYMCLKD_FE_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 4:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+				   SYMCLKE_FE_GATE_DISABLE, enable ? 0 : 1);
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL5,
+				   SYMCLKE_FE_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+static void dccg35_set_symclk_be_rcg(
+	struct dccg *dccg,
+	int inst,
+	bool enable)
+{
+
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	/* TBD add symclk_be in rcg control bits */
+	if (!dccg->ctx->dc->debug.root_clock_optimization.bits.symclk_fe && enable)
+		return;
+
+	switch (inst) {
+	case 0:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+				   SYMCLKA_GATE_DISABLE, enable ? 0 : 1);
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL5,
+				   SYMCLKA_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 1:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+				   SYMCLKB_GATE_DISABLE, enable ? 0 : 1);
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL5,
+				   SYMCLKB_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 2:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+				   SYMCLKC_GATE_DISABLE, enable ? 0 : 1);
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL5,
+				   SYMCLKC_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 3:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+				   SYMCLKD_GATE_DISABLE, enable ? 0 : 1);
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL5,
+				   SYMCLKD_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 4:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+				   SYMCLKE_GATE_DISABLE, enable ? 0 : 1);
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL5,
+				   SYMCLKE_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+static void dccg35_set_dtbclk_p_rcg(struct dccg *dccg, int inst, bool enable)
+{
+
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (!dccg->ctx->dc->debug.root_clock_optimization.bits.dpp && enable)
+		return;
+
+	switch (inst) {
+	case 0:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P0_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 1:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P1_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 2:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P2_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 3:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P3_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		break;
+	}
+}
+
+static void dccg35_set_dppclk_rcg(struct dccg *dccg, int inst, bool allow_rcg)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (!dccg->ctx->dc->debug.root_clock_optimization.bits.dpp && allow_rcg)
+		return;
+
+	switch (inst) {
+	case 0:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL6, DPPCLK0_ROOT_GATE_DISABLE, allow_rcg ? 0 : 1);
+		break;
+	case 1:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL6, DPPCLK1_ROOT_GATE_DISABLE, allow_rcg ? 0 : 1);
+		break;
+	case 2:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL6, DPPCLK2_ROOT_GATE_DISABLE, allow_rcg ? 0 : 1);
+		break;
+	case 3:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL6, DPPCLK3_ROOT_GATE_DISABLE, allow_rcg ? 0 : 1);
+		break;
+	default:
+	BREAK_TO_DEBUGGER();
+		break;
+	}
+
+	/* Wait for clock to ramp */
+	if (!allow_rcg)
+		udelay(10);
+}
+
+static void dccg35_set_dpstreamclk_rcg(
+	struct dccg *dccg,
+	int inst,
+	bool enable)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (!dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream && enable)
+		return;
+
+	switch (inst) {
+	case 0:
+		REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL5,
+					 DPSTREAMCLK0_GATE_DISABLE, enable ? 0 : 1,
+					 DPSTREAMCLK0_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 1:
+		REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL5,
+					 DPSTREAMCLK1_GATE_DISABLE, enable ? 0 : 1,
+					 DPSTREAMCLK1_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 2:
+		REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL5,
+				   DPSTREAMCLK2_GATE_DISABLE, enable ? 0 : 1,
+				   DPSTREAMCLK2_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 3:
+		REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL5,
+				   DPSTREAMCLK3_GATE_DISABLE, enable ? 0 : 1,
+				   DPSTREAMCLK3_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+static void dccg35_set_smclk32_se_rcg(
+		struct dccg *dccg,
+		int inst,
+		bool enable)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (!dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se && enable)
+		return;
+
+	switch (inst) {
+	case 0:
+		REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+					 SYMCLK32_SE0_GATE_DISABLE, enable ? 0 : 1,
+					 SYMCLK32_ROOT_SE0_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 1:
+		REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+					 SYMCLK32_SE1_GATE_DISABLE, enable ? 0 : 1,
+					 SYMCLK32_ROOT_SE1_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 2:
+		REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+					 SYMCLK32_SE2_GATE_DISABLE, enable ? 0 : 1,
+					 SYMCLK32_ROOT_SE2_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 3:
+		REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+					 SYMCLK32_SE3_GATE_DISABLE, enable ? 0 : 1,
+					 SYMCLK32_ROOT_SE3_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+static void dccg35_set_dsc_clk_src_new(struct dccg *dccg, int inst, enum dsc_clk_source src)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	/* DSCCLK#_EN=0 switches to refclock from functional clock */
+
+	switch (inst) {
+	case 0:
+		REG_UPDATE(DSCCLK_DTO_CTRL, DSCCLK0_EN, src);
+		break;
+	case 1:
+		REG_UPDATE(DSCCLK_DTO_CTRL, DSCCLK1_EN, src);
+		break;
+	case 2:
+		REG_UPDATE(DSCCLK_DTO_CTRL, DSCCLK2_EN, src);
+		break;
+	case 3:
+		REG_UPDATE(DSCCLK_DTO_CTRL, DSCCLK3_EN, src);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+static void dccg35_set_symclk32_se_src_new(
+	struct dccg *dccg,
+	int inst,
+	enum symclk32_se_clk_source src
+	)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	switch (inst) {
+	case 0:
+		REG_UPDATE_2(SYMCLK32_SE_CNTL,
+					 SYMCLK32_SE0_SRC_SEL, (src == SYMCLK32_SE_REFCLK) ? 0 : src,
+					 SYMCLK32_SE0_EN,  (src == SYMCLK32_SE_REFCLK) ? 0 : 1);
+		break;
+	case 1:
+		REG_UPDATE_2(SYMCLK32_SE_CNTL,
+					 SYMCLK32_SE1_SRC_SEL, (src == SYMCLK32_SE_REFCLK) ? 0 : src,
+					 SYMCLK32_SE1_EN, (src == SYMCLK32_SE_REFCLK) ? 0 : 1);
+		break;
+	case 2:
+		REG_UPDATE_2(SYMCLK32_SE_CNTL,
+					 SYMCLK32_SE2_SRC_SEL, (src == SYMCLK32_SE_REFCLK) ? 0 : src,
+					 SYMCLK32_SE2_EN, (src == SYMCLK32_SE_REFCLK) ? 0 : 1);
+		break;
+	case 3:
+		REG_UPDATE_2(SYMCLK32_SE_CNTL,
+					 SYMCLK32_SE3_SRC_SEL, (src == SYMCLK32_SE_REFCLK) ? 0 : src,
+					 SYMCLK32_SE3_EN, (src == SYMCLK32_SE_REFCLK) ? 0 : 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+static int
+dccg35_is_symclk32_se_src_functional_le_new(struct dccg *dccg, int symclk_32_se_inst, int symclk_32_le_inst)
+{
+	uint32_t en;
+	uint32_t src_sel;
+
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	REG_GET_2(SYMCLK32_SE_CNTL, SYMCLK32_SE3_SRC_SEL, &src_sel, SYMCLK32_SE3_EN, &en);
+
+	if (en == 1 && src_sel == symclk_32_le_inst)
+		return 1;
+
+	return 0;
+}
+
+
+static void dccg35_set_symclk32_le_src_new(
+	struct dccg *dccg,
+	int inst,
+	enum symclk32_le_clk_source src)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	switch (inst) {
+	case 0:
+		REG_UPDATE_2(SYMCLK32_LE_CNTL,
+					 SYMCLK32_LE0_SRC_SEL, (src == SYMCLK32_LE_REFCLK) ? 0 : src,
+					 SYMCLK32_LE0_EN, (src == SYMCLK32_LE_REFCLK) ? 0 : 1);
+		break;
+	case 1:
+		REG_UPDATE_2(SYMCLK32_LE_CNTL,
+					 SYMCLK32_LE1_SRC_SEL, (src == SYMCLK32_LE_REFCLK) ? 0 : src,
+					 SYMCLK32_LE1_EN, (src == SYMCLK32_LE_REFCLK) ? 0 : 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+static void dcn35_set_dppclk_src_new(struct dccg *dccg,
+				 int inst, enum dppclk_clock_source src)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	switch (inst) {
+	case 0:
+		REG_UPDATE(DPPCLK_CTRL, DPPCLK0_EN, src);
+		break;
+	case 1:
+		REG_UPDATE(DPPCLK_CTRL, DPPCLK1_EN, src);
+		break;
+	case 2:
+		REG_UPDATE(DPPCLK_CTRL, DPPCLK2_EN, src);
+		break;
+	case 3:
+		REG_UPDATE(DPPCLK_CTRL, DPPCLK3_EN, src);
+		break;
+	default:
+	BREAK_TO_DEBUGGER();
+		break;
+	}
+}
+
+static void dccg35_set_dtbclk_p_src_new(
+	struct dccg *dccg,
+	enum dtbclk_source src,
+	int inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	/* If DTBCLK_P#_EN is 0 refclock is selected as functional clock
+	 * If DTBCLK_P#_EN is 1 functional clock is selected as DTBCLK_P#_SRC_SEL
+	 */
+
+	switch (inst) {
+	case 0:
+		REG_UPDATE_2(DTBCLK_P_CNTL,
+					 DTBCLK_P0_SRC_SEL, (src == DTBCLK_REFCLK) ? 0 : src,
+					 DTBCLK_P0_EN, (src == DTBCLK_REFCLK) ? 0 : 1);
+		break;
+	case 1:
+		REG_UPDATE_2(DTBCLK_P_CNTL,
+					 DTBCLK_P1_SRC_SEL, (src == DTBCLK_REFCLK) ? 0 : src,
+					 DTBCLK_P1_EN, (src == DTBCLK_REFCLK) ? 0 : 1);
+		break;
+	case 2:
+		REG_UPDATE_2(DTBCLK_P_CNTL,
+					 DTBCLK_P2_SRC_SEL, (src == DTBCLK_REFCLK) ? 0 : src,
+					 DTBCLK_P2_EN, (src == DTBCLK_REFCLK) ? 0 : 1);
+		break;
+	case 3:
+		REG_UPDATE_2(DTBCLK_P_CNTL,
+					 DTBCLK_P3_SRC_SEL, (src == DTBCLK_REFCLK) ? 0 : src,
+					 DTBCLK_P3_EN, (src == DTBCLK_REFCLK) ? 0 : 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+static void dccg35_set_dpstreamclk_src_new(
+	struct dccg *dccg,
+	enum dp_stream_clk_source src,
+	int inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	switch (inst) {
+	case 0:
+		REG_UPDATE_2(DPSTREAMCLK_CNTL, DPSTREAMCLK0_EN,
+					 (src == DP_STREAM_REFCLK) ? 0 : 1,
+					 DPSTREAMCLK0_SRC_SEL,
+					 (src == DP_STREAM_REFCLK) ? 0 : src);
+		break;
+	case 1:
+		REG_UPDATE_2(DPSTREAMCLK_CNTL, DPSTREAMCLK1_EN,
+					 (src == DP_STREAM_REFCLK) ? 0 : 1,
+					 DPSTREAMCLK1_SRC_SEL,
+					 (src == DP_STREAM_REFCLK) ? 0 : src);
+
+		break;
+	case 2:
+		REG_UPDATE_2(DPSTREAMCLK_CNTL, DPSTREAMCLK2_EN,
+					 (src == DP_STREAM_REFCLK) ? 0 : 1,
+					 DPSTREAMCLK2_SRC_SEL,
+					 (src == DP_STREAM_REFCLK) ? 0 : src);
+
+		break;
+	case 3:
+		REG_UPDATE_2(DPSTREAMCLK_CNTL, DPSTREAMCLK3_EN,
+					 (src == DP_STREAM_REFCLK) ? 0 : 1,
+					 DPSTREAMCLK3_SRC_SEL,
+					 (src == DP_STREAM_REFCLK) ? 0 : src);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+static void dccg35_set_physymclk_src_new(
+	struct dccg *dccg,
+	enum physymclk_source src,
+	int inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	switch (inst) {
+	case 0:
+		REG_UPDATE_2(PHYASYMCLK_CLOCK_CNTL, PHYASYMCLK_EN,
+					 (src == PHYSYMCLK_REFCLK) ? 0 : 1,
+					 PHYASYMCLK_SRC_SEL,
+					 (src == PHYSYMCLK_REFCLK) ? 0 : src);
+		break;
+	case 1:
+		REG_UPDATE_2(PHYBSYMCLK_CLOCK_CNTL, PHYBSYMCLK_EN,
+					 (src == PHYSYMCLK_REFCLK) ? 0 : 1,
+					 PHYBSYMCLK_SRC_SEL,
+					 (src == PHYSYMCLK_REFCLK) ? 0 : src);
+		break;
+	case 2:
+		REG_UPDATE_2(PHYCSYMCLK_CLOCK_CNTL, PHYCSYMCLK_EN,
+					 (src == PHYSYMCLK_REFCLK) ? 0 : 1,
+					 PHYCSYMCLK_SRC_SEL,
+					 (src == PHYSYMCLK_REFCLK) ? 0 : src);
+		break;
+	case 3:
+		REG_UPDATE_2(PHYDSYMCLK_CLOCK_CNTL, PHYDSYMCLK_EN,
+					 (src == PHYSYMCLK_REFCLK) ? 0 : 1,
+					 PHYDSYMCLK_SRC_SEL,
+					 (src == PHYSYMCLK_REFCLK) ? 0 : src);
+		break;
+	case 4:
+		REG_UPDATE_2(PHYESYMCLK_CLOCK_CNTL, PHYESYMCLK_EN,
+					 (src == PHYSYMCLK_REFCLK) ? 0 : 1,
+					 PHYESYMCLK_SRC_SEL,
+					 (src == PHYSYMCLK_REFCLK) ? 0 : src);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+static void dccg35_set_symclk_be_src_new(
+	struct dccg *dccg,
+	enum symclk_be_source src,
+	int inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	switch (inst) {
+	case 0:
+		REG_UPDATE_2(SYMCLKA_CLOCK_ENABLE,
+					 SYMCLKA_CLOCK_ENABLE, (src == SYMCLK_BE_REFCLK) ? 0 : 1,
+					 SYMCLKA_SRC_SEL, (src == SYMCLK_BE_REFCLK) ? 0 : src);
+		break;
+	case 1:
+		REG_UPDATE_2(SYMCLKB_CLOCK_ENABLE,
+					 SYMCLKB_CLOCK_ENABLE, (src == SYMCLK_BE_REFCLK) ? 0 : 1,
+					 SYMCLKB_SRC_SEL, (src == SYMCLK_BE_REFCLK) ? 0 : src);
+		break;
+	case 2:
+		REG_UPDATE_2(SYMCLKC_CLOCK_ENABLE,
+					 SYMCLKC_CLOCK_ENABLE, (src == SYMCLK_BE_REFCLK) ? 0 : 1,
+					 SYMCLKC_SRC_SEL, (src == SYMCLK_BE_REFCLK) ? 0 : src);
+		break;
+	case 3:
+		REG_UPDATE_2(SYMCLKD_CLOCK_ENABLE,
+					 SYMCLKD_CLOCK_ENABLE, (src == SYMCLK_BE_REFCLK) ? 0 : 1,
+					 SYMCLKD_SRC_SEL, (src == SYMCLK_BE_REFCLK) ? 0 : src);
+		break;
+	case 4:
+		REG_UPDATE_2(SYMCLKE_CLOCK_ENABLE,
+					 SYMCLKE_CLOCK_ENABLE, (src == SYMCLK_BE_REFCLK) ? 0 : 1,
+					 SYMCLKE_SRC_SEL, (src == SYMCLK_BE_REFCLK) ? 0 : src);
+		break;
+	}
+}
+
+static int dccg35_is_symclk_fe_src_functional_be(struct dccg *dccg,
+												 int symclk_fe_inst,
+												 int symclk_be_inst)
+{
+
+	uint32_t en = 0;
+	uint32_t src_sel = 0;
+
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	switch (symclk_fe_inst) {
+	case 0:
+		REG_GET_2(SYMCLKA_CLOCK_ENABLE, SYMCLKA_FE_SRC_SEL, &src_sel, SYMCLKA_FE_EN, &en);
+		break;
+	case 1:
+		REG_GET_2(SYMCLKB_CLOCK_ENABLE, SYMCLKB_FE_SRC_SEL, &src_sel, SYMCLKB_FE_EN, &en);
+		break;
+	case 2:
+		REG_GET_2(SYMCLKC_CLOCK_ENABLE, SYMCLKC_FE_SRC_SEL, &src_sel, SYMCLKC_FE_EN, &en);
+		break;
+	case 3:
+		REG_GET_2(SYMCLKD_CLOCK_ENABLE, SYMCLKD_FE_SRC_SEL, &src_sel, SYMCLKD_FE_EN, &en);
+		break;
+	case 4:
+		REG_GET_2(SYMCLKE_CLOCK_ENABLE, SYMCLKE_FE_SRC_SEL, &src_sel, SYMCLKE_FE_EN, &en);
+		break;
+	}
+
+	if (en == 1 && src_sel == symclk_be_inst)
+		return 1;
+
+	return 0;
+}
+
+static void dccg35_set_symclk_fe_src_new(struct dccg *dccg, enum symclk_fe_source src, int inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	switch (inst) {
+	case 0:
+		REG_UPDATE_2(SYMCLKA_CLOCK_ENABLE,
+					 SYMCLKA_FE_EN, (src == SYMCLK_FE_REFCLK) ? 0 : 1,
+					 SYMCLKA_FE_SRC_SEL, (src == SYMCLK_FE_REFCLK) ? 0 : src);
+		break;
+	case 1:
+		REG_UPDATE_2(SYMCLKB_CLOCK_ENABLE,
+					 SYMCLKB_FE_EN, (src == SYMCLK_FE_REFCLK) ? 0 : 1,
+					 SYMCLKB_FE_SRC_SEL, (src == SYMCLK_FE_REFCLK) ? 0 : src);
+		break;
+	case 2:
+		REG_UPDATE_2(SYMCLKC_CLOCK_ENABLE,
+					 SYMCLKC_FE_EN, (src == SYMCLK_FE_REFCLK) ? 0 : 1,
+					 SYMCLKC_FE_SRC_SEL, (src == SYMCLK_FE_REFCLK) ? 0 : src);
+		break;
+	case 3:
+		REG_UPDATE_2(SYMCLKD_CLOCK_ENABLE,
+					 SYMCLKD_FE_EN, (src == SYMCLK_FE_REFCLK) ? 0 : 1,
+					 SYMCLKD_FE_SRC_SEL, (src == SYMCLK_FE_REFCLK) ? 0 : src);
+		break;
+	case 4:
+		REG_UPDATE_2(SYMCLKE_CLOCK_ENABLE,
+					 SYMCLKE_FE_EN, (src == SYMCLK_FE_REFCLK) ? 0 : 1,
+					 SYMCLKE_FE_SRC_SEL, (src == SYMCLK_FE_REFCLK) ? 0 : src);
+		break;
+	}
+}
+
+static uint32_t dccg35_is_fe_rcg(struct dccg *dccg, int inst)
+{
+	uint32_t enable = 0;
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	switch (inst) {
+	case 0:
+		REG_GET(DCCG_GATE_DISABLE_CNTL5,
+				SYMCLKA_FE_ROOT_GATE_DISABLE, &enable);
+		break;
+	case 1:
+		REG_GET(DCCG_GATE_DISABLE_CNTL5,
+				SYMCLKB_FE_ROOT_GATE_DISABLE, &enable);
+		break;
+	case 2:
+		REG_GET(DCCG_GATE_DISABLE_CNTL5,
+				SYMCLKC_FE_ROOT_GATE_DISABLE, &enable);
+		break;
+	case 3:
+		REG_GET(DCCG_GATE_DISABLE_CNTL5,
+				SYMCLKD_FE_ROOT_GATE_DISABLE, &enable);
+		break;
+	case 4:
+		REG_GET(DCCG_GATE_DISABLE_CNTL5,
+				SYMCLKE_FE_ROOT_GATE_DISABLE, &enable);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		break;
+	}
+	return enable;
+}
+
+static uint32_t dccg35_is_symclk32_se_rcg(struct dccg *dccg, int inst)
+{
+	uint32_t disable_l1 = 0;
+	uint32_t disable_l2 = 0;
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	switch (inst) {
+	case 0:
+		REG_GET_2(DCCG_GATE_DISABLE_CNTL3,
+				  SYMCLK32_SE0_GATE_DISABLE, &disable_l1,
+				  SYMCLK32_ROOT_SE0_GATE_DISABLE, &disable_l2);
+		break;
+	case 1:
+		REG_GET_2(DCCG_GATE_DISABLE_CNTL3,
+				  SYMCLK32_SE1_GATE_DISABLE, &disable_l1,
+				  SYMCLK32_ROOT_SE1_GATE_DISABLE, &disable_l2);
+		break;
+	case 2:
+		REG_GET_2(DCCG_GATE_DISABLE_CNTL3,
+				  SYMCLK32_SE2_GATE_DISABLE, &disable_l1,
+				  SYMCLK32_ROOT_SE2_GATE_DISABLE, &disable_l2);
+		break;
+	case 3:
+		REG_GET_2(DCCG_GATE_DISABLE_CNTL3,
+				  SYMCLK32_SE3_GATE_DISABLE, &disable_l1,
+				  SYMCLK32_ROOT_SE3_GATE_DISABLE, &disable_l2);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return 0;
+	}
+
+	/* return true if either block level or DCCG level gating is active */
+	return (disable_l1 | disable_l2);
+}
+
+static void dccg35_enable_symclk_fe_new(
+	struct dccg *dccg,
+	int inst,
+	enum symclk_fe_source src)
+{
+	dccg35_set_symclk_fe_rcg(dccg, inst, false);
+	dccg35_set_symclk_fe_src_new(dccg, src, inst);
+}
+
+static void dccg35_disable_symclk_fe_new(
+	struct dccg *dccg,
+	int inst)
+{
+	dccg35_set_symclk_fe_src_new(dccg, SYMCLK_FE_REFCLK, inst);
+	dccg35_set_symclk_fe_rcg(dccg, inst, true);
+}
+
+static void dccg35_enable_symclk_be_new(
+	struct dccg *dccg,
+	int inst,
+	enum symclk_be_source src)
+{
+	dccg35_set_symclk_be_rcg(dccg, inst, false);
+	dccg35_set_symclk_be_src_new(dccg, inst, src);
+}
+
+static void dccg35_disable_symclk_be_new(
+	struct dccg *dccg,
+	int inst)
+{
+	int i;
+
+	/* Switch from functional clock to refclock */
+	dccg35_set_symclk_be_src_new(dccg, inst, SYMCLK_BE_REFCLK);
+
+	/* Check if any other SE connected LE and disable them */
+	for (i = 0; i < 4; i++) {
+		/* Make sure FE is not already in RCG */
+		if (dccg35_is_fe_rcg(dccg, i) == 0) {
+			if (dccg35_is_symclk_fe_src_functional_be(dccg, i, inst))
+				dccg35_disable_symclk_fe_new(dccg, i);
+		}
+	}
+	/* Safe to RCG SYMCLK*/
+	dccg35_set_symclk_be_rcg(dccg, inst, true);
+}
+
+static void dccg35_enable_symclk32_se_new(
+	struct dccg *dccg,
+	int inst,
+	enum symclk32_se_clk_source src)
+{
+	dccg35_set_symclk32_se_rcg(dccg, inst, false);
+	dccg35_set_symclk32_se_src_new(dccg, inst, src);
+}
+
+static void dccg35_disable_symclk32_se_new(
+	struct dccg *dccg,
+	int inst)
+{
+	dccg35_set_symclk32_se_src_new(dccg, SYMCLK32_SE_REFCLK, inst);
+	dccg35_set_symclk32_se_rcg(dccg, inst, true);
+}
+
+static void dccg35_enable_symclk32_le_new(
+	struct dccg *dccg,
+	int inst,
+	enum symclk32_le_clk_source src)
+{
+	dccg35_set_symclk32_le_rcg(dccg, inst, false);
+	dccg35_set_symclk32_le_src_new(dccg, inst, src);
+}
+
+static void dccg35_disable_symclk32_le_new(
+	struct dccg *dccg,
+	int inst)
+{
+	int i;
+
+	/* Switch from functional clock to refclock */
+	dccg35_set_symclk32_le_src_new(dccg, inst, SYMCLK32_LE_REFCLK);
+
+	/* Check if any SE are connected and disable SE as well */
+	for (i = 0; i < 4; i++) {
+		/* Make sure FE is not already in RCG */
+		if (dccg35_is_symclk32_se_rcg(dccg, i) == 0) {
+			/* Disable and SE connected to this LE before RCG */
+			if (dccg35_is_symclk32_se_src_functional_le_new(dccg, i, inst))
+				dccg35_disable_symclk32_se_new(dccg, i);
+		}
+	}
+	/* Safe to RCG SYM32_LE*/
+	dccg35_set_symclk32_le_rcg(dccg, inst, true);
+}
+
+static void dccg35_enable_physymclk_new(struct dccg *dccg,
+					int inst,
+					enum physymclk_source src)
+{
+	dccg35_set_physymclk_rcg(dccg, inst, false);
+	dccg35_set_physymclk_src_new(dccg, src, inst);
+}
+
+static void dccg35_disable_physymclk_new(struct dccg *dccg,
+										 int inst)
+{
+	dccg35_set_physymclk_src_new(dccg, PHYSYMCLK_REFCLK, inst);
+	dccg35_set_physymclk_rcg(dccg, inst, true);
+}
+
+static void dccg35_enable_dpp_clk_new(
+	struct dccg *dccg,
+	int inst,
+	enum dppclk_clock_source src)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+	/* Sanitize inst before use in array de-ref */
+	if (inst < 0) {
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+	dccg35_set_dppclk_rcg(dccg, inst, false);
+	dcn35_set_dppclk_src_new(dccg, inst, src);
+	/* Switch DPP clock to DTO */
+	REG_SET_2(DPPCLK_DTO_PARAM[inst], 0,
+			  DPPCLK0_DTO_PHASE, 0xFF,
+			  DPPCLK0_DTO_MODULO, 0xFF);
+}
+
+
+static void dccg35_disable_dpp_clk_new(
+	struct dccg *dccg,
+	int inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+	/* Sanitize inst before use in array de-ref */
+	if (inst < 0) {
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+	dcn35_set_dppclk_src_new(dccg, inst, DPP_REFCLK);
+	REG_SET_2(DPPCLK_DTO_PARAM[inst], 0,
+			  DPPCLK0_DTO_PHASE, 0,
+			  DPPCLK0_DTO_MODULO, 1);
+	dccg35_set_dppclk_rcg(dccg, inst, true);
+}
+
+static void dccg35_disable_dscclk_new(struct dccg *dccg,
+									  int inst)
+{
+	dccg35_set_dsc_clk_src_new(dccg, inst, DSC_CLK_REF_CLK);
+	dccg35_set_dsc_clk_rcg(dccg, inst, true);
+}
+
+static void dccg35_enable_dscclk_new(struct dccg *dccg,
+									 int inst,
+									 enum dsc_clk_source src)
+{
+	dccg35_set_dsc_clk_rcg(dccg, inst, false);
+	dccg35_set_dsc_clk_src_new(dccg, inst, src);
+}
+
+static void dccg35_enable_dtbclk_p_new(struct dccg *dccg,
+									   enum dtbclk_source src,
+									   int inst)
+{
+	dccg35_set_dtbclk_p_rcg(dccg, inst, false);
+	dccg35_set_dtbclk_p_src_new(dccg, src, inst);
+}
+
+static void dccg35_disable_dtbclk_p_new(struct dccg *dccg,
+										int inst)
+{
+	dccg35_set_dtbclk_p_src_new(dccg, DTBCLK_REFCLK, inst);
+	dccg35_set_dtbclk_p_rcg(dccg, inst, true);
+}
+
+static void dccg35_disable_dpstreamclk_new(struct dccg *dccg,
+										  int inst)
+{
+	dccg35_set_dpstreamclk_src_new(dccg, DP_STREAM_REFCLK, inst);
+	dccg35_set_dpstreamclk_rcg(dccg, inst, true);
+}
+
+static void dccg35_enable_dpstreamclk_new(struct dccg *dccg,
+										   enum dp_stream_clk_source src,
+										   int inst)
+{
+	dccg35_set_dpstreamclk_rcg(dccg, inst, false);
+	dccg35_set_dpstreamclk_src_new(dccg, src, inst);
+}
+
+static void dccg35_trigger_dio_fifo_resync(struct dccg *dccg)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+	uint32_t dispclk_rdivider_value = 0;
+
+	REG_GET(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_RDIVIDER, &dispclk_rdivider_value);
+	if (dispclk_rdivider_value != 0)
+		REG_UPDATE(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_WDIVIDER, dispclk_rdivider_value);
+}
+
+static void dcn35_set_dppclk_enable(struct dccg *dccg,
+				 uint32_t dpp_inst, uint32_t enable)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+
+	switch (dpp_inst) {
+	case 0:
+		REG_UPDATE(DPPCLK_CTRL, DPPCLK0_EN, enable);
+		break;
+	case 1:
+		REG_UPDATE(DPPCLK_CTRL, DPPCLK1_EN, enable);
+		break;
+	case 2:
+		REG_UPDATE(DPPCLK_CTRL, DPPCLK2_EN, enable);
+		break;
+	case 3:
+		REG_UPDATE(DPPCLK_CTRL, DPPCLK3_EN, enable);
+		break;
+	default:
+		break;
+	}
+	DC_LOG_DEBUG("%s: dpp_inst(%d) DPPCLK_EN = %d\n", __func__, dpp_inst, enable);
+
+}
+
+static void dccg35_update_dpp_dto(struct dccg *dccg, int dpp_inst,
+				  int req_dppclk)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (dccg->dpp_clock_gated[dpp_inst]) {
+		/*
+		 * Do not update the DPPCLK DTO if the clock is stopped.
+		 */
+		return;
+	}
+
+	if (dccg->ref_dppclk && req_dppclk) {
+		int ref_dppclk = dccg->ref_dppclk;
+		int modulo, phase;
+
+		// phase / modulo = dpp pipe clk / dpp global clk
+		modulo = 0xff;   // use FF at the end
+		phase = ((modulo * req_dppclk) + ref_dppclk - 1) / ref_dppclk;
+
+		if (phase > 0xff) {
+			ASSERT(false);
+			phase = 0xff;
+		}
+		dccg35_set_dppclk_rcg(dccg, dpp_inst, false);
+
+		REG_SET_2(DPPCLK_DTO_PARAM[dpp_inst], 0,
+				DPPCLK0_DTO_PHASE, phase,
+				DPPCLK0_DTO_MODULO, modulo);
+
+		dcn35_set_dppclk_enable(dccg, dpp_inst, true);
+	} else {
+		dcn35_set_dppclk_enable(dccg, dpp_inst, false);
+		/*we have this in hwss: disable_plane*/
+		//dccg35_set_dppclk_rcg(dccg, dpp_inst, true);
+	}
+	dccg->pipe_dppclk_khz[dpp_inst] = req_dppclk;
+}
+
+static void dccg35_set_dppclk_root_clock_gating(struct dccg *dccg,
+		 uint32_t dpp_inst, uint32_t disallow_rcg)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (!dccg->ctx->dc->debug.root_clock_optimization.bits.dpp && !disallow_rcg)
+		return;
+
+
+	switch (dpp_inst) {
+	case 0:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL6, DPPCLK0_ROOT_GATE_DISABLE, disallow_rcg);
+		break;
+	case 1:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL6, DPPCLK1_ROOT_GATE_DISABLE, disallow_rcg);
+		break;
+	case 2:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL6, DPPCLK2_ROOT_GATE_DISABLE, disallow_rcg);
+		break;
+	case 3:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL6, DPPCLK3_ROOT_GATE_DISABLE, disallow_rcg);
+		break;
+	default:
+		break;
+	}
+
+	/* Wait for clock to ramp */
+	if (disallow_rcg)
+		udelay(10);
+}
+
+static void dccg35_get_pixel_rate_div(
+		struct dccg *dccg,
+		uint32_t otg_inst,
+		uint32_t *k1,
+		uint32_t *k2)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+	uint32_t val_k1 = PIXEL_RATE_DIV_NA, val_k2 = PIXEL_RATE_DIV_NA;
+
+	*k1 = PIXEL_RATE_DIV_NA;
+	*k2 = PIXEL_RATE_DIV_NA;
+
+	switch (otg_inst) {
+	case 0:
+		REG_GET_2(OTG_PIXEL_RATE_DIV,
+			OTG0_PIXEL_RATE_DIVK1, &val_k1,
+			OTG0_PIXEL_RATE_DIVK2, &val_k2);
+		break;
+	case 1:
+		REG_GET_2(OTG_PIXEL_RATE_DIV,
+			OTG1_PIXEL_RATE_DIVK1, &val_k1,
+			OTG1_PIXEL_RATE_DIVK2, &val_k2);
+		break;
+	case 2:
+		REG_GET_2(OTG_PIXEL_RATE_DIV,
+			OTG2_PIXEL_RATE_DIVK1, &val_k1,
+			OTG2_PIXEL_RATE_DIVK2, &val_k2);
+		break;
+	case 3:
+		REG_GET_2(OTG_PIXEL_RATE_DIV,
+			OTG3_PIXEL_RATE_DIVK1, &val_k1,
+			OTG3_PIXEL_RATE_DIVK2, &val_k2);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+	*k1 = val_k1;
+	*k2 = val_k2;
+}
+
+static void dccg35_set_pixel_rate_div(
+		struct dccg *dccg,
+		uint32_t otg_inst,
+		enum pixel_rate_div k1,
+		enum pixel_rate_div k2)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+	uint32_t cur_k1 = PIXEL_RATE_DIV_NA;
+	uint32_t cur_k2 = PIXEL_RATE_DIV_NA;
+
+
+	// Don't program 0xF into the register field. Not valid since
+	// K1 / K2 field is only 1 / 2 bits wide
+	if (k1 == PIXEL_RATE_DIV_NA || k2 == PIXEL_RATE_DIV_NA) {
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+	dccg35_get_pixel_rate_div(dccg, otg_inst, &cur_k1, &cur_k2);
+	if (k1 == cur_k1 && k2 == cur_k2)
+		return;
+
+	switch (otg_inst) {
+	case 0:
+		REG_UPDATE_2(OTG_PIXEL_RATE_DIV,
+				OTG0_PIXEL_RATE_DIVK1, k1,
+				OTG0_PIXEL_RATE_DIVK2, k2);
+		break;
+	case 1:
+		REG_UPDATE_2(OTG_PIXEL_RATE_DIV,
+				OTG1_PIXEL_RATE_DIVK1, k1,
+				OTG1_PIXEL_RATE_DIVK2, k2);
+		break;
+	case 2:
+		REG_UPDATE_2(OTG_PIXEL_RATE_DIV,
+				OTG2_PIXEL_RATE_DIVK1, k1,
+				OTG2_PIXEL_RATE_DIVK2, k2);
+		break;
+	case 3:
+		REG_UPDATE_2(OTG_PIXEL_RATE_DIV,
+				OTG3_PIXEL_RATE_DIVK1, k1,
+				OTG3_PIXEL_RATE_DIVK2, k2);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+static void dccg35_set_dtbclk_p_src(
+		struct dccg *dccg,
+		enum streamclk_source src,
+		uint32_t otg_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	uint32_t p_src_sel = 0; /* selects dprefclk */
+	if (src == DTBCLK0)
+		p_src_sel = 2;  /* selects dtbclk0 */
+
+	switch (otg_inst) {
+	case 0:
+		if (src == REFCLK)
+			REG_UPDATE(DTBCLK_P_CNTL,
+					DTBCLK_P0_EN, 0);
+		else
+			REG_UPDATE_2(DTBCLK_P_CNTL,
+					DTBCLK_P0_SRC_SEL, p_src_sel,
+					DTBCLK_P0_EN, 1);
+		break;
+	case 1:
+		if (src == REFCLK)
+			REG_UPDATE(DTBCLK_P_CNTL,
+					DTBCLK_P1_EN, 0);
+		else
+			REG_UPDATE_2(DTBCLK_P_CNTL,
+					DTBCLK_P1_SRC_SEL, p_src_sel,
+					DTBCLK_P1_EN, 1);
+		break;
+	case 2:
+		if (src == REFCLK)
+			REG_UPDATE(DTBCLK_P_CNTL,
+					DTBCLK_P2_EN, 0);
+		else
+			REG_UPDATE_2(DTBCLK_P_CNTL,
+					DTBCLK_P2_SRC_SEL, p_src_sel,
+					DTBCLK_P2_EN, 1);
+		break;
+	case 3:
+		if (src == REFCLK)
+			REG_UPDATE(DTBCLK_P_CNTL,
+					DTBCLK_P3_EN, 0);
+		else
+			REG_UPDATE_2(DTBCLK_P_CNTL,
+					DTBCLK_P3_SRC_SEL, p_src_sel,
+					DTBCLK_P3_EN, 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+}
+
+/* Controls the generation of pixel valid for OTG in (OTG -> HPO case) */
+static void dccg35_set_dtbclk_dto(
+		struct dccg *dccg,
+		const struct dtbclk_dto_params *params)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+	/* DTO Output Rate / Pixel Rate = 1/4 */
+	int req_dtbclk_khz = params->pixclk_khz / 4;
+
+	if (params->ref_dtbclk_khz && req_dtbclk_khz) {
+		uint32_t modulo, phase;
+
+		switch (params->otg_inst) {
+		case 0:
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P0_GATE_DISABLE, 1);
+			break;
+		case 1:
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P1_GATE_DISABLE, 1);
+			break;
+		case 2:
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P2_GATE_DISABLE, 1);
+			break;
+		case 3:
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P3_GATE_DISABLE, 1);
+			break;
+		}
+
+		// phase / modulo = dtbclk / dtbclk ref
+		modulo = params->ref_dtbclk_khz * 1000;
+		phase = req_dtbclk_khz * 1000;
+
+		REG_WRITE(DTBCLK_DTO_MODULO[params->otg_inst], modulo);
+		REG_WRITE(DTBCLK_DTO_PHASE[params->otg_inst], phase);
+
+		REG_UPDATE(OTG_PIXEL_RATE_CNTL[params->otg_inst],
+				DTBCLK_DTO_ENABLE[params->otg_inst], 1);
+
+		REG_WAIT(OTG_PIXEL_RATE_CNTL[params->otg_inst],
+				DTBCLKDTO_ENABLE_STATUS[params->otg_inst], 1,
+				1, 100);
+
+		/* program OTG_PIXEL_RATE_DIV for DIVK1 and DIVK2 fields */
+		dccg35_set_pixel_rate_div(dccg, params->otg_inst, PIXEL_RATE_DIV_BY_1, PIXEL_RATE_DIV_BY_1);
+
+		/* The recommended programming sequence to enable DTBCLK DTO to generate
+		 * valid pixel HPO DPSTREAM ENCODER, specifies that DTO source select should
+		 * be set only after DTO is enabled.
+		 * PIPEx_DTO_SRC_SEL should not be programmed during DTBCLK update since OTG may still be on, and the
+		 * programming is handled in program_pix_clk() regardless, so it can be removed from here.
+		 */
+		DC_LOG_DEBUG("%s: OTG%d DTBCLK DTO enabled: pixclk_khz=%d, ref_dtbclk_khz=%d, req_dtbclk_khz=%d, phase=%d, modulo=%d\n",
+				__func__, params->otg_inst, params->pixclk_khz,
+				params->ref_dtbclk_khz, req_dtbclk_khz, phase, modulo);
+
+	} else {
+		switch (params->otg_inst) {
+		case 0:
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P0_GATE_DISABLE, 0);
+			break;
+		case 1:
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P1_GATE_DISABLE, 0);
+			break;
+		case 2:
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P2_GATE_DISABLE, 0);
+			break;
+		case 3:
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P3_GATE_DISABLE, 0);
+			break;
+		}
+
+		/**
+		 * PIPEx_DTO_SRC_SEL should not be programmed during DTBCLK update since OTG may still be on, and the
+		 * programming is handled in program_pix_clk() regardless, so it can be removed from here.
+		 */
+		REG_UPDATE(OTG_PIXEL_RATE_CNTL[params->otg_inst],
+				DTBCLK_DTO_ENABLE[params->otg_inst], 0);
+
+		REG_WRITE(DTBCLK_DTO_MODULO[params->otg_inst], 0);
+		REG_WRITE(DTBCLK_DTO_PHASE[params->otg_inst], 0);
+
+		DC_LOG_DEBUG("%s: OTG%d DTBCLK DTO disabled\n", __func__, params->otg_inst);
+	}
+}
+
+static void dccg35_set_dpstreamclk(
+		struct dccg *dccg,
+		enum streamclk_source src,
+		int otg_inst,
+		int dp_hpo_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	/* set the dtbclk_p source */
+	dccg35_set_dtbclk_p_src(dccg, src, otg_inst);
+
+	/* enabled to select one of the DTBCLKs for pipe */
+	switch (dp_hpo_inst) {
+	case 0:
+		REG_UPDATE_2(DPSTREAMCLK_CNTL, DPSTREAMCLK0_EN,
+				(src == REFCLK) ? 0 : 1, DPSTREAMCLK0_SRC_SEL, otg_inst);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream)
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK0_ROOT_GATE_DISABLE, (src == REFCLK) ? 0 : 1);
+		break;
+	case 1:
+		REG_UPDATE_2(DPSTREAMCLK_CNTL, DPSTREAMCLK1_EN,
+				(src == REFCLK) ? 0 : 1, DPSTREAMCLK1_SRC_SEL, otg_inst);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream)
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK1_ROOT_GATE_DISABLE, (src == REFCLK) ? 0 : 1);
+		break;
+	case 2:
+		REG_UPDATE_2(DPSTREAMCLK_CNTL, DPSTREAMCLK2_EN,
+				(src == REFCLK) ? 0 : 1, DPSTREAMCLK2_SRC_SEL, otg_inst);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream)
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK2_ROOT_GATE_DISABLE, (src == REFCLK) ? 0 : 1);
+		break;
+	case 3:
+		REG_UPDATE_2(DPSTREAMCLK_CNTL, DPSTREAMCLK3_EN,
+				(src == REFCLK) ? 0 : 1, DPSTREAMCLK3_SRC_SEL, otg_inst);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream)
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK3_ROOT_GATE_DISABLE, (src == REFCLK) ? 0 : 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+	DC_LOG_DEBUG("%s: dp_hpo_inst(%d) DPSTREAMCLK_EN = %d, DPSTREAMCLK_SRC_SEL = %d\n",
+			__func__, dp_hpo_inst, (src == REFCLK) ? 0 : 1, otg_inst);
+}
+
+
+static void dccg35_set_dpstreamclk_root_clock_gating(
+		struct dccg *dccg,
+		int dp_hpo_inst,
+		bool enable)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	switch (dp_hpo_inst) {
+	case 0:
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream) {
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK0_ROOT_GATE_DISABLE, enable ? 1 : 0);
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK0_GATE_DISABLE, enable ? 1 : 0);
+		}
+		break;
+	case 1:
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream) {
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK1_ROOT_GATE_DISABLE, enable ? 1 : 0);
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK1_GATE_DISABLE, enable ? 1 : 0);
+		}
+		break;
+	case 2:
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream) {
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK2_ROOT_GATE_DISABLE, enable ? 1 : 0);
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK2_GATE_DISABLE, enable ? 1 : 0);
+		}
+		break;
+	case 3:
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream) {
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK3_ROOT_GATE_DISABLE, enable ? 1 : 0);
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK3_GATE_DISABLE, enable ? 1 : 0);
+		}
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+	DC_LOG_DEBUG("%s: dp_hpo_inst(%d) DPSTREAMCLK_ROOT_GATE_DISABLE = %d\n",
+			__func__, dp_hpo_inst, enable ? 1 : 0);
+}
+
+
+
+static void dccg35_set_physymclk_root_clock_gating(
+		struct dccg *dccg,
+		int phy_inst,
+		bool enable)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (!dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk)
+		return;
+
+	switch (phy_inst) {
+	case 0:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+				PHYASYMCLK_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 1:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+				PHYBSYMCLK_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 2:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+				PHYCSYMCLK_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 3:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+				PHYDSYMCLK_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	case 4:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+				PHYESYMCLK_ROOT_GATE_DISABLE, enable ? 0 : 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+	DC_LOG_DEBUG("%s: dpp_inst(%d) PHYESYMCLK_ROOT_GATE_DISABLE: %d\n", __func__, phy_inst, enable ? 0 : 1);
+
+}
+
+static void dccg35_set_physymclk(
+		struct dccg *dccg,
+		int phy_inst,
+		enum physymclk_clock_source clk_src,
+		bool force_enable)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	/* Force PHYSYMCLK on and Select phyd32clk as the source of clock which is output to PHY through DCIO */
+	switch (phy_inst) {
+	case 0:
+		if (force_enable) {
+			REG_UPDATE_2(PHYASYMCLK_CLOCK_CNTL,
+					PHYASYMCLK_EN, 1,
+					PHYASYMCLK_SRC_SEL, clk_src);
+		} else {
+			REG_UPDATE_2(PHYASYMCLK_CLOCK_CNTL,
+					PHYASYMCLK_EN, 0,
+					PHYASYMCLK_SRC_SEL, 0);
+		}
+		break;
+	case 1:
+		if (force_enable) {
+			REG_UPDATE_2(PHYBSYMCLK_CLOCK_CNTL,
+					PHYBSYMCLK_EN, 1,
+					PHYBSYMCLK_SRC_SEL, clk_src);
+		} else {
+			REG_UPDATE_2(PHYBSYMCLK_CLOCK_CNTL,
+					PHYBSYMCLK_EN, 0,
+					PHYBSYMCLK_SRC_SEL, 0);
+		}
+		break;
+	case 2:
+		if (force_enable) {
+			REG_UPDATE_2(PHYCSYMCLK_CLOCK_CNTL,
+					PHYCSYMCLK_EN, 1,
+					PHYCSYMCLK_SRC_SEL, clk_src);
+		} else {
+			REG_UPDATE_2(PHYCSYMCLK_CLOCK_CNTL,
+					PHYCSYMCLK_EN, 0,
+					PHYCSYMCLK_SRC_SEL, 0);
+		}
+		break;
+	case 3:
+		if (force_enable) {
+			REG_UPDATE_2(PHYDSYMCLK_CLOCK_CNTL,
+					PHYDSYMCLK_EN, 1,
+					PHYDSYMCLK_SRC_SEL, clk_src);
+		} else {
+			REG_UPDATE_2(PHYDSYMCLK_CLOCK_CNTL,
+					PHYDSYMCLK_EN, 0,
+					PHYDSYMCLK_SRC_SEL, 0);
+		}
+		break;
+	case 4:
+		if (force_enable) {
+			REG_UPDATE_2(PHYESYMCLK_CLOCK_CNTL,
+					PHYESYMCLK_EN, 1,
+					PHYESYMCLK_SRC_SEL, clk_src);
+		} else {
+			REG_UPDATE_2(PHYESYMCLK_CLOCK_CNTL,
+					PHYESYMCLK_EN, 0,
+					PHYESYMCLK_SRC_SEL, 0);
+		}
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+	DC_LOG_DEBUG("%s: phy_inst(%d) PHYxSYMCLK_EN = %d, PHYxSYMCLK_SRC_SEL = %d\n",
+			__func__, phy_inst, force_enable ? 1 : 0, clk_src);
+}
+
+static void dccg35_set_valid_pixel_rate(
+		struct dccg *dccg,
+		int ref_dtbclk_khz,
+		int otg_inst,
+		int pixclk_khz)
+{
+	struct dtbclk_dto_params dto_params = {0};
+
+	dto_params.ref_dtbclk_khz = ref_dtbclk_khz;
+	dto_params.otg_inst = otg_inst;
+	dto_params.pixclk_khz = pixclk_khz;
+	dto_params.is_hdmi = true;
+
+	dccg35_set_dtbclk_dto(dccg, &dto_params);
+}
+
+static void dccg35_dpp_root_clock_control(
+		struct dccg *dccg,
+		unsigned int dpp_inst,
+		bool clock_on)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (dccg->dpp_clock_gated[dpp_inst] == clock_on)
+		return;
+
+	if (clock_on) {
+		dccg35_set_dppclk_rcg(dccg, dpp_inst, false);
+
+		/* turn off the DTO and leave phase/modulo at max */
+		dcn35_set_dppclk_enable(dccg, dpp_inst, 1);
+		REG_SET_2(DPPCLK_DTO_PARAM[dpp_inst], 0,
+			  DPPCLK0_DTO_PHASE, 0xFF,
+			  DPPCLK0_DTO_MODULO, 0xFF);
+	} else {
+		dcn35_set_dppclk_enable(dccg, dpp_inst, 0);
+		/* turn on the DTO to generate a 0hz clock */
+		REG_SET_2(DPPCLK_DTO_PARAM[dpp_inst], 0,
+			  DPPCLK0_DTO_PHASE, 0,
+			  DPPCLK0_DTO_MODULO, 1);
+		/*we have this in hwss: disable_plane*/
+		//dccg35_set_dppclk_rcg(dccg, dpp_inst, true);
+	}
+
+	dccg->dpp_clock_gated[dpp_inst] = !clock_on;
+	DC_LOG_DEBUG("%s: dpp_inst(%d) clock_on = %d\n", __func__, dpp_inst, clock_on);
+}
+
+static void dccg35_disable_symclk32_se(
+		struct dccg *dccg,
+		int hpo_se_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	/* set refclk as the source for symclk32_se */
+	switch (hpo_se_inst) {
+	case 0:
+		REG_UPDATE_2(SYMCLK32_SE_CNTL,
+				SYMCLK32_SE0_SRC_SEL, 0,
+				SYMCLK32_SE0_EN, 0);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se) {
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_SE0_GATE_DISABLE, 0);
+//			REG_UPDATE(DCCG_GATE_DISABLE_CNTL3,
+//					SYMCLK32_ROOT_SE0_GATE_DISABLE, 0);
+		}
+		break;
+	case 1:
+		REG_UPDATE_2(SYMCLK32_SE_CNTL,
+				SYMCLK32_SE1_SRC_SEL, 0,
+				SYMCLK32_SE1_EN, 0);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se) {
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_SE1_GATE_DISABLE, 0);
+//			REG_UPDATE(DCCG_GATE_DISABLE_CNTL3,
+//					SYMCLK32_ROOT_SE1_GATE_DISABLE, 0);
+		}
+		break;
+	case 2:
+		REG_UPDATE_2(SYMCLK32_SE_CNTL,
+				SYMCLK32_SE2_SRC_SEL, 0,
+				SYMCLK32_SE2_EN, 0);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se) {
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_SE2_GATE_DISABLE, 0);
+//			REG_UPDATE(DCCG_GATE_DISABLE_CNTL3,
+//					SYMCLK32_ROOT_SE2_GATE_DISABLE, 0);
+		}
+		break;
+	case 3:
+		REG_UPDATE_2(SYMCLK32_SE_CNTL,
+				SYMCLK32_SE3_SRC_SEL, 0,
+				SYMCLK32_SE3_EN, 0);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se) {
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_SE3_GATE_DISABLE, 0);
+//			REG_UPDATE(DCCG_GATE_DISABLE_CNTL3,
+//					SYMCLK32_ROOT_SE3_GATE_DISABLE, 0);
+		}
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+}
+
+static void dccg35_init_cb(struct dccg *dccg)
+{
+	(void)dccg;
+	/* Any RCG should be done when driver enter low power mode*/
+}
+void dccg35_init(struct dccg *dccg)
+{
+	int otg_inst;
+	/* Set HPO stream encoder to use refclk to avoid case where PHY is
+	 * disabled and SYMCLK32 for HPO SE is sourced from PHYD32CLK which
+	 * will cause DCN to hang.
+	 */
+	for (otg_inst = 0; otg_inst < 4; otg_inst++)
+		dccg35_disable_symclk32_se(dccg, otg_inst);
+
+	if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_le)
+		for (otg_inst = 0; otg_inst < 2; otg_inst++) {
+			dccg31_disable_symclk32_le(dccg, otg_inst);
+			dccg31_set_symclk32_le_root_clock_gating(dccg, otg_inst, false);
+			DC_LOG_DEBUG("%s: OTG%d SYMCLK32_LE disabled and root clock gating disabled\n",
+					__func__, otg_inst);
+		}
+
+//	if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+//		for (otg_inst = 0; otg_inst < 4; otg_inst++)
+//			dccg35_disable_symclk_se(dccg, otg_inst, otg_inst);
+
+
+	if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream)
+		for (otg_inst = 0; otg_inst < 4; otg_inst++) {
+			dccg35_set_dpstreamclk(dccg, REFCLK, otg_inst,
+						otg_inst);
+			dccg35_set_dpstreamclk_root_clock_gating(dccg, otg_inst, false);
+			DC_LOG_DEBUG("%s: OTG%d DPSTREAMCLK disabled and root clock gating disabled\n",
+					__func__, otg_inst);
+		}
+
+/*
+	dccg35_enable_global_fgcg_rep(
+		dccg, dccg->ctx->dc->debug.enable_fine_grain_clock_gating.bits
+			      .dccg_global_fgcg_rep);*/
+}
+
+void dccg35_enable_global_fgcg_rep(struct dccg *dccg, bool value)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	REG_UPDATE(DCCG_GLOBAL_FGCG_REP_CNTL, DCCG_GLOBAL_FGCG_REP_DIS, !value);
+}
+
+static void dccg35_enable_dscclk(struct dccg *dccg, int inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	//Disable DTO
+	switch (inst) {
+	case 0:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL6, DSCCLK0_ROOT_GATE_DISABLE, 1);
+
+		REG_UPDATE_2(DSCCLK0_DTO_PARAM,
+				DSCCLK0_DTO_PHASE, 0,
+				DSCCLK0_DTO_MODULO, 0);
+		REG_UPDATE(DSCCLK_DTO_CTRL,	DSCCLK0_EN, 1);
+		break;
+	case 1:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL6, DSCCLK1_ROOT_GATE_DISABLE, 1);
+
+		REG_UPDATE_2(DSCCLK1_DTO_PARAM,
+				DSCCLK1_DTO_PHASE, 0,
+				DSCCLK1_DTO_MODULO, 0);
+		REG_UPDATE(DSCCLK_DTO_CTRL, DSCCLK1_EN, 1);
+		break;
+	case 2:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL6, DSCCLK2_ROOT_GATE_DISABLE, 1);
+
+		REG_UPDATE_2(DSCCLK2_DTO_PARAM,
+				DSCCLK2_DTO_PHASE, 0,
+				DSCCLK2_DTO_MODULO, 0);
+		REG_UPDATE(DSCCLK_DTO_CTRL, DSCCLK2_EN, 1);
+		break;
+	case 3:
+		REG_UPDATE(DCCG_GATE_DISABLE_CNTL6, DSCCLK3_ROOT_GATE_DISABLE, 1);
+
+		REG_UPDATE_2(DSCCLK3_DTO_PARAM,
+				DSCCLK3_DTO_PHASE, 0,
+				DSCCLK3_DTO_MODULO, 0);
+		REG_UPDATE(DSCCLK_DTO_CTRL, DSCCLK3_EN, 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+	/* Wait for clock to ramp */
+	udelay(10);
+}
+
+static void dccg35_disable_dscclk(struct dccg *dccg,
+				int inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	switch (inst) {
+	case 0:
+		REG_UPDATE(DSCCLK_DTO_CTRL, DSCCLK0_EN, 0);
+		REG_UPDATE_2(DSCCLK0_DTO_PARAM,
+				DSCCLK0_DTO_PHASE, 0,
+				DSCCLK0_DTO_MODULO, 1);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.dsc)
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL6, DSCCLK0_ROOT_GATE_DISABLE, 0);
+		break;
+	case 1:
+		REG_UPDATE(DSCCLK_DTO_CTRL, DSCCLK1_EN, 0);
+		REG_UPDATE_2(DSCCLK1_DTO_PARAM,
+				DSCCLK1_DTO_PHASE, 0,
+				DSCCLK1_DTO_MODULO, 1);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.dsc)
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL6, DSCCLK1_ROOT_GATE_DISABLE, 0);
+		break;
+	case 2:
+		REG_UPDATE(DSCCLK_DTO_CTRL, DSCCLK2_EN, 0);
+		REG_UPDATE_2(DSCCLK2_DTO_PARAM,
+				DSCCLK2_DTO_PHASE, 0,
+				DSCCLK2_DTO_MODULO, 1);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.dsc)
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL6, DSCCLK2_ROOT_GATE_DISABLE, 0);
+		break;
+	case 3:
+		REG_UPDATE(DSCCLK_DTO_CTRL, DSCCLK3_EN, 0);
+		REG_UPDATE_2(DSCCLK3_DTO_PARAM,
+				DSCCLK3_DTO_PHASE, 0,
+				DSCCLK3_DTO_MODULO, 1);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.dsc)
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL6, DSCCLK3_ROOT_GATE_DISABLE, 0);
+		break;
+	default:
+		return;
+	}
+
+	/* Wait for clock ramp */
+	udelay(10);
+}
+
+static void dccg35_enable_symclk_se(struct dccg *dccg, uint32_t stream_enc_inst, uint32_t link_enc_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	switch (link_enc_inst) {
+	case 0:
+		REG_UPDATE(SYMCLKA_CLOCK_ENABLE,
+				SYMCLKA_CLOCK_ENABLE, 1);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKA_ROOT_GATE_DISABLE, 1);
+		break;
+	case 1:
+		REG_UPDATE(SYMCLKB_CLOCK_ENABLE,
+				SYMCLKB_CLOCK_ENABLE, 1);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKB_ROOT_GATE_DISABLE, 1);
+		break;
+	case 2:
+		REG_UPDATE(SYMCLKC_CLOCK_ENABLE,
+				SYMCLKC_CLOCK_ENABLE, 1);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKC_ROOT_GATE_DISABLE, 1);
+		break;
+	case 3:
+		REG_UPDATE(SYMCLKD_CLOCK_ENABLE,
+				SYMCLKD_CLOCK_ENABLE, 1);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKD_ROOT_GATE_DISABLE, 1);
+		break;
+	case 4:
+		REG_UPDATE(SYMCLKE_CLOCK_ENABLE,
+				SYMCLKE_CLOCK_ENABLE, 1);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKE_ROOT_GATE_DISABLE, 1);
+		break;
+	}
+
+	switch (stream_enc_inst) {
+	case 0:
+		REG_UPDATE_2(SYMCLKA_CLOCK_ENABLE,
+				SYMCLKA_FE_EN, 1,
+				SYMCLKA_FE_SRC_SEL, link_enc_inst);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKA_FE_ROOT_GATE_DISABLE, 1);
+		break;
+	case 1:
+		REG_UPDATE_2(SYMCLKB_CLOCK_ENABLE,
+				SYMCLKB_FE_EN, 1,
+				SYMCLKB_FE_SRC_SEL, link_enc_inst);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKB_FE_ROOT_GATE_DISABLE, 1);
+		break;
+	case 2:
+		REG_UPDATE_2(SYMCLKC_CLOCK_ENABLE,
+				SYMCLKC_FE_EN, 1,
+				SYMCLKC_FE_SRC_SEL, link_enc_inst);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKC_FE_ROOT_GATE_DISABLE, 1);
+		break;
+	case 3:
+		REG_UPDATE_2(SYMCLKD_CLOCK_ENABLE,
+				SYMCLKD_FE_EN, 1,
+				SYMCLKD_FE_SRC_SEL, link_enc_inst);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKD_FE_ROOT_GATE_DISABLE, 1);
+		break;
+	case 4:
+		REG_UPDATE_2(SYMCLKE_CLOCK_ENABLE,
+				SYMCLKE_FE_EN, 1,
+				SYMCLKE_FE_SRC_SEL, link_enc_inst);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKE_FE_ROOT_GATE_DISABLE, 1);
+		break;
+	}
+}
+
+/*get other front end connected to this backend*/
+static uint8_t dccg35_get_number_enabled_symclk_fe_connected_to_be(struct dccg *dccg, uint32_t link_enc_inst)
+{
+	uint8_t num_enabled_symclk_fe = 0;
+	uint32_t fe_clk_en[5] = {0}, be_clk_sel[5] = {0};
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	REG_GET_2(SYMCLKA_CLOCK_ENABLE, SYMCLKA_FE_EN, &fe_clk_en[0],
+		SYMCLKA_FE_SRC_SEL, &be_clk_sel[0]);
+
+	REG_GET_2(SYMCLKB_CLOCK_ENABLE, SYMCLKB_FE_EN, &fe_clk_en[1],
+		SYMCLKB_FE_SRC_SEL, &be_clk_sel[1]);
+
+	REG_GET_2(SYMCLKC_CLOCK_ENABLE, SYMCLKC_FE_EN, &fe_clk_en[2],
+		SYMCLKC_FE_SRC_SEL, &be_clk_sel[2]);
+
+	REG_GET_2(SYMCLKD_CLOCK_ENABLE,	SYMCLKD_FE_EN, &fe_clk_en[3],
+		SYMCLKD_FE_SRC_SEL, &be_clk_sel[3]);
+
+	REG_GET_2(SYMCLKE_CLOCK_ENABLE,	SYMCLKE_FE_EN, &fe_clk_en[4],
+		SYMCLKE_FE_SRC_SEL, &be_clk_sel[4]);
+
+	uint8_t i;
+
+	for (i = 0; i < ARRAY_SIZE(fe_clk_en); i++) {
+		if (fe_clk_en[i] && be_clk_sel[i] == link_enc_inst)
+			num_enabled_symclk_fe++;
+	}
+	return num_enabled_symclk_fe;
+}
+
+static void dccg35_disable_symclk_se(struct dccg *dccg, uint32_t stream_enc_inst, uint32_t link_enc_inst)
+{
+	uint8_t num_enabled_symclk_fe = 0;
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	switch (stream_enc_inst) {
+	case 0:
+		REG_UPDATE_2(SYMCLKA_CLOCK_ENABLE,
+				SYMCLKA_FE_EN, 0,
+				SYMCLKA_FE_SRC_SEL, 0);
+//		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+//			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKA_FE_ROOT_GATE_DISABLE, 0);
+		break;
+	case 1:
+		REG_UPDATE_2(SYMCLKB_CLOCK_ENABLE,
+				SYMCLKB_FE_EN, 0,
+				SYMCLKB_FE_SRC_SEL, 0);
+//		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+//			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKB_FE_ROOT_GATE_DISABLE, 0);
+		break;
+	case 2:
+		REG_UPDATE_2(SYMCLKC_CLOCK_ENABLE,
+				SYMCLKC_FE_EN, 0,
+				SYMCLKC_FE_SRC_SEL, 0);
+//		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+//			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKC_FE_ROOT_GATE_DISABLE, 0);
+		break;
+	case 3:
+		REG_UPDATE_2(SYMCLKD_CLOCK_ENABLE,
+				SYMCLKD_FE_EN, 0,
+				SYMCLKD_FE_SRC_SEL, 0);
+//		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+//			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKD_FE_ROOT_GATE_DISABLE, 0);
+		break;
+	case 4:
+		REG_UPDATE_2(SYMCLKE_CLOCK_ENABLE,
+				SYMCLKE_FE_EN, 0,
+				SYMCLKE_FE_SRC_SEL, 0);
+//		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+//			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKE_FE_ROOT_GATE_DISABLE, 0);
+		break;
+	}
+
+	/*check other enabled symclk fe connected to this be */
+	num_enabled_symclk_fe = dccg35_get_number_enabled_symclk_fe_connected_to_be(dccg, link_enc_inst);
+	/*only turn off backend clk if other front end attached to this backend are all off,
+	 for mst, only turn off the backend if this is the last front end*/
+	if (num_enabled_symclk_fe == 0) {
+		switch (link_enc_inst) {
+		case 0:
+			REG_UPDATE(SYMCLKA_CLOCK_ENABLE,
+					SYMCLKA_CLOCK_ENABLE, 0);
+//			if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_le)
+//				REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKA_ROOT_GATE_DISABLE, 0);
+			break;
+		case 1:
+			REG_UPDATE(SYMCLKB_CLOCK_ENABLE,
+					SYMCLKB_CLOCK_ENABLE, 0);
+//			if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_le)
+//				REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKB_ROOT_GATE_DISABLE, 0);
+			break;
+		case 2:
+			REG_UPDATE(SYMCLKC_CLOCK_ENABLE,
+					SYMCLKC_CLOCK_ENABLE, 0);
+//			if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_le)
+//				REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKC_ROOT_GATE_DISABLE, 0);
+			break;
+		case 3:
+			REG_UPDATE(SYMCLKD_CLOCK_ENABLE,
+					SYMCLKD_CLOCK_ENABLE, 0);
+//			if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_le)
+//				REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKD_ROOT_GATE_DISABLE, 0);
+			break;
+		case 4:
+			REG_UPDATE(SYMCLKE_CLOCK_ENABLE,
+					SYMCLKE_CLOCK_ENABLE, 0);
+//			if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_le)
+//				REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKE_ROOT_GATE_DISABLE, 0);
+			break;
+		}
+	}
+}
+
+static void dccg35_set_dpstreamclk_cb(
+		struct dccg *dccg,
+		enum streamclk_source src,
+		int otg_inst,
+		int dp_hpo_inst)
+{
+
+	enum dtbclk_source dtb_clk_src;
+	enum dp_stream_clk_source dp_stream_clk_src;
+
+	switch (src) {
+	case REFCLK:
+		dtb_clk_src = DTBCLK_REFCLK;
+		dp_stream_clk_src = DP_STREAM_REFCLK;
+		break;
+	case DPREFCLK:
+		dtb_clk_src = DTBCLK_DPREFCLK;
+		dp_stream_clk_src = (enum dp_stream_clk_source)otg_inst;
+		break;
+	case DTBCLK0:
+		dtb_clk_src = DTBCLK_DTBCLK0;
+		dp_stream_clk_src = (enum dp_stream_clk_source)otg_inst;
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+	if (dtb_clk_src == DTBCLK_REFCLK &&
+		dp_stream_clk_src == DP_STREAM_REFCLK) {
+		dccg35_disable_dtbclk_p_new(dccg, otg_inst);
+		dccg35_disable_dpstreamclk_new(dccg, dp_hpo_inst);
+	} else {
+		dccg35_enable_dtbclk_p_new(dccg, dtb_clk_src, otg_inst);
+		dccg35_enable_dpstreamclk_new(dccg,
+										dp_stream_clk_src,
+										dp_hpo_inst);
+	}
+}
+
+static void dccg35_set_dpstreamclk_root_clock_gating_cb(
+	struct dccg *dccg,
+	int dp_hpo_inst,
+	bool power_on)
+{
+	/* power_on set indicates we need to ungate
+	 * Currently called from optimize_bandwidth and prepare_bandwidth calls
+	 * Since clock source is not passed restore to refclock on ungate
+	 * Instance 0 is implied here since only one streamclock resource
+	 * Redundant as gating when enabled is acheived through set_dpstreamclk
+	 */
+	if (power_on)
+		dccg35_enable_dpstreamclk_new(dccg,
+										DP_STREAM_REFCLK,
+										dp_hpo_inst);
+	else
+		dccg35_disable_dpstreamclk_new(dccg, dp_hpo_inst);
+}
+
+static void dccg35_update_dpp_dto_cb(struct dccg *dccg, int dpp_inst,
+				  int req_dppclk)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (dccg->dpp_clock_gated[dpp_inst]) {
+		/*
+		 * Do not update the DPPCLK DTO if the clock is stopped.
+		 */
+		return;
+	}
+
+	if (dccg->ref_dppclk && req_dppclk) {
+		int ref_dppclk = dccg->ref_dppclk;
+		int modulo, phase;
+
+		// phase / modulo = dpp pipe clk / dpp global clk
+		modulo = 0xff;   // use FF at the end
+		phase = ((modulo * req_dppclk) + ref_dppclk - 1) / ref_dppclk;
+
+		if (phase > 0xff) {
+			ASSERT(false);
+			phase = 0xff;
+		}
+
+		/* Enable DPP CLK DTO output */
+		dccg35_enable_dpp_clk_new(dccg, dpp_inst, DPP_DCCG_DTO);
+
+		/* Program DTO */
+		REG_SET_2(DPPCLK_DTO_PARAM[dpp_inst], 0,
+				DPPCLK0_DTO_PHASE, phase,
+				DPPCLK0_DTO_MODULO, modulo);
+	} else
+		dccg35_disable_dpp_clk_new(dccg, dpp_inst);
+
+	dccg->pipe_dppclk_khz[dpp_inst] = req_dppclk;
+}
+
+static void dccg35_dpp_root_clock_control_cb(
+	struct dccg *dccg,
+	unsigned int dpp_inst,
+	bool power_on)
+{
+	if (dccg->dpp_clock_gated[dpp_inst] == power_on)
+		return;
+	/* power_on set indicates we need to ungate
+	 * Currently called from optimize_bandwidth and prepare_bandwidth calls
+	 * Since clock source is not passed restore to refclock on ungate
+	 * Redundant as gating when enabled is acheived through update_dpp_dto
+	 */
+	dccg35_set_dppclk_rcg(dccg, dpp_inst, !power_on);
+
+	dccg->dpp_clock_gated[dpp_inst] = !power_on;
+}
+
+static void dccg35_enable_symclk32_se_cb(
+	struct dccg *dccg,
+	int inst,
+	enum phyd32clk_clock_source phyd32clk)
+{
+	dccg35_enable_symclk32_se_new(dccg, inst, (enum symclk32_se_clk_source)phyd32clk);
+}
+
+static void dccg35_disable_symclk32_se_cb(struct dccg *dccg, int inst)
+{
+	dccg35_disable_symclk32_se_new(dccg, inst);
+}
+
+static void dccg35_enable_symclk32_le_cb(
+			struct dccg *dccg,
+			int inst,
+			enum phyd32clk_clock_source src)
+{
+	dccg35_enable_symclk32_le_new(dccg, inst, (enum symclk32_le_clk_source) src);
+}
+
+static void dccg35_disable_symclk32_le_cb(struct dccg *dccg, int inst)
+{
+	dccg35_disable_symclk32_le_new(dccg, inst);
+}
+
+static void dccg35_set_symclk32_le_root_clock_gating_cb(
+	struct dccg *dccg,
+	int inst,
+	bool power_on)
+{
+	/* power_on set indicates we need to ungate
+	 * Currently called from optimize_bandwidth and prepare_bandwidth calls
+	 * Since clock source is not passed restore to refclock on ungate
+	 * Redundant as gating when enabled is acheived through disable_symclk32_le
+	 */
+	if (power_on)
+		dccg35_enable_symclk32_le_new(dccg, inst, SYMCLK32_LE_REFCLK);
+	else
+		dccg35_disable_symclk32_le_new(dccg, inst);
+}
+
+static void dccg35_set_physymclk_cb(
+	struct dccg *dccg,
+	int inst,
+	enum physymclk_clock_source clk_src,
+	bool force_enable)
+{
+	/* force_enable = 0 indicates we can switch to ref clock */
+	if (force_enable)
+		dccg35_enable_physymclk_new(dccg, inst, (enum physymclk_source)clk_src);
+	else
+		dccg35_disable_physymclk_new(dccg, inst);
+}
+
+static void dccg35_set_physymclk_root_clock_gating_cb(
+	struct dccg *dccg,
+	int inst,
+	bool power_on)
+{
+	/* Redundant RCG already done in disable_physymclk
+	 * power_on = 1 indicates we need to ungate
+	 */
+	if (power_on)
+		dccg35_enable_physymclk_new(dccg, inst, PHYSYMCLK_REFCLK);
+	else
+		dccg35_disable_physymclk_new(dccg, inst);
+}
+
+static void dccg35_set_symclk32_le_root_clock_gating(
+	struct dccg *dccg,
+	int inst,
+	bool power_on)
+{
+	/* power_on set indicates we need to ungate
+	 * Currently called from optimize_bandwidth and prepare_bandwidth calls
+	 * Since clock source is not passed restore to refclock on ungate
+	 * Redundant as gating when enabled is acheived through disable_symclk32_le
+	 */
+	if (power_on)
+		dccg35_enable_symclk32_le_new(dccg, inst, SYMCLK32_LE_REFCLK);
+	else
+		dccg35_disable_symclk32_le_new(dccg, inst);
+}
+
+static void dccg35_set_dtbclk_p_src_cb(
+		struct dccg *dccg,
+		enum streamclk_source src,
+		uint32_t inst)
+{
+	if (src == DTBCLK0)
+		dccg35_enable_dtbclk_p_new(dccg, DTBCLK_DTBCLK0, inst);
+	else
+		dccg35_disable_dtbclk_p_new(dccg, inst);
+}
+
+static void dccg35_set_dtbclk_dto_cb(
+		struct dccg *dccg,
+		const struct dtbclk_dto_params *params)
+{
+	/* set_dtbclk_p_src typ called earlier to switch to DTBCLK
+	 * if params->ref_dtbclk_khz and req_dtbclk_khz are 0 switch to ref-clock
+	 */
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+	/* DTO Output Rate / Pixel Rate = 1/4 */
+	int req_dtbclk_khz = params->pixclk_khz / 4;
+
+	if (params->ref_dtbclk_khz && req_dtbclk_khz) {
+		uint32_t modulo, phase;
+
+		dccg35_enable_dtbclk_p_new(dccg, DTBCLK_DTBCLK0, params->otg_inst);
+
+		// phase / modulo = dtbclk / dtbclk ref
+		modulo = params->ref_dtbclk_khz * 1000;
+		phase = req_dtbclk_khz * 1000;
+
+		REG_WRITE(DTBCLK_DTO_MODULO[params->otg_inst], modulo);
+		REG_WRITE(DTBCLK_DTO_PHASE[params->otg_inst], phase);
+
+		REG_UPDATE(OTG_PIXEL_RATE_CNTL[params->otg_inst],
+				DTBCLK_DTO_ENABLE[params->otg_inst], 1);
+
+		REG_WAIT(OTG_PIXEL_RATE_CNTL[params->otg_inst],
+				DTBCLKDTO_ENABLE_STATUS[params->otg_inst], 1,
+				1, 100);
+
+		/* program OTG_PIXEL_RATE_DIV for DIVK1 and DIVK2 fields */
+		dccg35_set_pixel_rate_div(dccg, params->otg_inst, PIXEL_RATE_DIV_BY_1, PIXEL_RATE_DIV_BY_1);
+
+		/* The recommended programming sequence to enable DTBCLK DTO to generate
+		 * valid pixel HPO DPSTREAM ENCODER, specifies that DTO source select should
+		 * be set only after DTO is enabled
+		 */
+		REG_UPDATE(OTG_PIXEL_RATE_CNTL[params->otg_inst],
+				PIPE_DTO_SRC_SEL[params->otg_inst], 2);
+	} else {
+		dccg35_disable_dtbclk_p_new(dccg, params->otg_inst);
+
+		REG_UPDATE_2(OTG_PIXEL_RATE_CNTL[params->otg_inst],
+					 DTBCLK_DTO_ENABLE[params->otg_inst], 0,
+					 PIPE_DTO_SRC_SEL[params->otg_inst], params->is_hdmi ? 0 : 1);
+
+		REG_WRITE(DTBCLK_DTO_MODULO[params->otg_inst], 0);
+		REG_WRITE(DTBCLK_DTO_PHASE[params->otg_inst], 0);
+	}
+}
+
+static void dccg35_disable_dscclk_cb(struct dccg *dccg,
+									 int inst)
+{
+	dccg35_disable_dscclk_new(dccg, inst);
+}
+
+static void dccg35_enable_dscclk_cb(struct dccg *dccg, int inst)
+{
+	dccg35_enable_dscclk_new(dccg, inst, DSC_DTO_TUNED_CK_GPU_DISCLK_3);
+}
+
+static void dccg35_enable_symclk_se_cb(struct dccg *dccg, uint32_t stream_enc_inst, uint32_t link_enc_inst)
+{
+	/* Switch to functional clock if already not selected */
+	dccg35_enable_symclk_be_new(dccg, SYMCLK_BE_PHYCLK, link_enc_inst);
+
+	dccg35_enable_symclk_fe_new(dccg, stream_enc_inst, (enum symclk_fe_source) link_enc_inst);
+
+}
+
+static void dccg35_disable_symclk_se_cb(
+			struct dccg *dccg,
+			uint32_t stream_enc_inst,
+			uint32_t link_enc_inst)
+{
+	dccg35_disable_symclk_fe_new(dccg, stream_enc_inst);
+
+	/* DMU PHY sequence switches SYMCLK_BE (link_enc_inst) to ref clock once PHY is turned off */
+}
+
+void dccg35_root_gate_disable_control(struct dccg *dccg, uint32_t pipe_idx, uint32_t disable_clock_gating)
+{
+	dccg35_set_dppclk_root_clock_gating(dccg, pipe_idx, disable_clock_gating);
+}
+
+static const struct dccg_funcs dccg35_funcs_new = {
+	.update_dpp_dto = dccg35_update_dpp_dto_cb,
+	.dpp_root_clock_control = dccg35_dpp_root_clock_control_cb,
+	.get_dccg_ref_freq = dccg31_get_dccg_ref_freq,
+	.dccg_init = dccg35_init_cb,
+	.set_dpstreamclk = dccg35_set_dpstreamclk_cb,
+	.set_dpstreamclk_root_clock_gating = dccg35_set_dpstreamclk_root_clock_gating_cb,
+	.enable_symclk32_se = dccg35_enable_symclk32_se_cb,
+	.disable_symclk32_se = dccg35_disable_symclk32_se_cb,
+	.enable_symclk32_le = dccg35_enable_symclk32_le_cb,
+	.disable_symclk32_le = dccg35_disable_symclk32_le_cb,
+	.set_symclk32_le_root_clock_gating = dccg35_set_symclk32_le_root_clock_gating_cb,
+	.set_physymclk = dccg35_set_physymclk_cb,
+	.set_physymclk_root_clock_gating = dccg35_set_physymclk_root_clock_gating_cb,
+	.set_dtbclk_dto = dccg35_set_dtbclk_dto_cb,
+	.set_audio_dtbclk_dto = dccg31_set_audio_dtbclk_dto,
+	.set_fifo_errdet_ovr_en = dccg2_set_fifo_errdet_ovr_en,
+	.otg_add_pixel = dccg31_otg_add_pixel,
+	.otg_drop_pixel = dccg31_otg_drop_pixel,
+	.set_dispclk_change_mode = dccg31_set_dispclk_change_mode,
+	.disable_dsc = dccg35_disable_dscclk_cb,
+	.enable_dsc = dccg35_enable_dscclk_cb,
+	.set_pixel_rate_div = dccg35_set_pixel_rate_div,
+	.get_pixel_rate_div = dccg35_get_pixel_rate_div,
+	.trigger_dio_fifo_resync = dccg35_trigger_dio_fifo_resync,
+	.set_valid_pixel_rate = dccg35_set_valid_pixel_rate,
+	.enable_symclk_se = dccg35_enable_symclk_se_cb,
+	.disable_symclk_se = dccg35_disable_symclk_se_cb,
+	.set_dtbclk_p_src = dccg35_set_dtbclk_p_src_cb,
+};
+
+static const struct dccg_funcs dccg35_funcs = {
+	.update_dpp_dto = dccg35_update_dpp_dto,
+	.dpp_root_clock_control = dccg35_dpp_root_clock_control,
+	.get_dccg_ref_freq = dccg31_get_dccg_ref_freq,
+	.dccg_init = dccg35_init,
+	.set_dpstreamclk = dccg35_set_dpstreamclk,
+	.set_dpstreamclk_root_clock_gating = dccg35_set_dpstreamclk_root_clock_gating,
+	.enable_symclk32_se = dccg31_enable_symclk32_se,
+	.disable_symclk32_se = dccg35_disable_symclk32_se,
+	.enable_symclk32_le = dccg31_enable_symclk32_le,
+	.disable_symclk32_le = dccg31_disable_symclk32_le,
+	.set_symclk32_le_root_clock_gating = dccg31_set_symclk32_le_root_clock_gating,
+	.set_physymclk = dccg35_set_physymclk,
+	.set_physymclk_root_clock_gating = dccg35_set_physymclk_root_clock_gating,
+	.set_dtbclk_dto = dccg35_set_dtbclk_dto,
+	.set_audio_dtbclk_dto = dccg31_set_audio_dtbclk_dto,
+	.set_fifo_errdet_ovr_en = dccg2_set_fifo_errdet_ovr_en,
+	.otg_add_pixel = dccg31_otg_add_pixel,
+	.otg_drop_pixel = dccg31_otg_drop_pixel,
+	.set_dispclk_change_mode = dccg31_set_dispclk_change_mode,
+	.disable_dsc = dccg35_disable_dscclk,
+	.enable_dsc = dccg35_enable_dscclk,
+	.set_pixel_rate_div = dccg35_set_pixel_rate_div,
+	.get_pixel_rate_div = dccg35_get_pixel_rate_div,
+	.trigger_dio_fifo_resync = dccg35_trigger_dio_fifo_resync,
+	.set_valid_pixel_rate = dccg35_set_valid_pixel_rate,
+	.enable_symclk_se = dccg35_enable_symclk_se,
+	.disable_symclk_se = dccg35_disable_symclk_se,
+	.set_dtbclk_p_src = dccg35_set_dtbclk_p_src,
+	.dccg_root_gate_disable_control = dccg35_root_gate_disable_control,
+};
+
+struct dccg *dccg35_create(
+	struct dc_context *ctx,
+	const struct dccg_registers *regs,
+	const struct dccg_shift *dccg_shift,
+	const struct dccg_mask *dccg_mask)
+{
+	struct dcn_dccg *dccg_dcn = kzalloc(sizeof(*dccg_dcn), GFP_KERNEL);
+	struct dccg *base;
+
+	if (dccg_dcn == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+	(void)&dccg35_disable_symclk_be_new;
+	(void)&dccg35_set_symclk32_le_root_clock_gating;
+	(void)&dccg35_set_smclk32_se_rcg;
+	(void)&dccg35_funcs_new;
+
+	base = &dccg_dcn->base;
+	base->ctx = ctx;
+	base->funcs = &dccg35_funcs;
+
+	dccg_dcn->regs = regs;
+	dccg_dcn->dccg_shift = dccg_shift;
+	dccg_dcn->dccg_mask = dccg_mask;
+
+	return &dccg_dcn->base;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.h b/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.h
new file mode 100644
index 000000000000..51f98c5c51c4
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.h
@@ -0,0 +1,247 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __DCN35_DCCG_H__
+#define __DCN35_DCCG_H__
+
+#include "dcn314/dcn314_dccg.h"
+
+#define DCCG_SFII(block, reg_name, field_prefix, field_name, inst, post_fix)\
+	.field_prefix ## _ ## field_name[inst] = block ## inst ## _ ## reg_name ## __ ## field_prefix ## inst ## _ ## field_name ## post_fix
+
+
+#define DCCG_REG_LIST_DCN35() \
+	DCCG_REG_LIST_DCN314(),\
+	SR(DPPCLK_CTRL),\
+	SR(DCCG_GATE_DISABLE_CNTL4),\
+	SR(DCCG_GATE_DISABLE_CNTL5),\
+	SR(DCCG_GATE_DISABLE_CNTL6),\
+	SR(DCCG_GLOBAL_FGCG_REP_CNTL),\
+	SR(SYMCLKA_CLOCK_ENABLE),\
+	SR(SYMCLKB_CLOCK_ENABLE),\
+	SR(SYMCLKC_CLOCK_ENABLE),\
+	SR(SYMCLKD_CLOCK_ENABLE),\
+	SR(SYMCLKE_CLOCK_ENABLE)
+
+#define DCCG_MASK_SH_LIST_DCN35(mask_sh) \
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 0, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 1, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 2, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 3, mask_sh),\
+	DCCG_SF(DPPCLK_CTRL, DPPCLK0_EN, mask_sh),\
+	DCCG_SF(DPPCLK_CTRL, DPPCLK1_EN, mask_sh),\
+	DCCG_SF(DPPCLK_CTRL, DPPCLK2_EN, mask_sh),\
+	DCCG_SF(DPPCLK_CTRL, DPPCLK3_EN, mask_sh),\
+	DCCG_SF(DPPCLK0_DTO_PARAM, DPPCLK0_DTO_PHASE, mask_sh),\
+	DCCG_SF(DPPCLK0_DTO_PARAM, DPPCLK0_DTO_MODULO, mask_sh),\
+	DCCG_SF(HDMICHARCLK0_CLOCK_CNTL, HDMICHARCLK0_EN, mask_sh),\
+	DCCG_SF(HDMICHARCLK0_CLOCK_CNTL, HDMICHARCLK0_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYASYMCLK_CLOCK_CNTL, PHYASYMCLK_EN, mask_sh),\
+	DCCG_SF(PHYASYMCLK_CLOCK_CNTL, PHYASYMCLK_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYBSYMCLK_CLOCK_CNTL, PHYBSYMCLK_EN, mask_sh),\
+	DCCG_SF(PHYBSYMCLK_CLOCK_CNTL, PHYBSYMCLK_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYCSYMCLK_CLOCK_CNTL, PHYCSYMCLK_EN, mask_sh),\
+	DCCG_SF(PHYCSYMCLK_CLOCK_CNTL, PHYCSYMCLK_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYDSYMCLK_CLOCK_CNTL, PHYDSYMCLK_EN, mask_sh),\
+	DCCG_SF(PHYDSYMCLK_CLOCK_CNTL, PHYDSYMCLK_SRC_SEL, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK0_EN, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK1_EN, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK2_EN, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK3_EN, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK0_SRC_SEL, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK1_SRC_SEL, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK2_SRC_SEL, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK3_SRC_SEL, mask_sh),\
+	DCCG_SF(HDMISTREAMCLK_CNTL, HDMISTREAMCLK0_EN, mask_sh),\
+	DCCG_SF(HDMISTREAMCLK_CNTL, HDMISTREAMCLK0_SRC_SEL, mask_sh),\
+	DCCG_SF(DSCCLK_DTO_CTRL, DSCCLK0_EN, mask_sh),\
+	DCCG_SF(DSCCLK_DTO_CTRL, DSCCLK1_EN, mask_sh),\
+	DCCG_SF(DSCCLK_DTO_CTRL, DSCCLK2_EN, mask_sh),\
+	DCCG_SF(DSCCLK_DTO_CTRL, DSCCLK3_EN, mask_sh),\
+	DCCG_SF(DSCCLK0_DTO_PARAM, DSCCLK0_DTO_PHASE, mask_sh),\
+	DCCG_SF(DSCCLK0_DTO_PARAM, DSCCLK0_DTO_MODULO, mask_sh),\
+	DCCG_SF(DSCCLK1_DTO_PARAM, DSCCLK1_DTO_PHASE, mask_sh),\
+	DCCG_SF(DSCCLK1_DTO_PARAM, DSCCLK1_DTO_MODULO, mask_sh),\
+	DCCG_SF(DSCCLK2_DTO_PARAM, DSCCLK2_DTO_PHASE, mask_sh),\
+	DCCG_SF(DSCCLK2_DTO_PARAM, DSCCLK2_DTO_MODULO, mask_sh),\
+	DCCG_SF(DSCCLK3_DTO_PARAM, DSCCLK3_DTO_PHASE, mask_sh),\
+	DCCG_SF(DSCCLK3_DTO_PARAM, DSCCLK3_DTO_MODULO, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE0_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE1_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE2_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE3_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE0_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE1_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE2_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE3_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_LE_CNTL, SYMCLK32_LE0_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_LE_CNTL, SYMCLK32_LE1_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_LE_CNTL, SYMCLK32_LE0_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_LE_CNTL, SYMCLK32_LE1_EN, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLK_DTO, ENABLE, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLK_DTO, ENABLE, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLK_DTO, ENABLE, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLK_DTO, ENABLE, 3, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLKDTO, ENABLE_STATUS, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLKDTO, ENABLE_STATUS, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLKDTO, ENABLE_STATUS, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DTBCLKDTO, ENABLE_STATUS, 3, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 3, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 3, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG0_PIXEL_RATE_DIVK1, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG0_PIXEL_RATE_DIVK2, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG1_PIXEL_RATE_DIVK1, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG1_PIXEL_RATE_DIVK2, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG2_PIXEL_RATE_DIVK1, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG2_PIXEL_RATE_DIVK2, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG3_PIXEL_RATE_DIVK1, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG3_PIXEL_RATE_DIVK2, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG3_PIXEL_RATE_DIVK2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, DROP_PIXEL, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, DROP_PIXEL, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, DROP_PIXEL, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, DROP_PIXEL, 3, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P0_SRC_SEL, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P0_EN, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P1_SRC_SEL, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P1_EN, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P2_SRC_SEL, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P2_EN, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P3_SRC_SEL, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P3_EN, mask_sh),\
+	DCCG_SF(PHYASYMCLK_CLOCK_CNTL, PHYASYMCLK_EN, mask_sh),\
+	DCCG_SF(PHYASYMCLK_CLOCK_CNTL, PHYASYMCLK_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYBSYMCLK_CLOCK_CNTL, PHYBSYMCLK_EN, mask_sh),\
+	DCCG_SF(PHYBSYMCLK_CLOCK_CNTL, PHYBSYMCLK_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYCSYMCLK_CLOCK_CNTL, PHYCSYMCLK_EN, mask_sh),\
+	DCCG_SF(PHYCSYMCLK_CLOCK_CNTL, PHYCSYMCLK_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYDSYMCLK_CLOCK_CNTL, PHYDSYMCLK_EN, mask_sh),\
+	DCCG_SF(PHYDSYMCLK_CLOCK_CNTL, PHYDSYMCLK_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYESYMCLK_CLOCK_CNTL, PHYESYMCLK_EN, mask_sh),\
+	DCCG_SF(PHYESYMCLK_CLOCK_CNTL, PHYESYMCLK_SRC_SEL, mask_sh),\
+	DCCG_SF(DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO_SEL, mask_sh),\
+	DCCG_SF(DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO0_SOURCE_SEL, mask_sh),\
+	DCCG_SF(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_CHG_DONE, mask_sh),\
+	DCCG_SF(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_RDIVIDER, mask_sh),\
+	DCCG_SF(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_WDIVIDER, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, PHYASYMCLK_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, PHYBSYMCLK_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, PHYCSYMCLK_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, PHYDSYMCLK_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, PHYESYMCLK_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GLOBAL_FGCG_REP_CNTL, DCCG_GLOBAL_FGCG_REP_DIS, mask_sh),\
+	DCCG_SF(SYMCLKA_CLOCK_ENABLE, SYMCLKA_CLOCK_ENABLE, mask_sh),\
+	DCCG_SF(SYMCLKB_CLOCK_ENABLE, SYMCLKB_CLOCK_ENABLE, mask_sh),\
+	DCCG_SF(SYMCLKC_CLOCK_ENABLE, SYMCLKC_CLOCK_ENABLE, mask_sh),\
+	DCCG_SF(SYMCLKD_CLOCK_ENABLE, SYMCLKD_CLOCK_ENABLE, mask_sh),\
+	DCCG_SF(SYMCLKE_CLOCK_ENABLE, SYMCLKE_CLOCK_ENABLE, mask_sh),\
+	DCCG_SF(SYMCLKA_CLOCK_ENABLE, SYMCLKA_FE_EN, mask_sh),\
+	DCCG_SF(SYMCLKB_CLOCK_ENABLE, SYMCLKB_FE_EN, mask_sh),\
+	DCCG_SF(SYMCLKC_CLOCK_ENABLE, SYMCLKC_FE_EN, mask_sh),\
+	DCCG_SF(SYMCLKD_CLOCK_ENABLE, SYMCLKD_FE_EN, mask_sh),\
+	DCCG_SF(SYMCLKE_CLOCK_ENABLE, SYMCLKE_FE_EN, mask_sh),\
+	DCCG_SF(SYMCLKA_CLOCK_ENABLE, SYMCLKA_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLKB_CLOCK_ENABLE, SYMCLKB_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLKC_CLOCK_ENABLE, SYMCLKC_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLKD_CLOCK_ENABLE, SYMCLKD_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLKE_CLOCK_ENABLE, SYMCLKE_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLKA_CLOCK_ENABLE, SYMCLKA_FE_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLKB_CLOCK_ENABLE, SYMCLKB_FE_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLKC_CLOCK_ENABLE, SYMCLKC_FE_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLKD_CLOCK_ENABLE, SYMCLKD_FE_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLKE_CLOCK_ENABLE, SYMCLKE_FE_SRC_SEL, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P1_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P2_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P3_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL6, DSCCLK0_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL6, DSCCLK1_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL6, DSCCLK2_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL6, DSCCLK3_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, SYMCLKA_FE_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, SYMCLKB_FE_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, SYMCLKC_FE_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, SYMCLKD_FE_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, SYMCLKE_FE_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL6, DPPCLK0_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL6, DPPCLK1_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL6, DPPCLK2_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL6, DPPCLK3_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, HDMICHARCLK0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL4, HDMICHARCLK0_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL6, HDMISTREAMCLK0_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, SYMCLKA_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, SYMCLKB_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, SYMCLKC_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, SYMCLKD_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, SYMCLKE_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_SE0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_SE1_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_SE2_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_SE3_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_LE0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_LE1_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_SE0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_SE1_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_SE2_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_SE3_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_LE0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_LE1_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL4, PHYA_REFCLK_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL4, PHYB_REFCLK_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL4, PHYC_REFCLK_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL4, PHYD_REFCLK_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL4, PHYE_REFCLK_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK0_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK1_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK2_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK3_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(HDMISTREAMCLK0_DTO_PARAM, HDMISTREAMCLK0_DTO_PHASE, mask_sh),\
+	DCCG_SF(HDMISTREAMCLK0_DTO_PARAM, HDMISTREAMCLK0_DTO_MODULO, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL, DISPCLK_DCCG_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, HDMISTREAMCLK0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK1_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK2_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK3_GATE_DISABLE, mask_sh),\
+
+struct dccg *dccg35_create(
+		struct dc_context *ctx,
+		const struct dccg_registers *regs,
+		const struct dccg_shift *dccg_shift,
+		const struct dccg_mask *dccg_mask);
+
+void dccg35_init(struct dccg *dccg);
+
+void dccg35_enable_global_fgcg_rep(struct dccg *dccg, bool value);
+void dccg35_root_gate_disable_control(struct dccg *dccg, uint32_t pipe_idx, uint32_t disable_clock_gating);
+
+
+#endif //__DCN35_DCCG_H__
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn401/dcn401_dccg.c b/drivers/gpu/drm/amd/display/dc/dccg/dcn401/dcn401_dccg.c
new file mode 100644
index 000000000000..0b8ed9b94d3c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn401/dcn401_dccg.c
@@ -0,0 +1,914 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+#include "core_types.h"
+#include "dcn401_dccg.h"
+#include "dcn31/dcn31_dccg.h"
+
+/*
+#include "dmub_common.h"
+#include "dmcub_reg_access_helper.h"
+
+#include "dmub401_common.h"
+#include "dmub401_regs.h"
+#include "dmub401_dccg.h"
+*/
+
+#define TO_DCN_DCCG(dccg)\
+	container_of(dccg, struct dcn_dccg, base)
+
+#define REG(reg) \
+	(dccg_dcn->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	dccg_dcn->dccg_shift->field_name, dccg_dcn->dccg_mask->field_name
+
+#define CTX \
+	dccg_dcn->base.ctx
+#define DC_LOGGER \
+	dccg->ctx->logger
+
+static void dcn401_set_dppclk_enable(struct dccg *dccg,
+				 uint32_t dpp_inst, uint32_t enable)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	switch (dpp_inst) {
+	case 0:
+		REG_UPDATE(DPPCLK_CTRL, DPPCLK0_EN, enable);
+		break;
+	case 1:
+		REG_UPDATE(DPPCLK_CTRL, DPPCLK1_EN, enable);
+		break;
+	case 2:
+		REG_UPDATE(DPPCLK_CTRL, DPPCLK2_EN, enable);
+		break;
+	case 3:
+		REG_UPDATE(DPPCLK_CTRL, DPPCLK3_EN, enable);
+		break;
+	default:
+		break;
+	}
+}
+void dccg401_update_dpp_dto(struct dccg *dccg, int dpp_inst, int req_dppclk)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	if (dccg->ref_dppclk && req_dppclk) {
+		int ref_dppclk = dccg->ref_dppclk;
+		int modulo, phase;
+
+		// phase / modulo = dpp pipe clk / dpp global clk
+		modulo = 0xff;   // use FF at the end
+		phase = ((modulo * req_dppclk) + ref_dppclk - 1) / ref_dppclk;
+
+		if (phase > 0xff) {
+			ASSERT(false);
+			phase = 0xff;
+		}
+
+		REG_SET_2(DPPCLK_DTO_PARAM[dpp_inst], 0,
+				DPPCLK0_DTO_PHASE, phase,
+				DPPCLK0_DTO_MODULO, modulo);
+		dcn401_set_dppclk_enable(dccg, dpp_inst, true);
+	} else {
+		dcn401_set_dppclk_enable(dccg, dpp_inst, false);
+	}
+
+	dccg->pipe_dppclk_khz[dpp_inst] = req_dppclk;
+}
+
+/* This function is a workaround for writing to OTG_PIXEL_RATE_DIV
+ * without the probability of causing a DIG FIFO error.
+ */
+static void dccg401_wait_for_dentist_change_done(
+	struct dccg *dccg)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	uint32_t dentist_dispclk_value = REG_READ(DENTIST_DISPCLK_CNTL);
+
+	REG_WRITE(DENTIST_DISPCLK_CNTL, dentist_dispclk_value);
+	REG_WAIT(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_CHG_DONE, 1, 50, 2000);
+}
+
+void dccg401_get_pixel_rate_div(
+		struct dccg *dccg,
+		uint32_t otg_inst,
+		uint32_t *tmds_div,
+		uint32_t *dp_dto_int)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+	uint32_t val_tmds_div = PIXEL_RATE_DIV_NA;
+
+	switch (otg_inst) {
+	case 0:
+		REG_GET_2(OTG_PIXEL_RATE_DIV,
+			OTG0_TMDS_PIXEL_RATE_DIV, &val_tmds_div,
+			DPDTO0_INT, dp_dto_int);
+		break;
+	case 1:
+		REG_GET_2(OTG_PIXEL_RATE_DIV,
+			OTG1_TMDS_PIXEL_RATE_DIV, &val_tmds_div,
+			DPDTO1_INT, dp_dto_int);
+		break;
+	case 2:
+		REG_GET_2(OTG_PIXEL_RATE_DIV,
+			OTG2_TMDS_PIXEL_RATE_DIV, &val_tmds_div,
+			DPDTO2_INT, dp_dto_int);
+		break;
+	case 3:
+		REG_GET_2(OTG_PIXEL_RATE_DIV,
+			OTG3_TMDS_PIXEL_RATE_DIV, &val_tmds_div,
+			DPDTO3_INT, dp_dto_int);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+	*tmds_div = val_tmds_div == 0 ? PIXEL_RATE_DIV_BY_2 : PIXEL_RATE_DIV_BY_4;
+}
+
+void dccg401_set_pixel_rate_div(
+		struct dccg *dccg,
+		uint32_t otg_inst,
+		enum pixel_rate_div tmds_div,
+		enum pixel_rate_div unused)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+	uint32_t cur_tmds_div = PIXEL_RATE_DIV_NA;
+	uint32_t dp_dto_int;
+	uint32_t reg_val;
+
+	// only 2 and 4 are valid on dcn401
+	if (tmds_div != PIXEL_RATE_DIV_BY_2 && tmds_div != PIXEL_RATE_DIV_BY_4) {
+		return;
+	}
+
+	dccg401_get_pixel_rate_div(dccg, otg_inst, &cur_tmds_div, &dp_dto_int);
+	if (tmds_div == cur_tmds_div)
+		return;
+
+	// encode enum to register value
+	reg_val = tmds_div == PIXEL_RATE_DIV_BY_4 ? 1 : 0;
+
+	switch (otg_inst) {
+	case 0:
+		REG_UPDATE(OTG_PIXEL_RATE_DIV,
+				OTG0_TMDS_PIXEL_RATE_DIV, reg_val);
+
+		dccg401_wait_for_dentist_change_done(dccg);
+		break;
+	case 1:
+		REG_UPDATE(OTG_PIXEL_RATE_DIV,
+				OTG1_TMDS_PIXEL_RATE_DIV, reg_val);
+
+		dccg401_wait_for_dentist_change_done(dccg);
+		break;
+	case 2:
+		REG_UPDATE(OTG_PIXEL_RATE_DIV,
+				OTG2_TMDS_PIXEL_RATE_DIV, reg_val);
+
+		dccg401_wait_for_dentist_change_done(dccg);
+		break;
+	case 3:
+		REG_UPDATE(OTG_PIXEL_RATE_DIV,
+				OTG3_TMDS_PIXEL_RATE_DIV, reg_val);
+
+		dccg401_wait_for_dentist_change_done(dccg);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+
+void dccg401_set_dtbclk_p_src(
+		struct dccg *dccg,
+		enum streamclk_source src,
+		uint32_t otg_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	uint32_t p_src_sel = 0; /* selects dprefclk */
+	if (src == DTBCLK0)
+		p_src_sel = 2;  /* selects dtbclk0 */
+
+	switch (otg_inst) {
+	case 0:
+		if (src == REFCLK)
+			REG_UPDATE(DTBCLK_P_CNTL,
+					DTBCLK_P0_EN, 0);
+		else
+			REG_UPDATE_2(DTBCLK_P_CNTL,
+					DTBCLK_P0_SRC_SEL, p_src_sel,
+					DTBCLK_P0_EN, 1);
+		break;
+	case 1:
+		if (src == REFCLK)
+			REG_UPDATE(DTBCLK_P_CNTL,
+					DTBCLK_P1_EN, 0);
+		else
+			REG_UPDATE_2(DTBCLK_P_CNTL,
+					DTBCLK_P1_SRC_SEL, p_src_sel,
+					DTBCLK_P1_EN, 1);
+		break;
+	case 2:
+		if (src == REFCLK)
+			REG_UPDATE(DTBCLK_P_CNTL,
+					DTBCLK_P2_EN, 0);
+		else
+			REG_UPDATE_2(DTBCLK_P_CNTL,
+					DTBCLK_P2_SRC_SEL, p_src_sel,
+					DTBCLK_P2_EN, 1);
+		break;
+	case 3:
+		if (src == REFCLK)
+			REG_UPDATE(DTBCLK_P_CNTL,
+					DTBCLK_P3_EN, 0);
+		else
+			REG_UPDATE_2(DTBCLK_P_CNTL,
+					DTBCLK_P3_SRC_SEL, p_src_sel,
+					DTBCLK_P3_EN, 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+}
+
+void dccg401_set_physymclk(
+		struct dccg *dccg,
+		int phy_inst,
+		enum physymclk_clock_source clk_src,
+		bool force_enable)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	/* Force PHYSYMCLK on and Select phyd32clk as the source of clock which is output to PHY through DCIO */
+	switch (phy_inst) {
+	case 0:
+		if (force_enable) {
+			REG_UPDATE_2(PHYASYMCLK_CLOCK_CNTL,
+					PHYASYMCLK_EN, 1,
+					PHYASYMCLK_SRC_SEL, clk_src);
+			if (dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk)
+				REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+					PHYASYMCLK_ROOT_GATE_DISABLE, 1);
+		} else {
+			REG_UPDATE_2(PHYASYMCLK_CLOCK_CNTL,
+					PHYASYMCLK_EN, 0,
+					PHYASYMCLK_SRC_SEL, 0);
+			if (dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk)
+				REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+					PHYASYMCLK_ROOT_GATE_DISABLE, 0);
+		}
+		break;
+	case 1:
+		if (force_enable) {
+			REG_UPDATE_2(PHYBSYMCLK_CLOCK_CNTL,
+					PHYBSYMCLK_EN, 1,
+					PHYBSYMCLK_SRC_SEL, clk_src);
+			if (dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk)
+				REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+					PHYBSYMCLK_ROOT_GATE_DISABLE, 1);
+		} else {
+			REG_UPDATE_2(PHYBSYMCLK_CLOCK_CNTL,
+					PHYBSYMCLK_EN, 0,
+					PHYBSYMCLK_SRC_SEL, 0);
+			if (dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk)
+				REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+					PHYBSYMCLK_ROOT_GATE_DISABLE, 0);
+		}
+		break;
+	case 2:
+		if (force_enable) {
+			REG_UPDATE_2(PHYCSYMCLK_CLOCK_CNTL,
+					PHYCSYMCLK_EN, 1,
+					PHYCSYMCLK_SRC_SEL, clk_src);
+			if (dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk)
+				REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+					PHYCSYMCLK_ROOT_GATE_DISABLE, 1);
+		} else {
+			REG_UPDATE_2(PHYCSYMCLK_CLOCK_CNTL,
+					PHYCSYMCLK_EN, 0,
+					PHYCSYMCLK_SRC_SEL, 0);
+			if (dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk)
+				REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+					PHYCSYMCLK_ROOT_GATE_DISABLE, 0);
+		}
+		break;
+	case 3:
+		if (force_enable) {
+			REG_UPDATE_2(PHYDSYMCLK_CLOCK_CNTL,
+					PHYDSYMCLK_EN, 1,
+					PHYDSYMCLK_SRC_SEL, clk_src);
+			if (dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk)
+				REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+					PHYDSYMCLK_ROOT_GATE_DISABLE, 1);
+		} else {
+			REG_UPDATE_2(PHYDSYMCLK_CLOCK_CNTL,
+					PHYDSYMCLK_EN, 0,
+					PHYDSYMCLK_SRC_SEL, 0);
+			if (dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk)
+				REG_UPDATE(DCCG_GATE_DISABLE_CNTL2,
+					PHYDSYMCLK_ROOT_GATE_DISABLE, 0);
+		}
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+void dccg401_get_dccg_ref_freq(struct dccg *dccg,
+		unsigned int xtalin_freq_inKhz,
+		unsigned int *dccg_ref_freq_inKhz)
+{
+	/*
+	 * Assume refclk is sourced from xtalin
+	 * expect 100MHz
+	 */
+	*dccg_ref_freq_inKhz = xtalin_freq_inKhz;
+	return;
+}
+
+static void dccg401_otg_add_pixel(struct dccg *dccg,
+		uint32_t otg_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	REG_UPDATE(OTG_PIXEL_RATE_CNTL[otg_inst],
+			OTG_ADD_PIXEL[otg_inst], 1);
+}
+
+static void dccg401_otg_drop_pixel(struct dccg *dccg,
+		uint32_t otg_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	REG_UPDATE(OTG_PIXEL_RATE_CNTL[otg_inst],
+			OTG_DROP_PIXEL[otg_inst], 1);
+}
+
+void dccg401_enable_symclk32_le(
+		struct dccg *dccg,
+		int hpo_le_inst,
+		enum phyd32clk_clock_source phyd32clk)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	/* select one of the PHYD32CLKs as the source for symclk32_le */
+	switch (hpo_le_inst) {
+	case 0:
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_le)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_LE0_GATE_DISABLE, 1,
+					SYMCLK32_ROOT_LE0_GATE_DISABLE, 1);
+		REG_UPDATE_2(SYMCLK32_LE_CNTL,
+				SYMCLK32_LE0_SRC_SEL, phyd32clk,
+				SYMCLK32_LE0_EN, 1);
+		break;
+	case 1:
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_le)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_LE1_GATE_DISABLE, 1,
+					SYMCLK32_ROOT_LE1_GATE_DISABLE, 1);
+		REG_UPDATE_2(SYMCLK32_LE_CNTL,
+				SYMCLK32_LE1_SRC_SEL, phyd32clk,
+				SYMCLK32_LE1_EN, 1);
+		break;
+	case 2:
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_le)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_LE2_GATE_DISABLE, 1,
+					SYMCLK32_ROOT_LE2_GATE_DISABLE, 1);
+		REG_UPDATE_2(SYMCLK32_LE_CNTL,
+				SYMCLK32_LE2_SRC_SEL, phyd32clk,
+				SYMCLK32_LE2_EN, 1);
+		break;
+	case 3:
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_le)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_LE3_GATE_DISABLE, 1,
+					SYMCLK32_ROOT_LE3_GATE_DISABLE, 1);
+		REG_UPDATE_2(SYMCLK32_LE_CNTL,
+				SYMCLK32_LE3_SRC_SEL, phyd32clk,
+				SYMCLK32_LE3_EN, 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+void dccg401_disable_symclk32_le(
+		struct dccg *dccg,
+		int hpo_le_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	/* set refclk as the source for symclk32_le */
+	switch (hpo_le_inst) {
+	case 0:
+		REG_UPDATE_2(SYMCLK32_LE_CNTL,
+				SYMCLK32_LE0_SRC_SEL, 0,
+				SYMCLK32_LE0_EN, 0);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_le)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_LE0_GATE_DISABLE, 0,
+					SYMCLK32_ROOT_LE0_GATE_DISABLE, 0);
+		break;
+	case 1:
+		REG_UPDATE_2(SYMCLK32_LE_CNTL,
+				SYMCLK32_LE1_SRC_SEL, 0,
+				SYMCLK32_LE1_EN, 0);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_le)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_LE1_GATE_DISABLE, 0,
+					SYMCLK32_ROOT_LE1_GATE_DISABLE, 0);
+		break;
+	case 2:
+		REG_UPDATE_2(SYMCLK32_LE_CNTL,
+				SYMCLK32_LE2_SRC_SEL, 0,
+				SYMCLK32_LE2_EN, 0);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_le)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_LE2_GATE_DISABLE, 0,
+					SYMCLK32_ROOT_LE2_GATE_DISABLE, 0);
+		break;
+	case 3:
+		REG_UPDATE_2(SYMCLK32_LE_CNTL,
+				SYMCLK32_LE3_SRC_SEL, 0,
+				SYMCLK32_LE3_EN, 0);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_le)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_LE3_GATE_DISABLE, 0,
+					SYMCLK32_ROOT_LE3_GATE_DISABLE, 0);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+static void dccg401_enable_dpstreamclk(struct dccg *dccg, int otg_inst, int dp_hpo_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	/* enabled to select one of the DTBCLKs for pipe */
+	switch (dp_hpo_inst) {
+	case 0:
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL5,
+					DPSTREAMCLK0_ROOT_GATE_DISABLE, 1,
+					DPSTREAMCLK0_GATE_DISABLE, 1);
+		REG_UPDATE_2(DPSTREAMCLK_CNTL,
+				DPSTREAMCLK0_SRC_SEL, otg_inst,
+				DPSTREAMCLK0_EN, 1);
+		break;
+	case 1:
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL5,
+					DPSTREAMCLK1_ROOT_GATE_DISABLE, 1,
+					DPSTREAMCLK1_GATE_DISABLE, 1);
+		REG_UPDATE_2(DPSTREAMCLK_CNTL,
+				DPSTREAMCLK1_SRC_SEL, otg_inst,
+				DPSTREAMCLK1_EN, 1);
+		break;
+	case 2:
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL5,
+					DPSTREAMCLK2_ROOT_GATE_DISABLE, 1,
+					DPSTREAMCLK2_GATE_DISABLE, 1);
+		REG_UPDATE_2(DPSTREAMCLK_CNTL,
+				DPSTREAMCLK2_SRC_SEL, otg_inst,
+				DPSTREAMCLK2_EN, 1);
+		break;
+	case 3:
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL5,
+					DPSTREAMCLK3_ROOT_GATE_DISABLE, 1,
+					DPSTREAMCLK3_GATE_DISABLE, 1);
+		REG_UPDATE_2(DPSTREAMCLK_CNTL,
+				DPSTREAMCLK3_SRC_SEL, otg_inst,
+				DPSTREAMCLK3_EN, 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+	if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream)
+		REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL3,
+			DPSTREAMCLK_GATE_DISABLE, 1,
+			DPSTREAMCLK_ROOT_GATE_DISABLE, 1);
+}
+
+void dccg401_disable_dpstreamclk(struct dccg *dccg, int dp_hpo_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	switch (dp_hpo_inst) {
+	case 0:
+		REG_UPDATE(DPSTREAMCLK_CNTL,
+				DPSTREAMCLK0_EN, 0);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL5,
+					DPSTREAMCLK0_ROOT_GATE_DISABLE, 0,
+					DPSTREAMCLK0_GATE_DISABLE, 0);
+		break;
+	case 1:
+		REG_UPDATE(DPSTREAMCLK_CNTL,
+				DPSTREAMCLK1_EN, 0);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL5,
+					DPSTREAMCLK1_ROOT_GATE_DISABLE, 0,
+					DPSTREAMCLK1_GATE_DISABLE, 0);
+		break;
+	case 2:
+		REG_UPDATE(DPSTREAMCLK_CNTL,
+				DPSTREAMCLK2_EN, 0);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL5,
+					DPSTREAMCLK2_ROOT_GATE_DISABLE, 0,
+					DPSTREAMCLK2_GATE_DISABLE, 0);
+		break;
+	case 3:
+		REG_UPDATE(DPSTREAMCLK_CNTL,
+				DPSTREAMCLK3_EN, 0);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream)
+			REG_UPDATE_2(DCCG_GATE_DISABLE_CNTL5,
+					DPSTREAMCLK3_ROOT_GATE_DISABLE, 0,
+					DPSTREAMCLK3_GATE_DISABLE, 0);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+void dccg401_set_dpstreamclk(
+		struct dccg *dccg,
+		enum streamclk_source src,
+		int otg_inst,
+		int dp_hpo_inst)
+{
+	/* enabled to select one of the DTBCLKs for pipe */
+	if (src == REFCLK)
+		dccg401_disable_dpstreamclk(dccg, dp_hpo_inst);
+	else
+		dccg401_enable_dpstreamclk(dccg, otg_inst, dp_hpo_inst);
+}
+
+void dccg401_set_dp_dto(
+		struct dccg *dccg,
+		const struct dp_dto_params *params)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	bool enable = false;
+
+	if (params->otg_inst > 3) {
+		/* dcn401 only has 4 instances */
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+	if (!params->refclk_hz) {
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+	if (!dc_is_tmds_signal(params->signal)) {
+		uint64_t dto_integer;
+		uint64_t dto_phase_hz;
+		uint64_t dto_modulo_hz = params->refclk_hz;
+
+		enable = true;
+
+		/* Set DTO values:
+		 * int = target_pix_rate / reference_clock
+		 * phase = target_pix_rate - int * reference_clock,
+		 * modulo = reference_clock */
+		dto_integer = div_u64(params->pixclk_hz, dto_modulo_hz);
+		dto_phase_hz = params->pixclk_hz - dto_integer * dto_modulo_hz;
+
+		if (dto_phase_hz <= 0 && dto_integer <= 0) {
+			/* negative pixel rate should never happen */
+			BREAK_TO_DEBUGGER();
+			return;
+		}
+
+		switch (params->otg_inst) {
+		case 0:
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P0_GATE_DISABLE, 1);
+			REG_UPDATE_4(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_SE0_GATE_DISABLE, 1,
+					SYMCLK32_ROOT_SE0_GATE_DISABLE, 1,
+					SYMCLK32_LE0_GATE_DISABLE, 1,
+					SYMCLK32_ROOT_LE0_GATE_DISABLE, 1);
+			break;
+		case 1:
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P1_GATE_DISABLE, 1);
+			REG_UPDATE_4(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_SE1_GATE_DISABLE, 1,
+					SYMCLK32_ROOT_SE1_GATE_DISABLE, 1,
+					SYMCLK32_LE1_GATE_DISABLE, 1,
+					SYMCLK32_ROOT_LE1_GATE_DISABLE, 1);
+			break;
+		case 2:
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P2_GATE_DISABLE, 1);
+			REG_UPDATE_4(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_SE2_GATE_DISABLE, 1,
+					SYMCLK32_ROOT_SE2_GATE_DISABLE, 1,
+					SYMCLK32_LE2_GATE_DISABLE, 1,
+					SYMCLK32_ROOT_LE2_GATE_DISABLE, 1);
+			break;
+		case 3:
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P3_GATE_DISABLE, 1);
+			REG_UPDATE_4(DCCG_GATE_DISABLE_CNTL3,
+					SYMCLK32_SE3_GATE_DISABLE, 1,
+					SYMCLK32_ROOT_SE3_GATE_DISABLE, 1,
+					SYMCLK32_LE3_GATE_DISABLE, 1,
+					SYMCLK32_ROOT_LE3_GATE_DISABLE, 1);
+			break;
+		}
+
+		dccg401_set_dtbclk_p_src(dccg, params->clk_src, params->otg_inst);
+
+		REG_WRITE(DP_DTO_PHASE[params->otg_inst], dto_phase_hz);
+		REG_WRITE(DP_DTO_MODULO[params->otg_inst], dto_modulo_hz);
+
+		switch (params->otg_inst) {
+		case 0:
+			REG_UPDATE(OTG_PIXEL_RATE_DIV,
+					DPDTO0_INT, dto_integer);
+			break;
+		case 1:
+			REG_UPDATE(OTG_PIXEL_RATE_DIV,
+					DPDTO1_INT, dto_integer);
+			break;
+		case 2:
+			REG_UPDATE(OTG_PIXEL_RATE_DIV,
+					DPDTO2_INT, dto_integer);
+			break;
+		case 3:
+			REG_UPDATE(OTG_PIXEL_RATE_DIV,
+					DPDTO3_INT, dto_integer);
+			break;
+		default:
+			BREAK_TO_DEBUGGER();
+			return;
+		}
+	}
+
+	/* Toggle DTO */
+	REG_UPDATE_2(OTG_PIXEL_RATE_CNTL[params->otg_inst],
+			DP_DTO_ENABLE[params->otg_inst], enable,
+			PIPE_DTO_SRC_SEL[params->otg_inst], enable);
+}
+
+void dccg401_init(struct dccg *dccg)
+{
+	/* Set HPO stream encoder to use refclk to avoid case where PHY is
+	 * disabled and SYMCLK32 for HPO SE is sourced from PHYD32CLK which
+	 * will cause DCN to hang.
+	 */
+	dccg31_disable_symclk32_se(dccg, 0);
+	dccg31_disable_symclk32_se(dccg, 1);
+	dccg31_disable_symclk32_se(dccg, 2);
+	dccg31_disable_symclk32_se(dccg, 3);
+
+	if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_le) {
+		dccg401_disable_symclk32_le(dccg, 0);
+		dccg401_disable_symclk32_le(dccg, 1);
+		dccg401_disable_symclk32_le(dccg, 2);
+		dccg401_disable_symclk32_le(dccg, 3);
+	}
+
+	if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream) {
+		dccg401_disable_dpstreamclk(dccg, 0);
+		dccg401_disable_dpstreamclk(dccg, 1);
+		dccg401_disable_dpstreamclk(dccg, 2);
+		dccg401_disable_dpstreamclk(dccg, 3);
+	}
+
+	if (dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk) {
+		dccg401_set_physymclk(dccg, 0, PHYSYMCLK_FORCE_SRC_SYMCLK, false);
+		dccg401_set_physymclk(dccg, 1, PHYSYMCLK_FORCE_SRC_SYMCLK, false);
+		dccg401_set_physymclk(dccg, 2, PHYSYMCLK_FORCE_SRC_SYMCLK, false);
+		dccg401_set_physymclk(dccg, 3, PHYSYMCLK_FORCE_SRC_SYMCLK, false);
+	}
+}
+
+void dccg401_set_dto_dscclk(struct dccg *dccg, uint32_t inst, uint32_t num_slices_h)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	switch (inst) {
+	case 0:
+		REG_UPDATE_2(DSCCLK0_DTO_PARAM,
+				DSCCLK0_DTO_PHASE, 1,
+				DSCCLK0_DTO_MODULO, 1);
+		REG_UPDATE(DSCCLK_DTO_CTRL, DSCCLK0_EN, 1);
+
+		break;
+	case 1:
+		REG_UPDATE_2(DSCCLK1_DTO_PARAM,
+				DSCCLK1_DTO_PHASE, 1,
+				DSCCLK1_DTO_MODULO, 1);
+		REG_UPDATE(DSCCLK_DTO_CTRL, DSCCLK1_EN, 1);
+		break;
+	case 2:
+		REG_UPDATE_2(DSCCLK2_DTO_PARAM,
+				DSCCLK2_DTO_PHASE, 1,
+				DSCCLK2_DTO_MODULO, 1);
+		REG_UPDATE(DSCCLK_DTO_CTRL, DSCCLK2_EN, 1);
+		break;
+	case 3:
+		REG_UPDATE_2(DSCCLK3_DTO_PARAM,
+				DSCCLK3_DTO_PHASE, 1,
+				DSCCLK3_DTO_MODULO, 1);
+		REG_UPDATE(DSCCLK_DTO_CTRL, DSCCLK3_EN, 1);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+}
+
+void dccg401_set_ref_dscclk(struct dccg *dccg,
+				uint32_t dsc_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	switch (dsc_inst) {
+	case 0:
+		REG_UPDATE(DSCCLK_DTO_CTRL, DSCCLK0_EN, 0);
+		REG_UPDATE_2(DSCCLK0_DTO_PARAM,
+				DSCCLK0_DTO_PHASE, 0,
+				DSCCLK0_DTO_MODULO, 0);
+		break;
+	case 1:
+		REG_UPDATE(DSCCLK_DTO_CTRL, DSCCLK1_EN, 0);
+		REG_UPDATE_2(DSCCLK1_DTO_PARAM,
+				DSCCLK1_DTO_PHASE, 0,
+				DSCCLK1_DTO_MODULO, 0);
+		break;
+	case 2:
+		REG_UPDATE(DSCCLK_DTO_CTRL, DSCCLK2_EN, 0);
+		REG_UPDATE_2(DSCCLK2_DTO_PARAM,
+				DSCCLK2_DTO_PHASE, 0,
+				DSCCLK2_DTO_MODULO, 0);
+		break;
+	case 3:
+		REG_UPDATE(DSCCLK_DTO_CTRL, DSCCLK3_EN, 0);
+		REG_UPDATE_2(DSCCLK3_DTO_PARAM,
+				DSCCLK3_DTO_PHASE, 0,
+				DSCCLK3_DTO_MODULO, 0);
+		break;
+	default:
+		return;
+	}
+}
+
+void dccg401_enable_symclk_se(struct dccg *dccg, uint32_t stream_enc_inst, uint32_t link_enc_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	switch (stream_enc_inst) {
+	case 0:
+		REG_UPDATE_2(SYMCLKA_CLOCK_ENABLE,
+				SYMCLKA_FE_EN, 1,
+				SYMCLKA_FE_SRC_SEL, link_enc_inst);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKA_FE_ROOT_GATE_DISABLE, 1);
+		break;
+	case 1:
+		REG_UPDATE_2(SYMCLKB_CLOCK_ENABLE,
+				SYMCLKB_FE_EN, 1,
+				SYMCLKB_FE_SRC_SEL, link_enc_inst);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKB_FE_ROOT_GATE_DISABLE, 1);
+		break;
+	case 2:
+		REG_UPDATE_2(SYMCLKC_CLOCK_ENABLE,
+				SYMCLKC_FE_EN, 1,
+				SYMCLKC_FE_SRC_SEL, link_enc_inst);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKC_FE_ROOT_GATE_DISABLE, 1);
+		break;
+	case 3:
+		REG_UPDATE_2(SYMCLKD_CLOCK_ENABLE,
+				SYMCLKD_FE_EN, 1,
+				SYMCLKD_FE_SRC_SEL, link_enc_inst);
+		if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_se)
+			REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, SYMCLKD_FE_ROOT_GATE_DISABLE, 1);
+		break;
+	}
+}
+
+void dccg401_disable_symclk_se(struct dccg *dccg, uint32_t stream_enc_inst, uint32_t link_enc_inst)
+{
+	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);
+
+	switch (stream_enc_inst) {
+	case 0:
+		REG_UPDATE_2(SYMCLKA_CLOCK_ENABLE,
+				SYMCLKA_FE_EN, 0,
+				SYMCLKA_FE_SRC_SEL, 0);
+		break;
+	case 1:
+		REG_UPDATE_2(SYMCLKB_CLOCK_ENABLE,
+				SYMCLKB_FE_EN, 0,
+				SYMCLKB_FE_SRC_SEL, 0);
+		break;
+	case 2:
+		REG_UPDATE_2(SYMCLKC_CLOCK_ENABLE,
+				SYMCLKC_FE_EN, 0,
+				SYMCLKC_FE_SRC_SEL, 0);
+		break;
+	case 3:
+		REG_UPDATE_2(SYMCLKD_CLOCK_ENABLE,
+				SYMCLKD_FE_EN, 0,
+				SYMCLKD_FE_SRC_SEL, 0);
+		break;
+	}
+}
+
+static const struct dccg_funcs dccg401_funcs = {
+	.update_dpp_dto = dccg401_update_dpp_dto,
+	.get_dccg_ref_freq = dccg401_get_dccg_ref_freq,
+	.dccg_init = dccg401_init,
+	.set_dpstreamclk = dccg401_set_dpstreamclk,
+	.enable_symclk32_se = dccg31_enable_symclk32_se,
+	.disable_symclk32_se = dccg31_disable_symclk32_se,
+	.enable_symclk32_le = dccg401_enable_symclk32_le,
+	.disable_symclk32_le = dccg401_disable_symclk32_le,
+	.set_physymclk = dccg401_set_physymclk,
+	.set_dtbclk_dto = NULL,
+	.set_dto_dscclk = dccg401_set_dto_dscclk,
+	.set_ref_dscclk = dccg401_set_ref_dscclk,
+	.set_valid_pixel_rate = NULL,
+	.set_fifo_errdet_ovr_en = dccg2_set_fifo_errdet_ovr_en,
+	.set_audio_dtbclk_dto = NULL,
+	.otg_add_pixel = dccg401_otg_add_pixel,
+	.otg_drop_pixel = dccg401_otg_drop_pixel,
+	.set_pixel_rate_div = dccg401_set_pixel_rate_div,
+	.get_pixel_rate_div = dccg401_get_pixel_rate_div,
+	.set_dp_dto = dccg401_set_dp_dto,
+	.enable_symclk_se = dccg401_enable_symclk_se,
+	.disable_symclk_se = dccg401_disable_symclk_se,
+	.set_dtbclk_p_src = dccg401_set_dtbclk_p_src,
+};
+
+struct dccg *dccg401_create(
+	struct dc_context *ctx,
+	const struct dccg_registers *regs,
+	const struct dccg_shift *dccg_shift,
+	const struct dccg_mask *dccg_mask)
+{
+	struct dcn_dccg *dccg_dcn = kzalloc(sizeof(*dccg_dcn), GFP_KERNEL);
+	struct dccg *base;
+
+	if (dccg_dcn == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	base = &dccg_dcn->base;
+	base->ctx = ctx;
+	base->funcs = &dccg401_funcs;
+
+	dccg_dcn->regs = regs;
+	dccg_dcn->dccg_shift = dccg_shift;
+	dccg_dcn->dccg_mask = dccg_mask;
+
+	return &dccg_dcn->base;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn401/dcn401_dccg.h b/drivers/gpu/drm/amd/display/dc/dccg/dcn401/dcn401_dccg.h
new file mode 100644
index 000000000000..5947a35363aa
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn401/dcn401_dccg.h
@@ -0,0 +1,249 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCN401_DCCG_H__
+#define __DCN401_DCCG_H__
+
+#include "dcn32/dcn32_dccg.h"
+
+#define DCCG_SFII(block, reg_name, field_prefix, field_name, inst, post_fix)\
+	.field_prefix ## _ ## field_name[inst] = block ## inst ## _ ## reg_name ## __ ## field_prefix ## inst ## _ ## field_name ## post_fix
+
+#define DCCG_MASK_SH_LIST_DCN401(mask_sh) \
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 0, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 1, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 2, mask_sh),\
+	DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 3, mask_sh),\
+	DCCG_SF(DPPCLK_CTRL, DPPCLK0_EN, mask_sh),\
+	DCCG_SF(DPPCLK_CTRL, DPPCLK1_EN, mask_sh),\
+	DCCG_SF(DPPCLK_CTRL, DPPCLK2_EN, mask_sh),\
+	DCCG_SF(DPPCLK_CTRL, DPPCLK3_EN, mask_sh),\
+	DCCG_SF(DPPCLK0_DTO_PARAM, DPPCLK0_DTO_PHASE, mask_sh),\
+	DCCG_SF(DPPCLK0_DTO_PARAM, DPPCLK0_DTO_MODULO, mask_sh),\
+	DCCG_SF(HDMICHARCLK0_CLOCK_CNTL, HDMICHARCLK0_EN, mask_sh),\
+	DCCG_SF(HDMICHARCLK0_CLOCK_CNTL, HDMICHARCLK0_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYASYMCLK_CLOCK_CNTL, PHYASYMCLK_EN, mask_sh),\
+	DCCG_SF(PHYASYMCLK_CLOCK_CNTL, PHYASYMCLK_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYBSYMCLK_CLOCK_CNTL, PHYBSYMCLK_EN, mask_sh),\
+	DCCG_SF(PHYBSYMCLK_CLOCK_CNTL, PHYBSYMCLK_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYCSYMCLK_CLOCK_CNTL, PHYCSYMCLK_EN, mask_sh),\
+	DCCG_SF(PHYCSYMCLK_CLOCK_CNTL, PHYCSYMCLK_SRC_SEL, mask_sh),\
+	DCCG_SF(PHYDSYMCLK_CLOCK_CNTL, PHYDSYMCLK_EN, mask_sh),\
+	DCCG_SF(PHYDSYMCLK_CLOCK_CNTL, PHYDSYMCLK_SRC_SEL, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK0_EN, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK1_EN, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK2_EN, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK3_EN, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK0_SRC_SEL, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK1_SRC_SEL, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK2_SRC_SEL, mask_sh),\
+	DCCG_SF(DPSTREAMCLK_CNTL, DPSTREAMCLK3_SRC_SEL, mask_sh),\
+	DCCG_SF(HDMISTREAMCLK_CNTL, HDMISTREAMCLK0_EN, mask_sh),\
+	DCCG_SF(HDMISTREAMCLK_CNTL, HDMISTREAMCLK0_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE0_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE1_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE2_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE3_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE0_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE1_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE2_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_SE_CNTL, SYMCLK32_SE3_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_LE_CNTL, SYMCLK32_LE0_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_LE_CNTL, SYMCLK32_LE1_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLK32_LE_CNTL, SYMCLK32_LE0_EN, mask_sh),\
+	DCCG_SF(SYMCLK32_LE_CNTL, SYMCLK32_LE1_EN, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 3, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, OTG, ADD_PIXEL, 3, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG0_TMDS_PIXEL_RATE_DIV, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, DPDTO0_INT, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG1_TMDS_PIXEL_RATE_DIV, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, DPDTO1_INT, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG2_TMDS_PIXEL_RATE_DIV, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, DPDTO2_INT, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, OTG3_TMDS_PIXEL_RATE_DIV, mask_sh),\
+	DCCG_SF(OTG_PIXEL_RATE_DIV, DPDTO3_INT, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P0_SRC_SEL, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P0_EN, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P1_SRC_SEL, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P1_EN, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P2_SRC_SEL, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P2_EN, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P3_SRC_SEL, mask_sh),\
+	DCCG_SF(DTBCLK_P_CNTL, DTBCLK_P3_EN, mask_sh),\
+	DCCG_SF(DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO_SEL, mask_sh),\
+	DCCG_SF(DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO0_SOURCE_SEL, mask_sh),\
+	DCCG_SF(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_CHG_DONE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, PHYASYMCLK_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, PHYBSYMCLK_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, PHYCSYMCLK_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, PHYDSYMCLK_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DP_DTO, ENABLE, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DP_DTO, ENABLE, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DP_DTO, ENABLE, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, DP_DTO, ENABLE, 3, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 0, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 1, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 2, mask_sh),\
+	DCCG_SFII(OTG, PIXEL_RATE_CNTL, PIPE, DTO_SRC_SEL, 3, mask_sh),\
+	DCCG_SF(DSCCLK_DTO_CTRL, DSCCLK0_EN, mask_sh),\
+	DCCG_SF(DSCCLK_DTO_CTRL, DSCCLK1_EN, mask_sh),\
+	DCCG_SF(DSCCLK_DTO_CTRL, DSCCLK2_EN, mask_sh),\
+	DCCG_SF(DSCCLK_DTO_CTRL, DSCCLK3_EN, mask_sh),\
+	DCCG_SF(DSCCLK0_DTO_PARAM, DSCCLK0_DTO_PHASE, mask_sh),\
+	DCCG_SF(DSCCLK0_DTO_PARAM, DSCCLK0_DTO_MODULO, mask_sh),\
+	DCCG_SF(DSCCLK1_DTO_PARAM, DSCCLK1_DTO_PHASE, mask_sh),\
+	DCCG_SF(DSCCLK1_DTO_PARAM, DSCCLK1_DTO_MODULO, mask_sh),\
+	DCCG_SF(DSCCLK2_DTO_PARAM, DSCCLK2_DTO_PHASE, mask_sh),\
+	DCCG_SF(DSCCLK2_DTO_PARAM, DSCCLK2_DTO_MODULO, mask_sh),\
+	DCCG_SF(DSCCLK3_DTO_PARAM, DSCCLK3_DTO_PHASE, mask_sh),\
+	DCCG_SF(DSCCLK3_DTO_PARAM, DSCCLK3_DTO_MODULO, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, HDMICHARCLK0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, HDMISTREAMCLK0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_SE0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_SE1_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_SE2_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_SE3_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_LE0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_LE1_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_LE2_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_LE3_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_SE0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_SE1_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_SE2_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_SE3_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_LE0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_LE1_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_LE2_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_LE3_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL4, HDMICHARCLK0_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL4, PHYA_REFCLK_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL4, PHYB_REFCLK_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL4, PHYC_REFCLK_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL4, PHYD_REFCLK_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P1_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P2_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P3_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, SYMCLKA_FE_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, SYMCLKB_FE_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, SYMCLKC_FE_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, SYMCLKD_FE_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, SYMCLKA_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, SYMCLKB_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, SYMCLKC_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, SYMCLKD_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK0_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK1_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK2_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK3_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK1_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK2_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK3_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL6, DSCCLK0_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL6, DSCCLK1_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL6, DSCCLK2_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL6, DSCCLK3_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL6, DPPCLK0_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL6, DPPCLK1_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL6, DPPCLK2_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL6, DPPCLK3_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL6, HDMISTREAMCLK0_ROOT_GATE_DISABLE, mask_sh),\
+	DCCG_SF(SYMCLKA_CLOCK_ENABLE, SYMCLKA_CLOCK_ENABLE, mask_sh),\
+	DCCG_SF(SYMCLKB_CLOCK_ENABLE, SYMCLKB_CLOCK_ENABLE, mask_sh),\
+	DCCG_SF(SYMCLKC_CLOCK_ENABLE, SYMCLKC_CLOCK_ENABLE, mask_sh),\
+	DCCG_SF(SYMCLKD_CLOCK_ENABLE, SYMCLKD_CLOCK_ENABLE, mask_sh),\
+	DCCG_SF(SYMCLKA_CLOCK_ENABLE, SYMCLKA_FE_EN, mask_sh),\
+	DCCG_SF(SYMCLKB_CLOCK_ENABLE, SYMCLKB_FE_EN, mask_sh),\
+	DCCG_SF(SYMCLKC_CLOCK_ENABLE, SYMCLKC_FE_EN, mask_sh),\
+	DCCG_SF(SYMCLKD_CLOCK_ENABLE, SYMCLKD_FE_EN, mask_sh),\
+	DCCG_SF(SYMCLKA_CLOCK_ENABLE, SYMCLKA_FE_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLKB_CLOCK_ENABLE, SYMCLKB_FE_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLKC_CLOCK_ENABLE, SYMCLKC_FE_SRC_SEL, mask_sh),\
+	DCCG_SF(SYMCLKD_CLOCK_ENABLE, SYMCLKD_FE_SRC_SEL, mask_sh),\
+
+void dccg401_init(struct dccg *dccg);
+
+void dccg401_update_dpp_dto(struct dccg *dccg, int dpp_inst, int req_dppclk);
+void dccg401_get_dccg_ref_freq(struct dccg *dccg,
+		unsigned int xtalin_freq_inKhz,
+		unsigned int *dccg_ref_freq_inKhz);
+void dccg401_set_dpstreamclk(
+		struct dccg *dccg,
+		enum streamclk_source src,
+		int otg_inst,
+		int dp_hpo_inst);
+void dccg401_enable_symclk32_le(
+		struct dccg *dccg,
+		int hpo_le_inst,
+		enum phyd32clk_clock_source phyd32clk);
+void dccg401_disable_symclk32_le(
+		struct dccg *dccg,
+		int hpo_le_inst);
+void dccg401_disable_dpstreamclk(struct dccg *dccg, int dp_hpo_inst);
+void dccg401_set_dto_dscclk(struct dccg *dccg, uint32_t inst, uint32_t num_slices_h);
+void dccg401_set_ref_dscclk(struct dccg *dccg,
+				uint32_t dsc_inst);
+void dccg401_set_src_sel(
+	struct dccg *dccg,
+	const struct dtbclk_dto_params *params);
+void dccg401_set_pixel_rate_div(
+		struct dccg *dccg,
+		uint32_t otg_inst,
+		enum pixel_rate_div tmds_div,
+		enum pixel_rate_div unused);
+void dccg401_get_pixel_rate_div(
+		struct dccg *dccg,
+		uint32_t otg_inst,
+		uint32_t *tmds_div,
+		uint32_t *dp_dto_int);
+void dccg401_set_dp_dto(
+		struct dccg *dccg,
+		const struct dp_dto_params *params);
+void dccg401_enable_symclk_se(struct dccg *dccg, uint32_t stream_enc_inst, uint32_t link_enc_inst);
+void dccg401_disable_symclk_se(struct dccg *dccg, uint32_t stream_enc_inst, uint32_t link_enc_inst);
+void dccg401_set_dtbclk_p_src(
+		struct dccg *dccg,
+		enum streamclk_source src,
+		uint32_t otg_inst);
+struct dccg *dccg401_create(
+	struct dc_context *ctx,
+	const struct dccg_registers *regs,
+	const struct dccg_shift *dccg_shift,
+	const struct dccg_mask *dccg_mask);
+
+void dccg401_set_physymclk(
+	struct dccg *dccg,
+	int phy_inst,
+	enum physymclk_clock_source clk_src,
+	bool force_enable);
+
+#endif //__DCN401_DCCG_H__
diff --git a/drivers/gpu/drm/amd/display/dc/dce/Makefile b/drivers/gpu/drm/amd/display/dc/dce/Makefile
new file mode 100644
index 000000000000..986e0e7abbc2
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/Makefile
@@ -0,0 +1,37 @@
+#
+# Copyright 2017 Advanced Micro Devices, Inc.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+#
+# Makefile for common 'dce' logic
+# HW object file under this folder follow similar pattern for HW programming
+#   - register offset and/or shift + mask stored in the dec_hw struct
+#   - register programming through common macros that look up register 
+#     offset/shift/mask stored in dce_hw struct
+
+DCE = dce_audio.o dce_stream_encoder.o dce_link_encoder.o \
+dce_mem_input.o dce_clock_source.o dce_scl_filters.o dce_transform.o \
+dce_opp.o dce_dmcu.o dce_abm.o dce_ipp.o dce_aux.o \
+dce_i2c.o dce_i2c_hw.o dce_i2c_sw.o dmub_psr.o dmub_abm.o dmub_abm_lcd.o dce_panel_cntl.o \
+dmub_hw_lock_mgr.o dmub_outbox.o dmub_replay.o
+
+AMD_DAL_DCE = $(addprefix $(AMDDALPATH)/dc/dce/,$(DCE))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCE)
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_abm.c b/drivers/gpu/drm/amd/display/dc/dce/dce_abm.c
new file mode 100644
index 000000000000..a6006776333d
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_abm.c
@@ -0,0 +1,306 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include <linux/slab.h>
+
+#include "dce_abm.h"
+#include "dm_services.h"
+#include "reg_helper.h"
+#include "fixed31_32.h"
+#include "dc.h"
+
+#include "atom.h"
+
+
+#define TO_DCE_ABM(abm)\
+	container_of(abm, struct dce_abm, base)
+
+#define REG(reg) \
+	(abm_dce->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	abm_dce->abm_shift->field_name, abm_dce->abm_mask->field_name
+
+#define DC_LOGGER \
+	abm->ctx->logger
+#define CTX \
+	abm_dce->base.ctx
+
+#define MCP_ABM_LEVEL_SET 0x65
+#define MCP_ABM_PIPE_SET 0x66
+#define MCP_BL_SET 0x67
+
+#define MCP_DISABLE_ABM_IMMEDIATELY 255
+
+static bool dce_abm_set_pipe(struct abm *abm, uint32_t controller_id, uint32_t panel_inst)
+{
+	struct dce_abm *abm_dce = TO_DCE_ABM(abm);
+	uint32_t rampingBoundary = 0xFFFF;
+
+	if (abm->dmcu_is_running == false)
+		return true;
+
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0,
+			1, 80000);
+
+	/* set ramping boundary */
+	REG_WRITE(MASTER_COMM_DATA_REG1, rampingBoundary);
+
+	/* setDMCUParam_Pipe */
+	REG_UPDATE_2(MASTER_COMM_CMD_REG,
+			MASTER_COMM_CMD_REG_BYTE0, MCP_ABM_PIPE_SET,
+			MASTER_COMM_CMD_REG_BYTE1, controller_id);
+
+	/* notifyDMCUMsg */
+	REG_UPDATE(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 1);
+
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0,
+			1, 80000);
+
+	return true;
+}
+
+static void dmcu_set_backlight_level(
+	struct dce_abm *abm_dce,
+	uint32_t backlight_pwm_u16_16,
+	uint32_t frame_ramp,
+	uint32_t controller_id,
+	uint32_t panel_id)
+{
+	unsigned int backlight_8_bit = 0;
+	uint32_t s2;
+
+	if (backlight_pwm_u16_16 & 0x10000)
+		// Check for max backlight condition
+		backlight_8_bit = 0xFF;
+	else
+		// Take MSB of fractional part since backlight is not max
+		backlight_8_bit = (backlight_pwm_u16_16 >> 8) & 0xFF;
+
+	dce_abm_set_pipe(&abm_dce->base, controller_id, panel_id);
+
+	/* waitDMCUReadyForCmd */
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT,
+			0, 1, 80000);
+
+	/* setDMCUParam_BL */
+	REG_UPDATE(BL1_PWM_USER_LEVEL, BL1_PWM_USER_LEVEL, backlight_pwm_u16_16);
+
+	/* write ramp */
+	if (controller_id == 0)
+		frame_ramp = 0;
+	REG_WRITE(MASTER_COMM_DATA_REG1, frame_ramp);
+
+	/* setDMCUParam_Cmd */
+	REG_UPDATE(MASTER_COMM_CMD_REG, MASTER_COMM_CMD_REG_BYTE0, MCP_BL_SET);
+
+	/* notifyDMCUMsg */
+	REG_UPDATE(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 1);
+
+	/* UpdateRequestedBacklightLevel */
+	s2 = REG_READ(BIOS_SCRATCH_2);
+
+	s2 &= ~ATOM_S2_CURRENT_BL_LEVEL_MASK;
+	backlight_8_bit &= (ATOM_S2_CURRENT_BL_LEVEL_MASK >>
+				ATOM_S2_CURRENT_BL_LEVEL_SHIFT);
+	s2 |= (backlight_8_bit << ATOM_S2_CURRENT_BL_LEVEL_SHIFT);
+
+	REG_WRITE(BIOS_SCRATCH_2, s2);
+
+	/* waitDMCUReadyForCmd */
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT,
+			0, 1, 80000);
+}
+
+static void dce_abm_init(struct abm *abm, uint32_t backlight, uint32_t user_level)
+{
+	struct dce_abm *abm_dce = TO_DCE_ABM(abm);
+
+	REG_WRITE(DC_ABM1_HG_SAMPLE_RATE, 0x103);
+	REG_WRITE(DC_ABM1_HG_SAMPLE_RATE, 0x101);
+	REG_WRITE(DC_ABM1_LS_SAMPLE_RATE, 0x103);
+	REG_WRITE(DC_ABM1_LS_SAMPLE_RATE, 0x101);
+	REG_WRITE(BL1_PWM_BL_UPDATE_SAMPLE_RATE, 0x101);
+
+	REG_SET_3(DC_ABM1_HG_MISC_CTRL, 0,
+			ABM1_HG_NUM_OF_BINS_SEL, 0,
+			ABM1_HG_VMAX_SEL, 1,
+			ABM1_HG_BIN_BITWIDTH_SIZE_SEL, 0);
+
+	REG_SET_3(DC_ABM1_IPCSC_COEFF_SEL, 0,
+			ABM1_IPCSC_COEFF_SEL_R, 2,
+			ABM1_IPCSC_COEFF_SEL_G, 4,
+			ABM1_IPCSC_COEFF_SEL_B, 2);
+
+	REG_UPDATE(BL1_PWM_CURRENT_ABM_LEVEL,
+			BL1_PWM_CURRENT_ABM_LEVEL, backlight);
+
+	REG_UPDATE(BL1_PWM_TARGET_ABM_LEVEL,
+			BL1_PWM_TARGET_ABM_LEVEL, backlight);
+
+	REG_UPDATE(BL1_PWM_USER_LEVEL,
+			BL1_PWM_USER_LEVEL, user_level);
+
+	REG_UPDATE_2(DC_ABM1_LS_MIN_MAX_PIXEL_VALUE_THRES,
+			ABM1_LS_MIN_PIXEL_VALUE_THRES, 0,
+			ABM1_LS_MAX_PIXEL_VALUE_THRES, 1000);
+
+	REG_SET_3(DC_ABM1_HGLS_REG_READ_PROGRESS, 0,
+			ABM1_HG_REG_READ_MISSED_FRAME_CLEAR, 1,
+			ABM1_LS_REG_READ_MISSED_FRAME_CLEAR, 1,
+			ABM1_BL_REG_READ_MISSED_FRAME_CLEAR, 1);
+}
+
+static unsigned int dce_abm_get_current_backlight(struct abm *abm)
+{
+	struct dce_abm *abm_dce = TO_DCE_ABM(abm);
+	unsigned int backlight = REG_READ(BL1_PWM_CURRENT_ABM_LEVEL);
+
+	/* return backlight in hardware format which is unsigned 17 bits, with
+	 * 1 bit integer and 16 bit fractional
+	 */
+	return backlight;
+}
+
+static unsigned int dce_abm_get_target_backlight(struct abm *abm)
+{
+	struct dce_abm *abm_dce = TO_DCE_ABM(abm);
+	unsigned int backlight = REG_READ(BL1_PWM_TARGET_ABM_LEVEL);
+
+	/* return backlight in hardware format which is unsigned 17 bits, with
+	 * 1 bit integer and 16 bit fractional
+	 */
+	return backlight;
+}
+
+static bool dce_abm_set_level(struct abm *abm, uint32_t level)
+{
+	struct dce_abm *abm_dce = TO_DCE_ABM(abm);
+
+	if (abm->dmcu_is_running == false)
+		return true;
+
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0,
+			1, 80000);
+
+	/* setDMCUParam_ABMLevel */
+	REG_UPDATE_2(MASTER_COMM_CMD_REG,
+			MASTER_COMM_CMD_REG_BYTE0, MCP_ABM_LEVEL_SET,
+			MASTER_COMM_CMD_REG_BYTE2, level);
+
+	/* notifyDMCUMsg */
+	REG_UPDATE(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 1);
+
+	return true;
+}
+
+static bool dce_abm_immediate_disable(struct abm *abm, uint32_t panel_inst)
+{
+	if (abm->dmcu_is_running == false)
+		return true;
+
+	dce_abm_set_pipe(abm, MCP_DISABLE_ABM_IMMEDIATELY, panel_inst);
+
+	return true;
+}
+
+static bool dce_abm_set_backlight_level_pwm(
+		struct abm *abm,
+		unsigned int backlight_pwm_u16_16,
+		unsigned int frame_ramp,
+		unsigned int controller_id,
+		unsigned int panel_inst)
+{
+	struct dce_abm *abm_dce = TO_DCE_ABM(abm);
+
+	DC_LOG_BACKLIGHT("New Backlight level: %d (0x%X)\n",
+			backlight_pwm_u16_16, backlight_pwm_u16_16);
+
+	dmcu_set_backlight_level(abm_dce,
+			backlight_pwm_u16_16,
+			frame_ramp,
+			controller_id,
+			panel_inst);
+
+	return true;
+}
+
+static const struct abm_funcs dce_funcs = {
+	.abm_init = dce_abm_init,
+	.set_abm_level = dce_abm_set_level,
+	.set_pipe = dce_abm_set_pipe,
+	.set_backlight_level_pwm = dce_abm_set_backlight_level_pwm,
+	.get_current_backlight = dce_abm_get_current_backlight,
+	.get_target_backlight = dce_abm_get_target_backlight,
+	.init_abm_config = NULL,
+	.set_abm_immediate_disable = dce_abm_immediate_disable,
+};
+
+static void dce_abm_construct(
+	struct dce_abm *abm_dce,
+	struct dc_context *ctx,
+	const struct dce_abm_registers *regs,
+	const struct dce_abm_shift *abm_shift,
+	const struct dce_abm_mask *abm_mask)
+{
+	struct abm *base = &abm_dce->base;
+
+	base->ctx = ctx;
+	base->funcs = &dce_funcs;
+	base->dmcu_is_running = false;
+
+	abm_dce->regs = regs;
+	abm_dce->abm_shift = abm_shift;
+	abm_dce->abm_mask = abm_mask;
+}
+
+struct abm *dce_abm_create(
+	struct dc_context *ctx,
+	const struct dce_abm_registers *regs,
+	const struct dce_abm_shift *abm_shift,
+	const struct dce_abm_mask *abm_mask)
+{
+	struct dce_abm *abm_dce = kzalloc(sizeof(*abm_dce), GFP_ATOMIC);
+
+	if (abm_dce == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	dce_abm_construct(abm_dce, ctx, regs, abm_shift, abm_mask);
+
+	abm_dce->base.funcs = &dce_funcs;
+
+	return &abm_dce->base;
+}
+
+void dce_abm_destroy(struct abm **abm)
+{
+	struct dce_abm *abm_dce = TO_DCE_ABM(*abm);
+
+	kfree(abm_dce);
+	*abm = NULL;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_abm.h b/drivers/gpu/drm/amd/display/dc/dce/dce_abm.h
new file mode 100644
index 000000000000..3d819fc5654c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_abm.h
@@ -0,0 +1,392 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+#ifndef _DCE_ABM_H_
+#define _DCE_ABM_H_
+
+#include "abm.h"
+
+#define ABM_COMMON_REG_LIST_DCE_BASE() \
+	SR(MASTER_COMM_CNTL_REG), \
+	SR(MASTER_COMM_CMD_REG), \
+	SR(MASTER_COMM_DATA_REG1)
+
+#define ABM_DCE110_COMMON_REG_LIST() \
+	ABM_COMMON_REG_LIST_DCE_BASE(), \
+	SR(DC_ABM1_HG_SAMPLE_RATE), \
+	SR(DC_ABM1_LS_SAMPLE_RATE), \
+	SR(BL1_PWM_BL_UPDATE_SAMPLE_RATE), \
+	SR(DC_ABM1_HG_MISC_CTRL), \
+	SR(DC_ABM1_IPCSC_COEFF_SEL), \
+	SR(BL1_PWM_CURRENT_ABM_LEVEL), \
+	SR(BL1_PWM_TARGET_ABM_LEVEL), \
+	SR(BL1_PWM_USER_LEVEL), \
+	SR(DC_ABM1_LS_MIN_MAX_PIXEL_VALUE_THRES), \
+	SR(DC_ABM1_HGLS_REG_READ_PROGRESS), \
+	SR(DC_ABM1_ACE_OFFSET_SLOPE_0), \
+	SR(DC_ABM1_ACE_THRES_12), \
+	SR(BIOS_SCRATCH_2)
+
+#define ABM_DCN10_REG_LIST(id)\
+	ABM_COMMON_REG_LIST_DCE_BASE(), \
+	SRI(DC_ABM1_HG_SAMPLE_RATE, ABM, id), \
+	SRI(DC_ABM1_LS_SAMPLE_RATE, ABM, id), \
+	SRI(BL1_PWM_BL_UPDATE_SAMPLE_RATE, ABM, id), \
+	SRI(DC_ABM1_HG_MISC_CTRL, ABM, id), \
+	SRI(DC_ABM1_IPCSC_COEFF_SEL, ABM, id), \
+	SRI(BL1_PWM_CURRENT_ABM_LEVEL, ABM, id), \
+	SRI(BL1_PWM_TARGET_ABM_LEVEL, ABM, id), \
+	SRI(BL1_PWM_USER_LEVEL, ABM, id), \
+	SRI(DC_ABM1_LS_MIN_MAX_PIXEL_VALUE_THRES, ABM, id), \
+	SRI(DC_ABM1_HGLS_REG_READ_PROGRESS, ABM, id), \
+	SRI(DC_ABM1_ACE_OFFSET_SLOPE_0, ABM, id), \
+	SRI(DC_ABM1_ACE_THRES_12, ABM, id), \
+	NBIO_SR(BIOS_SCRATCH_2)
+
+#define ABM_DCN20_REG_LIST() \
+	ABM_COMMON_REG_LIST_DCE_BASE(), \
+	SR(DC_ABM1_HG_SAMPLE_RATE), \
+	SR(DC_ABM1_LS_SAMPLE_RATE), \
+	SR(BL1_PWM_BL_UPDATE_SAMPLE_RATE), \
+	SR(DC_ABM1_HG_MISC_CTRL), \
+	SR(DC_ABM1_IPCSC_COEFF_SEL), \
+	SR(BL1_PWM_CURRENT_ABM_LEVEL), \
+	SR(BL1_PWM_TARGET_ABM_LEVEL), \
+	SR(BL1_PWM_USER_LEVEL), \
+	SR(DC_ABM1_LS_MIN_MAX_PIXEL_VALUE_THRES), \
+	SR(DC_ABM1_HGLS_REG_READ_PROGRESS), \
+	SR(DC_ABM1_ACE_OFFSET_SLOPE_0), \
+	SR(DC_ABM1_ACE_THRES_12), \
+	NBIO_SR(BIOS_SCRATCH_2)
+
+#define ABM_DCN301_REG_LIST(id)\
+	ABM_COMMON_REG_LIST_DCE_BASE(), \
+	SRI(DC_ABM1_HG_SAMPLE_RATE, ABM, id), \
+	SRI(DC_ABM1_LS_SAMPLE_RATE, ABM, id), \
+	SRI(BL1_PWM_BL_UPDATE_SAMPLE_RATE, ABM, id), \
+	SRI(DC_ABM1_HG_MISC_CTRL, ABM, id), \
+	SRI(DC_ABM1_IPCSC_COEFF_SEL, ABM, id), \
+	SRI(BL1_PWM_CURRENT_ABM_LEVEL, ABM, id), \
+	SRI(BL1_PWM_TARGET_ABM_LEVEL, ABM, id), \
+	SRI(BL1_PWM_USER_LEVEL, ABM, id), \
+	SRI(DC_ABM1_LS_MIN_MAX_PIXEL_VALUE_THRES, ABM, id), \
+	SRI(DC_ABM1_HGLS_REG_READ_PROGRESS, ABM, id), \
+	NBIO_SR(BIOS_SCRATCH_2)
+
+#define ABM_DCN302_REG_LIST(id)\
+	ABM_COMMON_REG_LIST_DCE_BASE(), \
+	SRI(DC_ABM1_HG_SAMPLE_RATE, ABM, id), \
+	SRI(DC_ABM1_LS_SAMPLE_RATE, ABM, id), \
+	SRI(BL1_PWM_BL_UPDATE_SAMPLE_RATE, ABM, id), \
+	SRI(DC_ABM1_HG_MISC_CTRL, ABM, id), \
+	SRI(DC_ABM1_IPCSC_COEFF_SEL, ABM, id), \
+	SRI(BL1_PWM_CURRENT_ABM_LEVEL, ABM, id), \
+	SRI(BL1_PWM_TARGET_ABM_LEVEL, ABM, id), \
+	SRI(BL1_PWM_USER_LEVEL, ABM, id), \
+	SRI(DC_ABM1_LS_MIN_MAX_PIXEL_VALUE_THRES, ABM, id), \
+	SRI(DC_ABM1_HGLS_REG_READ_PROGRESS, ABM, id), \
+	SRI(DC_ABM1_ACE_OFFSET_SLOPE_0, ABM, id), \
+	SRI(DC_ABM1_ACE_THRES_12, ABM, id), \
+	NBIO_SR(BIOS_SCRATCH_2)
+
+#define ABM_DCN30_REG_LIST(id)\
+	ABM_COMMON_REG_LIST_DCE_BASE(), \
+	SRI(DC_ABM1_HG_SAMPLE_RATE, ABM, id), \
+	SRI(DC_ABM1_LS_SAMPLE_RATE, ABM, id), \
+	SRI(BL1_PWM_BL_UPDATE_SAMPLE_RATE, ABM, id), \
+	SRI(DC_ABM1_HG_MISC_CTRL, ABM, id), \
+	SRI(DC_ABM1_IPCSC_COEFF_SEL, ABM, id), \
+	SRI(BL1_PWM_CURRENT_ABM_LEVEL, ABM, id), \
+	SRI(BL1_PWM_TARGET_ABM_LEVEL, ABM, id), \
+	SRI(BL1_PWM_USER_LEVEL, ABM, id), \
+	SRI(DC_ABM1_LS_MIN_MAX_PIXEL_VALUE_THRES, ABM, id), \
+	SRI(DC_ABM1_HGLS_REG_READ_PROGRESS, ABM, id), \
+	SRI(DC_ABM1_ACE_OFFSET_SLOPE_0, ABM, id), \
+	SRI(DC_ABM1_ACE_THRES_12, ABM, id), \
+	NBIO_SR(BIOS_SCRATCH_2)
+
+#define ABM_SF(reg_name, field_name, post_fix)\
+	.field_name = reg_name ## __ ## field_name ## post_fix
+
+#define ABM_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(mask_sh) \
+	ABM_SF(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, mask_sh), \
+	ABM_SF(MASTER_COMM_CMD_REG, MASTER_COMM_CMD_REG_BYTE0, mask_sh), \
+	ABM_SF(MASTER_COMM_CMD_REG, MASTER_COMM_CMD_REG_BYTE1, mask_sh), \
+	ABM_SF(MASTER_COMM_CMD_REG, MASTER_COMM_CMD_REG_BYTE2, mask_sh)
+
+#define ABM_MASK_SH_LIST_DCE110(mask_sh) \
+	ABM_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(mask_sh), \
+	ABM_SF(DC_ABM1_HG_MISC_CTRL, \
+			ABM1_HG_NUM_OF_BINS_SEL, mask_sh), \
+	ABM_SF(DC_ABM1_HG_MISC_CTRL, \
+			ABM1_HG_VMAX_SEL, mask_sh), \
+	ABM_SF(DC_ABM1_HG_MISC_CTRL, \
+			ABM1_HG_BIN_BITWIDTH_SIZE_SEL, mask_sh), \
+	ABM_SF(DC_ABM1_IPCSC_COEFF_SEL, \
+			ABM1_IPCSC_COEFF_SEL_R, mask_sh), \
+	ABM_SF(DC_ABM1_IPCSC_COEFF_SEL, \
+			ABM1_IPCSC_COEFF_SEL_G, mask_sh), \
+	ABM_SF(DC_ABM1_IPCSC_COEFF_SEL, \
+			ABM1_IPCSC_COEFF_SEL_B, mask_sh), \
+	ABM_SF(BL1_PWM_CURRENT_ABM_LEVEL, \
+			BL1_PWM_CURRENT_ABM_LEVEL, mask_sh), \
+	ABM_SF(BL1_PWM_TARGET_ABM_LEVEL, \
+			BL1_PWM_TARGET_ABM_LEVEL, mask_sh), \
+	ABM_SF(BL1_PWM_USER_LEVEL, \
+			BL1_PWM_USER_LEVEL, mask_sh), \
+	ABM_SF(DC_ABM1_LS_MIN_MAX_PIXEL_VALUE_THRES, \
+			ABM1_LS_MIN_PIXEL_VALUE_THRES, mask_sh), \
+	ABM_SF(DC_ABM1_LS_MIN_MAX_PIXEL_VALUE_THRES, \
+			ABM1_LS_MAX_PIXEL_VALUE_THRES, mask_sh), \
+	ABM_SF(DC_ABM1_HGLS_REG_READ_PROGRESS, \
+			ABM1_HG_REG_READ_MISSED_FRAME_CLEAR, mask_sh), \
+	ABM_SF(DC_ABM1_HGLS_REG_READ_PROGRESS, \
+			ABM1_LS_REG_READ_MISSED_FRAME_CLEAR, mask_sh), \
+	ABM_SF(DC_ABM1_HGLS_REG_READ_PROGRESS, \
+			ABM1_BL_REG_READ_MISSED_FRAME_CLEAR, mask_sh)
+
+#define ABM_MASK_SH_LIST_DCN10_COMMON(mask_sh) \
+	ABM_SF(ABM0_DC_ABM1_HG_MISC_CTRL, \
+			ABM1_HG_NUM_OF_BINS_SEL, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_HG_MISC_CTRL, \
+			ABM1_HG_VMAX_SEL, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_HG_MISC_CTRL, \
+			ABM1_HG_BIN_BITWIDTH_SIZE_SEL, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_IPCSC_COEFF_SEL, \
+			ABM1_IPCSC_COEFF_SEL_R, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_IPCSC_COEFF_SEL, \
+			ABM1_IPCSC_COEFF_SEL_G, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_IPCSC_COEFF_SEL, \
+			ABM1_IPCSC_COEFF_SEL_B, mask_sh), \
+	ABM_SF(ABM0_BL1_PWM_CURRENT_ABM_LEVEL, \
+			BL1_PWM_CURRENT_ABM_LEVEL, mask_sh), \
+	ABM_SF(ABM0_BL1_PWM_TARGET_ABM_LEVEL, \
+			BL1_PWM_TARGET_ABM_LEVEL, mask_sh), \
+	ABM_SF(ABM0_BL1_PWM_USER_LEVEL, \
+			BL1_PWM_USER_LEVEL, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_LS_MIN_MAX_PIXEL_VALUE_THRES, \
+			ABM1_LS_MIN_PIXEL_VALUE_THRES, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_LS_MIN_MAX_PIXEL_VALUE_THRES, \
+			ABM1_LS_MAX_PIXEL_VALUE_THRES, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_HGLS_REG_READ_PROGRESS, \
+			ABM1_HG_REG_READ_MISSED_FRAME_CLEAR, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_HGLS_REG_READ_PROGRESS, \
+			ABM1_LS_REG_READ_MISSED_FRAME_CLEAR, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_HGLS_REG_READ_PROGRESS, \
+			ABM1_BL_REG_READ_MISSED_FRAME_CLEAR, mask_sh)
+
+#define ABM_MASK_SH_LIST_DCN10(mask_sh) \
+	ABM_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(mask_sh), \
+	ABM_MASK_SH_LIST_DCN10_COMMON(mask_sh)
+
+#define ABM_MASK_SH_LIST_DCN20(mask_sh) ABM_MASK_SH_LIST_DCE110(mask_sh)
+#define ABM_MASK_SH_LIST_DCN30(mask_sh) ABM_MASK_SH_LIST_DCN10(mask_sh)
+#define ABM_MASK_SH_LIST_DCN35(mask_sh) ABM_MASK_SH_LIST_DCN10_COMMON(mask_sh)
+
+#define ABM_MASK_SH_LIST_DCN32(mask_sh) \
+	ABM_SF(ABM0_DC_ABM1_HG_MISC_CTRL, \
+			ABM1_HG_NUM_OF_BINS_SEL, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_HG_MISC_CTRL, \
+			ABM1_HG_VMAX_SEL, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_HG_MISC_CTRL, \
+			ABM1_HG_BIN_BITWIDTH_SIZE_SEL, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_IPCSC_COEFF_SEL, \
+			ABM1_IPCSC_COEFF_SEL_R, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_IPCSC_COEFF_SEL, \
+			ABM1_IPCSC_COEFF_SEL_G, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_IPCSC_COEFF_SEL, \
+			ABM1_IPCSC_COEFF_SEL_B, mask_sh), \
+	ABM_SF(ABM0_BL1_PWM_CURRENT_ABM_LEVEL, \
+			BL1_PWM_CURRENT_ABM_LEVEL, mask_sh), \
+	ABM_SF(ABM0_BL1_PWM_TARGET_ABM_LEVEL, \
+			BL1_PWM_TARGET_ABM_LEVEL, mask_sh), \
+	ABM_SF(ABM0_BL1_PWM_USER_LEVEL, \
+			BL1_PWM_USER_LEVEL, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_LS_MIN_MAX_PIXEL_VALUE_THRES, \
+			ABM1_LS_MIN_PIXEL_VALUE_THRES, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_LS_MIN_MAX_PIXEL_VALUE_THRES, \
+			ABM1_LS_MAX_PIXEL_VALUE_THRES, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_HGLS_REG_READ_PROGRESS, \
+			ABM1_HG_REG_READ_MISSED_FRAME_CLEAR, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_HGLS_REG_READ_PROGRESS, \
+			ABM1_LS_REG_READ_MISSED_FRAME_CLEAR, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_HGLS_REG_READ_PROGRESS, \
+			ABM1_BL_REG_READ_MISSED_FRAME_CLEAR, mask_sh)
+
+#define ABM_MASK_SH_LIST_DCN401(mask_sh) \
+	ABM_SF(ABM0_DC_ABM1_HG_MISC_CTRL, \
+			ABM1_HG_NUM_OF_BINS_SEL, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_HG_MISC_CTRL, \
+			ABM1_HG_VMAX_SEL, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_HG_MISC_CTRL, \
+			ABM1_HG_BIN_BITWIDTH_SIZE_SEL, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_IPCSC_COEFF_SEL, \
+			ABM1_IPCSC_COEFF_SEL_R, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_IPCSC_COEFF_SEL, \
+			ABM1_IPCSC_COEFF_SEL_G, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_IPCSC_COEFF_SEL, \
+			ABM1_IPCSC_COEFF_SEL_B, mask_sh), \
+	ABM_SF(ABM0_BL1_PWM_CURRENT_ABM_LEVEL, \
+			BL1_PWM_CURRENT_ABM_LEVEL, mask_sh), \
+	ABM_SF(ABM0_BL1_PWM_TARGET_ABM_LEVEL, \
+			BL1_PWM_TARGET_ABM_LEVEL, mask_sh), \
+	ABM_SF(ABM0_BL1_PWM_USER_LEVEL, \
+			BL1_PWM_USER_LEVEL, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_LS_MIN_MAX_PIXEL_VALUE_THRES, \
+			ABM1_LS_MIN_PIXEL_VALUE_THRES, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_LS_MIN_MAX_PIXEL_VALUE_THRES, \
+			ABM1_LS_MAX_PIXEL_VALUE_THRES, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_HGLS_REG_READ_PROGRESS, \
+			ABM1_HG_REG_READ_MISSED_FRAME_CLEAR, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_HGLS_REG_READ_PROGRESS, \
+			ABM1_LS_REG_READ_MISSED_FRAME_CLEAR, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_HGLS_REG_READ_PROGRESS, \
+			ABM1_BL_REG_READ_MISSED_FRAME_CLEAR, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_ACE_OFFSET_SLOPE_DATA, \
+			ABM1_ACE_SLOPE_DATA, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_ACE_OFFSET_SLOPE_DATA, \
+			ABM1_ACE_OFFSET_DATA, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_ACE_PWL_CNTL, \
+			ABM1_ACE_OFFSET_SLOPE_INDEX, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_ACE_PWL_CNTL, \
+			ABM1_ACE_THRES_INDEX, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_ACE_PWL_CNTL, \
+			ABM1_ACE_IGNORE_MASTER_LOCK_EN, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_ACE_PWL_CNTL, \
+			ABM1_ACE_READBACK_DB_REG_VALUE_EN, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_ACE_PWL_CNTL, \
+			ABM1_ACE_DBUF_REG_UPDATE_PENDING, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_ACE_PWL_CNTL, \
+			ABM1_ACE_LOCK, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_ACE_THRES_DATA, \
+			ABM1_ACE_THRES_DATA_1, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_ACE_THRES_DATA, \
+			ABM1_ACE_THRES_DATA_2, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_HG_RESULT_DATA, \
+			ABM1_HG_RESULT_DATA, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_HG_RESULT_INDEX, \
+			ABM1_HG_RESULT_INDEX, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_HG_BIN_33_40_SHIFT_INDEX, \
+			ABM1_HG_BIN_33_40_SHIFT_INDEX, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_HG_BIN_33_64_SHIFT_FLAG, \
+			ABM1_HG_BIN_33_64_SHIFT_FLAG, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_HG_BIN_41_48_SHIFT_INDEX, \
+			ABM1_HG_BIN_41_48_SHIFT_INDEX, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_HG_BIN_49_56_SHIFT_INDEX, \
+			ABM1_HG_BIN_49_56_SHIFT_INDEX, mask_sh), \
+	ABM_SF(ABM0_DC_ABM1_HG_BIN_57_64_SHIFT_INDEX, \
+			ABM1_HG_BIN_57_64_SHIFT_INDEX, mask_sh)
+
+#define ABM_REG_FIELD_LIST(type) \
+	type ABM1_HG_NUM_OF_BINS_SEL; \
+	type ABM1_HG_VMAX_SEL; \
+	type ABM1_HG_BIN_BITWIDTH_SIZE_SEL; \
+	type ABM1_IPCSC_COEFF_SEL_R; \
+	type ABM1_IPCSC_COEFF_SEL_G; \
+	type ABM1_IPCSC_COEFF_SEL_B; \
+	type BL1_PWM_CURRENT_ABM_LEVEL; \
+	type BL1_PWM_TARGET_ABM_LEVEL; \
+	type BL1_PWM_USER_LEVEL; \
+	type ABM1_LS_MIN_PIXEL_VALUE_THRES; \
+	type ABM1_LS_MAX_PIXEL_VALUE_THRES; \
+	type ABM1_HG_REG_READ_MISSED_FRAME_CLEAR; \
+	type ABM1_LS_REG_READ_MISSED_FRAME_CLEAR; \
+	type ABM1_BL_REG_READ_MISSED_FRAME_CLEAR; \
+	type MASTER_COMM_INTERRUPT; \
+	type MASTER_COMM_CMD_REG_BYTE0; \
+	type MASTER_COMM_CMD_REG_BYTE1; \
+	type MASTER_COMM_CMD_REG_BYTE2; \
+	type ABM1_HG_BIN_33_40_SHIFT_INDEX; \
+	type ABM1_HG_BIN_33_64_SHIFT_FLAG; \
+	type ABM1_HG_BIN_41_48_SHIFT_INDEX; \
+	type ABM1_HG_BIN_49_56_SHIFT_INDEX; \
+	type ABM1_HG_BIN_57_64_SHIFT_INDEX; \
+	type ABM1_HG_RESULT_DATA; \
+	type ABM1_HG_RESULT_INDEX; \
+	type ABM1_ACE_SLOPE_DATA; \
+	type ABM1_ACE_OFFSET_DATA; \
+	type ABM1_ACE_OFFSET_SLOPE_INDEX; \
+	type ABM1_ACE_THRES_INDEX; \
+	type ABM1_ACE_IGNORE_MASTER_LOCK_EN; \
+	type ABM1_ACE_READBACK_DB_REG_VALUE_EN; \
+	type ABM1_ACE_DBUF_REG_UPDATE_PENDING; \
+	type ABM1_ACE_LOCK; \
+	type ABM1_ACE_THRES_DATA_1; \
+	type ABM1_ACE_THRES_DATA_2
+
+struct dce_abm_shift {
+	ABM_REG_FIELD_LIST(uint8_t);
+};
+
+struct dce_abm_mask {
+	ABM_REG_FIELD_LIST(uint32_t);
+};
+
+struct dce_abm_registers {
+	uint32_t DC_ABM1_HG_SAMPLE_RATE;
+	uint32_t DC_ABM1_LS_SAMPLE_RATE;
+	uint32_t BL1_PWM_BL_UPDATE_SAMPLE_RATE;
+	uint32_t DC_ABM1_HG_MISC_CTRL;
+	uint32_t DC_ABM1_IPCSC_COEFF_SEL;
+	uint32_t BL1_PWM_CURRENT_ABM_LEVEL;
+	uint32_t BL1_PWM_TARGET_ABM_LEVEL;
+	uint32_t BL1_PWM_USER_LEVEL;
+	uint32_t DC_ABM1_LS_MIN_MAX_PIXEL_VALUE_THRES;
+	uint32_t DC_ABM1_HGLS_REG_READ_PROGRESS;
+	uint32_t DC_ABM1_ACE_OFFSET_SLOPE_0;
+	uint32_t DC_ABM1_ACE_OFFSET_SLOPE_DATA;
+	uint32_t DC_ABM1_ACE_PWL_CNTL;
+	uint32_t DC_ABM1_HG_BIN_33_40_SHIFT_INDEX;
+	uint32_t DC_ABM1_HG_BIN_33_64_SHIFT_FLAG;
+	uint32_t DC_ABM1_HG_BIN_41_48_SHIFT_INDEX;
+	uint32_t DC_ABM1_HG_BIN_49_56_SHIFT_INDEX;
+	uint32_t DC_ABM1_HG_BIN_57_64_SHIFT_INDEX;
+	uint32_t DC_ABM1_HG_RESULT_DATA;
+	uint32_t DC_ABM1_HG_RESULT_INDEX;
+	uint32_t DC_ABM1_ACE_THRES_DATA;
+	uint32_t DC_ABM1_ACE_THRES_12;
+	uint32_t MASTER_COMM_CNTL_REG;
+	uint32_t MASTER_COMM_CMD_REG;
+	uint32_t MASTER_COMM_DATA_REG1;
+	uint32_t BIOS_SCRATCH_2;
+};
+
+struct dce_abm {
+	struct abm base;
+	const struct dce_abm_registers *regs;
+	const struct dce_abm_shift *abm_shift;
+	const struct dce_abm_mask *abm_mask;
+};
+
+struct abm *dce_abm_create(
+	struct dc_context *ctx,
+	const struct dce_abm_registers *regs,
+	const struct dce_abm_shift *abm_shift,
+	const struct dce_abm_mask *abm_mask);
+
+void dce_abm_destroy(struct abm **abm);
+
+#endif /* _DCE_ABM_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_audio.c b/drivers/gpu/drm/amd/display/dc/dce/dce_audio.c
new file mode 100644
index 000000000000..eeed840073fe
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_audio.c
@@ -0,0 +1,1375 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+#include "dce_audio.h"
+#include "dce/dce_11_0_d.h"
+#include "dce/dce_11_0_sh_mask.h"
+
+#define DCE_AUD(audio)\
+	container_of(audio, struct dce_audio, base)
+
+#define CTX \
+	aud->base.ctx
+
+#define DC_LOGGER_INIT()
+
+#define REG(reg)\
+	(aud->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	aud->shifts->field_name, aud->masks->field_name
+
+#define IX_REG(reg)\
+	ix ## reg
+
+#define AZ_REG_READ(reg_name) \
+		read_indirect_azalia_reg(audio, IX_REG(reg_name))
+
+#define AZ_REG_WRITE(reg_name, value) \
+		write_indirect_azalia_reg(audio, IX_REG(reg_name), value)
+
+static void write_indirect_azalia_reg(struct audio *audio,
+	uint32_t reg_index,
+	uint32_t reg_data)
+{
+	struct dce_audio *aud = DCE_AUD(audio);
+
+	/* AZALIA_F0_CODEC_ENDPOINT_INDEX  endpoint index  */
+	REG_SET(AZALIA_F0_CODEC_ENDPOINT_INDEX, 0,
+			AZALIA_ENDPOINT_REG_INDEX, reg_index);
+
+	/* AZALIA_F0_CODEC_ENDPOINT_DATA  endpoint data  */
+	REG_SET(AZALIA_F0_CODEC_ENDPOINT_DATA, 0,
+			AZALIA_ENDPOINT_REG_DATA, reg_data);
+}
+
+static uint32_t read_indirect_azalia_reg(struct audio *audio, uint32_t reg_index)
+{
+	struct dce_audio *aud = DCE_AUD(audio);
+
+	uint32_t value = 0;
+
+	/* AZALIA_F0_CODEC_ENDPOINT_INDEX  endpoint index  */
+	REG_SET(AZALIA_F0_CODEC_ENDPOINT_INDEX, 0,
+			AZALIA_ENDPOINT_REG_INDEX, reg_index);
+
+	/* AZALIA_F0_CODEC_ENDPOINT_DATA  endpoint data  */
+	value = REG_READ(AZALIA_F0_CODEC_ENDPOINT_DATA);
+
+	return value;
+}
+
+static bool is_audio_format_supported(
+	const struct audio_info *audio_info,
+	enum audio_format_code audio_format_code,
+	uint32_t *format_index)
+{
+	uint32_t index;
+	uint32_t max_channe_index = 0;
+	bool found = false;
+
+	if (audio_info == NULL)
+		return found;
+
+	/* pass through whole array */
+	for (index = 0; index < audio_info->mode_count; index++) {
+		if (audio_info->modes[index].format_code == audio_format_code) {
+			if (found) {
+				/* format has multiply entries, choose one with
+				 *  highst number of channels */
+				if (audio_info->modes[index].channel_count >
+		audio_info->modes[max_channe_index].channel_count) {
+					max_channe_index = index;
+				}
+			} else {
+				/* format found, save it's index */
+				found = true;
+				max_channe_index = index;
+			}
+		}
+	}
+
+	/* return index */
+	if (found && format_index != NULL)
+		*format_index = max_channe_index;
+
+	return found;
+}
+
+/*For HDMI, calculate if specified sample rates can fit into a given timing */
+static void check_audio_bandwidth_hdmi(
+	const struct audio_crtc_info *crtc_info,
+	uint32_t channel_count,
+	union audio_sample_rates *sample_rates)
+{
+	uint32_t samples;
+	uint32_t  h_blank;
+	bool limit_freq_to_48_khz = false;
+	bool limit_freq_to_88_2_khz = false;
+	bool limit_freq_to_96_khz = false;
+	bool limit_freq_to_174_4_khz = false;
+	if (!crtc_info)
+		return;
+
+	/* For two channels supported return whatever sink support,unmodified*/
+	if (channel_count > 2) {
+
+		/* Based on HDMI spec 1.3 Table 7.5 */
+		if ((crtc_info->requested_pixel_clock_100Hz <= 270000) &&
+		(crtc_info->v_active <= 576) &&
+		!(crtc_info->interlaced) &&
+		!(crtc_info->pixel_repetition == 2 ||
+		crtc_info->pixel_repetition == 4)) {
+			limit_freq_to_48_khz = true;
+
+		} else if ((crtc_info->requested_pixel_clock_100Hz <= 270000) &&
+				(crtc_info->v_active <= 576) &&
+				(crtc_info->interlaced) &&
+				(crtc_info->pixel_repetition == 2)) {
+			limit_freq_to_88_2_khz = true;
+
+		} else if ((crtc_info->requested_pixel_clock_100Hz <= 540000) &&
+				(crtc_info->v_active <= 576) &&
+				!(crtc_info->interlaced)) {
+			limit_freq_to_174_4_khz = true;
+		}
+	}
+
+	/* Also do some calculation for the available Audio Bandwidth for the
+	 * 8 ch (i.e. for the Layout 1 => ch > 2)
+	 */
+	h_blank = crtc_info->h_total - crtc_info->h_active;
+
+	if (crtc_info->pixel_repetition)
+		h_blank *= crtc_info->pixel_repetition;
+
+	/*based on HDMI spec 1.3 Table 7.5 */
+	h_blank -= 58;
+	/*for Control Period */
+	h_blank -= 16;
+
+	samples = h_blank * 10;
+	/* Number of Audio Packets (multiplied by 10) per Line (for 8 ch number
+	 * of Audio samples per line multiplied by 10 - Layout 1)
+	 */
+	samples /= 32;
+	samples *= crtc_info->v_active;
+	/*Number of samples multiplied by 10, per second */
+	samples *= crtc_info->refresh_rate;
+	/*Number of Audio samples per second */
+	samples /= 10;
+
+	/* @todo do it after deep color is implemented
+	 * 8xx - deep color bandwidth scaling
+	 * Extra bandwidth is avaliable in deep color b/c link runs faster than
+	 * pixel rate. This has the effect of allowing more tmds characters to
+	 * be transmitted during blank
+	 */
+
+	switch (crtc_info->color_depth) {
+	case COLOR_DEPTH_888:
+		samples *= 4;
+		break;
+	case COLOR_DEPTH_101010:
+		samples *= 5;
+		break;
+	case COLOR_DEPTH_121212:
+		samples *= 6;
+		break;
+	default:
+		samples *= 4;
+		break;
+	}
+
+	samples /= 4;
+
+	/*check limitation*/
+	if (samples < 88200)
+		limit_freq_to_48_khz = true;
+	else if (samples < 96000)
+		limit_freq_to_88_2_khz = true;
+	else if (samples < 176400)
+		limit_freq_to_96_khz = true;
+	else if (samples < 192000)
+		limit_freq_to_174_4_khz = true;
+
+	if (sample_rates != NULL) {
+		/* limit frequencies */
+		if (limit_freq_to_174_4_khz)
+			sample_rates->rate.RATE_192 = 0;
+
+		if (limit_freq_to_96_khz) {
+			sample_rates->rate.RATE_192 = 0;
+			sample_rates->rate.RATE_176_4 = 0;
+		}
+		if (limit_freq_to_88_2_khz) {
+			sample_rates->rate.RATE_192 = 0;
+			sample_rates->rate.RATE_176_4 = 0;
+			sample_rates->rate.RATE_96 = 0;
+		}
+		if (limit_freq_to_48_khz) {
+			sample_rates->rate.RATE_192 = 0;
+			sample_rates->rate.RATE_176_4 = 0;
+			sample_rates->rate.RATE_96 = 0;
+			sample_rates->rate.RATE_88_2 = 0;
+		}
+	}
+}
+static struct fixed31_32 get_link_symbol_clk_freq_mhz(enum dc_link_rate link_rate)
+{
+	switch (link_rate) {
+	case LINK_RATE_LOW:
+		return dc_fixpt_from_int(162); /* 162 MHz */
+	case LINK_RATE_HIGH:
+		return dc_fixpt_from_int(270); /* 270 MHz */
+	case LINK_RATE_HIGH2:
+		return dc_fixpt_from_int(540); /* 540 MHz */
+	case LINK_RATE_HIGH3:
+		return dc_fixpt_from_int(810); /* 810 MHz */
+	case LINK_RATE_UHBR10:
+		return dc_fixpt_from_fraction(3125, 10); /* 312.5 MHz */
+	case LINK_RATE_UHBR13_5:
+		return dc_fixpt_from_fraction(421875, 1000); /* 421.875 MHz */
+	case LINK_RATE_UHBR20:
+		return dc_fixpt_from_int(625); /* 625 MHz */
+	default:
+		/* Unexpected case, this requires debug if encountered. */
+		ASSERT(0);
+		return dc_fixpt_from_int(0);
+	}
+}
+
+struct dp_audio_layout_config {
+	uint8_t layouts_per_sample_denom;
+	uint8_t symbols_per_layout;
+	uint8_t max_layouts_per_audio_sdp;
+};
+
+static void get_audio_layout_config(
+	uint32_t channel_count,
+	enum dp_link_encoding encoding,
+	struct dp_audio_layout_config *output)
+{
+	/* Assuming L-PCM audio. Current implementation uses max 1 layout per SDP,
+	 * with each layout being the same size (8ch layout).
+	 */
+	if (encoding == DP_8b_10b_ENCODING) {
+		if (channel_count == 2) {
+			output->layouts_per_sample_denom = 4;
+			output->symbols_per_layout = 40;
+			output->max_layouts_per_audio_sdp = 1;
+		} else if (channel_count == 8 || channel_count == 6) {
+			output->layouts_per_sample_denom = 1;
+			output->symbols_per_layout = 40;
+			output->max_layouts_per_audio_sdp = 1;
+		}
+	} else if (encoding == DP_128b_132b_ENCODING) {
+		if (channel_count == 2) {
+			output->layouts_per_sample_denom = 4;
+			output->symbols_per_layout = 10;
+			output->max_layouts_per_audio_sdp = 1;
+		} else if (channel_count == 8 || channel_count == 6) {
+			output->layouts_per_sample_denom = 1;
+			output->symbols_per_layout = 10;
+			output->max_layouts_per_audio_sdp = 1;
+		}
+	}
+}
+
+static uint32_t get_av_stream_map_lane_count(
+	enum dp_link_encoding encoding,
+	enum dc_lane_count lane_count,
+	bool is_mst)
+{
+	uint32_t av_stream_map_lane_count = 0;
+
+	if (encoding == DP_8b_10b_ENCODING) {
+		if (!is_mst)
+			av_stream_map_lane_count = lane_count;
+		else
+			av_stream_map_lane_count = 4;
+	} else if (encoding == DP_128b_132b_ENCODING) {
+		av_stream_map_lane_count = 4;
+	}
+
+	ASSERT(av_stream_map_lane_count != 0);
+
+	return av_stream_map_lane_count;
+}
+
+static uint32_t get_audio_sdp_overhead(
+	enum dp_link_encoding encoding,
+	enum dc_lane_count lane_count,
+	bool is_mst)
+{
+	uint32_t audio_sdp_overhead = 0;
+
+	if (encoding == DP_8b_10b_ENCODING) {
+		if (is_mst)
+			audio_sdp_overhead = 16; /* 4 * 2 + 8 */
+		else
+			audio_sdp_overhead = lane_count * 2 + 8;
+	} else if (encoding == DP_128b_132b_ENCODING) {
+		audio_sdp_overhead = 10; /* 4 x 2.5 */
+	}
+
+	ASSERT(audio_sdp_overhead != 0);
+
+	return audio_sdp_overhead;
+}
+
+static uint32_t calculate_required_audio_bw_in_symbols(
+	const struct audio_crtc_info *crtc_info,
+	const struct dp_audio_layout_config *layout_config,
+	uint32_t channel_count,
+	uint32_t sample_rate_hz,
+	uint32_t av_stream_map_lane_count,
+	uint32_t audio_sdp_overhead)
+{
+	/* DP spec recommends between 1.05 to 1.1 safety margin to prevent sample under-run */
+	struct fixed31_32 audio_sdp_margin = dc_fixpt_from_fraction(110, 100);
+	struct fixed31_32 horizontal_line_freq_khz = dc_fixpt_from_fraction(
+			crtc_info->requested_pixel_clock_100Hz, (long long)crtc_info->h_total * 10);
+	struct fixed31_32 samples_per_line;
+	struct fixed31_32 layouts_per_line;
+	struct fixed31_32 symbols_per_sdp_max_layout;
+	struct fixed31_32 remainder;
+	uint32_t num_sdp_with_max_layouts;
+	uint32_t required_symbols_per_hblank;
+
+	samples_per_line = dc_fixpt_from_fraction(sample_rate_hz, 1000);
+	samples_per_line = dc_fixpt_div(samples_per_line, horizontal_line_freq_khz);
+	layouts_per_line = dc_fixpt_div_int(samples_per_line, layout_config->layouts_per_sample_denom);
+
+	num_sdp_with_max_layouts = dc_fixpt_floor(
+			dc_fixpt_div_int(layouts_per_line, layout_config->max_layouts_per_audio_sdp));
+	symbols_per_sdp_max_layout = dc_fixpt_from_int(
+			layout_config->max_layouts_per_audio_sdp * layout_config->symbols_per_layout);
+	symbols_per_sdp_max_layout = dc_fixpt_add_int(symbols_per_sdp_max_layout, audio_sdp_overhead);
+	symbols_per_sdp_max_layout = dc_fixpt_mul(symbols_per_sdp_max_layout, audio_sdp_margin);
+	required_symbols_per_hblank = num_sdp_with_max_layouts;
+	required_symbols_per_hblank *= ((dc_fixpt_ceil(symbols_per_sdp_max_layout) + av_stream_map_lane_count) /
+			av_stream_map_lane_count) *	av_stream_map_lane_count;
+
+	if (num_sdp_with_max_layouts !=	dc_fixpt_ceil(
+			dc_fixpt_div_int(layouts_per_line, layout_config->max_layouts_per_audio_sdp))) {
+		remainder = dc_fixpt_sub_int(layouts_per_line,
+				num_sdp_with_max_layouts * layout_config->max_layouts_per_audio_sdp);
+		remainder = dc_fixpt_mul_int(remainder, layout_config->symbols_per_layout);
+		remainder = dc_fixpt_add_int(remainder, audio_sdp_overhead);
+		remainder = dc_fixpt_mul(remainder, audio_sdp_margin);
+		required_symbols_per_hblank += ((dc_fixpt_ceil(remainder) + av_stream_map_lane_count) /
+				av_stream_map_lane_count) * av_stream_map_lane_count;
+	}
+
+	return required_symbols_per_hblank;
+}
+
+/* Current calculation only applicable for 8b/10b MST and 128b/132b SST/MST.
+ */
+static uint32_t calculate_available_hblank_bw_in_symbols(
+	const struct audio_crtc_info *crtc_info,
+	const struct audio_dp_link_info *dp_link_info)
+{
+	uint64_t hblank = crtc_info->h_total - crtc_info->h_active;
+	struct fixed31_32 hblank_time_msec =
+			dc_fixpt_from_fraction(hblank * 10, crtc_info->requested_pixel_clock_100Hz);
+	struct fixed31_32 lsclkfreq_mhz =
+			get_link_symbol_clk_freq_mhz(dp_link_info->link_rate);
+	struct fixed31_32 average_stream_sym_bw_frac;
+	struct fixed31_32 peak_stream_bw_kbps;
+	struct fixed31_32 bits_per_pixel;
+	struct fixed31_32 link_bw_kbps;
+	struct fixed31_32 available_stream_sym_count;
+	uint32_t available_hblank_bw = 0; /* in stream symbols */
+
+	if (crtc_info->dsc_bits_per_pixel) {
+		bits_per_pixel = dc_fixpt_from_fraction(crtc_info->dsc_bits_per_pixel, 16);
+	} else {
+		switch (crtc_info->color_depth) {
+		case COLOR_DEPTH_666:
+			bits_per_pixel = dc_fixpt_from_int(6);
+			break;
+		case COLOR_DEPTH_888:
+			bits_per_pixel = dc_fixpt_from_int(8);
+			break;
+		case COLOR_DEPTH_101010:
+			bits_per_pixel = dc_fixpt_from_int(10);
+			break;
+		case COLOR_DEPTH_121212:
+			bits_per_pixel = dc_fixpt_from_int(12);
+			break;
+		default:
+			/* Default to commonly supported color depth. */
+			bits_per_pixel = dc_fixpt_from_int(8);
+			break;
+		}
+
+		bits_per_pixel = dc_fixpt_mul_int(bits_per_pixel, 3);
+
+		if (crtc_info->pixel_encoding == PIXEL_ENCODING_YCBCR422) {
+			bits_per_pixel = dc_fixpt_div_int(bits_per_pixel, 3);
+			bits_per_pixel = dc_fixpt_mul_int(bits_per_pixel, 2);
+		} else if (crtc_info->pixel_encoding == PIXEL_ENCODING_YCBCR420) {
+			bits_per_pixel = dc_fixpt_div_int(bits_per_pixel, 2);
+		}
+	}
+
+	/* Use simple stream BW calculation because mainlink overhead is
+	 * accounted for separately in the audio BW calculations.
+	 */
+	peak_stream_bw_kbps = dc_fixpt_from_fraction(crtc_info->requested_pixel_clock_100Hz, 10);
+	peak_stream_bw_kbps = dc_fixpt_mul(peak_stream_bw_kbps, bits_per_pixel);
+	link_bw_kbps = dc_fixpt_from_int(dp_link_info->link_bandwidth_kbps);
+	average_stream_sym_bw_frac = dc_fixpt_div(peak_stream_bw_kbps, link_bw_kbps);
+
+	available_stream_sym_count = dc_fixpt_mul_int(hblank_time_msec, 1000);
+	available_stream_sym_count = dc_fixpt_mul(available_stream_sym_count, lsclkfreq_mhz);
+	available_stream_sym_count = dc_fixpt_mul(available_stream_sym_count, average_stream_sym_bw_frac);
+	available_hblank_bw = dc_fixpt_floor(available_stream_sym_count);
+	available_hblank_bw *= dp_link_info->lane_count;
+	available_hblank_bw -= crtc_info->dsc_num_slices * 4; /* EOC overhead */
+
+	if (available_hblank_bw < dp_link_info->hblank_min_symbol_width)
+		/* Each symbol takes 4 frames */
+		available_hblank_bw = 4 * dp_link_info->hblank_min_symbol_width;
+
+	if (available_hblank_bw < 12)
+		available_hblank_bw = 0;
+	else
+		available_hblank_bw -= 12; /* Main link overhead */
+
+	return available_hblank_bw;
+}
+
+static void check_audio_bandwidth_dp(
+	const struct audio_crtc_info *crtc_info,
+	const struct audio_dp_link_info *dp_link_info,
+	uint32_t channel_count,
+	union audio_sample_rates *sample_rates)
+{
+	struct dp_audio_layout_config layout_config = {0};
+	uint32_t available_hblank_bw;
+	uint32_t av_stream_map_lane_count;
+	uint32_t audio_sdp_overhead;
+
+	/* TODO: Add validation for SST 8b/10 case  */
+	if (!dp_link_info->is_mst && dp_link_info->encoding == DP_8b_10b_ENCODING)
+		return;
+
+	available_hblank_bw = calculate_available_hblank_bw_in_symbols(
+			crtc_info, dp_link_info);
+	av_stream_map_lane_count = get_av_stream_map_lane_count(
+			dp_link_info->encoding, dp_link_info->lane_count, dp_link_info->is_mst);
+	audio_sdp_overhead = get_audio_sdp_overhead(
+			dp_link_info->encoding, dp_link_info->lane_count, dp_link_info->is_mst);
+	get_audio_layout_config(
+			channel_count, dp_link_info->encoding, &layout_config);
+
+	if (layout_config.max_layouts_per_audio_sdp == 0 ||
+		layout_config.symbols_per_layout == 0 ||
+		layout_config.layouts_per_sample_denom == 0) {
+		return;
+	}
+	if (available_hblank_bw < calculate_required_audio_bw_in_symbols(
+			crtc_info, &layout_config, channel_count, 192000,
+			av_stream_map_lane_count, audio_sdp_overhead))
+		sample_rates->rate.RATE_192 = 0;
+	if (available_hblank_bw < calculate_required_audio_bw_in_symbols(
+			crtc_info, &layout_config, channel_count, 176400,
+			av_stream_map_lane_count, audio_sdp_overhead))
+		sample_rates->rate.RATE_176_4 = 0;
+	if (available_hblank_bw < calculate_required_audio_bw_in_symbols(
+			crtc_info, &layout_config, channel_count, 96000,
+			av_stream_map_lane_count, audio_sdp_overhead))
+		sample_rates->rate.RATE_96 = 0;
+	if (available_hblank_bw < calculate_required_audio_bw_in_symbols(
+			crtc_info, &layout_config, channel_count, 88200,
+			av_stream_map_lane_count, audio_sdp_overhead))
+		sample_rates->rate.RATE_88_2 = 0;
+	if (available_hblank_bw < calculate_required_audio_bw_in_symbols(
+			crtc_info, &layout_config, channel_count, 48000,
+			av_stream_map_lane_count, audio_sdp_overhead))
+		sample_rates->rate.RATE_48 = 0;
+	if (available_hblank_bw < calculate_required_audio_bw_in_symbols(
+			crtc_info, &layout_config, channel_count, 44100,
+			av_stream_map_lane_count, audio_sdp_overhead))
+		sample_rates->rate.RATE_44_1 = 0;
+	if (available_hblank_bw < calculate_required_audio_bw_in_symbols(
+			crtc_info, &layout_config, channel_count, 32000,
+			av_stream_map_lane_count, audio_sdp_overhead))
+		sample_rates->rate.RATE_32 = 0;
+}
+
+static void check_audio_bandwidth(
+	const struct audio_crtc_info *crtc_info,
+	const struct audio_dp_link_info *dp_link_info,
+	uint32_t channel_count,
+	enum signal_type signal,
+	union audio_sample_rates *sample_rates)
+{
+	switch (signal) {
+	case SIGNAL_TYPE_HDMI_TYPE_A:
+		check_audio_bandwidth_hdmi(
+			crtc_info, channel_count, sample_rates);
+		break;
+	case SIGNAL_TYPE_EDP:
+	case SIGNAL_TYPE_DISPLAY_PORT:
+	case SIGNAL_TYPE_DISPLAY_PORT_MST:
+		check_audio_bandwidth_dp(
+			crtc_info, dp_link_info, channel_count, sample_rates);
+		break;
+	default:
+		break;
+	}
+}
+
+/* expose/not expose HBR capability to Audio driver */
+static void set_high_bit_rate_capable(
+	struct audio *audio,
+	bool capable)
+{
+	uint32_t value = 0;
+
+	/* set high bit rate audio capable*/
+	value = AZ_REG_READ(AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_HBR);
+
+	set_reg_field_value(value, capable,
+		AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_HBR,
+		HBR_CAPABLE);
+
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_HBR, value);
+}
+
+/* set video latency in ms/2+1 */
+static void set_video_latency(
+	struct audio *audio,
+	int latency_in_ms)
+{
+	uint32_t value = 0;
+
+	if ((latency_in_ms < 0) || (latency_in_ms > 255))
+		return;
+
+	value = AZ_REG_READ(AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC);
+
+	set_reg_field_value(value, latency_in_ms,
+		AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+		VIDEO_LIPSYNC);
+
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+		value);
+}
+
+/* set audio latency in ms/2+1 */
+static void set_audio_latency(
+	struct audio *audio,
+	int latency_in_ms)
+{
+	uint32_t value = 0;
+
+	if (latency_in_ms < 0)
+		latency_in_ms = 0;
+
+	if (latency_in_ms > 255)
+		latency_in_ms = 255;
+
+	value = AZ_REG_READ(AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC);
+
+	set_reg_field_value(value, latency_in_ms,
+		AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+		AUDIO_LIPSYNC);
+
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+		value);
+}
+
+void dce_aud_az_enable(struct audio *audio)
+{
+	uint32_t value = AZ_REG_READ(AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL);
+	DC_LOGGER_INIT();
+
+	set_reg_field_value(value, 1,
+			    AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL,
+			    CLOCK_GATING_DISABLE);
+	set_reg_field_value(value, 1,
+			    AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL,
+			    AUDIO_ENABLED);
+
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL, value);
+	set_reg_field_value(value, 0,
+			AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL,
+			CLOCK_GATING_DISABLE);
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL, value);
+
+	DC_LOG_HW_AUDIO("\n\t========= AUDIO:dce_aud_az_enable: index: %u  data: 0x%x\n",
+			audio->inst, value);
+}
+
+void dce_aud_az_disable_hbr_audio(struct audio *audio)
+{
+	set_high_bit_rate_capable(audio, false);
+}
+
+void dce_aud_az_disable(struct audio *audio)
+{
+	uint32_t value;
+	DC_LOGGER_INIT();
+
+	value = AZ_REG_READ(AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL);
+	set_reg_field_value(value, 1,
+			AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL,
+			CLOCK_GATING_DISABLE);
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL, value);
+
+	set_reg_field_value(value, 0,
+		AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL,
+		AUDIO_ENABLED);
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL, value);
+
+	set_reg_field_value(value, 0,
+			AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL,
+			CLOCK_GATING_DISABLE);
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL, value);
+	value = AZ_REG_READ(AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL);
+	DC_LOG_HW_AUDIO("\n\t========= AUDIO:dce_aud_az_disable: index: %u  data: 0x%x\n",
+			audio->inst, value);
+}
+
+void dce_aud_az_configure(
+	struct audio *audio,
+	enum signal_type signal,
+	const struct audio_crtc_info *crtc_info,
+	const struct audio_info *audio_info,
+	const struct audio_dp_link_info *dp_link_info)
+{
+	struct dce_audio *aud = DCE_AUD(audio);
+
+	uint32_t speakers = audio_info->flags.info.ALLSPEAKERS;
+	uint32_t value;
+	uint32_t field = 0;
+	enum audio_format_code audio_format_code;
+	uint32_t format_index;
+	uint32_t index;
+	bool is_ac3_supported = false;
+	union audio_sample_rates sample_rate;
+	uint32_t strlen = 0;
+
+	if (signal == SIGNAL_TYPE_VIRTUAL)
+		return;
+
+	value = AZ_REG_READ(AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL);
+	set_reg_field_value(value, 1,
+			AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL,
+			CLOCK_GATING_DISABLE);
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL, value);
+
+	/* Speaker Allocation */
+	/*
+	uint32_t value;
+	uint32_t field = 0;*/
+	value = AZ_REG_READ(AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER);
+
+	set_reg_field_value(value,
+		speakers,
+		AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+		SPEAKER_ALLOCATION);
+
+	/* LFE_PLAYBACK_LEVEL = LFEPBL
+	 * LFEPBL = 0 : Unknown or refer to other information
+	 * LFEPBL = 1 : 0dB playback
+	 * LFEPBL = 2 : +10dB playback
+	 * LFE_BL = 3 : Reserved
+	 */
+	set_reg_field_value(value,
+		0,
+		AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+		LFE_PLAYBACK_LEVEL);
+	/* todo: according to reg spec LFE_PLAYBACK_LEVEL is read only.
+	 *  why are we writing to it?  DCE8 does not write this */
+
+
+	set_reg_field_value(value,
+		0,
+		AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+		HDMI_CONNECTION);
+
+	set_reg_field_value(value,
+		0,
+		AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+		DP_CONNECTION);
+
+	field = get_reg_field_value(value,
+			AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+			EXTRA_CONNECTION_INFO);
+
+	field &= ~0x1;
+
+	set_reg_field_value(value,
+		field,
+		AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+		EXTRA_CONNECTION_INFO);
+
+	/* set audio for output signal */
+	switch (signal) {
+	case SIGNAL_TYPE_HDMI_TYPE_A:
+		set_reg_field_value(value,
+			1,
+			AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+			HDMI_CONNECTION);
+
+		break;
+
+	case SIGNAL_TYPE_EDP:
+	case SIGNAL_TYPE_DISPLAY_PORT:
+	case SIGNAL_TYPE_DISPLAY_PORT_MST:
+		set_reg_field_value(value,
+			1,
+			AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+			DP_CONNECTION);
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		break;
+	}
+
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER, value);
+
+	/*  ACP Data - Supports AI  */
+	value = AZ_REG_READ(AZALIA_F0_CODEC_PIN_CONTROL_ACP_DATA);
+
+	set_reg_field_value(
+		value,
+		audio_info->flags.info.SUPPORT_AI,
+		AZALIA_F0_CODEC_PIN_CONTROL_ACP_DATA,
+		SUPPORTS_AI);
+
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_ACP_DATA, value);
+
+	/*  Audio Descriptors   */
+	/* pass through all formats */
+	for (format_index = 0; format_index < AUDIO_FORMAT_CODE_COUNT;
+			format_index++) {
+		audio_format_code =
+			(AUDIO_FORMAT_CODE_FIRST + format_index);
+
+		/* those are unsupported, skip programming */
+		if (audio_format_code == AUDIO_FORMAT_CODE_1BITAUDIO ||
+			audio_format_code == AUDIO_FORMAT_CODE_DST)
+			continue;
+
+		value = 0;
+
+		/* check if supported */
+		if (is_audio_format_supported(
+				audio_info, audio_format_code, &index)) {
+			const struct audio_mode *audio_mode =
+					&audio_info->modes[index];
+			union audio_sample_rates sample_rates =
+					audio_mode->sample_rates;
+			uint8_t byte2 = audio_mode->max_bit_rate;
+			uint8_t channel_count = audio_mode->channel_count;
+
+			/* adjust specific properties */
+			switch (audio_format_code) {
+			case AUDIO_FORMAT_CODE_LINEARPCM: {
+
+				check_audio_bandwidth(
+					crtc_info,
+					dp_link_info,
+					channel_count,
+					signal,
+					&sample_rates);
+
+				byte2 = audio_mode->sample_size;
+
+				set_reg_field_value(value,
+						sample_rates.all,
+						AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+						SUPPORTED_FREQUENCIES_STEREO);
+				}
+				break;
+			case AUDIO_FORMAT_CODE_AC3:
+				is_ac3_supported = true;
+				break;
+			case AUDIO_FORMAT_CODE_DOLBYDIGITALPLUS:
+			case AUDIO_FORMAT_CODE_DTS_HD:
+			case AUDIO_FORMAT_CODE_MAT_MLP:
+			case AUDIO_FORMAT_CODE_DST:
+			case AUDIO_FORMAT_CODE_WMAPRO:
+				byte2 = audio_mode->vendor_specific;
+				break;
+			default:
+				break;
+			}
+
+			/* fill audio format data */
+			set_reg_field_value(value,
+					channel_count - 1,
+					AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+					MAX_CHANNELS);
+
+			set_reg_field_value(value,
+					sample_rates.all,
+					AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+					SUPPORTED_FREQUENCIES);
+
+			set_reg_field_value(value,
+					byte2,
+					AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+					DESCRIPTOR_BYTE_2);
+		} /* if */
+
+		AZ_REG_WRITE(
+				AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0 + format_index,
+				value);
+	} /* for */
+
+	if (is_ac3_supported)
+		/* todo: this reg global.  why program global register? */
+		REG_WRITE(AZALIA_F0_CODEC_FUNCTION_PARAMETER_STREAM_FORMATS,
+				0x05);
+
+	/* check for 192khz/8-Ch support for HBR requirements */
+	sample_rate.all = 0;
+	sample_rate.rate.RATE_192 = 1;
+
+	check_audio_bandwidth(
+		crtc_info,
+		dp_link_info,
+		8,
+		signal,
+		&sample_rate);
+
+	set_high_bit_rate_capable(audio, sample_rate.rate.RATE_192);
+
+	/* Audio and Video Lipsync */
+	set_video_latency(audio, audio_info->video_latency);
+	set_audio_latency(audio, audio_info->audio_latency);
+
+	value = 0;
+	set_reg_field_value(value, audio_info->manufacture_id,
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO0,
+		MANUFACTURER_ID);
+
+	set_reg_field_value(value, audio_info->product_id,
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO0,
+		PRODUCT_ID);
+
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO0,
+		value);
+
+	value = 0;
+
+	/*get display name string length */
+	while (audio_info->display_name[strlen++] != '\0') {
+		if (strlen >=
+		MAX_HW_AUDIO_INFO_DISPLAY_NAME_SIZE_IN_CHARS)
+			break;
+		}
+	set_reg_field_value(value, strlen,
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO1,
+		SINK_DESCRIPTION_LEN);
+
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO1,
+		value);
+	DC_LOG_HW_AUDIO("\n\tAUDIO:az_configure: index: %u data, 0x%x, displayName %s: \n",
+		audio->inst, value, audio_info->display_name);
+
+	/*
+	*write the port ID:
+	*PORT_ID0 = display index
+	*PORT_ID1 = 16bit BDF
+	*(format MSB->LSB: 8bit Bus, 5bit Device, 3bit Function)
+	*/
+
+	value = 0;
+
+	set_reg_field_value(value, audio_info->port_id[0],
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO2,
+		PORT_ID0);
+
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO2, value);
+
+	value = 0;
+	set_reg_field_value(value, audio_info->port_id[1],
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO3,
+		PORT_ID1);
+
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO3, value);
+
+	/*write the 18 char monitor string */
+
+	value = 0;
+	set_reg_field_value(value, audio_info->display_name[0],
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO4,
+		DESCRIPTION0);
+
+	set_reg_field_value(value, audio_info->display_name[1],
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO4,
+		DESCRIPTION1);
+
+	set_reg_field_value(value, audio_info->display_name[2],
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO4,
+		DESCRIPTION2);
+
+	set_reg_field_value(value, audio_info->display_name[3],
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO4,
+		DESCRIPTION3);
+
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO4, value);
+
+	value = 0;
+	set_reg_field_value(value, audio_info->display_name[4],
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO5,
+		DESCRIPTION4);
+
+	set_reg_field_value(value, audio_info->display_name[5],
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO5,
+		DESCRIPTION5);
+
+	set_reg_field_value(value, audio_info->display_name[6],
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO5,
+		DESCRIPTION6);
+
+	set_reg_field_value(value, audio_info->display_name[7],
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO5,
+		DESCRIPTION7);
+
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO5, value);
+
+	value = 0;
+	set_reg_field_value(value, audio_info->display_name[8],
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO6,
+		DESCRIPTION8);
+
+	set_reg_field_value(value, audio_info->display_name[9],
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO6,
+		DESCRIPTION9);
+
+	set_reg_field_value(value, audio_info->display_name[10],
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO6,
+		DESCRIPTION10);
+
+	set_reg_field_value(value, audio_info->display_name[11],
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO6,
+		DESCRIPTION11);
+
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO6, value);
+
+	value = 0;
+	set_reg_field_value(value, audio_info->display_name[12],
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO7,
+		DESCRIPTION12);
+
+	set_reg_field_value(value, audio_info->display_name[13],
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO7,
+		DESCRIPTION13);
+
+	set_reg_field_value(value, audio_info->display_name[14],
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO7,
+		DESCRIPTION14);
+
+	set_reg_field_value(value, audio_info->display_name[15],
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO7,
+		DESCRIPTION15);
+
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO7, value);
+
+	value = 0;
+	set_reg_field_value(value, audio_info->display_name[16],
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO8,
+		DESCRIPTION16);
+
+	set_reg_field_value(value, audio_info->display_name[17],
+		AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO8,
+		DESCRIPTION17);
+
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_SINK_INFO8, value);
+	value = AZ_REG_READ(AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL);
+	set_reg_field_value(value, 0,
+			AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL,
+			CLOCK_GATING_DISABLE);
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL, value);
+}
+
+/*
+* todo: wall clk related functionality probably belong to clock_src.
+*/
+
+/* search pixel clock value for Azalia HDMI Audio */
+static void get_azalia_clock_info_hdmi(
+	uint32_t crtc_pixel_clock_100hz,
+	uint32_t actual_pixel_clock_100Hz,
+	struct azalia_clock_info *azalia_clock_info)
+{
+	/* audio_dto_phase= 24 * 10,000;
+	 *   24MHz in [100Hz] units */
+	azalia_clock_info->audio_dto_phase =
+			24 * 10000;
+
+	/* audio_dto_module = PCLKFrequency * 10,000;
+	 *  [khz] -> [100Hz] */
+	azalia_clock_info->audio_dto_module =
+			actual_pixel_clock_100Hz;
+}
+
+static void get_azalia_clock_info_dp(
+	uint32_t requested_pixel_clock_100Hz,
+	const struct audio_pll_info *pll_info,
+	struct azalia_clock_info *azalia_clock_info)
+{
+	/* Reported dpDtoSourceClockInkhz value for
+	 * DCE8 already adjusted for SS, do not need any
+	 * adjustment here anymore
+	 */
+
+	/*audio_dto_phase = 24 * 10,000;
+	 * 24MHz in [100Hz] units */
+	azalia_clock_info->audio_dto_phase = 24 * 10000;
+
+	/*audio_dto_module = dpDtoSourceClockInkhz * 10,000;
+	 *  [khz] ->[100Hz] */
+	azalia_clock_info->audio_dto_module =
+		pll_info->audio_dto_source_clock_in_khz * 10;
+}
+
+void dce_aud_wall_dto_setup(
+	struct audio *audio,
+	enum signal_type signal,
+	const struct audio_crtc_info *crtc_info,
+	const struct audio_pll_info *pll_info)
+{
+	struct dce_audio *aud = DCE_AUD(audio);
+
+	struct azalia_clock_info clock_info = { 0 };
+
+	if (dc_is_hdmi_tmds_signal(signal)) {
+		uint32_t src_sel;
+
+		/*DTO0 Programming goal:
+		-generate 24MHz, 128*Fs from 24MHz
+		-use DTO0 when an active HDMI port is connected
+		(optionally a DP is connected) */
+
+		/* calculate DTO settings */
+		get_azalia_clock_info_hdmi(
+			crtc_info->requested_pixel_clock_100Hz,
+			crtc_info->calculated_pixel_clock_100Hz,
+			&clock_info);
+
+		DC_LOG_HW_AUDIO("\n%s:Input::requested_pixel_clock_100Hz = %d"\
+				"calculated_pixel_clock_100Hz =%d\n"\
+				"audio_dto_module = %d audio_dto_phase =%d \n\n", __func__,\
+				crtc_info->requested_pixel_clock_100Hz,\
+				crtc_info->calculated_pixel_clock_100Hz,\
+				clock_info.audio_dto_module,\
+				clock_info.audio_dto_phase);
+
+		/* On TN/SI, Program DTO source select and DTO select before
+		programming DTO modulo and DTO phase. These bits must be
+		programmed first, otherwise there will be no HDMI audio at boot
+		up. This is a HW sequence change (different from old ASICs).
+		Caution when changing this programming sequence.
+
+		HDMI enabled, using DTO0
+		program master CRTC for DTO0 */
+		src_sel = pll_info->dto_source - DTO_SOURCE_ID0;
+		REG_UPDATE_2(DCCG_AUDIO_DTO_SOURCE,
+			DCCG_AUDIO_DTO0_SOURCE_SEL, src_sel,
+			DCCG_AUDIO_DTO_SEL, 0);
+
+		/* module */
+		REG_UPDATE(DCCG_AUDIO_DTO0_MODULE,
+			DCCG_AUDIO_DTO0_MODULE, clock_info.audio_dto_module);
+
+		/* phase */
+		REG_UPDATE(DCCG_AUDIO_DTO0_PHASE,
+			DCCG_AUDIO_DTO0_PHASE, clock_info.audio_dto_phase);
+	} else {
+		/*DTO1 Programming goal:
+		-generate 24MHz, 512*Fs, 128*Fs from 24MHz
+		-default is to used DTO1, and switch to DTO0 when an audio
+		master HDMI port is connected
+		-use as default for DP
+
+		calculate DTO settings */
+		get_azalia_clock_info_dp(
+			crtc_info->requested_pixel_clock_100Hz,
+			pll_info,
+			&clock_info);
+
+		/* Program DTO select before programming DTO modulo and DTO
+		phase. default to use DTO1 */
+
+		REG_UPDATE(DCCG_AUDIO_DTO_SOURCE,
+				DCCG_AUDIO_DTO_SEL, 1);
+
+			/* DCCG_AUDIO_DTO2_USE_512FBR_DTO, 1)
+			 * Select 512fs for DP TODO: web register definition
+			 * does not match register header file
+			 * DCE11 version it's commented out while DCE8 it's set to 1
+			*/
+
+		/* module */
+		REG_UPDATE(DCCG_AUDIO_DTO1_MODULE,
+				DCCG_AUDIO_DTO1_MODULE, clock_info.audio_dto_module);
+
+		/* phase */
+		REG_UPDATE(DCCG_AUDIO_DTO1_PHASE,
+				DCCG_AUDIO_DTO1_PHASE, clock_info.audio_dto_phase);
+
+		REG_UPDATE(DCCG_AUDIO_DTO_SOURCE,
+				DCCG_AUDIO_DTO2_USE_512FBR_DTO, 1);
+
+	}
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static void dce60_aud_wall_dto_setup(
+	struct audio *audio,
+	enum signal_type signal,
+	const struct audio_crtc_info *crtc_info,
+	const struct audio_pll_info *pll_info)
+{
+	struct dce_audio *aud = DCE_AUD(audio);
+
+	struct azalia_clock_info clock_info = { 0 };
+
+	if (dc_is_hdmi_signal(signal)) {
+		uint32_t src_sel;
+
+		/*DTO0 Programming goal:
+		-generate 24MHz, 128*Fs from 24MHz
+		-use DTO0 when an active HDMI port is connected
+		(optionally a DP is connected) */
+
+		/* calculate DTO settings */
+		get_azalia_clock_info_hdmi(
+			crtc_info->requested_pixel_clock_100Hz,
+			crtc_info->calculated_pixel_clock_100Hz,
+			&clock_info);
+
+		DC_LOG_HW_AUDIO("\n%s:Input::requested_pixel_clock_100Hz = %d"\
+				"calculated_pixel_clock_100Hz =%d\n"\
+				"audio_dto_module = %d audio_dto_phase =%d \n\n", __func__,\
+				crtc_info->requested_pixel_clock_100Hz,\
+				crtc_info->calculated_pixel_clock_100Hz,\
+				clock_info.audio_dto_module,\
+				clock_info.audio_dto_phase);
+
+		/* On TN/SI, Program DTO source select and DTO select before
+		programming DTO modulo and DTO phase. These bits must be
+		programmed first, otherwise there will be no HDMI audio at boot
+		up. This is a HW sequence change (different from old ASICs).
+		Caution when changing this programming sequence.
+
+		HDMI enabled, using DTO0
+		program master CRTC for DTO0 */
+		src_sel = pll_info->dto_source - DTO_SOURCE_ID0;
+		REG_UPDATE_2(DCCG_AUDIO_DTO_SOURCE,
+			DCCG_AUDIO_DTO0_SOURCE_SEL, src_sel,
+			DCCG_AUDIO_DTO_SEL, 0);
+
+		/* module */
+		REG_UPDATE(DCCG_AUDIO_DTO0_MODULE,
+			DCCG_AUDIO_DTO0_MODULE, clock_info.audio_dto_module);
+
+		/* phase */
+		REG_UPDATE(DCCG_AUDIO_DTO0_PHASE,
+			DCCG_AUDIO_DTO0_PHASE, clock_info.audio_dto_phase);
+	} else {
+		/*DTO1 Programming goal:
+		-generate 24MHz, 128*Fs from 24MHz (DCE6 does not support 512*Fs)
+		-default is to used DTO1, and switch to DTO0 when an audio
+		master HDMI port is connected
+		-use as default for DP
+
+		calculate DTO settings */
+		get_azalia_clock_info_dp(
+			crtc_info->requested_pixel_clock_100Hz,
+			pll_info,
+			&clock_info);
+
+		/* Program DTO select before programming DTO modulo and DTO
+		phase. default to use DTO1 */
+
+		REG_UPDATE(DCCG_AUDIO_DTO_SOURCE,
+				DCCG_AUDIO_DTO_SEL, 1);
+
+			/* DCCG_AUDIO_DTO2_USE_512FBR_DTO, 1)
+			 * Cannot select 512fs for DP
+			 *
+			 * DCE6 has no DCCG_AUDIO_DTO2_USE_512FBR_DTO mask
+			*/
+
+		/* module */
+		REG_UPDATE(DCCG_AUDIO_DTO1_MODULE,
+				DCCG_AUDIO_DTO1_MODULE, clock_info.audio_dto_module);
+
+		/* phase */
+		REG_UPDATE(DCCG_AUDIO_DTO1_PHASE,
+				DCCG_AUDIO_DTO1_PHASE, clock_info.audio_dto_phase);
+
+		/* DCE6 has no DCCG_AUDIO_DTO2_USE_512FBR_DTO mask in DCCG_AUDIO_DTO_SOURCE reg */
+
+	}
+}
+#endif
+
+static bool dce_aud_endpoint_valid(struct audio *audio)
+{
+	uint32_t value;
+	uint32_t port_connectivity;
+
+	value = AZ_REG_READ(
+			AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT);
+
+	port_connectivity = get_reg_field_value(value,
+			AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT,
+			PORT_CONNECTIVITY);
+
+	return !(port_connectivity == 1);
+}
+
+/* initialize HW state */
+void dce_aud_hw_init(
+		struct audio *audio)
+{
+	uint32_t value;
+	struct dce_audio *aud = DCE_AUD(audio);
+
+	/* we only need to program the following registers once, so we only do
+	it for the inst 0*/
+	if (audio->inst != 0)
+		return;
+
+	/* Suport R5 - 32khz
+	 * Suport R6 - 44.1khz
+	 * Suport R7 - 48khz
+	 */
+	/*disable clock gating before write to endpoint register*/
+	value = AZ_REG_READ(AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL);
+	set_reg_field_value(value, 1,
+			AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL,
+			CLOCK_GATING_DISABLE);
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL, value);
+	REG_UPDATE(AZALIA_F0_CODEC_FUNCTION_PARAMETER_SUPPORTED_SIZE_RATES,
+			AUDIO_RATE_CAPABILITIES, 0x70);
+
+	/*Keep alive bit to verify HW block in BU. */
+	REG_UPDATE_2(AZALIA_F0_CODEC_FUNCTION_PARAMETER_POWER_STATES,
+			CLKSTOP, 1,
+			EPSS, 1);
+	set_reg_field_value(value, 0,
+			AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL,
+			CLOCK_GATING_DISABLE);
+	AZ_REG_WRITE(AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL, value);
+}
+
+static const struct audio_funcs funcs = {
+	.endpoint_valid = dce_aud_endpoint_valid,
+	.hw_init = dce_aud_hw_init,
+	.wall_dto_setup = dce_aud_wall_dto_setup,
+	.az_enable = dce_aud_az_enable,
+	.az_disable = dce_aud_az_disable,
+	.az_configure = dce_aud_az_configure,
+	.az_disable_hbr_audio = dce_aud_az_disable_hbr_audio,
+	.destroy = dce_aud_destroy,
+};
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static const struct audio_funcs dce60_funcs = {
+	.endpoint_valid = dce_aud_endpoint_valid,
+	.hw_init = dce_aud_hw_init,
+	.wall_dto_setup = dce60_aud_wall_dto_setup,
+	.az_enable = dce_aud_az_enable,
+	.az_disable = dce_aud_az_disable,
+	.az_configure = dce_aud_az_configure,
+	.destroy = dce_aud_destroy,
+};
+#endif
+
+void dce_aud_destroy(struct audio **audio)
+{
+	struct dce_audio *aud = DCE_AUD(*audio);
+
+	kfree(aud);
+	*audio = NULL;
+}
+
+struct audio *dce_audio_create(
+		struct dc_context *ctx,
+		unsigned int inst,
+		const struct dce_audio_registers *reg,
+		const struct dce_audio_shift *shifts,
+		const struct dce_audio_mask *masks
+		)
+{
+	struct dce_audio *audio = kzalloc(sizeof(*audio), GFP_KERNEL);
+
+	if (audio == NULL) {
+		ASSERT_CRITICAL(audio);
+		return NULL;
+	}
+
+	audio->base.ctx = ctx;
+	audio->base.inst = inst;
+	audio->base.funcs = &funcs;
+
+	audio->regs = reg;
+	audio->shifts = shifts;
+	audio->masks = masks;
+	return &audio->base;
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+struct audio *dce60_audio_create(
+		struct dc_context *ctx,
+		unsigned int inst,
+		const struct dce_audio_registers *reg,
+		const struct dce_audio_shift *shifts,
+		const struct dce_audio_mask *masks
+		)
+{
+	struct dce_audio *audio = kzalloc(sizeof(*audio), GFP_KERNEL);
+
+	if (audio == NULL) {
+		ASSERT_CRITICAL(audio);
+		return NULL;
+	}
+
+	audio->base.ctx = ctx;
+	audio->base.inst = inst;
+	audio->base.funcs = &dce60_funcs;
+
+	audio->regs = reg;
+	audio->shifts = shifts;
+	audio->masks = masks;
+	return &audio->base;
+}
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_audio.h b/drivers/gpu/drm/amd/display/dc/dce/dce_audio.h
new file mode 100644
index 000000000000..1b7b8b079af4
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_audio.h
@@ -0,0 +1,182 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+#ifndef __DAL_AUDIO_DCE_110_H__
+#define __DAL_AUDIO_DCE_110_H__
+
+#include "audio.h"
+
+#define AUD_COMMON_REG_LIST(id)\
+	SRI(AZALIA_F0_CODEC_ENDPOINT_INDEX, AZF0ENDPOINT, id),\
+	SRI(AZALIA_F0_CODEC_ENDPOINT_DATA, AZF0ENDPOINT, id),\
+	SR(AZALIA_F0_CODEC_FUNCTION_PARAMETER_STREAM_FORMATS),\
+	SR(AZALIA_F0_CODEC_FUNCTION_PARAMETER_SUPPORTED_SIZE_RATES),\
+	SR(AZALIA_F0_CODEC_FUNCTION_PARAMETER_POWER_STATES),\
+	SR(DCCG_AUDIO_DTO_SOURCE),\
+	SR(DCCG_AUDIO_DTO0_MODULE),\
+	SR(DCCG_AUDIO_DTO0_PHASE),\
+	SR(DCCG_AUDIO_DTO1_MODULE),\
+	SR(DCCG_AUDIO_DTO1_PHASE)
+
+
+ /* set field name */
+#define SF(reg_name, field_name, post_fix)\
+	.field_name = reg_name ## __ ## field_name ## post_fix
+
+
+#define AUD_COMMON_MASK_SH_LIST_BASE(mask_sh)\
+		SF(DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO0_SOURCE_SEL, mask_sh),\
+		SF(DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO_SEL, mask_sh),\
+		SF(DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO2_USE_512FBR_DTO, mask_sh),\
+		SF(DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO0_USE_512FBR_DTO, mask_sh),\
+		SF(DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO1_USE_512FBR_DTO, mask_sh),\
+		SF(DCCG_AUDIO_DTO0_MODULE, DCCG_AUDIO_DTO0_MODULE, mask_sh),\
+		SF(DCCG_AUDIO_DTO0_PHASE, DCCG_AUDIO_DTO0_PHASE, mask_sh),\
+		SF(DCCG_AUDIO_DTO1_MODULE, DCCG_AUDIO_DTO1_MODULE, mask_sh),\
+		SF(DCCG_AUDIO_DTO1_PHASE, DCCG_AUDIO_DTO1_PHASE, mask_sh),\
+		SF(AZALIA_F0_CODEC_FUNCTION_PARAMETER_SUPPORTED_SIZE_RATES, AUDIO_RATE_CAPABILITIES, mask_sh),\
+		SF(AZALIA_F0_CODEC_FUNCTION_PARAMETER_POWER_STATES, CLKSTOP, mask_sh),\
+		SF(AZALIA_F0_CODEC_FUNCTION_PARAMETER_POWER_STATES, EPSS, mask_sh)
+
+#define AUD_COMMON_MASK_SH_LIST(mask_sh)\
+		AUD_COMMON_MASK_SH_LIST_BASE(mask_sh),\
+		SF(AZALIA_F0_CODEC_ENDPOINT_INDEX, AZALIA_ENDPOINT_REG_INDEX, mask_sh),\
+		SF(AZALIA_F0_CODEC_ENDPOINT_DATA, AZALIA_ENDPOINT_REG_DATA, mask_sh)
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+#define AUD_DCE60_MASK_SH_LIST(mask_sh)\
+		SF(DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO0_SOURCE_SEL, mask_sh),\
+		SF(DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO_SEL, mask_sh),\
+		SF(DCCG_AUDIO_DTO0_MODULE, DCCG_AUDIO_DTO0_MODULE, mask_sh),\
+		SF(DCCG_AUDIO_DTO0_PHASE, DCCG_AUDIO_DTO0_PHASE, mask_sh),\
+		SF(DCCG_AUDIO_DTO1_MODULE, DCCG_AUDIO_DTO1_MODULE, mask_sh),\
+		SF(DCCG_AUDIO_DTO1_PHASE, DCCG_AUDIO_DTO1_PHASE, mask_sh),\
+		SF(AZALIA_F0_CODEC_FUNCTION_PARAMETER_SUPPORTED_SIZE_RATES, AUDIO_RATE_CAPABILITIES, mask_sh),\
+		SF(AZALIA_F0_CODEC_FUNCTION_PARAMETER_POWER_STATES, CLKSTOP, mask_sh),\
+		SF(AZALIA_F0_CODEC_FUNCTION_PARAMETER_POWER_STATES, EPSS, mask_sh), \
+		SF(AZALIA_F0_CODEC_ENDPOINT_INDEX, AZALIA_ENDPOINT_REG_INDEX, mask_sh),\
+		SF(AZALIA_F0_CODEC_ENDPOINT_DATA, AZALIA_ENDPOINT_REG_DATA, mask_sh)
+#endif
+
+struct dce_audio_registers {
+	uint32_t AZALIA_F0_CODEC_ENDPOINT_INDEX;
+	uint32_t AZALIA_F0_CODEC_ENDPOINT_DATA;
+
+	uint32_t AZALIA_F0_CODEC_FUNCTION_PARAMETER_STREAM_FORMATS;
+	uint32_t AZALIA_F0_CODEC_FUNCTION_PARAMETER_SUPPORTED_SIZE_RATES;
+	uint32_t AZALIA_F0_CODEC_FUNCTION_PARAMETER_POWER_STATES;
+
+	uint32_t DCCG_AUDIO_DTO_SOURCE;
+	uint32_t DCCG_AUDIO_DTO0_MODULE;
+	uint32_t DCCG_AUDIO_DTO0_PHASE;
+	uint32_t DCCG_AUDIO_DTO1_MODULE;
+	uint32_t DCCG_AUDIO_DTO1_PHASE;
+
+	uint32_t AUDIO_RATE_CAPABILITIES;
+};
+
+struct dce_audio_shift {
+	uint8_t AZALIA_ENDPOINT_REG_INDEX;
+	uint8_t AZALIA_ENDPOINT_REG_DATA;
+
+	uint8_t AUDIO_RATE_CAPABILITIES;
+	uint8_t CLKSTOP;
+	uint8_t EPSS;
+
+	uint8_t DCCG_AUDIO_DTO0_SOURCE_SEL;
+	uint8_t DCCG_AUDIO_DTO_SEL;
+	uint8_t DCCG_AUDIO_DTO0_MODULE;
+	uint8_t DCCG_AUDIO_DTO0_PHASE;
+	uint8_t DCCG_AUDIO_DTO1_MODULE;
+	uint8_t DCCG_AUDIO_DTO1_PHASE;
+	uint8_t DCCG_AUDIO_DTO2_USE_512FBR_DTO;
+	uint32_t DCCG_AUDIO_DTO0_USE_512FBR_DTO;
+	uint32_t DCCG_AUDIO_DTO1_USE_512FBR_DTO;
+	uint32_t CLOCK_GATING_DISABLE;
+};
+
+struct dce_audio_mask {
+	uint32_t AZALIA_ENDPOINT_REG_INDEX;
+	uint32_t AZALIA_ENDPOINT_REG_DATA;
+
+	uint32_t AUDIO_RATE_CAPABILITIES;
+	uint32_t CLKSTOP;
+	uint32_t EPSS;
+
+	uint32_t DCCG_AUDIO_DTO0_SOURCE_SEL;
+	uint32_t DCCG_AUDIO_DTO_SEL;
+	uint32_t DCCG_AUDIO_DTO0_MODULE;
+	uint32_t DCCG_AUDIO_DTO0_PHASE;
+	uint32_t DCCG_AUDIO_DTO1_MODULE;
+	uint32_t DCCG_AUDIO_DTO1_PHASE;
+	uint32_t DCCG_AUDIO_DTO2_USE_512FBR_DTO;
+	uint32_t DCCG_AUDIO_DTO0_USE_512FBR_DTO;
+	uint32_t DCCG_AUDIO_DTO1_USE_512FBR_DTO;
+	uint32_t CLOCK_GATING_DISABLE;
+
+};
+
+struct dce_audio {
+	struct audio base;
+	const struct dce_audio_registers *regs;
+	const struct dce_audio_shift *shifts;
+	const struct dce_audio_mask *masks;
+};
+
+struct audio *dce_audio_create(
+		struct dc_context *ctx,
+		unsigned int inst,
+		const struct dce_audio_registers *reg,
+		const struct dce_audio_shift *shifts,
+		const struct dce_audio_mask *masks);
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+struct audio *dce60_audio_create(
+		struct dc_context *ctx,
+		unsigned int inst,
+		const struct dce_audio_registers *reg,
+		const struct dce_audio_shift *shifts,
+		const struct dce_audio_mask *masks);
+#endif
+
+void dce_aud_destroy(struct audio **audio);
+
+void dce_aud_hw_init(struct audio *audio);
+
+void dce_aud_az_enable(struct audio *audio);
+void dce_aud_az_disable(struct audio *audio);
+void dce_aud_az_disable_hbr_audio(struct audio *audio);
+
+void dce_aud_az_configure(struct audio *audio,
+	enum signal_type signal,
+	const struct audio_crtc_info *crtc_info,
+	const struct audio_info *audio_info,
+	const struct audio_dp_link_info *dp_link_info);
+
+void dce_aud_wall_dto_setup(struct audio *audio,
+	enum signal_type signal,
+	const struct audio_crtc_info *crtc_info,
+	const struct audio_pll_info *pll_info);
+
+#endif   /*__DAL_AUDIO_DCE_110_H__*/
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_aux.c b/drivers/gpu/drm/amd/display/dc/dce/dce_aux.c
new file mode 100644
index 000000000000..673bb87d2c17
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_aux.c
@@ -0,0 +1,970 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+#include "core_types.h"
+#include "dce_aux.h"
+#include "dce/dce_11_0_sh_mask.h"
+#include "dm_event_log.h"
+#include "dm_helpers.h"
+#include "dmub/inc/dmub_cmd.h"
+
+#define CTX \
+	aux110->base.ctx
+#define REG(reg_name)\
+	(aux110->regs->reg_name)
+
+#define DC_LOGGER \
+	engine->ctx->logger
+
+#define DC_TRACE_LEVEL_MESSAGE(...) do { } while (0)
+#define IS_DC_I2CAUX_LOGGING_ENABLED() (false)
+#define LOG_FLAG_Error_I2cAux LOG_ERROR
+#define LOG_FLAG_I2cAux_DceAux LOG_I2C_AUX
+
+#include "reg_helper.h"
+
+#undef FN
+#define FN(reg_name, field_name) \
+	aux110->shift->field_name, aux110->mask->field_name
+
+#define FROM_AUX_ENGINE(ptr) \
+	container_of((ptr), struct aux_engine_dce110, base)
+
+#define FROM_ENGINE(ptr) \
+	FROM_AUX_ENGINE(container_of((ptr), struct dce_aux, base))
+
+#define FROM_AUX_ENGINE_ENGINE(ptr) \
+	container_of((ptr), struct dce_aux, base)
+enum {
+	AUX_INVALID_REPLY_RETRY_COUNTER = 1,
+	AUX_TIMED_OUT_RETRY_COUNTER = 2,
+	AUX_DEFER_RETRY_COUNTER = 6
+};
+
+#define TIME_OUT_INCREMENT        1016
+#define TIME_OUT_MULTIPLIER_8     8
+#define TIME_OUT_MULTIPLIER_16    16
+#define TIME_OUT_MULTIPLIER_32    32
+#define TIME_OUT_MULTIPLIER_64    64
+#define MAX_TIMEOUT_LENGTH        127
+#define DEFAULT_AUX_ENGINE_MULT   0
+#define DEFAULT_AUX_ENGINE_LENGTH 69
+
+#define DC_TRACE_LEVEL_MESSAGE(...) do { } while (0)
+
+static void release_engine(
+	struct dce_aux *engine)
+{
+	struct aux_engine_dce110 *aux110 = FROM_AUX_ENGINE(engine);
+
+	dal_ddc_close(engine->ddc);
+
+	engine->ddc = NULL;
+
+	REG_UPDATE_2(AUX_ARB_CONTROL, AUX_SW_DONE_USING_AUX_REG, 1,
+		AUX_SW_USE_AUX_REG_REQ, 0);
+}
+
+#define SW_CAN_ACCESS_AUX 1
+#define DMCU_CAN_ACCESS_AUX 2
+
+static bool is_engine_available(
+	struct dce_aux *engine)
+{
+	struct aux_engine_dce110 *aux110 = FROM_AUX_ENGINE(engine);
+
+	uint32_t value = REG_READ(AUX_ARB_CONTROL);
+	uint32_t field = get_reg_field_value(
+			value,
+			AUX_ARB_CONTROL,
+			AUX_REG_RW_CNTL_STATUS);
+
+	return (field != DMCU_CAN_ACCESS_AUX);
+}
+static bool acquire_engine(
+	struct dce_aux *engine)
+{
+	struct aux_engine_dce110 *aux110 = FROM_AUX_ENGINE(engine);
+
+	uint32_t value = REG_READ(AUX_ARB_CONTROL);
+	uint32_t field = get_reg_field_value(
+			value,
+			AUX_ARB_CONTROL,
+			AUX_REG_RW_CNTL_STATUS);
+	if (field == DMCU_CAN_ACCESS_AUX)
+		return false;
+	/* enable AUX before request SW to access AUX */
+	value = REG_READ(AUX_CONTROL);
+	field = get_reg_field_value(value,
+				AUX_CONTROL,
+				AUX_EN);
+
+	if (field == 0) {
+		set_reg_field_value(
+				value,
+				1,
+				AUX_CONTROL,
+				AUX_EN);
+
+		if (REG(AUX_RESET_MASK)) {
+			/*DP_AUX block as part of the enable sequence*/
+			set_reg_field_value(
+				value,
+				1,
+				AUX_CONTROL,
+				AUX_RESET);
+		}
+
+		REG_WRITE(AUX_CONTROL, value);
+
+		if (REG(AUX_RESET_MASK)) {
+			/*poll HW to make sure reset it done*/
+
+			REG_WAIT(AUX_CONTROL, AUX_RESET_DONE, 1,
+					1, 11);
+
+			set_reg_field_value(
+				value,
+				0,
+				AUX_CONTROL,
+				AUX_RESET);
+
+			REG_WRITE(AUX_CONTROL, value);
+
+			REG_WAIT(AUX_CONTROL, AUX_RESET_DONE, 0,
+					1, 11);
+		}
+	} /*if (field)*/
+
+	/* request SW to access AUX */
+	REG_UPDATE(AUX_ARB_CONTROL, AUX_SW_USE_AUX_REG_REQ, 1);
+
+	value = REG_READ(AUX_ARB_CONTROL);
+	field = get_reg_field_value(
+			value,
+			AUX_ARB_CONTROL,
+			AUX_REG_RW_CNTL_STATUS);
+
+	return (field == SW_CAN_ACCESS_AUX);
+}
+
+#define COMPOSE_AUX_SW_DATA_16_20(command, address) \
+	((command) | ((0xF0000 & (address)) >> 16))
+
+#define COMPOSE_AUX_SW_DATA_8_15(address) \
+	((0xFF00 & (address)) >> 8)
+
+#define COMPOSE_AUX_SW_DATA_0_7(address) \
+	(0xFF & (address))
+
+static void submit_channel_request(
+	struct dce_aux *engine,
+	struct aux_request_transaction_data *request)
+{
+	struct aux_engine_dce110 *aux110 = FROM_AUX_ENGINE(engine);
+	uint32_t value;
+	uint32_t length;
+
+	bool is_write =
+		((request->type == AUX_TRANSACTION_TYPE_DP) &&
+		 (request->action == I2CAUX_TRANSACTION_ACTION_DP_WRITE)) ||
+		((request->type == AUX_TRANSACTION_TYPE_I2C) &&
+		((request->action == I2CAUX_TRANSACTION_ACTION_I2C_WRITE) ||
+		 (request->action == I2CAUX_TRANSACTION_ACTION_I2C_WRITE_MOT)));
+	if (REG(AUXN_IMPCAL)) {
+		/* clear_aux_error */
+		REG_UPDATE_SEQ_2(AUXN_IMPCAL,
+				AUXN_CALOUT_ERROR_AK, 1,
+				AUXN_CALOUT_ERROR_AK, 0);
+
+		REG_UPDATE_SEQ_2(AUXP_IMPCAL,
+				AUXP_CALOUT_ERROR_AK, 1,
+				AUXP_CALOUT_ERROR_AK, 0);
+
+		/* force_default_calibrate */
+		REG_UPDATE_SEQ_2(AUXN_IMPCAL,
+				AUXN_IMPCAL_ENABLE, 1,
+				AUXN_IMPCAL_OVERRIDE_ENABLE, 0);
+
+		/* bug? why AUXN update EN and OVERRIDE_EN 1 by 1 while AUX P toggles OVERRIDE? */
+
+		REG_UPDATE_SEQ_2(AUXP_IMPCAL,
+				AUXP_IMPCAL_OVERRIDE_ENABLE, 1,
+				AUXP_IMPCAL_OVERRIDE_ENABLE, 0);
+	}
+
+	REG_UPDATE(AUX_INTERRUPT_CONTROL, AUX_SW_DONE_ACK, 1);
+
+	REG_WAIT(AUX_SW_STATUS, AUX_SW_DONE, 0,
+				10, aux110->polling_timeout_period/10);
+
+	/* set the delay and the number of bytes to write */
+
+	/* The length include
+	 * the 4 bit header and the 20 bit address
+	 * (that is 3 byte).
+	 * If the requested length is non zero this means
+	 * an addition byte specifying the length is required.
+	 */
+
+	length = request->length ? 4 : 3;
+	if (is_write)
+		length += request->length;
+
+	REG_UPDATE_2(AUX_SW_CONTROL,
+			AUX_SW_START_DELAY, request->delay,
+			AUX_SW_WR_BYTES, length);
+
+	/* program action and address and payload data (if 'is_write') */
+	value = REG_UPDATE_4(AUX_SW_DATA,
+			AUX_SW_INDEX, 0,
+			AUX_SW_DATA_RW, 0,
+			AUX_SW_AUTOINCREMENT_DISABLE, 1,
+			AUX_SW_DATA, COMPOSE_AUX_SW_DATA_16_20(request->action, request->address));
+
+	value = REG_SET_2(AUX_SW_DATA, value,
+			AUX_SW_AUTOINCREMENT_DISABLE, 0,
+			AUX_SW_DATA, COMPOSE_AUX_SW_DATA_8_15(request->address));
+
+	value = REG_SET(AUX_SW_DATA, value,
+			AUX_SW_DATA, COMPOSE_AUX_SW_DATA_0_7(request->address));
+
+	if (request->length) {
+		value = REG_SET(AUX_SW_DATA, value,
+				AUX_SW_DATA, request->length - 1);
+	}
+
+	if (is_write) {
+		/* Load the HW buffer with the Data to be sent.
+		 * This is relevant for write operation.
+		 * For read, the data recived data will be
+		 * processed in process_channel_reply().
+		 */
+		uint32_t i = 0;
+
+		while (i < request->length) {
+			value = REG_SET(AUX_SW_DATA, value,
+					AUX_SW_DATA, request->data[i]);
+
+			++i;
+		}
+	}
+
+	REG_UPDATE(AUX_SW_CONTROL, AUX_SW_GO, 1);
+	EVENT_LOG_AUX_REQ(engine->ddc->pin_data->en, EVENT_LOG_AUX_ORIGIN_NATIVE,
+					request->action, request->address, request->length, request->data);
+}
+
+static int read_channel_reply(struct dce_aux *engine, uint32_t size,
+			      uint8_t *buffer, uint8_t *reply_result,
+			      uint32_t *sw_status)
+{
+	struct aux_engine_dce110 *aux110 = FROM_AUX_ENGINE(engine);
+	uint32_t bytes_replied;
+	uint32_t reply_result_32;
+
+	*sw_status = REG_GET(AUX_SW_STATUS, AUX_SW_REPLY_BYTE_COUNT,
+			     &bytes_replied);
+
+	/* In case HPD is LOW, exit AUX transaction */
+	if ((*sw_status & AUX_SW_STATUS__AUX_SW_HPD_DISCON_MASK))
+		return -1;
+
+	/* Need at least the status byte */
+	if (!bytes_replied)
+		return -1;
+
+	REG_UPDATE_SEQ_3(AUX_SW_DATA,
+			  AUX_SW_INDEX, 0,
+			  AUX_SW_AUTOINCREMENT_DISABLE, 1,
+			  AUX_SW_DATA_RW, 1);
+
+	REG_GET(AUX_SW_DATA, AUX_SW_DATA, &reply_result_32);
+	reply_result_32 = reply_result_32 >> 4;
+	if (reply_result != NULL)
+		*reply_result = (uint8_t)reply_result_32;
+
+	if (reply_result_32 == 0) { /* ACK */
+		uint32_t i = 0;
+
+		/* First byte was already used to get the command status */
+		--bytes_replied;
+
+		/* Do not overflow buffer */
+		if (bytes_replied > size)
+			return -1;
+
+		while (i < bytes_replied) {
+			uint32_t aux_sw_data_val;
+
+			REG_GET(AUX_SW_DATA, AUX_SW_DATA, &aux_sw_data_val);
+			buffer[i] = aux_sw_data_val;
+			++i;
+		}
+
+		return i;
+	}
+
+	return 0;
+}
+
+static enum aux_return_code_type get_channel_status(
+	struct dce_aux *engine,
+	uint8_t *returned_bytes)
+{
+	struct aux_engine_dce110 *aux110 = FROM_AUX_ENGINE(engine);
+
+	uint32_t value;
+
+	if (returned_bytes == NULL) {
+		/*caller pass NULL pointer*/
+		ASSERT_CRITICAL(false);
+		return AUX_RET_ERROR_UNKNOWN;
+	}
+	*returned_bytes = 0;
+
+	/* poll to make sure that SW_DONE is asserted */
+	REG_WAIT(AUX_SW_STATUS, AUX_SW_DONE, 1,
+				10, aux110->polling_timeout_period/10);
+
+	value = REG_READ(AUX_SW_STATUS);
+	/* in case HPD is LOW, exit AUX transaction */
+	if ((value & AUX_SW_STATUS__AUX_SW_HPD_DISCON_MASK))
+		return AUX_RET_ERROR_HPD_DISCON;
+
+	/* Note that the following bits are set in 'status.bits'
+	 * during CTS 4.2.1.2 (FW 3.3.1):
+	 * AUX_SW_RX_MIN_COUNT_VIOL, AUX_SW_RX_INVALID_STOP,
+	 * AUX_SW_RX_RECV_NO_DET, AUX_SW_RX_RECV_INVALID_H.
+	 *
+	 * AUX_SW_RX_MIN_COUNT_VIOL is an internal,
+	 * HW debugging bit and should be ignored.
+	 */
+	if (value & AUX_SW_STATUS__AUX_SW_DONE_MASK) {
+		if ((value & AUX_SW_STATUS__AUX_SW_RX_TIMEOUT_STATE_MASK) ||
+			(value & AUX_SW_STATUS__AUX_SW_RX_TIMEOUT_MASK))
+			return AUX_RET_ERROR_TIMEOUT;
+
+		else if ((value & AUX_SW_STATUS__AUX_SW_RX_INVALID_STOP_MASK) ||
+			(value & AUX_SW_STATUS__AUX_SW_RX_RECV_NO_DET_MASK) ||
+			(value &
+				AUX_SW_STATUS__AUX_SW_RX_RECV_INVALID_H_MASK) ||
+			(value & AUX_SW_STATUS__AUX_SW_RX_RECV_INVALID_L_MASK))
+			return AUX_RET_ERROR_INVALID_REPLY;
+
+		*returned_bytes = get_reg_field_value(value,
+				AUX_SW_STATUS,
+				AUX_SW_REPLY_BYTE_COUNT);
+
+		if (*returned_bytes == 0)
+			return
+			AUX_RET_ERROR_INVALID_REPLY;
+		else {
+			*returned_bytes -= 1;
+			return AUX_RET_SUCCESS;
+		}
+	} else {
+		/*time_elapsed >= aux_engine->timeout_period
+		 *  AUX_SW_STATUS__AUX_SW_HPD_DISCON = at this point
+		 */
+		ASSERT_CRITICAL(false);
+		return AUX_RET_ERROR_TIMEOUT;
+	}
+}
+
+static bool acquire(
+	struct dce_aux *engine,
+	struct ddc *ddc)
+{
+	enum gpio_result result;
+
+	if ((engine == NULL) || !is_engine_available(engine))
+		return false;
+
+	result = dal_ddc_open(ddc, GPIO_MODE_HARDWARE,
+		GPIO_DDC_CONFIG_TYPE_MODE_AUX);
+
+	if (result != GPIO_RESULT_OK)
+		return false;
+
+	if (!acquire_engine(engine)) {
+		engine->ddc = ddc;
+		release_engine(engine);
+		return false;
+	}
+
+	engine->ddc = ddc;
+
+	return true;
+}
+
+void dce110_engine_destroy(struct dce_aux **engine)
+{
+
+	struct aux_engine_dce110 *engine110 = FROM_AUX_ENGINE(*engine);
+
+	kfree(engine110);
+	*engine = NULL;
+
+}
+
+static uint32_t dce_aux_configure_timeout(struct ddc_service *ddc,
+		uint32_t timeout_in_us)
+{
+	uint32_t multiplier = 0;
+	uint32_t length = 0;
+	uint32_t prev_length = 0;
+	uint32_t prev_mult = 0;
+	uint32_t prev_timeout_val = 0;
+	struct ddc *ddc_pin = ddc->ddc_pin;
+	struct dce_aux *aux_engine = ddc->ctx->dc->res_pool->engines[ddc_pin->pin_data->en];
+	struct aux_engine_dce110 *aux110 = FROM_AUX_ENGINE(aux_engine);
+
+	/* 1-Update polling timeout period */
+	aux110->polling_timeout_period = timeout_in_us * SW_AUX_TIMEOUT_PERIOD_MULTIPLIER;
+
+	/* 2-Update aux timeout period length and multiplier */
+	if (timeout_in_us == 0) {
+		multiplier = DEFAULT_AUX_ENGINE_MULT;
+		length = DEFAULT_AUX_ENGINE_LENGTH;
+	} else if (timeout_in_us <= TIME_OUT_INCREMENT) {
+		multiplier = 0;
+		length = timeout_in_us/TIME_OUT_MULTIPLIER_8;
+		if (timeout_in_us % TIME_OUT_MULTIPLIER_8 != 0)
+			length++;
+	} else if (timeout_in_us <= 2 * TIME_OUT_INCREMENT) {
+		multiplier = 1;
+		length = timeout_in_us/TIME_OUT_MULTIPLIER_16;
+		if (timeout_in_us % TIME_OUT_MULTIPLIER_16 != 0)
+			length++;
+	} else if (timeout_in_us <= 4 * TIME_OUT_INCREMENT) {
+		multiplier = 2;
+		length = timeout_in_us/TIME_OUT_MULTIPLIER_32;
+		if (timeout_in_us % TIME_OUT_MULTIPLIER_32 != 0)
+			length++;
+	} else if (timeout_in_us > 4 * TIME_OUT_INCREMENT) {
+		multiplier = 3;
+		length = timeout_in_us/TIME_OUT_MULTIPLIER_64;
+		if (timeout_in_us % TIME_OUT_MULTIPLIER_64 != 0)
+			length++;
+	}
+
+	length = (length < MAX_TIMEOUT_LENGTH) ? length : MAX_TIMEOUT_LENGTH;
+
+	REG_GET_2(AUX_DPHY_RX_CONTROL1, AUX_RX_TIMEOUT_LEN, &prev_length, AUX_RX_TIMEOUT_LEN_MUL, &prev_mult);
+
+	switch (prev_mult) {
+	case 0:
+		prev_timeout_val = prev_length * TIME_OUT_MULTIPLIER_8;
+		break;
+	case 1:
+		prev_timeout_val = prev_length * TIME_OUT_MULTIPLIER_16;
+		break;
+	case 2:
+		prev_timeout_val = prev_length * TIME_OUT_MULTIPLIER_32;
+		break;
+	case 3:
+		prev_timeout_val = prev_length * TIME_OUT_MULTIPLIER_64;
+		break;
+	default:
+		prev_timeout_val = DEFAULT_AUX_ENGINE_LENGTH * TIME_OUT_MULTIPLIER_8;
+		break;
+	}
+
+	REG_UPDATE_SEQ_2(AUX_DPHY_RX_CONTROL1, AUX_RX_TIMEOUT_LEN, length, AUX_RX_TIMEOUT_LEN_MUL, multiplier);
+
+	return prev_timeout_val;
+}
+
+static struct dce_aux_funcs aux_functions = {
+	.configure_timeout = NULL,
+	.destroy = NULL,
+};
+
+struct dce_aux *dce110_aux_engine_construct(struct aux_engine_dce110 *aux_engine110,
+		struct dc_context *ctx,
+		uint32_t inst,
+		uint32_t timeout_period,
+		const struct dce110_aux_registers *regs,
+		const struct dce110_aux_registers_mask *mask,
+		const struct dce110_aux_registers_shift *shift,
+		bool is_ext_aux_timeout_configurable)
+{
+	aux_engine110->base.ddc = NULL;
+	aux_engine110->base.ctx = ctx;
+	aux_engine110->base.delay = 0;
+	aux_engine110->base.max_defer_write_retry = 0;
+	aux_engine110->base.inst = inst;
+	aux_engine110->polling_timeout_period = timeout_period;
+	aux_engine110->regs = regs;
+
+	aux_engine110->mask = mask;
+	aux_engine110->shift = shift;
+	aux_engine110->base.funcs = &aux_functions;
+	if (is_ext_aux_timeout_configurable)
+		aux_engine110->base.funcs->configure_timeout = &dce_aux_configure_timeout;
+
+	return &aux_engine110->base;
+}
+
+static enum i2caux_transaction_action i2caux_action_from_payload(struct aux_payload *payload)
+{
+	if (payload->i2c_over_aux) {
+		if (payload->write_status_update) {
+			if (payload->mot)
+				return I2CAUX_TRANSACTION_ACTION_I2C_STATUS_REQUEST_MOT;
+			else
+				return I2CAUX_TRANSACTION_ACTION_I2C_STATUS_REQUEST;
+		}
+		if (payload->write) {
+			if (payload->mot)
+				return I2CAUX_TRANSACTION_ACTION_I2C_WRITE_MOT;
+			else
+				return I2CAUX_TRANSACTION_ACTION_I2C_WRITE;
+		}
+		if (payload->mot)
+			return I2CAUX_TRANSACTION_ACTION_I2C_READ_MOT;
+
+		return I2CAUX_TRANSACTION_ACTION_I2C_READ;
+	}
+	if (payload->write)
+		return I2CAUX_TRANSACTION_ACTION_DP_WRITE;
+
+	return I2CAUX_TRANSACTION_ACTION_DP_READ;
+}
+
+int dce_aux_transfer_raw(struct ddc_service *ddc,
+		struct aux_payload *payload,
+		enum aux_return_code_type *operation_result)
+{
+	struct ddc *ddc_pin = ddc->ddc_pin;
+	struct dce_aux *aux_engine;
+	struct aux_request_transaction_data aux_req;
+	uint8_t returned_bytes = 0;
+	int res = -1;
+	uint32_t status;
+
+	memset(&aux_req, 0, sizeof(aux_req));
+
+	if (ddc_pin == NULL) {
+		*operation_result = AUX_RET_ERROR_ENGINE_ACQUIRE;
+		return -1;
+	}
+
+	aux_engine = ddc->ctx->dc->res_pool->engines[ddc_pin->pin_data->en];
+	if (!acquire(aux_engine, ddc_pin)) {
+		*operation_result = AUX_RET_ERROR_ENGINE_ACQUIRE;
+		return -1;
+	}
+
+	if (payload->i2c_over_aux)
+		aux_req.type = AUX_TRANSACTION_TYPE_I2C;
+	else
+		aux_req.type = AUX_TRANSACTION_TYPE_DP;
+
+	aux_req.action = i2caux_action_from_payload(payload);
+
+	aux_req.address = payload->address;
+	aux_req.delay = 0;
+	aux_req.length = payload->length;
+	aux_req.data = payload->data;
+
+	submit_channel_request(aux_engine, &aux_req);
+	*operation_result = get_channel_status(aux_engine, &returned_bytes);
+
+	if (*operation_result == AUX_RET_SUCCESS) {
+		int __maybe_unused bytes_replied = 0;
+
+		bytes_replied = read_channel_reply(aux_engine, payload->length,
+					 payload->data, payload->reply,
+					 &status);
+		EVENT_LOG_AUX_REP(aux_engine->ddc->pin_data->en,
+					EVENT_LOG_AUX_ORIGIN_NATIVE, *payload->reply,
+					bytes_replied, payload->data);
+		res = returned_bytes;
+	} else {
+		res = -1;
+	}
+
+	release_engine(aux_engine);
+	return res;
+}
+
+int dce_aux_transfer_dmub_raw(struct ddc_service *ddc,
+		struct aux_payload *payload,
+		enum aux_return_code_type *operation_result)
+{
+	struct ddc *ddc_pin = ddc->ddc_pin;
+
+	if (ddc_pin != NULL) {
+		struct dce_aux *aux_engine = ddc->ctx->dc->res_pool->engines[ddc_pin->pin_data->en];
+		/* XXX: Workaround to configure ddc channels for aux transactions */
+		if (!acquire(aux_engine, ddc_pin)) {
+			*operation_result = AUX_RET_ERROR_ENGINE_ACQUIRE;
+			return -1;
+		}
+		release_engine(aux_engine);
+	}
+
+	return dm_helper_dmub_aux_transfer_sync(ddc->ctx, ddc->link, payload, operation_result);
+}
+
+#define AUX_MAX_RETRIES 7
+#define AUX_MIN_DEFER_RETRIES 7
+#define AUX_MAX_DEFER_TIMEOUT_MS 50
+#define AUX_MAX_I2C_DEFER_RETRIES 7
+#define AUX_MAX_INVALID_REPLY_RETRIES 2
+#define AUX_MAX_TIMEOUT_RETRIES 3
+#define AUX_DEFER_DELAY_FOR_DPIA 4 /*ms*/
+
+static void dce_aux_log_payload(const char *payload_name,
+	unsigned char *payload, uint32_t length, uint32_t max_length_to_log)
+{
+	if (!IS_DC_I2CAUX_LOGGING_ENABLED())
+		return;
+
+	if (payload && length) {
+		char hex_str[128] = {0};
+		char *hex_str_ptr = &hex_str[0];
+		uint32_t hex_str_remaining = sizeof(hex_str);
+		unsigned char *payload_ptr = payload;
+		unsigned char *payload_max_to_log_ptr = payload_ptr + min(max_length_to_log, length);
+		unsigned int count;
+		char *padding = "";
+
+		while (payload_ptr < payload_max_to_log_ptr) {
+			count = snprintf_count(hex_str_ptr, hex_str_remaining, "%s%02X", padding, *payload_ptr);
+			padding = " ";
+			hex_str_remaining -= count;
+			hex_str_ptr += count;
+			payload_ptr++;
+		}
+
+		count = snprintf_count(hex_str_ptr, hex_str_remaining, "   ");
+		hex_str_remaining -= count;
+		hex_str_ptr += count;
+
+		payload_ptr = payload;
+		while (payload_ptr < payload_max_to_log_ptr) {
+			count = snprintf_count(hex_str_ptr, hex_str_remaining, "%c",
+				*payload_ptr >= ' ' ? *payload_ptr : '.');
+			hex_str_remaining -= count;
+			hex_str_ptr += count;
+			payload_ptr++;
+		}
+
+		DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_VERBOSE,
+					LOG_FLAG_I2cAux_DceAux,
+					"dce_aux_log_payload: %s: length=%u: data: %s%s",
+					payload_name,
+					length,
+					hex_str,
+					(length > max_length_to_log ? " (...)" : " "));
+	} else {
+		DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_VERBOSE,
+					LOG_FLAG_I2cAux_DceAux,
+					"dce_aux_log_payload: %s: length=%u: data: <empty payload>",
+					payload_name,
+					length);
+	}
+}
+
+bool dce_aux_transfer_with_retries(struct ddc_service *ddc,
+		struct aux_payload *payload)
+{
+	int i, ret = 0;
+	uint8_t reply;
+	bool payload_reply = true;
+	enum aux_return_code_type operation_result;
+	bool retry_on_defer = false;
+	struct ddc *ddc_pin = ddc->ddc_pin;
+	struct dce_aux *aux_engine = NULL;
+	struct aux_engine_dce110 *aux110 = NULL;
+	uint32_t defer_time_in_ms = 0;
+
+	int aux_ack_retries = 0,
+		aux_defer_retries = 0,
+		aux_i2c_defer_retries = 0,
+		aux_timeout_retries = 0,
+		aux_invalid_reply_retries = 0,
+		aux_ack_m_retries = 0;
+
+	if (ddc_pin) {
+		aux_engine = ddc->ctx->dc->res_pool->engines[ddc_pin->pin_data->en];
+		aux110 = FROM_AUX_ENGINE(aux_engine);
+	}
+
+	if (!payload->reply) {
+		payload_reply = false;
+		payload->reply = &reply;
+	}
+
+	for (i = 0; i < AUX_MAX_RETRIES; i++) {
+		DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_INFORMATION,
+					LOG_FLAG_I2cAux_DceAux,
+					"dce_aux_transfer_with_retries: link_index=%u: START: retry %d of %d: "
+					"address=0x%04x length=%u write=%d mot=%d is_i2c=%d is_dpia=%d ddc_hw_inst=%d",
+					ddc && ddc->link ? ddc->link->link_index : UINT_MAX,
+					i + 1,
+					(int)AUX_MAX_RETRIES,
+					payload->address,
+					payload->length,
+					(unsigned int) payload->write,
+					(unsigned int) payload->mot,
+					payload->i2c_over_aux,
+					(ddc->link->ep_type == DISPLAY_ENDPOINT_USB4_DPIA) ? true : false,
+					ddc->link->ddc_hw_inst);
+		if (payload->write)
+			dce_aux_log_payload("  write", payload->data, payload->length, 16);
+
+		/* Check whether aux to be processed via dmub or dcn directly */
+		if (ddc->ctx->dc->debug.enable_dmub_aux_for_legacy_ddc
+			|| ddc->ddc_pin == NULL) {
+			ret = dce_aux_transfer_dmub_raw(ddc, payload, &operation_result);
+		} else {
+			ret = dce_aux_transfer_raw(ddc, payload, &operation_result);
+		}
+
+		DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_INFORMATION,
+					LOG_FLAG_I2cAux_DceAux,
+					"dce_aux_transfer_with_retries: link_index=%u: END: retry %d of %d: "
+					"address=0x%04x length=%u write=%d mot=%d: ret=%d operation_result=%d "
+					"payload->reply=%u  is_i2c=%d is_dpia=%d ddc_hw_inst=%d",
+					ddc && ddc->link ? ddc->link->link_index : UINT_MAX,
+					i + 1,
+					(int)AUX_MAX_RETRIES,
+					payload->address,
+					payload->length,
+					(unsigned int) payload->write,
+					(unsigned int) payload->mot,
+					ret,
+					(int)operation_result,
+					(unsigned int) *payload->reply,
+					payload->i2c_over_aux,
+					(ddc->link->ep_type == DISPLAY_ENDPOINT_USB4_DPIA) ? true : false,
+					ddc->link->ddc_hw_inst);
+		if (!payload->write)
+			dce_aux_log_payload("  read", payload->data, ret > 0 ? ret : 0, 16);
+
+		switch (operation_result) {
+		case AUX_RET_SUCCESS:
+			aux_timeout_retries = 0;
+			aux_invalid_reply_retries = 0;
+
+			switch (*payload->reply) {
+			case AUX_TRANSACTION_REPLY_AUX_ACK:
+				DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_INFORMATION,
+							LOG_FLAG_I2cAux_DceAux,
+							"dce_aux_transfer_with_retries: AUX_RET_SUCCESS: AUX_TRANSACTION_REPLY_AUX_ACK");
+				if (!payload->write && payload->length != ret) {
+					if (++aux_ack_retries >= AUX_MAX_RETRIES) {
+						DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_ERROR,
+									LOG_FLAG_Error_I2cAux,
+									"dce_aux_transfer_with_retries: FAILURE: aux_ack_retries=%d >= AUX_MAX_RETRIES=%d",
+									aux_defer_retries,
+									AUX_MAX_RETRIES);
+						goto fail;
+					} else
+						udelay(300);
+				} else if (payload->write && ret > 0) {
+					/* sink requested more time to complete the write via AUX_ACKM */
+					if (++aux_ack_m_retries >= AUX_MAX_RETRIES) {
+						DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_ERROR,
+								LOG_FLAG_Error_I2cAux,
+								"dce_aux_transfer_with_retries: FAILURE: aux_ack_m_retries=%d >= AUX_MAX_RETRIES=%d",
+								aux_ack_m_retries,
+								AUX_MAX_RETRIES);
+						goto fail;
+					}
+
+					/* retry reading the write status until complete
+					 * NOTE: payload is modified here
+					 */
+					payload->write = false;
+					payload->write_status_update = true;
+					payload->length = 0;
+					udelay(300);
+				} else
+					return true;
+			break;
+
+			case AUX_TRANSACTION_REPLY_AUX_DEFER:
+				DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_INFORMATION,
+							LOG_FLAG_I2cAux_DceAux,
+							"dce_aux_transfer_with_retries: AUX_RET_SUCCESS: AUX_TRANSACTION_REPLY_AUX_DEFER");
+
+				/* polling_timeout_period is in us */
+				if (aux110)
+					defer_time_in_ms += aux110->polling_timeout_period / 1000;
+				else
+					defer_time_in_ms += AUX_DEFER_DELAY_FOR_DPIA;
+				++aux_defer_retries;
+				fallthrough;
+			case AUX_TRANSACTION_REPLY_I2C_OVER_AUX_DEFER:
+				if (*payload->reply == AUX_TRANSACTION_REPLY_I2C_OVER_AUX_DEFER)
+					DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_INFORMATION,
+								LOG_FLAG_I2cAux_DceAux,
+								"dce_aux_transfer_with_retries: AUX_RET_SUCCESS: AUX_TRANSACTION_REPLY_I2C_OVER_AUX_DEFER");
+
+				retry_on_defer = true;
+
+				if (aux_defer_retries >= AUX_MIN_DEFER_RETRIES
+						&& defer_time_in_ms >= AUX_MAX_DEFER_TIMEOUT_MS) {
+					DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_ERROR,
+								LOG_FLAG_Error_I2cAux,
+								"dce_aux_transfer_with_retries: FAILURE: aux_defer_retries=%d >= AUX_MIN_DEFER_RETRIES=%d && defer_time_in_ms=%d >= AUX_MAX_DEFER_TIMEOUT_MS=%d",
+								aux_defer_retries,
+								AUX_MIN_DEFER_RETRIES,
+								defer_time_in_ms,
+								AUX_MAX_DEFER_TIMEOUT_MS);
+					goto fail;
+				} else {
+					if ((*payload->reply == AUX_TRANSACTION_REPLY_AUX_DEFER) ||
+						(*payload->reply == AUX_TRANSACTION_REPLY_I2C_OVER_AUX_DEFER)) {
+						DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_INFORMATION,
+									LOG_FLAG_I2cAux_DceAux,
+									"dce_aux_transfer_with_retries: payload->defer_delay=%u",
+									payload->defer_delay);
+						fsleep(payload->defer_delay * 1000);
+						defer_time_in_ms += payload->defer_delay;
+					}
+				}
+				break;
+			case AUX_TRANSACTION_REPLY_I2C_OVER_AUX_NACK:
+				DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_INFORMATION,
+							LOG_FLAG_I2cAux_DceAux,
+							"dce_aux_transfer_with_retries: FAILURE: AUX_TRANSACTION_REPLY_I2C_OVER_AUX_NACK");
+				goto fail;
+			case AUX_TRANSACTION_REPLY_I2C_DEFER:
+				DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_INFORMATION,
+							LOG_FLAG_I2cAux_DceAux,
+							"dce_aux_transfer_with_retries: AUX_RET_SUCCESS: AUX_TRANSACTION_REPLY_I2C_DEFER");
+
+				aux_defer_retries = 0;
+				if (++aux_i2c_defer_retries >= AUX_MAX_I2C_DEFER_RETRIES) {
+					DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_ERROR,
+								LOG_FLAG_Error_I2cAux,
+								"dce_aux_transfer_with_retries: FAILURE: aux_i2c_defer_retries=%d >= AUX_MAX_I2C_DEFER_RETRIES=%d",
+								aux_i2c_defer_retries,
+								AUX_MAX_I2C_DEFER_RETRIES);
+					goto fail;
+				}
+				break;
+
+			case AUX_TRANSACTION_REPLY_AUX_NACK:
+				DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_INFORMATION,
+							LOG_FLAG_I2cAux_DceAux,
+							"dce_aux_transfer_with_retries: AUX_RET_SUCCESS: AUX_TRANSACTION_REPLY_AUX_NACK");
+				goto fail;
+
+			case AUX_TRANSACTION_REPLY_HPD_DISCON:
+				DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_INFORMATION,
+							LOG_FLAG_I2cAux_DceAux,
+							"dce_aux_transfer_with_retries: AUX_RET_SUCCESS: AUX_TRANSACTION_REPLY_HPD_DISCON");
+				goto fail;
+
+			default:
+				DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_ERROR,
+							LOG_FLAG_Error_I2cAux,
+							"dce_aux_transfer_with_retries: AUX_RET_SUCCESS: FAILURE: AUX_TRANSACTION_REPLY_* unknown, default case. Reply: %d", *payload->reply);
+				goto fail;
+			}
+			break;
+
+		case AUX_RET_ERROR_INVALID_REPLY:
+			DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_INFORMATION,
+						LOG_FLAG_I2cAux_DceAux,
+						"dce_aux_transfer_with_retries: AUX_RET_ERROR_INVALID_REPLY");
+			if (++aux_invalid_reply_retries >= AUX_MAX_INVALID_REPLY_RETRIES) {
+				DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_ERROR,
+							LOG_FLAG_Error_I2cAux,
+							"dce_aux_transfer_with_retries: FAILURE: aux_invalid_reply_retries=%d >= AUX_MAX_INVALID_REPLY_RETRIES=%d",
+							aux_invalid_reply_retries,
+							AUX_MAX_INVALID_REPLY_RETRIES);
+				goto fail;
+			} else
+				udelay(400);
+			break;
+
+		case AUX_RET_ERROR_TIMEOUT:
+			DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_INFORMATION,
+						LOG_FLAG_I2cAux_DceAux,
+						"dce_aux_transfer_with_retries: AUX_RET_ERROR_TIMEOUT");
+			// Check whether a DEFER had occurred before the timeout.
+			// If so, treat timeout as a DEFER.
+			if (retry_on_defer) {
+				if (++aux_defer_retries >= AUX_MIN_DEFER_RETRIES) {
+					DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_ERROR,
+								LOG_FLAG_Error_I2cAux,
+								"dce_aux_transfer_with_retries: FAILURE: aux_defer_retries=%d >= AUX_MIN_DEFER_RETRIES=%d",
+								aux_defer_retries,
+								AUX_MIN_DEFER_RETRIES);
+					goto fail;
+				} else if (payload->defer_delay > 0) {
+					DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_INFORMATION,
+								LOG_FLAG_I2cAux_DceAux,
+								"dce_aux_transfer_with_retries: payload->defer_delay=%u",
+								payload->defer_delay);
+					msleep(payload->defer_delay);
+				}
+			} else {
+				if (++aux_timeout_retries >= AUX_MAX_TIMEOUT_RETRIES) {
+					DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_ERROR,
+								LOG_FLAG_Error_I2cAux,
+								"dce_aux_transfer_with_retries: FAILURE: aux_timeout_retries=%d >= AUX_MAX_TIMEOUT_RETRIES=%d",
+								aux_timeout_retries,
+								AUX_MAX_TIMEOUT_RETRIES);
+					goto fail;
+				} else {
+					/*
+					 * DP 1.4, 2.8.2:  AUX Transaction Response/Reply Timeouts
+					 * According to the DP spec there should be 3 retries total
+					 * with a 400us wait inbetween each. Hardware already waits
+					 * for 550us therefore no wait is required here.
+					 */
+				}
+			}
+			break;
+
+		case AUX_RET_ERROR_HPD_DISCON:
+		case AUX_RET_ERROR_ENGINE_ACQUIRE:
+		case AUX_RET_ERROR_UNKNOWN:
+		default:
+			goto fail;
+		}
+	}
+
+fail:
+	DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_ERROR,
+				LOG_FLAG_Error_I2cAux,
+				"%s: Failure: operation_result=%d",
+				__func__,
+				(int)operation_result);
+	if (!payload_reply)
+		payload->reply = NULL;
+
+	return false;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_aux.h b/drivers/gpu/drm/amd/display/dc/dce/dce_aux.h
new file mode 100644
index 000000000000..c850ed49281f
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_aux.h
@@ -0,0 +1,321 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_AUX_ENGINE_DCE110_H__
+#define __DAL_AUX_ENGINE_DCE110_H__
+
+#include "gpio_service_interface.h"
+#include "inc/hw/aux_engine.h"
+
+enum aux_return_code_type;
+
+#define AUX_COMMON_REG_LIST0(id)\
+	SRI(AUX_CONTROL, DP_AUX, id), \
+	SRI(AUX_ARB_CONTROL, DP_AUX, id), \
+	SRI(AUX_SW_DATA, DP_AUX, id), \
+	SRI(AUX_SW_CONTROL, DP_AUX, id), \
+	SRI(AUX_INTERRUPT_CONTROL, DP_AUX, id), \
+	SRI(AUX_DPHY_RX_CONTROL1, DP_AUX, id), \
+	SRI(AUX_SW_STATUS, DP_AUX, id)
+
+#define AUX_COMMON_REG_LIST(id)\
+	SRI(AUX_CONTROL, DP_AUX, id), \
+	SRI(AUX_ARB_CONTROL, DP_AUX, id), \
+	SRI(AUX_SW_DATA, DP_AUX, id), \
+	SRI(AUX_SW_CONTROL, DP_AUX, id), \
+	SRI(AUX_INTERRUPT_CONTROL, DP_AUX, id), \
+	SRI(AUX_SW_STATUS, DP_AUX, id), \
+	SR(AUXN_IMPCAL), \
+	SR(AUXP_IMPCAL)
+
+struct dce110_aux_registers {
+	uint32_t AUX_CONTROL;
+	uint32_t AUX_ARB_CONTROL;
+	uint32_t AUX_SW_DATA;
+	uint32_t AUX_SW_CONTROL;
+	uint32_t AUX_INTERRUPT_CONTROL;
+	uint32_t AUX_DPHY_RX_CONTROL1;
+	uint32_t AUX_SW_STATUS;
+	uint32_t AUXN_IMPCAL;
+	uint32_t AUXP_IMPCAL;
+
+	uint32_t AUX_RESET_MASK;
+};
+
+#define DCE_AUX_REG_FIELD_LIST(type)\
+	type AUX_EN;\
+	type AUX_RESET;\
+	type AUX_RESET_DONE;\
+	type AUX_REG_RW_CNTL_STATUS;\
+	type AUX_SW_USE_AUX_REG_REQ;\
+	type AUX_SW_DONE_USING_AUX_REG;\
+	type AUX_SW_AUTOINCREMENT_DISABLE;\
+	type AUX_SW_DATA_RW;\
+	type AUX_SW_INDEX;\
+	type AUX_SW_GO;\
+	type AUX_SW_DATA;\
+	type AUX_SW_REPLY_BYTE_COUNT;\
+	type AUX_SW_DONE;\
+	type AUX_SW_DONE_ACK;\
+	type AUXN_IMPCAL_ENABLE;\
+	type AUXP_IMPCAL_ENABLE;\
+	type AUXN_IMPCAL_OVERRIDE_ENABLE;\
+	type AUXP_IMPCAL_OVERRIDE_ENABLE;\
+	type AUX_RX_TIMEOUT_LEN;\
+	type AUX_RX_TIMEOUT_LEN_MUL;\
+	type AUXN_CALOUT_ERROR_AK;\
+	type AUXP_CALOUT_ERROR_AK;\
+	type AUX_SW_START_DELAY;\
+	type AUX_SW_WR_BYTES
+
+#define DCE10_AUX_MASK_SH_LIST(mask_sh)\
+	AUX_SF(AUX_CONTROL, AUX_EN, mask_sh),\
+	AUX_SF(AUX_ARB_CONTROL, AUX_REG_RW_CNTL_STATUS, mask_sh),\
+	AUX_SF(AUX_ARB_CONTROL, AUX_SW_USE_AUX_REG_REQ, mask_sh),\
+	AUX_SF(AUX_ARB_CONTROL, AUX_SW_DONE_USING_AUX_REG, mask_sh),\
+	AUX_SF(AUX_SW_CONTROL, AUX_SW_START_DELAY, mask_sh),\
+	AUX_SF(AUX_SW_CONTROL, AUX_SW_WR_BYTES, mask_sh),\
+	AUX_SF(AUX_SW_CONTROL, AUX_SW_GO, mask_sh),\
+	AUX_SF(AUX_SW_DATA, AUX_SW_AUTOINCREMENT_DISABLE, mask_sh),\
+	AUX_SF(AUX_SW_DATA, AUX_SW_DATA_RW, mask_sh),\
+	AUX_SF(AUX_SW_DATA, AUX_SW_INDEX, mask_sh),\
+	AUX_SF(AUX_SW_DATA, AUX_SW_DATA, mask_sh),\
+	AUX_SF(AUX_SW_STATUS, AUX_SW_REPLY_BYTE_COUNT, mask_sh),\
+	AUX_SF(AUX_SW_STATUS, AUX_SW_DONE, mask_sh),\
+	AUX_SF(AUX_INTERRUPT_CONTROL, AUX_SW_DONE_ACK, mask_sh),\
+	AUX_SF(AUXN_IMPCAL, AUXN_CALOUT_ERROR_AK, mask_sh),\
+	AUX_SF(AUXP_IMPCAL, AUXP_CALOUT_ERROR_AK, mask_sh),\
+	AUX_SF(AUXN_IMPCAL, AUXN_IMPCAL_ENABLE, mask_sh),\
+	AUX_SF(AUXP_IMPCAL, AUXP_IMPCAL_ENABLE, mask_sh),\
+	AUX_SF(AUXP_IMPCAL, AUXP_IMPCAL_OVERRIDE_ENABLE, mask_sh),\
+	AUX_SF(AUXN_IMPCAL, AUXN_IMPCAL_OVERRIDE_ENABLE, mask_sh)
+
+#define DCE_AUX_MASK_SH_LIST(mask_sh)\
+	AUX_SF(AUX_CONTROL, AUX_EN, mask_sh),\
+	AUX_SF(AUX_CONTROL, AUX_RESET, mask_sh),\
+	AUX_SF(AUX_CONTROL, AUX_RESET_DONE, mask_sh),\
+	AUX_SF(AUX_ARB_CONTROL, AUX_REG_RW_CNTL_STATUS, mask_sh),\
+	AUX_SF(AUX_ARB_CONTROL, AUX_SW_USE_AUX_REG_REQ, mask_sh),\
+	AUX_SF(AUX_ARB_CONTROL, AUX_SW_DONE_USING_AUX_REG, mask_sh),\
+	AUX_SF(AUX_SW_CONTROL, AUX_SW_START_DELAY, mask_sh),\
+	AUX_SF(AUX_SW_CONTROL, AUX_SW_WR_BYTES, mask_sh),\
+	AUX_SF(AUX_SW_CONTROL, AUX_SW_GO, mask_sh),\
+	AUX_SF(AUX_SW_DATA, AUX_SW_AUTOINCREMENT_DISABLE, mask_sh),\
+	AUX_SF(AUX_SW_DATA, AUX_SW_DATA_RW, mask_sh),\
+	AUX_SF(AUX_SW_DATA, AUX_SW_INDEX, mask_sh),\
+	AUX_SF(AUX_SW_DATA, AUX_SW_DATA, mask_sh),\
+	AUX_SF(AUX_SW_STATUS, AUX_SW_REPLY_BYTE_COUNT, mask_sh),\
+	AUX_SF(AUX_SW_STATUS, AUX_SW_DONE, mask_sh),\
+	AUX_SF(AUX_INTERRUPT_CONTROL, AUX_SW_DONE_ACK, mask_sh),\
+	AUX_SF(AUXN_IMPCAL, AUXN_CALOUT_ERROR_AK, mask_sh),\
+	AUX_SF(AUXP_IMPCAL, AUXP_CALOUT_ERROR_AK, mask_sh),\
+	AUX_SF(AUXN_IMPCAL, AUXN_IMPCAL_ENABLE, mask_sh),\
+	AUX_SF(AUXP_IMPCAL, AUXP_IMPCAL_ENABLE, mask_sh),\
+	AUX_SF(AUXP_IMPCAL, AUXP_IMPCAL_OVERRIDE_ENABLE, mask_sh),\
+	AUX_SF(AUXN_IMPCAL, AUXN_IMPCAL_OVERRIDE_ENABLE, mask_sh)
+
+#define DCE12_AUX_MASK_SH_LIST(mask_sh)\
+	AUX_SF(DP_AUX0_AUX_CONTROL, AUX_EN, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_CONTROL, AUX_RESET, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_CONTROL, AUX_RESET_DONE, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_ARB_CONTROL, AUX_REG_RW_CNTL_STATUS, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_ARB_CONTROL, AUX_SW_USE_AUX_REG_REQ, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_ARB_CONTROL, AUX_SW_DONE_USING_AUX_REG, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_CONTROL, AUX_SW_START_DELAY, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_CONTROL, AUX_SW_WR_BYTES, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_CONTROL, AUX_SW_GO, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_DATA, AUX_SW_AUTOINCREMENT_DISABLE, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_DATA, AUX_SW_DATA_RW, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_DATA, AUX_SW_AUTOINCREMENT_DISABLE, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_DATA, AUX_SW_INDEX, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_DATA, AUX_SW_DATA, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_STATUS, AUX_SW_REPLY_BYTE_COUNT, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_STATUS, AUX_SW_DONE, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_INTERRUPT_CONTROL, AUX_SW_DONE_ACK, mask_sh),\
+	AUX_SF(AUXN_IMPCAL, AUXN_CALOUT_ERROR_AK, mask_sh),\
+	AUX_SF(AUXP_IMPCAL, AUXP_CALOUT_ERROR_AK, mask_sh),\
+	AUX_SF(AUXN_IMPCAL, AUXN_IMPCAL_ENABLE, mask_sh),\
+	AUX_SF(AUXP_IMPCAL, AUXP_IMPCAL_ENABLE, mask_sh),\
+	AUX_SF(AUXP_IMPCAL, AUXP_IMPCAL_OVERRIDE_ENABLE, mask_sh),\
+	AUX_SF(AUXN_IMPCAL, AUXN_IMPCAL_OVERRIDE_ENABLE, mask_sh)
+
+/* DCN10 MASK */
+#define DCN10_AUX_MASK_SH_LIST(mask_sh)\
+	AUX_SF(DP_AUX0_AUX_CONTROL, AUX_EN, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_CONTROL, AUX_RESET, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_CONTROL, AUX_RESET_DONE, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_ARB_CONTROL, AUX_REG_RW_CNTL_STATUS, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_ARB_CONTROL, AUX_SW_USE_AUX_REG_REQ, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_ARB_CONTROL, AUX_SW_DONE_USING_AUX_REG, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_CONTROL, AUX_SW_START_DELAY, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_CONTROL, AUX_SW_WR_BYTES, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_CONTROL, AUX_SW_GO, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_DATA, AUX_SW_AUTOINCREMENT_DISABLE, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_DATA, AUX_SW_DATA_RW, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_DATA, AUX_SW_AUTOINCREMENT_DISABLE, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_DATA, AUX_SW_INDEX, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_DATA, AUX_SW_DATA, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_STATUS, AUX_SW_REPLY_BYTE_COUNT, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_STATUS, AUX_SW_DONE, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_INTERRUPT_CONTROL, AUX_SW_DONE_ACK, mask_sh),\
+	AUX_SF(AUXN_IMPCAL, AUXN_CALOUT_ERROR_AK, mask_sh),\
+	AUX_SF(AUXP_IMPCAL, AUXP_CALOUT_ERROR_AK, mask_sh),\
+	AUX_SF(AUXN_IMPCAL, AUXN_IMPCAL_ENABLE, mask_sh),\
+	AUX_SF(AUXP_IMPCAL, AUXP_IMPCAL_ENABLE, mask_sh),\
+	AUX_SF(AUXP_IMPCAL, AUXP_IMPCAL_OVERRIDE_ENABLE, mask_sh),\
+	AUX_SF(AUXN_IMPCAL, AUXN_IMPCAL_OVERRIDE_ENABLE, mask_sh)
+
+/* for all other DCN */
+#define DCN_AUX_MASK_SH_LIST(mask_sh)\
+	AUX_SF(DP_AUX0_AUX_CONTROL, AUX_EN, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_CONTROL, AUX_RESET, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_CONTROL, AUX_RESET_DONE, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_ARB_CONTROL, AUX_REG_RW_CNTL_STATUS, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_ARB_CONTROL, AUX_SW_USE_AUX_REG_REQ, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_ARB_CONTROL, AUX_SW_DONE_USING_AUX_REG, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_CONTROL, AUX_SW_START_DELAY, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_CONTROL, AUX_SW_WR_BYTES, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_CONTROL, AUX_SW_GO, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_DATA, AUX_SW_AUTOINCREMENT_DISABLE, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_DATA, AUX_SW_DATA_RW, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_DATA, AUX_SW_AUTOINCREMENT_DISABLE, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_DATA, AUX_SW_INDEX, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_DATA, AUX_SW_DATA, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_STATUS, AUX_SW_REPLY_BYTE_COUNT, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_SW_STATUS, AUX_SW_DONE, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_INTERRUPT_CONTROL, AUX_SW_DONE_ACK, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_DPHY_RX_CONTROL1, AUX_RX_TIMEOUT_LEN, mask_sh),\
+	AUX_SF(DP_AUX0_AUX_DPHY_RX_CONTROL1, AUX_RX_TIMEOUT_LEN_MUL, mask_sh)
+
+#define AUX_SF(reg_name, field_name, post_fix)\
+	.field_name = reg_name ## __ ## field_name ## post_fix
+
+enum {	/* This is the timeout as defined in DP 1.2a,
+	 * 2.3.4 "Detailed uPacket TX AUX CH State Description".
+	 */
+	AUX_TIMEOUT_PERIOD = 400,
+
+	/* Ideally, the SW timeout should be just above 550usec
+	 * which is programmed in HW.
+	 * But the SW timeout of 600usec is not reliable,
+	 * because on some systems, delay_in_microseconds()
+	 * returns faster than it should.
+	 * EPR #379763: by trial-and-error on different systems,
+	 * 700usec is the minimum reliable SW timeout for polling
+	 * the AUX_SW_STATUS.AUX_SW_DONE bit.
+	 * This timeout expires *only* when there is
+	 * AUX Error or AUX Timeout conditions - not during normal operation.
+	 * During normal operation, AUX_SW_STATUS.AUX_SW_DONE bit is set
+	 * at most within ~240usec. That means,
+	 * increasing this timeout will not affect normal operation,
+	 * and we'll timeout after
+	 * SW_AUX_TIMEOUT_PERIOD_MULTIPLIER * AUX_TIMEOUT_PERIOD = 2400usec.
+	 * This timeout is especially important for
+	 * converters, resume from S3, and CTS.
+	 */
+	SW_AUX_TIMEOUT_PERIOD_MULTIPLIER = 6
+};
+
+struct dce_aux {
+	uint32_t inst;
+	struct ddc *ddc;
+	struct dc_context *ctx;
+	/* following values are expressed in milliseconds */
+	uint32_t delay;
+	uint32_t max_defer_write_retry;
+
+	bool acquire_reset;
+	struct dce_aux_funcs *funcs;
+};
+
+struct dce110_aux_registers_mask {
+	DCE_AUX_REG_FIELD_LIST(uint32_t);
+};
+
+struct dce110_aux_registers_shift {
+	DCE_AUX_REG_FIELD_LIST(uint8_t);
+};
+
+
+struct aux_engine_dce110 {
+	struct dce_aux base;
+	const struct dce110_aux_registers *regs;
+	const struct dce110_aux_registers_mask *mask;
+	const struct dce110_aux_registers_shift *shift;
+	struct {
+		uint32_t aux_control;
+		uint32_t aux_arb_control;
+		uint32_t aux_sw_data;
+		uint32_t aux_sw_control;
+		uint32_t aux_interrupt_control;
+		uint32_t aux_dphy_rx_control1;
+		uint32_t aux_dphy_rx_control0;
+		uint32_t aux_sw_status;
+	} addr;
+	uint32_t polling_timeout_period;
+};
+
+struct aux_engine_dce110_init_data {
+	uint32_t engine_id;
+	uint32_t timeout_period;
+	struct dc_context *ctx;
+	const struct dce110_aux_registers *regs;
+};
+
+struct dce_aux *dce110_aux_engine_construct(struct aux_engine_dce110 *aux_engine110,
+		struct dc_context *ctx,
+		uint32_t inst,
+		uint32_t timeout_period,
+		const struct dce110_aux_registers *regs,
+
+		const struct dce110_aux_registers_mask *mask,
+		const struct dce110_aux_registers_shift *shift,
+		bool is_ext_aux_timeout_configurable);
+
+void dce110_engine_destroy(struct dce_aux **engine);
+
+bool dce110_aux_engine_acquire(
+	struct dce_aux *aux_engine,
+	struct ddc *ddc);
+
+int dce_aux_transfer_raw(struct ddc_service *ddc,
+		struct aux_payload *cmd,
+		enum aux_return_code_type *operation_result);
+
+int dce_aux_transfer_dmub_raw(struct ddc_service *ddc,
+		struct aux_payload *payload,
+		enum aux_return_code_type *operation_result);
+bool dce_aux_transfer_with_retries(struct ddc_service *ddc,
+		struct aux_payload *cmd);
+
+struct dce_aux_funcs {
+	uint32_t (*configure_timeout)
+		(struct ddc_service *ddc,
+		 uint32_t timeout);
+	void (*destroy)
+		(struct aux_engine **ptr);
+};
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/dce/dce_clk_mgr.c
new file mode 100644
index 000000000000..e7acd6eec1fd
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_clk_mgr.c
@@ -0,0 +1,967 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include <linux/slab.h>
+
+#include "dce_clk_mgr.h"
+
+#include "reg_helper.h"
+#include "dmcu.h"
+#include "core_types.h"
+#include "dal_asic_id.h"
+
+#define TO_DCE_CLK_MGR(clocks)\
+	container_of(clocks, struct dce_clk_mgr, base)
+
+#define REG(reg) \
+	(clk_mgr_dce->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	clk_mgr_dce->clk_mgr_shift->field_name, clk_mgr_dce->clk_mgr_mask->field_name
+
+#define CTX \
+	clk_mgr_dce->base.ctx
+#define DC_LOGGER \
+	clk_mgr->ctx->logger
+
+/* Max clock values for each state indexed by "enum clocks_state": */
+static const struct state_dependent_clocks dce80_max_clks_by_state[] = {
+/* ClocksStateInvalid - should not be used */
+{ .display_clk_khz = 0, .pixel_clk_khz = 0 },
+/* ClocksStateUltraLow - not expected to be used for DCE 8.0 */
+{ .display_clk_khz = 0, .pixel_clk_khz = 0 },
+/* ClocksStateLow */
+{ .display_clk_khz = 352000, .pixel_clk_khz = 330000},
+/* ClocksStateNominal */
+{ .display_clk_khz = 600000, .pixel_clk_khz = 400000 },
+/* ClocksStatePerformance */
+{ .display_clk_khz = 600000, .pixel_clk_khz = 400000 } };
+
+static const struct state_dependent_clocks dce110_max_clks_by_state[] = {
+/*ClocksStateInvalid - should not be used*/
+{ .display_clk_khz = 0, .pixel_clk_khz = 0 },
+/*ClocksStateUltraLow - currently by HW design team not supposed to be used*/
+{ .display_clk_khz = 352000, .pixel_clk_khz = 330000 },
+/*ClocksStateLow*/
+{ .display_clk_khz = 352000, .pixel_clk_khz = 330000 },
+/*ClocksStateNominal*/
+{ .display_clk_khz = 467000, .pixel_clk_khz = 400000 },
+/*ClocksStatePerformance*/
+{ .display_clk_khz = 643000, .pixel_clk_khz = 400000 } };
+
+static const struct state_dependent_clocks dce112_max_clks_by_state[] = {
+/*ClocksStateInvalid - should not be used*/
+{ .display_clk_khz = 0, .pixel_clk_khz = 0 },
+/*ClocksStateUltraLow - currently by HW design team not supposed to be used*/
+{ .display_clk_khz = 389189, .pixel_clk_khz = 346672 },
+/*ClocksStateLow*/
+{ .display_clk_khz = 459000, .pixel_clk_khz = 400000 },
+/*ClocksStateNominal*/
+{ .display_clk_khz = 667000, .pixel_clk_khz = 600000 },
+/*ClocksStatePerformance*/
+{ .display_clk_khz = 1132000, .pixel_clk_khz = 600000 } };
+
+static const struct state_dependent_clocks dce120_max_clks_by_state[] = {
+/*ClocksStateInvalid - should not be used*/
+{ .display_clk_khz = 0, .pixel_clk_khz = 0 },
+/*ClocksStateUltraLow - currently by HW design team not supposed to be used*/
+{ .display_clk_khz = 0, .pixel_clk_khz = 0 },
+/*ClocksStateLow*/
+{ .display_clk_khz = 460000, .pixel_clk_khz = 400000 },
+/*ClocksStateNominal*/
+{ .display_clk_khz = 670000, .pixel_clk_khz = 600000 },
+/*ClocksStatePerformance*/
+{ .display_clk_khz = 1133000, .pixel_clk_khz = 600000 } };
+
+int dentist_get_divider_from_did(int did)
+{
+	if (did < DENTIST_BASE_DID_1)
+		did = DENTIST_BASE_DID_1;
+	if (did > DENTIST_MAX_DID)
+		did = DENTIST_MAX_DID;
+
+	if (did < DENTIST_BASE_DID_2) {
+		return DENTIST_DIVIDER_RANGE_1_START + DENTIST_DIVIDER_RANGE_1_STEP
+							* (did - DENTIST_BASE_DID_1);
+	} else if (did < DENTIST_BASE_DID_3) {
+		return DENTIST_DIVIDER_RANGE_2_START + DENTIST_DIVIDER_RANGE_2_STEP
+							* (did - DENTIST_BASE_DID_2);
+	} else if (did < DENTIST_BASE_DID_4) {
+		return DENTIST_DIVIDER_RANGE_3_START + DENTIST_DIVIDER_RANGE_3_STEP
+							* (did - DENTIST_BASE_DID_3);
+	} else {
+		return DENTIST_DIVIDER_RANGE_4_START + DENTIST_DIVIDER_RANGE_4_STEP
+							* (did - DENTIST_BASE_DID_4);
+	}
+}
+
+/* SW will adjust DP REF Clock average value for all purposes
+ * (DP DTO / DP Audio DTO and DP GTC)
+ if clock is spread for all cases:
+ -if SS enabled on DP Ref clock and HW de-spreading enabled with SW
+ calculations for DS_INCR/DS_MODULO (this is planned to be default case)
+ -if SS enabled on DP Ref clock and HW de-spreading enabled with HW
+ calculations (not planned to be used, but average clock should still
+ be valid)
+ -if SS enabled on DP Ref clock and HW de-spreading disabled
+ (should not be case with CIK) then SW should program all rates
+ generated according to average value (case as with previous ASICs)
+  */
+static int clk_mgr_adjust_dp_ref_freq_for_ss(struct dce_clk_mgr *clk_mgr_dce, int dp_ref_clk_khz)
+{
+	if (clk_mgr_dce->ss_on_dprefclk && clk_mgr_dce->dprefclk_ss_divider != 0) {
+		struct fixed31_32 ss_percentage = dc_fixpt_div_int(
+				dc_fixpt_from_fraction(clk_mgr_dce->dprefclk_ss_percentage,
+							clk_mgr_dce->dprefclk_ss_divider), 200);
+		struct fixed31_32 adj_dp_ref_clk_khz;
+
+		ss_percentage = dc_fixpt_sub(dc_fixpt_one, ss_percentage);
+		adj_dp_ref_clk_khz = dc_fixpt_mul_int(ss_percentage, dp_ref_clk_khz);
+		dp_ref_clk_khz = dc_fixpt_floor(adj_dp_ref_clk_khz);
+	}
+	return dp_ref_clk_khz;
+}
+
+static int dce_get_dp_ref_freq_khz(struct clk_mgr *clk_mgr)
+{
+	struct dce_clk_mgr *clk_mgr_dce = TO_DCE_CLK_MGR(clk_mgr);
+	int dprefclk_wdivider;
+	int dprefclk_src_sel;
+	int dp_ref_clk_khz = 600000;
+	int target_div;
+
+	/* ASSERT DP Reference Clock source is from DFS*/
+	REG_GET(DPREFCLK_CNTL, DPREFCLK_SRC_SEL, &dprefclk_src_sel);
+	ASSERT(dprefclk_src_sel == 0);
+
+	/* Read the mmDENTIST_DISPCLK_CNTL to get the currently
+	 * programmed DID DENTIST_DPREFCLK_WDIVIDER*/
+	REG_GET(DENTIST_DISPCLK_CNTL, DENTIST_DPREFCLK_WDIVIDER, &dprefclk_wdivider);
+
+	/* Convert DENTIST_DPREFCLK_WDIVIDERto actual divider*/
+	target_div = dentist_get_divider_from_did(dprefclk_wdivider);
+
+	/* Calculate the current DFS clock, in kHz.*/
+	dp_ref_clk_khz = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+		* clk_mgr_dce->dentist_vco_freq_khz) / target_div;
+
+	return clk_mgr_adjust_dp_ref_freq_for_ss(clk_mgr_dce, dp_ref_clk_khz);
+}
+
+int dce12_get_dp_ref_freq_khz(struct clk_mgr *clk_mgr)
+{
+	struct dce_clk_mgr *clk_mgr_dce = TO_DCE_CLK_MGR(clk_mgr);
+
+	return clk_mgr_adjust_dp_ref_freq_for_ss(clk_mgr_dce, clk_mgr_dce->dprefclk_khz);
+}
+
+/* unit: in_khz before mode set, get pixel clock from context. ASIC register
+ * may not be programmed yet
+ */
+static uint32_t get_max_pixel_clock_for_all_paths(struct dc_state *context)
+{
+	uint32_t max_pix_clk = 0;
+	int i;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i];
+
+		if (pipe_ctx->stream == NULL)
+			continue;
+
+		/* do not check under lay */
+		if (pipe_ctx->top_pipe)
+			continue;
+
+		if (pipe_ctx->stream_res.pix_clk_params.requested_pix_clk_100hz / 10 > max_pix_clk)
+			max_pix_clk = pipe_ctx->stream_res.pix_clk_params.requested_pix_clk_100hz / 10;
+
+		/* raise clock state for HBR3/2 if required. Confirmed with HW DCE/DPCS
+		 * logic for HBR3 still needs Nominal (0.8V) on VDDC rail
+		 */
+		if (dc_is_dp_signal(pipe_ctx->stream->signal) &&
+				pipe_ctx->stream_res.pix_clk_params.requested_sym_clk > max_pix_clk)
+			max_pix_clk = pipe_ctx->stream_res.pix_clk_params.requested_sym_clk;
+	}
+
+	return max_pix_clk;
+}
+
+static enum dm_pp_clocks_state dce_get_required_clocks_state(
+	struct clk_mgr *clk_mgr,
+	struct dc_state *context)
+{
+	struct dce_clk_mgr *clk_mgr_dce = TO_DCE_CLK_MGR(clk_mgr);
+	int i;
+	enum dm_pp_clocks_state low_req_clk;
+	int max_pix_clk = get_max_pixel_clock_for_all_paths(context);
+
+	/* Iterate from highest supported to lowest valid state, and update
+	 * lowest RequiredState with the lowest state that satisfies
+	 * all required clocks
+	 */
+	for (i = clk_mgr_dce->max_clks_state; i >= DM_PP_CLOCKS_STATE_ULTRA_LOW; i--)
+		if (context->bw_ctx.bw.dce.dispclk_khz >
+				clk_mgr_dce->max_clks_by_state[i].display_clk_khz
+			|| max_pix_clk >
+				clk_mgr_dce->max_clks_by_state[i].pixel_clk_khz)
+			break;
+
+	low_req_clk = i + 1;
+	if (low_req_clk > clk_mgr_dce->max_clks_state) {
+		/* set max clock state for high phyclock, invalid on exceeding display clock */
+		if (clk_mgr_dce->max_clks_by_state[clk_mgr_dce->max_clks_state].display_clk_khz
+				< context->bw_ctx.bw.dce.dispclk_khz)
+			low_req_clk = DM_PP_CLOCKS_STATE_INVALID;
+		else
+			low_req_clk = clk_mgr_dce->max_clks_state;
+	}
+
+	return low_req_clk;
+}
+
+static int dce_set_clock(
+	struct clk_mgr *clk_mgr,
+	int requested_clk_khz)
+{
+	struct dce_clk_mgr *clk_mgr_dce = TO_DCE_CLK_MGR(clk_mgr);
+	struct bp_pixel_clock_parameters pxl_clk_params = { 0 };
+	struct dc_bios *bp = clk_mgr->ctx->dc_bios;
+	int actual_clock = requested_clk_khz;
+	struct dmcu *dmcu = clk_mgr_dce->base.ctx->dc->res_pool->dmcu;
+
+	/* Make sure requested clock isn't lower than minimum threshold*/
+	if (requested_clk_khz > 0)
+		requested_clk_khz = max(requested_clk_khz,
+				clk_mgr_dce->dentist_vco_freq_khz / 64);
+
+	/* Prepare to program display clock*/
+	pxl_clk_params.target_pixel_clock_100hz = requested_clk_khz * 10;
+	pxl_clk_params.pll_id = CLOCK_SOURCE_ID_DFS;
+
+	if (clk_mgr_dce->dfs_bypass_active)
+		pxl_clk_params.flags.SET_DISPCLK_DFS_BYPASS = true;
+
+	bp->funcs->program_display_engine_pll(bp, &pxl_clk_params);
+
+	if (clk_mgr_dce->dfs_bypass_active) {
+		/* Cache the fixed display clock*/
+		clk_mgr_dce->dfs_bypass_disp_clk =
+			pxl_clk_params.dfs_bypass_display_clock;
+		actual_clock = pxl_clk_params.dfs_bypass_display_clock;
+	}
+
+	/* from power down, we need mark the clock state as ClocksStateNominal
+	 * from HWReset, so when resume we will call pplib voltage regulator.*/
+	if (requested_clk_khz == 0)
+		clk_mgr_dce->cur_min_clks_state = DM_PP_CLOCKS_STATE_NOMINAL;
+
+	if (dmcu && dmcu->funcs->is_dmcu_initialized(dmcu))
+		dmcu->funcs->set_psr_wait_loop(dmcu, actual_clock / 1000 / 7);
+
+	return actual_clock;
+}
+
+int dce112_set_clock(struct clk_mgr *clk_mgr, int requested_clk_khz)
+{
+	struct dce_clk_mgr *clk_mgr_dce = TO_DCE_CLK_MGR(clk_mgr);
+	struct bp_set_dce_clock_parameters dce_clk_params;
+	struct dc_bios *bp = clk_mgr->ctx->dc_bios;
+	struct dc *core_dc = clk_mgr->ctx->dc;
+	struct dmcu *dmcu = core_dc->res_pool->dmcu;
+	int actual_clock = requested_clk_khz;
+	/* Prepare to program display clock*/
+	memset(&dce_clk_params, 0, sizeof(dce_clk_params));
+
+	/* Make sure requested clock isn't lower than minimum threshold*/
+	if (requested_clk_khz > 0)
+		requested_clk_khz = max(requested_clk_khz,
+				clk_mgr_dce->dentist_vco_freq_khz / 62);
+
+	dce_clk_params.target_clock_frequency = requested_clk_khz;
+	dce_clk_params.pll_id = CLOCK_SOURCE_ID_DFS;
+	dce_clk_params.clock_type = DCECLOCK_TYPE_DISPLAY_CLOCK;
+
+	bp->funcs->set_dce_clock(bp, &dce_clk_params);
+	actual_clock = dce_clk_params.target_clock_frequency;
+
+	/* from power down, we need mark the clock state as ClocksStateNominal
+	 * from HWReset, so when resume we will call pplib voltage regulator.*/
+	if (requested_clk_khz == 0)
+		clk_mgr_dce->cur_min_clks_state = DM_PP_CLOCKS_STATE_NOMINAL;
+
+	/*Program DP ref Clock*/
+	/*VBIOS will determine DPREFCLK frequency, so we don't set it*/
+	dce_clk_params.target_clock_frequency = 0;
+	dce_clk_params.clock_type = DCECLOCK_TYPE_DPREFCLK;
+
+	if (!((clk_mgr->ctx->asic_id.chip_family == FAMILY_AI) &&
+	       ASICREV_IS_VEGA20_P(clk_mgr->ctx->asic_id.hw_internal_rev)))
+		dce_clk_params.flags.USE_GENLOCK_AS_SOURCE_FOR_DPREFCLK =
+			(dce_clk_params.pll_id ==
+					CLOCK_SOURCE_COMBO_DISPLAY_PLL0);
+	else
+		dce_clk_params.flags.USE_GENLOCK_AS_SOURCE_FOR_DPREFCLK = false;
+
+	bp->funcs->set_dce_clock(bp, &dce_clk_params);
+
+	if (!IS_FPGA_MAXIMUS_DC(core_dc->ctx->dce_environment)) {
+		if (dmcu && dmcu->funcs->is_dmcu_initialized(dmcu)) {
+			if (clk_mgr_dce->dfs_bypass_disp_clk != actual_clock)
+				dmcu->funcs->set_psr_wait_loop(dmcu,
+						actual_clock / 1000 / 7);
+		}
+	}
+
+	clk_mgr_dce->dfs_bypass_disp_clk = actual_clock;
+	return actual_clock;
+}
+
+static void dce_clock_read_integrated_info(struct dce_clk_mgr *clk_mgr_dce)
+{
+	struct dc_debug_options *debug = &clk_mgr_dce->base.ctx->dc->debug;
+	struct dc_bios *bp = clk_mgr_dce->base.ctx->dc_bios;
+	struct integrated_info info = { { { 0 } } };
+	struct dc_firmware_info fw_info = { { 0 } };
+	int i;
+
+	if (bp->integrated_info)
+		info = *bp->integrated_info;
+
+	clk_mgr_dce->dentist_vco_freq_khz = info.dentist_vco_freq;
+	if (clk_mgr_dce->dentist_vco_freq_khz == 0) {
+		bp->funcs->get_firmware_info(bp, &fw_info);
+		clk_mgr_dce->dentist_vco_freq_khz =
+			fw_info.smu_gpu_pll_output_freq;
+		if (clk_mgr_dce->dentist_vco_freq_khz == 0)
+			clk_mgr_dce->dentist_vco_freq_khz = 3600000;
+	}
+
+	/*update the maximum display clock for each power state*/
+	for (i = 0; i < NUMBER_OF_DISP_CLK_VOLTAGE; ++i) {
+		enum dm_pp_clocks_state clk_state = DM_PP_CLOCKS_STATE_INVALID;
+
+		switch (i) {
+		case 0:
+			clk_state = DM_PP_CLOCKS_STATE_ULTRA_LOW;
+			break;
+
+		case 1:
+			clk_state = DM_PP_CLOCKS_STATE_LOW;
+			break;
+
+		case 2:
+			clk_state = DM_PP_CLOCKS_STATE_NOMINAL;
+			break;
+
+		case 3:
+			clk_state = DM_PP_CLOCKS_STATE_PERFORMANCE;
+			break;
+
+		default:
+			clk_state = DM_PP_CLOCKS_STATE_INVALID;
+			break;
+		}
+
+		/*Do not allow bad VBIOS/SBIOS to override with invalid values,
+		 * check for > 100MHz*/
+		if (info.disp_clk_voltage[i].max_supported_clk >= 100000)
+			clk_mgr_dce->max_clks_by_state[clk_state].display_clk_khz =
+				info.disp_clk_voltage[i].max_supported_clk;
+	}
+
+	if (!debug->disable_dfs_bypass && bp->integrated_info)
+		if (bp->integrated_info->gpu_cap_info & DFS_BYPASS_ENABLE)
+			clk_mgr_dce->dfs_bypass_enabled = true;
+}
+
+void dce_clock_read_ss_info(struct dce_clk_mgr *clk_mgr_dce)
+{
+	struct dc_bios *bp = clk_mgr_dce->base.ctx->dc_bios;
+	int ss_info_num = bp->funcs->get_ss_entry_number(
+			bp, AS_SIGNAL_TYPE_GPU_PLL);
+
+	if (ss_info_num) {
+		struct spread_spectrum_info info = { { 0 } };
+		enum bp_result result = bp->funcs->get_spread_spectrum_info(
+				bp, AS_SIGNAL_TYPE_GPU_PLL, 0, &info);
+
+		/* Based on VBIOS, VBIOS will keep entry for GPU PLL SS
+		 * even if SS not enabled and in that case
+		 * SSInfo.spreadSpectrumPercentage !=0 would be sign
+		 * that SS is enabled
+		 */
+		if (result == BP_RESULT_OK &&
+				info.spread_spectrum_percentage != 0) {
+			clk_mgr_dce->ss_on_dprefclk = true;
+			clk_mgr_dce->dprefclk_ss_divider = info.spread_percentage_divider;
+
+			if (info.type.CENTER_MODE == 0) {
+				/* TODO: Currently for DP Reference clock we
+				 * need only SS percentage for
+				 * downspread */
+				clk_mgr_dce->dprefclk_ss_percentage =
+						info.spread_spectrum_percentage;
+			}
+
+			return;
+		}
+
+		result = bp->funcs->get_spread_spectrum_info(
+				bp, AS_SIGNAL_TYPE_DISPLAY_PORT, 0, &info);
+
+		/* Based on VBIOS, VBIOS will keep entry for DPREFCLK SS
+		 * even if SS not enabled and in that case
+		 * SSInfo.spreadSpectrumPercentage !=0 would be sign
+		 * that SS is enabled
+		 */
+		if (result == BP_RESULT_OK &&
+				info.spread_spectrum_percentage != 0) {
+			clk_mgr_dce->ss_on_dprefclk = true;
+			clk_mgr_dce->dprefclk_ss_divider = info.spread_percentage_divider;
+
+			if (info.type.CENTER_MODE == 0) {
+				/* Currently for DP Reference clock we
+				 * need only SS percentage for
+				 * downspread */
+				clk_mgr_dce->dprefclk_ss_percentage =
+						info.spread_spectrum_percentage;
+			}
+			if (clk_mgr_dce->base.ctx->dc->debug.ignore_dpref_ss)
+				clk_mgr_dce->dprefclk_ss_percentage = 0;
+		}
+	}
+}
+
+/**
+ * dce121_clock_patch_xgmi_ss_info() - Save XGMI spread spectrum info
+ * @clk_mgr: clock manager base structure
+ *
+ * Reads from VBIOS the XGMI spread spectrum info and saves it within
+ * the dce clock manager. This operation will overwrite the existing dprefclk
+ * SS values if the vBIOS query succeeds. Otherwise, it does nothing. It also
+ * sets the ->xgmi_enabled flag.
+ */
+void dce121_clock_patch_xgmi_ss_info(struct clk_mgr *clk_mgr)
+{
+	struct dce_clk_mgr *clk_mgr_dce = TO_DCE_CLK_MGR(clk_mgr);
+	enum bp_result result;
+	struct spread_spectrum_info info = { { 0 } };
+	struct dc_bios *bp = clk_mgr_dce->base.ctx->dc_bios;
+
+	clk_mgr_dce->xgmi_enabled = false;
+
+	result = bp->funcs->get_spread_spectrum_info(bp, AS_SIGNAL_TYPE_XGMI,
+						     0, &info);
+	if (result == BP_RESULT_OK && info.spread_spectrum_percentage != 0) {
+		clk_mgr_dce->xgmi_enabled = true;
+		clk_mgr_dce->ss_on_dprefclk = true;
+		clk_mgr_dce->dprefclk_ss_divider =
+				info.spread_percentage_divider;
+
+		if (info.type.CENTER_MODE == 0) {
+			/* Currently for DP Reference clock we
+			 * need only SS percentage for
+			 * downspread */
+			clk_mgr_dce->dprefclk_ss_percentage =
+					info.spread_spectrum_percentage;
+		}
+	}
+}
+
+void dce110_fill_display_configs(
+	const struct dc_state *context,
+	struct dm_pp_display_configuration *pp_display_cfg)
+{
+	int j;
+	int num_cfgs = 0;
+
+	for (j = 0; j < context->stream_count; j++) {
+		int k;
+
+		const struct dc_stream_state *stream = context->streams[j];
+		struct dm_pp_single_disp_config *cfg =
+			&pp_display_cfg->disp_configs[num_cfgs];
+		const struct pipe_ctx *pipe_ctx = NULL;
+
+		for (k = 0; k < MAX_PIPES; k++)
+			if (stream == context->res_ctx.pipe_ctx[k].stream) {
+				pipe_ctx = &context->res_ctx.pipe_ctx[k];
+				break;
+			}
+
+		ASSERT(pipe_ctx != NULL);
+
+		/* only notify active stream */
+		if (stream->dpms_off)
+			continue;
+
+		num_cfgs++;
+		cfg->signal = pipe_ctx->stream->signal;
+		cfg->pipe_idx = pipe_ctx->stream_res.tg->inst;
+		cfg->src_height = stream->src.height;
+		cfg->src_width = stream->src.width;
+		cfg->ddi_channel_mapping =
+			stream->link->ddi_channel_mapping.raw;
+		cfg->transmitter =
+			stream->link->link_enc->transmitter;
+		cfg->link_settings.lane_count =
+			stream->link->cur_link_settings.lane_count;
+		cfg->link_settings.link_rate =
+			stream->link->cur_link_settings.link_rate;
+		cfg->link_settings.link_spread =
+			stream->link->cur_link_settings.link_spread;
+		cfg->sym_clock = stream->phy_pix_clk;
+		/* Round v_refresh*/
+		cfg->v_refresh = stream->timing.pix_clk_100hz * 100;
+		cfg->v_refresh /= stream->timing.h_total;
+		cfg->v_refresh = (cfg->v_refresh + stream->timing.v_total / 2)
+							/ stream->timing.v_total;
+	}
+
+	pp_display_cfg->display_count = num_cfgs;
+}
+
+static uint32_t dce110_get_min_vblank_time_us(const struct dc_state *context)
+{
+	uint8_t j;
+	uint32_t min_vertical_blank_time = -1;
+
+	for (j = 0; j < context->stream_count; j++) {
+		struct dc_stream_state *stream = context->streams[j];
+		uint32_t vertical_blank_in_pixels = 0;
+		uint32_t vertical_blank_time = 0;
+
+		vertical_blank_in_pixels = stream->timing.h_total *
+			(stream->timing.v_total
+			 - stream->timing.v_addressable);
+
+		vertical_blank_time = vertical_blank_in_pixels
+			* 10000 / stream->timing.pix_clk_100hz;
+
+		if (min_vertical_blank_time > vertical_blank_time)
+			min_vertical_blank_time = vertical_blank_time;
+	}
+
+	return min_vertical_blank_time;
+}
+
+static int determine_sclk_from_bounding_box(
+		const struct dc *dc,
+		int required_sclk)
+{
+	int i;
+
+	/*
+	 * Some asics do not give us sclk levels, so we just report the actual
+	 * required sclk
+	 */
+	if (dc->sclk_lvls.num_levels == 0)
+		return required_sclk;
+
+	for (i = 0; i < dc->sclk_lvls.num_levels; i++) {
+		if (dc->sclk_lvls.clocks_in_khz[i] >= required_sclk)
+			return dc->sclk_lvls.clocks_in_khz[i];
+	}
+	/*
+	 * even maximum level could not satisfy requirement, this
+	 * is unexpected at this stage, should have been caught at
+	 * validation time
+	 */
+	ASSERT(0);
+	return dc->sclk_lvls.clocks_in_khz[dc->sclk_lvls.num_levels - 1];
+}
+
+static void dce_pplib_apply_display_requirements(
+	struct dc *dc,
+	struct dc_state *context)
+{
+	struct dm_pp_display_configuration *pp_display_cfg = &context->pp_display_cfg;
+
+	pp_display_cfg->avail_mclk_switch_time_us = dce110_get_min_vblank_time_us(context);
+
+	dce110_fill_display_configs(context, pp_display_cfg);
+
+	if (memcmp(&dc->current_state->pp_display_cfg, pp_display_cfg, sizeof(*pp_display_cfg)) !=  0)
+		dm_pp_apply_display_requirements(dc->ctx, pp_display_cfg);
+}
+
+static void dce11_pplib_apply_display_requirements(
+	struct dc *dc,
+	struct dc_state *context)
+{
+	struct dm_pp_display_configuration *pp_display_cfg = &context->pp_display_cfg;
+
+	pp_display_cfg->all_displays_in_sync =
+		context->bw_ctx.bw.dce.all_displays_in_sync;
+	pp_display_cfg->nb_pstate_switch_disable =
+			context->bw_ctx.bw.dce.nbp_state_change_enable == false;
+	pp_display_cfg->cpu_cc6_disable =
+			context->bw_ctx.bw.dce.cpuc_state_change_enable == false;
+	pp_display_cfg->cpu_pstate_disable =
+			context->bw_ctx.bw.dce.cpup_state_change_enable == false;
+	pp_display_cfg->cpu_pstate_separation_time =
+			context->bw_ctx.bw.dce.blackout_recovery_time_us;
+
+	pp_display_cfg->min_memory_clock_khz = context->bw_ctx.bw.dce.yclk_khz
+		/ MEMORY_TYPE_MULTIPLIER_CZ;
+
+	pp_display_cfg->min_engine_clock_khz = determine_sclk_from_bounding_box(
+			dc,
+			context->bw_ctx.bw.dce.sclk_khz);
+
+	/*
+	 * As workaround for >4x4K lightup set dcfclock to min_engine_clock value.
+	 * This is not required for less than 5 displays,
+	 * thus don't request decfclk in dc to avoid impact
+	 * on power saving.
+	 *
+	 */
+	pp_display_cfg->min_dcfclock_khz = (context->stream_count > 4) ?
+			pp_display_cfg->min_engine_clock_khz : 0;
+
+	pp_display_cfg->min_engine_clock_deep_sleep_khz
+			= context->bw_ctx.bw.dce.sclk_deep_sleep_khz;
+
+	pp_display_cfg->avail_mclk_switch_time_us =
+						dce110_get_min_vblank_time_us(context);
+	/* TODO: dce11.2*/
+	pp_display_cfg->avail_mclk_switch_time_in_disp_active_us = 0;
+
+	pp_display_cfg->disp_clk_khz = dc->res_pool->clk_mgr->clks.dispclk_khz;
+
+	dce110_fill_display_configs(context, pp_display_cfg);
+
+	/* TODO: is this still applicable?*/
+	if (pp_display_cfg->display_count == 1) {
+		const struct dc_crtc_timing *timing =
+			&context->streams[0]->timing;
+
+		pp_display_cfg->crtc_index =
+			pp_display_cfg->disp_configs[0].pipe_idx;
+		pp_display_cfg->line_time_in_us = timing->h_total * 10000 / timing->pix_clk_100hz;
+	}
+
+	if (memcmp(&dc->current_state->pp_display_cfg, pp_display_cfg, sizeof(*pp_display_cfg)) !=  0)
+		dm_pp_apply_display_requirements(dc->ctx, pp_display_cfg);
+}
+
+static void dce_update_clocks(struct clk_mgr *clk_mgr,
+			struct dc_state *context,
+			bool safe_to_lower)
+{
+	struct dce_clk_mgr *clk_mgr_dce = TO_DCE_CLK_MGR(clk_mgr);
+	struct dm_pp_power_level_change_request level_change_req;
+	int patched_disp_clk = context->bw_ctx.bw.dce.dispclk_khz;
+
+	/*TODO: W/A for dal3 linux, investigate why this works */
+	if (!clk_mgr_dce->dfs_bypass_active)
+		patched_disp_clk = patched_disp_clk * 115 / 100;
+
+	level_change_req.power_level = dce_get_required_clocks_state(clk_mgr, context);
+	/* get max clock state from PPLIB */
+	if ((level_change_req.power_level < clk_mgr_dce->cur_min_clks_state && safe_to_lower)
+			|| level_change_req.power_level > clk_mgr_dce->cur_min_clks_state) {
+		if (dm_pp_apply_power_level_change_request(clk_mgr->ctx, &level_change_req))
+			clk_mgr_dce->cur_min_clks_state = level_change_req.power_level;
+	}
+
+	if (should_set_clock(safe_to_lower, patched_disp_clk, clk_mgr->clks.dispclk_khz)) {
+		patched_disp_clk = dce_set_clock(clk_mgr, patched_disp_clk);
+		clk_mgr->clks.dispclk_khz = patched_disp_clk;
+	}
+	dce_pplib_apply_display_requirements(clk_mgr->ctx->dc, context);
+}
+
+static void dce11_update_clocks(struct clk_mgr *clk_mgr,
+			struct dc_state *context,
+			bool safe_to_lower)
+{
+	struct dce_clk_mgr *clk_mgr_dce = TO_DCE_CLK_MGR(clk_mgr);
+	struct dm_pp_power_level_change_request level_change_req;
+	int patched_disp_clk = context->bw_ctx.bw.dce.dispclk_khz;
+
+	/*TODO: W/A for dal3 linux, investigate why this works */
+	if (!clk_mgr_dce->dfs_bypass_active)
+		patched_disp_clk = patched_disp_clk * 115 / 100;
+
+	level_change_req.power_level = dce_get_required_clocks_state(clk_mgr, context);
+	/* get max clock state from PPLIB */
+	if ((level_change_req.power_level < clk_mgr_dce->cur_min_clks_state && safe_to_lower)
+			|| level_change_req.power_level > clk_mgr_dce->cur_min_clks_state) {
+		if (dm_pp_apply_power_level_change_request(clk_mgr->ctx, &level_change_req))
+			clk_mgr_dce->cur_min_clks_state = level_change_req.power_level;
+	}
+
+	if (should_set_clock(safe_to_lower, patched_disp_clk, clk_mgr->clks.dispclk_khz)) {
+		context->bw_ctx.bw.dce.dispclk_khz = dce_set_clock(clk_mgr, patched_disp_clk);
+		clk_mgr->clks.dispclk_khz = patched_disp_clk;
+	}
+	dce11_pplib_apply_display_requirements(clk_mgr->ctx->dc, context);
+}
+
+static void dce112_update_clocks(struct clk_mgr *clk_mgr,
+			struct dc_state *context,
+			bool safe_to_lower)
+{
+	struct dce_clk_mgr *clk_mgr_dce = TO_DCE_CLK_MGR(clk_mgr);
+	struct dm_pp_power_level_change_request level_change_req;
+	int patched_disp_clk = context->bw_ctx.bw.dce.dispclk_khz;
+
+	/*TODO: W/A for dal3 linux, investigate why this works */
+	if (!clk_mgr_dce->dfs_bypass_active)
+		patched_disp_clk = patched_disp_clk * 115 / 100;
+
+	level_change_req.power_level = dce_get_required_clocks_state(clk_mgr, context);
+	/* get max clock state from PPLIB */
+	if ((level_change_req.power_level < clk_mgr_dce->cur_min_clks_state && safe_to_lower)
+			|| level_change_req.power_level > clk_mgr_dce->cur_min_clks_state) {
+		if (dm_pp_apply_power_level_change_request(clk_mgr->ctx, &level_change_req))
+			clk_mgr_dce->cur_min_clks_state = level_change_req.power_level;
+	}
+
+	if (should_set_clock(safe_to_lower, patched_disp_clk, clk_mgr->clks.dispclk_khz)) {
+		patched_disp_clk = dce112_set_clock(clk_mgr, patched_disp_clk);
+		clk_mgr->clks.dispclk_khz = patched_disp_clk;
+	}
+	dce11_pplib_apply_display_requirements(clk_mgr->ctx->dc, context);
+}
+
+static void dce12_update_clocks(struct clk_mgr *clk_mgr,
+			struct dc_state *context,
+			bool safe_to_lower)
+{
+	struct dce_clk_mgr *clk_mgr_dce = TO_DCE_CLK_MGR(clk_mgr);
+	struct dm_pp_clock_for_voltage_req clock_voltage_req = {0};
+	int max_pix_clk = get_max_pixel_clock_for_all_paths(context);
+	int patched_disp_clk = context->bw_ctx.bw.dce.dispclk_khz;
+
+	/*TODO: W/A for dal3 linux, investigate why this works */
+	if (!clk_mgr_dce->dfs_bypass_active)
+		patched_disp_clk = patched_disp_clk * 115 / 100;
+
+	if (should_set_clock(safe_to_lower, patched_disp_clk, clk_mgr->clks.dispclk_khz)) {
+		clock_voltage_req.clk_type = DM_PP_CLOCK_TYPE_DISPLAY_CLK;
+		/*
+		 * When xGMI is enabled, the display clk needs to be adjusted
+		 * with the WAFL link's SS percentage.
+		 */
+		if (clk_mgr_dce->xgmi_enabled)
+			patched_disp_clk = clk_mgr_adjust_dp_ref_freq_for_ss(
+					clk_mgr_dce, patched_disp_clk);
+		clock_voltage_req.clocks_in_khz = patched_disp_clk;
+		clk_mgr->clks.dispclk_khz = dce112_set_clock(clk_mgr, patched_disp_clk);
+
+		dm_pp_apply_clock_for_voltage_request(clk_mgr->ctx, &clock_voltage_req);
+	}
+
+	if (should_set_clock(safe_to_lower, max_pix_clk, clk_mgr->clks.phyclk_khz)) {
+		clock_voltage_req.clk_type = DM_PP_CLOCK_TYPE_DISPLAYPHYCLK;
+		clock_voltage_req.clocks_in_khz = max_pix_clk;
+		clk_mgr->clks.phyclk_khz = max_pix_clk;
+
+		dm_pp_apply_clock_for_voltage_request(clk_mgr->ctx, &clock_voltage_req);
+	}
+	dce11_pplib_apply_display_requirements(clk_mgr->ctx->dc, context);
+}
+
+static const struct clk_mgr_funcs dce120_funcs = {
+	.get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz,
+	.update_clocks = dce12_update_clocks
+};
+
+static const struct clk_mgr_funcs dce112_funcs = {
+	.get_dp_ref_clk_frequency = dce_get_dp_ref_freq_khz,
+	.update_clocks = dce112_update_clocks
+};
+
+static const struct clk_mgr_funcs dce110_funcs = {
+	.get_dp_ref_clk_frequency = dce_get_dp_ref_freq_khz,
+	.update_clocks = dce11_update_clocks,
+};
+
+static const struct clk_mgr_funcs dce_funcs = {
+	.get_dp_ref_clk_frequency = dce_get_dp_ref_freq_khz,
+	.update_clocks = dce_update_clocks
+};
+
+static void dce_clk_mgr_construct(
+	struct dce_clk_mgr *clk_mgr_dce,
+	struct dc_context *ctx,
+	const struct clk_mgr_registers *regs,
+	const struct clk_mgr_shift *clk_shift,
+	const struct clk_mgr_mask *clk_mask)
+{
+	struct clk_mgr *base = &clk_mgr_dce->base;
+	struct dm_pp_static_clock_info static_clk_info = {0};
+
+	base->ctx = ctx;
+	base->funcs = &dce_funcs;
+
+	clk_mgr_dce->regs = regs;
+	clk_mgr_dce->clk_mgr_shift = clk_shift;
+	clk_mgr_dce->clk_mgr_mask = clk_mask;
+
+	clk_mgr_dce->dfs_bypass_disp_clk = 0;
+
+	clk_mgr_dce->dprefclk_ss_percentage = 0;
+	clk_mgr_dce->dprefclk_ss_divider = 1000;
+	clk_mgr_dce->ss_on_dprefclk = false;
+
+
+	if (dm_pp_get_static_clocks(ctx, &static_clk_info))
+		clk_mgr_dce->max_clks_state = static_clk_info.max_clocks_state;
+	else
+		clk_mgr_dce->max_clks_state = DM_PP_CLOCKS_STATE_NOMINAL;
+	clk_mgr_dce->cur_min_clks_state = DM_PP_CLOCKS_STATE_INVALID;
+
+	dce_clock_read_integrated_info(clk_mgr_dce);
+	dce_clock_read_ss_info(clk_mgr_dce);
+}
+
+struct clk_mgr *dce_clk_mgr_create(
+	struct dc_context *ctx,
+	const struct clk_mgr_registers *regs,
+	const struct clk_mgr_shift *clk_shift,
+	const struct clk_mgr_mask *clk_mask)
+{
+	struct dce_clk_mgr *clk_mgr_dce = kzalloc(sizeof(*clk_mgr_dce), GFP_KERNEL);
+
+	if (clk_mgr_dce == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	memcpy(clk_mgr_dce->max_clks_by_state,
+		dce80_max_clks_by_state,
+		sizeof(dce80_max_clks_by_state));
+
+	dce_clk_mgr_construct(
+		clk_mgr_dce, ctx, regs, clk_shift, clk_mask);
+
+	return &clk_mgr_dce->base;
+}
+
+struct clk_mgr *dce110_clk_mgr_create(
+	struct dc_context *ctx,
+	const struct clk_mgr_registers *regs,
+	const struct clk_mgr_shift *clk_shift,
+	const struct clk_mgr_mask *clk_mask)
+{
+	struct dce_clk_mgr *clk_mgr_dce = kzalloc(sizeof(*clk_mgr_dce), GFP_KERNEL);
+
+	if (clk_mgr_dce == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	memcpy(clk_mgr_dce->max_clks_by_state,
+		dce110_max_clks_by_state,
+		sizeof(dce110_max_clks_by_state));
+
+	dce_clk_mgr_construct(
+		clk_mgr_dce, ctx, regs, clk_shift, clk_mask);
+
+	clk_mgr_dce->base.funcs = &dce110_funcs;
+
+	return &clk_mgr_dce->base;
+}
+
+struct clk_mgr *dce112_clk_mgr_create(
+	struct dc_context *ctx,
+	const struct clk_mgr_registers *regs,
+	const struct clk_mgr_shift *clk_shift,
+	const struct clk_mgr_mask *clk_mask)
+{
+	struct dce_clk_mgr *clk_mgr_dce = kzalloc(sizeof(*clk_mgr_dce), GFP_KERNEL);
+
+	if (clk_mgr_dce == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	memcpy(clk_mgr_dce->max_clks_by_state,
+		dce112_max_clks_by_state,
+		sizeof(dce112_max_clks_by_state));
+
+	dce_clk_mgr_construct(
+		clk_mgr_dce, ctx, regs, clk_shift, clk_mask);
+
+	clk_mgr_dce->base.funcs = &dce112_funcs;
+
+	return &clk_mgr_dce->base;
+}
+
+struct clk_mgr *dce120_clk_mgr_create(struct dc_context *ctx)
+{
+	struct dce_clk_mgr *clk_mgr_dce = kzalloc(sizeof(*clk_mgr_dce), GFP_KERNEL);
+
+	if (clk_mgr_dce == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	memcpy(clk_mgr_dce->max_clks_by_state,
+		dce120_max_clks_by_state,
+		sizeof(dce120_max_clks_by_state));
+
+	dce_clk_mgr_construct(
+		clk_mgr_dce, ctx, NULL, NULL, NULL);
+
+	clk_mgr_dce->dprefclk_khz = 600000;
+	clk_mgr_dce->base.funcs = &dce120_funcs;
+
+	return &clk_mgr_dce->base;
+}
+
+struct clk_mgr *dce121_clk_mgr_create(struct dc_context *ctx)
+{
+	struct dce_clk_mgr *clk_mgr_dce = kzalloc(sizeof(*clk_mgr_dce),
+						  GFP_KERNEL);
+
+	if (clk_mgr_dce == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	memcpy(clk_mgr_dce->max_clks_by_state, dce120_max_clks_by_state,
+	       sizeof(dce120_max_clks_by_state));
+
+	dce_clk_mgr_construct(clk_mgr_dce, ctx, NULL, NULL, NULL);
+
+	clk_mgr_dce->dprefclk_khz = 625000;
+	clk_mgr_dce->base.funcs = &dce120_funcs;
+
+	return &clk_mgr_dce->base;
+}
+
+void dce_clk_mgr_destroy(struct clk_mgr **clk_mgr)
+{
+	struct dce_clk_mgr *clk_mgr_dce = TO_DCE_CLK_MGR(*clk_mgr);
+
+	kfree(clk_mgr_dce);
+	*clk_mgr = NULL;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_clock_source.c b/drivers/gpu/drm/amd/display/dc/dce/dce_clock_source.c
new file mode 100644
index 000000000000..b4f5b4a6331a
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_clock_source.c
@@ -0,0 +1,1873 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+
+
+#include "dc_types.h"
+#include "core_types.h"
+
+#include "include/grph_object_id.h"
+#include "include/logger_interface.h"
+
+#include "dce_clock_source.h"
+#include "clk_mgr.h"
+#include "dccg.h"
+
+#include "reg_helper.h"
+
+#define REG(reg)\
+	(clk_src->regs->reg)
+
+#define CTX \
+	clk_src->base.ctx
+
+#define DC_LOGGER \
+	calc_pll_cs->ctx->logger
+#define DC_LOGGER_INIT() \
+	struct calc_pll_clock_source *calc_pll_cs = &clk_src->calc_pll
+
+#undef FN
+#define FN(reg_name, field_name) \
+	clk_src->cs_shift->field_name, clk_src->cs_mask->field_name
+
+#define FRACT_FB_DIVIDER_DEC_POINTS_MAX_NUM 6
+#define CALC_PLL_CLK_SRC_ERR_TOLERANCE 1
+#define MAX_PLL_CALC_ERROR 0xFFFFFFFF
+
+#define NUM_ELEMENTS(a) (sizeof(a) / sizeof((a)[0]))
+
+static const struct spread_spectrum_data *get_ss_data_entry(
+		struct dce110_clk_src *clk_src,
+		enum signal_type signal,
+		uint32_t pix_clk_khz)
+{
+
+	uint32_t entrys_num;
+	uint32_t i;
+	struct spread_spectrum_data *ss_parm = NULL;
+	struct spread_spectrum_data *ret = NULL;
+
+	switch (signal) {
+	case SIGNAL_TYPE_DVI_SINGLE_LINK:
+	case SIGNAL_TYPE_DVI_DUAL_LINK:
+		ss_parm = clk_src->dvi_ss_params;
+		entrys_num = clk_src->dvi_ss_params_cnt;
+		break;
+
+	case SIGNAL_TYPE_HDMI_TYPE_A:
+		ss_parm = clk_src->hdmi_ss_params;
+		entrys_num = clk_src->hdmi_ss_params_cnt;
+		break;
+
+	case SIGNAL_TYPE_LVDS:
+		ss_parm = clk_src->lvds_ss_params;
+		entrys_num = clk_src->lvds_ss_params_cnt;
+		break;
+
+	case SIGNAL_TYPE_DISPLAY_PORT:
+	case SIGNAL_TYPE_DISPLAY_PORT_MST:
+	case SIGNAL_TYPE_EDP:
+	case SIGNAL_TYPE_VIRTUAL:
+		ss_parm = clk_src->dp_ss_params;
+		entrys_num = clk_src->dp_ss_params_cnt;
+		break;
+
+	default:
+		ss_parm = NULL;
+		entrys_num = 0;
+		break;
+	}
+
+	if (ss_parm == NULL)
+		return ret;
+
+	for (i = 0; i < entrys_num; ++i, ++ss_parm) {
+		if (ss_parm->freq_range_khz >= pix_clk_khz) {
+			ret = ss_parm;
+			break;
+		}
+	}
+
+	return ret;
+}
+
+/**
+ * calculate_fb_and_fractional_fb_divider - Calculates feedback and fractional
+ *                                          feedback dividers values
+ *
+ * @calc_pll_cs:	    Pointer to clock source information
+ * @target_pix_clk_100hz:   Desired frequency in 100 Hz
+ * @ref_divider:            Reference divider (already known)
+ * @post_divider:           Post Divider (already known)
+ * @feedback_divider_param: Pointer where to store
+ *			    calculated feedback divider value
+ * @fract_feedback_divider_param: Pointer where to store
+ *			    calculated fract feedback divider value
+ *
+ * return:
+ * It fills the locations pointed by feedback_divider_param
+ *					and fract_feedback_divider_param
+ * It returns	- true if feedback divider not 0
+ *		- false should never happen)
+ */
+static bool calculate_fb_and_fractional_fb_divider(
+		struct calc_pll_clock_source *calc_pll_cs,
+		uint32_t target_pix_clk_100hz,
+		uint32_t ref_divider,
+		uint32_t post_divider,
+		uint32_t *feedback_divider_param,
+		uint32_t *fract_feedback_divider_param)
+{
+	uint64_t feedback_divider;
+
+	feedback_divider =
+		(uint64_t)target_pix_clk_100hz * ref_divider * post_divider;
+	feedback_divider *= 10;
+	/* additional factor, since we divide by 10 afterwards */
+	feedback_divider *= (uint64_t)(calc_pll_cs->fract_fb_divider_factor);
+	feedback_divider = div_u64(feedback_divider, calc_pll_cs->ref_freq_khz * 10ull);
+
+/*Round to the number of precision
+ * The following code replace the old code (ullfeedbackDivider + 5)/10
+ * for example if the difference between the number
+ * of fractional feedback decimal point and the fractional FB Divider precision
+ * is 2 then the equation becomes (ullfeedbackDivider + 5*100) / (10*100))*/
+
+	feedback_divider += 5ULL *
+			    calc_pll_cs->fract_fb_divider_precision_factor;
+	feedback_divider =
+		div_u64(feedback_divider,
+			calc_pll_cs->fract_fb_divider_precision_factor * 10);
+	feedback_divider *= (uint64_t)
+			(calc_pll_cs->fract_fb_divider_precision_factor);
+
+	*feedback_divider_param =
+		div_u64_rem(
+			feedback_divider,
+			calc_pll_cs->fract_fb_divider_factor,
+			fract_feedback_divider_param);
+
+	if (*feedback_divider_param != 0)
+		return true;
+	return false;
+}
+
+/**
+ * calc_fb_divider_checking_tolerance - Calculates Feedback and
+ *                                      Fractional Feedback divider values
+ *		                        for passed Reference and Post divider,
+ *                                      checking for tolerance.
+ * @calc_pll_cs:	Pointer to clock source information
+ * @pll_settings:	Pointer to PLL settings
+ * @ref_divider:	Reference divider (already known)
+ * @post_divider:	Post Divider (already known)
+ * @tolerance:		Tolerance for Calculated Pixel Clock to be within
+ *
+ * return:
+ *  It fills the PLLSettings structure with PLL Dividers values
+ *  if calculated values are within required tolerance
+ *  It returns	- true if error is within tolerance
+ *		- false if error is not within tolerance
+ */
+static bool calc_fb_divider_checking_tolerance(
+		struct calc_pll_clock_source *calc_pll_cs,
+		struct pll_settings *pll_settings,
+		uint32_t ref_divider,
+		uint32_t post_divider,
+		uint32_t tolerance)
+{
+	uint32_t feedback_divider;
+	uint32_t fract_feedback_divider;
+	uint32_t actual_calculated_clock_100hz;
+	uint32_t abs_err;
+	uint64_t actual_calc_clk_100hz;
+
+	calculate_fb_and_fractional_fb_divider(
+			calc_pll_cs,
+			pll_settings->adjusted_pix_clk_100hz,
+			ref_divider,
+			post_divider,
+			&feedback_divider,
+			&fract_feedback_divider);
+
+	/*Actual calculated value*/
+	actual_calc_clk_100hz = (uint64_t)feedback_divider *
+					calc_pll_cs->fract_fb_divider_factor +
+							fract_feedback_divider;
+	actual_calc_clk_100hz *= (uint64_t)calc_pll_cs->ref_freq_khz * 10;
+	actual_calc_clk_100hz =
+		div_u64(actual_calc_clk_100hz,
+			ref_divider * post_divider *
+				calc_pll_cs->fract_fb_divider_factor);
+
+	actual_calculated_clock_100hz = (uint32_t)(actual_calc_clk_100hz);
+
+	abs_err = (actual_calculated_clock_100hz >
+					pll_settings->adjusted_pix_clk_100hz)
+			? actual_calculated_clock_100hz -
+					pll_settings->adjusted_pix_clk_100hz
+			: pll_settings->adjusted_pix_clk_100hz -
+						actual_calculated_clock_100hz;
+
+	if (abs_err <= tolerance) {
+		/*found good values*/
+		pll_settings->reference_freq = calc_pll_cs->ref_freq_khz;
+		pll_settings->reference_divider = ref_divider;
+		pll_settings->feedback_divider = feedback_divider;
+		pll_settings->fract_feedback_divider = fract_feedback_divider;
+		pll_settings->pix_clk_post_divider = post_divider;
+		pll_settings->calculated_pix_clk_100hz =
+			actual_calculated_clock_100hz;
+		pll_settings->vco_freq =
+			div_u64((u64)actual_calculated_clock_100hz * post_divider, 10);
+		return true;
+	}
+	return false;
+}
+
+static bool calc_pll_dividers_in_range(
+		struct calc_pll_clock_source *calc_pll_cs,
+		struct pll_settings *pll_settings,
+		uint32_t min_ref_divider,
+		uint32_t max_ref_divider,
+		uint32_t min_post_divider,
+		uint32_t max_post_divider,
+		uint32_t err_tolerance)
+{
+	uint32_t ref_divider;
+	uint32_t post_divider;
+	uint32_t tolerance;
+
+/* This is err_tolerance / 10000 = 0.0025 - acceptable error of 0.25%
+ * This is errorTolerance / 10000 = 0.0001 - acceptable error of 0.01%*/
+	tolerance = (pll_settings->adjusted_pix_clk_100hz * err_tolerance) /
+									100000;
+	if (tolerance < CALC_PLL_CLK_SRC_ERR_TOLERANCE)
+		tolerance = CALC_PLL_CLK_SRC_ERR_TOLERANCE;
+
+	for (
+			post_divider = max_post_divider;
+			post_divider >= min_post_divider;
+			--post_divider) {
+		for (
+				ref_divider = min_ref_divider;
+				ref_divider <= max_ref_divider;
+				++ref_divider) {
+			if (calc_fb_divider_checking_tolerance(
+					calc_pll_cs,
+					pll_settings,
+					ref_divider,
+					post_divider,
+					tolerance)) {
+				return true;
+			}
+		}
+	}
+
+	return false;
+}
+
+static uint32_t calculate_pixel_clock_pll_dividers(
+		struct calc_pll_clock_source *calc_pll_cs,
+		struct pll_settings *pll_settings)
+{
+	uint32_t err_tolerance;
+	uint32_t min_post_divider;
+	uint32_t max_post_divider;
+	uint32_t min_ref_divider;
+	uint32_t max_ref_divider;
+
+	if (pll_settings->adjusted_pix_clk_100hz == 0) {
+		DC_LOG_ERROR(
+			"%s Bad requested pixel clock", __func__);
+		return MAX_PLL_CALC_ERROR;
+	}
+
+/* 1) Find Post divider ranges */
+	if (pll_settings->pix_clk_post_divider) {
+		min_post_divider = pll_settings->pix_clk_post_divider;
+		max_post_divider = pll_settings->pix_clk_post_divider;
+	} else {
+		min_post_divider = calc_pll_cs->min_pix_clock_pll_post_divider;
+		if (min_post_divider * pll_settings->adjusted_pix_clk_100hz <
+						calc_pll_cs->min_vco_khz * 10) {
+			min_post_divider = calc_pll_cs->min_vco_khz * 10 /
+					pll_settings->adjusted_pix_clk_100hz;
+			if ((min_post_divider *
+					pll_settings->adjusted_pix_clk_100hz) <
+						calc_pll_cs->min_vco_khz * 10)
+				min_post_divider++;
+		}
+
+		max_post_divider = calc_pll_cs->max_pix_clock_pll_post_divider;
+		if (max_post_divider * pll_settings->adjusted_pix_clk_100hz
+				> calc_pll_cs->max_vco_khz * 10)
+			max_post_divider = calc_pll_cs->max_vco_khz * 10 /
+					pll_settings->adjusted_pix_clk_100hz;
+	}
+
+/* 2) Find Reference divider ranges
+ * When SS is enabled, or for Display Port even without SS,
+ * pll_settings->referenceDivider is not zero.
+ * So calculate PPLL FB and fractional FB divider
+ * using the passed reference divider*/
+
+	if (pll_settings->reference_divider) {
+		min_ref_divider = pll_settings->reference_divider;
+		max_ref_divider = pll_settings->reference_divider;
+	} else {
+		min_ref_divider = ((calc_pll_cs->ref_freq_khz
+				/ calc_pll_cs->max_pll_input_freq_khz)
+				> calc_pll_cs->min_pll_ref_divider)
+			? calc_pll_cs->ref_freq_khz
+					/ calc_pll_cs->max_pll_input_freq_khz
+			: calc_pll_cs->min_pll_ref_divider;
+
+		max_ref_divider = ((calc_pll_cs->ref_freq_khz
+				/ calc_pll_cs->min_pll_input_freq_khz)
+				< calc_pll_cs->max_pll_ref_divider)
+			? calc_pll_cs->ref_freq_khz /
+					calc_pll_cs->min_pll_input_freq_khz
+			: calc_pll_cs->max_pll_ref_divider;
+	}
+
+/* If some parameters are invalid we could have scenario when  "min">"max"
+ * which produced endless loop later.
+ * We should investigate why we get the wrong parameters.
+ * But to follow the similar logic when "adjustedPixelClock" is set to be 0
+ * it is better to return here than cause system hang/watchdog timeout later.
+ *  ## SVS Wed 15 Jul 2009 */
+
+	if (min_post_divider > max_post_divider) {
+		DC_LOG_ERROR(
+			"%s Post divider range is invalid", __func__);
+		return MAX_PLL_CALC_ERROR;
+	}
+
+	if (min_ref_divider > max_ref_divider) {
+		DC_LOG_ERROR(
+			"%s Reference divider range is invalid", __func__);
+		return MAX_PLL_CALC_ERROR;
+	}
+
+/* 3) Try to find PLL dividers given ranges
+ * starting with minimal error tolerance.
+ * Increase error tolerance until PLL dividers found*/
+	err_tolerance = MAX_PLL_CALC_ERROR;
+
+	while (!calc_pll_dividers_in_range(
+			calc_pll_cs,
+			pll_settings,
+			min_ref_divider,
+			max_ref_divider,
+			min_post_divider,
+			max_post_divider,
+			err_tolerance))
+		err_tolerance += (err_tolerance > 10)
+				? (err_tolerance / 10)
+				: 1;
+
+	return err_tolerance;
+}
+
+static bool pll_adjust_pix_clk(
+		struct dce110_clk_src *clk_src,
+		struct pixel_clk_params *pix_clk_params,
+		struct pll_settings *pll_settings)
+{
+	uint32_t actual_pix_clk_100hz = 0;
+	uint32_t requested_clk_100hz = 0;
+	struct bp_adjust_pixel_clock_parameters bp_adjust_pixel_clock_params = {
+							0 };
+	enum bp_result bp_result;
+	switch (pix_clk_params->signal_type) {
+	case SIGNAL_TYPE_HDMI_TYPE_A: {
+		requested_clk_100hz = pix_clk_params->requested_pix_clk_100hz;
+		if (pix_clk_params->pixel_encoding != PIXEL_ENCODING_YCBCR422) {
+			switch (pix_clk_params->color_depth) {
+			case COLOR_DEPTH_101010:
+				requested_clk_100hz = (requested_clk_100hz * 5) >> 2;
+				break; /* x1.25*/
+			case COLOR_DEPTH_121212:
+				requested_clk_100hz = (requested_clk_100hz * 6) >> 2;
+				break; /* x1.5*/
+			case COLOR_DEPTH_161616:
+				requested_clk_100hz = requested_clk_100hz * 2;
+				break; /* x2.0*/
+			default:
+				break;
+			}
+		}
+		actual_pix_clk_100hz = requested_clk_100hz;
+	}
+		break;
+
+	case SIGNAL_TYPE_DISPLAY_PORT:
+	case SIGNAL_TYPE_DISPLAY_PORT_MST:
+	case SIGNAL_TYPE_EDP:
+		requested_clk_100hz = pix_clk_params->requested_sym_clk * 10;
+		actual_pix_clk_100hz = pix_clk_params->requested_pix_clk_100hz;
+		break;
+
+	default:
+		requested_clk_100hz = pix_clk_params->requested_pix_clk_100hz;
+		actual_pix_clk_100hz = pix_clk_params->requested_pix_clk_100hz;
+		break;
+	}
+
+	bp_adjust_pixel_clock_params.pixel_clock = requested_clk_100hz / 10;
+	bp_adjust_pixel_clock_params.
+		encoder_object_id = pix_clk_params->encoder_object_id;
+	bp_adjust_pixel_clock_params.signal_type = pix_clk_params->signal_type;
+	bp_adjust_pixel_clock_params.
+		ss_enable = pix_clk_params->flags.ENABLE_SS;
+	bp_result = clk_src->bios->funcs->adjust_pixel_clock(
+			clk_src->bios, &bp_adjust_pixel_clock_params);
+	if (bp_result == BP_RESULT_OK) {
+		pll_settings->actual_pix_clk_100hz = actual_pix_clk_100hz;
+		pll_settings->adjusted_pix_clk_100hz =
+			bp_adjust_pixel_clock_params.adjusted_pixel_clock * 10;
+		pll_settings->reference_divider =
+			bp_adjust_pixel_clock_params.reference_divider;
+		pll_settings->pix_clk_post_divider =
+			bp_adjust_pixel_clock_params.pixel_clock_post_divider;
+
+		return true;
+	}
+
+	return false;
+}
+
+/*
+ * Calculate PLL Dividers for given Clock Value.
+ * First will call VBIOS Adjust Exec table to check if requested Pixel clock
+ * will be Adjusted based on usage.
+ * Then it will calculate PLL Dividers for this Adjusted clock using preferred
+ * method (Maximum VCO frequency).
+ *
+ * \return
+ *     Calculation error in units of 0.01%
+ */
+
+static uint32_t dce110_get_pix_clk_dividers_helper (
+		struct dce110_clk_src *clk_src,
+		struct pll_settings *pll_settings,
+		struct pixel_clk_params *pix_clk_params)
+{
+	uint32_t field = 0;
+	uint32_t pll_calc_error = MAX_PLL_CALC_ERROR;
+	DC_LOGGER_INIT();
+	/* Check if reference clock is external (not pcie/xtalin)
+	* HW Dce80 spec:
+	* 00 - PCIE_REFCLK, 01 - XTALIN,    02 - GENERICA,    03 - GENERICB
+	* 04 - HSYNCA,      05 - GENLK_CLK, 06 - PCIE_REFCLK, 07 - DVOCLK0 */
+	REG_GET(PLL_CNTL, PLL_REF_DIV_SRC, &field);
+	pll_settings->use_external_clk = (field > 1);
+
+	/* VBIOS by default enables DP SS (spread on IDCLK) for DCE 8.0 always
+	 * (we do not care any more from SI for some older DP Sink which
+	 * does not report SS support, no known issues) */
+	if ((pix_clk_params->flags.ENABLE_SS) ||
+			(dc_is_dp_signal(pix_clk_params->signal_type))) {
+
+		const struct spread_spectrum_data *ss_data = get_ss_data_entry(
+					clk_src,
+					pix_clk_params->signal_type,
+					pll_settings->adjusted_pix_clk_100hz / 10);
+
+		if (NULL != ss_data)
+			pll_settings->ss_percentage = ss_data->percentage;
+	}
+
+	/* Check VBIOS AdjustPixelClock Exec table */
+	if (!pll_adjust_pix_clk(clk_src, pix_clk_params, pll_settings)) {
+		/* Should never happen, ASSERT and fill up values to be able
+		 * to continue. */
+		DC_LOG_ERROR(
+			"%s: Failed to adjust pixel clock!!", __func__);
+		pll_settings->actual_pix_clk_100hz =
+				pix_clk_params->requested_pix_clk_100hz;
+		pll_settings->adjusted_pix_clk_100hz =
+				pix_clk_params->requested_pix_clk_100hz;
+
+		if (dc_is_dp_signal(pix_clk_params->signal_type))
+			pll_settings->adjusted_pix_clk_100hz = 1000000;
+	}
+
+	/* Calculate Dividers */
+	if (pix_clk_params->signal_type == SIGNAL_TYPE_HDMI_TYPE_A)
+		/*Calculate Dividers by HDMI object, no SS case or SS case */
+		pll_calc_error =
+			calculate_pixel_clock_pll_dividers(
+					&clk_src->calc_pll_hdmi,
+					pll_settings);
+	else
+		/*Calculate Dividers by default object, no SS case or SS case */
+		pll_calc_error =
+			calculate_pixel_clock_pll_dividers(
+					&clk_src->calc_pll,
+					pll_settings);
+
+	return pll_calc_error;
+}
+
+static void dce112_get_pix_clk_dividers_helper (
+		struct dce110_clk_src *clk_src,
+		struct pll_settings *pll_settings,
+		struct pixel_clk_params *pix_clk_params)
+{
+	uint32_t actual_pixel_clock_100hz;
+
+	actual_pixel_clock_100hz = pix_clk_params->requested_pix_clk_100hz;
+	/* Calculate Dividers */
+	if (pix_clk_params->signal_type == SIGNAL_TYPE_HDMI_TYPE_A) {
+		switch (pix_clk_params->color_depth) {
+		case COLOR_DEPTH_101010:
+			actual_pixel_clock_100hz = (actual_pixel_clock_100hz * 5) >> 2;
+			actual_pixel_clock_100hz -= actual_pixel_clock_100hz % 10;
+			break;
+		case COLOR_DEPTH_121212:
+			actual_pixel_clock_100hz = (actual_pixel_clock_100hz * 6) >> 2;
+			actual_pixel_clock_100hz -= actual_pixel_clock_100hz % 10;
+			break;
+		case COLOR_DEPTH_161616:
+			actual_pixel_clock_100hz = actual_pixel_clock_100hz * 2;
+			break;
+		default:
+			break;
+		}
+	}
+	pll_settings->actual_pix_clk_100hz = actual_pixel_clock_100hz;
+	pll_settings->adjusted_pix_clk_100hz = actual_pixel_clock_100hz;
+	pll_settings->calculated_pix_clk_100hz = pix_clk_params->requested_pix_clk_100hz;
+}
+
+static uint32_t dce110_get_pix_clk_dividers(
+		struct clock_source *cs,
+		struct pixel_clk_params *pix_clk_params,
+		struct pll_settings *pll_settings)
+{
+	struct dce110_clk_src *clk_src = TO_DCE110_CLK_SRC(cs);
+	uint32_t pll_calc_error = MAX_PLL_CALC_ERROR;
+	DC_LOGGER_INIT();
+
+	if (pix_clk_params == NULL || pll_settings == NULL
+			|| pix_clk_params->requested_pix_clk_100hz == 0) {
+		DC_LOG_ERROR(
+			"%s: Invalid parameters!!\n", __func__);
+		return pll_calc_error;
+	}
+
+	memset(pll_settings, 0, sizeof(*pll_settings));
+
+	if (cs->id == CLOCK_SOURCE_ID_DP_DTO ||
+			cs->id == CLOCK_SOURCE_ID_EXTERNAL) {
+		pll_settings->adjusted_pix_clk_100hz = clk_src->ext_clk_khz * 10;
+		pll_settings->calculated_pix_clk_100hz = clk_src->ext_clk_khz * 10;
+		pll_settings->actual_pix_clk_100hz =
+					pix_clk_params->requested_pix_clk_100hz;
+		return 0;
+	}
+
+	pll_calc_error = dce110_get_pix_clk_dividers_helper(clk_src,
+			pll_settings, pix_clk_params);
+
+	return pll_calc_error;
+}
+
+static uint32_t dce112_get_pix_clk_dividers(
+		struct clock_source *cs,
+		struct pixel_clk_params *pix_clk_params,
+		struct pll_settings *pll_settings)
+{
+	struct dce110_clk_src *clk_src = TO_DCE110_CLK_SRC(cs);
+	DC_LOGGER_INIT();
+
+	if (pix_clk_params == NULL || pll_settings == NULL
+			|| pix_clk_params->requested_pix_clk_100hz == 0) {
+		DC_LOG_ERROR(
+			"%s: Invalid parameters!!\n", __func__);
+		return -1;
+	}
+
+	memset(pll_settings, 0, sizeof(*pll_settings));
+
+	if (cs->id == CLOCK_SOURCE_ID_DP_DTO ||
+			cs->id == CLOCK_SOURCE_ID_EXTERNAL) {
+		pll_settings->adjusted_pix_clk_100hz = clk_src->ext_clk_khz * 10;
+		pll_settings->calculated_pix_clk_100hz = clk_src->ext_clk_khz * 10;
+		pll_settings->actual_pix_clk_100hz =
+					pix_clk_params->requested_pix_clk_100hz;
+		return -1;
+	}
+
+	dce112_get_pix_clk_dividers_helper(clk_src,
+			pll_settings, pix_clk_params);
+
+	return 0;
+}
+
+static bool disable_spread_spectrum(struct dce110_clk_src *clk_src)
+{
+	enum bp_result result;
+	struct bp_spread_spectrum_parameters bp_ss_params = {0};
+
+	bp_ss_params.pll_id = clk_src->base.id;
+
+	/*Call ASICControl to process ATOMBIOS Exec table*/
+	result = clk_src->bios->funcs->enable_spread_spectrum_on_ppll(
+			clk_src->bios,
+			&bp_ss_params,
+			false);
+
+	return result == BP_RESULT_OK;
+}
+
+static bool calculate_ss(
+		const struct pll_settings *pll_settings,
+		const struct spread_spectrum_data *ss_data,
+		struct delta_sigma_data *ds_data)
+{
+	struct fixed31_32 fb_div;
+	struct fixed31_32 ss_amount;
+	struct fixed31_32 ss_nslip_amount;
+	struct fixed31_32 ss_ds_frac_amount;
+	struct fixed31_32 ss_step_size;
+	struct fixed31_32 modulation_time;
+
+	if (ds_data == NULL)
+		return false;
+	if (ss_data == NULL)
+		return false;
+	if (ss_data->percentage == 0)
+		return false;
+	if (pll_settings == NULL)
+		return false;
+
+	memset(ds_data, 0, sizeof(struct delta_sigma_data));
+
+	/* compute SS_AMOUNT_FBDIV & SS_AMOUNT_NFRAC_SLIP & SS_AMOUNT_DSFRAC*/
+	/* 6 decimal point support in fractional feedback divider */
+	fb_div  = dc_fixpt_from_fraction(
+		pll_settings->fract_feedback_divider, 1000000);
+	fb_div = dc_fixpt_add_int(fb_div, pll_settings->feedback_divider);
+
+	ds_data->ds_frac_amount = 0;
+	/*spreadSpectrumPercentage is in the unit of .01%,
+	 * so have to divided by 100 * 100*/
+	ss_amount = dc_fixpt_mul(
+		fb_div, dc_fixpt_from_fraction(ss_data->percentage,
+					100 * (long long)ss_data->percentage_divider));
+	ds_data->feedback_amount = dc_fixpt_floor(ss_amount);
+
+	ss_nslip_amount = dc_fixpt_sub(ss_amount,
+		dc_fixpt_from_int(ds_data->feedback_amount));
+	ss_nslip_amount = dc_fixpt_mul_int(ss_nslip_amount, 10);
+	ds_data->nfrac_amount = dc_fixpt_floor(ss_nslip_amount);
+
+	ss_ds_frac_amount = dc_fixpt_sub(ss_nslip_amount,
+		dc_fixpt_from_int(ds_data->nfrac_amount));
+	ss_ds_frac_amount = dc_fixpt_mul_int(ss_ds_frac_amount, 65536);
+	ds_data->ds_frac_amount = dc_fixpt_floor(ss_ds_frac_amount);
+
+	/* compute SS_STEP_SIZE_DSFRAC */
+	modulation_time = dc_fixpt_from_fraction(
+		pll_settings->reference_freq * (uint64_t)1000,
+		pll_settings->reference_divider * (uint64_t)ss_data->modulation_freq_hz);
+
+	if (ss_data->flags.CENTER_SPREAD)
+		modulation_time = dc_fixpt_div_int(modulation_time, 4);
+	else
+		modulation_time = dc_fixpt_div_int(modulation_time, 2);
+
+	ss_step_size = dc_fixpt_div(ss_amount, modulation_time);
+	/* SS_STEP_SIZE_DSFRAC_DEC = Int(SS_STEP_SIZE * 2 ^ 16 * 10)*/
+	ss_step_size = dc_fixpt_mul_int(ss_step_size, 65536 * 10);
+	ds_data->ds_frac_size =  dc_fixpt_floor(ss_step_size);
+
+	return true;
+}
+
+static bool enable_spread_spectrum(
+		struct dce110_clk_src *clk_src,
+		enum signal_type signal, struct pll_settings *pll_settings)
+{
+	struct bp_spread_spectrum_parameters bp_params = {0};
+	struct delta_sigma_data d_s_data;
+	const struct spread_spectrum_data *ss_data = NULL;
+
+	ss_data = get_ss_data_entry(
+			clk_src,
+			signal,
+			pll_settings->calculated_pix_clk_100hz / 10);
+
+/* Pixel clock PLL has been programmed to generate desired pixel clock,
+ * now enable SS on pixel clock */
+/* TODO is it OK to return true not doing anything ??*/
+	if (ss_data != NULL && pll_settings->ss_percentage != 0) {
+		if (calculate_ss(pll_settings, ss_data, &d_s_data)) {
+			bp_params.ds.feedback_amount =
+					d_s_data.feedback_amount;
+			bp_params.ds.nfrac_amount =
+					d_s_data.nfrac_amount;
+			bp_params.ds.ds_frac_size = d_s_data.ds_frac_size;
+			bp_params.ds_frac_amount =
+					d_s_data.ds_frac_amount;
+			bp_params.flags.DS_TYPE = 1;
+			bp_params.pll_id = clk_src->base.id;
+			bp_params.percentage = ss_data->percentage;
+			if (ss_data->flags.CENTER_SPREAD)
+				bp_params.flags.CENTER_SPREAD = 1;
+			if (ss_data->flags.EXTERNAL_SS)
+				bp_params.flags.EXTERNAL_SS = 1;
+
+			if (BP_RESULT_OK !=
+				clk_src->bios->funcs->
+					enable_spread_spectrum_on_ppll(
+							clk_src->bios,
+							&bp_params,
+							true))
+				return false;
+		} else
+			return false;
+	}
+	return true;
+}
+
+static void dce110_program_pixel_clk_resync(
+		struct dce110_clk_src *clk_src,
+		enum signal_type signal_type,
+		enum dc_color_depth colordepth)
+{
+	REG_UPDATE(RESYNC_CNTL,
+			DCCG_DEEP_COLOR_CNTL1, 0);
+	/*
+	 24 bit mode: TMDS clock = 1.0 x pixel clock  (1:1)
+	 30 bit mode: TMDS clock = 1.25 x pixel clock (5:4)
+	 36 bit mode: TMDS clock = 1.5 x pixel clock  (3:2)
+	 48 bit mode: TMDS clock = 2 x pixel clock    (2:1)
+	 */
+	if (signal_type != SIGNAL_TYPE_HDMI_TYPE_A)
+		return;
+
+	switch (colordepth) {
+	case COLOR_DEPTH_888:
+		REG_UPDATE(RESYNC_CNTL,
+				DCCG_DEEP_COLOR_CNTL1, 0);
+		break;
+	case COLOR_DEPTH_101010:
+		REG_UPDATE(RESYNC_CNTL,
+				DCCG_DEEP_COLOR_CNTL1, 1);
+		break;
+	case COLOR_DEPTH_121212:
+		REG_UPDATE(RESYNC_CNTL,
+				DCCG_DEEP_COLOR_CNTL1, 2);
+		break;
+	case COLOR_DEPTH_161616:
+		REG_UPDATE(RESYNC_CNTL,
+				DCCG_DEEP_COLOR_CNTL1, 3);
+		break;
+	default:
+		break;
+	}
+}
+
+static void dce112_program_pixel_clk_resync(
+		struct dce110_clk_src *clk_src,
+		enum signal_type signal_type,
+		enum dc_color_depth colordepth,
+		bool enable_ycbcr420)
+{
+	uint32_t deep_color_cntl = 0;
+	uint32_t double_rate_enable = 0;
+
+	/*
+	 24 bit mode: TMDS clock = 1.0 x pixel clock  (1:1)
+	 30 bit mode: TMDS clock = 1.25 x pixel clock (5:4)
+	 36 bit mode: TMDS clock = 1.5 x pixel clock  (3:2)
+	 48 bit mode: TMDS clock = 2 x pixel clock    (2:1)
+	 */
+	if (signal_type == SIGNAL_TYPE_HDMI_TYPE_A) {
+		double_rate_enable = enable_ycbcr420 ? 1 : 0;
+
+		switch (colordepth) {
+		case COLOR_DEPTH_888:
+			deep_color_cntl = 0;
+			break;
+		case COLOR_DEPTH_101010:
+			deep_color_cntl = 1;
+			break;
+		case COLOR_DEPTH_121212:
+			deep_color_cntl = 2;
+			break;
+		case COLOR_DEPTH_161616:
+			deep_color_cntl = 3;
+			break;
+		default:
+			break;
+		}
+	}
+
+	if (clk_src->cs_mask->PHYPLLA_PIXCLK_DOUBLE_RATE_ENABLE)
+		REG_UPDATE_2(PIXCLK_RESYNC_CNTL,
+				PHYPLLA_DCCG_DEEP_COLOR_CNTL, deep_color_cntl,
+				PHYPLLA_PIXCLK_DOUBLE_RATE_ENABLE, double_rate_enable);
+	else
+		REG_UPDATE(PIXCLK_RESYNC_CNTL,
+				PHYPLLA_DCCG_DEEP_COLOR_CNTL, deep_color_cntl);
+
+}
+
+static bool dce110_program_pix_clk(
+		struct clock_source *clock_source,
+		struct pixel_clk_params *pix_clk_params,
+		enum dp_link_encoding encoding,
+		struct pll_settings *pll_settings)
+{
+	struct dce110_clk_src *clk_src = TO_DCE110_CLK_SRC(clock_source);
+	struct bp_pixel_clock_parameters bp_pc_params = {0};
+
+	/* First disable SS
+	 * ATOMBIOS will enable by default SS on PLL for DP,
+	 * do not disable it here
+	 */
+	if (clock_source->id != CLOCK_SOURCE_ID_EXTERNAL &&
+			!dc_is_dp_signal(pix_clk_params->signal_type) &&
+			clock_source->ctx->dce_version <= DCE_VERSION_11_0)
+		disable_spread_spectrum(clk_src);
+
+	/*ATOMBIOS expects pixel rate adjusted by deep color ratio)*/
+	bp_pc_params.controller_id = pix_clk_params->controller_id;
+	bp_pc_params.pll_id = clock_source->id;
+	bp_pc_params.target_pixel_clock_100hz = pll_settings->actual_pix_clk_100hz;
+	bp_pc_params.encoder_object_id = pix_clk_params->encoder_object_id;
+	bp_pc_params.signal_type = pix_clk_params->signal_type;
+
+	bp_pc_params.reference_divider = pll_settings->reference_divider;
+	bp_pc_params.feedback_divider = pll_settings->feedback_divider;
+	bp_pc_params.fractional_feedback_divider =
+			pll_settings->fract_feedback_divider;
+	bp_pc_params.pixel_clock_post_divider =
+			pll_settings->pix_clk_post_divider;
+	bp_pc_params.flags.SET_EXTERNAL_REF_DIV_SRC =
+					pll_settings->use_external_clk;
+
+	switch (pix_clk_params->color_depth) {
+	case COLOR_DEPTH_101010:
+		bp_pc_params.color_depth = TRANSMITTER_COLOR_DEPTH_30;
+		break;
+	case COLOR_DEPTH_121212:
+		bp_pc_params.color_depth = TRANSMITTER_COLOR_DEPTH_36;
+		break;
+	case COLOR_DEPTH_161616:
+		bp_pc_params.color_depth = TRANSMITTER_COLOR_DEPTH_48;
+		break;
+	default:
+		break;
+	}
+
+	if (clk_src->bios->funcs->set_pixel_clock(
+			clk_src->bios, &bp_pc_params) != BP_RESULT_OK)
+		return false;
+	/* Enable SS
+	 * ATOMBIOS will enable by default SS for DP on PLL ( DP ID clock),
+	 * based on HW display PLL team, SS control settings should be programmed
+	 * during PLL Reset, but they do not have effect
+	 * until SS_EN is asserted.*/
+	if (clock_source->id != CLOCK_SOURCE_ID_EXTERNAL
+			&& !dc_is_dp_signal(pix_clk_params->signal_type)) {
+
+		if (pix_clk_params->flags.ENABLE_SS)
+			if (!enable_spread_spectrum(clk_src,
+							pix_clk_params->signal_type,
+							pll_settings))
+				return false;
+
+		/* Resync deep color DTO */
+		dce110_program_pixel_clk_resync(clk_src,
+					pix_clk_params->signal_type,
+					pix_clk_params->color_depth);
+	}
+
+	return true;
+}
+
+static bool dce112_program_pix_clk(
+		struct clock_source *clock_source,
+		struct pixel_clk_params *pix_clk_params,
+		enum dp_link_encoding encoding,
+		struct pll_settings *pll_settings)
+{
+	struct dce110_clk_src *clk_src = TO_DCE110_CLK_SRC(clock_source);
+	struct bp_pixel_clock_parameters bp_pc_params = {0};
+
+	/* First disable SS
+	 * ATOMBIOS will enable by default SS on PLL for DP,
+	 * do not disable it here
+	 */
+	if (clock_source->id != CLOCK_SOURCE_ID_EXTERNAL &&
+			!dc_is_dp_signal(pix_clk_params->signal_type) &&
+			clock_source->ctx->dce_version <= DCE_VERSION_11_0)
+		disable_spread_spectrum(clk_src);
+
+	/*ATOMBIOS expects pixel rate adjusted by deep color ratio)*/
+	bp_pc_params.controller_id = pix_clk_params->controller_id;
+	bp_pc_params.pll_id = clock_source->id;
+	bp_pc_params.target_pixel_clock_100hz = pll_settings->actual_pix_clk_100hz;
+	bp_pc_params.encoder_object_id = pix_clk_params->encoder_object_id;
+	bp_pc_params.signal_type = pix_clk_params->signal_type;
+
+	if (clock_source->id != CLOCK_SOURCE_ID_DP_DTO) {
+		bp_pc_params.flags.SET_GENLOCK_REF_DIV_SRC =
+						pll_settings->use_external_clk;
+		bp_pc_params.flags.SET_XTALIN_REF_SRC =
+						!pll_settings->use_external_clk;
+		if (pix_clk_params->flags.SUPPORT_YCBCR420) {
+			bp_pc_params.flags.SUPPORT_YUV_420 = 1;
+		}
+	}
+	if (clk_src->bios->funcs->set_pixel_clock(
+			clk_src->bios, &bp_pc_params) != BP_RESULT_OK)
+		return false;
+	/* Resync deep color DTO */
+	if (clock_source->id != CLOCK_SOURCE_ID_DP_DTO)
+		dce112_program_pixel_clk_resync(clk_src,
+					pix_clk_params->signal_type,
+					pix_clk_params->color_depth,
+					pix_clk_params->flags.SUPPORT_YCBCR420);
+
+	return true;
+}
+
+static bool dcn31_program_pix_clk(
+		struct clock_source *clock_source,
+		struct pixel_clk_params *pix_clk_params,
+		enum dp_link_encoding encoding,
+		struct pll_settings *pll_settings)
+{
+	struct dce110_clk_src *clk_src = TO_DCE110_CLK_SRC(clock_source);
+	unsigned int inst = pix_clk_params->controller_id - CONTROLLER_ID_D0;
+	unsigned int dp_dto_ref_khz = clock_source->ctx->dc->clk_mgr->dprefclk_khz;
+	const struct pixel_rate_range_table_entry *e =
+			look_up_in_video_optimized_rate_tlb(pix_clk_params->requested_pix_clk_100hz / 10);
+	struct bp_pixel_clock_parameters bp_pc_params = {0};
+	enum transmitter_color_depth bp_pc_colour_depth = TRANSMITTER_COLOR_DEPTH_24;
+
+	// Apply ssed(spread spectrum) dpref clock for edp and dp
+	if (clock_source->ctx->dc->clk_mgr->dp_dto_source_clock_in_khz != 0 &&
+		dc_is_dp_signal(pix_clk_params->signal_type) &&
+		encoding == DP_8b_10b_ENCODING)
+		dp_dto_ref_khz = clock_source->ctx->dc->clk_mgr->dp_dto_source_clock_in_khz;
+
+	// For these signal types Driver to program DP_DTO without calling VBIOS Command table
+	if (dc_is_dp_signal(pix_clk_params->signal_type) || dc_is_virtual_signal(pix_clk_params->signal_type)) {
+		if (e) {
+			/* Set DTO values: phase = target clock, modulo = reference clock*/
+			REG_WRITE(PHASE[inst], e->target_pixel_rate_khz * e->mult_factor);
+			REG_WRITE(MODULO[inst], dp_dto_ref_khz * e->div_factor);
+		} else {
+			/* Set DTO values: phase = target clock, modulo = reference clock*/
+			REG_WRITE(PHASE[inst], pll_settings->actual_pix_clk_100hz * 100);
+			REG_WRITE(MODULO[inst], dp_dto_ref_khz * 1000);
+		}
+		/* Enable DTO */
+		if (clk_src->cs_mask->PIPE0_DTO_SRC_SEL)
+			if (encoding == DP_128b_132b_ENCODING)
+				REG_UPDATE_2(PIXEL_RATE_CNTL[inst],
+						DP_DTO0_ENABLE, 1,
+						PIPE0_DTO_SRC_SEL, 2);
+			else
+				REG_UPDATE_2(PIXEL_RATE_CNTL[inst],
+						DP_DTO0_ENABLE, 1,
+						PIPE0_DTO_SRC_SEL, 1);
+		else
+			REG_UPDATE(PIXEL_RATE_CNTL[inst],
+					DP_DTO0_ENABLE, 1);
+	} else {
+
+		if (clk_src->cs_mask->PIPE0_DTO_SRC_SEL)
+			REG_UPDATE(PIXEL_RATE_CNTL[inst],
+					PIPE0_DTO_SRC_SEL, 0);
+
+		/*ATOMBIOS expects pixel rate adjusted by deep color ratio)*/
+		bp_pc_params.controller_id = pix_clk_params->controller_id;
+		bp_pc_params.pll_id = clock_source->id;
+		bp_pc_params.target_pixel_clock_100hz = pll_settings->actual_pix_clk_100hz;
+		bp_pc_params.encoder_object_id = pix_clk_params->encoder_object_id;
+		bp_pc_params.signal_type = pix_clk_params->signal_type;
+
+		// Make sure we send the correct color depth to DMUB for HDMI
+		if (pix_clk_params->signal_type == SIGNAL_TYPE_HDMI_TYPE_A) {
+			switch (pix_clk_params->color_depth) {
+			case COLOR_DEPTH_888:
+				bp_pc_colour_depth = TRANSMITTER_COLOR_DEPTH_24;
+				break;
+			case COLOR_DEPTH_101010:
+				bp_pc_colour_depth = TRANSMITTER_COLOR_DEPTH_30;
+				break;
+			case COLOR_DEPTH_121212:
+				bp_pc_colour_depth = TRANSMITTER_COLOR_DEPTH_36;
+				break;
+			case COLOR_DEPTH_161616:
+				bp_pc_colour_depth = TRANSMITTER_COLOR_DEPTH_48;
+				break;
+			default:
+				bp_pc_colour_depth = TRANSMITTER_COLOR_DEPTH_24;
+				break;
+			}
+			bp_pc_params.color_depth = bp_pc_colour_depth;
+		}
+
+		if (clock_source->id != CLOCK_SOURCE_ID_DP_DTO) {
+			bp_pc_params.flags.SET_GENLOCK_REF_DIV_SRC =
+							pll_settings->use_external_clk;
+			bp_pc_params.flags.SET_XTALIN_REF_SRC =
+							!pll_settings->use_external_clk;
+			if (pix_clk_params->flags.SUPPORT_YCBCR420) {
+				bp_pc_params.flags.SUPPORT_YUV_420 = 1;
+			}
+		}
+		if (clk_src->bios->funcs->set_pixel_clock(
+				clk_src->bios, &bp_pc_params) != BP_RESULT_OK)
+			return false;
+		/* Resync deep color DTO */
+		if (clock_source->id != CLOCK_SOURCE_ID_DP_DTO)
+			dce112_program_pixel_clk_resync(clk_src,
+						pix_clk_params->signal_type,
+						pix_clk_params->color_depth,
+						pix_clk_params->flags.SUPPORT_YCBCR420);
+	}
+
+	return true;
+}
+
+static bool dcn401_program_pix_clk(
+		struct clock_source *clock_source,
+		struct pixel_clk_params *pix_clk_params,
+		enum dp_link_encoding encoding,
+		struct pll_settings *pll_settings)
+{
+	struct dce110_clk_src *clk_src = TO_DCE110_CLK_SRC(clock_source);
+	unsigned int inst = pix_clk_params->controller_id - CONTROLLER_ID_D0;
+	const struct pixel_rate_range_table_entry *e =
+			look_up_in_video_optimized_rate_tlb(pix_clk_params->requested_pix_clk_100hz / 10);
+	struct bp_pixel_clock_parameters bp_pc_params = {0};
+	enum transmitter_color_depth bp_pc_colour_depth = TRANSMITTER_COLOR_DEPTH_24;
+	struct dp_dto_params dto_params = { 0 };
+
+	dto_params.otg_inst = inst;
+	dto_params.signal = pix_clk_params->signal_type;
+
+	// all but TMDS gets Driver to program DP_DTO without calling VBIOS Command table
+	if (!dc_is_tmds_signal(pix_clk_params->signal_type)) {
+		long long dtbclk_p_src_clk_khz;
+
+		dtbclk_p_src_clk_khz = clock_source->ctx->dc->clk_mgr->dprefclk_khz;
+		dto_params.clk_src = DPREFCLK;
+
+		if (e) {
+			dto_params.pixclk_hz = e->target_pixel_rate_khz;
+			dto_params.pixclk_hz *= e->mult_factor;
+			dto_params.refclk_hz = dtbclk_p_src_clk_khz;
+			dto_params.refclk_hz *= e->div_factor;
+		} else {
+			dto_params.pixclk_hz = pix_clk_params->requested_pix_clk_100hz;
+			dto_params.pixclk_hz *= 100;
+			dto_params.refclk_hz = dtbclk_p_src_clk_khz;
+			dto_params.refclk_hz *= 1000;
+		}
+
+		/* enable DP DTO */
+		clock_source->ctx->dc->res_pool->dccg->funcs->set_dp_dto(
+				clock_source->ctx->dc->res_pool->dccg,
+				&dto_params);
+
+	} else {
+		if (pll_settings->actual_pix_clk_100hz > 6000000UL)
+			return false;
+
+		/* disables DP DTO when provided with TMDS signal type */
+		clock_source->ctx->dc->res_pool->dccg->funcs->set_dp_dto(
+				clock_source->ctx->dc->res_pool->dccg,
+				&dto_params);
+
+		/*ATOMBIOS expects pixel rate adjusted by deep color ratio)*/
+		bp_pc_params.controller_id = pix_clk_params->controller_id;
+		bp_pc_params.pll_id = clock_source->id;
+		bp_pc_params.target_pixel_clock_100hz = pll_settings->actual_pix_clk_100hz;
+		bp_pc_params.encoder_object_id = pix_clk_params->encoder_object_id;
+		bp_pc_params.signal_type = pix_clk_params->signal_type;
+
+		// Make sure we send the correct color depth to DMUB for HDMI
+		if (pix_clk_params->signal_type == SIGNAL_TYPE_HDMI_TYPE_A) {
+			switch (pix_clk_params->color_depth) {
+			case COLOR_DEPTH_888:
+				bp_pc_colour_depth = TRANSMITTER_COLOR_DEPTH_24;
+				break;
+			case COLOR_DEPTH_101010:
+				bp_pc_colour_depth = TRANSMITTER_COLOR_DEPTH_30;
+				break;
+			case COLOR_DEPTH_121212:
+				bp_pc_colour_depth = TRANSMITTER_COLOR_DEPTH_36;
+				break;
+			case COLOR_DEPTH_161616:
+				bp_pc_colour_depth = TRANSMITTER_COLOR_DEPTH_48;
+				break;
+			default:
+				bp_pc_colour_depth = TRANSMITTER_COLOR_DEPTH_24;
+				break;
+			}
+			bp_pc_params.color_depth = bp_pc_colour_depth;
+		}
+
+		if (clock_source->id != CLOCK_SOURCE_ID_DP_DTO) {
+			bp_pc_params.flags.SET_GENLOCK_REF_DIV_SRC =
+							pll_settings->use_external_clk;
+			bp_pc_params.flags.SET_XTALIN_REF_SRC =
+							!pll_settings->use_external_clk;
+			if (pix_clk_params->flags.SUPPORT_YCBCR420) {
+				bp_pc_params.flags.SUPPORT_YUV_420 = 1;
+			}
+		}
+		if (clk_src->bios->funcs->set_pixel_clock(
+				clk_src->bios, &bp_pc_params) != BP_RESULT_OK)
+			return false;
+		/* Resync deep color DTO */
+		if (clock_source->id != CLOCK_SOURCE_ID_DP_DTO)
+			dce112_program_pixel_clk_resync(clk_src,
+						pix_clk_params->signal_type,
+						pix_clk_params->color_depth,
+						pix_clk_params->flags.SUPPORT_YCBCR420);
+	}
+
+	return true;
+}
+
+static bool dce110_clock_source_power_down(
+		struct clock_source *clk_src)
+{
+	struct dce110_clk_src *dce110_clk_src = TO_DCE110_CLK_SRC(clk_src);
+	enum bp_result bp_result;
+	struct bp_pixel_clock_parameters bp_pixel_clock_params = {0};
+
+	if (clk_src->dp_clk_src)
+		return true;
+
+	/* If Pixel Clock is 0 it means Power Down Pll*/
+	bp_pixel_clock_params.controller_id = CONTROLLER_ID_UNDEFINED;
+	bp_pixel_clock_params.pll_id = clk_src->id;
+	bp_pixel_clock_params.flags.FORCE_PROGRAMMING_OF_PLL = 1;
+
+	/*Call ASICControl to process ATOMBIOS Exec table*/
+	bp_result = dce110_clk_src->bios->funcs->set_pixel_clock(
+			dce110_clk_src->bios,
+			&bp_pixel_clock_params);
+
+	return bp_result == BP_RESULT_OK;
+}
+
+static bool get_pixel_clk_frequency_100hz(
+		const struct clock_source *clock_source,
+		unsigned int inst,
+		unsigned int *pixel_clk_khz)
+{
+	struct dce110_clk_src *clk_src = TO_DCE110_CLK_SRC(clock_source);
+	unsigned int clock_hz = 0;
+	unsigned int modulo_hz = 0;
+	unsigned int dp_dto_ref_khz = clock_source->ctx->dc->clk_mgr->dprefclk_khz;
+
+	if (clock_source->id == CLOCK_SOURCE_ID_DP_DTO) {
+		clock_hz = REG_READ(PHASE[inst]);
+
+		if (clock_source->ctx->dc->hwss.enable_vblanks_synchronization &&
+			clock_source->ctx->dc->config.vblank_alignment_max_frame_time_diff > 0) {
+			/* NOTE: In case VBLANK syncronization is enabled, MODULO may
+			 * not be programmed equal to DPREFCLK
+			 */
+			modulo_hz = REG_READ(MODULO[inst]);
+			if (modulo_hz)
+				*pixel_clk_khz = div_u64((uint64_t)clock_hz*
+					dp_dto_ref_khz*10,
+					modulo_hz);
+			else
+				*pixel_clk_khz = 0;
+		} else {
+			/* NOTE: There is agreement with VBIOS here that MODULO is
+			 * programmed equal to DPREFCLK, in which case PHASE will be
+			 * equivalent to pixel clock.
+			 */
+			*pixel_clk_khz = clock_hz / 100;
+		}
+		return true;
+	}
+
+	return false;
+}
+
+/* this table is use to find *1.001 and /1.001 pixel rates from non-precise pixel rate */
+const struct pixel_rate_range_table_entry video_optimized_pixel_rates[] = {
+	// /1.001 rates
+	{25170, 25180, 25200, 1000, 1001},	//25.2MHz   ->   25.17
+	{59340, 59350, 59400, 1000, 1001},	//59.4Mhz   ->   59.340
+	{74170, 74180, 74250, 1000, 1001},	//74.25Mhz  ->   74.1758
+	{89910, 90000, 90000, 1000, 1001},	//90Mhz     ->   89.91
+	{125870, 125880, 126000, 1000, 1001},	//126Mhz    ->  125.87
+	{148350, 148360, 148500, 1000, 1001},	//148.5Mhz  ->  148.3516
+	{167830, 167840, 168000, 1000, 1001},	//168Mhz    ->  167.83
+	{222520, 222530, 222750, 1000, 1001},	//222.75Mhz ->  222.527
+	{257140, 257150, 257400, 1000, 1001},	//257.4Mhz  ->  257.1429
+	{296700, 296710, 297000, 1000, 1001},	//297Mhz    ->  296.7033
+	{342850, 342860, 343200, 1000, 1001},	//343.2Mhz  ->  342.857
+	{395600, 395610, 396000, 1000, 1001},	//396Mhz    ->  395.6
+	{409090, 409100, 409500, 1000, 1001},	//409.5Mhz  ->  409.091
+	{445050, 445060, 445500, 1000, 1001},	//445.5Mhz  ->  445.055
+	{467530, 467540, 468000, 1000, 1001},	//468Mhz    ->  467.5325
+	{519230, 519240, 519750, 1000, 1001},	//519.75Mhz ->  519.231
+	{525970, 525980, 526500, 1000, 1001},	//526.5Mhz  ->  525.974
+	{545450, 545460, 546000, 1000, 1001},	//546Mhz    ->  545.455
+	{593400, 593410, 594000, 1000, 1001},	//594Mhz    ->  593.4066
+	{623370, 623380, 624000, 1000, 1001},	//624Mhz    ->  623.377
+	{692300, 692310, 693000, 1000, 1001},	//693Mhz    ->  692.308
+	{701290, 701300, 702000, 1000, 1001},	//702Mhz    ->  701.2987
+	{791200, 791210, 792000, 1000, 1001},	//792Mhz    ->  791.209
+	{890100, 890110, 891000, 1000, 1001},	//891Mhz    ->  890.1099
+	{1186810, 1186820, 1188000, 1000, 1001},//1188Mhz   -> 1186.8131
+
+	// *1.001 rates
+	{27020, 27030, 27000, 1001, 1000}, //27Mhz
+	{54050, 54060, 54000, 1001, 1000}, //54Mhz
+	{108100, 108110, 108000, 1001, 1000},//108Mhz
+};
+
+const struct pixel_rate_range_table_entry *look_up_in_video_optimized_rate_tlb(
+		unsigned int pixel_rate_khz)
+{
+	int i;
+
+	for (i = 0; i < NUM_ELEMENTS(video_optimized_pixel_rates); i++) {
+		const struct pixel_rate_range_table_entry *e = &video_optimized_pixel_rates[i];
+
+		if (e->range_min_khz <= pixel_rate_khz && pixel_rate_khz <= e->range_max_khz) {
+			return e;
+		}
+	}
+
+	return NULL;
+}
+
+static bool dcn20_program_pix_clk(
+		struct clock_source *clock_source,
+		struct pixel_clk_params *pix_clk_params,
+		enum dp_link_encoding encoding,
+		struct pll_settings *pll_settings)
+{
+	struct dce110_clk_src *clk_src = TO_DCE110_CLK_SRC(clock_source);
+	unsigned int inst = pix_clk_params->controller_id - CONTROLLER_ID_D0;
+
+	dce112_program_pix_clk(clock_source, pix_clk_params, encoding, pll_settings);
+
+	if (clock_source->ctx->dc->hwss.enable_vblanks_synchronization &&
+			clock_source->ctx->dc->config.vblank_alignment_max_frame_time_diff > 0) {
+		/* NOTE: In case VBLANK syncronization is enabled,
+		 * we need to set modulo to default DPREFCLK first
+		 * dce112_program_pix_clk does not set default DPREFCLK
+		 */
+		REG_WRITE(MODULO[inst],
+			clock_source->ctx->dc->clk_mgr->dprefclk_khz*1000);
+	}
+	return true;
+}
+
+static bool dcn20_override_dp_pix_clk(
+		struct clock_source *clock_source,
+		unsigned int inst,
+		unsigned int pixel_clk,
+		unsigned int ref_clk)
+{
+	struct dce110_clk_src *clk_src = TO_DCE110_CLK_SRC(clock_source);
+
+	REG_UPDATE(PIXEL_RATE_CNTL[inst], DP_DTO0_ENABLE, 0);
+	REG_WRITE(PHASE[inst], pixel_clk);
+	REG_WRITE(MODULO[inst], ref_clk);
+	REG_UPDATE(PIXEL_RATE_CNTL[inst], DP_DTO0_ENABLE, 1);
+	return true;
+}
+
+static const struct clock_source_funcs dcn20_clk_src_funcs = {
+	.cs_power_down = dce110_clock_source_power_down,
+	.program_pix_clk = dcn20_program_pix_clk,
+	.get_pix_clk_dividers = dce112_get_pix_clk_dividers,
+	.get_pixel_clk_frequency_100hz = get_pixel_clk_frequency_100hz,
+	.override_dp_pix_clk = dcn20_override_dp_pix_clk
+};
+
+static bool dcn3_program_pix_clk(
+		struct clock_source *clock_source,
+		struct pixel_clk_params *pix_clk_params,
+		enum dp_link_encoding encoding,
+		struct pll_settings *pll_settings)
+{
+	struct dce110_clk_src *clk_src = TO_DCE110_CLK_SRC(clock_source);
+	unsigned int inst = pix_clk_params->controller_id - CONTROLLER_ID_D0;
+	unsigned int dp_dto_ref_khz = clock_source->ctx->dc->clk_mgr->dprefclk_khz;
+	const struct pixel_rate_range_table_entry *e =
+			look_up_in_video_optimized_rate_tlb(pix_clk_params->requested_pix_clk_100hz / 10);
+
+	// For these signal types Driver to program DP_DTO without calling VBIOS Command table
+	if (dc_is_dp_signal(pix_clk_params->signal_type)) {
+		if (e) {
+			/* Set DTO values: phase = target clock, modulo = reference clock*/
+			REG_WRITE(PHASE[inst], e->target_pixel_rate_khz * e->mult_factor);
+			REG_WRITE(MODULO[inst], dp_dto_ref_khz * e->div_factor);
+		} else {
+			/* Set DTO values: phase = target clock, modulo = reference clock*/
+			REG_WRITE(PHASE[inst], pll_settings->actual_pix_clk_100hz * 100);
+			REG_WRITE(MODULO[inst], dp_dto_ref_khz * 1000);
+		}
+		/* Enable DTO */
+		if (clk_src->cs_mask->PIPE0_DTO_SRC_SEL)
+			REG_UPDATE_2(PIXEL_RATE_CNTL[inst],
+					DP_DTO0_ENABLE, 1,
+					PIPE0_DTO_SRC_SEL, 1);
+		else
+			REG_UPDATE(PIXEL_RATE_CNTL[inst],
+					DP_DTO0_ENABLE, 1);
+	} else
+		// For other signal types(HDMI_TYPE_A, DVI) Driver still to call VBIOS Command table
+		dce112_program_pix_clk(clock_source, pix_clk_params, encoding, pll_settings);
+
+	return true;
+}
+
+static uint32_t dcn3_get_pix_clk_dividers(
+		struct clock_source *cs,
+		struct pixel_clk_params *pix_clk_params,
+		struct pll_settings *pll_settings)
+{
+	unsigned long long actual_pix_clk_100Hz = pix_clk_params ? pix_clk_params->requested_pix_clk_100hz : 0;
+	struct dce110_clk_src *clk_src = TO_DCE110_CLK_SRC(cs);
+
+	DC_LOGGER_INIT();
+
+	if (pix_clk_params == NULL || pll_settings == NULL
+			|| pix_clk_params->requested_pix_clk_100hz == 0) {
+		DC_LOG_ERROR(
+			"%s: Invalid parameters!!\n", __func__);
+		return -1;
+	}
+
+	memset(pll_settings, 0, sizeof(*pll_settings));
+	/* Adjust for HDMI Type A deep color */
+	if (pix_clk_params->signal_type == SIGNAL_TYPE_HDMI_TYPE_A) {
+		switch (pix_clk_params->color_depth) {
+		case COLOR_DEPTH_101010:
+			actual_pix_clk_100Hz = (actual_pix_clk_100Hz * 5) >> 2;
+			break;
+		case COLOR_DEPTH_121212:
+			actual_pix_clk_100Hz = (actual_pix_clk_100Hz * 6) >> 2;
+			break;
+		case COLOR_DEPTH_161616:
+			actual_pix_clk_100Hz = actual_pix_clk_100Hz * 2;
+			break;
+		default:
+			break;
+		}
+	}
+	pll_settings->actual_pix_clk_100hz = (unsigned int) actual_pix_clk_100Hz;
+	pll_settings->adjusted_pix_clk_100hz = (unsigned int) actual_pix_clk_100Hz;
+	pll_settings->calculated_pix_clk_100hz = (unsigned int) actual_pix_clk_100Hz;
+
+	return 0;
+}
+
+static const struct clock_source_funcs dcn3_clk_src_funcs = {
+	.cs_power_down = dce110_clock_source_power_down,
+	.program_pix_clk = dcn3_program_pix_clk,
+	.get_pix_clk_dividers = dcn3_get_pix_clk_dividers,
+	.get_pixel_clk_frequency_100hz = get_pixel_clk_frequency_100hz
+};
+
+static const struct clock_source_funcs dcn31_clk_src_funcs = {
+	.cs_power_down = dce110_clock_source_power_down,
+	.program_pix_clk = dcn31_program_pix_clk,
+	.get_pix_clk_dividers = dcn3_get_pix_clk_dividers,
+	.get_pixel_clk_frequency_100hz = get_pixel_clk_frequency_100hz
+};
+
+static const struct clock_source_funcs dcn401_clk_src_funcs = {
+	.cs_power_down = dce110_clock_source_power_down,
+	.program_pix_clk = dcn401_program_pix_clk,
+	.get_pix_clk_dividers = dcn3_get_pix_clk_dividers,
+	.get_pixel_clk_frequency_100hz = get_pixel_clk_frequency_100hz
+};
+
+/*****************************************/
+/* Constructor                           */
+/*****************************************/
+
+static const struct clock_source_funcs dce112_clk_src_funcs = {
+	.cs_power_down = dce110_clock_source_power_down,
+	.program_pix_clk = dce112_program_pix_clk,
+	.get_pix_clk_dividers = dce112_get_pix_clk_dividers,
+	.get_pixel_clk_frequency_100hz = get_pixel_clk_frequency_100hz
+};
+static const struct clock_source_funcs dce110_clk_src_funcs = {
+	.cs_power_down = dce110_clock_source_power_down,
+	.program_pix_clk = dce110_program_pix_clk,
+	.get_pix_clk_dividers = dce110_get_pix_clk_dividers,
+	.get_pixel_clk_frequency_100hz = get_pixel_clk_frequency_100hz
+};
+
+
+static void get_ss_info_from_atombios(
+		struct dce110_clk_src *clk_src,
+		enum as_signal_type as_signal,
+		struct spread_spectrum_data *spread_spectrum_data[],
+		uint32_t *ss_entries_num)
+{
+	enum bp_result bp_result = BP_RESULT_FAILURE;
+	struct spread_spectrum_info *ss_info;
+	struct spread_spectrum_data *ss_data;
+	struct spread_spectrum_info *ss_info_cur;
+	struct spread_spectrum_data *ss_data_cur;
+	uint32_t i;
+	DC_LOGGER_INIT();
+	if (ss_entries_num == NULL) {
+		DC_LOG_SYNC(
+			"Invalid entry !!!\n");
+		return;
+	}
+	if (spread_spectrum_data == NULL) {
+		DC_LOG_SYNC(
+			"Invalid array pointer!!!\n");
+		return;
+	}
+
+	spread_spectrum_data[0] = NULL;
+	*ss_entries_num = 0;
+
+	*ss_entries_num = clk_src->bios->funcs->get_ss_entry_number(
+			clk_src->bios,
+			as_signal);
+
+	if (*ss_entries_num == 0)
+		return;
+
+	ss_info = kcalloc(*ss_entries_num,
+			  sizeof(struct spread_spectrum_info),
+			  GFP_KERNEL);
+	ss_info_cur = ss_info;
+	if (ss_info == NULL)
+		return;
+
+	ss_data = kcalloc(*ss_entries_num,
+			  sizeof(struct spread_spectrum_data),
+			  GFP_KERNEL);
+	if (ss_data == NULL)
+		goto out_free_info;
+
+	for (i = 0, ss_info_cur = ss_info;
+		i < (*ss_entries_num);
+		++i, ++ss_info_cur) {
+
+		bp_result = clk_src->bios->funcs->get_spread_spectrum_info(
+				clk_src->bios,
+				as_signal,
+				i,
+				ss_info_cur);
+
+		if (bp_result != BP_RESULT_OK)
+			goto out_free_data;
+	}
+
+	for (i = 0, ss_info_cur = ss_info, ss_data_cur = ss_data;
+		i < (*ss_entries_num);
+		++i, ++ss_info_cur, ++ss_data_cur) {
+
+		if (ss_info_cur->type.STEP_AND_DELAY_INFO != false) {
+			DC_LOG_SYNC(
+				"Invalid ATOMBIOS SS Table!!!\n");
+			goto out_free_data;
+		}
+
+		/* for HDMI check SS percentage,
+		 * if it is > 6 (0.06%), the ATOMBIOS table info is invalid*/
+		if (as_signal == AS_SIGNAL_TYPE_HDMI
+				&& ss_info_cur->spread_spectrum_percentage > 6){
+			/* invalid input, do nothing */
+			DC_LOG_SYNC(
+				"Invalid SS percentage ");
+			DC_LOG_SYNC(
+				"for HDMI in ATOMBIOS info Table!!!\n");
+			continue;
+		}
+		if (ss_info_cur->spread_percentage_divider == 1000) {
+			/* Keep previous precision from ATOMBIOS for these
+			* in case new precision set by ATOMBIOS for these
+			* (otherwise all code in DCE specific classes
+			* for all previous ASICs would need
+			* to be updated for SS calculations,
+			* Audio SS compensation and DP DTO SS compensation
+			* which assumes fixed SS percentage Divider = 100)*/
+			ss_info_cur->spread_spectrum_percentage /= 10;
+			ss_info_cur->spread_percentage_divider = 100;
+		}
+
+		ss_data_cur->freq_range_khz = ss_info_cur->target_clock_range;
+		ss_data_cur->percentage =
+				ss_info_cur->spread_spectrum_percentage;
+		ss_data_cur->percentage_divider =
+				ss_info_cur->spread_percentage_divider;
+		ss_data_cur->modulation_freq_hz =
+				ss_info_cur->spread_spectrum_range;
+
+		if (ss_info_cur->type.CENTER_MODE)
+			ss_data_cur->flags.CENTER_SPREAD = 1;
+
+		if (ss_info_cur->type.EXTERNAL)
+			ss_data_cur->flags.EXTERNAL_SS = 1;
+
+	}
+
+	*spread_spectrum_data = ss_data;
+	kfree(ss_info);
+	return;
+
+out_free_data:
+	kfree(ss_data);
+	*ss_entries_num = 0;
+out_free_info:
+	kfree(ss_info);
+}
+
+static void ss_info_from_atombios_create(
+	struct dce110_clk_src *clk_src)
+{
+	get_ss_info_from_atombios(
+		clk_src,
+		AS_SIGNAL_TYPE_DISPLAY_PORT,
+		&clk_src->dp_ss_params,
+		&clk_src->dp_ss_params_cnt);
+	get_ss_info_from_atombios(
+		clk_src,
+		AS_SIGNAL_TYPE_HDMI,
+		&clk_src->hdmi_ss_params,
+		&clk_src->hdmi_ss_params_cnt);
+	get_ss_info_from_atombios(
+		clk_src,
+		AS_SIGNAL_TYPE_DVI,
+		&clk_src->dvi_ss_params,
+		&clk_src->dvi_ss_params_cnt);
+	get_ss_info_from_atombios(
+		clk_src,
+		AS_SIGNAL_TYPE_LVDS,
+		&clk_src->lvds_ss_params,
+		&clk_src->lvds_ss_params_cnt);
+}
+
+static bool calc_pll_max_vco_construct(
+			struct calc_pll_clock_source *calc_pll_cs,
+			struct calc_pll_clock_source_init_data *init_data)
+{
+	uint32_t i;
+	struct dc_firmware_info *fw_info;
+	if (calc_pll_cs == NULL ||
+			init_data == NULL ||
+			init_data->bp == NULL)
+		return false;
+
+	if (!init_data->bp->fw_info_valid)
+		return false;
+
+	fw_info = &init_data->bp->fw_info;
+	calc_pll_cs->ctx = init_data->ctx;
+	calc_pll_cs->ref_freq_khz = fw_info->pll_info.crystal_frequency;
+	calc_pll_cs->min_vco_khz =
+			fw_info->pll_info.min_output_pxl_clk_pll_frequency;
+	calc_pll_cs->max_vco_khz =
+			fw_info->pll_info.max_output_pxl_clk_pll_frequency;
+
+	if (init_data->max_override_input_pxl_clk_pll_freq_khz != 0)
+		calc_pll_cs->max_pll_input_freq_khz =
+			init_data->max_override_input_pxl_clk_pll_freq_khz;
+	else
+		calc_pll_cs->max_pll_input_freq_khz =
+			fw_info->pll_info.max_input_pxl_clk_pll_frequency;
+
+	if (init_data->min_override_input_pxl_clk_pll_freq_khz != 0)
+		calc_pll_cs->min_pll_input_freq_khz =
+			init_data->min_override_input_pxl_clk_pll_freq_khz;
+	else
+		calc_pll_cs->min_pll_input_freq_khz =
+			fw_info->pll_info.min_input_pxl_clk_pll_frequency;
+
+	calc_pll_cs->min_pix_clock_pll_post_divider =
+			init_data->min_pix_clk_pll_post_divider;
+	calc_pll_cs->max_pix_clock_pll_post_divider =
+			init_data->max_pix_clk_pll_post_divider;
+	calc_pll_cs->min_pll_ref_divider =
+			init_data->min_pll_ref_divider;
+	calc_pll_cs->max_pll_ref_divider =
+			init_data->max_pll_ref_divider;
+
+	if (init_data->num_fract_fb_divider_decimal_point == 0 ||
+		init_data->num_fract_fb_divider_decimal_point_precision >
+				init_data->num_fract_fb_divider_decimal_point) {
+		DC_LOG_ERROR(
+			"The dec point num or precision is incorrect!");
+		return false;
+	}
+	if (init_data->num_fract_fb_divider_decimal_point_precision == 0) {
+		DC_LOG_ERROR(
+			"Incorrect fract feedback divider precision num!");
+		return false;
+	}
+
+	calc_pll_cs->fract_fb_divider_decimal_points_num =
+				init_data->num_fract_fb_divider_decimal_point;
+	calc_pll_cs->fract_fb_divider_precision =
+			init_data->num_fract_fb_divider_decimal_point_precision;
+	calc_pll_cs->fract_fb_divider_factor = 1;
+	for (i = 0; i < calc_pll_cs->fract_fb_divider_decimal_points_num; ++i)
+		calc_pll_cs->fract_fb_divider_factor *= 10;
+
+	calc_pll_cs->fract_fb_divider_precision_factor = 1;
+	for (
+		i = 0;
+		i < (calc_pll_cs->fract_fb_divider_decimal_points_num -
+				calc_pll_cs->fract_fb_divider_precision);
+		++i)
+		calc_pll_cs->fract_fb_divider_precision_factor *= 10;
+
+	return true;
+}
+
+bool dce110_clk_src_construct(
+	struct dce110_clk_src *clk_src,
+	struct dc_context *ctx,
+	struct dc_bios *bios,
+	enum clock_source_id id,
+	const struct dce110_clk_src_regs *regs,
+	const struct dce110_clk_src_shift *cs_shift,
+	const struct dce110_clk_src_mask *cs_mask)
+{
+	struct calc_pll_clock_source_init_data calc_pll_cs_init_data_hdmi;
+	struct calc_pll_clock_source_init_data calc_pll_cs_init_data;
+
+	clk_src->base.ctx = ctx;
+	clk_src->bios = bios;
+	clk_src->base.id = id;
+	clk_src->base.funcs = &dce110_clk_src_funcs;
+
+	clk_src->regs = regs;
+	clk_src->cs_shift = cs_shift;
+	clk_src->cs_mask = cs_mask;
+
+	if (!clk_src->bios->fw_info_valid) {
+		ASSERT_CRITICAL(false);
+		goto unexpected_failure;
+	}
+
+	clk_src->ext_clk_khz = clk_src->bios->fw_info.external_clock_source_frequency_for_dp;
+
+	/* structure normally used with PLL ranges from ATOMBIOS; DS on by default */
+	calc_pll_cs_init_data.bp = bios;
+	calc_pll_cs_init_data.min_pix_clk_pll_post_divider = 1;
+	calc_pll_cs_init_data.max_pix_clk_pll_post_divider =
+			clk_src->cs_mask->PLL_POST_DIV_PIXCLK;
+	calc_pll_cs_init_data.min_pll_ref_divider =	1;
+	calc_pll_cs_init_data.max_pll_ref_divider =	clk_src->cs_mask->PLL_REF_DIV;
+	/* when 0 use minInputPxlClkPLLFrequencyInKHz from firmwareInfo*/
+	calc_pll_cs_init_data.min_override_input_pxl_clk_pll_freq_khz =	0;
+	/* when 0 use maxInputPxlClkPLLFrequencyInKHz from firmwareInfo*/
+	calc_pll_cs_init_data.max_override_input_pxl_clk_pll_freq_khz =	0;
+	/*numberOfFractFBDividerDecimalPoints*/
+	calc_pll_cs_init_data.num_fract_fb_divider_decimal_point =
+			FRACT_FB_DIVIDER_DEC_POINTS_MAX_NUM;
+	/*number of decimal point to round off for fractional feedback divider value*/
+	calc_pll_cs_init_data.num_fract_fb_divider_decimal_point_precision =
+			FRACT_FB_DIVIDER_DEC_POINTS_MAX_NUM;
+	calc_pll_cs_init_data.ctx =	ctx;
+
+	/*structure for HDMI, no SS or SS% <= 0.06% for 27 MHz Ref clock */
+	calc_pll_cs_init_data_hdmi.bp = bios;
+	calc_pll_cs_init_data_hdmi.min_pix_clk_pll_post_divider = 1;
+	calc_pll_cs_init_data_hdmi.max_pix_clk_pll_post_divider =
+			clk_src->cs_mask->PLL_POST_DIV_PIXCLK;
+	calc_pll_cs_init_data_hdmi.min_pll_ref_divider = 1;
+	calc_pll_cs_init_data_hdmi.max_pll_ref_divider = clk_src->cs_mask->PLL_REF_DIV;
+	/* when 0 use minInputPxlClkPLLFrequencyInKHz from firmwareInfo*/
+	calc_pll_cs_init_data_hdmi.min_override_input_pxl_clk_pll_freq_khz = 13500;
+	/* when 0 use maxInputPxlClkPLLFrequencyInKHz from firmwareInfo*/
+	calc_pll_cs_init_data_hdmi.max_override_input_pxl_clk_pll_freq_khz = 27000;
+	/*numberOfFractFBDividerDecimalPoints*/
+	calc_pll_cs_init_data_hdmi.num_fract_fb_divider_decimal_point =
+			FRACT_FB_DIVIDER_DEC_POINTS_MAX_NUM;
+	/*number of decimal point to round off for fractional feedback divider value*/
+	calc_pll_cs_init_data_hdmi.num_fract_fb_divider_decimal_point_precision =
+			FRACT_FB_DIVIDER_DEC_POINTS_MAX_NUM;
+	calc_pll_cs_init_data_hdmi.ctx = ctx;
+
+	clk_src->ref_freq_khz = clk_src->bios->fw_info.pll_info.crystal_frequency;
+
+	if (clk_src->base.id == CLOCK_SOURCE_ID_EXTERNAL)
+		return true;
+
+	/* PLL only from here on */
+	ss_info_from_atombios_create(clk_src);
+
+	if (!calc_pll_max_vco_construct(
+			&clk_src->calc_pll,
+			&calc_pll_cs_init_data)) {
+		ASSERT_CRITICAL(false);
+		goto unexpected_failure;
+	}
+
+
+	calc_pll_cs_init_data_hdmi.
+			min_override_input_pxl_clk_pll_freq_khz = clk_src->ref_freq_khz/2;
+	calc_pll_cs_init_data_hdmi.
+			max_override_input_pxl_clk_pll_freq_khz = clk_src->ref_freq_khz;
+
+
+	if (!calc_pll_max_vco_construct(
+			&clk_src->calc_pll_hdmi, &calc_pll_cs_init_data_hdmi)) {
+		ASSERT_CRITICAL(false);
+		goto unexpected_failure;
+	}
+
+	return true;
+
+unexpected_failure:
+	return false;
+}
+
+bool dce112_clk_src_construct(
+	struct dce110_clk_src *clk_src,
+	struct dc_context *ctx,
+	struct dc_bios *bios,
+	enum clock_source_id id,
+	const struct dce110_clk_src_regs *regs,
+	const struct dce110_clk_src_shift *cs_shift,
+	const struct dce110_clk_src_mask *cs_mask)
+{
+	clk_src->base.ctx = ctx;
+	clk_src->bios = bios;
+	clk_src->base.id = id;
+	clk_src->base.funcs = &dce112_clk_src_funcs;
+
+	clk_src->regs = regs;
+	clk_src->cs_shift = cs_shift;
+	clk_src->cs_mask = cs_mask;
+
+	if (!clk_src->bios->fw_info_valid) {
+		ASSERT_CRITICAL(false);
+		return false;
+	}
+
+	clk_src->ext_clk_khz = clk_src->bios->fw_info.external_clock_source_frequency_for_dp;
+
+	return true;
+}
+
+bool dcn20_clk_src_construct(
+	struct dce110_clk_src *clk_src,
+	struct dc_context *ctx,
+	struct dc_bios *bios,
+	enum clock_source_id id,
+	const struct dce110_clk_src_regs *regs,
+	const struct dce110_clk_src_shift *cs_shift,
+	const struct dce110_clk_src_mask *cs_mask)
+{
+	bool ret = dce112_clk_src_construct(clk_src, ctx, bios, id, regs, cs_shift, cs_mask);
+
+	clk_src->base.funcs = &dcn20_clk_src_funcs;
+
+	return ret;
+}
+
+bool dcn3_clk_src_construct(
+	struct dce110_clk_src *clk_src,
+	struct dc_context *ctx,
+	struct dc_bios *bios,
+	enum clock_source_id id,
+	const struct dce110_clk_src_regs *regs,
+	const struct dce110_clk_src_shift *cs_shift,
+	const struct dce110_clk_src_mask *cs_mask)
+{
+	bool ret = dce112_clk_src_construct(clk_src, ctx, bios, id, regs, cs_shift, cs_mask);
+
+	clk_src->base.funcs = &dcn3_clk_src_funcs;
+
+	return ret;
+}
+
+bool dcn31_clk_src_construct(
+	struct dce110_clk_src *clk_src,
+	struct dc_context *ctx,
+	struct dc_bios *bios,
+	enum clock_source_id id,
+	const struct dce110_clk_src_regs *regs,
+	const struct dce110_clk_src_shift *cs_shift,
+	const struct dce110_clk_src_mask *cs_mask)
+{
+	bool ret = dce112_clk_src_construct(clk_src, ctx, bios, id, regs, cs_shift, cs_mask);
+
+	clk_src->base.funcs = &dcn31_clk_src_funcs;
+
+	return ret;
+}
+
+bool dcn401_clk_src_construct(
+	struct dce110_clk_src *clk_src,
+	struct dc_context *ctx,
+	struct dc_bios *bios,
+	enum clock_source_id id,
+	const struct dce110_clk_src_regs *regs,
+	const struct dce110_clk_src_shift *cs_shift,
+	const struct dce110_clk_src_mask *cs_mask)
+{
+	bool ret = dce112_clk_src_construct(clk_src, ctx, bios, id, regs, cs_shift, cs_mask);
+
+	clk_src->base.funcs = &dcn401_clk_src_funcs;
+
+	return ret;
+}
+bool dcn301_clk_src_construct(
+	struct dce110_clk_src *clk_src,
+	struct dc_context *ctx,
+	struct dc_bios *bios,
+	enum clock_source_id id,
+	const struct dce110_clk_src_regs *regs,
+	const struct dce110_clk_src_shift *cs_shift,
+	const struct dce110_clk_src_mask *cs_mask)
+{
+	bool ret = dce112_clk_src_construct(clk_src, ctx, bios, id, regs, cs_shift, cs_mask);
+
+	clk_src->base.funcs = &dcn3_clk_src_funcs;
+
+	return ret;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_clock_source.h b/drivers/gpu/drm/amd/display/dc/dce/dce_clock_source.h
new file mode 100644
index 000000000000..94128f7a18b1
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_clock_source.h
@@ -0,0 +1,331 @@
+/* Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_CLOCK_SOURCE_DCE_H__
+#define __DC_CLOCK_SOURCE_DCE_H__
+
+#include "../inc/clock_source.h"
+
+#define TO_DCE110_CLK_SRC(clk_src)\
+	container_of(clk_src, struct dce110_clk_src, base)
+
+#define CS_COMMON_REG_LIST_DCE_100_110(id) \
+		SRI(RESYNC_CNTL, PIXCLK, id), \
+		SRI(PLL_CNTL, BPHYC_PLL, id)
+
+#define CS_COMMON_REG_LIST_DCE_80(id) \
+		SRI(RESYNC_CNTL, PIXCLK, id), \
+		SRI(PLL_CNTL, DCCG_PLL, id)
+
+#define CS_COMMON_REG_LIST_DCE_112(id) \
+		SRI(PIXCLK_RESYNC_CNTL, PHYPLL, id)
+
+
+#define CS_SF(reg_name, field_name, post_fix)\
+	.field_name = reg_name ## __ ## field_name ## post_fix
+
+#define CS_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(mask_sh)\
+	CS_SF(PLL_CNTL, PLL_REF_DIV_SRC, mask_sh),\
+	CS_SF(PIXCLK1_RESYNC_CNTL, DCCG_DEEP_COLOR_CNTL1, mask_sh),\
+	CS_SF(PLL_POST_DIV, PLL_POST_DIV_PIXCLK, mask_sh),\
+	CS_SF(PLL_REF_DIV, PLL_REF_DIV, mask_sh)
+
+#define CS_COMMON_MASK_SH_LIST_DCE_112(mask_sh)\
+	CS_SF(PHYPLLA_PIXCLK_RESYNC_CNTL, PHYPLLA_DCCG_DEEP_COLOR_CNTL, mask_sh),\
+	CS_SF(PHYPLLA_PIXCLK_RESYNC_CNTL, PHYPLLA_PIXCLK_DOUBLE_RATE_ENABLE, mask_sh)
+
+#define CS_COMMON_REG_LIST_DCN2_0(index, pllid) \
+		SRI(PIXCLK_RESYNC_CNTL, PHYPLL, pllid),\
+		SRII(PHASE, DP_DTO, 0),\
+		SRII(PHASE, DP_DTO, 1),\
+		SRII(PHASE, DP_DTO, 2),\
+		SRII(PHASE, DP_DTO, 3),\
+		SRII(PHASE, DP_DTO, 4),\
+		SRII(PHASE, DP_DTO, 5),\
+		SRII(MODULO, DP_DTO, 0),\
+		SRII(MODULO, DP_DTO, 1),\
+		SRII(MODULO, DP_DTO, 2),\
+		SRII(MODULO, DP_DTO, 3),\
+		SRII(MODULO, DP_DTO, 4),\
+		SRII(MODULO, DP_DTO, 5),\
+		SRII(PIXEL_RATE_CNTL, OTG, 0),\
+		SRII(PIXEL_RATE_CNTL, OTG, 1),\
+		SRII(PIXEL_RATE_CNTL, OTG, 2),\
+		SRII(PIXEL_RATE_CNTL, OTG, 3),\
+		SRII(PIXEL_RATE_CNTL, OTG, 4),\
+		SRII(PIXEL_RATE_CNTL, OTG, 5)
+
+#define CS_COMMON_REG_LIST_DCN201(index, pllid) \
+		SRI(PIXCLK_RESYNC_CNTL, PHYPLL, pllid),\
+		SRII(PHASE, DP_DTO, 0),\
+		SRII(PHASE, DP_DTO, 1),\
+		SRII(MODULO, DP_DTO, 0),\
+		SRII(MODULO, DP_DTO, 1),\
+		SRII(PIXEL_RATE_CNTL, OTG, 0),\
+		SRII(PIXEL_RATE_CNTL, OTG, 1)
+
+#define CS_COMMON_REG_LIST_DCN2_1(index, pllid) \
+		SRI(PIXCLK_RESYNC_CNTL, PHYPLL, pllid),\
+		SRII(PHASE, DP_DTO, 0),\
+		SRII(PHASE, DP_DTO, 1),\
+		SRII(PHASE, DP_DTO, 2),\
+		SRII(PHASE, DP_DTO, 3),\
+		SRII(MODULO, DP_DTO, 0),\
+		SRII(MODULO, DP_DTO, 1),\
+		SRII(MODULO, DP_DTO, 2),\
+		SRII(MODULO, DP_DTO, 3),\
+		SRII(PIXEL_RATE_CNTL, OTG, 0),\
+		SRII(PIXEL_RATE_CNTL, OTG, 1),\
+		SRII(PIXEL_RATE_CNTL, OTG, 2),\
+		SRII(PIXEL_RATE_CNTL, OTG, 3)
+
+#define CS_COMMON_REG_LIST_DCN3_0(index, pllid) \
+		SRI(PIXCLK_RESYNC_CNTL, PHYPLL, pllid),\
+		SRII(PHASE, DP_DTO, 0),\
+		SRII(PHASE, DP_DTO, 1),\
+		SRII(PHASE, DP_DTO, 2),\
+		SRII(PHASE, DP_DTO, 3),\
+		SRII(MODULO, DP_DTO, 0),\
+		SRII(MODULO, DP_DTO, 1),\
+		SRII(MODULO, DP_DTO, 2),\
+		SRII(MODULO, DP_DTO, 3),\
+		SRII(PIXEL_RATE_CNTL, OTG, 0),\
+		SRII(PIXEL_RATE_CNTL, OTG, 1),\
+		SRII(PIXEL_RATE_CNTL, OTG, 2),\
+		SRII(PIXEL_RATE_CNTL, OTG, 3)
+
+#define CS_COMMON_REG_LIST_DCN3_01(index, pllid) \
+		SRI(PIXCLK_RESYNC_CNTL, PHYPLL, pllid),\
+		SRII(PHASE, DP_DTO, 0),\
+		SRII(PHASE, DP_DTO, 1),\
+		SRII(PHASE, DP_DTO, 2),\
+		SRII(PHASE, DP_DTO, 3),\
+		SRII(MODULO, DP_DTO, 0),\
+		SRII(MODULO, DP_DTO, 1),\
+		SRII(MODULO, DP_DTO, 2),\
+		SRII(MODULO, DP_DTO, 3),\
+		SRII(PIXEL_RATE_CNTL, OTG, 0),\
+		SRII(PIXEL_RATE_CNTL, OTG, 1),\
+		SRII(PIXEL_RATE_CNTL, OTG, 2),\
+		SRII(PIXEL_RATE_CNTL, OTG, 3)
+
+#define CS_COMMON_REG_LIST_DCN3_02(index, pllid) \
+		SRI(PIXCLK_RESYNC_CNTL, PHYPLL, pllid),\
+		SRII(PHASE, DP_DTO, 0),\
+		SRII(PHASE, DP_DTO, 1),\
+		SRII(PHASE, DP_DTO, 2),\
+		SRII(PHASE, DP_DTO, 3),\
+		SRII(PHASE, DP_DTO, 4),\
+		SRII(MODULO, DP_DTO, 0),\
+		SRII(MODULO, DP_DTO, 1),\
+		SRII(MODULO, DP_DTO, 2),\
+		SRII(MODULO, DP_DTO, 3),\
+		SRII(MODULO, DP_DTO, 4),\
+		SRII(PIXEL_RATE_CNTL, OTG, 0),\
+		SRII(PIXEL_RATE_CNTL, OTG, 1),\
+		SRII(PIXEL_RATE_CNTL, OTG, 2),\
+		SRII(PIXEL_RATE_CNTL, OTG, 3),\
+		SRII(PIXEL_RATE_CNTL, OTG, 4)
+
+#define CS_COMMON_REG_LIST_DCN3_03(index, pllid) \
+		SRI(PIXCLK_RESYNC_CNTL, PHYPLL, pllid),\
+		SRII(PHASE, DP_DTO, 0),\
+		SRII(PHASE, DP_DTO, 1),\
+		SRII(MODULO, DP_DTO, 0),\
+		SRII(MODULO, DP_DTO, 1),\
+		SRII(PIXEL_RATE_CNTL, OTG, 0),\
+		SRII(PIXEL_RATE_CNTL, OTG, 1)
+
+#define CS_COMMON_MASK_SH_LIST_DCN2_0(mask_sh)\
+	CS_SF(DP_DTO0_PHASE, DP_DTO0_PHASE, mask_sh),\
+	CS_SF(DP_DTO0_MODULO, DP_DTO0_MODULO, mask_sh),\
+	CS_SF(PHYPLLA_PIXCLK_RESYNC_CNTL, PHYPLLA_DCCG_DEEP_COLOR_CNTL, mask_sh),\
+	CS_SF(OTG0_PIXEL_RATE_CNTL, DP_DTO0_ENABLE, mask_sh)
+
+#define CS_COMMON_MASK_SH_LIST_DCN3_1_4(mask_sh)\
+	CS_COMMON_MASK_SH_LIST_DCN2_0(mask_sh),\
+	CS_SF(OTG0_PIXEL_RATE_CNTL, PIPE0_DTO_SRC_SEL, mask_sh),
+
+#define CS_COMMON_MASK_SH_LIST_DCN3_2(mask_sh)\
+	CS_COMMON_MASK_SH_LIST_DCN2_0(mask_sh),\
+	CS_SF(OTG0_PIXEL_RATE_CNTL, PIPE0_DTO_SRC_SEL, mask_sh)
+
+#define CS_COMMON_REG_LIST_DCN1_0(index, pllid) \
+		SRI(PIXCLK_RESYNC_CNTL, PHYPLL, pllid),\
+		SRII(PHASE, DP_DTO, 0),\
+		SRII(PHASE, DP_DTO, 1),\
+		SRII(PHASE, DP_DTO, 2),\
+		SRII(PHASE, DP_DTO, 3),\
+		SRII(MODULO, DP_DTO, 0),\
+		SRII(MODULO, DP_DTO, 1),\
+		SRII(MODULO, DP_DTO, 2),\
+		SRII(MODULO, DP_DTO, 3),\
+		SRII(PIXEL_RATE_CNTL, OTG, 0), \
+		SRII(PIXEL_RATE_CNTL, OTG, 1), \
+		SRII(PIXEL_RATE_CNTL, OTG, 2), \
+		SRII(PIXEL_RATE_CNTL, OTG, 3)
+
+#define CS_COMMON_MASK_SH_LIST_DCN1_0(mask_sh)\
+	CS_SF(DP_DTO0_PHASE, DP_DTO0_PHASE, mask_sh),\
+	CS_SF(DP_DTO0_MODULO, DP_DTO0_MODULO, mask_sh),\
+	CS_SF(PHYPLLA_PIXCLK_RESYNC_CNTL, PHYPLLA_DCCG_DEEP_COLOR_CNTL, mask_sh),\
+	CS_SF(OTG0_PIXEL_RATE_CNTL, DP_DTO0_ENABLE, mask_sh)
+
+
+#define CS_REG_FIELD_LIST(type) \
+	type PLL_REF_DIV_SRC; \
+	type DCCG_DEEP_COLOR_CNTL1; \
+	type PHYPLLA_DCCG_DEEP_COLOR_CNTL; \
+	type PHYPLLA_PIXCLK_DOUBLE_RATE_ENABLE; \
+	type PLL_POST_DIV_PIXCLK; \
+	type PLL_REF_DIV; \
+	type DP_DTO0_PHASE; \
+	type DP_DTO0_MODULO; \
+	type DP_DTO0_ENABLE;
+
+#define CS_REG_FIELD_LIST_DCN32(type) \
+	type PIPE0_DTO_SRC_SEL;
+
+struct dce110_clk_src_shift {
+	CS_REG_FIELD_LIST(uint8_t)
+	CS_REG_FIELD_LIST_DCN32(uint8_t)
+};
+
+struct dce110_clk_src_mask{
+	CS_REG_FIELD_LIST(uint32_t)
+	CS_REG_FIELD_LIST_DCN32(uint32_t)
+};
+
+struct dce110_clk_src_regs {
+	uint32_t RESYNC_CNTL;
+	uint32_t PIXCLK_RESYNC_CNTL;
+	uint32_t PLL_CNTL;
+
+	/* below are for DTO.
+	 * todo: should probably use different struct to not waste space
+	 */
+	uint32_t PHASE[MAX_PIPES];
+	uint32_t MODULO[MAX_PIPES];
+	uint32_t PIXEL_RATE_CNTL[MAX_PIPES];
+};
+
+struct dce110_clk_src {
+	struct clock_source base;
+	const struct dce110_clk_src_regs *regs;
+	const struct dce110_clk_src_mask *cs_mask;
+	const struct dce110_clk_src_shift *cs_shift;
+	struct dc_bios *bios;
+
+	struct spread_spectrum_data *dp_ss_params;
+	uint32_t dp_ss_params_cnt;
+	struct spread_spectrum_data *hdmi_ss_params;
+	uint32_t hdmi_ss_params_cnt;
+	struct spread_spectrum_data *dvi_ss_params;
+	uint32_t dvi_ss_params_cnt;
+	struct spread_spectrum_data *lvds_ss_params;
+	uint32_t lvds_ss_params_cnt;
+
+	uint32_t ext_clk_khz;
+	uint32_t ref_freq_khz;
+
+	struct calc_pll_clock_source calc_pll;
+	struct calc_pll_clock_source calc_pll_hdmi;
+};
+
+bool dce110_clk_src_construct(
+	struct dce110_clk_src *clk_src,
+	struct dc_context *ctx,
+	struct dc_bios *bios,
+	enum clock_source_id id,
+	const struct dce110_clk_src_regs *regs,
+	const struct dce110_clk_src_shift *cs_shift,
+	const struct dce110_clk_src_mask *cs_mask);
+
+bool dce112_clk_src_construct(
+	struct dce110_clk_src *clk_src,
+	struct dc_context *ctx,
+	struct dc_bios *bios,
+	enum clock_source_id id,
+	const struct dce110_clk_src_regs *regs,
+	const struct dce110_clk_src_shift *cs_shift,
+	const struct dce110_clk_src_mask *cs_mask);
+
+bool dcn20_clk_src_construct(
+	struct dce110_clk_src *clk_src,
+	struct dc_context *ctx,
+	struct dc_bios *bios,
+	enum clock_source_id id,
+	const struct dce110_clk_src_regs *regs,
+	const struct dce110_clk_src_shift *cs_shift,
+	const struct dce110_clk_src_mask *cs_mask);
+
+bool dcn3_clk_src_construct(
+	struct dce110_clk_src *clk_src,
+	struct dc_context *ctx,
+	struct dc_bios *bios,
+	enum clock_source_id id,
+	const struct dce110_clk_src_regs *regs,
+	const struct dce110_clk_src_shift *cs_shift,
+	const struct dce110_clk_src_mask *cs_mask);
+
+bool dcn301_clk_src_construct(
+	struct dce110_clk_src *clk_src,
+	struct dc_context *ctx,
+	struct dc_bios *bios,
+	enum clock_source_id id,
+	const struct dce110_clk_src_regs *regs,
+	const struct dce110_clk_src_shift *cs_shift,
+	const struct dce110_clk_src_mask *cs_mask);
+
+bool dcn31_clk_src_construct(
+	struct dce110_clk_src *clk_src,
+	struct dc_context *ctx,
+	struct dc_bios *bios,
+	enum clock_source_id id,
+	const struct dce110_clk_src_regs *regs,
+	const struct dce110_clk_src_shift *cs_shift,
+	const struct dce110_clk_src_mask *cs_mask);
+
+bool dcn401_clk_src_construct(
+	struct dce110_clk_src *clk_src,
+	struct dc_context *ctx,
+	struct dc_bios *bios,
+	enum clock_source_id id,
+	const struct dce110_clk_src_regs *regs,
+	const struct dce110_clk_src_shift *cs_shift,
+	const struct dce110_clk_src_mask *cs_mask);
+/* this table is use to find *1.001 and /1.001 pixel rates from non-precise pixel rate */
+struct pixel_rate_range_table_entry {
+	unsigned int range_min_khz;
+	unsigned int range_max_khz;
+	unsigned int target_pixel_rate_khz;
+	unsigned short mult_factor;
+	unsigned short div_factor;
+};
+
+extern const struct pixel_rate_range_table_entry video_optimized_pixel_rates[];
+const struct pixel_rate_range_table_entry *look_up_in_video_optimized_rate_tlb(
+		unsigned int pixel_rate_khz);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c b/drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c
new file mode 100644
index 000000000000..a8e79104b684
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c
@@ -0,0 +1,1192 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "core_types.h"
+#include "link_encoder.h"
+#include "dce_dmcu.h"
+#include "dm_services.h"
+#include "reg_helper.h"
+#include "fixed31_32.h"
+#include "dc.h"
+
+#define TO_DCE_DMCU(dmcu)\
+	container_of(dmcu, struct dce_dmcu, base)
+
+#define REG(reg) \
+	(dmcu_dce->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	dmcu_dce->dmcu_shift->field_name, dmcu_dce->dmcu_mask->field_name
+
+#define CTX \
+	dmcu_dce->base.ctx
+
+/* PSR related commands */
+#define PSR_ENABLE 0x20
+#define PSR_EXIT 0x21
+#define PSR_SET 0x23
+#define PSR_SET_WAITLOOP 0x31
+#define MCP_INIT_DMCU 0x88
+#define MCP_INIT_IRAM 0x89
+#define MCP_SYNC_PHY_LOCK 0x90
+#define MCP_SYNC_PHY_UNLOCK 0x91
+#define MCP_BL_SET_PWM_FRAC 0x6A  /* Enable or disable Fractional PWM */
+#define CRC_WIN_NOTIFY 0x92
+#define CRC_STOP_UPDATE 0x93
+#define MCP_SEND_EDID_CEA 0xA0
+#define EDID_CEA_CMD_ACK 1
+#define EDID_CEA_CMD_NACK 2
+#define MASTER_COMM_CNTL_REG__MASTER_COMM_INTERRUPT_MASK   0x00000001L
+
+// PSP FW version
+#define mmMP0_SMN_C2PMSG_58				0x1607A
+
+//Register access policy version
+#define mmMP0_SMN_C2PMSG_91				0x1609B
+
+static const uint32_t abm_gain_stepsize = 0x0060;
+
+static bool dce_dmcu_init(struct dmcu *dmcu)
+{
+	// Do nothing
+	return true;
+}
+
+static bool dce_dmcu_load_iram(struct dmcu *dmcu,
+			       unsigned int start_offset,
+			       const char *src,
+			       unsigned int bytes)
+{
+	struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(dmcu);
+	unsigned int count = 0;
+
+	/* Enable write access to IRAM */
+	REG_UPDATE_2(DMCU_RAM_ACCESS_CTRL,
+			IRAM_HOST_ACCESS_EN, 1,
+			IRAM_WR_ADDR_AUTO_INC, 1);
+
+	REG_WAIT(DCI_MEM_PWR_STATUS, DMCU_IRAM_MEM_PWR_STATE, 0, 2, 10);
+
+	REG_WRITE(DMCU_IRAM_WR_CTRL, start_offset);
+
+	for (count = 0; count < bytes; count++)
+		REG_WRITE(DMCU_IRAM_WR_DATA, src[count]);
+
+	/* Disable write access to IRAM to allow dynamic sleep state */
+	REG_UPDATE_2(DMCU_RAM_ACCESS_CTRL,
+			IRAM_HOST_ACCESS_EN, 0,
+			IRAM_WR_ADDR_AUTO_INC, 0);
+
+	return true;
+}
+
+static void dce_get_dmcu_psr_state(struct dmcu *dmcu, enum dc_psr_state *state)
+{
+	struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(dmcu);
+
+	uint32_t psr_state_offset = 0xf0;
+
+	/* Enable write access to IRAM */
+	REG_UPDATE(DMCU_RAM_ACCESS_CTRL, IRAM_HOST_ACCESS_EN, 1);
+
+	REG_WAIT(DCI_MEM_PWR_STATUS, DMCU_IRAM_MEM_PWR_STATE, 0, 2, 10);
+
+	/* Write address to IRAM_RD_ADDR in DMCU_IRAM_RD_CTRL */
+	REG_WRITE(DMCU_IRAM_RD_CTRL, psr_state_offset);
+
+	/* Read data from IRAM_RD_DATA in DMCU_IRAM_RD_DATA*/
+	*state = (enum dc_psr_state)REG_READ(DMCU_IRAM_RD_DATA);
+
+	/* Disable write access to IRAM after finished using IRAM
+	 * in order to allow dynamic sleep state
+	 */
+	REG_UPDATE(DMCU_RAM_ACCESS_CTRL, IRAM_HOST_ACCESS_EN, 0);
+}
+
+static void dce_dmcu_set_psr_enable(struct dmcu *dmcu, bool enable, bool wait)
+{
+	struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(dmcu);
+	unsigned int dmcu_max_retry_on_wait_reg_ready = 801;
+	unsigned int dmcu_wait_reg_ready_interval = 100;
+
+	unsigned int retryCount;
+	enum dc_psr_state state = PSR_STATE0;
+
+	/* waitDMCUReadyForCmd */
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0,
+				dmcu_wait_reg_ready_interval,
+				dmcu_max_retry_on_wait_reg_ready);
+
+	/* setDMCUParam_Cmd */
+	if (enable)
+		REG_UPDATE(MASTER_COMM_CMD_REG, MASTER_COMM_CMD_REG_BYTE0,
+				PSR_ENABLE);
+	else
+		REG_UPDATE(MASTER_COMM_CMD_REG, MASTER_COMM_CMD_REG_BYTE0,
+				PSR_EXIT);
+
+	/* notifyDMCUMsg */
+	REG_UPDATE(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 1);
+	if (wait == true) {
+		for (retryCount = 0; retryCount <= 100; retryCount++) {
+			dce_get_dmcu_psr_state(dmcu, &state);
+			if (enable) {
+				if (state != PSR_STATE0)
+					break;
+			} else {
+				if (state == PSR_STATE0)
+					break;
+			}
+			udelay(10);
+		}
+	}
+}
+
+static bool dce_dmcu_setup_psr(struct dmcu *dmcu,
+		struct dc_link *link,
+		struct psr_context *psr_context)
+{
+	struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(dmcu);
+
+	unsigned int dmcu_max_retry_on_wait_reg_ready = 801;
+	unsigned int dmcu_wait_reg_ready_interval = 100;
+
+	union dce_dmcu_psr_config_data_reg1 masterCmdData1;
+	union dce_dmcu_psr_config_data_reg2 masterCmdData2;
+	union dce_dmcu_psr_config_data_reg3 masterCmdData3;
+
+	link->link_enc->funcs->psr_program_dp_dphy_fast_training(link->link_enc,
+			psr_context->psrExitLinkTrainingRequired);
+
+	/* Enable static screen interrupts for PSR supported display */
+	/* Disable the interrupt coming from other displays. */
+	REG_UPDATE_4(DMCU_INTERRUPT_TO_UC_EN_MASK,
+			STATIC_SCREEN1_INT_TO_UC_EN, 0,
+			STATIC_SCREEN2_INT_TO_UC_EN, 0,
+			STATIC_SCREEN3_INT_TO_UC_EN, 0,
+			STATIC_SCREEN4_INT_TO_UC_EN, 0);
+
+	switch (psr_context->controllerId) {
+	/* Driver uses case 1 for unconfigured */
+	case 1:
+		REG_UPDATE(DMCU_INTERRUPT_TO_UC_EN_MASK,
+				STATIC_SCREEN1_INT_TO_UC_EN, 1);
+		break;
+	case 2:
+		REG_UPDATE(DMCU_INTERRUPT_TO_UC_EN_MASK,
+				STATIC_SCREEN2_INT_TO_UC_EN, 1);
+		break;
+	case 3:
+		REG_UPDATE(DMCU_INTERRUPT_TO_UC_EN_MASK,
+				STATIC_SCREEN3_INT_TO_UC_EN, 1);
+		break;
+	case 4:
+		REG_UPDATE(DMCU_INTERRUPT_TO_UC_EN_MASK,
+				STATIC_SCREEN4_INT_TO_UC_EN, 1);
+		break;
+	case 5:
+		/* CZ/NL only has 4 CRTC!!
+		 * really valid.
+		 * There is no interrupt enable mask for these instances.
+		 */
+		break;
+	case 6:
+		/* CZ/NL only has 4 CRTC!!
+		 * These are here because they are defined in HW regspec,
+		 * but not really valid. There is no interrupt enable mask
+		 * for these instances.
+		 */
+		break;
+	default:
+		REG_UPDATE(DMCU_INTERRUPT_TO_UC_EN_MASK,
+				STATIC_SCREEN1_INT_TO_UC_EN, 1);
+		break;
+	}
+
+	link->link_enc->funcs->psr_program_secondary_packet(link->link_enc,
+			psr_context->sdpTransmitLineNumDeadline);
+
+	/* waitDMCUReadyForCmd */
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0,
+					dmcu_wait_reg_ready_interval,
+					dmcu_max_retry_on_wait_reg_ready);
+
+	/* setDMCUParam_PSRHostConfigData */
+	masterCmdData1.u32All = 0;
+	masterCmdData1.bits.timehyst_frames = psr_context->timehyst_frames;
+	masterCmdData1.bits.hyst_lines = psr_context->hyst_lines;
+	masterCmdData1.bits.rfb_update_auto_en =
+			psr_context->rfb_update_auto_en;
+	masterCmdData1.bits.dp_port_num = psr_context->transmitterId;
+	masterCmdData1.bits.dcp_sel = psr_context->controllerId;
+	masterCmdData1.bits.phy_type  = psr_context->phyType;
+	masterCmdData1.bits.frame_cap_ind =
+			psr_context->psrFrameCaptureIndicationReq;
+	masterCmdData1.bits.aux_chan = psr_context->channel;
+	masterCmdData1.bits.aux_repeat = psr_context->aux_repeats;
+	dm_write_reg(dmcu->ctx, REG(MASTER_COMM_DATA_REG1),
+					masterCmdData1.u32All);
+
+	masterCmdData2.u32All = 0;
+	masterCmdData2.bits.dig_fe = psr_context->engineId;
+	masterCmdData2.bits.dig_be = psr_context->transmitterId;
+	masterCmdData2.bits.skip_wait_for_pll_lock =
+			psr_context->skipPsrWaitForPllLock;
+	masterCmdData2.bits.frame_delay = psr_context->frame_delay;
+	masterCmdData2.bits.smu_phy_id = psr_context->smuPhyId;
+	masterCmdData2.bits.num_of_controllers =
+			psr_context->numberOfControllers;
+	dm_write_reg(dmcu->ctx, REG(MASTER_COMM_DATA_REG2),
+			masterCmdData2.u32All);
+
+	masterCmdData3.u32All = 0;
+	masterCmdData3.bits.psr_level = psr_context->psr_level.u32all;
+	dm_write_reg(dmcu->ctx, REG(MASTER_COMM_DATA_REG3),
+			masterCmdData3.u32All);
+
+	/* setDMCUParam_Cmd */
+	REG_UPDATE(MASTER_COMM_CMD_REG,
+			MASTER_COMM_CMD_REG_BYTE0, PSR_SET);
+
+	/* notifyDMCUMsg */
+	REG_UPDATE(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 1);
+
+	return true;
+}
+
+static bool dce_is_dmcu_initialized(struct dmcu *dmcu)
+{
+	struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(dmcu);
+	unsigned int dmcu_uc_reset;
+
+	/* microcontroller is not running */
+	REG_GET(DMCU_STATUS, UC_IN_RESET, &dmcu_uc_reset);
+
+	/* DMCU is not running */
+	if (dmcu_uc_reset)
+		return false;
+
+	return true;
+}
+
+static void dce_psr_wait_loop(
+	struct dmcu *dmcu,
+	unsigned int wait_loop_number)
+{
+	struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(dmcu);
+	union dce_dmcu_psr_config_data_wait_loop_reg1 masterCmdData1;
+
+	if (dmcu->cached_wait_loop_number == wait_loop_number)
+		return;
+
+	/* DMCU is not running */
+	if (!dce_is_dmcu_initialized(dmcu))
+		return;
+
+	/* waitDMCUReadyForCmd */
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0, 1, 10000);
+
+	masterCmdData1.u32 = 0;
+	masterCmdData1.bits.wait_loop = wait_loop_number;
+	dmcu->cached_wait_loop_number = wait_loop_number;
+	dm_write_reg(dmcu->ctx, REG(MASTER_COMM_DATA_REG1), masterCmdData1.u32);
+
+	/* setDMCUParam_Cmd */
+	REG_UPDATE(MASTER_COMM_CMD_REG, MASTER_COMM_CMD_REG_BYTE0, PSR_SET_WAITLOOP);
+
+	/* notifyDMCUMsg */
+	REG_UPDATE(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 1);
+}
+
+static void dce_get_psr_wait_loop(
+		struct dmcu *dmcu, unsigned int *psr_wait_loop_number)
+{
+	*psr_wait_loop_number = dmcu->cached_wait_loop_number;
+	return;
+}
+
+static void dcn10_get_dmcu_version(struct dmcu *dmcu)
+{
+	struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(dmcu);
+	uint32_t dmcu_version_offset = 0xf1;
+
+	/* Enable write access to IRAM */
+	REG_UPDATE_2(DMCU_RAM_ACCESS_CTRL,
+			IRAM_HOST_ACCESS_EN, 1,
+			IRAM_RD_ADDR_AUTO_INC, 1);
+
+	REG_WAIT(DMU_MEM_PWR_CNTL, DMCU_IRAM_MEM_PWR_STATE, 0, 2, 10);
+
+	/* Write address to IRAM_RD_ADDR and read from DATA register */
+	REG_WRITE(DMCU_IRAM_RD_CTRL, dmcu_version_offset);
+	dmcu->dmcu_version.interface_version = REG_READ(DMCU_IRAM_RD_DATA);
+	dmcu->dmcu_version.abm_version = REG_READ(DMCU_IRAM_RD_DATA);
+	dmcu->dmcu_version.psr_version = REG_READ(DMCU_IRAM_RD_DATA);
+	dmcu->dmcu_version.build_version = ((REG_READ(DMCU_IRAM_RD_DATA) << 8) |
+						REG_READ(DMCU_IRAM_RD_DATA));
+
+	/* Disable write access to IRAM to allow dynamic sleep state */
+	REG_UPDATE_2(DMCU_RAM_ACCESS_CTRL,
+			IRAM_HOST_ACCESS_EN, 0,
+			IRAM_RD_ADDR_AUTO_INC, 0);
+}
+
+static void dcn10_dmcu_enable_fractional_pwm(struct dmcu *dmcu,
+		uint32_t fractional_pwm)
+{
+	struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(dmcu);
+
+	/* Wait until microcontroller is ready to process interrupt */
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0, 100, 800);
+
+	/* Set PWM fractional enable/disable */
+	REG_WRITE(MASTER_COMM_DATA_REG1, fractional_pwm);
+
+	/* Set command to enable or disable fractional PWM microcontroller */
+	REG_UPDATE(MASTER_COMM_CMD_REG, MASTER_COMM_CMD_REG_BYTE0,
+			MCP_BL_SET_PWM_FRAC);
+
+	/* Notify microcontroller of new command */
+	REG_UPDATE(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 1);
+
+	/* Ensure command has been executed before continuing */
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0, 100, 800);
+}
+
+static bool dcn10_dmcu_init(struct dmcu *dmcu)
+{
+	struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(dmcu);
+	const struct dc_config *config = &dmcu->ctx->dc->config;
+	bool status = false;
+	struct dc_context *ctx = dmcu->ctx;
+	unsigned int i;
+	//  5 4 3 2 1 0
+	//  F E D C B A - bit 0 is A, bit 5 is F
+	unsigned int tx_interrupt_mask = 0;
+
+	PERF_TRACE();
+	/*  Definition of DC_DMCU_SCRATCH
+	 *  0 : firmare not loaded
+	 *  1 : PSP load DMCU FW but not initialized
+	 *  2 : Firmware already initialized
+	 */
+	dmcu->dmcu_state = REG_READ(DC_DMCU_SCRATCH);
+
+	for (i = 0; i < ctx->dc->link_count; i++) {
+		if (ctx->dc->links[i]->link_enc->features.flags.bits.DP_IS_USB_C) {
+			if (ctx->dc->links[i]->link_enc->transmitter >= TRANSMITTER_UNIPHY_A &&
+					ctx->dc->links[i]->link_enc->transmitter <= TRANSMITTER_UNIPHY_F) {
+				tx_interrupt_mask |= 1 << ctx->dc->links[i]->link_enc->transmitter;
+			}
+		}
+	}
+
+	switch (dmcu->dmcu_state) {
+	case DMCU_UNLOADED:
+		status = false;
+		break;
+	case DMCU_LOADED_UNINITIALIZED:
+		/* Wait until microcontroller is ready to process interrupt */
+		REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0, 100, 800);
+
+		/* Set initialized ramping boundary value */
+		REG_WRITE(MASTER_COMM_DATA_REG1, 0xFFFF);
+
+		/* Set backlight ramping stepsize */
+		REG_WRITE(MASTER_COMM_DATA_REG2, abm_gain_stepsize);
+
+		REG_WRITE(MASTER_COMM_DATA_REG3, tx_interrupt_mask);
+
+		/* Set command to initialize microcontroller */
+		REG_UPDATE(MASTER_COMM_CMD_REG, MASTER_COMM_CMD_REG_BYTE0,
+			MCP_INIT_DMCU);
+
+		/* Notify microcontroller of new command */
+		REG_UPDATE(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 1);
+
+		/* Ensure command has been executed before continuing */
+		REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0, 100, 800);
+
+		// Check state is initialized
+		dmcu->dmcu_state = REG_READ(DC_DMCU_SCRATCH);
+
+		// If microcontroller is not in running state, fail
+		if (dmcu->dmcu_state == DMCU_RUNNING) {
+			/* Retrieve and cache the DMCU firmware version. */
+			dcn10_get_dmcu_version(dmcu);
+
+			/* Initialize DMCU to use fractional PWM or not */
+			dcn10_dmcu_enable_fractional_pwm(dmcu,
+				(config->disable_fractional_pwm == false) ? 1 : 0);
+			status = true;
+		} else {
+			status = false;
+		}
+
+		break;
+	case DMCU_RUNNING:
+		status = true;
+		break;
+	default:
+		status = false;
+		break;
+	}
+
+	PERF_TRACE();
+	return status;
+}
+
+static bool dcn21_dmcu_init(struct dmcu *dmcu)
+{
+	struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(dmcu);
+	uint32_t dmcub_psp_version = REG_READ(DMCUB_SCRATCH15);
+
+	if (dmcu->auto_load_dmcu && dmcub_psp_version == 0) {
+		return false;
+	}
+
+	return dcn10_dmcu_init(dmcu);
+}
+
+static bool dcn10_dmcu_load_iram(struct dmcu *dmcu,
+		unsigned int start_offset,
+		const char *src,
+		unsigned int bytes)
+{
+	struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(dmcu);
+	unsigned int count = 0;
+
+	/* If microcontroller is not running, do nothing */
+	if (dmcu->dmcu_state != DMCU_RUNNING)
+		return false;
+
+	/* Enable write access to IRAM */
+	REG_UPDATE_2(DMCU_RAM_ACCESS_CTRL,
+			IRAM_HOST_ACCESS_EN, 1,
+			IRAM_WR_ADDR_AUTO_INC, 1);
+
+	REG_WAIT(DMU_MEM_PWR_CNTL, DMCU_IRAM_MEM_PWR_STATE, 0, 2, 10);
+
+	REG_WRITE(DMCU_IRAM_WR_CTRL, start_offset);
+
+	for (count = 0; count < bytes; count++)
+		REG_WRITE(DMCU_IRAM_WR_DATA, src[count]);
+
+	/* Disable write access to IRAM to allow dynamic sleep state */
+	REG_UPDATE_2(DMCU_RAM_ACCESS_CTRL,
+			IRAM_HOST_ACCESS_EN, 0,
+			IRAM_WR_ADDR_AUTO_INC, 0);
+
+	/* Wait until microcontroller is ready to process interrupt */
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0, 100, 800);
+
+	/* Set command to signal IRAM is loaded and to initialize IRAM */
+	REG_UPDATE(MASTER_COMM_CMD_REG, MASTER_COMM_CMD_REG_BYTE0,
+			MCP_INIT_IRAM);
+
+	/* Notify microcontroller of new command */
+	REG_UPDATE(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 1);
+
+	/* Ensure command has been executed before continuing */
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0, 100, 800);
+
+	return true;
+}
+
+static void dcn10_get_dmcu_psr_state(struct dmcu *dmcu, enum dc_psr_state *state)
+{
+	struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(dmcu);
+
+	uint32_t psr_state_offset = 0xf0;
+
+	/* If microcontroller is not running, do nothing */
+	if (dmcu->dmcu_state != DMCU_RUNNING)
+		return;
+
+	/* Enable write access to IRAM */
+	REG_UPDATE(DMCU_RAM_ACCESS_CTRL, IRAM_HOST_ACCESS_EN, 1);
+
+	REG_WAIT(DMU_MEM_PWR_CNTL, DMCU_IRAM_MEM_PWR_STATE, 0, 2, 10);
+
+	/* Write address to IRAM_RD_ADDR in DMCU_IRAM_RD_CTRL */
+	REG_WRITE(DMCU_IRAM_RD_CTRL, psr_state_offset);
+
+	/* Read data from IRAM_RD_DATA in DMCU_IRAM_RD_DATA*/
+	*state = (enum dc_psr_state)REG_READ(DMCU_IRAM_RD_DATA);
+
+	/* Disable write access to IRAM after finished using IRAM
+	 * in order to allow dynamic sleep state
+	 */
+	REG_UPDATE(DMCU_RAM_ACCESS_CTRL, IRAM_HOST_ACCESS_EN, 0);
+}
+
+static void dcn10_dmcu_set_psr_enable(struct dmcu *dmcu, bool enable, bool wait)
+{
+	struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(dmcu);
+	unsigned int dmcu_max_retry_on_wait_reg_ready = 801;
+	unsigned int dmcu_wait_reg_ready_interval = 100;
+
+	unsigned int retryCount;
+	enum dc_psr_state state = PSR_STATE0;
+
+	/* If microcontroller is not running, do nothing */
+	if (dmcu->dmcu_state != DMCU_RUNNING)
+		return;
+
+	/* waitDMCUReadyForCmd */
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0,
+				dmcu_wait_reg_ready_interval,
+				dmcu_max_retry_on_wait_reg_ready);
+
+	/* setDMCUParam_Cmd */
+	if (enable)
+		REG_UPDATE(MASTER_COMM_CMD_REG, MASTER_COMM_CMD_REG_BYTE0,
+				PSR_ENABLE);
+	else
+		REG_UPDATE(MASTER_COMM_CMD_REG, MASTER_COMM_CMD_REG_BYTE0,
+				PSR_EXIT);
+
+	/* notifyDMCUMsg */
+	REG_UPDATE(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 1);
+
+	/* Below loops 1000 x 500us = 500 ms.
+	 *  Exit PSR may need to wait 1-2 frames to power up. Timeout after at
+	 *  least a few frames. Should never hit the max retry assert below.
+	 */
+	if (wait == true) {
+		for (retryCount = 0; retryCount <= 1000; retryCount++) {
+			dcn10_get_dmcu_psr_state(dmcu, &state);
+			if (enable) {
+				if (state != PSR_STATE0)
+					break;
+			} else {
+				if (state == PSR_STATE0)
+					break;
+			}
+			/* must *not* be fsleep - this can be called from high irq levels */
+			udelay(500);
+		}
+
+		/* assert if max retry hit */
+		if (retryCount >= 1000)
+			ASSERT(0);
+	}
+}
+
+static bool dcn10_dmcu_setup_psr(struct dmcu *dmcu,
+		struct dc_link *link,
+		struct psr_context *psr_context)
+{
+	struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(dmcu);
+
+	unsigned int dmcu_max_retry_on_wait_reg_ready = 801;
+	unsigned int dmcu_wait_reg_ready_interval = 100;
+
+	union dce_dmcu_psr_config_data_reg1 masterCmdData1;
+	union dce_dmcu_psr_config_data_reg2 masterCmdData2;
+	union dce_dmcu_psr_config_data_reg3 masterCmdData3;
+
+	/* If microcontroller is not running, do nothing */
+	if (dmcu->dmcu_state != DMCU_RUNNING)
+		return false;
+
+	link->link_enc->funcs->psr_program_dp_dphy_fast_training(link->link_enc,
+			psr_context->psrExitLinkTrainingRequired);
+
+	/* Enable static screen interrupts for PSR supported display */
+	/* Disable the interrupt coming from other displays. */
+	REG_UPDATE_4(DMCU_INTERRUPT_TO_UC_EN_MASK,
+			STATIC_SCREEN1_INT_TO_UC_EN, 0,
+			STATIC_SCREEN2_INT_TO_UC_EN, 0,
+			STATIC_SCREEN3_INT_TO_UC_EN, 0,
+			STATIC_SCREEN4_INT_TO_UC_EN, 0);
+
+	switch (psr_context->controllerId) {
+	/* Driver uses case 1 for unconfigured */
+	case 1:
+		REG_UPDATE(DMCU_INTERRUPT_TO_UC_EN_MASK,
+				STATIC_SCREEN1_INT_TO_UC_EN, 1);
+		break;
+	case 2:
+		REG_UPDATE(DMCU_INTERRUPT_TO_UC_EN_MASK,
+				STATIC_SCREEN2_INT_TO_UC_EN, 1);
+		break;
+	case 3:
+		REG_UPDATE(DMCU_INTERRUPT_TO_UC_EN_MASK,
+				STATIC_SCREEN3_INT_TO_UC_EN, 1);
+		break;
+	case 4:
+		REG_UPDATE(DMCU_INTERRUPT_TO_UC_EN_MASK,
+				STATIC_SCREEN4_INT_TO_UC_EN, 1);
+		break;
+	case 5:
+		/* CZ/NL only has 4 CRTC!!
+		 * really valid.
+		 * There is no interrupt enable mask for these instances.
+		 */
+		break;
+	case 6:
+		/* CZ/NL only has 4 CRTC!!
+		 * These are here because they are defined in HW regspec,
+		 * but not really valid. There is no interrupt enable mask
+		 * for these instances.
+		 */
+		break;
+	default:
+		REG_UPDATE(DMCU_INTERRUPT_TO_UC_EN_MASK,
+				STATIC_SCREEN1_INT_TO_UC_EN, 1);
+		break;
+	}
+
+	link->link_enc->funcs->psr_program_secondary_packet(link->link_enc,
+			psr_context->sdpTransmitLineNumDeadline);
+
+	if (psr_context->allow_smu_optimizations)
+		REG_UPDATE(SMU_INTERRUPT_CONTROL, DC_SMU_INT_ENABLE, 1);
+
+	/* waitDMCUReadyForCmd */
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0,
+			dmcu_wait_reg_ready_interval,
+			dmcu_max_retry_on_wait_reg_ready);
+
+	/* setDMCUParam_PSRHostConfigData */
+	masterCmdData1.u32All = 0;
+	masterCmdData1.bits.timehyst_frames = psr_context->timehyst_frames;
+	masterCmdData1.bits.hyst_lines = psr_context->hyst_lines;
+	masterCmdData1.bits.rfb_update_auto_en =
+			psr_context->rfb_update_auto_en;
+	masterCmdData1.bits.dp_port_num = psr_context->transmitterId;
+	masterCmdData1.bits.dcp_sel = psr_context->controllerId;
+	masterCmdData1.bits.phy_type  = psr_context->phyType;
+	masterCmdData1.bits.frame_cap_ind =
+			psr_context->psrFrameCaptureIndicationReq;
+	masterCmdData1.bits.aux_chan = psr_context->channel;
+	masterCmdData1.bits.aux_repeat = psr_context->aux_repeats;
+	masterCmdData1.bits.allow_smu_optimizations = psr_context->allow_smu_optimizations;
+	dm_write_reg(dmcu->ctx, REG(MASTER_COMM_DATA_REG1),
+					masterCmdData1.u32All);
+
+	masterCmdData2.u32All = 0;
+	masterCmdData2.bits.dig_fe = psr_context->engineId;
+	masterCmdData2.bits.dig_be = psr_context->transmitterId;
+	masterCmdData2.bits.skip_wait_for_pll_lock =
+			psr_context->skipPsrWaitForPllLock;
+	masterCmdData2.bits.frame_delay = psr_context->frame_delay;
+	masterCmdData2.bits.smu_phy_id = psr_context->smuPhyId;
+	masterCmdData2.bits.num_of_controllers =
+			psr_context->numberOfControllers;
+	dm_write_reg(dmcu->ctx, REG(MASTER_COMM_DATA_REG2),
+			masterCmdData2.u32All);
+
+	masterCmdData3.u32All = 0;
+	masterCmdData3.bits.psr_level = psr_context->psr_level.u32all;
+	dm_write_reg(dmcu->ctx, REG(MASTER_COMM_DATA_REG3),
+			masterCmdData3.u32All);
+
+
+	/* setDMCUParam_Cmd */
+	REG_UPDATE(MASTER_COMM_CMD_REG,
+			MASTER_COMM_CMD_REG_BYTE0, PSR_SET);
+
+	/* notifyDMCUMsg */
+	REG_UPDATE(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 1);
+
+	/* waitDMCUReadyForCmd */
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0, 1, 10000);
+
+	return true;
+}
+
+static void dcn10_psr_wait_loop(
+	struct dmcu *dmcu,
+	unsigned int wait_loop_number)
+{
+	struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(dmcu);
+	union dce_dmcu_psr_config_data_wait_loop_reg1 masterCmdData1;
+
+	/* If microcontroller is not running, do nothing */
+	if (dmcu->dmcu_state != DMCU_RUNNING)
+		return;
+
+	if (wait_loop_number != 0) {
+	/* waitDMCUReadyForCmd */
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0, 1, 10000);
+
+	masterCmdData1.u32 = 0;
+	masterCmdData1.bits.wait_loop = wait_loop_number;
+	dmcu->cached_wait_loop_number = wait_loop_number;
+	dm_write_reg(dmcu->ctx, REG(MASTER_COMM_DATA_REG1), masterCmdData1.u32);
+
+	/* setDMCUParam_Cmd */
+	REG_UPDATE(MASTER_COMM_CMD_REG, MASTER_COMM_CMD_REG_BYTE0, PSR_SET_WAITLOOP);
+
+	/* notifyDMCUMsg */
+	REG_UPDATE(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 1);
+	}
+}
+
+static void dcn10_get_psr_wait_loop(
+		struct dmcu *dmcu, unsigned int *psr_wait_loop_number)
+{
+	*psr_wait_loop_number = dmcu->cached_wait_loop_number;
+	return;
+}
+
+static bool dcn10_is_dmcu_initialized(struct dmcu *dmcu)
+{
+	/* microcontroller is not running */
+	if (dmcu->dmcu_state != DMCU_RUNNING)
+		return false;
+	return true;
+}
+
+
+
+static bool dcn20_lock_phy(struct dmcu *dmcu)
+{
+	struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(dmcu);
+
+	/* If microcontroller is not running, do nothing */
+	if (dmcu->dmcu_state != DMCU_RUNNING)
+		return false;
+
+	/* waitDMCUReadyForCmd */
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0, 1, 10000);
+
+	/* setDMCUParam_Cmd */
+	REG_UPDATE(MASTER_COMM_CMD_REG, MASTER_COMM_CMD_REG_BYTE0, MCP_SYNC_PHY_LOCK);
+
+	/* notifyDMCUMsg */
+	REG_UPDATE(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 1);
+
+	/* waitDMCUReadyForCmd */
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0, 1, 10000);
+
+	return true;
+}
+
+static bool dcn20_unlock_phy(struct dmcu *dmcu)
+{
+	struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(dmcu);
+
+	/* If microcontroller is not running, do nothing */
+	if (dmcu->dmcu_state != DMCU_RUNNING)
+		return false;
+
+	/* waitDMCUReadyForCmd */
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0, 1, 10000);
+
+	/* setDMCUParam_Cmd */
+	REG_UPDATE(MASTER_COMM_CMD_REG, MASTER_COMM_CMD_REG_BYTE0, MCP_SYNC_PHY_UNLOCK);
+
+	/* notifyDMCUMsg */
+	REG_UPDATE(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 1);
+
+	/* waitDMCUReadyForCmd */
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0, 1, 10000);
+
+	return true;
+}
+
+static bool dcn10_send_edid_cea(struct dmcu *dmcu,
+		int offset,
+		int total_length,
+		uint8_t *data,
+		int length)
+{
+	struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(dmcu);
+	uint32_t header, data1, data2;
+
+	/* If microcontroller is not running, do nothing */
+	if (dmcu->dmcu_state != DMCU_RUNNING)
+		return false;
+
+	if (length > 8 || length <= 0)
+		return false;
+
+	header = ((uint32_t)offset & 0xFFFF) << 16 | (total_length & 0xFFFF);
+	data1 = (((uint32_t)data[0]) << 24) | (((uint32_t)data[1]) << 16) |
+		(((uint32_t)data[2]) << 8) | ((uint32_t)data[3]);
+	data2 = (((uint32_t)data[4]) << 24) | (((uint32_t)data[5]) << 16) |
+		(((uint32_t)data[6]) << 8) | ((uint32_t)data[7]);
+
+	/* waitDMCUReadyForCmd */
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0, 1, 10000);
+
+	/* setDMCUParam_Cmd */
+	REG_UPDATE(MASTER_COMM_CMD_REG, MASTER_COMM_CMD_REG_BYTE0, MCP_SEND_EDID_CEA);
+
+	REG_WRITE(MASTER_COMM_DATA_REG1, header);
+	REG_WRITE(MASTER_COMM_DATA_REG2, data1);
+	REG_WRITE(MASTER_COMM_DATA_REG3, data2);
+
+	/* notifyDMCUMsg */
+	REG_UPDATE(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 1);
+
+	/* waitDMCUReadyForCmd */
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0, 1, 10000);
+
+	return true;
+}
+
+static bool dcn10_get_scp_results(struct dmcu *dmcu,
+		uint32_t *cmd,
+		uint32_t *data1,
+		uint32_t *data2,
+		uint32_t *data3)
+{
+	struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(dmcu);
+
+	/* If microcontroller is not running, do nothing */
+	if (dmcu->dmcu_state != DMCU_RUNNING)
+		return false;
+
+	*cmd = REG_READ(SLAVE_COMM_CMD_REG);
+	*data1 =  REG_READ(SLAVE_COMM_DATA_REG1);
+	*data2 =  REG_READ(SLAVE_COMM_DATA_REG2);
+	*data3 =  REG_READ(SLAVE_COMM_DATA_REG3);
+
+	/* clear SCP interrupt */
+	REG_UPDATE(SLAVE_COMM_CNTL_REG, SLAVE_COMM_INTERRUPT, 0);
+
+	return true;
+}
+
+static bool dcn10_recv_amd_vsdb(struct dmcu *dmcu,
+		int *version,
+		int *min_frame_rate,
+		int *max_frame_rate)
+{
+	uint32_t data[4];
+	int cmd, ack, len;
+
+	if (!dcn10_get_scp_results(dmcu, &data[0], &data[1], &data[2], &data[3]))
+		return false;
+
+	cmd = data[0] & 0x3FF;
+	len = (data[0] >> 10) & 0x3F;
+	ack = data[1];
+
+	if (cmd != MCP_SEND_EDID_CEA || ack != EDID_CEA_CMD_ACK || len != 12)
+		return false;
+
+	if ((data[2] & 0xFF)) {
+		*version = (data[2] >> 8) & 0xFF;
+		*min_frame_rate = (data[3] >> 16) & 0xFFFF;
+		*max_frame_rate = data[3] & 0xFFFF;
+		return true;
+	}
+
+	return false;
+}
+
+static bool dcn10_recv_edid_cea_ack(struct dmcu *dmcu, int *offset)
+{
+	uint32_t data[4];
+	int cmd, ack;
+
+	if (!dcn10_get_scp_results(dmcu,
+				&data[0], &data[1], &data[2], &data[3]))
+		return false;
+
+	cmd = data[0] & 0x3FF;
+	ack = data[1];
+
+	if (cmd != MCP_SEND_EDID_CEA)
+		return false;
+
+	if (ack == EDID_CEA_CMD_ACK)
+		return true;
+
+	*offset = data[2]; /* nack */
+	return false;
+}
+
+
+#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY)
+static void dcn10_forward_crc_window(struct dmcu *dmcu,
+					struct rect *rect,
+					struct otg_phy_mux *mux_mapping)
+{
+	struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(dmcu);
+	unsigned int dmcu_max_retry_on_wait_reg_ready = 801;
+	unsigned int dmcu_wait_reg_ready_interval = 100;
+	unsigned int crc_start = 0, crc_end = 0, otg_phy_mux = 0;
+	int x_start, y_start, x_end, y_end;
+
+	/* If microcontroller is not running, do nothing */
+	if (dmcu->dmcu_state != DMCU_RUNNING)
+		return;
+
+	if (!rect)
+		return;
+
+	/* waitDMCUReadyForCmd */
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0,
+				dmcu_wait_reg_ready_interval,
+				dmcu_max_retry_on_wait_reg_ready);
+
+	x_start = rect->x;
+	y_start = rect->y;
+	x_end = x_start + rect->width;
+	y_end = y_start + rect->height;
+
+	/* build up nitification data */
+	crc_start = (((unsigned int) x_start) << 16) | y_start;
+	crc_end = (((unsigned int) x_end) << 16) | y_end;
+	otg_phy_mux =
+		(((unsigned int) mux_mapping->otg_output_num) << 16) | mux_mapping->phy_output_num;
+
+	dm_write_reg(dmcu->ctx, REG(MASTER_COMM_DATA_REG1),
+					crc_start);
+
+	dm_write_reg(dmcu->ctx, REG(MASTER_COMM_DATA_REG2),
+			crc_end);
+
+	dm_write_reg(dmcu->ctx, REG(MASTER_COMM_DATA_REG3),
+			otg_phy_mux);
+
+	/* setDMCUParam_Cmd */
+	REG_UPDATE(MASTER_COMM_CMD_REG, MASTER_COMM_CMD_REG_BYTE0,
+				CRC_WIN_NOTIFY);
+
+	/* notifyDMCUMsg */
+	REG_UPDATE(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 1);
+}
+
+static void dcn10_stop_crc_win_update(struct dmcu *dmcu,
+					struct otg_phy_mux *mux_mapping)
+{
+	struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(dmcu);
+	unsigned int dmcu_max_retry_on_wait_reg_ready = 801;
+	unsigned int dmcu_wait_reg_ready_interval = 100;
+	unsigned int otg_phy_mux = 0;
+
+	/* If microcontroller is not running, do nothing */
+	if (dmcu->dmcu_state != DMCU_RUNNING)
+		return;
+
+	/* waitDMCUReadyForCmd */
+	REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0,
+				dmcu_wait_reg_ready_interval,
+				dmcu_max_retry_on_wait_reg_ready);
+
+	/* build up nitification data */
+	otg_phy_mux =
+		(((unsigned int) mux_mapping->otg_output_num) << 16) | mux_mapping->phy_output_num;
+
+	dm_write_reg(dmcu->ctx, REG(MASTER_COMM_DATA_REG1),
+					otg_phy_mux);
+
+	/* setDMCUParam_Cmd */
+	REG_UPDATE(MASTER_COMM_CMD_REG, MASTER_COMM_CMD_REG_BYTE0,
+				CRC_STOP_UPDATE);
+
+	/* notifyDMCUMsg */
+	REG_UPDATE(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 1);
+}
+#endif
+
+static const struct dmcu_funcs dce_funcs = {
+	.dmcu_init = dce_dmcu_init,
+	.load_iram = dce_dmcu_load_iram,
+	.set_psr_enable = dce_dmcu_set_psr_enable,
+	.setup_psr = dce_dmcu_setup_psr,
+	.get_psr_state = dce_get_dmcu_psr_state,
+	.set_psr_wait_loop = dce_psr_wait_loop,
+	.get_psr_wait_loop = dce_get_psr_wait_loop,
+	.is_dmcu_initialized = dce_is_dmcu_initialized
+};
+
+static const struct dmcu_funcs dcn10_funcs = {
+	.dmcu_init = dcn10_dmcu_init,
+	.load_iram = dcn10_dmcu_load_iram,
+	.set_psr_enable = dcn10_dmcu_set_psr_enable,
+	.setup_psr = dcn10_dmcu_setup_psr,
+	.get_psr_state = dcn10_get_dmcu_psr_state,
+	.set_psr_wait_loop = dcn10_psr_wait_loop,
+	.get_psr_wait_loop = dcn10_get_psr_wait_loop,
+	.send_edid_cea = dcn10_send_edid_cea,
+	.recv_amd_vsdb = dcn10_recv_amd_vsdb,
+	.recv_edid_cea_ack = dcn10_recv_edid_cea_ack,
+#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY)
+	.forward_crc_window = dcn10_forward_crc_window,
+	.stop_crc_win_update = dcn10_stop_crc_win_update,
+#endif
+	.is_dmcu_initialized = dcn10_is_dmcu_initialized
+};
+
+static const struct dmcu_funcs dcn20_funcs = {
+	.dmcu_init = dcn10_dmcu_init,
+	.load_iram = dcn10_dmcu_load_iram,
+	.set_psr_enable = dcn10_dmcu_set_psr_enable,
+	.setup_psr = dcn10_dmcu_setup_psr,
+	.get_psr_state = dcn10_get_dmcu_psr_state,
+	.set_psr_wait_loop = dcn10_psr_wait_loop,
+	.get_psr_wait_loop = dcn10_get_psr_wait_loop,
+	.is_dmcu_initialized = dcn10_is_dmcu_initialized,
+	.lock_phy = dcn20_lock_phy,
+	.unlock_phy = dcn20_unlock_phy
+};
+
+static const struct dmcu_funcs dcn21_funcs = {
+	.dmcu_init = dcn21_dmcu_init,
+	.load_iram = dcn10_dmcu_load_iram,
+	.set_psr_enable = dcn10_dmcu_set_psr_enable,
+	.setup_psr = dcn10_dmcu_setup_psr,
+	.get_psr_state = dcn10_get_dmcu_psr_state,
+	.set_psr_wait_loop = dcn10_psr_wait_loop,
+	.get_psr_wait_loop = dcn10_get_psr_wait_loop,
+	.is_dmcu_initialized = dcn10_is_dmcu_initialized,
+	.lock_phy = dcn20_lock_phy,
+	.unlock_phy = dcn20_unlock_phy
+};
+
+static void dce_dmcu_construct(
+	struct dce_dmcu *dmcu_dce,
+	struct dc_context *ctx,
+	const struct dce_dmcu_registers *regs,
+	const struct dce_dmcu_shift *dmcu_shift,
+	const struct dce_dmcu_mask *dmcu_mask)
+{
+	struct dmcu *base = &dmcu_dce->base;
+
+	base->ctx = ctx;
+	base->funcs = &dce_funcs;
+	base->cached_wait_loop_number = 0;
+
+	dmcu_dce->regs = regs;
+	dmcu_dce->dmcu_shift = dmcu_shift;
+	dmcu_dce->dmcu_mask = dmcu_mask;
+}
+
+static void dcn21_dmcu_construct(
+		struct dce_dmcu *dmcu_dce,
+		struct dc_context *ctx,
+		const struct dce_dmcu_registers *regs,
+		const struct dce_dmcu_shift *dmcu_shift,
+		const struct dce_dmcu_mask *dmcu_mask)
+{
+	uint32_t psp_version = 0;
+
+	dce_dmcu_construct(dmcu_dce, ctx, regs, dmcu_shift, dmcu_mask);
+
+	psp_version = dm_read_reg(ctx, mmMP0_SMN_C2PMSG_58);
+	dmcu_dce->base.auto_load_dmcu = ((psp_version & 0x00FF00FF) > 0x00110029);
+	dmcu_dce->base.psp_version = psp_version;
+}
+
+struct dmcu *dce_dmcu_create(
+	struct dc_context *ctx,
+	const struct dce_dmcu_registers *regs,
+	const struct dce_dmcu_shift *dmcu_shift,
+	const struct dce_dmcu_mask *dmcu_mask)
+{
+	struct dce_dmcu *dmcu_dce = kzalloc(sizeof(*dmcu_dce), GFP_KERNEL);
+
+	if (dmcu_dce == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	dce_dmcu_construct(
+		dmcu_dce, ctx, regs, dmcu_shift, dmcu_mask);
+
+	dmcu_dce->base.funcs = &dce_funcs;
+
+	return &dmcu_dce->base;
+}
+
+struct dmcu *dcn10_dmcu_create(
+	struct dc_context *ctx,
+	const struct dce_dmcu_registers *regs,
+	const struct dce_dmcu_shift *dmcu_shift,
+	const struct dce_dmcu_mask *dmcu_mask)
+{
+	struct dce_dmcu *dmcu_dce = kzalloc(sizeof(*dmcu_dce), GFP_ATOMIC);
+
+	if (dmcu_dce == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	dce_dmcu_construct(
+		dmcu_dce, ctx, regs, dmcu_shift, dmcu_mask);
+
+	dmcu_dce->base.funcs = &dcn10_funcs;
+
+	return &dmcu_dce->base;
+}
+
+struct dmcu *dcn20_dmcu_create(
+	struct dc_context *ctx,
+	const struct dce_dmcu_registers *regs,
+	const struct dce_dmcu_shift *dmcu_shift,
+	const struct dce_dmcu_mask *dmcu_mask)
+{
+	struct dce_dmcu *dmcu_dce = kzalloc(sizeof(*dmcu_dce), GFP_ATOMIC);
+
+	if (dmcu_dce == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	dce_dmcu_construct(
+		dmcu_dce, ctx, regs, dmcu_shift, dmcu_mask);
+
+	dmcu_dce->base.funcs = &dcn20_funcs;
+
+	return &dmcu_dce->base;
+}
+
+struct dmcu *dcn21_dmcu_create(
+	struct dc_context *ctx,
+	const struct dce_dmcu_registers *regs,
+	const struct dce_dmcu_shift *dmcu_shift,
+	const struct dce_dmcu_mask *dmcu_mask)
+{
+	struct dce_dmcu *dmcu_dce = kzalloc(sizeof(*dmcu_dce), GFP_ATOMIC);
+
+	if (dmcu_dce == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	dcn21_dmcu_construct(
+		dmcu_dce, ctx, regs, dmcu_shift, dmcu_mask);
+
+	dmcu_dce->base.funcs = &dcn21_funcs;
+
+	return &dmcu_dce->base;
+}
+
+void dce_dmcu_destroy(struct dmcu **dmcu)
+{
+	struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(*dmcu);
+
+	kfree(dmcu_dce);
+	*dmcu = NULL;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.h b/drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.h
new file mode 100644
index 000000000000..ff726b35ef6a
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.h
@@ -0,0 +1,331 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+#ifndef _DCE_DMCU_H_
+#define _DCE_DMCU_H_
+
+#include "dmcu.h"
+
+#define DMCU_COMMON_REG_LIST_DCE_BASE() \
+	SR(DMCU_CTRL), \
+	SR(DMCU_STATUS), \
+	SR(DMCU_RAM_ACCESS_CTRL), \
+	SR(DMCU_IRAM_WR_CTRL), \
+	SR(DMCU_IRAM_WR_DATA), \
+	SR(MASTER_COMM_DATA_REG1), \
+	SR(MASTER_COMM_DATA_REG2), \
+	SR(MASTER_COMM_DATA_REG3), \
+	SR(MASTER_COMM_CMD_REG), \
+	SR(MASTER_COMM_CNTL_REG), \
+	SR(SLAVE_COMM_DATA_REG1), \
+	SR(SLAVE_COMM_DATA_REG2), \
+	SR(SLAVE_COMM_DATA_REG3), \
+	SR(SLAVE_COMM_CMD_REG), \
+	SR(DMCU_IRAM_RD_CTRL), \
+	SR(DMCU_IRAM_RD_DATA), \
+	SR(DMCU_INTERRUPT_TO_UC_EN_MASK), \
+	SR(SMU_INTERRUPT_CONTROL), \
+	SR(DC_DMCU_SCRATCH)
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+#define DMCU_DCE60_REG_LIST() \
+	SR(DMCU_CTRL), \
+	SR(DMCU_STATUS), \
+	SR(DMCU_RAM_ACCESS_CTRL), \
+	SR(DMCU_IRAM_WR_CTRL), \
+	SR(DMCU_IRAM_WR_DATA), \
+	SR(MASTER_COMM_DATA_REG1), \
+	SR(MASTER_COMM_DATA_REG2), \
+	SR(MASTER_COMM_DATA_REG3), \
+	SR(MASTER_COMM_CMD_REG), \
+	SR(MASTER_COMM_CNTL_REG), \
+	SR(DMCU_IRAM_RD_CTRL), \
+	SR(DMCU_IRAM_RD_DATA), \
+	SR(DMCU_INTERRUPT_TO_UC_EN_MASK), \
+	SR(DC_DMCU_SCRATCH)
+#endif
+
+#define DMCU_DCE80_REG_LIST() \
+	SR(DMCU_CTRL), \
+	SR(DMCU_STATUS), \
+	SR(DMCU_RAM_ACCESS_CTRL), \
+	SR(DMCU_IRAM_WR_CTRL), \
+	SR(DMCU_IRAM_WR_DATA), \
+	SR(MASTER_COMM_DATA_REG1), \
+	SR(MASTER_COMM_DATA_REG2), \
+	SR(MASTER_COMM_DATA_REG3), \
+	SR(MASTER_COMM_CMD_REG), \
+	SR(MASTER_COMM_CNTL_REG), \
+	SR(DMCU_IRAM_RD_CTRL), \
+	SR(DMCU_IRAM_RD_DATA), \
+	SR(DMCU_INTERRUPT_TO_UC_EN_MASK), \
+	SR(SMU_INTERRUPT_CONTROL), \
+	SR(DC_DMCU_SCRATCH)
+
+#define DMCU_DCE110_COMMON_REG_LIST() \
+	DMCU_COMMON_REG_LIST_DCE_BASE(), \
+	SR(DCI_MEM_PWR_STATUS)
+
+#define DMCU_DCN10_REG_LIST()\
+	DMCU_COMMON_REG_LIST_DCE_BASE(), \
+	SR(DMU_MEM_PWR_CNTL)
+
+#define DMCU_DCN20_REG_LIST()\
+	DMCU_DCN10_REG_LIST(), \
+	SR(DMCUB_SCRATCH15)
+
+#define DMCU_SF(reg_name, field_name, post_fix)\
+	.field_name = reg_name ## __ ## field_name ## post_fix
+
+#define DMCU_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(mask_sh) \
+	DMCU_SF(DMCU_CTRL, \
+			DMCU_ENABLE, mask_sh), \
+	DMCU_SF(DMCU_STATUS, \
+			UC_IN_STOP_MODE, mask_sh), \
+	DMCU_SF(DMCU_STATUS, \
+			UC_IN_RESET, mask_sh), \
+	DMCU_SF(DMCU_RAM_ACCESS_CTRL, \
+			IRAM_HOST_ACCESS_EN, mask_sh), \
+	DMCU_SF(DMCU_RAM_ACCESS_CTRL, \
+			IRAM_WR_ADDR_AUTO_INC, mask_sh), \
+	DMCU_SF(DMCU_RAM_ACCESS_CTRL, \
+			IRAM_RD_ADDR_AUTO_INC, mask_sh), \
+	DMCU_SF(MASTER_COMM_CMD_REG, \
+			MASTER_COMM_CMD_REG_BYTE0, mask_sh), \
+	DMCU_SF(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, mask_sh), \
+	DMCU_SF(SLAVE_COMM_CNTL_REG, SLAVE_COMM_INTERRUPT, mask_sh), \
+	DMCU_SF(DMCU_INTERRUPT_TO_UC_EN_MASK, \
+			STATIC_SCREEN1_INT_TO_UC_EN, mask_sh), \
+	DMCU_SF(DMCU_INTERRUPT_TO_UC_EN_MASK, \
+			STATIC_SCREEN2_INT_TO_UC_EN, mask_sh), \
+	DMCU_SF(DMCU_INTERRUPT_TO_UC_EN_MASK, \
+			STATIC_SCREEN3_INT_TO_UC_EN, mask_sh), \
+	DMCU_SF(DMCU_INTERRUPT_TO_UC_EN_MASK, \
+			STATIC_SCREEN4_INT_TO_UC_EN, mask_sh), \
+	DMCU_SF(SMU_INTERRUPT_CONTROL, DC_SMU_INT_ENABLE, mask_sh)
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+#define DMCU_MASK_SH_LIST_DCE60(mask_sh) \
+	DMCU_SF(DMCU_CTRL, \
+			DMCU_ENABLE, mask_sh), \
+	DMCU_SF(DMCU_STATUS, \
+			UC_IN_STOP_MODE, mask_sh), \
+	DMCU_SF(DMCU_STATUS, \
+			UC_IN_RESET, mask_sh), \
+	DMCU_SF(DMCU_RAM_ACCESS_CTRL, \
+			IRAM_HOST_ACCESS_EN, mask_sh), \
+	DMCU_SF(DMCU_RAM_ACCESS_CTRL, \
+			IRAM_WR_ADDR_AUTO_INC, mask_sh), \
+	DMCU_SF(DMCU_RAM_ACCESS_CTRL, \
+			IRAM_RD_ADDR_AUTO_INC, mask_sh), \
+	DMCU_SF(MASTER_COMM_CMD_REG, \
+			MASTER_COMM_CMD_REG_BYTE0, mask_sh), \
+	DMCU_SF(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, mask_sh)
+#endif
+
+#define DMCU_MASK_SH_LIST_DCE80(mask_sh) \
+	DMCU_SF(DMCU_CTRL, \
+			DMCU_ENABLE, mask_sh), \
+	DMCU_SF(DMCU_STATUS, \
+			UC_IN_STOP_MODE, mask_sh), \
+	DMCU_SF(DMCU_STATUS, \
+			UC_IN_RESET, mask_sh), \
+	DMCU_SF(DMCU_RAM_ACCESS_CTRL, \
+			IRAM_HOST_ACCESS_EN, mask_sh), \
+	DMCU_SF(DMCU_RAM_ACCESS_CTRL, \
+			IRAM_WR_ADDR_AUTO_INC, mask_sh), \
+	DMCU_SF(DMCU_RAM_ACCESS_CTRL, \
+			IRAM_RD_ADDR_AUTO_INC, mask_sh), \
+	DMCU_SF(MASTER_COMM_CMD_REG, \
+			MASTER_COMM_CMD_REG_BYTE0, mask_sh), \
+	DMCU_SF(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, mask_sh), \
+	DMCU_SF(SMU_INTERRUPT_CONTROL, DC_SMU_INT_ENABLE, mask_sh)
+
+#define DMCU_MASK_SH_LIST_DCE110(mask_sh) \
+	DMCU_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(mask_sh), \
+	DMCU_SF(DCI_MEM_PWR_STATUS, \
+		DMCU_IRAM_MEM_PWR_STATE, mask_sh)
+
+#define DMCU_MASK_SH_LIST_DCN10(mask_sh) \
+	DMCU_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(mask_sh), \
+	DMCU_SF(DMU_MEM_PWR_CNTL, \
+			DMCU_IRAM_MEM_PWR_STATE, mask_sh)
+
+#define DMCU_REG_FIELD_LIST(type) \
+	type DMCU_IRAM_MEM_PWR_STATE; \
+	type IRAM_HOST_ACCESS_EN; \
+	type IRAM_WR_ADDR_AUTO_INC; \
+	type IRAM_RD_ADDR_AUTO_INC; \
+	type DMCU_ENABLE; \
+	type UC_IN_STOP_MODE; \
+	type UC_IN_RESET; \
+	type MASTER_COMM_CMD_REG_BYTE0; \
+	type MASTER_COMM_INTERRUPT; \
+	type SLAVE_COMM_INTERRUPT; \
+	type DPHY_RX_FAST_TRAINING_CAPABLE; \
+	type DPHY_LOAD_BS_COUNT; \
+	type STATIC_SCREEN1_INT_TO_UC_EN; \
+	type STATIC_SCREEN2_INT_TO_UC_EN; \
+	type STATIC_SCREEN3_INT_TO_UC_EN; \
+	type STATIC_SCREEN4_INT_TO_UC_EN; \
+	type DP_SEC_GSP0_LINE_NUM; \
+	type DP_SEC_GSP0_PRIORITY; \
+	type DC_SMU_INT_ENABLE
+
+struct dce_dmcu_shift {
+	DMCU_REG_FIELD_LIST(uint8_t);
+};
+
+struct dce_dmcu_mask {
+	DMCU_REG_FIELD_LIST(uint32_t);
+};
+
+struct dce_dmcu_registers {
+	uint32_t DMCU_CTRL;
+	uint32_t DMCU_STATUS;
+	uint32_t DMCU_RAM_ACCESS_CTRL;
+	uint32_t DCI_MEM_PWR_STATUS;
+	uint32_t DMU_MEM_PWR_CNTL;
+	uint32_t DMCU_IRAM_WR_CTRL;
+	uint32_t DMCU_IRAM_WR_DATA;
+
+	uint32_t MASTER_COMM_DATA_REG1;
+	uint32_t MASTER_COMM_DATA_REG2;
+	uint32_t MASTER_COMM_DATA_REG3;
+	uint32_t MASTER_COMM_CMD_REG;
+	uint32_t MASTER_COMM_CNTL_REG;
+	uint32_t SLAVE_COMM_DATA_REG1;
+	uint32_t SLAVE_COMM_DATA_REG2;
+	uint32_t SLAVE_COMM_DATA_REG3;
+	uint32_t SLAVE_COMM_CMD_REG;
+	uint32_t SLAVE_COMM_CNTL_REG;
+	uint32_t DMCU_IRAM_RD_CTRL;
+	uint32_t DMCU_IRAM_RD_DATA;
+	uint32_t DMCU_INTERRUPT_TO_UC_EN_MASK;
+	uint32_t SMU_INTERRUPT_CONTROL;
+	uint32_t DC_DMCU_SCRATCH;
+	uint32_t DMCUB_SCRATCH15;
+};
+
+struct dce_dmcu {
+	struct dmcu base;
+	const struct dce_dmcu_registers *regs;
+	const struct dce_dmcu_shift *dmcu_shift;
+	const struct dce_dmcu_mask *dmcu_mask;
+};
+
+/*******************************************************************
+ *   MASTER_COMM_DATA_REG1   Bit position    Data
+ *                           7:0	            hyst_frames[7:0]
+ *                           14:8	        hyst_lines[6:0]
+ *                           15	            RFB_UPDATE_AUTO_EN
+ *                           18:16	        phy_num[2:0]
+ *                           21:19	        dcp_sel[2:0]
+ *                           22	            phy_type
+ *                           23	            frame_cap_ind
+ *                           26:24	        aux_chan[2:0]
+ *                           30:27	        aux_repeat[3:0]
+ *                           31:31	        reserved[31:31]
+ ******************************************************************/
+union dce_dmcu_psr_config_data_reg1 {
+	struct {
+		unsigned int timehyst_frames:8;                  /*[7:0]*/
+		unsigned int hyst_lines:7;                       /*[14:8]*/
+		unsigned int rfb_update_auto_en:1;               /*[15:15]*/
+		unsigned int dp_port_num:3;                      /*[18:16]*/
+		unsigned int dcp_sel:3;                          /*[21:19]*/
+		unsigned int phy_type:1;                         /*[22:22]*/
+		unsigned int frame_cap_ind:1;                    /*[23:23]*/
+		unsigned int aux_chan:3;                         /*[26:24]*/
+		unsigned int aux_repeat:4;                       /*[30:27]*/
+		unsigned int allow_smu_optimizations:1;         /*[31:31]*/
+	} bits;
+	unsigned int u32All;
+};
+
+/*******************************************************************
+ *   MASTER_COMM_DATA_REG2
+ *******************************************************************/
+union dce_dmcu_psr_config_data_reg2 {
+	struct {
+		unsigned int dig_fe:3;                  /*[2:0]*/
+		unsigned int dig_be:3;                  /*[5:3]*/
+		unsigned int skip_wait_for_pll_lock:1;  /*[6:6]*/
+		unsigned int reserved:9;                /*[15:7]*/
+		unsigned int frame_delay:8;             /*[23:16]*/
+		unsigned int smu_phy_id:4;              /*[27:24]*/
+		unsigned int num_of_controllers:4;      /*[31:28]*/
+	} bits;
+	unsigned int u32All;
+};
+
+/*******************************************************************
+ *   MASTER_COMM_DATA_REG3
+ *******************************************************************/
+union dce_dmcu_psr_config_data_reg3 {
+	struct {
+		unsigned int psr_level:16;      /*[15:0]*/
+		unsigned int link_rate:4;       /*[19:16]*/
+		unsigned int reserved:12;        /*[31:20]*/
+	} bits;
+	unsigned int u32All;
+};
+
+union dce_dmcu_psr_config_data_wait_loop_reg1 {
+	struct {
+		unsigned int wait_loop:16; /* [15:0] */
+		unsigned int reserved:16; /* [31:16] */
+	} bits;
+	unsigned int u32;
+};
+
+struct dmcu *dce_dmcu_create(
+	struct dc_context *ctx,
+	const struct dce_dmcu_registers *regs,
+	const struct dce_dmcu_shift *dmcu_shift,
+	const struct dce_dmcu_mask *dmcu_mask);
+
+struct dmcu *dcn10_dmcu_create(
+	struct dc_context *ctx,
+	const struct dce_dmcu_registers *regs,
+	const struct dce_dmcu_shift *dmcu_shift,
+	const struct dce_dmcu_mask *dmcu_mask);
+
+struct dmcu *dcn20_dmcu_create(
+	struct dc_context *ctx,
+	const struct dce_dmcu_registers *regs,
+	const struct dce_dmcu_shift *dmcu_shift,
+	const struct dce_dmcu_mask *dmcu_mask);
+
+struct dmcu *dcn21_dmcu_create(
+	struct dc_context *ctx,
+	const struct dce_dmcu_registers *regs,
+	const struct dce_dmcu_shift *dmcu_shift,
+	const struct dce_dmcu_mask *dmcu_mask);
+
+void dce_dmcu_destroy(struct dmcu **dmcu);
+
+#endif /* _DCE_ABM_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_i2c.c b/drivers/gpu/drm/amd/display/dc/dce/dce_i2c.c
new file mode 100644
index 000000000000..f5cd2392fc5f
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_i2c.c
@@ -0,0 +1,83 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+#include "dce_i2c.h"
+#include "reg_helper.h"
+
+bool dce_i2c_oem_device_present(
+	struct resource_pool *pool,
+	struct ddc_service *ddc,
+	size_t slave_address
+)
+{
+	struct dc *dc = ddc->ctx->dc;
+	struct dc_bios *dcb = dc->ctx->dc_bios;
+	struct graphics_object_id id = {0};
+	struct graphics_object_i2c_info i2c_info;
+
+	if (!dc->ctx->dc_bios->fw_info.oem_i2c_present)
+		return false;
+
+	id.id = dc->ctx->dc_bios->fw_info.oem_i2c_obj_id;
+	id.enum_id = 0;
+	id.type = OBJECT_TYPE_GENERIC;
+	if (dcb->funcs->get_i2c_info(dcb, id, &i2c_info) != BP_RESULT_OK)
+		return false;
+
+	if (i2c_info.i2c_slave_address != slave_address)
+		return false;
+
+	return true;
+}
+
+bool dce_i2c_submit_command(
+	struct resource_pool *pool,
+	struct ddc *ddc,
+	struct i2c_command *cmd)
+{
+	struct dce_i2c_hw *dce_i2c_hw;
+	struct dce_i2c_sw dce_i2c_sw = {0};
+
+	if (!ddc) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+
+	if (!cmd) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+
+	dce_i2c_hw = acquire_i2c_hw_engine(pool, ddc);
+
+	if (dce_i2c_hw)
+		return dce_i2c_submit_command_hw(pool, ddc, cmd, dce_i2c_hw);
+
+	dce_i2c_sw.ctx = ddc->ctx;
+	if (dce_i2c_engine_acquire_sw(&dce_i2c_sw, ddc)) {
+		return dce_i2c_submit_command_sw(pool, ddc, cmd, &dce_i2c_sw);
+	}
+
+	return false;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_i2c.h b/drivers/gpu/drm/amd/display/dc/dce/dce_i2c.h
new file mode 100644
index 000000000000..535fd58de450
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_i2c.h
@@ -0,0 +1,44 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCE_I2C_H__
+#define __DCE_I2C_H__
+
+#include "inc/core_types.h"
+#include "dce_i2c_hw.h"
+#include "dce_i2c_sw.h"
+
+bool dce_i2c_oem_device_present(
+	struct resource_pool *pool,
+	struct ddc_service *ddc,
+	size_t slave_address
+);
+
+bool dce_i2c_submit_command(
+	struct resource_pool *pool,
+	struct ddc *ddc,
+	struct i2c_command *cmd);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_i2c_hw.c b/drivers/gpu/drm/amd/display/dc/dce/dce_i2c_hw.c
new file mode 100644
index 000000000000..365dd2e37aea
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_i2c_hw.c
@@ -0,0 +1,756 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+#include "resource.h"
+#include "dce_i2c.h"
+#include "dce_i2c_hw.h"
+#include "reg_helper.h"
+#include "include/gpio_service_interface.h"
+
+#define CTX \
+	dce_i2c_hw->ctx
+#define REG(reg)\
+	dce_i2c_hw->regs->reg
+
+#undef FN
+#define FN(reg_name, field_name) \
+	dce_i2c_hw->shifts->field_name, dce_i2c_hw->masks->field_name
+
+static void execute_transaction(
+	struct dce_i2c_hw *dce_i2c_hw)
+{
+	REG_UPDATE_N(SETUP, 5,
+		     FN(DC_I2C_DDC1_SETUP, DC_I2C_DDC1_DATA_DRIVE_EN), 0,
+		     FN(DC_I2C_DDC1_SETUP, DC_I2C_DDC1_CLK_DRIVE_EN), 0,
+		     FN(DC_I2C_DDC1_SETUP, DC_I2C_DDC1_DATA_DRIVE_SEL), 0,
+		     FN(DC_I2C_DDC1_SETUP, DC_I2C_DDC1_INTRA_TRANSACTION_DELAY), 0,
+		     FN(DC_I2C_DDC1_SETUP, DC_I2C_DDC1_INTRA_BYTE_DELAY), 0);
+
+
+	REG_UPDATE_5(DC_I2C_CONTROL,
+		     DC_I2C_SOFT_RESET, 0,
+		     DC_I2C_SW_STATUS_RESET, 0,
+		     DC_I2C_SEND_RESET, 0,
+		     DC_I2C_GO, 0,
+		     DC_I2C_TRANSACTION_COUNT, dce_i2c_hw->transaction_count - 1);
+
+	/* start I2C transfer */
+	REG_UPDATE(DC_I2C_CONTROL, DC_I2C_GO, 1);
+
+	/* all transactions were executed and HW buffer became empty
+	 * (even though it actually happens when status becomes DONE)
+	 */
+	dce_i2c_hw->transaction_count = 0;
+	dce_i2c_hw->buffer_used_bytes = 0;
+}
+
+static enum i2c_channel_operation_result get_channel_status(
+	struct dce_i2c_hw *dce_i2c_hw,
+	uint8_t *returned_bytes)
+{
+	uint32_t i2c_sw_status = 0;
+	uint32_t value =
+		REG_GET(DC_I2C_SW_STATUS, DC_I2C_SW_STATUS, &i2c_sw_status);
+	if (i2c_sw_status == DC_I2C_STATUS__DC_I2C_STATUS_USED_BY_SW)
+		return I2C_CHANNEL_OPERATION_ENGINE_BUSY;
+	else if (value & dce_i2c_hw->masks->DC_I2C_SW_STOPPED_ON_NACK)
+		return I2C_CHANNEL_OPERATION_NO_RESPONSE;
+	else if (value & dce_i2c_hw->masks->DC_I2C_SW_TIMEOUT)
+		return I2C_CHANNEL_OPERATION_TIMEOUT;
+	else if (value & dce_i2c_hw->masks->DC_I2C_SW_ABORTED)
+		return I2C_CHANNEL_OPERATION_FAILED;
+	else if (value & dce_i2c_hw->masks->DC_I2C_SW_DONE)
+		return I2C_CHANNEL_OPERATION_SUCCEEDED;
+
+	/*
+	 * this is the case when HW used for communication, I2C_SW_STATUS
+	 * could be zero
+	 */
+	return I2C_CHANNEL_OPERATION_SUCCEEDED;
+}
+
+static uint32_t get_hw_buffer_available_size(
+	const struct dce_i2c_hw *dce_i2c_hw)
+{
+	return dce_i2c_hw->buffer_size -
+			dce_i2c_hw->buffer_used_bytes;
+}
+
+static void process_channel_reply(
+	struct dce_i2c_hw *dce_i2c_hw,
+	struct i2c_payload *reply)
+{
+	uint32_t length = reply->length;
+	uint8_t *buffer = reply->data;
+
+	REG_SET_3(DC_I2C_DATA, 0,
+		 DC_I2C_INDEX, dce_i2c_hw->buffer_used_write,
+		 DC_I2C_DATA_RW, 1,
+		 DC_I2C_INDEX_WRITE, 1);
+
+	while (length) {
+		/* after reading the status,
+		 * if the I2C operation executed successfully
+		 * (i.e. DC_I2C_STATUS_DONE = 1) then the I2C controller
+		 * should read data bytes from I2C circular data buffer
+		 */
+
+		uint32_t i2c_data;
+
+		REG_GET(DC_I2C_DATA, DC_I2C_DATA, &i2c_data);
+		*buffer++ = i2c_data;
+
+		--length;
+	}
+}
+
+static bool is_engine_available(struct dce_i2c_hw *dce_i2c_hw)
+{
+	unsigned int arbitrate;
+	unsigned int i2c_hw_status;
+
+	REG_GET(HW_STATUS, DC_I2C_DDC1_HW_STATUS, &i2c_hw_status);
+	if (i2c_hw_status == DC_I2C_STATUS__DC_I2C_STATUS_USED_BY_HW)
+		return false;
+
+	REG_GET(DC_I2C_ARBITRATION, DC_I2C_REG_RW_CNTL_STATUS, &arbitrate);
+	if (arbitrate == DC_I2C_REG_RW_CNTL_STATUS_DMCU_ONLY)
+		return false;
+
+	return true;
+}
+
+static bool is_hw_busy(struct dce_i2c_hw *dce_i2c_hw)
+{
+	uint32_t i2c_sw_status = 0;
+
+	REG_GET(DC_I2C_SW_STATUS, DC_I2C_SW_STATUS, &i2c_sw_status);
+	if (i2c_sw_status == DC_I2C_STATUS__DC_I2C_STATUS_IDLE)
+		return false;
+
+	if (is_engine_available(dce_i2c_hw))
+		return false;
+
+	return true;
+}
+
+static bool process_transaction(
+	struct dce_i2c_hw *dce_i2c_hw,
+	struct i2c_request_transaction_data *request)
+{
+	uint32_t length = request->length;
+	uint8_t *buffer = request->data;
+
+	bool last_transaction = false;
+	uint32_t value = 0;
+
+	if (is_hw_busy(dce_i2c_hw)) {
+		request->status = I2C_CHANNEL_OPERATION_ENGINE_BUSY;
+		return false;
+	}
+
+	last_transaction = ((dce_i2c_hw->transaction_count == 3) ||
+			(request->action == DCE_I2C_TRANSACTION_ACTION_I2C_WRITE) ||
+			(request->action & DCE_I2C_TRANSACTION_ACTION_I2C_READ));
+
+
+	switch (dce_i2c_hw->transaction_count) {
+	case 0:
+		REG_UPDATE_5(DC_I2C_TRANSACTION0,
+				 DC_I2C_STOP_ON_NACK0, 1,
+				 DC_I2C_START0, 1,
+				 DC_I2C_RW0, 0 != (request->action & DCE_I2C_TRANSACTION_ACTION_I2C_READ),
+				 DC_I2C_COUNT0, length,
+				 DC_I2C_STOP0, last_transaction ? 1 : 0);
+		break;
+	case 1:
+		REG_UPDATE_5(DC_I2C_TRANSACTION1,
+				 DC_I2C_STOP_ON_NACK0, 1,
+				 DC_I2C_START0, 1,
+				 DC_I2C_RW0, 0 != (request->action & DCE_I2C_TRANSACTION_ACTION_I2C_READ),
+				 DC_I2C_COUNT0, length,
+				 DC_I2C_STOP0, last_transaction ? 1 : 0);
+		break;
+	case 2:
+		REG_UPDATE_5(DC_I2C_TRANSACTION2,
+				 DC_I2C_STOP_ON_NACK0, 1,
+				 DC_I2C_START0, 1,
+				 DC_I2C_RW0, 0 != (request->action & DCE_I2C_TRANSACTION_ACTION_I2C_READ),
+				 DC_I2C_COUNT0, length,
+				 DC_I2C_STOP0, last_transaction ? 1 : 0);
+		break;
+	case 3:
+		REG_UPDATE_5(DC_I2C_TRANSACTION3,
+				 DC_I2C_STOP_ON_NACK0, 1,
+				 DC_I2C_START0, 1,
+				 DC_I2C_RW0, 0 != (request->action & DCE_I2C_TRANSACTION_ACTION_I2C_READ),
+				 DC_I2C_COUNT0, length,
+				 DC_I2C_STOP0, last_transaction ? 1 : 0);
+		break;
+	default:
+		/* TODO Warning ? */
+		break;
+	}
+
+	/* Write the I2C address and I2C data
+	 * into the hardware circular buffer, one byte per entry.
+	 * As an example, the 7-bit I2C slave address for CRT monitor
+	 * for reading DDC/EDID information is 0b1010001.
+	 * For an I2C send operation, the LSB must be programmed to 0;
+	 * for I2C receive operation, the LSB must be programmed to 1.
+	 */
+	if (dce_i2c_hw->transaction_count == 0) {
+		value = REG_SET_4(DC_I2C_DATA, 0,
+				  DC_I2C_DATA_RW, false,
+				  DC_I2C_DATA, request->address,
+				  DC_I2C_INDEX, 0,
+				  DC_I2C_INDEX_WRITE, 1);
+		dce_i2c_hw->buffer_used_write = 0;
+	} else
+		value = REG_SET_2(DC_I2C_DATA, 0,
+			  DC_I2C_DATA_RW, false,
+			  DC_I2C_DATA, request->address);
+
+	dce_i2c_hw->buffer_used_write++;
+
+	if (!(request->action & DCE_I2C_TRANSACTION_ACTION_I2C_READ)) {
+		while (length) {
+			REG_SET_2(DC_I2C_DATA, value,
+				  DC_I2C_INDEX_WRITE, 0,
+				  DC_I2C_DATA, *buffer++);
+			dce_i2c_hw->buffer_used_write++;
+			--length;
+		}
+	}
+
+	++dce_i2c_hw->transaction_count;
+	dce_i2c_hw->buffer_used_bytes += length + 1;
+
+	return last_transaction;
+}
+
+static inline void reset_hw_engine(struct dce_i2c_hw *dce_i2c_hw)
+{
+	REG_UPDATE_2(DC_I2C_CONTROL,
+		     DC_I2C_SW_STATUS_RESET, 1,
+		     DC_I2C_SW_STATUS_RESET, 1);
+}
+
+static void set_speed(
+	struct dce_i2c_hw *dce_i2c_hw,
+	uint32_t speed)
+{
+	uint32_t xtal_ref_div = 0, ref_base_div = 0;
+	uint32_t prescale = 0;
+	uint32_t i2c_ref_clock = 0;
+
+	if (speed == 0)
+		return;
+
+	REG_GET_2(MICROSECOND_TIME_BASE_DIV, MICROSECOND_TIME_BASE_DIV, &ref_base_div,
+		XTAL_REF_DIV, &xtal_ref_div);
+
+	if (xtal_ref_div == 0)
+		xtal_ref_div = 2;
+
+	if (ref_base_div == 0)
+		i2c_ref_clock = (dce_i2c_hw->reference_frequency * 2);
+	else
+		i2c_ref_clock = ref_base_div * 1000;
+
+	prescale = (i2c_ref_clock / xtal_ref_div) / speed;
+
+	if (dce_i2c_hw->masks->DC_I2C_DDC1_START_STOP_TIMING_CNTL)
+		REG_UPDATE_N(SPEED, 3,
+			     FN(DC_I2C_DDC1_SPEED, DC_I2C_DDC1_PRESCALE), prescale,
+			     FN(DC_I2C_DDC1_SPEED, DC_I2C_DDC1_THRESHOLD), 2,
+			     FN(DC_I2C_DDC1_SPEED, DC_I2C_DDC1_START_STOP_TIMING_CNTL), speed > 50 ? 2:1);
+	else
+		REG_UPDATE_N(SPEED, 2,
+			     FN(DC_I2C_DDC1_SPEED, DC_I2C_DDC1_PRESCALE), prescale,
+			     FN(DC_I2C_DDC1_SPEED, DC_I2C_DDC1_THRESHOLD), 2);
+}
+
+static bool acquire_engine(struct dce_i2c_hw *dce_i2c_hw)
+{
+	uint32_t arbitrate = 0;
+
+	REG_GET(DC_I2C_ARBITRATION, DC_I2C_REG_RW_CNTL_STATUS, &arbitrate);
+	switch (arbitrate) {
+	case DC_I2C_STATUS__DC_I2C_STATUS_USED_BY_SW:
+		return true;
+	case DC_I2C_STATUS__DC_I2C_STATUS_USED_BY_HW:
+		return false;
+	case DC_I2C_STATUS__DC_I2C_STATUS_IDLE:
+	default:
+		break;
+	}
+
+	REG_UPDATE(DC_I2C_ARBITRATION, DC_I2C_SW_USE_I2C_REG_REQ, true);
+	REG_GET(DC_I2C_ARBITRATION, DC_I2C_REG_RW_CNTL_STATUS, &arbitrate);
+	if (arbitrate != DC_I2C_STATUS__DC_I2C_STATUS_USED_BY_SW)
+		return false;
+
+	return true;
+}
+
+static bool setup_engine(
+	struct dce_i2c_hw *dce_i2c_hw)
+{
+	// Deassert soft reset to unblock I2C engine registers
+	REG_UPDATE(DC_I2C_CONTROL, DC_I2C_SOFT_RESET, false);
+
+	uint32_t i2c_setup_limit = I2C_SETUP_TIME_LIMIT_DCE;
+	uint32_t  reset_length = 0;
+
+	if (dce_i2c_hw->ctx->dc->debug.enable_mem_low_power.bits.i2c) {
+		if (dce_i2c_hw->regs->DIO_MEM_PWR_CTRL) {
+			REG_UPDATE(DIO_MEM_PWR_CTRL, I2C_LIGHT_SLEEP_FORCE, 0);
+			REG_WAIT(DIO_MEM_PWR_STATUS, I2C_MEM_PWR_STATE, 0, 0, 5);
+		}
+	}
+
+	if (dce_i2c_hw->masks->DC_I2C_DDC1_CLK_EN)
+		REG_UPDATE_N(SETUP, 1,
+			     FN(DC_I2C_DDC1_SETUP, DC_I2C_DDC1_CLK_EN), 1);
+
+	if (!acquire_engine(dce_i2c_hw))
+		return false;
+
+	/*set SW requested I2c speed to default, if API calls in it will be override later*/
+	set_speed(dce_i2c_hw, dce_i2c_hw->ctx->dc->caps.i2c_speed_in_khz);
+
+	if (dce_i2c_hw->setup_limit != 0)
+		i2c_setup_limit = dce_i2c_hw->setup_limit;
+
+	/* Program pin select */
+	REG_UPDATE_5(DC_I2C_CONTROL,
+		     DC_I2C_GO, 0,
+		     DC_I2C_SEND_RESET, 0,
+		     DC_I2C_SW_STATUS_RESET, 1,
+		     DC_I2C_TRANSACTION_COUNT, 0,
+		     DC_I2C_DDC_SELECT, dce_i2c_hw->engine_id);
+
+	/* Program time limit */
+	if (dce_i2c_hw->send_reset_length == 0) {
+		/*pre-dcn*/
+		REG_UPDATE_N(SETUP, 2,
+			     FN(DC_I2C_DDC1_SETUP, DC_I2C_DDC1_TIME_LIMIT), i2c_setup_limit,
+			     FN(DC_I2C_DDC1_SETUP, DC_I2C_DDC1_ENABLE), 1);
+	} else {
+		reset_length = dce_i2c_hw->send_reset_length;
+		REG_UPDATE_N(SETUP, 3,
+			     FN(DC_I2C_DDC1_SETUP, DC_I2C_DDC1_TIME_LIMIT), i2c_setup_limit,
+			     FN(DC_I2C_DDC1_SETUP, DC_I2C_DDC1_SEND_RESET_LENGTH), reset_length,
+			     FN(DC_I2C_DDC1_SETUP, DC_I2C_DDC1_ENABLE), 1);
+	}
+	/* Program HW priority
+	 * set to High - interrupt software I2C at any time
+	 * Enable restart of SW I2C that was interrupted by HW
+	 * disable queuing of software while I2C is in use by HW
+	 */
+	REG_UPDATE(DC_I2C_ARBITRATION,
+			DC_I2C_NO_QUEUED_SW_GO, 0);
+
+	return true;
+}
+
+/**
+ * cntl_stuck_hw_workaround - Workaround for I2C engine stuck state
+ * @dce_i2c_hw: Pointer to dce_i2c_hw structure
+ *
+ * If we boot without an HDMI display, the I2C engine does not get initialized
+ * correctly. One of its symptoms is that SW_USE_I2C does not get cleared after
+ * acquire. After setting SW_DONE_USING_I2C on release, the engine gets
+ * immediately reacquired by SW, preventing DMUB from using it.
+ *
+ * This function checks the I2C arbitration status and applies a release
+ * workaround if necessary.
+ */
+static void cntl_stuck_hw_workaround(struct dce_i2c_hw *dce_i2c_hw)
+{
+	uint32_t arbitrate = 0;
+
+	REG_GET(DC_I2C_ARBITRATION, DC_I2C_REG_RW_CNTL_STATUS, &arbitrate);
+	if (arbitrate != DC_I2C_STATUS__DC_I2C_STATUS_USED_BY_SW)
+		return;
+
+	// Still acquired after release, release again as a workaround
+	REG_UPDATE(DC_I2C_ARBITRATION, DC_I2C_SW_DONE_USING_I2C_REG, true);
+	REG_GET(DC_I2C_ARBITRATION, DC_I2C_REG_RW_CNTL_STATUS, &arbitrate);
+	ASSERT(arbitrate != DC_I2C_STATUS__DC_I2C_STATUS_USED_BY_SW);
+}
+
+static void release_engine(
+	struct dce_i2c_hw *dce_i2c_hw)
+{
+	bool safe_to_reset;
+
+
+	/* Reset HW engine */
+	{
+		uint32_t i2c_sw_status = 0;
+
+		REG_GET(DC_I2C_SW_STATUS, DC_I2C_SW_STATUS, &i2c_sw_status);
+		/* if used by SW, safe to reset */
+		safe_to_reset = (i2c_sw_status == 1);
+	}
+
+	if (safe_to_reset)
+		REG_UPDATE_2(DC_I2C_CONTROL,
+			     DC_I2C_SOFT_RESET, 1,
+			     DC_I2C_SW_STATUS_RESET, 1);
+	else
+		REG_UPDATE(DC_I2C_CONTROL, DC_I2C_SW_STATUS_RESET, 1);
+	/* HW I2c engine - clock gating feature */
+	if (!dce_i2c_hw->engine_keep_power_up_count)
+		REG_UPDATE_N(SETUP, 1, FN(SETUP, DC_I2C_DDC1_ENABLE), 0);
+
+	/*for HW HDCP Ri polling failure w/a test*/
+	set_speed(dce_i2c_hw, dce_i2c_hw->ctx->dc->caps.i2c_speed_in_khz_hdcp);
+	// Release I2C engine so it can be used by HW or DMCU, automatically clears SW_USE_I2C
+	REG_UPDATE(DC_I2C_ARBITRATION, DC_I2C_SW_DONE_USING_I2C_REG, true);
+	cntl_stuck_hw_workaround(dce_i2c_hw);
+
+	if (dce_i2c_hw->ctx->dc->debug.enable_mem_low_power.bits.i2c) {
+		if (dce_i2c_hw->regs->DIO_MEM_PWR_CTRL)
+			REG_UPDATE(DIO_MEM_PWR_CTRL, I2C_LIGHT_SLEEP_FORCE, 1);
+	}
+}
+
+struct dce_i2c_hw *acquire_i2c_hw_engine(
+	struct resource_pool *pool,
+	struct ddc *ddc)
+{
+	uint32_t counter = 0;
+	enum gpio_result result;
+	struct dce_i2c_hw *dce_i2c_hw = NULL;
+
+	if (!ddc)
+		return NULL;
+
+	if (ddc->hw_info.hw_supported) {
+		enum gpio_ddc_line line = dal_ddc_get_line(ddc);
+
+		if (line < pool->res_cap->num_ddc)
+			dce_i2c_hw = pool->hw_i2cs[line];
+	}
+
+	if (!dce_i2c_hw)
+		return NULL;
+
+	if (pool->i2c_hw_buffer_in_use || !is_engine_available(dce_i2c_hw))
+		return NULL;
+
+	do {
+		result = dal_ddc_open(ddc, GPIO_MODE_HARDWARE,
+			GPIO_DDC_CONFIG_TYPE_MODE_I2C);
+
+		if (result == GPIO_RESULT_OK)
+			break;
+
+		/* i2c_engine is busy by VBios, lets wait and retry */
+
+		udelay(10);
+
+		++counter;
+	} while (counter < 2);
+
+	if (result != GPIO_RESULT_OK)
+		return NULL;
+
+	dce_i2c_hw->ddc = ddc;
+
+	if (!setup_engine(dce_i2c_hw)) {
+		release_engine(dce_i2c_hw);
+		return NULL;
+	}
+
+	pool->i2c_hw_buffer_in_use = true;
+	return dce_i2c_hw;
+}
+
+static enum i2c_channel_operation_result dce_i2c_hw_engine_wait_on_operation_result(struct dce_i2c_hw *dce_i2c_hw,
+										    uint32_t timeout,
+										    enum i2c_channel_operation_result expected_result)
+{
+	enum i2c_channel_operation_result result;
+	uint32_t i = 0;
+
+	if (!timeout)
+		return I2C_CHANNEL_OPERATION_SUCCEEDED;
+
+	do {
+
+		result = get_channel_status(
+				dce_i2c_hw, NULL);
+
+		if (result != expected_result)
+			break;
+
+		udelay(1);
+
+		++i;
+	} while (i < timeout);
+	return result;
+}
+
+static void submit_channel_request_hw(
+	struct dce_i2c_hw *dce_i2c_hw,
+	struct i2c_request_transaction_data *request)
+{
+	request->status = I2C_CHANNEL_OPERATION_SUCCEEDED;
+
+	if (!process_transaction(dce_i2c_hw, request))
+		return;
+
+	if (is_hw_busy(dce_i2c_hw)) {
+		request->status = I2C_CHANNEL_OPERATION_ENGINE_BUSY;
+		return;
+	}
+	reset_hw_engine(dce_i2c_hw);
+
+	execute_transaction(dce_i2c_hw);
+
+
+}
+
+static uint32_t get_transaction_timeout_hw(
+	const struct dce_i2c_hw *dce_i2c_hw,
+	uint32_t length,
+	uint32_t speed)
+{
+	uint32_t period_timeout;
+	uint32_t num_of_clock_stretches;
+
+	if (!speed)
+		return 0;
+
+	period_timeout = (1000 * TRANSACTION_TIMEOUT_IN_I2C_CLOCKS) / speed;
+
+	num_of_clock_stretches = 1 + (length << 3) + 1;
+	num_of_clock_stretches +=
+		(dce_i2c_hw->buffer_used_bytes << 3) +
+		(dce_i2c_hw->transaction_count << 1);
+
+	return period_timeout * num_of_clock_stretches;
+}
+
+static bool dce_i2c_hw_engine_submit_payload(struct dce_i2c_hw *dce_i2c_hw,
+					     struct i2c_payload *payload,
+					     bool middle_of_transaction,
+					     uint32_t speed)
+{
+
+	struct i2c_request_transaction_data request;
+
+	uint32_t transaction_timeout;
+
+	enum i2c_channel_operation_result operation_result;
+
+	bool result = false;
+
+	/* We need following:
+	 * transaction length will not exceed
+	 * the number of free bytes in HW buffer (minus one for address)
+	 */
+
+	if (payload->length >=
+			get_hw_buffer_available_size(dce_i2c_hw)) {
+		return false;
+	}
+
+	if (!payload->write)
+		request.action = middle_of_transaction ?
+			DCE_I2C_TRANSACTION_ACTION_I2C_READ_MOT :
+			DCE_I2C_TRANSACTION_ACTION_I2C_READ;
+	else
+		request.action = middle_of_transaction ?
+			DCE_I2C_TRANSACTION_ACTION_I2C_WRITE_MOT :
+			DCE_I2C_TRANSACTION_ACTION_I2C_WRITE;
+
+
+	request.address = (uint8_t) ((payload->address << 1) | (payload->write ? 0 : 1));
+	request.length = payload->length;
+	request.data = payload->data;
+
+	/* obtain timeout value before submitting request */
+
+	transaction_timeout = get_transaction_timeout_hw(
+		dce_i2c_hw, payload->length + 1, speed);
+
+	submit_channel_request_hw(
+		dce_i2c_hw, &request);
+
+	if ((request.status == I2C_CHANNEL_OPERATION_FAILED) ||
+		(request.status == I2C_CHANNEL_OPERATION_ENGINE_BUSY))
+		return false;
+
+	/* wait until transaction proceed */
+
+	operation_result = dce_i2c_hw_engine_wait_on_operation_result(
+		dce_i2c_hw,
+		transaction_timeout,
+		I2C_CHANNEL_OPERATION_ENGINE_BUSY);
+
+	/* update transaction status */
+
+	if (operation_result == I2C_CHANNEL_OPERATION_SUCCEEDED)
+		result = true;
+
+	if (result && (!payload->write))
+		process_channel_reply(dce_i2c_hw, payload);
+
+	return result;
+}
+
+bool dce_i2c_submit_command_hw(
+	struct resource_pool *pool,
+	struct ddc *ddc,
+	struct i2c_command *cmd,
+	struct dce_i2c_hw *dce_i2c_hw)
+{
+	uint8_t index_of_payload = 0;
+	bool result;
+
+	set_speed(dce_i2c_hw, cmd->speed);
+
+	result = true;
+
+	while (index_of_payload < cmd->number_of_payloads) {
+		bool mot = (index_of_payload != cmd->number_of_payloads - 1);
+
+		struct i2c_payload *payload = cmd->payloads + index_of_payload;
+
+		if (!dce_i2c_hw_engine_submit_payload(
+				dce_i2c_hw, payload, mot, cmd->speed)) {
+			result = false;
+			break;
+		}
+
+		++index_of_payload;
+	}
+
+	pool->i2c_hw_buffer_in_use = false;
+
+	release_engine(dce_i2c_hw);
+	dal_ddc_close(dce_i2c_hw->ddc);
+
+	dce_i2c_hw->ddc = NULL;
+
+	return result;
+}
+
+void dce_i2c_hw_construct(
+	struct dce_i2c_hw *dce_i2c_hw,
+	struct dc_context *ctx,
+	uint32_t engine_id,
+	const struct dce_i2c_registers *regs,
+	const struct dce_i2c_shift *shifts,
+	const struct dce_i2c_mask *masks)
+{
+	dce_i2c_hw->ctx = ctx;
+	dce_i2c_hw->engine_id = engine_id;
+	dce_i2c_hw->reference_frequency = (ctx->dc_bios->fw_info.pll_info.crystal_frequency) >> 1;
+	dce_i2c_hw->regs = regs;
+	dce_i2c_hw->shifts = shifts;
+	dce_i2c_hw->masks = masks;
+	dce_i2c_hw->buffer_used_bytes = 0;
+	dce_i2c_hw->transaction_count = 0;
+	dce_i2c_hw->engine_keep_power_up_count = 1;
+	dce_i2c_hw->default_speed = DEFAULT_I2C_HW_SPEED;
+	dce_i2c_hw->send_reset_length = 0;
+	dce_i2c_hw->setup_limit = I2C_SETUP_TIME_LIMIT_DCE;
+	dce_i2c_hw->buffer_size = I2C_HW_BUFFER_SIZE_DCE;
+}
+
+void dce100_i2c_hw_construct(
+	struct dce_i2c_hw *dce_i2c_hw,
+	struct dc_context *ctx,
+	uint32_t engine_id,
+	const struct dce_i2c_registers *regs,
+	const struct dce_i2c_shift *shifts,
+	const struct dce_i2c_mask *masks)
+{
+	dce_i2c_hw_construct(dce_i2c_hw,
+			ctx,
+			engine_id,
+			regs,
+			shifts,
+			masks);
+	dce_i2c_hw->buffer_size = I2C_HW_BUFFER_SIZE_DCE100;
+}
+
+void dce112_i2c_hw_construct(
+	struct dce_i2c_hw *dce_i2c_hw,
+	struct dc_context *ctx,
+	uint32_t engine_id,
+	const struct dce_i2c_registers *regs,
+	const struct dce_i2c_shift *shifts,
+	const struct dce_i2c_mask *masks)
+{
+	dce100_i2c_hw_construct(dce_i2c_hw,
+			ctx,
+			engine_id,
+			regs,
+			shifts,
+			masks);
+	dce_i2c_hw->default_speed = DEFAULT_I2C_HW_SPEED_100KHZ;
+}
+
+void dcn1_i2c_hw_construct(
+	struct dce_i2c_hw *dce_i2c_hw,
+	struct dc_context *ctx,
+	uint32_t engine_id,
+	const struct dce_i2c_registers *regs,
+	const struct dce_i2c_shift *shifts,
+	const struct dce_i2c_mask *masks)
+{
+	dce112_i2c_hw_construct(dce_i2c_hw,
+			ctx,
+			engine_id,
+			regs,
+			shifts,
+			masks);
+	dce_i2c_hw->setup_limit = I2C_SETUP_TIME_LIMIT_DCN;
+}
+
+void dcn2_i2c_hw_construct(
+	struct dce_i2c_hw *dce_i2c_hw,
+	struct dc_context *ctx,
+	uint32_t engine_id,
+	const struct dce_i2c_registers *regs,
+	const struct dce_i2c_shift *shifts,
+	const struct dce_i2c_mask *masks)
+{
+	dcn1_i2c_hw_construct(dce_i2c_hw,
+			ctx,
+			engine_id,
+			regs,
+			shifts,
+			masks);
+	dce_i2c_hw->send_reset_length = I2C_SEND_RESET_LENGTH_9;
+	if (ctx->dc->debug.scl_reset_length10)
+		dce_i2c_hw->send_reset_length = I2C_SEND_RESET_LENGTH_10;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_i2c_hw.h b/drivers/gpu/drm/amd/display/dc/dce/dce_i2c_hw.h
new file mode 100644
index 000000000000..a9a16f645994
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_i2c_hw.h
@@ -0,0 +1,355 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCE_I2C_HW_H__
+#define __DCE_I2C_HW_H__
+
+enum dc_i2c_status {
+	DC_I2C_STATUS__DC_I2C_STATUS_IDLE,
+	DC_I2C_STATUS__DC_I2C_STATUS_USED_BY_SW,
+	DC_I2C_STATUS__DC_I2C_STATUS_USED_BY_HW,
+	DC_I2C_REG_RW_CNTL_STATUS_DMCU_ONLY = 2,
+};
+
+enum dc_i2c_arbitration {
+	DC_I2C_ARBITRATION__DC_I2C_SW_PRIORITY_NORMAL,
+	DC_I2C_ARBITRATION__DC_I2C_SW_PRIORITY_HIGH
+};
+
+enum i2c_channel_operation_result {
+	I2C_CHANNEL_OPERATION_SUCCEEDED,
+	I2C_CHANNEL_OPERATION_FAILED,
+	I2C_CHANNEL_OPERATION_NOT_GRANTED,
+	I2C_CHANNEL_OPERATION_IS_BUSY,
+	I2C_CHANNEL_OPERATION_NO_HANDLE_PROVIDED,
+	I2C_CHANNEL_OPERATION_CHANNEL_IN_USE,
+	I2C_CHANNEL_OPERATION_CHANNEL_CLIENT_MAX_ALLOWED,
+	I2C_CHANNEL_OPERATION_ENGINE_BUSY,
+	I2C_CHANNEL_OPERATION_TIMEOUT,
+	I2C_CHANNEL_OPERATION_NO_RESPONSE,
+	I2C_CHANNEL_OPERATION_HW_REQUEST_I2C_BUS,
+	I2C_CHANNEL_OPERATION_WRONG_PARAMETER,
+	I2C_CHANNEL_OPERATION_OUT_NB_OF_RETRIES,
+	I2C_CHANNEL_OPERATION_NOT_STARTED
+};
+
+
+enum dce_i2c_transaction_action {
+	DCE_I2C_TRANSACTION_ACTION_I2C_WRITE = 0x00,
+	DCE_I2C_TRANSACTION_ACTION_I2C_READ = 0x10,
+	DCE_I2C_TRANSACTION_ACTION_I2C_STATUS_REQUEST = 0x20,
+
+	DCE_I2C_TRANSACTION_ACTION_I2C_WRITE_MOT = 0x40,
+	DCE_I2C_TRANSACTION_ACTION_I2C_READ_MOT = 0x50,
+	DCE_I2C_TRANSACTION_ACTION_I2C_STATUS_REQUEST_MOT = 0x60,
+
+	DCE_I2C_TRANSACTION_ACTION_DP_WRITE = 0x80,
+	DCE_I2C_TRANSACTION_ACTION_DP_READ = 0x90
+};
+
+enum {
+	I2C_SETUP_TIME_LIMIT_DCE = 255,
+	I2C_SETUP_TIME_LIMIT_DCN = 3,
+	I2C_HW_BUFFER_SIZE_DCE100 = 538,
+	I2C_HW_BUFFER_SIZE_DCE = 144,
+	I2C_SEND_RESET_LENGTH_9 = 9,
+	I2C_SEND_RESET_LENGTH_10 = 10,
+	DEFAULT_I2C_HW_SPEED = 50,
+	DEFAULT_I2C_HW_SPEED_100KHZ = 100,
+	TRANSACTION_TIMEOUT_IN_I2C_CLOCKS = 32,
+};
+
+#define I2C_HW_ENGINE_COMMON_REG_LIST(id)\
+	SRI(SETUP, DC_I2C_DDC, id),\
+	SRI(SPEED, DC_I2C_DDC, id),\
+	SRI(HW_STATUS, DC_I2C_DDC, id),\
+	SR(DC_I2C_ARBITRATION),\
+	SR(DC_I2C_CONTROL),\
+	SR(DC_I2C_SW_STATUS),\
+	SR(DC_I2C_TRANSACTION0),\
+	SR(DC_I2C_TRANSACTION1),\
+	SR(DC_I2C_TRANSACTION2),\
+	SR(DC_I2C_TRANSACTION3),\
+	SR(DC_I2C_DATA),\
+	SR(MICROSECOND_TIME_BASE_DIV)
+
+#define I2C_HW_ENGINE_COMMON_REG_LIST_DCN30(id)\
+	I2C_HW_ENGINE_COMMON_REG_LIST(id),\
+	SR(DIO_MEM_PWR_CTRL),\
+	SR(DIO_MEM_PWR_STATUS)
+
+#define I2C_SF(reg_name, field_name, post_fix)\
+	.field_name = reg_name ## __ ## field_name ## post_fix
+
+#define I2C_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(mask_sh)\
+	I2C_SF(DC_I2C_DDC1_SETUP, DC_I2C_DDC1_ENABLE, mask_sh),\
+	I2C_SF(DC_I2C_DDC1_SETUP, DC_I2C_DDC1_TIME_LIMIT, mask_sh),\
+	I2C_SF(DC_I2C_DDC1_SETUP, DC_I2C_DDC1_DATA_DRIVE_EN, mask_sh),\
+	I2C_SF(DC_I2C_DDC1_SETUP, DC_I2C_DDC1_CLK_DRIVE_EN, mask_sh),\
+	I2C_SF(DC_I2C_DDC1_SETUP, DC_I2C_DDC1_DATA_DRIVE_SEL, mask_sh),\
+	I2C_SF(DC_I2C_DDC1_SETUP, DC_I2C_DDC1_INTRA_TRANSACTION_DELAY, mask_sh),\
+	I2C_SF(DC_I2C_DDC1_SETUP, DC_I2C_DDC1_INTRA_BYTE_DELAY, mask_sh),\
+	I2C_SF(DC_I2C_DDC1_HW_STATUS, DC_I2C_DDC1_HW_STATUS, mask_sh),\
+	I2C_SF(DC_I2C_ARBITRATION, DC_I2C_SW_USE_I2C_REG_REQ, mask_sh),\
+	I2C_SF(DC_I2C_ARBITRATION, DC_I2C_SW_DONE_USING_I2C_REG, mask_sh),\
+	I2C_SF(DC_I2C_ARBITRATION, DC_I2C_NO_QUEUED_SW_GO, mask_sh),\
+	I2C_SF(DC_I2C_ARBITRATION, DC_I2C_SW_PRIORITY, mask_sh),\
+	I2C_SF(DC_I2C_CONTROL, DC_I2C_SOFT_RESET, mask_sh),\
+	I2C_SF(DC_I2C_CONTROL, DC_I2C_SW_STATUS_RESET, mask_sh),\
+	I2C_SF(DC_I2C_CONTROL, DC_I2C_GO, mask_sh),\
+	I2C_SF(DC_I2C_CONTROL, DC_I2C_SEND_RESET, mask_sh),\
+	I2C_SF(DC_I2C_CONTROL, DC_I2C_TRANSACTION_COUNT, mask_sh),\
+	I2C_SF(DC_I2C_CONTROL, DC_I2C_DDC_SELECT, mask_sh),\
+	I2C_SF(DC_I2C_DDC1_SPEED, DC_I2C_DDC1_PRESCALE, mask_sh),\
+	I2C_SF(DC_I2C_DDC1_SPEED, DC_I2C_DDC1_THRESHOLD, mask_sh),\
+	I2C_SF(DC_I2C_SW_STATUS, DC_I2C_SW_STOPPED_ON_NACK, mask_sh),\
+	I2C_SF(DC_I2C_SW_STATUS, DC_I2C_SW_TIMEOUT, mask_sh),\
+	I2C_SF(DC_I2C_SW_STATUS, DC_I2C_SW_ABORTED, mask_sh),\
+	I2C_SF(DC_I2C_SW_STATUS, DC_I2C_SW_DONE, mask_sh),\
+	I2C_SF(DC_I2C_SW_STATUS, DC_I2C_SW_STATUS, mask_sh),\
+	I2C_SF(DC_I2C_TRANSACTION0, DC_I2C_STOP_ON_NACK0, mask_sh),\
+	I2C_SF(DC_I2C_TRANSACTION0, DC_I2C_START0, mask_sh),\
+	I2C_SF(DC_I2C_TRANSACTION0, DC_I2C_RW0, mask_sh),\
+	I2C_SF(DC_I2C_TRANSACTION0, DC_I2C_STOP0, mask_sh),\
+	I2C_SF(DC_I2C_TRANSACTION0, DC_I2C_COUNT0, mask_sh),\
+	I2C_SF(DC_I2C_DATA, DC_I2C_DATA_RW, mask_sh),\
+	I2C_SF(DC_I2C_DATA, DC_I2C_DATA, mask_sh),\
+	I2C_SF(DC_I2C_DATA, DC_I2C_INDEX, mask_sh),\
+	I2C_SF(DC_I2C_DATA, DC_I2C_INDEX_WRITE, mask_sh),\
+	I2C_SF(MICROSECOND_TIME_BASE_DIV, XTAL_REF_DIV, mask_sh),\
+	I2C_SF(MICROSECOND_TIME_BASE_DIV, MICROSECOND_TIME_BASE_DIV, mask_sh),\
+	I2C_SF(DC_I2C_ARBITRATION, DC_I2C_REG_RW_CNTL_STATUS, mask_sh)
+
+#define I2C_COMMON_MASK_SH_LIST_DCE110(mask_sh)\
+	I2C_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(mask_sh),\
+	I2C_SF(DC_I2C_DDC1_SPEED, DC_I2C_DDC1_START_STOP_TIMING_CNTL, mask_sh)
+
+struct dce_i2c_shift {
+	uint8_t DC_I2C_DDC1_ENABLE;
+	uint8_t DC_I2C_DDC1_TIME_LIMIT;
+	uint8_t DC_I2C_DDC1_DATA_DRIVE_EN;
+	uint8_t DC_I2C_DDC1_CLK_DRIVE_EN;
+	uint8_t DC_I2C_DDC1_DATA_DRIVE_SEL;
+	uint8_t DC_I2C_DDC1_INTRA_TRANSACTION_DELAY;
+	uint8_t DC_I2C_DDC1_INTRA_BYTE_DELAY;
+	uint8_t DC_I2C_DDC1_HW_STATUS;
+	uint8_t DC_I2C_SW_DONE_USING_I2C_REG;
+	uint8_t DC_I2C_SW_USE_I2C_REG_REQ;
+	uint8_t DC_I2C_NO_QUEUED_SW_GO;
+	uint8_t DC_I2C_SW_PRIORITY;
+	uint8_t DC_I2C_SOFT_RESET;
+	uint8_t DC_I2C_SW_STATUS_RESET;
+	uint8_t DC_I2C_GO;
+	uint8_t DC_I2C_SEND_RESET;
+	uint8_t DC_I2C_TRANSACTION_COUNT;
+	uint8_t DC_I2C_DDC_SELECT;
+	uint8_t DC_I2C_DDC1_PRESCALE;
+	uint8_t DC_I2C_DDC1_THRESHOLD;
+	uint8_t DC_I2C_DDC1_START_STOP_TIMING_CNTL;
+	uint8_t DC_I2C_SW_STOPPED_ON_NACK;
+	uint8_t DC_I2C_SW_TIMEOUT;
+	uint8_t DC_I2C_SW_ABORTED;
+	uint8_t DC_I2C_SW_DONE;
+	uint8_t DC_I2C_SW_STATUS;
+	uint8_t DC_I2C_STOP_ON_NACK0;
+	uint8_t DC_I2C_START0;
+	uint8_t DC_I2C_RW0;
+	uint8_t DC_I2C_STOP0;
+	uint8_t DC_I2C_COUNT0;
+	uint8_t DC_I2C_DATA_RW;
+	uint8_t DC_I2C_DATA;
+	uint8_t DC_I2C_INDEX;
+	uint8_t DC_I2C_INDEX_WRITE;
+	uint8_t XTAL_REF_DIV;
+	uint8_t MICROSECOND_TIME_BASE_DIV;
+	uint8_t DC_I2C_DDC1_SEND_RESET_LENGTH;
+	uint8_t DC_I2C_REG_RW_CNTL_STATUS;
+	uint8_t I2C_LIGHT_SLEEP_FORCE;
+	uint8_t I2C_MEM_PWR_STATE;
+	uint8_t DC_I2C_DDC1_CLK_EN;
+};
+
+struct dce_i2c_mask {
+	uint32_t DC_I2C_DDC1_ENABLE;
+	uint32_t DC_I2C_DDC1_TIME_LIMIT;
+	uint32_t DC_I2C_DDC1_DATA_DRIVE_EN;
+	uint32_t DC_I2C_DDC1_CLK_DRIVE_EN;
+	uint32_t DC_I2C_DDC1_DATA_DRIVE_SEL;
+	uint32_t DC_I2C_DDC1_INTRA_TRANSACTION_DELAY;
+	uint32_t DC_I2C_DDC1_INTRA_BYTE_DELAY;
+	uint32_t DC_I2C_DDC1_HW_STATUS;
+	uint32_t DC_I2C_SW_DONE_USING_I2C_REG;
+	uint32_t DC_I2C_SW_USE_I2C_REG_REQ;
+	uint32_t DC_I2C_NO_QUEUED_SW_GO;
+	uint32_t DC_I2C_SW_PRIORITY;
+	uint32_t DC_I2C_SOFT_RESET;
+	uint32_t DC_I2C_SW_STATUS_RESET;
+	uint32_t DC_I2C_GO;
+	uint32_t DC_I2C_SEND_RESET;
+	uint32_t DC_I2C_TRANSACTION_COUNT;
+	uint32_t DC_I2C_DDC_SELECT;
+	uint32_t DC_I2C_DDC1_PRESCALE;
+	uint32_t DC_I2C_DDC1_THRESHOLD;
+	uint32_t DC_I2C_DDC1_START_STOP_TIMING_CNTL;
+	uint32_t DC_I2C_SW_STOPPED_ON_NACK;
+	uint32_t DC_I2C_SW_TIMEOUT;
+	uint32_t DC_I2C_SW_ABORTED;
+	uint32_t DC_I2C_SW_DONE;
+	uint32_t DC_I2C_SW_STATUS;
+	uint32_t DC_I2C_STOP_ON_NACK0;
+	uint32_t DC_I2C_START0;
+	uint32_t DC_I2C_RW0;
+	uint32_t DC_I2C_STOP0;
+	uint32_t DC_I2C_COUNT0;
+	uint32_t DC_I2C_DATA_RW;
+	uint32_t DC_I2C_DATA;
+	uint32_t DC_I2C_INDEX;
+	uint32_t DC_I2C_INDEX_WRITE;
+	uint32_t XTAL_REF_DIV;
+	uint32_t MICROSECOND_TIME_BASE_DIV;
+	uint32_t DC_I2C_DDC1_SEND_RESET_LENGTH;
+	uint32_t DC_I2C_REG_RW_CNTL_STATUS;
+	uint32_t I2C_LIGHT_SLEEP_FORCE;
+	uint32_t I2C_MEM_PWR_STATE;
+	uint32_t DC_I2C_DDC1_CLK_EN;
+};
+
+#define I2C_COMMON_MASK_SH_LIST_DCN2(mask_sh)\
+	I2C_COMMON_MASK_SH_LIST_DCE110(mask_sh),\
+	I2C_SF(DC_I2C_DDC1_SETUP, DC_I2C_DDC1_SEND_RESET_LENGTH, mask_sh)
+
+#define I2C_COMMON_MASK_SH_LIST_DCN30(mask_sh)\
+	I2C_COMMON_MASK_SH_LIST_DCN2(mask_sh),\
+	I2C_SF(DIO_MEM_PWR_CTRL, I2C_LIGHT_SLEEP_FORCE, mask_sh),\
+	I2C_SF(DIO_MEM_PWR_STATUS, I2C_MEM_PWR_STATE, mask_sh)
+
+#define I2C_COMMON_MASK_SH_LIST_DCN35(mask_sh)\
+	I2C_COMMON_MASK_SH_LIST_DCN30(mask_sh),\
+	I2C_SF(DC_I2C_DDC1_SETUP, DC_I2C_DDC1_CLK_EN, mask_sh)
+
+#define I2C_COMMON_MASK_SH_LIST_DCN401(mask_sh)\
+	I2C_COMMON_MASK_SH_LIST_DCN30(mask_sh),\
+	I2C_SF(DC_I2C_DDC1_SETUP, DC_I2C_DDC1_CLK_EN, mask_sh)
+
+struct dce_i2c_registers {
+	uint32_t SETUP;
+	uint32_t SPEED;
+	uint32_t HW_STATUS;
+	uint32_t DC_I2C_ARBITRATION;
+	uint32_t DC_I2C_CONTROL;
+	uint32_t DC_I2C_SW_STATUS;
+	uint32_t DC_I2C_TRANSACTION0;
+	uint32_t DC_I2C_TRANSACTION1;
+	uint32_t DC_I2C_TRANSACTION2;
+	uint32_t DC_I2C_TRANSACTION3;
+	uint32_t DC_I2C_DATA;
+	uint32_t MICROSECOND_TIME_BASE_DIV;
+	uint32_t DIO_MEM_PWR_CTRL;
+	uint32_t DIO_MEM_PWR_STATUS;
+};
+
+enum dce_i2c_transaction_address_space {
+	DCE_I2C_TRANSACTION_ADDRESS_SPACE_I2C = 1,
+	DCE_I2C_TRANSACTION_ADDRESS_SPACE_DPCD
+};
+
+struct i2c_request_transaction_data {
+	enum dce_i2c_transaction_action action;
+	enum i2c_channel_operation_result status;
+	uint8_t address;
+	uint32_t length;
+	uint8_t *data;
+};
+
+struct dce_i2c_hw {
+	struct ddc *ddc;
+	uint32_t engine_keep_power_up_count;
+	uint32_t transaction_count;
+	uint32_t buffer_used_bytes;
+	uint32_t buffer_used_write;
+	uint32_t reference_frequency;
+	uint32_t default_speed;
+	uint32_t engine_id;
+	uint32_t setup_limit;
+	uint32_t send_reset_length;
+	uint32_t buffer_size;
+	struct dc_context *ctx;
+
+	const struct dce_i2c_registers *regs;
+	const struct dce_i2c_shift *shifts;
+	const struct dce_i2c_mask *masks;
+};
+
+void dce_i2c_hw_construct(
+	struct dce_i2c_hw *dce_i2c_hw,
+	struct dc_context *ctx,
+	uint32_t engine_id,
+	const struct dce_i2c_registers *regs,
+	const struct dce_i2c_shift *shifts,
+	const struct dce_i2c_mask *masks);
+
+void dce100_i2c_hw_construct(
+	struct dce_i2c_hw *dce_i2c_hw,
+	struct dc_context *ctx,
+	uint32_t engine_id,
+	const struct dce_i2c_registers *regs,
+	const struct dce_i2c_shift *shifts,
+	const struct dce_i2c_mask *masks);
+
+void dce112_i2c_hw_construct(
+	struct dce_i2c_hw *dce_i2c_hw,
+	struct dc_context *ctx,
+	uint32_t engine_id,
+	const struct dce_i2c_registers *regs,
+	const struct dce_i2c_shift *shifts,
+	const struct dce_i2c_mask *masks);
+
+void dcn1_i2c_hw_construct(
+	struct dce_i2c_hw *dce_i2c_hw,
+	struct dc_context *ctx,
+	uint32_t engine_id,
+	const struct dce_i2c_registers *regs,
+	const struct dce_i2c_shift *shifts,
+	const struct dce_i2c_mask *masks);
+
+void dcn2_i2c_hw_construct(
+	struct dce_i2c_hw *dce_i2c_hw,
+	struct dc_context *ctx,
+	uint32_t engine_id,
+	const struct dce_i2c_registers *regs,
+	const struct dce_i2c_shift *shifts,
+	const struct dce_i2c_mask *masks);
+
+bool dce_i2c_submit_command_hw(
+	struct resource_pool *pool,
+	struct ddc *ddc,
+	struct i2c_command *cmd,
+	struct dce_i2c_hw *dce_i2c_hw);
+
+struct dce_i2c_hw *acquire_i2c_hw_engine(
+	struct resource_pool *pool,
+	struct ddc *ddc);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_i2c_sw.c b/drivers/gpu/drm/amd/display/dc/dce/dce_i2c_sw.c
new file mode 100644
index 000000000000..2d73b94c515c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_i2c_sw.c
@@ -0,0 +1,496 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dce_i2c.h"
+#include "dce_i2c_sw.h"
+#include "include/gpio_service_interface.h"
+#define SCL false
+#define SDA true
+
+void dce_i2c_sw_construct(
+	struct dce_i2c_sw *dce_i2c_sw,
+	struct dc_context *ctx)
+{
+	dce_i2c_sw->ctx = ctx;
+}
+
+static inline bool read_bit_from_ddc(
+	struct ddc *ddc,
+	bool data_nor_clock)
+{
+	uint32_t value = 0;
+
+	if (data_nor_clock)
+		dal_gpio_get_value(ddc->pin_data, &value);
+	else
+		dal_gpio_get_value(ddc->pin_clock, &value);
+
+	return (value != 0);
+}
+
+static inline void write_bit_to_ddc(
+	struct ddc *ddc,
+	bool data_nor_clock,
+	bool bit)
+{
+	uint32_t value = bit ? 1 : 0;
+
+	if (data_nor_clock)
+		dal_gpio_set_value(ddc->pin_data, value);
+	else
+		dal_gpio_set_value(ddc->pin_clock, value);
+}
+
+static void release_engine_dce_sw(
+	struct resource_pool *pool,
+	struct dce_i2c_sw *dce_i2c_sw)
+{
+	dal_ddc_close(dce_i2c_sw->ddc);
+	dce_i2c_sw->ddc = NULL;
+}
+
+static bool wait_for_scl_high_sw(
+	struct dc_context *ctx,
+	struct ddc *ddc,
+	uint16_t clock_delay_div_4)
+{
+	uint32_t scl_retry = 0;
+	uint32_t scl_retry_max = I2C_SW_TIMEOUT_DELAY / clock_delay_div_4;
+
+	udelay(clock_delay_div_4);
+
+	do {
+		if (read_bit_from_ddc(ddc, SCL))
+			return true;
+
+		udelay(clock_delay_div_4);
+
+		++scl_retry;
+	} while (scl_retry <= scl_retry_max);
+
+	return false;
+}
+static bool write_byte_sw(
+	struct dc_context *ctx,
+	struct ddc *ddc_handle,
+	uint16_t clock_delay_div_4,
+	uint8_t byte)
+{
+	int32_t shift = 7;
+	bool ack;
+
+	/* bits are transmitted serially, starting from MSB */
+
+	do {
+		udelay(clock_delay_div_4);
+
+		write_bit_to_ddc(ddc_handle, SDA, (byte >> shift) & 1);
+
+		udelay(clock_delay_div_4);
+
+		write_bit_to_ddc(ddc_handle, SCL, true);
+
+		if (!wait_for_scl_high_sw(ctx, ddc_handle, clock_delay_div_4))
+			return false;
+
+		write_bit_to_ddc(ddc_handle, SCL, false);
+
+		--shift;
+	} while (shift >= 0);
+
+	/* The display sends ACK by preventing the SDA from going high
+	 * after the SCL pulse we use to send our last data bit.
+	 * If the SDA goes high after that bit, it's a NACK
+	 */
+
+	udelay(clock_delay_div_4);
+
+	write_bit_to_ddc(ddc_handle, SDA, true);
+
+	udelay(clock_delay_div_4);
+
+	write_bit_to_ddc(ddc_handle, SCL, true);
+
+	if (!wait_for_scl_high_sw(ctx, ddc_handle, clock_delay_div_4))
+		return false;
+
+	/* read ACK bit */
+
+	ack = !read_bit_from_ddc(ddc_handle, SDA);
+
+	udelay(clock_delay_div_4 << 1);
+
+	write_bit_to_ddc(ddc_handle, SCL, false);
+
+	udelay(clock_delay_div_4 << 1);
+
+	return ack;
+}
+
+static bool read_byte_sw(
+	struct dc_context *ctx,
+	struct ddc *ddc_handle,
+	uint16_t clock_delay_div_4,
+	uint8_t *byte,
+	bool more)
+{
+	int32_t shift = 7;
+
+	uint8_t data = 0;
+
+	/* The data bits are read from MSB to LSB;
+	 * bit is read while SCL is high
+	 */
+
+	do {
+		write_bit_to_ddc(ddc_handle, SCL, true);
+
+		if (!wait_for_scl_high_sw(ctx, ddc_handle, clock_delay_div_4))
+			return false;
+
+		if (read_bit_from_ddc(ddc_handle, SDA))
+			data |= (1 << shift);
+
+		write_bit_to_ddc(ddc_handle, SCL, false);
+
+		udelay(clock_delay_div_4 << 1);
+
+		--shift;
+	} while (shift >= 0);
+
+	/* read only whole byte */
+
+	*byte = data;
+
+	udelay(clock_delay_div_4);
+
+	/* send the acknowledge bit:
+	 * SDA low means ACK, SDA high means NACK
+	 */
+
+	write_bit_to_ddc(ddc_handle, SDA, !more);
+
+	udelay(clock_delay_div_4);
+
+	write_bit_to_ddc(ddc_handle, SCL, true);
+
+	if (!wait_for_scl_high_sw(ctx, ddc_handle, clock_delay_div_4))
+		return false;
+
+	write_bit_to_ddc(ddc_handle, SCL, false);
+
+	udelay(clock_delay_div_4);
+
+	write_bit_to_ddc(ddc_handle, SDA, true);
+
+	udelay(clock_delay_div_4);
+
+	return true;
+}
+static bool stop_sync_sw(
+	struct dc_context *ctx,
+	struct ddc *ddc_handle,
+	uint16_t clock_delay_div_4)
+{
+	uint32_t retry = 0;
+
+	/* The I2C communications stop signal is:
+	 * the SDA going high from low, while the SCL is high.
+	 */
+
+	write_bit_to_ddc(ddc_handle, SCL, false);
+
+	udelay(clock_delay_div_4);
+
+	write_bit_to_ddc(ddc_handle, SDA, false);
+
+	udelay(clock_delay_div_4);
+
+	write_bit_to_ddc(ddc_handle, SCL, true);
+
+	if (!wait_for_scl_high_sw(ctx, ddc_handle, clock_delay_div_4))
+		return false;
+
+	write_bit_to_ddc(ddc_handle, SDA, true);
+
+	do {
+		udelay(clock_delay_div_4);
+
+		if (read_bit_from_ddc(ddc_handle, SDA))
+			return true;
+
+		++retry;
+	} while (retry <= 2);
+
+	return false;
+}
+static bool i2c_write_sw(
+	struct dc_context *ctx,
+	struct ddc *ddc_handle,
+	uint16_t clock_delay_div_4,
+	uint8_t address,
+	uint32_t length,
+	const uint8_t *data)
+{
+	uint32_t i = 0;
+
+	if (!write_byte_sw(ctx, ddc_handle, clock_delay_div_4, address))
+		return false;
+
+	while (i < length) {
+		if (!write_byte_sw(ctx, ddc_handle, clock_delay_div_4, data[i]))
+			return false;
+		++i;
+	}
+
+	return true;
+}
+
+static bool i2c_read_sw(
+	struct dc_context *ctx,
+	struct ddc *ddc_handle,
+	uint16_t clock_delay_div_4,
+	uint8_t address,
+	uint32_t length,
+	uint8_t *data)
+{
+	uint32_t i = 0;
+
+	if (!write_byte_sw(ctx, ddc_handle, clock_delay_div_4, address))
+		return false;
+
+	while (i < length) {
+		if (!read_byte_sw(ctx, ddc_handle, clock_delay_div_4, data + i,
+			i < length - 1))
+			return false;
+		++i;
+	}
+
+	return true;
+}
+
+
+
+static bool start_sync_sw(
+	struct dc_context *ctx,
+	struct ddc *ddc_handle,
+	uint16_t clock_delay_div_4)
+{
+	uint32_t retry = 0;
+
+	/* The I2C communications start signal is:
+	 * the SDA going low from high, while the SCL is high.
+	 */
+
+	write_bit_to_ddc(ddc_handle, SCL, true);
+
+	udelay(clock_delay_div_4);
+
+	do {
+		write_bit_to_ddc(ddc_handle, SDA, true);
+
+		if (!read_bit_from_ddc(ddc_handle, SDA)) {
+			++retry;
+			continue;
+		}
+
+		udelay(clock_delay_div_4);
+
+		write_bit_to_ddc(ddc_handle, SCL, true);
+
+		if (!wait_for_scl_high_sw(ctx, ddc_handle, clock_delay_div_4))
+			break;
+
+		write_bit_to_ddc(ddc_handle, SDA, false);
+
+		udelay(clock_delay_div_4);
+
+		write_bit_to_ddc(ddc_handle, SCL, false);
+
+		udelay(clock_delay_div_4);
+
+		return true;
+	} while (retry <= I2C_SW_RETRIES);
+
+	return false;
+}
+
+static void dce_i2c_sw_engine_set_speed(
+	struct dce_i2c_sw *engine,
+	uint32_t speed)
+{
+	ASSERT(speed);
+
+	engine->speed = speed ? speed : DCE_I2C_DEFAULT_I2C_SW_SPEED;
+
+	engine->clock_delay = 1000 / engine->speed;
+
+	if (engine->clock_delay < 12)
+		engine->clock_delay = 12;
+}
+
+static bool dce_i2c_sw_engine_acquire_engine(
+	struct dce_i2c_sw *engine,
+	struct ddc *ddc)
+{
+	enum gpio_result result;
+
+	result = dal_ddc_open(ddc, GPIO_MODE_FAST_OUTPUT,
+		GPIO_DDC_CONFIG_TYPE_MODE_I2C);
+
+	if (result != GPIO_RESULT_OK)
+		return false;
+
+	engine->ddc = ddc;
+
+	return true;
+}
+
+bool dce_i2c_engine_acquire_sw(
+	struct dce_i2c_sw *dce_i2c_sw,
+	struct ddc *ddc_handle)
+{
+	uint32_t counter = 0;
+	bool result;
+
+	do {
+
+		result = dce_i2c_sw_engine_acquire_engine(
+				dce_i2c_sw, ddc_handle);
+
+		if (result)
+			break;
+
+		/* i2c_engine is busy by VBios, lets wait and retry */
+
+		udelay(10);
+
+		++counter;
+	} while (counter < 2);
+
+	return result;
+}
+
+static void dce_i2c_sw_engine_submit_channel_request(struct dce_i2c_sw *engine,
+						     struct i2c_request_transaction_data *req)
+{
+	struct ddc *ddc = engine->ddc;
+	uint16_t clock_delay_div_4 = engine->clock_delay >> 2;
+
+	/* send sync (start / repeated start) */
+
+	bool result = start_sync_sw(engine->ctx, ddc, clock_delay_div_4);
+
+	/* process payload */
+
+	if (result) {
+		switch (req->action) {
+		case DCE_I2C_TRANSACTION_ACTION_I2C_WRITE:
+		case DCE_I2C_TRANSACTION_ACTION_I2C_WRITE_MOT:
+			result = i2c_write_sw(engine->ctx, ddc, clock_delay_div_4,
+				req->address, req->length, req->data);
+		break;
+		case DCE_I2C_TRANSACTION_ACTION_I2C_READ:
+		case DCE_I2C_TRANSACTION_ACTION_I2C_READ_MOT:
+			result = i2c_read_sw(engine->ctx, ddc, clock_delay_div_4,
+				req->address, req->length, req->data);
+		break;
+		default:
+			result = false;
+		break;
+		}
+	}
+
+	/* send stop if not 'mot' or operation failed */
+
+	if (!result ||
+		(req->action == DCE_I2C_TRANSACTION_ACTION_I2C_WRITE) ||
+		(req->action == DCE_I2C_TRANSACTION_ACTION_I2C_READ))
+		if (!stop_sync_sw(engine->ctx, ddc, clock_delay_div_4))
+			result = false;
+
+	req->status = result ?
+		I2C_CHANNEL_OPERATION_SUCCEEDED :
+		I2C_CHANNEL_OPERATION_FAILED;
+}
+
+static bool dce_i2c_sw_engine_submit_payload(struct dce_i2c_sw *engine,
+					     struct i2c_payload *payload,
+					     bool middle_of_transaction)
+{
+	struct i2c_request_transaction_data request;
+
+	if (!payload->write)
+		request.action = middle_of_transaction ?
+			DCE_I2C_TRANSACTION_ACTION_I2C_READ_MOT :
+			DCE_I2C_TRANSACTION_ACTION_I2C_READ;
+	else
+		request.action = middle_of_transaction ?
+			DCE_I2C_TRANSACTION_ACTION_I2C_WRITE_MOT :
+			DCE_I2C_TRANSACTION_ACTION_I2C_WRITE;
+
+	request.address = (uint8_t) ((payload->address << 1) | (payload->write ? 0 : 1));
+	request.length = payload->length;
+	request.data = payload->data;
+
+	dce_i2c_sw_engine_submit_channel_request(engine, &request);
+
+	if ((request.status == I2C_CHANNEL_OPERATION_ENGINE_BUSY) ||
+		(request.status == I2C_CHANNEL_OPERATION_FAILED))
+		return false;
+
+	return true;
+}
+bool dce_i2c_submit_command_sw(
+	struct resource_pool *pool,
+	struct ddc *ddc,
+	struct i2c_command *cmd,
+	struct dce_i2c_sw *dce_i2c_sw)
+{
+	uint8_t index_of_payload = 0;
+	bool result;
+
+	dce_i2c_sw_engine_set_speed(dce_i2c_sw, cmd->speed);
+
+	result = true;
+
+	while (index_of_payload < cmd->number_of_payloads) {
+		bool mot = (index_of_payload != cmd->number_of_payloads - 1);
+
+		struct i2c_payload *payload = cmd->payloads + index_of_payload;
+
+		if (!dce_i2c_sw_engine_submit_payload(
+			dce_i2c_sw, payload, mot)) {
+			result = false;
+			break;
+		}
+
+		++index_of_payload;
+	}
+
+	release_engine_dce_sw(pool, dce_i2c_sw);
+
+	return result;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_i2c_sw.h b/drivers/gpu/drm/amd/display/dc/dce/dce_i2c_sw.h
new file mode 100644
index 000000000000..019fc47bb767
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_i2c_sw.h
@@ -0,0 +1,57 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCE_I2C_SW_H__
+#define __DCE_I2C_SW_H__
+
+enum {
+	DCE_I2C_DEFAULT_I2C_SW_SPEED = 50,
+	I2C_SW_RETRIES = 10,
+	I2C_SW_TIMEOUT_DELAY = 3000,
+};
+
+struct dce_i2c_sw {
+	struct ddc *ddc;
+	struct dc_context *ctx;
+	uint32_t clock_delay;
+	uint32_t speed;
+};
+
+void dce_i2c_sw_construct(
+	struct dce_i2c_sw *dce_i2c_sw,
+	struct dc_context *ctx);
+
+bool dce_i2c_submit_command_sw(
+	struct resource_pool *pool,
+	struct ddc *ddc,
+	struct i2c_command *cmd,
+	struct dce_i2c_sw *dce_i2c_sw);
+
+bool dce_i2c_engine_acquire_sw(
+	struct dce_i2c_sw *dce_i2c_sw,
+	struct ddc *ddc_handle);
+
+#endif
+
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_ipp.c b/drivers/gpu/drm/amd/display/dc/dce/dce_ipp.c
new file mode 100644
index 000000000000..34bff9aef66c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_ipp.c
@@ -0,0 +1,311 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dce_ipp.h"
+#include "reg_helper.h"
+#include "dm_services.h"
+
+#define REG(reg) \
+	(ipp_dce->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	ipp_dce->ipp_shift->field_name, ipp_dce->ipp_mask->field_name
+
+#define CTX \
+	ipp_dce->base.ctx
+
+
+static void dce_ipp_cursor_set_position(
+	struct input_pixel_processor *ipp,
+	const struct dc_cursor_position *position,
+	const struct dc_cursor_mi_param *param)
+{
+	struct dce_ipp *ipp_dce = TO_DCE_IPP(ipp);
+
+	/* lock cursor registers */
+	REG_UPDATE(CUR_UPDATE, CURSOR_UPDATE_LOCK, true);
+
+	/* Flag passed in structure differentiates cursor enable/disable. */
+	/* Update if it differs from cached state. */
+	REG_UPDATE(CUR_CONTROL, CURSOR_EN, position->enable);
+
+	REG_SET_2(CUR_POSITION, 0,
+		CURSOR_X_POSITION, position->x,
+		CURSOR_Y_POSITION, position->y);
+
+	REG_SET_2(CUR_HOT_SPOT, 0,
+		CURSOR_HOT_SPOT_X, position->x_hotspot,
+		CURSOR_HOT_SPOT_Y, position->y_hotspot);
+
+	/* unlock cursor registers */
+	REG_UPDATE(CUR_UPDATE, CURSOR_UPDATE_LOCK, false);
+}
+
+static void dce_ipp_cursor_set_attributes(
+	struct input_pixel_processor *ipp,
+	const struct dc_cursor_attributes *attributes)
+{
+	struct dce_ipp *ipp_dce = TO_DCE_IPP(ipp);
+	int mode;
+
+	/* Lock cursor registers */
+	REG_UPDATE(CUR_UPDATE, CURSOR_UPDATE_LOCK, true);
+
+	/* Program cursor control */
+	switch (attributes->color_format) {
+	case CURSOR_MODE_MONO:
+		mode = 0;
+		break;
+	case CURSOR_MODE_COLOR_1BIT_AND:
+		mode = 1;
+		break;
+	case CURSOR_MODE_COLOR_PRE_MULTIPLIED_ALPHA:
+		mode = 2;
+		break;
+	case CURSOR_MODE_COLOR_UN_PRE_MULTIPLIED_ALPHA:
+		mode = 3;
+		break;
+	default:
+		BREAK_TO_DEBUGGER(); /* unsupported */
+		mode = 0;
+	}
+
+	REG_UPDATE_3(CUR_CONTROL,
+		CURSOR_MODE, mode,
+		CURSOR_2X_MAGNIFY, attributes->attribute_flags.bits.ENABLE_MAGNIFICATION,
+		CUR_INV_TRANS_CLAMP, attributes->attribute_flags.bits.INVERSE_TRANSPARENT_CLAMPING);
+
+	if (attributes->color_format == CURSOR_MODE_MONO) {
+		REG_SET_3(CUR_COLOR1, 0,
+			CUR_COLOR1_BLUE, 0,
+			CUR_COLOR1_GREEN, 0,
+			CUR_COLOR1_RED, 0);
+
+		REG_SET_3(CUR_COLOR2, 0,
+			CUR_COLOR2_BLUE, 0xff,
+			CUR_COLOR2_GREEN, 0xff,
+			CUR_COLOR2_RED, 0xff);
+	}
+
+	/*
+	 * Program cursor size -- NOTE: HW spec specifies that HW register
+	 * stores size as (height - 1, width - 1)
+	 */
+	REG_SET_2(CUR_SIZE, 0,
+		CURSOR_WIDTH, attributes->width-1,
+		CURSOR_HEIGHT, attributes->height-1);
+
+	/* Program cursor surface address */
+	/* SURFACE_ADDRESS_HIGH: Higher order bits (39:32) of hardware cursor
+	 * surface base address in byte. It is 4K byte aligned.
+	 * The correct way to program cursor surface address is to first write
+	 * to CUR_SURFACE_ADDRESS_HIGH, and then write to CUR_SURFACE_ADDRESS
+	 */
+	REG_SET(CUR_SURFACE_ADDRESS_HIGH, 0,
+		CURSOR_SURFACE_ADDRESS_HIGH, attributes->address.high_part);
+
+	REG_SET(CUR_SURFACE_ADDRESS, 0,
+		CURSOR_SURFACE_ADDRESS, attributes->address.low_part);
+
+	/* Unlock Cursor registers. */
+	REG_UPDATE(CUR_UPDATE, CURSOR_UPDATE_LOCK, false);
+}
+
+
+static void dce_ipp_program_prescale(struct input_pixel_processor *ipp,
+				     struct ipp_prescale_params *params)
+{
+	struct dce_ipp *ipp_dce = TO_DCE_IPP(ipp);
+
+	/* set to bypass mode first before change */
+	REG_UPDATE(PRESCALE_GRPH_CONTROL,
+		   GRPH_PRESCALE_BYPASS, 1);
+
+	REG_SET_2(PRESCALE_VALUES_GRPH_R, 0,
+		  GRPH_PRESCALE_SCALE_R, params->scale,
+		  GRPH_PRESCALE_BIAS_R, params->bias);
+
+	REG_SET_2(PRESCALE_VALUES_GRPH_G, 0,
+		  GRPH_PRESCALE_SCALE_G, params->scale,
+		  GRPH_PRESCALE_BIAS_G, params->bias);
+
+	REG_SET_2(PRESCALE_VALUES_GRPH_B, 0,
+		  GRPH_PRESCALE_SCALE_B, params->scale,
+		  GRPH_PRESCALE_BIAS_B, params->bias);
+
+	if (params->mode != IPP_PRESCALE_MODE_BYPASS) {
+		REG_UPDATE(PRESCALE_GRPH_CONTROL,
+			   GRPH_PRESCALE_BYPASS, 0);
+
+		/* If prescale is in use, then legacy lut should be bypassed */
+		REG_UPDATE(INPUT_GAMMA_CONTROL,
+			   GRPH_INPUT_GAMMA_MODE, 1);
+	}
+}
+
+static void dce_ipp_program_input_lut(
+	struct input_pixel_processor *ipp,
+	const struct dc_gamma *gamma)
+{
+	int i;
+	struct dce_ipp *ipp_dce = TO_DCE_IPP(ipp);
+
+	/* power on LUT memory */
+	if (REG(DCFE_MEM_PWR_CTRL))
+		REG_SET(DCFE_MEM_PWR_CTRL, 0, DCP_LUT_MEM_PWR_DIS, 1);
+
+	/* enable all */
+	REG_SET(DC_LUT_WRITE_EN_MASK, 0, DC_LUT_WRITE_EN_MASK, 0x7);
+
+	/* 256 entry mode */
+	REG_UPDATE(DC_LUT_RW_MODE, DC_LUT_RW_MODE, 0);
+
+	/* LUT-256, unsigned, integer, new u0.12 format */
+	REG_SET_3(DC_LUT_CONTROL, 0,
+		DC_LUT_DATA_R_FORMAT, 3,
+		DC_LUT_DATA_G_FORMAT, 3,
+		DC_LUT_DATA_B_FORMAT, 3);
+
+	/* start from index 0 */
+	REG_SET(DC_LUT_RW_INDEX, 0,
+		DC_LUT_RW_INDEX, 0);
+
+	for (i = 0; i < gamma->num_entries; i++) {
+		REG_SET(DC_LUT_SEQ_COLOR, 0, DC_LUT_SEQ_COLOR,
+				dc_fixpt_round(
+					gamma->entries.red[i]));
+		REG_SET(DC_LUT_SEQ_COLOR, 0, DC_LUT_SEQ_COLOR,
+				dc_fixpt_round(
+					gamma->entries.green[i]));
+		REG_SET(DC_LUT_SEQ_COLOR, 0, DC_LUT_SEQ_COLOR,
+				dc_fixpt_round(
+					gamma->entries.blue[i]));
+	}
+
+	/* power off LUT memory */
+	if (REG(DCFE_MEM_PWR_CTRL))
+		REG_SET(DCFE_MEM_PWR_CTRL, 0, DCP_LUT_MEM_PWR_DIS, 0);
+
+	/* bypass prescale, enable legacy LUT */
+	REG_UPDATE(PRESCALE_GRPH_CONTROL, GRPH_PRESCALE_BYPASS, 1);
+	REG_UPDATE(INPUT_GAMMA_CONTROL, GRPH_INPUT_GAMMA_MODE, 0);
+}
+
+static void dce_ipp_set_degamma(
+	struct input_pixel_processor *ipp,
+	enum ipp_degamma_mode mode)
+{
+	struct dce_ipp *ipp_dce = TO_DCE_IPP(ipp);
+	uint32_t degamma_type = (mode == IPP_DEGAMMA_MODE_HW_sRGB) ? 1 : 0;
+
+	ASSERT(mode == IPP_DEGAMMA_MODE_BYPASS || mode == IPP_DEGAMMA_MODE_HW_sRGB);
+
+	REG_SET_3(DEGAMMA_CONTROL, 0,
+		  GRPH_DEGAMMA_MODE, degamma_type,
+		  CURSOR_DEGAMMA_MODE, degamma_type,
+		  CURSOR2_DEGAMMA_MODE, degamma_type);
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static void dce60_ipp_set_degamma(
+	struct input_pixel_processor *ipp,
+	enum ipp_degamma_mode mode)
+{
+	struct dce_ipp *ipp_dce = TO_DCE_IPP(ipp);
+	uint32_t degamma_type = (mode == IPP_DEGAMMA_MODE_HW_sRGB) ? 1 : 0;
+
+	ASSERT(mode == IPP_DEGAMMA_MODE_BYPASS || mode == IPP_DEGAMMA_MODE_HW_sRGB);
+	/* DCE6 does not have CURSOR2_DEGAMMA_MODE bit in DEGAMMA_CONTROL reg */
+	REG_SET_2(DEGAMMA_CONTROL, 0,
+		  GRPH_DEGAMMA_MODE, degamma_type,
+		  CURSOR_DEGAMMA_MODE, degamma_type);
+}
+#endif
+
+static const struct ipp_funcs dce_ipp_funcs = {
+	.ipp_cursor_set_attributes = dce_ipp_cursor_set_attributes,
+	.ipp_cursor_set_position = dce_ipp_cursor_set_position,
+	.ipp_program_prescale = dce_ipp_program_prescale,
+	.ipp_program_input_lut = dce_ipp_program_input_lut,
+	.ipp_set_degamma = dce_ipp_set_degamma
+};
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static const struct ipp_funcs dce60_ipp_funcs = {
+	.ipp_cursor_set_attributes = dce_ipp_cursor_set_attributes,
+	.ipp_cursor_set_position = dce_ipp_cursor_set_position,
+	.ipp_program_prescale = dce_ipp_program_prescale,
+	.ipp_program_input_lut = dce_ipp_program_input_lut,
+	.ipp_set_degamma = dce60_ipp_set_degamma
+};
+#endif
+
+
+/*****************************************/
+/* Constructor, Destructor               */
+/*****************************************/
+
+void dce_ipp_construct(
+	struct dce_ipp *ipp_dce,
+	struct dc_context *ctx,
+	int inst,
+	const struct dce_ipp_registers *regs,
+	const struct dce_ipp_shift *ipp_shift,
+	const struct dce_ipp_mask *ipp_mask)
+{
+	ipp_dce->base.ctx = ctx;
+	ipp_dce->base.inst = inst;
+	ipp_dce->base.funcs = &dce_ipp_funcs;
+
+	ipp_dce->regs = regs;
+	ipp_dce->ipp_shift = ipp_shift;
+	ipp_dce->ipp_mask = ipp_mask;
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+void dce60_ipp_construct(
+	struct dce_ipp *ipp_dce,
+	struct dc_context *ctx,
+	int inst,
+	const struct dce_ipp_registers *regs,
+	const struct dce_ipp_shift *ipp_shift,
+	const struct dce_ipp_mask *ipp_mask)
+{
+	ipp_dce->base.ctx = ctx;
+	ipp_dce->base.inst = inst;
+	ipp_dce->base.funcs = &dce60_ipp_funcs;
+
+	ipp_dce->regs = regs;
+	ipp_dce->ipp_shift = ipp_shift;
+	ipp_dce->ipp_mask = ipp_mask;
+}
+#endif
+
+void dce_ipp_destroy(struct input_pixel_processor **ipp)
+{
+	kfree(TO_DCE_IPP(*ipp));
+	*ipp = NULL;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_ipp.h b/drivers/gpu/drm/amd/display/dc/dce/dce_ipp.h
new file mode 100644
index 000000000000..0028d4bdd81b
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_ipp.h
@@ -0,0 +1,287 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef _DCE_IPP_H_
+#define _DCE_IPP_H_
+
+#include "ipp.h"
+
+#define TO_DCE_IPP(ipp)\
+	container_of(ipp, struct dce_ipp, base)
+
+#define IPP_COMMON_REG_LIST_DCE_BASE(id) \
+	SRI(CUR_UPDATE, DCP, id), \
+	SRI(CUR_CONTROL, DCP, id), \
+	SRI(CUR_POSITION, DCP, id), \
+	SRI(CUR_HOT_SPOT, DCP, id), \
+	SRI(CUR_COLOR1, DCP, id), \
+	SRI(CUR_COLOR2, DCP, id), \
+	SRI(CUR_SIZE, DCP, id), \
+	SRI(CUR_SURFACE_ADDRESS_HIGH, DCP, id), \
+	SRI(CUR_SURFACE_ADDRESS, DCP, id), \
+	SRI(PRESCALE_GRPH_CONTROL, DCP, id), \
+	SRI(PRESCALE_VALUES_GRPH_R, DCP, id), \
+	SRI(PRESCALE_VALUES_GRPH_G, DCP, id), \
+	SRI(PRESCALE_VALUES_GRPH_B, DCP, id), \
+	SRI(INPUT_GAMMA_CONTROL, DCP, id), \
+	SRI(DC_LUT_WRITE_EN_MASK, DCP, id), \
+	SRI(DC_LUT_RW_MODE, DCP, id), \
+	SRI(DC_LUT_CONTROL, DCP, id), \
+	SRI(DC_LUT_RW_INDEX, DCP, id), \
+	SRI(DC_LUT_SEQ_COLOR, DCP, id), \
+	SRI(DEGAMMA_CONTROL, DCP, id)
+
+#define IPP_DCE100_REG_LIST_DCE_BASE(id) \
+	IPP_COMMON_REG_LIST_DCE_BASE(id), \
+	SRI(DCFE_MEM_PWR_CTRL, CRTC, id)
+
+#define IPP_DCE110_REG_LIST_DCE_BASE(id) \
+	IPP_COMMON_REG_LIST_DCE_BASE(id), \
+	SRI(DCFE_MEM_PWR_CTRL, DCFE, id)
+
+#define IPP_SF(reg_name, field_name, post_fix)\
+	.field_name = reg_name ## __ ## field_name ## post_fix
+
+#define IPP_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(mask_sh) \
+	IPP_SF(CUR_UPDATE, CURSOR_UPDATE_LOCK, mask_sh), \
+	IPP_SF(CUR_CONTROL, CURSOR_EN, mask_sh), \
+	IPP_SF(CUR_CONTROL, CURSOR_MODE, mask_sh), \
+	IPP_SF(CUR_CONTROL, CURSOR_2X_MAGNIFY, mask_sh), \
+	IPP_SF(CUR_CONTROL, CUR_INV_TRANS_CLAMP, mask_sh), \
+	IPP_SF(CUR_POSITION, CURSOR_X_POSITION, mask_sh), \
+	IPP_SF(CUR_POSITION, CURSOR_Y_POSITION, mask_sh), \
+	IPP_SF(CUR_HOT_SPOT, CURSOR_HOT_SPOT_X, mask_sh), \
+	IPP_SF(CUR_HOT_SPOT, CURSOR_HOT_SPOT_Y, mask_sh), \
+	IPP_SF(CUR_COLOR1, CUR_COLOR1_BLUE, mask_sh), \
+	IPP_SF(CUR_COLOR1, CUR_COLOR1_GREEN, mask_sh), \
+	IPP_SF(CUR_COLOR1, CUR_COLOR1_RED, mask_sh), \
+	IPP_SF(CUR_COLOR2, CUR_COLOR2_BLUE, mask_sh), \
+	IPP_SF(CUR_COLOR2, CUR_COLOR2_GREEN, mask_sh), \
+	IPP_SF(CUR_COLOR2, CUR_COLOR2_RED, mask_sh), \
+	IPP_SF(CUR_SIZE, CURSOR_WIDTH, mask_sh), \
+	IPP_SF(CUR_SIZE, CURSOR_HEIGHT, mask_sh), \
+	IPP_SF(CUR_SURFACE_ADDRESS_HIGH, CURSOR_SURFACE_ADDRESS_HIGH, mask_sh), \
+	IPP_SF(CUR_SURFACE_ADDRESS, CURSOR_SURFACE_ADDRESS, mask_sh), \
+	IPP_SF(PRESCALE_GRPH_CONTROL, GRPH_PRESCALE_BYPASS, mask_sh), \
+	IPP_SF(PRESCALE_VALUES_GRPH_R, GRPH_PRESCALE_SCALE_R, mask_sh), \
+	IPP_SF(PRESCALE_VALUES_GRPH_R, GRPH_PRESCALE_BIAS_R, mask_sh), \
+	IPP_SF(PRESCALE_VALUES_GRPH_G, GRPH_PRESCALE_SCALE_G, mask_sh), \
+	IPP_SF(PRESCALE_VALUES_GRPH_G, GRPH_PRESCALE_BIAS_G, mask_sh), \
+	IPP_SF(PRESCALE_VALUES_GRPH_B, GRPH_PRESCALE_SCALE_B, mask_sh), \
+	IPP_SF(PRESCALE_VALUES_GRPH_B, GRPH_PRESCALE_BIAS_B, mask_sh), \
+	IPP_SF(INPUT_GAMMA_CONTROL, GRPH_INPUT_GAMMA_MODE, mask_sh), \
+	IPP_SF(DC_LUT_WRITE_EN_MASK, DC_LUT_WRITE_EN_MASK, mask_sh), \
+	IPP_SF(DC_LUT_RW_MODE, DC_LUT_RW_MODE, mask_sh), \
+	IPP_SF(DC_LUT_CONTROL, DC_LUT_DATA_R_FORMAT, mask_sh), \
+	IPP_SF(DC_LUT_CONTROL, DC_LUT_DATA_G_FORMAT, mask_sh), \
+	IPP_SF(DC_LUT_CONTROL, DC_LUT_DATA_B_FORMAT, mask_sh), \
+	IPP_SF(DC_LUT_RW_INDEX, DC_LUT_RW_INDEX, mask_sh), \
+	IPP_SF(DC_LUT_SEQ_COLOR, DC_LUT_SEQ_COLOR, mask_sh), \
+	IPP_SF(DEGAMMA_CONTROL, GRPH_DEGAMMA_MODE, mask_sh), \
+	IPP_SF(DEGAMMA_CONTROL, CURSOR_DEGAMMA_MODE, mask_sh), \
+	IPP_SF(DEGAMMA_CONTROL, CURSOR2_DEGAMMA_MODE, mask_sh)
+
+#define IPP_DCE100_MASK_SH_LIST_DCE_COMMON_BASE(mask_sh) \
+	IPP_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(mask_sh), \
+	IPP_SF(DCFE_MEM_PWR_CTRL, DCP_LUT_MEM_PWR_DIS, mask_sh)
+
+#define IPP_DCE120_MASK_SH_LIST_SOC_BASE(mask_sh) \
+	IPP_SF(DCP0_CUR_UPDATE, CURSOR_UPDATE_LOCK, mask_sh), \
+	IPP_SF(DCP0_CUR_CONTROL, CURSOR_EN, mask_sh), \
+	IPP_SF(DCP0_CUR_CONTROL, CURSOR_MODE, mask_sh), \
+	IPP_SF(DCP0_CUR_CONTROL, CURSOR_2X_MAGNIFY, mask_sh), \
+	IPP_SF(DCP0_CUR_CONTROL, CUR_INV_TRANS_CLAMP, mask_sh), \
+	IPP_SF(DCP0_CUR_POSITION, CURSOR_X_POSITION, mask_sh), \
+	IPP_SF(DCP0_CUR_POSITION, CURSOR_Y_POSITION, mask_sh), \
+	IPP_SF(DCP0_CUR_HOT_SPOT, CURSOR_HOT_SPOT_X, mask_sh), \
+	IPP_SF(DCP0_CUR_HOT_SPOT, CURSOR_HOT_SPOT_Y, mask_sh), \
+	IPP_SF(DCP0_CUR_COLOR1, CUR_COLOR1_BLUE, mask_sh), \
+	IPP_SF(DCP0_CUR_COLOR1, CUR_COLOR1_GREEN, mask_sh), \
+	IPP_SF(DCP0_CUR_COLOR1, CUR_COLOR1_RED, mask_sh), \
+	IPP_SF(DCP0_CUR_COLOR2, CUR_COLOR2_BLUE, mask_sh), \
+	IPP_SF(DCP0_CUR_COLOR2, CUR_COLOR2_GREEN, mask_sh), \
+	IPP_SF(DCP0_CUR_COLOR2, CUR_COLOR2_RED, mask_sh), \
+	IPP_SF(DCP0_CUR_SIZE, CURSOR_WIDTH, mask_sh), \
+	IPP_SF(DCP0_CUR_SIZE, CURSOR_HEIGHT, mask_sh), \
+	IPP_SF(DCP0_CUR_SURFACE_ADDRESS_HIGH, CURSOR_SURFACE_ADDRESS_HIGH, mask_sh), \
+	IPP_SF(DCP0_CUR_SURFACE_ADDRESS, CURSOR_SURFACE_ADDRESS, mask_sh), \
+	IPP_SF(DCP0_PRESCALE_GRPH_CONTROL, GRPH_PRESCALE_BYPASS, mask_sh), \
+	IPP_SF(DCP0_PRESCALE_VALUES_GRPH_R, GRPH_PRESCALE_SCALE_R, mask_sh), \
+	IPP_SF(DCP0_PRESCALE_VALUES_GRPH_R, GRPH_PRESCALE_BIAS_R, mask_sh), \
+	IPP_SF(DCP0_PRESCALE_VALUES_GRPH_G, GRPH_PRESCALE_SCALE_G, mask_sh), \
+	IPP_SF(DCP0_PRESCALE_VALUES_GRPH_G, GRPH_PRESCALE_BIAS_G, mask_sh), \
+	IPP_SF(DCP0_PRESCALE_VALUES_GRPH_B, GRPH_PRESCALE_SCALE_B, mask_sh), \
+	IPP_SF(DCP0_PRESCALE_VALUES_GRPH_B, GRPH_PRESCALE_BIAS_B, mask_sh), \
+	IPP_SF(DCP0_INPUT_GAMMA_CONTROL, GRPH_INPUT_GAMMA_MODE, mask_sh), \
+	IPP_SF(DCFE0_DCFE_MEM_PWR_CTRL, DCP_LUT_MEM_PWR_DIS, mask_sh), \
+	IPP_SF(DCP0_DC_LUT_WRITE_EN_MASK, DC_LUT_WRITE_EN_MASK, mask_sh), \
+	IPP_SF(DCP0_DC_LUT_RW_MODE, DC_LUT_RW_MODE, mask_sh), \
+	IPP_SF(DCP0_DC_LUT_CONTROL, DC_LUT_DATA_R_FORMAT, mask_sh), \
+	IPP_SF(DCP0_DC_LUT_CONTROL, DC_LUT_DATA_G_FORMAT, mask_sh), \
+	IPP_SF(DCP0_DC_LUT_CONTROL, DC_LUT_DATA_B_FORMAT, mask_sh), \
+	IPP_SF(DCP0_DC_LUT_RW_INDEX, DC_LUT_RW_INDEX, mask_sh), \
+	IPP_SF(DCP0_DC_LUT_SEQ_COLOR, DC_LUT_SEQ_COLOR, mask_sh), \
+	IPP_SF(DCP0_DEGAMMA_CONTROL, GRPH_DEGAMMA_MODE, mask_sh), \
+	IPP_SF(DCP0_DEGAMMA_CONTROL, CURSOR_DEGAMMA_MODE, mask_sh), \
+	IPP_SF(DCP0_DEGAMMA_CONTROL, CURSOR2_DEGAMMA_MODE, mask_sh)
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+#define IPP_DCE60_MASK_SH_LIST_DCE_COMMON_BASE(mask_sh) \
+	IPP_SF(CUR_UPDATE, CURSOR_UPDATE_LOCK, mask_sh), \
+	IPP_SF(CUR_CONTROL, CURSOR_EN, mask_sh), \
+	IPP_SF(CUR_CONTROL, CURSOR_MODE, mask_sh), \
+	IPP_SF(CUR_CONTROL, CURSOR_2X_MAGNIFY, mask_sh), \
+	IPP_SF(CUR_CONTROL, CUR_INV_TRANS_CLAMP, mask_sh), \
+	IPP_SF(CUR_POSITION, CURSOR_X_POSITION, mask_sh), \
+	IPP_SF(CUR_POSITION, CURSOR_Y_POSITION, mask_sh), \
+	IPP_SF(CUR_HOT_SPOT, CURSOR_HOT_SPOT_X, mask_sh), \
+	IPP_SF(CUR_HOT_SPOT, CURSOR_HOT_SPOT_Y, mask_sh), \
+	IPP_SF(CUR_COLOR1, CUR_COLOR1_BLUE, mask_sh), \
+	IPP_SF(CUR_COLOR1, CUR_COLOR1_GREEN, mask_sh), \
+	IPP_SF(CUR_COLOR1, CUR_COLOR1_RED, mask_sh), \
+	IPP_SF(CUR_COLOR2, CUR_COLOR2_BLUE, mask_sh), \
+	IPP_SF(CUR_COLOR2, CUR_COLOR2_GREEN, mask_sh), \
+	IPP_SF(CUR_COLOR2, CUR_COLOR2_RED, mask_sh), \
+	IPP_SF(CUR_SIZE, CURSOR_WIDTH, mask_sh), \
+	IPP_SF(CUR_SIZE, CURSOR_HEIGHT, mask_sh), \
+	IPP_SF(CUR_SURFACE_ADDRESS_HIGH, CURSOR_SURFACE_ADDRESS_HIGH, mask_sh), \
+	IPP_SF(CUR_SURFACE_ADDRESS, CURSOR_SURFACE_ADDRESS, mask_sh), \
+	IPP_SF(PRESCALE_GRPH_CONTROL, GRPH_PRESCALE_BYPASS, mask_sh), \
+	IPP_SF(PRESCALE_VALUES_GRPH_R, GRPH_PRESCALE_SCALE_R, mask_sh), \
+	IPP_SF(PRESCALE_VALUES_GRPH_R, GRPH_PRESCALE_BIAS_R, mask_sh), \
+	IPP_SF(PRESCALE_VALUES_GRPH_G, GRPH_PRESCALE_SCALE_G, mask_sh), \
+	IPP_SF(PRESCALE_VALUES_GRPH_G, GRPH_PRESCALE_BIAS_G, mask_sh), \
+	IPP_SF(PRESCALE_VALUES_GRPH_B, GRPH_PRESCALE_SCALE_B, mask_sh), \
+	IPP_SF(PRESCALE_VALUES_GRPH_B, GRPH_PRESCALE_BIAS_B, mask_sh), \
+	IPP_SF(INPUT_GAMMA_CONTROL, GRPH_INPUT_GAMMA_MODE, mask_sh), \
+	IPP_SF(DC_LUT_WRITE_EN_MASK, DC_LUT_WRITE_EN_MASK, mask_sh), \
+	IPP_SF(DC_LUT_RW_MODE, DC_LUT_RW_MODE, mask_sh), \
+	IPP_SF(DC_LUT_CONTROL, DC_LUT_DATA_R_FORMAT, mask_sh), \
+	IPP_SF(DC_LUT_CONTROL, DC_LUT_DATA_G_FORMAT, mask_sh), \
+	IPP_SF(DC_LUT_CONTROL, DC_LUT_DATA_B_FORMAT, mask_sh), \
+	IPP_SF(DC_LUT_RW_INDEX, DC_LUT_RW_INDEX, mask_sh), \
+	IPP_SF(DC_LUT_SEQ_COLOR, DC_LUT_SEQ_COLOR, mask_sh), \
+	IPP_SF(DEGAMMA_CONTROL, GRPH_DEGAMMA_MODE, mask_sh), \
+	IPP_SF(DEGAMMA_CONTROL, CURSOR_DEGAMMA_MODE, mask_sh)
+#endif
+
+#define IPP_REG_FIELD_LIST(type) \
+	type CURSOR_UPDATE_LOCK; \
+	type CURSOR_EN; \
+	type CURSOR_X_POSITION; \
+	type CURSOR_Y_POSITION; \
+	type CURSOR_HOT_SPOT_X; \
+	type CURSOR_HOT_SPOT_Y; \
+	type CURSOR_MODE; \
+	type CURSOR_2X_MAGNIFY; \
+	type CUR_INV_TRANS_CLAMP; \
+	type CUR_COLOR1_BLUE; \
+	type CUR_COLOR1_GREEN; \
+	type CUR_COLOR1_RED; \
+	type CUR_COLOR2_BLUE; \
+	type CUR_COLOR2_GREEN; \
+	type CUR_COLOR2_RED; \
+	type CURSOR_WIDTH; \
+	type CURSOR_HEIGHT; \
+	type CURSOR_SURFACE_ADDRESS_HIGH; \
+	type CURSOR_SURFACE_ADDRESS; \
+	type GRPH_PRESCALE_BYPASS; \
+	type GRPH_PRESCALE_SCALE_R; \
+	type GRPH_PRESCALE_BIAS_R; \
+	type GRPH_PRESCALE_SCALE_G; \
+	type GRPH_PRESCALE_BIAS_G; \
+	type GRPH_PRESCALE_SCALE_B; \
+	type GRPH_PRESCALE_BIAS_B; \
+	type GRPH_INPUT_GAMMA_MODE; \
+	type DCP_LUT_MEM_PWR_DIS; \
+	type DC_LUT_WRITE_EN_MASK; \
+	type DC_LUT_RW_MODE; \
+	type DC_LUT_DATA_R_FORMAT; \
+	type DC_LUT_DATA_G_FORMAT; \
+	type DC_LUT_DATA_B_FORMAT; \
+	type DC_LUT_RW_INDEX; \
+	type DC_LUT_SEQ_COLOR; \
+	type GRPH_DEGAMMA_MODE; \
+	type CURSOR_DEGAMMA_MODE; \
+	type CURSOR2_DEGAMMA_MODE
+
+struct dce_ipp_shift {
+	IPP_REG_FIELD_LIST(uint8_t);
+};
+
+struct dce_ipp_mask {
+	IPP_REG_FIELD_LIST(uint32_t);
+};
+
+struct dce_ipp_registers {
+	uint32_t CUR_UPDATE;
+	uint32_t CUR_CONTROL;
+	uint32_t CUR_POSITION;
+	uint32_t CUR_HOT_SPOT;
+	uint32_t CUR_COLOR1;
+	uint32_t CUR_COLOR2;
+	uint32_t CUR_SIZE;
+	uint32_t CUR_SURFACE_ADDRESS_HIGH;
+	uint32_t CUR_SURFACE_ADDRESS;
+	uint32_t PRESCALE_GRPH_CONTROL;
+	uint32_t PRESCALE_VALUES_GRPH_R;
+	uint32_t PRESCALE_VALUES_GRPH_G;
+	uint32_t PRESCALE_VALUES_GRPH_B;
+	uint32_t INPUT_GAMMA_CONTROL;
+	uint32_t DCFE_MEM_PWR_CTRL;
+	uint32_t DC_LUT_WRITE_EN_MASK;
+	uint32_t DC_LUT_RW_MODE;
+	uint32_t DC_LUT_CONTROL;
+	uint32_t DC_LUT_RW_INDEX;
+	uint32_t DC_LUT_SEQ_COLOR;
+	uint32_t DEGAMMA_CONTROL;
+};
+
+struct dce_ipp {
+	struct input_pixel_processor base;
+	const struct dce_ipp_registers *regs;
+	const struct dce_ipp_shift *ipp_shift;
+	const struct dce_ipp_mask *ipp_mask;
+};
+
+void dce_ipp_construct(struct dce_ipp *ipp_dce,
+	struct dc_context *ctx,
+	int inst,
+	const struct dce_ipp_registers *regs,
+	const struct dce_ipp_shift *ipp_shift,
+	const struct dce_ipp_mask *ipp_mask);
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+void dce60_ipp_construct(struct dce_ipp *ipp_dce,
+	struct dc_context *ctx,
+	int inst,
+	const struct dce_ipp_registers *regs,
+	const struct dce_ipp_shift *ipp_shift,
+	const struct dce_ipp_mask *ipp_mask);
+#endif
+
+void dce_ipp_destroy(struct input_pixel_processor **ipp);
+
+#endif /* _DCE_IPP_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.c b/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.c
new file mode 100644
index 000000000000..0c50fe266c8a
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.c
@@ -0,0 +1,1816 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+
+#include "core_types.h"
+#include "link_encoder.h"
+#include "dce_link_encoder.h"
+#include "stream_encoder.h"
+#include "dc_bios_types.h"
+
+#include "gpio_service_interface.h"
+
+#include "dce/dce_11_0_d.h"
+#include "dce/dce_11_0_sh_mask.h"
+#include "dce/dce_11_0_enum.h"
+
+#ifndef DMU_MEM_PWR_CNTL__DMCU_IRAM_MEM_PWR_STATE__SHIFT
+#define DMU_MEM_PWR_CNTL__DMCU_IRAM_MEM_PWR_STATE__SHIFT 0xa
+#endif
+
+#ifndef DMU_MEM_PWR_CNTL__DMCU_IRAM_MEM_PWR_STATE_MASK
+#define DMU_MEM_PWR_CNTL__DMCU_IRAM_MEM_PWR_STATE_MASK 0x00000400L
+#endif
+
+#ifndef HPD0_DC_HPD_CONTROL__DC_HPD_EN_MASK
+#define HPD0_DC_HPD_CONTROL__DC_HPD_EN_MASK  0x10000000L
+#endif
+
+#ifndef HPD0_DC_HPD_CONTROL__DC_HPD_EN__SHIFT
+#define HPD0_DC_HPD_CONTROL__DC_HPD_EN__SHIFT  0x1c
+#endif
+
+#define CTX \
+	enc110->base.ctx
+#define DC_LOGGER \
+	enc110->base.ctx->logger
+
+#define REG(reg)\
+	(enc110->link_regs->reg)
+
+#define AUX_REG(reg)\
+	(enc110->aux_regs->reg)
+
+#define HPD_REG(reg)\
+	(enc110->hpd_regs->reg)
+
+#define DEFAULT_AUX_MAX_DATA_SIZE 16
+#define AUX_MAX_DEFER_WRITE_RETRY 20
+/*
+ * @brief
+ * Trigger Source Select
+ * ASIC-dependent, actual values for register programming
+ */
+#define DCE110_DIG_FE_SOURCE_SELECT_INVALID 0x0
+#define DCE110_DIG_FE_SOURCE_SELECT_DIGA 0x1
+#define DCE110_DIG_FE_SOURCE_SELECT_DIGB 0x2
+#define DCE110_DIG_FE_SOURCE_SELECT_DIGC 0x4
+#define DCE110_DIG_FE_SOURCE_SELECT_DIGD 0x08
+#define DCE110_DIG_FE_SOURCE_SELECT_DIGE 0x10
+#define DCE110_DIG_FE_SOURCE_SELECT_DIGF 0x20
+#define DCE110_DIG_FE_SOURCE_SELECT_DIGG 0x40
+
+enum {
+	DP_MST_UPDATE_MAX_RETRY = 50
+};
+
+#define DIG_REG(reg)\
+	(reg + enc110->offsets.dig)
+
+#define DP_REG(reg)\
+	(reg + enc110->offsets.dp)
+
+static const struct link_encoder_funcs dce110_lnk_enc_funcs = {
+	.validate_output_with_stream =
+		dce110_link_encoder_validate_output_with_stream,
+	.hw_init = dce110_link_encoder_hw_init,
+	.setup = dce110_link_encoder_setup,
+	.enable_tmds_output = dce110_link_encoder_enable_tmds_output,
+	.enable_dp_output = dce110_link_encoder_enable_dp_output,
+	.enable_dp_mst_output = dce110_link_encoder_enable_dp_mst_output,
+	.enable_lvds_output = dce110_link_encoder_enable_lvds_output,
+	.disable_output = dce110_link_encoder_disable_output,
+	.dp_set_lane_settings = dce110_link_encoder_dp_set_lane_settings,
+	.dp_set_phy_pattern = dce110_link_encoder_dp_set_phy_pattern,
+	.update_mst_stream_allocation_table =
+		dce110_link_encoder_update_mst_stream_allocation_table,
+	.psr_program_dp_dphy_fast_training =
+			dce110_psr_program_dp_dphy_fast_training,
+	.psr_program_secondary_packet = dce110_psr_program_secondary_packet,
+	.connect_dig_be_to_fe = dce110_link_encoder_connect_dig_be_to_fe,
+	.enable_hpd = dce110_link_encoder_enable_hpd,
+	.disable_hpd = dce110_link_encoder_disable_hpd,
+	.is_dig_enabled = dce110_is_dig_enabled,
+	.destroy = dce110_link_encoder_destroy,
+	.get_max_link_cap = dce110_link_encoder_get_max_link_cap,
+	.get_dig_frontend = dce110_get_dig_frontend,
+};
+
+static enum bp_result link_transmitter_control(
+	struct dce110_link_encoder *enc110,
+	struct bp_transmitter_control *cntl)
+{
+	enum bp_result result;
+	struct dc_bios *bp = enc110->base.ctx->dc_bios;
+
+	result = bp->funcs->transmitter_control(bp, cntl);
+
+	return result;
+}
+
+static void enable_phy_bypass_mode(
+	struct dce110_link_encoder *enc110,
+	bool enable)
+{
+	/* This register resides in DP back end block;
+	 * transmitter is used for the offset */
+
+	REG_UPDATE(DP_DPHY_CNTL, DPHY_BYPASS, enable);
+
+}
+
+static void disable_prbs_symbols(
+	struct dce110_link_encoder *enc110,
+	bool disable)
+{
+	/* This register resides in DP back end block;
+	 * transmitter is used for the offset */
+
+	REG_UPDATE_4(DP_DPHY_CNTL,
+			DPHY_ATEST_SEL_LANE0, disable,
+			DPHY_ATEST_SEL_LANE1, disable,
+			DPHY_ATEST_SEL_LANE2, disable,
+			DPHY_ATEST_SEL_LANE3, disable);
+}
+
+static void disable_prbs_mode(
+	struct dce110_link_encoder *enc110)
+{
+	REG_UPDATE(DP_DPHY_PRBS_CNTL, DPHY_PRBS_EN, 0);
+}
+
+static void program_pattern_symbols(
+	struct dce110_link_encoder *enc110,
+	uint16_t pattern_symbols[8])
+{
+	/* This register resides in DP back end block;
+	 * transmitter is used for the offset */
+
+	REG_SET_3(DP_DPHY_SYM0, 0,
+			DPHY_SYM1, pattern_symbols[0],
+			DPHY_SYM2, pattern_symbols[1],
+			DPHY_SYM3, pattern_symbols[2]);
+
+	/* This register resides in DP back end block;
+	 * transmitter is used for the offset */
+
+	REG_SET_3(DP_DPHY_SYM1, 0,
+			DPHY_SYM4, pattern_symbols[3],
+			DPHY_SYM5, pattern_symbols[4],
+			DPHY_SYM6, pattern_symbols[5]);
+
+	/* This register resides in DP back end block;
+	 * transmitter is used for the offset */
+
+	REG_SET_2(DP_DPHY_SYM2, 0,
+			DPHY_SYM7, pattern_symbols[6],
+			DPHY_SYM8, pattern_symbols[7]);
+}
+
+static void set_dp_phy_pattern_d102(
+	struct dce110_link_encoder *enc110)
+{
+	/* Disable PHY Bypass mode to setup the test pattern */
+	enable_phy_bypass_mode(enc110, false);
+
+	/* For 10-bit PRBS or debug symbols
+	 * please use the following sequence: */
+
+	/* Enable debug symbols on the lanes */
+
+	disable_prbs_symbols(enc110, true);
+
+	/* Disable PRBS mode */
+	disable_prbs_mode(enc110);
+
+	/* Program debug symbols to be output */
+	{
+		uint16_t pattern_symbols[8] = {
+			0x2AA, 0x2AA, 0x2AA, 0x2AA,
+			0x2AA, 0x2AA, 0x2AA, 0x2AA
+		};
+
+		program_pattern_symbols(enc110, pattern_symbols);
+	}
+
+	/* Enable phy bypass mode to enable the test pattern */
+
+	enable_phy_bypass_mode(enc110, true);
+}
+
+static void set_link_training_complete(
+	struct dce110_link_encoder *enc110,
+	bool complete)
+{
+	/* This register resides in DP back end block;
+	 * transmitter is used for the offset */
+
+	REG_UPDATE(DP_LINK_CNTL, DP_LINK_TRAINING_COMPLETE, complete);
+
+}
+
+unsigned int dce110_get_dig_frontend(struct link_encoder *enc)
+{
+	struct dce110_link_encoder *enc110 = TO_DCE110_LINK_ENC(enc);
+	u32 value;
+	enum engine_id result;
+
+	REG_GET(DIG_BE_CNTL, DIG_FE_SOURCE_SELECT, &value);
+
+	switch (value) {
+	case DCE110_DIG_FE_SOURCE_SELECT_DIGA:
+		result = ENGINE_ID_DIGA;
+		break;
+	case DCE110_DIG_FE_SOURCE_SELECT_DIGB:
+		result = ENGINE_ID_DIGB;
+		break;
+	case DCE110_DIG_FE_SOURCE_SELECT_DIGC:
+		result = ENGINE_ID_DIGC;
+		break;
+	case DCE110_DIG_FE_SOURCE_SELECT_DIGD:
+		result = ENGINE_ID_DIGD;
+		break;
+	case DCE110_DIG_FE_SOURCE_SELECT_DIGE:
+		result = ENGINE_ID_DIGE;
+		break;
+	case DCE110_DIG_FE_SOURCE_SELECT_DIGF:
+		result = ENGINE_ID_DIGF;
+		break;
+	case DCE110_DIG_FE_SOURCE_SELECT_DIGG:
+		result = ENGINE_ID_DIGG;
+		break;
+	default:
+		// invalid source select DIG
+		result = ENGINE_ID_UNKNOWN;
+	}
+
+	return result;
+}
+
+void dce110_link_encoder_set_dp_phy_pattern_training_pattern(
+	struct link_encoder *enc,
+	uint32_t index)
+{
+	struct dce110_link_encoder *enc110 = TO_DCE110_LINK_ENC(enc);
+	/* Write Training Pattern */
+
+	REG_WRITE(DP_DPHY_TRAINING_PATTERN_SEL, index);
+
+	/* Set HW Register Training Complete to false */
+
+	set_link_training_complete(enc110, false);
+
+	/* Disable PHY Bypass mode to output Training Pattern */
+
+	enable_phy_bypass_mode(enc110, false);
+
+	/* Disable PRBS mode */
+	disable_prbs_mode(enc110);
+}
+
+static void setup_panel_mode(
+	struct dce110_link_encoder *enc110,
+	enum dp_panel_mode panel_mode)
+{
+	uint32_t value;
+	struct dc_context *ctx = enc110->base.ctx;
+
+	/* if psp set panel mode, dal should be program it */
+	if (ctx->dc->caps.psp_setup_panel_mode)
+		return;
+
+	ASSERT(REG(DP_DPHY_INTERNAL_CTRL));
+	value = REG_READ(DP_DPHY_INTERNAL_CTRL);
+
+	switch (panel_mode) {
+	case DP_PANEL_MODE_EDP:
+		value = 0x1;
+		break;
+	case DP_PANEL_MODE_SPECIAL:
+		value = 0x11;
+		break;
+	default:
+		value = 0x0;
+		break;
+	}
+
+	REG_WRITE(DP_DPHY_INTERNAL_CTRL, value);
+}
+
+static void set_dp_phy_pattern_symbol_error(
+	struct dce110_link_encoder *enc110)
+{
+	/* Disable PHY Bypass mode to setup the test pattern */
+	enable_phy_bypass_mode(enc110, false);
+
+	/* program correct panel mode*/
+	setup_panel_mode(enc110, DP_PANEL_MODE_DEFAULT);
+
+	/* A PRBS23 pattern is used for most DP electrical measurements. */
+
+	/* Enable PRBS symbols on the lanes */
+	disable_prbs_symbols(enc110, false);
+
+	/* For PRBS23 Set bit DPHY_PRBS_SEL=1 and Set bit DPHY_PRBS_EN=1 */
+	REG_UPDATE_2(DP_DPHY_PRBS_CNTL,
+			DPHY_PRBS_SEL, 1,
+			DPHY_PRBS_EN, 1);
+
+	/* Enable phy bypass mode to enable the test pattern */
+	enable_phy_bypass_mode(enc110, true);
+}
+
+static void set_dp_phy_pattern_prbs7(
+	struct dce110_link_encoder *enc110)
+{
+	/* Disable PHY Bypass mode to setup the test pattern */
+	enable_phy_bypass_mode(enc110, false);
+
+	/* A PRBS7 pattern is used for most DP electrical measurements. */
+
+	/* Enable PRBS symbols on the lanes */
+	disable_prbs_symbols(enc110, false);
+
+	/* For PRBS7 Set bit DPHY_PRBS_SEL=0 and Set bit DPHY_PRBS_EN=1 */
+	REG_UPDATE_2(DP_DPHY_PRBS_CNTL,
+			DPHY_PRBS_SEL, 0,
+			DPHY_PRBS_EN, 1);
+
+	/* Enable phy bypass mode to enable the test pattern */
+	enable_phy_bypass_mode(enc110, true);
+}
+
+static void set_dp_phy_pattern_80bit_custom(
+	struct dce110_link_encoder *enc110,
+	const uint8_t *pattern)
+{
+	/* Disable PHY Bypass mode to setup the test pattern */
+	enable_phy_bypass_mode(enc110, false);
+
+	/* Enable debug symbols on the lanes */
+
+	disable_prbs_symbols(enc110, true);
+
+	/* Enable PHY bypass mode to enable the test pattern */
+	/* TODO is it really needed ? */
+
+	enable_phy_bypass_mode(enc110, true);
+
+	/* Program 80 bit custom pattern */
+	{
+		uint16_t pattern_symbols[8];
+
+		pattern_symbols[0] =
+			((pattern[1] & 0x03) << 8) | pattern[0];
+		pattern_symbols[1] =
+			((pattern[2] & 0x0f) << 6) | ((pattern[1] >> 2) & 0x3f);
+		pattern_symbols[2] =
+			((pattern[3] & 0x3f) << 4) | ((pattern[2] >> 4) & 0x0f);
+		pattern_symbols[3] =
+			(pattern[4] << 2) | ((pattern[3] >> 6) & 0x03);
+		pattern_symbols[4] =
+			((pattern[6] & 0x03) << 8) | pattern[5];
+		pattern_symbols[5] =
+			((pattern[7] & 0x0f) << 6) | ((pattern[6] >> 2) & 0x3f);
+		pattern_symbols[6] =
+			((pattern[8] & 0x3f) << 4) | ((pattern[7] >> 4) & 0x0f);
+		pattern_symbols[7] =
+			(pattern[9] << 2) | ((pattern[8] >> 6) & 0x03);
+
+		program_pattern_symbols(enc110, pattern_symbols);
+	}
+
+	/* Enable phy bypass mode to enable the test pattern */
+
+	enable_phy_bypass_mode(enc110, true);
+}
+
+static void set_dp_phy_pattern_hbr2_compliance_cp2520_2(
+	struct dce110_link_encoder *enc110,
+	unsigned int cp2520_pattern)
+{
+
+	/* previously there is a register DP_HBR2_EYE_PATTERN
+	 * that is enabled to get the pattern.
+	 * But it does not work with the latest spec change,
+	 * so we are programming the following registers manually.
+	 *
+	 * The following settings have been confirmed
+	 * by Nick Chorney and Sandra Liu */
+
+	/* Disable PHY Bypass mode to setup the test pattern */
+
+	enable_phy_bypass_mode(enc110, false);
+
+	/* Setup DIG encoder in DP SST mode */
+	enc110->base.funcs->setup(&enc110->base, SIGNAL_TYPE_DISPLAY_PORT);
+
+	/* ensure normal panel mode. */
+	setup_panel_mode(enc110, DP_PANEL_MODE_DEFAULT);
+
+	/* no vbid after BS (SR)
+	 * DP_LINK_FRAMING_CNTL changed history Sandra Liu
+	 * 11000260 / 11000104 / 110000FC */
+	REG_UPDATE_3(DP_LINK_FRAMING_CNTL,
+			DP_IDLE_BS_INTERVAL, 0xFC,
+			DP_VBID_DISABLE, 1,
+			DP_VID_ENHANCED_FRAME_MODE, 1);
+
+	/* swap every BS with SR */
+	REG_UPDATE(DP_DPHY_SCRAM_CNTL, DPHY_SCRAMBLER_BS_COUNT, 0);
+
+	/* select cp2520 patterns */
+	if (REG(DP_DPHY_HBR2_PATTERN_CONTROL))
+		REG_UPDATE(DP_DPHY_HBR2_PATTERN_CONTROL,
+				DP_DPHY_HBR2_PATTERN_CONTROL, cp2520_pattern);
+	else
+		/* pre-DCE11 can only generate CP2520 pattern 2 */
+		ASSERT(cp2520_pattern == 2);
+
+	/* set link training complete */
+	set_link_training_complete(enc110, true);
+
+	/* disable video stream */
+	REG_UPDATE(DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, 0);
+
+	/* Disable PHY Bypass mode to setup the test pattern */
+	enable_phy_bypass_mode(enc110, false);
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static void dce60_set_dp_phy_pattern_hbr2_compliance_cp2520_2(
+	struct dce110_link_encoder *enc110,
+	unsigned int cp2520_pattern)
+{
+
+	/* previously there is a register DP_HBR2_EYE_PATTERN
+	 * that is enabled to get the pattern.
+	 * But it does not work with the latest spec change,
+	 * so we are programming the following registers manually.
+	 *
+	 * The following settings have been confirmed
+	 * by Nick Chorney and Sandra Liu */
+
+	/* Disable PHY Bypass mode to setup the test pattern */
+
+	enable_phy_bypass_mode(enc110, false);
+
+	/* Setup DIG encoder in DP SST mode */
+	enc110->base.funcs->setup(&enc110->base, SIGNAL_TYPE_DISPLAY_PORT);
+
+	/* ensure normal panel mode. */
+	setup_panel_mode(enc110, DP_PANEL_MODE_DEFAULT);
+
+	/* no vbid after BS (SR)
+	 * DP_LINK_FRAMING_CNTL changed history Sandra Liu
+	 * 11000260 / 11000104 / 110000FC */
+	REG_UPDATE_3(DP_LINK_FRAMING_CNTL,
+			DP_IDLE_BS_INTERVAL, 0xFC,
+			DP_VBID_DISABLE, 1,
+			DP_VID_ENHANCED_FRAME_MODE, 1);
+
+	/* DCE6 has no DP_DPHY_SCRAM_CNTL register, skip swap BS with SR */
+
+	/* select cp2520 patterns */
+	if (REG(DP_DPHY_HBR2_PATTERN_CONTROL))
+		REG_UPDATE(DP_DPHY_HBR2_PATTERN_CONTROL,
+				DP_DPHY_HBR2_PATTERN_CONTROL, cp2520_pattern);
+	else
+		/* pre-DCE11 can only generate CP2520 pattern 2 */
+		ASSERT(cp2520_pattern == 2);
+
+	/* set link training complete */
+	set_link_training_complete(enc110, true);
+
+	/* disable video stream */
+	REG_UPDATE(DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, 0);
+
+	/* Disable PHY Bypass mode to setup the test pattern */
+	enable_phy_bypass_mode(enc110, false);
+}
+#endif
+
+static void set_dp_phy_pattern_passthrough_mode(
+	struct dce110_link_encoder *enc110,
+	enum dp_panel_mode panel_mode)
+{
+	/* program correct panel mode */
+	setup_panel_mode(enc110, panel_mode);
+
+	/* restore LINK_FRAMING_CNTL and DPHY_SCRAMBLER_BS_COUNT
+	 * in case we were doing HBR2 compliance pattern before
+	 */
+	REG_UPDATE_3(DP_LINK_FRAMING_CNTL,
+			DP_IDLE_BS_INTERVAL, 0x2000,
+			DP_VBID_DISABLE, 0,
+			DP_VID_ENHANCED_FRAME_MODE, 1);
+
+	REG_UPDATE(DP_DPHY_SCRAM_CNTL, DPHY_SCRAMBLER_BS_COUNT, 0x1FF);
+
+	/* set link training complete */
+	set_link_training_complete(enc110, true);
+
+	/* Disable PHY Bypass mode to setup the test pattern */
+	enable_phy_bypass_mode(enc110, false);
+
+	/* Disable PRBS mode */
+	disable_prbs_mode(enc110);
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static void dce60_set_dp_phy_pattern_passthrough_mode(
+	struct dce110_link_encoder *enc110,
+	enum dp_panel_mode panel_mode)
+{
+	/* program correct panel mode */
+	setup_panel_mode(enc110, panel_mode);
+
+	/* restore LINK_FRAMING_CNTL
+	 * in case we were doing HBR2 compliance pattern before
+	 */
+	REG_UPDATE_3(DP_LINK_FRAMING_CNTL,
+			DP_IDLE_BS_INTERVAL, 0x2000,
+			DP_VBID_DISABLE, 0,
+			DP_VID_ENHANCED_FRAME_MODE, 1);
+
+	/* DCE6 has no DP_DPHY_SCRAM_CNTL register, skip DPHY_SCRAMBLER_BS_COUNT restore */
+
+	/* set link training complete */
+	set_link_training_complete(enc110, true);
+
+	/* Disable PHY Bypass mode to setup the test pattern */
+	enable_phy_bypass_mode(enc110, false);
+
+	/* Disable PRBS mode */
+	disable_prbs_mode(enc110);
+}
+#endif
+
+/* return value is bit-vector */
+static uint8_t get_frontend_source(
+	enum engine_id engine)
+{
+	switch (engine) {
+	case ENGINE_ID_DIGA:
+		return DCE110_DIG_FE_SOURCE_SELECT_DIGA;
+	case ENGINE_ID_DIGB:
+		return DCE110_DIG_FE_SOURCE_SELECT_DIGB;
+	case ENGINE_ID_DIGC:
+		return DCE110_DIG_FE_SOURCE_SELECT_DIGC;
+	case ENGINE_ID_DIGD:
+		return DCE110_DIG_FE_SOURCE_SELECT_DIGD;
+	case ENGINE_ID_DIGE:
+		return DCE110_DIG_FE_SOURCE_SELECT_DIGE;
+	case ENGINE_ID_DIGF:
+		return DCE110_DIG_FE_SOURCE_SELECT_DIGF;
+	case ENGINE_ID_DIGG:
+		return DCE110_DIG_FE_SOURCE_SELECT_DIGG;
+	default:
+		ASSERT_CRITICAL(false);
+		return DCE110_DIG_FE_SOURCE_SELECT_INVALID;
+	}
+}
+
+static void configure_encoder(
+	struct dce110_link_encoder *enc110,
+	const struct dc_link_settings *link_settings)
+{
+	/* set number of lanes */
+
+	REG_SET(DP_CONFIG, 0,
+			DP_UDI_LANES, link_settings->lane_count - LANE_COUNT_ONE);
+
+	/* setup scrambler */
+	REG_UPDATE(DP_DPHY_SCRAM_CNTL, DPHY_SCRAMBLER_ADVANCE, 1);
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static void dce60_configure_encoder(
+	struct dce110_link_encoder *enc110,
+	const struct dc_link_settings *link_settings)
+{
+	/* set number of lanes */
+
+	REG_SET(DP_CONFIG, 0,
+			DP_UDI_LANES, link_settings->lane_count - LANE_COUNT_ONE);
+
+	/* DCE6 has no DP_DPHY_SCRAM_CNTL register, skip setup scrambler */
+}
+#endif
+
+static void aux_initialize(
+	struct dce110_link_encoder *enc110)
+{
+	struct dc_context *ctx = enc110->base.ctx;
+	enum hpd_source_id hpd_source = enc110->base.hpd_source;
+	uint32_t addr = AUX_REG(AUX_CONTROL);
+	uint32_t value = dm_read_reg(ctx, addr);
+
+	set_reg_field_value(value, hpd_source, AUX_CONTROL, AUX_HPD_SEL);
+	set_reg_field_value(value, 0, AUX_CONTROL, AUX_LS_READ_EN);
+	dm_write_reg(ctx, addr, value);
+
+	addr = AUX_REG(AUX_DPHY_RX_CONTROL0);
+	value = dm_read_reg(ctx, addr);
+
+	/* 1/4 window (the maximum allowed) */
+	set_reg_field_value(value, 1,
+			AUX_DPHY_RX_CONTROL0, AUX_RX_RECEIVE_WINDOW);
+	dm_write_reg(ctx, addr, value);
+
+}
+
+void dce110_psr_program_dp_dphy_fast_training(struct link_encoder *enc,
+			bool exit_link_training_required)
+{
+	struct dce110_link_encoder *enc110 = TO_DCE110_LINK_ENC(enc);
+
+	if (exit_link_training_required)
+		REG_UPDATE(DP_DPHY_FAST_TRAINING,
+				DPHY_RX_FAST_TRAINING_CAPABLE, 1);
+	else {
+		REG_UPDATE(DP_DPHY_FAST_TRAINING,
+				DPHY_RX_FAST_TRAINING_CAPABLE, 0);
+		/*In DCE 11, we are able to pre-program a Force SR register
+		 * to be able to trigger SR symbol after 5 idle patterns
+		 * transmitted. Upon PSR Exit, DMCU can trigger
+		 * DPHY_LOAD_BS_COUNT_START = 1. Upon writing 1 to
+		 * DPHY_LOAD_BS_COUNT_START and the internal counter
+		 * reaches DPHY_LOAD_BS_COUNT, the next BS symbol will be
+		 * replaced by SR symbol once.
+		 */
+
+		REG_UPDATE(DP_DPHY_BS_SR_SWAP_CNTL, DPHY_LOAD_BS_COUNT, 0x5);
+	}
+}
+
+void dce110_psr_program_secondary_packet(struct link_encoder *enc,
+			unsigned int sdp_transmit_line_num_deadline)
+{
+	struct dce110_link_encoder *enc110 = TO_DCE110_LINK_ENC(enc);
+
+	REG_UPDATE_2(DP_SEC_CNTL1,
+		DP_SEC_GSP0_LINE_NUM, sdp_transmit_line_num_deadline,
+		DP_SEC_GSP0_PRIORITY, 1);
+}
+
+bool dce110_is_dig_enabled(struct link_encoder *enc)
+{
+	struct dce110_link_encoder *enc110 = TO_DCE110_LINK_ENC(enc);
+	uint32_t value;
+
+	REG_GET(DIG_BE_EN_CNTL, DIG_ENABLE, &value);
+	return value;
+}
+
+static void link_encoder_disable(struct dce110_link_encoder *enc110)
+{
+	/* reset training pattern */
+	REG_SET(DP_DPHY_TRAINING_PATTERN_SEL, 0,
+			DPHY_TRAINING_PATTERN_SEL, 0);
+
+	/* reset training complete */
+	REG_UPDATE(DP_LINK_CNTL, DP_LINK_TRAINING_COMPLETE, 0);
+
+	/* reset panel mode */
+	setup_panel_mode(enc110, DP_PANEL_MODE_DEFAULT);
+}
+
+static void hpd_initialize(
+	struct dce110_link_encoder *enc110)
+{
+	/* Associate HPD with DIG_BE */
+	enum hpd_source_id hpd_source = enc110->base.hpd_source;
+
+	REG_UPDATE(DIG_BE_CNTL, DIG_HPD_SELECT, hpd_source);
+}
+
+bool dce110_link_encoder_validate_dvi_output(
+	const struct dce110_link_encoder *enc110,
+	enum signal_type connector_signal,
+	enum signal_type signal,
+	const struct dc_crtc_timing *crtc_timing)
+{
+	uint32_t max_pixel_clock = TMDS_MAX_PIXEL_CLOCK;
+
+	if (signal == SIGNAL_TYPE_DVI_DUAL_LINK)
+		max_pixel_clock *= 2;
+
+	/* This handles the case of HDMI downgrade to DVI we don't want to
+	 * we don't want to cap the pixel clock if the DDI is not DVI.
+	 */
+	if (connector_signal != SIGNAL_TYPE_DVI_DUAL_LINK &&
+			connector_signal != SIGNAL_TYPE_DVI_SINGLE_LINK)
+		max_pixel_clock = enc110->base.features.max_hdmi_pixel_clock;
+
+	/* DVI only support RGB pixel encoding */
+	if (crtc_timing->pixel_encoding != PIXEL_ENCODING_RGB)
+		return false;
+
+	/*connect DVI via adpater's HDMI connector*/
+	if ((connector_signal == SIGNAL_TYPE_DVI_SINGLE_LINK ||
+		connector_signal == SIGNAL_TYPE_HDMI_TYPE_A) &&
+		signal != SIGNAL_TYPE_HDMI_TYPE_A &&
+		crtc_timing->pix_clk_100hz > (TMDS_MAX_PIXEL_CLOCK * 10))
+		return false;
+	if (crtc_timing->pix_clk_100hz < (TMDS_MIN_PIXEL_CLOCK * 10))
+		return false;
+
+	if (crtc_timing->pix_clk_100hz > (max_pixel_clock * 10))
+		return false;
+
+	/* DVI supports 6/8bpp single-link and 10/16bpp dual-link */
+	switch (crtc_timing->display_color_depth) {
+	case COLOR_DEPTH_666:
+	case COLOR_DEPTH_888:
+	break;
+	case COLOR_DEPTH_101010:
+	case COLOR_DEPTH_161616:
+		if (signal != SIGNAL_TYPE_DVI_DUAL_LINK)
+			return false;
+	break;
+	default:
+		return false;
+	}
+
+	return true;
+}
+
+static bool dce110_link_encoder_validate_hdmi_output(
+	const struct dce110_link_encoder *enc110,
+	const struct dc_crtc_timing *crtc_timing,
+	int adjusted_pix_clk_khz)
+{
+	enum dc_color_depth max_deep_color =
+			enc110->base.features.max_hdmi_deep_color;
+
+	if (max_deep_color < crtc_timing->display_color_depth)
+		return false;
+
+	if (crtc_timing->display_color_depth < COLOR_DEPTH_888)
+		return false;
+	if (adjusted_pix_clk_khz < TMDS_MIN_PIXEL_CLOCK)
+		return false;
+
+	if ((adjusted_pix_clk_khz == 0) ||
+		(adjusted_pix_clk_khz > enc110->base.features.max_hdmi_pixel_clock))
+		return false;
+
+	/* DCE11 HW does not support 420 */
+	if (!enc110->base.features.hdmi_ycbcr420_supported &&
+			crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR420)
+		return false;
+
+	if ((!enc110->base.features.flags.bits.HDMI_6GB_EN ||
+			enc110->base.ctx->dc->debug.hdmi20_disable) &&
+			adjusted_pix_clk_khz >= 300000)
+		return false;
+	if (enc110->base.ctx->dc->debug.hdmi20_disable &&
+		crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR420)
+		return false;
+	return true;
+}
+
+bool dce110_link_encoder_validate_dp_output(
+	const struct dce110_link_encoder *enc110,
+	const struct dc_crtc_timing *crtc_timing)
+{
+	if (crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR420)
+		return false;
+
+	return true;
+}
+
+void dce110_link_encoder_construct(
+	struct dce110_link_encoder *enc110,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dce110_link_enc_registers *link_regs,
+	const struct dce110_link_enc_aux_registers *aux_regs,
+	const struct dce110_link_enc_hpd_registers *hpd_regs)
+{
+	struct bp_encoder_cap_info bp_cap_info = {0};
+	const struct dc_vbios_funcs *bp_funcs = init_data->ctx->dc_bios->funcs;
+	enum bp_result result = BP_RESULT_OK;
+
+	enc110->base.funcs = &dce110_lnk_enc_funcs;
+	enc110->base.ctx = init_data->ctx;
+	enc110->base.id = init_data->encoder;
+
+	enc110->base.hpd_source = init_data->hpd_source;
+	enc110->base.connector = init_data->connector;
+
+	enc110->base.preferred_engine = ENGINE_ID_UNKNOWN;
+
+	enc110->base.features = *enc_features;
+
+	enc110->base.transmitter = init_data->transmitter;
+
+	/* set the flag to indicate whether driver poll the I2C data pin
+	 * while doing the DP sink detect
+	 */
+
+/*	if (dal_adapter_service_is_feature_supported(as,
+		FEATURE_DP_SINK_DETECT_POLL_DATA_PIN))
+		enc110->base.features.flags.bits.
+			DP_SINK_DETECT_POLL_DATA_PIN = true;*/
+
+	enc110->base.output_signals =
+		SIGNAL_TYPE_DVI_SINGLE_LINK |
+		SIGNAL_TYPE_DVI_DUAL_LINK |
+		SIGNAL_TYPE_LVDS |
+		SIGNAL_TYPE_DISPLAY_PORT |
+		SIGNAL_TYPE_DISPLAY_PORT_MST |
+		SIGNAL_TYPE_EDP |
+		SIGNAL_TYPE_HDMI_TYPE_A;
+
+	/* For DCE 8.0 and 8.1, by design, UNIPHY is hardwired to DIG_BE.
+	 * SW always assign DIG_FE 1:1 mapped to DIG_FE for non-MST UNIPHY.
+	 * SW assign DIG_FE to non-MST UNIPHY first and MST last. So prefer
+	 * DIG is per UNIPHY and used by SST DP, eDP, HDMI, DVI and LVDS.
+	 * Prefer DIG assignment is decided by board design.
+	 * For DCE 8.0, there are only max 6 UNIPHYs, we assume board design
+	 * and VBIOS will filter out 7 UNIPHY for DCE 8.0.
+	 * By this, adding DIGG should not hurt DCE 8.0.
+	 * This will let DCE 8.1 share DCE 8.0 as much as possible
+	 */
+
+	enc110->link_regs = link_regs;
+	enc110->aux_regs = aux_regs;
+	enc110->hpd_regs = hpd_regs;
+
+	switch (enc110->base.transmitter) {
+	case TRANSMITTER_UNIPHY_A:
+		enc110->base.preferred_engine = ENGINE_ID_DIGA;
+	break;
+	case TRANSMITTER_UNIPHY_B:
+		enc110->base.preferred_engine = ENGINE_ID_DIGB;
+	break;
+	case TRANSMITTER_UNIPHY_C:
+		enc110->base.preferred_engine = ENGINE_ID_DIGC;
+	break;
+	case TRANSMITTER_UNIPHY_D:
+		enc110->base.preferred_engine = ENGINE_ID_DIGD;
+	break;
+	case TRANSMITTER_UNIPHY_E:
+		enc110->base.preferred_engine = ENGINE_ID_DIGE;
+	break;
+	case TRANSMITTER_UNIPHY_F:
+		enc110->base.preferred_engine = ENGINE_ID_DIGF;
+	break;
+	case TRANSMITTER_UNIPHY_G:
+		enc110->base.preferred_engine = ENGINE_ID_DIGG;
+	break;
+	default:
+		ASSERT_CRITICAL(false);
+		enc110->base.preferred_engine = ENGINE_ID_UNKNOWN;
+	}
+
+	/* default to one to mirror Windows behavior */
+	enc110->base.features.flags.bits.HDMI_6GB_EN = 1;
+
+	result = bp_funcs->get_encoder_cap_info(enc110->base.ctx->dc_bios,
+						enc110->base.id, &bp_cap_info);
+
+	/* Override features with DCE-specific values */
+	if (result == BP_RESULT_OK) {
+		enc110->base.features.flags.bits.IS_HBR2_CAPABLE =
+				bp_cap_info.DP_HBR2_EN;
+		enc110->base.features.flags.bits.IS_HBR3_CAPABLE =
+				bp_cap_info.DP_HBR3_EN;
+		enc110->base.features.flags.bits.HDMI_6GB_EN = bp_cap_info.HDMI_6GB_EN;
+	} else if (result != BP_RESULT_NORECORD) {
+		DC_LOG_WARNING("%s: Failed to get encoder_cap_info from VBIOS with error code %d!\n",
+				__func__,
+				result);
+	}
+	if (enc110->base.ctx->dc->debug.hdmi20_disable) {
+		enc110->base.features.flags.bits.HDMI_6GB_EN = 0;
+	}
+}
+
+bool dce110_link_encoder_validate_output_with_stream(
+	struct link_encoder *enc,
+	const struct dc_stream_state *stream)
+{
+	struct dce110_link_encoder *enc110 = TO_DCE110_LINK_ENC(enc);
+	bool is_valid;
+
+	switch (stream->signal) {
+	case SIGNAL_TYPE_DVI_SINGLE_LINK:
+	case SIGNAL_TYPE_DVI_DUAL_LINK:
+		is_valid = dce110_link_encoder_validate_dvi_output(
+			enc110,
+			stream->link->connector_signal,
+			stream->signal,
+			&stream->timing);
+	break;
+	case SIGNAL_TYPE_HDMI_TYPE_A:
+		is_valid = dce110_link_encoder_validate_hdmi_output(
+				enc110,
+				&stream->timing,
+				stream->phy_pix_clk);
+	break;
+	case SIGNAL_TYPE_DISPLAY_PORT:
+	case SIGNAL_TYPE_DISPLAY_PORT_MST:
+		is_valid = dce110_link_encoder_validate_dp_output(
+					enc110, &stream->timing);
+	break;
+	case SIGNAL_TYPE_EDP:
+	case SIGNAL_TYPE_LVDS:
+		is_valid = stream->timing.pixel_encoding == PIXEL_ENCODING_RGB;
+	break;
+	case SIGNAL_TYPE_VIRTUAL:
+		is_valid = true;
+		break;
+	default:
+		is_valid = false;
+	break;
+	}
+
+	return is_valid;
+}
+
+void dce110_link_encoder_hw_init(
+	struct link_encoder *enc)
+{
+	struct dce110_link_encoder *enc110 = TO_DCE110_LINK_ENC(enc);
+	struct bp_transmitter_control cntl = { 0 };
+	enum bp_result result;
+
+	cntl.action = TRANSMITTER_CONTROL_INIT;
+	cntl.engine_id = ENGINE_ID_UNKNOWN;
+	cntl.transmitter = enc110->base.transmitter;
+	cntl.connector_obj_id = enc110->base.connector;
+	cntl.lanes_number = LANE_COUNT_FOUR;
+	cntl.coherent = false;
+	cntl.hpd_sel = enc110->base.hpd_source;
+
+	if (enc110->base.connector.id == CONNECTOR_ID_EDP)
+		cntl.signal = SIGNAL_TYPE_EDP;
+
+	result = link_transmitter_control(enc110, &cntl);
+
+	if (result != BP_RESULT_OK) {
+		DC_LOG_ERROR("%s: Failed to execute VBIOS command table!\n",
+			__func__);
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+	if (enc110->base.connector.id == CONNECTOR_ID_LVDS) {
+		cntl.action = TRANSMITTER_CONTROL_BACKLIGHT_BRIGHTNESS;
+
+		result = link_transmitter_control(enc110, &cntl);
+
+		ASSERT(result == BP_RESULT_OK);
+
+	}
+	aux_initialize(enc110);
+
+	/* reinitialize HPD.
+	 * hpd_initialize() will pass DIG_FE id to HW context.
+	 * All other routine within HW context will use fe_engine_offset
+	 * as DIG_FE id even caller pass DIG_FE id.
+	 * So this routine must be called first. */
+	hpd_initialize(enc110);
+}
+
+void dce110_link_encoder_destroy(struct link_encoder **enc)
+{
+	kfree(TO_DCE110_LINK_ENC(*enc));
+	*enc = NULL;
+}
+
+void dce110_link_encoder_setup(
+	struct link_encoder *enc,
+	enum signal_type signal)
+{
+	struct dce110_link_encoder *enc110 = TO_DCE110_LINK_ENC(enc);
+
+	switch (signal) {
+	case SIGNAL_TYPE_EDP:
+	case SIGNAL_TYPE_DISPLAY_PORT:
+		/* DP SST */
+		REG_UPDATE(DIG_BE_CNTL, DIG_MODE, 0);
+		break;
+	case SIGNAL_TYPE_LVDS:
+		/* LVDS */
+		REG_UPDATE(DIG_BE_CNTL, DIG_MODE, 1);
+		break;
+	case SIGNAL_TYPE_DVI_SINGLE_LINK:
+	case SIGNAL_TYPE_DVI_DUAL_LINK:
+		/* TMDS-DVI */
+		REG_UPDATE(DIG_BE_CNTL, DIG_MODE, 2);
+		break;
+	case SIGNAL_TYPE_HDMI_TYPE_A:
+		/* TMDS-HDMI */
+		REG_UPDATE(DIG_BE_CNTL, DIG_MODE, 3);
+		break;
+	case SIGNAL_TYPE_DISPLAY_PORT_MST:
+		/* DP MST */
+		REG_UPDATE(DIG_BE_CNTL, DIG_MODE, 5);
+		break;
+	default:
+		ASSERT_CRITICAL(false);
+		/* invalid mode ! */
+		break;
+	}
+
+}
+
+/* TODO: still need depth or just pass in adjusted pixel clock? */
+void dce110_link_encoder_enable_tmds_output(
+	struct link_encoder *enc,
+	enum clock_source_id clock_source,
+	enum dc_color_depth color_depth,
+	enum signal_type signal,
+	uint32_t pixel_clock)
+{
+	struct dce110_link_encoder *enc110 = TO_DCE110_LINK_ENC(enc);
+	struct bp_transmitter_control cntl = { 0 };
+	enum bp_result result;
+
+	/* Enable the PHY */
+	cntl.connector_obj_id = enc110->base.connector;
+	cntl.action = TRANSMITTER_CONTROL_ENABLE;
+	cntl.engine_id = enc->preferred_engine;
+	cntl.transmitter = enc110->base.transmitter;
+	cntl.pll_id = clock_source;
+	cntl.signal = signal;
+	if (cntl.signal == SIGNAL_TYPE_DVI_DUAL_LINK)
+		cntl.lanes_number = 8;
+	else
+		cntl.lanes_number = 4;
+
+	cntl.hpd_sel = enc110->base.hpd_source;
+
+	cntl.pixel_clock = pixel_clock;
+	cntl.color_depth = color_depth;
+
+	result = link_transmitter_control(enc110, &cntl);
+
+	if (result != BP_RESULT_OK) {
+		DC_LOG_ERROR("%s: Failed to execute VBIOS command table!\n",
+			__func__);
+		BREAK_TO_DEBUGGER();
+	}
+}
+
+/* TODO: still need depth or just pass in adjusted pixel clock? */
+void dce110_link_encoder_enable_lvds_output(
+	struct link_encoder *enc,
+	enum clock_source_id clock_source,
+	uint32_t pixel_clock)
+{
+	struct dce110_link_encoder *enc110 = TO_DCE110_LINK_ENC(enc);
+	struct bp_transmitter_control cntl = { 0 };
+	enum bp_result result;
+
+	/* Enable the PHY */
+	cntl.connector_obj_id = enc110->base.connector;
+	cntl.action = TRANSMITTER_CONTROL_ENABLE;
+	cntl.engine_id = enc->preferred_engine;
+	cntl.transmitter = enc110->base.transmitter;
+	cntl.pll_id = clock_source;
+	cntl.signal = SIGNAL_TYPE_LVDS;
+	cntl.lanes_number = 4;
+
+	cntl.hpd_sel = enc110->base.hpd_source;
+
+	cntl.pixel_clock = pixel_clock;
+
+	result = link_transmitter_control(enc110, &cntl);
+
+	if (result != BP_RESULT_OK) {
+		DC_LOG_ERROR("%s: Failed to execute VBIOS command table!\n",
+			__func__);
+		BREAK_TO_DEBUGGER();
+	}
+}
+
+/* enables DP PHY output */
+void dce110_link_encoder_enable_dp_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source)
+{
+	struct dce110_link_encoder *enc110 = TO_DCE110_LINK_ENC(enc);
+	struct bp_transmitter_control cntl = { 0 };
+	enum bp_result result;
+
+	/* Enable the PHY */
+
+	/* number_of_lanes is used for pixel clock adjust,
+	 * but it's not passed to asic_control.
+	 * We need to set number of lanes manually.
+	 */
+	configure_encoder(enc110, link_settings);
+	cntl.connector_obj_id = enc110->base.connector;
+	cntl.action = TRANSMITTER_CONTROL_ENABLE;
+	cntl.engine_id = enc->preferred_engine;
+	cntl.transmitter = enc110->base.transmitter;
+	cntl.pll_id = clock_source;
+	cntl.signal = SIGNAL_TYPE_DISPLAY_PORT;
+	cntl.lanes_number = link_settings->lane_count;
+	cntl.hpd_sel = enc110->base.hpd_source;
+	cntl.pixel_clock = link_settings->link_rate
+						* LINK_RATE_REF_FREQ_IN_KHZ;
+	/* TODO: check if undefined works */
+	cntl.color_depth = COLOR_DEPTH_UNDEFINED;
+
+	result = link_transmitter_control(enc110, &cntl);
+
+	if (result != BP_RESULT_OK) {
+		DC_LOG_ERROR("%s: Failed to execute VBIOS command table!\n",
+			__func__);
+		BREAK_TO_DEBUGGER();
+	}
+}
+
+/* enables DP PHY output in MST mode */
+void dce110_link_encoder_enable_dp_mst_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source)
+{
+	struct dce110_link_encoder *enc110 = TO_DCE110_LINK_ENC(enc);
+	struct bp_transmitter_control cntl = { 0 };
+	enum bp_result result;
+
+	/* Enable the PHY */
+
+	/* number_of_lanes is used for pixel clock adjust,
+	 * but it's not passed to asic_control.
+	 * We need to set number of lanes manually.
+	 */
+	configure_encoder(enc110, link_settings);
+
+	cntl.action = TRANSMITTER_CONTROL_ENABLE;
+	cntl.engine_id = ENGINE_ID_UNKNOWN;
+	cntl.transmitter = enc110->base.transmitter;
+	cntl.pll_id = clock_source;
+	cntl.signal = SIGNAL_TYPE_DISPLAY_PORT_MST;
+	cntl.lanes_number = link_settings->lane_count;
+	cntl.hpd_sel = enc110->base.hpd_source;
+	cntl.pixel_clock = link_settings->link_rate
+						* LINK_RATE_REF_FREQ_IN_KHZ;
+	/* TODO: check if undefined works */
+	cntl.color_depth = COLOR_DEPTH_UNDEFINED;
+
+	result = link_transmitter_control(enc110, &cntl);
+
+	if (result != BP_RESULT_OK) {
+		DC_LOG_ERROR("%s: Failed to execute VBIOS command table!\n",
+			__func__);
+		BREAK_TO_DEBUGGER();
+	}
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+/* enables DP PHY output */
+static void dce60_link_encoder_enable_dp_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source)
+{
+	struct dce110_link_encoder *enc110 = TO_DCE110_LINK_ENC(enc);
+	struct bp_transmitter_control cntl = { 0 };
+	enum bp_result result;
+
+	/* Enable the PHY */
+
+	/* number_of_lanes is used for pixel clock adjust,
+	 * but it's not passed to asic_control.
+	 * We need to set number of lanes manually.
+	 */
+	dce60_configure_encoder(enc110, link_settings);
+	cntl.connector_obj_id = enc110->base.connector;
+	cntl.action = TRANSMITTER_CONTROL_ENABLE;
+	cntl.engine_id = enc->preferred_engine;
+	cntl.transmitter = enc110->base.transmitter;
+	cntl.pll_id = clock_source;
+	cntl.signal = SIGNAL_TYPE_DISPLAY_PORT;
+	cntl.lanes_number = link_settings->lane_count;
+	cntl.hpd_sel = enc110->base.hpd_source;
+	cntl.pixel_clock = link_settings->link_rate
+						* LINK_RATE_REF_FREQ_IN_KHZ;
+	/* TODO: check if undefined works */
+	cntl.color_depth = COLOR_DEPTH_UNDEFINED;
+
+	result = link_transmitter_control(enc110, &cntl);
+
+	if (result != BP_RESULT_OK) {
+		DC_LOG_ERROR("%s: Failed to execute VBIOS command table!\n",
+			__func__);
+		BREAK_TO_DEBUGGER();
+	}
+}
+
+/* enables DP PHY output in MST mode */
+static void dce60_link_encoder_enable_dp_mst_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source)
+{
+	struct dce110_link_encoder *enc110 = TO_DCE110_LINK_ENC(enc);
+	struct bp_transmitter_control cntl = { 0 };
+	enum bp_result result;
+
+	/* Enable the PHY */
+
+	/* number_of_lanes is used for pixel clock adjust,
+	 * but it's not passed to asic_control.
+	 * We need to set number of lanes manually.
+	 */
+	dce60_configure_encoder(enc110, link_settings);
+
+	cntl.action = TRANSMITTER_CONTROL_ENABLE;
+	cntl.engine_id = ENGINE_ID_UNKNOWN;
+	cntl.transmitter = enc110->base.transmitter;
+	cntl.pll_id = clock_source;
+	cntl.signal = SIGNAL_TYPE_DISPLAY_PORT_MST;
+	cntl.lanes_number = link_settings->lane_count;
+	cntl.hpd_sel = enc110->base.hpd_source;
+	cntl.pixel_clock = link_settings->link_rate
+						* LINK_RATE_REF_FREQ_IN_KHZ;
+	/* TODO: check if undefined works */
+	cntl.color_depth = COLOR_DEPTH_UNDEFINED;
+
+	result = link_transmitter_control(enc110, &cntl);
+
+	if (result != BP_RESULT_OK) {
+		DC_LOG_ERROR("%s: Failed to execute VBIOS command table!\n",
+			__func__);
+		BREAK_TO_DEBUGGER();
+	}
+}
+#endif
+
+/*
+ * @brief
+ * Disable transmitter and its encoder
+ */
+void dce110_link_encoder_disable_output(
+	struct link_encoder *enc,
+	enum signal_type signal)
+{
+	struct dce110_link_encoder *enc110 = TO_DCE110_LINK_ENC(enc);
+	struct bp_transmitter_control cntl = { 0 };
+	enum bp_result result;
+
+	if (!dce110_is_dig_enabled(enc)) {
+		/* OF_SKIP_POWER_DOWN_INACTIVE_ENCODER */
+		return;
+	}
+	/* Power-down RX and disable GPU PHY should be paired.
+	 * Disabling PHY without powering down RX may cause
+	 * symbol lock loss, on which we will get DP Sink interrupt. */
+
+	/* There is a case for the DP active dongles
+	 * where we want to disable the PHY but keep RX powered,
+	 * for those we need to ignore DP Sink interrupt
+	 * by checking lane count that has been set
+	 * on the last do_enable_output(). */
+
+	/* disable transmitter */
+	cntl.action = TRANSMITTER_CONTROL_DISABLE;
+	cntl.transmitter = enc110->base.transmitter;
+	cntl.hpd_sel = enc110->base.hpd_source;
+	cntl.signal = signal;
+	cntl.connector_obj_id = enc110->base.connector;
+
+	result = link_transmitter_control(enc110, &cntl);
+
+	if (result != BP_RESULT_OK) {
+		DC_LOG_ERROR("%s: Failed to execute VBIOS command table!\n",
+			__func__);
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+	/* disable encoder */
+	if (dc_is_dp_signal(signal))
+		link_encoder_disable(enc110);
+}
+
+void dce110_link_encoder_dp_set_lane_settings(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	const struct dc_lane_settings lane_settings[LANE_COUNT_DP_MAX])
+{
+	struct dce110_link_encoder *enc110 = TO_DCE110_LINK_ENC(enc);
+	union dpcd_training_lane_set training_lane_set = { { 0 } };
+	int32_t lane = 0;
+	struct bp_transmitter_control cntl = { 0 };
+
+	if (!link_settings) {
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+	cntl.action = TRANSMITTER_CONTROL_SET_VOLTAGE_AND_PREEMPASIS;
+	cntl.transmitter = enc110->base.transmitter;
+	cntl.connector_obj_id = enc110->base.connector;
+	cntl.lanes_number = link_settings->lane_count;
+	cntl.hpd_sel = enc110->base.hpd_source;
+	cntl.pixel_clock = link_settings->link_rate *
+						LINK_RATE_REF_FREQ_IN_KHZ;
+
+	for (lane = 0; lane < link_settings->lane_count; lane++) {
+		/* translate lane settings */
+
+		training_lane_set.bits.VOLTAGE_SWING_SET =
+				lane_settings[lane].VOLTAGE_SWING;
+		training_lane_set.bits.PRE_EMPHASIS_SET =
+				lane_settings[lane].PRE_EMPHASIS;
+
+		/* post cursor 2 setting only applies to HBR2 link rate */
+		if (link_settings->link_rate == LINK_RATE_HIGH2) {
+			/* this is passed to VBIOS
+			 * to program post cursor 2 level */
+
+			training_lane_set.bits.POST_CURSOR2_SET =
+					lane_settings[lane].POST_CURSOR2;
+		}
+
+		cntl.lane_select = lane;
+		cntl.lane_settings = training_lane_set.raw;
+
+		/* call VBIOS table to set voltage swing and pre-emphasis */
+		if (link_transmitter_control(enc110, &cntl) != BP_RESULT_OK) {
+			DC_LOG_ERROR("%s: Failed to execute VBIOS command table!\n", __func__);
+			BREAK_TO_DEBUGGER();
+		}
+	}
+}
+
+/* set DP PHY test and training patterns */
+void dce110_link_encoder_dp_set_phy_pattern(
+	struct link_encoder *enc,
+	const struct encoder_set_dp_phy_pattern_param *param)
+{
+	struct dce110_link_encoder *enc110 = TO_DCE110_LINK_ENC(enc);
+
+	switch (param->dp_phy_pattern) {
+	case DP_TEST_PATTERN_TRAINING_PATTERN1:
+		dce110_link_encoder_set_dp_phy_pattern_training_pattern(enc, 0);
+		break;
+	case DP_TEST_PATTERN_TRAINING_PATTERN2:
+		dce110_link_encoder_set_dp_phy_pattern_training_pattern(enc, 1);
+		break;
+	case DP_TEST_PATTERN_TRAINING_PATTERN3:
+		dce110_link_encoder_set_dp_phy_pattern_training_pattern(enc, 2);
+		break;
+	case DP_TEST_PATTERN_TRAINING_PATTERN4:
+		dce110_link_encoder_set_dp_phy_pattern_training_pattern(enc, 3);
+		break;
+	case DP_TEST_PATTERN_D102:
+		set_dp_phy_pattern_d102(enc110);
+		break;
+	case DP_TEST_PATTERN_SYMBOL_ERROR:
+		set_dp_phy_pattern_symbol_error(enc110);
+		break;
+	case DP_TEST_PATTERN_PRBS7:
+		set_dp_phy_pattern_prbs7(enc110);
+		break;
+	case DP_TEST_PATTERN_80BIT_CUSTOM:
+		set_dp_phy_pattern_80bit_custom(
+			enc110, param->custom_pattern);
+		break;
+	case DP_TEST_PATTERN_CP2520_1:
+		set_dp_phy_pattern_hbr2_compliance_cp2520_2(enc110, 1);
+		break;
+	case DP_TEST_PATTERN_CP2520_2:
+		set_dp_phy_pattern_hbr2_compliance_cp2520_2(enc110, 2);
+		break;
+	case DP_TEST_PATTERN_CP2520_3:
+		set_dp_phy_pattern_hbr2_compliance_cp2520_2(enc110, 3);
+		break;
+	case DP_TEST_PATTERN_VIDEO_MODE: {
+		set_dp_phy_pattern_passthrough_mode(
+			enc110, param->dp_panel_mode);
+		break;
+	}
+
+	default:
+		/* invalid phy pattern */
+		ASSERT_CRITICAL(false);
+		break;
+	}
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+/* set DP PHY test and training patterns */
+static void dce60_link_encoder_dp_set_phy_pattern(
+	struct link_encoder *enc,
+	const struct encoder_set_dp_phy_pattern_param *param)
+{
+	struct dce110_link_encoder *enc110 = TO_DCE110_LINK_ENC(enc);
+
+	switch (param->dp_phy_pattern) {
+	case DP_TEST_PATTERN_TRAINING_PATTERN1:
+		dce110_link_encoder_set_dp_phy_pattern_training_pattern(enc, 0);
+		break;
+	case DP_TEST_PATTERN_TRAINING_PATTERN2:
+		dce110_link_encoder_set_dp_phy_pattern_training_pattern(enc, 1);
+		break;
+	case DP_TEST_PATTERN_TRAINING_PATTERN3:
+		dce110_link_encoder_set_dp_phy_pattern_training_pattern(enc, 2);
+		break;
+	case DP_TEST_PATTERN_TRAINING_PATTERN4:
+		dce110_link_encoder_set_dp_phy_pattern_training_pattern(enc, 3);
+		break;
+	case DP_TEST_PATTERN_D102:
+		set_dp_phy_pattern_d102(enc110);
+		break;
+	case DP_TEST_PATTERN_SYMBOL_ERROR:
+		set_dp_phy_pattern_symbol_error(enc110);
+		break;
+	case DP_TEST_PATTERN_PRBS7:
+		set_dp_phy_pattern_prbs7(enc110);
+		break;
+	case DP_TEST_PATTERN_80BIT_CUSTOM:
+		set_dp_phy_pattern_80bit_custom(
+			enc110, param->custom_pattern);
+		break;
+	case DP_TEST_PATTERN_CP2520_1:
+		dce60_set_dp_phy_pattern_hbr2_compliance_cp2520_2(enc110, 1);
+		break;
+	case DP_TEST_PATTERN_CP2520_2:
+		dce60_set_dp_phy_pattern_hbr2_compliance_cp2520_2(enc110, 2);
+		break;
+	case DP_TEST_PATTERN_CP2520_3:
+		dce60_set_dp_phy_pattern_hbr2_compliance_cp2520_2(enc110, 3);
+		break;
+	case DP_TEST_PATTERN_VIDEO_MODE: {
+		dce60_set_dp_phy_pattern_passthrough_mode(
+			enc110, param->dp_panel_mode);
+		break;
+	}
+
+	default:
+		/* invalid phy pattern */
+		ASSERT_CRITICAL(false);
+		break;
+	}
+}
+#endif
+
+static void fill_stream_allocation_row_info(
+	const struct link_mst_stream_allocation *stream_allocation,
+	uint32_t *src,
+	uint32_t *slots)
+{
+	const struct stream_encoder *stream_enc = stream_allocation->stream_enc;
+
+	if (stream_enc) {
+		*src = stream_enc->id;
+		*slots = stream_allocation->slot_count;
+	} else {
+		*src = 0;
+		*slots = 0;
+	}
+}
+
+/* programs DP MST VC payload allocation */
+void dce110_link_encoder_update_mst_stream_allocation_table(
+	struct link_encoder *enc,
+	const struct link_mst_stream_allocation_table *table)
+{
+	struct dce110_link_encoder *enc110 = TO_DCE110_LINK_ENC(enc);
+	uint32_t value1 = 0;
+	uint32_t value2 = 0;
+	uint32_t slots = 0;
+	uint32_t src = 0;
+	uint32_t retries = 0;
+
+	/* For CZ, there are only 3 pipes. So Virtual channel is up 3.*/
+
+	/* --- Set MSE Stream Attribute -
+	 * Setup VC Payload Table on Tx Side,
+	 * Issue allocation change trigger
+	 * to commit payload on both tx and rx side */
+
+	/* we should clean-up table each time */
+
+	if (table->stream_count >= 1) {
+		fill_stream_allocation_row_info(
+			&table->stream_allocations[0],
+			&src,
+			&slots);
+	} else {
+		src = 0;
+		slots = 0;
+	}
+
+	REG_UPDATE_2(DP_MSE_SAT0,
+			DP_MSE_SAT_SRC0, src,
+			DP_MSE_SAT_SLOT_COUNT0, slots);
+
+	if (table->stream_count >= 2) {
+		fill_stream_allocation_row_info(
+			&table->stream_allocations[1],
+			&src,
+			&slots);
+	} else {
+		src = 0;
+		slots = 0;
+	}
+
+	REG_UPDATE_2(DP_MSE_SAT0,
+			DP_MSE_SAT_SRC1, src,
+			DP_MSE_SAT_SLOT_COUNT1, slots);
+
+	if (table->stream_count >= 3) {
+		fill_stream_allocation_row_info(
+			&table->stream_allocations[2],
+			&src,
+			&slots);
+	} else {
+		src = 0;
+		slots = 0;
+	}
+
+	REG_UPDATE_2(DP_MSE_SAT1,
+			DP_MSE_SAT_SRC2, src,
+			DP_MSE_SAT_SLOT_COUNT2, slots);
+
+	if (table->stream_count >= 4) {
+		fill_stream_allocation_row_info(
+			&table->stream_allocations[3],
+			&src,
+			&slots);
+	} else {
+		src = 0;
+		slots = 0;
+	}
+
+	REG_UPDATE_2(DP_MSE_SAT1,
+			DP_MSE_SAT_SRC3, src,
+			DP_MSE_SAT_SLOT_COUNT3, slots);
+
+	/* --- wait for transaction finish */
+
+	/* send allocation change trigger (ACT) ?
+	 * this step first sends the ACT,
+	 * then double buffers the SAT into the hardware
+	 * making the new allocation active on the DP MST mode link */
+
+
+	/* DP_MSE_SAT_UPDATE:
+	 * 0 - No Action
+	 * 1 - Update SAT with trigger
+	 * 2 - Update SAT without trigger */
+
+	REG_UPDATE(DP_MSE_SAT_UPDATE,
+			DP_MSE_SAT_UPDATE, 1);
+
+	/* wait for update to complete
+	 * (i.e. DP_MSE_SAT_UPDATE field is reset to 0)
+	 * then wait for the transmission
+	 * of at least 16 MTP headers on immediate local link.
+	 * i.e. DP_MSE_16_MTP_KEEPOUT field (read only) is reset to 0
+	 * a value of 1 indicates that DP MST mode
+	 * is in the 16 MTP keepout region after a VC has been added.
+	 * MST stream bandwidth (VC rate) can be configured
+	 * after this bit is cleared */
+
+	do {
+		udelay(10);
+
+		REG_READ(DP_MSE_SAT_UPDATE);
+
+		REG_GET(DP_MSE_SAT_UPDATE,
+				DP_MSE_SAT_UPDATE, &value1);
+
+		REG_GET(DP_MSE_SAT_UPDATE,
+				DP_MSE_16_MTP_KEEPOUT, &value2);
+
+		/* bit field DP_MSE_SAT_UPDATE is set to 1 already */
+		if (!value1 && !value2)
+			break;
+		++retries;
+	} while (retries < DP_MST_UPDATE_MAX_RETRY);
+}
+
+void dce110_link_encoder_connect_dig_be_to_fe(
+	struct link_encoder *enc,
+	enum engine_id engine,
+	bool connect)
+{
+	struct dce110_link_encoder *enc110 = TO_DCE110_LINK_ENC(enc);
+	uint32_t field;
+
+	if (engine != ENGINE_ID_UNKNOWN) {
+
+		REG_GET(DIG_BE_CNTL, DIG_FE_SOURCE_SELECT, &field);
+
+		if (connect)
+			field |= get_frontend_source(engine);
+		else
+			field &= ~get_frontend_source(engine);
+
+		REG_UPDATE(DIG_BE_CNTL, DIG_FE_SOURCE_SELECT, field);
+	}
+}
+
+void dce110_link_encoder_enable_hpd(struct link_encoder *enc)
+{
+	struct dce110_link_encoder *enc110 = TO_DCE110_LINK_ENC(enc);
+	struct dc_context *ctx = enc110->base.ctx;
+	uint32_t addr = HPD_REG(DC_HPD_CONTROL);
+	uint32_t hpd_enable = 0;
+	uint32_t value = dm_read_reg(ctx, addr);
+
+	hpd_enable = get_reg_field_value(hpd_enable, DC_HPD_CONTROL, DC_HPD_EN);
+
+	if (hpd_enable == 0)
+		set_reg_field_value(value, 1, DC_HPD_CONTROL, DC_HPD_EN);
+}
+
+void dce110_link_encoder_disable_hpd(struct link_encoder *enc)
+{
+	struct dce110_link_encoder *enc110 = TO_DCE110_LINK_ENC(enc);
+	struct dc_context *ctx = enc110->base.ctx;
+	uint32_t addr = HPD_REG(DC_HPD_CONTROL);
+	uint32_t value = dm_read_reg(ctx, addr);
+
+	set_reg_field_value(value, 0, DC_HPD_CONTROL, DC_HPD_EN);
+}
+
+void dce110_link_encoder_get_max_link_cap(struct link_encoder *enc,
+	struct dc_link_settings *link_settings)
+{
+	/* Set Default link settings */
+	struct dc_link_settings max_link_cap = {LANE_COUNT_FOUR, LINK_RATE_HIGH,
+			LINK_SPREAD_05_DOWNSPREAD_30KHZ, false, 0};
+
+	/* Higher link settings based on feature supported */
+	if (enc->features.flags.bits.IS_HBR2_CAPABLE)
+		max_link_cap.link_rate = LINK_RATE_HIGH2;
+
+	if (enc->features.flags.bits.IS_HBR3_CAPABLE)
+		max_link_cap.link_rate = LINK_RATE_HIGH3;
+
+	*link_settings = max_link_cap;
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static const struct link_encoder_funcs dce60_lnk_enc_funcs = {
+	.validate_output_with_stream =
+		dce110_link_encoder_validate_output_with_stream,
+	.hw_init = dce110_link_encoder_hw_init,
+	.setup = dce110_link_encoder_setup,
+	.enable_tmds_output = dce110_link_encoder_enable_tmds_output,
+	.enable_dp_output = dce60_link_encoder_enable_dp_output,
+	.enable_dp_mst_output = dce60_link_encoder_enable_dp_mst_output,
+	.enable_lvds_output = dce110_link_encoder_enable_lvds_output,
+	.disable_output = dce110_link_encoder_disable_output,
+	.dp_set_lane_settings = dce110_link_encoder_dp_set_lane_settings,
+	.dp_set_phy_pattern = dce60_link_encoder_dp_set_phy_pattern,
+	.update_mst_stream_allocation_table =
+		dce110_link_encoder_update_mst_stream_allocation_table,
+	.psr_program_dp_dphy_fast_training =
+			dce110_psr_program_dp_dphy_fast_training,
+	.psr_program_secondary_packet = dce110_psr_program_secondary_packet,
+	.connect_dig_be_to_fe = dce110_link_encoder_connect_dig_be_to_fe,
+	.enable_hpd = dce110_link_encoder_enable_hpd,
+	.disable_hpd = dce110_link_encoder_disable_hpd,
+	.is_dig_enabled = dce110_is_dig_enabled,
+	.destroy = dce110_link_encoder_destroy,
+	.get_max_link_cap = dce110_link_encoder_get_max_link_cap,
+	.get_dig_frontend = dce110_get_dig_frontend
+};
+
+void dce60_link_encoder_construct(
+	struct dce110_link_encoder *enc110,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dce110_link_enc_registers *link_regs,
+	const struct dce110_link_enc_aux_registers *aux_regs,
+	const struct dce110_link_enc_hpd_registers *hpd_regs)
+{
+	struct bp_encoder_cap_info bp_cap_info = {0};
+	const struct dc_vbios_funcs *bp_funcs = init_data->ctx->dc_bios->funcs;
+	enum bp_result result = BP_RESULT_OK;
+
+	enc110->base.funcs = &dce60_lnk_enc_funcs;
+	enc110->base.ctx = init_data->ctx;
+	enc110->base.id = init_data->encoder;
+
+	enc110->base.hpd_source = init_data->hpd_source;
+	enc110->base.connector = init_data->connector;
+
+	enc110->base.preferred_engine = ENGINE_ID_UNKNOWN;
+
+	enc110->base.features = *enc_features;
+
+	enc110->base.transmitter = init_data->transmitter;
+
+	/* set the flag to indicate whether driver poll the I2C data pin
+	 * while doing the DP sink detect
+	 */
+
+/*	if (dal_adapter_service_is_feature_supported(as,
+		FEATURE_DP_SINK_DETECT_POLL_DATA_PIN))
+		enc110->base.features.flags.bits.
+			DP_SINK_DETECT_POLL_DATA_PIN = true;*/
+
+	enc110->base.output_signals =
+		SIGNAL_TYPE_DVI_SINGLE_LINK |
+		SIGNAL_TYPE_DVI_DUAL_LINK |
+		SIGNAL_TYPE_LVDS |
+		SIGNAL_TYPE_DISPLAY_PORT |
+		SIGNAL_TYPE_DISPLAY_PORT_MST |
+		SIGNAL_TYPE_EDP |
+		SIGNAL_TYPE_HDMI_TYPE_A;
+
+	/* For DCE 8.0 and 8.1, by design, UNIPHY is hardwired to DIG_BE.
+	 * SW always assign DIG_FE 1:1 mapped to DIG_FE for non-MST UNIPHY.
+	 * SW assign DIG_FE to non-MST UNIPHY first and MST last. So prefer
+	 * DIG is per UNIPHY and used by SST DP, eDP, HDMI, DVI and LVDS.
+	 * Prefer DIG assignment is decided by board design.
+	 * For DCE 8.0, there are only max 6 UNIPHYs, we assume board design
+	 * and VBIOS will filter out 7 UNIPHY for DCE 8.0.
+	 * By this, adding DIGG should not hurt DCE 8.0.
+	 * This will let DCE 8.1 share DCE 8.0 as much as possible
+	 */
+
+	enc110->link_regs = link_regs;
+	enc110->aux_regs = aux_regs;
+	enc110->hpd_regs = hpd_regs;
+
+	switch (enc110->base.transmitter) {
+	case TRANSMITTER_UNIPHY_A:
+		enc110->base.preferred_engine = ENGINE_ID_DIGA;
+	break;
+	case TRANSMITTER_UNIPHY_B:
+		enc110->base.preferred_engine = ENGINE_ID_DIGB;
+	break;
+	case TRANSMITTER_UNIPHY_C:
+		enc110->base.preferred_engine = ENGINE_ID_DIGC;
+	break;
+	case TRANSMITTER_UNIPHY_D:
+		enc110->base.preferred_engine = ENGINE_ID_DIGD;
+	break;
+	case TRANSMITTER_UNIPHY_E:
+		enc110->base.preferred_engine = ENGINE_ID_DIGE;
+	break;
+	case TRANSMITTER_UNIPHY_F:
+		enc110->base.preferred_engine = ENGINE_ID_DIGF;
+	break;
+	case TRANSMITTER_UNIPHY_G:
+		enc110->base.preferred_engine = ENGINE_ID_DIGG;
+	break;
+	default:
+		ASSERT_CRITICAL(false);
+		enc110->base.preferred_engine = ENGINE_ID_UNKNOWN;
+	}
+
+	/* default to one to mirror Windows behavior */
+	enc110->base.features.flags.bits.HDMI_6GB_EN = 1;
+
+	result = bp_funcs->get_encoder_cap_info(enc110->base.ctx->dc_bios,
+						enc110->base.id, &bp_cap_info);
+
+	/* Override features with DCE-specific values */
+	if (result == BP_RESULT_OK) {
+		enc110->base.features.flags.bits.IS_HBR2_CAPABLE =
+				bp_cap_info.DP_HBR2_EN;
+		enc110->base.features.flags.bits.IS_HBR3_CAPABLE =
+				bp_cap_info.DP_HBR3_EN;
+		enc110->base.features.flags.bits.HDMI_6GB_EN = bp_cap_info.HDMI_6GB_EN;
+	} else if (result != BP_RESULT_NORECORD) {
+		DC_LOG_WARNING("%s: Failed to get encoder_cap_info from VBIOS with error code %d!\n",
+				__func__,
+				result);
+	}
+	if (enc110->base.ctx->dc->debug.hdmi20_disable) {
+		enc110->base.features.flags.bits.HDMI_6GB_EN = 0;
+	}
+}
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.h b/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.h
new file mode 100644
index 000000000000..261c70e01e33
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.h
@@ -0,0 +1,320 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_LINK_ENCODER__DCE110_H__
+#define __DC_LINK_ENCODER__DCE110_H__
+
+#include "link_encoder.h"
+
+#define TO_DCE110_LINK_ENC(link_encoder)\
+	container_of(link_encoder, struct dce110_link_encoder, base)
+
+/* Not found regs in dce120 spec
+ * BIOS_SCRATCH_2
+ * DP_DPHY_INTERNAL_CTRL
+ */
+
+#define AUX_REG_LIST(id)\
+	SRI(AUX_CONTROL, DP_AUX, id), \
+	SRI(AUX_DPHY_RX_CONTROL0, DP_AUX, id), \
+	SRI(AUX_DPHY_RX_CONTROL1, DP_AUX, id)
+
+#define HPD_REG_LIST(id)\
+	SRI(DC_HPD_CONTROL, HPD, id)
+
+#define LE_COMMON_REG_LIST_BASE(id) \
+	SR(DMCU_RAM_ACCESS_CTRL), \
+	SR(DMCU_IRAM_RD_CTRL), \
+	SR(DMCU_IRAM_RD_DATA), \
+	SR(DMCU_INTERRUPT_TO_UC_EN_MASK), \
+	SRI(DIG_BE_CNTL, DIG, id), \
+	SRI(DIG_BE_EN_CNTL, DIG, id), \
+	SRI(DP_CONFIG, DP, id), \
+	SRI(DP_DPHY_CNTL, DP, id), \
+	SRI(DP_DPHY_PRBS_CNTL, DP, id), \
+	SRI(DP_DPHY_SCRAM_CNTL, DP, id),\
+	SRI(DP_DPHY_SYM0, DP, id), \
+	SRI(DP_DPHY_SYM1, DP, id), \
+	SRI(DP_DPHY_SYM2, DP, id), \
+	SRI(DP_DPHY_TRAINING_PATTERN_SEL, DP, id), \
+	SRI(DP_LINK_CNTL, DP, id), \
+	SRI(DP_LINK_FRAMING_CNTL, DP, id), \
+	SRI(DP_MSE_SAT0, DP, id), \
+	SRI(DP_MSE_SAT1, DP, id), \
+	SRI(DP_MSE_SAT2, DP, id), \
+	SRI(DP_MSE_SAT_UPDATE, DP, id), \
+	SRI(DP_SEC_CNTL, DP, id), \
+	SRI(DP_VID_STREAM_CNTL, DP, id), \
+	SRI(DP_DPHY_FAST_TRAINING, DP, id), \
+	SRI(DP_SEC_CNTL1, DP, id)
+
+#define LE_COMMON_REG_LIST(id)\
+	LE_COMMON_REG_LIST_BASE(id), \
+	SRI(DP_DPHY_BS_SR_SWAP_CNTL, DP, id), \
+	SRI(DP_DPHY_INTERNAL_CTRL, DP, id), \
+	SR(DCI_MEM_PWR_STATUS)
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+#define LE_DCE60_REG_LIST(id)\
+	SRI(DP_DPHY_INTERNAL_CTRL, DP, id), \
+	SR(DMCU_RAM_ACCESS_CTRL), \
+	SR(DMCU_IRAM_RD_CTRL), \
+	SR(DMCU_IRAM_RD_DATA), \
+	SR(DMCU_INTERRUPT_TO_UC_EN_MASK), \
+	SRI(DIG_BE_CNTL, DIG, id), \
+	SRI(DIG_BE_EN_CNTL, DIG, id), \
+	SRI(DP_CONFIG, DP, id), \
+	SRI(DP_DPHY_CNTL, DP, id), \
+	SRI(DP_DPHY_PRBS_CNTL, DP, id), \
+	SRI(DP_DPHY_SYM0, DP, id), \
+	SRI(DP_DPHY_SYM1, DP, id), \
+	SRI(DP_DPHY_SYM2, DP, id), \
+	SRI(DP_DPHY_TRAINING_PATTERN_SEL, DP, id), \
+	SRI(DP_LINK_CNTL, DP, id), \
+	SRI(DP_LINK_FRAMING_CNTL, DP, id), \
+	SRI(DP_MSE_SAT0, DP, id), \
+	SRI(DP_MSE_SAT1, DP, id), \
+	SRI(DP_MSE_SAT2, DP, id), \
+	SRI(DP_MSE_SAT_UPDATE, DP, id), \
+	SRI(DP_SEC_CNTL, DP, id), \
+	SRI(DP_VID_STREAM_CNTL, DP, id), \
+	SRI(DP_DPHY_FAST_TRAINING, DP, id), \
+	SRI(DP_SEC_CNTL1, DP, id)
+#endif
+
+#define LE_DCE80_REG_LIST(id)\
+	SRI(DP_DPHY_INTERNAL_CTRL, DP, id), \
+	LE_COMMON_REG_LIST_BASE(id)
+
+#define LE_DCE100_REG_LIST(id)\
+	LE_COMMON_REG_LIST_BASE(id), \
+	SRI(DP_DPHY_BS_SR_SWAP_CNTL, DP, id), \
+	SRI(DP_DPHY_INTERNAL_CTRL, DP, id), \
+	SR(DCI_MEM_PWR_STATUS)
+
+#define LE_DCE110_REG_LIST(id)\
+	LE_COMMON_REG_LIST_BASE(id), \
+	SRI(DP_DPHY_BS_SR_SWAP_CNTL, DP, id), \
+	SRI(DP_DPHY_INTERNAL_CTRL, DP, id), \
+	SRI(DP_DPHY_HBR2_PATTERN_CONTROL, DP, id), \
+	SR(DCI_MEM_PWR_STATUS)
+
+#define LE_DCE120_REG_LIST(id)\
+	LE_COMMON_REG_LIST_BASE(id), \
+	SRI(DP_DPHY_BS_SR_SWAP_CNTL, DP, id), \
+	SRI(DP_DPHY_HBR2_PATTERN_CONTROL, DP, id), \
+	SR(DCI_MEM_PWR_STATUS)
+
+#define LE_DCN10_REG_LIST(id)\
+	LE_COMMON_REG_LIST_BASE(id), \
+	SRI(DP_DPHY_BS_SR_SWAP_CNTL, DP, id), \
+	SRI(DP_DPHY_INTERNAL_CTRL, DP, id), \
+	SRI(DP_DPHY_HBR2_PATTERN_CONTROL, DP, id)
+
+struct dce110_link_enc_aux_registers {
+	uint32_t AUX_CONTROL;
+	uint32_t AUX_DPHY_RX_CONTROL0;
+	uint32_t AUX_DPHY_RX_CONTROL1;
+};
+
+struct dce110_link_enc_hpd_registers {
+	uint32_t DC_HPD_CONTROL;
+};
+
+struct dce110_link_enc_registers {
+	/* DMCU registers */
+	uint32_t MASTER_COMM_DATA_REG1;
+	uint32_t MASTER_COMM_DATA_REG2;
+	uint32_t MASTER_COMM_DATA_REG3;
+	uint32_t MASTER_COMM_CMD_REG;
+	uint32_t MASTER_COMM_CNTL_REG;
+	uint32_t DMCU_RAM_ACCESS_CTRL;
+	uint32_t DCI_MEM_PWR_STATUS;
+	uint32_t DMU_MEM_PWR_CNTL;
+	uint32_t DMCU_IRAM_RD_CTRL;
+	uint32_t DMCU_IRAM_RD_DATA;
+	uint32_t DMCU_INTERRUPT_TO_UC_EN_MASK;
+
+	/* Common DP registers */
+	uint32_t DIG_BE_CNTL;
+	uint32_t DIG_BE_EN_CNTL;
+	uint32_t DP_CONFIG;
+	uint32_t DP_DPHY_CNTL;
+	uint32_t DP_DPHY_INTERNAL_CTRL;
+	uint32_t DP_DPHY_PRBS_CNTL;
+	uint32_t DP_DPHY_SCRAM_CNTL;
+	uint32_t DP_DPHY_SYM0;
+	uint32_t DP_DPHY_SYM1;
+	uint32_t DP_DPHY_SYM2;
+	uint32_t DP_DPHY_TRAINING_PATTERN_SEL;
+	uint32_t DP_LINK_CNTL;
+	uint32_t DP_LINK_FRAMING_CNTL;
+	uint32_t DP_MSE_SAT0;
+	uint32_t DP_MSE_SAT1;
+	uint32_t DP_MSE_SAT2;
+	uint32_t DP_MSE_SAT_UPDATE;
+	uint32_t DP_SEC_CNTL;
+	uint32_t DP_VID_STREAM_CNTL;
+	uint32_t DP_DPHY_FAST_TRAINING;
+	uint32_t DP_DPHY_BS_SR_SWAP_CNTL;
+	uint32_t DP_DPHY_HBR2_PATTERN_CONTROL;
+	uint32_t DP_SEC_CNTL1;
+};
+
+struct dce110_link_encoder {
+	struct link_encoder base;
+	const struct dce110_link_enc_registers *link_regs;
+	const struct dce110_link_enc_aux_registers *aux_regs;
+	const struct dce110_link_enc_hpd_registers *hpd_regs;
+};
+
+
+void dce110_link_encoder_construct(
+	struct dce110_link_encoder *enc110,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dce110_link_enc_registers *link_regs,
+	const struct dce110_link_enc_aux_registers *aux_regs,
+	const struct dce110_link_enc_hpd_registers *hpd_regs);
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+void dce60_link_encoder_construct(
+	struct dce110_link_encoder *enc110,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dce110_link_enc_registers *link_regs,
+	const struct dce110_link_enc_aux_registers *aux_regs,
+	const struct dce110_link_enc_hpd_registers *hpd_regs);
+#endif
+
+bool dce110_link_encoder_validate_dvi_output(
+	const struct dce110_link_encoder *enc110,
+	enum signal_type connector_signal,
+	enum signal_type signal,
+	const struct dc_crtc_timing *crtc_timing);
+
+bool dce110_link_encoder_validate_rgb_output(
+	const struct dce110_link_encoder *enc110,
+	const struct dc_crtc_timing *crtc_timing);
+
+bool dce110_link_encoder_validate_dp_output(
+	const struct dce110_link_encoder *enc110,
+	const struct dc_crtc_timing *crtc_timing);
+
+bool dce110_link_encoder_validate_wireless_output(
+	const struct dce110_link_encoder *enc110,
+	const struct dc_crtc_timing *crtc_timing);
+
+bool dce110_link_encoder_validate_output_with_stream(
+	struct link_encoder *enc,
+	const struct dc_stream_state *stream);
+
+/****************** HW programming ************************/
+
+/* initialize HW */  /* why do we initialze aux in here? */
+void dce110_link_encoder_hw_init(struct link_encoder *enc);
+
+void dce110_link_encoder_destroy(struct link_encoder **enc);
+
+/* program DIG_MODE in DIG_BE */
+/* TODO can this be combined with enable_output? */
+void dce110_link_encoder_setup(
+	struct link_encoder *enc,
+	enum signal_type signal);
+
+/* enables TMDS PHY output */
+/* TODO: still need depth or just pass in adjusted pixel clock? */
+void dce110_link_encoder_enable_tmds_output(
+	struct link_encoder *enc,
+	enum clock_source_id clock_source,
+	enum dc_color_depth color_depth,
+	enum signal_type signal,
+	uint32_t pixel_clock);
+
+/* enables DP PHY output */
+void dce110_link_encoder_enable_dp_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source);
+
+/* enables DP PHY output in MST mode */
+void dce110_link_encoder_enable_dp_mst_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source);
+
+/* enables LVDS PHY output */
+void dce110_link_encoder_enable_lvds_output(
+	struct link_encoder *enc,
+	enum clock_source_id clock_source,
+	uint32_t pixel_clock);
+
+/* disable PHY output */
+void dce110_link_encoder_disable_output(
+	struct link_encoder *enc,
+	enum signal_type signal);
+
+/* set DP lane settings */
+void dce110_link_encoder_dp_set_lane_settings(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	const struct dc_lane_settings lane_settings[LANE_COUNT_DP_MAX]);
+
+void dce110_link_encoder_dp_set_phy_pattern(
+	struct link_encoder *enc,
+	const struct encoder_set_dp_phy_pattern_param *param);
+
+/* programs DP MST VC payload allocation */
+void dce110_link_encoder_update_mst_stream_allocation_table(
+	struct link_encoder *enc,
+	const struct link_mst_stream_allocation_table *table);
+
+void dce110_link_encoder_connect_dig_be_to_fe(
+	struct link_encoder *enc,
+	enum engine_id engine,
+	bool connect);
+
+unsigned int dce110_get_dig_frontend(struct link_encoder *enc);
+
+void dce110_link_encoder_set_dp_phy_pattern_training_pattern(
+	struct link_encoder *enc,
+	uint32_t index);
+
+void dce110_link_encoder_enable_hpd(struct link_encoder *enc);
+
+void dce110_link_encoder_disable_hpd(struct link_encoder *enc);
+
+void dce110_psr_program_dp_dphy_fast_training(struct link_encoder *enc,
+			bool exit_link_training_required);
+
+void dce110_psr_program_secondary_packet(struct link_encoder *enc,
+			unsigned int sdp_transmit_line_num_deadline);
+
+bool dce110_is_dig_enabled(struct link_encoder *enc);
+
+void dce110_link_encoder_get_max_link_cap(struct link_encoder *enc,
+	struct dc_link_settings *link_settings);
+
+#endif /* __DC_LINK_ENCODER__DCE110_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_mem_input.c b/drivers/gpu/drm/amd/display/dc/dce/dce_mem_input.c
new file mode 100644
index 000000000000..1c2009e38aa1
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_mem_input.c
@@ -0,0 +1,1006 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dce_mem_input.h"
+#include "reg_helper.h"
+#include "basics/conversion.h"
+
+#define CTX \
+	dce_mi->base.ctx
+#define REG(reg)\
+	dce_mi->regs->reg
+
+#undef FN
+#define FN(reg_name, field_name) \
+	dce_mi->shifts->field_name, dce_mi->masks->field_name
+
+struct pte_setting {
+	unsigned int bpp;
+	unsigned int page_width;
+	unsigned int page_height;
+	unsigned char min_pte_before_flip_horiz_scan;
+	unsigned char min_pte_before_flip_vert_scan;
+	unsigned char pte_req_per_chunk;
+	unsigned char param_6;
+	unsigned char param_7;
+	unsigned char param_8;
+};
+
+enum mi_bits_per_pixel {
+	mi_bpp_8 = 0,
+	mi_bpp_16,
+	mi_bpp_32,
+	mi_bpp_64,
+	mi_bpp_count,
+};
+
+enum mi_tiling_format {
+	mi_tiling_linear = 0,
+	mi_tiling_1D,
+	mi_tiling_2D,
+	mi_tiling_count,
+};
+
+static const struct pte_setting pte_settings[mi_tiling_count][mi_bpp_count] = {
+	[mi_tiling_linear] = {
+		{  8, 4096, 1, 8, 0, 1, 0, 0, 0},
+		{ 16, 2048, 1, 8, 0, 1, 0, 0, 0},
+		{ 32, 1024, 1, 8, 0, 1, 0, 0, 0},
+		{ 64,  512, 1, 8, 0, 1, 0, 0, 0}, /* new for 64bpp from HW */
+	},
+	[mi_tiling_1D] = {
+		{  8, 512, 8, 1, 0, 1, 0, 0, 0},  /* 0 for invalid */
+		{ 16, 256, 8, 2, 0, 1, 0, 0, 0},
+		{ 32, 128, 8, 4, 0, 1, 0, 0, 0},
+		{ 64,  64, 8, 4, 0, 1, 0, 0, 0}, /* fake */
+	},
+	[mi_tiling_2D] = {
+		{  8, 64, 64,  8,  8, 1, 4, 0, 0},
+		{ 16, 64, 32,  8, 16, 1, 8, 0, 0},
+		{ 32, 32, 32, 16, 16, 1, 8, 0, 0},
+		{ 64,  8, 32, 16, 16, 1, 8, 0, 0}, /* fake */
+	},
+};
+
+static enum mi_bits_per_pixel get_mi_bpp(
+		enum surface_pixel_format format)
+{
+	if (format >= SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616)
+		return mi_bpp_64;
+	else if (format >= SURFACE_PIXEL_FORMAT_GRPH_ARGB8888)
+		return mi_bpp_32;
+	else if (format >= SURFACE_PIXEL_FORMAT_GRPH_ARGB1555)
+		return mi_bpp_16;
+	else
+		return mi_bpp_8;
+}
+
+static enum mi_tiling_format get_mi_tiling(
+		struct dc_tiling_info *tiling_info)
+{
+	switch (tiling_info->gfx8.array_mode) {
+	case DC_ARRAY_1D_TILED_THIN1:
+	case DC_ARRAY_1D_TILED_THICK:
+	case DC_ARRAY_PRT_TILED_THIN1:
+		return mi_tiling_1D;
+	case DC_ARRAY_2D_TILED_THIN1:
+	case DC_ARRAY_2D_TILED_THICK:
+	case DC_ARRAY_2D_TILED_X_THICK:
+	case DC_ARRAY_PRT_2D_TILED_THIN1:
+	case DC_ARRAY_PRT_2D_TILED_THICK:
+		return mi_tiling_2D;
+	case DC_ARRAY_LINEAR_GENERAL:
+	case DC_ARRAY_LINEAR_ALLIGNED:
+		return mi_tiling_linear;
+	default:
+		return mi_tiling_2D;
+	}
+}
+
+static bool is_vert_scan(enum dc_rotation_angle rotation)
+{
+	switch (rotation) {
+	case ROTATION_ANGLE_90:
+	case ROTATION_ANGLE_270:
+		return true;
+	default:
+		return false;
+	}
+}
+
+static void dce_mi_program_pte_vm(
+		struct mem_input *mi,
+		enum surface_pixel_format format,
+		struct dc_tiling_info *tiling_info,
+		enum dc_rotation_angle rotation)
+{
+	struct dce_mem_input *dce_mi = TO_DCE_MEM_INPUT(mi);
+	enum mi_bits_per_pixel mi_bpp = get_mi_bpp(format);
+	enum mi_tiling_format mi_tiling = get_mi_tiling(tiling_info);
+	const struct pte_setting *pte = &pte_settings[mi_tiling][mi_bpp];
+
+	unsigned int page_width = log_2(pte->page_width);
+	unsigned int page_height = log_2(pte->page_height);
+	unsigned int min_pte_before_flip = is_vert_scan(rotation) ?
+			pte->min_pte_before_flip_vert_scan :
+			pte->min_pte_before_flip_horiz_scan;
+
+	REG_UPDATE(GRPH_PIPE_OUTSTANDING_REQUEST_LIMIT,
+			GRPH_PIPE_OUTSTANDING_REQUEST_LIMIT, 0x7f);
+
+	REG_UPDATE_3(DVMM_PTE_CONTROL,
+			DVMM_PAGE_WIDTH, page_width,
+			DVMM_PAGE_HEIGHT, page_height,
+			DVMM_MIN_PTE_BEFORE_FLIP, min_pte_before_flip);
+
+	REG_UPDATE_2(DVMM_PTE_ARB_CONTROL,
+			DVMM_PTE_REQ_PER_CHUNK, pte->pte_req_per_chunk,
+			DVMM_MAX_PTE_REQ_OUTSTANDING, 0x7f);
+}
+
+static void program_urgency_watermark(
+	struct dce_mem_input *dce_mi,
+	uint32_t wm_select,
+	uint32_t urgency_low_wm,
+	uint32_t urgency_high_wm)
+{
+	REG_UPDATE(DPG_WATERMARK_MASK_CONTROL,
+		URGENCY_WATERMARK_MASK, wm_select);
+
+	REG_SET_2(DPG_PIPE_URGENCY_CONTROL, 0,
+		URGENCY_LOW_WATERMARK, urgency_low_wm,
+		URGENCY_HIGH_WATERMARK, urgency_high_wm);
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static void dce60_program_urgency_watermark(
+	struct dce_mem_input *dce_mi,
+	uint32_t wm_select,
+	uint32_t urgency_low_wm,
+	uint32_t urgency_high_wm)
+{
+	REG_UPDATE(DPG_PIPE_ARBITRATION_CONTROL3,
+		URGENCY_WATERMARK_MASK, wm_select);
+
+	REG_SET_2(DPG_PIPE_URGENCY_CONTROL, 0,
+		URGENCY_LOW_WATERMARK, urgency_low_wm,
+		URGENCY_HIGH_WATERMARK, urgency_high_wm);
+}
+#endif
+
+static void dce120_program_urgency_watermark(
+	struct dce_mem_input *dce_mi,
+	uint32_t wm_select,
+	uint32_t urgency_low_wm,
+	uint32_t urgency_high_wm)
+{
+	REG_UPDATE(DPG_WATERMARK_MASK_CONTROL,
+		URGENCY_WATERMARK_MASK, wm_select);
+
+	REG_SET_2(DPG_PIPE_URGENCY_CONTROL, 0,
+		URGENCY_LOW_WATERMARK, urgency_low_wm,
+		URGENCY_HIGH_WATERMARK, urgency_high_wm);
+
+	REG_SET_2(DPG_PIPE_URGENT_LEVEL_CONTROL, 0,
+		URGENT_LEVEL_LOW_WATERMARK, urgency_low_wm,
+		URGENT_LEVEL_HIGH_WATERMARK, urgency_high_wm);
+
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static void dce60_program_nbp_watermark(
+	struct dce_mem_input *dce_mi,
+	uint32_t wm_select,
+	uint32_t nbp_wm)
+{
+	REG_UPDATE(DPG_PIPE_NB_PSTATE_CHANGE_CONTROL,
+		NB_PSTATE_CHANGE_WATERMARK_MASK, wm_select);
+
+	REG_UPDATE_3(DPG_PIPE_NB_PSTATE_CHANGE_CONTROL,
+		NB_PSTATE_CHANGE_ENABLE, 1,
+		NB_PSTATE_CHANGE_URGENT_DURING_REQUEST, 1,
+		NB_PSTATE_CHANGE_NOT_SELF_REFRESH_DURING_REQUEST, 1);
+
+	REG_UPDATE(DPG_PIPE_NB_PSTATE_CHANGE_CONTROL,
+		NB_PSTATE_CHANGE_WATERMARK, nbp_wm);
+}
+#endif
+
+static void program_nbp_watermark(
+	struct dce_mem_input *dce_mi,
+	uint32_t wm_select,
+	uint32_t nbp_wm)
+{
+	if (REG(DPG_PIPE_NB_PSTATE_CHANGE_CONTROL)) {
+		REG_UPDATE(DPG_WATERMARK_MASK_CONTROL,
+				NB_PSTATE_CHANGE_WATERMARK_MASK, wm_select);
+
+		REG_UPDATE_3(DPG_PIPE_NB_PSTATE_CHANGE_CONTROL,
+				NB_PSTATE_CHANGE_ENABLE, 1,
+				NB_PSTATE_CHANGE_URGENT_DURING_REQUEST, 1,
+				NB_PSTATE_CHANGE_NOT_SELF_REFRESH_DURING_REQUEST, 1);
+
+		REG_UPDATE(DPG_PIPE_NB_PSTATE_CHANGE_CONTROL,
+				NB_PSTATE_CHANGE_WATERMARK, nbp_wm);
+	}
+
+	if (REG(DPG_PIPE_LOW_POWER_CONTROL)) {
+		REG_UPDATE(DPG_WATERMARK_MASK_CONTROL,
+				PSTATE_CHANGE_WATERMARK_MASK, wm_select);
+
+		REG_UPDATE_3(DPG_PIPE_LOW_POWER_CONTROL,
+				PSTATE_CHANGE_ENABLE, 1,
+				PSTATE_CHANGE_URGENT_DURING_REQUEST, 1,
+				PSTATE_CHANGE_NOT_SELF_REFRESH_DURING_REQUEST, 1);
+
+		REG_UPDATE(DPG_PIPE_LOW_POWER_CONTROL,
+				PSTATE_CHANGE_WATERMARK, nbp_wm);
+	}
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static void dce60_program_stutter_watermark(
+	struct dce_mem_input *dce_mi,
+	uint32_t wm_select,
+	uint32_t stutter_mark)
+{
+	REG_UPDATE(DPG_PIPE_STUTTER_CONTROL,
+		STUTTER_EXIT_SELF_REFRESH_WATERMARK_MASK, wm_select);
+
+	REG_UPDATE(DPG_PIPE_STUTTER_CONTROL,
+		STUTTER_EXIT_SELF_REFRESH_WATERMARK, stutter_mark);
+}
+#endif
+
+static void dce120_program_stutter_watermark(
+	struct dce_mem_input *dce_mi,
+	uint32_t wm_select,
+	uint32_t stutter_mark,
+	uint32_t stutter_entry)
+{
+	REG_UPDATE(DPG_WATERMARK_MASK_CONTROL,
+		STUTTER_EXIT_SELF_REFRESH_WATERMARK_MASK, wm_select);
+
+	if (REG(DPG_PIPE_STUTTER_CONTROL2))
+		REG_UPDATE_2(DPG_PIPE_STUTTER_CONTROL2,
+				STUTTER_EXIT_SELF_REFRESH_WATERMARK, stutter_mark,
+				STUTTER_ENTER_SELF_REFRESH_WATERMARK, stutter_entry);
+	else
+		REG_UPDATE_2(DPG_PIPE_STUTTER_CONTROL,
+				STUTTER_EXIT_SELF_REFRESH_WATERMARK, stutter_mark,
+				STUTTER_ENTER_SELF_REFRESH_WATERMARK, stutter_entry);
+}
+
+static void program_stutter_watermark(
+	struct dce_mem_input *dce_mi,
+	uint32_t wm_select,
+	uint32_t stutter_mark)
+{
+	REG_UPDATE(DPG_WATERMARK_MASK_CONTROL,
+		STUTTER_EXIT_SELF_REFRESH_WATERMARK_MASK, wm_select);
+
+	if (REG(DPG_PIPE_STUTTER_CONTROL2))
+		REG_UPDATE(DPG_PIPE_STUTTER_CONTROL2,
+				STUTTER_EXIT_SELF_REFRESH_WATERMARK, stutter_mark);
+	else
+		REG_UPDATE(DPG_PIPE_STUTTER_CONTROL,
+				STUTTER_EXIT_SELF_REFRESH_WATERMARK, stutter_mark);
+}
+
+static void dce_mi_program_display_marks(
+	struct mem_input *mi,
+	struct dce_watermarks nbp,
+	struct dce_watermarks stutter_exit,
+	struct dce_watermarks stutter_enter,
+	struct dce_watermarks urgent,
+	uint32_t total_dest_line_time_ns)
+{
+	struct dce_mem_input *dce_mi = TO_DCE_MEM_INPUT(mi);
+	uint32_t stutter_en = mi->ctx->dc->debug.disable_stutter ? 0 : 1;
+
+	program_urgency_watermark(dce_mi, 2, /* set a */
+			urgent.a_mark, total_dest_line_time_ns);
+	program_urgency_watermark(dce_mi, 1, /* set d */
+			urgent.d_mark, total_dest_line_time_ns);
+
+	REG_UPDATE_2(DPG_PIPE_STUTTER_CONTROL,
+		STUTTER_ENABLE, stutter_en,
+		STUTTER_IGNORE_FBC, 1);
+	program_nbp_watermark(dce_mi, 2, nbp.a_mark); /* set a */
+	program_nbp_watermark(dce_mi, 1, nbp.d_mark); /* set d */
+
+	program_stutter_watermark(dce_mi, 2, stutter_exit.a_mark); /* set a */
+	program_stutter_watermark(dce_mi, 1, stutter_exit.d_mark); /* set d */
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static void dce60_mi_program_display_marks(
+	struct mem_input *mi,
+	struct dce_watermarks nbp,
+	struct dce_watermarks stutter_exit,
+	struct dce_watermarks stutter_enter,
+	struct dce_watermarks urgent,
+	uint32_t total_dest_line_time_ns)
+{
+	struct dce_mem_input *dce_mi = TO_DCE_MEM_INPUT(mi);
+	uint32_t stutter_en = mi->ctx->dc->debug.disable_stutter ? 0 : 1;
+
+	dce60_program_urgency_watermark(dce_mi, 2, /* set a */
+			urgent.a_mark, total_dest_line_time_ns);
+	dce60_program_urgency_watermark(dce_mi, 1, /* set d */
+			urgent.d_mark, total_dest_line_time_ns);
+
+	REG_UPDATE_2(DPG_PIPE_STUTTER_CONTROL,
+		STUTTER_ENABLE, stutter_en,
+		STUTTER_IGNORE_FBC, 1);
+	dce60_program_nbp_watermark(dce_mi, 2, nbp.a_mark); /* set a */
+	dce60_program_nbp_watermark(dce_mi, 1, nbp.d_mark); /* set d */
+
+	dce60_program_stutter_watermark(dce_mi, 2, stutter_exit.a_mark); /* set a */
+	dce60_program_stutter_watermark(dce_mi, 1, stutter_exit.d_mark); /* set d */
+}
+#endif
+
+static void dce112_mi_program_display_marks(struct mem_input *mi,
+	struct dce_watermarks nbp,
+	struct dce_watermarks stutter_exit,
+	struct dce_watermarks stutter_entry,
+	struct dce_watermarks urgent,
+	uint32_t total_dest_line_time_ns)
+{
+	struct dce_mem_input *dce_mi = TO_DCE_MEM_INPUT(mi);
+	uint32_t stutter_en = mi->ctx->dc->debug.disable_stutter ? 0 : 1;
+
+	program_urgency_watermark(dce_mi, 0, /* set a */
+			urgent.a_mark, total_dest_line_time_ns);
+	program_urgency_watermark(dce_mi, 1, /* set b */
+			urgent.b_mark, total_dest_line_time_ns);
+	program_urgency_watermark(dce_mi, 2, /* set c */
+			urgent.c_mark, total_dest_line_time_ns);
+	program_urgency_watermark(dce_mi, 3, /* set d */
+			urgent.d_mark, total_dest_line_time_ns);
+
+	REG_UPDATE_2(DPG_PIPE_STUTTER_CONTROL,
+		STUTTER_ENABLE, stutter_en,
+		STUTTER_IGNORE_FBC, 1);
+	program_nbp_watermark(dce_mi, 0, nbp.a_mark); /* set a */
+	program_nbp_watermark(dce_mi, 1, nbp.b_mark); /* set b */
+	program_nbp_watermark(dce_mi, 2, nbp.c_mark); /* set c */
+	program_nbp_watermark(dce_mi, 3, nbp.d_mark); /* set d */
+
+	program_stutter_watermark(dce_mi, 0, stutter_exit.a_mark); /* set a */
+	program_stutter_watermark(dce_mi, 1, stutter_exit.b_mark); /* set b */
+	program_stutter_watermark(dce_mi, 2, stutter_exit.c_mark); /* set c */
+	program_stutter_watermark(dce_mi, 3, stutter_exit.d_mark); /* set d */
+}
+
+static void dce120_mi_program_display_marks(struct mem_input *mi,
+	struct dce_watermarks nbp,
+	struct dce_watermarks stutter_exit,
+	struct dce_watermarks stutter_entry,
+	struct dce_watermarks urgent,
+	uint32_t total_dest_line_time_ns)
+{
+	struct dce_mem_input *dce_mi = TO_DCE_MEM_INPUT(mi);
+	uint32_t stutter_en = mi->ctx->dc->debug.disable_stutter ? 0 : 1;
+
+	dce120_program_urgency_watermark(dce_mi, 0, /* set a */
+			urgent.a_mark, total_dest_line_time_ns);
+	dce120_program_urgency_watermark(dce_mi, 1, /* set b */
+			urgent.b_mark, total_dest_line_time_ns);
+	dce120_program_urgency_watermark(dce_mi, 2, /* set c */
+			urgent.c_mark, total_dest_line_time_ns);
+	dce120_program_urgency_watermark(dce_mi, 3, /* set d */
+			urgent.d_mark, total_dest_line_time_ns);
+
+	REG_UPDATE_2(DPG_PIPE_STUTTER_CONTROL,
+		STUTTER_ENABLE, stutter_en,
+		STUTTER_IGNORE_FBC, 1);
+	program_nbp_watermark(dce_mi, 0, nbp.a_mark); /* set a */
+	program_nbp_watermark(dce_mi, 1, nbp.b_mark); /* set b */
+	program_nbp_watermark(dce_mi, 2, nbp.c_mark); /* set c */
+	program_nbp_watermark(dce_mi, 3, nbp.d_mark); /* set d */
+
+	dce120_program_stutter_watermark(dce_mi, 0, stutter_exit.a_mark, stutter_entry.a_mark); /* set a */
+	dce120_program_stutter_watermark(dce_mi, 1, stutter_exit.b_mark, stutter_entry.b_mark); /* set b */
+	dce120_program_stutter_watermark(dce_mi, 2, stutter_exit.c_mark, stutter_entry.c_mark); /* set c */
+	dce120_program_stutter_watermark(dce_mi, 3, stutter_exit.d_mark, stutter_entry.d_mark); /* set d */
+}
+
+static void program_tiling(
+	struct dce_mem_input *dce_mi, const struct dc_tiling_info *info)
+{
+	if (dce_mi->masks->GRPH_SW_MODE) { /* GFX9 */
+		REG_UPDATE_6(GRPH_CONTROL,
+				GRPH_SW_MODE, info->gfx9.swizzle,
+				GRPH_NUM_BANKS, log_2(info->gfx9.num_banks),
+				GRPH_NUM_SHADER_ENGINES, log_2(info->gfx9.num_shader_engines),
+				GRPH_NUM_PIPES, log_2(info->gfx9.num_pipes),
+				GRPH_COLOR_EXPANSION_MODE, 1,
+				GRPH_SE_ENABLE, info->gfx9.shaderEnable);
+		/* TODO: DCP0_GRPH_CONTROL__GRPH_SE_ENABLE where to get info
+		GRPH_SE_ENABLE, 1,
+		GRPH_Z, 0);
+		 */
+	}
+
+	if (dce_mi->masks->GRPH_MICRO_TILE_MODE) { /* GFX8 */
+		REG_UPDATE_9(GRPH_CONTROL,
+				GRPH_NUM_BANKS, info->gfx8.num_banks,
+				GRPH_BANK_WIDTH, info->gfx8.bank_width,
+				GRPH_BANK_HEIGHT, info->gfx8.bank_height,
+				GRPH_MACRO_TILE_ASPECT, info->gfx8.tile_aspect,
+				GRPH_TILE_SPLIT, info->gfx8.tile_split,
+				GRPH_MICRO_TILE_MODE, info->gfx8.tile_mode,
+				GRPH_PIPE_CONFIG, info->gfx8.pipe_config,
+				GRPH_ARRAY_MODE, info->gfx8.array_mode,
+				GRPH_COLOR_EXPANSION_MODE, 1);
+		/* 01 - DCP_GRPH_COLOR_EXPANSION_MODE_ZEXP: zero expansion for YCbCr */
+		/*
+				GRPH_Z, 0);
+				*/
+	}
+
+	if (dce_mi->masks->GRPH_ARRAY_MODE) { /* GFX6 but reuses gfx8 struct */
+		REG_UPDATE_8(GRPH_CONTROL,
+				GRPH_NUM_BANKS, info->gfx8.num_banks,
+				GRPH_BANK_WIDTH, info->gfx8.bank_width,
+				GRPH_BANK_HEIGHT, info->gfx8.bank_height,
+				GRPH_MACRO_TILE_ASPECT, info->gfx8.tile_aspect,
+				GRPH_TILE_SPLIT, info->gfx8.tile_split,
+				/* DCE6 has no GRPH_MICRO_TILE_MODE mask */
+				GRPH_PIPE_CONFIG, info->gfx8.pipe_config,
+				GRPH_ARRAY_MODE, info->gfx8.array_mode,
+				GRPH_COLOR_EXPANSION_MODE, 1);
+		/* 01 - DCP_GRPH_COLOR_EXPANSION_MODE_ZEXP: zero expansion for YCbCr */
+		/*
+				GRPH_Z, 0);
+				*/
+	}
+}
+
+static void program_size_and_rotation(
+	struct dce_mem_input *dce_mi,
+	enum dc_rotation_angle rotation,
+	const struct plane_size *plane_size)
+{
+	const struct rect *in_rect = &plane_size->surface_size;
+	struct rect hw_rect = plane_size->surface_size;
+	const uint32_t rotation_angles[ROTATION_ANGLE_COUNT] = {
+			[ROTATION_ANGLE_0] = 0,
+			[ROTATION_ANGLE_90] = 1,
+			[ROTATION_ANGLE_180] = 2,
+			[ROTATION_ANGLE_270] = 3,
+	};
+
+	if (rotation == ROTATION_ANGLE_90 || rotation == ROTATION_ANGLE_270) {
+		hw_rect.x = in_rect->y;
+		hw_rect.y = in_rect->x;
+
+		hw_rect.height = in_rect->width;
+		hw_rect.width = in_rect->height;
+	}
+
+	REG_SET(GRPH_X_START, 0,
+			GRPH_X_START, hw_rect.x);
+
+	REG_SET(GRPH_Y_START, 0,
+			GRPH_Y_START, hw_rect.y);
+
+	REG_SET(GRPH_X_END, 0,
+			GRPH_X_END, hw_rect.width);
+
+	REG_SET(GRPH_Y_END, 0,
+			GRPH_Y_END, hw_rect.height);
+
+	REG_SET(GRPH_PITCH, 0,
+			GRPH_PITCH, plane_size->surface_pitch);
+
+	REG_SET(HW_ROTATION, 0,
+			GRPH_ROTATION_ANGLE, rotation_angles[rotation]);
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static void dce60_program_size(
+	struct dce_mem_input *dce_mi,
+	enum dc_rotation_angle rotation, /* not used in DCE6 */
+	const struct plane_size *plane_size)
+{
+	struct rect hw_rect = plane_size->surface_size;
+	/* DCE6 has no HW rotation, skip rotation_angles declaration */
+
+	/* DCE6 has no HW rotation, skip ROTATION_ANGLE_* processing */
+
+	REG_SET(GRPH_X_START, 0,
+			GRPH_X_START, hw_rect.x);
+
+	REG_SET(GRPH_Y_START, 0,
+			GRPH_Y_START, hw_rect.y);
+
+	REG_SET(GRPH_X_END, 0,
+			GRPH_X_END, hw_rect.width);
+
+	REG_SET(GRPH_Y_END, 0,
+			GRPH_Y_END, hw_rect.height);
+
+	REG_SET(GRPH_PITCH, 0,
+			GRPH_PITCH, plane_size->surface_pitch);
+
+	/* DCE6 has no HW_ROTATION register, skip setting rotation_angles */
+}
+#endif
+
+static void program_grph_pixel_format(
+	struct dce_mem_input *dce_mi,
+	enum surface_pixel_format format)
+{
+	uint32_t red_xbar = 0, blue_xbar = 0; /* no swap */
+	uint32_t grph_depth = 0, grph_format = 0;
+	uint32_t sign = 0, floating = 0;
+
+	if (format == SURFACE_PIXEL_FORMAT_GRPH_ABGR8888 ||
+			/*todo: doesn't look like we handle BGRA here,
+			 *  should problem swap endian*/
+		format == SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010 ||
+		format == SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010_XR_BIAS ||
+		format == SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616 ||
+		format == SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F) {
+		/* ABGR formats */
+		red_xbar = 2;
+		blue_xbar = 2;
+	}
+
+	REG_SET_2(GRPH_SWAP_CNTL, 0,
+			GRPH_RED_CROSSBAR, red_xbar,
+			GRPH_BLUE_CROSSBAR, blue_xbar);
+
+	switch (format) {
+	case SURFACE_PIXEL_FORMAT_GRPH_PALETA_256_COLORS:
+		grph_depth = 0;
+		grph_format = 0;
+		break;
+	case SURFACE_PIXEL_FORMAT_GRPH_ARGB1555:
+		grph_depth = 1;
+		grph_format = 0;
+		break;
+	case SURFACE_PIXEL_FORMAT_GRPH_RGB565:
+		grph_depth = 1;
+		grph_format = 1;
+		break;
+	case SURFACE_PIXEL_FORMAT_GRPH_ARGB8888:
+	case SURFACE_PIXEL_FORMAT_GRPH_ABGR8888:
+		grph_depth = 2;
+		grph_format = 0;
+		break;
+	case SURFACE_PIXEL_FORMAT_GRPH_ARGB2101010:
+	case SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010:
+	case SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010_XR_BIAS:
+		grph_depth = 2;
+		grph_format = 1;
+		break;
+	case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F:
+		sign = 1;
+		floating = 1;
+		fallthrough;
+	case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616F: /* shouldn't this get float too? */
+	case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616:
+	case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616:
+		grph_depth = 3;
+		grph_format = 0;
+		break;
+	default:
+		DC_ERR("unsupported grph pixel format");
+		break;
+	}
+
+	REG_UPDATE_2(GRPH_CONTROL,
+			GRPH_DEPTH, grph_depth,
+			GRPH_FORMAT, grph_format);
+
+	REG_UPDATE_4(PRESCALE_GRPH_CONTROL,
+			GRPH_PRESCALE_SELECT, floating,
+			GRPH_PRESCALE_R_SIGN, sign,
+			GRPH_PRESCALE_G_SIGN, sign,
+			GRPH_PRESCALE_B_SIGN, sign);
+}
+
+static void dce_mi_clear_tiling(
+	struct mem_input *mi)
+{
+	struct dce_mem_input *dce_mi = TO_DCE_MEM_INPUT(mi);
+
+	if (dce_mi->masks->GRPH_SW_MODE) { /* GFX9 */
+		REG_UPDATE(GRPH_CONTROL,
+			   GRPH_SW_MODE, DC_SW_LINEAR);
+	}
+
+	if (dce_mi->masks->GRPH_MICRO_TILE_MODE) { /* GFX8 */
+		REG_UPDATE(GRPH_CONTROL,
+			   GRPH_ARRAY_MODE, DC_SW_LINEAR);
+	}
+
+	if (dce_mi->masks->GRPH_ARRAY_MODE) { /* GFX6 but reuses gfx8 struct */
+		REG_UPDATE(GRPH_CONTROL,
+			   GRPH_ARRAY_MODE, DC_SW_LINEAR);
+	}
+}
+
+static void dce_mi_program_surface_config(
+	struct mem_input *mi,
+	enum surface_pixel_format format,
+	struct dc_tiling_info *tiling_info,
+	struct plane_size *plane_size,
+	enum dc_rotation_angle rotation,
+	struct dc_plane_dcc_param *dcc,
+	bool horizontal_mirror)
+{
+	struct dce_mem_input *dce_mi = TO_DCE_MEM_INPUT(mi);
+	REG_UPDATE(GRPH_ENABLE, GRPH_ENABLE, 1);
+
+	program_tiling(dce_mi, tiling_info);
+	program_size_and_rotation(dce_mi, rotation, plane_size);
+
+	if (format < SURFACE_PIXEL_FORMAT_VIDEO_BEGIN)
+		program_grph_pixel_format(dce_mi, format);
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static void dce60_mi_program_surface_config(
+	struct mem_input *mi,
+	enum surface_pixel_format format,
+	struct dc_tiling_info *tiling_info,
+	struct plane_size *plane_size,
+	enum dc_rotation_angle rotation, /* not used in DCE6 */
+	struct dc_plane_dcc_param *dcc,
+	bool horizontal_mirror)
+{
+	struct dce_mem_input *dce_mi = TO_DCE_MEM_INPUT(mi);
+	REG_UPDATE(GRPH_ENABLE, GRPH_ENABLE, 1);
+
+	program_tiling(dce_mi, tiling_info);
+	dce60_program_size(dce_mi, rotation, plane_size);
+
+	if (format < SURFACE_PIXEL_FORMAT_VIDEO_BEGIN)
+		program_grph_pixel_format(dce_mi, format);
+}
+#endif
+
+static uint32_t get_dmif_switch_time_us(
+	uint32_t h_total,
+	uint32_t v_total,
+	uint32_t pix_clk_khz)
+{
+	uint32_t frame_time;
+	uint32_t pixels_per_second;
+	uint32_t pixels_per_frame;
+	uint32_t refresh_rate;
+	const uint32_t us_in_sec = 1000000;
+	const uint32_t min_single_frame_time_us = 30000;
+	/*return double of frame time*/
+	const uint32_t single_frame_time_multiplier = 2;
+
+	if (!h_total || v_total || !pix_clk_khz)
+		return single_frame_time_multiplier * min_single_frame_time_us;
+
+	/*TODO: should we use pixel format normalized pixel clock here?*/
+	pixels_per_second = pix_clk_khz * 1000;
+	pixels_per_frame = h_total * v_total;
+
+	if (!pixels_per_second || !pixels_per_frame) {
+		/* avoid division by zero */
+		ASSERT(pixels_per_frame);
+		ASSERT(pixels_per_second);
+		return single_frame_time_multiplier * min_single_frame_time_us;
+	}
+
+	refresh_rate = pixels_per_second / pixels_per_frame;
+
+	if (!refresh_rate) {
+		/* avoid division by zero*/
+		ASSERT(refresh_rate);
+		return single_frame_time_multiplier * min_single_frame_time_us;
+	}
+
+	frame_time = us_in_sec / refresh_rate;
+
+	if (frame_time < min_single_frame_time_us)
+		frame_time = min_single_frame_time_us;
+
+	frame_time *= single_frame_time_multiplier;
+
+	return frame_time;
+}
+
+static void dce_mi_allocate_dmif(
+	struct mem_input *mi,
+	uint32_t h_total,
+	uint32_t v_total,
+	uint32_t pix_clk_khz,
+	uint32_t total_stream_num)
+{
+	struct dce_mem_input *dce_mi = TO_DCE_MEM_INPUT(mi);
+	const uint32_t retry_delay = 10;
+	uint32_t retry_count = get_dmif_switch_time_us(
+			h_total,
+			v_total,
+			pix_clk_khz) / retry_delay;
+
+	uint32_t pix_dur;
+	uint32_t buffers_allocated;
+	uint32_t dmif_buffer_control;
+
+	dmif_buffer_control = REG_GET(DMIF_BUFFER_CONTROL,
+			DMIF_BUFFERS_ALLOCATED, &buffers_allocated);
+
+	if (buffers_allocated == 2)
+		return;
+
+	REG_SET(DMIF_BUFFER_CONTROL, dmif_buffer_control,
+			DMIF_BUFFERS_ALLOCATED, 2);
+
+	REG_WAIT(DMIF_BUFFER_CONTROL,
+			DMIF_BUFFERS_ALLOCATION_COMPLETED, 1,
+			retry_delay, retry_count);
+
+	if (pix_clk_khz != 0) {
+		pix_dur = 1000000000ULL / pix_clk_khz;
+
+		REG_UPDATE(DPG_PIPE_ARBITRATION_CONTROL1,
+			PIXEL_DURATION, pix_dur);
+	}
+
+	if (dce_mi->wa.single_head_rdreq_dmif_limit) {
+		uint32_t enable =  (total_stream_num > 1) ? 0 :
+				dce_mi->wa.single_head_rdreq_dmif_limit;
+
+		REG_UPDATE(MC_HUB_RDREQ_DMIF_LIMIT,
+				ENABLE, enable);
+	}
+}
+
+static void dce_mi_free_dmif(
+		struct mem_input *mi,
+		uint32_t total_stream_num)
+{
+	struct dce_mem_input *dce_mi = TO_DCE_MEM_INPUT(mi);
+	uint32_t buffers_allocated;
+	uint32_t dmif_buffer_control;
+
+	dmif_buffer_control = REG_GET(DMIF_BUFFER_CONTROL,
+			DMIF_BUFFERS_ALLOCATED, &buffers_allocated);
+
+	if (buffers_allocated == 0)
+		return;
+
+	REG_SET(DMIF_BUFFER_CONTROL, dmif_buffer_control,
+			DMIF_BUFFERS_ALLOCATED, 0);
+
+	REG_WAIT(DMIF_BUFFER_CONTROL,
+			DMIF_BUFFERS_ALLOCATION_COMPLETED, 1,
+			10, 3500);
+
+	if (dce_mi->wa.single_head_rdreq_dmif_limit) {
+		uint32_t enable =  (total_stream_num > 1) ? 0 :
+				dce_mi->wa.single_head_rdreq_dmif_limit;
+
+		REG_UPDATE(MC_HUB_RDREQ_DMIF_LIMIT,
+				ENABLE, enable);
+	}
+}
+
+
+static void program_sec_addr(
+	struct dce_mem_input *dce_mi,
+	PHYSICAL_ADDRESS_LOC address)
+{
+	/*high register MUST be programmed first*/
+	REG_SET(GRPH_SECONDARY_SURFACE_ADDRESS_HIGH, 0,
+		GRPH_SECONDARY_SURFACE_ADDRESS_HIGH,
+		address.high_part);
+
+	REG_SET_2(GRPH_SECONDARY_SURFACE_ADDRESS, 0,
+		GRPH_SECONDARY_SURFACE_ADDRESS, address.low_part >> 8,
+		GRPH_SECONDARY_DFQ_ENABLE, 0);
+}
+
+static void program_pri_addr(
+	struct dce_mem_input *dce_mi,
+	PHYSICAL_ADDRESS_LOC address)
+{
+	/*high register MUST be programmed first*/
+	REG_SET(GRPH_PRIMARY_SURFACE_ADDRESS_HIGH, 0,
+		GRPH_PRIMARY_SURFACE_ADDRESS_HIGH,
+		address.high_part);
+
+	REG_SET(GRPH_PRIMARY_SURFACE_ADDRESS, 0,
+		GRPH_PRIMARY_SURFACE_ADDRESS,
+		address.low_part >> 8);
+}
+
+
+static bool dce_mi_is_flip_pending(struct mem_input *mem_input)
+{
+	struct dce_mem_input *dce_mi = TO_DCE_MEM_INPUT(mem_input);
+	uint32_t update_pending;
+
+	REG_GET(GRPH_UPDATE, GRPH_SURFACE_UPDATE_PENDING, &update_pending);
+	if (update_pending)
+		return true;
+
+	mem_input->current_address = mem_input->request_address;
+	return false;
+}
+
+static bool dce_mi_program_surface_flip_and_addr(
+	struct mem_input *mem_input,
+	const struct dc_plane_address *address,
+	bool flip_immediate)
+{
+	struct dce_mem_input *dce_mi = TO_DCE_MEM_INPUT(mem_input);
+
+	REG_UPDATE(GRPH_UPDATE, GRPH_UPDATE_LOCK, 1);
+
+	REG_UPDATE(
+		GRPH_FLIP_CONTROL,
+		GRPH_SURFACE_UPDATE_H_RETRACE_EN, flip_immediate ? 1 : 0);
+
+	switch (address->type) {
+	case PLN_ADDR_TYPE_GRAPHICS:
+		if (address->grph.addr.quad_part == 0)
+			break;
+		program_pri_addr(dce_mi, address->grph.addr);
+		break;
+	case PLN_ADDR_TYPE_GRPH_STEREO:
+		if (address->grph_stereo.left_addr.quad_part == 0 ||
+		    address->grph_stereo.right_addr.quad_part == 0)
+			break;
+		program_pri_addr(dce_mi, address->grph_stereo.left_addr);
+		program_sec_addr(dce_mi, address->grph_stereo.right_addr);
+		break;
+	default:
+		/* not supported */
+		BREAK_TO_DEBUGGER();
+		break;
+	}
+
+	mem_input->request_address = *address;
+
+	if (flip_immediate)
+		mem_input->current_address = *address;
+
+	REG_UPDATE(GRPH_UPDATE, GRPH_UPDATE_LOCK, 0);
+
+	return true;
+}
+
+static const struct mem_input_funcs dce_mi_funcs = {
+	.mem_input_program_display_marks = dce_mi_program_display_marks,
+	.allocate_mem_input = dce_mi_allocate_dmif,
+	.free_mem_input = dce_mi_free_dmif,
+	.mem_input_program_surface_flip_and_addr =
+			dce_mi_program_surface_flip_and_addr,
+	.mem_input_program_pte_vm = dce_mi_program_pte_vm,
+	.mem_input_program_surface_config =
+			dce_mi_program_surface_config,
+	.mem_input_is_flip_pending = dce_mi_is_flip_pending,
+	.mem_input_clear_tiling = dce_mi_clear_tiling,
+};
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static const struct mem_input_funcs dce60_mi_funcs = {
+	.mem_input_program_display_marks = dce60_mi_program_display_marks,
+	.allocate_mem_input = dce_mi_allocate_dmif,
+	.free_mem_input = dce_mi_free_dmif,
+	.mem_input_program_surface_flip_and_addr =
+			dce_mi_program_surface_flip_and_addr,
+	.mem_input_program_pte_vm = dce_mi_program_pte_vm,
+	.mem_input_program_surface_config =
+			dce60_mi_program_surface_config,
+	.mem_input_is_flip_pending = dce_mi_is_flip_pending,
+	.mem_input_clear_tiling = dce_mi_clear_tiling,
+};
+#endif
+
+static const struct mem_input_funcs dce112_mi_funcs = {
+	.mem_input_program_display_marks = dce112_mi_program_display_marks,
+	.allocate_mem_input = dce_mi_allocate_dmif,
+	.free_mem_input = dce_mi_free_dmif,
+	.mem_input_program_surface_flip_and_addr =
+			dce_mi_program_surface_flip_and_addr,
+	.mem_input_program_pte_vm = dce_mi_program_pte_vm,
+	.mem_input_program_surface_config =
+			dce_mi_program_surface_config,
+	.mem_input_is_flip_pending = dce_mi_is_flip_pending,
+	.mem_input_clear_tiling = dce_mi_clear_tiling,
+};
+
+static const struct mem_input_funcs dce120_mi_funcs = {
+	.mem_input_program_display_marks = dce120_mi_program_display_marks,
+	.allocate_mem_input = dce_mi_allocate_dmif,
+	.free_mem_input = dce_mi_free_dmif,
+	.mem_input_program_surface_flip_and_addr =
+			dce_mi_program_surface_flip_and_addr,
+	.mem_input_program_pte_vm = dce_mi_program_pte_vm,
+	.mem_input_program_surface_config =
+			dce_mi_program_surface_config,
+	.mem_input_is_flip_pending = dce_mi_is_flip_pending,
+	.mem_input_clear_tiling = dce_mi_clear_tiling,
+};
+
+void dce_mem_input_construct(
+	struct dce_mem_input *dce_mi,
+	struct dc_context *ctx,
+	int inst,
+	const struct dce_mem_input_registers *regs,
+	const struct dce_mem_input_shift *mi_shift,
+	const struct dce_mem_input_mask *mi_mask)
+{
+	dce_mi->base.ctx = ctx;
+
+	dce_mi->base.inst = inst;
+	dce_mi->base.funcs = &dce_mi_funcs;
+
+	dce_mi->regs = regs;
+	dce_mi->shifts = mi_shift;
+	dce_mi->masks = mi_mask;
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+void dce60_mem_input_construct(
+	struct dce_mem_input *dce_mi,
+	struct dc_context *ctx,
+	int inst,
+	const struct dce_mem_input_registers *regs,
+	const struct dce_mem_input_shift *mi_shift,
+	const struct dce_mem_input_mask *mi_mask)
+{
+	dce_mem_input_construct(dce_mi, ctx, inst, regs, mi_shift, mi_mask);
+	dce_mi->base.funcs = &dce60_mi_funcs;
+}
+#endif
+
+void dce112_mem_input_construct(
+	struct dce_mem_input *dce_mi,
+	struct dc_context *ctx,
+	int inst,
+	const struct dce_mem_input_registers *regs,
+	const struct dce_mem_input_shift *mi_shift,
+	const struct dce_mem_input_mask *mi_mask)
+{
+	dce_mem_input_construct(dce_mi, ctx, inst, regs, mi_shift, mi_mask);
+	dce_mi->base.funcs = &dce112_mi_funcs;
+}
+
+void dce120_mem_input_construct(
+	struct dce_mem_input *dce_mi,
+	struct dc_context *ctx,
+	int inst,
+	const struct dce_mem_input_registers *regs,
+	const struct dce_mem_input_shift *mi_shift,
+	const struct dce_mem_input_mask *mi_mask)
+{
+	dce_mem_input_construct(dce_mi, ctx, inst, regs, mi_shift, mi_mask);
+	dce_mi->base.funcs = &dce120_mi_funcs;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_mem_input.h b/drivers/gpu/drm/amd/display/dc/dce/dce_mem_input.h
new file mode 100644
index 000000000000..e34e445a4013
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_mem_input.h
@@ -0,0 +1,467 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+#ifndef __DCE_MEM_INPUT_H__
+#define __DCE_MEM_INPUT_H__
+
+#include "dc_hw_types.h"
+#include "mem_input.h"
+
+#define TO_DCE_MEM_INPUT(mem_input)\
+	container_of(mem_input, struct dce_mem_input, base)
+
+#define MI_DCE_BASE_REG_LIST(id)\
+	SRI(GRPH_ENABLE, DCP, id),\
+	SRI(GRPH_CONTROL, DCP, id),\
+	SRI(GRPH_X_START, DCP, id),\
+	SRI(GRPH_Y_START, DCP, id),\
+	SRI(GRPH_X_END, DCP, id),\
+	SRI(GRPH_Y_END, DCP, id),\
+	SRI(GRPH_PITCH, DCP, id),\
+	SRI(HW_ROTATION, DCP, id),\
+	SRI(GRPH_SWAP_CNTL, DCP, id),\
+	SRI(PRESCALE_GRPH_CONTROL, DCP, id),\
+	SRI(GRPH_UPDATE, DCP, id),\
+	SRI(GRPH_FLIP_CONTROL, DCP, id),\
+	SRI(GRPH_PRIMARY_SURFACE_ADDRESS, DCP, id),\
+	SRI(GRPH_PRIMARY_SURFACE_ADDRESS_HIGH, DCP, id),\
+	SRI(GRPH_SECONDARY_SURFACE_ADDRESS, DCP, id),\
+	SRI(GRPH_SECONDARY_SURFACE_ADDRESS_HIGH, DCP, id),\
+	SRI(DPG_PIPE_ARBITRATION_CONTROL1, DMIF_PG, id),\
+	SRI(DPG_WATERMARK_MASK_CONTROL, DMIF_PG, id),\
+	SRI(DPG_PIPE_URGENCY_CONTROL, DMIF_PG, id),\
+	SRI(DPG_PIPE_STUTTER_CONTROL, DMIF_PG, id),\
+	SRI(DMIF_BUFFER_CONTROL, PIPE, id)
+
+#define MI_DCE_PTE_REG_LIST(id)\
+	SRI(DVMM_PTE_CONTROL, DCP, id),\
+	SRI(DVMM_PTE_ARB_CONTROL, DCP, id)
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+#define MI_DCE6_REG_LIST(id)\
+	SRI(GRPH_ENABLE, DCP, id),\
+	SRI(GRPH_CONTROL, DCP, id),\
+	SRI(GRPH_X_START, DCP, id),\
+	SRI(GRPH_Y_START, DCP, id),\
+	SRI(GRPH_X_END, DCP, id),\
+	SRI(GRPH_Y_END, DCP, id),\
+	SRI(GRPH_PITCH, DCP, id),\
+	SRI(GRPH_SWAP_CNTL, DCP, id),\
+	SRI(PRESCALE_GRPH_CONTROL, DCP, id),\
+	SRI(GRPH_UPDATE, DCP, id),\
+	SRI(GRPH_FLIP_CONTROL, DCP, id),\
+	SRI(GRPH_PRIMARY_SURFACE_ADDRESS, DCP, id),\
+	SRI(GRPH_PRIMARY_SURFACE_ADDRESS_HIGH, DCP, id),\
+	SRI(GRPH_SECONDARY_SURFACE_ADDRESS, DCP, id),\
+	SRI(GRPH_SECONDARY_SURFACE_ADDRESS_HIGH, DCP, id),\
+	SRI(DPG_PIPE_ARBITRATION_CONTROL1, DMIF_PG, id),\
+	SRI(DPG_PIPE_ARBITRATION_CONTROL3, DMIF_PG, id),\
+	SRI(DPG_PIPE_NB_PSTATE_CHANGE_CONTROL, DMIF_PG, id),\
+	SRI(DPG_PIPE_URGENCY_CONTROL, DMIF_PG, id),\
+	SRI(DPG_PIPE_STUTTER_CONTROL, DMIF_PG, id),\
+	SRI(DMIF_BUFFER_CONTROL, PIPE, id)
+#endif
+
+#define MI_DCE8_REG_LIST(id)\
+	MI_DCE_BASE_REG_LIST(id),\
+	SRI(DPG_PIPE_NB_PSTATE_CHANGE_CONTROL, DMIF_PG, id)
+
+#define MI_DCE11_2_REG_LIST(id)\
+	MI_DCE8_REG_LIST(id),\
+	SRI(GRPH_PIPE_OUTSTANDING_REQUEST_LIMIT, DCP, id)
+
+#define MI_DCE11_REG_LIST(id)\
+	MI_DCE11_2_REG_LIST(id),\
+	MI_DCE_PTE_REG_LIST(id)
+
+#define MI_DCE12_REG_LIST(id)\
+	MI_DCE_BASE_REG_LIST(id),\
+	MI_DCE_PTE_REG_LIST(id),\
+	SRI(GRPH_PIPE_OUTSTANDING_REQUEST_LIMIT, DCP, id),\
+	SRI(DPG_PIPE_STUTTER_CONTROL2, DMIF_PG, id),\
+	SRI(DPG_PIPE_LOW_POWER_CONTROL, DMIF_PG, id),\
+	SR(DCHUB_FB_LOCATION),\
+	SR(DCHUB_AGP_BASE),\
+	SR(DCHUB_AGP_BOT),\
+	SR(DCHUB_AGP_TOP)
+
+struct dce_mem_input_registers {
+	/* DCP */
+	uint32_t GRPH_ENABLE;
+	uint32_t GRPH_CONTROL;
+	uint32_t GRPH_X_START;
+	uint32_t GRPH_Y_START;
+	uint32_t GRPH_X_END;
+	uint32_t GRPH_Y_END;
+	uint32_t GRPH_PITCH;
+	uint32_t HW_ROTATION;
+	uint32_t GRPH_SWAP_CNTL;
+	uint32_t PRESCALE_GRPH_CONTROL;
+	uint32_t GRPH_PIPE_OUTSTANDING_REQUEST_LIMIT;
+	uint32_t DVMM_PTE_CONTROL;
+	uint32_t DVMM_PTE_ARB_CONTROL;
+	uint32_t GRPH_UPDATE;
+	uint32_t GRPH_FLIP_CONTROL;
+	uint32_t GRPH_PRIMARY_SURFACE_ADDRESS;
+	uint32_t GRPH_PRIMARY_SURFACE_ADDRESS_HIGH;
+	uint32_t GRPH_SECONDARY_SURFACE_ADDRESS;
+	uint32_t GRPH_SECONDARY_SURFACE_ADDRESS_HIGH;
+	/* DMIF_PG */
+	uint32_t DPG_PIPE_ARBITRATION_CONTROL1;
+#if defined(CONFIG_DRM_AMD_DC_SI)
+	uint32_t DPG_PIPE_ARBITRATION_CONTROL3;
+#endif
+	uint32_t DPG_WATERMARK_MASK_CONTROL;
+	uint32_t DPG_PIPE_URGENCY_CONTROL;
+	uint32_t DPG_PIPE_URGENT_LEVEL_CONTROL;
+	uint32_t DPG_PIPE_NB_PSTATE_CHANGE_CONTROL;
+	uint32_t DPG_PIPE_LOW_POWER_CONTROL;
+	uint32_t DPG_PIPE_STUTTER_CONTROL;
+	uint32_t DPG_PIPE_STUTTER_CONTROL2;
+	/* DCI */
+	uint32_t DMIF_BUFFER_CONTROL;
+	/* MC_HUB */
+	uint32_t MC_HUB_RDREQ_DMIF_LIMIT;
+	/*DCHUB*/
+	uint32_t DCHUB_FB_LOCATION;
+	uint32_t DCHUB_AGP_BASE;
+	uint32_t DCHUB_AGP_BOT;
+	uint32_t DCHUB_AGP_TOP;
+};
+
+/* Set_Filed_for_Block */
+#define SFB(blk_name, reg_name, field_name, post_fix)\
+	.field_name = blk_name ## reg_name ## __ ## field_name ## post_fix
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+#define MI_GFX6_TILE_MASK_SH_LIST(mask_sh, blk)\
+	SFB(blk, GRPH_CONTROL, GRPH_NUM_BANKS, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_BANK_WIDTH, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_BANK_HEIGHT, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_MACRO_TILE_ASPECT, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_TILE_SPLIT, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_PIPE_CONFIG, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_ARRAY_MODE, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_COLOR_EXPANSION_MODE, mask_sh)
+#endif
+
+#define MI_GFX8_TILE_MASK_SH_LIST(mask_sh, blk)\
+	SFB(blk, GRPH_CONTROL, GRPH_NUM_BANKS, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_BANK_WIDTH, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_BANK_HEIGHT, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_MACRO_TILE_ASPECT, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_TILE_SPLIT, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_MICRO_TILE_MODE, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_PIPE_CONFIG, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_ARRAY_MODE, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_COLOR_EXPANSION_MODE, mask_sh)
+
+#define MI_DCP_MASK_SH_LIST(mask_sh, blk)\
+	SFB(blk, GRPH_ENABLE, GRPH_ENABLE, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_DEPTH, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_FORMAT, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_NUM_BANKS, mask_sh),\
+	SFB(blk, GRPH_X_START, GRPH_X_START, mask_sh),\
+	SFB(blk, GRPH_Y_START, GRPH_Y_START, mask_sh),\
+	SFB(blk, GRPH_X_END, GRPH_X_END, mask_sh),\
+	SFB(blk, GRPH_Y_END, GRPH_Y_END, mask_sh),\
+	SFB(blk, GRPH_PITCH, GRPH_PITCH, mask_sh),\
+	SFB(blk, HW_ROTATION, GRPH_ROTATION_ANGLE, mask_sh),\
+	SFB(blk, GRPH_SWAP_CNTL, GRPH_RED_CROSSBAR, mask_sh),\
+	SFB(blk, GRPH_SWAP_CNTL, GRPH_BLUE_CROSSBAR, mask_sh),\
+	SFB(blk, PRESCALE_GRPH_CONTROL, GRPH_PRESCALE_SELECT, mask_sh),\
+	SFB(blk, PRESCALE_GRPH_CONTROL, GRPH_PRESCALE_R_SIGN, mask_sh),\
+	SFB(blk, PRESCALE_GRPH_CONTROL, GRPH_PRESCALE_G_SIGN, mask_sh),\
+	SFB(blk, PRESCALE_GRPH_CONTROL, GRPH_PRESCALE_B_SIGN, mask_sh),\
+	SFB(blk, GRPH_SECONDARY_SURFACE_ADDRESS_HIGH, GRPH_SECONDARY_SURFACE_ADDRESS_HIGH, mask_sh),\
+	SFB(blk, GRPH_SECONDARY_SURFACE_ADDRESS, GRPH_SECONDARY_SURFACE_ADDRESS, mask_sh),\
+	SFB(blk, GRPH_PRIMARY_SURFACE_ADDRESS_HIGH, GRPH_PRIMARY_SURFACE_ADDRESS_HIGH, mask_sh),\
+	SFB(blk, GRPH_PRIMARY_SURFACE_ADDRESS, GRPH_PRIMARY_SURFACE_ADDRESS, mask_sh),\
+	SFB(blk, GRPH_UPDATE, GRPH_SURFACE_UPDATE_PENDING, mask_sh),\
+	SFB(blk, GRPH_UPDATE, GRPH_UPDATE_LOCK, mask_sh),\
+	SFB(blk, GRPH_FLIP_CONTROL, GRPH_SURFACE_UPDATE_H_RETRACE_EN, mask_sh)
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+#define MI_DCP_MASK_SH_LIST_DCE6(mask_sh, blk)\
+	SFB(blk, GRPH_ENABLE, GRPH_ENABLE, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_DEPTH, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_FORMAT, mask_sh),\
+	SFB(blk, GRPH_X_START, GRPH_X_START, mask_sh),\
+	SFB(blk, GRPH_Y_START, GRPH_Y_START, mask_sh),\
+	SFB(blk, GRPH_X_END, GRPH_X_END, mask_sh),\
+	SFB(blk, GRPH_Y_END, GRPH_Y_END, mask_sh),\
+	SFB(blk, GRPH_PITCH, GRPH_PITCH, mask_sh),\
+	SFB(blk, GRPH_SWAP_CNTL, GRPH_RED_CROSSBAR, mask_sh),\
+	SFB(blk, GRPH_SWAP_CNTL, GRPH_BLUE_CROSSBAR, mask_sh),\
+	SFB(blk, PRESCALE_GRPH_CONTROL, GRPH_PRESCALE_SELECT, mask_sh),\
+	SFB(blk, PRESCALE_GRPH_CONTROL, GRPH_PRESCALE_R_SIGN, mask_sh),\
+	SFB(blk, PRESCALE_GRPH_CONTROL, GRPH_PRESCALE_G_SIGN, mask_sh),\
+	SFB(blk, PRESCALE_GRPH_CONTROL, GRPH_PRESCALE_B_SIGN, mask_sh),\
+	SFB(blk, GRPH_SECONDARY_SURFACE_ADDRESS_HIGH, GRPH_SECONDARY_SURFACE_ADDRESS_HIGH, mask_sh),\
+	SFB(blk, GRPH_SECONDARY_SURFACE_ADDRESS, GRPH_SECONDARY_SURFACE_ADDRESS, mask_sh),\
+	SFB(blk, GRPH_PRIMARY_SURFACE_ADDRESS_HIGH, GRPH_PRIMARY_SURFACE_ADDRESS_HIGH, mask_sh),\
+	SFB(blk, GRPH_PRIMARY_SURFACE_ADDRESS, GRPH_PRIMARY_SURFACE_ADDRESS, mask_sh),\
+	SFB(blk, GRPH_UPDATE, GRPH_SURFACE_UPDATE_PENDING, mask_sh),\
+	SFB(blk, GRPH_UPDATE, GRPH_UPDATE_LOCK, mask_sh),\
+	SFB(blk, GRPH_FLIP_CONTROL, GRPH_SURFACE_UPDATE_H_RETRACE_EN, mask_sh)
+#endif
+
+#define MI_DCP_DCE11_MASK_SH_LIST(mask_sh, blk)\
+	SFB(blk, GRPH_PIPE_OUTSTANDING_REQUEST_LIMIT, GRPH_PIPE_OUTSTANDING_REQUEST_LIMIT, mask_sh)
+
+#define MI_DCP_PTE_MASK_SH_LIST(mask_sh, blk)\
+	SFB(blk, DVMM_PTE_CONTROL, DVMM_PAGE_WIDTH, mask_sh),\
+	SFB(blk, DVMM_PTE_CONTROL, DVMM_PAGE_HEIGHT, mask_sh),\
+	SFB(blk, DVMM_PTE_CONTROL, DVMM_MIN_PTE_BEFORE_FLIP, mask_sh),\
+	SFB(blk, DVMM_PTE_ARB_CONTROL, DVMM_PTE_REQ_PER_CHUNK, mask_sh),\
+	SFB(blk, DVMM_PTE_ARB_CONTROL, DVMM_MAX_PTE_REQ_OUTSTANDING, mask_sh)
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+#define MI_DMIF_PG_MASK_SH_LIST_DCE6(mask_sh, blk)\
+	SFB(blk, DPG_PIPE_ARBITRATION_CONTROL1, PIXEL_DURATION, mask_sh),\
+	SFB(blk, DPG_PIPE_URGENCY_CONTROL, URGENCY_LOW_WATERMARK, mask_sh),\
+	SFB(blk, DPG_PIPE_URGENCY_CONTROL, URGENCY_HIGH_WATERMARK, mask_sh),\
+	SFB(blk, DPG_PIPE_STUTTER_CONTROL, STUTTER_ENABLE, mask_sh),\
+	SFB(blk, DPG_PIPE_STUTTER_CONTROL, STUTTER_IGNORE_FBC, mask_sh),\
+	SF(PIPE0_DMIF_BUFFER_CONTROL, DMIF_BUFFERS_ALLOCATED, mask_sh),\
+	SF(PIPE0_DMIF_BUFFER_CONTROL, DMIF_BUFFERS_ALLOCATION_COMPLETED, mask_sh)
+
+#define MI_DMIF_PG_MASK_SH_DCE6(mask_sh, blk)\
+	SFB(blk, DPG_PIPE_ARBITRATION_CONTROL3, URGENCY_WATERMARK_MASK, mask_sh),\
+	SFB(blk, DPG_PIPE_STUTTER_CONTROL, STUTTER_EXIT_SELF_REFRESH_WATERMARK_MASK, mask_sh),\
+	SFB(blk, DPG_PIPE_STUTTER_CONTROL, STUTTER_EXIT_SELF_REFRESH_WATERMARK, mask_sh),\
+	SFB(blk, DPG_PIPE_NB_PSTATE_CHANGE_CONTROL, NB_PSTATE_CHANGE_WATERMARK_MASK, mask_sh),\
+	SFB(blk, DPG_PIPE_NB_PSTATE_CHANGE_CONTROL, NB_PSTATE_CHANGE_ENABLE, mask_sh),\
+	SFB(blk, DPG_PIPE_NB_PSTATE_CHANGE_CONTROL, NB_PSTATE_CHANGE_URGENT_DURING_REQUEST, mask_sh),\
+	SFB(blk, DPG_PIPE_NB_PSTATE_CHANGE_CONTROL, NB_PSTATE_CHANGE_NOT_SELF_REFRESH_DURING_REQUEST, mask_sh),\
+	SFB(blk, DPG_PIPE_NB_PSTATE_CHANGE_CONTROL, NB_PSTATE_CHANGE_WATERMARK, mask_sh)
+
+#define MI_DCE6_MASK_SH_LIST(mask_sh)\
+	MI_DCP_MASK_SH_LIST_DCE6(mask_sh, ),\
+	MI_DMIF_PG_MASK_SH_LIST_DCE6(mask_sh, ),\
+	MI_DMIF_PG_MASK_SH_DCE6(mask_sh, ),\
+	MI_GFX6_TILE_MASK_SH_LIST(mask_sh, )
+#endif
+
+#define MI_DMIF_PG_MASK_SH_LIST(mask_sh, blk)\
+	SFB(blk, DPG_PIPE_ARBITRATION_CONTROL1, PIXEL_DURATION, mask_sh),\
+	SFB(blk, DPG_WATERMARK_MASK_CONTROL, URGENCY_WATERMARK_MASK, mask_sh),\
+	SFB(blk, DPG_WATERMARK_MASK_CONTROL, STUTTER_EXIT_SELF_REFRESH_WATERMARK_MASK, mask_sh),\
+	SFB(blk, DPG_PIPE_URGENCY_CONTROL, URGENCY_LOW_WATERMARK, mask_sh),\
+	SFB(blk, DPG_PIPE_URGENCY_CONTROL, URGENCY_HIGH_WATERMARK, mask_sh),\
+	SFB(blk, DPG_PIPE_STUTTER_CONTROL, STUTTER_ENABLE, mask_sh),\
+	SFB(blk, DPG_PIPE_STUTTER_CONTROL, STUTTER_IGNORE_FBC, mask_sh),\
+	SF(PIPE0_DMIF_BUFFER_CONTROL, DMIF_BUFFERS_ALLOCATED, mask_sh),\
+	SF(PIPE0_DMIF_BUFFER_CONTROL, DMIF_BUFFERS_ALLOCATION_COMPLETED, mask_sh)
+
+#define MI_DMIF_PG_MASK_SH_DCE(mask_sh, blk)\
+	SFB(blk, DPG_PIPE_STUTTER_CONTROL, STUTTER_EXIT_SELF_REFRESH_WATERMARK, mask_sh),\
+	SFB(blk, DPG_WATERMARK_MASK_CONTROL, NB_PSTATE_CHANGE_WATERMARK_MASK, mask_sh),\
+	SFB(blk, DPG_PIPE_NB_PSTATE_CHANGE_CONTROL, NB_PSTATE_CHANGE_ENABLE, mask_sh),\
+	SFB(blk, DPG_PIPE_NB_PSTATE_CHANGE_CONTROL, NB_PSTATE_CHANGE_URGENT_DURING_REQUEST, mask_sh),\
+	SFB(blk, DPG_PIPE_NB_PSTATE_CHANGE_CONTROL, NB_PSTATE_CHANGE_NOT_SELF_REFRESH_DURING_REQUEST, mask_sh),\
+	SFB(blk, DPG_PIPE_NB_PSTATE_CHANGE_CONTROL, NB_PSTATE_CHANGE_WATERMARK, mask_sh)
+
+#define MI_DCE8_MASK_SH_LIST(mask_sh)\
+	MI_DCP_MASK_SH_LIST(mask_sh, ),\
+	MI_DMIF_PG_MASK_SH_LIST(mask_sh, ),\
+	MI_DMIF_PG_MASK_SH_DCE(mask_sh, ),\
+	MI_GFX8_TILE_MASK_SH_LIST(mask_sh, )
+
+#define MI_DCE11_2_MASK_SH_LIST(mask_sh)\
+	MI_DCE8_MASK_SH_LIST(mask_sh),\
+	MI_DCP_DCE11_MASK_SH_LIST(mask_sh, )
+
+#define MI_DCE11_MASK_SH_LIST(mask_sh)\
+	MI_DCE11_2_MASK_SH_LIST(mask_sh),\
+	MI_DCP_PTE_MASK_SH_LIST(mask_sh, )
+
+#define MI_GFX9_TILE_MASK_SH_LIST(mask_sh, blk)\
+	SFB(blk, GRPH_CONTROL, GRPH_NUM_BANKS, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_SW_MODE, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_SE_ENABLE, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_NUM_SHADER_ENGINES, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_NUM_PIPES, mask_sh),\
+	SFB(blk, GRPH_CONTROL, GRPH_COLOR_EXPANSION_MODE, mask_sh)
+
+#define MI_DCE12_DMIF_PG_MASK_SH_LIST(mask_sh, blk)\
+	SFB(blk, DPG_PIPE_STUTTER_CONTROL2, STUTTER_EXIT_SELF_REFRESH_WATERMARK, mask_sh),\
+	SFB(blk, DPG_PIPE_STUTTER_CONTROL2, STUTTER_ENTER_SELF_REFRESH_WATERMARK, mask_sh),\
+	SFB(blk, DPG_PIPE_URGENT_LEVEL_CONTROL, URGENT_LEVEL_LOW_WATERMARK, mask_sh),\
+	SFB(blk, DPG_PIPE_URGENT_LEVEL_CONTROL, URGENT_LEVEL_HIGH_WATERMARK, mask_sh),\
+	SFB(blk, DPG_PIPE_URGENCY_CONTROL, URGENCY_LOW_WATERMARK, mask_sh),\
+	SFB(blk, DPG_PIPE_URGENCY_CONTROL, URGENCY_HIGH_WATERMARK, mask_sh),\
+	SFB(blk, DPG_WATERMARK_MASK_CONTROL, PSTATE_CHANGE_WATERMARK_MASK, mask_sh),\
+	SFB(blk, DPG_PIPE_LOW_POWER_CONTROL, PSTATE_CHANGE_ENABLE, mask_sh),\
+	SFB(blk, DPG_PIPE_LOW_POWER_CONTROL, PSTATE_CHANGE_URGENT_DURING_REQUEST, mask_sh),\
+	SFB(blk, DPG_PIPE_LOW_POWER_CONTROL, PSTATE_CHANGE_NOT_SELF_REFRESH_DURING_REQUEST, mask_sh),\
+	SFB(blk, DPG_PIPE_LOW_POWER_CONTROL, PSTATE_CHANGE_WATERMARK, mask_sh)
+
+#define MI_GFX9_DCHUB_MASK_SH_LIST(mask_sh)\
+	SF(DCHUB_FB_LOCATION, FB_TOP, mask_sh),\
+	SF(DCHUB_FB_LOCATION, FB_BASE, mask_sh),\
+	SF(DCHUB_AGP_BASE, AGP_BASE, mask_sh),\
+	SF(DCHUB_AGP_BOT, AGP_BOT, mask_sh),\
+	SF(DCHUB_AGP_TOP, AGP_TOP, mask_sh)
+
+#define MI_DCE12_MASK_SH_LIST(mask_sh)\
+	MI_DCP_MASK_SH_LIST(mask_sh, DCP0_),\
+	SF(DCP0_GRPH_SECONDARY_SURFACE_ADDRESS, GRPH_SECONDARY_DFQ_ENABLE, mask_sh),\
+	MI_DCP_DCE11_MASK_SH_LIST(mask_sh, DCP0_),\
+	MI_DCP_PTE_MASK_SH_LIST(mask_sh, DCP0_),\
+	MI_DMIF_PG_MASK_SH_LIST(mask_sh, DMIF_PG0_),\
+	MI_DCE12_DMIF_PG_MASK_SH_LIST(mask_sh, DMIF_PG0_),\
+	MI_GFX9_TILE_MASK_SH_LIST(mask_sh, DCP0_),\
+	MI_GFX9_DCHUB_MASK_SH_LIST(mask_sh)
+
+#define MI_REG_FIELD_LIST(type) \
+	type GRPH_ENABLE; \
+	type GRPH_X_START; \
+	type GRPH_Y_START; \
+	type GRPH_X_END; \
+	type GRPH_Y_END; \
+	type GRPH_PITCH; \
+	type GRPH_ROTATION_ANGLE; \
+	type GRPH_RED_CROSSBAR; \
+	type GRPH_BLUE_CROSSBAR; \
+	type GRPH_PRESCALE_SELECT; \
+	type GRPH_PRESCALE_R_SIGN; \
+	type GRPH_PRESCALE_G_SIGN; \
+	type GRPH_PRESCALE_B_SIGN; \
+	type GRPH_PIPE_OUTSTANDING_REQUEST_LIMIT; \
+	type DVMM_PAGE_WIDTH; \
+	type DVMM_PAGE_HEIGHT; \
+	type DVMM_MIN_PTE_BEFORE_FLIP; \
+	type DVMM_PTE_REQ_PER_CHUNK; \
+	type DVMM_MAX_PTE_REQ_OUTSTANDING; \
+	type GRPH_DEPTH; \
+	type GRPH_FORMAT; \
+	type GRPH_NUM_BANKS; \
+	type GRPH_BANK_WIDTH;\
+	type GRPH_BANK_HEIGHT;\
+	type GRPH_MACRO_TILE_ASPECT;\
+	type GRPH_TILE_SPLIT;\
+	type GRPH_MICRO_TILE_MODE;\
+	type GRPH_PIPE_CONFIG;\
+	type GRPH_ARRAY_MODE;\
+	type GRPH_COLOR_EXPANSION_MODE;\
+	type GRPH_SW_MODE; \
+	type GRPH_SE_ENABLE; \
+	type GRPH_NUM_SHADER_ENGINES; \
+	type GRPH_NUM_PIPES; \
+	type GRPH_SECONDARY_SURFACE_ADDRESS_HIGH; \
+	type GRPH_SECONDARY_SURFACE_ADDRESS; \
+	type GRPH_SECONDARY_DFQ_ENABLE; \
+	type GRPH_PRIMARY_SURFACE_ADDRESS_HIGH; \
+	type GRPH_PRIMARY_SURFACE_ADDRESS; \
+	type GRPH_SURFACE_UPDATE_PENDING; \
+	type GRPH_SURFACE_UPDATE_H_RETRACE_EN; \
+	type GRPH_UPDATE_LOCK; \
+	type PIXEL_DURATION; \
+	type URGENCY_WATERMARK_MASK; \
+	type PSTATE_CHANGE_WATERMARK_MASK; \
+	type NB_PSTATE_CHANGE_WATERMARK_MASK; \
+	type STUTTER_EXIT_SELF_REFRESH_WATERMARK_MASK; \
+	type URGENCY_LOW_WATERMARK; \
+	type URGENCY_HIGH_WATERMARK; \
+	type URGENT_LEVEL_LOW_WATERMARK;\
+	type URGENT_LEVEL_HIGH_WATERMARK;\
+	type NB_PSTATE_CHANGE_ENABLE; \
+	type NB_PSTATE_CHANGE_URGENT_DURING_REQUEST; \
+	type NB_PSTATE_CHANGE_NOT_SELF_REFRESH_DURING_REQUEST; \
+	type NB_PSTATE_CHANGE_WATERMARK; \
+	type PSTATE_CHANGE_ENABLE; \
+	type PSTATE_CHANGE_URGENT_DURING_REQUEST; \
+	type PSTATE_CHANGE_NOT_SELF_REFRESH_DURING_REQUEST; \
+	type PSTATE_CHANGE_WATERMARK; \
+	type STUTTER_ENABLE; \
+	type STUTTER_IGNORE_FBC; \
+	type STUTTER_EXIT_SELF_REFRESH_WATERMARK; \
+	type STUTTER_ENTER_SELF_REFRESH_WATERMARK; \
+	type DMIF_BUFFERS_ALLOCATED; \
+	type DMIF_BUFFERS_ALLOCATION_COMPLETED; \
+	type ENABLE; /* MC_HUB_RDREQ_DMIF_LIMIT */\
+	type FB_BASE; \
+	type FB_TOP; \
+	type AGP_BASE; \
+	type AGP_TOP; \
+	type AGP_BOT; \
+
+struct dce_mem_input_shift {
+	MI_REG_FIELD_LIST(uint8_t)
+};
+
+struct dce_mem_input_mask {
+	MI_REG_FIELD_LIST(uint32_t)
+};
+
+struct dce_mem_input_wa {
+	uint8_t single_head_rdreq_dmif_limit;
+};
+
+struct dce_mem_input {
+	struct mem_input base;
+
+	const struct dce_mem_input_registers *regs;
+	const struct dce_mem_input_shift *shifts;
+	const struct dce_mem_input_mask *masks;
+
+	struct dce_mem_input_wa wa;
+};
+
+void dce_mem_input_construct(
+	struct dce_mem_input *dce_mi,
+	struct dc_context *ctx,
+	int inst,
+	const struct dce_mem_input_registers *regs,
+	const struct dce_mem_input_shift *mi_shift,
+	const struct dce_mem_input_mask *mi_mask);
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+void dce60_mem_input_construct(
+	struct dce_mem_input *dce_mi,
+	struct dc_context *ctx,
+	int inst,
+	const struct dce_mem_input_registers *regs,
+	const struct dce_mem_input_shift *mi_shift,
+	const struct dce_mem_input_mask *mi_mask);
+#endif
+
+void dce112_mem_input_construct(
+	struct dce_mem_input *dce_mi,
+	struct dc_context *ctx,
+	int inst,
+	const struct dce_mem_input_registers *regs,
+	const struct dce_mem_input_shift *mi_shift,
+	const struct dce_mem_input_mask *mi_mask);
+
+void dce120_mem_input_construct(
+	struct dce_mem_input *dce_mi,
+	struct dc_context *ctx,
+	int inst,
+	const struct dce_mem_input_registers *regs,
+	const struct dce_mem_input_shift *mi_shift,
+	const struct dce_mem_input_mask *mi_mask);
+
+#endif /*__DCE_MEM_INPUT_H__*/
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_opp.c b/drivers/gpu/drm/amd/display/dc/dce/dce_opp.c
new file mode 100644
index 000000000000..f342da5a5e50
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_opp.c
@@ -0,0 +1,768 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+#include "basics/conversion.h"
+
+#include "dce_opp.h"
+
+#include "reg_helper.h"
+
+#define REG(reg)\
+	(opp110->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	opp110->opp_shift->field_name, opp110->opp_mask->field_name
+
+#define CTX \
+	opp110->base.ctx
+
+enum {
+	MAX_PWL_ENTRY = 128,
+	MAX_REGIONS_NUMBER = 16
+};
+
+enum {
+	MAX_LUT_ENTRY = 256,
+	MAX_NUMBER_OF_ENTRIES = 256
+};
+
+
+enum {
+	OUTPUT_CSC_MATRIX_SIZE = 12
+};
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+/*
+ *****************************************************************************
+ *  Function: regamma_config_regions_and_segments
+ *
+ *     build regamma curve by using predefined hw points
+ *     uses interface parameters ,like EDID coeff.
+ *
+ * @param   : parameters   interface parameters
+ *  @return void
+ *
+ *  @note
+ *
+ *  @see
+ *
+ *****************************************************************************
+ */
+
+
+
+/*
+ *	set_truncation
+ *	1) set truncation depth: 0 for 18 bpp or 1 for 24 bpp
+ *	2) enable truncation
+ *	3) HW remove 12bit FMT support for DCE11 power saving reason.
+ */
+static void set_truncation(
+		struct dce110_opp *opp110,
+		const struct bit_depth_reduction_params *params)
+{
+	/*Disable truncation*/
+	REG_UPDATE_3(FMT_BIT_DEPTH_CONTROL,
+			FMT_TRUNCATE_EN, 0,
+			FMT_TRUNCATE_DEPTH, 0,
+			FMT_TRUNCATE_MODE, 0);
+
+
+	if (params->pixel_encoding == PIXEL_ENCODING_YCBCR422) {
+		/*  8bpc trunc on YCbCr422*/
+		if (params->flags.TRUNCATE_DEPTH == 1)
+			REG_UPDATE_3(FMT_BIT_DEPTH_CONTROL,
+					FMT_TRUNCATE_EN, 1,
+					FMT_TRUNCATE_DEPTH, 1,
+					FMT_TRUNCATE_MODE, 0);
+		else if (params->flags.TRUNCATE_DEPTH == 2)
+			/*  10bpc trunc on YCbCr422*/
+			REG_UPDATE_3(FMT_BIT_DEPTH_CONTROL,
+					FMT_TRUNCATE_EN, 1,
+					FMT_TRUNCATE_DEPTH, 2,
+					FMT_TRUNCATE_MODE, 0);
+		return;
+	}
+	/* on other format-to do */
+	if (params->flags.TRUNCATE_ENABLED == 0)
+		return;
+	/*Set truncation depth and Enable truncation*/
+	REG_UPDATE_3(FMT_BIT_DEPTH_CONTROL,
+				FMT_TRUNCATE_EN, 1,
+				FMT_TRUNCATE_DEPTH,
+				params->flags.TRUNCATE_DEPTH,
+				FMT_TRUNCATE_MODE,
+				params->flags.TRUNCATE_MODE);
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+/*
+ *	dce60_set_truncation
+ *	1) set truncation depth: 0 for 18 bpp or 1 for 24 bpp
+ *	2) enable truncation
+ *	3) HW remove 12bit FMT support for DCE11 power saving reason.
+ */
+static void dce60_set_truncation(
+		struct dce110_opp *opp110,
+		const struct bit_depth_reduction_params *params)
+{
+	/* DCE6 has no FMT_TRUNCATE_MODE bit in FMT_BIT_DEPTH_CONTROL reg */
+
+	/*Disable truncation*/
+	REG_UPDATE_2(FMT_BIT_DEPTH_CONTROL,
+			FMT_TRUNCATE_EN, 0,
+			FMT_TRUNCATE_DEPTH, 0);
+
+	if (params->pixel_encoding == PIXEL_ENCODING_YCBCR422) {
+		/*  8bpc trunc on YCbCr422*/
+		if (params->flags.TRUNCATE_DEPTH == 1)
+			REG_UPDATE_2(FMT_BIT_DEPTH_CONTROL,
+					FMT_TRUNCATE_EN, 1,
+					FMT_TRUNCATE_DEPTH, 1);
+		else if (params->flags.TRUNCATE_DEPTH == 2)
+			/*  10bpc trunc on YCbCr422*/
+			REG_UPDATE_2(FMT_BIT_DEPTH_CONTROL,
+					FMT_TRUNCATE_EN, 1,
+					FMT_TRUNCATE_DEPTH, 2);
+		return;
+	}
+	/* on other format-to do */
+	if (params->flags.TRUNCATE_ENABLED == 0)
+		return;
+	/*Set truncation depth and Enable truncation*/
+	REG_UPDATE_2(FMT_BIT_DEPTH_CONTROL,
+				FMT_TRUNCATE_EN, 1,
+				FMT_TRUNCATE_DEPTH,
+				params->flags.TRUNCATE_DEPTH);
+}
+#endif
+
+/*
+ *	set_spatial_dither
+ *	1) set spatial dithering mode: pattern of seed
+ *	2) set spatial dithering depth: 0 for 18bpp or 1 for 24bpp
+ *	3) set random seed
+ *	4) set random mode
+ *		lfsr is reset every frame or not reset
+ *		RGB dithering method
+ *		0: RGB data are all dithered with x^28+x^3+1
+ *		1: R data is dithered with x^28+x^3+1
+ *		G data is dithered with x^28+X^9+1
+ *		B data is dithered with x^28+x^13+1
+ *		enable high pass filter or not
+ *	5) enable spatical dithering
+ */
+static void set_spatial_dither(
+	struct dce110_opp *opp110,
+	const struct bit_depth_reduction_params *params)
+{
+	/*Disable spatial (random) dithering*/
+	REG_UPDATE_3(FMT_BIT_DEPTH_CONTROL,
+		FMT_SPATIAL_DITHER_EN, 0,
+		FMT_SPATIAL_DITHER_DEPTH, 0,
+		FMT_SPATIAL_DITHER_MODE, 0);
+
+	REG_UPDATE_3(FMT_BIT_DEPTH_CONTROL,
+		FMT_HIGHPASS_RANDOM_ENABLE, 0,
+		FMT_FRAME_RANDOM_ENABLE, 0,
+		FMT_RGB_RANDOM_ENABLE, 0);
+
+	REG_UPDATE(FMT_BIT_DEPTH_CONTROL,
+		FMT_TEMPORAL_DITHER_EN, 0);
+
+	if (params->flags.SPATIAL_DITHER_ENABLED == 0)
+		return;
+
+	/* only use FRAME_COUNTER_MAX if frameRandom == 1*/
+
+	if (opp110->opp_mask->FMT_SPATIAL_DITHER_FRAME_COUNTER_MAX &&
+			opp110->opp_mask->FMT_SPATIAL_DITHER_FRAME_COUNTER_BIT_SWAP) {
+		if (params->flags.FRAME_RANDOM == 1) {
+			if (params->flags.SPATIAL_DITHER_DEPTH == 0 ||
+			params->flags.SPATIAL_DITHER_DEPTH == 1) {
+				REG_UPDATE_2(FMT_CONTROL,
+					FMT_SPATIAL_DITHER_FRAME_COUNTER_MAX, 15,
+					FMT_SPATIAL_DITHER_FRAME_COUNTER_BIT_SWAP, 2);
+			} else if (params->flags.SPATIAL_DITHER_DEPTH == 2) {
+				REG_UPDATE_2(FMT_CONTROL,
+					FMT_SPATIAL_DITHER_FRAME_COUNTER_MAX, 3,
+					FMT_SPATIAL_DITHER_FRAME_COUNTER_BIT_SWAP, 1);
+			} else
+				return;
+		} else {
+			REG_UPDATE_2(FMT_CONTROL,
+					FMT_SPATIAL_DITHER_FRAME_COUNTER_MAX, 0,
+					FMT_SPATIAL_DITHER_FRAME_COUNTER_BIT_SWAP, 0);
+		}
+	}
+	/* Set seed for random values for
+	 * spatial dithering for R,G,B channels
+	 */
+	REG_UPDATE(FMT_DITHER_RAND_R_SEED,
+			FMT_RAND_R_SEED, params->r_seed_value);
+
+	REG_UPDATE(FMT_DITHER_RAND_G_SEED,
+			FMT_RAND_G_SEED, params->g_seed_value);
+
+	REG_UPDATE(FMT_DITHER_RAND_B_SEED,
+			FMT_RAND_B_SEED, params->b_seed_value);
+
+	/* FMT_OFFSET_R_Cr  31:16 0x0 Setting the zero
+	 * offset for the R/Cr channel, lower 4LSB
+	 * is forced to zeros. Typically set to 0
+	 * RGB and 0x80000 YCbCr.
+	 */
+	/* FMT_OFFSET_G_Y   31:16 0x0 Setting the zero
+	 * offset for the G/Y  channel, lower 4LSB is
+	 * forced to zeros. Typically set to 0 RGB
+	 * and 0x80000 YCbCr.
+	 */
+	/* FMT_OFFSET_B_Cb  31:16 0x0 Setting the zero
+	 * offset for the B/Cb channel, lower 4LSB is
+	 * forced to zeros. Typically set to 0 RGB and
+	 * 0x80000 YCbCr.
+	 */
+
+	/* Disable High pass filter
+	 * Reset only at startup
+	 * Set RGB data dithered with x^28+x^3+1
+	 */
+	REG_UPDATE_3(FMT_BIT_DEPTH_CONTROL,
+		FMT_HIGHPASS_RANDOM_ENABLE, params->flags.HIGHPASS_RANDOM,
+		FMT_FRAME_RANDOM_ENABLE, params->flags.FRAME_RANDOM,
+		FMT_RGB_RANDOM_ENABLE, params->flags.RGB_RANDOM);
+
+	/* Set spatial dithering bit depth
+	 * Set spatial dithering mode
+	 * (default is Seed patterrn AAAA...)
+	 * Enable spatial dithering
+	 */
+	REG_UPDATE_3(FMT_BIT_DEPTH_CONTROL,
+		FMT_SPATIAL_DITHER_DEPTH, params->flags.SPATIAL_DITHER_DEPTH,
+		FMT_SPATIAL_DITHER_MODE, params->flags.SPATIAL_DITHER_MODE,
+		FMT_SPATIAL_DITHER_EN, 1);
+}
+
+/*
+ *	SetTemporalDither (Frame Modulation)
+ *	1) set temporal dither depth
+ *	2) select pattern: from hard-coded pattern or programmable pattern
+ *	3) select optimized strips for BGR or RGB LCD sub-pixel
+ *	4) set s matrix
+ *	5) set t matrix
+ *	6) set grey level for 0.25, 0.5, 0.75
+ *	7) enable temporal dithering
+ */
+
+static void set_temporal_dither(
+	struct dce110_opp *opp110,
+	const struct bit_depth_reduction_params *params)
+{
+	/*Disable temporal (frame modulation) dithering first*/
+	REG_UPDATE_3(FMT_BIT_DEPTH_CONTROL,
+		FMT_TEMPORAL_DITHER_EN, 0,
+		FMT_TEMPORAL_DITHER_RESET, 0,
+		FMT_TEMPORAL_DITHER_OFFSET, 0);
+
+	REG_UPDATE_2(FMT_BIT_DEPTH_CONTROL,
+		FMT_TEMPORAL_DITHER_DEPTH, 0,
+		FMT_TEMPORAL_LEVEL, 0);
+
+	REG_UPDATE_3(FMT_BIT_DEPTH_CONTROL,
+		FMT_25FRC_SEL, 0,
+		FMT_50FRC_SEL, 0,
+		FMT_75FRC_SEL, 0);
+
+	/* no 10bpc dither on DCE11*/
+	if (params->flags.FRAME_MODULATION_ENABLED == 0 ||
+		params->flags.FRAME_MODULATION_DEPTH == 2)
+		return;
+
+	/* Set temporal dithering depth*/
+	REG_UPDATE_3(FMT_BIT_DEPTH_CONTROL,
+		FMT_TEMPORAL_DITHER_DEPTH, params->flags.FRAME_MODULATION_DEPTH,
+		FMT_TEMPORAL_DITHER_RESET, 0,
+		FMT_TEMPORAL_DITHER_OFFSET, 0);
+
+	/*Select legacy pattern based on FRC and Temporal level*/
+	if (REG(FMT_TEMPORAL_DITHER_PATTERN_CONTROL)) {
+		REG_WRITE(FMT_TEMPORAL_DITHER_PATTERN_CONTROL, 0);
+		/*Set s matrix*/
+		REG_WRITE(FMT_TEMPORAL_DITHER_PROGRAMMABLE_PATTERN_S_MATRIX, 0);
+		/*Set t matrix*/
+		REG_WRITE(FMT_TEMPORAL_DITHER_PROGRAMMABLE_PATTERN_T_MATRIX, 0);
+	}
+
+	/*Select patterns for 0.25, 0.5 and 0.75 grey level*/
+	REG_UPDATE(FMT_BIT_DEPTH_CONTROL,
+		FMT_TEMPORAL_LEVEL, params->flags.TEMPORAL_LEVEL);
+
+	REG_UPDATE_3(FMT_BIT_DEPTH_CONTROL,
+		FMT_25FRC_SEL, params->flags.FRC25,
+		FMT_50FRC_SEL, params->flags.FRC50,
+		FMT_75FRC_SEL, params->flags.FRC75);
+
+	/*Enable bit reduction by temporal (frame modulation) dithering*/
+	REG_UPDATE(FMT_BIT_DEPTH_CONTROL,
+		FMT_TEMPORAL_DITHER_EN, 1);
+}
+
+/*
+ *	Set Clamping
+ *	1) Set clamping format based on bpc - 0 for 6bpc (No clamping)
+ *		1 for 8 bpc
+ *		2 for 10 bpc
+ *		3 for 12 bpc
+ *		7 for programable
+ *	2) Enable clamp if Limited range requested
+ */
+void dce110_opp_set_clamping(
+	struct dce110_opp *opp110,
+	const struct clamping_and_pixel_encoding_params *params)
+{
+	REG_SET_2(FMT_CLAMP_CNTL, 0,
+		FMT_CLAMP_DATA_EN, 0,
+		FMT_CLAMP_COLOR_FORMAT, 0);
+
+	switch (params->clamping_level) {
+	case CLAMPING_FULL_RANGE:
+		break;
+	case CLAMPING_LIMITED_RANGE_8BPC:
+		REG_SET_2(FMT_CLAMP_CNTL, 0,
+			FMT_CLAMP_DATA_EN, 1,
+			FMT_CLAMP_COLOR_FORMAT, 1);
+		break;
+	case CLAMPING_LIMITED_RANGE_10BPC:
+		REG_SET_2(FMT_CLAMP_CNTL, 0,
+			FMT_CLAMP_DATA_EN, 1,
+			FMT_CLAMP_COLOR_FORMAT, 2);
+		break;
+	case CLAMPING_LIMITED_RANGE_12BPC:
+		REG_SET_2(FMT_CLAMP_CNTL, 0,
+			FMT_CLAMP_DATA_EN, 1,
+			FMT_CLAMP_COLOR_FORMAT, 3);
+		break;
+	case CLAMPING_LIMITED_RANGE_PROGRAMMABLE:
+		/*Set clamp control*/
+		REG_SET_2(FMT_CLAMP_CNTL, 0,
+			FMT_CLAMP_DATA_EN, 1,
+			FMT_CLAMP_COLOR_FORMAT, 7);
+
+		/*set the defaults*/
+		REG_SET_2(FMT_CLAMP_COMPONENT_R, 0,
+			FMT_CLAMP_LOWER_R, 0x10,
+			FMT_CLAMP_UPPER_R, 0xFEF);
+
+		REG_SET_2(FMT_CLAMP_COMPONENT_G, 0,
+			FMT_CLAMP_LOWER_G, 0x10,
+			FMT_CLAMP_UPPER_G, 0xFEF);
+
+		REG_SET_2(FMT_CLAMP_COMPONENT_B, 0,
+			FMT_CLAMP_LOWER_B, 0x10,
+			FMT_CLAMP_UPPER_B, 0xFEF);
+		break;
+	default:
+		break;
+	}
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+/*
+ *	Set Clamping for DCE6 parts
+ *	1) Set clamping format based on bpc - 0 for 6bpc (No clamping)
+ *		1 for 8 bpc
+ *		2 for 10 bpc
+ *		3 for 12 bpc
+ *		7 for programable
+ *	2) Enable clamp if Limited range requested
+ */
+static void dce60_opp_set_clamping(
+	struct dce110_opp *opp110,
+	const struct clamping_and_pixel_encoding_params *params)
+{
+	REG_SET_2(FMT_CLAMP_CNTL, 0,
+		FMT_CLAMP_DATA_EN, 0,
+		FMT_CLAMP_COLOR_FORMAT, 0);
+
+	switch (params->clamping_level) {
+	case CLAMPING_FULL_RANGE:
+		break;
+	case CLAMPING_LIMITED_RANGE_8BPC:
+		REG_SET_2(FMT_CLAMP_CNTL, 0,
+			FMT_CLAMP_DATA_EN, 1,
+			FMT_CLAMP_COLOR_FORMAT, 1);
+		break;
+	case CLAMPING_LIMITED_RANGE_10BPC:
+		REG_SET_2(FMT_CLAMP_CNTL, 0,
+			FMT_CLAMP_DATA_EN, 1,
+			FMT_CLAMP_COLOR_FORMAT, 2);
+		break;
+	case CLAMPING_LIMITED_RANGE_12BPC:
+		REG_SET_2(FMT_CLAMP_CNTL, 0,
+			FMT_CLAMP_DATA_EN, 1,
+			FMT_CLAMP_COLOR_FORMAT, 3);
+		break;
+	case CLAMPING_LIMITED_RANGE_PROGRAMMABLE:
+		/*Set clamp control*/
+		REG_SET_2(FMT_CLAMP_CNTL, 0,
+			FMT_CLAMP_DATA_EN, 1,
+			FMT_CLAMP_COLOR_FORMAT, 7);
+
+		/* DCE6 does have FMT_CLAMP_COMPONENT_{R,G,B} registers */
+
+		break;
+	default:
+		break;
+	}
+}
+#endif
+
+/*
+ *	set_pixel_encoding
+ *
+ *	Set Pixel Encoding
+ *		0: RGB 4:4:4 or YCbCr 4:4:4 or YOnly
+ *		1: YCbCr 4:2:2
+ */
+static void set_pixel_encoding(
+	struct dce110_opp *opp110,
+	const struct clamping_and_pixel_encoding_params *params)
+{
+	if (opp110->opp_mask->FMT_CBCR_BIT_REDUCTION_BYPASS)
+		REG_UPDATE_3(FMT_CONTROL,
+				FMT_PIXEL_ENCODING, 0,
+				FMT_SUBSAMPLING_MODE, 0,
+				FMT_CBCR_BIT_REDUCTION_BYPASS, 0);
+	else
+		REG_UPDATE_2(FMT_CONTROL,
+				FMT_PIXEL_ENCODING, 0,
+				FMT_SUBSAMPLING_MODE, 0);
+
+	if (params->pixel_encoding == PIXEL_ENCODING_YCBCR422) {
+		REG_UPDATE_2(FMT_CONTROL,
+				FMT_PIXEL_ENCODING, 1,
+				FMT_SUBSAMPLING_ORDER, 0);
+	}
+	if (params->pixel_encoding == PIXEL_ENCODING_YCBCR420) {
+		REG_UPDATE_3(FMT_CONTROL,
+				FMT_PIXEL_ENCODING, 2,
+				FMT_SUBSAMPLING_MODE, 2,
+				FMT_CBCR_BIT_REDUCTION_BYPASS, 1);
+	}
+
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+/*
+ *	dce60_set_pixel_encoding
+ *	DCE6 has no FMT_SUBSAMPLING_{MODE,ORDER} bits in FMT_CONTROL reg
+ *	Set Pixel Encoding
+ *		0: RGB 4:4:4 or YCbCr 4:4:4 or YOnly
+ *		1: YCbCr 4:2:2
+ */
+static void dce60_set_pixel_encoding(
+	struct dce110_opp *opp110,
+	const struct clamping_and_pixel_encoding_params *params)
+{
+	if (opp110->opp_mask->FMT_CBCR_BIT_REDUCTION_BYPASS)
+		REG_UPDATE_2(FMT_CONTROL,
+				FMT_PIXEL_ENCODING, 0,
+				FMT_CBCR_BIT_REDUCTION_BYPASS, 0);
+	else
+		REG_UPDATE(FMT_CONTROL,
+				FMT_PIXEL_ENCODING, 0);
+
+	if (params->pixel_encoding == PIXEL_ENCODING_YCBCR422) {
+		REG_UPDATE(FMT_CONTROL,
+				FMT_PIXEL_ENCODING, 1);
+	}
+	if (params->pixel_encoding == PIXEL_ENCODING_YCBCR420) {
+		REG_UPDATE_2(FMT_CONTROL,
+				FMT_PIXEL_ENCODING, 2,
+				FMT_CBCR_BIT_REDUCTION_BYPASS, 1);
+	}
+
+}
+#endif
+
+void dce110_opp_program_bit_depth_reduction(
+	struct output_pixel_processor *opp,
+	const struct bit_depth_reduction_params *params)
+{
+	struct dce110_opp *opp110 = TO_DCE110_OPP(opp);
+
+	set_truncation(opp110, params);
+	set_spatial_dither(opp110, params);
+	set_temporal_dither(opp110, params);
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static void dce60_opp_program_bit_depth_reduction(
+	struct output_pixel_processor *opp,
+	const struct bit_depth_reduction_params *params)
+{
+	struct dce110_opp *opp110 = TO_DCE110_OPP(opp);
+
+	dce60_set_truncation(opp110, params);
+	set_spatial_dither(opp110, params);
+	set_temporal_dither(opp110, params);
+}
+#endif
+
+void dce110_opp_program_clamping_and_pixel_encoding(
+	struct output_pixel_processor *opp,
+	const struct clamping_and_pixel_encoding_params *params)
+{
+	struct dce110_opp *opp110 = TO_DCE110_OPP(opp);
+
+	dce110_opp_set_clamping(opp110, params);
+	set_pixel_encoding(opp110, params);
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static void dce60_opp_program_clamping_and_pixel_encoding(
+	struct output_pixel_processor *opp,
+	const struct clamping_and_pixel_encoding_params *params)
+{
+	struct dce110_opp *opp110 = TO_DCE110_OPP(opp);
+
+	dce60_opp_set_clamping(opp110, params);
+	dce60_set_pixel_encoding(opp110, params);
+}
+#endif
+
+static void program_formatter_420_memory(struct output_pixel_processor *opp)
+{
+	struct dce110_opp *opp110 = TO_DCE110_OPP(opp);
+	uint32_t fmt_mem_cntl_value;
+
+	/* Program source select*/
+	/* Use HW default source select for FMT_MEMORYx_CONTROL */
+	/* Use that value for FMT_SRC_SELECT as well*/
+	REG_GET(CONTROL,
+			FMT420_MEM0_SOURCE_SEL, &fmt_mem_cntl_value);
+
+	REG_UPDATE(FMT_CONTROL,
+			FMT_SRC_SELECT, fmt_mem_cntl_value);
+
+	/* Turn on the memory */
+	REG_UPDATE(CONTROL,
+			FMT420_MEM0_PWR_FORCE, 0);
+}
+
+void dce110_opp_set_dyn_expansion(
+	struct output_pixel_processor *opp,
+	enum dc_color_space color_sp,
+	enum dc_color_depth color_dpth,
+	enum signal_type signal)
+{
+	struct dce110_opp *opp110 = TO_DCE110_OPP(opp);
+
+	REG_UPDATE_2(FMT_DYNAMIC_EXP_CNTL,
+			FMT_DYNAMIC_EXP_EN, 0,
+			FMT_DYNAMIC_EXP_MODE, 0);
+
+	/*00 - 10-bit -> 12-bit dynamic expansion*/
+	/*01 - 8-bit  -> 12-bit dynamic expansion*/
+	if (signal == SIGNAL_TYPE_HDMI_TYPE_A ||
+		signal == SIGNAL_TYPE_DISPLAY_PORT ||
+		signal == SIGNAL_TYPE_DISPLAY_PORT_MST) {
+		switch (color_dpth) {
+		case COLOR_DEPTH_888:
+			REG_UPDATE_2(FMT_DYNAMIC_EXP_CNTL,
+				FMT_DYNAMIC_EXP_EN, 1,
+				FMT_DYNAMIC_EXP_MODE, 1);
+			break;
+		case COLOR_DEPTH_101010:
+			REG_UPDATE_2(FMT_DYNAMIC_EXP_CNTL,
+				FMT_DYNAMIC_EXP_EN, 1,
+				FMT_DYNAMIC_EXP_MODE, 0);
+			break;
+		case COLOR_DEPTH_121212:
+			REG_UPDATE_2(
+				FMT_DYNAMIC_EXP_CNTL,
+				FMT_DYNAMIC_EXP_EN, 1,/*otherwise last two bits are zero*/
+				FMT_DYNAMIC_EXP_MODE, 0);
+			break;
+		default:
+			break;
+		}
+	}
+}
+
+static void program_formatter_reset_dig_resync_fifo(struct output_pixel_processor *opp)
+{
+	struct dce110_opp *opp110 = TO_DCE110_OPP(opp);
+
+	/* clear previous phase lock status*/
+	REG_UPDATE(FMT_CONTROL,
+			FMT_420_PIXEL_PHASE_LOCKED_CLEAR, 1);
+
+	/* poll until FMT_420_PIXEL_PHASE_LOCKED become 1*/
+	REG_WAIT(FMT_CONTROL, FMT_420_PIXEL_PHASE_LOCKED, 1, 10, 10);
+
+}
+
+void dce110_opp_program_fmt(
+	struct output_pixel_processor *opp,
+	struct bit_depth_reduction_params *fmt_bit_depth,
+	struct clamping_and_pixel_encoding_params *clamping)
+{
+	/* dithering is affected by <CrtcSourceSelect>, hence should be
+	 * programmed afterwards */
+
+	if (clamping->pixel_encoding == PIXEL_ENCODING_YCBCR420)
+		program_formatter_420_memory(opp);
+
+	dce110_opp_program_bit_depth_reduction(
+		opp,
+		fmt_bit_depth);
+
+	dce110_opp_program_clamping_and_pixel_encoding(
+		opp,
+		clamping);
+
+	if (clamping->pixel_encoding == PIXEL_ENCODING_YCBCR420)
+		program_formatter_reset_dig_resync_fifo(opp);
+
+	return;
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static void dce60_opp_program_fmt(
+	struct output_pixel_processor *opp,
+	struct bit_depth_reduction_params *fmt_bit_depth,
+	struct clamping_and_pixel_encoding_params *clamping)
+{
+	/* dithering is affected by <CrtcSourceSelect>, hence should be
+	 * programmed afterwards */
+
+	if (clamping->pixel_encoding == PIXEL_ENCODING_YCBCR420)
+		program_formatter_420_memory(opp);
+
+	dce60_opp_program_bit_depth_reduction(
+		opp,
+		fmt_bit_depth);
+
+	dce60_opp_program_clamping_and_pixel_encoding(
+		opp,
+		clamping);
+
+	if (clamping->pixel_encoding == PIXEL_ENCODING_YCBCR420)
+		program_formatter_reset_dig_resync_fifo(opp);
+
+	return;
+}
+#endif
+
+
+
+/*****************************************/
+/* Constructor, Destructor               */
+/*****************************************/
+
+static const struct opp_funcs funcs = {
+	.opp_set_dyn_expansion = dce110_opp_set_dyn_expansion,
+	.opp_destroy = dce110_opp_destroy,
+	.opp_program_fmt = dce110_opp_program_fmt,
+	.opp_program_bit_depth_reduction = dce110_opp_program_bit_depth_reduction
+};
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static const struct opp_funcs dce60_opp_funcs = {
+	.opp_set_dyn_expansion = dce110_opp_set_dyn_expansion,
+	.opp_destroy = dce110_opp_destroy,
+	.opp_program_fmt = dce60_opp_program_fmt,
+	.opp_program_bit_depth_reduction = dce60_opp_program_bit_depth_reduction
+};
+#endif
+
+void dce110_opp_construct(struct dce110_opp *opp110,
+	struct dc_context *ctx,
+	uint32_t inst,
+	const struct dce_opp_registers *regs,
+	const struct dce_opp_shift *opp_shift,
+	const struct dce_opp_mask *opp_mask)
+{
+	opp110->base.funcs = &funcs;
+
+	opp110->base.ctx = ctx;
+
+	opp110->base.inst = inst;
+
+	opp110->regs = regs;
+	opp110->opp_shift = opp_shift;
+	opp110->opp_mask = opp_mask;
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+void dce60_opp_construct(struct dce110_opp *opp110,
+	struct dc_context *ctx,
+	uint32_t inst,
+	const struct dce_opp_registers *regs,
+	const struct dce_opp_shift *opp_shift,
+	const struct dce_opp_mask *opp_mask)
+{
+	opp110->base.funcs = &dce60_opp_funcs;
+
+	opp110->base.ctx = ctx;
+
+	opp110->base.inst = inst;
+
+	opp110->regs = regs;
+	opp110->opp_shift = opp_shift;
+	opp110->opp_mask = opp_mask;
+}
+#endif
+
+void dce110_opp_destroy(struct output_pixel_processor **opp)
+{
+	if (*opp)
+		kfree(FROM_DCE11_OPP(*opp));
+	*opp = NULL;
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_opp.h b/drivers/gpu/drm/amd/display/dc/dce/dce_opp.h
new file mode 100644
index 000000000000..3d9be87aae45
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_opp.h
@@ -0,0 +1,358 @@
+/* Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_OPP_DCE_H__
+#define __DC_OPP_DCE_H__
+
+#include "dc_types.h"
+#include "opp.h"
+#include "core_types.h"
+
+#define FROM_DCE11_OPP(opp)\
+	container_of(opp, struct dce110_opp, base)
+
+enum dce110_opp_reg_type {
+	DCE110_OPP_REG_DCP = 0,
+	DCE110_OPP_REG_DCFE,
+	DCE110_OPP_REG_FMT,
+
+	DCE110_OPP_REG_MAX
+};
+
+#define OPP_COMMON_REG_LIST_BASE(id) \
+	SRI(FMT_DYNAMIC_EXP_CNTL, FMT, id), \
+	SRI(FMT_BIT_DEPTH_CONTROL, FMT, id), \
+	SRI(FMT_CONTROL, FMT, id), \
+	SRI(FMT_DITHER_RAND_R_SEED, FMT, id), \
+	SRI(FMT_DITHER_RAND_G_SEED, FMT, id), \
+	SRI(FMT_DITHER_RAND_B_SEED, FMT, id), \
+	SRI(FMT_CLAMP_CNTL, FMT, id), \
+	SRI(FMT_CLAMP_COMPONENT_R, FMT, id), \
+	SRI(FMT_CLAMP_COMPONENT_G, FMT, id), \
+	SRI(FMT_CLAMP_COMPONENT_B, FMT, id)
+
+#define OPP_DCE_80_REG_LIST(id) \
+	OPP_COMMON_REG_LIST_BASE(id), \
+	SRI(FMT_TEMPORAL_DITHER_PATTERN_CONTROL, FMT, id), \
+	SRI(FMT_TEMPORAL_DITHER_PROGRAMMABLE_PATTERN_S_MATRIX, FMT, id), \
+	SRI(FMT_TEMPORAL_DITHER_PROGRAMMABLE_PATTERN_T_MATRIX, FMT, id)
+
+#define OPP_DCE_100_REG_LIST(id) \
+	OPP_COMMON_REG_LIST_BASE(id), \
+	SRI(FMT_TEMPORAL_DITHER_PATTERN_CONTROL, FMT, id), \
+	SRI(FMT_TEMPORAL_DITHER_PROGRAMMABLE_PATTERN_S_MATRIX, FMT, id), \
+	SRI(FMT_TEMPORAL_DITHER_PROGRAMMABLE_PATTERN_T_MATRIX, FMT, id)
+
+#define OPP_DCE_110_REG_LIST(id) \
+	OPP_COMMON_REG_LIST_BASE(id), \
+	SRI(FMT_TEMPORAL_DITHER_PATTERN_CONTROL, FMT, id), \
+	SRI(FMT_TEMPORAL_DITHER_PROGRAMMABLE_PATTERN_S_MATRIX, FMT, id), \
+	SRI(FMT_TEMPORAL_DITHER_PROGRAMMABLE_PATTERN_T_MATRIX, FMT, id)
+
+#define OPP_DCE_112_REG_LIST(id) \
+	OPP_COMMON_REG_LIST_BASE(id), \
+	SRI(FMT_TEMPORAL_DITHER_PATTERN_CONTROL, FMT, id), \
+	SRI(FMT_TEMPORAL_DITHER_PROGRAMMABLE_PATTERN_S_MATRIX, FMT, id), \
+	SRI(FMT_TEMPORAL_DITHER_PROGRAMMABLE_PATTERN_T_MATRIX, FMT, id), \
+	SRI(CONTROL, FMT_MEMORY, id)
+
+#define OPP_DCE_120_REG_LIST(id) \
+	OPP_COMMON_REG_LIST_BASE(id), \
+	SRI(CONTROL, FMT_MEMORY, id)
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+#define OPP_DCE_60_REG_LIST(id) \
+	SRI(FMT_DYNAMIC_EXP_CNTL, FMT, id), \
+	SRI(FMT_BIT_DEPTH_CONTROL, FMT, id), \
+	SRI(FMT_CONTROL, FMT, id), \
+	SRI(FMT_DITHER_RAND_R_SEED, FMT, id), \
+	SRI(FMT_DITHER_RAND_G_SEED, FMT, id), \
+	SRI(FMT_DITHER_RAND_B_SEED, FMT, id), \
+	SRI(FMT_CLAMP_CNTL, FMT, id)
+#endif
+
+#define OPP_SF(reg_name, field_name, post_fix)\
+	.field_name = reg_name ## __ ## field_name ## post_fix
+
+#define OPP_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(mask_sh)\
+	OPP_SF(FMT_DYNAMIC_EXP_CNTL, FMT_DYNAMIC_EXP_EN, mask_sh),\
+	OPP_SF(FMT_DYNAMIC_EXP_CNTL, FMT_DYNAMIC_EXP_MODE, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_DEPTH, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_MODE, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_EN, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_DEPTH, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_MODE, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_HIGHPASS_RANDOM_ENABLE, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_FRAME_RANDOM_ENABLE, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_RGB_RANDOM_ENABLE, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_TEMPORAL_DITHER_EN, mask_sh),\
+	OPP_SF(FMT_DITHER_RAND_R_SEED, FMT_RAND_R_SEED, mask_sh),\
+	OPP_SF(FMT_DITHER_RAND_G_SEED, FMT_RAND_G_SEED, mask_sh),\
+	OPP_SF(FMT_DITHER_RAND_B_SEED, FMT_RAND_B_SEED, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_TEMPORAL_DITHER_EN, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_TEMPORAL_DITHER_RESET, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_TEMPORAL_DITHER_OFFSET, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_TEMPORAL_DITHER_DEPTH, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_TEMPORAL_LEVEL, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_25FRC_SEL, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_50FRC_SEL, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_75FRC_SEL, mask_sh),\
+	OPP_SF(FMT_CONTROL, FMT_SRC_SELECT, mask_sh),\
+	OPP_SF(FMT_CLAMP_CNTL, FMT_CLAMP_DATA_EN, mask_sh),\
+	OPP_SF(FMT_CLAMP_CNTL, FMT_CLAMP_COLOR_FORMAT, mask_sh),\
+	OPP_SF(FMT_CLAMP_COMPONENT_R, FMT_CLAMP_LOWER_R, mask_sh),\
+	OPP_SF(FMT_CLAMP_COMPONENT_R, FMT_CLAMP_UPPER_R, mask_sh),\
+	OPP_SF(FMT_CLAMP_COMPONENT_G, FMT_CLAMP_LOWER_G, mask_sh),\
+	OPP_SF(FMT_CLAMP_COMPONENT_G, FMT_CLAMP_UPPER_G, mask_sh),\
+	OPP_SF(FMT_CLAMP_COMPONENT_B, FMT_CLAMP_LOWER_B, mask_sh),\
+	OPP_SF(FMT_CLAMP_COMPONENT_B, FMT_CLAMP_UPPER_B, mask_sh),\
+	OPP_SF(FMT_CONTROL, FMT_PIXEL_ENCODING, mask_sh),\
+	OPP_SF(FMT_CONTROL, FMT_SUBSAMPLING_MODE, mask_sh),\
+	OPP_SF(FMT_CONTROL, FMT_SUBSAMPLING_ORDER, mask_sh)
+
+#define OPP_COMMON_MASK_SH_LIST_DCE_110(mask_sh)\
+	OPP_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(mask_sh),\
+	OPP_SF(FMT_CONTROL, FMT_SPATIAL_DITHER_FRAME_COUNTER_MAX, mask_sh),\
+	OPP_SF(FMT_CONTROL, FMT_SPATIAL_DITHER_FRAME_COUNTER_BIT_SWAP, mask_sh),\
+	OPP_SF(FMT_CONTROL, FMT_STEREOSYNC_OVERRIDE, mask_sh)
+
+#define OPP_COMMON_MASK_SH_LIST_DCE_100(mask_sh)\
+	OPP_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(mask_sh),\
+	OPP_SF(FMT_CONTROL, FMT_SPATIAL_DITHER_FRAME_COUNTER_MAX, mask_sh),\
+	OPP_SF(FMT_CONTROL, FMT_SPATIAL_DITHER_FRAME_COUNTER_BIT_SWAP, mask_sh),\
+	OPP_SF(FMT_CONTROL, FMT_STEREOSYNC_OVERRIDE, mask_sh)
+
+#define OPP_COMMON_MASK_SH_LIST_DCE_112(mask_sh)\
+	OPP_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(mask_sh),\
+	OPP_SF(FMT_MEMORY0_CONTROL, FMT420_MEM0_SOURCE_SEL, mask_sh),\
+	OPP_SF(FMT_MEMORY0_CONTROL, FMT420_MEM0_PWR_FORCE, mask_sh),\
+	OPP_SF(FMT_CONTROL, FMT_420_PIXEL_PHASE_LOCKED_CLEAR, mask_sh),\
+	OPP_SF(FMT_CONTROL, FMT_420_PIXEL_PHASE_LOCKED, mask_sh),\
+	OPP_SF(FMT_CONTROL, FMT_CBCR_BIT_REDUCTION_BYPASS, mask_sh),\
+	OPP_SF(FMT_CONTROL, FMT_SPATIAL_DITHER_FRAME_COUNTER_MAX, mask_sh),\
+	OPP_SF(FMT_CONTROL, FMT_SPATIAL_DITHER_FRAME_COUNTER_BIT_SWAP, mask_sh),\
+	OPP_SF(FMT_CONTROL, FMT_STEREOSYNC_OVERRIDE, mask_sh)
+
+#define OPP_COMMON_MASK_SH_LIST_DCE_80(mask_sh)\
+	OPP_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(mask_sh)
+
+#define OPP_COMMON_MASK_SH_LIST_DCE_120(mask_sh)\
+	OPP_SF(FMT0_FMT_DYNAMIC_EXP_CNTL, FMT_DYNAMIC_EXP_EN, mask_sh),\
+	OPP_SF(FMT0_FMT_DYNAMIC_EXP_CNTL, FMT_DYNAMIC_EXP_MODE, mask_sh),\
+	OPP_SF(FMT0_FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, mask_sh),\
+	OPP_SF(FMT0_FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_DEPTH, mask_sh),\
+	OPP_SF(FMT0_FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_MODE, mask_sh),\
+	OPP_SF(FMT0_FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_EN, mask_sh),\
+	OPP_SF(FMT0_FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_DEPTH, mask_sh),\
+	OPP_SF(FMT0_FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_MODE, mask_sh),\
+	OPP_SF(FMT0_FMT_BIT_DEPTH_CONTROL, FMT_TEMPORAL_DITHER_EN, mask_sh),\
+	OPP_SF(FMT0_FMT_BIT_DEPTH_CONTROL, FMT_TEMPORAL_DITHER_RESET, mask_sh),\
+	OPP_SF(FMT0_FMT_BIT_DEPTH_CONTROL, FMT_TEMPORAL_DITHER_OFFSET, mask_sh),\
+	OPP_SF(FMT0_FMT_BIT_DEPTH_CONTROL, FMT_TEMPORAL_DITHER_DEPTH, mask_sh),\
+	OPP_SF(FMT0_FMT_BIT_DEPTH_CONTROL, FMT_TEMPORAL_LEVEL, mask_sh),\
+	OPP_SF(FMT0_FMT_BIT_DEPTH_CONTROL, FMT_25FRC_SEL, mask_sh),\
+	OPP_SF(FMT0_FMT_BIT_DEPTH_CONTROL, FMT_50FRC_SEL, mask_sh),\
+	OPP_SF(FMT0_FMT_BIT_DEPTH_CONTROL, FMT_75FRC_SEL, mask_sh),\
+	OPP_SF(FMT0_FMT_BIT_DEPTH_CONTROL, FMT_HIGHPASS_RANDOM_ENABLE, mask_sh),\
+	OPP_SF(FMT0_FMT_BIT_DEPTH_CONTROL, FMT_FRAME_RANDOM_ENABLE, mask_sh),\
+	OPP_SF(FMT0_FMT_BIT_DEPTH_CONTROL, FMT_RGB_RANDOM_ENABLE, mask_sh),\
+	OPP_SF(FMT0_FMT_BIT_DEPTH_CONTROL, FMT_TEMPORAL_DITHER_EN, mask_sh),\
+	OPP_SF(FMT0_FMT_CONTROL, FMT_SPATIAL_DITHER_FRAME_COUNTER_MAX, mask_sh),\
+	OPP_SF(FMT0_FMT_CONTROL, FMT_SPATIAL_DITHER_FRAME_COUNTER_BIT_SWAP, mask_sh),\
+	OPP_SF(FMT0_FMT_CONTROL, FMT_STEREOSYNC_OVERRIDE, mask_sh),\
+	OPP_SF(FMT0_FMT_DITHER_RAND_R_SEED, FMT_RAND_R_SEED, mask_sh),\
+	OPP_SF(FMT0_FMT_DITHER_RAND_G_SEED, FMT_RAND_G_SEED, mask_sh),\
+	OPP_SF(FMT0_FMT_DITHER_RAND_B_SEED, FMT_RAND_B_SEED, mask_sh),\
+	OPP_SF(FMT_MEMORY0_CONTROL, FMT420_MEM0_SOURCE_SEL, mask_sh),\
+	OPP_SF(FMT_MEMORY0_CONTROL, FMT420_MEM0_PWR_FORCE, mask_sh),\
+	OPP_SF(FMT0_FMT_CONTROL, FMT_SRC_SELECT, mask_sh),\
+	OPP_SF(FMT0_FMT_CONTROL, FMT_420_PIXEL_PHASE_LOCKED_CLEAR, mask_sh),\
+	OPP_SF(FMT0_FMT_CONTROL, FMT_420_PIXEL_PHASE_LOCKED, mask_sh),\
+	OPP_SF(FMT0_FMT_CLAMP_CNTL, FMT_CLAMP_DATA_EN, mask_sh),\
+	OPP_SF(FMT0_FMT_CLAMP_CNTL, FMT_CLAMP_COLOR_FORMAT, mask_sh),\
+	OPP_SF(FMT0_FMT_CLAMP_COMPONENT_R, FMT_CLAMP_LOWER_R, mask_sh),\
+	OPP_SF(FMT0_FMT_CLAMP_COMPONENT_R, FMT_CLAMP_UPPER_R, mask_sh),\
+	OPP_SF(FMT0_FMT_CLAMP_COMPONENT_G, FMT_CLAMP_LOWER_G, mask_sh),\
+	OPP_SF(FMT0_FMT_CLAMP_COMPONENT_G, FMT_CLAMP_UPPER_G, mask_sh),\
+	OPP_SF(FMT0_FMT_CLAMP_COMPONENT_B, FMT_CLAMP_LOWER_B, mask_sh),\
+	OPP_SF(FMT0_FMT_CLAMP_COMPONENT_B, FMT_CLAMP_UPPER_B, mask_sh),\
+	OPP_SF(FMT0_FMT_CONTROL, FMT_PIXEL_ENCODING, mask_sh),\
+	OPP_SF(FMT0_FMT_CONTROL, FMT_SUBSAMPLING_MODE, mask_sh),\
+	OPP_SF(FMT0_FMT_CONTROL, FMT_SUBSAMPLING_ORDER, mask_sh),\
+	OPP_SF(FMT0_FMT_CONTROL, FMT_CBCR_BIT_REDUCTION_BYPASS, mask_sh)
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+#define OPP_COMMON_MASK_SH_LIST_DCE_60(mask_sh)\
+	OPP_SF(FMT_DYNAMIC_EXP_CNTL, FMT_DYNAMIC_EXP_EN, mask_sh),\
+	OPP_SF(FMT_DYNAMIC_EXP_CNTL, FMT_DYNAMIC_EXP_MODE, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_DEPTH, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_EN, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_DEPTH, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_MODE, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_HIGHPASS_RANDOM_ENABLE, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_FRAME_RANDOM_ENABLE, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_RGB_RANDOM_ENABLE, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_TEMPORAL_DITHER_EN, mask_sh),\
+	OPP_SF(FMT_DITHER_RAND_R_SEED, FMT_RAND_R_SEED, mask_sh),\
+	OPP_SF(FMT_DITHER_RAND_G_SEED, FMT_RAND_G_SEED, mask_sh),\
+	OPP_SF(FMT_DITHER_RAND_B_SEED, FMT_RAND_B_SEED, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_TEMPORAL_DITHER_RESET, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_TEMPORAL_DITHER_OFFSET, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_TEMPORAL_DITHER_DEPTH, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_TEMPORAL_LEVEL, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_25FRC_SEL, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_50FRC_SEL, mask_sh),\
+	OPP_SF(FMT_BIT_DEPTH_CONTROL, FMT_75FRC_SEL, mask_sh),\
+	OPP_SF(FMT_CLAMP_CNTL, FMT_CLAMP_DATA_EN, mask_sh),\
+	OPP_SF(FMT_CLAMP_CNTL, FMT_CLAMP_COLOR_FORMAT, mask_sh),\
+	OPP_SF(FMT_CONTROL, FMT_PIXEL_ENCODING, mask_sh)
+#endif
+
+#define OPP_REG_FIELD_LIST(type) \
+	type FMT_DYNAMIC_EXP_EN; \
+	type FMT_DYNAMIC_EXP_MODE; \
+	type FMT_TRUNCATE_EN; \
+	type FMT_TRUNCATE_DEPTH; \
+	type FMT_TRUNCATE_MODE; \
+	type FMT_SPATIAL_DITHER_EN; \
+	type FMT_SPATIAL_DITHER_DEPTH; \
+	type FMT_SPATIAL_DITHER_MODE; \
+	type FMT_TEMPORAL_DITHER_EN; \
+	type FMT_TEMPORAL_DITHER_RESET; \
+	type FMT_TEMPORAL_DITHER_OFFSET; \
+	type FMT_TEMPORAL_DITHER_DEPTH; \
+	type FMT_TEMPORAL_LEVEL; \
+	type FMT_25FRC_SEL; \
+	type FMT_50FRC_SEL; \
+	type FMT_75FRC_SEL; \
+	type FMT_HIGHPASS_RANDOM_ENABLE; \
+	type FMT_FRAME_RANDOM_ENABLE; \
+	type FMT_RGB_RANDOM_ENABLE; \
+	type FMT_SPATIAL_DITHER_FRAME_COUNTER_MAX; \
+	type FMT_SPATIAL_DITHER_FRAME_COUNTER_BIT_SWAP; \
+	type FMT_STEREOSYNC_OVERRIDE; \
+	type FMT_RAND_R_SEED; \
+	type FMT_RAND_G_SEED; \
+	type FMT_RAND_B_SEED; \
+	type FMT420_MEM0_SOURCE_SEL; \
+	type FMT420_MEM0_PWR_FORCE; \
+	type FMT_SRC_SELECT; \
+	type FMT_420_PIXEL_PHASE_LOCKED_CLEAR; \
+	type FMT_420_PIXEL_PHASE_LOCKED; \
+	type FMT_CLAMP_DATA_EN; \
+	type FMT_CLAMP_COLOR_FORMAT; \
+	type FMT_CLAMP_LOWER_R; \
+	type FMT_CLAMP_UPPER_R; \
+	type FMT_CLAMP_LOWER_G; \
+	type FMT_CLAMP_UPPER_G; \
+	type FMT_CLAMP_LOWER_B; \
+	type FMT_CLAMP_UPPER_B; \
+	type FMT_PIXEL_ENCODING; \
+	type FMT_SUBSAMPLING_ORDER; \
+	type FMT_SUBSAMPLING_MODE; \
+	type FMT_CBCR_BIT_REDUCTION_BYPASS;\
+
+struct dce_opp_shift {
+	OPP_REG_FIELD_LIST(uint8_t)
+};
+
+struct dce_opp_mask {
+	OPP_REG_FIELD_LIST(uint32_t)
+};
+
+struct dce_opp_registers {
+	uint32_t FMT_DYNAMIC_EXP_CNTL;
+	uint32_t FMT_BIT_DEPTH_CONTROL;
+	uint32_t FMT_CONTROL;
+	uint32_t FMT_DITHER_RAND_R_SEED;
+	uint32_t FMT_DITHER_RAND_G_SEED;
+	uint32_t FMT_DITHER_RAND_B_SEED;
+	uint32_t FMT_TEMPORAL_DITHER_PATTERN_CONTROL;
+	uint32_t FMT_TEMPORAL_DITHER_PROGRAMMABLE_PATTERN_S_MATRIX;
+	uint32_t FMT_TEMPORAL_DITHER_PROGRAMMABLE_PATTERN_T_MATRIX;
+	uint32_t CONTROL;
+	uint32_t FMT_CLAMP_CNTL;
+	uint32_t FMT_CLAMP_COMPONENT_R;
+	uint32_t FMT_CLAMP_COMPONENT_G;
+	uint32_t FMT_CLAMP_COMPONENT_B;
+};
+
+/* OPP RELATED */
+#define TO_DCE110_OPP(opp)\
+	container_of(opp, struct dce110_opp, base)
+
+struct dce110_opp {
+	struct output_pixel_processor base;
+	const struct dce_opp_registers *regs;
+	const struct dce_opp_shift *opp_shift;
+	const struct dce_opp_mask *opp_mask;
+};
+
+void dce110_opp_construct(struct dce110_opp *opp110,
+	struct dc_context *ctx,
+	uint32_t inst,
+	const struct dce_opp_registers *regs,
+	const struct dce_opp_shift *opp_shift,
+	const struct dce_opp_mask *opp_mask);
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+void dce60_opp_construct(struct dce110_opp *opp110,
+	struct dc_context *ctx,
+	uint32_t inst,
+	const struct dce_opp_registers *regs,
+	const struct dce_opp_shift *opp_shift,
+	const struct dce_opp_mask *opp_mask);
+#endif
+
+void dce110_opp_destroy(struct output_pixel_processor **opp);
+
+
+
+/* FORMATTER RELATED */
+void dce110_opp_program_bit_depth_reduction(
+	struct output_pixel_processor *opp,
+	const struct bit_depth_reduction_params *params);
+
+void dce110_opp_program_clamping_and_pixel_encoding(
+	struct output_pixel_processor *opp,
+	const struct clamping_and_pixel_encoding_params *params);
+
+void dce110_opp_set_dyn_expansion(
+	struct output_pixel_processor *opp,
+	enum dc_color_space color_sp,
+	enum dc_color_depth color_dpth,
+	enum signal_type signal);
+
+void dce110_opp_program_fmt(
+	struct output_pixel_processor *opp,
+	struct bit_depth_reduction_params *fmt_bit_depth,
+	struct clamping_and_pixel_encoding_params *clamping);
+
+void dce110_opp_set_clamping(
+	struct dce110_opp *opp110,
+	const struct clamping_and_pixel_encoding_params *params);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_panel_cntl.c b/drivers/gpu/drm/amd/display/dc/dce/dce_panel_cntl.c
new file mode 100644
index 000000000000..de31fb1b6819
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_panel_cntl.c
@@ -0,0 +1,294 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+#include "core_types.h"
+#include "dc_dmub_srv.h"
+#include "panel_cntl.h"
+#include "dce_panel_cntl.h"
+#include "atom.h"
+
+#define TO_DCE_PANEL_CNTL(panel_cntl)\
+	container_of(panel_cntl, struct dce_panel_cntl, base)
+
+#define CTX \
+	dce_panel_cntl->base.ctx
+
+#define DC_LOGGER \
+	dce_panel_cntl->base.ctx->logger
+
+#define REG(reg)\
+	dce_panel_cntl->regs->reg
+
+#undef FN
+#define FN(reg_name, field_name) \
+	dce_panel_cntl->shift->field_name, dce_panel_cntl->mask->field_name
+
+static unsigned int dce_get_16_bit_backlight_from_pwm(struct panel_cntl *panel_cntl)
+{
+	uint64_t current_backlight;
+	uint32_t bl_period, bl_int_count;
+	uint32_t bl_pwm, fractional_duty_cycle_en;
+	uint32_t bl_period_mask, bl_pwm_mask;
+	struct dce_panel_cntl *dce_panel_cntl = TO_DCE_PANEL_CNTL(panel_cntl);
+
+	REG_READ(BL_PWM_PERIOD_CNTL);
+	REG_GET(BL_PWM_PERIOD_CNTL, BL_PWM_PERIOD, &bl_period);
+	REG_GET(BL_PWM_PERIOD_CNTL, BL_PWM_PERIOD_BITCNT, &bl_int_count);
+
+	REG_READ(BL_PWM_CNTL);
+	REG_GET(BL_PWM_CNTL, BL_ACTIVE_INT_FRAC_CNT, (uint32_t *)(&bl_pwm));
+	REG_GET(BL_PWM_CNTL, BL_PWM_FRACTIONAL_EN, &fractional_duty_cycle_en);
+
+	if (bl_int_count == 0)
+		bl_int_count = 16;
+
+	bl_period_mask = (1 << bl_int_count) - 1;
+	bl_period &= bl_period_mask;
+
+	bl_pwm_mask = bl_period_mask << (16 - bl_int_count);
+
+	if (fractional_duty_cycle_en == 0)
+		bl_pwm &= bl_pwm_mask;
+	else
+		bl_pwm &= 0xFFFF;
+
+	current_backlight = (uint64_t)bl_pwm << (1 + bl_int_count);
+
+	if (bl_period == 0)
+		bl_period = 0xFFFF;
+
+	current_backlight = div_u64(current_backlight, bl_period);
+	current_backlight = (current_backlight + 1) >> 1;
+
+	return (uint32_t)(current_backlight);
+}
+
+static uint32_t dce_panel_cntl_hw_init(struct panel_cntl *panel_cntl)
+{
+	struct dce_panel_cntl *dce_panel_cntl = TO_DCE_PANEL_CNTL(panel_cntl);
+	uint32_t value;
+	uint32_t current_backlight;
+
+	/* It must not be 0, so we have to restore them
+	 * Bios bug w/a - period resets to zero,
+	 * restoring to cache values which is always correct
+	 */
+	REG_GET(BL_PWM_CNTL, BL_ACTIVE_INT_FRAC_CNT, &value);
+
+	if (panel_cntl->stored_backlight_registers.BL_PWM_CNTL != 0) {
+		REG_WRITE(BL_PWM_CNTL,
+				panel_cntl->stored_backlight_registers.BL_PWM_CNTL);
+		REG_WRITE(BL_PWM_CNTL2,
+				panel_cntl->stored_backlight_registers.BL_PWM_CNTL2);
+		REG_WRITE(BL_PWM_PERIOD_CNTL,
+				panel_cntl->stored_backlight_registers.BL_PWM_PERIOD_CNTL);
+		REG_UPDATE(PWRSEQ_REF_DIV,
+			BL_PWM_REF_DIV,
+			panel_cntl->stored_backlight_registers.LVTMA_PWRSEQ_REF_DIV_BL_PWM_REF_DIV);
+	} else if ((value != 0) && (value != 1)) {
+		panel_cntl->stored_backlight_registers.BL_PWM_CNTL =
+				REG_READ(BL_PWM_CNTL);
+		panel_cntl->stored_backlight_registers.BL_PWM_CNTL2 =
+				REG_READ(BL_PWM_CNTL2);
+		panel_cntl->stored_backlight_registers.BL_PWM_PERIOD_CNTL =
+				REG_READ(BL_PWM_PERIOD_CNTL);
+
+		REG_GET(PWRSEQ_REF_DIV, BL_PWM_REF_DIV,
+				&panel_cntl->stored_backlight_registers.LVTMA_PWRSEQ_REF_DIV_BL_PWM_REF_DIV);
+	} else {
+		/* TODO: Note: This should not really happen since VBIOS
+		 * should have initialized PWM registers on boot.
+		 */
+		REG_WRITE(BL_PWM_CNTL, 0x8000FA00);
+		REG_WRITE(BL_PWM_PERIOD_CNTL, 0x000C0FA0);
+	}
+
+	// Have driver take backlight control
+	// TakeBacklightControl(true)
+	value = REG_READ(BIOS_SCRATCH_2);
+	value |= ATOM_S2_VRI_BRIGHT_ENABLE;
+	REG_WRITE(BIOS_SCRATCH_2, value);
+
+	// Enable the backlight output
+	REG_UPDATE(BL_PWM_CNTL, BL_PWM_EN, 1);
+
+	// Unlock group 2 backlight registers
+	REG_UPDATE(BL_PWM_GRP1_REG_LOCK,
+			BL_PWM_GRP1_REG_LOCK, 0);
+
+	current_backlight = dce_get_16_bit_backlight_from_pwm(panel_cntl);
+
+	return current_backlight;
+}
+
+static bool dce_is_panel_backlight_on(struct panel_cntl *panel_cntl)
+{
+	struct dce_panel_cntl *dce_panel_cntl = TO_DCE_PANEL_CNTL(panel_cntl);
+	uint32_t blon, blon_ovrd, pwrseq_target_state;
+
+	REG_GET_2(PWRSEQ_CNTL, LVTMA_BLON, &blon, LVTMA_BLON_OVRD, &blon_ovrd);
+	REG_GET(PWRSEQ_CNTL, LVTMA_PWRSEQ_TARGET_STATE, &pwrseq_target_state);
+
+	if (blon_ovrd)
+		return blon;
+	else
+		return pwrseq_target_state;
+}
+
+static bool dce_is_panel_powered_on(struct panel_cntl *panel_cntl)
+{
+	struct dce_panel_cntl *dce_panel_cntl = TO_DCE_PANEL_CNTL(panel_cntl);
+	uint32_t pwr_seq_state, dig_on, dig_on_ovrd;
+
+	REG_GET(PWRSEQ_STATE, LVTMA_PWRSEQ_TARGET_STATE_R, &pwr_seq_state);
+
+	REG_GET_2(PWRSEQ_CNTL, LVTMA_DIGON, &dig_on, LVTMA_DIGON_OVRD, &dig_on_ovrd);
+
+	return (pwr_seq_state == 1) || (dig_on == 1 && dig_on_ovrd == 1);
+}
+
+static void dce_store_backlight_level(struct panel_cntl *panel_cntl)
+{
+	struct dce_panel_cntl *dce_panel_cntl = TO_DCE_PANEL_CNTL(panel_cntl);
+
+	panel_cntl->stored_backlight_registers.BL_PWM_CNTL =
+		REG_READ(BL_PWM_CNTL);
+	panel_cntl->stored_backlight_registers.BL_PWM_CNTL2 =
+		REG_READ(BL_PWM_CNTL2);
+	panel_cntl->stored_backlight_registers.BL_PWM_PERIOD_CNTL =
+		REG_READ(BL_PWM_PERIOD_CNTL);
+
+	REG_GET(PWRSEQ_REF_DIV, BL_PWM_REF_DIV,
+		&panel_cntl->stored_backlight_registers.LVTMA_PWRSEQ_REF_DIV_BL_PWM_REF_DIV);
+}
+
+static void dce_driver_set_backlight(struct panel_cntl *panel_cntl,
+		uint32_t backlight_pwm_u16_16)
+{
+	uint32_t backlight_16bit;
+	uint32_t masked_pwm_period;
+	uint8_t bit_count;
+	uint64_t active_duty_cycle;
+	uint32_t pwm_period_bitcnt;
+	struct dce_panel_cntl *dce_panel_cntl = TO_DCE_PANEL_CNTL(panel_cntl);
+
+	/*
+	 * 1. Find  16 bit backlight active duty cycle, where 0 <= backlight
+	 * active duty cycle <= backlight period
+	 */
+
+	/* 1.1 Apply bitmask for backlight period value based on value of BITCNT
+	 */
+	REG_GET_2(BL_PWM_PERIOD_CNTL,
+			BL_PWM_PERIOD_BITCNT, &pwm_period_bitcnt,
+			BL_PWM_PERIOD, &masked_pwm_period);
+
+	if (pwm_period_bitcnt == 0)
+		bit_count = 16;
+	else
+		bit_count = pwm_period_bitcnt;
+
+	/* e.g. maskedPwmPeriod = 0x24 when bitCount is 6 */
+	masked_pwm_period = masked_pwm_period & ((1 << bit_count) - 1);
+
+	/* 1.2 Calculate integer active duty cycle required upper 16 bits
+	 * contain integer component, lower 16 bits contain fractional component
+	 * of active duty cycle e.g. 0x21BDC0 = 0xEFF0 * 0x24
+	 */
+	active_duty_cycle = backlight_pwm_u16_16 * (uint64_t)masked_pwm_period;
+
+	/* 1.3 Calculate 16 bit active duty cycle from integer and fractional
+	 * components shift by bitCount then mask 16 bits and add rounding bit
+	 * from MSB of fraction e.g. 0x86F7 = ((0x21BDC0 >> 6) & 0xFFF) + 0
+	 */
+	backlight_16bit = active_duty_cycle >> bit_count;
+	backlight_16bit &= 0xFFFF;
+	backlight_16bit += (active_duty_cycle >> (bit_count - 1)) & 0x1;
+
+	/*
+	 * 2. Program register with updated value
+	 */
+
+	/* 2.1 Lock group 2 backlight registers */
+
+	REG_UPDATE_2(BL_PWM_GRP1_REG_LOCK,
+			BL_PWM_GRP1_IGNORE_MASTER_LOCK_EN, 1,
+			BL_PWM_GRP1_REG_LOCK, 1);
+
+	// 2.2 Write new active duty cycle
+	REG_UPDATE(BL_PWM_CNTL, BL_ACTIVE_INT_FRAC_CNT, backlight_16bit);
+
+	/* 2.3 Unlock group 2 backlight registers */
+	REG_UPDATE(BL_PWM_GRP1_REG_LOCK,
+			BL_PWM_GRP1_REG_LOCK, 0);
+
+	/* 3 Wait for pending bit to be cleared */
+	REG_WAIT(BL_PWM_GRP1_REG_LOCK,
+			BL_PWM_GRP1_REG_UPDATE_PENDING, 0,
+			1, 10000);
+}
+
+static void dce_panel_cntl_destroy(struct panel_cntl **panel_cntl)
+{
+	struct dce_panel_cntl *dce_panel_cntl = TO_DCE_PANEL_CNTL(*panel_cntl);
+
+	kfree(dce_panel_cntl);
+	*panel_cntl = NULL;
+}
+
+static const struct panel_cntl_funcs dce_link_panel_cntl_funcs = {
+	.destroy = dce_panel_cntl_destroy,
+	.hw_init = dce_panel_cntl_hw_init,
+	.is_panel_backlight_on = dce_is_panel_backlight_on,
+	.is_panel_powered_on = dce_is_panel_powered_on,
+	.store_backlight_level = dce_store_backlight_level,
+	.driver_set_backlight = dce_driver_set_backlight,
+	.get_current_backlight = dce_get_16_bit_backlight_from_pwm,
+};
+
+void dce_panel_cntl_construct(
+	struct dce_panel_cntl *dce_panel_cntl,
+	const struct panel_cntl_init_data *init_data,
+	const struct dce_panel_cntl_registers *regs,
+	const struct dce_panel_cntl_shift *shift,
+	const struct dce_panel_cntl_mask *mask)
+{
+	struct panel_cntl *base = &dce_panel_cntl->base;
+
+	base->stored_backlight_registers.BL_PWM_CNTL = 0;
+	base->stored_backlight_registers.BL_PWM_CNTL2 = 0;
+	base->stored_backlight_registers.BL_PWM_PERIOD_CNTL = 0;
+	base->stored_backlight_registers.LVTMA_PWRSEQ_REF_DIV_BL_PWM_REF_DIV = 0;
+
+	dce_panel_cntl->regs = regs;
+	dce_panel_cntl->shift = shift;
+	dce_panel_cntl->mask = mask;
+
+	dce_panel_cntl->base.funcs = &dce_link_panel_cntl_funcs;
+	dce_panel_cntl->base.ctx = init_data->ctx;
+	dce_panel_cntl->base.inst = init_data->inst;
+	dce_panel_cntl->base.pwrseq_inst = 0;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_panel_cntl.h b/drivers/gpu/drm/amd/display/dc/dce/dce_panel_cntl.h
new file mode 100644
index 000000000000..6bd1196083a3
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_panel_cntl.h
@@ -0,0 +1,129 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_PANEL_CNTL__DCE_H__
+#define __DC_PANEL_CNTL__DCE_H__
+
+#include "panel_cntl.h"
+
+/* set register offset with instance */
+#define DCE_PANEL_CNTL_SR(reg_name, block)\
+	.reg_name = mm ## block ## _ ## reg_name
+
+#define DCE_PANEL_CNTL_REG_LIST()\
+	DCE_PANEL_CNTL_SR(PWRSEQ_CNTL, LVTMA), \
+	DCE_PANEL_CNTL_SR(PWRSEQ_STATE, LVTMA), \
+	DCE_PANEL_CNTL_SR(PWRSEQ_REF_DIV, LVTMA), \
+	SR(BL_PWM_CNTL), \
+	SR(BL_PWM_CNTL2), \
+	SR(BL_PWM_PERIOD_CNTL), \
+	SR(BL_PWM_GRP1_REG_LOCK), \
+	SR(BIOS_SCRATCH_2)
+
+#define DCN_PANEL_CNTL_SR(reg_name, block)\
+	.reg_name = BASE(mm ## block ## _ ## reg_name ## _BASE_IDX) + \
+					mm ## block ## _ ## reg_name
+
+#define DCN_PANEL_CNTL_REG_LIST()\
+	DCN_PANEL_CNTL_SR(PWRSEQ_CNTL, LVTMA), \
+	DCN_PANEL_CNTL_SR(PWRSEQ_STATE, LVTMA), \
+	DCN_PANEL_CNTL_SR(PWRSEQ_REF_DIV, LVTMA), \
+	SR(BL_PWM_CNTL), \
+	SR(BL_PWM_CNTL2), \
+	SR(BL_PWM_PERIOD_CNTL), \
+	SR(BL_PWM_GRP1_REG_LOCK), \
+	NBIO_SR(BIOS_SCRATCH_2)
+
+#define DCE_PANEL_CNTL_SF(reg_name, field_name, post_fix)\
+	.field_name = reg_name ## __ ## field_name ## post_fix
+
+#define DCE_PANEL_CNTL_MASK_SH_LIST(mask_sh) \
+	DCE_PANEL_CNTL_SF(LVTMA_PWRSEQ_CNTL, LVTMA_BLON, mask_sh),\
+	DCE_PANEL_CNTL_SF(LVTMA_PWRSEQ_CNTL, LVTMA_BLON_OVRD, mask_sh),\
+	DCE_PANEL_CNTL_SF(LVTMA_PWRSEQ_CNTL, LVTMA_DIGON, mask_sh),\
+	DCE_PANEL_CNTL_SF(LVTMA_PWRSEQ_CNTL, LVTMA_DIGON_OVRD, mask_sh),\
+	DCE_PANEL_CNTL_SF(LVTMA_PWRSEQ_CNTL, LVTMA_PWRSEQ_TARGET_STATE, mask_sh), \
+	DCE_PANEL_CNTL_SF(LVTMA_PWRSEQ_STATE, LVTMA_PWRSEQ_TARGET_STATE_R, mask_sh), \
+	DCE_PANEL_CNTL_SF(LVTMA_PWRSEQ_REF_DIV, BL_PWM_REF_DIV, mask_sh), \
+	DCE_PANEL_CNTL_SF(BL_PWM_PERIOD_CNTL, BL_PWM_PERIOD, mask_sh), \
+	DCE_PANEL_CNTL_SF(BL_PWM_PERIOD_CNTL, BL_PWM_PERIOD_BITCNT, mask_sh), \
+	DCE_PANEL_CNTL_SF(BL_PWM_CNTL, BL_ACTIVE_INT_FRAC_CNT, mask_sh), \
+	DCE_PANEL_CNTL_SF(BL_PWM_CNTL, BL_PWM_FRACTIONAL_EN, mask_sh), \
+	DCE_PANEL_CNTL_SF(BL_PWM_CNTL, BL_PWM_EN, mask_sh), \
+	DCE_PANEL_CNTL_SF(BL_PWM_GRP1_REG_LOCK, BL_PWM_GRP1_IGNORE_MASTER_LOCK_EN, mask_sh), \
+	DCE_PANEL_CNTL_SF(BL_PWM_GRP1_REG_LOCK, BL_PWM_GRP1_REG_LOCK, mask_sh), \
+	DCE_PANEL_CNTL_SF(BL_PWM_GRP1_REG_LOCK, BL_PWM_GRP1_REG_UPDATE_PENDING, mask_sh)
+
+#define DCE_PANEL_CNTL_REG_FIELD_LIST(type) \
+	type LVTMA_BLON;\
+	type LVTMA_BLON_OVRD;\
+	type LVTMA_DIGON;\
+	type LVTMA_DIGON_OVRD;\
+	type LVTMA_PWRSEQ_TARGET_STATE; \
+	type LVTMA_PWRSEQ_TARGET_STATE_R; \
+	type BL_PWM_REF_DIV; \
+	type BL_PWM_EN; \
+	type BL_ACTIVE_INT_FRAC_CNT; \
+	type BL_PWM_FRACTIONAL_EN; \
+	type BL_PWM_PERIOD; \
+	type BL_PWM_PERIOD_BITCNT; \
+	type BL_PWM_GRP1_IGNORE_MASTER_LOCK_EN; \
+	type BL_PWM_GRP1_REG_LOCK; \
+	type BL_PWM_GRP1_REG_UPDATE_PENDING
+
+struct dce_panel_cntl_shift {
+	DCE_PANEL_CNTL_REG_FIELD_LIST(uint8_t);
+};
+
+struct dce_panel_cntl_mask {
+	DCE_PANEL_CNTL_REG_FIELD_LIST(uint32_t);
+};
+
+struct dce_panel_cntl_registers {
+	uint32_t PWRSEQ_CNTL;
+	uint32_t PWRSEQ_STATE;
+	uint32_t BL_PWM_CNTL;
+	uint32_t BL_PWM_CNTL2;
+	uint32_t BL_PWM_PERIOD_CNTL;
+	uint32_t BL_PWM_GRP1_REG_LOCK;
+	uint32_t PWRSEQ_REF_DIV;
+	uint32_t BIOS_SCRATCH_2;
+};
+
+struct dce_panel_cntl {
+	struct panel_cntl base;
+	const struct dce_panel_cntl_registers *regs;
+	const struct dce_panel_cntl_shift *shift;
+	const struct dce_panel_cntl_mask *mask;
+};
+
+void dce_panel_cntl_construct(
+	struct dce_panel_cntl *panel_cntl,
+	const struct panel_cntl_init_data *init_data,
+	const struct dce_panel_cntl_registers *regs,
+	const struct dce_panel_cntl_shift *shift,
+	const struct dce_panel_cntl_mask *mask);
+
+#endif /* __DC_PANEL_CNTL__DCE_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_scl_filters.c b/drivers/gpu/drm/amd/display/dc/dce/dce_scl_filters.c
new file mode 100644
index 000000000000..7311f312369f
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_scl_filters.c
@@ -0,0 +1,1445 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+#include "transform.h"
+
+//=========================================
+// <num_taps>    = 2
+// <num_phases>  = 16
+// <scale_ratio> = 0.833333 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = s1.10
+// <CoefOut>     = s1.12
+//=========================================
+static const uint16_t filter_2tap_16p[18] = {
+		0x1000, 0x0000,
+		0x0FF0, 0x0010,
+		0x0FB0, 0x0050,
+		0x0F34, 0x00CC,
+		0x0E68, 0x0198,
+		0x0D44, 0x02BC,
+		0x0BC4, 0x043C,
+		0x09FC, 0x0604,
+		0x0800, 0x0800
+};
+
+//=========================================
+// <num_taps>    = 3
+// <num_phases>  = 16
+// <scale_ratio> = 0.83333 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_3tap_16p_upscale[27] = {
+		0x0804, 0x07FC, 0x0000,
+		0x06AC, 0x0978, 0x3FDC,
+		0x055C, 0x0AF0, 0x3FB4,
+		0x0420, 0x0C50, 0x3F90,
+		0x0300, 0x0D88, 0x3F78,
+		0x0200, 0x0E90, 0x3F70,
+		0x0128, 0x0F5C, 0x3F7C,
+		0x007C, 0x0FD8, 0x3FAC,
+		0x0000, 0x1000, 0x0000
+};
+
+//=========================================
+// <num_taps>    = 3
+// <num_phases>  = 16
+// <scale_ratio> = 1.16666 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_3tap_16p_116[27] = {
+		0x0804, 0x07FC, 0x0000,
+		0x0700, 0x0914, 0x3FEC,
+		0x0604, 0x0A1C, 0x3FE0,
+		0x050C, 0x0B14, 0x3FE0,
+		0x041C, 0x0BF4, 0x3FF0,
+		0x0340, 0x0CB0, 0x0010,
+		0x0274, 0x0D3C, 0x0050,
+		0x01C0, 0x0D94, 0x00AC,
+		0x0128, 0x0DB4, 0x0124
+};
+
+//=========================================
+// <num_taps>    = 3
+// <num_phases>  = 16
+// <scale_ratio> = 1.49999 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_3tap_16p_149[27] = {
+		0x0804, 0x07FC, 0x0000,
+		0x0730, 0x08CC, 0x0004,
+		0x0660, 0x098C, 0x0014,
+		0x0590, 0x0A3C, 0x0034,
+		0x04C4, 0x0AD4, 0x0068,
+		0x0400, 0x0B54, 0x00AC,
+		0x0348, 0x0BB0, 0x0108,
+		0x029C, 0x0BEC, 0x0178,
+		0x0200, 0x0C00, 0x0200
+};
+
+//=========================================
+// <num_taps>    = 3
+// <num_phases>  = 16
+// <scale_ratio> = 1.83332 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_3tap_16p_183[27] = {
+		0x0804, 0x07FC, 0x0000,
+		0x0754, 0x0880, 0x002C,
+		0x06A8, 0x08F0, 0x0068,
+		0x05FC, 0x0954, 0x00B0,
+		0x0550, 0x09AC, 0x0104,
+		0x04A8, 0x09F0, 0x0168,
+		0x0408, 0x0A20, 0x01D8,
+		0x036C, 0x0A40, 0x0254,
+		0x02DC, 0x0A48, 0x02DC
+};
+
+//=========================================
+// <num_taps>    = 4
+// <num_phases>  = 16
+// <scale_ratio> = 0.83333 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_4tap_16p_upscale[36] = {
+		0x0000, 0x1000, 0x0000, 0x0000,
+		0x3F74, 0x0FDC, 0x00B4, 0x3FFC,
+		0x3F0C, 0x0F70, 0x0194, 0x3FF0,
+		0x3ECC, 0x0EC4, 0x0298, 0x3FD8,
+		0x3EAC, 0x0DE4, 0x03B8, 0x3FB8,
+		0x3EA4, 0x0CD8, 0x04F4, 0x3F90,
+		0x3EB8, 0x0BA0, 0x0644, 0x3F64,
+		0x3ED8, 0x0A54, 0x07A0, 0x3F34,
+		0x3F00, 0x08FC, 0x0900, 0x3F04
+};
+
+//=========================================
+// <num_taps>    = 4
+// <num_phases>  = 16
+// <scale_ratio> = 1.16666 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_4tap_16p_116[36] = {
+		0x01A8, 0x0CB4, 0x01A4, 0x0000,
+		0x0110, 0x0CB0, 0x0254, 0x3FEC,
+		0x0090, 0x0C80, 0x031C, 0x3FD4,
+		0x0024, 0x0C2C, 0x03F4, 0x3FBC,
+		0x3FD8, 0x0BAC, 0x04DC, 0x3FA0,
+		0x3F9C, 0x0B14, 0x05CC, 0x3F84,
+		0x3F70, 0x0A60, 0x06C4, 0x3F6C,
+		0x3F5C, 0x098C, 0x07BC, 0x3F5C,
+		0x3F54, 0x08AC, 0x08AC, 0x3F54
+};
+
+//=========================================
+// <num_taps>    = 4
+// <num_phases>  = 16
+// <scale_ratio> = 1.49999 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_4tap_16p_149[36] = {
+		0x02B8, 0x0A90, 0x02B8, 0x0000,
+		0x0230, 0x0A90, 0x0350, 0x3FF0,
+		0x01B8, 0x0A78, 0x03F0, 0x3FE0,
+		0x0148, 0x0A48, 0x049C, 0x3FD4,
+		0x00E8, 0x0A00, 0x054C, 0x3FCC,
+		0x0098, 0x09A0, 0x0600, 0x3FC8,
+		0x0054, 0x0928, 0x06B4, 0x3FD0,
+		0x001C, 0x08A4, 0x0760, 0x3FE0,
+		0x3FFC, 0x0804, 0x0804, 0x3FFC
+};
+
+//=========================================
+// <num_taps>    = 4
+// <num_phases>  = 16
+// <scale_ratio> = 1.83332 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_4tap_16p_183[36] = {
+		0x03B0, 0x08A0, 0x03B0, 0x0000,
+		0x0348, 0x0898, 0x041C, 0x0004,
+		0x02DC, 0x0884, 0x0490, 0x0010,
+		0x0278, 0x0864, 0x0500, 0x0024,
+		0x021C, 0x0838, 0x0570, 0x003C,
+		0x01C8, 0x07FC, 0x05E0, 0x005C,
+		0x0178, 0x07B8, 0x064C, 0x0084,
+		0x0130, 0x076C, 0x06B0, 0x00B4,
+		0x00F0, 0x0714, 0x0710, 0x00EC
+};
+
+//=========================================
+// <num_taps>    = 2
+// <num_phases>  = 64
+// <scale_ratio> = 0.833333 (input/output)
+// <sharpness>   = 0
+// <CoefType> = ModifiedLanczos
+// <CoefQuant> = s1.10
+// <CoefOut> = s1.12
+//=========================================
+static const uint16_t filter_2tap_64p[66] = {
+		0x1000, 0x0000,
+		0x1000, 0x0000,
+		0x0FFC, 0x0004,
+		0x0FF8, 0x0008,
+		0x0FF0, 0x0010,
+		0x0FE4, 0x001C,
+		0x0FD8, 0x0028,
+		0x0FC4, 0x003C,
+		0x0FB0, 0x0050,
+		0x0F98, 0x0068,
+		0x0F7C, 0x0084,
+		0x0F58, 0x00A8,
+		0x0F34, 0x00CC,
+		0x0F08, 0x00F8,
+		0x0ED8, 0x0128,
+		0x0EA4, 0x015C,
+		0x0E68, 0x0198,
+		0x0E28, 0x01D8,
+		0x0DE4, 0x021C,
+		0x0D98, 0x0268,
+		0x0D44, 0x02BC,
+		0x0CEC, 0x0314,
+		0x0C90, 0x0370,
+		0x0C2C, 0x03D4,
+		0x0BC4, 0x043C,
+		0x0B58, 0x04A8,
+		0x0AE8, 0x0518,
+		0x0A74, 0x058C,
+		0x09FC, 0x0604,
+		0x0980, 0x0680,
+		0x0900, 0x0700,
+		0x0880, 0x0780,
+		0x0800, 0x0800
+};
+
+//=========================================
+// <num_taps>    = 3
+// <num_phases>  = 64
+// <scale_ratio> = 0.83333 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_3tap_64p_upscale[99] = {
+		0x0804, 0x07FC, 0x0000,
+		0x07A8, 0x0860, 0x3FF8,
+		0x0754, 0x08BC, 0x3FF0,
+		0x0700, 0x0918, 0x3FE8,
+		0x06AC, 0x0978, 0x3FDC,
+		0x0654, 0x09D8, 0x3FD4,
+		0x0604, 0x0A34, 0x3FC8,
+		0x05B0, 0x0A90, 0x3FC0,
+		0x055C, 0x0AF0, 0x3FB4,
+		0x050C, 0x0B48, 0x3FAC,
+		0x04BC, 0x0BA0, 0x3FA4,
+		0x0470, 0x0BF4, 0x3F9C,
+		0x0420, 0x0C50, 0x3F90,
+		0x03D8, 0x0C9C, 0x3F8C,
+		0x038C, 0x0CF0, 0x3F84,
+		0x0344, 0x0D40, 0x3F7C,
+		0x0300, 0x0D88, 0x3F78,
+		0x02BC, 0x0DD0, 0x3F74,
+		0x027C, 0x0E14, 0x3F70,
+		0x023C, 0x0E54, 0x3F70,
+		0x0200, 0x0E90, 0x3F70,
+		0x01C8, 0x0EC8, 0x3F70,
+		0x0190, 0x0EFC, 0x3F74,
+		0x015C, 0x0F2C, 0x3F78,
+		0x0128, 0x0F5C, 0x3F7C,
+		0x00FC, 0x0F7C, 0x3F88,
+		0x00CC, 0x0FA4, 0x3F90,
+		0x00A4, 0x0FC0, 0x3F9C,
+		0x007C, 0x0FD8, 0x3FAC,
+		0x0058, 0x0FE8, 0x3FC0,
+		0x0038, 0x0FF4, 0x3FD4,
+		0x0018, 0x1000, 0x3FE8,
+		0x0000, 0x1000, 0x0000
+};
+
+//=========================================
+// <num_taps>    = 3
+// <num_phases>  = 64
+// <scale_ratio> = 1.16666 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_3tap_64p_116[99] = {
+		0x0804, 0x07FC, 0x0000,
+		0x07C0, 0x0844, 0x3FFC,
+		0x0780, 0x0888, 0x3FF8,
+		0x0740, 0x08D0, 0x3FF0,
+		0x0700, 0x0914, 0x3FEC,
+		0x06C0, 0x0958, 0x3FE8,
+		0x0684, 0x0998, 0x3FE4,
+		0x0644, 0x09DC, 0x3FE0,
+		0x0604, 0x0A1C, 0x3FE0,
+		0x05C4, 0x0A5C, 0x3FE0,
+		0x0588, 0x0A9C, 0x3FDC,
+		0x0548, 0x0ADC, 0x3FDC,
+		0x050C, 0x0B14, 0x3FE0,
+		0x04CC, 0x0B54, 0x3FE0,
+		0x0490, 0x0B8C, 0x3FE4,
+		0x0458, 0x0BC0, 0x3FE8,
+		0x041C, 0x0BF4, 0x3FF0,
+		0x03E0, 0x0C28, 0x3FF8,
+		0x03A8, 0x0C58, 0x0000,
+		0x0374, 0x0C88, 0x0004,
+		0x0340, 0x0CB0, 0x0010,
+		0x0308, 0x0CD8, 0x0020,
+		0x02D8, 0x0CFC, 0x002C,
+		0x02A0, 0x0D20, 0x0040,
+		0x0274, 0x0D3C, 0x0050,
+		0x0244, 0x0D58, 0x0064,
+		0x0214, 0x0D70, 0x007C,
+		0x01E8, 0x0D84, 0x0094,
+		0x01C0, 0x0D94, 0x00AC,
+		0x0198, 0x0DA0, 0x00C8,
+		0x0170, 0x0DAC, 0x00E4,
+		0x014C, 0x0DB0, 0x0104,
+		0x0128, 0x0DB4, 0x0124
+};
+
+//=========================================
+// <num_taps>    = 3
+// <num_phases>  = 64
+// <scale_ratio> = 1.49999 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_3tap_64p_149[99] = {
+		0x0804, 0x07FC, 0x0000,
+		0x07CC, 0x0834, 0x0000,
+		0x0798, 0x0868, 0x0000,
+		0x0764, 0x089C, 0x0000,
+		0x0730, 0x08CC, 0x0004,
+		0x0700, 0x08FC, 0x0004,
+		0x06CC, 0x092C, 0x0008,
+		0x0698, 0x095C, 0x000C,
+		0x0660, 0x098C, 0x0014,
+		0x062C, 0x09B8, 0x001C,
+		0x05FC, 0x09E4, 0x0020,
+		0x05C4, 0x0A10, 0x002C,
+		0x0590, 0x0A3C, 0x0034,
+		0x055C, 0x0A64, 0x0040,
+		0x0528, 0x0A8C, 0x004C,
+		0x04F8, 0x0AB0, 0x0058,
+		0x04C4, 0x0AD4, 0x0068,
+		0x0490, 0x0AF8, 0x0078,
+		0x0460, 0x0B18, 0x0088,
+		0x0430, 0x0B38, 0x0098,
+		0x0400, 0x0B54, 0x00AC,
+		0x03D0, 0x0B6C, 0x00C4,
+		0x03A0, 0x0B88, 0x00D8,
+		0x0374, 0x0B9C, 0x00F0,
+		0x0348, 0x0BB0, 0x0108,
+		0x0318, 0x0BC4, 0x0124,
+		0x02EC, 0x0BD4, 0x0140,
+		0x02C4, 0x0BE0, 0x015C,
+		0x029C, 0x0BEC, 0x0178,
+		0x0274, 0x0BF4, 0x0198,
+		0x024C, 0x0BFC, 0x01B8,
+		0x0228, 0x0BFC, 0x01DC,
+		0x0200, 0x0C00, 0x0200
+};
+
+//=========================================
+// <num_taps>    = 3
+// <num_phases>  = 64
+// <scale_ratio> = 1.83332 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_3tap_64p_183[99] = {
+		0x0804, 0x07FC, 0x0000,
+		0x07D4, 0x0824, 0x0008,
+		0x07AC, 0x0840, 0x0014,
+		0x0780, 0x0860, 0x0020,
+		0x0754, 0x0880, 0x002C,
+		0x0728, 0x089C, 0x003C,
+		0x0700, 0x08B8, 0x0048,
+		0x06D4, 0x08D4, 0x0058,
+		0x06A8, 0x08F0, 0x0068,
+		0x067C, 0x090C, 0x0078,
+		0x0650, 0x0924, 0x008C,
+		0x0628, 0x093C, 0x009C,
+		0x05FC, 0x0954, 0x00B0,
+		0x05D0, 0x096C, 0x00C4,
+		0x05A8, 0x0980, 0x00D8,
+		0x0578, 0x0998, 0x00F0,
+		0x0550, 0x09AC, 0x0104,
+		0x0528, 0x09BC, 0x011C,
+		0x04FC, 0x09D0, 0x0134,
+		0x04D4, 0x09E0, 0x014C,
+		0x04A8, 0x09F0, 0x0168,
+		0x0480, 0x09FC, 0x0184,
+		0x045C, 0x0A08, 0x019C,
+		0x0434, 0x0A14, 0x01B8,
+		0x0408, 0x0A20, 0x01D8,
+		0x03E0, 0x0A2C, 0x01F4,
+		0x03B8, 0x0A34, 0x0214,
+		0x0394, 0x0A38, 0x0234,
+		0x036C, 0x0A40, 0x0254,
+		0x0348, 0x0A44, 0x0274,
+		0x0324, 0x0A48, 0x0294,
+		0x0300, 0x0A48, 0x02B8,
+		0x02DC, 0x0A48, 0x02DC
+};
+
+//=========================================
+// <num_taps>    = 4
+// <num_phases>  = 64
+// <scale_ratio> = 0.83333 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_4tap_64p_upscale[132] = {
+		0x0000, 0x1000, 0x0000, 0x0000,
+		0x3FDC, 0x0FFC, 0x0028, 0x0000,
+		0x3FB4, 0x0FF8, 0x0054, 0x0000,
+		0x3F94, 0x0FE8, 0x0084, 0x0000,
+		0x3F74, 0x0FDC, 0x00B4, 0x3FFC,
+		0x3F58, 0x0FC4, 0x00E8, 0x3FFC,
+		0x3F3C, 0x0FAC, 0x0120, 0x3FF8,
+		0x3F24, 0x0F90, 0x0158, 0x3FF4,
+		0x3F0C, 0x0F70, 0x0194, 0x3FF0,
+		0x3EF8, 0x0F4C, 0x01D0, 0x3FEC,
+		0x3EE8, 0x0F20, 0x0210, 0x3FE8,
+		0x3ED8, 0x0EF4, 0x0254, 0x3FE0,
+		0x3ECC, 0x0EC4, 0x0298, 0x3FD8,
+		0x3EC0, 0x0E90, 0x02DC, 0x3FD4,
+		0x3EB8, 0x0E58, 0x0324, 0x3FCC,
+		0x3EB0, 0x0E20, 0x036C, 0x3FC4,
+		0x3EAC, 0x0DE4, 0x03B8, 0x3FB8,
+		0x3EA8, 0x0DA4, 0x0404, 0x3FB0,
+		0x3EA4, 0x0D60, 0x0454, 0x3FA8,
+		0x3EA4, 0x0D1C, 0x04A4, 0x3F9C,
+		0x3EA4, 0x0CD8, 0x04F4, 0x3F90,
+		0x3EA8, 0x0C88, 0x0548, 0x3F88,
+		0x3EAC, 0x0C3C, 0x059C, 0x3F7C,
+		0x3EB0, 0x0BF0, 0x05F0, 0x3F70,
+		0x3EB8, 0x0BA0, 0x0644, 0x3F64,
+		0x3EBC, 0x0B54, 0x0698, 0x3F58,
+		0x3EC4, 0x0B00, 0x06F0, 0x3F4C,
+		0x3ECC, 0x0AAC, 0x0748, 0x3F40,
+		0x3ED8, 0x0A54, 0x07A0, 0x3F34,
+		0x3EE0, 0x0A04, 0x07F8, 0x3F24,
+		0x3EEC, 0x09AC, 0x0850, 0x3F18,
+		0x3EF8, 0x0954, 0x08A8, 0x3F0C,
+		0x3F00, 0x08FC, 0x0900, 0x3F04
+};
+
+//=========================================
+// <num_taps>    = 4
+// <num_phases>  = 64
+// <scale_ratio> = 1.16666 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_4tap_64p_116[132] = {
+		0x01A8, 0x0CB4, 0x01A4, 0x0000,
+		0x017C, 0x0CB8, 0x01D0, 0x3FFC,
+		0x0158, 0x0CB8, 0x01F8, 0x3FF8,
+		0x0130, 0x0CB4, 0x0228, 0x3FF4,
+		0x0110, 0x0CB0, 0x0254, 0x3FEC,
+		0x00EC, 0x0CA8, 0x0284, 0x3FE8,
+		0x00CC, 0x0C9C, 0x02B4, 0x3FE4,
+		0x00AC, 0x0C90, 0x02E8, 0x3FDC,
+		0x0090, 0x0C80, 0x031C, 0x3FD4,
+		0x0070, 0x0C70, 0x0350, 0x3FD0,
+		0x0058, 0x0C5C, 0x0384, 0x3FC8,
+		0x003C, 0x0C48, 0x03BC, 0x3FC0,
+		0x0024, 0x0C2C, 0x03F4, 0x3FBC,
+		0x0010, 0x0C10, 0x042C, 0x3FB4,
+		0x3FFC, 0x0BF4, 0x0464, 0x3FAC,
+		0x3FE8, 0x0BD4, 0x04A0, 0x3FA4,
+		0x3FD8, 0x0BAC, 0x04DC, 0x3FA0,
+		0x3FC4, 0x0B8C, 0x0518, 0x3F98,
+		0x3FB4, 0x0B68, 0x0554, 0x3F90,
+		0x3FA8, 0x0B40, 0x0590, 0x3F88,
+		0x3F9C, 0x0B14, 0x05CC, 0x3F84,
+		0x3F90, 0x0AEC, 0x0608, 0x3F7C,
+		0x3F84, 0x0ABC, 0x0648, 0x3F78,
+		0x3F7C, 0x0A90, 0x0684, 0x3F70,
+		0x3F70, 0x0A60, 0x06C4, 0x3F6C,
+		0x3F6C, 0x0A2C, 0x0700, 0x3F68,
+		0x3F64, 0x09F8, 0x0740, 0x3F64,
+		0x3F60, 0x09C4, 0x077C, 0x3F60,
+		0x3F5C, 0x098C, 0x07BC, 0x3F5C,
+		0x3F58, 0x0958, 0x07F8, 0x3F58,
+		0x3F58, 0x091C, 0x0834, 0x3F58,
+		0x3F54, 0x08E4, 0x0870, 0x3F58,
+		0x3F54, 0x08AC, 0x08AC, 0x3F54
+};
+
+//=========================================
+// <num_taps>    = 4
+// <num_phases>  = 64
+// <scale_ratio> = 1.49999 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_4tap_64p_149[132] = {
+		0x02B8, 0x0A90, 0x02B8, 0x0000,
+		0x0294, 0x0A94, 0x02DC, 0x3FFC,
+		0x0274, 0x0A94, 0x0300, 0x3FF8,
+		0x0250, 0x0A94, 0x0328, 0x3FF4,
+		0x0230, 0x0A90, 0x0350, 0x3FF0,
+		0x0214, 0x0A8C, 0x0374, 0x3FEC,
+		0x01F0, 0x0A88, 0x03A0, 0x3FE8,
+		0x01D4, 0x0A80, 0x03C8, 0x3FE4,
+		0x01B8, 0x0A78, 0x03F0, 0x3FE0,
+		0x0198, 0x0A70, 0x041C, 0x3FDC,
+		0x0180, 0x0A64, 0x0444, 0x3FD8,
+		0x0164, 0x0A54, 0x0470, 0x3FD8,
+		0x0148, 0x0A48, 0x049C, 0x3FD4,
+		0x0130, 0x0A38, 0x04C8, 0x3FD0,
+		0x0118, 0x0A24, 0x04F4, 0x3FD0,
+		0x0100, 0x0A14, 0x0520, 0x3FCC,
+		0x00E8, 0x0A00, 0x054C, 0x3FCC,
+		0x00D4, 0x09E8, 0x057C, 0x3FC8,
+		0x00C0, 0x09D0, 0x05A8, 0x3FC8,
+		0x00AC, 0x09B8, 0x05D4, 0x3FC8,
+		0x0098, 0x09A0, 0x0600, 0x3FC8,
+		0x0084, 0x0984, 0x0630, 0x3FC8,
+		0x0074, 0x0964, 0x065C, 0x3FCC,
+		0x0064, 0x0948, 0x0688, 0x3FCC,
+		0x0054, 0x0928, 0x06B4, 0x3FD0,
+		0x0044, 0x0908, 0x06E0, 0x3FD4,
+		0x0038, 0x08E8, 0x070C, 0x3FD4,
+		0x002C, 0x08C4, 0x0738, 0x3FD8,
+		0x001C, 0x08A4, 0x0760, 0x3FE0,
+		0x0014, 0x087C, 0x078C, 0x3FE4,
+		0x0008, 0x0858, 0x07B4, 0x3FEC,
+		0x0000, 0x0830, 0x07DC, 0x3FF4,
+		0x3FFC, 0x0804, 0x0804, 0x3FFC
+};
+
+//=========================================
+// <num_taps>    = 4
+// <num_phases>  = 64
+// <scale_ratio> = 1.83332 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_4tap_64p_183[132] = {
+		0x03B0, 0x08A0, 0x03B0, 0x0000,
+		0x0394, 0x08A0, 0x03CC, 0x0000,
+		0x037C, 0x089C, 0x03E8, 0x0000,
+		0x0360, 0x089C, 0x0400, 0x0004,
+		0x0348, 0x0898, 0x041C, 0x0004,
+		0x032C, 0x0894, 0x0438, 0x0008,
+		0x0310, 0x0890, 0x0454, 0x000C,
+		0x02F8, 0x0888, 0x0474, 0x000C,
+		0x02DC, 0x0884, 0x0490, 0x0010,
+		0x02C4, 0x087C, 0x04AC, 0x0014,
+		0x02AC, 0x0874, 0x04C8, 0x0018,
+		0x0290, 0x086C, 0x04E4, 0x0020,
+		0x0278, 0x0864, 0x0500, 0x0024,
+		0x0264, 0x0858, 0x051C, 0x0028,
+		0x024C, 0x084C, 0x0538, 0x0030,
+		0x0234, 0x0844, 0x0554, 0x0034,
+		0x021C, 0x0838, 0x0570, 0x003C,
+		0x0208, 0x0828, 0x058C, 0x0044,
+		0x01F0, 0x081C, 0x05A8, 0x004C,
+		0x01DC, 0x080C, 0x05C4, 0x0054,
+		0x01C8, 0x07FC, 0x05E0, 0x005C,
+		0x01B4, 0x07EC, 0x05FC, 0x0064,
+		0x019C, 0x07DC, 0x0618, 0x0070,
+		0x018C, 0x07CC, 0x0630, 0x0078,
+		0x0178, 0x07B8, 0x064C, 0x0084,
+		0x0164, 0x07A8, 0x0664, 0x0090,
+		0x0150, 0x0794, 0x0680, 0x009C,
+		0x0140, 0x0780, 0x0698, 0x00A8,
+		0x0130, 0x076C, 0x06B0, 0x00B4,
+		0x0120, 0x0758, 0x06C8, 0x00C0,
+		0x0110, 0x0740, 0x06E0, 0x00D0,
+		0x0100, 0x072C, 0x06F8, 0x00DC,
+		0x00F0, 0x0714, 0x0710, 0x00EC
+};
+
+//=========================================
+// <num_taps>    = 5
+// <num_phases>  = 64
+// <scale_ratio> = 0.83333 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_5tap_64p_upscale[165] = {
+		0x3E40, 0x09C0, 0x09C0, 0x3E40, 0x0000,
+		0x3E50, 0x0964, 0x0A18, 0x3E34, 0x0000,
+		0x3E5C, 0x0908, 0x0A6C, 0x3E2C, 0x0004,
+		0x3E6C, 0x08AC, 0x0AC0, 0x3E20, 0x0008,
+		0x3E78, 0x0850, 0x0B14, 0x3E18, 0x000C,
+		0x3E88, 0x07F4, 0x0B60, 0x3E14, 0x0010,
+		0x3E98, 0x0798, 0x0BB0, 0x3E0C, 0x0014,
+		0x3EA8, 0x073C, 0x0C00, 0x3E08, 0x0014,
+		0x3EB8, 0x06E4, 0x0C48, 0x3E04, 0x0018,
+		0x3ECC, 0x0684, 0x0C90, 0x3E04, 0x001C,
+		0x3EDC, 0x062C, 0x0CD4, 0x3E04, 0x0020,
+		0x3EEC, 0x05D4, 0x0D1C, 0x3E04, 0x0020,
+		0x3EFC, 0x057C, 0x0D5C, 0x3E08, 0x0024,
+		0x3F0C, 0x0524, 0x0D98, 0x3E10, 0x0028,
+		0x3F20, 0x04CC, 0x0DD8, 0x3E14, 0x0028,
+		0x3F30, 0x0478, 0x0E14, 0x3E1C, 0x0028,
+		0x3F40, 0x0424, 0x0E48, 0x3E28, 0x002C,
+		0x3F50, 0x03D4, 0x0E7C, 0x3E34, 0x002C,
+		0x3F60, 0x0384, 0x0EAC, 0x3E44, 0x002C,
+		0x3F6C, 0x0338, 0x0EDC, 0x3E54, 0x002C,
+		0x3F7C, 0x02E8, 0x0F08, 0x3E68, 0x002C,
+		0x3F8C, 0x02A0, 0x0F2C, 0x3E7C, 0x002C,
+		0x3F98, 0x0258, 0x0F50, 0x3E94, 0x002C,
+		0x3FA4, 0x0210, 0x0F74, 0x3EB0, 0x0028,
+		0x3FB0, 0x01CC, 0x0F90, 0x3ECC, 0x0028,
+		0x3FC0, 0x018C, 0x0FA8, 0x3EE8, 0x0024,
+		0x3FC8, 0x014C, 0x0FC0, 0x3F0C, 0x0020,
+		0x3FD4, 0x0110, 0x0FD4, 0x3F2C, 0x001C,
+		0x3FE0, 0x00D4, 0x0FE0, 0x3F54, 0x0018,
+		0x3FE8, 0x009C, 0x0FF0, 0x3F7C, 0x0010,
+		0x3FF0, 0x0064, 0x0FFC, 0x3FA4, 0x000C,
+		0x3FFC, 0x0030, 0x0FFC, 0x3FD4, 0x0004,
+		0x0000, 0x0000, 0x1000, 0x0000, 0x0000
+};
+
+//=========================================
+// <num_taps>    = 5
+// <num_phases>  = 64
+// <scale_ratio> = 1.16666 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_5tap_64p_116[165] = {
+		0x3EDC, 0x0924, 0x0924, 0x3EDC, 0x0000,
+		0x3ED8, 0x08EC, 0x095C, 0x3EE0, 0x0000,
+		0x3ED4, 0x08B0, 0x0994, 0x3EE8, 0x0000,
+		0x3ED0, 0x0878, 0x09C8, 0x3EF0, 0x0000,
+		0x3ED0, 0x083C, 0x09FC, 0x3EF8, 0x0000,
+		0x3ED0, 0x0800, 0x0A2C, 0x3F04, 0x0000,
+		0x3ED0, 0x07C4, 0x0A5C, 0x3F10, 0x0000,
+		0x3ED0, 0x0788, 0x0A8C, 0x3F1C, 0x0000,
+		0x3ED0, 0x074C, 0x0AC0, 0x3F28, 0x3FFC,
+		0x3ED4, 0x0710, 0x0AE8, 0x3F38, 0x3FFC,
+		0x3ED8, 0x06D0, 0x0B18, 0x3F48, 0x3FF8,
+		0x3EDC, 0x0694, 0x0B3C, 0x3F5C, 0x3FF8,
+		0x3EE0, 0x0658, 0x0B68, 0x3F6C, 0x3FF4,
+		0x3EE4, 0x061C, 0x0B90, 0x3F80, 0x3FF0,
+		0x3EEC, 0x05DC, 0x0BB4, 0x3F98, 0x3FEC,
+		0x3EF0, 0x05A0, 0x0BD8, 0x3FB0, 0x3FE8,
+		0x3EF8, 0x0564, 0x0BF8, 0x3FC8, 0x3FE4,
+		0x3EFC, 0x0528, 0x0C1C, 0x3FE0, 0x3FE0,
+		0x3F04, 0x04EC, 0x0C38, 0x3FFC, 0x3FDC,
+		0x3F0C, 0x04B4, 0x0C54, 0x0014, 0x3FD8,
+		0x3F14, 0x047C, 0x0C70, 0x0030, 0x3FD0,
+		0x3F1C, 0x0440, 0x0C88, 0x0050, 0x3FCC,
+		0x3F24, 0x0408, 0x0CA0, 0x0070, 0x3FC4,
+		0x3F2C, 0x03D0, 0x0CB0, 0x0094, 0x3FC0,
+		0x3F34, 0x0398, 0x0CC4, 0x00B8, 0x3FB8,
+		0x3F3C, 0x0364, 0x0CD4, 0x00DC, 0x3FB0,
+		0x3F48, 0x032C, 0x0CE0, 0x0100, 0x3FAC,
+		0x3F50, 0x02F8, 0x0CEC, 0x0128, 0x3FA4,
+		0x3F58, 0x02C4, 0x0CF8, 0x0150, 0x3F9C,
+		0x3F60, 0x0290, 0x0D00, 0x017C, 0x3F94,
+		0x3F68, 0x0260, 0x0D04, 0x01A8, 0x3F8C,
+		0x3F74, 0x0230, 0x0D04, 0x01D4, 0x3F84,
+		0x3F7C, 0x0200, 0x0D08, 0x0200, 0x3F7C
+};
+
+//=========================================
+// <num_taps>    = 5
+// <num_phases>  = 64
+// <scale_ratio> = 1.49999 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_5tap_64p_149[165] = {
+		0x3FF4, 0x080C, 0x080C, 0x3FF4, 0x0000,
+		0x3FE8, 0x07E8, 0x0830, 0x0000, 0x0000,
+		0x3FDC, 0x07C8, 0x0850, 0x0010, 0x3FFC,
+		0x3FD0, 0x07A4, 0x0878, 0x001C, 0x3FF8,
+		0x3FC4, 0x0780, 0x0898, 0x0030, 0x3FF4,
+		0x3FB8, 0x075C, 0x08B8, 0x0040, 0x3FF4,
+		0x3FB0, 0x0738, 0x08D8, 0x0050, 0x3FF0,
+		0x3FA8, 0x0710, 0x08F8, 0x0064, 0x3FEC,
+		0x3FA0, 0x06EC, 0x0914, 0x0078, 0x3FE8,
+		0x3F98, 0x06C4, 0x0934, 0x008C, 0x3FE4,
+		0x3F90, 0x06A0, 0x094C, 0x00A4, 0x3FE0,
+		0x3F8C, 0x0678, 0x0968, 0x00B8, 0x3FDC,
+		0x3F84, 0x0650, 0x0984, 0x00D0, 0x3FD8,
+		0x3F80, 0x0628, 0x099C, 0x00E8, 0x3FD4,
+		0x3F7C, 0x0600, 0x09B8, 0x0100, 0x3FCC,
+		0x3F78, 0x05D8, 0x09D0, 0x0118, 0x3FC8,
+		0x3F74, 0x05B0, 0x09E4, 0x0134, 0x3FC4,
+		0x3F70, 0x0588, 0x09F8, 0x0150, 0x3FC0,
+		0x3F70, 0x0560, 0x0A08, 0x016C, 0x3FBC,
+		0x3F6C, 0x0538, 0x0A20, 0x0188, 0x3FB4,
+		0x3F6C, 0x0510, 0x0A30, 0x01A4, 0x3FB0,
+		0x3F6C, 0x04E8, 0x0A3C, 0x01C4, 0x3FAC,
+		0x3F6C, 0x04C0, 0x0A48, 0x01E4, 0x3FA8,
+		0x3F6C, 0x0498, 0x0A58, 0x0200, 0x3FA4,
+		0x3F6C, 0x0470, 0x0A60, 0x0224, 0x3FA0,
+		0x3F6C, 0x0448, 0x0A70, 0x0244, 0x3F98,
+		0x3F70, 0x0420, 0x0A78, 0x0264, 0x3F94,
+		0x3F70, 0x03F8, 0x0A80, 0x0288, 0x3F90,
+		0x3F74, 0x03D4, 0x0A84, 0x02A8, 0x3F8C,
+		0x3F74, 0x03AC, 0x0A8C, 0x02CC, 0x3F88,
+		0x3F78, 0x0384, 0x0A90, 0x02F0, 0x3F84,
+		0x3F7C, 0x0360, 0x0A90, 0x0314, 0x3F80,
+		0x3F7C, 0x033C, 0x0A90, 0x033C, 0x3F7C
+};
+
+//=========================================
+// <num_taps>    = 5
+// <num_phases>  = 64
+// <scale_ratio> = 1.83332 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_5tap_64p_183[165] = {
+		0x0168, 0x069C, 0x0698, 0x0164, 0x0000,
+		0x0154, 0x068C, 0x06AC, 0x0174, 0x0000,
+		0x0144, 0x0674, 0x06C0, 0x0188, 0x0000,
+		0x0138, 0x0664, 0x06D0, 0x0198, 0x3FFC,
+		0x0128, 0x0654, 0x06E0, 0x01A8, 0x3FFC,
+		0x0118, 0x0640, 0x06F0, 0x01BC, 0x3FFC,
+		0x010C, 0x0630, 0x0700, 0x01CC, 0x3FF8,
+		0x00FC, 0x061C, 0x0710, 0x01E0, 0x3FF8,
+		0x00F0, 0x060C, 0x071C, 0x01F0, 0x3FF8,
+		0x00E4, 0x05F4, 0x072C, 0x0204, 0x3FF8,
+		0x00D8, 0x05E4, 0x0738, 0x0218, 0x3FF4,
+		0x00CC, 0x05D0, 0x0744, 0x022C, 0x3FF4,
+		0x00C0, 0x05B8, 0x0754, 0x0240, 0x3FF4,
+		0x00B4, 0x05A4, 0x0760, 0x0254, 0x3FF4,
+		0x00A8, 0x0590, 0x076C, 0x0268, 0x3FF4,
+		0x009C, 0x057C, 0x0778, 0x027C, 0x3FF4,
+		0x0094, 0x0564, 0x0780, 0x0294, 0x3FF4,
+		0x0088, 0x0550, 0x0788, 0x02A8, 0x3FF8,
+		0x0080, 0x0538, 0x0794, 0x02BC, 0x3FF8,
+		0x0074, 0x0524, 0x079C, 0x02D4, 0x3FF8,
+		0x006C, 0x0510, 0x07A4, 0x02E8, 0x3FF8,
+		0x0064, 0x04F4, 0x07AC, 0x0300, 0x3FFC,
+		0x005C, 0x04E4, 0x07B0, 0x0314, 0x3FFC,
+		0x0054, 0x04C8, 0x07B8, 0x032C, 0x0000,
+		0x004C, 0x04B4, 0x07C0, 0x0340, 0x0000,
+		0x0044, 0x04A0, 0x07C4, 0x0358, 0x0000,
+		0x003C, 0x0488, 0x07C8, 0x0370, 0x0004,
+		0x0038, 0x0470, 0x07CC, 0x0384, 0x0008,
+		0x0030, 0x045C, 0x07D0, 0x039C, 0x0008,
+		0x002C, 0x0444, 0x07D0, 0x03B4, 0x000C,
+		0x0024, 0x042C, 0x07D4, 0x03CC, 0x0010,
+		0x0020, 0x0414, 0x07D4, 0x03E0, 0x0018,
+		0x001C, 0x03FC, 0x07D4, 0x03F8, 0x001C
+};
+
+//=========================================
+// <num_taps>    = 6
+// <num_phases>  = 64
+// <scale_ratio> = 0.83333 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_6tap_64p_upscale[198] = {
+		0x0000, 0x0000, 0x1000, 0x0000, 0x0000, 0x0000,
+		0x000C, 0x3FD0, 0x0FFC, 0x0034, 0x3FF4, 0x0000,
+		0x0018, 0x3F9C, 0x0FF8, 0x006C, 0x3FE8, 0x0000,
+		0x0024, 0x3F6C, 0x0FF0, 0x00A8, 0x3FD8, 0x0000,
+		0x002C, 0x3F44, 0x0FE4, 0x00E4, 0x3FC8, 0x0000,
+		0x0038, 0x3F18, 0x0FD4, 0x0124, 0x3FB8, 0x0000,
+		0x0040, 0x3EF0, 0x0FC0, 0x0164, 0x3FA8, 0x0004,
+		0x0048, 0x3EC8, 0x0FAC, 0x01A8, 0x3F98, 0x0004,
+		0x0050, 0x3EA8, 0x0F94, 0x01EC, 0x3F84, 0x0004,
+		0x0058, 0x3E84, 0x0F74, 0x0234, 0x3F74, 0x0008,
+		0x0060, 0x3E68, 0x0F54, 0x027C, 0x3F60, 0x0008,
+		0x0064, 0x3E4C, 0x0F30, 0x02C8, 0x3F4C, 0x000C,
+		0x006C, 0x3E30, 0x0F04, 0x0314, 0x3F3C, 0x0010,
+		0x0070, 0x3E18, 0x0EDC, 0x0360, 0x3F28, 0x0014,
+		0x0074, 0x3E04, 0x0EB0, 0x03B0, 0x3F14, 0x0014,
+		0x0078, 0x3DF0, 0x0E80, 0x0400, 0x3F00, 0x0018,
+		0x0078, 0x3DE0, 0x0E4C, 0x0454, 0x3EEC, 0x001C,
+		0x007C, 0x3DD0, 0x0E14, 0x04A8, 0x3ED8, 0x0020,
+		0x007C, 0x3DC4, 0x0DDC, 0x04FC, 0x3EC4, 0x0024,
+		0x007C, 0x3DBC, 0x0DA0, 0x0550, 0x3EB0, 0x0028,
+		0x0080, 0x3DB4, 0x0D5C, 0x05A8, 0x3E9C, 0x002C,
+		0x0080, 0x3DAC, 0x0D1C, 0x0600, 0x3E88, 0x0030,
+		0x007C, 0x3DA8, 0x0CDC, 0x0658, 0x3E74, 0x0034,
+		0x007C, 0x3DA4, 0x0C94, 0x06B0, 0x3E64, 0x0038,
+		0x007C, 0x3DA4, 0x0C48, 0x0708, 0x3E50, 0x0040,
+		0x0078, 0x3DA4, 0x0C00, 0x0760, 0x3E40, 0x0044,
+		0x0078, 0x3DA8, 0x0BB4, 0x07B8, 0x3E2C, 0x0048,
+		0x0074, 0x3DAC, 0x0B68, 0x0810, 0x3E1C, 0x004C,
+		0x0070, 0x3DB4, 0x0B18, 0x0868, 0x3E0C, 0x0050,
+		0x006C, 0x3DBC, 0x0AC4, 0x08C4, 0x3DFC, 0x0054,
+		0x0068, 0x3DC4, 0x0A74, 0x0918, 0x3DF0, 0x0058,
+		0x0068, 0x3DCC, 0x0A20, 0x0970, 0x3DE0, 0x005C,
+		0x0064, 0x3DD4, 0x09C8, 0x09C8, 0x3DD4, 0x0064
+};
+
+//=========================================
+// <num_taps>    = 6
+// <num_phases>  = 64
+// <scale_ratio> = 1.16666 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_6tap_64p_116[198] = {
+		0x3F0C, 0x0240, 0x0D68, 0x0240, 0x3F0C, 0x0000,
+		0x3F18, 0x0210, 0x0D64, 0x0274, 0x3F00, 0x0000,
+		0x3F24, 0x01E0, 0x0D58, 0x02A8, 0x3EF8, 0x0004,
+		0x3F2C, 0x01B0, 0x0D58, 0x02DC, 0x3EEC, 0x0004,
+		0x3F38, 0x0180, 0x0D50, 0x0310, 0x3EE0, 0x0008,
+		0x3F44, 0x0154, 0x0D40, 0x0348, 0x3ED8, 0x0008,
+		0x3F50, 0x0128, 0x0D34, 0x037C, 0x3ECC, 0x000C,
+		0x3F5C, 0x00FC, 0x0D20, 0x03B4, 0x3EC4, 0x0010,
+		0x3F64, 0x00D4, 0x0D14, 0x03EC, 0x3EB8, 0x0010,
+		0x3F70, 0x00AC, 0x0CFC, 0x0424, 0x3EB0, 0x0014,
+		0x3F78, 0x0084, 0x0CE8, 0x0460, 0x3EA8, 0x0014,
+		0x3F84, 0x0060, 0x0CCC, 0x0498, 0x3EA0, 0x0018,
+		0x3F90, 0x003C, 0x0CB4, 0x04D0, 0x3E98, 0x0018,
+		0x3F98, 0x0018, 0x0C9C, 0x050C, 0x3E90, 0x0018,
+		0x3FA0, 0x3FFC, 0x0C78, 0x0548, 0x3E88, 0x001C,
+		0x3FAC, 0x3FDC, 0x0C54, 0x0584, 0x3E84, 0x001C,
+		0x3FB4, 0x3FBC, 0x0C3C, 0x05BC, 0x3E7C, 0x001C,
+		0x3FBC, 0x3FA0, 0x0C14, 0x05F8, 0x3E78, 0x0020,
+		0x3FC4, 0x3F84, 0x0BF0, 0x0634, 0x3E74, 0x0020,
+		0x3FCC, 0x3F68, 0x0BCC, 0x0670, 0x3E70, 0x0020,
+		0x3FD4, 0x3F50, 0x0BA4, 0x06AC, 0x3E6C, 0x0020,
+		0x3FDC, 0x3F38, 0x0B78, 0x06E8, 0x3E6C, 0x0020,
+		0x3FE0, 0x3F24, 0x0B50, 0x0724, 0x3E68, 0x0020,
+		0x3FE8, 0x3F0C, 0x0B24, 0x0760, 0x3E68, 0x0020,
+		0x3FF0, 0x3EFC, 0x0AF4, 0x0798, 0x3E68, 0x0020,
+		0x3FF4, 0x3EE8, 0x0AC8, 0x07D4, 0x3E68, 0x0020,
+		0x3FFC, 0x3ED8, 0x0A94, 0x0810, 0x3E6C, 0x001C,
+		0x0000, 0x3EC8, 0x0A64, 0x0848, 0x3E70, 0x001C,
+		0x0000, 0x3EB8, 0x0A38, 0x0880, 0x3E74, 0x001C,
+		0x0004, 0x3EAC, 0x0A04, 0x08BC, 0x3E78, 0x0018,
+		0x0008, 0x3EA4, 0x09D0, 0x08F4, 0x3E7C, 0x0014,
+		0x000C, 0x3E98, 0x0998, 0x092C, 0x3E84, 0x0014,
+		0x0010, 0x3E90, 0x0964, 0x0960, 0x3E8C, 0x0010
+};
+
+//=========================================
+// <num_taps>    = 6
+// <num_phases>  = 64
+// <scale_ratio> = 1.49999 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_6tap_64p_149[198] = {
+		0x3F14, 0x0394, 0x0AB0, 0x0394, 0x3F14, 0x0000,
+		0x3F18, 0x036C, 0x0AB0, 0x03B8, 0x3F14, 0x0000,
+		0x3F18, 0x0348, 0x0AAC, 0x03E0, 0x3F14, 0x0000,
+		0x3F1C, 0x0320, 0x0AAC, 0x0408, 0x3F10, 0x0000,
+		0x3F20, 0x02FC, 0x0AA8, 0x042C, 0x3F10, 0x0000,
+		0x3F24, 0x02D8, 0x0AA0, 0x0454, 0x3F10, 0x0000,
+		0x3F28, 0x02B4, 0x0A98, 0x047C, 0x3F10, 0x0000,
+		0x3F28, 0x0290, 0x0A90, 0x04A4, 0x3F14, 0x0000,
+		0x3F30, 0x026C, 0x0A84, 0x04CC, 0x3F14, 0x0000,
+		0x3F34, 0x024C, 0x0A7C, 0x04F4, 0x3F14, 0x3FFC,
+		0x3F38, 0x0228, 0x0A70, 0x051C, 0x3F18, 0x3FFC,
+		0x3F3C, 0x0208, 0x0A64, 0x0544, 0x3F1C, 0x3FF8,
+		0x3F40, 0x01E8, 0x0A54, 0x056C, 0x3F20, 0x3FF8,
+		0x3F44, 0x01C8, 0x0A48, 0x0594, 0x3F24, 0x3FF4,
+		0x3F4C, 0x01A8, 0x0A34, 0x05BC, 0x3F28, 0x3FF4,
+		0x3F50, 0x0188, 0x0A28, 0x05E4, 0x3F2C, 0x3FF0,
+		0x3F54, 0x016C, 0x0A10, 0x060C, 0x3F34, 0x3FF0,
+		0x3F5C, 0x014C, 0x09FC, 0x0634, 0x3F3C, 0x3FEC,
+		0x3F60, 0x0130, 0x09EC, 0x065C, 0x3F40, 0x3FE8,
+		0x3F68, 0x0114, 0x09D0, 0x0684, 0x3F48, 0x3FE8,
+		0x3F6C, 0x00F8, 0x09B8, 0x06AC, 0x3F54, 0x3FE4,
+		0x3F74, 0x00E0, 0x09A0, 0x06D0, 0x3F5C, 0x3FE0,
+		0x3F78, 0x00C4, 0x098C, 0x06F8, 0x3F64, 0x3FDC,
+		0x3F7C, 0x00AC, 0x0970, 0x0720, 0x3F70, 0x3FD8,
+		0x3F84, 0x0094, 0x0954, 0x0744, 0x3F7C, 0x3FD4,
+		0x3F88, 0x007C, 0x093C, 0x0768, 0x3F88, 0x3FD0,
+		0x3F90, 0x0064, 0x091C, 0x0790, 0x3F94, 0x3FCC,
+		0x3F94, 0x0050, 0x08FC, 0x07B4, 0x3FA4, 0x3FC8,
+		0x3F98, 0x003C, 0x08E0, 0x07D8, 0x3FB0, 0x3FC4,
+		0x3FA0, 0x0024, 0x08C0, 0x07FC, 0x3FC0, 0x3FC0,
+		0x3FA4, 0x0014, 0x08A4, 0x081C, 0x3FD0, 0x3FB8,
+		0x3FAC, 0x0000, 0x0880, 0x0840, 0x3FE0, 0x3FB4,
+		0x3FB0, 0x3FF0, 0x0860, 0x0860, 0x3FF0, 0x3FB0
+};
+
+//=========================================
+// <num_taps>    = 6
+// <num_phases>  = 64
+// <scale_ratio> = 1.83332 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_6tap_64p_183[198] = {
+		0x002C, 0x0420, 0x076C, 0x041C, 0x002C, 0x0000,
+		0x0028, 0x040C, 0x0768, 0x0430, 0x0034, 0x0000,
+		0x0020, 0x03F8, 0x0768, 0x0448, 0x003C, 0x3FFC,
+		0x0018, 0x03E4, 0x0768, 0x045C, 0x0044, 0x3FFC,
+		0x0014, 0x03D0, 0x0768, 0x0470, 0x004C, 0x3FF8,
+		0x000C, 0x03BC, 0x0764, 0x0484, 0x0058, 0x3FF8,
+		0x0008, 0x03A4, 0x0764, 0x049C, 0x0060, 0x3FF4,
+		0x0004, 0x0390, 0x0760, 0x04B0, 0x0068, 0x3FF4,
+		0x0000, 0x037C, 0x0760, 0x04C4, 0x0070, 0x3FF0,
+		0x3FFC, 0x0364, 0x075C, 0x04D8, 0x007C, 0x3FF0,
+		0x3FF8, 0x0350, 0x0758, 0x04F0, 0x0084, 0x3FEC,
+		0x3FF4, 0x033C, 0x0750, 0x0504, 0x0090, 0x3FEC,
+		0x3FF0, 0x0328, 0x074C, 0x0518, 0x009C, 0x3FE8,
+		0x3FEC, 0x0314, 0x0744, 0x052C, 0x00A8, 0x3FE8,
+		0x3FE8, 0x0304, 0x0740, 0x0540, 0x00B0, 0x3FE4,
+		0x3FE4, 0x02EC, 0x073C, 0x0554, 0x00BC, 0x3FE4,
+		0x3FE0, 0x02DC, 0x0734, 0x0568, 0x00C8, 0x3FE0,
+		0x3FE0, 0x02C4, 0x072C, 0x057C, 0x00D4, 0x3FE0,
+		0x3FDC, 0x02B4, 0x0724, 0x058C, 0x00E4, 0x3FDC,
+		0x3FDC, 0x02A0, 0x0718, 0x05A0, 0x00F0, 0x3FDC,
+		0x3FD8, 0x028C, 0x0714, 0x05B4, 0x00FC, 0x3FD8,
+		0x3FD8, 0x0278, 0x0704, 0x05C8, 0x010C, 0x3FD8,
+		0x3FD4, 0x0264, 0x0700, 0x05D8, 0x0118, 0x3FD8,
+		0x3FD4, 0x0254, 0x06F0, 0x05EC, 0x0128, 0x3FD4,
+		0x3FD0, 0x0244, 0x06E8, 0x05FC, 0x0134, 0x3FD4,
+		0x3FD0, 0x0230, 0x06DC, 0x060C, 0x0144, 0x3FD4,
+		0x3FD0, 0x021C, 0x06D0, 0x0620, 0x0154, 0x3FD0,
+		0x3FD0, 0x0208, 0x06C4, 0x0630, 0x0164, 0x3FD0,
+		0x3FD0, 0x01F8, 0x06B8, 0x0640, 0x0170, 0x3FD0,
+		0x3FCC, 0x01E8, 0x06AC, 0x0650, 0x0180, 0x3FD0,
+		0x3FCC, 0x01D8, 0x069C, 0x0660, 0x0190, 0x3FD0,
+		0x3FCC, 0x01C4, 0x068C, 0x0670, 0x01A4, 0x3FD0,
+		0x3FCC, 0x01B8, 0x0680, 0x067C, 0x01B4, 0x3FCC
+};
+
+//=========================================
+// <num_taps>    = 7
+// <num_phases>  = 64
+// <scale_ratio> = 0.83333 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_7tap_64p_upscale[231] = {
+		0x00B0, 0x3D98, 0x09BC, 0x09B8, 0x3D94, 0x00B0, 0x0000,
+		0x00AC, 0x3DA0, 0x0968, 0x0A10, 0x3D88, 0x00B4, 0x0000,
+		0x00A8, 0x3DAC, 0x0914, 0x0A60, 0x3D80, 0x00B8, 0x0000,
+		0x00A4, 0x3DB8, 0x08C0, 0x0AB4, 0x3D78, 0x00BC, 0x3FFC,
+		0x00A0, 0x3DC8, 0x0868, 0x0B00, 0x3D74, 0x00C0, 0x3FFC,
+		0x0098, 0x3DD8, 0x0818, 0x0B54, 0x3D6C, 0x00C0, 0x3FF8,
+		0x0094, 0x3DE8, 0x07C0, 0x0B9C, 0x3D6C, 0x00C4, 0x3FF8,
+		0x008C, 0x3DFC, 0x0768, 0x0BEC, 0x3D68, 0x00C4, 0x3FF8,
+		0x0088, 0x3E0C, 0x0714, 0x0C38, 0x3D68, 0x00C4, 0x3FF4,
+		0x0080, 0x3E20, 0x06BC, 0x0C80, 0x3D6C, 0x00C4, 0x3FF4,
+		0x0078, 0x3E34, 0x0668, 0x0CC4, 0x3D70, 0x00C4, 0x3FF4,
+		0x0074, 0x3E48, 0x0610, 0x0D08, 0x3D78, 0x00C4, 0x3FF0,
+		0x006C, 0x3E5C, 0x05BC, 0x0D48, 0x3D80, 0x00C4, 0x3FF0,
+		0x0068, 0x3E74, 0x0568, 0x0D84, 0x3D88, 0x00C0, 0x3FF0,
+		0x0060, 0x3E88, 0x0514, 0x0DC8, 0x3D94, 0x00BC, 0x3FEC,
+		0x0058, 0x3E9C, 0x04C0, 0x0E04, 0x3DA4, 0x00B8, 0x3FEC,
+		0x0054, 0x3EB4, 0x046C, 0x0E38, 0x3DB4, 0x00B4, 0x3FEC,
+		0x004C, 0x3ECC, 0x0418, 0x0E6C, 0x3DC8, 0x00B0, 0x3FEC,
+		0x0044, 0x3EE0, 0x03C8, 0x0EA4, 0x3DDC, 0x00A8, 0x3FEC,
+		0x0040, 0x3EF8, 0x0378, 0x0ED0, 0x3DF4, 0x00A0, 0x3FEC,
+		0x0038, 0x3F0C, 0x032C, 0x0EFC, 0x3E10, 0x0098, 0x3FEC,
+		0x0034, 0x3F24, 0x02DC, 0x0F24, 0x3E2C, 0x0090, 0x3FEC,
+		0x002C, 0x3F38, 0x0294, 0x0F4C, 0x3E48, 0x0088, 0x3FEC,
+		0x0028, 0x3F50, 0x0248, 0x0F68, 0x3E6C, 0x007C, 0x3FF0,
+		0x0020, 0x3F64, 0x0200, 0x0F88, 0x3E90, 0x0074, 0x3FF0,
+		0x001C, 0x3F7C, 0x01B8, 0x0FA4, 0x3EB4, 0x0068, 0x3FF0,
+		0x0018, 0x3F90, 0x0174, 0x0FBC, 0x3EDC, 0x0058, 0x3FF4,
+		0x0014, 0x3FA4, 0x0130, 0x0FD0, 0x3F08, 0x004C, 0x3FF4,
+		0x000C, 0x3FB8, 0x00F0, 0x0FE4, 0x3F34, 0x003C, 0x3FF8,
+		0x0008, 0x3FCC, 0x00B0, 0x0FF0, 0x3F64, 0x0030, 0x3FF8,
+		0x0004, 0x3FDC, 0x0070, 0x0FFC, 0x3F98, 0x0020, 0x3FFC,
+		0x0000, 0x3FF0, 0x0038, 0x0FFC, 0x3FCC, 0x0010, 0x0000,
+		0x0000, 0x0000, 0x0000, 0x1000, 0x0000, 0x0000, 0x0000
+};
+
+//=========================================
+// <num_taps>    = 7
+// <num_phases>  = 64
+// <scale_ratio> = 1.16666 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_7tap_64p_116[231] = {
+		0x0020, 0x3E58, 0x0988, 0x0988, 0x3E58, 0x0020, 0x0000,
+		0x0024, 0x3E4C, 0x0954, 0x09C0, 0x3E64, 0x0018, 0x0000,
+		0x002C, 0x3E44, 0x091C, 0x09F4, 0x3E70, 0x0010, 0x0000,
+		0x0030, 0x3E3C, 0x08E8, 0x0A24, 0x3E80, 0x0008, 0x0000,
+		0x0034, 0x3E34, 0x08AC, 0x0A5C, 0x3E90, 0x0000, 0x0000,
+		0x003C, 0x3E30, 0x0870, 0x0A84, 0x3EA0, 0x3FFC, 0x0004,
+		0x0040, 0x3E28, 0x0838, 0x0AB4, 0x3EB4, 0x3FF4, 0x0004,
+		0x0044, 0x3E24, 0x07FC, 0x0AE4, 0x3EC8, 0x3FEC, 0x0004,
+		0x0048, 0x3E24, 0x07C4, 0x0B08, 0x3EDC, 0x3FE4, 0x0008,
+		0x0048, 0x3E20, 0x0788, 0x0B3C, 0x3EF4, 0x3FD8, 0x0008,
+		0x004C, 0x3E20, 0x074C, 0x0B60, 0x3F0C, 0x3FD0, 0x000C,
+		0x0050, 0x3E20, 0x0710, 0x0B8C, 0x3F24, 0x3FC4, 0x000C,
+		0x0050, 0x3E20, 0x06D4, 0x0BB0, 0x3F40, 0x3FBC, 0x0010,
+		0x0054, 0x3E24, 0x0698, 0x0BD4, 0x3F5C, 0x3FB0, 0x0010,
+		0x0054, 0x3E24, 0x065C, 0x0BFC, 0x3F78, 0x3FA4, 0x0014,
+		0x0054, 0x3E28, 0x0624, 0x0C1C, 0x3F98, 0x3F98, 0x0014,
+		0x0058, 0x3E2C, 0x05E4, 0x0C3C, 0x3FB8, 0x3F8C, 0x0018,
+		0x0058, 0x3E34, 0x05A8, 0x0C58, 0x3FD8, 0x3F80, 0x001C,
+		0x0058, 0x3E38, 0x0570, 0x0C78, 0x3FF8, 0x3F74, 0x001C,
+		0x0058, 0x3E40, 0x0534, 0x0C94, 0x0018, 0x3F68, 0x0020,
+		0x0058, 0x3E48, 0x04F4, 0x0CAC, 0x0040, 0x3F5C, 0x0024,
+		0x0058, 0x3E50, 0x04BC, 0x0CC4, 0x0064, 0x3F50, 0x0024,
+		0x0054, 0x3E58, 0x0484, 0x0CD8, 0x008C, 0x3F44, 0x0028,
+		0x0054, 0x3E60, 0x0448, 0x0CEC, 0x00B4, 0x3F38, 0x002C,
+		0x0054, 0x3E68, 0x0410, 0x0CFC, 0x00E0, 0x3F28, 0x0030,
+		0x0054, 0x3E74, 0x03D4, 0x0D0C, 0x010C, 0x3F1C, 0x0030,
+		0x0050, 0x3E7C, 0x03A0, 0x0D18, 0x0138, 0x3F10, 0x0034,
+		0x0050, 0x3E88, 0x0364, 0x0D24, 0x0164, 0x3F04, 0x0038,
+		0x004C, 0x3E94, 0x0330, 0x0D30, 0x0194, 0x3EF4, 0x0038,
+		0x004C, 0x3EA0, 0x02F8, 0x0D34, 0x01C4, 0x3EE8, 0x003C,
+		0x0048, 0x3EAC, 0x02C0, 0x0D3C, 0x01F4, 0x3EDC, 0x0040,
+		0x0048, 0x3EB8, 0x0290, 0x0D3C, 0x0224, 0x3ED0, 0x0040,
+		0x0044, 0x3EC4, 0x0258, 0x0D40, 0x0258, 0x3EC4, 0x0044
+};
+
+//=========================================
+// <num_taps>    = 7
+// <num_phases>  = 64
+// <scale_ratio> = 1.49999 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_7tap_64p_149[231] = {
+		0x3F68, 0x3FEC, 0x08A8, 0x08AC, 0x3FF0, 0x3F68, 0x0000,
+		0x3F70, 0x3FDC, 0x0888, 0x08CC, 0x0000, 0x3F60, 0x0000,
+		0x3F74, 0x3FC8, 0x0868, 0x08F0, 0x0014, 0x3F58, 0x0000,
+		0x3F7C, 0x3FB4, 0x0844, 0x0908, 0x002C, 0x3F54, 0x0004,
+		0x3F84, 0x3FA4, 0x0820, 0x0924, 0x0044, 0x3F4C, 0x0004,
+		0x3F88, 0x3F90, 0x0800, 0x0944, 0x005C, 0x3F44, 0x0004,
+		0x3F90, 0x3F80, 0x07D8, 0x095C, 0x0074, 0x3F40, 0x0008,
+		0x3F98, 0x3F70, 0x07B0, 0x097C, 0x008C, 0x3F38, 0x0008,
+		0x3F9C, 0x3F60, 0x0790, 0x0994, 0x00A8, 0x3F30, 0x0008,
+		0x3FA4, 0x3F54, 0x0764, 0x09B0, 0x00C4, 0x3F28, 0x0008,
+		0x3FA8, 0x3F48, 0x0740, 0x09C4, 0x00DC, 0x3F24, 0x000C,
+		0x3FB0, 0x3F38, 0x0718, 0x09DC, 0x00FC, 0x3F1C, 0x000C,
+		0x3FB4, 0x3F2C, 0x06F0, 0x09F4, 0x0118, 0x3F18, 0x000C,
+		0x3FBC, 0x3F24, 0x06C8, 0x0A08, 0x0134, 0x3F10, 0x000C,
+		0x3FC0, 0x3F18, 0x06A0, 0x0A1C, 0x0154, 0x3F08, 0x0010,
+		0x3FC8, 0x3F10, 0x0678, 0x0A2C, 0x0170, 0x3F04, 0x0010,
+		0x3FCC, 0x3F04, 0x0650, 0x0A40, 0x0190, 0x3F00, 0x0010,
+		0x3FD0, 0x3EFC, 0x0628, 0x0A54, 0x01B0, 0x3EF8, 0x0010,
+		0x3FD4, 0x3EF4, 0x0600, 0x0A64, 0x01D0, 0x3EF4, 0x0010,
+		0x3FDC, 0x3EEC, 0x05D8, 0x0A6C, 0x01F4, 0x3EF0, 0x0010,
+		0x3FE0, 0x3EE8, 0x05B0, 0x0A7C, 0x0214, 0x3EE8, 0x0010,
+		0x3FE4, 0x3EE0, 0x0588, 0x0A88, 0x0238, 0x3EE4, 0x0010,
+		0x3FE8, 0x3EDC, 0x055C, 0x0A98, 0x0258, 0x3EE0, 0x0010,
+		0x3FEC, 0x3ED8, 0x0534, 0x0AA0, 0x027C, 0x3EDC, 0x0010,
+		0x3FF0, 0x3ED4, 0x050C, 0x0AAC, 0x02A0, 0x3ED8, 0x000C,
+		0x3FF4, 0x3ED0, 0x04E4, 0x0AB4, 0x02C4, 0x3ED4, 0x000C,
+		0x3FF4, 0x3ECC, 0x04C0, 0x0ABC, 0x02E8, 0x3ED0, 0x000C,
+		0x3FF8, 0x3ECC, 0x0494, 0x0AC0, 0x030C, 0x3ED0, 0x000C,
+		0x3FFC, 0x3EC8, 0x046C, 0x0AC8, 0x0334, 0x3ECC, 0x0008,
+		0x0000, 0x3EC8, 0x0444, 0x0AC8, 0x0358, 0x3ECC, 0x0008,
+		0x0000, 0x3EC8, 0x041C, 0x0ACC, 0x0380, 0x3EC8, 0x0008,
+		0x0000, 0x3EC8, 0x03F4, 0x0AD0, 0x03A8, 0x3EC8, 0x0004,
+		0x0004, 0x3EC8, 0x03CC, 0x0AD0, 0x03CC, 0x3EC8, 0x0004
+};
+
+//=========================================
+// <num_taps>    = 7
+// <num_phases>  = 64
+// <scale_ratio> = 1.83332 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_7tap_64p_183[231] = {
+		0x3FA4, 0x01E8, 0x0674, 0x0674, 0x01E8, 0x3FA4, 0x0000,
+		0x3FA4, 0x01D4, 0x0668, 0x0684, 0x01F8, 0x3FA4, 0x0000,
+		0x3FA4, 0x01C4, 0x0658, 0x0690, 0x0208, 0x3FA8, 0x0000,
+		0x3FA0, 0x01B4, 0x064C, 0x06A0, 0x021C, 0x3FA8, 0x3FFC,
+		0x3FA0, 0x01A4, 0x063C, 0x06AC, 0x022C, 0x3FAC, 0x3FFC,
+		0x3FA0, 0x0194, 0x0630, 0x06B4, 0x0240, 0x3FAC, 0x3FFC,
+		0x3FA0, 0x0184, 0x0620, 0x06C4, 0x0250, 0x3FB0, 0x3FF8,
+		0x3FA0, 0x0174, 0x0614, 0x06CC, 0x0264, 0x3FB0, 0x3FF8,
+		0x3FA0, 0x0164, 0x0604, 0x06D8, 0x0278, 0x3FB4, 0x3FF4,
+		0x3FA0, 0x0154, 0x05F4, 0x06E4, 0x0288, 0x3FB8, 0x3FF4,
+		0x3FA0, 0x0148, 0x05E4, 0x06EC, 0x029C, 0x3FBC, 0x3FF0,
+		0x3FA0, 0x0138, 0x05D4, 0x06F4, 0x02B0, 0x3FC0, 0x3FF0,
+		0x3FA0, 0x0128, 0x05C4, 0x0704, 0x02C4, 0x3FC0, 0x3FEC,
+		0x3FA0, 0x011C, 0x05B4, 0x0708, 0x02D8, 0x3FC4, 0x3FEC,
+		0x3FA4, 0x010C, 0x05A4, 0x0714, 0x02E8, 0x3FC8, 0x3FE8,
+		0x3FA4, 0x0100, 0x0590, 0x0718, 0x02FC, 0x3FD0, 0x3FE8,
+		0x3FA4, 0x00F0, 0x0580, 0x0724, 0x0310, 0x3FD4, 0x3FE4,
+		0x3FA4, 0x00E4, 0x056C, 0x072C, 0x0324, 0x3FD8, 0x3FE4,
+		0x3FA8, 0x00D8, 0x055C, 0x0730, 0x0338, 0x3FDC, 0x3FE0,
+		0x3FA8, 0x00CC, 0x0548, 0x0738, 0x034C, 0x3FE4, 0x3FDC,
+		0x3FA8, 0x00BC, 0x0538, 0x0740, 0x0360, 0x3FE8, 0x3FDC,
+		0x3FAC, 0x00B0, 0x0528, 0x0744, 0x0374, 0x3FEC, 0x3FD8,
+		0x3FAC, 0x00A4, 0x0514, 0x0748, 0x0388, 0x3FF4, 0x3FD8,
+		0x3FB0, 0x0098, 0x0500, 0x074C, 0x039C, 0x3FFC, 0x3FD4,
+		0x3FB0, 0x0090, 0x04EC, 0x0750, 0x03B0, 0x0000, 0x3FD4,
+		0x3FB0, 0x0084, 0x04DC, 0x0758, 0x03C4, 0x0004, 0x3FD0,
+		0x3FB4, 0x0078, 0x04CC, 0x0758, 0x03D8, 0x000C, 0x3FCC,
+		0x3FB4, 0x006C, 0x04B8, 0x075C, 0x03EC, 0x0014, 0x3FCC,
+		0x3FB8, 0x0064, 0x04A0, 0x0760, 0x0400, 0x001C, 0x3FC8,
+		0x3FB8, 0x0058, 0x0490, 0x0760, 0x0414, 0x0024, 0x3FC8,
+		0x3FBC, 0x0050, 0x047C, 0x0760, 0x0428, 0x002C, 0x3FC4,
+		0x3FBC, 0x0048, 0x0464, 0x0764, 0x043C, 0x0034, 0x3FC4,
+		0x3FC0, 0x003C, 0x0454, 0x0764, 0x0450, 0x003C, 0x3FC0
+};
+
+//=========================================
+// <num_taps>    = 8
+// <num_phases>  = 64
+// <scale_ratio> = 0.83333 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_8tap_64p_upscale[264] = {
+		0x0000, 0x0000, 0x0000, 0x1000, 0x0000, 0x0000, 0x0000, 0x0000,
+		0x3FFC, 0x0014, 0x3FC8, 0x1000, 0x0038, 0x3FEC, 0x0004, 0x0000,
+		0x3FF4, 0x0024, 0x3F94, 0x0FFC, 0x0074, 0x3FD8, 0x000C, 0x0000,
+		0x3FF0, 0x0038, 0x3F60, 0x0FEC, 0x00B4, 0x3FC4, 0x0014, 0x0000,
+		0x3FEC, 0x004C, 0x3F2C, 0x0FE4, 0x00F4, 0x3FAC, 0x0018, 0x0000,
+		0x3FE4, 0x005C, 0x3F00, 0x0FD4, 0x0138, 0x3F94, 0x0020, 0x0000,
+		0x3FE0, 0x006C, 0x3ED0, 0x0FC4, 0x017C, 0x3F7C, 0x0028, 0x0000,
+		0x3FDC, 0x007C, 0x3EA8, 0x0FA4, 0x01C4, 0x3F68, 0x0030, 0x0000,
+		0x3FD8, 0x0088, 0x3E80, 0x0F90, 0x020C, 0x3F50, 0x0038, 0x3FFC,
+		0x3FD4, 0x0098, 0x3E58, 0x0F70, 0x0258, 0x3F38, 0x0040, 0x3FFC,
+		0x3FD0, 0x00A4, 0x3E34, 0x0F54, 0x02A0, 0x3F1C, 0x004C, 0x3FFC,
+		0x3FD0, 0x00B0, 0x3E14, 0x0F28, 0x02F0, 0x3F04, 0x0054, 0x3FFC,
+		0x3FCC, 0x00BC, 0x3DF4, 0x0F08, 0x033C, 0x3EEC, 0x005C, 0x3FF8,
+		0x3FC8, 0x00C8, 0x3DD8, 0x0EDC, 0x038C, 0x3ED4, 0x0064, 0x3FF8,
+		0x3FC8, 0x00D0, 0x3DC0, 0x0EAC, 0x03E0, 0x3EBC, 0x006C, 0x3FF4,
+		0x3FC4, 0x00D8, 0x3DA8, 0x0E7C, 0x0430, 0x3EA4, 0x0078, 0x3FF4,
+		0x3FC4, 0x00E0, 0x3D94, 0x0E48, 0x0484, 0x3E8C, 0x0080, 0x3FF0,
+		0x3FC4, 0x00E8, 0x3D80, 0x0E10, 0x04D8, 0x3E74, 0x0088, 0x3FF0,
+		0x3FC4, 0x00F0, 0x3D70, 0x0DD8, 0x052C, 0x3E5C, 0x0090, 0x3FEC,
+		0x3FC0, 0x00F4, 0x3D60, 0x0DA0, 0x0584, 0x3E44, 0x0098, 0x3FEC,
+		0x3FC0, 0x00F8, 0x3D54, 0x0D68, 0x05D8, 0x3E2C, 0x00A0, 0x3FE8,
+		0x3FC0, 0x00FC, 0x3D48, 0x0D20, 0x0630, 0x3E18, 0x00AC, 0x3FE8,
+		0x3FC0, 0x0100, 0x3D40, 0x0CE0, 0x0688, 0x3E00, 0x00B4, 0x3FE4,
+		0x3FC4, 0x0100, 0x3D3C, 0x0C98, 0x06DC, 0x3DEC, 0x00BC, 0x3FE4,
+		0x3FC4, 0x0100, 0x3D38, 0x0C58, 0x0734, 0x3DD8, 0x00C0, 0x3FE0,
+		0x3FC4, 0x0104, 0x3D38, 0x0C0C, 0x078C, 0x3DC4, 0x00C8, 0x3FDC,
+		0x3FC4, 0x0100, 0x3D38, 0x0BC4, 0x07E4, 0x3DB0, 0x00D0, 0x3FDC,
+		0x3FC4, 0x0100, 0x3D38, 0x0B78, 0x083C, 0x3DA0, 0x00D8, 0x3FD8,
+		0x3FC8, 0x0100, 0x3D3C, 0x0B28, 0x0890, 0x3D90, 0x00DC, 0x3FD8,
+		0x3FC8, 0x00FC, 0x3D40, 0x0ADC, 0x08E8, 0x3D80, 0x00E4, 0x3FD4,
+		0x3FCC, 0x00FC, 0x3D48, 0x0A84, 0x093C, 0x3D74, 0x00E8, 0x3FD4,
+		0x3FCC, 0x00F8, 0x3D50, 0x0A38, 0x0990, 0x3D64, 0x00F0, 0x3FD0,
+		0x3FD0, 0x00F4, 0x3D58, 0x09E0, 0x09E4, 0x3D5C, 0x00F4, 0x3FD0
+};
+
+//=========================================
+// <num_taps>    = 8
+// <num_phases>  = 64
+// <scale_ratio> = 1.16666 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_8tap_64p_116[264] = {
+		0x0080, 0x3E90, 0x0268, 0x0D14, 0x0264, 0x3E90, 0x0080, 0x0000,
+		0x007C, 0x3E9C, 0x0238, 0x0D14, 0x0298, 0x3E84, 0x0080, 0x0000,
+		0x0078, 0x3EAC, 0x0200, 0x0D10, 0x02D0, 0x3E78, 0x0084, 0x0000,
+		0x0078, 0x3EB8, 0x01D0, 0x0D0C, 0x0304, 0x3E6C, 0x0084, 0x0000,
+		0x0074, 0x3EC8, 0x01A0, 0x0D00, 0x033C, 0x3E60, 0x0088, 0x0000,
+		0x0070, 0x3ED4, 0x0170, 0x0D00, 0x0374, 0x3E54, 0x0088, 0x3FFC,
+		0x006C, 0x3EE4, 0x0140, 0x0CF8, 0x03AC, 0x3E48, 0x0088, 0x3FFC,
+		0x006C, 0x3EF0, 0x0114, 0x0CE8, 0x03E4, 0x3E3C, 0x008C, 0x3FFC,
+		0x0068, 0x3F00, 0x00E8, 0x0CD8, 0x041C, 0x3E34, 0x008C, 0x3FFC,
+		0x0064, 0x3F10, 0x00BC, 0x0CCC, 0x0454, 0x3E28, 0x008C, 0x3FFC,
+		0x0060, 0x3F1C, 0x0090, 0x0CBC, 0x0490, 0x3E20, 0x008C, 0x3FFC,
+		0x005C, 0x3F2C, 0x0068, 0x0CA4, 0x04CC, 0x3E18, 0x008C, 0x3FFC,
+		0x0058, 0x3F38, 0x0040, 0x0C94, 0x0504, 0x3E10, 0x008C, 0x3FFC,
+		0x0054, 0x3F48, 0x001C, 0x0C7C, 0x0540, 0x3E08, 0x0088, 0x3FFC,
+		0x0050, 0x3F54, 0x3FF8, 0x0C60, 0x057C, 0x3E04, 0x0088, 0x3FFC,
+		0x004C, 0x3F64, 0x3FD4, 0x0C44, 0x05B8, 0x3DFC, 0x0088, 0x3FFC,
+		0x0048, 0x3F70, 0x3FB4, 0x0C28, 0x05F4, 0x3DF8, 0x0084, 0x3FFC,
+		0x0044, 0x3F80, 0x3F90, 0x0C0C, 0x0630, 0x3DF4, 0x0080, 0x3FFC,
+		0x0040, 0x3F8C, 0x3F70, 0x0BE8, 0x066C, 0x3DF4, 0x0080, 0x3FFC,
+		0x003C, 0x3F9C, 0x3F50, 0x0BC8, 0x06A8, 0x3DF0, 0x007C, 0x3FFC,
+		0x0038, 0x3FA8, 0x3F34, 0x0BA0, 0x06E4, 0x3DF0, 0x0078, 0x0000,
+		0x0034, 0x3FB4, 0x3F18, 0x0B80, 0x071C, 0x3DF0, 0x0074, 0x0000,
+		0x0030, 0x3FC0, 0x3EFC, 0x0B5C, 0x0758, 0x3DF0, 0x0070, 0x0000,
+		0x002C, 0x3FCC, 0x3EE4, 0x0B34, 0x0794, 0x3DF4, 0x0068, 0x0000,
+		0x002C, 0x3FDC, 0x3ECC, 0x0B08, 0x07CC, 0x3DF4, 0x0064, 0x0000,
+		0x0028, 0x3FE4, 0x3EB4, 0x0AE0, 0x0808, 0x3DF8, 0x0060, 0x0000,
+		0x0024, 0x3FF0, 0x3EA0, 0x0AB0, 0x0840, 0x3E00, 0x0058, 0x0004,
+		0x0020, 0x3FFC, 0x3E90, 0x0A84, 0x0878, 0x3E04, 0x0050, 0x0004,
+		0x001C, 0x0004, 0x3E7C, 0x0A54, 0x08B0, 0x3E0C, 0x004C, 0x0008,
+		0x0018, 0x000C, 0x3E68, 0x0A28, 0x08E8, 0x3E18, 0x0044, 0x0008,
+		0x0018, 0x0018, 0x3E54, 0x09F4, 0x0920, 0x3E20, 0x003C, 0x000C,
+		0x0014, 0x0020, 0x3E48, 0x09C0, 0x0954, 0x3E2C, 0x0034, 0x0010,
+		0x0010, 0x002C, 0x3E3C, 0x098C, 0x0988, 0x3E38, 0x002C, 0x0010
+};
+
+//=========================================
+// <num_taps>    = 8
+// <num_phases>  = 64
+// <scale_ratio> = 1.49999 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_8tap_64p_149[264] = {
+		0x0008, 0x3E8C, 0x03F8, 0x0AE8, 0x03F8, 0x3E8C, 0x0008, 0x0000,
+		0x000C, 0x3E8C, 0x03D0, 0x0AE8, 0x0420, 0x3E90, 0x0000, 0x0000,
+		0x000C, 0x3E8C, 0x03AC, 0x0AE8, 0x0444, 0x3E90, 0x0000, 0x0000,
+		0x0010, 0x3E90, 0x0384, 0x0AE0, 0x046C, 0x3E94, 0x3FFC, 0x0000,
+		0x0014, 0x3E90, 0x035C, 0x0ADC, 0x0494, 0x3E94, 0x3FF8, 0x0004,
+		0x0018, 0x3E90, 0x0334, 0x0AD8, 0x04BC, 0x3E98, 0x3FF4, 0x0004,
+		0x001C, 0x3E94, 0x0310, 0x0AD0, 0x04E4, 0x3E9C, 0x3FEC, 0x0004,
+		0x0020, 0x3E98, 0x02E8, 0x0AC4, 0x050C, 0x3EA0, 0x3FE8, 0x0008,
+		0x0020, 0x3E98, 0x02C4, 0x0AC0, 0x0534, 0x3EA4, 0x3FE4, 0x0008,
+		0x0024, 0x3E9C, 0x02A0, 0x0AB4, 0x055C, 0x3EAC, 0x3FDC, 0x0008,
+		0x0024, 0x3EA0, 0x027C, 0x0AA8, 0x0584, 0x3EB0, 0x3FD8, 0x000C,
+		0x0028, 0x3EA4, 0x0258, 0x0A9C, 0x05AC, 0x3EB8, 0x3FD0, 0x000C,
+		0x0028, 0x3EA8, 0x0234, 0x0A90, 0x05D4, 0x3EC0, 0x3FC8, 0x0010,
+		0x002C, 0x3EAC, 0x0210, 0x0A80, 0x05FC, 0x3EC8, 0x3FC4, 0x0010,
+		0x002C, 0x3EB4, 0x01F0, 0x0A70, 0x0624, 0x3ED0, 0x3FBC, 0x0010,
+		0x002C, 0x3EB8, 0x01CC, 0x0A60, 0x064C, 0x3EDC, 0x3FB4, 0x0014,
+		0x0030, 0x3EBC, 0x01A8, 0x0A50, 0x0674, 0x3EE4, 0x3FB0, 0x0014,
+		0x0030, 0x3EC4, 0x0188, 0x0A38, 0x069C, 0x3EF0, 0x3FA8, 0x0018,
+		0x0030, 0x3ECC, 0x0168, 0x0A28, 0x06C0, 0x3EFC, 0x3FA0, 0x0018,
+		0x0030, 0x3ED0, 0x0148, 0x0A14, 0x06E8, 0x3F08, 0x3F98, 0x001C,
+		0x0030, 0x3ED8, 0x012C, 0x0A00, 0x070C, 0x3F14, 0x3F90, 0x001C,
+		0x0034, 0x3EE0, 0x0108, 0x09E4, 0x0734, 0x3F24, 0x3F8C, 0x001C,
+		0x0034, 0x3EE4, 0x00EC, 0x09CC, 0x0758, 0x3F34, 0x3F84, 0x0020,
+		0x0034, 0x3EEC, 0x00D0, 0x09B8, 0x077C, 0x3F40, 0x3F7C, 0x0020,
+		0x0034, 0x3EF4, 0x00B4, 0x0998, 0x07A4, 0x3F50, 0x3F74, 0x0024,
+		0x0030, 0x3EFC, 0x0098, 0x0980, 0x07C8, 0x3F64, 0x3F6C, 0x0024,
+		0x0030, 0x3F04, 0x0080, 0x0968, 0x07E8, 0x3F74, 0x3F64, 0x0024,
+		0x0030, 0x3F0C, 0x0060, 0x094C, 0x080C, 0x3F88, 0x3F5C, 0x0028,
+		0x0030, 0x3F14, 0x0048, 0x0930, 0x0830, 0x3F98, 0x3F54, 0x0028,
+		0x0030, 0x3F1C, 0x0030, 0x0914, 0x0850, 0x3FAC, 0x3F4C, 0x0028,
+		0x0030, 0x3F24, 0x0018, 0x08F0, 0x0874, 0x3FC0, 0x3F44, 0x002C,
+		0x002C, 0x3F2C, 0x0000, 0x08D4, 0x0894, 0x3FD8, 0x3F3C, 0x002C,
+		0x002C, 0x3F34, 0x3FEC, 0x08B4, 0x08B4, 0x3FEC, 0x3F34, 0x002C
+};
+
+//=========================================
+// <num_taps>    = 8
+// <num_phases>  = 64
+// <scale_ratio> = 1.83332 (input/output)
+// <sharpness>   = 0
+// <CoefType>    = ModifiedLanczos
+// <CoefQuant>   = 1.10
+// <CoefOut>     = 1.12
+//=========================================
+static const uint16_t filter_8tap_64p_183[264] = {
+		0x3F88, 0x0048, 0x047C, 0x0768, 0x047C, 0x0048, 0x3F88, 0x0000,
+		0x3F88, 0x003C, 0x0468, 0x076C, 0x0490, 0x0054, 0x3F84, 0x0000,
+		0x3F8C, 0x0034, 0x0454, 0x0768, 0x04A4, 0x005C, 0x3F84, 0x0000,
+		0x3F8C, 0x0028, 0x0444, 0x076C, 0x04B4, 0x0068, 0x3F80, 0x0000,
+		0x3F90, 0x0020, 0x042C, 0x0768, 0x04C8, 0x0074, 0x3F80, 0x0000,
+		0x3F90, 0x0018, 0x041C, 0x0764, 0x04DC, 0x0080, 0x3F7C, 0x0000,
+		0x3F94, 0x0010, 0x0408, 0x075C, 0x04F0, 0x008C, 0x3F7C, 0x0000,
+		0x3F94, 0x0004, 0x03F8, 0x0760, 0x0500, 0x0098, 0x3F7C, 0x3FFC,
+		0x3F98, 0x0000, 0x03E0, 0x075C, 0x0514, 0x00A4, 0x3F78, 0x3FFC,
+		0x3F9C, 0x3FF8, 0x03CC, 0x0754, 0x0528, 0x00B0, 0x3F78, 0x3FFC,
+		0x3F9C, 0x3FF0, 0x03B8, 0x0754, 0x0538, 0x00BC, 0x3F78, 0x3FFC,
+		0x3FA0, 0x3FE8, 0x03A4, 0x0750, 0x054C, 0x00CC, 0x3F74, 0x3FF8,
+		0x3FA4, 0x3FE0, 0x0390, 0x074C, 0x055C, 0x00D8, 0x3F74, 0x3FF8,
+		0x3FA4, 0x3FDC, 0x037C, 0x0744, 0x0570, 0x00E4, 0x3F74, 0x3FF8,
+		0x3FA8, 0x3FD4, 0x0368, 0x0740, 0x0580, 0x00F4, 0x3F74, 0x3FF4,
+		0x3FA8, 0x3FCC, 0x0354, 0x073C, 0x0590, 0x0104, 0x3F74, 0x3FF4,
+		0x3FAC, 0x3FC8, 0x0340, 0x0730, 0x05A4, 0x0110, 0x3F74, 0x3FF4,
+		0x3FB0, 0x3FC0, 0x0330, 0x0728, 0x05B4, 0x0120, 0x3F74, 0x3FF0,
+		0x3FB0, 0x3FBC, 0x031C, 0x0724, 0x05C4, 0x0130, 0x3F70, 0x3FF0,
+		0x3FB4, 0x3FB4, 0x0308, 0x0720, 0x05D4, 0x013C, 0x3F70, 0x3FF0,
+		0x3FB8, 0x3FB0, 0x02F4, 0x0714, 0x05E4, 0x014C, 0x3F74, 0x3FEC,
+		0x3FB8, 0x3FAC, 0x02E0, 0x0708, 0x05F8, 0x015C, 0x3F74, 0x3FEC,
+		0x3FBC, 0x3FA8, 0x02CC, 0x0704, 0x0604, 0x016C, 0x3F74, 0x3FE8,
+		0x3FC0, 0x3FA0, 0x02BC, 0x06F8, 0x0614, 0x017C, 0x3F74, 0x3FE8,
+		0x3FC0, 0x3F9C, 0x02A8, 0x06F4, 0x0624, 0x018C, 0x3F74, 0x3FE4,
+		0x3FC4, 0x3F98, 0x0294, 0x06E8, 0x0634, 0x019C, 0x3F74, 0x3FE4,
+		0x3FC8, 0x3F94, 0x0284, 0x06D8, 0x0644, 0x01AC, 0x3F78, 0x3FE0,
+		0x3FC8, 0x3F90, 0x0270, 0x06D4, 0x0650, 0x01BC, 0x3F78, 0x3FE0,
+		0x3FCC, 0x3F8C, 0x025C, 0x06C8, 0x0660, 0x01D0, 0x3F78, 0x3FDC,
+		0x3FCC, 0x3F8C, 0x024C, 0x06B8, 0x066C, 0x01E0, 0x3F7C, 0x3FDC,
+		0x3FD0, 0x3F88, 0x0238, 0x06B0, 0x067C, 0x01F0, 0x3F7C, 0x3FD8,
+		0x3FD4, 0x3F84, 0x0228, 0x069C, 0x0688, 0x0204, 0x3F80, 0x3FD8,
+		0x3FD4, 0x3F84, 0x0214, 0x0694, 0x0694, 0x0214, 0x3F84, 0x3FD4
+};
+
+const uint16_t *get_filter_3tap_16p(struct fixed31_32 ratio)
+{
+	if (ratio.value < dc_fixpt_one.value)
+		return filter_3tap_16p_upscale;
+	else if (ratio.value < dc_fixpt_from_fraction(4, 3).value)
+		return filter_3tap_16p_116;
+	else if (ratio.value < dc_fixpt_from_fraction(5, 3).value)
+		return filter_3tap_16p_149;
+	else
+		return filter_3tap_16p_183;
+}
+
+const uint16_t *get_filter_3tap_64p(struct fixed31_32 ratio)
+{
+	if (ratio.value < dc_fixpt_one.value)
+		return filter_3tap_64p_upscale;
+	else if (ratio.value < dc_fixpt_from_fraction(4, 3).value)
+		return filter_3tap_64p_116;
+	else if (ratio.value < dc_fixpt_from_fraction(5, 3).value)
+		return filter_3tap_64p_149;
+	else
+		return filter_3tap_64p_183;
+}
+
+const uint16_t *get_filter_4tap_16p(struct fixed31_32 ratio)
+{
+	if (ratio.value < dc_fixpt_one.value)
+		return filter_4tap_16p_upscale;
+	else if (ratio.value < dc_fixpt_from_fraction(4, 3).value)
+		return filter_4tap_16p_116;
+	else if (ratio.value < dc_fixpt_from_fraction(5, 3).value)
+		return filter_4tap_16p_149;
+	else
+		return filter_4tap_16p_183;
+}
+
+const uint16_t *get_filter_4tap_64p(struct fixed31_32 ratio)
+{
+	if (ratio.value < dc_fixpt_one.value)
+		return filter_4tap_64p_upscale;
+	else if (ratio.value < dc_fixpt_from_fraction(4, 3).value)
+		return filter_4tap_64p_116;
+	else if (ratio.value < dc_fixpt_from_fraction(5, 3).value)
+		return filter_4tap_64p_149;
+	else
+		return filter_4tap_64p_183;
+}
+
+const uint16_t *get_filter_5tap_64p(struct fixed31_32 ratio)
+{
+	if (ratio.value < dc_fixpt_one.value)
+		return filter_5tap_64p_upscale;
+	else if (ratio.value < dc_fixpt_from_fraction(4, 3).value)
+		return filter_5tap_64p_116;
+	else if (ratio.value < dc_fixpt_from_fraction(5, 3).value)
+		return filter_5tap_64p_149;
+	else
+		return filter_5tap_64p_183;
+}
+
+const uint16_t *get_filter_6tap_64p(struct fixed31_32 ratio)
+{
+	if (ratio.value < dc_fixpt_one.value)
+		return filter_6tap_64p_upscale;
+	else if (ratio.value < dc_fixpt_from_fraction(4, 3).value)
+		return filter_6tap_64p_116;
+	else if (ratio.value < dc_fixpt_from_fraction(5, 3).value)
+		return filter_6tap_64p_149;
+	else
+		return filter_6tap_64p_183;
+}
+
+const uint16_t *get_filter_7tap_64p(struct fixed31_32 ratio)
+{
+	if (ratio.value < dc_fixpt_one.value)
+		return filter_7tap_64p_upscale;
+	else if (ratio.value < dc_fixpt_from_fraction(4, 3).value)
+		return filter_7tap_64p_116;
+	else if (ratio.value < dc_fixpt_from_fraction(5, 3).value)
+		return filter_7tap_64p_149;
+	else
+		return filter_7tap_64p_183;
+}
+
+const uint16_t *get_filter_8tap_64p(struct fixed31_32 ratio)
+{
+	if (ratio.value < dc_fixpt_one.value)
+		return filter_8tap_64p_upscale;
+	else if (ratio.value < dc_fixpt_from_fraction(4, 3).value)
+		return filter_8tap_64p_116;
+	else if (ratio.value < dc_fixpt_from_fraction(5, 3).value)
+		return filter_8tap_64p_149;
+	else
+		return filter_8tap_64p_183;
+}
+
+const uint16_t *get_filter_2tap_16p(void)
+{
+	return filter_2tap_16p;
+}
+
+const uint16_t *get_filter_2tap_64p(void)
+{
+	return filter_2tap_64p;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_scl_filters_old.c b/drivers/gpu/drm/amd/display/dc/dce/dce_scl_filters_old.c
new file mode 100644
index 000000000000..bb0e1b80ec3c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_scl_filters_old.c
@@ -0,0 +1,25 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.c b/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.c
new file mode 100644
index 000000000000..1130d7619b26
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.c
@@ -0,0 +1,1569 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dc_bios_types.h"
+#include "dce_stream_encoder.h"
+#include "reg_helper.h"
+#include "hw_shared.h"
+
+#define DC_LOGGER \
+		enc110->base.ctx->logger
+
+#define REG(reg)\
+	(enc110->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	enc110->se_shift->field_name, enc110->se_mask->field_name
+
+#define VBI_LINE_0 0
+#define DP_BLANK_MAX_RETRY 20
+#define HDMI_CLOCK_CHANNEL_RATE_MORE_340M 340000
+
+#ifndef TMDS_CNTL__TMDS_PIXEL_ENCODING_MASK
+	#define TMDS_CNTL__TMDS_PIXEL_ENCODING_MASK       0x00000010L
+	#define TMDS_CNTL__TMDS_COLOR_FORMAT_MASK         0x00000300L
+	#define	TMDS_CNTL__TMDS_PIXEL_ENCODING__SHIFT     0x00000004
+	#define	TMDS_CNTL__TMDS_COLOR_FORMAT__SHIFT       0x00000008
+#endif
+
+enum {
+	DP_MST_UPDATE_MAX_RETRY = 50
+};
+
+#define DCE110_SE(audio)\
+	container_of(audio, struct dce110_stream_encoder, base)
+
+#define CTX \
+	enc110->base.ctx
+
+static void dce110_update_generic_info_packet(
+	struct dce110_stream_encoder *enc110,
+	uint32_t packet_index,
+	const struct dc_info_packet *info_packet)
+{
+	/* TODOFPGA Figure out a proper number for max_retries polling for lock
+	 * use 50 for now.
+	 */
+	uint32_t max_retries = 50;
+
+	/*we need turn on clock before programming AFMT block*/
+	if (REG(AFMT_CNTL))
+		REG_UPDATE(AFMT_CNTL, AFMT_AUDIO_CLOCK_EN, 1);
+
+	if (REG(AFMT_VBI_PACKET_CONTROL1)) {
+		if (packet_index >= 8)
+			ASSERT(0);
+
+		/* poll dig_update_lock is not locked -> asic internal signal
+		 * assume otg master lock will unlock it
+		 */
+/*		REG_WAIT(AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_LOCK_STATUS,
+				0, 10, max_retries);*/
+
+		/* check if HW reading GSP memory */
+		REG_WAIT(AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_CONFLICT,
+				0, 10, max_retries);
+
+		/* HW does is not reading GSP memory not reading too long ->
+		 * something wrong. clear GPS memory access and notify?
+		 * hw SW is writing to GSP memory
+		 */
+		REG_UPDATE(AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_CONFLICT_CLR, 1);
+	}
+	/* choose which generic packet to use */
+	{
+		REG_READ(AFMT_VBI_PACKET_CONTROL);
+		REG_UPDATE(AFMT_VBI_PACKET_CONTROL,
+				AFMT_GENERIC_INDEX, packet_index);
+	}
+
+	/* write generic packet header
+	 * (4th byte is for GENERIC0 only) */
+	{
+		REG_SET_4(AFMT_GENERIC_HDR, 0,
+				AFMT_GENERIC_HB0, info_packet->hb0,
+				AFMT_GENERIC_HB1, info_packet->hb1,
+				AFMT_GENERIC_HB2, info_packet->hb2,
+				AFMT_GENERIC_HB3, info_packet->hb3);
+	}
+
+	/* write generic packet contents
+	 * (we never use last 4 bytes)
+	 * there are 8 (0-7) mmDIG0_AFMT_GENERIC0_x registers */
+	{
+		const uint32_t *content =
+			(const uint32_t *) &info_packet->sb[0];
+
+		REG_WRITE(AFMT_GENERIC_0, *content++);
+		REG_WRITE(AFMT_GENERIC_1, *content++);
+		REG_WRITE(AFMT_GENERIC_2, *content++);
+		REG_WRITE(AFMT_GENERIC_3, *content++);
+		REG_WRITE(AFMT_GENERIC_4, *content++);
+		REG_WRITE(AFMT_GENERIC_5, *content++);
+		REG_WRITE(AFMT_GENERIC_6, *content++);
+		REG_WRITE(AFMT_GENERIC_7, *content);
+	}
+
+	if (!REG(AFMT_VBI_PACKET_CONTROL1)) {
+		/* force double-buffered packet update */
+		REG_UPDATE_2(AFMT_VBI_PACKET_CONTROL,
+			AFMT_GENERIC0_UPDATE, (packet_index == 0),
+			AFMT_GENERIC2_UPDATE, (packet_index == 2));
+	}
+
+	if (REG(AFMT_VBI_PACKET_CONTROL1)) {
+		switch (packet_index) {
+		case 0:
+			REG_UPDATE(AFMT_VBI_PACKET_CONTROL1,
+					AFMT_GENERIC0_FRAME_UPDATE, 1);
+			break;
+		case 1:
+			REG_UPDATE(AFMT_VBI_PACKET_CONTROL1,
+					AFMT_GENERIC1_FRAME_UPDATE, 1);
+			break;
+		case 2:
+			REG_UPDATE(AFMT_VBI_PACKET_CONTROL1,
+					AFMT_GENERIC2_FRAME_UPDATE, 1);
+			break;
+		case 3:
+			REG_UPDATE(AFMT_VBI_PACKET_CONTROL1,
+					AFMT_GENERIC3_FRAME_UPDATE, 1);
+			break;
+		case 4:
+			REG_UPDATE(AFMT_VBI_PACKET_CONTROL1,
+					AFMT_GENERIC4_FRAME_UPDATE, 1);
+			break;
+		case 5:
+			REG_UPDATE(AFMT_VBI_PACKET_CONTROL1,
+					AFMT_GENERIC5_FRAME_UPDATE, 1);
+			break;
+		case 6:
+			REG_UPDATE(AFMT_VBI_PACKET_CONTROL1,
+					AFMT_GENERIC6_FRAME_UPDATE, 1);
+			break;
+		case 7:
+			REG_UPDATE(AFMT_VBI_PACKET_CONTROL1,
+					AFMT_GENERIC7_FRAME_UPDATE, 1);
+			break;
+		default:
+			break;
+		}
+	}
+}
+
+static void dce110_update_hdmi_info_packet(
+	struct dce110_stream_encoder *enc110,
+	uint32_t packet_index,
+	const struct dc_info_packet *info_packet)
+{
+	uint32_t cont, send, line;
+
+	if (info_packet->valid) {
+		dce110_update_generic_info_packet(
+			enc110,
+			packet_index,
+			info_packet);
+
+		/* enable transmission of packet(s) -
+		 * packet transmission begins on the next frame */
+		cont = 1;
+		/* send packet(s) every frame */
+		send = 1;
+		/* select line number to send packets on */
+		line = 2;
+	} else {
+		cont = 0;
+		send = 0;
+		line = 0;
+	}
+
+	/* choose which generic packet control to use */
+	switch (packet_index) {
+	case 0:
+		REG_UPDATE_3(HDMI_GENERIC_PACKET_CONTROL0,
+				HDMI_GENERIC0_CONT, cont,
+				HDMI_GENERIC0_SEND, send,
+				HDMI_GENERIC0_LINE, line);
+		break;
+	case 1:
+		REG_UPDATE_3(HDMI_GENERIC_PACKET_CONTROL0,
+				HDMI_GENERIC1_CONT, cont,
+				HDMI_GENERIC1_SEND, send,
+				HDMI_GENERIC1_LINE, line);
+		break;
+	case 2:
+		REG_UPDATE_3(HDMI_GENERIC_PACKET_CONTROL1,
+				HDMI_GENERIC0_CONT, cont,
+				HDMI_GENERIC0_SEND, send,
+				HDMI_GENERIC0_LINE, line);
+		break;
+	case 3:
+		REG_UPDATE_3(HDMI_GENERIC_PACKET_CONTROL1,
+				HDMI_GENERIC1_CONT, cont,
+				HDMI_GENERIC1_SEND, send,
+				HDMI_GENERIC1_LINE, line);
+		break;
+	case 4:
+		if (REG(HDMI_GENERIC_PACKET_CONTROL2))
+			REG_UPDATE_3(HDMI_GENERIC_PACKET_CONTROL2,
+					HDMI_GENERIC0_CONT, cont,
+					HDMI_GENERIC0_SEND, send,
+					HDMI_GENERIC0_LINE, line);
+		break;
+	case 5:
+		if (REG(HDMI_GENERIC_PACKET_CONTROL2))
+			REG_UPDATE_3(HDMI_GENERIC_PACKET_CONTROL2,
+					HDMI_GENERIC1_CONT, cont,
+					HDMI_GENERIC1_SEND, send,
+					HDMI_GENERIC1_LINE, line);
+		break;
+	case 6:
+		if (REG(HDMI_GENERIC_PACKET_CONTROL3))
+			REG_UPDATE_3(HDMI_GENERIC_PACKET_CONTROL3,
+					HDMI_GENERIC0_CONT, cont,
+					HDMI_GENERIC0_SEND, send,
+					HDMI_GENERIC0_LINE, line);
+		break;
+	case 7:
+		if (REG(HDMI_GENERIC_PACKET_CONTROL3))
+			REG_UPDATE_3(HDMI_GENERIC_PACKET_CONTROL3,
+					HDMI_GENERIC1_CONT, cont,
+					HDMI_GENERIC1_SEND, send,
+					HDMI_GENERIC1_LINE, line);
+		break;
+	default:
+		/* invalid HW packet index */
+		DC_LOG_WARNING(
+			"Invalid HW packet index: %s()\n",
+			__func__);
+		return;
+	}
+}
+
+/* setup stream encoder in dp mode */
+static void dce110_stream_encoder_dp_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	enum dc_color_space output_color_space,
+	bool use_vsc_sdp_for_colorimetry,
+	uint32_t enable_sdp_splitting)
+{
+	uint32_t h_active_start;
+	uint32_t v_active_start;
+	uint32_t misc0 = 0;
+	uint32_t misc1 = 0;
+	uint32_t h_blank;
+	uint32_t h_back_porch;
+	uint8_t colorimetry_bpc;
+	uint8_t dynamic_range_rgb = 0; /*full range*/
+	uint8_t dynamic_range_ycbcr = 1; /*bt709*/
+
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+	struct dc_crtc_timing hw_crtc_timing = *crtc_timing;
+	if (hw_crtc_timing.flags.INTERLACE) {
+		/*the input timing is in VESA spec format with Interlace flag =1*/
+		hw_crtc_timing.v_total /= 2;
+		hw_crtc_timing.v_border_top /= 2;
+		hw_crtc_timing.v_addressable /= 2;
+		hw_crtc_timing.v_border_bottom /= 2;
+		hw_crtc_timing.v_front_porch /= 2;
+		hw_crtc_timing.v_sync_width /= 2;
+	}
+	/* set pixel encoding */
+	switch (hw_crtc_timing.pixel_encoding) {
+	case PIXEL_ENCODING_YCBCR422:
+		REG_UPDATE(DP_PIXEL_FORMAT, DP_PIXEL_ENCODING,
+				DP_PIXEL_ENCODING_TYPE_YCBCR422);
+		break;
+	case PIXEL_ENCODING_YCBCR444:
+		REG_UPDATE(DP_PIXEL_FORMAT, DP_PIXEL_ENCODING,
+				DP_PIXEL_ENCODING_TYPE_YCBCR444);
+
+		if (hw_crtc_timing.flags.Y_ONLY)
+			if (hw_crtc_timing.display_color_depth != COLOR_DEPTH_666)
+				/* HW testing only, no use case yet.
+				 * Color depth of Y-only could be
+				 * 8, 10, 12, 16 bits */
+				REG_UPDATE(DP_PIXEL_FORMAT, DP_PIXEL_ENCODING,
+						DP_PIXEL_ENCODING_TYPE_Y_ONLY);
+		/* Note: DP_MSA_MISC1 bit 7 is the indicator
+		 * of Y-only mode.
+		 * This bit is set in HW if register
+		 * DP_PIXEL_ENCODING is programmed to 0x4 */
+		break;
+	case PIXEL_ENCODING_YCBCR420:
+		REG_UPDATE(DP_PIXEL_FORMAT, DP_PIXEL_ENCODING,
+				DP_PIXEL_ENCODING_TYPE_YCBCR420);
+		if (enc110->se_mask->DP_VID_M_DOUBLE_VALUE_EN)
+			REG_UPDATE(DP_VID_TIMING, DP_VID_M_DOUBLE_VALUE_EN, 1);
+
+		if (enc110->se_mask->DP_VID_N_MUL)
+			REG_UPDATE(DP_VID_TIMING, DP_VID_N_MUL, 1);
+		break;
+	default:
+		REG_UPDATE(DP_PIXEL_FORMAT, DP_PIXEL_ENCODING,
+				DP_PIXEL_ENCODING_TYPE_RGB444);
+		break;
+	}
+
+	if (REG(DP_MSA_MISC))
+		misc1 = REG_READ(DP_MSA_MISC);
+
+	/* set color depth */
+
+	switch (hw_crtc_timing.display_color_depth) {
+	case COLOR_DEPTH_666:
+		REG_UPDATE(DP_PIXEL_FORMAT, DP_COMPONENT_DEPTH,
+				0);
+		break;
+	case COLOR_DEPTH_888:
+		REG_UPDATE(DP_PIXEL_FORMAT, DP_COMPONENT_DEPTH,
+				DP_COMPONENT_PIXEL_DEPTH_8BPC);
+		break;
+	case COLOR_DEPTH_101010:
+		REG_UPDATE(DP_PIXEL_FORMAT, DP_COMPONENT_DEPTH,
+				DP_COMPONENT_PIXEL_DEPTH_10BPC);
+
+		break;
+	case COLOR_DEPTH_121212:
+		REG_UPDATE(DP_PIXEL_FORMAT, DP_COMPONENT_DEPTH,
+				DP_COMPONENT_PIXEL_DEPTH_12BPC);
+		break;
+	default:
+		REG_UPDATE(DP_PIXEL_FORMAT, DP_COMPONENT_DEPTH,
+				DP_COMPONENT_PIXEL_DEPTH_6BPC);
+		break;
+	}
+
+	/* set dynamic range and YCbCr range */
+
+
+	switch (hw_crtc_timing.display_color_depth) {
+	case COLOR_DEPTH_666:
+		colorimetry_bpc = 0;
+		break;
+	case COLOR_DEPTH_888:
+		colorimetry_bpc = 1;
+		break;
+	case COLOR_DEPTH_101010:
+		colorimetry_bpc = 2;
+		break;
+	case COLOR_DEPTH_121212:
+		colorimetry_bpc = 3;
+		break;
+	default:
+		colorimetry_bpc = 0;
+		break;
+	}
+
+	misc0 = colorimetry_bpc << 5;
+
+	if (REG(DP_MSA_TIMING_PARAM1)) {
+		switch (output_color_space) {
+		case COLOR_SPACE_SRGB:
+			misc0 = misc0 | 0x0;
+			misc1 = misc1 & ~0x80; /* bit7 = 0*/
+			dynamic_range_rgb = 0; /*full range*/
+			break;
+		case COLOR_SPACE_SRGB_LIMITED:
+			misc0 = misc0 | 0x8; /* bit3=1 */
+			misc1 = misc1 & ~0x80; /* bit7 = 0*/
+			dynamic_range_rgb = 1; /*limited range*/
+			break;
+		case COLOR_SPACE_YCBCR601:
+		case COLOR_SPACE_YCBCR601_LIMITED:
+			misc0 = misc0 | 0x8; /* bit3=1, bit4=0 */
+			misc1 = misc1 & ~0x80; /* bit7 = 0*/
+			dynamic_range_ycbcr = 0; /*bt601*/
+			if (hw_crtc_timing.pixel_encoding == PIXEL_ENCODING_YCBCR422)
+				misc0 = misc0 | 0x2; /* bit2=0, bit1=1 */
+			else if (hw_crtc_timing.pixel_encoding == PIXEL_ENCODING_YCBCR444)
+				misc0 = misc0 | 0x4; /* bit2=1, bit1=0 */
+			break;
+		case COLOR_SPACE_YCBCR709:
+		case COLOR_SPACE_YCBCR709_LIMITED:
+		case COLOR_SPACE_YCBCR709_BLACK:
+			misc0 = misc0 | 0x18; /* bit3=1, bit4=1 */
+			misc1 = misc1 & ~0x80; /* bit7 = 0*/
+			dynamic_range_ycbcr = 1; /*bt709*/
+			if (hw_crtc_timing.pixel_encoding == PIXEL_ENCODING_YCBCR422)
+				misc0 = misc0 | 0x2; /* bit2=0, bit1=1 */
+			else if (hw_crtc_timing.pixel_encoding == PIXEL_ENCODING_YCBCR444)
+				misc0 = misc0 | 0x4; /* bit2=1, bit1=0 */
+			break;
+		case COLOR_SPACE_2020_RGB_LIMITEDRANGE:
+			dynamic_range_rgb = 1; /*limited range*/
+			break;
+		case COLOR_SPACE_2020_RGB_FULLRANGE:
+		case COLOR_SPACE_2020_YCBCR_LIMITED:
+		case COLOR_SPACE_XR_RGB:
+		case COLOR_SPACE_MSREF_SCRGB:
+		case COLOR_SPACE_ADOBERGB:
+		case COLOR_SPACE_DCIP3:
+		case COLOR_SPACE_XV_YCC_709:
+		case COLOR_SPACE_XV_YCC_601:
+		case COLOR_SPACE_DISPLAYNATIVE:
+		case COLOR_SPACE_DOLBYVISION:
+		case COLOR_SPACE_APPCTRL:
+		case COLOR_SPACE_CUSTOMPOINTS:
+		case COLOR_SPACE_UNKNOWN:
+		default:
+			/* do nothing */
+			break;
+		}
+		if (enc110->se_mask->DP_DYN_RANGE && enc110->se_mask->DP_YCBCR_RANGE)
+			REG_UPDATE_2(
+				DP_PIXEL_FORMAT,
+				DP_DYN_RANGE, dynamic_range_rgb,
+				DP_YCBCR_RANGE, dynamic_range_ycbcr);
+
+		if (REG(DP_MSA_COLORIMETRY))
+			REG_SET(DP_MSA_COLORIMETRY, 0, DP_MSA_MISC0, misc0);
+
+		if (REG(DP_MSA_MISC))
+			REG_WRITE(DP_MSA_MISC, misc1);   /* MSA_MISC1 */
+
+	/* dcn new register
+	 * dc_crtc_timing is vesa dmt struct. data from edid
+	 */
+		if (REG(DP_MSA_TIMING_PARAM1))
+			REG_SET_2(DP_MSA_TIMING_PARAM1, 0,
+					DP_MSA_HTOTAL, hw_crtc_timing.h_total,
+					DP_MSA_VTOTAL, hw_crtc_timing.v_total);
+
+		/* calcuate from vesa timing parameters
+		 * h_active_start related to leading edge of sync
+		 */
+
+		h_blank = hw_crtc_timing.h_total - hw_crtc_timing.h_border_left -
+				hw_crtc_timing.h_addressable - hw_crtc_timing.h_border_right;
+
+		h_back_porch = h_blank - hw_crtc_timing.h_front_porch -
+				hw_crtc_timing.h_sync_width;
+
+		/* start at begining of left border */
+		h_active_start = hw_crtc_timing.h_sync_width + h_back_porch;
+
+
+		v_active_start = hw_crtc_timing.v_total - hw_crtc_timing.v_border_top -
+				hw_crtc_timing.v_addressable - hw_crtc_timing.v_border_bottom -
+				hw_crtc_timing.v_front_porch;
+
+
+		/* start at begining of left border */
+		if (REG(DP_MSA_TIMING_PARAM2))
+			REG_SET_2(DP_MSA_TIMING_PARAM2, 0,
+				DP_MSA_HSTART, h_active_start,
+				DP_MSA_VSTART, v_active_start);
+
+		if (REG(DP_MSA_TIMING_PARAM3))
+			REG_SET_4(DP_MSA_TIMING_PARAM3, 0,
+					DP_MSA_HSYNCWIDTH,
+					hw_crtc_timing.h_sync_width,
+					DP_MSA_HSYNCPOLARITY,
+					!hw_crtc_timing.flags.HSYNC_POSITIVE_POLARITY,
+					DP_MSA_VSYNCWIDTH,
+					hw_crtc_timing.v_sync_width,
+					DP_MSA_VSYNCPOLARITY,
+					!hw_crtc_timing.flags.VSYNC_POSITIVE_POLARITY);
+
+		/* HWDITH include border or overscan */
+		if (REG(DP_MSA_TIMING_PARAM4))
+			REG_SET_2(DP_MSA_TIMING_PARAM4, 0,
+				DP_MSA_HWIDTH, hw_crtc_timing.h_border_left +
+				hw_crtc_timing.h_addressable + hw_crtc_timing.h_border_right,
+				DP_MSA_VHEIGHT, hw_crtc_timing.v_border_top +
+				hw_crtc_timing.v_addressable + hw_crtc_timing.v_border_bottom);
+	}
+}
+
+static void dce110_stream_encoder_set_stream_attribute_helper(
+		struct dce110_stream_encoder *enc110,
+		struct dc_crtc_timing *crtc_timing)
+{
+	if (enc110->regs->TMDS_CNTL) {
+		switch (crtc_timing->pixel_encoding) {
+		case PIXEL_ENCODING_YCBCR422:
+			REG_UPDATE(TMDS_CNTL, TMDS_PIXEL_ENCODING, 1);
+			break;
+		default:
+			REG_UPDATE(TMDS_CNTL, TMDS_PIXEL_ENCODING, 0);
+			break;
+		}
+		REG_UPDATE(TMDS_CNTL, TMDS_COLOR_FORMAT, 0);
+	} else if (enc110->regs->DIG_FE_CNTL) {
+		switch (crtc_timing->pixel_encoding) {
+		case PIXEL_ENCODING_YCBCR422:
+			REG_UPDATE(DIG_FE_CNTL, TMDS_PIXEL_ENCODING, 1);
+			break;
+		default:
+			REG_UPDATE(DIG_FE_CNTL, TMDS_PIXEL_ENCODING, 0);
+			break;
+		}
+		REG_UPDATE(DIG_FE_CNTL, TMDS_COLOR_FORMAT, 0);
+	}
+
+}
+
+/* setup stream encoder in hdmi mode */
+static void dce110_stream_encoder_hdmi_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	int actual_pix_clk_khz,
+	bool enable_audio)
+{
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+	struct bp_encoder_control cntl = {0};
+
+	cntl.action = ENCODER_CONTROL_SETUP;
+	cntl.engine_id = enc110->base.id;
+	cntl.signal = SIGNAL_TYPE_HDMI_TYPE_A;
+	cntl.enable_dp_audio = enable_audio;
+	cntl.pixel_clock = actual_pix_clk_khz;
+	cntl.lanes_number = LANE_COUNT_FOUR;
+	cntl.color_depth = crtc_timing->display_color_depth;
+
+	if (enc110->base.bp->funcs->encoder_control(
+			enc110->base.bp, &cntl) != BP_RESULT_OK)
+		return;
+
+	dce110_stream_encoder_set_stream_attribute_helper(enc110, crtc_timing);
+
+	/* setup HDMI engine */
+	if (!enc110->se_mask->HDMI_DATA_SCRAMBLE_EN) {
+		REG_UPDATE_3(HDMI_CONTROL,
+			HDMI_PACKET_GEN_VERSION, 1,
+			HDMI_KEEPOUT_MODE, 1,
+			HDMI_DEEP_COLOR_ENABLE, 0);
+	} else if (enc110->regs->DIG_FE_CNTL) {
+		REG_UPDATE_5(HDMI_CONTROL,
+			HDMI_PACKET_GEN_VERSION, 1,
+			HDMI_KEEPOUT_MODE, 1,
+			HDMI_DEEP_COLOR_ENABLE, 0,
+			HDMI_DATA_SCRAMBLE_EN, 0,
+			HDMI_CLOCK_CHANNEL_RATE, 0);
+	}
+
+	switch (crtc_timing->display_color_depth) {
+	case COLOR_DEPTH_888:
+		REG_UPDATE(HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 0);
+		break;
+	case COLOR_DEPTH_101010:
+		if (crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR422) {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 1,
+					HDMI_DEEP_COLOR_ENABLE, 0);
+		} else {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 1,
+					HDMI_DEEP_COLOR_ENABLE, 1);
+			}
+		break;
+	case COLOR_DEPTH_121212:
+		if (crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR422) {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 2,
+					HDMI_DEEP_COLOR_ENABLE, 0);
+		} else {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 2,
+					HDMI_DEEP_COLOR_ENABLE, 1);
+			}
+		break;
+	case COLOR_DEPTH_161616:
+		REG_UPDATE_2(HDMI_CONTROL,
+				HDMI_DEEP_COLOR_DEPTH, 3,
+				HDMI_DEEP_COLOR_ENABLE, 1);
+		break;
+	default:
+		break;
+	}
+
+	if (enc110->se_mask->HDMI_DATA_SCRAMBLE_EN) {
+		if (actual_pix_clk_khz >= HDMI_CLOCK_CHANNEL_RATE_MORE_340M) {
+			/* enable HDMI data scrambler
+			 * HDMI_CLOCK_CHANNEL_RATE_MORE_340M
+			 * Clock channel frequency is 1/4 of character rate.
+			 */
+			REG_UPDATE_2(HDMI_CONTROL,
+				HDMI_DATA_SCRAMBLE_EN, 1,
+				HDMI_CLOCK_CHANNEL_RATE, 1);
+		} else if (crtc_timing->flags.LTE_340MCSC_SCRAMBLE) {
+
+			/* TODO: New feature for DCE11, still need to implement */
+
+			/* enable HDMI data scrambler
+			 * HDMI_CLOCK_CHANNEL_FREQ_EQUAL_TO_CHAR_RATE
+			 * Clock channel frequency is the same
+			 * as character rate
+			 */
+			REG_UPDATE_2(HDMI_CONTROL,
+				HDMI_DATA_SCRAMBLE_EN, 1,
+				HDMI_CLOCK_CHANNEL_RATE, 0);
+		}
+	}
+
+	REG_UPDATE_3(HDMI_VBI_PACKET_CONTROL,
+		HDMI_GC_CONT, 1,
+		HDMI_GC_SEND, 1,
+		HDMI_NULL_SEND, 1);
+
+	REG_UPDATE(HDMI_VBI_PACKET_CONTROL, HDMI_ACP_SEND, 0);
+
+	/* following belongs to audio */
+	REG_UPDATE(HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, 1);
+
+	REG_UPDATE(AFMT_INFOFRAME_CONTROL0, AFMT_AUDIO_INFO_UPDATE, 1);
+
+	REG_UPDATE(HDMI_INFOFRAME_CONTROL1, HDMI_AUDIO_INFO_LINE,
+				VBI_LINE_0 + 2);
+
+	REG_UPDATE(HDMI_GC, HDMI_GC_AVMUTE, 0);
+
+}
+
+/* setup stream encoder in dvi mode */
+static void dce110_stream_encoder_dvi_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	bool is_dual_link)
+{
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+	struct bp_encoder_control cntl = {0};
+
+	cntl.action = ENCODER_CONTROL_SETUP;
+	cntl.engine_id = enc110->base.id;
+	cntl.signal = is_dual_link ?
+			SIGNAL_TYPE_DVI_DUAL_LINK : SIGNAL_TYPE_DVI_SINGLE_LINK;
+	cntl.enable_dp_audio = false;
+	cntl.pixel_clock = crtc_timing->pix_clk_100hz / 10;
+	cntl.lanes_number = (is_dual_link) ? LANE_COUNT_EIGHT : LANE_COUNT_FOUR;
+
+	if (enc110->base.bp->funcs->encoder_control(
+			enc110->base.bp, &cntl) != BP_RESULT_OK)
+		return;
+
+	ASSERT(crtc_timing->pixel_encoding == PIXEL_ENCODING_RGB);
+	ASSERT(crtc_timing->display_color_depth == COLOR_DEPTH_888);
+	dce110_stream_encoder_set_stream_attribute_helper(enc110, crtc_timing);
+}
+
+/* setup stream encoder in LVDS mode */
+static void dce110_stream_encoder_lvds_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing)
+{
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+	struct bp_encoder_control cntl = {0};
+
+	cntl.action = ENCODER_CONTROL_SETUP;
+	cntl.engine_id = enc110->base.id;
+	cntl.signal = SIGNAL_TYPE_LVDS;
+	cntl.enable_dp_audio = false;
+	cntl.pixel_clock = crtc_timing->pix_clk_100hz / 10;
+	cntl.lanes_number = LANE_COUNT_FOUR;
+
+	if (enc110->base.bp->funcs->encoder_control(
+			enc110->base.bp, &cntl) != BP_RESULT_OK)
+		return;
+
+	ASSERT(crtc_timing->pixel_encoding == PIXEL_ENCODING_RGB);
+}
+
+static void dce110_stream_encoder_set_throttled_vcp_size(
+	struct stream_encoder *enc,
+	struct fixed31_32 avg_time_slots_per_mtp)
+{
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+	uint32_t x = dc_fixpt_floor(
+		avg_time_slots_per_mtp);
+	uint32_t y = dc_fixpt_ceil(
+		dc_fixpt_shl(
+			dc_fixpt_sub_int(
+				avg_time_slots_per_mtp,
+				x),
+			26));
+
+	{
+		REG_SET_2(DP_MSE_RATE_CNTL, 0,
+			DP_MSE_RATE_X, x,
+			DP_MSE_RATE_Y, y);
+	}
+
+	/* wait for update to be completed on the link */
+	/* i.e. DP_MSE_RATE_UPDATE_PENDING field (read only) */
+	/* is reset to 0 (not pending) */
+	REG_WAIT(DP_MSE_RATE_UPDATE, DP_MSE_RATE_UPDATE_PENDING,
+			0,
+			10, DP_MST_UPDATE_MAX_RETRY);
+}
+
+static void dce110_stream_encoder_update_hdmi_info_packets(
+	struct stream_encoder *enc,
+	const struct encoder_info_frame *info_frame)
+{
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+
+	if (enc110->se_mask->HDMI_AVI_INFO_CONT &&
+			enc110->se_mask->HDMI_AVI_INFO_SEND) {
+
+		if (info_frame->avi.valid) {
+			const uint32_t *content =
+				(const uint32_t *) &info_frame->avi.sb[0];
+			/*we need turn on clock before programming AFMT block*/
+			if (REG(AFMT_CNTL))
+				REG_UPDATE(AFMT_CNTL, AFMT_AUDIO_CLOCK_EN, 1);
+
+			REG_WRITE(AFMT_AVI_INFO0, content[0]);
+
+			REG_WRITE(AFMT_AVI_INFO1, content[1]);
+
+			REG_WRITE(AFMT_AVI_INFO2, content[2]);
+
+			REG_WRITE(AFMT_AVI_INFO3, content[3]);
+
+			REG_UPDATE(AFMT_AVI_INFO3, AFMT_AVI_INFO_VERSION,
+						info_frame->avi.hb1);
+
+			REG_UPDATE_2(HDMI_INFOFRAME_CONTROL0,
+					HDMI_AVI_INFO_SEND, 1,
+					HDMI_AVI_INFO_CONT, 1);
+
+			REG_UPDATE(HDMI_INFOFRAME_CONTROL1, HDMI_AVI_INFO_LINE,
+							VBI_LINE_0 + 2);
+
+		} else {
+			REG_UPDATE_2(HDMI_INFOFRAME_CONTROL0,
+				HDMI_AVI_INFO_SEND, 0,
+				HDMI_AVI_INFO_CONT, 0);
+		}
+	}
+
+	if (enc110->se_mask->HDMI_AVI_INFO_CONT &&
+			enc110->se_mask->HDMI_AVI_INFO_SEND) {
+		dce110_update_hdmi_info_packet(enc110, 0, &info_frame->vendor);
+		dce110_update_hdmi_info_packet(enc110, 1, &info_frame->gamut);
+		dce110_update_hdmi_info_packet(enc110, 2, &info_frame->spd);
+		dce110_update_hdmi_info_packet(enc110, 3, &info_frame->hdrsmd);
+	}
+
+	if (enc110->se_mask->HDMI_DB_DISABLE) {
+		/* for bring up, disable dp double  TODO */
+		if (REG(HDMI_DB_CONTROL))
+			REG_UPDATE(HDMI_DB_CONTROL, HDMI_DB_DISABLE, 1);
+
+		dce110_update_hdmi_info_packet(enc110, 0, &info_frame->avi);
+		dce110_update_hdmi_info_packet(enc110, 1, &info_frame->vendor);
+		dce110_update_hdmi_info_packet(enc110, 2, &info_frame->gamut);
+		dce110_update_hdmi_info_packet(enc110, 3, &info_frame->spd);
+		dce110_update_hdmi_info_packet(enc110, 4, &info_frame->hdrsmd);
+	}
+}
+
+static void dce110_stream_encoder_stop_hdmi_info_packets(
+	struct stream_encoder *enc)
+{
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+
+	/* stop generic packets 0 & 1 on HDMI */
+	REG_SET_6(HDMI_GENERIC_PACKET_CONTROL0, 0,
+		HDMI_GENERIC1_CONT, 0,
+		HDMI_GENERIC1_LINE, 0,
+		HDMI_GENERIC1_SEND, 0,
+		HDMI_GENERIC0_CONT, 0,
+		HDMI_GENERIC0_LINE, 0,
+		HDMI_GENERIC0_SEND, 0);
+
+	/* stop generic packets 2 & 3 on HDMI */
+	REG_SET_6(HDMI_GENERIC_PACKET_CONTROL1, 0,
+		HDMI_GENERIC0_CONT, 0,
+		HDMI_GENERIC0_LINE, 0,
+		HDMI_GENERIC0_SEND, 0,
+		HDMI_GENERIC1_CONT, 0,
+		HDMI_GENERIC1_LINE, 0,
+		HDMI_GENERIC1_SEND, 0);
+
+	/* stop generic packets 2 & 3 on HDMI */
+	if (REG(HDMI_GENERIC_PACKET_CONTROL2))
+		REG_SET_6(HDMI_GENERIC_PACKET_CONTROL2, 0,
+			HDMI_GENERIC0_CONT, 0,
+			HDMI_GENERIC0_LINE, 0,
+			HDMI_GENERIC0_SEND, 0,
+			HDMI_GENERIC1_CONT, 0,
+			HDMI_GENERIC1_LINE, 0,
+			HDMI_GENERIC1_SEND, 0);
+
+	if (REG(HDMI_GENERIC_PACKET_CONTROL3))
+		REG_SET_6(HDMI_GENERIC_PACKET_CONTROL3, 0,
+			HDMI_GENERIC0_CONT, 0,
+			HDMI_GENERIC0_LINE, 0,
+			HDMI_GENERIC0_SEND, 0,
+			HDMI_GENERIC1_CONT, 0,
+			HDMI_GENERIC1_LINE, 0,
+			HDMI_GENERIC1_SEND, 0);
+}
+
+static void dce110_stream_encoder_update_dp_info_packets(
+	struct stream_encoder *enc,
+	const struct encoder_info_frame *info_frame)
+{
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+	uint32_t value = 0;
+
+	if (info_frame->vsc.valid)
+		dce110_update_generic_info_packet(
+					enc110,
+					0,  /* packetIndex */
+					&info_frame->vsc);
+
+	if (info_frame->spd.valid)
+		dce110_update_generic_info_packet(
+				enc110,
+				2,  /* packetIndex */
+				&info_frame->spd);
+
+	if (info_frame->hdrsmd.valid)
+		dce110_update_generic_info_packet(
+				enc110,
+				3,  /* packetIndex */
+				&info_frame->hdrsmd);
+
+	/* enable/disable transmission of packet(s).
+	*  If enabled, packet transmission begins on the next frame
+	*/
+	REG_UPDATE(DP_SEC_CNTL, DP_SEC_GSP0_ENABLE, info_frame->vsc.valid);
+	REG_UPDATE(DP_SEC_CNTL, DP_SEC_GSP2_ENABLE, info_frame->spd.valid);
+	REG_UPDATE(DP_SEC_CNTL, DP_SEC_GSP3_ENABLE, info_frame->hdrsmd.valid);
+
+	/* This bit is the master enable bit.
+	* When enabling secondary stream engine,
+	* this master bit must also be set.
+	* This register shared with audio info frame.
+	* Therefore we need to enable master bit
+	* if at least on of the fields is not 0
+	*/
+	value = REG_READ(DP_SEC_CNTL);
+	if (value)
+		REG_UPDATE(DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, 1);
+}
+
+static void dce110_stream_encoder_stop_dp_info_packets(
+	struct stream_encoder *enc)
+{
+	/* stop generic packets on DP */
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+	uint32_t value = 0;
+
+	if (enc110->se_mask->DP_SEC_AVI_ENABLE) {
+		REG_SET_7(DP_SEC_CNTL, 0,
+			DP_SEC_GSP0_ENABLE, 0,
+			DP_SEC_GSP1_ENABLE, 0,
+			DP_SEC_GSP2_ENABLE, 0,
+			DP_SEC_GSP3_ENABLE, 0,
+			DP_SEC_AVI_ENABLE, 0,
+			DP_SEC_MPG_ENABLE, 0,
+			DP_SEC_STREAM_ENABLE, 0);
+	}
+
+	/* this register shared with audio info frame.
+	 * therefore we need to keep master enabled
+	 * if at least one of the fields is not 0 */
+	value = REG_READ(DP_SEC_CNTL);
+	if (value)
+		REG_UPDATE(DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, 1);
+
+}
+
+static void dce110_stream_encoder_dp_blank(
+	struct dc_link *link,
+	struct stream_encoder *enc)
+{
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+	uint32_t  reg1 = 0;
+	uint32_t max_retries = DP_BLANK_MAX_RETRY * 10;
+
+	/* Note: For CZ, we are changing driver default to disable
+	 * stream deferred to next VBLANK. If results are positive, we
+	 * will make the same change to all DCE versions. There are a
+	 * handful of panels that cannot handle disable stream at
+	 * HBLANK and will result in a white line flash across the
+	 * screen on stream disable. */
+	REG_GET(DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, &reg1);
+	if ((reg1 & 0x1) == 0)
+		/*stream not enabled*/
+		return;
+	/* Specify the video stream disable point
+	 * (2 = start of the next vertical blank) */
+	REG_UPDATE(DP_VID_STREAM_CNTL, DP_VID_STREAM_DIS_DEFER, 2);
+	/* Larger delay to wait until VBLANK - use max retry of
+	 * 10us*3000=30ms. This covers 16.6ms of typical 60 Hz mode +
+	 * a little more because we may not trust delay accuracy.
+	 */
+	max_retries = DP_BLANK_MAX_RETRY * 150;
+
+	/* disable DP stream */
+	REG_UPDATE(DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, 0);
+
+	/* the encoder stops sending the video stream
+	 * at the start of the vertical blanking.
+	 * Poll for DP_VID_STREAM_STATUS == 0
+	 */
+
+	REG_WAIT(DP_VID_STREAM_CNTL, DP_VID_STREAM_STATUS,
+			0,
+			10, max_retries);
+
+	/* Tell the DP encoder to ignore timing from CRTC, must be done after
+	 * the polling. If we set DP_STEER_FIFO_RESET before DP stream blank is
+	 * complete, stream status will be stuck in video stream enabled state,
+	 * i.e. DP_VID_STREAM_STATUS stuck at 1.
+	 */
+
+	REG_UPDATE(DP_STEER_FIFO, DP_STEER_FIFO_RESET, true);
+}
+
+/* output video stream to link encoder */
+static void dce110_stream_encoder_dp_unblank(
+	struct dc_link *link,
+	struct stream_encoder *enc,
+	const struct encoder_unblank_param *param)
+{
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+
+	if (param->link_settings.link_rate != LINK_RATE_UNKNOWN) {
+		uint32_t n_vid = 0x8000;
+		uint32_t m_vid;
+
+		/* M / N = Fstream / Flink
+		* m_vid / n_vid = pixel rate / link rate
+		*/
+
+		uint64_t m_vid_l = n_vid;
+
+		m_vid_l *= param->timing.pix_clk_100hz / 10;
+		m_vid_l = div_u64(m_vid_l,
+			param->link_settings.link_rate
+				* LINK_RATE_REF_FREQ_IN_KHZ);
+
+		m_vid = (uint32_t) m_vid_l;
+
+		/* enable auto measurement */
+
+		REG_UPDATE(DP_VID_TIMING, DP_VID_M_N_GEN_EN, 0);
+
+		/* auto measurement need 1 full 0x8000 symbol cycle to kick in,
+		 * therefore program initial value for Mvid and Nvid
+		 */
+
+		REG_UPDATE(DP_VID_N, DP_VID_N, n_vid);
+
+		REG_UPDATE(DP_VID_M, DP_VID_M, m_vid);
+
+		REG_UPDATE(DP_VID_TIMING, DP_VID_M_N_GEN_EN, 1);
+	}
+
+	/* set DIG_START to 0x1 to resync FIFO */
+
+	REG_UPDATE(DIG_FE_CNTL, DIG_START, 1);
+
+	/* switch DP encoder to CRTC data */
+
+	REG_UPDATE(DP_STEER_FIFO, DP_STEER_FIFO_RESET, 0);
+
+	/* wait 100us for DIG/DP logic to prime
+	* (i.e. a few video lines)
+	*/
+	udelay(100);
+
+	/* the hardware would start sending video at the start of the next DP
+	* frame (i.e. rising edge of the vblank).
+	* NOTE: We used to program DP_VID_STREAM_DIS_DEFER = 2 here, but this
+	* register has no effect on enable transition! HW always guarantees
+	* VID_STREAM enable at start of next frame, and this is not
+	* programmable
+	*/
+
+	REG_UPDATE(DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, true);
+}
+
+static void dce110_stream_encoder_set_avmute(
+	struct stream_encoder *enc,
+	bool enable)
+{
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+	unsigned int value = enable ? 1 : 0;
+
+	REG_UPDATE(HDMI_GC, HDMI_GC_AVMUTE, value);
+}
+
+
+static void dce110_reset_hdmi_stream_attribute(
+	struct stream_encoder *enc)
+{
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+
+	if (enc110->se_mask->HDMI_DATA_SCRAMBLE_EN)
+		REG_UPDATE_5(HDMI_CONTROL,
+			HDMI_PACKET_GEN_VERSION, 1,
+			HDMI_KEEPOUT_MODE, 1,
+			HDMI_DEEP_COLOR_ENABLE, 0,
+			HDMI_DATA_SCRAMBLE_EN, 0,
+			HDMI_CLOCK_CHANNEL_RATE, 0);
+	else
+		REG_UPDATE_3(HDMI_CONTROL,
+			HDMI_PACKET_GEN_VERSION, 1,
+			HDMI_KEEPOUT_MODE, 1,
+			HDMI_DEEP_COLOR_ENABLE, 0);
+}
+
+#define DP_SEC_AUD_N__DP_SEC_AUD_N__DEFAULT 0x8000
+#define DP_SEC_TIMESTAMP__DP_SEC_TIMESTAMP_MODE__AUTO_CALC 1
+
+#include "include/audio_types.h"
+
+
+/* 25.2MHz/1.001*/
+/* 25.2MHz/1.001*/
+/* 25.2MHz*/
+/* 27MHz */
+/* 27MHz*1.001*/
+/* 27MHz*1.001*/
+/* 54MHz*/
+/* 54MHz*1.001*/
+/* 74.25MHz/1.001*/
+/* 74.25MHz*/
+/* 148.5MHz/1.001*/
+/* 148.5MHz*/
+
+static const struct audio_clock_info audio_clock_info_table[16] = {
+	{2517, 4576, 28125, 7007, 31250, 6864, 28125},
+	{2518, 4576, 28125, 7007, 31250, 6864, 28125},
+	{2520, 4096, 25200, 6272, 28000, 6144, 25200},
+	{2700, 4096, 27000, 6272, 30000, 6144, 27000},
+	{2702, 4096, 27027, 6272, 30030, 6144, 27027},
+	{2703, 4096, 27027, 6272, 30030, 6144, 27027},
+	{5400, 4096, 54000, 6272, 60000, 6144, 54000},
+	{5405, 4096, 54054, 6272, 60060, 6144, 54054},
+	{7417, 11648, 210937, 17836, 234375, 11648, 140625},
+	{7425, 4096, 74250, 6272, 82500, 6144, 74250},
+	{14835, 11648, 421875, 8918, 234375, 5824, 140625},
+	{14850, 4096, 148500, 6272, 165000, 6144, 148500},
+	{29670, 5824, 421875, 4459, 234375, 5824, 281250},
+	{29700, 3072, 222750, 4704, 247500, 5120, 247500},
+	{59340, 5824, 843750, 8918, 937500, 5824, 562500},
+	{59400, 3072, 445500, 9408, 990000, 6144, 594000}
+};
+
+static const struct audio_clock_info audio_clock_info_table_36bpc[14] = {
+	{2517,  9152,  84375,  7007,  48875,  9152,  56250},
+	{2518,  9152,  84375,  7007,  48875,  9152,  56250},
+	{2520,  4096,  37800,  6272,  42000,  6144,  37800},
+	{2700,  4096,  40500,  6272,  45000,  6144,  40500},
+	{2702,  8192,  81081,  6272,  45045,  8192,  54054},
+	{2703,  8192,  81081,  6272,  45045,  8192,  54054},
+	{5400,  4096,  81000,  6272,  90000,  6144,  81000},
+	{5405,  4096,  81081,  6272,  90090,  6144,  81081},
+	{7417, 11648, 316406, 17836, 351562, 11648, 210937},
+	{7425, 4096, 111375,  6272, 123750,  6144, 111375},
+	{14835, 11648, 632812, 17836, 703125, 11648, 421875},
+	{14850, 4096, 222750,  6272, 247500,  6144, 222750},
+	{29670, 5824, 632812,  8918, 703125,  5824, 421875},
+	{29700, 4096, 445500,  4704, 371250,  5120, 371250}
+};
+
+static const struct audio_clock_info audio_clock_info_table_48bpc[14] = {
+	{2517,  4576,  56250,  7007,  62500,  6864,  56250},
+	{2518,  4576,  56250,  7007,  62500,  6864,  56250},
+	{2520,  4096,  50400,  6272,  56000,  6144,  50400},
+	{2700,  4096,  54000,  6272,  60000,  6144,  54000},
+	{2702,  4096,  54054,  6267,  60060,  8192,  54054},
+	{2703,  4096,  54054,  6272,  60060,  8192,  54054},
+	{5400,  4096, 108000,  6272, 120000,  6144, 108000},
+	{5405,  4096, 108108,  6272, 120120,  6144, 108108},
+	{7417, 11648, 421875, 17836, 468750, 11648, 281250},
+	{7425,  4096, 148500,  6272, 165000,  6144, 148500},
+	{14835, 11648, 843750,  8918, 468750, 11648, 281250},
+	{14850, 4096, 297000,  6272, 330000,  6144, 297000},
+	{29670, 5824, 843750,  4459, 468750,  5824, 562500},
+	{29700, 3072, 445500,  4704, 495000,  5120, 495000}
+
+
+};
+
+static union audio_cea_channels speakers_to_channels(
+	struct audio_speaker_flags speaker_flags)
+{
+	union audio_cea_channels cea_channels = {0};
+
+	/* these are one to one */
+	cea_channels.channels.FL = speaker_flags.FL_FR;
+	cea_channels.channels.FR = speaker_flags.FL_FR;
+	cea_channels.channels.LFE = speaker_flags.LFE;
+	cea_channels.channels.FC = speaker_flags.FC;
+
+	/* if Rear Left and Right exist move RC speaker to channel 7
+	 * otherwise to channel 5
+	 */
+	if (speaker_flags.RL_RR) {
+		cea_channels.channels.RL_RC = speaker_flags.RL_RR;
+		cea_channels.channels.RR = speaker_flags.RL_RR;
+		cea_channels.channels.RC_RLC_FLC = speaker_flags.RC;
+	} else {
+		cea_channels.channels.RL_RC = speaker_flags.RC;
+	}
+
+	/* FRONT Left Right Center and REAR Left Right Center are exclusive */
+	if (speaker_flags.FLC_FRC) {
+		cea_channels.channels.RC_RLC_FLC = speaker_flags.FLC_FRC;
+		cea_channels.channels.RRC_FRC = speaker_flags.FLC_FRC;
+	} else {
+		cea_channels.channels.RC_RLC_FLC = speaker_flags.RLC_RRC;
+		cea_channels.channels.RRC_FRC = speaker_flags.RLC_RRC;
+	}
+
+	return cea_channels;
+}
+
+static uint32_t calc_max_audio_packets_per_line(
+	const struct audio_crtc_info *crtc_info)
+{
+	uint32_t max_packets_per_line;
+
+	max_packets_per_line =
+		crtc_info->h_total - crtc_info->h_active;
+
+	if (crtc_info->pixel_repetition)
+		max_packets_per_line *= crtc_info->pixel_repetition;
+
+	/* for other hdmi features */
+	max_packets_per_line -= 58;
+	/* for Control Period */
+	max_packets_per_line -= 16;
+	/* Number of Audio Packets per Line */
+	max_packets_per_line /= 32;
+
+	return max_packets_per_line;
+}
+
+static void get_audio_clock_info(
+	enum dc_color_depth color_depth,
+	uint32_t crtc_pixel_clock_100Hz,
+	uint32_t actual_pixel_clock_100Hz,
+	struct audio_clock_info *audio_clock_info)
+{
+	const struct audio_clock_info *clock_info;
+	uint32_t index;
+	uint32_t crtc_pixel_clock_in_10khz = crtc_pixel_clock_100Hz / 100;
+	uint32_t audio_array_size;
+
+	switch (color_depth) {
+	case COLOR_DEPTH_161616:
+		clock_info = audio_clock_info_table_48bpc;
+		audio_array_size = ARRAY_SIZE(
+				audio_clock_info_table_48bpc);
+		break;
+	case COLOR_DEPTH_121212:
+		clock_info = audio_clock_info_table_36bpc;
+		audio_array_size = ARRAY_SIZE(
+				audio_clock_info_table_36bpc);
+		break;
+	default:
+		clock_info = audio_clock_info_table;
+		audio_array_size = ARRAY_SIZE(
+				audio_clock_info_table);
+		break;
+	}
+
+	if (clock_info != NULL) {
+		/* search for exact pixel clock in table */
+		for (index = 0; index < audio_array_size; index++) {
+			if (clock_info[index].pixel_clock_in_10khz >
+				crtc_pixel_clock_in_10khz)
+				break;  /* not match */
+			else if (clock_info[index].pixel_clock_in_10khz ==
+					crtc_pixel_clock_in_10khz) {
+				/* match found */
+				*audio_clock_info = clock_info[index];
+				return;
+			}
+		}
+	}
+
+	/* not found */
+	if (actual_pixel_clock_100Hz == 0)
+		actual_pixel_clock_100Hz = crtc_pixel_clock_100Hz;
+
+	/* See HDMI spec  the table entry under
+	 *  pixel clock of "Other". */
+	audio_clock_info->pixel_clock_in_10khz =
+			actual_pixel_clock_100Hz / 100;
+	audio_clock_info->cts_32khz = actual_pixel_clock_100Hz / 10;
+	audio_clock_info->cts_44khz = actual_pixel_clock_100Hz / 10;
+	audio_clock_info->cts_48khz = actual_pixel_clock_100Hz / 10;
+
+	audio_clock_info->n_32khz = 4096;
+	audio_clock_info->n_44khz = 6272;
+	audio_clock_info->n_48khz = 6144;
+}
+
+static void dce110_se_audio_setup(
+	struct stream_encoder *enc,
+	unsigned int az_inst,
+	struct audio_info *audio_info)
+{
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+
+	uint32_t channels = 0;
+
+	ASSERT(audio_info);
+	if (audio_info == NULL)
+		/* This should not happen.it does so we don't get BSOD*/
+		return;
+
+	channels = speakers_to_channels(audio_info->flags.speaker_flags).all;
+
+	/* setup the audio stream source select (audio -> dig mapping) */
+	REG_SET(AFMT_AUDIO_SRC_CONTROL, 0, AFMT_AUDIO_SRC_SELECT, az_inst);
+
+	/* Channel allocation */
+	REG_UPDATE(AFMT_AUDIO_PACKET_CONTROL2, AFMT_AUDIO_CHANNEL_ENABLE, channels);
+}
+
+static void dce110_se_setup_hdmi_audio(
+	struct stream_encoder *enc,
+	const struct audio_crtc_info *crtc_info)
+{
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+
+	struct audio_clock_info audio_clock_info = {0};
+	uint32_t max_packets_per_line;
+
+	/* For now still do calculation, although this field is ignored when
+	above HDMI_PACKET_GEN_VERSION set to 1 */
+	max_packets_per_line = calc_max_audio_packets_per_line(crtc_info);
+
+	/* HDMI_AUDIO_PACKET_CONTROL */
+	REG_UPDATE_2(HDMI_AUDIO_PACKET_CONTROL,
+			HDMI_AUDIO_PACKETS_PER_LINE, max_packets_per_line,
+			HDMI_AUDIO_DELAY_EN, 1);
+
+	/* AFMT_AUDIO_PACKET_CONTROL */
+	REG_UPDATE(AFMT_AUDIO_PACKET_CONTROL, AFMT_60958_CS_UPDATE, 1);
+
+	/* AFMT_AUDIO_PACKET_CONTROL2 */
+	REG_UPDATE_2(AFMT_AUDIO_PACKET_CONTROL2,
+			AFMT_AUDIO_LAYOUT_OVRD, 0,
+			AFMT_60958_OSF_OVRD, 0);
+
+	/* HDMI_ACR_PACKET_CONTROL */
+	REG_UPDATE_3(HDMI_ACR_PACKET_CONTROL,
+			HDMI_ACR_AUTO_SEND, 1,
+			HDMI_ACR_SOURCE, 0,
+			HDMI_ACR_AUDIO_PRIORITY, 0);
+
+	/* Program audio clock sample/regeneration parameters */
+	get_audio_clock_info(crtc_info->color_depth,
+			     crtc_info->requested_pixel_clock_100Hz,
+			     crtc_info->calculated_pixel_clock_100Hz,
+			     &audio_clock_info);
+	DC_LOG_HW_AUDIO(
+			"\n%s:Input::requested_pixel_clock_100Hz = %d"	\
+			"calculated_pixel_clock_100Hz = %d \n", __func__,	\
+			crtc_info->requested_pixel_clock_100Hz,		\
+			crtc_info->calculated_pixel_clock_100Hz);
+
+	/* HDMI_ACR_32_0__HDMI_ACR_CTS_32_MASK */
+	REG_UPDATE(HDMI_ACR_32_0, HDMI_ACR_CTS_32, audio_clock_info.cts_32khz);
+
+	/* HDMI_ACR_32_1__HDMI_ACR_N_32_MASK */
+	REG_UPDATE(HDMI_ACR_32_1, HDMI_ACR_N_32, audio_clock_info.n_32khz);
+
+	/* HDMI_ACR_44_0__HDMI_ACR_CTS_44_MASK */
+	REG_UPDATE(HDMI_ACR_44_0, HDMI_ACR_CTS_44, audio_clock_info.cts_44khz);
+
+	/* HDMI_ACR_44_1__HDMI_ACR_N_44_MASK */
+	REG_UPDATE(HDMI_ACR_44_1, HDMI_ACR_N_44, audio_clock_info.n_44khz);
+
+	/* HDMI_ACR_48_0__HDMI_ACR_CTS_48_MASK */
+	REG_UPDATE(HDMI_ACR_48_0, HDMI_ACR_CTS_48, audio_clock_info.cts_48khz);
+
+	/* HDMI_ACR_48_1__HDMI_ACR_N_48_MASK */
+	REG_UPDATE(HDMI_ACR_48_1, HDMI_ACR_N_48, audio_clock_info.n_48khz);
+
+	/* Video driver cannot know in advance which sample rate will
+	   be used by HD Audio driver
+	   HDMI_ACR_PACKET_CONTROL__HDMI_ACR_N_MULTIPLE field is
+	   programmed below in interruppt callback */
+
+	/* AFMT_60958_0__AFMT_60958_CS_CHANNEL_NUMBER_L_MASK &
+	AFMT_60958_0__AFMT_60958_CS_CLOCK_ACCURACY_MASK */
+	REG_UPDATE_2(AFMT_60958_0,
+			AFMT_60958_CS_CHANNEL_NUMBER_L, 1,
+			AFMT_60958_CS_CLOCK_ACCURACY, 0);
+
+	/* AFMT_60958_1 AFMT_60958_CS_CHALNNEL_NUMBER_R */
+	REG_UPDATE(AFMT_60958_1, AFMT_60958_CS_CHANNEL_NUMBER_R, 2);
+
+	/*AFMT_60958_2 now keep this settings until
+	 *  Programming guide comes out*/
+	REG_UPDATE_6(AFMT_60958_2,
+			AFMT_60958_CS_CHANNEL_NUMBER_2, 3,
+			AFMT_60958_CS_CHANNEL_NUMBER_3, 4,
+			AFMT_60958_CS_CHANNEL_NUMBER_4, 5,
+			AFMT_60958_CS_CHANNEL_NUMBER_5, 6,
+			AFMT_60958_CS_CHANNEL_NUMBER_6, 7,
+			AFMT_60958_CS_CHANNEL_NUMBER_7, 8);
+}
+
+static void dce110_se_setup_dp_audio(
+	struct stream_encoder *enc)
+{
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+
+	/* --- DP Audio packet configurations --- */
+
+	/* ATP Configuration */
+	REG_SET(DP_SEC_AUD_N, 0,
+			DP_SEC_AUD_N, DP_SEC_AUD_N__DP_SEC_AUD_N__DEFAULT);
+
+	/* Async/auto-calc timestamp mode */
+	REG_SET(DP_SEC_TIMESTAMP, 0, DP_SEC_TIMESTAMP_MODE,
+			DP_SEC_TIMESTAMP__DP_SEC_TIMESTAMP_MODE__AUTO_CALC);
+
+	/* --- The following are the registers
+	 *  copied from the SetupHDMI --- */
+
+	/* AFMT_AUDIO_PACKET_CONTROL */
+	REG_UPDATE(AFMT_AUDIO_PACKET_CONTROL, AFMT_60958_CS_UPDATE, 1);
+
+	/* AFMT_AUDIO_PACKET_CONTROL2 */
+	/* Program the ATP and AIP next */
+	REG_UPDATE_2(AFMT_AUDIO_PACKET_CONTROL2,
+			AFMT_AUDIO_LAYOUT_OVRD, 0,
+			AFMT_60958_OSF_OVRD, 0);
+
+	/* AFMT_INFOFRAME_CONTROL0 */
+	REG_UPDATE(AFMT_INFOFRAME_CONTROL0, AFMT_AUDIO_INFO_UPDATE, 1);
+
+	/* AFMT_60958_0__AFMT_60958_CS_CLOCK_ACCURACY_MASK */
+	REG_UPDATE(AFMT_60958_0, AFMT_60958_CS_CLOCK_ACCURACY, 0);
+}
+
+static void dce110_se_enable_audio_clock(
+	struct stream_encoder *enc,
+	bool enable)
+{
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+
+	if (REG(AFMT_CNTL) == 0)
+		return;   /* DCE8/10 does not have this register */
+
+	REG_UPDATE(AFMT_CNTL, AFMT_AUDIO_CLOCK_EN, !!enable);
+
+	/* wait for AFMT clock to turn on,
+	 * expectation: this should complete in 1-2 reads
+	 *
+	 * REG_WAIT(AFMT_CNTL, AFMT_AUDIO_CLOCK_ON, !!enable, 1, 10);
+	 *
+	 * TODO: wait for clock_on does not work well. May need HW
+	 * program sequence. But audio seems work normally even without wait
+	 * for clock_on status change
+	 */
+}
+
+static void dce110_se_enable_dp_audio(
+	struct stream_encoder *enc)
+{
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+
+	/* Enable Audio packets */
+	REG_UPDATE(DP_SEC_CNTL, DP_SEC_ASP_ENABLE, 1);
+
+	/* Program the ATP and AIP next */
+	REG_UPDATE_2(DP_SEC_CNTL,
+			DP_SEC_ATP_ENABLE, 1,
+			DP_SEC_AIP_ENABLE, 1);
+
+	/* Program STREAM_ENABLE after all the other enables. */
+	REG_UPDATE(DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, 1);
+}
+
+static void dce110_se_disable_dp_audio(
+	struct stream_encoder *enc)
+{
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+	uint32_t value = 0;
+
+	/* Disable Audio packets */
+	REG_UPDATE_5(DP_SEC_CNTL,
+			DP_SEC_ASP_ENABLE, 0,
+			DP_SEC_ATP_ENABLE, 0,
+			DP_SEC_AIP_ENABLE, 0,
+			DP_SEC_ACM_ENABLE, 0,
+			DP_SEC_STREAM_ENABLE, 0);
+
+	/* This register shared with encoder info frame. Therefore we need to
+	keep master enabled if at least on of the fields is not 0 */
+	value = REG_READ(DP_SEC_CNTL);
+	if (value != 0)
+		REG_UPDATE(DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, 1);
+
+}
+
+void dce110_se_audio_mute_control(
+	struct stream_encoder *enc,
+	bool mute)
+{
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+
+	REG_UPDATE(AFMT_AUDIO_PACKET_CONTROL, AFMT_AUDIO_SAMPLE_SEND, !mute);
+}
+
+void dce110_se_dp_audio_setup(
+	struct stream_encoder *enc,
+	unsigned int az_inst,
+	struct audio_info *info)
+{
+	dce110_se_audio_setup(enc, az_inst, info);
+}
+
+void dce110_se_dp_audio_enable(
+	struct stream_encoder *enc)
+{
+	dce110_se_enable_audio_clock(enc, true);
+	dce110_se_setup_dp_audio(enc);
+	dce110_se_enable_dp_audio(enc);
+}
+
+void dce110_se_dp_audio_disable(
+	struct stream_encoder *enc)
+{
+	dce110_se_disable_dp_audio(enc);
+	dce110_se_enable_audio_clock(enc, false);
+}
+
+void dce110_se_hdmi_audio_setup(
+	struct stream_encoder *enc,
+	unsigned int az_inst,
+	struct audio_info *info,
+	struct audio_crtc_info *audio_crtc_info)
+{
+	dce110_se_enable_audio_clock(enc, true);
+	dce110_se_setup_hdmi_audio(enc, audio_crtc_info);
+	dce110_se_audio_setup(enc, az_inst, info);
+}
+
+void dce110_se_hdmi_audio_disable(
+	struct stream_encoder *enc)
+{
+	dce110_se_enable_audio_clock(enc, false);
+}
+
+
+static void setup_stereo_sync(
+	struct stream_encoder *enc,
+	int tg_inst, bool enable)
+{
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+	REG_UPDATE(DIG_FE_CNTL, DIG_STEREOSYNC_SELECT, tg_inst);
+	REG_UPDATE(DIG_FE_CNTL, DIG_STEREOSYNC_GATE_EN, !enable);
+}
+
+static void dig_connect_to_otg(
+	struct stream_encoder *enc,
+	int tg_inst)
+{
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+
+	REG_UPDATE(DIG_FE_CNTL, DIG_SOURCE_SELECT, tg_inst);
+}
+
+static unsigned int dig_source_otg(
+	struct stream_encoder *enc)
+{
+	uint32_t tg_inst = 0;
+	struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc);
+
+	REG_GET(DIG_FE_CNTL, DIG_SOURCE_SELECT, &tg_inst);
+
+	return tg_inst;
+}
+
+static const struct stream_encoder_funcs dce110_str_enc_funcs = {
+	.dp_set_stream_attribute =
+		dce110_stream_encoder_dp_set_stream_attribute,
+	.hdmi_set_stream_attribute =
+		dce110_stream_encoder_hdmi_set_stream_attribute,
+	.dvi_set_stream_attribute =
+		dce110_stream_encoder_dvi_set_stream_attribute,
+	.lvds_set_stream_attribute =
+		dce110_stream_encoder_lvds_set_stream_attribute,
+	.set_throttled_vcp_size =
+		dce110_stream_encoder_set_throttled_vcp_size,
+	.update_hdmi_info_packets =
+		dce110_stream_encoder_update_hdmi_info_packets,
+	.stop_hdmi_info_packets =
+		dce110_stream_encoder_stop_hdmi_info_packets,
+	.update_dp_info_packets =
+		dce110_stream_encoder_update_dp_info_packets,
+	.stop_dp_info_packets =
+		dce110_stream_encoder_stop_dp_info_packets,
+	.dp_blank =
+		dce110_stream_encoder_dp_blank,
+	.dp_unblank =
+		dce110_stream_encoder_dp_unblank,
+	.audio_mute_control = dce110_se_audio_mute_control,
+
+	.dp_audio_setup = dce110_se_dp_audio_setup,
+	.dp_audio_enable = dce110_se_dp_audio_enable,
+	.dp_audio_disable = dce110_se_dp_audio_disable,
+
+	.hdmi_audio_setup = dce110_se_hdmi_audio_setup,
+	.hdmi_audio_disable = dce110_se_hdmi_audio_disable,
+	.setup_stereo_sync  = setup_stereo_sync,
+	.set_avmute = dce110_stream_encoder_set_avmute,
+	.dig_connect_to_otg  = dig_connect_to_otg,
+	.hdmi_reset_stream_attribute = dce110_reset_hdmi_stream_attribute,
+	.dig_source_otg = dig_source_otg,
+};
+
+void dce110_stream_encoder_construct(
+	struct dce110_stream_encoder *enc110,
+	struct dc_context *ctx,
+	struct dc_bios *bp,
+	enum engine_id eng_id,
+	const struct dce110_stream_enc_registers *regs,
+	const struct dce_stream_encoder_shift *se_shift,
+	const struct dce_stream_encoder_mask *se_mask)
+{
+	enc110->base.funcs = &dce110_str_enc_funcs;
+	enc110->base.ctx = ctx;
+	enc110->base.id = eng_id;
+	enc110->base.bp = bp;
+	enc110->regs = regs;
+	enc110->se_shift = se_shift;
+	enc110->se_mask = se_mask;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.h b/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.h
new file mode 100644
index 000000000000..cc5020a8e1e1
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.h
@@ -0,0 +1,735 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_STREAM_ENCODER_DCE110_H__
+#define __DC_STREAM_ENCODER_DCE110_H__
+
+#include "stream_encoder.h"
+
+#define DCE110STRENC_FROM_STRENC(stream_encoder)\
+	container_of(stream_encoder, struct dce110_stream_encoder, base)
+
+#ifndef TMDS_CNTL__TMDS_PIXEL_ENCODING_MASK
+	#define TMDS_CNTL__TMDS_PIXEL_ENCODING_MASK       0x00000010L
+	#define TMDS_CNTL__TMDS_COLOR_FORMAT_MASK         0x00000300L
+	#define	TMDS_CNTL__TMDS_PIXEL_ENCODING__SHIFT     0x00000004
+	#define	TMDS_CNTL__TMDS_COLOR_FORMAT__SHIFT       0x00000008
+#endif
+
+
+#define SE_COMMON_REG_LIST_DCE_BASE(id) \
+	SE_COMMON_REG_LIST_BASE(id),\
+	SRI(AFMT_AVI_INFO0, DIG, id), \
+	SRI(AFMT_AVI_INFO1, DIG, id), \
+	SRI(AFMT_AVI_INFO2, DIG, id), \
+	SRI(AFMT_AVI_INFO3, DIG, id)
+
+#define SE_COMMON_REG_LIST_BASE(id) \
+	SRI(AFMT_GENERIC_0, DIG, id), \
+	SRI(AFMT_GENERIC_1, DIG, id), \
+	SRI(AFMT_GENERIC_2, DIG, id), \
+	SRI(AFMT_GENERIC_3, DIG, id), \
+	SRI(AFMT_GENERIC_4, DIG, id), \
+	SRI(AFMT_GENERIC_5, DIG, id), \
+	SRI(AFMT_GENERIC_6, DIG, id), \
+	SRI(AFMT_GENERIC_7, DIG, id), \
+	SRI(AFMT_GENERIC_HDR, DIG, id), \
+	SRI(AFMT_INFOFRAME_CONTROL0, DIG, id), \
+	SRI(AFMT_VBI_PACKET_CONTROL, DIG, id), \
+	SRI(AFMT_AUDIO_PACKET_CONTROL, DIG, id), \
+	SRI(AFMT_AUDIO_PACKET_CONTROL2, DIG, id), \
+	SRI(AFMT_AUDIO_SRC_CONTROL, DIG, id), \
+	SRI(AFMT_60958_0, DIG, id), \
+	SRI(AFMT_60958_1, DIG, id), \
+	SRI(AFMT_60958_2, DIG, id), \
+	SRI(DIG_FE_CNTL, DIG, id), \
+	SRI(HDMI_CONTROL, DIG, id), \
+	SRI(HDMI_GC, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL0, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL1, DIG, id), \
+	SRI(HDMI_INFOFRAME_CONTROL0, DIG, id), \
+	SRI(HDMI_INFOFRAME_CONTROL1, DIG, id), \
+	SRI(HDMI_VBI_PACKET_CONTROL, DIG, id), \
+	SRI(HDMI_AUDIO_PACKET_CONTROL, DIG, id),\
+	SRI(HDMI_ACR_PACKET_CONTROL, DIG, id),\
+	SRI(HDMI_ACR_32_0, DIG, id),\
+	SRI(HDMI_ACR_32_1, DIG, id),\
+	SRI(HDMI_ACR_44_0, DIG, id),\
+	SRI(HDMI_ACR_44_1, DIG, id),\
+	SRI(HDMI_ACR_48_0, DIG, id),\
+	SRI(HDMI_ACR_48_1, DIG, id),\
+	SRI(TMDS_CNTL, DIG, id), \
+	SRI(DP_MSE_RATE_CNTL, DP, id), \
+	SRI(DP_MSE_RATE_UPDATE, DP, id), \
+	SRI(DP_PIXEL_FORMAT, DP, id), \
+	SRI(DP_SEC_CNTL, DP, id), \
+	SRI(DP_STEER_FIFO, DP, id), \
+	SRI(DP_VID_M, DP, id), \
+	SRI(DP_VID_N, DP, id), \
+	SRI(DP_VID_STREAM_CNTL, DP, id), \
+	SRI(DP_VID_TIMING, DP, id), \
+	SRI(DP_SEC_AUD_N, DP, id), \
+	SRI(DP_SEC_TIMESTAMP, DP, id)
+
+#define SE_COMMON_REG_LIST(id)\
+	SE_COMMON_REG_LIST_DCE_BASE(id), \
+	SRI(AFMT_CNTL, DIG, id)
+
+#define SE_DCN_REG_LIST(id)\
+	SE_COMMON_REG_LIST_BASE(id),\
+	SRI(AFMT_CNTL, DIG, id),\
+	SRI(AFMT_VBI_PACKET_CONTROL1, DIG, id),\
+	SRI(HDMI_GENERIC_PACKET_CONTROL2, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL3, DIG, id), \
+	SRI(DP_DB_CNTL, DP, id), \
+	SRI(DP_MSA_MISC, DP, id), \
+	SRI(DP_MSA_COLORIMETRY, DP, id), \
+	SRI(DP_MSA_TIMING_PARAM1, DP, id), \
+	SRI(DP_MSA_TIMING_PARAM2, DP, id), \
+	SRI(DP_MSA_TIMING_PARAM3, DP, id), \
+	SRI(DP_MSA_TIMING_PARAM4, DP, id), \
+	SRI(HDMI_DB_CONTROL, DIG, id)
+
+#define SE_SF(reg_name, field_name, post_fix)\
+	.field_name = reg_name ## __ ## field_name ## post_fix
+
+#define SE_COMMON_MASK_SH_LIST_DCE_COMMON(mask_sh)\
+	SE_SF(AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_INDEX, mask_sh),\
+	SE_SF(AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC0_UPDATE, mask_sh),\
+	SE_SF(AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC2_UPDATE, mask_sh),\
+	SE_SF(AFMT_GENERIC_HDR, AFMT_GENERIC_HB0, mask_sh),\
+	SE_SF(AFMT_GENERIC_HDR, AFMT_GENERIC_HB1, mask_sh),\
+	SE_SF(AFMT_GENERIC_HDR, AFMT_GENERIC_HB2, mask_sh),\
+	SE_SF(AFMT_GENERIC_HDR, AFMT_GENERIC_HB3, mask_sh),\
+	SE_SF(HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC0_CONT, mask_sh),\
+	SE_SF(HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC0_SEND, mask_sh),\
+	SE_SF(HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC0_LINE, mask_sh),\
+	SE_SF(HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC1_CONT, mask_sh),\
+	SE_SF(HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC1_SEND, mask_sh),\
+	SE_SF(HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC1_LINE, mask_sh),\
+	SE_SF(DP_PIXEL_FORMAT, DP_PIXEL_ENCODING, mask_sh),\
+	SE_SF(DP_PIXEL_FORMAT, DP_COMPONENT_DEPTH, mask_sh),\
+	SE_SF(DP_PIXEL_FORMAT, DP_DYN_RANGE, mask_sh),\
+	SE_SF(DP_PIXEL_FORMAT, DP_YCBCR_RANGE, mask_sh),\
+	SE_SF(HDMI_CONTROL, HDMI_PACKET_GEN_VERSION, mask_sh),\
+	SE_SF(HDMI_CONTROL, HDMI_KEEPOUT_MODE, mask_sh),\
+	SE_SF(HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, mask_sh),\
+	SE_SF(HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, mask_sh),\
+	SE_SF(HDMI_VBI_PACKET_CONTROL, HDMI_GC_CONT, mask_sh),\
+	SE_SF(HDMI_VBI_PACKET_CONTROL, HDMI_GC_SEND, mask_sh),\
+	SE_SF(HDMI_VBI_PACKET_CONTROL, HDMI_NULL_SEND, mask_sh),\
+	SE_SF(HDMI_VBI_PACKET_CONTROL, HDMI_ACP_SEND, mask_sh),\
+	SE_SF(HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, mask_sh),\
+	SE_SF(AFMT_INFOFRAME_CONTROL0, AFMT_AUDIO_INFO_UPDATE, mask_sh),\
+	SE_SF(HDMI_INFOFRAME_CONTROL1, HDMI_AUDIO_INFO_LINE, mask_sh),\
+	SE_SF(HDMI_GC, HDMI_GC_AVMUTE, mask_sh),\
+	SE_SF(DP_MSE_RATE_CNTL, DP_MSE_RATE_X, mask_sh),\
+	SE_SF(DP_MSE_RATE_CNTL, DP_MSE_RATE_Y, mask_sh),\
+	SE_SF(DP_MSE_RATE_UPDATE, DP_MSE_RATE_UPDATE_PENDING, mask_sh),\
+	SE_SF(AFMT_AVI_INFO3, AFMT_AVI_INFO_VERSION, mask_sh),\
+	SE_SF(HDMI_INFOFRAME_CONTROL0, HDMI_AVI_INFO_SEND, mask_sh),\
+	SE_SF(HDMI_INFOFRAME_CONTROL0, HDMI_AVI_INFO_CONT, mask_sh),\
+	SE_SF(HDMI_INFOFRAME_CONTROL1, HDMI_AVI_INFO_LINE, mask_sh),\
+	SE_SF(DP_SEC_CNTL, DP_SEC_GSP0_ENABLE, mask_sh),\
+	SE_SF(DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, mask_sh),\
+	SE_SF(DP_SEC_CNTL, DP_SEC_GSP1_ENABLE, mask_sh),\
+	SE_SF(DP_SEC_CNTL, DP_SEC_GSP2_ENABLE, mask_sh),\
+	SE_SF(DP_SEC_CNTL, DP_SEC_GSP3_ENABLE, mask_sh),\
+	SE_SF(DP_SEC_CNTL, DP_SEC_AVI_ENABLE, mask_sh),\
+	SE_SF(DP_SEC_CNTL, DP_SEC_MPG_ENABLE, mask_sh),\
+	SE_SF(DP_VID_STREAM_CNTL, DP_VID_STREAM_DIS_DEFER, mask_sh),\
+	SE_SF(DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, mask_sh),\
+	SE_SF(DP_VID_STREAM_CNTL, DP_VID_STREAM_STATUS, mask_sh),\
+	SE_SF(DP_STEER_FIFO, DP_STEER_FIFO_RESET, mask_sh),\
+	SE_SF(DP_VID_TIMING, DP_VID_M_N_GEN_EN, mask_sh),\
+	SE_SF(DP_VID_N, DP_VID_N, mask_sh),\
+	SE_SF(DP_VID_M, DP_VID_M, mask_sh),\
+	SE_SF(DIG_FE_CNTL, DIG_START, mask_sh),\
+	SE_SF(DIG_FE_CNTL, DIG_STEREOSYNC_SELECT, mask_sh),\
+	SE_SF(DIG_FE_CNTL, DIG_STEREOSYNC_GATE_EN, mask_sh),\
+	SE_SF(AFMT_AUDIO_SRC_CONTROL, AFMT_AUDIO_SRC_SELECT, mask_sh),\
+	SE_SF(AFMT_AUDIO_PACKET_CONTROL2, AFMT_AUDIO_CHANNEL_ENABLE, mask_sh),\
+	SE_SF(HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_PACKETS_PER_LINE, mask_sh),\
+	SE_SF(HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_DELAY_EN, mask_sh),\
+	SE_SF(AFMT_AUDIO_PACKET_CONTROL, AFMT_60958_CS_UPDATE, mask_sh),\
+	SE_SF(AFMT_AUDIO_PACKET_CONTROL2, AFMT_AUDIO_LAYOUT_OVRD, mask_sh),\
+	SE_SF(AFMT_AUDIO_PACKET_CONTROL2, AFMT_60958_OSF_OVRD, mask_sh),\
+	SE_SF(HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUTO_SEND, mask_sh),\
+	SE_SF(HDMI_ACR_PACKET_CONTROL, HDMI_ACR_SOURCE, mask_sh),\
+	SE_SF(HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUDIO_PRIORITY, mask_sh),\
+	SE_SF(HDMI_ACR_32_0, HDMI_ACR_CTS_32, mask_sh),\
+	SE_SF(HDMI_ACR_32_1, HDMI_ACR_N_32, mask_sh),\
+	SE_SF(HDMI_ACR_44_0, HDMI_ACR_CTS_44, mask_sh),\
+	SE_SF(HDMI_ACR_44_1, HDMI_ACR_N_44, mask_sh),\
+	SE_SF(HDMI_ACR_48_0, HDMI_ACR_CTS_48, mask_sh),\
+	SE_SF(HDMI_ACR_48_1, HDMI_ACR_N_48, mask_sh),\
+	SE_SF(AFMT_60958_0, AFMT_60958_CS_CHANNEL_NUMBER_L, mask_sh),\
+	SE_SF(AFMT_60958_0, AFMT_60958_CS_CLOCK_ACCURACY, mask_sh),\
+	SE_SF(AFMT_60958_1, AFMT_60958_CS_CHANNEL_NUMBER_R, mask_sh),\
+	SE_SF(AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_2, mask_sh),\
+	SE_SF(AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_3, mask_sh),\
+	SE_SF(AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_4, mask_sh),\
+	SE_SF(AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_5, mask_sh),\
+	SE_SF(AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_6, mask_sh),\
+	SE_SF(AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_7, mask_sh),\
+	SE_SF(DP_SEC_AUD_N, DP_SEC_AUD_N, mask_sh),\
+	SE_SF(DP_SEC_TIMESTAMP, DP_SEC_TIMESTAMP_MODE, mask_sh),\
+	SE_SF(DP_SEC_CNTL, DP_SEC_ASP_ENABLE, mask_sh),\
+	SE_SF(DP_SEC_CNTL, DP_SEC_ATP_ENABLE, mask_sh),\
+	SE_SF(DP_SEC_CNTL, DP_SEC_AIP_ENABLE, mask_sh),\
+	SE_SF(DP_SEC_CNTL, DP_SEC_ACM_ENABLE, mask_sh),\
+	SE_SF(AFMT_AUDIO_PACKET_CONTROL, AFMT_AUDIO_SAMPLE_SEND, mask_sh),\
+	SE_SF(DIG_FE_CNTL, DIG_SOURCE_SELECT, mask_sh)
+
+#define SE_COMMON_MASK_SH_LIST_SOC(mask_sh)\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_INDEX, mask_sh),\
+	SE_SF(DIG0_AFMT_GENERIC_HDR, AFMT_GENERIC_HB0, mask_sh),\
+	SE_SF(DIG0_AFMT_GENERIC_HDR, AFMT_GENERIC_HB1, mask_sh),\
+	SE_SF(DIG0_AFMT_GENERIC_HDR, AFMT_GENERIC_HB2, mask_sh),\
+	SE_SF(DIG0_AFMT_GENERIC_HDR, AFMT_GENERIC_HB3, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC0_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC0_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC0_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC1_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC1_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC1_LINE, mask_sh),\
+	SE_SF(DP0_DP_PIXEL_FORMAT, DP_PIXEL_ENCODING, mask_sh),\
+	SE_SF(DP0_DP_PIXEL_FORMAT, DP_COMPONENT_DEPTH, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_PACKET_GEN_VERSION, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_KEEPOUT_MODE, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_DATA_SCRAMBLE_EN, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_GC_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_GC_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_NULL_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_ACP_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, mask_sh),\
+	SE_SF(DIG0_AFMT_INFOFRAME_CONTROL0, AFMT_AUDIO_INFO_UPDATE, mask_sh),\
+	SE_SF(DIG0_HDMI_INFOFRAME_CONTROL1, HDMI_AUDIO_INFO_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GC, HDMI_GC_AVMUTE, mask_sh),\
+	SE_SF(DP0_DP_MSE_RATE_CNTL, DP_MSE_RATE_X, mask_sh),\
+	SE_SF(DP0_DP_MSE_RATE_CNTL, DP_MSE_RATE_Y, mask_sh),\
+	SE_SF(DP0_DP_MSE_RATE_UPDATE, DP_MSE_RATE_UPDATE_PENDING, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP0_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP1_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP2_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP3_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_MPG_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_DIS_DEFER, mask_sh),\
+	SE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_STATUS, mask_sh),\
+	SE_SF(DP0_DP_STEER_FIFO, DP_STEER_FIFO_RESET, mask_sh),\
+	SE_SF(DP0_DP_VID_TIMING, DP_VID_M_N_GEN_EN, mask_sh),\
+	SE_SF(DP0_DP_VID_N, DP_VID_N, mask_sh),\
+	SE_SF(DP0_DP_VID_M, DP_VID_M, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_START, mask_sh),\
+	SE_SF(DIG0_AFMT_AUDIO_SRC_CONTROL, AFMT_AUDIO_SRC_SELECT, mask_sh),\
+	SE_SF(DIG0_AFMT_AUDIO_PACKET_CONTROL2, AFMT_AUDIO_CHANNEL_ENABLE, mask_sh),\
+	SE_SF(DIG0_HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_PACKETS_PER_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_DELAY_EN, mask_sh),\
+	SE_SF(DIG0_AFMT_AUDIO_PACKET_CONTROL, AFMT_60958_CS_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_AUDIO_PACKET_CONTROL2, AFMT_AUDIO_LAYOUT_OVRD, mask_sh),\
+	SE_SF(DIG0_AFMT_AUDIO_PACKET_CONTROL2, AFMT_60958_OSF_OVRD, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUTO_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_PACKET_CONTROL, HDMI_ACR_SOURCE, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUDIO_PRIORITY, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_32_0, HDMI_ACR_CTS_32, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_32_1, HDMI_ACR_N_32, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_44_0, HDMI_ACR_CTS_44, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_44_1, HDMI_ACR_N_44, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_48_0, HDMI_ACR_CTS_48, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_48_1, HDMI_ACR_N_48, mask_sh),\
+	SE_SF(DIG0_AFMT_60958_0, AFMT_60958_CS_CHANNEL_NUMBER_L, mask_sh),\
+	SE_SF(DIG0_AFMT_60958_0, AFMT_60958_CS_CLOCK_ACCURACY, mask_sh),\
+	SE_SF(DIG0_AFMT_60958_1, AFMT_60958_CS_CHANNEL_NUMBER_R, mask_sh),\
+	SE_SF(DIG0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_2, mask_sh),\
+	SE_SF(DIG0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_3, mask_sh),\
+	SE_SF(DIG0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_4, mask_sh),\
+	SE_SF(DIG0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_5, mask_sh),\
+	SE_SF(DIG0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_6, mask_sh),\
+	SE_SF(DIG0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_7, mask_sh),\
+	SE_SF(DP0_DP_SEC_AUD_N, DP_SEC_AUD_N, mask_sh),\
+	SE_SF(DP0_DP_SEC_TIMESTAMP, DP_SEC_TIMESTAMP_MODE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_ASP_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_ATP_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_AIP_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_ACM_ENABLE, mask_sh),\
+	SE_SF(DIG0_AFMT_AUDIO_PACKET_CONTROL, AFMT_AUDIO_SAMPLE_SEND, mask_sh),\
+	SE_SF(DIG0_AFMT_CNTL, AFMT_AUDIO_CLOCK_EN, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_CLOCK_CHANNEL_RATE, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, TMDS_PIXEL_ENCODING, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, TMDS_COLOR_FORMAT, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_STEREOSYNC_SELECT, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_STEREOSYNC_GATE_EN, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_SOURCE_SELECT, mask_sh)
+
+#define SE_COMMON_MASK_SH_LIST_DCE80_100(mask_sh)\
+	SE_COMMON_MASK_SH_LIST_DCE_COMMON(mask_sh),\
+	SE_SF(TMDS_CNTL, TMDS_PIXEL_ENCODING, mask_sh),\
+	SE_SF(TMDS_CNTL, TMDS_COLOR_FORMAT, mask_sh)
+
+#define SE_COMMON_MASK_SH_LIST_DCE110(mask_sh)\
+	SE_COMMON_MASK_SH_LIST_DCE_COMMON(mask_sh),\
+	SE_SF(AFMT_CNTL, AFMT_AUDIO_CLOCK_EN, mask_sh),\
+	SE_SF(HDMI_CONTROL, HDMI_CLOCK_CHANNEL_RATE, mask_sh),\
+	SE_SF(HDMI_CONTROL, HDMI_DATA_SCRAMBLE_EN, mask_sh),\
+	SE_SF(DIG_FE_CNTL, TMDS_PIXEL_ENCODING, mask_sh),\
+	SE_SF(DIG_FE_CNTL, TMDS_COLOR_FORMAT, mask_sh),\
+	SE_SF(DIG_FE_CNTL, DIG_STEREOSYNC_SELECT, mask_sh),\
+	SE_SF(DIG_FE_CNTL, DIG_STEREOSYNC_GATE_EN, mask_sh)
+
+#define SE_COMMON_MASK_SH_LIST_DCE112(mask_sh)\
+	SE_COMMON_MASK_SH_LIST_DCE_COMMON(mask_sh),\
+	SE_SF(AFMT_CNTL, AFMT_AUDIO_CLOCK_EN, mask_sh),\
+	SE_SF(HDMI_CONTROL, HDMI_CLOCK_CHANNEL_RATE, mask_sh),\
+	SE_SF(HDMI_CONTROL, HDMI_DATA_SCRAMBLE_EN, mask_sh),\
+	SE_SF(DIG_FE_CNTL, TMDS_PIXEL_ENCODING, mask_sh),\
+	SE_SF(DIG_FE_CNTL, TMDS_COLOR_FORMAT, mask_sh),\
+	SE_SF(DP_VID_TIMING, DP_VID_M_DOUBLE_VALUE_EN, mask_sh)
+
+#define SE_COMMON_MASK_SH_LIST_DCE120(mask_sh)\
+	SE_COMMON_MASK_SH_LIST_SOC(mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC0_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC2_UPDATE, mask_sh),\
+	SE_SF(DP0_DP_PIXEL_FORMAT, DP_DYN_RANGE, mask_sh),\
+	SE_SF(DP0_DP_PIXEL_FORMAT, DP_YCBCR_RANGE, mask_sh),\
+	SE_SF(DIG0_HDMI_INFOFRAME_CONTROL0, HDMI_AVI_INFO_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_INFOFRAME_CONTROL0, HDMI_AVI_INFO_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_INFOFRAME_CONTROL1, HDMI_AVI_INFO_LINE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_AVI_ENABLE, mask_sh),\
+	SE_SF(DIG0_AFMT_AVI_INFO3, AFMT_AVI_INFO_VERSION, mask_sh),\
+	SE_SF(DP0_DP_VID_TIMING, DP_VID_M_DOUBLE_VALUE_EN, mask_sh)
+
+#define SE_COMMON_MASK_SH_LIST_DCN10(mask_sh)\
+	SE_COMMON_MASK_SH_LIST_SOC(mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_LOCK_STATUS, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_CONFLICT, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_CONFLICT_CLR, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC0_FRAME_UPDATE_PENDING, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC1_FRAME_UPDATE_PENDING, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC2_FRAME_UPDATE_PENDING, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC3_FRAME_UPDATE_PENDING, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC4_FRAME_UPDATE_PENDING, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC5_FRAME_UPDATE_PENDING, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC6_FRAME_UPDATE_PENDING, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC7_FRAME_UPDATE_PENDING, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC0_FRAME_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC1_FRAME_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC2_FRAME_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC3_FRAME_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC4_FRAME_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC5_FRAME_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC6_FRAME_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC7_FRAME_UPDATE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP4_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP5_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP6_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP7_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_DB_CNTL, DP_DB_DISABLE, mask_sh),\
+	SE_SF(DP0_DP_MSA_COLORIMETRY, DP_MSA_MISC0, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM1, DP_MSA_HTOTAL, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM1, DP_MSA_VTOTAL, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM2, DP_MSA_HSTART, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM2, DP_MSA_VSTART, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_HSYNCWIDTH, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_HSYNCPOLARITY, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_VSYNCWIDTH, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_VSYNCPOLARITY, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM4, DP_MSA_HWIDTH, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM4, DP_MSA_VHEIGHT, mask_sh),\
+	SE_SF(DIG0_HDMI_DB_CONTROL, HDMI_DB_DISABLE, mask_sh),\
+	SE_SF(DP0_DP_VID_TIMING, DP_VID_N_MUL, mask_sh)
+
+struct dce_stream_encoder_shift {
+	uint8_t AFMT_GENERIC_INDEX;
+	uint8_t AFMT_GENERIC0_UPDATE;
+	uint8_t AFMT_GENERIC2_UPDATE;
+	uint8_t AFMT_GENERIC_HB0;
+	uint8_t AFMT_GENERIC_HB1;
+	uint8_t AFMT_GENERIC_HB2;
+	uint8_t AFMT_GENERIC_HB3;
+	uint8_t AFMT_GENERIC_LOCK_STATUS;
+	uint8_t AFMT_GENERIC_CONFLICT;
+	uint8_t AFMT_GENERIC_CONFLICT_CLR;
+	uint8_t AFMT_GENERIC0_FRAME_UPDATE_PENDING;
+	uint8_t AFMT_GENERIC1_FRAME_UPDATE_PENDING;
+	uint8_t AFMT_GENERIC2_FRAME_UPDATE_PENDING;
+	uint8_t AFMT_GENERIC3_FRAME_UPDATE_PENDING;
+	uint8_t AFMT_GENERIC4_FRAME_UPDATE_PENDING;
+	uint8_t AFMT_GENERIC5_FRAME_UPDATE_PENDING;
+	uint8_t AFMT_GENERIC6_FRAME_UPDATE_PENDING;
+	uint8_t AFMT_GENERIC7_FRAME_UPDATE_PENDING;
+	uint8_t AFMT_GENERIC0_FRAME_UPDATE;
+	uint8_t AFMT_GENERIC1_FRAME_UPDATE;
+	uint8_t AFMT_GENERIC2_FRAME_UPDATE;
+	uint8_t AFMT_GENERIC3_FRAME_UPDATE;
+	uint8_t AFMT_GENERIC4_FRAME_UPDATE;
+	uint8_t AFMT_GENERIC5_FRAME_UPDATE;
+	uint8_t AFMT_GENERIC6_FRAME_UPDATE;
+	uint8_t AFMT_GENERIC7_FRAME_UPDATE;
+	uint8_t HDMI_GENERIC0_CONT;
+	uint8_t HDMI_GENERIC0_SEND;
+	uint8_t HDMI_GENERIC0_LINE;
+	uint8_t HDMI_GENERIC1_CONT;
+	uint8_t HDMI_GENERIC1_SEND;
+	uint8_t HDMI_GENERIC1_LINE;
+	uint8_t DP_PIXEL_ENCODING;
+	uint8_t DP_COMPONENT_DEPTH;
+	uint8_t DP_DYN_RANGE;
+	uint8_t DP_YCBCR_RANGE;
+	uint8_t HDMI_PACKET_GEN_VERSION;
+	uint8_t HDMI_KEEPOUT_MODE;
+	uint8_t HDMI_DEEP_COLOR_ENABLE;
+	uint8_t HDMI_CLOCK_CHANNEL_RATE;
+	uint8_t HDMI_DEEP_COLOR_DEPTH;
+	uint8_t HDMI_GC_CONT;
+	uint8_t HDMI_GC_SEND;
+	uint8_t HDMI_NULL_SEND;
+	uint8_t HDMI_DATA_SCRAMBLE_EN;
+	uint8_t HDMI_ACP_SEND;
+	uint8_t HDMI_AUDIO_INFO_SEND;
+	uint8_t AFMT_AUDIO_INFO_UPDATE;
+	uint8_t HDMI_AUDIO_INFO_LINE;
+	uint8_t HDMI_GC_AVMUTE;
+	uint8_t DP_MSE_RATE_X;
+	uint8_t DP_MSE_RATE_Y;
+	uint8_t DP_MSE_RATE_UPDATE_PENDING;
+	uint8_t AFMT_AVI_INFO_VERSION;
+	uint8_t HDMI_AVI_INFO_SEND;
+	uint8_t HDMI_AVI_INFO_CONT;
+	uint8_t HDMI_AVI_INFO_LINE;
+	uint8_t DP_SEC_GSP0_ENABLE;
+	uint8_t DP_SEC_STREAM_ENABLE;
+	uint8_t DP_SEC_GSP1_ENABLE;
+	uint8_t DP_SEC_GSP2_ENABLE;
+	uint8_t DP_SEC_GSP3_ENABLE;
+	uint8_t DP_SEC_GSP4_ENABLE;
+	uint8_t DP_SEC_GSP5_ENABLE;
+	uint8_t DP_SEC_GSP6_ENABLE;
+	uint8_t DP_SEC_GSP7_ENABLE;
+	uint8_t DP_SEC_AVI_ENABLE;
+	uint8_t DP_SEC_MPG_ENABLE;
+	uint8_t DP_VID_STREAM_DIS_DEFER;
+	uint8_t DP_VID_STREAM_ENABLE;
+	uint8_t DP_VID_STREAM_STATUS;
+	uint8_t DP_STEER_FIFO_RESET;
+	uint8_t DP_VID_M_N_GEN_EN;
+	uint8_t DP_VID_N;
+	uint8_t DP_VID_M;
+	uint8_t DIG_START;
+	uint8_t AFMT_AUDIO_SRC_SELECT;
+	uint8_t AFMT_AUDIO_CHANNEL_ENABLE;
+	uint8_t HDMI_AUDIO_PACKETS_PER_LINE;
+	uint8_t HDMI_AUDIO_DELAY_EN;
+	uint8_t AFMT_60958_CS_UPDATE;
+	uint8_t AFMT_AUDIO_LAYOUT_OVRD;
+	uint8_t AFMT_60958_OSF_OVRD;
+	uint8_t HDMI_ACR_AUTO_SEND;
+	uint8_t HDMI_ACR_SOURCE;
+	uint8_t HDMI_ACR_AUDIO_PRIORITY;
+	uint8_t HDMI_ACR_CTS_32;
+	uint8_t HDMI_ACR_N_32;
+	uint8_t HDMI_ACR_CTS_44;
+	uint8_t HDMI_ACR_N_44;
+	uint8_t HDMI_ACR_CTS_48;
+	uint8_t HDMI_ACR_N_48;
+	uint8_t AFMT_60958_CS_CHANNEL_NUMBER_L;
+	uint8_t AFMT_60958_CS_CLOCK_ACCURACY;
+	uint8_t AFMT_60958_CS_CHANNEL_NUMBER_R;
+	uint8_t AFMT_60958_CS_CHANNEL_NUMBER_2;
+	uint8_t AFMT_60958_CS_CHANNEL_NUMBER_3;
+	uint8_t AFMT_60958_CS_CHANNEL_NUMBER_4;
+	uint8_t AFMT_60958_CS_CHANNEL_NUMBER_5;
+	uint8_t AFMT_60958_CS_CHANNEL_NUMBER_6;
+	uint8_t AFMT_60958_CS_CHANNEL_NUMBER_7;
+	uint8_t DP_SEC_AUD_N;
+	uint8_t DP_SEC_TIMESTAMP_MODE;
+	uint8_t DP_SEC_ASP_ENABLE;
+	uint8_t DP_SEC_ATP_ENABLE;
+	uint8_t DP_SEC_AIP_ENABLE;
+	uint8_t DP_SEC_ACM_ENABLE;
+	uint8_t AFMT_AUDIO_SAMPLE_SEND;
+	uint8_t AFMT_AUDIO_CLOCK_EN;
+	uint8_t TMDS_PIXEL_ENCODING;
+	uint8_t TMDS_COLOR_FORMAT;
+	uint8_t DIG_STEREOSYNC_SELECT;
+	uint8_t DIG_STEREOSYNC_GATE_EN;
+	uint8_t DP_DB_DISABLE;
+	uint8_t DP_MSA_MISC0;
+	uint8_t DP_MSA_HTOTAL;
+	uint8_t DP_MSA_VTOTAL;
+	uint8_t DP_MSA_HSTART;
+	uint8_t DP_MSA_VSTART;
+	uint8_t DP_MSA_HSYNCWIDTH;
+	uint8_t DP_MSA_HSYNCPOLARITY;
+	uint8_t DP_MSA_VSYNCWIDTH;
+	uint8_t DP_MSA_VSYNCPOLARITY;
+	uint8_t DP_MSA_HWIDTH;
+	uint8_t DP_MSA_VHEIGHT;
+	uint8_t HDMI_DB_DISABLE;
+	uint8_t DP_VID_N_MUL;
+	uint8_t DP_VID_M_DOUBLE_VALUE_EN;
+	uint8_t DIG_SOURCE_SELECT;
+};
+
+struct dce_stream_encoder_mask {
+	uint32_t AFMT_GENERIC_INDEX;
+	uint32_t AFMT_GENERIC0_UPDATE;
+	uint32_t AFMT_GENERIC2_UPDATE;
+	uint32_t AFMT_GENERIC_HB0;
+	uint32_t AFMT_GENERIC_HB1;
+	uint32_t AFMT_GENERIC_HB2;
+	uint32_t AFMT_GENERIC_HB3;
+	uint32_t AFMT_GENERIC_LOCK_STATUS;
+	uint32_t AFMT_GENERIC_CONFLICT;
+	uint32_t AFMT_GENERIC_CONFLICT_CLR;
+	uint32_t AFMT_GENERIC0_FRAME_UPDATE_PENDING;
+	uint32_t AFMT_GENERIC1_FRAME_UPDATE_PENDING;
+	uint32_t AFMT_GENERIC2_FRAME_UPDATE_PENDING;
+	uint32_t AFMT_GENERIC3_FRAME_UPDATE_PENDING;
+	uint32_t AFMT_GENERIC4_FRAME_UPDATE_PENDING;
+	uint32_t AFMT_GENERIC5_FRAME_UPDATE_PENDING;
+	uint32_t AFMT_GENERIC6_FRAME_UPDATE_PENDING;
+	uint32_t AFMT_GENERIC7_FRAME_UPDATE_PENDING;
+	uint32_t AFMT_GENERIC0_FRAME_UPDATE;
+	uint32_t AFMT_GENERIC1_FRAME_UPDATE;
+	uint32_t AFMT_GENERIC2_FRAME_UPDATE;
+	uint32_t AFMT_GENERIC3_FRAME_UPDATE;
+	uint32_t AFMT_GENERIC4_FRAME_UPDATE;
+	uint32_t AFMT_GENERIC5_FRAME_UPDATE;
+	uint32_t AFMT_GENERIC6_FRAME_UPDATE;
+	uint32_t AFMT_GENERIC7_FRAME_UPDATE;
+	uint32_t HDMI_GENERIC0_CONT;
+	uint32_t HDMI_GENERIC0_SEND;
+	uint32_t HDMI_GENERIC0_LINE;
+	uint32_t HDMI_GENERIC1_CONT;
+	uint32_t HDMI_GENERIC1_SEND;
+	uint32_t HDMI_GENERIC1_LINE;
+	uint32_t DP_PIXEL_ENCODING;
+	uint32_t DP_COMPONENT_DEPTH;
+	uint32_t DP_DYN_RANGE;
+	uint32_t DP_YCBCR_RANGE;
+	uint32_t HDMI_PACKET_GEN_VERSION;
+	uint32_t HDMI_KEEPOUT_MODE;
+	uint32_t HDMI_DEEP_COLOR_ENABLE;
+	uint32_t HDMI_CLOCK_CHANNEL_RATE;
+	uint32_t HDMI_DEEP_COLOR_DEPTH;
+	uint32_t HDMI_GC_CONT;
+	uint32_t HDMI_GC_SEND;
+	uint32_t HDMI_NULL_SEND;
+	uint32_t HDMI_DATA_SCRAMBLE_EN;
+	uint32_t HDMI_ACP_SEND;
+	uint32_t HDMI_AUDIO_INFO_SEND;
+	uint32_t AFMT_AUDIO_INFO_UPDATE;
+	uint32_t HDMI_AUDIO_INFO_LINE;
+	uint32_t HDMI_GC_AVMUTE;
+	uint32_t DP_MSE_RATE_X;
+	uint32_t DP_MSE_RATE_Y;
+	uint32_t DP_MSE_RATE_UPDATE_PENDING;
+	uint32_t AFMT_AVI_INFO_VERSION;
+	uint32_t HDMI_AVI_INFO_SEND;
+	uint32_t HDMI_AVI_INFO_CONT;
+	uint32_t HDMI_AVI_INFO_LINE;
+	uint32_t DP_SEC_GSP0_ENABLE;
+	uint32_t DP_SEC_STREAM_ENABLE;
+	uint32_t DP_SEC_GSP1_ENABLE;
+	uint32_t DP_SEC_GSP2_ENABLE;
+	uint32_t DP_SEC_GSP3_ENABLE;
+	uint32_t DP_SEC_GSP4_ENABLE;
+	uint32_t DP_SEC_GSP5_ENABLE;
+	uint32_t DP_SEC_GSP6_ENABLE;
+	uint32_t DP_SEC_GSP7_ENABLE;
+	uint32_t DP_SEC_AVI_ENABLE;
+	uint32_t DP_SEC_MPG_ENABLE;
+	uint32_t DP_VID_STREAM_DIS_DEFER;
+	uint32_t DP_VID_STREAM_ENABLE;
+	uint32_t DP_VID_STREAM_STATUS;
+	uint32_t DP_STEER_FIFO_RESET;
+	uint32_t DP_VID_M_N_GEN_EN;
+	uint32_t DP_VID_N;
+	uint32_t DP_VID_M;
+	uint32_t DIG_START;
+	uint32_t AFMT_AUDIO_SRC_SELECT;
+	uint32_t AFMT_AUDIO_CHANNEL_ENABLE;
+	uint32_t HDMI_AUDIO_PACKETS_PER_LINE;
+	uint32_t HDMI_AUDIO_DELAY_EN;
+	uint32_t AFMT_60958_CS_UPDATE;
+	uint32_t AFMT_AUDIO_LAYOUT_OVRD;
+	uint32_t AFMT_60958_OSF_OVRD;
+	uint32_t HDMI_ACR_AUTO_SEND;
+	uint32_t HDMI_ACR_SOURCE;
+	uint32_t HDMI_ACR_AUDIO_PRIORITY;
+	uint32_t HDMI_ACR_CTS_32;
+	uint32_t HDMI_ACR_N_32;
+	uint32_t HDMI_ACR_CTS_44;
+	uint32_t HDMI_ACR_N_44;
+	uint32_t HDMI_ACR_CTS_48;
+	uint32_t HDMI_ACR_N_48;
+	uint32_t AFMT_60958_CS_CHANNEL_NUMBER_L;
+	uint32_t AFMT_60958_CS_CLOCK_ACCURACY;
+	uint32_t AFMT_60958_CS_CHANNEL_NUMBER_R;
+	uint32_t AFMT_60958_CS_CHANNEL_NUMBER_2;
+	uint32_t AFMT_60958_CS_CHANNEL_NUMBER_3;
+	uint32_t AFMT_60958_CS_CHANNEL_NUMBER_4;
+	uint32_t AFMT_60958_CS_CHANNEL_NUMBER_5;
+	uint32_t AFMT_60958_CS_CHANNEL_NUMBER_6;
+	uint32_t AFMT_60958_CS_CHANNEL_NUMBER_7;
+	uint32_t DP_SEC_AUD_N;
+	uint32_t DP_SEC_TIMESTAMP_MODE;
+	uint32_t DP_SEC_ASP_ENABLE;
+	uint32_t DP_SEC_ATP_ENABLE;
+	uint32_t DP_SEC_AIP_ENABLE;
+	uint32_t DP_SEC_ACM_ENABLE;
+	uint32_t AFMT_AUDIO_SAMPLE_SEND;
+	uint32_t AFMT_AUDIO_CLOCK_EN;
+	uint32_t TMDS_PIXEL_ENCODING;
+	uint32_t DIG_STEREOSYNC_SELECT;
+	uint32_t DIG_STEREOSYNC_GATE_EN;
+	uint32_t TMDS_COLOR_FORMAT;
+	uint32_t DP_DB_DISABLE;
+	uint32_t DP_MSA_MISC0;
+	uint32_t DP_MSA_HTOTAL;
+	uint32_t DP_MSA_VTOTAL;
+	uint32_t DP_MSA_HSTART;
+	uint32_t DP_MSA_VSTART;
+	uint32_t DP_MSA_HSYNCWIDTH;
+	uint32_t DP_MSA_HSYNCPOLARITY;
+	uint32_t DP_MSA_VSYNCWIDTH;
+	uint32_t DP_MSA_VSYNCPOLARITY;
+	uint32_t DP_MSA_HWIDTH;
+	uint32_t DP_MSA_VHEIGHT;
+	uint32_t HDMI_DB_DISABLE;
+	uint32_t DP_VID_N_MUL;
+	uint32_t DP_VID_M_DOUBLE_VALUE_EN;
+	uint32_t DIG_SOURCE_SELECT;
+};
+
+struct dce110_stream_enc_registers {
+	uint32_t AFMT_CNTL;
+	uint32_t AFMT_AVI_INFO0;
+	uint32_t AFMT_AVI_INFO1;
+	uint32_t AFMT_AVI_INFO2;
+	uint32_t AFMT_AVI_INFO3;
+	uint32_t AFMT_GENERIC_0;
+	uint32_t AFMT_GENERIC_1;
+	uint32_t AFMT_GENERIC_2;
+	uint32_t AFMT_GENERIC_3;
+	uint32_t AFMT_GENERIC_4;
+	uint32_t AFMT_GENERIC_5;
+	uint32_t AFMT_GENERIC_6;
+	uint32_t AFMT_GENERIC_7;
+	uint32_t AFMT_GENERIC_HDR;
+	uint32_t AFMT_INFOFRAME_CONTROL0;
+	uint32_t AFMT_VBI_PACKET_CONTROL;
+	uint32_t AFMT_VBI_PACKET_CONTROL1;
+	uint32_t AFMT_AUDIO_PACKET_CONTROL;
+	uint32_t AFMT_AUDIO_PACKET_CONTROL2;
+	uint32_t AFMT_AUDIO_SRC_CONTROL;
+	uint32_t AFMT_60958_0;
+	uint32_t AFMT_60958_1;
+	uint32_t AFMT_60958_2;
+	uint32_t DIG_FE_CNTL;
+	uint32_t DP_MSE_RATE_CNTL;
+	uint32_t DP_MSE_RATE_UPDATE;
+	uint32_t DP_PIXEL_FORMAT;
+	uint32_t DP_SEC_CNTL;
+	uint32_t DP_STEER_FIFO;
+	uint32_t DP_VID_M;
+	uint32_t DP_VID_N;
+	uint32_t DP_VID_STREAM_CNTL;
+	uint32_t DP_VID_TIMING;
+	uint32_t DP_SEC_AUD_N;
+	uint32_t DP_SEC_TIMESTAMP;
+	uint32_t HDMI_CONTROL;
+	uint32_t HDMI_GC;
+	uint32_t HDMI_GENERIC_PACKET_CONTROL0;
+	uint32_t HDMI_GENERIC_PACKET_CONTROL1;
+	uint32_t HDMI_GENERIC_PACKET_CONTROL2;
+	uint32_t HDMI_GENERIC_PACKET_CONTROL3;
+	uint32_t HDMI_INFOFRAME_CONTROL0;
+	uint32_t HDMI_INFOFRAME_CONTROL1;
+	uint32_t HDMI_VBI_PACKET_CONTROL;
+	uint32_t HDMI_AUDIO_PACKET_CONTROL;
+	uint32_t HDMI_ACR_PACKET_CONTROL;
+	uint32_t HDMI_ACR_32_0;
+	uint32_t HDMI_ACR_32_1;
+	uint32_t HDMI_ACR_44_0;
+	uint32_t HDMI_ACR_44_1;
+	uint32_t HDMI_ACR_48_0;
+	uint32_t HDMI_ACR_48_1;
+	uint32_t TMDS_CNTL;
+	uint32_t DP_DB_CNTL;
+	uint32_t DP_MSA_MISC;
+	uint32_t DP_MSA_COLORIMETRY;
+	uint32_t DP_MSA_TIMING_PARAM1;
+	uint32_t DP_MSA_TIMING_PARAM2;
+	uint32_t DP_MSA_TIMING_PARAM3;
+	uint32_t DP_MSA_TIMING_PARAM4;
+	uint32_t HDMI_DB_CONTROL;
+};
+
+struct dce110_stream_encoder {
+	struct stream_encoder base;
+	const struct dce110_stream_enc_registers *regs;
+	const struct dce_stream_encoder_shift *se_shift;
+	const struct dce_stream_encoder_mask *se_mask;
+};
+
+void dce110_stream_encoder_construct(
+	struct dce110_stream_encoder *enc110,
+	struct dc_context *ctx,
+	struct dc_bios *bp,
+	enum engine_id eng_id,
+	const struct dce110_stream_enc_registers *regs,
+	const struct dce_stream_encoder_shift *se_shift,
+	const struct dce_stream_encoder_mask *se_mask);
+
+
+void dce110_se_audio_mute_control(
+	struct stream_encoder *enc, bool mute);
+
+void dce110_se_dp_audio_setup(
+	struct stream_encoder *enc,
+	unsigned int az_inst,
+	struct audio_info *info);
+
+void dce110_se_dp_audio_enable(
+		struct stream_encoder *enc);
+
+void dce110_se_dp_audio_disable(
+		struct stream_encoder *enc);
+
+void dce110_se_hdmi_audio_setup(
+	struct stream_encoder *enc,
+	unsigned int az_inst,
+	struct audio_info *info,
+	struct audio_crtc_info *audio_crtc_info);
+
+void dce110_se_hdmi_audio_disable(
+	struct stream_encoder *enc);
+
+#endif /* __DC_STREAM_ENCODER_DCE110_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_transform.c b/drivers/gpu/drm/amd/display/dc/dce/dce_transform.c
new file mode 100644
index 000000000000..1ab5ae9b5ea5
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_transform.c
@@ -0,0 +1,1689 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dce_transform.h"
+#include "reg_helper.h"
+#include "opp.h"
+#include "basics/conversion.h"
+#include "dc.h"
+
+#define REG(reg) \
+	(xfm_dce->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	xfm_dce->xfm_shift->field_name, xfm_dce->xfm_mask->field_name
+
+#define CTX \
+	xfm_dce->base.ctx
+#define DC_LOGGER \
+	xfm_dce->base.ctx->logger
+
+#define IDENTITY_RATIO(ratio) (dc_fixpt_u2d19(ratio) == (1 << 19))
+#define GAMUT_MATRIX_SIZE 12
+#define SCL_PHASES 16
+
+enum dcp_out_trunc_round_mode {
+	DCP_OUT_TRUNC_ROUND_MODE_TRUNCATE,
+	DCP_OUT_TRUNC_ROUND_MODE_ROUND
+};
+
+enum dcp_out_trunc_round_depth {
+	DCP_OUT_TRUNC_ROUND_DEPTH_14BIT,
+	DCP_OUT_TRUNC_ROUND_DEPTH_13BIT,
+	DCP_OUT_TRUNC_ROUND_DEPTH_12BIT,
+	DCP_OUT_TRUNC_ROUND_DEPTH_11BIT,
+	DCP_OUT_TRUNC_ROUND_DEPTH_10BIT,
+	DCP_OUT_TRUNC_ROUND_DEPTH_9BIT,
+	DCP_OUT_TRUNC_ROUND_DEPTH_8BIT
+};
+
+/*  defines the various methods of bit reduction available for use */
+enum dcp_bit_depth_reduction_mode {
+	DCP_BIT_DEPTH_REDUCTION_MODE_DITHER,
+	DCP_BIT_DEPTH_REDUCTION_MODE_ROUND,
+	DCP_BIT_DEPTH_REDUCTION_MODE_TRUNCATE,
+	DCP_BIT_DEPTH_REDUCTION_MODE_DISABLED,
+	DCP_BIT_DEPTH_REDUCTION_MODE_INVALID
+};
+
+enum dcp_spatial_dither_mode {
+	DCP_SPATIAL_DITHER_MODE_AAAA,
+	DCP_SPATIAL_DITHER_MODE_A_AA_A,
+	DCP_SPATIAL_DITHER_MODE_AABBAABB,
+	DCP_SPATIAL_DITHER_MODE_AABBCCAABBCC,
+	DCP_SPATIAL_DITHER_MODE_INVALID
+};
+
+enum dcp_spatial_dither_depth {
+	DCP_SPATIAL_DITHER_DEPTH_30BPP,
+	DCP_SPATIAL_DITHER_DEPTH_24BPP
+};
+
+enum csc_color_mode {
+	/* 00 - BITS2:0 Bypass */
+	CSC_COLOR_MODE_GRAPHICS_BYPASS,
+	/* 01 - hard coded coefficient TV RGB */
+	CSC_COLOR_MODE_GRAPHICS_PREDEFINED,
+	/* 04 - programmable OUTPUT CSC coefficient */
+	CSC_COLOR_MODE_GRAPHICS_OUTPUT_CSC,
+};
+
+enum grph_color_adjust_option {
+	GRPH_COLOR_MATRIX_HW_DEFAULT = 1,
+	GRPH_COLOR_MATRIX_SW
+};
+
+static const struct out_csc_color_matrix global_color_matrix[] = {
+{ COLOR_SPACE_SRGB,
+	{ 0x2000, 0, 0, 0, 0, 0x2000, 0, 0, 0, 0, 0x2000, 0} },
+{ COLOR_SPACE_SRGB_LIMITED,
+	{ 0x1B60, 0, 0, 0x200, 0, 0x1B60, 0, 0x200, 0, 0, 0x1B60, 0x200} },
+{ COLOR_SPACE_YCBCR601,
+	{ 0xE00, 0xF447, 0xFDB9, 0x1000, 0x82F, 0x1012, 0x31F, 0x200, 0xFB47,
+		0xF6B9, 0xE00, 0x1000} },
+{ COLOR_SPACE_YCBCR709, { 0xE00, 0xF349, 0xFEB7, 0x1000, 0x5D2, 0x1394, 0x1FA,
+	0x200, 0xFCCB, 0xF535, 0xE00, 0x1000} },
+/* TODO: correct values below */
+{ COLOR_SPACE_YCBCR601_LIMITED, { 0xE00, 0xF447, 0xFDB9, 0x1000, 0x991,
+	0x12C9, 0x3A6, 0x200, 0xFB47, 0xF6B9, 0xE00, 0x1000} },
+{ COLOR_SPACE_YCBCR709_LIMITED, { 0xE00, 0xF349, 0xFEB7, 0x1000, 0x6CE, 0x16E3,
+	0x24F, 0x200, 0xFCCB, 0xF535, 0xE00, 0x1000} }
+};
+
+static bool setup_scaling_configuration(
+	struct dce_transform *xfm_dce,
+	const struct scaler_data *data)
+{
+	REG_SET(SCL_BYPASS_CONTROL, 0, SCL_BYPASS_MODE, 0);
+
+	if (data->taps.h_taps + data->taps.v_taps <= 2) {
+		/* Set bypass */
+		if (xfm_dce->xfm_mask->SCL_PSCL_EN != 0)
+			REG_UPDATE_2(SCL_MODE, SCL_MODE, 0, SCL_PSCL_EN, 0);
+		else
+			REG_UPDATE(SCL_MODE, SCL_MODE, 0);
+		return false;
+	}
+
+	REG_SET_2(SCL_TAP_CONTROL, 0,
+			SCL_H_NUM_OF_TAPS, data->taps.h_taps - 1,
+			SCL_V_NUM_OF_TAPS, data->taps.v_taps - 1);
+
+	if (data->format <= PIXEL_FORMAT_GRPH_END)
+		REG_UPDATE(SCL_MODE, SCL_MODE, 1);
+	else
+		REG_UPDATE(SCL_MODE, SCL_MODE, 2);
+
+	if (xfm_dce->xfm_mask->SCL_PSCL_EN != 0)
+		REG_UPDATE(SCL_MODE, SCL_PSCL_EN, 1);
+
+	/* 1 - Replace out of bound pixels with edge */
+	REG_SET(SCL_CONTROL, 0, SCL_BOUNDARY_MODE, 1);
+
+	return true;
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static bool dce60_setup_scaling_configuration(
+	struct dce_transform *xfm_dce,
+	const struct scaler_data *data)
+{
+	REG_SET(SCL_BYPASS_CONTROL, 0, SCL_BYPASS_MODE, 0);
+
+	if (data->taps.h_taps + data->taps.v_taps <= 2) {
+		/* Disable scaler functionality */
+		REG_WRITE(SCL_SCALER_ENABLE, 0);
+
+		/* Clear registers that can cause glitches even when the scaler is off */
+		REG_WRITE(SCL_TAP_CONTROL, 0);
+		REG_WRITE(SCL_AUTOMATIC_MODE_CONTROL, 0);
+		REG_WRITE(SCL_F_SHARP_CONTROL, 0);
+		return false;
+	}
+
+	REG_SET_2(SCL_TAP_CONTROL, 0,
+			SCL_H_NUM_OF_TAPS, data->taps.h_taps - 1,
+			SCL_V_NUM_OF_TAPS, data->taps.v_taps - 1);
+
+	REG_WRITE(SCL_SCALER_ENABLE, 1);
+
+	/* DCE6 has no SCL_BOUNDARY_MODE bit, skip replace out of bound pixels */
+
+	return true;
+}
+#endif
+
+static void program_overscan(
+		struct dce_transform *xfm_dce,
+		const struct scaler_data *data)
+{
+	int overscan_right = data->h_active
+			- data->recout.x - data->recout.width;
+	int overscan_bottom = data->v_active
+			- data->recout.y - data->recout.height;
+
+	if (xfm_dce->base.ctx->dc->debug.visual_confirm != VISUAL_CONFIRM_DISABLE) {
+		overscan_bottom += 2;
+		overscan_right += 2;
+	}
+
+	if (overscan_right < 0) {
+		BREAK_TO_DEBUGGER();
+		overscan_right = 0;
+	}
+	if (overscan_bottom < 0) {
+		BREAK_TO_DEBUGGER();
+		overscan_bottom = 0;
+	}
+
+	REG_SET_2(EXT_OVERSCAN_LEFT_RIGHT, 0,
+			EXT_OVERSCAN_LEFT, data->recout.x,
+			EXT_OVERSCAN_RIGHT, overscan_right);
+	REG_SET_2(EXT_OVERSCAN_TOP_BOTTOM, 0,
+			EXT_OVERSCAN_TOP, data->recout.y,
+			EXT_OVERSCAN_BOTTOM, overscan_bottom);
+}
+
+static void program_multi_taps_filter(
+	struct dce_transform *xfm_dce,
+	int taps,
+	const uint16_t *coeffs,
+	enum ram_filter_type filter_type)
+{
+	int phase, pair;
+	int array_idx = 0;
+	int taps_pairs = (taps + 1) / 2;
+	int phases_to_program = SCL_PHASES / 2 + 1;
+
+	uint32_t power_ctl = 0;
+
+	if (!coeffs)
+		return;
+
+	/*We need to disable power gating on coeff memory to do programming*/
+	if (REG(DCFE_MEM_PWR_CTRL)) {
+		power_ctl = REG_READ(DCFE_MEM_PWR_CTRL);
+		REG_SET(DCFE_MEM_PWR_CTRL, power_ctl, SCL_COEFF_MEM_PWR_DIS, 1);
+
+		REG_WAIT(DCFE_MEM_PWR_STATUS, SCL_COEFF_MEM_PWR_STATE, 0, 1, 10);
+	}
+	for (phase = 0; phase < phases_to_program; phase++) {
+		/*we always program N/2 + 1 phases, total phases N, but N/2-1 are just mirror
+		phase 0 is unique and phase N/2 is unique if N is even*/
+		for (pair = 0; pair < taps_pairs; pair++) {
+			uint16_t odd_coeff = 0;
+			uint16_t even_coeff = coeffs[array_idx];
+
+			REG_SET_3(SCL_COEF_RAM_SELECT, 0,
+					SCL_C_RAM_FILTER_TYPE, filter_type,
+					SCL_C_RAM_PHASE, phase,
+					SCL_C_RAM_TAP_PAIR_IDX, pair);
+
+			if (taps % 2 && pair == taps_pairs - 1)
+				array_idx++;
+			else {
+				odd_coeff = coeffs[array_idx + 1];
+				array_idx += 2;
+			}
+
+			REG_SET_4(SCL_COEF_RAM_TAP_DATA, 0,
+					SCL_C_RAM_EVEN_TAP_COEF_EN, 1,
+					SCL_C_RAM_EVEN_TAP_COEF, even_coeff,
+					SCL_C_RAM_ODD_TAP_COEF_EN, 1,
+					SCL_C_RAM_ODD_TAP_COEF, odd_coeff);
+		}
+	}
+
+	/*We need to restore power gating on coeff memory to initial state*/
+	if (REG(DCFE_MEM_PWR_CTRL))
+		REG_WRITE(DCFE_MEM_PWR_CTRL, power_ctl);
+}
+
+static void program_viewport(
+	struct dce_transform *xfm_dce,
+	const struct rect *view_port)
+{
+	REG_SET_2(VIEWPORT_START, 0,
+			VIEWPORT_X_START, view_port->x,
+			VIEWPORT_Y_START, view_port->y);
+
+	REG_SET_2(VIEWPORT_SIZE, 0,
+			VIEWPORT_HEIGHT, view_port->height,
+			VIEWPORT_WIDTH, view_port->width);
+
+	/* TODO: add stereo support */
+}
+
+static void calculate_inits(
+	struct dce_transform *xfm_dce,
+	const struct scaler_data *data,
+	struct scl_ratios_inits *inits)
+{
+	struct fixed31_32 h_init;
+	struct fixed31_32 v_init;
+
+	inits->h_int_scale_ratio =
+		dc_fixpt_u2d19(data->ratios.horz) << 5;
+	inits->v_int_scale_ratio =
+		dc_fixpt_u2d19(data->ratios.vert) << 5;
+
+	h_init =
+		dc_fixpt_div_int(
+			dc_fixpt_add(
+				data->ratios.horz,
+				dc_fixpt_from_int(data->taps.h_taps + 1)),
+				2);
+	inits->h_init.integer = dc_fixpt_floor(h_init);
+	inits->h_init.fraction = dc_fixpt_u0d19(h_init) << 5;
+
+	v_init =
+		dc_fixpt_div_int(
+			dc_fixpt_add(
+				data->ratios.vert,
+				dc_fixpt_from_int(data->taps.v_taps + 1)),
+				2);
+	inits->v_init.integer = dc_fixpt_floor(v_init);
+	inits->v_init.fraction = dc_fixpt_u0d19(v_init) << 5;
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static void dce60_calculate_inits(
+	struct dce_transform *xfm_dce,
+	const struct scaler_data *data,
+	struct sclh_ratios_inits *inits)
+{
+	struct fixed31_32 v_init;
+
+	inits->h_int_scale_ratio =
+		dc_fixpt_u2d19(data->ratios.horz) << 5;
+	inits->v_int_scale_ratio =
+		dc_fixpt_u2d19(data->ratios.vert) << 5;
+
+	/* DCE6 h_init_luma setting inspired by DCE110 */
+	inits->h_init_luma.integer = 1;
+
+	/* DCE6 h_init_chroma setting inspired by DCE110 */
+	inits->h_init_chroma.integer = 1;
+
+	v_init =
+		dc_fixpt_div_int(
+			dc_fixpt_add(
+				data->ratios.vert,
+				dc_fixpt_from_int(data->taps.v_taps + 1)),
+				2);
+	inits->v_init.integer = dc_fixpt_floor(v_init);
+	inits->v_init.fraction = dc_fixpt_u0d19(v_init) << 5;
+}
+#endif
+
+static void program_scl_ratios_inits(
+	struct dce_transform *xfm_dce,
+	struct scl_ratios_inits *inits)
+{
+
+	REG_SET(SCL_HORZ_FILTER_SCALE_RATIO, 0,
+			SCL_H_SCALE_RATIO, inits->h_int_scale_ratio);
+
+	REG_SET(SCL_VERT_FILTER_SCALE_RATIO, 0,
+			SCL_V_SCALE_RATIO, inits->v_int_scale_ratio);
+
+	REG_SET_2(SCL_HORZ_FILTER_INIT, 0,
+			SCL_H_INIT_INT, inits->h_init.integer,
+			SCL_H_INIT_FRAC, inits->h_init.fraction);
+
+	REG_SET_2(SCL_VERT_FILTER_INIT, 0,
+			SCL_V_INIT_INT, inits->v_init.integer,
+			SCL_V_INIT_FRAC, inits->v_init.fraction);
+
+	REG_WRITE(SCL_AUTOMATIC_MODE_CONTROL, 0);
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static void dce60_program_scl_ratios_inits(
+	struct dce_transform *xfm_dce,
+	struct sclh_ratios_inits *inits)
+{
+
+	REG_SET(SCL_HORZ_FILTER_SCALE_RATIO, 0,
+			SCL_H_SCALE_RATIO, inits->h_int_scale_ratio);
+
+	REG_SET(SCL_VERT_FILTER_SCALE_RATIO, 0,
+			SCL_V_SCALE_RATIO, inits->v_int_scale_ratio);
+
+	/* DCE6 has SCL_HORZ_FILTER_INIT_RGB_LUMA register */
+	REG_SET_2(SCL_HORZ_FILTER_INIT_RGB_LUMA, 0,
+			SCL_H_INIT_INT_RGB_Y, inits->h_init_luma.integer,
+			SCL_H_INIT_FRAC_RGB_Y, inits->h_init_luma.fraction);
+
+	/* DCE6 has SCL_HORZ_FILTER_INIT_CHROMA register */
+	REG_SET_2(SCL_HORZ_FILTER_INIT_CHROMA, 0,
+			SCL_H_INIT_INT_CBCR, inits->h_init_chroma.integer,
+			SCL_H_INIT_FRAC_CBCR, inits->h_init_chroma.fraction);
+
+	REG_SET_2(SCL_VERT_FILTER_INIT, 0,
+			SCL_V_INIT_INT, inits->v_init.integer,
+			SCL_V_INIT_FRAC, inits->v_init.fraction);
+
+	REG_WRITE(SCL_AUTOMATIC_MODE_CONTROL, 0);
+}
+#endif
+
+static const uint16_t *get_filter_coeffs_16p(int taps, struct fixed31_32 ratio)
+{
+	if (taps == 4)
+		return get_filter_4tap_16p(ratio);
+	else if (taps == 3)
+		return get_filter_3tap_16p(ratio);
+	else if (taps == 2)
+		return get_filter_2tap_16p();
+	else if (taps == 1)
+		return NULL;
+	else {
+		/* should never happen, bug */
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+}
+
+static void dce_transform_set_scaler(
+	struct transform *xfm,
+	const struct scaler_data *data)
+{
+	struct dce_transform *xfm_dce = TO_DCE_TRANSFORM(xfm);
+	bool is_scaling_required;
+	bool filter_updated = false;
+	const uint16_t *coeffs_v, *coeffs_h;
+
+	/*Use all three pieces of memory always*/
+	REG_SET_2(LB_MEMORY_CTRL, 0,
+			LB_MEMORY_CONFIG, 0,
+			LB_MEMORY_SIZE, xfm_dce->lb_memory_size);
+
+	/* Clear SCL_F_SHARP_CONTROL value to 0 */
+	REG_WRITE(SCL_F_SHARP_CONTROL, 0);
+
+	/* 1. Program overscan */
+	program_overscan(xfm_dce, data);
+
+	/* 2. Program taps and configuration */
+	is_scaling_required = setup_scaling_configuration(xfm_dce, data);
+
+	if (is_scaling_required) {
+		/* 3. Calculate and program ratio, filter initialization */
+		struct scl_ratios_inits inits = { 0 };
+
+		calculate_inits(xfm_dce, data, &inits);
+
+		program_scl_ratios_inits(xfm_dce, &inits);
+
+		coeffs_v = get_filter_coeffs_16p(data->taps.v_taps, data->ratios.vert);
+		coeffs_h = get_filter_coeffs_16p(data->taps.h_taps, data->ratios.horz);
+
+		if (coeffs_v != xfm_dce->filter_v || coeffs_h != xfm_dce->filter_h) {
+			/* 4. Program vertical filters */
+			if (xfm_dce->filter_v == NULL)
+				REG_SET(SCL_VERT_FILTER_CONTROL, 0,
+						SCL_V_2TAP_HARDCODE_COEF_EN, 0);
+			program_multi_taps_filter(
+					xfm_dce,
+					data->taps.v_taps,
+					coeffs_v,
+					FILTER_TYPE_RGB_Y_VERTICAL);
+			program_multi_taps_filter(
+					xfm_dce,
+					data->taps.v_taps,
+					coeffs_v,
+					FILTER_TYPE_ALPHA_VERTICAL);
+
+			/* 5. Program horizontal filters */
+			if (xfm_dce->filter_h == NULL)
+				REG_SET(SCL_HORZ_FILTER_CONTROL, 0,
+						SCL_H_2TAP_HARDCODE_COEF_EN, 0);
+			program_multi_taps_filter(
+					xfm_dce,
+					data->taps.h_taps,
+					coeffs_h,
+					FILTER_TYPE_RGB_Y_HORIZONTAL);
+			program_multi_taps_filter(
+					xfm_dce,
+					data->taps.h_taps,
+					coeffs_h,
+					FILTER_TYPE_ALPHA_HORIZONTAL);
+
+			xfm_dce->filter_v = coeffs_v;
+			xfm_dce->filter_h = coeffs_h;
+			filter_updated = true;
+		}
+	}
+
+	/* 6. Program the viewport */
+	program_viewport(xfm_dce, &data->viewport);
+
+	/* 7. Set bit to flip to new coefficient memory */
+	if (filter_updated)
+		REG_UPDATE(SCL_UPDATE, SCL_COEF_UPDATE_COMPLETE, 1);
+
+	REG_UPDATE(LB_DATA_FORMAT, ALPHA_EN, data->lb_params.alpha_en);
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static void dce60_transform_set_scaler(
+	struct transform *xfm,
+	const struct scaler_data *data)
+{
+	struct dce_transform *xfm_dce = TO_DCE_TRANSFORM(xfm);
+	bool is_scaling_required;
+	const uint16_t *coeffs_v, *coeffs_h;
+
+	/*Use whole line buffer memory always*/
+	REG_SET(DC_LB_MEMORY_SPLIT, 0,
+		DC_LB_MEMORY_CONFIG, 0);
+
+	REG_SET(DC_LB_MEM_SIZE, 0,
+		DC_LB_MEM_SIZE, xfm_dce->lb_memory_size);
+
+	REG_WRITE(SCL_UPDATE, 0x00010000);
+
+	/* Clear SCL_F_SHARP_CONTROL value to 0 */
+	REG_WRITE(SCL_F_SHARP_CONTROL, 0);
+
+	/* 1. Program overscan */
+	program_overscan(xfm_dce, data);
+
+	/* 2. Program taps and configuration */
+	is_scaling_required = dce60_setup_scaling_configuration(xfm_dce, data);
+
+	if (is_scaling_required) {
+		/* 3. Calculate and program ratio, DCE6 filter initialization */
+		struct sclh_ratios_inits inits = { 0 };
+
+		/* DCE6 has specific calculate_inits() function */
+		dce60_calculate_inits(xfm_dce, data, &inits);
+
+		/* DCE6 has specific program_scl_ratios_inits() function */
+		dce60_program_scl_ratios_inits(xfm_dce, &inits);
+
+		coeffs_v = get_filter_coeffs_16p(data->taps.v_taps, data->ratios.vert);
+		coeffs_h = get_filter_coeffs_16p(data->taps.h_taps, data->ratios.horz);
+
+		if (coeffs_v != xfm_dce->filter_v || coeffs_h != xfm_dce->filter_h) {
+			/* 4. Program vertical filters */
+			if (xfm_dce->filter_v == NULL)
+				REG_WRITE(SCL_VERT_FILTER_CONTROL, 0);
+			program_multi_taps_filter(
+					xfm_dce,
+					data->taps.v_taps,
+					coeffs_v,
+					FILTER_TYPE_RGB_Y_VERTICAL);
+			program_multi_taps_filter(
+					xfm_dce,
+					data->taps.v_taps,
+					coeffs_v,
+					FILTER_TYPE_ALPHA_VERTICAL);
+
+			/* 5. Program horizontal filters */
+			if (xfm_dce->filter_h == NULL)
+				REG_WRITE(SCL_HORZ_FILTER_CONTROL, 0);
+			program_multi_taps_filter(
+					xfm_dce,
+					data->taps.h_taps,
+					coeffs_h,
+					FILTER_TYPE_RGB_Y_HORIZONTAL);
+			program_multi_taps_filter(
+					xfm_dce,
+					data->taps.h_taps,
+					coeffs_h,
+					FILTER_TYPE_ALPHA_HORIZONTAL);
+
+			xfm_dce->filter_v = coeffs_v;
+			xfm_dce->filter_h = coeffs_h;
+		}
+	}
+
+	/* 6. Program the viewport */
+	program_viewport(xfm_dce, &data->viewport);
+
+	/* DCE6 has no SCL_COEF_UPDATE_COMPLETE bit to flip to new coefficient memory */
+
+	/* DCE6 DATA_FORMAT register does not support ALPHA_EN */
+
+	REG_WRITE(SCL_UPDATE, 0);
+}
+#endif
+
+/*****************************************************************************
+ * set_clamp
+ *
+ * @param depth : bit depth to set the clamp to (should match denorm)
+ *
+ * @brief
+ *     Programs clamp according to panel bit depth.
+ *
+ *******************************************************************************/
+static void set_clamp(
+	struct dce_transform *xfm_dce,
+	enum dc_color_depth depth)
+{
+	int clamp_max = 0;
+
+	/* At the clamp block the data will be MSB aligned, so we set the max
+	 * clamp accordingly.
+	 * For example, the max value for 6 bits MSB aligned (14 bit bus) would
+	 * be "11 1111 0000 0000" in binary, so 0x3F00.
+	 */
+	switch (depth) {
+	case COLOR_DEPTH_666:
+		/* 6bit MSB aligned on 14 bit bus '11 1111 0000 0000' */
+		clamp_max = 0x3F00;
+		break;
+	case COLOR_DEPTH_888:
+		/* 8bit MSB aligned on 14 bit bus '11 1111 1100 0000' */
+		clamp_max = 0x3FC0;
+		break;
+	case COLOR_DEPTH_101010:
+		/* 10bit MSB aligned on 14 bit bus '11 1111 1111 0000' */
+		clamp_max = 0x3FF0;
+		break;
+	case COLOR_DEPTH_121212:
+		/* 12bit MSB aligned on 14 bit bus '11 1111 1111 1100' */
+		clamp_max = 0x3FFC;
+		break;
+	default:
+		clamp_max = 0x3FC0;
+		BREAK_TO_DEBUGGER(); /* Invalid clamp bit depth */
+	}
+	REG_SET_2(OUT_CLAMP_CONTROL_B_CB, 0,
+			OUT_CLAMP_MIN_B_CB, 0,
+			OUT_CLAMP_MAX_B_CB, clamp_max);
+
+	REG_SET_2(OUT_CLAMP_CONTROL_G_Y, 0,
+			OUT_CLAMP_MIN_G_Y, 0,
+			OUT_CLAMP_MAX_G_Y, clamp_max);
+
+	REG_SET_2(OUT_CLAMP_CONTROL_R_CR, 0,
+			OUT_CLAMP_MIN_R_CR, 0,
+			OUT_CLAMP_MAX_R_CR, clamp_max);
+}
+
+/*******************************************************************************
+ * set_round
+ *
+ * @brief
+ *     Programs Round/Truncate
+ *
+ * @param [in] mode  :round or truncate
+ * @param [in] depth :bit depth to round/truncate to
+ OUT_ROUND_TRUNC_MODE 3:0 0xA Output data round or truncate mode
+ POSSIBLE VALUES:
+      00 - truncate to u0.12
+      01 - truncate to u0.11
+      02 - truncate to u0.10
+      03 - truncate to u0.9
+      04 - truncate to u0.8
+      05 - reserved
+      06 - truncate to u0.14
+      07 - truncate to u0.13		set_reg_field_value(
+			value,
+			clamp_max,
+			OUT_CLAMP_CONTROL_R_CR,
+			OUT_CLAMP_MAX_R_CR);
+      08 - round to u0.12
+      09 - round to u0.11
+      10 - round to u0.10
+      11 - round to u0.9
+      12 - round to u0.8
+      13 - reserved
+      14 - round to u0.14
+      15 - round to u0.13
+
+ ******************************************************************************/
+static void set_round(
+	struct dce_transform *xfm_dce,
+	enum dcp_out_trunc_round_mode mode,
+	enum dcp_out_trunc_round_depth depth)
+{
+	int depth_bits = 0;
+	int mode_bit = 0;
+
+	/*  set up bit depth */
+	switch (depth) {
+	case DCP_OUT_TRUNC_ROUND_DEPTH_14BIT:
+		depth_bits = 6;
+		break;
+	case DCP_OUT_TRUNC_ROUND_DEPTH_13BIT:
+		depth_bits = 7;
+		break;
+	case DCP_OUT_TRUNC_ROUND_DEPTH_12BIT:
+		depth_bits = 0;
+		break;
+	case DCP_OUT_TRUNC_ROUND_DEPTH_11BIT:
+		depth_bits = 1;
+		break;
+	case DCP_OUT_TRUNC_ROUND_DEPTH_10BIT:
+		depth_bits = 2;
+		break;
+	case DCP_OUT_TRUNC_ROUND_DEPTH_9BIT:
+		depth_bits = 3;
+		break;
+	case DCP_OUT_TRUNC_ROUND_DEPTH_8BIT:
+		depth_bits = 4;
+		break;
+	default:
+		depth_bits = 4;
+		BREAK_TO_DEBUGGER(); /* Invalid dcp_out_trunc_round_depth */
+	}
+
+	/*  set up round or truncate */
+	switch (mode) {
+	case DCP_OUT_TRUNC_ROUND_MODE_TRUNCATE:
+		mode_bit = 0;
+		break;
+	case DCP_OUT_TRUNC_ROUND_MODE_ROUND:
+		mode_bit = 1;
+		break;
+	default:
+		BREAK_TO_DEBUGGER(); /* Invalid dcp_out_trunc_round_mode */
+	}
+
+	depth_bits |= mode_bit << 3;
+
+	REG_SET(OUT_ROUND_CONTROL, 0, OUT_ROUND_TRUNC_MODE, depth_bits);
+}
+
+/*****************************************************************************
+ * set_dither
+ *
+ * @brief
+ *     Programs Dither
+ *
+ * @param [in] dither_enable        : enable dither
+ * @param [in] dither_mode           : dither mode to set
+ * @param [in] dither_depth          : bit depth to dither to
+ * @param [in] frame_random_enable    : enable frame random
+ * @param [in] rgb_random_enable      : enable rgb random
+ * @param [in] highpass_random_enable : enable highpass random
+ *
+ ******************************************************************************/
+
+static void set_dither(
+	struct dce_transform *xfm_dce,
+	bool dither_enable,
+	enum dcp_spatial_dither_mode dither_mode,
+	enum dcp_spatial_dither_depth dither_depth,
+	bool frame_random_enable,
+	bool rgb_random_enable,
+	bool highpass_random_enable)
+{
+	int dither_depth_bits = 0;
+	int dither_mode_bits = 0;
+
+	switch (dither_mode) {
+	case DCP_SPATIAL_DITHER_MODE_AAAA:
+		dither_mode_bits = 0;
+		break;
+	case DCP_SPATIAL_DITHER_MODE_A_AA_A:
+		dither_mode_bits = 1;
+		break;
+	case DCP_SPATIAL_DITHER_MODE_AABBAABB:
+		dither_mode_bits = 2;
+		break;
+	case DCP_SPATIAL_DITHER_MODE_AABBCCAABBCC:
+		dither_mode_bits = 3;
+		break;
+	default:
+		/* Invalid dcp_spatial_dither_mode */
+		BREAK_TO_DEBUGGER();
+	}
+
+	switch (dither_depth) {
+	case DCP_SPATIAL_DITHER_DEPTH_30BPP:
+		dither_depth_bits = 0;
+		break;
+	case DCP_SPATIAL_DITHER_DEPTH_24BPP:
+		dither_depth_bits = 1;
+		break;
+	default:
+		/* Invalid dcp_spatial_dither_depth */
+		BREAK_TO_DEBUGGER();
+	}
+
+	/*  write the register */
+	REG_SET_6(DCP_SPATIAL_DITHER_CNTL, 0,
+			DCP_SPATIAL_DITHER_EN, dither_enable,
+			DCP_SPATIAL_DITHER_MODE, dither_mode_bits,
+			DCP_SPATIAL_DITHER_DEPTH, dither_depth_bits,
+			DCP_FRAME_RANDOM_ENABLE, frame_random_enable,
+			DCP_RGB_RANDOM_ENABLE, rgb_random_enable,
+			DCP_HIGHPASS_RANDOM_ENABLE, highpass_random_enable);
+}
+
+/*****************************************************************************
+ * dce_transform_bit_depth_reduction_program
+ *
+ * @brief
+ *     Programs the DCP bit depth reduction registers (Clamp, Round/Truncate,
+ *      Dither) for dce
+ *
+ * @param depth : bit depth to set the clamp to (should match denorm)
+ *
+ ******************************************************************************/
+static void program_bit_depth_reduction(
+	struct dce_transform *xfm_dce,
+	enum dc_color_depth depth,
+	const struct bit_depth_reduction_params *bit_depth_params)
+{
+	enum dcp_out_trunc_round_depth trunc_round_depth;
+	enum dcp_out_trunc_round_mode trunc_mode;
+	bool spatial_dither_enable;
+
+	ASSERT(depth <= COLOR_DEPTH_121212); /* Invalid clamp bit depth */
+
+	spatial_dither_enable = bit_depth_params->flags.SPATIAL_DITHER_ENABLED;
+	/* Default to 12 bit truncation without rounding */
+	trunc_round_depth = DCP_OUT_TRUNC_ROUND_DEPTH_12BIT;
+	trunc_mode = DCP_OUT_TRUNC_ROUND_MODE_TRUNCATE;
+
+	if (bit_depth_params->flags.TRUNCATE_ENABLED) {
+		/* Don't enable dithering if truncation is enabled */
+		spatial_dither_enable = false;
+		trunc_mode = bit_depth_params->flags.TRUNCATE_MODE ?
+			     DCP_OUT_TRUNC_ROUND_MODE_ROUND :
+			     DCP_OUT_TRUNC_ROUND_MODE_TRUNCATE;
+
+		if (bit_depth_params->flags.TRUNCATE_DEPTH == 0 ||
+		    bit_depth_params->flags.TRUNCATE_DEPTH == 1)
+			trunc_round_depth = DCP_OUT_TRUNC_ROUND_DEPTH_8BIT;
+		else if (bit_depth_params->flags.TRUNCATE_DEPTH == 2)
+			trunc_round_depth = DCP_OUT_TRUNC_ROUND_DEPTH_10BIT;
+		else {
+			/*
+			 * Invalid truncate/round depth. Setting here to 12bit
+			 * to prevent use-before-initialize errors.
+			 */
+			trunc_round_depth = DCP_OUT_TRUNC_ROUND_DEPTH_12BIT;
+			BREAK_TO_DEBUGGER();
+		}
+	}
+
+	set_clamp(xfm_dce, depth);
+	set_round(xfm_dce, trunc_mode, trunc_round_depth);
+	set_dither(xfm_dce,
+		   spatial_dither_enable,
+		   DCP_SPATIAL_DITHER_MODE_A_AA_A,
+		   DCP_SPATIAL_DITHER_DEPTH_30BPP,
+		   bit_depth_params->flags.FRAME_RANDOM,
+		   bit_depth_params->flags.RGB_RANDOM,
+		   bit_depth_params->flags.HIGHPASS_RANDOM);
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+/*****************************************************************************
+ * dce60_transform_bit_depth_reduction program
+ *
+ * @brief
+ *     Programs the DCP bit depth reduction registers (Clamp, Round/Truncate,
+ *      Dither) for dce
+ *
+ * @param depth : bit depth to set the clamp to (should match denorm)
+ *
+ ******************************************************************************/
+static void dce60_program_bit_depth_reduction(
+	struct dce_transform *xfm_dce,
+	enum dc_color_depth depth,
+	const struct bit_depth_reduction_params *bit_depth_params)
+{
+	enum dcp_out_trunc_round_depth trunc_round_depth;
+	enum dcp_out_trunc_round_mode trunc_mode;
+	bool spatial_dither_enable;
+
+	ASSERT(depth <= COLOR_DEPTH_121212); /* Invalid clamp bit depth */
+
+	spatial_dither_enable = bit_depth_params->flags.SPATIAL_DITHER_ENABLED;
+	/* Default to 12 bit truncation without rounding */
+	trunc_round_depth = DCP_OUT_TRUNC_ROUND_DEPTH_12BIT;
+	trunc_mode = DCP_OUT_TRUNC_ROUND_MODE_TRUNCATE;
+
+	if (bit_depth_params->flags.TRUNCATE_ENABLED) {
+		/* Don't enable dithering if truncation is enabled */
+		spatial_dither_enable = false;
+		trunc_mode = bit_depth_params->flags.TRUNCATE_MODE ?
+			     DCP_OUT_TRUNC_ROUND_MODE_ROUND :
+			     DCP_OUT_TRUNC_ROUND_MODE_TRUNCATE;
+
+		if (bit_depth_params->flags.TRUNCATE_DEPTH == 0 ||
+		    bit_depth_params->flags.TRUNCATE_DEPTH == 1)
+			trunc_round_depth = DCP_OUT_TRUNC_ROUND_DEPTH_8BIT;
+		else if (bit_depth_params->flags.TRUNCATE_DEPTH == 2)
+			trunc_round_depth = DCP_OUT_TRUNC_ROUND_DEPTH_10BIT;
+		else {
+			/*
+			 * Invalid truncate/round depth. Setting here to 12bit
+			 * to prevent use-before-initialize errors.
+			 */
+			trunc_round_depth = DCP_OUT_TRUNC_ROUND_DEPTH_12BIT;
+			BREAK_TO_DEBUGGER();
+		}
+	}
+
+	/* DCE6 has no OUT_CLAMP_CONTROL_* registers - set_clamp() is skipped */
+	set_round(xfm_dce, trunc_mode, trunc_round_depth);
+	set_dither(xfm_dce,
+		   spatial_dither_enable,
+		   DCP_SPATIAL_DITHER_MODE_A_AA_A,
+		   DCP_SPATIAL_DITHER_DEPTH_30BPP,
+		   bit_depth_params->flags.FRAME_RANDOM,
+		   bit_depth_params->flags.RGB_RANDOM,
+		   bit_depth_params->flags.HIGHPASS_RANDOM);
+}
+#endif
+
+static int dce_transform_get_max_num_of_supported_lines(
+	struct dce_transform *xfm_dce,
+	enum lb_pixel_depth depth,
+	int pixel_width)
+{
+	int pixels_per_entries = 0;
+	int max_pixels_supports = 0;
+
+	ASSERT(pixel_width);
+
+	/* Find number of pixels that can fit into a single LB entry and
+	 * take floor of the value since we cannot store a single pixel
+	 * across multiple entries. */
+	switch (depth) {
+	case LB_PIXEL_DEPTH_18BPP:
+		pixels_per_entries = xfm_dce->lb_bits_per_entry / 18;
+		break;
+
+	case LB_PIXEL_DEPTH_24BPP:
+		pixels_per_entries = xfm_dce->lb_bits_per_entry / 24;
+		break;
+
+	case LB_PIXEL_DEPTH_30BPP:
+		pixels_per_entries = xfm_dce->lb_bits_per_entry / 30;
+		break;
+
+	case LB_PIXEL_DEPTH_36BPP:
+		pixels_per_entries = xfm_dce->lb_bits_per_entry / 36;
+		break;
+
+	default:
+		DC_LOG_WARNING("%s: Invalid LB pixel depth",
+			__func__);
+		BREAK_TO_DEBUGGER();
+		break;
+	}
+
+	ASSERT(pixels_per_entries);
+
+	max_pixels_supports =
+			pixels_per_entries *
+			xfm_dce->lb_memory_size;
+
+	return (max_pixels_supports / pixel_width);
+}
+
+static void set_denormalization(
+	struct dce_transform *xfm_dce,
+	enum dc_color_depth depth)
+{
+	int denorm_mode = 0;
+
+	switch (depth) {
+	case COLOR_DEPTH_666:
+		/* 63/64 for 6 bit output color depth */
+		denorm_mode = 1;
+		break;
+	case COLOR_DEPTH_888:
+		/* Unity for 8 bit output color depth
+		 * because prescale is disabled by default */
+		denorm_mode = 0;
+		break;
+	case COLOR_DEPTH_101010:
+		/* 1023/1024 for 10 bit output color depth */
+		denorm_mode = 3;
+		break;
+	case COLOR_DEPTH_121212:
+		/* 4095/4096 for 12 bit output color depth */
+		denorm_mode = 5;
+		break;
+	case COLOR_DEPTH_141414:
+	case COLOR_DEPTH_161616:
+	default:
+		/* not valid used case! */
+		break;
+	}
+
+	REG_SET(DENORM_CONTROL, 0, DENORM_MODE, denorm_mode);
+}
+
+static void dce_transform_set_pixel_storage_depth(
+	struct transform *xfm,
+	enum lb_pixel_depth depth,
+	const struct bit_depth_reduction_params *bit_depth_params)
+{
+	struct dce_transform *xfm_dce = TO_DCE_TRANSFORM(xfm);
+	int pixel_depth, expan_mode;
+	enum dc_color_depth color_depth;
+
+	switch (depth) {
+	case LB_PIXEL_DEPTH_18BPP:
+		color_depth = COLOR_DEPTH_666;
+		pixel_depth = 2;
+		expan_mode  = 1;
+		break;
+	case LB_PIXEL_DEPTH_24BPP:
+		color_depth = COLOR_DEPTH_888;
+		pixel_depth = 1;
+		expan_mode  = 1;
+		break;
+	case LB_PIXEL_DEPTH_30BPP:
+		color_depth = COLOR_DEPTH_101010;
+		pixel_depth = 0;
+		expan_mode  = 1;
+		break;
+	case LB_PIXEL_DEPTH_36BPP:
+		color_depth = COLOR_DEPTH_121212;
+		pixel_depth = 3;
+		expan_mode  = 0;
+		break;
+	default:
+		color_depth = COLOR_DEPTH_101010;
+		pixel_depth = 0;
+		expan_mode  = 1;
+		DC_LOG_DC("The pixel depth %d is not valid, set COLOR_DEPTH_101010 instead.", depth);
+		break;
+	}
+
+	set_denormalization(xfm_dce, color_depth);
+	program_bit_depth_reduction(xfm_dce, color_depth, bit_depth_params);
+
+	REG_UPDATE_2(LB_DATA_FORMAT,
+			PIXEL_DEPTH, pixel_depth,
+			PIXEL_EXPAN_MODE, expan_mode);
+
+	if (!(xfm_dce->lb_pixel_depth_supported & depth)) {
+		/*we should use unsupported capabilities
+		 *  unless it is required by w/a*/
+		DC_LOG_DC("%s: Capability not supported", __func__);
+	}
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static void dce60_transform_set_pixel_storage_depth(
+	struct transform *xfm,
+	enum lb_pixel_depth depth,
+	const struct bit_depth_reduction_params *bit_depth_params)
+{
+	struct dce_transform *xfm_dce = TO_DCE_TRANSFORM(xfm);
+	enum dc_color_depth color_depth;
+
+	switch (depth) {
+	case LB_PIXEL_DEPTH_18BPP:
+		color_depth = COLOR_DEPTH_666;
+		break;
+	case LB_PIXEL_DEPTH_24BPP:
+		color_depth = COLOR_DEPTH_888;
+		break;
+	case LB_PIXEL_DEPTH_30BPP:
+		color_depth = COLOR_DEPTH_101010;
+		break;
+	case LB_PIXEL_DEPTH_36BPP:
+		color_depth = COLOR_DEPTH_121212;
+		break;
+	default:
+		color_depth = COLOR_DEPTH_101010;
+		BREAK_TO_DEBUGGER();
+		break;
+	}
+
+	set_denormalization(xfm_dce, color_depth);
+	dce60_program_bit_depth_reduction(xfm_dce, color_depth, bit_depth_params);
+
+	/* DATA_FORMAT in DCE6 does not have PIXEL_DEPTH and PIXEL_EXPAN_MODE masks */
+
+	if (!(xfm_dce->lb_pixel_depth_supported & depth)) {
+		/*we should use unsupported capabilities
+		 *  unless it is required by w/a*/
+		DC_LOG_WARNING("%s: Capability not supported",
+			__func__);
+	}
+}
+#endif
+
+static void program_gamut_remap(
+	struct dce_transform *xfm_dce,
+	const uint16_t *reg_val)
+{
+	if (reg_val) {
+		REG_SET_2(GAMUT_REMAP_C11_C12, 0,
+				GAMUT_REMAP_C11, reg_val[0],
+				GAMUT_REMAP_C12, reg_val[1]);
+		REG_SET_2(GAMUT_REMAP_C13_C14, 0,
+				GAMUT_REMAP_C13, reg_val[2],
+				GAMUT_REMAP_C14, reg_val[3]);
+		REG_SET_2(GAMUT_REMAP_C21_C22, 0,
+				GAMUT_REMAP_C21, reg_val[4],
+				GAMUT_REMAP_C22, reg_val[5]);
+		REG_SET_2(GAMUT_REMAP_C23_C24, 0,
+				GAMUT_REMAP_C23, reg_val[6],
+				GAMUT_REMAP_C24, reg_val[7]);
+		REG_SET_2(GAMUT_REMAP_C31_C32, 0,
+				GAMUT_REMAP_C31, reg_val[8],
+				GAMUT_REMAP_C32, reg_val[9]);
+		REG_SET_2(GAMUT_REMAP_C33_C34, 0,
+				GAMUT_REMAP_C33, reg_val[10],
+				GAMUT_REMAP_C34, reg_val[11]);
+
+		REG_SET(GAMUT_REMAP_CONTROL, 0, GRPH_GAMUT_REMAP_MODE, 1);
+	} else
+		REG_SET(GAMUT_REMAP_CONTROL, 0, GRPH_GAMUT_REMAP_MODE, 0);
+
+}
+
+/*
+ *****************************************************************************
+ *  Function: dal_transform_wide_gamut_set_gamut_remap
+ *
+ *  @param [in] const struct xfm_grph_csc_adjustment *adjust
+ *
+ *  @return
+ *     void
+ *
+ *  @note calculate and apply color temperature adjustment to in Rgb color space
+ *
+ *  @see
+ *
+ *****************************************************************************
+ */
+static void dce_transform_set_gamut_remap(
+	struct transform *xfm,
+	const struct xfm_grph_csc_adjustment *adjust)
+{
+	struct dce_transform *xfm_dce = TO_DCE_TRANSFORM(xfm);
+	int i = 0;
+
+	if (adjust->gamut_adjust_type != GRAPHICS_GAMUT_ADJUST_TYPE_SW)
+		/* Bypass if type is bypass or hw */
+		program_gamut_remap(xfm_dce, NULL);
+	else {
+		struct fixed31_32 arr_matrix[GAMUT_MATRIX_SIZE];
+		uint16_t arr_reg_val[GAMUT_MATRIX_SIZE];
+
+		for (i = 0; i < GAMUT_MATRIX_SIZE; i++)
+			arr_matrix[i] = adjust->temperature_matrix[i];
+
+		convert_float_matrix(
+			arr_reg_val, arr_matrix, GAMUT_MATRIX_SIZE);
+
+		program_gamut_remap(xfm_dce, arr_reg_val);
+	}
+}
+
+static uint32_t decide_taps(struct fixed31_32 ratio, uint32_t in_taps, bool chroma)
+{
+	uint32_t taps;
+
+	if (IDENTITY_RATIO(ratio)) {
+		return 1;
+	} else if (in_taps != 0) {
+		taps = in_taps;
+	} else {
+		taps = 4;
+	}
+
+	if (chroma) {
+		taps /= 2;
+		if (taps < 2)
+			taps = 2;
+	}
+
+	return taps;
+}
+
+
+bool dce_transform_get_optimal_number_of_taps(
+	struct transform *xfm,
+	struct scaler_data *scl_data,
+	const struct scaling_taps *in_taps)
+{
+	struct dce_transform *xfm_dce = TO_DCE_TRANSFORM(xfm);
+	int pixel_width = scl_data->viewport.width;
+	int max_num_of_lines;
+
+	if (xfm_dce->prescaler_on &&
+			(scl_data->viewport.width > scl_data->recout.width))
+		pixel_width = scl_data->recout.width;
+
+	max_num_of_lines = dce_transform_get_max_num_of_supported_lines(
+		xfm_dce,
+		scl_data->lb_params.depth,
+		pixel_width);
+
+	/* Fail if in_taps are impossible */
+	if (in_taps->v_taps >= max_num_of_lines)
+		return false;
+
+	/*
+	 * Set taps according to this policy (in this order)
+	 * - Use 1 for no scaling
+	 * - Use input taps
+	 * - Use 4 and reduce as required by line buffer size
+	 * - Decide chroma taps if chroma is scaled
+	 *
+	 * Ignore input chroma taps. Decide based on non-chroma
+	 */
+	scl_data->taps.h_taps = decide_taps(scl_data->ratios.horz, in_taps->h_taps, false);
+	scl_data->taps.v_taps = decide_taps(scl_data->ratios.vert, in_taps->v_taps, false);
+	scl_data->taps.h_taps_c = decide_taps(scl_data->ratios.horz_c, in_taps->h_taps, true);
+	scl_data->taps.v_taps_c = decide_taps(scl_data->ratios.vert_c, in_taps->v_taps, true);
+
+	if (!IDENTITY_RATIO(scl_data->ratios.vert)) {
+		/* reduce v_taps if needed but ensure we have at least two */
+		if (in_taps->v_taps == 0
+				&& max_num_of_lines <= scl_data->taps.v_taps
+				&& scl_data->taps.v_taps > 1) {
+			scl_data->taps.v_taps = max_num_of_lines - 1;
+		}
+
+		if (scl_data->taps.v_taps <= 1)
+			return false;
+	}
+
+	if (!IDENTITY_RATIO(scl_data->ratios.vert_c)) {
+		/* reduce chroma v_taps if needed but ensure we have at least two */
+		if (max_num_of_lines <= scl_data->taps.v_taps_c && scl_data->taps.v_taps_c > 1) {
+			scl_data->taps.v_taps_c = max_num_of_lines - 1;
+		}
+
+		if (scl_data->taps.v_taps_c <= 1)
+			return false;
+	}
+
+	/* we've got valid taps */
+	return true;
+}
+
+static void dce_transform_reset(struct transform *xfm)
+{
+	struct dce_transform *xfm_dce = TO_DCE_TRANSFORM(xfm);
+
+	xfm_dce->filter_h = NULL;
+	xfm_dce->filter_v = NULL;
+}
+
+static void program_color_matrix(
+	struct dce_transform *xfm_dce,
+	const struct out_csc_color_matrix *tbl_entry,
+	enum grph_color_adjust_option options)
+{
+	{
+		REG_SET_2(OUTPUT_CSC_C11_C12, 0,
+			OUTPUT_CSC_C11, tbl_entry->regval[0],
+			OUTPUT_CSC_C12, tbl_entry->regval[1]);
+	}
+	{
+		REG_SET_2(OUTPUT_CSC_C13_C14, 0,
+			OUTPUT_CSC_C11, tbl_entry->regval[2],
+			OUTPUT_CSC_C12, tbl_entry->regval[3]);
+	}
+	{
+		REG_SET_2(OUTPUT_CSC_C21_C22, 0,
+			OUTPUT_CSC_C11, tbl_entry->regval[4],
+			OUTPUT_CSC_C12, tbl_entry->regval[5]);
+	}
+	{
+		REG_SET_2(OUTPUT_CSC_C23_C24, 0,
+			OUTPUT_CSC_C11, tbl_entry->regval[6],
+			OUTPUT_CSC_C12, tbl_entry->regval[7]);
+	}
+	{
+		REG_SET_2(OUTPUT_CSC_C31_C32, 0,
+			OUTPUT_CSC_C11, tbl_entry->regval[8],
+			OUTPUT_CSC_C12, tbl_entry->regval[9]);
+	}
+	{
+		REG_SET_2(OUTPUT_CSC_C33_C34, 0,
+			OUTPUT_CSC_C11, tbl_entry->regval[10],
+			OUTPUT_CSC_C12, tbl_entry->regval[11]);
+	}
+}
+
+static bool configure_graphics_mode(
+	struct dce_transform *xfm_dce,
+	enum csc_color_mode config,
+	enum graphics_csc_adjust_type csc_adjust_type,
+	enum dc_color_space color_space)
+{
+	REG_SET(OUTPUT_CSC_CONTROL, 0,
+		OUTPUT_CSC_GRPH_MODE, 0);
+
+	if (csc_adjust_type == GRAPHICS_CSC_ADJUST_TYPE_SW) {
+		if (config == CSC_COLOR_MODE_GRAPHICS_OUTPUT_CSC) {
+			REG_SET(OUTPUT_CSC_CONTROL, 0,
+				OUTPUT_CSC_GRPH_MODE, 4);
+		} else {
+
+			switch (color_space) {
+			case COLOR_SPACE_SRGB:
+				/* by pass */
+				REG_SET(OUTPUT_CSC_CONTROL, 0,
+					OUTPUT_CSC_GRPH_MODE, 0);
+				break;
+			case COLOR_SPACE_SRGB_LIMITED:
+				/* TV RGB */
+				REG_SET(OUTPUT_CSC_CONTROL, 0,
+					OUTPUT_CSC_GRPH_MODE, 1);
+				break;
+			case COLOR_SPACE_YCBCR601:
+			case COLOR_SPACE_YCBCR601_LIMITED:
+				/* YCbCr601 */
+				REG_SET(OUTPUT_CSC_CONTROL, 0,
+					OUTPUT_CSC_GRPH_MODE, 2);
+				break;
+			case COLOR_SPACE_YCBCR709:
+			case COLOR_SPACE_YCBCR709_LIMITED:
+				/* YCbCr709 */
+				REG_SET(OUTPUT_CSC_CONTROL, 0,
+					OUTPUT_CSC_GRPH_MODE, 3);
+				break;
+			default:
+				return false;
+			}
+		}
+	} else if (csc_adjust_type == GRAPHICS_CSC_ADJUST_TYPE_HW) {
+		switch (color_space) {
+		case COLOR_SPACE_SRGB:
+			/* by pass */
+			REG_SET(OUTPUT_CSC_CONTROL, 0,
+				OUTPUT_CSC_GRPH_MODE, 0);
+			break;
+		case COLOR_SPACE_SRGB_LIMITED:
+			/* TV RGB */
+			REG_SET(OUTPUT_CSC_CONTROL, 0,
+				OUTPUT_CSC_GRPH_MODE, 1);
+			break;
+		case COLOR_SPACE_YCBCR601:
+		case COLOR_SPACE_YCBCR601_LIMITED:
+			/* YCbCr601 */
+			REG_SET(OUTPUT_CSC_CONTROL, 0,
+				OUTPUT_CSC_GRPH_MODE, 2);
+			break;
+		case COLOR_SPACE_YCBCR709:
+		case COLOR_SPACE_YCBCR709_LIMITED:
+			 /* YCbCr709 */
+			REG_SET(OUTPUT_CSC_CONTROL, 0,
+				OUTPUT_CSC_GRPH_MODE, 3);
+			break;
+		default:
+			return false;
+		}
+
+	} else
+		/* by pass */
+		REG_SET(OUTPUT_CSC_CONTROL, 0,
+			OUTPUT_CSC_GRPH_MODE, 0);
+
+	return true;
+}
+
+void dce110_opp_set_csc_adjustment(
+	struct transform *xfm,
+	const struct out_csc_color_matrix *tbl_entry)
+{
+	struct dce_transform *xfm_dce = TO_DCE_TRANSFORM(xfm);
+	enum csc_color_mode config =
+			CSC_COLOR_MODE_GRAPHICS_OUTPUT_CSC;
+
+	program_color_matrix(
+			xfm_dce, tbl_entry, GRPH_COLOR_MATRIX_SW);
+
+	/*  We did everything ,now program DxOUTPUT_CSC_CONTROL */
+	configure_graphics_mode(xfm_dce, config, GRAPHICS_CSC_ADJUST_TYPE_SW,
+			tbl_entry->color_space);
+}
+
+void dce110_opp_set_csc_default(
+	struct transform *xfm,
+	const struct default_adjustment *default_adjust)
+{
+	struct dce_transform *xfm_dce = TO_DCE_TRANSFORM(xfm);
+	enum csc_color_mode config =
+			CSC_COLOR_MODE_GRAPHICS_PREDEFINED;
+
+	if (default_adjust->force_hw_default == false) {
+		const struct out_csc_color_matrix *elm;
+		/* currently parameter not in use */
+		enum grph_color_adjust_option option =
+			GRPH_COLOR_MATRIX_HW_DEFAULT;
+		uint32_t i;
+		/*
+		 * HW default false we program locally defined matrix
+		 * HW default true  we use predefined hw matrix and we
+		 * do not need to program matrix
+		 * OEM wants the HW default via runtime parameter.
+		 */
+		option = GRPH_COLOR_MATRIX_SW;
+
+		for (i = 0; i < ARRAY_SIZE(global_color_matrix); ++i) {
+			elm = &global_color_matrix[i];
+			if (elm->color_space != default_adjust->out_color_space)
+				continue;
+			/* program the matrix with default values from this
+			 * file */
+			program_color_matrix(xfm_dce, elm, option);
+			config = CSC_COLOR_MODE_GRAPHICS_OUTPUT_CSC;
+			break;
+		}
+	}
+
+	/* configure the what we programmed :
+	 * 1. Default values from this file
+	 * 2. Use hardware default from ROM_A and we do not need to program
+	 * matrix */
+
+	configure_graphics_mode(xfm_dce, config,
+		default_adjust->csc_adjust_type,
+		default_adjust->out_color_space);
+}
+
+static void program_pwl(struct dce_transform *xfm_dce,
+			const struct pwl_params *params)
+{
+	uint32_t retval;
+	uint8_t max_tries = 10;
+	uint8_t counter = 0;
+	uint32_t i = 0;
+	const struct pwl_result_data *rgb = params->rgb_resulted;
+
+	/* Power on LUT memory */
+	if (REG(DCFE_MEM_PWR_CTRL))
+		REG_UPDATE(DCFE_MEM_PWR_CTRL,
+			   DCP_REGAMMA_MEM_PWR_DIS, 1);
+	else
+		REG_UPDATE(DCFE_MEM_LIGHT_SLEEP_CNTL,
+			   REGAMMA_LUT_LIGHT_SLEEP_DIS, 1);
+
+	while (counter < max_tries) {
+		if (REG(DCFE_MEM_PWR_STATUS)) {
+			REG_GET(DCFE_MEM_PWR_STATUS,
+				DCP_REGAMMA_MEM_PWR_STATE,
+				&retval);
+
+			if (retval == 0)
+				break;
+			++counter;
+		} else {
+			REG_GET(DCFE_MEM_LIGHT_SLEEP_CNTL,
+				REGAMMA_LUT_MEM_PWR_STATE,
+				&retval);
+
+			if (retval == 0)
+				break;
+			++counter;
+		}
+	}
+
+	if (counter == max_tries) {
+		DC_LOG_WARNING("%s: regamma lut was not powered on "
+				"in a timely manner,"
+				" programming still proceeds\n",
+				__func__);
+	}
+
+	REG_UPDATE(REGAMMA_LUT_WRITE_EN_MASK,
+		   REGAMMA_LUT_WRITE_EN_MASK, 7);
+
+	REG_WRITE(REGAMMA_LUT_INDEX, 0);
+
+	/* Program REGAMMA_LUT_DATA */
+	while (i != params->hw_points_num) {
+
+		REG_WRITE(REGAMMA_LUT_DATA, rgb->red_reg);
+		REG_WRITE(REGAMMA_LUT_DATA, rgb->green_reg);
+		REG_WRITE(REGAMMA_LUT_DATA, rgb->blue_reg);
+		REG_WRITE(REGAMMA_LUT_DATA, rgb->delta_red_reg);
+		REG_WRITE(REGAMMA_LUT_DATA, rgb->delta_green_reg);
+		REG_WRITE(REGAMMA_LUT_DATA, rgb->delta_blue_reg);
+
+		++rgb;
+		++i;
+	}
+
+	/*  we are done with DCP LUT memory; re-enable low power mode */
+	if (REG(DCFE_MEM_PWR_CTRL))
+		REG_UPDATE(DCFE_MEM_PWR_CTRL,
+			   DCP_REGAMMA_MEM_PWR_DIS, 0);
+	else
+		REG_UPDATE(DCFE_MEM_LIGHT_SLEEP_CNTL,
+			   REGAMMA_LUT_LIGHT_SLEEP_DIS, 0);
+}
+
+static void regamma_config_regions_and_segments(struct dce_transform *xfm_dce,
+						const struct pwl_params *params)
+{
+	const struct gamma_curve *curve;
+
+	REG_SET_2(REGAMMA_CNTLA_START_CNTL, 0,
+		  REGAMMA_CNTLA_EXP_REGION_START, params->arr_points[0].custom_float_x,
+		  REGAMMA_CNTLA_EXP_REGION_START_SEGMENT, 0);
+
+	REG_SET(REGAMMA_CNTLA_SLOPE_CNTL, 0,
+		REGAMMA_CNTLA_EXP_REGION_LINEAR_SLOPE, params->arr_points[0].custom_float_slope);
+
+	REG_SET(REGAMMA_CNTLA_END_CNTL1, 0,
+		REGAMMA_CNTLA_EXP_REGION_END, params->arr_points[1].custom_float_x);
+
+	REG_SET_2(REGAMMA_CNTLA_END_CNTL2, 0,
+		  REGAMMA_CNTLA_EXP_REGION_END_BASE, params->arr_points[1].custom_float_y,
+		  REGAMMA_CNTLA_EXP_REGION_END_SLOPE, params->arr_points[1].custom_float_slope);
+
+	curve = params->arr_curve_points;
+
+	REG_SET_4(REGAMMA_CNTLA_REGION_0_1, 0,
+		  REGAMMA_CNTLA_EXP_REGION0_LUT_OFFSET, curve[0].offset,
+		  REGAMMA_CNTLA_EXP_REGION0_NUM_SEGMENTS, curve[0].segments_num,
+		  REGAMMA_CNTLA_EXP_REGION1_LUT_OFFSET, curve[1].offset,
+		  REGAMMA_CNTLA_EXP_REGION1_NUM_SEGMENTS, curve[1].segments_num);
+	curve += 2;
+
+	REG_SET_4(REGAMMA_CNTLA_REGION_2_3, 0,
+		  REGAMMA_CNTLA_EXP_REGION0_LUT_OFFSET, curve[0].offset,
+		  REGAMMA_CNTLA_EXP_REGION0_NUM_SEGMENTS, curve[0].segments_num,
+		  REGAMMA_CNTLA_EXP_REGION1_LUT_OFFSET, curve[1].offset,
+		  REGAMMA_CNTLA_EXP_REGION1_NUM_SEGMENTS, curve[1].segments_num);
+	curve += 2;
+
+	REG_SET_4(REGAMMA_CNTLA_REGION_4_5, 0,
+		  REGAMMA_CNTLA_EXP_REGION0_LUT_OFFSET, curve[0].offset,
+		  REGAMMA_CNTLA_EXP_REGION0_NUM_SEGMENTS, curve[0].segments_num,
+		  REGAMMA_CNTLA_EXP_REGION1_LUT_OFFSET, curve[1].offset,
+		  REGAMMA_CNTLA_EXP_REGION1_NUM_SEGMENTS, curve[1].segments_num);
+	curve += 2;
+
+	REG_SET_4(REGAMMA_CNTLA_REGION_6_7, 0,
+		  REGAMMA_CNTLA_EXP_REGION0_LUT_OFFSET, curve[0].offset,
+		  REGAMMA_CNTLA_EXP_REGION0_NUM_SEGMENTS, curve[0].segments_num,
+		  REGAMMA_CNTLA_EXP_REGION1_LUT_OFFSET, curve[1].offset,
+		  REGAMMA_CNTLA_EXP_REGION1_NUM_SEGMENTS, curve[1].segments_num);
+	curve += 2;
+
+	REG_SET_4(REGAMMA_CNTLA_REGION_8_9, 0,
+		  REGAMMA_CNTLA_EXP_REGION0_LUT_OFFSET, curve[0].offset,
+		  REGAMMA_CNTLA_EXP_REGION0_NUM_SEGMENTS, curve[0].segments_num,
+		  REGAMMA_CNTLA_EXP_REGION1_LUT_OFFSET, curve[1].offset,
+		  REGAMMA_CNTLA_EXP_REGION1_NUM_SEGMENTS, curve[1].segments_num);
+	curve += 2;
+
+	REG_SET_4(REGAMMA_CNTLA_REGION_10_11, 0,
+		  REGAMMA_CNTLA_EXP_REGION0_LUT_OFFSET, curve[0].offset,
+		  REGAMMA_CNTLA_EXP_REGION0_NUM_SEGMENTS, curve[0].segments_num,
+		  REGAMMA_CNTLA_EXP_REGION1_LUT_OFFSET, curve[1].offset,
+		  REGAMMA_CNTLA_EXP_REGION1_NUM_SEGMENTS, curve[1].segments_num);
+	curve += 2;
+
+	REG_SET_4(REGAMMA_CNTLA_REGION_12_13, 0,
+		  REGAMMA_CNTLA_EXP_REGION0_LUT_OFFSET, curve[0].offset,
+		  REGAMMA_CNTLA_EXP_REGION0_NUM_SEGMENTS, curve[0].segments_num,
+		  REGAMMA_CNTLA_EXP_REGION1_LUT_OFFSET, curve[1].offset,
+		  REGAMMA_CNTLA_EXP_REGION1_NUM_SEGMENTS, curve[1].segments_num);
+	curve += 2;
+
+	REG_SET_4(REGAMMA_CNTLA_REGION_14_15, 0,
+		  REGAMMA_CNTLA_EXP_REGION0_LUT_OFFSET, curve[0].offset,
+		  REGAMMA_CNTLA_EXP_REGION0_NUM_SEGMENTS, curve[0].segments_num,
+		  REGAMMA_CNTLA_EXP_REGION1_LUT_OFFSET, curve[1].offset,
+		  REGAMMA_CNTLA_EXP_REGION1_NUM_SEGMENTS, curve[1].segments_num);
+}
+
+
+
+void dce110_opp_program_regamma_pwl(struct transform *xfm,
+				    const struct pwl_params *params)
+{
+	struct dce_transform *xfm_dce = TO_DCE_TRANSFORM(xfm);
+
+	/* Setup regions */
+	regamma_config_regions_and_segments(xfm_dce, params);
+
+	/* Program PWL */
+	program_pwl(xfm_dce, params);
+}
+
+void dce110_opp_power_on_regamma_lut(struct transform *xfm,
+				     bool power_on)
+{
+	struct dce_transform *xfm_dce = TO_DCE_TRANSFORM(xfm);
+
+	if (REG(DCFE_MEM_PWR_CTRL))
+		REG_UPDATE_2(DCFE_MEM_PWR_CTRL,
+			     DCP_REGAMMA_MEM_PWR_DIS, power_on,
+			     DCP_LUT_MEM_PWR_DIS, power_on);
+	else
+		REG_UPDATE_2(DCFE_MEM_LIGHT_SLEEP_CNTL,
+			    REGAMMA_LUT_LIGHT_SLEEP_DIS, power_on,
+			    DCP_LUT_LIGHT_SLEEP_DIS, power_on);
+
+}
+
+void dce110_opp_set_regamma_mode(struct transform *xfm,
+				 enum opp_regamma mode)
+{
+	struct dce_transform *xfm_dce = TO_DCE_TRANSFORM(xfm);
+
+	REG_SET(REGAMMA_CONTROL, 0,
+		GRPH_REGAMMA_MODE, mode);
+}
+
+static const struct transform_funcs dce_transform_funcs = {
+	.transform_reset = dce_transform_reset,
+	.transform_set_scaler = dce_transform_set_scaler,
+	.transform_set_gamut_remap = dce_transform_set_gamut_remap,
+	.opp_set_csc_adjustment = dce110_opp_set_csc_adjustment,
+	.opp_set_csc_default = dce110_opp_set_csc_default,
+	.opp_power_on_regamma_lut = dce110_opp_power_on_regamma_lut,
+	.opp_program_regamma_pwl = dce110_opp_program_regamma_pwl,
+	.opp_set_regamma_mode = dce110_opp_set_regamma_mode,
+	.transform_set_pixel_storage_depth = dce_transform_set_pixel_storage_depth,
+	.transform_get_optimal_number_of_taps = dce_transform_get_optimal_number_of_taps
+};
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+static const struct transform_funcs dce60_transform_funcs = {
+	.transform_reset = dce_transform_reset,
+	.transform_set_scaler = dce60_transform_set_scaler,
+	.transform_set_gamut_remap = dce_transform_set_gamut_remap,
+	.opp_set_csc_adjustment = dce110_opp_set_csc_adjustment,
+	.opp_set_csc_default = dce110_opp_set_csc_default,
+	.opp_power_on_regamma_lut = dce110_opp_power_on_regamma_lut,
+	.opp_program_regamma_pwl = dce110_opp_program_regamma_pwl,
+	.opp_set_regamma_mode = dce110_opp_set_regamma_mode,
+	.transform_set_pixel_storage_depth = dce60_transform_set_pixel_storage_depth,
+	.transform_get_optimal_number_of_taps = dce_transform_get_optimal_number_of_taps
+};
+#endif
+
+/*****************************************/
+/* Constructor, Destructor               */
+/*****************************************/
+
+void dce_transform_construct(
+	struct dce_transform *xfm_dce,
+	struct dc_context *ctx,
+	uint32_t inst,
+	const struct dce_transform_registers *regs,
+	const struct dce_transform_shift *xfm_shift,
+	const struct dce_transform_mask *xfm_mask)
+{
+	xfm_dce->base.ctx = ctx;
+
+	xfm_dce->base.inst = inst;
+	xfm_dce->base.funcs = &dce_transform_funcs;
+
+	xfm_dce->regs = regs;
+	xfm_dce->xfm_shift = xfm_shift;
+	xfm_dce->xfm_mask = xfm_mask;
+
+	xfm_dce->prescaler_on = true;
+	xfm_dce->lb_pixel_depth_supported =
+			LB_PIXEL_DEPTH_18BPP |
+			LB_PIXEL_DEPTH_24BPP |
+			LB_PIXEL_DEPTH_30BPP |
+			LB_PIXEL_DEPTH_36BPP;
+
+	xfm_dce->lb_bits_per_entry = LB_BITS_PER_ENTRY;
+	xfm_dce->lb_memory_size = LB_TOTAL_NUMBER_OF_ENTRIES; /*0x6B0*/
+}
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+void dce60_transform_construct(
+	struct dce_transform *xfm_dce,
+	struct dc_context *ctx,
+	uint32_t inst,
+	const struct dce_transform_registers *regs,
+	const struct dce_transform_shift *xfm_shift,
+	const struct dce_transform_mask *xfm_mask)
+{
+	xfm_dce->base.ctx = ctx;
+
+	xfm_dce->base.inst = inst;
+	xfm_dce->base.funcs = &dce60_transform_funcs;
+
+	xfm_dce->regs = regs;
+	xfm_dce->xfm_shift = xfm_shift;
+	xfm_dce->xfm_mask = xfm_mask;
+
+	xfm_dce->prescaler_on = true;
+	xfm_dce->lb_pixel_depth_supported =
+			LB_PIXEL_DEPTH_18BPP |
+			LB_PIXEL_DEPTH_24BPP |
+			LB_PIXEL_DEPTH_30BPP |
+			LB_PIXEL_DEPTH_36BPP;
+
+	xfm_dce->lb_bits_per_entry = LB_BITS_PER_ENTRY;
+	xfm_dce->lb_memory_size = LB_TOTAL_NUMBER_OF_ENTRIES; /*0x6B0*/
+}
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_transform.h b/drivers/gpu/drm/amd/display/dc/dce/dce_transform.h
new file mode 100644
index 000000000000..eb716e8337e2
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_transform.h
@@ -0,0 +1,699 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef _DCE_DCE_TRANSFORM_H_
+#define _DCE_DCE_TRANSFORM_H_
+
+
+#include "transform.h"
+
+#define TO_DCE_TRANSFORM(transform)\
+	container_of(transform, struct dce_transform, base)
+
+#define LB_TOTAL_NUMBER_OF_ENTRIES 1712
+#define LB_BITS_PER_ENTRY 144
+
+#define XFM_COMMON_REG_LIST_DCE_BASE(id) \
+	SRI(LB_DATA_FORMAT, LB, id), \
+	SRI(GAMUT_REMAP_CONTROL, DCP, id), \
+	SRI(GAMUT_REMAP_C11_C12, DCP, id), \
+	SRI(GAMUT_REMAP_C13_C14, DCP, id), \
+	SRI(GAMUT_REMAP_C21_C22, DCP, id), \
+	SRI(GAMUT_REMAP_C23_C24, DCP, id), \
+	SRI(GAMUT_REMAP_C31_C32, DCP, id), \
+	SRI(GAMUT_REMAP_C33_C34, DCP, id), \
+	SRI(OUTPUT_CSC_C11_C12, DCP, id), \
+	SRI(OUTPUT_CSC_C13_C14, DCP, id), \
+	SRI(OUTPUT_CSC_C21_C22, DCP, id), \
+	SRI(OUTPUT_CSC_C23_C24, DCP, id), \
+	SRI(OUTPUT_CSC_C31_C32, DCP, id), \
+	SRI(OUTPUT_CSC_C33_C34, DCP, id), \
+	SRI(OUTPUT_CSC_CONTROL, DCP, id), \
+	SRI(REGAMMA_CNTLA_START_CNTL, DCP, id), \
+	SRI(REGAMMA_CNTLA_SLOPE_CNTL, DCP, id), \
+	SRI(REGAMMA_CNTLA_END_CNTL1, DCP, id), \
+	SRI(REGAMMA_CNTLA_END_CNTL2, DCP, id), \
+	SRI(REGAMMA_CNTLA_REGION_0_1, DCP, id), \
+	SRI(REGAMMA_CNTLA_REGION_2_3, DCP, id), \
+	SRI(REGAMMA_CNTLA_REGION_4_5, DCP, id), \
+	SRI(REGAMMA_CNTLA_REGION_6_7, DCP, id), \
+	SRI(REGAMMA_CNTLA_REGION_8_9, DCP, id), \
+	SRI(REGAMMA_CNTLA_REGION_10_11, DCP, id), \
+	SRI(REGAMMA_CNTLA_REGION_12_13, DCP, id), \
+	SRI(REGAMMA_CNTLA_REGION_14_15, DCP, id), \
+	SRI(REGAMMA_LUT_WRITE_EN_MASK, DCP, id), \
+	SRI(REGAMMA_LUT_INDEX, DCP, id), \
+	SRI(REGAMMA_LUT_DATA, DCP, id), \
+	SRI(REGAMMA_CONTROL, DCP, id), \
+	SRI(DENORM_CONTROL, DCP, id), \
+	SRI(DCP_SPATIAL_DITHER_CNTL, DCP, id), \
+	SRI(OUT_ROUND_CONTROL, DCP, id), \
+	SRI(OUT_CLAMP_CONTROL_R_CR, DCP, id), \
+	SRI(OUT_CLAMP_CONTROL_G_Y, DCP, id), \
+	SRI(OUT_CLAMP_CONTROL_B_CB, DCP, id), \
+	SRI(SCL_MODE, SCL, id), \
+	SRI(SCL_TAP_CONTROL, SCL, id), \
+	SRI(SCL_CONTROL, SCL, id), \
+	SRI(SCL_BYPASS_CONTROL, SCL, id), \
+	SRI(EXT_OVERSCAN_LEFT_RIGHT, SCL, id), \
+	SRI(EXT_OVERSCAN_TOP_BOTTOM, SCL, id), \
+	SRI(SCL_VERT_FILTER_CONTROL, SCL, id), \
+	SRI(SCL_HORZ_FILTER_CONTROL, SCL, id), \
+	SRI(SCL_COEF_RAM_SELECT, SCL, id), \
+	SRI(SCL_COEF_RAM_TAP_DATA, SCL, id), \
+	SRI(VIEWPORT_START, SCL, id), \
+	SRI(VIEWPORT_SIZE, SCL, id), \
+	SRI(SCL_HORZ_FILTER_SCALE_RATIO, SCL, id), \
+	SRI(SCL_VERT_FILTER_SCALE_RATIO, SCL, id), \
+	SRI(SCL_HORZ_FILTER_INIT, SCL, id), \
+	SRI(SCL_VERT_FILTER_INIT, SCL, id), \
+	SRI(SCL_AUTOMATIC_MODE_CONTROL, SCL, id), \
+	SRI(LB_MEMORY_CTRL, LB, id), \
+	SRI(SCL_UPDATE, SCL, id), \
+	SRI(SCL_F_SHARP_CONTROL, SCL, id)
+
+#define XFM_COMMON_REG_LIST_DCE80(id) \
+	XFM_COMMON_REG_LIST_DCE_BASE(id), \
+	SRI(DCFE_MEM_LIGHT_SLEEP_CNTL, CRTC, id)
+
+#define XFM_COMMON_REG_LIST_DCE100(id) \
+	XFM_COMMON_REG_LIST_DCE_BASE(id), \
+	SRI(DCFE_MEM_PWR_CTRL, CRTC, id), \
+	SRI(DCFE_MEM_PWR_STATUS, CRTC, id)
+
+#define XFM_COMMON_REG_LIST_DCE110(id) \
+	XFM_COMMON_REG_LIST_DCE_BASE(id), \
+	SRI(DCFE_MEM_PWR_CTRL, DCFE, id), \
+	SRI(DCFE_MEM_PWR_STATUS, DCFE, id)
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+#define XFM_COMMON_REG_LIST_DCE60_BASE(id) \
+	SRI(DATA_FORMAT, LB, id), \
+	SRI(GAMUT_REMAP_CONTROL, DCP, id), \
+	SRI(GAMUT_REMAP_C11_C12, DCP, id), \
+	SRI(GAMUT_REMAP_C13_C14, DCP, id), \
+	SRI(GAMUT_REMAP_C21_C22, DCP, id), \
+	SRI(GAMUT_REMAP_C23_C24, DCP, id), \
+	SRI(GAMUT_REMAP_C31_C32, DCP, id), \
+	SRI(GAMUT_REMAP_C33_C34, DCP, id), \
+	SRI(OUTPUT_CSC_C11_C12, DCP, id), \
+	SRI(OUTPUT_CSC_C13_C14, DCP, id), \
+	SRI(OUTPUT_CSC_C21_C22, DCP, id), \
+	SRI(OUTPUT_CSC_C23_C24, DCP, id), \
+	SRI(OUTPUT_CSC_C31_C32, DCP, id), \
+	SRI(OUTPUT_CSC_C33_C34, DCP, id), \
+	SRI(OUTPUT_CSC_CONTROL, DCP, id), \
+	SRI(REGAMMA_CNTLA_START_CNTL, DCP, id), \
+	SRI(REGAMMA_CNTLA_SLOPE_CNTL, DCP, id), \
+	SRI(REGAMMA_CNTLA_END_CNTL1, DCP, id), \
+	SRI(REGAMMA_CNTLA_END_CNTL2, DCP, id), \
+	SRI(REGAMMA_CNTLA_REGION_0_1, DCP, id), \
+	SRI(REGAMMA_CNTLA_REGION_2_3, DCP, id), \
+	SRI(REGAMMA_CNTLA_REGION_4_5, DCP, id), \
+	SRI(REGAMMA_CNTLA_REGION_6_7, DCP, id), \
+	SRI(REGAMMA_CNTLA_REGION_8_9, DCP, id), \
+	SRI(REGAMMA_CNTLA_REGION_10_11, DCP, id), \
+	SRI(REGAMMA_CNTLA_REGION_12_13, DCP, id), \
+	SRI(REGAMMA_CNTLA_REGION_14_15, DCP, id), \
+	SRI(REGAMMA_LUT_WRITE_EN_MASK, DCP, id), \
+	SRI(REGAMMA_LUT_INDEX, DCP, id), \
+	SRI(REGAMMA_LUT_DATA, DCP, id), \
+	SRI(REGAMMA_CONTROL, DCP, id), \
+	SRI(DENORM_CONTROL, DCP, id), \
+	SRI(DCP_SPATIAL_DITHER_CNTL, DCP, id), \
+	SRI(OUT_ROUND_CONTROL, DCP, id), \
+	SRI(SCL_TAP_CONTROL, SCL, id), \
+	SRI(SCL_CONTROL, SCL, id), \
+	SRI(SCL_BYPASS_CONTROL, SCL, id), \
+	SRI(EXT_OVERSCAN_LEFT_RIGHT, SCL, id), \
+	SRI(EXT_OVERSCAN_TOP_BOTTOM, SCL, id), \
+	SRI(SCL_VERT_FILTER_CONTROL, SCL, id), \
+	SRI(SCL_HORZ_FILTER_CONTROL, SCL, id), \
+	SRI(SCL_COEF_RAM_SELECT, SCL, id), \
+	SRI(SCL_COEF_RAM_TAP_DATA, SCL, id), \
+	SRI(VIEWPORT_START, SCL, id), \
+	SRI(VIEWPORT_SIZE, SCL, id), \
+	SRI(SCL_SCALER_ENABLE, SCL, id), \
+	SRI(SCL_HORZ_FILTER_INIT_RGB_LUMA, SCL, id), \
+	SRI(SCL_HORZ_FILTER_INIT_CHROMA, SCL, id), \
+	SRI(SCL_HORZ_FILTER_SCALE_RATIO, SCL, id), \
+	SRI(SCL_VERT_FILTER_SCALE_RATIO, SCL, id), \
+	SRI(SCL_VERT_FILTER_INIT, SCL, id), \
+	SRI(SCL_AUTOMATIC_MODE_CONTROL, SCL, id), \
+	SRI(DC_LB_MEMORY_SPLIT, LB, id), \
+	SRI(DC_LB_MEM_SIZE, LB, id), \
+	SRI(DCFE_MEM_LIGHT_SLEEP_CNTL, CRTC, id), \
+	SRI(SCL_UPDATE, SCL, id), \
+	SRI(SCL_F_SHARP_CONTROL, SCL, id)
+
+#define XFM_COMMON_REG_LIST_DCE60(id) \
+	XFM_COMMON_REG_LIST_DCE60_BASE(id), \
+	SRI(DCFE_MEM_LIGHT_SLEEP_CNTL, CRTC, id)
+#endif
+
+#define XFM_SF(reg_name, field_name, post_fix)\
+	.field_name = reg_name ## __ ## field_name ## post_fix
+
+#define XFM_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(mask_sh) \
+	XFM_SF(OUT_CLAMP_CONTROL_B_CB, OUT_CLAMP_MIN_B_CB, mask_sh), \
+	XFM_SF(OUT_CLAMP_CONTROL_B_CB, OUT_CLAMP_MAX_B_CB, mask_sh), \
+	XFM_SF(OUT_CLAMP_CONTROL_G_Y, OUT_CLAMP_MIN_G_Y, mask_sh), \
+	XFM_SF(OUT_CLAMP_CONTROL_G_Y, OUT_CLAMP_MAX_G_Y, mask_sh), \
+	XFM_SF(OUT_CLAMP_CONTROL_R_CR, OUT_CLAMP_MIN_R_CR, mask_sh), \
+	XFM_SF(OUT_CLAMP_CONTROL_R_CR, OUT_CLAMP_MAX_R_CR, mask_sh), \
+	XFM_SF(OUT_ROUND_CONTROL, OUT_ROUND_TRUNC_MODE, mask_sh), \
+	XFM_SF(DCP_SPATIAL_DITHER_CNTL, DCP_SPATIAL_DITHER_EN, mask_sh), \
+	XFM_SF(DCP_SPATIAL_DITHER_CNTL, DCP_SPATIAL_DITHER_MODE, mask_sh), \
+	XFM_SF(DCP_SPATIAL_DITHER_CNTL, DCP_SPATIAL_DITHER_DEPTH, mask_sh), \
+	XFM_SF(DCP_SPATIAL_DITHER_CNTL, DCP_FRAME_RANDOM_ENABLE, mask_sh), \
+	XFM_SF(DCP_SPATIAL_DITHER_CNTL, DCP_RGB_RANDOM_ENABLE, mask_sh), \
+	XFM_SF(DCP_SPATIAL_DITHER_CNTL, DCP_HIGHPASS_RANDOM_ENABLE, mask_sh), \
+	XFM_SF(DENORM_CONTROL, DENORM_MODE, mask_sh), \
+	XFM_SF(LB_DATA_FORMAT, PIXEL_DEPTH, mask_sh), \
+	XFM_SF(LB_DATA_FORMAT, PIXEL_EXPAN_MODE, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C11_C12, GAMUT_REMAP_C11, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C11_C12, GAMUT_REMAP_C12, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C13_C14, GAMUT_REMAP_C13, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C13_C14, GAMUT_REMAP_C14, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C21_C22, GAMUT_REMAP_C21, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C21_C22, GAMUT_REMAP_C22, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C23_C24, GAMUT_REMAP_C23, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C23_C24, GAMUT_REMAP_C24, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C31_C32, GAMUT_REMAP_C31, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C31_C32, GAMUT_REMAP_C32, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C33_C34, GAMUT_REMAP_C33, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C33_C34, GAMUT_REMAP_C34, mask_sh), \
+	XFM_SF(GAMUT_REMAP_CONTROL, GRPH_GAMUT_REMAP_MODE, mask_sh), \
+	XFM_SF(OUTPUT_CSC_C11_C12, OUTPUT_CSC_C11, mask_sh),\
+	XFM_SF(OUTPUT_CSC_C11_C12, OUTPUT_CSC_C12, mask_sh),\
+	XFM_SF(OUTPUT_CSC_CONTROL, OUTPUT_CSC_GRPH_MODE, mask_sh),\
+	XFM_SF(REGAMMA_CNTLA_START_CNTL, REGAMMA_CNTLA_EXP_REGION_START, mask_sh),\
+	XFM_SF(REGAMMA_CNTLA_START_CNTL, REGAMMA_CNTLA_EXP_REGION_START_SEGMENT, mask_sh),\
+	XFM_SF(REGAMMA_CNTLA_SLOPE_CNTL, REGAMMA_CNTLA_EXP_REGION_LINEAR_SLOPE, mask_sh),\
+	XFM_SF(REGAMMA_CNTLA_END_CNTL1, REGAMMA_CNTLA_EXP_REGION_END, mask_sh),\
+	XFM_SF(REGAMMA_CNTLA_END_CNTL2, REGAMMA_CNTLA_EXP_REGION_END_BASE, mask_sh),\
+	XFM_SF(REGAMMA_CNTLA_END_CNTL2, REGAMMA_CNTLA_EXP_REGION_END_SLOPE, mask_sh),\
+	XFM_SF(REGAMMA_CNTLA_REGION_0_1, REGAMMA_CNTLA_EXP_REGION0_LUT_OFFSET, mask_sh),\
+	XFM_SF(REGAMMA_CNTLA_REGION_0_1, REGAMMA_CNTLA_EXP_REGION0_NUM_SEGMENTS, mask_sh),\
+	XFM_SF(REGAMMA_CNTLA_REGION_0_1, REGAMMA_CNTLA_EXP_REGION1_LUT_OFFSET, mask_sh),\
+	XFM_SF(REGAMMA_CNTLA_REGION_0_1, REGAMMA_CNTLA_EXP_REGION1_NUM_SEGMENTS, mask_sh),\
+	XFM_SF(REGAMMA_LUT_WRITE_EN_MASK, REGAMMA_LUT_WRITE_EN_MASK, mask_sh),\
+	XFM_SF(REGAMMA_CONTROL, GRPH_REGAMMA_MODE, mask_sh),\
+	XFM_SF(SCL_MODE, SCL_MODE, mask_sh), \
+	XFM_SF(SCL_TAP_CONTROL, SCL_H_NUM_OF_TAPS, mask_sh), \
+	XFM_SF(SCL_TAP_CONTROL, SCL_V_NUM_OF_TAPS, mask_sh), \
+	XFM_SF(SCL_CONTROL, SCL_BOUNDARY_MODE, mask_sh), \
+	XFM_SF(SCL_BYPASS_CONTROL, SCL_BYPASS_MODE, mask_sh), \
+	XFM_SF(EXT_OVERSCAN_LEFT_RIGHT, EXT_OVERSCAN_LEFT, mask_sh), \
+	XFM_SF(EXT_OVERSCAN_LEFT_RIGHT, EXT_OVERSCAN_RIGHT, mask_sh), \
+	XFM_SF(EXT_OVERSCAN_TOP_BOTTOM, EXT_OVERSCAN_TOP, mask_sh), \
+	XFM_SF(EXT_OVERSCAN_TOP_BOTTOM, EXT_OVERSCAN_BOTTOM, mask_sh), \
+	XFM_SF(SCL_COEF_RAM_SELECT, SCL_C_RAM_FILTER_TYPE, mask_sh), \
+	XFM_SF(SCL_COEF_RAM_SELECT, SCL_C_RAM_PHASE, mask_sh), \
+	XFM_SF(SCL_COEF_RAM_SELECT, SCL_C_RAM_TAP_PAIR_IDX, mask_sh), \
+	XFM_SF(SCL_COEF_RAM_TAP_DATA, SCL_C_RAM_EVEN_TAP_COEF_EN, mask_sh), \
+	XFM_SF(SCL_COEF_RAM_TAP_DATA, SCL_C_RAM_EVEN_TAP_COEF, mask_sh), \
+	XFM_SF(SCL_COEF_RAM_TAP_DATA, SCL_C_RAM_ODD_TAP_COEF_EN, mask_sh), \
+	XFM_SF(SCL_COEF_RAM_TAP_DATA, SCL_C_RAM_ODD_TAP_COEF, mask_sh), \
+	XFM_SF(VIEWPORT_START, VIEWPORT_X_START, mask_sh), \
+	XFM_SF(VIEWPORT_START, VIEWPORT_Y_START, mask_sh), \
+	XFM_SF(VIEWPORT_SIZE, VIEWPORT_HEIGHT, mask_sh), \
+	XFM_SF(VIEWPORT_SIZE, VIEWPORT_WIDTH, mask_sh), \
+	XFM_SF(SCL_HORZ_FILTER_SCALE_RATIO, SCL_H_SCALE_RATIO, mask_sh), \
+	XFM_SF(SCL_VERT_FILTER_SCALE_RATIO, SCL_V_SCALE_RATIO, mask_sh), \
+	XFM_SF(SCL_HORZ_FILTER_INIT, SCL_H_INIT_INT, mask_sh), \
+	XFM_SF(SCL_HORZ_FILTER_INIT, SCL_H_INIT_FRAC, mask_sh), \
+	XFM_SF(SCL_VERT_FILTER_INIT, SCL_V_INIT_INT, mask_sh), \
+	XFM_SF(SCL_VERT_FILTER_INIT, SCL_V_INIT_FRAC, mask_sh), \
+	XFM_SF(LB_MEMORY_CTRL, LB_MEMORY_CONFIG, mask_sh), \
+	XFM_SF(LB_MEMORY_CTRL, LB_MEMORY_SIZE, mask_sh), \
+	XFM_SF(SCL_VERT_FILTER_CONTROL, SCL_V_2TAP_HARDCODE_COEF_EN, mask_sh), \
+	XFM_SF(SCL_HORZ_FILTER_CONTROL, SCL_H_2TAP_HARDCODE_COEF_EN, mask_sh), \
+	XFM_SF(SCL_UPDATE, SCL_COEF_UPDATE_COMPLETE, mask_sh), \
+	XFM_SF(LB_DATA_FORMAT, ALPHA_EN, mask_sh)
+
+#define XFM_COMMON_MASK_SH_LIST_DCE80(mask_sh) \
+	XFM_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(mask_sh), \
+	OPP_SF(DCFE_MEM_LIGHT_SLEEP_CNTL, REGAMMA_LUT_LIGHT_SLEEP_DIS, mask_sh),\
+	OPP_SF(DCFE_MEM_LIGHT_SLEEP_CNTL, DCP_LUT_LIGHT_SLEEP_DIS, mask_sh),\
+	OPP_SF(DCFE_MEM_LIGHT_SLEEP_CNTL, REGAMMA_LUT_MEM_PWR_STATE, mask_sh)
+
+#define XFM_COMMON_MASK_SH_LIST_DCE110(mask_sh) \
+	XFM_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(mask_sh), \
+	XFM_SF(DCFE_MEM_PWR_CTRL, SCL_COEFF_MEM_PWR_DIS, mask_sh), \
+	XFM_SF(DCFE_MEM_PWR_STATUS, SCL_COEFF_MEM_PWR_STATE, mask_sh), \
+	XFM_SF(DCFE_MEM_PWR_CTRL, DCP_REGAMMA_MEM_PWR_DIS, mask_sh),\
+	XFM_SF(DCFE_MEM_PWR_CTRL, DCP_LUT_MEM_PWR_DIS, mask_sh),\
+	XFM_SF(DCFE_MEM_PWR_STATUS, DCP_REGAMMA_MEM_PWR_STATE, mask_sh),\
+	XFM_SF(SCL_MODE, SCL_PSCL_EN, mask_sh)
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+#define XFM_COMMON_MASK_SH_LIST_DCE60(mask_sh) \
+	XFM_COMMON_MASK_SH_LIST_DCE60_COMMON_BASE(mask_sh), \
+	OPP_SF(DCFE_MEM_LIGHT_SLEEP_CNTL, REGAMMA_LUT_LIGHT_SLEEP_DIS, mask_sh),\
+	OPP_SF(DCFE_MEM_LIGHT_SLEEP_CNTL, DCP_LUT_LIGHT_SLEEP_DIS, mask_sh),\
+	OPP_SF(DCFE_MEM_LIGHT_SLEEP_CNTL, REGAMMA_LUT_MEM_PWR_STATE, mask_sh)
+
+#define XFM_COMMON_MASK_SH_LIST_DCE60_COMMON_BASE(mask_sh) \
+	XFM_SF(OUT_ROUND_CONTROL, OUT_ROUND_TRUNC_MODE, mask_sh), \
+	XFM_SF(DCP_SPATIAL_DITHER_CNTL, DCP_SPATIAL_DITHER_EN, mask_sh), \
+	XFM_SF(DCP_SPATIAL_DITHER_CNTL, DCP_SPATIAL_DITHER_MODE, mask_sh), \
+	XFM_SF(DCP_SPATIAL_DITHER_CNTL, DCP_SPATIAL_DITHER_DEPTH, mask_sh), \
+	XFM_SF(DCP_SPATIAL_DITHER_CNTL, DCP_FRAME_RANDOM_ENABLE, mask_sh), \
+	XFM_SF(DCP_SPATIAL_DITHER_CNTL, DCP_RGB_RANDOM_ENABLE, mask_sh), \
+	XFM_SF(DCP_SPATIAL_DITHER_CNTL, DCP_HIGHPASS_RANDOM_ENABLE, mask_sh), \
+	XFM_SF(DENORM_CONTROL, DENORM_MODE, mask_sh), \
+	XFM_SF(DATA_FORMAT, INTERLEAVE_EN, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C11_C12, GAMUT_REMAP_C11, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C11_C12, GAMUT_REMAP_C12, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C13_C14, GAMUT_REMAP_C13, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C13_C14, GAMUT_REMAP_C14, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C21_C22, GAMUT_REMAP_C21, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C21_C22, GAMUT_REMAP_C22, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C23_C24, GAMUT_REMAP_C23, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C23_C24, GAMUT_REMAP_C24, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C31_C32, GAMUT_REMAP_C31, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C31_C32, GAMUT_REMAP_C32, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C33_C34, GAMUT_REMAP_C33, mask_sh), \
+	XFM_SF(GAMUT_REMAP_C33_C34, GAMUT_REMAP_C34, mask_sh), \
+	XFM_SF(GAMUT_REMAP_CONTROL, GRPH_GAMUT_REMAP_MODE, mask_sh), \
+	XFM_SF(OUTPUT_CSC_C11_C12, OUTPUT_CSC_C11, mask_sh),\
+	XFM_SF(OUTPUT_CSC_C11_C12, OUTPUT_CSC_C12, mask_sh),\
+	XFM_SF(OUTPUT_CSC_CONTROL, OUTPUT_CSC_GRPH_MODE, mask_sh),\
+	XFM_SF(REGAMMA_CNTLA_START_CNTL, REGAMMA_CNTLA_EXP_REGION_START, mask_sh),\
+	XFM_SF(REGAMMA_CNTLA_START_CNTL, REGAMMA_CNTLA_EXP_REGION_START_SEGMENT, mask_sh),\
+	XFM_SF(REGAMMA_CNTLA_SLOPE_CNTL, REGAMMA_CNTLA_EXP_REGION_LINEAR_SLOPE, mask_sh),\
+	XFM_SF(REGAMMA_CNTLA_END_CNTL1, REGAMMA_CNTLA_EXP_REGION_END, mask_sh),\
+	XFM_SF(REGAMMA_CNTLA_END_CNTL2, REGAMMA_CNTLA_EXP_REGION_END_BASE, mask_sh),\
+	XFM_SF(REGAMMA_CNTLA_END_CNTL2, REGAMMA_CNTLA_EXP_REGION_END_SLOPE, mask_sh),\
+	XFM_SF(REGAMMA_CNTLA_REGION_0_1, REGAMMA_CNTLA_EXP_REGION0_LUT_OFFSET, mask_sh),\
+	XFM_SF(REGAMMA_CNTLA_REGION_0_1, REGAMMA_CNTLA_EXP_REGION0_NUM_SEGMENTS, mask_sh),\
+	XFM_SF(REGAMMA_CNTLA_REGION_0_1, REGAMMA_CNTLA_EXP_REGION1_LUT_OFFSET, mask_sh),\
+	XFM_SF(REGAMMA_CNTLA_REGION_0_1, REGAMMA_CNTLA_EXP_REGION1_NUM_SEGMENTS, mask_sh),\
+	XFM_SF(REGAMMA_LUT_WRITE_EN_MASK, REGAMMA_LUT_WRITE_EN_MASK, mask_sh),\
+	XFM_SF(REGAMMA_CONTROL, GRPH_REGAMMA_MODE, mask_sh),\
+	XFM_SF(SCL_TAP_CONTROL, SCL_H_NUM_OF_TAPS, mask_sh), \
+	XFM_SF(SCL_TAP_CONTROL, SCL_V_NUM_OF_TAPS, mask_sh), \
+	XFM_SF(SCL_BYPASS_CONTROL, SCL_BYPASS_MODE, mask_sh), \
+	XFM_SF(EXT_OVERSCAN_LEFT_RIGHT, EXT_OVERSCAN_LEFT, mask_sh), \
+	XFM_SF(EXT_OVERSCAN_LEFT_RIGHT, EXT_OVERSCAN_RIGHT, mask_sh), \
+	XFM_SF(EXT_OVERSCAN_TOP_BOTTOM, EXT_OVERSCAN_TOP, mask_sh), \
+	XFM_SF(EXT_OVERSCAN_TOP_BOTTOM, EXT_OVERSCAN_BOTTOM, mask_sh), \
+	XFM_SF(SCL_COEF_RAM_SELECT, SCL_C_RAM_FILTER_TYPE, mask_sh), \
+	XFM_SF(SCL_COEF_RAM_SELECT, SCL_C_RAM_PHASE, mask_sh), \
+	XFM_SF(SCL_COEF_RAM_SELECT, SCL_C_RAM_TAP_PAIR_IDX, mask_sh), \
+	XFM_SF(SCL_COEF_RAM_TAP_DATA, SCL_C_RAM_EVEN_TAP_COEF_EN, mask_sh), \
+	XFM_SF(SCL_COEF_RAM_TAP_DATA, SCL_C_RAM_EVEN_TAP_COEF, mask_sh), \
+	XFM_SF(SCL_COEF_RAM_TAP_DATA, SCL_C_RAM_ODD_TAP_COEF_EN, mask_sh), \
+	XFM_SF(SCL_COEF_RAM_TAP_DATA, SCL_C_RAM_ODD_TAP_COEF, mask_sh), \
+	XFM_SF(VIEWPORT_START, VIEWPORT_X_START, mask_sh), \
+	XFM_SF(VIEWPORT_START, VIEWPORT_Y_START, mask_sh), \
+	XFM_SF(VIEWPORT_SIZE, VIEWPORT_HEIGHT, mask_sh), \
+	XFM_SF(VIEWPORT_SIZE, VIEWPORT_WIDTH, mask_sh), \
+	XFM_SF(SCL_HORZ_FILTER_SCALE_RATIO, SCL_H_SCALE_RATIO, mask_sh), \
+	XFM_SF(SCL_VERT_FILTER_SCALE_RATIO, SCL_V_SCALE_RATIO, mask_sh), \
+	XFM_SF(SCL_HORZ_FILTER_INIT_RGB_LUMA, SCL_H_INIT_INT_RGB_Y, mask_sh), \
+	XFM_SF(SCL_HORZ_FILTER_INIT_RGB_LUMA, SCL_H_INIT_FRAC_RGB_Y, mask_sh), \
+	XFM_SF(SCL_HORZ_FILTER_INIT_CHROMA, SCL_H_INIT_INT_CBCR, mask_sh), \
+	XFM_SF(SCL_HORZ_FILTER_INIT_CHROMA, SCL_H_INIT_FRAC_CBCR, mask_sh), \
+	XFM_SF(SCL_VERT_FILTER_INIT, SCL_V_INIT_INT, mask_sh), \
+	XFM_SF(SCL_VERT_FILTER_INIT, SCL_V_INIT_FRAC, mask_sh), \
+	XFM_SF(SCL_HORZ_FILTER_CONTROL, SCL_H_FILTER_PICK_NEAREST, mask_sh), \
+	XFM_SF(SCL_VERT_FILTER_CONTROL, SCL_V_FILTER_PICK_NEAREST, mask_sh), \
+	XFM_SF(DC_LB_MEMORY_SPLIT, DC_LB_MEMORY_CONFIG, mask_sh), \
+	XFM_SF(DC_LB_MEM_SIZE, DC_LB_MEM_SIZE, mask_sh)
+#endif
+
+#define XFM_COMMON_MASK_SH_LIST_SOC_BASE(mask_sh) \
+	XFM_SF(DCP0_OUT_CLAMP_CONTROL_B_CB, OUT_CLAMP_MIN_B_CB, mask_sh), \
+	XFM_SF(DCP0_OUT_CLAMP_CONTROL_B_CB, OUT_CLAMP_MAX_B_CB, mask_sh), \
+	XFM_SF(DCP0_OUT_CLAMP_CONTROL_G_Y, OUT_CLAMP_MIN_G_Y, mask_sh), \
+	XFM_SF(DCP0_OUT_CLAMP_CONTROL_G_Y, OUT_CLAMP_MAX_G_Y, mask_sh), \
+	XFM_SF(DCP0_OUT_CLAMP_CONTROL_R_CR, OUT_CLAMP_MIN_R_CR, mask_sh), \
+	XFM_SF(DCP0_OUT_CLAMP_CONTROL_R_CR, OUT_CLAMP_MAX_R_CR, mask_sh), \
+	XFM_SF(DCP0_OUT_ROUND_CONTROL, OUT_ROUND_TRUNC_MODE, mask_sh), \
+	XFM_SF(DCP0_DCP_SPATIAL_DITHER_CNTL, DCP_SPATIAL_DITHER_EN, mask_sh), \
+	XFM_SF(DCP0_DCP_SPATIAL_DITHER_CNTL, DCP_SPATIAL_DITHER_MODE, mask_sh), \
+	XFM_SF(DCP0_DCP_SPATIAL_DITHER_CNTL, DCP_SPATIAL_DITHER_DEPTH, mask_sh), \
+	XFM_SF(DCP0_DCP_SPATIAL_DITHER_CNTL, DCP_FRAME_RANDOM_ENABLE, mask_sh), \
+	XFM_SF(DCP0_DCP_SPATIAL_DITHER_CNTL, DCP_RGB_RANDOM_ENABLE, mask_sh), \
+	XFM_SF(DCP0_DCP_SPATIAL_DITHER_CNTL, DCP_HIGHPASS_RANDOM_ENABLE, mask_sh), \
+	XFM_SF(DCP0_DENORM_CONTROL, DENORM_MODE, mask_sh), \
+	XFM_SF(LB0_LB_DATA_FORMAT, PIXEL_DEPTH, mask_sh), \
+	XFM_SF(LB0_LB_DATA_FORMAT, PIXEL_EXPAN_MODE, mask_sh), \
+	XFM_SF(DCP0_GAMUT_REMAP_C11_C12, GAMUT_REMAP_C11, mask_sh), \
+	XFM_SF(DCP0_GAMUT_REMAP_C11_C12, GAMUT_REMAP_C12, mask_sh), \
+	XFM_SF(DCP0_GAMUT_REMAP_C13_C14, GAMUT_REMAP_C13, mask_sh), \
+	XFM_SF(DCP0_GAMUT_REMAP_C13_C14, GAMUT_REMAP_C14, mask_sh), \
+	XFM_SF(DCP0_GAMUT_REMAP_C21_C22, GAMUT_REMAP_C21, mask_sh), \
+	XFM_SF(DCP0_GAMUT_REMAP_C21_C22, GAMUT_REMAP_C22, mask_sh), \
+	XFM_SF(DCP0_GAMUT_REMAP_C23_C24, GAMUT_REMAP_C23, mask_sh), \
+	XFM_SF(DCP0_GAMUT_REMAP_C23_C24, GAMUT_REMAP_C24, mask_sh), \
+	XFM_SF(DCP0_GAMUT_REMAP_C31_C32, GAMUT_REMAP_C31, mask_sh), \
+	XFM_SF(DCP0_GAMUT_REMAP_C31_C32, GAMUT_REMAP_C32, mask_sh), \
+	XFM_SF(DCP0_GAMUT_REMAP_C33_C34, GAMUT_REMAP_C33, mask_sh), \
+	XFM_SF(DCP0_GAMUT_REMAP_C33_C34, GAMUT_REMAP_C34, mask_sh), \
+	XFM_SF(DCP0_GAMUT_REMAP_CONTROL, GRPH_GAMUT_REMAP_MODE, mask_sh), \
+	XFM_SF(DCP0_OUTPUT_CSC_C11_C12, OUTPUT_CSC_C11, mask_sh),\
+	XFM_SF(DCP0_OUTPUT_CSC_C11_C12, OUTPUT_CSC_C12, mask_sh),\
+	XFM_SF(DCP0_OUTPUT_CSC_CONTROL, OUTPUT_CSC_GRPH_MODE, mask_sh),\
+	XFM_SF(DCP0_REGAMMA_CNTLA_START_CNTL, REGAMMA_CNTLA_EXP_REGION_START, mask_sh),\
+	XFM_SF(DCP0_REGAMMA_CNTLA_START_CNTL, REGAMMA_CNTLA_EXP_REGION_START_SEGMENT, mask_sh),\
+	XFM_SF(DCP0_REGAMMA_CNTLA_SLOPE_CNTL, REGAMMA_CNTLA_EXP_REGION_LINEAR_SLOPE, mask_sh),\
+	XFM_SF(DCP0_REGAMMA_CNTLA_END_CNTL1, REGAMMA_CNTLA_EXP_REGION_END, mask_sh),\
+	XFM_SF(DCP0_REGAMMA_CNTLA_END_CNTL2, REGAMMA_CNTLA_EXP_REGION_END_BASE, mask_sh),\
+	XFM_SF(DCP0_REGAMMA_CNTLA_END_CNTL2, REGAMMA_CNTLA_EXP_REGION_END_SLOPE, mask_sh),\
+	XFM_SF(DCP0_REGAMMA_CNTLA_REGION_0_1, REGAMMA_CNTLA_EXP_REGION0_LUT_OFFSET, mask_sh),\
+	XFM_SF(DCP0_REGAMMA_CNTLA_REGION_0_1, REGAMMA_CNTLA_EXP_REGION0_NUM_SEGMENTS, mask_sh),\
+	XFM_SF(DCP0_REGAMMA_CNTLA_REGION_0_1, REGAMMA_CNTLA_EXP_REGION1_LUT_OFFSET, mask_sh),\
+	XFM_SF(DCP0_REGAMMA_CNTLA_REGION_0_1, REGAMMA_CNTLA_EXP_REGION1_NUM_SEGMENTS, mask_sh),\
+	XFM_SF(DCP0_REGAMMA_CONTROL, GRPH_REGAMMA_MODE, mask_sh),\
+	XFM_SF(DCP0_REGAMMA_LUT_WRITE_EN_MASK, REGAMMA_LUT_WRITE_EN_MASK, mask_sh),\
+	XFM_SF(SCL0_SCL_MODE, SCL_MODE, mask_sh), \
+	XFM_SF(SCL0_SCL_TAP_CONTROL, SCL_H_NUM_OF_TAPS, mask_sh), \
+	XFM_SF(SCL0_SCL_TAP_CONTROL, SCL_V_NUM_OF_TAPS, mask_sh), \
+	XFM_SF(SCL0_SCL_CONTROL, SCL_BOUNDARY_MODE, mask_sh), \
+	XFM_SF(SCL0_SCL_BYPASS_CONTROL, SCL_BYPASS_MODE, mask_sh), \
+	XFM_SF(SCL0_EXT_OVERSCAN_LEFT_RIGHT, EXT_OVERSCAN_LEFT, mask_sh), \
+	XFM_SF(SCL0_EXT_OVERSCAN_LEFT_RIGHT, EXT_OVERSCAN_RIGHT, mask_sh), \
+	XFM_SF(SCL0_EXT_OVERSCAN_TOP_BOTTOM, EXT_OVERSCAN_TOP, mask_sh), \
+	XFM_SF(SCL0_EXT_OVERSCAN_TOP_BOTTOM, EXT_OVERSCAN_BOTTOM, mask_sh), \
+	XFM_SF(SCL0_SCL_COEF_RAM_SELECT, SCL_C_RAM_FILTER_TYPE, mask_sh), \
+	XFM_SF(SCL0_SCL_COEF_RAM_SELECT, SCL_C_RAM_PHASE, mask_sh), \
+	XFM_SF(SCL0_SCL_COEF_RAM_SELECT, SCL_C_RAM_TAP_PAIR_IDX, mask_sh), \
+	XFM_SF(SCL0_SCL_COEF_RAM_TAP_DATA, SCL_C_RAM_EVEN_TAP_COEF_EN, mask_sh), \
+	XFM_SF(SCL0_SCL_COEF_RAM_TAP_DATA, SCL_C_RAM_EVEN_TAP_COEF, mask_sh), \
+	XFM_SF(SCL0_SCL_COEF_RAM_TAP_DATA, SCL_C_RAM_ODD_TAP_COEF_EN, mask_sh), \
+	XFM_SF(SCL0_SCL_COEF_RAM_TAP_DATA, SCL_C_RAM_ODD_TAP_COEF, mask_sh), \
+	XFM_SF(SCL0_VIEWPORT_START, VIEWPORT_X_START, mask_sh), \
+	XFM_SF(SCL0_VIEWPORT_START, VIEWPORT_Y_START, mask_sh), \
+	XFM_SF(SCL0_VIEWPORT_SIZE, VIEWPORT_HEIGHT, mask_sh), \
+	XFM_SF(SCL0_VIEWPORT_SIZE, VIEWPORT_WIDTH, mask_sh), \
+	XFM_SF(SCL0_SCL_HORZ_FILTER_SCALE_RATIO, SCL_H_SCALE_RATIO, mask_sh), \
+	XFM_SF(SCL0_SCL_VERT_FILTER_SCALE_RATIO, SCL_V_SCALE_RATIO, mask_sh), \
+	XFM_SF(SCL0_SCL_HORZ_FILTER_INIT, SCL_H_INIT_INT, mask_sh), \
+	XFM_SF(SCL0_SCL_HORZ_FILTER_INIT, SCL_H_INIT_FRAC, mask_sh), \
+	XFM_SF(SCL0_SCL_VERT_FILTER_INIT, SCL_V_INIT_INT, mask_sh), \
+	XFM_SF(SCL0_SCL_VERT_FILTER_INIT, SCL_V_INIT_FRAC, mask_sh), \
+	XFM_SF(LB0_LB_MEMORY_CTRL, LB_MEMORY_CONFIG, mask_sh), \
+	XFM_SF(LB0_LB_MEMORY_CTRL, LB_MEMORY_SIZE, mask_sh), \
+	XFM_SF(SCL0_SCL_VERT_FILTER_CONTROL, SCL_V_2TAP_HARDCODE_COEF_EN, mask_sh), \
+	XFM_SF(SCL0_SCL_HORZ_FILTER_CONTROL, SCL_H_2TAP_HARDCODE_COEF_EN, mask_sh), \
+	XFM_SF(SCL0_SCL_UPDATE, SCL_COEF_UPDATE_COMPLETE, mask_sh), \
+	XFM_SF(LB0_LB_DATA_FORMAT, ALPHA_EN, mask_sh), \
+	XFM_SF(DCFE0_DCFE_MEM_PWR_CTRL, SCL_COEFF_MEM_PWR_DIS, mask_sh), \
+	XFM_SF(DCFE0_DCFE_MEM_PWR_CTRL, DCP_REGAMMA_MEM_PWR_DIS, mask_sh),\
+	XFM_SF(DCFE0_DCFE_MEM_PWR_CTRL, DCP_LUT_MEM_PWR_DIS, mask_sh),\
+	XFM_SF(DCFE0_DCFE_MEM_PWR_STATUS, SCL_COEFF_MEM_PWR_STATE, mask_sh), \
+	XFM_SF(SCL0_SCL_MODE, SCL_PSCL_EN, mask_sh)
+
+#define XFM_REG_FIELD_LIST(type) \
+	type OUT_CLAMP_MIN_B_CB; \
+	type OUT_CLAMP_MAX_B_CB; \
+	type OUT_CLAMP_MIN_G_Y; \
+	type OUT_CLAMP_MAX_G_Y; \
+	type OUT_CLAMP_MIN_R_CR; \
+	type OUT_CLAMP_MAX_R_CR; \
+	type OUT_ROUND_TRUNC_MODE; \
+	type DCP_SPATIAL_DITHER_EN; \
+	type DCP_SPATIAL_DITHER_MODE; \
+	type DCP_SPATIAL_DITHER_DEPTH; \
+	type DCP_FRAME_RANDOM_ENABLE; \
+	type DCP_RGB_RANDOM_ENABLE; \
+	type DCP_HIGHPASS_RANDOM_ENABLE; \
+	type DENORM_MODE; \
+	type INTERLEAVE_EN; \
+	type PIXEL_DEPTH; \
+	type PIXEL_EXPAN_MODE; \
+	type GAMUT_REMAP_C11; \
+	type GAMUT_REMAP_C12; \
+	type GAMUT_REMAP_C13; \
+	type GAMUT_REMAP_C14; \
+	type GAMUT_REMAP_C21; \
+	type GAMUT_REMAP_C22; \
+	type GAMUT_REMAP_C23; \
+	type GAMUT_REMAP_C24; \
+	type GAMUT_REMAP_C31; \
+	type GAMUT_REMAP_C32; \
+	type GAMUT_REMAP_C33; \
+	type GAMUT_REMAP_C34; \
+	type GRPH_GAMUT_REMAP_MODE; \
+	type OUTPUT_CSC_C11; \
+	type OUTPUT_CSC_C12; \
+	type OUTPUT_CSC_GRPH_MODE; \
+	type DCP_REGAMMA_MEM_PWR_DIS; \
+	type DCP_LUT_MEM_PWR_DIS; \
+	type REGAMMA_LUT_LIGHT_SLEEP_DIS; \
+	type DCP_LUT_LIGHT_SLEEP_DIS; \
+	type REGAMMA_CNTLA_EXP_REGION_START; \
+	type REGAMMA_CNTLA_EXP_REGION_START_SEGMENT; \
+	type REGAMMA_CNTLA_EXP_REGION_LINEAR_SLOPE; \
+	type REGAMMA_CNTLA_EXP_REGION_END; \
+	type REGAMMA_CNTLA_EXP_REGION_END_BASE; \
+	type REGAMMA_CNTLA_EXP_REGION_END_SLOPE; \
+	type REGAMMA_CNTLA_EXP_REGION0_LUT_OFFSET; \
+	type REGAMMA_CNTLA_EXP_REGION0_NUM_SEGMENTS; \
+	type REGAMMA_CNTLA_EXP_REGION1_LUT_OFFSET; \
+	type REGAMMA_CNTLA_EXP_REGION1_NUM_SEGMENTS; \
+	type DCP_REGAMMA_MEM_PWR_STATE; \
+	type REGAMMA_LUT_MEM_PWR_STATE; \
+	type REGAMMA_LUT_WRITE_EN_MASK; \
+	type GRPH_REGAMMA_MODE; \
+	type SCL_MODE; \
+	type SCL_BYPASS_MODE; \
+	type SCL_PSCL_EN; \
+	type SCL_H_NUM_OF_TAPS; \
+	type SCL_V_NUM_OF_TAPS; \
+	type SCL_BOUNDARY_MODE; \
+	type EXT_OVERSCAN_LEFT; \
+	type EXT_OVERSCAN_RIGHT; \
+	type EXT_OVERSCAN_TOP; \
+	type EXT_OVERSCAN_BOTTOM; \
+	type SCL_COEFF_MEM_PWR_DIS; \
+	type SCL_COEFF_MEM_PWR_STATE; \
+	type SCL_C_RAM_FILTER_TYPE; \
+	type SCL_C_RAM_PHASE; \
+	type SCL_C_RAM_TAP_PAIR_IDX; \
+	type SCL_C_RAM_EVEN_TAP_COEF_EN; \
+	type SCL_C_RAM_EVEN_TAP_COEF; \
+	type SCL_C_RAM_ODD_TAP_COEF_EN; \
+	type SCL_C_RAM_ODD_TAP_COEF; \
+	type VIEWPORT_X_START; \
+	type VIEWPORT_Y_START; \
+	type VIEWPORT_HEIGHT; \
+	type VIEWPORT_WIDTH; \
+	type SCL_H_SCALE_RATIO; \
+	type SCL_V_SCALE_RATIO; \
+	type SCL_H_INIT_INT; \
+	type SCL_H_INIT_FRAC; \
+	type SCL_H_INIT_INT_RGB_Y; \
+	type SCL_H_INIT_FRAC_RGB_Y; \
+	type SCL_H_INIT_INT_CBCR; \
+	type SCL_H_INIT_FRAC_CBCR; \
+	type SCL_V_INIT_INT; \
+	type SCL_V_INIT_FRAC; \
+	type DC_LB_MEMORY_CONFIG; \
+	type DC_LB_MEM_SIZE; \
+	type LB_MEMORY_CONFIG; \
+	type LB_MEMORY_SIZE; \
+	type SCL_V_2TAP_HARDCODE_COEF_EN; \
+	type SCL_H_2TAP_HARDCODE_COEF_EN; \
+	type SCL_V_FILTER_PICK_NEAREST; \
+	type SCL_H_FILTER_PICK_NEAREST; \
+	type SCL_COEF_UPDATE_COMPLETE; \
+	type ALPHA_EN
+
+struct dce_transform_shift {
+	XFM_REG_FIELD_LIST(uint8_t);
+};
+
+struct dce_transform_mask {
+	XFM_REG_FIELD_LIST(uint32_t);
+};
+
+struct dce_transform_registers {
+#if defined(CONFIG_DRM_AMD_DC_SI)
+	uint32_t DATA_FORMAT;
+#endif
+	uint32_t LB_DATA_FORMAT;
+	uint32_t GAMUT_REMAP_CONTROL;
+	uint32_t GAMUT_REMAP_C11_C12;
+	uint32_t GAMUT_REMAP_C13_C14;
+	uint32_t GAMUT_REMAP_C21_C22;
+	uint32_t GAMUT_REMAP_C23_C24;
+	uint32_t GAMUT_REMAP_C31_C32;
+	uint32_t GAMUT_REMAP_C33_C34;
+	uint32_t OUTPUT_CSC_C11_C12;
+	uint32_t OUTPUT_CSC_C13_C14;
+	uint32_t OUTPUT_CSC_C21_C22;
+	uint32_t OUTPUT_CSC_C23_C24;
+	uint32_t OUTPUT_CSC_C31_C32;
+	uint32_t OUTPUT_CSC_C33_C34;
+	uint32_t OUTPUT_CSC_CONTROL;
+	uint32_t DCFE_MEM_LIGHT_SLEEP_CNTL;
+	uint32_t REGAMMA_CNTLA_START_CNTL;
+	uint32_t REGAMMA_CNTLA_SLOPE_CNTL;
+	uint32_t REGAMMA_CNTLA_END_CNTL1;
+	uint32_t REGAMMA_CNTLA_END_CNTL2;
+	uint32_t REGAMMA_CNTLA_REGION_0_1;
+	uint32_t REGAMMA_CNTLA_REGION_2_3;
+	uint32_t REGAMMA_CNTLA_REGION_4_5;
+	uint32_t REGAMMA_CNTLA_REGION_6_7;
+	uint32_t REGAMMA_CNTLA_REGION_8_9;
+	uint32_t REGAMMA_CNTLA_REGION_10_11;
+	uint32_t REGAMMA_CNTLA_REGION_12_13;
+	uint32_t REGAMMA_CNTLA_REGION_14_15;
+	uint32_t REGAMMA_LUT_WRITE_EN_MASK;
+	uint32_t REGAMMA_LUT_INDEX;
+	uint32_t REGAMMA_LUT_DATA;
+	uint32_t REGAMMA_CONTROL;
+	uint32_t DENORM_CONTROL;
+	uint32_t DCP_SPATIAL_DITHER_CNTL;
+	uint32_t OUT_ROUND_CONTROL;
+	uint32_t OUT_CLAMP_CONTROL_R_CR;
+	uint32_t OUT_CLAMP_CONTROL_G_Y;
+	uint32_t OUT_CLAMP_CONTROL_B_CB;
+	uint32_t SCL_MODE;
+	uint32_t SCL_TAP_CONTROL;
+	uint32_t SCL_CONTROL;
+	uint32_t SCL_BYPASS_CONTROL;
+	uint32_t EXT_OVERSCAN_LEFT_RIGHT;
+	uint32_t EXT_OVERSCAN_TOP_BOTTOM;
+	uint32_t SCL_VERT_FILTER_CONTROL;
+	uint32_t SCL_HORZ_FILTER_CONTROL;
+	uint32_t DCFE_MEM_PWR_CTRL;
+	uint32_t DCFE_MEM_PWR_STATUS;
+	uint32_t SCL_COEF_RAM_SELECT;
+	uint32_t SCL_COEF_RAM_TAP_DATA;
+	uint32_t VIEWPORT_START;
+	uint32_t VIEWPORT_SIZE;
+	uint32_t SCL_HORZ_FILTER_SCALE_RATIO;
+	uint32_t SCL_VERT_FILTER_SCALE_RATIO;
+	uint32_t SCL_HORZ_FILTER_INIT;
+#if defined(CONFIG_DRM_AMD_DC_SI)
+	uint32_t SCL_SCALER_ENABLE;
+	uint32_t SCL_HORZ_FILTER_INIT_RGB_LUMA;
+	uint32_t SCL_HORZ_FILTER_INIT_CHROMA;
+#endif
+	uint32_t SCL_VERT_FILTER_INIT;
+	uint32_t SCL_AUTOMATIC_MODE_CONTROL;
+#if defined(CONFIG_DRM_AMD_DC_SI)
+	uint32_t DC_LB_MEMORY_SPLIT;
+	uint32_t DC_LB_MEM_SIZE;
+#endif
+	uint32_t LB_MEMORY_CTRL;
+	uint32_t SCL_UPDATE;
+	uint32_t SCL_F_SHARP_CONTROL;
+};
+
+struct init_int_and_frac {
+	uint32_t integer;
+	uint32_t fraction;
+};
+
+struct scl_ratios_inits {
+	uint32_t h_int_scale_ratio;
+	uint32_t v_int_scale_ratio;
+	struct init_int_and_frac h_init;
+	struct init_int_and_frac v_init;
+};
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+struct sclh_ratios_inits {
+	uint32_t h_int_scale_ratio;
+	uint32_t v_int_scale_ratio;
+	struct init_int_and_frac h_init_luma;
+	struct init_int_and_frac h_init_chroma;
+	struct init_int_and_frac v_init;
+};
+#endif
+
+enum ram_filter_type {
+	FILTER_TYPE_RGB_Y_VERTICAL	= 0, /* 0 - RGB/Y Vertical filter */
+	FILTER_TYPE_CBCR_VERTICAL	= 1, /* 1 - CbCr  Vertical filter */
+	FILTER_TYPE_RGB_Y_HORIZONTAL	= 2, /* 1 - RGB/Y Horizontal filter */
+	FILTER_TYPE_CBCR_HORIZONTAL	= 3, /* 3 - CbCr  Horizontal filter */
+	FILTER_TYPE_ALPHA_VERTICAL	= 4, /* 4 - Alpha Vertical filter. */
+	FILTER_TYPE_ALPHA_HORIZONTAL	= 5, /* 5 - Alpha Horizontal filter. */
+};
+
+struct dce_transform {
+	struct transform base;
+	const struct dce_transform_registers *regs;
+	const struct dce_transform_shift *xfm_shift;
+	const struct dce_transform_mask *xfm_mask;
+
+	const uint16_t *filter_v;
+	const uint16_t *filter_h;
+	const uint16_t *filter_v_c;
+	const uint16_t *filter_h_c;
+	int lb_pixel_depth_supported;
+	int lb_memory_size;
+	int lb_bits_per_entry;
+	bool prescaler_on;
+};
+
+void dce_transform_construct(struct dce_transform *xfm_dce,
+	struct dc_context *ctx,
+	uint32_t inst,
+	const struct dce_transform_registers *regs,
+	const struct dce_transform_shift *xfm_shift,
+	const struct dce_transform_mask *xfm_mask);
+
+#if defined(CONFIG_DRM_AMD_DC_SI)
+void dce60_transform_construct(struct dce_transform *xfm_dce,
+	struct dc_context *ctx,
+	uint32_t inst,
+	const struct dce_transform_registers *regs,
+	const struct dce_transform_shift *xfm_shift,
+	const struct dce_transform_mask *xfm_mask);
+#endif
+
+bool dce_transform_get_optimal_number_of_taps(
+	struct transform *xfm,
+	struct scaler_data *scl_data,
+	const struct scaling_taps *in_taps);
+
+void dce110_opp_set_csc_adjustment(
+	struct transform *xfm,
+	const struct out_csc_color_matrix *tbl_entry);
+
+void dce110_opp_set_csc_default(
+	struct transform *xfm,
+	const struct default_adjustment *default_adjust);
+
+/* REGAMMA RELATED */
+void dce110_opp_power_on_regamma_lut(
+	struct transform *xfm,
+	bool power_on);
+
+void dce110_opp_program_regamma_pwl(
+	struct transform *xfm,
+	const struct pwl_params *params);
+
+void dce110_opp_set_regamma_mode(struct transform *xfm,
+		enum opp_regamma mode);
+
+#endif /* _DCE_DCE_TRANSFORM_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dmub_abm.c b/drivers/gpu/drm/amd/display/dc/dce/dmub_abm.c
new file mode 100644
index 000000000000..3b9011ef9b68
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dmub_abm.c
@@ -0,0 +1,245 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dmub_abm.h"
+#include "dmub_abm_lcd.h"
+#include "dc.h"
+#include "core_types.h"
+#include "dmub_cmd.h"
+#include "dc_dmub_srv.h"
+#include "dmub/dmub_srv.h"
+
+#define TO_DMUB_ABM(abm)\
+	container_of(abm, struct dce_abm, base)
+
+#define ABM_FEATURE_NO_SUPPORT	0
+#define ABM_LCD_SUPPORT			1
+
+static unsigned int abm_feature_support(struct abm *abm, unsigned int panel_inst)
+{
+	struct dc_context *dc = abm->ctx;
+	struct dc_link *edp_links[MAX_NUM_EDP];
+	int i;
+	int edp_num;
+	unsigned int ret = ABM_FEATURE_NO_SUPPORT;
+
+	dc_get_edp_links(dc->dc, edp_links, &edp_num);
+
+	for (i = 0; i < edp_num; i++) {
+		if (panel_inst == i)
+			break;
+	}
+
+	if (i < edp_num) {
+		ret = ABM_LCD_SUPPORT;
+	}
+
+	return ret;
+}
+
+static void dmub_abm_init_ex(struct abm *abm, uint32_t backlight, uint32_t user_level)
+{
+	dmub_abm_init(abm, backlight, user_level);
+}
+
+static unsigned int dmub_abm_get_current_backlight_ex(struct abm *abm)
+{
+	dc_allow_idle_optimizations(abm->ctx->dc, false);
+
+	return dmub_abm_get_current_backlight(abm);
+}
+
+static unsigned int dmub_abm_get_target_backlight_ex(struct abm *abm)
+{
+	dc_allow_idle_optimizations(abm->ctx->dc, false);
+
+	return dmub_abm_get_target_backlight(abm);
+}
+
+static bool dmub_abm_set_level_ex(struct abm *abm, uint32_t level)
+{
+	bool ret = false;
+	unsigned int feature_support, i;
+	uint8_t panel_mask0 = 0;
+
+	for (i = 0; i < MAX_NUM_EDP; i++) {
+		feature_support = abm_feature_support(abm, i);
+
+		if (feature_support == ABM_LCD_SUPPORT)
+			panel_mask0 |= (0x01 << i);
+	}
+
+	if (panel_mask0)
+		ret = dmub_abm_set_level(abm, level, panel_mask0);
+
+	return ret;
+}
+
+static bool dmub_abm_init_config_ex(struct abm *abm,
+	const char *src,
+	unsigned int bytes,
+	unsigned int inst)
+{
+	unsigned int feature_support;
+
+	feature_support = abm_feature_support(abm, inst);
+
+	if (feature_support == ABM_LCD_SUPPORT)
+		dmub_abm_init_config(abm, src, bytes, inst);
+
+	return true;
+}
+
+static bool dmub_abm_set_pause_ex(struct abm *abm, bool pause, unsigned int panel_inst, unsigned int stream_inst)
+{
+	bool ret = false;
+	unsigned int feature_support;
+
+	feature_support = abm_feature_support(abm, panel_inst);
+
+	if (feature_support == ABM_LCD_SUPPORT)
+		ret = dmub_abm_set_pause(abm, pause, panel_inst, stream_inst);
+
+	return ret;
+}
+
+/*****************************************************************************
+ *  dmub_abm_save_restore_ex() - calls dmub_abm_save_restore for preserving DMUB's
+ *                              Varibright states for LCD only. OLED is TBD
+ *  @abm: used to check get dc context
+ *  @panel_inst: panel instance index
+ *  @pData: contains command to pause/un-pause abm and abm parameters
+ *
+ *
+ ***************************************************************************/
+static bool dmub_abm_save_restore_ex(
+		struct abm *abm,
+		unsigned int panel_inst,
+		struct abm_save_restore *pData)
+{
+	bool ret = false;
+	unsigned int feature_support;
+	struct dc_context *dc = abm->ctx;
+
+	feature_support = abm_feature_support(abm, panel_inst);
+
+	if (feature_support == ABM_LCD_SUPPORT)
+		ret = dmub_abm_save_restore(dc, panel_inst, pData);
+
+	return ret;
+}
+
+static bool dmub_abm_set_pipe_ex(struct abm *abm,
+		uint32_t otg_inst,
+		uint32_t option,
+		uint32_t panel_inst,
+		uint32_t pwrseq_inst)
+{
+	bool ret = false;
+	unsigned int feature_support;
+
+	feature_support = abm_feature_support(abm, panel_inst);
+
+	if (feature_support == ABM_LCD_SUPPORT)
+		ret = dmub_abm_set_pipe(abm, otg_inst, option, panel_inst, pwrseq_inst);
+
+	return ret;
+}
+
+static bool dmub_abm_set_backlight_level_pwm_ex(struct abm *abm,
+		unsigned int backlight_pwm_u16_16,
+		unsigned int frame_ramp,
+		unsigned int controller_id,
+		unsigned int panel_inst)
+{
+	bool ret = false;
+	unsigned int feature_support;
+
+	feature_support = abm_feature_support(abm, panel_inst);
+
+	if (feature_support == ABM_LCD_SUPPORT)
+		ret = dmub_abm_set_backlight_level(abm, backlight_pwm_u16_16, frame_ramp, panel_inst);
+
+	return ret;
+}
+
+static const struct abm_funcs abm_funcs = {
+	.abm_init = dmub_abm_init_ex,
+	.set_abm_level = dmub_abm_set_level_ex,
+	.get_current_backlight = dmub_abm_get_current_backlight_ex,
+	.get_target_backlight = dmub_abm_get_target_backlight_ex,
+	.init_abm_config = dmub_abm_init_config_ex,
+	.set_abm_pause = dmub_abm_set_pause_ex,
+	.save_restore = dmub_abm_save_restore_ex,
+	.set_pipe_ex = dmub_abm_set_pipe_ex,
+	.set_backlight_level_pwm = dmub_abm_set_backlight_level_pwm_ex,
+};
+
+static void dmub_abm_construct(
+	struct dce_abm *abm_dce,
+	struct dc_context *ctx,
+	const struct dce_abm_registers *regs,
+	const struct dce_abm_shift *abm_shift,
+	const struct dce_abm_mask *abm_mask)
+{
+	struct abm *base = &abm_dce->base;
+
+	base->ctx = ctx;
+	base->funcs = &abm_funcs;
+	base->dmcu_is_running = false;
+
+	abm_dce->regs = regs;
+	abm_dce->abm_shift = abm_shift;
+	abm_dce->abm_mask = abm_mask;
+}
+
+struct abm *dmub_abm_create(
+	struct dc_context *ctx,
+	const struct dce_abm_registers *regs,
+	const struct dce_abm_shift *abm_shift,
+	const struct dce_abm_mask *abm_mask)
+{
+	if (ctx->dc->caps.dmcub_support) {
+		struct dce_abm *abm_dce = kzalloc(sizeof(*abm_dce), GFP_KERNEL);
+
+		if (abm_dce == NULL) {
+			BREAK_TO_DEBUGGER();
+			return NULL;
+		}
+
+		dmub_abm_construct(abm_dce, ctx, regs, abm_shift, abm_mask);
+
+		return &abm_dce->base;
+	}
+	return NULL;
+}
+
+void dmub_abm_destroy(struct abm **abm)
+{
+	struct dce_abm *abm_dce = TO_DMUB_ABM(*abm);
+
+	kfree(abm_dce);
+	*abm = NULL;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dmub_abm.h b/drivers/gpu/drm/amd/display/dc/dce/dmub_abm.h
new file mode 100644
index 000000000000..3a5d5ac7a86e
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dmub_abm.h
@@ -0,0 +1,40 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DMUB_ABM_H__
+#define __DMUB_ABM_H__
+
+#include "abm.h"
+#include "dce_abm.h"
+
+struct abm *dmub_abm_create(
+	struct dc_context *ctx,
+	const struct dce_abm_registers *regs,
+	const struct dce_abm_shift *abm_shift,
+	const struct dce_abm_mask *abm_mask);
+
+void dmub_abm_destroy(struct abm **abm);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dmub_abm_lcd.c b/drivers/gpu/drm/amd/display/dc/dce/dmub_abm_lcd.c
new file mode 100644
index 000000000000..a641ae04450c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dmub_abm_lcd.c
@@ -0,0 +1,318 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dmub_abm.h"
+#include "dmub_abm_lcd.h"
+#include "dce_abm.h"
+#include "dc.h"
+#include "dc_dmub_srv.h"
+#include "dmub/dmub_srv.h"
+#include "core_types.h"
+#include "dm_services.h"
+#include "reg_helper.h"
+#include "fixed31_32.h"
+
+#include "atom.h"
+
+#define TO_DMUB_ABM(abm)\
+	container_of(abm, struct dce_abm, base)
+
+#define REG(reg) \
+	(dce_abm->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	dce_abm->abm_shift->field_name, dce_abm->abm_mask->field_name
+
+#define CTX \
+	dce_abm->base.ctx
+
+#define DISABLE_ABM_IMMEDIATELY 255
+
+
+
+static void dmub_abm_enable_fractional_pwm(struct dc_context *dc)
+{
+	union dmub_rb_cmd cmd;
+	uint32_t fractional_pwm = (dc->dc->config.disable_fractional_pwm == false) ? 1 : 0;
+	uint32_t edp_id_count = dc->dc_edp_id_count;
+	int i;
+	uint8_t panel_mask = 0;
+
+	for (i = 0; i < edp_id_count; i++)
+		panel_mask |= 0x01 << i;
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.abm_set_pwm_frac.header.type = DMUB_CMD__ABM;
+	cmd.abm_set_pwm_frac.header.sub_type = DMUB_CMD__ABM_SET_PWM_FRAC;
+	cmd.abm_set_pwm_frac.abm_set_pwm_frac_data.fractional_pwm = fractional_pwm;
+	cmd.abm_set_pwm_frac.abm_set_pwm_frac_data.version = DMUB_CMD_ABM_CONTROL_VERSION_1;
+	cmd.abm_set_pwm_frac.abm_set_pwm_frac_data.panel_mask = panel_mask;
+	cmd.abm_set_pwm_frac.header.payload_bytes = sizeof(struct dmub_cmd_abm_set_pwm_frac_data);
+
+	dc_wake_and_execute_dmub_cmd(dc, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+void dmub_abm_init(struct abm *abm, uint32_t backlight, uint32_t user_level)
+{
+	struct dce_abm *dce_abm = TO_DMUB_ABM(abm);
+
+	REG_WRITE(DC_ABM1_HG_SAMPLE_RATE, 0x3);
+	REG_WRITE(DC_ABM1_HG_SAMPLE_RATE, 0x1);
+	REG_WRITE(DC_ABM1_LS_SAMPLE_RATE, 0x3);
+	REG_WRITE(DC_ABM1_LS_SAMPLE_RATE, 0x1);
+	REG_WRITE(BL1_PWM_BL_UPDATE_SAMPLE_RATE, 0x1);
+
+	REG_SET_3(DC_ABM1_HG_MISC_CTRL, 0,
+			ABM1_HG_NUM_OF_BINS_SEL, 0,
+			ABM1_HG_VMAX_SEL, 1,
+			ABM1_HG_BIN_BITWIDTH_SIZE_SEL, 0);
+
+	REG_SET_3(DC_ABM1_IPCSC_COEFF_SEL, 0,
+			ABM1_IPCSC_COEFF_SEL_R, 2,
+			ABM1_IPCSC_COEFF_SEL_G, 4,
+			ABM1_IPCSC_COEFF_SEL_B, 2);
+
+	REG_UPDATE(BL1_PWM_CURRENT_ABM_LEVEL,
+			BL1_PWM_CURRENT_ABM_LEVEL, backlight);
+
+	REG_UPDATE(BL1_PWM_TARGET_ABM_LEVEL,
+			BL1_PWM_TARGET_ABM_LEVEL, backlight);
+
+	REG_UPDATE(BL1_PWM_USER_LEVEL,
+			BL1_PWM_USER_LEVEL, user_level);
+
+	REG_UPDATE_2(DC_ABM1_LS_MIN_MAX_PIXEL_VALUE_THRES,
+			ABM1_LS_MIN_PIXEL_VALUE_THRES, 0,
+			ABM1_LS_MAX_PIXEL_VALUE_THRES, 1000);
+
+	REG_SET_3(DC_ABM1_HGLS_REG_READ_PROGRESS, 0,
+			ABM1_HG_REG_READ_MISSED_FRAME_CLEAR, 1,
+			ABM1_LS_REG_READ_MISSED_FRAME_CLEAR, 1,
+			ABM1_BL_REG_READ_MISSED_FRAME_CLEAR, 1);
+
+	dmub_abm_enable_fractional_pwm(abm->ctx);
+}
+
+unsigned int dmub_abm_get_current_backlight(struct abm *abm)
+{
+	struct dce_abm *dce_abm = TO_DMUB_ABM(abm);
+	unsigned int backlight = REG_READ(BL1_PWM_CURRENT_ABM_LEVEL);
+
+	/* return backlight in hardware format which is unsigned 17 bits, with
+	 * 1 bit integer and 16 bit fractional
+	 */
+	return backlight;
+}
+
+unsigned int dmub_abm_get_target_backlight(struct abm *abm)
+{
+	struct dce_abm *dce_abm = TO_DMUB_ABM(abm);
+	unsigned int backlight = REG_READ(BL1_PWM_TARGET_ABM_LEVEL);
+
+	/* return backlight in hardware format which is unsigned 17 bits, with
+	 * 1 bit integer and 16 bit fractional
+	 */
+	return backlight;
+}
+
+bool dmub_abm_set_level(struct abm *abm, uint32_t level, uint8_t panel_mask)
+{
+	union dmub_rb_cmd cmd;
+	struct dc_context *dc = abm->ctx;
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.abm_set_level.header.type = DMUB_CMD__ABM;
+	cmd.abm_set_level.header.sub_type = DMUB_CMD__ABM_SET_LEVEL;
+	cmd.abm_set_level.abm_set_level_data.level = level;
+	cmd.abm_set_level.abm_set_level_data.version = DMUB_CMD_ABM_CONTROL_VERSION_1;
+	cmd.abm_set_level.abm_set_level_data.panel_mask = panel_mask;
+	cmd.abm_set_level.header.payload_bytes = sizeof(struct dmub_cmd_abm_set_level_data);
+
+	dc_wake_and_execute_dmub_cmd(dc, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+
+	return true;
+}
+
+void dmub_abm_init_config(struct abm *abm,
+	const char *src,
+	unsigned int bytes,
+	unsigned int inst)
+{
+	union dmub_rb_cmd cmd;
+	struct dc_context *dc = abm->ctx;
+	uint8_t panel_mask = 0x01 << inst;
+
+	// TODO: Optimize by only reading back final 4 bytes
+	dmub_flush_buffer_mem(&dc->dmub_srv->dmub->scratch_mem_fb);
+
+	// Copy iramtable into cw7
+	memcpy(dc->dmub_srv->dmub->scratch_mem_fb.cpu_addr, (void *)src, bytes);
+
+	memset(&cmd, 0, sizeof(cmd));
+	// Fw will copy from cw7 to fw_state
+	cmd.abm_init_config.header.type = DMUB_CMD__ABM;
+	cmd.abm_init_config.header.sub_type = DMUB_CMD__ABM_INIT_CONFIG;
+	cmd.abm_init_config.abm_init_config_data.src.quad_part = dc->dmub_srv->dmub->scratch_mem_fb.gpu_addr;
+	cmd.abm_init_config.abm_init_config_data.bytes = bytes;
+	cmd.abm_init_config.abm_init_config_data.version = DMUB_CMD_ABM_CONTROL_VERSION_1;
+	cmd.abm_init_config.abm_init_config_data.panel_mask = panel_mask;
+
+	cmd.abm_init_config.header.payload_bytes = sizeof(struct dmub_cmd_abm_init_config_data);
+
+	dc_wake_and_execute_dmub_cmd(dc, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+
+}
+
+bool dmub_abm_set_pause(struct abm *abm, bool pause, unsigned int panel_inst, unsigned int stream_inst)
+{
+	union dmub_rb_cmd cmd;
+	struct dc_context *dc = abm->ctx;
+	uint8_t panel_mask = 0x01 << panel_inst;
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.abm_pause.header.type = DMUB_CMD__ABM;
+	cmd.abm_pause.header.sub_type = DMUB_CMD__ABM_PAUSE;
+	cmd.abm_pause.abm_pause_data.enable = pause;
+	cmd.abm_pause.abm_pause_data.panel_mask = panel_mask;
+	cmd.abm_set_level.header.payload_bytes = sizeof(struct dmub_cmd_abm_pause_data);
+
+	dc_wake_and_execute_dmub_cmd(dc, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+
+	return true;
+}
+
+
+/*****************************************************************************
+ *  dmub_abm_save_restore() - dmub interface for abm save+pause and restore+
+ *                           un-pause
+ *  @dc: dc context
+ *  @panel_inst: panel instance index
+ *  @pData: contains command to pause/un-pause abm and exchange abm parameters
+ *
+ *  When called Pause will get abm data and store in pData, and un-pause will
+ *  set/apply abm data stored in pData.
+ *
+ *****************************************************************************/
+bool dmub_abm_save_restore(
+		struct dc_context *dc,
+		unsigned int panel_inst,
+		struct abm_save_restore *pData)
+{
+	union dmub_rb_cmd cmd;
+	uint8_t panel_mask = 0x01 << panel_inst;
+	unsigned int bytes = sizeof(struct abm_save_restore);
+
+	// TODO: Optimize by only reading back final 4 bytes
+	dmub_flush_buffer_mem(&dc->dmub_srv->dmub->scratch_mem_fb);
+
+	// Copy iramtable into cw7
+	memcpy(dc->dmub_srv->dmub->scratch_mem_fb.cpu_addr, (void *)pData, bytes);
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.abm_save_restore.header.type = DMUB_CMD__ABM;
+	cmd.abm_save_restore.header.sub_type = DMUB_CMD__ABM_SAVE_RESTORE;
+
+	cmd.abm_save_restore.abm_init_config_data.src.quad_part = dc->dmub_srv->dmub->scratch_mem_fb.gpu_addr;
+	cmd.abm_save_restore.abm_init_config_data.bytes = bytes;
+	cmd.abm_save_restore.abm_init_config_data.version = DMUB_CMD_ABM_CONTROL_VERSION_1;
+	cmd.abm_save_restore.abm_init_config_data.panel_mask = panel_mask;
+
+	cmd.abm_save_restore.header.payload_bytes =
+			sizeof(struct dmub_rb_cmd_abm_save_restore) - sizeof(struct dmub_cmd_header);
+
+	dc_wake_and_execute_dmub_cmd(dc, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+
+	// Copy iramtable data into local structure
+	memcpy((void *)pData, dc->dmub_srv->dmub->scratch_mem_fb.cpu_addr, bytes);
+
+	return true;
+}
+
+bool dmub_abm_set_pipe(struct abm *abm,
+		uint32_t otg_inst,
+		uint32_t option,
+		uint32_t panel_inst,
+		uint32_t pwrseq_inst)
+{
+	union dmub_rb_cmd cmd;
+	struct dc_context *dc = abm->ctx;
+	uint8_t ramping_boundary = 0xFF;
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.abm_set_pipe.header.type = DMUB_CMD__ABM;
+	cmd.abm_set_pipe.header.sub_type = DMUB_CMD__ABM_SET_PIPE;
+	cmd.abm_set_pipe.abm_set_pipe_data.otg_inst = otg_inst;
+	cmd.abm_set_pipe.abm_set_pipe_data.pwrseq_inst = pwrseq_inst;
+	cmd.abm_set_pipe.abm_set_pipe_data.set_pipe_option = option;
+	cmd.abm_set_pipe.abm_set_pipe_data.panel_inst = panel_inst;
+	cmd.abm_set_pipe.abm_set_pipe_data.ramping_boundary = ramping_boundary;
+	cmd.abm_set_pipe.header.payload_bytes = sizeof(struct dmub_cmd_abm_set_pipe_data);
+
+	dc_wake_and_execute_dmub_cmd(dc, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+
+	return true;
+}
+
+bool dmub_abm_set_backlight_level(struct abm *abm,
+		unsigned int backlight_pwm_u16_16,
+		unsigned int frame_ramp,
+		unsigned int panel_inst)
+{
+	union dmub_rb_cmd cmd;
+	struct dc_context *dc = abm->ctx;
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.abm_set_backlight.header.type = DMUB_CMD__ABM;
+	cmd.abm_set_backlight.header.sub_type = DMUB_CMD__ABM_SET_BACKLIGHT;
+	cmd.abm_set_backlight.abm_set_backlight_data.frame_ramp = frame_ramp;
+	cmd.abm_set_backlight.abm_set_backlight_data.backlight_user_level = backlight_pwm_u16_16;
+	cmd.abm_set_backlight.abm_set_backlight_data.version = DMUB_CMD_ABM_CONTROL_VERSION_1;
+	cmd.abm_set_backlight.abm_set_backlight_data.panel_mask = (0x01 << panel_inst);
+	cmd.abm_set_backlight.header.payload_bytes = sizeof(struct dmub_cmd_abm_set_backlight_data);
+
+	dc_wake_and_execute_dmub_cmd(dc, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+
+	return true;
+}
+
+bool dmub_abm_set_event(struct abm *abm, unsigned int scaling_enable, unsigned int scaling_strength_map,
+		unsigned int panel_inst)
+{
+	union dmub_rb_cmd cmd;
+	struct dc_context *dc = abm->ctx;
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.abm_set_event.header.type = DMUB_CMD__ABM;
+	cmd.abm_set_event.header.sub_type = DMUB_CMD__ABM_SET_EVENT;
+	cmd.abm_set_event.abm_set_event_data.vb_scaling_enable = scaling_enable;
+	cmd.abm_set_event.abm_set_event_data.vb_scaling_strength_mapping = scaling_strength_map;
+	cmd.abm_set_event.abm_set_event_data.panel_mask = (1<<panel_inst);
+	cmd.abm_set_event.header.payload_bytes = sizeof(struct dmub_cmd_abm_set_event_data);
+
+	dc_wake_and_execute_dmub_cmd(dc, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+
+	return true;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dmub_abm_lcd.h b/drivers/gpu/drm/amd/display/dc/dce/dmub_abm_lcd.h
new file mode 100644
index 000000000000..13f54f1df780
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dmub_abm_lcd.h
@@ -0,0 +1,54 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DMUB_ABM_LCD_H__
+#define __DMUB_ABM_LCD_H__
+
+#include "abm.h"
+
+struct abm_save_restore;
+
+void dmub_abm_init(struct abm *abm, uint32_t backlight, uint32_t user_level);
+bool dmub_abm_set_level(struct abm *abm, uint32_t level, uint8_t panel_mask);
+unsigned int dmub_abm_get_current_backlight(struct abm *abm);
+unsigned int dmub_abm_get_target_backlight(struct abm *abm);
+void dmub_abm_init_config(struct abm *abm,
+	const char *src,
+	unsigned int bytes,
+	unsigned int inst);
+
+bool dmub_abm_set_pause(struct abm *abm, bool pause, unsigned int panel_inst, unsigned int stream_inst);
+bool dmub_abm_save_restore(
+		struct dc_context *dc,
+		unsigned int panel_inst,
+		struct abm_save_restore *pData);
+bool dmub_abm_set_pipe(struct abm *abm, uint32_t otg_inst, uint32_t option, uint32_t panel_inst, uint32_t pwrseq_inst);
+bool dmub_abm_set_backlight_level(struct abm *abm,
+		unsigned int backlight_pwm_u16_16,
+		unsigned int frame_ramp,
+		unsigned int panel_inst);
+bool dmub_abm_set_event(struct abm *abm, unsigned int scaling_enable, unsigned int scaling_strength_map,
+		unsigned int panel_inst);
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dmub_hw_lock_mgr.c b/drivers/gpu/drm/amd/display/dc/dce/dmub_hw_lock_mgr.c
new file mode 100644
index 000000000000..d37ecfdde4f1
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dmub_hw_lock_mgr.c
@@ -0,0 +1,88 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dmub_hw_lock_mgr.h"
+#include "dc_dmub_srv.h"
+#include "dc_types.h"
+#include "core_types.h"
+
+void dmub_hw_lock_mgr_cmd(struct dc_dmub_srv *dmub_srv,
+				bool lock,
+				union dmub_hw_lock_flags *hw_locks,
+				struct dmub_hw_lock_inst_flags *inst_flags)
+{
+	union dmub_rb_cmd cmd;
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.lock_hw.header.type = DMUB_CMD__HW_LOCK;
+	cmd.lock_hw.header.sub_type = 0;
+	cmd.lock_hw.header.payload_bytes = sizeof(struct dmub_cmd_lock_hw_data);
+	cmd.lock_hw.lock_hw_data.client = HW_LOCK_CLIENT_DRIVER;
+	cmd.lock_hw.lock_hw_data.lock = lock;
+	cmd.lock_hw.lock_hw_data.hw_locks.u8All = hw_locks->u8All;
+	memcpy(&cmd.lock_hw.lock_hw_data.inst_flags, inst_flags, sizeof(struct dmub_hw_lock_inst_flags));
+
+	if (!lock)
+		cmd.lock_hw.lock_hw_data.should_release = 1;
+
+	dc_wake_and_execute_dmub_cmd(dmub_srv->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+void dmub_hw_lock_mgr_inbox0_cmd(struct dc_dmub_srv *dmub_srv,
+		union dmub_inbox0_cmd_lock_hw hw_lock_cmd)
+{
+	union dmub_inbox0_data_register data = { 0 };
+
+	data.inbox0_cmd_lock_hw = hw_lock_cmd;
+	dc_dmub_srv_clear_inbox0_ack(dmub_srv);
+	dc_dmub_srv_send_inbox0_cmd(dmub_srv, data);
+	dc_dmub_srv_wait_for_inbox0_ack(dmub_srv);
+}
+
+bool should_use_dmub_lock(struct dc_link *link)
+{
+	/* ASIC doesn't support DMUB */
+	if (!link->ctx->dmub_srv)
+		return false;
+
+	if (link->psr_settings.psr_version == DC_PSR_VERSION_SU_1)
+		return true;
+
+	if (link->replay_settings.replay_feature_enabled)
+		return true;
+
+	/* only use HW lock for PSR1 on single eDP */
+	if (link->psr_settings.psr_version == DC_PSR_VERSION_1) {
+		struct dc_link *edp_links[MAX_NUM_EDP];
+		int edp_num;
+
+		dc_get_edp_links(link->dc, edp_links, &edp_num);
+
+		if (edp_num == 1)
+			return true;
+	}
+
+	return false;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dmub_hw_lock_mgr.h b/drivers/gpu/drm/amd/display/dc/dce/dmub_hw_lock_mgr.h
new file mode 100644
index 000000000000..5a72b168fb4a
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dmub_hw_lock_mgr.h
@@ -0,0 +1,42 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef _DMUB_HW_LOCK_MGR_H_
+#define _DMUB_HW_LOCK_MGR_H_
+
+#include "dc_dmub_srv.h"
+#include "core_types.h"
+
+void dmub_hw_lock_mgr_cmd(struct dc_dmub_srv *dmub_srv,
+				bool lock,
+				union dmub_hw_lock_flags *hw_locks,
+				struct dmub_hw_lock_inst_flags *inst_flags);
+
+void dmub_hw_lock_mgr_inbox0_cmd(struct dc_dmub_srv *dmub_srv,
+		union dmub_inbox0_cmd_lock_hw hw_lock_cmd);
+
+bool should_use_dmub_lock(struct dc_link *link);
+
+#endif /*_DMUB_HW_LOCK_MGR_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dmub_outbox.c b/drivers/gpu/drm/amd/display/dc/dce/dmub_outbox.c
new file mode 100644
index 000000000000..98a778996e1a
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dmub_outbox.c
@@ -0,0 +1,52 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ */
+
+#include "dc.h"
+#include "dc_dmub_srv.h"
+#include "dmub_outbox.h"
+#include "dmub/inc/dmub_cmd.h"
+
+/*
+ *  Function: dmub_enable_outbox_notification
+ *
+ *  @brief
+ *		Sends inbox cmd to dmub for enabling outbox notifications to x86.
+ *
+ *  @param
+ *		[in] dmub_srv: dmub_srv structure
+ */
+void dmub_enable_outbox_notification(struct dc_dmub_srv *dmub_srv)
+{
+	union dmub_rb_cmd cmd;
+
+	memset(&cmd, 0x0, sizeof(cmd));
+	cmd.outbox1_enable.header.type = DMUB_CMD__OUTBOX1_ENABLE;
+	cmd.outbox1_enable.header.sub_type = 0;
+	cmd.outbox1_enable.header.payload_bytes =
+		sizeof(cmd.outbox1_enable) -
+		sizeof(cmd.outbox1_enable.header);
+	cmd.outbox1_enable.enable = true;
+
+	dc_wake_and_execute_dmub_cmd(dmub_srv->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dmub_outbox.h b/drivers/gpu/drm/amd/display/dc/dce/dmub_outbox.h
new file mode 100644
index 000000000000..58ceabb9d497
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dmub_outbox.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef _DMUB_OUTBOX_H_
+#define _DMUB_OUTBOX_H_
+
+struct dc_dmub_srv;
+
+void dmub_enable_outbox_notification(struct dc_dmub_srv *dmub_srv);
+
+#endif /* _DMUB_OUTBOX_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c b/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c
new file mode 100644
index 000000000000..87af4fdc04a6
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c
@@ -0,0 +1,510 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dmub_psr.h"
+#include "dc.h"
+#include "dc_dmub_srv.h"
+#include "dmub/dmub_srv.h"
+#include "core_types.h"
+
+#define DC_TRACE_LEVEL_MESSAGE(...)	do {} while (0) /* do nothing */
+
+#define MAX_PIPES 6
+
+static const uint8_t DP_SINK_DEVICE_STR_ID_1[] = {7, 1, 8, 7, 3};
+static const uint8_t DP_SINK_DEVICE_STR_ID_2[] = {7, 1, 8, 7, 5};
+static const uint8_t DP_SINK_DEVICE_STR_ID_3[] = {0x42, 0x61, 0x6c, 0x73, 0x61};
+
+/*
+ * Convert dmcub psr state to dmcu psr state.
+ */
+static enum dc_psr_state convert_psr_state(uint32_t raw_state)
+{
+	enum dc_psr_state state = PSR_STATE0;
+
+	if (raw_state == 0)
+		state = PSR_STATE0;
+	else if (raw_state == 0x10)
+		state = PSR_STATE1;
+	else if (raw_state == 0x11)
+		state = PSR_STATE1a;
+	else if (raw_state == 0x20)
+		state = PSR_STATE2;
+	else if (raw_state == 0x21)
+		state = PSR_STATE2a;
+	else if (raw_state == 0x22)
+		state = PSR_STATE2b;
+	else if (raw_state == 0x30)
+		state = PSR_STATE3;
+	else if (raw_state == 0x31)
+		state = PSR_STATE3Init;
+	else if (raw_state == 0x40)
+		state = PSR_STATE4;
+	else if (raw_state == 0x41)
+		state = PSR_STATE4a;
+	else if (raw_state == 0x42)
+		state = PSR_STATE4b;
+	else if (raw_state == 0x43)
+		state = PSR_STATE4c;
+	else if (raw_state == 0x44)
+		state = PSR_STATE4d;
+	else if (raw_state == 0x50)
+		state = PSR_STATE5;
+	else if (raw_state == 0x51)
+		state = PSR_STATE5a;
+	else if (raw_state == 0x52)
+		state = PSR_STATE5b;
+	else if (raw_state == 0x53)
+		state = PSR_STATE5c;
+	else if (raw_state == 0x4A)
+		state = PSR_STATE4_FULL_FRAME;
+	else if (raw_state == 0x4B)
+		state = PSR_STATE4a_FULL_FRAME;
+	else if (raw_state == 0x4C)
+		state = PSR_STATE4b_FULL_FRAME;
+	else if (raw_state == 0x4D)
+		state = PSR_STATE4c_FULL_FRAME;
+	else if (raw_state == 0x4E)
+		state = PSR_STATE4_FULL_FRAME_POWERUP;
+	else if (raw_state == 0x4F)
+		state = PSR_STATE4_FULL_FRAME_HW_LOCK;
+	else if (raw_state == 0x60)
+		state = PSR_STATE_HWLOCK_MGR;
+	else if (raw_state == 0x61)
+		state = PSR_STATE_POLLVUPDATE;
+	else if (raw_state == 0x62)
+		state = PSR_STATE_RELEASE_HWLOCK_MGR_FULL_FRAME;
+	else
+		state = PSR_STATE_INVALID;
+
+	return state;
+}
+
+/*
+ * Get PSR state from firmware.
+ */
+static void dmub_psr_get_state(struct dmub_psr *dmub, enum dc_psr_state *state, uint8_t panel_inst)
+{
+	uint32_t raw_state = 0;
+	uint32_t retry_count = 0;
+
+	do {
+		// Send gpint command and wait for ack
+		if (dc_wake_and_execute_gpint(dmub->ctx, DMUB_GPINT__GET_PSR_STATE, panel_inst, &raw_state,
+					      DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY)) {
+			*state = convert_psr_state(raw_state);
+		} else {
+			// Return invalid state when GPINT times out
+			*state = PSR_STATE_INVALID;
+		}
+	} while (++retry_count <= 1000 && *state == PSR_STATE_INVALID);
+
+	// Assert if max retry hit
+	if (retry_count >= 1000 && *state == PSR_STATE_INVALID) {
+		ASSERT(0);
+		DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_ERROR,
+				WPP_BIT_FLAG_Firmware_PsrState,
+				"Unable to get PSR state from FW.");
+	} else
+		DC_TRACE_LEVEL_MESSAGE(DAL_TRACE_LEVEL_VERBOSE,
+				WPP_BIT_FLAG_Firmware_PsrState,
+				"Got PSR state from FW. PSR state: %d, Retry count: %d",
+				*state, retry_count);
+}
+
+/*
+ * Set PSR version.
+ */
+static bool dmub_psr_set_version(struct dmub_psr *dmub, struct dc_stream_state *stream, uint8_t panel_inst)
+{
+	union dmub_rb_cmd cmd;
+	struct dc_context *dc = dmub->ctx;
+
+	if (stream->link->psr_settings.psr_version == DC_PSR_VERSION_UNSUPPORTED)
+		return false;
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.psr_set_version.header.type = DMUB_CMD__PSR;
+	cmd.psr_set_version.header.sub_type = DMUB_CMD__PSR_SET_VERSION;
+	switch (stream->link->psr_settings.psr_version) {
+	case DC_PSR_VERSION_1:
+		cmd.psr_set_version.psr_set_version_data.version = PSR_VERSION_1;
+		break;
+	case DC_PSR_VERSION_SU_1:
+		cmd.psr_set_version.psr_set_version_data.version = PSR_VERSION_SU_1;
+		break;
+	case DC_PSR_VERSION_UNSUPPORTED:
+	default:
+		cmd.psr_set_version.psr_set_version_data.version = PSR_VERSION_UNSUPPORTED;
+		break;
+	}
+
+	if (cmd.psr_set_version.psr_set_version_data.version == PSR_VERSION_UNSUPPORTED)
+		return false;
+
+	cmd.psr_set_version.psr_set_version_data.cmd_version = DMUB_CMD_PSR_CONTROL_VERSION_1;
+	cmd.psr_set_version.psr_set_version_data.panel_inst = panel_inst;
+	cmd.psr_set_version.header.payload_bytes = sizeof(struct dmub_cmd_psr_set_version_data);
+
+	dc_wake_and_execute_dmub_cmd(dc, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+
+	return true;
+}
+
+/*
+ * Enable/Disable PSR.
+ */
+static void dmub_psr_enable(struct dmub_psr *dmub, bool enable, bool wait, uint8_t panel_inst)
+{
+	union dmub_rb_cmd cmd;
+	struct dc_context *dc = dmub->ctx;
+	uint32_t retry_count;
+	enum dc_psr_state state = PSR_STATE0;
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.psr_enable.header.type = DMUB_CMD__PSR;
+
+	cmd.psr_enable.data.cmd_version = DMUB_CMD_PSR_CONTROL_VERSION_1;
+	cmd.psr_enable.data.panel_inst = panel_inst;
+
+	if (enable)
+		cmd.psr_enable.header.sub_type = DMUB_CMD__PSR_ENABLE;
+	else
+		cmd.psr_enable.header.sub_type = DMUB_CMD__PSR_DISABLE;
+
+	cmd.psr_enable.header.payload_bytes = 0; // Send header only
+
+	dc_wake_and_execute_dmub_cmd(dc->dmub_srv->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+
+	/* Below loops 1000 x 500us = 500 ms.
+	 *  Exit PSR may need to wait 1-2 frames to power up. Timeout after at
+	 *  least a few frames. Should never hit the max retry assert below.
+	 */
+	if (wait) {
+		for (retry_count = 0; retry_count <= 1000; retry_count++) {
+			dmub_psr_get_state(dmub, &state, panel_inst);
+
+			if (enable) {
+				if (state != PSR_STATE0)
+					break;
+			} else {
+				if (state == PSR_STATE0)
+					break;
+			}
+
+			/* must *not* be fsleep - this can be called from high irq levels */
+			udelay(500);
+		}
+
+		/* assert if max retry hit */
+		if (retry_count >= 1000)
+			ASSERT(0);
+	}
+}
+
+/*
+ * Set PSR level.
+ */
+static void dmub_psr_set_level(struct dmub_psr *dmub, uint16_t psr_level, uint8_t panel_inst)
+{
+	union dmub_rb_cmd cmd;
+	enum dc_psr_state state = PSR_STATE0;
+	struct dc_context *dc = dmub->ctx;
+
+	dmub_psr_get_state(dmub, &state, panel_inst);
+
+	if (state == PSR_STATE0)
+		return;
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.psr_set_level.header.type = DMUB_CMD__PSR;
+	cmd.psr_set_level.header.sub_type = DMUB_CMD__PSR_SET_LEVEL;
+	cmd.psr_set_level.header.payload_bytes = sizeof(struct dmub_cmd_psr_set_level_data);
+	cmd.psr_set_level.psr_set_level_data.psr_level = psr_level;
+	cmd.psr_set_level.psr_set_level_data.cmd_version = DMUB_CMD_PSR_CONTROL_VERSION_1;
+	cmd.psr_set_level.psr_set_level_data.panel_inst = panel_inst;
+	dc_wake_and_execute_dmub_cmd(dc, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+/*
+ * Set PSR vtotal requirement for FreeSync PSR.
+ */
+static void dmub_psr_set_sink_vtotal_in_psr_active(struct dmub_psr *dmub,
+		uint16_t psr_vtotal_idle, uint16_t psr_vtotal_su)
+{
+	union dmub_rb_cmd cmd;
+	struct dc_context *dc = dmub->ctx;
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.psr_set_vtotal.header.type = DMUB_CMD__PSR;
+	cmd.psr_set_vtotal.header.sub_type = DMUB_CMD__SET_SINK_VTOTAL_IN_PSR_ACTIVE;
+	cmd.psr_set_vtotal.header.payload_bytes = sizeof(struct dmub_cmd_psr_set_vtotal_data);
+	cmd.psr_set_vtotal.psr_set_vtotal_data.psr_vtotal_idle = psr_vtotal_idle;
+	cmd.psr_set_vtotal.psr_set_vtotal_data.psr_vtotal_su = psr_vtotal_su;
+
+	dc_wake_and_execute_dmub_cmd(dc, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+/*
+ * Set PSR power optimization flags.
+ */
+static void dmub_psr_set_power_opt(struct dmub_psr *dmub, unsigned int power_opt, uint8_t panel_inst)
+{
+	union dmub_rb_cmd cmd;
+	struct dc_context *dc = dmub->ctx;
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.psr_set_power_opt.header.type = DMUB_CMD__PSR;
+	cmd.psr_set_power_opt.header.sub_type = DMUB_CMD__SET_PSR_POWER_OPT;
+	cmd.psr_set_power_opt.header.payload_bytes = sizeof(struct dmub_cmd_psr_set_power_opt_data);
+	cmd.psr_set_power_opt.psr_set_power_opt_data.cmd_version = DMUB_CMD_PSR_CONTROL_VERSION_1;
+	cmd.psr_set_power_opt.psr_set_power_opt_data.power_opt = power_opt;
+	cmd.psr_set_power_opt.psr_set_power_opt_data.panel_inst = panel_inst;
+
+	dc_wake_and_execute_dmub_cmd(dc, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+/*
+ * Setup PSR by programming phy registers and sending psr hw context values to firmware.
+ */
+static bool dmub_psr_copy_settings(struct dmub_psr *dmub,
+		struct dc_link *link,
+		struct psr_context *psr_context,
+		uint8_t panel_inst)
+{
+	union dmub_rb_cmd cmd = { 0 };
+	struct dc_context *dc = dmub->ctx;
+	struct dmub_cmd_psr_copy_settings_data *copy_settings_data
+		= &cmd.psr_copy_settings.psr_copy_settings_data;
+	struct pipe_ctx *pipe_ctx = NULL;
+	struct resource_context *res_ctx = &link->ctx->dc->current_state->res_ctx;
+	int i = 0;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		if (res_ctx->pipe_ctx[i].stream &&
+		    res_ctx->pipe_ctx[i].stream->link == link &&
+		    res_ctx->pipe_ctx[i].stream->link->connector_signal == SIGNAL_TYPE_EDP) {
+			pipe_ctx = &res_ctx->pipe_ctx[i];
+			//TODO: refactor for multi edp support
+			break;
+		}
+	}
+
+	if (!pipe_ctx)
+		return false;
+
+	// First, set the psr version
+	if (!dmub_psr_set_version(dmub, pipe_ctx->stream, panel_inst))
+		return false;
+
+	// Program DP DPHY fast training registers
+	link->link_enc->funcs->psr_program_dp_dphy_fast_training(link->link_enc,
+			psr_context->psrExitLinkTrainingRequired);
+
+	// Program DP_SEC_CNTL1 register to set transmission GPS0 line num and priority to high
+	link->link_enc->funcs->psr_program_secondary_packet(link->link_enc,
+			psr_context->sdpTransmitLineNumDeadline);
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.psr_copy_settings.header.type = DMUB_CMD__PSR;
+	cmd.psr_copy_settings.header.sub_type = DMUB_CMD__PSR_COPY_SETTINGS;
+	cmd.psr_copy_settings.header.payload_bytes = sizeof(struct dmub_cmd_psr_copy_settings_data);
+
+	// Hw insts
+	copy_settings_data->dpphy_inst				= psr_context->transmitterId;
+	copy_settings_data->aux_inst				= psr_context->channel;
+	copy_settings_data->digfe_inst				= psr_context->engineId;
+	copy_settings_data->digbe_inst				= psr_context->transmitterId;
+
+	copy_settings_data->mpcc_inst				= pipe_ctx->plane_res.mpcc_inst;
+
+	if (pipe_ctx->plane_res.dpp)
+		copy_settings_data->dpp_inst			= pipe_ctx->plane_res.dpp->inst;
+	else
+		copy_settings_data->dpp_inst			= 0;
+	if (pipe_ctx->stream_res.opp)
+		copy_settings_data->opp_inst			= pipe_ctx->stream_res.opp->inst;
+	else
+		copy_settings_data->opp_inst			= 0;
+	if (pipe_ctx->stream_res.tg)
+		copy_settings_data->otg_inst			= pipe_ctx->stream_res.tg->inst;
+	else
+		copy_settings_data->otg_inst			= 0;
+
+	// Misc
+	copy_settings_data->use_phy_fsm             = link->ctx->dc->debug.psr_power_use_phy_fsm;
+	copy_settings_data->psr_level				= psr_context->psr_level.u32all;
+	copy_settings_data->smu_optimizations_en		= psr_context->allow_smu_optimizations;
+	copy_settings_data->multi_disp_optimizations_en	= psr_context->allow_multi_disp_optimizations;
+	copy_settings_data->frame_delay				= psr_context->frame_delay;
+	copy_settings_data->frame_cap_ind			= psr_context->psrFrameCaptureIndicationReq;
+	copy_settings_data->init_sdp_deadline			= psr_context->sdpTransmitLineNumDeadline;
+	copy_settings_data->debug.u32All = 0;
+	copy_settings_data->debug.bitfields.visual_confirm	= dc->dc->debug.visual_confirm == VISUAL_CONFIRM_PSR;
+	copy_settings_data->debug.bitfields.use_hw_lock_mgr		= 1;
+	copy_settings_data->debug.bitfields.force_full_frame_update	= 0;
+	copy_settings_data->debug.bitfields.enable_ips_visual_confirm = dc->dc->debug.enable_ips_visual_confirm;
+
+	if (psr_context->su_granularity_required == 0)
+		copy_settings_data->su_y_granularity = 0;
+	else
+		copy_settings_data->su_y_granularity = psr_context->su_y_granularity;
+
+	copy_settings_data->line_capture_indication = 0;
+	copy_settings_data->line_time_in_us = psr_context->line_time_in_us;
+	copy_settings_data->rate_control_caps = psr_context->rate_control_caps;
+	copy_settings_data->fec_enable_status = (link->fec_state == dc_link_fec_enabled);
+	copy_settings_data->fec_enable_delay_in100us = link->dc->debug.fec_enable_delay_in100us;
+	copy_settings_data->cmd_version =  DMUB_CMD_PSR_CONTROL_VERSION_1;
+	copy_settings_data->panel_inst = panel_inst;
+	copy_settings_data->dsc_enable_status = (pipe_ctx->stream->timing.flags.DSC == 1);
+
+	/**
+	 * WA for PSRSU+DSC on specific TCON, if DSC is enabled, force PSRSU as ffu mode(full frame update)
+	 * Note that PSRSU+DSC is still under development.
+	 */
+	if (copy_settings_data->dsc_enable_status &&
+		link->dpcd_caps.sink_dev_id == DP_DEVICE_ID_38EC11 &&
+		!memcmp(link->dpcd_caps.sink_dev_id_str, DP_SINK_DEVICE_STR_ID_1,
+			sizeof(DP_SINK_DEVICE_STR_ID_1)))
+		link->psr_settings.force_ffu_mode = 1;
+
+	copy_settings_data->force_ffu_mode = link->psr_settings.force_ffu_mode || psr_context->os_request_force_ffu;
+
+	if (((link->dpcd_caps.fec_cap.bits.FEC_CAPABLE &&
+		!link->dc->debug.disable_fec) &&
+		(link->dpcd_caps.dsc_caps.dsc_basic_caps.fields.dsc_support.DSC_SUPPORT &&
+		!link->panel_config.dsc.disable_dsc_edp &&
+		link->dc->caps.edp_dsc_support)) &&
+		link->dpcd_caps.sink_dev_id == DP_DEVICE_ID_38EC11 &&
+		(!memcmp(link->dpcd_caps.sink_dev_id_str, DP_SINK_DEVICE_STR_ID_1,
+			sizeof(DP_SINK_DEVICE_STR_ID_1)) ||
+		!memcmp(link->dpcd_caps.sink_dev_id_str, DP_SINK_DEVICE_STR_ID_2,
+			sizeof(DP_SINK_DEVICE_STR_ID_2))))
+		copy_settings_data->debug.bitfields.force_wakeup_by_tps3 = 1;
+	else
+		copy_settings_data->debug.bitfields.force_wakeup_by_tps3 = 0;
+
+	if (link->psr_settings.psr_version == DC_PSR_VERSION_1 &&
+		link->dpcd_caps.sink_dev_id == DP_DEVICE_ID_0022B9 &&
+		!memcmp(link->dpcd_caps.sink_dev_id_str, DP_SINK_DEVICE_STR_ID_3,
+			sizeof(DP_SINK_DEVICE_STR_ID_3))) {
+		copy_settings_data->poweroff_before_vertical_line = 16;
+	}
+
+	//WA for PSR1 on specific TCON, require frame delay for frame re-lock
+	copy_settings_data->relock_delay_frame_cnt = 0;
+	if (link->dpcd_caps.sink_dev_id == DP_BRANCH_DEVICE_ID_001CF8)
+		copy_settings_data->relock_delay_frame_cnt = 2;
+
+	copy_settings_data->power_down_phy_before_disable_stream =
+		link->psr_settings.power_down_phy_before_disable_stream;
+
+	copy_settings_data->dsc_slice_height = psr_context->dsc_slice_height;
+
+	dc_wake_and_execute_dmub_cmd(dc, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+
+	return true;
+}
+
+/*
+ * Send command to PSR to force static ENTER and ignore all state changes until exit
+ */
+static void dmub_psr_force_static(struct dmub_psr *dmub, uint8_t panel_inst)
+{
+	union dmub_rb_cmd cmd;
+	struct dc_context *dc = dmub->ctx;
+
+	memset(&cmd, 0, sizeof(cmd));
+
+	cmd.psr_force_static.psr_force_static_data.panel_inst = panel_inst;
+	cmd.psr_force_static.psr_force_static_data.cmd_version = DMUB_CMD_PSR_CONTROL_VERSION_1;
+	cmd.psr_force_static.header.type = DMUB_CMD__PSR;
+	cmd.psr_force_static.header.sub_type = DMUB_CMD__PSR_FORCE_STATIC;
+	cmd.psr_enable.header.payload_bytes = 0;
+
+	dc_wake_and_execute_dmub_cmd(dc, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+/*
+ * Get PSR residency from firmware.
+ */
+static void dmub_psr_get_residency(struct dmub_psr *dmub, uint32_t *residency,
+	uint8_t panel_inst, enum psr_residency_mode mode)
+{
+	uint16_t param = (uint16_t)(panel_inst << 8);
+
+	param |= mode;
+
+	/* Send gpint command and wait for ack */
+	dc_wake_and_execute_gpint(dmub->ctx, DMUB_GPINT__PSR_RESIDENCY, param, residency,
+				  DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY);
+}
+
+static const struct dmub_psr_funcs psr_funcs = {
+	.psr_copy_settings		= dmub_psr_copy_settings,
+	.psr_enable			= dmub_psr_enable,
+	.psr_get_state			= dmub_psr_get_state,
+	.psr_set_level			= dmub_psr_set_level,
+	.psr_force_static		= dmub_psr_force_static,
+	.psr_get_residency		= dmub_psr_get_residency,
+	.psr_set_sink_vtotal_in_psr_active	= dmub_psr_set_sink_vtotal_in_psr_active,
+	.psr_set_power_opt		= dmub_psr_set_power_opt,
+};
+
+/*
+ * Construct PSR object.
+ */
+static void dmub_psr_construct(struct dmub_psr *psr, struct dc_context *ctx)
+{
+	psr->ctx = ctx;
+	psr->funcs = &psr_funcs;
+}
+
+/*
+ * Allocate and initialize PSR object.
+ */
+struct dmub_psr *dmub_psr_create(struct dc_context *ctx)
+{
+	struct dmub_psr *psr = kzalloc(sizeof(struct dmub_psr), GFP_KERNEL);
+
+	if (psr == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	dmub_psr_construct(psr, ctx);
+
+	return psr;
+}
+
+/*
+ * Deallocate PSR object.
+ */
+void dmub_psr_destroy(struct dmub_psr **dmub)
+{
+	kfree(*dmub);
+	*dmub = NULL;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.h b/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.h
new file mode 100644
index 000000000000..a6e282d950c3
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.h
@@ -0,0 +1,60 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef _DMUB_PSR_H_
+#define _DMUB_PSR_H_
+
+#include "dc_types.h"
+#include "dmub_cmd.h"
+struct dc_link;
+struct dmub_psr_funcs;
+
+struct dmub_psr {
+	struct dc_context *ctx;
+	const struct dmub_psr_funcs *funcs;
+};
+
+struct dmub_psr_funcs {
+	bool (*psr_copy_settings)(struct dmub_psr *dmub, struct dc_link *link,
+	struct psr_context *psr_context, uint8_t panel_inst);
+	void (*psr_enable)(struct dmub_psr *dmub, bool enable, bool wait,
+	uint8_t panel_inst);
+	void (*psr_get_state)(struct dmub_psr *dmub, enum dc_psr_state *dc_psr_state,
+	uint8_t panel_inst);
+	void (*psr_set_level)(struct dmub_psr *dmub, uint16_t psr_level,
+	uint8_t panel_inst);
+	void (*psr_force_static)(struct dmub_psr *dmub, uint8_t panel_inst);
+	void (*psr_get_residency)(struct dmub_psr *dmub, uint32_t *residency,
+	uint8_t panel_inst,  enum psr_residency_mode mode);
+	void (*psr_set_sink_vtotal_in_psr_active)(struct dmub_psr *dmub,
+	uint16_t psr_vtotal_idle, uint16_t psr_vtotal_su);
+	void (*psr_set_power_opt)(struct dmub_psr *dmub, unsigned int power_opt, uint8_t panel_inst);
+};
+
+struct dmub_psr *dmub_psr_create(struct dc_context *ctx);
+void dmub_psr_destroy(struct dmub_psr **dmub);
+
+
+#endif /* _DMUB_PSR_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dmub_replay.c b/drivers/gpu/drm/amd/display/dc/dce/dmub_replay.c
new file mode 100644
index 000000000000..f9542edff14b
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dmub_replay.c
@@ -0,0 +1,457 @@
+// SPDX-License-Identifier: MIT
+//
+// Copyright 2024 Advanced Micro Devices, Inc.
+
+#include "dc.h"
+#include "link_service.h"
+#include "dc_dmub_srv.h"
+#include "dmub/dmub_srv.h"
+#include "core_types.h"
+#include "dmub_replay.h"
+
+#define DC_TRACE_LEVEL_MESSAGE(...) /* do nothing */
+
+#define MAX_PIPES 6
+
+#define GPINT_RETRY_NUM 20
+
+static const uint8_t DP_SINK_DEVICE_STR_ID_1[] = {7, 1, 8, 7, 3};
+static const uint8_t DP_SINK_DEVICE_STR_ID_2[] = {7, 1, 8, 7, 5};
+
+/*
+ * Get Replay state from firmware.
+ */
+static void dmub_replay_get_state(struct dmub_replay *dmub, enum replay_state *state, uint8_t panel_inst)
+{
+	uint32_t retry_count = 0;
+
+	do {
+		// Send gpint command and wait for ack
+		if (!dc_wake_and_execute_gpint(dmub->ctx, DMUB_GPINT__GET_REPLAY_STATE, panel_inst,
+					       (uint32_t *)state, DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY)) {
+			// Return invalid state when GPINT times out
+			*state = REPLAY_STATE_INVALID;
+		}
+	} while (++retry_count <= 1000 && *state == REPLAY_STATE_INVALID);
+
+	// Assert if max retry hit
+	if (retry_count >= 1000 && *state == REPLAY_STATE_INVALID) {
+		ASSERT(0);
+		/* To-do: Add retry fail log */
+	}
+}
+
+/*
+ * Enable/Disable Replay.
+ */
+static void dmub_replay_enable(struct dmub_replay *dmub, bool enable, bool wait, uint8_t panel_inst)
+{
+	union dmub_rb_cmd cmd;
+	struct dc_context *dc = dmub->ctx;
+	uint32_t retry_count;
+	enum replay_state state = REPLAY_STATE_0;
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.replay_enable.header.type = DMUB_CMD__REPLAY;
+	cmd.replay_enable.data.panel_inst = panel_inst;
+
+	cmd.replay_enable.header.sub_type = DMUB_CMD__REPLAY_ENABLE;
+	if (enable)
+		cmd.replay_enable.data.enable = REPLAY_ENABLE;
+	else
+		cmd.replay_enable.data.enable = REPLAY_DISABLE;
+
+	cmd.replay_enable.header.payload_bytes = sizeof(struct dmub_rb_cmd_replay_enable_data);
+
+	dc_wake_and_execute_dmub_cmd(dc, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+
+	/* Below loops 1000 x 500us = 500 ms.
+	 *  Exit REPLAY may need to wait 1-2 frames to power up. Timeout after at
+	 *  least a few frames. Should never hit the max retry assert below.
+	 */
+	if (wait) {
+		for (retry_count = 0; retry_count <= 1000; retry_count++) {
+			dmub_replay_get_state(dmub, &state, panel_inst);
+
+			if (enable) {
+				if (state != REPLAY_STATE_0)
+					break;
+			} else {
+				if (state == REPLAY_STATE_0)
+					break;
+			}
+
+			/* must *not* be fsleep - this can be called from high irq levels */
+			udelay(500);
+		}
+
+		/* assert if max retry hit */
+		if (retry_count >= 1000)
+			ASSERT(0);
+	}
+}
+
+/*
+ * Set REPLAY power optimization flags.
+ */
+static void dmub_replay_set_power_opt(struct dmub_replay *dmub, unsigned int power_opt, uint8_t panel_inst)
+{
+	union dmub_rb_cmd cmd;
+	struct dc_context *dc = dmub->ctx;
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.replay_set_power_opt.header.type = DMUB_CMD__REPLAY;
+	cmd.replay_set_power_opt.header.sub_type = DMUB_CMD__SET_REPLAY_POWER_OPT;
+	cmd.replay_set_power_opt.header.payload_bytes = sizeof(struct dmub_cmd_replay_set_power_opt_data);
+	cmd.replay_set_power_opt.replay_set_power_opt_data.power_opt = power_opt;
+	cmd.replay_set_power_opt.replay_set_power_opt_data.panel_inst = panel_inst;
+
+	dc_wake_and_execute_dmub_cmd(dc, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+/*
+ * Setup Replay by programming phy registers and sending replay hw context values to firmware.
+ */
+static bool dmub_replay_copy_settings(struct dmub_replay *dmub,
+	struct dc_link *link,
+	struct replay_context *replay_context,
+	uint8_t panel_inst)
+{
+	union dmub_rb_cmd cmd;
+	struct dc_context *dc = dmub->ctx;
+	struct dmub_cmd_replay_copy_settings_data *copy_settings_data
+		= &cmd.replay_copy_settings.replay_copy_settings_data;
+	struct pipe_ctx *pipe_ctx = NULL;
+	struct resource_context *res_ctx = &link->ctx->dc->current_state->res_ctx;
+	int i = 0;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		if (res_ctx &&
+			res_ctx->pipe_ctx[i].stream &&
+			res_ctx->pipe_ctx[i].stream->link &&
+			res_ctx->pipe_ctx[i].stream->link == link &&
+			res_ctx->pipe_ctx[i].stream->link->connector_signal == SIGNAL_TYPE_EDP) {
+			pipe_ctx = &res_ctx->pipe_ctx[i];
+			//TODO: refactor for multi edp support
+			break;
+		}
+	}
+
+	if (!pipe_ctx)
+		return false;
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.replay_copy_settings.header.type = DMUB_CMD__REPLAY;
+	cmd.replay_copy_settings.header.sub_type = DMUB_CMD__REPLAY_COPY_SETTINGS;
+	cmd.replay_copy_settings.header.payload_bytes = sizeof(struct dmub_cmd_replay_copy_settings_data);
+
+	// HW insts
+	copy_settings_data->aux_inst				= replay_context->aux_inst;
+	copy_settings_data->digbe_inst				= replay_context->digbe_inst;
+	copy_settings_data->digfe_inst				= replay_context->digfe_inst;
+
+	if (pipe_ctx->plane_res.dpp)
+		copy_settings_data->dpp_inst			= pipe_ctx->plane_res.dpp->inst;
+	else
+		copy_settings_data->dpp_inst			= 0;
+	if (pipe_ctx->stream_res.tg)
+		copy_settings_data->otg_inst			= pipe_ctx->stream_res.tg->inst;
+	else
+		copy_settings_data->otg_inst			= 0;
+
+	copy_settings_data->dpphy_inst				= link->link_enc->transmitter;
+
+	// Misc
+	copy_settings_data->line_time_in_ns			= replay_context->line_time_in_ns;
+	copy_settings_data->panel_inst				= panel_inst;
+	copy_settings_data->debug.u32All			= link->replay_settings.config.debug_flags;
+	copy_settings_data->pixel_deviation_per_line		= link->dpcd_caps.pr_info.pixel_deviation_per_line;
+	copy_settings_data->max_deviation_line			= link->dpcd_caps.pr_info.max_deviation_line;
+	copy_settings_data->smu_optimizations_en		= link->replay_settings.replay_smu_opt_enable;
+	copy_settings_data->replay_timing_sync_supported = link->replay_settings.config.replay_timing_sync_supported;
+	copy_settings_data->replay_support_fast_resync_in_ultra_sleep_mode = link->replay_settings.config.replay_support_fast_resync_in_ultra_sleep_mode;
+
+	copy_settings_data->debug.bitfields.enable_ips_visual_confirm = dc->dc->debug.enable_ips_visual_confirm;
+
+	copy_settings_data->flags.u32All = 0;
+	copy_settings_data->flags.bitfields.fec_enable_status = (link->fec_state == dc_link_fec_enabled);
+	copy_settings_data->flags.bitfields.dsc_enable_status = (pipe_ctx->stream->timing.flags.DSC == 1);
+	// WA for PSRSU+DSC on specific TCON, if DSC is enabled, force PSRSU as ffu mode(full frame update)
+	if (((link->dpcd_caps.fec_cap.bits.FEC_CAPABLE &&
+		!link->dc->debug.disable_fec) &&
+		(link->dpcd_caps.dsc_caps.dsc_basic_caps.fields.dsc_support.DSC_SUPPORT &&
+		!link->panel_config.dsc.disable_dsc_edp &&
+		link->dc->caps.edp_dsc_support)) &&
+		link->dpcd_caps.sink_dev_id == DP_DEVICE_ID_38EC11 &&
+		(!memcmp(link->dpcd_caps.sink_dev_id_str, DP_SINK_DEVICE_STR_ID_1,
+			sizeof(DP_SINK_DEVICE_STR_ID_1)) ||
+		!memcmp(link->dpcd_caps.sink_dev_id_str, DP_SINK_DEVICE_STR_ID_2,
+			sizeof(DP_SINK_DEVICE_STR_ID_2))))
+		copy_settings_data->flags.bitfields.force_wakeup_by_tps3 = 1;
+	else
+		copy_settings_data->flags.bitfields.force_wakeup_by_tps3 = 0;
+
+	copy_settings_data->flags.bitfields.alpm_mode = (enum dmub_alpm_mode)link->replay_settings.config.alpm_mode;
+	if (link->replay_settings.config.alpm_mode == DC_ALPM_AUXLESS) {
+		copy_settings_data->auxless_alpm_data.lfps_setup_ns = dc->dc->debug.auxless_alpm_lfps_setup_ns;
+		copy_settings_data->auxless_alpm_data.lfps_period_ns = dc->dc->debug.auxless_alpm_lfps_period_ns;
+		copy_settings_data->auxless_alpm_data.lfps_silence_ns = dc->dc->debug.auxless_alpm_lfps_silence_ns;
+		copy_settings_data->auxless_alpm_data.lfps_t1_t2_override_us =
+			dc->dc->debug.auxless_alpm_lfps_t1t2_us;
+		copy_settings_data->auxless_alpm_data.lfps_t1_t2_offset_us =
+			dc->dc->debug.auxless_alpm_lfps_t1t2_offset_us;
+		copy_settings_data->auxless_alpm_data.lttpr_count = link->dc->link_srv->dp_get_lttpr_count(link);
+	}
+
+	dc_wake_and_execute_dmub_cmd(dc, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+
+	return true;
+}
+
+/*
+ * Set coasting vtotal.
+ */
+static void dmub_replay_set_coasting_vtotal(struct dmub_replay *dmub,
+		uint32_t coasting_vtotal,
+		uint8_t panel_inst)
+{
+	union dmub_rb_cmd cmd;
+	struct dc_context *dc = dmub->ctx;
+	struct dmub_rb_cmd_replay_set_coasting_vtotal *pCmd = NULL;
+
+	pCmd = &(cmd.replay_set_coasting_vtotal);
+
+	memset(&cmd, 0, sizeof(cmd));
+	pCmd->header.type = DMUB_CMD__REPLAY;
+	pCmd->header.sub_type = DMUB_CMD__REPLAY_SET_COASTING_VTOTAL;
+	pCmd->header.payload_bytes = sizeof(struct dmub_cmd_replay_set_coasting_vtotal_data);
+	pCmd->replay_set_coasting_vtotal_data.coasting_vtotal = (coasting_vtotal & 0xFFFF);
+	pCmd->replay_set_coasting_vtotal_data.coasting_vtotal_high = (coasting_vtotal & 0xFFFF0000) >> 16;
+
+	dc_wake_and_execute_dmub_cmd(dc, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+/*
+ * Get Replay residency from firmware.
+ */
+static void dmub_replay_residency(struct dmub_replay *dmub, uint8_t panel_inst,
+	uint32_t *residency, const bool is_start, enum pr_residency_mode mode)
+{
+	uint16_t param = (uint16_t)(panel_inst << 8);
+	uint32_t i = 0;
+
+	switch (mode) {
+	case PR_RESIDENCY_MODE_PHY:
+		param |= REPLAY_RESIDENCY_FIELD_MODE_PHY;
+		break;
+	case PR_RESIDENCY_MODE_ALPM:
+		param |= REPLAY_RESIDENCY_FIELD_MODE_ALPM;
+		break;
+	case PR_RESIDENCY_MODE_IPS2:
+		param |= REPLAY_RESIDENCY_REVISION_1;
+		param |= REPLAY_RESIDENCY_FIELD_MODE2_IPS;
+		break;
+	case PR_RESIDENCY_MODE_FRAME_CNT:
+		param |= REPLAY_RESIDENCY_REVISION_1;
+		param |= REPLAY_RESIDENCY_FIELD_MODE2_FRAME_CNT;
+		break;
+	case PR_RESIDENCY_MODE_ENABLEMENT_PERIOD:
+		param |= REPLAY_RESIDENCY_REVISION_1;
+		param |= REPLAY_RESIDENCY_FIELD_MODE2_EN_PERIOD;
+		break;
+	default:
+		break;
+	}
+
+	if (is_start)
+		param |= REPLAY_RESIDENCY_ENABLE;
+
+	for (i = 0; i < GPINT_RETRY_NUM; i++) {
+		// Send gpint command and wait for ack
+		if (dc_wake_and_execute_gpint(dmub->ctx, DMUB_GPINT__REPLAY_RESIDENCY, param,
+			residency, DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY))
+			return;
+
+		udelay(100);
+	}
+
+	// it means gpint retry many times
+	*residency = 0;
+}
+
+/*
+ * Set REPLAY power optimization flags and coasting vtotal.
+ */
+static void dmub_replay_set_power_opt_and_coasting_vtotal(struct dmub_replay *dmub,
+		unsigned int power_opt, uint8_t panel_inst, uint32_t coasting_vtotal)
+{
+	union dmub_rb_cmd cmd;
+	struct dc_context *dc = dmub->ctx;
+	struct dmub_rb_cmd_replay_set_power_opt_and_coasting_vtotal *pCmd = NULL;
+
+	pCmd = &(cmd.replay_set_power_opt_and_coasting_vtotal);
+
+	memset(&cmd, 0, sizeof(cmd));
+	pCmd->header.type = DMUB_CMD__REPLAY;
+	pCmd->header.sub_type = DMUB_CMD__REPLAY_SET_POWER_OPT_AND_COASTING_VTOTAL;
+	pCmd->header.payload_bytes =
+			sizeof(struct dmub_rb_cmd_replay_set_power_opt_and_coasting_vtotal) -
+			sizeof(struct dmub_cmd_header);
+	pCmd->replay_set_power_opt_data.power_opt = power_opt;
+	pCmd->replay_set_power_opt_data.panel_inst = panel_inst;
+	pCmd->replay_set_coasting_vtotal_data.coasting_vtotal = (coasting_vtotal & 0xFFFF);
+	pCmd->replay_set_coasting_vtotal_data.coasting_vtotal_high = (coasting_vtotal & 0xFFFF0000) >> 16;
+
+	dc_wake_and_execute_dmub_cmd(dc, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+/*
+ * send Replay general cmd to DMUB.
+ */
+static void dmub_replay_send_cmd(struct dmub_replay *dmub,
+		enum replay_FW_Message_type msg, union dmub_replay_cmd_set *cmd_element)
+{
+	union dmub_rb_cmd cmd;
+	struct dc_context *ctx = NULL;
+
+	if (dmub == NULL || cmd_element == NULL)
+		return;
+
+	ctx = dmub->ctx;
+	if (ctx != NULL) {
+
+		if (msg != Replay_Msg_Not_Support) {
+			memset(&cmd, 0, sizeof(cmd));
+			//Header
+			cmd.replay_set_timing_sync.header.type = DMUB_CMD__REPLAY;
+		} else
+			return;
+	} else
+		return;
+
+	switch (msg) {
+	case Replay_Set_Timing_Sync_Supported:
+		//Header
+		cmd.replay_set_timing_sync.header.sub_type =
+			DMUB_CMD__REPLAY_SET_TIMING_SYNC_SUPPORTED;
+		cmd.replay_set_timing_sync.header.payload_bytes =
+			sizeof(struct dmub_rb_cmd_replay_set_timing_sync) -
+			sizeof(struct dmub_cmd_header);
+		//Cmd Body
+		cmd.replay_set_timing_sync.replay_set_timing_sync_data.panel_inst =
+						cmd_element->sync_data.panel_inst;
+		cmd.replay_set_timing_sync.replay_set_timing_sync_data.timing_sync_supported =
+						cmd_element->sync_data.timing_sync_supported;
+		break;
+	case Replay_Set_Residency_Frameupdate_Timer:
+		//Header
+		cmd.replay_set_frameupdate_timer.header.sub_type =
+			DMUB_CMD__REPLAY_SET_RESIDENCY_FRAMEUPDATE_TIMER;
+		cmd.replay_set_frameupdate_timer.header.payload_bytes =
+			sizeof(struct dmub_rb_cmd_replay_set_frameupdate_timer) -
+			sizeof(struct dmub_cmd_header);
+		//Cmd Body
+		cmd.replay_set_frameupdate_timer.data.panel_inst =
+						cmd_element->panel_inst;
+		cmd.replay_set_frameupdate_timer.data.enable =
+						cmd_element->timer_data.enable;
+		cmd.replay_set_frameupdate_timer.data.frameupdate_count =
+						cmd_element->timer_data.frameupdate_count;
+		break;
+	case Replay_Set_Pseudo_VTotal:
+		//Header
+		cmd.replay_set_pseudo_vtotal.header.sub_type =
+			DMUB_CMD__REPLAY_SET_PSEUDO_VTOTAL;
+		cmd.replay_set_pseudo_vtotal.header.payload_bytes =
+			sizeof(struct dmub_rb_cmd_replay_set_pseudo_vtotal) -
+			sizeof(struct dmub_cmd_header);
+		//Cmd Body
+		cmd.replay_set_pseudo_vtotal.data.panel_inst =
+			cmd_element->pseudo_vtotal_data.panel_inst;
+		cmd.replay_set_pseudo_vtotal.data.vtotal =
+			cmd_element->pseudo_vtotal_data.vtotal;
+		break;
+	case Replay_Disabled_Adaptive_Sync_SDP:
+		//Header
+		cmd.replay_disabled_adaptive_sync_sdp.header.sub_type =
+			DMUB_CMD__REPLAY_DISABLED_ADAPTIVE_SYNC_SDP;
+		cmd.replay_disabled_adaptive_sync_sdp.header.payload_bytes =
+			sizeof(struct dmub_rb_cmd_replay_disabled_adaptive_sync_sdp) -
+			sizeof(struct dmub_cmd_header);
+		//Cmd Body
+		cmd.replay_disabled_adaptive_sync_sdp.data.panel_inst =
+			cmd_element->disabled_adaptive_sync_sdp_data.panel_inst;
+		cmd.replay_disabled_adaptive_sync_sdp.data.force_disabled =
+			cmd_element->disabled_adaptive_sync_sdp_data.force_disabled;
+		break;
+	case Replay_Set_General_Cmd:
+		//Header
+		cmd.replay_set_general_cmd.header.sub_type =
+			DMUB_CMD__REPLAY_SET_GENERAL_CMD;
+		cmd.replay_set_general_cmd.header.payload_bytes =
+			sizeof(struct dmub_rb_cmd_replay_set_general_cmd) -
+			sizeof(struct dmub_cmd_header);
+		//Cmd Body
+		cmd.replay_set_general_cmd.data.panel_inst =
+			cmd_element->set_general_cmd_data.panel_inst;
+		cmd.replay_set_general_cmd.data.subtype =
+			cmd_element->set_general_cmd_data.subtype;
+		cmd.replay_set_general_cmd.data.param1 =
+			cmd_element->set_general_cmd_data.param1;
+		cmd.replay_set_general_cmd.data.param2 =
+			cmd_element->set_general_cmd_data.param2;
+		break;
+	case Replay_Msg_Not_Support:
+	default:
+		return;
+		break;
+	}
+
+	dc_wake_and_execute_dmub_cmd(ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+}
+
+static const struct dmub_replay_funcs replay_funcs = {
+	.replay_copy_settings				= dmub_replay_copy_settings,
+	.replay_enable					= dmub_replay_enable,
+	.replay_get_state				= dmub_replay_get_state,
+	.replay_set_power_opt				= dmub_replay_set_power_opt,
+	.replay_set_coasting_vtotal			= dmub_replay_set_coasting_vtotal,
+	.replay_residency				= dmub_replay_residency,
+	.replay_set_power_opt_and_coasting_vtotal	= dmub_replay_set_power_opt_and_coasting_vtotal,
+	.replay_send_cmd				= dmub_replay_send_cmd,
+};
+
+/*
+ * Construct Replay object.
+ */
+static void dmub_replay_construct(struct dmub_replay *replay, struct dc_context *ctx)
+{
+	replay->ctx = ctx;
+	replay->funcs = &replay_funcs;
+}
+
+/*
+ * Allocate and initialize Replay object.
+ */
+struct dmub_replay *dmub_replay_create(struct dc_context *ctx)
+{
+	struct dmub_replay *replay = kzalloc(sizeof(struct dmub_replay), GFP_KERNEL);
+
+	if (replay == NULL) {
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+
+	dmub_replay_construct(replay, ctx);
+
+	return replay;
+}
+
+/*
+ * Deallocate Replay object.
+ */
+void dmub_replay_destroy(struct dmub_replay **dmub)
+{
+	kfree(*dmub);
+	*dmub = NULL;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dmub_replay.h b/drivers/gpu/drm/amd/display/dc/dce/dmub_replay.h
new file mode 100644
index 000000000000..e6346c0ffc0e
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce/dmub_replay.h
@@ -0,0 +1,41 @@
+// SPDX-License-Identifier: MIT
+//
+// Copyright 2024 Advanced Micro Devices, Inc.
+
+#ifndef _DMUB_REPLAY_H_
+#define _DMUB_REPLAY_H_
+
+#include "dc_types.h"
+#include "dmub_cmd.h"
+struct dc_link;
+struct dmub_replay_funcs;
+
+struct dmub_replay {
+	struct dc_context *ctx;
+	const struct dmub_replay_funcs *funcs;
+};
+
+struct dmub_replay_funcs {
+	void (*replay_get_state)(struct dmub_replay *dmub, enum replay_state *state,
+		uint8_t panel_inst);
+	void (*replay_enable)(struct dmub_replay *dmub, bool enable, bool wait,
+		uint8_t panel_inst);
+	bool (*replay_copy_settings)(struct dmub_replay *dmub, struct dc_link *link,
+		struct replay_context *replay_context, uint8_t panel_inst);
+	void (*replay_set_power_opt)(struct dmub_replay *dmub, unsigned int power_opt,
+		uint8_t panel_inst);
+	void (*replay_send_cmd)(struct dmub_replay *dmub,
+		enum replay_FW_Message_type msg, union dmub_replay_cmd_set *cmd_element);
+	void (*replay_set_coasting_vtotal)(struct dmub_replay *dmub, uint32_t coasting_vtotal,
+		uint8_t panel_inst);
+	void (*replay_residency)(struct dmub_replay *dmub,
+		uint8_t panel_inst, uint32_t *residency, const bool is_start, const enum pr_residency_mode mode);
+	void (*replay_set_power_opt_and_coasting_vtotal)(struct dmub_replay *dmub,
+		unsigned int power_opt, uint8_t panel_inst, uint32_t coasting_vtotal);
+};
+
+struct dmub_replay *dmub_replay_create(struct dc_context *ctx);
+void dmub_replay_destroy(struct dmub_replay **dmub);
+
+
+#endif /* _DMUB_REPLAY_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/Makefile b/drivers/gpu/drm/amd/display/dc/dce110/Makefile
new file mode 100644
index 000000000000..c307f040e48f
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce110/Makefile
@@ -0,0 +1,35 @@
+#
+# Copyright 2017 Advanced Micro Devices, Inc.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+#
+# Makefile for the 'controller' sub-component of DAL.
+# It provides the control and status of HW CRTC block.
+
+CFLAGS_$(AMDDALPATH)/dc/dce110/dce110_resource.o = -Wno-override-init
+
+DCE110 = dce110_timing_generator.o \
+dce110_compressor.o dce110_opp_regamma_v.o \
+dce110_opp_csc_v.o dce110_timing_generator_v.o \
+dce110_mem_input_v.o dce110_opp_v.o dce110_transform_v.o
+
+AMD_DAL_DCE110 = $(addprefix $(AMDDALPATH)/dc/dce110/,$(DCE110))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCE110)
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_compressor.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_compressor.c
new file mode 100644
index 000000000000..59a0961b49da
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_compressor.c
@@ -0,0 +1,458 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+
+#include "dce/dce_11_0_d.h"
+#include "dce/dce_11_0_sh_mask.h"
+#include "gmc/gmc_8_2_sh_mask.h"
+#include "gmc/gmc_8_2_d.h"
+
+#include "include/logger_interface.h"
+
+#include "dce110_compressor.h"
+
+#define DC_LOGGER \
+		cp110->base.ctx->logger
+#define DCP_REG(reg)\
+	(reg + cp110->offsets.dcp_offset)
+#define DMIF_REG(reg)\
+	(reg + cp110->offsets.dmif_offset)
+
+static const struct dce110_compressor_reg_offsets reg_offsets[] = {
+{
+	.dcp_offset = (mmDCP0_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
+	.dmif_offset =
+		(mmDMIF_PG0_DPG_PIPE_DPM_CONTROL
+			- mmDMIF_PG0_DPG_PIPE_DPM_CONTROL),
+},
+{
+	.dcp_offset = (mmDCP1_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
+	.dmif_offset =
+		(mmDMIF_PG1_DPG_PIPE_DPM_CONTROL
+			- mmDMIF_PG0_DPG_PIPE_DPM_CONTROL),
+},
+{
+	.dcp_offset = (mmDCP2_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
+	.dmif_offset =
+		(mmDMIF_PG2_DPG_PIPE_DPM_CONTROL
+			- mmDMIF_PG0_DPG_PIPE_DPM_CONTROL),
+}
+};
+
+static uint32_t align_to_chunks_number_per_line(uint32_t pixels)
+{
+	return 256 * ((pixels + 255) / 256);
+}
+
+static void reset_lb_on_vblank(struct compressor *compressor, uint32_t crtc_inst)
+{
+	uint32_t value;
+	uint32_t frame_count;
+	uint32_t status_pos;
+	uint32_t retry = 0;
+	struct dce110_compressor *cp110 = TO_DCE110_COMPRESSOR(compressor);
+
+	cp110->offsets = reg_offsets[crtc_inst];
+
+	status_pos = dm_read_reg(compressor->ctx, DCP_REG(mmCRTC_STATUS_POSITION));
+
+
+	/* Only if CRTC is enabled and counter is moving we wait for one frame. */
+	if (status_pos != dm_read_reg(compressor->ctx, DCP_REG(mmCRTC_STATUS_POSITION))) {
+		/* Resetting LB on VBlank */
+		value = dm_read_reg(compressor->ctx, DCP_REG(mmLB_SYNC_RESET_SEL));
+		set_reg_field_value(value, 3, LB_SYNC_RESET_SEL, LB_SYNC_RESET_SEL);
+		set_reg_field_value(value, 1, LB_SYNC_RESET_SEL, LB_SYNC_RESET_SEL2);
+		dm_write_reg(compressor->ctx, DCP_REG(mmLB_SYNC_RESET_SEL), value);
+
+		frame_count = dm_read_reg(compressor->ctx, DCP_REG(mmCRTC_STATUS_FRAME_COUNT));
+
+
+		for (retry = 10000; retry > 0; retry--) {
+			if (frame_count != dm_read_reg(compressor->ctx, DCP_REG(mmCRTC_STATUS_FRAME_COUNT)))
+				break;
+			udelay(10);
+		}
+		if (!retry)
+			dm_error("Frame count did not increase for 100ms.\n");
+
+		/* Resetting LB on VBlank */
+		value = dm_read_reg(compressor->ctx, DCP_REG(mmLB_SYNC_RESET_SEL));
+		set_reg_field_value(value, 2, LB_SYNC_RESET_SEL, LB_SYNC_RESET_SEL);
+		set_reg_field_value(value, 0, LB_SYNC_RESET_SEL, LB_SYNC_RESET_SEL2);
+		dm_write_reg(compressor->ctx, DCP_REG(mmLB_SYNC_RESET_SEL), value);
+	}
+}
+
+static void wait_for_fbc_state_changed(
+	struct dce110_compressor *cp110,
+	bool enabled)
+{
+	uint32_t counter = 0;
+	uint32_t addr = mmFBC_STATUS;
+	uint32_t value;
+
+	while (counter < 1000) {
+		value = dm_read_reg(cp110->base.ctx, addr);
+		if (get_reg_field_value(
+			value,
+			FBC_STATUS,
+			FBC_ENABLE_STATUS) == enabled)
+			break;
+		udelay(100);
+		counter++;
+	}
+
+	if (counter == 1000) {
+		DC_LOG_WARNING("%s: wait counter exceeded, changes to HW not applied",
+			__func__);
+	} else {
+		DC_LOG_SYNC("FBC status changed to %d", enabled);
+	}
+
+
+}
+
+void dce110_compressor_power_up_fbc(struct compressor *compressor)
+{
+	uint32_t value;
+	uint32_t addr;
+
+	addr = mmFBC_CNTL;
+	value = dm_read_reg(compressor->ctx, addr);
+	set_reg_field_value(value, 0, FBC_CNTL, FBC_GRPH_COMP_EN);
+	set_reg_field_value(value, 1, FBC_CNTL, FBC_EN);
+	set_reg_field_value(value, 2, FBC_CNTL, FBC_COHERENCY_MODE);
+	if (compressor->options.bits.CLK_GATING_DISABLED == 1) {
+		/* HW needs to do power measurement comparison. */
+		set_reg_field_value(
+			value,
+			0,
+			FBC_CNTL,
+			FBC_COMP_CLK_GATE_EN);
+	}
+	dm_write_reg(compressor->ctx, addr, value);
+
+	addr = mmFBC_COMP_MODE;
+	value = dm_read_reg(compressor->ctx, addr);
+	set_reg_field_value(value, 1, FBC_COMP_MODE, FBC_RLE_EN);
+	set_reg_field_value(value, 1, FBC_COMP_MODE, FBC_DPCM4_RGB_EN);
+	set_reg_field_value(value, 1, FBC_COMP_MODE, FBC_IND_EN);
+	dm_write_reg(compressor->ctx, addr, value);
+
+	addr = mmFBC_COMP_CNTL;
+	value = dm_read_reg(compressor->ctx, addr);
+	set_reg_field_value(value, 1, FBC_COMP_CNTL, FBC_DEPTH_RGB08_EN);
+	dm_write_reg(compressor->ctx, addr, value);
+	/*FBC_MIN_COMPRESSION 0 ==> 2:1 */
+	/*                    1 ==> 4:1 */
+	/*                    2 ==> 8:1 */
+	/*                  0xF ==> 1:1 */
+	set_reg_field_value(value, 0xF, FBC_COMP_CNTL, FBC_MIN_COMPRESSION);
+	dm_write_reg(compressor->ctx, addr, value);
+	compressor->min_compress_ratio = FBC_COMPRESS_RATIO_1TO1;
+
+	value = 0;
+	dm_write_reg(compressor->ctx, mmFBC_IND_LUT0, value);
+
+	value = 0xFFFFFF;
+	dm_write_reg(compressor->ctx, mmFBC_IND_LUT1, value);
+}
+
+void dce110_compressor_enable_fbc(
+	struct compressor *compressor,
+	struct compr_addr_and_pitch_params *params)
+{
+	struct dce110_compressor *cp110 = TO_DCE110_COMPRESSOR(compressor);
+
+	if (compressor->options.bits.FBC_SUPPORT &&
+		(!dce110_compressor_is_fbc_enabled_in_hw(compressor, NULL))) {
+
+		uint32_t addr;
+		uint32_t value, misc_value;
+
+		addr = mmFBC_CNTL;
+		value = dm_read_reg(compressor->ctx, addr);
+		set_reg_field_value(value, 1, FBC_CNTL, FBC_GRPH_COMP_EN);
+		/* params->inst is valid HW CRTC instance start from 0 */
+		set_reg_field_value(
+			value,
+			params->inst,
+			FBC_CNTL, FBC_SRC_SEL);
+		dm_write_reg(compressor->ctx, addr, value);
+
+		/* Keep track of enum controller_id FBC is attached to */
+		compressor->is_enabled = true;
+		/* attached_inst is SW CRTC instance start from 1
+		 * 0 = CONTROLLER_ID_UNDEFINED means not attached crtc
+		 */
+		compressor->attached_inst = params->inst + CONTROLLER_ID_D0;
+
+		/* Toggle it as there is bug in HW */
+		set_reg_field_value(value, 0, FBC_CNTL, FBC_GRPH_COMP_EN);
+		dm_write_reg(compressor->ctx, addr, value);
+
+		/* FBC usage with scatter & gather for dce110 */
+		misc_value = dm_read_reg(compressor->ctx, mmFBC_MISC);
+
+		set_reg_field_value(misc_value, 1,
+				FBC_MISC, FBC_INVALIDATE_ON_ERROR);
+		set_reg_field_value(misc_value, 1,
+				FBC_MISC, FBC_DECOMPRESS_ERROR_CLEAR);
+		set_reg_field_value(misc_value, 0x14,
+				FBC_MISC, FBC_SLOW_REQ_INTERVAL);
+
+		dm_write_reg(compressor->ctx, mmFBC_MISC, misc_value);
+
+		/* Enable FBC */
+		set_reg_field_value(value, 1, FBC_CNTL, FBC_GRPH_COMP_EN);
+		dm_write_reg(compressor->ctx, addr, value);
+
+		wait_for_fbc_state_changed(cp110, true);
+	}
+}
+
+void dce110_compressor_disable_fbc(struct compressor *compressor)
+{
+	struct dce110_compressor *cp110 = TO_DCE110_COMPRESSOR(compressor);
+	uint32_t crtc_inst = 0;
+
+	if (compressor->options.bits.FBC_SUPPORT) {
+		if (dce110_compressor_is_fbc_enabled_in_hw(compressor, &crtc_inst)) {
+			uint32_t reg_data;
+			/* Turn off compression */
+			reg_data = dm_read_reg(compressor->ctx, mmFBC_CNTL);
+			set_reg_field_value(reg_data, 0, FBC_CNTL, FBC_GRPH_COMP_EN);
+			dm_write_reg(compressor->ctx, mmFBC_CNTL, reg_data);
+
+			/* Reset enum controller_id to undefined */
+			compressor->attached_inst = 0;
+			compressor->is_enabled = false;
+
+			wait_for_fbc_state_changed(cp110, false);
+		}
+
+		/* Sync line buffer which fbc was attached to dce100/110 only */
+		if (crtc_inst > CONTROLLER_ID_UNDEFINED && crtc_inst < CONTROLLER_ID_D3)
+			reset_lb_on_vblank(compressor,
+					crtc_inst - CONTROLLER_ID_D0);
+	}
+}
+
+bool dce110_compressor_is_fbc_enabled_in_hw(
+	struct compressor *compressor,
+	uint32_t *inst)
+{
+	/* Check the hardware register */
+	uint32_t value;
+
+	value = dm_read_reg(compressor->ctx, mmFBC_STATUS);
+	if (get_reg_field_value(value, FBC_STATUS, FBC_ENABLE_STATUS)) {
+		if (inst != NULL)
+			*inst = compressor->attached_inst;
+		return true;
+	}
+
+	value = dm_read_reg(compressor->ctx, mmFBC_MISC);
+	if (get_reg_field_value(value, FBC_MISC, FBC_STOP_ON_HFLIP_EVENT)) {
+		value = dm_read_reg(compressor->ctx, mmFBC_CNTL);
+
+		if (get_reg_field_value(value, FBC_CNTL, FBC_GRPH_COMP_EN)) {
+			if (inst != NULL)
+				*inst =
+					compressor->attached_inst;
+			return true;
+		}
+	}
+	return false;
+}
+
+
+void dce110_compressor_program_compressed_surface_address_and_pitch(
+	struct compressor *compressor,
+	struct compr_addr_and_pitch_params *params)
+{
+	struct dce110_compressor *cp110 = TO_DCE110_COMPRESSOR(compressor);
+	uint32_t value = 0;
+	uint32_t fbc_pitch = 0;
+	uint32_t compressed_surf_address_low_part =
+		compressor->compr_surface_address.addr.low_part;
+
+	cp110->offsets = reg_offsets[params->inst];
+
+	/* Clear content first. */
+	dm_write_reg(
+		compressor->ctx,
+		DCP_REG(mmGRPH_COMPRESS_SURFACE_ADDRESS_HIGH),
+		0);
+	dm_write_reg(compressor->ctx,
+		DCP_REG(mmGRPH_COMPRESS_SURFACE_ADDRESS), 0);
+
+	/* Write address, HIGH has to be first. */
+	dm_write_reg(compressor->ctx,
+		DCP_REG(mmGRPH_COMPRESS_SURFACE_ADDRESS_HIGH),
+		compressor->compr_surface_address.addr.high_part);
+	dm_write_reg(compressor->ctx,
+		DCP_REG(mmGRPH_COMPRESS_SURFACE_ADDRESS),
+		compressed_surf_address_low_part);
+
+	fbc_pitch = align_to_chunks_number_per_line(params->source_view_width);
+
+	if (compressor->min_compress_ratio == FBC_COMPRESS_RATIO_1TO1)
+		fbc_pitch = fbc_pitch / 8;
+	else
+		DC_LOG_WARNING("%s: Unexpected DCE11 compression ratio",
+			__func__);
+
+	/* Clear content first. */
+	dm_write_reg(compressor->ctx, DCP_REG(mmGRPH_COMPRESS_PITCH), 0);
+
+	/* Write FBC Pitch. */
+	set_reg_field_value(
+		value,
+		fbc_pitch,
+		GRPH_COMPRESS_PITCH,
+		GRPH_COMPRESS_PITCH);
+	dm_write_reg(compressor->ctx, DCP_REG(mmGRPH_COMPRESS_PITCH), value);
+
+}
+
+void dce110_compressor_set_fbc_invalidation_triggers(
+	struct compressor *compressor,
+	uint32_t fbc_trigger)
+{
+	/* Disable region hit event, FBC_MEMORY_REGION_MASK = 0 (bits 16-19)
+	 * for DCE 11 regions cannot be used - does not work with S/G
+	 */
+	uint32_t addr = mmFBC_CLIENT_REGION_MASK;
+	uint32_t value = dm_read_reg(compressor->ctx, addr);
+
+	set_reg_field_value(
+		value,
+		0,
+		FBC_CLIENT_REGION_MASK,
+		FBC_MEMORY_REGION_MASK);
+	dm_write_reg(compressor->ctx, addr, value);
+
+	/* Setup events when to clear all CSM entries (effectively marking
+	 * current compressed data invalid)
+	 * For DCE 11 CSM metadata 11111 means - "Not Compressed"
+	 * Used as the initial value of the metadata sent to the compressor
+	 * after invalidation, to indicate that the compressor should attempt
+	 * to compress all chunks on the current pass.  Also used when the chunk
+	 * is not successfully written to memory.
+	 * When this CSM value is detected, FBC reads from the uncompressed
+	 * buffer. Set events according to passed in value, these events are
+	 * valid for DCE11:
+	 *     - bit  0 - display register updated
+	 *     - bit 28 - memory write from any client except from MCIF
+	 *     - bit 29 - CG static screen signal is inactive
+	 * In addition, DCE11.1 also needs to set new DCE11.1 specific events
+	 * that are used to trigger invalidation on certain register changes,
+	 * for example enabling of Alpha Compression may trigger invalidation of
+	 * FBC once bit is set. These events are as follows:
+	 *      - Bit 2 - FBC_GRPH_COMP_EN register updated
+	 *      - Bit 3 - FBC_SRC_SEL register updated
+	 *      - Bit 4 - FBC_MIN_COMPRESSION register updated
+	 *      - Bit 5 - FBC_ALPHA_COMP_EN register updated
+	 *      - Bit 6 - FBC_ZERO_ALPHA_CHUNK_SKIP_EN register updated
+	 *      - Bit 7 - FBC_FORCE_COPY_TO_COMP_BUF register updated
+	 */
+	addr = mmFBC_IDLE_FORCE_CLEAR_MASK;
+	value = dm_read_reg(compressor->ctx, addr);
+	set_reg_field_value(
+		value,
+		fbc_trigger,
+		FBC_IDLE_FORCE_CLEAR_MASK,
+		FBC_IDLE_FORCE_CLEAR_MASK);
+	dm_write_reg(compressor->ctx, addr, value);
+}
+
+struct compressor *dce110_compressor_create(struct dc_context *ctx)
+{
+	struct dce110_compressor *cp110 =
+		kzalloc(sizeof(struct dce110_compressor), GFP_KERNEL);
+
+	if (!cp110)
+		return NULL;
+
+	dce110_compressor_construct(cp110, ctx);
+	return &cp110->base;
+}
+
+void dce110_compressor_destroy(struct compressor **compressor)
+{
+	kfree(TO_DCE110_COMPRESSOR(*compressor));
+	*compressor = NULL;
+}
+
+static const struct compressor_funcs dce110_compressor_funcs = {
+	.power_up_fbc = dce110_compressor_power_up_fbc,
+	.enable_fbc = dce110_compressor_enable_fbc,
+	.disable_fbc = dce110_compressor_disable_fbc,
+	.set_fbc_invalidation_triggers = dce110_compressor_set_fbc_invalidation_triggers,
+	.surface_address_and_pitch = dce110_compressor_program_compressed_surface_address_and_pitch,
+	.is_fbc_enabled_in_hw = dce110_compressor_is_fbc_enabled_in_hw
+};
+
+
+void dce110_compressor_construct(struct dce110_compressor *compressor,
+	struct dc_context *ctx)
+{
+
+	compressor->base.options.raw = 0;
+	compressor->base.options.bits.FBC_SUPPORT = true;
+
+	/* for dce 11 always use one dram channel for lpt */
+	compressor->base.lpt_channels_num = 1;
+	compressor->base.options.bits.DUMMY_BACKEND = false;
+
+	/*
+	 * check if this system has more than 1 dram channel; if only 1 then lpt
+	 * should not be supported
+	 */
+
+
+	compressor->base.options.bits.CLK_GATING_DISABLED = false;
+
+	compressor->base.ctx = ctx;
+	compressor->base.embedded_panel_h_size = 0;
+	compressor->base.embedded_panel_v_size = 0;
+	compressor->base.memory_bus_width = ctx->asic_id.vram_width;
+	compressor->base.allocated_size = 0;
+	compressor->base.preferred_requested_size = 0;
+	compressor->base.min_compress_ratio = FBC_COMPRESS_RATIO_INVALID;
+	compressor->base.banks_num = 0;
+	compressor->base.raw_size = 0;
+	compressor->base.channel_interleave_size = 0;
+	compressor->base.dram_channels_num = 0;
+	compressor->base.lpt_channels_num = 0;
+	compressor->base.attached_inst = CONTROLLER_ID_UNDEFINED;
+	compressor->base.is_enabled = false;
+	compressor->base.funcs = &dce110_compressor_funcs;
+
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_compressor.h b/drivers/gpu/drm/amd/display/dc/dce110/dce110_compressor.h
new file mode 100644
index 000000000000..223c57941e92
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_compressor.h
@@ -0,0 +1,79 @@
+/* Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_COMPRESSOR_DCE110_H__
+#define __DC_COMPRESSOR_DCE110_H__
+
+#include "../inc/compressor.h"
+
+#define TO_DCE110_COMPRESSOR(compressor)\
+	container_of(compressor, struct dce110_compressor, base)
+
+struct dce110_compressor_reg_offsets {
+	uint32_t dcp_offset;
+	uint32_t dmif_offset;
+};
+
+struct dce110_compressor {
+	struct compressor base;
+	struct dce110_compressor_reg_offsets offsets;
+};
+
+struct compressor *dce110_compressor_create(struct dc_context *ctx);
+
+void dce110_compressor_construct(struct dce110_compressor *cp110,
+	struct dc_context *ctx);
+
+void dce110_compressor_destroy(struct compressor **cp);
+
+/* FBC RELATED */
+void dce110_compressor_power_up_fbc(struct compressor *cp);
+
+void dce110_compressor_enable_fbc(struct compressor *cp,
+	struct compr_addr_and_pitch_params *params);
+
+void dce110_compressor_disable_fbc(struct compressor *cp);
+
+void dce110_compressor_set_fbc_invalidation_triggers(struct compressor *cp,
+	uint32_t fbc_trigger);
+
+void dce110_compressor_program_compressed_surface_address_and_pitch(
+	struct compressor *cp,
+	struct compr_addr_and_pitch_params *params);
+
+bool dce110_compressor_is_fbc_enabled_in_hw(struct compressor *cp,
+	uint32_t *fbc_mapped_crtc_id);
+
+/* LPT RELATED */
+void dce110_compressor_enable_lpt(struct compressor *cp);
+
+void dce110_compressor_disable_lpt(struct compressor *cp);
+
+void dce110_compressor_program_lpt_control(struct compressor *cp,
+	struct compr_addr_and_pitch_params *params);
+
+bool dce110_compressor_is_lpt_enabled_in_hw(struct compressor *cp);
+
+#endif
+
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_mem_input_v.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_mem_input_v.c
new file mode 100644
index 000000000000..2c43c2422638
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_mem_input_v.c
@@ -0,0 +1,1041 @@
+/*
+ * Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+#include "dm_services.h"
+
+#include "dce/dce_11_0_d.h"
+#include "dce/dce_11_0_sh_mask.h"
+/* TODO: this needs to be looked at, used by Stella's workaround*/
+#include "gmc/gmc_8_2_d.h"
+#include "gmc/gmc_8_2_sh_mask.h"
+
+#include "include/logger_interface.h"
+#include "inc/dce_calcs.h"
+
+#include "dce/dce_mem_input.h"
+#include "dce110_mem_input_v.h"
+
+static void set_flip_control(
+	struct dce_mem_input *mem_input110,
+	bool immediate)
+{
+	uint32_t value = 0;
+
+	value = dm_read_reg(
+			mem_input110->base.ctx,
+			mmUNP_FLIP_CONTROL);
+
+	set_reg_field_value(value, 1,
+			UNP_FLIP_CONTROL,
+			GRPH_SURFACE_UPDATE_PENDING_MODE);
+
+	dm_write_reg(
+			mem_input110->base.ctx,
+			mmUNP_FLIP_CONTROL,
+			value);
+}
+
+/* chroma part */
+static void program_pri_addr_c(
+	struct dce_mem_input *mem_input110,
+	PHYSICAL_ADDRESS_LOC address)
+{
+	uint32_t value = 0;
+	uint32_t temp = 0;
+	/*high register MUST be programmed first*/
+	temp = address.high_part &
+UNP_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH_C__GRPH_PRIMARY_SURFACE_ADDRESS_HIGH_C_MASK;
+
+	set_reg_field_value(value, temp,
+		UNP_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH_C,
+		GRPH_PRIMARY_SURFACE_ADDRESS_HIGH_C);
+
+	dm_write_reg(
+		mem_input110->base.ctx,
+		mmUNP_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH_C,
+		value);
+
+	value = 0;
+	temp = address.low_part >>
+	UNP_GRPH_PRIMARY_SURFACE_ADDRESS_C__GRPH_PRIMARY_SURFACE_ADDRESS_C__SHIFT;
+
+	set_reg_field_value(value, temp,
+		UNP_GRPH_PRIMARY_SURFACE_ADDRESS_C,
+		GRPH_PRIMARY_SURFACE_ADDRESS_C);
+
+	dm_write_reg(
+		mem_input110->base.ctx,
+		mmUNP_GRPH_PRIMARY_SURFACE_ADDRESS_C,
+		value);
+}
+
+/* luma part */
+static void program_pri_addr_l(
+	struct dce_mem_input *mem_input110,
+	PHYSICAL_ADDRESS_LOC address)
+{
+	uint32_t value = 0;
+	uint32_t temp = 0;
+
+	/*high register MUST be programmed first*/
+	temp = address.high_part &
+UNP_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH_L__GRPH_PRIMARY_SURFACE_ADDRESS_HIGH_L_MASK;
+
+	set_reg_field_value(value, temp,
+		UNP_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH_L,
+		GRPH_PRIMARY_SURFACE_ADDRESS_HIGH_L);
+
+	dm_write_reg(
+		mem_input110->base.ctx,
+		mmUNP_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH_L,
+		value);
+
+	value = 0;
+	temp = address.low_part >>
+	UNP_GRPH_PRIMARY_SURFACE_ADDRESS_L__GRPH_PRIMARY_SURFACE_ADDRESS_L__SHIFT;
+
+	set_reg_field_value(value, temp,
+		UNP_GRPH_PRIMARY_SURFACE_ADDRESS_L,
+		GRPH_PRIMARY_SURFACE_ADDRESS_L);
+
+	dm_write_reg(
+		mem_input110->base.ctx,
+		mmUNP_GRPH_PRIMARY_SURFACE_ADDRESS_L,
+		value);
+}
+
+static void program_addr(
+	struct dce_mem_input *mem_input110,
+	const struct dc_plane_address *addr)
+{
+	switch (addr->type) {
+	case PLN_ADDR_TYPE_GRAPHICS:
+		program_pri_addr_l(
+			mem_input110,
+			addr->grph.addr);
+		break;
+	case PLN_ADDR_TYPE_VIDEO_PROGRESSIVE:
+		program_pri_addr_c(
+			mem_input110,
+			addr->video_progressive.chroma_addr);
+		program_pri_addr_l(
+			mem_input110,
+			addr->video_progressive.luma_addr);
+		break;
+	default:
+		/* not supported */
+		BREAK_TO_DEBUGGER();
+	}
+}
+
+static void enable(struct dce_mem_input *mem_input110)
+{
+	uint32_t value = 0;
+
+	value = dm_read_reg(mem_input110->base.ctx, mmUNP_GRPH_ENABLE);
+	set_reg_field_value(value, 1, UNP_GRPH_ENABLE, GRPH_ENABLE);
+	dm_write_reg(mem_input110->base.ctx,
+		mmUNP_GRPH_ENABLE,
+		value);
+}
+
+static void program_tiling(
+	struct dce_mem_input *mem_input110,
+	const struct dc_tiling_info *info,
+	const enum surface_pixel_format pixel_format)
+{
+	uint32_t value = 0;
+
+	set_reg_field_value(value, info->gfx8.num_banks,
+		UNP_GRPH_CONTROL, GRPH_NUM_BANKS);
+
+	set_reg_field_value(value, info->gfx8.bank_width,
+		UNP_GRPH_CONTROL, GRPH_BANK_WIDTH_L);
+
+	set_reg_field_value(value, info->gfx8.bank_height,
+		UNP_GRPH_CONTROL, GRPH_BANK_HEIGHT_L);
+
+	set_reg_field_value(value, info->gfx8.tile_aspect,
+		UNP_GRPH_CONTROL, GRPH_MACRO_TILE_ASPECT_L);
+
+	set_reg_field_value(value, info->gfx8.tile_split,
+		UNP_GRPH_CONTROL, GRPH_TILE_SPLIT_L);
+
+	set_reg_field_value(value, info->gfx8.tile_mode,
+		UNP_GRPH_CONTROL, GRPH_MICRO_TILE_MODE_L);
+
+	set_reg_field_value(value, info->gfx8.pipe_config,
+		UNP_GRPH_CONTROL, GRPH_PIPE_CONFIG);
+
+	set_reg_field_value(value, info->gfx8.array_mode,
+		UNP_GRPH_CONTROL, GRPH_ARRAY_MODE);
+
+	set_reg_field_value(value, 1,
+		UNP_GRPH_CONTROL, GRPH_COLOR_EXPANSION_MODE);
+
+	set_reg_field_value(value, 0,
+		UNP_GRPH_CONTROL, GRPH_Z);
+
+	dm_write_reg(
+		mem_input110->base.ctx,
+		mmUNP_GRPH_CONTROL,
+		value);
+
+	value = 0;
+
+	set_reg_field_value(value, info->gfx8.bank_width_c,
+		UNP_GRPH_CONTROL_C, GRPH_BANK_WIDTH_C);
+
+	set_reg_field_value(value, info->gfx8.bank_height_c,
+		UNP_GRPH_CONTROL_C, GRPH_BANK_HEIGHT_C);
+
+	set_reg_field_value(value, info->gfx8.tile_aspect_c,
+		UNP_GRPH_CONTROL_C, GRPH_MACRO_TILE_ASPECT_C);
+
+	set_reg_field_value(value, info->gfx8.tile_split_c,
+		UNP_GRPH_CONTROL_C, GRPH_TILE_SPLIT_C);
+
+	set_reg_field_value(value, info->gfx8.tile_mode_c,
+		UNP_GRPH_CONTROL_C, GRPH_MICRO_TILE_MODE_C);
+
+	dm_write_reg(
+		mem_input110->base.ctx,
+		mmUNP_GRPH_CONTROL_C,
+		value);
+}
+
+static void program_size_and_rotation(
+	struct dce_mem_input *mem_input110,
+	enum dc_rotation_angle rotation,
+	const struct plane_size *plane_size)
+{
+	uint32_t value = 0;
+	struct plane_size local_size = *plane_size;
+
+	if (rotation == ROTATION_ANGLE_90 ||
+		rotation == ROTATION_ANGLE_270) {
+
+		swap(local_size.surface_size.x,
+		     local_size.surface_size.y);
+		swap(local_size.surface_size.width,
+		     local_size.surface_size.height);
+		swap(local_size.chroma_size.x,
+		     local_size.chroma_size.y);
+		swap(local_size.chroma_size.width,
+		     local_size.chroma_size.height);
+	}
+
+	value = 0;
+	set_reg_field_value(value, local_size.surface_pitch,
+			UNP_GRPH_PITCH_L, GRPH_PITCH_L);
+
+	dm_write_reg(
+		mem_input110->base.ctx,
+		mmUNP_GRPH_PITCH_L,
+		value);
+
+	value = 0;
+	set_reg_field_value(value, local_size.chroma_pitch,
+			UNP_GRPH_PITCH_C, GRPH_PITCH_C);
+	dm_write_reg(
+		mem_input110->base.ctx,
+		mmUNP_GRPH_PITCH_C,
+		value);
+
+	value = 0;
+	set_reg_field_value(value, 0,
+			UNP_GRPH_X_START_L, GRPH_X_START_L);
+	dm_write_reg(
+		mem_input110->base.ctx,
+		mmUNP_GRPH_X_START_L,
+		value);
+
+	value = 0;
+	set_reg_field_value(value, 0,
+			UNP_GRPH_X_START_C, GRPH_X_START_C);
+	dm_write_reg(
+		mem_input110->base.ctx,
+		mmUNP_GRPH_X_START_C,
+		value);
+
+	value = 0;
+	set_reg_field_value(value, 0,
+			UNP_GRPH_Y_START_L, GRPH_Y_START_L);
+	dm_write_reg(
+		mem_input110->base.ctx,
+		mmUNP_GRPH_Y_START_L,
+		value);
+
+	value = 0;
+	set_reg_field_value(value, 0,
+			UNP_GRPH_Y_START_C, GRPH_Y_START_C);
+	dm_write_reg(
+		mem_input110->base.ctx,
+		mmUNP_GRPH_Y_START_C,
+		value);
+
+	value = 0;
+	set_reg_field_value(value, local_size.surface_size.x +
+			local_size.surface_size.width,
+			UNP_GRPH_X_END_L, GRPH_X_END_L);
+	dm_write_reg(
+		mem_input110->base.ctx,
+		mmUNP_GRPH_X_END_L,
+		value);
+
+	value = 0;
+	set_reg_field_value(value, local_size.chroma_size.x +
+			local_size.chroma_size.width,
+			UNP_GRPH_X_END_C, GRPH_X_END_C);
+	dm_write_reg(
+		mem_input110->base.ctx,
+		mmUNP_GRPH_X_END_C,
+		value);
+
+	value = 0;
+	set_reg_field_value(value, local_size.surface_size.y +
+			local_size.surface_size.height,
+			UNP_GRPH_Y_END_L, GRPH_Y_END_L);
+	dm_write_reg(
+		mem_input110->base.ctx,
+		mmUNP_GRPH_Y_END_L,
+		value);
+
+	value = 0;
+	set_reg_field_value(value, local_size.chroma_size.y +
+			local_size.chroma_size.height,
+			UNP_GRPH_Y_END_C, GRPH_Y_END_C);
+	dm_write_reg(
+		mem_input110->base.ctx,
+		mmUNP_GRPH_Y_END_C,
+		value);
+
+	value = 0;
+	switch (rotation) {
+	case ROTATION_ANGLE_90:
+		set_reg_field_value(value, 3,
+			UNP_HW_ROTATION, ROTATION_ANGLE);
+		break;
+	case ROTATION_ANGLE_180:
+		set_reg_field_value(value, 2,
+			UNP_HW_ROTATION, ROTATION_ANGLE);
+		break;
+	case ROTATION_ANGLE_270:
+		set_reg_field_value(value, 1,
+			UNP_HW_ROTATION, ROTATION_ANGLE);
+		break;
+	default:
+		set_reg_field_value(value, 0,
+			UNP_HW_ROTATION, ROTATION_ANGLE);
+		break;
+	}
+
+	dm_write_reg(
+		mem_input110->base.ctx,
+		mmUNP_HW_ROTATION,
+		value);
+}
+
+static void program_pixel_format(
+	struct dce_mem_input *mem_input110,
+	enum surface_pixel_format format)
+{
+	if (format < SURFACE_PIXEL_FORMAT_VIDEO_BEGIN) {
+		uint32_t value;
+		uint8_t grph_depth;
+		uint8_t grph_format;
+
+		value =	dm_read_reg(
+				mem_input110->base.ctx,
+				mmUNP_GRPH_CONTROL);
+
+		switch (format) {
+		case SURFACE_PIXEL_FORMAT_GRPH_PALETA_256_COLORS:
+			grph_depth = 0;
+			grph_format = 0;
+			break;
+		case SURFACE_PIXEL_FORMAT_GRPH_RGB565:
+			grph_depth = 1;
+			grph_format = 1;
+			break;
+		case SURFACE_PIXEL_FORMAT_GRPH_ARGB8888:
+		case SURFACE_PIXEL_FORMAT_GRPH_ABGR8888:
+			grph_depth = 2;
+			grph_format = 0;
+			break;
+		case SURFACE_PIXEL_FORMAT_GRPH_ARGB2101010:
+		case SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010:
+		case SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010_XR_BIAS:
+			grph_depth = 2;
+			grph_format = 1;
+			break;
+		case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616:
+		case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616:
+		case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F:
+		case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616F:
+			grph_depth = 3;
+			grph_format = 0;
+			break;
+		default:
+			grph_depth = 2;
+			grph_format = 0;
+			break;
+		}
+
+		set_reg_field_value(
+				value,
+				grph_depth,
+				UNP_GRPH_CONTROL,
+				GRPH_DEPTH);
+		set_reg_field_value(
+				value,
+				grph_format,
+				UNP_GRPH_CONTROL,
+				GRPH_FORMAT);
+
+		dm_write_reg(
+				mem_input110->base.ctx,
+				mmUNP_GRPH_CONTROL,
+				value);
+
+		value =	dm_read_reg(
+				mem_input110->base.ctx,
+				mmUNP_GRPH_CONTROL_EXP);
+
+		/* VIDEO FORMAT 0 */
+		set_reg_field_value(
+				value,
+				0,
+				UNP_GRPH_CONTROL_EXP,
+				VIDEO_FORMAT);
+		dm_write_reg(
+				mem_input110->base.ctx,
+				mmUNP_GRPH_CONTROL_EXP,
+				value);
+
+	} else {
+		/* Video 422 and 420 needs UNP_GRPH_CONTROL_EXP programmed */
+		uint32_t value;
+		uint8_t video_format;
+
+		value =	dm_read_reg(
+				mem_input110->base.ctx,
+				mmUNP_GRPH_CONTROL_EXP);
+
+		switch (format) {
+		case SURFACE_PIXEL_FORMAT_VIDEO_420_YCbCr:
+			video_format = 2;
+			break;
+		case SURFACE_PIXEL_FORMAT_VIDEO_420_YCrCb:
+			video_format = 3;
+			break;
+		default:
+			video_format = 0;
+			break;
+		}
+
+		set_reg_field_value(
+			value,
+			video_format,
+			UNP_GRPH_CONTROL_EXP,
+			VIDEO_FORMAT);
+
+		dm_write_reg(
+			mem_input110->base.ctx,
+			mmUNP_GRPH_CONTROL_EXP,
+			value);
+	}
+}
+
+static bool dce_mem_input_v_is_surface_pending(struct mem_input *mem_input)
+{
+	struct dce_mem_input *mem_input110 = TO_DCE_MEM_INPUT(mem_input);
+	uint32_t value;
+
+	value = dm_read_reg(mem_input110->base.ctx, mmUNP_GRPH_UPDATE);
+
+	if (get_reg_field_value(value, UNP_GRPH_UPDATE,
+			GRPH_SURFACE_UPDATE_PENDING))
+		return true;
+
+	mem_input->current_address = mem_input->request_address;
+	return false;
+}
+
+static bool dce_mem_input_v_program_surface_flip_and_addr(
+	struct mem_input *mem_input,
+	const struct dc_plane_address *address,
+	bool flip_immediate)
+{
+	struct dce_mem_input *mem_input110 = TO_DCE_MEM_INPUT(mem_input);
+
+	set_flip_control(mem_input110, flip_immediate);
+	program_addr(mem_input110,
+		address);
+
+	mem_input->request_address = *address;
+
+	return true;
+}
+
+/* Scatter Gather param tables */
+static const unsigned int dvmm_Hw_Setting_2DTiling[4][9] = {
+		{  8, 64, 64,  8,  8, 1, 4, 0, 0},
+		{ 16, 64, 32,  8, 16, 1, 8, 0, 0},
+		{ 32, 32, 32, 16, 16, 1, 8, 0, 0},
+		{ 64,  8, 32, 16, 16, 1, 8, 0, 0}, /* fake */
+};
+
+static const unsigned int dvmm_Hw_Setting_1DTiling[4][9] = {
+		{  8, 512, 8, 1, 0, 1, 0, 0, 0},  /* 0 for invalid */
+		{ 16, 256, 8, 2, 0, 1, 0, 0, 0},
+		{ 32, 128, 8, 4, 0, 1, 0, 0, 0},
+		{ 64,  64, 8, 4, 0, 1, 0, 0, 0}, /* fake */
+};
+
+static const unsigned int dvmm_Hw_Setting_Linear[4][9] = {
+		{  8, 4096, 1, 8, 0, 1, 0, 0, 0},
+		{ 16, 2048, 1, 8, 0, 1, 0, 0, 0},
+		{ 32, 1024, 1, 8, 0, 1, 0, 0, 0},
+		{ 64,  512, 1, 8, 0, 1, 0, 0, 0}, /* new for 64bpp from HW */
+};
+
+/* Helper to get table entry from surface info */
+static const unsigned int *get_dvmm_hw_setting(
+		struct dc_tiling_info *tiling_info,
+		enum surface_pixel_format format,
+		bool chroma)
+{
+	enum bits_per_pixel {
+		bpp_8 = 0,
+		bpp_16,
+		bpp_32,
+		bpp_64
+	} bpp;
+
+	if (format >= SURFACE_PIXEL_FORMAT_INVALID)
+		bpp = bpp_32;
+	else if (format >= SURFACE_PIXEL_FORMAT_VIDEO_BEGIN)
+		bpp = chroma ? bpp_16 : bpp_8;
+	else
+		bpp = bpp_8;
+
+	switch (tiling_info->gfx8.array_mode) {
+	case DC_ARRAY_1D_TILED_THIN1:
+	case DC_ARRAY_1D_TILED_THICK:
+	case DC_ARRAY_PRT_TILED_THIN1:
+		return dvmm_Hw_Setting_1DTiling[bpp];
+	case DC_ARRAY_2D_TILED_THIN1:
+	case DC_ARRAY_2D_TILED_THICK:
+	case DC_ARRAY_2D_TILED_X_THICK:
+	case DC_ARRAY_PRT_2D_TILED_THIN1:
+	case DC_ARRAY_PRT_2D_TILED_THICK:
+		return dvmm_Hw_Setting_2DTiling[bpp];
+	case DC_ARRAY_LINEAR_GENERAL:
+	case DC_ARRAY_LINEAR_ALLIGNED:
+		return dvmm_Hw_Setting_Linear[bpp];
+	default:
+		return dvmm_Hw_Setting_2DTiling[bpp];
+	}
+}
+
+static void dce_mem_input_v_program_pte_vm(
+		struct mem_input *mem_input,
+		enum surface_pixel_format format,
+		struct dc_tiling_info *tiling_info,
+		enum dc_rotation_angle rotation)
+{
+	struct dce_mem_input *mem_input110 = TO_DCE_MEM_INPUT(mem_input);
+	const unsigned int *pte = get_dvmm_hw_setting(tiling_info, format, false);
+	const unsigned int *pte_chroma = get_dvmm_hw_setting(tiling_info, format, true);
+
+	unsigned int page_width = 0;
+	unsigned int page_height = 0;
+	unsigned int page_width_chroma = 0;
+	unsigned int page_height_chroma = 0;
+	unsigned int temp_page_width = pte[1];
+	unsigned int temp_page_height = pte[2];
+	unsigned int min_pte_before_flip = 0;
+	unsigned int min_pte_before_flip_chroma = 0;
+	uint32_t value = 0;
+
+	while ((temp_page_width >>= 1) != 0)
+		page_width++;
+	while ((temp_page_height >>= 1) != 0)
+		page_height++;
+
+	temp_page_width = pte_chroma[1];
+	temp_page_height = pte_chroma[2];
+	while ((temp_page_width >>= 1) != 0)
+		page_width_chroma++;
+	while ((temp_page_height >>= 1) != 0)
+		page_height_chroma++;
+
+	switch (rotation) {
+	case ROTATION_ANGLE_90:
+	case ROTATION_ANGLE_270:
+		min_pte_before_flip = pte[4];
+		min_pte_before_flip_chroma = pte_chroma[4];
+		break;
+	default:
+		min_pte_before_flip = pte[3];
+		min_pte_before_flip_chroma = pte_chroma[3];
+		break;
+	}
+
+	value = dm_read_reg(mem_input110->base.ctx, mmUNP_PIPE_OUTSTANDING_REQUEST_LIMIT);
+	/* TODO: un-hardcode requestlimit */
+	set_reg_field_value(value, 0xff, UNP_PIPE_OUTSTANDING_REQUEST_LIMIT, UNP_PIPE_OUTSTANDING_REQUEST_LIMIT_L);
+	set_reg_field_value(value, 0xff, UNP_PIPE_OUTSTANDING_REQUEST_LIMIT, UNP_PIPE_OUTSTANDING_REQUEST_LIMIT_C);
+	dm_write_reg(mem_input110->base.ctx, mmUNP_PIPE_OUTSTANDING_REQUEST_LIMIT, value);
+
+	value = dm_read_reg(mem_input110->base.ctx, mmUNP_DVMM_PTE_CONTROL);
+	set_reg_field_value(value, page_width, UNP_DVMM_PTE_CONTROL, DVMM_PAGE_WIDTH);
+	set_reg_field_value(value, page_height, UNP_DVMM_PTE_CONTROL, DVMM_PAGE_HEIGHT);
+	set_reg_field_value(value, min_pte_before_flip, UNP_DVMM_PTE_CONTROL, DVMM_MIN_PTE_BEFORE_FLIP);
+	dm_write_reg(mem_input110->base.ctx, mmUNP_DVMM_PTE_CONTROL, value);
+
+	value = dm_read_reg(mem_input110->base.ctx, mmUNP_DVMM_PTE_ARB_CONTROL);
+	set_reg_field_value(value, pte[5], UNP_DVMM_PTE_ARB_CONTROL, DVMM_PTE_REQ_PER_CHUNK);
+	set_reg_field_value(value, 0xff, UNP_DVMM_PTE_ARB_CONTROL, DVMM_MAX_PTE_REQ_OUTSTANDING);
+	dm_write_reg(mem_input110->base.ctx, mmUNP_DVMM_PTE_ARB_CONTROL, value);
+
+	value = dm_read_reg(mem_input110->base.ctx, mmUNP_DVMM_PTE_CONTROL_C);
+	set_reg_field_value(value, page_width_chroma, UNP_DVMM_PTE_CONTROL_C, DVMM_PAGE_WIDTH_C);
+	set_reg_field_value(value, page_height_chroma, UNP_DVMM_PTE_CONTROL_C, DVMM_PAGE_HEIGHT_C);
+	set_reg_field_value(value, min_pte_before_flip_chroma, UNP_DVMM_PTE_CONTROL_C, DVMM_MIN_PTE_BEFORE_FLIP_C);
+	dm_write_reg(mem_input110->base.ctx, mmUNP_DVMM_PTE_CONTROL_C, value);
+
+	value = dm_read_reg(mem_input110->base.ctx, mmUNP_DVMM_PTE_ARB_CONTROL_C);
+	set_reg_field_value(value, pte_chroma[5], UNP_DVMM_PTE_ARB_CONTROL_C, DVMM_PTE_REQ_PER_CHUNK_C);
+	set_reg_field_value(value, 0xff, UNP_DVMM_PTE_ARB_CONTROL_C, DVMM_MAX_PTE_REQ_OUTSTANDING_C);
+	dm_write_reg(mem_input110->base.ctx, mmUNP_DVMM_PTE_ARB_CONTROL_C, value);
+}
+
+static void dce_mem_input_v_program_surface_config(
+	struct mem_input *mem_input,
+	enum surface_pixel_format format,
+	struct dc_tiling_info *tiling_info,
+	struct plane_size *plane_size,
+	enum dc_rotation_angle rotation,
+	struct dc_plane_dcc_param *dcc,
+	bool horizotal_mirror)
+{
+	struct dce_mem_input *mem_input110 = TO_DCE_MEM_INPUT(mem_input);
+
+	enable(mem_input110);
+	program_tiling(mem_input110, tiling_info, format);
+	program_size_and_rotation(mem_input110, rotation, plane_size);
+	program_pixel_format(mem_input110, format);
+}
+
+static void program_urgency_watermark(
+	const struct dc_context *ctx,
+	const uint32_t urgency_addr,
+	const uint32_t wm_addr,
+	struct dce_watermarks marks_low,
+	uint32_t total_dest_line_time_ns)
+{
+	/* register value */
+	uint32_t urgency_cntl = 0;
+	uint32_t wm_mask_cntl = 0;
+
+	/*Write mask to enable reading/writing of watermark set A*/
+	wm_mask_cntl = dm_read_reg(ctx, wm_addr);
+	set_reg_field_value(wm_mask_cntl,
+			1,
+			DPGV0_WATERMARK_MASK_CONTROL,
+			URGENCY_WATERMARK_MASK);
+	dm_write_reg(ctx, wm_addr, wm_mask_cntl);
+
+	urgency_cntl = dm_read_reg(ctx, urgency_addr);
+
+	set_reg_field_value(
+		urgency_cntl,
+		marks_low.a_mark,
+		DPGV0_PIPE_URGENCY_CONTROL,
+		URGENCY_LOW_WATERMARK);
+
+	set_reg_field_value(
+		urgency_cntl,
+		total_dest_line_time_ns,
+		DPGV0_PIPE_URGENCY_CONTROL,
+		URGENCY_HIGH_WATERMARK);
+	dm_write_reg(ctx, urgency_addr, urgency_cntl);
+
+	/*Write mask to enable reading/writing of watermark set B*/
+	wm_mask_cntl = dm_read_reg(ctx, wm_addr);
+	set_reg_field_value(wm_mask_cntl,
+			2,
+			DPGV0_WATERMARK_MASK_CONTROL,
+			URGENCY_WATERMARK_MASK);
+	dm_write_reg(ctx, wm_addr, wm_mask_cntl);
+
+	urgency_cntl = dm_read_reg(ctx, urgency_addr);
+
+	set_reg_field_value(urgency_cntl,
+		marks_low.b_mark,
+		DPGV0_PIPE_URGENCY_CONTROL,
+		URGENCY_LOW_WATERMARK);
+
+	set_reg_field_value(urgency_cntl,
+		total_dest_line_time_ns,
+		DPGV0_PIPE_URGENCY_CONTROL,
+		URGENCY_HIGH_WATERMARK);
+
+	dm_write_reg(ctx, urgency_addr, urgency_cntl);
+}
+
+static void program_urgency_watermark_l(
+	const struct dc_context *ctx,
+	struct dce_watermarks marks_low,
+	uint32_t total_dest_line_time_ns)
+{
+	program_urgency_watermark(
+		ctx,
+		mmDPGV0_PIPE_URGENCY_CONTROL,
+		mmDPGV0_WATERMARK_MASK_CONTROL,
+		marks_low,
+		total_dest_line_time_ns);
+}
+
+static void program_urgency_watermark_c(
+	const struct dc_context *ctx,
+	struct dce_watermarks marks_low,
+	uint32_t total_dest_line_time_ns)
+{
+	program_urgency_watermark(
+		ctx,
+		mmDPGV1_PIPE_URGENCY_CONTROL,
+		mmDPGV1_WATERMARK_MASK_CONTROL,
+		marks_low,
+		total_dest_line_time_ns);
+}
+
+static void program_stutter_watermark(
+	const struct dc_context *ctx,
+	const uint32_t stutter_addr,
+	const uint32_t wm_addr,
+	struct dce_watermarks marks)
+{
+	/* register value */
+	uint32_t stutter_cntl = 0;
+	uint32_t wm_mask_cntl = 0;
+
+	/*Write mask to enable reading/writing of watermark set A*/
+
+	wm_mask_cntl = dm_read_reg(ctx, wm_addr);
+	set_reg_field_value(wm_mask_cntl,
+		1,
+		DPGV0_WATERMARK_MASK_CONTROL,
+		STUTTER_EXIT_SELF_REFRESH_WATERMARK_MASK);
+	dm_write_reg(ctx, wm_addr, wm_mask_cntl);
+
+	stutter_cntl = dm_read_reg(ctx, stutter_addr);
+
+	if (ctx->dc->debug.disable_stutter) {
+		set_reg_field_value(stutter_cntl,
+			0,
+			DPGV0_PIPE_STUTTER_CONTROL,
+			STUTTER_ENABLE);
+	} else {
+		set_reg_field_value(stutter_cntl,
+			1,
+			DPGV0_PIPE_STUTTER_CONTROL,
+			STUTTER_ENABLE);
+	}
+
+	set_reg_field_value(stutter_cntl,
+		1,
+		DPGV0_PIPE_STUTTER_CONTROL,
+		STUTTER_IGNORE_FBC);
+
+	/*Write watermark set A*/
+	set_reg_field_value(stutter_cntl,
+		marks.a_mark,
+		DPGV0_PIPE_STUTTER_CONTROL,
+		STUTTER_EXIT_SELF_REFRESH_WATERMARK);
+	dm_write_reg(ctx, stutter_addr, stutter_cntl);
+
+	/*Write mask to enable reading/writing of watermark set B*/
+	wm_mask_cntl = dm_read_reg(ctx, wm_addr);
+	set_reg_field_value(wm_mask_cntl,
+		2,
+		DPGV0_WATERMARK_MASK_CONTROL,
+		STUTTER_EXIT_SELF_REFRESH_WATERMARK_MASK);
+	dm_write_reg(ctx, wm_addr, wm_mask_cntl);
+
+	stutter_cntl = dm_read_reg(ctx, stutter_addr);
+	/*Write watermark set B*/
+	set_reg_field_value(stutter_cntl,
+		marks.b_mark,
+		DPGV0_PIPE_STUTTER_CONTROL,
+		STUTTER_EXIT_SELF_REFRESH_WATERMARK);
+	dm_write_reg(ctx, stutter_addr, stutter_cntl);
+}
+
+static void program_stutter_watermark_l(
+	const struct dc_context *ctx,
+	struct dce_watermarks marks)
+{
+	program_stutter_watermark(ctx,
+			mmDPGV0_PIPE_STUTTER_CONTROL,
+			mmDPGV0_WATERMARK_MASK_CONTROL,
+			marks);
+}
+
+static void program_stutter_watermark_c(
+	const struct dc_context *ctx,
+	struct dce_watermarks marks)
+{
+	program_stutter_watermark(ctx,
+			mmDPGV1_PIPE_STUTTER_CONTROL,
+			mmDPGV1_WATERMARK_MASK_CONTROL,
+			marks);
+}
+
+static void program_nbp_watermark(
+	const struct dc_context *ctx,
+	const uint32_t wm_mask_ctrl_addr,
+	const uint32_t nbp_pstate_ctrl_addr,
+	struct dce_watermarks marks)
+{
+	uint32_t value;
+
+	/* Write mask to enable reading/writing of watermark set A */
+
+	value = dm_read_reg(ctx, wm_mask_ctrl_addr);
+
+	set_reg_field_value(
+		value,
+		1,
+		DPGV0_WATERMARK_MASK_CONTROL,
+		NB_PSTATE_CHANGE_WATERMARK_MASK);
+	dm_write_reg(ctx, wm_mask_ctrl_addr, value);
+
+	value = dm_read_reg(ctx, nbp_pstate_ctrl_addr);
+
+	set_reg_field_value(
+		value,
+		1,
+		DPGV0_PIPE_NB_PSTATE_CHANGE_CONTROL,
+		NB_PSTATE_CHANGE_ENABLE);
+	set_reg_field_value(
+		value,
+		1,
+		DPGV0_PIPE_NB_PSTATE_CHANGE_CONTROL,
+		NB_PSTATE_CHANGE_URGENT_DURING_REQUEST);
+	set_reg_field_value(
+		value,
+		1,
+		DPGV0_PIPE_NB_PSTATE_CHANGE_CONTROL,
+		NB_PSTATE_CHANGE_NOT_SELF_REFRESH_DURING_REQUEST);
+	dm_write_reg(ctx, nbp_pstate_ctrl_addr, value);
+
+	/* Write watermark set A */
+	value = dm_read_reg(ctx, nbp_pstate_ctrl_addr);
+	set_reg_field_value(
+		value,
+		marks.a_mark,
+		DPGV0_PIPE_NB_PSTATE_CHANGE_CONTROL,
+		NB_PSTATE_CHANGE_WATERMARK);
+	dm_write_reg(ctx, nbp_pstate_ctrl_addr, value);
+
+	/* Write mask to enable reading/writing of watermark set B */
+	value = dm_read_reg(ctx, wm_mask_ctrl_addr);
+	set_reg_field_value(
+		value,
+		2,
+		DPGV0_WATERMARK_MASK_CONTROL,
+		NB_PSTATE_CHANGE_WATERMARK_MASK);
+	dm_write_reg(ctx, wm_mask_ctrl_addr, value);
+
+	value = dm_read_reg(ctx, nbp_pstate_ctrl_addr);
+	set_reg_field_value(
+		value,
+		1,
+		DPGV0_PIPE_NB_PSTATE_CHANGE_CONTROL,
+		NB_PSTATE_CHANGE_ENABLE);
+	set_reg_field_value(
+		value,
+		1,
+		DPGV0_PIPE_NB_PSTATE_CHANGE_CONTROL,
+		NB_PSTATE_CHANGE_URGENT_DURING_REQUEST);
+	set_reg_field_value(
+		value,
+		1,
+		DPGV0_PIPE_NB_PSTATE_CHANGE_CONTROL,
+		NB_PSTATE_CHANGE_NOT_SELF_REFRESH_DURING_REQUEST);
+	dm_write_reg(ctx, nbp_pstate_ctrl_addr, value);
+
+	/* Write watermark set B */
+	value = dm_read_reg(ctx, nbp_pstate_ctrl_addr);
+	set_reg_field_value(
+		value,
+		marks.b_mark,
+		DPGV0_PIPE_NB_PSTATE_CHANGE_CONTROL,
+		NB_PSTATE_CHANGE_WATERMARK);
+	dm_write_reg(ctx, nbp_pstate_ctrl_addr, value);
+}
+
+static void program_nbp_watermark_l(
+	const struct dc_context *ctx,
+	struct dce_watermarks marks)
+{
+	program_nbp_watermark(ctx,
+			mmDPGV0_WATERMARK_MASK_CONTROL,
+			mmDPGV0_PIPE_NB_PSTATE_CHANGE_CONTROL,
+			marks);
+}
+
+static void program_nbp_watermark_c(
+	const struct dc_context *ctx,
+	struct dce_watermarks marks)
+{
+	program_nbp_watermark(ctx,
+			mmDPGV1_WATERMARK_MASK_CONTROL,
+			mmDPGV1_PIPE_NB_PSTATE_CHANGE_CONTROL,
+			marks);
+}
+
+static void dce_mem_input_v_program_display_marks(
+	struct mem_input *mem_input,
+	struct dce_watermarks nbp,
+	struct dce_watermarks stutter,
+	struct dce_watermarks stutter_enter,
+	struct dce_watermarks urgent,
+	uint32_t total_dest_line_time_ns)
+{
+	program_urgency_watermark_l(
+		mem_input->ctx,
+		urgent,
+		total_dest_line_time_ns);
+
+	program_nbp_watermark_l(
+		mem_input->ctx,
+		nbp);
+
+	program_stutter_watermark_l(
+		mem_input->ctx,
+		stutter);
+
+}
+
+static void dce_mem_input_program_chroma_display_marks(
+	struct mem_input *mem_input,
+	struct dce_watermarks nbp,
+	struct dce_watermarks stutter,
+	struct dce_watermarks urgent,
+	uint32_t total_dest_line_time_ns)
+{
+	program_urgency_watermark_c(
+		mem_input->ctx,
+		urgent,
+		total_dest_line_time_ns);
+
+	program_nbp_watermark_c(
+		mem_input->ctx,
+		nbp);
+
+	program_stutter_watermark_c(
+		mem_input->ctx,
+		stutter);
+}
+
+static void dce110_allocate_mem_input_v(
+	struct mem_input *mi,
+	uint32_t h_total,/* for current stream */
+	uint32_t v_total,/* for current stream */
+	uint32_t pix_clk_khz,/* for current stream */
+	uint32_t total_stream_num)
+{
+	uint32_t addr;
+	uint32_t value;
+	uint32_t pix_dur;
+	if (pix_clk_khz != 0) {
+		addr = mmDPGV0_PIPE_ARBITRATION_CONTROL1;
+		value = dm_read_reg(mi->ctx, addr);
+		pix_dur = 1000000000ULL / pix_clk_khz;
+		set_reg_field_value(
+			value,
+			pix_dur,
+			DPGV0_PIPE_ARBITRATION_CONTROL1,
+			PIXEL_DURATION);
+		dm_write_reg(mi->ctx, addr, value);
+
+		addr = mmDPGV1_PIPE_ARBITRATION_CONTROL1;
+		value = dm_read_reg(mi->ctx, addr);
+		pix_dur = 1000000000ULL / pix_clk_khz;
+		set_reg_field_value(
+			value,
+			pix_dur,
+			DPGV1_PIPE_ARBITRATION_CONTROL1,
+			PIXEL_DURATION);
+		dm_write_reg(mi->ctx, addr, value);
+
+		addr = mmDPGV0_PIPE_ARBITRATION_CONTROL2;
+		value = 0x4000800;
+		dm_write_reg(mi->ctx, addr, value);
+
+		addr = mmDPGV1_PIPE_ARBITRATION_CONTROL2;
+		value = 0x4000800;
+		dm_write_reg(mi->ctx, addr, value);
+	}
+
+}
+
+static void dce110_free_mem_input_v(
+	struct mem_input *mi,
+	uint32_t total_stream_num)
+{
+}
+
+static const struct mem_input_funcs dce110_mem_input_v_funcs = {
+	.mem_input_program_display_marks =
+			dce_mem_input_v_program_display_marks,
+	.mem_input_program_chroma_display_marks =
+			dce_mem_input_program_chroma_display_marks,
+	.allocate_mem_input = dce110_allocate_mem_input_v,
+	.free_mem_input = dce110_free_mem_input_v,
+	.mem_input_program_surface_flip_and_addr =
+			dce_mem_input_v_program_surface_flip_and_addr,
+	.mem_input_program_pte_vm =
+			dce_mem_input_v_program_pte_vm,
+	.mem_input_program_surface_config =
+			dce_mem_input_v_program_surface_config,
+	.mem_input_is_flip_pending =
+			dce_mem_input_v_is_surface_pending
+};
+/*****************************************/
+/* Constructor, Destructor               */
+/*****************************************/
+
+void dce110_mem_input_v_construct(
+	struct dce_mem_input *dce_mi,
+	struct dc_context *ctx)
+{
+	dce_mi->base.funcs = &dce110_mem_input_v_funcs;
+	dce_mi->base.ctx = ctx;
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_mem_input_v.h b/drivers/gpu/drm/amd/display/dc/dce110/dce110_mem_input_v.h
new file mode 100644
index 000000000000..f01d4a607fea
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_mem_input_v.h
@@ -0,0 +1,35 @@
+/* Copyright 2012-16 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_MEM_INPUT_V_DCE110_H__
+#define __DC_MEM_INPUT_V_DCE110_H__
+
+#include "mem_input.h"
+#include "dce/dce_mem_input.h"
+
+void dce110_mem_input_v_construct(
+	struct dce_mem_input *dce_mi,
+	struct dc_context *ctx);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_opp_csc_v.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_opp_csc_v.c
new file mode 100644
index 000000000000..e096d2b95ef9
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_opp_csc_v.c
@@ -0,0 +1,737 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+#include "dce110_transform_v.h"
+#include "basics/conversion.h"
+
+/* include DCE11 register header files */
+#include "dce/dce_11_0_d.h"
+#include "dce/dce_11_0_sh_mask.h"
+#include "dce/dce_11_0_enum.h"
+
+enum {
+	OUTPUT_CSC_MATRIX_SIZE = 12
+};
+
+/* constrast:0 - 2.0, default 1.0 */
+#define UNDERLAY_CONTRAST_DEFAULT 100
+#define UNDERLAY_CONTRAST_MAX     200
+#define UNDERLAY_CONTRAST_MIN       0
+#define UNDERLAY_CONTRAST_STEP      1
+#define UNDERLAY_CONTRAST_DIVIDER 100
+
+/* Saturation: 0 - 2.0; default 1.0 */
+#define UNDERLAY_SATURATION_DEFAULT   100 /*1.00*/
+#define UNDERLAY_SATURATION_MIN         0
+#define UNDERLAY_SATURATION_MAX       200 /* 2.00 */
+#define UNDERLAY_SATURATION_STEP        1 /* 0.01 */
+/*actual max overlay saturation
+ * value = UNDERLAY_SATURATION_MAX /UNDERLAY_SATURATION_DIVIDER
+ */
+
+/* Hue */
+#define  UNDERLAY_HUE_DEFAULT      0
+#define  UNDERLAY_HUE_MIN       -300
+#define  UNDERLAY_HUE_MAX        300
+#define  UNDERLAY_HUE_STEP         5
+#define  UNDERLAY_HUE_DIVIDER   10 /* HW range: -30 ~ +30 */
+#define UNDERLAY_SATURATION_DIVIDER   100
+
+/* Brightness: in DAL usually -.25 ~ .25.
+ * In MMD is -100 to +100 in 16-235 range; which when scaled to full range is
+ *  ~-116 to +116. When normalized this is about 0.4566.
+ * With 100 divider this becomes 46, but we may use another for better precision
+ * The ideal one is 100/219 ((100/255)*(255/219)),
+ * i.e. min/max = +-100, divider = 219
+ * default 0.0
+ */
+#define  UNDERLAY_BRIGHTNESS_DEFAULT    0
+#define  UNDERLAY_BRIGHTNESS_MIN      -46 /* ~116/255 */
+#define  UNDERLAY_BRIGHTNESS_MAX       46
+#define  UNDERLAY_BRIGHTNESS_STEP       1 /*  .01 */
+#define  UNDERLAY_BRIGHTNESS_DIVIDER  100
+
+static const struct out_csc_color_matrix global_color_matrix[] = {
+{ COLOR_SPACE_SRGB,
+	{ 0x2000, 0, 0, 0, 0, 0x2000, 0, 0, 0, 0, 0x2000, 0} },
+{ COLOR_SPACE_SRGB_LIMITED,
+	{ 0x1B60, 0, 0, 0x200, 0, 0x1B60, 0, 0x200, 0, 0, 0x1B60, 0x200} },
+{ COLOR_SPACE_YCBCR601,
+	{ 0xE00, 0xF447, 0xFDB9, 0x1000, 0x82F, 0x1012, 0x31F, 0x200, 0xFB47,
+		0xF6B9, 0xE00, 0x1000} },
+{ COLOR_SPACE_YCBCR709, { 0xE00, 0xF349, 0xFEB7, 0x1000, 0x5D2, 0x1394, 0x1FA,
+	0x200, 0xFCCB, 0xF535, 0xE00, 0x1000} },
+/* TODO: correct values below */
+{ COLOR_SPACE_YCBCR601_LIMITED, { 0xE00, 0xF447, 0xFDB9, 0x1000, 0x991,
+	0x12C9, 0x3A6, 0x200, 0xFB47, 0xF6B9, 0xE00, 0x1000} },
+{ COLOR_SPACE_YCBCR709_LIMITED, { 0xE00, 0xF349, 0xFEB7, 0x1000, 0x6CE, 0x16E3,
+	0x24F, 0x200, 0xFCCB, 0xF535, 0xE00, 0x1000} }
+};
+
+enum csc_color_mode {
+	/* 00 - BITS2:0 Bypass */
+	CSC_COLOR_MODE_GRAPHICS_BYPASS,
+	/* 01 - hard coded coefficient TV RGB */
+	CSC_COLOR_MODE_GRAPHICS_PREDEFINED,
+	/* 04 - programmable OUTPUT CSC coefficient */
+	CSC_COLOR_MODE_GRAPHICS_OUTPUT_CSC,
+};
+
+enum grph_color_adjust_option {
+	GRPH_COLOR_MATRIX_HW_DEFAULT = 1,
+	GRPH_COLOR_MATRIX_SW
+};
+
+static void program_color_matrix_v(
+	struct dce_transform *xfm_dce,
+	const struct out_csc_color_matrix *tbl_entry,
+	enum grph_color_adjust_option options)
+{
+	struct dc_context *ctx = xfm_dce->base.ctx;
+	uint32_t cntl_value = dm_read_reg(ctx, mmCOL_MAN_OUTPUT_CSC_CONTROL);
+	bool use_set_a = (get_reg_field_value(cntl_value,
+			COL_MAN_OUTPUT_CSC_CONTROL,
+			OUTPUT_CSC_MODE) != 4);
+
+	set_reg_field_value(
+			cntl_value,
+		0,
+		COL_MAN_OUTPUT_CSC_CONTROL,
+		OUTPUT_CSC_MODE);
+
+	if (use_set_a) {
+		{
+			uint32_t value = 0;
+			uint32_t addr = mmOUTPUT_CSC_C11_C12_A;
+			/* fixed S2.13 format */
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[0],
+				OUTPUT_CSC_C11_C12_A,
+				OUTPUT_CSC_C11_A);
+
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[1],
+				OUTPUT_CSC_C11_C12_A,
+				OUTPUT_CSC_C12_A);
+
+			dm_write_reg(ctx, addr, value);
+		}
+		{
+			uint32_t value = 0;
+			uint32_t addr = mmOUTPUT_CSC_C13_C14_A;
+			/* fixed S2.13 format */
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[2],
+				OUTPUT_CSC_C13_C14_A,
+				OUTPUT_CSC_C13_A);
+			/* fixed S0.13 format */
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[3],
+				OUTPUT_CSC_C13_C14_A,
+				OUTPUT_CSC_C14_A);
+
+			dm_write_reg(ctx, addr, value);
+		}
+		{
+			uint32_t value = 0;
+			uint32_t addr = mmOUTPUT_CSC_C21_C22_A;
+			/* fixed S2.13 format */
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[4],
+				OUTPUT_CSC_C21_C22_A,
+				OUTPUT_CSC_C21_A);
+			/* fixed S2.13 format */
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[5],
+				OUTPUT_CSC_C21_C22_A,
+				OUTPUT_CSC_C22_A);
+
+			dm_write_reg(ctx, addr, value);
+		}
+		{
+			uint32_t value = 0;
+			uint32_t addr = mmOUTPUT_CSC_C23_C24_A;
+			/* fixed S2.13 format */
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[6],
+				OUTPUT_CSC_C23_C24_A,
+				OUTPUT_CSC_C23_A);
+			/* fixed S0.13 format */
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[7],
+				OUTPUT_CSC_C23_C24_A,
+				OUTPUT_CSC_C24_A);
+
+			dm_write_reg(ctx, addr, value);
+		}
+		{
+			uint32_t value = 0;
+			uint32_t addr = mmOUTPUT_CSC_C31_C32_A;
+			/* fixed S2.13 format */
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[8],
+				OUTPUT_CSC_C31_C32_A,
+				OUTPUT_CSC_C31_A);
+			/* fixed S0.13 format */
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[9],
+				OUTPUT_CSC_C31_C32_A,
+				OUTPUT_CSC_C32_A);
+
+			dm_write_reg(ctx, addr, value);
+		}
+		{
+			uint32_t value = 0;
+			uint32_t addr = mmOUTPUT_CSC_C33_C34_A;
+			/* fixed S2.13 format */
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[10],
+				OUTPUT_CSC_C33_C34_A,
+				OUTPUT_CSC_C33_A);
+			/* fixed S0.13 format */
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[11],
+				OUTPUT_CSC_C33_C34_A,
+				OUTPUT_CSC_C34_A);
+
+			dm_write_reg(ctx, addr, value);
+		}
+		set_reg_field_value(
+			cntl_value,
+			4,
+			COL_MAN_OUTPUT_CSC_CONTROL,
+			OUTPUT_CSC_MODE);
+	} else {
+		{
+			uint32_t value = 0;
+			uint32_t addr = mmOUTPUT_CSC_C11_C12_B;
+			/* fixed S2.13 format */
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[0],
+				OUTPUT_CSC_C11_C12_B,
+				OUTPUT_CSC_C11_B);
+
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[1],
+				OUTPUT_CSC_C11_C12_B,
+				OUTPUT_CSC_C12_B);
+
+			dm_write_reg(ctx, addr, value);
+		}
+		{
+			uint32_t value = 0;
+			uint32_t addr = mmOUTPUT_CSC_C13_C14_B;
+			/* fixed S2.13 format */
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[2],
+				OUTPUT_CSC_C13_C14_B,
+				OUTPUT_CSC_C13_B);
+			/* fixed S0.13 format */
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[3],
+				OUTPUT_CSC_C13_C14_B,
+				OUTPUT_CSC_C14_B);
+
+			dm_write_reg(ctx, addr, value);
+		}
+		{
+			uint32_t value = 0;
+			uint32_t addr = mmOUTPUT_CSC_C21_C22_B;
+			/* fixed S2.13 format */
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[4],
+				OUTPUT_CSC_C21_C22_B,
+				OUTPUT_CSC_C21_B);
+			/* fixed S2.13 format */
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[5],
+				OUTPUT_CSC_C21_C22_B,
+				OUTPUT_CSC_C22_B);
+
+			dm_write_reg(ctx, addr, value);
+		}
+		{
+			uint32_t value = 0;
+			uint32_t addr = mmOUTPUT_CSC_C23_C24_B;
+			/* fixed S2.13 format */
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[6],
+				OUTPUT_CSC_C23_C24_B,
+				OUTPUT_CSC_C23_B);
+			/* fixed S0.13 format */
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[7],
+				OUTPUT_CSC_C23_C24_B,
+				OUTPUT_CSC_C24_B);
+
+			dm_write_reg(ctx, addr, value);
+		}
+		{
+			uint32_t value = 0;
+			uint32_t addr = mmOUTPUT_CSC_C31_C32_B;
+			/* fixed S2.13 format */
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[8],
+				OUTPUT_CSC_C31_C32_B,
+				OUTPUT_CSC_C31_B);
+			/* fixed S0.13 format */
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[9],
+				OUTPUT_CSC_C31_C32_B,
+				OUTPUT_CSC_C32_B);
+
+			dm_write_reg(ctx, addr, value);
+		}
+		{
+			uint32_t value = 0;
+			uint32_t addr = mmOUTPUT_CSC_C33_C34_B;
+			/* fixed S2.13 format */
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[10],
+				OUTPUT_CSC_C33_C34_B,
+				OUTPUT_CSC_C33_B);
+			/* fixed S0.13 format */
+			set_reg_field_value(
+				value,
+				tbl_entry->regval[11],
+				OUTPUT_CSC_C33_C34_B,
+				OUTPUT_CSC_C34_B);
+
+			dm_write_reg(ctx, addr, value);
+		}
+		set_reg_field_value(
+			cntl_value,
+			5,
+			COL_MAN_OUTPUT_CSC_CONTROL,
+			OUTPUT_CSC_MODE);
+	}
+
+	dm_write_reg(ctx, mmCOL_MAN_OUTPUT_CSC_CONTROL, cntl_value);
+}
+
+static bool configure_graphics_mode_v(
+	struct dce_transform *xfm_dce,
+	enum csc_color_mode config,
+	enum graphics_csc_adjust_type csc_adjust_type,
+	enum dc_color_space color_space)
+{
+	struct dc_context *ctx = xfm_dce->base.ctx;
+	uint32_t addr = mmCOL_MAN_OUTPUT_CSC_CONTROL;
+	uint32_t value = dm_read_reg(ctx, addr);
+
+	set_reg_field_value(
+		value,
+		0,
+		COL_MAN_OUTPUT_CSC_CONTROL,
+		OUTPUT_CSC_MODE);
+
+	if (csc_adjust_type == GRAPHICS_CSC_ADJUST_TYPE_SW) {
+		if (config == CSC_COLOR_MODE_GRAPHICS_OUTPUT_CSC)
+			return true;
+
+		switch (color_space) {
+		case COLOR_SPACE_SRGB:
+			/* by pass */
+			set_reg_field_value(
+				value,
+				0,
+				COL_MAN_OUTPUT_CSC_CONTROL,
+				OUTPUT_CSC_MODE);
+			break;
+		case COLOR_SPACE_SRGB_LIMITED:
+			/* not supported for underlay on CZ */
+			return false;
+
+		case COLOR_SPACE_YCBCR601_LIMITED:
+			/* YCbCr601 */
+			set_reg_field_value(
+				value,
+				2,
+				COL_MAN_OUTPUT_CSC_CONTROL,
+				OUTPUT_CSC_MODE);
+			break;
+		case COLOR_SPACE_YCBCR709:
+		case COLOR_SPACE_YCBCR709_LIMITED:
+			/* YCbCr709 */
+			set_reg_field_value(
+				value,
+				3,
+				COL_MAN_OUTPUT_CSC_CONTROL,
+				OUTPUT_CSC_MODE);
+			break;
+		default:
+			return false;
+		}
+
+	} else if (csc_adjust_type == GRAPHICS_CSC_ADJUST_TYPE_HW) {
+		switch (color_space) {
+		case COLOR_SPACE_SRGB:
+			/* by pass */
+			set_reg_field_value(
+				value,
+				0,
+				COL_MAN_OUTPUT_CSC_CONTROL,
+				OUTPUT_CSC_MODE);
+			break;
+		case COLOR_SPACE_SRGB_LIMITED:
+			/* not supported for underlay on CZ */
+			return false;
+		case COLOR_SPACE_YCBCR601:
+		case COLOR_SPACE_YCBCR601_LIMITED:
+			/* YCbCr601 */
+			set_reg_field_value(
+				value,
+				2,
+				COL_MAN_OUTPUT_CSC_CONTROL,
+				OUTPUT_CSC_MODE);
+			break;
+		case COLOR_SPACE_YCBCR709:
+		case COLOR_SPACE_YCBCR709_LIMITED:
+			 /* YCbCr709 */
+			set_reg_field_value(
+				value,
+				3,
+				COL_MAN_OUTPUT_CSC_CONTROL,
+				OUTPUT_CSC_MODE);
+			break;
+		default:
+			return false;
+		}
+
+	} else
+		/* by pass */
+		set_reg_field_value(
+			value,
+			0,
+			COL_MAN_OUTPUT_CSC_CONTROL,
+			OUTPUT_CSC_MODE);
+
+	addr = mmCOL_MAN_OUTPUT_CSC_CONTROL;
+	dm_write_reg(ctx, addr, value);
+
+	return true;
+}
+
+/*TODO: color depth is not correct when this is called*/
+static void set_Denormalization(struct transform *xfm,
+		enum dc_color_depth color_depth)
+{
+	uint32_t value = dm_read_reg(xfm->ctx, mmDENORM_CLAMP_CONTROL);
+
+	switch (color_depth) {
+	case COLOR_DEPTH_888:
+		/* 255/256 for 8 bit output color depth */
+		set_reg_field_value(
+			value,
+			1,
+			DENORM_CLAMP_CONTROL,
+			DENORM_MODE);
+		break;
+	case COLOR_DEPTH_101010:
+		/* 1023/1024 for 10 bit output color depth */
+		set_reg_field_value(
+			value,
+			2,
+			DENORM_CLAMP_CONTROL,
+			DENORM_MODE);
+		break;
+	case COLOR_DEPTH_121212:
+		/* 4095/4096 for 12 bit output color depth */
+		set_reg_field_value(
+			value,
+			3,
+			DENORM_CLAMP_CONTROL,
+			DENORM_MODE);
+		break;
+	default:
+		/* not valid case */
+		break;
+	}
+
+	set_reg_field_value(
+		value,
+		1,
+		DENORM_CLAMP_CONTROL,
+		DENORM_10BIT_OUT);
+
+	dm_write_reg(xfm->ctx, mmDENORM_CLAMP_CONTROL, value);
+}
+
+struct input_csc_matrix {
+	enum dc_color_space color_space;
+	uint32_t regval[12];
+};
+
+static const struct input_csc_matrix input_csc_matrix[] = {
+	{COLOR_SPACE_SRGB,
+/*1_1   1_2   1_3   1_4   2_1   2_2   2_3   2_4   3_1   3_2   3_3   3_4 */
+		{0x2000, 0, 0, 0, 0, 0x2000, 0, 0, 0, 0, 0x2000, 0} },
+	{COLOR_SPACE_SRGB_LIMITED,
+		{0x2000, 0, 0, 0, 0, 0x2000, 0, 0, 0, 0, 0x2000, 0} },
+	{COLOR_SPACE_YCBCR601,
+		{0x2cdd, 0x2000, 0x0, 0xe991, 0xe926, 0x2000, 0xf4fd, 0x10ef,
+						0x0, 0x2000, 0x38b4, 0xe3a6} },
+	{COLOR_SPACE_YCBCR601_LIMITED,
+		{0x3353, 0x2568, 0x0, 0xe400, 0xe5dc, 0x2568, 0xf367, 0x1108,
+						0x0, 0x2568, 0x40de, 0xdd3a} },
+	{COLOR_SPACE_YCBCR709,
+		{0x3265, 0x2000, 0, 0xe6ce, 0xf105, 0x2000, 0xfa01, 0xa7d, 0,
+						0x2000, 0x3b61, 0xe24f} },
+	{COLOR_SPACE_YCBCR709_LIMITED,
+		{0x39a6, 0x2568, 0, 0xe0d6, 0xeedd, 0x2568, 0xf925, 0x9a8, 0,
+						0x2568, 0x43ee, 0xdbb2} }
+};
+
+static void program_input_csc(
+		struct transform *xfm, enum dc_color_space color_space)
+{
+	int arr_size = sizeof(input_csc_matrix)/sizeof(struct input_csc_matrix);
+	struct dc_context *ctx = xfm->ctx;
+	const uint32_t *regval = NULL;
+	bool use_set_a;
+	uint32_t value;
+	int i;
+
+	for (i = 0; i < arr_size; i++)
+		if (input_csc_matrix[i].color_space == color_space) {
+			regval = input_csc_matrix[i].regval;
+			break;
+		}
+	if (regval == NULL) {
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+	/*
+	 * 1 == set A, the logic is 'if currently we're not using set A,
+	 * then use set A, otherwise use set B'
+	 */
+	value = dm_read_reg(ctx, mmCOL_MAN_INPUT_CSC_CONTROL);
+	use_set_a = get_reg_field_value(
+		value, COL_MAN_INPUT_CSC_CONTROL, INPUT_CSC_MODE) != 1;
+
+	if (use_set_a) {
+		/* fixed S2.13 format */
+		value = 0;
+		set_reg_field_value(
+			value, regval[0], INPUT_CSC_C11_C12_A, INPUT_CSC_C11_A);
+		set_reg_field_value(
+			value, regval[1], INPUT_CSC_C11_C12_A, INPUT_CSC_C12_A);
+		dm_write_reg(ctx, mmINPUT_CSC_C11_C12_A, value);
+
+		value = 0;
+		set_reg_field_value(
+			value, regval[2], INPUT_CSC_C13_C14_A, INPUT_CSC_C13_A);
+		set_reg_field_value(
+			value, regval[3], INPUT_CSC_C13_C14_A, INPUT_CSC_C14_A);
+		dm_write_reg(ctx, mmINPUT_CSC_C13_C14_A, value);
+
+		value = 0;
+		set_reg_field_value(
+			value, regval[4], INPUT_CSC_C21_C22_A, INPUT_CSC_C21_A);
+		set_reg_field_value(
+			value, regval[5], INPUT_CSC_C21_C22_A, INPUT_CSC_C22_A);
+		dm_write_reg(ctx, mmINPUT_CSC_C21_C22_A, value);
+
+		value = 0;
+		set_reg_field_value(
+			value, regval[6], INPUT_CSC_C23_C24_A, INPUT_CSC_C23_A);
+		set_reg_field_value(
+			value, regval[7], INPUT_CSC_C23_C24_A, INPUT_CSC_C24_A);
+		dm_write_reg(ctx, mmINPUT_CSC_C23_C24_A, value);
+
+		value = 0;
+		set_reg_field_value(
+			value, regval[8], INPUT_CSC_C31_C32_A, INPUT_CSC_C31_A);
+		set_reg_field_value(
+			value, regval[9], INPUT_CSC_C31_C32_A, INPUT_CSC_C32_A);
+		dm_write_reg(ctx, mmINPUT_CSC_C31_C32_A, value);
+
+		value = 0;
+		set_reg_field_value(
+			value, regval[10], INPUT_CSC_C33_C34_A, INPUT_CSC_C33_A);
+		set_reg_field_value(
+			value, regval[11], INPUT_CSC_C33_C34_A, INPUT_CSC_C34_A);
+		dm_write_reg(ctx, mmINPUT_CSC_C33_C34_A, value);
+	} else {
+		/* fixed S2.13 format */
+		value = 0;
+		set_reg_field_value(
+			value, regval[0], INPUT_CSC_C11_C12_B, INPUT_CSC_C11_B);
+		set_reg_field_value(
+			value, regval[1], INPUT_CSC_C11_C12_B, INPUT_CSC_C12_B);
+		dm_write_reg(ctx, mmINPUT_CSC_C11_C12_B, value);
+
+		value = 0;
+		set_reg_field_value(
+			value, regval[2], INPUT_CSC_C13_C14_B, INPUT_CSC_C13_B);
+		set_reg_field_value(
+			value, regval[3], INPUT_CSC_C13_C14_B, INPUT_CSC_C14_B);
+		dm_write_reg(ctx, mmINPUT_CSC_C13_C14_B, value);
+
+		value = 0;
+		set_reg_field_value(
+			value, regval[4], INPUT_CSC_C21_C22_B, INPUT_CSC_C21_B);
+		set_reg_field_value(
+			value, regval[5], INPUT_CSC_C21_C22_B, INPUT_CSC_C22_B);
+		dm_write_reg(ctx, mmINPUT_CSC_C21_C22_B, value);
+
+		value = 0;
+		set_reg_field_value(
+			value, regval[6], INPUT_CSC_C23_C24_B, INPUT_CSC_C23_B);
+		set_reg_field_value(
+			value, regval[7], INPUT_CSC_C23_C24_B, INPUT_CSC_C24_B);
+		dm_write_reg(ctx, mmINPUT_CSC_C23_C24_B, value);
+
+		value = 0;
+		set_reg_field_value(
+			value, regval[8], INPUT_CSC_C31_C32_B, INPUT_CSC_C31_B);
+		set_reg_field_value(
+			value, regval[9], INPUT_CSC_C31_C32_B, INPUT_CSC_C32_B);
+		dm_write_reg(ctx, mmINPUT_CSC_C31_C32_B, value);
+
+		value = 0;
+		set_reg_field_value(
+			value, regval[10], INPUT_CSC_C33_C34_B, INPUT_CSC_C33_B);
+		set_reg_field_value(
+			value, regval[11], INPUT_CSC_C33_C34_B, INPUT_CSC_C34_B);
+		dm_write_reg(ctx, mmINPUT_CSC_C33_C34_B, value);
+	}
+
+	/* KK: leave INPUT_CSC_CONVERSION_MODE at default */
+	value = 0;
+	/*
+	 * select 8.4 input type instead of default 12.0. From the discussion
+	 * with HW team, this format depends on the UNP surface format, so for
+	 * 8-bit we should select 8.4 (4 bits truncated). For 10 it should be
+	 * 10.2. For Carrizo we only support 8-bit surfaces on underlay pipe
+	 * so we can always keep this at 8.4 (input_type=2). If the later asics
+	 * start supporting 10+ bits, we will have a problem: surface
+	 * programming including UNP_GRPH* is being done in DalISR after this,
+	 * so either we pass surface format to here, or move this logic to ISR
+	 */
+
+	set_reg_field_value(
+		value, 2, COL_MAN_INPUT_CSC_CONTROL, INPUT_CSC_INPUT_TYPE);
+	set_reg_field_value(
+		value,
+		use_set_a ? 1 : 2,
+		COL_MAN_INPUT_CSC_CONTROL,
+		INPUT_CSC_MODE);
+
+	dm_write_reg(ctx, mmCOL_MAN_INPUT_CSC_CONTROL, value);
+}
+
+void dce110_opp_v_set_csc_default(
+	struct transform *xfm,
+	const struct default_adjustment *default_adjust)
+{
+	struct dce_transform *xfm_dce = TO_DCE_TRANSFORM(xfm);
+	enum csc_color_mode config =
+			CSC_COLOR_MODE_GRAPHICS_PREDEFINED;
+
+	if (default_adjust->force_hw_default == false) {
+		const struct out_csc_color_matrix *elm;
+		/* currently parameter not in use */
+		enum grph_color_adjust_option option;
+		uint32_t i;
+		/*
+		 * HW default false we program locally defined matrix
+		 * HW default true  we use predefined hw matrix and we
+		 * do not need to program matrix
+		 * OEM wants the HW default via runtime parameter.
+		 */
+		option = GRPH_COLOR_MATRIX_SW;
+
+		for (i = 0; i < ARRAY_SIZE(global_color_matrix); ++i) {
+			elm = &global_color_matrix[i];
+			if (elm->color_space != default_adjust->out_color_space)
+				continue;
+			/* program the matrix with default values from this
+			 * file
+			 */
+			program_color_matrix_v(xfm_dce, elm, option);
+			config = CSC_COLOR_MODE_GRAPHICS_OUTPUT_CSC;
+			break;
+		}
+	}
+
+	program_input_csc(xfm, default_adjust->in_color_space);
+
+	/* configure the what we programmed :
+	 * 1. Default values from this file
+	 * 2. Use hardware default from ROM_A and we do not need to program
+	 * matrix
+	 */
+
+	configure_graphics_mode_v(xfm_dce, config,
+		default_adjust->csc_adjust_type,
+		default_adjust->out_color_space);
+
+	set_Denormalization(xfm, default_adjust->color_depth);
+}
+
+void dce110_opp_v_set_csc_adjustment(
+	struct transform *xfm,
+	const struct out_csc_color_matrix *tbl_entry)
+{
+	struct dce_transform *xfm_dce = TO_DCE_TRANSFORM(xfm);
+	enum csc_color_mode config =
+			CSC_COLOR_MODE_GRAPHICS_OUTPUT_CSC;
+
+	program_color_matrix_v(
+			xfm_dce, tbl_entry, GRPH_COLOR_MATRIX_SW);
+
+	/*  We did everything ,now program DxOUTPUT_CSC_CONTROL */
+	configure_graphics_mode_v(xfm_dce, config, GRAPHICS_CSC_ADJUST_TYPE_SW,
+			tbl_entry->color_space);
+
+	/*TODO: Check if denormalization is needed*/
+	/*set_Denormalization(opp, adjust->color_depth);*/
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_opp_regamma_v.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_opp_regamma_v.c
new file mode 100644
index 000000000000..9b65b77e8823
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_opp_regamma_v.c
@@ -0,0 +1,555 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+
+/* include DCE11 register header files */
+#include "dce/dce_11_0_d.h"
+#include "dce/dce_11_0_sh_mask.h"
+
+#include "dce110_transform_v.h"
+
+static void power_on_lut(struct transform *xfm,
+	bool power_on, bool inputgamma, bool regamma)
+{
+	uint32_t value = dm_read_reg(xfm->ctx, mmDCFEV_MEM_PWR_CTRL);
+	int i;
+
+	if (power_on) {
+		if (inputgamma)
+			set_reg_field_value(
+				value,
+				1,
+				DCFEV_MEM_PWR_CTRL,
+				COL_MAN_INPUT_GAMMA_MEM_PWR_DIS);
+		if (regamma)
+			set_reg_field_value(
+				value,
+				1,
+				DCFEV_MEM_PWR_CTRL,
+				COL_MAN_GAMMA_CORR_MEM_PWR_DIS);
+	} else {
+		if (inputgamma)
+			set_reg_field_value(
+				value,
+				0,
+				DCFEV_MEM_PWR_CTRL,
+				COL_MAN_INPUT_GAMMA_MEM_PWR_DIS);
+		if (regamma)
+			set_reg_field_value(
+				value,
+				0,
+				DCFEV_MEM_PWR_CTRL,
+				COL_MAN_GAMMA_CORR_MEM_PWR_DIS);
+	}
+
+	dm_write_reg(xfm->ctx, mmDCFEV_MEM_PWR_CTRL, value);
+
+	for (i = 0; i < 3; i++) {
+		value = dm_read_reg(xfm->ctx, mmDCFEV_MEM_PWR_CTRL);
+		if (get_reg_field_value(value,
+				DCFEV_MEM_PWR_CTRL,
+				COL_MAN_INPUT_GAMMA_MEM_PWR_DIS) &&
+			get_reg_field_value(value,
+					DCFEV_MEM_PWR_CTRL,
+					COL_MAN_GAMMA_CORR_MEM_PWR_DIS))
+			break;
+
+		udelay(2);
+	}
+}
+
+static void set_bypass_input_gamma(struct dce_transform *xfm_dce)
+{
+	uint32_t value;
+
+	value = dm_read_reg(xfm_dce->base.ctx,
+			mmCOL_MAN_INPUT_GAMMA_CONTROL1);
+
+	set_reg_field_value(
+				value,
+				0,
+				COL_MAN_INPUT_GAMMA_CONTROL1,
+				INPUT_GAMMA_MODE);
+
+	dm_write_reg(xfm_dce->base.ctx,
+			mmCOL_MAN_INPUT_GAMMA_CONTROL1, value);
+}
+
+static void configure_regamma_mode(struct dce_transform *xfm_dce, uint32_t mode)
+{
+	uint32_t value = 0;
+
+	set_reg_field_value(
+				value,
+				mode,
+				GAMMA_CORR_CONTROL,
+				GAMMA_CORR_MODE);
+
+	dm_write_reg(xfm_dce->base.ctx, mmGAMMA_CORR_CONTROL, 0);
+}
+
+/*
+ *****************************************************************************
+ *  Function: regamma_config_regions_and_segments
+ *
+ *     build regamma curve by using predefined hw points
+ *     uses interface parameters ,like EDID coeff.
+ *
+ * @param   : parameters   interface parameters
+ *  @return void
+ *
+ *  @note
+ *
+ *  @see
+ *
+ *****************************************************************************
+ */
+static void regamma_config_regions_and_segments(
+	struct dce_transform *xfm_dce, const struct pwl_params *params)
+{
+	const struct gamma_curve *curve;
+	uint32_t value = 0;
+
+	{
+		set_reg_field_value(
+			value,
+			params->arr_points[0].custom_float_x,
+			GAMMA_CORR_CNTLA_START_CNTL,
+			GAMMA_CORR_CNTLA_EXP_REGION_START);
+
+		set_reg_field_value(
+			value,
+			0,
+			GAMMA_CORR_CNTLA_START_CNTL,
+			GAMMA_CORR_CNTLA_EXP_REGION_START_SEGMENT);
+
+		dm_write_reg(xfm_dce->base.ctx, mmGAMMA_CORR_CNTLA_START_CNTL,
+				value);
+	}
+	{
+		value = 0;
+		set_reg_field_value(
+			value,
+			params->arr_points[0].custom_float_slope,
+			GAMMA_CORR_CNTLA_SLOPE_CNTL,
+			GAMMA_CORR_CNTLA_EXP_REGION_LINEAR_SLOPE);
+
+		dm_write_reg(xfm_dce->base.ctx,
+			mmGAMMA_CORR_CNTLA_SLOPE_CNTL, value);
+	}
+	{
+		value = 0;
+		set_reg_field_value(
+			value,
+			params->arr_points[1].custom_float_x,
+			GAMMA_CORR_CNTLA_END_CNTL1,
+			GAMMA_CORR_CNTLA_EXP_REGION_END);
+
+		dm_write_reg(xfm_dce->base.ctx,
+			mmGAMMA_CORR_CNTLA_END_CNTL1, value);
+	}
+	{
+		value = 0;
+		set_reg_field_value(
+			value,
+			params->arr_points[1].custom_float_slope,
+			GAMMA_CORR_CNTLA_END_CNTL2,
+			GAMMA_CORR_CNTLA_EXP_REGION_END_BASE);
+
+		set_reg_field_value(
+			value,
+			params->arr_points[1].custom_float_y,
+			GAMMA_CORR_CNTLA_END_CNTL2,
+			GAMMA_CORR_CNTLA_EXP_REGION_END_SLOPE);
+
+		dm_write_reg(xfm_dce->base.ctx,
+			mmGAMMA_CORR_CNTLA_END_CNTL2, value);
+	}
+
+	curve = params->arr_curve_points;
+
+	{
+		value = 0;
+		set_reg_field_value(
+			value,
+			curve[0].offset,
+			GAMMA_CORR_CNTLA_REGION_0_1,
+			GAMMA_CORR_CNTLA_EXP_REGION0_LUT_OFFSET);
+
+		set_reg_field_value(
+			value,
+			curve[0].segments_num,
+			GAMMA_CORR_CNTLA_REGION_0_1,
+			GAMMA_CORR_CNTLA_EXP_REGION0_NUM_SEGMENTS);
+
+		set_reg_field_value(
+			value,
+			curve[1].offset,
+			GAMMA_CORR_CNTLA_REGION_0_1,
+			GAMMA_CORR_CNTLA_EXP_REGION1_LUT_OFFSET);
+
+		set_reg_field_value(
+			value,
+			curve[1].segments_num,
+			GAMMA_CORR_CNTLA_REGION_0_1,
+			GAMMA_CORR_CNTLA_EXP_REGION1_NUM_SEGMENTS);
+
+		dm_write_reg(
+				xfm_dce->base.ctx,
+			mmGAMMA_CORR_CNTLA_REGION_0_1,
+			value);
+	}
+
+	curve += 2;
+	{
+		value = 0;
+		set_reg_field_value(
+			value,
+			curve[0].offset,
+			GAMMA_CORR_CNTLA_REGION_2_3,
+			GAMMA_CORR_CNTLA_EXP_REGION2_LUT_OFFSET);
+
+		set_reg_field_value(
+			value,
+			curve[0].segments_num,
+			GAMMA_CORR_CNTLA_REGION_2_3,
+			GAMMA_CORR_CNTLA_EXP_REGION2_NUM_SEGMENTS);
+
+		set_reg_field_value(
+			value,
+			curve[1].offset,
+			GAMMA_CORR_CNTLA_REGION_2_3,
+			GAMMA_CORR_CNTLA_EXP_REGION3_LUT_OFFSET);
+
+		set_reg_field_value(
+			value,
+			curve[1].segments_num,
+			GAMMA_CORR_CNTLA_REGION_2_3,
+			GAMMA_CORR_CNTLA_EXP_REGION3_NUM_SEGMENTS);
+
+		dm_write_reg(xfm_dce->base.ctx,
+			mmGAMMA_CORR_CNTLA_REGION_2_3,
+			value);
+	}
+
+	curve += 2;
+	{
+		value = 0;
+		set_reg_field_value(
+			value,
+			curve[0].offset,
+			GAMMA_CORR_CNTLA_REGION_4_5,
+			GAMMA_CORR_CNTLA_EXP_REGION4_LUT_OFFSET);
+
+		set_reg_field_value(
+			value,
+			curve[0].segments_num,
+			GAMMA_CORR_CNTLA_REGION_4_5,
+			GAMMA_CORR_CNTLA_EXP_REGION4_NUM_SEGMENTS);
+
+		set_reg_field_value(
+			value,
+			curve[1].offset,
+			GAMMA_CORR_CNTLA_REGION_4_5,
+			GAMMA_CORR_CNTLA_EXP_REGION5_LUT_OFFSET);
+
+		set_reg_field_value(
+			value,
+			curve[1].segments_num,
+			GAMMA_CORR_CNTLA_REGION_4_5,
+			GAMMA_CORR_CNTLA_EXP_REGION5_NUM_SEGMENTS);
+
+		dm_write_reg(xfm_dce->base.ctx,
+			mmGAMMA_CORR_CNTLA_REGION_4_5,
+			value);
+	}
+
+	curve += 2;
+	{
+		value = 0;
+		set_reg_field_value(
+			value,
+			curve[0].offset,
+			GAMMA_CORR_CNTLA_REGION_6_7,
+			GAMMA_CORR_CNTLA_EXP_REGION6_LUT_OFFSET);
+
+		set_reg_field_value(
+			value,
+			curve[0].segments_num,
+			GAMMA_CORR_CNTLA_REGION_6_7,
+			GAMMA_CORR_CNTLA_EXP_REGION6_NUM_SEGMENTS);
+
+		set_reg_field_value(
+			value,
+			curve[1].offset,
+			GAMMA_CORR_CNTLA_REGION_6_7,
+			GAMMA_CORR_CNTLA_EXP_REGION7_LUT_OFFSET);
+
+		set_reg_field_value(
+			value,
+			curve[1].segments_num,
+			GAMMA_CORR_CNTLA_REGION_6_7,
+			GAMMA_CORR_CNTLA_EXP_REGION7_NUM_SEGMENTS);
+
+		dm_write_reg(xfm_dce->base.ctx,
+			mmGAMMA_CORR_CNTLA_REGION_6_7,
+			value);
+	}
+
+	curve += 2;
+	{
+		value = 0;
+		set_reg_field_value(
+			value,
+			curve[0].offset,
+			GAMMA_CORR_CNTLA_REGION_8_9,
+			GAMMA_CORR_CNTLA_EXP_REGION8_LUT_OFFSET);
+
+		set_reg_field_value(
+			value,
+			curve[0].segments_num,
+			GAMMA_CORR_CNTLA_REGION_8_9,
+			GAMMA_CORR_CNTLA_EXP_REGION8_NUM_SEGMENTS);
+
+		set_reg_field_value(
+			value,
+			curve[1].offset,
+			GAMMA_CORR_CNTLA_REGION_8_9,
+			GAMMA_CORR_CNTLA_EXP_REGION9_LUT_OFFSET);
+
+		set_reg_field_value(
+			value,
+			curve[1].segments_num,
+			GAMMA_CORR_CNTLA_REGION_8_9,
+			GAMMA_CORR_CNTLA_EXP_REGION9_NUM_SEGMENTS);
+
+		dm_write_reg(xfm_dce->base.ctx,
+			mmGAMMA_CORR_CNTLA_REGION_8_9,
+			value);
+	}
+
+	curve += 2;
+	{
+		value = 0;
+		set_reg_field_value(
+			value,
+			curve[0].offset,
+			GAMMA_CORR_CNTLA_REGION_10_11,
+			GAMMA_CORR_CNTLA_EXP_REGION10_LUT_OFFSET);
+
+		set_reg_field_value(
+			value,
+			curve[0].segments_num,
+			GAMMA_CORR_CNTLA_REGION_10_11,
+			GAMMA_CORR_CNTLA_EXP_REGION10_NUM_SEGMENTS);
+
+		set_reg_field_value(
+			value,
+			curve[1].offset,
+			GAMMA_CORR_CNTLA_REGION_10_11,
+			GAMMA_CORR_CNTLA_EXP_REGION11_LUT_OFFSET);
+
+		set_reg_field_value(
+			value,
+			curve[1].segments_num,
+			GAMMA_CORR_CNTLA_REGION_10_11,
+			GAMMA_CORR_CNTLA_EXP_REGION11_NUM_SEGMENTS);
+
+		dm_write_reg(xfm_dce->base.ctx,
+			mmGAMMA_CORR_CNTLA_REGION_10_11,
+			value);
+	}
+
+	curve += 2;
+	{
+		value = 0;
+		set_reg_field_value(
+			value,
+			curve[0].offset,
+			GAMMA_CORR_CNTLA_REGION_12_13,
+			GAMMA_CORR_CNTLA_EXP_REGION12_LUT_OFFSET);
+
+		set_reg_field_value(
+			value,
+			curve[0].segments_num,
+			GAMMA_CORR_CNTLA_REGION_12_13,
+			GAMMA_CORR_CNTLA_EXP_REGION12_NUM_SEGMENTS);
+
+		set_reg_field_value(
+			value,
+			curve[1].offset,
+			GAMMA_CORR_CNTLA_REGION_12_13,
+			GAMMA_CORR_CNTLA_EXP_REGION13_LUT_OFFSET);
+
+		set_reg_field_value(
+			value,
+			curve[1].segments_num,
+			GAMMA_CORR_CNTLA_REGION_12_13,
+			GAMMA_CORR_CNTLA_EXP_REGION13_NUM_SEGMENTS);
+
+		dm_write_reg(xfm_dce->base.ctx,
+			mmGAMMA_CORR_CNTLA_REGION_12_13,
+			value);
+	}
+
+	curve += 2;
+	{
+		value = 0;
+		set_reg_field_value(
+			value,
+			curve[0].offset,
+			GAMMA_CORR_CNTLA_REGION_14_15,
+			GAMMA_CORR_CNTLA_EXP_REGION14_LUT_OFFSET);
+
+		set_reg_field_value(
+			value,
+			curve[0].segments_num,
+			GAMMA_CORR_CNTLA_REGION_14_15,
+			GAMMA_CORR_CNTLA_EXP_REGION14_NUM_SEGMENTS);
+
+		set_reg_field_value(
+			value,
+			curve[1].offset,
+			GAMMA_CORR_CNTLA_REGION_14_15,
+			GAMMA_CORR_CNTLA_EXP_REGION15_LUT_OFFSET);
+
+		set_reg_field_value(
+			value,
+			curve[1].segments_num,
+			GAMMA_CORR_CNTLA_REGION_14_15,
+			GAMMA_CORR_CNTLA_EXP_REGION15_NUM_SEGMENTS);
+
+		dm_write_reg(xfm_dce->base.ctx,
+			mmGAMMA_CORR_CNTLA_REGION_14_15,
+			value);
+	}
+}
+
+static void program_pwl(struct dce_transform *xfm_dce,
+		const struct pwl_params *params)
+{
+	uint32_t value = 0;
+
+	set_reg_field_value(
+		value,
+		7,
+		GAMMA_CORR_LUT_WRITE_EN_MASK,
+		GAMMA_CORR_LUT_WRITE_EN_MASK);
+
+	dm_write_reg(xfm_dce->base.ctx,
+		mmGAMMA_CORR_LUT_WRITE_EN_MASK, value);
+
+	dm_write_reg(xfm_dce->base.ctx,
+		mmGAMMA_CORR_LUT_INDEX, 0);
+
+	/* Program REGAMMA_LUT_DATA */
+	{
+		const uint32_t addr = mmGAMMA_CORR_LUT_DATA;
+		uint32_t i = 0;
+		const struct pwl_result_data *rgb =
+				params->rgb_resulted;
+
+		while (i != params->hw_points_num) {
+			dm_write_reg(xfm_dce->base.ctx, addr, rgb->red_reg);
+			dm_write_reg(xfm_dce->base.ctx, addr, rgb->green_reg);
+			dm_write_reg(xfm_dce->base.ctx, addr, rgb->blue_reg);
+
+			dm_write_reg(xfm_dce->base.ctx, addr,
+				rgb->delta_red_reg);
+			dm_write_reg(xfm_dce->base.ctx, addr,
+				rgb->delta_green_reg);
+			dm_write_reg(xfm_dce->base.ctx, addr,
+				rgb->delta_blue_reg);
+
+			++rgb;
+			++i;
+		}
+	}
+}
+
+void dce110_opp_program_regamma_pwl_v(
+	struct transform *xfm,
+	const struct pwl_params *params)
+{
+	struct dce_transform *xfm_dce = TO_DCE_TRANSFORM(xfm);
+
+	/* Setup regions */
+	regamma_config_regions_and_segments(xfm_dce, params);
+
+	set_bypass_input_gamma(xfm_dce);
+
+	/* Power on gamma LUT memory */
+	power_on_lut(xfm, true, false, true);
+
+	/* Program PWL */
+	program_pwl(xfm_dce, params);
+
+	/* program regamma config */
+	configure_regamma_mode(xfm_dce, 1);
+
+	/* Power return to auto back */
+	power_on_lut(xfm, false, false, true);
+}
+
+void dce110_opp_power_on_regamma_lut_v(
+	struct transform *xfm,
+	bool power_on)
+{
+	uint32_t value = dm_read_reg(xfm->ctx, mmDCFEV_MEM_PWR_CTRL);
+
+	set_reg_field_value(
+		value,
+		0,
+		DCFEV_MEM_PWR_CTRL,
+		COL_MAN_GAMMA_CORR_MEM_PWR_FORCE);
+
+	set_reg_field_value(
+		value,
+		power_on,
+		DCFEV_MEM_PWR_CTRL,
+		COL_MAN_GAMMA_CORR_MEM_PWR_DIS);
+
+	set_reg_field_value(
+		value,
+		0,
+		DCFEV_MEM_PWR_CTRL,
+		COL_MAN_INPUT_GAMMA_MEM_PWR_FORCE);
+
+	set_reg_field_value(
+		value,
+		power_on,
+		DCFEV_MEM_PWR_CTRL,
+		COL_MAN_INPUT_GAMMA_MEM_PWR_DIS);
+
+	dm_write_reg(xfm->ctx, mmDCFEV_MEM_PWR_CTRL, value);
+}
+
+void dce110_opp_set_regamma_mode_v(
+	struct transform *xfm,
+	enum opp_regamma mode)
+{
+	// TODO: need to implement the function
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_opp_v.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_opp_v.c
new file mode 100644
index 000000000000..3545e43a4b77
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_opp_v.c
@@ -0,0 +1,54 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+
+/* include DCE11 register header files */
+#include "dce/dce_11_0_d.h"
+#include "dce/dce_11_0_sh_mask.h"
+
+#include "dce/dce_opp.h"
+#include "dce110_opp_v.h"
+
+/*****************************************/
+/* Constructor, Destructor               */
+/*****************************************/
+
+static const struct opp_funcs funcs = {
+		.opp_set_dyn_expansion = dce110_opp_set_dyn_expansion,
+		.opp_destroy = dce110_opp_destroy,
+		.opp_program_fmt = dce110_opp_program_fmt,
+		.opp_program_bit_depth_reduction =
+				dce110_opp_program_bit_depth_reduction
+};
+
+void dce110_opp_v_construct(struct dce110_opp *opp110,
+	struct dc_context *ctx)
+{
+	opp110->base.funcs = &funcs;
+
+	opp110->base.ctx = ctx;
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_opp_v.h b/drivers/gpu/drm/amd/display/dc/dce110/dce110_opp_v.h
new file mode 100644
index 000000000000..152af4c418cb
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_opp_v.h
@@ -0,0 +1,39 @@
+/* Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_OPP_DCE110_V_H__
+#define __DC_OPP_DCE110_V_H__
+
+#include "dc_types.h"
+#include "opp.h"
+#include "core_types.h"
+
+void dce110_opp_v_construct(struct dce110_opp *opp110,
+	struct dc_context *ctx);
+
+/* underlay callbacks */
+
+
+
+#endif /* __DC_OPP_DCE110_V_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator.c
new file mode 100644
index 000000000000..61b0807693fb
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator.c
@@ -0,0 +1,2362 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+
+/* include DCE11 register header files */
+#include "dce/dce_11_0_d.h"
+#include "dce/dce_11_0_sh_mask.h"
+
+#include "dc_types.h"
+#include "dc_bios_types.h"
+#include "dc.h"
+
+#include "include/grph_object_id.h"
+#include "include/logger_interface.h"
+#include "dce110_timing_generator.h"
+
+#include "timing_generator.h"
+
+
+#define NUMBER_OF_FRAME_TO_WAIT_ON_TRIGGERED_RESET 10
+
+#define MAX_H_TOTAL (CRTC_H_TOTAL__CRTC_H_TOTAL_MASK + 1)
+#define MAX_V_TOTAL (CRTC_V_TOTAL__CRTC_V_TOTAL_MASKhw + 1)
+
+#define CRTC_REG(reg) (reg + tg110->offsets.crtc)
+#define DCP_REG(reg) (reg + tg110->offsets.dcp)
+
+/* Flowing register offsets are same in files of
+ * dce/dce_11_0_d.h
+ * dce/vi_polaris10_p/vi_polaris10_d.h
+ *
+ * So we can create dce110 timing generator to use it.
+ */
+
+
+/*
+* apply_front_porch_workaround
+*
+* This is a workaround for a bug that has existed since R5xx and has not been
+* fixed keep Front porch at minimum 2 for Interlaced mode or 1 for progressive.
+*/
+static void dce110_timing_generator_apply_front_porch_workaround(
+	struct timing_generator *tg,
+	struct dc_crtc_timing *timing)
+{
+	if (timing->flags.INTERLACE == 1) {
+		if (timing->v_front_porch < 2)
+			timing->v_front_porch = 2;
+	} else {
+		if (timing->v_front_porch < 1)
+			timing->v_front_porch = 1;
+	}
+}
+
+/*
+ *****************************************************************************
+ *  Function: is_in_vertical_blank
+ *
+ *  @brief
+ *     check the current status of CRTC to check if we are in Vertical Blank
+ *     regioneased" state
+ *
+ *  @return
+ *     true if currently in blank region, false otherwise
+ *
+ *****************************************************************************
+ */
+static bool dce110_timing_generator_is_in_vertical_blank(
+		struct timing_generator *tg)
+{
+	uint32_t addr = 0;
+	uint32_t value = 0;
+	uint32_t field = 0;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	addr = CRTC_REG(mmCRTC_STATUS);
+	value = dm_read_reg(tg->ctx, addr);
+	field = get_reg_field_value(value, CRTC_STATUS, CRTC_V_BLANK);
+	return field == 1;
+}
+
+void dce110_timing_generator_set_early_control(
+		struct timing_generator *tg,
+		uint32_t early_cntl)
+{
+	uint32_t regval;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t address = CRTC_REG(mmCRTC_CONTROL);
+
+	regval = dm_read_reg(tg->ctx, address);
+	set_reg_field_value(regval, early_cntl,
+			CRTC_CONTROL, CRTC_HBLANK_EARLY_CONTROL);
+	dm_write_reg(tg->ctx, address, regval);
+}
+
+/*
+ * Enable CRTC
+ * Enable CRTC - call ASIC Control Object to enable Timing generator.
+ */
+bool dce110_timing_generator_enable_crtc(struct timing_generator *tg)
+{
+	enum bp_result result;
+
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t value = 0;
+
+	/*
+	 * 3 is used to make sure V_UPDATE occurs at the beginning of the first
+	 * line of vertical front porch
+	 */
+	set_reg_field_value(
+		value,
+		0,
+		CRTC_MASTER_UPDATE_MODE,
+		MASTER_UPDATE_MODE);
+
+	dm_write_reg(tg->ctx, CRTC_REG(mmCRTC_MASTER_UPDATE_MODE), value);
+
+	/* TODO: may want this on to catch underflow */
+	value = 0;
+	dm_write_reg(tg->ctx, CRTC_REG(mmCRTC_MASTER_UPDATE_LOCK), value);
+
+	result = tg->bp->funcs->enable_crtc(tg->bp, tg110->controller_id, true);
+
+	return result == BP_RESULT_OK;
+}
+
+void dce110_timing_generator_program_blank_color(
+		struct timing_generator *tg,
+		const struct tg_color *black_color)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t addr = CRTC_REG(mmCRTC_BLACK_COLOR);
+	uint32_t value = dm_read_reg(tg->ctx, addr);
+
+	set_reg_field_value(
+		value,
+		black_color->color_b_cb,
+		CRTC_BLACK_COLOR,
+		CRTC_BLACK_COLOR_B_CB);
+	set_reg_field_value(
+		value,
+		black_color->color_g_y,
+		CRTC_BLACK_COLOR,
+		CRTC_BLACK_COLOR_G_Y);
+	set_reg_field_value(
+		value,
+		black_color->color_r_cr,
+		CRTC_BLACK_COLOR,
+		CRTC_BLACK_COLOR_R_CR);
+
+	dm_write_reg(tg->ctx, addr, value);
+}
+
+/*
+ *****************************************************************************
+ *  Function: disable_stereo
+ *
+ *  @brief
+ *     Disables active stereo on controller
+ *     Frame Packing need to be disabled in vBlank or when CRTC not running
+ *****************************************************************************
+ */
+#if 0
+@TODOSTEREO
+static void disable_stereo(struct timing_generator *tg)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t addr = CRTC_REG(mmCRTC_3D_STRUCTURE_CONTROL);
+	uint32_t value = 0;
+	uint32_t test = 0;
+	uint32_t field = 0;
+	uint32_t struc_en = 0;
+	uint32_t struc_stereo_sel_ovr = 0;
+
+	value = dm_read_reg(tg->ctx, addr);
+	struc_en = get_reg_field_value(
+			value,
+			CRTC_3D_STRUCTURE_CONTROL,
+			CRTC_3D_STRUCTURE_EN);
+
+	struc_stereo_sel_ovr = get_reg_field_value(
+			value,
+			CRTC_3D_STRUCTURE_CONTROL,
+			CRTC_3D_STRUCTURE_STEREO_SEL_OVR);
+
+	/*
+	 * When disabling Frame Packing in 2 step mode, we need to program both
+	 * registers at the same frame
+	 * Programming it in the beginning of VActive makes sure we are ok
+	 */
+
+	if (struc_en != 0 && struc_stereo_sel_ovr == 0) {
+		tg->funcs->wait_for_vblank(tg);
+		tg->funcs->wait_for_vactive(tg);
+	}
+
+	value = 0;
+	dm_write_reg(tg->ctx, addr, value);
+
+	addr = tg->regs[IDX_CRTC_STEREO_CONTROL];
+	dm_write_reg(tg->ctx, addr, value);
+}
+#endif
+
+/*
+ * disable_crtc - call ASIC Control Object to disable Timing generator.
+ */
+bool dce110_timing_generator_disable_crtc(struct timing_generator *tg)
+{
+	enum bp_result result;
+
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	result = tg->bp->funcs->enable_crtc(tg->bp, tg110->controller_id, false);
+
+	/* Need to make sure stereo is disabled according to the DCE5.0 spec */
+
+	/*
+	 * @TODOSTEREO call this when adding stereo support
+	 * tg->funcs->disable_stereo(tg);
+	 */
+
+	return result == BP_RESULT_OK;
+}
+
+/*
+ * program_horz_count_by_2
+ * Programs DxCRTC_HORZ_COUNT_BY2_EN - 1 for DVI 30bpp mode, 0 otherwise
+ */
+static void program_horz_count_by_2(
+	struct timing_generator *tg,
+	const struct dc_crtc_timing *timing)
+{
+	uint32_t regval;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	regval = dm_read_reg(tg->ctx,
+			CRTC_REG(mmCRTC_COUNT_CONTROL));
+
+	set_reg_field_value(regval, 0, CRTC_COUNT_CONTROL,
+			CRTC_HORZ_COUNT_BY2_EN);
+
+	if (timing->flags.HORZ_COUNT_BY_TWO)
+		set_reg_field_value(regval, 1, CRTC_COUNT_CONTROL,
+					CRTC_HORZ_COUNT_BY2_EN);
+
+	dm_write_reg(tg->ctx,
+			CRTC_REG(mmCRTC_COUNT_CONTROL), regval);
+}
+
+/*
+ * program_timing_generator
+ * Program CRTC Timing Registers - DxCRTC_H_*, DxCRTC_V_*, Pixel repetition.
+ * Call ASIC Control Object to program Timings.
+ */
+bool dce110_timing_generator_program_timing_generator(
+	struct timing_generator *tg,
+	const struct dc_crtc_timing *dc_crtc_timing)
+{
+	enum bp_result result;
+	struct bp_hw_crtc_timing_parameters bp_params;
+	struct dc_crtc_timing patched_crtc_timing;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	uint32_t vsync_offset = dc_crtc_timing->v_border_bottom +
+			dc_crtc_timing->v_front_porch;
+	uint32_t v_sync_start = dc_crtc_timing->v_addressable + vsync_offset;
+
+	uint32_t hsync_offset = dc_crtc_timing->h_border_right +
+			dc_crtc_timing->h_front_porch;
+	uint32_t h_sync_start = dc_crtc_timing->h_addressable + hsync_offset;
+
+	memset(&bp_params, 0, sizeof(struct bp_hw_crtc_timing_parameters));
+
+	/* Due to an asic bug we need to apply the Front Porch workaround prior
+	 * to programming the timing.
+	 */
+
+	patched_crtc_timing = *dc_crtc_timing;
+
+	dce110_timing_generator_apply_front_porch_workaround(tg, &patched_crtc_timing);
+
+	bp_params.controller_id = tg110->controller_id;
+
+	bp_params.h_total = patched_crtc_timing.h_total;
+	bp_params.h_addressable =
+		patched_crtc_timing.h_addressable;
+	bp_params.v_total = patched_crtc_timing.v_total;
+	bp_params.v_addressable = patched_crtc_timing.v_addressable;
+
+	bp_params.h_sync_start = h_sync_start;
+	bp_params.h_sync_width = patched_crtc_timing.h_sync_width;
+	bp_params.v_sync_start = v_sync_start;
+	bp_params.v_sync_width = patched_crtc_timing.v_sync_width;
+
+	/* Set overscan */
+	bp_params.h_overscan_left =
+		patched_crtc_timing.h_border_left;
+	bp_params.h_overscan_right =
+		patched_crtc_timing.h_border_right;
+	bp_params.v_overscan_top = patched_crtc_timing.v_border_top;
+	bp_params.v_overscan_bottom =
+		patched_crtc_timing.v_border_bottom;
+
+	/* Set flags */
+	if (patched_crtc_timing.flags.HSYNC_POSITIVE_POLARITY == 1)
+		bp_params.flags.HSYNC_POSITIVE_POLARITY = 1;
+
+	if (patched_crtc_timing.flags.VSYNC_POSITIVE_POLARITY == 1)
+		bp_params.flags.VSYNC_POSITIVE_POLARITY = 1;
+
+	if (patched_crtc_timing.flags.INTERLACE == 1)
+		bp_params.flags.INTERLACE = 1;
+
+	if (patched_crtc_timing.flags.HORZ_COUNT_BY_TWO == 1)
+		bp_params.flags.HORZ_COUNT_BY_TWO = 1;
+
+	result = tg->bp->funcs->program_crtc_timing(tg->bp, &bp_params);
+
+	program_horz_count_by_2(tg, &patched_crtc_timing);
+
+	tg110->base.funcs->enable_advanced_request(tg, true, &patched_crtc_timing);
+
+	/* Enable stereo - only when we need to pack 3D frame. Other types
+	 * of stereo handled in explicit call */
+
+	return result == BP_RESULT_OK;
+}
+
+/*
+ *****************************************************************************
+ *  Function: set_drr
+ *
+ *  @brief
+ *     Program dynamic refresh rate registers m_DxCRTC_V_TOTAL_*.
+ *
+ *  @param [in] pHwCrtcTiming: point to H
+ *  wCrtcTiming struct
+ *****************************************************************************
+ */
+void dce110_timing_generator_set_drr(
+	struct timing_generator *tg,
+	const struct drr_params *params)
+{
+	/* register values */
+	uint32_t v_total_min = 0;
+	uint32_t v_total_max = 0;
+	uint32_t v_total_cntl = 0;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	uint32_t addr = 0;
+
+	addr = CRTC_REG(mmCRTC_V_TOTAL_MIN);
+	v_total_min = dm_read_reg(tg->ctx, addr);
+
+	addr = CRTC_REG(mmCRTC_V_TOTAL_MAX);
+	v_total_max = dm_read_reg(tg->ctx, addr);
+
+	addr = CRTC_REG(mmCRTC_V_TOTAL_CONTROL);
+	v_total_cntl = dm_read_reg(tg->ctx, addr);
+
+	if (params != NULL &&
+		params->vertical_total_max > 0 &&
+		params->vertical_total_min > 0) {
+
+		set_reg_field_value(v_total_max,
+				params->vertical_total_max - 1,
+				CRTC_V_TOTAL_MAX,
+				CRTC_V_TOTAL_MAX);
+
+		set_reg_field_value(v_total_min,
+				params->vertical_total_min - 1,
+				CRTC_V_TOTAL_MIN,
+				CRTC_V_TOTAL_MIN);
+
+		set_reg_field_value(v_total_cntl,
+				1,
+				CRTC_V_TOTAL_CONTROL,
+				CRTC_V_TOTAL_MIN_SEL);
+
+		set_reg_field_value(v_total_cntl,
+				1,
+				CRTC_V_TOTAL_CONTROL,
+				CRTC_V_TOTAL_MAX_SEL);
+
+		set_reg_field_value(v_total_cntl,
+				0,
+				CRTC_V_TOTAL_CONTROL,
+				CRTC_FORCE_LOCK_ON_EVENT);
+		set_reg_field_value(v_total_cntl,
+				0,
+				CRTC_V_TOTAL_CONTROL,
+				CRTC_FORCE_LOCK_TO_MASTER_VSYNC);
+
+		set_reg_field_value(v_total_cntl,
+				0,
+				CRTC_V_TOTAL_CONTROL,
+				CRTC_SET_V_TOTAL_MIN_MASK_EN);
+
+		set_reg_field_value(v_total_cntl,
+				0,
+				CRTC_V_TOTAL_CONTROL,
+				CRTC_SET_V_TOTAL_MIN_MASK);
+	} else {
+		set_reg_field_value(v_total_cntl,
+			0,
+			CRTC_V_TOTAL_CONTROL,
+			CRTC_SET_V_TOTAL_MIN_MASK);
+		set_reg_field_value(v_total_cntl,
+				0,
+				CRTC_V_TOTAL_CONTROL,
+				CRTC_V_TOTAL_MIN_SEL);
+		set_reg_field_value(v_total_cntl,
+				0,
+				CRTC_V_TOTAL_CONTROL,
+				CRTC_V_TOTAL_MAX_SEL);
+		set_reg_field_value(v_total_min,
+				0,
+				CRTC_V_TOTAL_MIN,
+				CRTC_V_TOTAL_MIN);
+		set_reg_field_value(v_total_max,
+				0,
+				CRTC_V_TOTAL_MAX,
+				CRTC_V_TOTAL_MAX);
+		set_reg_field_value(v_total_cntl,
+				0,
+				CRTC_V_TOTAL_CONTROL,
+				CRTC_FORCE_LOCK_ON_EVENT);
+		set_reg_field_value(v_total_cntl,
+				0,
+				CRTC_V_TOTAL_CONTROL,
+				CRTC_FORCE_LOCK_TO_MASTER_VSYNC);
+	}
+
+	addr = CRTC_REG(mmCRTC_V_TOTAL_MIN);
+	dm_write_reg(tg->ctx, addr, v_total_min);
+
+	addr = CRTC_REG(mmCRTC_V_TOTAL_MAX);
+	dm_write_reg(tg->ctx, addr, v_total_max);
+
+	addr = CRTC_REG(mmCRTC_V_TOTAL_CONTROL);
+	dm_write_reg(tg->ctx, addr, v_total_cntl);
+}
+
+void dce110_timing_generator_set_static_screen_control(
+	struct timing_generator *tg,
+	uint32_t event_triggers,
+	uint32_t num_frames)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t static_screen_cntl = 0;
+	uint32_t addr = 0;
+
+	// By register spec, it only takes 8 bit value
+	if (num_frames > 0xFF)
+		num_frames = 0xFF;
+
+	addr = CRTC_REG(mmCRTC_STATIC_SCREEN_CONTROL);
+	static_screen_cntl = dm_read_reg(tg->ctx, addr);
+
+	set_reg_field_value(static_screen_cntl,
+				event_triggers,
+				CRTC_STATIC_SCREEN_CONTROL,
+				CRTC_STATIC_SCREEN_EVENT_MASK);
+
+	set_reg_field_value(static_screen_cntl,
+				num_frames,
+				CRTC_STATIC_SCREEN_CONTROL,
+				CRTC_STATIC_SCREEN_FRAME_COUNT);
+
+	dm_write_reg(tg->ctx, addr, static_screen_cntl);
+}
+
+/*
+ * get_vblank_counter
+ *
+ * @brief
+ * Get counter for vertical blanks. use register CRTC_STATUS_FRAME_COUNT which
+ * holds the counter of frames.
+ *
+ * @param
+ * struct timing_generator *tg - [in] timing generator which controls the
+ * desired CRTC
+ *
+ * @return
+ * Counter of frames, which should equal to number of vblanks.
+ */
+uint32_t dce110_timing_generator_get_vblank_counter(struct timing_generator *tg)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t addr = CRTC_REG(mmCRTC_STATUS_FRAME_COUNT);
+	uint32_t value = dm_read_reg(tg->ctx, addr);
+	uint32_t field = get_reg_field_value(
+			value, CRTC_STATUS_FRAME_COUNT, CRTC_FRAME_COUNT);
+
+	return field;
+}
+
+/*
+ *****************************************************************************
+ *  Function: dce110_timing_generator_get_position
+ *
+ *  @brief
+ *     Returns CRTC vertical/horizontal counters
+ *
+ *  @param [out] position
+ *****************************************************************************
+ */
+void dce110_timing_generator_get_position(struct timing_generator *tg,
+	struct crtc_position *position)
+{
+	uint32_t value;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	value = dm_read_reg(tg->ctx, CRTC_REG(mmCRTC_STATUS_POSITION));
+
+	position->horizontal_count = get_reg_field_value(
+			value,
+			CRTC_STATUS_POSITION,
+			CRTC_HORZ_COUNT);
+
+	position->vertical_count = get_reg_field_value(
+			value,
+			CRTC_STATUS_POSITION,
+			CRTC_VERT_COUNT);
+
+	value = dm_read_reg(tg->ctx, CRTC_REG(mmCRTC_NOM_VERT_POSITION));
+
+	position->nominal_vcount = get_reg_field_value(
+			value,
+			CRTC_NOM_VERT_POSITION,
+			CRTC_VERT_COUNT_NOM);
+}
+
+/*
+ *****************************************************************************
+ *  Function: get_crtc_scanoutpos
+ *
+ *  @brief
+ *     Returns CRTC vertical/horizontal counters
+ *
+ *  @param [out] vpos, hpos
+ *****************************************************************************
+ */
+void dce110_timing_generator_get_crtc_scanoutpos(
+	struct timing_generator *tg,
+	uint32_t *v_blank_start,
+	uint32_t *v_blank_end,
+	uint32_t *h_position,
+	uint32_t *v_position)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	struct crtc_position position;
+
+	uint32_t value  = dm_read_reg(tg->ctx,
+			CRTC_REG(mmCRTC_V_BLANK_START_END));
+
+	*v_blank_start = get_reg_field_value(value,
+					     CRTC_V_BLANK_START_END,
+					     CRTC_V_BLANK_START);
+	*v_blank_end = get_reg_field_value(value,
+					   CRTC_V_BLANK_START_END,
+					   CRTC_V_BLANK_END);
+
+	dce110_timing_generator_get_position(
+			tg, &position);
+
+	*h_position = position.horizontal_count;
+	*v_position = position.vertical_count;
+}
+
+/* TODO: is it safe to assume that mask/shift of Primary and Underlay
+ * are the same?
+ * For example: today CRTC_H_TOTAL == CRTCV_H_TOTAL but is it always
+ * guaranteed? */
+void dce110_timing_generator_program_blanking(
+	struct timing_generator *tg,
+	const struct dc_crtc_timing *timing)
+{
+	uint32_t vsync_offset = timing->v_border_bottom +
+			timing->v_front_porch;
+	uint32_t v_sync_start = timing->v_addressable + vsync_offset;
+
+	uint32_t hsync_offset = timing->h_border_right +
+			timing->h_front_porch;
+	uint32_t h_sync_start = timing->h_addressable + hsync_offset;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	struct dc_context *ctx = tg->ctx;
+	uint32_t value = 0;
+	uint32_t addr = 0;
+	uint32_t tmp = 0;
+
+	addr = CRTC_REG(mmCRTC_H_TOTAL);
+	value = dm_read_reg(ctx, addr);
+	set_reg_field_value(
+		value,
+		timing->h_total - 1,
+		CRTC_H_TOTAL,
+		CRTC_H_TOTAL);
+	dm_write_reg(ctx, addr, value);
+
+	addr = CRTC_REG(mmCRTC_V_TOTAL);
+	value = dm_read_reg(ctx, addr);
+	set_reg_field_value(
+		value,
+		timing->v_total - 1,
+		CRTC_V_TOTAL,
+		CRTC_V_TOTAL);
+	dm_write_reg(ctx, addr, value);
+
+	/* In case of V_TOTAL_CONTROL is on, make sure V_TOTAL_MAX and
+	 * V_TOTAL_MIN are equal to V_TOTAL.
+	 */
+	addr = CRTC_REG(mmCRTC_V_TOTAL_MAX);
+	value = dm_read_reg(ctx, addr);
+	set_reg_field_value(
+		value,
+		timing->v_total - 1,
+		CRTC_V_TOTAL_MAX,
+		CRTC_V_TOTAL_MAX);
+	dm_write_reg(ctx, addr, value);
+
+	addr = CRTC_REG(mmCRTC_V_TOTAL_MIN);
+	value = dm_read_reg(ctx, addr);
+	set_reg_field_value(
+		value,
+		timing->v_total - 1,
+		CRTC_V_TOTAL_MIN,
+		CRTC_V_TOTAL_MIN);
+	dm_write_reg(ctx, addr, value);
+
+	addr = CRTC_REG(mmCRTC_H_BLANK_START_END);
+	value = dm_read_reg(ctx, addr);
+
+	tmp = timing->h_total -
+		(h_sync_start + timing->h_border_left);
+
+	set_reg_field_value(
+		value,
+		tmp,
+		CRTC_H_BLANK_START_END,
+		CRTC_H_BLANK_END);
+
+	tmp = tmp + timing->h_addressable +
+		timing->h_border_left + timing->h_border_right;
+
+	set_reg_field_value(
+		value,
+		tmp,
+		CRTC_H_BLANK_START_END,
+		CRTC_H_BLANK_START);
+
+	dm_write_reg(ctx, addr, value);
+
+	addr = CRTC_REG(mmCRTC_V_BLANK_START_END);
+	value = dm_read_reg(ctx, addr);
+
+	tmp = timing->v_total - (v_sync_start + timing->v_border_top);
+
+	set_reg_field_value(
+		value,
+		tmp,
+		CRTC_V_BLANK_START_END,
+		CRTC_V_BLANK_END);
+
+	tmp = tmp + timing->v_addressable + timing->v_border_top +
+		timing->v_border_bottom;
+
+	set_reg_field_value(
+		value,
+		tmp,
+		CRTC_V_BLANK_START_END,
+		CRTC_V_BLANK_START);
+
+	dm_write_reg(ctx, addr, value);
+}
+
+void dce110_timing_generator_set_test_pattern(
+	struct timing_generator *tg,
+	/* TODO: replace 'controller_dp_test_pattern' by 'test_pattern_mode'
+	 * because this is not DP-specific (which is probably somewhere in DP
+	 * encoder) */
+	enum controller_dp_test_pattern test_pattern,
+	enum dc_color_depth color_depth)
+{
+	struct dc_context *ctx = tg->ctx;
+	uint32_t value;
+	uint32_t addr;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	enum test_pattern_color_format bit_depth;
+	enum test_pattern_dyn_range dyn_range;
+	enum test_pattern_mode mode;
+	/* color ramp generator mixes 16-bits color */
+	uint32_t src_bpc = 16;
+	/* requested bpc */
+	uint32_t dst_bpc;
+	uint32_t index;
+	/* RGB values of the color bars.
+	 * Produce two RGB colors: RGB0 - white (all Fs)
+	 * and RGB1 - black (all 0s)
+	 * (three RGB components for two colors)
+	 */
+	uint16_t src_color[6] = {0xFFFF, 0xFFFF, 0xFFFF, 0x0000,
+						0x0000, 0x0000};
+	/* dest color (converted to the specified color format) */
+	uint16_t dst_color[6];
+	uint32_t inc_base;
+
+	/* translate to bit depth */
+	switch (color_depth) {
+	case COLOR_DEPTH_666:
+		bit_depth = TEST_PATTERN_COLOR_FORMAT_BPC_6;
+	break;
+	case COLOR_DEPTH_888:
+		bit_depth = TEST_PATTERN_COLOR_FORMAT_BPC_8;
+	break;
+	case COLOR_DEPTH_101010:
+		bit_depth = TEST_PATTERN_COLOR_FORMAT_BPC_10;
+	break;
+	case COLOR_DEPTH_121212:
+		bit_depth = TEST_PATTERN_COLOR_FORMAT_BPC_12;
+	break;
+	default:
+		bit_depth = TEST_PATTERN_COLOR_FORMAT_BPC_8;
+	break;
+	}
+
+	switch (test_pattern) {
+	case CONTROLLER_DP_TEST_PATTERN_COLORSQUARES:
+	case CONTROLLER_DP_TEST_PATTERN_COLORSQUARES_CEA:
+	{
+		dyn_range = (test_pattern ==
+				CONTROLLER_DP_TEST_PATTERN_COLORSQUARES_CEA ?
+				TEST_PATTERN_DYN_RANGE_CEA :
+				TEST_PATTERN_DYN_RANGE_VESA);
+		mode = TEST_PATTERN_MODE_COLORSQUARES_RGB;
+		value = 0;
+		addr = CRTC_REG(mmCRTC_TEST_PATTERN_PARAMETERS);
+
+		set_reg_field_value(
+			value,
+			6,
+			CRTC_TEST_PATTERN_PARAMETERS,
+			CRTC_TEST_PATTERN_VRES);
+		set_reg_field_value(
+			value,
+			6,
+			CRTC_TEST_PATTERN_PARAMETERS,
+			CRTC_TEST_PATTERN_HRES);
+
+		dm_write_reg(ctx, addr, value);
+
+		addr = CRTC_REG(mmCRTC_TEST_PATTERN_CONTROL);
+		value = 0;
+
+		set_reg_field_value(
+			value,
+			1,
+			CRTC_TEST_PATTERN_CONTROL,
+			CRTC_TEST_PATTERN_EN);
+
+		set_reg_field_value(
+			value,
+			mode,
+			CRTC_TEST_PATTERN_CONTROL,
+			CRTC_TEST_PATTERN_MODE);
+
+		set_reg_field_value(
+			value,
+			dyn_range,
+			CRTC_TEST_PATTERN_CONTROL,
+			CRTC_TEST_PATTERN_DYNAMIC_RANGE);
+		set_reg_field_value(
+			value,
+			bit_depth,
+			CRTC_TEST_PATTERN_CONTROL,
+			CRTC_TEST_PATTERN_COLOR_FORMAT);
+		dm_write_reg(ctx, addr, value);
+	}
+	break;
+
+	case CONTROLLER_DP_TEST_PATTERN_VERTICALBARS:
+	case CONTROLLER_DP_TEST_PATTERN_HORIZONTALBARS:
+	{
+		mode = (test_pattern ==
+			CONTROLLER_DP_TEST_PATTERN_VERTICALBARS ?
+			TEST_PATTERN_MODE_VERTICALBARS :
+			TEST_PATTERN_MODE_HORIZONTALBARS);
+
+		switch (bit_depth) {
+		case TEST_PATTERN_COLOR_FORMAT_BPC_6:
+			dst_bpc = 6;
+		break;
+		case TEST_PATTERN_COLOR_FORMAT_BPC_8:
+			dst_bpc = 8;
+		break;
+		case TEST_PATTERN_COLOR_FORMAT_BPC_10:
+			dst_bpc = 10;
+		break;
+		default:
+			dst_bpc = 8;
+		break;
+		}
+
+		/* adjust color to the required colorFormat */
+		for (index = 0; index < 6; index++) {
+			/* dst = 2^dstBpc * src / 2^srcBpc = src >>
+			 * (srcBpc - dstBpc);
+			 */
+			dst_color[index] =
+				src_color[index] >> (src_bpc - dst_bpc);
+		/* CRTC_TEST_PATTERN_DATA has 16 bits,
+		 * lowest 6 are hardwired to ZERO
+		 * color bits should be left aligned aligned to MSB
+		 * XXXXXXXXXX000000 for 10 bit,
+		 * XXXXXXXX00000000 for 8 bit and XXXXXX0000000000 for 6
+		 */
+			dst_color[index] <<= (16 - dst_bpc);
+		}
+
+		value = 0;
+		addr = CRTC_REG(mmCRTC_TEST_PATTERN_PARAMETERS);
+		dm_write_reg(ctx, addr, value);
+
+		/* We have to write the mask before data, similar to pipeline.
+		 * For example, for 8 bpc, if we want RGB0 to be magenta,
+		 * and RGB1 to be cyan,
+		 * we need to make 7 writes:
+		 * MASK   DATA
+		 * 000001 00000000 00000000                     set mask to R0
+		 * 000010 11111111 00000000     R0 255, 0xFF00, set mask to G0
+		 * 000100 00000000 00000000     G0 0,   0x0000, set mask to B0
+		 * 001000 11111111 00000000     B0 255, 0xFF00, set mask to R1
+		 * 010000 00000000 00000000     R1 0,   0x0000, set mask to G1
+		 * 100000 11111111 00000000     G1 255, 0xFF00, set mask to B1
+		 * 100000 11111111 00000000     B1 255, 0xFF00
+		 *
+		 * we will make a loop of 6 in which we prepare the mask,
+		 * then write, then prepare the color for next write.
+		 * first iteration will write mask only,
+		 * but each next iteration color prepared in
+		 * previous iteration will be written within new mask,
+		 * the last component will written separately,
+		 * mask is not changing between 6th and 7th write
+		 * and color will be prepared by last iteration
+		 */
+
+		/* write color, color values mask in CRTC_TEST_PATTERN_MASK
+		 * is B1, G1, R1, B0, G0, R0
+		 */
+		value = 0;
+		addr = CRTC_REG(mmCRTC_TEST_PATTERN_COLOR);
+		for (index = 0; index < 6; index++) {
+			/* prepare color mask, first write PATTERN_DATA
+			 * will have all zeros
+			 */
+			set_reg_field_value(
+				value,
+				(1 << index),
+				CRTC_TEST_PATTERN_COLOR,
+				CRTC_TEST_PATTERN_MASK);
+			/* write color component */
+			dm_write_reg(ctx, addr, value);
+			/* prepare next color component,
+			 * will be written in the next iteration
+			 */
+			set_reg_field_value(
+				value,
+				dst_color[index],
+				CRTC_TEST_PATTERN_COLOR,
+				CRTC_TEST_PATTERN_DATA);
+		}
+		/* write last color component,
+		 * it's been already prepared in the loop
+		 */
+		dm_write_reg(ctx, addr, value);
+
+		/* enable test pattern */
+		addr = CRTC_REG(mmCRTC_TEST_PATTERN_CONTROL);
+		value = 0;
+
+		set_reg_field_value(
+			value,
+			1,
+			CRTC_TEST_PATTERN_CONTROL,
+			CRTC_TEST_PATTERN_EN);
+
+		set_reg_field_value(
+			value,
+			mode,
+			CRTC_TEST_PATTERN_CONTROL,
+			CRTC_TEST_PATTERN_MODE);
+
+		set_reg_field_value(
+			value,
+			0,
+			CRTC_TEST_PATTERN_CONTROL,
+			CRTC_TEST_PATTERN_DYNAMIC_RANGE);
+
+		set_reg_field_value(
+			value,
+			bit_depth,
+			CRTC_TEST_PATTERN_CONTROL,
+			CRTC_TEST_PATTERN_COLOR_FORMAT);
+
+		dm_write_reg(ctx, addr, value);
+	}
+	break;
+
+	case CONTROLLER_DP_TEST_PATTERN_COLORRAMP:
+	{
+		mode = (bit_depth ==
+			TEST_PATTERN_COLOR_FORMAT_BPC_10 ?
+			TEST_PATTERN_MODE_DUALRAMP_RGB :
+			TEST_PATTERN_MODE_SINGLERAMP_RGB);
+
+		switch (bit_depth) {
+		case TEST_PATTERN_COLOR_FORMAT_BPC_6:
+			dst_bpc = 6;
+		break;
+		case TEST_PATTERN_COLOR_FORMAT_BPC_8:
+			dst_bpc = 8;
+		break;
+		case TEST_PATTERN_COLOR_FORMAT_BPC_10:
+			dst_bpc = 10;
+		break;
+		default:
+			dst_bpc = 8;
+		break;
+		}
+
+		/* increment for the first ramp for one color gradation
+		 * 1 gradation for 6-bit color is 2^10
+		 * gradations in 16-bit color
+		 */
+		inc_base = (src_bpc - dst_bpc);
+
+		value = 0;
+		addr = CRTC_REG(mmCRTC_TEST_PATTERN_PARAMETERS);
+
+		switch (bit_depth) {
+		case TEST_PATTERN_COLOR_FORMAT_BPC_6:
+		{
+			set_reg_field_value(
+				value,
+				inc_base,
+				CRTC_TEST_PATTERN_PARAMETERS,
+				CRTC_TEST_PATTERN_INC0);
+			set_reg_field_value(
+				value,
+				0,
+				CRTC_TEST_PATTERN_PARAMETERS,
+				CRTC_TEST_PATTERN_INC1);
+			set_reg_field_value(
+				value,
+				6,
+				CRTC_TEST_PATTERN_PARAMETERS,
+				CRTC_TEST_PATTERN_HRES);
+			set_reg_field_value(
+				value,
+				6,
+				CRTC_TEST_PATTERN_PARAMETERS,
+				CRTC_TEST_PATTERN_VRES);
+			set_reg_field_value(
+				value,
+				0,
+				CRTC_TEST_PATTERN_PARAMETERS,
+				CRTC_TEST_PATTERN_RAMP0_OFFSET);
+		}
+		break;
+		case TEST_PATTERN_COLOR_FORMAT_BPC_8:
+		{
+			set_reg_field_value(
+				value,
+				inc_base,
+				CRTC_TEST_PATTERN_PARAMETERS,
+				CRTC_TEST_PATTERN_INC0);
+			set_reg_field_value(
+				value,
+				0,
+				CRTC_TEST_PATTERN_PARAMETERS,
+				CRTC_TEST_PATTERN_INC1);
+			set_reg_field_value(
+				value,
+				8,
+				CRTC_TEST_PATTERN_PARAMETERS,
+				CRTC_TEST_PATTERN_HRES);
+			set_reg_field_value(
+				value,
+				6,
+				CRTC_TEST_PATTERN_PARAMETERS,
+				CRTC_TEST_PATTERN_VRES);
+			set_reg_field_value(
+				value,
+				0,
+				CRTC_TEST_PATTERN_PARAMETERS,
+				CRTC_TEST_PATTERN_RAMP0_OFFSET);
+		}
+		break;
+		case TEST_PATTERN_COLOR_FORMAT_BPC_10:
+		{
+			set_reg_field_value(
+				value,
+				inc_base,
+				CRTC_TEST_PATTERN_PARAMETERS,
+				CRTC_TEST_PATTERN_INC0);
+			set_reg_field_value(
+				value,
+				inc_base + 2,
+				CRTC_TEST_PATTERN_PARAMETERS,
+				CRTC_TEST_PATTERN_INC1);
+			set_reg_field_value(
+				value,
+				8,
+				CRTC_TEST_PATTERN_PARAMETERS,
+				CRTC_TEST_PATTERN_HRES);
+			set_reg_field_value(
+				value,
+				5,
+				CRTC_TEST_PATTERN_PARAMETERS,
+				CRTC_TEST_PATTERN_VRES);
+			set_reg_field_value(
+				value,
+				384 << 6,
+				CRTC_TEST_PATTERN_PARAMETERS,
+				CRTC_TEST_PATTERN_RAMP0_OFFSET);
+		}
+		break;
+		default:
+		break;
+		}
+		dm_write_reg(ctx, addr, value);
+
+		value = 0;
+		addr = CRTC_REG(mmCRTC_TEST_PATTERN_COLOR);
+		dm_write_reg(ctx, addr, value);
+
+		/* enable test pattern */
+		addr = CRTC_REG(mmCRTC_TEST_PATTERN_CONTROL);
+		value = 0;
+
+		set_reg_field_value(
+			value,
+			1,
+			CRTC_TEST_PATTERN_CONTROL,
+			CRTC_TEST_PATTERN_EN);
+
+		set_reg_field_value(
+			value,
+			mode,
+			CRTC_TEST_PATTERN_CONTROL,
+			CRTC_TEST_PATTERN_MODE);
+
+		set_reg_field_value(
+			value,
+			0,
+			CRTC_TEST_PATTERN_CONTROL,
+			CRTC_TEST_PATTERN_DYNAMIC_RANGE);
+		/* add color depth translation here */
+		set_reg_field_value(
+			value,
+			bit_depth,
+			CRTC_TEST_PATTERN_CONTROL,
+			CRTC_TEST_PATTERN_COLOR_FORMAT);
+
+		dm_write_reg(ctx, addr, value);
+	}
+	break;
+	case CONTROLLER_DP_TEST_PATTERN_VIDEOMODE:
+	{
+		value = 0;
+		dm_write_reg(ctx, CRTC_REG(mmCRTC_TEST_PATTERN_CONTROL), value);
+		dm_write_reg(ctx, CRTC_REG(mmCRTC_TEST_PATTERN_COLOR), value);
+		dm_write_reg(ctx, CRTC_REG(mmCRTC_TEST_PATTERN_PARAMETERS),
+				value);
+	}
+	break;
+	default:
+	break;
+	}
+}
+
+/*
+ * dce110_timing_generator_validate_timing
+ * The timing generators support a maximum display size of is 8192 x 8192 pixels,
+ * including both active display and blanking periods. Check H Total and V Total.
+ */
+bool dce110_timing_generator_validate_timing(
+	struct timing_generator *tg,
+	const struct dc_crtc_timing *timing,
+	enum signal_type signal)
+{
+	uint32_t h_blank;
+	uint32_t h_back_porch, hsync_offset, h_sync_start;
+
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	ASSERT(timing != NULL);
+
+	if (!timing)
+		return false;
+
+	hsync_offset = timing->h_border_right + timing->h_front_porch;
+	h_sync_start = timing->h_addressable + hsync_offset;
+
+	/* Currently we don't support 3D, so block all 3D timings */
+	if (timing->timing_3d_format != TIMING_3D_FORMAT_NONE)
+		return false;
+
+	/* Temporarily blocking interlacing mode until it's supported */
+	if (timing->flags.INTERLACE == 1)
+		return false;
+
+	/* Check maximum number of pixels supported by Timing Generator
+	 * (Currently will never fail, in order to fail needs display which
+	 * needs more than 8192 horizontal and
+	 * more than 8192 vertical total pixels)
+	 */
+	if (timing->h_total > tg110->max_h_total ||
+		timing->v_total > tg110->max_v_total)
+		return false;
+
+	h_blank = (timing->h_total - timing->h_addressable -
+		timing->h_border_right -
+		timing->h_border_left);
+
+	if (h_blank < tg110->min_h_blank)
+		return false;
+
+	if (timing->h_front_porch < tg110->min_h_front_porch)
+		return false;
+
+	h_back_porch = h_blank - (h_sync_start -
+		timing->h_addressable -
+		timing->h_border_right -
+		timing->h_sync_width);
+
+	if (h_back_porch < tg110->min_h_back_porch)
+		return false;
+
+	return true;
+}
+
+/*
+ * Wait till we are at the beginning of VBlank.
+ */
+void dce110_timing_generator_wait_for_vblank(struct timing_generator *tg)
+{
+	/* We want to catch beginning of VBlank here, so if the first try are
+	 * in VBlank, we might be very close to Active, in this case wait for
+	 * another frame
+	 */
+	while (dce110_timing_generator_is_in_vertical_blank(tg)) {
+		if (!dce110_timing_generator_is_counter_moving(tg)) {
+			/* error - no point to wait if counter is not moving */
+			break;
+		}
+	}
+
+	while (!dce110_timing_generator_is_in_vertical_blank(tg)) {
+		if (!dce110_timing_generator_is_counter_moving(tg)) {
+			/* error - no point to wait if counter is not moving */
+			break;
+		}
+	}
+}
+
+/*
+ * Wait till we are in VActive (anywhere in VActive)
+ */
+void dce110_timing_generator_wait_for_vactive(struct timing_generator *tg)
+{
+	while (dce110_timing_generator_is_in_vertical_blank(tg)) {
+		if (!dce110_timing_generator_is_counter_moving(tg)) {
+			/* error - no point to wait if counter is not moving */
+			break;
+		}
+	}
+}
+
+/*
+ *****************************************************************************
+ *  Function: dce110_timing_generator_setup_global_swap_lock
+ *
+ *  @brief
+ *     Setups Global Swap Lock group for current pipe
+ *     Pipe can join or leave GSL group, become a TimingServer or TimingClient
+ *
+ *  @param [in] gsl_params: setup data
+ *****************************************************************************
+ */
+void dce110_timing_generator_setup_global_swap_lock(
+	struct timing_generator *tg,
+	const struct dcp_gsl_params *gsl_params)
+{
+	uint32_t value;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t address = DCP_REG(mmDCP_GSL_CONTROL);
+	uint32_t check_point = FLIP_READY_BACK_LOOKUP;
+
+	value = dm_read_reg(tg->ctx, address);
+
+	/* This pipe will belong to GSL Group zero. */
+	set_reg_field_value(value,
+			    1,
+			    DCP_GSL_CONTROL,
+			    DCP_GSL0_EN);
+
+	set_reg_field_value(value,
+			    gsl_params->gsl_master == tg->inst,
+			    DCP_GSL_CONTROL,
+			    DCP_GSL_MASTER_EN);
+
+	set_reg_field_value(value,
+			    HFLIP_READY_DELAY,
+			    DCP_GSL_CONTROL,
+			    DCP_GSL_HSYNC_FLIP_FORCE_DELAY);
+
+	/* Keep signal low (pending high) during 6 lines.
+	 * Also defines minimum interval before re-checking signal. */
+	set_reg_field_value(value,
+			    HFLIP_CHECK_DELAY,
+			    DCP_GSL_CONTROL,
+			    DCP_GSL_HSYNC_FLIP_CHECK_DELAY);
+
+	dm_write_reg(tg->ctx, CRTC_REG(mmDCP_GSL_CONTROL), value);
+	value = 0;
+
+	set_reg_field_value(value,
+			    gsl_params->gsl_master,
+			    DCIO_GSL0_CNTL,
+			    DCIO_GSL0_VSYNC_SEL);
+
+	set_reg_field_value(value,
+			    0,
+			    DCIO_GSL0_CNTL,
+			    DCIO_GSL0_TIMING_SYNC_SEL);
+
+	set_reg_field_value(value,
+			    0,
+			    DCIO_GSL0_CNTL,
+			    DCIO_GSL0_GLOBAL_UNLOCK_SEL);
+
+	dm_write_reg(tg->ctx, CRTC_REG(mmDCIO_GSL0_CNTL), value);
+
+
+	{
+		uint32_t value_crtc_vtotal;
+
+		value_crtc_vtotal = dm_read_reg(tg->ctx,
+				CRTC_REG(mmCRTC_V_TOTAL));
+
+		set_reg_field_value(value,
+				    0,/* DCP_GSL_PURPOSE_SURFACE_FLIP */
+				    DCP_GSL_CONTROL,
+				    DCP_GSL_SYNC_SOURCE);
+
+		/* Checkpoint relative to end of frame */
+		check_point = get_reg_field_value(value_crtc_vtotal,
+						  CRTC_V_TOTAL,
+						  CRTC_V_TOTAL);
+
+		dm_write_reg(tg->ctx, CRTC_REG(mmCRTC_GSL_WINDOW), 0);
+	}
+
+	set_reg_field_value(value,
+			    1,
+			    DCP_GSL_CONTROL,
+			    DCP_GSL_DELAY_SURFACE_UPDATE_PENDING);
+
+	dm_write_reg(tg->ctx, address, value);
+
+	/********************************************************************/
+	address = CRTC_REG(mmCRTC_GSL_CONTROL);
+
+	value = dm_read_reg(tg->ctx, address);
+	set_reg_field_value(value,
+			    check_point - FLIP_READY_BACK_LOOKUP,
+			    CRTC_GSL_CONTROL,
+			    CRTC_GSL_CHECK_LINE_NUM);
+
+	set_reg_field_value(value,
+			    VFLIP_READY_DELAY,
+			    CRTC_GSL_CONTROL,
+			    CRTC_GSL_FORCE_DELAY);
+
+	dm_write_reg(tg->ctx, address, value);
+}
+
+void dce110_timing_generator_tear_down_global_swap_lock(
+	struct timing_generator *tg)
+{
+	/* Clear all the register writes done by
+	 * dce110_timing_generator_setup_global_swap_lock
+	 */
+
+	uint32_t value;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t address = DCP_REG(mmDCP_GSL_CONTROL);
+
+	value = 0;
+
+	/* This pipe will belong to GSL Group zero. */
+	/* Settig HW default values from reg specs */
+	set_reg_field_value(value,
+			0,
+			DCP_GSL_CONTROL,
+			DCP_GSL0_EN);
+
+	set_reg_field_value(value,
+			0,
+			DCP_GSL_CONTROL,
+			DCP_GSL_MASTER_EN);
+
+	set_reg_field_value(value,
+			0x2,
+			DCP_GSL_CONTROL,
+			DCP_GSL_HSYNC_FLIP_FORCE_DELAY);
+
+	set_reg_field_value(value,
+			0x6,
+			DCP_GSL_CONTROL,
+			DCP_GSL_HSYNC_FLIP_CHECK_DELAY);
+
+	/* Restore DCP_GSL_PURPOSE_SURFACE_FLIP */
+	{
+		dm_read_reg(tg->ctx, CRTC_REG(mmCRTC_V_TOTAL));
+
+		set_reg_field_value(value,
+				0,
+				DCP_GSL_CONTROL,
+				DCP_GSL_SYNC_SOURCE);
+	}
+
+	set_reg_field_value(value,
+			0,
+			DCP_GSL_CONTROL,
+			DCP_GSL_DELAY_SURFACE_UPDATE_PENDING);
+
+	dm_write_reg(tg->ctx, address, value);
+
+	/********************************************************************/
+	address = CRTC_REG(mmCRTC_GSL_CONTROL);
+
+	value = 0;
+	set_reg_field_value(value,
+			0,
+			CRTC_GSL_CONTROL,
+			CRTC_GSL_CHECK_LINE_NUM);
+
+	set_reg_field_value(value,
+			0x2,
+			CRTC_GSL_CONTROL,
+			CRTC_GSL_FORCE_DELAY);
+
+	dm_write_reg(tg->ctx, address, value);
+}
+/*
+ *****************************************************************************
+ *  Function: is_counter_moving
+ *
+ *  @brief
+ *     check if the timing generator is currently going
+ *
+ *  @return
+ *     true if currently going, false if currently paused or stopped.
+ *
+ *****************************************************************************
+ */
+bool dce110_timing_generator_is_counter_moving(struct timing_generator *tg)
+{
+	struct crtc_position position1, position2;
+
+	tg->funcs->get_position(tg, &position1);
+	tg->funcs->get_position(tg, &position2);
+
+	if (position1.horizontal_count == position2.horizontal_count &&
+		position1.vertical_count == position2.vertical_count)
+		return false;
+	else
+		return true;
+}
+
+void dce110_timing_generator_enable_advanced_request(
+	struct timing_generator *tg,
+	bool enable,
+	const struct dc_crtc_timing *timing)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t addr = CRTC_REG(mmCRTC_START_LINE_CONTROL);
+	uint32_t value = dm_read_reg(tg->ctx, addr);
+
+	if (enable) {
+		set_reg_field_value(
+			value,
+			0,
+			CRTC_START_LINE_CONTROL,
+			CRTC_LEGACY_REQUESTOR_EN);
+	} else {
+		set_reg_field_value(
+			value,
+			1,
+			CRTC_START_LINE_CONTROL,
+			CRTC_LEGACY_REQUESTOR_EN);
+	}
+
+	if ((timing->v_sync_width + timing->v_front_porch) <= 3) {
+		set_reg_field_value(
+			value,
+			3,
+			CRTC_START_LINE_CONTROL,
+			CRTC_ADVANCED_START_LINE_POSITION);
+		set_reg_field_value(
+			value,
+			0,
+			CRTC_START_LINE_CONTROL,
+			CRTC_PREFETCH_EN);
+	} else {
+		set_reg_field_value(
+			value,
+			4,
+			CRTC_START_LINE_CONTROL,
+			CRTC_ADVANCED_START_LINE_POSITION);
+		set_reg_field_value(
+			value,
+			1,
+			CRTC_START_LINE_CONTROL,
+			CRTC_PREFETCH_EN);
+	}
+
+	set_reg_field_value(
+		value,
+		1,
+		CRTC_START_LINE_CONTROL,
+		CRTC_PROGRESSIVE_START_LINE_EARLY);
+
+	set_reg_field_value(
+		value,
+		1,
+		CRTC_START_LINE_CONTROL,
+		CRTC_INTERLACE_START_LINE_EARLY);
+
+	dm_write_reg(tg->ctx, addr, value);
+}
+
+/*TODO: Figure out if we need this function. */
+void dce110_timing_generator_set_lock_master(struct timing_generator *tg,
+		bool lock)
+{
+	struct dc_context *ctx = tg->ctx;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t addr = CRTC_REG(mmCRTC_MASTER_UPDATE_LOCK);
+	uint32_t value = dm_read_reg(ctx, addr);
+
+	set_reg_field_value(
+		value,
+		lock ? 1 : 0,
+		CRTC_MASTER_UPDATE_LOCK,
+		MASTER_UPDATE_LOCK);
+
+	dm_write_reg(ctx, addr, value);
+}
+
+void dce110_timing_generator_enable_reset_trigger(
+	struct timing_generator *tg,
+	int source_tg_inst)
+{
+	uint32_t value;
+	uint32_t rising_edge = 0;
+	uint32_t falling_edge = 0;
+	enum trigger_source_select trig_src_select = TRIGGER_SOURCE_SELECT_LOGIC_ZERO;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	/* Setup trigger edge */
+	{
+		uint32_t pol_value = dm_read_reg(tg->ctx,
+				CRTC_REG(mmCRTC_V_SYNC_A_CNTL));
+
+		/* Register spec has reversed definition:
+		 *	0 for positive, 1 for negative */
+		if (get_reg_field_value(pol_value,
+				CRTC_V_SYNC_A_CNTL,
+				CRTC_V_SYNC_A_POL) == 0) {
+			rising_edge = 1;
+		} else {
+			falling_edge = 1;
+		}
+	}
+
+	value = dm_read_reg(tg->ctx, CRTC_REG(mmCRTC_TRIGB_CNTL));
+
+	trig_src_select = TRIGGER_SOURCE_SELECT_GSL_GROUP0;
+
+	set_reg_field_value(value,
+			trig_src_select,
+			CRTC_TRIGB_CNTL,
+			CRTC_TRIGB_SOURCE_SELECT);
+
+	set_reg_field_value(value,
+			TRIGGER_POLARITY_SELECT_LOGIC_ZERO,
+			CRTC_TRIGB_CNTL,
+			CRTC_TRIGB_POLARITY_SELECT);
+
+	set_reg_field_value(value,
+			rising_edge,
+			CRTC_TRIGB_CNTL,
+			CRTC_TRIGB_RISING_EDGE_DETECT_CNTL);
+
+	set_reg_field_value(value,
+			falling_edge,
+			CRTC_TRIGB_CNTL,
+			CRTC_TRIGB_FALLING_EDGE_DETECT_CNTL);
+
+	set_reg_field_value(value,
+			0, /* send every signal */
+			CRTC_TRIGB_CNTL,
+			CRTC_TRIGB_FREQUENCY_SELECT);
+
+	set_reg_field_value(value,
+			0, /* no delay */
+			CRTC_TRIGB_CNTL,
+			CRTC_TRIGB_DELAY);
+
+	set_reg_field_value(value,
+			1, /* clear trigger status */
+			CRTC_TRIGB_CNTL,
+			CRTC_TRIGB_CLEAR);
+
+	dm_write_reg(tg->ctx, CRTC_REG(mmCRTC_TRIGB_CNTL), value);
+
+	/**************************************************************/
+
+	value = dm_read_reg(tg->ctx, CRTC_REG(mmCRTC_FORCE_COUNT_NOW_CNTL));
+
+	set_reg_field_value(value,
+			2, /* force H count to H_TOTAL and V count to V_TOTAL */
+			CRTC_FORCE_COUNT_NOW_CNTL,
+			CRTC_FORCE_COUNT_NOW_MODE);
+
+	set_reg_field_value(value,
+			1, /* TriggerB - we never use TriggerA */
+			CRTC_FORCE_COUNT_NOW_CNTL,
+			CRTC_FORCE_COUNT_NOW_TRIG_SEL);
+
+	set_reg_field_value(value,
+			1, /* clear trigger status */
+			CRTC_FORCE_COUNT_NOW_CNTL,
+			CRTC_FORCE_COUNT_NOW_CLEAR);
+
+	dm_write_reg(tg->ctx, CRTC_REG(mmCRTC_FORCE_COUNT_NOW_CNTL), value);
+}
+
+void dce110_timing_generator_enable_crtc_reset(
+		struct timing_generator *tg,
+		int source_tg_inst,
+		struct crtc_trigger_info *crtc_tp)
+{
+	uint32_t value = 0;
+	uint32_t rising_edge = 0;
+	uint32_t falling_edge = 0;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	/* Setup trigger edge */
+	switch (crtc_tp->event) {
+	case CRTC_EVENT_VSYNC_RISING:
+			rising_edge = 1;
+			break;
+
+	case CRTC_EVENT_VSYNC_FALLING:
+		falling_edge = 1;
+		break;
+	}
+
+	value = dm_read_reg(tg->ctx, CRTC_REG(mmCRTC_TRIGB_CNTL));
+
+	set_reg_field_value(value,
+			    source_tg_inst,
+			    CRTC_TRIGB_CNTL,
+			    CRTC_TRIGB_SOURCE_SELECT);
+
+	set_reg_field_value(value,
+			    TRIGGER_POLARITY_SELECT_LOGIC_ZERO,
+			    CRTC_TRIGB_CNTL,
+			    CRTC_TRIGB_POLARITY_SELECT);
+
+	set_reg_field_value(value,
+			    rising_edge,
+			    CRTC_TRIGB_CNTL,
+			    CRTC_TRIGB_RISING_EDGE_DETECT_CNTL);
+
+	set_reg_field_value(value,
+			    falling_edge,
+			    CRTC_TRIGB_CNTL,
+			    CRTC_TRIGB_FALLING_EDGE_DETECT_CNTL);
+
+	set_reg_field_value(value,
+			    1, /* clear trigger status */
+			    CRTC_TRIGB_CNTL,
+			    CRTC_TRIGB_CLEAR);
+
+	dm_write_reg(tg->ctx, CRTC_REG(mmCRTC_TRIGB_CNTL), value);
+
+	/**************************************************************/
+
+	switch (crtc_tp->delay) {
+	case TRIGGER_DELAY_NEXT_LINE:
+		value = dm_read_reg(tg->ctx, CRTC_REG(mmCRTC_FORCE_COUNT_NOW_CNTL));
+
+		set_reg_field_value(value,
+				    0, /* force H count to H_TOTAL and V count to V_TOTAL */
+				    CRTC_FORCE_COUNT_NOW_CNTL,
+				    CRTC_FORCE_COUNT_NOW_MODE);
+
+		set_reg_field_value(value,
+				    0, /* TriggerB - we never use TriggerA */
+				    CRTC_FORCE_COUNT_NOW_CNTL,
+				    CRTC_FORCE_COUNT_NOW_TRIG_SEL);
+
+		set_reg_field_value(value,
+				    1, /* clear trigger status */
+				    CRTC_FORCE_COUNT_NOW_CNTL,
+				    CRTC_FORCE_COUNT_NOW_CLEAR);
+
+		dm_write_reg(tg->ctx, CRTC_REG(mmCRTC_FORCE_COUNT_NOW_CNTL), value);
+
+		value = dm_read_reg(tg->ctx, CRTC_REG(mmCRTC_VERT_SYNC_CONTROL));
+
+		set_reg_field_value(value,
+				    1,
+				    CRTC_VERT_SYNC_CONTROL,
+				    CRTC_FORCE_VSYNC_NEXT_LINE_CLEAR);
+
+		set_reg_field_value(value,
+				    2,
+				    CRTC_VERT_SYNC_CONTROL,
+				    CRTC_AUTO_FORCE_VSYNC_MODE);
+
+		break;
+
+	case TRIGGER_DELAY_NEXT_PIXEL:
+		value = dm_read_reg(tg->ctx, CRTC_REG(mmCRTC_VERT_SYNC_CONTROL));
+
+		set_reg_field_value(value,
+				    1,
+				    CRTC_VERT_SYNC_CONTROL,
+				    CRTC_FORCE_VSYNC_NEXT_LINE_CLEAR);
+
+		set_reg_field_value(value,
+				    0,
+				    CRTC_VERT_SYNC_CONTROL,
+				    CRTC_AUTO_FORCE_VSYNC_MODE);
+
+		dm_write_reg(tg->ctx, CRTC_REG(mmCRTC_VERT_SYNC_CONTROL), value);
+
+		value = dm_read_reg(tg->ctx, CRTC_REG(mmCRTC_FORCE_COUNT_NOW_CNTL));
+
+		set_reg_field_value(value,
+				    2, /* force H count to H_TOTAL and V count to V_TOTAL */
+				    CRTC_FORCE_COUNT_NOW_CNTL,
+				    CRTC_FORCE_COUNT_NOW_MODE);
+
+		set_reg_field_value(value,
+				    1, /* TriggerB - we never use TriggerA */
+				    CRTC_FORCE_COUNT_NOW_CNTL,
+				    CRTC_FORCE_COUNT_NOW_TRIG_SEL);
+
+		set_reg_field_value(value,
+				    1, /* clear trigger status */
+				    CRTC_FORCE_COUNT_NOW_CNTL,
+				    CRTC_FORCE_COUNT_NOW_CLEAR);
+
+		dm_write_reg(tg->ctx, CRTC_REG(mmCRTC_FORCE_COUNT_NOW_CNTL), value);
+		break;
+	}
+
+	value = dm_read_reg(tg->ctx, CRTC_REG(mmCRTC_MASTER_UPDATE_MODE));
+
+	set_reg_field_value(value,
+			    2,
+			    CRTC_MASTER_UPDATE_MODE,
+			    MASTER_UPDATE_MODE);
+
+	dm_write_reg(tg->ctx, CRTC_REG(mmCRTC_MASTER_UPDATE_MODE), value);
+}
+void dce110_timing_generator_disable_reset_trigger(
+	struct timing_generator *tg)
+{
+	uint32_t value;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	value = dm_read_reg(tg->ctx, CRTC_REG(mmCRTC_FORCE_COUNT_NOW_CNTL));
+
+	set_reg_field_value(value,
+			    0, /* force counter now mode is disabled */
+			    CRTC_FORCE_COUNT_NOW_CNTL,
+			    CRTC_FORCE_COUNT_NOW_MODE);
+
+	set_reg_field_value(value,
+			    1, /* clear trigger status */
+			    CRTC_FORCE_COUNT_NOW_CNTL,
+			    CRTC_FORCE_COUNT_NOW_CLEAR);
+
+	dm_write_reg(tg->ctx, CRTC_REG(mmCRTC_FORCE_COUNT_NOW_CNTL), value);
+
+	value = dm_read_reg(tg->ctx, CRTC_REG(mmCRTC_VERT_SYNC_CONTROL));
+
+	set_reg_field_value(value,
+			    1,
+			    CRTC_VERT_SYNC_CONTROL,
+			    CRTC_FORCE_VSYNC_NEXT_LINE_CLEAR);
+
+	set_reg_field_value(value,
+			    0,
+			    CRTC_VERT_SYNC_CONTROL,
+			    CRTC_AUTO_FORCE_VSYNC_MODE);
+
+	dm_write_reg(tg->ctx, CRTC_REG(mmCRTC_VERT_SYNC_CONTROL), value);
+
+	/********************************************************************/
+	value = dm_read_reg(tg->ctx, CRTC_REG(mmCRTC_TRIGB_CNTL));
+
+	set_reg_field_value(value,
+			    TRIGGER_SOURCE_SELECT_LOGIC_ZERO,
+			    CRTC_TRIGB_CNTL,
+			    CRTC_TRIGB_SOURCE_SELECT);
+
+	set_reg_field_value(value,
+			    TRIGGER_POLARITY_SELECT_LOGIC_ZERO,
+			    CRTC_TRIGB_CNTL,
+			    CRTC_TRIGB_POLARITY_SELECT);
+
+	set_reg_field_value(value,
+			    1, /* clear trigger status */
+			    CRTC_TRIGB_CNTL,
+			    CRTC_TRIGB_CLEAR);
+
+	dm_write_reg(tg->ctx, CRTC_REG(mmCRTC_TRIGB_CNTL), value);
+}
+
+/*
+ *****************************************************************************
+ *  @brief
+ *     Checks whether CRTC triggered reset occurred
+ *
+ *  @return
+ *     true if triggered reset occurred, false otherwise
+ *****************************************************************************
+ */
+bool dce110_timing_generator_did_triggered_reset_occur(
+	struct timing_generator *tg)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t value = dm_read_reg(tg->ctx,
+			CRTC_REG(mmCRTC_FORCE_COUNT_NOW_CNTL));
+	uint32_t value1 = dm_read_reg(tg->ctx,
+			CRTC_REG(mmCRTC_VERT_SYNC_CONTROL));
+	bool force = get_reg_field_value(value,
+					 CRTC_FORCE_COUNT_NOW_CNTL,
+					 CRTC_FORCE_COUNT_NOW_OCCURRED) != 0;
+	bool vert_sync = get_reg_field_value(value1,
+					     CRTC_VERT_SYNC_CONTROL,
+					     CRTC_FORCE_VSYNC_NEXT_LINE_OCCURRED) != 0;
+
+	return (force || vert_sync);
+}
+
+/*
+ * dce110_timing_generator_disable_vga
+ * Turn OFF VGA Mode and Timing  - DxVGA_CONTROL
+ * VGA Mode and VGA Timing is used by VBIOS on CRT Monitors;
+ */
+void dce110_timing_generator_disable_vga(
+	struct timing_generator *tg)
+{
+	uint32_t addr = 0;
+	uint32_t value = 0;
+
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	switch (tg110->controller_id) {
+	case CONTROLLER_ID_D0:
+		addr = mmD1VGA_CONTROL;
+		break;
+	case CONTROLLER_ID_D1:
+		addr = mmD2VGA_CONTROL;
+		break;
+	case CONTROLLER_ID_D2:
+		addr = mmD3VGA_CONTROL;
+		break;
+	case CONTROLLER_ID_D3:
+		addr = mmD4VGA_CONTROL;
+		break;
+	case CONTROLLER_ID_D4:
+		addr = mmD5VGA_CONTROL;
+		break;
+	case CONTROLLER_ID_D5:
+		addr = mmD6VGA_CONTROL;
+		break;
+	default:
+		break;
+	}
+	value = dm_read_reg(tg->ctx, addr);
+
+	set_reg_field_value(value, 0, D1VGA_CONTROL, D1VGA_MODE_ENABLE);
+	set_reg_field_value(value, 0, D1VGA_CONTROL, D1VGA_TIMING_SELECT);
+	set_reg_field_value(
+			value, 0, D1VGA_CONTROL, D1VGA_SYNC_POLARITY_SELECT);
+	set_reg_field_value(value, 0, D1VGA_CONTROL, D1VGA_OVERSCAN_COLOR_EN);
+
+	dm_write_reg(tg->ctx, addr, value);
+}
+
+/*
+ * set_overscan_color_black
+ *
+ * @param :black_color is one of the color space
+ *    :this routine will set overscan black color according to the color space.
+ * @return none
+ */
+void dce110_timing_generator_set_overscan_color_black(
+	struct timing_generator *tg,
+	const struct tg_color *color)
+{
+	struct dc_context *ctx = tg->ctx;
+	uint32_t addr;
+	uint32_t value = 0;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	set_reg_field_value(
+			value,
+			color->color_b_cb,
+			CRTC_OVERSCAN_COLOR,
+			CRTC_OVERSCAN_COLOR_BLUE);
+
+	set_reg_field_value(
+			value,
+			color->color_r_cr,
+			CRTC_OVERSCAN_COLOR,
+			CRTC_OVERSCAN_COLOR_RED);
+
+	set_reg_field_value(
+			value,
+			color->color_g_y,
+			CRTC_OVERSCAN_COLOR,
+			CRTC_OVERSCAN_COLOR_GREEN);
+
+	addr = CRTC_REG(mmCRTC_OVERSCAN_COLOR);
+	dm_write_reg(ctx, addr, value);
+	addr = CRTC_REG(mmCRTC_BLACK_COLOR);
+	dm_write_reg(ctx, addr, value);
+	/* This is desirable to have a constant DAC output voltage during the
+	 * blank time that is higher than the 0 volt reference level that the
+	 * DAC outputs when the NBLANK signal
+	 * is asserted low, such as for output to an analog TV. */
+	addr = CRTC_REG(mmCRTC_BLANK_DATA_COLOR);
+	dm_write_reg(ctx, addr, value);
+
+	/* TO DO we have to program EXT registers and we need to know LB DATA
+	 * format because it is used when more 10 , i.e. 12 bits per color
+	 *
+	 * m_mmDxCRTC_OVERSCAN_COLOR_EXT
+	 * m_mmDxCRTC_BLACK_COLOR_EXT
+	 * m_mmDxCRTC_BLANK_DATA_COLOR_EXT
+	 */
+
+}
+
+void dce110_tg_program_blank_color(struct timing_generator *tg,
+		const struct tg_color *black_color)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t addr = CRTC_REG(mmCRTC_BLACK_COLOR);
+	uint32_t value = dm_read_reg(tg->ctx, addr);
+
+	set_reg_field_value(
+		value,
+		black_color->color_b_cb,
+		CRTC_BLACK_COLOR,
+		CRTC_BLACK_COLOR_B_CB);
+	set_reg_field_value(
+		value,
+		black_color->color_g_y,
+		CRTC_BLACK_COLOR,
+		CRTC_BLACK_COLOR_G_Y);
+	set_reg_field_value(
+		value,
+		black_color->color_r_cr,
+		CRTC_BLACK_COLOR,
+		CRTC_BLACK_COLOR_R_CR);
+
+	dm_write_reg(tg->ctx, addr, value);
+
+	addr = CRTC_REG(mmCRTC_BLANK_DATA_COLOR);
+	dm_write_reg(tg->ctx, addr, value);
+}
+
+void dce110_tg_set_overscan_color(struct timing_generator *tg,
+	const struct tg_color *overscan_color)
+{
+	struct dc_context *ctx = tg->ctx;
+	uint32_t value = 0;
+	uint32_t addr;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	set_reg_field_value(
+		value,
+		overscan_color->color_b_cb,
+		CRTC_OVERSCAN_COLOR,
+		CRTC_OVERSCAN_COLOR_BLUE);
+
+	set_reg_field_value(
+		value,
+		overscan_color->color_g_y,
+		CRTC_OVERSCAN_COLOR,
+		CRTC_OVERSCAN_COLOR_GREEN);
+
+	set_reg_field_value(
+		value,
+		overscan_color->color_r_cr,
+		CRTC_OVERSCAN_COLOR,
+		CRTC_OVERSCAN_COLOR_RED);
+
+	addr = CRTC_REG(mmCRTC_OVERSCAN_COLOR);
+	dm_write_reg(ctx, addr, value);
+}
+
+void dce110_tg_program_timing(struct timing_generator *tg,
+	const struct dc_crtc_timing *timing,
+	int vready_offset,
+	int vstartup_start,
+	int vupdate_offset,
+	int vupdate_width,
+	int pstate_keepout,
+	const enum signal_type signal,
+	bool use_vbios)
+{
+	if (use_vbios)
+		dce110_timing_generator_program_timing_generator(tg, timing);
+	else
+		dce110_timing_generator_program_blanking(tg, timing);
+}
+
+bool dce110_tg_is_blanked(struct timing_generator *tg)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t value = dm_read_reg(tg->ctx, CRTC_REG(mmCRTC_BLANK_CONTROL));
+
+	if (get_reg_field_value(
+			value,
+			CRTC_BLANK_CONTROL,
+			CRTC_BLANK_DATA_EN) == 1 &&
+		get_reg_field_value(
+			value,
+			CRTC_BLANK_CONTROL,
+			CRTC_CURRENT_BLANK_STATE) == 1)
+		return true;
+	return false;
+}
+
+void dce110_tg_set_blank(struct timing_generator *tg,
+		bool enable_blanking)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t value = 0;
+
+	set_reg_field_value(
+		value,
+		1,
+		CRTC_DOUBLE_BUFFER_CONTROL,
+		CRTC_BLANK_DATA_DOUBLE_BUFFER_EN);
+
+	dm_write_reg(tg->ctx, CRTC_REG(mmCRTC_DOUBLE_BUFFER_CONTROL), value);
+	value = 0;
+
+	if (enable_blanking) {
+		set_reg_field_value(
+			value,
+			1,
+			CRTC_BLANK_CONTROL,
+			CRTC_BLANK_DATA_EN);
+
+		dm_write_reg(tg->ctx, CRTC_REG(mmCRTC_BLANK_CONTROL), value);
+
+	} else
+		dm_write_reg(tg->ctx, CRTC_REG(mmCRTC_BLANK_CONTROL), 0);
+}
+
+bool dce110_tg_validate_timing(struct timing_generator *tg,
+	const struct dc_crtc_timing *timing)
+{
+	return dce110_timing_generator_validate_timing(tg, timing, SIGNAL_TYPE_NONE);
+}
+
+/* "Container" vs. "pixel" is a concept within HW blocks, mostly those closer to the back-end. It works like this:
+ *
+ * - In most of the formats (RGB or YCbCr 4:4:4, 4:2:2 uncompressed and DSC 4:2:2 Simple) pixel rate is the same as
+ *   container rate.
+ *
+ * - In 4:2:0 (DSC or uncompressed) there are two pixels per container, hence the target container rate has to be
+ *   halved to maintain the correct pixel rate.
+ *
+ * - Unlike 4:2:2 uncompressed, DSC 4:2:2 Native also has two pixels per container (this happens when DSC is applied
+ *   to it) and has to be treated the same as 4:2:0, i.e. target containter rate has to be halved in this case as well.
+ *
+ */
+bool dce110_is_two_pixels_per_container(const struct dc_crtc_timing *timing)
+{
+	return timing->pixel_encoding == PIXEL_ENCODING_YCBCR420;
+}
+
+void dce110_tg_wait_for_state(struct timing_generator *tg,
+	enum crtc_state state)
+{
+	switch (state) {
+	case CRTC_STATE_VBLANK:
+		dce110_timing_generator_wait_for_vblank(tg);
+		break;
+
+	case CRTC_STATE_VACTIVE:
+		dce110_timing_generator_wait_for_vactive(tg);
+		break;
+
+	default:
+		break;
+	}
+}
+
+void dce110_tg_set_colors(struct timing_generator *tg,
+	const struct tg_color *blank_color,
+	const struct tg_color *overscan_color)
+{
+	if (blank_color != NULL)
+		dce110_tg_program_blank_color(tg, blank_color);
+	if (overscan_color != NULL)
+		dce110_tg_set_overscan_color(tg, overscan_color);
+}
+
+/* Gets first line of blank region of the display timing for CRTC
+ * and programms is as a trigger to fire vertical interrupt
+ */
+bool dce110_arm_vert_intr(struct timing_generator *tg, uint8_t width)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t v_blank_start = 0;
+	uint32_t v_blank_end = 0;
+	uint32_t val = 0;
+	uint32_t h_position, v_position;
+
+	tg->funcs->get_scanoutpos(
+			tg,
+			&v_blank_start,
+			&v_blank_end,
+			&h_position,
+			&v_position);
+
+	if (v_blank_start == 0 || v_blank_end == 0)
+		return false;
+
+	set_reg_field_value(
+		val,
+		v_blank_start,
+		CRTC_VERTICAL_INTERRUPT0_POSITION,
+		CRTC_VERTICAL_INTERRUPT0_LINE_START);
+
+	/* Set interval width for interrupt to fire to 1 scanline */
+	set_reg_field_value(
+		val,
+		v_blank_start + width,
+		CRTC_VERTICAL_INTERRUPT0_POSITION,
+		CRTC_VERTICAL_INTERRUPT0_LINE_END);
+
+	dm_write_reg(tg->ctx, CRTC_REG(mmCRTC_VERTICAL_INTERRUPT0_POSITION), val);
+
+	return true;
+}
+
+static bool dce110_is_tg_enabled(struct timing_generator *tg)
+{
+	uint32_t addr = 0;
+	uint32_t value = 0;
+	uint32_t field = 0;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	addr = CRTC_REG(mmCRTC_CONTROL);
+	value = dm_read_reg(tg->ctx, addr);
+	field = get_reg_field_value(value, CRTC_CONTROL,
+				    CRTC_CURRENT_MASTER_EN_STATE);
+	return field == 1;
+}
+
+bool dce110_configure_crc(struct timing_generator *tg,
+			  const struct crc_params *params)
+{
+	uint32_t cntl_addr = 0;
+	uint32_t addr = 0;
+	uint32_t value;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	/* Cannot configure crc on a CRTC that is disabled */
+	if (!dce110_is_tg_enabled(tg))
+		return false;
+
+	cntl_addr = CRTC_REG(mmCRTC_CRC_CNTL);
+
+	if (!params->enable || params->reset)
+		/* First, disable CRC before we configure it. */
+		dm_write_reg(tg->ctx, cntl_addr, 0);
+
+	if (!params->enable)
+		return true;
+
+	/* Program frame boundaries */
+	switch (params->crc_eng_inst) {
+	case 0:
+		/* Window A x axis start and end. */
+		value = 0;
+		addr = CRTC_REG(mmCRTC_CRC0_WINDOWA_X_CONTROL);
+		set_reg_field_value(value, params->windowa_x_start,
+				    CRTC_CRC0_WINDOWA_X_CONTROL,
+				    CRTC_CRC0_WINDOWA_X_START);
+		set_reg_field_value(value, params->windowa_x_end,
+				    CRTC_CRC0_WINDOWA_X_CONTROL,
+				    CRTC_CRC0_WINDOWA_X_END);
+		dm_write_reg(tg->ctx, addr, value);
+
+		/* Window A y axis start and end. */
+		value = 0;
+		addr = CRTC_REG(mmCRTC_CRC0_WINDOWA_Y_CONTROL);
+		set_reg_field_value(value, params->windowa_y_start,
+				    CRTC_CRC0_WINDOWA_Y_CONTROL,
+				    CRTC_CRC0_WINDOWA_Y_START);
+		set_reg_field_value(value, params->windowa_y_end,
+				    CRTC_CRC0_WINDOWA_Y_CONTROL,
+				    CRTC_CRC0_WINDOWA_Y_END);
+		dm_write_reg(tg->ctx, addr, value);
+
+		/* Window B x axis start and end. */
+		value = 0;
+		addr = CRTC_REG(mmCRTC_CRC0_WINDOWB_X_CONTROL);
+		set_reg_field_value(value, params->windowb_x_start,
+				    CRTC_CRC0_WINDOWB_X_CONTROL,
+				    CRTC_CRC0_WINDOWB_X_START);
+		set_reg_field_value(value, params->windowb_x_end,
+				    CRTC_CRC0_WINDOWB_X_CONTROL,
+				    CRTC_CRC0_WINDOWB_X_END);
+		dm_write_reg(tg->ctx, addr, value);
+
+		/* Window B y axis start and end. */
+		value = 0;
+		addr = CRTC_REG(mmCRTC_CRC0_WINDOWB_Y_CONTROL);
+		set_reg_field_value(value, params->windowb_y_start,
+				    CRTC_CRC0_WINDOWB_Y_CONTROL,
+				    CRTC_CRC0_WINDOWB_Y_START);
+		set_reg_field_value(value, params->windowb_y_end,
+				    CRTC_CRC0_WINDOWB_Y_CONTROL,
+				    CRTC_CRC0_WINDOWB_Y_END);
+		dm_write_reg(tg->ctx, addr, value);
+
+		/* Set crc mode and selection, and enable.*/
+		value = 0;
+		set_reg_field_value(value, params->continuous_mode ? 1 : 0,
+				    CRTC_CRC_CNTL, CRTC_CRC_CONT_EN);
+		set_reg_field_value(value, params->selection,
+				    CRTC_CRC_CNTL, CRTC_CRC0_SELECT);
+		set_reg_field_value(value, 1, CRTC_CRC_CNTL, CRTC_CRC_EN);
+		dm_write_reg(tg->ctx, cntl_addr, value);
+		break;
+	case 1:
+		/* Window A x axis start and end. */
+		value = 0;
+		addr = CRTC_REG(mmCRTC_CRC1_WINDOWA_X_CONTROL);
+		set_reg_field_value(value, params->windowa_x_start,
+				    CRTC_CRC1_WINDOWA_X_CONTROL,
+				    CRTC_CRC1_WINDOWA_X_START);
+		set_reg_field_value(value, params->windowa_x_end,
+				    CRTC_CRC1_WINDOWA_X_CONTROL,
+				    CRTC_CRC1_WINDOWA_X_END);
+		dm_write_reg(tg->ctx, addr, value);
+
+		/* Window A y axis start and end. */
+		value = 0;
+		addr = CRTC_REG(mmCRTC_CRC1_WINDOWA_Y_CONTROL);
+		set_reg_field_value(value, params->windowa_y_start,
+				    CRTC_CRC1_WINDOWA_Y_CONTROL,
+				    CRTC_CRC1_WINDOWA_Y_START);
+		set_reg_field_value(value, params->windowa_y_end,
+				    CRTC_CRC1_WINDOWA_Y_CONTROL,
+				    CRTC_CRC1_WINDOWA_Y_END);
+		dm_write_reg(tg->ctx, addr, value);
+
+		/* Window B x axis start and end. */
+		value = 0;
+		addr = CRTC_REG(mmCRTC_CRC1_WINDOWB_X_CONTROL);
+		set_reg_field_value(value, params->windowb_x_start,
+				    CRTC_CRC1_WINDOWB_X_CONTROL,
+				    CRTC_CRC1_WINDOWB_X_START);
+		set_reg_field_value(value, params->windowb_x_end,
+				    CRTC_CRC1_WINDOWB_X_CONTROL,
+				    CRTC_CRC1_WINDOWB_X_END);
+		dm_write_reg(tg->ctx, addr, value);
+
+		/* Window B y axis start and end. */
+		value = 0;
+		addr = CRTC_REG(mmCRTC_CRC1_WINDOWB_Y_CONTROL);
+		set_reg_field_value(value, params->windowb_y_start,
+				    CRTC_CRC1_WINDOWB_Y_CONTROL,
+				    CRTC_CRC1_WINDOWB_Y_START);
+		set_reg_field_value(value, params->windowb_y_end,
+				    CRTC_CRC1_WINDOWB_Y_CONTROL,
+				    CRTC_CRC1_WINDOWB_Y_END);
+		dm_write_reg(tg->ctx, addr, value);
+
+		/* Set crc mode and selection, and enable.*/
+		value = 0;
+		set_reg_field_value(value, params->continuous_mode ? 1 : 0,
+				    CRTC_CRC_CNTL, CRTC_CRC_CONT_EN);
+		set_reg_field_value(value, params->selection,
+				    CRTC_CRC_CNTL, CRTC_CRC1_SELECT);
+		set_reg_field_value(value, 1, CRTC_CRC_CNTL, CRTC_CRC_EN);
+		dm_write_reg(tg->ctx, cntl_addr, value);
+		break;
+	default:
+		return false;
+	}
+
+	return true;
+}
+
+bool dce110_get_crc(struct timing_generator *tg, uint8_t idx,
+		    uint32_t *r_cr, uint32_t *g_y, uint32_t *b_cb)
+{
+	uint32_t addr = 0;
+	uint32_t value = 0;
+	uint32_t field = 0;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	addr = CRTC_REG(mmCRTC_CRC_CNTL);
+	value = dm_read_reg(tg->ctx, addr);
+	field = get_reg_field_value(value, CRTC_CRC_CNTL, CRTC_CRC_EN);
+
+	/* Early return if CRC is not enabled for this CRTC */
+	if (!field)
+		return false;
+
+	switch (idx) {
+	case 0:
+		addr = CRTC_REG(mmCRTC_CRC0_DATA_RG);
+		value = dm_read_reg(tg->ctx, addr);
+		*r_cr = get_reg_field_value(value, CRTC_CRC0_DATA_RG, CRC0_R_CR);
+		*g_y = get_reg_field_value(value, CRTC_CRC0_DATA_RG, CRC0_G_Y);
+
+		addr = CRTC_REG(mmCRTC_CRC0_DATA_B);
+		value = dm_read_reg(tg->ctx, addr);
+		*b_cb = get_reg_field_value(value, CRTC_CRC0_DATA_B, CRC0_B_CB);
+		break;
+	case 1:
+		addr = CRTC_REG(mmCRTC_CRC1_DATA_RG);
+		value = dm_read_reg(tg->ctx, addr);
+		*r_cr = get_reg_field_value(value, CRTC_CRC1_DATA_RG, CRC1_R_CR);
+		*g_y = get_reg_field_value(value, CRTC_CRC1_DATA_RG, CRC1_G_Y);
+
+		addr = CRTC_REG(mmCRTC_CRC1_DATA_B);
+		value = dm_read_reg(tg->ctx, addr);
+		*b_cb = get_reg_field_value(value, CRTC_CRC1_DATA_B, CRC1_B_CB);
+		break;
+	default:
+		return false;
+	}
+
+	return true;
+}
+
+static const struct timing_generator_funcs dce110_tg_funcs = {
+		.validate_timing = dce110_tg_validate_timing,
+		.program_timing = dce110_tg_program_timing,
+		.enable_crtc = dce110_timing_generator_enable_crtc,
+		.disable_crtc = dce110_timing_generator_disable_crtc,
+		.is_counter_moving = dce110_timing_generator_is_counter_moving,
+		.get_position = dce110_timing_generator_get_position,
+		.get_frame_count = dce110_timing_generator_get_vblank_counter,
+		.get_scanoutpos = dce110_timing_generator_get_crtc_scanoutpos,
+		.set_early_control = dce110_timing_generator_set_early_control,
+		.wait_for_state = dce110_tg_wait_for_state,
+		.set_blank = dce110_tg_set_blank,
+		.is_blanked = dce110_tg_is_blanked,
+		.set_colors = dce110_tg_set_colors,
+		.set_overscan_blank_color =
+				dce110_timing_generator_set_overscan_color_black,
+		.set_blank_color = dce110_timing_generator_program_blank_color,
+		.disable_vga = dce110_timing_generator_disable_vga,
+		.did_triggered_reset_occur =
+				dce110_timing_generator_did_triggered_reset_occur,
+		.setup_global_swap_lock =
+				dce110_timing_generator_setup_global_swap_lock,
+		.enable_reset_trigger = dce110_timing_generator_enable_reset_trigger,
+		.enable_crtc_reset = dce110_timing_generator_enable_crtc_reset,
+		.disable_reset_trigger = dce110_timing_generator_disable_reset_trigger,
+		.tear_down_global_swap_lock =
+				dce110_timing_generator_tear_down_global_swap_lock,
+		.enable_advanced_request =
+				dce110_timing_generator_enable_advanced_request,
+		.set_drr =
+				dce110_timing_generator_set_drr,
+		.get_last_used_drr_vtotal = NULL,
+		.set_static_screen_control =
+			dce110_timing_generator_set_static_screen_control,
+		.set_test_pattern = dce110_timing_generator_set_test_pattern,
+		.arm_vert_intr = dce110_arm_vert_intr,
+		.is_tg_enabled = dce110_is_tg_enabled,
+		.configure_crc = dce110_configure_crc,
+		.get_crc = dce110_get_crc,
+		.is_two_pixels_per_container = dce110_is_two_pixels_per_container,
+};
+
+void dce110_timing_generator_construct(
+	struct dce110_timing_generator *tg110,
+	struct dc_context *ctx,
+	uint32_t instance,
+	const struct dce110_timing_generator_offsets *offsets)
+{
+	tg110->controller_id = CONTROLLER_ID_D0 + instance;
+	tg110->base.inst = instance;
+
+	tg110->offsets = *offsets;
+
+	tg110->base.funcs = &dce110_tg_funcs;
+
+	tg110->base.ctx = ctx;
+	tg110->base.bp = ctx->dc_bios;
+
+	tg110->max_h_total = CRTC_H_TOTAL__CRTC_H_TOTAL_MASK + 1;
+	tg110->max_v_total = CRTC_V_TOTAL__CRTC_V_TOTAL_MASK + 1;
+
+	tg110->min_h_blank = 56;
+	tg110->min_h_front_porch = 4;
+	tg110->min_h_back_porch = 4;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator.h b/drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator.h
new file mode 100644
index 000000000000..e4f5cad64f32
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator.h
@@ -0,0 +1,294 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_TIMING_GENERATOR_DCE110_H__
+#define __DC_TIMING_GENERATOR_DCE110_H__
+
+#include "timing_generator.h"
+#include "../include/grph_object_id.h"
+
+/* GSL Sync related values */
+
+/* In VSync mode, after 4 units of time, master pipe will generate
+ * flip_ready signal */
+#define VFLIP_READY_DELAY 4
+/* In HSync mode, after 2 units of time, master pipe will generate
+ * flip_ready signal */
+#define HFLIP_READY_DELAY 2
+/* 6 lines delay between forcing flip and checking all pipes ready */
+#define HFLIP_CHECK_DELAY 6
+/* 3 lines before end of frame */
+#define FLIP_READY_BACK_LOOKUP 3
+
+/* Trigger Source Select - ASIC-defendant, actual values for the
+ * register programming */
+enum trigger_source_select {
+	TRIGGER_SOURCE_SELECT_LOGIC_ZERO = 0,
+	TRIGGER_SOURCE_SELECT_CRTC_VSYNCA = 1,
+	TRIGGER_SOURCE_SELECT_CRTC_HSYNCA = 2,
+	TRIGGER_SOURCE_SELECT_CRTC_VSYNCB = 3,
+	TRIGGER_SOURCE_SELECT_CRTC_HSYNCB = 4,
+	TRIGGER_SOURCE_SELECT_GENERICF = 5,
+	TRIGGER_SOURCE_SELECT_GENERICE = 6,
+	TRIGGER_SOURCE_SELECT_VSYNCA = 7,
+	TRIGGER_SOURCE_SELECT_HSYNCA = 8,
+	TRIGGER_SOURCE_SELECT_VSYNCB = 9,
+	TRIGGER_SOURCE_SELECT_HSYNCB = 10,
+	TRIGGER_SOURCE_SELECT_HPD1 = 11,
+	TRIGGER_SOURCE_SELECT_HPD2 = 12,
+	TRIGGER_SOURCE_SELECT_GENERICD = 13,
+	TRIGGER_SOURCE_SELECT_GENERICC = 14,
+	TRIGGER_SOURCE_SELECT_VIDEO_CAPTURE = 15,
+	TRIGGER_SOURCE_SELECT_GSL_GROUP0 = 16,
+	TRIGGER_SOURCE_SELECT_GSL_GROUP1 = 17,
+	TRIGGER_SOURCE_SELECT_GSL_GROUP2 = 18,
+	TRIGGER_SOURCE_SELECT_BLONY = 19,
+	TRIGGER_SOURCE_SELECT_GENERICA = 20,
+	TRIGGER_SOURCE_SELECT_GENERICB = 21,
+	TRIGGER_SOURCE_SELECT_GSL_ALLOW_FLIP = 22,
+	TRIGGER_SOURCE_SELECT_MANUAL_TRIGGER = 23
+};
+
+/* Trigger Source Select - ASIC-dependant, actual values for the
+ * register programming */
+enum trigger_polarity_select {
+	TRIGGER_POLARITY_SELECT_LOGIC_ZERO = 0,
+	TRIGGER_POLARITY_SELECT_CRTC = 1,
+	TRIGGER_POLARITY_SELECT_GENERICA = 2,
+	TRIGGER_POLARITY_SELECT_GENERICB = 3,
+	TRIGGER_POLARITY_SELECT_HSYNCA = 4,
+	TRIGGER_POLARITY_SELECT_HSYNCB = 5,
+	TRIGGER_POLARITY_SELECT_VIDEO_CAPTURE = 6,
+	TRIGGER_POLARITY_SELECT_GENERICC = 7
+};
+
+
+struct dce110_timing_generator_offsets {
+	int32_t crtc;
+	int32_t dcp;
+
+	/* DCE80 use only */
+	int32_t dmif;
+};
+
+struct dce110_timing_generator {
+	struct timing_generator base;
+	struct dce110_timing_generator_offsets offsets;
+	struct dce110_timing_generator_offsets derived_offsets;
+
+	enum controller_id controller_id;
+
+	uint32_t max_h_total;
+	uint32_t max_v_total;
+
+	uint32_t min_h_blank;
+	uint32_t min_h_front_porch;
+	uint32_t min_h_back_porch;
+
+	/* DCE 12 */
+	uint32_t min_h_sync_width;
+	uint32_t min_v_sync_width;
+	uint32_t min_v_blank;
+
+};
+
+#define DCE110TG_FROM_TG(tg)\
+	container_of(tg, struct dce110_timing_generator, base)
+
+void dce110_timing_generator_construct(
+	struct dce110_timing_generator *tg,
+	struct dc_context *ctx,
+	uint32_t instance,
+	const struct dce110_timing_generator_offsets *offsets);
+
+/* determine if given timing can be supported by TG */
+bool dce110_timing_generator_validate_timing(
+	struct timing_generator *tg,
+	const struct dc_crtc_timing *timing,
+	enum signal_type signal);
+
+/******** HW programming ************/
+
+/* Program timing generator with given timing */
+bool dce110_timing_generator_program_timing_generator(
+	struct timing_generator *tg,
+	const struct dc_crtc_timing *dc_crtc_timing);
+
+/* Disable/Enable Timing Generator */
+bool dce110_timing_generator_enable_crtc(struct timing_generator *tg);
+bool dce110_timing_generator_disable_crtc(struct timing_generator *tg);
+
+void dce110_timing_generator_set_early_control(
+		struct timing_generator *tg,
+		uint32_t early_cntl);
+
+/**************** TG current status ******************/
+
+/* return the current frame counter. Used by Linux kernel DRM */
+uint32_t dce110_timing_generator_get_vblank_counter(
+		struct timing_generator *tg);
+
+void dce110_timing_generator_get_position(
+	struct timing_generator *tg,
+	struct crtc_position *position);
+
+/* return true if TG counter is moving. false if TG is stopped */
+bool dce110_timing_generator_is_counter_moving(struct timing_generator *tg);
+
+/* wait until TG is in beginning of vertical blank region */
+void dce110_timing_generator_wait_for_vblank(struct timing_generator *tg);
+
+/* wait until TG is in beginning of active region */
+void dce110_timing_generator_wait_for_vactive(struct timing_generator *tg);
+
+/*********** Timing Generator Synchronization routines ****/
+
+/* Setups Global Swap Lock group, TimingServer or TimingClient*/
+void dce110_timing_generator_setup_global_swap_lock(
+	struct timing_generator *tg,
+	const struct dcp_gsl_params *gsl_params);
+
+/* Clear all the register writes done by setup_global_swap_lock */
+void dce110_timing_generator_tear_down_global_swap_lock(
+	struct timing_generator *tg);
+
+/* Reset crtc position on master VSync */
+void dce110_timing_generator_enable_crtc_reset(
+	struct timing_generator *tg,
+	int source,
+	struct crtc_trigger_info *crtc_tp);
+
+/* Reset slave controllers on master VSync */
+void dce110_timing_generator_enable_reset_trigger(
+	struct timing_generator *tg,
+	int source);
+
+/* disabling trigger-reset */
+void dce110_timing_generator_disable_reset_trigger(
+	struct timing_generator *tg);
+
+/* Checks whether CRTC triggered reset occurred */
+bool dce110_timing_generator_did_triggered_reset_occur(
+	struct timing_generator *tg);
+
+/******** Stuff to move to other virtual HW objects *****************/
+/* Move to enable accelerated mode */
+void dce110_timing_generator_disable_vga(struct timing_generator *tg);
+/* TODO: Should we move it to transform */
+/* Fully program CRTC timing in timing generator */
+void dce110_timing_generator_program_blanking(
+	struct timing_generator *tg,
+	const struct dc_crtc_timing *timing);
+
+/* TODO: Should we move it to opp? */
+/* Combine with below and move YUV/RGB color conversion to SW layer */
+void dce110_timing_generator_program_blank_color(
+	struct timing_generator *tg,
+	const struct tg_color *black_color);
+/* Combine with above and move YUV/RGB color conversion to SW layer */
+void dce110_timing_generator_set_overscan_color_black(
+	struct timing_generator *tg,
+	const struct tg_color *color);
+void dce110_timing_generator_color_space_to_black_color(
+		enum dc_color_space colorspace,
+	struct tg_color *black_color);
+/*************** End-of-move ********************/
+
+/* Not called yet */
+void dce110_timing_generator_set_test_pattern(
+	struct timing_generator *tg,
+	/* TODO: replace 'controller_dp_test_pattern' by 'test_pattern_mode'
+	 * because this is not DP-specific (which is probably somewhere in DP
+	 * encoder) */
+	enum controller_dp_test_pattern test_pattern,
+	enum dc_color_depth color_depth);
+
+void dce110_timing_generator_set_drr(
+	struct timing_generator *tg,
+	const struct drr_params *params);
+
+void dce110_timing_generator_set_static_screen_control(
+	struct timing_generator *tg,
+	uint32_t event_triggers,
+	uint32_t num_frames);
+
+void dce110_timing_generator_get_crtc_scanoutpos(
+	struct timing_generator *tg,
+	uint32_t *v_blank_start,
+	uint32_t *v_blank_end,
+	uint32_t *h_position,
+	uint32_t *v_position);
+
+void dce110_timing_generator_enable_advanced_request(
+	struct timing_generator *tg,
+	bool enable,
+	const struct dc_crtc_timing *timing);
+
+void dce110_timing_generator_set_lock_master(struct timing_generator *tg,
+		bool lock);
+
+void dce110_tg_program_blank_color(struct timing_generator *tg,
+	const struct tg_color *black_color);
+
+void dce110_tg_set_overscan_color(struct timing_generator *tg,
+	const struct tg_color *overscan_color);
+
+void dce110_tg_program_timing(struct timing_generator *tg,
+	const struct dc_crtc_timing *timing,
+	int vready_offset,
+	int vstartup_start,
+	int vupdate_offset,
+	int vupdate_width,
+	int pstate_keepout,
+	const enum signal_type signal,
+	bool use_vbios);
+
+bool dce110_tg_is_blanked(struct timing_generator *tg);
+
+void dce110_tg_set_blank(struct timing_generator *tg,
+		bool enable_blanking);
+
+bool dce110_tg_validate_timing(struct timing_generator *tg,
+	const struct dc_crtc_timing *timing);
+
+void dce110_tg_wait_for_state(struct timing_generator *tg,
+	enum crtc_state state);
+
+void dce110_tg_set_colors(struct timing_generator *tg,
+	const struct tg_color *blank_color,
+	const struct tg_color *overscan_color);
+
+bool dce110_arm_vert_intr(
+		struct timing_generator *tg, uint8_t width);
+
+bool dce110_configure_crc(struct timing_generator *tg,
+			  const struct crc_params *params);
+
+bool dce110_get_crc(struct timing_generator *tg, uint8_t idx,
+		    uint32_t *r_cr, uint32_t *g_y, uint32_t *b_cb);
+
+bool dce110_is_two_pixels_per_container(const struct dc_crtc_timing *timing);
+
+#endif /* __DC_TIMING_GENERATOR_DCE110_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator_v.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator_v.c
new file mode 100644
index 000000000000..9837dec837ff
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator_v.c
@@ -0,0 +1,707 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "dm_services.h"
+
+/* include DCE11 register header files */
+#include "dce/dce_11_0_d.h"
+#include "dce/dce_11_0_sh_mask.h"
+
+#include "dc_types.h"
+#include "dc_bios_types.h"
+#include "dc.h"
+
+#include "include/grph_object_id.h"
+#include "include/logger_interface.h"
+#include "dce110_timing_generator.h"
+#include "dce110_timing_generator_v.h"
+
+#include "timing_generator.h"
+
+#define DC_LOGGER \
+	tg->ctx->logger
+/** ********************************************************************************
+ *
+ * DCE11 Timing Generator Implementation
+ *
+ **********************************************************************************/
+
+/*
+ * Enable CRTCV
+ */
+
+static bool dce110_timing_generator_v_enable_crtc(struct timing_generator *tg)
+{
+/*
+ * Set MASTER_UPDATE_MODE to 0
+ * This is needed for DRR, and also suggested to be default value by Syed.
+ */
+	uint32_t value;
+
+	value = 0;
+	set_reg_field_value(value, 0,
+			CRTCV_MASTER_UPDATE_MODE, MASTER_UPDATE_MODE);
+	dm_write_reg(tg->ctx,
+			mmCRTCV_MASTER_UPDATE_MODE, value);
+
+	/* TODO: may want this on for looking for underflow */
+	value = 0;
+	dm_write_reg(tg->ctx, mmCRTCV_MASTER_UPDATE_MODE, value);
+
+	value = 0;
+	set_reg_field_value(value, 1,
+			CRTCV_MASTER_EN, CRTC_MASTER_EN);
+	dm_write_reg(tg->ctx,
+			mmCRTCV_MASTER_EN, value);
+
+	return true;
+}
+
+static bool dce110_timing_generator_v_disable_crtc(struct timing_generator *tg)
+{
+	uint32_t value;
+
+	value = dm_read_reg(tg->ctx,
+			mmCRTCV_CONTROL);
+	set_reg_field_value(value, 0,
+			CRTCV_CONTROL, CRTC_DISABLE_POINT_CNTL);
+	set_reg_field_value(value, 0,
+				CRTCV_CONTROL, CRTC_MASTER_EN);
+	dm_write_reg(tg->ctx,
+			mmCRTCV_CONTROL, value);
+	/*
+	 * TODO: call this when adding stereo support
+	 * tg->funcs->disable_stereo(tg);
+	 */
+	return true;
+}
+
+static void dce110_timing_generator_v_blank_crtc(struct timing_generator *tg)
+{
+	uint32_t addr = mmCRTCV_BLANK_CONTROL;
+	uint32_t value = dm_read_reg(tg->ctx, addr);
+
+	set_reg_field_value(
+		value,
+		1,
+		CRTCV_BLANK_CONTROL,
+		CRTC_BLANK_DATA_EN);
+
+	set_reg_field_value(
+		value,
+		0,
+		CRTCV_BLANK_CONTROL,
+		CRTC_BLANK_DE_MODE);
+
+	dm_write_reg(tg->ctx, addr, value);
+}
+
+static void dce110_timing_generator_v_unblank_crtc(struct timing_generator *tg)
+{
+	uint32_t addr = mmCRTCV_BLANK_CONTROL;
+	uint32_t value = dm_read_reg(tg->ctx, addr);
+
+	set_reg_field_value(
+		value,
+		0,
+		CRTCV_BLANK_CONTROL,
+		CRTC_BLANK_DATA_EN);
+
+	set_reg_field_value(
+		value,
+		0,
+		CRTCV_BLANK_CONTROL,
+		CRTC_BLANK_DE_MODE);
+
+	dm_write_reg(tg->ctx, addr, value);
+}
+
+static bool dce110_timing_generator_v_is_in_vertical_blank(
+		struct timing_generator *tg)
+{
+	uint32_t addr = 0;
+	uint32_t value = 0;
+	uint32_t field = 0;
+
+	addr = mmCRTCV_STATUS;
+	value = dm_read_reg(tg->ctx, addr);
+	field = get_reg_field_value(value, CRTCV_STATUS, CRTC_V_BLANK);
+	return field == 1;
+}
+
+static bool dce110_timing_generator_v_is_counter_moving(struct timing_generator *tg)
+{
+	uint32_t value;
+	uint32_t h1 = 0;
+	uint32_t h2 = 0;
+	uint32_t v1 = 0;
+	uint32_t v2 = 0;
+
+	value = dm_read_reg(tg->ctx, mmCRTCV_STATUS_POSITION);
+
+	h1 = get_reg_field_value(
+			value,
+			CRTCV_STATUS_POSITION,
+			CRTC_HORZ_COUNT);
+
+	v1 = get_reg_field_value(
+			value,
+			CRTCV_STATUS_POSITION,
+			CRTC_VERT_COUNT);
+
+	value = dm_read_reg(tg->ctx, mmCRTCV_STATUS_POSITION);
+
+	h2 = get_reg_field_value(
+			value,
+			CRTCV_STATUS_POSITION,
+			CRTC_HORZ_COUNT);
+
+	v2 = get_reg_field_value(
+			value,
+			CRTCV_STATUS_POSITION,
+			CRTC_VERT_COUNT);
+
+	if (h1 == h2 && v1 == v2)
+		return false;
+	else
+		return true;
+}
+
+static void dce110_timing_generator_v_wait_for_vblank(struct timing_generator *tg)
+{
+	/* We want to catch beginning of VBlank here, so if the first try are
+	 * in VBlank, we might be very close to Active, in this case wait for
+	 * another frame
+	 */
+	while (dce110_timing_generator_v_is_in_vertical_blank(tg)) {
+		if (!dce110_timing_generator_v_is_counter_moving(tg)) {
+			/* error - no point to wait if counter is not moving */
+			break;
+		}
+	}
+
+	while (!dce110_timing_generator_v_is_in_vertical_blank(tg)) {
+		if (!dce110_timing_generator_v_is_counter_moving(tg)) {
+			/* error - no point to wait if counter is not moving */
+			break;
+		}
+	}
+}
+
+/*
+ * Wait till we are in VActive (anywhere in VActive)
+ */
+static void dce110_timing_generator_v_wait_for_vactive(struct timing_generator *tg)
+{
+	while (dce110_timing_generator_v_is_in_vertical_blank(tg)) {
+		if (!dce110_timing_generator_v_is_counter_moving(tg)) {
+			/* error - no point to wait if counter is not moving */
+			break;
+		}
+	}
+}
+
+static void dce110_timing_generator_v_wait_for_state(struct timing_generator *tg,
+	enum crtc_state state)
+{
+	switch (state) {
+	case CRTC_STATE_VBLANK:
+		dce110_timing_generator_v_wait_for_vblank(tg);
+		break;
+
+	case CRTC_STATE_VACTIVE:
+		dce110_timing_generator_v_wait_for_vactive(tg);
+		break;
+
+	default:
+		break;
+	}
+}
+
+static void dce110_timing_generator_v_program_blanking(
+	struct timing_generator *tg,
+	const struct dc_crtc_timing *timing)
+{
+	uint32_t vsync_offset = timing->v_border_bottom +
+			timing->v_front_porch;
+	uint32_t v_sync_start = timing->v_addressable + vsync_offset;
+
+	uint32_t hsync_offset = timing->h_border_right +
+			timing->h_front_porch;
+	uint32_t h_sync_start = timing->h_addressable + hsync_offset;
+
+	struct dc_context *ctx = tg->ctx;
+	uint32_t value = 0;
+	uint32_t addr = 0;
+	uint32_t tmp = 0;
+
+	addr = mmCRTCV_H_TOTAL;
+	value = dm_read_reg(ctx, addr);
+	set_reg_field_value(
+		value,
+		timing->h_total - 1,
+		CRTCV_H_TOTAL,
+		CRTC_H_TOTAL);
+	dm_write_reg(ctx, addr, value);
+
+	addr = mmCRTCV_V_TOTAL;
+	value = dm_read_reg(ctx, addr);
+	set_reg_field_value(
+		value,
+		timing->v_total - 1,
+		CRTCV_V_TOTAL,
+		CRTC_V_TOTAL);
+	dm_write_reg(ctx, addr, value);
+
+	addr = mmCRTCV_H_BLANK_START_END;
+	value = dm_read_reg(ctx, addr);
+
+	tmp = timing->h_total -
+		(h_sync_start + timing->h_border_left);
+
+	set_reg_field_value(
+		value,
+		tmp,
+		CRTCV_H_BLANK_START_END,
+		CRTC_H_BLANK_END);
+
+	tmp = tmp + timing->h_addressable +
+		timing->h_border_left + timing->h_border_right;
+
+	set_reg_field_value(
+		value,
+		tmp,
+		CRTCV_H_BLANK_START_END,
+		CRTC_H_BLANK_START);
+
+	dm_write_reg(ctx, addr, value);
+
+	addr = mmCRTCV_V_BLANK_START_END;
+	value = dm_read_reg(ctx, addr);
+
+	tmp = timing->v_total - (v_sync_start + timing->v_border_top);
+
+	set_reg_field_value(
+		value,
+		tmp,
+		CRTCV_V_BLANK_START_END,
+		CRTC_V_BLANK_END);
+
+	tmp = tmp + timing->v_addressable + timing->v_border_top +
+		timing->v_border_bottom;
+
+	set_reg_field_value(
+		value,
+		tmp,
+		CRTCV_V_BLANK_START_END,
+		CRTC_V_BLANK_START);
+
+	dm_write_reg(ctx, addr, value);
+
+	addr = mmCRTCV_H_SYNC_A;
+	value = 0;
+	set_reg_field_value(
+		value,
+		timing->h_sync_width,
+		CRTCV_H_SYNC_A,
+		CRTC_H_SYNC_A_END);
+	dm_write_reg(ctx, addr, value);
+
+	addr = mmCRTCV_H_SYNC_A_CNTL;
+	value = dm_read_reg(ctx, addr);
+	if (timing->flags.HSYNC_POSITIVE_POLARITY) {
+		set_reg_field_value(
+			value,
+			0,
+			CRTCV_H_SYNC_A_CNTL,
+			CRTC_H_SYNC_A_POL);
+	} else {
+		set_reg_field_value(
+			value,
+			1,
+			CRTCV_H_SYNC_A_CNTL,
+			CRTC_H_SYNC_A_POL);
+	}
+	dm_write_reg(ctx, addr, value);
+
+	addr = mmCRTCV_V_SYNC_A;
+	value = 0;
+	set_reg_field_value(
+		value,
+		timing->v_sync_width,
+		CRTCV_V_SYNC_A,
+		CRTC_V_SYNC_A_END);
+	dm_write_reg(ctx, addr, value);
+
+	addr = mmCRTCV_V_SYNC_A_CNTL;
+	value = dm_read_reg(ctx, addr);
+	if (timing->flags.VSYNC_POSITIVE_POLARITY) {
+		set_reg_field_value(
+			value,
+			0,
+			CRTCV_V_SYNC_A_CNTL,
+			CRTC_V_SYNC_A_POL);
+	} else {
+		set_reg_field_value(
+			value,
+			1,
+			CRTCV_V_SYNC_A_CNTL,
+			CRTC_V_SYNC_A_POL);
+	}
+	dm_write_reg(ctx, addr, value);
+
+	addr = mmCRTCV_INTERLACE_CONTROL;
+	value = dm_read_reg(ctx, addr);
+	set_reg_field_value(
+		value,
+		timing->flags.INTERLACE,
+		CRTCV_INTERLACE_CONTROL,
+		CRTC_INTERLACE_ENABLE);
+	dm_write_reg(ctx, addr, value);
+}
+
+static void dce110_timing_generator_v_enable_advanced_request(
+	struct timing_generator *tg,
+	bool enable,
+	const struct dc_crtc_timing *timing)
+{
+	uint32_t addr = mmCRTCV_START_LINE_CONTROL;
+	uint32_t value = dm_read_reg(tg->ctx, addr);
+
+	if (enable) {
+		if ((timing->v_sync_width + timing->v_front_porch) <= 3) {
+			set_reg_field_value(
+				value,
+				3,
+				CRTCV_START_LINE_CONTROL,
+				CRTC_ADVANCED_START_LINE_POSITION);
+		} else {
+			set_reg_field_value(
+				value,
+				4,
+				CRTCV_START_LINE_CONTROL,
+				CRTC_ADVANCED_START_LINE_POSITION);
+		}
+		set_reg_field_value(
+			value,
+			0,
+			CRTCV_START_LINE_CONTROL,
+			CRTC_LEGACY_REQUESTOR_EN);
+	} else {
+		set_reg_field_value(
+			value,
+			2,
+			CRTCV_START_LINE_CONTROL,
+			CRTC_ADVANCED_START_LINE_POSITION);
+		set_reg_field_value(
+			value,
+			1,
+			CRTCV_START_LINE_CONTROL,
+			CRTC_LEGACY_REQUESTOR_EN);
+	}
+
+	dm_write_reg(tg->ctx, addr, value);
+}
+
+static void dce110_timing_generator_v_set_blank(struct timing_generator *tg,
+		bool enable_blanking)
+{
+	if (enable_blanking)
+		dce110_timing_generator_v_blank_crtc(tg);
+	else
+		dce110_timing_generator_v_unblank_crtc(tg);
+}
+
+static void dce110_timing_generator_v_program_timing(struct timing_generator *tg,
+	const struct dc_crtc_timing *timing,
+	int vready_offset,
+	int vstartup_start,
+	int vupdate_offset,
+	int vupdate_width,
+	int pstate_keepout,
+	const enum signal_type signal,
+	bool use_vbios)
+{
+	if (use_vbios)
+		dce110_timing_generator_program_timing_generator(tg, timing);
+	else
+		dce110_timing_generator_v_program_blanking(tg, timing);
+}
+
+static void dce110_timing_generator_v_program_blank_color(
+		struct timing_generator *tg,
+		const struct tg_color *black_color)
+{
+	uint32_t addr = mmCRTCV_BLACK_COLOR;
+	uint32_t value = dm_read_reg(tg->ctx, addr);
+
+	set_reg_field_value(
+		value,
+		black_color->color_b_cb,
+		CRTCV_BLACK_COLOR,
+		CRTC_BLACK_COLOR_B_CB);
+	set_reg_field_value(
+		value,
+		black_color->color_g_y,
+		CRTCV_BLACK_COLOR,
+		CRTC_BLACK_COLOR_G_Y);
+	set_reg_field_value(
+		value,
+		black_color->color_r_cr,
+		CRTCV_BLACK_COLOR,
+		CRTC_BLACK_COLOR_R_CR);
+
+	dm_write_reg(tg->ctx, addr, value);
+}
+
+static void dce110_timing_generator_v_set_overscan_color_black(
+	struct timing_generator *tg,
+	const struct tg_color *color)
+{
+	struct dc_context *ctx = tg->ctx;
+	uint32_t addr;
+	uint32_t value = 0;
+
+	set_reg_field_value(
+			value,
+			color->color_b_cb,
+			CRTC_OVERSCAN_COLOR,
+			CRTC_OVERSCAN_COLOR_BLUE);
+
+	set_reg_field_value(
+			value,
+			color->color_r_cr,
+			CRTC_OVERSCAN_COLOR,
+			CRTC_OVERSCAN_COLOR_RED);
+
+	set_reg_field_value(
+			value,
+			color->color_g_y,
+			CRTC_OVERSCAN_COLOR,
+			CRTC_OVERSCAN_COLOR_GREEN);
+
+	addr = mmCRTCV_OVERSCAN_COLOR;
+	dm_write_reg(ctx, addr, value);
+	addr = mmCRTCV_BLACK_COLOR;
+	dm_write_reg(ctx, addr, value);
+	/* This is desirable to have a constant DAC output voltage during the
+	 * blank time that is higher than the 0 volt reference level that the
+	 * DAC outputs when the NBLANK signal
+	 * is asserted low, such as for output to an analog TV. */
+	addr = mmCRTCV_BLANK_DATA_COLOR;
+	dm_write_reg(ctx, addr, value);
+
+	/* TO DO we have to program EXT registers and we need to know LB DATA
+	 * format because it is used when more 10 , i.e. 12 bits per color
+	 *
+	 * m_mmDxCRTC_OVERSCAN_COLOR_EXT
+	 * m_mmDxCRTC_BLACK_COLOR_EXT
+	 * m_mmDxCRTC_BLANK_DATA_COLOR_EXT
+	 */
+}
+
+static void dce110_tg_v_program_blank_color(struct timing_generator *tg,
+		const struct tg_color *black_color)
+{
+	uint32_t addr = mmCRTCV_BLACK_COLOR;
+	uint32_t value = dm_read_reg(tg->ctx, addr);
+
+	set_reg_field_value(
+		value,
+		black_color->color_b_cb,
+		CRTCV_BLACK_COLOR,
+		CRTC_BLACK_COLOR_B_CB);
+	set_reg_field_value(
+		value,
+		black_color->color_g_y,
+		CRTCV_BLACK_COLOR,
+		CRTC_BLACK_COLOR_G_Y);
+	set_reg_field_value(
+		value,
+		black_color->color_r_cr,
+		CRTCV_BLACK_COLOR,
+		CRTC_BLACK_COLOR_R_CR);
+
+	dm_write_reg(tg->ctx, addr, value);
+
+	addr = mmCRTCV_BLANK_DATA_COLOR;
+	dm_write_reg(tg->ctx, addr, value);
+}
+
+static void dce110_timing_generator_v_set_overscan_color(struct timing_generator *tg,
+	const struct tg_color *overscan_color)
+{
+	struct dc_context *ctx = tg->ctx;
+	uint32_t value = 0;
+	uint32_t addr;
+
+	set_reg_field_value(
+		value,
+		overscan_color->color_b_cb,
+		CRTCV_OVERSCAN_COLOR,
+		CRTC_OVERSCAN_COLOR_BLUE);
+
+	set_reg_field_value(
+		value,
+		overscan_color->color_g_y,
+		CRTCV_OVERSCAN_COLOR,
+		CRTC_OVERSCAN_COLOR_GREEN);
+
+	set_reg_field_value(
+		value,
+		overscan_color->color_r_cr,
+		CRTCV_OVERSCAN_COLOR,
+		CRTC_OVERSCAN_COLOR_RED);
+
+	addr = mmCRTCV_OVERSCAN_COLOR;
+	dm_write_reg(ctx, addr, value);
+}
+
+static void dce110_timing_generator_v_set_colors(struct timing_generator *tg,
+	const struct tg_color *blank_color,
+	const struct tg_color *overscan_color)
+{
+	if (blank_color != NULL)
+		dce110_tg_v_program_blank_color(tg, blank_color);
+	if (overscan_color != NULL)
+		dce110_timing_generator_v_set_overscan_color(tg, overscan_color);
+}
+
+static void dce110_timing_generator_v_set_early_control(
+		struct timing_generator *tg,
+		uint32_t early_cntl)
+{
+	uint32_t regval;
+	uint32_t address = mmCRTC_CONTROL;
+
+	regval = dm_read_reg(tg->ctx, address);
+	set_reg_field_value(regval, early_cntl,
+			CRTCV_CONTROL, CRTC_HBLANK_EARLY_CONTROL);
+	dm_write_reg(tg->ctx, address, regval);
+}
+
+static uint32_t dce110_timing_generator_v_get_vblank_counter(struct timing_generator *tg)
+{
+	uint32_t addr = mmCRTCV_STATUS_FRAME_COUNT;
+	uint32_t value = dm_read_reg(tg->ctx, addr);
+	uint32_t field = get_reg_field_value(
+			value, CRTCV_STATUS_FRAME_COUNT, CRTC_FRAME_COUNT);
+
+	return field;
+}
+
+static bool dce110_timing_generator_v_did_triggered_reset_occur(
+	struct timing_generator *tg)
+{
+	DC_LOG_ERROR("Timing Sync not supported on underlay pipe\n");
+	return false;
+}
+
+static void dce110_timing_generator_v_setup_global_swap_lock(
+	struct timing_generator *tg,
+	const struct dcp_gsl_params *gsl_params)
+{
+	DC_LOG_ERROR("Timing Sync not supported on underlay pipe\n");
+	return;
+}
+
+static void dce110_timing_generator_v_enable_reset_trigger(
+	struct timing_generator *tg,
+	int source_tg_inst)
+{
+	DC_LOG_ERROR("Timing Sync not supported on underlay pipe\n");
+	return;
+}
+
+static void dce110_timing_generator_v_disable_reset_trigger(
+	struct timing_generator *tg)
+{
+	DC_LOG_ERROR("Timing Sync not supported on underlay pipe\n");
+	return;
+}
+
+static void dce110_timing_generator_v_tear_down_global_swap_lock(
+	struct timing_generator *tg)
+{
+	DC_LOG_ERROR("Timing Sync not supported on underlay pipe\n");
+	return;
+}
+
+static void dce110_timing_generator_v_disable_vga(
+	struct timing_generator *tg)
+{
+	return;
+}
+
+/** ********************************************************************************************
+ *
+ * DCE11 Timing Generator Constructor / Destructor
+ *
+ *********************************************************************************************/
+static const struct timing_generator_funcs dce110_tg_v_funcs = {
+		.validate_timing = dce110_tg_validate_timing,
+		.program_timing = dce110_timing_generator_v_program_timing,
+		.enable_crtc = dce110_timing_generator_v_enable_crtc,
+		.disable_crtc = dce110_timing_generator_v_disable_crtc,
+		.is_counter_moving = dce110_timing_generator_v_is_counter_moving,
+		.get_position = NULL, /* Not to be implemented for underlay*/
+		.get_frame_count = dce110_timing_generator_v_get_vblank_counter,
+		.set_early_control = dce110_timing_generator_v_set_early_control,
+		.wait_for_state = dce110_timing_generator_v_wait_for_state,
+		.set_blank = dce110_timing_generator_v_set_blank,
+		.set_colors = dce110_timing_generator_v_set_colors,
+		.set_overscan_blank_color =
+				dce110_timing_generator_v_set_overscan_color_black,
+		.set_blank_color = dce110_timing_generator_v_program_blank_color,
+		.disable_vga = dce110_timing_generator_v_disable_vga,
+		.did_triggered_reset_occur =
+				dce110_timing_generator_v_did_triggered_reset_occur,
+		.setup_global_swap_lock =
+				dce110_timing_generator_v_setup_global_swap_lock,
+		.enable_reset_trigger = dce110_timing_generator_v_enable_reset_trigger,
+		.disable_reset_trigger = dce110_timing_generator_v_disable_reset_trigger,
+		.tear_down_global_swap_lock =
+				dce110_timing_generator_v_tear_down_global_swap_lock,
+		.enable_advanced_request =
+				dce110_timing_generator_v_enable_advanced_request,
+		.is_two_pixels_per_container = dce110_is_two_pixels_per_container,
+};
+
+void dce110_timing_generator_v_construct(
+	struct dce110_timing_generator *tg110,
+	struct dc_context *ctx)
+{
+	tg110->controller_id = CONTROLLER_ID_UNDERLAY0;
+
+	tg110->base.funcs = &dce110_tg_v_funcs;
+
+	tg110->base.ctx = ctx;
+	tg110->base.bp = ctx->dc_bios;
+
+	tg110->max_h_total = CRTC_H_TOTAL__CRTC_H_TOTAL_MASK + 1;
+	tg110->max_v_total = CRTC_V_TOTAL__CRTC_V_TOTAL_MASK + 1;
+
+	tg110->min_h_blank = 56;
+	tg110->min_h_front_porch = 4;
+	tg110->min_h_back_porch = 4;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator_v.h b/drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator_v.h
new file mode 100644
index 000000000000..d2623a5994e8
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator_v.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_TIMING_GENERATOR_V_DCE110_H__
+#define __DC_TIMING_GENERATOR_V_DCE110_H__
+
+void dce110_timing_generator_v_construct(
+	struct dce110_timing_generator *tg110,
+	struct dc_context *ctx);
+
+#endif /* __DC_TIMING_GENERATOR_V_DCE110_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_transform_v.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_transform_v.c
new file mode 100644
index 000000000000..28d3b2663cd3
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_transform_v.c
@@ -0,0 +1,717 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dce110_transform_v.h"
+#include "dm_services.h"
+#include "dc.h"
+#include "dce/dce_11_0_d.h"
+#include "dce/dce_11_0_sh_mask.h"
+
+#define SCLV_PHASES 64
+#define DC_LOGGER \
+	xfm->ctx->logger
+
+struct sclv_ratios_inits {
+	uint32_t h_int_scale_ratio_luma;
+	uint32_t h_int_scale_ratio_chroma;
+	uint32_t v_int_scale_ratio_luma;
+	uint32_t v_int_scale_ratio_chroma;
+	struct init_int_and_frac h_init_luma;
+	struct init_int_and_frac h_init_chroma;
+	struct init_int_and_frac v_init_luma;
+	struct init_int_and_frac v_init_chroma;
+};
+
+static void calculate_viewport(
+		const struct scaler_data *scl_data,
+		struct rect *luma_viewport,
+		struct rect *chroma_viewport)
+{
+	/*Do not set chroma vp for rgb444 pixel format*/
+	luma_viewport->x = scl_data->viewport.x - scl_data->viewport.x % 2;
+	luma_viewport->y = scl_data->viewport.y - scl_data->viewport.y % 2;
+	luma_viewport->width =
+		scl_data->viewport.width - scl_data->viewport.width % 2;
+	luma_viewport->height =
+		scl_data->viewport.height - scl_data->viewport.height % 2;
+	chroma_viewport->x = luma_viewport->x;
+	chroma_viewport->y = luma_viewport->y;
+	chroma_viewport->height = luma_viewport->height;
+	chroma_viewport->width = luma_viewport->width;
+
+	if (scl_data->format == PIXEL_FORMAT_420BPP8) {
+		luma_viewport->height += luma_viewport->height % 2;
+		luma_viewport->width += luma_viewport->width % 2;
+		/*for 420 video chroma is 1/4 the area of luma, scaled
+		 *vertically and horizontally
+		 */
+		chroma_viewport->x = luma_viewport->x / 2;
+		chroma_viewport->y = luma_viewport->y / 2;
+		chroma_viewport->height = luma_viewport->height / 2;
+		chroma_viewport->width = luma_viewport->width / 2;
+	}
+}
+
+static void program_viewport(
+	struct dce_transform *xfm_dce,
+	struct rect *luma_view_port,
+	struct rect *chroma_view_port)
+{
+	struct dc_context *ctx = xfm_dce->base.ctx;
+	uint32_t value = 0;
+	uint32_t addr = 0;
+
+	if (luma_view_port->width != 0 && luma_view_port->height != 0) {
+		addr = mmSCLV_VIEWPORT_START;
+		value = 0;
+		set_reg_field_value(
+			value,
+			luma_view_port->x,
+			SCLV_VIEWPORT_START,
+			VIEWPORT_X_START);
+		set_reg_field_value(
+			value,
+			luma_view_port->y,
+			SCLV_VIEWPORT_START,
+			VIEWPORT_Y_START);
+		dm_write_reg(ctx, addr, value);
+
+		addr = mmSCLV_VIEWPORT_SIZE;
+		value = 0;
+		set_reg_field_value(
+			value,
+			luma_view_port->height,
+			SCLV_VIEWPORT_SIZE,
+			VIEWPORT_HEIGHT);
+		set_reg_field_value(
+			value,
+			luma_view_port->width,
+			SCLV_VIEWPORT_SIZE,
+			VIEWPORT_WIDTH);
+		dm_write_reg(ctx, addr, value);
+	}
+
+	if (chroma_view_port->width != 0 && chroma_view_port->height != 0) {
+		addr = mmSCLV_VIEWPORT_START_C;
+		value = 0;
+		set_reg_field_value(
+			value,
+			chroma_view_port->x,
+			SCLV_VIEWPORT_START_C,
+			VIEWPORT_X_START_C);
+		set_reg_field_value(
+			value,
+			chroma_view_port->y,
+			SCLV_VIEWPORT_START_C,
+			VIEWPORT_Y_START_C);
+		dm_write_reg(ctx, addr, value);
+
+		addr = mmSCLV_VIEWPORT_SIZE_C;
+		value = 0;
+		set_reg_field_value(
+			value,
+			chroma_view_port->height,
+			SCLV_VIEWPORT_SIZE_C,
+			VIEWPORT_HEIGHT_C);
+		set_reg_field_value(
+			value,
+			chroma_view_port->width,
+			SCLV_VIEWPORT_SIZE_C,
+			VIEWPORT_WIDTH_C);
+		dm_write_reg(ctx, addr, value);
+	}
+}
+
+/*
+ * Function:
+ * void setup_scaling_configuration
+ *
+ * Purpose: setup scaling mode : bypass, RGb, YCbCr and nummber of taps
+ * Input:   data
+ *
+ * Output:
+ *  void
+ */
+static bool setup_scaling_configuration(
+	struct dce_transform *xfm_dce,
+	const struct scaler_data *data)
+{
+	bool is_scaling_needed = false;
+	struct dc_context *ctx = xfm_dce->base.ctx;
+	uint32_t value = 0;
+
+	set_reg_field_value(value, data->taps.h_taps - 1,
+			SCLV_TAP_CONTROL, SCL_H_NUM_OF_TAPS);
+	set_reg_field_value(value, data->taps.v_taps - 1,
+			SCLV_TAP_CONTROL, SCL_V_NUM_OF_TAPS);
+	set_reg_field_value(value, data->taps.h_taps_c - 1,
+			SCLV_TAP_CONTROL, SCL_H_NUM_OF_TAPS_C);
+	set_reg_field_value(value, data->taps.v_taps_c - 1,
+			SCLV_TAP_CONTROL, SCL_V_NUM_OF_TAPS_C);
+	dm_write_reg(ctx, mmSCLV_TAP_CONTROL, value);
+
+	value = 0;
+	if (data->taps.h_taps + data->taps.v_taps > 2) {
+		set_reg_field_value(value, 1, SCLV_MODE, SCL_MODE);
+		set_reg_field_value(value, 1, SCLV_MODE, SCL_PSCL_EN);
+		is_scaling_needed = true;
+	} else {
+		set_reg_field_value(value, 0, SCLV_MODE, SCL_MODE);
+		set_reg_field_value(value, 0, SCLV_MODE, SCL_PSCL_EN);
+	}
+
+	if (data->taps.h_taps_c + data->taps.v_taps_c > 2) {
+		set_reg_field_value(value, 1, SCLV_MODE, SCL_MODE_C);
+		set_reg_field_value(value, 1, SCLV_MODE, SCL_PSCL_EN_C);
+		is_scaling_needed = true;
+	} else if (data->format != PIXEL_FORMAT_420BPP8) {
+		set_reg_field_value(
+			value,
+			get_reg_field_value(value, SCLV_MODE, SCL_MODE),
+			SCLV_MODE,
+			SCL_MODE_C);
+		set_reg_field_value(
+			value,
+			get_reg_field_value(value, SCLV_MODE, SCL_PSCL_EN),
+			SCLV_MODE,
+			SCL_PSCL_EN_C);
+	} else {
+		set_reg_field_value(value, 0, SCLV_MODE, SCL_MODE_C);
+		set_reg_field_value(value, 0, SCLV_MODE, SCL_PSCL_EN_C);
+	}
+	dm_write_reg(ctx, mmSCLV_MODE, value);
+
+	value = 0;
+	/*
+	 * 0 - Replaced out of bound pixels with black pixel
+	 * (or any other required color)
+	 * 1 - Replaced out of bound pixels with the edge pixel
+	 */
+	set_reg_field_value(value, 1, SCLV_CONTROL, SCL_BOUNDARY_MODE);
+	dm_write_reg(ctx, mmSCLV_CONTROL, value);
+
+	return is_scaling_needed;
+}
+
+/*
+ * Function:
+ * void program_overscan
+ *
+ * Purpose: Programs overscan border
+ * Input:   overscan
+ *
+ * Output: void
+ */
+static void program_overscan(
+		struct dce_transform *xfm_dce,
+		const struct scaler_data *data)
+{
+	uint32_t overscan_left_right = 0;
+	uint32_t overscan_top_bottom = 0;
+
+	int overscan_right = data->h_active - data->recout.x - data->recout.width;
+	int overscan_bottom = data->v_active - data->recout.y - data->recout.height;
+
+	if (xfm_dce->base.ctx->dc->debug.visual_confirm != VISUAL_CONFIRM_DISABLE) {
+		overscan_bottom += 2;
+		overscan_right += 2;
+	}
+
+	if (overscan_right < 0) {
+		BREAK_TO_DEBUGGER();
+		overscan_right = 0;
+	}
+	if (overscan_bottom < 0) {
+		BREAK_TO_DEBUGGER();
+		overscan_bottom = 0;
+	}
+
+	set_reg_field_value(overscan_left_right, data->recout.x,
+			EXT_OVERSCAN_LEFT_RIGHT, EXT_OVERSCAN_LEFT);
+
+	set_reg_field_value(overscan_left_right, overscan_right,
+			EXT_OVERSCAN_LEFT_RIGHT, EXT_OVERSCAN_RIGHT);
+
+	set_reg_field_value(overscan_top_bottom, data->recout.y,
+			EXT_OVERSCAN_TOP_BOTTOM, EXT_OVERSCAN_TOP);
+
+	set_reg_field_value(overscan_top_bottom, overscan_bottom,
+			EXT_OVERSCAN_TOP_BOTTOM, EXT_OVERSCAN_BOTTOM);
+
+	dm_write_reg(xfm_dce->base.ctx,
+			mmSCLV_EXT_OVERSCAN_LEFT_RIGHT,
+			overscan_left_right);
+
+	dm_write_reg(xfm_dce->base.ctx,
+			mmSCLV_EXT_OVERSCAN_TOP_BOTTOM,
+			overscan_top_bottom);
+}
+
+static void set_coeff_update_complete(
+		struct dce_transform *xfm_dce)
+{
+	uint32_t value;
+
+	value = dm_read_reg(xfm_dce->base.ctx, mmSCLV_UPDATE);
+	set_reg_field_value(value, 1, SCLV_UPDATE, SCL_COEF_UPDATE_COMPLETE);
+	dm_write_reg(xfm_dce->base.ctx, mmSCLV_UPDATE, value);
+}
+
+static void program_multi_taps_filter(
+	struct dce_transform *xfm_dce,
+	int taps,
+	const uint16_t *coeffs,
+	enum ram_filter_type filter_type)
+{
+	struct dc_context *ctx = xfm_dce->base.ctx;
+	int i, phase, pair;
+	int array_idx = 0;
+	int taps_pairs = (taps + 1) / 2;
+	int phases_to_program = SCLV_PHASES / 2 + 1;
+
+	uint32_t select = 0;
+	uint32_t power_ctl, power_ctl_off;
+
+	if (!coeffs)
+		return;
+
+	/*We need to disable power gating on coeff memory to do programming*/
+	power_ctl = dm_read_reg(ctx, mmDCFEV_MEM_PWR_CTRL);
+	power_ctl_off = power_ctl;
+	set_reg_field_value(power_ctl_off, 1, DCFEV_MEM_PWR_CTRL, SCLV_COEFF_MEM_PWR_DIS);
+	dm_write_reg(ctx, mmDCFEV_MEM_PWR_CTRL, power_ctl_off);
+
+	/*Wait to disable gating:*/
+	for (i = 0; i < 10; i++) {
+		if (get_reg_field_value(
+				dm_read_reg(ctx, mmDCFEV_MEM_PWR_STATUS),
+				DCFEV_MEM_PWR_STATUS,
+				SCLV_COEFF_MEM_PWR_STATE) == 0)
+			break;
+
+		udelay(1);
+	}
+
+	set_reg_field_value(select, filter_type, SCLV_COEF_RAM_SELECT, SCL_C_RAM_FILTER_TYPE);
+
+	for (phase = 0; phase < phases_to_program; phase++) {
+		/*we always program N/2 + 1 phases, total phases N, but N/2-1 are just mirror
+		phase 0 is unique and phase N/2 is unique if N is even*/
+		set_reg_field_value(select, phase, SCLV_COEF_RAM_SELECT, SCL_C_RAM_PHASE);
+		for (pair = 0; pair < taps_pairs; pair++) {
+			uint32_t data = 0;
+
+			set_reg_field_value(select, pair,
+					SCLV_COEF_RAM_SELECT, SCL_C_RAM_TAP_PAIR_IDX);
+
+			dm_write_reg(ctx, mmSCLV_COEF_RAM_SELECT, select);
+
+			set_reg_field_value(
+					data, 1,
+					SCLV_COEF_RAM_TAP_DATA,
+					SCL_C_RAM_EVEN_TAP_COEF_EN);
+			set_reg_field_value(
+					data, coeffs[array_idx],
+					SCLV_COEF_RAM_TAP_DATA,
+					SCL_C_RAM_EVEN_TAP_COEF);
+
+			if (taps % 2 && pair == taps_pairs - 1) {
+				set_reg_field_value(
+						data, 0,
+						SCLV_COEF_RAM_TAP_DATA,
+						SCL_C_RAM_ODD_TAP_COEF_EN);
+				array_idx++;
+			} else {
+				set_reg_field_value(
+						data, 1,
+						SCLV_COEF_RAM_TAP_DATA,
+						SCL_C_RAM_ODD_TAP_COEF_EN);
+				set_reg_field_value(
+						data, coeffs[array_idx + 1],
+						SCLV_COEF_RAM_TAP_DATA,
+						SCL_C_RAM_ODD_TAP_COEF);
+
+				array_idx += 2;
+			}
+
+			dm_write_reg(ctx, mmSCLV_COEF_RAM_TAP_DATA, data);
+		}
+	}
+
+	/*We need to restore power gating on coeff memory to initial state*/
+	dm_write_reg(ctx, mmDCFEV_MEM_PWR_CTRL, power_ctl);
+}
+
+static void calculate_inits(
+	struct dce_transform *xfm_dce,
+	const struct scaler_data *data,
+	struct sclv_ratios_inits *inits,
+	struct rect *luma_viewport,
+	struct rect *chroma_viewport)
+{
+	inits->h_int_scale_ratio_luma =
+		dc_fixpt_u2d19(data->ratios.horz) << 5;
+	inits->v_int_scale_ratio_luma =
+		dc_fixpt_u2d19(data->ratios.vert) << 5;
+	inits->h_int_scale_ratio_chroma =
+		dc_fixpt_u2d19(data->ratios.horz_c) << 5;
+	inits->v_int_scale_ratio_chroma =
+		dc_fixpt_u2d19(data->ratios.vert_c) << 5;
+
+	inits->h_init_luma.integer = 1;
+	inits->v_init_luma.integer = 1;
+	inits->h_init_chroma.integer = 1;
+	inits->v_init_chroma.integer = 1;
+}
+
+static void program_scl_ratios_inits(
+	struct dce_transform *xfm_dce,
+	struct sclv_ratios_inits *inits)
+{
+	struct dc_context *ctx = xfm_dce->base.ctx;
+	uint32_t addr = mmSCLV_HORZ_FILTER_SCALE_RATIO;
+	uint32_t value = 0;
+
+	set_reg_field_value(
+		value,
+		inits->h_int_scale_ratio_luma,
+		SCLV_HORZ_FILTER_SCALE_RATIO,
+		SCL_H_SCALE_RATIO);
+	dm_write_reg(ctx, addr, value);
+
+	addr = mmSCLV_VERT_FILTER_SCALE_RATIO;
+	value = 0;
+	set_reg_field_value(
+		value,
+		inits->v_int_scale_ratio_luma,
+		SCLV_VERT_FILTER_SCALE_RATIO,
+		SCL_V_SCALE_RATIO);
+	dm_write_reg(ctx, addr, value);
+
+	addr = mmSCLV_HORZ_FILTER_SCALE_RATIO_C;
+	value = 0;
+	set_reg_field_value(
+		value,
+		inits->h_int_scale_ratio_chroma,
+		SCLV_HORZ_FILTER_SCALE_RATIO_C,
+		SCL_H_SCALE_RATIO_C);
+	dm_write_reg(ctx, addr, value);
+
+	addr = mmSCLV_VERT_FILTER_SCALE_RATIO_C;
+	value = 0;
+	set_reg_field_value(
+		value,
+		inits->v_int_scale_ratio_chroma,
+		SCLV_VERT_FILTER_SCALE_RATIO_C,
+		SCL_V_SCALE_RATIO_C);
+	dm_write_reg(ctx, addr, value);
+
+	addr = mmSCLV_HORZ_FILTER_INIT;
+	value = 0;
+	set_reg_field_value(
+		value,
+		inits->h_init_luma.fraction,
+		SCLV_HORZ_FILTER_INIT,
+		SCL_H_INIT_FRAC);
+	set_reg_field_value(
+		value,
+		inits->h_init_luma.integer,
+		SCLV_HORZ_FILTER_INIT,
+		SCL_H_INIT_INT);
+	dm_write_reg(ctx, addr, value);
+
+	addr = mmSCLV_VERT_FILTER_INIT;
+	value = 0;
+	set_reg_field_value(
+		value,
+		inits->v_init_luma.fraction,
+		SCLV_VERT_FILTER_INIT,
+		SCL_V_INIT_FRAC);
+	set_reg_field_value(
+		value,
+		inits->v_init_luma.integer,
+		SCLV_VERT_FILTER_INIT,
+		SCL_V_INIT_INT);
+	dm_write_reg(ctx, addr, value);
+
+	addr = mmSCLV_HORZ_FILTER_INIT_C;
+	value = 0;
+	set_reg_field_value(
+		value,
+		inits->h_init_chroma.fraction,
+		SCLV_HORZ_FILTER_INIT_C,
+		SCL_H_INIT_FRAC_C);
+	set_reg_field_value(
+		value,
+		inits->h_init_chroma.integer,
+		SCLV_HORZ_FILTER_INIT_C,
+		SCL_H_INIT_INT_C);
+	dm_write_reg(ctx, addr, value);
+
+	addr = mmSCLV_VERT_FILTER_INIT_C;
+	value = 0;
+	set_reg_field_value(
+		value,
+		inits->v_init_chroma.fraction,
+		SCLV_VERT_FILTER_INIT_C,
+		SCL_V_INIT_FRAC_C);
+	set_reg_field_value(
+		value,
+		inits->v_init_chroma.integer,
+		SCLV_VERT_FILTER_INIT_C,
+		SCL_V_INIT_INT_C);
+	dm_write_reg(ctx, addr, value);
+}
+
+static const uint16_t *get_filter_coeffs_64p(int taps, struct fixed31_32 ratio)
+{
+	if (taps == 4)
+		return get_filter_4tap_64p(ratio);
+	else if (taps == 2)
+		return get_filter_2tap_64p();
+	else if (taps == 1)
+		return NULL;
+	else {
+		/* should never happen, bug */
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+}
+
+static bool dce110_xfmv_power_up_line_buffer(struct transform *xfm)
+{
+	struct dce_transform *xfm_dce = TO_DCE_TRANSFORM(xfm);
+	uint32_t value;
+
+	value = dm_read_reg(xfm_dce->base.ctx, mmLBV_MEMORY_CTRL);
+
+	/*Use all three pieces of memory always*/
+	set_reg_field_value(value, 0, LBV_MEMORY_CTRL, LB_MEMORY_CONFIG);
+	/*hard coded number DCE11 1712(0x6B0) Partitions: 720/960/1712*/
+	set_reg_field_value(value, xfm_dce->lb_memory_size, LBV_MEMORY_CTRL,
+			LB_MEMORY_SIZE);
+
+	dm_write_reg(xfm_dce->base.ctx, mmLBV_MEMORY_CTRL, value);
+
+	return true;
+}
+
+static void dce110_xfmv_set_scaler(
+	struct transform *xfm,
+	const struct scaler_data *data)
+{
+	struct dce_transform *xfm_dce = TO_DCE_TRANSFORM(xfm);
+	bool is_scaling_required = false;
+	bool filter_updated = false;
+	const uint16_t *coeffs_v, *coeffs_h, *coeffs_h_c, *coeffs_v_c;
+	struct rect luma_viewport = {0};
+	struct rect chroma_viewport = {0};
+
+	dce110_xfmv_power_up_line_buffer(xfm);
+	/* 1. Calculate viewport, viewport programming should happen after init
+	 * calculations as they may require an adjustment in the viewport.
+	 */
+
+	calculate_viewport(data, &luma_viewport, &chroma_viewport);
+
+	/* 2. Program overscan */
+	program_overscan(xfm_dce, data);
+
+	/* 3. Program taps and configuration */
+	is_scaling_required = setup_scaling_configuration(xfm_dce, data);
+
+	if (is_scaling_required) {
+		/* 4. Calculate and program ratio, filter initialization */
+
+		struct sclv_ratios_inits inits = { 0 };
+
+		calculate_inits(
+			xfm_dce,
+			data,
+			&inits,
+			&luma_viewport,
+			&chroma_viewport);
+
+		program_scl_ratios_inits(xfm_dce, &inits);
+
+		coeffs_v = get_filter_coeffs_64p(data->taps.v_taps, data->ratios.vert);
+		coeffs_h = get_filter_coeffs_64p(data->taps.h_taps, data->ratios.horz);
+		coeffs_v_c = get_filter_coeffs_64p(data->taps.v_taps_c, data->ratios.vert_c);
+		coeffs_h_c = get_filter_coeffs_64p(data->taps.h_taps_c, data->ratios.horz_c);
+
+		if (coeffs_v != xfm_dce->filter_v
+				|| coeffs_v_c != xfm_dce->filter_v_c
+				|| coeffs_h != xfm_dce->filter_h
+				|| coeffs_h_c != xfm_dce->filter_h_c) {
+		/* 5. Program vertical filters */
+			program_multi_taps_filter(
+					xfm_dce,
+					data->taps.v_taps,
+					coeffs_v,
+					FILTER_TYPE_RGB_Y_VERTICAL);
+			program_multi_taps_filter(
+					xfm_dce,
+					data->taps.v_taps_c,
+					coeffs_v_c,
+					FILTER_TYPE_CBCR_VERTICAL);
+
+		/* 6. Program horizontal filters */
+			program_multi_taps_filter(
+					xfm_dce,
+					data->taps.h_taps,
+					coeffs_h,
+					FILTER_TYPE_RGB_Y_HORIZONTAL);
+			program_multi_taps_filter(
+					xfm_dce,
+					data->taps.h_taps_c,
+					coeffs_h_c,
+					FILTER_TYPE_CBCR_HORIZONTAL);
+
+			xfm_dce->filter_v = coeffs_v;
+			xfm_dce->filter_v_c = coeffs_v_c;
+			xfm_dce->filter_h = coeffs_h;
+			xfm_dce->filter_h_c = coeffs_h_c;
+			filter_updated = true;
+		}
+	}
+
+	/* 7. Program the viewport */
+	program_viewport(xfm_dce, &luma_viewport, &chroma_viewport);
+
+	/* 8. Set bit to flip to new coefficient memory */
+	if (filter_updated)
+		set_coeff_update_complete(xfm_dce);
+}
+
+static void dce110_xfmv_reset(struct transform *xfm)
+{
+	struct dce_transform *xfm_dce = TO_DCE_TRANSFORM(xfm);
+
+	xfm_dce->filter_h = NULL;
+	xfm_dce->filter_v = NULL;
+	xfm_dce->filter_h_c = NULL;
+	xfm_dce->filter_v_c = NULL;
+}
+
+static void dce110_xfmv_set_gamut_remap(
+	struct transform *xfm,
+	const struct xfm_grph_csc_adjustment *adjust)
+{
+	/* DO NOTHING*/
+}
+
+static void dce110_xfmv_set_pixel_storage_depth(
+	struct transform *xfm,
+	enum lb_pixel_depth depth,
+	const struct bit_depth_reduction_params *bit_depth_params)
+{
+	struct dce_transform *xfm_dce = TO_DCE_TRANSFORM(xfm);
+	int pixel_depth = 0;
+	int expan_mode = 0;
+	uint32_t reg_data = 0;
+
+	switch (depth) {
+	case LB_PIXEL_DEPTH_18BPP:
+		pixel_depth = 2;
+		expan_mode  = 1;
+		break;
+	case LB_PIXEL_DEPTH_24BPP:
+		pixel_depth = 1;
+		expan_mode  = 1;
+		break;
+	case LB_PIXEL_DEPTH_30BPP:
+		pixel_depth = 0;
+		expan_mode  = 1;
+		break;
+	case LB_PIXEL_DEPTH_36BPP:
+		pixel_depth = 3;
+		expan_mode  = 0;
+		break;
+	default:
+		BREAK_TO_DEBUGGER();
+		break;
+	}
+
+	set_reg_field_value(
+		reg_data,
+		expan_mode,
+		LBV_DATA_FORMAT,
+		PIXEL_EXPAN_MODE);
+
+	set_reg_field_value(
+		reg_data,
+		pixel_depth,
+		LBV_DATA_FORMAT,
+		PIXEL_DEPTH);
+
+	dm_write_reg(xfm->ctx, mmLBV_DATA_FORMAT, reg_data);
+
+	if (!(xfm_dce->lb_pixel_depth_supported & depth)) {
+		/*we should use unsupported capabilities
+		 *  unless it is required by w/a*/
+		DC_LOG_WARNING("%s: Capability not supported",
+			__func__);
+	}
+}
+
+static const struct transform_funcs dce110_xfmv_funcs = {
+	.transform_reset = dce110_xfmv_reset,
+	.transform_set_scaler = dce110_xfmv_set_scaler,
+	.transform_set_gamut_remap =
+		dce110_xfmv_set_gamut_remap,
+	.opp_set_csc_default = dce110_opp_v_set_csc_default,
+	.opp_set_csc_adjustment = dce110_opp_v_set_csc_adjustment,
+	.opp_power_on_regamma_lut = dce110_opp_power_on_regamma_lut_v,
+	.opp_program_regamma_pwl = dce110_opp_program_regamma_pwl_v,
+	.opp_set_regamma_mode = dce110_opp_set_regamma_mode_v,
+	.transform_set_pixel_storage_depth =
+			dce110_xfmv_set_pixel_storage_depth,
+	.transform_get_optimal_number_of_taps =
+		dce_transform_get_optimal_number_of_taps
+};
+/*****************************************/
+/* Constructor, Destructor               */
+/*****************************************/
+
+bool dce110_transform_v_construct(
+	struct dce_transform *xfm_dce,
+	struct dc_context *ctx)
+{
+	xfm_dce->base.ctx = ctx;
+
+	xfm_dce->base.funcs = &dce110_xfmv_funcs;
+
+	xfm_dce->lb_pixel_depth_supported =
+			LB_PIXEL_DEPTH_18BPP |
+			LB_PIXEL_DEPTH_24BPP |
+			LB_PIXEL_DEPTH_30BPP |
+			LB_PIXEL_DEPTH_36BPP;
+
+	xfm_dce->prescaler_on = true;
+	xfm_dce->lb_bits_per_entry = LB_BITS_PER_ENTRY;
+	xfm_dce->lb_memory_size = LB_TOTAL_NUMBER_OF_ENTRIES; /*0x6B0*/
+
+	return true;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_transform_v.h b/drivers/gpu/drm/amd/display/dc/dce110/dce110_transform_v.h
new file mode 100644
index 000000000000..b70780210aad
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_transform_v.h
@@ -0,0 +1,58 @@
+/* Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_TRANSFORM_V_DCE110_H__
+#define __DAL_TRANSFORM_V_DCE110_H__
+
+#include "../dce/dce_transform.h"
+
+#define LB_TOTAL_NUMBER_OF_ENTRIES 1712
+#define LB_BITS_PER_ENTRY 144
+
+bool dce110_transform_v_construct(
+	struct dce_transform *xfm110,
+	struct dc_context *ctx);
+
+void dce110_opp_v_set_csc_default(
+	struct transform *xfm,
+	const struct default_adjustment *default_adjust);
+
+void dce110_opp_v_set_csc_adjustment(
+		struct transform *xfm,
+	const struct out_csc_color_matrix *tbl_entry);
+
+
+void dce110_opp_program_regamma_pwl_v(
+	struct transform *xfm,
+	const struct pwl_params *params);
+
+void dce110_opp_power_on_regamma_lut_v(
+	struct transform *xfm,
+	bool power_on);
+
+void dce110_opp_set_regamma_mode_v(
+	struct transform *xfm,
+	enum opp_regamma mode);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dce112/Makefile b/drivers/gpu/drm/amd/display/dc/dce112/Makefile
new file mode 100644
index 000000000000..683866797709
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce112/Makefile
@@ -0,0 +1,32 @@
+#
+# Copyright 2017 Advanced Micro Devices, Inc.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+#
+# Makefile for the 'controller' sub-component of DAL.
+# It provides the control and status of HW CRTC block.
+
+CFLAGS_$(AMDDALPATH)/dc/dce112/dce112_resource.o = -Wno-override-init
+
+DCE112 = dce112_compressor.o
+
+AMD_DAL_DCE112 = $(addprefix $(AMDDALPATH)/dc/dce112/,$(DCE112))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCE112)
diff --git a/drivers/gpu/drm/amd/display/dc/dce112/dce112_compressor.c b/drivers/gpu/drm/amd/display/dc/dce112/dce112_compressor.c
new file mode 100644
index 000000000000..faae12cf7968
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce112/dce112_compressor.c
@@ -0,0 +1,847 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+
+#include "dce/dce_11_2_d.h"
+#include "dce/dce_11_2_sh_mask.h"
+#include "gmc/gmc_8_1_sh_mask.h"
+#include "gmc/gmc_8_1_d.h"
+
+#include "include/logger_interface.h"
+
+#include "dce112_compressor.h"
+#define DC_LOGGER \
+		cp110->base.ctx->logger
+#define DCP_REG(reg)\
+	(reg + cp110->offsets.dcp_offset)
+#define DMIF_REG(reg)\
+	(reg + cp110->offsets.dmif_offset)
+
+static const struct dce112_compressor_reg_offsets reg_offsets[] = {
+{
+	.dcp_offset = (mmDCP0_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
+	.dmif_offset =
+		(mmDMIF_PG0_DPG_PIPE_DPM_CONTROL
+			- mmDMIF_PG0_DPG_PIPE_DPM_CONTROL),
+},
+{
+	.dcp_offset = (mmDCP1_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
+	.dmif_offset =
+		(mmDMIF_PG1_DPG_PIPE_DPM_CONTROL
+			- mmDMIF_PG0_DPG_PIPE_DPM_CONTROL),
+},
+{
+	.dcp_offset = (mmDCP2_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
+	.dmif_offset =
+		(mmDMIF_PG2_DPG_PIPE_DPM_CONTROL
+			- mmDMIF_PG0_DPG_PIPE_DPM_CONTROL),
+}
+};
+
+static const uint32_t dce11_one_lpt_channel_max_resolution = 2560 * 1600;
+
+enum fbc_idle_force {
+	/* Bit 0 - Display registers updated */
+	FBC_IDLE_FORCE_DISPLAY_REGISTER_UPDATE = 0x00000001,
+
+	/* Bit 2 - FBC_GRPH_COMP_EN register updated */
+	FBC_IDLE_FORCE_GRPH_COMP_EN = 0x00000002,
+	/* Bit 3 - FBC_SRC_SEL register updated */
+	FBC_IDLE_FORCE_SRC_SEL_CHANGE = 0x00000004,
+	/* Bit 4 - FBC_MIN_COMPRESSION register updated */
+	FBC_IDLE_FORCE_MIN_COMPRESSION_CHANGE = 0x00000008,
+	/* Bit 5 - FBC_ALPHA_COMP_EN register updated */
+	FBC_IDLE_FORCE_ALPHA_COMP_EN = 0x00000010,
+	/* Bit 6 - FBC_ZERO_ALPHA_CHUNK_SKIP_EN register updated */
+	FBC_IDLE_FORCE_ZERO_ALPHA_CHUNK_SKIP_EN = 0x00000020,
+	/* Bit 7 - FBC_FORCE_COPY_TO_COMP_BUF register updated */
+	FBC_IDLE_FORCE_FORCE_COPY_TO_COMP_BUF = 0x00000040,
+
+	/* Bit 24 - Memory write to region 0 defined by MC registers. */
+	FBC_IDLE_FORCE_MEMORY_WRITE_TO_REGION0 = 0x01000000,
+	/* Bit 25 - Memory write to region 1 defined by MC registers */
+	FBC_IDLE_FORCE_MEMORY_WRITE_TO_REGION1 = 0x02000000,
+	/* Bit 26 - Memory write to region 2 defined by MC registers */
+	FBC_IDLE_FORCE_MEMORY_WRITE_TO_REGION2 = 0x04000000,
+	/* Bit 27 - Memory write to region 3 defined by MC registers. */
+	FBC_IDLE_FORCE_MEMORY_WRITE_TO_REGION3 = 0x08000000,
+
+	/* Bit 28 - Memory write from any client other than MCIF */
+	FBC_IDLE_FORCE_MEMORY_WRITE_OTHER_THAN_MCIF = 0x10000000,
+	/* Bit 29 - CG statics screen signal is inactive */
+	FBC_IDLE_FORCE_CG_STATIC_SCREEN_IS_INACTIVE = 0x20000000,
+};
+
+static uint32_t lpt_size_alignment(struct dce112_compressor *cp110)
+{
+	/*LPT_ALIGNMENT (in bytes) = ROW_SIZE * #BANKS * # DRAM CHANNELS. */
+	return cp110->base.raw_size * cp110->base.banks_num *
+		cp110->base.dram_channels_num;
+}
+
+static uint32_t lpt_memory_control_config(struct dce112_compressor *cp110,
+	uint32_t lpt_control)
+{
+	/*LPT MC Config */
+	if (cp110->base.options.bits.LPT_MC_CONFIG == 1) {
+		/* POSSIBLE VALUES for LPT NUM_PIPES (DRAM CHANNELS):
+		 * 00 - 1 CHANNEL
+		 * 01 - 2 CHANNELS
+		 * 02 - 4 OR 6 CHANNELS
+		 * (Only for discrete GPU, N/A for CZ)
+		 * 03 - 8 OR 12 CHANNELS
+		 * (Only for discrete GPU, N/A for CZ) */
+		switch (cp110->base.dram_channels_num) {
+		case 2:
+			set_reg_field_value(
+				lpt_control,
+				1,
+				LOW_POWER_TILING_CONTROL,
+				LOW_POWER_TILING_NUM_PIPES);
+			break;
+		case 1:
+			set_reg_field_value(
+				lpt_control,
+				0,
+				LOW_POWER_TILING_CONTROL,
+				LOW_POWER_TILING_NUM_PIPES);
+			break;
+		default:
+			DC_LOG_WARNING(
+				"%s: Invalid LPT NUM_PIPES!!!",
+				__func__);
+			break;
+		}
+
+		/* The mapping for LPT NUM_BANKS is in
+		 * GRPH_CONTROL.GRPH_NUM_BANKS register field
+		 * Specifies the number of memory banks for tiling
+		 * purposes. Only applies to 2D and 3D tiling modes.
+		 * POSSIBLE VALUES:
+		 * 00 - DCP_GRPH_NUM_BANKS_2BANK: ADDR_SURF_2_BANK
+		 * 01 - DCP_GRPH_NUM_BANKS_4BANK: ADDR_SURF_4_BANK
+		 * 02 - DCP_GRPH_NUM_BANKS_8BANK: ADDR_SURF_8_BANK
+		 * 03 - DCP_GRPH_NUM_BANKS_16BANK: ADDR_SURF_16_BANK */
+		switch (cp110->base.banks_num) {
+		case 16:
+			set_reg_field_value(
+				lpt_control,
+				3,
+				LOW_POWER_TILING_CONTROL,
+				LOW_POWER_TILING_NUM_BANKS);
+			break;
+		case 8:
+			set_reg_field_value(
+				lpt_control,
+				2,
+				LOW_POWER_TILING_CONTROL,
+				LOW_POWER_TILING_NUM_BANKS);
+			break;
+		case 4:
+			set_reg_field_value(
+				lpt_control,
+				1,
+				LOW_POWER_TILING_CONTROL,
+				LOW_POWER_TILING_NUM_BANKS);
+			break;
+		case 2:
+			set_reg_field_value(
+				lpt_control,
+				0,
+				LOW_POWER_TILING_CONTROL,
+				LOW_POWER_TILING_NUM_BANKS);
+			break;
+		default:
+			DC_LOG_WARNING(
+				"%s: Invalid LPT NUM_BANKS!!!",
+				__func__);
+			break;
+		}
+
+		/* The mapping is in DMIF_ADDR_CALC.
+		 * ADDR_CONFIG_PIPE_INTERLEAVE_SIZE register field for
+		 * Carrizo specifies the memory interleave per pipe.
+		 * It effectively specifies the location of pipe bits in
+		 * the memory address.
+		 * POSSIBLE VALUES:
+		 * 00 - ADDR_CONFIG_PIPE_INTERLEAVE_256B: 256 byte
+		 * interleave
+		 * 01 - ADDR_CONFIG_PIPE_INTERLEAVE_512B: 512 byte
+		 * interleave
+		 */
+		switch (cp110->base.channel_interleave_size) {
+		case 256: /*256B */
+			set_reg_field_value(
+				lpt_control,
+				0,
+				LOW_POWER_TILING_CONTROL,
+				LOW_POWER_TILING_PIPE_INTERLEAVE_SIZE);
+			break;
+		case 512: /*512B */
+			set_reg_field_value(
+				lpt_control,
+				1,
+				LOW_POWER_TILING_CONTROL,
+				LOW_POWER_TILING_PIPE_INTERLEAVE_SIZE);
+			break;
+		default:
+			DC_LOG_WARNING(
+				"%s: Invalid LPT INTERLEAVE_SIZE!!!",
+				__func__);
+			break;
+		}
+
+		/* The mapping for LOW_POWER_TILING_ROW_SIZE is in
+		 * DMIF_ADDR_CALC.ADDR_CONFIG_ROW_SIZE register field
+		 * for Carrizo. Specifies the size of dram row in bytes.
+		 * This should match up with NOOFCOLS field in
+		 * MC_ARB_RAMCFG (ROW_SIZE = 4 * 2 ^^ columns).
+		 * This register DMIF_ADDR_CALC is not used by the
+		 * hardware as it is only used for addrlib assertions.
+		 * POSSIBLE VALUES:
+		 * 00 - ADDR_CONFIG_1KB_ROW: Treat 1KB as DRAM row
+		 * boundary
+		 * 01 - ADDR_CONFIG_2KB_ROW: Treat 2KB as DRAM row
+		 * boundary
+		 * 02 - ADDR_CONFIG_4KB_ROW: Treat 4KB as DRAM row
+		 * boundary */
+		switch (cp110->base.raw_size) {
+		case 4096: /*4 KB */
+			set_reg_field_value(
+				lpt_control,
+				2,
+				LOW_POWER_TILING_CONTROL,
+				LOW_POWER_TILING_ROW_SIZE);
+			break;
+		case 2048:
+			set_reg_field_value(
+				lpt_control,
+				1,
+				LOW_POWER_TILING_CONTROL,
+				LOW_POWER_TILING_ROW_SIZE);
+			break;
+		case 1024:
+			set_reg_field_value(
+				lpt_control,
+				0,
+				LOW_POWER_TILING_CONTROL,
+				LOW_POWER_TILING_ROW_SIZE);
+			break;
+		default:
+			DC_LOG_WARNING(
+				"%s: Invalid LPT ROW_SIZE!!!",
+				__func__);
+			break;
+		}
+	} else {
+		DC_LOG_WARNING(
+			"%s: LPT MC Configuration is not provided",
+			__func__);
+	}
+
+	return lpt_control;
+}
+
+static bool is_source_bigger_than_epanel_size(
+	struct dce112_compressor *cp110,
+	uint32_t source_view_width,
+	uint32_t source_view_height)
+{
+	if (cp110->base.embedded_panel_h_size != 0 &&
+		cp110->base.embedded_panel_v_size != 0 &&
+		((source_view_width * source_view_height) >
+		(cp110->base.embedded_panel_h_size *
+			cp110->base.embedded_panel_v_size)))
+		return true;
+
+	return false;
+}
+
+static uint32_t align_to_chunks_number_per_line(
+	struct dce112_compressor *cp110,
+	uint32_t pixels)
+{
+	return 256 * ((pixels + 255) / 256);
+}
+
+static void wait_for_fbc_state_changed(
+	struct dce112_compressor *cp110,
+	bool enabled)
+{
+	uint8_t counter = 0;
+	uint32_t addr = mmFBC_STATUS;
+	uint32_t value;
+
+	while (counter < 10) {
+		value = dm_read_reg(cp110->base.ctx, addr);
+		if (get_reg_field_value(
+			value,
+			FBC_STATUS,
+			FBC_ENABLE_STATUS) == enabled)
+			break;
+		udelay(10);
+		counter++;
+	}
+
+	if (counter == 10) {
+		DC_LOG_WARNING(
+			"%s: wait counter exceeded, changes to HW not applied",
+			__func__);
+	}
+}
+
+void dce112_compressor_power_up_fbc(struct compressor *compressor)
+{
+	uint32_t value;
+	uint32_t addr;
+
+	addr = mmFBC_CNTL;
+	value = dm_read_reg(compressor->ctx, addr);
+	set_reg_field_value(value, 0, FBC_CNTL, FBC_GRPH_COMP_EN);
+	set_reg_field_value(value, 1, FBC_CNTL, FBC_EN);
+	set_reg_field_value(value, 2, FBC_CNTL, FBC_COHERENCY_MODE);
+	if (compressor->options.bits.CLK_GATING_DISABLED == 1) {
+		/* HW needs to do power measurement comparison. */
+		set_reg_field_value(
+			value,
+			0,
+			FBC_CNTL,
+			FBC_COMP_CLK_GATE_EN);
+	}
+	dm_write_reg(compressor->ctx, addr, value);
+
+	addr = mmFBC_COMP_MODE;
+	value = dm_read_reg(compressor->ctx, addr);
+	set_reg_field_value(value, 1, FBC_COMP_MODE, FBC_RLE_EN);
+	set_reg_field_value(value, 1, FBC_COMP_MODE, FBC_DPCM4_RGB_EN);
+	set_reg_field_value(value, 1, FBC_COMP_MODE, FBC_IND_EN);
+	dm_write_reg(compressor->ctx, addr, value);
+
+	addr = mmFBC_COMP_CNTL;
+	value = dm_read_reg(compressor->ctx, addr);
+	set_reg_field_value(value, 1, FBC_COMP_CNTL, FBC_DEPTH_RGB08_EN);
+	dm_write_reg(compressor->ctx, addr, value);
+	/*FBC_MIN_COMPRESSION 0 ==> 2:1 */
+	/*                    1 ==> 4:1 */
+	/*                    2 ==> 8:1 */
+	/*                  0xF ==> 1:1 */
+	set_reg_field_value(value, 0xF, FBC_COMP_CNTL, FBC_MIN_COMPRESSION);
+	dm_write_reg(compressor->ctx, addr, value);
+	compressor->min_compress_ratio = FBC_COMPRESS_RATIO_1TO1;
+
+	value = 0;
+	dm_write_reg(compressor->ctx, mmFBC_IND_LUT0, value);
+
+	value = 0xFFFFFF;
+	dm_write_reg(compressor->ctx, mmFBC_IND_LUT1, value);
+}
+
+void dce112_compressor_enable_fbc(
+	struct compressor *compressor,
+	uint32_t paths_num,
+	struct compr_addr_and_pitch_params *params)
+{
+	struct dce112_compressor *cp110 = TO_DCE112_COMPRESSOR(compressor);
+
+	if (compressor->options.bits.FBC_SUPPORT &&
+		(compressor->options.bits.DUMMY_BACKEND == 0) &&
+		(!dce112_compressor_is_fbc_enabled_in_hw(compressor, NULL)) &&
+		(!is_source_bigger_than_epanel_size(
+			cp110,
+			params->source_view_width,
+			params->source_view_height))) {
+
+		uint32_t addr;
+		uint32_t value;
+
+		/* Before enabling FBC first need to enable LPT if applicable
+		 * LPT state should always be changed (enable/disable) while FBC
+		 * is disabled */
+		if (compressor->options.bits.LPT_SUPPORT && (paths_num < 2) &&
+			(params->source_view_width *
+				params->source_view_height <=
+				dce11_one_lpt_channel_max_resolution)) {
+			dce112_compressor_enable_lpt(compressor);
+		}
+
+		addr = mmFBC_CNTL;
+		value = dm_read_reg(compressor->ctx, addr);
+		set_reg_field_value(value, 1, FBC_CNTL, FBC_GRPH_COMP_EN);
+		set_reg_field_value(
+			value,
+			params->inst,
+			FBC_CNTL, FBC_SRC_SEL);
+		dm_write_reg(compressor->ctx, addr, value);
+
+		/* Keep track of enum controller_id FBC is attached to */
+		compressor->is_enabled = true;
+		compressor->attached_inst = params->inst;
+		cp110->offsets = reg_offsets[params->inst];
+
+		/*Toggle it as there is bug in HW */
+		set_reg_field_value(value, 0, FBC_CNTL, FBC_GRPH_COMP_EN);
+		dm_write_reg(compressor->ctx, addr, value);
+		set_reg_field_value(value, 1, FBC_CNTL, FBC_GRPH_COMP_EN);
+		dm_write_reg(compressor->ctx, addr, value);
+
+		wait_for_fbc_state_changed(cp110, true);
+	}
+}
+
+void dce112_compressor_disable_fbc(struct compressor *compressor)
+{
+	struct dce112_compressor *cp110 = TO_DCE112_COMPRESSOR(compressor);
+
+	if (compressor->options.bits.FBC_SUPPORT &&
+		dce112_compressor_is_fbc_enabled_in_hw(compressor, NULL)) {
+		uint32_t reg_data;
+		/* Turn off compression */
+		reg_data = dm_read_reg(compressor->ctx, mmFBC_CNTL);
+		set_reg_field_value(reg_data, 0, FBC_CNTL, FBC_GRPH_COMP_EN);
+		dm_write_reg(compressor->ctx, mmFBC_CNTL, reg_data);
+
+		/* Reset enum controller_id to undefined */
+		compressor->attached_inst = 0;
+		compressor->is_enabled = false;
+
+		/* Whenever disabling FBC make sure LPT is disabled if LPT
+		 * supported */
+		if (compressor->options.bits.LPT_SUPPORT)
+			dce112_compressor_disable_lpt(compressor);
+
+		wait_for_fbc_state_changed(cp110, false);
+	}
+}
+
+bool dce112_compressor_is_fbc_enabled_in_hw(
+	struct compressor *compressor,
+	uint32_t *inst)
+{
+	/* Check the hardware register */
+	uint32_t value;
+
+	value = dm_read_reg(compressor->ctx, mmFBC_STATUS);
+	if (get_reg_field_value(value, FBC_STATUS, FBC_ENABLE_STATUS)) {
+		if (inst != NULL)
+			*inst = compressor->attached_inst;
+		return true;
+	}
+
+	value = dm_read_reg(compressor->ctx, mmFBC_MISC);
+	if (get_reg_field_value(value, FBC_MISC, FBC_STOP_ON_HFLIP_EVENT)) {
+		value = dm_read_reg(compressor->ctx, mmFBC_CNTL);
+
+		if (get_reg_field_value(value, FBC_CNTL, FBC_GRPH_COMP_EN)) {
+			if (inst != NULL)
+				*inst =
+					compressor->attached_inst;
+			return true;
+		}
+	}
+	return false;
+}
+
+bool dce112_compressor_is_lpt_enabled_in_hw(struct compressor *compressor)
+{
+	/* Check the hardware register */
+	uint32_t value = dm_read_reg(compressor->ctx,
+		mmLOW_POWER_TILING_CONTROL);
+
+	return get_reg_field_value(
+		value,
+		LOW_POWER_TILING_CONTROL,
+		LOW_POWER_TILING_ENABLE);
+}
+
+void dce112_compressor_program_compressed_surface_address_and_pitch(
+	struct compressor *compressor,
+	struct compr_addr_and_pitch_params *params)
+{
+	struct dce112_compressor *cp110 = TO_DCE112_COMPRESSOR(compressor);
+	uint32_t value = 0;
+	uint32_t fbc_pitch = 0;
+	uint32_t compressed_surf_address_low_part =
+		compressor->compr_surface_address.addr.low_part;
+
+	/* Clear content first. */
+	dm_write_reg(
+		compressor->ctx,
+		DCP_REG(mmGRPH_COMPRESS_SURFACE_ADDRESS_HIGH),
+		0);
+	dm_write_reg(compressor->ctx,
+		DCP_REG(mmGRPH_COMPRESS_SURFACE_ADDRESS), 0);
+
+	if (compressor->options.bits.LPT_SUPPORT) {
+		uint32_t lpt_alignment = lpt_size_alignment(cp110);
+
+		if (lpt_alignment != 0) {
+			compressed_surf_address_low_part =
+				((compressed_surf_address_low_part
+					+ (lpt_alignment - 1)) / lpt_alignment)
+					* lpt_alignment;
+		}
+	}
+
+	/* Write address, HIGH has to be first. */
+	dm_write_reg(compressor->ctx,
+		DCP_REG(mmGRPH_COMPRESS_SURFACE_ADDRESS_HIGH),
+		compressor->compr_surface_address.addr.high_part);
+	dm_write_reg(compressor->ctx,
+		DCP_REG(mmGRPH_COMPRESS_SURFACE_ADDRESS),
+		compressed_surf_address_low_part);
+
+	fbc_pitch = align_to_chunks_number_per_line(
+		cp110,
+		params->source_view_width);
+
+	if (compressor->min_compress_ratio == FBC_COMPRESS_RATIO_1TO1)
+		fbc_pitch = fbc_pitch / 8;
+	else
+		DC_LOG_WARNING(
+			"%s: Unexpected DCE11 compression ratio",
+			__func__);
+
+	/* Clear content first. */
+	dm_write_reg(compressor->ctx, DCP_REG(mmGRPH_COMPRESS_PITCH), 0);
+
+	/* Write FBC Pitch. */
+	set_reg_field_value(
+		value,
+		fbc_pitch,
+		GRPH_COMPRESS_PITCH,
+		GRPH_COMPRESS_PITCH);
+	dm_write_reg(compressor->ctx, DCP_REG(mmGRPH_COMPRESS_PITCH), value);
+
+}
+
+void dce112_compressor_disable_lpt(struct compressor *compressor)
+{
+	struct dce112_compressor *cp110 = TO_DCE112_COMPRESSOR(compressor);
+	uint32_t value;
+	uint32_t addr;
+	uint32_t inx;
+
+	/* Disable all pipes LPT Stutter */
+	for (inx = 0; inx < 3; inx++) {
+		value =
+			dm_read_reg(
+				compressor->ctx,
+				DMIF_REG(mmDPG_PIPE_STUTTER_CONTROL_NONLPTCH));
+		set_reg_field_value(
+			value,
+			0,
+			DPG_PIPE_STUTTER_CONTROL_NONLPTCH,
+			STUTTER_ENABLE_NONLPTCH);
+		dm_write_reg(
+			compressor->ctx,
+			DMIF_REG(mmDPG_PIPE_STUTTER_CONTROL_NONLPTCH),
+			value);
+	}
+	/* Disable Underlay pipe LPT Stutter */
+	addr = mmDPGV0_PIPE_STUTTER_CONTROL_NONLPTCH;
+	value = dm_read_reg(compressor->ctx, addr);
+	set_reg_field_value(
+		value,
+		0,
+		DPGV0_PIPE_STUTTER_CONTROL_NONLPTCH,
+		STUTTER_ENABLE_NONLPTCH);
+	dm_write_reg(compressor->ctx, addr, value);
+
+	/* Disable LPT */
+	addr = mmLOW_POWER_TILING_CONTROL;
+	value = dm_read_reg(compressor->ctx, addr);
+	set_reg_field_value(
+		value,
+		0,
+		LOW_POWER_TILING_CONTROL,
+		LOW_POWER_TILING_ENABLE);
+	dm_write_reg(compressor->ctx, addr, value);
+
+	/* Clear selection of Channel(s) containing Compressed Surface */
+	addr = mmGMCON_LPT_TARGET;
+	value = dm_read_reg(compressor->ctx, addr);
+	set_reg_field_value(
+		value,
+		0xFFFFFFFF,
+		GMCON_LPT_TARGET,
+		STCTRL_LPT_TARGET);
+	dm_write_reg(compressor->ctx, mmGMCON_LPT_TARGET, value);
+}
+
+void dce112_compressor_enable_lpt(struct compressor *compressor)
+{
+	struct dce112_compressor *cp110 = TO_DCE112_COMPRESSOR(compressor);
+	uint32_t value;
+	uint32_t addr;
+	uint32_t value_control;
+	uint32_t channels;
+
+	/* Enable LPT Stutter from Display pipe */
+	value = dm_read_reg(compressor->ctx,
+		DMIF_REG(mmDPG_PIPE_STUTTER_CONTROL_NONLPTCH));
+	set_reg_field_value(
+		value,
+		1,
+		DPG_PIPE_STUTTER_CONTROL_NONLPTCH,
+		STUTTER_ENABLE_NONLPTCH);
+	dm_write_reg(compressor->ctx,
+		DMIF_REG(mmDPG_PIPE_STUTTER_CONTROL_NONLPTCH), value);
+
+	/* Enable Underlay pipe LPT Stutter */
+	addr = mmDPGV0_PIPE_STUTTER_CONTROL_NONLPTCH;
+	value = dm_read_reg(compressor->ctx, addr);
+	set_reg_field_value(
+		value,
+		1,
+		DPGV0_PIPE_STUTTER_CONTROL_NONLPTCH,
+		STUTTER_ENABLE_NONLPTCH);
+	dm_write_reg(compressor->ctx, addr, value);
+
+	/* Selection of Channel(s) containing Compressed Surface: 0xfffffff
+	 * will disable LPT.
+	 * STCTRL_LPT_TARGETn corresponds to channel n. */
+	addr = mmLOW_POWER_TILING_CONTROL;
+	value_control = dm_read_reg(compressor->ctx, addr);
+	channels = get_reg_field_value(value_control,
+			LOW_POWER_TILING_CONTROL,
+			LOW_POWER_TILING_MODE);
+
+	addr = mmGMCON_LPT_TARGET;
+	value = dm_read_reg(compressor->ctx, addr);
+	set_reg_field_value(
+		value,
+		channels + 1, /* not mentioned in programming guide,
+				but follow DCE8.1 */
+		GMCON_LPT_TARGET,
+		STCTRL_LPT_TARGET);
+	dm_write_reg(compressor->ctx, addr, value);
+
+	/* Enable LPT */
+	addr = mmLOW_POWER_TILING_CONTROL;
+	value = dm_read_reg(compressor->ctx, addr);
+	set_reg_field_value(
+		value,
+		1,
+		LOW_POWER_TILING_CONTROL,
+		LOW_POWER_TILING_ENABLE);
+	dm_write_reg(compressor->ctx, addr, value);
+}
+
+void dce112_compressor_program_lpt_control(
+	struct compressor *compressor,
+	struct compr_addr_and_pitch_params *params)
+{
+	struct dce112_compressor *cp110 = TO_DCE112_COMPRESSOR(compressor);
+	uint32_t rows_per_channel;
+	uint32_t lpt_alignment;
+	uint32_t source_view_width;
+	uint32_t source_view_height;
+	uint32_t lpt_control = 0;
+
+	if (!compressor->options.bits.LPT_SUPPORT)
+		return;
+
+	lpt_control = dm_read_reg(compressor->ctx,
+		mmLOW_POWER_TILING_CONTROL);
+
+	/* POSSIBLE VALUES for Low Power Tiling Mode:
+	 * 00 - Use channel 0
+	 * 01 - Use Channel 0 and 1
+	 * 02 - Use Channel 0,1,2,3
+	 * 03 - reserved */
+	switch (compressor->lpt_channels_num) {
+	/* case 2:
+	 * Use Channel 0 & 1 / Not used for DCE 11 */
+	case 1:
+		/*Use Channel 0 for LPT for DCE 11 */
+		set_reg_field_value(
+			lpt_control,
+			0,
+			LOW_POWER_TILING_CONTROL,
+			LOW_POWER_TILING_MODE);
+		break;
+	default:
+		DC_LOG_WARNING(
+			"%s: Invalid selected DRAM channels for LPT!!!",
+			__func__);
+		break;
+	}
+
+	lpt_control = lpt_memory_control_config(cp110, lpt_control);
+
+	/* Program LOW_POWER_TILING_ROWS_PER_CHAN field which depends on
+	 * FBC compressed surface pitch.
+	 * LOW_POWER_TILING_ROWS_PER_CHAN = Roundup ((Surface Height *
+	 * Surface Pitch) / (Row Size * Number of Channels *
+	 * Number of Banks)). */
+	rows_per_channel = 0;
+	lpt_alignment = lpt_size_alignment(cp110);
+	source_view_width =
+		align_to_chunks_number_per_line(
+			cp110,
+			params->source_view_width);
+	source_view_height = (params->source_view_height + 1) & (~0x1);
+
+	if (lpt_alignment != 0) {
+		rows_per_channel = source_view_width * source_view_height * 4;
+		rows_per_channel =
+			(rows_per_channel % lpt_alignment) ?
+				(rows_per_channel / lpt_alignment + 1) :
+				rows_per_channel / lpt_alignment;
+	}
+
+	set_reg_field_value(
+		lpt_control,
+		rows_per_channel,
+		LOW_POWER_TILING_CONTROL,
+		LOW_POWER_TILING_ROWS_PER_CHAN);
+
+	dm_write_reg(compressor->ctx,
+		mmLOW_POWER_TILING_CONTROL, lpt_control);
+}
+
+/*
+ * DCE 11 Frame Buffer Compression Implementation
+ */
+
+void dce112_compressor_set_fbc_invalidation_triggers(
+	struct compressor *compressor,
+	uint32_t fbc_trigger)
+{
+	/* Disable region hit event, FBC_MEMORY_REGION_MASK = 0 (bits 16-19)
+	 * for DCE 11 regions cannot be used - does not work with S/G
+	 */
+	uint32_t addr = mmFBC_CLIENT_REGION_MASK;
+	uint32_t value = dm_read_reg(compressor->ctx, addr);
+
+	set_reg_field_value(
+		value,
+		0,
+		FBC_CLIENT_REGION_MASK,
+		FBC_MEMORY_REGION_MASK);
+	dm_write_reg(compressor->ctx, addr, value);
+
+	/* Setup events when to clear all CSM entries (effectively marking
+	 * current compressed data invalid)
+	 * For DCE 11 CSM metadata 11111 means - "Not Compressed"
+	 * Used as the initial value of the metadata sent to the compressor
+	 * after invalidation, to indicate that the compressor should attempt
+	 * to compress all chunks on the current pass.  Also used when the chunk
+	 * is not successfully written to memory.
+	 * When this CSM value is detected, FBC reads from the uncompressed
+	 * buffer. Set events according to passed in value, these events are
+	 * valid for DCE11:
+	 *     - bit  0 - display register updated
+	 *     - bit 28 - memory write from any client except from MCIF
+	 *     - bit 29 - CG static screen signal is inactive
+	 * In addition, DCE11.1 also needs to set new DCE11.1 specific events
+	 * that are used to trigger invalidation on certain register changes,
+	 * for example enabling of Alpha Compression may trigger invalidation of
+	 * FBC once bit is set. These events are as follows:
+	 *      - Bit 2 - FBC_GRPH_COMP_EN register updated
+	 *      - Bit 3 - FBC_SRC_SEL register updated
+	 *      - Bit 4 - FBC_MIN_COMPRESSION register updated
+	 *      - Bit 5 - FBC_ALPHA_COMP_EN register updated
+	 *      - Bit 6 - FBC_ZERO_ALPHA_CHUNK_SKIP_EN register updated
+	 *      - Bit 7 - FBC_FORCE_COPY_TO_COMP_BUF register updated
+	 */
+	addr = mmFBC_IDLE_FORCE_CLEAR_MASK;
+	value = dm_read_reg(compressor->ctx, addr);
+	set_reg_field_value(
+		value,
+		fbc_trigger |
+		FBC_IDLE_FORCE_GRPH_COMP_EN |
+		FBC_IDLE_FORCE_SRC_SEL_CHANGE |
+		FBC_IDLE_FORCE_MIN_COMPRESSION_CHANGE |
+		FBC_IDLE_FORCE_ALPHA_COMP_EN |
+		FBC_IDLE_FORCE_ZERO_ALPHA_CHUNK_SKIP_EN |
+		FBC_IDLE_FORCE_FORCE_COPY_TO_COMP_BUF,
+		FBC_IDLE_FORCE_CLEAR_MASK,
+		FBC_IDLE_FORCE_CLEAR_MASK);
+	dm_write_reg(compressor->ctx, addr, value);
+}
+
+void dce112_compressor_construct(struct dce112_compressor *compressor,
+	struct dc_context *ctx)
+{
+	struct dc_bios *bp = ctx->dc_bios;
+	struct embedded_panel_info panel_info;
+
+	compressor->base.options.raw = 0;
+	compressor->base.options.bits.FBC_SUPPORT = true;
+	compressor->base.options.bits.LPT_SUPPORT = true;
+	 /* For DCE 11 always use one DRAM channel for LPT */
+	compressor->base.lpt_channels_num = 1;
+	compressor->base.options.bits.DUMMY_BACKEND = false;
+
+	/* Check if this system has more than 1 DRAM channel; if only 1 then LPT
+	 * should not be supported */
+	if (compressor->base.memory_bus_width == 64)
+		compressor->base.options.bits.LPT_SUPPORT = false;
+
+	compressor->base.options.bits.CLK_GATING_DISABLED = false;
+
+	compressor->base.ctx = ctx;
+	compressor->base.embedded_panel_h_size = 0;
+	compressor->base.embedded_panel_v_size = 0;
+	compressor->base.memory_bus_width = ctx->asic_id.vram_width;
+	compressor->base.allocated_size = 0;
+	compressor->base.preferred_requested_size = 0;
+	compressor->base.min_compress_ratio = FBC_COMPRESS_RATIO_INVALID;
+	compressor->base.banks_num = 0;
+	compressor->base.raw_size = 0;
+	compressor->base.channel_interleave_size = 0;
+	compressor->base.dram_channels_num = 0;
+	compressor->base.lpt_channels_num = 0;
+	compressor->base.attached_inst = 0;
+	compressor->base.is_enabled = false;
+
+	if (BP_RESULT_OK ==
+			bp->funcs->get_embedded_panel_info(bp, &panel_info)) {
+		compressor->base.embedded_panel_h_size =
+			panel_info.lcd_timing.horizontal_addressable;
+		compressor->base.embedded_panel_v_size =
+			panel_info.lcd_timing.vertical_addressable;
+	}
+}
+
+struct compressor *dce112_compressor_create(struct dc_context *ctx)
+{
+	struct dce112_compressor *cp110 =
+		kzalloc(sizeof(struct dce112_compressor), GFP_KERNEL);
+
+	if (!cp110)
+		return NULL;
+
+	dce112_compressor_construct(cp110, ctx);
+	return &cp110->base;
+}
+
+void dce112_compressor_destroy(struct compressor **compressor)
+{
+	kfree(TO_DCE112_COMPRESSOR(*compressor));
+	*compressor = NULL;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce112/dce112_compressor.h b/drivers/gpu/drm/amd/display/dc/dce112/dce112_compressor.h
new file mode 100644
index 000000000000..f1227133f6df
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce112/dce112_compressor.h
@@ -0,0 +1,78 @@
+/* Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_COMPRESSOR_DCE112_H__
+#define __DC_COMPRESSOR_DCE112_H__
+
+#include "../inc/compressor.h"
+
+#define TO_DCE112_COMPRESSOR(compressor)\
+	container_of(compressor, struct dce112_compressor, base)
+
+struct dce112_compressor_reg_offsets {
+	uint32_t dcp_offset;
+	uint32_t dmif_offset;
+};
+
+struct dce112_compressor {
+	struct compressor base;
+	struct dce112_compressor_reg_offsets offsets;
+};
+
+struct compressor *dce112_compressor_create(struct dc_context *ctx);
+
+void dce112_compressor_construct(struct dce112_compressor *cp110,
+	struct dc_context *ctx);
+
+void dce112_compressor_destroy(struct compressor **cp);
+
+/* FBC RELATED */
+void dce112_compressor_power_up_fbc(struct compressor *cp);
+
+void dce112_compressor_enable_fbc(struct compressor *cp, uint32_t paths_num,
+	struct compr_addr_and_pitch_params *params);
+
+void dce112_compressor_disable_fbc(struct compressor *cp);
+
+void dce112_compressor_set_fbc_invalidation_triggers(struct compressor *cp,
+	uint32_t fbc_trigger);
+
+void dce112_compressor_program_compressed_surface_address_and_pitch(
+	struct compressor *cp,
+	struct compr_addr_and_pitch_params *params);
+
+bool dce112_compressor_is_fbc_enabled_in_hw(struct compressor *cp,
+	uint32_t *fbc_mapped_crtc_id);
+
+/* LPT RELATED */
+void dce112_compressor_enable_lpt(struct compressor *cp);
+
+void dce112_compressor_disable_lpt(struct compressor *cp);
+
+void dce112_compressor_program_lpt_control(struct compressor *cp,
+	struct compr_addr_and_pitch_params *params);
+
+bool dce112_compressor_is_lpt_enabled_in_hw(struct compressor *cp);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dce120/Makefile b/drivers/gpu/drm/amd/display/dc/dce120/Makefile
new file mode 100644
index 000000000000..8f508e662748
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce120/Makefile
@@ -0,0 +1,33 @@
+#
+# Copyright 2017 Advanced Micro Devices, Inc.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+#
+# Makefile for the 'controller' sub-component of DAL.
+# It provides the control and status of HW CRTC block.
+
+
+CFLAGS_$(AMDDALPATH)/dc/dce120/dce120_resource.o = -Wno-override-init
+
+DCE120 = dce120_timing_generator.o
+
+AMD_DAL_DCE120 = $(addprefix $(AMDDALPATH)/dc/dce120/,$(DCE120))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCE120)
diff --git a/drivers/gpu/drm/amd/display/dc/dce120/dce120_timing_generator.c b/drivers/gpu/drm/amd/display/dc/dce120/dce120_timing_generator.c
new file mode 100644
index 000000000000..31c4f44ceaac
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce120/dce120_timing_generator.c
@@ -0,0 +1,1284 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+
+#include "dce/dce_12_0_offset.h"
+#include "dce/dce_12_0_sh_mask.h"
+#include "soc15_hw_ip.h"
+#include "vega10_ip_offset.h"
+
+#include "dc_types.h"
+#include "dc_bios_types.h"
+
+#include "include/grph_object_id.h"
+#include "include/logger_interface.h"
+#include "dce120_timing_generator.h"
+
+#include "timing_generator.h"
+
+#define CRTC_REG_UPDATE_N(reg_name, n, ...)	\
+		generic_reg_update_soc15(tg110->base.ctx, tg110->offsets.crtc, reg_name, n, __VA_ARGS__)
+
+#define CRTC_REG_SET_N(reg_name, n, ...)	\
+		generic_reg_set_soc15(tg110->base.ctx, tg110->offsets.crtc, reg_name, n, __VA_ARGS__)
+
+#define CRTC_REG_UPDATE(reg, field, val)	\
+		CRTC_REG_UPDATE_N(reg, 1, FD(reg##__##field), val)
+
+#define CRTC_REG_UPDATE_2(reg, field1, val1, field2, val2)	\
+		CRTC_REG_UPDATE_N(reg, 2, FD(reg##__##field1), val1, FD(reg##__##field2), val2)
+
+#define CRTC_REG_UPDATE_3(reg, field1, val1, field2, val2, field3, val3)	\
+		CRTC_REG_UPDATE_N(reg, 3, FD(reg##__##field1), val1, FD(reg##__##field2), val2, FD(reg##__##field3), val3)
+
+#define CRTC_REG_UPDATE_4(reg, field1, val1, field2, val2, field3, val3, field4, val4)	\
+		CRTC_REG_UPDATE_N(reg, 3, FD(reg##__##field1), val1, FD(reg##__##field2), val2, FD(reg##__##field3), val3, FD(reg##__##field4), val4)
+
+#define CRTC_REG_UPDATE_5(reg, field1, val1, field2, val2, field3, val3, field4, val4, field5, val5)	\
+		CRTC_REG_UPDATE_N(reg, 3, FD(reg##__##field1), val1, FD(reg##__##field2), val2, FD(reg##__##field3), val3, FD(reg##__##field4), val4, FD(reg##__##field5), val5)
+
+#define CRTC_REG_SET(reg, field, val)	\
+		CRTC_REG_SET_N(reg, 1, FD(reg##__##field), val)
+
+#define CRTC_REG_SET_2(reg, field1, val1, field2, val2)	\
+		CRTC_REG_SET_N(reg, 2, FD(reg##__##field1), val1, FD(reg##__##field2), val2)
+
+#define CRTC_REG_SET_3(reg, field1, val1, field2, val2, field3, val3)	\
+		CRTC_REG_SET_N(reg, 3, FD(reg##__##field1), val1, FD(reg##__##field2), val2, FD(reg##__##field3), val3)
+
+/*
+ *****************************************************************************
+ *  Function: is_in_vertical_blank
+ *
+ *  @brief
+ *     check the current status of CRTC to check if we are in Vertical Blank
+ *     regioneased" state
+ *
+ *  @return
+ *     true if currently in blank region, false otherwise
+ *
+ *****************************************************************************
+ */
+static bool dce120_timing_generator_is_in_vertical_blank(
+		struct timing_generator *tg)
+{
+	uint32_t field = 0;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t value = dm_read_reg_soc15(
+					tg->ctx,
+					mmCRTC0_CRTC_STATUS,
+					tg110->offsets.crtc);
+
+	field = get_reg_field_value(value, CRTC0_CRTC_STATUS, CRTC_V_BLANK);
+	return field == 1;
+}
+
+
+/* determine if given timing can be supported by TG */
+static bool dce120_timing_generator_validate_timing(
+	struct timing_generator *tg,
+	const struct dc_crtc_timing *timing,
+	enum signal_type signal)
+{
+	uint32_t interlace_factor = timing->flags.INTERLACE ? 2 : 1;
+	uint32_t v_blank =
+					(timing->v_total - timing->v_addressable -
+					timing->v_border_top - timing->v_border_bottom) *
+					interlace_factor;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	if (!dce110_timing_generator_validate_timing(
+					tg,
+					timing,
+					signal))
+		return false;
+
+
+	if (v_blank < tg110->min_v_blank	||
+		 timing->h_sync_width  < tg110->min_h_sync_width ||
+		 timing->v_sync_width  < tg110->min_v_sync_width)
+		return false;
+
+	return true;
+}
+
+static bool dce120_tg_validate_timing(struct timing_generator *tg,
+	const struct dc_crtc_timing *timing)
+{
+	return dce120_timing_generator_validate_timing(tg, timing, SIGNAL_TYPE_NONE);
+}
+
+/******** HW programming ************/
+/* Disable/Enable Timing Generator */
+static bool dce120_timing_generator_enable_crtc(struct timing_generator *tg)
+{
+	enum bp_result result;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	/* Set MASTER_UPDATE_MODE to 0
+	 * This is needed for DRR, and also suggested to be default value by Syed.*/
+
+	CRTC_REG_UPDATE(CRTC0_CRTC_MASTER_UPDATE_MODE,
+			MASTER_UPDATE_MODE, 0);
+
+	CRTC_REG_UPDATE(CRTC0_CRTC_MASTER_UPDATE_LOCK,
+			UNDERFLOW_UPDATE_LOCK, 0);
+
+	/* TODO API for AtomFirmware didn't change*/
+	result = tg->bp->funcs->enable_crtc(tg->bp, tg110->controller_id, true);
+
+	return result == BP_RESULT_OK;
+}
+
+static void dce120_timing_generator_set_early_control(
+		struct timing_generator *tg,
+		uint32_t early_cntl)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	CRTC_REG_UPDATE(CRTC0_CRTC_CONTROL,
+			CRTC_HBLANK_EARLY_CONTROL, early_cntl);
+}
+
+/**************** TG current status ******************/
+
+/* return the current frame counter. Used by Linux kernel DRM */
+static uint32_t dce120_timing_generator_get_vblank_counter(
+		struct timing_generator *tg)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t value = dm_read_reg_soc15(
+				tg->ctx,
+				mmCRTC0_CRTC_STATUS_FRAME_COUNT,
+				tg110->offsets.crtc);
+	uint32_t field = get_reg_field_value(
+				value, CRTC0_CRTC_STATUS_FRAME_COUNT, CRTC_FRAME_COUNT);
+
+	return field;
+}
+
+/* Get current H and V position */
+static void dce120_timing_generator_get_crtc_position(
+	struct timing_generator *tg,
+	struct crtc_position *position)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t value = dm_read_reg_soc15(
+				tg->ctx,
+				mmCRTC0_CRTC_STATUS_POSITION,
+				tg110->offsets.crtc);
+
+	position->horizontal_count = get_reg_field_value(value,
+			CRTC0_CRTC_STATUS_POSITION, CRTC_HORZ_COUNT);
+
+	position->vertical_count = get_reg_field_value(value,
+			CRTC0_CRTC_STATUS_POSITION, CRTC_VERT_COUNT);
+
+	value = dm_read_reg_soc15(
+				tg->ctx,
+				mmCRTC0_CRTC_NOM_VERT_POSITION,
+				tg110->offsets.crtc);
+
+	position->nominal_vcount = get_reg_field_value(value,
+			CRTC0_CRTC_NOM_VERT_POSITION, CRTC_VERT_COUNT_NOM);
+}
+
+/* wait until TG is in beginning of vertical blank region */
+static void dce120_timing_generator_wait_for_vblank(struct timing_generator *tg)
+{
+	/* We want to catch beginning of VBlank here, so if the first try are
+	 * in VBlank, we might be very close to Active, in this case wait for
+	 * another frame
+	 */
+	while (dce120_timing_generator_is_in_vertical_blank(tg)) {
+		if (!tg->funcs->is_counter_moving(tg)) {
+			/* error - no point to wait if counter is not moving */
+			break;
+		}
+	}
+
+	while (!dce120_timing_generator_is_in_vertical_blank(tg)) {
+		if (!tg->funcs->is_counter_moving(tg)) {
+			/* error - no point to wait if counter is not moving */
+			break;
+		}
+	}
+}
+
+/* wait until TG is in beginning of active region */
+static void dce120_timing_generator_wait_for_vactive(struct timing_generator *tg)
+{
+	while (dce120_timing_generator_is_in_vertical_blank(tg)) {
+		if (!tg->funcs->is_counter_moving(tg)) {
+			/* error - no point to wait if counter is not moving */
+			break;
+		}
+	}
+}
+
+/*********** Timing Generator Synchronization routines ****/
+
+/* Setups Global Swap Lock group, TimingServer or TimingClient*/
+static void dce120_timing_generator_setup_global_swap_lock(
+	struct timing_generator *tg,
+	const struct dcp_gsl_params *gsl_params)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t value_crtc_vtotal =
+							dm_read_reg_soc15(tg->ctx,
+							mmCRTC0_CRTC_V_TOTAL,
+							tg110->offsets.crtc);
+	/* Checkpoint relative to end of frame */
+	uint32_t check_point =
+							get_reg_field_value(value_crtc_vtotal,
+							CRTC0_CRTC_V_TOTAL,
+							CRTC_V_TOTAL);
+
+
+	dm_write_reg_soc15(tg->ctx, mmCRTC0_CRTC_GSL_WINDOW, tg110->offsets.crtc, 0);
+
+	CRTC_REG_UPDATE_N(DCP0_DCP_GSL_CONTROL, 6,
+		/* This pipe will belong to GSL Group zero. */
+		FD(DCP0_DCP_GSL_CONTROL__DCP_GSL0_EN), 1,
+		FD(DCP0_DCP_GSL_CONTROL__DCP_GSL_MASTER_EN), gsl_params->gsl_master == tg->inst,
+		FD(DCP0_DCP_GSL_CONTROL__DCP_GSL_HSYNC_FLIP_FORCE_DELAY), HFLIP_READY_DELAY,
+		/* Keep signal low (pending high) during 6 lines.
+		 * Also defines minimum interval before re-checking signal. */
+		FD(DCP0_DCP_GSL_CONTROL__DCP_GSL_HSYNC_FLIP_CHECK_DELAY), HFLIP_CHECK_DELAY,
+		/* DCP_GSL_PURPOSE_SURFACE_FLIP */
+		FD(DCP0_DCP_GSL_CONTROL__DCP_GSL_SYNC_SOURCE), 0,
+		FD(DCP0_DCP_GSL_CONTROL__DCP_GSL_DELAY_SURFACE_UPDATE_PENDING), 1);
+
+	CRTC_REG_SET_2(
+			CRTC0_CRTC_GSL_CONTROL,
+			CRTC_GSL_CHECK_LINE_NUM, check_point - FLIP_READY_BACK_LOOKUP,
+			CRTC_GSL_FORCE_DELAY, VFLIP_READY_DELAY);
+}
+
+/* Clear all the register writes done by setup_global_swap_lock */
+static void dce120_timing_generator_tear_down_global_swap_lock(
+	struct timing_generator *tg)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	/* Settig HW default values from reg specs */
+	CRTC_REG_SET_N(DCP0_DCP_GSL_CONTROL, 6,
+			FD(DCP0_DCP_GSL_CONTROL__DCP_GSL0_EN), 0,
+			FD(DCP0_DCP_GSL_CONTROL__DCP_GSL_MASTER_EN), 0,
+			FD(DCP0_DCP_GSL_CONTROL__DCP_GSL_HSYNC_FLIP_FORCE_DELAY), HFLIP_READY_DELAY,
+			FD(DCP0_DCP_GSL_CONTROL__DCP_GSL_HSYNC_FLIP_CHECK_DELAY), HFLIP_CHECK_DELAY,
+			/* DCP_GSL_PURPOSE_SURFACE_FLIP */
+			FD(DCP0_DCP_GSL_CONTROL__DCP_GSL_SYNC_SOURCE), 0,
+			FD(DCP0_DCP_GSL_CONTROL__DCP_GSL_DELAY_SURFACE_UPDATE_PENDING), 0);
+
+	CRTC_REG_SET_2(CRTC0_CRTC_GSL_CONTROL,
+		       CRTC_GSL_CHECK_LINE_NUM, 0,
+		       CRTC_GSL_FORCE_DELAY, 0x2); /*TODO Why this value here ?*/
+}
+
+/* Reset slave controllers on master VSync */
+static void dce120_timing_generator_enable_reset_trigger(
+	struct timing_generator *tg,
+	int source)
+{
+	enum trigger_source_select trig_src_select = TRIGGER_SOURCE_SELECT_LOGIC_ZERO;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t rising_edge = 0;
+	uint32_t falling_edge = 0;
+	/* Setup trigger edge */
+	uint32_t pol_value = dm_read_reg_soc15(
+									tg->ctx,
+									mmCRTC0_CRTC_V_SYNC_A_CNTL,
+									tg110->offsets.crtc);
+
+	/* Register spec has reversed definition:
+	 *	0 for positive, 1 for negative */
+	if (get_reg_field_value(pol_value,
+			CRTC0_CRTC_V_SYNC_A_CNTL,
+			CRTC_V_SYNC_A_POL) == 0) {
+		rising_edge = 1;
+	} else {
+		falling_edge = 1;
+	}
+
+	/* TODO What about other sources ?*/
+	trig_src_select = TRIGGER_SOURCE_SELECT_GSL_GROUP0;
+
+	CRTC_REG_UPDATE_N(CRTC0_CRTC_TRIGB_CNTL, 7,
+		FD(CRTC0_CRTC_TRIGB_CNTL__CRTC_TRIGB_SOURCE_SELECT), trig_src_select,
+		FD(CRTC0_CRTC_TRIGB_CNTL__CRTC_TRIGB_POLARITY_SELECT), TRIGGER_POLARITY_SELECT_LOGIC_ZERO,
+		FD(CRTC0_CRTC_TRIGB_CNTL__CRTC_TRIGB_RISING_EDGE_DETECT_CNTL), rising_edge,
+		FD(CRTC0_CRTC_TRIGB_CNTL__CRTC_TRIGB_FALLING_EDGE_DETECT_CNTL), falling_edge,
+		/* send every signal */
+		FD(CRTC0_CRTC_TRIGB_CNTL__CRTC_TRIGB_FREQUENCY_SELECT), 0,
+		/* no delay */
+		FD(CRTC0_CRTC_TRIGB_CNTL__CRTC_TRIGB_DELAY), 0,
+		/* clear trigger status */
+		FD(CRTC0_CRTC_TRIGB_CNTL__CRTC_TRIGB_CLEAR), 1);
+
+	CRTC_REG_UPDATE_3(
+			CRTC0_CRTC_FORCE_COUNT_NOW_CNTL,
+			CRTC_FORCE_COUNT_NOW_MODE, 2,
+			CRTC_FORCE_COUNT_NOW_TRIG_SEL, 1,
+			CRTC_FORCE_COUNT_NOW_CLEAR, 1);
+}
+
+/* disabling trigger-reset */
+static void dce120_timing_generator_disable_reset_trigger(
+	struct timing_generator *tg)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	CRTC_REG_UPDATE_2(
+		CRTC0_CRTC_FORCE_COUNT_NOW_CNTL,
+		CRTC_FORCE_COUNT_NOW_MODE, 0,
+		CRTC_FORCE_COUNT_NOW_CLEAR, 1);
+
+	CRTC_REG_UPDATE_3(
+		CRTC0_CRTC_TRIGB_CNTL,
+		CRTC_TRIGB_SOURCE_SELECT, TRIGGER_SOURCE_SELECT_LOGIC_ZERO,
+		CRTC_TRIGB_POLARITY_SELECT, TRIGGER_POLARITY_SELECT_LOGIC_ZERO,
+		/* clear trigger status */
+		CRTC_TRIGB_CLEAR, 1);
+
+}
+
+/* Checks whether CRTC triggered reset occurred */
+static bool dce120_timing_generator_did_triggered_reset_occur(
+	struct timing_generator *tg)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t value = dm_read_reg_soc15(
+			tg->ctx,
+			mmCRTC0_CRTC_FORCE_COUNT_NOW_CNTL,
+			tg110->offsets.crtc);
+
+	return get_reg_field_value(value,
+			CRTC0_CRTC_FORCE_COUNT_NOW_CNTL,
+			CRTC_FORCE_COUNT_NOW_OCCURRED) != 0;
+}
+
+
+/******** Stuff to move to other virtual HW objects *****************/
+/* Move to enable accelerated mode */
+static void dce120_timing_generator_disable_vga(struct timing_generator *tg)
+{
+	uint32_t offset = 0;
+	uint32_t value = 0;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	switch (tg110->controller_id) {
+	case CONTROLLER_ID_D0:
+		offset = 0;
+		break;
+	case CONTROLLER_ID_D1:
+		offset = mmD2VGA_CONTROL - mmD1VGA_CONTROL;
+		break;
+	case CONTROLLER_ID_D2:
+		offset = mmD3VGA_CONTROL - mmD1VGA_CONTROL;
+		break;
+	case CONTROLLER_ID_D3:
+		offset = mmD4VGA_CONTROL - mmD1VGA_CONTROL;
+		break;
+	case CONTROLLER_ID_D4:
+		offset = mmD5VGA_CONTROL - mmD1VGA_CONTROL;
+		break;
+	case CONTROLLER_ID_D5:
+		offset = mmD6VGA_CONTROL - mmD1VGA_CONTROL;
+		break;
+	default:
+		break;
+	}
+
+	value = dm_read_reg_soc15(tg->ctx, mmD1VGA_CONTROL, offset);
+
+	set_reg_field_value(value, 0, D1VGA_CONTROL, D1VGA_MODE_ENABLE);
+	set_reg_field_value(value, 0, D1VGA_CONTROL, D1VGA_TIMING_SELECT);
+	set_reg_field_value(
+			value, 0, D1VGA_CONTROL, D1VGA_SYNC_POLARITY_SELECT);
+	set_reg_field_value(value, 0, D1VGA_CONTROL, D1VGA_OVERSCAN_COLOR_EN);
+
+	dm_write_reg_soc15(tg->ctx, mmD1VGA_CONTROL, offset, value);
+}
+/* TODO: Should we move it to transform */
+/* Fully program CRTC timing in timing generator */
+static void dce120_timing_generator_program_blanking(
+	struct timing_generator *tg,
+	const struct dc_crtc_timing *timing)
+{
+	uint32_t tmp1 = 0;
+	uint32_t tmp2 = 0;
+	uint32_t vsync_offset = timing->v_border_bottom +
+			timing->v_front_porch;
+	uint32_t v_sync_start = timing->v_addressable + vsync_offset;
+
+	uint32_t hsync_offset = timing->h_border_right +
+			timing->h_front_porch;
+	uint32_t h_sync_start = timing->h_addressable + hsync_offset;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	CRTC_REG_UPDATE(
+		CRTC0_CRTC_H_TOTAL,
+		CRTC_H_TOTAL,
+		timing->h_total - 1);
+
+	CRTC_REG_UPDATE(
+		CRTC0_CRTC_V_TOTAL,
+		CRTC_V_TOTAL,
+		timing->v_total - 1);
+
+	/* In case of V_TOTAL_CONTROL is on, make sure V_TOTAL_MAX and
+	 * V_TOTAL_MIN are equal to V_TOTAL.
+	 */
+	CRTC_REG_UPDATE(
+		CRTC0_CRTC_V_TOTAL_MAX,
+		CRTC_V_TOTAL_MAX,
+		timing->v_total - 1);
+
+	CRTC_REG_UPDATE(
+		CRTC0_CRTC_V_TOTAL_MIN,
+		CRTC_V_TOTAL_MIN,
+		timing->v_total - 1);
+
+	tmp1 = timing->h_total -
+			(h_sync_start + timing->h_border_left);
+	tmp2 = tmp1 + timing->h_addressable +
+			timing->h_border_left + timing->h_border_right;
+
+	CRTC_REG_UPDATE_2(
+			CRTC0_CRTC_H_BLANK_START_END,
+			CRTC_H_BLANK_END, tmp1,
+			CRTC_H_BLANK_START, tmp2);
+
+	tmp1 = timing->v_total - (v_sync_start + timing->v_border_top);
+	tmp2 = tmp1 + timing->v_addressable + timing->v_border_top +
+			timing->v_border_bottom;
+
+	CRTC_REG_UPDATE_2(
+		CRTC0_CRTC_V_BLANK_START_END,
+		CRTC_V_BLANK_END, tmp1,
+		CRTC_V_BLANK_START, tmp2);
+}
+
+/* TODO: Should we move it to opp? */
+/* Combine with below and move YUV/RGB color conversion to SW layer */
+static void dce120_timing_generator_program_blank_color(
+	struct timing_generator *tg,
+	const struct tg_color *black_color)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	CRTC_REG_UPDATE_3(
+		CRTC0_CRTC_BLACK_COLOR,
+		CRTC_BLACK_COLOR_B_CB, black_color->color_b_cb,
+		CRTC_BLACK_COLOR_G_Y, black_color->color_g_y,
+		CRTC_BLACK_COLOR_R_CR, black_color->color_r_cr);
+}
+/* Combine with above and move YUV/RGB color conversion to SW layer */
+static void dce120_timing_generator_set_overscan_color_black(
+	struct timing_generator *tg,
+	const struct tg_color *color)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t value = 0;
+	CRTC_REG_SET_3(
+		CRTC0_CRTC_OVERSCAN_COLOR,
+		CRTC_OVERSCAN_COLOR_BLUE, color->color_b_cb,
+		CRTC_OVERSCAN_COLOR_GREEN, color->color_g_y,
+		CRTC_OVERSCAN_COLOR_RED, color->color_r_cr);
+
+	value = dm_read_reg_soc15(
+			tg->ctx,
+			mmCRTC0_CRTC_OVERSCAN_COLOR,
+			tg110->offsets.crtc);
+
+	dm_write_reg_soc15(
+			tg->ctx,
+			mmCRTC0_CRTC_BLACK_COLOR,
+			tg110->offsets.crtc,
+			value);
+
+	/* This is desirable to have a constant DAC output voltage during the
+	 * blank time that is higher than the 0 volt reference level that the
+	 * DAC outputs when the NBLANK signal
+	 * is asserted low, such as for output to an analog TV. */
+	dm_write_reg_soc15(
+		tg->ctx,
+		mmCRTC0_CRTC_BLANK_DATA_COLOR,
+		tg110->offsets.crtc,
+		value);
+
+	/* TO DO we have to program EXT registers and we need to know LB DATA
+	 * format because it is used when more 10 , i.e. 12 bits per color
+	 *
+	 * m_mmDxCRTC_OVERSCAN_COLOR_EXT
+	 * m_mmDxCRTC_BLACK_COLOR_EXT
+	 * m_mmDxCRTC_BLANK_DATA_COLOR_EXT
+	 */
+}
+
+static void dce120_timing_generator_set_drr(
+	struct timing_generator *tg,
+	const struct drr_params *params)
+{
+
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	if (params != NULL &&
+		params->vertical_total_max > 0 &&
+		params->vertical_total_min > 0) {
+
+		CRTC_REG_UPDATE(
+				CRTC0_CRTC_V_TOTAL_MIN,
+				CRTC_V_TOTAL_MIN, params->vertical_total_min - 1);
+		CRTC_REG_UPDATE(
+				CRTC0_CRTC_V_TOTAL_MAX,
+				CRTC_V_TOTAL_MAX, params->vertical_total_max - 1);
+		CRTC_REG_SET_N(CRTC0_CRTC_V_TOTAL_CONTROL, 6,
+				FD(CRTC0_CRTC_V_TOTAL_CONTROL__CRTC_V_TOTAL_MIN_SEL), 1,
+				FD(CRTC0_CRTC_V_TOTAL_CONTROL__CRTC_V_TOTAL_MAX_SEL), 1,
+				FD(CRTC0_CRTC_V_TOTAL_CONTROL__CRTC_FORCE_LOCK_ON_EVENT), 0,
+				FD(CRTC0_CRTC_V_TOTAL_CONTROL__CRTC_FORCE_LOCK_TO_MASTER_VSYNC), 0,
+				FD(CRTC0_CRTC_V_TOTAL_CONTROL__CRTC_SET_V_TOTAL_MIN_MASK_EN), 0,
+				FD(CRTC0_CRTC_V_TOTAL_CONTROL__CRTC_SET_V_TOTAL_MIN_MASK), 0);
+		CRTC_REG_UPDATE(
+				CRTC0_CRTC_STATIC_SCREEN_CONTROL,
+				CRTC_STATIC_SCREEN_EVENT_MASK,
+				0x180);
+
+	} else {
+		CRTC_REG_SET_N(CRTC0_CRTC_V_TOTAL_CONTROL, 5,
+				FD(CRTC0_CRTC_V_TOTAL_CONTROL__CRTC_V_TOTAL_MIN_SEL), 0,
+				FD(CRTC0_CRTC_V_TOTAL_CONTROL__CRTC_V_TOTAL_MAX_SEL), 0,
+				FD(CRTC0_CRTC_V_TOTAL_CONTROL__CRTC_FORCE_LOCK_ON_EVENT), 0,
+				FD(CRTC0_CRTC_V_TOTAL_CONTROL__CRTC_FORCE_LOCK_TO_MASTER_VSYNC), 0,
+				FD(CRTC0_CRTC_V_TOTAL_CONTROL__CRTC_SET_V_TOTAL_MIN_MASK), 0);
+		CRTC_REG_UPDATE(
+				CRTC0_CRTC_V_TOTAL_MIN,
+				CRTC_V_TOTAL_MIN, 0);
+		CRTC_REG_UPDATE(
+				CRTC0_CRTC_V_TOTAL_MAX,
+				CRTC_V_TOTAL_MAX, 0);
+		CRTC_REG_UPDATE(
+				CRTC0_CRTC_STATIC_SCREEN_CONTROL,
+				CRTC_STATIC_SCREEN_EVENT_MASK,
+				0);
+	}
+}
+
+static void dce120_timing_generator_get_crtc_scanoutpos(
+	struct timing_generator *tg,
+	uint32_t *v_blank_start,
+	uint32_t *v_blank_end,
+	uint32_t *h_position,
+	uint32_t *v_position)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	struct crtc_position position;
+
+	uint32_t v_blank_start_end = dm_read_reg_soc15(
+			tg->ctx,
+			mmCRTC0_CRTC_V_BLANK_START_END,
+			tg110->offsets.crtc);
+
+	*v_blank_start = get_reg_field_value(v_blank_start_end,
+					     CRTC0_CRTC_V_BLANK_START_END,
+					     CRTC_V_BLANK_START);
+	*v_blank_end = get_reg_field_value(v_blank_start_end,
+					   CRTC0_CRTC_V_BLANK_START_END,
+					   CRTC_V_BLANK_END);
+
+	dce120_timing_generator_get_crtc_position(
+			tg, &position);
+
+	*h_position = position.horizontal_count;
+	*v_position = position.vertical_count;
+}
+
+static void dce120_timing_generator_enable_advanced_request(
+	struct timing_generator *tg,
+	bool enable,
+	const struct dc_crtc_timing *timing)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t v_sync_width_and_b_porch =
+				timing->v_total - timing->v_addressable -
+				timing->v_border_bottom - timing->v_front_porch;
+	uint32_t value = dm_read_reg_soc15(
+				tg->ctx,
+				mmCRTC0_CRTC_START_LINE_CONTROL,
+				tg110->offsets.crtc);
+
+	set_reg_field_value(
+		value,
+		enable ? 0 : 1,
+		CRTC0_CRTC_START_LINE_CONTROL,
+		CRTC_LEGACY_REQUESTOR_EN);
+
+	/* Program advanced line position acc.to the best case from fetching data perspective to hide MC latency
+	 * and prefilling Line Buffer in V Blank (to 10 lines as LB can store max 10 lines)
+	 */
+	if (v_sync_width_and_b_porch > 10)
+		v_sync_width_and_b_porch = 10;
+
+	set_reg_field_value(
+		value,
+		v_sync_width_and_b_porch,
+		CRTC0_CRTC_START_LINE_CONTROL,
+		CRTC_ADVANCED_START_LINE_POSITION);
+
+	dm_write_reg_soc15(tg->ctx,
+			mmCRTC0_CRTC_START_LINE_CONTROL,
+			tg110->offsets.crtc,
+			value);
+}
+
+static void dce120_tg_program_blank_color(struct timing_generator *tg,
+	const struct tg_color *black_color)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t value = 0;
+
+	CRTC_REG_UPDATE_3(
+		CRTC0_CRTC_BLACK_COLOR,
+		CRTC_BLACK_COLOR_B_CB, black_color->color_b_cb,
+		CRTC_BLACK_COLOR_G_Y, black_color->color_g_y,
+		CRTC_BLACK_COLOR_R_CR, black_color->color_r_cr);
+
+	value = dm_read_reg_soc15(
+				tg->ctx,
+				mmCRTC0_CRTC_BLACK_COLOR,
+				tg110->offsets.crtc);
+	dm_write_reg_soc15(
+		tg->ctx,
+		mmCRTC0_CRTC_BLANK_DATA_COLOR,
+		tg110->offsets.crtc,
+		value);
+}
+
+static void dce120_tg_set_overscan_color(struct timing_generator *tg,
+	const struct tg_color *overscan_color)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	CRTC_REG_SET_3(
+		CRTC0_CRTC_OVERSCAN_COLOR,
+		CRTC_OVERSCAN_COLOR_BLUE, overscan_color->color_b_cb,
+		CRTC_OVERSCAN_COLOR_GREEN, overscan_color->color_g_y,
+		CRTC_OVERSCAN_COLOR_RED, overscan_color->color_r_cr);
+}
+
+static void dce120_tg_program_timing(struct timing_generator *tg,
+	const struct dc_crtc_timing *timing,
+	int vready_offset,
+	int vstartup_start,
+	int vupdate_offset,
+	int vupdate_width,
+	int pstate_keepout,
+	const enum signal_type signal,
+	bool use_vbios)
+{
+	if (use_vbios)
+		dce110_timing_generator_program_timing_generator(tg, timing);
+	else
+		dce120_timing_generator_program_blanking(tg, timing);
+}
+
+static bool dce120_tg_is_blanked(struct timing_generator *tg)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t value = dm_read_reg_soc15(
+			tg->ctx,
+			mmCRTC0_CRTC_BLANK_CONTROL,
+			tg110->offsets.crtc);
+
+	if (get_reg_field_value(
+		value,
+		CRTC0_CRTC_BLANK_CONTROL,
+		CRTC_BLANK_DATA_EN) == 1 &&
+	    get_reg_field_value(
+		value,
+		CRTC0_CRTC_BLANK_CONTROL,
+		CRTC_CURRENT_BLANK_STATE) == 1)
+			return true;
+
+	return false;
+}
+
+static void dce120_tg_set_blank(struct timing_generator *tg,
+		bool enable_blanking)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	CRTC_REG_SET(
+		CRTC0_CRTC_DOUBLE_BUFFER_CONTROL,
+		CRTC_BLANK_DATA_DOUBLE_BUFFER_EN, 1);
+
+	if (enable_blanking)
+		CRTC_REG_SET(CRTC0_CRTC_BLANK_CONTROL, CRTC_BLANK_DATA_EN, 1);
+	else
+		dm_write_reg_soc15(tg->ctx, mmCRTC0_CRTC_BLANK_CONTROL,
+			tg110->offsets.crtc, 0);
+}
+
+bool dce120_tg_validate_timing(struct timing_generator *tg,
+	const struct dc_crtc_timing *timing);
+
+static void dce120_tg_wait_for_state(struct timing_generator *tg,
+	enum crtc_state state)
+{
+	switch (state) {
+	case CRTC_STATE_VBLANK:
+		dce120_timing_generator_wait_for_vblank(tg);
+		break;
+
+	case CRTC_STATE_VACTIVE:
+		dce120_timing_generator_wait_for_vactive(tg);
+		break;
+
+	default:
+		break;
+	}
+}
+
+static void dce120_tg_set_colors(struct timing_generator *tg,
+	const struct tg_color *blank_color,
+	const struct tg_color *overscan_color)
+{
+	if (blank_color != NULL)
+		dce120_tg_program_blank_color(tg, blank_color);
+
+	if (overscan_color != NULL)
+		dce120_tg_set_overscan_color(tg, overscan_color);
+}
+
+static void dce120_timing_generator_set_static_screen_control(
+	struct timing_generator *tg,
+	uint32_t event_triggers,
+	uint32_t num_frames)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	// By register spec, it only takes 8 bit value
+	if (num_frames > 0xFF)
+		num_frames = 0xFF;
+
+	CRTC_REG_UPDATE_2(CRTC0_CRTC_STATIC_SCREEN_CONTROL,
+			CRTC_STATIC_SCREEN_EVENT_MASK, event_triggers,
+			CRTC_STATIC_SCREEN_FRAME_COUNT, num_frames);
+}
+
+static void dce120_timing_generator_set_test_pattern(
+	struct timing_generator *tg,
+	/* TODO: replace 'controller_dp_test_pattern' by 'test_pattern_mode'
+	 * because this is not DP-specific (which is probably somewhere in DP
+	 * encoder) */
+	enum controller_dp_test_pattern test_pattern,
+	enum dc_color_depth color_depth)
+{
+	struct dc_context *ctx = tg->ctx;
+	uint32_t value;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	enum test_pattern_color_format bit_depth;
+	enum test_pattern_dyn_range dyn_range;
+	enum test_pattern_mode mode;
+	/* color ramp generator mixes 16-bits color */
+	uint32_t src_bpc = 16;
+	/* requested bpc */
+	uint32_t dst_bpc;
+	uint32_t index;
+	/* RGB values of the color bars.
+	 * Produce two RGB colors: RGB0 - white (all Fs)
+	 * and RGB1 - black (all 0s)
+	 * (three RGB components for two colors)
+	 */
+	uint16_t src_color[6] = {0xFFFF, 0xFFFF, 0xFFFF, 0x0000,
+						0x0000, 0x0000};
+	/* dest color (converted to the specified color format) */
+	uint16_t dst_color[6];
+	uint32_t inc_base;
+
+	/* translate to bit depth */
+	switch (color_depth) {
+	case COLOR_DEPTH_666:
+		bit_depth = TEST_PATTERN_COLOR_FORMAT_BPC_6;
+	break;
+	case COLOR_DEPTH_888:
+		bit_depth = TEST_PATTERN_COLOR_FORMAT_BPC_8;
+	break;
+	case COLOR_DEPTH_101010:
+		bit_depth = TEST_PATTERN_COLOR_FORMAT_BPC_10;
+	break;
+	case COLOR_DEPTH_121212:
+		bit_depth = TEST_PATTERN_COLOR_FORMAT_BPC_12;
+	break;
+	default:
+		bit_depth = TEST_PATTERN_COLOR_FORMAT_BPC_8;
+	break;
+	}
+
+	switch (test_pattern) {
+	case CONTROLLER_DP_TEST_PATTERN_COLORSQUARES:
+	case CONTROLLER_DP_TEST_PATTERN_COLORSQUARES_CEA:
+	{
+		dyn_range = (test_pattern ==
+				CONTROLLER_DP_TEST_PATTERN_COLORSQUARES_CEA ?
+				TEST_PATTERN_DYN_RANGE_CEA :
+				TEST_PATTERN_DYN_RANGE_VESA);
+		mode = TEST_PATTERN_MODE_COLORSQUARES_RGB;
+
+		CRTC_REG_UPDATE_2(CRTC0_CRTC_TEST_PATTERN_PARAMETERS,
+				CRTC_TEST_PATTERN_VRES, 6,
+				CRTC_TEST_PATTERN_HRES, 6);
+
+		CRTC_REG_UPDATE_4(CRTC0_CRTC_TEST_PATTERN_CONTROL,
+				CRTC_TEST_PATTERN_EN, 1,
+				CRTC_TEST_PATTERN_MODE, mode,
+				CRTC_TEST_PATTERN_DYNAMIC_RANGE, dyn_range,
+				CRTC_TEST_PATTERN_COLOR_FORMAT, bit_depth);
+	}
+	break;
+
+	case CONTROLLER_DP_TEST_PATTERN_VERTICALBARS:
+	case CONTROLLER_DP_TEST_PATTERN_HORIZONTALBARS:
+	{
+		mode = (test_pattern ==
+			CONTROLLER_DP_TEST_PATTERN_VERTICALBARS ?
+			TEST_PATTERN_MODE_VERTICALBARS :
+			TEST_PATTERN_MODE_HORIZONTALBARS);
+
+		switch (bit_depth) {
+		case TEST_PATTERN_COLOR_FORMAT_BPC_6:
+			dst_bpc = 6;
+		break;
+		case TEST_PATTERN_COLOR_FORMAT_BPC_8:
+			dst_bpc = 8;
+		break;
+		case TEST_PATTERN_COLOR_FORMAT_BPC_10:
+			dst_bpc = 10;
+		break;
+		default:
+			dst_bpc = 8;
+		break;
+		}
+
+		/* adjust color to the required colorFormat */
+		for (index = 0; index < 6; index++) {
+			/* dst = 2^dstBpc * src / 2^srcBpc = src >>
+			 * (srcBpc - dstBpc);
+			 */
+			dst_color[index] =
+				src_color[index] >> (src_bpc - dst_bpc);
+		/* CRTC_TEST_PATTERN_DATA has 16 bits,
+		 * lowest 6 are hardwired to ZERO
+		 * color bits should be left aligned aligned to MSB
+		 * XXXXXXXXXX000000 for 10 bit,
+		 * XXXXXXXX00000000 for 8 bit and XXXXXX0000000000 for 6
+		 */
+			dst_color[index] <<= (16 - dst_bpc);
+		}
+
+		dm_write_reg_soc15(ctx, mmCRTC0_CRTC_TEST_PATTERN_PARAMETERS, tg110->offsets.crtc, 0);
+
+		/* We have to write the mask before data, similar to pipeline.
+		 * For example, for 8 bpc, if we want RGB0 to be magenta,
+		 * and RGB1 to be cyan,
+		 * we need to make 7 writes:
+		 * MASK   DATA
+		 * 000001 00000000 00000000                     set mask to R0
+		 * 000010 11111111 00000000     R0 255, 0xFF00, set mask to G0
+		 * 000100 00000000 00000000     G0 0,   0x0000, set mask to B0
+		 * 001000 11111111 00000000     B0 255, 0xFF00, set mask to R1
+		 * 010000 00000000 00000000     R1 0,   0x0000, set mask to G1
+		 * 100000 11111111 00000000     G1 255, 0xFF00, set mask to B1
+		 * 100000 11111111 00000000     B1 255, 0xFF00
+		 *
+		 * we will make a loop of 6 in which we prepare the mask,
+		 * then write, then prepare the color for next write.
+		 * first iteration will write mask only,
+		 * but each next iteration color prepared in
+		 * previous iteration will be written within new mask,
+		 * the last component will written separately,
+		 * mask is not changing between 6th and 7th write
+		 * and color will be prepared by last iteration
+		 */
+
+		/* write color, color values mask in CRTC_TEST_PATTERN_MASK
+		 * is B1, G1, R1, B0, G0, R0
+		 */
+		value = 0;
+		for (index = 0; index < 6; index++) {
+			/* prepare color mask, first write PATTERN_DATA
+			 * will have all zeros
+			 */
+			set_reg_field_value(
+				value,
+				(1 << index),
+				CRTC0_CRTC_TEST_PATTERN_COLOR,
+				CRTC_TEST_PATTERN_MASK);
+			/* write color component */
+			dm_write_reg_soc15(ctx, mmCRTC0_CRTC_TEST_PATTERN_COLOR, tg110->offsets.crtc, value);
+			/* prepare next color component,
+			 * will be written in the next iteration
+			 */
+			set_reg_field_value(
+				value,
+				dst_color[index],
+				CRTC0_CRTC_TEST_PATTERN_COLOR,
+				CRTC_TEST_PATTERN_DATA);
+		}
+		/* write last color component,
+		 * it's been already prepared in the loop
+		 */
+		dm_write_reg_soc15(ctx, mmCRTC0_CRTC_TEST_PATTERN_COLOR, tg110->offsets.crtc, value);
+
+		/* enable test pattern */
+		CRTC_REG_UPDATE_4(CRTC0_CRTC_TEST_PATTERN_CONTROL,
+				CRTC_TEST_PATTERN_EN, 1,
+				CRTC_TEST_PATTERN_MODE, mode,
+				CRTC_TEST_PATTERN_DYNAMIC_RANGE, 0,
+				CRTC_TEST_PATTERN_COLOR_FORMAT, bit_depth);
+	}
+	break;
+
+	case CONTROLLER_DP_TEST_PATTERN_COLORRAMP:
+	{
+		mode = (bit_depth ==
+			TEST_PATTERN_COLOR_FORMAT_BPC_10 ?
+			TEST_PATTERN_MODE_DUALRAMP_RGB :
+			TEST_PATTERN_MODE_SINGLERAMP_RGB);
+
+		switch (bit_depth) {
+		case TEST_PATTERN_COLOR_FORMAT_BPC_6:
+			dst_bpc = 6;
+		break;
+		case TEST_PATTERN_COLOR_FORMAT_BPC_8:
+			dst_bpc = 8;
+		break;
+		case TEST_PATTERN_COLOR_FORMAT_BPC_10:
+			dst_bpc = 10;
+		break;
+		default:
+			dst_bpc = 8;
+		break;
+		}
+
+		/* increment for the first ramp for one color gradation
+		 * 1 gradation for 6-bit color is 2^10
+		 * gradations in 16-bit color
+		 */
+		inc_base = (src_bpc - dst_bpc);
+
+		switch (bit_depth) {
+		case TEST_PATTERN_COLOR_FORMAT_BPC_6:
+		{
+			CRTC_REG_UPDATE_5(CRTC0_CRTC_TEST_PATTERN_PARAMETERS,
+					CRTC_TEST_PATTERN_INC0, inc_base,
+					CRTC_TEST_PATTERN_INC1, 0,
+					CRTC_TEST_PATTERN_HRES, 6,
+					CRTC_TEST_PATTERN_VRES, 6,
+					CRTC_TEST_PATTERN_RAMP0_OFFSET, 0);
+		}
+		break;
+		case TEST_PATTERN_COLOR_FORMAT_BPC_8:
+		{
+			CRTC_REG_UPDATE_5(CRTC0_CRTC_TEST_PATTERN_PARAMETERS,
+					CRTC_TEST_PATTERN_INC0, inc_base,
+					CRTC_TEST_PATTERN_INC1, 0,
+					CRTC_TEST_PATTERN_HRES, 8,
+					CRTC_TEST_PATTERN_VRES, 6,
+					CRTC_TEST_PATTERN_RAMP0_OFFSET, 0);
+		}
+		break;
+		case TEST_PATTERN_COLOR_FORMAT_BPC_10:
+		{
+			CRTC_REG_UPDATE_5(CRTC0_CRTC_TEST_PATTERN_PARAMETERS,
+					CRTC_TEST_PATTERN_INC0, inc_base,
+					CRTC_TEST_PATTERN_INC1, inc_base + 2,
+					CRTC_TEST_PATTERN_HRES, 8,
+					CRTC_TEST_PATTERN_VRES, 5,
+					CRTC_TEST_PATTERN_RAMP0_OFFSET, 384 << 6);
+		}
+		break;
+		default:
+		break;
+		}
+
+		dm_write_reg_soc15(ctx, mmCRTC0_CRTC_TEST_PATTERN_COLOR, tg110->offsets.crtc, 0);
+
+		/* enable test pattern */
+		dm_write_reg_soc15(ctx, mmCRTC0_CRTC_TEST_PATTERN_CONTROL, tg110->offsets.crtc, 0);
+
+		CRTC_REG_UPDATE_4(CRTC0_CRTC_TEST_PATTERN_CONTROL,
+				CRTC_TEST_PATTERN_EN, 1,
+				CRTC_TEST_PATTERN_MODE, mode,
+				CRTC_TEST_PATTERN_DYNAMIC_RANGE, 0,
+				CRTC_TEST_PATTERN_COLOR_FORMAT, bit_depth);
+	}
+	break;
+	case CONTROLLER_DP_TEST_PATTERN_VIDEOMODE:
+	{
+		value = 0;
+		dm_write_reg_soc15(ctx, mmCRTC0_CRTC_TEST_PATTERN_CONTROL, tg110->offsets.crtc,  value);
+		dm_write_reg_soc15(ctx, mmCRTC0_CRTC_TEST_PATTERN_COLOR, tg110->offsets.crtc, value);
+		dm_write_reg_soc15(ctx, mmCRTC0_CRTC_TEST_PATTERN_PARAMETERS, tg110->offsets.crtc, value);
+	}
+	break;
+	default:
+	break;
+	}
+}
+
+static bool dce120_arm_vert_intr(
+		struct timing_generator *tg,
+		uint8_t width)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t v_blank_start, v_blank_end, h_position, v_position;
+
+	tg->funcs->get_scanoutpos(
+				tg,
+				&v_blank_start,
+				&v_blank_end,
+				&h_position,
+				&v_position);
+
+	if (v_blank_start == 0 || v_blank_end == 0)
+		return false;
+
+	CRTC_REG_SET_2(
+			CRTC0_CRTC_VERTICAL_INTERRUPT0_POSITION,
+			CRTC_VERTICAL_INTERRUPT0_LINE_START, v_blank_start,
+			CRTC_VERTICAL_INTERRUPT0_LINE_END, v_blank_start + width);
+
+	return true;
+}
+
+
+static bool dce120_is_tg_enabled(struct timing_generator *tg)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t value, field;
+
+	value = dm_read_reg_soc15(tg->ctx, mmCRTC0_CRTC_CONTROL,
+				  tg110->offsets.crtc);
+	field = get_reg_field_value(value, CRTC0_CRTC_CONTROL,
+				    CRTC_CURRENT_MASTER_EN_STATE);
+
+	return field == 1;
+}
+
+static bool dce120_configure_crc(struct timing_generator *tg,
+				 const struct crc_params *params)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	/* Cannot configure crc on a CRTC that is disabled */
+	if (!dce120_is_tg_enabled(tg))
+		return false;
+
+	if (!params->enable || params->reset)
+		/* First, disable CRC before we configure it. */
+		dm_write_reg_soc15(tg->ctx, mmCRTC0_CRTC_CRC_CNTL,
+				   tg110->offsets.crtc, 0);
+
+	if (!params->enable)
+		return true;
+
+	/* Program frame boundaries */
+	switch (params->crc_eng_inst) {
+	case 0:
+		/* Window A x axis start and end. */
+		CRTC_REG_UPDATE_2(CRTC0_CRTC_CRC0_WINDOWA_X_CONTROL,
+				  CRTC_CRC0_WINDOWA_X_START, params->windowa_x_start,
+				  CRTC_CRC0_WINDOWA_X_END, params->windowa_x_end);
+
+		/* Window A y axis start and end. */
+		CRTC_REG_UPDATE_2(CRTC0_CRTC_CRC0_WINDOWA_Y_CONTROL,
+				  CRTC_CRC0_WINDOWA_Y_START, params->windowa_y_start,
+				  CRTC_CRC0_WINDOWA_Y_END, params->windowa_y_end);
+
+		/* Window B x axis start and end. */
+		CRTC_REG_UPDATE_2(CRTC0_CRTC_CRC0_WINDOWB_X_CONTROL,
+				  CRTC_CRC0_WINDOWB_X_START, params->windowb_x_start,
+				  CRTC_CRC0_WINDOWB_X_END, params->windowb_x_end);
+
+		/* Window B y axis start and end. */
+		CRTC_REG_UPDATE_2(CRTC0_CRTC_CRC0_WINDOWB_Y_CONTROL,
+				  CRTC_CRC0_WINDOWB_Y_START, params->windowb_y_start,
+				  CRTC_CRC0_WINDOWB_Y_END, params->windowb_y_end);
+
+		/* Set crc mode and selection, and enable.*/
+		CRTC_REG_UPDATE_3(CRTC0_CRTC_CRC_CNTL,
+				  CRTC_CRC_CONT_EN, params->continuous_mode ? 1 : 0,
+				  CRTC_CRC0_SELECT, params->selection,
+				  CRTC_CRC_EN, 1);
+		break;
+	case 1:
+		/* Window A x axis start and end. */
+		CRTC_REG_UPDATE_2(CRTC0_CRTC_CRC1_WINDOWA_X_CONTROL,
+				  CRTC_CRC1_WINDOWA_X_START, params->windowa_x_start,
+				  CRTC_CRC1_WINDOWA_X_END, params->windowa_x_end);
+
+		/* Window A y axis start and end. */
+		CRTC_REG_UPDATE_2(CRTC0_CRTC_CRC1_WINDOWA_Y_CONTROL,
+				  CRTC_CRC1_WINDOWA_Y_START, params->windowa_y_start,
+				  CRTC_CRC1_WINDOWA_Y_END, params->windowa_y_end);
+
+		/* Window B x axis start and end. */
+		CRTC_REG_UPDATE_2(CRTC0_CRTC_CRC1_WINDOWB_X_CONTROL,
+				  CRTC_CRC1_WINDOWB_X_START, params->windowb_x_start,
+				  CRTC_CRC1_WINDOWB_X_END, params->windowb_x_end);
+
+		/* Window B y axis start and end. */
+		CRTC_REG_UPDATE_2(CRTC0_CRTC_CRC1_WINDOWB_Y_CONTROL,
+				  CRTC_CRC1_WINDOWB_Y_START, params->windowb_y_start,
+				  CRTC_CRC1_WINDOWB_Y_END, params->windowb_y_end);
+
+		/* Set crc mode and selection, and enable */
+		CRTC_REG_UPDATE_3(CRTC0_CRTC_CRC_CNTL,
+				  CRTC_CRC_CONT_EN, params->continuous_mode ? 1 : 0,
+				  CRTC_CRC1_SELECT, params->selection,
+				  CRTC_CRC_EN, 1);
+		break;
+	default:
+		return false;
+	}
+
+	return true;
+}
+
+static bool dce120_get_crc(struct timing_generator *tg, uint8_t idx,
+			uint32_t *r_cr, uint32_t *g_y, uint32_t *b_cb)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t value, field;
+
+	value = dm_read_reg_soc15(tg->ctx, mmCRTC0_CRTC_CRC_CNTL,
+				  tg110->offsets.crtc);
+	field = get_reg_field_value(value, CRTC0_CRTC_CRC_CNTL, CRTC_CRC_EN);
+
+	/* Early return if CRC is not enabled for this CRTC */
+	if (!field)
+		return false;
+
+	switch (idx) {
+	case 0:
+		value = dm_read_reg_soc15(tg->ctx, mmCRTC0_CRTC_CRC0_DATA_RG,
+					  tg110->offsets.crtc);
+		*r_cr = get_reg_field_value(value, CRTC0_CRTC_CRC0_DATA_RG, CRC0_R_CR);
+		*g_y = get_reg_field_value(value, CRTC0_CRTC_CRC0_DATA_RG, CRC0_G_Y);
+
+		value = dm_read_reg_soc15(tg->ctx, mmCRTC0_CRTC_CRC0_DATA_B,
+					  tg110->offsets.crtc);
+		*b_cb = get_reg_field_value(value, CRTC0_CRTC_CRC0_DATA_B, CRC0_B_CB);
+		break;
+	case 1:
+		value = dm_read_reg_soc15(tg->ctx, mmCRTC0_CRTC_CRC1_DATA_RG,
+					  tg110->offsets.crtc);
+		*r_cr = get_reg_field_value(value, CRTC0_CRTC_CRC1_DATA_RG, CRC1_R_CR);
+		*g_y = get_reg_field_value(value, CRTC0_CRTC_CRC1_DATA_RG, CRC1_G_Y);
+
+		value = dm_read_reg_soc15(tg->ctx, mmCRTC0_CRTC_CRC1_DATA_B,
+					  tg110->offsets.crtc);
+		*b_cb = get_reg_field_value(value, CRTC0_CRTC_CRC1_DATA_B, CRC1_B_CB);
+		break;
+	default:
+		return false;
+	}
+
+	return true;
+}
+
+static const struct timing_generator_funcs dce120_tg_funcs = {
+		.validate_timing = dce120_tg_validate_timing,
+		.program_timing = dce120_tg_program_timing,
+		.enable_crtc = dce120_timing_generator_enable_crtc,
+		.disable_crtc = dce110_timing_generator_disable_crtc,
+		/* used by enable_timing_synchronization. Not need for FPGA */
+		.is_counter_moving = dce110_timing_generator_is_counter_moving,
+		/* never be called */
+		.get_position = dce120_timing_generator_get_crtc_position,
+		.get_frame_count = dce120_timing_generator_get_vblank_counter,
+		.get_scanoutpos = dce120_timing_generator_get_crtc_scanoutpos,
+		.set_early_control = dce120_timing_generator_set_early_control,
+		/* used by enable_timing_synchronization. Not need for FPGA */
+		.wait_for_state = dce120_tg_wait_for_state,
+		.set_blank = dce120_tg_set_blank,
+		.is_blanked = dce120_tg_is_blanked,
+		/* never be called */
+		.set_colors = dce120_tg_set_colors,
+		.set_overscan_blank_color = dce120_timing_generator_set_overscan_color_black,
+		.set_blank_color = dce120_timing_generator_program_blank_color,
+		.disable_vga = dce120_timing_generator_disable_vga,
+		.did_triggered_reset_occur = dce120_timing_generator_did_triggered_reset_occur,
+		.setup_global_swap_lock = dce120_timing_generator_setup_global_swap_lock,
+		.enable_reset_trigger = dce120_timing_generator_enable_reset_trigger,
+		.disable_reset_trigger = dce120_timing_generator_disable_reset_trigger,
+		.tear_down_global_swap_lock = dce120_timing_generator_tear_down_global_swap_lock,
+		.enable_advanced_request = dce120_timing_generator_enable_advanced_request,
+		.set_drr = dce120_timing_generator_set_drr,
+		.get_last_used_drr_vtotal = NULL,
+		.set_static_screen_control = dce120_timing_generator_set_static_screen_control,
+		.set_test_pattern = dce120_timing_generator_set_test_pattern,
+		.arm_vert_intr = dce120_arm_vert_intr,
+		.is_tg_enabled = dce120_is_tg_enabled,
+		.configure_crc = dce120_configure_crc,
+		.get_crc = dce120_get_crc,
+		.is_two_pixels_per_container = dce110_is_two_pixels_per_container,
+};
+
+
+void dce120_timing_generator_construct(
+	struct dce110_timing_generator *tg110,
+	struct dc_context *ctx,
+	uint32_t instance,
+	const struct dce110_timing_generator_offsets *offsets)
+{
+	tg110->controller_id = CONTROLLER_ID_D0 + instance;
+	tg110->base.inst = instance;
+
+	tg110->offsets = *offsets;
+
+	tg110->base.funcs = &dce120_tg_funcs;
+
+	tg110->base.ctx = ctx;
+	tg110->base.bp = ctx->dc_bios;
+
+	tg110->max_h_total = CRTC0_CRTC_H_TOTAL__CRTC_H_TOTAL_MASK + 1;
+	tg110->max_v_total = CRTC0_CRTC_V_TOTAL__CRTC_V_TOTAL_MASK + 1;
+
+	/*//CRTC requires a minimum HBLANK = 32 pixels and o
+	 * Minimum HSYNC = 8 pixels*/
+	tg110->min_h_blank = 32;
+	/*DCE12_CRTC_Block_ARch.doc*/
+	tg110->min_h_front_porch = 0;
+	tg110->min_h_back_porch = 0;
+
+	tg110->min_h_sync_width = 4;
+	tg110->min_v_sync_width = 1;
+	tg110->min_v_blank = 3;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dce120/dce120_timing_generator.h b/drivers/gpu/drm/amd/display/dc/dce120/dce120_timing_generator.h
new file mode 100644
index 000000000000..549d70b23e82
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce120/dce120_timing_generator.h
@@ -0,0 +1,40 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_TIMING_GENERATOR_DCE120_H__
+#define __DC_TIMING_GENERATOR_DCE120_H__
+
+#include "timing_generator.h"
+#include "../include/grph_object_id.h"
+#include "dce110/dce110_timing_generator.h"
+
+
+void dce120_timing_generator_construct(
+	struct dce110_timing_generator *tg110,
+	struct dc_context *ctx,
+	uint32_t instance,
+	const struct dce110_timing_generator_offsets *offsets);
+
+#endif /* __DC_TIMING_GENERATOR_DCE120_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dce60/Makefile b/drivers/gpu/drm/amd/display/dc/dce60/Makefile
new file mode 100644
index 000000000000..824f73eb3326
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce60/Makefile
@@ -0,0 +1,35 @@
+#
+# Copyright 2020 Mauro Rossi <issor.oruam@gmail.com>
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+#
+# Makefile for the 'controller' sub-component of DAL.
+# It provides the control and status of HW CRTC block.
+
+CFLAGS_$(AMDDALPATH)/dc/dce60/dce60_resource.o = -Wno-override-init
+
+DCE60 = dce60_timing_generator.o
+
+AMD_DAL_DCE60 = $(addprefix $(AMDDALPATH)/dc/dce60/,$(DCE60))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCE60)
+
+
+
diff --git a/drivers/gpu/drm/amd/display/dc/dce60/dce60_timing_generator.c b/drivers/gpu/drm/amd/display/dc/dce60/dce60_timing_generator.c
new file mode 100644
index 000000000000..e691a1cf3356
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce60/dce60_timing_generator.c
@@ -0,0 +1,268 @@
+/*
+ * Copyright 2020 Mauro Rossi <issor.oruam@gmail.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+
+/* include DCE6 register header files */
+#include "dce/dce_6_0_d.h"
+#include "dce/dce_6_0_sh_mask.h"
+
+#include "dc_types.h"
+
+#include "include/grph_object_id.h"
+#include "include/logger_interface.h"
+#include "../dce110/dce110_timing_generator.h"
+#include "dce60_timing_generator.h"
+
+#include "timing_generator.h"
+
+enum black_color_format {
+	BLACK_COLOR_FORMAT_RGB_FULLRANGE = 0,	/* used as index in array */
+	BLACK_COLOR_FORMAT_RGB_LIMITED,
+	BLACK_COLOR_FORMAT_YUV_TV,
+	BLACK_COLOR_FORMAT_YUV_CV,
+	BLACK_COLOR_FORMAT_YUV_SUPER_AA,
+
+	BLACK_COLOR_FORMAT_COUNT
+};
+
+static const struct dce110_timing_generator_offsets reg_offsets[] = {
+{
+	.crtc = (mmCRTC0_DCFE_MEM_LIGHT_SLEEP_CNTL - mmCRTC0_DCFE_MEM_LIGHT_SLEEP_CNTL),
+	.dcp = (mmDCP0_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
+},
+{
+	.crtc = (mmCRTC1_DCFE_MEM_LIGHT_SLEEP_CNTL - mmCRTC0_DCFE_MEM_LIGHT_SLEEP_CNTL),
+	.dcp = (mmDCP1_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
+},
+{
+	.crtc = (mmCRTC2_DCFE_MEM_LIGHT_SLEEP_CNTL - mmCRTC0_DCFE_MEM_LIGHT_SLEEP_CNTL),
+	.dcp = (mmDCP2_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
+},
+{
+	.crtc = (mmCRTC3_DCFE_MEM_LIGHT_SLEEP_CNTL - mmCRTC0_DCFE_MEM_LIGHT_SLEEP_CNTL),
+	.dcp = (mmDCP3_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
+},
+{
+	.crtc = (mmCRTC4_DCFE_MEM_LIGHT_SLEEP_CNTL - mmCRTC0_DCFE_MEM_LIGHT_SLEEP_CNTL),
+	.dcp = (mmDCP4_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
+},
+{
+	.crtc = (mmCRTC5_DCFE_MEM_LIGHT_SLEEP_CNTL - mmCRTC0_DCFE_MEM_LIGHT_SLEEP_CNTL),
+	.dcp = (mmDCP5_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
+}
+};
+
+#define NUMBER_OF_FRAME_TO_WAIT_ON_TRIGGERED_RESET 10
+
+#define MAX_H_TOTAL (CRTC_H_TOTAL__CRTC_H_TOTAL_MASK + 1)
+#define MAX_V_TOTAL (CRTC_V_TOTAL__CRTC_V_TOTAL_MASKhw + 1)
+
+#define CRTC_REG(reg) (reg + tg110->offsets.crtc)
+#define DCP_REG(reg) (reg + tg110->offsets.dcp)
+#define DMIF_REG(reg) (reg + tg110->offsets.dmif)
+
+static void program_pix_dur(struct timing_generator *tg, uint32_t pix_clk_100hz)
+{
+	uint64_t pix_dur;
+	uint32_t addr = mmDMIF_PG0_DPG_PIPE_ARBITRATION_CONTROL1
+					+ DCE110TG_FROM_TG(tg)->offsets.dmif;
+	uint32_t value = dm_read_reg(tg->ctx, addr);
+
+	if (pix_clk_100hz == 0)
+		return;
+
+	pix_dur = div_u64(10000000000ull, pix_clk_100hz);
+
+	set_reg_field_value(
+		value,
+		pix_dur,
+		DPG_PIPE_ARBITRATION_CONTROL1,
+		PIXEL_DURATION);
+
+	dm_write_reg(tg->ctx, addr, value);
+}
+
+static void program_timing(struct timing_generator *tg,
+	const struct dc_crtc_timing *timing,
+	int vready_offset,
+	int vstartup_start,
+	int vupdate_offset,
+	int vupdate_width,
+	int pstate_keepout,
+	const enum signal_type signal,
+	bool use_vbios)
+{
+	if (!use_vbios)
+		program_pix_dur(tg, timing->pix_clk_100hz);
+
+	dce110_tg_program_timing(tg, timing, 0, 0, 0, 0, 0, 0, use_vbios);
+}
+
+static void dce60_timing_generator_enable_advanced_request(
+	struct timing_generator *tg,
+	bool enable,
+	const struct dc_crtc_timing *timing)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t addr = CRTC_REG(mmCRTC_START_LINE_CONTROL);
+	uint32_t value = dm_read_reg(tg->ctx, addr);
+	/* DCE6 has CRTC_PREFETCH_EN bit in CRTC_CONTROL register */
+	uint32_t addr2 = CRTC_REG(mmCRTC_CONTROL);
+	uint32_t value2 = dm_read_reg(tg->ctx, addr2);
+
+	/* DCE6 does not support CRTC_LEGACY_REQUESTOR_EN bit
+	   so here is not possible to set bit based on enable argument */
+
+	if ((timing->v_sync_width + timing->v_front_porch) <= 3) {
+		set_reg_field_value(
+			value,
+			3,
+			CRTC_START_LINE_CONTROL,
+			CRTC_ADVANCED_START_LINE_POSITION);
+		set_reg_field_value(
+			value2,
+			0,
+			CRTC_CONTROL,
+			CRTC_PREFETCH_EN);
+	} else {
+		set_reg_field_value(
+			value,
+			4,
+			CRTC_START_LINE_CONTROL,
+			CRTC_ADVANCED_START_LINE_POSITION);
+		set_reg_field_value(
+			value2,
+			1,
+			CRTC_CONTROL,
+			CRTC_PREFETCH_EN);
+	}
+
+	set_reg_field_value(
+		value,
+		1,
+		CRTC_START_LINE_CONTROL,
+		CRTC_PROGRESSIVE_START_LINE_EARLY);
+
+	set_reg_field_value(
+		value,
+		1,
+		CRTC_START_LINE_CONTROL,
+		CRTC_INTERLACE_START_LINE_EARLY);
+
+	dm_write_reg(tg->ctx, addr, value);
+	dm_write_reg(tg->ctx, addr2, value2);
+}
+
+static bool dce60_is_tg_enabled(struct timing_generator *tg)
+{
+	uint32_t addr = 0;
+	uint32_t value = 0;
+	uint32_t field = 0;
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+
+	addr = CRTC_REG(mmCRTC_CONTROL);
+	value = dm_read_reg(tg->ctx, addr);
+	field = get_reg_field_value(value, CRTC_CONTROL,
+				    CRTC_CURRENT_MASTER_EN_STATE);
+	return field == 1;
+}
+
+static bool dce60_configure_crc(struct timing_generator *tg,
+				const struct crc_params *params)
+{
+	/* Cannot configure crc on a CRTC that is disabled */
+	if (!dce60_is_tg_enabled(tg))
+		return false;
+
+	/* DCE6 has no CRTC_CRC_CNTL register, nothing to do */
+
+	return true;
+}
+
+static const struct timing_generator_funcs dce60_tg_funcs = {
+		.validate_timing = dce110_tg_validate_timing,
+		.program_timing = program_timing,
+		.enable_crtc = dce110_timing_generator_enable_crtc,
+		.disable_crtc = dce110_timing_generator_disable_crtc,
+		.is_counter_moving = dce110_timing_generator_is_counter_moving,
+		.get_position = dce110_timing_generator_get_position,
+		.get_frame_count = dce110_timing_generator_get_vblank_counter,
+		.get_scanoutpos = dce110_timing_generator_get_crtc_scanoutpos,
+		.set_early_control = dce110_timing_generator_set_early_control,
+		.wait_for_state = dce110_tg_wait_for_state,
+		.set_blank = dce110_tg_set_blank,
+		.is_blanked = dce110_tg_is_blanked,
+		.set_colors = dce110_tg_set_colors,
+		.set_overscan_blank_color =
+				dce110_timing_generator_set_overscan_color_black,
+		.set_blank_color = dce110_timing_generator_program_blank_color,
+		.disable_vga = dce110_timing_generator_disable_vga,
+		.did_triggered_reset_occur =
+				dce110_timing_generator_did_triggered_reset_occur,
+		.setup_global_swap_lock =
+				dce110_timing_generator_setup_global_swap_lock,
+		.enable_reset_trigger = dce110_timing_generator_enable_reset_trigger,
+		.disable_reset_trigger = dce110_timing_generator_disable_reset_trigger,
+		.tear_down_global_swap_lock =
+				dce110_timing_generator_tear_down_global_swap_lock,
+		.set_drr = dce110_timing_generator_set_drr,
+		.set_static_screen_control =
+			dce110_timing_generator_set_static_screen_control,
+		.set_test_pattern = dce110_timing_generator_set_test_pattern,
+		.arm_vert_intr = dce110_arm_vert_intr,
+
+		/* DCE6.0 overrides */
+		.enable_advanced_request =
+				dce60_timing_generator_enable_advanced_request,
+		.configure_crc = dce60_configure_crc,
+		.get_crc = dce110_get_crc,
+		.is_two_pixels_per_container = dce110_is_two_pixels_per_container,
+};
+
+void dce60_timing_generator_construct(
+	struct dce110_timing_generator *tg110,
+	struct dc_context *ctx,
+	uint32_t instance,
+	const struct dce110_timing_generator_offsets *offsets)
+{
+	tg110->controller_id = CONTROLLER_ID_D0 + instance;
+	tg110->base.inst = instance;
+	tg110->offsets = *offsets;
+	tg110->derived_offsets = reg_offsets[instance];
+
+	tg110->base.funcs = &dce60_tg_funcs;
+
+	tg110->base.ctx = ctx;
+	tg110->base.bp = ctx->dc_bios;
+
+	tg110->max_h_total = CRTC_H_TOTAL__CRTC_H_TOTAL_MASK + 1;
+	tg110->max_v_total = CRTC_V_TOTAL__CRTC_V_TOTAL_MASK + 1;
+
+	tg110->min_h_blank = 56;
+	tg110->min_h_front_porch = 4;
+	tg110->min_h_back_porch = 4;
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/dce60/dce60_timing_generator.h b/drivers/gpu/drm/amd/display/dc/dce60/dce60_timing_generator.h
new file mode 100644
index 000000000000..81d831233cc5
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce60/dce60_timing_generator.h
@@ -0,0 +1,39 @@
+/*
+ * Copyright 2020 Mauro Rossi <issor.oruam@gmail.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_TIMING_GENERATOR_DCE60_H__
+#define __DC_TIMING_GENERATOR_DCE60_H__
+
+#include "timing_generator.h"
+#include "../include/grph_object_id.h"
+
+/* DCE6.0 implementation inherits from DCE11.0 */
+void dce60_timing_generator_construct(
+	struct dce110_timing_generator *tg,
+	struct dc_context *ctx,
+	uint32_t instance,
+	const struct dce110_timing_generator_offsets *offsets);
+
+#endif /* __DC_TIMING_GENERATOR_DCE60_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dce80/Makefile b/drivers/gpu/drm/amd/display/dc/dce80/Makefile
new file mode 100644
index 000000000000..fba189d26652
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce80/Makefile
@@ -0,0 +1,35 @@
+#
+# Copyright 2017 Advanced Micro Devices, Inc.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+#
+# Makefile for the 'controller' sub-component of DAL.
+# It provides the control and status of HW CRTC block.
+
+CFLAGS_$(AMDDALPATH)/dc/dce80/dce80_resource.o = -Wno-override-init
+
+DCE80 = dce80_timing_generator.o
+
+AMD_DAL_DCE80 = $(addprefix $(AMDDALPATH)/dc/dce80/,$(DCE80))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCE80)
+
+
+
diff --git a/drivers/gpu/drm/amd/display/dc/dce80/dce80_timing_generator.c b/drivers/gpu/drm/amd/display/dc/dce80/dce80_timing_generator.c
new file mode 100644
index 000000000000..88e7a1fc9a30
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce80/dce80_timing_generator.c
@@ -0,0 +1,250 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+
+/* include DCE8 register header files */
+#include "dce/dce_8_0_d.h"
+#include "dce/dce_8_0_sh_mask.h"
+
+#include "dc_types.h"
+
+#include "include/grph_object_id.h"
+#include "include/logger_interface.h"
+#include "../dce110/dce110_timing_generator.h"
+#include "dce80_timing_generator.h"
+
+#include "timing_generator.h"
+
+enum black_color_format {
+	BLACK_COLOR_FORMAT_RGB_FULLRANGE = 0,	/* used as index in array */
+	BLACK_COLOR_FORMAT_RGB_LIMITED,
+	BLACK_COLOR_FORMAT_YUV_TV,
+	BLACK_COLOR_FORMAT_YUV_CV,
+	BLACK_COLOR_FORMAT_YUV_SUPER_AA,
+
+	BLACK_COLOR_FORMAT_COUNT
+};
+
+static const struct dce110_timing_generator_offsets reg_offsets[] = {
+{
+	.crtc = (mmCRTC0_DCFE_MEM_LIGHT_SLEEP_CNTL - mmCRTC0_DCFE_MEM_LIGHT_SLEEP_CNTL),
+	.dcp = (mmDCP0_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
+},
+{
+	.crtc = (mmCRTC1_DCFE_MEM_LIGHT_SLEEP_CNTL - mmCRTC0_DCFE_MEM_LIGHT_SLEEP_CNTL),
+	.dcp = (mmDCP1_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
+},
+{
+	.crtc = (mmCRTC2_DCFE_MEM_LIGHT_SLEEP_CNTL - mmCRTC0_DCFE_MEM_LIGHT_SLEEP_CNTL),
+	.dcp = (mmDCP2_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
+},
+{
+	.crtc = (mmCRTC3_DCFE_MEM_LIGHT_SLEEP_CNTL - mmCRTC0_DCFE_MEM_LIGHT_SLEEP_CNTL),
+	.dcp = (mmDCP3_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
+},
+{
+	.crtc = (mmCRTC4_DCFE_MEM_LIGHT_SLEEP_CNTL - mmCRTC0_DCFE_MEM_LIGHT_SLEEP_CNTL),
+	.dcp = (mmDCP4_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
+},
+{
+	.crtc = (mmCRTC5_DCFE_MEM_LIGHT_SLEEP_CNTL - mmCRTC0_DCFE_MEM_LIGHT_SLEEP_CNTL),
+	.dcp = (mmDCP5_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
+}
+};
+
+#define NUMBER_OF_FRAME_TO_WAIT_ON_TRIGGERED_RESET 10
+
+#define MAX_H_TOTAL (CRTC_H_TOTAL__CRTC_H_TOTAL_MASK + 1)
+#define MAX_V_TOTAL (CRTC_V_TOTAL__CRTC_V_TOTAL_MASKhw + 1)
+
+#define CRTC_REG(reg) (reg + tg110->offsets.crtc)
+#define DCP_REG(reg) (reg + tg110->offsets.dcp)
+#define DMIF_REG(reg) (reg + tg110->offsets.dmif)
+
+static void program_pix_dur(struct timing_generator *tg, uint32_t pix_clk_100hz)
+{
+	uint64_t pix_dur;
+	uint32_t addr = mmDMIF_PG0_DPG_PIPE_ARBITRATION_CONTROL1
+					+ DCE110TG_FROM_TG(tg)->offsets.dmif;
+	uint32_t value = dm_read_reg(tg->ctx, addr);
+
+	if (pix_clk_100hz == 0)
+		return;
+
+	pix_dur = div_u64(10000000000ull, pix_clk_100hz);
+
+	set_reg_field_value(
+		value,
+		pix_dur,
+		DPG_PIPE_ARBITRATION_CONTROL1,
+		PIXEL_DURATION);
+
+	dm_write_reg(tg->ctx, addr, value);
+}
+
+static void dce80_timing_generator_program_timing(struct timing_generator *tg,
+	const struct dc_crtc_timing *timing,
+	int vready_offset,
+	int vstartup_start,
+	int vupdate_offset,
+	int vupdate_width,
+	int pstate_keepout,
+	const enum signal_type signal,
+	bool use_vbios)
+{
+	if (!use_vbios)
+		program_pix_dur(tg, timing->pix_clk_100hz);
+
+	dce110_tg_program_timing(tg, timing, 0, 0, 0, 0, 0, 0, use_vbios);
+}
+
+static void dce80_timing_generator_enable_advanced_request(
+	struct timing_generator *tg,
+	bool enable,
+	const struct dc_crtc_timing *timing)
+{
+	struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+	uint32_t addr = CRTC_REG(mmCRTC_START_LINE_CONTROL);
+	uint32_t value = dm_read_reg(tg->ctx, addr);
+
+	if (enable) {
+		set_reg_field_value(
+			value,
+			0,
+			CRTC_START_LINE_CONTROL,
+			CRTC_LEGACY_REQUESTOR_EN);
+	} else {
+		set_reg_field_value(
+			value,
+			1,
+			CRTC_START_LINE_CONTROL,
+			CRTC_LEGACY_REQUESTOR_EN);
+	}
+
+	if ((timing->v_sync_width + timing->v_front_porch) <= 3) {
+		set_reg_field_value(
+			value,
+			3,
+			CRTC_START_LINE_CONTROL,
+			CRTC_ADVANCED_START_LINE_POSITION);
+		set_reg_field_value(
+			value,
+			0,
+			CRTC_START_LINE_CONTROL,
+			CRTC_PREFETCH_EN);
+	} else {
+		set_reg_field_value(
+			value,
+			4,
+			CRTC_START_LINE_CONTROL,
+			CRTC_ADVANCED_START_LINE_POSITION);
+		set_reg_field_value(
+			value,
+			1,
+			CRTC_START_LINE_CONTROL,
+			CRTC_PREFETCH_EN);
+	}
+
+	set_reg_field_value(
+		value,
+		1,
+		CRTC_START_LINE_CONTROL,
+		CRTC_PROGRESSIVE_START_LINE_EARLY);
+
+	set_reg_field_value(
+		value,
+		1,
+		CRTC_START_LINE_CONTROL,
+		CRTC_INTERLACE_START_LINE_EARLY);
+
+	dm_write_reg(tg->ctx, addr, value);
+}
+
+static const struct timing_generator_funcs dce80_tg_funcs = {
+		.validate_timing = dce110_tg_validate_timing,
+		.program_timing = dce80_timing_generator_program_timing,
+		.enable_crtc = dce110_timing_generator_enable_crtc,
+		.disable_crtc = dce110_timing_generator_disable_crtc,
+		.is_counter_moving = dce110_timing_generator_is_counter_moving,
+		.get_position = dce110_timing_generator_get_position,
+		.get_frame_count = dce110_timing_generator_get_vblank_counter,
+		.get_scanoutpos = dce110_timing_generator_get_crtc_scanoutpos,
+		.set_early_control = dce110_timing_generator_set_early_control,
+		.wait_for_state = dce110_tg_wait_for_state,
+		.set_blank = dce110_tg_set_blank,
+		.is_blanked = dce110_tg_is_blanked,
+		.set_colors = dce110_tg_set_colors,
+		.set_overscan_blank_color =
+				dce110_timing_generator_set_overscan_color_black,
+		.set_blank_color = dce110_timing_generator_program_blank_color,
+		.disable_vga = dce110_timing_generator_disable_vga,
+		.did_triggered_reset_occur =
+				dce110_timing_generator_did_triggered_reset_occur,
+		.setup_global_swap_lock =
+				dce110_timing_generator_setup_global_swap_lock,
+		.enable_reset_trigger = dce110_timing_generator_enable_reset_trigger,
+		.disable_reset_trigger = dce110_timing_generator_disable_reset_trigger,
+		.tear_down_global_swap_lock =
+				dce110_timing_generator_tear_down_global_swap_lock,
+		.set_drr = dce110_timing_generator_set_drr,
+		.get_last_used_drr_vtotal = NULL,
+		.set_static_screen_control =
+			dce110_timing_generator_set_static_screen_control,
+		.set_test_pattern = dce110_timing_generator_set_test_pattern,
+		.arm_vert_intr = dce110_arm_vert_intr,
+
+		/* DCE8.0 overrides */
+		.enable_advanced_request =
+				dce80_timing_generator_enable_advanced_request,
+		.configure_crc = dce110_configure_crc,
+		.get_crc = dce110_get_crc,
+		.is_two_pixels_per_container = dce110_is_two_pixels_per_container,
+};
+
+void dce80_timing_generator_construct(
+	struct dce110_timing_generator *tg110,
+	struct dc_context *ctx,
+	uint32_t instance,
+	const struct dce110_timing_generator_offsets *offsets)
+{
+	tg110->controller_id = CONTROLLER_ID_D0 + instance;
+	tg110->base.inst = instance;
+	tg110->offsets = *offsets;
+	tg110->derived_offsets = reg_offsets[instance];
+
+	tg110->base.funcs = &dce80_tg_funcs;
+
+	tg110->base.ctx = ctx;
+	tg110->base.bp = ctx->dc_bios;
+
+	tg110->max_h_total = CRTC_H_TOTAL__CRTC_H_TOTAL_MASK + 1;
+	tg110->max_v_total = CRTC_V_TOTAL__CRTC_V_TOTAL_MASK + 1;
+
+	tg110->min_h_blank = 56;
+	tg110->min_h_front_porch = 4;
+	tg110->min_h_back_porch = 4;
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/dce80/dce80_timing_generator.h b/drivers/gpu/drm/amd/display/dc/dce80/dce80_timing_generator.h
new file mode 100644
index 000000000000..8ff1b06bcd8b
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dce80/dce80_timing_generator.h
@@ -0,0 +1,39 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_TIMING_GENERATOR_DCE80_H__
+#define __DC_TIMING_GENERATOR_DCE80_H__
+
+#include "timing_generator.h"
+#include "../include/grph_object_id.h"
+
+/* DCE8.0 implementation inherits from DCE11.0 */
+void dce80_timing_generator_construct(
+	struct dce110_timing_generator *tg,
+	struct dc_context *ctx,
+	uint32_t instance,
+	const struct dce110_timing_generator_offsets *offsets);
+
+#endif /* __DC_TIMING_GENERATOR_DCE80_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/Makefile b/drivers/gpu/drm/amd/display/dc/dcn10/Makefile
new file mode 100644
index 000000000000..e1f6623d4936
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn10/Makefile
@@ -0,0 +1,31 @@
+#
+# Copyright 2017 Advanced Micro Devices, Inc.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+#
+# Makefile for DCN.
+
+DCN10 = dcn10_ipp.o \
+		dcn10_hw_sequencer_debug.o \
+		dcn10_cm_common.o \
+
+AMD_DAL_DCN10 = $(addprefix $(AMDDALPATH)/dc/dcn10/,$(DCN10))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCN10)
diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_cm_common.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_cm_common.c
new file mode 100644
index 000000000000..dcd2cdfe91eb
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_cm_common.c
@@ -0,0 +1,683 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+#include "dc.h"
+#include "reg_helper.h"
+#include "dcn10/dcn10_dpp.h"
+
+#include "dcn10_cm_common.h"
+#include "custom_float.h"
+
+#define REG(reg) reg
+
+#define CTX \
+	ctx
+
+#undef FN
+#define FN(reg_name, field_name) \
+	reg->shifts.field_name, reg->masks.field_name
+
+void cm_helper_program_color_matrices(
+		struct dc_context *ctx,
+		const uint16_t *regval,
+		const struct color_matrices_reg *reg)
+{
+	uint32_t cur_csc_reg;
+	unsigned int i = 0;
+
+	for (cur_csc_reg = reg->csc_c11_c12;
+			cur_csc_reg <= reg->csc_c33_c34;
+			cur_csc_reg++) {
+
+		const uint16_t *regval0 = &(regval[2 * i]);
+		const uint16_t *regval1 = &(regval[(2 * i) + 1]);
+
+		REG_SET_2(cur_csc_reg, 0,
+				csc_c11, *regval0,
+				csc_c12, *regval1);
+
+		i++;
+	}
+
+}
+
+void cm_helper_read_color_matrices(struct dc_context *ctx,
+				   uint16_t *regval,
+				   const struct color_matrices_reg *reg)
+{
+	uint32_t cur_csc_reg, regval0, regval1;
+	unsigned int i = 0;
+
+	for (cur_csc_reg = reg->csc_c11_c12;
+	     cur_csc_reg <= reg->csc_c33_c34; cur_csc_reg++) {
+		REG_GET_2(cur_csc_reg,
+				csc_c11, &regval0,
+				csc_c12, &regval1);
+
+		regval[2 * i] = regval0;
+		regval[(2 * i) + 1] = regval1;
+
+		i++;
+	}
+}
+
+void cm_helper_program_xfer_func(
+		struct dc_context *ctx,
+		const struct pwl_params *params,
+		const struct xfer_func_reg *reg)
+{
+	uint32_t reg_region_cur;
+	unsigned int i = 0;
+
+	REG_SET_2(reg->start_cntl_b, 0,
+			exp_region_start, params->corner_points[0].blue.custom_float_x,
+			exp_resion_start_segment, 0);
+	REG_SET_2(reg->start_cntl_g, 0,
+			exp_region_start, params->corner_points[0].green.custom_float_x,
+			exp_resion_start_segment, 0);
+	REG_SET_2(reg->start_cntl_r, 0,
+			exp_region_start, params->corner_points[0].red.custom_float_x,
+			exp_resion_start_segment, 0);
+
+	REG_SET(reg->start_slope_cntl_b, 0,
+			field_region_linear_slope, params->corner_points[0].blue.custom_float_slope);
+	REG_SET(reg->start_slope_cntl_g, 0,
+			field_region_linear_slope, params->corner_points[0].green.custom_float_slope);
+	REG_SET(reg->start_slope_cntl_r, 0,
+			field_region_linear_slope, params->corner_points[0].red.custom_float_slope);
+
+	REG_SET(reg->start_end_cntl1_b, 0,
+			field_region_end, params->corner_points[1].blue.custom_float_x);
+	REG_SET_2(reg->start_end_cntl2_b, 0,
+			field_region_end_slope, params->corner_points[1].blue.custom_float_slope,
+			field_region_end_base, params->corner_points[1].blue.custom_float_y);
+
+	REG_SET(reg->start_end_cntl1_g, 0,
+			field_region_end, params->corner_points[1].green.custom_float_x);
+	REG_SET_2(reg->start_end_cntl2_g, 0,
+			field_region_end_slope, params->corner_points[1].green.custom_float_slope,
+		field_region_end_base, params->corner_points[1].green.custom_float_y);
+
+	REG_SET(reg->start_end_cntl1_r, 0,
+			field_region_end, params->corner_points[1].red.custom_float_x);
+	REG_SET_2(reg->start_end_cntl2_r, 0,
+			field_region_end_slope, params->corner_points[1].red.custom_float_slope,
+		field_region_end_base, params->corner_points[1].red.custom_float_y);
+
+	for (reg_region_cur = reg->region_start;
+			reg_region_cur <= reg->region_end;
+			reg_region_cur++) {
+
+		const struct gamma_curve *curve0 = &(params->arr_curve_points[2 * i]);
+		const struct gamma_curve *curve1 = &(params->arr_curve_points[(2 * i) + 1]);
+
+		REG_SET_4(reg_region_cur, 0,
+				exp_region0_lut_offset, curve0->offset,
+				exp_region0_num_segments, curve0->segments_num,
+				exp_region1_lut_offset, curve1->offset,
+				exp_region1_num_segments, curve1->segments_num);
+
+		i++;
+	}
+
+}
+
+
+
+bool cm_helper_convert_to_custom_float(
+		struct pwl_result_data *rgb_resulted,
+		struct curve_points3 *corner_points,
+		uint32_t hw_points_num,
+		bool fixpoint)
+{
+	struct custom_float_format fmt;
+
+	struct pwl_result_data *rgb = rgb_resulted;
+
+	uint32_t i = 0;
+
+	fmt.exponenta_bits = 6;
+	fmt.mantissa_bits = 12;
+	fmt.sign = false;
+
+	/* corner_points[0] - beginning base, slope offset for R,G,B
+	 * corner_points[1] - end base, slope offset for R,G,B
+	 */
+	if (!convert_to_custom_float_format(corner_points[0].red.x, &fmt,
+				&corner_points[0].red.custom_float_x)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+	if (!convert_to_custom_float_format(corner_points[0].green.x, &fmt,
+				&corner_points[0].green.custom_float_x)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+	if (!convert_to_custom_float_format(corner_points[0].blue.x, &fmt,
+				&corner_points[0].blue.custom_float_x)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+
+	if (!convert_to_custom_float_format(corner_points[0].red.offset, &fmt,
+				&corner_points[0].red.custom_float_offset)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+	if (!convert_to_custom_float_format(corner_points[0].green.offset, &fmt,
+				&corner_points[0].green.custom_float_offset)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+	if (!convert_to_custom_float_format(corner_points[0].blue.offset, &fmt,
+				&corner_points[0].blue.custom_float_offset)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+
+	if (!convert_to_custom_float_format(corner_points[0].red.slope, &fmt,
+				&corner_points[0].red.custom_float_slope)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+	if (!convert_to_custom_float_format(corner_points[0].green.slope, &fmt,
+				&corner_points[0].green.custom_float_slope)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+	if (!convert_to_custom_float_format(corner_points[0].blue.slope, &fmt,
+				&corner_points[0].blue.custom_float_slope)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+
+	fmt.mantissa_bits = 10;
+	fmt.sign = false;
+
+	if (!convert_to_custom_float_format(corner_points[1].red.x, &fmt,
+				&corner_points[1].red.custom_float_x)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+	if (!convert_to_custom_float_format(corner_points[1].green.x, &fmt,
+				&corner_points[1].green.custom_float_x)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+	if (!convert_to_custom_float_format(corner_points[1].blue.x, &fmt,
+				&corner_points[1].blue.custom_float_x)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+
+	if (fixpoint == true) {
+		corner_points[1].red.custom_float_y =
+				dc_fixpt_clamp_u0d14(corner_points[1].red.y);
+		corner_points[1].green.custom_float_y =
+				dc_fixpt_clamp_u0d14(corner_points[1].green.y);
+		corner_points[1].blue.custom_float_y =
+				dc_fixpt_clamp_u0d14(corner_points[1].blue.y);
+	} else {
+		if (!convert_to_custom_float_format(corner_points[1].red.y,
+				&fmt, &corner_points[1].red.custom_float_y)) {
+			BREAK_TO_DEBUGGER();
+			return false;
+		}
+		if (!convert_to_custom_float_format(corner_points[1].green.y,
+				&fmt, &corner_points[1].green.custom_float_y)) {
+			BREAK_TO_DEBUGGER();
+			return false;
+		}
+		if (!convert_to_custom_float_format(corner_points[1].blue.y,
+				&fmt, &corner_points[1].blue.custom_float_y)) {
+			BREAK_TO_DEBUGGER();
+			return false;
+		}
+	}
+
+	if (!convert_to_custom_float_format(corner_points[1].red.slope, &fmt,
+				&corner_points[1].red.custom_float_slope)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+	if (!convert_to_custom_float_format(corner_points[1].green.slope, &fmt,
+				&corner_points[1].green.custom_float_slope)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+	if (!convert_to_custom_float_format(corner_points[1].blue.slope, &fmt,
+				&corner_points[1].blue.custom_float_slope)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+
+	if (hw_points_num == 0 || rgb_resulted == NULL || fixpoint == true)
+		return true;
+
+	fmt.mantissa_bits = 12;
+	fmt.sign = true;
+
+	while (i != hw_points_num) {
+		if (!convert_to_custom_float_format(rgb->red, &fmt,
+						    &rgb->red_reg)) {
+			BREAK_TO_DEBUGGER();
+			return false;
+		}
+
+		if (!convert_to_custom_float_format(rgb->green, &fmt,
+						    &rgb->green_reg)) {
+			BREAK_TO_DEBUGGER();
+			return false;
+		}
+
+		if (!convert_to_custom_float_format(rgb->blue, &fmt,
+						    &rgb->blue_reg)) {
+			BREAK_TO_DEBUGGER();
+			return false;
+		}
+
+		if (!convert_to_custom_float_format(rgb->delta_red, &fmt,
+						    &rgb->delta_red_reg)) {
+			BREAK_TO_DEBUGGER();
+			return false;
+		}
+
+		if (!convert_to_custom_float_format(rgb->delta_green, &fmt,
+						    &rgb->delta_green_reg)) {
+			BREAK_TO_DEBUGGER();
+			return false;
+		}
+
+		if (!convert_to_custom_float_format(rgb->delta_blue, &fmt,
+						    &rgb->delta_blue_reg)) {
+			BREAK_TO_DEBUGGER();
+			return false;
+		}
+
+		++rgb;
+		++i;
+	}
+
+	return true;
+}
+
+/* driver uses 32 regions or less, but DCN HW has 34, extra 2 are set to 0 */
+#define MAX_REGIONS_NUMBER 34
+#define MAX_LOW_POINT      25
+#define NUMBER_REGIONS     32
+#define NUMBER_SW_SEGMENTS 16
+
+#define DC_LOGGER \
+		ctx->logger
+
+bool cm_helper_translate_curve_to_hw_format(struct dc_context *ctx,
+				const struct dc_transfer_func *output_tf,
+				struct pwl_params *lut_params, bool fixpoint)
+{
+	struct curve_points3 *corner_points;
+	struct pwl_result_data *rgb_resulted;
+	struct pwl_result_data *rgb;
+	struct pwl_result_data *rgb_plus_1;
+	struct pwl_result_data *rgb_minus_1;
+
+	int32_t region_start, region_end;
+	int32_t i;
+	uint32_t j, k, seg_distr[MAX_REGIONS_NUMBER], increment, start_index, hw_points;
+
+	if (output_tf == NULL || lut_params == NULL || output_tf->type == TF_TYPE_BYPASS)
+		return false;
+
+	corner_points = lut_params->corner_points;
+	rgb_resulted = lut_params->rgb_resulted;
+	hw_points = 0;
+
+	memset(lut_params, 0, sizeof(struct pwl_params));
+	memset(seg_distr, 0, sizeof(seg_distr));
+
+	if (output_tf->tf == TRANSFER_FUNCTION_PQ || output_tf->tf == TRANSFER_FUNCTION_GAMMA22) {
+		/* 32 segments
+		 * segments are from 2^-25 to 2^7
+		 */
+		for (i = 0; i < NUMBER_REGIONS ; i++)
+			seg_distr[i] = 3;
+
+		region_start = -MAX_LOW_POINT;
+		region_end   = NUMBER_REGIONS - MAX_LOW_POINT;
+	} else {
+		/* 13 segments
+		 * segment is from 2^-12 to 2^0
+		 * There are less than 256 points, for optimization
+		 */
+		const uint8_t SEG_COUNT = 12;
+
+		for (i = 0; i < SEG_COUNT; i++)
+			seg_distr[i] = 4;
+
+		seg_distr[SEG_COUNT] = 1;
+
+		region_start = -SEG_COUNT;
+		region_end = 1;
+	}
+
+	for (i = region_end - region_start; i < MAX_REGIONS_NUMBER ; i++)
+		seg_distr[i] = -1;
+
+	for (k = 0; k < MAX_REGIONS_NUMBER; k++) {
+		if (seg_distr[k] != -1)
+			hw_points += (1 << seg_distr[k]);
+	}
+
+	j = 0;
+	for (k = 0; k < (region_end - region_start); k++) {
+		increment = NUMBER_SW_SEGMENTS / (1 << seg_distr[k]);
+		start_index = (region_start + k + MAX_LOW_POINT) *
+				NUMBER_SW_SEGMENTS;
+		for (i = start_index; i < start_index + NUMBER_SW_SEGMENTS;
+				i += increment) {
+			if (j == hw_points - 1)
+				break;
+			if (i >= TRANSFER_FUNC_POINTS) {
+				DC_LOG_ERROR("Index out of bounds: i=%d, TRANSFER_FUNC_POINTS=%d\n",
+					     i, TRANSFER_FUNC_POINTS);
+				return false;
+			}
+			rgb_resulted[j].red = output_tf->tf_pts.red[i];
+			rgb_resulted[j].green = output_tf->tf_pts.green[i];
+			rgb_resulted[j].blue = output_tf->tf_pts.blue[i];
+			j++;
+		}
+	}
+
+	/* last point */
+	start_index = (region_end + MAX_LOW_POINT) * NUMBER_SW_SEGMENTS;
+	rgb_resulted[hw_points - 1].red = output_tf->tf_pts.red[start_index];
+	rgb_resulted[hw_points - 1].green = output_tf->tf_pts.green[start_index];
+	rgb_resulted[hw_points - 1].blue = output_tf->tf_pts.blue[start_index];
+
+	rgb_resulted[hw_points].red = rgb_resulted[hw_points - 1].red;
+	rgb_resulted[hw_points].green = rgb_resulted[hw_points - 1].green;
+	rgb_resulted[hw_points].blue = rgb_resulted[hw_points - 1].blue;
+
+	// All 3 color channels have same x
+	corner_points[0].red.x = dc_fixpt_pow(dc_fixpt_from_int(2),
+					     dc_fixpt_from_int(region_start));
+	corner_points[0].green.x = corner_points[0].red.x;
+	corner_points[0].blue.x = corner_points[0].red.x;
+
+	corner_points[1].red.x = dc_fixpt_pow(dc_fixpt_from_int(2),
+					     dc_fixpt_from_int(region_end));
+	corner_points[1].green.x = corner_points[1].red.x;
+	corner_points[1].blue.x = corner_points[1].red.x;
+
+	corner_points[0].red.y = rgb_resulted[0].red;
+	corner_points[0].green.y = rgb_resulted[0].green;
+	corner_points[0].blue.y = rgb_resulted[0].blue;
+
+	corner_points[0].red.slope = dc_fixpt_div(corner_points[0].red.y,
+			corner_points[0].red.x);
+	corner_points[0].green.slope = dc_fixpt_div(corner_points[0].green.y,
+			corner_points[0].green.x);
+	corner_points[0].blue.slope = dc_fixpt_div(corner_points[0].blue.y,
+			corner_points[0].blue.x);
+
+	/* see comment above, m_arrPoints[1].y should be the Y value for the
+	 * region end (m_numOfHwPoints), not last HW point(m_numOfHwPoints - 1)
+	 */
+	corner_points[1].red.y = rgb_resulted[hw_points - 1].red;
+	corner_points[1].green.y = rgb_resulted[hw_points - 1].green;
+	corner_points[1].blue.y = rgb_resulted[hw_points - 1].blue;
+	corner_points[1].red.slope = dc_fixpt_zero;
+	corner_points[1].green.slope = dc_fixpt_zero;
+	corner_points[1].blue.slope = dc_fixpt_zero;
+
+	if (output_tf->tf == TRANSFER_FUNCTION_PQ) {
+		/* for PQ, we want to have a straight line from last HW X point,
+		 * and the slope to be such that we hit 1.0 at 10000 nits.
+		 */
+		const struct fixed31_32 end_value =
+				dc_fixpt_from_int(125);
+
+		corner_points[1].red.slope = dc_fixpt_div(
+			dc_fixpt_sub(dc_fixpt_one, corner_points[1].red.y),
+			dc_fixpt_sub(end_value, corner_points[1].red.x));
+		corner_points[1].green.slope = dc_fixpt_div(
+			dc_fixpt_sub(dc_fixpt_one, corner_points[1].green.y),
+			dc_fixpt_sub(end_value, corner_points[1].green.x));
+		corner_points[1].blue.slope = dc_fixpt_div(
+			dc_fixpt_sub(dc_fixpt_one, corner_points[1].blue.y),
+			dc_fixpt_sub(end_value, corner_points[1].blue.x));
+	}
+
+	lut_params->hw_points_num = hw_points;
+
+	k = 0;
+	for (i = 1; i < MAX_REGIONS_NUMBER; i++) {
+		if (seg_distr[k] != -1) {
+			lut_params->arr_curve_points[k].segments_num =
+					seg_distr[k];
+			lut_params->arr_curve_points[i].offset =
+					lut_params->arr_curve_points[k].offset + (1 << seg_distr[k]);
+		}
+		k++;
+	}
+
+	if (seg_distr[k] != -1)
+		lut_params->arr_curve_points[k].segments_num = seg_distr[k];
+
+	rgb = rgb_resulted;
+	rgb_plus_1 = rgb_resulted + 1;
+	rgb_minus_1 = rgb;
+
+	i = 1;
+	while (i != hw_points + 1) {
+
+		if (i >= hw_points - 1) {
+			if (dc_fixpt_lt(rgb_plus_1->red, rgb->red))
+				rgb_plus_1->red = dc_fixpt_add(rgb->red, rgb_minus_1->delta_red);
+			if (dc_fixpt_lt(rgb_plus_1->green, rgb->green))
+				rgb_plus_1->green = dc_fixpt_add(rgb->green, rgb_minus_1->delta_green);
+			if (dc_fixpt_lt(rgb_plus_1->blue, rgb->blue))
+				rgb_plus_1->blue = dc_fixpt_add(rgb->blue, rgb_minus_1->delta_blue);
+		}
+
+		rgb->delta_red   = dc_fixpt_sub(rgb_plus_1->red,   rgb->red);
+		rgb->delta_green = dc_fixpt_sub(rgb_plus_1->green, rgb->green);
+		rgb->delta_blue  = dc_fixpt_sub(rgb_plus_1->blue,  rgb->blue);
+
+
+		if (fixpoint == true) {
+			uint32_t red_clamp = dc_fixpt_clamp_u0d14(rgb->delta_red);
+			uint32_t green_clamp = dc_fixpt_clamp_u0d14(rgb->delta_green);
+			uint32_t blue_clamp = dc_fixpt_clamp_u0d14(rgb->delta_blue);
+
+			if (red_clamp >> 10 || green_clamp >> 10 || blue_clamp >> 10)
+				DC_LOG_WARNING("Losing delta precision while programming shaper LUT.");
+
+			rgb->delta_red_reg   = red_clamp & 0x3ff;
+			rgb->delta_green_reg = green_clamp & 0x3ff;
+			rgb->delta_blue_reg  = blue_clamp & 0x3ff;
+			rgb->red_reg         = dc_fixpt_clamp_u0d14(rgb->red);
+			rgb->green_reg       = dc_fixpt_clamp_u0d14(rgb->green);
+			rgb->blue_reg        = dc_fixpt_clamp_u0d14(rgb->blue);
+		}
+
+		++rgb_plus_1;
+		rgb_minus_1 = rgb;
+		++rgb;
+		++i;
+	}
+	cm_helper_convert_to_custom_float(rgb_resulted,
+						lut_params->corner_points,
+						hw_points, fixpoint);
+
+	return true;
+}
+
+#define NUM_DEGAMMA_REGIONS    12
+
+
+bool cm_helper_translate_curve_to_degamma_hw_format(
+				const struct dc_transfer_func *output_tf,
+				struct pwl_params *lut_params)
+{
+	struct curve_points3 *corner_points;
+	struct pwl_result_data *rgb_resulted;
+	struct pwl_result_data *rgb;
+	struct pwl_result_data *rgb_plus_1;
+
+	int32_t region_start, region_end;
+	int32_t i;
+	uint32_t j, k, seg_distr[MAX_REGIONS_NUMBER], increment, start_index, hw_points;
+
+	if (output_tf == NULL || lut_params == NULL || output_tf->type == TF_TYPE_BYPASS)
+		return false;
+
+	corner_points = lut_params->corner_points;
+	rgb_resulted = lut_params->rgb_resulted;
+	hw_points = 0;
+
+	memset(lut_params, 0, sizeof(struct pwl_params));
+	memset(seg_distr, 0, sizeof(seg_distr));
+
+	region_start = -NUM_DEGAMMA_REGIONS;
+	region_end   = 0;
+
+
+	for (i = region_end - region_start; i < MAX_REGIONS_NUMBER ; i++)
+		seg_distr[i] = -1;
+	/* 12 segments
+	 * segments are from 2^-12 to 0
+	 */
+	for (i = 0; i < NUM_DEGAMMA_REGIONS ; i++)
+		seg_distr[i] = 4;
+
+	for (k = 0; k < MAX_REGIONS_NUMBER; k++) {
+		if (seg_distr[k] != -1)
+			hw_points += (1 << seg_distr[k]);
+	}
+
+	j = 0;
+	for (k = 0; k < (region_end - region_start); k++) {
+		increment = NUMBER_SW_SEGMENTS / (1 << seg_distr[k]);
+		start_index = (region_start + k + MAX_LOW_POINT) *
+				NUMBER_SW_SEGMENTS;
+		for (i = start_index; i < start_index + NUMBER_SW_SEGMENTS;
+				i += increment) {
+			if (j == hw_points - 1)
+				break;
+			if (i >= TRANSFER_FUNC_POINTS)
+				return false;
+			rgb_resulted[j].red = output_tf->tf_pts.red[i];
+			rgb_resulted[j].green = output_tf->tf_pts.green[i];
+			rgb_resulted[j].blue = output_tf->tf_pts.blue[i];
+			j++;
+		}
+	}
+
+	/* last point */
+	start_index = (region_end + MAX_LOW_POINT) * NUMBER_SW_SEGMENTS;
+	rgb_resulted[hw_points - 1].red = output_tf->tf_pts.red[start_index];
+	rgb_resulted[hw_points - 1].green = output_tf->tf_pts.green[start_index];
+	rgb_resulted[hw_points - 1].blue = output_tf->tf_pts.blue[start_index];
+
+	rgb_resulted[hw_points].red = rgb_resulted[hw_points - 1].red;
+	rgb_resulted[hw_points].green = rgb_resulted[hw_points - 1].green;
+	rgb_resulted[hw_points].blue = rgb_resulted[hw_points - 1].blue;
+
+	corner_points[0].red.x = dc_fixpt_pow(dc_fixpt_from_int(2),
+					     dc_fixpt_from_int(region_start));
+	corner_points[0].green.x = corner_points[0].red.x;
+	corner_points[0].blue.x = corner_points[0].red.x;
+	corner_points[1].red.x = dc_fixpt_pow(dc_fixpt_from_int(2),
+					     dc_fixpt_from_int(region_end));
+	corner_points[1].green.x = corner_points[1].red.x;
+	corner_points[1].blue.x = corner_points[1].red.x;
+
+	corner_points[0].red.y = rgb_resulted[0].red;
+	corner_points[0].green.y = rgb_resulted[0].green;
+	corner_points[0].blue.y = rgb_resulted[0].blue;
+
+	/* see comment above, m_arrPoints[1].y should be the Y value for the
+	 * region end (m_numOfHwPoints), not last HW point(m_numOfHwPoints - 1)
+	 */
+	corner_points[1].red.y = rgb_resulted[hw_points - 1].red;
+	corner_points[1].green.y = rgb_resulted[hw_points - 1].green;
+	corner_points[1].blue.y = rgb_resulted[hw_points - 1].blue;
+	corner_points[1].red.slope = dc_fixpt_zero;
+	corner_points[1].green.slope = dc_fixpt_zero;
+	corner_points[1].blue.slope = dc_fixpt_zero;
+
+	if (output_tf->tf == TRANSFER_FUNCTION_PQ) {
+		/* for PQ, we want to have a straight line from last HW X point,
+		 * and the slope to be such that we hit 1.0 at 10000 nits.
+		 */
+		const struct fixed31_32 end_value =
+				dc_fixpt_from_int(125);
+
+		corner_points[1].red.slope = dc_fixpt_div(
+			dc_fixpt_sub(dc_fixpt_one, corner_points[1].red.y),
+			dc_fixpt_sub(end_value, corner_points[1].red.x));
+		corner_points[1].green.slope = dc_fixpt_div(
+			dc_fixpt_sub(dc_fixpt_one, corner_points[1].green.y),
+			dc_fixpt_sub(end_value, corner_points[1].green.x));
+		corner_points[1].blue.slope = dc_fixpt_div(
+			dc_fixpt_sub(dc_fixpt_one, corner_points[1].blue.y),
+			dc_fixpt_sub(end_value, corner_points[1].blue.x));
+	}
+
+	lut_params->hw_points_num = hw_points;
+
+	k = 0;
+	for (i = 1; i < MAX_REGIONS_NUMBER; i++) {
+		if (seg_distr[k] != -1) {
+			lut_params->arr_curve_points[k].segments_num =
+					seg_distr[k];
+			lut_params->arr_curve_points[i].offset =
+					lut_params->arr_curve_points[k].offset + (1 << seg_distr[k]);
+		}
+		k++;
+	}
+
+	if (seg_distr[k] != -1)
+		lut_params->arr_curve_points[k].segments_num = seg_distr[k];
+
+	rgb = rgb_resulted;
+	rgb_plus_1 = rgb_resulted + 1;
+
+	i = 1;
+	while (i != hw_points + 1) {
+		rgb->delta_red   = dc_fixpt_sub(rgb_plus_1->red,   rgb->red);
+		rgb->delta_green = dc_fixpt_sub(rgb_plus_1->green, rgb->green);
+		rgb->delta_blue  = dc_fixpt_sub(rgb_plus_1->blue,  rgb->blue);
+
+		++rgb_plus_1;
+		++rgb;
+		++i;
+	}
+	cm_helper_convert_to_custom_float(rgb_resulted,
+						lut_params->corner_points,
+						hw_points, false);
+
+	return true;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_cm_common.h b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_cm_common.h
new file mode 100644
index 000000000000..decc50b1ac53
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_cm_common.h
@@ -0,0 +1,120 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_DCN10_CM_COMMON_H__
+#define __DAL_DCN10_CM_COMMON_H__
+
+#define TF_HELPER_REG_FIELD_LIST(type) \
+	type exp_region0_lut_offset; \
+	type exp_region0_num_segments; \
+	type exp_region1_lut_offset; \
+	type exp_region1_num_segments;\
+	type field_region_end;\
+	type field_region_end_slope;\
+	type field_region_end_base;\
+	type exp_region_start;\
+	type exp_resion_start_segment;\
+	type field_region_linear_slope
+
+#define TF_HELPER_REG_LIST \
+	uint32_t start_cntl_b; \
+	uint32_t start_cntl_g; \
+	uint32_t start_cntl_r; \
+	uint32_t start_slope_cntl_b; \
+	uint32_t start_slope_cntl_g; \
+	uint32_t start_slope_cntl_r; \
+	uint32_t start_end_cntl1_b; \
+	uint32_t start_end_cntl2_b; \
+	uint32_t start_end_cntl1_g; \
+	uint32_t start_end_cntl2_g; \
+	uint32_t start_end_cntl1_r; \
+	uint32_t start_end_cntl2_r; \
+	uint32_t region_start; \
+	uint32_t region_end
+
+#define TF_CM_REG_FIELD_LIST(type) \
+	type csc_c11; \
+	type csc_c12
+
+struct xfer_func_shift {
+	TF_HELPER_REG_FIELD_LIST(uint8_t);
+};
+
+struct xfer_func_mask {
+	TF_HELPER_REG_FIELD_LIST(uint32_t);
+};
+
+struct xfer_func_reg {
+	struct xfer_func_shift shifts;
+	struct xfer_func_mask masks;
+
+	TF_HELPER_REG_LIST;
+};
+
+struct cm_color_matrix_shift {
+	TF_CM_REG_FIELD_LIST(uint8_t);
+};
+
+struct cm_color_matrix_mask {
+	TF_CM_REG_FIELD_LIST(uint32_t);
+};
+
+struct color_matrices_reg{
+	struct cm_color_matrix_shift shifts;
+	struct cm_color_matrix_mask masks;
+
+	uint32_t csc_c11_c12;
+	uint32_t csc_c33_c34;
+};
+
+void cm_helper_program_color_matrices(
+		struct dc_context *ctx,
+		const uint16_t *regval,
+		const struct color_matrices_reg *reg);
+
+void cm_helper_program_xfer_func(
+		struct dc_context *ctx,
+		const struct pwl_params *params,
+		const struct xfer_func_reg *reg);
+
+bool cm_helper_convert_to_custom_float(
+		struct pwl_result_data *rgb_resulted,
+		struct curve_points3 *corner_points,
+		uint32_t hw_points_num,
+		bool fixpoint);
+
+bool cm_helper_translate_curve_to_hw_format(
+		struct dc_context *ctx,
+		const struct dc_transfer_func *output_tf,
+		struct pwl_params *lut_params, bool fixpoint);
+
+bool cm_helper_translate_curve_to_degamma_hw_format(
+				const struct dc_transfer_func *output_tf,
+				struct pwl_params *lut_params);
+
+void cm_helper_read_color_matrices(struct dc_context *ctx,
+				   uint16_t *regval,
+				   const struct color_matrices_reg *reg);
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dwb.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dwb.c
new file mode 100644
index 000000000000..365a3215f6d5
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dwb.c
@@ -0,0 +1,129 @@
+/*
+ * Copyright 2012-17 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+#include "resource.h"
+#include "dwb.h"
+#include "dcn10_dwb.h"
+
+
+#define REG(reg)\
+	dwbc10->dwbc_regs->reg
+
+#define CTX \
+	dwbc10->base.ctx
+
+#undef FN
+#define FN(reg_name, field_name) \
+	dwbc10->dwbc_shift->field_name, dwbc10->dwbc_mask->field_name
+
+#define TO_DCN10_DWBC(dwbc_base) \
+	container_of(dwbc_base, struct dcn10_dwbc, base)
+
+static bool dwb1_get_caps(struct dwbc *dwbc, struct dwb_caps *caps)
+{
+	if (caps) {
+		caps->adapter_id = 0;	/* we only support 1 adapter currently */
+		caps->hw_version = DCN_VERSION_1_0;
+		caps->num_pipes = 2;
+		memset(&caps->reserved, 0, sizeof(caps->reserved));
+		memset(&caps->reserved2, 0, sizeof(caps->reserved2));
+		caps->sw_version = dwb_ver_1_0;
+		caps->caps.support_dwb = true;
+		caps->caps.support_ogam = false;
+		caps->caps.support_wbscl = true;
+		caps->caps.support_ocsc = false;
+		return true;
+	} else {
+		return false;
+	}
+}
+
+static bool dwb1_enable(struct dwbc *dwbc, struct dc_dwb_params *params)
+{
+	struct dcn10_dwbc *dwbc10 = TO_DCN10_DWBC(dwbc);
+
+	/* disable first. */
+	dwbc->funcs->disable(dwbc);
+
+	/* disable power gating */
+	REG_UPDATE_5(WB_EC_CONFIG, DISPCLK_R_WB_GATE_DIS, 1,
+		 DISPCLK_G_WB_GATE_DIS, 1, DISPCLK_G_WBSCL_GATE_DIS, 1,
+		 WB_LB_LS_DIS, 1, WB_LUT_LS_DIS, 1);
+
+	REG_UPDATE(WB_ENABLE, WB_ENABLE, 1);
+
+	return true;
+}
+
+static bool dwb1_disable(struct dwbc *dwbc)
+{
+	struct dcn10_dwbc *dwbc10 = TO_DCN10_DWBC(dwbc);
+
+	/* disable CNV */
+	REG_UPDATE(CNV_MODE, CNV_FRAME_CAPTURE_EN, 0);
+
+	/* disable WB */
+	REG_UPDATE(WB_ENABLE, WB_ENABLE, 0);
+
+	/* soft reset */
+	REG_UPDATE(WB_SOFT_RESET, WB_SOFT_RESET, 1);
+	REG_UPDATE(WB_SOFT_RESET, WB_SOFT_RESET, 0);
+
+	/* enable power gating */
+	REG_UPDATE_5(WB_EC_CONFIG, DISPCLK_R_WB_GATE_DIS, 0,
+		 DISPCLK_G_WB_GATE_DIS, 0, DISPCLK_G_WBSCL_GATE_DIS, 0,
+		 WB_LB_LS_DIS, 0, WB_LUT_LS_DIS, 0);
+
+	return true;
+}
+
+const struct dwbc_funcs dcn10_dwbc_funcs = {
+	.get_caps			= dwb1_get_caps,
+	.enable				= dwb1_enable,
+	.disable			= dwb1_disable,
+	.update				= NULL,
+	.set_stereo			= NULL,
+	.set_new_content		= NULL,
+	.set_warmup			= NULL,
+	.dwb_set_scaler			= NULL,
+};
+
+void dcn10_dwbc_construct(struct dcn10_dwbc *dwbc10,
+		struct dc_context *ctx,
+		const struct dcn10_dwbc_registers *dwbc_regs,
+		const struct dcn10_dwbc_shift *dwbc_shift,
+		const struct dcn10_dwbc_mask *dwbc_mask,
+		int inst)
+{
+	dwbc10->base.ctx = ctx;
+
+	dwbc10->base.inst = inst;
+	dwbc10->base.funcs = &dcn10_dwbc_funcs;
+
+	dwbc10->dwbc_regs = dwbc_regs;
+	dwbc10->dwbc_shift = dwbc_shift;
+	dwbc10->dwbc_mask = dwbc_mask;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dwb.h b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dwb.h
new file mode 100644
index 000000000000..5268c46ae907
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dwb.h
@@ -0,0 +1,267 @@
+/* Copyright 2012-17 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+#ifndef __DC_DWBC_DCN10_H__
+#define __DC_DWBC_DCN10_H__
+
+/* DCN */
+#define BASE_INNER(seg) \
+	DCE_BASE__INST0_SEG ## seg
+
+#define BASE(seg) \
+	BASE_INNER(seg)
+
+#define SR(reg_name)\
+		.reg_name = BASE(mm ## reg_name ## _BASE_IDX) +  \
+					mm ## reg_name
+
+#define SRI(reg_name, block, id)\
+	.reg_name = BASE(mm ## block ## id ## _ ## reg_name ## _BASE_IDX) + \
+					mm ## block ## id ## _ ## reg_name
+
+
+#define SRII(reg_name, block, id)\
+	.reg_name[id] = BASE(mm ## block ## id ## _ ## reg_name ## _BASE_IDX) + \
+					mm ## block ## id ## _ ## reg_name
+
+#define SF(reg_name, field_name, post_fix)\
+	.field_name = reg_name ## __ ## field_name ## post_fix
+
+
+#define DWBC_COMMON_REG_LIST_DCN1_0(inst) \
+	SRI(WB_ENABLE, CNV, inst),\
+	SRI(WB_EC_CONFIG, CNV, inst),\
+	SRI(CNV_MODE, CNV, inst),\
+	SRI(WB_SOFT_RESET, CNV, inst),\
+	SRI(MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_PITCH, MCIF_WB, inst),\
+	SRI(MCIF_WB_ARBITRATION_CONTROL, MCIF_WB, inst),\
+	SRI(MCIF_WB_SCLK_CHANGE, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_1_ADDR_Y, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_1_ADDR_Y_OFFSET, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_1_ADDR_C, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_1_ADDR_C_OFFSET, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_2_ADDR_Y, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_2_ADDR_Y_OFFSET, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_2_ADDR_C, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_2_ADDR_C_OFFSET, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_3_ADDR_Y, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_3_ADDR_Y_OFFSET, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_3_ADDR_C, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_3_ADDR_C_OFFSET, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_4_ADDR_Y, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_4_ADDR_Y_OFFSET, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_4_ADDR_C, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_4_ADDR_C_OFFSET, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUFMGR_VCE_CONTROL, MCIF_WB, inst),\
+	SRI(MCIF_WB_NB_PSTATE_LATENCY_WATERMARK, MCIF_WB, inst),\
+	SRI(MCIF_WB_NB_PSTATE_CONTROL, MCIF_WB, inst),\
+	SRI(MCIF_WB_WATERMARK, MCIF_WB, inst),\
+	SRI(MCIF_WB_WARM_UP_CNTL, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_LUMA_SIZE, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_CHROMA_SIZE, MCIF_WB, inst)
+
+#define DWBC_COMMON_MASK_SH_LIST_DCN1_0(mask_sh) \
+	SF(CNV0_WB_ENABLE, WB_ENABLE, mask_sh),\
+	SF(CNV0_WB_EC_CONFIG, DISPCLK_R_WB_GATE_DIS, mask_sh),\
+	SF(CNV0_WB_EC_CONFIG, DISPCLK_G_WB_GATE_DIS, mask_sh),\
+	SF(CNV0_WB_EC_CONFIG, DISPCLK_G_WBSCL_GATE_DIS, mask_sh),\
+	SF(CNV0_WB_EC_CONFIG, WB_LB_LS_DIS, mask_sh),\
+	SF(CNV0_WB_EC_CONFIG, WB_LUT_LS_DIS, mask_sh),\
+	SF(CNV0_CNV_MODE, CNV_WINDOW_CROP_EN, mask_sh),\
+	SF(CNV0_CNV_MODE, CNV_STEREO_TYPE, mask_sh),\
+	SF(CNV0_CNV_MODE, CNV_INTERLACED_MODE, mask_sh),\
+	SF(CNV0_CNV_MODE, CNV_EYE_SELECTION, mask_sh),\
+	SF(CNV0_CNV_MODE, CNV_STEREO_POLARITY, mask_sh),\
+	SF(CNV0_CNV_MODE, CNV_INTERLACED_FIELD_ORDER, mask_sh),\
+	SF(CNV0_CNV_MODE, CNV_STEREO_SPLIT, mask_sh),\
+	SF(CNV0_CNV_MODE, CNV_NEW_CONTENT, mask_sh),\
+	SF(CNV0_CNV_MODE, CNV_FRAME_CAPTURE_EN, mask_sh),\
+	SF(CNV0_WB_SOFT_RESET, WB_SOFT_RESET, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUFMGR_ENABLE, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUF_DUALSIZE_REQ, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUFMGR_SW_INT_EN, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUFMGR_SW_INT_ACK, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUFMGR_SW_SLICE_INT_EN, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUFMGR_SW_OVERRUN_INT_EN, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUFMGR_SW_LOCK, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_P_VMID, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUF_ADDR_FENCE_EN, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_PITCH, MCIF_WB_BUF_LUMA_PITCH, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_PITCH, MCIF_WB_BUF_CHROMA_PITCH, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_ARBITRATION_CONTROL, MCIF_WB_CLIENT_ARBITRATION_SLICE, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_ARBITRATION_CONTROL, MCIF_WB_TIME_PER_PIXEL, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_SCLK_CHANGE, WM_CHANGE_ACK_FORCE_ON, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_SCLK_CHANGE, MCIF_WB_CLI_WATERMARK_MASK, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_ADDR_Y, MCIF_WB_BUF_1_ADDR_Y, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_ADDR_Y_OFFSET, MCIF_WB_BUF_1_ADDR_Y_OFFSET, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_ADDR_C, MCIF_WB_BUF_1_ADDR_C, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_ADDR_C_OFFSET, MCIF_WB_BUF_1_ADDR_C_OFFSET, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_ADDR_Y, MCIF_WB_BUF_2_ADDR_Y, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_ADDR_Y_OFFSET, MCIF_WB_BUF_2_ADDR_Y_OFFSET, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_ADDR_C, MCIF_WB_BUF_2_ADDR_C, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_ADDR_C_OFFSET, MCIF_WB_BUF_2_ADDR_C_OFFSET, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_ADDR_Y, MCIF_WB_BUF_3_ADDR_Y, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_ADDR_Y_OFFSET, MCIF_WB_BUF_3_ADDR_Y_OFFSET, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_ADDR_C, MCIF_WB_BUF_3_ADDR_C, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_ADDR_C_OFFSET, MCIF_WB_BUF_3_ADDR_C_OFFSET, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_ADDR_Y, MCIF_WB_BUF_4_ADDR_Y, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_ADDR_Y_OFFSET, MCIF_WB_BUF_4_ADDR_Y_OFFSET, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_ADDR_C, MCIF_WB_BUF_4_ADDR_C, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_ADDR_C_OFFSET, MCIF_WB_BUF_4_ADDR_C_OFFSET, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_VCE_CONTROL, MCIF_WB_BUFMGR_VCE_LOCK_IGNORE, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_VCE_CONTROL, MCIF_WB_BUFMGR_VCE_INT_EN, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_VCE_CONTROL, MCIF_WB_BUFMGR_VCE_INT_ACK, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_VCE_CONTROL, MCIF_WB_BUFMGR_VCE_SLICE_INT_EN, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_VCE_CONTROL, MCIF_WB_BUFMGR_VCE_LOCK, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_VCE_CONTROL, MCIF_WB_BUFMGR_SLICE_SIZE, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_NB_PSTATE_LATENCY_WATERMARK, NB_PSTATE_CHANGE_REFRESH_WATERMARK, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_NB_PSTATE_CONTROL, NB_PSTATE_CHANGE_URGENT_DURING_REQUEST, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_NB_PSTATE_CONTROL, NB_PSTATE_CHANGE_FORCE_ON, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_NB_PSTATE_CONTROL, NB_PSTATE_ALLOW_FOR_URGENT, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_NB_PSTATE_CONTROL, NB_PSTATE_CHANGE_WATERMARK_MASK, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_WATERMARK, MCIF_WB_CLI_WATERMARK, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_WARM_UP_CNTL, MCIF_WB_PITCH_SIZE_WARMUP, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_LUMA_SIZE, MCIF_WB_BUF_LUMA_SIZE, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_CHROMA_SIZE, MCIF_WB_BUF_CHROMA_SIZE, mask_sh)
+
+#define DWBC_REG_FIELD_LIST(type) \
+	type WB_ENABLE;\
+	type DISPCLK_R_WB_GATE_DIS;\
+	type DISPCLK_G_WB_GATE_DIS;\
+	type DISPCLK_G_WBSCL_GATE_DIS;\
+	type WB_LB_LS_DIS;\
+	type WB_LB_SD_DIS;\
+	type WB_LUT_LS_DIS;\
+	type CNV_WINDOW_CROP_EN;\
+	type CNV_STEREO_TYPE;\
+	type CNV_INTERLACED_MODE;\
+	type CNV_EYE_SELECTION;\
+	type CNV_STEREO_POLARITY;\
+	type CNV_INTERLACED_FIELD_ORDER;\
+	type CNV_STEREO_SPLIT;\
+	type CNV_NEW_CONTENT;\
+	type CNV_FRAME_CAPTURE_EN;\
+	type WB_SOFT_RESET;\
+	type MCIF_WB_BUFMGR_ENABLE;\
+	type MCIF_WB_BUF_DUALSIZE_REQ;\
+	type MCIF_WB_BUFMGR_SW_INT_EN;\
+	type MCIF_WB_BUFMGR_SW_INT_ACK;\
+	type MCIF_WB_BUFMGR_SW_SLICE_INT_EN;\
+	type MCIF_WB_BUFMGR_SW_OVERRUN_INT_EN;\
+	type MCIF_WB_BUFMGR_SW_LOCK;\
+	type MCIF_WB_P_VMID;\
+	type MCIF_WB_BUF_ADDR_FENCE_EN;\
+	type MCIF_WB_BUF_LUMA_PITCH;\
+	type MCIF_WB_BUF_CHROMA_PITCH;\
+	type MCIF_WB_CLIENT_ARBITRATION_SLICE;\
+	type MCIF_WB_TIME_PER_PIXEL;\
+	type WM_CHANGE_ACK_FORCE_ON;\
+	type MCIF_WB_CLI_WATERMARK_MASK;\
+	type MCIF_WB_BUF_1_ADDR_Y;\
+	type MCIF_WB_BUF_1_ADDR_Y_OFFSET;\
+	type MCIF_WB_BUF_1_ADDR_C;\
+	type MCIF_WB_BUF_1_ADDR_C_OFFSET;\
+	type MCIF_WB_BUF_2_ADDR_Y;\
+	type MCIF_WB_BUF_2_ADDR_Y_OFFSET;\
+	type MCIF_WB_BUF_2_ADDR_C;\
+	type MCIF_WB_BUF_2_ADDR_C_OFFSET;\
+	type MCIF_WB_BUF_3_ADDR_Y;\
+	type MCIF_WB_BUF_3_ADDR_Y_OFFSET;\
+	type MCIF_WB_BUF_3_ADDR_C;\
+	type MCIF_WB_BUF_3_ADDR_C_OFFSET;\
+	type MCIF_WB_BUF_4_ADDR_Y;\
+	type MCIF_WB_BUF_4_ADDR_Y_OFFSET;\
+	type MCIF_WB_BUF_4_ADDR_C;\
+	type MCIF_WB_BUF_4_ADDR_C_OFFSET;\
+	type MCIF_WB_BUFMGR_VCE_LOCK_IGNORE;\
+	type MCIF_WB_BUFMGR_VCE_INT_EN;\
+	type MCIF_WB_BUFMGR_VCE_INT_ACK;\
+	type MCIF_WB_BUFMGR_VCE_SLICE_INT_EN;\
+	type MCIF_WB_BUFMGR_VCE_LOCK;\
+	type MCIF_WB_BUFMGR_SLICE_SIZE;\
+	type NB_PSTATE_CHANGE_REFRESH_WATERMARK;\
+	type NB_PSTATE_CHANGE_URGENT_DURING_REQUEST;\
+	type NB_PSTATE_CHANGE_FORCE_ON;\
+	type NB_PSTATE_ALLOW_FOR_URGENT;\
+	type NB_PSTATE_CHANGE_WATERMARK_MASK;\
+	type MCIF_WB_CLI_WATERMARK;\
+	type MCIF_WB_CLI_CLOCK_GATER_OVERRIDE;\
+	type MCIF_WB_PITCH_SIZE_WARMUP;\
+	type MCIF_WB_BUF_LUMA_SIZE;\
+	type MCIF_WB_BUF_CHROMA_SIZE;\
+
+struct dcn10_dwbc_registers {
+	uint32_t WB_ENABLE;
+	uint32_t WB_EC_CONFIG;
+	uint32_t CNV_MODE;
+	uint32_t WB_SOFT_RESET;
+	uint32_t MCIF_WB_BUFMGR_SW_CONTROL;
+	uint32_t MCIF_WB_BUF_PITCH;
+	uint32_t MCIF_WB_ARBITRATION_CONTROL;
+	uint32_t MCIF_WB_SCLK_CHANGE;
+	uint32_t MCIF_WB_BUF_1_ADDR_Y;
+	uint32_t MCIF_WB_BUF_1_ADDR_Y_OFFSET;
+	uint32_t MCIF_WB_BUF_1_ADDR_C;
+	uint32_t MCIF_WB_BUF_1_ADDR_C_OFFSET;
+	uint32_t MCIF_WB_BUF_2_ADDR_Y;
+	uint32_t MCIF_WB_BUF_2_ADDR_Y_OFFSET;
+	uint32_t MCIF_WB_BUF_2_ADDR_C;
+	uint32_t MCIF_WB_BUF_2_ADDR_C_OFFSET;
+	uint32_t MCIF_WB_BUF_3_ADDR_Y;
+	uint32_t MCIF_WB_BUF_3_ADDR_Y_OFFSET;
+	uint32_t MCIF_WB_BUF_3_ADDR_C;
+	uint32_t MCIF_WB_BUF_3_ADDR_C_OFFSET;
+	uint32_t MCIF_WB_BUF_4_ADDR_Y;
+	uint32_t MCIF_WB_BUF_4_ADDR_Y_OFFSET;
+	uint32_t MCIF_WB_BUF_4_ADDR_C;
+	uint32_t MCIF_WB_BUF_4_ADDR_C_OFFSET;
+	uint32_t MCIF_WB_BUFMGR_VCE_CONTROL;
+	uint32_t MCIF_WB_NB_PSTATE_LATENCY_WATERMARK;
+	uint32_t MCIF_WB_NB_PSTATE_CONTROL;
+	uint32_t MCIF_WB_WATERMARK;
+	uint32_t MCIF_WB_WARM_UP_CNTL;
+	uint32_t MCIF_WB_BUF_LUMA_SIZE;
+	uint32_t MCIF_WB_BUF_CHROMA_SIZE;
+};
+struct dcn10_dwbc_mask {
+	DWBC_REG_FIELD_LIST(uint32_t)
+};
+struct dcn10_dwbc_shift {
+	DWBC_REG_FIELD_LIST(uint8_t)
+};
+struct dcn10_dwbc {
+	struct dwbc base;
+	const struct dcn10_dwbc_registers *dwbc_regs;
+	const struct dcn10_dwbc_shift *dwbc_shift;
+	const struct dcn10_dwbc_mask *dwbc_mask;
+};
+
+void dcn10_dwbc_construct(struct dcn10_dwbc *dwbc10,
+		struct dc_context *ctx,
+		const struct dcn10_dwbc_registers *dwbc_regs,
+		const struct dcn10_dwbc_shift *dwbc_shift,
+		const struct dcn10_dwbc_mask *dwbc_mask,
+		int inst);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer_debug.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer_debug.c
new file mode 100644
index 000000000000..baf663b661c8
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer_debug.c
@@ -0,0 +1,622 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+#include "core_types.h"
+#include "resource.h"
+#include "custom_float.h"
+#include "dcn10/dcn10_hwseq.h"
+#include "dce110/dce110_hwseq.h"
+#include "dce/dce_hwseq.h"
+#include "abm.h"
+#include "dmcu.h"
+#include "dcn10/dcn10_optc.h"
+#include "dcn10/dcn10_dpp.h"
+#include "dcn10/dcn10_mpc.h"
+#include "timing_generator.h"
+#include "opp.h"
+#include "ipp.h"
+#include "mpc.h"
+#include "reg_helper.h"
+#include "dcn10/dcn10_hubp.h"
+#include "dcn10/dcn10_hubbub.h"
+#include "dcn10_cm_common.h"
+#include "clk_mgr.h"
+
+__printf(3, 4)
+unsigned int snprintf_count(char *pbuf, unsigned int bufsize, const char *fmt, ...)
+{
+	int ret_vsnprintf;
+	unsigned int chars_printed;
+
+	va_list args;
+	va_start(args, fmt);
+
+	ret_vsnprintf = vsnprintf(pbuf, bufsize, fmt, args);
+
+	va_end(args);
+
+	if (ret_vsnprintf > 0) {
+		if (ret_vsnprintf < bufsize)
+			chars_printed = ret_vsnprintf;
+		else
+			chars_printed = bufsize - 1;
+	} else
+		chars_printed = 0;
+
+	return chars_printed;
+}
+
+static unsigned int dcn10_get_hubbub_state(struct dc *dc, char *pBuf, unsigned int bufSize)
+{
+	struct dc_context *dc_ctx = dc->ctx;
+	struct dcn_hubbub_wm wm;
+	int i;
+
+	unsigned int chars_printed = 0;
+	unsigned int remaining_buffer = bufSize;
+
+	const uint32_t ref_clk_mhz = dc_ctx->dc->res_pool->ref_clocks.dchub_ref_clock_inKhz / 1000;
+	static const unsigned int frac = 1000;
+
+	memset(&wm, 0, sizeof(struct dcn_hubbub_wm));
+	dc->res_pool->hubbub->funcs->wm_read_state(dc->res_pool->hubbub, &wm);
+
+	chars_printed = snprintf_count(pBuf, remaining_buffer, "wm_set_index,data_urgent,pte_meta_urgent,sr_enter,sr_exit,dram_clk_change\n");
+	remaining_buffer -= chars_printed;
+	pBuf += chars_printed;
+
+	for (i = 0; i < 4; i++) {
+		struct dcn_hubbub_wm_set *s;
+
+		s = &wm.sets[i];
+
+		chars_printed = snprintf_count(pBuf, remaining_buffer, "%x,%d.%03d,%d.%03d,%d.%03d,%d.%03d,%d.%03d\n",
+			s->wm_set,
+			(s->data_urgent * frac) / ref_clk_mhz / frac, (s->data_urgent * frac) / ref_clk_mhz % frac,
+			(s->pte_meta_urgent * frac) / ref_clk_mhz / frac, (s->pte_meta_urgent * frac) / ref_clk_mhz % frac,
+			(s->sr_enter * frac) / ref_clk_mhz / frac, (s->sr_enter * frac) / ref_clk_mhz % frac,
+			(s->sr_exit * frac) / ref_clk_mhz / frac, (s->sr_exit * frac) / ref_clk_mhz % frac,
+			(s->dram_clk_change * frac) / ref_clk_mhz / frac, (s->dram_clk_change * frac) / ref_clk_mhz % frac);
+		remaining_buffer -= chars_printed;
+		pBuf += chars_printed;
+	}
+
+	return bufSize - remaining_buffer;
+}
+
+static unsigned int dcn10_get_hubp_states(struct dc *dc, char *pBuf, unsigned int bufSize, bool invarOnly)
+{
+	struct dc_context *dc_ctx = dc->ctx;
+	struct resource_pool *pool = dc->res_pool;
+	int i;
+
+	unsigned int chars_printed = 0;
+	unsigned int remaining_buffer = bufSize;
+
+	const uint32_t ref_clk_mhz = dc_ctx->dc->res_pool->ref_clocks.dchub_ref_clock_inKhz / 1000;
+	static const unsigned int frac = 1000;
+
+	if (invarOnly)
+		chars_printed = snprintf_count(pBuf, remaining_buffer, "instance,format,addr_hi,width,height,rotation,mirror,sw_mode,dcc_en,blank_en,ttu_dis,underflow,"
+			"min_ttu_vblank,qos_low_wm,qos_high_wm"
+			"\n");
+	else
+		chars_printed = snprintf_count(pBuf, remaining_buffer, "instance,format,addr_hi,addr_lo,width,height,rotation,mirror,sw_mode,dcc_en,blank_en,ttu_dis,underflow,"
+					"min_ttu_vblank,qos_low_wm,qos_high_wm"
+					"\n");
+
+	remaining_buffer -= chars_printed;
+	pBuf += chars_printed;
+
+	for (i = 0; i < pool->pipe_count; i++) {
+		struct hubp *hubp = pool->hubps[i];
+		struct dcn_hubp_state *s = &(TO_DCN10_HUBP(hubp)->state);
+
+		hubp->funcs->hubp_read_state(hubp);
+
+		if (!s->blank_en) {
+			if (invarOnly)
+				chars_printed = snprintf_count(pBuf, remaining_buffer, "%x,%x,%x,%d,%d,%x,%x,%x,%x,%x,%x,%x,"
+					"%d.%03d,%d.%03d,%d.%03d"
+					"\n",
+					hubp->inst,
+					s->pixel_format,
+					s->inuse_addr_hi,
+					s->viewport_width,
+					s->viewport_height,
+					s->rotation_angle,
+					s->h_mirror_en,
+					s->sw_mode,
+					s->dcc_en,
+					s->blank_en,
+					s->ttu_disable,
+					s->underflow_status,
+					(s->min_ttu_vblank * frac) / ref_clk_mhz / frac, (s->min_ttu_vblank * frac) / ref_clk_mhz % frac,
+					(s->qos_level_low_wm * frac) / ref_clk_mhz / frac, (s->qos_level_low_wm * frac) / ref_clk_mhz % frac,
+					(s->qos_level_high_wm * frac) / ref_clk_mhz / frac, (s->qos_level_high_wm * frac) / ref_clk_mhz % frac);
+			else
+				chars_printed = snprintf_count(pBuf, remaining_buffer, "%x,%x,%x,%x,%d,%d,%x,%x,%x,%x,%x,%x,%x,"
+					"%d.%03d,%d.%03d,%d.%03d"
+					"\n",
+					hubp->inst,
+					s->pixel_format,
+					s->inuse_addr_hi,
+					s->inuse_addr_lo,
+					s->viewport_width,
+					s->viewport_height,
+					s->rotation_angle,
+					s->h_mirror_en,
+					s->sw_mode,
+					s->dcc_en,
+					s->blank_en,
+					s->ttu_disable,
+					s->underflow_status,
+					(s->min_ttu_vblank * frac) / ref_clk_mhz / frac, (s->min_ttu_vblank * frac) / ref_clk_mhz % frac,
+					(s->qos_level_low_wm * frac) / ref_clk_mhz / frac, (s->qos_level_low_wm * frac) / ref_clk_mhz % frac,
+					(s->qos_level_high_wm * frac) / ref_clk_mhz / frac, (s->qos_level_high_wm * frac) / ref_clk_mhz % frac);
+
+			remaining_buffer -= chars_printed;
+			pBuf += chars_printed;
+		}
+	}
+
+	return bufSize - remaining_buffer;
+}
+
+static unsigned int dcn10_get_rq_states(struct dc *dc, char *pBuf, unsigned int bufSize)
+{
+	struct resource_pool *pool = dc->res_pool;
+	int i;
+
+	unsigned int chars_printed = 0;
+	unsigned int remaining_buffer = bufSize;
+
+	chars_printed = snprintf_count(pBuf, remaining_buffer, "instance,drq_exp_m,prq_exp_m,mrq_exp_m,crq_exp_m,plane1_ba,"
+		"luma_chunk_s,luma_min_chu_s,luma_meta_ch_s,luma_min_m_c_s,luma_dpte_gr_s,luma_mpte_gr_s,luma_swath_hei,luma_pte_row_h,"
+		"chroma_chunk_s,chroma_min_chu_s,chroma_meta_ch_s,chroma_min_m_c_s,chroma_dpte_gr_s,chroma_mpte_gr_s,chroma_swath_hei,chroma_pte_row_h"
+		"\n");
+	remaining_buffer -= chars_printed;
+	pBuf += chars_printed;
+
+	for (i = 0; i < pool->pipe_count; i++) {
+		struct dcn_hubp_state *s = &(TO_DCN10_HUBP(pool->hubps[i])->state);
+		struct _vcs_dpi_display_rq_regs_st *rq_regs = &s->rq_regs;
+
+		if (!s->blank_en) {
+			chars_printed = snprintf_count(pBuf, remaining_buffer, "%x,%x,%x,%x,%x,%x,"
+				"%x,%x,%x,%x,%x,%x,%x,%x,"
+				"%x,%x,%x,%x,%x,%x,%x,%x"
+				"\n",
+				pool->hubps[i]->inst, rq_regs->drq_expansion_mode, rq_regs->prq_expansion_mode, rq_regs->mrq_expansion_mode,
+				rq_regs->crq_expansion_mode, rq_regs->plane1_base_address, rq_regs->rq_regs_l.chunk_size,
+				rq_regs->rq_regs_l.min_chunk_size, rq_regs->rq_regs_l.meta_chunk_size,
+				rq_regs->rq_regs_l.min_meta_chunk_size, rq_regs->rq_regs_l.dpte_group_size,
+				rq_regs->rq_regs_l.mpte_group_size, rq_regs->rq_regs_l.swath_height,
+				rq_regs->rq_regs_l.pte_row_height_linear, rq_regs->rq_regs_c.chunk_size, rq_regs->rq_regs_c.min_chunk_size,
+				rq_regs->rq_regs_c.meta_chunk_size, rq_regs->rq_regs_c.min_meta_chunk_size,
+				rq_regs->rq_regs_c.dpte_group_size, rq_regs->rq_regs_c.mpte_group_size,
+				rq_regs->rq_regs_c.swath_height, rq_regs->rq_regs_c.pte_row_height_linear);
+
+			remaining_buffer -= chars_printed;
+			pBuf += chars_printed;
+		}
+	}
+
+	return bufSize - remaining_buffer;
+}
+
+static unsigned int dcn10_get_dlg_states(struct dc *dc, char *pBuf, unsigned int bufSize)
+{
+	struct resource_pool *pool = dc->res_pool;
+	int i;
+
+	unsigned int chars_printed = 0;
+	unsigned int remaining_buffer = bufSize;
+
+	chars_printed = snprintf_count(pBuf, remaining_buffer, "instance,rc_hbe,dlg_vbe,min_d_y_n,rc_per_ht,rc_x_a_s,"
+		"dst_y_a_s,dst_y_pf,dst_y_vvb,dst_y_rvb,dst_y_vfl,dst_y_rfl,rf_pix_fq,"
+		"vratio_pf,vrat_pf_c,rc_pg_vbl,rc_pg_vbc,rc_mc_vbl,rc_mc_vbc,rc_pg_fll,"
+		"rc_pg_flc,rc_mc_fll,rc_mc_flc,pr_nom_l,pr_nom_c,rc_pg_nl,rc_pg_nc,"
+		"mr_nom_l,mr_nom_c,rc_mc_nl,rc_mc_nc,rc_ld_pl,rc_ld_pc,rc_ld_l,"
+		"rc_ld_c,cha_cur0,ofst_cur1,cha_cur1,vr_af_vc0,ddrq_limt,x_rt_dlay,x_rp_dlay,x_rr_sfl"
+		"\n");
+	remaining_buffer -= chars_printed;
+	pBuf += chars_printed;
+
+	for (i = 0; i < pool->pipe_count; i++) {
+		struct dcn_hubp_state *s = &(TO_DCN10_HUBP(pool->hubps[i])->state);
+		struct _vcs_dpi_display_dlg_regs_st *dlg_regs = &s->dlg_attr;
+
+		if (!s->blank_en) {
+			chars_printed = snprintf_count(pBuf, remaining_buffer, "%x,%x,%x,%x,%x,"
+				"%x,%x,%x,%x,%x,%x,%x,"
+				"%x,%x,%x,%x,%x,%x,%x,"
+				"%x,%x,%x,%x,%x,%x,%x,"
+				"%x,%x,%x,%x,%x,%x,%x,"
+				"%x,%x,%x,%x,%x,%x,%x,%x,%x,%x"
+				"\n",
+				pool->hubps[i]->inst, dlg_regs->refcyc_h_blank_end, dlg_regs->dlg_vblank_end, dlg_regs->min_dst_y_next_start,
+				dlg_regs->refcyc_per_htotal, dlg_regs->refcyc_x_after_scaler, dlg_regs->dst_y_after_scaler,
+				dlg_regs->dst_y_prefetch, dlg_regs->dst_y_per_vm_vblank, dlg_regs->dst_y_per_row_vblank,
+				dlg_regs->dst_y_per_vm_flip, dlg_regs->dst_y_per_row_flip, dlg_regs->ref_freq_to_pix_freq,
+				dlg_regs->vratio_prefetch, dlg_regs->vratio_prefetch_c, dlg_regs->refcyc_per_pte_group_vblank_l,
+				dlg_regs->refcyc_per_pte_group_vblank_c, dlg_regs->refcyc_per_meta_chunk_vblank_l,
+				dlg_regs->refcyc_per_meta_chunk_vblank_c, dlg_regs->refcyc_per_pte_group_flip_l,
+				dlg_regs->refcyc_per_pte_group_flip_c, dlg_regs->refcyc_per_meta_chunk_flip_l,
+				dlg_regs->refcyc_per_meta_chunk_flip_c, dlg_regs->dst_y_per_pte_row_nom_l,
+				dlg_regs->dst_y_per_pte_row_nom_c, dlg_regs->refcyc_per_pte_group_nom_l,
+				dlg_regs->refcyc_per_pte_group_nom_c, dlg_regs->dst_y_per_meta_row_nom_l,
+				dlg_regs->dst_y_per_meta_row_nom_c, dlg_regs->refcyc_per_meta_chunk_nom_l,
+				dlg_regs->refcyc_per_meta_chunk_nom_c, dlg_regs->refcyc_per_line_delivery_pre_l,
+				dlg_regs->refcyc_per_line_delivery_pre_c, dlg_regs->refcyc_per_line_delivery_l,
+				dlg_regs->refcyc_per_line_delivery_c, dlg_regs->chunk_hdl_adjust_cur0, dlg_regs->dst_y_offset_cur1,
+				dlg_regs->chunk_hdl_adjust_cur1, dlg_regs->vready_after_vcount0, dlg_regs->dst_y_delta_drq_limit,
+				dlg_regs->xfc_reg_transfer_delay, dlg_regs->xfc_reg_precharge_delay,
+				dlg_regs->xfc_reg_remote_surface_flip_latency);
+
+			remaining_buffer -= chars_printed;
+			pBuf += chars_printed;
+		}
+	}
+
+	return bufSize - remaining_buffer;
+}
+
+static unsigned int dcn10_get_ttu_states(struct dc *dc, char *pBuf, unsigned int bufSize)
+{
+	struct resource_pool *pool = dc->res_pool;
+	int i;
+
+	unsigned int chars_printed = 0;
+	unsigned int remaining_buffer = bufSize;
+
+	chars_printed = snprintf_count(pBuf, remaining_buffer, "instance,qos_ll_wm,qos_lh_wm,mn_ttu_vb,qos_l_flp,rc_rd_p_l,rc_rd_l,rc_rd_p_c,"
+		"rc_rd_c,rc_rd_c0,rc_rd_pc0,rc_rd_c1,rc_rd_pc1,qos_lf_l,qos_rds_l,"
+		"qos_lf_c,qos_rds_c,qos_lf_c0,qos_rds_c0,qos_lf_c1,qos_rds_c1"
+		"\n");
+	remaining_buffer -= chars_printed;
+	pBuf += chars_printed;
+
+	for (i = 0; i < pool->pipe_count; i++) {
+		struct dcn_hubp_state *s = &(TO_DCN10_HUBP(pool->hubps[i])->state);
+		struct _vcs_dpi_display_ttu_regs_st *ttu_regs = &s->ttu_attr;
+
+		if (!s->blank_en) {
+			chars_printed = snprintf_count(pBuf, remaining_buffer, "%x,%x,%x,%x,%x,%x,%x,%x,"
+				"%x,%x,%x,%x,%x,%x,%x,"
+				"%x,%x,%x,%x,%x,%x"
+				"\n",
+				pool->hubps[i]->inst, ttu_regs->qos_level_low_wm, ttu_regs->qos_level_high_wm, ttu_regs->min_ttu_vblank,
+				ttu_regs->qos_level_flip, ttu_regs->refcyc_per_req_delivery_pre_l, ttu_regs->refcyc_per_req_delivery_l,
+				ttu_regs->refcyc_per_req_delivery_pre_c, ttu_regs->refcyc_per_req_delivery_c, ttu_regs->refcyc_per_req_delivery_cur0,
+				ttu_regs->refcyc_per_req_delivery_pre_cur0, ttu_regs->refcyc_per_req_delivery_cur1,
+				ttu_regs->refcyc_per_req_delivery_pre_cur1, ttu_regs->qos_level_fixed_l, ttu_regs->qos_ramp_disable_l,
+				ttu_regs->qos_level_fixed_c, ttu_regs->qos_ramp_disable_c, ttu_regs->qos_level_fixed_cur0,
+				ttu_regs->qos_ramp_disable_cur0, ttu_regs->qos_level_fixed_cur1, ttu_regs->qos_ramp_disable_cur1);
+
+			remaining_buffer -= chars_printed;
+			pBuf += chars_printed;
+		}
+	}
+
+	return bufSize - remaining_buffer;
+}
+
+static unsigned int dcn10_get_cm_states(struct dc *dc, char *pBuf, unsigned int bufSize)
+{
+	struct resource_pool *pool = dc->res_pool;
+	int i;
+
+	unsigned int chars_printed = 0;
+	unsigned int remaining_buffer = bufSize;
+
+	chars_printed = snprintf_count(pBuf, remaining_buffer, "instance,igam_format,igam_mode,dgam_mode,rgam_mode,gamut_mode,"
+		"c11_c12,c13_c14,c21_c22,c23_c24,c31_c32,c33_c34"
+		"\n");
+	remaining_buffer -= chars_printed;
+	pBuf += chars_printed;
+
+	for (i = 0; i < pool->pipe_count; i++) {
+		struct dpp *dpp = pool->dpps[i];
+		struct dcn_dpp_state s = {0};
+
+		dpp->funcs->dpp_read_state(dpp, &s);
+
+		if (s.is_enabled) {
+			chars_printed = snprintf_count(pBuf, remaining_buffer, "%x,%x,"
+					"%s,%s,%s,"
+					"%x,%08x,%08x,%08x,%08x,%08x,%08x"
+				"\n",
+				dpp->inst, s.igam_input_format,
+				(s.igam_lut_mode == 0) ? "BypassFixed" :
+					((s.igam_lut_mode == 1) ? "BypassFloat" :
+					((s.igam_lut_mode == 2) ? "RAM" :
+					((s.igam_lut_mode == 3) ? "RAM" :
+								 "Unknown"))),
+				(s.dgam_lut_mode == 0) ? "Bypass" :
+					((s.dgam_lut_mode == 1) ? "sRGB" :
+					((s.dgam_lut_mode == 2) ? "Ycc" :
+					((s.dgam_lut_mode == 3) ? "RAM" :
+					((s.dgam_lut_mode == 4) ? "RAM" :
+								 "Unknown")))),
+				(s.rgam_lut_mode == 0) ? "Bypass" :
+					((s.rgam_lut_mode == 1) ? "sRGB" :
+					((s.rgam_lut_mode == 2) ? "Ycc" :
+					((s.rgam_lut_mode == 3) ? "RAM" :
+					((s.rgam_lut_mode == 4) ? "RAM" :
+								 "Unknown")))),
+				s.gamut_remap_mode, s.gamut_remap_c11_c12,
+				s.gamut_remap_c13_c14, s.gamut_remap_c21_c22, s.gamut_remap_c23_c24,
+				s.gamut_remap_c31_c32, s.gamut_remap_c33_c34);
+
+			remaining_buffer -= chars_printed;
+			pBuf += chars_printed;
+		}
+	}
+
+	return bufSize - remaining_buffer;
+}
+
+static unsigned int dcn10_get_mpcc_states(struct dc *dc, char *pBuf, unsigned int bufSize)
+{
+	struct resource_pool *pool = dc->res_pool;
+	int i;
+
+	unsigned int chars_printed = 0;
+	unsigned int remaining_buffer = bufSize;
+
+	chars_printed = snprintf_count(pBuf, remaining_buffer, "instance,opp,dpp,mpccbot,mode,alpha_mode,premult,overlap_only,idle\n");
+	remaining_buffer -= chars_printed;
+	pBuf += chars_printed;
+
+	for (i = 0; i < pool->mpcc_count; i++) {
+		struct mpcc_state s = {0};
+
+		pool->mpc->funcs->read_mpcc_state(pool->mpc, i, &s);
+
+		if (s.opp_id != 0xf) {
+			chars_printed = snprintf_count(pBuf, remaining_buffer, "%x,%x,%x,%x,%x,%x,%x,%x,%x\n",
+				i, s.opp_id, s.dpp_id, s.bot_mpcc_id,
+				s.mode, s.alpha_mode, s.pre_multiplied_alpha, s.overlap_only,
+				s.idle);
+
+			remaining_buffer -= chars_printed;
+			pBuf += chars_printed;
+		}
+	}
+
+	return bufSize - remaining_buffer;
+}
+
+static unsigned int dcn10_get_otg_states(struct dc *dc, char *pBuf, unsigned int bufSize)
+{
+	struct resource_pool *pool = dc->res_pool;
+	int i;
+
+	unsigned int chars_printed = 0;
+	unsigned int remaining_buffer = bufSize;
+
+	chars_printed = snprintf_count(pBuf, remaining_buffer, "instance,v_bs,v_be,v_ss,v_se,vpol,vmax,vmin,vmax_sel,vmin_sel,"
+			"h_bs,h_be,h_ss,h_se,hpol,htot,vtot,underflow,pixelclk[khz]\n");
+	remaining_buffer -= chars_printed;
+	pBuf += chars_printed;
+
+	for (i = 0; i < pool->timing_generator_count; i++) {
+		struct timing_generator *tg = pool->timing_generators[i];
+		struct dcn_otg_state s = {0};
+		int pix_clk = 0;
+
+		if (tg->funcs->read_otg_state)
+			tg->funcs->read_otg_state(tg, &s);
+
+		pix_clk = dc->current_state->res_ctx.pipe_ctx[i].stream_res.pix_clk_params.requested_pix_clk_100hz / 10;
+
+		//only print if OTG master is enabled
+		if (s.otg_enabled & 1) {
+			chars_printed = snprintf_count(pBuf, remaining_buffer, "%x,%d,%d,%d,%d,%d,%d,%d,%d,%d,"
+				"%d,%d,%d,%d,%d,%d,%d,%d,%d"
+				"\n",
+				tg->inst,
+				s.v_blank_start,
+				s.v_blank_end,
+				s.v_sync_a_start,
+				s.v_sync_a_end,
+				s.v_sync_a_pol,
+				s.v_total_max,
+				s.v_total_min,
+				s.v_total_max_sel,
+				s.v_total_min_sel,
+				s.h_blank_start,
+				s.h_blank_end,
+				s.h_sync_a_start,
+				s.h_sync_a_end,
+				s.h_sync_a_pol,
+				s.h_total,
+				s.v_total,
+				s.underflow_occurred_status,
+				pix_clk);
+
+			remaining_buffer -= chars_printed;
+			pBuf += chars_printed;
+		}
+	}
+
+	return bufSize - remaining_buffer;
+}
+
+static unsigned int dcn10_get_clock_states(struct dc *dc, char *pBuf, unsigned int bufSize)
+{
+	unsigned int chars_printed = 0;
+	unsigned int remaining_buffer = bufSize;
+
+	chars_printed = snprintf_count(pBuf, bufSize, "dcfclk,dcfclk_deep_sleep,dispclk,"
+		"dppclk,fclk,socclk\n"
+		"%d,%d,%d,%d,%d,%d\n",
+		dc->current_state->bw_ctx.bw.dcn.clk.dcfclk_khz,
+		dc->current_state->bw_ctx.bw.dcn.clk.dcfclk_deep_sleep_khz,
+		dc->current_state->bw_ctx.bw.dcn.clk.dispclk_khz,
+		dc->current_state->bw_ctx.bw.dcn.clk.dppclk_khz,
+		dc->current_state->bw_ctx.bw.dcn.clk.fclk_khz,
+		dc->current_state->bw_ctx.bw.dcn.clk.socclk_khz);
+
+	remaining_buffer -= chars_printed;
+	pBuf += chars_printed;
+
+	return bufSize - remaining_buffer;
+}
+
+static void dcn10_clear_otpc_underflow(struct dc *dc)
+{
+	struct resource_pool *pool = dc->res_pool;
+	int i;
+
+	for (i = 0; i < pool->timing_generator_count; i++) {
+		struct timing_generator *tg = pool->timing_generators[i];
+		struct dcn_otg_state s = {0};
+
+		if (tg->funcs->read_otg_state)
+			tg->funcs->read_otg_state(tg, &s);
+
+		if (s.otg_enabled & 1)
+			tg->funcs->clear_optc_underflow(tg);
+	}
+}
+
+static void dcn10_clear_hubp_underflow(struct dc *dc)
+{
+	struct resource_pool *pool = dc->res_pool;
+	int i;
+
+	for (i = 0; i < pool->pipe_count; i++) {
+		struct hubp *hubp = pool->hubps[i];
+		struct dcn_hubp_state *s = &(TO_DCN10_HUBP(hubp)->state);
+
+		hubp->funcs->hubp_read_state(hubp);
+
+		if (!s->blank_en)
+			hubp->funcs->hubp_clear_underflow(hubp);
+	}
+}
+
+void dcn10_clear_status_bits(struct dc *dc, unsigned int mask)
+{
+	/*
+	 *  Mask Format
+	 *  Bit 0 - 31: Status bit to clear
+	 *
+	 *  Mask = 0x0 means clear all status bits
+	 */
+	const unsigned int DC_HW_STATE_MASK_HUBP_UNDERFLOW	= 0x1;
+	const unsigned int DC_HW_STATE_MASK_OTPC_UNDERFLOW	= 0x2;
+
+	if (mask == 0x0)
+		mask = 0xFFFFFFFF;
+
+	if (mask & DC_HW_STATE_MASK_HUBP_UNDERFLOW)
+		dcn10_clear_hubp_underflow(dc);
+
+	if (mask & DC_HW_STATE_MASK_OTPC_UNDERFLOW)
+		dcn10_clear_otpc_underflow(dc);
+}
+
+void dcn10_get_hw_state(struct dc *dc, char *pBuf, unsigned int bufSize, unsigned int mask)
+{
+	/*
+	 *  Mask Format
+	 *  Bit 0 - 15: Hardware block mask
+	 *  Bit 15: 1 = Invariant Only, 0 = All
+	 */
+	const unsigned int DC_HW_STATE_MASK_HUBBUB			= 0x1;
+	const unsigned int DC_HW_STATE_MASK_HUBP			= 0x2;
+	const unsigned int DC_HW_STATE_MASK_RQ				= 0x4;
+	const unsigned int DC_HW_STATE_MASK_DLG				= 0x8;
+	const unsigned int DC_HW_STATE_MASK_TTU				= 0x10;
+	const unsigned int DC_HW_STATE_MASK_CM				= 0x20;
+	const unsigned int DC_HW_STATE_MASK_MPCC			= 0x40;
+	const unsigned int DC_HW_STATE_MASK_OTG				= 0x80;
+	const unsigned int DC_HW_STATE_MASK_CLOCKS			= 0x100;
+	const unsigned int DC_HW_STATE_INVAR_ONLY			= 0x8000;
+
+	unsigned int chars_printed = 0;
+	unsigned int remaining_buf_size = bufSize;
+
+	if (mask == 0x0)
+		mask = 0xFFFF; // Default, capture all, invariant only
+
+	if ((mask & DC_HW_STATE_MASK_HUBBUB) && remaining_buf_size > 0) {
+		chars_printed = dcn10_get_hubbub_state(dc, pBuf, remaining_buf_size);
+		pBuf += chars_printed;
+		remaining_buf_size -= chars_printed;
+	}
+
+	if ((mask & DC_HW_STATE_MASK_HUBP) && remaining_buf_size > 0) {
+		chars_printed = dcn10_get_hubp_states(dc, pBuf, remaining_buf_size, mask & DC_HW_STATE_INVAR_ONLY);
+		pBuf += chars_printed;
+		remaining_buf_size -= chars_printed;
+	}
+
+	if ((mask & DC_HW_STATE_MASK_RQ) && remaining_buf_size > 0) {
+		chars_printed = dcn10_get_rq_states(dc, pBuf, remaining_buf_size);
+		pBuf += chars_printed;
+		remaining_buf_size -= chars_printed;
+	}
+
+	if ((mask & DC_HW_STATE_MASK_DLG) && remaining_buf_size > 0) {
+		chars_printed = dcn10_get_dlg_states(dc, pBuf, remaining_buf_size);
+		pBuf += chars_printed;
+		remaining_buf_size -= chars_printed;
+	}
+
+	if ((mask & DC_HW_STATE_MASK_TTU) && remaining_buf_size > 0) {
+		chars_printed = dcn10_get_ttu_states(dc, pBuf, remaining_buf_size);
+		pBuf += chars_printed;
+		remaining_buf_size -= chars_printed;
+	}
+
+	if ((mask & DC_HW_STATE_MASK_CM) && remaining_buf_size > 0) {
+		chars_printed = dcn10_get_cm_states(dc, pBuf, remaining_buf_size);
+		pBuf += chars_printed;
+		remaining_buf_size -= chars_printed;
+	}
+
+	if ((mask & DC_HW_STATE_MASK_MPCC) && remaining_buf_size > 0) {
+		chars_printed = dcn10_get_mpcc_states(dc, pBuf, remaining_buf_size);
+		pBuf += chars_printed;
+		remaining_buf_size -= chars_printed;
+	}
+
+	if ((mask & DC_HW_STATE_MASK_OTG) && remaining_buf_size > 0) {
+		chars_printed = dcn10_get_otg_states(dc, pBuf, remaining_buf_size);
+		pBuf += chars_printed;
+		remaining_buf_size -= chars_printed;
+	}
+
+	if ((mask & DC_HW_STATE_MASK_CLOCKS) && remaining_buf_size > 0) {
+		chars_printed = dcn10_get_clock_states(dc, pBuf, remaining_buf_size);
+		pBuf += chars_printed;
+		remaining_buf_size -= chars_printed;
+	}
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer_debug.h b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer_debug.h
new file mode 100644
index 000000000000..596f95c22e85
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer_debug.h
@@ -0,0 +1,43 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_HWSS_DCN10_DEBUG_H__
+#define __DC_HWSS_DCN10_DEBUG_H__
+
+#include "core_types.h"
+
+struct dc;
+
+void dcn10_clear_status_bits(struct dc *dc, unsigned int mask);
+
+void dcn10_log_hw_state(struct dc *dc,
+		struct dc_log_buffer_ctx *log_ctx);
+
+void dcn10_get_hw_state(struct dc *dc,
+		char *pBuf,
+		unsigned int bufSize,
+		unsigned int mask);
+
+#endif /* __DC_HWSS_DCN10_DEBUG_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_ipp.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_ipp.c
new file mode 100644
index 000000000000..24b68337d76e
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_ipp.c
@@ -0,0 +1,90 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+#include "dcn10_ipp.h"
+#include "reg_helper.h"
+
+#define REG(reg) \
+	(ippn10->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	ippn10->ipp_shift->field_name, ippn10->ipp_mask->field_name
+
+#define CTX \
+	ippn10->base.ctx
+
+/*****************************************/
+/* Constructor, Destructor               */
+/*****************************************/
+
+static void dcn10_ipp_destroy(struct input_pixel_processor **ipp)
+{
+	kfree(TO_DCN10_IPP(*ipp));
+	*ipp = NULL;
+}
+
+static const struct ipp_funcs dcn10_ipp_funcs = {
+	.ipp_destroy			= dcn10_ipp_destroy
+};
+
+static const struct ipp_funcs dcn20_ipp_funcs = {
+	.ipp_destroy			= dcn10_ipp_destroy
+};
+
+void dcn10_ipp_construct(
+	struct dcn10_ipp *ippn10,
+	struct dc_context *ctx,
+	int inst,
+	const struct dcn10_ipp_registers *regs,
+	const struct dcn10_ipp_shift *ipp_shift,
+	const struct dcn10_ipp_mask *ipp_mask)
+{
+	ippn10->base.ctx = ctx;
+	ippn10->base.inst = inst;
+	ippn10->base.funcs = &dcn10_ipp_funcs;
+
+	ippn10->regs = regs;
+	ippn10->ipp_shift = ipp_shift;
+	ippn10->ipp_mask = ipp_mask;
+}
+
+void dcn20_ipp_construct(
+	struct dcn10_ipp *ippn10,
+	struct dc_context *ctx,
+	int inst,
+	const struct dcn10_ipp_registers *regs,
+	const struct dcn10_ipp_shift *ipp_shift,
+	const struct dcn10_ipp_mask *ipp_mask)
+{
+	ippn10->base.ctx = ctx;
+	ippn10->base.inst = inst;
+	ippn10->base.funcs = &dcn20_ipp_funcs;
+
+	ippn10->regs = regs;
+	ippn10->ipp_shift = ipp_shift;
+	ippn10->ipp_mask = ipp_mask;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_ipp.h b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_ipp.h
new file mode 100644
index 000000000000..e2508d637e0a
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_ipp.h
@@ -0,0 +1,223 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef _DCN10_IPP_H_
+#define _DCN10_IPP_H_
+
+#include "ipp.h"
+
+#define TO_DCN10_IPP(ipp)\
+	container_of(ipp, struct dcn10_ipp, base)
+
+#define IPP_REG_LIST_DCN(id) \
+	SRI(FORMAT_CONTROL, CNVC_CFG, id), \
+	SRI(CNVC_SURFACE_PIXEL_FORMAT, CNVC_CFG, id), \
+	SRI(CURSOR0_CONTROL, CNVC_CUR, id), \
+	SRI(CURSOR0_COLOR0, CNVC_CUR, id), \
+	SRI(CURSOR0_COLOR1, CNVC_CUR, id)
+
+#define IPP_REG_LIST_DCN10(id) \
+	IPP_REG_LIST_DCN(id), \
+	SRI(CURSOR_SETTINS, HUBPREQ, id), \
+	SRI(CURSOR_SURFACE_ADDRESS_HIGH, CURSOR, id), \
+	SRI(CURSOR_SURFACE_ADDRESS, CURSOR, id), \
+	SRI(CURSOR_SIZE, CURSOR, id), \
+	SRI(CURSOR_CONTROL, CURSOR, id), \
+	SRI(CURSOR_POSITION, CURSOR, id), \
+	SRI(CURSOR_HOT_SPOT, CURSOR, id), \
+	SRI(CURSOR_DST_OFFSET, CURSOR, id)
+
+#define IPP_REG_LIST_DCN20(id) \
+	IPP_REG_LIST_DCN(id), \
+	SRI(CURSOR_SETTINGS, HUBPREQ, id), \
+	SRI(CURSOR_SURFACE_ADDRESS_HIGH, CURSOR0_, id), \
+	SRI(CURSOR_SURFACE_ADDRESS, CURSOR0_, id), \
+	SRI(CURSOR_SIZE, CURSOR0_, id), \
+	SRI(CURSOR_CONTROL, CURSOR0_, id), \
+	SRI(CURSOR_POSITION, CURSOR0_, id), \
+	SRI(CURSOR_HOT_SPOT, CURSOR0_, id), \
+	SRI(CURSOR_DST_OFFSET, CURSOR0_, id)
+
+#define IPP_REG_LIST_DCN201(id) \
+	IPP_REG_LIST_DCN(id), \
+	SRI(CURSOR_SURFACE_ADDRESS_HIGH, CURSOR0_, id), \
+	SRI(CURSOR_SURFACE_ADDRESS, CURSOR0_, id), \
+	SRI(CURSOR_SIZE, CURSOR0_, id), \
+	SRI(CURSOR_CONTROL, CURSOR0_, id), \
+	SRI(CURSOR_POSITION, CURSOR0_, id), \
+	SRI(CURSOR_HOT_SPOT, CURSOR0_, id), \
+	SRI(CURSOR_DST_OFFSET, CURSOR0_, id)
+
+#define CURSOR0_CURSOR_CONTROL__CURSOR_2X_MAGNIFY__SHIFT	0x4
+#define CURSOR0_CURSOR_CONTROL__CURSOR_2X_MAGNIFY_MASK		0x00000010L
+
+#define IPP_SF(reg_name, field_name, post_fix)\
+	.field_name = reg_name ## __ ## field_name ## post_fix
+
+#define IPP_MASK_SH_LIST_DCN(mask_sh) \
+	IPP_SF(CNVC_CFG0_CNVC_SURFACE_PIXEL_FORMAT, CNVC_SURFACE_PIXEL_FORMAT, mask_sh), \
+	IPP_SF(CNVC_CFG0_FORMAT_CONTROL, CNVC_BYPASS, mask_sh), \
+	IPP_SF(CNVC_CFG0_FORMAT_CONTROL, ALPHA_EN, mask_sh), \
+	IPP_SF(CNVC_CFG0_FORMAT_CONTROL, FORMAT_EXPANSION_MODE, mask_sh), \
+	IPP_SF(CNVC_CUR0_CURSOR0_CONTROL, CUR0_MODE, mask_sh), \
+	IPP_SF(CNVC_CUR0_CURSOR0_COLOR0, CUR0_COLOR0, mask_sh), \
+	IPP_SF(CNVC_CUR0_CURSOR0_COLOR1, CUR0_COLOR1, mask_sh), \
+	IPP_SF(CNVC_CUR0_CURSOR0_CONTROL, CUR0_EXPANSION_MODE, mask_sh), \
+	IPP_SF(CNVC_CUR0_CURSOR0_CONTROL, CUR0_ENABLE, mask_sh)
+
+#define IPP_MASK_SH_LIST_DCN10(mask_sh) \
+	IPP_MASK_SH_LIST_DCN(mask_sh),\
+	IPP_SF(HUBPREQ0_CURSOR_SETTINS, CURSOR0_DST_Y_OFFSET, mask_sh), \
+	IPP_SF(HUBPREQ0_CURSOR_SETTINS, CURSOR0_CHUNK_HDL_ADJUST, mask_sh), \
+	IPP_SF(CURSOR0_CURSOR_SURFACE_ADDRESS_HIGH, CURSOR_SURFACE_ADDRESS_HIGH, mask_sh), \
+	IPP_SF(CURSOR0_CURSOR_SURFACE_ADDRESS, CURSOR_SURFACE_ADDRESS, mask_sh), \
+	IPP_SF(CURSOR0_CURSOR_SIZE, CURSOR_WIDTH, mask_sh), \
+	IPP_SF(CURSOR0_CURSOR_SIZE, CURSOR_HEIGHT, mask_sh), \
+	IPP_SF(CURSOR0_CURSOR_CONTROL, CURSOR_MODE, mask_sh), \
+	IPP_SF(CURSOR0_CURSOR_CONTROL, CURSOR_2X_MAGNIFY, mask_sh), \
+	IPP_SF(CURSOR0_CURSOR_CONTROL, CURSOR_PITCH, mask_sh), \
+	IPP_SF(CURSOR0_CURSOR_CONTROL, CURSOR_LINES_PER_CHUNK, mask_sh), \
+	IPP_SF(CURSOR0_CURSOR_CONTROL, CURSOR_ENABLE, mask_sh), \
+	IPP_SF(CURSOR0_CURSOR_POSITION, CURSOR_X_POSITION, mask_sh), \
+	IPP_SF(CURSOR0_CURSOR_POSITION, CURSOR_Y_POSITION, mask_sh), \
+	IPP_SF(CURSOR0_CURSOR_HOT_SPOT, CURSOR_HOT_SPOT_X, mask_sh), \
+	IPP_SF(CURSOR0_CURSOR_HOT_SPOT, CURSOR_HOT_SPOT_Y, mask_sh), \
+	IPP_SF(CURSOR0_CURSOR_DST_OFFSET, CURSOR_DST_X_OFFSET, mask_sh), \
+	IPP_SF(CNVC_CFG0_FORMAT_CONTROL, OUTPUT_FP, mask_sh)
+
+#define IPP_MASK_SH_LIST_DCN20(mask_sh) \
+	IPP_MASK_SH_LIST_DCN(mask_sh), \
+	IPP_SF(HUBPREQ0_CURSOR_SETTINGS, CURSOR0_DST_Y_OFFSET, mask_sh), \
+	IPP_SF(HUBPREQ0_CURSOR_SETTINGS, CURSOR0_CHUNK_HDL_ADJUST, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_SURFACE_ADDRESS_HIGH, CURSOR_SURFACE_ADDRESS_HIGH, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_SURFACE_ADDRESS, CURSOR_SURFACE_ADDRESS, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_SIZE, CURSOR_WIDTH, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_SIZE, CURSOR_HEIGHT, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_CONTROL, CURSOR_MODE, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_CONTROL, CURSOR_2X_MAGNIFY, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_CONTROL, CURSOR_PITCH, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_CONTROL, CURSOR_LINES_PER_CHUNK, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_CONTROL, CURSOR_ENABLE, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_POSITION, CURSOR_X_POSITION, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_POSITION, CURSOR_Y_POSITION, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_HOT_SPOT, CURSOR_HOT_SPOT_X, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_HOT_SPOT, CURSOR_HOT_SPOT_Y, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_DST_OFFSET, CURSOR_DST_X_OFFSET, mask_sh)
+
+#define IPP_MASK_SH_LIST_DCN201(mask_sh) \
+	IPP_MASK_SH_LIST_DCN(mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_SURFACE_ADDRESS_HIGH, CURSOR_SURFACE_ADDRESS_HIGH, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_SURFACE_ADDRESS, CURSOR_SURFACE_ADDRESS, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_SIZE, CURSOR_WIDTH, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_SIZE, CURSOR_HEIGHT, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_CONTROL, CURSOR_MODE, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_CONTROL, CURSOR_2X_MAGNIFY, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_CONTROL, CURSOR_PITCH, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_CONTROL, CURSOR_LINES_PER_CHUNK, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_CONTROL, CURSOR_ENABLE, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_POSITION, CURSOR_X_POSITION, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_POSITION, CURSOR_Y_POSITION, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_HOT_SPOT, CURSOR_HOT_SPOT_X, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_HOT_SPOT, CURSOR_HOT_SPOT_Y, mask_sh), \
+	IPP_SF(CURSOR0_0_CURSOR_DST_OFFSET, CURSOR_DST_X_OFFSET, mask_sh)
+
+#define IPP_DCN10_REG_FIELD_LIST(type) \
+	type CNVC_SURFACE_PIXEL_FORMAT; \
+	type CNVC_BYPASS; \
+	type ALPHA_EN; \
+	type FORMAT_EXPANSION_MODE; \
+	type CURSOR0_DST_Y_OFFSET; \
+	type CURSOR0_CHUNK_HDL_ADJUST; \
+	type CUR0_MODE; \
+	type CUR0_COLOR0; \
+	type CUR0_COLOR1; \
+	type CUR0_EXPANSION_MODE; \
+	type CURSOR_SURFACE_ADDRESS_HIGH; \
+	type CURSOR_SURFACE_ADDRESS; \
+	type CURSOR_WIDTH; \
+	type CURSOR_HEIGHT; \
+	type CURSOR_MODE; \
+	type CURSOR_2X_MAGNIFY; \
+	type CURSOR_PITCH; \
+	type CURSOR_LINES_PER_CHUNK; \
+	type CURSOR_ENABLE; \
+	type CUR0_ENABLE; \
+	type CURSOR_X_POSITION; \
+	type CURSOR_Y_POSITION; \
+	type CURSOR_HOT_SPOT_X; \
+	type CURSOR_HOT_SPOT_Y; \
+	type CURSOR_DST_X_OFFSET; \
+	type OUTPUT_FP
+
+struct dcn10_ipp_shift {
+	IPP_DCN10_REG_FIELD_LIST(uint8_t);
+};
+
+struct dcn10_ipp_mask {
+	IPP_DCN10_REG_FIELD_LIST(uint32_t);
+};
+
+struct dcn10_ipp_registers {
+	uint32_t CURSOR_SETTINS;
+	uint32_t CURSOR_SETTINGS;
+	uint32_t CNVC_SURFACE_PIXEL_FORMAT;
+	uint32_t CURSOR0_CONTROL;
+	uint32_t CURSOR0_COLOR0;
+	uint32_t CURSOR0_COLOR1;
+	uint32_t FORMAT_CONTROL;
+	uint32_t CURSOR_SURFACE_ADDRESS_HIGH;
+	uint32_t CURSOR_SURFACE_ADDRESS;
+	uint32_t CURSOR_SIZE;
+	uint32_t CURSOR_CONTROL;
+	uint32_t CURSOR_POSITION;
+	uint32_t CURSOR_HOT_SPOT;
+	uint32_t CURSOR_DST_OFFSET;
+};
+
+struct dcn10_ipp {
+	struct input_pixel_processor base;
+
+	const struct dcn10_ipp_registers *regs;
+	const struct dcn10_ipp_shift *ipp_shift;
+	const struct dcn10_ipp_mask *ipp_mask;
+
+	struct dc_cursor_attributes curs_attr;
+};
+
+void dcn10_ipp_construct(struct dcn10_ipp *ippn10,
+	struct dc_context *ctx,
+	int inst,
+	const struct dcn10_ipp_registers *regs,
+	const struct dcn10_ipp_shift *ipp_shift,
+	const struct dcn10_ipp_mask *ipp_mask);
+
+void dcn20_ipp_construct(struct dcn10_ipp *ippn10,
+	struct dc_context *ctx,
+	int inst,
+	const struct dcn10_ipp_registers *regs,
+	const struct dcn10_ipp_shift *ipp_shift,
+	const struct dcn10_ipp_mask *ipp_mask);
+
+#endif /* _DCN10_IPP_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/Makefile b/drivers/gpu/drm/amd/display/dc/dcn20/Makefile
new file mode 100644
index 000000000000..25ba0d310d46
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn20/Makefile
@@ -0,0 +1,8 @@
+# SPDX-License-Identifier: MIT
+# Copyright © 2019-2024 Advanced Micro Devices, Inc. All rights reserved.
+
+DCN20 = dcn20_vmid.o dcn20_dwb.o dcn20_dwb_scl.o
+
+AMD_DAL_DCN20 = $(addprefix $(AMDDALPATH)/dc/dcn20/,$(DCN20))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCN20)
diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dwb.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dwb.c
new file mode 100644
index 000000000000..80779e85e2c5
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dwb.c
@@ -0,0 +1,343 @@
+/*
+ * Copyright 2012-17 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+#include "reg_helper.h"
+#include "resource.h"
+#include "dwb.h"
+#include "dcn20_dwb.h"
+
+
+#define REG(reg)\
+	dwbc20->dwbc_regs->reg
+
+#define CTX \
+	dwbc20->base.ctx
+
+#define DC_LOGGER \
+	dwbc20->base.ctx->logger
+#undef FN
+#define FN(reg_name, field_name) \
+	dwbc20->dwbc_shift->field_name, dwbc20->dwbc_mask->field_name
+
+enum dwb_outside_pix_strategy {
+	DWB_OUTSIDE_PIX_STRATEGY_BLACK = 0,
+	DWB_OUTSIDE_PIX_STRATEGY_EDGE  = 1
+};
+
+static bool dwb2_get_caps(struct dwbc *dwbc, struct dwb_caps *caps)
+{
+	struct dcn20_dwbc *dwbc20 = TO_DCN20_DWBC(dwbc);
+	if (caps) {
+		caps->adapter_id = 0;	/* we only support 1 adapter currently */
+		caps->hw_version = DCN_VERSION_2_0;
+		caps->num_pipes = 1;
+		memset(&caps->reserved, 0, sizeof(caps->reserved));
+		memset(&caps->reserved2, 0, sizeof(caps->reserved2));
+		caps->sw_version = dwb_ver_1_0;
+		caps->caps.support_dwb = true;
+		caps->caps.support_ogam = false;
+		caps->caps.support_wbscl = false;
+		caps->caps.support_ocsc = false;
+		DC_LOG_DWB("%s SUPPORTED! inst = %d", __func__, dwbc20->base.inst);
+		return true;
+	} else {
+		DC_LOG_DWB("%s NOT SUPPORTED! inst = %d", __func__, dwbc20->base.inst);
+		return false;
+	}
+}
+
+void dwb2_config_dwb_cnv(struct dwbc *dwbc, struct dc_dwb_params *params)
+{
+	struct dcn20_dwbc *dwbc20 = TO_DCN20_DWBC(dwbc);
+	DC_LOG_DWB("%s inst = %d", __func__, dwbc20->base.inst);
+
+	/* Set DWB source size */
+	REG_UPDATE_2(CNV_SOURCE_SIZE, CNV_SOURCE_WIDTH, params->cnv_params.src_width,
+			CNV_SOURCE_HEIGHT, params->cnv_params.src_height);
+
+	/* source size is not equal the source size, then enable cropping. */
+	if (params->cnv_params.crop_en) {
+		REG_UPDATE(CNV_MODE, CNV_WINDOW_CROP_EN, 1);
+		REG_UPDATE(CNV_WINDOW_START, CNV_WINDOW_START_X, params->cnv_params.crop_x);
+		REG_UPDATE(CNV_WINDOW_START, CNV_WINDOW_START_Y, params->cnv_params.crop_y);
+		REG_UPDATE(CNV_WINDOW_SIZE,  CNV_WINDOW_WIDTH,   params->cnv_params.crop_width);
+		REG_UPDATE(CNV_WINDOW_SIZE,  CNV_WINDOW_HEIGHT,  params->cnv_params.crop_height);
+	} else {
+		REG_UPDATE(CNV_MODE, CNV_WINDOW_CROP_EN, 0);
+	}
+
+	/* Set CAPTURE_RATE */
+	REG_UPDATE(CNV_MODE, CNV_FRAME_CAPTURE_RATE, params->capture_rate);
+
+	/* Set CNV output pixel depth */
+	REG_UPDATE(CNV_MODE, CNV_OUT_BPC, params->cnv_params.cnv_out_bpc);
+}
+
+static bool dwb2_enable(struct dwbc *dwbc, struct dc_dwb_params *params)
+{
+	struct dcn20_dwbc *dwbc20 = TO_DCN20_DWBC(dwbc);
+
+	/* Only chroma scaling (sub-sampling) is supported in DCN2 */
+	if ((params->cnv_params.src_width  != params->dest_width) ||
+	    (params->cnv_params.src_height != params->dest_height)) {
+
+		DC_LOG_DWB("%s inst = %d, FAILED!LUMA SCALING NOT SUPPORTED", __func__, dwbc20->base.inst);
+		return false;
+	}
+	DC_LOG_DWB("%s inst = %d, ENABLED", __func__, dwbc20->base.inst);
+
+	/* disable power gating */
+	//REG_UPDATE_5(WB_EC_CONFIG, DISPCLK_R_WB_GATE_DIS, 1,
+	//			 DISPCLK_G_WB_GATE_DIS, 1, DISPCLK_G_WBSCL_GATE_DIS, 1,
+	//			 WB_LB_LS_DIS, 1, WB_LUT_LS_DIS, 1);
+
+	/* Set WB_ENABLE (not double buffered; capture not enabled) */
+	REG_UPDATE(WB_ENABLE, WB_ENABLE, 1);
+
+	/* Set CNV parameters */
+	dwb2_config_dwb_cnv(dwbc, params);
+
+	/* Set scaling parameters */
+	dwb2_set_scaler(dwbc, params);
+
+	/* Enable DWB capture enable (double buffered) */
+	REG_UPDATE(CNV_MODE, CNV_FRAME_CAPTURE_EN, DWB_FRAME_CAPTURE_ENABLE);
+
+	// disable warmup
+	REG_UPDATE(WB_WARM_UP_MODE_CTL1, GMC_WARM_UP_ENABLE, 0);
+
+	return true;
+}
+
+bool dwb2_disable(struct dwbc *dwbc)
+{
+	struct dcn20_dwbc *dwbc20 = TO_DCN20_DWBC(dwbc);
+	DC_LOG_DWB("%s inst = %d, Disabled", __func__, dwbc20->base.inst);
+
+	/* disable CNV */
+	REG_UPDATE(CNV_MODE, CNV_FRAME_CAPTURE_EN, DWB_FRAME_CAPTURE_DISABLE);
+
+	/* disable WB */
+	REG_UPDATE(WB_ENABLE, WB_ENABLE, 0);
+
+	/* soft reset */
+	REG_UPDATE(WB_SOFT_RESET, WB_SOFT_RESET, 1);
+	REG_UPDATE(WB_SOFT_RESET, WB_SOFT_RESET, 0);
+
+	/* enable power gating */
+	//REG_UPDATE_5(WB_EC_CONFIG, DISPCLK_R_WB_GATE_DIS, 0,
+	//			 DISPCLK_G_WB_GATE_DIS, 0, DISPCLK_G_WBSCL_GATE_DIS, 0,
+	//			 WB_LB_LS_DIS, 0, WB_LUT_LS_DIS, 0);
+
+	return true;
+}
+
+static bool dwb2_update(struct dwbc *dwbc, struct dc_dwb_params *params)
+{
+	struct dcn20_dwbc *dwbc20 = TO_DCN20_DWBC(dwbc);
+	unsigned int pre_locked;
+
+	/* Only chroma scaling (sub-sampling) is supported in DCN2 */
+	if ((params->cnv_params.src_width != params->dest_width) ||
+			(params->cnv_params.src_height != params->dest_height)) {
+		DC_LOG_DWB("%s inst = %d, FAILED!LUMA SCALING NOT SUPPORTED", __func__, dwbc20->base.inst);
+		return false;
+	}
+	DC_LOG_DWB("%s inst = %d, scaling", __func__, dwbc20->base.inst);
+
+	/*
+	 * Check if the caller has already locked CNV registers.
+	 * If so: assume the caller will unlock, so don't touch the lock.
+	 * If not: lock them for this update, then unlock after the
+	 * update is complete.
+	 */
+	REG_GET(CNV_UPDATE, CNV_UPDATE_LOCK, &pre_locked);
+
+	if (pre_locked == 0) {
+		/* Lock DWB registers */
+		REG_UPDATE(CNV_UPDATE, CNV_UPDATE_LOCK, 1);
+	}
+
+	/* Set CNV parameters */
+	dwb2_config_dwb_cnv(dwbc, params);
+
+	/* Set scaling parameters */
+	dwb2_set_scaler(dwbc, params);
+
+	if (pre_locked == 0) {
+		/* Unlock DWB registers */
+		REG_UPDATE(CNV_UPDATE, CNV_UPDATE_LOCK, 0);
+	}
+
+	return true;
+}
+
+bool dwb2_is_enabled(struct dwbc *dwbc)
+{
+	struct dcn20_dwbc *dwbc20 = TO_DCN20_DWBC(dwbc);
+	unsigned int wb_enabled = 0;
+	unsigned int cnv_frame_capture_en = 0;
+
+	REG_GET(WB_ENABLE, WB_ENABLE, &wb_enabled);
+	REG_GET(CNV_MODE, CNV_FRAME_CAPTURE_EN, &cnv_frame_capture_en);
+
+	return ((wb_enabled != 0) && (cnv_frame_capture_en != 0));
+}
+
+void dwb2_set_stereo(struct dwbc *dwbc,
+		struct dwb_stereo_params *stereo_params)
+{
+	struct dcn20_dwbc *dwbc20 = TO_DCN20_DWBC(dwbc);
+	DC_LOG_DWB("%s inst = %d, enabled =%d", __func__,\
+		dwbc20->base.inst, stereo_params->stereo_enabled);
+
+	if (stereo_params->stereo_enabled) {
+		REG_UPDATE(CNV_MODE, CNV_STEREO_TYPE,     stereo_params->stereo_type);
+		REG_UPDATE(CNV_MODE, CNV_EYE_SELECTION,   stereo_params->stereo_eye_select);
+		REG_UPDATE(CNV_MODE, CNV_STEREO_POLARITY, stereo_params->stereo_polarity);
+	} else {
+		REG_UPDATE(CNV_MODE, CNV_EYE_SELECTION, 0);
+	}
+}
+
+void dwb2_set_new_content(struct dwbc *dwbc,
+						bool is_new_content)
+{
+	struct dcn20_dwbc *dwbc20 = TO_DCN20_DWBC(dwbc);
+	DC_LOG_DWB("%s inst = %d", __func__, dwbc20->base.inst);
+
+	REG_UPDATE(CNV_MODE, CNV_NEW_CONTENT, is_new_content);
+}
+
+static void dwb2_set_warmup(struct dwbc *dwbc,
+		struct dwb_warmup_params *warmup_params)
+{
+	struct dcn20_dwbc *dwbc20 = TO_DCN20_DWBC(dwbc);
+	DC_LOG_DWB("%s inst = %d", __func__, dwbc20->base.inst);
+
+	REG_UPDATE(WB_WARM_UP_MODE_CTL1, GMC_WARM_UP_ENABLE, warmup_params->warmup_en);
+	REG_UPDATE(WB_WARM_UP_MODE_CTL1, WIDTH_WARMUP, warmup_params->warmup_width);
+	REG_UPDATE(WB_WARM_UP_MODE_CTL1, HEIGHT_WARMUP, warmup_params->warmup_height);
+
+	REG_UPDATE(WB_WARM_UP_MODE_CTL2, DATA_VALUE_WARMUP, warmup_params->warmup_data);
+	REG_UPDATE(WB_WARM_UP_MODE_CTL2, MODE_WARMUP, warmup_params->warmup_mode);
+	REG_UPDATE(WB_WARM_UP_MODE_CTL2, DATA_DEPTH_WARMUP, warmup_params->warmup_depth);
+}
+
+void dwb2_set_scaler(struct dwbc *dwbc, struct dc_dwb_params *params)
+{
+	struct dcn20_dwbc *dwbc20 = TO_DCN20_DWBC(dwbc);
+	DC_LOG_DWB("%s inst = %d", __func__, dwbc20->base.inst);
+
+	/* Program scaling mode */
+	REG_UPDATE_2(WBSCL_MODE, WBSCL_MODE, params->out_format,
+			WBSCL_OUT_BIT_DEPTH, params->output_depth);
+
+	if (params->out_format != dwb_scaler_mode_bypass444) {
+		/* Program output size */
+		REG_UPDATE(WBSCL_DEST_SIZE, WBSCL_DEST_WIDTH,	params->dest_width);
+		REG_UPDATE(WBSCL_DEST_SIZE, WBSCL_DEST_HEIGHT,	params->dest_height);
+
+		/* Program round offsets */
+		REG_UPDATE(WBSCL_ROUND_OFFSET, WBSCL_ROUND_OFFSET_Y_RGB, 0x40);
+		REG_UPDATE(WBSCL_ROUND_OFFSET, WBSCL_ROUND_OFFSET_CBCR,  0x200);
+
+		/* Program clamp values */
+		REG_UPDATE(WBSCL_CLAMP_Y_RGB,	WBSCL_CLAMP_UPPER_Y_RGB,	0x3fe);
+		REG_UPDATE(WBSCL_CLAMP_Y_RGB,	WBSCL_CLAMP_LOWER_Y_RGB,	0x1);
+		REG_UPDATE(WBSCL_CLAMP_CBCR,	WBSCL_CLAMP_UPPER_CBCR,		0x3fe);
+		REG_UPDATE(WBSCL_CLAMP_CBCR,	WBSCL_CLAMP_LOWER_CBCR,		0x1);
+
+		/* Program outside pixel strategy to use edge pixels */
+		REG_UPDATE(WBSCL_OUTSIDE_PIX_STRATEGY, WBSCL_OUTSIDE_PIX_STRATEGY, DWB_OUTSIDE_PIX_STRATEGY_EDGE);
+
+		if (params->cnv_params.crop_en) {
+			/* horizontal scale */
+			dwb_program_horz_scalar(dwbc20, params->cnv_params.crop_width,
+							params->dest_width,
+							params->scaler_taps);
+
+			/* vertical scale */
+			dwb_program_vert_scalar(dwbc20, params->cnv_params.crop_height,
+							params->dest_height,
+							params->scaler_taps,
+							params->subsample_position);
+		} else {
+			/* horizontal scale */
+			dwb_program_horz_scalar(dwbc20, params->cnv_params.src_width,
+							params->dest_width,
+							params->scaler_taps);
+
+			/* vertical scale */
+			dwb_program_vert_scalar(dwbc20, params->cnv_params.src_height,
+							params->dest_height,
+							params->scaler_taps,
+							params->subsample_position);
+		}
+	}
+
+
+	if (dwbc20->dwbc_mask->WBSCL_COEF_RAM_SEL) {
+		/* Swap double buffered coefficient set */
+		uint32_t wbscl_mode = REG_READ(WBSCL_MODE);
+		bool coef_ram_current = get_reg_field_value_ex(
+			wbscl_mode, dwbc20->dwbc_mask->WBSCL_COEF_RAM_SEL_CURRENT,
+			dwbc20->dwbc_shift->WBSCL_COEF_RAM_SEL_CURRENT);
+
+		REG_UPDATE(WBSCL_MODE, WBSCL_COEF_RAM_SEL, !coef_ram_current);
+	}
+
+}
+
+static const struct dwbc_funcs dcn20_dwbc_funcs = {
+	.get_caps		= dwb2_get_caps,
+	.enable			= dwb2_enable,
+	.disable		= dwb2_disable,
+	.update			= dwb2_update,
+	.is_enabled		= dwb2_is_enabled,
+	.set_stereo		= dwb2_set_stereo,
+	.set_new_content	= dwb2_set_new_content,
+	.set_warmup		= dwb2_set_warmup,
+	.dwb_set_scaler		= dwb2_set_scaler,
+};
+
+void dcn20_dwbc_construct(struct dcn20_dwbc *dwbc20,
+		struct dc_context *ctx,
+		const struct dcn20_dwbc_registers *dwbc_regs,
+		const struct dcn20_dwbc_shift *dwbc_shift,
+		const struct dcn20_dwbc_mask *dwbc_mask,
+		int inst)
+{
+	dwbc20->base.ctx = ctx;
+
+	dwbc20->base.inst = inst;
+	dwbc20->base.funcs = &dcn20_dwbc_funcs;
+
+	dwbc20->dwbc_regs = dwbc_regs;
+	dwbc20->dwbc_shift = dwbc_shift;
+	dwbc20->dwbc_mask = dwbc_mask;
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dwb.h b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dwb.h
new file mode 100644
index 000000000000..a9dd9ae23ec9
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dwb.h
@@ -0,0 +1,431 @@
+/* Copyright 2012-17 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+#ifndef __DC_DWBC_DCN20_H__
+#define __DC_DWBC_DCN20_H__
+
+#define TO_DCN20_DWBC(dwbc_base) \
+	container_of(dwbc_base, struct dcn20_dwbc, base)
+
+#define DWBC_COMMON_REG_LIST_DCN2_0(inst) \
+	SRI2_DWB(WB_ENABLE, CNV, inst),\
+	SRI2_DWB(WB_EC_CONFIG, CNV, inst),\
+	SRI2_DWB(CNV_MODE, CNV, inst),\
+	SRI2_DWB(CNV_WINDOW_START, CNV, inst),\
+	SRI2_DWB(CNV_WINDOW_SIZE, CNV, inst),\
+	SRI2_DWB(CNV_UPDATE, CNV, inst),\
+	SRI2_DWB(CNV_SOURCE_SIZE, CNV, inst),\
+	SRI2_DWB(CNV_TEST_CNTL, CNV, inst),\
+	SRI2_DWB(CNV_TEST_CRC_RED, CNV, inst),\
+	SRI2_DWB(CNV_TEST_CRC_GREEN, CNV, inst),\
+	SRI2_DWB(CNV_TEST_CRC_BLUE, CNV, inst),\
+	SRI2_DWB(WBSCL_COEF_RAM_SELECT, WBSCL, inst),\
+	SRI2_DWB(WBSCL_COEF_RAM_TAP_DATA, WBSCL, inst),\
+	SRI2_DWB(WBSCL_MODE, WBSCL, inst),\
+	SRI2_DWB(WBSCL_TAP_CONTROL, WBSCL, inst),\
+	SRI2_DWB(WBSCL_DEST_SIZE, WBSCL, inst),\
+	SRI2_DWB(WBSCL_HORZ_FILTER_SCALE_RATIO, WBSCL, inst),\
+	SRI2_DWB(WBSCL_HORZ_FILTER_INIT_Y_RGB, WBSCL, inst),\
+	SRI2_DWB(WBSCL_HORZ_FILTER_INIT_CBCR, WBSCL, inst),\
+	SRI2_DWB(WBSCL_VERT_FILTER_SCALE_RATIO, WBSCL, inst),\
+	SRI2_DWB(WBSCL_VERT_FILTER_INIT_Y_RGB, WBSCL, inst),\
+	SRI2_DWB(WBSCL_VERT_FILTER_INIT_CBCR, WBSCL, inst),\
+	SRI2_DWB(WBSCL_ROUND_OFFSET, WBSCL, inst),\
+	SRI2_DWB(WBSCL_OVERFLOW_STATUS, WBSCL, inst),\
+	SRI2_DWB(WBSCL_COEF_RAM_CONFLICT_STATUS, WBSCL, inst),\
+	SRI2_DWB(WBSCL_TEST_CNTL, WBSCL, inst),\
+	SRI2_DWB(WBSCL_TEST_CRC_RED, WBSCL, inst),\
+	SRI2_DWB(WBSCL_TEST_CRC_GREEN, WBSCL, inst),\
+	SRI2_DWB(WBSCL_TEST_CRC_BLUE, WBSCL, inst),\
+	SRI2_DWB(WBSCL_BACKPRESSURE_CNT_EN, WBSCL, inst),\
+	SRI2_DWB(WB_MCIF_BACKPRESSURE_CNT, WBSCL, inst),\
+	SRI2_DWB(WBSCL_CLAMP_Y_RGB, WBSCL, inst),\
+	SRI2_DWB(WBSCL_CLAMP_CBCR, WBSCL, inst),\
+	SRI2_DWB(WBSCL_OUTSIDE_PIX_STRATEGY, WBSCL, inst),\
+	SRI2_DWB(WBSCL_OUTSIDE_PIX_STRATEGY_CBCR, WBSCL, inst),\
+	SRI2_DWB(WBSCL_DEBUG, WBSCL, inst),\
+	SRI2_DWB(WBSCL_TEST_DEBUG_INDEX, WBSCL, inst),\
+	SRI2_DWB(WBSCL_TEST_DEBUG_DATA, WBSCL, inst),\
+	SRI2_DWB(WB_DEBUG_CTRL, CNV, inst),\
+	SRI2_DWB(WB_DBG_MODE, CNV, inst),\
+	SRI2_DWB(WB_HW_DEBUG, CNV, inst),\
+	SRI2_DWB(CNV_TEST_DEBUG_INDEX, CNV, inst),\
+	SRI2_DWB(CNV_TEST_DEBUG_DATA, CNV, inst),\
+	SRI2_DWB(WB_SOFT_RESET, CNV, inst),\
+	SRI2_DWB(WB_WARM_UP_MODE_CTL1, CNV, inst),\
+	SRI2_DWB(WB_WARM_UP_MODE_CTL2, CNV, inst)
+
+#define DWBC_COMMON_MASK_SH_LIST_DCN2_0(mask_sh) \
+	SF_DWB(WB_ENABLE, WB_ENABLE, mask_sh),\
+	SF_DWB(WB_EC_CONFIG, DISPCLK_R_WB_GATE_DIS, mask_sh),\
+	SF_DWB(WB_EC_CONFIG, DISPCLK_G_WB_GATE_DIS, mask_sh),\
+	SF_DWB(WB_EC_CONFIG, DISPCLK_G_WBSCL_GATE_DIS, mask_sh),\
+	SF_DWB(WB_EC_CONFIG, WB_TEST_CLK_SEL, mask_sh),\
+	SF_DWB(WB_EC_CONFIG, WB_LB_LS_DIS, mask_sh),\
+	SF_DWB(WB_EC_CONFIG, WB_LB_SD_DIS, mask_sh),\
+	SF_DWB(WB_EC_CONFIG, WB_LUT_LS_DIS, mask_sh),\
+	SF_DWB(WB_EC_CONFIG, WBSCL_LB_MEM_PWR_MODE_SEL, mask_sh),\
+	SF_DWB(WB_EC_CONFIG, WBSCL_LB_MEM_PWR_DIS, mask_sh),\
+	SF_DWB(WB_EC_CONFIG, WBSCL_LB_MEM_PWR_FORCE, mask_sh),\
+	SF_DWB(WB_EC_CONFIG, WBSCL_LB_MEM_PWR_STATE, mask_sh),\
+	SF_DWB(WB_EC_CONFIG, WB_RAM_PW_SAVE_MODE, mask_sh),\
+	SF_DWB(WB_EC_CONFIG, WBSCL_LUT_MEM_PWR_STATE, mask_sh),\
+	SF_DWB(CNV_MODE, CNV_OUT_BPC, mask_sh),\
+	SF_DWB(CNV_MODE, CNV_FRAME_CAPTURE_RATE, mask_sh),\
+	SF_DWB(CNV_MODE, CNV_WINDOW_CROP_EN, mask_sh),\
+	SF_DWB(CNV_MODE, CNV_STEREO_TYPE, mask_sh),\
+	SF_DWB(CNV_MODE, CNV_INTERLACED_MODE, mask_sh),\
+	SF_DWB(CNV_MODE, CNV_EYE_SELECTION, mask_sh),\
+	SF_DWB(CNV_MODE, CNV_STEREO_POLARITY, mask_sh),\
+	SF_DWB(CNV_MODE, CNV_INTERLACED_FIELD_ORDER, mask_sh),\
+	SF_DWB(CNV_MODE, CNV_STEREO_SPLIT, mask_sh),\
+	SF_DWB(CNV_MODE, CNV_NEW_CONTENT, mask_sh),\
+	SF_DWB(CNV_MODE, CNV_FRAME_CAPTURE_EN_CURRENT, mask_sh),\
+	SF_DWB(CNV_MODE, CNV_FRAME_CAPTURE_EN, mask_sh),\
+	SF_DWB(CNV_WINDOW_START, CNV_WINDOW_START_X, mask_sh),\
+	SF_DWB(CNV_WINDOW_START, CNV_WINDOW_START_Y, mask_sh),\
+	SF_DWB(CNV_WINDOW_SIZE, CNV_WINDOW_WIDTH, mask_sh),\
+	SF_DWB(CNV_WINDOW_SIZE, CNV_WINDOW_HEIGHT, mask_sh),\
+	SF_DWB(CNV_UPDATE, CNV_UPDATE_PENDING, mask_sh),\
+	SF_DWB(CNV_UPDATE, CNV_UPDATE_TAKEN, mask_sh),\
+	SF_DWB(CNV_UPDATE, CNV_UPDATE_LOCK, mask_sh),\
+	SF_DWB(CNV_SOURCE_SIZE, CNV_SOURCE_WIDTH, mask_sh),\
+	SF_DWB(CNV_SOURCE_SIZE, CNV_SOURCE_HEIGHT, mask_sh),\
+	SF_DWB(CNV_TEST_CNTL, CNV_TEST_CRC_EN, mask_sh),\
+	SF_DWB(CNV_TEST_CNTL, CNV_TEST_CRC_CONT_EN, mask_sh),\
+	SF_DWB(CNV_TEST_CRC_RED, CNV_TEST_CRC_RED_MASK, mask_sh),\
+	SF_DWB(CNV_TEST_CRC_RED, CNV_TEST_CRC_SIG_RED, mask_sh),\
+	SF_DWB(CNV_TEST_CRC_GREEN, CNV_TEST_CRC_GREEN_MASK, mask_sh),\
+	SF_DWB(CNV_TEST_CRC_GREEN, CNV_TEST_CRC_SIG_GREEN, mask_sh),\
+	SF_DWB(CNV_TEST_CRC_BLUE, CNV_TEST_CRC_BLUE_MASK, mask_sh),\
+	SF_DWB(CNV_TEST_CRC_BLUE, CNV_TEST_CRC_SIG_BLUE, mask_sh),\
+	SF_DWB(WB_DEBUG_CTRL, WB_DEBUG_EN, mask_sh),\
+	SF_DWB(WB_DEBUG_CTRL, WB_DEBUG_SEL, mask_sh),\
+	SF_DWB(WB_DBG_MODE, WB_DBG_MODE_EN, mask_sh),\
+	SF_DWB(WB_DBG_MODE, WB_DBG_DIN_FMT, mask_sh),\
+	SF_DWB(WB_DBG_MODE, WB_DBG_36MODE, mask_sh),\
+	SF_DWB(WB_DBG_MODE, WB_DBG_CMAP, mask_sh),\
+	SF_DWB(WB_DBG_MODE, WB_DBG_PXLRATE_ERROR, mask_sh),\
+	SF_DWB(WB_DBG_MODE, WB_DBG_SOURCE_WIDTH, mask_sh),\
+	SF_DWB(WB_HW_DEBUG, WB_HW_DEBUG, mask_sh),\
+	SF_DWB(WB_SOFT_RESET, WB_SOFT_RESET, mask_sh),\
+	SF_DWB(CNV_TEST_DEBUG_INDEX, CNV_TEST_DEBUG_INDEX, mask_sh),\
+	SF_DWB(CNV_TEST_DEBUG_INDEX, CNV_TEST_DEBUG_WRITE_EN, mask_sh),\
+	SF_DWB(CNV_TEST_DEBUG_DATA, CNV_TEST_DEBUG_DATA, mask_sh),\
+	SF_DWB(WBSCL_COEF_RAM_SELECT, WBSCL_COEF_RAM_TAP_PAIR_IDX, mask_sh),\
+	SF_DWB(WBSCL_COEF_RAM_SELECT, WBSCL_COEF_RAM_PHASE, mask_sh),\
+	SF_DWB(WBSCL_COEF_RAM_SELECT, WBSCL_COEF_RAM_FILTER_TYPE, mask_sh),\
+	SF_DWB(WBSCL_COEF_RAM_TAP_DATA, WBSCL_COEF_RAM_EVEN_TAP_COEF, mask_sh),\
+	SF_DWB(WBSCL_COEF_RAM_TAP_DATA, WBSCL_COEF_RAM_EVEN_TAP_COEF_EN, mask_sh),\
+	SF_DWB(WBSCL_COEF_RAM_TAP_DATA, WBSCL_COEF_RAM_ODD_TAP_COEF, mask_sh),\
+	SF_DWB(WBSCL_COEF_RAM_TAP_DATA, WBSCL_COEF_RAM_ODD_TAP_COEF_EN, mask_sh),\
+	SF_DWB(WBSCL_MODE, WBSCL_MODE, mask_sh),\
+	SF_DWB(WBSCL_MODE, WBSCL_OUT_BIT_DEPTH, mask_sh),\
+	SF_DWB(WBSCL_TAP_CONTROL, WBSCL_V_NUM_OF_TAPS_Y_RGB, mask_sh),\
+	SF_DWB(WBSCL_TAP_CONTROL, WBSCL_V_NUM_OF_TAPS_CBCR, mask_sh),\
+	SF_DWB(WBSCL_TAP_CONTROL, WBSCL_H_NUM_OF_TAPS_Y_RGB, mask_sh),\
+	SF_DWB(WBSCL_TAP_CONTROL, WBSCL_H_NUM_OF_TAPS_CBCR, mask_sh),\
+	SF_DWB(WBSCL_DEST_SIZE, WBSCL_DEST_HEIGHT, mask_sh),\
+	SF_DWB(WBSCL_DEST_SIZE, WBSCL_DEST_WIDTH, mask_sh),\
+	SF_DWB(WBSCL_HORZ_FILTER_SCALE_RATIO, WBSCL_H_SCALE_RATIO, mask_sh),\
+	SF_DWB(WBSCL_HORZ_FILTER_INIT_Y_RGB, WBSCL_H_INIT_FRAC_Y_RGB, mask_sh),\
+	SF_DWB(WBSCL_HORZ_FILTER_INIT_Y_RGB, WBSCL_H_INIT_INT_Y_RGB, mask_sh),\
+	SF_DWB(WBSCL_HORZ_FILTER_INIT_CBCR, WBSCL_H_INIT_FRAC_CBCR, mask_sh),\
+	SF_DWB(WBSCL_HORZ_FILTER_INIT_CBCR, WBSCL_H_INIT_INT_CBCR, mask_sh),\
+	SF_DWB(WBSCL_VERT_FILTER_SCALE_RATIO, WBSCL_V_SCALE_RATIO, mask_sh),\
+	SF_DWB(WBSCL_VERT_FILTER_INIT_Y_RGB, WBSCL_V_INIT_FRAC_Y_RGB, mask_sh),\
+	SF_DWB(WBSCL_VERT_FILTER_INIT_Y_RGB, WBSCL_V_INIT_INT_Y_RGB, mask_sh),\
+	SF_DWB(WBSCL_VERT_FILTER_INIT_CBCR, WBSCL_V_INIT_FRAC_CBCR, mask_sh),\
+	SF_DWB(WBSCL_VERT_FILTER_INIT_CBCR, WBSCL_V_INIT_INT_CBCR, mask_sh),\
+	SF_DWB(WBSCL_ROUND_OFFSET, WBSCL_ROUND_OFFSET_Y_RGB, mask_sh),\
+	SF_DWB(WBSCL_ROUND_OFFSET, WBSCL_ROUND_OFFSET_CBCR, mask_sh),\
+	SF_DWB(WBSCL_OVERFLOW_STATUS, WBSCL_DATA_OVERFLOW_FLAG, mask_sh),\
+	SF_DWB(WBSCL_OVERFLOW_STATUS, WBSCL_DATA_OVERFLOW_ACK, mask_sh),\
+	SF_DWB(WBSCL_OVERFLOW_STATUS, WBSCL_DATA_OVERFLOW_MASK, mask_sh),\
+	SF_DWB(WBSCL_OVERFLOW_STATUS, WBSCL_DATA_OVERFLOW_INT_STATUS, mask_sh),\
+	SF_DWB(WBSCL_OVERFLOW_STATUS, WBSCL_DATA_OVERFLOW_INT_TYPE, mask_sh),\
+	SF_DWB(WBSCL_COEF_RAM_CONFLICT_STATUS, WBSCL_HOST_CONFLICT_FLAG, mask_sh),\
+	SF_DWB(WBSCL_COEF_RAM_CONFLICT_STATUS, WBSCL_HOST_CONFLICT_ACK, mask_sh),\
+	SF_DWB(WBSCL_COEF_RAM_CONFLICT_STATUS, WBSCL_HOST_CONFLICT_MASK, mask_sh),\
+	SF_DWB(WBSCL_COEF_RAM_CONFLICT_STATUS, WBSCL_HOST_CONFLICT_INT_STATUS, mask_sh),\
+	SF_DWB(WBSCL_COEF_RAM_CONFLICT_STATUS, WBSCL_HOST_CONFLICT_INT_TYPE, mask_sh),\
+	SF_DWB(WBSCL_TEST_CNTL, WBSCL_TEST_CRC_EN, mask_sh),\
+	SF_DWB(WBSCL_TEST_CNTL, WBSCL_TEST_CRC_CONT_EN, mask_sh),\
+	SF_DWB(WBSCL_TEST_CRC_RED, WBSCL_TEST_CRC_RED_MASK, mask_sh),\
+	SF_DWB(WBSCL_TEST_CRC_RED, WBSCL_TEST_CRC_SIG_RED, mask_sh),\
+	SF_DWB(WBSCL_TEST_CRC_GREEN, WBSCL_TEST_CRC_GREEN_MASK, mask_sh),\
+	SF_DWB(WBSCL_TEST_CRC_GREEN, WBSCL_TEST_CRC_SIG_GREEN, mask_sh),\
+	SF_DWB(WBSCL_TEST_CRC_BLUE, WBSCL_TEST_CRC_BLUE_MASK, mask_sh),\
+	SF_DWB(WBSCL_TEST_CRC_BLUE, WBSCL_TEST_CRC_SIG_BLUE, mask_sh),\
+	SF_DWB(WBSCL_BACKPRESSURE_CNT_EN, WBSCL_BACKPRESSURE_CNT_EN, mask_sh),\
+	SF_DWB(WB_MCIF_BACKPRESSURE_CNT, WB_MCIF_Y_MAX_BACKPRESSURE, mask_sh),\
+	SF_DWB(WB_MCIF_BACKPRESSURE_CNT, WB_MCIF_C_MAX_BACKPRESSURE, mask_sh),\
+	SF_DWB(WBSCL_CLAMP_Y_RGB, WBSCL_CLAMP_UPPER_Y_RGB, mask_sh),\
+	SF_DWB(WBSCL_CLAMP_Y_RGB, WBSCL_CLAMP_LOWER_Y_RGB, mask_sh),\
+	SF_DWB(WBSCL_CLAMP_CBCR, WBSCL_CLAMP_UPPER_CBCR, mask_sh),\
+	SF_DWB(WBSCL_CLAMP_CBCR, WBSCL_CLAMP_LOWER_CBCR, mask_sh),\
+	SF_DWB(WBSCL_OUTSIDE_PIX_STRATEGY, WBSCL_OUTSIDE_PIX_STRATEGY, mask_sh),\
+	SF_DWB(WBSCL_OUTSIDE_PIX_STRATEGY, WBSCL_BLACK_COLOR_G_Y, mask_sh),\
+	SF_DWB(WBSCL_OUTSIDE_PIX_STRATEGY_CBCR, WBSCL_BLACK_COLOR_B_CB, mask_sh),\
+	SF_DWB(WBSCL_OUTSIDE_PIX_STRATEGY_CBCR, WBSCL_BLACK_COLOR_R_CR, mask_sh),\
+	SF_DWB(WBSCL_DEBUG, WBSCL_DEBUG, mask_sh),\
+	SF_DWB(WBSCL_TEST_DEBUG_INDEX, WBSCL_TEST_DEBUG_INDEX, mask_sh),\
+	SF_DWB(WBSCL_TEST_DEBUG_INDEX, WBSCL_TEST_DEBUG_WRITE_EN, mask_sh),\
+	SF_DWB(WBSCL_TEST_DEBUG_DATA, WBSCL_TEST_DEBUG_DATA, mask_sh),\
+	SF_DWB(WB_WARM_UP_MODE_CTL1, WIDTH_WARMUP, mask_sh),\
+	SF_DWB(WB_WARM_UP_MODE_CTL1, HEIGHT_WARMUP, mask_sh),\
+	SF_DWB(WB_WARM_UP_MODE_CTL1, GMC_WARM_UP_ENABLE, mask_sh),\
+	SF_DWB(WB_WARM_UP_MODE_CTL2, DATA_VALUE_WARMUP, mask_sh),\
+	SF_DWB(WB_WARM_UP_MODE_CTL2, MODE_WARMUP, mask_sh),\
+	SF_DWB(WB_WARM_UP_MODE_CTL2, DATA_DEPTH_WARMUP, mask_sh)
+
+#define DWBC_REG_FIELD_LIST_DCN2_0(type) \
+	type WB_ENABLE;\
+	type DISPCLK_R_WB_GATE_DIS;\
+	type DISPCLK_G_WB_GATE_DIS;\
+	type DISPCLK_G_WBSCL_GATE_DIS;\
+	type WB_TEST_CLK_SEL;\
+	type WB_LB_LS_DIS;\
+	type WB_LB_SD_DIS;\
+	type WB_LUT_LS_DIS;\
+	type WBSCL_LB_MEM_PWR_MODE_SEL;\
+	type WBSCL_LB_MEM_PWR_DIS;\
+	type WBSCL_LB_MEM_PWR_FORCE;\
+	type WBSCL_LB_MEM_PWR_STATE;\
+	type WB_RAM_PW_SAVE_MODE;\
+	type WBSCL_LUT_MEM_PWR_STATE;\
+	type CNV_OUT_BPC;\
+	type CNV_FRAME_CAPTURE_RATE;\
+	type CNV_WINDOW_CROP_EN;\
+	type CNV_STEREO_TYPE;\
+	type CNV_INTERLACED_MODE;\
+	type CNV_EYE_SELECTION;\
+	type CNV_STEREO_POLARITY;\
+	type CNV_INTERLACED_FIELD_ORDER;\
+	type CNV_STEREO_SPLIT;\
+	type CNV_NEW_CONTENT;\
+	type CNV_FRAME_CAPTURE_EN_CURRENT;\
+	type CNV_FRAME_CAPTURE_EN;\
+	type CNV_WINDOW_START_X;\
+	type CNV_WINDOW_START_Y;\
+	type CNV_WINDOW_WIDTH;\
+	type CNV_WINDOW_HEIGHT;\
+	type CNV_UPDATE_PENDING;\
+	type CNV_UPDATE_TAKEN;\
+	type CNV_UPDATE_LOCK;\
+	type CNV_SOURCE_WIDTH;\
+	type CNV_SOURCE_HEIGHT;\
+	type CNV_TEST_CRC_EN;\
+	type CNV_TEST_CRC_CONT_EN;\
+	type CNV_TEST_CRC_RED_MASK;\
+	type CNV_TEST_CRC_SIG_RED;\
+	type CNV_TEST_CRC_GREEN_MASK;\
+	type CNV_TEST_CRC_SIG_GREEN;\
+	type CNV_TEST_CRC_BLUE_MASK;\
+	type CNV_TEST_CRC_SIG_BLUE;\
+	type WB_DEBUG_EN;\
+	type WB_DEBUG_SEL;\
+	type WB_DBG_MODE_EN;\
+	type WB_DBG_DIN_FMT;\
+	type WB_DBG_36MODE;\
+	type WB_DBG_CMAP;\
+	type WB_DBG_PXLRATE_ERROR;\
+	type WB_DBG_SOURCE_WIDTH;\
+	type WB_HW_DEBUG;\
+	type CNV_TEST_DEBUG_INDEX;\
+	type CNV_TEST_DEBUG_WRITE_EN;\
+	type CNV_TEST_DEBUG_DATA;\
+	type WB_SOFT_RESET;\
+	type WBSCL_COEF_RAM_TAP_PAIR_IDX;\
+	type WBSCL_COEF_RAM_PHASE;\
+	type WBSCL_COEF_RAM_FILTER_TYPE;\
+	type WBSCL_COEF_RAM_SEL;\
+	type WBSCL_COEF_RAM_SEL_CURRENT;\
+	type WBSCL_COEF_RAM_RD_SEL;\
+	type WBSCL_COEF_RAM_EVEN_TAP_COEF;\
+	type WBSCL_COEF_RAM_EVEN_TAP_COEF_EN;\
+	type WBSCL_COEF_RAM_ODD_TAP_COEF;\
+	type WBSCL_COEF_RAM_ODD_TAP_COEF_EN;\
+	type WBSCL_MODE;\
+	type WBSCL_OUT_BIT_DEPTH;\
+	type WBSCL_V_NUM_OF_TAPS_Y_RGB;\
+	type WBSCL_V_NUM_OF_TAPS_CBCR;\
+	type WBSCL_H_NUM_OF_TAPS_Y_RGB;\
+	type WBSCL_H_NUM_OF_TAPS_CBCR;\
+	type WBSCL_DEST_HEIGHT;\
+	type WBSCL_DEST_WIDTH;\
+	type WBSCL_H_SCALE_RATIO;\
+	type WBSCL_H_INIT_FRAC_Y_RGB;\
+	type WBSCL_H_INIT_INT_Y_RGB;\
+	type WBSCL_H_INIT_FRAC_CBCR;\
+	type WBSCL_H_INIT_INT_CBCR;\
+	type WBSCL_V_SCALE_RATIO;\
+	type WBSCL_V_INIT_FRAC_Y_RGB;\
+	type WBSCL_V_INIT_INT_Y_RGB;\
+	type WBSCL_V_INIT_FRAC_CBCR;\
+	type WBSCL_V_INIT_INT_CBCR;\
+	type WBSCL_ROUND_OFFSET_Y_RGB;\
+	type WBSCL_ROUND_OFFSET_CBCR;\
+	type WBSCL_DATA_OVERFLOW_FLAG;\
+	type WBSCL_DATA_OVERFLOW_ACK;\
+	type WBSCL_DATA_OVERFLOW_MASK;\
+	type WBSCL_DATA_OVERFLOW_INT_STATUS;\
+	type WBSCL_DATA_OVERFLOW_INT_TYPE;\
+	type WBSCL_HOST_CONFLICT_FLAG;\
+	type WBSCL_HOST_CONFLICT_ACK;\
+	type WBSCL_HOST_CONFLICT_MASK;\
+	type WBSCL_HOST_CONFLICT_INT_STATUS;\
+	type WBSCL_HOST_CONFLICT_INT_TYPE;\
+	type WBSCL_TEST_CRC_EN;\
+	type WBSCL_TEST_CRC_CONT_EN;\
+	type WBSCL_TEST_CRC_RED_MASK;\
+	type WBSCL_TEST_CRC_SIG_RED;\
+	type WBSCL_TEST_CRC_GREEN_MASK;\
+	type WBSCL_TEST_CRC_SIG_GREEN;\
+	type WBSCL_TEST_CRC_BLUE_MASK;\
+	type WBSCL_TEST_CRC_SIG_BLUE;\
+	type WBSCL_BACKPRESSURE_CNT_EN;\
+	type WB_MCIF_Y_MAX_BACKPRESSURE;\
+	type WB_MCIF_C_MAX_BACKPRESSURE;\
+	type WBSCL_CLAMP_UPPER_Y_RGB;\
+	type WBSCL_CLAMP_LOWER_Y_RGB;\
+	type WBSCL_CLAMP_UPPER_CBCR;\
+	type WBSCL_CLAMP_LOWER_CBCR;\
+	type WBSCL_OUTSIDE_PIX_STRATEGY;\
+	type WBSCL_BLACK_COLOR_G_Y;\
+	type WBSCL_BLACK_COLOR_B_CB;\
+	type WBSCL_BLACK_COLOR_R_CR;\
+	type WBSCL_DEBUG;\
+	type WBSCL_TEST_DEBUG_INDEX;\
+	type WBSCL_TEST_DEBUG_WRITE_EN;\
+	type WBSCL_TEST_DEBUG_DATA;\
+	type WIDTH_WARMUP;\
+	type HEIGHT_WARMUP;\
+	type GMC_WARM_UP_ENABLE;\
+	type DATA_VALUE_WARMUP;\
+	type MODE_WARMUP;\
+	type DATA_DEPTH_WARMUP; \
+
+struct dcn20_dwbc_registers {
+	/* DCN2.0 */
+	uint32_t WB_ENABLE;
+	uint32_t WB_EC_CONFIG;
+	uint32_t CNV_MODE;
+	uint32_t CNV_WINDOW_START;
+	uint32_t CNV_WINDOW_SIZE;
+	uint32_t CNV_UPDATE;
+	uint32_t CNV_SOURCE_SIZE;
+	uint32_t CNV_TEST_CNTL;
+	uint32_t CNV_TEST_CRC_RED;
+	uint32_t CNV_TEST_CRC_GREEN;
+	uint32_t CNV_TEST_CRC_BLUE;
+	uint32_t WB_DEBUG_CTRL;
+	uint32_t WB_DBG_MODE;
+	uint32_t WB_HW_DEBUG;
+	uint32_t CNV_TEST_DEBUG_INDEX;
+	uint32_t CNV_TEST_DEBUG_DATA;
+	uint32_t WB_SOFT_RESET;
+	uint32_t WBSCL_COEF_RAM_SELECT;
+	uint32_t WBSCL_COEF_RAM_TAP_DATA;
+	uint32_t WBSCL_MODE;
+	uint32_t WBSCL_TAP_CONTROL;
+	uint32_t WBSCL_DEST_SIZE;
+	uint32_t WBSCL_HORZ_FILTER_SCALE_RATIO;
+	uint32_t WBSCL_HORZ_FILTER_INIT_Y_RGB;
+	uint32_t WBSCL_HORZ_FILTER_INIT_CBCR;
+	uint32_t WBSCL_VERT_FILTER_SCALE_RATIO;
+	uint32_t WBSCL_VERT_FILTER_INIT_Y_RGB;
+	uint32_t WBSCL_VERT_FILTER_INIT_CBCR;
+	uint32_t WBSCL_ROUND_OFFSET;
+	uint32_t WBSCL_OVERFLOW_STATUS;
+	uint32_t WBSCL_COEF_RAM_CONFLICT_STATUS;
+	uint32_t WBSCL_TEST_CNTL;
+	uint32_t WBSCL_TEST_CRC_RED;
+	uint32_t WBSCL_TEST_CRC_GREEN;
+	uint32_t WBSCL_TEST_CRC_BLUE;
+	uint32_t WBSCL_BACKPRESSURE_CNT_EN;
+	uint32_t WB_MCIF_BACKPRESSURE_CNT;
+	uint32_t WBSCL_CLAMP_Y_RGB;
+	uint32_t WBSCL_CLAMP_CBCR;
+	uint32_t WBSCL_OUTSIDE_PIX_STRATEGY;
+	uint32_t WBSCL_OUTSIDE_PIX_STRATEGY_CBCR;
+	uint32_t WBSCL_DEBUG;
+	uint32_t WBSCL_TEST_DEBUG_INDEX;
+	uint32_t WBSCL_TEST_DEBUG_DATA;
+	uint32_t WB_WARM_UP_MODE_CTL1;
+	uint32_t WB_WARM_UP_MODE_CTL2;
+};
+
+
+struct dcn20_dwbc_mask {
+	DWBC_REG_FIELD_LIST_DCN2_0(uint32_t)
+};
+
+struct dcn20_dwbc_shift {
+	DWBC_REG_FIELD_LIST_DCN2_0(uint8_t)
+};
+
+struct dcn20_dwbc {
+	struct dwbc base;
+	const struct dcn20_dwbc_registers *dwbc_regs;
+	const struct dcn20_dwbc_shift *dwbc_shift;
+	const struct dcn20_dwbc_mask *dwbc_mask;
+};
+
+void dcn20_dwbc_construct(struct dcn20_dwbc *dwbc20,
+	struct dc_context *ctx,
+	const struct dcn20_dwbc_registers *dwbc_regs,
+	const struct dcn20_dwbc_shift *dwbc_shift,
+	const struct dcn20_dwbc_mask *dwbc_mask,
+	int inst);
+
+bool dwb2_disable(struct dwbc *dwbc);
+
+bool dwb2_is_enabled(struct dwbc *dwbc);
+
+void dwb2_set_stereo(struct dwbc *dwbc,
+	struct dwb_stereo_params *stereo_params);
+
+void dwb2_set_new_content(struct dwbc *dwbc,
+	bool is_new_content);
+
+void dwb2_config_dwb_cnv(struct dwbc *dwbc,
+	struct dc_dwb_params *params);
+
+void dwb2_set_scaler(struct dwbc *dwbc, struct dc_dwb_params *params);
+
+bool dwb_program_vert_scalar(struct dcn20_dwbc *dwbc20,
+	uint32_t src_height,
+	uint32_t dest_height,
+	struct scaling_taps num_taps,
+	enum dwb_subsample_position subsample_position);
+
+bool dwb_program_horz_scalar(struct dcn20_dwbc *dwbc20,
+	uint32_t src_width,
+	uint32_t dest_width,
+	struct scaling_taps num_taps);
+
+
+#endif
+
+
diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dwb_scl.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dwb_scl.c
new file mode 100644
index 000000000000..a0d437f0ce2b
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dwb_scl.c
@@ -0,0 +1,876 @@
+/*
+ * Copyright 2012-17 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+#include "fixed31_32.h"
+#include "resource.h"
+#include "dwb.h"
+#include "dcn20_dwb.h"
+
+#define NUM_PHASES    16
+#define HORZ_MAX_TAPS 12
+#define VERT_MAX_TAPS 12
+
+#define REG(reg)\
+	dwbc20->dwbc_regs->reg
+
+#define CTX \
+	dwbc20->base.ctx
+
+#undef FN
+#define FN(reg_name, field_name) \
+	dwbc20->dwbc_shift->field_name, dwbc20->dwbc_mask->field_name
+
+#define TO_DCN20_DWBC(dwbc_base) \
+	container_of(dwbc_base, struct dcn20_dwbc, base)
+
+
+static const uint16_t filter_3tap_16p_upscale[27] = {
+	2048, 2048, 0,
+	1708, 2424, 16348,
+	1372, 2796, 16308,
+	1056, 3148, 16272,
+	768, 3464, 16244,
+	512, 3728, 16236,
+	296, 3928, 16252,
+	124, 4052, 16296,
+	0, 4096, 0
+};
+
+static const uint16_t filter_3tap_16p_117[27] = {
+	2048, 2048, 0,
+	1824, 2276, 16376,
+	1600, 2496, 16380,
+	1376, 2700, 16,
+	1156, 2880, 52,
+	948, 3032, 108,
+	756, 3144, 192,
+	580, 3212, 296,
+	428, 3236, 428
+};
+
+static const uint16_t filter_3tap_16p_150[27] = {
+	2048, 2048, 0,
+	1872, 2184, 36,
+	1692, 2308, 88,
+	1516, 2420, 156,
+	1340, 2516, 236,
+	1168, 2592, 328,
+	1004, 2648, 440,
+	844, 2684, 560,
+	696, 2696, 696
+};
+
+static const uint16_t filter_3tap_16p_183[27] = {
+	2048, 2048, 0,
+	1892, 2104, 92,
+	1744, 2152, 196,
+	1592, 2196, 300,
+	1448, 2232, 412,
+	1304, 2256, 528,
+	1168, 2276, 648,
+	1032, 2288, 772,
+	900, 2292, 900
+};
+
+static const uint16_t filter_4tap_16p_upscale[36] = {
+	0, 4096, 0, 0,
+	16240, 4056, 180, 16380,
+	16136, 3952, 404, 16364,
+	16072, 3780, 664, 16344,
+	16040, 3556, 952, 16312,
+	16036, 3284, 1268, 16272,
+	16052, 2980, 1604, 16224,
+	16084, 2648, 1952, 16176,
+	16128, 2304, 2304, 16128
+};
+
+static const uint16_t filter_4tap_16p_117[36] = {
+	428, 3236, 428, 0,
+	276, 3232, 604, 16364,
+	148, 3184, 800, 16340,
+	44, 3104, 1016, 16312,
+	16344, 2984, 1244, 16284,
+	16284, 2832, 1488, 16256,
+	16244, 2648, 1732, 16236,
+	16220, 2440, 1976, 16220,
+	16212, 2216, 2216, 16212
+};
+
+static const uint16_t filter_4tap_16p_150[36] = {
+	696, 2700, 696, 0,
+	560, 2700, 848, 16364,
+	436, 2676, 1008, 16348,
+	328, 2628, 1180, 16336,
+	232, 2556, 1356, 16328,
+	152, 2460, 1536, 16328,
+	84, 2344, 1716, 16332,
+	28, 2208, 1888, 16348,
+	16376, 2052, 2052, 16376
+};
+
+static const uint16_t filter_4tap_16p_183[36] = {
+	940, 2208, 940, 0,
+	832, 2200, 1052, 4,
+	728, 2180, 1164, 16,
+	628, 2148, 1280, 36,
+	536, 2100, 1392, 60,
+	448, 2044, 1504, 92,
+	368, 1976, 1612, 132,
+	296, 1900, 1716, 176,
+	232, 1812, 1812, 232
+};
+
+static const uint16_t filter_5tap_16p_upscale[45] = {
+	15936, 2496, 2496, 15936, 0,
+	15992, 2128, 2832, 15896, 12,
+	16056, 1760, 3140, 15876, 24,
+	16120, 1404, 3420, 15876, 36,
+	16188, 1060, 3652, 15908, 44,
+	16248, 744, 3844, 15972, 44,
+	16304, 460, 3980, 16072, 40,
+	16348, 212, 4064, 16208, 24,
+	0, 0, 4096, 0, 0,
+};
+
+static const uint16_t filter_5tap_16p_117[45] = {
+	16056, 2372, 2372, 16056, 0,
+	16052, 2124, 2600, 16076, 0,
+	16060, 1868, 2808, 16120, 0,
+	16080, 1612, 2992, 16180, 16376,
+	16112, 1356, 3144, 16268, 16364,
+	16144, 1108, 3268, 16376, 16344,
+	16184, 872, 3356, 124, 16320,
+	16220, 656, 3412, 276, 16292,
+	16256, 456, 3428, 456, 16256,
+};
+
+static const uint16_t filter_5tap_16p_150[45] = {
+	16368, 2064, 2064, 16368, 0,
+	16316, 1924, 2204, 44, 16372,
+	16280, 1772, 2328, 116, 16356,
+	16256, 1616, 2440, 204, 16340,
+	16240, 1456, 2536, 304, 16320,
+	16232, 1296, 2612, 416, 16300,
+	16232, 1132, 2664, 544, 16284,
+	16240, 976, 2700, 680, 16264,
+	16248, 824, 2708, 824, 16248,
+};
+
+static const uint16_t filter_5tap_16p_183[45] = {
+	228, 1816, 1816, 228, 0,
+	168, 1728, 1904, 300, 16372,
+	116, 1632, 1988, 376, 16360,
+	72, 1528, 2060, 460, 16348,
+	36, 1424, 2120, 552, 16340,
+	4, 1312, 2168, 652, 16336,
+	16368, 1200, 2204, 752, 16332,
+	16352, 1084, 2224, 860, 16332,
+	16340, 972, 2232, 972, 16340,
+};
+
+static const uint16_t filter_6tap_16p_upscale[54] = {
+	0, 0, 4092, 0, 0, 0,
+	44, 16188, 4064, 228, 16324, 0,
+	80, 16036, 3980, 492, 16256, 4,
+	108, 15916, 3844, 788, 16184, 16,
+	120, 15836, 3656, 1108, 16104, 28,
+	128, 15792, 3420, 1448, 16024, 44,
+	124, 15776, 3144, 1800, 15948, 64,
+	112, 15792, 2836, 2152, 15880, 80,
+	100, 15828, 2504, 2504, 15828, 100,
+};
+
+static const uint16_t filter_6tap_16p_117[54] = {
+	16168, 476, 3568, 476, 16168, 0,
+	16216, 280, 3540, 692, 16116, 8,
+	16264, 104, 3472, 924, 16068, 16,
+	16304, 16340, 3372, 1168, 16024, 28,
+	16344, 16212, 3236, 1424, 15988, 36,
+	16372, 16112, 3072, 1680, 15956, 44,
+	12, 16036, 2880, 1936, 15940, 48,
+	28, 15984, 2668, 2192, 15936, 48,
+	40, 15952, 2436, 2436, 15952, 40,
+};
+
+static const uint16_t filter_6tap_16p_150[54] = {
+	16148, 920, 2724, 920, 16148, 0,
+	16156, 768, 2712, 1072, 16144, 0,
+	16172, 628, 2684, 1232, 16148, 16380,
+	16192, 492, 2632, 1388, 16160, 16372,
+	16212, 368, 2564, 1548, 16180, 16364,
+	16232, 256, 2480, 1704, 16212, 16352,
+	16256, 156, 2380, 1856, 16256, 16336,
+	16276, 64, 2268, 2004, 16308, 16320,
+	16300, 16372, 2140, 2140, 16372, 16300,
+};
+
+static const uint16_t filter_6tap_16p_183[54] = {
+	16296, 1032, 2196, 1032, 16296, 0,
+	16284, 924, 2196, 1144, 16320, 16376,
+	16272, 820, 2180, 1256, 16348, 16364,
+	16268, 716, 2156, 1364, 16380, 16352,
+	16264, 620, 2116, 1472, 36, 16340,
+	16268, 524, 2068, 1576, 88, 16328,
+	16272, 436, 2008, 1680, 144, 16316,
+	16280, 352, 1940, 1772, 204, 16304,
+	16292, 276, 1860, 1860, 276, 16292,
+};
+
+static const uint16_t filter_7tap_16p_upscale[63] = {
+	176, 15760, 2488, 2488, 15760, 176, 0,
+	160, 15812, 2152, 2816, 15728, 192, 16376,
+	136, 15884, 1812, 3124, 15720, 196, 16368,
+	108, 15964, 1468, 3400, 15740, 196, 16364,
+	84, 16048, 1132, 3640, 15792, 180, 16360,
+	56, 16140, 812, 3832, 15884, 152, 16360,
+	32, 16228, 512, 3976, 16012, 116, 16364,
+	12, 16308, 240, 4064, 16180, 60, 16372,
+	0, 0, 0, 4096, 0, 0, 0,
+};
+
+static const uint16_t filter_7tap_16p_117[63] = {
+	92, 15868, 2464, 2464, 15868, 92, 0,
+	108, 15852, 2216, 2700, 15904, 72, 0,
+	112, 15856, 1960, 2916, 15964, 44, 0,
+	116, 15876, 1696, 3108, 16048, 8, 8,
+	112, 15908, 1428, 3268, 16156, 16348, 12,
+	104, 15952, 1168, 3400, 16288, 16300, 24,
+	92, 16004, 916, 3496, 64, 16244, 36,
+	80, 16064, 676, 3556, 248, 16184, 48,
+	64, 16124, 452, 3576, 452, 16124, 64,
+};
+
+static const uint16_t filter_7tap_16p_150[63] = {
+	16224, 16380, 2208, 2208, 16380, 16224, 0,
+	16252, 16304, 2072, 2324, 84, 16196, 4,
+	16276, 16240, 1924, 2432, 184, 16172, 8,
+	16300, 16184, 1772, 2524, 296, 16144, 12,
+	16324, 16144, 1616, 2600, 416, 16124, 12,
+	16344, 16112, 1456, 2660, 548, 16104, 12,
+	16360, 16092, 1296, 2704, 688, 16088, 12,
+	16372, 16080, 1140, 2732, 832, 16080, 8,
+	0, 16076, 984, 2740, 984, 16076, 0,
+};
+
+static const uint16_t filter_7tap_16p_183[63] = {
+	16216, 324, 1884, 1884, 324, 16216, 0,
+	16228, 248, 1804, 1960, 408, 16212, 16380,
+	16240, 176, 1716, 2028, 496, 16208, 16376,
+	16252, 112, 1624, 2084, 588, 16208, 16372,
+	16264, 56, 1524, 2132, 684, 16212, 16364,
+	16280, 4, 1424, 2168, 788, 16220, 16356,
+	16292, 16344, 1320, 2196, 892, 16232, 16344,
+	16308, 16308, 1212, 2212, 996, 16252, 16332,
+	16320, 16276, 1104, 2216, 1104, 16276, 16320,
+};
+
+static const uint16_t filter_8tap_16p_upscale[72] = {
+	0, 0, 0, 4096, 0, 0, 0, 0,
+	16360, 76, 16172, 4064, 244, 16296, 24, 16380,
+	16340, 136, 15996, 3980, 524, 16204, 56, 16380,
+	16328, 188, 15860, 3844, 828, 16104, 92, 16372,
+	16320, 224, 15760, 3656, 1156, 16008, 128, 16368,
+	16320, 248, 15696, 3428, 1496, 15912, 160, 16360,
+	16320, 256, 15668, 3156, 1844, 15828, 192, 16348,
+	16324, 256, 15672, 2856, 2192, 15756, 220, 16340,
+	16332, 244, 15704, 2532, 2532, 15704, 244, 16332,
+};
+
+static const uint16_t filter_8tap_16p_117[72] = {
+	116, 16100, 428, 3564, 428, 16100, 116, 0,
+	96, 16168, 220, 3548, 656, 16032, 136, 16376,
+	76, 16236, 32, 3496, 904, 15968, 152, 16372,
+	56, 16300, 16252, 3408, 1164, 15908, 164, 16368,
+	36, 16360, 16116, 3284, 1428, 15856, 172, 16364,
+	20, 28, 16000, 3124, 1700, 15820, 176, 16364,
+	4, 76, 15912, 2940, 1972, 15800, 172, 16364,
+	16380, 112, 15848, 2724, 2236, 15792, 160, 16364,
+	16372, 140, 15812, 2488, 2488, 15812, 140, 16372,
+};
+
+static const uint16_t filter_8tap_16p_150[72] = {
+	16380, 16020, 1032, 2756, 1032, 16020, 16380, 0,
+	12, 16020, 876, 2744, 1184, 16032, 16364, 4,
+	24, 16028, 728, 2716, 1344, 16052, 16340, 8,
+	36, 16040, 584, 2668, 1500, 16080, 16316, 16,
+	40, 16060, 448, 2608, 1652, 16120, 16288, 20,
+	44, 16080, 320, 2528, 1804, 16168, 16260, 28,
+	48, 16108, 204, 2436, 1948, 16232, 16228, 32,
+	44, 16136, 100, 2328, 2084, 16304, 16200, 40,
+	44, 16168, 4, 2212, 2212, 4, 16168, 44,
+};
+
+static const uint16_t filter_8tap_16p_183[72] = {
+	16264, 16264, 1164, 2244, 1164, 16264, 16264, 0,
+	16280, 16232, 1056, 2236, 1268, 16300, 16248, 0,
+	16296, 16204, 948, 2220, 1372, 16348, 16232, 0,
+	16312, 16184, 844, 2192, 1472, 12, 16216, 4,
+	16328, 16172, 740, 2156, 1572, 72, 16200, 0,
+	16340, 16160, 640, 2108, 1668, 136, 16188, 0,
+	16352, 16156, 544, 2052, 1756, 204, 16176, 16380,
+	16360, 16156, 452, 1988, 1840, 280, 16164, 16376,
+	16368, 16160, 364, 1920, 1920, 364, 16160, 16368,
+};
+
+static const uint16_t filter_9tap_16p_upscale[81] = {
+	16284, 296, 15660, 2572, 2572, 15660, 296, 16284, 0,
+	16296, 272, 15712, 2228, 2896, 15632, 304, 16276, 4,
+	16308, 240, 15788, 1876, 3192, 15632, 304, 16276, 4,
+	16320, 204, 15876, 1520, 3452, 15664, 288, 16280, 8,
+	16336, 164, 15976, 1176, 3676, 15732, 260, 16288, 12,
+	16348, 120, 16080, 844, 3856, 15840, 216, 16300, 12,
+	16364, 76, 16188, 532, 3988, 15984, 156, 16324, 8,
+	16376, 36, 16288, 252, 4068, 16164, 84, 16352, 4,
+	0, 0, 0, 0, 4096, 0, 0, 0, 0,
+};
+
+static const uint16_t filter_9tap_16p_117[81] = {
+	16356, 172, 15776, 2504, 2504, 15776, 172, 16356, 0,
+	16344, 200, 15756, 2252, 2740, 15816, 136, 16372, 16380,
+	16336, 216, 15756, 1988, 2956, 15884, 92, 8, 16380,
+	16332, 224, 15780, 1720, 3144, 15976, 40, 28, 16376,
+	16328, 224, 15816, 1448, 3304, 16096, 16364, 52, 16372,
+	16328, 216, 15868, 1180, 3432, 16240, 16296, 80, 16364,
+	16332, 200, 15928, 916, 3524, 24, 16224, 108, 16356,
+	16336, 184, 15996, 668, 3580, 220, 16148, 132, 16352,
+	16344, 160, 16072, 436, 3600, 436, 16072, 160, 16344,
+};
+
+static const uint16_t filter_9tap_16p_150[81] = {
+	84, 16128, 0, 2216, 2216, 0, 16128, 84, 0,
+	80, 16160, 16296, 2088, 2332, 100, 16092, 84, 0,
+	76, 16196, 16220, 1956, 2432, 208, 16064, 80, 0,
+	72, 16232, 16152, 1812, 2524, 328, 16036, 76, 4,
+	64, 16264, 16096, 1664, 2600, 460, 16012, 64, 8,
+	56, 16300, 16052, 1508, 2656, 596, 15996, 52, 12,
+	48, 16328, 16020, 1356, 2700, 740, 15984, 36, 20,
+	40, 16356, 15996, 1196, 2728, 888, 15980, 20, 24,
+	32, 0, 15984, 1044, 2736, 1044, 15984, 0, 32,
+};
+
+static const uint16_t filter_9tap_16p_183[81] = {
+	16356, 16112, 388, 1952, 1952, 388, 16112, 16356, 0,
+	16368, 16116, 304, 1876, 2020, 480, 16112, 16344, 4,
+	16376, 16124, 224, 1792, 2080, 576, 16116, 16328, 8,
+	0, 16136, 148, 1700, 2132, 672, 16124, 16312, 8,
+	8, 16148, 80, 1604, 2176, 772, 16140, 16296, 12,
+	12, 16164, 16, 1504, 2208, 876, 16156, 16276, 16,
+	16, 16180, 16344, 1404, 2232, 980, 16184, 16256, 20,
+	20, 16200, 16296, 1300, 2244, 1088, 16212, 16240, 20,
+	20, 16220, 16252, 1196, 2252, 1196, 16252, 16220, 20,
+};
+
+static const uint16_t filter_10tap_16p_upscale[90] = {
+	0, 0, 0, 0, 4096, 0, 0, 0, 0, 0,
+	12, 16344, 88, 16160, 4068, 252, 16280, 44, 16368, 0,
+	24, 16308, 168, 15976, 3988, 540, 16176, 92, 16348, 0,
+	32, 16280, 236, 15828, 3852, 852, 16064, 140, 16328, 4,
+	36, 16260, 284, 15720, 3672, 1184, 15956, 188, 16308, 8,
+	36, 16244, 320, 15648, 3448, 1528, 15852, 236, 16288, 12,
+	36, 16240, 336, 15612, 3184, 1880, 15764, 276, 16272, 20,
+	32, 16240, 340, 15608, 2888, 2228, 15688, 308, 16256, 24,
+	28, 16244, 332, 15636, 2568, 2568, 15636, 332, 16244, 28,
+};
+
+static const uint16_t filter_10tap_16p_117[90] = {
+	16308, 196, 16048, 440, 3636, 440, 16048, 196, 16308, 0,
+	16316, 164, 16132, 220, 3612, 676, 15972, 220, 16300, 0,
+	16324, 132, 16212, 20, 3552, 932, 15900, 240, 16296, 4,
+	16336, 100, 16292, 16232, 3456, 1192, 15836, 256, 16296, 4,
+	16348, 68, 16364, 16084, 3324, 1464, 15784, 264, 16296, 8,
+	16356, 36, 48, 15960, 3164, 1736, 15748, 260, 16304, 4,
+	16364, 8, 108, 15864, 2972, 2008, 15728, 252, 16312, 4,
+	16372, 16368, 160, 15792, 2756, 2268, 15724, 228, 16328, 0,
+	16380, 16344, 200, 15748, 2520, 2520, 15748, 200, 16344, 16380,
+};
+
+static const uint16_t filter_10tap_16p_150[90] = {
+	64, 0, 15956, 1048, 2716, 1048, 15956, 0, 64, 0,
+	52, 24, 15952, 896, 2708, 1204, 15972, 16356, 72, 16380,
+	44, 48, 15952, 748, 2684, 1360, 16000, 16320, 84, 16380,
+	32, 68, 15964, 604, 2644, 1516, 16032, 16288, 92, 16376,
+	24, 88, 15980, 464, 2588, 1668, 16080, 16248, 100, 16376,
+	16, 100, 16004, 332, 2516, 1816, 16140, 16212, 108, 16376,
+	8, 108, 16032, 212, 2428, 1956, 16208, 16172, 112, 16376,
+	4, 116, 16060, 100, 2328, 2092, 16288, 16132, 116, 16380,
+	0, 116, 16096, 16380, 2216, 2216, 16380, 16096, 116, 0,
+};
+
+static const uint16_t filter_10tap_16p_183[90] = {
+	40, 16180, 16240, 1216, 2256, 1216, 16240, 16180, 40, 0,
+	44, 16204, 16200, 1112, 2252, 1320, 16288, 16160, 36, 0,
+	44, 16224, 16168, 1004, 2236, 1424, 16344, 16144, 28, 4,
+	44, 16248, 16136, 900, 2208, 1524, 16, 16124, 24, 8,
+	44, 16268, 16116, 796, 2176, 1620, 84, 16108, 12, 12,
+	40, 16288, 16100, 692, 2132, 1712, 156, 16096, 4, 16,
+	36, 16308, 16088, 592, 2080, 1796, 232, 16088, 16376, 20,
+	32, 16328, 16080, 496, 2020, 1876, 316, 16080, 16360, 24,
+	28, 16344, 16080, 404, 1952, 1952, 404, 16080, 16344, 28,
+};
+
+static const uint16_t filter_11tap_16p_upscale[99] = {
+	60, 16216, 356, 15620, 2556, 2556, 15620, 356, 16216, 60, 0,
+	52, 16224, 336, 15672, 2224, 2876, 15592, 368, 16208, 64, 16380,
+	44, 16244, 304, 15744, 1876, 3176, 15596, 364, 16212, 64, 16376,
+	36, 16264, 260, 15836, 1532, 3440, 15636, 340, 16220, 60, 16376,
+	28, 16288, 212, 15940, 1188, 3668, 15708, 304, 16236, 56, 16376,
+	20, 16312, 160, 16052, 856, 3848, 15820, 248, 16264, 48, 16376,
+	12, 16336, 104, 16164, 544, 3984, 15968, 180, 16296, 36, 16376,
+	4, 16360, 48, 16276, 256, 4068, 16160, 96, 16336, 16, 16380,
+	0, 0, 0, 0, 0, 4096, 0, 0, 0, 0, 0,
+};
+
+static const uint16_t filter_11tap_16p_117[99] = {
+	16380, 16332, 220, 15728, 2536, 2536, 15728, 220, 16332, 16380, 0,
+	4, 16308, 256, 15704, 2280, 2768, 15772, 176, 16360, 16368, 0,
+	12, 16292, 280, 15704, 2016, 2984, 15848, 120, 8, 16356, 0,
+	20, 16276, 292, 15724, 1744, 3172, 15948, 56, 40, 16340, 4,
+	24, 16268, 292, 15760, 1468, 3328, 16072, 16368, 80, 16324, 8,
+	24, 16264, 288, 15816, 1196, 3456, 16224, 16288, 116, 16312, 12,
+	24, 16264, 272, 15880, 932, 3548, 16, 16208, 152, 16296, 16,
+	24, 16268, 248, 15956, 676, 3604, 216, 16120, 188, 16284, 20,
+	24, 16276, 220, 16036, 436, 3624, 436, 16036, 220, 16276, 24,
+};
+
+static const uint16_t filter_11tap_16p_150[99] = {
+	0, 144, 16072, 0, 2212, 2212, 0, 16072, 144, 0, 0,
+	16376, 144, 16112, 16288, 2092, 2324, 104, 16036, 140, 8, 16380,
+	16368, 144, 16152, 16204, 1960, 2424, 216, 16004, 132, 16, 16376,
+	16364, 140, 16192, 16132, 1820, 2512, 340, 15976, 116, 28, 16376,
+	16364, 132, 16232, 16072, 1676, 2584, 476, 15952, 100, 40, 16372,
+	16360, 124, 16272, 16020, 1528, 2644, 612, 15936, 80, 52, 16368,
+	16360, 116, 16312, 15980, 1372, 2684, 760, 15928, 56, 64, 16364,
+	16360, 104, 16348, 15952, 1216, 2712, 908, 15928, 28, 76, 16364,
+	16360, 92, 0, 15936, 1064, 2720, 1064, 15936, 0, 92, 16360,
+};
+
+static const uint16_t filter_11tap_16p_183[99] = {
+	60, 16336, 16052, 412, 1948, 1948, 412, 16052, 16336, 60, 0,
+	56, 16356, 16052, 324, 1876, 2016, 504, 16056, 16316, 64, 0,
+	48, 16372, 16060, 240, 1796, 2072, 604, 16064, 16292, 64, 0,
+	44, 4, 16068, 160, 1712, 2124, 700, 16080, 16272, 68, 0,
+	40, 20, 16080, 84, 1620, 2164, 804, 16096, 16248, 68, 4,
+	32, 32, 16096, 16, 1524, 2200, 908, 16124, 16224, 68, 4,
+	28, 40, 16112, 16340, 1428, 2220, 1012, 16152, 16200, 64, 8,
+	24, 52, 16132, 16284, 1328, 2236, 1120, 16192, 16176, 64, 12,
+	16, 56, 16156, 16236, 1224, 2240, 1224, 16236, 16156, 56, 16,
+};
+
+static const uint16_t filter_12tap_16p_upscale[108] = {
+	0, 0, 0, 0, 0, 4096, 0, 0, 0, 0, 0, 0,
+	16376, 24, 16332, 100, 16156, 4068, 260, 16272, 56, 16356, 8, 0,
+	16368, 44, 16284, 188, 15964, 3988, 548, 16156, 112, 16328, 20, 16380,
+	16360, 64, 16248, 260, 15812, 3856, 864, 16040, 172, 16296, 32, 16380,
+	16360, 76, 16216, 320, 15696, 3672, 1196, 15928, 228, 16268, 44, 16376,
+	16356, 84, 16196, 360, 15620, 3448, 1540, 15820, 280, 16240, 56, 16372,
+	16356, 88, 16184, 384, 15580, 3188, 1888, 15728, 324, 16216, 68, 16368,
+	16360, 88, 16180, 392, 15576, 2892, 2236, 15652, 360, 16200, 80, 16364,
+	16360, 84, 16188, 384, 15600, 2576, 2576, 15600, 384, 16188, 84, 16360,
+};
+
+static const uint16_t filter_12tap_16p_117[108] = {
+	48, 16248, 240, 16028, 436, 3612, 436, 16028, 240, 16248, 48, 0,
+	44, 16260, 208, 16116, 212, 3596, 676, 15944, 272, 16240, 48, 16380,
+	40, 16276, 168, 16204, 12, 3540, 932, 15868, 296, 16240, 48, 16380,
+	36, 16292, 128, 16288, 16220, 3452, 1196, 15800, 312, 16240, 44, 16380,
+	28, 16308, 84, 16372, 16064, 3324, 1472, 15748, 316, 16244, 40, 16380,
+	24, 16328, 44, 64, 15936, 3168, 1744, 15708, 312, 16256, 32, 16380,
+	16, 16344, 8, 132, 15836, 2980, 2016, 15688, 300, 16272, 20, 0,
+	12, 16364, 16356, 188, 15760, 2768, 2280, 15688, 272, 16296, 8, 4,
+	8, 16380, 16324, 236, 15712, 2532, 2532, 15712, 236, 16324, 16380, 8,
+};
+
+static const uint16_t filter_12tap_16p_150[108] = {
+	16340, 116, 0, 15916, 1076, 2724, 1076, 15916, 0, 116, 16340, 0,
+	16340, 100, 32, 15908, 920, 2716, 1232, 15936, 16344, 128, 16340, 0,
+	16344, 84, 64, 15908, 772, 2692, 1388, 15968, 16304, 140, 16344, 16380,
+	16344, 68, 92, 15912, 624, 2652, 1540, 16008, 16264, 152, 16344, 16380,
+	16348, 52, 112, 15928, 484, 2592, 1688, 16060, 16220, 160, 16348, 16380,
+	16352, 40, 132, 15952, 348, 2520, 1836, 16124, 16176, 168, 16356, 16376,
+	16356, 24, 148, 15980, 224, 2436, 1976, 16200, 16132, 172, 16364, 16372,
+	16360, 12, 160, 16012, 108, 2336, 2104, 16288, 16088, 172, 16372, 16368,
+	16364, 0, 168, 16048, 0, 2228, 2228, 0, 16048, 168, 0, 16364,
+};
+
+static const uint16_t filter_12tap_16p_183[108] = {
+	36, 72, 16132, 16228, 1224, 2224, 1224, 16228, 16132, 72, 36, 0,
+	28, 80, 16156, 16184, 1120, 2224, 1328, 16280, 16112, 64, 40, 16380,
+	24, 84, 16180, 16144, 1016, 2208, 1428, 16340, 16092, 52, 48, 16380,
+	16, 88, 16208, 16112, 912, 2188, 1524, 16, 16072, 36, 56, 16380,
+	12, 92, 16232, 16084, 812, 2156, 1620, 88, 16056, 24, 64, 16380,
+	8, 92, 16256, 16064, 708, 2116, 1708, 164, 16044, 4, 68, 16380,
+	4, 88, 16280, 16048, 608, 2068, 1792, 244, 16036, 16372, 76, 16380,
+	0, 88, 16308, 16036, 512, 2008, 1872, 328, 16032, 16352, 80, 16380,
+	0, 84, 16328, 16032, 416, 1944, 1944, 416, 16032, 16328, 84, 0,
+};
+
+static const uint16_t *wbscl_get_filter_3tap_16p(struct fixed31_32 ratio)
+{
+	if (ratio.value < dc_fixpt_one.value)
+		return filter_3tap_16p_upscale;
+	else if (ratio.value < dc_fixpt_from_fraction(4, 3).value)
+		return filter_3tap_16p_117;
+	else if (ratio.value < dc_fixpt_from_fraction(5, 3).value)
+		return filter_3tap_16p_150;
+	else
+		return filter_3tap_16p_183;
+}
+
+static const uint16_t *wbscl_get_filter_4tap_16p(struct fixed31_32 ratio)
+{
+	if (ratio.value < dc_fixpt_one.value)
+		return filter_4tap_16p_upscale;
+	else if (ratio.value < dc_fixpt_from_fraction(4, 3).value)
+		return filter_4tap_16p_117;
+	else if (ratio.value < dc_fixpt_from_fraction(5, 3).value)
+		return filter_4tap_16p_150;
+	else
+		return filter_4tap_16p_183;
+}
+
+static const uint16_t *wbscl_get_filter_5tap_16p(struct fixed31_32 ratio)
+{
+	if (ratio.value < dc_fixpt_one.value)
+		return filter_5tap_16p_upscale;
+	else if (ratio.value < dc_fixpt_from_fraction(4, 3).value)
+		return filter_5tap_16p_117;
+	else if (ratio.value < dc_fixpt_from_fraction(5, 3).value)
+		return filter_5tap_16p_150;
+	else
+		return filter_5tap_16p_183;
+}
+
+static const uint16_t *wbscl_get_filter_6tap_16p(struct fixed31_32 ratio)
+{
+	if (ratio.value < dc_fixpt_one.value)
+		return filter_6tap_16p_upscale;
+	else if (ratio.value < dc_fixpt_from_fraction(4, 3).value)
+		return filter_6tap_16p_117;
+	else if (ratio.value < dc_fixpt_from_fraction(5, 3).value)
+		return filter_6tap_16p_150;
+	else
+		return filter_6tap_16p_183;
+}
+
+static const uint16_t *wbscl_get_filter_7tap_16p(struct fixed31_32 ratio)
+{
+	if (ratio.value < dc_fixpt_one.value)
+		return filter_7tap_16p_upscale;
+	else if (ratio.value < dc_fixpt_from_fraction(4, 3).value)
+		return filter_7tap_16p_117;
+	else if (ratio.value < dc_fixpt_from_fraction(5, 3).value)
+		return filter_7tap_16p_150;
+	else
+		return filter_7tap_16p_183;
+}
+
+static const uint16_t *wbscl_get_filter_8tap_16p(struct fixed31_32 ratio)
+{
+	if (ratio.value < dc_fixpt_one.value)
+		return filter_8tap_16p_upscale;
+	else if (ratio.value < dc_fixpt_from_fraction(4, 3).value)
+		return filter_8tap_16p_117;
+	else if (ratio.value < dc_fixpt_from_fraction(5, 3).value)
+		return filter_8tap_16p_150;
+	else
+		return filter_8tap_16p_183;
+}
+
+static const uint16_t *wbscl_get_filter_9tap_16p(struct fixed31_32 ratio)
+{
+	if (ratio.value < dc_fixpt_one.value)
+		return filter_9tap_16p_upscale;
+	else if (ratio.value < dc_fixpt_from_fraction(4, 3).value)
+		return filter_9tap_16p_117;
+	else if (ratio.value < dc_fixpt_from_fraction(5, 3).value)
+		return filter_9tap_16p_150;
+	else
+		return filter_9tap_16p_183;
+}
+static const uint16_t *wbscl_get_filter_10tap_16p(struct fixed31_32 ratio)
+{
+	if (ratio.value < dc_fixpt_one.value)
+		return filter_10tap_16p_upscale;
+	else if (ratio.value < dc_fixpt_from_fraction(4, 3).value)
+		return filter_10tap_16p_117;
+	else if (ratio.value < dc_fixpt_from_fraction(5, 3).value)
+		return filter_10tap_16p_150;
+	else
+		return filter_10tap_16p_183;
+}
+
+static const uint16_t *wbscl_get_filter_11tap_16p(struct fixed31_32 ratio)
+{
+	if (ratio.value < dc_fixpt_one.value)
+		return filter_11tap_16p_upscale;
+	else if (ratio.value < dc_fixpt_from_fraction(4, 3).value)
+		return filter_11tap_16p_117;
+	else if (ratio.value < dc_fixpt_from_fraction(5, 3).value)
+		return filter_11tap_16p_150;
+	else
+		return filter_11tap_16p_183;
+}
+
+static const uint16_t *wbscl_get_filter_12tap_16p(struct fixed31_32 ratio)
+{
+	if (ratio.value < dc_fixpt_one.value)
+		return filter_12tap_16p_upscale;
+	else if (ratio.value < dc_fixpt_from_fraction(4, 3).value)
+		return filter_12tap_16p_117;
+	else if (ratio.value < dc_fixpt_from_fraction(5, 3).value)
+		return filter_12tap_16p_150;
+	else
+		return filter_12tap_16p_183;
+}
+
+static const uint16_t *wbscl_get_filter_coeffs_16p(int taps, struct fixed31_32 ratio)
+{
+	if (taps == 12)
+		return wbscl_get_filter_12tap_16p(ratio);
+	else if (taps == 11)
+		return wbscl_get_filter_11tap_16p(ratio);
+	else if (taps == 10)
+		return wbscl_get_filter_10tap_16p(ratio);
+	else if (taps == 9)
+		return wbscl_get_filter_9tap_16p(ratio);
+	else if (taps == 8)
+		return wbscl_get_filter_8tap_16p(ratio);
+	else if (taps == 7)
+		return wbscl_get_filter_7tap_16p(ratio);
+	else if (taps == 6)
+		return wbscl_get_filter_6tap_16p(ratio);
+	else if (taps == 5)
+		return wbscl_get_filter_5tap_16p(ratio);
+	else if (taps == 4)
+		return wbscl_get_filter_4tap_16p(ratio);
+	else if (taps == 3)
+		return wbscl_get_filter_3tap_16p(ratio);
+	else if (taps == 2)
+		return get_filter_2tap_16p();
+	else if (taps == 1)
+		return NULL;
+	else {
+		/* should never happen, bug */
+		BREAK_TO_DEBUGGER();
+		return NULL;
+	}
+}
+
+static void wbscl_set_scaler_filter(
+	struct dcn20_dwbc *dwbc20,
+	uint32_t taps,
+	enum wbscl_coef_filter_type_sel filter_type,
+	const uint16_t *filter)
+{
+	const int tap_pairs = (taps + 1) / 2;
+	int phase;
+	int pair;
+	uint16_t odd_coef, even_coef;
+
+	if (!filter)
+		return;
+
+	for (phase = 0; phase < (NUM_PHASES / 2 + 1); phase++) {
+		for (pair = 0; pair < tap_pairs; pair++) {
+			even_coef = filter[phase * taps + 2 * pair];
+			if ((pair * 2 + 1) < taps)
+				odd_coef = filter[phase * taps + 2 * pair + 1];
+			else
+				odd_coef = 0;
+
+			REG_SET_3(WBSCL_COEF_RAM_SELECT, 0,
+				WBSCL_COEF_RAM_TAP_PAIR_IDX, pair,
+				WBSCL_COEF_RAM_PHASE, phase,
+				WBSCL_COEF_RAM_FILTER_TYPE, filter_type);
+
+			REG_SET_4(WBSCL_COEF_RAM_TAP_DATA, 0,
+				/* Even tap coefficient (bits 1:0 fixed to 0) */
+				WBSCL_COEF_RAM_EVEN_TAP_COEF, even_coef,
+				/* Write/read control for even coefficient */
+				WBSCL_COEF_RAM_EVEN_TAP_COEF_EN, 1,
+				/* Odd tap coefficient (bits 1:0 fixed to 0) */
+				WBSCL_COEF_RAM_ODD_TAP_COEF, odd_coef,
+				/* Write/read control for odd coefficient */
+				WBSCL_COEF_RAM_ODD_TAP_COEF_EN, 1);
+		}
+	}
+}
+
+bool dwb_program_horz_scalar(struct dcn20_dwbc *dwbc20,
+		uint32_t src_width,
+		uint32_t dest_width,
+		struct scaling_taps num_taps)
+{
+	uint32_t h_ratio_luma = 1;
+	uint32_t h_taps_luma = num_taps.h_taps;
+	uint32_t h_taps_chroma = num_taps.h_taps_c;
+	int32_t h_init_phase_luma = 0;
+	int32_t h_init_phase_chroma = 0;
+	uint32_t h_init_phase_luma_int = 0;
+	uint32_t h_init_phase_luma_frac = 0;
+	uint32_t h_init_phase_chroma_int = 0;
+	uint32_t h_init_phase_chroma_frac = 0;
+	const uint16_t *filter_h = NULL;
+	const uint16_t *filter_h_c = NULL;
+
+
+	struct fixed31_32 tmp_h_init_phase_luma = dc_fixpt_from_int(0);
+	struct fixed31_32 tmp_h_init_phase_chroma = dc_fixpt_from_int(0);
+
+
+	/*Calculate ratio*/
+	struct fixed31_32 tmp_h_ratio_luma = dc_fixpt_from_fraction(
+		src_width, dest_width);
+
+	if (dc_fixpt_floor(tmp_h_ratio_luma) == 8)
+		h_ratio_luma = -1;
+	else
+		h_ratio_luma = dc_fixpt_u3d19(tmp_h_ratio_luma) << 5;
+
+	/*Program ratio*/
+	REG_UPDATE(WBSCL_HORZ_FILTER_SCALE_RATIO, WBSCL_H_SCALE_RATIO, h_ratio_luma);
+
+	/* Program taps*/
+	REG_UPDATE(WBSCL_TAP_CONTROL, WBSCL_H_NUM_OF_TAPS_Y_RGB, h_taps_luma - 1);
+	REG_UPDATE(WBSCL_TAP_CONTROL, WBSCL_H_NUM_OF_TAPS_CBCR, h_taps_chroma - 1);
+
+	/* Calculate phase*/
+	tmp_h_init_phase_luma = dc_fixpt_add_int(tmp_h_ratio_luma, h_taps_luma + 1);
+	tmp_h_init_phase_luma = dc_fixpt_div_int(tmp_h_init_phase_luma, 2);
+	tmp_h_init_phase_luma = dc_fixpt_sub_int(tmp_h_init_phase_luma, h_taps_luma);
+
+	h_init_phase_luma = dc_fixpt_s4d19(tmp_h_init_phase_luma);
+	h_init_phase_luma_int = (h_init_phase_luma >> 19) & 0x1f;
+	h_init_phase_luma_frac = (h_init_phase_luma & 0x7ffff) << 5;
+
+	tmp_h_init_phase_chroma = dc_fixpt_mul_int(tmp_h_ratio_luma, 2);
+	tmp_h_init_phase_chroma = dc_fixpt_add_int(tmp_h_init_phase_chroma, h_taps_chroma + 1);
+	tmp_h_init_phase_chroma = dc_fixpt_div_int(tmp_h_init_phase_chroma, 2);
+	tmp_h_init_phase_chroma = dc_fixpt_sub_int(tmp_h_init_phase_chroma, h_taps_chroma);
+	tmp_h_init_phase_chroma = dc_fixpt_add(tmp_h_init_phase_chroma, dc_fixpt_from_fraction(1, 4));
+
+	h_init_phase_chroma = dc_fixpt_s4d19(tmp_h_init_phase_chroma);
+	h_init_phase_chroma_int = (h_init_phase_chroma >> 19) & 0x1f;
+	h_init_phase_chroma_frac = (h_init_phase_chroma & 0x7ffff) << 5;
+
+	/* Program phase*/
+	REG_UPDATE(WBSCL_HORZ_FILTER_INIT_Y_RGB, WBSCL_H_INIT_INT_Y_RGB, h_init_phase_luma_int);
+	REG_UPDATE(WBSCL_HORZ_FILTER_INIT_Y_RGB, WBSCL_H_INIT_FRAC_Y_RGB, h_init_phase_luma_frac);
+	REG_UPDATE(WBSCL_HORZ_FILTER_INIT_CBCR, WBSCL_H_INIT_INT_CBCR, h_init_phase_chroma_int);
+	REG_UPDATE(WBSCL_HORZ_FILTER_INIT_CBCR, WBSCL_H_INIT_FRAC_CBCR, h_init_phase_chroma_frac);
+
+	/* Program LUT coefficients*/
+	filter_h = wbscl_get_filter_coeffs_16p(
+		h_taps_luma, tmp_h_ratio_luma);
+	filter_h_c = wbscl_get_filter_coeffs_16p(
+		h_taps_chroma, dc_fixpt_from_int(h_ratio_luma * 2));
+
+	wbscl_set_scaler_filter(dwbc20, h_taps_luma,
+		WBSCL_COEF_LUMA_HORZ_FILTER, filter_h);
+
+	wbscl_set_scaler_filter(dwbc20, h_taps_chroma,
+		WBSCL_COEF_CHROMA_HORZ_FILTER, filter_h_c);
+
+	return true;
+}
+
+bool dwb_program_vert_scalar(struct dcn20_dwbc *dwbc20,
+		uint32_t src_height,
+		uint32_t dest_height,
+		struct scaling_taps num_taps,
+		enum dwb_subsample_position subsample_position)
+{
+	uint32_t v_ratio_luma = 1;
+	uint32_t v_taps_luma = num_taps.v_taps;
+	uint32_t v_taps_chroma = num_taps.v_taps_c;
+	int32_t v_init_phase_luma = 0;
+	int32_t v_init_phase_chroma = 0;
+	uint32_t v_init_phase_luma_int = 0;
+	uint32_t v_init_phase_luma_frac = 0;
+	uint32_t v_init_phase_chroma_int = 0;
+	uint32_t v_init_phase_chroma_frac = 0;
+
+	const uint16_t *filter_v = NULL;
+	const uint16_t *filter_v_c = NULL;
+
+	struct fixed31_32 tmp_v_init_phase_luma = dc_fixpt_from_int(0);
+	struct fixed31_32 tmp_v_init_phase_chroma = dc_fixpt_from_int(0);
+
+	/*Calculate ratio*/
+	struct fixed31_32 tmp_v_ratio_luma = dc_fixpt_from_fraction(
+		src_height, dest_height);
+
+	if (dc_fixpt_floor(tmp_v_ratio_luma) == 8)
+		v_ratio_luma = -1;
+	else
+		v_ratio_luma = dc_fixpt_u3d19(tmp_v_ratio_luma) << 5;
+
+	/*Program ratio*/
+	REG_UPDATE(WBSCL_VERT_FILTER_SCALE_RATIO, WBSCL_V_SCALE_RATIO, v_ratio_luma);
+
+	/* Program taps*/
+	REG_UPDATE(WBSCL_TAP_CONTROL, WBSCL_V_NUM_OF_TAPS_Y_RGB, v_taps_luma - 1);
+	REG_UPDATE(WBSCL_TAP_CONTROL, WBSCL_V_NUM_OF_TAPS_CBCR, v_taps_chroma - 1);
+
+	/* Calculate phase*/
+	tmp_v_init_phase_luma = dc_fixpt_add_int(tmp_v_ratio_luma, v_taps_luma + 1);
+	tmp_v_init_phase_luma = dc_fixpt_div_int(tmp_v_init_phase_luma, 2);
+	tmp_v_init_phase_luma = dc_fixpt_sub_int(tmp_v_init_phase_luma, v_taps_luma);
+
+	v_init_phase_luma = dc_fixpt_s4d19(tmp_v_init_phase_luma);
+	v_init_phase_luma_int = (v_init_phase_luma >> 19) & 0x1f;
+	v_init_phase_luma_frac = (v_init_phase_luma & 0x7ffff) << 5;
+
+	tmp_v_init_phase_chroma = dc_fixpt_mul_int(tmp_v_ratio_luma, 2);
+	tmp_v_init_phase_chroma = dc_fixpt_add_int(tmp_v_init_phase_chroma, v_taps_chroma + 1);
+	tmp_v_init_phase_chroma = dc_fixpt_div_int(tmp_v_init_phase_chroma, 2);
+	tmp_v_init_phase_chroma = dc_fixpt_sub_int(tmp_v_init_phase_chroma, v_taps_chroma);
+	if (subsample_position == DWB_COSITED_SUBSAMPLING)
+		tmp_v_init_phase_chroma = dc_fixpt_add(tmp_v_init_phase_chroma, dc_fixpt_from_fraction(1, 4));
+
+	v_init_phase_chroma = dc_fixpt_s4d19(tmp_v_init_phase_chroma);
+	v_init_phase_chroma_int = (v_init_phase_chroma >> 19) & 0x1f;
+	v_init_phase_chroma_frac = (v_init_phase_chroma & 0x7ffff) << 5;
+
+	/* Program phase*/
+	REG_UPDATE(WBSCL_VERT_FILTER_INIT_Y_RGB, WBSCL_V_INIT_INT_Y_RGB, v_init_phase_luma_int);
+	REG_UPDATE(WBSCL_VERT_FILTER_INIT_Y_RGB, WBSCL_V_INIT_FRAC_Y_RGB, v_init_phase_luma_frac);
+	REG_UPDATE(WBSCL_VERT_FILTER_INIT_CBCR, WBSCL_V_INIT_INT_CBCR, v_init_phase_chroma_int);
+	REG_UPDATE(WBSCL_VERT_FILTER_INIT_CBCR, WBSCL_V_INIT_FRAC_CBCR, v_init_phase_chroma_frac);
+
+
+	/* Program LUT coefficients*/
+	filter_v  = wbscl_get_filter_coeffs_16p(
+		v_taps_luma, tmp_v_ratio_luma);
+	filter_v_c = wbscl_get_filter_coeffs_16p(
+		v_taps_chroma, dc_fixpt_from_int(v_ratio_luma * 2));
+	wbscl_set_scaler_filter(dwbc20, v_taps_luma,
+		WBSCL_COEF_LUMA_VERT_FILTER, filter_v);
+
+	wbscl_set_scaler_filter(dwbc20, v_taps_chroma,
+		WBSCL_COEF_CHROMA_VERT_FILTER, filter_v_c);
+	return true;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_vmid.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_vmid.c
new file mode 100644
index 000000000000..5bc3bc60a2ac
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_vmid.c
@@ -0,0 +1,99 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include <linux/delay.h>
+
+#include "dcn20_vmid.h"
+#include "reg_helper.h"
+
+#define REG(reg)\
+	vmid->regs->reg
+
+#define CTX \
+	vmid->ctx
+
+#undef FN
+#define FN(reg_name, field_name) \
+	vmid->shifts->field_name, vmid->masks->field_name
+
+#define DC_LOGGER \
+	CTX->logger
+
+static void dcn20_wait_for_vmid_ready(struct dcn20_vmid *vmid)
+{
+	/* According the hardware spec, we need to poll for the lowest
+	 * bit of PAGE_TABLE_BASE_ADDR_LO32 = 1 any time a GPUVM
+	 * context is updated. We can't use REG_WAIT here since we
+	 * don't have a seperate field to wait on.
+	 *
+	 * TODO: Confirm timeout / poll interval with hardware team
+	 */
+
+	int max_times = 10000;
+	int delay_us  = 5;
+	int i;
+
+	for (i = 0; i < max_times; ++i) {
+		uint32_t entry_lo32;
+
+		REG_GET(PAGE_TABLE_BASE_ADDR_LO32,
+			VM_CONTEXT0_PAGE_DIRECTORY_ENTRY_LO32,
+			&entry_lo32);
+
+		if (entry_lo32 & 0x1)
+			return;
+
+		udelay(delay_us);
+	}
+
+	/* VM setup timed out */
+	DC_LOG_WARNING("Timeout while waiting for GPUVM context update\n");
+	ASSERT(0);
+}
+
+void dcn20_vmid_setup(struct dcn20_vmid *vmid, const struct dcn_vmid_page_table_config *config)
+{
+	REG_SET(PAGE_TABLE_START_ADDR_HI32, 0,
+			VM_CONTEXT0_START_LOGICAL_PAGE_NUMBER_HI4, (config->page_table_start_addr >> 32) & 0xF);
+	REG_SET(PAGE_TABLE_START_ADDR_LO32, 0,
+			VM_CONTEXT0_START_LOGICAL_PAGE_NUMBER_LO32, config->page_table_start_addr & 0xFFFFFFFF);
+
+	REG_SET(PAGE_TABLE_END_ADDR_HI32, 0,
+			VM_CONTEXT0_END_LOGICAL_PAGE_NUMBER_HI4, (config->page_table_end_addr >> 32) & 0xF);
+	REG_SET(PAGE_TABLE_END_ADDR_LO32, 0,
+			VM_CONTEXT0_END_LOGICAL_PAGE_NUMBER_LO32, config->page_table_end_addr & 0xFFFFFFFF);
+
+	REG_SET_2(CNTL, 0,
+			VM_CONTEXT0_PAGE_TABLE_DEPTH, config->depth,
+			VM_CONTEXT0_PAGE_TABLE_BLOCK_SIZE, config->block_size);
+
+	REG_SET(PAGE_TABLE_BASE_ADDR_HI32, 0,
+			VM_CONTEXT0_PAGE_DIRECTORY_ENTRY_HI32, (config->page_table_base_addr >> 32) & 0xFFFFFFFF);
+	/* Note: per hardware spec PAGE_TABLE_BASE_ADDR_LO32 must be programmed last in sequence */
+	REG_SET(PAGE_TABLE_BASE_ADDR_LO32, 0,
+			VM_CONTEXT0_PAGE_DIRECTORY_ENTRY_LO32, config->page_table_base_addr & 0xFFFFFFFF);
+
+	dcn20_wait_for_vmid_ready(vmid);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_vmid.h b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_vmid.h
new file mode 100644
index 000000000000..e7a1b7fa2cce
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_vmid.h
@@ -0,0 +1,77 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DAL_DC_DCN20_DCN20_VMID_H_
+#define DAL_DC_DCN20_DCN20_VMID_H_
+
+#include "vmid.h"
+
+#define DCN20_VMID_REG_LIST(id)\
+	SRI(CNTL, DCN_VM_CONTEXT, id),\
+	SRI(PAGE_TABLE_BASE_ADDR_HI32, DCN_VM_CONTEXT, id),\
+	SRI(PAGE_TABLE_BASE_ADDR_LO32, DCN_VM_CONTEXT, id),\
+	SRI(PAGE_TABLE_START_ADDR_HI32, DCN_VM_CONTEXT, id),\
+	SRI(PAGE_TABLE_START_ADDR_LO32, DCN_VM_CONTEXT, id),\
+	SRI(PAGE_TABLE_END_ADDR_HI32, DCN_VM_CONTEXT, id),\
+	SRI(PAGE_TABLE_END_ADDR_LO32, DCN_VM_CONTEXT, id)
+
+#define DCN20_VMID_MASK_SH_LIST(mask_sh)\
+	SF(DCN_VM_CONTEXT0_CNTL, VM_CONTEXT0_PAGE_TABLE_DEPTH, mask_sh),\
+	SF(DCN_VM_CONTEXT0_CNTL, VM_CONTEXT0_PAGE_TABLE_BLOCK_SIZE, mask_sh),\
+	SF(DCN_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32, VM_CONTEXT0_PAGE_DIRECTORY_ENTRY_HI32, mask_sh),\
+	SF(DCN_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32, VM_CONTEXT0_PAGE_DIRECTORY_ENTRY_LO32, mask_sh),\
+	SF(DCN_VM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32, VM_CONTEXT0_START_LOGICAL_PAGE_NUMBER_HI4, mask_sh),\
+	SF(DCN_VM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32, VM_CONTEXT0_START_LOGICAL_PAGE_NUMBER_LO32, mask_sh),\
+	SF(DCN_VM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32, VM_CONTEXT0_END_LOGICAL_PAGE_NUMBER_HI4, mask_sh),\
+	SF(DCN_VM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32, VM_CONTEXT0_END_LOGICAL_PAGE_NUMBER_LO32, mask_sh)
+
+#define DCN20_VMID_REG_FIELD_LIST(type)\
+	type VM_CONTEXT0_PAGE_TABLE_DEPTH;\
+	type VM_CONTEXT0_PAGE_TABLE_BLOCK_SIZE;\
+	type VM_CONTEXT0_PAGE_DIRECTORY_ENTRY_HI32;\
+	type VM_CONTEXT0_PAGE_DIRECTORY_ENTRY_LO32;\
+	type VM_CONTEXT0_START_LOGICAL_PAGE_NUMBER_HI4;\
+	type VM_CONTEXT0_START_LOGICAL_PAGE_NUMBER_LO32;\
+	type VM_CONTEXT0_END_LOGICAL_PAGE_NUMBER_HI4;\
+	type VM_CONTEXT0_END_LOGICAL_PAGE_NUMBER_LO32
+
+struct dcn20_vmid_shift {
+	DCN20_VMID_REG_FIELD_LIST(uint8_t);
+};
+
+struct dcn20_vmid_mask {
+	DCN20_VMID_REG_FIELD_LIST(uint32_t);
+};
+
+struct dcn20_vmid {
+	struct dc_context *ctx;
+	const struct dcn_vmid_registers *regs;
+	const struct dcn20_vmid_shift *shifts;
+	const struct dcn20_vmid_mask *masks;
+};
+
+void dcn20_vmid_setup(struct dcn20_vmid *vmid, const struct dcn_vmid_page_table_config *config);
+
+#endif /* DAL_DC_DCN20_DCN20_VMID_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/dcn201/Makefile b/drivers/gpu/drm/amd/display/dc/dcn201/Makefile
new file mode 100644
index 000000000000..c9f4a5a9f522
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn201/Makefile
@@ -0,0 +1,9 @@
+# SPDX-License-Identifier: MIT
+# Copyright © 2021-2024 Advanced Micro Devices, Inc. All rights reserved.
+
+DCN201 = dcn201_mpc.o dcn201_opp.o \
+	dcn201_link_encoder.o
+
+AMD_DAL_DCN201 = $(addprefix $(AMDDALPATH)/dc/dcn201/,$(DCN201))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCN201)
diff --git a/drivers/gpu/drm/amd/display/dc/dcn201/dcn201_link_encoder.c b/drivers/gpu/drm/amd/display/dc/dcn201/dcn201_link_encoder.c
new file mode 100644
index 000000000000..8d31fa131cd6
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn201/dcn201_link_encoder.c
@@ -0,0 +1,208 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+
+#include "core_types.h"
+#include "link_encoder.h"
+#include "dcn201_link_encoder.h"
+#include "stream_encoder.h"
+#include "dc_bios_types.h"
+
+#include "gpio_service_interface.h"
+
+#define CTX \
+	enc10->base.ctx
+
+#define DC_LOGGER \
+	enc10->base.ctx->logger
+
+#define REG(reg)\
+	(enc10->link_regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	enc10->link_shift->field_name, enc10->link_mask->field_name
+
+#define IND_REG(index) \
+	(enc10->link_regs->index)
+
+static void dcn201_link_encoder_get_max_link_cap(struct link_encoder *enc,
+						 struct dc_link_settings *link_settings)
+{
+	uint32_t value1, value2;
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	dcn10_link_encoder_get_max_link_cap(enc, link_settings);
+	REG_GET_2(RDPCSTX_PHY_CNTL2, RDPCS_PHY_DPALT_DISABLE, &value1,
+		 RDPCS_PHY_DPALT_DP4, &value2);
+	/*limit to combo_phy*/
+	if (enc->usbc_combo_phy) {
+		if (!value1 && !value2 && link_settings->lane_count > LANE_COUNT_TWO)
+			link_settings->lane_count = LANE_COUNT_TWO;
+	}
+}
+
+static bool dcn201_link_encoder_is_in_alt_mode(struct link_encoder *enc)
+{
+	uint32_t value;
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	REG_GET(RDPCSTX_PHY_CNTL2, RDPCS_PHY_DPALT_DISABLE, &value);
+
+	// if value == 1 alt mode is disabled, otherwise it is enabled
+	return !value;
+}
+
+static const struct link_encoder_funcs dcn201_link_enc_funcs = {
+	.read_state = link_enc2_read_state,
+	.validate_output_with_stream =
+		dcn10_link_encoder_validate_output_with_stream,
+	.hw_init = enc2_hw_init,
+	.setup = dcn10_link_encoder_setup,
+	.enable_tmds_output = dcn10_link_encoder_enable_tmds_output,
+	.enable_dp_output = dcn10_link_encoder_enable_dp_output,
+	.enable_dp_mst_output = dcn10_link_encoder_enable_dp_mst_output,
+	.disable_output = dcn10_link_encoder_disable_output,
+	.dp_set_lane_settings = dcn10_link_encoder_dp_set_lane_settings,
+	.dp_set_phy_pattern = dcn10_link_encoder_dp_set_phy_pattern,
+	.update_mst_stream_allocation_table =
+		dcn10_link_encoder_update_mst_stream_allocation_table,
+	.psr_program_dp_dphy_fast_training =
+			dcn10_psr_program_dp_dphy_fast_training,
+	.psr_program_secondary_packet = dcn10_psr_program_secondary_packet,
+	.connect_dig_be_to_fe = dcn10_link_encoder_connect_dig_be_to_fe,
+	.enable_hpd = dcn10_link_encoder_enable_hpd,
+	.disable_hpd = dcn10_link_encoder_disable_hpd,
+	.is_dig_enabled = dcn10_is_dig_enabled,
+	.destroy = dcn10_link_encoder_destroy,
+	.fec_set_enable = enc2_fec_set_enable,
+	.fec_set_ready = enc2_fec_set_ready,
+	.get_dig_frontend = dcn10_get_dig_frontend,
+	.fec_is_active = enc2_fec_is_active,
+	.is_in_alt_mode = dcn201_link_encoder_is_in_alt_mode,
+	.get_max_link_cap = dcn201_link_encoder_get_max_link_cap,
+};
+
+void dcn201_link_encoder_construct(
+	struct dcn20_link_encoder *enc20,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	const struct dcn10_link_enc_aux_registers *aux_regs,
+	const struct dcn10_link_enc_hpd_registers *hpd_regs,
+	const struct dcn10_link_enc_shift *link_shift,
+	const struct dcn10_link_enc_mask *link_mask)
+{
+	struct bp_encoder_cap_info bp_cap_info = {0};
+	const struct dc_vbios_funcs *bp_funcs = init_data->ctx->dc_bios->funcs;
+	enum bp_result result = BP_RESULT_OK;
+	struct dcn10_link_encoder *enc10 = &enc20->enc10;
+
+	enc10->base.funcs = &dcn201_link_enc_funcs;
+	enc10->base.ctx = init_data->ctx;
+	enc10->base.id = init_data->encoder;
+
+	enc10->base.hpd_source = init_data->hpd_source;
+	enc10->base.connector = init_data->connector;
+
+	enc10->base.preferred_engine = ENGINE_ID_UNKNOWN;
+
+	enc10->base.features = *enc_features;
+
+	enc10->base.transmitter = init_data->transmitter;
+
+	/* set the flag to indicate whether driver poll the I2C data pin
+	 * while doing the DP sink detect
+	 */
+
+	/* if (dal_adapter_service_is_feature_supported(as,
+	 * FEATURE_DP_SINK_DETECT_POLL_DATA_PIN))
+	 * enc10->base.features.flags.bits.
+	 * DP_SINK_DETECT_POLL_DATA_PIN = true;
+	 */
+
+	enc10->base.output_signals =
+		SIGNAL_TYPE_DVI_SINGLE_LINK |
+		SIGNAL_TYPE_DVI_DUAL_LINK |
+		SIGNAL_TYPE_LVDS |
+		SIGNAL_TYPE_DISPLAY_PORT |
+		SIGNAL_TYPE_DISPLAY_PORT_MST |
+		SIGNAL_TYPE_EDP |
+		SIGNAL_TYPE_HDMI_TYPE_A;
+
+	/* For DCE 8.0 and 8.1, by design, UNIPHY is hardwired to DIG_BE.
+	 * SW always assign DIG_FE 1:1 mapped to DIG_FE for non-MST UNIPHY.
+	 * SW assign DIG_FE to non-MST UNIPHY first and MST last. So prefer
+	 * DIG is per UNIPHY and used by SST DP, eDP, HDMI, DVI and LVDS.
+	 * Prefer DIG assignment is decided by board design.
+	 * For DCE 8.0, there are only max 6 UNIPHYs, we assume board design
+	 * and VBIOS will filter out 7 UNIPHY for DCE 8.0.
+	 * By this, adding DIGG should not hurt DCE 8.0.
+	 * This will let DCE 8.1 share DCE 8.0 as much as possible
+	 */
+
+	enc10->link_regs = link_regs;
+	enc10->aux_regs = aux_regs;
+	enc10->hpd_regs = hpd_regs;
+	enc10->link_shift = link_shift;
+	enc10->link_mask = link_mask;
+
+	switch (enc10->base.transmitter) {
+	case TRANSMITTER_UNIPHY_A:
+		enc10->base.preferred_engine = ENGINE_ID_DIGA;
+	break;
+	case TRANSMITTER_UNIPHY_B:
+		enc10->base.preferred_engine = ENGINE_ID_DIGB;
+	break;
+	default:
+		ASSERT_CRITICAL(false);
+		enc10->base.preferred_engine = ENGINE_ID_UNKNOWN;
+	}
+
+	/* default to one to mirror Windows behavior */
+	enc10->base.features.flags.bits.HDMI_6GB_EN = 1;
+
+	result = bp_funcs->get_encoder_cap_info(enc10->base.ctx->dc_bios,
+						enc10->base.id, &bp_cap_info);
+
+	/* Override features with DCE-specific values */
+	if (result == BP_RESULT_OK) {
+		enc10->base.features.flags.bits.IS_HBR2_CAPABLE =
+				bp_cap_info.DP_HBR2_EN;
+		enc10->base.features.flags.bits.IS_HBR3_CAPABLE =
+				bp_cap_info.DP_HBR3_EN;
+		enc10->base.features.flags.bits.HDMI_6GB_EN = bp_cap_info.HDMI_6GB_EN;
+		enc10->base.features.flags.bits.DP_IS_USB_C =
+				bp_cap_info.DP_IS_USB_C;
+	} else {
+		DC_LOG_WARNING("%s: Failed to get encoder_cap_info from VBIOS with error code %d!\n",
+				__func__,
+				result);
+	}
+	if (enc10->base.ctx->dc->debug.hdmi20_disable) {
+		enc10->base.features.flags.bits.HDMI_6GB_EN = 0;
+	}
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dcn201/dcn201_link_encoder.h b/drivers/gpu/drm/amd/display/dc/dcn201/dcn201_link_encoder.h
new file mode 100644
index 000000000000..be25e8dc0636
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn201/dcn201_link_encoder.h
@@ -0,0 +1,71 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_LINK_ENCODER__DCN201_H__
+#define __DC_LINK_ENCODER__DCN201_H__
+
+#include "dcn20/dcn20_link_encoder.h"
+
+#define DPCS_DCN201_MASK_SH_LIST(mask_sh)\
+	DPCS_MASK_SH_LIST(mask_sh),\
+	LE_SF(DPCSSYS_CR0_RAWLANE0_DIG_PCS_XF_RX_OVRD_IN_2, VCO_LD_VAL_OVRD, mask_sh),\
+	LE_SF(DPCSSYS_CR0_RAWLANE0_DIG_PCS_XF_RX_OVRD_IN_2, VCO_LD_VAL_OVRD_EN, mask_sh),\
+	LE_SF(DPCSSYS_CR0_RAWLANE0_DIG_PCS_XF_RX_OVRD_IN_3, REF_LD_VAL_OVRD, mask_sh),\
+	LE_SF(DPCSSYS_CR0_RAWLANE0_DIG_PCS_XF_RX_OVRD_IN_3, REF_LD_VAL_OVRD_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL2, RDPCS_PHY_DPALT_DISABLE_ACK, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL2, RDPCS_PHY_DPALT_DISABLE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL2, RDPCS_PHY_DPALT_DP4, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL5, RDPCS_PHY_DP_TX0_PSTATE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL5, RDPCS_PHY_DP_TX1_PSTATE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL5, RDPCS_PHY_DP_TX0_MPLL_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL5, RDPCS_PHY_DP_TX1_MPLL_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DP_TX2_WIDTH, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DP_TX2_RATE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DP_TX3_WIDTH, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DP_TX3_RATE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL11, RDPCS_PHY_DP_REF_CLK_EN, mask_sh)
+
+#define DPCS_DCN201_REG_LIST(id) \
+	DPCS_DCN2_CMN_REG_LIST(id), \
+	SRI_IX(RAWLANE0_DIG_PCS_XF_RX_OVRD_IN_2, DPCSSYS_CR, id), \
+	SRI_IX(RAWLANE0_DIG_PCS_XF_RX_OVRD_IN_3, DPCSSYS_CR, id), \
+	SRI_IX(RAWLANE1_DIG_PCS_XF_RX_OVRD_IN_2, DPCSSYS_CR, id), \
+	SRI_IX(RAWLANE1_DIG_PCS_XF_RX_OVRD_IN_3, DPCSSYS_CR, id), \
+	SRI_IX(RAWLANE2_DIG_PCS_XF_RX_OVRD_IN_2, DPCSSYS_CR, id), \
+	SRI_IX(RAWLANE2_DIG_PCS_XF_RX_OVRD_IN_3, DPCSSYS_CR, id), \
+	SRI_IX(RAWLANE3_DIG_PCS_XF_RX_OVRD_IN_2, DPCSSYS_CR, id), \
+	SRI_IX(RAWLANE3_DIG_PCS_XF_RX_OVRD_IN_3, DPCSSYS_CR, id)
+
+void dcn201_link_encoder_construct(
+	struct dcn20_link_encoder *enc20,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	const struct dcn10_link_enc_aux_registers *aux_regs,
+	const struct dcn10_link_enc_hpd_registers *hpd_regs,
+	const struct dcn10_link_enc_shift *link_shift,
+	const struct dcn10_link_enc_mask *link_mask);
+
+#endif /* __DC_LINK_ENCODER__DCN201_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dcn201/dcn201_mpc.c b/drivers/gpu/drm/amd/display/dc/dcn201/dcn201_mpc.c
new file mode 100644
index 000000000000..1af03a86ec9b
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn201/dcn201_mpc.c
@@ -0,0 +1,125 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+#include "dcn201_mpc.h"
+
+#define REG(reg)\
+	mpc201->mpc_regs->reg
+
+#define CTX \
+	mpc201->base.ctx
+
+#define DC_LOGGER \
+	mpc201->base.ctx->logger
+
+#undef FN
+#define FN(reg_name, field_name) \
+	mpc201->mpc_shift->field_name, mpc201->mpc_mask->field_name
+
+static void mpc201_set_out_rate_control(
+	struct mpc *mpc,
+	int opp_id,
+	bool enable,
+	bool rate_2x_mode,
+	struct mpc_dwb_flow_control *flow_control)
+{
+	struct dcn201_mpc *mpc201 = TO_DCN201_MPC(mpc);
+
+	REG_UPDATE_2(MUX[opp_id],
+			MPC_OUT_RATE_CONTROL_DISABLE, !enable,
+			MPC_OUT_RATE_CONTROL, rate_2x_mode);
+
+	if (flow_control)
+		REG_UPDATE_3(MUX[opp_id],
+			MPC_OUT_FLOW_CONTROL_MODE, flow_control->flow_ctrl_mode,
+			MPC_OUT_FLOW_CONTROL_COUNT0, flow_control->flow_ctrl_cnt0,
+			MPC_OUT_FLOW_CONTROL_COUNT1, flow_control->flow_ctrl_cnt1);
+}
+
+static void mpc201_init_mpcc(struct mpcc *mpcc, int mpcc_inst)
+{
+	mpcc->mpcc_id = mpcc_inst;
+	mpcc->dpp_id = 0xf;
+	mpcc->mpcc_bot = NULL;
+	mpcc->blnd_cfg.overlap_only = false;
+	mpcc->blnd_cfg.global_alpha = 0xff;
+	mpcc->blnd_cfg.global_gain = 0xff;
+	mpcc->blnd_cfg.background_color_bpc = 4;
+	mpcc->blnd_cfg.bottom_gain_mode = 0;
+	mpcc->blnd_cfg.top_gain = 0x1f000;
+	mpcc->blnd_cfg.bottom_inside_gain = 0x1f000;
+	mpcc->blnd_cfg.bottom_outside_gain = 0x1f000;
+	mpcc->sm_cfg.enable = false;
+	mpcc->shared_bottom = false;
+}
+
+static const struct mpc_funcs dcn201_mpc_funcs = {
+	.read_mpcc_state = mpc1_read_mpcc_state,
+	.insert_plane = mpc1_insert_plane,
+	.remove_mpcc = mpc1_remove_mpcc,
+	.mpc_init = mpc1_mpc_init,
+	.mpc_init_single_inst = mpc1_mpc_init_single_inst,
+	.update_blending = mpc2_update_blending,
+	.cursor_lock = mpc1_cursor_lock,
+	.get_mpcc_for_dpp = mpc1_get_mpcc_for_dpp,
+	.get_mpcc_for_dpp_from_secondary = NULL,
+	.wait_for_idle = mpc2_assert_idle_mpcc,
+	.assert_mpcc_idle_before_connect = mpc2_assert_mpcc_idle_before_connect,
+	.init_mpcc_list_from_hw = mpc1_init_mpcc_list_from_hw,
+	.set_denorm = mpc2_set_denorm,
+	.set_denorm_clamp = mpc2_set_denorm_clamp,
+	.set_output_csc = mpc2_set_output_csc,
+	.set_ocsc_default = mpc2_set_ocsc_default,
+	.set_output_gamma = mpc2_set_output_gamma,
+	.set_out_rate_control = mpc201_set_out_rate_control,
+	.power_on_mpc_mem_pwr = mpc20_power_on_ogam_lut,
+	.get_mpc_out_mux = mpc1_get_mpc_out_mux,
+	.set_bg_color = mpc1_set_bg_color,
+};
+
+void dcn201_mpc_construct(struct dcn201_mpc *mpc201,
+	struct dc_context *ctx,
+	const struct dcn201_mpc_registers *mpc_regs,
+	const struct dcn201_mpc_shift *mpc_shift,
+	const struct dcn201_mpc_mask *mpc_mask,
+	int num_mpcc)
+{
+	int i;
+
+	mpc201->base.ctx = ctx;
+
+	mpc201->base.funcs = &dcn201_mpc_funcs;
+
+	mpc201->mpc_regs = mpc_regs;
+	mpc201->mpc_shift = mpc_shift;
+	mpc201->mpc_mask = mpc_mask;
+
+	mpc201->mpcc_in_use_mask = 0;
+	mpc201->num_mpcc = num_mpcc;
+
+	for (i = 0; i < MAX_MPCC; i++)
+		mpc201_init_mpcc(&mpc201->base.mpcc_array[i], i);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dcn201/dcn201_mpc.h b/drivers/gpu/drm/amd/display/dc/dcn201/dcn201_mpc.h
new file mode 100644
index 000000000000..b9ce0c1ba5c7
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn201/dcn201_mpc.h
@@ -0,0 +1,86 @@
+/* Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_MPCC_DCN201_H__
+#define __DC_MPCC_DCN201_H__
+
+#include "dcn20/dcn20_mpc.h"
+
+#define TO_DCN201_MPC(mpc_base) \
+	container_of(mpc_base, struct dcn201_mpc, base)
+
+#define MPC_REG_LIST_DCN201(inst) \
+	MPC_REG_LIST_DCN2_0(inst)
+
+#define MPC_OUT_MUX_REG_LIST_DCN201(inst) \
+	MPC_OUT_MUX_REG_LIST_DCN2_0(inst)
+
+#define MPC_REG_VARIABLE_LIST_DCN201 \
+	MPC_REG_VARIABLE_LIST_DCN2_0
+
+#define MPC_COMMON_MASK_SH_LIST_DCN201(mask_sh) \
+	MPC_COMMON_MASK_SH_LIST_DCN2_0(mask_sh),\
+	SF(MPC_OUT0_MUX, MPC_OUT_RATE_CONTROL, mask_sh),\
+	SF(MPC_OUT0_MUX, MPC_OUT_RATE_CONTROL_DISABLE, mask_sh),\
+	SF(MPC_OUT0_MUX, MPC_OUT_FLOW_CONTROL_MODE, mask_sh),\
+	SF(MPC_OUT0_MUX, MPC_OUT_FLOW_CONTROL_COUNT0, mask_sh),\
+	SF(MPC_OUT0_MUX, MPC_OUT_FLOW_CONTROL_COUNT1, mask_sh)
+
+#define MPC_REG_FIELD_LIST_DCN201(type) \
+	MPC_REG_FIELD_LIST_DCN2_0(type) \
+	type MPC_OUT_RATE_CONTROL;\
+	type MPC_OUT_RATE_CONTROL_DISABLE;\
+	type MPC_OUT_FLOW_CONTROL_MODE;\
+	type MPC_OUT_FLOW_CONTROL_COUNT0;\
+	type MPC_OUT_FLOW_CONTROL_COUNT1;
+
+struct dcn201_mpc_registers {
+	MPC_REG_VARIABLE_LIST_DCN201
+};
+
+struct dcn201_mpc_shift {
+	MPC_REG_FIELD_LIST_DCN201(uint8_t)
+};
+
+struct dcn201_mpc_mask {
+	MPC_REG_FIELD_LIST_DCN201(uint32_t)
+};
+
+struct dcn201_mpc {
+	struct mpc base;
+	int mpcc_in_use_mask;
+	int num_mpcc;
+	const struct dcn201_mpc_registers *mpc_regs;
+	const struct dcn201_mpc_shift *mpc_shift;
+	const struct dcn201_mpc_mask *mpc_mask;
+};
+
+void dcn201_mpc_construct(struct dcn201_mpc *mpc201,
+	struct dc_context *ctx,
+	const struct dcn201_mpc_registers *mpc_regs,
+	const struct dcn201_mpc_shift *mpc_shift,
+	const struct dcn201_mpc_mask *mpc_mask,
+	int num_mpcc);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dcn201/dcn201_opp.c b/drivers/gpu/drm/amd/display/dc/dcn201/dcn201_opp.c
new file mode 100644
index 000000000000..e83367a9b6b3
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn201/dcn201_opp.c
@@ -0,0 +1,74 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+#include "dcn201_opp.h"
+#include "reg_helper.h"
+
+#define REG(reg) \
+	(oppn201->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	oppn201->opp_shift->field_name, oppn201->opp_mask->field_name
+
+#define CTX \
+	oppn201->base.ctx
+
+/*****************************************/
+/* Constructor, Destructor               */
+/*****************************************/
+
+static struct opp_funcs dcn201_opp_funcs = {
+		.opp_set_dyn_expansion = opp1_set_dyn_expansion,
+		.opp_program_fmt = opp1_program_fmt,
+		.opp_program_bit_depth_reduction = opp1_program_bit_depth_reduction,
+		.opp_program_stereo = opp1_program_stereo,
+		.opp_pipe_clock_control = opp1_pipe_clock_control,
+		.opp_set_disp_pattern_generator = opp2_set_disp_pattern_generator,
+		.opp_program_dpg_dimensions = opp2_program_dpg_dimensions,
+		.dpg_is_blanked = opp2_dpg_is_blanked,
+		.dpg_is_pending = opp2_dpg_is_pending,
+		.opp_dpg_set_blank_color = opp2_dpg_set_blank_color,
+		.opp_destroy = opp1_destroy,
+		.opp_program_left_edge_extra_pixel = opp2_program_left_edge_extra_pixel,
+		.opp_get_left_edge_extra_pixel_count = opp2_get_left_edge_extra_pixel_count,
+};
+
+void dcn201_opp_construct(struct dcn201_opp *oppn201,
+	struct dc_context *ctx,
+	uint32_t inst,
+	const struct dcn201_opp_registers *regs,
+	const struct dcn201_opp_shift *opp_shift,
+	const struct dcn201_opp_mask *opp_mask)
+{
+	oppn201->base.ctx = ctx;
+	oppn201->base.inst = inst;
+	oppn201->base.funcs = &dcn201_opp_funcs;
+
+	oppn201->regs = regs;
+	oppn201->opp_shift = opp_shift;
+	oppn201->opp_mask = opp_mask;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dcn201/dcn201_opp.h b/drivers/gpu/drm/amd/display/dc/dcn201/dcn201_opp.h
new file mode 100644
index 000000000000..edb7f9653cb6
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn201/dcn201_opp.h
@@ -0,0 +1,73 @@
+/* Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_OPP_DCN201_H__
+#define __DC_OPP_DCN201_H__
+
+#include "dcn20/dcn20_opp.h"
+
+#define TO_DCN201_OPP(opp)\
+	container_of(opp, struct dcn201_opp, base)
+
+#define OPP_SF(reg_name, field_name, post_fix)\
+	.field_name = reg_name ## __ ## field_name ## post_fix
+
+#define OPP_REG_LIST_DCN201(id) \
+	OPP_REG_LIST_DCN10(id), \
+	OPP_DPG_REG_LIST(id), \
+	SRI(FMT_422_CONTROL, FMT, id)
+
+#define OPP_MASK_SH_LIST_DCN201(mask_sh) \
+	OPP_MASK_SH_LIST_DCN20(mask_sh)
+
+#define OPP_DCN201_REG_FIELD_LIST(type) \
+	OPP_DCN20_REG_FIELD_LIST(type)
+struct dcn201_opp_shift {
+	OPP_DCN201_REG_FIELD_LIST(uint8_t);
+};
+
+struct dcn201_opp_mask {
+	OPP_DCN201_REG_FIELD_LIST(uint32_t);
+};
+
+struct dcn201_opp_registers {
+	OPP_REG_VARIABLE_LIST_DCN2_0;
+};
+
+struct dcn201_opp {
+	struct output_pixel_processor base;
+	const struct  dcn201_opp_registers *regs;
+	const struct  dcn201_opp_shift *opp_shift;
+	const struct  dcn201_opp_mask *opp_mask;
+	bool is_write_to_ram_a_safe;
+};
+
+void dcn201_opp_construct(struct dcn201_opp *oppn201,
+	struct dc_context *ctx,
+	uint32_t inst,
+	const struct dcn201_opp_registers *regs,
+	const struct dcn201_opp_shift *opp_shift,
+	const struct dcn201_opp_mask *opp_mask);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dcn21/Makefile b/drivers/gpu/drm/amd/display/dc/dcn21/Makefile
new file mode 100644
index 000000000000..c215f3cc6e44
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn21/Makefile
@@ -0,0 +1,8 @@
+# SPDX-License-Identifier: MIT
+# Copyright © 2019-2024 Advanced Micro Devices, Inc. All rights reserved.
+
+DCN21 = dcn21_link_encoder.o
+
+AMD_DAL_DCN21 = $(addprefix $(AMDDALPATH)/dc/dcn21/,$(DCN21))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCN21)
diff --git a/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_link_encoder.c b/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_link_encoder.c
new file mode 100644
index 000000000000..eb9abb9f9698
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_link_encoder.c
@@ -0,0 +1,443 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+
+#include <linux/delay.h>
+#include "core_types.h"
+#include "link_encoder.h"
+#include "dcn21_link_encoder.h"
+#include "stream_encoder.h"
+
+#include "dc_bios_types.h"
+
+#include "gpio_service_interface.h"
+
+#define CTX \
+	enc10->base.ctx
+#define DC_LOGGER \
+	enc10->base.ctx->logger
+
+#define REG(reg)\
+	(enc10->link_regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	enc10->link_shift->field_name, enc10->link_mask->field_name
+
+#define IND_REG(index) \
+	(enc10->link_regs->index)
+
+static struct mpll_cfg dcn21_mpll_cfg_ref[] = {
+	// RBR
+	{
+		.hdmimode_enable = 0,
+		.ref_range = 1,
+		.ref_clk_mpllb_div = 1,
+		.mpllb_ssc_en = 1,
+		.mpllb_div5_clk_en = 1,
+		.mpllb_multiplier = 238,
+		.mpllb_fracn_en = 0,
+		.mpllb_fracn_quot = 0,
+		.mpllb_fracn_rem = 0,
+		.mpllb_fracn_den = 1,
+		.mpllb_ssc_up_spread = 0,
+		.mpllb_ssc_peak = 44237,
+		.mpllb_ssc_stepsize = 59454,
+		.mpllb_div_clk_en = 0,
+		.mpllb_div_multiplier = 0,
+		.mpllb_hdmi_div = 0,
+		.mpllb_tx_clk_div = 2,
+		.tx_vboost_lvl = 5,
+		.mpllb_pmix_en = 1,
+		.mpllb_word_div2_en = 0,
+		.mpllb_ana_v2i = 2,
+		.mpllb_ana_freq_vco = 2,
+		.mpllb_ana_cp_int = 9,
+		.mpllb_ana_cp_prop = 15,
+		.hdmi_pixel_clk_div = 0,
+	},
+	// HBR
+	{
+		.hdmimode_enable = 0,
+		.ref_range = 1,
+		.ref_clk_mpllb_div = 1,
+		.mpllb_ssc_en = 1,
+		.mpllb_div5_clk_en = 1,
+		.mpllb_multiplier = 192,
+		.mpllb_fracn_en = 1,
+		.mpllb_fracn_quot = 32768,
+		.mpllb_fracn_rem = 0,
+		.mpllb_fracn_den = 1,
+		.mpllb_ssc_up_spread = 0,
+		.mpllb_ssc_peak = 36864,
+		.mpllb_ssc_stepsize = 49545,
+		.mpllb_div_clk_en = 0,
+		.mpllb_div_multiplier = 0,
+		.mpllb_hdmi_div = 0,
+		.mpllb_tx_clk_div = 1,
+		.tx_vboost_lvl = 5,
+		.mpllb_pmix_en = 1,
+		.mpllb_word_div2_en = 0,
+		.mpllb_ana_v2i = 2,
+		.mpllb_ana_freq_vco = 3,
+		.mpllb_ana_cp_int = 9,
+		.mpllb_ana_cp_prop = 15,
+		.hdmi_pixel_clk_div = 0,
+	},
+	//HBR2
+	{
+		.hdmimode_enable = 0,
+		.ref_range = 1,
+		.ref_clk_mpllb_div = 1,
+		.mpllb_ssc_en = 1,
+		.mpllb_div5_clk_en = 1,
+		.mpllb_multiplier = 192,
+		.mpllb_fracn_en = 1,
+		.mpllb_fracn_quot = 32768,
+		.mpllb_fracn_rem = 0,
+		.mpllb_fracn_den = 1,
+		.mpllb_ssc_up_spread = 0,
+		.mpllb_ssc_peak = 36864,
+		.mpllb_ssc_stepsize = 49545,
+		.mpllb_div_clk_en = 0,
+		.mpllb_div_multiplier = 0,
+		.mpllb_hdmi_div = 0,
+		.mpllb_tx_clk_div = 0,
+		.tx_vboost_lvl = 5,
+		.mpllb_pmix_en = 1,
+		.mpllb_word_div2_en = 0,
+		.mpllb_ana_v2i = 2,
+		.mpllb_ana_freq_vco = 3,
+		.mpllb_ana_cp_int = 9,
+		.mpllb_ana_cp_prop = 15,
+		.hdmi_pixel_clk_div = 0,
+	},
+	//HBR3
+	{
+		.hdmimode_enable = 0,
+		.ref_range = 1,
+		.ref_clk_mpllb_div = 1,
+		.mpllb_ssc_en = 1,
+		.mpllb_div5_clk_en = 1,
+		.mpllb_multiplier = 304,
+		.mpllb_fracn_en = 1,
+		.mpllb_fracn_quot = 49152,
+		.mpllb_fracn_rem = 0,
+		.mpllb_fracn_den = 1,
+		.mpllb_ssc_up_spread = 0,
+		.mpllb_ssc_peak = 55296,
+		.mpllb_ssc_stepsize = 74318,
+		.mpllb_div_clk_en = 0,
+		.mpllb_div_multiplier = 0,
+		.mpllb_hdmi_div = 0,
+		.mpllb_tx_clk_div = 0,
+		.tx_vboost_lvl = 5,
+		.mpllb_pmix_en = 1,
+		.mpllb_word_div2_en = 0,
+		.mpllb_ana_v2i = 2,
+		.mpllb_ana_freq_vco = 1,
+		.mpllb_ana_cp_int = 7,
+		.mpllb_ana_cp_prop = 16,
+		.hdmi_pixel_clk_div = 0,
+	},
+};
+
+
+static bool update_cfg_data(
+		struct dcn10_link_encoder *enc10,
+		const struct dc_link_settings *link_settings,
+		struct dpcssys_phy_seq_cfg *cfg)
+{
+	int i;
+
+	cfg->load_sram_fw = false;
+	cfg->use_calibration_setting = true;
+
+	//TODO: need to implement a proper lane mapping for Renoir.
+	for (i = 0; i < 4; i++)
+		cfg->lane_en[i] = true;
+
+	switch (link_settings->link_rate) {
+	case LINK_RATE_LOW:
+		cfg->mpll_cfg = dcn21_mpll_cfg_ref[0];
+		break;
+	case LINK_RATE_HIGH:
+		cfg->mpll_cfg = dcn21_mpll_cfg_ref[1];
+		break;
+	case LINK_RATE_HIGH2:
+		cfg->mpll_cfg = dcn21_mpll_cfg_ref[2];
+		break;
+	case LINK_RATE_HIGH3:
+		cfg->mpll_cfg = dcn21_mpll_cfg_ref[3];
+		break;
+	default:
+		DC_LOG_ERROR("%s: No supported link rate found %X!\n",
+				__func__, link_settings->link_rate);
+		return false;
+	}
+
+	return true;
+}
+
+static bool dcn21_link_encoder_acquire_phy(struct link_encoder *enc)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	int value;
+
+	if (enc->features.flags.bits.DP_IS_USB_C) {
+		REG_GET(RDPCSTX_PHY_CNTL6,
+				RDPCS_PHY_DPALT_DISABLE, &value);
+
+		if (value == 1) {
+			ASSERT(0);
+			return false;
+		}
+		REG_UPDATE(RDPCSTX_PHY_CNTL6,
+				RDPCS_PHY_DPALT_DISABLE_ACK, 0);
+
+		udelay(40);
+
+		REG_GET(RDPCSTX_PHY_CNTL6,
+						RDPCS_PHY_DPALT_DISABLE, &value);
+		if (value == 1) {
+			ASSERT(0);
+			REG_UPDATE(RDPCSTX_PHY_CNTL6,
+					RDPCS_PHY_DPALT_DISABLE_ACK, 1);
+			return false;
+		}
+	}
+
+	REG_UPDATE(RDPCSTX_PHY_CNTL6, RDPCS_PHY_DP_REF_CLK_EN, 1);
+
+	return true;
+}
+
+
+
+static void dcn21_link_encoder_release_phy(struct link_encoder *enc)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	if (enc->features.flags.bits.DP_IS_USB_C) {
+		REG_UPDATE(RDPCSTX_PHY_CNTL6,
+				RDPCS_PHY_DPALT_DISABLE_ACK, 1);
+	}
+
+	REG_UPDATE(RDPCSTX_PHY_CNTL6, RDPCS_PHY_DP_REF_CLK_EN, 0);
+
+}
+
+void dcn21_link_encoder_enable_dp_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	struct dcn21_link_encoder *enc21 = (struct dcn21_link_encoder *) enc10;
+	struct dpcssys_phy_seq_cfg *cfg = &enc21->phy_seq_cfg;
+
+	if (!dcn21_link_encoder_acquire_phy(enc))
+		return;
+
+	if (!enc->ctx->dc->debug.avoid_vbios_exec_table) {
+		dcn10_link_encoder_enable_dp_output(enc, link_settings, clock_source);
+		return;
+	}
+
+	if (!update_cfg_data(enc10, link_settings, cfg))
+		return;
+
+	enc1_configure_encoder(enc10, link_settings);
+
+	dcn10_link_encoder_setup(enc, SIGNAL_TYPE_DISPLAY_PORT);
+
+}
+
+static void dcn21_link_encoder_enable_dp_mst_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source)
+{
+	if (!dcn21_link_encoder_acquire_phy(enc))
+		return;
+
+	dcn10_link_encoder_enable_dp_mst_output(enc, link_settings, clock_source);
+}
+
+static void dcn21_link_encoder_disable_output(struct link_encoder *enc,
+					      enum signal_type signal)
+{
+	dcn10_link_encoder_disable_output(enc, signal);
+
+	if (dc_is_dp_signal(signal))
+		dcn21_link_encoder_release_phy(enc);
+}
+
+
+static const struct link_encoder_funcs dcn21_link_enc_funcs = {
+	.read_state = link_enc2_read_state,
+	.validate_output_with_stream =
+		dcn10_link_encoder_validate_output_with_stream,
+	.hw_init = enc2_hw_init,
+	.setup = dcn10_link_encoder_setup,
+	.enable_tmds_output = dcn10_link_encoder_enable_tmds_output,
+	.enable_dp_output = dcn21_link_encoder_enable_dp_output,
+	.enable_dp_mst_output = dcn21_link_encoder_enable_dp_mst_output,
+	.disable_output = dcn21_link_encoder_disable_output,
+	.dp_set_lane_settings = dcn10_link_encoder_dp_set_lane_settings,
+	.dp_set_phy_pattern = dcn10_link_encoder_dp_set_phy_pattern,
+	.update_mst_stream_allocation_table =
+		dcn10_link_encoder_update_mst_stream_allocation_table,
+	.psr_program_dp_dphy_fast_training =
+			dcn10_psr_program_dp_dphy_fast_training,
+	.psr_program_secondary_packet = dcn10_psr_program_secondary_packet,
+	.connect_dig_be_to_fe = dcn10_link_encoder_connect_dig_be_to_fe,
+	.enable_hpd = dcn10_link_encoder_enable_hpd,
+	.disable_hpd = dcn10_link_encoder_disable_hpd,
+	.is_dig_enabled = dcn10_is_dig_enabled,
+	.destroy = dcn10_link_encoder_destroy,
+	.fec_set_enable = enc2_fec_set_enable,
+	.fec_set_ready = enc2_fec_set_ready,
+	.fec_is_active = enc2_fec_is_active,
+	.get_dig_frontend = dcn10_get_dig_frontend,
+	.is_in_alt_mode = dcn20_link_encoder_is_in_alt_mode,
+	.get_max_link_cap = dcn20_link_encoder_get_max_link_cap,
+};
+
+void dcn21_link_encoder_construct(
+	struct dcn21_link_encoder *enc21,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	const struct dcn10_link_enc_aux_registers *aux_regs,
+	const struct dcn10_link_enc_hpd_registers *hpd_regs,
+	const struct dcn10_link_enc_shift *link_shift,
+	const struct dcn10_link_enc_mask *link_mask)
+{
+	struct bp_encoder_cap_info bp_cap_info = {0};
+	const struct dc_vbios_funcs *bp_funcs = init_data->ctx->dc_bios->funcs;
+	enum bp_result result = BP_RESULT_OK;
+	struct dcn10_link_encoder *enc10 = &enc21->enc10;
+
+	enc10->base.funcs = &dcn21_link_enc_funcs;
+	enc10->base.ctx = init_data->ctx;
+	enc10->base.id = init_data->encoder;
+
+	enc10->base.hpd_source = init_data->hpd_source;
+	enc10->base.connector = init_data->connector;
+
+	enc10->base.preferred_engine = ENGINE_ID_UNKNOWN;
+
+	enc10->base.features = *enc_features;
+
+	enc10->base.transmitter = init_data->transmitter;
+
+	/* set the flag to indicate whether driver poll the I2C data pin
+	 * while doing the DP sink detect
+	 */
+
+/*	if (dal_adapter_service_is_feature_supported(as,
+		FEATURE_DP_SINK_DETECT_POLL_DATA_PIN))
+		enc10->base.features.flags.bits.
+			DP_SINK_DETECT_POLL_DATA_PIN = true;*/
+
+	enc10->base.output_signals =
+		SIGNAL_TYPE_DVI_SINGLE_LINK |
+		SIGNAL_TYPE_DVI_DUAL_LINK |
+		SIGNAL_TYPE_LVDS |
+		SIGNAL_TYPE_DISPLAY_PORT |
+		SIGNAL_TYPE_DISPLAY_PORT_MST |
+		SIGNAL_TYPE_EDP |
+		SIGNAL_TYPE_HDMI_TYPE_A;
+
+	/* For DCE 8.0 and 8.1, by design, UNIPHY is hardwired to DIG_BE.
+	 * SW always assign DIG_FE 1:1 mapped to DIG_FE for non-MST UNIPHY.
+	 * SW assign DIG_FE to non-MST UNIPHY first and MST last. So prefer
+	 * DIG is per UNIPHY and used by SST DP, eDP, HDMI, DVI and LVDS.
+	 * Prefer DIG assignment is decided by board design.
+	 * For DCE 8.0, there are only max 6 UNIPHYs, we assume board design
+	 * and VBIOS will filter out 7 UNIPHY for DCE 8.0.
+	 * By this, adding DIGG should not hurt DCE 8.0.
+	 * This will let DCE 8.1 share DCE 8.0 as much as possible
+	 */
+
+	enc10->link_regs = link_regs;
+	enc10->aux_regs = aux_regs;
+	enc10->hpd_regs = hpd_regs;
+	enc10->link_shift = link_shift;
+	enc10->link_mask = link_mask;
+
+	switch (enc10->base.transmitter) {
+	case TRANSMITTER_UNIPHY_A:
+		enc10->base.preferred_engine = ENGINE_ID_DIGA;
+	break;
+	case TRANSMITTER_UNIPHY_B:
+		enc10->base.preferred_engine = ENGINE_ID_DIGB;
+	break;
+	case TRANSMITTER_UNIPHY_C:
+		enc10->base.preferred_engine = ENGINE_ID_DIGC;
+	break;
+	case TRANSMITTER_UNIPHY_D:
+		enc10->base.preferred_engine = ENGINE_ID_DIGD;
+	break;
+	case TRANSMITTER_UNIPHY_E:
+		enc10->base.preferred_engine = ENGINE_ID_DIGE;
+	break;
+	case TRANSMITTER_UNIPHY_F:
+		enc10->base.preferred_engine = ENGINE_ID_DIGF;
+	break;
+	case TRANSMITTER_UNIPHY_G:
+		enc10->base.preferred_engine = ENGINE_ID_DIGG;
+	break;
+	default:
+		ASSERT_CRITICAL(false);
+		enc10->base.preferred_engine = ENGINE_ID_UNKNOWN;
+	}
+
+	/* default to one to mirror Windows behavior */
+	enc10->base.features.flags.bits.HDMI_6GB_EN = 1;
+
+	result = bp_funcs->get_encoder_cap_info(enc10->base.ctx->dc_bios,
+						enc10->base.id, &bp_cap_info);
+
+	/* Override features with DCE-specific values */
+	if (result == BP_RESULT_OK) {
+		enc10->base.features.flags.bits.IS_HBR2_CAPABLE =
+				bp_cap_info.DP_HBR2_EN;
+		enc10->base.features.flags.bits.IS_HBR3_CAPABLE =
+				bp_cap_info.DP_HBR3_EN;
+		enc10->base.features.flags.bits.HDMI_6GB_EN = bp_cap_info.HDMI_6GB_EN;
+		enc10->base.features.flags.bits.DP_IS_USB_C =
+				bp_cap_info.DP_IS_USB_C;
+	} else {
+		DC_LOG_WARNING("%s: Failed to get encoder_cap_info from VBIOS with error code %d!\n",
+				__func__,
+				result);
+	}
+	if (enc10->base.ctx->dc->debug.hdmi20_disable) {
+		enc10->base.features.flags.bits.HDMI_6GB_EN = 0;
+	}
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_link_encoder.h b/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_link_encoder.h
new file mode 100644
index 000000000000..033d5d76f195
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_link_encoder.h
@@ -0,0 +1,100 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_LINK_ENCODER__DCN21_H__
+#define __DC_LINK_ENCODER__DCN21_H__
+
+#include "dcn20/dcn20_link_encoder.h"
+
+struct dcn21_link_encoder {
+	struct dcn10_link_encoder enc10;
+	struct dpcssys_phy_seq_cfg phy_seq_cfg;
+};
+
+#define DPCS_DCN21_MASK_SH_LIST(mask_sh)\
+	DPCS_DCN2_MASK_SH_LIST(mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE3, RDPCS_PHY_TX_VBOOST_LVL, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE2, RDPCS_PHY_DP_MPLLB_CP_PROP_GS, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE0, RDPCS_PHY_RX_VREF_CTRL, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE0, RDPCS_PHY_DP_MPLLB_CP_INT_GS, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_DMCU_DPALT_DIS_BLOCK_REG, RDPCS_DMCU_DPALT_DIS_BLOCK_REG, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL15, RDPCS_PHY_SUP_PRE_HP, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL15, RDPCS_PHY_DP_TX0_VREGDRV_BYP, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL15, RDPCS_PHY_DP_TX1_VREGDRV_BYP, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL15, RDPCS_PHY_DP_TX2_VREGDRV_BYP, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL15, RDPCS_PHY_DP_TX3_VREGDRV_BYP, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DPALT_DP4, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DPALT_DISABLE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE0, RDPCS_PHY_DP_TX0_EQ_MAIN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE0, RDPCS_PHY_DP_TX0_EQ_PRE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE0, RDPCS_PHY_DP_TX0_EQ_POST, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE1, RDPCS_PHY_DP_TX1_EQ_MAIN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE1, RDPCS_PHY_DP_TX1_EQ_PRE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE1, RDPCS_PHY_DP_TX1_EQ_POST, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE2, RDPCS_PHY_DP_TX2_EQ_MAIN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE2, RDPCS_PHY_DP_TX2_EQ_PRE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE2, RDPCS_PHY_DP_TX2_EQ_POST, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE3, RDPCS_PHY_DP_TX3_EQ_MAIN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE3, RDPCS_PHY_DCO_FINETUNE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE3, RDPCS_PHY_DCO_RANGE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE3, RDPCS_PHY_DP_TX3_EQ_PRE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE3, RDPCS_PHY_DP_TX3_EQ_POST, mask_sh),\
+	LE_SF(DCIO_SOFT_RESET, UNIPHYA_SOFT_RESET, mask_sh),\
+	LE_SF(DCIO_SOFT_RESET, UNIPHYB_SOFT_RESET, mask_sh),\
+	LE_SF(DCIO_SOFT_RESET, UNIPHYC_SOFT_RESET, mask_sh),\
+	LE_SF(DCIO_SOFT_RESET, UNIPHYD_SOFT_RESET, mask_sh),\
+	LE_SF(DCIO_SOFT_RESET, UNIPHYE_SOFT_RESET, mask_sh)
+
+#define DPCS_DCN21_REG_LIST(id) \
+	DPCS_DCN2_REG_LIST(id),\
+	SRI(RDPCSTX_PHY_CNTL15, RDPCSTX, id),\
+	SRI(RDPCSTX_DMCU_DPALT_DIS_BLOCK_REG, RDPCSTX, id)
+
+#define LINK_ENCODER_MASK_SH_LIST_DCN21(mask_sh)\
+	LINK_ENCODER_MASK_SH_LIST_DCN20(mask_sh),\
+	LE_SF(UNIPHYA_CHANNEL_XBAR_CNTL, UNIPHY_CHANNEL0_XBAR_SOURCE, mask_sh),\
+	LE_SF(UNIPHYA_CHANNEL_XBAR_CNTL, UNIPHY_CHANNEL1_XBAR_SOURCE, mask_sh),\
+	LE_SF(UNIPHYA_CHANNEL_XBAR_CNTL, UNIPHY_CHANNEL2_XBAR_SOURCE, mask_sh),\
+	LE_SF(UNIPHYA_CHANNEL_XBAR_CNTL, UNIPHY_CHANNEL3_XBAR_SOURCE, mask_sh), \
+	SRI(RDPCSTX_PHY_FUSE2, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_FUSE3, RDPCSTX, id), \
+	SR(RDPCSTX0_RDPCSTX_SCRATCH)
+
+void dcn21_link_encoder_enable_dp_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source);
+
+void dcn21_link_encoder_construct(
+	struct dcn21_link_encoder *enc21,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	const struct dcn10_link_enc_aux_registers *aux_regs,
+	const struct dcn10_link_enc_hpd_registers *hpd_regs,
+	const struct dcn10_link_enc_shift *link_shift,
+	const struct dcn10_link_enc_mask *link_mask);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dcn30/Makefile b/drivers/gpu/drm/amd/display/dc/dcn30/Makefile
new file mode 100644
index 000000000000..b17277de0340
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn30/Makefile
@@ -0,0 +1,33 @@
+#
+# Copyright 2020 Advanced Micro Devices, Inc.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+# Authors: AMD
+#
+#
+
+DCN30 := dcn30_vpg.o \
+	dcn30_afmt.o \
+	dcn30_cm_common.o \
+	dcn30_mmhubbub.o \
+
+AMD_DAL_DCN30 = $(addprefix $(AMDDALPATH)/dc/dcn30/,$(DCN30))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCN30)
diff --git a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_afmt.c b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_afmt.c
new file mode 100644
index 000000000000..55e388c4c98b
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_afmt.c
@@ -0,0 +1,215 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+#include "dc_bios_types.h"
+#include "hw_shared.h"
+#include "dcn30_afmt.h"
+#include "reg_helper.h"
+
+#define DC_LOGGER \
+		afmt3->base.ctx->logger
+
+#define REG(reg)\
+	(afmt3->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	afmt3->afmt_shift->field_name, afmt3->afmt_mask->field_name
+
+
+#define CTX \
+	afmt3->base.ctx
+
+
+void afmt3_setup_hdmi_audio(
+	struct afmt *afmt)
+{
+	struct dcn30_afmt *afmt3 = DCN30_AFMT_FROM_AFMT(afmt);
+
+	if (afmt->funcs->afmt_poweron)
+		afmt->funcs->afmt_poweron(afmt);
+
+	/* AFMT_AUDIO_PACKET_CONTROL */
+	REG_UPDATE(AFMT_AUDIO_PACKET_CONTROL, AFMT_60958_CS_UPDATE, 1);
+
+	/* AFMT_AUDIO_PACKET_CONTROL2 */
+	REG_UPDATE_2(AFMT_AUDIO_PACKET_CONTROL2,
+			AFMT_AUDIO_LAYOUT_OVRD, 0,
+			AFMT_60958_OSF_OVRD, 0);
+
+	/* AFMT_60958_0__AFMT_60958_CS_CHANNEL_NUMBER_L_MASK &
+	 * AFMT_60958_0__AFMT_60958_CS_CLOCK_ACCURACY_MASK
+	 */
+	REG_UPDATE_2(AFMT_60958_0,
+			AFMT_60958_CS_CHANNEL_NUMBER_L, 1,
+			AFMT_60958_CS_CLOCK_ACCURACY, 0);
+
+	/* AFMT_60958_1 AFMT_60958_CS_CHALNNEL_NUMBER_R */
+	REG_UPDATE(AFMT_60958_1, AFMT_60958_CS_CHANNEL_NUMBER_R, 2);
+
+	/* AFMT_60958_2 now keep this settings until
+	 * Programming guide comes out
+	 */
+	REG_UPDATE_6(AFMT_60958_2,
+			AFMT_60958_CS_CHANNEL_NUMBER_2, 3,
+			AFMT_60958_CS_CHANNEL_NUMBER_3, 4,
+			AFMT_60958_CS_CHANNEL_NUMBER_4, 5,
+			AFMT_60958_CS_CHANNEL_NUMBER_5, 6,
+			AFMT_60958_CS_CHANNEL_NUMBER_6, 7,
+			AFMT_60958_CS_CHANNEL_NUMBER_7, 8);
+}
+
+static union audio_cea_channels speakers_to_channels(
+	struct audio_speaker_flags speaker_flags)
+{
+	union audio_cea_channels cea_channels = {0};
+
+	/* these are one to one */
+	cea_channels.channels.FL = speaker_flags.FL_FR;
+	cea_channels.channels.FR = speaker_flags.FL_FR;
+	cea_channels.channels.LFE = speaker_flags.LFE;
+	cea_channels.channels.FC = speaker_flags.FC;
+
+	/* if Rear Left and Right exist move RC speaker to channel 7
+	 * otherwise to channel 5
+	 */
+	if (speaker_flags.RL_RR) {
+		cea_channels.channels.RL_RC = speaker_flags.RL_RR;
+		cea_channels.channels.RR = speaker_flags.RL_RR;
+		cea_channels.channels.RC_RLC_FLC = speaker_flags.RC;
+	} else {
+		cea_channels.channels.RL_RC = speaker_flags.RC;
+	}
+
+	/* FRONT Left Right Center and REAR Left Right Center are exclusive */
+	if (speaker_flags.FLC_FRC) {
+		cea_channels.channels.RC_RLC_FLC = speaker_flags.FLC_FRC;
+		cea_channels.channels.RRC_FRC = speaker_flags.FLC_FRC;
+	} else {
+		cea_channels.channels.RC_RLC_FLC = speaker_flags.RLC_RRC;
+		cea_channels.channels.RRC_FRC = speaker_flags.RLC_RRC;
+	}
+
+	return cea_channels;
+}
+
+void afmt3_se_audio_setup(
+	struct afmt *afmt,
+	unsigned int az_inst,
+	struct audio_info *audio_info)
+{
+	struct dcn30_afmt *afmt3 = DCN30_AFMT_FROM_AFMT(afmt);
+
+	uint32_t channels = 0;
+
+	ASSERT(audio_info);
+	/* This should not happen.it does so we don't get BSOD*/
+	if (audio_info == NULL)
+		return;
+
+	channels = speakers_to_channels(audio_info->flags.speaker_flags).all;
+
+	/* setup the audio stream source select (audio -> dig mapping) */
+	REG_SET(AFMT_AUDIO_SRC_CONTROL, 0, AFMT_AUDIO_SRC_SELECT, az_inst);
+
+	/* Channel allocation */
+	REG_UPDATE(AFMT_AUDIO_PACKET_CONTROL2, AFMT_AUDIO_CHANNEL_ENABLE, channels);
+
+	/* Disable forced mem power off */
+	if (afmt->funcs->afmt_poweron == NULL)
+		REG_UPDATE(AFMT_MEM_PWR, AFMT_MEM_PWR_FORCE, 0);
+}
+
+void afmt3_audio_mute_control(
+	struct afmt *afmt,
+	bool mute)
+{
+	struct dcn30_afmt *afmt3 = DCN30_AFMT_FROM_AFMT(afmt);
+	if (mute && afmt->funcs->afmt_powerdown)
+		afmt->funcs->afmt_powerdown(afmt);
+	if (!mute && afmt->funcs->afmt_poweron)
+		afmt->funcs->afmt_poweron(afmt);
+	/* enable/disable transmission of audio packets */
+	REG_UPDATE(AFMT_AUDIO_PACKET_CONTROL, AFMT_AUDIO_SAMPLE_SEND, !mute);
+}
+
+void afmt3_audio_info_immediate_update(
+	struct afmt *afmt)
+{
+	struct dcn30_afmt *afmt3 = DCN30_AFMT_FROM_AFMT(afmt);
+
+	/* update double-buffered AUDIO_INFO registers immediately */
+	REG_UPDATE(AFMT_INFOFRAME_CONTROL0, AFMT_AUDIO_INFO_UPDATE, 1);
+}
+
+void afmt3_setup_dp_audio(
+		struct afmt *afmt)
+{
+	struct dcn30_afmt *afmt3 = DCN30_AFMT_FROM_AFMT(afmt);
+
+	if (afmt->funcs->afmt_poweron)
+		afmt->funcs->afmt_poweron(afmt);
+
+	/* AFMT_AUDIO_PACKET_CONTROL */
+	REG_UPDATE(AFMT_AUDIO_PACKET_CONTROL, AFMT_60958_CS_UPDATE, 1);
+
+	/* AFMT_AUDIO_PACKET_CONTROL2 */
+	/* Program the ATP and AIP next */
+	REG_UPDATE_2(AFMT_AUDIO_PACKET_CONTROL2,
+			AFMT_AUDIO_LAYOUT_OVRD, 0,
+			AFMT_60958_OSF_OVRD, 0);
+
+	/* AFMT_INFOFRAME_CONTROL0 */
+	REG_UPDATE(AFMT_INFOFRAME_CONTROL0, AFMT_AUDIO_INFO_UPDATE, 1);
+
+	/* AFMT_60958_0__AFMT_60958_CS_CLOCK_ACCURACY_MASK */
+	REG_UPDATE(AFMT_60958_0, AFMT_60958_CS_CLOCK_ACCURACY, 0);
+}
+
+static struct afmt_funcs dcn30_afmt_funcs = {
+	.setup_hdmi_audio		= afmt3_setup_hdmi_audio,
+	.se_audio_setup			= afmt3_se_audio_setup,
+	.audio_mute_control		= afmt3_audio_mute_control,
+	.audio_info_immediate_update	= afmt3_audio_info_immediate_update,
+	.setup_dp_audio			= afmt3_setup_dp_audio,
+};
+
+void afmt3_construct(struct dcn30_afmt *afmt3,
+	struct dc_context *ctx,
+	uint32_t inst,
+	const struct dcn30_afmt_registers *afmt_regs,
+	const struct dcn30_afmt_shift *afmt_shift,
+	const struct dcn30_afmt_mask *afmt_mask)
+{
+	afmt3->base.ctx = ctx;
+
+	afmt3->base.inst = inst;
+	afmt3->base.funcs = &dcn30_afmt_funcs;
+
+	afmt3->regs = afmt_regs;
+	afmt3->afmt_shift = afmt_shift;
+	afmt3->afmt_mask = afmt_mask;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_afmt.h b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_afmt.h
new file mode 100644
index 000000000000..97e0cf62f98e
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_afmt.h
@@ -0,0 +1,171 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_DCN30_AFMT_H__
+#define __DAL_DCN30_AFMT_H__
+
+
+#define DCN30_AFMT_FROM_AFMT(afmt)\
+	container_of(afmt, struct dcn30_afmt, base)
+
+#define AFMT_DCN3_REG_LIST(id) \
+	SRI(AFMT_INFOFRAME_CONTROL0, AFMT, id), \
+	SRI(AFMT_VBI_PACKET_CONTROL, AFMT, id), \
+	SRI(AFMT_AUDIO_PACKET_CONTROL, AFMT, id), \
+	SRI(AFMT_AUDIO_PACKET_CONTROL2, AFMT, id), \
+	SRI(AFMT_AUDIO_SRC_CONTROL, AFMT, id), \
+	SRI(AFMT_60958_0, AFMT, id), \
+	SRI(AFMT_60958_1, AFMT, id), \
+	SRI(AFMT_60958_2, AFMT, id), \
+	SRI(AFMT_MEM_PWR, AFMT, id)
+
+struct dcn30_afmt_registers {
+	uint32_t AFMT_INFOFRAME_CONTROL0;
+	uint32_t AFMT_VBI_PACKET_CONTROL;
+	uint32_t AFMT_AUDIO_PACKET_CONTROL;
+	uint32_t AFMT_AUDIO_PACKET_CONTROL2;
+	uint32_t AFMT_AUDIO_SRC_CONTROL;
+	uint32_t AFMT_60958_0;
+	uint32_t AFMT_60958_1;
+	uint32_t AFMT_60958_2;
+	uint32_t AFMT_MEM_PWR;
+};
+
+#define DCN3_AFMT_MASK_SH_LIST(mask_sh)\
+	SE_SF(AFMT0_AFMT_INFOFRAME_CONTROL0, AFMT_AUDIO_INFO_UPDATE, mask_sh),\
+	SE_SF(AFMT0_AFMT_AUDIO_SRC_CONTROL, AFMT_AUDIO_SRC_SELECT, mask_sh),\
+	SE_SF(AFMT0_AFMT_AUDIO_PACKET_CONTROL2, AFMT_AUDIO_CHANNEL_ENABLE, mask_sh),\
+	SE_SF(AFMT0_AFMT_AUDIO_PACKET_CONTROL, AFMT_60958_CS_UPDATE, mask_sh),\
+	SE_SF(AFMT0_AFMT_AUDIO_PACKET_CONTROL2, AFMT_AUDIO_LAYOUT_OVRD, mask_sh),\
+	SE_SF(AFMT0_AFMT_AUDIO_PACKET_CONTROL2, AFMT_60958_OSF_OVRD, mask_sh),\
+	SE_SF(AFMT0_AFMT_60958_0, AFMT_60958_CS_CHANNEL_NUMBER_L, mask_sh),\
+	SE_SF(AFMT0_AFMT_60958_0, AFMT_60958_CS_CLOCK_ACCURACY, mask_sh),\
+	SE_SF(AFMT0_AFMT_60958_1, AFMT_60958_CS_CHANNEL_NUMBER_R, mask_sh),\
+	SE_SF(AFMT0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_2, mask_sh),\
+	SE_SF(AFMT0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_3, mask_sh),\
+	SE_SF(AFMT0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_4, mask_sh),\
+	SE_SF(AFMT0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_5, mask_sh),\
+	SE_SF(AFMT0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_6, mask_sh),\
+	SE_SF(AFMT0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_7, mask_sh),\
+	SE_SF(AFMT0_AFMT_AUDIO_PACKET_CONTROL, AFMT_AUDIO_SAMPLE_SEND, mask_sh),\
+	SE_SF(AFMT0_AFMT_MEM_PWR, AFMT_MEM_PWR_FORCE, mask_sh)
+
+#define AFMT_DCN3_REG_FIELD_LIST(type) \
+		type AFMT_AUDIO_INFO_UPDATE;\
+		type AFMT_AUDIO_SRC_SELECT;\
+		type AFMT_AUDIO_CHANNEL_ENABLE;\
+		type AFMT_60958_CS_UPDATE;\
+		type AFMT_AUDIO_LAYOUT_OVRD;\
+		type AFMT_60958_OSF_OVRD;\
+		type AFMT_60958_CS_CHANNEL_NUMBER_L;\
+		type AFMT_60958_CS_CLOCK_ACCURACY;\
+		type AFMT_60958_CS_CHANNEL_NUMBER_R;\
+		type AFMT_60958_CS_CHANNEL_NUMBER_2;\
+		type AFMT_60958_CS_CHANNEL_NUMBER_3;\
+		type AFMT_60958_CS_CHANNEL_NUMBER_4;\
+		type AFMT_60958_CS_CHANNEL_NUMBER_5;\
+		type AFMT_60958_CS_CHANNEL_NUMBER_6;\
+		type AFMT_60958_CS_CHANNEL_NUMBER_7;\
+		type AFMT_AUDIO_SAMPLE_SEND;\
+		type AFMT_MEM_PWR_FORCE
+
+struct dcn30_afmt_shift {
+	AFMT_DCN3_REG_FIELD_LIST(uint8_t);
+};
+
+struct dcn30_afmt_mask {
+	AFMT_DCN3_REG_FIELD_LIST(uint32_t);
+};
+
+
+struct afmt;
+
+struct afmt_funcs {
+
+	void (*setup_hdmi_audio)(
+		struct afmt *afmt);
+
+	void (*se_audio_setup)(
+		struct afmt *afmt,
+		unsigned int az_inst,
+		struct audio_info *audio_info);
+
+	void (*audio_mute_control)(
+		struct afmt *afmt,
+		bool mute);
+
+	void (*audio_info_immediate_update)(
+		struct afmt *afmt);
+
+	void (*setup_dp_audio)(
+		struct afmt *afmt);
+
+	void (*afmt_poweron)(
+		struct afmt *afmt);
+
+	void (*afmt_powerdown)(
+		struct afmt *afmt);
+};
+
+struct afmt {
+	const struct afmt_funcs *funcs;
+	struct dc_context *ctx;
+	int inst;
+};
+
+struct dcn30_afmt {
+	struct afmt base;
+	const struct dcn30_afmt_registers *regs;
+	const struct dcn30_afmt_shift *afmt_shift;
+	const struct dcn30_afmt_mask *afmt_mask;
+};
+
+void afmt3_setup_hdmi_audio(
+	struct afmt *afmt);
+
+void afmt3_se_audio_setup(
+	struct afmt *afmt,
+	unsigned int az_inst,
+	struct audio_info *audio_info);
+
+void afmt3_audio_mute_control(
+	struct afmt *afmt,
+	bool mute);
+
+void afmt3_audio_info_immediate_update(
+	struct afmt *afmt);
+
+void afmt3_setup_dp_audio(
+		struct afmt *afmt);
+
+void afmt3_construct(struct dcn30_afmt *afmt3,
+	struct dc_context *ctx,
+	uint32_t inst,
+	const struct dcn30_afmt_registers *afmt_regs,
+	const struct dcn30_afmt_shift *afmt_shift,
+	const struct dcn30_afmt_mask *afmt_mask);
+
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_cm_common.c b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_cm_common.c
new file mode 100644
index 000000000000..0690c346f2c5
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_cm_common.c
@@ -0,0 +1,456 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+#include "core_types.h"
+#include "reg_helper.h"
+#include "dcn30/dcn30_dpp.h"
+#include "basics/conversion.h"
+#include "dcn30/dcn30_cm_common.h"
+#include "custom_float.h"
+
+#define REG(reg) reg
+
+#define CTX \
+	ctx //dpp->base.ctx
+
+#undef FN
+#define FN(reg_name, field_name) \
+	reg->shifts.field_name, reg->masks.field_name
+
+void cm_helper_program_gamcor_xfer_func(
+		struct dc_context *ctx,
+		const struct pwl_params *params,
+		const struct dcn3_xfer_func_reg *reg)
+{
+	uint32_t reg_region_cur;
+	unsigned int i = 0;
+
+	REG_SET_2(reg->start_cntl_b, 0,
+		exp_region_start, params->corner_points[0].blue.custom_float_x,
+		exp_resion_start_segment, 0);
+	REG_SET_2(reg->start_cntl_g, 0,
+		exp_region_start, params->corner_points[0].green.custom_float_x,
+		exp_resion_start_segment, 0);
+	REG_SET_2(reg->start_cntl_r, 0,
+		exp_region_start, params->corner_points[0].red.custom_float_x,
+		exp_resion_start_segment, 0);
+
+	REG_SET(reg->start_slope_cntl_b, 0, //linear slope at start of curve
+		field_region_linear_slope, params->corner_points[0].blue.custom_float_slope);
+	REG_SET(reg->start_slope_cntl_g, 0,
+		field_region_linear_slope, params->corner_points[0].green.custom_float_slope);
+	REG_SET(reg->start_slope_cntl_r, 0,
+		field_region_linear_slope, params->corner_points[0].red.custom_float_slope);
+
+	REG_SET(reg->start_end_cntl1_b, 0,
+		field_region_end_base, params->corner_points[1].blue.custom_float_y);
+	REG_SET(reg->start_end_cntl1_g, 0,
+		field_region_end_base, params->corner_points[1].green.custom_float_y);
+	REG_SET(reg->start_end_cntl1_r, 0,
+		field_region_end_base, params->corner_points[1].red.custom_float_y);
+
+	REG_SET_2(reg->start_end_cntl2_b, 0,
+		field_region_end_slope, params->corner_points[1].blue.custom_float_slope,
+		field_region_end, params->corner_points[1].blue.custom_float_x);
+	REG_SET_2(reg->start_end_cntl2_g, 0,
+		field_region_end_slope, params->corner_points[1].green.custom_float_slope,
+		field_region_end, params->corner_points[1].green.custom_float_x);
+	REG_SET_2(reg->start_end_cntl2_r, 0,
+		field_region_end_slope, params->corner_points[1].red.custom_float_slope,
+		field_region_end, params->corner_points[1].red.custom_float_x);
+
+	for (reg_region_cur = reg->region_start;
+		reg_region_cur <= reg->region_end;
+		reg_region_cur++) {
+
+		const struct gamma_curve *curve0 = &(params->arr_curve_points[2 * i]);
+		const struct gamma_curve *curve1 = &(params->arr_curve_points[(2 * i) + 1]);
+
+		REG_SET_4(reg_region_cur, 0,
+			exp_region0_lut_offset, curve0->offset,
+			exp_region0_num_segments, curve0->segments_num,
+			exp_region1_lut_offset, curve1->offset,
+			exp_region1_num_segments, curve1->segments_num);
+
+		i++;
+	}
+}
+
+/* driver uses 32 regions or less, but DCN HW has 34, extra 2 are set to 0 */
+#define MAX_REGIONS_NUMBER 34
+#define MAX_LOW_POINT      25
+#define NUMBER_REGIONS     32
+#define NUMBER_SW_SEGMENTS 16
+
+bool cm3_helper_translate_curve_to_hw_format(
+				const struct dc_transfer_func *output_tf,
+				struct pwl_params *lut_params, bool fixpoint)
+{
+	struct curve_points3 *corner_points;
+	struct pwl_result_data *rgb_resulted;
+	struct pwl_result_data *rgb;
+	struct pwl_result_data *rgb_plus_1;
+	struct pwl_result_data *rgb_minus_1;
+
+	int32_t region_start, region_end;
+	int32_t i;
+	uint32_t j, k, seg_distr[MAX_REGIONS_NUMBER], increment, start_index, hw_points;
+
+	if (output_tf == NULL || lut_params == NULL || output_tf->type == TF_TYPE_BYPASS)
+		return false;
+
+	corner_points = lut_params->corner_points;
+	rgb_resulted = lut_params->rgb_resulted;
+	hw_points = 0;
+
+	memset(lut_params, 0, sizeof(struct pwl_params));
+	memset(seg_distr, 0, sizeof(seg_distr));
+
+	if (output_tf->tf == TRANSFER_FUNCTION_PQ || output_tf->tf == TRANSFER_FUNCTION_GAMMA22 ||
+		output_tf->tf == TRANSFER_FUNCTION_HLG) {
+		/* 32 segments
+		 * segments are from 2^-25 to 2^7
+		 */
+		for (i = 0; i < NUMBER_REGIONS ; i++)
+			seg_distr[i] = 3;
+
+		region_start = -MAX_LOW_POINT;
+		region_end   = NUMBER_REGIONS - MAX_LOW_POINT;
+	} else {
+		/* 13 segments
+		 * segment is from 2^-12 to 2^0
+		 * There are less than 256 points, for optimization
+		 */
+		const uint8_t SEG_COUNT = 12;
+
+		for (i = 0; i < SEG_COUNT; i++)
+			seg_distr[i] = 4;
+
+		seg_distr[SEG_COUNT] = 1;
+
+		region_start = -SEG_COUNT;
+		region_end = 1;
+	}
+
+	for (i = region_end - region_start; i < MAX_REGIONS_NUMBER ; i++)
+		seg_distr[i] = -1;
+
+	for (k = 0; k < MAX_REGIONS_NUMBER; k++) {
+		if (seg_distr[k] != -1)
+			hw_points += (1 << seg_distr[k]);
+	}
+
+	j = 0;
+	for (k = 0; k < (region_end - region_start); k++) {
+		increment = NUMBER_SW_SEGMENTS / (1 << seg_distr[k]);
+		start_index = (region_start + k + MAX_LOW_POINT) *
+				NUMBER_SW_SEGMENTS;
+		for (i = start_index; i < start_index + NUMBER_SW_SEGMENTS;
+				i += increment) {
+			if (j == hw_points)
+				break;
+			if (i >= TRANSFER_FUNC_POINTS)
+				return false;
+			rgb_resulted[j].red = output_tf->tf_pts.red[i];
+			rgb_resulted[j].green = output_tf->tf_pts.green[i];
+			rgb_resulted[j].blue = output_tf->tf_pts.blue[i];
+			j++;
+		}
+	}
+
+	/* last point */
+	start_index = (region_end + MAX_LOW_POINT) * NUMBER_SW_SEGMENTS;
+	rgb_resulted[hw_points].red = output_tf->tf_pts.red[start_index];
+	rgb_resulted[hw_points].green = output_tf->tf_pts.green[start_index];
+	rgb_resulted[hw_points].blue = output_tf->tf_pts.blue[start_index];
+
+	rgb_resulted[hw_points+1].red = rgb_resulted[hw_points].red;
+	rgb_resulted[hw_points+1].green = rgb_resulted[hw_points].green;
+	rgb_resulted[hw_points+1].blue = rgb_resulted[hw_points].blue;
+
+	// All 3 color channels have same x
+	corner_points[0].red.x = dc_fixpt_pow(dc_fixpt_from_int(2),
+					     dc_fixpt_from_int(region_start));
+	corner_points[0].green.x = corner_points[0].red.x;
+	corner_points[0].blue.x = corner_points[0].red.x;
+
+	corner_points[1].red.x = dc_fixpt_pow(dc_fixpt_from_int(2),
+					     dc_fixpt_from_int(region_end));
+	corner_points[1].green.x = corner_points[1].red.x;
+	corner_points[1].blue.x = corner_points[1].red.x;
+
+	corner_points[0].red.y = rgb_resulted[0].red;
+	corner_points[0].green.y = rgb_resulted[0].green;
+	corner_points[0].blue.y = rgb_resulted[0].blue;
+
+	corner_points[0].red.slope = dc_fixpt_div(corner_points[0].red.y,
+			corner_points[0].red.x);
+	corner_points[0].green.slope = dc_fixpt_div(corner_points[0].green.y,
+			corner_points[0].green.x);
+	corner_points[0].blue.slope = dc_fixpt_div(corner_points[0].blue.y,
+			corner_points[0].blue.x);
+
+	/* see comment above, m_arrPoints[1].y should be the Y value for the
+	 * region end (m_numOfHwPoints), not last HW point(m_numOfHwPoints - 1)
+	 */
+	corner_points[1].red.y = rgb_resulted[hw_points].red;
+	corner_points[1].green.y = rgb_resulted[hw_points].green;
+	corner_points[1].blue.y = rgb_resulted[hw_points].blue;
+	corner_points[1].red.slope = dc_fixpt_zero;
+	corner_points[1].green.slope = dc_fixpt_zero;
+	corner_points[1].blue.slope = dc_fixpt_zero;
+
+	// DCN3+ have 257 pts in lieu of no separate slope registers
+	// Prior HW had 256 base+slope pairs
+	lut_params->hw_points_num = hw_points + 1;
+
+	k = 0;
+	for (i = 1; i < MAX_REGIONS_NUMBER; i++) {
+		if (seg_distr[k] != -1) {
+			lut_params->arr_curve_points[k].segments_num =
+					seg_distr[k];
+			lut_params->arr_curve_points[i].offset =
+					lut_params->arr_curve_points[k].offset + (1 << seg_distr[k]);
+		}
+		k++;
+	}
+
+	if (seg_distr[k] != -1)
+		lut_params->arr_curve_points[k].segments_num = seg_distr[k];
+
+	rgb = rgb_resulted;
+	rgb_plus_1 = rgb_resulted + 1;
+	rgb_minus_1 = rgb;
+
+	if (fixpoint == true) {
+		i = 1;
+		while (i != hw_points + 2) {
+			if (i >= hw_points) {
+				if (dc_fixpt_lt(rgb_plus_1->red, rgb->red))
+					rgb_plus_1->red = dc_fixpt_add(rgb->red,
+							rgb_minus_1->delta_red);
+				if (dc_fixpt_lt(rgb_plus_1->green, rgb->green))
+					rgb_plus_1->green = dc_fixpt_add(rgb->green,
+							rgb_minus_1->delta_green);
+				if (dc_fixpt_lt(rgb_plus_1->blue, rgb->blue))
+					rgb_plus_1->blue = dc_fixpt_add(rgb->blue,
+							rgb_minus_1->delta_blue);
+			}
+
+			rgb->delta_red_reg   = dc_fixpt_clamp_u0d10(rgb->delta_red);
+			rgb->delta_green_reg = dc_fixpt_clamp_u0d10(rgb->delta_green);
+			rgb->delta_blue_reg  = dc_fixpt_clamp_u0d10(rgb->delta_blue);
+			rgb->red_reg         = dc_fixpt_clamp_u0d14(rgb->red);
+			rgb->green_reg       = dc_fixpt_clamp_u0d14(rgb->green);
+			rgb->blue_reg        = dc_fixpt_clamp_u0d14(rgb->blue);
+
+			++rgb_plus_1;
+			rgb_minus_1 = rgb;
+			++rgb;
+			++i;
+		}
+	}
+	cm3_helper_convert_to_custom_float(rgb_resulted,
+						lut_params->corner_points,
+						hw_points+1, fixpoint);
+
+	return true;
+}
+
+bool cm3_helper_convert_to_custom_float(
+		struct pwl_result_data *rgb_resulted,
+		struct curve_points3 *corner_points,
+		uint32_t hw_points_num,
+		bool fixpoint)
+{
+	struct custom_float_format fmt;
+
+	struct pwl_result_data *rgb = rgb_resulted;
+
+	uint32_t i = 0;
+
+	fmt.exponenta_bits = 6;
+	fmt.mantissa_bits = 12;
+	fmt.sign = false;
+
+	/* corner_points[0] - beginning base, slope offset for R,G,B
+	 * corner_points[1] - end base, slope offset for R,G,B
+	 */
+	if (!convert_to_custom_float_format(corner_points[0].red.x, &fmt,
+				&corner_points[0].red.custom_float_x)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+	if (!convert_to_custom_float_format(corner_points[0].green.x, &fmt,
+				&corner_points[0].green.custom_float_x)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+	if (!convert_to_custom_float_format(corner_points[0].blue.x, &fmt,
+				&corner_points[0].blue.custom_float_x)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+
+	if (!convert_to_custom_float_format(corner_points[0].red.offset, &fmt,
+				&corner_points[0].red.custom_float_offset)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+	if (!convert_to_custom_float_format(corner_points[0].green.offset, &fmt,
+				&corner_points[0].green.custom_float_offset)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+	if (!convert_to_custom_float_format(corner_points[0].blue.offset, &fmt,
+				&corner_points[0].blue.custom_float_offset)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+
+	if (!convert_to_custom_float_format(corner_points[0].red.slope, &fmt,
+				&corner_points[0].red.custom_float_slope)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+	if (!convert_to_custom_float_format(corner_points[0].green.slope, &fmt,
+				&corner_points[0].green.custom_float_slope)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+	if (!convert_to_custom_float_format(corner_points[0].blue.slope, &fmt,
+				&corner_points[0].blue.custom_float_slope)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+
+	if (fixpoint == true) {
+		corner_points[1].red.custom_float_y =
+				dc_fixpt_clamp_u0d14(corner_points[1].red.y);
+		corner_points[1].green.custom_float_y =
+				dc_fixpt_clamp_u0d14(corner_points[1].green.y);
+		corner_points[1].blue.custom_float_y =
+				dc_fixpt_clamp_u0d14(corner_points[1].blue.y);
+	} else {
+		if (!convert_to_custom_float_format(corner_points[1].red.y,
+				&fmt, &corner_points[1].red.custom_float_y)) {
+			BREAK_TO_DEBUGGER();
+			return false;
+		}
+		if (!convert_to_custom_float_format(corner_points[1].green.y,
+				&fmt, &corner_points[1].green.custom_float_y)) {
+			BREAK_TO_DEBUGGER();
+			return false;
+		}
+		if (!convert_to_custom_float_format(corner_points[1].blue.y,
+				&fmt, &corner_points[1].blue.custom_float_y)) {
+			BREAK_TO_DEBUGGER();
+			return false;
+		}
+	}
+
+	fmt.mantissa_bits = 10;
+	fmt.sign = false;
+
+	if (!convert_to_custom_float_format(corner_points[1].red.x, &fmt,
+				&corner_points[1].red.custom_float_x)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+	if (!convert_to_custom_float_format(corner_points[1].green.x, &fmt,
+				&corner_points[1].green.custom_float_x)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+	if (!convert_to_custom_float_format(corner_points[1].blue.x, &fmt,
+				&corner_points[1].blue.custom_float_x)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+
+	if (!convert_to_custom_float_format(corner_points[1].red.slope, &fmt,
+				&corner_points[1].red.custom_float_slope)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+	if (!convert_to_custom_float_format(corner_points[1].green.slope, &fmt,
+				&corner_points[1].green.custom_float_slope)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+	if (!convert_to_custom_float_format(corner_points[1].blue.slope, &fmt,
+				&corner_points[1].blue.custom_float_slope)) {
+		BREAK_TO_DEBUGGER();
+		return false;
+	}
+
+	if (hw_points_num == 0 || rgb_resulted == NULL || fixpoint == true)
+		return true;
+
+	fmt.mantissa_bits = 12;
+
+	while (i != hw_points_num) {
+		if (!convert_to_custom_float_format(rgb->red, &fmt,
+						    &rgb->red_reg)) {
+			BREAK_TO_DEBUGGER();
+			return false;
+		}
+
+		if (!convert_to_custom_float_format(rgb->green, &fmt,
+						    &rgb->green_reg)) {
+			BREAK_TO_DEBUGGER();
+			return false;
+		}
+
+		if (!convert_to_custom_float_format(rgb->blue, &fmt,
+						    &rgb->blue_reg)) {
+			BREAK_TO_DEBUGGER();
+			return false;
+		}
+
+		++rgb;
+		++i;
+	}
+
+	return true;
+}
+
+bool is_rgb_equal(const struct pwl_result_data *rgb, uint32_t num)
+{
+	uint32_t i;
+	bool ret = true;
+
+	for (i = 0 ; i < num; i++) {
+		if (rgb[i].red_reg != rgb[i].green_reg ||
+		rgb[i].blue_reg != rgb[i].red_reg  ||
+		rgb[i].blue_reg != rgb[i].green_reg) {
+			ret = false;
+			break;
+		}
+	}
+	return ret;
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_mmhubbub.c b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_mmhubbub.c
new file mode 100644
index 000000000000..6f2a0d5d963b
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_mmhubbub.c
@@ -0,0 +1,239 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+#include "reg_helper.h"
+#include "resource.h"
+#include "mcif_wb.h"
+#include "dcn30_mmhubbub.h"
+
+
+#define REG(reg)\
+	mcif_wb30->mcif_wb_regs->reg
+
+#define CTX \
+	mcif_wb30->base.ctx
+
+#undef FN
+#define FN(reg_name, field_name) \
+	mcif_wb30->mcif_wb_shift->field_name, mcif_wb30->mcif_wb_mask->field_name
+
+#define MCIF_ADDR(addr) (((unsigned long long)addr & 0xffffffffff) + 0xFE) >> 8
+#define MCIF_ADDR_HIGH(addr) (unsigned long long)addr >> 40
+
+/* wbif programming guide:
+ * 1. set up wbif parameter:
+ *    unsigned long long   luma_address[4];       //4 frame buffer
+ *    unsigned long long   chroma_address[4];
+ *    unsigned int	   luma_pitch;
+ *    unsigned int	   chroma_pitch;
+ *    unsigned int         warmup_pitch=0x10;     //256B align, the page size is 4KB when it is 0x10
+ *    unsigned int	   slice_lines;           //slice size
+ *    unsigned int         time_per_pixel;        // time per pixel, in ns
+ *    unsigned int         arbitration_slice;     // 0: 2048 bytes 1: 4096 bytes 2: 8192 Bytes
+ *    unsigned int         max_scaled_time;       // used for QOS generation
+ *    unsigned int         swlock=0x0;
+ *    unsigned int         cli_watermark[4];      //4 group urgent watermark
+ *    unsigned int         pstate_watermark[4];   //4 group pstate watermark
+ *    unsigned int         sw_int_en;             // Software interrupt enable, frame end and overflow
+ *    unsigned int         sw_slice_int_en;       // slice end interrupt enable
+ *    unsigned int         sw_overrun_int_en;     // overrun error interrupt enable
+ *    unsigned int         vce_int_en;            // VCE interrupt enable, frame end and overflow
+ *    unsigned int         vce_slice_int_en;      // VCE slice end interrupt enable, frame end and overflow
+ *
+ * 2. configure wbif register
+ *    a. call mmhubbub_config_wbif()
+ *
+ * 3. Enable wbif
+ *    call set_wbif_bufmgr_enable();
+ *
+ * 4. wbif_dump_status(), option, for debug purpose
+ *    the bufmgr status can show the progress of write back, can be used for debug purpose
+ */
+
+static void mmhubbub3_warmup_mcif(struct mcif_wb *mcif_wb,
+		struct mcif_warmup_params *params)
+{
+	struct dcn30_mmhubbub *mcif_wb30 = TO_DCN30_MMHUBBUB(mcif_wb);
+	union large_integer start_address_shift = {.quad_part = params->start_address.quad_part >> 5};
+
+	/* Set base address and region size for warmup */
+	REG_SET(MMHUBBUB_WARMUP_BASE_ADDR_HIGH, 0, MMHUBBUB_WARMUP_BASE_ADDR_HIGH, start_address_shift.high_part);
+	REG_SET(MMHUBBUB_WARMUP_BASE_ADDR_LOW, 0, MMHUBBUB_WARMUP_BASE_ADDR_LOW, start_address_shift.low_part);
+	REG_SET(MMHUBBUB_WARMUP_ADDR_REGION, 0, MMHUBBUB_WARMUP_ADDR_REGION, params->region_size >> 5);
+//	REG_SET(MMHUBBUB_WARMUP_P_VMID, 0, MMHUBBUB_WARMUP_P_VMID, params->p_vmid);
+
+	/* Set address increment and enable warmup */
+	REG_SET_3(MMHUBBUB_WARMUP_CONTROL_STATUS, 0, MMHUBBUB_WARMUP_EN, true,
+			MMHUBBUB_WARMUP_SW_INT_EN, true,
+			MMHUBBUB_WARMUP_INC_ADDR, params->address_increment >> 5);
+
+	/* Wait for an interrupt to signal warmup is completed */
+	REG_WAIT(MMHUBBUB_WARMUP_CONTROL_STATUS, MMHUBBUB_WARMUP_SW_INT_STATUS, 1, 20, 100);
+
+	/* Acknowledge interrupt */
+	REG_UPDATE(MMHUBBUB_WARMUP_CONTROL_STATUS, MMHUBBUB_WARMUP_SW_INT_ACK, 1);
+
+	/* Disable warmup */
+	REG_UPDATE(MMHUBBUB_WARMUP_CONTROL_STATUS, MMHUBBUB_WARMUP_EN, false);
+}
+
+static void mmhubbub3_config_mcif_buf(struct mcif_wb *mcif_wb,
+		struct mcif_buf_params *params,
+		unsigned int dest_height)
+{
+	struct dcn30_mmhubbub *mcif_wb30 = TO_DCN30_MMHUBBUB(mcif_wb);
+
+	/* buffer address for packing mode or Luma in planar mode */
+	REG_UPDATE(MCIF_WB_BUF_1_ADDR_Y, MCIF_WB_BUF_1_ADDR_Y, MCIF_ADDR(params->luma_address[0]));
+	REG_UPDATE(MCIF_WB_BUF_1_ADDR_Y_HIGH, MCIF_WB_BUF_1_ADDR_Y_HIGH, MCIF_ADDR_HIGH(params->luma_address[0]));
+
+	/* buffer address for Chroma in planar mode (unused in packing mode) */
+	REG_UPDATE(MCIF_WB_BUF_1_ADDR_C, MCIF_WB_BUF_1_ADDR_C, MCIF_ADDR(params->chroma_address[0]));
+	REG_UPDATE(MCIF_WB_BUF_1_ADDR_C_HIGH, MCIF_WB_BUF_1_ADDR_C_HIGH, MCIF_ADDR_HIGH(params->chroma_address[0]));
+
+	/* buffer address for packing mode or Luma in planar mode */
+	REG_UPDATE(MCIF_WB_BUF_2_ADDR_Y, MCIF_WB_BUF_2_ADDR_Y, MCIF_ADDR(params->luma_address[1]));
+	REG_UPDATE(MCIF_WB_BUF_2_ADDR_Y_HIGH, MCIF_WB_BUF_2_ADDR_Y_HIGH, MCIF_ADDR_HIGH(params->luma_address[1]));
+
+	/* buffer address for Chroma in planar mode (unused in packing mode) */
+	REG_UPDATE(MCIF_WB_BUF_2_ADDR_C, MCIF_WB_BUF_2_ADDR_C, MCIF_ADDR(params->chroma_address[1]));
+	REG_UPDATE(MCIF_WB_BUF_2_ADDR_C_HIGH, MCIF_WB_BUF_2_ADDR_C_HIGH, MCIF_ADDR_HIGH(params->chroma_address[1]));
+
+	/* buffer address for packing mode or Luma in planar mode */
+	REG_UPDATE(MCIF_WB_BUF_3_ADDR_Y, MCIF_WB_BUF_3_ADDR_Y, MCIF_ADDR(params->luma_address[2]));
+	REG_UPDATE(MCIF_WB_BUF_3_ADDR_Y_HIGH, MCIF_WB_BUF_3_ADDR_Y_HIGH, MCIF_ADDR_HIGH(params->luma_address[2]));
+
+	/* buffer address for Chroma in planar mode (unused in packing mode) */
+	REG_UPDATE(MCIF_WB_BUF_3_ADDR_C, MCIF_WB_BUF_3_ADDR_C, MCIF_ADDR(params->chroma_address[2]));
+	REG_UPDATE(MCIF_WB_BUF_3_ADDR_C_HIGH, MCIF_WB_BUF_3_ADDR_C_HIGH, MCIF_ADDR_HIGH(params->chroma_address[2]));
+
+	/* buffer address for packing mode or Luma in planar mode */
+	REG_UPDATE(MCIF_WB_BUF_4_ADDR_Y, MCIF_WB_BUF_4_ADDR_Y, MCIF_ADDR(params->luma_address[3]));
+	REG_UPDATE(MCIF_WB_BUF_4_ADDR_Y_HIGH, MCIF_WB_BUF_4_ADDR_Y_HIGH, MCIF_ADDR_HIGH(params->luma_address[3]));
+
+	/* buffer address for Chroma in planar mode (unused in packing mode) */
+	REG_UPDATE(MCIF_WB_BUF_4_ADDR_C, MCIF_WB_BUF_4_ADDR_C, MCIF_ADDR(params->chroma_address[3]));
+	REG_UPDATE(MCIF_WB_BUF_4_ADDR_C_HIGH, MCIF_WB_BUF_4_ADDR_C_HIGH, MCIF_ADDR_HIGH(params->chroma_address[3]));
+
+	/* setup luma & chroma size
+	 * should be enough to contain a whole frame Luma data,
+	 * the programmed value is frame buffer size [27:8], 256-byte aligned
+	 */
+	REG_UPDATE(MCIF_WB_BUF_LUMA_SIZE, MCIF_WB_BUF_LUMA_SIZE, (params->luma_pitch>>8) * dest_height);
+	REG_UPDATE(MCIF_WB_BUF_CHROMA_SIZE, MCIF_WB_BUF_CHROMA_SIZE, (params->chroma_pitch>>8) * dest_height);
+
+	/* enable address fence */
+	REG_UPDATE(MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUF_ADDR_FENCE_EN, 1);
+
+	/* setup pitch, the programmed value is [15:8], 256B align */
+	REG_UPDATE_2(MCIF_WB_BUF_PITCH, MCIF_WB_BUF_LUMA_PITCH, params->luma_pitch >> 8,
+			MCIF_WB_BUF_CHROMA_PITCH, params->chroma_pitch >> 8);
+}
+
+static void mmhubbub3_config_mcif_arb(struct mcif_wb *mcif_wb,
+		struct mcif_arb_params *params)
+{
+	struct dcn30_mmhubbub *mcif_wb30 = TO_DCN30_MMHUBBUB(mcif_wb);
+
+	/* Programmed by the video driver based on the CRTC timing (for DWB) */
+	REG_UPDATE(MCIF_WB_ARBITRATION_CONTROL, MCIF_WB_TIME_PER_PIXEL, params->time_per_pixel);
+
+	/* Programming dwb watermark */
+	/* Watermark to generate urgent in MCIF_WB_CLI, value is determined by MCIF_WB_CLI_WATERMARK_MASK. */
+	/* Program in ns. A formula will be provided in the pseudo code to calculate the value. */
+	REG_UPDATE(MCIF_WB_WATERMARK, MCIF_WB_CLI_WATERMARK_MASK, 0x0);
+	/* urgent_watermarkA */
+	REG_UPDATE(MCIF_WB_WATERMARK, MCIF_WB_CLI_WATERMARK,  params->cli_watermark[0]);
+	REG_UPDATE(MCIF_WB_WATERMARK, MCIF_WB_CLI_WATERMARK_MASK, 0x1);
+	/* urgent_watermarkB */
+	REG_UPDATE(MCIF_WB_WATERMARK, MCIF_WB_CLI_WATERMARK,  params->cli_watermark[1]);
+	REG_UPDATE(MCIF_WB_WATERMARK, MCIF_WB_CLI_WATERMARK_MASK, 0x2);
+	/* urgent_watermarkC */
+	REG_UPDATE(MCIF_WB_WATERMARK, MCIF_WB_CLI_WATERMARK,  params->cli_watermark[2]);
+	REG_UPDATE(MCIF_WB_WATERMARK, MCIF_WB_CLI_WATERMARK_MASK, 0x3);
+	/* urgent_watermarkD */
+	REG_UPDATE(MCIF_WB_WATERMARK, MCIF_WB_CLI_WATERMARK,  params->cli_watermark[3]);
+
+	/* Programming nb pstate watermark */
+	/* nbp_state_change_watermarkA */
+	REG_UPDATE(MCIF_WB_NB_PSTATE_LATENCY_WATERMARK, NB_PSTATE_CHANGE_WATERMARK_MASK, 0x0);
+	REG_UPDATE(MCIF_WB_NB_PSTATE_LATENCY_WATERMARK,
+			NB_PSTATE_CHANGE_REFRESH_WATERMARK, params->pstate_watermark[0]);
+	/* nbp_state_change_watermarkB */
+	REG_UPDATE(MCIF_WB_NB_PSTATE_LATENCY_WATERMARK, NB_PSTATE_CHANGE_WATERMARK_MASK, 0x1);
+	REG_UPDATE(MCIF_WB_NB_PSTATE_LATENCY_WATERMARK,
+			NB_PSTATE_CHANGE_REFRESH_WATERMARK, params->pstate_watermark[1]);
+	/* nbp_state_change_watermarkC */
+	REG_UPDATE(MCIF_WB_NB_PSTATE_LATENCY_WATERMARK, NB_PSTATE_CHANGE_WATERMARK_MASK, 0x2);
+	REG_UPDATE(MCIF_WB_NB_PSTATE_LATENCY_WATERMARK,
+			NB_PSTATE_CHANGE_REFRESH_WATERMARK, params->pstate_watermark[2]);
+	/* nbp_state_change_watermarkD */
+	REG_UPDATE(MCIF_WB_NB_PSTATE_LATENCY_WATERMARK, NB_PSTATE_CHANGE_WATERMARK_MASK, 0x3);
+	REG_UPDATE(MCIF_WB_NB_PSTATE_LATENCY_WATERMARK,
+			NB_PSTATE_CHANGE_REFRESH_WATERMARK, params->pstate_watermark[3]);
+
+	/* dram_speed_change_duration */
+	REG_UPDATE(MCIF_WB_DRAM_SPEED_CHANGE_DURATION_VBI,
+			MCIF_WB_DRAM_SPEED_CHANGE_DURATION_VBI, params->dram_speed_change_duration);
+
+	/* max_scaled_time */
+	REG_UPDATE(MULTI_LEVEL_QOS_CTRL, MAX_SCALED_TIME_TO_URGENT, params->max_scaled_time);
+
+	/* slice_lines */
+	REG_UPDATE(MCIF_WB_BUFMGR_VCE_CONTROL, MCIF_WB_BUFMGR_SLICE_SIZE, params->slice_lines-1);
+
+	/* Set arbitration unit for Luma/Chroma */
+	/* arb_unit=2 should be chosen for more efficiency */
+	/* Arbitration size, 0: 2048 bytes 1: 4096 bytes 2: 8192 Bytes */
+	REG_UPDATE(MCIF_WB_ARBITRATION_CONTROL, MCIF_WB_CLIENT_ARBITRATION_SLICE,  params->arbitration_slice);
+}
+
+static const struct mcif_wb_funcs dcn30_mmhubbub_funcs = {
+	.warmup_mcif		= mmhubbub3_warmup_mcif,
+	.enable_mcif		= mmhubbub2_enable_mcif,
+	.disable_mcif		= mmhubbub2_disable_mcif,
+	.config_mcif_buf	= mmhubbub3_config_mcif_buf,
+	.config_mcif_arb	= mmhubbub3_config_mcif_arb,
+	.config_mcif_irq	= mmhubbub2_config_mcif_irq,
+	.dump_frame		= mcifwb2_dump_frame,
+};
+
+void dcn30_mmhubbub_construct(struct dcn30_mmhubbub *mcif_wb30,
+		struct dc_context *ctx,
+		const struct dcn30_mmhubbub_registers *mcif_wb_regs,
+		const struct dcn30_mmhubbub_shift *mcif_wb_shift,
+		const struct dcn30_mmhubbub_mask *mcif_wb_mask,
+		int inst)
+{
+	mcif_wb30->base.ctx = ctx;
+
+	mcif_wb30->base.inst = inst;
+	mcif_wb30->base.funcs = &dcn30_mmhubbub_funcs;
+
+	mcif_wb30->mcif_wb_regs = mcif_wb_regs;
+	mcif_wb30->mcif_wb_shift = mcif_wb_shift;
+	mcif_wb30->mcif_wb_mask = mcif_wb_mask;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_mmhubbub.h b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_mmhubbub.h
new file mode 100644
index 000000000000..376620a8f02f
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_mmhubbub.h
@@ -0,0 +1,448 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_MCIF_WB_DCN30_H__
+#define __DC_MCIF_WB_DCN30_H__
+
+#include "dcn20/dcn20_mmhubbub.h"
+
+#define TO_DCN30_MMHUBBUB(mcif_wb_base) \
+	container_of(mcif_wb_base, struct dcn30_mmhubbub, base)
+
+#define MCIF_WB_COMMON_REG_LIST_DCN3_0(inst) \
+	SRI(MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUFMGR_STATUS, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_PITCH, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_1_STATUS, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_1_STATUS2, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_2_STATUS, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_2_STATUS2, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_3_STATUS, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_3_STATUS2, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_4_STATUS, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_4_STATUS2, MCIF_WB, inst),\
+	SRI(MCIF_WB_ARBITRATION_CONTROL, MCIF_WB, inst),\
+	SRI(MCIF_WB_SCLK_CHANGE, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_1_ADDR_Y, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_1_ADDR_C, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_2_ADDR_Y, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_2_ADDR_C, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_3_ADDR_Y, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_3_ADDR_C, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_4_ADDR_Y, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_4_ADDR_C, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUFMGR_VCE_CONTROL, MCIF_WB, inst),\
+	SRI2(MCIF_WB_NB_PSTATE_LATENCY_WATERMARK, MMHUBBUB, inst),\
+	SRI(MCIF_WB_NB_PSTATE_CONTROL, MCIF_WB, inst),\
+	SRI2(MCIF_WB_WATERMARK, MMHUBBUB, inst),\
+	SRI(MCIF_WB_CLOCK_GATER_CONTROL, MCIF_WB, inst),\
+	SRI(MCIF_WB_SELF_REFRESH_CONTROL, MCIF_WB, inst),\
+	SRI(MULTI_LEVEL_QOS_CTRL, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_LUMA_SIZE, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_CHROMA_SIZE, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_1_ADDR_Y_HIGH, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_1_ADDR_C_HIGH, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_2_ADDR_Y_HIGH, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_2_ADDR_C_HIGH, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_3_ADDR_Y_HIGH, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_3_ADDR_C_HIGH, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_4_ADDR_Y_HIGH, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_4_ADDR_C_HIGH, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_1_RESOLUTION, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_2_RESOLUTION, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_3_RESOLUTION, MCIF_WB, inst),\
+	SRI(MCIF_WB_BUF_4_RESOLUTION, MCIF_WB, inst),\
+	SRI2(MMHUBBUB_MEM_PWR_CNTL, MMHUBBUB, inst),\
+	SRI2(MMHUBBUB_WARMUP_ADDR_REGION, MMHUBBUB, inst),\
+	SRI2(MMHUBBUB_WARMUP_BASE_ADDR_HIGH, MMHUBBUB, inst),\
+	SRI2(MMHUBBUB_WARMUP_BASE_ADDR_LOW, MMHUBBUB, inst),\
+	SRI2(MMHUBBUB_WARMUP_CONTROL_STATUS, MMHUBBUB, inst),\
+	SRI2(MMHUBBUB_WARMUP_P_VMID, MMHUBBUB, inst),\
+	SRI(MCIF_WB_DRAM_SPEED_CHANGE_DURATION_VBI, MCIF_WB, inst)
+
+#define MCIF_WB_COMMON_REG_LIST_DCN30(inst) \
+	SRI2(MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUFMGR_STATUS, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_PITCH, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_1_STATUS, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_1_STATUS2, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_2_STATUS, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_2_STATUS2, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_3_STATUS, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_3_STATUS2, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_4_STATUS, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_4_STATUS2, MCIF_WB, inst),\
+	SRI2(MCIF_WB_ARBITRATION_CONTROL, MCIF_WB, inst),\
+	SRI2(MCIF_WB_SCLK_CHANGE, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_1_ADDR_Y, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_1_ADDR_C, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_2_ADDR_Y, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_2_ADDR_C, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_3_ADDR_Y, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_3_ADDR_C, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_4_ADDR_Y, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_4_ADDR_C, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUFMGR_VCE_CONTROL, MCIF_WB, inst),\
+	SRI2(MCIF_WB_NB_PSTATE_LATENCY_WATERMARK, MMHUBBUB, inst),\
+	SRI2(MCIF_WB_NB_PSTATE_CONTROL, MCIF_WB, inst),\
+	SRI2(MCIF_WB_WATERMARK, MMHUBBUB, inst),\
+	SRI2(MCIF_WB_CLOCK_GATER_CONTROL, MCIF_WB, inst),\
+	SRI2(MCIF_WB_SELF_REFRESH_CONTROL, MCIF_WB, inst),\
+	SRI2(MULTI_LEVEL_QOS_CTRL, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_LUMA_SIZE, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_CHROMA_SIZE, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_1_ADDR_Y_HIGH, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_1_ADDR_C_HIGH, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_2_ADDR_Y_HIGH, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_2_ADDR_C_HIGH, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_3_ADDR_Y_HIGH, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_3_ADDR_C_HIGH, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_4_ADDR_Y_HIGH, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_4_ADDR_C_HIGH, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_1_RESOLUTION, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_2_RESOLUTION, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_3_RESOLUTION, MCIF_WB, inst),\
+	SRI2(MCIF_WB_BUF_4_RESOLUTION, MCIF_WB, inst),\
+	SRI2(MMHUBBUB_MEM_PWR_CNTL, MMHUBBUB, inst),\
+	SRI2(MMHUBBUB_WARMUP_ADDR_REGION, MMHUBBUB, inst),\
+	SRI2(MMHUBBUB_WARMUP_BASE_ADDR_HIGH, MMHUBBUB, inst),\
+	SRI2(MMHUBBUB_WARMUP_BASE_ADDR_LOW, MMHUBBUB, inst),\
+	SRI2(MMHUBBUB_WARMUP_CONTROL_STATUS, MMHUBBUB, inst),\
+	SRI2(MCIF_WB_DRAM_SPEED_CHANGE_DURATION_VBI, MCIF_WB, inst)
+
+#define MCIF_WB_COMMON_MASK_SH_LIST_DCN3_0(mask_sh) \
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUFMGR_ENABLE, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUFMGR_SW_INT_EN, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUFMGR_SW_INT_ACK, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUFMGR_SW_SLICE_INT_EN, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUFMGR_SW_OVERRUN_INT_EN, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUFMGR_SW_LOCK, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUF_ADDR_FENCE_EN, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_STATUS, MCIF_WB_BUFMGR_VCE_INT_STATUS, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_STATUS, MCIF_WB_BUFMGR_SW_INT_STATUS, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_STATUS, MCIF_WB_BUFMGR_SW_OVERRUN_INT_STATUS, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_STATUS, MCIF_WB_BUFMGR_CUR_BUF, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_STATUS, MCIF_WB_BUFMGR_BUFTAG, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_STATUS, MCIF_WB_BUFMGR_CUR_LINE_L, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_STATUS, MCIF_WB_BUFMGR_NEXT_BUF, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_PITCH, MCIF_WB_BUF_LUMA_PITCH, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_PITCH, MCIF_WB_BUF_CHROMA_PITCH, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_STATUS, MCIF_WB_BUF_1_ACTIVE, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_STATUS, MCIF_WB_BUF_1_SW_LOCKED, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_STATUS, MCIF_WB_BUF_1_VCE_LOCKED, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_STATUS, MCIF_WB_BUF_1_OVERFLOW, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_STATUS, MCIF_WB_BUF_1_DISABLE, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_STATUS, MCIF_WB_BUF_1_MODE, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_STATUS, MCIF_WB_BUF_1_BUFTAG, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_STATUS, MCIF_WB_BUF_1_NXT_BUF, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_STATUS, MCIF_WB_BUF_1_CUR_LINE_L, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_STATUS2, MCIF_WB_BUF_1_NEW_CONTENT, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_STATUS2, MCIF_WB_BUF_1_COLOR_DEPTH, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_STATUS2, MCIF_WB_BUF_1_TMZ_BLACK_PIXEL, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_STATUS2, MCIF_WB_BUF_1_TMZ, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_STATUS2, MCIF_WB_BUF_1_Y_OVERRUN, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_STATUS2, MCIF_WB_BUF_1_C_OVERRUN, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_STATUS, MCIF_WB_BUF_2_ACTIVE, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_STATUS, MCIF_WB_BUF_2_SW_LOCKED, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_STATUS, MCIF_WB_BUF_2_VCE_LOCKED, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_STATUS, MCIF_WB_BUF_2_OVERFLOW, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_STATUS, MCIF_WB_BUF_2_DISABLE, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_STATUS, MCIF_WB_BUF_2_MODE, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_STATUS, MCIF_WB_BUF_2_BUFTAG, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_STATUS, MCIF_WB_BUF_2_NXT_BUF, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_STATUS, MCIF_WB_BUF_2_CUR_LINE_L, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_STATUS2, MCIF_WB_BUF_2_NEW_CONTENT, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_STATUS2, MCIF_WB_BUF_2_COLOR_DEPTH, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_STATUS2, MCIF_WB_BUF_2_TMZ_BLACK_PIXEL, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_STATUS2, MCIF_WB_BUF_2_TMZ, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_STATUS2, MCIF_WB_BUF_2_Y_OVERRUN, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_STATUS2, MCIF_WB_BUF_2_C_OVERRUN, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_STATUS, MCIF_WB_BUF_3_ACTIVE, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_STATUS, MCIF_WB_BUF_3_SW_LOCKED, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_STATUS, MCIF_WB_BUF_3_VCE_LOCKED, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_STATUS, MCIF_WB_BUF_3_OVERFLOW, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_STATUS, MCIF_WB_BUF_3_DISABLE, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_STATUS, MCIF_WB_BUF_3_MODE, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_STATUS, MCIF_WB_BUF_3_BUFTAG, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_STATUS, MCIF_WB_BUF_3_NXT_BUF, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_STATUS, MCIF_WB_BUF_3_CUR_LINE_L, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_STATUS2, MCIF_WB_BUF_3_NEW_CONTENT, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_STATUS2, MCIF_WB_BUF_3_COLOR_DEPTH, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_STATUS2, MCIF_WB_BUF_3_TMZ_BLACK_PIXEL, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_STATUS2, MCIF_WB_BUF_3_TMZ, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_STATUS2, MCIF_WB_BUF_3_Y_OVERRUN, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_STATUS2, MCIF_WB_BUF_3_C_OVERRUN, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_STATUS, MCIF_WB_BUF_4_ACTIVE, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_STATUS, MCIF_WB_BUF_4_SW_LOCKED, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_STATUS, MCIF_WB_BUF_4_VCE_LOCKED, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_STATUS, MCIF_WB_BUF_4_OVERFLOW, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_STATUS, MCIF_WB_BUF_4_DISABLE, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_STATUS, MCIF_WB_BUF_4_MODE, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_STATUS, MCIF_WB_BUF_4_BUFTAG, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_STATUS, MCIF_WB_BUF_4_NXT_BUF, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_STATUS, MCIF_WB_BUF_4_CUR_LINE_L, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_STATUS2, MCIF_WB_BUF_4_NEW_CONTENT, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_STATUS2, MCIF_WB_BUF_4_COLOR_DEPTH, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_STATUS2, MCIF_WB_BUF_4_TMZ_BLACK_PIXEL, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_STATUS2, MCIF_WB_BUF_4_TMZ, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_STATUS2, MCIF_WB_BUF_4_Y_OVERRUN, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_STATUS2, MCIF_WB_BUF_4_C_OVERRUN, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_ARBITRATION_CONTROL, MCIF_WB_CLIENT_ARBITRATION_SLICE, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_ARBITRATION_CONTROL, MCIF_WB_TIME_PER_PIXEL, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_SCLK_CHANGE, WM_CHANGE_ACK_FORCE_ON, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_ADDR_Y, MCIF_WB_BUF_1_ADDR_Y, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_ADDR_C, MCIF_WB_BUF_1_ADDR_C, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_ADDR_Y, MCIF_WB_BUF_2_ADDR_Y, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_ADDR_C, MCIF_WB_BUF_2_ADDR_C, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_ADDR_Y, MCIF_WB_BUF_3_ADDR_Y, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_ADDR_C, MCIF_WB_BUF_3_ADDR_C, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_ADDR_Y, MCIF_WB_BUF_4_ADDR_Y, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_ADDR_C, MCIF_WB_BUF_4_ADDR_C, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_VCE_CONTROL, MCIF_WB_BUFMGR_VCE_SLICE_INT_EN, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUFMGR_VCE_CONTROL, MCIF_WB_BUFMGR_SLICE_SIZE, mask_sh),\
+	SF(MCIF_WB_NB_PSTATE_LATENCY_WATERMARK, NB_PSTATE_CHANGE_REFRESH_WATERMARK, mask_sh),\
+	SF(MCIF_WB_NB_PSTATE_LATENCY_WATERMARK, NB_PSTATE_CHANGE_WATERMARK_MASK, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_NB_PSTATE_CONTROL, NB_PSTATE_CHANGE_URGENT_DURING_REQUEST, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_NB_PSTATE_CONTROL, NB_PSTATE_CHANGE_FORCE_ON, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_NB_PSTATE_CONTROL, NB_PSTATE_ALLOW_FOR_URGENT, mask_sh),\
+	SF(MCIF_WB_WATERMARK, MCIF_WB_CLI_WATERMARK, mask_sh),\
+	SF(MCIF_WB_WATERMARK, MCIF_WB_CLI_WATERMARK_MASK, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_CLOCK_GATER_CONTROL, MCIF_WB_CLI_CLOCK_GATER_OVERRIDE, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_SELF_REFRESH_CONTROL, DIS_REFRESH_UNDER_NBPREQ, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_SELF_REFRESH_CONTROL, PERFRAME_SELF_REFRESH, mask_sh),\
+	SF(MCIF_WB0_MULTI_LEVEL_QOS_CTRL, MAX_SCALED_TIME_TO_URGENT, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_LUMA_SIZE, MCIF_WB_BUF_LUMA_SIZE, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_CHROMA_SIZE, MCIF_WB_BUF_CHROMA_SIZE, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_ADDR_Y_HIGH, MCIF_WB_BUF_1_ADDR_Y_HIGH, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_ADDR_C_HIGH, MCIF_WB_BUF_1_ADDR_C_HIGH, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_ADDR_Y_HIGH, MCIF_WB_BUF_2_ADDR_Y_HIGH, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_ADDR_C_HIGH, MCIF_WB_BUF_2_ADDR_C_HIGH, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_ADDR_Y_HIGH, MCIF_WB_BUF_3_ADDR_Y_HIGH, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_ADDR_C_HIGH, MCIF_WB_BUF_3_ADDR_C_HIGH, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_ADDR_Y_HIGH, MCIF_WB_BUF_4_ADDR_Y_HIGH, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_ADDR_C_HIGH, MCIF_WB_BUF_4_ADDR_C_HIGH, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_RESOLUTION, MCIF_WB_BUF_1_RESOLUTION_WIDTH, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_1_RESOLUTION, MCIF_WB_BUF_1_RESOLUTION_HEIGHT, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_RESOLUTION, MCIF_WB_BUF_2_RESOLUTION_WIDTH, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_2_RESOLUTION, MCIF_WB_BUF_2_RESOLUTION_HEIGHT, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_RESOLUTION, MCIF_WB_BUF_3_RESOLUTION_WIDTH, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_3_RESOLUTION, MCIF_WB_BUF_3_RESOLUTION_HEIGHT, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_RESOLUTION, MCIF_WB_BUF_4_RESOLUTION_WIDTH, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_BUF_4_RESOLUTION, MCIF_WB_BUF_4_RESOLUTION_HEIGHT, mask_sh),\
+	SF(MMHUBBUB_MEM_PWR_CNTL, WBIF_WHOLE_BUF_MODE, mask_sh),\
+	SF(MMHUBBUB_WARMUP_ADDR_REGION, MMHUBBUB_WARMUP_ADDR_REGION, mask_sh),\
+	SF(MMHUBBUB_WARMUP_BASE_ADDR_HIGH, MMHUBBUB_WARMUP_BASE_ADDR_HIGH, mask_sh),\
+	SF(MMHUBBUB_WARMUP_BASE_ADDR_LOW, MMHUBBUB_WARMUP_BASE_ADDR_LOW, mask_sh),\
+	SF(MMHUBBUB_WARMUP_CONTROL_STATUS, MMHUBBUB_WARMUP_EN, mask_sh),\
+	SF(MMHUBBUB_WARMUP_CONTROL_STATUS, MMHUBBUB_WARMUP_SW_INT_EN, mask_sh),\
+	SF(MMHUBBUB_WARMUP_CONTROL_STATUS, MMHUBBUB_WARMUP_SW_INT_STATUS, mask_sh),\
+	SF(MMHUBBUB_WARMUP_CONTROL_STATUS, MMHUBBUB_WARMUP_SW_INT_ACK, mask_sh),\
+	SF(MMHUBBUB_WARMUP_CONTROL_STATUS, MMHUBBUB_WARMUP_INC_ADDR, mask_sh),\
+	SF(MMHUBBUB_WARMUP_P_VMID, MMHUBBUB_WARMUP_P_VMID, mask_sh),\
+	SF(MCIF_WB0_MCIF_WB_DRAM_SPEED_CHANGE_DURATION_VBI, MCIF_WB_DRAM_SPEED_CHANGE_DURATION_VBI, mask_sh)
+
+
+#define MCIF_WB_COMMON_MASK_SH_LIST_DCN30(mask_sh) \
+	SF(MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUFMGR_ENABLE, mask_sh),\
+	SF(MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUFMGR_SW_INT_EN, mask_sh),\
+	SF(MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUFMGR_SW_INT_ACK, mask_sh),\
+	SF(MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUFMGR_SW_SLICE_INT_EN, mask_sh),\
+	SF(MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUFMGR_SW_OVERRUN_INT_EN, mask_sh),\
+	SF(MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUFMGR_SW_LOCK, mask_sh),\
+	SF(MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB_BUF_ADDR_FENCE_EN, mask_sh),\
+	SF(MCIF_WB_BUFMGR_STATUS, MCIF_WB_BUFMGR_VCE_INT_STATUS, mask_sh),\
+	SF(MCIF_WB_BUFMGR_STATUS, MCIF_WB_BUFMGR_SW_INT_STATUS, mask_sh),\
+	SF(MCIF_WB_BUFMGR_STATUS, MCIF_WB_BUFMGR_SW_OVERRUN_INT_STATUS, mask_sh),\
+	SF(MCIF_WB_BUFMGR_STATUS, MCIF_WB_BUFMGR_CUR_BUF, mask_sh),\
+	SF(MCIF_WB_BUFMGR_STATUS, MCIF_WB_BUFMGR_BUFTAG, mask_sh),\
+	SF(MCIF_WB_BUFMGR_STATUS, MCIF_WB_BUFMGR_CUR_LINE_L, mask_sh),\
+	SF(MCIF_WB_BUFMGR_STATUS, MCIF_WB_BUFMGR_NEXT_BUF, mask_sh),\
+	SF(MCIF_WB_BUF_PITCH, MCIF_WB_BUF_LUMA_PITCH, mask_sh),\
+	SF(MCIF_WB_BUF_PITCH, MCIF_WB_BUF_CHROMA_PITCH, mask_sh),\
+	SF(MCIF_WB_BUF_1_STATUS, MCIF_WB_BUF_1_ACTIVE, mask_sh),\
+	SF(MCIF_WB_BUF_1_STATUS, MCIF_WB_BUF_1_SW_LOCKED, mask_sh),\
+	SF(MCIF_WB_BUF_1_STATUS, MCIF_WB_BUF_1_VCE_LOCKED, mask_sh),\
+	SF(MCIF_WB_BUF_1_STATUS, MCIF_WB_BUF_1_OVERFLOW, mask_sh),\
+	SF(MCIF_WB_BUF_1_STATUS, MCIF_WB_BUF_1_DISABLE, mask_sh),\
+	SF(MCIF_WB_BUF_1_STATUS, MCIF_WB_BUF_1_MODE, mask_sh),\
+	SF(MCIF_WB_BUF_1_STATUS, MCIF_WB_BUF_1_BUFTAG, mask_sh),\
+	SF(MCIF_WB_BUF_1_STATUS, MCIF_WB_BUF_1_NXT_BUF, mask_sh),\
+	SF(MCIF_WB_BUF_1_STATUS, MCIF_WB_BUF_1_CUR_LINE_L, mask_sh),\
+	SF(MCIF_WB_BUF_1_STATUS2, MCIF_WB_BUF_1_NEW_CONTENT, mask_sh),\
+	SF(MCIF_WB_BUF_1_STATUS2, MCIF_WB_BUF_1_COLOR_DEPTH, mask_sh),\
+	SF(MCIF_WB_BUF_1_STATUS2, MCIF_WB_BUF_1_TMZ_BLACK_PIXEL, mask_sh),\
+	SF(MCIF_WB_BUF_1_STATUS2, MCIF_WB_BUF_1_TMZ, mask_sh),\
+	SF(MCIF_WB_BUF_1_STATUS2, MCIF_WB_BUF_1_Y_OVERRUN, mask_sh),\
+	SF(MCIF_WB_BUF_1_STATUS2, MCIF_WB_BUF_1_C_OVERRUN, mask_sh),\
+	SF(MCIF_WB_BUF_2_STATUS, MCIF_WB_BUF_2_ACTIVE, mask_sh),\
+	SF(MCIF_WB_BUF_2_STATUS, MCIF_WB_BUF_2_SW_LOCKED, mask_sh),\
+	SF(MCIF_WB_BUF_2_STATUS, MCIF_WB_BUF_2_VCE_LOCKED, mask_sh),\
+	SF(MCIF_WB_BUF_2_STATUS, MCIF_WB_BUF_2_OVERFLOW, mask_sh),\
+	SF(MCIF_WB_BUF_2_STATUS, MCIF_WB_BUF_2_DISABLE, mask_sh),\
+	SF(MCIF_WB_BUF_2_STATUS, MCIF_WB_BUF_2_MODE, mask_sh),\
+	SF(MCIF_WB_BUF_2_STATUS, MCIF_WB_BUF_2_BUFTAG, mask_sh),\
+	SF(MCIF_WB_BUF_2_STATUS, MCIF_WB_BUF_2_NXT_BUF, mask_sh),\
+	SF(MCIF_WB_BUF_2_STATUS, MCIF_WB_BUF_2_CUR_LINE_L, mask_sh),\
+	SF(MCIF_WB_BUF_2_STATUS2, MCIF_WB_BUF_2_NEW_CONTENT, mask_sh),\
+	SF(MCIF_WB_BUF_2_STATUS2, MCIF_WB_BUF_2_COLOR_DEPTH, mask_sh),\
+	SF(MCIF_WB_BUF_2_STATUS2, MCIF_WB_BUF_2_TMZ_BLACK_PIXEL, mask_sh),\
+	SF(MCIF_WB_BUF_2_STATUS2, MCIF_WB_BUF_2_TMZ, mask_sh),\
+	SF(MCIF_WB_BUF_2_STATUS2, MCIF_WB_BUF_2_Y_OVERRUN, mask_sh),\
+	SF(MCIF_WB_BUF_2_STATUS2, MCIF_WB_BUF_2_C_OVERRUN, mask_sh),\
+	SF(MCIF_WB_BUF_3_STATUS, MCIF_WB_BUF_3_ACTIVE, mask_sh),\
+	SF(MCIF_WB_BUF_3_STATUS, MCIF_WB_BUF_3_SW_LOCKED, mask_sh),\
+	SF(MCIF_WB_BUF_3_STATUS, MCIF_WB_BUF_3_VCE_LOCKED, mask_sh),\
+	SF(MCIF_WB_BUF_3_STATUS, MCIF_WB_BUF_3_OVERFLOW, mask_sh),\
+	SF(MCIF_WB_BUF_3_STATUS, MCIF_WB_BUF_3_DISABLE, mask_sh),\
+	SF(MCIF_WB_BUF_3_STATUS, MCIF_WB_BUF_3_MODE, mask_sh),\
+	SF(MCIF_WB_BUF_3_STATUS, MCIF_WB_BUF_3_BUFTAG, mask_sh),\
+	SF(MCIF_WB_BUF_3_STATUS, MCIF_WB_BUF_3_NXT_BUF, mask_sh),\
+	SF(MCIF_WB_BUF_3_STATUS, MCIF_WB_BUF_3_CUR_LINE_L, mask_sh),\
+	SF(MCIF_WB_BUF_3_STATUS2, MCIF_WB_BUF_3_NEW_CONTENT, mask_sh),\
+	SF(MCIF_WB_BUF_3_STATUS2, MCIF_WB_BUF_3_COLOR_DEPTH, mask_sh),\
+	SF(MCIF_WB_BUF_3_STATUS2, MCIF_WB_BUF_3_TMZ_BLACK_PIXEL, mask_sh),\
+	SF(MCIF_WB_BUF_3_STATUS2, MCIF_WB_BUF_3_TMZ, mask_sh),\
+	SF(MCIF_WB_BUF_3_STATUS2, MCIF_WB_BUF_3_Y_OVERRUN, mask_sh),\
+	SF(MCIF_WB_BUF_3_STATUS2, MCIF_WB_BUF_3_C_OVERRUN, mask_sh),\
+	SF(MCIF_WB_BUF_4_STATUS, MCIF_WB_BUF_4_ACTIVE, mask_sh),\
+	SF(MCIF_WB_BUF_4_STATUS, MCIF_WB_BUF_4_SW_LOCKED, mask_sh),\
+	SF(MCIF_WB_BUF_4_STATUS, MCIF_WB_BUF_4_VCE_LOCKED, mask_sh),\
+	SF(MCIF_WB_BUF_4_STATUS, MCIF_WB_BUF_4_OVERFLOW, mask_sh),\
+	SF(MCIF_WB_BUF_4_STATUS, MCIF_WB_BUF_4_DISABLE, mask_sh),\
+	SF(MCIF_WB_BUF_4_STATUS, MCIF_WB_BUF_4_MODE, mask_sh),\
+	SF(MCIF_WB_BUF_4_STATUS, MCIF_WB_BUF_4_BUFTAG, mask_sh),\
+	SF(MCIF_WB_BUF_4_STATUS, MCIF_WB_BUF_4_NXT_BUF, mask_sh),\
+	SF(MCIF_WB_BUF_4_STATUS, MCIF_WB_BUF_4_CUR_LINE_L, mask_sh),\
+	SF(MCIF_WB_BUF_4_STATUS2, MCIF_WB_BUF_4_NEW_CONTENT, mask_sh),\
+	SF(MCIF_WB_BUF_4_STATUS2, MCIF_WB_BUF_4_COLOR_DEPTH, mask_sh),\
+	SF(MCIF_WB_BUF_4_STATUS2, MCIF_WB_BUF_4_TMZ_BLACK_PIXEL, mask_sh),\
+	SF(MCIF_WB_BUF_4_STATUS2, MCIF_WB_BUF_4_TMZ, mask_sh),\
+	SF(MCIF_WB_BUF_4_STATUS2, MCIF_WB_BUF_4_Y_OVERRUN, mask_sh),\
+	SF(MCIF_WB_BUF_4_STATUS2, MCIF_WB_BUF_4_C_OVERRUN, mask_sh),\
+	SF(MCIF_WB_ARBITRATION_CONTROL, MCIF_WB_CLIENT_ARBITRATION_SLICE, mask_sh),\
+	SF(MCIF_WB_ARBITRATION_CONTROL, MCIF_WB_TIME_PER_PIXEL, mask_sh),\
+	SF(MCIF_WB_SCLK_CHANGE, WM_CHANGE_ACK_FORCE_ON, mask_sh),\
+	SF(MCIF_WB_BUF_1_ADDR_Y, MCIF_WB_BUF_1_ADDR_Y, mask_sh),\
+	SF(MCIF_WB_BUF_1_ADDR_C, MCIF_WB_BUF_1_ADDR_C, mask_sh),\
+	SF(MCIF_WB_BUF_2_ADDR_Y, MCIF_WB_BUF_2_ADDR_Y, mask_sh),\
+	SF(MCIF_WB_BUF_2_ADDR_C, MCIF_WB_BUF_2_ADDR_C, mask_sh),\
+	SF(MCIF_WB_BUF_3_ADDR_Y, MCIF_WB_BUF_3_ADDR_Y, mask_sh),\
+	SF(MCIF_WB_BUF_3_ADDR_C, MCIF_WB_BUF_3_ADDR_C, mask_sh),\
+	SF(MCIF_WB_BUF_4_ADDR_Y, MCIF_WB_BUF_4_ADDR_Y, mask_sh),\
+	SF(MCIF_WB_BUF_4_ADDR_C, MCIF_WB_BUF_4_ADDR_C, mask_sh),\
+	SF(MCIF_WB_BUFMGR_VCE_CONTROL, MCIF_WB_BUFMGR_VCE_SLICE_INT_EN, mask_sh),\
+	SF(MCIF_WB_BUFMGR_VCE_CONTROL, MCIF_WB_BUFMGR_SLICE_SIZE, mask_sh),\
+	SF(MCIF_WB_NB_PSTATE_LATENCY_WATERMARK, NB_PSTATE_CHANGE_REFRESH_WATERMARK, mask_sh),\
+	SF(MCIF_WB_NB_PSTATE_LATENCY_WATERMARK, NB_PSTATE_CHANGE_WATERMARK_MASK, mask_sh),\
+	SF(MCIF_WB_NB_PSTATE_CONTROL, NB_PSTATE_CHANGE_URGENT_DURING_REQUEST, mask_sh),\
+	SF(MCIF_WB_NB_PSTATE_CONTROL, NB_PSTATE_CHANGE_FORCE_ON, mask_sh),\
+	SF(MCIF_WB_NB_PSTATE_CONTROL, NB_PSTATE_ALLOW_FOR_URGENT, mask_sh),\
+	SF(MCIF_WB_WATERMARK, MCIF_WB_CLI_WATERMARK, mask_sh),\
+	SF(MCIF_WB_WATERMARK, MCIF_WB_CLI_WATERMARK_MASK, mask_sh),\
+	SF(MCIF_WB_CLOCK_GATER_CONTROL, MCIF_WB_CLI_CLOCK_GATER_OVERRIDE, mask_sh),\
+	SF(MCIF_WB_SELF_REFRESH_CONTROL, DIS_REFRESH_UNDER_NBPREQ, mask_sh),\
+	SF(MCIF_WB_SELF_REFRESH_CONTROL, PERFRAME_SELF_REFRESH, mask_sh),\
+	SF(MULTI_LEVEL_QOS_CTRL, MAX_SCALED_TIME_TO_URGENT, mask_sh),\
+	SF(MCIF_WB_BUF_LUMA_SIZE, MCIF_WB_BUF_LUMA_SIZE, mask_sh),\
+	SF(MCIF_WB_BUF_CHROMA_SIZE, MCIF_WB_BUF_CHROMA_SIZE, mask_sh),\
+	SF(MCIF_WB_BUF_1_ADDR_Y_HIGH, MCIF_WB_BUF_1_ADDR_Y_HIGH, mask_sh),\
+	SF(MCIF_WB_BUF_1_ADDR_C_HIGH, MCIF_WB_BUF_1_ADDR_C_HIGH, mask_sh),\
+	SF(MCIF_WB_BUF_2_ADDR_Y_HIGH, MCIF_WB_BUF_2_ADDR_Y_HIGH, mask_sh),\
+	SF(MCIF_WB_BUF_2_ADDR_C_HIGH, MCIF_WB_BUF_2_ADDR_C_HIGH, mask_sh),\
+	SF(MCIF_WB_BUF_3_ADDR_Y_HIGH, MCIF_WB_BUF_3_ADDR_Y_HIGH, mask_sh),\
+	SF(MCIF_WB_BUF_3_ADDR_C_HIGH, MCIF_WB_BUF_3_ADDR_C_HIGH, mask_sh),\
+	SF(MCIF_WB_BUF_4_ADDR_Y_HIGH, MCIF_WB_BUF_4_ADDR_Y_HIGH, mask_sh),\
+	SF(MCIF_WB_BUF_4_ADDR_C_HIGH, MCIF_WB_BUF_4_ADDR_C_HIGH, mask_sh),\
+	SF(MCIF_WB_BUF_1_RESOLUTION, MCIF_WB_BUF_1_RESOLUTION_WIDTH, mask_sh),\
+	SF(MCIF_WB_BUF_1_RESOLUTION, MCIF_WB_BUF_1_RESOLUTION_HEIGHT, mask_sh),\
+	SF(MCIF_WB_BUF_2_RESOLUTION, MCIF_WB_BUF_2_RESOLUTION_WIDTH, mask_sh),\
+	SF(MCIF_WB_BUF_2_RESOLUTION, MCIF_WB_BUF_2_RESOLUTION_HEIGHT, mask_sh),\
+	SF(MCIF_WB_BUF_3_RESOLUTION, MCIF_WB_BUF_3_RESOLUTION_WIDTH, mask_sh),\
+	SF(MCIF_WB_BUF_3_RESOLUTION, MCIF_WB_BUF_3_RESOLUTION_HEIGHT, mask_sh),\
+	SF(MCIF_WB_BUF_4_RESOLUTION, MCIF_WB_BUF_4_RESOLUTION_WIDTH, mask_sh),\
+	SF(MCIF_WB_BUF_4_RESOLUTION, MCIF_WB_BUF_4_RESOLUTION_HEIGHT, mask_sh),\
+	SF(MMHUBBUB_WARMUP_ADDR_REGION, MMHUBBUB_WARMUP_ADDR_REGION, mask_sh),\
+	SF(MMHUBBUB_WARMUP_BASE_ADDR_HIGH, MMHUBBUB_WARMUP_BASE_ADDR_HIGH, mask_sh),\
+	SF(MMHUBBUB_WARMUP_BASE_ADDR_LOW, MMHUBBUB_WARMUP_BASE_ADDR_LOW, mask_sh),\
+	SF(MMHUBBUB_WARMUP_CONTROL_STATUS, MMHUBBUB_WARMUP_EN, mask_sh),\
+	SF(MMHUBBUB_WARMUP_CONTROL_STATUS, MMHUBBUB_WARMUP_SW_INT_EN, mask_sh),\
+	SF(MMHUBBUB_WARMUP_CONTROL_STATUS, MMHUBBUB_WARMUP_SW_INT_STATUS, mask_sh),\
+	SF(MMHUBBUB_WARMUP_CONTROL_STATUS, MMHUBBUB_WARMUP_SW_INT_ACK, mask_sh),\
+	SF(MMHUBBUB_WARMUP_CONTROL_STATUS, MMHUBBUB_WARMUP_INC_ADDR, mask_sh),\
+	SF(MCIF_WB_DRAM_SPEED_CHANGE_DURATION_VBI, MCIF_WB_DRAM_SPEED_CHANGE_DURATION_VBI, mask_sh)
+
+
+#define MCIF_WB_REG_FIELD_LIST_DCN3_0(type) \
+	MCIF_WB_REG_FIELD_LIST_DCN2_0(type);\
+	type WBIF_WHOLE_BUF_MODE;\
+	type MMHUBBUB_WARMUP_ADDR_REGION;\
+	type MMHUBBUB_WARMUP_BASE_ADDR_HIGH;\
+	type MMHUBBUB_WARMUP_BASE_ADDR_LOW;\
+	type MMHUBBUB_WARMUP_EN;\
+	type MMHUBBUB_WARMUP_SW_INT_EN;\
+	type MMHUBBUB_WARMUP_SW_INT_STATUS;\
+	type MMHUBBUB_WARMUP_SW_INT_ACK;\
+	type MMHUBBUB_WARMUP_INC_ADDR;\
+	type MMHUBBUB_WARMUP_P_VMID;\
+	type MCIF_WB_DRAM_SPEED_CHANGE_DURATION_VBI
+
+#define MCIF_WB_REG_VARIABLE_LIST_DCN3_0 \
+	MCIF_WB_REG_VARIABLE_LIST_DCN2_0; \
+	uint32_t MMHUBBUB_MEM_PWR_CNTL;\
+	uint32_t MMHUBBUB_WARMUP_ADDR_REGION;\
+	uint32_t MMHUBBUB_WARMUP_BASE_ADDR_HIGH;\
+	uint32_t MMHUBBUB_WARMUP_BASE_ADDR_LOW;\
+	uint32_t MMHUBBUB_WARMUP_CONTROL_STATUS;\
+	uint32_t MMHUBBUB_WARMUP_P_VMID;\
+	uint32_t MCIF_WB_DRAM_SPEED_CHANGE_DURATION_VBI
+
+struct dcn30_mmhubbub_registers {
+	MCIF_WB_REG_VARIABLE_LIST_DCN3_0;
+};
+
+
+struct dcn30_mmhubbub_mask {
+	MCIF_WB_REG_FIELD_LIST_DCN3_0(uint32_t);
+};
+
+struct dcn30_mmhubbub_shift {
+	MCIF_WB_REG_FIELD_LIST_DCN3_0(uint8_t);
+};
+
+struct dcn30_mmhubbub {
+	struct mcif_wb base;
+	const struct dcn30_mmhubbub_registers *mcif_wb_regs;
+	const struct dcn30_mmhubbub_shift *mcif_wb_shift;
+	const struct dcn30_mmhubbub_mask *mcif_wb_mask;
+};
+
+void dcn30_mmhubbub_construct(struct dcn30_mmhubbub *mcif_wb30,
+	struct dc_context *ctx,
+	const struct dcn30_mmhubbub_registers *mcif_wb_regs,
+	const struct dcn30_mmhubbub_shift *mcif_wb_shift,
+	const struct dcn30_mmhubbub_mask *mcif_wb_mask,
+	int inst);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_opp.h b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_opp.h
new file mode 100644
index 000000000000..78f7cf772bbc
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_opp.h
@@ -0,0 +1,36 @@
+/* Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_OPP_DCN30_H__
+#define __DC_OPP_DCN30_H__
+
+#include "dcn20/dcn20_opp.h"
+
+
+#define OPP_REG_LIST_DCN30(id) \
+	OPP_REG_LIST_DCN10(id), \
+	OPP_DPG_REG_LIST(id), \
+	SRI(FMT_422_CONTROL, FMT, id)
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_vpg.c b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_vpg.c
new file mode 100644
index 000000000000..14bc44b1f886
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_vpg.c
@@ -0,0 +1,264 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+#include "dc_bios_types.h"
+#include "dcn30_vpg.h"
+#include "reg_helper.h"
+
+#define DC_LOGGER \
+		vpg3->base.ctx->logger
+
+#define REG(reg)\
+	(vpg3->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	vpg3->vpg_shift->field_name, vpg3->vpg_mask->field_name
+
+
+#define CTX \
+	vpg3->base.ctx
+
+
+void vpg3_update_generic_info_packet(
+	struct vpg *vpg,
+	uint32_t packet_index,
+	const struct dc_info_packet *info_packet,
+	bool immediate_update)
+{
+	struct dcn30_vpg *vpg3 = DCN30_VPG_FROM_VPG(vpg);
+	uint32_t i;
+
+	/* TODOFPGA Figure out a proper number for max_retries polling for lock
+	 * use 50 for now.
+	 */
+	uint32_t max_retries = 50;
+
+	if (packet_index > 14)
+		ASSERT(0);
+
+	/* poll dig_update_lock is not locked -> asic internal signal
+	 * assume otg master lock will unlock it
+	 */
+	/* REG_WAIT(AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_LOCK_STATUS,
+	 * 		0, 10, max_retries);
+	 */
+
+	/* TODO: Check if this is required */
+	/* check if HW reading GSP memory */
+	REG_WAIT(VPG_GENERIC_STATUS, VPG_GENERIC_CONFLICT_OCCURED,
+			0, 10, max_retries);
+
+	/* HW does is not reading GSP memory not reading too long ->
+	 * something wrong. clear GPS memory access and notify?
+	 * hw SW is writing to GSP memory
+	 */
+	REG_UPDATE(VPG_GENERIC_STATUS, VPG_GENERIC_CONFLICT_CLR, 1);
+
+	/* choose which generic packet to use */
+	REG_UPDATE(VPG_GENERIC_PACKET_ACCESS_CTRL,
+			VPG_GENERIC_DATA_INDEX, packet_index*9);
+
+	/* write generic packet header
+	 * (4th byte is for GENERIC0 only)
+	 */
+	REG_SET_4(VPG_GENERIC_PACKET_DATA, 0,
+			VPG_GENERIC_DATA_BYTE0, info_packet->hb0,
+			VPG_GENERIC_DATA_BYTE1, info_packet->hb1,
+			VPG_GENERIC_DATA_BYTE2, info_packet->hb2,
+			VPG_GENERIC_DATA_BYTE3, info_packet->hb3);
+
+	/* write generic packet contents
+	 * (we never use last 4 bytes)
+	 * there are 8 (0-7) mmDIG0_AFMT_GENERIC0_x registers
+	 */
+	{
+		const uint32_t *content =
+			(const uint32_t *) &info_packet->sb[0];
+
+		for (i = 0; i < 8; i++) {
+			REG_WRITE(VPG_GENERIC_PACKET_DATA, *content++);
+		}
+	}
+
+	/* atomically update double-buffered GENERIC0 registers in immediate mode
+	 * (update at next block_update when block_update_lock == 0).
+	 */
+	if (immediate_update) {
+		switch (packet_index) {
+		case 0:
+			REG_UPDATE(VPG_GSP_IMMEDIATE_UPDATE_CTRL,
+					VPG_GENERIC0_IMMEDIATE_UPDATE, 1);
+			break;
+		case 1:
+			REG_UPDATE(VPG_GSP_IMMEDIATE_UPDATE_CTRL,
+					VPG_GENERIC1_IMMEDIATE_UPDATE, 1);
+			break;
+		case 2:
+			REG_UPDATE(VPG_GSP_IMMEDIATE_UPDATE_CTRL,
+					VPG_GENERIC2_IMMEDIATE_UPDATE, 1);
+			break;
+		case 3:
+			REG_UPDATE(VPG_GSP_IMMEDIATE_UPDATE_CTRL,
+					VPG_GENERIC3_IMMEDIATE_UPDATE, 1);
+			break;
+		case 4:
+			REG_UPDATE(VPG_GSP_IMMEDIATE_UPDATE_CTRL,
+					VPG_GENERIC4_IMMEDIATE_UPDATE, 1);
+			break;
+		case 5:
+			REG_UPDATE(VPG_GSP_IMMEDIATE_UPDATE_CTRL,
+					VPG_GENERIC5_IMMEDIATE_UPDATE, 1);
+			break;
+		case 6:
+			REG_UPDATE(VPG_GSP_IMMEDIATE_UPDATE_CTRL,
+					VPG_GENERIC6_IMMEDIATE_UPDATE, 1);
+			break;
+		case 7:
+			REG_UPDATE(VPG_GSP_IMMEDIATE_UPDATE_CTRL,
+					VPG_GENERIC7_IMMEDIATE_UPDATE, 1);
+			break;
+		case 8:
+			REG_UPDATE(VPG_GSP_IMMEDIATE_UPDATE_CTRL,
+					VPG_GENERIC8_IMMEDIATE_UPDATE, 1);
+			break;
+		case 9:
+			REG_UPDATE(VPG_GSP_IMMEDIATE_UPDATE_CTRL,
+					VPG_GENERIC9_IMMEDIATE_UPDATE, 1);
+			break;
+		case 10:
+			REG_UPDATE(VPG_GSP_IMMEDIATE_UPDATE_CTRL,
+					VPG_GENERIC10_IMMEDIATE_UPDATE, 1);
+			break;
+		case 11:
+			REG_UPDATE(VPG_GSP_IMMEDIATE_UPDATE_CTRL,
+					VPG_GENERIC11_IMMEDIATE_UPDATE, 1);
+			break;
+		case 12:
+			REG_UPDATE(VPG_GSP_IMMEDIATE_UPDATE_CTRL,
+					VPG_GENERIC12_IMMEDIATE_UPDATE, 1);
+			break;
+		case 13:
+			REG_UPDATE(VPG_GSP_IMMEDIATE_UPDATE_CTRL,
+					VPG_GENERIC13_IMMEDIATE_UPDATE, 1);
+			break;
+		case 14:
+			REG_UPDATE(VPG_GSP_IMMEDIATE_UPDATE_CTRL,
+					VPG_GENERIC14_IMMEDIATE_UPDATE, 1);
+			break;
+		default:
+			break;
+		}
+	} else {
+		switch (packet_index) {
+		case 0:
+			REG_UPDATE(VPG_GSP_FRAME_UPDATE_CTRL,
+					VPG_GENERIC0_FRAME_UPDATE, 1);
+			break;
+		case 1:
+			REG_UPDATE(VPG_GSP_FRAME_UPDATE_CTRL,
+					VPG_GENERIC1_FRAME_UPDATE, 1);
+			break;
+		case 2:
+			REG_UPDATE(VPG_GSP_FRAME_UPDATE_CTRL,
+					VPG_GENERIC2_FRAME_UPDATE, 1);
+			break;
+		case 3:
+			REG_UPDATE(VPG_GSP_FRAME_UPDATE_CTRL,
+					VPG_GENERIC3_FRAME_UPDATE, 1);
+			break;
+		case 4:
+			REG_UPDATE(VPG_GSP_FRAME_UPDATE_CTRL,
+					VPG_GENERIC4_FRAME_UPDATE, 1);
+			break;
+		case 5:
+			REG_UPDATE(VPG_GSP_FRAME_UPDATE_CTRL,
+					VPG_GENERIC5_FRAME_UPDATE, 1);
+			break;
+		case 6:
+			REG_UPDATE(VPG_GSP_FRAME_UPDATE_CTRL,
+					VPG_GENERIC6_FRAME_UPDATE, 1);
+			break;
+		case 7:
+			REG_UPDATE(VPG_GSP_FRAME_UPDATE_CTRL,
+					VPG_GENERIC7_FRAME_UPDATE, 1);
+			break;
+		case 8:
+			REG_UPDATE(VPG_GSP_FRAME_UPDATE_CTRL,
+					VPG_GENERIC8_FRAME_UPDATE, 1);
+			break;
+		case 9:
+			REG_UPDATE(VPG_GSP_FRAME_UPDATE_CTRL,
+					VPG_GENERIC9_FRAME_UPDATE, 1);
+			break;
+		case 10:
+			REG_UPDATE(VPG_GSP_FRAME_UPDATE_CTRL,
+					VPG_GENERIC10_FRAME_UPDATE, 1);
+			break;
+		case 11:
+			REG_UPDATE(VPG_GSP_FRAME_UPDATE_CTRL,
+					VPG_GENERIC11_FRAME_UPDATE, 1);
+			break;
+		case 12:
+			REG_UPDATE(VPG_GSP_FRAME_UPDATE_CTRL,
+					VPG_GENERIC12_FRAME_UPDATE, 1);
+			break;
+		case 13:
+			REG_UPDATE(VPG_GSP_FRAME_UPDATE_CTRL,
+					VPG_GENERIC13_FRAME_UPDATE, 1);
+			break;
+		case 14:
+			REG_UPDATE(VPG_GSP_FRAME_UPDATE_CTRL,
+					VPG_GENERIC14_FRAME_UPDATE, 1);
+			break;
+
+		default:
+			break;
+		}
+
+	}
+}
+
+static struct vpg_funcs dcn30_vpg_funcs = {
+	.update_generic_info_packet	= vpg3_update_generic_info_packet,
+};
+
+void vpg3_construct(struct dcn30_vpg *vpg3,
+	struct dc_context *ctx,
+	uint32_t inst,
+	const struct dcn30_vpg_registers *vpg_regs,
+	const struct dcn30_vpg_shift *vpg_shift,
+	const struct dcn30_vpg_mask *vpg_mask)
+{
+	vpg3->base.ctx = ctx;
+
+	vpg3->base.inst = inst;
+	vpg3->base.funcs = &dcn30_vpg_funcs;
+
+	vpg3->regs = vpg_regs;
+	vpg3->vpg_shift = vpg_shift;
+	vpg3->vpg_mask = vpg_mask;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_vpg.h b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_vpg.h
new file mode 100644
index 000000000000..466ba20b9c61
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_vpg.h
@@ -0,0 +1,157 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_DCN30_VPG_H__
+#define __DAL_DCN30_VPG_H__
+
+#include "vpg.h"
+
+#define DCN30_VPG_FROM_VPG(vpg)\
+	container_of(vpg, struct dcn30_vpg, base)
+
+#define VPG_DCN3_REG_LIST(id) \
+	SRI(VPG_GENERIC_STATUS, VPG, id), \
+	SRI(VPG_GENERIC_PACKET_ACCESS_CTRL, VPG, id), \
+	SRI(VPG_GENERIC_PACKET_DATA, VPG, id), \
+	SRI(VPG_GSP_FRAME_UPDATE_CTRL, VPG, id), \
+	SRI(VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG, id)
+
+struct dcn30_vpg_registers {
+	uint32_t VPG_GENERIC_STATUS;
+	uint32_t VPG_GENERIC_PACKET_ACCESS_CTRL;
+	uint32_t VPG_GENERIC_PACKET_DATA;
+	uint32_t VPG_GSP_FRAME_UPDATE_CTRL;
+	uint32_t VPG_GSP_IMMEDIATE_UPDATE_CTRL;
+};
+
+#define DCN3_VPG_MASK_SH_LIST(mask_sh)\
+	SE_SF(VPG0_VPG_GENERIC_STATUS, VPG_GENERIC_CONFLICT_OCCURED, mask_sh),\
+	SE_SF(VPG0_VPG_GENERIC_STATUS, VPG_GENERIC_CONFLICT_CLR, mask_sh),\
+	SE_SF(VPG0_VPG_GENERIC_PACKET_ACCESS_CTRL, VPG_GENERIC_DATA_INDEX, mask_sh),\
+	SE_SF(VPG0_VPG_GENERIC_PACKET_DATA, VPG_GENERIC_DATA_BYTE0, mask_sh),\
+	SE_SF(VPG0_VPG_GENERIC_PACKET_DATA, VPG_GENERIC_DATA_BYTE1, mask_sh),\
+	SE_SF(VPG0_VPG_GENERIC_PACKET_DATA, VPG_GENERIC_DATA_BYTE2, mask_sh),\
+	SE_SF(VPG0_VPG_GENERIC_PACKET_DATA, VPG_GENERIC_DATA_BYTE3, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC0_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC1_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC2_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC3_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC4_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC5_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC6_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC7_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC8_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC9_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC10_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC11_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC12_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC13_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC14_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC0_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC1_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC2_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC3_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC4_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC5_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC6_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC7_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC8_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC9_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC10_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC11_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC12_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC13_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC14_IMMEDIATE_UPDATE, mask_sh)
+
+#define VPG_DCN3_REG_FIELD_LIST(type) \
+	type VPG_GENERIC_CONFLICT_OCCURED;\
+	type VPG_GENERIC_CONFLICT_CLR;\
+	type VPG_GENERIC_DATA_INDEX;\
+	type VPG_GENERIC_DATA_BYTE0;\
+	type VPG_GENERIC_DATA_BYTE1;\
+	type VPG_GENERIC_DATA_BYTE2;\
+	type VPG_GENERIC_DATA_BYTE3;\
+	type VPG_GENERIC0_FRAME_UPDATE;\
+	type VPG_GENERIC1_FRAME_UPDATE;\
+	type VPG_GENERIC2_FRAME_UPDATE;\
+	type VPG_GENERIC3_FRAME_UPDATE;\
+	type VPG_GENERIC4_FRAME_UPDATE;\
+	type VPG_GENERIC5_FRAME_UPDATE;\
+	type VPG_GENERIC6_FRAME_UPDATE;\
+	type VPG_GENERIC7_FRAME_UPDATE;\
+	type VPG_GENERIC8_FRAME_UPDATE;\
+	type VPG_GENERIC9_FRAME_UPDATE;\
+	type VPG_GENERIC10_FRAME_UPDATE;\
+	type VPG_GENERIC11_FRAME_UPDATE;\
+	type VPG_GENERIC12_FRAME_UPDATE;\
+	type VPG_GENERIC13_FRAME_UPDATE;\
+	type VPG_GENERIC14_FRAME_UPDATE;\
+	type VPG_GENERIC0_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC1_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC2_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC3_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC4_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC5_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC6_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC7_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC8_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC9_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC10_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC11_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC12_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC13_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC14_IMMEDIATE_UPDATE
+
+
+struct dcn30_vpg_shift {
+	VPG_DCN3_REG_FIELD_LIST(uint8_t);
+};
+
+struct dcn30_vpg_mask {
+	VPG_DCN3_REG_FIELD_LIST(uint32_t);
+};
+
+struct dcn30_vpg {
+	struct vpg base;
+	const struct dcn30_vpg_registers *regs;
+	const struct dcn30_vpg_shift *vpg_shift;
+	const struct dcn30_vpg_mask *vpg_mask;
+};
+
+void vpg3_update_generic_info_packet(
+	struct vpg *vpg,
+	uint32_t packet_index,
+	const struct dc_info_packet *info_packet,
+	bool immediate_update);
+
+void vpg3_construct(struct dcn30_vpg *vpg3,
+	struct dc_context *ctx,
+	uint32_t inst,
+	const struct dcn30_vpg_registers *vpg_regs,
+	const struct dcn30_vpg_shift *vpg_shift,
+	const struct dcn30_vpg_mask *vpg_mask);
+
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dcn301/Makefile b/drivers/gpu/drm/amd/display/dc/dcn301/Makefile
new file mode 100644
index 000000000000..fb4814ab3f05
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn301/Makefile
@@ -0,0 +1,10 @@
+# SPDX-License-Identifier: MIT
+# Copyright © 2024 Advanced Micro Devices, Inc. All rights reserved.
+#
+# Makefile for dcn30.
+
+DCN301 = dcn301_panel_cntl.o
+
+AMD_DAL_DCN301 = $(addprefix $(AMDDALPATH)/dc/dcn301/,$(DCN301))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCN301)
diff --git a/drivers/gpu/drm/amd/display/dc/dcn301/dcn301_panel_cntl.c b/drivers/gpu/drm/amd/display/dc/dcn301/dcn301_panel_cntl.c
new file mode 100644
index 000000000000..9e96a3ace207
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn301/dcn301_panel_cntl.c
@@ -0,0 +1,219 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+#include "core_types.h"
+#include "dc_dmub_srv.h"
+#include "dcn301_panel_cntl.h"
+#include "atom.h"
+
+#define TO_DCN301_PANEL_CNTL(panel_cntl)\
+	container_of(panel_cntl, struct dcn301_panel_cntl, base)
+
+#define CTX \
+	dcn301_panel_cntl->base.ctx
+
+#define DC_LOGGER \
+	dcn301_panel_cntl->base.ctx->logger
+
+#define REG(reg)\
+	dcn301_panel_cntl->regs->reg
+
+#undef FN
+#define FN(reg_name, field_name) \
+	dcn301_panel_cntl->shift->field_name, dcn301_panel_cntl->mask->field_name
+
+static unsigned int dcn301_get_16_bit_backlight_from_pwm(struct panel_cntl *panel_cntl)
+{
+	uint64_t current_backlight;
+	uint32_t round_result;
+	uint32_t bl_period, bl_int_count;
+	uint32_t bl_pwm, fractional_duty_cycle_en;
+	uint32_t bl_period_mask, bl_pwm_mask;
+	struct dcn301_panel_cntl *dcn301_panel_cntl = TO_DCN301_PANEL_CNTL(panel_cntl);
+
+	REG_GET(BL_PWM_PERIOD_CNTL, BL_PWM_PERIOD, &bl_period);
+	REG_GET(BL_PWM_PERIOD_CNTL, BL_PWM_PERIOD_BITCNT, &bl_int_count);
+
+	REG_GET(BL_PWM_CNTL, BL_ACTIVE_INT_FRAC_CNT, &bl_pwm);
+	REG_GET(BL_PWM_CNTL, BL_PWM_FRACTIONAL_EN, &fractional_duty_cycle_en);
+
+	if (bl_int_count == 0)
+		bl_int_count = 16;
+
+	bl_period_mask = (1 << bl_int_count) - 1;
+	bl_period &= bl_period_mask;
+
+	bl_pwm_mask = bl_period_mask << (16 - bl_int_count);
+
+	if (fractional_duty_cycle_en == 0)
+		bl_pwm &= bl_pwm_mask;
+	else
+		bl_pwm &= 0xFFFF;
+
+	current_backlight = (uint64_t)bl_pwm << (1 + bl_int_count);
+
+	if (bl_period == 0)
+		bl_period = 0xFFFF;
+
+	current_backlight = div_u64(current_backlight, bl_period);
+	current_backlight = (current_backlight + 1) >> 1;
+
+	current_backlight = (uint64_t)(current_backlight) * bl_period;
+
+	round_result = (uint32_t)(current_backlight & 0xFFFFFFFF);
+
+	round_result = (round_result >> (bl_int_count-1)) & 1;
+
+	current_backlight >>= bl_int_count;
+	current_backlight += round_result;
+
+	return (uint32_t)(current_backlight);
+}
+
+static uint32_t dcn301_panel_cntl_hw_init(struct panel_cntl *panel_cntl)
+{
+	struct dcn301_panel_cntl *dcn301_panel_cntl = TO_DCN301_PANEL_CNTL(panel_cntl);
+	uint32_t value;
+	uint32_t current_backlight;
+
+	/* It must not be 0, so we have to restore them
+	 * Bios bug w/a - period resets to zero,
+	 * restoring to cache values which is always correct
+	 */
+	REG_GET(BL_PWM_CNTL, BL_ACTIVE_INT_FRAC_CNT, &value);
+
+	if (value == 0 || value == 1) {
+		if (panel_cntl->stored_backlight_registers.BL_PWM_CNTL != 0) {
+			REG_WRITE(BL_PWM_CNTL,
+					panel_cntl->stored_backlight_registers.BL_PWM_CNTL);
+			REG_WRITE(BL_PWM_CNTL2,
+					panel_cntl->stored_backlight_registers.BL_PWM_CNTL2);
+			REG_WRITE(BL_PWM_PERIOD_CNTL,
+					panel_cntl->stored_backlight_registers.BL_PWM_PERIOD_CNTL);
+			REG_UPDATE(PWRSEQ_REF_DIV,
+				BL_PWM_REF_DIV,
+				panel_cntl->stored_backlight_registers.LVTMA_PWRSEQ_REF_DIV_BL_PWM_REF_DIV);
+		} else {
+			/* TODO: Note: This should not really happen since VBIOS
+			 * should have initialized PWM registers on boot.
+			 */
+			REG_WRITE(BL_PWM_CNTL, 0xC000FA00);
+			REG_WRITE(BL_PWM_PERIOD_CNTL, 0x000C0FA0);
+		}
+	} else {
+		panel_cntl->stored_backlight_registers.BL_PWM_CNTL =
+				REG_READ(BL_PWM_CNTL);
+		panel_cntl->stored_backlight_registers.BL_PWM_CNTL2 =
+				REG_READ(BL_PWM_CNTL2);
+		panel_cntl->stored_backlight_registers.BL_PWM_PERIOD_CNTL =
+				REG_READ(BL_PWM_PERIOD_CNTL);
+
+		REG_GET(PWRSEQ_REF_DIV, BL_PWM_REF_DIV,
+				&panel_cntl->stored_backlight_registers.LVTMA_PWRSEQ_REF_DIV_BL_PWM_REF_DIV);
+	}
+
+	// Enable the backlight output
+	REG_UPDATE(BL_PWM_CNTL, BL_PWM_EN, 1);
+
+	// Unlock group 2 backlight registers
+	REG_UPDATE(BL_PWM_GRP1_REG_LOCK,
+			BL_PWM_GRP1_REG_LOCK, 0);
+
+	current_backlight = dcn301_get_16_bit_backlight_from_pwm(panel_cntl);
+
+	return current_backlight;
+}
+
+static void dcn301_panel_cntl_destroy(struct panel_cntl **panel_cntl)
+{
+	struct dcn301_panel_cntl *dcn301_panel_cntl = TO_DCN301_PANEL_CNTL(*panel_cntl);
+
+	kfree(dcn301_panel_cntl);
+	*panel_cntl = NULL;
+}
+
+static bool dcn301_is_panel_backlight_on(struct panel_cntl *panel_cntl)
+{
+	struct dcn301_panel_cntl *dcn301_panel_cntl = TO_DCN301_PANEL_CNTL(panel_cntl);
+	uint32_t value;
+
+	REG_GET(PWRSEQ_CNTL, PANEL_BLON, &value);
+
+	return value;
+}
+
+static bool dcn301_is_panel_powered_on(struct panel_cntl *panel_cntl)
+{
+	struct dcn301_panel_cntl *dcn301_panel_cntl = TO_DCN301_PANEL_CNTL(panel_cntl);
+	uint32_t pwr_seq_state, dig_on, dig_on_ovrd;
+
+	REG_GET(PWRSEQ_STATE, PANEL_PWRSEQ_TARGET_STATE_R, &pwr_seq_state);
+
+	REG_GET_2(PWRSEQ_CNTL, PANEL_DIGON, &dig_on, PANEL_DIGON_OVRD, &dig_on_ovrd);
+
+	return (pwr_seq_state == 1) || (dig_on == 1 && dig_on_ovrd == 1);
+}
+
+static void dcn301_store_backlight_level(struct panel_cntl *panel_cntl)
+{
+	struct dcn301_panel_cntl *dcn301_panel_cntl = TO_DCN301_PANEL_CNTL(panel_cntl);
+
+	panel_cntl->stored_backlight_registers.BL_PWM_CNTL =
+		REG_READ(BL_PWM_CNTL);
+	panel_cntl->stored_backlight_registers.BL_PWM_CNTL2 =
+		REG_READ(BL_PWM_CNTL2);
+	panel_cntl->stored_backlight_registers.BL_PWM_PERIOD_CNTL =
+		REG_READ(BL_PWM_PERIOD_CNTL);
+
+	REG_GET(PWRSEQ_REF_DIV, BL_PWM_REF_DIV,
+		&panel_cntl->stored_backlight_registers.LVTMA_PWRSEQ_REF_DIV_BL_PWM_REF_DIV);
+}
+
+static const struct panel_cntl_funcs dcn301_link_panel_cntl_funcs = {
+	.destroy = dcn301_panel_cntl_destroy,
+	.hw_init = dcn301_panel_cntl_hw_init,
+	.is_panel_backlight_on = dcn301_is_panel_backlight_on,
+	.is_panel_powered_on = dcn301_is_panel_powered_on,
+	.store_backlight_level = dcn301_store_backlight_level,
+	.get_current_backlight = dcn301_get_16_bit_backlight_from_pwm,
+};
+
+void dcn301_panel_cntl_construct(
+	struct dcn301_panel_cntl *dcn301_panel_cntl,
+	const struct panel_cntl_init_data *init_data,
+	const struct dce_panel_cntl_registers *regs,
+	const struct dcn301_panel_cntl_shift *shift,
+	const struct dcn301_panel_cntl_mask *mask)
+{
+	dcn301_panel_cntl->regs = regs;
+	dcn301_panel_cntl->shift = shift;
+	dcn301_panel_cntl->mask = mask;
+
+	dcn301_panel_cntl->base.funcs = &dcn301_link_panel_cntl_funcs;
+	dcn301_panel_cntl->base.ctx = init_data->ctx;
+	dcn301_panel_cntl->base.inst = init_data->inst;
+	dcn301_panel_cntl->base.pwrseq_inst = 0;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dcn301/dcn301_panel_cntl.h b/drivers/gpu/drm/amd/display/dc/dcn301/dcn301_panel_cntl.h
new file mode 100644
index 000000000000..ffc875da7467
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn301/dcn301_panel_cntl.h
@@ -0,0 +1,97 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_PANEL_CNTL__DCN301_H__
+#define __DC_PANEL_CNTL__DCN301_H__
+
+#include "panel_cntl.h"
+#include "dce/dce_panel_cntl.h"
+
+
+#define DCN301_PANEL_CNTL_REG_LIST(id)\
+	SRIR(PWRSEQ_CNTL, CNTL, PANEL_PWRSEQ, id), \
+	SRIR(PWRSEQ_STATE, STATE, PANEL_PWRSEQ, id), \
+	SRIR(PWRSEQ_REF_DIV, REF_DIV, PANEL_PWRSEQ, id), \
+	SRIR(BL_PWM_CNTL, CNTL, BL_PWM, id), \
+	SRIR(BL_PWM_CNTL2, CNTL2, BL_PWM, id), \
+	SRIR(BL_PWM_PERIOD_CNTL, PERIOD_CNTL, BL_PWM, id), \
+	SRIR(BL_PWM_GRP1_REG_LOCK, GRP1_REG_LOCK, BL_PWM, id)
+
+#define DCN301_PANEL_CNTL_SF(reg_name, field_name, post_fix)\
+	.field_name = reg_name ## __ ## field_name ## post_fix
+
+#define DCN301_PANEL_CNTL_MASK_SH_LIST(mask_sh) \
+	DCN301_PANEL_CNTL_SF(PANEL_PWRSEQ0_CNTL, PANEL_BLON, mask_sh),\
+	DCN301_PANEL_CNTL_SF(PANEL_PWRSEQ0_CNTL, PANEL_DIGON, mask_sh),\
+	DCN301_PANEL_CNTL_SF(PANEL_PWRSEQ0_CNTL, PANEL_DIGON_OVRD, mask_sh),\
+	DCN301_PANEL_CNTL_SF(PANEL_PWRSEQ0_STATE, PANEL_PWRSEQ_TARGET_STATE_R, mask_sh), \
+	DCN301_PANEL_CNTL_SF(PANEL_PWRSEQ0_REF_DIV, BL_PWM_REF_DIV, mask_sh), \
+	DCN301_PANEL_CNTL_SF(BL_PWM0_PERIOD_CNTL, BL_PWM_PERIOD, mask_sh), \
+	DCN301_PANEL_CNTL_SF(BL_PWM0_PERIOD_CNTL, BL_PWM_PERIOD_BITCNT, mask_sh), \
+	DCN301_PANEL_CNTL_SF(BL_PWM0_CNTL, BL_ACTIVE_INT_FRAC_CNT, mask_sh), \
+	DCN301_PANEL_CNTL_SF(BL_PWM0_CNTL, BL_PWM_FRACTIONAL_EN, mask_sh), \
+	DCN301_PANEL_CNTL_SF(BL_PWM0_CNTL, BL_PWM_EN, mask_sh), \
+	DCN301_PANEL_CNTL_SF(BL_PWM0_GRP1_REG_LOCK, BL_PWM_GRP1_IGNORE_MASTER_LOCK_EN, mask_sh), \
+	DCN301_PANEL_CNTL_SF(BL_PWM0_GRP1_REG_LOCK, BL_PWM_GRP1_REG_LOCK, mask_sh), \
+	DCN301_PANEL_CNTL_SF(BL_PWM0_GRP1_REG_LOCK, BL_PWM_GRP1_REG_UPDATE_PENDING, mask_sh)
+
+#define DCN301_PANEL_CNTL_REG_FIELD_LIST(type) \
+	type PANEL_BLON;\
+	type PANEL_DIGON;\
+	type PANEL_DIGON_OVRD;\
+	type PANEL_PWRSEQ_TARGET_STATE_R; \
+	type BL_PWM_EN; \
+	type BL_ACTIVE_INT_FRAC_CNT; \
+	type BL_PWM_FRACTIONAL_EN; \
+	type BL_PWM_PERIOD; \
+	type BL_PWM_PERIOD_BITCNT; \
+	type BL_PWM_GRP1_IGNORE_MASTER_LOCK_EN; \
+	type BL_PWM_GRP1_REG_LOCK; \
+	type BL_PWM_GRP1_REG_UPDATE_PENDING; \
+	type BL_PWM_REF_DIV
+
+struct dcn301_panel_cntl_shift {
+	DCN301_PANEL_CNTL_REG_FIELD_LIST(uint8_t);
+};
+
+struct dcn301_panel_cntl_mask {
+	DCN301_PANEL_CNTL_REG_FIELD_LIST(uint32_t);
+};
+
+struct dcn301_panel_cntl {
+	struct panel_cntl base;
+	const struct dce_panel_cntl_registers *regs;
+	const struct dcn301_panel_cntl_shift *shift;
+	const struct dcn301_panel_cntl_mask *mask;
+};
+
+void dcn301_panel_cntl_construct(
+	struct dcn301_panel_cntl *panel_cntl,
+	const struct panel_cntl_init_data *init_data,
+	const struct dce_panel_cntl_registers *regs,
+	const struct dcn301_panel_cntl_shift *shift,
+	const struct dcn301_panel_cntl_mask *mask);
+
+#endif /* __DC_PANEL_CNTL__DCN301_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dcn31/Makefile b/drivers/gpu/drm/amd/display/dc/dcn31/Makefile
new file mode 100644
index 000000000000..d510e4652c18
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn31/Makefile
@@ -0,0 +1,13 @@
+# SPDX-License-Identifier: MIT
+# Copyright © 2024 Advanced Micro Devices, Inc. All rights reserved.
+#
+#
+# Makefile for dcn31.
+
+DCN31 = dcn31_panel_cntl.o \
+	dcn31_apg.o \
+	dcn31_afmt.o dcn31_vpg.o
+
+AMD_DAL_DCN31 = $(addprefix $(AMDDALPATH)/dc/dcn31/,$(DCN31))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DCN31)
diff --git a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_afmt.c b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_afmt.c
new file mode 100644
index 000000000000..d380a8ec2184
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_afmt.c
@@ -0,0 +1,92 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dc_bios_types.h"
+#include "hw_shared.h"
+#include "dcn30/dcn30_afmt.h"
+#include "dcn31_afmt.h"
+#include "reg_helper.h"
+#include "dc/dc.h"
+
+#define DC_LOGGER \
+		afmt31->base.ctx->logger
+
+#define REG(reg)\
+	(afmt31->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	afmt31->afmt_shift->field_name, afmt31->afmt_mask->field_name
+
+
+#define CTX \
+	afmt31->base.ctx
+
+static struct afmt_funcs dcn31_afmt_funcs = {
+	.setup_hdmi_audio		= afmt3_setup_hdmi_audio,
+	.se_audio_setup			= afmt3_se_audio_setup,
+	.audio_mute_control		= afmt3_audio_mute_control,
+	.audio_info_immediate_update	= afmt3_audio_info_immediate_update,
+	.setup_dp_audio			= afmt3_setup_dp_audio,
+	.afmt_powerdown			= afmt31_powerdown,
+	.afmt_poweron			= afmt31_poweron
+};
+
+void afmt31_powerdown(struct afmt *afmt)
+{
+	struct dcn31_afmt *afmt31 = DCN31_AFMT_FROM_AFMT(afmt);
+
+	if (afmt->ctx->dc->debug.enable_mem_low_power.bits.afmt == false)
+		return;
+
+	REG_UPDATE_2(AFMT_MEM_PWR, AFMT_MEM_PWR_DIS, 0, AFMT_MEM_PWR_FORCE, 1);
+}
+
+void afmt31_poweron(struct afmt *afmt)
+{
+	struct dcn31_afmt *afmt31 = DCN31_AFMT_FROM_AFMT(afmt);
+
+	if (afmt->ctx->dc->debug.enable_mem_low_power.bits.afmt == false)
+		return;
+
+	REG_UPDATE_2(AFMT_MEM_PWR, AFMT_MEM_PWR_DIS, 1, AFMT_MEM_PWR_FORCE, 0);
+}
+
+void afmt31_construct(struct dcn31_afmt *afmt31,
+	struct dc_context *ctx,
+	uint32_t inst,
+	const struct dcn31_afmt_registers *afmt_regs,
+	const struct dcn31_afmt_shift *afmt_shift,
+	const struct dcn31_afmt_mask *afmt_mask)
+{
+	afmt31->base.ctx = ctx;
+
+	afmt31->base.inst = inst;
+	afmt31->base.funcs = &dcn31_afmt_funcs;
+
+	afmt31->regs = afmt_regs;
+	afmt31->afmt_shift = afmt_shift;
+	afmt31->afmt_mask = afmt_mask;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_afmt.h b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_afmt.h
new file mode 100644
index 000000000000..802cb05b6ab9
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_afmt.h
@@ -0,0 +1,126 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_DCN31_AFMT_H__
+#define __DAL_DCN31_AFMT_H__
+
+
+#define DCN31_AFMT_FROM_AFMT(afmt)\
+	container_of(afmt, struct dcn31_afmt, base)
+
+#define AFMT_DCN31_REG_LIST(id) \
+	SRI(AFMT_INFOFRAME_CONTROL0, AFMT, id), \
+	SRI(AFMT_VBI_PACKET_CONTROL, AFMT, id), \
+	SRI(AFMT_AUDIO_PACKET_CONTROL, AFMT, id), \
+	SRI(AFMT_AUDIO_PACKET_CONTROL2, AFMT, id), \
+	SRI(AFMT_AUDIO_SRC_CONTROL, AFMT, id), \
+	SRI(AFMT_60958_0, AFMT, id), \
+	SRI(AFMT_60958_1, AFMT, id), \
+	SRI(AFMT_60958_2, AFMT, id), \
+	SRI(AFMT_MEM_PWR, AFMT, id)
+
+struct dcn31_afmt_registers {
+	uint32_t AFMT_INFOFRAME_CONTROL0;
+	uint32_t AFMT_VBI_PACKET_CONTROL;
+	uint32_t AFMT_AUDIO_PACKET_CONTROL;
+	uint32_t AFMT_AUDIO_PACKET_CONTROL2;
+	uint32_t AFMT_AUDIO_SRC_CONTROL;
+	uint32_t AFMT_60958_0;
+	uint32_t AFMT_60958_1;
+	uint32_t AFMT_60958_2;
+	uint32_t AFMT_MEM_PWR;
+};
+
+#define DCN31_AFMT_MASK_SH_LIST(mask_sh)\
+	SE_SF(AFMT0_AFMT_INFOFRAME_CONTROL0, AFMT_AUDIO_INFO_UPDATE, mask_sh),\
+	SE_SF(AFMT0_AFMT_AUDIO_SRC_CONTROL, AFMT_AUDIO_SRC_SELECT, mask_sh),\
+	SE_SF(AFMT0_AFMT_AUDIO_PACKET_CONTROL2, AFMT_AUDIO_CHANNEL_ENABLE, mask_sh),\
+	SE_SF(AFMT0_AFMT_AUDIO_PACKET_CONTROL, AFMT_60958_CS_UPDATE, mask_sh),\
+	SE_SF(AFMT0_AFMT_AUDIO_PACKET_CONTROL2, AFMT_AUDIO_LAYOUT_OVRD, mask_sh),\
+	SE_SF(AFMT0_AFMT_AUDIO_PACKET_CONTROL2, AFMT_60958_OSF_OVRD, mask_sh),\
+	SE_SF(AFMT0_AFMT_60958_0, AFMT_60958_CS_CHANNEL_NUMBER_L, mask_sh),\
+	SE_SF(AFMT0_AFMT_60958_0, AFMT_60958_CS_CLOCK_ACCURACY, mask_sh),\
+	SE_SF(AFMT0_AFMT_60958_1, AFMT_60958_CS_CHANNEL_NUMBER_R, mask_sh),\
+	SE_SF(AFMT0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_2, mask_sh),\
+	SE_SF(AFMT0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_3, mask_sh),\
+	SE_SF(AFMT0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_4, mask_sh),\
+	SE_SF(AFMT0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_5, mask_sh),\
+	SE_SF(AFMT0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_6, mask_sh),\
+	SE_SF(AFMT0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_7, mask_sh),\
+	SE_SF(AFMT0_AFMT_AUDIO_PACKET_CONTROL, AFMT_AUDIO_SAMPLE_SEND, mask_sh),\
+	SE_SF(AFMT0_AFMT_MEM_PWR, AFMT_MEM_PWR_FORCE, mask_sh),\
+	SE_SF(AFMT0_AFMT_MEM_PWR, AFMT_MEM_PWR_DIS, mask_sh),\
+	SE_SF(AFMT0_AFMT_MEM_PWR, AFMT_MEM_PWR_STATE, mask_sh)
+
+#define AFMT_DCN31_REG_FIELD_LIST(type) \
+		type AFMT_AUDIO_INFO_UPDATE;\
+		type AFMT_AUDIO_SRC_SELECT;\
+		type AFMT_AUDIO_CHANNEL_ENABLE;\
+		type AFMT_60958_CS_UPDATE;\
+		type AFMT_AUDIO_LAYOUT_OVRD;\
+		type AFMT_60958_OSF_OVRD;\
+		type AFMT_60958_CS_CHANNEL_NUMBER_L;\
+		type AFMT_60958_CS_CLOCK_ACCURACY;\
+		type AFMT_60958_CS_CHANNEL_NUMBER_R;\
+		type AFMT_60958_CS_CHANNEL_NUMBER_2;\
+		type AFMT_60958_CS_CHANNEL_NUMBER_3;\
+		type AFMT_60958_CS_CHANNEL_NUMBER_4;\
+		type AFMT_60958_CS_CHANNEL_NUMBER_5;\
+		type AFMT_60958_CS_CHANNEL_NUMBER_6;\
+		type AFMT_60958_CS_CHANNEL_NUMBER_7;\
+		type AFMT_AUDIO_SAMPLE_SEND;\
+		type AFMT_MEM_PWR_FORCE;\
+		type AFMT_MEM_PWR_DIS;\
+		type AFMT_MEM_PWR_STATE
+
+struct dcn31_afmt_shift {
+	AFMT_DCN31_REG_FIELD_LIST(uint8_t);
+};
+
+struct dcn31_afmt_mask {
+	AFMT_DCN31_REG_FIELD_LIST(uint32_t);
+};
+
+struct dcn31_afmt {
+	struct afmt base;
+	const struct dcn31_afmt_registers *regs;
+	const struct dcn31_afmt_shift *afmt_shift;
+	const struct dcn31_afmt_mask *afmt_mask;
+};
+
+void afmt31_poweron(
+		struct afmt *afmt);
+
+void afmt31_powerdown(
+		struct afmt *afmt);
+
+void afmt31_construct(struct dcn31_afmt *afmt31,
+	struct dc_context *ctx,
+	uint32_t inst,
+	const struct dcn31_afmt_registers *afmt_regs,
+	const struct dcn31_afmt_shift *afmt_shift,
+	const struct dcn31_afmt_mask *afmt_mask);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_apg.c b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_apg.c
new file mode 100644
index 000000000000..05aac3e444b4
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_apg.c
@@ -0,0 +1,119 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+#include "dc_bios_types.h"
+#include "hw_shared.h"
+#include "dcn31_apg.h"
+#include "reg_helper.h"
+
+#define DC_LOGGER \
+		apg31->base.ctx->logger
+
+#define REG(reg)\
+	(apg31->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	apg31->apg_shift->field_name, apg31->apg_mask->field_name
+
+
+#define CTX \
+	apg31->base.ctx
+
+
+static void apg31_enable(
+	struct apg *apg)
+{
+	struct dcn31_apg *apg31 = DCN31_APG_FROM_APG(apg);
+
+	/* Reset APG */
+	REG_UPDATE(APG_CONTROL, APG_RESET, 1);
+	REG_WAIT(APG_CONTROL,
+			APG_RESET_DONE, 1,
+			1, 10);
+	REG_UPDATE(APG_CONTROL, APG_RESET, 0);
+	REG_WAIT(APG_CONTROL,
+			APG_RESET_DONE, 0,
+			1, 10);
+
+	/* Enable APG */
+	REG_UPDATE(APG_CONTROL2, APG_ENABLE, 1);
+}
+
+static void apg31_disable(
+	struct apg *apg)
+{
+	struct dcn31_apg *apg31 = DCN31_APG_FROM_APG(apg);
+
+	/* Disable APG */
+	REG_UPDATE(APG_CONTROL2, APG_ENABLE, 0);
+}
+
+static void apg31_se_audio_setup(
+	struct apg *apg,
+	unsigned int az_inst,
+	struct audio_info *audio_info)
+{
+	struct dcn31_apg *apg31 = DCN31_APG_FROM_APG(apg);
+
+	ASSERT(audio_info);
+	/* This should not happen.it does so we don't get BSOD*/
+	if (audio_info == NULL)
+		return;
+
+	/* DisplayPort only allows for one audio stream with stream ID 0 */
+	REG_UPDATE(APG_CONTROL2, APG_DP_AUDIO_STREAM_ID, 0);
+
+	/* When running in "pair mode", pairs of audio channels have their own enable
+	 * this is for really old audio drivers */
+	REG_UPDATE(APG_DBG_GEN_CONTROL, APG_DBG_AUDIO_CHANNEL_ENABLE, 0xFF);
+
+	/* Disable forced mem power off */
+	REG_UPDATE(APG_MEM_PWR, APG_MEM_PWR_FORCE, 0);
+}
+
+static struct apg_funcs dcn31_apg_funcs = {
+	.se_audio_setup			= apg31_se_audio_setup,
+	.enable_apg			= apg31_enable,
+	.disable_apg			= apg31_disable,
+};
+
+void apg31_construct(struct dcn31_apg *apg31,
+	struct dc_context *ctx,
+	uint32_t inst,
+	const struct dcn31_apg_registers *apg_regs,
+	const struct dcn31_apg_shift *apg_shift,
+	const struct dcn31_apg_mask *apg_mask)
+{
+	apg31->base.ctx = ctx;
+
+	apg31->base.inst = inst;
+	apg31->base.funcs = &dcn31_apg_funcs;
+
+	apg31->regs = apg_regs;
+	apg31->apg_shift = apg_shift;
+	apg31->apg_mask = apg_mask;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_apg.h b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_apg.h
new file mode 100644
index 000000000000..1b81f6773c53
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_apg.h
@@ -0,0 +1,111 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_DCN31_AGP_H__
+#define __DAL_DCN31_AGP_H__
+
+
+#define DCN31_APG_FROM_APG(apg)\
+	container_of(apg, struct dcn31_apg, base)
+
+#define APG_DCN31_REG_LIST(id) \
+	SRI(APG_CONTROL, APG, id), \
+	SRI(APG_CONTROL2, APG, id),\
+	SRI(APG_MEM_PWR, APG, id),\
+	SRI(APG_DBG_GEN_CONTROL, APG, id)
+
+struct dcn31_apg_registers {
+	uint32_t APG_CONTROL;
+	uint32_t APG_CONTROL2;
+	uint32_t APG_MEM_PWR;
+	uint32_t APG_DBG_GEN_CONTROL;
+};
+
+
+#define DCN31_APG_MASK_SH_LIST(mask_sh)\
+	SE_SF(APG0_APG_CONTROL, APG_RESET, mask_sh),\
+	SE_SF(APG0_APG_CONTROL, APG_RESET_DONE, mask_sh),\
+	SE_SF(APG0_APG_CONTROL2, APG_ENABLE, mask_sh),\
+	SE_SF(APG0_APG_CONTROL2, APG_DP_AUDIO_STREAM_ID, mask_sh),\
+	SE_SF(APG0_APG_DBG_GEN_CONTROL, APG_DBG_AUDIO_CHANNEL_ENABLE, mask_sh),\
+	SE_SF(APG0_APG_MEM_PWR, APG_MEM_PWR_FORCE, mask_sh)
+
+#define APG_DCN31_REG_FIELD_LIST(type) \
+		type APG_RESET;\
+		type APG_RESET_DONE;\
+		type APG_ENABLE;\
+		type APG_DP_AUDIO_STREAM_ID;\
+		type APG_DBG_AUDIO_CHANNEL_ENABLE;\
+		type APG_MEM_PWR_FORCE
+
+struct dcn31_apg_shift {
+	APG_DCN31_REG_FIELD_LIST(uint8_t);
+};
+
+struct dcn31_apg_mask {
+	APG_DCN31_REG_FIELD_LIST(uint32_t);
+};
+
+struct apg {
+	const struct apg_funcs *funcs;
+	struct dc_context *ctx;
+	int inst;
+};
+
+struct apg_funcs {
+
+	void (*setup_hdmi_audio)(
+		struct apg *apg);
+
+	void (*se_audio_setup)(
+		struct apg *apg,
+		unsigned int az_inst,
+		struct audio_info *audio_info);
+
+	void (*enable_apg)(
+		struct apg *apg);
+
+	void (*disable_apg)(
+		struct apg *apg);
+};
+
+
+
+struct dcn31_apg {
+	struct apg base;
+	const struct dcn31_apg_registers *regs;
+	const struct dcn31_apg_shift *apg_shift;
+	const struct dcn31_apg_mask *apg_mask;
+};
+
+void apg31_construct(struct dcn31_apg *apg3,
+	struct dc_context *ctx,
+	uint32_t inst,
+	const struct dcn31_apg_registers *apg_regs,
+	const struct dcn31_apg_shift *apg_shift,
+	const struct dcn31_apg_mask *apg_mask);
+
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_panel_cntl.c b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_panel_cntl.c
new file mode 100644
index 000000000000..f9961a6446f3
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_panel_cntl.c
@@ -0,0 +1,200 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+#include "core_types.h"
+#include "dc_dmub_srv.h"
+#include "dcn31_panel_cntl.h"
+#include "atom.h"
+
+#define TO_DCN31_PANEL_CNTL(panel_cntl)\
+	container_of(panel_cntl, struct dcn31_panel_cntl, base)
+
+#define CTX \
+	dcn31_panel_cntl->base.ctx
+
+#define DC_LOGGER \
+	dcn31_panel_cntl->base.ctx->logger
+
+static bool dcn31_query_backlight_info(struct panel_cntl *panel_cntl, union dmub_rb_cmd *cmd)
+{
+	struct dcn31_panel_cntl *dcn31_panel_cntl = TO_DCN31_PANEL_CNTL(panel_cntl);
+	struct dc_dmub_srv *dc_dmub_srv = panel_cntl->ctx->dmub_srv;
+
+	if (!dc_dmub_srv)
+		return false;
+
+	memset(cmd, 0, sizeof(*cmd));
+	cmd->panel_cntl.header.type = DMUB_CMD__PANEL_CNTL;
+	cmd->panel_cntl.header.sub_type = DMUB_CMD__PANEL_CNTL_QUERY_BACKLIGHT_INFO;
+	cmd->panel_cntl.header.payload_bytes = sizeof(cmd->panel_cntl.data);
+	cmd->panel_cntl.data.pwrseq_inst = dcn31_panel_cntl->base.pwrseq_inst;
+
+	return dc_wake_and_execute_dmub_cmd(dc_dmub_srv->ctx, cmd, DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY);
+}
+
+static uint32_t dcn31_get_16_bit_backlight_from_pwm(struct panel_cntl *panel_cntl)
+{
+	union dmub_rb_cmd cmd;
+
+	if (!dcn31_query_backlight_info(panel_cntl, &cmd))
+		return 0;
+
+	return cmd.panel_cntl.data.current_backlight;
+}
+
+static uint32_t dcn31_panel_cntl_hw_init(struct panel_cntl *panel_cntl)
+{
+	struct dcn31_panel_cntl *dcn31_panel_cntl = TO_DCN31_PANEL_CNTL(panel_cntl);
+	struct dc_dmub_srv *dc_dmub_srv = panel_cntl->ctx->dmub_srv;
+	union dmub_rb_cmd cmd;
+	uint32_t freq_to_set = panel_cntl->ctx->dc->debug.pwm_freq;
+
+	if (!dc_dmub_srv)
+		return 0;
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.panel_cntl.header.type = DMUB_CMD__PANEL_CNTL;
+	cmd.panel_cntl.header.sub_type = DMUB_CMD__PANEL_CNTL_HW_INIT;
+	cmd.panel_cntl.header.payload_bytes = sizeof(cmd.panel_cntl.data);
+	cmd.panel_cntl.data.pwrseq_inst = dcn31_panel_cntl->base.pwrseq_inst;
+	cmd.panel_cntl.data.bl_pwm_cntl = panel_cntl->stored_backlight_registers.BL_PWM_CNTL;
+	cmd.panel_cntl.data.bl_pwm_period_cntl = panel_cntl->stored_backlight_registers.BL_PWM_PERIOD_CNTL;
+	cmd.panel_cntl.data.bl_pwm_ref_div1 =
+		panel_cntl->stored_backlight_registers.LVTMA_PWRSEQ_REF_DIV_BL_PWM_REF_DIV;
+	cmd.panel_cntl.data.bl_pwm_ref_div2 =
+		panel_cntl->stored_backlight_registers.PANEL_PWRSEQ_REF_DIV2;
+	if (!dc_wake_and_execute_dmub_cmd(dc_dmub_srv->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY))
+		return 0;
+
+	panel_cntl->stored_backlight_registers.BL_PWM_CNTL = cmd.panel_cntl.data.bl_pwm_cntl;
+	panel_cntl->stored_backlight_registers.BL_PWM_CNTL2 = 0; /* unused */
+	panel_cntl->stored_backlight_registers.BL_PWM_PERIOD_CNTL = cmd.panel_cntl.data.bl_pwm_period_cntl;
+	panel_cntl->stored_backlight_registers.LVTMA_PWRSEQ_REF_DIV_BL_PWM_REF_DIV =
+		cmd.panel_cntl.data.bl_pwm_ref_div1;
+	panel_cntl->stored_backlight_registers.PANEL_PWRSEQ_REF_DIV2 =
+		cmd.panel_cntl.data.bl_pwm_ref_div2;
+
+	if (freq_to_set >= MIN_DEBUG_FREQ_HZ && freq_to_set <= MAX_DEBUG_FREQ_HZ) {
+		uint32_t xtal = panel_cntl->ctx->dc->res_pool->ref_clocks.dccg_ref_clock_inKhz;
+
+		memset(&cmd, 0, sizeof(cmd));
+		cmd.panel_cntl.header.type = DMUB_CMD__PANEL_CNTL;
+		cmd.panel_cntl.header.sub_type = DMUB_CMD__PANEL_DEBUG_PWM_FREQ;
+		cmd.panel_cntl.header.payload_bytes = sizeof(cmd.panel_cntl.data);
+		cmd.panel_cntl.data.pwrseq_inst = dcn31_panel_cntl->base.pwrseq_inst;
+		cmd.panel_cntl.data.bl_pwm_cntl = xtal;
+		cmd.panel_cntl.data.bl_pwm_period_cntl = freq_to_set;
+		if (!dc_wake_and_execute_dmub_cmd(dc_dmub_srv->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY))
+			return 0;
+	}
+	return cmd.panel_cntl.data.current_backlight;
+}
+
+static void dcn31_panel_cntl_destroy(struct panel_cntl **panel_cntl)
+{
+	struct dcn31_panel_cntl *dcn31_panel_cntl = TO_DCN31_PANEL_CNTL(*panel_cntl);
+
+	kfree(dcn31_panel_cntl);
+	*panel_cntl = NULL;
+}
+
+static bool dcn31_is_panel_backlight_on(struct panel_cntl *panel_cntl)
+{
+	union dmub_rb_cmd cmd;
+
+	if (!dcn31_query_backlight_info(panel_cntl, &cmd))
+		return false;
+
+	return cmd.panel_cntl.data.is_backlight_on;
+}
+
+static bool dcn31_is_panel_powered_on(struct panel_cntl *panel_cntl)
+{
+	union dmub_rb_cmd cmd;
+
+	if (!dcn31_query_backlight_info(panel_cntl, &cmd))
+		return false;
+
+	return cmd.panel_cntl.data.is_powered_on;
+}
+
+static void dcn31_store_backlight_level(struct panel_cntl *panel_cntl)
+{
+	union dmub_rb_cmd cmd;
+
+	if (!dcn31_query_backlight_info(panel_cntl, &cmd))
+		return;
+
+	panel_cntl->stored_backlight_registers.BL_PWM_CNTL = cmd.panel_cntl.data.bl_pwm_cntl;
+	panel_cntl->stored_backlight_registers.BL_PWM_CNTL2 = 0; /* unused */
+	panel_cntl->stored_backlight_registers.BL_PWM_PERIOD_CNTL = cmd.panel_cntl.data.bl_pwm_period_cntl;
+	panel_cntl->stored_backlight_registers.LVTMA_PWRSEQ_REF_DIV_BL_PWM_REF_DIV =
+		cmd.panel_cntl.data.bl_pwm_ref_div1;
+}
+
+static const struct panel_cntl_funcs dcn31_link_panel_cntl_funcs = {
+	.destroy = dcn31_panel_cntl_destroy,
+	.hw_init = dcn31_panel_cntl_hw_init,
+	.is_panel_backlight_on = dcn31_is_panel_backlight_on,
+	.is_panel_powered_on = dcn31_is_panel_powered_on,
+	.store_backlight_level = dcn31_store_backlight_level,
+	.get_current_backlight = dcn31_get_16_bit_backlight_from_pwm,
+};
+
+void dcn31_panel_cntl_construct(
+	struct dcn31_panel_cntl *dcn31_panel_cntl,
+	const struct panel_cntl_init_data *init_data)
+{
+
+	dcn31_panel_cntl->base.funcs = &dcn31_link_panel_cntl_funcs;
+	dcn31_panel_cntl->base.ctx = init_data->ctx;
+	dcn31_panel_cntl->base.inst = init_data->inst;
+
+	if (dcn31_panel_cntl->base.ctx->dc->config.support_edp0_on_dp1) {
+		//If supported, power sequencer mapping shall follow the DIG instance
+		uint8_t pwrseq_inst = 0xF;
+
+		switch (init_data->eng_id) {
+		case ENGINE_ID_DIGA:
+			pwrseq_inst = 0;
+			break;
+		case ENGINE_ID_DIGB:
+			pwrseq_inst = 1;
+			break;
+		default:
+			DC_LOG_WARNING("Unsupported pwrseq engine id: %d!\n", init_data->eng_id);
+			ASSERT(false);
+			break;
+		}
+
+		dcn31_panel_cntl->base.pwrseq_inst = pwrseq_inst;
+	} else {
+		/* If not supported, pwrseq will be assigned in order,
+		 * so first pwrseq will be assigned to first panel instance (legacy behavior)
+		 */
+		dcn31_panel_cntl->base.pwrseq_inst = dcn31_panel_cntl->base.inst;
+	}
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_panel_cntl.h b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_panel_cntl.h
new file mode 100644
index 000000000000..8cf0259e211e
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_panel_cntl.h
@@ -0,0 +1,43 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_PANEL_CNTL__DCN31_H__
+#define __DC_PANEL_CNTL__DCN31_H__
+
+#include "panel_cntl.h"
+#include "dce/dce_panel_cntl.h"
+
+#define MIN_DEBUG_FREQ_HZ 200
+#define MAX_DEBUG_FREQ_HZ 6250
+
+struct dcn31_panel_cntl {
+	struct panel_cntl base;
+};
+
+void dcn31_panel_cntl_construct(
+	struct dcn31_panel_cntl *dcn31_panel_cntl,
+	const struct panel_cntl_init_data *init_data);
+
+#endif /* __DC_PANEL_CNTL__DCN31_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_vpg.c b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_vpg.c
new file mode 100644
index 000000000000..cfb923d85630
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_vpg.c
@@ -0,0 +1,92 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dc_bios_types.h"
+#include "dcn30/dcn30_vpg.h"
+#include "dcn31_vpg.h"
+#include "reg_helper.h"
+#include "dc/dc.h"
+
+#define DC_LOGGER \
+		vpg31->base.ctx->logger
+
+#define REG(reg)\
+	(vpg31->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	vpg31->vpg_shift->field_name, vpg31->vpg_mask->field_name
+
+
+#define CTX \
+	vpg31->base.ctx
+
+static struct vpg_funcs dcn31_vpg_funcs = {
+	.update_generic_info_packet	= vpg3_update_generic_info_packet,
+	.vpg_poweron = vpg31_poweron,
+	.vpg_powerdown = vpg31_powerdown,
+};
+
+void vpg31_powerdown(struct vpg *vpg)
+{
+	struct dcn31_vpg *vpg31 = DCN31_VPG_FROM_VPG(vpg);
+
+	if (vpg->ctx->dc->debug.enable_mem_low_power.bits.vpg == false)
+		return;
+
+	REG_UPDATE_2(VPG_MEM_PWR, VPG_GSP_MEM_LIGHT_SLEEP_DIS, 0, VPG_GSP_LIGHT_SLEEP_FORCE, 1);
+}
+
+void vpg31_poweron(struct vpg *vpg)
+{
+	struct dcn31_vpg *vpg31 = DCN31_VPG_FROM_VPG(vpg);
+
+	uint32_t vpg_gsp_mem_pwr_state;
+
+	REG_GET(VPG_MEM_PWR, VPG_GSP_MEM_PWR_STATE, &vpg_gsp_mem_pwr_state);
+
+	if (vpg->ctx->dc->debug.enable_mem_low_power.bits.vpg == false &&
+			vpg_gsp_mem_pwr_state == 0)
+		return;
+
+	REG_UPDATE_2(VPG_MEM_PWR, VPG_GSP_MEM_LIGHT_SLEEP_DIS, 1, VPG_GSP_LIGHT_SLEEP_FORCE, 0);
+}
+
+void vpg31_construct(struct dcn31_vpg *vpg31,
+	struct dc_context *ctx,
+	uint32_t inst,
+	const struct dcn31_vpg_registers *vpg_regs,
+	const struct dcn31_vpg_shift *vpg_shift,
+	const struct dcn31_vpg_mask *vpg_mask)
+{
+	vpg31->base.ctx = ctx;
+
+	vpg31->base.inst = inst;
+	vpg31->base.funcs = &dcn31_vpg_funcs;
+
+	vpg31->regs = vpg_regs;
+	vpg31->vpg_shift = vpg_shift;
+	vpg31->vpg_mask = vpg_mask;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_vpg.h b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_vpg.h
new file mode 100644
index 000000000000..609e58dbd056
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_vpg.h
@@ -0,0 +1,163 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DAL_DCN31_VPG_H__
+#define __DAL_DCN31_VPG_H__
+
+#include "vpg.h"
+
+#define DCN31_VPG_FROM_VPG(vpg)\
+	container_of(vpg, struct dcn31_vpg, base)
+
+#define VPG_DCN31_REG_LIST(id) \
+	SRI(VPG_GENERIC_STATUS, VPG, id), \
+	SRI(VPG_GENERIC_PACKET_ACCESS_CTRL, VPG, id), \
+	SRI(VPG_GENERIC_PACKET_DATA, VPG, id), \
+	SRI(VPG_GSP_FRAME_UPDATE_CTRL, VPG, id), \
+	SRI(VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG, id), \
+	SRI(VPG_MEM_PWR, VPG, id)
+
+struct dcn31_vpg_registers {
+	uint32_t VPG_GENERIC_STATUS;
+	uint32_t VPG_GENERIC_PACKET_ACCESS_CTRL;
+	uint32_t VPG_GENERIC_PACKET_DATA;
+	uint32_t VPG_GSP_FRAME_UPDATE_CTRL;
+	uint32_t VPG_GSP_IMMEDIATE_UPDATE_CTRL;
+	uint32_t VPG_MEM_PWR;
+};
+
+#define DCN31_VPG_MASK_SH_LIST(mask_sh)\
+	SE_SF(VPG0_VPG_GENERIC_STATUS, VPG_GENERIC_CONFLICT_OCCURED, mask_sh),\
+	SE_SF(VPG0_VPG_GENERIC_STATUS, VPG_GENERIC_CONFLICT_CLR, mask_sh),\
+	SE_SF(VPG0_VPG_GENERIC_PACKET_ACCESS_CTRL, VPG_GENERIC_DATA_INDEX, mask_sh),\
+	SE_SF(VPG0_VPG_GENERIC_PACKET_DATA, VPG_GENERIC_DATA_BYTE0, mask_sh),\
+	SE_SF(VPG0_VPG_GENERIC_PACKET_DATA, VPG_GENERIC_DATA_BYTE1, mask_sh),\
+	SE_SF(VPG0_VPG_GENERIC_PACKET_DATA, VPG_GENERIC_DATA_BYTE2, mask_sh),\
+	SE_SF(VPG0_VPG_GENERIC_PACKET_DATA, VPG_GENERIC_DATA_BYTE3, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC0_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC1_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC2_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC3_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC4_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC5_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC6_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC7_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC8_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC9_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC10_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC11_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC12_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC13_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_FRAME_UPDATE_CTRL, VPG_GENERIC14_FRAME_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC0_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC1_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC2_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC3_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC4_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC5_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC6_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC7_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC8_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC9_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC10_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC11_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC12_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC13_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_GSP_IMMEDIATE_UPDATE_CTRL, VPG_GENERIC14_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(VPG0_VPG_MEM_PWR, VPG_GSP_MEM_LIGHT_SLEEP_DIS, mask_sh),\
+	SE_SF(VPG0_VPG_MEM_PWR, VPG_GSP_LIGHT_SLEEP_FORCE, mask_sh),\
+	SE_SF(VPG0_VPG_MEM_PWR, VPG_GSP_MEM_PWR_STATE, mask_sh)
+
+#define VPG_DCN31_REG_FIELD_LIST(type) \
+	type VPG_GENERIC_CONFLICT_OCCURED;\
+	type VPG_GENERIC_CONFLICT_CLR;\
+	type VPG_GENERIC_DATA_INDEX;\
+	type VPG_GENERIC_DATA_BYTE0;\
+	type VPG_GENERIC_DATA_BYTE1;\
+	type VPG_GENERIC_DATA_BYTE2;\
+	type VPG_GENERIC_DATA_BYTE3;\
+	type VPG_GENERIC0_FRAME_UPDATE;\
+	type VPG_GENERIC1_FRAME_UPDATE;\
+	type VPG_GENERIC2_FRAME_UPDATE;\
+	type VPG_GENERIC3_FRAME_UPDATE;\
+	type VPG_GENERIC4_FRAME_UPDATE;\
+	type VPG_GENERIC5_FRAME_UPDATE;\
+	type VPG_GENERIC6_FRAME_UPDATE;\
+	type VPG_GENERIC7_FRAME_UPDATE;\
+	type VPG_GENERIC8_FRAME_UPDATE;\
+	type VPG_GENERIC9_FRAME_UPDATE;\
+	type VPG_GENERIC10_FRAME_UPDATE;\
+	type VPG_GENERIC11_FRAME_UPDATE;\
+	type VPG_GENERIC12_FRAME_UPDATE;\
+	type VPG_GENERIC13_FRAME_UPDATE;\
+	type VPG_GENERIC14_FRAME_UPDATE;\
+	type VPG_GENERIC0_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC1_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC2_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC3_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC4_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC5_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC6_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC7_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC8_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC9_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC10_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC11_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC12_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC13_IMMEDIATE_UPDATE;\
+	type VPG_GENERIC14_IMMEDIATE_UPDATE;\
+	type VPG_GSP_MEM_LIGHT_SLEEP_DIS;\
+	type VPG_GSP_LIGHT_SLEEP_FORCE;\
+	type VPG_GSP_MEM_PWR_STATE
+
+struct dcn31_vpg_shift {
+	VPG_DCN31_REG_FIELD_LIST(uint8_t);
+};
+
+struct dcn31_vpg_mask {
+	VPG_DCN31_REG_FIELD_LIST(uint32_t);
+};
+
+struct dcn31_vpg {
+	struct vpg base;
+	const struct dcn31_vpg_registers *regs;
+	const struct dcn31_vpg_shift *vpg_shift;
+	const struct dcn31_vpg_mask *vpg_mask;
+};
+
+void vpg31_poweron(
+		struct vpg *vpg);
+
+void vpg31_powerdown(
+		struct vpg *vpg);
+
+void vpg31_construct(struct dcn31_vpg *vpg31,
+	struct dc_context *ctx,
+	uint32_t inst,
+	const struct dcn31_vpg_registers *vpg_regs,
+	const struct dcn31_vpg_shift *vpg_shift,
+	const struct dcn31_vpg_mask *vpg_mask);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dio/Makefile b/drivers/gpu/drm/amd/display/dc/dio/Makefile
new file mode 100644
index 000000000000..0dfd480976f7
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/Makefile
@@ -0,0 +1,117 @@
+#
+# Copyright 2020 Advanced Micro Devices, Inc.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+# Authors: AMD
+#
+#
+
+ifdef CONFIG_DRM_AMD_DC_FP
+###############################################################################
+# DCN10
+###############################################################################
+DIO_DCN10 = dcn10_link_encoder.o dcn10_stream_encoder.o
+
+AMD_DAL_DIO_DCN10 = $(addprefix $(AMDDALPATH)/dc/dio/dcn10/,$(DIO_DCN10))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DIO_DCN10)
+
+###############################################################################
+# DCN20
+###############################################################################
+DIO_DCN20 = dcn20_link_encoder.o dcn20_stream_encoder.o
+
+AMD_DAL_DIO_DCN20 = $(addprefix $(AMDDALPATH)/dc/dio/dcn20/,$(DIO_DCN20))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DIO_DCN20)
+
+###############################################################################
+# DCN30
+###############################################################################
+DIO_DCN30 = dcn30_dio_link_encoder.o dcn30_dio_stream_encoder.o
+
+AMD_DAL_DIO_DCN30 = $(addprefix $(AMDDALPATH)/dc/dio/dcn30/,$(DIO_DCN30))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DIO_DCN30)
+
+###############################################################################
+# DCN301
+###############################################################################
+DIO_DCN301 = dcn301_dio_link_encoder.o
+
+AMD_DAL_DIO_DCN301 = $(addprefix $(AMDDALPATH)/dc/dio/dcn301/,$(DIO_DCN301))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DIO_DCN301)
+
+###############################################################################
+# DCN31
+###############################################################################
+DIO_DCN31 = dcn31_dio_link_encoder.o
+
+AMD_DAL_DIO_DCN31 = $(addprefix $(AMDDALPATH)/dc/dio/dcn31/,$(DIO_DCN31))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DIO_DCN31)
+
+###############################################################################
+# DCN314
+###############################################################################
+DIO_DCN314 = dcn314_dio_stream_encoder.o
+
+AMD_DAL_DIO_DCN314 = $(addprefix $(AMDDALPATH)/dc/dio/dcn314/,$(DIO_DCN314))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DIO_DCN314)
+
+###############################################################################
+# DCN32
+###############################################################################
+DIO_DCN32 = dcn32_dio_link_encoder.o dcn32_dio_stream_encoder.o
+
+AMD_DAL_DIO_DCN32 = $(addprefix $(AMDDALPATH)/dc/dio/dcn32/,$(DIO_DCN32))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DIO_DCN32)
+
+###############################################################################
+# DCN35
+###############################################################################
+DIO_DCN35 = dcn35_dio_link_encoder.o dcn35_dio_stream_encoder.o
+
+AMD_DAL_DIO_DCN35 = $(addprefix $(AMDDALPATH)/dc/dio/dcn35/,$(DIO_DCN35))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DIO_DCN35)
+
+###############################################################################
+# DCN321
+###############################################################################
+DIO_DCN321 = dcn321_dio_link_encoder.o
+
+AMD_DAL_DIO_DCN321 = $(addprefix $(AMDDALPATH)/dc/dio/dcn321/,$(DIO_DCN321))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DIO_DCN321)
+
+
+###############################################################################
+# DCN401
+###############################################################################
+DIO_DCN401 = dcn401_dio_link_encoder.o dcn401_dio_stream_encoder.o
+
+AMD_DAL_DIO_DCN401 = $(addprefix $(AMDDALPATH)/dc/dio/dcn401/,$(DIO_DCN401))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DIO_DCN401)
+endif
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn10/dcn10_link_encoder.c b/drivers/gpu/drm/amd/display/dc/dio/dcn10/dcn10_link_encoder.c
new file mode 100644
index 000000000000..1c1228116487
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn10/dcn10_link_encoder.c
@@ -0,0 +1,1474 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+
+#include "core_types.h"
+#include "link_encoder.h"
+#include "dcn10_link_encoder.h"
+#include "stream_encoder.h"
+#include "dc_bios_types.h"
+
+#include "gpio_service_interface.h"
+
+#define CTX \
+	enc10->base.ctx
+#define DC_LOGGER \
+	enc10->base.ctx->logger
+
+#define REG(reg)\
+	(enc10->link_regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	enc10->link_shift->field_name, enc10->link_mask->field_name
+
+
+/*
+ * @brief
+ * Trigger Source Select
+ * ASIC-dependent, actual values for register programming
+ */
+#define DCN10_DIG_FE_SOURCE_SELECT_INVALID 0x0
+#define DCN10_DIG_FE_SOURCE_SELECT_DIGA 0x1
+#define DCN10_DIG_FE_SOURCE_SELECT_DIGB 0x2
+#define DCN10_DIG_FE_SOURCE_SELECT_DIGC 0x4
+#define DCN10_DIG_FE_SOURCE_SELECT_DIGD 0x08
+#define DCN10_DIG_FE_SOURCE_SELECT_DIGE 0x10
+#define DCN10_DIG_FE_SOURCE_SELECT_DIGF 0x20
+#define DCN10_DIG_FE_SOURCE_SELECT_DIGG 0x40
+
+enum {
+	DP_MST_UPDATE_MAX_RETRY = 50
+};
+
+static const struct link_encoder_funcs dcn10_lnk_enc_funcs = {
+	.validate_output_with_stream =
+		dcn10_link_encoder_validate_output_with_stream,
+	.hw_init = dcn10_link_encoder_hw_init,
+	.setup = dcn10_link_encoder_setup,
+	.enable_tmds_output = dcn10_link_encoder_enable_tmds_output,
+	.enable_dp_output = dcn10_link_encoder_enable_dp_output,
+	.enable_dp_mst_output = dcn10_link_encoder_enable_dp_mst_output,
+	.disable_output = dcn10_link_encoder_disable_output,
+	.dp_set_lane_settings = dcn10_link_encoder_dp_set_lane_settings,
+	.dp_set_phy_pattern = dcn10_link_encoder_dp_set_phy_pattern,
+	.update_mst_stream_allocation_table =
+		dcn10_link_encoder_update_mst_stream_allocation_table,
+	.psr_program_dp_dphy_fast_training =
+			dcn10_psr_program_dp_dphy_fast_training,
+	.psr_program_secondary_packet = dcn10_psr_program_secondary_packet,
+	.connect_dig_be_to_fe = dcn10_link_encoder_connect_dig_be_to_fe,
+	.enable_hpd = dcn10_link_encoder_enable_hpd,
+	.disable_hpd = dcn10_link_encoder_disable_hpd,
+	.is_dig_enabled = dcn10_is_dig_enabled,
+	.get_dig_frontend = dcn10_get_dig_frontend,
+	.get_dig_mode = dcn10_get_dig_mode,
+	.destroy = dcn10_link_encoder_destroy,
+	.get_max_link_cap = dcn10_link_encoder_get_max_link_cap,
+};
+
+static enum bp_result link_transmitter_control(
+	struct dcn10_link_encoder *enc10,
+	struct bp_transmitter_control *cntl)
+{
+	enum bp_result result;
+	struct dc_bios *bp = enc10->base.ctx->dc_bios;
+
+	result = bp->funcs->transmitter_control(bp, cntl);
+
+	return result;
+}
+
+static void enable_phy_bypass_mode(
+	struct dcn10_link_encoder *enc10,
+	bool enable)
+{
+	/* This register resides in DP back end block;
+	 * transmitter is used for the offset
+	 */
+	REG_UPDATE(DP_DPHY_CNTL, DPHY_BYPASS, enable);
+
+}
+
+static void disable_prbs_symbols(
+	struct dcn10_link_encoder *enc10,
+	bool disable)
+{
+	/* This register resides in DP back end block;
+	 * transmitter is used for the offset
+	 */
+	REG_UPDATE_4(DP_DPHY_CNTL,
+			DPHY_ATEST_SEL_LANE0, disable,
+			DPHY_ATEST_SEL_LANE1, disable,
+			DPHY_ATEST_SEL_LANE2, disable,
+			DPHY_ATEST_SEL_LANE3, disable);
+}
+
+static void disable_prbs_mode(
+	struct dcn10_link_encoder *enc10)
+{
+	REG_UPDATE(DP_DPHY_PRBS_CNTL, DPHY_PRBS_EN, 0);
+}
+
+static void program_pattern_symbols(
+	struct dcn10_link_encoder *enc10,
+	uint16_t pattern_symbols[8])
+{
+	/* This register resides in DP back end block;
+	 * transmitter is used for the offset
+	 */
+	REG_SET_3(DP_DPHY_SYM0, 0,
+			DPHY_SYM1, pattern_symbols[0],
+			DPHY_SYM2, pattern_symbols[1],
+			DPHY_SYM3, pattern_symbols[2]);
+
+	/* This register resides in DP back end block;
+	 * transmitter is used for the offset
+	 */
+	REG_SET_3(DP_DPHY_SYM1, 0,
+			DPHY_SYM4, pattern_symbols[3],
+			DPHY_SYM5, pattern_symbols[4],
+			DPHY_SYM6, pattern_symbols[5]);
+
+	/* This register resides in DP back end block;
+	 * transmitter is used for the offset
+	 */
+	REG_SET_2(DP_DPHY_SYM2, 0,
+			DPHY_SYM7, pattern_symbols[6],
+			DPHY_SYM8, pattern_symbols[7]);
+}
+
+static void set_dp_phy_pattern_d102(
+	struct dcn10_link_encoder *enc10)
+{
+	/* Disable PHY Bypass mode to setup the test pattern */
+	enable_phy_bypass_mode(enc10, false);
+
+	/* For 10-bit PRBS or debug symbols
+	 * please use the following sequence:
+	 *
+	 * Enable debug symbols on the lanes
+	 */
+	disable_prbs_symbols(enc10, true);
+
+	/* Disable PRBS mode */
+	disable_prbs_mode(enc10);
+
+	/* Program debug symbols to be output */
+	{
+		uint16_t pattern_symbols[8] = {
+			0x2AA, 0x2AA, 0x2AA, 0x2AA,
+			0x2AA, 0x2AA, 0x2AA, 0x2AA
+		};
+
+		program_pattern_symbols(enc10, pattern_symbols);
+	}
+
+	/* Enable phy bypass mode to enable the test pattern */
+
+	enable_phy_bypass_mode(enc10, true);
+}
+
+static void set_link_training_complete(
+	struct dcn10_link_encoder *enc10,
+	bool complete)
+{
+	/* This register resides in DP back end block;
+	 * transmitter is used for the offset
+	 */
+	REG_UPDATE(DP_LINK_CNTL, DP_LINK_TRAINING_COMPLETE, complete);
+
+}
+
+void dcn10_link_encoder_set_dp_phy_pattern_training_pattern(
+	struct link_encoder *enc,
+	uint32_t index)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	/* Write Training Pattern */
+
+	REG_WRITE(DP_DPHY_TRAINING_PATTERN_SEL, index);
+
+	/* Set HW Register Training Complete to false */
+
+	set_link_training_complete(enc10, false);
+
+	/* Disable PHY Bypass mode to output Training Pattern */
+
+	enable_phy_bypass_mode(enc10, false);
+
+	/* Disable PRBS mode */
+	disable_prbs_mode(enc10);
+}
+
+static void setup_panel_mode(
+	struct dcn10_link_encoder *enc10,
+	enum dp_panel_mode panel_mode)
+{
+	uint32_t value;
+
+	if (!REG(DP_DPHY_INTERNAL_CTRL))
+		return;
+
+	value = REG_READ(DP_DPHY_INTERNAL_CTRL);
+
+	switch (panel_mode) {
+	case DP_PANEL_MODE_EDP:
+		value = 0x1;
+		break;
+	case DP_PANEL_MODE_SPECIAL:
+		value = 0x11;
+		break;
+	default:
+		value = 0x0;
+		break;
+	}
+
+	REG_WRITE(DP_DPHY_INTERNAL_CTRL, value);
+}
+
+static void set_dp_phy_pattern_symbol_error(
+	struct dcn10_link_encoder *enc10)
+{
+	/* Disable PHY Bypass mode to setup the test pattern */
+	enable_phy_bypass_mode(enc10, false);
+
+	/* program correct panel mode*/
+	setup_panel_mode(enc10, DP_PANEL_MODE_DEFAULT);
+
+	/* A PRBS23 pattern is used for most DP electrical measurements. */
+
+	/* Enable PRBS symbols on the lanes */
+	disable_prbs_symbols(enc10, false);
+
+	/* For PRBS23 Set bit DPHY_PRBS_SEL=1 and Set bit DPHY_PRBS_EN=1 */
+	REG_UPDATE_2(DP_DPHY_PRBS_CNTL,
+			DPHY_PRBS_SEL, 1,
+			DPHY_PRBS_EN, 1);
+
+	/* Enable phy bypass mode to enable the test pattern */
+	enable_phy_bypass_mode(enc10, true);
+}
+
+static void set_dp_phy_pattern_prbs7(
+	struct dcn10_link_encoder *enc10)
+{
+	/* Disable PHY Bypass mode to setup the test pattern */
+	enable_phy_bypass_mode(enc10, false);
+
+	/* A PRBS7 pattern is used for most DP electrical measurements. */
+
+	/* Enable PRBS symbols on the lanes */
+	disable_prbs_symbols(enc10, false);
+
+	/* For PRBS7 Set bit DPHY_PRBS_SEL=0 and Set bit DPHY_PRBS_EN=1 */
+	REG_UPDATE_2(DP_DPHY_PRBS_CNTL,
+			DPHY_PRBS_SEL, 0,
+			DPHY_PRBS_EN, 1);
+
+	/* Enable phy bypass mode to enable the test pattern */
+	enable_phy_bypass_mode(enc10, true);
+}
+
+static void set_dp_phy_pattern_80bit_custom(
+	struct dcn10_link_encoder *enc10,
+	const uint8_t *pattern)
+{
+	/* Disable PHY Bypass mode to setup the test pattern */
+	enable_phy_bypass_mode(enc10, false);
+
+	/* Enable debug symbols on the lanes */
+
+	disable_prbs_symbols(enc10, true);
+
+	/* Enable PHY bypass mode to enable the test pattern */
+	/* TODO is it really needed ? */
+
+	enable_phy_bypass_mode(enc10, true);
+
+	/* Program 80 bit custom pattern */
+	{
+		uint16_t pattern_symbols[8];
+
+		pattern_symbols[0] =
+			((pattern[1] & 0x03) << 8) | pattern[0];
+		pattern_symbols[1] =
+			((pattern[2] & 0x0f) << 6) | ((pattern[1] >> 2) & 0x3f);
+		pattern_symbols[2] =
+			((pattern[3] & 0x3f) << 4) | ((pattern[2] >> 4) & 0x0f);
+		pattern_symbols[3] =
+			(pattern[4] << 2) | ((pattern[3] >> 6) & 0x03);
+		pattern_symbols[4] =
+			((pattern[6] & 0x03) << 8) | pattern[5];
+		pattern_symbols[5] =
+			((pattern[7] & 0x0f) << 6) | ((pattern[6] >> 2) & 0x3f);
+		pattern_symbols[6] =
+			((pattern[8] & 0x3f) << 4) | ((pattern[7] >> 4) & 0x0f);
+		pattern_symbols[7] =
+			(pattern[9] << 2) | ((pattern[8] >> 6) & 0x03);
+
+		program_pattern_symbols(enc10, pattern_symbols);
+	}
+
+	/* Enable phy bypass mode to enable the test pattern */
+
+	enable_phy_bypass_mode(enc10, true);
+}
+
+static void set_dp_phy_pattern_hbr2_compliance_cp2520_2(
+	struct dcn10_link_encoder *enc10,
+	unsigned int cp2520_pattern)
+{
+
+	/* previously there is a register DP_HBR2_EYE_PATTERN
+	 * that is enabled to get the pattern.
+	 * But it does not work with the latest spec change,
+	 * so we are programming the following registers manually.
+	 *
+	 * The following settings have been confirmed
+	 * by Nick Chorney and Sandra Liu
+	 */
+
+	/* Disable PHY Bypass mode to setup the test pattern */
+
+	enable_phy_bypass_mode(enc10, false);
+
+	/* Setup DIG encoder in DP SST mode */
+	enc10->base.funcs->setup(&enc10->base, SIGNAL_TYPE_DISPLAY_PORT);
+
+	/* ensure normal panel mode. */
+	setup_panel_mode(enc10, DP_PANEL_MODE_DEFAULT);
+
+	/* no vbid after BS (SR)
+	 * DP_LINK_FRAMING_CNTL changed history Sandra Liu
+	 * 11000260 / 11000104 / 110000FC
+	 */
+	REG_UPDATE_3(DP_LINK_FRAMING_CNTL,
+			DP_IDLE_BS_INTERVAL, 0xFC,
+			DP_VBID_DISABLE, 1,
+			DP_VID_ENHANCED_FRAME_MODE, 1);
+
+	/* swap every BS with SR */
+	REG_UPDATE(DP_DPHY_SCRAM_CNTL, DPHY_SCRAMBLER_BS_COUNT, 0);
+
+	/* select cp2520 patterns */
+	if (REG(DP_DPHY_HBR2_PATTERN_CONTROL))
+		REG_UPDATE(DP_DPHY_HBR2_PATTERN_CONTROL,
+				DP_DPHY_HBR2_PATTERN_CONTROL, cp2520_pattern);
+	else
+		/* pre-DCE11 can only generate CP2520 pattern 2 */
+		ASSERT(cp2520_pattern == 2);
+
+	/* set link training complete */
+	set_link_training_complete(enc10, true);
+
+	/* disable video stream */
+	REG_UPDATE(DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, 0);
+
+	/* Disable PHY Bypass mode to setup the test pattern */
+	enable_phy_bypass_mode(enc10, false);
+}
+
+static void set_dp_phy_pattern_passthrough_mode(
+	struct dcn10_link_encoder *enc10,
+	enum dp_panel_mode panel_mode)
+{
+	/* program correct panel mode */
+	setup_panel_mode(enc10, panel_mode);
+
+	/* restore LINK_FRAMING_CNTL and DPHY_SCRAMBLER_BS_COUNT
+	 * in case we were doing HBR2 compliance pattern before
+	 */
+	REG_UPDATE_3(DP_LINK_FRAMING_CNTL,
+			DP_IDLE_BS_INTERVAL, 0x2000,
+			DP_VBID_DISABLE, 0,
+			DP_VID_ENHANCED_FRAME_MODE, 1);
+
+	REG_UPDATE(DP_DPHY_SCRAM_CNTL, DPHY_SCRAMBLER_BS_COUNT, 0x1FF);
+
+	/* set link training complete */
+	set_link_training_complete(enc10, true);
+
+	/* Disable PHY Bypass mode to setup the test pattern */
+	enable_phy_bypass_mode(enc10, false);
+
+	/* Disable PRBS mode */
+	disable_prbs_mode(enc10);
+}
+
+/* return value is bit-vector */
+static uint8_t get_frontend_source(
+	enum engine_id engine)
+{
+	switch (engine) {
+	case ENGINE_ID_DIGA:
+		return DCN10_DIG_FE_SOURCE_SELECT_DIGA;
+	case ENGINE_ID_DIGB:
+		return DCN10_DIG_FE_SOURCE_SELECT_DIGB;
+	case ENGINE_ID_DIGC:
+		return DCN10_DIG_FE_SOURCE_SELECT_DIGC;
+	case ENGINE_ID_DIGD:
+		return DCN10_DIG_FE_SOURCE_SELECT_DIGD;
+	case ENGINE_ID_DIGE:
+		return DCN10_DIG_FE_SOURCE_SELECT_DIGE;
+	case ENGINE_ID_DIGF:
+		return DCN10_DIG_FE_SOURCE_SELECT_DIGF;
+	case ENGINE_ID_DIGG:
+		return DCN10_DIG_FE_SOURCE_SELECT_DIGG;
+	default:
+		ASSERT_CRITICAL(false);
+		return DCN10_DIG_FE_SOURCE_SELECT_INVALID;
+	}
+}
+
+unsigned int dcn10_get_dig_frontend(struct link_encoder *enc)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	int32_t value;
+	enum engine_id result;
+
+	REG_GET(DIG_BE_CNTL, DIG_FE_SOURCE_SELECT, &value);
+
+	switch (value) {
+	case DCN10_DIG_FE_SOURCE_SELECT_DIGA:
+		result = ENGINE_ID_DIGA;
+		break;
+	case DCN10_DIG_FE_SOURCE_SELECT_DIGB:
+		result = ENGINE_ID_DIGB;
+		break;
+	case DCN10_DIG_FE_SOURCE_SELECT_DIGC:
+		result = ENGINE_ID_DIGC;
+		break;
+	case DCN10_DIG_FE_SOURCE_SELECT_DIGD:
+		result = ENGINE_ID_DIGD;
+		break;
+	case DCN10_DIG_FE_SOURCE_SELECT_DIGE:
+		result = ENGINE_ID_DIGE;
+		break;
+	case DCN10_DIG_FE_SOURCE_SELECT_DIGF:
+		result = ENGINE_ID_DIGF;
+		break;
+	case DCN10_DIG_FE_SOURCE_SELECT_DIGG:
+		result = ENGINE_ID_DIGG;
+		break;
+	default:
+		// invalid source select DIG
+		result = ENGINE_ID_UNKNOWN;
+	}
+
+	return result;
+
+}
+
+void enc1_configure_encoder(
+	struct dcn10_link_encoder *enc10,
+	const struct dc_link_settings *link_settings)
+{
+	/* set number of lanes */
+	REG_SET(DP_CONFIG, 0,
+			DP_UDI_LANES, link_settings->lane_count - LANE_COUNT_ONE);
+
+	/* setup scrambler */
+	REG_UPDATE(DP_DPHY_SCRAM_CNTL, DPHY_SCRAMBLER_ADVANCE, 1);
+}
+
+void dcn10_psr_program_dp_dphy_fast_training(struct link_encoder *enc,
+			bool exit_link_training_required)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	if (exit_link_training_required)
+		REG_UPDATE(DP_DPHY_FAST_TRAINING,
+				DPHY_RX_FAST_TRAINING_CAPABLE, 1);
+	else {
+		REG_UPDATE(DP_DPHY_FAST_TRAINING,
+				DPHY_RX_FAST_TRAINING_CAPABLE, 0);
+		/*In DCE 11, we are able to pre-program a Force SR register
+		 * to be able to trigger SR symbol after 5 idle patterns
+		 * transmitted. Upon PSR Exit, DMCU can trigger
+		 * DPHY_LOAD_BS_COUNT_START = 1. Upon writing 1 to
+		 * DPHY_LOAD_BS_COUNT_START and the internal counter
+		 * reaches DPHY_LOAD_BS_COUNT, the next BS symbol will be
+		 * replaced by SR symbol once.
+		 */
+
+		REG_UPDATE(DP_DPHY_BS_SR_SWAP_CNTL, DPHY_LOAD_BS_COUNT, 0x5);
+	}
+}
+
+void dcn10_psr_program_secondary_packet(struct link_encoder *enc,
+			unsigned int sdp_transmit_line_num_deadline)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	REG_UPDATE_2(DP_SEC_CNTL1,
+		DP_SEC_GSP0_LINE_NUM, sdp_transmit_line_num_deadline,
+		DP_SEC_GSP0_PRIORITY, 1);
+}
+
+bool dcn10_is_dig_enabled(struct link_encoder *enc)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	uint32_t value;
+
+	REG_GET(DIG_BE_EN_CNTL, DIG_ENABLE, &value);
+	return value;
+}
+
+static void link_encoder_disable(struct dcn10_link_encoder *enc10)
+{
+	/* reset training pattern */
+	REG_SET(DP_DPHY_TRAINING_PATTERN_SEL, 0,
+			DPHY_TRAINING_PATTERN_SEL, 0);
+
+	/* reset training complete */
+	REG_UPDATE(DP_LINK_CNTL, DP_LINK_TRAINING_COMPLETE, 0);
+
+	/* reset panel mode */
+	setup_panel_mode(enc10, DP_PANEL_MODE_DEFAULT);
+}
+
+static void hpd_initialize(
+	struct dcn10_link_encoder *enc10)
+{
+	/* Associate HPD with DIG_BE */
+	enum hpd_source_id hpd_source = enc10->base.hpd_source;
+
+	REG_UPDATE(DIG_BE_CNTL, DIG_HPD_SELECT, hpd_source);
+}
+
+bool dcn10_link_encoder_validate_dvi_output(
+	const struct dcn10_link_encoder *enc10,
+	enum signal_type connector_signal,
+	enum signal_type signal,
+	const struct dc_crtc_timing *crtc_timing)
+{
+	uint32_t max_pixel_clock = TMDS_MAX_PIXEL_CLOCK;
+
+	if (signal == SIGNAL_TYPE_DVI_DUAL_LINK)
+		max_pixel_clock *= 2;
+
+	/* This handles the case of HDMI downgrade to DVI we don't want to
+	 * we don't want to cap the pixel clock if the DDI is not DVI.
+	 */
+	if (connector_signal != SIGNAL_TYPE_DVI_DUAL_LINK &&
+			connector_signal != SIGNAL_TYPE_DVI_SINGLE_LINK)
+		max_pixel_clock = enc10->base.features.max_hdmi_pixel_clock;
+
+	/* DVI only support RGB pixel encoding */
+	if (crtc_timing->pixel_encoding != PIXEL_ENCODING_RGB)
+		return false;
+
+	/*connect DVI via adpater's HDMI connector*/
+	if ((connector_signal == SIGNAL_TYPE_DVI_SINGLE_LINK ||
+		connector_signal == SIGNAL_TYPE_HDMI_TYPE_A) &&
+		signal != SIGNAL_TYPE_HDMI_TYPE_A &&
+		crtc_timing->pix_clk_100hz > (TMDS_MAX_PIXEL_CLOCK * 10))
+		return false;
+	if (crtc_timing->pix_clk_100hz < (TMDS_MIN_PIXEL_CLOCK * 10))
+		return false;
+
+	if (crtc_timing->pix_clk_100hz > (max_pixel_clock * 10))
+		return false;
+
+	/* DVI supports 6/8bpp single-link and 10/16bpp dual-link */
+	switch (crtc_timing->display_color_depth) {
+	case COLOR_DEPTH_666:
+	case COLOR_DEPTH_888:
+	break;
+	case COLOR_DEPTH_101010:
+	case COLOR_DEPTH_161616:
+		if (signal != SIGNAL_TYPE_DVI_DUAL_LINK)
+			return false;
+	break;
+	default:
+		return false;
+	}
+
+	return true;
+}
+
+static bool dcn10_link_encoder_validate_hdmi_output(
+	const struct dcn10_link_encoder *enc10,
+	const struct dc_crtc_timing *crtc_timing,
+	const struct dc_edid_caps *edid_caps,
+	int adjusted_pix_clk_100hz)
+{
+	enum dc_color_depth max_deep_color =
+			enc10->base.features.max_hdmi_deep_color;
+
+	// check pixel clock against edid specified max TMDS clk
+	if (edid_caps->max_tmds_clk_mhz != 0 &&
+			adjusted_pix_clk_100hz > edid_caps->max_tmds_clk_mhz * 10000)
+		return false;
+
+	if (max_deep_color < crtc_timing->display_color_depth)
+		return false;
+
+	if (crtc_timing->display_color_depth < COLOR_DEPTH_888)
+		return false;
+	if (adjusted_pix_clk_100hz < (TMDS_MIN_PIXEL_CLOCK * 10))
+		return false;
+
+	if ((adjusted_pix_clk_100hz == 0) ||
+		(adjusted_pix_clk_100hz > (enc10->base.features.max_hdmi_pixel_clock * 10)))
+		return false;
+
+	/* DCE11 HW does not support 420 */
+	if (!enc10->base.features.hdmi_ycbcr420_supported &&
+			crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR420)
+		return false;
+
+	if ((!enc10->base.features.flags.bits.HDMI_6GB_EN ||
+			enc10->base.ctx->dc->debug.hdmi20_disable) &&
+			adjusted_pix_clk_100hz >= 3000000)
+		return false;
+	if (enc10->base.ctx->dc->debug.hdmi20_disable &&
+		crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR420)
+		return false;
+	return true;
+}
+
+bool dcn10_link_encoder_validate_dp_output(
+	const struct dcn10_link_encoder *enc10,
+	const struct dc_crtc_timing *crtc_timing)
+{
+	if (crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR420) {
+		if (!enc10->base.features.dp_ycbcr420_supported)
+			return false;
+	}
+
+	return true;
+}
+
+void dcn10_link_encoder_construct(
+	struct dcn10_link_encoder *enc10,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	const struct dcn10_link_enc_aux_registers *aux_regs,
+	const struct dcn10_link_enc_hpd_registers *hpd_regs,
+	const struct dcn10_link_enc_shift *link_shift,
+	const struct dcn10_link_enc_mask *link_mask)
+{
+	struct bp_encoder_cap_info bp_cap_info = {0};
+	const struct dc_vbios_funcs *bp_funcs = init_data->ctx->dc_bios->funcs;
+	enum bp_result result = BP_RESULT_OK;
+
+	enc10->base.funcs = &dcn10_lnk_enc_funcs;
+	enc10->base.ctx = init_data->ctx;
+	enc10->base.id = init_data->encoder;
+
+	enc10->base.hpd_source = init_data->hpd_source;
+	enc10->base.connector = init_data->connector;
+
+	enc10->base.preferred_engine = ENGINE_ID_UNKNOWN;
+
+	enc10->base.features = *enc_features;
+
+	enc10->base.transmitter = init_data->transmitter;
+
+	/* set the flag to indicate whether driver poll the I2C data pin
+	 * while doing the DP sink detect
+	 */
+
+/*	if (dal_adapter_service_is_feature_supported(as,
+		FEATURE_DP_SINK_DETECT_POLL_DATA_PIN))
+		enc10->base.features.flags.bits.
+			DP_SINK_DETECT_POLL_DATA_PIN = true;*/
+
+	enc10->base.output_signals =
+		SIGNAL_TYPE_DVI_SINGLE_LINK |
+		SIGNAL_TYPE_DVI_DUAL_LINK |
+		SIGNAL_TYPE_LVDS |
+		SIGNAL_TYPE_DISPLAY_PORT |
+		SIGNAL_TYPE_DISPLAY_PORT_MST |
+		SIGNAL_TYPE_EDP |
+		SIGNAL_TYPE_HDMI_TYPE_A;
+
+	/* For DCE 8.0 and 8.1, by design, UNIPHY is hardwired to DIG_BE.
+	 * SW always assign DIG_FE 1:1 mapped to DIG_FE for non-MST UNIPHY.
+	 * SW assign DIG_FE to non-MST UNIPHY first and MST last. So prefer
+	 * DIG is per UNIPHY and used by SST DP, eDP, HDMI, DVI and LVDS.
+	 * Prefer DIG assignment is decided by board design.
+	 * For DCE 8.0, there are only max 6 UNIPHYs, we assume board design
+	 * and VBIOS will filter out 7 UNIPHY for DCE 8.0.
+	 * By this, adding DIGG should not hurt DCE 8.0.
+	 * This will let DCE 8.1 share DCE 8.0 as much as possible
+	 */
+
+	enc10->link_regs = link_regs;
+	enc10->aux_regs = aux_regs;
+	enc10->hpd_regs = hpd_regs;
+	enc10->link_shift = link_shift;
+	enc10->link_mask = link_mask;
+
+	switch (enc10->base.transmitter) {
+	case TRANSMITTER_UNIPHY_A:
+		enc10->base.preferred_engine = ENGINE_ID_DIGA;
+	break;
+	case TRANSMITTER_UNIPHY_B:
+		enc10->base.preferred_engine = ENGINE_ID_DIGB;
+	break;
+	case TRANSMITTER_UNIPHY_C:
+		enc10->base.preferred_engine = ENGINE_ID_DIGC;
+	break;
+	case TRANSMITTER_UNIPHY_D:
+		enc10->base.preferred_engine = ENGINE_ID_DIGD;
+	break;
+	case TRANSMITTER_UNIPHY_E:
+		enc10->base.preferred_engine = ENGINE_ID_DIGE;
+	break;
+	case TRANSMITTER_UNIPHY_F:
+		enc10->base.preferred_engine = ENGINE_ID_DIGF;
+	break;
+	case TRANSMITTER_UNIPHY_G:
+		enc10->base.preferred_engine = ENGINE_ID_DIGG;
+	break;
+	default:
+		ASSERT_CRITICAL(false);
+		enc10->base.preferred_engine = ENGINE_ID_UNKNOWN;
+	}
+
+	/* default to one to mirror Windows behavior */
+	enc10->base.features.flags.bits.HDMI_6GB_EN = 1;
+
+	result = bp_funcs->get_encoder_cap_info(enc10->base.ctx->dc_bios,
+						enc10->base.id, &bp_cap_info);
+
+	/* Override features with DCE-specific values */
+	if (result == BP_RESULT_OK) {
+		enc10->base.features.flags.bits.IS_HBR2_CAPABLE =
+				bp_cap_info.DP_HBR2_EN;
+		enc10->base.features.flags.bits.IS_HBR3_CAPABLE =
+				bp_cap_info.DP_HBR3_EN;
+		enc10->base.features.flags.bits.HDMI_6GB_EN = bp_cap_info.HDMI_6GB_EN;
+		enc10->base.features.flags.bits.DP_IS_USB_C =
+				bp_cap_info.DP_IS_USB_C;
+	} else {
+		DC_LOG_WARNING("%s: Failed to get encoder_cap_info from VBIOS with error code %d!\n",
+				__func__,
+				result);
+	}
+	if (enc10->base.ctx->dc->debug.hdmi20_disable) {
+		enc10->base.features.flags.bits.HDMI_6GB_EN = 0;
+	}
+}
+
+bool dcn10_link_encoder_validate_output_with_stream(
+	struct link_encoder *enc,
+	const struct dc_stream_state *stream)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	bool is_valid;
+
+	//if SCDC (340-600MHz) is disabled, set to HDMI 1.4 timing limit
+	if (stream->sink->edid_caps.panel_patch.skip_scdc_overwrite &&
+		enc10->base.features.max_hdmi_pixel_clock > 300000)
+		enc10->base.features.max_hdmi_pixel_clock = 300000;
+
+	switch (stream->signal) {
+	case SIGNAL_TYPE_DVI_SINGLE_LINK:
+	case SIGNAL_TYPE_DVI_DUAL_LINK:
+		is_valid = dcn10_link_encoder_validate_dvi_output(
+			enc10,
+			stream->link->connector_signal,
+			stream->signal,
+			&stream->timing);
+	break;
+	case SIGNAL_TYPE_HDMI_TYPE_A:
+		is_valid = dcn10_link_encoder_validate_hdmi_output(
+				enc10,
+				&stream->timing,
+				&stream->sink->edid_caps,
+				stream->phy_pix_clk * 10);
+	break;
+	case SIGNAL_TYPE_DISPLAY_PORT:
+	case SIGNAL_TYPE_DISPLAY_PORT_MST:
+		is_valid = dcn10_link_encoder_validate_dp_output(
+					enc10, &stream->timing);
+	break;
+	case SIGNAL_TYPE_EDP:
+		is_valid = stream->timing.pixel_encoding == PIXEL_ENCODING_RGB;
+	break;
+	case SIGNAL_TYPE_VIRTUAL:
+		is_valid = true;
+		break;
+	default:
+		is_valid = false;
+	break;
+	}
+
+	return is_valid;
+}
+
+void dcn10_link_encoder_hw_init(
+	struct link_encoder *enc)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	struct bp_transmitter_control cntl = { 0 };
+	enum bp_result result;
+
+	cntl.action = TRANSMITTER_CONTROL_INIT;
+	cntl.engine_id = ENGINE_ID_UNKNOWN;
+	cntl.transmitter = enc10->base.transmitter;
+	cntl.connector_obj_id = enc10->base.connector;
+	cntl.lanes_number = LANE_COUNT_FOUR;
+	cntl.coherent = false;
+	cntl.hpd_sel = enc10->base.hpd_source;
+
+	if (enc10->base.connector.id == CONNECTOR_ID_EDP)
+		cntl.signal = SIGNAL_TYPE_EDP;
+
+	result = link_transmitter_control(enc10, &cntl);
+
+	if (result != BP_RESULT_OK) {
+		DC_LOG_ERROR("%s: Failed to execute VBIOS command table!\n",
+			__func__);
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+	if (enc10->base.connector.id == CONNECTOR_ID_LVDS) {
+		cntl.action = TRANSMITTER_CONTROL_BACKLIGHT_BRIGHTNESS;
+
+		result = link_transmitter_control(enc10, &cntl);
+
+		ASSERT(result == BP_RESULT_OK);
+
+	}
+	dcn10_aux_initialize(enc10);
+
+	/* reinitialize HPD.
+	 * hpd_initialize() will pass DIG_FE id to HW context.
+	 * All other routine within HW context will use fe_engine_offset
+	 * as DIG_FE id even caller pass DIG_FE id.
+	 * So this routine must be called first.
+	 */
+	hpd_initialize(enc10);
+}
+
+void dcn10_link_encoder_destroy(struct link_encoder **enc)
+{
+	kfree(TO_DCN10_LINK_ENC(*enc));
+	*enc = NULL;
+}
+
+void dcn10_link_encoder_setup(
+	struct link_encoder *enc,
+	enum signal_type signal)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	switch (signal) {
+	case SIGNAL_TYPE_EDP:
+	case SIGNAL_TYPE_DISPLAY_PORT:
+		/* DP SST */
+		REG_UPDATE(DIG_BE_CNTL, DIG_MODE, 0);
+		break;
+	case SIGNAL_TYPE_LVDS:
+		/* LVDS */
+		REG_UPDATE(DIG_BE_CNTL, DIG_MODE, 1);
+		break;
+	case SIGNAL_TYPE_DVI_SINGLE_LINK:
+	case SIGNAL_TYPE_DVI_DUAL_LINK:
+		/* TMDS-DVI */
+		REG_UPDATE(DIG_BE_CNTL, DIG_MODE, 2);
+		break;
+	case SIGNAL_TYPE_HDMI_TYPE_A:
+		/* TMDS-HDMI */
+		REG_UPDATE(DIG_BE_CNTL, DIG_MODE, 3);
+		break;
+	case SIGNAL_TYPE_DISPLAY_PORT_MST:
+		/* DP MST */
+		REG_UPDATE(DIG_BE_CNTL, DIG_MODE, 5);
+		break;
+	default:
+		ASSERT_CRITICAL(false);
+		/* invalid mode ! */
+		break;
+	}
+
+}
+
+/* TODO: still need depth or just pass in adjusted pixel clock? */
+void dcn10_link_encoder_enable_tmds_output(
+	struct link_encoder *enc,
+	enum clock_source_id clock_source,
+	enum dc_color_depth color_depth,
+	enum signal_type signal,
+	uint32_t pixel_clock)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	struct bp_transmitter_control cntl = { 0 };
+	enum bp_result result;
+
+	/* Enable the PHY */
+
+	cntl.action = TRANSMITTER_CONTROL_ENABLE;
+	cntl.engine_id = enc->preferred_engine;
+	cntl.transmitter = enc10->base.transmitter;
+	cntl.pll_id = clock_source;
+	cntl.signal = signal;
+	if (cntl.signal == SIGNAL_TYPE_DVI_DUAL_LINK)
+		cntl.lanes_number = 8;
+	else
+		cntl.lanes_number = 4;
+
+	cntl.hpd_sel = enc10->base.hpd_source;
+
+	cntl.pixel_clock = pixel_clock;
+	cntl.color_depth = color_depth;
+
+	result = link_transmitter_control(enc10, &cntl);
+
+	if (result != BP_RESULT_OK) {
+		DC_LOG_ERROR("%s: Failed to execute VBIOS command table!\n",
+			__func__);
+		BREAK_TO_DEBUGGER();
+	}
+}
+
+void dcn10_link_encoder_enable_tmds_output_with_clk_pattern_wa(
+	struct link_encoder *enc,
+	enum clock_source_id clock_source,
+	enum dc_color_depth color_depth,
+	enum signal_type signal,
+	uint32_t pixel_clock)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	dcn10_link_encoder_enable_tmds_output(
+		enc, clock_source, color_depth, signal, pixel_clock);
+
+	REG_UPDATE(DIG_CLOCK_PATTERN, DIG_CLOCK_PATTERN, 0x1F);
+}
+
+/* enables DP PHY output */
+void dcn10_link_encoder_enable_dp_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	struct bp_transmitter_control cntl = { 0 };
+	enum bp_result result;
+
+	/* Enable the PHY */
+
+	/* number_of_lanes is used for pixel clock adjust,
+	 * but it's not passed to asic_control.
+	 * We need to set number of lanes manually.
+	 */
+	enc1_configure_encoder(enc10, link_settings);
+
+	cntl.action = TRANSMITTER_CONTROL_ENABLE;
+	cntl.engine_id = enc->preferred_engine;
+	cntl.transmitter = enc10->base.transmitter;
+	cntl.pll_id = clock_source;
+	cntl.signal = SIGNAL_TYPE_DISPLAY_PORT;
+	cntl.lanes_number = link_settings->lane_count;
+	cntl.hpd_sel = enc10->base.hpd_source;
+	cntl.pixel_clock = link_settings->link_rate
+						* LINK_RATE_REF_FREQ_IN_KHZ;
+	/* TODO: check if undefined works */
+	cntl.color_depth = COLOR_DEPTH_UNDEFINED;
+
+	result = link_transmitter_control(enc10, &cntl);
+
+	if (result != BP_RESULT_OK) {
+		DC_LOG_ERROR("%s: Failed to execute VBIOS command table!\n",
+			__func__);
+		BREAK_TO_DEBUGGER();
+	}
+}
+
+/* enables DP PHY output in MST mode */
+void dcn10_link_encoder_enable_dp_mst_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	struct bp_transmitter_control cntl = { 0 };
+	enum bp_result result;
+
+	/* Enable the PHY */
+
+	/* number_of_lanes is used for pixel clock adjust,
+	 * but it's not passed to asic_control.
+	 * We need to set number of lanes manually.
+	 */
+	enc1_configure_encoder(enc10, link_settings);
+
+	cntl.action = TRANSMITTER_CONTROL_ENABLE;
+	cntl.engine_id = ENGINE_ID_UNKNOWN;
+	cntl.transmitter = enc10->base.transmitter;
+	cntl.pll_id = clock_source;
+	cntl.signal = SIGNAL_TYPE_DISPLAY_PORT_MST;
+	cntl.lanes_number = link_settings->lane_count;
+	cntl.hpd_sel = enc10->base.hpd_source;
+	cntl.pixel_clock = link_settings->link_rate
+						* LINK_RATE_REF_FREQ_IN_KHZ;
+	/* TODO: check if undefined works */
+	cntl.color_depth = COLOR_DEPTH_UNDEFINED;
+
+	result = link_transmitter_control(enc10, &cntl);
+
+	if (result != BP_RESULT_OK) {
+		DC_LOG_ERROR("%s: Failed to execute VBIOS command table!\n",
+			__func__);
+		BREAK_TO_DEBUGGER();
+	}
+}
+/*
+ * @brief
+ * Disable transmitter and its encoder
+ */
+void dcn10_link_encoder_disable_output(
+	struct link_encoder *enc,
+	enum signal_type signal)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	struct bp_transmitter_control cntl = { 0 };
+	enum bp_result result;
+
+	if (enc->funcs->is_dig_enabled && !enc->funcs->is_dig_enabled(enc)) {
+		/* OF_SKIP_POWER_DOWN_INACTIVE_ENCODER */
+	/*in DP_Alt_No_Connect case, we turn off the dig already,
+	after excuation the PHY w/a sequence, not allow touch PHY any more*/
+		return;
+	}
+	/* Power-down RX and disable GPU PHY should be paired.
+	 * Disabling PHY without powering down RX may cause
+	 * symbol lock loss, on which we will get DP Sink interrupt.
+	 */
+
+	/* There is a case for the DP active dongles
+	 * where we want to disable the PHY but keep RX powered,
+	 * for those we need to ignore DP Sink interrupt
+	 * by checking lane count that has been set
+	 * on the last do_enable_output().
+	 */
+
+	/* disable transmitter */
+	cntl.action = TRANSMITTER_CONTROL_DISABLE;
+	cntl.transmitter = enc10->base.transmitter;
+	cntl.hpd_sel = enc10->base.hpd_source;
+	cntl.signal = signal;
+	cntl.connector_obj_id = enc10->base.connector;
+
+	result = link_transmitter_control(enc10, &cntl);
+
+	if (result != BP_RESULT_OK) {
+		DC_LOG_ERROR("%s: Failed to execute VBIOS command table!\n",
+			__func__);
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+	/* disable encoder */
+	if (dc_is_dp_signal(signal))
+		link_encoder_disable(enc10);
+}
+
+void dcn10_link_encoder_dp_set_lane_settings(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	const struct dc_lane_settings lane_settings[LANE_COUNT_DP_MAX])
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	union dpcd_training_lane_set training_lane_set = { { 0 } };
+	int32_t lane = 0;
+	struct bp_transmitter_control cntl = { 0 };
+	enum bp_result result;
+
+	if (!link_settings) {
+		BREAK_TO_DEBUGGER();
+		return;
+	}
+
+	cntl.action = TRANSMITTER_CONTROL_SET_VOLTAGE_AND_PREEMPASIS;
+	cntl.transmitter = enc10->base.transmitter;
+	cntl.connector_obj_id = enc10->base.connector;
+	cntl.lanes_number = link_settings->lane_count;
+	cntl.hpd_sel = enc10->base.hpd_source;
+	cntl.pixel_clock = link_settings->link_rate * LINK_RATE_REF_FREQ_IN_KHZ;
+
+	for (lane = 0; lane < link_settings->lane_count; lane++) {
+		/* translate lane settings */
+
+		training_lane_set.bits.VOLTAGE_SWING_SET =
+				lane_settings[lane].VOLTAGE_SWING;
+		training_lane_set.bits.PRE_EMPHASIS_SET =
+				lane_settings[lane].PRE_EMPHASIS;
+
+		/* post cursor 2 setting only applies to HBR2 link rate */
+		if (link_settings->link_rate == LINK_RATE_HIGH2) {
+			/* this is passed to VBIOS
+			 * to program post cursor 2 level
+			 */
+			training_lane_set.bits.POST_CURSOR2_SET =
+					lane_settings[lane].POST_CURSOR2;
+		}
+
+		cntl.lane_select = lane;
+		cntl.lane_settings = training_lane_set.raw;
+
+		/* call VBIOS table to set voltage swing and pre-emphasis */
+		result = link_transmitter_control(enc10, &cntl);
+
+		if (result != BP_RESULT_OK) {
+			DC_LOG_ERROR("%s: Failed to execute VBIOS command table!\n", __func__);
+			BREAK_TO_DEBUGGER();
+		}
+	}
+}
+
+/* set DP PHY test and training patterns */
+void dcn10_link_encoder_dp_set_phy_pattern(
+	struct link_encoder *enc,
+	const struct encoder_set_dp_phy_pattern_param *param)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	switch (param->dp_phy_pattern) {
+	case DP_TEST_PATTERN_TRAINING_PATTERN1:
+		dcn10_link_encoder_set_dp_phy_pattern_training_pattern(enc, 0);
+		break;
+	case DP_TEST_PATTERN_TRAINING_PATTERN2:
+		dcn10_link_encoder_set_dp_phy_pattern_training_pattern(enc, 1);
+		break;
+	case DP_TEST_PATTERN_TRAINING_PATTERN3:
+		dcn10_link_encoder_set_dp_phy_pattern_training_pattern(enc, 2);
+		break;
+	case DP_TEST_PATTERN_TRAINING_PATTERN4:
+		dcn10_link_encoder_set_dp_phy_pattern_training_pattern(enc, 3);
+		break;
+	case DP_TEST_PATTERN_D102:
+		set_dp_phy_pattern_d102(enc10);
+		break;
+	case DP_TEST_PATTERN_SYMBOL_ERROR:
+		set_dp_phy_pattern_symbol_error(enc10);
+		break;
+	case DP_TEST_PATTERN_PRBS7:
+		set_dp_phy_pattern_prbs7(enc10);
+		break;
+	case DP_TEST_PATTERN_80BIT_CUSTOM:
+		set_dp_phy_pattern_80bit_custom(
+			enc10, param->custom_pattern);
+		break;
+	case DP_TEST_PATTERN_CP2520_1:
+		set_dp_phy_pattern_hbr2_compliance_cp2520_2(enc10, 1);
+		break;
+	case DP_TEST_PATTERN_CP2520_2:
+		set_dp_phy_pattern_hbr2_compliance_cp2520_2(enc10, 2);
+		break;
+	case DP_TEST_PATTERN_CP2520_3:
+		set_dp_phy_pattern_hbr2_compliance_cp2520_2(enc10, 3);
+		break;
+	case DP_TEST_PATTERN_VIDEO_MODE: {
+		set_dp_phy_pattern_passthrough_mode(
+			enc10, param->dp_panel_mode);
+		break;
+	}
+
+	default:
+		/* invalid phy pattern */
+		ASSERT_CRITICAL(false);
+		break;
+	}
+}
+
+static void fill_stream_allocation_row_info(
+	const struct link_mst_stream_allocation *stream_allocation,
+	uint32_t *src,
+	uint32_t *slots)
+{
+	const struct stream_encoder *stream_enc = stream_allocation->stream_enc;
+
+	if (stream_enc) {
+		*src = stream_enc->id;
+		*slots = stream_allocation->slot_count;
+	} else {
+		*src = 0;
+		*slots = 0;
+	}
+}
+
+/* programs DP MST VC payload allocation */
+void dcn10_link_encoder_update_mst_stream_allocation_table(
+	struct link_encoder *enc,
+	const struct link_mst_stream_allocation_table *table)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	uint32_t value1 = 0;
+	uint32_t value2 = 0;
+	uint32_t slots = 0;
+	uint32_t src = 0;
+	uint32_t retries = 0;
+
+	/* For CZ, there are only 3 pipes. So Virtual channel is up 3.*/
+
+	/* --- Set MSE Stream Attribute -
+	 * Setup VC Payload Table on Tx Side,
+	 * Issue allocation change trigger
+	 * to commit payload on both tx and rx side
+	 */
+
+	/* we should clean-up table each time */
+
+	if (table->stream_count >= 1) {
+		fill_stream_allocation_row_info(
+			&table->stream_allocations[0],
+			&src,
+			&slots);
+	} else {
+		src = 0;
+		slots = 0;
+	}
+
+	REG_UPDATE_2(DP_MSE_SAT0,
+			DP_MSE_SAT_SRC0, src,
+			DP_MSE_SAT_SLOT_COUNT0, slots);
+
+	if (table->stream_count >= 2) {
+		fill_stream_allocation_row_info(
+			&table->stream_allocations[1],
+			&src,
+			&slots);
+	} else {
+		src = 0;
+		slots = 0;
+	}
+
+	REG_UPDATE_2(DP_MSE_SAT0,
+			DP_MSE_SAT_SRC1, src,
+			DP_MSE_SAT_SLOT_COUNT1, slots);
+
+	if (table->stream_count >= 3) {
+		fill_stream_allocation_row_info(
+			&table->stream_allocations[2],
+			&src,
+			&slots);
+	} else {
+		src = 0;
+		slots = 0;
+	}
+
+	REG_UPDATE_2(DP_MSE_SAT1,
+			DP_MSE_SAT_SRC2, src,
+			DP_MSE_SAT_SLOT_COUNT2, slots);
+
+	if (table->stream_count >= 4) {
+		fill_stream_allocation_row_info(
+			&table->stream_allocations[3],
+			&src,
+			&slots);
+	} else {
+		src = 0;
+		slots = 0;
+	}
+
+	REG_UPDATE_2(DP_MSE_SAT1,
+			DP_MSE_SAT_SRC3, src,
+			DP_MSE_SAT_SLOT_COUNT3, slots);
+
+	/* --- wait for transaction finish */
+
+	/* send allocation change trigger (ACT) ?
+	 * this step first sends the ACT,
+	 * then double buffers the SAT into the hardware
+	 * making the new allocation active on the DP MST mode link
+	 */
+
+	/* DP_MSE_SAT_UPDATE:
+	 * 0 - No Action
+	 * 1 - Update SAT with trigger
+	 * 2 - Update SAT without trigger
+	 */
+	REG_UPDATE(DP_MSE_SAT_UPDATE,
+			DP_MSE_SAT_UPDATE, 1);
+
+	/* wait for update to complete
+	 * (i.e. DP_MSE_SAT_UPDATE field is reset to 0)
+	 * then wait for the transmission
+	 * of at least 16 MTP headers on immediate local link.
+	 * i.e. DP_MSE_16_MTP_KEEPOUT field (read only) is reset to 0
+	 * a value of 1 indicates that DP MST mode
+	 * is in the 16 MTP keepout region after a VC has been added.
+	 * MST stream bandwidth (VC rate) can be configured
+	 * after this bit is cleared
+	 */
+	do {
+		udelay(10);
+
+		REG_READ(DP_MSE_SAT_UPDATE);
+
+		REG_GET(DP_MSE_SAT_UPDATE,
+				DP_MSE_SAT_UPDATE, &value1);
+
+		REG_GET(DP_MSE_SAT_UPDATE,
+				DP_MSE_16_MTP_KEEPOUT, &value2);
+
+		/* bit field DP_MSE_SAT_UPDATE is set to 1 already */
+		if (!value1 && !value2)
+			break;
+		++retries;
+	} while (retries < DP_MST_UPDATE_MAX_RETRY);
+}
+
+void dcn10_link_encoder_connect_dig_be_to_fe(
+	struct link_encoder *enc,
+	enum engine_id engine,
+	bool connect)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	uint32_t field;
+
+	if (engine != ENGINE_ID_UNKNOWN) {
+
+		REG_GET(DIG_BE_CNTL, DIG_FE_SOURCE_SELECT, &field);
+
+		if (connect)
+			field |= get_frontend_source(engine);
+		else
+			field &= ~get_frontend_source(engine);
+
+		REG_UPDATE(DIG_BE_CNTL, DIG_FE_SOURCE_SELECT, field);
+	}
+}
+
+
+#define HPD_REG(reg)\
+	(enc10->hpd_regs->reg)
+
+#define HPD_REG_READ(reg_name) \
+		dm_read_reg(CTX, HPD_REG(reg_name))
+
+#define HPD_REG_UPDATE_N(reg_name, n, ...)	\
+		generic_reg_update_ex(CTX, \
+				HPD_REG(reg_name), \
+				n, __VA_ARGS__)
+
+#define HPD_REG_UPDATE(reg_name, field, val)	\
+		HPD_REG_UPDATE_N(reg_name, 1, \
+				FN(reg_name, field), val)
+
+void dcn10_link_encoder_enable_hpd(struct link_encoder *enc)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	HPD_REG_UPDATE(DC_HPD_CONTROL,
+			DC_HPD_EN, 1);
+}
+
+void dcn10_link_encoder_disable_hpd(struct link_encoder *enc)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	HPD_REG_UPDATE(DC_HPD_CONTROL,
+			DC_HPD_EN, 0);
+}
+
+#define AUX_REG(reg)\
+	(enc10->aux_regs->reg)
+
+#define AUX_REG_READ(reg_name) \
+		dm_read_reg(CTX, AUX_REG(reg_name))
+
+#define AUX_REG_UPDATE_N(reg_name, n, ...)	\
+		generic_reg_update_ex(CTX, \
+				AUX_REG(reg_name), \
+				n, __VA_ARGS__)
+
+#define AUX_REG_UPDATE(reg_name, field, val)	\
+		AUX_REG_UPDATE_N(reg_name, 1, \
+				FN(reg_name, field), val)
+
+#define AUX_REG_UPDATE_2(reg, f1, v1, f2, v2)	\
+		AUX_REG_UPDATE_N(reg, 2,\
+				FN(reg, f1), v1,\
+				FN(reg, f2), v2)
+
+void dcn10_aux_initialize(struct dcn10_link_encoder *enc10)
+{
+	enum hpd_source_id hpd_source = enc10->base.hpd_source;
+
+	AUX_REG_UPDATE_2(AUX_CONTROL,
+			AUX_HPD_SEL, hpd_source,
+			AUX_LS_READ_EN, 0);
+
+	/* 1/4 window (the maximum allowed) */
+	AUX_REG_UPDATE(AUX_DPHY_RX_CONTROL0,
+			AUX_RX_RECEIVE_WINDOW, 0);
+}
+
+enum signal_type dcn10_get_dig_mode(
+	struct link_encoder *enc)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	uint32_t value;
+	REG_GET(DIG_BE_CNTL, DIG_MODE, &value);
+	switch (value) {
+	case 1:
+		return SIGNAL_TYPE_DISPLAY_PORT;
+	case 2:
+		return SIGNAL_TYPE_DVI_SINGLE_LINK;
+	case 3:
+		return SIGNAL_TYPE_HDMI_TYPE_A;
+	case 5:
+		return SIGNAL_TYPE_DISPLAY_PORT_MST;
+	default:
+		return SIGNAL_TYPE_NONE;
+	}
+}
+
+void dcn10_link_encoder_get_max_link_cap(struct link_encoder *enc,
+	struct dc_link_settings *link_settings)
+{
+	/* Set Default link settings */
+	struct dc_link_settings max_link_cap = {LANE_COUNT_FOUR, LINK_RATE_HIGH,
+			LINK_SPREAD_05_DOWNSPREAD_30KHZ, false, 0};
+
+	/* Higher link settings based on feature supported */
+	if (enc->features.flags.bits.IS_HBR2_CAPABLE)
+		max_link_cap.link_rate = LINK_RATE_HIGH2;
+
+	if (enc->features.flags.bits.IS_HBR3_CAPABLE)
+		max_link_cap.link_rate = LINK_RATE_HIGH3;
+
+	if (enc->features.flags.bits.IS_UHBR10_CAPABLE)
+		max_link_cap.link_rate = LINK_RATE_UHBR10;
+
+	if (enc->features.flags.bits.IS_UHBR13_5_CAPABLE)
+		max_link_cap.link_rate = LINK_RATE_UHBR13_5;
+
+	if (enc->features.flags.bits.IS_UHBR20_CAPABLE)
+		max_link_cap.link_rate = LINK_RATE_UHBR20;
+
+	*link_settings = max_link_cap;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn10/dcn10_link_encoder.h b/drivers/gpu/drm/amd/display/dc/dio/dcn10/dcn10_link_encoder.h
new file mode 100644
index 000000000000..b7a89c39f445
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn10/dcn10_link_encoder.h
@@ -0,0 +1,659 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_LINK_ENCODER__DCN10_H__
+#define __DC_LINK_ENCODER__DCN10_H__
+
+#include "link_encoder.h"
+
+#define TO_DCN10_LINK_ENC(link_encoder)\
+	container_of(link_encoder, struct dcn10_link_encoder, base)
+
+#define AUX_REG_LIST(id)\
+	SRI(AUX_CONTROL, DP_AUX, id), \
+	SRI(AUX_DPHY_RX_CONTROL0, DP_AUX, id), \
+	SRI(AUX_DPHY_RX_CONTROL1, DP_AUX, id)
+
+#define HPD_REG_LIST(id)\
+	SRI(DC_HPD_CONTROL, HPD, id)
+
+#define LE_DCN_COMMON_REG_LIST(id) \
+	SRI(DIG_BE_CNTL, DIG, id), \
+	SRI(DIG_BE_EN_CNTL, DIG, id), \
+	SRI(DIG_CLOCK_PATTERN, DIG, id), \
+	SRI(TMDS_CTL_BITS, DIG, id), \
+	SRI(DP_CONFIG, DP, id), \
+	SRI(DP_DPHY_CNTL, DP, id), \
+	SRI(DP_DPHY_PRBS_CNTL, DP, id), \
+	SRI(DP_DPHY_SCRAM_CNTL, DP, id),\
+	SRI(DP_DPHY_SYM0, DP, id), \
+	SRI(DP_DPHY_SYM1, DP, id), \
+	SRI(DP_DPHY_SYM2, DP, id), \
+	SRI(DP_DPHY_TRAINING_PATTERN_SEL, DP, id), \
+	SRI(DP_LINK_CNTL, DP, id), \
+	SRI(DP_LINK_FRAMING_CNTL, DP, id), \
+	SRI(DP_MSE_SAT0, DP, id), \
+	SRI(DP_MSE_SAT1, DP, id), \
+	SRI(DP_MSE_SAT2, DP, id), \
+	SRI(DP_MSE_SAT_UPDATE, DP, id), \
+	SRI(DP_SEC_CNTL, DP, id), \
+	SRI(DP_VID_STREAM_CNTL, DP, id), \
+	SRI(DP_DPHY_FAST_TRAINING, DP, id), \
+	SRI(DP_SEC_CNTL1, DP, id), \
+	SRI(DP_DPHY_BS_SR_SWAP_CNTL, DP, id), \
+	SRI(DP_DPHY_HBR2_PATTERN_CONTROL, DP, id)
+
+
+#define LE_DCN10_REG_LIST(id)\
+	SRI(DP_DPHY_INTERNAL_CTRL, DP, id), \
+	LE_DCN_COMMON_REG_LIST(id)
+
+struct dcn10_link_enc_aux_registers {
+	uint32_t AUX_CONTROL;
+	uint32_t AUX_DPHY_RX_CONTROL0;
+	uint32_t AUX_DPHY_TX_CONTROL;
+	uint32_t AUX_DPHY_RX_CONTROL1;
+};
+
+struct dcn10_link_enc_hpd_registers {
+	uint32_t DC_HPD_CONTROL;
+};
+
+struct dcn10_link_enc_registers {
+	uint32_t DIG_BE_CNTL;
+	uint32_t DIG_BE_EN_CNTL;
+	uint32_t DIG_CLOCK_PATTERN;
+	uint32_t DP_CONFIG;
+	uint32_t DP_DPHY_CNTL;
+	uint32_t DP_DPHY_INTERNAL_CTRL;
+	uint32_t DP_DPHY_PRBS_CNTL;
+	uint32_t DP_DPHY_SCRAM_CNTL;
+	uint32_t DP_DPHY_SYM0;
+	uint32_t DP_DPHY_SYM1;
+	uint32_t DP_DPHY_SYM2;
+	uint32_t DP_DPHY_TRAINING_PATTERN_SEL;
+	uint32_t DP_LINK_CNTL;
+	uint32_t DP_LINK_FRAMING_CNTL;
+	uint32_t DP_MSE_SAT0;
+	uint32_t DP_MSE_SAT1;
+	uint32_t DP_MSE_SAT2;
+	uint32_t DP_MSE_SAT_UPDATE;
+	uint32_t DP_SEC_CNTL;
+	uint32_t DP_VID_STREAM_CNTL;
+	uint32_t DP_DPHY_FAST_TRAINING;
+	uint32_t DP_DPHY_BS_SR_SWAP_CNTL;
+	uint32_t DP_DPHY_HBR2_PATTERN_CONTROL;
+	uint32_t DP_SEC_CNTL1;
+	uint32_t TMDS_CTL_BITS;
+	/* DCCG  */
+	uint32_t CLOCK_ENABLE;
+	/* DIG */
+	uint32_t DIG_LANE_ENABLE;
+	/* UNIPHY */
+	uint32_t CHANNEL_XBAR_CNTL;
+	/* DPCS */
+	uint32_t RDPCSTX_PHY_CNTL3;
+	uint32_t RDPCSTX_PHY_CNTL4;
+	uint32_t RDPCSTX_PHY_CNTL5;
+	uint32_t RDPCSTX_PHY_CNTL6;
+	uint32_t RDPCSPIPE_PHY_CNTL6;
+	uint32_t RDPCSTX_PHY_CNTL7;
+	uint32_t RDPCSTX_PHY_CNTL8;
+	uint32_t RDPCSTX_PHY_CNTL9;
+	uint32_t RDPCSTX_PHY_CNTL10;
+	uint32_t RDPCSTX_PHY_CNTL11;
+	uint32_t RDPCSTX_PHY_CNTL12;
+	uint32_t RDPCSTX_PHY_CNTL13;
+	uint32_t RDPCSTX_PHY_CNTL14;
+	uint32_t RDPCSTX_PHY_CNTL15;
+	uint32_t RDPCSTX_CNTL;
+	uint32_t RDPCSTX_CLOCK_CNTL;
+	uint32_t RDPCSTX_PHY_CNTL0;
+	uint32_t RDPCSTX_PHY_CNTL2;
+	uint32_t RDPCSTX_PLL_UPDATE_DATA;
+	uint32_t RDPCS_TX_CR_ADDR;
+	uint32_t RDPCS_TX_CR_DATA;
+	uint32_t DPCSTX_TX_CLOCK_CNTL;
+	uint32_t DPCSTX_TX_CNTL;
+	uint32_t RDPCSTX_INTERRUPT_CONTROL;
+	uint32_t RDPCSTX_PHY_FUSE0;
+	uint32_t RDPCSTX_PHY_FUSE1;
+	uint32_t RDPCSTX_PHY_FUSE2;
+	uint32_t RDPCSTX_PHY_FUSE3;
+	uint32_t RDPCSTX_PHY_RX_LD_VAL;
+	uint32_t DPCSTX_DEBUG_CONFIG;
+	uint32_t RDPCSTX_DEBUG_CONFIG;
+	uint32_t RDPCSTX0_RDPCSTX_SCRATCH;
+	uint32_t RDPCSTX_DMCU_DPALT_DIS_BLOCK_REG;
+	uint32_t DCIO_SOFT_RESET;
+	/* indirect registers */
+	uint32_t RAWLANE0_DIG_PCS_XF_RX_OVRD_IN_2;
+	uint32_t RAWLANE0_DIG_PCS_XF_RX_OVRD_IN_3;
+	uint32_t RAWLANE1_DIG_PCS_XF_RX_OVRD_IN_2;
+	uint32_t RAWLANE1_DIG_PCS_XF_RX_OVRD_IN_3;
+	uint32_t RAWLANE2_DIG_PCS_XF_RX_OVRD_IN_2;
+	uint32_t RAWLANE2_DIG_PCS_XF_RX_OVRD_IN_3;
+	uint32_t RAWLANE3_DIG_PCS_XF_RX_OVRD_IN_2;
+	uint32_t RAWLANE3_DIG_PCS_XF_RX_OVRD_IN_3;
+	uint32_t TMDS_DCBALANCER_CONTROL;
+	uint32_t PHYA_LINK_CNTL2;
+	uint32_t PHYB_LINK_CNTL2;
+	uint32_t PHYC_LINK_CNTL2;
+	uint32_t DIO_LINKA_CNTL;
+	uint32_t DIO_LINKB_CNTL;
+	uint32_t DIO_LINKC_CNTL;
+	uint32_t DIO_LINKD_CNTL;
+	uint32_t DIO_LINKE_CNTL;
+	uint32_t DIO_LINKF_CNTL;
+	uint32_t DIO_CLK_CNTL;
+	uint32_t DIG_BE_CLK_CNTL;
+};
+
+#define LE_SF(reg_name, field_name, post_fix)\
+	.field_name = reg_name ## __ ## field_name ## post_fix
+
+#define LINK_ENCODER_MASK_SH_LIST_DCN10(mask_sh)\
+	LE_SF(DIG0_DIG_BE_EN_CNTL, DIG_ENABLE, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CNTL, DIG_HPD_SELECT, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CNTL, DIG_MODE, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CNTL, DIG_FE_SOURCE_SELECT, mask_sh),\
+	LE_SF(DIG0_DIG_CLOCK_PATTERN, DIG_CLOCK_PATTERN, mask_sh),\
+	LE_SF(DIG0_TMDS_CTL_BITS, TMDS_CTL0, mask_sh), \
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_BYPASS, mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_ATEST_SEL_LANE0, mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_ATEST_SEL_LANE1, mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_ATEST_SEL_LANE2, mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_ATEST_SEL_LANE3, mask_sh),\
+	LE_SF(DP0_DP_DPHY_PRBS_CNTL, DPHY_PRBS_EN, mask_sh),\
+	LE_SF(DP0_DP_DPHY_PRBS_CNTL, DPHY_PRBS_SEL, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM0, DPHY_SYM1, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM0, DPHY_SYM2, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM0, DPHY_SYM3, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM1, DPHY_SYM4, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM1, DPHY_SYM5, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM1, DPHY_SYM6, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM2, DPHY_SYM7, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM2, DPHY_SYM8, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SCRAM_CNTL, DPHY_SCRAMBLER_BS_COUNT, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SCRAM_CNTL, DPHY_SCRAMBLER_ADVANCE, mask_sh),\
+	LE_SF(DP0_DP_DPHY_FAST_TRAINING, DPHY_RX_FAST_TRAINING_CAPABLE, mask_sh),\
+	LE_SF(DP0_DP_DPHY_BS_SR_SWAP_CNTL, DPHY_LOAD_BS_COUNT, mask_sh),\
+	LE_SF(DP0_DP_DPHY_TRAINING_PATTERN_SEL, DPHY_TRAINING_PATTERN_SEL, mask_sh),\
+	LE_SF(DP0_DP_DPHY_HBR2_PATTERN_CONTROL, DP_DPHY_HBR2_PATTERN_CONTROL, mask_sh),\
+	LE_SF(DP0_DP_LINK_CNTL, DP_LINK_TRAINING_COMPLETE, mask_sh),\
+	LE_SF(DP0_DP_LINK_FRAMING_CNTL, DP_IDLE_BS_INTERVAL, mask_sh),\
+	LE_SF(DP0_DP_LINK_FRAMING_CNTL, DP_VBID_DISABLE, mask_sh),\
+	LE_SF(DP0_DP_LINK_FRAMING_CNTL, DP_VID_ENHANCED_FRAME_MODE, mask_sh),\
+	LE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, mask_sh),\
+	LE_SF(DP0_DP_CONFIG, DP_UDI_LANES, mask_sh),\
+	LE_SF(DP0_DP_SEC_CNTL1, DP_SEC_GSP0_LINE_NUM, mask_sh),\
+	LE_SF(DP0_DP_SEC_CNTL1, DP_SEC_GSP0_PRIORITY, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT0, DP_MSE_SAT_SRC0, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT0, DP_MSE_SAT_SRC1, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT0, DP_MSE_SAT_SLOT_COUNT0, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT0, DP_MSE_SAT_SLOT_COUNT1, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT1, DP_MSE_SAT_SRC2, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT1, DP_MSE_SAT_SRC3, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT1, DP_MSE_SAT_SLOT_COUNT2, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT1, DP_MSE_SAT_SLOT_COUNT3, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT_UPDATE, DP_MSE_SAT_UPDATE, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT_UPDATE, DP_MSE_16_MTP_KEEPOUT, mask_sh),\
+	LE_SF(DP_AUX0_AUX_CONTROL, AUX_HPD_SEL, mask_sh),\
+	LE_SF(DP_AUX0_AUX_CONTROL, AUX_LS_READ_EN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_RECEIVE_WINDOW, mask_sh),\
+	LE_SF(HPD0_DC_HPD_CONTROL, DC_HPD_EN, mask_sh)
+
+#define DCN_LINK_ENCODER_REG_FIELD_LIST(type) \
+	type DIG_ENABLE;\
+	type DIG_HPD_SELECT;\
+	type DIG_MODE;\
+	type DIG_FE_SOURCE_SELECT;\
+	type DIG_CLOCK_PATTERN;\
+	type DPHY_BYPASS;\
+	type DPHY_ATEST_SEL_LANE0;\
+	type DPHY_ATEST_SEL_LANE1;\
+	type DPHY_ATEST_SEL_LANE2;\
+	type DPHY_ATEST_SEL_LANE3;\
+	type DPHY_PRBS_EN;\
+	type DPHY_PRBS_SEL;\
+	type DPHY_SYM1;\
+	type DPHY_SYM2;\
+	type DPHY_SYM3;\
+	type DPHY_SYM4;\
+	type DPHY_SYM5;\
+	type DPHY_SYM6;\
+	type DPHY_SYM7;\
+	type DPHY_SYM8;\
+	type DPHY_SCRAMBLER_BS_COUNT;\
+	type DPHY_SCRAMBLER_ADVANCE;\
+	type DPHY_RX_FAST_TRAINING_CAPABLE;\
+	type DPHY_LOAD_BS_COUNT;\
+	type DPHY_TRAINING_PATTERN_SEL;\
+	type DP_DPHY_HBR2_PATTERN_CONTROL;\
+	type DP_LINK_TRAINING_COMPLETE;\
+	type DP_IDLE_BS_INTERVAL;\
+	type DP_VBID_DISABLE;\
+	type DP_VID_ENHANCED_FRAME_MODE;\
+	type DP_VID_STREAM_ENABLE;\
+	type DP_UDI_LANES;\
+	type DP_SEC_GSP0_LINE_NUM;\
+	type DP_SEC_GSP0_PRIORITY;\
+	type DP_MSE_SAT_SRC0;\
+	type DP_MSE_SAT_SRC1;\
+	type DP_MSE_SAT_SRC2;\
+	type DP_MSE_SAT_SRC3;\
+	type DP_MSE_SAT_SLOT_COUNT0;\
+	type DP_MSE_SAT_SLOT_COUNT1;\
+	type DP_MSE_SAT_SLOT_COUNT2;\
+	type DP_MSE_SAT_SLOT_COUNT3;\
+	type DP_MSE_SAT_UPDATE;\
+	type DP_MSE_16_MTP_KEEPOUT;\
+	type DC_HPD_EN;\
+	type TMDS_CTL0;\
+	type AUX_HPD_SEL;\
+	type AUX_LS_READ_EN;\
+	type AUX_RX_RECEIVE_WINDOW
+
+
+#define DCN20_LINK_ENCODER_DPCS_REG_FIELD_LIST(type) \
+		type RDPCS_PHY_DP_TX0_DATA_EN;\
+		type RDPCS_PHY_DP_TX1_DATA_EN;\
+		type RDPCS_PHY_DP_TX2_DATA_EN;\
+		type RDPCS_PHY_DP_TX3_DATA_EN;\
+		type RDPCS_PHY_DP_TX0_PSTATE;\
+		type RDPCS_PHY_DP_TX1_PSTATE;\
+		type RDPCS_PHY_DP_TX2_PSTATE;\
+		type RDPCS_PHY_DP_TX3_PSTATE;\
+		type RDPCS_PHY_DP_TX0_MPLL_EN;\
+		type RDPCS_PHY_DP_TX1_MPLL_EN;\
+		type RDPCS_PHY_DP_TX2_MPLL_EN;\
+		type RDPCS_PHY_DP_TX3_MPLL_EN;\
+		type RDPCS_TX_FIFO_LANE0_EN;\
+		type RDPCS_TX_FIFO_LANE1_EN;\
+		type RDPCS_TX_FIFO_LANE2_EN;\
+		type RDPCS_TX_FIFO_LANE3_EN;\
+		type RDPCS_EXT_REFCLK_EN;\
+		type RDPCS_TX_FIFO_EN;\
+		type UNIPHY_LINK_ENABLE;\
+		type UNIPHY_CHANNEL0_XBAR_SOURCE;\
+		type UNIPHY_CHANNEL1_XBAR_SOURCE;\
+		type UNIPHY_CHANNEL2_XBAR_SOURCE;\
+		type UNIPHY_CHANNEL3_XBAR_SOURCE;\
+		type UNIPHY_CHANNEL0_INVERT;\
+		type UNIPHY_CHANNEL1_INVERT;\
+		type UNIPHY_CHANNEL2_INVERT;\
+		type UNIPHY_CHANNEL3_INVERT;\
+		type UNIPHY_LINK_ENABLE_HPD_MASK;\
+		type UNIPHY_LANE_STAGGER_DELAY;\
+		type RDPCS_SRAMCLK_BYPASS;\
+		type RDPCS_SRAMCLK_EN;\
+		type RDPCS_SRAMCLK_CLOCK_ON;\
+		type DPCS_TX_FIFO_EN;\
+		type RDPCS_PHY_DP_TX0_DISABLE;\
+		type RDPCS_PHY_DP_TX1_DISABLE;\
+		type RDPCS_PHY_DP_TX2_DISABLE;\
+		type RDPCS_PHY_DP_TX3_DISABLE;\
+		type RDPCS_PHY_DP_TX0_CLK_RDY;\
+		type RDPCS_PHY_DP_TX1_CLK_RDY;\
+		type RDPCS_PHY_DP_TX2_CLK_RDY;\
+		type RDPCS_PHY_DP_TX3_CLK_RDY;\
+		type RDPCS_PHY_DP_TX0_REQ;\
+		type RDPCS_PHY_DP_TX1_REQ;\
+		type RDPCS_PHY_DP_TX2_REQ;\
+		type RDPCS_PHY_DP_TX3_REQ;\
+		type RDPCS_PHY_DP_TX0_ACK;\
+		type RDPCS_PHY_DP_TX1_ACK;\
+		type RDPCS_PHY_DP_TX2_ACK;\
+		type RDPCS_PHY_DP_TX3_ACK;\
+		type RDPCS_PHY_DP_TX0_RESET;\
+		type RDPCS_PHY_DP_TX1_RESET;\
+		type RDPCS_PHY_DP_TX2_RESET;\
+		type RDPCS_PHY_DP_TX3_RESET;\
+		type RDPCS_PHY_RESET;\
+		type RDPCS_PHY_CR_MUX_SEL;\
+		type RDPCS_PHY_REF_RANGE;\
+		type RDPCS_PHY_DP4_POR;\
+		type RDPCS_SRAM_BYPASS;\
+		type RDPCS_SRAM_EXT_LD_DONE;\
+		type RDPCS_PHY_DP_TX0_TERM_CTRL;\
+		type RDPCS_PHY_DP_TX1_TERM_CTRL;\
+		type RDPCS_PHY_DP_TX2_TERM_CTRL;\
+		type RDPCS_PHY_DP_TX3_TERM_CTRL;\
+		type RDPCS_PHY_DP_REF_CLK_MPLLB_DIV;\
+		type RDPCS_PHY_DP_MPLLB_MULTIPLIER;\
+		type RDPCS_PHY_DP_MPLLB_SSC_EN;\
+		type RDPCS_PHY_DP_MPLLB_DIV5_CLK_EN;\
+		type RDPCS_PHY_DP_MPLLB_TX_CLK_DIV;\
+		type RDPCS_PHY_DP_MPLLB_WORD_DIV2_EN;\
+		type RDPCS_PHY_DP_MPLLB_FRACN_EN;\
+		type RDPCS_PHY_DP_MPLLB_PMIX_EN;\
+		type RDPCS_PHY_DP_MPLLB_FRACN_QUOT;\
+		type RDPCS_PHY_DP_MPLLB_FRACN_DEN;\
+		type RDPCS_PHY_DP_MPLLB_FRACN_REM;\
+		type RDPCS_PHY_DP_MPLLB_SSC_UP_SPREAD;\
+		type RDPCS_PHY_DP_MPLLB_SSC_STEPSIZE;\
+		type RDPCS_PHY_DP_MPLLB_SSC_PEAK;\
+		type RDPCS_PHY_DP_MPLLB_DIV_CLK_EN;\
+		type RDPCS_PHY_DP_MPLLB_DIV_MULTIPLIER;\
+		type RDPCS_PHY_TX_VBOOST_LVL;\
+		type RDPCS_PHY_HDMIMODE_ENABLE;\
+		type RDPCS_PHY_DP_REF_CLK_EN;\
+		type RDPCS_PLL_UPDATE_DATA;\
+		type RDPCS_SRAM_INIT_DONE;\
+		type RDPCS_TX_CR_ADDR;\
+		type RDPCS_TX_CR_DATA;\
+		type RDPCS_PHY_HDMI_MPLLB_HDMI_DIV;\
+		type RDPCS_PHY_DP_MPLLB_STATE;\
+		type RDPCS_PHY_DP_TX0_WIDTH;\
+		type RDPCS_PHY_DP_TX0_RATE;\
+		type RDPCS_PHY_DP_TX1_WIDTH;\
+		type RDPCS_PHY_DP_TX1_RATE;\
+		type RDPCS_PHY_DP_TX2_WIDTH;\
+		type RDPCS_PHY_DP_TX2_RATE;\
+		type RDPCS_PHY_DP_TX3_WIDTH;\
+		type RDPCS_PHY_DP_TX3_RATE;\
+		type DPCS_SYMCLK_CLOCK_ON;\
+		type DPCS_SYMCLK_GATE_DIS;\
+		type DPCS_SYMCLK_EN;\
+		type RDPCS_SYMCLK_DIV2_CLOCK_ON;\
+		type RDPCS_SYMCLK_DIV2_GATE_DIS;\
+		type RDPCS_SYMCLK_DIV2_EN;\
+		type DPCS_TX_DATA_SWAP;\
+		type DPCS_TX_DATA_ORDER_INVERT;\
+		type DPCS_TX_FIFO_RD_START_DELAY;\
+		type RDPCS_TX_FIFO_RD_START_DELAY;\
+		type RDPCS_REG_FIFO_ERROR_MASK;\
+		type RDPCS_TX_FIFO_ERROR_MASK;\
+		type RDPCS_DPALT_DISABLE_TOGGLE_MASK;\
+		type RDPCS_DPALT_4LANE_TOGGLE_MASK;\
+		type RDPCS_PHY_DPALT_DP4;\
+		type RDPCS_PHY_DPALT_DISABLE;\
+		type RDPCS_PHY_DPALT_DISABLE_ACK;\
+		type RDPCS_PHY_DP_MPLLB_V2I;\
+		type RDPCS_PHY_DP_MPLLB_FREQ_VCO;\
+		type RDPCS_PHY_DP_MPLLB_CP_INT_GS;\
+		type RDPCS_PHY_RX_VREF_CTRL;\
+		type RDPCS_PHY_DP_MPLLB_CP_INT;\
+		type RDPCS_PHY_DP_MPLLB_CP_PROP;\
+		type RDPCS_PHY_RX_REF_LD_VAL;\
+		type RDPCS_PHY_RX_VCO_LD_VAL;\
+		type DPCSTX_DEBUG_CONFIG; \
+		type RDPCSTX_DEBUG_CONFIG; \
+		type RDPCS_PHY_DP_TX0_EQ_MAIN;\
+		type RDPCS_PHY_DP_TX0_EQ_PRE;\
+		type RDPCS_PHY_DP_TX0_EQ_POST;\
+		type RDPCS_PHY_DP_TX1_EQ_MAIN;\
+		type RDPCS_PHY_DP_TX1_EQ_PRE;\
+		type RDPCS_PHY_DP_TX1_EQ_POST;\
+		type RDPCS_PHY_DP_TX2_EQ_MAIN;\
+		type RDPCS_PHY_DP_MPLLB_CP_PROP_GS;\
+		type RDPCS_PHY_DP_TX2_EQ_PRE;\
+		type RDPCS_PHY_DP_TX2_EQ_POST;\
+		type RDPCS_PHY_DP_TX3_EQ_MAIN;\
+		type RDPCS_PHY_DCO_RANGE;\
+		type RDPCS_PHY_DCO_FINETUNE;\
+		type RDPCS_PHY_DP_TX3_EQ_PRE;\
+		type RDPCS_PHY_DP_TX3_EQ_POST;\
+		type RDPCS_PHY_SUP_PRE_HP;\
+		type RDPCS_PHY_DP_TX0_VREGDRV_BYP;\
+		type RDPCS_PHY_DP_TX1_VREGDRV_BYP;\
+		type RDPCS_PHY_DP_TX2_VREGDRV_BYP;\
+		type RDPCS_PHY_DP_TX3_VREGDRV_BYP;\
+		type RDPCS_DMCU_DPALT_DIS_BLOCK_REG;\
+		type UNIPHYA_SOFT_RESET;\
+		type UNIPHYB_SOFT_RESET;\
+		type UNIPHYC_SOFT_RESET;\
+		type UNIPHYD_SOFT_RESET;\
+		type UNIPHYE_SOFT_RESET;\
+		type UNIPHYF_SOFT_RESET
+
+#define DCN20_LINK_ENCODER_REG_FIELD_LIST(type) \
+	type DIG_LANE0EN;\
+	type DIG_LANE1EN;\
+	type DIG_LANE2EN;\
+	type DIG_LANE3EN;\
+	type DIG_CLK_EN;\
+	type SYMCLKA_CLOCK_ENABLE;\
+	type DPHY_FEC_EN;\
+	type DPHY_FEC_READY_SHADOW;\
+	type DPHY_FEC_ACTIVE_STATUS;\
+	DCN20_LINK_ENCODER_DPCS_REG_FIELD_LIST(type);\
+	type VCO_LD_VAL_OVRD;\
+	type VCO_LD_VAL_OVRD_EN;\
+	type REF_LD_VAL_OVRD;\
+	type REF_LD_VAL_OVRD_EN;\
+	type AUX_RX_START_WINDOW; \
+	type AUX_RX_HALF_SYM_DETECT_LEN; \
+	type AUX_RX_TRANSITION_FILTER_EN; \
+	type AUX_RX_ALLOW_BELOW_THRESHOLD_PHASE_DETECT; \
+	type AUX_RX_ALLOW_BELOW_THRESHOLD_START; \
+	type AUX_RX_ALLOW_BELOW_THRESHOLD_STOP; \
+	type AUX_RX_PHASE_DETECT_LEN; \
+	type AUX_RX_DETECTION_THRESHOLD; \
+	type AUX_TX_PRECHARGE_LEN; \
+	type AUX_TX_PRECHARGE_SYMBOLS; \
+	type AUX_MODE_DET_CHECK_DELAY;\
+	type DPCS_DBG_CBUS_DIS;\
+	type AUX_RX_PRECHARGE_SKIP;\
+	type AUX_RX_TIMEOUT_LEN;\
+	type AUX_RX_TIMEOUT_LEN_MUL
+
+#define DCN30_LINK_ENCODER_REG_FIELD_LIST(type) \
+	type TMDS_SYNC_DCBAL_EN;\
+	type PHY_HPO_DIG_SRC_SEL;\
+	type PHY_HPO_ENC_SRC_SEL;\
+	type DPCS_TX_HDMI_FRL_MODE;\
+	type DPCS_TX_DATA_SWAP_10_BIT;\
+	type DPCS_TX_DATA_ORDER_INVERT_18_BIT;\
+	type RDPCS_TX_CLK_EN
+
+#define DCN31_LINK_ENCODER_REG_FIELD_LIST(type) \
+	type ENC_TYPE_SEL;\
+	type HPO_DP_ENC_SEL;\
+	type HPO_HDMI_ENC_SEL
+
+#define DCN35_LINK_ENCODER_REG_FIELD_LIST(type) \
+	type DIG_BE_ENABLE;\
+	type DIG_RB_SWITCH_EN;\
+	type DIG_BE_MODE;\
+	type DIG_BE_CLK_EN;\
+	type DIG_BE_SOFT_RESET;\
+	type HDCP_SOFT_RESET;\
+	type DIG_BE_SYMCLK_G_CLOCK_ON;\
+	type DIG_BE_SYMCLK_G_HDCP_CLOCK_ON;\
+	type DIG_BE_SYMCLK_G_TMDS_CLOCK_ON;\
+	type DISPCLK_R_GATE_DIS;\
+	type DISPCLK_G_GATE_DIS;\
+	type REFCLK_R_GATE_DIS;\
+	type REFCLK_G_GATE_DIS;\
+	type SOCCLK_G_GATE_DIS;\
+	type SYMCLK_FE_R_GATE_DIS;\
+	type SYMCLK_FE_G_GATE_DIS;\
+	type SYMCLK_R_GATE_DIS;\
+	type SYMCLK_G_GATE_DIS;\
+	type DIO_FGCG_REP_DIS;\
+	type DISPCLK_G_HDCP_GATE_DIS;\
+	type SYMCLKA_G_HDCP_GATE_DIS;\
+	type SYMCLKB_G_HDCP_GATE_DIS;\
+	type SYMCLKC_G_HDCP_GATE_DIS;\
+	type SYMCLKD_G_HDCP_GATE_DIS;\
+	type SYMCLKE_G_HDCP_GATE_DIS;\
+	type SYMCLKF_G_HDCP_GATE_DIS;\
+	type SYMCLKG_G_HDCP_GATE_DIS
+
+struct dcn10_link_enc_shift {
+	DCN_LINK_ENCODER_REG_FIELD_LIST(uint8_t);
+	DCN20_LINK_ENCODER_REG_FIELD_LIST(uint8_t);
+	DCN30_LINK_ENCODER_REG_FIELD_LIST(uint8_t);
+	DCN31_LINK_ENCODER_REG_FIELD_LIST(uint8_t);
+	DCN35_LINK_ENCODER_REG_FIELD_LIST(uint8_t);
+};
+
+struct dcn10_link_enc_mask {
+	DCN_LINK_ENCODER_REG_FIELD_LIST(uint32_t);
+	DCN20_LINK_ENCODER_REG_FIELD_LIST(uint32_t);
+	DCN30_LINK_ENCODER_REG_FIELD_LIST(uint32_t);
+	DCN31_LINK_ENCODER_REG_FIELD_LIST(uint32_t);
+	DCN35_LINK_ENCODER_REG_FIELD_LIST(uint32_t);
+};
+
+struct dcn10_link_encoder {
+	struct link_encoder base;
+	const struct dcn10_link_enc_registers *link_regs;
+	const struct dcn10_link_enc_aux_registers *aux_regs;
+	const struct dcn10_link_enc_hpd_registers *hpd_regs;
+	const struct dcn10_link_enc_shift *link_shift;
+	const struct dcn10_link_enc_mask *link_mask;
+};
+
+
+void dcn10_link_encoder_construct(
+	struct dcn10_link_encoder *enc10,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	const struct dcn10_link_enc_aux_registers *aux_regs,
+	const struct dcn10_link_enc_hpd_registers *hpd_regs,
+	const struct dcn10_link_enc_shift *link_shift,
+	const struct dcn10_link_enc_mask *link_mask);
+
+bool dcn10_link_encoder_validate_dvi_output(
+	const struct dcn10_link_encoder *enc10,
+	enum signal_type connector_signal,
+	enum signal_type signal,
+	const struct dc_crtc_timing *crtc_timing);
+
+bool dcn10_link_encoder_validate_rgb_output(
+	const struct dcn10_link_encoder *enc10,
+	const struct dc_crtc_timing *crtc_timing);
+
+bool dcn10_link_encoder_validate_dp_output(
+	const struct dcn10_link_encoder *enc10,
+	const struct dc_crtc_timing *crtc_timing);
+
+bool dcn10_link_encoder_validate_wireless_output(
+	const struct dcn10_link_encoder *enc10,
+	const struct dc_crtc_timing *crtc_timing);
+
+bool dcn10_link_encoder_validate_output_with_stream(
+	struct link_encoder *enc,
+	const struct dc_stream_state *stream);
+
+/****************** HW programming ************************/
+
+/* initialize HW */  /* why do we initialze aux in here? */
+void dcn10_link_encoder_hw_init(struct link_encoder *enc);
+
+void dcn10_link_encoder_destroy(struct link_encoder **enc);
+
+/* program DIG_MODE in DIG_BE */
+/* TODO can this be combined with enable_output? */
+void dcn10_link_encoder_setup(
+	struct link_encoder *enc,
+	enum signal_type signal);
+
+void enc1_configure_encoder(
+	struct dcn10_link_encoder *enc10,
+	const struct dc_link_settings *link_settings);
+
+/* enables TMDS PHY output */
+/* TODO: still need depth or just pass in adjusted pixel clock? */
+void dcn10_link_encoder_enable_tmds_output(
+	struct link_encoder *enc,
+	enum clock_source_id clock_source,
+	enum dc_color_depth color_depth,
+	enum signal_type signal,
+	uint32_t pixel_clock);
+
+void dcn10_link_encoder_enable_tmds_output_with_clk_pattern_wa(
+	struct link_encoder *enc,
+	enum clock_source_id clock_source,
+	enum dc_color_depth color_depth,
+	enum signal_type signal,
+	uint32_t pixel_clock);
+
+/* enables DP PHY output */
+void dcn10_link_encoder_enable_dp_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source);
+
+/* enables DP PHY output in MST mode */
+void dcn10_link_encoder_enable_dp_mst_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source);
+
+/* disable PHY output */
+void dcn10_link_encoder_disable_output(
+	struct link_encoder *enc,
+	enum signal_type signal);
+
+/* set DP lane settings */
+void dcn10_link_encoder_dp_set_lane_settings(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	const struct dc_lane_settings lane_settings[LANE_COUNT_DP_MAX]);
+
+void dcn10_link_encoder_dp_set_phy_pattern(
+	struct link_encoder *enc,
+	const struct encoder_set_dp_phy_pattern_param *param);
+
+/* programs DP MST VC payload allocation */
+void dcn10_link_encoder_update_mst_stream_allocation_table(
+	struct link_encoder *enc,
+	const struct link_mst_stream_allocation_table *table);
+
+void dcn10_link_encoder_connect_dig_be_to_fe(
+	struct link_encoder *enc,
+	enum engine_id engine,
+	bool connect);
+
+void dcn10_link_encoder_set_dp_phy_pattern_training_pattern(
+	struct link_encoder *enc,
+	uint32_t index);
+
+void dcn10_link_encoder_enable_hpd(struct link_encoder *enc);
+
+void dcn10_link_encoder_disable_hpd(struct link_encoder *enc);
+
+void dcn10_psr_program_dp_dphy_fast_training(struct link_encoder *enc,
+			bool exit_link_training_required);
+
+void dcn10_psr_program_secondary_packet(struct link_encoder *enc,
+			unsigned int sdp_transmit_line_num_deadline);
+
+bool dcn10_is_dig_enabled(struct link_encoder *enc);
+
+unsigned int dcn10_get_dig_frontend(struct link_encoder *enc);
+
+void dcn10_aux_initialize(struct dcn10_link_encoder *enc10);
+
+enum signal_type dcn10_get_dig_mode(
+	struct link_encoder *enc);
+
+void dcn10_link_encoder_get_max_link_cap(struct link_encoder *enc,
+	struct dc_link_settings *link_settings);
+#endif /* __DC_LINK_ENCODER__DCN10_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn10/dcn10_stream_encoder.c b/drivers/gpu/drm/amd/display/dc/dio/dcn10/dcn10_stream_encoder.c
new file mode 100644
index 000000000000..d928b4dcf6b8
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn10/dcn10_stream_encoder.c
@@ -0,0 +1,1627 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+#include "dc_bios_types.h"
+#include "dcn10_stream_encoder.h"
+#include "reg_helper.h"
+#include "hw_shared.h"
+#include "link_service.h"
+#include "dpcd_defs.h"
+#include "dcn30/dcn30_afmt.h"
+
+#define DC_LOGGER \
+		enc1->base.ctx->logger
+
+#define REG(reg)\
+	(enc1->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	enc1->se_shift->field_name, enc1->se_mask->field_name
+
+#define VBI_LINE_0 0
+#define DP_BLANK_MAX_RETRY 20
+#define HDMI_CLOCK_CHANNEL_RATE_MORE_340M 340000
+
+
+enum {
+	DP_MST_UPDATE_MAX_RETRY = 50
+};
+
+#define CTX \
+	enc1->base.ctx
+
+void enc1_update_generic_info_packet(
+	struct dcn10_stream_encoder *enc1,
+	uint32_t packet_index,
+	const struct dc_info_packet *info_packet)
+{
+	/* TODOFPGA Figure out a proper number for max_retries polling for lock
+	 * use 50 for now.
+	 */
+	uint32_t max_retries = 50;
+
+	/*we need turn on clock before programming AFMT block*/
+	REG_UPDATE(AFMT_CNTL, AFMT_AUDIO_CLOCK_EN, 1);
+
+	if (packet_index >= 8)
+		ASSERT(0);
+
+	/* poll dig_update_lock is not locked -> asic internal signal
+	 * assume otg master lock will unlock it
+	 */
+/*		REG_WAIT(AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_LOCK_STATUS,
+			0, 10, max_retries);*/
+
+	/* check if HW reading GSP memory */
+	REG_WAIT(AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_CONFLICT,
+			0, 10, max_retries);
+
+	/* HW does is not reading GSP memory not reading too long ->
+	 * something wrong. clear GPS memory access and notify?
+	 * hw SW is writing to GSP memory
+	 */
+	REG_UPDATE(AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_CONFLICT_CLR, 1);
+
+	/* choose which generic packet to use */
+	REG_UPDATE(AFMT_VBI_PACKET_CONTROL,
+			AFMT_GENERIC_INDEX, packet_index);
+
+	/* write generic packet header
+	 * (4th byte is for GENERIC0 only)
+	 */
+	REG_SET_4(AFMT_GENERIC_HDR, 0,
+			AFMT_GENERIC_HB0, info_packet->hb0,
+			AFMT_GENERIC_HB1, info_packet->hb1,
+			AFMT_GENERIC_HB2, info_packet->hb2,
+			AFMT_GENERIC_HB3, info_packet->hb3);
+
+	/* write generic packet contents
+	 * (we never use last 4 bytes)
+	 * there are 8 (0-7) mmDIG0_AFMT_GENERIC0_x registers
+	 */
+	{
+		const uint32_t *content =
+			(const uint32_t *) &info_packet->sb[0];
+
+		REG_WRITE(AFMT_GENERIC_0, *content++);
+		REG_WRITE(AFMT_GENERIC_1, *content++);
+		REG_WRITE(AFMT_GENERIC_2, *content++);
+		REG_WRITE(AFMT_GENERIC_3, *content++);
+		REG_WRITE(AFMT_GENERIC_4, *content++);
+		REG_WRITE(AFMT_GENERIC_5, *content++);
+		REG_WRITE(AFMT_GENERIC_6, *content++);
+		REG_WRITE(AFMT_GENERIC_7, *content);
+	}
+
+	switch (packet_index) {
+	case 0:
+		REG_UPDATE(AFMT_VBI_PACKET_CONTROL1,
+				AFMT_GENERIC0_IMMEDIATE_UPDATE, 1);
+		break;
+	case 1:
+		REG_UPDATE(AFMT_VBI_PACKET_CONTROL1,
+				AFMT_GENERIC1_IMMEDIATE_UPDATE, 1);
+		break;
+	case 2:
+		REG_UPDATE(AFMT_VBI_PACKET_CONTROL1,
+				AFMT_GENERIC2_IMMEDIATE_UPDATE, 1);
+		break;
+	case 3:
+		REG_UPDATE(AFMT_VBI_PACKET_CONTROL1,
+				AFMT_GENERIC3_IMMEDIATE_UPDATE, 1);
+		break;
+	case 4:
+		REG_UPDATE(AFMT_VBI_PACKET_CONTROL1,
+				AFMT_GENERIC4_IMMEDIATE_UPDATE, 1);
+		break;
+	case 5:
+		REG_UPDATE(AFMT_VBI_PACKET_CONTROL1,
+				AFMT_GENERIC5_IMMEDIATE_UPDATE, 1);
+		break;
+	case 6:
+		REG_UPDATE(AFMT_VBI_PACKET_CONTROL1,
+				AFMT_GENERIC6_IMMEDIATE_UPDATE, 1);
+		break;
+	case 7:
+		REG_UPDATE(AFMT_VBI_PACKET_CONTROL1,
+				AFMT_GENERIC7_IMMEDIATE_UPDATE, 1);
+		break;
+	default:
+		break;
+	}
+}
+
+static void enc1_update_hdmi_info_packet(
+	struct dcn10_stream_encoder *enc1,
+	uint32_t packet_index,
+	const struct dc_info_packet *info_packet)
+{
+	uint32_t cont, send, line;
+
+	if (info_packet->valid) {
+		enc1_update_generic_info_packet(
+			enc1,
+			packet_index,
+			info_packet);
+
+		/* enable transmission of packet(s) -
+		 * packet transmission begins on the next frame
+		 */
+		cont = 1;
+		/* send packet(s) every frame */
+		send = 1;
+		/* select line number to send packets on */
+		line = 2;
+	} else {
+		cont = 0;
+		send = 0;
+		line = 0;
+	}
+
+	/* choose which generic packet control to use */
+	switch (packet_index) {
+	case 0:
+		REG_UPDATE_3(HDMI_GENERIC_PACKET_CONTROL0,
+				HDMI_GENERIC0_CONT, cont,
+				HDMI_GENERIC0_SEND, send,
+				HDMI_GENERIC0_LINE, line);
+		break;
+	case 1:
+		REG_UPDATE_3(HDMI_GENERIC_PACKET_CONTROL0,
+				HDMI_GENERIC1_CONT, cont,
+				HDMI_GENERIC1_SEND, send,
+				HDMI_GENERIC1_LINE, line);
+		break;
+	case 2:
+		REG_UPDATE_3(HDMI_GENERIC_PACKET_CONTROL1,
+				HDMI_GENERIC0_CONT, cont,
+				HDMI_GENERIC0_SEND, send,
+				HDMI_GENERIC0_LINE, line);
+		break;
+	case 3:
+		REG_UPDATE_3(HDMI_GENERIC_PACKET_CONTROL1,
+				HDMI_GENERIC1_CONT, cont,
+				HDMI_GENERIC1_SEND, send,
+				HDMI_GENERIC1_LINE, line);
+		break;
+	case 4:
+		REG_UPDATE_3(HDMI_GENERIC_PACKET_CONTROL2,
+				HDMI_GENERIC0_CONT, cont,
+				HDMI_GENERIC0_SEND, send,
+				HDMI_GENERIC0_LINE, line);
+		break;
+	case 5:
+		REG_UPDATE_3(HDMI_GENERIC_PACKET_CONTROL2,
+				HDMI_GENERIC1_CONT, cont,
+				HDMI_GENERIC1_SEND, send,
+				HDMI_GENERIC1_LINE, line);
+		break;
+	case 6:
+		REG_UPDATE_3(HDMI_GENERIC_PACKET_CONTROL3,
+				HDMI_GENERIC0_CONT, cont,
+				HDMI_GENERIC0_SEND, send,
+				HDMI_GENERIC0_LINE, line);
+		break;
+	case 7:
+		REG_UPDATE_3(HDMI_GENERIC_PACKET_CONTROL3,
+				HDMI_GENERIC1_CONT, cont,
+				HDMI_GENERIC1_SEND, send,
+				HDMI_GENERIC1_LINE, line);
+		break;
+	default:
+		/* invalid HW packet index */
+		DC_LOG_WARNING(
+			"Invalid HW packet index: %s()\n",
+			__func__);
+		return;
+	}
+}
+
+/* setup stream encoder in dp mode */
+void enc1_stream_encoder_dp_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	enum dc_color_space output_color_space,
+	bool use_vsc_sdp_for_colorimetry,
+	uint32_t enable_sdp_splitting)
+{
+	uint32_t h_active_start;
+	uint32_t v_active_start;
+	uint32_t misc0 = 0;
+	uint32_t misc1 = 0;
+	uint32_t h_blank;
+	uint32_t h_back_porch;
+	uint8_t colorimetry_bpc;
+	uint8_t dp_pixel_encoding = 0;
+	uint8_t dp_component_depth = 0;
+
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+	struct dc_crtc_timing hw_crtc_timing = *crtc_timing;
+
+	if (hw_crtc_timing.flags.INTERLACE) {
+		/*the input timing is in VESA spec format with Interlace flag =1*/
+		hw_crtc_timing.v_total /= 2;
+		hw_crtc_timing.v_border_top /= 2;
+		hw_crtc_timing.v_addressable /= 2;
+		hw_crtc_timing.v_border_bottom /= 2;
+		hw_crtc_timing.v_front_porch /= 2;
+		hw_crtc_timing.v_sync_width /= 2;
+	}
+
+
+	/* set pixel encoding */
+	switch (hw_crtc_timing.pixel_encoding) {
+	case PIXEL_ENCODING_YCBCR422:
+		dp_pixel_encoding = DP_PIXEL_ENCODING_TYPE_YCBCR422;
+		break;
+	case PIXEL_ENCODING_YCBCR444:
+		dp_pixel_encoding = DP_PIXEL_ENCODING_TYPE_YCBCR444;
+
+		if (hw_crtc_timing.flags.Y_ONLY)
+			if (hw_crtc_timing.display_color_depth != COLOR_DEPTH_666)
+				/* HW testing only, no use case yet.
+				 * Color depth of Y-only could be
+				 * 8, 10, 12, 16 bits
+				 */
+				dp_pixel_encoding = DP_PIXEL_ENCODING_TYPE_Y_ONLY;
+
+		/* Note: DP_MSA_MISC1 bit 7 is the indicator
+		 * of Y-only mode.
+		 * This bit is set in HW if register
+		 * DP_PIXEL_ENCODING is programmed to 0x4
+		 */
+		break;
+	case PIXEL_ENCODING_YCBCR420:
+		dp_pixel_encoding = DP_PIXEL_ENCODING_TYPE_YCBCR420;
+		break;
+	default:
+		dp_pixel_encoding = DP_PIXEL_ENCODING_TYPE_RGB444;
+		break;
+	}
+
+	misc1 = REG_READ(DP_MSA_MISC);
+	/* For YCbCr420 and BT2020 Colorimetry Formats, VSC SDP shall be used.
+	 * When MISC1, bit 6, is Set to 1, a Source device uses a VSC SDP to indicate the
+	 * Pixel Encoding/Colorimetry Format and that a Sink device shall ignore MISC1, bit 7,
+	 * and MISC0, bits 7:1 (MISC1, bit 7, and MISC0, bits 7:1, become "don't care").
+	 */
+	if (use_vsc_sdp_for_colorimetry)
+		misc1 = misc1 | 0x40;
+	else
+		misc1 = misc1 & ~0x40;
+
+	/* set color depth */
+	switch (hw_crtc_timing.display_color_depth) {
+	case COLOR_DEPTH_666:
+		dp_component_depth = DP_COMPONENT_PIXEL_DEPTH_6BPC;
+		break;
+	case COLOR_DEPTH_888:
+		dp_component_depth = DP_COMPONENT_PIXEL_DEPTH_8BPC;
+		break;
+	case COLOR_DEPTH_101010:
+		dp_component_depth = DP_COMPONENT_PIXEL_DEPTH_10BPC;
+		break;
+	case COLOR_DEPTH_121212:
+		dp_component_depth = DP_COMPONENT_PIXEL_DEPTH_12BPC;
+		break;
+	case COLOR_DEPTH_161616:
+		dp_component_depth = DP_COMPONENT_PIXEL_DEPTH_16BPC;
+		break;
+	default:
+		dp_component_depth = DP_COMPONENT_PIXEL_DEPTH_6BPC;
+		break;
+	}
+
+	/* Set DP pixel encoding and component depth */
+	REG_UPDATE_2(DP_PIXEL_FORMAT,
+			DP_PIXEL_ENCODING, dp_pixel_encoding,
+			DP_COMPONENT_DEPTH, dp_component_depth);
+
+	/* set dynamic range and YCbCr range */
+
+	switch (hw_crtc_timing.display_color_depth) {
+	case COLOR_DEPTH_666:
+		colorimetry_bpc = 0;
+		break;
+	case COLOR_DEPTH_888:
+		colorimetry_bpc = 1;
+		break;
+	case COLOR_DEPTH_101010:
+		colorimetry_bpc = 2;
+		break;
+	case COLOR_DEPTH_121212:
+		colorimetry_bpc = 3;
+		break;
+	default:
+		colorimetry_bpc = 0;
+		break;
+	}
+
+	misc0 = colorimetry_bpc << 5;
+
+	switch (output_color_space) {
+	case COLOR_SPACE_SRGB:
+		misc1 = misc1 & ~0x80; /* bit7 = 0*/
+		break;
+	case COLOR_SPACE_SRGB_LIMITED:
+		misc0 = misc0 | 0x8; /* bit3=1 */
+		misc1 = misc1 & ~0x80; /* bit7 = 0*/
+		break;
+	case COLOR_SPACE_YCBCR601:
+	case COLOR_SPACE_YCBCR601_LIMITED:
+		misc0 = misc0 | 0x8; /* bit3=1, bit4=0 */
+		misc1 = misc1 & ~0x80; /* bit7 = 0*/
+		if (hw_crtc_timing.pixel_encoding == PIXEL_ENCODING_YCBCR422)
+			misc0 = misc0 | 0x2; /* bit2=0, bit1=1 */
+		else if (hw_crtc_timing.pixel_encoding == PIXEL_ENCODING_YCBCR444)
+			misc0 = misc0 | 0x4; /* bit2=1, bit1=0 */
+		break;
+	case COLOR_SPACE_YCBCR709:
+	case COLOR_SPACE_YCBCR709_LIMITED:
+		misc0 = misc0 | 0x18; /* bit3=1, bit4=1 */
+		misc1 = misc1 & ~0x80; /* bit7 = 0*/
+		if (hw_crtc_timing.pixel_encoding == PIXEL_ENCODING_YCBCR422)
+			misc0 = misc0 | 0x2; /* bit2=0, bit1=1 */
+		else if (hw_crtc_timing.pixel_encoding == PIXEL_ENCODING_YCBCR444)
+			misc0 = misc0 | 0x4; /* bit2=1, bit1=0 */
+		break;
+	case COLOR_SPACE_2020_RGB_LIMITEDRANGE:
+	case COLOR_SPACE_2020_RGB_FULLRANGE:
+	case COLOR_SPACE_2020_YCBCR_LIMITED:
+	case COLOR_SPACE_XR_RGB:
+	case COLOR_SPACE_MSREF_SCRGB:
+	case COLOR_SPACE_ADOBERGB:
+	case COLOR_SPACE_DCIP3:
+	case COLOR_SPACE_XV_YCC_709:
+	case COLOR_SPACE_XV_YCC_601:
+	case COLOR_SPACE_DISPLAYNATIVE:
+	case COLOR_SPACE_DOLBYVISION:
+	case COLOR_SPACE_APPCTRL:
+	case COLOR_SPACE_CUSTOMPOINTS:
+	case COLOR_SPACE_UNKNOWN:
+	case COLOR_SPACE_YCBCR709_BLACK:
+	default:
+		/* do nothing */
+		break;
+	}
+
+	REG_SET(DP_MSA_COLORIMETRY, 0, DP_MSA_MISC0, misc0);
+	REG_WRITE(DP_MSA_MISC, misc1);   /* MSA_MISC1 */
+
+	/* dcn new register
+	 * dc_crtc_timing is vesa dmt struct. data from edid
+	 */
+	REG_SET_2(DP_MSA_TIMING_PARAM1, 0,
+			DP_MSA_HTOTAL, hw_crtc_timing.h_total,
+			DP_MSA_VTOTAL, hw_crtc_timing.v_total);
+
+	/* calculate from vesa timing parameters
+	 * h_active_start related to leading edge of sync
+	 */
+
+	h_blank = hw_crtc_timing.h_total - hw_crtc_timing.h_border_left -
+			hw_crtc_timing.h_addressable - hw_crtc_timing.h_border_right;
+
+	h_back_porch = h_blank - hw_crtc_timing.h_front_porch -
+			hw_crtc_timing.h_sync_width;
+
+	/* start at beginning of left border */
+	h_active_start = hw_crtc_timing.h_sync_width + h_back_porch;
+
+
+	v_active_start = hw_crtc_timing.v_total - hw_crtc_timing.v_border_top -
+			hw_crtc_timing.v_addressable - hw_crtc_timing.v_border_bottom -
+			hw_crtc_timing.v_front_porch;
+
+
+	/* start at beginning of left border */
+	REG_SET_2(DP_MSA_TIMING_PARAM2, 0,
+		DP_MSA_HSTART, h_active_start,
+		DP_MSA_VSTART, v_active_start);
+
+	REG_SET_4(DP_MSA_TIMING_PARAM3, 0,
+			DP_MSA_HSYNCWIDTH,
+			hw_crtc_timing.h_sync_width,
+			DP_MSA_HSYNCPOLARITY,
+			!hw_crtc_timing.flags.HSYNC_POSITIVE_POLARITY,
+			DP_MSA_VSYNCWIDTH,
+			hw_crtc_timing.v_sync_width,
+			DP_MSA_VSYNCPOLARITY,
+			!hw_crtc_timing.flags.VSYNC_POSITIVE_POLARITY);
+
+	/* HWDITH include border or overscan */
+	REG_SET_2(DP_MSA_TIMING_PARAM4, 0,
+		DP_MSA_HWIDTH, hw_crtc_timing.h_border_left +
+		hw_crtc_timing.h_addressable + hw_crtc_timing.h_border_right,
+		DP_MSA_VHEIGHT, hw_crtc_timing.v_border_top +
+		hw_crtc_timing.v_addressable + hw_crtc_timing.v_border_bottom);
+}
+
+void enc1_stream_encoder_set_stream_attribute_helper(
+		struct dcn10_stream_encoder *enc1,
+		struct dc_crtc_timing *crtc_timing)
+{
+	switch (crtc_timing->pixel_encoding) {
+	case PIXEL_ENCODING_YCBCR422:
+		REG_UPDATE(DIG_FE_CNTL, TMDS_PIXEL_ENCODING, 1);
+		break;
+	default:
+		REG_UPDATE(DIG_FE_CNTL, TMDS_PIXEL_ENCODING, 0);
+		break;
+	}
+	REG_UPDATE(DIG_FE_CNTL, TMDS_COLOR_FORMAT, 0);
+}
+
+/* setup stream encoder in hdmi mode */
+void enc1_stream_encoder_hdmi_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	int actual_pix_clk_khz,
+	bool enable_audio)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+	struct bp_encoder_control cntl = {0};
+
+	cntl.action = ENCODER_CONTROL_SETUP;
+	cntl.engine_id = enc1->base.id;
+	cntl.signal = SIGNAL_TYPE_HDMI_TYPE_A;
+	cntl.enable_dp_audio = enable_audio;
+	cntl.pixel_clock = actual_pix_clk_khz;
+	cntl.lanes_number = LANE_COUNT_FOUR;
+
+	if (enc1->base.bp->funcs->encoder_control(
+			enc1->base.bp, &cntl) != BP_RESULT_OK)
+		return;
+
+	enc1_stream_encoder_set_stream_attribute_helper(enc1, crtc_timing);
+
+	/* setup HDMI engine */
+	REG_UPDATE_6(HDMI_CONTROL,
+		HDMI_PACKET_GEN_VERSION, 1,
+		HDMI_KEEPOUT_MODE, 1,
+		HDMI_DEEP_COLOR_ENABLE, 0,
+		HDMI_DATA_SCRAMBLE_EN, 0,
+		HDMI_NO_EXTRA_NULL_PACKET_FILLED, 1,
+		HDMI_CLOCK_CHANNEL_RATE, 0);
+
+
+	switch (crtc_timing->display_color_depth) {
+	case COLOR_DEPTH_888:
+		REG_UPDATE(HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 0);
+		DC_LOG_DEBUG("HDMI source set to 24BPP deep color depth\n");
+		break;
+	case COLOR_DEPTH_101010:
+		if (crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR422) {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 1,
+					HDMI_DEEP_COLOR_ENABLE, 0);
+			DC_LOG_DEBUG("HDMI source 30BPP deep color depth"  \
+				"disabled for YCBCR422 pixel encoding\n");
+		} else {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 1,
+					HDMI_DEEP_COLOR_ENABLE, 1);
+			DC_LOG_DEBUG("HDMI source 30BPP deep color depth"  \
+				"enabled for YCBCR422 non-pixel encoding\n");
+			}
+		break;
+	case COLOR_DEPTH_121212:
+		if (crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR422) {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 2,
+					HDMI_DEEP_COLOR_ENABLE, 0);
+			DC_LOG_DEBUG("HDMI source 36BPP deep color depth"  \
+				"disabled for YCBCR422 pixel encoding\n");
+		} else {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 2,
+					HDMI_DEEP_COLOR_ENABLE, 1);
+			DC_LOG_DEBUG("HDMI source 36BPP deep color depth"  \
+				"enabled for non-pixel YCBCR422 encoding\n");
+			}
+		break;
+	case COLOR_DEPTH_161616:
+		REG_UPDATE_2(HDMI_CONTROL,
+				HDMI_DEEP_COLOR_DEPTH, 3,
+				HDMI_DEEP_COLOR_ENABLE, 1);
+		DC_LOG_DEBUG("HDMI source deep color depth enabled in"  \
+				"reserved mode\n");
+		break;
+	default:
+		break;
+	}
+
+	if (actual_pix_clk_khz >= HDMI_CLOCK_CHANNEL_RATE_MORE_340M) {
+		/* enable HDMI data scrambler
+		 * HDMI_CLOCK_CHANNEL_RATE_MORE_340M
+		 * Clock channel frequency is 1/4 of character rate.
+		 */
+		REG_UPDATE_2(HDMI_CONTROL,
+			HDMI_DATA_SCRAMBLE_EN, 1,
+			HDMI_CLOCK_CHANNEL_RATE, 1);
+	} else if (crtc_timing->flags.LTE_340MCSC_SCRAMBLE) {
+
+		/* TODO: New feature for DCE11, still need to implement */
+
+		/* enable HDMI data scrambler
+		 * HDMI_CLOCK_CHANNEL_FREQ_EQUAL_TO_CHAR_RATE
+		 * Clock channel frequency is the same
+		 * as character rate
+		 */
+		REG_UPDATE_2(HDMI_CONTROL,
+			HDMI_DATA_SCRAMBLE_EN, 1,
+			HDMI_CLOCK_CHANNEL_RATE, 0);
+	}
+
+
+	REG_UPDATE_3(HDMI_VBI_PACKET_CONTROL,
+		HDMI_GC_CONT, 1,
+		HDMI_GC_SEND, 1,
+		HDMI_NULL_SEND, 1);
+
+	REG_UPDATE(HDMI_VBI_PACKET_CONTROL, HDMI_ACP_SEND, 0);
+
+	/* following belongs to audio */
+	REG_UPDATE(HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, 1);
+
+	REG_UPDATE(AFMT_INFOFRAME_CONTROL0, AFMT_AUDIO_INFO_UPDATE, 1);
+
+	REG_UPDATE(HDMI_INFOFRAME_CONTROL1, HDMI_AUDIO_INFO_LINE,
+				VBI_LINE_0 + 2);
+
+	REG_UPDATE(HDMI_GC, HDMI_GC_AVMUTE, 0);
+}
+
+/* setup stream encoder in dvi mode */
+void enc1_stream_encoder_dvi_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	bool is_dual_link)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+	struct bp_encoder_control cntl = {0};
+
+	cntl.action = ENCODER_CONTROL_SETUP;
+	cntl.engine_id = enc1->base.id;
+	cntl.signal = is_dual_link ?
+			SIGNAL_TYPE_DVI_DUAL_LINK : SIGNAL_TYPE_DVI_SINGLE_LINK;
+	cntl.enable_dp_audio = false;
+	cntl.pixel_clock = crtc_timing->pix_clk_100hz / 10;
+	cntl.lanes_number = (is_dual_link) ? LANE_COUNT_EIGHT : LANE_COUNT_FOUR;
+
+	if (enc1->base.bp->funcs->encoder_control(
+			enc1->base.bp, &cntl) != BP_RESULT_OK)
+		return;
+
+	ASSERT(crtc_timing->pixel_encoding == PIXEL_ENCODING_RGB);
+	ASSERT(crtc_timing->display_color_depth == COLOR_DEPTH_888);
+	enc1_stream_encoder_set_stream_attribute_helper(enc1, crtc_timing);
+}
+
+void enc1_stream_encoder_set_throttled_vcp_size(
+	struct stream_encoder *enc,
+	struct fixed31_32 avg_time_slots_per_mtp)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+	uint32_t x = dc_fixpt_floor(
+		avg_time_slots_per_mtp);
+	uint32_t y = dc_fixpt_ceil(
+		dc_fixpt_shl(
+			dc_fixpt_sub_int(
+				avg_time_slots_per_mtp,
+				x),
+			26));
+
+	// If y rounds up to integer, carry it over to x.
+	if (y >> 26) {
+		x += 1;
+		y = 0;
+	}
+
+	REG_SET_2(DP_MSE_RATE_CNTL, 0,
+		DP_MSE_RATE_X, x,
+		DP_MSE_RATE_Y, y);
+
+	/* wait for update to be completed on the link */
+	/* i.e. DP_MSE_RATE_UPDATE_PENDING field (read only) */
+	/* is reset to 0 (not pending) */
+	REG_WAIT(DP_MSE_RATE_UPDATE, DP_MSE_RATE_UPDATE_PENDING,
+			0,
+			10, DP_MST_UPDATE_MAX_RETRY);
+}
+
+static void enc1_stream_encoder_update_hdmi_info_packets(
+	struct stream_encoder *enc,
+	const struct encoder_info_frame *info_frame)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	/* for bring up, disable dp double  TODO */
+	REG_UPDATE(HDMI_DB_CONTROL, HDMI_DB_DISABLE, 1);
+
+	/*Always add mandatory packets first followed by optional ones*/
+	enc1_update_hdmi_info_packet(enc1, 0, &info_frame->avi);
+	enc1_update_hdmi_info_packet(enc1, 1, &info_frame->hfvsif);
+	enc1_update_hdmi_info_packet(enc1, 2, &info_frame->gamut);
+	enc1_update_hdmi_info_packet(enc1, 3, &info_frame->vendor);
+	enc1_update_hdmi_info_packet(enc1, 4, &info_frame->spd);
+	enc1_update_hdmi_info_packet(enc1, 5, &info_frame->hdrsmd);
+}
+
+static void enc1_stream_encoder_stop_hdmi_info_packets(
+	struct stream_encoder *enc)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	/* stop generic packets 0 & 1 on HDMI */
+	REG_SET_6(HDMI_GENERIC_PACKET_CONTROL0, 0,
+		HDMI_GENERIC1_CONT, 0,
+		HDMI_GENERIC1_LINE, 0,
+		HDMI_GENERIC1_SEND, 0,
+		HDMI_GENERIC0_CONT, 0,
+		HDMI_GENERIC0_LINE, 0,
+		HDMI_GENERIC0_SEND, 0);
+
+	/* stop generic packets 2 & 3 on HDMI */
+	REG_SET_6(HDMI_GENERIC_PACKET_CONTROL1, 0,
+		HDMI_GENERIC0_CONT, 0,
+		HDMI_GENERIC0_LINE, 0,
+		HDMI_GENERIC0_SEND, 0,
+		HDMI_GENERIC1_CONT, 0,
+		HDMI_GENERIC1_LINE, 0,
+		HDMI_GENERIC1_SEND, 0);
+
+	/* stop generic packets 2 & 3 on HDMI */
+	REG_SET_6(HDMI_GENERIC_PACKET_CONTROL2, 0,
+		HDMI_GENERIC0_CONT, 0,
+		HDMI_GENERIC0_LINE, 0,
+		HDMI_GENERIC0_SEND, 0,
+		HDMI_GENERIC1_CONT, 0,
+		HDMI_GENERIC1_LINE, 0,
+		HDMI_GENERIC1_SEND, 0);
+
+	REG_SET_6(HDMI_GENERIC_PACKET_CONTROL3, 0,
+		HDMI_GENERIC0_CONT, 0,
+		HDMI_GENERIC0_LINE, 0,
+		HDMI_GENERIC0_SEND, 0,
+		HDMI_GENERIC1_CONT, 0,
+		HDMI_GENERIC1_LINE, 0,
+		HDMI_GENERIC1_SEND, 0);
+}
+
+void enc1_stream_encoder_update_dp_info_packets(
+	struct stream_encoder *enc,
+	const struct encoder_info_frame *info_frame)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+	uint32_t value = 0;
+
+	if (info_frame->vsc.valid)
+		enc1_update_generic_info_packet(
+					enc1,
+					0,  /* packetIndex */
+					&info_frame->vsc);
+
+	/* VSC SDP at packetIndex 1 is used by PSR in DMCUB FW.
+	 * Note that the enablement of GSP1 is not done below,
+	 * it's done in FW.
+	 */
+	if (info_frame->vsc.valid)
+		enc1_update_generic_info_packet(
+					enc1,
+					1,  /* packetIndex */
+					&info_frame->vsc);
+
+	if (info_frame->spd.valid)
+		enc1_update_generic_info_packet(
+				enc1,
+				2,  /* packetIndex */
+				&info_frame->spd);
+
+	if (info_frame->hdrsmd.valid)
+		enc1_update_generic_info_packet(
+				enc1,
+				3,  /* packetIndex */
+				&info_frame->hdrsmd);
+
+	/* packetIndex 4 is used for send immediate sdp message, and please
+	 * use other packetIndex (such as 5,6) for other info packet
+	 */
+
+	if (info_frame->adaptive_sync.valid)
+		enc1_update_generic_info_packet(
+				enc1,
+				5,  /* packetIndex */
+				&info_frame->adaptive_sync);
+
+	/* enable/disable transmission of packet(s).
+	 * If enabled, packet transmission begins on the next frame
+	 */
+	REG_UPDATE(DP_SEC_CNTL, DP_SEC_GSP0_ENABLE, info_frame->vsc.valid);
+	REG_UPDATE(DP_SEC_CNTL, DP_SEC_GSP2_ENABLE, info_frame->spd.valid);
+	REG_UPDATE(DP_SEC_CNTL, DP_SEC_GSP3_ENABLE, info_frame->hdrsmd.valid);
+	REG_UPDATE(DP_SEC_CNTL, DP_SEC_GSP5_ENABLE, info_frame->adaptive_sync.valid);
+
+	/* This bit is the master enable bit.
+	 * When enabling secondary stream engine,
+	 * this master bit must also be set.
+	 * This register shared with audio info frame.
+	 * Therefore we need to enable master bit
+	 * if at least on of the fields is not 0
+	 */
+	value = REG_READ(DP_SEC_CNTL);
+	if (value)
+		REG_UPDATE(DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, 1);
+}
+
+void enc1_stream_encoder_send_immediate_sdp_message(
+	struct stream_encoder *enc,
+	const uint8_t *custom_sdp_message,
+	unsigned int sdp_message_size)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+	uint32_t value = 0;
+
+	/* TODOFPGA Figure out a proper number for max_retries polling for lock
+	 * use 50 for now.
+	 */
+	uint32_t max_retries = 50;
+
+	/* check if GSP4 is transmitted */
+	REG_WAIT(DP_SEC_CNTL2, DP_SEC_GSP4_SEND_PENDING,
+		0, 10, max_retries);
+
+	/* disable GSP4 transmitting */
+	REG_UPDATE(DP_SEC_CNTL2, DP_SEC_GSP4_SEND, 0);
+
+	/* transmit GSP4 at the earliest time in a frame */
+	REG_UPDATE(DP_SEC_CNTL2, DP_SEC_GSP4_SEND_ANY_LINE, 1);
+
+	/*we need turn on clock before programming AFMT block*/
+	REG_UPDATE(AFMT_CNTL, AFMT_AUDIO_CLOCK_EN, 1);
+
+	/* check if HW reading GSP memory */
+	REG_WAIT(AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_CONFLICT,
+			0, 10, max_retries);
+
+	/* HW does is not reading GSP memory not reading too long ->
+	 * something wrong. clear GPS memory access and notify?
+	 * hw SW is writing to GSP memory
+	 */
+	REG_UPDATE(AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_CONFLICT_CLR, 1);
+
+	/* use generic packet 4 for immediate sdp message */
+	REG_UPDATE(AFMT_VBI_PACKET_CONTROL,
+			AFMT_GENERIC_INDEX, 4);
+
+	/* write generic packet header
+	 * (4th byte is for GENERIC0 only)
+	 */
+	REG_SET_4(AFMT_GENERIC_HDR, 0,
+			AFMT_GENERIC_HB0, custom_sdp_message[0],
+			AFMT_GENERIC_HB1, custom_sdp_message[1],
+			AFMT_GENERIC_HB2, custom_sdp_message[2],
+			AFMT_GENERIC_HB3, custom_sdp_message[3]);
+
+	/* write generic packet contents
+	 * (we never use last 4 bytes)
+	 * there are 8 (0-7) mmDIG0_AFMT_GENERIC0_x registers
+	 */
+	{
+		const uint32_t *content =
+			(const uint32_t *) &custom_sdp_message[4];
+
+		REG_WRITE(AFMT_GENERIC_0, *content++);
+		REG_WRITE(AFMT_GENERIC_1, *content++);
+		REG_WRITE(AFMT_GENERIC_2, *content++);
+		REG_WRITE(AFMT_GENERIC_3, *content++);
+		REG_WRITE(AFMT_GENERIC_4, *content++);
+		REG_WRITE(AFMT_GENERIC_5, *content++);
+		REG_WRITE(AFMT_GENERIC_6, *content++);
+		REG_WRITE(AFMT_GENERIC_7, *content);
+	}
+
+	/* check whether GENERIC4 registers double buffer update in immediate mode
+	 * is pending
+	 */
+	REG_WAIT(AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC4_IMMEDIATE_UPDATE_PENDING,
+			0, 10, max_retries);
+
+	/* atomically update double-buffered GENERIC4 registers in immediate mode
+	 * (update immediately)
+	 */
+	REG_UPDATE(AFMT_VBI_PACKET_CONTROL1,
+			AFMT_GENERIC4_IMMEDIATE_UPDATE, 1);
+
+	/* enable GSP4 transmitting */
+	REG_UPDATE(DP_SEC_CNTL2, DP_SEC_GSP4_SEND, 1);
+
+	/* This bit is the master enable bit.
+	 * When enabling secondary stream engine,
+	 * this master bit must also be set.
+	 * This register shared with audio info frame.
+	 * Therefore we need to enable master bit
+	 * if at least on of the fields is not 0
+	 */
+	value = REG_READ(DP_SEC_CNTL);
+	if (value)
+		REG_UPDATE(DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, 1);
+}
+
+void enc1_stream_encoder_stop_dp_info_packets(
+	struct stream_encoder *enc)
+{
+	/* stop generic packets on DP */
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+	uint32_t value = 0;
+
+	REG_SET_10(DP_SEC_CNTL, 0,
+		DP_SEC_GSP0_ENABLE, 0,
+		DP_SEC_GSP1_ENABLE, 0,
+		DP_SEC_GSP2_ENABLE, 0,
+		DP_SEC_GSP3_ENABLE, 0,
+		DP_SEC_GSP4_ENABLE, 0,
+		DP_SEC_GSP5_ENABLE, 0,
+		DP_SEC_GSP6_ENABLE, 0,
+		DP_SEC_GSP7_ENABLE, 0,
+		DP_SEC_MPG_ENABLE, 0,
+		DP_SEC_STREAM_ENABLE, 0);
+
+	/* this register shared with audio info frame.
+	 * therefore we need to keep master enabled
+	 * if at least one of the fields is not 0 */
+	value = REG_READ(DP_SEC_CNTL);
+	if (value)
+		REG_UPDATE(DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, 1);
+
+}
+
+void enc1_stream_encoder_dp_blank(
+	struct dc_link *link,
+	struct stream_encoder *enc)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+	uint32_t  reg1 = 0;
+	uint32_t max_retries = DP_BLANK_MAX_RETRY * 10;
+
+	/* Note: For CZ, we are changing driver default to disable
+	 * stream deferred to next VBLANK. If results are positive, we
+	 * will make the same change to all DCE versions. There are a
+	 * handful of panels that cannot handle disable stream at
+	 * HBLANK and will result in a white line flash across the
+	 * screen on stream disable.
+	 */
+	REG_GET(DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, &reg1);
+	if ((reg1 & 0x1) == 0)
+		/*stream not enabled*/
+		return;
+	/* Specify the video stream disable point
+	 * (2 = start of the next vertical blank)
+	 */
+	REG_UPDATE(DP_VID_STREAM_CNTL, DP_VID_STREAM_DIS_DEFER, 2);
+	/* Larger delay to wait until VBLANK - use max retry of
+	 * 10us*10200=102ms. This covers 100.0ms of minimum 10 Hz mode +
+	 * a little more because we may not trust delay accuracy.
+	 */
+	max_retries = DP_BLANK_MAX_RETRY * 501;
+
+	/* disable DP stream */
+	REG_UPDATE(DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, 0);
+
+	link->dc->link_srv->dp_trace_source_sequence(link, DPCD_SOURCE_SEQ_AFTER_DISABLE_DP_VID_STREAM);
+
+	/* the encoder stops sending the video stream
+	 * at the start of the vertical blanking.
+	 * Poll for DP_VID_STREAM_STATUS == 0
+	 */
+
+	REG_WAIT(DP_VID_STREAM_CNTL, DP_VID_STREAM_STATUS,
+			0,
+			10, max_retries);
+
+	/* Tell the DP encoder to ignore timing from CRTC, must be done after
+	 * the polling. If we set DP_STEER_FIFO_RESET before DP stream blank is
+	 * complete, stream status will be stuck in video stream enabled state,
+	 * i.e. DP_VID_STREAM_STATUS stuck at 1.
+	 */
+
+	REG_UPDATE(DP_STEER_FIFO, DP_STEER_FIFO_RESET, true);
+
+	link->dc->link_srv->dp_trace_source_sequence(link, DPCD_SOURCE_SEQ_AFTER_FIFO_STEER_RESET);
+}
+
+/* output video stream to link encoder */
+void enc1_stream_encoder_dp_unblank(
+	struct dc_link *link,
+	struct stream_encoder *enc,
+	const struct encoder_unblank_param *param)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (param->link_settings.link_rate != LINK_RATE_UNKNOWN) {
+		uint32_t n_vid = 0x8000;
+		uint32_t m_vid;
+		uint32_t n_multiply = 0;
+		uint64_t m_vid_l = n_vid;
+
+		/* YCbCr 4:2:0 : Computed VID_M will be 2X the input rate */
+		if (param->timing.pixel_encoding == PIXEL_ENCODING_YCBCR420) {
+			/*this param->pixel_clk_khz is half of 444 rate for 420 already*/
+			n_multiply = 1;
+		}
+		/* M / N = Fstream / Flink
+		 * m_vid / n_vid = pixel rate / link rate
+		 */
+
+		m_vid_l *= param->timing.pix_clk_100hz / 10;
+		m_vid_l = div_u64(m_vid_l,
+			param->link_settings.link_rate
+				* LINK_RATE_REF_FREQ_IN_KHZ);
+
+		m_vid = (uint32_t) m_vid_l;
+
+		/* enable auto measurement */
+
+		REG_UPDATE(DP_VID_TIMING, DP_VID_M_N_GEN_EN, 0);
+
+		/* auto measurement need 1 full 0x8000 symbol cycle to kick in,
+		 * therefore program initial value for Mvid and Nvid
+		 */
+
+		REG_UPDATE(DP_VID_N, DP_VID_N, n_vid);
+
+		REG_UPDATE(DP_VID_M, DP_VID_M, m_vid);
+
+		REG_UPDATE_2(DP_VID_TIMING,
+				DP_VID_M_N_GEN_EN, 1,
+				DP_VID_N_MUL, n_multiply);
+	}
+
+	/* set DIG_START to 0x1 to resync FIFO */
+
+	REG_UPDATE(DIG_FE_CNTL, DIG_START, 1);
+
+	/* switch DP encoder to CRTC data */
+
+	REG_UPDATE(DP_STEER_FIFO, DP_STEER_FIFO_RESET, 0);
+
+	/* wait 100us for DIG/DP logic to prime
+	 * (i.e. a few video lines)
+	 */
+	udelay(100);
+
+	/* the hardware would start sending video at the start of the next DP
+	 * frame (i.e. rising edge of the vblank).
+	 * NOTE: We used to program DP_VID_STREAM_DIS_DEFER = 2 here, but this
+	 * register has no effect on enable transition! HW always guarantees
+	 * VID_STREAM enable at start of next frame, and this is not
+	 * programmable
+	 */
+
+	REG_UPDATE(DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, true);
+
+	link->dc->link_srv->dp_trace_source_sequence(link,
+			DPCD_SOURCE_SEQ_AFTER_ENABLE_DP_VID_STREAM);
+}
+
+void enc1_stream_encoder_set_avmute(
+	struct stream_encoder *enc,
+	bool enable)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+	unsigned int value = enable ? 1 : 0;
+
+	REG_UPDATE(HDMI_GC, HDMI_GC_AVMUTE, value);
+}
+
+void enc1_reset_hdmi_stream_attribute(
+	struct stream_encoder *enc)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	REG_UPDATE_5(HDMI_CONTROL,
+		HDMI_PACKET_GEN_VERSION, 1,
+		HDMI_KEEPOUT_MODE, 1,
+		HDMI_DEEP_COLOR_ENABLE, 0,
+		HDMI_DATA_SCRAMBLE_EN, 0,
+		HDMI_CLOCK_CHANNEL_RATE, 0);
+}
+
+
+#define DP_SEC_AUD_N__DP_SEC_AUD_N__DEFAULT 0x8000
+#define DP_SEC_TIMESTAMP__DP_SEC_TIMESTAMP_MODE__AUTO_CALC 1
+
+#include "include/audio_types.h"
+
+
+/* 25.2MHz/1.001*/
+/* 25.2MHz/1.001*/
+/* 25.2MHz*/
+/* 27MHz */
+/* 27MHz*1.001*/
+/* 27MHz*1.001*/
+/* 54MHz*/
+/* 54MHz*1.001*/
+/* 74.25MHz/1.001*/
+/* 74.25MHz*/
+/* 148.5MHz/1.001*/
+/* 148.5MHz*/
+
+static const struct audio_clock_info audio_clock_info_table[16] = {
+	{2517, 4576, 28125, 7007, 31250, 6864, 28125},
+	{2518, 4576, 28125, 7007, 31250, 6864, 28125},
+	{2520, 4096, 25200, 6272, 28000, 6144, 25200},
+	{2700, 4096, 27000, 6272, 30000, 6144, 27000},
+	{2702, 4096, 27027, 6272, 30030, 6144, 27027},
+	{2703, 4096, 27027, 6272, 30030, 6144, 27027},
+	{5400, 4096, 54000, 6272, 60000, 6144, 54000},
+	{5405, 4096, 54054, 6272, 60060, 6144, 54054},
+	{7417, 11648, 210937, 17836, 234375, 11648, 140625},
+	{7425, 4096, 74250, 6272, 82500, 6144, 74250},
+	{14835, 11648, 421875, 8918, 234375, 5824, 140625},
+	{14850, 4096, 148500, 6272, 165000, 6144, 148500},
+	{29670, 5824, 421875, 4459, 234375, 5824, 281250},
+	{29700, 3072, 222750, 4704, 247500, 5120, 247500},
+	{59340, 5824, 843750, 8918, 937500, 5824, 562500},
+	{59400, 3072, 445500, 9408, 990000, 6144, 594000}
+};
+
+static const struct audio_clock_info audio_clock_info_table_36bpc[14] = {
+	{2517,  9152,  84375,  7007,  48875,  9152,  56250},
+	{2518,  9152,  84375,  7007,  48875,  9152,  56250},
+	{2520,  4096,  37800,  6272,  42000,  6144,  37800},
+	{2700,  4096,  40500,  6272,  45000,  6144,  40500},
+	{2702,  8192,  81081,  6272,  45045,  8192,  54054},
+	{2703,  8192,  81081,  6272,  45045,  8192,  54054},
+	{5400,  4096,  81000,  6272,  90000,  6144,  81000},
+	{5405,  4096,  81081,  6272,  90090,  6144,  81081},
+	{7417, 11648, 316406, 17836, 351562, 11648, 210937},
+	{7425, 4096, 111375,  6272, 123750,  6144, 111375},
+	{14835, 11648, 632812, 17836, 703125, 11648, 421875},
+	{14850, 4096, 222750,  6272, 247500,  6144, 222750},
+	{29670, 5824, 632812,  8918, 703125,  5824, 421875},
+	{29700, 4096, 445500,  4704, 371250,  5120, 371250}
+};
+
+static const struct audio_clock_info audio_clock_info_table_48bpc[14] = {
+	{2517,  4576,  56250,  7007,  62500,  6864,  56250},
+	{2518,  4576,  56250,  7007,  62500,  6864,  56250},
+	{2520,  4096,  50400,  6272,  56000,  6144,  50400},
+	{2700,  4096,  54000,  6272,  60000,  6144,  54000},
+	{2702,  4096,  54054,  6267,  60060,  8192,  54054},
+	{2703,  4096,  54054,  6272,  60060,  8192,  54054},
+	{5400,  4096, 108000,  6272, 120000,  6144, 108000},
+	{5405,  4096, 108108,  6272, 120120,  6144, 108108},
+	{7417, 11648, 421875, 17836, 468750, 11648, 281250},
+	{7425,  4096, 148500,  6272, 165000,  6144, 148500},
+	{14835, 11648, 843750,  8918, 468750, 11648, 281250},
+	{14850, 4096, 297000,  6272, 330000,  6144, 297000},
+	{29670, 5824, 843750,  4459, 468750,  5824, 562500},
+	{29700, 3072, 445500,  4704, 495000,  5120, 495000}
+
+
+};
+
+static union audio_cea_channels speakers_to_channels(
+	struct audio_speaker_flags speaker_flags)
+{
+	union audio_cea_channels cea_channels = {0};
+
+	/* these are one to one */
+	cea_channels.channels.FL = speaker_flags.FL_FR;
+	cea_channels.channels.FR = speaker_flags.FL_FR;
+	cea_channels.channels.LFE = speaker_flags.LFE;
+	cea_channels.channels.FC = speaker_flags.FC;
+
+	/* if Rear Left and Right exist move RC speaker to channel 7
+	 * otherwise to channel 5
+	 */
+	if (speaker_flags.RL_RR) {
+		cea_channels.channels.RL_RC = speaker_flags.RL_RR;
+		cea_channels.channels.RR = speaker_flags.RL_RR;
+		cea_channels.channels.RC_RLC_FLC = speaker_flags.RC;
+	} else {
+		cea_channels.channels.RL_RC = speaker_flags.RC;
+	}
+
+	/* FRONT Left Right Center and REAR Left Right Center are exclusive */
+	if (speaker_flags.FLC_FRC) {
+		cea_channels.channels.RC_RLC_FLC = speaker_flags.FLC_FRC;
+		cea_channels.channels.RRC_FRC = speaker_flags.FLC_FRC;
+	} else {
+		cea_channels.channels.RC_RLC_FLC = speaker_flags.RLC_RRC;
+		cea_channels.channels.RRC_FRC = speaker_flags.RLC_RRC;
+	}
+
+	return cea_channels;
+}
+
+void get_audio_clock_info(
+	enum dc_color_depth color_depth,
+	uint32_t crtc_pixel_clock_100Hz,
+	uint32_t actual_pixel_clock_100Hz,
+	struct audio_clock_info *audio_clock_info)
+{
+	const struct audio_clock_info *clock_info;
+	uint32_t index;
+	uint32_t crtc_pixel_clock_in_10khz = crtc_pixel_clock_100Hz / 100;
+	uint32_t audio_array_size;
+
+	switch (color_depth) {
+	case COLOR_DEPTH_161616:
+		clock_info = audio_clock_info_table_48bpc;
+		audio_array_size = ARRAY_SIZE(
+				audio_clock_info_table_48bpc);
+		break;
+	case COLOR_DEPTH_121212:
+		clock_info = audio_clock_info_table_36bpc;
+		audio_array_size = ARRAY_SIZE(
+				audio_clock_info_table_36bpc);
+		break;
+	default:
+		clock_info = audio_clock_info_table;
+		audio_array_size = ARRAY_SIZE(
+				audio_clock_info_table);
+		break;
+	}
+
+	if (clock_info != NULL) {
+		/* search for exact pixel clock in table */
+		for (index = 0; index < audio_array_size; index++) {
+			if (clock_info[index].pixel_clock_in_10khz >
+				crtc_pixel_clock_in_10khz)
+				break;  /* not match */
+			else if (clock_info[index].pixel_clock_in_10khz ==
+					crtc_pixel_clock_in_10khz) {
+				/* match found */
+				*audio_clock_info = clock_info[index];
+				return;
+			}
+		}
+	}
+
+	/* not found */
+	if (actual_pixel_clock_100Hz == 0)
+		actual_pixel_clock_100Hz = crtc_pixel_clock_100Hz;
+
+	/* See HDMI spec  the table entry under
+	 *  pixel clock of "Other". */
+	audio_clock_info->pixel_clock_in_10khz =
+			actual_pixel_clock_100Hz / 100;
+	audio_clock_info->cts_32khz = actual_pixel_clock_100Hz / 10;
+	audio_clock_info->cts_44khz = actual_pixel_clock_100Hz / 10;
+	audio_clock_info->cts_48khz = actual_pixel_clock_100Hz / 10;
+
+	audio_clock_info->n_32khz = 4096;
+	audio_clock_info->n_44khz = 6272;
+	audio_clock_info->n_48khz = 6144;
+}
+
+static void enc1_se_audio_setup(
+	struct stream_encoder *enc,
+	unsigned int az_inst,
+	struct audio_info *audio_info)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	uint32_t channels = 0;
+
+	ASSERT(audio_info);
+	if (audio_info == NULL)
+		/* This should not happen.it does so we don't get BSOD*/
+		return;
+
+	channels = speakers_to_channels(audio_info->flags.speaker_flags).all;
+
+	/* setup the audio stream source select (audio -> dig mapping) */
+	REG_SET(AFMT_AUDIO_SRC_CONTROL, 0, AFMT_AUDIO_SRC_SELECT, az_inst);
+
+	/* Channel allocation */
+	REG_UPDATE(AFMT_AUDIO_PACKET_CONTROL2, AFMT_AUDIO_CHANNEL_ENABLE, channels);
+}
+
+static void enc1_se_setup_hdmi_audio(
+	struct stream_encoder *enc,
+	const struct audio_crtc_info *crtc_info)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	struct audio_clock_info audio_clock_info = {0};
+
+	/* HDMI_AUDIO_PACKET_CONTROL */
+	REG_UPDATE(HDMI_AUDIO_PACKET_CONTROL,
+			HDMI_AUDIO_DELAY_EN, 1);
+
+	/* AFMT_AUDIO_PACKET_CONTROL */
+	REG_UPDATE(AFMT_AUDIO_PACKET_CONTROL, AFMT_60958_CS_UPDATE, 1);
+
+	/* AFMT_AUDIO_PACKET_CONTROL2 */
+	REG_UPDATE_2(AFMT_AUDIO_PACKET_CONTROL2,
+			AFMT_AUDIO_LAYOUT_OVRD, 0,
+			AFMT_60958_OSF_OVRD, 0);
+
+	/* HDMI_ACR_PACKET_CONTROL */
+	REG_UPDATE_3(HDMI_ACR_PACKET_CONTROL,
+			HDMI_ACR_AUTO_SEND, 1,
+			HDMI_ACR_SOURCE, 0,
+			HDMI_ACR_AUDIO_PRIORITY, 0);
+
+	/* Program audio clock sample/regeneration parameters */
+	get_audio_clock_info(crtc_info->color_depth,
+			     crtc_info->requested_pixel_clock_100Hz,
+			     crtc_info->calculated_pixel_clock_100Hz,
+			     &audio_clock_info);
+	DC_LOG_HW_AUDIO(
+			"\n%s:Input::requested_pixel_clock_100Hz = %d"	\
+			"calculated_pixel_clock_100Hz = %d \n", __func__,	\
+			crtc_info->requested_pixel_clock_100Hz,		\
+			crtc_info->calculated_pixel_clock_100Hz);
+
+	/* HDMI_ACR_32_0__HDMI_ACR_CTS_32_MASK */
+	REG_UPDATE(HDMI_ACR_32_0, HDMI_ACR_CTS_32, audio_clock_info.cts_32khz);
+
+	/* HDMI_ACR_32_1__HDMI_ACR_N_32_MASK */
+	REG_UPDATE(HDMI_ACR_32_1, HDMI_ACR_N_32, audio_clock_info.n_32khz);
+
+	/* HDMI_ACR_44_0__HDMI_ACR_CTS_44_MASK */
+	REG_UPDATE(HDMI_ACR_44_0, HDMI_ACR_CTS_44, audio_clock_info.cts_44khz);
+
+	/* HDMI_ACR_44_1__HDMI_ACR_N_44_MASK */
+	REG_UPDATE(HDMI_ACR_44_1, HDMI_ACR_N_44, audio_clock_info.n_44khz);
+
+	/* HDMI_ACR_48_0__HDMI_ACR_CTS_48_MASK */
+	REG_UPDATE(HDMI_ACR_48_0, HDMI_ACR_CTS_48, audio_clock_info.cts_48khz);
+
+	/* HDMI_ACR_48_1__HDMI_ACR_N_48_MASK */
+	REG_UPDATE(HDMI_ACR_48_1, HDMI_ACR_N_48, audio_clock_info.n_48khz);
+
+	/* Video driver cannot know in advance which sample rate will
+	 * be used by HD Audio driver
+	 * HDMI_ACR_PACKET_CONTROL__HDMI_ACR_N_MULTIPLE field is
+	 * programmed below in interruppt callback
+	 */
+
+	/* AFMT_60958_0__AFMT_60958_CS_CHANNEL_NUMBER_L_MASK &
+	 * AFMT_60958_0__AFMT_60958_CS_CLOCK_ACCURACY_MASK
+	 */
+	REG_UPDATE_2(AFMT_60958_0,
+			AFMT_60958_CS_CHANNEL_NUMBER_L, 1,
+			AFMT_60958_CS_CLOCK_ACCURACY, 0);
+
+	/* AFMT_60958_1 AFMT_60958_CS_CHALNNEL_NUMBER_R */
+	REG_UPDATE(AFMT_60958_1, AFMT_60958_CS_CHANNEL_NUMBER_R, 2);
+
+	/* AFMT_60958_2 now keep this settings until
+	 * Programming guide comes out
+	 */
+	REG_UPDATE_6(AFMT_60958_2,
+			AFMT_60958_CS_CHANNEL_NUMBER_2, 3,
+			AFMT_60958_CS_CHANNEL_NUMBER_3, 4,
+			AFMT_60958_CS_CHANNEL_NUMBER_4, 5,
+			AFMT_60958_CS_CHANNEL_NUMBER_5, 6,
+			AFMT_60958_CS_CHANNEL_NUMBER_6, 7,
+			AFMT_60958_CS_CHANNEL_NUMBER_7, 8);
+}
+
+static void enc1_se_setup_dp_audio(
+	struct stream_encoder *enc)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	/* --- DP Audio packet configurations --- */
+
+	/* ATP Configuration */
+	REG_SET(DP_SEC_AUD_N, 0,
+			DP_SEC_AUD_N, DP_SEC_AUD_N__DP_SEC_AUD_N__DEFAULT);
+
+	/* Async/auto-calc timestamp mode */
+	REG_SET(DP_SEC_TIMESTAMP, 0, DP_SEC_TIMESTAMP_MODE,
+			DP_SEC_TIMESTAMP__DP_SEC_TIMESTAMP_MODE__AUTO_CALC);
+
+	/* --- The following are the registers
+	 *  copied from the SetupHDMI ---
+	 */
+
+	/* AFMT_AUDIO_PACKET_CONTROL */
+	REG_UPDATE(AFMT_AUDIO_PACKET_CONTROL, AFMT_60958_CS_UPDATE, 1);
+
+	/* AFMT_AUDIO_PACKET_CONTROL2 */
+	/* Program the ATP and AIP next */
+	REG_UPDATE_2(AFMT_AUDIO_PACKET_CONTROL2,
+			AFMT_AUDIO_LAYOUT_OVRD, 0,
+			AFMT_60958_OSF_OVRD, 0);
+
+	/* AFMT_INFOFRAME_CONTROL0 */
+	REG_UPDATE(AFMT_INFOFRAME_CONTROL0, AFMT_AUDIO_INFO_UPDATE, 1);
+
+	/* AFMT_60958_0__AFMT_60958_CS_CLOCK_ACCURACY_MASK */
+	REG_UPDATE(AFMT_60958_0, AFMT_60958_CS_CLOCK_ACCURACY, 0);
+}
+
+void enc1_se_enable_audio_clock(
+	struct stream_encoder *enc,
+	bool enable)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (REG(AFMT_CNTL) == 0)
+		return;   /* DCE8/10 does not have this register */
+
+	REG_UPDATE(AFMT_CNTL, AFMT_AUDIO_CLOCK_EN, !!enable);
+
+	/* wait for AFMT clock to turn on,
+	 * expectation: this should complete in 1-2 reads
+	 *
+	 * REG_WAIT(AFMT_CNTL, AFMT_AUDIO_CLOCK_ON, !!enable, 1, 10);
+	 *
+	 * TODO: wait for clock_on does not work well. May need HW
+	 * program sequence. But audio seems work normally even without wait
+	 * for clock_on status change
+	 */
+}
+
+void enc1_se_enable_dp_audio(
+	struct stream_encoder *enc)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	/* Enable Audio packets */
+	REG_UPDATE(DP_SEC_CNTL, DP_SEC_ASP_ENABLE, 1);
+
+	/* Program the ATP and AIP next */
+	REG_UPDATE_2(DP_SEC_CNTL,
+			DP_SEC_ATP_ENABLE, 1,
+			DP_SEC_AIP_ENABLE, 1);
+
+	/* Program STREAM_ENABLE after all the other enables. */
+	REG_UPDATE(DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, 1);
+}
+
+static void enc1_se_disable_dp_audio(
+	struct stream_encoder *enc)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+	uint32_t value = 0;
+
+	/* Disable Audio packets */
+	REG_UPDATE_5(DP_SEC_CNTL,
+			DP_SEC_ASP_ENABLE, 0,
+			DP_SEC_ATP_ENABLE, 0,
+			DP_SEC_AIP_ENABLE, 0,
+			DP_SEC_ACM_ENABLE, 0,
+			DP_SEC_STREAM_ENABLE, 0);
+
+	/* This register shared with encoder info frame. Therefore we need to
+	 * keep master enabled if at least on of the fields is not 0
+	 */
+	value = REG_READ(DP_SEC_CNTL);
+	if (value != 0)
+		REG_UPDATE(DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, 1);
+
+}
+
+void enc1_se_audio_mute_control(
+	struct stream_encoder *enc,
+	bool mute)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	REG_UPDATE(AFMT_AUDIO_PACKET_CONTROL, AFMT_AUDIO_SAMPLE_SEND, !mute);
+}
+
+void enc1_se_dp_audio_setup(
+	struct stream_encoder *enc,
+	unsigned int az_inst,
+	struct audio_info *info)
+{
+	enc1_se_audio_setup(enc, az_inst, info);
+}
+
+void enc1_se_dp_audio_enable(
+	struct stream_encoder *enc)
+{
+	enc1_se_enable_audio_clock(enc, true);
+	enc1_se_setup_dp_audio(enc);
+	enc1_se_enable_dp_audio(enc);
+}
+
+void enc1_se_dp_audio_disable(
+	struct stream_encoder *enc)
+{
+	enc1_se_disable_dp_audio(enc);
+	enc1_se_enable_audio_clock(enc, false);
+}
+
+void enc1_se_hdmi_audio_setup(
+	struct stream_encoder *enc,
+	unsigned int az_inst,
+	struct audio_info *info,
+	struct audio_crtc_info *audio_crtc_info)
+{
+	enc1_se_enable_audio_clock(enc, true);
+	enc1_se_setup_hdmi_audio(enc, audio_crtc_info);
+	enc1_se_audio_setup(enc, az_inst, info);
+}
+
+void enc1_se_hdmi_audio_disable(
+	struct stream_encoder *enc)
+{
+	if (enc->afmt && enc->afmt->funcs->afmt_powerdown)
+		enc->afmt->funcs->afmt_powerdown(enc->afmt);
+
+	enc1_se_enable_audio_clock(enc, false);
+}
+
+
+void enc1_setup_stereo_sync(
+	struct stream_encoder *enc,
+	int tg_inst, bool enable)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+	REG_UPDATE(DIG_FE_CNTL, DIG_STEREOSYNC_SELECT, tg_inst);
+	REG_UPDATE(DIG_FE_CNTL, DIG_STEREOSYNC_GATE_EN, !enable);
+}
+
+void enc1_dig_connect_to_otg(
+	struct stream_encoder *enc,
+	int tg_inst)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	REG_UPDATE(DIG_FE_CNTL, DIG_SOURCE_SELECT, tg_inst);
+}
+
+unsigned int enc1_dig_source_otg(
+	struct stream_encoder *enc)
+{
+	uint32_t tg_inst = 0;
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	REG_GET(DIG_FE_CNTL, DIG_SOURCE_SELECT, &tg_inst);
+
+	return tg_inst;
+}
+
+bool enc1_stream_encoder_dp_get_pixel_format(
+	struct stream_encoder *enc,
+	enum dc_pixel_encoding *encoding,
+	enum dc_color_depth *depth)
+{
+	uint32_t hw_encoding = 0;
+	uint32_t hw_depth = 0;
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (enc == NULL ||
+		encoding == NULL ||
+		depth == NULL)
+		return false;
+
+	REG_GET_2(DP_PIXEL_FORMAT,
+		DP_PIXEL_ENCODING, &hw_encoding,
+		DP_COMPONENT_DEPTH, &hw_depth);
+
+	switch (hw_depth) {
+	case DP_COMPONENT_PIXEL_DEPTH_6BPC:
+		*depth = COLOR_DEPTH_666;
+		break;
+	case DP_COMPONENT_PIXEL_DEPTH_8BPC:
+		*depth = COLOR_DEPTH_888;
+		break;
+	case DP_COMPONENT_PIXEL_DEPTH_10BPC:
+		*depth = COLOR_DEPTH_101010;
+		break;
+	case DP_COMPONENT_PIXEL_DEPTH_12BPC:
+		*depth = COLOR_DEPTH_121212;
+		break;
+	case DP_COMPONENT_PIXEL_DEPTH_16BPC:
+		*depth = COLOR_DEPTH_161616;
+		break;
+	default:
+		*depth = COLOR_DEPTH_UNDEFINED;
+		break;
+	}
+
+	switch (hw_encoding) {
+	case DP_PIXEL_ENCODING_TYPE_RGB444:
+		*encoding = PIXEL_ENCODING_RGB;
+		break;
+	case DP_PIXEL_ENCODING_TYPE_YCBCR422:
+		*encoding = PIXEL_ENCODING_YCBCR422;
+		break;
+	case DP_PIXEL_ENCODING_TYPE_YCBCR444:
+	case DP_PIXEL_ENCODING_TYPE_Y_ONLY:
+		*encoding = PIXEL_ENCODING_YCBCR444;
+		break;
+	case DP_PIXEL_ENCODING_TYPE_YCBCR420:
+		*encoding = PIXEL_ENCODING_YCBCR420;
+		break;
+	default:
+		*encoding = PIXEL_ENCODING_UNDEFINED;
+		break;
+	}
+	return true;
+}
+
+static const struct stream_encoder_funcs dcn10_str_enc_funcs = {
+	.dp_set_stream_attribute =
+		enc1_stream_encoder_dp_set_stream_attribute,
+	.hdmi_set_stream_attribute =
+		enc1_stream_encoder_hdmi_set_stream_attribute,
+	.dvi_set_stream_attribute =
+		enc1_stream_encoder_dvi_set_stream_attribute,
+	.set_throttled_vcp_size =
+		enc1_stream_encoder_set_throttled_vcp_size,
+	.update_hdmi_info_packets =
+		enc1_stream_encoder_update_hdmi_info_packets,
+	.stop_hdmi_info_packets =
+		enc1_stream_encoder_stop_hdmi_info_packets,
+	.update_dp_info_packets =
+		enc1_stream_encoder_update_dp_info_packets,
+	.send_immediate_sdp_message =
+		enc1_stream_encoder_send_immediate_sdp_message,
+	.stop_dp_info_packets =
+		enc1_stream_encoder_stop_dp_info_packets,
+	.dp_blank =
+		enc1_stream_encoder_dp_blank,
+	.dp_unblank =
+		enc1_stream_encoder_dp_unblank,
+	.audio_mute_control = enc1_se_audio_mute_control,
+
+	.dp_audio_setup = enc1_se_dp_audio_setup,
+	.dp_audio_enable = enc1_se_dp_audio_enable,
+	.dp_audio_disable = enc1_se_dp_audio_disable,
+
+	.hdmi_audio_setup = enc1_se_hdmi_audio_setup,
+	.hdmi_audio_disable = enc1_se_hdmi_audio_disable,
+	.setup_stereo_sync  = enc1_setup_stereo_sync,
+	.set_avmute = enc1_stream_encoder_set_avmute,
+	.dig_connect_to_otg  = enc1_dig_connect_to_otg,
+	.hdmi_reset_stream_attribute = enc1_reset_hdmi_stream_attribute,
+	.dig_source_otg = enc1_dig_source_otg,
+
+	.dp_get_pixel_format  = enc1_stream_encoder_dp_get_pixel_format,
+};
+
+void dcn10_stream_encoder_construct(
+	struct dcn10_stream_encoder *enc1,
+	struct dc_context *ctx,
+	struct dc_bios *bp,
+	enum engine_id eng_id,
+	const struct dcn10_stream_enc_registers *regs,
+	const struct dcn10_stream_encoder_shift *se_shift,
+	const struct dcn10_stream_encoder_mask *se_mask)
+{
+	enc1->base.funcs = &dcn10_str_enc_funcs;
+	enc1->base.ctx = ctx;
+	enc1->base.id = eng_id;
+	enc1->base.bp = bp;
+	enc1->regs = regs;
+	enc1->se_shift = se_shift;
+	enc1->se_mask = se_mask;
+	enc1->base.stream_enc_inst = eng_id - ENGINE_ID_DIGA;
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn10/dcn10_stream_encoder.h b/drivers/gpu/drm/amd/display/dc/dio/dcn10/dcn10_stream_encoder.h
new file mode 100644
index 000000000000..54a6a4ebd636
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn10/dcn10_stream_encoder.h
@@ -0,0 +1,749 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_STREAM_ENCODER_DCN10_H__
+#define __DC_STREAM_ENCODER_DCN10_H__
+
+#include "stream_encoder.h"
+
+#define DCN10STRENC_FROM_STRENC(stream_encoder)\
+	container_of(stream_encoder, struct dcn10_stream_encoder, base)
+
+#define SE_COMMON_DCN_REG_LIST(id) \
+	SRI(AFMT_CNTL, DIG, id), \
+	SRI(AFMT_GENERIC_0, DIG, id), \
+	SRI(AFMT_GENERIC_1, DIG, id), \
+	SRI(AFMT_GENERIC_2, DIG, id), \
+	SRI(AFMT_GENERIC_3, DIG, id), \
+	SRI(AFMT_GENERIC_4, DIG, id), \
+	SRI(AFMT_GENERIC_5, DIG, id), \
+	SRI(AFMT_GENERIC_6, DIG, id), \
+	SRI(AFMT_GENERIC_7, DIG, id), \
+	SRI(AFMT_GENERIC_HDR, DIG, id), \
+	SRI(AFMT_INFOFRAME_CONTROL0, DIG, id), \
+	SRI(AFMT_VBI_PACKET_CONTROL, DIG, id), \
+	SRI(AFMT_VBI_PACKET_CONTROL1, DIG, id), \
+	SRI(AFMT_AUDIO_PACKET_CONTROL, DIG, id), \
+	SRI(AFMT_AUDIO_PACKET_CONTROL2, DIG, id), \
+	SRI(AFMT_AUDIO_SRC_CONTROL, DIG, id), \
+	SRI(AFMT_60958_0, DIG, id), \
+	SRI(AFMT_60958_1, DIG, id), \
+	SRI(AFMT_60958_2, DIG, id), \
+	SRI(DIG_FE_CNTL, DIG, id), \
+	SRI(DIG_FIFO_STATUS, DIG, id), \
+	SRI(HDMI_CONTROL, DIG, id), \
+	SRI(HDMI_DB_CONTROL, DIG, id), \
+	SRI(HDMI_GC, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL0, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL1, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL2, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL3, DIG, id), \
+	SRI(HDMI_INFOFRAME_CONTROL0, DIG, id), \
+	SRI(HDMI_INFOFRAME_CONTROL1, DIG, id), \
+	SRI(HDMI_VBI_PACKET_CONTROL, DIG, id), \
+	SRI(HDMI_AUDIO_PACKET_CONTROL, DIG, id),\
+	SRI(HDMI_ACR_PACKET_CONTROL, DIG, id),\
+	SRI(HDMI_ACR_32_0, DIG, id),\
+	SRI(HDMI_ACR_32_1, DIG, id),\
+	SRI(HDMI_ACR_44_0, DIG, id),\
+	SRI(HDMI_ACR_44_1, DIG, id),\
+	SRI(HDMI_ACR_48_0, DIG, id),\
+	SRI(HDMI_ACR_48_1, DIG, id),\
+	SRI(DP_DB_CNTL, DP, id), \
+	SRI(DP_MSA_MISC, DP, id), \
+	SRI(DP_MSA_VBID_MISC, DP, id), \
+	SRI(DP_MSA_COLORIMETRY, DP, id), \
+	SRI(DP_MSA_TIMING_PARAM1, DP, id), \
+	SRI(DP_MSA_TIMING_PARAM2, DP, id), \
+	SRI(DP_MSA_TIMING_PARAM3, DP, id), \
+	SRI(DP_MSA_TIMING_PARAM4, DP, id), \
+	SRI(DP_MSE_RATE_CNTL, DP, id), \
+	SRI(DP_MSE_RATE_UPDATE, DP, id), \
+	SRI(DP_PIXEL_FORMAT, DP, id), \
+	SRI(DP_SEC_CNTL, DP, id), \
+	SRI(DP_SEC_CNTL1, DP, id), \
+	SRI(DP_SEC_CNTL2, DP, id), \
+	SRI(DP_SEC_CNTL5, DP, id), \
+	SRI(DP_SEC_CNTL6, DP, id), \
+	SRI(DP_STEER_FIFO, DP, id), \
+	SRI(DP_VID_M, DP, id), \
+	SRI(DP_VID_N, DP, id), \
+	SRI(DP_VID_STREAM_CNTL, DP, id), \
+	SRI(DP_VID_TIMING, DP, id), \
+	SRI(DP_SEC_AUD_N, DP, id), \
+	SRI(DP_SEC_AUD_N_READBACK, DP, id), \
+	SRI(DP_SEC_AUD_M_READBACK, DP, id), \
+	SRI(DP_SEC_TIMESTAMP, DP, id), \
+	SRI(DIG_CLOCK_PATTERN, DIG, id)
+
+#define SE_DCN_REG_LIST(id)\
+	SE_COMMON_DCN_REG_LIST(id)
+
+
+struct dcn10_stream_enc_registers {
+	uint32_t AFMT_CNTL;
+	uint32_t AFMT_AVI_INFO0;
+	uint32_t AFMT_AVI_INFO1;
+	uint32_t AFMT_AVI_INFO2;
+	uint32_t AFMT_AVI_INFO3;
+	uint32_t AFMT_GENERIC_0;
+	uint32_t AFMT_GENERIC_1;
+	uint32_t AFMT_GENERIC_2;
+	uint32_t AFMT_GENERIC_3;
+	uint32_t AFMT_GENERIC_4;
+	uint32_t AFMT_GENERIC_5;
+	uint32_t AFMT_GENERIC_6;
+	uint32_t AFMT_GENERIC_7;
+	uint32_t AFMT_GENERIC_HDR;
+	uint32_t AFMT_INFOFRAME_CONTROL0;
+	uint32_t AFMT_VBI_PACKET_CONTROL;
+	uint32_t AFMT_VBI_PACKET_CONTROL1;
+	uint32_t AFMT_AUDIO_PACKET_CONTROL;
+	uint32_t AFMT_AUDIO_PACKET_CONTROL2;
+	uint32_t AFMT_AUDIO_SRC_CONTROL;
+	uint32_t AFMT_60958_0;
+	uint32_t AFMT_60958_1;
+	uint32_t AFMT_60958_2;
+	uint32_t DIG_FE_CNTL;
+	uint32_t DIG_FIFO_STATUS;
+	uint32_t DP_MSE_RATE_CNTL;
+	uint32_t DP_MSE_RATE_UPDATE;
+	uint32_t DP_PIXEL_FORMAT;
+	uint32_t DP_SEC_CNTL;
+	uint32_t DP_SEC_CNTL1;
+	uint32_t DP_SEC_CNTL2;
+	uint32_t DP_SEC_CNTL5;
+	uint32_t DP_SEC_CNTL6;
+	uint32_t DP_STEER_FIFO;
+	uint32_t DP_VID_M;
+	uint32_t DP_VID_N;
+	uint32_t DP_VID_STREAM_CNTL;
+	uint32_t DP_VID_TIMING;
+	uint32_t DP_SEC_AUD_N;
+	uint32_t DP_SEC_AUD_N_READBACK;
+	uint32_t DP_SEC_AUD_M_READBACK;
+	uint32_t DP_SEC_TIMESTAMP;
+	uint32_t HDMI_CONTROL;
+	uint32_t HDMI_GC;
+	uint32_t HDMI_GENERIC_PACKET_CONTROL0;
+	uint32_t HDMI_GENERIC_PACKET_CONTROL1;
+	uint32_t HDMI_GENERIC_PACKET_CONTROL2;
+	uint32_t HDMI_GENERIC_PACKET_CONTROL3;
+	uint32_t HDMI_GENERIC_PACKET_CONTROL4;
+	uint32_t HDMI_GENERIC_PACKET_CONTROL5;
+	uint32_t HDMI_INFOFRAME_CONTROL0;
+	uint32_t HDMI_INFOFRAME_CONTROL1;
+	uint32_t HDMI_VBI_PACKET_CONTROL;
+	uint32_t HDMI_AUDIO_PACKET_CONTROL;
+	uint32_t HDMI_ACR_PACKET_CONTROL;
+	uint32_t HDMI_ACR_32_0;
+	uint32_t HDMI_ACR_32_1;
+	uint32_t HDMI_ACR_44_0;
+	uint32_t HDMI_ACR_44_1;
+	uint32_t HDMI_ACR_48_0;
+	uint32_t HDMI_ACR_48_1;
+	uint32_t DP_DB_CNTL;
+	uint32_t DP_MSA_MISC;
+	uint32_t DP_MSA_VBID_MISC;
+	uint32_t DP_MSA_COLORIMETRY;
+	uint32_t DP_MSA_TIMING_PARAM1;
+	uint32_t DP_MSA_TIMING_PARAM2;
+	uint32_t DP_MSA_TIMING_PARAM3;
+	uint32_t DP_MSA_TIMING_PARAM4;
+	uint32_t HDMI_DB_CONTROL;
+	uint32_t DP_DSC_CNTL;
+	uint32_t DP_DSC_BYTES_PER_PIXEL;
+	uint32_t DME_CONTROL;
+	uint32_t DP_SEC_METADATA_TRANSMISSION;
+	uint32_t HDMI_METADATA_PACKET_CONTROL;
+	uint32_t DP_SEC_FRAMING4;
+	uint32_t DP_GSP11_CNTL;
+	uint32_t HDMI_GENERIC_PACKET_CONTROL6;
+	uint32_t HDMI_GENERIC_PACKET_CONTROL7;
+	uint32_t HDMI_GENERIC_PACKET_CONTROL8;
+	uint32_t HDMI_GENERIC_PACKET_CONTROL9;
+	uint32_t HDMI_GENERIC_PACKET_CONTROL10;
+	uint32_t DIG_CLOCK_PATTERN;
+	uint32_t DIG_FIFO_CTRL0;
+	uint32_t DIG_FE_CLK_CNTL;
+	uint32_t DIG_FE_EN_CNTL;
+	uint32_t STREAM_MAPPER_CONTROL;
+};
+
+
+#define SE_SF(reg_name, field_name, post_fix)\
+	.field_name = reg_name ## __ ## field_name ## post_fix
+
+#define SE_COMMON_MASK_SH_LIST_SOC(mask_sh)\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_INDEX, mask_sh),\
+	SE_SF(DIG0_AFMT_GENERIC_HDR, AFMT_GENERIC_HB0, mask_sh),\
+	SE_SF(DIG0_AFMT_GENERIC_HDR, AFMT_GENERIC_HB1, mask_sh),\
+	SE_SF(DIG0_AFMT_GENERIC_HDR, AFMT_GENERIC_HB2, mask_sh),\
+	SE_SF(DIG0_AFMT_GENERIC_HDR, AFMT_GENERIC_HB3, mask_sh),\
+	SE_SF(DP0_DP_PIXEL_FORMAT, DP_PIXEL_ENCODING, mask_sh),\
+	SE_SF(DP0_DP_PIXEL_FORMAT, DP_COMPONENT_DEPTH, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_PACKET_GEN_VERSION, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_KEEPOUT_MODE, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_DATA_SCRAMBLE_EN, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_NO_EXTRA_NULL_PACKET_FILLED, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_GC_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_GC_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_NULL_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_ACP_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, mask_sh),\
+	SE_SF(DIG0_AFMT_INFOFRAME_CONTROL0, AFMT_AUDIO_INFO_UPDATE, mask_sh),\
+	SE_SF(DIG0_HDMI_INFOFRAME_CONTROL1, HDMI_AUDIO_INFO_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GC, HDMI_GC_AVMUTE, mask_sh),\
+	SE_SF(DP0_DP_MSE_RATE_CNTL, DP_MSE_RATE_X, mask_sh),\
+	SE_SF(DP0_DP_MSE_RATE_CNTL, DP_MSE_RATE_Y, mask_sh),\
+	SE_SF(DP0_DP_MSE_RATE_UPDATE, DP_MSE_RATE_UPDATE_PENDING, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP0_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP1_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP2_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP3_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_MPG_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP4_SEND, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP4_SEND_PENDING, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL4, DP_SEC_GSP4_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP4_SEND_ANY_LINE, mask_sh),\
+	SE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_DIS_DEFER, mask_sh),\
+	SE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_STATUS, mask_sh),\
+	SE_SF(DP0_DP_STEER_FIFO, DP_STEER_FIFO_RESET, mask_sh),\
+	SE_SF(DP0_DP_VID_TIMING, DP_VID_M_N_GEN_EN, mask_sh),\
+	SE_SF(DP0_DP_VID_N, DP_VID_N, mask_sh),\
+	SE_SF(DP0_DP_VID_M, DP_VID_M, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_START, mask_sh),\
+	SE_SF(DIG0_AFMT_AUDIO_SRC_CONTROL, AFMT_AUDIO_SRC_SELECT, mask_sh),\
+	SE_SF(DIG0_AFMT_AUDIO_PACKET_CONTROL2, AFMT_AUDIO_CHANNEL_ENABLE, mask_sh),\
+	SE_SF(DIG0_HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_PACKETS_PER_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_DELAY_EN, mask_sh),\
+	SE_SF(DIG0_AFMT_AUDIO_PACKET_CONTROL, AFMT_60958_CS_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_AUDIO_PACKET_CONTROL2, AFMT_AUDIO_LAYOUT_OVRD, mask_sh),\
+	SE_SF(DIG0_AFMT_AUDIO_PACKET_CONTROL2, AFMT_60958_OSF_OVRD, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUTO_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_PACKET_CONTROL, HDMI_ACR_SOURCE, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUDIO_PRIORITY, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_32_0, HDMI_ACR_CTS_32, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_32_1, HDMI_ACR_N_32, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_44_0, HDMI_ACR_CTS_44, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_44_1, HDMI_ACR_N_44, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_48_0, HDMI_ACR_CTS_48, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_48_1, HDMI_ACR_N_48, mask_sh),\
+	SE_SF(DIG0_AFMT_60958_0, AFMT_60958_CS_CHANNEL_NUMBER_L, mask_sh),\
+	SE_SF(DIG0_AFMT_60958_0, AFMT_60958_CS_CLOCK_ACCURACY, mask_sh),\
+	SE_SF(DIG0_AFMT_60958_1, AFMT_60958_CS_CHANNEL_NUMBER_R, mask_sh),\
+	SE_SF(DIG0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_2, mask_sh),\
+	SE_SF(DIG0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_3, mask_sh),\
+	SE_SF(DIG0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_4, mask_sh),\
+	SE_SF(DIG0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_5, mask_sh),\
+	SE_SF(DIG0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_6, mask_sh),\
+	SE_SF(DIG0_AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_7, mask_sh),\
+	SE_SF(DP0_DP_SEC_AUD_N, DP_SEC_AUD_N, mask_sh),\
+	SE_SF(DP0_DP_SEC_AUD_N_READBACK, DP_SEC_AUD_N_READBACK, mask_sh),\
+	SE_SF(DP0_DP_SEC_AUD_M_READBACK, DP_SEC_AUD_M_READBACK, mask_sh),\
+	SE_SF(DP0_DP_SEC_TIMESTAMP, DP_SEC_TIMESTAMP_MODE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_ASP_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_ATP_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_AIP_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_ACM_ENABLE, mask_sh),\
+	SE_SF(DIG0_AFMT_AUDIO_PACKET_CONTROL, AFMT_AUDIO_SAMPLE_SEND, mask_sh),\
+	SE_SF(DIG0_AFMT_CNTL, AFMT_AUDIO_CLOCK_EN, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_CLOCK_CHANNEL_RATE, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, TMDS_PIXEL_ENCODING, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, TMDS_COLOR_FORMAT, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_STEREOSYNC_SELECT, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_STEREOSYNC_GATE_EN, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_STATUS, DIG_FIFO_LEVEL_ERROR, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_STATUS, DIG_FIFO_USE_OVERWRITE_LEVEL, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_STATUS, DIG_FIFO_OVERWRITE_LEVEL, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_STATUS, DIG_FIFO_ERROR_ACK, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_STATUS, DIG_FIFO_CAL_AVERAGE_LEVEL, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_STATUS, DIG_FIFO_MAXIMUM_LEVEL, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_STATUS, DIG_FIFO_MINIMUM_LEVEL, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_STATUS, DIG_FIFO_READ_CLOCK_SRC, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_STATUS, DIG_FIFO_CALIBRATED, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_STATUS, DIG_FIFO_FORCE_RECAL_AVERAGE, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_STATUS, DIG_FIFO_FORCE_RECOMP_MINMAX, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_LOCK_STATUS, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_CONFLICT, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_CONFLICT_CLR, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC0_FRAME_UPDATE_PENDING, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC1_FRAME_UPDATE_PENDING, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC2_FRAME_UPDATE_PENDING, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC3_FRAME_UPDATE_PENDING, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC4_FRAME_UPDATE_PENDING, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC4_IMMEDIATE_UPDATE_PENDING, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC5_FRAME_UPDATE_PENDING, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC6_FRAME_UPDATE_PENDING, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC7_FRAME_UPDATE_PENDING, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC0_FRAME_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC1_FRAME_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC2_FRAME_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC3_FRAME_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC4_FRAME_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC0_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC1_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC2_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC3_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC4_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC5_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC6_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC7_IMMEDIATE_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC5_FRAME_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC6_FRAME_UPDATE, mask_sh),\
+	SE_SF(DIG0_AFMT_VBI_PACKET_CONTROL1, AFMT_GENERIC7_FRAME_UPDATE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP4_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP5_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP6_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP7_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP7_PPS, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP7_SEND, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL6, DP_SEC_GSP7_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_DB_CNTL, DP_DB_DISABLE, mask_sh),\
+	SE_SF(DP0_DP_MSA_COLORIMETRY, DP_MSA_MISC0, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM1, DP_MSA_HTOTAL, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM1, DP_MSA_VTOTAL, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM2, DP_MSA_HSTART, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM2, DP_MSA_VSTART, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_HSYNCWIDTH, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_HSYNCPOLARITY, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_VSYNCWIDTH, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_VSYNCPOLARITY, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM4, DP_MSA_HWIDTH, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM4, DP_MSA_VHEIGHT, mask_sh),\
+	SE_SF(DIG0_HDMI_DB_CONTROL, HDMI_DB_DISABLE, mask_sh),\
+	SE_SF(DP0_DP_VID_TIMING, DP_VID_N_MUL, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_SOURCE_SELECT, mask_sh),\
+	SE_SF(DIG0_DIG_CLOCK_PATTERN, DIG_CLOCK_PATTERN, mask_sh)
+
+#define SE_COMMON_MASK_SH_LIST_DCN10(mask_sh)\
+	SE_COMMON_MASK_SH_LIST_SOC(mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC0_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC0_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC0_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC1_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC1_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC1_LINE, mask_sh)
+
+
+#define SE_REG_FIELD_LIST_DCN1_0(type) \
+	type AFMT_GENERIC_INDEX;\
+	type AFMT_GENERIC_HB0;\
+	type AFMT_GENERIC_HB1;\
+	type AFMT_GENERIC_HB2;\
+	type AFMT_GENERIC_HB3;\
+	type AFMT_GENERIC_LOCK_STATUS;\
+	type AFMT_GENERIC_CONFLICT;\
+	type AFMT_GENERIC_CONFLICT_CLR;\
+	type AFMT_GENERIC0_FRAME_UPDATE_PENDING;\
+	type AFMT_GENERIC1_FRAME_UPDATE_PENDING;\
+	type AFMT_GENERIC2_FRAME_UPDATE_PENDING;\
+	type AFMT_GENERIC3_FRAME_UPDATE_PENDING;\
+	type AFMT_GENERIC4_FRAME_UPDATE_PENDING;\
+	type AFMT_GENERIC4_IMMEDIATE_UPDATE_PENDING;\
+	type AFMT_GENERIC5_FRAME_UPDATE_PENDING;\
+	type AFMT_GENERIC6_FRAME_UPDATE_PENDING;\
+	type AFMT_GENERIC7_FRAME_UPDATE_PENDING;\
+	type AFMT_GENERIC0_FRAME_UPDATE;\
+	type AFMT_GENERIC1_FRAME_UPDATE;\
+	type AFMT_GENERIC2_FRAME_UPDATE;\
+	type AFMT_GENERIC3_FRAME_UPDATE;\
+	type AFMT_GENERIC4_FRAME_UPDATE;\
+	type AFMT_GENERIC0_IMMEDIATE_UPDATE;\
+	type AFMT_GENERIC1_IMMEDIATE_UPDATE;\
+	type AFMT_GENERIC2_IMMEDIATE_UPDATE;\
+	type AFMT_GENERIC3_IMMEDIATE_UPDATE;\
+	type AFMT_GENERIC4_IMMEDIATE_UPDATE;\
+	type AFMT_GENERIC5_IMMEDIATE_UPDATE;\
+	type AFMT_GENERIC6_IMMEDIATE_UPDATE;\
+	type AFMT_GENERIC7_IMMEDIATE_UPDATE;\
+	type AFMT_GENERIC5_FRAME_UPDATE;\
+	type AFMT_GENERIC6_FRAME_UPDATE;\
+	type AFMT_GENERIC7_FRAME_UPDATE;\
+	type HDMI_GENERIC0_CONT;\
+	type HDMI_GENERIC0_SEND;\
+	type HDMI_GENERIC0_LINE;\
+	type HDMI_GENERIC1_CONT;\
+	type HDMI_GENERIC1_SEND;\
+	type HDMI_GENERIC1_LINE;\
+	type HDMI_GENERIC2_CONT;\
+	type HDMI_GENERIC2_SEND;\
+	type HDMI_GENERIC2_LINE;\
+	type HDMI_GENERIC3_CONT;\
+	type HDMI_GENERIC3_SEND;\
+	type HDMI_GENERIC3_LINE;\
+	type HDMI_GENERIC4_CONT;\
+	type HDMI_GENERIC4_SEND;\
+	type HDMI_GENERIC4_LINE;\
+	type HDMI_GENERIC5_CONT;\
+	type HDMI_GENERIC5_SEND;\
+	type HDMI_GENERIC5_LINE;\
+	type HDMI_GENERIC6_CONT;\
+	type HDMI_GENERIC6_SEND;\
+	type HDMI_GENERIC6_LINE;\
+	type HDMI_GENERIC7_CONT;\
+	type HDMI_GENERIC7_SEND;\
+	type HDMI_GENERIC7_LINE;\
+	type DP_PIXEL_ENCODING;\
+	type DP_COMPONENT_DEPTH;\
+	type HDMI_PACKET_GEN_VERSION;\
+	type HDMI_KEEPOUT_MODE;\
+	type HDMI_DEEP_COLOR_ENABLE;\
+	type HDMI_CLOCK_CHANNEL_RATE;\
+	type HDMI_DEEP_COLOR_DEPTH;\
+	type HDMI_GC_CONT;\
+	type HDMI_GC_SEND;\
+	type HDMI_NULL_SEND;\
+	type HDMI_DATA_SCRAMBLE_EN;\
+	type HDMI_NO_EXTRA_NULL_PACKET_FILLED;\
+	type HDMI_AUDIO_INFO_SEND;\
+	type AFMT_AUDIO_INFO_UPDATE;\
+	type HDMI_AUDIO_INFO_LINE;\
+	type HDMI_GC_AVMUTE;\
+	type DP_MSE_RATE_X;\
+	type DP_MSE_RATE_Y;\
+	type DP_MSE_RATE_UPDATE_PENDING;\
+	type DP_SEC_GSP0_ENABLE;\
+	type DP_SEC_STREAM_ENABLE;\
+	type DP_SEC_GSP1_ENABLE;\
+	type DP_SEC_GSP2_ENABLE;\
+	type DP_SEC_GSP3_ENABLE;\
+	type DP_SEC_GSP4_ENABLE;\
+	type DP_SEC_GSP5_ENABLE;\
+	type DP_SEC_GSP5_LINE_NUM;\
+	type DP_SEC_GSP5_LINE_REFERENCE;\
+	type DP_SEC_GSP6_ENABLE;\
+	type DP_SEC_GSP7_ENABLE;\
+	type DP_SEC_GSP7_PPS;\
+	type DP_SEC_GSP7_SEND;\
+	type DP_SEC_GSP4_SEND;\
+	type DP_SEC_GSP4_SEND_PENDING;\
+	type DP_SEC_GSP4_LINE_NUM;\
+	type DP_SEC_GSP4_SEND_ANY_LINE;\
+	type DP_SEC_MPG_ENABLE;\
+	type DP_VID_STREAM_DIS_DEFER;\
+	type DP_VID_STREAM_ENABLE;\
+	type DP_VID_STREAM_STATUS;\
+	type DP_STEER_FIFO_RESET;\
+	type DP_VID_M_N_GEN_EN;\
+	type DP_VID_N;\
+	type DP_VID_M;\
+	type DIG_START;\
+	type AFMT_AUDIO_SRC_SELECT;\
+	type AFMT_AUDIO_CHANNEL_ENABLE;\
+	type HDMI_AUDIO_PACKETS_PER_LINE;\
+	type HDMI_AUDIO_DELAY_EN;\
+	type AFMT_60958_CS_UPDATE;\
+	type AFMT_AUDIO_LAYOUT_OVRD;\
+	type AFMT_60958_OSF_OVRD;\
+	type HDMI_ACR_AUTO_SEND;\
+	type HDMI_ACR_SOURCE;\
+	type HDMI_ACR_AUDIO_PRIORITY;\
+	type HDMI_ACR_CTS_32;\
+	type HDMI_ACR_N_32;\
+	type HDMI_ACR_CTS_44;\
+	type HDMI_ACR_N_44;\
+	type HDMI_ACR_CTS_48;\
+	type HDMI_ACR_N_48;\
+	type AFMT_60958_CS_CHANNEL_NUMBER_L;\
+	type AFMT_60958_CS_CLOCK_ACCURACY;\
+	type AFMT_60958_CS_CHANNEL_NUMBER_R;\
+	type AFMT_60958_CS_CHANNEL_NUMBER_2;\
+	type AFMT_60958_CS_CHANNEL_NUMBER_3;\
+	type AFMT_60958_CS_CHANNEL_NUMBER_4;\
+	type AFMT_60958_CS_CHANNEL_NUMBER_5;\
+	type AFMT_60958_CS_CHANNEL_NUMBER_6;\
+	type AFMT_60958_CS_CHANNEL_NUMBER_7;\
+	type DP_SEC_AUD_N;\
+	type DP_SEC_AUD_N_READBACK;\
+	type DP_SEC_AUD_M_READBACK;\
+	type DP_SEC_TIMESTAMP_MODE;\
+	type DP_SEC_ASP_ENABLE;\
+	type DP_SEC_ATP_ENABLE;\
+	type DP_SEC_AIP_ENABLE;\
+	type DP_SEC_ACM_ENABLE;\
+	type DP_SEC_GSP7_LINE_NUM;\
+	type AFMT_AUDIO_SAMPLE_SEND;\
+	type AFMT_AUDIO_CLOCK_EN;\
+	type TMDS_PIXEL_ENCODING;\
+	type TMDS_COLOR_FORMAT;\
+	type DIG_STEREOSYNC_SELECT;\
+	type DIG_STEREOSYNC_GATE_EN;\
+	type DP_DB_DISABLE;\
+	type DP_MSA_MISC0;\
+	type DP_MSA_HTOTAL;\
+	type DP_MSA_VTOTAL;\
+	type DP_MSA_HSTART;\
+	type DP_MSA_VSTART;\
+	type DP_MSA_HSYNCWIDTH;\
+	type DP_MSA_HSYNCPOLARITY;\
+	type DP_MSA_VSYNCWIDTH;\
+	type DP_MSA_VSYNCPOLARITY;\
+	type DP_MSA_HWIDTH;\
+	type DP_MSA_VHEIGHT;\
+	type HDMI_DB_DISABLE;\
+	type DP_VID_N_MUL;\
+	type DP_VID_M_DOUBLE_VALUE_EN;\
+	type DIG_SOURCE_SELECT;\
+	type DIG_FIFO_LEVEL_ERROR;\
+	type DIG_FIFO_USE_OVERWRITE_LEVEL;\
+	type DIG_FIFO_OVERWRITE_LEVEL;\
+	type DIG_FIFO_ERROR_ACK;\
+	type DIG_FIFO_CAL_AVERAGE_LEVEL;\
+	type DIG_FIFO_MAXIMUM_LEVEL;\
+	type DIG_FIFO_MINIMUM_LEVEL;\
+	type DIG_FIFO_READ_CLOCK_SRC;\
+	type DIG_FIFO_CALIBRATED;\
+	type DIG_FIFO_FORCE_RECAL_AVERAGE;\
+	type DIG_FIFO_FORCE_RECOMP_MINMAX;\
+	type DIG_CLOCK_PATTERN
+
+#define SE_REG_FIELD_LIST_DCN2_0(type) \
+	type DP_DSC_MODE;\
+	type DP_DSC_SLICE_WIDTH;\
+	type DP_DSC_BYTES_PER_PIXEL;\
+	type DP_VBID6_LINE_REFERENCE;\
+	type DP_VBID6_LINE_NUM;\
+	type METADATA_ENGINE_EN;\
+	type METADATA_HUBP_REQUESTOR_ID;\
+	type METADATA_STREAM_TYPE;\
+	type DP_SEC_METADATA_PACKET_ENABLE;\
+	type DP_SEC_METADATA_PACKET_LINE_REFERENCE;\
+	type DP_SEC_METADATA_PACKET_LINE;\
+	type HDMI_METADATA_PACKET_ENABLE;\
+	type HDMI_METADATA_PACKET_LINE_REFERENCE;\
+	type HDMI_METADATA_PACKET_LINE;\
+	type DOLBY_VISION_EN;\
+	type DP_PIXEL_COMBINE;\
+	type DP_SST_SDP_SPLITTING
+
+#define SE_REG_FIELD_LIST_DCN3_0(type) \
+	type HDMI_GENERIC8_CONT;\
+	type HDMI_GENERIC8_SEND;\
+	type HDMI_GENERIC8_LINE;\
+	type HDMI_GENERIC9_CONT;\
+	type HDMI_GENERIC9_SEND;\
+	type HDMI_GENERIC9_LINE;\
+	type HDMI_GENERIC10_CONT;\
+	type HDMI_GENERIC10_SEND;\
+	type HDMI_GENERIC10_LINE;\
+	type HDMI_GENERIC11_CONT;\
+	type HDMI_GENERIC11_SEND;\
+	type HDMI_GENERIC11_LINE;\
+	type HDMI_GENERIC12_CONT;\
+	type HDMI_GENERIC12_SEND;\
+	type HDMI_GENERIC12_LINE;\
+	type HDMI_GENERIC13_CONT;\
+	type HDMI_GENERIC13_SEND;\
+	type HDMI_GENERIC13_LINE;\
+	type HDMI_GENERIC14_CONT;\
+	type HDMI_GENERIC14_SEND;\
+	type HDMI_GENERIC14_LINE;\
+	type DP_SEC_GSP11_PPS;\
+	type DP_SEC_GSP11_ENABLE;\
+	type DP_SEC_GSP11_LINE_NUM
+
+#define SE_REG_FIELD_LIST_DCN3_1_COMMON(type) \
+	type DIG_FIFO_OUTPUT_PIXEL_MODE;\
+	type DP_PIXEL_PER_CYCLE_PROCESSING_MODE;\
+	type DIG_SYMCLK_FE_ON;\
+	type DIG_FIFO_READ_START_LEVEL;\
+	type DIG_FIFO_ENABLE;\
+	type DIG_FIFO_RESET;\
+	type DIG_FIFO_RESET_DONE;\
+	type PIXEL_ENCODING_TYPE;\
+	type UNCOMPRESSED_PIXEL_FORMAT;\
+	type UNCOMPRESSED_COMPONENT_DEPTH
+
+#define SE_REG_FIELD_LIST_DCN3_5_COMMON(type) \
+	type DIG_FE_CLK_EN;\
+	type DIG_FE_MODE;\
+	type DIG_FE_SOFT_RESET;\
+	type DIG_FE_ENABLE;\
+	type DIG_FE_SYMCLK_FE_G_CLOCK_ON;\
+	type DIG_FE_DISPCLK_G_CLOCK_ON;\
+	type DIG_FE_SYMCLK_FE_G_AFMT_CLOCK_ON;\
+	type DIG_FE_SYMCLK_FE_G_TMDS_CLOCK_ON;\
+	type DIG_FE_SOCCLK_G_AFMT_CLOCK_ON;\
+	type DIG_STREAM_LINK_TARGET
+
+#define SE_REG_FIELD_LIST_DCN4_01_COMMON(type) \
+	type COMPRESSED_PIXEL_FORMAT;\
+	type DP_VID_N_INTERVAL;\
+	type DIG_FIFO_OUTPUT_PIXEL_PER_CYCLE;\
+	type DP_STEER_FIFO_ENABLE
+struct dcn10_stream_encoder_shift {
+	SE_REG_FIELD_LIST_DCN1_0(uint8_t);
+	uint8_t HDMI_ACP_SEND;
+	SE_REG_FIELD_LIST_DCN2_0(uint8_t);
+	SE_REG_FIELD_LIST_DCN3_0(uint8_t);
+	SE_REG_FIELD_LIST_DCN3_1_COMMON(uint8_t);
+	SE_REG_FIELD_LIST_DCN3_5_COMMON(uint8_t);
+	SE_REG_FIELD_LIST_DCN4_01_COMMON(uint32_t);
+};
+
+struct dcn10_stream_encoder_mask {
+	SE_REG_FIELD_LIST_DCN1_0(uint32_t);
+	uint32_t HDMI_ACP_SEND;
+	SE_REG_FIELD_LIST_DCN2_0(uint32_t);
+	SE_REG_FIELD_LIST_DCN3_0(uint32_t);
+	SE_REG_FIELD_LIST_DCN3_1_COMMON(uint32_t);
+	SE_REG_FIELD_LIST_DCN3_5_COMMON(uint32_t);
+	SE_REG_FIELD_LIST_DCN4_01_COMMON(uint32_t);
+};
+
+struct dcn10_stream_encoder {
+	struct stream_encoder base;
+	const struct dcn10_stream_enc_registers *regs;
+	const struct dcn10_stream_encoder_shift *se_shift;
+	const struct dcn10_stream_encoder_mask *se_mask;
+};
+
+void dcn10_stream_encoder_construct(
+	struct dcn10_stream_encoder *enc1,
+	struct dc_context *ctx,
+	struct dc_bios *bp,
+	enum engine_id eng_id,
+	const struct dcn10_stream_enc_registers *regs,
+	const struct dcn10_stream_encoder_shift *se_shift,
+	const struct dcn10_stream_encoder_mask *se_mask);
+
+void enc1_update_generic_info_packet(
+	struct dcn10_stream_encoder *enc1,
+	uint32_t packet_index,
+	const struct dc_info_packet *info_packet);
+
+void enc1_stream_encoder_dp_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	enum dc_color_space output_color_space,
+	bool use_vsc_sdp_for_colorimetry,
+	uint32_t enable_sdp_splitting);
+
+void enc1_stream_encoder_hdmi_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	int actual_pix_clk_khz,
+	bool enable_audio);
+
+void enc1_stream_encoder_dvi_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	bool is_dual_link);
+
+void enc1_stream_encoder_set_throttled_vcp_size(
+	struct stream_encoder *enc,
+	struct fixed31_32 avg_time_slots_per_mtp);
+
+void enc1_stream_encoder_update_dp_info_packets(
+	struct stream_encoder *enc,
+	const struct encoder_info_frame *info_frame);
+
+void enc1_stream_encoder_send_immediate_sdp_message(
+	struct stream_encoder *enc,
+	const uint8_t *custom_sdp_message,
+				unsigned int sdp_message_size);
+
+void enc1_stream_encoder_stop_dp_info_packets(
+	struct stream_encoder *enc);
+
+void enc1_stream_encoder_dp_blank(
+	struct dc_link *link,
+	struct stream_encoder *enc);
+
+void enc1_stream_encoder_dp_unblank(
+	struct dc_link *link,
+	struct stream_encoder *enc,
+	const struct encoder_unblank_param *param);
+
+void enc1_setup_stereo_sync(
+	struct stream_encoder *enc,
+	int tg_inst, bool enable);
+
+void enc1_stream_encoder_set_avmute(
+	struct stream_encoder *enc,
+	bool enable);
+
+void enc1_se_audio_mute_control(
+	struct stream_encoder *enc,
+	bool mute);
+
+void enc1_se_dp_audio_setup(
+	struct stream_encoder *enc,
+	unsigned int az_inst,
+	struct audio_info *info);
+
+void enc1_se_dp_audio_enable(
+	struct stream_encoder *enc);
+
+void enc1_se_dp_audio_disable(
+	struct stream_encoder *enc);
+
+void enc1_se_hdmi_audio_setup(
+	struct stream_encoder *enc,
+	unsigned int az_inst,
+	struct audio_info *info,
+	struct audio_crtc_info *audio_crtc_info);
+
+void enc1_se_hdmi_audio_disable(
+	struct stream_encoder *enc);
+
+void enc1_dig_connect_to_otg(
+	struct stream_encoder *enc,
+	int tg_inst);
+
+unsigned int enc1_dig_source_otg(
+	struct stream_encoder *enc);
+
+void enc1_stream_encoder_set_stream_attribute_helper(
+	struct dcn10_stream_encoder *enc1,
+	struct dc_crtc_timing *crtc_timing);
+
+void enc1_se_enable_audio_clock(
+	struct stream_encoder *enc,
+	bool enable);
+
+void enc1_se_enable_dp_audio(
+	struct stream_encoder *enc);
+
+void get_audio_clock_info(
+	enum dc_color_depth color_depth,
+	uint32_t crtc_pixel_clock_100Hz,
+	uint32_t actual_pixel_clock_100Hz,
+	struct audio_clock_info *audio_clock_info);
+
+void enc1_reset_hdmi_stream_attribute(
+	struct stream_encoder *enc);
+
+bool enc1_stream_encoder_dp_get_pixel_format(
+	struct stream_encoder *enc,
+	enum dc_pixel_encoding *encoding,
+	enum dc_color_depth *depth);
+
+#endif /* __DC_STREAM_ENCODER_DCN10_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn20/dcn20_link_encoder.c b/drivers/gpu/drm/amd/display/dc/dio/dcn20/dcn20_link_encoder.c
new file mode 100644
index 000000000000..51a57dae1811
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn20/dcn20_link_encoder.c
@@ -0,0 +1,502 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+
+#include "core_types.h"
+#include "link_encoder.h"
+#include "dcn20_link_encoder.h"
+#include "stream_encoder.h"
+#include "dc_bios_types.h"
+
+#include "gpio_service_interface.h"
+
+#define CTX \
+	enc10->base.ctx
+#define DC_LOGGER \
+	enc10->base.ctx->logger
+
+#define REG(reg)\
+	(enc10->link_regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	enc10->link_shift->field_name, enc10->link_mask->field_name
+
+#define IND_REG(index) \
+	(enc10->link_regs->index)
+
+#ifndef MAX
+#define MAX(X, Y) ((X) > (Y) ? (X) : (Y))
+#endif
+#ifndef MIN
+#define MIN(X, Y) ((X) < (Y) ? (X) : (Y))
+#endif
+
+static struct mpll_cfg dcn2_mpll_cfg[] = {
+	// RBR
+	{
+		.hdmimode_enable = 1,
+		.ref_range = 3,
+		.ref_clk_mpllb_div = 2,
+		.mpllb_ssc_en = 1,
+		.mpllb_div5_clk_en = 1,
+		.mpllb_multiplier = 226,
+		.mpllb_fracn_en = 1,
+		.mpllb_fracn_quot = 39321,
+		.mpllb_fracn_rem = 3,
+		.mpllb_fracn_den = 5,
+		.mpllb_ssc_up_spread = 0,
+		.mpllb_ssc_peak = 38221,
+		.mpllb_ssc_stepsize = 49314,
+		.mpllb_div_clk_en = 0,
+		.mpllb_div_multiplier = 0,
+		.mpllb_hdmi_div = 0,
+		.mpllb_tx_clk_div = 2,
+		.tx_vboost_lvl = 4,
+		.mpllb_pmix_en = 1,
+		.mpllb_word_div2_en = 0,
+		.mpllb_ana_v2i = 2,
+		.mpllb_ana_freq_vco = 2,
+		.mpllb_ana_cp_int = 7,
+		.mpllb_ana_cp_prop = 18,
+		.hdmi_pixel_clk_div = 0,
+	},
+	// HBR
+	{
+		.hdmimode_enable = 1,
+		.ref_range = 3,
+		.ref_clk_mpllb_div = 2,
+		.mpllb_ssc_en = 1,
+		.mpllb_div5_clk_en = 1,
+		.mpllb_multiplier = 184,
+		.mpllb_fracn_en = 0,
+		.mpllb_fracn_quot = 0,
+		.mpllb_fracn_rem = 0,
+		.mpllb_fracn_den = 1,
+		.mpllb_ssc_up_spread = 0,
+		.mpllb_ssc_peak = 31850,
+		.mpllb_ssc_stepsize = 41095,
+		.mpllb_div_clk_en = 0,
+		.mpllb_div_multiplier = 0,
+		.mpllb_hdmi_div = 0,
+		.mpllb_tx_clk_div = 1,
+		.tx_vboost_lvl = 4,
+		.mpllb_pmix_en = 1,
+		.mpllb_word_div2_en = 0,
+		.mpllb_ana_v2i = 2,
+		.mpllb_ana_freq_vco = 3,
+		.mpllb_ana_cp_int = 7,
+		.mpllb_ana_cp_prop = 18,
+		.hdmi_pixel_clk_div = 0,
+	},
+	//HBR2
+	{
+		.hdmimode_enable = 1,
+		.ref_range = 3,
+		.ref_clk_mpllb_div = 2,
+		.mpllb_ssc_en = 1,
+		.mpllb_div5_clk_en = 1,
+		.mpllb_multiplier = 184,
+		.mpllb_fracn_en = 0,
+		.mpllb_fracn_quot = 0,
+		.mpllb_fracn_rem = 0,
+		.mpllb_fracn_den = 1,
+		.mpllb_ssc_up_spread = 0,
+		.mpllb_ssc_peak = 31850,
+		.mpllb_ssc_stepsize = 41095,
+		.mpllb_div_clk_en = 0,
+		.mpllb_div_multiplier = 0,
+		.mpllb_hdmi_div = 0,
+		.mpllb_tx_clk_div = 0,
+		.tx_vboost_lvl = 4,
+		.mpllb_pmix_en = 1,
+		.mpllb_word_div2_en = 0,
+		.mpllb_ana_v2i = 2,
+		.mpllb_ana_freq_vco = 3,
+		.mpllb_ana_cp_int = 7,
+		.mpllb_ana_cp_prop = 18,
+		.hdmi_pixel_clk_div = 0,
+	},
+	//HBR3
+	{
+		.hdmimode_enable = 1,
+		.ref_range = 3,
+		.ref_clk_mpllb_div = 2,
+		.mpllb_ssc_en = 1,
+		.mpllb_div5_clk_en = 1,
+		.mpllb_multiplier = 292,
+		.mpllb_fracn_en = 0,
+		.mpllb_fracn_quot = 0,
+		.mpllb_fracn_rem = 0,
+		.mpllb_fracn_den = 1,
+		.mpllb_ssc_up_spread = 0,
+		.mpllb_ssc_peak = 47776,
+		.mpllb_ssc_stepsize = 61642,
+		.mpllb_div_clk_en = 0,
+		.mpllb_div_multiplier = 0,
+		.mpllb_hdmi_div = 0,
+		.mpllb_tx_clk_div = 0,
+		.tx_vboost_lvl = 4,
+		.mpllb_pmix_en = 1,
+		.mpllb_word_div2_en = 0,
+		.mpllb_ana_v2i = 2,
+		.mpllb_ana_freq_vco = 0,
+		.mpllb_ana_cp_int = 7,
+		.mpllb_ana_cp_prop = 18,
+		.hdmi_pixel_clk_div = 0,
+	},
+};
+
+void enc2_fec_set_enable(struct link_encoder *enc, bool enable)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	DC_LOG_DSC("%s FEC at link encoder inst %d",
+			enable ? "Enabling" : "Disabling", enc->id.enum_id);
+	REG_UPDATE(DP_DPHY_CNTL, DPHY_FEC_EN, enable);
+}
+
+void enc2_fec_set_ready(struct link_encoder *enc, bool ready)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	REG_UPDATE(DP_DPHY_CNTL, DPHY_FEC_READY_SHADOW, ready);
+}
+
+bool enc2_fec_is_active(struct link_encoder *enc)
+{
+	uint32_t active = 0;
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	REG_GET(DP_DPHY_CNTL, DPHY_FEC_ACTIVE_STATUS, &active);
+
+	return (active != 0);
+}
+
+/* this function reads dsc related register fields to be logged later in dcn10_log_hw_state
+ * into a dcn_dsc_state struct.
+ */
+void link_enc2_read_state(struct link_encoder *enc, struct link_enc_state *s)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	REG_GET(DP_DPHY_CNTL, DPHY_FEC_EN, &s->dphy_fec_en);
+	REG_GET(DP_DPHY_CNTL, DPHY_FEC_READY_SHADOW, &s->dphy_fec_ready_shadow);
+	REG_GET(DP_DPHY_CNTL, DPHY_FEC_ACTIVE_STATUS, &s->dphy_fec_active_status);
+	REG_GET(DP_LINK_CNTL, DP_LINK_TRAINING_COMPLETE, &s->dp_link_training_complete);
+}
+
+static bool update_cfg_data(
+		struct dcn10_link_encoder *enc10,
+		const struct dc_link_settings *link_settings,
+		struct dpcssys_phy_seq_cfg *cfg)
+{
+	int i;
+
+	cfg->load_sram_fw = false;
+
+	for (i = 0; i < link_settings->lane_count; i++)
+		cfg->lane_en[i] = true;
+
+	switch (link_settings->link_rate) {
+	case LINK_RATE_LOW:
+		cfg->mpll_cfg = dcn2_mpll_cfg[0];
+		break;
+	case LINK_RATE_HIGH:
+		cfg->mpll_cfg = dcn2_mpll_cfg[1];
+		break;
+	case LINK_RATE_HIGH2:
+		cfg->mpll_cfg = dcn2_mpll_cfg[2];
+		break;
+	case LINK_RATE_HIGH3:
+		cfg->mpll_cfg = dcn2_mpll_cfg[3];
+		break;
+	default:
+		DC_LOG_ERROR("%s: No supported link rate found %X!\n",
+				__func__, link_settings->link_rate);
+		return false;
+	}
+
+	return true;
+}
+
+void dcn20_link_encoder_enable_dp_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	struct dcn20_link_encoder *enc20 = (struct dcn20_link_encoder *) enc10;
+	struct dpcssys_phy_seq_cfg *cfg = &enc20->phy_seq_cfg;
+
+	if (!enc->ctx->dc->debug.avoid_vbios_exec_table) {
+		dcn10_link_encoder_enable_dp_output(enc, link_settings, clock_source);
+		return;
+	}
+
+	if (!update_cfg_data(enc10, link_settings, cfg))
+		return;
+
+	enc1_configure_encoder(enc10, link_settings);
+
+	dcn10_link_encoder_setup(enc, SIGNAL_TYPE_DISPLAY_PORT);
+
+}
+
+void dcn20_link_encoder_get_max_link_cap(struct link_encoder *enc,
+	struct dc_link_settings *link_settings)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	uint32_t is_in_usb_c_dp4_mode = 0;
+
+	dcn10_link_encoder_get_max_link_cap(enc, link_settings);
+
+	/* in usb c dp2 mode, max lane count is 2 */
+	if (enc->funcs->is_in_alt_mode && enc->funcs->is_in_alt_mode(enc)) {
+		REG_GET(RDPCSTX_PHY_CNTL6, RDPCS_PHY_DPALT_DP4, &is_in_usb_c_dp4_mode);
+		if (!is_in_usb_c_dp4_mode)
+			link_settings->lane_count = MIN(LANE_COUNT_TWO, link_settings->lane_count);
+	}
+
+}
+
+bool dcn20_link_encoder_is_in_alt_mode(struct link_encoder *enc)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	uint32_t dp_alt_mode_disable = 0;
+	bool is_usb_c_alt_mode = false;
+
+	if (enc->features.flags.bits.DP_IS_USB_C) {
+		/* if value == 1 alt mode is disabled, otherwise it is enabled */
+		REG_GET(RDPCSTX_PHY_CNTL6, RDPCS_PHY_DPALT_DISABLE, &dp_alt_mode_disable);
+		is_usb_c_alt_mode = (dp_alt_mode_disable == 0);
+	}
+
+	return is_usb_c_alt_mode;
+}
+
+#define AUX_REG(reg)\
+	(enc10->aux_regs->reg)
+
+#define AUX_REG_READ(reg_name) \
+		dm_read_reg(CTX, AUX_REG(reg_name))
+
+#define AUX_REG_WRITE(reg_name, val) \
+			dm_write_reg(CTX, AUX_REG(reg_name), val)
+void enc2_hw_init(struct link_encoder *enc)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+/*
+	00 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__1to2 : 1/2
+	01 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__3to4 : 3/4
+	02 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__7to8 : 7/8
+	03 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__15to16 : 15/16
+	04 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__31to32 : 31/32
+	05 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__63to64 : 63/64
+	06 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__127to128 : 127/128
+	07 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__255to256 : 255/256
+*/
+
+/*
+	AUX_REG_UPDATE_5(AUX_DPHY_RX_CONTROL0,
+	AUX_RX_START_WINDOW = 1 [6:4]
+	AUX_RX_RECEIVE_WINDOW = 1 default is 2 [10:8]
+	AUX_RX_HALF_SYM_DETECT_LEN  = 1 [13:12] default is 1
+	AUX_RX_TRANSITION_FILTER_EN = 1 [16] default is 1
+	AUX_RX_ALLOW_BELOW_THRESHOLD_PHASE_DETECT [17] is 0  default is 0
+	AUX_RX_ALLOW_BELOW_THRESHOLD_START [18] is 1  default is 1
+	AUX_RX_ALLOW_BELOW_THRESHOLD_STOP [19] is 1  default is 1
+	AUX_RX_PHASE_DETECT_LEN,  [21,20] = 0x3 default is 3
+	AUX_RX_DETECTION_THRESHOLD [30:28] = 1
+*/
+	if (enc->ctx->dc_bios->golden_table.dc_golden_table_ver > 0) {
+		AUX_REG_WRITE(AUX_DPHY_RX_CONTROL0, enc->ctx->dc_bios->golden_table.aux_dphy_rx_control0_val);
+
+		AUX_REG_WRITE(AUX_DPHY_TX_CONTROL, enc->ctx->dc_bios->golden_table.aux_dphy_tx_control_val);
+
+		AUX_REG_WRITE(AUX_DPHY_RX_CONTROL1, enc->ctx->dc_bios->golden_table.aux_dphy_rx_control1_val);
+	} else {
+		AUX_REG_WRITE(AUX_DPHY_RX_CONTROL0, 0x103d1110);
+
+		AUX_REG_WRITE(AUX_DPHY_TX_CONTROL, 0x21c7a);
+	}
+
+	//AUX_DPHY_TX_REF_CONTROL'AUX_TX_REF_DIV HW default is 0x32;
+	// Set AUX_TX_REF_DIV Divider to generate 2 MHz reference from refclk
+	// 27MHz -> 0xd
+	// 100MHz -> 0x32
+	// 48MHz -> 0x18
+
+	// Set TMDS_CTL0 to 1.  This is a legacy setting.
+	REG_UPDATE(TMDS_CTL_BITS, TMDS_CTL0, 1);
+
+	dcn10_aux_initialize(enc10);
+}
+
+static const struct link_encoder_funcs dcn20_link_enc_funcs = {
+	.read_state = link_enc2_read_state,
+	.validate_output_with_stream =
+		dcn10_link_encoder_validate_output_with_stream,
+	.hw_init = enc2_hw_init,
+	.setup = dcn10_link_encoder_setup,
+	.enable_tmds_output = dcn10_link_encoder_enable_tmds_output_with_clk_pattern_wa,
+	.enable_dp_output = dcn20_link_encoder_enable_dp_output,
+	.enable_dp_mst_output = dcn10_link_encoder_enable_dp_mst_output,
+	.disable_output = dcn10_link_encoder_disable_output,
+	.dp_set_lane_settings = dcn10_link_encoder_dp_set_lane_settings,
+	.dp_set_phy_pattern = dcn10_link_encoder_dp_set_phy_pattern,
+	.update_mst_stream_allocation_table =
+		dcn10_link_encoder_update_mst_stream_allocation_table,
+	.psr_program_dp_dphy_fast_training =
+			dcn10_psr_program_dp_dphy_fast_training,
+	.psr_program_secondary_packet = dcn10_psr_program_secondary_packet,
+	.connect_dig_be_to_fe = dcn10_link_encoder_connect_dig_be_to_fe,
+	.enable_hpd = dcn10_link_encoder_enable_hpd,
+	.disable_hpd = dcn10_link_encoder_disable_hpd,
+	.is_dig_enabled = dcn10_is_dig_enabled,
+	.destroy = dcn10_link_encoder_destroy,
+	.fec_set_enable = enc2_fec_set_enable,
+	.fec_set_ready = enc2_fec_set_ready,
+	.fec_is_active = enc2_fec_is_active,
+	.get_dig_mode = dcn10_get_dig_mode,
+	.get_dig_frontend = dcn10_get_dig_frontend,
+	.is_in_alt_mode = dcn20_link_encoder_is_in_alt_mode,
+	.get_max_link_cap = dcn20_link_encoder_get_max_link_cap,
+};
+
+void dcn20_link_encoder_construct(
+	struct dcn20_link_encoder *enc20,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	const struct dcn10_link_enc_aux_registers *aux_regs,
+	const struct dcn10_link_enc_hpd_registers *hpd_regs,
+	const struct dcn10_link_enc_shift *link_shift,
+	const struct dcn10_link_enc_mask *link_mask)
+{
+	struct bp_encoder_cap_info bp_cap_info = {0};
+	const struct dc_vbios_funcs *bp_funcs = init_data->ctx->dc_bios->funcs;
+	enum bp_result result = BP_RESULT_OK;
+	struct dcn10_link_encoder *enc10 = &enc20->enc10;
+
+	enc10->base.funcs = &dcn20_link_enc_funcs;
+	enc10->base.ctx = init_data->ctx;
+	enc10->base.id = init_data->encoder;
+
+	enc10->base.hpd_source = init_data->hpd_source;
+	enc10->base.connector = init_data->connector;
+
+	enc10->base.preferred_engine = ENGINE_ID_UNKNOWN;
+
+	enc10->base.features = *enc_features;
+
+	enc10->base.transmitter = init_data->transmitter;
+
+	/* set the flag to indicate whether driver poll the I2C data pin
+	 * while doing the DP sink detect
+	 */
+
+/*	if (dal_adapter_service_is_feature_supported(as,
+		FEATURE_DP_SINK_DETECT_POLL_DATA_PIN))
+		enc10->base.features.flags.bits.
+			DP_SINK_DETECT_POLL_DATA_PIN = true;*/
+
+	enc10->base.output_signals =
+		SIGNAL_TYPE_DVI_SINGLE_LINK |
+		SIGNAL_TYPE_DVI_DUAL_LINK |
+		SIGNAL_TYPE_LVDS |
+		SIGNAL_TYPE_DISPLAY_PORT |
+		SIGNAL_TYPE_DISPLAY_PORT_MST |
+		SIGNAL_TYPE_EDP |
+		SIGNAL_TYPE_HDMI_TYPE_A;
+
+	/* For DCE 8.0 and 8.1, by design, UNIPHY is hardwired to DIG_BE.
+	 * SW always assign DIG_FE 1:1 mapped to DIG_FE for non-MST UNIPHY.
+	 * SW assign DIG_FE to non-MST UNIPHY first and MST last. So prefer
+	 * DIG is per UNIPHY and used by SST DP, eDP, HDMI, DVI and LVDS.
+	 * Prefer DIG assignment is decided by board design.
+	 * For DCE 8.0, there are only max 6 UNIPHYs, we assume board design
+	 * and VBIOS will filter out 7 UNIPHY for DCE 8.0.
+	 * By this, adding DIGG should not hurt DCE 8.0.
+	 * This will let DCE 8.1 share DCE 8.0 as much as possible
+	 */
+
+	enc10->link_regs = link_regs;
+	enc10->aux_regs = aux_regs;
+	enc10->hpd_regs = hpd_regs;
+	enc10->link_shift = link_shift;
+	enc10->link_mask = link_mask;
+
+	switch (enc10->base.transmitter) {
+	case TRANSMITTER_UNIPHY_A:
+		enc10->base.preferred_engine = ENGINE_ID_DIGA;
+	break;
+	case TRANSMITTER_UNIPHY_B:
+		enc10->base.preferred_engine = ENGINE_ID_DIGB;
+	break;
+	case TRANSMITTER_UNIPHY_C:
+		enc10->base.preferred_engine = ENGINE_ID_DIGC;
+	break;
+	case TRANSMITTER_UNIPHY_D:
+		enc10->base.preferred_engine = ENGINE_ID_DIGD;
+	break;
+	case TRANSMITTER_UNIPHY_E:
+		enc10->base.preferred_engine = ENGINE_ID_DIGE;
+	break;
+	case TRANSMITTER_UNIPHY_F:
+		enc10->base.preferred_engine = ENGINE_ID_DIGF;
+	break;
+	case TRANSMITTER_UNIPHY_G:
+		enc10->base.preferred_engine = ENGINE_ID_DIGG;
+	break;
+	default:
+		ASSERT_CRITICAL(false);
+		enc10->base.preferred_engine = ENGINE_ID_UNKNOWN;
+	}
+
+	/* default to one to mirror Windows behavior */
+	enc10->base.features.flags.bits.HDMI_6GB_EN = 1;
+
+	result = bp_funcs->get_encoder_cap_info(enc10->base.ctx->dc_bios,
+						enc10->base.id, &bp_cap_info);
+
+	/* Override features with DCE-specific values */
+	if (result == BP_RESULT_OK) {
+		enc10->base.features.flags.bits.IS_HBR2_CAPABLE =
+				bp_cap_info.DP_HBR2_EN;
+		enc10->base.features.flags.bits.IS_HBR3_CAPABLE =
+				bp_cap_info.DP_HBR3_EN;
+		enc10->base.features.flags.bits.HDMI_6GB_EN = bp_cap_info.HDMI_6GB_EN;
+		enc10->base.features.flags.bits.DP_IS_USB_C =
+				bp_cap_info.DP_IS_USB_C;
+	} else {
+		DC_LOG_WARNING("%s: Failed to get encoder_cap_info from VBIOS with error code %d!\n",
+				__func__,
+				result);
+	}
+	if (enc10->base.ctx->dc->debug.hdmi20_disable) {
+		enc10->base.features.flags.bits.HDMI_6GB_EN = 0;
+	}
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn20/dcn20_link_encoder.h b/drivers/gpu/drm/amd/display/dc/dio/dcn20/dcn20_link_encoder.h
new file mode 100644
index 000000000000..762c579fcb44
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn20/dcn20_link_encoder.h
@@ -0,0 +1,366 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_LINK_ENCODER__DCN20_H__
+#define __DC_LINK_ENCODER__DCN20_H__
+
+#include "dcn10/dcn10_link_encoder.h"
+
+#define DCN2_AUX_REG_LIST(id)\
+	AUX_REG_LIST(id), \
+	SRI(AUX_DPHY_TX_CONTROL, DP_AUX, id)
+
+#define UNIPHY_MASK_SH_LIST(mask_sh)\
+	LE_SF(SYMCLKA_CLOCK_ENABLE, SYMCLKA_CLOCK_ENABLE, mask_sh),\
+	LE_SF(UNIPHYA_CHANNEL_XBAR_CNTL, UNIPHY_LINK_ENABLE, mask_sh),\
+	LE_SF(UNIPHYA_CHANNEL_XBAR_CNTL, UNIPHY_CHANNEL0_XBAR_SOURCE, mask_sh),\
+	LE_SF(UNIPHYA_CHANNEL_XBAR_CNTL, UNIPHY_CHANNEL1_XBAR_SOURCE, mask_sh),\
+	LE_SF(UNIPHYA_CHANNEL_XBAR_CNTL, UNIPHY_CHANNEL2_XBAR_SOURCE, mask_sh),\
+	LE_SF(UNIPHYA_CHANNEL_XBAR_CNTL, UNIPHY_CHANNEL3_XBAR_SOURCE, mask_sh)
+
+#define DPCS_MASK_SH_LIST(mask_sh)\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX0_CLK_RDY, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX0_DATA_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX1_CLK_RDY, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX1_DATA_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX2_CLK_RDY, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX2_DATA_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX3_CLK_RDY, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX3_DATA_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL4, RDPCS_PHY_DP_TX0_TERM_CTRL, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL4, RDPCS_PHY_DP_TX1_TERM_CTRL, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL4, RDPCS_PHY_DP_TX2_TERM_CTRL, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL4, RDPCS_PHY_DP_TX3_TERM_CTRL, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL11, RDPCS_PHY_DP_MPLLB_MULTIPLIER, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL5, RDPCS_PHY_DP_TX0_WIDTH, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL5, RDPCS_PHY_DP_TX0_RATE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL5, RDPCS_PHY_DP_TX1_WIDTH, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL5, RDPCS_PHY_DP_TX1_RATE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DP_TX2_PSTATE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DP_TX3_PSTATE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DP_TX2_MPLL_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DP_TX3_MPLL_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL7, RDPCS_PHY_DP_MPLLB_FRACN_QUOT, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL7, RDPCS_PHY_DP_MPLLB_FRACN_DEN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL8, RDPCS_PHY_DP_MPLLB_SSC_PEAK, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL9, RDPCS_PHY_DP_MPLLB_SSC_UP_SPREAD, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL9, RDPCS_PHY_DP_MPLLB_SSC_STEPSIZE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL10, RDPCS_PHY_DP_MPLLB_FRACN_REM, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL11, RDPCS_PHY_DP_REF_CLK_MPLLB_DIV, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL11, RDPCS_PHY_HDMI_MPLLB_HDMI_DIV, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL12, RDPCS_PHY_DP_MPLLB_SSC_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL12, RDPCS_PHY_DP_MPLLB_DIV5_CLK_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL12, RDPCS_PHY_DP_MPLLB_TX_CLK_DIV, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL12, RDPCS_PHY_DP_MPLLB_WORD_DIV2_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL12, RDPCS_PHY_DP_MPLLB_STATE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL13, RDPCS_PHY_DP_MPLLB_DIV_CLK_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL13, RDPCS_PHY_DP_MPLLB_DIV_MULTIPLIER, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL14, RDPCS_PHY_DP_MPLLB_FRACN_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL14, RDPCS_PHY_DP_MPLLB_PMIX_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CNTL, RDPCS_TX_FIFO_LANE0_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CNTL, RDPCS_TX_FIFO_LANE1_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CNTL, RDPCS_TX_FIFO_LANE2_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CNTL, RDPCS_TX_FIFO_LANE3_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CNTL, RDPCS_TX_FIFO_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CNTL, RDPCS_TX_FIFO_RD_START_DELAY, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CLOCK_CNTL, RDPCS_EXT_REFCLK_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CLOCK_CNTL, RDPCS_SRAMCLK_BYPASS, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CLOCK_CNTL, RDPCS_SRAMCLK_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CLOCK_CNTL, RDPCS_SRAMCLK_CLOCK_ON, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CLOCK_CNTL, RDPCS_SYMCLK_DIV2_CLOCK_ON, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CLOCK_CNTL, RDPCS_SYMCLK_DIV2_GATE_DIS, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CLOCK_CNTL, RDPCS_SYMCLK_DIV2_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX0_DISABLE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX1_DISABLE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX2_DISABLE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX3_DISABLE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX0_REQ, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX1_REQ, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX2_REQ, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX3_REQ, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX0_ACK, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX1_ACK, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX2_ACK, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX3_ACK, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX0_RESET, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX1_RESET, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX2_RESET, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX3_RESET, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL0, RDPCS_PHY_RESET, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL0, RDPCS_PHY_CR_MUX_SEL, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL0, RDPCS_PHY_REF_RANGE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL0, RDPCS_SRAM_BYPASS, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL0, RDPCS_SRAM_EXT_LD_DONE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL0, RDPCS_PHY_HDMIMODE_ENABLE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL0, RDPCS_SRAM_INIT_DONE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL2, RDPCS_PHY_DP4_POR, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PLL_UPDATE_DATA, RDPCS_PLL_UPDATE_DATA, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_INTERRUPT_CONTROL, RDPCS_REG_FIFO_ERROR_MASK, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_INTERRUPT_CONTROL, RDPCS_TX_FIFO_ERROR_MASK, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_INTERRUPT_CONTROL, RDPCS_DPALT_DISABLE_TOGGLE_MASK, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_INTERRUPT_CONTROL, RDPCS_DPALT_4LANE_TOGGLE_MASK, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCS_TX_CR_ADDR, RDPCS_TX_CR_ADDR, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCS_TX_CR_DATA, RDPCS_TX_CR_DATA, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE0, RDPCS_PHY_DP_MPLLB_V2I, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE0, RDPCS_PHY_DP_TX0_EQ_MAIN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE0, RDPCS_PHY_DP_TX0_EQ_PRE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE0, RDPCS_PHY_DP_TX0_EQ_POST, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE0, RDPCS_PHY_DP_MPLLB_FREQ_VCO, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE1, RDPCS_PHY_DP_MPLLB_CP_INT, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE1, RDPCS_PHY_DP_MPLLB_CP_PROP, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE1, RDPCS_PHY_DP_TX1_EQ_MAIN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE1, RDPCS_PHY_DP_TX1_EQ_PRE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE1, RDPCS_PHY_DP_TX1_EQ_POST, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE2, RDPCS_PHY_DP_TX2_EQ_MAIN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE2, RDPCS_PHY_DP_TX2_EQ_PRE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE2, RDPCS_PHY_DP_TX2_EQ_POST, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE3, RDPCS_PHY_DP_TX3_EQ_MAIN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE3, RDPCS_PHY_DCO_FINETUNE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE3, RDPCS_PHY_DCO_RANGE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE3, RDPCS_PHY_DP_TX3_EQ_PRE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE3, RDPCS_PHY_DP_TX3_EQ_POST, mask_sh),\
+	LE_SF(DPCSTX0_DPCSTX_TX_CLOCK_CNTL, DPCS_SYMCLK_CLOCK_ON, mask_sh),\
+	LE_SF(DPCSTX0_DPCSTX_TX_CLOCK_CNTL, DPCS_SYMCLK_GATE_DIS, mask_sh),\
+	LE_SF(DPCSTX0_DPCSTX_TX_CLOCK_CNTL, DPCS_SYMCLK_EN, mask_sh),\
+	LE_SF(DPCSTX0_DPCSTX_TX_CNTL, DPCS_TX_DATA_SWAP, mask_sh),\
+	LE_SF(DPCSTX0_DPCSTX_TX_CNTL, DPCS_TX_DATA_ORDER_INVERT, mask_sh),\
+	LE_SF(DPCSTX0_DPCSTX_TX_CNTL, DPCS_TX_FIFO_EN, mask_sh),\
+	LE_SF(DPCSTX0_DPCSTX_TX_CNTL, DPCS_TX_FIFO_RD_START_DELAY, mask_sh),\
+	LE_SF(DPCSTX0_DPCSTX_DEBUG_CONFIG, DPCS_DBG_CBUS_DIS, mask_sh)
+
+#define DPCS_DCN2_MASK_SH_LIST(mask_sh)\
+	DPCS_MASK_SH_LIST(mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_RX_LD_VAL, RDPCS_PHY_RX_REF_LD_VAL, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_RX_LD_VAL, RDPCS_PHY_RX_VCO_LD_VAL, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DPALT_DISABLE_ACK, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DP_TX0_PSTATE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DP_TX1_PSTATE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DP_TX0_MPLL_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DP_TX1_MPLL_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DP_REF_CLK_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL5, RDPCS_PHY_DP_TX2_WIDTH, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL5, RDPCS_PHY_DP_TX2_RATE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL5, RDPCS_PHY_DP_TX3_WIDTH, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL5, RDPCS_PHY_DP_TX3_RATE, mask_sh),\
+	LE_SF(DCIO_SOFT_RESET, UNIPHYA_SOFT_RESET, mask_sh),\
+	LE_SF(DCIO_SOFT_RESET, UNIPHYB_SOFT_RESET, mask_sh),\
+	LE_SF(DCIO_SOFT_RESET, UNIPHYC_SOFT_RESET, mask_sh),\
+	LE_SF(DCIO_SOFT_RESET, UNIPHYD_SOFT_RESET, mask_sh),\
+	LE_SF(DCIO_SOFT_RESET, UNIPHYE_SOFT_RESET, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DPALT_DP4, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DPALT_DISABLE, mask_sh)
+
+#define LINK_ENCODER_MASK_SH_LIST_DCN20(mask_sh)\
+	LINK_ENCODER_MASK_SH_LIST_DCN10(mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_FEC_EN, mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_FEC_READY_SHADOW, mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_FEC_ACTIVE_STATUS, mask_sh),\
+	LE_SF(DIG0_DIG_LANE_ENABLE, DIG_LANE0EN, mask_sh),\
+	LE_SF(DIG0_DIG_LANE_ENABLE, DIG_LANE1EN, mask_sh),\
+	LE_SF(DIG0_DIG_LANE_ENABLE, DIG_LANE2EN, mask_sh),\
+	LE_SF(DIG0_DIG_LANE_ENABLE, DIG_LANE3EN, mask_sh),\
+	LE_SF(DIG0_DIG_LANE_ENABLE, DIG_CLK_EN, mask_sh),\
+	LE_SF(DIG0_TMDS_CTL_BITS, TMDS_CTL0, mask_sh), \
+	UNIPHY_MASK_SH_LIST(mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_START_WINDOW, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_HALF_SYM_DETECT_LEN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_TRANSITION_FILTER_EN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_ALLOW_BELOW_THRESHOLD_PHASE_DETECT, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_ALLOW_BELOW_THRESHOLD_START, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_ALLOW_BELOW_THRESHOLD_STOP, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_PHASE_DETECT_LEN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_DETECTION_THRESHOLD, mask_sh), \
+	LE_SF(DP_AUX0_AUX_DPHY_TX_CONTROL, AUX_TX_PRECHARGE_LEN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_TX_CONTROL, AUX_TX_PRECHARGE_SYMBOLS, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_TX_CONTROL, AUX_MODE_DET_CHECK_DELAY, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL1, AUX_RX_PRECHARGE_SKIP, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL1, AUX_RX_TIMEOUT_LEN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL1, AUX_RX_TIMEOUT_LEN_MUL, mask_sh)
+
+#define UNIPHY_DCN2_REG_LIST(id) \
+	SRI(CLOCK_ENABLE, SYMCLK, id), \
+	SRI(CHANNEL_XBAR_CNTL, UNIPHY, id)
+
+#define DPCS_DCN2_CMN_REG_LIST(id) \
+	SRI(DIG_LANE_ENABLE, DIG, id), \
+	SRI(TMDS_CTL_BITS, DIG, id), \
+	SRI(RDPCSTX_PHY_CNTL3, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL4, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL5, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL6, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL7, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL8, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL9, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL10, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL11, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL12, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL13, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL14, RDPCSTX, id), \
+	SRI(RDPCSTX_CNTL, RDPCSTX, id), \
+	SRI(RDPCSTX_CLOCK_CNTL, RDPCSTX, id), \
+	SRI(RDPCSTX_INTERRUPT_CONTROL, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL0, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL2, RDPCSTX, id), \
+	SRI(RDPCSTX_PLL_UPDATE_DATA, RDPCSTX, id), \
+	SRI(RDPCS_TX_CR_ADDR, RDPCSTX, id), \
+	SRI(RDPCS_TX_CR_DATA, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_FUSE0, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_FUSE1, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_FUSE2, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_FUSE3, RDPCSTX, id), \
+	SRI(DPCSTX_TX_CLOCK_CNTL, DPCSTX, id), \
+	SRI(DPCSTX_TX_CNTL, DPCSTX, id), \
+	SRI(DPCSTX_DEBUG_CONFIG, DPCSTX, id), \
+	SRI(RDPCSTX_DEBUG_CONFIG, RDPCSTX, id), \
+	SR(RDPCSTX0_RDPCSTX_SCRATCH)
+
+
+#define DPCS_DCN2_REG_LIST(id) \
+	DPCS_DCN2_CMN_REG_LIST(id), \
+	SRI(RDPCSTX_PHY_RX_LD_VAL, RDPCSTX, id),\
+	SRI(RDPCSTX_DMCU_DPALT_DIS_BLOCK_REG, RDPCSTX, id)
+
+#define LE_DCN2_REG_LIST(id) \
+		LE_DCN10_REG_LIST(id), \
+		SR(DCIO_SOFT_RESET)
+
+struct mpll_cfg {
+	uint32_t mpllb_ana_v2i;
+	uint32_t mpllb_ana_freq_vco;
+	uint32_t mpllb_ana_cp_int;
+	uint32_t mpllb_ana_cp_prop;
+	uint32_t mpllb_multiplier;
+	uint32_t ref_clk_mpllb_div;
+	bool mpllb_word_div2_en;
+	bool mpllb_ssc_en;
+	bool mpllb_div5_clk_en;
+	bool mpllb_div_clk_en;
+	bool mpllb_fracn_en;
+	bool mpllb_pmix_en;
+	uint32_t mpllb_div_multiplier;
+	uint32_t mpllb_tx_clk_div;
+	uint32_t mpllb_fracn_quot;
+	uint32_t mpllb_fracn_den;
+	uint32_t mpllb_ssc_peak;
+	uint32_t mpllb_ssc_stepsize;
+	uint32_t mpllb_ssc_up_spread;
+	uint32_t mpllb_fracn_rem;
+	uint32_t mpllb_hdmi_div;
+	// TODO: May not mpll params, need to figure out.
+	uint32_t tx_vboost_lvl;
+	uint32_t hdmi_pixel_clk_div;
+	uint32_t ref_range;
+	uint32_t ref_clk;
+	bool hdmimode_enable;
+	bool sup_pre_hp;
+	bool dp_tx0_vergdrv_byp;
+	bool dp_tx1_vergdrv_byp;
+	bool dp_tx2_vergdrv_byp;
+	bool dp_tx3_vergdrv_byp;
+	uint32_t tx_peaking_lvl;
+	uint32_t ctr_reqs_pll;
+
+};
+
+struct dpcssys_phy_seq_cfg {
+	bool program_fuse;
+	bool bypass_sram;
+	bool lane_en[4];
+	bool use_calibration_setting;
+	struct mpll_cfg mpll_cfg;
+	bool load_sram_fw;
+#if 0
+
+	bool hdmimode_enable;
+	bool silver2;
+	bool ext_refclk_en;
+	uint32_t dp_tx0_term_ctrl;
+	uint32_t dp_tx1_term_ctrl;
+	uint32_t dp_tx2_term_ctrl;
+	uint32_t dp_tx3_term_ctrl;
+	uint32_t fw_data[0x1000];
+	uint32_t dp_tx0_width;
+	uint32_t dp_tx1_width;
+	uint32_t dp_tx2_width;
+	uint32_t dp_tx3_width;
+	uint32_t dp_tx0_rate;
+	uint32_t dp_tx1_rate;
+	uint32_t dp_tx2_rate;
+	uint32_t dp_tx3_rate;
+	uint32_t dp_tx0_eq_main;
+	uint32_t dp_tx0_eq_pre;
+	uint32_t dp_tx0_eq_post;
+	uint32_t dp_tx1_eq_main;
+	uint32_t dp_tx1_eq_pre;
+	uint32_t dp_tx1_eq_post;
+	uint32_t dp_tx2_eq_main;
+	uint32_t dp_tx2_eq_pre;
+	uint32_t dp_tx2_eq_post;
+	uint32_t dp_tx3_eq_main;
+	uint32_t dp_tx3_eq_pre;
+	uint32_t dp_tx3_eq_post;
+	bool data_swap_en;
+	bool data_order_invert_en;
+	uint32_t ldpcs_fifo_start_delay;
+	uint32_t rdpcs_fifo_start_delay;
+	bool rdpcs_reg_fifo_error_mask;
+	bool rdpcs_tx_fifo_error_mask;
+	bool rdpcs_dpalt_disable_mask;
+	bool rdpcs_dpalt_4lane_mask;
+#endif
+};
+
+struct dcn20_link_encoder {
+	struct dcn10_link_encoder enc10;
+	struct dpcssys_phy_seq_cfg phy_seq_cfg;
+};
+
+void enc2_fec_set_enable(struct link_encoder *enc, bool enable);
+void enc2_fec_set_ready(struct link_encoder *enc, bool ready);
+bool enc2_fec_is_active(struct link_encoder *enc);
+void enc2_hw_init(struct link_encoder *enc);
+
+void link_enc2_read_state(struct link_encoder *enc, struct link_enc_state *s);
+
+void dcn20_link_encoder_enable_dp_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source);
+
+bool dcn20_link_encoder_is_in_alt_mode(struct link_encoder *enc);
+void dcn20_link_encoder_get_max_link_cap(struct link_encoder *enc,
+	struct dc_link_settings *link_settings);
+
+void dcn20_link_encoder_construct(
+	struct dcn20_link_encoder *enc20,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	const struct dcn10_link_enc_aux_registers *aux_regs,
+	const struct dcn10_link_enc_hpd_registers *hpd_regs,
+	const struct dcn10_link_enc_shift *link_shift,
+	const struct dcn10_link_enc_mask *link_mask);
+
+#endif /* __DC_LINK_ENCODER__DCN20_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn20/dcn20_stream_encoder.c b/drivers/gpu/drm/amd/display/dc/dio/dcn20/dcn20_stream_encoder.c
new file mode 100644
index 000000000000..bec0b4aaeb2b
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn20/dcn20_stream_encoder.c
@@ -0,0 +1,661 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include <linux/delay.h>
+
+#include "dc_bios_types.h"
+#include "dcn20_stream_encoder.h"
+#include "reg_helper.h"
+#include "hw_shared.h"
+#include "link_service.h"
+#include "dpcd_defs.h"
+
+#define DC_LOGGER \
+		enc1->base.ctx->logger
+
+#define REG(reg)\
+	(enc1->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	enc1->se_shift->field_name, enc1->se_mask->field_name
+
+
+#define CTX \
+	enc1->base.ctx
+
+
+static void enc2_update_hdmi_info_packet(
+	struct dcn10_stream_encoder *enc1,
+	uint32_t packet_index,
+	const struct dc_info_packet *info_packet)
+{
+	uint32_t cont, send, line;
+
+	if (info_packet->valid) {
+		enc1_update_generic_info_packet(
+			enc1,
+			packet_index,
+			info_packet);
+
+		/* enable transmission of packet(s) -
+		 * packet transmission begins on the next frame */
+		cont = 1;
+		/* send packet(s) every frame */
+		send = 1;
+		/* select line number to send packets on */
+		line = 2;
+	} else {
+		cont = 0;
+		send = 0;
+		line = 0;
+	}
+
+	/* DP_SEC_GSP[x]_LINE_REFERENCE - keep default value REFER_TO_DP_SOF */
+
+	/* choose which generic packet control to use */
+	switch (packet_index) {
+	case 0:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL0,
+				HDMI_GENERIC0_CONT, cont,
+				HDMI_GENERIC0_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL1,
+				HDMI_GENERIC0_LINE, line);
+		break;
+	case 1:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL0,
+				HDMI_GENERIC1_CONT, cont,
+				HDMI_GENERIC1_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL1,
+				HDMI_GENERIC1_LINE, line);
+		break;
+	case 2:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL0,
+				HDMI_GENERIC2_CONT, cont,
+				HDMI_GENERIC2_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL2,
+				HDMI_GENERIC2_LINE, line);
+		break;
+	case 3:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL0,
+				HDMI_GENERIC3_CONT, cont,
+				HDMI_GENERIC3_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL2,
+				HDMI_GENERIC3_LINE, line);
+		break;
+	case 4:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL0,
+				HDMI_GENERIC4_CONT, cont,
+				HDMI_GENERIC4_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL3,
+				HDMI_GENERIC4_LINE, line);
+		break;
+	case 5:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL0,
+				HDMI_GENERIC5_CONT, cont,
+				HDMI_GENERIC5_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL3,
+				HDMI_GENERIC5_LINE, line);
+		break;
+	case 6:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL0,
+				HDMI_GENERIC6_CONT, cont,
+				HDMI_GENERIC6_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL4,
+				HDMI_GENERIC6_LINE, line);
+		break;
+	case 7:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL0,
+				HDMI_GENERIC7_CONT, cont,
+				HDMI_GENERIC7_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL4,
+				HDMI_GENERIC7_LINE, line);
+		break;
+	default:
+		/* invalid HW packet index */
+		DC_LOG_WARNING(
+			"Invalid HW packet index: %s()\n",
+			__func__);
+		return;
+	}
+}
+
+static void enc2_stream_encoder_update_hdmi_info_packets(
+	struct stream_encoder *enc,
+	const struct encoder_info_frame *info_frame)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	/* for bring up, disable dp double  TODO */
+	REG_UPDATE(HDMI_DB_CONTROL, HDMI_DB_DISABLE, 1);
+
+	/*Always add mandatory packets first followed by optional ones*/
+	enc2_update_hdmi_info_packet(enc1, 0, &info_frame->avi);
+	enc2_update_hdmi_info_packet(enc1, 1, &info_frame->hfvsif);
+	enc2_update_hdmi_info_packet(enc1, 2, &info_frame->gamut);
+	enc2_update_hdmi_info_packet(enc1, 3, &info_frame->vendor);
+	enc2_update_hdmi_info_packet(enc1, 4, &info_frame->spd);
+	enc2_update_hdmi_info_packet(enc1, 5, &info_frame->hdrsmd);
+	enc2_update_hdmi_info_packet(enc1, 6, &info_frame->vtem);
+}
+
+static void enc2_stream_encoder_stop_hdmi_info_packets(
+	struct stream_encoder *enc)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	/* stop generic packets 0,1 on HDMI */
+	REG_SET_4(HDMI_GENERIC_PACKET_CONTROL0, 0,
+		HDMI_GENERIC0_CONT, 0,
+		HDMI_GENERIC0_SEND, 0,
+		HDMI_GENERIC1_CONT, 0,
+		HDMI_GENERIC1_SEND, 0);
+	REG_SET_2(HDMI_GENERIC_PACKET_CONTROL1, 0,
+		HDMI_GENERIC0_LINE, 0,
+		HDMI_GENERIC1_LINE, 0);
+
+	/* stop generic packets 2,3 on HDMI */
+	REG_SET_4(HDMI_GENERIC_PACKET_CONTROL0, 0,
+		HDMI_GENERIC2_CONT, 0,
+		HDMI_GENERIC2_SEND, 0,
+		HDMI_GENERIC3_CONT, 0,
+		HDMI_GENERIC3_SEND, 0);
+	REG_SET_2(HDMI_GENERIC_PACKET_CONTROL2, 0,
+		HDMI_GENERIC2_LINE, 0,
+		HDMI_GENERIC3_LINE, 0);
+
+	/* stop generic packets 4,5 on HDMI */
+	REG_SET_4(HDMI_GENERIC_PACKET_CONTROL0, 0,
+		HDMI_GENERIC4_CONT, 0,
+		HDMI_GENERIC4_SEND, 0,
+		HDMI_GENERIC5_CONT, 0,
+		HDMI_GENERIC5_SEND, 0);
+	REG_SET_2(HDMI_GENERIC_PACKET_CONTROL3, 0,
+		HDMI_GENERIC4_LINE, 0,
+		HDMI_GENERIC5_LINE, 0);
+
+	/* stop generic packets 6,7 on HDMI */
+	REG_SET_4(HDMI_GENERIC_PACKET_CONTROL0, 0,
+		HDMI_GENERIC6_CONT, 0,
+		HDMI_GENERIC6_SEND, 0,
+		HDMI_GENERIC7_CONT, 0,
+		HDMI_GENERIC7_SEND, 0);
+	REG_SET_2(HDMI_GENERIC_PACKET_CONTROL4, 0,
+		HDMI_GENERIC6_LINE, 0,
+		HDMI_GENERIC7_LINE, 0);
+}
+
+
+/* Update GSP7 SDP 128 byte long */
+static void enc2_update_gsp7_128_info_packet(
+	struct dcn10_stream_encoder *enc1,
+	const struct dc_info_packet_128 *info_packet,
+	bool immediate_update)
+{
+	uint32_t i;
+
+	/* TODOFPGA Figure out a proper number for max_retries polling for lock
+	 * use 50 for now.
+	 */
+	uint32_t max_retries = 50;
+	const uint32_t *content = (const uint32_t *) &info_packet->sb[0];
+
+	ASSERT(info_packet->hb1  == DC_DP_INFOFRAME_TYPE_PPS);
+
+	/* Configure for PPS packet size (128 bytes) */
+	REG_UPDATE(DP_SEC_CNTL2, DP_SEC_GSP7_PPS, 1);
+
+	/* We need turn on clock before programming AFMT block*/
+	REG_UPDATE(AFMT_CNTL, AFMT_AUDIO_CLOCK_EN, 1);
+
+	/* Poll dig_update_lock is not locked -> asic internal signal
+	 * assumes otg master lock will unlock it
+	 */
+	/*REG_WAIT(AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_LOCK_STATUS, 0, 10, max_retries);*/
+
+	/* Wait for HW/SW GSP memory access conflict to go away */
+	REG_WAIT(AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_CONFLICT,
+			0, 10, max_retries);
+
+	/* Clear HW/SW memory access conflict flag */
+	REG_UPDATE(AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_CONFLICT_CLR, 1);
+
+	/* write generic packet header */
+	REG_UPDATE(AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_INDEX, 7);
+	REG_SET_4(AFMT_GENERIC_HDR, 0,
+			AFMT_GENERIC_HB0, info_packet->hb0,
+			AFMT_GENERIC_HB1, info_packet->hb1,
+			AFMT_GENERIC_HB2, info_packet->hb2,
+			AFMT_GENERIC_HB3, info_packet->hb3);
+
+	/* Write generic packet content 128 bytes long. Four sets are used (indexes 7
+	 * through 10) to fit 128 bytes.
+	 */
+	for (i = 0; i < 4; i++) {
+		uint32_t packet_index = 7 + i;
+		REG_UPDATE(AFMT_VBI_PACKET_CONTROL, AFMT_GENERIC_INDEX, packet_index);
+
+		REG_WRITE(AFMT_GENERIC_0, *content++);
+		REG_WRITE(AFMT_GENERIC_1, *content++);
+		REG_WRITE(AFMT_GENERIC_2, *content++);
+		REG_WRITE(AFMT_GENERIC_3, *content++);
+		REG_WRITE(AFMT_GENERIC_4, *content++);
+		REG_WRITE(AFMT_GENERIC_5, *content++);
+		REG_WRITE(AFMT_GENERIC_6, *content++);
+		REG_WRITE(AFMT_GENERIC_7, *content++);
+	}
+
+	REG_UPDATE_2(AFMT_VBI_PACKET_CONTROL1,
+			AFMT_GENERIC7_FRAME_UPDATE, !immediate_update,
+			AFMT_GENERIC7_IMMEDIATE_UPDATE, immediate_update);
+}
+
+/* Set DSC-related configuration.
+ *   dsc_mode: 0 disables DSC, other values enable DSC in specified format
+ *   sc_bytes_per_pixel: Bytes per pixel in u3.28 format
+ *   dsc_slice_width: Slice width in pixels
+ */
+static void enc2_dp_set_dsc_config(struct stream_encoder *enc,
+					enum optc_dsc_mode dsc_mode,
+					uint32_t dsc_bytes_per_pixel,
+					uint32_t dsc_slice_width)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	REG_UPDATE_2(DP_DSC_CNTL,
+			DP_DSC_MODE, dsc_mode,
+			DP_DSC_SLICE_WIDTH, dsc_slice_width);
+
+	REG_SET(DP_DSC_BYTES_PER_PIXEL, 0,
+		DP_DSC_BYTES_PER_PIXEL, dsc_bytes_per_pixel);
+}
+
+
+static void enc2_dp_set_dsc_pps_info_packet(struct stream_encoder *enc,
+					bool enable,
+					uint8_t *dsc_packed_pps,
+					bool immediate_update)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (enable) {
+		struct dc_info_packet_128 pps_sdp;
+
+		ASSERT(dsc_packed_pps);
+
+		/* Load PPS into infoframe (SDP) registers */
+		pps_sdp.valid = true;
+		pps_sdp.hb0 = 0;
+		pps_sdp.hb1 = DC_DP_INFOFRAME_TYPE_PPS;
+		pps_sdp.hb2 = 127;
+		pps_sdp.hb3 = 0;
+		memcpy(&pps_sdp.sb[0], dsc_packed_pps, sizeof(pps_sdp.sb));
+		enc2_update_gsp7_128_info_packet(enc1, &pps_sdp, immediate_update);
+
+		/* Enable Generic Stream Packet 7 (GSP) transmission */
+		//REG_UPDATE(DP_SEC_CNTL,
+		//	DP_SEC_GSP7_ENABLE, 1);
+
+		/* SW should make sure VBID[6] update line number is bigger
+		 * than PPS transmit line number
+		 */
+		REG_UPDATE(DP_SEC_CNTL6,
+				DP_SEC_GSP7_LINE_NUM, 2);
+		REG_UPDATE_2(DP_MSA_VBID_MISC,
+				DP_VBID6_LINE_REFERENCE, 0,
+				DP_VBID6_LINE_NUM, 3);
+
+		/* Send PPS data at the line number specified above.
+		 * DP spec requires PPS to be sent only when it changes, however since
+		 * decoder has to be able to handle its change on every frame, we're
+		 * sending it always (i.e. on every frame) to reduce the chance it'd be
+		 * missed by decoder. If it turns out required to send PPS only when it
+		 * changes, we can use DP_SEC_GSP7_SEND register.
+		 */
+		REG_UPDATE_2(DP_SEC_CNTL,
+			DP_SEC_GSP7_ENABLE, 1,
+			DP_SEC_STREAM_ENABLE, 1);
+	} else {
+		/* Disable Generic Stream Packet 7 (GSP) transmission */
+		REG_UPDATE(DP_SEC_CNTL, DP_SEC_GSP7_ENABLE, 0);
+		REG_UPDATE(DP_SEC_CNTL2, DP_SEC_GSP7_PPS, 0);
+	}
+}
+
+
+/* this function read dsc related register fields to be logged later in dcn10_log_hw_state
+ * into a dcn_dsc_state struct.
+ */
+static void enc2_read_state(struct stream_encoder *enc, struct enc_state *s)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	//if dsc is enabled, continue to read
+	REG_GET(DP_DSC_CNTL, DP_DSC_MODE, &s->dsc_mode);
+	if (s->dsc_mode) {
+		REG_GET(DP_DSC_CNTL, DP_DSC_SLICE_WIDTH, &s->dsc_slice_width);
+		REG_GET(DP_SEC_CNTL6, DP_SEC_GSP7_LINE_NUM, &s->sec_gsp_pps_line_num);
+
+		REG_GET(DP_MSA_VBID_MISC, DP_VBID6_LINE_REFERENCE, &s->vbid6_line_reference);
+		REG_GET(DP_MSA_VBID_MISC, DP_VBID6_LINE_NUM, &s->vbid6_line_num);
+
+		REG_GET(DP_SEC_CNTL, DP_SEC_GSP7_ENABLE, &s->sec_gsp_pps_enable);
+		REG_GET(DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, &s->sec_stream_enable);
+	}
+}
+
+/* Set Dynamic Metadata-configuration.
+ *   enable_dme:         TRUE: enables Dynamic Metadata Enfine, FALSE: disables DME
+ *   hubp_requestor_id:  HUBP physical instance that is the source of dynamic metadata
+ *                       only needs to be set when enable_dme is TRUE
+ *   dmdata_mode:        dynamic metadata packet type: DP, HDMI, or Dolby Vision
+ *
+ *   Ensure the OTG master update lock is set when changing DME configuration.
+ */
+void enc2_set_dynamic_metadata(struct stream_encoder *enc,
+		bool enable_dme,
+		uint32_t hubp_requestor_id,
+		enum dynamic_metadata_mode dmdata_mode)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (enable_dme) {
+		REG_UPDATE_2(DME_CONTROL,
+				METADATA_HUBP_REQUESTOR_ID, hubp_requestor_id,
+				METADATA_STREAM_TYPE, (dmdata_mode == dmdata_dolby_vision) ? 1 : 0);
+
+		/* Use default line reference DP_SOF for bringup.
+		 * Should use OTG_SOF for DRR cases
+		 */
+		if (dmdata_mode == dmdata_dp)
+			REG_UPDATE_3(DP_SEC_METADATA_TRANSMISSION,
+					DP_SEC_METADATA_PACKET_ENABLE, 1,
+					DP_SEC_METADATA_PACKET_LINE_REFERENCE, 0,
+					DP_SEC_METADATA_PACKET_LINE, 20);
+		else {
+			REG_UPDATE_3(HDMI_METADATA_PACKET_CONTROL,
+					HDMI_METADATA_PACKET_ENABLE, 1,
+					HDMI_METADATA_PACKET_LINE_REFERENCE, 0,
+					HDMI_METADATA_PACKET_LINE, 2);
+
+			if (dmdata_mode == dmdata_dolby_vision)
+				REG_UPDATE(DIG_FE_CNTL,
+						DOLBY_VISION_EN, 1);
+		}
+
+		REG_UPDATE(DME_CONTROL,
+				METADATA_ENGINE_EN, 1);
+	} else {
+		REG_UPDATE(DME_CONTROL,
+				METADATA_ENGINE_EN, 0);
+
+		if (dmdata_mode == dmdata_dp)
+			REG_UPDATE(DP_SEC_METADATA_TRANSMISSION,
+					DP_SEC_METADATA_PACKET_ENABLE, 0);
+		else {
+			REG_UPDATE(HDMI_METADATA_PACKET_CONTROL,
+					HDMI_METADATA_PACKET_ENABLE, 0);
+			REG_UPDATE(DIG_FE_CNTL,
+					DOLBY_VISION_EN, 0);
+		}
+	}
+}
+
+static void enc2_stream_encoder_update_dp_info_packets_sdp_line_num(
+		struct stream_encoder *enc,
+		struct encoder_info_frame *info_frame)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (info_frame->adaptive_sync.valid == true &&
+		info_frame->sdp_line_num.adaptive_sync_line_num_valid == true) {
+		//00: REFER_TO_DP_SOF, 01: REFER_TO_OTG_SOF
+		REG_UPDATE(DP_SEC_CNTL1, DP_SEC_GSP5_LINE_REFERENCE, 1);
+
+		REG_UPDATE(DP_SEC_CNTL5, DP_SEC_GSP5_LINE_NUM,
+					info_frame->sdp_line_num.adaptive_sync_line_num);
+	}
+}
+
+static void enc2_stream_encoder_update_dp_info_packets(
+	struct stream_encoder *enc,
+	const struct encoder_info_frame *info_frame)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+	uint32_t dmdata_packet_enabled = 0;
+
+	enc1_stream_encoder_update_dp_info_packets(enc, info_frame);
+
+	/* check if dynamic metadata packet transmission is enabled */
+	REG_GET(DP_SEC_METADATA_TRANSMISSION,
+			DP_SEC_METADATA_PACKET_ENABLE, &dmdata_packet_enabled);
+
+	if (dmdata_packet_enabled)
+		REG_UPDATE(DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, 1);
+}
+
+static bool is_two_pixels_per_containter(const struct dc_crtc_timing *timing)
+{
+	bool two_pix = timing->pixel_encoding == PIXEL_ENCODING_YCBCR420;
+
+	two_pix = two_pix || (timing->flags.DSC && timing->pixel_encoding == PIXEL_ENCODING_YCBCR422
+			&& !timing->dsc_cfg.ycbcr422_simple);
+	return two_pix;
+}
+
+void enc2_stream_encoder_dp_unblank(
+		struct dc_link *link,
+		struct stream_encoder *enc,
+		const struct encoder_unblank_param *param)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (param->link_settings.link_rate != LINK_RATE_UNKNOWN) {
+		uint32_t n_vid = 0x8000;
+		uint32_t m_vid;
+		uint32_t n_multiply = 0;
+		uint64_t m_vid_l = n_vid;
+
+		/* YCbCr 4:2:0 : Computed VID_M will be 2X the input rate */
+		if (is_two_pixels_per_containter(&param->timing) || param->opp_cnt > 1) {
+			/*this logic should be the same in get_pixel_clock_parameters() */
+			n_multiply = 1;
+		}
+		/* M / N = Fstream / Flink
+		 * m_vid / n_vid = pixel rate / link rate
+		 */
+
+		m_vid_l *= param->timing.pix_clk_100hz / 10;
+		m_vid_l = div_u64(m_vid_l,
+			param->link_settings.link_rate
+				* LINK_RATE_REF_FREQ_IN_KHZ);
+
+		m_vid = (uint32_t) m_vid_l;
+
+		/* enable auto measurement */
+
+		REG_UPDATE(DP_VID_TIMING, DP_VID_M_N_GEN_EN, 0);
+
+		/* auto measurement need 1 full 0x8000 symbol cycle to kick in,
+		 * therefore program initial value for Mvid and Nvid
+		 */
+
+		REG_UPDATE(DP_VID_N, DP_VID_N, n_vid);
+
+		REG_UPDATE(DP_VID_M, DP_VID_M, m_vid);
+
+		REG_UPDATE_2(DP_VID_TIMING,
+				DP_VID_M_N_GEN_EN, 1,
+				DP_VID_N_MUL, n_multiply);
+	}
+
+	/* make sure stream is disabled before resetting steer fifo */
+	REG_UPDATE(DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, false);
+	REG_WAIT(DP_VID_STREAM_CNTL, DP_VID_STREAM_STATUS, 0, 10, 5000);
+
+	/* set DIG_START to 0x1 to reset FIFO */
+	REG_UPDATE(DIG_FE_CNTL, DIG_START, 1);
+	udelay(1);
+
+	/* write 0 to take the FIFO out of reset */
+
+	REG_UPDATE(DIG_FE_CNTL, DIG_START, 0);
+
+	/* switch DP encoder to CRTC data, but reset it the fifo first. It may happen
+	 * that it overflows during mode transition, and sometimes doesn't recover.
+	 */
+	REG_UPDATE(DP_STEER_FIFO, DP_STEER_FIFO_RESET, 1);
+	udelay(10);
+
+	REG_UPDATE(DP_STEER_FIFO, DP_STEER_FIFO_RESET, 0);
+
+	/* wait 100us for DIG/DP logic to prime
+	 * (i.e. a few video lines)
+	 */
+	udelay(100);
+
+	/* the hardware would start sending video at the start of the next DP
+	 * frame (i.e. rising edge of the vblank).
+	 * NOTE: We used to program DP_VID_STREAM_DIS_DEFER = 2 here, but this
+	 * register has no effect on enable transition! HW always guarantees
+	 * VID_STREAM enable at start of next frame, and this is not
+	 * programmable
+	 */
+
+	REG_UPDATE(DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, true);
+
+	link->dc->link_srv->dp_trace_source_sequence(link,
+			DPCD_SOURCE_SEQ_AFTER_ENABLE_DP_VID_STREAM);
+}
+
+static void enc2_dp_set_odm_combine(
+	struct stream_encoder *enc,
+	bool odm_combine)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	REG_UPDATE(DP_PIXEL_FORMAT, DP_PIXEL_COMBINE, odm_combine);
+}
+
+void enc2_stream_encoder_dp_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	enum dc_color_space output_color_space,
+	bool use_vsc_sdp_for_colorimetry,
+	uint32_t enable_sdp_splitting)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	enc1_stream_encoder_dp_set_stream_attribute(enc,
+			crtc_timing,
+			output_color_space,
+			use_vsc_sdp_for_colorimetry,
+			enable_sdp_splitting);
+
+	REG_UPDATE(DP_SEC_FRAMING4,
+		DP_SST_SDP_SPLITTING, enable_sdp_splitting);
+}
+
+uint32_t enc2_get_fifo_cal_average_level(
+		struct stream_encoder *enc)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+	uint32_t fifo_level;
+
+	REG_GET(DIG_FIFO_STATUS,
+			DIG_FIFO_CAL_AVERAGE_LEVEL, &fifo_level);
+	return fifo_level;
+}
+
+static const struct stream_encoder_funcs dcn20_str_enc_funcs = {
+	.dp_set_odm_combine =
+		enc2_dp_set_odm_combine,
+	.dp_set_stream_attribute =
+		enc2_stream_encoder_dp_set_stream_attribute,
+	.hdmi_set_stream_attribute =
+		enc1_stream_encoder_hdmi_set_stream_attribute,
+	.dvi_set_stream_attribute =
+		enc1_stream_encoder_dvi_set_stream_attribute,
+	.set_throttled_vcp_size =
+		enc1_stream_encoder_set_throttled_vcp_size,
+	.update_hdmi_info_packets =
+		enc2_stream_encoder_update_hdmi_info_packets,
+	.stop_hdmi_info_packets =
+		enc2_stream_encoder_stop_hdmi_info_packets,
+	.update_dp_info_packets_sdp_line_num =
+		enc2_stream_encoder_update_dp_info_packets_sdp_line_num,
+	.update_dp_info_packets =
+		enc2_stream_encoder_update_dp_info_packets,
+	.send_immediate_sdp_message =
+		enc1_stream_encoder_send_immediate_sdp_message,
+	.stop_dp_info_packets =
+		enc1_stream_encoder_stop_dp_info_packets,
+	.dp_blank =
+		enc1_stream_encoder_dp_blank,
+	.dp_unblank =
+		enc2_stream_encoder_dp_unblank,
+	.audio_mute_control = enc1_se_audio_mute_control,
+
+	.dp_audio_setup = enc1_se_dp_audio_setup,
+	.dp_audio_enable = enc1_se_dp_audio_enable,
+	.dp_audio_disable = enc1_se_dp_audio_disable,
+
+	.hdmi_audio_setup = enc1_se_hdmi_audio_setup,
+	.hdmi_audio_disable = enc1_se_hdmi_audio_disable,
+	.setup_stereo_sync  = enc1_setup_stereo_sync,
+	.set_avmute = enc1_stream_encoder_set_avmute,
+	.dig_connect_to_otg  = enc1_dig_connect_to_otg,
+	.dig_source_otg = enc1_dig_source_otg,
+
+	.dp_get_pixel_format =
+		enc1_stream_encoder_dp_get_pixel_format,
+
+	.enc_read_state = enc2_read_state,
+	.dp_set_dsc_config = enc2_dp_set_dsc_config,
+	.dp_set_dsc_pps_info_packet = enc2_dp_set_dsc_pps_info_packet,
+	.set_dynamic_metadata = enc2_set_dynamic_metadata,
+	.hdmi_reset_stream_attribute = enc1_reset_hdmi_stream_attribute,
+	.get_fifo_cal_average_level = enc2_get_fifo_cal_average_level,
+};
+
+void dcn20_stream_encoder_construct(
+	struct dcn10_stream_encoder *enc1,
+	struct dc_context *ctx,
+	struct dc_bios *bp,
+	enum engine_id eng_id,
+	const struct dcn10_stream_enc_registers *regs,
+	const struct dcn10_stream_encoder_shift *se_shift,
+	const struct dcn10_stream_encoder_mask *se_mask)
+{
+	enc1->base.funcs = &dcn20_str_enc_funcs;
+	enc1->base.ctx = ctx;
+	enc1->base.id = eng_id;
+	enc1->base.bp = bp;
+	enc1->regs = regs;
+	enc1->se_shift = se_shift;
+	enc1->se_mask = se_mask;
+	enc1->base.stream_enc_inst = eng_id - ENGINE_ID_DIGA;
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn20/dcn20_stream_encoder.h b/drivers/gpu/drm/amd/display/dc/dio/dcn20/dcn20_stream_encoder.h
new file mode 100644
index 000000000000..baa1e539f341
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn20/dcn20_stream_encoder.h
@@ -0,0 +1,119 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_STREAM_ENCODER_DCN20_H__
+#define __DC_STREAM_ENCODER_DCN20_H__
+
+#include "stream_encoder.h"
+#include "dcn10/dcn10_stream_encoder.h"
+
+
+#define SE_DCN2_REG_LIST(id)\
+	SE_COMMON_DCN_REG_LIST(id),\
+	SRI(HDMI_GENERIC_PACKET_CONTROL4, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL5, DIG, id), \
+	SRI(DP_DSC_CNTL, DP, id), \
+	SRI(DP_DSC_BYTES_PER_PIXEL, DP, id), \
+	SRI(DME_CONTROL, DIG, id),\
+	SRI(DP_SEC_METADATA_TRANSMISSION, DP, id), \
+	SRI(HDMI_METADATA_PACKET_CONTROL, DIG, id), \
+	SRI(DP_SEC_FRAMING4, DP, id)
+
+#define SE_COMMON_MASK_SH_LIST_DCN20(mask_sh)\
+	SE_COMMON_MASK_SH_LIST_SOC(mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC0_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC0_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC1_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC1_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC2_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC2_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC3_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC3_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC4_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC4_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC5_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC5_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC6_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC6_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC7_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC7_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL1, HDMI_GENERIC0_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL1, HDMI_GENERIC1_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL2, HDMI_GENERIC2_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL2, HDMI_GENERIC3_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL3, HDMI_GENERIC4_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL3, HDMI_GENERIC5_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL4, HDMI_GENERIC6_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL4, HDMI_GENERIC7_LINE, mask_sh),\
+	SE_SF(DP0_DP_DSC_CNTL, DP_DSC_MODE, mask_sh),\
+	SE_SF(DP0_DP_DSC_CNTL, DP_DSC_SLICE_WIDTH, mask_sh),\
+	SE_SF(DP0_DP_DSC_BYTES_PER_PIXEL, DP_DSC_BYTES_PER_PIXEL, mask_sh),\
+	SE_SF(DP0_DP_MSA_VBID_MISC, DP_VBID6_LINE_REFERENCE, mask_sh),\
+	SE_SF(DP0_DP_MSA_VBID_MISC, DP_VBID6_LINE_NUM, mask_sh),\
+	SE_SF(DIG0_DME_CONTROL, METADATA_ENGINE_EN, mask_sh),\
+	SE_SF(DIG0_DME_CONTROL, METADATA_HUBP_REQUESTOR_ID, mask_sh),\
+	SE_SF(DIG0_DME_CONTROL, METADATA_STREAM_TYPE, mask_sh),\
+	SE_SF(DP0_DP_SEC_METADATA_TRANSMISSION, DP_SEC_METADATA_PACKET_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_METADATA_TRANSMISSION, DP_SEC_METADATA_PACKET_LINE_REFERENCE, mask_sh),\
+	SE_SF(DP0_DP_SEC_METADATA_TRANSMISSION, DP_SEC_METADATA_PACKET_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_METADATA_PACKET_CONTROL, HDMI_METADATA_PACKET_ENABLE, mask_sh),\
+	SE_SF(DIG0_HDMI_METADATA_PACKET_CONTROL, HDMI_METADATA_PACKET_LINE_REFERENCE, mask_sh),\
+	SE_SF(DIG0_HDMI_METADATA_PACKET_CONTROL, HDMI_METADATA_PACKET_LINE, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DOLBY_VISION_EN, mask_sh),\
+	SE_SF(DP0_DP_PIXEL_FORMAT, DP_PIXEL_COMBINE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL1, DP_SEC_GSP5_LINE_REFERENCE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL5, DP_SEC_GSP5_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_SEC_FRAMING4, DP_SST_SDP_SPLITTING, mask_sh)
+
+void dcn20_stream_encoder_construct(
+	struct dcn10_stream_encoder *enc1,
+	struct dc_context *ctx,
+	struct dc_bios *bp,
+	enum engine_id eng_id,
+	const struct dcn10_stream_enc_registers *regs,
+	const struct dcn10_stream_encoder_shift *se_shift,
+	const struct dcn10_stream_encoder_mask *se_mask);
+
+void enc2_stream_encoder_dp_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	enum dc_color_space output_color_space,
+	bool use_vsc_sdp_for_colorimetry,
+	uint32_t enable_sdp_splitting);
+
+void enc2_stream_encoder_dp_unblank(
+	struct dc_link *link,
+	struct stream_encoder *enc,
+	const struct encoder_unblank_param *param);
+
+void enc2_set_dynamic_metadata(struct stream_encoder *enc,
+		bool enable_dme,
+		uint32_t hubp_requestor_id,
+		enum dynamic_metadata_mode dmdata_mode);
+
+uint32_t enc2_get_fifo_cal_average_level(
+		struct stream_encoder *enc);
+
+#endif /* __DC_STREAM_ENCODER_DCN20_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn30/dcn30_dio_link_encoder.c b/drivers/gpu/drm/amd/display/dc/dio/dcn30/dcn30_dio_link_encoder.c
new file mode 100644
index 000000000000..b8e31b5ea114
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn30/dcn30_dio_link_encoder.c
@@ -0,0 +1,257 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+
+#include "core_types.h"
+#include "link_encoder.h"
+#include "dcn30_dio_link_encoder.h"
+#include "stream_encoder.h"
+#include "dc_bios_types.h"
+#include "gpio_service_interface.h"
+
+#define CTX \
+	enc10->base.ctx
+#define DC_LOGGER \
+	enc10->base.ctx->logger
+
+#define REG(reg)\
+	(enc10->link_regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	enc10->link_shift->field_name, enc10->link_mask->field_name
+
+#define IND_REG(index) \
+	(enc10->link_regs->index)
+
+
+bool dcn30_link_encoder_validate_output_with_stream(
+	struct link_encoder *enc,
+	const struct dc_stream_state *stream)
+{
+		return dcn10_link_encoder_validate_output_with_stream(enc, stream);
+}
+
+static const struct link_encoder_funcs dcn30_link_enc_funcs = {
+	.read_state = link_enc2_read_state,
+	.validate_output_with_stream =
+			dcn30_link_encoder_validate_output_with_stream,
+	.hw_init = enc3_hw_init,
+	.setup = dcn10_link_encoder_setup,
+	.enable_tmds_output = dcn10_link_encoder_enable_tmds_output,
+	.enable_dp_output = dcn20_link_encoder_enable_dp_output,
+	.enable_dp_mst_output = dcn10_link_encoder_enable_dp_mst_output,
+	.disable_output = dcn10_link_encoder_disable_output,
+	.dp_set_lane_settings = dcn10_link_encoder_dp_set_lane_settings,
+	.dp_set_phy_pattern = dcn10_link_encoder_dp_set_phy_pattern,
+	.update_mst_stream_allocation_table =
+		dcn10_link_encoder_update_mst_stream_allocation_table,
+	.psr_program_dp_dphy_fast_training =
+			dcn10_psr_program_dp_dphy_fast_training,
+	.psr_program_secondary_packet = dcn10_psr_program_secondary_packet,
+	.connect_dig_be_to_fe = dcn10_link_encoder_connect_dig_be_to_fe,
+	.enable_hpd = dcn10_link_encoder_enable_hpd,
+	.disable_hpd = dcn10_link_encoder_disable_hpd,
+	.is_dig_enabled = dcn10_is_dig_enabled,
+	.destroy = dcn10_link_encoder_destroy,
+	.fec_set_enable = enc2_fec_set_enable,
+	.fec_set_ready = enc2_fec_set_ready,
+	.fec_is_active = enc2_fec_is_active,
+	.get_dig_frontend = dcn10_get_dig_frontend,
+	.get_dig_mode = dcn10_get_dig_mode,
+	.is_in_alt_mode = dcn20_link_encoder_is_in_alt_mode,
+	.get_max_link_cap = dcn20_link_encoder_get_max_link_cap,
+};
+
+void dcn30_link_encoder_construct(
+	struct dcn20_link_encoder *enc20,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	const struct dcn10_link_enc_aux_registers *aux_regs,
+	const struct dcn10_link_enc_hpd_registers *hpd_regs,
+	const struct dcn10_link_enc_shift *link_shift,
+	const struct dcn10_link_enc_mask *link_mask)
+{
+	struct bp_encoder_cap_info bp_cap_info = {0};
+	const struct dc_vbios_funcs *bp_funcs = init_data->ctx->dc_bios->funcs;
+	enum bp_result result = BP_RESULT_OK;
+	struct dcn10_link_encoder *enc10 = &enc20->enc10;
+
+	enc10->base.funcs = &dcn30_link_enc_funcs;
+	enc10->base.ctx = init_data->ctx;
+	enc10->base.id = init_data->encoder;
+
+	enc10->base.hpd_source = init_data->hpd_source;
+	enc10->base.connector = init_data->connector;
+
+	enc10->base.preferred_engine = ENGINE_ID_UNKNOWN;
+
+	enc10->base.features = *enc_features;
+
+	enc10->base.transmitter = init_data->transmitter;
+
+	/* set the flag to indicate whether driver poll the I2C data pin
+	 * while doing the DP sink detect
+	 */
+
+/*	if (dal_adapter_service_is_feature_supported(as,
+		FEATURE_DP_SINK_DETECT_POLL_DATA_PIN))
+		enc10->base.features.flags.bits.
+			DP_SINK_DETECT_POLL_DATA_PIN = true;*/
+
+	enc10->base.output_signals =
+		SIGNAL_TYPE_DVI_SINGLE_LINK |
+		SIGNAL_TYPE_DVI_DUAL_LINK |
+		SIGNAL_TYPE_LVDS |
+		SIGNAL_TYPE_DISPLAY_PORT |
+		SIGNAL_TYPE_DISPLAY_PORT_MST |
+		SIGNAL_TYPE_EDP |
+		SIGNAL_TYPE_HDMI_TYPE_A;
+
+	/* For DCE 8.0 and 8.1, by design, UNIPHY is hardwired to DIG_BE.
+	 * SW always assign DIG_FE 1:1 mapped to DIG_FE for non-MST UNIPHY.
+	 * SW assign DIG_FE to non-MST UNIPHY first and MST last. So prefer
+	 * DIG is per UNIPHY and used by SST DP, eDP, HDMI, DVI and LVDS.
+	 * Prefer DIG assignment is decided by board design.
+	 * For DCE 8.0, there are only max 6 UNIPHYs, we assume board design
+	 * and VBIOS will filter out 7 UNIPHY for DCE 8.0.
+	 * By this, adding DIGG should not hurt DCE 8.0.
+	 * This will let DCE 8.1 share DCE 8.0 as much as possible
+	 */
+
+	enc10->link_regs = link_regs;
+	enc10->aux_regs = aux_regs;
+	enc10->hpd_regs = hpd_regs;
+	enc10->link_shift = link_shift;
+	enc10->link_mask = link_mask;
+
+	switch (enc10->base.transmitter) {
+	case TRANSMITTER_UNIPHY_A:
+		enc10->base.preferred_engine = ENGINE_ID_DIGA;
+	break;
+	case TRANSMITTER_UNIPHY_B:
+		enc10->base.preferred_engine = ENGINE_ID_DIGB;
+	break;
+	case TRANSMITTER_UNIPHY_C:
+		enc10->base.preferred_engine = ENGINE_ID_DIGC;
+	break;
+	case TRANSMITTER_UNIPHY_D:
+		enc10->base.preferred_engine = ENGINE_ID_DIGD;
+	break;
+	case TRANSMITTER_UNIPHY_E:
+		enc10->base.preferred_engine = ENGINE_ID_DIGE;
+	break;
+	case TRANSMITTER_UNIPHY_F:
+		enc10->base.preferred_engine = ENGINE_ID_DIGF;
+	break;
+	case TRANSMITTER_UNIPHY_G:
+		enc10->base.preferred_engine = ENGINE_ID_DIGG;
+	break;
+	default:
+		ASSERT_CRITICAL(false);
+		enc10->base.preferred_engine = ENGINE_ID_UNKNOWN;
+	}
+
+	/* default to one to mirror Windows behavior */
+	enc10->base.features.flags.bits.HDMI_6GB_EN = 1;
+
+	result = bp_funcs->get_encoder_cap_info(enc10->base.ctx->dc_bios,
+						enc10->base.id, &bp_cap_info);
+
+	/* Override features with DCE-specific values */
+	if (result == BP_RESULT_OK) {
+		enc10->base.features.flags.bits.IS_HBR2_CAPABLE =
+				bp_cap_info.DP_HBR2_EN;
+		enc10->base.features.flags.bits.IS_HBR3_CAPABLE =
+				bp_cap_info.DP_HBR3_EN;
+		enc10->base.features.flags.bits.HDMI_6GB_EN = bp_cap_info.HDMI_6GB_EN;
+		enc10->base.features.flags.bits.IS_DP2_CAPABLE = bp_cap_info.IS_DP2_CAPABLE;
+		enc10->base.features.flags.bits.IS_UHBR10_CAPABLE = bp_cap_info.DP_UHBR10_EN;
+		enc10->base.features.flags.bits.IS_UHBR13_5_CAPABLE = bp_cap_info.DP_UHBR13_5_EN;
+		enc10->base.features.flags.bits.IS_UHBR20_CAPABLE = bp_cap_info.DP_UHBR20_EN;
+		enc10->base.features.flags.bits.DP_IS_USB_C =
+				bp_cap_info.DP_IS_USB_C;
+	} else {
+		DC_LOG_WARNING("%s: Failed to get encoder_cap_info from VBIOS with error code %d!\n",
+				__func__,
+				result);
+	}
+	if (enc10->base.ctx->dc->debug.hdmi20_disable) {
+		enc10->base.features.flags.bits.HDMI_6GB_EN = 0;
+	}
+}
+
+#define AUX_REG(reg)\
+	(enc10->aux_regs->reg)
+
+#define AUX_REG_READ(reg_name) \
+		dm_read_reg(CTX, AUX_REG(reg_name))
+
+#define AUX_REG_WRITE(reg_name, val) \
+			dm_write_reg(CTX, AUX_REG(reg_name), val)
+void enc3_hw_init(struct link_encoder *enc)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+/*
+	00 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__1to2 : 1/2
+	01 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__3to4 : 3/4
+	02 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__7to8 : 7/8
+	03 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__15to16 : 15/16
+	04 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__31to32 : 31/32
+	05 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__63to64 : 63/64
+	06 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__127to128 : 127/128
+	07 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__255to256 : 255/256
+*/
+
+/*
+	AUX_REG_UPDATE_5(AUX_DPHY_RX_CONTROL0,
+	AUX_RX_START_WINDOW = 1 [6:4]
+	AUX_RX_RECEIVE_WINDOW = 1 default is 2 [10:8]
+	AUX_RX_HALF_SYM_DETECT_LEN  = 1 [13:12] default is 1
+	AUX_RX_TRANSITION_FILTER_EN = 1 [16] default is 1
+	AUX_RX_ALLOW_BELOW_THRESHOLD_PHASE_DETECT [17] is 0  default is 0
+	AUX_RX_ALLOW_BELOW_THRESHOLD_START [18] is 1  default is 1
+	AUX_RX_ALLOW_BELOW_THRESHOLD_STOP [19] is 1  default is 1
+	AUX_RX_PHASE_DETECT_LEN,  [21,20] = 0x3 default is 3
+	AUX_RX_DETECTION_THRESHOLD [30:28] = 1
+*/
+	AUX_REG_WRITE(AUX_DPHY_RX_CONTROL0, 0x103d1110);
+
+	AUX_REG_WRITE(AUX_DPHY_TX_CONTROL, 0x21c7a);
+
+	//AUX_DPHY_TX_REF_CONTROL'AUX_TX_REF_DIV HW default is 0x32;
+	// Set AUX_TX_REF_DIV Divider to generate 2 MHz reference from refclk
+	// 27MHz -> 0xd
+	// 100MHz -> 0x32
+	// 48MHz -> 0x18
+
+	// Set TMDS_CTL0 to 1.  This is a legacy setting.
+	REG_UPDATE(TMDS_CTL_BITS, TMDS_CTL0, 1);
+
+	dcn10_aux_initialize(enc10);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn30/dcn30_dio_link_encoder.h b/drivers/gpu/drm/amd/display/dc/dio/dcn30/dcn30_dio_link_encoder.h
new file mode 100644
index 000000000000..5b6177c2ae98
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn30/dcn30_dio_link_encoder.h
@@ -0,0 +1,86 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_LINK_ENCODER__DCN30_H__
+#define __DC_LINK_ENCODER__DCN30_H__
+
+#include "dcn20/dcn20_link_encoder.h"
+
+#define LE_DCN3_REG_LIST(id)\
+	SRI(DIG_BE_CNTL, DIG, id), \
+	SRI(DIG_BE_EN_CNTL, DIG, id), \
+	SRI(TMDS_CTL_BITS, DIG, id), \
+	SRI(TMDS_DCBALANCER_CONTROL, DIG, id), \
+	SRI(DP_CONFIG, DP, id), \
+	SRI(DP_DPHY_CNTL, DP, id), \
+	SRI(DP_DPHY_PRBS_CNTL, DP, id), \
+	SRI(DP_DPHY_SCRAM_CNTL, DP, id),\
+	SRI(DP_DPHY_SYM0, DP, id), \
+	SRI(DP_DPHY_SYM1, DP, id), \
+	SRI(DP_DPHY_SYM2, DP, id), \
+	SRI(DP_DPHY_TRAINING_PATTERN_SEL, DP, id), \
+	SRI(DP_LINK_CNTL, DP, id), \
+	SRI(DP_LINK_FRAMING_CNTL, DP, id), \
+	SRI(DP_MSE_SAT0, DP, id), \
+	SRI(DP_MSE_SAT1, DP, id), \
+	SRI(DP_MSE_SAT2, DP, id), \
+	SRI(DP_MSE_SAT_UPDATE, DP, id), \
+	SRI(DP_SEC_CNTL, DP, id), \
+	SRI(DP_VID_STREAM_CNTL, DP, id), \
+	SRI(DP_DPHY_FAST_TRAINING, DP, id), \
+	SRI(DP_SEC_CNTL1, DP, id), \
+	SRI(DP_DPHY_BS_SR_SWAP_CNTL, DP, id), \
+	SRI(DP_DPHY_HBR2_PATTERN_CONTROL, DP, id)
+
+#define LINK_ENCODER_MASK_SH_LIST_DCN30(mask_sh) \
+	LINK_ENCODER_MASK_SH_LIST_DCN20(mask_sh),\
+	LE_SF(DIG0_TMDS_DCBALANCER_CONTROL, TMDS_SYNC_DCBAL_EN, mask_sh)
+
+#define DPCS_DCN3_MASK_SH_LIST(mask_sh)\
+	DPCS_DCN2_MASK_SH_LIST(mask_sh),\
+	LE_SF(DPCSTX0_DPCSTX_TX_CNTL, DPCS_TX_DATA_ORDER_INVERT_18_BIT, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL0, RDPCS_PHY_TX_VBOOST_LVL, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CLOCK_CNTL, RDPCS_TX_CLK_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DPALT_DP4, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DPALT_DISABLE, mask_sh)
+
+
+void dcn30_link_encoder_construct(
+	struct dcn20_link_encoder *enc20,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	const struct dcn10_link_enc_aux_registers *aux_regs,
+	const struct dcn10_link_enc_hpd_registers *hpd_regs,
+	const struct dcn10_link_enc_shift *link_shift,
+	const struct dcn10_link_enc_mask *link_mask);
+
+void enc3_hw_init(struct link_encoder *enc);
+
+bool dcn30_link_encoder_validate_output_with_stream(
+	struct link_encoder *enc,
+	const struct dc_stream_state *stream);
+
+#endif /* __DC_LINK_ENCODER__DCN30_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn30/dcn30_dio_stream_encoder.c b/drivers/gpu/drm/amd/display/dc/dio/dcn30/dcn30_dio_stream_encoder.c
new file mode 100644
index 000000000000..e93be7b6d9b0
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn30/dcn30_dio_stream_encoder.c
@@ -0,0 +1,899 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+#include "dc_bios_types.h"
+#include "dcn30_dio_stream_encoder.h"
+#include "reg_helper.h"
+#include "hw_shared.h"
+#include "dc.h"
+
+#define DC_LOGGER \
+		enc1->base.ctx->logger
+
+#define REG(reg)\
+	(enc1->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	enc1->se_shift->field_name, enc1->se_mask->field_name
+
+#define VBI_LINE_0 0
+#define HDMI_CLOCK_CHANNEL_RATE_MORE_340M 340000
+
+#define CTX \
+	enc1->base.ctx
+
+
+void enc3_update_hdmi_info_packet(
+	struct dcn10_stream_encoder *enc1,
+	uint32_t packet_index,
+	const struct dc_info_packet *info_packet)
+{
+	uint32_t cont, send, line;
+
+	if (info_packet->valid) {
+		enc1->base.vpg->funcs->update_generic_info_packet(
+				enc1->base.vpg,
+				packet_index,
+				info_packet,
+				true);
+
+		/* enable transmission of packet(s) -
+		 * packet transmission begins on the next frame */
+		cont = 1;
+		/* send packet(s) every frame */
+		send = 1;
+		/* select line number to send packets on */
+		line = 2;
+	} else {
+		cont = 0;
+		send = 0;
+		line = 0;
+	}
+
+	/* DP_SEC_GSP[x]_LINE_REFERENCE - keep default value REFER_TO_DP_SOF */
+
+	/* choose which generic packet control to use */
+	switch (packet_index) {
+	case 0:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL0,
+				HDMI_GENERIC0_CONT, cont,
+				HDMI_GENERIC0_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL1,
+				HDMI_GENERIC0_LINE, line);
+		break;
+	case 1:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL0,
+				HDMI_GENERIC1_CONT, cont,
+				HDMI_GENERIC1_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL1,
+				HDMI_GENERIC1_LINE, line);
+		break;
+	case 2:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL0,
+				HDMI_GENERIC2_CONT, cont,
+				HDMI_GENERIC2_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL2,
+				HDMI_GENERIC2_LINE, line);
+		break;
+	case 3:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL0,
+				HDMI_GENERIC3_CONT, cont,
+				HDMI_GENERIC3_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL2,
+				HDMI_GENERIC3_LINE, line);
+		break;
+	case 4:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL0,
+				HDMI_GENERIC4_CONT, cont,
+				HDMI_GENERIC4_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL3,
+				HDMI_GENERIC4_LINE, line);
+		break;
+	case 5:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL0,
+				HDMI_GENERIC5_CONT, cont,
+				HDMI_GENERIC5_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL3,
+				HDMI_GENERIC5_LINE, line);
+		break;
+	case 6:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL0,
+				HDMI_GENERIC6_CONT, cont,
+				HDMI_GENERIC6_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL4,
+				HDMI_GENERIC6_LINE, line);
+		break;
+	case 7:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL0,
+				HDMI_GENERIC7_CONT, cont,
+				HDMI_GENERIC7_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL4,
+				HDMI_GENERIC7_LINE, line);
+		break;
+	case 8:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL6,
+				HDMI_GENERIC8_CONT, cont,
+				HDMI_GENERIC8_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL7,
+				HDMI_GENERIC8_LINE, line);
+		break;
+	case 9:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL6,
+				HDMI_GENERIC9_CONT, cont,
+				HDMI_GENERIC9_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL7,
+				HDMI_GENERIC9_LINE, line);
+		break;
+	case 10:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL6,
+				HDMI_GENERIC10_CONT, cont,
+				HDMI_GENERIC10_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL8,
+				HDMI_GENERIC10_LINE, line);
+		break;
+	case 11:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL6,
+				HDMI_GENERIC11_CONT, cont,
+				HDMI_GENERIC11_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL8,
+				HDMI_GENERIC11_LINE, line);
+		break;
+	case 12:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL6,
+				HDMI_GENERIC12_CONT, cont,
+				HDMI_GENERIC12_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL9,
+				HDMI_GENERIC12_LINE, line);
+		break;
+	case 13:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL6,
+				HDMI_GENERIC13_CONT, cont,
+				HDMI_GENERIC13_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL9,
+				HDMI_GENERIC13_LINE, line);
+		break;
+	case 14:
+		REG_UPDATE_2(HDMI_GENERIC_PACKET_CONTROL6,
+				HDMI_GENERIC14_CONT, cont,
+				HDMI_GENERIC14_SEND, send);
+		REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL10,
+				HDMI_GENERIC14_LINE, line);
+		break;
+	default:
+		/* invalid HW packet index */
+		DC_LOG_WARNING(
+			"Invalid HW packet index: %s()\n",
+			__func__);
+		return;
+	}
+}
+
+void enc3_stream_encoder_update_hdmi_info_packets(
+	struct stream_encoder *enc,
+	const struct encoder_info_frame *info_frame)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	/* for bring up, disable dp double  TODO */
+	REG_UPDATE(HDMI_DB_CONTROL, HDMI_DB_DISABLE, 1);
+	REG_UPDATE(AFMT_CNTL, AFMT_AUDIO_CLOCK_EN, 1);
+
+	/*Always add mandatory packets first followed by optional ones*/
+	enc3_update_hdmi_info_packet(enc1, 0, &info_frame->avi);
+	enc3_update_hdmi_info_packet(enc1, 5, &info_frame->hfvsif);
+	enc3_update_hdmi_info_packet(enc1, 2, &info_frame->gamut);
+	enc3_update_hdmi_info_packet(enc1, 1, &info_frame->vendor);
+	enc3_update_hdmi_info_packet(enc1, 3, &info_frame->spd);
+	enc3_update_hdmi_info_packet(enc1, 4, &info_frame->hdrsmd);
+	enc3_update_hdmi_info_packet(enc1, 6, &info_frame->vtem);
+}
+
+void enc3_stream_encoder_stop_hdmi_info_packets(
+	struct stream_encoder *enc)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	/* stop generic packets 0,1 on HDMI */
+	REG_SET_4(HDMI_GENERIC_PACKET_CONTROL0, 0,
+		HDMI_GENERIC0_CONT, 0,
+		HDMI_GENERIC0_SEND, 0,
+		HDMI_GENERIC1_CONT, 0,
+		HDMI_GENERIC1_SEND, 0);
+	REG_SET_2(HDMI_GENERIC_PACKET_CONTROL1, 0,
+		HDMI_GENERIC0_LINE, 0,
+		HDMI_GENERIC1_LINE, 0);
+
+	/* stop generic packets 2,3 on HDMI */
+	REG_SET_4(HDMI_GENERIC_PACKET_CONTROL0, 0,
+		HDMI_GENERIC2_CONT, 0,
+		HDMI_GENERIC2_SEND, 0,
+		HDMI_GENERIC3_CONT, 0,
+		HDMI_GENERIC3_SEND, 0);
+	REG_SET_2(HDMI_GENERIC_PACKET_CONTROL2, 0,
+		HDMI_GENERIC2_LINE, 0,
+		HDMI_GENERIC3_LINE, 0);
+
+	/* stop generic packets 4,5 on HDMI */
+	REG_SET_4(HDMI_GENERIC_PACKET_CONTROL0, 0,
+		HDMI_GENERIC4_CONT, 0,
+		HDMI_GENERIC4_SEND, 0,
+		HDMI_GENERIC5_CONT, 0,
+		HDMI_GENERIC5_SEND, 0);
+	REG_SET_2(HDMI_GENERIC_PACKET_CONTROL3, 0,
+		HDMI_GENERIC4_LINE, 0,
+		HDMI_GENERIC5_LINE, 0);
+
+	/* stop generic packets 6,7 on HDMI */
+	REG_SET_4(HDMI_GENERIC_PACKET_CONTROL0, 0,
+		HDMI_GENERIC6_CONT, 0,
+		HDMI_GENERIC6_SEND, 0,
+		HDMI_GENERIC7_CONT, 0,
+		HDMI_GENERIC7_SEND, 0);
+	REG_SET_2(HDMI_GENERIC_PACKET_CONTROL4, 0,
+		HDMI_GENERIC6_LINE, 0,
+		HDMI_GENERIC7_LINE, 0);
+
+	/* stop generic packets 8,9 on HDMI */
+	REG_SET_4(HDMI_GENERIC_PACKET_CONTROL6, 0,
+		HDMI_GENERIC8_CONT, 0,
+		HDMI_GENERIC8_SEND, 0,
+		HDMI_GENERIC9_CONT, 0,
+		HDMI_GENERIC9_SEND, 0);
+	REG_SET_2(HDMI_GENERIC_PACKET_CONTROL7, 0,
+		HDMI_GENERIC8_LINE, 0,
+		HDMI_GENERIC9_LINE, 0);
+
+	/* stop generic packets 10,11 on HDMI */
+	REG_SET_4(HDMI_GENERIC_PACKET_CONTROL6, 0,
+		HDMI_GENERIC10_CONT, 0,
+		HDMI_GENERIC10_SEND, 0,
+		HDMI_GENERIC11_CONT, 0,
+		HDMI_GENERIC11_SEND, 0);
+	REG_SET_2(HDMI_GENERIC_PACKET_CONTROL8, 0,
+		HDMI_GENERIC10_LINE, 0,
+		HDMI_GENERIC11_LINE, 0);
+
+	/* stop generic packets 12,13 on HDMI */
+	REG_SET_4(HDMI_GENERIC_PACKET_CONTROL6, 0,
+		HDMI_GENERIC12_CONT, 0,
+		HDMI_GENERIC12_SEND, 0,
+		HDMI_GENERIC13_CONT, 0,
+		HDMI_GENERIC13_SEND, 0);
+	REG_SET_2(HDMI_GENERIC_PACKET_CONTROL9, 0,
+		HDMI_GENERIC12_LINE, 0,
+		HDMI_GENERIC13_LINE, 0);
+
+	/* stop generic packet 14 on HDMI */
+	REG_SET_2(HDMI_GENERIC_PACKET_CONTROL6, 0,
+		HDMI_GENERIC14_CONT, 0,
+		HDMI_GENERIC14_SEND, 0);
+	REG_UPDATE(HDMI_GENERIC_PACKET_CONTROL10,
+		HDMI_GENERIC14_LINE, 0);
+}
+
+/* Set DSC-related configuration.
+ *   dsc_mode: 0 disables DSC, other values enable DSC in specified format
+ *   sc_bytes_per_pixel: Bytes per pixel in u3.28 format
+ *   dsc_slice_width: Slice width in pixels
+ */
+static void enc3_dp_set_dsc_config(struct stream_encoder *enc,
+					enum optc_dsc_mode dsc_mode,
+					uint32_t dsc_bytes_per_pixel,
+					uint32_t dsc_slice_width)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	REG_UPDATE_2(DP_DSC_CNTL,
+			DP_DSC_MODE, dsc_mode,
+			DP_DSC_SLICE_WIDTH, dsc_slice_width);
+
+	REG_SET(DP_DSC_BYTES_PER_PIXEL, 0,
+		DP_DSC_BYTES_PER_PIXEL, dsc_bytes_per_pixel);
+}
+
+
+void enc3_dp_set_dsc_pps_info_packet(struct stream_encoder *enc,
+					bool enable,
+					uint8_t *dsc_packed_pps,
+					bool immediate_update)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (enable) {
+		struct dc_info_packet pps_sdp;
+		int i;
+
+		/* Configure for PPS packet size (128 bytes) */
+		REG_UPDATE(DP_SEC_CNTL2, DP_SEC_GSP11_PPS, 1);
+
+		/* We need turn on clock before programming AFMT block
+		 *
+		 * TODO: We may not need this here anymore since update_generic_info_packet
+		 * no longer touches AFMT
+		 */
+		REG_UPDATE(AFMT_CNTL, AFMT_AUDIO_CLOCK_EN, 1);
+
+		/* Load PPS into infoframe (SDP) registers */
+		pps_sdp.valid = true;
+		pps_sdp.hb0 = 0;
+		pps_sdp.hb1 = DC_DP_INFOFRAME_TYPE_PPS;
+		pps_sdp.hb2 = 127;
+		pps_sdp.hb3 = 0;
+
+		for (i = 0; i < 4; i++) {
+			memcpy(pps_sdp.sb, &dsc_packed_pps[i * 32], 32);
+			enc1->base.vpg->funcs->update_generic_info_packet(
+							enc1->base.vpg,
+							11 + i,
+							&pps_sdp,
+							immediate_update);
+		}
+
+		/* SW should make sure VBID[6] update line number is bigger
+		 * than PPS transmit line number
+		 */
+		REG_UPDATE(DP_GSP11_CNTL,
+				DP_SEC_GSP11_LINE_NUM, 2);
+		REG_UPDATE_2(DP_MSA_VBID_MISC,
+				DP_VBID6_LINE_REFERENCE, 0,
+				DP_VBID6_LINE_NUM, 3);
+
+		/* Send PPS data at the line number specified above.
+		 * DP spec requires PPS to be sent only when it changes, however since
+		 * decoder has to be able to handle its change on every frame, we're
+		 * sending it always (i.e. on every frame) to reduce the chance it'd be
+		 * missed by decoder. If it turns out required to send PPS only when it
+		 * changes, we can use DP_SEC_GSP11_SEND register.
+		 */
+		REG_UPDATE(DP_GSP11_CNTL,
+			DP_SEC_GSP11_ENABLE, 1);
+		REG_UPDATE(DP_SEC_CNTL,
+			DP_SEC_STREAM_ENABLE, 1);
+	} else {
+		/* Disable Generic Stream Packet 11 (GSP) transmission */
+		REG_UPDATE(DP_GSP11_CNTL, DP_SEC_GSP11_ENABLE, 0);
+		REG_UPDATE(DP_SEC_CNTL2, DP_SEC_GSP11_PPS, 0);
+	}
+}
+
+
+/* this function read dsc related register fields to be logged later in dcn10_log_hw_state
+ * into a dcn_dsc_state struct.
+ */
+static void enc3_read_state(struct stream_encoder *enc, struct enc_state *s)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	//if dsc is enabled, continue to read
+	REG_GET(DP_DSC_CNTL, DP_DSC_MODE, &s->dsc_mode);
+	if (s->dsc_mode) {
+		REG_GET(DP_DSC_CNTL, DP_DSC_SLICE_WIDTH, &s->dsc_slice_width);
+		REG_GET(DP_GSP11_CNTL, DP_SEC_GSP11_LINE_NUM, &s->sec_gsp_pps_line_num);
+
+		REG_GET(DP_MSA_VBID_MISC, DP_VBID6_LINE_REFERENCE, &s->vbid6_line_reference);
+		REG_GET(DP_MSA_VBID_MISC, DP_VBID6_LINE_NUM, &s->vbid6_line_num);
+
+		REG_GET(DP_GSP11_CNTL, DP_SEC_GSP11_ENABLE, &s->sec_gsp_pps_enable);
+		REG_GET(DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, &s->sec_stream_enable);
+	}
+}
+
+void enc3_stream_encoder_update_dp_info_packets_sdp_line_num(
+		struct stream_encoder *enc,
+		struct encoder_info_frame *info_frame)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (info_frame->adaptive_sync.valid == true &&
+		info_frame->sdp_line_num.adaptive_sync_line_num_valid == true) {
+		//00: REFER_TO_DP_SOF, 01: REFER_TO_OTG_SOF
+		REG_UPDATE(DP_SEC_CNTL1, DP_SEC_GSP5_LINE_REFERENCE, 1);
+
+		REG_UPDATE(DP_SEC_CNTL5, DP_SEC_GSP5_LINE_NUM,
+					info_frame->sdp_line_num.adaptive_sync_line_num);
+	}
+}
+
+void enc3_stream_encoder_update_dp_info_packets(
+	struct stream_encoder *enc,
+	const struct encoder_info_frame *info_frame)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+	uint32_t value = 0;
+	uint32_t dmdata_packet_enabled = 0;
+
+	if (info_frame->vsc.valid) {
+		enc->vpg->funcs->update_generic_info_packet(
+				enc->vpg,
+				0,  /* packetIndex */
+				&info_frame->vsc,
+				true);
+	}
+	/* TODO: VSC SDP at packetIndex 1 should be retricted only if PSR-SU on.
+	 * There should have another Infopacket type (e.g. vsc_psrsu) for PSR_SU.
+	 * In addition, currently the driver check the valid bit then update and
+	 * send the corresponding Infopacket. For PSR-SU, the SDP only be sent
+	 * while entering PSR-SU mode. So we need another parameter(e.g. send)
+	 * in dc_info_packet to indicate which infopacket should be enabled by
+	 * default here.
+	 */
+	if (info_frame->vsc.valid) {
+		enc->vpg->funcs->update_generic_info_packet(
+				enc->vpg,
+				1,  /* packetIndex */
+				&info_frame->vsc,
+				true);
+	}
+	/* TODO: VSC SDP at packetIndex 1 should be restricted only if PSR-SU on.
+	 * There should have another Infopacket type (e.g. vsc_psrsu) for PSR_SU.
+	 * In addition, currently the driver check the valid bit then update and
+	 * send the corresponding Infopacket. For PSR-SU, the SDP only be sent
+	 * while entering PSR-SU mode. So we need another parameter(e.g. send)
+	 * in dc_info_packet to indicate which infopacket should be enabled by
+	 * default here.
+	 */
+	if (info_frame->vsc.valid) {
+		enc->vpg->funcs->update_generic_info_packet(
+				enc->vpg,
+				1,  /* packetIndex */
+				&info_frame->vsc,
+				true);
+	}
+	if (info_frame->spd.valid) {
+		enc->vpg->funcs->update_generic_info_packet(
+				enc->vpg,
+				2,  /* packetIndex */
+				&info_frame->spd,
+				true);
+	}
+	if (info_frame->hdrsmd.valid) {
+		enc->vpg->funcs->update_generic_info_packet(
+				enc->vpg,
+				3,  /* packetIndex */
+				&info_frame->hdrsmd,
+				true);
+	}
+	/* packetIndex 4 is used for send immediate sdp message, and please
+	 * use other packetIndex (such as 5,6) for other info packet
+	 */
+
+	if (info_frame->adaptive_sync.valid)
+		enc->vpg->funcs->update_generic_info_packet(
+				enc->vpg,
+				5,  /* packetIndex */
+				&info_frame->adaptive_sync,
+				true);
+
+	/* enable/disable transmission of packet(s).
+	 * If enabled, packet transmission begins on the next frame
+	 */
+	REG_UPDATE(DP_SEC_CNTL, DP_SEC_GSP0_ENABLE, info_frame->vsc.valid);
+	REG_UPDATE(DP_SEC_CNTL, DP_SEC_GSP2_ENABLE, info_frame->spd.valid);
+	REG_UPDATE(DP_SEC_CNTL, DP_SEC_GSP3_ENABLE, info_frame->hdrsmd.valid);
+	REG_UPDATE(DP_SEC_CNTL, DP_SEC_GSP5_ENABLE, info_frame->adaptive_sync.valid);
+
+	/* This bit is the master enable bit.
+	 * When enabling secondary stream engine,
+	 * this master bit must also be set.
+	 * This register shared with audio info frame.
+	 * Therefore we need to enable master bit
+	 * if at least on of the fields is not 0
+	 */
+	value = REG_READ(DP_SEC_CNTL);
+	if (value)
+		REG_UPDATE(DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, 1);
+
+	/* check if dynamic metadata packet transmission is enabled */
+	REG_GET(DP_SEC_METADATA_TRANSMISSION,
+			DP_SEC_METADATA_PACKET_ENABLE, &dmdata_packet_enabled);
+
+	if (dmdata_packet_enabled)
+		REG_UPDATE(DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, 1);
+}
+
+static void enc3_dp_set_odm_combine(
+	struct stream_encoder *enc,
+	bool odm_combine)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	REG_UPDATE(DP_PIXEL_FORMAT, DP_PIXEL_COMBINE, odm_combine);
+}
+
+/* setup stream encoder in dvi mode */
+static void enc3_stream_encoder_dvi_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	bool is_dual_link)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (!enc->ctx->dc->debug.avoid_vbios_exec_table) {
+		struct bp_encoder_control cntl = {0};
+
+		cntl.action = ENCODER_CONTROL_SETUP;
+		cntl.engine_id = enc1->base.id;
+		cntl.signal = is_dual_link ?
+			SIGNAL_TYPE_DVI_DUAL_LINK : SIGNAL_TYPE_DVI_SINGLE_LINK;
+		cntl.enable_dp_audio = false;
+		cntl.pixel_clock = crtc_timing->pix_clk_100hz / 10;
+		cntl.lanes_number = (is_dual_link) ? LANE_COUNT_EIGHT : LANE_COUNT_FOUR;
+
+		if (enc1->base.bp->funcs->encoder_control(
+				enc1->base.bp, &cntl) != BP_RESULT_OK)
+			return;
+
+	} else {
+
+		//Set pattern for clock channel, default vlue 0x63 does not work
+		REG_UPDATE(DIG_CLOCK_PATTERN, DIG_CLOCK_PATTERN, 0x1F);
+
+		//DIG_BE_TMDS_DVI_MODE : TMDS-DVI mode is already set in link_encoder_setup
+
+		//DIG_SOURCE_SELECT is already set in dig_connect_to_otg
+
+		/* set DIG_START to 0x1 to reset FIFO */
+		REG_UPDATE(DIG_FE_CNTL, DIG_START, 1);
+		udelay(1);
+
+		/* write 0 to take the FIFO out of reset */
+		REG_UPDATE(DIG_FE_CNTL, DIG_START, 0);
+		udelay(1);
+	}
+
+	ASSERT(crtc_timing->pixel_encoding == PIXEL_ENCODING_RGB);
+	ASSERT(crtc_timing->display_color_depth == COLOR_DEPTH_888);
+	enc1_stream_encoder_set_stream_attribute_helper(enc1, crtc_timing);
+}
+
+/* setup stream encoder in hdmi mode */
+static void enc3_stream_encoder_hdmi_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	int actual_pix_clk_khz,
+	bool enable_audio)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (!enc->ctx->dc->debug.avoid_vbios_exec_table) {
+		struct bp_encoder_control cntl = {0};
+
+		cntl.action = ENCODER_CONTROL_SETUP;
+		cntl.engine_id = enc1->base.id;
+		cntl.signal = SIGNAL_TYPE_HDMI_TYPE_A;
+		cntl.enable_dp_audio = enable_audio;
+		cntl.pixel_clock = actual_pix_clk_khz;
+		cntl.lanes_number = LANE_COUNT_FOUR;
+
+		if (enc1->base.bp->funcs->encoder_control(
+				enc1->base.bp, &cntl) != BP_RESULT_OK)
+			return;
+
+	} else {
+
+		//Set pattern for clock channel, default vlue 0x63 does not work
+		REG_UPDATE(DIG_CLOCK_PATTERN, DIG_CLOCK_PATTERN, 0x1F);
+
+		//DIG_BE_TMDS_HDMI_MODE : TMDS-HDMI mode is already set in link_encoder_setup
+
+		//DIG_SOURCE_SELECT is already set in dig_connect_to_otg
+
+		/* set DIG_START to 0x1 to reset FIFO */
+		REG_UPDATE(DIG_FE_CNTL, DIG_START, 1);
+		udelay(1);
+
+		/* write 0 to take the FIFO out of reset */
+		REG_UPDATE(DIG_FE_CNTL, DIG_START, 0);
+		udelay(1);
+	}
+
+	/* Configure pixel encoding */
+	enc1_stream_encoder_set_stream_attribute_helper(enc1, crtc_timing);
+
+	/* setup HDMI engine */
+	REG_UPDATE_6(HDMI_CONTROL,
+		HDMI_PACKET_GEN_VERSION, 1,
+		HDMI_KEEPOUT_MODE, 1,
+		HDMI_DEEP_COLOR_ENABLE, 0,
+		HDMI_DATA_SCRAMBLE_EN, 0,
+		HDMI_NO_EXTRA_NULL_PACKET_FILLED, 1,
+		HDMI_CLOCK_CHANNEL_RATE, 0);
+
+	/* Configure color depth */
+	switch (crtc_timing->display_color_depth) {
+	case COLOR_DEPTH_888:
+		REG_UPDATE(HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 0);
+		break;
+	case COLOR_DEPTH_101010:
+		if (crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR422) {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 1,
+					HDMI_DEEP_COLOR_ENABLE, 0);
+		} else {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 1,
+					HDMI_DEEP_COLOR_ENABLE, 1);
+			}
+		break;
+	case COLOR_DEPTH_121212:
+		if (crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR422) {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 2,
+					HDMI_DEEP_COLOR_ENABLE, 0);
+		} else {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 2,
+					HDMI_DEEP_COLOR_ENABLE, 1);
+			}
+		break;
+	case COLOR_DEPTH_161616:
+		REG_UPDATE_2(HDMI_CONTROL,
+				HDMI_DEEP_COLOR_DEPTH, 3,
+				HDMI_DEEP_COLOR_ENABLE, 1);
+		break;
+	default:
+		break;
+	}
+
+	if (actual_pix_clk_khz >= HDMI_CLOCK_CHANNEL_RATE_MORE_340M) {
+		/* enable HDMI data scrambler
+		 * HDMI_CLOCK_CHANNEL_RATE_MORE_340M
+		 * Clock channel frequency is 1/4 of character rate.
+		 */
+		REG_UPDATE_2(HDMI_CONTROL,
+			HDMI_DATA_SCRAMBLE_EN, 1,
+			HDMI_CLOCK_CHANNEL_RATE, 1);
+	} else if (crtc_timing->flags.LTE_340MCSC_SCRAMBLE) {
+
+		/* TODO: New feature for DCE11, still need to implement */
+
+		/* enable HDMI data scrambler
+		 * HDMI_CLOCK_CHANNEL_FREQ_EQUAL_TO_CHAR_RATE
+		 * Clock channel frequency is the same
+		 * as character rate
+		 */
+		REG_UPDATE_2(HDMI_CONTROL,
+			HDMI_DATA_SCRAMBLE_EN, 1,
+			HDMI_CLOCK_CHANNEL_RATE, 0);
+	}
+
+
+	/* Enable transmission of General Control packet on every frame */
+	REG_UPDATE_3(HDMI_VBI_PACKET_CONTROL,
+		HDMI_GC_CONT, 1,
+		HDMI_GC_SEND, 1,
+		HDMI_NULL_SEND, 1);
+
+	/* Disable Audio Content Protection packet transmission */
+	REG_UPDATE(HDMI_VBI_PACKET_CONTROL, HDMI_ACP_SEND, 0);
+
+	/* following belongs to audio */
+	/* Enable Audio InfoFrame packet transmission. */
+	REG_UPDATE(HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, 1);
+
+	/* update double-buffered AUDIO_INFO registers immediately */
+	ASSERT (enc->afmt);
+	enc->afmt->funcs->audio_info_immediate_update(enc->afmt);
+
+	/* Select line number on which to send Audio InfoFrame packets */
+	REG_UPDATE(HDMI_INFOFRAME_CONTROL1, HDMI_AUDIO_INFO_LINE,
+				VBI_LINE_0 + 2);
+
+	/* set HDMI GC AVMUTE */
+	REG_UPDATE(HDMI_GC, HDMI_GC_AVMUTE, 0);
+}
+
+void enc3_audio_mute_control(
+	struct stream_encoder *enc,
+	bool mute)
+{
+	ASSERT (enc->afmt);
+	enc->afmt->funcs->audio_mute_control(enc->afmt, mute);
+}
+
+void enc3_se_dp_audio_setup(
+	struct stream_encoder *enc,
+	unsigned int az_inst,
+	struct audio_info *info)
+{
+	ASSERT (enc->afmt);
+	enc->afmt->funcs->se_audio_setup(enc->afmt, az_inst, info);
+}
+
+#define DP_SEC_AUD_N__DP_SEC_AUD_N__DEFAULT 0x8000
+#define DP_SEC_TIMESTAMP__DP_SEC_TIMESTAMP_MODE__AUTO_CALC 1
+
+static void enc3_se_setup_dp_audio(
+	struct stream_encoder *enc)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	/* --- DP Audio packet configurations --- */
+
+	/* ATP Configuration */
+	REG_SET(DP_SEC_AUD_N, 0,
+			DP_SEC_AUD_N, DP_SEC_AUD_N__DP_SEC_AUD_N__DEFAULT);
+
+	/* Async/auto-calc timestamp mode */
+	REG_SET(DP_SEC_TIMESTAMP, 0, DP_SEC_TIMESTAMP_MODE,
+			DP_SEC_TIMESTAMP__DP_SEC_TIMESTAMP_MODE__AUTO_CALC);
+
+	ASSERT (enc->afmt);
+	enc->afmt->funcs->setup_dp_audio(enc->afmt);
+}
+
+void enc3_se_dp_audio_enable(
+	struct stream_encoder *enc)
+{
+	enc1_se_enable_audio_clock(enc, true);
+	enc3_se_setup_dp_audio(enc);
+	enc1_se_enable_dp_audio(enc);
+}
+
+static void enc3_se_setup_hdmi_audio(
+	struct stream_encoder *enc,
+	const struct audio_crtc_info *crtc_info)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	struct audio_clock_info audio_clock_info = {0};
+
+	/* Setup audio in AFMT - program AFMT block associated with DIO */
+	ASSERT (enc->afmt);
+	enc->afmt->funcs->setup_hdmi_audio(enc->afmt);
+
+	/* HDMI_AUDIO_PACKET_CONTROL */
+	REG_UPDATE(HDMI_AUDIO_PACKET_CONTROL,
+			HDMI_AUDIO_DELAY_EN, 1);
+
+	/* HDMI_ACR_PACKET_CONTROL */
+	REG_UPDATE_3(HDMI_ACR_PACKET_CONTROL,
+			HDMI_ACR_AUTO_SEND, 1,
+			HDMI_ACR_SOURCE, 0,
+			HDMI_ACR_AUDIO_PRIORITY, 0);
+
+	/* Program audio clock sample/regeneration parameters */
+	get_audio_clock_info(crtc_info->color_depth,
+			     crtc_info->requested_pixel_clock_100Hz,
+			     crtc_info->calculated_pixel_clock_100Hz,
+			     &audio_clock_info);
+	DC_LOG_HW_AUDIO(
+			"\n%s:Input::requested_pixel_clock_100Hz = %d"	\
+			"calculated_pixel_clock_100Hz = %d \n", __func__,	\
+			crtc_info->requested_pixel_clock_100Hz,		\
+			crtc_info->calculated_pixel_clock_100Hz);
+
+	/* HDMI_ACR_32_0__HDMI_ACR_CTS_32_MASK */
+	REG_UPDATE(HDMI_ACR_32_0, HDMI_ACR_CTS_32, audio_clock_info.cts_32khz);
+
+	/* HDMI_ACR_32_1__HDMI_ACR_N_32_MASK */
+	REG_UPDATE(HDMI_ACR_32_1, HDMI_ACR_N_32, audio_clock_info.n_32khz);
+
+	/* HDMI_ACR_44_0__HDMI_ACR_CTS_44_MASK */
+	REG_UPDATE(HDMI_ACR_44_0, HDMI_ACR_CTS_44, audio_clock_info.cts_44khz);
+
+	/* HDMI_ACR_44_1__HDMI_ACR_N_44_MASK */
+	REG_UPDATE(HDMI_ACR_44_1, HDMI_ACR_N_44, audio_clock_info.n_44khz);
+
+	/* HDMI_ACR_48_0__HDMI_ACR_CTS_48_MASK */
+	REG_UPDATE(HDMI_ACR_48_0, HDMI_ACR_CTS_48, audio_clock_info.cts_48khz);
+
+	/* HDMI_ACR_48_1__HDMI_ACR_N_48_MASK */
+	REG_UPDATE(HDMI_ACR_48_1, HDMI_ACR_N_48, audio_clock_info.n_48khz);
+
+	/* Video driver cannot know in advance which sample rate will
+	 * be used by HD Audio driver
+	 * HDMI_ACR_PACKET_CONTROL__HDMI_ACR_N_MULTIPLE field is
+	 * programmed below in interruppt callback
+	 */
+}
+
+void enc3_se_hdmi_audio_setup(
+	struct stream_encoder *enc,
+	unsigned int az_inst,
+	struct audio_info *info,
+	struct audio_crtc_info *audio_crtc_info)
+{
+	enc1_se_enable_audio_clock(enc, true);
+	enc3_se_setup_hdmi_audio(enc, audio_crtc_info);
+	ASSERT (enc->afmt);
+	enc->afmt->funcs->se_audio_setup(enc->afmt, az_inst, info);
+}
+
+
+static const struct stream_encoder_funcs dcn30_str_enc_funcs = {
+	.dp_set_odm_combine =
+		enc3_dp_set_odm_combine,
+	.dp_set_stream_attribute =
+		enc2_stream_encoder_dp_set_stream_attribute,
+	.hdmi_set_stream_attribute =
+		enc3_stream_encoder_hdmi_set_stream_attribute,
+	.dvi_set_stream_attribute =
+		enc3_stream_encoder_dvi_set_stream_attribute,
+	.set_throttled_vcp_size =
+		enc1_stream_encoder_set_throttled_vcp_size,
+	.update_hdmi_info_packets =
+		enc3_stream_encoder_update_hdmi_info_packets,
+	.stop_hdmi_info_packets =
+		enc3_stream_encoder_stop_hdmi_info_packets,
+	.update_dp_info_packets_sdp_line_num =
+		enc3_stream_encoder_update_dp_info_packets_sdp_line_num,
+	.update_dp_info_packets =
+		enc3_stream_encoder_update_dp_info_packets,
+	.stop_dp_info_packets =
+		enc1_stream_encoder_stop_dp_info_packets,
+	.dp_blank =
+		enc1_stream_encoder_dp_blank,
+	.dp_unblank =
+		enc2_stream_encoder_dp_unblank,
+	.audio_mute_control = enc3_audio_mute_control,
+
+	.dp_audio_setup = enc3_se_dp_audio_setup,
+	.dp_audio_enable = enc3_se_dp_audio_enable,
+	.dp_audio_disable = enc1_se_dp_audio_disable,
+
+	.hdmi_audio_setup = enc3_se_hdmi_audio_setup,
+	.hdmi_audio_disable = enc1_se_hdmi_audio_disable,
+	.setup_stereo_sync  = enc1_setup_stereo_sync,
+	.set_avmute = enc1_stream_encoder_set_avmute,
+	.dig_connect_to_otg = enc1_dig_connect_to_otg,
+	.dig_source_otg = enc1_dig_source_otg,
+
+	.dp_get_pixel_format  = enc1_stream_encoder_dp_get_pixel_format,
+
+	.enc_read_state = enc3_read_state,
+	.dp_set_dsc_config = enc3_dp_set_dsc_config,
+	.dp_set_dsc_pps_info_packet = enc3_dp_set_dsc_pps_info_packet,
+	.set_dynamic_metadata = enc2_set_dynamic_metadata,
+	.hdmi_reset_stream_attribute = enc1_reset_hdmi_stream_attribute,
+
+	.get_fifo_cal_average_level = enc2_get_fifo_cal_average_level,
+};
+
+void dcn30_dio_stream_encoder_construct(
+	struct dcn10_stream_encoder *enc1,
+	struct dc_context *ctx,
+	struct dc_bios *bp,
+	enum engine_id eng_id,
+	struct vpg *vpg,
+	struct afmt *afmt,
+	const struct dcn10_stream_enc_registers *regs,
+	const struct dcn10_stream_encoder_shift *se_shift,
+	const struct dcn10_stream_encoder_mask *se_mask)
+{
+	enc1->base.funcs = &dcn30_str_enc_funcs;
+	enc1->base.ctx = ctx;
+	enc1->base.id = eng_id;
+	enc1->base.bp = bp;
+	enc1->base.vpg = vpg;
+	enc1->base.afmt = afmt;
+	enc1->regs = regs;
+	enc1->se_shift = se_shift;
+	enc1->se_mask = se_mask;
+	enc1->base.stream_enc_inst = vpg->inst;
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn30/dcn30_dio_stream_encoder.h b/drivers/gpu/drm/amd/display/dc/dio/dcn30/dcn30_dio_stream_encoder.h
new file mode 100644
index 000000000000..830ce7e47035
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn30/dcn30_dio_stream_encoder.h
@@ -0,0 +1,331 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_DIO_STREAM_ENCODER_DCN30_H__
+#define __DC_DIO_STREAM_ENCODER_DCN30_H__
+
+#include "dcn30/dcn30_vpg.h"
+#include "dcn30/dcn30_afmt.h"
+#include "stream_encoder.h"
+#include "dcn20/dcn20_stream_encoder.h"
+
+/* Register bit field name change */
+#define RDPCSTX0_RDPCSTX_CLOCK_CNTL__RDPCS_SYMCLK_DIV2_GATE_DIS__SHIFT                                        0x8
+#define RDPCSTX0_RDPCSTX_CLOCK_CNTL__RDPCS_SYMCLK_DIV2_EN__SHIFT                                              0x9
+#define RDPCSTX0_RDPCSTX_CLOCK_CNTL__RDPCS_SYMCLK_DIV2_CLOCK_ON__SHIFT                                        0xa
+#define DPCSTX0_DPCSTX_TX_CNTL__DPCS_TX_DATA_SWAP__SHIFT                                                      0xe
+#define DPCSTX0_DPCSTX_TX_CNTL__DPCS_TX_DATA_ORDER_INVERT__SHIFT                                              0xf
+
+#define RDPCSTX0_RDPCSTX_CLOCK_CNTL__RDPCS_SYMCLK_DIV2_GATE_DIS_MASK                                          0x00000100L
+#define RDPCSTX0_RDPCSTX_CLOCK_CNTL__RDPCS_SYMCLK_DIV2_EN_MASK                                                0x00000200L
+#define RDPCSTX0_RDPCSTX_CLOCK_CNTL__RDPCS_SYMCLK_DIV2_CLOCK_ON_MASK                                          0x00000400L
+#define DPCSTX0_DPCSTX_TX_CNTL__DPCS_TX_DATA_SWAP_MASK                                                        0x00004000L
+#define DPCSTX0_DPCSTX_TX_CNTL__DPCS_TX_DATA_ORDER_INVERT_MASK                                                0x00008000L
+
+
+#define SE_DCN3_REG_LIST(id)\
+	SRI(AFMT_CNTL, DIG, id), \
+	SRI(DIG_FE_CNTL, DIG, id), \
+	SRI(HDMI_CONTROL, DIG, id), \
+	SRI(HDMI_DB_CONTROL, DIG, id), \
+	SRI(HDMI_GC, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL0, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL1, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL2, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL3, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL4, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL5, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL6, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL7, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL8, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL9, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL10, DIG, id), \
+	SRI(HDMI_INFOFRAME_CONTROL0, DIG, id), \
+	SRI(HDMI_INFOFRAME_CONTROL1, DIG, id), \
+	SRI(HDMI_VBI_PACKET_CONTROL, DIG, id), \
+	SRI(HDMI_AUDIO_PACKET_CONTROL, DIG, id),\
+	SRI(HDMI_ACR_PACKET_CONTROL, DIG, id),\
+	SRI(HDMI_ACR_32_0, DIG, id),\
+	SRI(HDMI_ACR_32_1, DIG, id),\
+	SRI(HDMI_ACR_44_0, DIG, id),\
+	SRI(HDMI_ACR_44_1, DIG, id),\
+	SRI(HDMI_ACR_48_0, DIG, id),\
+	SRI(HDMI_ACR_48_1, DIG, id),\
+	SRI(DP_DB_CNTL, DP, id), \
+	SRI(DP_MSA_MISC, DP, id), \
+	SRI(DP_MSA_VBID_MISC, DP, id), \
+	SRI(DP_MSA_COLORIMETRY, DP, id), \
+	SRI(DP_MSA_TIMING_PARAM1, DP, id), \
+	SRI(DP_MSA_TIMING_PARAM2, DP, id), \
+	SRI(DP_MSA_TIMING_PARAM3, DP, id), \
+	SRI(DP_MSA_TIMING_PARAM4, DP, id), \
+	SRI(DP_MSE_RATE_CNTL, DP, id), \
+	SRI(DP_MSE_RATE_UPDATE, DP, id), \
+	SRI(DP_PIXEL_FORMAT, DP, id), \
+	SRI(DP_SEC_CNTL, DP, id), \
+	SRI(DP_SEC_CNTL1, DP, id), \
+	SRI(DP_SEC_CNTL2, DP, id), \
+	SRI(DP_SEC_CNTL5, DP, id), \
+	SRI(DP_SEC_CNTL6, DP, id), \
+	SRI(DP_STEER_FIFO, DP, id), \
+	SRI(DP_VID_M, DP, id), \
+	SRI(DP_VID_N, DP, id), \
+	SRI(DP_VID_STREAM_CNTL, DP, id), \
+	SRI(DP_VID_TIMING, DP, id), \
+	SRI(DP_SEC_AUD_N, DP, id), \
+	SRI(DP_SEC_AUD_N_READBACK, DP, id), \
+	SRI(DP_SEC_AUD_M_READBACK, DP, id), \
+	SRI(DP_SEC_TIMESTAMP, DP, id), \
+	SRI(DP_DSC_CNTL, DP, id), \
+	SRI(DP_DSC_BYTES_PER_PIXEL, DP, id), \
+	SRI(DP_SEC_METADATA_TRANSMISSION, DP, id), \
+	SRI(HDMI_METADATA_PACKET_CONTROL, DIG, id), \
+	SRI(DP_SEC_FRAMING4, DP, id), \
+	SRI(DP_GSP11_CNTL, DP, id), \
+	SRI(DME_CONTROL, DME, id),\
+	SRI(DP_SEC_METADATA_TRANSMISSION, DP, id), \
+	SRI(HDMI_METADATA_PACKET_CONTROL, DIG, id), \
+	SRI(DIG_FE_CNTL, DIG, id), \
+	SRI(DIG_FIFO_STATUS, DIG, id), \
+	SRI(DIG_CLOCK_PATTERN, DIG, id)
+
+
+#define SE_COMMON_MASK_SH_LIST_DCN30(mask_sh)\
+	SE_SF(DP0_DP_PIXEL_FORMAT, DP_PIXEL_ENCODING, mask_sh),\
+	SE_SF(DP0_DP_PIXEL_FORMAT, DP_COMPONENT_DEPTH, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_PACKET_GEN_VERSION, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_KEEPOUT_MODE, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_DATA_SCRAMBLE_EN, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_NO_EXTRA_NULL_PACKET_FILLED, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_GC_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_GC_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_NULL_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_ACP_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_INFOFRAME_CONTROL1, HDMI_AUDIO_INFO_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GC, HDMI_GC_AVMUTE, mask_sh),\
+	SE_SF(DP0_DP_MSE_RATE_CNTL, DP_MSE_RATE_X, mask_sh),\
+	SE_SF(DP0_DP_MSE_RATE_CNTL, DP_MSE_RATE_Y, mask_sh),\
+	SE_SF(DP0_DP_MSE_RATE_UPDATE, DP_MSE_RATE_UPDATE_PENDING, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP0_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP1_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP2_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP3_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_MPG_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP4_SEND, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP4_SEND_PENDING, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL4, DP_SEC_GSP4_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP4_SEND_ANY_LINE, mask_sh),\
+	SE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_DIS_DEFER, mask_sh),\
+	SE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_STATUS, mask_sh),\
+	SE_SF(DP0_DP_STEER_FIFO, DP_STEER_FIFO_RESET, mask_sh),\
+	SE_SF(DP0_DP_VID_TIMING, DP_VID_M_N_GEN_EN, mask_sh),\
+	SE_SF(DP0_DP_VID_N, DP_VID_N, mask_sh),\
+	SE_SF(DP0_DP_VID_M, DP_VID_M, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_START, mask_sh),\
+	SE_SF(DIG0_HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_DELAY_EN, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUTO_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_PACKET_CONTROL, HDMI_ACR_SOURCE, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUDIO_PRIORITY, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_32_0, HDMI_ACR_CTS_32, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_32_1, HDMI_ACR_N_32, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_44_0, HDMI_ACR_CTS_44, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_44_1, HDMI_ACR_N_44, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_48_0, HDMI_ACR_CTS_48, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_48_1, HDMI_ACR_N_48, mask_sh),\
+	SE_SF(DP0_DP_SEC_AUD_N, DP_SEC_AUD_N, mask_sh),\
+	SE_SF(DP0_DP_SEC_AUD_N_READBACK, DP_SEC_AUD_N_READBACK, mask_sh),\
+	SE_SF(DP0_DP_SEC_AUD_M_READBACK, DP_SEC_AUD_M_READBACK, mask_sh),\
+	SE_SF(DP0_DP_SEC_TIMESTAMP, DP_SEC_TIMESTAMP_MODE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_ASP_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_ATP_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_AIP_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_ACM_ENABLE, mask_sh),\
+	SE_SF(DIG0_AFMT_CNTL, AFMT_AUDIO_CLOCK_EN, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_CLOCK_CHANNEL_RATE, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, TMDS_PIXEL_ENCODING, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, TMDS_COLOR_FORMAT, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_STEREOSYNC_SELECT, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_STEREOSYNC_GATE_EN, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_STATUS, DIG_FIFO_LEVEL_ERROR, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_STATUS, DIG_FIFO_USE_OVERWRITE_LEVEL, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_STATUS, DIG_FIFO_OVERWRITE_LEVEL, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_STATUS, DIG_FIFO_ERROR_ACK, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_STATUS, DIG_FIFO_CAL_AVERAGE_LEVEL, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_STATUS, DIG_FIFO_MAXIMUM_LEVEL, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_STATUS, DIG_FIFO_MINIMUM_LEVEL, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_STATUS, DIG_FIFO_READ_CLOCK_SRC, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_STATUS, DIG_FIFO_CALIBRATED, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_STATUS, DIG_FIFO_FORCE_RECAL_AVERAGE, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_STATUS, DIG_FIFO_FORCE_RECOMP_MINMAX, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP4_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP5_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP6_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP7_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL1, DP_SEC_GSP5_LINE_REFERENCE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP7_SEND, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL5, DP_SEC_GSP5_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL6, DP_SEC_GSP7_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP11_PPS, mask_sh),\
+	SE_SF(DP0_DP_GSP11_CNTL, DP_SEC_GSP11_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_GSP11_CNTL, DP_SEC_GSP11_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_DB_CNTL, DP_DB_DISABLE, mask_sh),\
+	SE_SF(DP0_DP_MSA_COLORIMETRY, DP_MSA_MISC0, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM1, DP_MSA_HTOTAL, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM1, DP_MSA_VTOTAL, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM2, DP_MSA_HSTART, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM2, DP_MSA_VSTART, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_HSYNCWIDTH, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_HSYNCPOLARITY, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_VSYNCWIDTH, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_VSYNCPOLARITY, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM4, DP_MSA_HWIDTH, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM4, DP_MSA_VHEIGHT, mask_sh),\
+	SE_SF(DIG0_HDMI_DB_CONTROL, HDMI_DB_DISABLE, mask_sh),\
+	SE_SF(DP0_DP_VID_TIMING, DP_VID_N_MUL, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_SOURCE_SELECT, mask_sh), \
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC0_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC0_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC1_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC1_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC2_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC2_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC3_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC3_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC4_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC4_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC5_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC5_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC6_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC6_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC7_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC7_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC8_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC8_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC9_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC9_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC10_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC10_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC11_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC11_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC12_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC12_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC13_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC13_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC14_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC14_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL1, HDMI_GENERIC0_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL1, HDMI_GENERIC1_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL2, HDMI_GENERIC2_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL2, HDMI_GENERIC3_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL3, HDMI_GENERIC4_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL3, HDMI_GENERIC5_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL4, HDMI_GENERIC6_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL4, HDMI_GENERIC7_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL7, HDMI_GENERIC8_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL7, HDMI_GENERIC9_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL8, HDMI_GENERIC10_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL8, HDMI_GENERIC11_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL9, HDMI_GENERIC12_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL9, HDMI_GENERIC13_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL10, HDMI_GENERIC14_LINE, mask_sh),\
+	SE_SF(DP0_DP_DSC_CNTL, DP_DSC_MODE, mask_sh),\
+	SE_SF(DP0_DP_DSC_CNTL, DP_DSC_SLICE_WIDTH, mask_sh),\
+	SE_SF(DP0_DP_DSC_BYTES_PER_PIXEL, DP_DSC_BYTES_PER_PIXEL, mask_sh),\
+	SE_SF(DP0_DP_MSA_VBID_MISC, DP_VBID6_LINE_REFERENCE, mask_sh),\
+	SE_SF(DP0_DP_MSA_VBID_MISC, DP_VBID6_LINE_NUM, mask_sh),\
+	SE_SF(DME0_DME_CONTROL, METADATA_ENGINE_EN, mask_sh),\
+	SE_SF(DME0_DME_CONTROL, METADATA_HUBP_REQUESTOR_ID, mask_sh),\
+	SE_SF(DME0_DME_CONTROL, METADATA_STREAM_TYPE, mask_sh),\
+	SE_SF(DP0_DP_SEC_METADATA_TRANSMISSION, DP_SEC_METADATA_PACKET_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_METADATA_TRANSMISSION, DP_SEC_METADATA_PACKET_LINE_REFERENCE, mask_sh),\
+	SE_SF(DP0_DP_SEC_METADATA_TRANSMISSION, DP_SEC_METADATA_PACKET_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_METADATA_PACKET_CONTROL, HDMI_METADATA_PACKET_ENABLE, mask_sh),\
+	SE_SF(DIG0_HDMI_METADATA_PACKET_CONTROL, HDMI_METADATA_PACKET_LINE_REFERENCE, mask_sh),\
+	SE_SF(DIG0_HDMI_METADATA_PACKET_CONTROL, HDMI_METADATA_PACKET_LINE, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DOLBY_VISION_EN, mask_sh),\
+	SE_SF(DP0_DP_PIXEL_FORMAT, DP_PIXEL_COMBINE, mask_sh),\
+	SE_SF(DP0_DP_SEC_FRAMING4, DP_SST_SDP_SPLITTING, mask_sh),\
+	SE_SF(DIG0_DIG_CLOCK_PATTERN, DIG_CLOCK_PATTERN, mask_sh)
+
+void dcn30_dio_stream_encoder_construct(
+	struct dcn10_stream_encoder *enc1,
+	struct dc_context *ctx,
+	struct dc_bios *bp,
+	enum engine_id eng_id,
+	struct vpg *vpg,
+	struct afmt *afmt,
+	const struct dcn10_stream_enc_registers *regs,
+	const struct dcn10_stream_encoder_shift *se_shift,
+	const struct dcn10_stream_encoder_mask *se_mask);
+
+void enc3_stream_encoder_update_hdmi_info_packets(
+	struct stream_encoder *enc,
+	const struct encoder_info_frame *info_frame);
+
+void enc3_stream_encoder_stop_hdmi_info_packets(
+	struct stream_encoder *enc);
+
+void enc3_stream_encoder_update_dp_info_packets_sdp_line_num(
+		struct stream_encoder *enc,
+		struct encoder_info_frame *info_frame);
+
+void enc3_stream_encoder_update_dp_info_packets(
+	struct stream_encoder *enc,
+	const struct encoder_info_frame *info_frame);
+
+void enc3_audio_mute_control(
+	struct stream_encoder *enc,
+	bool mute);
+
+void enc3_se_dp_audio_setup(
+	struct stream_encoder *enc,
+	unsigned int az_inst,
+	struct audio_info *info);
+
+void enc3_se_dp_audio_enable(
+	struct stream_encoder *enc);
+
+void enc3_se_hdmi_audio_setup(
+	struct stream_encoder *enc,
+	unsigned int az_inst,
+	struct audio_info *info,
+	struct audio_crtc_info *audio_crtc_info);
+
+void enc3_dp_set_dsc_pps_info_packet(
+	struct stream_encoder *enc,
+	bool enable,
+	uint8_t *dsc_packed_pps,
+	bool immediate_update);
+void enc3_update_hdmi_info_packet(
+	struct dcn10_stream_encoder *enc1,
+	uint32_t packet_index,
+	const struct dc_info_packet *info_packet);
+
+#endif /* __DC_DIO_STREAM_ENCODER_DCN30_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn301/dcn301_dio_link_encoder.c b/drivers/gpu/drm/amd/display/dc/dio/dcn301/dcn301_dio_link_encoder.c
new file mode 100644
index 000000000000..1b39a6e8a1ac
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn301/dcn301_dio_link_encoder.c
@@ -0,0 +1,191 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+
+#include "core_types.h"
+#include "link_encoder.h"
+#include "dcn301_dio_link_encoder.h"
+#include "stream_encoder.h"
+#include "dc_bios_types.h"
+#include "gpio_service_interface.h"
+
+#define CTX \
+	enc10->base.ctx
+#define DC_LOGGER \
+	enc10->base.ctx->logger
+
+#define REG(reg)\
+	(enc10->link_regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	enc10->link_shift->field_name, enc10->link_mask->field_name
+
+#define IND_REG(index) \
+	(enc10->link_regs->index)
+
+static const struct link_encoder_funcs dcn301_link_enc_funcs = {
+	.read_state = link_enc2_read_state,
+	.validate_output_with_stream = dcn10_link_encoder_validate_output_with_stream,
+	.hw_init = enc3_hw_init,
+	.setup = dcn10_link_encoder_setup,
+	.enable_tmds_output = dcn10_link_encoder_enable_tmds_output,
+	.enable_dp_output = dcn20_link_encoder_enable_dp_output,
+	.enable_dp_mst_output = dcn10_link_encoder_enable_dp_mst_output,
+	.disable_output = dcn10_link_encoder_disable_output,
+	.dp_set_lane_settings = dcn10_link_encoder_dp_set_lane_settings,
+	.dp_set_phy_pattern = dcn10_link_encoder_dp_set_phy_pattern,
+	.update_mst_stream_allocation_table = dcn10_link_encoder_update_mst_stream_allocation_table,
+	.psr_program_dp_dphy_fast_training = dcn10_psr_program_dp_dphy_fast_training,
+	.psr_program_secondary_packet = dcn10_psr_program_secondary_packet,
+	.connect_dig_be_to_fe = dcn10_link_encoder_connect_dig_be_to_fe,
+	.enable_hpd = dcn10_link_encoder_enable_hpd,
+	.disable_hpd = dcn10_link_encoder_disable_hpd,
+	.is_dig_enabled = dcn10_is_dig_enabled,
+	.destroy = dcn10_link_encoder_destroy,
+	.fec_set_enable = enc2_fec_set_enable,
+	.fec_set_ready = enc2_fec_set_ready,
+	.fec_is_active = enc2_fec_is_active,
+	.get_dig_frontend = dcn10_get_dig_frontend,
+	.get_dig_mode = dcn10_get_dig_mode,
+	.is_in_alt_mode = dcn20_link_encoder_is_in_alt_mode,
+	.get_max_link_cap = dcn20_link_encoder_get_max_link_cap,
+};
+
+void dcn301_link_encoder_construct(
+	struct dcn20_link_encoder *enc20,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	const struct dcn10_link_enc_aux_registers *aux_regs,
+	const struct dcn10_link_enc_hpd_registers *hpd_regs,
+	const struct dcn10_link_enc_shift *link_shift,
+	const struct dcn10_link_enc_mask *link_mask)
+{
+	struct bp_encoder_cap_info bp_cap_info = {0};
+	const struct dc_vbios_funcs *bp_funcs = init_data->ctx->dc_bios->funcs;
+	enum bp_result result = BP_RESULT_OK;
+	struct dcn10_link_encoder *enc10 = &enc20->enc10;
+
+	enc10->base.funcs = &dcn301_link_enc_funcs;
+	enc10->base.ctx = init_data->ctx;
+	enc10->base.id = init_data->encoder;
+
+	enc10->base.hpd_source = init_data->hpd_source;
+	enc10->base.connector = init_data->connector;
+
+	enc10->base.preferred_engine = ENGINE_ID_UNKNOWN;
+
+	enc10->base.features = *enc_features;
+
+	enc10->base.transmitter = init_data->transmitter;
+
+	/* set the flag to indicate whether driver poll the I2C data pin
+	 * while doing the DP sink detect
+	 */
+
+/*	if (dal_adapter_service_is_feature_supported(as,
+		FEATURE_DP_SINK_DETECT_POLL_DATA_PIN))
+		enc10->base.features.flags.bits.
+			DP_SINK_DETECT_POLL_DATA_PIN = true;*/
+
+	enc10->base.output_signals =
+		SIGNAL_TYPE_DVI_SINGLE_LINK |
+		SIGNAL_TYPE_DVI_DUAL_LINK |
+		SIGNAL_TYPE_LVDS |
+		SIGNAL_TYPE_DISPLAY_PORT |
+		SIGNAL_TYPE_DISPLAY_PORT_MST |
+		SIGNAL_TYPE_EDP |
+		SIGNAL_TYPE_HDMI_TYPE_A;
+
+	/* For DCE 8.0 and 8.1, by design, UNIPHY is hardwired to DIG_BE.
+	 * SW always assign DIG_FE 1:1 mapped to DIG_FE for non-MST UNIPHY.
+	 * SW assign DIG_FE to non-MST UNIPHY first and MST last. So prefer
+	 * DIG is per UNIPHY and used by SST DP, eDP, HDMI, DVI and LVDS.
+	 * Prefer DIG assignment is decided by board design.
+	 * For DCE 8.0, there are only max 6 UNIPHYs, we assume board design
+	 * and VBIOS will filter out 7 UNIPHY for DCE 8.0.
+	 * By this, adding DIGG should not hurt DCE 8.0.
+	 * This will let DCE 8.1 share DCE 8.0 as much as possible
+	 */
+
+	enc10->link_regs = link_regs;
+	enc10->aux_regs = aux_regs;
+	enc10->hpd_regs = hpd_regs;
+	enc10->link_shift = link_shift;
+	enc10->link_mask = link_mask;
+
+	switch (enc10->base.transmitter) {
+	case TRANSMITTER_UNIPHY_A:
+		enc10->base.preferred_engine = ENGINE_ID_DIGA;
+	break;
+	case TRANSMITTER_UNIPHY_B:
+		enc10->base.preferred_engine = ENGINE_ID_DIGB;
+	break;
+	case TRANSMITTER_UNIPHY_C:
+		enc10->base.preferred_engine = ENGINE_ID_DIGC;
+	break;
+	case TRANSMITTER_UNIPHY_D:
+		enc10->base.preferred_engine = ENGINE_ID_DIGD;
+	break;
+	case TRANSMITTER_UNIPHY_E:
+		enc10->base.preferred_engine = ENGINE_ID_DIGE;
+	break;
+	case TRANSMITTER_UNIPHY_F:
+		enc10->base.preferred_engine = ENGINE_ID_DIGF;
+	break;
+	case TRANSMITTER_UNIPHY_G:
+		enc10->base.preferred_engine = ENGINE_ID_DIGG;
+	break;
+	default:
+		ASSERT_CRITICAL(false);
+		enc10->base.preferred_engine = ENGINE_ID_UNKNOWN;
+	}
+
+	/* default to one to mirror Windows behavior */
+	enc10->base.features.flags.bits.HDMI_6GB_EN = 1;
+
+	result = bp_funcs->get_encoder_cap_info(enc10->base.ctx->dc_bios,
+						enc10->base.id, &bp_cap_info);
+
+	/* Override features with DCE-specific values */
+	if (result == BP_RESULT_OK) {
+		enc10->base.features.flags.bits.IS_HBR2_CAPABLE =
+				bp_cap_info.DP_HBR2_EN;
+		enc10->base.features.flags.bits.IS_HBR3_CAPABLE =
+				bp_cap_info.DP_HBR3_EN;
+		enc10->base.features.flags.bits.HDMI_6GB_EN = bp_cap_info.HDMI_6GB_EN;
+		enc10->base.features.flags.bits.DP_IS_USB_C =
+				bp_cap_info.DP_IS_USB_C;
+	} else {
+		DC_LOG_WARNING("%s: Failed to get encoder_cap_info from VBIOS with error code %d!\n",
+				__func__,
+				result);
+	}
+	if (enc10->base.ctx->dc->debug.hdmi20_disable) {
+		enc10->base.features.flags.bits.HDMI_6GB_EN = 0;
+	}
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn301/dcn301_dio_link_encoder.h b/drivers/gpu/drm/amd/display/dc/dio/dcn301/dcn301_dio_link_encoder.h
new file mode 100644
index 000000000000..49f8d91d4951
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn301/dcn301_dio_link_encoder.h
@@ -0,0 +1,82 @@
+/*
+ * Copyright 2020 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_LINK_ENCODER__DCN301_H__
+#define __DC_LINK_ENCODER__DCN301_H__
+
+#include "dcn20/dcn20_link_encoder.h"
+
+
+#define LE_DCN301_REG_LIST(id)\
+	SRI(DIG_BE_CNTL, DIG, id), \
+	SRI(DIG_BE_EN_CNTL, DIG, id), \
+	SRI(TMDS_CTL_BITS, DIG, id), \
+	SRI(TMDS_DCBALANCER_CONTROL, DIG, id), \
+	SRI(DP_CONFIG, DP, id), \
+	SRI(DP_DPHY_CNTL, DP, id), \
+	SRI(DP_DPHY_PRBS_CNTL, DP, id), \
+	SRI(DP_DPHY_SCRAM_CNTL, DP, id),\
+	SRI(DP_DPHY_SYM0, DP, id), \
+	SRI(DP_DPHY_SYM1, DP, id), \
+	SRI(DP_DPHY_SYM2, DP, id), \
+	SRI(DP_DPHY_TRAINING_PATTERN_SEL, DP, id), \
+	SRI(DP_LINK_CNTL, DP, id), \
+	SRI(DP_LINK_FRAMING_CNTL, DP, id), \
+	SRI(DP_MSE_SAT0, DP, id), \
+	SRI(DP_MSE_SAT1, DP, id), \
+	SRI(DP_MSE_SAT2, DP, id), \
+	SRI(DP_MSE_SAT_UPDATE, DP, id), \
+	SRI(DP_SEC_CNTL, DP, id), \
+	SRI(DP_VID_STREAM_CNTL, DP, id), \
+	SRI(DP_DPHY_FAST_TRAINING, DP, id), \
+	SRI(DP_SEC_CNTL1, DP, id), \
+	SRI(DP_DPHY_BS_SR_SWAP_CNTL, DP, id), \
+	SRI(DP_DPHY_HBR2_PATTERN_CONTROL, DP, id)
+
+#define LINK_ENCODER_MASK_SH_LIST_DCN301(mask_sh) \
+	LINK_ENCODER_MASK_SH_LIST_DCN20(mask_sh),\
+	LE_SF(DIG0_TMDS_DCBALANCER_CONTROL, TMDS_SYNC_DCBAL_EN, mask_sh)
+
+#define DPCS_DCN301_MASK_SH_LIST(mask_sh)\
+	DPCS_DCN2_MASK_SH_LIST(mask_sh),\
+	LE_SF(DPCSTX0_DPCSTX_TX_CNTL, DPCS_TX_HDMI_FRL_MODE, mask_sh),\
+	LE_SF(DPCSTX0_DPCSTX_TX_CNTL, DPCS_TX_DATA_SWAP_10_BIT, mask_sh),\
+	LE_SF(DPCSTX0_DPCSTX_TX_CNTL, DPCS_TX_DATA_ORDER_INVERT_18_BIT, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL0, RDPCS_PHY_TX_VBOOST_LVL, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CLOCK_CNTL, RDPCS_TX_CLK_EN, mask_sh)
+
+void dcn301_link_encoder_construct(
+	struct dcn20_link_encoder *enc20,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	const struct dcn10_link_enc_aux_registers *aux_regs,
+	const struct dcn10_link_enc_hpd_registers *hpd_regs,
+	const struct dcn10_link_enc_shift *link_shift,
+	const struct dcn10_link_enc_mask *link_mask);
+
+void enc3_hw_init(struct link_encoder *enc);
+
+#endif /* __DC_LINK_ENCODER__DCN301_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn31/dcn31_dio_link_encoder.c b/drivers/gpu/drm/amd/display/dc/dio/dcn31/dcn31_dio_link_encoder.c
new file mode 100644
index 000000000000..84cc2ddc52fe
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn31/dcn31_dio_link_encoder.c
@@ -0,0 +1,682 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+#include "reg_helper.h"
+
+#include "core_types.h"
+#include "link_encoder.h"
+#include "dcn31_dio_link_encoder.h"
+#include "stream_encoder.h"
+#include "dc_bios_types.h"
+
+#include "gpio_service_interface.h"
+
+#include "link_enc_cfg.h"
+#include "dc_dmub_srv.h"
+#include "dal_asic_id.h"
+#include "link_service.h"
+
+#define CTX \
+	enc10->base.ctx
+#define DC_LOGGER \
+	enc10->base.ctx->logger
+
+#define REG(reg)\
+	(enc10->link_regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	enc10->link_shift->field_name, enc10->link_mask->field_name
+
+#define IND_REG(index) \
+	(enc10->link_regs->index)
+
+#define AUX_REG(reg)\
+	(enc10->aux_regs->reg)
+
+#define AUX_REG_READ(reg_name) \
+		dm_read_reg(CTX, AUX_REG(reg_name))
+
+#define AUX_REG_WRITE(reg_name, val) \
+			dm_write_reg(CTX, AUX_REG(reg_name), val)
+
+#ifndef MIN
+#define MIN(X, Y) ((X) < (Y) ? (X) : (Y))
+#endif
+
+static uint8_t phy_id_from_transmitter(enum transmitter t)
+{
+	uint8_t phy_id;
+
+	switch (t) {
+	case TRANSMITTER_UNIPHY_A:
+		phy_id = 0;
+		break;
+	case TRANSMITTER_UNIPHY_B:
+		phy_id = 1;
+		break;
+	case TRANSMITTER_UNIPHY_C:
+		phy_id = 2;
+		break;
+	case TRANSMITTER_UNIPHY_D:
+		phy_id = 3;
+		break;
+	case TRANSMITTER_UNIPHY_E:
+		phy_id = 4;
+		break;
+	case TRANSMITTER_UNIPHY_F:
+		phy_id = 5;
+		break;
+	case TRANSMITTER_UNIPHY_G:
+		phy_id = 6;
+		break;
+	default:
+		phy_id = 0;
+		break;
+	}
+	return phy_id;
+}
+
+static bool has_query_dp_alt(struct link_encoder *enc)
+{
+	struct dc_dmub_srv *dc_dmub_srv = enc->ctx->dmub_srv;
+
+	if (enc->ctx->dce_version >= DCN_VERSION_3_15)
+		return true;
+
+	/* Supports development firmware and firmware >= 4.0.11 */
+	return dc_dmub_srv &&
+	       !(dc_dmub_srv->dmub->fw_version >= DMUB_FW_VERSION(4, 0, 0) &&
+		 dc_dmub_srv->dmub->fw_version <= DMUB_FW_VERSION(4, 0, 10));
+}
+
+static bool query_dp_alt_from_dmub(struct link_encoder *enc,
+				   union dmub_rb_cmd *cmd)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	memset(cmd, 0, sizeof(*cmd));
+	cmd->query_dp_alt.header.type = DMUB_CMD__VBIOS;
+	cmd->query_dp_alt.header.sub_type =
+		DMUB_CMD__VBIOS_TRANSMITTER_QUERY_DP_ALT;
+	cmd->query_dp_alt.header.payload_bytes = sizeof(cmd->query_dp_alt.data);
+	cmd->query_dp_alt.data.phy_id = phy_id_from_transmitter(enc10->base.transmitter);
+
+	if (!dc_wake_and_execute_dmub_cmd(enc->ctx, cmd, DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY))
+		return false;
+
+	return true;
+}
+
+void dcn31_link_encoder_set_dio_phy_mux(
+	struct link_encoder *enc,
+	enum encoder_type_select sel,
+	uint32_t hpo_inst)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	switch (enc->transmitter) {
+	case TRANSMITTER_UNIPHY_A:
+		if (sel == ENCODER_TYPE_HDMI_FRL)
+			REG_UPDATE(DIO_LINKA_CNTL,
+					HPO_HDMI_ENC_SEL, hpo_inst);
+		else if (sel == ENCODER_TYPE_DP_128B132B)
+			REG_UPDATE(DIO_LINKA_CNTL,
+					HPO_DP_ENC_SEL, hpo_inst);
+		REG_UPDATE(DIO_LINKA_CNTL,
+				ENC_TYPE_SEL, sel);
+		break;
+	case TRANSMITTER_UNIPHY_B:
+		if (sel == ENCODER_TYPE_HDMI_FRL)
+			REG_UPDATE(DIO_LINKB_CNTL,
+					HPO_HDMI_ENC_SEL, hpo_inst);
+		else if (sel == ENCODER_TYPE_DP_128B132B)
+			REG_UPDATE(DIO_LINKB_CNTL,
+					HPO_DP_ENC_SEL, hpo_inst);
+		REG_UPDATE(DIO_LINKB_CNTL,
+				ENC_TYPE_SEL, sel);
+		break;
+	case TRANSMITTER_UNIPHY_C:
+		if (sel == ENCODER_TYPE_HDMI_FRL)
+			REG_UPDATE(DIO_LINKC_CNTL,
+					HPO_HDMI_ENC_SEL, hpo_inst);
+		else if (sel == ENCODER_TYPE_DP_128B132B)
+			REG_UPDATE(DIO_LINKC_CNTL,
+					HPO_DP_ENC_SEL, hpo_inst);
+		REG_UPDATE(DIO_LINKC_CNTL,
+				ENC_TYPE_SEL, sel);
+		break;
+	case TRANSMITTER_UNIPHY_D:
+		if (sel == ENCODER_TYPE_HDMI_FRL)
+			REG_UPDATE(DIO_LINKD_CNTL,
+					HPO_HDMI_ENC_SEL, hpo_inst);
+		else if (sel == ENCODER_TYPE_DP_128B132B)
+			REG_UPDATE(DIO_LINKD_CNTL,
+					HPO_DP_ENC_SEL, hpo_inst);
+		REG_UPDATE(DIO_LINKD_CNTL,
+				ENC_TYPE_SEL, sel);
+		break;
+	case TRANSMITTER_UNIPHY_E:
+		if (sel == ENCODER_TYPE_HDMI_FRL)
+			REG_UPDATE(DIO_LINKE_CNTL,
+					HPO_HDMI_ENC_SEL, hpo_inst);
+		else if (sel == ENCODER_TYPE_DP_128B132B)
+			REG_UPDATE(DIO_LINKE_CNTL,
+					HPO_DP_ENC_SEL, hpo_inst);
+		REG_UPDATE(DIO_LINKE_CNTL,
+				ENC_TYPE_SEL, sel);
+		break;
+	case TRANSMITTER_UNIPHY_F:
+		if (sel == ENCODER_TYPE_HDMI_FRL)
+			REG_UPDATE(DIO_LINKF_CNTL,
+					HPO_HDMI_ENC_SEL, hpo_inst);
+		else if (sel == ENCODER_TYPE_DP_128B132B)
+			REG_UPDATE(DIO_LINKF_CNTL,
+					HPO_DP_ENC_SEL, hpo_inst);
+		REG_UPDATE(DIO_LINKF_CNTL,
+				ENC_TYPE_SEL, sel);
+		break;
+	default:
+		/* Do nothing */
+		break;
+	}
+}
+
+void enc31_hw_init(struct link_encoder *enc)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+/*
+	00 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__1to2 : 1/2
+	01 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__3to4 : 3/4
+	02 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__7to8 : 7/8
+	03 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__15to16 : 15/16
+	04 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__31to32 : 31/32
+	05 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__63to64 : 63/64
+	06 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__127to128 : 127/128
+	07 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__255to256 : 255/256
+*/
+
+/*
+	AUX_REG_UPDATE_5(AUX_DPHY_RX_CONTROL0,
+	AUX_RX_START_WINDOW = 1 [6:4]
+	AUX_RX_RECEIVE_WINDOW = 1 default is 2 [10:8]
+	AUX_RX_HALF_SYM_DETECT_LEN  = 1 [13:12] default is 1
+	AUX_RX_TRANSITION_FILTER_EN = 1 [16] default is 1
+	AUX_RX_ALLOW_BELOW_THRESHOLD_PHASE_DETECT [17] is 0  default is 0
+	AUX_RX_ALLOW_BELOW_THRESHOLD_START [18] is 1  default is 1
+	AUX_RX_ALLOW_BELOW_THRESHOLD_STOP [19] is 1  default is 1
+	AUX_RX_PHASE_DETECT_LEN,  [21,20] = 0x3 default is 3
+	AUX_RX_DETECTION_THRESHOLD [30:28] = 1
+*/
+	// dmub will read AUX_DPHY_RX_CONTROL0/AUX_DPHY_TX_CONTROL from vbios table in dp_aux_init
+
+	//AUX_DPHY_TX_REF_CONTROL'AUX_TX_REF_DIV HW default is 0x32;
+	// Set AUX_TX_REF_DIV Divider to generate 2 MHz reference from refclk
+	// 27MHz -> 0xd
+	// 100MHz -> 0x32
+	// 48MHz -> 0x18
+
+	// Set TMDS_CTL0 to 1.  This is a legacy setting.
+	REG_UPDATE(TMDS_CTL_BITS, TMDS_CTL0, 1);
+
+	dcn10_aux_initialize(enc10);
+}
+
+static const struct link_encoder_funcs dcn31_link_enc_funcs = {
+	.read_state = link_enc2_read_state,
+	.validate_output_with_stream =
+			dcn30_link_encoder_validate_output_with_stream,
+	.hw_init = enc31_hw_init,
+	.setup = dcn10_link_encoder_setup,
+	.enable_tmds_output = dcn10_link_encoder_enable_tmds_output,
+	.enable_dp_output = dcn31_link_encoder_enable_dp_output,
+	.enable_dp_mst_output = dcn31_link_encoder_enable_dp_mst_output,
+	.disable_output = dcn31_link_encoder_disable_output,
+	.dp_set_lane_settings = dcn10_link_encoder_dp_set_lane_settings,
+	.dp_set_phy_pattern = dcn10_link_encoder_dp_set_phy_pattern,
+	.update_mst_stream_allocation_table =
+		dcn10_link_encoder_update_mst_stream_allocation_table,
+	.psr_program_dp_dphy_fast_training =
+			dcn10_psr_program_dp_dphy_fast_training,
+	.psr_program_secondary_packet = dcn10_psr_program_secondary_packet,
+	.connect_dig_be_to_fe = dcn10_link_encoder_connect_dig_be_to_fe,
+	.enable_hpd = dcn10_link_encoder_enable_hpd,
+	.disable_hpd = dcn10_link_encoder_disable_hpd,
+	.is_dig_enabled = dcn10_is_dig_enabled,
+	.destroy = dcn10_link_encoder_destroy,
+	.fec_set_enable = enc2_fec_set_enable,
+	.fec_set_ready = enc2_fec_set_ready,
+	.fec_is_active = enc2_fec_is_active,
+	.get_dig_frontend = dcn10_get_dig_frontend,
+	.get_dig_mode = dcn10_get_dig_mode,
+	.is_in_alt_mode = dcn31_link_encoder_is_in_alt_mode,
+	.get_max_link_cap = dcn31_link_encoder_get_max_link_cap,
+	.set_dio_phy_mux = dcn31_link_encoder_set_dio_phy_mux,
+};
+
+void dcn31_link_encoder_construct(
+	struct dcn20_link_encoder *enc20,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	const struct dcn10_link_enc_aux_registers *aux_regs,
+	const struct dcn10_link_enc_hpd_registers *hpd_regs,
+	const struct dcn10_link_enc_shift *link_shift,
+	const struct dcn10_link_enc_mask *link_mask)
+{
+	struct bp_encoder_cap_info bp_cap_info = {0};
+	const struct dc_vbios_funcs *bp_funcs = init_data->ctx->dc_bios->funcs;
+	enum bp_result result = BP_RESULT_OK;
+	struct dcn10_link_encoder *enc10 = &enc20->enc10;
+
+	enc10->base.funcs = &dcn31_link_enc_funcs;
+	enc10->base.ctx = init_data->ctx;
+	enc10->base.id = init_data->encoder;
+
+	enc10->base.hpd_source = init_data->hpd_source;
+	enc10->base.connector = init_data->connector;
+
+	enc10->base.preferred_engine = ENGINE_ID_UNKNOWN;
+
+	enc10->base.features = *enc_features;
+
+	enc10->base.transmitter = init_data->transmitter;
+
+	/* set the flag to indicate whether driver poll the I2C data pin
+	 * while doing the DP sink detect
+	 */
+
+/*	if (dal_adapter_service_is_feature_supported(as,
+		FEATURE_DP_SINK_DETECT_POLL_DATA_PIN))
+		enc10->base.features.flags.bits.
+			DP_SINK_DETECT_POLL_DATA_PIN = true;*/
+
+	enc10->base.output_signals =
+		SIGNAL_TYPE_DVI_SINGLE_LINK |
+		SIGNAL_TYPE_DVI_DUAL_LINK |
+		SIGNAL_TYPE_LVDS |
+		SIGNAL_TYPE_DISPLAY_PORT |
+		SIGNAL_TYPE_DISPLAY_PORT_MST |
+		SIGNAL_TYPE_EDP |
+		SIGNAL_TYPE_HDMI_TYPE_A;
+
+	/* For DCE 8.0 and 8.1, by design, UNIPHY is hardwired to DIG_BE.
+	 * SW always assign DIG_FE 1:1 mapped to DIG_FE for non-MST UNIPHY.
+	 * SW assign DIG_FE to non-MST UNIPHY first and MST last. So prefer
+	 * DIG is per UNIPHY and used by SST DP, eDP, HDMI, DVI and LVDS.
+	 * Prefer DIG assignment is decided by board design.
+	 * For DCE 8.0, there are only max 6 UNIPHYs, we assume board design
+	 * and VBIOS will filter out 7 UNIPHY for DCE 8.0.
+	 * By this, adding DIGG should not hurt DCE 8.0.
+	 * This will let DCE 8.1 share DCE 8.0 as much as possible
+	 */
+
+	enc10->link_regs = link_regs;
+	enc10->aux_regs = aux_regs;
+	enc10->hpd_regs = hpd_regs;
+	enc10->link_shift = link_shift;
+	enc10->link_mask = link_mask;
+
+	switch (enc10->base.transmitter) {
+	case TRANSMITTER_UNIPHY_A:
+		enc10->base.preferred_engine = ENGINE_ID_DIGA;
+	break;
+	case TRANSMITTER_UNIPHY_B:
+		enc10->base.preferred_engine = ENGINE_ID_DIGB;
+	break;
+	case TRANSMITTER_UNIPHY_C:
+		enc10->base.preferred_engine = ENGINE_ID_DIGC;
+	break;
+	case TRANSMITTER_UNIPHY_D:
+		enc10->base.preferred_engine = ENGINE_ID_DIGD;
+	break;
+	case TRANSMITTER_UNIPHY_E:
+		enc10->base.preferred_engine = ENGINE_ID_DIGE;
+	break;
+	case TRANSMITTER_UNIPHY_F:
+		enc10->base.preferred_engine = ENGINE_ID_DIGF;
+	break;
+	default:
+		ASSERT_CRITICAL(false);
+		enc10->base.preferred_engine = ENGINE_ID_UNKNOWN;
+	}
+
+	/* default to one to mirror Windows behavior */
+	enc10->base.features.flags.bits.HDMI_6GB_EN = 1;
+
+	result = bp_funcs->get_encoder_cap_info(enc10->base.ctx->dc_bios,
+						enc10->base.id, &bp_cap_info);
+
+	/* Override features with DCE-specific values */
+	if (result == BP_RESULT_OK) {
+		enc10->base.features.flags.bits.IS_HBR2_CAPABLE =
+				bp_cap_info.DP_HBR2_EN;
+		enc10->base.features.flags.bits.IS_HBR3_CAPABLE =
+				bp_cap_info.DP_HBR3_EN;
+		enc10->base.features.flags.bits.HDMI_6GB_EN = bp_cap_info.HDMI_6GB_EN;
+		enc10->base.features.flags.bits.IS_DP2_CAPABLE = bp_cap_info.IS_DP2_CAPABLE;
+		enc10->base.features.flags.bits.IS_UHBR10_CAPABLE = bp_cap_info.DP_UHBR10_EN;
+		enc10->base.features.flags.bits.IS_UHBR13_5_CAPABLE = bp_cap_info.DP_UHBR13_5_EN;
+		enc10->base.features.flags.bits.IS_UHBR20_CAPABLE = bp_cap_info.DP_UHBR20_EN;
+		enc10->base.features.flags.bits.DP_IS_USB_C =
+				bp_cap_info.DP_IS_USB_C;
+	} else {
+		DC_LOG_WARNING("%s: Failed to get encoder_cap_info from VBIOS with error code %d!\n",
+				__func__,
+				result);
+	}
+	if (enc10->base.ctx->dc->debug.hdmi20_disable) {
+		enc10->base.features.flags.bits.HDMI_6GB_EN = 0;
+	}
+}
+
+void dcn31_link_encoder_construct_minimal(
+	struct dcn20_link_encoder *enc20,
+	struct dc_context *ctx,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	enum engine_id eng_id)
+{
+	struct dcn10_link_encoder *enc10 = &enc20->enc10;
+
+	enc10->base.funcs = &dcn31_link_enc_funcs;
+	enc10->base.ctx = ctx;
+	enc10->base.id.type = OBJECT_TYPE_ENCODER;
+	enc10->base.hpd_source = HPD_SOURCEID_UNKNOWN;
+	enc10->base.connector.type = OBJECT_TYPE_CONNECTOR;
+	enc10->base.preferred_engine = eng_id;
+	enc10->base.features = *enc_features;
+	enc10->base.transmitter = TRANSMITTER_UNKNOWN;
+	enc10->link_regs = link_regs;
+
+	enc10->base.output_signals =
+		SIGNAL_TYPE_DISPLAY_PORT |
+		SIGNAL_TYPE_DISPLAY_PORT_MST |
+		SIGNAL_TYPE_EDP;
+}
+
+/* DPIA equivalent of link_transmitter_control. */
+static bool link_dpia_control(struct dc_context *dc_ctx,
+	struct dmub_cmd_dig_dpia_control_data *dpia_control)
+{
+	union dmub_rb_cmd cmd;
+
+	memset(&cmd, 0, sizeof(cmd));
+
+	cmd.dig1_dpia_control.header.type = DMUB_CMD__DPIA;
+	cmd.dig1_dpia_control.header.sub_type =
+			DMUB_CMD__DPIA_DIG1_DPIA_CONTROL;
+	cmd.dig1_dpia_control.header.payload_bytes =
+		sizeof(cmd.dig1_dpia_control) -
+		sizeof(cmd.dig1_dpia_control.header);
+
+	cmd.dig1_dpia_control.dpia_control = *dpia_control;
+
+	dc_wake_and_execute_dmub_cmd(dc_ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+
+	return true;
+}
+
+static void link_encoder_disable(struct dcn10_link_encoder *enc10)
+{
+	/* reset training complete */
+	REG_UPDATE(DP_LINK_CNTL, DP_LINK_TRAINING_COMPLETE, 0);
+}
+
+void dcn31_link_encoder_enable_dp_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	/* Enable transmitter and encoder. */
+	if (!link_enc_cfg_is_transmitter_mappable(enc->ctx->dc, enc)) {
+
+		DC_LOG_DEBUG("%s: enc_id(%d)\n", __func__, enc->preferred_engine);
+		dcn20_link_encoder_enable_dp_output(enc, link_settings, clock_source);
+
+	} else {
+
+		struct dmub_cmd_dig_dpia_control_data dpia_control = { 0 };
+		struct dc_link *link;
+
+		link = link_enc_cfg_get_link_using_link_enc(enc->ctx->dc, enc->preferred_engine);
+
+		enc1_configure_encoder(enc10, link_settings);
+
+		dpia_control.action = (uint8_t)TRANSMITTER_CONTROL_ENABLE;
+		dpia_control.enc_id = enc->preferred_engine;
+		dpia_control.mode_laneset.digmode = 0; /* 0 for SST; 5 for MST */
+		dpia_control.lanenum = (uint8_t)link_settings->lane_count;
+		dpia_control.symclk_10khz = link_settings->link_rate *
+				LINK_RATE_REF_FREQ_IN_KHZ / 10;
+		/* DIG_BE_CNTL.DIG_HPD_SELECT set to 5 (hpdsel - 1) to indicate HPD pin
+		 * unused by DPIA.
+		 */
+		dpia_control.hpdsel = 6;
+
+		if (link) {
+			dpia_control.dpia_id = link->ddc_hw_inst;
+			dpia_control.fec_rdy = link->dc->link_srv->dp_should_enable_fec(link);
+		} else {
+			DC_LOG_ERROR("%s: Failed to execute DPIA enable DMUB command.\n", __func__);
+			BREAK_TO_DEBUGGER();
+			return;
+		}
+
+		DC_LOG_DEBUG("%s: DPIA(%d) - enc_id(%d)\n", __func__, dpia_control.dpia_id, dpia_control.enc_id);
+		link_dpia_control(enc->ctx, &dpia_control);
+	}
+}
+
+void dcn31_link_encoder_enable_dp_mst_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	/* Enable transmitter and encoder. */
+	if (!link_enc_cfg_is_transmitter_mappable(enc->ctx->dc, enc)) {
+
+		DC_LOG_DEBUG("%s: enc_id(%d)\n", __func__, enc->preferred_engine);
+		dcn10_link_encoder_enable_dp_mst_output(enc, link_settings, clock_source);
+
+	} else {
+
+		struct dmub_cmd_dig_dpia_control_data dpia_control = { 0 };
+		struct dc_link *link;
+
+		link = link_enc_cfg_get_link_using_link_enc(enc->ctx->dc, enc->preferred_engine);
+
+		enc1_configure_encoder(enc10, link_settings);
+
+		dpia_control.action = (uint8_t)TRANSMITTER_CONTROL_ENABLE;
+		dpia_control.enc_id = enc->preferred_engine;
+		dpia_control.mode_laneset.digmode = 5; /* 0 for SST; 5 for MST */
+		dpia_control.lanenum = (uint8_t)link_settings->lane_count;
+		dpia_control.symclk_10khz = link_settings->link_rate *
+				LINK_RATE_REF_FREQ_IN_KHZ / 10;
+		/* DIG_BE_CNTL.DIG_HPD_SELECT set to 5 (hpdsel - 1) to indicate HPD pin
+		 * unused by DPIA.
+		 */
+		dpia_control.hpdsel = 6;
+
+		if (link) {
+			dpia_control.dpia_id = link->ddc_hw_inst;
+			dpia_control.fec_rdy = link->dc->link_srv->dp_should_enable_fec(link);
+		} else {
+			DC_LOG_ERROR("%s: Failed to execute DPIA enable DMUB command.\n", __func__);
+			BREAK_TO_DEBUGGER();
+			return;
+		}
+
+		DC_LOG_DEBUG("%s: DPIA(%d) - enc_id(%d)\n", __func__, dpia_control.dpia_id, dpia_control.enc_id);
+		link_dpia_control(enc->ctx, &dpia_control);
+	}
+}
+
+void dcn31_link_encoder_disable_output(
+	struct link_encoder *enc,
+	enum signal_type signal)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	/* Disable transmitter and encoder. */
+	if (!link_enc_cfg_is_transmitter_mappable(enc->ctx->dc, enc)) {
+
+		DC_LOG_DEBUG("%s: enc_id(%d)\n", __func__, enc->preferred_engine);
+		dcn10_link_encoder_disable_output(enc, signal);
+
+	} else {
+
+		struct dmub_cmd_dig_dpia_control_data dpia_control = { 0 };
+		struct dc_link *link;
+
+		if (enc->funcs->is_dig_enabled && !enc->funcs->is_dig_enabled(enc))
+			return;
+
+		link = link_enc_cfg_get_link_using_link_enc(enc->ctx->dc, enc->preferred_engine);
+
+		dpia_control.action = (uint8_t)TRANSMITTER_CONTROL_DISABLE;
+		dpia_control.enc_id = enc->preferred_engine;
+		if (signal == SIGNAL_TYPE_DISPLAY_PORT) {
+			dpia_control.mode_laneset.digmode = 0; /* 0 for SST; 5 for MST */
+		} else if (signal == SIGNAL_TYPE_DISPLAY_PORT_MST) {
+			dpia_control.mode_laneset.digmode = 5; /* 0 for SST; 5 for MST */
+		} else {
+			DC_LOG_ERROR("%s: USB4 DPIA only supports DisplayPort.\n", __func__);
+			BREAK_TO_DEBUGGER();
+		}
+
+		if (link) {
+			dpia_control.dpia_id = link->ddc_hw_inst;
+		} else {
+			DC_LOG_ERROR("%s: Failed to execute DPIA enable DMUB command.\n", __func__);
+			BREAK_TO_DEBUGGER();
+			return;
+		}
+
+		DC_LOG_DEBUG("%s: DPIA(%d) - enc_id(%d)\n", __func__, dpia_control.dpia_id, dpia_control.enc_id);
+		link_dpia_control(enc->ctx, &dpia_control);
+
+		link_encoder_disable(enc10);
+	}
+}
+
+bool dcn31_link_encoder_is_in_alt_mode(struct link_encoder *enc)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	union dmub_rb_cmd cmd;
+	uint32_t dp_alt_mode_disable;
+
+	/* Only applicable to USB-C PHY. */
+	if (!enc->features.flags.bits.DP_IS_USB_C)
+		return false;
+
+	/*
+	 * Use the new interface from DMCUB if available.
+	 * Avoids hanging the RDCPSPIPE if DMCUB wasn't already running.
+	 */
+	if (has_query_dp_alt(enc)) {
+		if (!query_dp_alt_from_dmub(enc, &cmd))
+			return false;
+
+		return (cmd.query_dp_alt.data.is_dp_alt_disable == 0);
+	}
+
+	/* Legacy path, avoid if possible. */
+	if (enc->ctx->asic_id.hw_internal_rev != YELLOW_CARP_B0) {
+		REG_GET(RDPCSTX_PHY_CNTL6, RDPCS_PHY_DPALT_DISABLE,
+			&dp_alt_mode_disable);
+	} else {
+		/*
+		 * B0 phys use a new set of registers to check whether alt mode is disabled.
+		 * if value == 1 alt mode is disabled, otherwise it is enabled.
+		 */
+		if ((enc10->base.transmitter == TRANSMITTER_UNIPHY_A) ||
+		    (enc10->base.transmitter == TRANSMITTER_UNIPHY_B) ||
+		    (enc10->base.transmitter == TRANSMITTER_UNIPHY_E)) {
+			REG_GET(RDPCSTX_PHY_CNTL6, RDPCS_PHY_DPALT_DISABLE,
+				&dp_alt_mode_disable);
+		} else {
+			REG_GET(RDPCSPIPE_PHY_CNTL6, RDPCS_PHY_DPALT_DISABLE,
+				&dp_alt_mode_disable);
+		}
+	}
+
+	return (dp_alt_mode_disable == 0);
+}
+
+void dcn31_link_encoder_get_max_link_cap(struct link_encoder *enc, struct dc_link_settings *link_settings)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	union dmub_rb_cmd cmd;
+	uint32_t is_in_usb_c_dp4_mode = 0;
+
+	dcn10_link_encoder_get_max_link_cap(enc, link_settings);
+
+	/* Take the link cap directly if not USB */
+	if (!enc->features.flags.bits.DP_IS_USB_C)
+		return;
+
+	/*
+	 * Use the new interface from DMCUB if available.
+	 * Avoids hanging the RDCPSPIPE if DMCUB wasn't already running.
+	 */
+	if (has_query_dp_alt(enc)) {
+		if (!query_dp_alt_from_dmub(enc, &cmd))
+			return;
+
+		if (cmd.query_dp_alt.data.is_dp_alt_disable == 0 &&
+				cmd.query_dp_alt.data.is_usb &&
+				cmd.query_dp_alt.data.is_dp4 == 0)
+			link_settings->lane_count = MIN(LANE_COUNT_TWO, link_settings->lane_count);
+
+		return;
+	}
+
+	/* Legacy path, avoid if possible. */
+	if (enc->ctx->asic_id.hw_internal_rev != YELLOW_CARP_B0) {
+		REG_GET(RDPCSTX_PHY_CNTL6, RDPCS_PHY_DPALT_DP4,
+			&is_in_usb_c_dp4_mode);
+	} else {
+		if ((enc10->base.transmitter == TRANSMITTER_UNIPHY_A) ||
+		    (enc10->base.transmitter == TRANSMITTER_UNIPHY_B) ||
+		    (enc10->base.transmitter == TRANSMITTER_UNIPHY_E)) {
+			REG_GET(RDPCSTX_PHY_CNTL6, RDPCS_PHY_DPALT_DP4,
+				&is_in_usb_c_dp4_mode);
+		} else {
+			REG_GET(RDPCSPIPE_PHY_CNTL6, RDPCS_PHY_DPALT_DP4,
+				&is_in_usb_c_dp4_mode);
+		}
+	}
+
+	if (!is_in_usb_c_dp4_mode)
+		link_settings->lane_count = MIN(LANE_COUNT_TWO, link_settings->lane_count);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn31/dcn31_dio_link_encoder.h b/drivers/gpu/drm/amd/display/dc/dio/dcn31/dcn31_dio_link_encoder.h
new file mode 100644
index 000000000000..ee78ba80797c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn31/dcn31_dio_link_encoder.h
@@ -0,0 +1,290 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_LINK_ENCODER__DCN31_H__
+#define __DC_LINK_ENCODER__DCN31_H__
+
+#include "dcn30/dcn30_dio_link_encoder.h"
+
+
+#define LE_DCN31_REG_LIST(id)\
+	LE_DCN3_REG_LIST(id),\
+	SRI(DP_DPHY_INTERNAL_CTRL, DP, id), \
+	SR(DIO_LINKA_CNTL), \
+	SR(DIO_LINKB_CNTL), \
+	SR(DIO_LINKC_CNTL), \
+	SR(DIO_LINKD_CNTL), \
+	SR(DIO_LINKE_CNTL), \
+	SR(DIO_LINKF_CNTL)
+
+#define LINK_ENCODER_MASK_SH_LIST_DCN31(mask_sh) \
+	LINK_ENCODER_MASK_SH_LIST_DCN10(mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_FEC_EN, mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_FEC_READY_SHADOW, mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_FEC_ACTIVE_STATUS, mask_sh),\
+	LE_SF(DIG0_TMDS_CTL_BITS, TMDS_CTL0, mask_sh), \
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_START_WINDOW, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_HALF_SYM_DETECT_LEN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_TRANSITION_FILTER_EN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_ALLOW_BELOW_THRESHOLD_PHASE_DETECT, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_ALLOW_BELOW_THRESHOLD_START, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_ALLOW_BELOW_THRESHOLD_STOP, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_PHASE_DETECT_LEN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_DETECTION_THRESHOLD, mask_sh), \
+	LE_SF(DP_AUX0_AUX_DPHY_TX_CONTROL, AUX_TX_PRECHARGE_LEN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_TX_CONTROL, AUX_TX_PRECHARGE_SYMBOLS, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_TX_CONTROL, AUX_MODE_DET_CHECK_DELAY, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL1, AUX_RX_PRECHARGE_SKIP, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL1, AUX_RX_TIMEOUT_LEN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL1, AUX_RX_TIMEOUT_LEN_MUL, mask_sh),\
+	LE_SF(DIO_LINKA_CNTL, ENC_TYPE_SEL, mask_sh),\
+	LE_SF(DIO_LINKA_CNTL, HPO_DP_ENC_SEL, mask_sh),\
+	LE_SF(DIO_LINKA_CNTL, HPO_HDMI_ENC_SEL, mask_sh)
+
+#define DPCS_DCN31_REG_LIST(id) \
+	SRI(TMDS_CTL_BITS, DIG, id), \
+	SRI(RDPCSTX_PHY_CNTL3, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL4, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL5, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL6, RDPCSTX, id), \
+	SRI(RDPCSPIPE_PHY_CNTL6, RDPCSPIPE, id), \
+	SRI(RDPCSTX_PHY_CNTL7, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL8, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL9, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL10, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL11, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL12, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL13, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL14, RDPCSTX, id), \
+	SRI(RDPCSTX_CNTL, RDPCSTX, id), \
+	SRI(RDPCSTX_CLOCK_CNTL, RDPCSTX, id), \
+	SRI(RDPCSTX_INTERRUPT_CONTROL, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL0, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL2, RDPCSTX, id), \
+	SRI(RDPCS_TX_CR_ADDR, RDPCSTX, id), \
+	SRI(RDPCS_TX_CR_DATA, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_FUSE0, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_FUSE1, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_FUSE2, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_FUSE3, RDPCSTX, id), \
+	SRI(RDPCSTX_DEBUG_CONFIG, RDPCSTX, id), \
+	SR(RDPCSTX0_RDPCSTX_SCRATCH), \
+	SRI(RDPCSTX_PHY_RX_LD_VAL, RDPCSTX, id),\
+	SRI(RDPCSTX_DMCU_DPALT_DIS_BLOCK_REG, RDPCSTX, id)
+
+#define DPCS_DCN31_MASK_SH_LIST(mask_sh)\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX0_CLK_RDY, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX0_DATA_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX1_CLK_RDY, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX1_DATA_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX2_CLK_RDY, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX2_DATA_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX3_CLK_RDY, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX3_DATA_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL4, RDPCS_PHY_DP_TX0_TERM_CTRL, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL4, RDPCS_PHY_DP_TX1_TERM_CTRL, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL4, RDPCS_PHY_DP_TX2_TERM_CTRL, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL4, RDPCS_PHY_DP_TX3_TERM_CTRL, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL11, RDPCS_PHY_DP_MPLLB_MULTIPLIER, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL5, RDPCS_PHY_DP_TX0_WIDTH, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL5, RDPCS_PHY_DP_TX0_RATE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL5, RDPCS_PHY_DP_TX1_WIDTH, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL5, RDPCS_PHY_DP_TX1_RATE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DP_TX2_PSTATE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DP_TX3_PSTATE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DP_TX2_MPLL_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DP_TX3_MPLL_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL6, RDPCS_PHY_DPALT_DP4, mask_sh),\
+	LE_SF(RDPCSPIPE0_RDPCSPIPE_PHY_CNTL6, RDPCS_PHY_DPALT_DP4, mask_sh),\
+	LE_SF(RDPCSPIPE0_RDPCSPIPE_PHY_CNTL6, RDPCS_PHY_DPALT_DISABLE, mask_sh),\
+	LE_SF(RDPCSPIPE0_RDPCSPIPE_PHY_CNTL6, RDPCS_PHY_DPALT_DISABLE_ACK, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL7, RDPCS_PHY_DP_MPLLB_FRACN_QUOT, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL7, RDPCS_PHY_DP_MPLLB_FRACN_DEN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL8, RDPCS_PHY_DP_MPLLB_SSC_PEAK, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL9, RDPCS_PHY_DP_MPLLB_SSC_UP_SPREAD, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL9, RDPCS_PHY_DP_MPLLB_SSC_STEPSIZE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL10, RDPCS_PHY_DP_MPLLB_FRACN_REM, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL11, RDPCS_PHY_DP_REF_CLK_MPLLB_DIV, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL11, RDPCS_PHY_HDMI_MPLLB_HDMI_DIV, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL12, RDPCS_PHY_DP_MPLLB_SSC_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL12, RDPCS_PHY_DP_MPLLB_DIV5_CLK_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL12, RDPCS_PHY_DP_MPLLB_TX_CLK_DIV, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL12, RDPCS_PHY_DP_MPLLB_WORD_DIV2_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL12, RDPCS_PHY_DP_MPLLB_STATE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL13, RDPCS_PHY_DP_MPLLB_DIV_CLK_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL13, RDPCS_PHY_DP_MPLLB_DIV_MULTIPLIER, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL14, RDPCS_PHY_DP_MPLLB_FRACN_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL14, RDPCS_PHY_DP_MPLLB_PMIX_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CNTL, RDPCS_TX_FIFO_LANE0_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CNTL, RDPCS_TX_FIFO_LANE1_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CNTL, RDPCS_TX_FIFO_LANE2_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CNTL, RDPCS_TX_FIFO_LANE3_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CNTL, RDPCS_TX_FIFO_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CNTL, RDPCS_TX_FIFO_RD_START_DELAY, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CLOCK_CNTL, RDPCS_EXT_REFCLK_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CLOCK_CNTL, RDPCS_SRAMCLK_BYPASS, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CLOCK_CNTL, RDPCS_SRAMCLK_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CLOCK_CNTL, RDPCS_SRAMCLK_CLOCK_ON, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CLOCK_CNTL, RDPCS_SYMCLK_DIV2_CLOCK_ON, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CLOCK_CNTL, RDPCS_SYMCLK_DIV2_GATE_DIS, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_CLOCK_CNTL, RDPCS_SYMCLK_DIV2_EN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX0_DISABLE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX1_DISABLE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX2_DISABLE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX3_DISABLE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX0_REQ, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX1_REQ, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX2_REQ, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX3_REQ, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX0_ACK, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX1_ACK, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX2_ACK, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX3_ACK, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX0_RESET, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX1_RESET, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX2_RESET, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL3, RDPCS_PHY_DP_TX3_RESET, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL0, RDPCS_PHY_RESET, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL0, RDPCS_PHY_CR_MUX_SEL, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL0, RDPCS_PHY_REF_RANGE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL0, RDPCS_SRAM_BYPASS, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL0, RDPCS_SRAM_EXT_LD_DONE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL0, RDPCS_PHY_HDMIMODE_ENABLE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL0, RDPCS_SRAM_INIT_DONE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_CNTL2, RDPCS_PHY_DP4_POR, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_INTERRUPT_CONTROL, RDPCS_REG_FIFO_ERROR_MASK, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_INTERRUPT_CONTROL, RDPCS_TX_FIFO_ERROR_MASK, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_INTERRUPT_CONTROL, RDPCS_DPALT_DISABLE_TOGGLE_MASK, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_INTERRUPT_CONTROL, RDPCS_DPALT_4LANE_TOGGLE_MASK, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCS_TX_CR_ADDR, RDPCS_TX_CR_ADDR, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCS_TX_CR_DATA, RDPCS_TX_CR_DATA, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE0, RDPCS_PHY_DP_MPLLB_V2I, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE0, RDPCS_PHY_DP_TX0_EQ_MAIN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE0, RDPCS_PHY_DP_TX0_EQ_PRE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE0, RDPCS_PHY_DP_TX0_EQ_POST, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE0, RDPCS_PHY_DP_MPLLB_FREQ_VCO, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE1, RDPCS_PHY_DP_MPLLB_CP_INT, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE1, RDPCS_PHY_DP_MPLLB_CP_PROP, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE1, RDPCS_PHY_DP_TX1_EQ_MAIN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE1, RDPCS_PHY_DP_TX1_EQ_PRE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE1, RDPCS_PHY_DP_TX1_EQ_POST, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE2, RDPCS_PHY_DP_TX2_EQ_MAIN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE2, RDPCS_PHY_DP_TX2_EQ_PRE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE2, RDPCS_PHY_DP_TX2_EQ_POST, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE3, RDPCS_PHY_DP_TX3_EQ_MAIN, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE3, RDPCS_PHY_DCO_FINETUNE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE3, RDPCS_PHY_DCO_RANGE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE3, RDPCS_PHY_DP_TX3_EQ_PRE, mask_sh),\
+	LE_SF(RDPCSTX0_RDPCSTX_PHY_FUSE3, RDPCS_PHY_DP_TX3_EQ_POST, mask_sh)
+
+#define DPCS_DCN314_REG_LIST(id) \
+	SRI(TMDS_CTL_BITS, DIG, id), \
+	SRI(RDPCSTX_PHY_CNTL3, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL4, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL5, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL7, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL8, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL9, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL10, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL11, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL12, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL13, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL14, RDPCSTX, id), \
+	SRI(RDPCSTX_CNTL, RDPCSTX, id), \
+	SRI(RDPCSTX_CLOCK_CNTL, RDPCSTX, id), \
+	SRI(RDPCSTX_INTERRUPT_CONTROL, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL0, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_CNTL2, RDPCSTX, id), \
+	SRI(RDPCS_TX_CR_ADDR, RDPCSTX, id), \
+	SRI(RDPCS_TX_CR_DATA, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_FUSE0, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_FUSE1, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_FUSE2, RDPCSTX, id), \
+	SRI(RDPCSTX_PHY_FUSE3, RDPCSTX, id), \
+	SRI(RDPCSTX_DEBUG_CONFIG, RDPCSTX, id), \
+	SR(RDPCSTX0_RDPCSTX_SCRATCH), \
+	SRI(RDPCSTX_PHY_RX_LD_VAL, RDPCSTX, id),\
+	SRI(RDPCSTX_DMCU_DPALT_DIS_BLOCK_REG, RDPCSTX, id)
+
+void dcn31_link_encoder_construct(
+	struct dcn20_link_encoder *enc20,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	const struct dcn10_link_enc_aux_registers *aux_regs,
+	const struct dcn10_link_enc_hpd_registers *hpd_regs,
+	const struct dcn10_link_enc_shift *link_shift,
+	const struct dcn10_link_enc_mask *link_mask);
+
+/*
+ * Create a minimal link encoder object with no dc_link object associated with it.
+ */
+void dcn31_link_encoder_construct_minimal(
+	struct dcn20_link_encoder *enc20,
+	struct dc_context *ctx,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	enum engine_id eng_id);
+
+void dcn31_link_encoder_set_dio_phy_mux(
+	struct link_encoder *enc,
+	enum encoder_type_select sel,
+	uint32_t hpo_inst);
+
+/*
+ * Enable DP transmitter and its encoder.
+ */
+void dcn31_link_encoder_enable_dp_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source);
+
+/*
+ * Enable DP transmitter and its encoder in MST mode.
+ */
+void dcn31_link_encoder_enable_dp_mst_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source);
+
+/*
+ * Disable transmitter and its encoder.
+ */
+void dcn31_link_encoder_disable_output(
+	struct link_encoder *enc,
+	enum signal_type signal);
+
+/*
+ * Check whether USB-C DP Alt mode is disabled
+ */
+bool dcn31_link_encoder_is_in_alt_mode(
+	struct link_encoder *enc);
+
+void dcn31_link_encoder_get_max_link_cap(struct link_encoder *enc,
+	struct dc_link_settings *link_settings);
+
+void enc31_hw_init(struct link_encoder *enc);
+
+#endif /* __DC_LINK_ENCODER__DCN31_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn314/dcn314_dio_stream_encoder.c b/drivers/gpu/drm/amd/display/dc/dio/dcn314/dcn314_dio_stream_encoder.c
new file mode 100644
index 000000000000..3e85e9c3d2cb
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn314/dcn314_dio_stream_encoder.c
@@ -0,0 +1,506 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+#include "dc_bios_types.h"
+#include "dcn30/dcn30_dio_stream_encoder.h"
+#include "dcn314_dio_stream_encoder.h"
+#include "reg_helper.h"
+#include "hw_shared.h"
+#include "link_service.h"
+#include "dpcd_defs.h"
+
+#define DC_LOGGER \
+		enc1->base.ctx->logger
+
+#define REG(reg)\
+	(enc1->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	enc1->se_shift->field_name, enc1->se_mask->field_name
+
+#define VBI_LINE_0 0
+#define HDMI_CLOCK_CHANNEL_RATE_MORE_340M 340000
+
+#define CTX \
+	enc1->base.ctx
+
+void enc314_reset_fifo(struct stream_encoder *enc, bool reset)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+	uint32_t reset_val = reset ? 1 : 0;
+	uint32_t is_symclk_on;
+
+	REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_RESET, reset_val);
+	REG_GET(DIG_FE_CNTL, DIG_SYMCLK_FE_ON, &is_symclk_on);
+
+	if (is_symclk_on)
+		REG_WAIT(DIG_FIFO_CTRL0, DIG_FIFO_RESET_DONE, reset_val, 10, 5000);
+	else
+		udelay(10);
+}
+
+void enc314_enable_fifo(struct stream_encoder *enc)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_READ_START_LEVEL, 0x7);
+
+	enc314_reset_fifo(enc, true);
+	enc314_reset_fifo(enc, false);
+
+	REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_ENABLE, 1);
+}
+
+void enc314_disable_fifo(struct stream_encoder *enc)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_ENABLE, 0);
+}
+
+static bool enc314_is_fifo_enabled(struct stream_encoder *enc)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+	uint32_t reset_val;
+
+	REG_GET(DIG_FIFO_CTRL0, DIG_FIFO_ENABLE, &reset_val);
+	return (reset_val != 0);
+}
+
+void enc314_dp_set_odm_combine(
+	struct stream_encoder *enc,
+	bool odm_combine)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	REG_UPDATE(DP_PIXEL_FORMAT, DP_PIXEL_PER_CYCLE_PROCESSING_MODE, odm_combine);
+}
+
+/* setup stream encoder in dvi mode */
+void enc314_stream_encoder_dvi_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	bool is_dual_link)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (!enc->ctx->dc->debug.avoid_vbios_exec_table) {
+		struct bp_encoder_control cntl = {0};
+
+		cntl.action = ENCODER_CONTROL_SETUP;
+		cntl.engine_id = enc1->base.id;
+		cntl.signal = is_dual_link ?
+			SIGNAL_TYPE_DVI_DUAL_LINK : SIGNAL_TYPE_DVI_SINGLE_LINK;
+		cntl.enable_dp_audio = false;
+		cntl.pixel_clock = crtc_timing->pix_clk_100hz / 10;
+		cntl.lanes_number = (is_dual_link) ? LANE_COUNT_EIGHT : LANE_COUNT_FOUR;
+
+		if (enc1->base.bp->funcs->encoder_control(
+				enc1->base.bp, &cntl) != BP_RESULT_OK)
+			return;
+
+	} else {
+
+		//Set pattern for clock channel, default vlue 0x63 does not work
+		REG_UPDATE(DIG_CLOCK_PATTERN, DIG_CLOCK_PATTERN, 0x1F);
+
+		//DIG_BE_TMDS_DVI_MODE : TMDS-DVI mode is already set in link_encoder_setup
+
+		//DIG_SOURCE_SELECT is already set in dig_connect_to_otg
+
+		enc314_enable_fifo(enc);
+	}
+
+	ASSERT(crtc_timing->pixel_encoding == PIXEL_ENCODING_RGB);
+	ASSERT(crtc_timing->display_color_depth == COLOR_DEPTH_888);
+	enc1_stream_encoder_set_stream_attribute_helper(enc1, crtc_timing);
+}
+
+/* setup stream encoder in hdmi mode */
+void enc314_stream_encoder_hdmi_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	int actual_pix_clk_khz,
+	bool enable_audio)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (!enc->ctx->dc->debug.avoid_vbios_exec_table) {
+		struct bp_encoder_control cntl = {0};
+
+		cntl.action = ENCODER_CONTROL_SETUP;
+		cntl.engine_id = enc1->base.id;
+		cntl.signal = SIGNAL_TYPE_HDMI_TYPE_A;
+		cntl.enable_dp_audio = enable_audio;
+		cntl.pixel_clock = actual_pix_clk_khz;
+		cntl.lanes_number = LANE_COUNT_FOUR;
+
+		if (enc1->base.bp->funcs->encoder_control(
+				enc1->base.bp, &cntl) != BP_RESULT_OK)
+			return;
+
+	} else {
+
+		//Set pattern for clock channel, default vlue 0x63 does not work
+		REG_UPDATE(DIG_CLOCK_PATTERN, DIG_CLOCK_PATTERN, 0x1F);
+
+		//DIG_BE_TMDS_HDMI_MODE : TMDS-HDMI mode is already set in link_encoder_setup
+
+		//DIG_SOURCE_SELECT is already set in dig_connect_to_otg
+
+		enc314_enable_fifo(enc);
+	}
+
+	/* Configure pixel encoding */
+	enc1_stream_encoder_set_stream_attribute_helper(enc1, crtc_timing);
+
+	/* setup HDMI engine */
+	REG_UPDATE_6(HDMI_CONTROL,
+		HDMI_PACKET_GEN_VERSION, 1,
+		HDMI_KEEPOUT_MODE, 1,
+		HDMI_DEEP_COLOR_ENABLE, 0,
+		HDMI_DATA_SCRAMBLE_EN, 0,
+		HDMI_NO_EXTRA_NULL_PACKET_FILLED, 1,
+		HDMI_CLOCK_CHANNEL_RATE, 0);
+
+	/* Configure color depth */
+	switch (crtc_timing->display_color_depth) {
+	case COLOR_DEPTH_888:
+		REG_UPDATE(HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 0);
+		break;
+	case COLOR_DEPTH_101010:
+		if (crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR422) {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 1,
+					HDMI_DEEP_COLOR_ENABLE, 0);
+		} else {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 1,
+					HDMI_DEEP_COLOR_ENABLE, 1);
+			}
+		break;
+	case COLOR_DEPTH_121212:
+		if (crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR422) {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 2,
+					HDMI_DEEP_COLOR_ENABLE, 0);
+		} else {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 2,
+					HDMI_DEEP_COLOR_ENABLE, 1);
+			}
+		break;
+	case COLOR_DEPTH_161616:
+		REG_UPDATE_2(HDMI_CONTROL,
+				HDMI_DEEP_COLOR_DEPTH, 3,
+				HDMI_DEEP_COLOR_ENABLE, 1);
+		break;
+	default:
+		break;
+	}
+
+	if (actual_pix_clk_khz >= HDMI_CLOCK_CHANNEL_RATE_MORE_340M) {
+		/* enable HDMI data scrambler
+		 * HDMI_CLOCK_CHANNEL_RATE_MORE_340M
+		 * Clock channel frequency is 1/4 of character rate.
+		 */
+		REG_UPDATE_2(HDMI_CONTROL,
+			HDMI_DATA_SCRAMBLE_EN, 1,
+			HDMI_CLOCK_CHANNEL_RATE, 1);
+	} else if (crtc_timing->flags.LTE_340MCSC_SCRAMBLE) {
+
+		/* TODO: New feature for DCE11, still need to implement */
+
+		/* enable HDMI data scrambler
+		 * HDMI_CLOCK_CHANNEL_FREQ_EQUAL_TO_CHAR_RATE
+		 * Clock channel frequency is the same
+		 * as character rate
+		 */
+		REG_UPDATE_2(HDMI_CONTROL,
+			HDMI_DATA_SCRAMBLE_EN, 1,
+			HDMI_CLOCK_CHANNEL_RATE, 0);
+	}
+
+
+	/* Enable transmission of General Control packet on every frame */
+	REG_UPDATE_3(HDMI_VBI_PACKET_CONTROL,
+		HDMI_GC_CONT, 1,
+		HDMI_GC_SEND, 1,
+		HDMI_NULL_SEND, 1);
+
+	/* Disable Audio Content Protection packet transmission */
+	REG_UPDATE(HDMI_VBI_PACKET_CONTROL, HDMI_ACP_SEND, 0);
+
+	/* following belongs to audio */
+	/* Enable Audio InfoFrame packet transmission. */
+	REG_UPDATE(HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, 1);
+
+	/* update double-buffered AUDIO_INFO registers immediately */
+	ASSERT(enc->afmt);
+	enc->afmt->funcs->audio_info_immediate_update(enc->afmt);
+
+	/* Select line number on which to send Audio InfoFrame packets */
+	REG_UPDATE(HDMI_INFOFRAME_CONTROL1, HDMI_AUDIO_INFO_LINE,
+				VBI_LINE_0 + 2);
+
+	/* set HDMI GC AVMUTE */
+	REG_UPDATE(HDMI_GC, HDMI_GC_AVMUTE, 0);
+}
+
+
+
+static bool is_two_pixels_per_containter(const struct dc_crtc_timing *timing)
+{
+	bool two_pix = timing->pixel_encoding == PIXEL_ENCODING_YCBCR420;
+
+	two_pix = two_pix || (timing->flags.DSC && timing->pixel_encoding == PIXEL_ENCODING_YCBCR422
+			&& !timing->dsc_cfg.ycbcr422_simple);
+	return two_pix;
+}
+
+void enc314_stream_encoder_dp_blank(
+	struct dc_link *link,
+	struct stream_encoder *enc)
+{
+	enc1_stream_encoder_dp_blank(link, enc);
+
+	/* Disable FIFO after the DP vid stream is disabled to avoid corruption. */
+	if (enc->ctx->dc->debug.dig_fifo_off_in_blank)
+		enc314_disable_fifo(enc);
+}
+
+void enc314_stream_encoder_dp_unblank(
+		struct dc_link *link,
+		struct stream_encoder *enc,
+		const struct encoder_unblank_param *param)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (param->link_settings.link_rate != LINK_RATE_UNKNOWN) {
+		uint32_t n_vid = 0x8000;
+		uint32_t m_vid;
+		uint32_t n_multiply = 0;
+		uint32_t pix_per_cycle = 0;
+		uint64_t m_vid_l = n_vid;
+
+		/* YCbCr 4:2:0 : Computed VID_M will be 2X the input rate */
+		if (is_two_pixels_per_containter(&param->timing) || param->opp_cnt > 1) {
+			/*this logic should be the same in get_pixel_clock_parameters() */
+			n_multiply = 1;
+			pix_per_cycle = 1;
+		}
+		/* M / N = Fstream / Flink
+		 * m_vid / n_vid = pixel rate / link rate
+		 */
+
+		m_vid_l *= param->timing.pix_clk_100hz / 10;
+		m_vid_l = div_u64(m_vid_l,
+			param->link_settings.link_rate
+				* LINK_RATE_REF_FREQ_IN_KHZ);
+
+		m_vid = (uint32_t) m_vid_l;
+
+		/* enable auto measurement */
+
+		REG_UPDATE(DP_VID_TIMING, DP_VID_M_N_GEN_EN, 0);
+
+		/* auto measurement need 1 full 0x8000 symbol cycle to kick in,
+		 * therefore program initial value for Mvid and Nvid
+		 */
+
+		REG_UPDATE(DP_VID_N, DP_VID_N, n_vid);
+
+		REG_UPDATE(DP_VID_M, DP_VID_M, m_vid);
+
+		REG_UPDATE_2(DP_VID_TIMING,
+				DP_VID_M_N_GEN_EN, 1,
+				DP_VID_N_MUL, n_multiply);
+
+		REG_UPDATE(DP_PIXEL_FORMAT,
+				DP_PIXEL_PER_CYCLE_PROCESSING_MODE,
+				pix_per_cycle);
+	}
+
+	/* make sure stream is disabled before resetting steer fifo */
+	REG_UPDATE(DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, false);
+	REG_WAIT(DP_VID_STREAM_CNTL, DP_VID_STREAM_STATUS, 0, 10, 5000);
+
+	/* DIG_START is removed from the register spec */
+
+	/* switch DP encoder to CRTC data, but reset it the fifo first. It may happen
+	 * that it overflows during mode transition, and sometimes doesn't recover.
+	 */
+	REG_UPDATE(DP_STEER_FIFO, DP_STEER_FIFO_RESET, 1);
+	udelay(10);
+
+	REG_UPDATE(DP_STEER_FIFO, DP_STEER_FIFO_RESET, 0);
+
+	/* wait 100us for DIG/DP logic to prime
+	 * (i.e. a few video lines)
+	 */
+	udelay(100);
+
+	/* the hardware would start sending video at the start of the next DP
+	 * frame (i.e. rising edge of the vblank).
+	 * NOTE: We used to program DP_VID_STREAM_DIS_DEFER = 2 here, but this
+	 * register has no effect on enable transition! HW always guarantees
+	 * VID_STREAM enable at start of next frame, and this is not
+	 * programmable
+	 */
+
+	REG_UPDATE(DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, true);
+
+	/*
+	 * DIG Resync FIFO now needs to be explicitly enabled.
+	 * This should come after DP_VID_STREAM_ENABLE per HW docs.
+	 */
+	enc314_enable_fifo(enc);
+
+	link->dc->link_srv->dp_trace_source_sequence(link, DPCD_SOURCE_SEQ_AFTER_ENABLE_DP_VID_STREAM);
+}
+
+/* Set DSC-related configuration.
+ *   dsc_mode: 0 disables DSC, other values enable DSC in specified format
+ *   sc_bytes_per_pixel: DP_DSC_BYTES_PER_PIXEL removed in DCN32
+ *   dsc_slice_width: DP_DSC_SLICE_WIDTH removed in DCN32
+ */
+void enc314_dp_set_dsc_config(struct stream_encoder *enc,
+					enum optc_dsc_mode dsc_mode,
+					uint32_t dsc_bytes_per_pixel,
+					uint32_t dsc_slice_width)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	REG_UPDATE(DP_DSC_CNTL,	DP_DSC_MODE, dsc_mode == OPTC_DSC_DISABLED ? 0 : 1);
+}
+
+/* this function read dsc related register fields to be logged later in dcn10_log_hw_state
+ * into a dcn_dsc_state struct.
+ */
+void enc314_read_state(struct stream_encoder *enc, struct enc_state *s)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	//if dsc is enabled, continue to read
+	REG_GET(DP_DSC_CNTL, DP_DSC_MODE, &s->dsc_mode);
+	if (s->dsc_mode) {
+		REG_GET(DP_GSP11_CNTL, DP_SEC_GSP11_LINE_NUM, &s->sec_gsp_pps_line_num);
+
+		REG_GET(DP_MSA_VBID_MISC, DP_VBID6_LINE_REFERENCE, &s->vbid6_line_reference);
+		REG_GET(DP_MSA_VBID_MISC, DP_VBID6_LINE_NUM, &s->vbid6_line_num);
+
+		REG_GET(DP_GSP11_CNTL, DP_SEC_GSP11_ENABLE, &s->sec_gsp_pps_enable);
+		REG_GET(DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, &s->sec_stream_enable);
+	}
+}
+
+void enc314_set_dig_input_mode(struct stream_encoder *enc, unsigned int pix_per_container)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	// The naming of this field is confusing, what it means is the output mode of otg, which
+	// is the input mode of the dig
+	REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_OUTPUT_PIXEL_MODE, pix_per_container == 2 ? 0x1 : 0x0);
+}
+
+
+static const struct stream_encoder_funcs dcn314_str_enc_funcs = {
+	.dp_set_odm_combine =
+		enc314_dp_set_odm_combine,
+	.dp_set_stream_attribute =
+		enc2_stream_encoder_dp_set_stream_attribute,
+	.hdmi_set_stream_attribute =
+		enc314_stream_encoder_hdmi_set_stream_attribute,
+	.dvi_set_stream_attribute =
+		enc314_stream_encoder_dvi_set_stream_attribute,
+	.set_throttled_vcp_size =
+		enc1_stream_encoder_set_throttled_vcp_size,
+	.update_hdmi_info_packets =
+		enc3_stream_encoder_update_hdmi_info_packets,
+	.stop_hdmi_info_packets =
+		enc3_stream_encoder_stop_hdmi_info_packets,
+	.update_dp_info_packets_sdp_line_num =
+		enc3_stream_encoder_update_dp_info_packets_sdp_line_num,
+	.update_dp_info_packets =
+		enc3_stream_encoder_update_dp_info_packets,
+	.stop_dp_info_packets =
+		enc1_stream_encoder_stop_dp_info_packets,
+	.dp_blank =
+		enc314_stream_encoder_dp_blank,
+	.dp_unblank =
+		enc314_stream_encoder_dp_unblank,
+	.audio_mute_control = enc3_audio_mute_control,
+
+	.dp_audio_setup = enc3_se_dp_audio_setup,
+	.dp_audio_enable = enc3_se_dp_audio_enable,
+	.dp_audio_disable = enc1_se_dp_audio_disable,
+
+	.hdmi_audio_setup = enc3_se_hdmi_audio_setup,
+	.hdmi_audio_disable = enc1_se_hdmi_audio_disable,
+	.setup_stereo_sync  = enc1_setup_stereo_sync,
+	.set_avmute = enc1_stream_encoder_set_avmute,
+	.dig_connect_to_otg = enc1_dig_connect_to_otg,
+	.dig_source_otg = enc1_dig_source_otg,
+
+	.dp_get_pixel_format  = enc1_stream_encoder_dp_get_pixel_format,
+
+	.enc_read_state = enc314_read_state,
+	.dp_set_dsc_config = enc314_dp_set_dsc_config,
+	.dp_set_dsc_pps_info_packet = enc3_dp_set_dsc_pps_info_packet,
+	.set_dynamic_metadata = enc2_set_dynamic_metadata,
+	.hdmi_reset_stream_attribute = enc1_reset_hdmi_stream_attribute,
+
+	.enable_fifo = enc314_enable_fifo,
+	.disable_fifo = enc314_disable_fifo,
+	.is_fifo_enabled = enc314_is_fifo_enabled,
+	.set_input_mode = enc314_set_dig_input_mode,
+};
+
+void dcn314_dio_stream_encoder_construct(
+	struct dcn10_stream_encoder *enc1,
+	struct dc_context *ctx,
+	struct dc_bios *bp,
+	enum engine_id eng_id,
+	struct vpg *vpg,
+	struct afmt *afmt,
+	const struct dcn10_stream_enc_registers *regs,
+	const struct dcn10_stream_encoder_shift *se_shift,
+	const struct dcn10_stream_encoder_mask *se_mask)
+{
+	enc1->base.funcs = &dcn314_str_enc_funcs;
+	enc1->base.ctx = ctx;
+	enc1->base.id = eng_id;
+	enc1->base.bp = bp;
+	enc1->base.vpg = vpg;
+	enc1->base.afmt = afmt;
+	enc1->regs = regs;
+	enc1->se_shift = se_shift;
+	enc1->se_mask = se_mask;
+	enc1->base.stream_enc_inst = vpg->inst;
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn314/dcn314_dio_stream_encoder.h b/drivers/gpu/drm/amd/display/dc/dio/dcn314/dcn314_dio_stream_encoder.h
new file mode 100644
index 000000000000..86548be591be
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn314/dcn314_dio_stream_encoder.h
@@ -0,0 +1,355 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_DIO_STREAM_ENCODER_DCN314_H__
+#define __DC_DIO_STREAM_ENCODER_DCN314_H__
+
+#include "dcn30/dcn30_vpg.h"
+#include "dcn30/dcn30_afmt.h"
+#include "stream_encoder.h"
+#include "dcn20/dcn20_stream_encoder.h"
+
+/* Register bit field name change */
+#define RDPCSTX0_RDPCSTX_CLOCK_CNTL__RDPCS_SYMCLK_DIV2_GATE_DIS__SHIFT                                        0x8
+#define RDPCSTX0_RDPCSTX_CLOCK_CNTL__RDPCS_SYMCLK_DIV2_EN__SHIFT                                              0x9
+#define RDPCSTX0_RDPCSTX_CLOCK_CNTL__RDPCS_SYMCLK_DIV2_CLOCK_ON__SHIFT                                        0xa
+#define DPCSTX0_DPCSTX_TX_CNTL__DPCS_TX_DATA_SWAP__SHIFT                                                      0xe
+#define DPCSTX0_DPCSTX_TX_CNTL__DPCS_TX_DATA_ORDER_INVERT__SHIFT                                              0xf
+
+#define RDPCSTX0_RDPCSTX_CLOCK_CNTL__RDPCS_SYMCLK_DIV2_GATE_DIS_MASK                                          0x00000100L
+#define RDPCSTX0_RDPCSTX_CLOCK_CNTL__RDPCS_SYMCLK_DIV2_EN_MASK                                                0x00000200L
+#define RDPCSTX0_RDPCSTX_CLOCK_CNTL__RDPCS_SYMCLK_DIV2_CLOCK_ON_MASK                                          0x00000400L
+#define DPCSTX0_DPCSTX_TX_CNTL__DPCS_TX_DATA_SWAP_MASK                                                        0x00004000L
+#define DPCSTX0_DPCSTX_TX_CNTL__DPCS_TX_DATA_ORDER_INVERT_MASK                                                0x00008000L
+
+
+#define SE_DCN314_REG_LIST(id)\
+	SRI(AFMT_CNTL, DIG, id), \
+	SRI(DIG_FE_CNTL, DIG, id), \
+	SRI(HDMI_CONTROL, DIG, id), \
+	SRI(HDMI_DB_CONTROL, DIG, id), \
+	SRI(HDMI_GC, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL0, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL1, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL2, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL3, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL4, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL5, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL6, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL7, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL8, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL9, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL10, DIG, id), \
+	SRI(HDMI_INFOFRAME_CONTROL0, DIG, id), \
+	SRI(HDMI_INFOFRAME_CONTROL1, DIG, id), \
+	SRI(HDMI_VBI_PACKET_CONTROL, DIG, id), \
+	SRI(HDMI_AUDIO_PACKET_CONTROL, DIG, id),\
+	SRI(HDMI_ACR_PACKET_CONTROL, DIG, id),\
+	SRI(HDMI_ACR_32_0, DIG, id),\
+	SRI(HDMI_ACR_32_1, DIG, id),\
+	SRI(HDMI_ACR_44_0, DIG, id),\
+	SRI(HDMI_ACR_44_1, DIG, id),\
+	SRI(HDMI_ACR_48_0, DIG, id),\
+	SRI(HDMI_ACR_48_1, DIG, id),\
+	SRI(DP_DB_CNTL, DP, id), \
+	SRI(DP_MSA_MISC, DP, id), \
+	SRI(DP_MSA_VBID_MISC, DP, id), \
+	SRI(DP_MSA_COLORIMETRY, DP, id), \
+	SRI(DP_MSA_TIMING_PARAM1, DP, id), \
+	SRI(DP_MSA_TIMING_PARAM2, DP, id), \
+	SRI(DP_MSA_TIMING_PARAM3, DP, id), \
+	SRI(DP_MSA_TIMING_PARAM4, DP, id), \
+	SRI(DP_MSE_RATE_CNTL, DP, id), \
+	SRI(DP_MSE_RATE_UPDATE, DP, id), \
+	SRI(DP_PIXEL_FORMAT, DP, id), \
+	SRI(DP_SEC_CNTL, DP, id), \
+	SRI(DP_SEC_CNTL1, DP, id), \
+	SRI(DP_SEC_CNTL2, DP, id), \
+	SRI(DP_SEC_CNTL5, DP, id), \
+	SRI(DP_SEC_CNTL6, DP, id), \
+	SRI(DP_STEER_FIFO, DP, id), \
+	SRI(DP_VID_M, DP, id), \
+	SRI(DP_VID_N, DP, id), \
+	SRI(DP_VID_STREAM_CNTL, DP, id), \
+	SRI(DP_VID_TIMING, DP, id), \
+	SRI(DP_SEC_AUD_N, DP, id), \
+	SRI(DP_SEC_TIMESTAMP, DP, id), \
+	SRI(DP_DSC_CNTL, DP, id), \
+	SRI(DP_SEC_METADATA_TRANSMISSION, DP, id), \
+	SRI(HDMI_METADATA_PACKET_CONTROL, DIG, id), \
+	SRI(DP_SEC_FRAMING4, DP, id), \
+	SRI(DP_GSP11_CNTL, DP, id), \
+	SRI(DME_CONTROL, DME, id),\
+	SRI(DP_SEC_METADATA_TRANSMISSION, DP, id), \
+	SRI(HDMI_METADATA_PACKET_CONTROL, DIG, id), \
+	SRI(DIG_FE_CNTL, DIG, id), \
+	SRI(DIG_CLOCK_PATTERN, DIG, id), \
+	SRI(DIG_FIFO_CTRL0, DIG, id)
+
+
+#define SE_COMMON_MASK_SH_LIST_DCN314(mask_sh)\
+	SE_SF(DP0_DP_PIXEL_FORMAT, DP_PIXEL_ENCODING, mask_sh),\
+	SE_SF(DP0_DP_PIXEL_FORMAT, DP_COMPONENT_DEPTH, mask_sh),\
+	SE_SF(DP0_DP_PIXEL_FORMAT, DP_PIXEL_PER_CYCLE_PROCESSING_MODE, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_PACKET_GEN_VERSION, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_KEEPOUT_MODE, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_DATA_SCRAMBLE_EN, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_NO_EXTRA_NULL_PACKET_FILLED, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_GC_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_GC_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_NULL_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_ACP_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_INFOFRAME_CONTROL1, HDMI_AUDIO_INFO_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GC, HDMI_GC_AVMUTE, mask_sh),\
+	SE_SF(DP0_DP_MSE_RATE_CNTL, DP_MSE_RATE_X, mask_sh),\
+	SE_SF(DP0_DP_MSE_RATE_CNTL, DP_MSE_RATE_Y, mask_sh),\
+	SE_SF(DP0_DP_MSE_RATE_UPDATE, DP_MSE_RATE_UPDATE_PENDING, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP0_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP1_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP2_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP3_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_MPG_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL1, DP_SEC_GSP5_LINE_REFERENCE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP4_SEND, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP4_SEND_PENDING, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL4, DP_SEC_GSP4_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL5, DP_SEC_GSP5_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP4_SEND_ANY_LINE, mask_sh),\
+	SE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_DIS_DEFER, mask_sh),\
+	SE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_STATUS, mask_sh),\
+	SE_SF(DP0_DP_STEER_FIFO, DP_STEER_FIFO_RESET, mask_sh),\
+	SE_SF(DP0_DP_VID_TIMING, DP_VID_M_N_GEN_EN, mask_sh),\
+	SE_SF(DP0_DP_VID_N, DP_VID_N, mask_sh),\
+	SE_SF(DP0_DP_VID_M, DP_VID_M, mask_sh),\
+	SE_SF(DIG0_HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_DELAY_EN, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUTO_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_PACKET_CONTROL, HDMI_ACR_SOURCE, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUDIO_PRIORITY, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_32_0, HDMI_ACR_CTS_32, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_32_1, HDMI_ACR_N_32, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_44_0, HDMI_ACR_CTS_44, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_44_1, HDMI_ACR_N_44, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_48_0, HDMI_ACR_CTS_48, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_48_1, HDMI_ACR_N_48, mask_sh),\
+	SE_SF(DP0_DP_SEC_AUD_N, DP_SEC_AUD_N, mask_sh),\
+	SE_SF(DP0_DP_SEC_TIMESTAMP, DP_SEC_TIMESTAMP_MODE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_ASP_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_ATP_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_AIP_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_ACM_ENABLE, mask_sh),\
+	SE_SF(DIG0_AFMT_CNTL, AFMT_AUDIO_CLOCK_EN, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_CLOCK_CHANNEL_RATE, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, TMDS_PIXEL_ENCODING, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, TMDS_COLOR_FORMAT, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_STEREOSYNC_SELECT, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_STEREOSYNC_GATE_EN, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP4_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP5_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP6_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP7_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP7_SEND, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL6, DP_SEC_GSP7_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP11_PPS, mask_sh),\
+	SE_SF(DP0_DP_GSP11_CNTL, DP_SEC_GSP11_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_GSP11_CNTL, DP_SEC_GSP11_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_DB_CNTL, DP_DB_DISABLE, mask_sh),\
+	SE_SF(DP0_DP_MSA_COLORIMETRY, DP_MSA_MISC0, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM1, DP_MSA_HTOTAL, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM1, DP_MSA_VTOTAL, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM2, DP_MSA_HSTART, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM2, DP_MSA_VSTART, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_HSYNCWIDTH, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_HSYNCPOLARITY, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_VSYNCWIDTH, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_VSYNCPOLARITY, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM4, DP_MSA_HWIDTH, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM4, DP_MSA_VHEIGHT, mask_sh),\
+	SE_SF(DIG0_HDMI_DB_CONTROL, HDMI_DB_DISABLE, mask_sh),\
+	SE_SF(DP0_DP_VID_TIMING, DP_VID_N_MUL, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_SOURCE_SELECT, mask_sh), \
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC0_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC0_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC1_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC1_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC2_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC2_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC3_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC3_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC4_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC4_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC5_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC5_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC6_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC6_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC7_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC7_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC8_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC8_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC9_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC9_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC10_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC10_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC11_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC11_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC12_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC12_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC13_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC13_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC14_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC14_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL1, HDMI_GENERIC0_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL1, HDMI_GENERIC1_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL2, HDMI_GENERIC2_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL2, HDMI_GENERIC3_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL3, HDMI_GENERIC4_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL3, HDMI_GENERIC5_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL4, HDMI_GENERIC6_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL4, HDMI_GENERIC7_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL7, HDMI_GENERIC8_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL7, HDMI_GENERIC9_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL8, HDMI_GENERIC10_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL8, HDMI_GENERIC11_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL9, HDMI_GENERIC12_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL9, HDMI_GENERIC13_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL10, HDMI_GENERIC14_LINE, mask_sh),\
+	SE_SF(DP0_DP_DSC_CNTL, DP_DSC_MODE, mask_sh),\
+	SE_SF(DP0_DP_MSA_VBID_MISC, DP_VBID6_LINE_REFERENCE, mask_sh),\
+	SE_SF(DP0_DP_MSA_VBID_MISC, DP_VBID6_LINE_NUM, mask_sh),\
+	SE_SF(DME0_DME_CONTROL, METADATA_ENGINE_EN, mask_sh),\
+	SE_SF(DME0_DME_CONTROL, METADATA_HUBP_REQUESTOR_ID, mask_sh),\
+	SE_SF(DME0_DME_CONTROL, METADATA_STREAM_TYPE, mask_sh),\
+	SE_SF(DP0_DP_SEC_METADATA_TRANSMISSION, DP_SEC_METADATA_PACKET_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_METADATA_TRANSMISSION, DP_SEC_METADATA_PACKET_LINE_REFERENCE, mask_sh),\
+	SE_SF(DP0_DP_SEC_METADATA_TRANSMISSION, DP_SEC_METADATA_PACKET_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_METADATA_PACKET_CONTROL, HDMI_METADATA_PACKET_ENABLE, mask_sh),\
+	SE_SF(DIG0_HDMI_METADATA_PACKET_CONTROL, HDMI_METADATA_PACKET_LINE_REFERENCE, mask_sh),\
+	SE_SF(DIG0_HDMI_METADATA_PACKET_CONTROL, HDMI_METADATA_PACKET_LINE, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DOLBY_VISION_EN, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_SYMCLK_FE_ON, mask_sh),\
+	SE_SF(DP0_DP_SEC_FRAMING4, DP_SST_SDP_SPLITTING, mask_sh),\
+	SE_SF(DIG0_DIG_CLOCK_PATTERN, DIG_CLOCK_PATTERN, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_CTRL0, DIG_FIFO_OUTPUT_PIXEL_MODE, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_CTRL0, DIG_FIFO_READ_START_LEVEL, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_CTRL0, DIG_FIFO_ENABLE, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_CTRL0, DIG_FIFO_RESET, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_CTRL0, DIG_FIFO_RESET_DONE, mask_sh)
+
+void dcn314_dio_stream_encoder_construct(
+	struct dcn10_stream_encoder *enc1,
+	struct dc_context *ctx,
+	struct dc_bios *bp,
+	enum engine_id eng_id,
+	struct vpg *vpg,
+	struct afmt *afmt,
+	const struct dcn10_stream_enc_registers *regs,
+	const struct dcn10_stream_encoder_shift *se_shift,
+	const struct dcn10_stream_encoder_mask *se_mask);
+
+void enc3_stream_encoder_update_hdmi_info_packets(
+	struct stream_encoder *enc,
+	const struct encoder_info_frame *info_frame);
+
+void enc3_stream_encoder_stop_hdmi_info_packets(
+	struct stream_encoder *enc);
+
+void enc3_stream_encoder_update_dp_info_packets_sdp_line_num(
+		struct stream_encoder *enc,
+		struct encoder_info_frame *info_frame);
+
+void enc3_stream_encoder_update_dp_info_packets(
+	struct stream_encoder *enc,
+	const struct encoder_info_frame *info_frame);
+
+void enc3_audio_mute_control(
+	struct stream_encoder *enc,
+	bool mute);
+
+void enc3_se_dp_audio_setup(
+	struct stream_encoder *enc,
+	unsigned int az_inst,
+	struct audio_info *info);
+
+void enc3_se_dp_audio_enable(
+	struct stream_encoder *enc);
+
+void enc3_se_hdmi_audio_setup(
+	struct stream_encoder *enc,
+	unsigned int az_inst,
+	struct audio_info *info,
+	struct audio_crtc_info *audio_crtc_info);
+
+void enc3_dp_set_dsc_pps_info_packet(
+	struct stream_encoder *enc,
+	bool enable,
+	uint8_t *dsc_packed_pps,
+	bool immediate_update);
+
+void enc314_stream_encoder_dvi_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	bool is_dual_link);
+
+void enc314_stream_encoder_hdmi_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	int actual_pix_clk_khz,
+	bool enable_audio);
+
+void enc314_stream_encoder_dp_blank(
+	struct dc_link *link,
+	struct stream_encoder *enc);
+
+void enc314_stream_encoder_dp_unblank(
+		struct dc_link *link,
+		struct stream_encoder *enc,
+		const struct encoder_unblank_param *param);
+
+void enc314_reset_fifo(struct stream_encoder *enc, bool reset);
+
+void enc314_enable_fifo(struct stream_encoder *enc);
+
+void enc314_disable_fifo(struct stream_encoder *enc);
+
+void enc314_set_dig_input_mode(struct stream_encoder *enc, unsigned int pix_per_container);
+
+void enc314_read_state(struct stream_encoder *enc, struct enc_state *s);
+
+void enc314_dp_set_odm_combine(
+	struct stream_encoder *enc,
+	bool odm_combine);
+
+void enc314_dp_set_dsc_config(
+	struct stream_encoder *enc,
+	enum optc_dsc_mode dsc_mode,
+	uint32_t dsc_bytes_per_pixel,
+	uint32_t dsc_slice_width);
+
+#endif /* __DC_DIO_STREAM_ENCODER_DCN314_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn32/dcn32_dio_link_encoder.c b/drivers/gpu/drm/amd/display/dc/dio/dcn32/dcn32_dio_link_encoder.c
new file mode 100644
index 000000000000..06907e8a4eda
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn32/dcn32_dio_link_encoder.c
@@ -0,0 +1,328 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+#include "reg_helper.h"
+
+#include "core_types.h"
+#include "link_encoder.h"
+#include "dcn31/dcn31_dio_link_encoder.h"
+#include "dcn32_dio_link_encoder.h"
+#include "stream_encoder.h"
+#include "dc_bios_types.h"
+#include "link_enc_cfg.h"
+
+#include "dc_dmub_srv.h"
+#include "gpio_service_interface.h"
+
+#ifndef MIN
+#define MIN(X, Y) ((X) < (Y) ? (X) : (Y))
+#endif
+
+#define CTX \
+	enc10->base.ctx
+#define DC_LOGGER \
+	enc10->base.ctx->logger
+
+#define REG(reg)\
+	(enc10->link_regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	enc10->link_shift->field_name, enc10->link_mask->field_name
+
+#define AUX_REG(reg)\
+	(enc10->aux_regs->reg)
+
+#define AUX_REG_READ(reg_name) \
+		dm_read_reg(CTX, AUX_REG(reg_name))
+
+#define AUX_REG_WRITE(reg_name, val) \
+			dm_write_reg(CTX, AUX_REG(reg_name), val)
+
+static uint8_t phy_id_from_transmitter(enum transmitter t)
+{
+	uint8_t phy_id;
+
+	switch (t) {
+	case TRANSMITTER_UNIPHY_A:
+		phy_id = 0;
+		break;
+	case TRANSMITTER_UNIPHY_B:
+		phy_id = 1;
+		break;
+	case TRANSMITTER_UNIPHY_C:
+		phy_id = 2;
+		break;
+	case TRANSMITTER_UNIPHY_D:
+		phy_id = 3;
+		break;
+	case TRANSMITTER_UNIPHY_E:
+		phy_id = 4;
+		break;
+	case TRANSMITTER_UNIPHY_F:
+		phy_id = 5;
+		break;
+	case TRANSMITTER_UNIPHY_G:
+		phy_id = 6;
+		break;
+	default:
+		phy_id = 0;
+		break;
+	}
+	return phy_id;
+}
+
+void enc32_hw_init(struct link_encoder *enc)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+/*
+	00 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__1to2 : 1/2
+	01 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__3to4 : 3/4
+	02 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__7to8 : 7/8
+	03 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__15to16 : 15/16
+	04 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__31to32 : 31/32
+	05 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__63to64 : 63/64
+	06 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__127to128 : 127/128
+	07 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__255to256 : 255/256
+*/
+
+/*
+	AUX_REG_UPDATE_5(AUX_DPHY_RX_CONTROL0,
+	AUX_RX_START_WINDOW = 1 [6:4]
+	AUX_RX_RECEIVE_WINDOW = 1 default is 2 [10:8]
+	AUX_RX_HALF_SYM_DETECT_LEN  = 1 [13:12] default is 1
+	AUX_RX_TRANSITION_FILTER_EN = 1 [16] default is 1
+	AUX_RX_ALLOW_BELOW_THRESHOLD_PHASE_DETECT [17] is 0  default is 0
+	AUX_RX_ALLOW_BELOW_THRESHOLD_START [18] is 1  default is 1
+	AUX_RX_ALLOW_BELOW_THRESHOLD_STOP [19] is 1  default is 1
+	AUX_RX_PHASE_DETECT_LEN,  [21,20] = 0x3 default is 3
+	AUX_RX_DETECTION_THRESHOLD [30:28] = 1
+*/
+	AUX_REG_WRITE(AUX_DPHY_RX_CONTROL0, 0x103d1110);
+
+	AUX_REG_WRITE(AUX_DPHY_TX_CONTROL, 0x21c7a);
+
+	//AUX_DPHY_TX_REF_CONTROL'AUX_TX_REF_DIV HW default is 0x32;
+	// Set AUX_TX_REF_DIV Divider to generate 2 MHz reference from refclk
+	// 27MHz -> 0xd
+	// 100MHz -> 0x32
+	// 48MHz -> 0x18
+
+	// Set TMDS_CTL0 to 1.  This is a legacy setting.
+	REG_UPDATE(TMDS_CTL_BITS, TMDS_CTL0, 1);
+
+	dcn10_aux_initialize(enc10);
+}
+
+
+void dcn32_link_encoder_enable_dp_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source)
+{
+	if (!enc->ctx->dc->debug.avoid_vbios_exec_table) {
+		dcn10_link_encoder_enable_dp_output(enc, link_settings, clock_source);
+		return;
+	}
+}
+
+static bool query_dp_alt_from_dmub(struct link_encoder *enc,
+	union dmub_rb_cmd *cmd)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	memset(cmd, 0, sizeof(*cmd));
+	cmd->query_dp_alt.header.type = DMUB_CMD__VBIOS;
+	cmd->query_dp_alt.header.sub_type =
+		DMUB_CMD__VBIOS_TRANSMITTER_QUERY_DP_ALT;
+	cmd->query_dp_alt.header.payload_bytes = sizeof(cmd->query_dp_alt.data);
+	cmd->query_dp_alt.data.phy_id = phy_id_from_transmitter(enc10->base.transmitter);
+
+	if (!dc_wake_and_execute_dmub_cmd(enc->ctx, cmd, DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY))
+		return false;
+
+	return true;
+}
+
+bool dcn32_link_encoder_is_in_alt_mode(struct link_encoder *enc)
+{
+	union dmub_rb_cmd cmd;
+
+	if (!query_dp_alt_from_dmub(enc, &cmd))
+		return false;
+
+	return (cmd.query_dp_alt.data.is_dp_alt_disable == 0);
+}
+
+void dcn32_link_encoder_get_max_link_cap(struct link_encoder *enc,
+	struct dc_link_settings *link_settings)
+{
+	union dmub_rb_cmd cmd;
+
+	dcn10_link_encoder_get_max_link_cap(enc, link_settings);
+
+	if (!query_dp_alt_from_dmub(enc, &cmd))
+		return;
+
+	if (cmd.query_dp_alt.data.is_usb &&
+			cmd.query_dp_alt.data.is_dp4 == 0)
+		link_settings->lane_count = MIN(LANE_COUNT_TWO, link_settings->lane_count);
+}
+
+
+static const struct link_encoder_funcs dcn32_link_enc_funcs = {
+	.read_state = link_enc2_read_state,
+	.validate_output_with_stream =
+			dcn30_link_encoder_validate_output_with_stream,
+	.hw_init = enc32_hw_init,
+	.setup = dcn10_link_encoder_setup,
+	.enable_tmds_output = dcn10_link_encoder_enable_tmds_output,
+	.enable_dp_output = dcn32_link_encoder_enable_dp_output,
+	.enable_dp_mst_output = dcn10_link_encoder_enable_dp_mst_output,
+	.disable_output = dcn10_link_encoder_disable_output,
+	.dp_set_lane_settings = dcn10_link_encoder_dp_set_lane_settings,
+	.dp_set_phy_pattern = dcn10_link_encoder_dp_set_phy_pattern,
+	.update_mst_stream_allocation_table =
+		dcn10_link_encoder_update_mst_stream_allocation_table,
+	.psr_program_dp_dphy_fast_training =
+			dcn10_psr_program_dp_dphy_fast_training,
+	.psr_program_secondary_packet = dcn10_psr_program_secondary_packet,
+	.connect_dig_be_to_fe = dcn10_link_encoder_connect_dig_be_to_fe,
+	.enable_hpd = dcn10_link_encoder_enable_hpd,
+	.disable_hpd = dcn10_link_encoder_disable_hpd,
+	.is_dig_enabled = dcn10_is_dig_enabled,
+	.destroy = dcn10_link_encoder_destroy,
+	.fec_set_enable = enc2_fec_set_enable,
+	.fec_set_ready = enc2_fec_set_ready,
+	.fec_is_active = enc2_fec_is_active,
+	.get_dig_frontend = dcn10_get_dig_frontend,
+	.get_dig_mode = dcn10_get_dig_mode,
+	.is_in_alt_mode = dcn32_link_encoder_is_in_alt_mode,
+	.get_max_link_cap = dcn32_link_encoder_get_max_link_cap,
+	.set_dio_phy_mux = dcn31_link_encoder_set_dio_phy_mux,
+};
+
+void dcn32_link_encoder_construct(
+	struct dcn20_link_encoder *enc20,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	const struct dcn10_link_enc_aux_registers *aux_regs,
+	const struct dcn10_link_enc_hpd_registers *hpd_regs,
+	const struct dcn10_link_enc_shift *link_shift,
+	const struct dcn10_link_enc_mask *link_mask)
+{
+	struct bp_connector_speed_cap_info bp_cap_info = {0};
+	const struct dc_vbios_funcs *bp_funcs = init_data->ctx->dc_bios->funcs;
+	enum bp_result result = BP_RESULT_OK;
+	struct dcn10_link_encoder *enc10 = &enc20->enc10;
+
+	enc10->base.funcs = &dcn32_link_enc_funcs;
+	enc10->base.ctx = init_data->ctx;
+	enc10->base.id = init_data->encoder;
+
+	enc10->base.hpd_source = init_data->hpd_source;
+	enc10->base.connector = init_data->connector;
+
+	enc10->base.preferred_engine = ENGINE_ID_UNKNOWN;
+
+	enc10->base.features = *enc_features;
+
+	enc10->base.transmitter = init_data->transmitter;
+
+	/* set the flag to indicate whether driver poll the I2C data pin
+	 * while doing the DP sink detect
+	 */
+
+/*	if (dal_adapter_service_is_feature_supported(as,
+		FEATURE_DP_SINK_DETECT_POLL_DATA_PIN))
+		enc10->base.features.flags.bits.
+			DP_SINK_DETECT_POLL_DATA_PIN = true;*/
+
+	enc10->base.output_signals =
+		SIGNAL_TYPE_DVI_SINGLE_LINK |
+		SIGNAL_TYPE_DVI_DUAL_LINK |
+		SIGNAL_TYPE_LVDS |
+		SIGNAL_TYPE_DISPLAY_PORT |
+		SIGNAL_TYPE_DISPLAY_PORT_MST |
+		SIGNAL_TYPE_EDP |
+		SIGNAL_TYPE_HDMI_TYPE_A;
+
+	enc10->link_regs = link_regs;
+	enc10->aux_regs = aux_regs;
+	enc10->hpd_regs = hpd_regs;
+	enc10->link_shift = link_shift;
+	enc10->link_mask = link_mask;
+
+	switch (enc10->base.transmitter) {
+	case TRANSMITTER_UNIPHY_A:
+		enc10->base.preferred_engine = ENGINE_ID_DIGA;
+	break;
+	case TRANSMITTER_UNIPHY_B:
+		enc10->base.preferred_engine = ENGINE_ID_DIGB;
+	break;
+	case TRANSMITTER_UNIPHY_C:
+		enc10->base.preferred_engine = ENGINE_ID_DIGC;
+	break;
+	case TRANSMITTER_UNIPHY_D:
+		enc10->base.preferred_engine = ENGINE_ID_DIGD;
+	break;
+	case TRANSMITTER_UNIPHY_E:
+		enc10->base.preferred_engine = ENGINE_ID_DIGE;
+	break;
+	default:
+		ASSERT_CRITICAL(false);
+		enc10->base.preferred_engine = ENGINE_ID_UNKNOWN;
+	}
+
+	/* default to one to mirror Windows behavior */
+	enc10->base.features.flags.bits.HDMI_6GB_EN = 1;
+
+	if (bp_funcs->get_connector_speed_cap_info)
+		result = bp_funcs->get_connector_speed_cap_info(enc10->base.ctx->dc_bios,
+						enc10->base.connector, &bp_cap_info);
+
+	/* Override features with DCE-specific values */
+	if (result == BP_RESULT_OK) {
+		enc10->base.features.flags.bits.IS_HBR2_CAPABLE =
+				bp_cap_info.DP_HBR2_EN;
+		enc10->base.features.flags.bits.IS_HBR3_CAPABLE =
+				bp_cap_info.DP_HBR3_EN;
+		enc10->base.features.flags.bits.HDMI_6GB_EN = bp_cap_info.HDMI_6GB_EN;
+		enc10->base.features.flags.bits.IS_DP2_CAPABLE = 1;
+		enc10->base.features.flags.bits.IS_UHBR10_CAPABLE = bp_cap_info.DP_UHBR10_EN;
+		enc10->base.features.flags.bits.IS_UHBR13_5_CAPABLE = bp_cap_info.DP_UHBR13_5_EN;
+		enc10->base.features.flags.bits.IS_UHBR20_CAPABLE = bp_cap_info.DP_UHBR20_EN;
+	} else {
+		DC_LOG_WARNING("%s: Failed to get encoder_cap_info from VBIOS with error code %d!\n",
+				__func__,
+				result);
+	}
+	if (enc10->base.ctx->dc->debug.hdmi20_disable) {
+		enc10->base.features.flags.bits.HDMI_6GB_EN = 0;
+	}
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn32/dcn32_dio_link_encoder.h b/drivers/gpu/drm/amd/display/dc/dio/dcn32/dcn32_dio_link_encoder.h
new file mode 100644
index 000000000000..35d23d9db45e
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn32/dcn32_dio_link_encoder.h
@@ -0,0 +1,53 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_LINK_ENCODER__DCN32_H__
+#define __DC_LINK_ENCODER__DCN32_H__
+
+#include "dcn30/dcn30_dio_link_encoder.h"
+
+void dcn32_link_encoder_construct(
+	struct dcn20_link_encoder *enc20,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	const struct dcn10_link_enc_aux_registers *aux_regs,
+	const struct dcn10_link_enc_hpd_registers *hpd_regs,
+	const struct dcn10_link_enc_shift *link_shift,
+	const struct dcn10_link_enc_mask *link_mask);
+
+void enc32_hw_init(struct link_encoder *enc);
+
+void dcn32_link_encoder_enable_dp_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source);
+
+bool dcn32_link_encoder_is_in_alt_mode(struct link_encoder *enc);
+
+void dcn32_link_encoder_get_max_link_cap(struct link_encoder *enc,
+	struct dc_link_settings *link_settings);
+
+#endif /* __DC_LINK_ENCODER__DCN32_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn32/dcn32_dio_stream_encoder.c b/drivers/gpu/drm/amd/display/dc/dio/dcn32/dcn32_dio_stream_encoder.c
new file mode 100644
index 000000000000..3523d1cdc1a3
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn32/dcn32_dio_stream_encoder.c
@@ -0,0 +1,493 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+#include "dc_bios_types.h"
+#include "dcn30/dcn30_dio_stream_encoder.h"
+#include "dcn32_dio_stream_encoder.h"
+#include "reg_helper.h"
+#include "hw_shared.h"
+#include "link_service.h"
+#include "dpcd_defs.h"
+
+#define DC_LOGGER \
+		enc1->base.ctx->logger
+
+#define REG(reg)\
+	(enc1->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	enc1->se_shift->field_name, enc1->se_mask->field_name
+
+#define VBI_LINE_0 0
+#define HDMI_CLOCK_CHANNEL_RATE_MORE_340M 340000
+
+#define CTX \
+	enc1->base.ctx
+
+
+
+static void enc32_dp_set_odm_combine(
+	struct stream_encoder *enc,
+	bool two_pixel_per_cyle)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	REG_UPDATE(DP_PIXEL_FORMAT, DP_PIXEL_PER_CYCLE_PROCESSING_MODE, two_pixel_per_cyle ? 1 : 0);
+}
+
+/* setup stream encoder in dvi mode */
+static void enc32_stream_encoder_dvi_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	bool is_dual_link)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (!enc->ctx->dc->debug.avoid_vbios_exec_table) {
+		struct bp_encoder_control cntl = {0};
+
+		cntl.action = ENCODER_CONTROL_SETUP;
+		cntl.engine_id = enc1->base.id;
+		cntl.signal = is_dual_link ?
+			SIGNAL_TYPE_DVI_DUAL_LINK : SIGNAL_TYPE_DVI_SINGLE_LINK;
+		cntl.enable_dp_audio = false;
+		cntl.pixel_clock = crtc_timing->pix_clk_100hz / 10;
+		cntl.lanes_number = (is_dual_link) ? LANE_COUNT_EIGHT : LANE_COUNT_FOUR;
+
+		if (enc1->base.bp->funcs->encoder_control(
+				enc1->base.bp, &cntl) != BP_RESULT_OK)
+			return;
+
+	} else {
+
+		//Set pattern for clock channel, default vlue 0x63 does not work
+		REG_UPDATE(DIG_CLOCK_PATTERN, DIG_CLOCK_PATTERN, 0x1F);
+
+		//DIG_BE_TMDS_DVI_MODE : TMDS-DVI mode is already set in link_encoder_setup
+
+		//DIG_SOURCE_SELECT is already set in dig_connect_to_otg
+
+		/* DIG_START is removed from the register spec */
+	}
+
+	ASSERT(crtc_timing->pixel_encoding == PIXEL_ENCODING_RGB);
+	ASSERT(crtc_timing->display_color_depth == COLOR_DEPTH_888);
+	enc1_stream_encoder_set_stream_attribute_helper(enc1, crtc_timing);
+}
+
+/* setup stream encoder in hdmi mode */
+static void enc32_stream_encoder_hdmi_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	int actual_pix_clk_khz,
+	bool enable_audio)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (!enc->ctx->dc->debug.avoid_vbios_exec_table) {
+		struct bp_encoder_control cntl = {0};
+
+		cntl.action = ENCODER_CONTROL_SETUP;
+		cntl.engine_id = enc1->base.id;
+		cntl.signal = SIGNAL_TYPE_HDMI_TYPE_A;
+		cntl.enable_dp_audio = enable_audio;
+		cntl.pixel_clock = actual_pix_clk_khz;
+		cntl.lanes_number = LANE_COUNT_FOUR;
+
+		if (enc1->base.bp->funcs->encoder_control(
+				enc1->base.bp, &cntl) != BP_RESULT_OK)
+			return;
+
+	} else {
+
+		//Set pattern for clock channel, default vlue 0x63 does not work
+		REG_UPDATE(DIG_CLOCK_PATTERN, DIG_CLOCK_PATTERN, 0x1F);
+
+		//DIG_BE_TMDS_HDMI_MODE : TMDS-HDMI mode is already set in link_encoder_setup
+
+		//DIG_SOURCE_SELECT is already set in dig_connect_to_otg
+
+		/* DIG_START is removed from the register spec */
+	}
+
+	/* Configure pixel encoding */
+	enc1_stream_encoder_set_stream_attribute_helper(enc1, crtc_timing);
+
+	/* setup HDMI engine */
+	REG_UPDATE_6(HDMI_CONTROL,
+		HDMI_PACKET_GEN_VERSION, 1,
+		HDMI_KEEPOUT_MODE, 1,
+		HDMI_DEEP_COLOR_ENABLE, 0,
+		HDMI_DATA_SCRAMBLE_EN, 0,
+		HDMI_NO_EXTRA_NULL_PACKET_FILLED, 1,
+		HDMI_CLOCK_CHANNEL_RATE, 0);
+
+	/* Configure color depth */
+	switch (crtc_timing->display_color_depth) {
+	case COLOR_DEPTH_888:
+		REG_UPDATE(HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 0);
+		break;
+	case COLOR_DEPTH_101010:
+		if (crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR422) {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 1,
+					HDMI_DEEP_COLOR_ENABLE, 0);
+		} else {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 1,
+					HDMI_DEEP_COLOR_ENABLE, 1);
+			}
+		break;
+	case COLOR_DEPTH_121212:
+		if (crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR422) {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 2,
+					HDMI_DEEP_COLOR_ENABLE, 0);
+		} else {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 2,
+					HDMI_DEEP_COLOR_ENABLE, 1);
+			}
+		break;
+	case COLOR_DEPTH_161616:
+		REG_UPDATE_2(HDMI_CONTROL,
+				HDMI_DEEP_COLOR_DEPTH, 3,
+				HDMI_DEEP_COLOR_ENABLE, 1);
+		break;
+	default:
+		break;
+	}
+
+	if (actual_pix_clk_khz >= HDMI_CLOCK_CHANNEL_RATE_MORE_340M) {
+		/* enable HDMI data scrambler
+		 * HDMI_CLOCK_CHANNEL_RATE_MORE_340M
+		 * Clock channel frequency is 1/4 of character rate.
+		 */
+		REG_UPDATE_2(HDMI_CONTROL,
+			HDMI_DATA_SCRAMBLE_EN, 1,
+			HDMI_CLOCK_CHANNEL_RATE, 1);
+	} else if (crtc_timing->flags.LTE_340MCSC_SCRAMBLE) {
+
+		/* TODO: New feature for DCE11, still need to implement */
+
+		/* enable HDMI data scrambler
+		 * HDMI_CLOCK_CHANNEL_FREQ_EQUAL_TO_CHAR_RATE
+		 * Clock channel frequency is the same
+		 * as character rate
+		 */
+		REG_UPDATE_2(HDMI_CONTROL,
+			HDMI_DATA_SCRAMBLE_EN, 1,
+			HDMI_CLOCK_CHANNEL_RATE, 0);
+	}
+
+
+	/* Enable transmission of General Control packet on every frame */
+	REG_UPDATE_3(HDMI_VBI_PACKET_CONTROL,
+		HDMI_GC_CONT, 1,
+		HDMI_GC_SEND, 1,
+		HDMI_NULL_SEND, 1);
+
+	/* Disable Audio Content Protection packet transmission */
+	REG_UPDATE(HDMI_VBI_PACKET_CONTROL, HDMI_ACP_SEND, 0);
+
+	/* following belongs to audio */
+	/* Enable Audio InfoFrame packet transmission. */
+	REG_UPDATE(HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, 1);
+
+	/* update double-buffered AUDIO_INFO registers immediately */
+	ASSERT(enc->afmt);
+	enc->afmt->funcs->audio_info_immediate_update(enc->afmt);
+
+	/* Select line number on which to send Audio InfoFrame packets */
+	REG_UPDATE(HDMI_INFOFRAME_CONTROL1, HDMI_AUDIO_INFO_LINE,
+				VBI_LINE_0 + 2);
+
+	/* set HDMI GC AVMUTE */
+	REG_UPDATE(HDMI_GC, HDMI_GC_AVMUTE, 0);
+}
+
+
+
+static bool is_two_pixels_per_containter(const struct dc_crtc_timing *timing)
+{
+	bool two_pix = timing->pixel_encoding == PIXEL_ENCODING_YCBCR420;
+
+	two_pix = two_pix || (timing->flags.DSC && timing->pixel_encoding == PIXEL_ENCODING_YCBCR422
+			&& !timing->dsc_cfg.ycbcr422_simple);
+	return two_pix;
+}
+
+void enc32_stream_encoder_dp_unblank(
+	struct dc_link *link,
+	struct stream_encoder *enc,
+	const struct encoder_unblank_param *param)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (param->link_settings.link_rate != LINK_RATE_UNKNOWN) {
+		uint32_t n_vid = 0x8000;
+		uint32_t m_vid;
+		uint32_t n_multiply = 0;
+		uint32_t pix_per_cycle = 0;
+		uint64_t m_vid_l = n_vid;
+
+		/* YCbCr 4:2:0 : Computed VID_M will be 2X the input rate */
+		if (is_two_pixels_per_containter(&param->timing) || param->opp_cnt > 1
+			|| param->pix_per_cycle > 1) {
+			/*this logic should be the same in get_pixel_clock_parameters() */
+			n_multiply = 1;
+			pix_per_cycle = 1;
+		}
+		/* M / N = Fstream / Flink
+		 * m_vid / n_vid = pixel rate / link rate
+		 */
+
+		m_vid_l *= param->timing.pix_clk_100hz / 10;
+		m_vid_l = div_u64(m_vid_l,
+			param->link_settings.link_rate
+				* LINK_RATE_REF_FREQ_IN_KHZ);
+
+		m_vid = (uint32_t) m_vid_l;
+
+		/* enable auto measurement */
+
+		REG_UPDATE(DP_VID_TIMING, DP_VID_M_N_GEN_EN, 0);
+
+		/* auto measurement need 1 full 0x8000 symbol cycle to kick in,
+		 * therefore program initial value for Mvid and Nvid
+		 */
+
+		REG_UPDATE(DP_VID_N, DP_VID_N, n_vid);
+
+		REG_UPDATE(DP_VID_M, DP_VID_M, m_vid);
+
+		REG_UPDATE_2(DP_VID_TIMING,
+				DP_VID_M_N_GEN_EN, 1,
+				DP_VID_N_MUL, n_multiply);
+
+		REG_UPDATE(DP_PIXEL_FORMAT,
+				DP_PIXEL_PER_CYCLE_PROCESSING_MODE,
+				pix_per_cycle);
+	}
+
+	/* make sure stream is disabled before resetting steer fifo */
+	REG_UPDATE(DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, false);
+	REG_WAIT(DP_VID_STREAM_CNTL, DP_VID_STREAM_STATUS, 0, 10, 5000);
+
+	/* DIG_START is removed from the register spec */
+
+	/* switch DP encoder to CRTC data, but reset it the fifo first. It may happen
+	 * that it overflows during mode transition, and sometimes doesn't recover.
+	 */
+	REG_UPDATE(DP_STEER_FIFO, DP_STEER_FIFO_RESET, 1);
+	udelay(10);
+
+	REG_UPDATE(DP_STEER_FIFO, DP_STEER_FIFO_RESET, 0);
+
+	/* DIG Resync FIFO now needs to be explicitly enabled
+	 */
+	// TODO: Confirm if we need to wait for DIG_SYMCLK_FE_ON
+	REG_WAIT(DIG_FE_CNTL, DIG_SYMCLK_FE_ON, 1, 10, 5000);
+
+	/* read start level = 0 will bring underflow / overflow and DIG_FIFO_ERROR = 1
+	 * so set it to 1/2 full = 7 before reset as suggested by hardware team.
+	 */
+	REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_READ_START_LEVEL, 0x7);
+
+	REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_RESET, 1);
+
+	REG_WAIT(DIG_FIFO_CTRL0, DIG_FIFO_RESET_DONE, 1, 10, 5000);
+
+	REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_RESET, 0);
+
+	REG_WAIT(DIG_FIFO_CTRL0, DIG_FIFO_RESET_DONE, 0, 10, 5000);
+
+	REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_ENABLE, 1);
+
+	/* wait 100us for DIG/DP logic to prime
+	 * (i.e. a few video lines)
+	 */
+	udelay(100);
+
+	/* the hardware would start sending video at the start of the next DP
+	 * frame (i.e. rising edge of the vblank).
+	 * NOTE: We used to program DP_VID_STREAM_DIS_DEFER = 2 here, but this
+	 * register has no effect on enable transition! HW always guarantees
+	 * VID_STREAM enable at start of next frame, and this is not
+	 * programmable
+	 */
+
+	REG_UPDATE(DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, true);
+
+	link->dc->link_srv->dp_trace_source_sequence(link, DPCD_SOURCE_SEQ_AFTER_ENABLE_DP_VID_STREAM);
+}
+
+/* Set DSC-related configuration.
+ *   dsc_mode: 0 disables DSC, other values enable DSC in specified format
+ *   sc_bytes_per_pixel: DP_DSC_BYTES_PER_PIXEL removed in DCN32
+ *   dsc_slice_width: DP_DSC_SLICE_WIDTH removed in DCN32
+ */
+static void enc32_dp_set_dsc_config(struct stream_encoder *enc,
+					enum optc_dsc_mode dsc_mode,
+					uint32_t dsc_bytes_per_pixel,
+					uint32_t dsc_slice_width)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	REG_UPDATE(DP_DSC_CNTL,	DP_DSC_MODE, dsc_mode == OPTC_DSC_DISABLED ? 0 : 1);
+}
+
+/* this function read dsc related register fields to be logged later in dcn10_log_hw_state
+ * into a dcn_dsc_state struct.
+ */
+static void enc32_read_state(struct stream_encoder *enc, struct enc_state *s)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	//if dsc is enabled, continue to read
+	REG_GET(DP_DSC_CNTL, DP_DSC_MODE, &s->dsc_mode);
+	if (s->dsc_mode) {
+		REG_GET(DP_GSP11_CNTL, DP_SEC_GSP11_LINE_NUM, &s->sec_gsp_pps_line_num);
+
+		REG_GET(DP_MSA_VBID_MISC, DP_VBID6_LINE_REFERENCE, &s->vbid6_line_reference);
+		REG_GET(DP_MSA_VBID_MISC, DP_VBID6_LINE_NUM, &s->vbid6_line_num);
+
+		REG_GET(DP_GSP11_CNTL, DP_SEC_GSP11_ENABLE, &s->sec_gsp_pps_enable);
+		REG_GET(DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, &s->sec_stream_enable);
+	}
+}
+
+static void enc32_set_dig_input_mode(struct stream_encoder *enc, unsigned int pix_per_container)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	/* The naming of this field is confusing, what it means is the output mode of otg, which
+	 * is the input mode of the dig
+	 */
+	REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_OUTPUT_PIXEL_MODE, pix_per_container == 2 ? 0x1 : 0x0);
+}
+
+static void enc32_reset_fifo(struct stream_encoder *enc, bool reset)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+	uint32_t reset_val = reset ? 1 : 0;
+	uint32_t is_symclk_on;
+
+	REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_RESET, reset_val);
+	REG_GET(DIG_FE_CNTL, DIG_SYMCLK_FE_ON, &is_symclk_on);
+
+	if (is_symclk_on)
+		REG_WAIT(DIG_FIFO_CTRL0, DIG_FIFO_RESET_DONE, reset_val, 10, 5000);
+	else
+		udelay(10);
+}
+
+void enc32_enable_fifo(struct stream_encoder *enc)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_READ_START_LEVEL, 0x7);
+
+	enc32_reset_fifo(enc, true);
+	enc32_reset_fifo(enc, false);
+
+	REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_ENABLE, 1);
+}
+
+static const struct stream_encoder_funcs dcn32_str_enc_funcs = {
+	.dp_set_odm_combine =
+		enc32_dp_set_odm_combine,
+	.dp_set_stream_attribute =
+		enc2_stream_encoder_dp_set_stream_attribute,
+	.hdmi_set_stream_attribute =
+		enc32_stream_encoder_hdmi_set_stream_attribute,
+	.dvi_set_stream_attribute =
+		enc32_stream_encoder_dvi_set_stream_attribute,
+	.set_throttled_vcp_size =
+		enc1_stream_encoder_set_throttled_vcp_size,
+	.update_hdmi_info_packets =
+		enc3_stream_encoder_update_hdmi_info_packets,
+	.stop_hdmi_info_packets =
+		enc3_stream_encoder_stop_hdmi_info_packets,
+	.update_dp_info_packets_sdp_line_num =
+		enc3_stream_encoder_update_dp_info_packets_sdp_line_num,
+	.update_dp_info_packets =
+		enc3_stream_encoder_update_dp_info_packets,
+	.stop_dp_info_packets =
+		enc1_stream_encoder_stop_dp_info_packets,
+	.dp_blank =
+		enc1_stream_encoder_dp_blank,
+	.dp_unblank =
+		enc32_stream_encoder_dp_unblank,
+	.audio_mute_control = enc3_audio_mute_control,
+
+	.dp_audio_setup = enc3_se_dp_audio_setup,
+	.dp_audio_enable = enc3_se_dp_audio_enable,
+	.dp_audio_disable = enc1_se_dp_audio_disable,
+
+	.hdmi_audio_setup = enc3_se_hdmi_audio_setup,
+	.hdmi_audio_disable = enc1_se_hdmi_audio_disable,
+	.setup_stereo_sync  = enc1_setup_stereo_sync,
+	.set_avmute = enc1_stream_encoder_set_avmute,
+	.dig_connect_to_otg = enc1_dig_connect_to_otg,
+	.dig_source_otg = enc1_dig_source_otg,
+
+	.dp_get_pixel_format  = enc1_stream_encoder_dp_get_pixel_format,
+
+	.enc_read_state = enc32_read_state,
+	.dp_set_dsc_config = enc32_dp_set_dsc_config,
+	.dp_set_dsc_pps_info_packet = enc3_dp_set_dsc_pps_info_packet,
+	.set_dynamic_metadata = enc2_set_dynamic_metadata,
+	.hdmi_reset_stream_attribute = enc1_reset_hdmi_stream_attribute,
+
+	.set_input_mode = enc32_set_dig_input_mode,
+	.enable_fifo = enc32_enable_fifo,
+};
+
+void dcn32_dio_stream_encoder_construct(
+	struct dcn10_stream_encoder *enc1,
+	struct dc_context *ctx,
+	struct dc_bios *bp,
+	enum engine_id eng_id,
+	struct vpg *vpg,
+	struct afmt *afmt,
+	const struct dcn10_stream_enc_registers *regs,
+	const struct dcn10_stream_encoder_shift *se_shift,
+	const struct dcn10_stream_encoder_mask *se_mask)
+{
+	enc1->base.funcs = &dcn32_str_enc_funcs;
+	enc1->base.ctx = ctx;
+	enc1->base.id = eng_id;
+	enc1->base.bp = bp;
+	enc1->base.vpg = vpg;
+	enc1->base.afmt = afmt;
+	enc1->regs = regs;
+	enc1->se_shift = se_shift;
+	enc1->se_mask = se_mask;
+	enc1->base.stream_enc_inst = vpg->inst;
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn32/dcn32_dio_stream_encoder.h b/drivers/gpu/drm/amd/display/dc/dio/dcn32/dcn32_dio_stream_encoder.h
new file mode 100644
index 000000000000..ca53d39561d2
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn32/dcn32_dio_stream_encoder.h
@@ -0,0 +1,206 @@
+/*
+ * Copyright 2021 - Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_DIO_STREAM_ENCODER_DCN32_H__
+#define __DC_DIO_STREAM_ENCODER_DCN32_H__
+
+#include "dcn30/dcn30_vpg.h"
+#include "dcn30/dcn30_afmt.h"
+#include "stream_encoder.h"
+#include "dcn20/dcn20_stream_encoder.h"
+
+#define SE_COMMON_MASK_SH_LIST_DCN32(mask_sh)\
+	SE_SF(DP0_DP_PIXEL_FORMAT, DP_PIXEL_ENCODING, mask_sh),\
+	SE_SF(DP0_DP_PIXEL_FORMAT, DP_COMPONENT_DEPTH, mask_sh),\
+	SE_SF(DP0_DP_PIXEL_FORMAT, DP_PIXEL_PER_CYCLE_PROCESSING_MODE, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_PACKET_GEN_VERSION, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_KEEPOUT_MODE, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_DATA_SCRAMBLE_EN, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_NO_EXTRA_NULL_PACKET_FILLED, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_GC_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_GC_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_NULL_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_ACP_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_INFOFRAME_CONTROL1, HDMI_AUDIO_INFO_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GC, HDMI_GC_AVMUTE, mask_sh),\
+	SE_SF(DP0_DP_MSE_RATE_CNTL, DP_MSE_RATE_X, mask_sh),\
+	SE_SF(DP0_DP_MSE_RATE_CNTL, DP_MSE_RATE_Y, mask_sh),\
+	SE_SF(DP0_DP_MSE_RATE_UPDATE, DP_MSE_RATE_UPDATE_PENDING, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP0_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP1_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP2_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP3_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_MPG_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL1, DP_SEC_GSP5_LINE_REFERENCE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP4_SEND, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP4_SEND_PENDING, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL4, DP_SEC_GSP4_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL5, DP_SEC_GSP5_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP4_SEND_ANY_LINE, mask_sh),\
+	SE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_DIS_DEFER, mask_sh),\
+	SE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_STATUS, mask_sh),\
+	SE_SF(DP0_DP_STEER_FIFO, DP_STEER_FIFO_RESET, mask_sh),\
+	SE_SF(DP0_DP_VID_TIMING, DP_VID_M_N_GEN_EN, mask_sh),\
+	SE_SF(DP0_DP_VID_N, DP_VID_N, mask_sh),\
+	SE_SF(DP0_DP_VID_M, DP_VID_M, mask_sh),\
+	SE_SF(DIG0_HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_DELAY_EN, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUTO_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_PACKET_CONTROL, HDMI_ACR_SOURCE, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUDIO_PRIORITY, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_32_0, HDMI_ACR_CTS_32, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_32_1, HDMI_ACR_N_32, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_44_0, HDMI_ACR_CTS_44, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_44_1, HDMI_ACR_N_44, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_48_0, HDMI_ACR_CTS_48, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_48_1, HDMI_ACR_N_48, mask_sh),\
+	SE_SF(DP0_DP_SEC_AUD_N, DP_SEC_AUD_N, mask_sh),\
+	SE_SF(DP0_DP_SEC_TIMESTAMP, DP_SEC_TIMESTAMP_MODE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_ASP_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_ATP_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_AIP_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_ACM_ENABLE, mask_sh),\
+	SE_SF(DIG0_AFMT_CNTL, AFMT_AUDIO_CLOCK_EN, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_CLOCK_CHANNEL_RATE, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, TMDS_PIXEL_ENCODING, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, TMDS_COLOR_FORMAT, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_STEREOSYNC_SELECT, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_STEREOSYNC_GATE_EN, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP4_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP5_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP6_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP7_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP7_SEND, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL6, DP_SEC_GSP7_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP11_PPS, mask_sh),\
+	SE_SF(DP0_DP_GSP11_CNTL, DP_SEC_GSP11_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_GSP11_CNTL, DP_SEC_GSP11_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_DB_CNTL, DP_DB_DISABLE, mask_sh),\
+	SE_SF(DP0_DP_MSA_COLORIMETRY, DP_MSA_MISC0, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM1, DP_MSA_HTOTAL, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM1, DP_MSA_VTOTAL, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM2, DP_MSA_HSTART, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM2, DP_MSA_VSTART, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_HSYNCWIDTH, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_HSYNCPOLARITY, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_VSYNCWIDTH, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_VSYNCPOLARITY, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM4, DP_MSA_HWIDTH, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM4, DP_MSA_VHEIGHT, mask_sh),\
+	SE_SF(DIG0_HDMI_DB_CONTROL, HDMI_DB_DISABLE, mask_sh),\
+	SE_SF(DP0_DP_VID_TIMING, DP_VID_N_MUL, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_SOURCE_SELECT, mask_sh), \
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC0_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC0_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC1_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC1_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC2_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC2_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC3_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC3_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC4_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC4_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC5_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC5_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC6_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC6_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC7_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC7_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC8_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC8_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC9_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC9_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC10_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC10_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC11_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC11_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC12_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC12_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC13_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC13_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC14_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC14_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL1, HDMI_GENERIC0_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL1, HDMI_GENERIC1_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL2, HDMI_GENERIC2_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL2, HDMI_GENERIC3_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL3, HDMI_GENERIC4_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL3, HDMI_GENERIC5_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL4, HDMI_GENERIC6_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL4, HDMI_GENERIC7_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL7, HDMI_GENERIC8_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL7, HDMI_GENERIC9_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL8, HDMI_GENERIC10_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL8, HDMI_GENERIC11_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL9, HDMI_GENERIC12_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL9, HDMI_GENERIC13_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL10, HDMI_GENERIC14_LINE, mask_sh),\
+	SE_SF(DP0_DP_DSC_CNTL, DP_DSC_MODE, mask_sh),\
+	SE_SF(DP0_DP_MSA_VBID_MISC, DP_VBID6_LINE_REFERENCE, mask_sh),\
+	SE_SF(DP0_DP_MSA_VBID_MISC, DP_VBID6_LINE_NUM, mask_sh),\
+	SE_SF(DME0_DME_CONTROL, METADATA_ENGINE_EN, mask_sh),\
+	SE_SF(DME0_DME_CONTROL, METADATA_HUBP_REQUESTOR_ID, mask_sh),\
+	SE_SF(DME0_DME_CONTROL, METADATA_STREAM_TYPE, mask_sh),\
+	SE_SF(DP0_DP_SEC_METADATA_TRANSMISSION, DP_SEC_METADATA_PACKET_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_METADATA_TRANSMISSION, DP_SEC_METADATA_PACKET_LINE_REFERENCE, mask_sh),\
+	SE_SF(DP0_DP_SEC_METADATA_TRANSMISSION, DP_SEC_METADATA_PACKET_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_METADATA_PACKET_CONTROL, HDMI_METADATA_PACKET_ENABLE, mask_sh),\
+	SE_SF(DIG0_HDMI_METADATA_PACKET_CONTROL, HDMI_METADATA_PACKET_LINE_REFERENCE, mask_sh),\
+	SE_SF(DIG0_HDMI_METADATA_PACKET_CONTROL, HDMI_METADATA_PACKET_LINE, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DOLBY_VISION_EN, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_SYMCLK_FE_ON, mask_sh),\
+	SE_SF(DP0_DP_SEC_FRAMING4, DP_SST_SDP_SPLITTING, mask_sh),\
+	SE_SF(DIG0_DIG_CLOCK_PATTERN, DIG_CLOCK_PATTERN, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_CTRL0, DIG_FIFO_OUTPUT_PIXEL_MODE, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_CTRL0, DIG_FIFO_READ_START_LEVEL, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_CTRL0, DIG_FIFO_ENABLE, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_CTRL0, DIG_FIFO_RESET, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_CTRL0, DIG_FIFO_RESET_DONE, mask_sh)
+
+
+void dcn32_dio_stream_encoder_construct(
+	struct dcn10_stream_encoder *enc1,
+	struct dc_context *ctx,
+	struct dc_bios *bp,
+	enum engine_id eng_id,
+	struct vpg *vpg,
+	struct afmt *afmt,
+	const struct dcn10_stream_enc_registers *regs,
+	const struct dcn10_stream_encoder_shift *se_shift,
+	const struct dcn10_stream_encoder_mask *se_mask);
+
+
+void enc32_enable_fifo(struct stream_encoder *enc);
+
+void enc32_stream_encoder_dp_unblank(
+		struct dc_link *link,
+		struct stream_encoder *enc,
+		const struct encoder_unblank_param *param);
+
+#endif /* __DC_DIO_STREAM_ENCODER_DCN32_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn321/dcn321_dio_link_encoder.c b/drivers/gpu/drm/amd/display/dc/dio/dcn321/dcn321_dio_link_encoder.c
new file mode 100644
index 000000000000..2ed382a8e79c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn321/dcn321_dio_link_encoder.c
@@ -0,0 +1,190 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "reg_helper.h"
+
+#include "core_types.h"
+#include "link_encoder.h"
+#include "dcn321_dio_link_encoder.h"
+#include "dcn31/dcn31_dio_link_encoder.h"
+#include "stream_encoder.h"
+#include "dc_bios_types.h"
+
+#include "gpio_service_interface.h"
+
+#ifndef MIN
+#define MIN(X, Y) ((X) < (Y) ? (X) : (Y))
+#endif
+
+#define CTX \
+	enc10->base.ctx
+#define DC_LOGGER \
+	enc10->base.ctx->logger
+
+#define REG(reg)\
+	(enc10->link_regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	enc10->link_shift->field_name, enc10->link_mask->field_name
+
+#define AUX_REG(reg)\
+	(enc10->aux_regs->reg)
+
+#define AUX_REG_READ(reg_name) \
+		dm_read_reg(CTX, AUX_REG(reg_name))
+
+#define AUX_REG_WRITE(reg_name, val) \
+			dm_write_reg(CTX, AUX_REG(reg_name), val)
+
+static const struct link_encoder_funcs dcn321_link_enc_funcs = {
+	.read_state = link_enc2_read_state,
+	.validate_output_with_stream =
+			dcn30_link_encoder_validate_output_with_stream,
+	.hw_init = enc32_hw_init,
+	.setup = dcn10_link_encoder_setup,
+	.enable_tmds_output = dcn10_link_encoder_enable_tmds_output,
+	.enable_dp_output = dcn32_link_encoder_enable_dp_output,
+	.enable_dp_mst_output = dcn10_link_encoder_enable_dp_mst_output,
+	.disable_output = dcn10_link_encoder_disable_output,
+	.dp_set_lane_settings = dcn10_link_encoder_dp_set_lane_settings,
+	.dp_set_phy_pattern = dcn10_link_encoder_dp_set_phy_pattern,
+	.update_mst_stream_allocation_table =
+		dcn10_link_encoder_update_mst_stream_allocation_table,
+	.psr_program_dp_dphy_fast_training =
+			dcn10_psr_program_dp_dphy_fast_training,
+	.psr_program_secondary_packet = dcn10_psr_program_secondary_packet,
+	.connect_dig_be_to_fe = dcn10_link_encoder_connect_dig_be_to_fe,
+	.enable_hpd = dcn10_link_encoder_enable_hpd,
+	.disable_hpd = dcn10_link_encoder_disable_hpd,
+	.is_dig_enabled = dcn10_is_dig_enabled,
+	.destroy = dcn10_link_encoder_destroy,
+	.fec_set_enable = enc2_fec_set_enable,
+	.fec_set_ready = enc2_fec_set_ready,
+	.fec_is_active = enc2_fec_is_active,
+	.get_dig_frontend = dcn10_get_dig_frontend,
+	.get_dig_mode = dcn10_get_dig_mode,
+	.is_in_alt_mode = dcn20_link_encoder_is_in_alt_mode,
+	.get_max_link_cap = dcn20_link_encoder_get_max_link_cap,
+	.set_dio_phy_mux = dcn31_link_encoder_set_dio_phy_mux,
+};
+
+void dcn321_link_encoder_construct(
+	struct dcn20_link_encoder *enc20,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	const struct dcn10_link_enc_aux_registers *aux_regs,
+	const struct dcn10_link_enc_hpd_registers *hpd_regs,
+	const struct dcn10_link_enc_shift *link_shift,
+	const struct dcn10_link_enc_mask *link_mask)
+{
+	struct bp_connector_speed_cap_info bp_cap_info = {0};
+	const struct dc_vbios_funcs *bp_funcs = init_data->ctx->dc_bios->funcs;
+	enum bp_result result = BP_RESULT_OK;
+	struct dcn10_link_encoder *enc10 = &enc20->enc10;
+
+	enc10->base.funcs = &dcn321_link_enc_funcs;
+	enc10->base.ctx = init_data->ctx;
+	enc10->base.id = init_data->encoder;
+
+	enc10->base.hpd_source = init_data->hpd_source;
+	enc10->base.connector = init_data->connector;
+
+	if (enc10->base.connector.id == CONNECTOR_ID_USBC)
+		enc10->base.features.flags.bits.DP_IS_USB_C = 1;
+
+	enc10->base.preferred_engine = ENGINE_ID_UNKNOWN;
+
+	enc10->base.features = *enc_features;
+
+	enc10->base.transmitter = init_data->transmitter;
+
+	/* set the flag to indicate whether driver poll the I2C data pin
+	 * while doing the DP sink detect
+	 */
+
+	enc10->base.output_signals =
+		SIGNAL_TYPE_DVI_SINGLE_LINK |
+		SIGNAL_TYPE_DVI_DUAL_LINK |
+		SIGNAL_TYPE_LVDS |
+		SIGNAL_TYPE_DISPLAY_PORT |
+		SIGNAL_TYPE_DISPLAY_PORT_MST |
+		SIGNAL_TYPE_EDP |
+		SIGNAL_TYPE_HDMI_TYPE_A;
+
+	enc10->link_regs = link_regs;
+	enc10->aux_regs = aux_regs;
+	enc10->hpd_regs = hpd_regs;
+	enc10->link_shift = link_shift;
+	enc10->link_mask = link_mask;
+
+	switch (enc10->base.transmitter) {
+	case TRANSMITTER_UNIPHY_A:
+		enc10->base.preferred_engine = ENGINE_ID_DIGA;
+	break;
+	case TRANSMITTER_UNIPHY_B:
+		enc10->base.preferred_engine = ENGINE_ID_DIGB;
+	break;
+	case TRANSMITTER_UNIPHY_C:
+		enc10->base.preferred_engine = ENGINE_ID_DIGC;
+	break;
+	case TRANSMITTER_UNIPHY_D:
+		enc10->base.preferred_engine = ENGINE_ID_DIGD;
+	break;
+	case TRANSMITTER_UNIPHY_E:
+		enc10->base.preferred_engine = ENGINE_ID_DIGE;
+	break;
+	default:
+		ASSERT_CRITICAL(false);
+		enc10->base.preferred_engine = ENGINE_ID_UNKNOWN;
+	}
+
+	/* default to one to mirror Windows behavior */
+	enc10->base.features.flags.bits.HDMI_6GB_EN = 1;
+
+	if (bp_funcs->get_connector_speed_cap_info)
+		result = bp_funcs->get_connector_speed_cap_info(enc10->base.ctx->dc_bios,
+						enc10->base.connector, &bp_cap_info);
+
+	/* Override features with DCE-specific values */
+	if (result == BP_RESULT_OK) {
+		enc10->base.features.flags.bits.IS_HBR2_CAPABLE =
+				bp_cap_info.DP_HBR2_EN;
+		enc10->base.features.flags.bits.IS_HBR3_CAPABLE =
+				bp_cap_info.DP_HBR3_EN;
+		enc10->base.features.flags.bits.HDMI_6GB_EN = bp_cap_info.HDMI_6GB_EN;
+		enc10->base.features.flags.bits.IS_DP2_CAPABLE = 1;
+		enc10->base.features.flags.bits.IS_UHBR10_CAPABLE = bp_cap_info.DP_UHBR10_EN;
+		enc10->base.features.flags.bits.IS_UHBR13_5_CAPABLE = bp_cap_info.DP_UHBR13_5_EN;
+		enc10->base.features.flags.bits.IS_UHBR20_CAPABLE = bp_cap_info.DP_UHBR20_EN;
+	} else {
+		DC_LOG_WARNING("%s: Failed to get encoder_cap_info from VBIOS with error code %d!\n",
+				__func__,
+				result);
+	}
+	if (enc10->base.ctx->dc->debug.hdmi20_disable)
+		enc10->base.features.flags.bits.HDMI_6GB_EN = 0;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn321/dcn321_dio_link_encoder.h b/drivers/gpu/drm/amd/display/dc/dio/dcn321/dcn321_dio_link_encoder.h
new file mode 100644
index 000000000000..2205f39b0a24
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn321/dcn321_dio_link_encoder.h
@@ -0,0 +1,42 @@
+/*
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_LINK_ENCODER__DCN321_H__
+#define __DC_LINK_ENCODER__DCN321_H__
+
+#include "dcn32/dcn32_dio_link_encoder.h"
+
+void dcn321_link_encoder_construct(
+	struct dcn20_link_encoder *enc20,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	const struct dcn10_link_enc_aux_registers *aux_regs,
+	const struct dcn10_link_enc_hpd_registers *hpd_regs,
+	const struct dcn10_link_enc_shift *link_shift,
+	const struct dcn10_link_enc_mask *link_mask);
+
+
+#endif /* __DC_LINK_ENCODER__DCN321_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn35/dcn35_dio_link_encoder.c b/drivers/gpu/drm/amd/display/dc/dio/dcn35/dcn35_dio_link_encoder.c
new file mode 100644
index 000000000000..9972911330b6
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn35/dcn35_dio_link_encoder.c
@@ -0,0 +1,391 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "reg_helper.h"
+
+#include "core_types.h"
+#include "link_encoder.h"
+#include "dcn31/dcn31_dio_link_encoder.h"
+#include "dcn35_dio_link_encoder.h"
+#include "dc_dmub_srv.h"
+#define CTX \
+	enc10->base.ctx
+#define DC_LOGGER \
+	enc10->base.ctx->logger
+
+#define REG(reg)\
+	(enc10->link_regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	enc10->link_shift->field_name, enc10->link_mask->field_name
+/*
+ * @brief
+ * Trigger Source Select
+ * ASIC-dependent, actual values for register programming
+ */
+#define DCN35_DIG_FE_SOURCE_SELECT_INVALID 0x0
+#define DCN35_DIG_FE_SOURCE_SELECT_DIGA 0x1
+#define DCN35_DIG_FE_SOURCE_SELECT_DIGB 0x2
+#define DCN35_DIG_FE_SOURCE_SELECT_DIGC 0x4
+#define DCN35_DIG_FE_SOURCE_SELECT_DIGD 0x08
+#define DCN35_DIG_FE_SOURCE_SELECT_DIGE 0x10
+
+
+bool dcn35_is_dig_enabled(struct link_encoder *enc)
+{
+	uint32_t enabled;
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	REG_GET(DIG_BE_CLK_CNTL, DIG_BE_CLK_EN, &enabled);
+	return (enabled == 1);
+}
+
+enum signal_type dcn35_get_dig_mode(
+	struct link_encoder *enc)
+{
+	uint32_t value;
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	REG_GET(DIG_BE_CLK_CNTL, DIG_BE_MODE, &value);
+	switch (value) {
+	case 0:
+		return SIGNAL_TYPE_DISPLAY_PORT;
+	case 2:
+		return SIGNAL_TYPE_DVI_SINGLE_LINK;
+	case 3:
+		return SIGNAL_TYPE_HDMI_TYPE_A;
+	case 5:
+		return SIGNAL_TYPE_DISPLAY_PORT_MST;
+	default:
+		return SIGNAL_TYPE_NONE;
+	}
+}
+
+void dcn35_link_encoder_setup(
+	struct link_encoder *enc,
+	enum signal_type signal)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	switch (signal) {
+	case SIGNAL_TYPE_EDP:
+	case SIGNAL_TYPE_DISPLAY_PORT:
+		/* DP SST */
+		REG_UPDATE(DIG_BE_CLK_CNTL, DIG_BE_MODE, 0);
+		break;
+	case SIGNAL_TYPE_DVI_SINGLE_LINK:
+	case SIGNAL_TYPE_DVI_DUAL_LINK:
+		/* TMDS-DVI */
+		REG_UPDATE(DIG_BE_CLK_CNTL, DIG_BE_MODE, 2);
+		break;
+	case SIGNAL_TYPE_HDMI_TYPE_A:
+		/* TMDS-HDMI */
+		REG_UPDATE(DIG_BE_CLK_CNTL, DIG_BE_MODE, 3);
+		break;
+	case SIGNAL_TYPE_DISPLAY_PORT_MST:
+		/* DP MST */
+		REG_UPDATE(DIG_BE_CLK_CNTL, DIG_BE_MODE, 5);
+		break;
+	default:
+		ASSERT_CRITICAL(false);
+		/* invalid mode ! */
+		break;
+	}
+	REG_UPDATE(DIG_BE_CLK_CNTL, DIG_BE_CLK_EN, 1);
+
+}
+
+void dcn35_link_encoder_init(struct link_encoder *enc)
+{
+	enc31_hw_init(enc);
+	dcn35_link_encoder_set_fgcg(enc, enc->ctx->dc->debug.enable_fine_grain_clock_gating.bits.dio);
+}
+
+void dcn35_link_encoder_set_fgcg(struct link_encoder *enc, bool enable)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	REG_UPDATE(DIO_CLK_CNTL, DIO_FGCG_REP_DIS, !enable);
+}
+
+static const struct link_encoder_funcs dcn35_link_enc_funcs = {
+	.read_state = link_enc2_read_state,
+	.validate_output_with_stream =
+			dcn30_link_encoder_validate_output_with_stream,
+	.hw_init = dcn35_link_encoder_init,
+	.setup = dcn35_link_encoder_setup,
+	.enable_tmds_output = dcn10_link_encoder_enable_tmds_output,
+	.enable_dp_output = dcn35_link_encoder_enable_dp_output,
+	.enable_dp_mst_output = dcn35_link_encoder_enable_dp_mst_output,
+	.disable_output = dcn35_link_encoder_disable_output,
+	.dp_set_lane_settings = dcn10_link_encoder_dp_set_lane_settings,
+	.dp_set_phy_pattern = dcn10_link_encoder_dp_set_phy_pattern,
+	.update_mst_stream_allocation_table =
+		dcn10_link_encoder_update_mst_stream_allocation_table,
+	.psr_program_dp_dphy_fast_training =
+			dcn10_psr_program_dp_dphy_fast_training,
+	.psr_program_secondary_packet = dcn10_psr_program_secondary_packet,
+	.connect_dig_be_to_fe = dcn10_link_encoder_connect_dig_be_to_fe,
+	.enable_hpd = dcn10_link_encoder_enable_hpd,
+	.disable_hpd = dcn10_link_encoder_disable_hpd,
+	.is_dig_enabled = dcn35_is_dig_enabled,
+	.destroy = dcn10_link_encoder_destroy,
+	.fec_set_enable = enc2_fec_set_enable,
+	.fec_set_ready = enc2_fec_set_ready,
+	.fec_is_active = enc2_fec_is_active,
+	.get_dig_frontend = dcn10_get_dig_frontend,
+	.get_dig_mode = dcn35_get_dig_mode,
+	.is_in_alt_mode = dcn31_link_encoder_is_in_alt_mode,
+	.get_max_link_cap = dcn31_link_encoder_get_max_link_cap,
+	.set_dio_phy_mux = dcn31_link_encoder_set_dio_phy_mux,
+	.enable_dpia_output = dcn35_link_encoder_enable_dpia_output,
+	.disable_dpia_output = dcn35_link_encoder_disable_dpia_output,
+};
+
+void dcn35_link_encoder_construct(
+	struct dcn20_link_encoder *enc20,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	const struct dcn10_link_enc_aux_registers *aux_regs,
+	const struct dcn10_link_enc_hpd_registers *hpd_regs,
+	const struct dcn10_link_enc_shift *link_shift,
+	const struct dcn10_link_enc_mask *link_mask)
+{
+	struct bp_connector_speed_cap_info bp_cap_info = {0};
+	const struct dc_vbios_funcs *bp_funcs = init_data->ctx->dc_bios->funcs;
+	enum bp_result result = BP_RESULT_OK;
+	struct dcn10_link_encoder *enc10 = &enc20->enc10;
+
+	enc10->base.funcs = &dcn35_link_enc_funcs;
+	enc10->base.ctx = init_data->ctx;
+	enc10->base.id = init_data->encoder;
+
+	enc10->base.hpd_source = init_data->hpd_source;
+	enc10->base.connector = init_data->connector;
+
+	enc10->base.preferred_engine = ENGINE_ID_UNKNOWN;
+
+	enc10->base.features = *enc_features;
+
+	if (enc10->base.connector.id == CONNECTOR_ID_USBC)
+		enc10->base.features.flags.bits.DP_IS_USB_C = 1;
+
+	enc10->base.transmitter = init_data->transmitter;
+
+	/* set the flag to indicate whether driver poll the I2C data pin
+	 * while doing the DP sink detect
+	 */
+
+/*	if (dal_adapter_service_is_feature_supported(as,
+ *		FEATURE_DP_SINK_DETECT_POLL_DATA_PIN))
+ *		enc10->base.features.flags.bits.
+ *			DP_SINK_DETECT_POLL_DATA_PIN = true;
+ */
+
+	enc10->base.output_signals =
+		SIGNAL_TYPE_DVI_SINGLE_LINK |
+		SIGNAL_TYPE_DVI_DUAL_LINK |
+		SIGNAL_TYPE_LVDS |
+		SIGNAL_TYPE_DISPLAY_PORT |
+		SIGNAL_TYPE_DISPLAY_PORT_MST |
+		SIGNAL_TYPE_EDP |
+		SIGNAL_TYPE_HDMI_TYPE_A;
+
+	enc10->link_regs = link_regs;
+	enc10->aux_regs = aux_regs;
+	enc10->hpd_regs = hpd_regs;
+	enc10->link_shift = link_shift;
+	enc10->link_mask = link_mask;
+
+	switch (enc10->base.transmitter) {
+	case TRANSMITTER_UNIPHY_A:
+		enc10->base.preferred_engine = ENGINE_ID_DIGA;
+	break;
+	case TRANSMITTER_UNIPHY_B:
+		enc10->base.preferred_engine = ENGINE_ID_DIGB;
+	break;
+	case TRANSMITTER_UNIPHY_C:
+		enc10->base.preferred_engine = ENGINE_ID_DIGC;
+	break;
+	case TRANSMITTER_UNIPHY_D:
+		enc10->base.preferred_engine = ENGINE_ID_DIGD;
+	break;
+	case TRANSMITTER_UNIPHY_E:
+		enc10->base.preferred_engine = ENGINE_ID_DIGE;
+	break;
+	default:
+		ASSERT_CRITICAL(false);
+		enc10->base.preferred_engine = ENGINE_ID_UNKNOWN;
+	}
+
+	enc10->base.features.flags.bits.HDMI_6GB_EN = 1;
+
+	if (bp_funcs->get_connector_speed_cap_info)
+		result = bp_funcs->get_connector_speed_cap_info(enc10->base.ctx->dc_bios,
+						enc10->base.connector, &bp_cap_info);
+
+	/* Override features with DCE-specific values */
+	if (result == BP_RESULT_OK) {
+		enc10->base.features.flags.bits.IS_HBR2_CAPABLE =
+				bp_cap_info.DP_HBR2_EN;
+		enc10->base.features.flags.bits.IS_HBR3_CAPABLE =
+				bp_cap_info.DP_HBR3_EN;
+		enc10->base.features.flags.bits.HDMI_6GB_EN = bp_cap_info.HDMI_6GB_EN;
+		enc10->base.features.flags.bits.IS_DP2_CAPABLE = 1;
+		enc10->base.features.flags.bits.IS_UHBR10_CAPABLE = bp_cap_info.DP_UHBR10_EN;
+		enc10->base.features.flags.bits.IS_UHBR13_5_CAPABLE = bp_cap_info.DP_UHBR13_5_EN;
+		enc10->base.features.flags.bits.IS_UHBR20_CAPABLE = bp_cap_info.DP_UHBR20_EN;
+
+	} else {
+		DC_LOG_WARNING("%s: Failed to get encoder_cap_info from VBIOS with error code %d!\n",
+				__func__,
+				result);
+	}
+	if (enc10->base.ctx->dc->debug.hdmi20_disable)
+		enc10->base.features.flags.bits.HDMI_6GB_EN = 0;
+
+}
+
+/* DPIA equivalent of link_transmitter_control. */
+static bool link_dpia_control(struct dc_context *dc_ctx,
+	struct dmub_cmd_dig_dpia_control_data *dpia_control)
+{
+	union dmub_rb_cmd cmd;
+
+	memset(&cmd, 0, sizeof(cmd));
+
+	cmd.dig1_dpia_control.header.type = DMUB_CMD__DPIA;
+	cmd.dig1_dpia_control.header.sub_type =
+			DMUB_CMD__DPIA_DIG1_DPIA_CONTROL;
+	cmd.dig1_dpia_control.header.payload_bytes =
+		sizeof(cmd.dig1_dpia_control) -
+		sizeof(cmd.dig1_dpia_control.header);
+
+	cmd.dig1_dpia_control.dpia_control = *dpia_control;
+
+	dc_wake_and_execute_dmub_cmd(dc_ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT);
+
+	return true;
+}
+
+static void link_encoder_disable(struct dcn10_link_encoder *enc10)
+{
+	/* reset training complete */
+	REG_UPDATE(DP_LINK_CNTL, DP_LINK_TRAINING_COMPLETE, 0);
+}
+
+void dcn35_link_encoder_enable_dp_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	if (!enc->ctx->dc->config.unify_link_enc_assignment)
+		dcn31_link_encoder_enable_dp_output(enc, link_settings, clock_source);
+	else {
+		DC_LOG_DEBUG("%s: enc_id(%d)\n", __func__, enc->preferred_engine);
+		dcn20_link_encoder_enable_dp_output(enc, link_settings, clock_source);
+	}
+}
+
+void dcn35_link_encoder_enable_dp_mst_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	if (!enc->ctx->dc->config.unify_link_enc_assignment)
+		dcn31_link_encoder_enable_dp_mst_output(enc, link_settings, clock_source);
+	else {
+		DC_LOG_DEBUG("%s: enc_id(%d)\n", __func__, enc->preferred_engine);
+		dcn10_link_encoder_enable_dp_mst_output(enc, link_settings, clock_source);
+	}
+}
+
+void dcn35_link_encoder_disable_output(
+	struct link_encoder *enc,
+	enum signal_type signal)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	if (!enc->ctx->dc->config.unify_link_enc_assignment)
+		dcn31_link_encoder_disable_output(enc, signal);
+	else {
+		DC_LOG_DEBUG("%s: enc_id(%d)\n", __func__, enc->preferred_engine);
+		dcn10_link_encoder_disable_output(enc, signal);
+	}
+}
+
+void dcn35_link_encoder_enable_dpia_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	uint8_t dpia_id,
+	uint8_t digmode,
+	uint8_t fec_rdy)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	struct dmub_cmd_dig_dpia_control_data dpia_control = { 0 };
+
+	enc1_configure_encoder(enc10, link_settings);
+
+	dpia_control.action = (uint8_t)TRANSMITTER_CONTROL_ENABLE;
+	dpia_control.enc_id = enc->preferred_engine;
+	dpia_control.mode_laneset.digmode = digmode;
+	dpia_control.lanenum = (uint8_t)link_settings->lane_count;
+	dpia_control.symclk_10khz = link_settings->link_rate *
+			LINK_RATE_REF_FREQ_IN_KHZ / 10;
+	/* DIG_BE_CNTL.DIG_HPD_SELECT set to 5 (hpdsel - 1) to indicate HPD pin unused by DPIA. */
+	dpia_control.hpdsel = 6;
+	dpia_control.dpia_id = dpia_id;
+	dpia_control.fec_rdy = fec_rdy;
+
+	DC_LOG_DEBUG("%s: DPIA(%d) - enc_id(%d)\n", __func__, dpia_control.dpia_id, dpia_control.enc_id);
+	link_dpia_control(enc->ctx, &dpia_control);
+}
+
+void dcn35_link_encoder_disable_dpia_output(
+	struct link_encoder *enc,
+	uint8_t dpia_id,
+	uint8_t digmode)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	struct dmub_cmd_dig_dpia_control_data dpia_control = { 0 };
+
+	if (enc->funcs->is_dig_enabled && !enc->funcs->is_dig_enabled(enc))
+		return;
+
+	dpia_control.action = (uint8_t)TRANSMITTER_CONTROL_DISABLE;
+	dpia_control.enc_id = enc->preferred_engine;
+	dpia_control.mode_laneset.digmode = digmode;
+	dpia_control.dpia_id = dpia_id;
+
+	DC_LOG_DEBUG("%s: DPIA(%d) - enc_id(%d)\n", __func__, dpia_control.dpia_id, dpia_control.enc_id);
+	link_dpia_control(enc->ctx, &dpia_control);
+
+	link_encoder_disable(enc10);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn35/dcn35_dio_link_encoder.h b/drivers/gpu/drm/amd/display/dc/dio/dcn35/dcn35_dio_link_encoder.h
new file mode 100644
index 000000000000..5712e6553fab
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn35/dcn35_dio_link_encoder.h
@@ -0,0 +1,188 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __DC_LINK_ENCODER__DCN35_H__
+#define __DC_LINK_ENCODER__DCN35_H__
+
+#include "dcn32/dcn32_dio_link_encoder.h"
+#include "dcn30/dcn30_dio_link_encoder.h"
+#include "dcn31/dcn31_dio_link_encoder.h"
+
+#define LINK_ENCODER_MASK_SH_LIST_DCN35(mask_sh) \
+	LE_SF(DIG0_DIG_BE_EN_CNTL, DIG_BE_ENABLE, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CNTL, DIG_RB_SWITCH_EN, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CNTL, DIG_HPD_SELECT, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CNTL, DIG_FE_SOURCE_SELECT, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CLK_CNTL, DIG_BE_MODE, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CLK_CNTL, DIG_BE_CLK_EN, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CLK_CNTL, DIG_BE_SOFT_RESET, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CLK_CNTL, HDCP_SOFT_RESET, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CLK_CNTL, DIG_BE_SYMCLK_G_CLOCK_ON, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CLK_CNTL, DIG_BE_SYMCLK_G_HDCP_CLOCK_ON, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CLK_CNTL, DIG_BE_SYMCLK_G_TMDS_CLOCK_ON, mask_sh),\
+	LE_SF(DIG0_DIG_CLOCK_PATTERN, DIG_CLOCK_PATTERN, mask_sh),\
+	LE_SF(DIG0_TMDS_CTL_BITS, TMDS_CTL0, mask_sh), \
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_BYPASS, mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_ATEST_SEL_LANE0, mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_ATEST_SEL_LANE1, mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_ATEST_SEL_LANE2, mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_ATEST_SEL_LANE3, mask_sh),\
+	LE_SF(DP0_DP_DPHY_PRBS_CNTL, DPHY_PRBS_EN, mask_sh),\
+	LE_SF(DP0_DP_DPHY_PRBS_CNTL, DPHY_PRBS_SEL, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM0, DPHY_SYM1, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM0, DPHY_SYM2, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM0, DPHY_SYM3, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM1, DPHY_SYM4, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM1, DPHY_SYM5, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM1, DPHY_SYM6, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM2, DPHY_SYM7, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM2, DPHY_SYM8, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SCRAM_CNTL, DPHY_SCRAMBLER_BS_COUNT, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SCRAM_CNTL, DPHY_SCRAMBLER_ADVANCE, mask_sh),\
+	LE_SF(DP0_DP_DPHY_FAST_TRAINING, DPHY_RX_FAST_TRAINING_CAPABLE, mask_sh),\
+	LE_SF(DP0_DP_DPHY_BS_SR_SWAP_CNTL, DPHY_LOAD_BS_COUNT, mask_sh),\
+	LE_SF(DP0_DP_DPHY_TRAINING_PATTERN_SEL, DPHY_TRAINING_PATTERN_SEL, mask_sh),\
+	LE_SF(DP0_DP_DPHY_HBR2_PATTERN_CONTROL, DP_DPHY_HBR2_PATTERN_CONTROL, mask_sh),\
+	LE_SF(DP0_DP_LINK_CNTL, DP_LINK_TRAINING_COMPLETE, mask_sh),\
+	LE_SF(DP0_DP_LINK_FRAMING_CNTL, DP_IDLE_BS_INTERVAL, mask_sh),\
+	LE_SF(DP0_DP_LINK_FRAMING_CNTL, DP_VBID_DISABLE, mask_sh),\
+	LE_SF(DP0_DP_LINK_FRAMING_CNTL, DP_VID_ENHANCED_FRAME_MODE, mask_sh),\
+	LE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, mask_sh),\
+	LE_SF(DP0_DP_CONFIG, DP_UDI_LANES, mask_sh),\
+	LE_SF(DP0_DP_SEC_CNTL1, DP_SEC_GSP0_LINE_NUM, mask_sh),\
+	LE_SF(DP0_DP_SEC_CNTL1, DP_SEC_GSP0_PRIORITY, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT0, DP_MSE_SAT_SRC0, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT0, DP_MSE_SAT_SRC1, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT0, DP_MSE_SAT_SLOT_COUNT0, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT0, DP_MSE_SAT_SLOT_COUNT1, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT1, DP_MSE_SAT_SRC2, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT1, DP_MSE_SAT_SRC3, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT1, DP_MSE_SAT_SLOT_COUNT2, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT1, DP_MSE_SAT_SLOT_COUNT3, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT_UPDATE, DP_MSE_SAT_UPDATE, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT_UPDATE, DP_MSE_16_MTP_KEEPOUT, mask_sh),\
+	LE_SF(DP_AUX0_AUX_CONTROL, AUX_HPD_SEL, mask_sh),\
+	LE_SF(DP_AUX0_AUX_CONTROL, AUX_LS_READ_EN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_RECEIVE_WINDOW, mask_sh),\
+	LE_SF(HPD0_DC_HPD_CONTROL, DC_HPD_EN, mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_FEC_EN, mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_FEC_READY_SHADOW, mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_FEC_ACTIVE_STATUS, mask_sh),\
+	LE_SF(DIG0_TMDS_CTL_BITS, TMDS_CTL0, mask_sh), \
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_START_WINDOW, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_HALF_SYM_DETECT_LEN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_TRANSITION_FILTER_EN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_ALLOW_BELOW_THRESHOLD_PHASE_DETECT, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_ALLOW_BELOW_THRESHOLD_START, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_ALLOW_BELOW_THRESHOLD_STOP, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_PHASE_DETECT_LEN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_DETECTION_THRESHOLD, mask_sh), \
+	LE_SF(DP_AUX0_AUX_DPHY_TX_CONTROL, AUX_TX_PRECHARGE_LEN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_TX_CONTROL, AUX_TX_PRECHARGE_SYMBOLS, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_TX_CONTROL, AUX_MODE_DET_CHECK_DELAY, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL1, AUX_RX_PRECHARGE_SKIP, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL1, AUX_RX_TIMEOUT_LEN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL1, AUX_RX_TIMEOUT_LEN_MUL, mask_sh),\
+	LE_SF(DIO_LINKA_CNTL, ENC_TYPE_SEL, mask_sh),\
+	LE_SF(DIO_LINKA_CNTL, HPO_DP_ENC_SEL, mask_sh),\
+	LE_SF(DIO_LINKA_CNTL, HPO_HDMI_ENC_SEL, mask_sh),\
+	LE_SF(DIO_CLK_CNTL, DISPCLK_R_GATE_DIS, mask_sh),\
+	LE_SF(DIO_CLK_CNTL, DISPCLK_G_GATE_DIS, mask_sh),\
+	LE_SF(DIO_CLK_CNTL, REFCLK_R_GATE_DIS, mask_sh),\
+	LE_SF(DIO_CLK_CNTL, REFCLK_G_GATE_DIS, mask_sh),\
+	LE_SF(DIO_CLK_CNTL, SOCCLK_G_GATE_DIS, mask_sh),\
+	LE_SF(DIO_CLK_CNTL, SYMCLK_FE_R_GATE_DIS, mask_sh),\
+	LE_SF(DIO_CLK_CNTL, SYMCLK_FE_G_GATE_DIS, mask_sh),\
+	LE_SF(DIO_CLK_CNTL, SYMCLK_R_GATE_DIS, mask_sh),\
+	LE_SF(DIO_CLK_CNTL, SYMCLK_G_GATE_DIS, mask_sh),\
+	LE_SF(DIO_CLK_CNTL, DIO_FGCG_REP_DIS, mask_sh),\
+	LE_SF(DIO_CLK_CNTL, DISPCLK_G_HDCP_GATE_DIS, mask_sh),\
+	LE_SF(DIO_CLK_CNTL, SYMCLKA_G_HDCP_GATE_DIS, mask_sh),\
+	LE_SF(DIO_CLK_CNTL, SYMCLKB_G_HDCP_GATE_DIS, mask_sh),\
+	LE_SF(DIO_CLK_CNTL, SYMCLKC_G_HDCP_GATE_DIS, mask_sh),\
+	LE_SF(DIO_CLK_CNTL, SYMCLKD_G_HDCP_GATE_DIS, mask_sh),\
+	LE_SF(DIO_CLK_CNTL, SYMCLKE_G_HDCP_GATE_DIS, mask_sh),\
+	LE_SF(DIO_CLK_CNTL, SYMCLKF_G_HDCP_GATE_DIS, mask_sh),\
+	LE_SF(DIO_CLK_CNTL, SYMCLKG_G_HDCP_GATE_DIS, mask_sh)
+
+
+void dcn35_link_encoder_construct(
+	struct dcn20_link_encoder *enc20,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	const struct dcn10_link_enc_aux_registers *aux_regs,
+	const struct dcn10_link_enc_hpd_registers *hpd_regs,
+	const struct dcn10_link_enc_shift *link_shift,
+	const struct dcn10_link_enc_mask *link_mask);
+
+void dcn35_link_encoder_init(struct link_encoder *enc);
+void dcn35_link_encoder_set_fgcg(struct link_encoder *enc, bool enabled);
+bool dcn35_is_dig_enabled(struct link_encoder *enc);
+
+enum signal_type dcn35_get_dig_mode(struct link_encoder *enc);
+void dcn35_link_encoder_setup(struct link_encoder *enc, enum signal_type signal);
+
+/*
+ * Enable DP transmitter and its encoder.
+ */
+void dcn35_link_encoder_enable_dp_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source);
+
+/*
+ * Enable DP transmitter and its encoder in MST mode.
+ */
+void dcn35_link_encoder_enable_dp_mst_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source);
+
+/*
+ * Disable transmitter and its encoder.
+ */
+void dcn35_link_encoder_disable_output(
+	struct link_encoder *enc,
+	enum signal_type signal);
+
+/*
+ * Enable DP transmitter and its encoder for dpia port.
+ */
+void dcn35_link_encoder_enable_dpia_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	uint8_t dpia_id,
+	uint8_t digmode,
+	uint8_t fec_rdy);
+
+/*
+ * Disable transmitter and its encoder for dpia port.
+ */
+void dcn35_link_encoder_disable_dpia_output(
+	struct link_encoder *enc,
+	uint8_t dpia_id,
+	uint8_t digmode);
+
+#endif /* __DC_LINK_ENCODER__DCN35_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn35/dcn35_dio_stream_encoder.c b/drivers/gpu/drm/amd/display/dc/dio/dcn35/dcn35_dio_stream_encoder.c
new file mode 100644
index 000000000000..fd5d1dbf9dc6
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn35/dcn35_dio_stream_encoder.c
@@ -0,0 +1,520 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "dc_bios_types.h"
+#include "dcn30/dcn30_dio_stream_encoder.h"
+#include "dcn314/dcn314_dio_stream_encoder.h"
+#include "dcn32/dcn32_dio_stream_encoder.h"
+#include "dcn35_dio_stream_encoder.h"
+#include "reg_helper.h"
+#include "hw_shared.h"
+#include "link_service.h"
+#include "dpcd_defs.h"
+
+#define DC_LOGGER \
+		enc1->base.ctx->logger
+
+#define REG(reg)\
+	(enc1->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	enc1->se_shift->field_name, enc1->se_mask->field_name
+
+#define VBI_LINE_0 0
+#define HDMI_CLOCK_CHANNEL_RATE_MORE_340M 340000
+
+#define CTX \
+	enc1->base.ctx
+/* setup stream encoder in dvi mode */
+static void enc35_stream_encoder_dvi_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	bool is_dual_link)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (!enc->ctx->dc->debug.avoid_vbios_exec_table) {
+		struct bp_encoder_control cntl = {0};
+
+		cntl.action = ENCODER_CONTROL_SETUP;
+		cntl.engine_id = enc1->base.id;
+		cntl.signal = is_dual_link ?
+			SIGNAL_TYPE_DVI_DUAL_LINK : SIGNAL_TYPE_DVI_SINGLE_LINK;
+		cntl.enable_dp_audio = false;
+		cntl.pixel_clock = crtc_timing->pix_clk_100hz / 10;
+		cntl.lanes_number = (is_dual_link) ? LANE_COUNT_EIGHT : LANE_COUNT_FOUR;
+
+		if (enc1->base.bp->funcs->encoder_control(
+				enc1->base.bp, &cntl) != BP_RESULT_OK)
+			return;
+
+	} else {
+
+		//Set pattern for clock channel, default vlue 0x63 does not work
+		REG_UPDATE(DIG_CLOCK_PATTERN, DIG_CLOCK_PATTERN, 0x1F);
+
+		//DIG_BE_TMDS_DVI_MODE : TMDS-DVI mode is already set in link_encoder_setup
+
+		//DIG_SOURCE_SELECT is already set in dig_connect_to_otg
+
+		/* DIG_START is removed from the register spec */
+	}
+
+	ASSERT(crtc_timing->pixel_encoding == PIXEL_ENCODING_RGB);
+	ASSERT(crtc_timing->display_color_depth == COLOR_DEPTH_888);
+	enc1_stream_encoder_set_stream_attribute_helper(enc1, crtc_timing);
+}
+/* setup stream encoder in hdmi mode */
+static void enc35_stream_encoder_hdmi_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	int actual_pix_clk_khz,
+	bool enable_audio)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (!enc->ctx->dc->debug.avoid_vbios_exec_table) {
+		struct bp_encoder_control cntl = {0};
+
+		cntl.action = ENCODER_CONTROL_SETUP;
+		cntl.engine_id = enc1->base.id;
+		cntl.signal = SIGNAL_TYPE_HDMI_TYPE_A;
+		cntl.enable_dp_audio = enable_audio;
+		cntl.pixel_clock = actual_pix_clk_khz;
+		cntl.lanes_number = LANE_COUNT_FOUR;
+
+		if (enc1->base.bp->funcs->encoder_control(
+				enc1->base.bp, &cntl) != BP_RESULT_OK)
+			return;
+
+	} else {
+
+		//Set pattern for clock channel, default vlue 0x63 does not work
+		REG_UPDATE(DIG_CLOCK_PATTERN, DIG_CLOCK_PATTERN, 0x1F);
+
+		//DIG_BE_TMDS_HDMI_MODE : TMDS-HDMI mode is already set in link_encoder_setup
+
+		//DIG_SOURCE_SELECT is already set in dig_connect_to_otg
+
+		/* DIG_START is removed from the register spec */
+		enc314_enable_fifo(enc);
+	}
+
+	/* Configure pixel encoding */
+	enc1_stream_encoder_set_stream_attribute_helper(enc1, crtc_timing);
+
+	/* setup HDMI engine */
+	REG_UPDATE_6(HDMI_CONTROL,
+		HDMI_PACKET_GEN_VERSION, 1,
+		HDMI_KEEPOUT_MODE, 1,
+		HDMI_DEEP_COLOR_ENABLE, 0,
+		HDMI_DATA_SCRAMBLE_EN, 0,
+		HDMI_NO_EXTRA_NULL_PACKET_FILLED, 1,
+		HDMI_CLOCK_CHANNEL_RATE, 0);
+
+	/* Configure color depth */
+	switch (crtc_timing->display_color_depth) {
+	case COLOR_DEPTH_888:
+		REG_UPDATE(HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 0);
+		break;
+	case COLOR_DEPTH_101010:
+		if (crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR422) {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 1,
+					HDMI_DEEP_COLOR_ENABLE, 0);
+		} else {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 1,
+					HDMI_DEEP_COLOR_ENABLE, 1);
+			}
+		break;
+	case COLOR_DEPTH_121212:
+		if (crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR422) {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 2,
+					HDMI_DEEP_COLOR_ENABLE, 0);
+		} else {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 2,
+					HDMI_DEEP_COLOR_ENABLE, 1);
+			}
+		break;
+	case COLOR_DEPTH_161616:
+		REG_UPDATE_2(HDMI_CONTROL,
+				HDMI_DEEP_COLOR_DEPTH, 3,
+				HDMI_DEEP_COLOR_ENABLE, 1);
+		break;
+	default:
+		break;
+	}
+
+	if (actual_pix_clk_khz >= HDMI_CLOCK_CHANNEL_RATE_MORE_340M) {
+		/* enable HDMI data scrambler
+		 * HDMI_CLOCK_CHANNEL_RATE_MORE_340M
+		 * Clock channel frequency is 1/4 of character rate.
+		 */
+		REG_UPDATE_2(HDMI_CONTROL,
+			HDMI_DATA_SCRAMBLE_EN, 1,
+			HDMI_CLOCK_CHANNEL_RATE, 1);
+	} else if (crtc_timing->flags.LTE_340MCSC_SCRAMBLE) {
+
+		/* TODO: New feature for DCE11, still need to implement */
+
+		/* enable HDMI data scrambler
+		 * HDMI_CLOCK_CHANNEL_FREQ_EQUAL_TO_CHAR_RATE
+		 * Clock channel frequency is the same
+		 * as character rate
+		 */
+		REG_UPDATE_2(HDMI_CONTROL,
+			HDMI_DATA_SCRAMBLE_EN, 1,
+			HDMI_CLOCK_CHANNEL_RATE, 0);
+	}
+
+
+	/* Enable transmission of General Control packet on every frame */
+	REG_UPDATE_3(HDMI_VBI_PACKET_CONTROL,
+		HDMI_GC_CONT, 1,
+		HDMI_GC_SEND, 1,
+		HDMI_NULL_SEND, 1);
+
+	/* Disable Audio Content Protection packet transmission */
+	REG_UPDATE(HDMI_VBI_PACKET_CONTROL, HDMI_ACP_SEND, 0);
+
+	/* following belongs to audio */
+	/* Enable Audio InfoFrame packet transmission. */
+	REG_UPDATE(HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, 1);
+
+	/* update double-buffered AUDIO_INFO registers immediately */
+	ASSERT(enc->afmt);
+	enc->afmt->funcs->audio_info_immediate_update(enc->afmt);
+
+	/* Select line number on which to send Audio InfoFrame packets */
+	REG_UPDATE(HDMI_INFOFRAME_CONTROL1, HDMI_AUDIO_INFO_LINE,
+				VBI_LINE_0 + 2);
+
+	/* set HDMI GC AVMUTE */
+	REG_UPDATE(HDMI_GC, HDMI_GC_AVMUTE, 0);
+	switch (crtc_timing->pixel_encoding) {
+	case PIXEL_ENCODING_YCBCR422:
+		REG_UPDATE(HDMI_CONTROL, TMDS_PIXEL_ENCODING, 1);
+	break;
+	default:
+		REG_UPDATE(HDMI_CONTROL, TMDS_PIXEL_ENCODING, 0);
+	break;
+	}
+	REG_UPDATE(HDMI_CONTROL, TMDS_COLOR_FORMAT, 0);
+}
+
+
+
+static void enc35_stream_encoder_enable(
+	struct stream_encoder *enc,
+	enum signal_type signal,
+	bool enable)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (enable) {
+		switch (signal) {
+		case SIGNAL_TYPE_DVI_SINGLE_LINK:
+		case SIGNAL_TYPE_DVI_DUAL_LINK:
+			/* TMDS-DVI */
+			REG_UPDATE(DIG_FE_CLK_CNTL, DIG_FE_MODE, 2);
+			break;
+		case SIGNAL_TYPE_HDMI_TYPE_A:
+			/* TMDS-HDMI */
+			REG_UPDATE(DIG_FE_CLK_CNTL, DIG_FE_MODE, 3);
+			break;
+		case SIGNAL_TYPE_DISPLAY_PORT_MST:
+			/* DP MST */
+			REG_UPDATE(DIG_FE_CLK_CNTL, DIG_FE_MODE, 5);
+			break;
+		case SIGNAL_TYPE_EDP:
+		case SIGNAL_TYPE_DISPLAY_PORT:
+		case SIGNAL_TYPE_VIRTUAL:
+			/* DP SST */
+			REG_UPDATE(DIG_FE_CLK_CNTL, DIG_FE_MODE, 0);
+			break;
+		default:
+			/* invalid mode ! */
+			ASSERT_CRITICAL(false);
+		}
+	}
+}
+
+static bool is_two_pixels_per_containter(const struct dc_crtc_timing *timing)
+{
+	bool two_pix = timing->pixel_encoding == PIXEL_ENCODING_YCBCR420;
+
+	two_pix = two_pix || (timing->flags.DSC && timing->pixel_encoding == PIXEL_ENCODING_YCBCR422
+			&& !timing->dsc_cfg.ycbcr422_simple);
+	return two_pix;
+}
+
+static void enc35_stream_encoder_dp_unblank(
+		struct dc_link *link,
+		struct stream_encoder *enc,
+		const struct encoder_unblank_param *param)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (param->link_settings.link_rate != LINK_RATE_UNKNOWN) {
+		uint32_t n_vid = 0x8000;
+		uint32_t m_vid;
+		uint32_t n_multiply = 0;
+		uint32_t pix_per_cycle = 0;
+		uint64_t m_vid_l = n_vid;
+
+		/* YCbCr 4:2:0 : Computed VID_M will be 2X the input rate */
+		if (is_two_pixels_per_containter(&param->timing) || param->opp_cnt > 1
+			|| param->pix_per_cycle > 1) {
+			/*this logic should be the same in get_pixel_clock_parameters() */
+			n_multiply = 1;
+			pix_per_cycle = 1;
+		}
+		/* M / N = Fstream / Flink
+		 * m_vid / n_vid = pixel rate / link rate
+		 */
+
+		m_vid_l *= param->timing.pix_clk_100hz / 10;
+		m_vid_l = div_u64(m_vid_l,
+			param->link_settings.link_rate
+				* LINK_RATE_REF_FREQ_IN_KHZ);
+
+		m_vid = (uint32_t) m_vid_l;
+
+		/* enable auto measurement */
+
+		REG_UPDATE(DP_VID_TIMING, DP_VID_M_N_GEN_EN, 0);
+
+		/* auto measurement need 1 full 0x8000 symbol cycle to kick in,
+		 * therefore program initial value for Mvid and Nvid
+		 */
+
+		REG_UPDATE(DP_VID_N, DP_VID_N, n_vid);
+
+		REG_UPDATE(DP_VID_M, DP_VID_M, m_vid);
+
+		REG_UPDATE_2(DP_VID_TIMING,
+				DP_VID_M_N_GEN_EN, 1,
+				DP_VID_N_MUL, n_multiply);
+
+		REG_UPDATE(DP_PIXEL_FORMAT,
+				DP_PIXEL_PER_CYCLE_PROCESSING_MODE,
+				pix_per_cycle);
+	}
+
+	/* make sure stream is disabled before resetting steer fifo */
+	REG_UPDATE(DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, false);
+	REG_WAIT(DP_VID_STREAM_CNTL, DP_VID_STREAM_STATUS, 0, 10, 5000);
+
+	/* DIG_START is removed from the register spec */
+
+	/* switch DP encoder to CRTC data, but reset it the fifo first. It may happen
+	 * that it overflows during mode transition, and sometimes doesn't recover.
+	 */
+	REG_UPDATE(DP_STEER_FIFO, DP_STEER_FIFO_RESET, 1);
+	udelay(10);
+
+	REG_UPDATE(DP_STEER_FIFO, DP_STEER_FIFO_RESET, 0);
+
+	/* wait 100us for DIG/DP logic to prime
+	 * (i.e. a few video lines)
+	 */
+	udelay(100);
+
+	/* the hardware would start sending video at the start of the next DP
+	 * frame (i.e. rising edge of the vblank).
+	 * NOTE: We used to program DP_VID_STREAM_DIS_DEFER = 2 here, but this
+	 * register has no effect on enable transition! HW always makes sure
+	 * VID_STREAM enable at start of next frame, and this is not
+	 * programmable
+	 */
+
+	REG_UPDATE(DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, true);
+
+	/*
+	 * DIG Resync FIFO now needs to be explicitly enabled.
+	 * This should come after DP_VID_STREAM_ENABLE per HW docs.
+	 */
+	enc314_enable_fifo(enc);
+
+	link->dc->link_srv->dp_trace_source_sequence(link, DPCD_SOURCE_SEQ_AFTER_ENABLE_DP_VID_STREAM);
+}
+
+static void enc35_stream_encoder_map_to_link(
+		struct stream_encoder *enc,
+		uint32_t stream_enc_inst,
+		uint32_t link_enc_inst)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	ASSERT(stream_enc_inst < 5 && link_enc_inst < 5);
+	REG_UPDATE(STREAM_MAPPER_CONTROL,
+				DIG_STREAM_LINK_TARGET, link_enc_inst);
+}
+
+static void enc35_reset_fifo(struct stream_encoder *enc, bool reset)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+	uint32_t reset_val = reset ? 1 : 0;
+	uint32_t is_symclk_on;
+
+	REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_RESET, reset_val);
+	REG_GET(DIG_FE_CLK_CNTL, DIG_FE_SYMCLK_FE_G_CLOCK_ON, &is_symclk_on);
+
+	if (is_symclk_on)
+		REG_WAIT(DIG_FIFO_CTRL0, DIG_FIFO_RESET_DONE, reset_val, 10, 5000);
+	else
+		udelay(10);
+}
+
+static bool enc35_is_fifo_enabled(struct stream_encoder *enc)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+	uint32_t reset_val;
+
+	REG_GET(DIG_FIFO_CTRL0, DIG_FIFO_ENABLE, &reset_val);
+	return reset_val != 0;
+}
+void enc35_disable_fifo(struct stream_encoder *enc)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_ENABLE, 0);
+	REG_UPDATE(DIG_FE_EN_CNTL, DIG_FE_ENABLE, 0);
+	REG_UPDATE(DIG_FE_CLK_CNTL, DIG_FE_CLK_EN, 0);
+}
+
+void enc35_enable_fifo(struct stream_encoder *enc)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_READ_START_LEVEL, 0x7);
+	REG_UPDATE(DIG_FE_CLK_CNTL, DIG_FE_CLK_EN, 1);
+	REG_UPDATE(DIG_FE_EN_CNTL, DIG_FE_ENABLE, 1);
+
+	enc35_reset_fifo(enc, true);
+	enc35_reset_fifo(enc, false);
+
+	REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_ENABLE, 1);
+}
+
+static uint32_t enc35_get_pixels_per_cycle(struct stream_encoder *enc)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+	uint32_t value;
+
+	REG_GET(DIG_FIFO_CTRL0, DIG_FIFO_OUTPUT_PIXEL_MODE, &value);
+
+	switch (value) {
+	case 0:
+		return 1;
+	case 1:
+		return 2;
+	default:
+		ASSERT_CRITICAL(false);
+		return 1;
+	}
+}
+
+static const struct stream_encoder_funcs dcn35_str_enc_funcs = {
+	.dp_set_odm_combine =
+		enc314_dp_set_odm_combine,
+	.dp_set_stream_attribute =
+		enc2_stream_encoder_dp_set_stream_attribute,
+	.hdmi_set_stream_attribute =
+		enc35_stream_encoder_hdmi_set_stream_attribute,
+	.dvi_set_stream_attribute =
+		enc35_stream_encoder_dvi_set_stream_attribute,
+	.set_throttled_vcp_size =
+		enc1_stream_encoder_set_throttled_vcp_size,
+	.update_hdmi_info_packets =
+		enc3_stream_encoder_update_hdmi_info_packets,
+	.stop_hdmi_info_packets =
+		enc3_stream_encoder_stop_hdmi_info_packets,
+	.update_dp_info_packets_sdp_line_num =
+		enc3_stream_encoder_update_dp_info_packets_sdp_line_num,
+	.update_dp_info_packets =
+		enc3_stream_encoder_update_dp_info_packets,
+	.stop_dp_info_packets =
+		enc1_stream_encoder_stop_dp_info_packets,
+	.dp_blank =
+		enc314_stream_encoder_dp_blank,
+	.dp_unblank =
+		enc35_stream_encoder_dp_unblank,
+	.audio_mute_control = enc3_audio_mute_control,
+
+	.dp_audio_setup = enc3_se_dp_audio_setup,
+	.dp_audio_enable = enc3_se_dp_audio_enable,
+	.dp_audio_disable = enc1_se_dp_audio_disable,
+
+	.hdmi_audio_setup = enc3_se_hdmi_audio_setup,
+	.hdmi_audio_disable = enc1_se_hdmi_audio_disable,
+	.setup_stereo_sync  = enc1_setup_stereo_sync,
+	.set_avmute = enc1_stream_encoder_set_avmute,
+	.dig_connect_to_otg = enc1_dig_connect_to_otg,
+	.dig_source_otg = enc1_dig_source_otg,
+
+	.dp_get_pixel_format  = enc1_stream_encoder_dp_get_pixel_format,
+
+	.enc_read_state = enc314_read_state,
+	.dp_set_dsc_config = enc314_dp_set_dsc_config,
+	.dp_set_dsc_pps_info_packet = enc3_dp_set_dsc_pps_info_packet,
+	.set_dynamic_metadata = enc2_set_dynamic_metadata,
+	.hdmi_reset_stream_attribute = enc1_reset_hdmi_stream_attribute,
+	.enable_stream = enc35_stream_encoder_enable,
+
+	.set_input_mode = enc314_set_dig_input_mode,
+	.enable_fifo = enc35_enable_fifo,
+	.disable_fifo = enc35_disable_fifo,
+	.is_fifo_enabled = enc35_is_fifo_enabled,
+	.map_stream_to_link = enc35_stream_encoder_map_to_link,
+	.get_pixels_per_cycle = enc35_get_pixels_per_cycle,
+};
+
+void dcn35_dio_stream_encoder_construct(
+	struct dcn10_stream_encoder *enc1,
+	struct dc_context *ctx,
+	struct dc_bios *bp,
+	enum engine_id eng_id,
+	struct vpg *vpg,
+	struct afmt *afmt,
+	const struct dcn10_stream_enc_registers *regs,
+	const struct dcn10_stream_encoder_shift *se_shift,
+	const struct dcn10_stream_encoder_mask *se_mask)
+{
+	enc1->base.funcs = &dcn35_str_enc_funcs;
+	enc1->base.ctx = ctx;
+	enc1->base.id = eng_id;
+	enc1->base.bp = bp;
+	enc1->base.vpg = vpg;
+	enc1->base.afmt = afmt;
+	enc1->regs = regs;
+	enc1->se_shift = se_shift;
+	enc1->se_mask = se_mask;
+	enc1->base.stream_enc_inst = vpg->inst;
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn35/dcn35_dio_stream_encoder.h b/drivers/gpu/drm/amd/display/dc/dio/dcn35/dcn35_dio_stream_encoder.h
new file mode 100644
index 000000000000..ddb33fdfb4ee
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn35/dcn35_dio_stream_encoder.h
@@ -0,0 +1,332 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __DC_DIO_STREAM_ENCODER_DCN35_H__
+#define __DC_DIO_STREAM_ENCODER_DCN35_H__
+
+#include "dcn30/dcn30_vpg.h"
+#include "dcn30/dcn30_afmt.h"
+#include "stream_encoder.h"
+#include "dcn20/dcn20_stream_encoder.h"
+
+/* Register bit field name change */
+#define RDPCSTX0_RDPCSTX_CLOCK_CNTL__RDPCS_SYMCLK_DIV2_GATE_DIS__SHIFT                                        0x8
+#define RDPCSTX0_RDPCSTX_CLOCK_CNTL__RDPCS_SYMCLK_DIV2_EN__SHIFT                                              0x9
+#define RDPCSTX0_RDPCSTX_CLOCK_CNTL__RDPCS_SYMCLK_DIV2_CLOCK_ON__SHIFT                                        0xa
+#define DPCSTX0_DPCSTX_TX_CNTL__DPCS_TX_DATA_SWAP__SHIFT                                                      0xe
+#define DPCSTX0_DPCSTX_TX_CNTL__DPCS_TX_DATA_ORDER_INVERT__SHIFT                                              0xf
+
+#define RDPCSTX0_RDPCSTX_CLOCK_CNTL__RDPCS_SYMCLK_DIV2_GATE_DIS_MASK                                          0x00000100L
+#define RDPCSTX0_RDPCSTX_CLOCK_CNTL__RDPCS_SYMCLK_DIV2_EN_MASK                                                0x00000200L
+#define RDPCSTX0_RDPCSTX_CLOCK_CNTL__RDPCS_SYMCLK_DIV2_CLOCK_ON_MASK                                          0x00000400L
+#define DPCSTX0_DPCSTX_TX_CNTL__DPCS_TX_DATA_SWAP_MASK                                                        0x00004000L
+#define DPCSTX0_DPCSTX_TX_CNTL__DPCS_TX_DATA_ORDER_INVERT_MASK                                                0x00008000L
+
+
+#define SE_DCN35_REG_LIST(id)\
+	SRI(AFMT_CNTL, DIG, id), \
+	SRI(DIG_FE_CNTL, DIG, id), \
+	SRI(HDMI_CONTROL, DIG, id), \
+	SRI(HDMI_DB_CONTROL, DIG, id), \
+	SRI(HDMI_GC, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL0, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL1, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL2, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL3, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL4, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL5, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL6, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL7, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL8, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL9, DIG, id), \
+	SRI(HDMI_GENERIC_PACKET_CONTROL10, DIG, id), \
+	SRI(HDMI_INFOFRAME_CONTROL0, DIG, id), \
+	SRI(HDMI_INFOFRAME_CONTROL1, DIG, id), \
+	SRI(HDMI_VBI_PACKET_CONTROL, DIG, id), \
+	SRI(HDMI_AUDIO_PACKET_CONTROL, DIG, id),\
+	SRI(HDMI_ACR_PACKET_CONTROL, DIG, id),\
+	SRI(HDMI_ACR_32_0, DIG, id),\
+	SRI(HDMI_ACR_32_1, DIG, id),\
+	SRI(HDMI_ACR_44_0, DIG, id),\
+	SRI(HDMI_ACR_44_1, DIG, id),\
+	SRI(HDMI_ACR_48_0, DIG, id),\
+	SRI(HDMI_ACR_48_1, DIG, id),\
+	SRI(DP_DB_CNTL, DP, id), \
+	SRI(DP_MSA_MISC, DP, id), \
+	SRI(DP_MSA_VBID_MISC, DP, id), \
+	SRI(DP_MSA_COLORIMETRY, DP, id), \
+	SRI(DP_MSA_TIMING_PARAM1, DP, id), \
+	SRI(DP_MSA_TIMING_PARAM2, DP, id), \
+	SRI(DP_MSA_TIMING_PARAM3, DP, id), \
+	SRI(DP_MSA_TIMING_PARAM4, DP, id), \
+	SRI(DP_MSE_RATE_CNTL, DP, id), \
+	SRI(DP_MSE_RATE_UPDATE, DP, id), \
+	SRI(DP_PIXEL_FORMAT, DP, id), \
+	SRI(DP_SEC_CNTL, DP, id), \
+	SRI(DP_SEC_CNTL1, DP, id), \
+	SRI(DP_SEC_CNTL2, DP, id), \
+	SRI(DP_SEC_CNTL5, DP, id), \
+	SRI(DP_SEC_CNTL6, DP, id), \
+	SRI(DP_STEER_FIFO, DP, id), \
+	SRI(DP_VID_M, DP, id), \
+	SRI(DP_VID_N, DP, id), \
+	SRI(DP_VID_STREAM_CNTL, DP, id), \
+	SRI(DP_VID_TIMING, DP, id), \
+	SRI(DP_SEC_AUD_N, DP, id), \
+	SRI(DP_SEC_TIMESTAMP, DP, id), \
+	SRI(DP_DSC_CNTL, DP, id), \
+	SRI(DP_SEC_METADATA_TRANSMISSION, DP, id), \
+	SRI(HDMI_METADATA_PACKET_CONTROL, DIG, id), \
+	SRI(DP_SEC_FRAMING4, DP, id), \
+	SRI(DP_GSP11_CNTL, DP, id), \
+	SRI(DME_CONTROL, DME, id),\
+	SRI(DP_SEC_METADATA_TRANSMISSION, DP, id), \
+	SRI(HDMI_METADATA_PACKET_CONTROL, DIG, id), \
+	SRI(DIG_FE_CNTL, DIG, id), \
+	SRI(DIG_FE_EN_CNTL, DIG, id), \
+	SRI(DIG_FE_CLK_CNTL, DIG, id), \
+	SRI(DIG_CLOCK_PATTERN, DIG, id), \
+	SRI(DIG_FIFO_CTRL0, DIG, id),\
+	SRI(STREAM_MAPPER_CONTROL, DIG, id)
+
+
+#define SE_COMMON_MASK_SH_LIST_DCN35(mask_sh)\
+	SE_SF(DP0_DP_PIXEL_FORMAT, DP_PIXEL_ENCODING, mask_sh),\
+	SE_SF(DP0_DP_PIXEL_FORMAT, DP_COMPONENT_DEPTH, mask_sh),\
+	SE_SF(DP0_DP_PIXEL_FORMAT, DP_PIXEL_PER_CYCLE_PROCESSING_MODE, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_PACKET_GEN_VERSION, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_KEEPOUT_MODE, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_DATA_SCRAMBLE_EN, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_NO_EXTRA_NULL_PACKET_FILLED, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_GC_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_GC_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_NULL_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_ACP_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_INFOFRAME_CONTROL1, HDMI_AUDIO_INFO_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GC, HDMI_GC_AVMUTE, mask_sh),\
+	SE_SF(DP0_DP_MSE_RATE_CNTL, DP_MSE_RATE_X, mask_sh),\
+	SE_SF(DP0_DP_MSE_RATE_CNTL, DP_MSE_RATE_Y, mask_sh),\
+	SE_SF(DP0_DP_MSE_RATE_UPDATE, DP_MSE_RATE_UPDATE_PENDING, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP0_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP1_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP2_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP3_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_MPG_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL1, DP_SEC_GSP5_LINE_REFERENCE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP4_SEND, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP4_SEND_PENDING, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL4, DP_SEC_GSP4_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL5, DP_SEC_GSP5_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP4_SEND_ANY_LINE, mask_sh),\
+	SE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_DIS_DEFER, mask_sh),\
+	SE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_STATUS, mask_sh),\
+	SE_SF(DP0_DP_STEER_FIFO, DP_STEER_FIFO_RESET, mask_sh),\
+	SE_SF(DP0_DP_VID_TIMING, DP_VID_M_N_GEN_EN, mask_sh),\
+	SE_SF(DP0_DP_VID_N, DP_VID_N, mask_sh),\
+	SE_SF(DP0_DP_VID_M, DP_VID_M, mask_sh),\
+	SE_SF(DIG0_HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_DELAY_EN, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUTO_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_PACKET_CONTROL, HDMI_ACR_SOURCE, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUDIO_PRIORITY, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_32_0, HDMI_ACR_CTS_32, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_32_1, HDMI_ACR_N_32, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_44_0, HDMI_ACR_CTS_44, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_44_1, HDMI_ACR_N_44, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_48_0, HDMI_ACR_CTS_48, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_48_1, HDMI_ACR_N_48, mask_sh),\
+	SE_SF(DP0_DP_SEC_AUD_N, DP_SEC_AUD_N, mask_sh),\
+	SE_SF(DP0_DP_SEC_TIMESTAMP, DP_SEC_TIMESTAMP_MODE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_ASP_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_ATP_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_AIP_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_ACM_ENABLE, mask_sh),\
+	SE_SF(DIG0_AFMT_CNTL, AFMT_AUDIO_CLOCK_EN, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_CLOCK_CHANNEL_RATE, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, TMDS_PIXEL_ENCODING, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, TMDS_COLOR_FORMAT, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_STEREOSYNC_SELECT, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_STEREOSYNC_GATE_EN, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP4_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP5_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP6_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP7_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP7_SEND, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL6, DP_SEC_GSP7_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP11_PPS, mask_sh),\
+	SE_SF(DP0_DP_GSP11_CNTL, DP_SEC_GSP11_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_GSP11_CNTL, DP_SEC_GSP11_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_DB_CNTL, DP_DB_DISABLE, mask_sh),\
+	SE_SF(DP0_DP_MSA_COLORIMETRY, DP_MSA_MISC0, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM1, DP_MSA_HTOTAL, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM1, DP_MSA_VTOTAL, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM2, DP_MSA_HSTART, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM2, DP_MSA_VSTART, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_HSYNCWIDTH, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_HSYNCPOLARITY, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_VSYNCWIDTH, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_VSYNCPOLARITY, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM4, DP_MSA_HWIDTH, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM4, DP_MSA_VHEIGHT, mask_sh),\
+	SE_SF(DIG0_HDMI_DB_CONTROL, HDMI_DB_DISABLE, mask_sh),\
+	SE_SF(DP0_DP_VID_TIMING, DP_VID_N_MUL, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_SOURCE_SELECT, mask_sh), \
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC0_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC0_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC1_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC1_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC2_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC2_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC3_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC3_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC4_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC4_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC5_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC5_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC6_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC6_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC7_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC7_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC8_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC8_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC9_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC9_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC10_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC10_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC11_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC11_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC12_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC12_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC13_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC13_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC14_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC14_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL1, HDMI_GENERIC0_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL1, HDMI_GENERIC1_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL2, HDMI_GENERIC2_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL2, HDMI_GENERIC3_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL3, HDMI_GENERIC4_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL3, HDMI_GENERIC5_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL4, HDMI_GENERIC6_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL4, HDMI_GENERIC7_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL7, HDMI_GENERIC8_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL7, HDMI_GENERIC9_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL8, HDMI_GENERIC10_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL8, HDMI_GENERIC11_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL9, HDMI_GENERIC12_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL9, HDMI_GENERIC13_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL10, HDMI_GENERIC14_LINE, mask_sh),\
+	SE_SF(DP0_DP_DSC_CNTL, DP_DSC_MODE, mask_sh),\
+	SE_SF(DP0_DP_MSA_VBID_MISC, DP_VBID6_LINE_REFERENCE, mask_sh),\
+	SE_SF(DP0_DP_MSA_VBID_MISC, DP_VBID6_LINE_NUM, mask_sh),\
+	SE_SF(DME0_DME_CONTROL, METADATA_ENGINE_EN, mask_sh),\
+	SE_SF(DME0_DME_CONTROL, METADATA_HUBP_REQUESTOR_ID, mask_sh),\
+	SE_SF(DME0_DME_CONTROL, METADATA_STREAM_TYPE, mask_sh),\
+	SE_SF(DP0_DP_SEC_METADATA_TRANSMISSION, DP_SEC_METADATA_PACKET_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_METADATA_TRANSMISSION, DP_SEC_METADATA_PACKET_LINE_REFERENCE, mask_sh),\
+	SE_SF(DP0_DP_SEC_METADATA_TRANSMISSION, DP_SEC_METADATA_PACKET_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_METADATA_PACKET_CONTROL, HDMI_METADATA_PACKET_ENABLE, mask_sh),\
+	SE_SF(DIG0_HDMI_METADATA_PACKET_CONTROL, HDMI_METADATA_PACKET_LINE_REFERENCE, mask_sh),\
+	SE_SF(DIG0_HDMI_METADATA_PACKET_CONTROL, HDMI_METADATA_PACKET_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, DOLBY_VISION_EN, mask_sh),\
+	SE_SF(DIG0_DIG_FE_EN_CNTL, DIG_FE_ENABLE, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CLK_CNTL, DIG_FE_MODE, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CLK_CNTL, DIG_FE_CLK_EN, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CLK_CNTL, DIG_FE_SOFT_RESET, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CLK_CNTL, DIG_FE_DISPCLK_G_CLOCK_ON, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CLK_CNTL, DIG_FE_SYMCLK_FE_G_CLOCK_ON, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CLK_CNTL, DIG_FE_SYMCLK_FE_G_AFMT_CLOCK_ON, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CLK_CNTL, DIG_FE_SYMCLK_FE_G_TMDS_CLOCK_ON, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CLK_CNTL, DIG_FE_SOCCLK_G_AFMT_CLOCK_ON, mask_sh),\
+	SE_SF(DP0_DP_SEC_FRAMING4, DP_SST_SDP_SPLITTING, mask_sh),\
+	SE_SF(DIG0_DIG_CLOCK_PATTERN, DIG_CLOCK_PATTERN, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_CTRL0, DIG_FIFO_OUTPUT_PIXEL_MODE, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_CTRL0, DIG_FIFO_READ_START_LEVEL, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_CTRL0, DIG_FIFO_ENABLE, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_CTRL0, DIG_FIFO_RESET, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_CTRL0, DIG_FIFO_RESET_DONE, mask_sh),\
+	SE_SF(DIG0_STREAM_MAPPER_CONTROL, DIG_STREAM_LINK_TARGET, mask_sh),
+
+void dcn35_dio_stream_encoder_construct(
+	struct dcn10_stream_encoder *enc1,
+	struct dc_context *ctx,
+	struct dc_bios *bp,
+	enum engine_id eng_id,
+	struct vpg *vpg,
+	struct afmt *afmt,
+	const struct dcn10_stream_enc_registers *regs,
+	const struct dcn10_stream_encoder_shift *se_shift,
+	const struct dcn10_stream_encoder_mask *se_mask);
+
+void enc3_stream_encoder_update_hdmi_info_packets(
+	struct stream_encoder *enc,
+	const struct encoder_info_frame *info_frame);
+
+void enc3_stream_encoder_stop_hdmi_info_packets(
+	struct stream_encoder *enc);
+
+void enc3_stream_encoder_update_dp_info_packets_sdp_line_num(
+		struct stream_encoder *enc,
+		struct encoder_info_frame *info_frame);
+
+void enc3_stream_encoder_update_dp_info_packets(
+	struct stream_encoder *enc,
+	const struct encoder_info_frame *info_frame);
+
+void enc3_audio_mute_control(
+	struct stream_encoder *enc,
+	bool mute);
+
+void enc3_se_dp_audio_setup(
+	struct stream_encoder *enc,
+	unsigned int az_inst,
+	struct audio_info *info);
+
+void enc3_se_dp_audio_enable(
+	struct stream_encoder *enc);
+
+void enc3_se_hdmi_audio_setup(
+	struct stream_encoder *enc,
+	unsigned int az_inst,
+	struct audio_info *info,
+	struct audio_crtc_info *audio_crtc_info);
+
+void enc3_dp_set_dsc_pps_info_packet(
+	struct stream_encoder *enc,
+	bool enable,
+	uint8_t *dsc_packed_pps,
+	bool immediate_update);
+
+void enc35_disable_fifo(
+	struct stream_encoder *enc);
+
+void enc35_enable_fifo(
+	struct stream_encoder *enc);
+
+
+#endif /* __DC_DIO_STREAM_ENCODER_DCN35_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn401/dcn401_dio_link_encoder.c b/drivers/gpu/drm/amd/display/dc/dio/dcn401/dcn401_dio_link_encoder.c
new file mode 100644
index 000000000000..7e558ca195ef
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn401/dcn401_dio_link_encoder.c
@@ -0,0 +1,322 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+#include "reg_helper.h"
+
+#include "core_types.h"
+#include "link_encoder.h"
+#include "dcn31/dcn31_dio_link_encoder.h"
+#include "dcn32/dcn32_dio_link_encoder.h"
+#include "dcn401_dio_link_encoder.h"
+#include "stream_encoder.h"
+#include "dc_bios_types.h"
+
+#include "gpio_service_interface.h"
+
+#ifndef MIN
+#define MIN(X, Y) ((X) < (Y) ? (X) : (Y))
+#endif
+
+#define CTX \
+	enc10->base.ctx
+#define DC_LOGGER \
+	enc10->base.ctx->logger
+
+#define REG(reg)\
+	(enc10->link_regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	enc10->link_shift->field_name, enc10->link_mask->field_name
+
+#define AUX_REG(reg)\
+	(enc10->aux_regs->reg)
+
+#define AUX_REG_READ(reg_name) \
+		dm_read_reg(CTX, AUX_REG(reg_name))
+
+#define AUX_REG_WRITE(reg_name, val) \
+			dm_write_reg(CTX, AUX_REG(reg_name), val)
+
+#ifndef MIN
+#define MIN(X, Y) ((X) < (Y) ? (X) : (Y))
+#endif
+
+void enc401_hw_init(struct link_encoder *enc)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+/*
+	00 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__1to2 : 1/2
+	01 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__3to4 : 3/4
+	02 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__7to8 : 7/8
+	03 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__15to16 : 15/16
+	04 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__31to32 : 31/32
+	05 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__63to64 : 63/64
+	06 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__127to128 : 127/128
+	07 - DP_AUX_DPHY_RX_DETECTION_THRESHOLD__255to256 : 255/256
+*/
+
+/*
+	AUX_REG_UPDATE_5(AUX_DPHY_RX_CONTROL0,
+	AUX_RX_START_WINDOW = 1 [6:4]
+	AUX_RX_RECEIVE_WINDOW = 1 default is 2 [10:8]
+	AUX_RX_HALF_SYM_DETECT_LEN  = 1 [13:12] default is 1
+	AUX_RX_TRANSITION_FILTER_EN = 1 [16] default is 1
+	AUX_RX_ALLOW_BELOW_THRESHOLD_PHASE_DETECT [17] is 0  default is 0
+	AUX_RX_ALLOW_BELOW_THRESHOLD_START [18] is 1  default is 1
+	AUX_RX_ALLOW_BELOW_THRESHOLD_STOP [19] is 1  default is 1
+	AUX_RX_PHASE_DETECT_LEN,  [21,20] = 0x3 default is 3
+	AUX_RX_DETECTION_THRESHOLD [30:28] = 1
+*/
+	AUX_REG_WRITE(AUX_DPHY_RX_CONTROL0, 0x103d1110);
+
+	AUX_REG_WRITE(AUX_DPHY_TX_CONTROL, 0x21c7a);
+
+	//AUX_DPHY_TX_REF_CONTROL'AUX_TX_REF_DIV HW default is 0x32;
+	// Set AUX_TX_REF_DIV Divider to generate 2 MHz reference from refclk
+	// 27MHz -> 0xd
+	// 100MHz -> 0x32
+	// 48MHz -> 0x18
+
+	// Set TMDS_CTL0 to 1.  This is a legacy setting.
+	REG_UPDATE(TMDS_CTL_BITS, TMDS_CTL0, 1);
+
+	dcn10_aux_initialize(enc10);
+}
+
+
+void dcn401_link_encoder_enable_dp_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source)
+{
+	if (!enc->ctx->dc->debug.avoid_vbios_exec_table) {
+		dcn10_link_encoder_enable_dp_output(enc, link_settings, clock_source);
+		return;
+	}
+}
+
+void dcn401_link_encoder_setup(
+	struct link_encoder *enc,
+	enum signal_type signal)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	switch (signal) {
+	case SIGNAL_TYPE_EDP:
+	case SIGNAL_TYPE_DISPLAY_PORT:
+		/* DP SST */
+		REG_UPDATE(DIG_BE_CLK_CNTL, DIG_BE_MODE, 0);
+		break;
+	case SIGNAL_TYPE_DVI_SINGLE_LINK:
+	case SIGNAL_TYPE_DVI_DUAL_LINK:
+		/* TMDS-DVI */
+		REG_UPDATE(DIG_BE_CLK_CNTL, DIG_BE_MODE, 2);
+		break;
+	case SIGNAL_TYPE_HDMI_TYPE_A:
+		/* TMDS-HDMI */
+		REG_UPDATE(DIG_BE_CLK_CNTL, DIG_BE_MODE, 3);
+		break;
+	case SIGNAL_TYPE_DISPLAY_PORT_MST:
+		/* DP MST */
+		REG_UPDATE(DIG_BE_CLK_CNTL, DIG_BE_MODE, 5);
+		break;
+	default:
+		ASSERT_CRITICAL(false);
+		/* invalid mode ! */
+		break;
+	}
+	REG_UPDATE(DIG_BE_CLK_CNTL, DIG_BE_CLK_EN, 1);
+	REG_UPDATE(DIG_BE_EN_CNTL, DIG_BE_ENABLE, 1);
+}
+
+bool dcn401_is_dig_enabled(struct link_encoder *enc)
+{
+	uint32_t clk_enabled;
+	uint32_t dig_enabled;
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+
+	REG_GET(DIG_BE_CLK_CNTL, DIG_BE_CLK_EN, &clk_enabled);
+	REG_GET(DIG_BE_EN_CNTL, DIG_BE_ENABLE, &dig_enabled);
+	return (clk_enabled == 1 && dig_enabled == 1);
+}
+
+enum signal_type dcn401_get_dig_mode(
+	struct link_encoder *enc)
+{
+	struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc);
+	uint32_t value;
+	REG_GET(DIG_BE_CLK_CNTL, DIG_BE_MODE, &value);
+	switch (value) {
+	case 0:
+		return SIGNAL_TYPE_DISPLAY_PORT;
+	case 2:
+		return SIGNAL_TYPE_DVI_SINGLE_LINK;
+	case 3:
+		return SIGNAL_TYPE_HDMI_TYPE_A;
+	case 5:
+		return SIGNAL_TYPE_DISPLAY_PORT_MST;
+	default:
+		return SIGNAL_TYPE_NONE;
+	}
+}
+
+static const struct link_encoder_funcs dcn401_link_enc_funcs = {
+	.read_state = link_enc2_read_state,
+	.validate_output_with_stream =
+			dcn30_link_encoder_validate_output_with_stream,
+	.hw_init = enc401_hw_init,
+	.setup = dcn401_link_encoder_setup,
+	.enable_tmds_output = dcn10_link_encoder_enable_tmds_output,
+	.enable_dp_output = dcn401_link_encoder_enable_dp_output,
+	.enable_dp_mst_output = dcn10_link_encoder_enable_dp_mst_output,
+	.disable_output = dcn10_link_encoder_disable_output,
+	.dp_set_lane_settings = dcn10_link_encoder_dp_set_lane_settings,
+	.dp_set_phy_pattern = dcn10_link_encoder_dp_set_phy_pattern,
+	.update_mst_stream_allocation_table =
+		dcn10_link_encoder_update_mst_stream_allocation_table,
+	.psr_program_dp_dphy_fast_training =
+			dcn10_psr_program_dp_dphy_fast_training,
+	.psr_program_secondary_packet = dcn10_psr_program_secondary_packet,
+	.connect_dig_be_to_fe = dcn10_link_encoder_connect_dig_be_to_fe,
+	.enable_hpd = dcn10_link_encoder_enable_hpd,
+	.disable_hpd = dcn10_link_encoder_disable_hpd,
+	.is_dig_enabled = dcn401_is_dig_enabled,
+	.destroy = dcn10_link_encoder_destroy,
+	.fec_set_enable = enc2_fec_set_enable,
+	.fec_set_ready = enc2_fec_set_ready,
+	.fec_is_active = enc2_fec_is_active,
+	.get_dig_frontend = dcn10_get_dig_frontend,
+	.get_dig_mode = dcn401_get_dig_mode,
+	.is_in_alt_mode = dcn32_link_encoder_is_in_alt_mode,
+	.get_max_link_cap = dcn32_link_encoder_get_max_link_cap,
+	.set_dio_phy_mux = dcn31_link_encoder_set_dio_phy_mux,
+};
+
+void dcn401_link_encoder_construct(
+	struct dcn20_link_encoder *enc20,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	const struct dcn10_link_enc_aux_registers *aux_regs,
+	const struct dcn10_link_enc_hpd_registers *hpd_regs,
+	const struct dcn10_link_enc_shift *link_shift,
+	const struct dcn10_link_enc_mask *link_mask)
+{
+	struct bp_connector_speed_cap_info bp_cap_info = {0};
+	const struct dc_vbios_funcs *bp_funcs = init_data->ctx->dc_bios->funcs;
+	enum bp_result result = BP_RESULT_OK;
+	struct dcn10_link_encoder *enc10 = &enc20->enc10;
+
+	enc10->base.funcs = &dcn401_link_enc_funcs;
+	enc10->base.ctx = init_data->ctx;
+	enc10->base.id = init_data->encoder;
+
+	enc10->base.hpd_source = init_data->hpd_source;
+	enc10->base.connector = init_data->connector;
+
+
+	enc10->base.preferred_engine = ENGINE_ID_UNKNOWN;
+
+	enc10->base.features = *enc_features;
+	if (enc10->base.connector.id == CONNECTOR_ID_USBC)
+		enc10->base.features.flags.bits.DP_IS_USB_C = 1;
+
+	enc10->base.transmitter = init_data->transmitter;
+
+	/* set the flag to indicate whether driver poll the I2C data pin
+	 * while doing the DP sink detect
+	 */
+
+/*	if (dal_adapter_service_is_feature_supported(as,
+		FEATURE_DP_SINK_DETECT_POLL_DATA_PIN))
+		enc10->base.features.flags.bits.
+			DP_SINK_DETECT_POLL_DATA_PIN = true;*/
+
+	enc10->base.output_signals =
+		SIGNAL_TYPE_DVI_SINGLE_LINK |
+		SIGNAL_TYPE_DVI_DUAL_LINK |
+		SIGNAL_TYPE_LVDS |
+		SIGNAL_TYPE_DISPLAY_PORT |
+		SIGNAL_TYPE_DISPLAY_PORT_MST |
+		SIGNAL_TYPE_EDP |
+		SIGNAL_TYPE_HDMI_TYPE_A;
+
+	enc10->link_regs = link_regs;
+	enc10->aux_regs = aux_regs;
+	enc10->hpd_regs = hpd_regs;
+	enc10->link_shift = link_shift;
+	enc10->link_mask = link_mask;
+
+	switch (enc10->base.transmitter) {
+	case TRANSMITTER_UNIPHY_A:
+		enc10->base.preferred_engine = ENGINE_ID_DIGA;
+	break;
+	case TRANSMITTER_UNIPHY_B:
+		enc10->base.preferred_engine = ENGINE_ID_DIGB;
+	break;
+	case TRANSMITTER_UNIPHY_C:
+		enc10->base.preferred_engine = ENGINE_ID_DIGC;
+	break;
+	case TRANSMITTER_UNIPHY_D:
+		enc10->base.preferred_engine = ENGINE_ID_DIGD;
+	break;
+	case TRANSMITTER_UNIPHY_E:
+		enc10->base.preferred_engine = ENGINE_ID_DIGE;
+	break;
+	default:
+		ASSERT_CRITICAL(false);
+		enc10->base.preferred_engine = ENGINE_ID_UNKNOWN;
+	}
+
+	/* default to one to mirror Windows behavior */
+	enc10->base.features.flags.bits.HDMI_6GB_EN = 1;
+
+	if (bp_funcs->get_connector_speed_cap_info)
+		result = bp_funcs->get_connector_speed_cap_info(enc10->base.ctx->dc_bios,
+						enc10->base.connector, &bp_cap_info);
+
+	/* Override features with DCE-specific values */
+	if (result == BP_RESULT_OK) {
+		enc10->base.features.flags.bits.IS_HBR2_CAPABLE =
+				bp_cap_info.DP_HBR2_EN;
+		enc10->base.features.flags.bits.IS_HBR3_CAPABLE =
+				bp_cap_info.DP_HBR3_EN;
+		enc10->base.features.flags.bits.HDMI_6GB_EN = bp_cap_info.HDMI_6GB_EN;
+		enc10->base.features.flags.bits.IS_DP2_CAPABLE = 1;
+		enc10->base.features.flags.bits.IS_UHBR10_CAPABLE = bp_cap_info.DP_UHBR10_EN;
+		enc10->base.features.flags.bits.IS_UHBR13_5_CAPABLE = bp_cap_info.DP_UHBR13_5_EN;
+		enc10->base.features.flags.bits.IS_UHBR20_CAPABLE = bp_cap_info.DP_UHBR20_EN;
+	} else {
+		DC_LOG_WARNING("%s: Failed to get encoder_cap_info from VBIOS with error code %d!\n",
+				__func__,
+				result);
+	}
+	if (enc10->base.ctx->dc->debug.hdmi20_disable) {
+		enc10->base.features.flags.bits.HDMI_6GB_EN = 0;
+	}
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn401/dcn401_dio_link_encoder.h b/drivers/gpu/drm/amd/display/dc/dio/dcn401/dcn401_dio_link_encoder.h
new file mode 100644
index 000000000000..6baab8302b81
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn401/dcn401_dio_link_encoder.h
@@ -0,0 +1,134 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_LINK_ENCODER__DCN401_H__
+#define __DC_LINK_ENCODER__DCN401_H__
+
+#include "dcn30/dcn30_dio_link_encoder.h"
+
+#define LINK_ENCODER_MASK_SH_LIST_DCN401(mask_sh) \
+	LE_SF(DIG0_DIG_BE_EN_CNTL, DIG_BE_ENABLE, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CNTL, DIG_RB_SWITCH_EN, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CNTL, DIG_HPD_SELECT, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CNTL, DIG_FE_SOURCE_SELECT, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CLK_CNTL, DIG_BE_MODE, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CLK_CNTL, DIG_BE_CLK_EN, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CLK_CNTL, DIG_BE_SOFT_RESET, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CLK_CNTL, HDCP_SOFT_RESET, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CLK_CNTL, DIG_BE_SYMCLK_G_CLOCK_ON, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CLK_CNTL, DIG_BE_SYMCLK_G_HDCP_CLOCK_ON, mask_sh),\
+	LE_SF(DIG0_DIG_BE_CLK_CNTL, DIG_BE_SYMCLK_G_TMDS_CLOCK_ON, mask_sh),\
+	LE_SF(DIG0_DIG_CLOCK_PATTERN, DIG_CLOCK_PATTERN, mask_sh),\
+	LE_SF(DIG0_TMDS_CTL_BITS, TMDS_CTL0, mask_sh), \
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_BYPASS, mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_ATEST_SEL_LANE0, mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_ATEST_SEL_LANE1, mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_ATEST_SEL_LANE2, mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_ATEST_SEL_LANE3, mask_sh),\
+	LE_SF(DP0_DP_DPHY_PRBS_CNTL, DPHY_PRBS_EN, mask_sh),\
+	LE_SF(DP0_DP_DPHY_PRBS_CNTL, DPHY_PRBS_SEL, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM0, DPHY_SYM1, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM0, DPHY_SYM2, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM0, DPHY_SYM3, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM1, DPHY_SYM4, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM1, DPHY_SYM5, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM1, DPHY_SYM6, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM2, DPHY_SYM7, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SYM2, DPHY_SYM8, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SCRAM_CNTL, DPHY_SCRAMBLER_BS_COUNT, mask_sh),\
+	LE_SF(DP0_DP_DPHY_SCRAM_CNTL, DPHY_SCRAMBLER_ADVANCE, mask_sh),\
+	LE_SF(DP0_DP_DPHY_FAST_TRAINING, DPHY_RX_FAST_TRAINING_CAPABLE, mask_sh),\
+	LE_SF(DP0_DP_DPHY_BS_SR_SWAP_CNTL, DPHY_LOAD_BS_COUNT, mask_sh),\
+	LE_SF(DP0_DP_DPHY_TRAINING_PATTERN_SEL, DPHY_TRAINING_PATTERN_SEL, mask_sh),\
+	LE_SF(DP0_DP_DPHY_HBR2_PATTERN_CONTROL, DP_DPHY_HBR2_PATTERN_CONTROL, mask_sh),\
+	LE_SF(DP0_DP_LINK_CNTL, DP_LINK_TRAINING_COMPLETE, mask_sh),\
+	LE_SF(DP0_DP_LINK_FRAMING_CNTL, DP_IDLE_BS_INTERVAL, mask_sh),\
+	LE_SF(DP0_DP_LINK_FRAMING_CNTL, DP_VBID_DISABLE, mask_sh),\
+	LE_SF(DP0_DP_LINK_FRAMING_CNTL, DP_VID_ENHANCED_FRAME_MODE, mask_sh),\
+	LE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, mask_sh),\
+	LE_SF(DP0_DP_CONFIG, DP_UDI_LANES, mask_sh),\
+	LE_SF(DP0_DP_SEC_CNTL1, DP_SEC_GSP0_LINE_NUM, mask_sh),\
+	LE_SF(DP0_DP_SEC_CNTL1, DP_SEC_GSP0_PRIORITY, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT0, DP_MSE_SAT_SRC0, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT0, DP_MSE_SAT_SRC1, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT0, DP_MSE_SAT_SLOT_COUNT0, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT0, DP_MSE_SAT_SLOT_COUNT1, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT1, DP_MSE_SAT_SRC2, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT1, DP_MSE_SAT_SRC3, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT1, DP_MSE_SAT_SLOT_COUNT2, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT1, DP_MSE_SAT_SLOT_COUNT3, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT_UPDATE, DP_MSE_SAT_UPDATE, mask_sh),\
+	LE_SF(DP0_DP_MSE_SAT_UPDATE, DP_MSE_16_MTP_KEEPOUT, mask_sh),\
+	LE_SF(DP_AUX0_AUX_CONTROL, AUX_HPD_SEL, mask_sh),\
+	LE_SF(DP_AUX0_AUX_CONTROL, AUX_LS_READ_EN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_RECEIVE_WINDOW, mask_sh),\
+	LE_SF(HPD0_DC_HPD_CONTROL, DC_HPD_EN, mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_FEC_EN, mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_FEC_READY_SHADOW, mask_sh),\
+	LE_SF(DP0_DP_DPHY_CNTL, DPHY_FEC_ACTIVE_STATUS, mask_sh),\
+	LE_SF(DIG0_TMDS_CTL_BITS, TMDS_CTL0, mask_sh), \
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_START_WINDOW, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_HALF_SYM_DETECT_LEN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_TRANSITION_FILTER_EN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_ALLOW_BELOW_THRESHOLD_PHASE_DETECT, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_ALLOW_BELOW_THRESHOLD_START, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_ALLOW_BELOW_THRESHOLD_STOP, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_PHASE_DETECT_LEN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_DETECTION_THRESHOLD, mask_sh), \
+	LE_SF(DP_AUX0_AUX_DPHY_TX_CONTROL, AUX_TX_PRECHARGE_LEN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_TX_CONTROL, AUX_TX_PRECHARGE_SYMBOLS, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_TX_CONTROL, AUX_MODE_DET_CHECK_DELAY, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL1, AUX_RX_PRECHARGE_SKIP, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL1, AUX_RX_TIMEOUT_LEN, mask_sh),\
+	LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL1, AUX_RX_TIMEOUT_LEN_MUL, mask_sh)
+
+void dcn401_link_encoder_construct(
+	struct dcn20_link_encoder *enc20,
+	const struct encoder_init_data *init_data,
+	const struct encoder_feature_support *enc_features,
+	const struct dcn10_link_enc_registers *link_regs,
+	const struct dcn10_link_enc_aux_registers *aux_regs,
+	const struct dcn10_link_enc_hpd_registers *hpd_regs,
+	const struct dcn10_link_enc_shift *link_shift,
+	const struct dcn10_link_enc_mask *link_mask);
+
+void enc401_hw_init(struct link_encoder *enc);
+
+void dcn401_link_encoder_enable_dp_output(
+	struct link_encoder *enc,
+	const struct dc_link_settings *link_settings,
+	enum clock_source_id clock_source);
+
+void dcn401_link_encoder_setup(
+	struct link_encoder *enc,
+	enum signal_type signal);
+
+enum signal_type dcn401_get_dig_mode(
+	struct link_encoder *enc);
+
+bool dcn401_is_dig_enabled(struct link_encoder *enc);
+
+enum signal_type dcn401_get_dig_mode(struct link_encoder *enc);
+#endif /* __DC_LINK_ENCODER__DCN401_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn401/dcn401_dio_stream_encoder.c b/drivers/gpu/drm/amd/display/dc/dio/dcn401/dcn401_dio_stream_encoder.c
new file mode 100644
index 000000000000..99aab70ef3e1
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn401/dcn401_dio_stream_encoder.c
@@ -0,0 +1,856 @@
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+#include "dc_bios_types.h"
+#include "dcn30/dcn30_dio_stream_encoder.h"
+#include "dcn32/dcn32_dio_stream_encoder.h"
+#include "dcn35/dcn35_dio_stream_encoder.h"
+
+#include "dcn401_dio_stream_encoder.h"
+#include "reg_helper.h"
+#include "hw_shared.h"
+#include "link_service.h"
+#include "dpcd_defs.h"
+
+#define DC_LOGGER \
+		enc1->base.ctx->logger
+
+#define REG(reg)\
+	(enc1->regs->reg)
+
+#undef FN
+#define FN(reg_name, field_name) \
+	enc1->se_shift->field_name, enc1->se_mask->field_name
+
+#define VBI_LINE_0 0
+#define HDMI_CLOCK_CHANNEL_RATE_MORE_340M 340000
+
+#define CTX \
+	enc1->base.ctx
+
+
+
+static void enc401_dp_set_odm_combine(
+	struct stream_encoder *enc,
+	bool odm_combine)
+{
+}
+
+/* setup stream encoder in dvi mode */
+void enc401_stream_encoder_dvi_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	bool is_dual_link)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (!enc->ctx->dc->debug.avoid_vbios_exec_table) {
+		struct bp_encoder_control cntl = {0};
+
+		cntl.action = ENCODER_CONTROL_SETUP;
+		cntl.engine_id = enc1->base.id;
+		cntl.signal = is_dual_link ?
+			SIGNAL_TYPE_DVI_DUAL_LINK : SIGNAL_TYPE_DVI_SINGLE_LINK;
+		cntl.enable_dp_audio = false;
+		cntl.pixel_clock = crtc_timing->pix_clk_100hz / 10;
+		cntl.lanes_number = (is_dual_link) ? LANE_COUNT_EIGHT : LANE_COUNT_FOUR;
+
+		if (enc1->base.bp->funcs->encoder_control(
+				enc1->base.bp, &cntl) != BP_RESULT_OK)
+			return;
+
+	} else {
+
+		//Set pattern for clock channel, default vlue 0x63 does not work
+		REG_UPDATE(DIG_CLOCK_PATTERN, DIG_CLOCK_PATTERN, 0x1F);
+
+		//DIG_BE_TMDS_DVI_MODE : TMDS-DVI mode is already set in link_encoder_setup
+
+		//DIG_SOURCE_SELECT is already set in dig_connect_to_otg
+
+		/* DIG_START is removed from the register spec */
+	}
+
+	ASSERT(crtc_timing->pixel_encoding == PIXEL_ENCODING_RGB);
+	ASSERT(crtc_timing->display_color_depth == COLOR_DEPTH_888);
+	enc401_stream_encoder_set_stream_attribute_helper(enc1, crtc_timing);
+}
+
+/* setup stream encoder in hdmi mode */
+void enc401_stream_encoder_hdmi_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	int actual_pix_clk_khz,
+	bool enable_audio)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (!enc->ctx->dc->debug.avoid_vbios_exec_table) {
+		struct bp_encoder_control cntl = {0};
+
+		cntl.action = ENCODER_CONTROL_SETUP;
+		cntl.engine_id = enc1->base.id;
+		cntl.signal = SIGNAL_TYPE_HDMI_TYPE_A;
+		cntl.enable_dp_audio = enable_audio;
+		cntl.pixel_clock = actual_pix_clk_khz;
+		cntl.lanes_number = LANE_COUNT_FOUR;
+
+		if (enc1->base.bp->funcs->encoder_control(
+				enc1->base.bp, &cntl) != BP_RESULT_OK)
+			return;
+
+	} else {
+
+		//Set pattern for clock channel, default vlue 0x63 does not work
+		REG_UPDATE(DIG_CLOCK_PATTERN, DIG_CLOCK_PATTERN, 0x1F);
+
+		//DIG_BE_TMDS_HDMI_MODE : TMDS-HDMI mode is already set in link_encoder_setup
+
+		//DIG_SOURCE_SELECT is already set in dig_connect_to_otg
+
+		/* DIG_START is removed from the register spec */
+	}
+
+	/* Configure pixel encoding */
+	enc401_stream_encoder_set_stream_attribute_helper(enc1, crtc_timing);
+
+	/* setup HDMI engine */
+	REG_UPDATE_6(HDMI_CONTROL,
+		HDMI_PACKET_GEN_VERSION, 1,
+		HDMI_KEEPOUT_MODE, 1,
+		HDMI_DEEP_COLOR_ENABLE, 0,
+		HDMI_DATA_SCRAMBLE_EN, 0,
+		HDMI_NO_EXTRA_NULL_PACKET_FILLED, 1,
+		HDMI_CLOCK_CHANNEL_RATE, 0);
+
+	/* Configure color depth */
+	switch (crtc_timing->display_color_depth) {
+	case COLOR_DEPTH_888:
+		REG_UPDATE(HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 0);
+		break;
+	case COLOR_DEPTH_101010:
+		if (crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR422) {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 1,
+					HDMI_DEEP_COLOR_ENABLE, 0);
+		} else {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 1,
+					HDMI_DEEP_COLOR_ENABLE, 1);
+			}
+		break;
+	case COLOR_DEPTH_121212:
+		if (crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR422) {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 2,
+					HDMI_DEEP_COLOR_ENABLE, 0);
+		} else {
+			REG_UPDATE_2(HDMI_CONTROL,
+					HDMI_DEEP_COLOR_DEPTH, 2,
+					HDMI_DEEP_COLOR_ENABLE, 1);
+			}
+		break;
+	case COLOR_DEPTH_161616:
+		REG_UPDATE_2(HDMI_CONTROL,
+				HDMI_DEEP_COLOR_DEPTH, 3,
+				HDMI_DEEP_COLOR_ENABLE, 1);
+		break;
+	default:
+		break;
+	}
+
+	if (actual_pix_clk_khz >= HDMI_CLOCK_CHANNEL_RATE_MORE_340M) {
+		/* enable HDMI data scrambler
+		 * HDMI_CLOCK_CHANNEL_RATE_MORE_340M
+		 * Clock channel frequency is 1/4 of character rate.
+		 */
+		REG_UPDATE_2(HDMI_CONTROL,
+			HDMI_DATA_SCRAMBLE_EN, 1,
+			HDMI_CLOCK_CHANNEL_RATE, 1);
+	} else if (crtc_timing->flags.LTE_340MCSC_SCRAMBLE) {
+
+		/* TODO: New feature for DCE11, still need to implement */
+
+		/* enable HDMI data scrambler
+		 * HDMI_CLOCK_CHANNEL_FREQ_EQUAL_TO_CHAR_RATE
+		 * Clock channel frequency is the same
+		 * as character rate
+		 */
+		REG_UPDATE_2(HDMI_CONTROL,
+			HDMI_DATA_SCRAMBLE_EN, 1,
+			HDMI_CLOCK_CHANNEL_RATE, 0);
+	}
+
+
+	/* Enable transmission of General Control packet on every frame */
+	REG_UPDATE_3(HDMI_VBI_PACKET_CONTROL,
+		HDMI_GC_CONT, 1,
+		HDMI_GC_SEND, 1,
+		HDMI_NULL_SEND, 1);
+
+	/* Disable Audio Content Protection packet transmission */
+	REG_UPDATE(HDMI_VBI_PACKET_CONTROL, HDMI_ACP_SEND, 0);
+	/* following belongs to audio */
+	/* Enable Audio InfoFrame packet transmission. */
+	REG_UPDATE(HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, 1);
+
+	/* update double-buffered AUDIO_INFO registers immediately */
+	ASSERT(enc->afmt);
+	enc->afmt->funcs->audio_info_immediate_update(enc->afmt);
+
+	/* Select line number on which to send Audio InfoFrame packets */
+	REG_UPDATE(HDMI_INFOFRAME_CONTROL1, HDMI_AUDIO_INFO_LINE,
+				VBI_LINE_0 + 2);
+
+	/* set HDMI GC AVMUTE */
+	REG_UPDATE(HDMI_GC, HDMI_GC_AVMUTE, 0);
+}
+
+void enc401_set_dig_input_mode(struct stream_encoder *enc, unsigned int pix_per_container)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	// The naming of this field is confusing, what it means is the output mode of otg, which
+	// is the input mode of the dig
+	switch (pix_per_container)	{
+	case 2:
+		REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_OUTPUT_PIXEL_PER_CYCLE, 0x1);
+		break;
+	case 4:
+		REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_OUTPUT_PIXEL_PER_CYCLE, 0x2);
+		break;
+	case 8:
+		REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_OUTPUT_PIXEL_PER_CYCLE, 0x3);
+		break;
+	default:
+		REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_OUTPUT_PIXEL_PER_CYCLE, 0x0);
+		break;
+	}
+}
+
+static bool is_two_pixels_per_containter(const struct dc_crtc_timing *timing)
+{
+	bool two_pix = timing->pixel_encoding == PIXEL_ENCODING_YCBCR420;
+
+	two_pix = two_pix || (timing->flags.DSC && timing->pixel_encoding == PIXEL_ENCODING_YCBCR422
+			&& !timing->dsc_cfg.ycbcr422_simple);
+	return two_pix;
+}
+
+void enc401_stream_encoder_dp_unblank(
+		struct dc_link *link,
+		struct stream_encoder *enc,
+		const struct encoder_unblank_param *param)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (param->link_settings.link_rate != LINK_RATE_UNKNOWN) {
+		uint32_t n_vid = 0x8000;
+		uint32_t m_vid;
+		uint32_t pix_per_container = 1;
+		uint64_t m_vid_l = n_vid;
+
+		/* YCbCr 4:2:0 or YCbCr4:2:2 simple + DSC: Computed VID_M will be 2X the input rate */
+		if (is_two_pixels_per_containter(&param->timing)) {
+			pix_per_container = 2;
+		}
+
+		/* M / N = Fstream / Flink
+		 * m_vid / n_vid = pixel rate / link rate
+		 */
+		m_vid_l *= param->timing.pix_clk_100hz / pix_per_container / 10;
+		m_vid_l = div_u64(m_vid_l,
+			param->link_settings.link_rate
+				* LINK_RATE_REF_FREQ_IN_KHZ);
+
+		m_vid = (uint32_t) m_vid_l;
+
+		/* enable auto measurement */
+
+		REG_UPDATE(DP_VID_TIMING, DP_VID_M_N_GEN_EN, 0);
+
+		/* auto measurement need 1 full 0x8000 symbol cycle to kick in,
+		 * therefore program initial value for Mvid and Nvid
+		 */
+
+		REG_UPDATE(DP_VID_N, DP_VID_N, n_vid);
+
+		REG_UPDATE(DP_VID_M, DP_VID_M, m_vid);
+
+		/* reduce jitter based on read rate */
+		switch (param->pix_per_cycle)	{
+		case 2:
+			REG_UPDATE(DP_VID_TIMING, DP_VID_N_INTERVAL, 0x1);
+			break;
+		case 4:
+			REG_UPDATE(DP_VID_TIMING, DP_VID_N_INTERVAL, 0x2);
+			break;
+		case 8:
+			REG_UPDATE(DP_VID_TIMING, DP_VID_N_INTERVAL, 0x3);
+			break;
+		default:
+			REG_UPDATE(DP_VID_TIMING, DP_VID_N_INTERVAL, 0x0);
+			break;
+		}
+
+		REG_UPDATE(DP_VID_TIMING, DP_VID_M_N_GEN_EN, 1);
+	}
+
+	/* make sure stream is disabled before resetting steer fifo */
+	REG_UPDATE(DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, false);
+	REG_WAIT(DP_VID_STREAM_CNTL, DP_VID_STREAM_STATUS, 0, 10, 5000);
+
+	/* DIG_START is removed from the register spec */
+
+	/* switch DP encoder to CRTC data, but reset it the fifo first. It may happen
+	 * that it overflows during mode transition, and sometimes doesn't recover.
+	 */
+	REG_UPDATE(DP_STEER_FIFO, DP_STEER_FIFO_RESET, 1);
+	udelay(10);
+
+	REG_UPDATE(DP_STEER_FIFO, DP_STEER_FIFO_RESET, 0);
+
+	REG_UPDATE(DP_STEER_FIFO, DP_STEER_FIFO_ENABLE, 1);
+
+	REG_UPDATE_2(DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, 1, DP_VID_STREAM_DIS_DEFER, 2);
+	udelay(200);
+
+	/* DIG Resync FIFO now needs to be explicitly enabled
+	 */
+	/* read start level = 0 will bring underflow / overflow and DIG_FIFO_ERROR = 1
+	 * so set it to 1/2 full = 7 before reset as suggested by hardware team.
+	 */
+	REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_READ_START_LEVEL, 0x7);
+
+	REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_RESET, 1);
+
+	REG_WAIT(DIG_FIFO_CTRL0, DIG_FIFO_RESET_DONE, 1, 10, 5000);
+
+	REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_RESET, 0);
+
+	REG_WAIT(DIG_FIFO_CTRL0, DIG_FIFO_RESET_DONE, 0, 10, 5000);
+
+	REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_ENABLE, 1);
+
+	/* wait 100us for DIG/DP logic to prime
+	 * (i.e. a few video lines)
+	 */
+	udelay(100);
+
+	/* the hardware would start sending video at the start of the next DP
+	 * frame (i.e. rising edge of the vblank).
+	 * NOTE: We used to program DP_VID_STREAM_DIS_DEFER = 2 here, but this
+	 * register has no effect on enable transition! HW always guarantees
+	 * VID_STREAM enable at start of next frame, and this is not
+	 * programmable
+	 */
+
+	REG_UPDATE(DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, true);
+
+	link->dc->link_srv->dp_trace_source_sequence(link, DPCD_SOURCE_SEQ_AFTER_ENABLE_DP_VID_STREAM);
+}
+
+/* this function read dsc related register fields to be logged later in dcn10_log_hw_state
+ * into a dcn_dsc_state struct.
+ */
+void enc401_read_state(struct stream_encoder *enc, struct enc_state *s)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	//if dsc is enabled, continue to read
+	REG_GET(DP_PIXEL_FORMAT, PIXEL_ENCODING_TYPE, &s->dsc_mode);
+
+	if (s->dsc_mode) {
+		REG_GET(DP_GSP11_CNTL, DP_SEC_GSP11_LINE_NUM, &s->sec_gsp_pps_line_num);
+
+		REG_GET(DP_MSA_VBID_MISC, DP_VBID6_LINE_REFERENCE, &s->vbid6_line_reference);
+		REG_GET(DP_MSA_VBID_MISC, DP_VBID6_LINE_NUM, &s->vbid6_line_num);
+
+		REG_GET(DP_GSP11_CNTL, DP_SEC_GSP11_ENABLE, &s->sec_gsp_pps_enable);
+		REG_GET(DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, &s->sec_stream_enable);
+	}
+}
+
+void enc401_stream_encoder_enable(
+	struct stream_encoder *enc,
+	enum signal_type signal,
+	bool enable)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+	if (enable) {
+		switch (signal) {
+		case SIGNAL_TYPE_DVI_SINGLE_LINK:
+		case SIGNAL_TYPE_DVI_DUAL_LINK:
+			/* TMDS-DVI */
+			REG_UPDATE(DIG_FE_CLK_CNTL, DIG_FE_MODE, 2);
+			break;
+		case SIGNAL_TYPE_HDMI_TYPE_A:
+			/* TMDS-HDMI */
+			REG_UPDATE(DIG_FE_CLK_CNTL, DIG_FE_MODE, 3);
+			break;
+		case SIGNAL_TYPE_DISPLAY_PORT_MST:
+			/* DP MST */
+			REG_UPDATE(DIG_FE_CLK_CNTL, DIG_FE_MODE, 5);
+			break;
+		case SIGNAL_TYPE_EDP:
+		case SIGNAL_TYPE_DISPLAY_PORT:
+		case SIGNAL_TYPE_VIRTUAL:
+			/* DP SST */
+			REG_UPDATE(DIG_FE_CLK_CNTL, DIG_FE_MODE, 0);
+			break;
+		default:
+			/* invalid mode ! */
+			ASSERT_CRITICAL(false);
+		}
+
+		REG_UPDATE(DIG_FE_CLK_CNTL, DIG_FE_CLK_EN, 1);
+		REG_UPDATE(DIG_FE_EN_CNTL, DIG_FE_ENABLE, 1);
+	} else {
+		REG_UPDATE(DIG_FE_EN_CNTL, DIG_FE_ENABLE, 0);
+		REG_UPDATE(DIG_FE_CLK_CNTL, DIG_FE_CLK_EN, 0);
+	}
+}
+
+void enc401_stream_encoder_dp_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	enum dc_color_space output_color_space,
+	bool use_vsc_sdp_for_colorimetry,
+	uint32_t enable_sdp_splitting)
+{
+	uint32_t h_active_start;
+	uint32_t v_active_start;
+	uint32_t misc0 = 0;
+	uint32_t misc1 = 0;
+	uint32_t h_blank;
+	uint32_t h_back_porch;
+	uint8_t colorimetry_bpc;
+	uint8_t dp_pixel_encoding = 0;
+	uint8_t dp_component_depth = 0;
+	uint8_t dp_translate_pixel_enc = 0;
+	// Fix set but not used warnings
+	//uint8_t dp_pixel_encoding_type = 0;
+	uint8_t dp_compressed_pixel_format = 0;
+
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+	struct dc_crtc_timing hw_crtc_timing = *crtc_timing;
+
+	if (hw_crtc_timing.flags.INTERLACE) {
+		/*the input timing is in VESA spec format with Interlace flag =1*/
+		hw_crtc_timing.v_total /= 2;
+		hw_crtc_timing.v_border_top /= 2;
+		hw_crtc_timing.v_addressable /= 2;
+		hw_crtc_timing.v_border_bottom /= 2;
+		hw_crtc_timing.v_front_porch /= 2;
+		hw_crtc_timing.v_sync_width /= 2;
+	}
+
+
+	/* set pixel encoding */
+	switch (hw_crtc_timing.pixel_encoding) {
+	case PIXEL_ENCODING_YCBCR422:
+		dp_pixel_encoding = DP_PIXEL_ENCODING_TYPE_YCBCR422;
+		break;
+	case PIXEL_ENCODING_YCBCR444:
+		dp_pixel_encoding = DP_PIXEL_ENCODING_TYPE_YCBCR444;
+
+		if (hw_crtc_timing.flags.Y_ONLY)
+			if (hw_crtc_timing.display_color_depth != COLOR_DEPTH_666)
+				/* HW testing only, no use case yet.
+				 * Color depth of Y-only could be
+				 * 8, 10, 12, 16 bits
+				 */
+				dp_pixel_encoding = DP_PIXEL_ENCODING_TYPE_Y_ONLY;
+
+		/* Note: DP_MSA_MISC1 bit 7 is the indicator
+		 * of Y-only mode.
+		 * This bit is set in HW if register
+		 * DP_PIXEL_ENCODING is programmed to 0x4
+		 */
+		break;
+	case PIXEL_ENCODING_YCBCR420:
+		dp_pixel_encoding = DP_PIXEL_ENCODING_TYPE_YCBCR420;
+		break;
+	default:
+		dp_pixel_encoding = DP_PIXEL_ENCODING_TYPE_RGB444;
+		break;
+	}
+
+	misc1 = REG_READ(DP_MSA_MISC);
+	/* For YCbCr420 and BT2020 Colorimetry Formats, VSC SDP shall be used.
+	 * When MISC1, bit 6, is Set to 1, a Source device uses a VSC SDP to indicate the
+	 * Pixel Encoding/Colorimetry Format and that a Sink device shall ignore MISC1, bit 7,
+	 * and MISC0, bits 7:1 (MISC1, bit 7, and MISC0, bits 7:1, become "don't care").
+	 */
+	if (use_vsc_sdp_for_colorimetry)
+		misc1 = misc1 | 0x40;
+	else
+		misc1 = misc1 & ~0x40;
+
+	/* set color depth */
+	switch (hw_crtc_timing.display_color_depth) {
+	case COLOR_DEPTH_666:
+		dp_component_depth = DP_COMPONENT_PIXEL_DEPTH_6BPC;
+		break;
+	case COLOR_DEPTH_888:
+		dp_component_depth = DP_COMPONENT_PIXEL_DEPTH_8BPC;
+		break;
+	case COLOR_DEPTH_101010:
+		dp_component_depth = DP_COMPONENT_PIXEL_DEPTH_10BPC;
+		break;
+	case COLOR_DEPTH_121212:
+		dp_component_depth = DP_COMPONENT_PIXEL_DEPTH_12BPC;
+		break;
+	case COLOR_DEPTH_161616:
+		dp_component_depth = DP_COMPONENT_PIXEL_DEPTH_16BPC;
+		break;
+	default:
+		dp_component_depth = DP_COMPONENT_PIXEL_DEPTH_6BPC;
+		break;
+	}
+
+	if (hw_crtc_timing.flags.DSC) {
+		// Fix set but not used error
+		//dp_pixel_encoding_type = 1;
+		switch (hw_crtc_timing.pixel_encoding) {
+		case PIXEL_ENCODING_YCBCR444:
+			dp_compressed_pixel_format = 0;
+			break;
+		case PIXEL_ENCODING_YCBCR422:
+			dp_compressed_pixel_format = 1;
+			if (hw_crtc_timing.dsc_cfg.ycbcr422_simple)
+				dp_compressed_pixel_format = 0;
+			break;
+		case PIXEL_ENCODING_YCBCR420:
+			dp_compressed_pixel_format = 1;
+			break;
+		default:
+			dp_compressed_pixel_format = 0;
+			break;
+		}
+	} else {
+		// Fix set but not used error
+		//dp_pixel_encoding_type = 0;
+		switch (dp_pixel_encoding) {
+		case DP_PIXEL_ENCODING_TYPE_RGB444:
+			dp_translate_pixel_enc = 0;
+			break;
+		case DP_PIXEL_ENCODING_TYPE_YCBCR422:
+			dp_translate_pixel_enc = 1;
+			break;
+		case DP_PIXEL_ENCODING_TYPE_YCBCR444:
+			dp_translate_pixel_enc = 0;
+			break;
+		case DP_PIXEL_ENCODING_TYPE_Y_ONLY:
+			dp_translate_pixel_enc = 3;
+			break;
+		case DP_PIXEL_ENCODING_TYPE_YCBCR420:
+			dp_translate_pixel_enc = 2;
+			break;
+		default:
+			ASSERT(0);
+			break;
+		}
+	}
+	/* Set DP pixel encoding and component depth */
+	REG_UPDATE_4(DP_PIXEL_FORMAT,
+			PIXEL_ENCODING_TYPE, hw_crtc_timing.flags.DSC ? 1 : 0,
+			UNCOMPRESSED_PIXEL_FORMAT, dp_translate_pixel_enc,
+			UNCOMPRESSED_COMPONENT_DEPTH, dp_component_depth,
+			COMPRESSED_PIXEL_FORMAT, dp_compressed_pixel_format);
+
+	/* set dynamic range and YCbCr range */
+
+	switch (hw_crtc_timing.display_color_depth) {
+	case COLOR_DEPTH_666:
+		colorimetry_bpc = 0;
+		break;
+	case COLOR_DEPTH_888:
+		colorimetry_bpc = 1;
+		break;
+	case COLOR_DEPTH_101010:
+		colorimetry_bpc = 2;
+		break;
+	case COLOR_DEPTH_121212:
+		colorimetry_bpc = 3;
+		break;
+	default:
+		colorimetry_bpc = 0;
+		break;
+	}
+
+	misc0 = colorimetry_bpc << 5;
+
+	switch (output_color_space) {
+	case COLOR_SPACE_SRGB:
+		misc1 = misc1 & ~0x80; /* bit7 = 0*/
+		break;
+	case COLOR_SPACE_SRGB_LIMITED:
+		misc0 = misc0 | 0x8; /* bit3=1 */
+		misc1 = misc1 & ~0x80; /* bit7 = 0*/
+		break;
+	case COLOR_SPACE_YCBCR601:
+	case COLOR_SPACE_YCBCR601_LIMITED:
+		misc0 = misc0 | 0x8; /* bit3=1, bit4=0 */
+		misc1 = misc1 & ~0x80; /* bit7 = 0*/
+		if (hw_crtc_timing.pixel_encoding == PIXEL_ENCODING_YCBCR422)
+			misc0 = misc0 | 0x2; /* bit2=0, bit1=1 */
+		else if (hw_crtc_timing.pixel_encoding == PIXEL_ENCODING_YCBCR444)
+			misc0 = misc0 | 0x4; /* bit2=1, bit1=0 */
+		break;
+	case COLOR_SPACE_YCBCR709:
+	case COLOR_SPACE_YCBCR709_LIMITED:
+		misc0 = misc0 | 0x18; /* bit3=1, bit4=1 */
+		misc1 = misc1 & ~0x80; /* bit7 = 0*/
+		if (hw_crtc_timing.pixel_encoding == PIXEL_ENCODING_YCBCR422)
+			misc0 = misc0 | 0x2; /* bit2=0, bit1=1 */
+		else if (hw_crtc_timing.pixel_encoding == PIXEL_ENCODING_YCBCR444)
+			misc0 = misc0 | 0x4; /* bit2=1, bit1=0 */
+		break;
+	case COLOR_SPACE_2020_RGB_LIMITEDRANGE:
+	case COLOR_SPACE_2020_RGB_FULLRANGE:
+	case COLOR_SPACE_2020_YCBCR_LIMITED:
+	case COLOR_SPACE_XR_RGB:
+	case COLOR_SPACE_MSREF_SCRGB:
+	case COLOR_SPACE_ADOBERGB:
+	case COLOR_SPACE_DCIP3:
+	case COLOR_SPACE_XV_YCC_709:
+	case COLOR_SPACE_XV_YCC_601:
+	case COLOR_SPACE_DISPLAYNATIVE:
+	case COLOR_SPACE_DOLBYVISION:
+	case COLOR_SPACE_APPCTRL:
+	case COLOR_SPACE_CUSTOMPOINTS:
+	case COLOR_SPACE_UNKNOWN:
+	case COLOR_SPACE_YCBCR709_BLACK:
+	default:
+		/* do nothing */
+		break;
+	}
+
+	REG_SET(DP_MSA_COLORIMETRY, 0, DP_MSA_MISC0, misc0);
+	REG_WRITE(DP_MSA_MISC, misc1);   /* MSA_MISC1 */
+
+	/* dcn new register
+	 * dc_crtc_timing is vesa dmt struct. data from edid
+	 */
+	REG_SET_2(DP_MSA_TIMING_PARAM1, 0,
+			DP_MSA_HTOTAL, hw_crtc_timing.h_total,
+			DP_MSA_VTOTAL, hw_crtc_timing.v_total);
+
+	/* calculate from vesa timing parameters
+	 * h_active_start related to leading edge of sync
+	 */
+
+	h_blank = hw_crtc_timing.h_total - hw_crtc_timing.h_border_left -
+			hw_crtc_timing.h_addressable - hw_crtc_timing.h_border_right;
+
+	h_back_porch = h_blank - hw_crtc_timing.h_front_porch -
+			hw_crtc_timing.h_sync_width;
+
+	/* start at beginning of left border */
+	h_active_start = hw_crtc_timing.h_sync_width + h_back_porch;
+
+
+	v_active_start = hw_crtc_timing.v_total - hw_crtc_timing.v_border_top -
+			hw_crtc_timing.v_addressable - hw_crtc_timing.v_border_bottom -
+			hw_crtc_timing.v_front_porch;
+
+
+	/* start at beginning of left border */
+	REG_SET_2(DP_MSA_TIMING_PARAM2, 0,
+		DP_MSA_HSTART, h_active_start,
+		DP_MSA_VSTART, v_active_start);
+
+	REG_SET_4(DP_MSA_TIMING_PARAM3, 0,
+			DP_MSA_HSYNCWIDTH,
+			hw_crtc_timing.h_sync_width,
+			DP_MSA_HSYNCPOLARITY,
+			!hw_crtc_timing.flags.HSYNC_POSITIVE_POLARITY,
+			DP_MSA_VSYNCWIDTH,
+			hw_crtc_timing.v_sync_width,
+			DP_MSA_VSYNCPOLARITY,
+			!hw_crtc_timing.flags.VSYNC_POSITIVE_POLARITY);
+
+	/* HWDITH include border or overscan */
+	REG_SET_2(DP_MSA_TIMING_PARAM4, 0,
+		DP_MSA_HWIDTH, hw_crtc_timing.h_border_left +
+		hw_crtc_timing.h_addressable + hw_crtc_timing.h_border_right,
+		DP_MSA_VHEIGHT, hw_crtc_timing.v_border_top +
+		hw_crtc_timing.v_addressable + hw_crtc_timing.v_border_bottom);
+
+	REG_UPDATE(DP_SEC_FRAMING4,
+		DP_SST_SDP_SPLITTING, enable_sdp_splitting);
+}
+
+void enc401_stream_encoder_map_to_link(
+		struct stream_encoder *enc,
+		uint32_t stream_enc_inst,
+		uint32_t link_enc_inst)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	REG_UPDATE(STREAM_MAPPER_CONTROL,
+				DIG_STREAM_LINK_TARGET, link_enc_inst);
+}
+
+static const struct stream_encoder_funcs dcn401_str_enc_funcs = {
+	.dp_set_odm_combine =
+		enc401_dp_set_odm_combine,
+	.dp_set_stream_attribute =
+		enc401_stream_encoder_dp_set_stream_attribute,
+	.hdmi_set_stream_attribute =
+		enc401_stream_encoder_hdmi_set_stream_attribute,
+	.dvi_set_stream_attribute =
+		enc401_stream_encoder_dvi_set_stream_attribute,
+	.set_throttled_vcp_size =
+		enc1_stream_encoder_set_throttled_vcp_size,
+	.update_hdmi_info_packets =
+		enc3_stream_encoder_update_hdmi_info_packets,
+	.stop_hdmi_info_packets =
+		enc3_stream_encoder_stop_hdmi_info_packets,
+	.update_dp_info_packets_sdp_line_num =
+		enc3_stream_encoder_update_dp_info_packets_sdp_line_num,
+	.update_dp_info_packets =
+		enc3_stream_encoder_update_dp_info_packets,
+	.stop_dp_info_packets =
+		enc1_stream_encoder_stop_dp_info_packets,
+	.dp_blank =
+		enc1_stream_encoder_dp_blank,
+	.dp_unblank =
+		enc401_stream_encoder_dp_unblank,
+	.audio_mute_control = enc3_audio_mute_control,
+
+	.dp_audio_setup = enc3_se_dp_audio_setup,
+	.dp_audio_enable = enc3_se_dp_audio_enable,
+	.dp_audio_disable = enc1_se_dp_audio_disable,
+
+	.hdmi_audio_setup = enc3_se_hdmi_audio_setup,
+	.hdmi_audio_disable = enc1_se_hdmi_audio_disable,
+	.setup_stereo_sync  = enc1_setup_stereo_sync,
+	.set_avmute = enc1_stream_encoder_set_avmute,
+	.dig_connect_to_otg = enc1_dig_connect_to_otg,
+	.dig_source_otg = enc1_dig_source_otg,
+
+	.dp_get_pixel_format  = enc1_stream_encoder_dp_get_pixel_format,
+
+	.enc_read_state = enc401_read_state,
+	.dp_set_dsc_config = NULL,
+	.dp_set_dsc_pps_info_packet = enc3_dp_set_dsc_pps_info_packet,
+	.set_dynamic_metadata = enc401_set_dynamic_metadata,
+	.hdmi_reset_stream_attribute = enc1_reset_hdmi_stream_attribute,
+	.enable_stream = enc401_stream_encoder_enable,
+
+	.set_input_mode = enc401_set_dig_input_mode,
+	.enable_fifo = enc35_enable_fifo,
+	.disable_fifo = enc35_disable_fifo,
+	.map_stream_to_link = enc401_stream_encoder_map_to_link,
+};
+
+void dcn401_dio_stream_encoder_construct(
+	struct dcn10_stream_encoder *enc1,
+	struct dc_context *ctx,
+	struct dc_bios *bp,
+	enum engine_id eng_id,
+	struct vpg *vpg,
+	struct afmt *afmt,
+	const struct dcn10_stream_enc_registers *regs,
+	const struct dcn10_stream_encoder_shift *se_shift,
+	const struct dcn10_stream_encoder_mask *se_mask)
+{
+	enc1->base.funcs = &dcn401_str_enc_funcs;
+	enc1->base.ctx = ctx;
+	enc1->base.id = eng_id;
+	enc1->base.bp = bp;
+	enc1->base.vpg = vpg;
+	enc1->base.afmt = afmt;
+	enc1->regs = regs;
+	enc1->se_shift = se_shift;
+	enc1->se_mask = se_mask;
+	enc1->base.stream_enc_inst = vpg->inst;
+}
+
+void enc401_set_dynamic_metadata(struct stream_encoder *enc,
+		bool enable_dme,
+		uint32_t hubp_requestor_id,
+		enum dynamic_metadata_mode dmdata_mode)
+{
+	struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc);
+
+	if (enable_dme) {
+		REG_UPDATE_2(DME_CONTROL,
+				METADATA_HUBP_REQUESTOR_ID, hubp_requestor_id,
+				METADATA_STREAM_TYPE, (dmdata_mode == dmdata_dolby_vision) ? 1 : 0);
+
+		/* Use default line reference DP_SOF for bringup.
+		 * Should use OTG_SOF for DRR cases
+		 */
+		if (dmdata_mode == dmdata_dp)
+			REG_UPDATE_3(DP_SEC_METADATA_TRANSMISSION,
+					DP_SEC_METADATA_PACKET_ENABLE, 1,
+					DP_SEC_METADATA_PACKET_LINE_REFERENCE, 0,
+					DP_SEC_METADATA_PACKET_LINE, 20);
+		else {
+			REG_UPDATE_3(HDMI_METADATA_PACKET_CONTROL,
+					HDMI_METADATA_PACKET_ENABLE, 1,
+					HDMI_METADATA_PACKET_LINE_REFERENCE, 0,
+					HDMI_METADATA_PACKET_LINE, 2);
+
+			if (dmdata_mode == dmdata_dolby_vision)
+				REG_UPDATE(HDMI_CONTROL,
+						DOLBY_VISION_EN, 1);
+		}
+
+		REG_UPDATE(DME_CONTROL,
+				METADATA_ENGINE_EN, 1);
+	} else {
+		REG_UPDATE(DME_CONTROL,
+				METADATA_ENGINE_EN, 0);
+
+		if (dmdata_mode == dmdata_dp)
+			REG_UPDATE(DP_SEC_METADATA_TRANSMISSION,
+					DP_SEC_METADATA_PACKET_ENABLE, 0);
+		else {
+			REG_UPDATE(HDMI_METADATA_PACKET_CONTROL,
+					HDMI_METADATA_PACKET_ENABLE, 0);
+			REG_UPDATE(HDMI_CONTROL,
+					DOLBY_VISION_EN, 0);
+		}
+	}
+}
+void enc401_stream_encoder_set_stream_attribute_helper(
+		struct dcn10_stream_encoder *enc1,
+		struct dc_crtc_timing *crtc_timing)
+{
+	switch (crtc_timing->pixel_encoding) {
+	case PIXEL_ENCODING_YCBCR422:
+		REG_UPDATE(HDMI_CONTROL, TMDS_PIXEL_ENCODING, 1);
+		break;
+	default:
+		REG_UPDATE(HDMI_CONTROL, TMDS_PIXEL_ENCODING, 0);
+		break;
+	}
+	REG_UPDATE(HDMI_CONTROL, TMDS_COLOR_FORMAT, 0);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn401/dcn401_dio_stream_encoder.h b/drivers/gpu/drm/amd/display/dc/dio/dcn401/dcn401_dio_stream_encoder.h
new file mode 100644
index 000000000000..d6b00cd246b1
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dio/dcn401/dcn401_dio_stream_encoder.h
@@ -0,0 +1,240 @@
+/*
+ * Copyright 2021 - Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ *  and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DC_DIO_STREAM_ENCODER_DCN401_H__
+#define __DC_DIO_STREAM_ENCODER_DCN401_H__
+
+#include "dcn30/dcn30_vpg.h"
+#include "dcn30/dcn30_afmt.h"
+#include "stream_encoder.h"
+#include "dcn20/dcn20_stream_encoder.h"
+
+#define SE_COMMON_MASK_SH_LIST_DCN401(mask_sh)\
+	SE_SF(DP0_DP_PIXEL_FORMAT, PIXEL_ENCODING_TYPE, mask_sh),\
+	SE_SF(DP0_DP_PIXEL_FORMAT, UNCOMPRESSED_PIXEL_FORMAT, mask_sh),\
+	SE_SF(DP0_DP_PIXEL_FORMAT, UNCOMPRESSED_COMPONENT_DEPTH, mask_sh),\
+	SE_SF(DP0_DP_PIXEL_FORMAT, COMPRESSED_PIXEL_FORMAT, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_PACKET_GEN_VERSION, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_KEEPOUT_MODE, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_DATA_SCRAMBLE_EN, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_NO_EXTRA_NULL_PACKET_FILLED, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_GC_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_GC_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_NULL_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_VBI_PACKET_CONTROL, HDMI_ACP_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_INFOFRAME_CONTROL1, HDMI_AUDIO_INFO_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GC, HDMI_GC_AVMUTE, mask_sh),\
+	SE_SF(DP0_DP_MSE_RATE_CNTL, DP_MSE_RATE_X, mask_sh),\
+	SE_SF(DP0_DP_MSE_RATE_CNTL, DP_MSE_RATE_Y, mask_sh),\
+	SE_SF(DP0_DP_MSE_RATE_UPDATE, DP_MSE_RATE_UPDATE_PENDING, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP0_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP1_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP2_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP3_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_MPG_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL1, DP_SEC_GSP5_LINE_REFERENCE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP4_SEND, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP4_SEND_PENDING, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL4, DP_SEC_GSP4_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL5, DP_SEC_GSP5_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP4_SEND_ANY_LINE, mask_sh),\
+	SE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_DIS_DEFER, mask_sh),\
+	SE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_STATUS, mask_sh),\
+	SE_SF(DP0_DP_STEER_FIFO, DP_STEER_FIFO_RESET, mask_sh),\
+	SE_SF(DP0_DP_STEER_FIFO, DP_STEER_FIFO_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_VID_TIMING, DP_VID_M_N_GEN_EN, mask_sh),\
+	SE_SF(DP0_DP_VID_N, DP_VID_N, mask_sh),\
+	SE_SF(DP0_DP_VID_M, DP_VID_M, mask_sh),\
+	SE_SF(DIG0_HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_DELAY_EN, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUTO_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_PACKET_CONTROL, HDMI_ACR_SOURCE, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUDIO_PRIORITY, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_32_0, HDMI_ACR_CTS_32, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_32_1, HDMI_ACR_N_32, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_44_0, HDMI_ACR_CTS_44, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_44_1, HDMI_ACR_N_44, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_48_0, HDMI_ACR_CTS_48, mask_sh),\
+	SE_SF(DIG0_HDMI_ACR_48_1, HDMI_ACR_N_48, mask_sh),\
+	SE_SF(DP0_DP_SEC_AUD_N, DP_SEC_AUD_N, mask_sh),\
+	SE_SF(DP0_DP_SEC_TIMESTAMP, DP_SEC_TIMESTAMP_MODE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_ASP_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_ATP_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_AIP_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_ACM_ENABLE, mask_sh),\
+	SE_SF(DIG0_AFMT_CNTL, AFMT_AUDIO_CLOCK_EN, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, HDMI_CLOCK_CHANNEL_RATE, mask_sh),\
+	SE_SF(DIG1_HDMI_CONTROL, TMDS_PIXEL_ENCODING, mask_sh),\
+	SE_SF(DIG1_HDMI_CONTROL, TMDS_COLOR_FORMAT, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_STEREOSYNC_SELECT, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_STEREOSYNC_GATE_EN, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP4_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP5_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP6_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL, DP_SEC_GSP7_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP7_SEND, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL6, DP_SEC_GSP7_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_SEC_CNTL2, DP_SEC_GSP11_PPS, mask_sh),\
+	SE_SF(DP0_DP_GSP11_CNTL, DP_SEC_GSP11_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_GSP11_CNTL, DP_SEC_GSP11_LINE_NUM, mask_sh),\
+	SE_SF(DP0_DP_DB_CNTL, DP_DB_DISABLE, mask_sh),\
+	SE_SF(DP0_DP_MSA_COLORIMETRY, DP_MSA_MISC0, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM1, DP_MSA_HTOTAL, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM1, DP_MSA_VTOTAL, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM2, DP_MSA_HSTART, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM2, DP_MSA_VSTART, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_HSYNCWIDTH, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_HSYNCPOLARITY, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_VSYNCWIDTH, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM3, DP_MSA_VSYNCPOLARITY, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM4, DP_MSA_HWIDTH, mask_sh),\
+	SE_SF(DP0_DP_MSA_TIMING_PARAM4, DP_MSA_VHEIGHT, mask_sh),\
+	SE_SF(DIG0_HDMI_DB_CONTROL, HDMI_DB_DISABLE, mask_sh),\
+	SE_SF(DP0_DP_VID_TIMING, DP_VID_N_INTERVAL, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_SOURCE_SELECT, mask_sh), \
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC0_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC0_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC1_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC1_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC2_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC2_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC3_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC3_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC4_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC4_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC5_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC5_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC6_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC6_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC7_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL0, HDMI_GENERIC7_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC8_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC8_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC9_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC9_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC10_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC10_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC11_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC11_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC12_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC12_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC13_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC13_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC14_CONT, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL6, HDMI_GENERIC14_SEND, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL1, HDMI_GENERIC0_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL1, HDMI_GENERIC1_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL2, HDMI_GENERIC2_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL2, HDMI_GENERIC3_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL3, HDMI_GENERIC4_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL3, HDMI_GENERIC5_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL4, HDMI_GENERIC6_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL4, HDMI_GENERIC7_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL7, HDMI_GENERIC8_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL7, HDMI_GENERIC9_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL8, HDMI_GENERIC10_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL8, HDMI_GENERIC11_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL9, HDMI_GENERIC12_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL9, HDMI_GENERIC13_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_GENERIC_PACKET_CONTROL10, HDMI_GENERIC14_LINE, mask_sh),\
+	SE_SF(DP0_DP_MSA_VBID_MISC, DP_VBID6_LINE_REFERENCE, mask_sh),\
+	SE_SF(DP0_DP_MSA_VBID_MISC, DP_VBID6_LINE_NUM, mask_sh),\
+	SE_SF(DME0_DME_CONTROL, METADATA_ENGINE_EN, mask_sh),\
+	SE_SF(DME0_DME_CONTROL, METADATA_HUBP_REQUESTOR_ID, mask_sh),\
+	SE_SF(DME0_DME_CONTROL, METADATA_STREAM_TYPE, mask_sh),\
+	SE_SF(DP0_DP_SEC_METADATA_TRANSMISSION, DP_SEC_METADATA_PACKET_ENABLE, mask_sh),\
+	SE_SF(DP0_DP_SEC_METADATA_TRANSMISSION, DP_SEC_METADATA_PACKET_LINE_REFERENCE, mask_sh),\
+	SE_SF(DP0_DP_SEC_METADATA_TRANSMISSION, DP_SEC_METADATA_PACKET_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_METADATA_PACKET_CONTROL, HDMI_METADATA_PACKET_ENABLE, mask_sh),\
+	SE_SF(DIG0_HDMI_METADATA_PACKET_CONTROL, HDMI_METADATA_PACKET_LINE_REFERENCE, mask_sh),\
+	SE_SF(DIG0_HDMI_METADATA_PACKET_CONTROL, HDMI_METADATA_PACKET_LINE, mask_sh),\
+	SE_SF(DIG0_HDMI_CONTROL, DOLBY_VISION_EN, mask_sh),\
+	SE_SF(DIG0_DIG_FE_EN_CNTL, DIG_FE_ENABLE, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CLK_CNTL, DIG_FE_MODE, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CLK_CNTL, DIG_FE_CLK_EN, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CLK_CNTL, DIG_FE_SOFT_RESET, mask_sh),\
+	SE_SF(DIG0_DIG_FE_CNTL, DIG_STEREOSYNC_GATE_EN, mask_sh),\
+	SE_SF(DP0_DP_SEC_FRAMING4, DP_SST_SDP_SPLITTING, mask_sh),\
+	SE_SF(DIG0_DIG_CLOCK_PATTERN, DIG_CLOCK_PATTERN, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_CTRL0, DIG_FIFO_OUTPUT_PIXEL_PER_CYCLE, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_CTRL0, DIG_FIFO_READ_START_LEVEL, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_CTRL0, DIG_FIFO_ENABLE, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_CTRL0, DIG_FIFO_RESET, mask_sh),\
+	SE_SF(DIG0_DIG_FIFO_CTRL0, DIG_FIFO_RESET_DONE, mask_sh),\
+	SE_SF(DIG0_STREAM_MAPPER_CONTROL, DIG_STREAM_LINK_TARGET, mask_sh),
+
+
+void dcn401_dio_stream_encoder_construct(
+	struct dcn10_stream_encoder *enc1,
+	struct dc_context *ctx,
+	struct dc_bios *bp,
+	enum engine_id eng_id,
+	struct vpg *vpg,
+	struct afmt *afmt,
+	const struct dcn10_stream_enc_registers *regs,
+	const struct dcn10_stream_encoder_shift *se_shift,
+	const struct dcn10_stream_encoder_mask *se_mask);
+
+void enc401_set_dynamic_metadata(struct stream_encoder *enc,
+		bool enable_dme,
+		uint32_t hubp_requestor_id,
+		enum dynamic_metadata_mode dmdata_mode);
+void enc401_stream_encoder_set_stream_attribute_helper(
+		struct dcn10_stream_encoder *enc1,
+		struct dc_crtc_timing *crtc_timing);
+void enc401_stream_encoder_dp_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	enum dc_color_space output_color_space,
+	bool use_vsc_sdp_for_colorimetry,
+	uint32_t enable_sdp_splitting);
+void enc401_stream_encoder_dvi_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	bool is_dual_link);
+void enc401_stream_encoder_dp_unblank(
+		struct dc_link *link,
+		struct stream_encoder *enc,
+		const struct encoder_unblank_param *param);
+void enc401_stream_encoder_enable(
+	struct stream_encoder *enc,
+	enum signal_type signal,
+	bool enable);
+void enc401_set_dig_input_mode(struct stream_encoder *enc, unsigned int pix_per_container);
+void enc401_stream_encoder_map_to_link(
+		struct stream_encoder *enc,
+		uint32_t stream_enc_inst,
+		uint32_t link_enc_inst);
+void enc401_read_state(struct stream_encoder *enc, struct enc_state *s);
+void enc401_stream_encoder_hdmi_set_stream_attribute(
+	struct stream_encoder *enc,
+	struct dc_crtc_timing *crtc_timing,
+	int actual_pix_clk_khz,
+	bool enable_audio);
+#endif /* __DC_DIO_STREAM_ENCODER_DCN401_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dm_cp_psp.h b/drivers/gpu/drm/amd/display/dc/dm_cp_psp.h
new file mode 100644
index 000000000000..f4d3f04ec857
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dm_cp_psp.h
@@ -0,0 +1,61 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DM_CP_PSP_IF__H
+#define DM_CP_PSP_IF__H
+
+/*
+ * Interface to CPLIB/PSP to enable ASSR
+ */
+struct dc_link;
+
+struct cp_psp_stream_config {
+	uint8_t otg_inst;
+	uint8_t dig_be;
+	uint8_t dig_fe;
+	uint8_t link_enc_idx;
+	uint8_t stream_enc_idx;
+	uint8_t dio_output_idx;
+	uint8_t phy_idx;
+	uint8_t assr_enabled;
+	uint8_t mst_enabled;
+	uint8_t dp2_enabled;
+	uint8_t usb4_enabled;
+	void *dm_stream_ctx;
+	bool dpms_off;
+};
+
+struct cp_psp_funcs {
+	bool (*enable_assr)(void *handle, struct dc_link *link);
+	void (*update_stream_config)(void *handle, struct cp_psp_stream_config *config);
+};
+
+struct cp_psp {
+	void *handle;
+	struct cp_psp_funcs funcs;
+};
+
+
+#endif /* DM_CP_PSP_IF__H */
diff --git a/drivers/gpu/drm/amd/display/dc/dm_event_log.h b/drivers/gpu/drm/amd/display/dc/dm_event_log.h
new file mode 100644
index 000000000000..65663f4d93e1
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dm_event_log.h
@@ -0,0 +1,39 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+/**
+ * This file defines external dependencies of Display Core.
+ */
+
+#ifndef __DM_EVENT_LOG_H__
+
+#define __DM_EVENT_LOG_H__
+
+#define EVENT_LOG_AUX_REQ(ddc, type, action, address, len, data)
+#define EVENT_LOG_AUX_REP(ddc, type, replyStatus, len, data)
+#define EVENT_LOG_CUST_MSG(tag, a, ...)
+
+#endif
+
diff --git a/drivers/gpu/drm/amd/display/dc/dm_helpers.h b/drivers/gpu/drm/amd/display/dc/dm_helpers.h
new file mode 100644
index 000000000000..9d160b39e8c5
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dm_helpers.h
@@ -0,0 +1,229 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+/**
+ * This file defines helper functions provided by the Display Manager to
+ * Display Core.
+ */
+#ifndef __DM_HELPERS__
+#define __DM_HELPERS__
+
+#include "dc_types.h"
+#include "dc.h"
+
+struct dc_dp_mst_stream_allocation_table;
+struct aux_payload;
+enum aux_return_code_type;
+enum set_config_status;
+
+/*
+ * Allocate memory accessible by the GPU
+ *
+ * frame buffer allocations must be aligned to a 4096-byte boundary
+ *
+ * Returns virtual address, sets addr to physical address
+ */
+void *dm_helpers_allocate_gpu_mem(
+		struct dc_context *ctx,
+		enum dc_gpu_mem_alloc_type type,
+		size_t size,
+		long long *addr);
+
+/*
+ * Free the GPU-accessible memory at the virtual address pvMem
+ */
+void dm_helpers_free_gpu_mem(
+		struct dc_context *ctx,
+		enum dc_gpu_mem_alloc_type type,
+		void *pvMem);
+
+enum dc_edid_status dm_helpers_parse_edid_caps(
+	struct dc_link *link,
+	const struct dc_edid *edid,
+	struct dc_edid_caps *edid_caps);
+
+
+/*
+ * Update DP branch info
+ */
+void dm_helpers_dp_update_branch_info(
+		struct dc_context *ctx,
+		const struct dc_link *link);
+
+/*
+ * Writes payload allocation table in immediate downstream device.
+ */
+bool dm_helpers_dp_mst_write_payload_allocation_table(
+		struct dc_context *ctx,
+		const struct dc_stream_state *stream,
+		struct dc_dp_mst_stream_allocation_table *proposed_table,
+		bool enable);
+
+/*
+ * poll pending down reply
+ */
+void dm_helpers_dp_mst_poll_pending_down_reply(
+	struct dc_context *ctx,
+	const struct dc_link *link);
+
+/*
+ * Clear payload allocation table before enable MST DP link.
+ */
+void dm_helpers_dp_mst_clear_payload_allocation_table(
+	struct dc_context *ctx,
+	const struct dc_link *link);
+
+/*
+ * Polls for ACT (allocation change trigger) handled and
+ */
+enum act_return_status dm_helpers_dp_mst_poll_for_allocation_change_trigger(
+		struct dc_context *ctx,
+		const struct dc_stream_state *stream);
+/*
+ * Sends ALLOCATE_PAYLOAD message.
+ */
+void dm_helpers_dp_mst_send_payload_allocation(
+		struct dc_context *ctx,
+		const struct dc_stream_state *stream);
+
+/*
+ * Update mst manager relevant variables
+ */
+void dm_helpers_dp_mst_update_mst_mgr_for_deallocation(
+		struct dc_context *ctx,
+		const struct dc_stream_state *stream);
+
+bool dm_helpers_dp_mst_start_top_mgr(
+		struct dc_context *ctx,
+		const struct dc_link *link,
+		bool boot);
+
+bool dm_helpers_dp_mst_stop_top_mgr(
+		struct dc_context *ctx,
+		struct dc_link *link);
+
+void dm_helpers_dp_mst_update_branch_bandwidth(
+		struct dc_context *ctx,
+		struct dc_link *link);
+
+/**
+ * OS specific aux read callback.
+ */
+bool dm_helpers_dp_read_dpcd(
+		struct dc_context *ctx,
+		const struct dc_link *link,
+		uint32_t address,
+		uint8_t *data,
+		uint32_t size);
+
+/**
+ * OS specific aux write callback.
+ */
+bool dm_helpers_dp_write_dpcd(
+		struct dc_context *ctx,
+		const struct dc_link *link,
+		uint32_t address,
+		const uint8_t *data,
+		uint32_t size);
+
+bool dm_helpers_submit_i2c(
+		struct dc_context *ctx,
+		const struct dc_link *link,
+		struct i2c_command *cmd);
+
+bool dm_helpers_execute_fused_io(
+		struct dc_context *ctx,
+		struct dc_link *link,
+		union dmub_rb_cmd *commands,
+		uint8_t count,
+		uint32_t timeout_us
+);
+
+bool dm_helpers_dp_write_dsc_enable(
+		struct dc_context *ctx,
+		const struct dc_stream_state *stream,
+		bool enable
+);
+
+bool dm_helpers_dp_write_hblank_reduction(
+		struct dc_context *ctx,
+		const struct dc_stream_state *stream);
+
+bool dm_helpers_is_dp_sink_present(
+		struct dc_link *link);
+
+void dm_helpers_mst_enable_stream_features(const struct dc_stream_state *stream);
+
+enum dc_edid_status dm_helpers_read_local_edid(
+		struct dc_context *ctx,
+		struct dc_link *link,
+		struct dc_sink *sink);
+
+bool dm_helpers_dp_handle_test_pattern_request(
+		struct dc_context *ctx,
+		const struct dc_link *link,
+		union link_test_pattern dpcd_test_pattern,
+		union test_misc dpcd_test_params);
+
+void dm_set_dcn_clocks(
+		struct dc_context *ctx,
+		struct dc_clocks *clks);
+
+void dm_helpers_enable_periodic_detection(struct dc_context *ctx, bool enable);
+
+void dm_set_phyd32clk(struct dc_context *ctx, int freq_khz);
+
+bool dm_helpers_dmub_outbox_interrupt_control(struct dc_context *ctx, bool enable);
+
+void dm_helpers_smu_timeout(struct dc_context *ctx, unsigned int msg_id, unsigned int param, unsigned int timeout_us);
+
+// 0x1 = Result_OK, 0xFE = Result_UnkmownCmd, 0x0 = Status_Busy
+#define IS_SMU_TIMEOUT(result) \
+	(result == 0x0)
+void dm_helpers_init_panel_settings(
+	struct dc_context *ctx,
+	struct dc_panel_config *config,
+	struct dc_sink *sink);
+void dm_helpers_override_panel_settings(
+	struct dc_context *ctx,
+	struct dc_panel_config *config);
+int dm_helper_dmub_aux_transfer_sync(
+		struct dc_context *ctx,
+		const struct dc_link *link,
+		struct aux_payload *payload,
+		enum aux_return_code_type *operation_result);
+
+int dm_helpers_dmub_set_config_sync(struct dc_context *ctx,
+		const struct dc_link *link,
+		struct set_config_cmd_payload *payload,
+		enum set_config_status *operation_result);
+enum adaptive_sync_type dm_get_adaptive_sync_support_type(struct dc_link *link);
+
+enum dc_edid_status dm_helpers_get_sbios_edid(struct dc_link *link, struct dc_edid *edid);
+
+bool dm_helpers_is_fullscreen(struct dc_context *ctx, struct dc_stream_state *stream);
+bool dm_helpers_is_hdr_on(struct dc_context *ctx, struct dc_stream_state *stream);
+
+#endif /* __DM_HELPERS__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dm_pp_smu.h b/drivers/gpu/drm/amd/display/dc/dm_pp_smu.h
new file mode 100644
index 000000000000..b0e17a594ec3
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dm_pp_smu.h
@@ -0,0 +1,321 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef DM_PP_SMU_IF__H
+#define DM_PP_SMU_IF__H
+
+/*
+ * interface to PPLIB/SMU to setup clocks and pstate requirements on SoC
+ */
+
+enum pp_smu_ver {
+	/*
+	 * PP_SMU_INTERFACE_X should be interpreted as the interface defined
+	 * starting from X, where X is some family of ASICs.  This is as
+	 * opposed to interfaces used only for X.  There will be some degree
+	 * of interface sharing between families of ASIcs.
+	 */
+	PP_SMU_UNSUPPORTED,
+	PP_SMU_VER_RV,
+	PP_SMU_VER_NV,
+
+	PP_SMU_VER_RN,
+	PP_SMU_VER_VG,
+	PP_SMU_VER_MAX
+};
+
+struct pp_smu {
+	enum pp_smu_ver ver;
+	const void *pp;
+
+	/*
+	 * interim extra handle for backwards compatibility
+	 * as some existing functionality not yet implemented
+	 * by ppsmu
+	 */
+	const void *dm;
+};
+
+enum pp_smu_status {
+	PP_SMU_RESULT_UNDEFINED = 0,
+	PP_SMU_RESULT_OK = 1,
+	PP_SMU_RESULT_FAIL,
+	PP_SMU_RESULT_UNSUPPORTED
+};
+
+#define PP_SMU_WM_SET_RANGE_CLK_UNCONSTRAINED_MIN 0x0
+#define PP_SMU_WM_SET_RANGE_CLK_UNCONSTRAINED_MAX 0xFFFF
+
+enum wm_type {
+	WM_TYPE_PSTATE_CHG = 0,
+	WM_TYPE_RETRAINING = 1,
+};
+
+/* This structure is a copy of WatermarkRowGeneric_t defined by smuxx_driver_if.h*/
+struct pp_smu_wm_set_range {
+	uint16_t min_fill_clk_mhz;
+	uint16_t max_fill_clk_mhz;
+	uint16_t min_drain_clk_mhz;
+	uint16_t max_drain_clk_mhz;
+
+	uint8_t wm_inst;
+	uint8_t wm_type;
+};
+
+#define MAX_WATERMARK_SETS 4
+
+struct pp_smu_wm_range_sets {
+	unsigned int num_reader_wm_sets;
+	struct pp_smu_wm_set_range reader_wm_sets[MAX_WATERMARK_SETS];
+
+	unsigned int num_writer_wm_sets;
+	struct pp_smu_wm_set_range writer_wm_sets[MAX_WATERMARK_SETS];
+};
+
+struct pp_smu_funcs_rv {
+	struct pp_smu pp_smu;
+
+	/* PPSMC_MSG_SetDisplayCount
+	 * 0 triggers S0i2 optimization
+	 */
+
+	void (*set_display_count)(struct pp_smu *pp, int count);
+
+	/* reader and writer WM's are sent together as part of one table*/
+	/*
+	 * PPSMC_MSG_SetDriverDramAddrHigh
+	 * PPSMC_MSG_SetDriverDramAddrLow
+	 * PPSMC_MSG_TransferTableDram2Smu
+	 *
+	 * */
+	void (*set_wm_ranges)(struct pp_smu *pp,
+			struct pp_smu_wm_range_sets *ranges);
+
+	/* PPSMC_MSG_SetHardMinDcfclkByFreq
+	 * fixed clock at requested freq, either from FCH bypass or DFS
+	 */
+	void (*set_hard_min_dcfclk_by_freq)(struct pp_smu *pp, int mhz);
+
+	/* PPSMC_MSG_SetMinDeepSleepDcfclk
+	 * when DF is in cstate, dcf clock is further divided down
+	 * to just above given frequency
+	 */
+	void (*set_min_deep_sleep_dcfclk)(struct pp_smu *pp, int mhz);
+
+	/* PPSMC_MSG_SetHardMinFclkByFreq
+	 * FCLK will vary with DPM, but never below requested hard min
+	 */
+	void (*set_hard_min_fclk_by_freq)(struct pp_smu *pp, int mhz);
+
+	/* PPSMC_MSG_SetHardMinSocclkByFreq
+	 * Needed for DWB support
+	 */
+	void (*set_hard_min_socclk_by_freq)(struct pp_smu *pp, int mhz);
+
+	/* PME w/a */
+	void (*set_pme_wa_enable)(struct pp_smu *pp);
+};
+
+/* Used by pp_smu_funcs_nv.set_voltage_by_freq
+ *
+ */
+enum pp_smu_nv_clock_id {
+	PP_SMU_NV_DISPCLK,
+	PP_SMU_NV_PHYCLK,
+	PP_SMU_NV_PIXELCLK
+};
+
+/*
+ * Used by pp_smu_funcs_nv.get_maximum_sustainable_clocks
+ */
+struct pp_smu_nv_clock_table {
+	// voltage managed SMU, freq set by driver
+	unsigned int    displayClockInKhz;
+	unsigned int	dppClockInKhz;
+	unsigned int    phyClockInKhz;
+	unsigned int    pixelClockInKhz;
+	unsigned int	dscClockInKhz;
+
+	// freq/voltage managed by SMU
+	unsigned int	fabricClockInKhz;
+	unsigned int	socClockInKhz;
+	unsigned int    dcfClockInKhz;
+	unsigned int    uClockInKhz;
+};
+
+struct pp_smu_funcs_nv {
+	struct pp_smu pp_smu;
+
+	/* PPSMC_MSG_SetDisplayCount
+	 * 0 triggers S0i2 optimization
+	 */
+	enum pp_smu_status (*set_display_count)(struct pp_smu *pp, int count);
+
+	/* PPSMC_MSG_SetHardMinDcfclkByFreq
+	 * fixed clock at requested freq, either from FCH bypass or DFS
+	 */
+	enum pp_smu_status (*set_hard_min_dcfclk_by_freq)(struct pp_smu *pp, int Mhz);
+
+	/* PPSMC_MSG_SetMinDeepSleepDcfclk
+	 * when DF is in cstate, dcf clock is further divided down
+	 * to just above given frequency
+	 */
+	enum pp_smu_status (*set_min_deep_sleep_dcfclk)(struct pp_smu *pp, int Mhz);
+
+	/* PPSMC_MSG_SetHardMinUclkByFreq
+	 * UCLK will vary with DPM, but never below requested hard min
+	 */
+	enum pp_smu_status (*set_hard_min_uclk_by_freq)(struct pp_smu *pp, int Mhz);
+
+	/* PPSMC_MSG_SetHardMinSocclkByFreq
+	 * Needed for DWB support
+	 */
+	enum pp_smu_status (*set_hard_min_socclk_by_freq)(struct pp_smu *pp, int Mhz);
+
+	/* PME w/a */
+	enum pp_smu_status (*set_pme_wa_enable)(struct pp_smu *pp);
+
+	/* PPSMC_MSG_SetHardMinByFreq
+	 * Needed to set ASIC voltages for clocks programmed by DAL
+	 */
+	enum pp_smu_status (*set_voltage_by_freq)(struct pp_smu *pp,
+			enum pp_smu_nv_clock_id clock_id, int Mhz);
+
+	/* reader and writer WM's are sent together as part of one table*/
+	/*
+	 * PPSMC_MSG_SetDriverDramAddrHigh
+	 * PPSMC_MSG_SetDriverDramAddrLow
+	 * PPSMC_MSG_TransferTableDram2Smu
+	 *
+	 * on DCN20:
+	 * 	reader fill clk = uclk
+	 * 	reader drain clk = dcfclk
+	 * 	writer fill clk = socclk
+	 * 	writer drain clk = uclk
+	 * */
+	enum pp_smu_status (*set_wm_ranges)(struct pp_smu *pp,
+			struct pp_smu_wm_range_sets *ranges);
+
+	/* Not a single SMU message.  This call should return maximum sustainable limit for all
+	 * clocks that DC depends on.  These will be used as basis for mode enumeration.
+	 */
+	enum pp_smu_status (*get_maximum_sustainable_clocks)(struct pp_smu *pp,
+			struct pp_smu_nv_clock_table *max_clocks);
+
+	/* This call should return the discrete uclk DPM states available
+	 */
+	enum pp_smu_status (*get_uclk_dpm_states)(struct pp_smu *pp,
+			unsigned int *clock_values_in_khz, unsigned int *num_states);
+
+	/* Not a single SMU message.  This call informs PPLIB that display will not be able
+	 * to perform pstate handshaking in its current state.  Typically this handshake
+	 * is used to perform uCLK switching, so disabling pstate disables uCLK switching.
+	 *
+	 * Note that when setting handshake to unsupported, the call is pre-emptive.  That means
+	 * DC will make the call BEFORE setting up the display state which would cause pstate
+	 * request to go un-acked.  Only when the call completes should such a state be applied to
+	 * DC hardware
+	 */
+	enum pp_smu_status (*set_pstate_handshake_support)(struct pp_smu *pp,
+			bool pstate_handshake_supported);
+};
+
+#define PP_SMU_NUM_SOCCLK_DPM_LEVELS  8
+#define PP_SMU_NUM_DCFCLK_DPM_LEVELS  8
+#define PP_SMU_NUM_FCLK_DPM_LEVELS    4
+#define PP_SMU_NUM_MEMCLK_DPM_LEVELS  4
+#define PP_SMU_NUM_DCLK_DPM_LEVELS    8
+#define PP_SMU_NUM_VCLK_DPM_LEVELS    8
+#define PP_SMU_NUM_VPECLK_DPM_LEVELS  8
+
+struct dpm_clock {
+  uint32_t  Freq;    // In MHz
+  uint32_t  Vol;     // Millivolts with 2 fractional bits
+};
+
+
+/* this is a copy of the structure defined in smuxx_driver_if.h*/
+struct dpm_clocks {
+	struct dpm_clock DcfClocks[PP_SMU_NUM_DCFCLK_DPM_LEVELS];
+	struct dpm_clock SocClocks[PP_SMU_NUM_SOCCLK_DPM_LEVELS];
+	struct dpm_clock FClocks[PP_SMU_NUM_FCLK_DPM_LEVELS];
+	struct dpm_clock MemClocks[PP_SMU_NUM_MEMCLK_DPM_LEVELS];
+	struct dpm_clock VClocks[PP_SMU_NUM_VCLK_DPM_LEVELS];
+	struct dpm_clock DClocks[PP_SMU_NUM_DCLK_DPM_LEVELS];
+	struct dpm_clock VPEClocks[PP_SMU_NUM_VPECLK_DPM_LEVELS];
+};
+
+
+struct pp_smu_funcs_rn {
+	struct pp_smu pp_smu;
+
+	/*
+	 * reader and writer WM's are sent together as part of one table
+	 *
+	 * PPSMC_MSG_SetDriverDramAddrHigh
+	 * PPSMC_MSG_SetDriverDramAddrLow
+	 * PPSMC_MSG_TransferTableDram2Smu
+	 *
+	 */
+	enum pp_smu_status (*set_wm_ranges)(struct pp_smu *pp,
+			struct pp_smu_wm_range_sets *ranges);
+
+	enum pp_smu_status (*get_dpm_clock_table) (struct pp_smu *pp,
+			struct dpm_clocks *clock_table);
+};
+
+struct pp_smu_funcs_vgh {
+	struct pp_smu pp_smu;
+
+	/*
+	 * reader and writer WM's are sent together as part of one table
+	 *
+	 * PPSMC_MSG_SetDriverDramAddrHigh
+	 * PPSMC_MSG_SetDriverDramAddrLow
+	 * PPSMC_MSG_TransferTableDram2Smu
+	 *
+	 */
+	// TODO: Check whether this is moved to DAL, and remove as needed
+	enum pp_smu_status (*set_wm_ranges)(struct pp_smu *pp,
+			struct pp_smu_wm_range_sets *ranges);
+
+	// TODO: Check whether this is moved to DAL, and remove as needed
+	enum pp_smu_status (*get_dpm_clock_table) (struct pp_smu *pp,
+			struct dpm_clocks *clock_table);
+
+	enum pp_smu_status (*notify_smu_timeout) (struct pp_smu *pp);
+};
+
+struct pp_smu_funcs {
+	struct pp_smu ctx;
+	union {
+		struct pp_smu_funcs_rv rv_funcs;
+		struct pp_smu_funcs_nv nv_funcs;
+		struct pp_smu_funcs_rn rn_funcs;
+		struct pp_smu_funcs_vgh vgh_funcs;
+	};
+};
+
+#endif /* DM_PP_SMU_IF__H */
diff --git a/drivers/gpu/drm/amd/display/dc/dm_services.h b/drivers/gpu/drm/amd/display/dc/dm_services.h
new file mode 100644
index 000000000000..fbbf9c757b3c
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dm_services.h
@@ -0,0 +1,317 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+/**
+ * This file defines external dependencies of Display Core.
+ */
+
+#ifndef __DM_SERVICES_H__
+
+#define __DM_SERVICES_H__
+
+/* TODO: remove when DC is complete. */
+#include "dm_services_types.h"
+#include "logger_interface.h"
+#include "link_service_types.h"
+
+#undef DEPRECATED
+
+struct dmub_srv;
+struct dc_dmub_srv;
+union dmub_rb_cmd;
+
+irq_handler_idx dm_register_interrupt(
+	struct dc_context *ctx,
+	struct dc_interrupt_params *int_params,
+	interrupt_handler ih,
+	void *handler_args);
+
+/*
+ *
+ * GPU registers access
+ *
+ */
+uint32_t dm_read_reg_func(const struct dc_context *ctx, uint32_t address,
+			  const char *func_name);
+
+/* enable for debugging new code, this adds 50k to the driver size. */
+/* #define DM_CHECK_ADDR_0 */
+
+void dm_write_reg_func(const struct dc_context *ctx, uint32_t address,
+		       uint32_t value, const char *func_name);
+
+#define dm_read_reg(ctx, address)	\
+	dm_read_reg_func(ctx, address, __func__)
+
+#define dm_write_reg(ctx, address, value)	\
+	dm_write_reg_func(ctx, address, value, __func__)
+
+static inline uint32_t dm_read_index_reg(
+	const struct dc_context *ctx,
+	enum cgs_ind_reg addr_space,
+	uint32_t index)
+{
+	return cgs_read_ind_register(ctx->cgs_device, addr_space, index);
+}
+
+static inline void dm_write_index_reg(
+	const struct dc_context *ctx,
+	enum cgs_ind_reg addr_space,
+	uint32_t index,
+	uint32_t value)
+{
+	cgs_write_ind_register(ctx->cgs_device, addr_space, index, value);
+}
+
+static inline uint32_t get_reg_field_value_ex(
+	uint32_t reg_value,
+	uint32_t mask,
+	uint8_t shift)
+{
+	return (mask & reg_value) >> shift;
+}
+
+#define get_reg_field_value(reg_value, reg_name, reg_field)\
+	get_reg_field_value_ex(\
+		(reg_value),\
+		reg_name ## __ ## reg_field ## _MASK,\
+		reg_name ## __ ## reg_field ## __SHIFT)
+
+static inline uint32_t set_reg_field_value_ex(
+	uint32_t reg_value,
+	uint32_t value,
+	uint32_t mask,
+	uint8_t shift)
+{
+	ASSERT(mask != 0);
+	return (reg_value & ~mask) | (mask & (value << shift));
+}
+
+#define set_reg_field_value(reg_value, value, reg_name, reg_field)\
+	(reg_value) = set_reg_field_value_ex(\
+		(reg_value),\
+		(value),\
+		reg_name ## __ ## reg_field ## _MASK,\
+		reg_name ## __ ## reg_field ## __SHIFT)
+
+uint32_t generic_reg_set_ex(const struct dc_context *ctx,
+		uint32_t addr, uint32_t reg_val, int n,
+		uint8_t shift1, uint32_t mask1, uint32_t field_value1, ...);
+
+uint32_t generic_reg_update_ex(const struct dc_context *ctx,
+		uint32_t addr, int n,
+		uint8_t shift1, uint32_t mask1, uint32_t field_value1, ...);
+
+struct dc_dmub_srv *dc_dmub_srv_create(struct dc *dc, struct dmub_srv *dmub);
+void dc_dmub_srv_destroy(struct dc_dmub_srv **dmub_srv);
+
+void reg_sequence_start_gather(const struct dc_context *ctx);
+void reg_sequence_start_execute(const struct dc_context *ctx);
+void reg_sequence_wait_done(const struct dc_context *ctx);
+
+#define FD(reg_field)	reg_field ## __SHIFT, \
+						reg_field ## _MASK
+
+/*
+ * return number of poll before condition is met
+ * return 0 if condition is not meet after specified time out tries
+ */
+void generic_reg_wait(const struct dc_context *ctx,
+	uint32_t addr, uint32_t mask, uint32_t shift, uint32_t condition_value,
+	unsigned int delay_between_poll_us, unsigned int time_out_num_tries,
+	const char *func_name, int line);
+
+unsigned int snprintf_count(char *pBuf, unsigned int bufSize, const char *fmt, ...);
+
+/* These macros need to be used with soc15 registers in order to retrieve
+ * the actual offset.
+ */
+#define dm_write_reg_soc15(ctx, reg, inst_offset, value)	\
+		dm_write_reg_func(ctx, reg + DCE_BASE.instance[0].segment[reg##_BASE_IDX] + inst_offset, value, __func__)
+
+#define dm_read_reg_soc15(ctx, reg, inst_offset)	\
+		dm_read_reg_func(ctx, reg + DCE_BASE.instance[0].segment[reg##_BASE_IDX] + inst_offset, __func__)
+
+#define generic_reg_update_soc15(ctx, inst_offset, reg_name, n, ...)\
+		generic_reg_update_ex(ctx, DCE_BASE.instance[0].segment[mm##reg_name##_BASE_IDX] +  mm##reg_name + inst_offset, \
+		n, __VA_ARGS__)
+
+#define generic_reg_set_soc15(ctx, inst_offset, reg_name, n, ...)\
+		generic_reg_set_ex(ctx, DCE_BASE.instance[0].segment[mm##reg_name##_BASE_IDX] + mm##reg_name + inst_offset, 0, \
+		n, __VA_ARGS__)
+
+#define get_reg_field_value_soc15(reg_value, block, reg_num, reg_name, reg_field)\
+	get_reg_field_value_ex(\
+		(reg_value),\
+		block ## reg_num ## _ ## reg_name ## __ ## reg_field ## _MASK,\
+		block ## reg_num ## _ ## reg_name ## __ ## reg_field ## __SHIFT)
+
+#define set_reg_field_value_soc15(reg_value, value, block, reg_num, reg_name, reg_field)\
+	(reg_value) = set_reg_field_value_ex(\
+		(reg_value),\
+		(value),\
+		block ## reg_num ## _ ## reg_name ## __ ## reg_field ## _MASK,\
+		block ## reg_num ## _ ## reg_name ## __ ## reg_field ## __SHIFT)
+
+/**************************************
+ * Power Play (PP) interfaces
+ **************************************/
+
+/* Gets valid clocks levels from pplib
+ *
+ * input: clk_type - display clk / sclk / mem clk
+ *
+ * output: array of valid clock levels for given type in ascending order,
+ * with invalid levels filtered out
+ *
+ */
+bool dm_pp_get_clock_levels_by_type(
+	const struct dc_context *ctx,
+	enum dm_pp_clock_type clk_type,
+	struct dm_pp_clock_levels *clk_level_info);
+
+bool dm_pp_get_clock_levels_by_type_with_latency(
+	const struct dc_context *ctx,
+	enum dm_pp_clock_type clk_type,
+	struct dm_pp_clock_levels_with_latency *clk_level_info);
+
+bool dm_pp_get_clock_levels_by_type_with_voltage(
+	const struct dc_context *ctx,
+	enum dm_pp_clock_type clk_type,
+	struct dm_pp_clock_levels_with_voltage *clk_level_info);
+
+bool dm_pp_notify_wm_clock_changes(
+	const struct dc_context *ctx,
+	struct dm_pp_wm_sets_with_clock_ranges *wm_with_clock_ranges);
+
+void dm_pp_get_funcs(struct dc_context *ctx,
+		struct pp_smu_funcs *funcs);
+
+/* DAL calls this function to notify PP about completion of Mode Set.
+ * For PP it means that current DCE clocks are those which were returned
+ * by dc_service_pp_pre_dce_clock_change(), in the 'output' parameter.
+ *
+ * If the clocks are higher than before, then PP does nothing.
+ *
+ * If the clocks are lower than before, then PP reduces the voltage.
+ *
+ * \returns	true - call is successful
+ *		false - call failed
+ */
+bool dm_pp_apply_display_requirements(
+	const struct dc_context *ctx,
+	const struct dm_pp_display_configuration *pp_display_cfg);
+
+bool dm_pp_apply_power_level_change_request(
+	const struct dc_context *ctx,
+	struct dm_pp_power_level_change_request *level_change_req);
+
+bool dm_pp_apply_clock_for_voltage_request(
+	const struct dc_context *ctx,
+	struct dm_pp_clock_for_voltage_req *clock_for_voltage_req);
+
+bool dm_pp_get_static_clocks(
+	const struct dc_context *ctx,
+	struct dm_pp_static_clock_info *static_clk_info);
+
+/****** end of PP interfaces ******/
+
+struct persistent_data_flag {
+	bool save_per_link;
+	bool save_per_edid;
+};
+
+bool dm_query_extended_brightness_caps
+	(struct dc_context *ctx, enum dm_acpi_display_type display,
+			struct dm_acpi_atif_backlight_caps *pCaps);
+
+bool dm_dmcu_set_pipe(struct dc_context *ctx, unsigned int controller_id);
+
+/*
+ *
+ * print-out services
+ *
+ */
+#define dm_log_to_buffer(buffer, size, fmt, args)\
+	vsnprintf(buffer, size, fmt, args)
+
+static inline unsigned long long dm_get_timestamp(struct dc_context *ctx)
+{
+	return ktime_get_raw_ns();
+}
+
+unsigned long long dm_get_elapse_time_in_ns(struct dc_context *ctx,
+		unsigned long long current_time_stamp,
+		unsigned long long last_time_stamp);
+
+/*
+ * performance tracing
+ */
+void dm_perf_trace_timestamp(const char *func_name, unsigned int line, struct dc_context *ctx);
+
+#define PERF_TRACE()	dm_perf_trace_timestamp(__func__, __LINE__, CTX)
+#define PERF_TRACE_CTX(__CTX)	dm_perf_trace_timestamp(__func__, __LINE__, __CTX)
+
+/*
+ * SMU message tracing
+ */
+void dm_trace_smu_enter(uint32_t msg_id, uint32_t param_in, unsigned int delay, struct dc_context *ctx);
+void dm_trace_smu_exit(bool success, uint32_t response, struct dc_context *ctx);
+
+#define TRACE_SMU_MSG_DELAY(msg_id, param_in, delay, ctx)	dm_trace_smu_enter(msg_id, param_in, delay, ctx)
+#define TRACE_SMU_MSG(msg_id, param_in, ctx)	dm_trace_smu_enter(msg_id, param_in, 0, ctx)
+#define TRACE_SMU_MSG_ENTER(msg_id, param_in, ctx)	dm_trace_smu_enter(msg_id, param_in, 0, ctx)
+#define TRACE_SMU_MSG_EXIT(success, response, ctx)	dm_trace_smu_exit(success, response, ctx)
+
+/*
+ * DMUB Interfaces
+ */
+bool dm_execute_dmub_cmd(const struct dc_context *ctx, union dmub_rb_cmd *cmd, enum dm_dmub_wait_type wait_type);
+bool dm_execute_dmub_cmd_list(const struct dc_context *ctx, unsigned int count, union dmub_rb_cmd *cmd, enum dm_dmub_wait_type wait_type);
+
+/*
+ * ACPI Interfaces
+ */
+void dm_acpi_process_phy_transition_interlock(
+	const struct dc_context *ctx,
+	struct dm_process_phy_transition_init_params process_phy_transition_init_params);
+
+/*
+ * Debug and verification hooks
+ */
+
+void dm_dtn_log_begin(struct dc_context *ctx,
+	struct dc_log_buffer_ctx *log_ctx);
+void dm_dtn_log_append_v(struct dc_context *ctx,
+	struct dc_log_buffer_ctx *log_ctx,
+	const char *msg, ...);
+void dm_dtn_log_end(struct dc_context *ctx,
+	struct dc_log_buffer_ctx *log_ctx);
+
+char *dce_version_to_string(const int version);
+
+bool dc_supports_vrr(const enum dce_version v);
+
+#endif /* __DM_SERVICES_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dm_services_types.h b/drivers/gpu/drm/amd/display/dc/dm_services_types.h
new file mode 100644
index 000000000000..3b093b8699ab
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dm_services_types.h
@@ -0,0 +1,304 @@
+/*
+ * Copyright 2012-15 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DM_SERVICES_TYPES_H__
+#define __DM_SERVICES_TYPES_H__
+
+#include "os_types.h"
+#include "dc_types.h"
+
+struct pp_smu_funcs;
+
+struct dm_pp_clock_range {
+	int min_khz;
+	int max_khz;
+};
+
+enum dm_pp_clocks_state {
+	DM_PP_CLOCKS_STATE_INVALID,
+	DM_PP_CLOCKS_STATE_ULTRA_LOW,
+	DM_PP_CLOCKS_STATE_LOW,
+	DM_PP_CLOCKS_STATE_NOMINAL,
+	DM_PP_CLOCKS_STATE_PERFORMANCE,
+
+	/* Starting from DCE11, Max 8 levels of DPM state supported. */
+	DM_PP_CLOCKS_DPM_STATE_LEVEL_INVALID = DM_PP_CLOCKS_STATE_INVALID,
+	DM_PP_CLOCKS_DPM_STATE_LEVEL_0,
+	DM_PP_CLOCKS_DPM_STATE_LEVEL_1,
+	DM_PP_CLOCKS_DPM_STATE_LEVEL_2,
+	/* to be backward compatible */
+	DM_PP_CLOCKS_DPM_STATE_LEVEL_3,
+	DM_PP_CLOCKS_DPM_STATE_LEVEL_4,
+	DM_PP_CLOCKS_DPM_STATE_LEVEL_5,
+	DM_PP_CLOCKS_DPM_STATE_LEVEL_6,
+	DM_PP_CLOCKS_DPM_STATE_LEVEL_7,
+
+	DM_PP_CLOCKS_MAX_STATES
+};
+
+struct dm_pp_gpu_clock_range {
+	enum dm_pp_clocks_state clock_state;
+	struct dm_pp_clock_range sclk;
+	struct dm_pp_clock_range mclk;
+	struct dm_pp_clock_range eclk;
+	struct dm_pp_clock_range dclk;
+};
+
+enum dm_pp_clock_type {
+	DM_PP_CLOCK_TYPE_DISPLAY_CLK = 1,
+	DM_PP_CLOCK_TYPE_ENGINE_CLK, /* System clock */
+	DM_PP_CLOCK_TYPE_MEMORY_CLK,
+	DM_PP_CLOCK_TYPE_DCFCLK,
+	DM_PP_CLOCK_TYPE_DCEFCLK,
+	DM_PP_CLOCK_TYPE_SOCCLK,
+	DM_PP_CLOCK_TYPE_PIXELCLK,
+	DM_PP_CLOCK_TYPE_DISPLAYPHYCLK,
+	DM_PP_CLOCK_TYPE_DPPCLK,
+	DM_PP_CLOCK_TYPE_FCLK,
+};
+
+#define DC_DECODE_PP_CLOCK_TYPE(clk_type) \
+	(clk_type) == DM_PP_CLOCK_TYPE_DISPLAY_CLK ? "Display" : \
+	(clk_type) == DM_PP_CLOCK_TYPE_ENGINE_CLK ? "Engine" : \
+	(clk_type) == DM_PP_CLOCK_TYPE_MEMORY_CLK ? "Memory" : \
+	(clk_type) == DM_PP_CLOCK_TYPE_DCFCLK ? "DCF" : \
+	(clk_type) == DM_PP_CLOCK_TYPE_DCEFCLK ? "DCEF" : \
+	(clk_type) == DM_PP_CLOCK_TYPE_SOCCLK ? "SoC" : \
+	(clk_type) == DM_PP_CLOCK_TYPE_PIXELCLK ? "Pixel" : \
+	(clk_type) == DM_PP_CLOCK_TYPE_DISPLAYPHYCLK ? "Display PHY" : \
+	(clk_type) == DM_PP_CLOCK_TYPE_DPPCLK ? "DPP" : \
+	(clk_type) == DM_PP_CLOCK_TYPE_FCLK ? "F" : \
+	"Invalid"
+
+#define DM_PP_MAX_CLOCK_LEVELS 16
+
+struct dm_pp_clock_levels {
+	uint32_t num_levels;
+	uint32_t clocks_in_khz[DM_PP_MAX_CLOCK_LEVELS];
+};
+
+struct dm_pp_clock_with_latency {
+	uint32_t clocks_in_khz;
+	uint32_t latency_in_us;
+};
+
+struct dm_pp_clock_levels_with_latency {
+	uint32_t num_levels;
+	struct dm_pp_clock_with_latency data[DM_PP_MAX_CLOCK_LEVELS];
+};
+
+struct dm_pp_clock_with_voltage {
+	uint32_t clocks_in_khz;
+	uint32_t voltage_in_mv;
+};
+
+struct dm_pp_clock_levels_with_voltage {
+	uint32_t num_levels;
+	struct dm_pp_clock_with_voltage data[DM_PP_MAX_CLOCK_LEVELS];
+};
+
+struct dm_pp_single_disp_config {
+	enum signal_type signal;
+	uint8_t transmitter;
+	uint8_t ddi_channel_mapping;
+	uint8_t pipe_idx;
+	uint32_t src_height;
+	uint32_t src_width;
+	uint32_t v_refresh;
+	uint32_t pixel_clock; /* Pixel clock in KHz (for HDMI only: normalized) */
+	struct dc_link_settings link_settings; /* DP only */
+};
+
+#define MAX_WM_SETS 4
+
+enum dm_pp_wm_set_id {
+	WM_SET_A = 0,
+	WM_SET_B,
+	WM_SET_C,
+	WM_SET_D,
+	WM_SET_INVALID = 0xffff,
+};
+
+struct dm_pp_clock_range_for_wm_set {
+	enum dm_pp_wm_set_id wm_set_id;
+	uint32_t wm_min_eng_clk_in_khz;
+	uint32_t wm_max_eng_clk_in_khz;
+	uint32_t wm_min_mem_clk_in_khz;
+	uint32_t wm_max_mem_clk_in_khz;
+};
+
+struct dm_pp_wm_sets_with_clock_ranges {
+	uint32_t num_wm_sets;
+	struct dm_pp_clock_range_for_wm_set wm_clk_ranges[MAX_WM_SETS];
+};
+
+struct dm_pp_clock_range_for_dmif_wm_set_soc15 {
+	enum dm_pp_wm_set_id wm_set_id;
+	uint32_t wm_min_dcfclk_clk_in_khz;
+	uint32_t wm_max_dcfclk_clk_in_khz;
+	uint32_t wm_min_mem_clk_in_khz;
+	uint32_t wm_max_mem_clk_in_khz;
+};
+
+struct dm_pp_clock_range_for_mcif_wm_set_soc15 {
+	enum dm_pp_wm_set_id wm_set_id;
+	uint32_t wm_min_socclk_clk_in_khz;
+	uint32_t wm_max_socclk_clk_in_khz;
+	uint32_t wm_min_mem_clk_in_khz;
+	uint32_t wm_max_mem_clk_in_khz;
+};
+
+struct dm_pp_wm_sets_with_clock_ranges_soc15 {
+	uint32_t num_wm_dmif_sets;
+	uint32_t num_wm_mcif_sets;
+	struct dm_pp_clock_range_for_dmif_wm_set_soc15
+		wm_dmif_clocks_ranges[MAX_WM_SETS];
+	struct dm_pp_clock_range_for_mcif_wm_set_soc15
+		wm_mcif_clocks_ranges[MAX_WM_SETS];
+};
+
+#define MAX_DISPLAY_CONFIGS 6
+
+struct dm_pp_display_configuration {
+	bool nb_pstate_switch_disable;/* controls NB PState switch */
+	bool cpu_cc6_disable; /* controls CPU CState switch ( on or off) */
+	bool cpu_pstate_disable;
+	uint32_t cpu_pstate_separation_time;
+
+	uint32_t min_memory_clock_khz;
+	uint32_t min_engine_clock_khz;
+	uint32_t min_engine_clock_deep_sleep_khz;
+
+	uint32_t avail_mclk_switch_time_us;
+	uint32_t avail_mclk_switch_time_in_disp_active_us;
+	uint32_t min_dcfclock_khz;
+	uint32_t min_dcfc_deep_sleep_clock_khz;
+
+	uint32_t disp_clk_khz;
+
+	bool all_displays_in_sync;
+
+	uint8_t display_count;
+	struct dm_pp_single_disp_config disp_configs[MAX_DISPLAY_CONFIGS];
+
+	/*Controller Index of primary display - used in MCLK SMC switching hang
+	 * SW Workaround*/
+	uint8_t crtc_index;
+	/*htotal*1000/pixelclk - used in MCLK SMC switching hang SW Workaround*/
+	uint32_t line_time_in_us;
+};
+
+struct dm_bl_data_point {
+		/* Brightness level in percentage */
+		uint8_t luminance;
+		/* Brightness level as effective value in range 0-255,
+		 * corresponding to above percentage
+		 */
+		uint8_t signal_level;
+};
+
+/* Total size of the structure should not exceed 256 bytes */
+#define BL_DATA_POINTS 99
+struct dm_acpi_atif_backlight_caps {
+	uint16_t size; /* Bytes 0-1 (2 bytes) */
+	uint16_t flags; /* Byted 2-3 (2 bytes) */
+	uint8_t  error_code; /* Byte 4 */
+	uint8_t  ac_level_percentage; /* Byte 5 */
+	uint8_t  dc_level_percentage; /* Byte 6 */
+	uint8_t  min_input_signal; /* Byte 7 */
+	uint8_t  max_input_signal; /* Byte 8 */
+	uint8_t  num_data_points; /* Byte 9 */
+	struct dm_bl_data_point data_points[BL_DATA_POINTS]; /* Bytes 10-207 (198 bytes)*/
+};
+
+enum dm_acpi_display_type {
+	AcpiDisplayType_LCD1 = 0,
+	AcpiDisplayType_CRT1 = 1,
+	AcpiDisplayType_DFP1 = 3,
+	AcpiDisplayType_CRT2 = 4,
+	AcpiDisplayType_LCD2 = 5,
+	AcpiDisplayType_DFP2 = 7,
+	AcpiDisplayType_DFP3 = 9,
+	AcpiDisplayType_DFP4 = 10,
+	AcpiDisplayType_DFP5 = 11,
+	AcpiDisplayType_DFP6 = 12
+};
+
+struct dm_pp_power_level_change_request {
+	enum dm_pp_clocks_state power_level;
+};
+
+struct dm_pp_clock_for_voltage_req {
+	enum dm_pp_clock_type clk_type;
+	uint32_t clocks_in_khz;
+};
+
+struct dm_pp_static_clock_info {
+	uint32_t max_sclk_khz;
+	uint32_t max_mclk_khz;
+
+	/* max possible display block clocks state */
+	enum dm_pp_clocks_state max_clocks_state;
+};
+
+struct dtn_min_clk_info {
+	uint32_t disp_clk_khz;
+	uint32_t min_engine_clock_khz;
+	uint32_t min_memory_clock_khz;
+};
+
+enum dm_dmub_wait_type {
+	DM_DMUB_WAIT_TYPE_NO_WAIT,
+	DM_DMUB_WAIT_TYPE_WAIT,
+	DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY,
+};
+
+enum dm_acpi_transition_link_type {
+	hdmi_tmds,
+	hdmi_frl,
+	dp_8b_10b,
+	dp_128b_132b,
+	none
+};
+
+struct dm_process_phy_transition_init_params {
+	uint32_t phy_id;
+	uint8_t action;
+	uint32_t sym_clock_10khz;
+	enum signal_type signal;
+	enum dc_lane_count display_port_lanes_count;
+	enum dc_link_rate display_port_link_rate;
+	uint32_t transition_bitmask;
+	uint8_t hdmi_frl_num_lanes;
+};
+
+struct dm_process_phy_transition_input_params {
+	uint32_t phy_id;
+	uint32_t transition_id;
+	uint32_t phy_configuration;
+	uint32_t data_rate;
+};
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dml/Makefile b/drivers/gpu/drm/amd/display/dc/dml/Makefile
new file mode 100644
index 000000000000..b357683b4255
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dml/Makefile
@@ -0,0 +1,143 @@
+#
+# Copyright 2017 Advanced Micro Devices, Inc.
+# Copyright 2019 Raptor Engineering, LLC
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+#
+# Makefile for the 'utils' sub-component of DAL.
+# It provides the general basic services required by other DAL
+# subcomponents.
+
+dml_ccflags := $(CC_FLAGS_FPU)
+dml_rcflags := $(CC_FLAGS_NO_FPU)
+
+ifneq ($(CONFIG_FRAME_WARN),0)
+    ifeq ($(filter y,$(CONFIG_KASAN)$(CONFIG_KCSAN)),y)
+        frame_warn_limit := 3072
+    else
+        frame_warn_limit := 2048
+    endif
+
+    ifeq ($(call test-lt, $(CONFIG_FRAME_WARN), $(frame_warn_limit)),y)
+        frame_warn_flag := -Wframe-larger-than=$(frame_warn_limit)
+    endif
+endif
+
+CFLAGS_$(AMDDALPATH)/dc/dml/display_mode_lib.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/display_mode_vba.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn10/dcn10_fpu.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn20/dcn20_fpu.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn20/display_mode_vba_20.o := $(dml_ccflags) $(frame_warn_flag)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn20/display_rq_dlg_calc_20.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn20/display_mode_vba_20v2.o := $(dml_ccflags) $(frame_warn_flag)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn20/display_rq_dlg_calc_20v2.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn21/display_mode_vba_21.o := $(dml_ccflags) $(frame_warn_flag)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn21/display_rq_dlg_calc_21.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn30/display_mode_vba_30.o := $(dml_ccflags) $(frame_warn_flag)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn30/display_rq_dlg_calc_30.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn31/display_mode_vba_31.o := $(dml_ccflags) $(frame_warn_flag)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn31/display_rq_dlg_calc_31.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn314/display_mode_vba_314.o := $(dml_ccflags) $(frame_warn_flag)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn314/display_rq_dlg_calc_314.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn314/dcn314_fpu.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn30/dcn30_fpu.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn32/dcn32_fpu.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn32/display_mode_vba_32.o := $(dml_ccflags) $(frame_warn_flag)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn32/display_rq_dlg_calc_32.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn32/display_mode_vba_util_32.o := $(dml_ccflags) $(frame_warn_flag)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn321/dcn321_fpu.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn35/dcn35_fpu.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn351/dcn351_fpu.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn31/dcn31_fpu.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn301/dcn301_fpu.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn302/dcn302_fpu.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/dcn303/dcn303_fpu.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/dsc/rc_calc_fpu.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/calcs/dcn_calcs.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/calcs/dcn_calc_auto.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/calcs/dcn_calc_math.o := $(dml_ccflags) -Wno-tautological-compare
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/display_mode_vba.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn20/display_mode_vba_20.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn20/display_rq_dlg_calc_20.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn20/display_mode_vba_20v2.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn20/display_rq_dlg_calc_20v2.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn21/display_mode_vba_21.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn21/display_rq_dlg_calc_21.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn30/display_mode_vba_30.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn30/display_rq_dlg_calc_30.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn31/display_mode_vba_31.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn31/display_rq_dlg_calc_31.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn32/display_mode_vba_32.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn32/display_rq_dlg_calc_32.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn32/display_mode_vba_util_32.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn314/display_mode_vba_314.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn314/display_rq_dlg_calc_314.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn301/dcn301_fpu.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/display_mode_lib.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dsc/rc_calc_fpu.o  := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn10/dcn10_fpu.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn20/dcn20_fpu.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn314/display_mode_vba_314.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn314/display_rq_dlg_calc_314.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn314/dcn314_fpu.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn30/dcn30_fpu.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn32/dcn32_fpu.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn321/dcn321_fpu.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn35/dcn35_fpu.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn351/dcn351_fpu.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn31/dcn31_fpu.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn302/dcn302_fpu.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn303/dcn303_fpu.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/calcs/dcn_calc_math.o := $(dml_rcflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/dml1_display_rq_dlg_calc.o := $(dml_ccflags)
+CFLAGS_$(AMDDALPATH)/dc/dml/display_rq_dlg_helpers.o := $(dml_ccflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dml1_display_rq_dlg_calc.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/display_rq_dlg_helpers.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/calcs/dcn_calcs.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/calcs/dcn_calc_auto.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/calcs/dcn_calc_math.o := $(dml_rcflags)
+
+ifdef CONFIG_DRM_AMD_DC_FP
+DML += display_mode_lib.o display_rq_dlg_helpers.o dml1_display_rq_dlg_calc.o
+DML += dcn10/dcn10_fpu.o
+DML += dcn20/dcn20_fpu.o
+DML += display_mode_vba.o dcn20/display_rq_dlg_calc_20.o dcn20/display_mode_vba_20.o
+DML += dcn20/display_rq_dlg_calc_20v2.o dcn20/display_mode_vba_20v2.o
+DML += dcn21/display_rq_dlg_calc_21.o dcn21/display_mode_vba_21.o
+DML += dcn30/dcn30_fpu.o dcn30/display_mode_vba_30.o dcn30/display_rq_dlg_calc_30.o
+DML += dcn31/display_mode_vba_31.o dcn31/display_rq_dlg_calc_31.o
+DML += dcn314/display_mode_vba_314.o dcn314/display_rq_dlg_calc_314.o
+DML += dcn32/display_mode_vba_32.o dcn32/display_rq_dlg_calc_32.o dcn32/display_mode_vba_util_32.o
+DML += dcn31/dcn31_fpu.o
+DML += dcn32/dcn32_fpu.o
+DML += dcn321/dcn321_fpu.o
+DML += dcn301/dcn301_fpu.o
+DML += dcn302/dcn302_fpu.o
+DML += dcn303/dcn303_fpu.o
+DML += dcn314/dcn314_fpu.o
+DML += dcn35/dcn35_fpu.o
+DML += dcn351/dcn351_fpu.o
+DML += dsc/rc_calc_fpu.o
+DML += calcs/dcn_calcs.o calcs/dcn_calc_math.o calcs/dcn_calc_auto.o
+endif
+
+AMD_DAL_DML = $(addprefix $(AMDDALPATH)/dc/dml/,$(DML))
+
+AMD_DISPLAY_FILES += $(AMD_DAL_DML)
diff --git a/drivers/gpu/drm/amd/display/dc/dml/calcs/dcn_calc_auto.c b/drivers/gpu/drm/amd/display/dc/dml/calcs/dcn_calc_auto.c
new file mode 100644
index 000000000000..288d22a16cf2
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dml/calcs/dcn_calc_auto.c
@@ -0,0 +1,1933 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+#include "dcn_calc_auto.h"
+#include "dcn_calc_math.h"
+
+/*
+ * NOTE:
+ *   This file is gcc-parseable HW gospel, coming straight from HW engineers.
+ *
+ * It doesn't adhere to Linux kernel style and sometimes will do things in odd
+ * ways. Unless there is something clearly wrong with it the code should
+ * remain as-is as it provides us with a guarantee from HW that it is correct.
+ */
+
+/*REVISION#250*/
+void scaler_settings_calculation(struct dcn_bw_internal_vars *v)
+{
+	int k;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->allow_different_hratio_vratio == dcn_bw_yes) {
+			if (v->source_scan[k] == dcn_bw_hor) {
+				v->h_ratio[k] = v->viewport_width[k] / v->scaler_rec_out_width[k];
+				v->v_ratio[k] = v->viewport_height[k] / v->scaler_recout_height[k];
+			}
+			else {
+				v->h_ratio[k] = v->viewport_height[k] / v->scaler_rec_out_width[k];
+				v->v_ratio[k] = v->viewport_width[k] / v->scaler_recout_height[k];
+			}
+		}
+		else {
+			if (v->source_scan[k] == dcn_bw_hor) {
+				v->h_ratio[k] =dcn_bw_max2(v->viewport_width[k] / v->scaler_rec_out_width[k], v->viewport_height[k] / v->scaler_recout_height[k]);
+			}
+			else {
+				v->h_ratio[k] =dcn_bw_max2(v->viewport_height[k] / v->scaler_rec_out_width[k], v->viewport_width[k] / v->scaler_recout_height[k]);
+			}
+			v->v_ratio[k] = v->h_ratio[k];
+		}
+		if (v->interlace_output[k] == 1.0) {
+			v->v_ratio[k] = 2.0 * v->v_ratio[k];
+		}
+		if (v->underscan_output[k] == 1.0) {
+			v->h_ratio[k] = v->h_ratio[k] * v->under_scan_factor;
+			v->v_ratio[k] = v->v_ratio[k] * v->under_scan_factor;
+		}
+	}
+	/*scaler taps calculation*/
+
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->h_ratio[k] > 1.0) {
+			v->acceptable_quality_hta_ps =dcn_bw_min2(v->max_hscl_taps, 2.0 *dcn_bw_ceil2(v->h_ratio[k], 1.0));
+		}
+		else if (v->h_ratio[k] < 1.0) {
+			v->acceptable_quality_hta_ps = 4.0;
+		}
+		else {
+			v->acceptable_quality_hta_ps = 1.0;
+		}
+		if (v->ta_pscalculation == dcn_bw_override) {
+			v->htaps[k] = v->override_hta_ps[k];
+		}
+		else {
+			v->htaps[k] = v->acceptable_quality_hta_ps;
+		}
+		if (v->v_ratio[k] > 1.0) {
+			v->acceptable_quality_vta_ps =dcn_bw_min2(v->max_vscl_taps, 2.0 *dcn_bw_ceil2(v->v_ratio[k], 1.0));
+		}
+		else if (v->v_ratio[k] < 1.0) {
+			v->acceptable_quality_vta_ps = 4.0;
+		}
+		else {
+			v->acceptable_quality_vta_ps = 1.0;
+		}
+		if (v->ta_pscalculation == dcn_bw_override) {
+			v->vtaps[k] = v->override_vta_ps[k];
+		}
+		else {
+			v->vtaps[k] = v->acceptable_quality_vta_ps;
+		}
+		if (v->source_pixel_format[k] == dcn_bw_rgb_sub_64 || v->source_pixel_format[k] == dcn_bw_rgb_sub_32 || v->source_pixel_format[k] == dcn_bw_rgb_sub_16) {
+			v->vta_pschroma[k] = 0.0;
+			v->hta_pschroma[k] = 0.0;
+		}
+		else {
+			if (v->ta_pscalculation == dcn_bw_override) {
+				v->vta_pschroma[k] = v->override_vta_pschroma[k];
+				v->hta_pschroma[k] = v->override_hta_pschroma[k];
+			}
+			else {
+				v->vta_pschroma[k] = v->acceptable_quality_vta_ps;
+				v->hta_pschroma[k] = v->acceptable_quality_hta_ps;
+			}
+		}
+	}
+}
+
+void mode_support_and_system_configuration(struct dcn_bw_internal_vars *v)
+{
+	int i;
+	int j;
+	int k;
+	/*mode support, voltage state and soc configuration*/
+
+	/*scale ratio support check*/
+
+	v->scale_ratio_support = dcn_bw_yes;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->h_ratio[k] > v->max_hscl_ratio || v->v_ratio[k] > v->max_vscl_ratio || v->h_ratio[k] > v->htaps[k] || v->v_ratio[k] > v->vtaps[k] || (v->source_pixel_format[k] != dcn_bw_rgb_sub_64 && v->source_pixel_format[k] != dcn_bw_rgb_sub_32 && v->source_pixel_format[k] != dcn_bw_rgb_sub_16 && (v->h_ratio[k] / 2.0 > v->hta_pschroma[k] || v->v_ratio[k] / 2.0 > v->vta_pschroma[k]))) {
+			v->scale_ratio_support = dcn_bw_no;
+		}
+	}
+	/*source format, pixel format and scan support check*/
+
+	v->source_format_pixel_and_scan_support = dcn_bw_yes;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if ((v->source_surface_mode[k] == dcn_bw_sw_linear && v->source_scan[k] != dcn_bw_hor) || ((v->source_surface_mode[k] == dcn_bw_sw_4_kb_d || v->source_surface_mode[k] == dcn_bw_sw_4_kb_d_x || v->source_surface_mode[k] == dcn_bw_sw_64_kb_d || v->source_surface_mode[k] == dcn_bw_sw_64_kb_d_t || v->source_surface_mode[k] == dcn_bw_sw_64_kb_d_x || v->source_surface_mode[k] == dcn_bw_sw_var_d || v->source_surface_mode[k] == dcn_bw_sw_var_d_x) && v->source_pixel_format[k] != dcn_bw_rgb_sub_64)) {
+			v->source_format_pixel_and_scan_support = dcn_bw_no;
+		}
+	}
+	/*bandwidth support check*/
+
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->source_scan[k] == dcn_bw_hor) {
+			v->swath_width_ysingle_dpp[k] = v->viewport_width[k];
+		}
+		else {
+			v->swath_width_ysingle_dpp[k] = v->viewport_height[k];
+		}
+		if (v->source_pixel_format[k] == dcn_bw_rgb_sub_64) {
+			v->byte_per_pixel_in_dety[k] = 8.0;
+			v->byte_per_pixel_in_detc[k] = 0.0;
+		}
+		else if (v->source_pixel_format[k] == dcn_bw_rgb_sub_32) {
+			v->byte_per_pixel_in_dety[k] = 4.0;
+			v->byte_per_pixel_in_detc[k] = 0.0;
+		}
+		else if (v->source_pixel_format[k] == dcn_bw_rgb_sub_16) {
+			v->byte_per_pixel_in_dety[k] = 2.0;
+			v->byte_per_pixel_in_detc[k] = 0.0;
+		}
+		else if (v->source_pixel_format[k] == dcn_bw_yuv420_sub_8) {
+			v->byte_per_pixel_in_dety[k] = 1.0;
+			v->byte_per_pixel_in_detc[k] = 2.0;
+		}
+		else {
+			v->byte_per_pixel_in_dety[k] = 4.0f / 3.0f;
+			v->byte_per_pixel_in_detc[k] = 8.0f / 3.0f;
+		}
+	}
+	v->total_read_bandwidth_consumed_gbyte_per_second = 0.0;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		v->read_bandwidth[k] = v->swath_width_ysingle_dpp[k] * (dcn_bw_ceil2(v->byte_per_pixel_in_dety[k], 1.0) * v->v_ratio[k] +dcn_bw_ceil2(v->byte_per_pixel_in_detc[k], 2.0) / 2.0 * v->v_ratio[k] / 2) / (v->htotal[k] / v->pixel_clock[k]);
+		if (v->dcc_enable[k] == dcn_bw_yes) {
+			v->read_bandwidth[k] = v->read_bandwidth[k] * (1 + 1 / 256);
+		}
+		if (v->pte_enable == dcn_bw_yes && v->source_scan[k] != dcn_bw_hor && (v->source_surface_mode[k] == dcn_bw_sw_4_kb_s || v->source_surface_mode[k] == dcn_bw_sw_4_kb_s_x || v->source_surface_mode[k] == dcn_bw_sw_4_kb_d || v->source_surface_mode[k] == dcn_bw_sw_4_kb_d_x)) {
+			v->read_bandwidth[k] = v->read_bandwidth[k] * (1 + 1 / 64);
+		}
+		else if (v->pte_enable == dcn_bw_yes && v->source_scan[k] == dcn_bw_hor && (v->source_pixel_format[k] == dcn_bw_rgb_sub_64 || v->source_pixel_format[k] == dcn_bw_rgb_sub_32) && (v->source_surface_mode[k] == dcn_bw_sw_64_kb_s || v->source_surface_mode[k] == dcn_bw_sw_64_kb_s_t || v->source_surface_mode[k] == dcn_bw_sw_64_kb_s_x || v->source_surface_mode[k] == dcn_bw_sw_64_kb_d || v->source_surface_mode[k] == dcn_bw_sw_64_kb_d_t || v->source_surface_mode[k] == dcn_bw_sw_64_kb_d_x)) {
+			v->read_bandwidth[k] = v->read_bandwidth[k] * (1 + 1 / 256);
+		}
+		else if (v->pte_enable == dcn_bw_yes) {
+			v->read_bandwidth[k] = v->read_bandwidth[k] * (1 + 1 / 512);
+		}
+		v->total_read_bandwidth_consumed_gbyte_per_second = v->total_read_bandwidth_consumed_gbyte_per_second + v->read_bandwidth[k] / 1000.0;
+	}
+	v->total_write_bandwidth_consumed_gbyte_per_second = 0.0;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->output[k] == dcn_bw_writeback && v->output_format[k] == dcn_bw_444) {
+			v->write_bandwidth[k] = v->scaler_rec_out_width[k] / (v->htotal[k] / v->pixel_clock[k]) * 4.0;
+		}
+		else if (v->output[k] == dcn_bw_writeback) {
+			v->write_bandwidth[k] = v->scaler_rec_out_width[k] / (v->htotal[k] / v->pixel_clock[k]) * 1.5;
+		}
+		else {
+			v->write_bandwidth[k] = 0.0;
+		}
+		v->total_write_bandwidth_consumed_gbyte_per_second = v->total_write_bandwidth_consumed_gbyte_per_second + v->write_bandwidth[k] / 1000.0;
+	}
+	v->total_bandwidth_consumed_gbyte_per_second = v->total_read_bandwidth_consumed_gbyte_per_second + v->total_write_bandwidth_consumed_gbyte_per_second;
+	v->dcc_enabled_in_any_plane = dcn_bw_no;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->dcc_enable[k] == dcn_bw_yes) {
+			v->dcc_enabled_in_any_plane = dcn_bw_yes;
+		}
+	}
+	for (i = 0; i <= number_of_states_plus_one; i++) {
+		v->return_bw_todcn_per_state =dcn_bw_min2(v->return_bus_width * v->dcfclk_per_state[i], v->fabric_and_dram_bandwidth_per_state[i] * 1000.0 * v->percent_of_ideal_drambw_received_after_urg_latency / 100.0);
+		v->return_bw_per_state[i] = v->return_bw_todcn_per_state;
+		if (v->dcc_enabled_in_any_plane == dcn_bw_yes && v->return_bw_todcn_per_state > v->dcfclk_per_state[i] * v->return_bus_width / 4.0) {
+			v->return_bw_per_state[i] =dcn_bw_min2(v->return_bw_per_state[i], v->return_bw_todcn_per_state * 4.0 * (1.0 - v->urgent_latency / ((v->rob_buffer_size_in_kbyte - v->pixel_chunk_size_in_kbyte) * 1024.0 / (v->return_bw_todcn_per_state - v->dcfclk_per_state[i] * v->return_bus_width / 4.0) + v->urgent_latency)));
+		}
+		v->critical_point = 2.0 * v->return_bus_width * v->dcfclk_per_state[i] * v->urgent_latency / (v->return_bw_todcn_per_state * v->urgent_latency + (v->rob_buffer_size_in_kbyte - v->pixel_chunk_size_in_kbyte) * 1024.0);
+		if (v->dcc_enabled_in_any_plane == dcn_bw_yes && v->critical_point > 1.0 && v->critical_point < 4.0) {
+			v->return_bw_per_state[i] =dcn_bw_min2(v->return_bw_per_state[i], dcn_bw_pow(4.0 * v->return_bw_todcn_per_state * (v->rob_buffer_size_in_kbyte - v->pixel_chunk_size_in_kbyte) * 1024.0 * v->return_bus_width * v->dcfclk_per_state[i] * v->urgent_latency / (v->return_bw_todcn_per_state * v->urgent_latency + (v->rob_buffer_size_in_kbyte - v->pixel_chunk_size_in_kbyte) * 1024.0), 2));
+		}
+		v->return_bw_todcn_per_state =dcn_bw_min2(v->return_bus_width * v->dcfclk_per_state[i], v->fabric_and_dram_bandwidth_per_state[i] * 1000.0);
+		if (v->dcc_enabled_in_any_plane == dcn_bw_yes && v->return_bw_todcn_per_state > v->dcfclk_per_state[i] * v->return_bus_width / 4.0) {
+			v->return_bw_per_state[i] =dcn_bw_min2(v->return_bw_per_state[i], v->return_bw_todcn_per_state * 4.0 * (1.0 - v->urgent_latency / ((v->rob_buffer_size_in_kbyte - v->pixel_chunk_size_in_kbyte) * 1024.0 / (v->return_bw_todcn_per_state - v->dcfclk_per_state[i] * v->return_bus_width / 4.0) + v->urgent_latency)));
+		}
+		v->critical_point = 2.0 * v->return_bus_width * v->dcfclk_per_state[i] * v->urgent_latency / (v->return_bw_todcn_per_state * v->urgent_latency + (v->rob_buffer_size_in_kbyte - v->pixel_chunk_size_in_kbyte) * 1024.0);
+		if (v->dcc_enabled_in_any_plane == dcn_bw_yes && v->critical_point > 1.0 && v->critical_point < 4.0) {
+			v->return_bw_per_state[i] =dcn_bw_min2(v->return_bw_per_state[i], dcn_bw_pow(4.0 * v->return_bw_todcn_per_state * (v->rob_buffer_size_in_kbyte - v->pixel_chunk_size_in_kbyte) * 1024.0 * v->return_bus_width * v->dcfclk_per_state[i] * v->urgent_latency / (v->return_bw_todcn_per_state * v->urgent_latency + (v->rob_buffer_size_in_kbyte - v->pixel_chunk_size_in_kbyte) * 1024.0), 2));
+		}
+	}
+	for (i = 0; i <= number_of_states_plus_one; i++) {
+		if ((v->total_read_bandwidth_consumed_gbyte_per_second * 1000.0 <= v->return_bw_per_state[i]) && (v->total_bandwidth_consumed_gbyte_per_second * 1000.0 <= v->fabric_and_dram_bandwidth_per_state[i] * 1000.0 * v->percent_of_ideal_drambw_received_after_urg_latency / 100.0)) {
+			v->bandwidth_support[i] = dcn_bw_yes;
+		}
+		else {
+			v->bandwidth_support[i] = dcn_bw_no;
+		}
+	}
+	/*writeback latency support check*/
+
+	v->writeback_latency_support = dcn_bw_yes;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->output[k] == dcn_bw_writeback && v->output_format[k] == dcn_bw_444 && v->scaler_rec_out_width[k] / (v->htotal[k] / v->pixel_clock[k]) * 4.0 > (v->writeback_luma_buffer_size + v->writeback_chroma_buffer_size) * 1024.0 / v->write_back_latency) {
+			v->writeback_latency_support = dcn_bw_no;
+		}
+		else if (v->output[k] == dcn_bw_writeback && v->scaler_rec_out_width[k] / (v->htotal[k] / v->pixel_clock[k]) >dcn_bw_min2(v->writeback_luma_buffer_size, 2.0 * v->writeback_chroma_buffer_size) * 1024.0 / v->write_back_latency) {
+			v->writeback_latency_support = dcn_bw_no;
+		}
+	}
+	/*re-ordering buffer support check*/
+
+	for (i = 0; i <= number_of_states_plus_one; i++) {
+		v->urgent_round_trip_and_out_of_order_latency_per_state[i] = (v->round_trip_ping_latency_cycles + 32.0) / v->dcfclk_per_state[i] + v->urgent_out_of_order_return_per_channel * v->number_of_channels / v->return_bw_per_state[i];
+		if ((v->rob_buffer_size_in_kbyte - v->pixel_chunk_size_in_kbyte) * 1024.0 / v->return_bw_per_state[i] > v->urgent_round_trip_and_out_of_order_latency_per_state[i]) {
+			v->rob_support[i] = dcn_bw_yes;
+		}
+		else {
+			v->rob_support[i] = dcn_bw_no;
+		}
+	}
+	/*display io support check*/
+
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->output[k] == dcn_bw_dp && v->dsc_capability == dcn_bw_yes) {
+			if (v->output_format[k] == dcn_bw_420) {
+				v->required_output_bw = v->pixel_clock[k] / 2.0;
+			}
+			else {
+				v->required_output_bw = v->pixel_clock[k];
+			}
+		}
+		else if (v->output_format[k] == dcn_bw_420) {
+			v->required_output_bw = v->pixel_clock[k] * 3.0 / 2.0;
+		}
+		else {
+			v->required_output_bw = v->pixel_clock[k] * 3.0;
+		}
+		if (v->output[k] == dcn_bw_hdmi) {
+			v->required_phyclk[k] = v->required_output_bw;
+			switch (v->output_deep_color[k]) {
+			case dcn_bw_encoder_10bpc:
+				v->required_phyclk[k] =  v->required_phyclk[k] * 5.0 / 4;
+			break;
+			case dcn_bw_encoder_12bpc:
+				v->required_phyclk[k] =  v->required_phyclk[k] * 3.0 / 2;
+				break;
+			default:
+				break;
+			}
+			v->required_phyclk[k] = v->required_phyclk[k] / 3.0;
+		}
+		else if (v->output[k] == dcn_bw_dp) {
+			v->required_phyclk[k] = v->required_output_bw / 4.0;
+		}
+		else {
+			v->required_phyclk[k] = 0.0;
+		}
+	}
+	for (i = 0; i <= number_of_states_plus_one; i++) {
+		v->dio_support[i] = dcn_bw_yes;
+		for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+			if (v->required_phyclk[k] > v->phyclk_per_state[i] || (v->output[k] == dcn_bw_hdmi && v->required_phyclk[k] > 600.0)) {
+				v->dio_support[i] = dcn_bw_no;
+			}
+		}
+	}
+	/*total available writeback support check*/
+
+	v->total_number_of_active_writeback = 0.0;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->output[k] == dcn_bw_writeback) {
+			v->total_number_of_active_writeback = v->total_number_of_active_writeback + 1.0;
+		}
+	}
+	if (v->total_number_of_active_writeback <= v->max_num_writeback) {
+		v->total_available_writeback_support = dcn_bw_yes;
+	}
+	else {
+		v->total_available_writeback_support = dcn_bw_no;
+	}
+	/*maximum dispclk/dppclk support check*/
+
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->h_ratio[k] > 1.0) {
+			v->pscl_factor[k] =dcn_bw_min2(v->max_dchub_topscl_throughput, v->max_pscl_tolb_throughput * v->h_ratio[k] /dcn_bw_ceil2(v->htaps[k] / 6.0, 1.0));
+		}
+		else {
+			v->pscl_factor[k] =dcn_bw_min2(v->max_dchub_topscl_throughput, v->max_pscl_tolb_throughput);
+		}
+		if (v->byte_per_pixel_in_detc[k] == 0.0) {
+			v->pscl_factor_chroma[k] = 0.0;
+			v->min_dppclk_using_single_dpp[k] = v->pixel_clock[k] *dcn_bw_max3(v->vtaps[k] / 6.0 *dcn_bw_min2(1.0, v->h_ratio[k]), v->h_ratio[k] * v->v_ratio[k] / v->pscl_factor[k], 1.0);
+		}
+		else {
+			if (v->h_ratio[k] / 2.0 > 1.0) {
+				v->pscl_factor_chroma[k] =dcn_bw_min2(v->max_dchub_topscl_throughput, v->max_pscl_tolb_throughput * v->h_ratio[k] / 2.0 /dcn_bw_ceil2(v->hta_pschroma[k] / 6.0, 1.0));
+			}
+			else {
+				v->pscl_factor_chroma[k] =dcn_bw_min2(v->max_dchub_topscl_throughput, v->max_pscl_tolb_throughput);
+			}
+			v->min_dppclk_using_single_dpp[k] = v->pixel_clock[k] *dcn_bw_max5(v->vtaps[k] / 6.0 *dcn_bw_min2(1.0, v->h_ratio[k]), v->h_ratio[k] * v->v_ratio[k] / v->pscl_factor[k], v->vta_pschroma[k] / 6.0 *dcn_bw_min2(1.0, v->h_ratio[k] / 2.0), v->h_ratio[k] * v->v_ratio[k] / 4.0 / v->pscl_factor_chroma[k], 1.0);
+		}
+	}
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if ((v->source_pixel_format[k] == dcn_bw_rgb_sub_64 || v->source_pixel_format[k] == dcn_bw_rgb_sub_32 || v->source_pixel_format[k] == dcn_bw_rgb_sub_16)) {
+			if (v->source_surface_mode[k] == dcn_bw_sw_linear) {
+				v->read256_block_height_y[k] = 1.0;
+			}
+			else if (v->source_pixel_format[k] == dcn_bw_rgb_sub_64) {
+				v->read256_block_height_y[k] = 4.0;
+			}
+			else {
+				v->read256_block_height_y[k] = 8.0;
+			}
+			v->read256_block_width_y[k] = 256.0 /dcn_bw_ceil2(v->byte_per_pixel_in_dety[k], 1.0) / v->read256_block_height_y[k];
+			v->read256_block_height_c[k] = 0.0;
+			v->read256_block_width_c[k] = 0.0;
+		}
+		else {
+			if (v->source_surface_mode[k] == dcn_bw_sw_linear) {
+				v->read256_block_height_y[k] = 1.0;
+				v->read256_block_height_c[k] = 1.0;
+			}
+			else if (v->source_pixel_format[k] == dcn_bw_yuv420_sub_8) {
+				v->read256_block_height_y[k] = 16.0;
+				v->read256_block_height_c[k] = 8.0;
+			}
+			else {
+				v->read256_block_height_y[k] = 8.0;
+				v->read256_block_height_c[k] = 8.0;
+			}
+			v->read256_block_width_y[k] = 256.0 /dcn_bw_ceil2(v->byte_per_pixel_in_dety[k], 1.0) / v->read256_block_height_y[k];
+			v->read256_block_width_c[k] = 256.0 /dcn_bw_ceil2(v->byte_per_pixel_in_detc[k], 2.0) / v->read256_block_height_c[k];
+		}
+		if (v->source_scan[k] == dcn_bw_hor) {
+			v->max_swath_height_y[k] = v->read256_block_height_y[k];
+			v->max_swath_height_c[k] = v->read256_block_height_c[k];
+		}
+		else {
+			v->max_swath_height_y[k] = v->read256_block_width_y[k];
+			v->max_swath_height_c[k] = v->read256_block_width_c[k];
+		}
+		if ((v->source_pixel_format[k] == dcn_bw_rgb_sub_64 || v->source_pixel_format[k] == dcn_bw_rgb_sub_32 || v->source_pixel_format[k] == dcn_bw_rgb_sub_16)) {
+			if (v->source_surface_mode[k] == dcn_bw_sw_linear || (v->source_pixel_format[k] == dcn_bw_rgb_sub_64 && (v->source_surface_mode[k] == dcn_bw_sw_4_kb_s || v->source_surface_mode[k] == dcn_bw_sw_4_kb_s_x || v->source_surface_mode[k] == dcn_bw_sw_64_kb_s || v->source_surface_mode[k] == dcn_bw_sw_64_kb_s_t || v->source_surface_mode[k] == dcn_bw_sw_64_kb_s_x || v->source_surface_mode[k] == dcn_bw_sw_var_s || v->source_surface_mode[k] == dcn_bw_sw_var_s_x) && v->source_scan[k] == dcn_bw_hor)) {
+				v->min_swath_height_y[k] = v->max_swath_height_y[k];
+			}
+			else {
+				v->min_swath_height_y[k] = v->max_swath_height_y[k] / 2.0;
+			}
+			v->min_swath_height_c[k] = v->max_swath_height_c[k];
+		}
+		else {
+			if (v->source_surface_mode[k] == dcn_bw_sw_linear) {
+				v->min_swath_height_y[k] = v->max_swath_height_y[k];
+				v->min_swath_height_c[k] = v->max_swath_height_c[k];
+			}
+			else if (v->source_pixel_format[k] == dcn_bw_yuv420_sub_8 && v->source_scan[k] == dcn_bw_hor) {
+				v->min_swath_height_y[k] = v->max_swath_height_y[k] / 2.0;
+				if (v->bug_forcing_luma_and_chroma_request_to_same_size_fixed == dcn_bw_yes) {
+					v->min_swath_height_c[k] = v->max_swath_height_c[k];
+				}
+				else {
+					v->min_swath_height_c[k] = v->max_swath_height_c[k] / 2.0;
+				}
+			}
+			else if (v->source_pixel_format[k] == dcn_bw_yuv420_sub_10 && v->source_scan[k] == dcn_bw_hor) {
+				v->min_swath_height_c[k] = v->max_swath_height_c[k] / 2.0;
+				if (v->bug_forcing_luma_and_chroma_request_to_same_size_fixed == dcn_bw_yes) {
+					v->min_swath_height_y[k] = v->max_swath_height_y[k];
+				}
+				else {
+					v->min_swath_height_y[k] = v->max_swath_height_y[k] / 2.0;
+				}
+			}
+			else {
+				v->min_swath_height_y[k] = v->max_swath_height_y[k];
+				v->min_swath_height_c[k] = v->max_swath_height_c[k];
+			}
+		}
+		if (v->source_surface_mode[k] == dcn_bw_sw_linear) {
+			v->maximum_swath_width = 8192.0;
+		}
+		else {
+			v->maximum_swath_width = 5120.0;
+		}
+		v->number_of_dpp_required_for_det_size =dcn_bw_ceil2(v->swath_width_ysingle_dpp[k] /dcn_bw_min2(v->maximum_swath_width, v->det_buffer_size_in_kbyte * 1024.0 / 2.0 / (v->byte_per_pixel_in_dety[k] * v->min_swath_height_y[k] + v->byte_per_pixel_in_detc[k] / 2.0 * v->min_swath_height_c[k])), 1.0);
+		if (v->byte_per_pixel_in_detc[k] == 0.0) {
+			v->number_of_dpp_required_for_lb_size =dcn_bw_ceil2((v->vtaps[k] +dcn_bw_max2(dcn_bw_ceil2(v->v_ratio[k], 1.0) - 2, 0.0)) * v->swath_width_ysingle_dpp[k] /dcn_bw_max2(v->h_ratio[k], 1.0) * v->lb_bit_per_pixel[k] / v->line_buffer_size, 1.0);
+		}
+		else {
+			v->number_of_dpp_required_for_lb_size =dcn_bw_max2(dcn_bw_ceil2((v->vtaps[k] +dcn_bw_max2(dcn_bw_ceil2(v->v_ratio[k], 1.0) - 2, 0.0)) * v->swath_width_ysingle_dpp[k] /dcn_bw_max2(v->h_ratio[k], 1.0) * v->lb_bit_per_pixel[k] / v->line_buffer_size, 1.0),dcn_bw_ceil2((v->vta_pschroma[k] +dcn_bw_max2(dcn_bw_ceil2(v->v_ratio[k] / 2.0, 1.0) - 2, 0.0)) * v->swath_width_ysingle_dpp[k] / 2.0 /dcn_bw_max2(v->h_ratio[k] / 2.0, 1.0) * v->lb_bit_per_pixel[k] / v->line_buffer_size, 1.0));
+		}
+		v->number_of_dpp_required_for_det_and_lb_size[k] =dcn_bw_max2(v->number_of_dpp_required_for_det_size, v->number_of_dpp_required_for_lb_size);
+	}
+	for (i = 0; i <= number_of_states_plus_one; i++) {
+		for (j = 0; j <= 1; j++) {
+			v->total_number_of_active_dpp[i][j] = 0.0;
+			v->required_dispclk[i][j] = 0.0;
+			v->dispclk_dppclk_support[i][j] = dcn_bw_yes;
+			for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+				v->min_dispclk_using_single_dpp =dcn_bw_max2(v->pixel_clock[k], v->min_dppclk_using_single_dpp[k] * (j + 1)) * (1.0 + v->downspreading / 100.0);
+				if (v->odm_capability == dcn_bw_yes) {
+					v->min_dispclk_using_dual_dpp =dcn_bw_max2(v->pixel_clock[k] / 2.0, v->min_dppclk_using_single_dpp[k] / 2.0 * (j + 1)) * (1.0 + v->downspreading / 100.0);
+				}
+				else {
+					v->min_dispclk_using_dual_dpp =dcn_bw_max2(v->pixel_clock[k], v->min_dppclk_using_single_dpp[k] / 2.0 * (j + 1)) * (1.0 + v->downspreading / 100.0);
+				}
+				if (i < number_of_states) {
+					v->min_dispclk_using_single_dpp = v->min_dispclk_using_single_dpp * (1.0 + v->dispclk_ramping_margin / 100.0);
+					v->min_dispclk_using_dual_dpp = v->min_dispclk_using_dual_dpp * (1.0 + v->dispclk_ramping_margin / 100.0);
+				}
+				if (v->min_dispclk_using_single_dpp <=dcn_bw_min2(v->max_dispclk[i], (j + 1) * v->max_dppclk[i]) && v->number_of_dpp_required_for_det_and_lb_size[k] <= 1.0) {
+					v->no_of_dpp[i][j][k] = 1.0;
+					v->required_dispclk[i][j] =dcn_bw_max2(v->required_dispclk[i][j], v->min_dispclk_using_single_dpp);
+				}
+				else if (v->min_dispclk_using_dual_dpp <=dcn_bw_min2(v->max_dispclk[i], (j + 1) * v->max_dppclk[i])) {
+					v->no_of_dpp[i][j][k] = 2.0;
+					v->required_dispclk[i][j] =dcn_bw_max2(v->required_dispclk[i][j], v->min_dispclk_using_dual_dpp);
+				}
+				else {
+					v->no_of_dpp[i][j][k] = 2.0;
+					v->required_dispclk[i][j] =dcn_bw_max2(v->required_dispclk[i][j], v->min_dispclk_using_dual_dpp);
+					v->dispclk_dppclk_support[i][j] = dcn_bw_no;
+				}
+				v->total_number_of_active_dpp[i][j] = v->total_number_of_active_dpp[i][j] + v->no_of_dpp[i][j][k];
+			}
+			if (v->total_number_of_active_dpp[i][j] > v->max_num_dpp) {
+				v->total_number_of_active_dpp[i][j] = 0.0;
+				v->required_dispclk[i][j] = 0.0;
+				v->dispclk_dppclk_support[i][j] = dcn_bw_yes;
+				for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+					v->min_dispclk_using_single_dpp =dcn_bw_max2(v->pixel_clock[k], v->min_dppclk_using_single_dpp[k] * (j + 1)) * (1.0 + v->downspreading / 100.0);
+					v->min_dispclk_using_dual_dpp =dcn_bw_max2(v->pixel_clock[k], v->min_dppclk_using_single_dpp[k] / 2.0 * (j + 1)) * (1.0 + v->downspreading / 100.0);
+					if (i < number_of_states) {
+						v->min_dispclk_using_single_dpp = v->min_dispclk_using_single_dpp * (1.0 + v->dispclk_ramping_margin / 100.0);
+						v->min_dispclk_using_dual_dpp = v->min_dispclk_using_dual_dpp * (1.0 + v->dispclk_ramping_margin / 100.0);
+					}
+					if (v->number_of_dpp_required_for_det_and_lb_size[k] <= 1.0) {
+						v->no_of_dpp[i][j][k] = 1.0;
+						v->required_dispclk[i][j] =dcn_bw_max2(v->required_dispclk[i][j], v->min_dispclk_using_single_dpp);
+						if (v->min_dispclk_using_single_dpp >dcn_bw_min2(v->max_dispclk[i], (j + 1) * v->max_dppclk[i])) {
+							v->dispclk_dppclk_support[i][j] = dcn_bw_no;
+						}
+					}
+					else {
+						v->no_of_dpp[i][j][k] = 2.0;
+						v->required_dispclk[i][j] =dcn_bw_max2(v->required_dispclk[i][j], v->min_dispclk_using_dual_dpp);
+						if (v->min_dispclk_using_dual_dpp >dcn_bw_min2(v->max_dispclk[i], (j + 1) * v->max_dppclk[i])) {
+							v->dispclk_dppclk_support[i][j] = dcn_bw_no;
+						}
+					}
+					v->total_number_of_active_dpp[i][j] = v->total_number_of_active_dpp[i][j] + v->no_of_dpp[i][j][k];
+				}
+			}
+		}
+	}
+	/*viewport size check*/
+
+	v->viewport_size_support = dcn_bw_yes;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->number_of_dpp_required_for_det_and_lb_size[k] > 2.0) {
+			v->viewport_size_support = dcn_bw_no;
+		}
+	}
+	/*total available pipes support check*/
+
+	for (i = 0; i <= number_of_states_plus_one; i++) {
+		for (j = 0; j <= 1; j++) {
+			if (v->total_number_of_active_dpp[i][j] <= v->max_num_dpp) {
+				v->total_available_pipes_support[i][j] = dcn_bw_yes;
+			}
+			else {
+				v->total_available_pipes_support[i][j] = dcn_bw_no;
+			}
+		}
+	}
+	/*urgent latency support check*/
+
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		for (i = 0; i <= number_of_states_plus_one; i++) {
+			for (j = 0; j <= 1; j++) {
+				v->swath_width_yper_state[i][j][k] = v->swath_width_ysingle_dpp[k] / v->no_of_dpp[i][j][k];
+				v->swath_width_granularity_y = 256.0 /dcn_bw_ceil2(v->byte_per_pixel_in_dety[k], 1.0) / v->max_swath_height_y[k];
+				v->rounded_up_max_swath_size_bytes_y = (dcn_bw_ceil2(v->swath_width_yper_state[i][j][k] - 1.0, v->swath_width_granularity_y) + v->swath_width_granularity_y) * v->byte_per_pixel_in_dety[k] * v->max_swath_height_y[k];
+				if (v->source_pixel_format[k] == dcn_bw_yuv420_sub_10) {
+					v->rounded_up_max_swath_size_bytes_y =dcn_bw_ceil2(v->rounded_up_max_swath_size_bytes_y, 256.0) + 256;
+				}
+				if (v->max_swath_height_c[k] > 0.0) {
+					v->swath_width_granularity_c = 256.0 /dcn_bw_ceil2(v->byte_per_pixel_in_detc[k], 2.0) / v->max_swath_height_c[k];
+					v->rounded_up_max_swath_size_bytes_c = (dcn_bw_ceil2(v->swath_width_yper_state[i][j][k] / 2.0 - 1.0, v->swath_width_granularity_c) + v->swath_width_granularity_c) * v->byte_per_pixel_in_detc[k] * v->max_swath_height_c[k];
+					if (v->source_pixel_format[k] == dcn_bw_yuv420_sub_10) {
+						v->rounded_up_max_swath_size_bytes_c = dcn_bw_ceil2(v->rounded_up_max_swath_size_bytes_c, 256.0) + 256;
+					}
+				}
+				if (v->rounded_up_max_swath_size_bytes_y + v->rounded_up_max_swath_size_bytes_c <= v->det_buffer_size_in_kbyte * 1024.0 / 2.0) {
+					v->swath_height_yper_state[i][j][k] = v->max_swath_height_y[k];
+					v->swath_height_cper_state[i][j][k] = v->max_swath_height_c[k];
+				}
+				else {
+					v->swath_height_yper_state[i][j][k] = v->min_swath_height_y[k];
+					v->swath_height_cper_state[i][j][k] = v->min_swath_height_c[k];
+				}
+				if (v->byte_per_pixel_in_detc[k] == 0.0) {
+					v->lines_in_det_luma = v->det_buffer_size_in_kbyte * 1024.0 / v->byte_per_pixel_in_dety[k] / v->swath_width_yper_state[i][j][k];
+					v->lines_in_det_chroma = 0.0;
+				}
+				else if (v->swath_height_yper_state[i][j][k] <= v->swath_height_cper_state[i][j][k]) {
+					v->lines_in_det_luma = v->det_buffer_size_in_kbyte * 1024.0 / 2.0 / v->byte_per_pixel_in_dety[k] / v->swath_width_yper_state[i][j][k];
+					v->lines_in_det_chroma = v->det_buffer_size_in_kbyte * 1024.0 / 2.0 / v->byte_per_pixel_in_detc[k] / (v->swath_width_yper_state[i][j][k] / 2.0);
+				}
+				else {
+					v->lines_in_det_luma = v->det_buffer_size_in_kbyte * 1024.0 * 2.0 / 3.0 / v->byte_per_pixel_in_dety[k] / v->swath_width_yper_state[i][j][k];
+					v->lines_in_det_chroma = v->det_buffer_size_in_kbyte * 1024.0 / 3.0 / v->byte_per_pixel_in_dety[k] / (v->swath_width_yper_state[i][j][k] / 2.0);
+				}
+				v->effective_lb_latency_hiding_source_lines_luma =dcn_bw_min2(v->max_line_buffer_lines,dcn_bw_floor2(v->line_buffer_size / v->lb_bit_per_pixel[k] / (v->swath_width_yper_state[i][j][k] /dcn_bw_max2(v->h_ratio[k], 1.0)), 1.0)) - (v->vtaps[k] - 1.0);
+				v->effective_detlb_lines_luma =dcn_bw_floor2(v->lines_in_det_luma +dcn_bw_min2(v->lines_in_det_luma * v->required_dispclk[i][j] * v->byte_per_pixel_in_dety[k] * v->pscl_factor[k] / v->return_bw_per_state[i], v->effective_lb_latency_hiding_source_lines_luma), v->swath_height_yper_state[i][j][k]);
+				if (v->byte_per_pixel_in_detc[k] == 0.0) {
+					v->urgent_latency_support_us_per_state[i][j][k] = v->effective_detlb_lines_luma * (v->htotal[k] / v->pixel_clock[k]) / v->v_ratio[k] - v->effective_detlb_lines_luma * v->swath_width_yper_state[i][j][k] *dcn_bw_ceil2(v->byte_per_pixel_in_dety[k], 1.0) / (v->return_bw_per_state[i] / v->no_of_dpp[i][j][k]);
+				}
+				else {
+					v->effective_lb_latency_hiding_source_lines_chroma = dcn_bw_min2(v->max_line_buffer_lines, dcn_bw_floor2(v->line_buffer_size / v->lb_bit_per_pixel[k] / (v->swath_width_yper_state[i][j][k] / 2.0 / dcn_bw_max2(v->h_ratio[k] / 2.0, 1.0)), 1.0)) - (v->vta_pschroma[k] - 1.0);
+					v->effective_detlb_lines_chroma = dcn_bw_floor2(v->lines_in_det_chroma + dcn_bw_min2(v->lines_in_det_chroma * v->required_dispclk[i][j] * v->byte_per_pixel_in_detc[k] * v->pscl_factor_chroma[k] / v->return_bw_per_state[i], v->effective_lb_latency_hiding_source_lines_chroma), v->swath_height_cper_state[i][j][k]);
+					v->urgent_latency_support_us_per_state[i][j][k] = dcn_bw_min2(v->effective_detlb_lines_luma * (v->htotal[k] / v->pixel_clock[k]) / v->v_ratio[k] - v->effective_detlb_lines_luma * v->swath_width_yper_state[i][j][k] * dcn_bw_ceil2(v->byte_per_pixel_in_dety[k], 1.0) / (v->return_bw_per_state[i] / v->no_of_dpp[i][j][k]), v->effective_detlb_lines_chroma * (v->htotal[k] / v->pixel_clock[k]) / (v->v_ratio[k] / 2.0) - v->effective_detlb_lines_chroma * v->swath_width_yper_state[i][j][k] / 2.0 * dcn_bw_ceil2(v->byte_per_pixel_in_detc[k], 2.0) / (v->return_bw_per_state[i] / v->no_of_dpp[i][j][k]));
+				}
+			}
+		}
+	}
+	for (i = 0; i <= number_of_states_plus_one; i++) {
+		for (j = 0; j <= 1; j++) {
+			v->urgent_latency_support[i][j] = dcn_bw_yes;
+			for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+				if (v->urgent_latency_support_us_per_state[i][j][k] < v->urgent_latency / 1.0) {
+					v->urgent_latency_support[i][j] = dcn_bw_no;
+				}
+			}
+		}
+	}
+	/*prefetch check*/
+
+	for (i = 0; i <= number_of_states_plus_one; i++) {
+		for (j = 0; j <= 1; j++) {
+			v->total_number_of_dcc_active_dpp[i][j] = 0.0;
+			for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+				if (v->dcc_enable[k] == dcn_bw_yes) {
+					v->total_number_of_dcc_active_dpp[i][j] = v->total_number_of_dcc_active_dpp[i][j] + v->no_of_dpp[i][j][k];
+				}
+			}
+		}
+	}
+	for (i = 0; i <= number_of_states_plus_one; i++) {
+		for (j = 0; j <= 1; j++) {
+			v->projected_dcfclk_deep_sleep = 8.0;
+			for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+				v->projected_dcfclk_deep_sleep =dcn_bw_max2(v->projected_dcfclk_deep_sleep, v->pixel_clock[k] / 16.0);
+				if (v->byte_per_pixel_in_detc[k] == 0.0) {
+					if (v->v_ratio[k] <= 1.0) {
+						v->projected_dcfclk_deep_sleep =dcn_bw_max2(v->projected_dcfclk_deep_sleep, 1.1 *dcn_bw_ceil2(v->byte_per_pixel_in_dety[k], 1.0) / 64.0 * v->h_ratio[k] * v->pixel_clock[k] / v->no_of_dpp[i][j][k]);
+					}
+					else {
+						v->projected_dcfclk_deep_sleep =dcn_bw_max2(v->projected_dcfclk_deep_sleep, 1.1 *dcn_bw_ceil2(v->byte_per_pixel_in_dety[k], 1.0) / 64.0 * v->pscl_factor[k] * v->required_dispclk[i][j] / (1 + j));
+					}
+				}
+				else {
+					if (v->v_ratio[k] <= 1.0) {
+						v->projected_dcfclk_deep_sleep =dcn_bw_max2(v->projected_dcfclk_deep_sleep, 1.1 *dcn_bw_ceil2(v->byte_per_pixel_in_dety[k], 1.0) / 32.0 * v->h_ratio[k] * v->pixel_clock[k] / v->no_of_dpp[i][j][k]);
+					}
+					else {
+						v->projected_dcfclk_deep_sleep =dcn_bw_max2(v->projected_dcfclk_deep_sleep, 1.1 *dcn_bw_ceil2(v->byte_per_pixel_in_dety[k], 1.0) / 32.0 * v->pscl_factor[k] * v->required_dispclk[i][j] / (1 + j));
+					}
+					if (v->v_ratio[k] / 2.0 <= 1.0) {
+						v->projected_dcfclk_deep_sleep =dcn_bw_max2(v->projected_dcfclk_deep_sleep, 1.1 *dcn_bw_ceil2(v->byte_per_pixel_in_detc[k], 2.0) / 32.0 * v->h_ratio[k] / 2.0 * v->pixel_clock[k] / v->no_of_dpp[i][j][k]);
+					}
+					else {
+						v->projected_dcfclk_deep_sleep =dcn_bw_max2(v->projected_dcfclk_deep_sleep, 1.1 *dcn_bw_ceil2(v->byte_per_pixel_in_detc[k], 2.0) / 32.0 * v->pscl_factor_chroma[k] * v->required_dispclk[i][j] / (1 + j));
+					}
+				}
+			}
+			for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+				if (v->dcc_enable[k] == dcn_bw_yes) {
+					v->meta_req_height_y = 8.0 * v->read256_block_height_y[k];
+					v->meta_req_width_y = 64.0 * 256.0 /dcn_bw_ceil2(v->byte_per_pixel_in_dety[k], 1.0) / v->meta_req_height_y;
+					v->meta_surface_width_y =dcn_bw_ceil2(v->viewport_width[k] / v->no_of_dpp[i][j][k] - 1.0, v->meta_req_width_y) + v->meta_req_width_y;
+					v->meta_surface_height_y =dcn_bw_ceil2(v->viewport_height[k] - 1.0, v->meta_req_height_y) + v->meta_req_height_y;
+					if (v->pte_enable == dcn_bw_yes) {
+						v->meta_pte_bytes_per_frame_y = (dcn_bw_ceil2((v->meta_surface_width_y * v->meta_surface_height_y *dcn_bw_ceil2(v->byte_per_pixel_in_dety[k], 1.0) / 256.0 - 4096.0) / 8.0 / 4096.0, 1.0) + 1) * 64.0;
+					}
+					else {
+						v->meta_pte_bytes_per_frame_y = 0.0;
+					}
+					if (v->source_scan[k] == dcn_bw_hor) {
+						v->meta_row_bytes_y = v->meta_surface_width_y * v->meta_req_height_y *dcn_bw_ceil2(v->byte_per_pixel_in_dety[k], 1.0) / 256.0;
+					}
+					else {
+						v->meta_row_bytes_y = v->meta_surface_height_y * v->meta_req_width_y *dcn_bw_ceil2(v->byte_per_pixel_in_dety[k], 1.0) / 256.0;
+					}
+				}
+				else {
+					v->meta_pte_bytes_per_frame_y = 0.0;
+					v->meta_row_bytes_y = 0.0;
+				}
+				if (v->pte_enable == dcn_bw_yes) {
+					if (v->source_surface_mode[k] == dcn_bw_sw_linear) {
+						v->macro_tile_block_size_bytes_y = 256.0;
+						v->macro_tile_block_height_y = 1.0;
+					}
+					else if (v->source_surface_mode[k] == dcn_bw_sw_4_kb_s || v->source_surface_mode[k] == dcn_bw_sw_4_kb_s_x || v->source_surface_mode[k] == dcn_bw_sw_4_kb_d || v->source_surface_mode[k] == dcn_bw_sw_4_kb_d_x) {
+						v->macro_tile_block_size_bytes_y = 4096.0;
+						v->macro_tile_block_height_y = 4.0 * v->read256_block_height_y[k];
+					}
+					else if (v->source_surface_mode[k] == dcn_bw_sw_64_kb_s || v->source_surface_mode[k] == dcn_bw_sw_64_kb_s_t || v->source_surface_mode[k] == dcn_bw_sw_64_kb_s_x || v->source_surface_mode[k] == dcn_bw_sw_64_kb_d || v->source_surface_mode[k] == dcn_bw_sw_64_kb_d_t || v->source_surface_mode[k] == dcn_bw_sw_64_kb_d_x) {
+						v->macro_tile_block_size_bytes_y = 64.0 * 1024;
+						v->macro_tile_block_height_y = 16.0 * v->read256_block_height_y[k];
+					}
+					else {
+						v->macro_tile_block_size_bytes_y = 256.0 * 1024;
+						v->macro_tile_block_height_y = 32.0 * v->read256_block_height_y[k];
+					}
+					if (v->macro_tile_block_size_bytes_y <= 65536.0) {
+						v->data_pte_req_height_y = v->macro_tile_block_height_y;
+					}
+					else {
+						v->data_pte_req_height_y = 16.0 * v->read256_block_height_y[k];
+					}
+					v->data_pte_req_width_y = 4096.0 /dcn_bw_ceil2(v->byte_per_pixel_in_dety[k], 1.0) / v->data_pte_req_height_y * 8;
+					if (v->source_surface_mode[k] == dcn_bw_sw_linear) {
+						v->dpte_bytes_per_row_y = 64.0 * (dcn_bw_ceil2((v->viewport_width[k] / v->no_of_dpp[i][j][k] *dcn_bw_min2(128.0, dcn_bw_pow(2.0,dcn_bw_floor2(dcn_bw_log(v->pte_buffer_size_in_requests * v->data_pte_req_width_y / (v->viewport_width[k] / v->no_of_dpp[i][j][k]), 2.0), 1.0))) - 1.0) / v->data_pte_req_width_y, 1.0) + 1);
+					}
+					else if (v->source_scan[k] == dcn_bw_hor) {
+						v->dpte_bytes_per_row_y = 64.0 * (dcn_bw_ceil2((v->viewport_width[k] / v->no_of_dpp[i][j][k] - 1.0) / v->data_pte_req_width_y, 1.0) + 1);
+					}
+					else {
+						v->dpte_bytes_per_row_y = 64.0 * (dcn_bw_ceil2((v->viewport_height[k] - 1.0) / v->data_pte_req_height_y, 1.0) + 1);
+					}
+				}
+				else {
+					v->dpte_bytes_per_row_y = 0.0;
+				}
+				if ((v->source_pixel_format[k] != dcn_bw_rgb_sub_64 && v->source_pixel_format[k] != dcn_bw_rgb_sub_32 && v->source_pixel_format[k] != dcn_bw_rgb_sub_16)) {
+					if (v->dcc_enable[k] == dcn_bw_yes) {
+						v->meta_req_height_c = 8.0 * v->read256_block_height_c[k];
+						v->meta_req_width_c = 64.0 * 256.0 /dcn_bw_ceil2(v->byte_per_pixel_in_detc[k], 2.0) / v->meta_req_height_c;
+						v->meta_surface_width_c =dcn_bw_ceil2(v->viewport_width[k] / v->no_of_dpp[i][j][k] / 2.0 - 1.0, v->meta_req_width_c) + v->meta_req_width_c;
+						v->meta_surface_height_c =dcn_bw_ceil2(v->viewport_height[k] / 2.0 - 1.0, v->meta_req_height_c) + v->meta_req_height_c;
+						if (v->pte_enable == dcn_bw_yes) {
+							v->meta_pte_bytes_per_frame_c = (dcn_bw_ceil2((v->meta_surface_width_c * v->meta_surface_height_c *dcn_bw_ceil2(v->byte_per_pixel_in_detc[k], 2.0) / 256.0 - 4096.0) / 8.0 / 4096.0, 1.0) + 1) * 64.0;
+						}
+						else {
+							v->meta_pte_bytes_per_frame_c = 0.0;
+						}
+						if (v->source_scan[k] == dcn_bw_hor) {
+							v->meta_row_bytes_c = v->meta_surface_width_c * v->meta_req_height_c *dcn_bw_ceil2(v->byte_per_pixel_in_detc[k], 2.0) / 256.0;
+						}
+						else {
+							v->meta_row_bytes_c = v->meta_surface_height_c * v->meta_req_width_c *dcn_bw_ceil2(v->byte_per_pixel_in_detc[k], 2.0) / 256.0;
+						}
+					}
+					else {
+						v->meta_pte_bytes_per_frame_c = 0.0;
+						v->meta_row_bytes_c = 0.0;
+					}
+					if (v->pte_enable == dcn_bw_yes) {
+						if (v->source_surface_mode[k] == dcn_bw_sw_linear) {
+							v->macro_tile_block_size_bytes_c = 256.0;
+							v->macro_tile_block_height_c = 1.0;
+						}
+						else if (v->source_surface_mode[k] == dcn_bw_sw_4_kb_s || v->source_surface_mode[k] == dcn_bw_sw_4_kb_s_x || v->source_surface_mode[k] == dcn_bw_sw_4_kb_d || v->source_surface_mode[k] == dcn_bw_sw_4_kb_d_x) {
+							v->macro_tile_block_size_bytes_c = 4096.0;
+							v->macro_tile_block_height_c = 4.0 * v->read256_block_height_c[k];
+						}
+						else if (v->source_surface_mode[k] == dcn_bw_sw_64_kb_s || v->source_surface_mode[k] == dcn_bw_sw_64_kb_s_t || v->source_surface_mode[k] == dcn_bw_sw_64_kb_s_x || v->source_surface_mode[k] == dcn_bw_sw_64_kb_d || v->source_surface_mode[k] == dcn_bw_sw_64_kb_d_t || v->source_surface_mode[k] == dcn_bw_sw_64_kb_d_x) {
+							v->macro_tile_block_size_bytes_c = 64.0 * 1024;
+							v->macro_tile_block_height_c = 16.0 * v->read256_block_height_c[k];
+						}
+						else {
+							v->macro_tile_block_size_bytes_c = 256.0 * 1024;
+							v->macro_tile_block_height_c = 32.0 * v->read256_block_height_c[k];
+						}
+						v->macro_tile_block_width_c = v->macro_tile_block_size_bytes_c /dcn_bw_ceil2(v->byte_per_pixel_in_detc[k], 2.0) / v->macro_tile_block_height_c;
+						if (v->macro_tile_block_size_bytes_c <= 65536.0) {
+							v->data_pte_req_height_c = v->macro_tile_block_height_c;
+						}
+						else {
+							v->data_pte_req_height_c = 16.0 * v->read256_block_height_c[k];
+						}
+						v->data_pte_req_width_c = 4096.0 /dcn_bw_ceil2(v->byte_per_pixel_in_detc[k], 2.0) / v->data_pte_req_height_c * 8;
+						if (v->source_surface_mode[k] == dcn_bw_sw_linear) {
+							v->dpte_bytes_per_row_c = 64.0 * (dcn_bw_ceil2((v->viewport_width[k] / v->no_of_dpp[i][j][k] / 2.0 * dcn_bw_min2(128.0, dcn_bw_pow(2.0,dcn_bw_floor2(dcn_bw_log(v->pte_buffer_size_in_requests * v->data_pte_req_width_c / (v->viewport_width[k] / v->no_of_dpp[i][j][k] / 2.0), 2.0), 1.0))) - 1.0) / v->data_pte_req_width_c, 1.0) + 1);
+						}
+						else if (v->source_scan[k] == dcn_bw_hor) {
+							v->dpte_bytes_per_row_c = 64.0 * (dcn_bw_ceil2((v->viewport_width[k] / v->no_of_dpp[i][j][k] / 2.0 - 1.0) / v->data_pte_req_width_c, 1.0) + 1);
+						}
+						else {
+							v->dpte_bytes_per_row_c = 64.0 * (dcn_bw_ceil2((v->viewport_height[k] / 2.0 - 1.0) / v->data_pte_req_height_c, 1.0) + 1);
+						}
+					}
+					else {
+						v->dpte_bytes_per_row_c = 0.0;
+					}
+				}
+				else {
+					v->dpte_bytes_per_row_c = 0.0;
+					v->meta_pte_bytes_per_frame_c = 0.0;
+					v->meta_row_bytes_c = 0.0;
+				}
+				v->dpte_bytes_per_row[k] = v->dpte_bytes_per_row_y + v->dpte_bytes_per_row_c;
+				v->meta_pte_bytes_per_frame[k] = v->meta_pte_bytes_per_frame_y + v->meta_pte_bytes_per_frame_c;
+				v->meta_row_bytes[k] = v->meta_row_bytes_y + v->meta_row_bytes_c;
+				v->v_init_y = (v->v_ratio[k] + v->vtaps[k] + 1.0 + v->interlace_output[k] * 0.5 * v->v_ratio[k]) / 2.0;
+				v->prefill_y[k] =dcn_bw_floor2(v->v_init_y, 1.0);
+				v->max_num_sw_y[k] =dcn_bw_ceil2((v->prefill_y[k] - 1.0) / v->swath_height_yper_state[i][j][k], 1.0) + 1;
+				if (v->prefill_y[k] > 1.0) {
+					v->max_partial_sw_y =dcn_bw_mod((v->prefill_y[k] - 2.0), v->swath_height_yper_state[i][j][k]);
+				}
+				else {
+					v->max_partial_sw_y =dcn_bw_mod((v->prefill_y[k] + v->swath_height_yper_state[i][j][k] - 2.0), v->swath_height_yper_state[i][j][k]);
+				}
+				v->max_partial_sw_y =dcn_bw_max2(1.0, v->max_partial_sw_y);
+				v->prefetch_lines_y[k] = v->max_num_sw_y[k] * v->swath_height_yper_state[i][j][k] + v->max_partial_sw_y;
+				if ((v->source_pixel_format[k] != dcn_bw_rgb_sub_64 && v->source_pixel_format[k] != dcn_bw_rgb_sub_32 && v->source_pixel_format[k] != dcn_bw_rgb_sub_16)) {
+					v->v_init_c = (v->v_ratio[k] / 2.0 + v->vtaps[k] + 1.0 + v->interlace_output[k] * 0.5 * v->v_ratio[k] / 2.0) / 2.0;
+					v->prefill_c[k] =dcn_bw_floor2(v->v_init_c, 1.0);
+					v->max_num_sw_c[k] =dcn_bw_ceil2((v->prefill_c[k] - 1.0) / v->swath_height_cper_state[i][j][k], 1.0) + 1;
+					if (v->prefill_c[k] > 1.0) {
+						v->max_partial_sw_c =dcn_bw_mod((v->prefill_c[k] - 2.0), v->swath_height_cper_state[i][j][k]);
+					}
+					else {
+						v->max_partial_sw_c =dcn_bw_mod((v->prefill_c[k] + v->swath_height_cper_state[i][j][k] - 2.0), v->swath_height_cper_state[i][j][k]);
+					}
+					v->max_partial_sw_c =dcn_bw_max2(1.0, v->max_partial_sw_c);
+					v->prefetch_lines_c[k] = v->max_num_sw_c[k] * v->swath_height_cper_state[i][j][k] + v->max_partial_sw_c;
+				}
+				else {
+					v->prefetch_lines_c[k] = 0.0;
+				}
+				v->dst_x_after_scaler = 90.0 * v->pixel_clock[k] / (v->required_dispclk[i][j] / (j + 1)) + 42.0 * v->pixel_clock[k] / v->required_dispclk[i][j];
+				if (v->no_of_dpp[i][j][k] > 1.0) {
+					v->dst_x_after_scaler = v->dst_x_after_scaler + v->scaler_rec_out_width[k] / 2.0;
+				}
+				if (v->output_format[k] == dcn_bw_420) {
+					v->dst_y_after_scaler = 1.0;
+				}
+				else {
+					v->dst_y_after_scaler = 0.0;
+				}
+				v->time_calc = 24.0 / v->projected_dcfclk_deep_sleep;
+				v->v_update_offset[k][j] = dcn_bw_ceil2(v->htotal[k] / 4.0, 1.0);
+				v->total_repeater_delay = v->max_inter_dcn_tile_repeaters * (2.0 / (v->required_dispclk[i][j] / (j + 1)) + 3.0 / v->required_dispclk[i][j]);
+				v->v_update_width[k][j] = (14.0 / v->projected_dcfclk_deep_sleep + 12.0 / (v->required_dispclk[i][j] / (j + 1)) + v->total_repeater_delay) * v->pixel_clock[k];
+				v->v_ready_offset[k][j] = dcn_bw_max2(150.0 / (v->required_dispclk[i][j] / (j + 1)), v->total_repeater_delay + 20.0 / v->projected_dcfclk_deep_sleep + 10.0 / (v->required_dispclk[i][j] / (j + 1))) * v->pixel_clock[k];
+				v->time_setup = (v->v_update_offset[k][j] + v->v_update_width[k][j] + v->v_ready_offset[k][j]) / v->pixel_clock[k];
+				v->extra_latency = v->urgent_round_trip_and_out_of_order_latency_per_state[i] + (v->total_number_of_active_dpp[i][j] * v->pixel_chunk_size_in_kbyte + v->total_number_of_dcc_active_dpp[i][j] * v->meta_chunk_size) * 1024.0 / v->return_bw_per_state[i];
+				if (v->pte_enable == dcn_bw_yes) {
+					v->extra_latency = v->extra_latency + v->total_number_of_active_dpp[i][j] * v->pte_chunk_size * 1024.0 / v->return_bw_per_state[i];
+				}
+				if (v->can_vstartup_lines_exceed_vsync_plus_back_porch_lines_minus_one == dcn_bw_yes) {
+					v->maximum_vstartup = v->vtotal[k] - v->vactive[k] - 1.0;
+				}
+				else {
+					v->maximum_vstartup = v->v_sync_plus_back_porch[k] - 1.0;
+				}
+
+				do {
+					v->line_times_for_prefetch[k] = v->maximum_vstartup - v->urgent_latency / (v->htotal[k] / v->pixel_clock[k]) - (v->time_calc + v->time_setup) / (v->htotal[k] / v->pixel_clock[k]) - (v->dst_y_after_scaler + v->dst_x_after_scaler / v->htotal[k]);
+					v->line_times_for_prefetch[k] =dcn_bw_floor2(4.0 * (v->line_times_for_prefetch[k] + 0.125), 1.0) / 4;
+					v->prefetch_bw[k] = (v->meta_pte_bytes_per_frame[k] + 2.0 * v->meta_row_bytes[k] + 2.0 * v->dpte_bytes_per_row[k] + v->prefetch_lines_y[k] * v->swath_width_yper_state[i][j][k] *dcn_bw_ceil2(v->byte_per_pixel_in_dety[k], 1.0) + v->prefetch_lines_c[k] * v->swath_width_yper_state[i][j][k] / 2.0 *dcn_bw_ceil2(v->byte_per_pixel_in_detc[k], 2.0)) / (v->line_times_for_prefetch[k] * v->htotal[k] / v->pixel_clock[k]);
+
+					if (v->pte_enable == dcn_bw_yes && v->dcc_enable[k] == dcn_bw_yes) {
+						v->time_for_meta_pte_without_immediate_flip = dcn_bw_max3(
+								v->meta_pte_bytes_frame[k] / v->prefetch_bw[k],
+								v->extra_latency,
+								v->htotal[k] / v->pixel_clock[k] / 4.0);
+					} else {
+						v->time_for_meta_pte_without_immediate_flip = v->htotal[k] / v->pixel_clock[k] / 4.0;
+					}
+
+					if (v->pte_enable == dcn_bw_yes || v->dcc_enable[k] == dcn_bw_yes) {
+						v->time_for_meta_and_dpte_row_without_immediate_flip = dcn_bw_max3((
+								v->meta_row_bytes[k] + v->dpte_bytes_per_row[k]) / v->prefetch_bw[k],
+								v->htotal[k] / v->pixel_clock[k] - v->time_for_meta_pte_without_immediate_flip,
+								v->extra_latency);
+					} else {
+						v->time_for_meta_and_dpte_row_without_immediate_flip = dcn_bw_max2(
+								v->htotal[k] / v->pixel_clock[k] - v->time_for_meta_pte_without_immediate_flip,
+								v->extra_latency - v->time_for_meta_pte_with_immediate_flip);
+					}
+
+					v->lines_for_meta_pte_without_immediate_flip[k] =dcn_bw_floor2(4.0 * (v->time_for_meta_pte_without_immediate_flip / (v->htotal[k] / v->pixel_clock[k]) + 0.125), 1.0) / 4;
+					v->lines_for_meta_and_dpte_row_without_immediate_flip[k] =dcn_bw_floor2(4.0 * (v->time_for_meta_and_dpte_row_without_immediate_flip / (v->htotal[k] / v->pixel_clock[k]) + 0.125), 1.0) / 4;
+					v->maximum_vstartup = v->maximum_vstartup - 1;
+
+					if (v->lines_for_meta_pte_without_immediate_flip[k] < 32.0 && v->lines_for_meta_and_dpte_row_without_immediate_flip[k] < 16.0)
+						break;
+
+				} while(1);
+			}
+			v->bw_available_for_immediate_flip = v->return_bw_per_state[i];
+			for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+				v->bw_available_for_immediate_flip = v->bw_available_for_immediate_flip -dcn_bw_max2(v->read_bandwidth[k], v->prefetch_bw[k]);
+			}
+			for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+				v->total_immediate_flip_bytes[k] = 0.0;
+				if ((v->source_pixel_format[k] != dcn_bw_yuv420_sub_8 && v->source_pixel_format[k] != dcn_bw_yuv420_sub_10)) {
+					v->total_immediate_flip_bytes[k] = v->total_immediate_flip_bytes[k] + v->meta_pte_bytes_per_frame[k] + v->meta_row_bytes[k] + v->dpte_bytes_per_row[k];
+				}
+			}
+			for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+				if (v->pte_enable == dcn_bw_yes && v->dcc_enable[k] == dcn_bw_yes) {
+					v->time_for_meta_pte_with_immediate_flip =dcn_bw_max5(v->meta_pte_bytes_per_frame[k] / v->prefetch_bw[k], v->meta_pte_bytes_per_frame[k] * v->total_immediate_flip_bytes[k] / (v->bw_available_for_immediate_flip * (v->meta_pte_bytes_per_frame[k] + v->meta_row_bytes[k] + v->dpte_bytes_per_row[k])), v->extra_latency, v->urgent_latency, v->htotal[k] / v->pixel_clock[k] / 4.0);
+				}
+				else {
+					v->time_for_meta_pte_with_immediate_flip = v->htotal[k] / v->pixel_clock[k] / 4.0;
+				}
+				if (v->pte_enable == dcn_bw_yes || v->dcc_enable[k] == dcn_bw_yes) {
+					v->time_for_meta_and_dpte_row_with_immediate_flip =dcn_bw_max5((v->meta_row_bytes[k] + v->dpte_bytes_per_row[k]) / v->prefetch_bw[k], (v->meta_row_bytes[k] + v->dpte_bytes_per_row[k]) * v->total_immediate_flip_bytes[k] / (v->bw_available_for_immediate_flip * (v->meta_pte_bytes_per_frame[k] + v->meta_row_bytes[k] + v->dpte_bytes_per_row[k])), v->htotal[k] / v->pixel_clock[k] - v->time_for_meta_pte_with_immediate_flip, v->extra_latency, 2.0 * v->urgent_latency);
+				}
+				else {
+					v->time_for_meta_and_dpte_row_with_immediate_flip =dcn_bw_max2(v->htotal[k] / v->pixel_clock[k] - v->time_for_meta_pte_with_immediate_flip, v->extra_latency - v->time_for_meta_pte_with_immediate_flip);
+				}
+				v->lines_for_meta_pte_with_immediate_flip[k] =dcn_bw_floor2(4.0 * (v->time_for_meta_pte_with_immediate_flip / (v->htotal[k] / v->pixel_clock[k]) + 0.125), 1.0) / 4;
+				v->lines_for_meta_and_dpte_row_with_immediate_flip[k] =dcn_bw_floor2(4.0 * (v->time_for_meta_and_dpte_row_with_immediate_flip / (v->htotal[k] / v->pixel_clock[k]) + 0.125), 1.0) / 4;
+				v->line_times_to_request_prefetch_pixel_data_with_immediate_flip = v->line_times_for_prefetch[k] - v->lines_for_meta_pte_with_immediate_flip[k] - v->lines_for_meta_and_dpte_row_with_immediate_flip[k];
+				v->line_times_to_request_prefetch_pixel_data_without_immediate_flip = v->line_times_for_prefetch[k] - v->lines_for_meta_pte_without_immediate_flip[k] - v->lines_for_meta_and_dpte_row_without_immediate_flip[k];
+				if (v->line_times_to_request_prefetch_pixel_data_with_immediate_flip > 0.0) {
+					v->v_ratio_pre_ywith_immediate_flip[i][j][k] = v->prefetch_lines_y[k] / v->line_times_to_request_prefetch_pixel_data_with_immediate_flip;
+					if ((v->swath_height_yper_state[i][j][k] > 4.0)) {
+						if (v->line_times_to_request_prefetch_pixel_data_with_immediate_flip - (v->prefill_y[k] - 3.0) / 2.0 > 0.0) {
+							v->v_ratio_pre_ywith_immediate_flip[i][j][k] =dcn_bw_max2(v->v_ratio_pre_ywith_immediate_flip[i][j][k], (v->max_num_sw_y[k] * v->swath_height_yper_state[i][j][k]) / (v->line_times_to_request_prefetch_pixel_data_with_immediate_flip - (v->prefill_y[k] - 3.0) / 2.0));
+						}
+						else {
+							v->v_ratio_pre_ywith_immediate_flip[i][j][k] = 999999.0;
+						}
+					}
+					v->v_ratio_pre_cwith_immediate_flip[i][j][k] = v->prefetch_lines_c[k] / v->line_times_to_request_prefetch_pixel_data_with_immediate_flip;
+					if ((v->swath_height_cper_state[i][j][k] > 4.0)) {
+						if (v->line_times_to_request_prefetch_pixel_data_with_immediate_flip - (v->prefill_c[k] - 3.0) / 2.0 > 0.0) {
+							v->v_ratio_pre_cwith_immediate_flip[i][j][k] =dcn_bw_max2(v->v_ratio_pre_cwith_immediate_flip[i][j][k], (v->max_num_sw_c[k] * v->swath_height_cper_state[i][j][k]) / (v->line_times_to_request_prefetch_pixel_data_with_immediate_flip - (v->prefill_c[k] - 3.0) / 2.0));
+						}
+						else {
+							v->v_ratio_pre_cwith_immediate_flip[i][j][k] = 999999.0;
+						}
+					}
+					v->required_prefetch_pixel_data_bw_with_immediate_flip[i][j][k] = v->no_of_dpp[i][j][k] * (v->prefetch_lines_y[k] / v->line_times_to_request_prefetch_pixel_data_with_immediate_flip *dcn_bw_ceil2(v->byte_per_pixel_in_dety[k], 1.0) + v->prefetch_lines_c[k] / v->line_times_to_request_prefetch_pixel_data_with_immediate_flip *dcn_bw_ceil2(v->byte_per_pixel_in_detc[k], 2.0) / 2.0) * v->swath_width_yper_state[i][j][k] / (v->htotal[k] / v->pixel_clock[k]);
+				}
+				else {
+					v->v_ratio_pre_ywith_immediate_flip[i][j][k] = 999999.0;
+					v->v_ratio_pre_cwith_immediate_flip[i][j][k] = 999999.0;
+					v->required_prefetch_pixel_data_bw_with_immediate_flip[i][j][k] = 999999.0;
+				}
+				if (v->line_times_to_request_prefetch_pixel_data_without_immediate_flip > 0.0) {
+					v->v_ratio_pre_ywithout_immediate_flip[i][j][k] = v->prefetch_lines_y[k] / v->line_times_to_request_prefetch_pixel_data_without_immediate_flip;
+					if ((v->swath_height_yper_state[i][j][k] > 4.0)) {
+						if (v->line_times_to_request_prefetch_pixel_data_without_immediate_flip - (v->prefill_y[k] - 3.0) / 2.0 > 0.0) {
+							v->v_ratio_pre_ywithout_immediate_flip[i][j][k] =dcn_bw_max2(v->v_ratio_pre_ywithout_immediate_flip[i][j][k], (v->max_num_sw_y[k] * v->swath_height_yper_state[i][j][k]) / (v->line_times_to_request_prefetch_pixel_data_without_immediate_flip - (v->prefill_y[k] - 3.0) / 2.0));
+						}
+						else {
+							v->v_ratio_pre_ywithout_immediate_flip[i][j][k] = 999999.0;
+						}
+					}
+					v->v_ratio_pre_cwithout_immediate_flip[i][j][k] = v->prefetch_lines_c[k] / v->line_times_to_request_prefetch_pixel_data_without_immediate_flip;
+					if ((v->swath_height_cper_state[i][j][k] > 4.0)) {
+						if (v->line_times_to_request_prefetch_pixel_data_without_immediate_flip - (v->prefill_c[k] - 3.0) / 2.0 > 0.0) {
+							v->v_ratio_pre_cwithout_immediate_flip[i][j][k] =dcn_bw_max2(v->v_ratio_pre_cwithout_immediate_flip[i][j][k], (v->max_num_sw_c[k] * v->swath_height_cper_state[i][j][k]) / (v->line_times_to_request_prefetch_pixel_data_without_immediate_flip - (v->prefill_c[k] - 3.0) / 2.0));
+						}
+						else {
+							v->v_ratio_pre_cwithout_immediate_flip[i][j][k] = 999999.0;
+						}
+					}
+					v->required_prefetch_pixel_data_bw_without_immediate_flip[i][j][k] = v->no_of_dpp[i][j][k] * (v->prefetch_lines_y[k] / v->line_times_to_request_prefetch_pixel_data_without_immediate_flip *dcn_bw_ceil2(v->byte_per_pixel_in_dety[k], 1.0) + v->prefetch_lines_c[k] / v->line_times_to_request_prefetch_pixel_data_without_immediate_flip *dcn_bw_ceil2(v->byte_per_pixel_in_detc[k], 2.0) / 2.0) * v->swath_width_yper_state[i][j][k] / (v->htotal[k] / v->pixel_clock[k]);
+				}
+				else {
+					v->v_ratio_pre_ywithout_immediate_flip[i][j][k] = 999999.0;
+					v->v_ratio_pre_cwithout_immediate_flip[i][j][k] = 999999.0;
+					v->required_prefetch_pixel_data_bw_without_immediate_flip[i][j][k] = 999999.0;
+				}
+			}
+			v->maximum_read_bandwidth_with_prefetch_with_immediate_flip = 0.0;
+			for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+				if ((v->source_pixel_format[k] != dcn_bw_yuv420_sub_8 && v->source_pixel_format[k] != dcn_bw_yuv420_sub_10)) {
+					v->maximum_read_bandwidth_with_prefetch_with_immediate_flip = v->maximum_read_bandwidth_with_prefetch_with_immediate_flip +dcn_bw_max2(v->read_bandwidth[k], v->required_prefetch_pixel_data_bw_with_immediate_flip[i][j][k]) +dcn_bw_max2(v->meta_pte_bytes_per_frame[k] / (v->lines_for_meta_pte_with_immediate_flip[k] * v->htotal[k] / v->pixel_clock[k]), (v->meta_row_bytes[k] + v->dpte_bytes_per_row[k]) / (v->lines_for_meta_and_dpte_row_with_immediate_flip[k] * v->htotal[k] / v->pixel_clock[k]));
+				}
+				else {
+					v->maximum_read_bandwidth_with_prefetch_with_immediate_flip = v->maximum_read_bandwidth_with_prefetch_with_immediate_flip +dcn_bw_max2(v->read_bandwidth[k], v->required_prefetch_pixel_data_bw_without_immediate_flip[i][j][k]);
+				}
+			}
+			v->maximum_read_bandwidth_with_prefetch_without_immediate_flip = 0.0;
+			for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+				v->maximum_read_bandwidth_with_prefetch_without_immediate_flip = v->maximum_read_bandwidth_with_prefetch_without_immediate_flip +dcn_bw_max2(v->read_bandwidth[k], v->required_prefetch_pixel_data_bw_without_immediate_flip[i][j][k]);
+			}
+			v->prefetch_supported_with_immediate_flip[i][j] = dcn_bw_yes;
+			if (v->maximum_read_bandwidth_with_prefetch_with_immediate_flip > v->return_bw_per_state[i]) {
+				v->prefetch_supported_with_immediate_flip[i][j] = dcn_bw_no;
+			}
+			for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+				if (v->line_times_for_prefetch[k] < 2.0 || v->lines_for_meta_pte_with_immediate_flip[k] >= 8.0 || v->lines_for_meta_and_dpte_row_with_immediate_flip[k] >= 16.0) {
+					v->prefetch_supported_with_immediate_flip[i][j] = dcn_bw_no;
+				}
+			}
+			v->prefetch_supported_without_immediate_flip[i][j] = dcn_bw_yes;
+			if (v->maximum_read_bandwidth_with_prefetch_without_immediate_flip > v->return_bw_per_state[i]) {
+				v->prefetch_supported_without_immediate_flip[i][j] = dcn_bw_no;
+			}
+			for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+				if (v->line_times_for_prefetch[k] < 2.0 || v->lines_for_meta_pte_without_immediate_flip[k] >= 8.0 || v->lines_for_meta_and_dpte_row_without_immediate_flip[k] >= 16.0) {
+					v->prefetch_supported_without_immediate_flip[i][j] = dcn_bw_no;
+				}
+			}
+		}
+	}
+	for (i = 0; i <= number_of_states_plus_one; i++) {
+		for (j = 0; j <= 1; j++) {
+			v->v_ratio_in_prefetch_supported_with_immediate_flip[i][j] = dcn_bw_yes;
+			for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+				if ((((v->source_pixel_format[k] != dcn_bw_yuv420_sub_8 && v->source_pixel_format[k] != dcn_bw_yuv420_sub_10) && (v->v_ratio_pre_ywith_immediate_flip[i][j][k] > 4.0 || v->v_ratio_pre_cwith_immediate_flip[i][j][k] > 4.0)) || ((v->source_pixel_format[k] == dcn_bw_yuv420_sub_8 || v->source_pixel_format[k] == dcn_bw_yuv420_sub_10) && (v->v_ratio_pre_ywithout_immediate_flip[i][j][k] > 4.0 || v->v_ratio_pre_cwithout_immediate_flip[i][j][k] > 4.0)))) {
+					v->v_ratio_in_prefetch_supported_with_immediate_flip[i][j] = dcn_bw_no;
+				}
+			}
+			v->v_ratio_in_prefetch_supported_without_immediate_flip[i][j] = dcn_bw_yes;
+			for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+				if ((v->v_ratio_pre_ywithout_immediate_flip[i][j][k] > 4.0 || v->v_ratio_pre_cwithout_immediate_flip[i][j][k] > 4.0)) {
+					v->v_ratio_in_prefetch_supported_without_immediate_flip[i][j] = dcn_bw_no;
+				}
+			}
+		}
+	}
+	/*mode support, voltage state and soc configuration*/
+
+	for (i = number_of_states_plus_one; i >= 0; i--) {
+		for (j = 0; j <= 1; j++) {
+			if (v->scale_ratio_support == dcn_bw_yes && v->source_format_pixel_and_scan_support == dcn_bw_yes && v->viewport_size_support == dcn_bw_yes && v->bandwidth_support[i] == dcn_bw_yes && v->dio_support[i] == dcn_bw_yes && v->urgent_latency_support[i][j] == dcn_bw_yes && v->rob_support[i] == dcn_bw_yes && v->dispclk_dppclk_support[i][j] == dcn_bw_yes && v->total_available_pipes_support[i][j] == dcn_bw_yes && v->total_available_writeback_support == dcn_bw_yes && v->writeback_latency_support == dcn_bw_yes) {
+				if (v->prefetch_supported_with_immediate_flip[i][j] == dcn_bw_yes && v->v_ratio_in_prefetch_supported_with_immediate_flip[i][j] == dcn_bw_yes) {
+					v->mode_support_with_immediate_flip[i][j] = dcn_bw_yes;
+				}
+				else {
+					v->mode_support_with_immediate_flip[i][j] = dcn_bw_no;
+				}
+				if (v->prefetch_supported_without_immediate_flip[i][j] == dcn_bw_yes && v->v_ratio_in_prefetch_supported_without_immediate_flip[i][j] == dcn_bw_yes) {
+					v->mode_support_without_immediate_flip[i][j] = dcn_bw_yes;
+				}
+				else {
+					v->mode_support_without_immediate_flip[i][j] = dcn_bw_no;
+				}
+			}
+			else {
+				v->mode_support_with_immediate_flip[i][j] = dcn_bw_no;
+				v->mode_support_without_immediate_flip[i][j] = dcn_bw_no;
+			}
+		}
+	}
+	for (i = number_of_states_plus_one; i >= 0; i--) {
+		if ((i == number_of_states_plus_one || v->mode_support_with_immediate_flip[i][1] == dcn_bw_yes || v->mode_support_with_immediate_flip[i][0] == dcn_bw_yes) && i >= v->voltage_override_level) {
+			v->voltage_level_with_immediate_flip = i;
+		}
+	}
+	for (i = number_of_states_plus_one; i >= 0; i--) {
+		if ((i == number_of_states_plus_one || v->mode_support_without_immediate_flip[i][1] == dcn_bw_yes || v->mode_support_without_immediate_flip[i][0] == dcn_bw_yes) && i >= v->voltage_override_level) {
+			v->voltage_level_without_immediate_flip = i;
+		}
+	}
+	if (v->voltage_level_with_immediate_flip == number_of_states_plus_one) {
+		v->immediate_flip_supported = dcn_bw_no;
+		v->voltage_level = v->voltage_level_without_immediate_flip;
+	}
+	else {
+		v->immediate_flip_supported = dcn_bw_yes;
+		v->voltage_level = v->voltage_level_with_immediate_flip;
+	}
+	v->dcfclk = v->dcfclk_per_state[v->voltage_level];
+	v->fabric_and_dram_bandwidth = v->fabric_and_dram_bandwidth_per_state[v->voltage_level];
+	for (j = 0; j <= 1; j++) {
+		v->required_dispclk_per_ratio[j] = v->required_dispclk[v->voltage_level][j];
+		for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+			v->dpp_per_plane_per_ratio[j][k] = v->no_of_dpp[v->voltage_level][j][k];
+		}
+		v->dispclk_dppclk_support_per_ratio[j] = v->dispclk_dppclk_support[v->voltage_level][j];
+	}
+	v->max_phyclk = v->phyclk_per_state[v->voltage_level];
+}
+void display_pipe_configuration(struct dcn_bw_internal_vars *v)
+{
+	int j;
+	int k;
+	/*display pipe configuration*/
+
+	for (j = 0; j <= 1; j++) {
+		v->total_number_of_active_dpp_per_ratio[j] = 0.0;
+		for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+			v->total_number_of_active_dpp_per_ratio[j] = v->total_number_of_active_dpp_per_ratio[j] + v->dpp_per_plane_per_ratio[j][k];
+		}
+	}
+	if ((v->dispclk_dppclk_support_per_ratio[0] == dcn_bw_yes && v->dispclk_dppclk_support_per_ratio[1] == dcn_bw_no) || (v->dispclk_dppclk_support_per_ratio[0] == v->dispclk_dppclk_support_per_ratio[1] && (v->total_number_of_active_dpp_per_ratio[0] < v->total_number_of_active_dpp_per_ratio[1] || (((v->total_number_of_active_dpp_per_ratio[0] == v->total_number_of_active_dpp_per_ratio[1]) && v->required_dispclk_per_ratio[0] <= 0.5 * v->required_dispclk_per_ratio[1]))))) {
+		v->dispclk_dppclk_ratio = 1;
+		v->final_error_message = v->error_message[0];
+	}
+	else {
+		v->dispclk_dppclk_ratio = 2;
+		v->final_error_message = v->error_message[1];
+	}
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		v->dpp_per_plane[k] = v->dpp_per_plane_per_ratio[v->dispclk_dppclk_ratio - 1][k];
+	}
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->source_pixel_format[k] == dcn_bw_rgb_sub_64) {
+			v->byte_per_pix_dety = 8.0;
+			v->byte_per_pix_detc = 0.0;
+		}
+		else if (v->source_pixel_format[k] == dcn_bw_rgb_sub_32) {
+			v->byte_per_pix_dety = 4.0;
+			v->byte_per_pix_detc = 0.0;
+		}
+		else if (v->source_pixel_format[k] == dcn_bw_rgb_sub_16) {
+			v->byte_per_pix_dety = 2.0;
+			v->byte_per_pix_detc = 0.0;
+		}
+		else if (v->source_pixel_format[k] == dcn_bw_yuv420_sub_8) {
+			v->byte_per_pix_dety = 1.0;
+			v->byte_per_pix_detc = 2.0;
+		}
+		else {
+			v->byte_per_pix_dety = 4.0f / 3.0f;
+			v->byte_per_pix_detc = 8.0f / 3.0f;
+		}
+		if ((v->source_pixel_format[k] == dcn_bw_rgb_sub_64 || v->source_pixel_format[k] == dcn_bw_rgb_sub_32 || v->source_pixel_format[k] == dcn_bw_rgb_sub_16)) {
+			if (v->source_surface_mode[k] == dcn_bw_sw_linear) {
+				v->read256_bytes_block_height_y = 1.0;
+			}
+			else if (v->source_pixel_format[k] == dcn_bw_rgb_sub_64) {
+				v->read256_bytes_block_height_y = 4.0;
+			}
+			else {
+				v->read256_bytes_block_height_y = 8.0;
+			}
+			v->read256_bytes_block_width_y = 256.0 /dcn_bw_ceil2(v->byte_per_pix_dety, 1.0) / v->read256_bytes_block_height_y;
+			v->read256_bytes_block_height_c = 0.0;
+			v->read256_bytes_block_width_c = 0.0;
+		}
+		else {
+			if (v->source_surface_mode[k] == dcn_bw_sw_linear) {
+				v->read256_bytes_block_height_y = 1.0;
+				v->read256_bytes_block_height_c = 1.0;
+			}
+			else if (v->source_pixel_format[k] == dcn_bw_yuv420_sub_8) {
+				v->read256_bytes_block_height_y = 16.0;
+				v->read256_bytes_block_height_c = 8.0;
+			}
+			else {
+				v->read256_bytes_block_height_y = 8.0;
+				v->read256_bytes_block_height_c = 8.0;
+			}
+			v->read256_bytes_block_width_y = 256.0 /dcn_bw_ceil2(v->byte_per_pix_dety, 1.0) / v->read256_bytes_block_height_y;
+			v->read256_bytes_block_width_c = 256.0 /dcn_bw_ceil2(v->byte_per_pix_detc, 2.0) / v->read256_bytes_block_height_c;
+		}
+		if (v->source_scan[k] == dcn_bw_hor) {
+			v->maximum_swath_height_y = v->read256_bytes_block_height_y;
+			v->maximum_swath_height_c = v->read256_bytes_block_height_c;
+		}
+		else {
+			v->maximum_swath_height_y = v->read256_bytes_block_width_y;
+			v->maximum_swath_height_c = v->read256_bytes_block_width_c;
+		}
+		if ((v->source_pixel_format[k] == dcn_bw_rgb_sub_64 || v->source_pixel_format[k] == dcn_bw_rgb_sub_32 || v->source_pixel_format[k] == dcn_bw_rgb_sub_16)) {
+			if (v->source_surface_mode[k] == dcn_bw_sw_linear || (v->source_pixel_format[k] == dcn_bw_rgb_sub_64 && (v->source_surface_mode[k] == dcn_bw_sw_4_kb_s || v->source_surface_mode[k] == dcn_bw_sw_4_kb_s_x || v->source_surface_mode[k] == dcn_bw_sw_64_kb_s || v->source_surface_mode[k] == dcn_bw_sw_64_kb_s_t || v->source_surface_mode[k] == dcn_bw_sw_64_kb_s_x || v->source_surface_mode[k] == dcn_bw_sw_var_s || v->source_surface_mode[k] == dcn_bw_sw_var_s_x) && v->source_scan[k] == dcn_bw_hor)) {
+				v->minimum_swath_height_y = v->maximum_swath_height_y;
+			}
+			else {
+				v->minimum_swath_height_y = v->maximum_swath_height_y / 2.0;
+			}
+			v->minimum_swath_height_c = v->maximum_swath_height_c;
+		}
+		else {
+			if (v->source_surface_mode[k] == dcn_bw_sw_linear) {
+				v->minimum_swath_height_y = v->maximum_swath_height_y;
+				v->minimum_swath_height_c = v->maximum_swath_height_c;
+			}
+			else if (v->source_pixel_format[k] == dcn_bw_yuv420_sub_8 && v->source_scan[k] == dcn_bw_hor) {
+				v->minimum_swath_height_y = v->maximum_swath_height_y / 2.0;
+				if (v->bug_forcing_luma_and_chroma_request_to_same_size_fixed == dcn_bw_yes) {
+					v->minimum_swath_height_c = v->maximum_swath_height_c;
+				}
+				else {
+					v->minimum_swath_height_c = v->maximum_swath_height_c / 2.0;
+				}
+			}
+			else if (v->source_pixel_format[k] == dcn_bw_yuv420_sub_10 && v->source_scan[k] == dcn_bw_hor) {
+				v->minimum_swath_height_c = v->maximum_swath_height_c / 2.0;
+				if (v->bug_forcing_luma_and_chroma_request_to_same_size_fixed == dcn_bw_yes) {
+					v->minimum_swath_height_y = v->maximum_swath_height_y;
+				}
+				else {
+					v->minimum_swath_height_y = v->maximum_swath_height_y / 2.0;
+				}
+			}
+			else {
+				v->minimum_swath_height_y = v->maximum_swath_height_y;
+				v->minimum_swath_height_c = v->maximum_swath_height_c;
+			}
+		}
+		if (v->source_scan[k] == dcn_bw_hor) {
+			v->swath_width = v->viewport_width[k] / v->dpp_per_plane[k];
+		}
+		else {
+			v->swath_width = v->viewport_height[k] / v->dpp_per_plane[k];
+		}
+		v->swath_width_granularity_y = 256.0 /dcn_bw_ceil2(v->byte_per_pix_dety, 1.0) / v->maximum_swath_height_y;
+		v->rounded_up_max_swath_size_bytes_y = (dcn_bw_ceil2(v->swath_width - 1.0, v->swath_width_granularity_y) + v->swath_width_granularity_y) * v->byte_per_pix_dety * v->maximum_swath_height_y;
+		if (v->source_pixel_format[k] == dcn_bw_yuv420_sub_10) {
+			v->rounded_up_max_swath_size_bytes_y =dcn_bw_ceil2(v->rounded_up_max_swath_size_bytes_y, 256.0) + 256;
+		}
+		if (v->maximum_swath_height_c > 0.0) {
+			v->swath_width_granularity_c = 256.0 /dcn_bw_ceil2(v->byte_per_pix_detc, 2.0) / v->maximum_swath_height_c;
+			v->rounded_up_max_swath_size_bytes_c = (dcn_bw_ceil2(v->swath_width / 2.0 - 1.0, v->swath_width_granularity_c) + v->swath_width_granularity_c) * v->byte_per_pix_detc * v->maximum_swath_height_c;
+			if (v->source_pixel_format[k] == dcn_bw_yuv420_sub_10) {
+				v->rounded_up_max_swath_size_bytes_c = dcn_bw_ceil2(v->rounded_up_max_swath_size_bytes_c, 256.0) + 256;
+			}
+		}
+		if (v->rounded_up_max_swath_size_bytes_y + v->rounded_up_max_swath_size_bytes_c <= v->det_buffer_size_in_kbyte * 1024.0 / 2.0) {
+			v->swath_height_y[k] = v->maximum_swath_height_y;
+			v->swath_height_c[k] = v->maximum_swath_height_c;
+		}
+		else {
+			v->swath_height_y[k] = v->minimum_swath_height_y;
+			v->swath_height_c[k] = v->minimum_swath_height_c;
+		}
+		if (v->swath_height_c[k] == 0.0) {
+			v->det_buffer_size_y[k] = v->det_buffer_size_in_kbyte * 1024.0;
+			v->det_buffer_size_c[k] = 0.0;
+		}
+		else if (v->swath_height_y[k] <= v->swath_height_c[k]) {
+			v->det_buffer_size_y[k] = v->det_buffer_size_in_kbyte * 1024.0 / 2.0;
+			v->det_buffer_size_c[k] = v->det_buffer_size_in_kbyte * 1024.0 / 2.0;
+		}
+		else {
+			v->det_buffer_size_y[k] = v->det_buffer_size_in_kbyte * 1024.0 * 2.0 / 3.0;
+			v->det_buffer_size_c[k] = v->det_buffer_size_in_kbyte * 1024.0 / 3.0;
+		}
+	}
+}
+void dispclkdppclkdcfclk_deep_sleep_prefetch_parameters_watermarks_and_performance_calculation(struct dcn_bw_internal_vars *v)
+{
+	int k;
+	/*dispclk and dppclk calculation*/
+
+	v->dispclk_with_ramping = 0.0;
+	v->dispclk_without_ramping = 0.0;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->h_ratio[k] > 1.0) {
+			v->pscl_throughput[k] =dcn_bw_min2(v->max_dchub_topscl_throughput, v->max_pscl_tolb_throughput * v->h_ratio[k] /dcn_bw_ceil2(v->htaps[k] / 6.0, 1.0));
+		}
+		else {
+			v->pscl_throughput[k] =dcn_bw_min2(v->max_dchub_topscl_throughput, v->max_pscl_tolb_throughput);
+		}
+		v->dppclk_using_single_dpp_luma = v->pixel_clock[k] *dcn_bw_max3(v->vtaps[k] / 6.0 *dcn_bw_min2(1.0, v->h_ratio[k]), v->h_ratio[k] * v->v_ratio[k] / v->pscl_throughput[k], 1.0);
+		if ((v->source_pixel_format[k] != dcn_bw_yuv420_sub_8 && v->source_pixel_format[k] != dcn_bw_yuv420_sub_10)) {
+			v->pscl_throughput_chroma[k] = 0.0;
+			v->dppclk_using_single_dpp = v->dppclk_using_single_dpp_luma;
+		}
+		else {
+			if (v->h_ratio[k] > 1.0) {
+				v->pscl_throughput_chroma[k] =dcn_bw_min2(v->max_dchub_topscl_throughput, v->max_pscl_tolb_throughput * v->h_ratio[k] / 2.0 /dcn_bw_ceil2(v->hta_pschroma[k] / 6.0, 1.0));
+			}
+			else {
+				v->pscl_throughput_chroma[k] =dcn_bw_min2(v->max_dchub_topscl_throughput, v->max_pscl_tolb_throughput);
+			}
+			v->dppclk_using_single_dpp_chroma = v->pixel_clock[k] *dcn_bw_max3(v->vta_pschroma[k] / 6.0 *dcn_bw_min2(1.0, v->h_ratio[k] / 2.0), v->h_ratio[k] * v->v_ratio[k] / 4.0 / v->pscl_throughput_chroma[k], 1.0);
+			v->dppclk_using_single_dpp =dcn_bw_max2(v->dppclk_using_single_dpp_luma, v->dppclk_using_single_dpp_chroma);
+		}
+		if (v->odm_capable == dcn_bw_yes) {
+			v->dispclk_with_ramping =dcn_bw_max2(v->dispclk_with_ramping,dcn_bw_max2(v->dppclk_using_single_dpp / v->dpp_per_plane[k] * v->dispclk_dppclk_ratio, v->pixel_clock[k] / v->dpp_per_plane[k]) * (1.0 + v->downspreading / 100.0) * (1.0 + v->dispclk_ramping_margin / 100.0));
+			v->dispclk_without_ramping =dcn_bw_max2(v->dispclk_without_ramping,dcn_bw_max2(v->dppclk_using_single_dpp / v->dpp_per_plane[k] * v->dispclk_dppclk_ratio, v->pixel_clock[k] / v->dpp_per_plane[k]) * (1.0 + v->downspreading / 100.0));
+		}
+		else {
+			v->dispclk_with_ramping =dcn_bw_max2(v->dispclk_with_ramping,dcn_bw_max2(v->dppclk_using_single_dpp / v->dpp_per_plane[k] * v->dispclk_dppclk_ratio, v->pixel_clock[k]) * (1.0 + v->downspreading / 100.0) * (1.0 + v->dispclk_ramping_margin / 100.0));
+			v->dispclk_without_ramping =dcn_bw_max2(v->dispclk_without_ramping,dcn_bw_max2(v->dppclk_using_single_dpp / v->dpp_per_plane[k] * v->dispclk_dppclk_ratio, v->pixel_clock[k]) * (1.0 + v->downspreading / 100.0));
+		}
+	}
+	if (v->dispclk_without_ramping > v->max_dispclk[number_of_states]) {
+		v->dispclk = v->dispclk_without_ramping;
+	}
+	else if (v->dispclk_with_ramping > v->max_dispclk[number_of_states]) {
+		v->dispclk = v->max_dispclk[number_of_states];
+	}
+	else {
+		v->dispclk = v->dispclk_with_ramping;
+	}
+	v->dppclk = v->dispclk / v->dispclk_dppclk_ratio;
+	/*urgent watermark*/
+
+	v->return_bandwidth_to_dcn =dcn_bw_min2(v->return_bus_width * v->dcfclk, v->fabric_and_dram_bandwidth * 1000.0 * v->percent_of_ideal_drambw_received_after_urg_latency / 100.0);
+	v->dcc_enabled_any_plane = dcn_bw_no;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->dcc_enable[k] == dcn_bw_yes) {
+			v->dcc_enabled_any_plane = dcn_bw_yes;
+		}
+	}
+	v->return_bw = v->return_bandwidth_to_dcn;
+	if (v->dcc_enabled_any_plane == dcn_bw_yes && v->return_bandwidth_to_dcn > v->dcfclk * v->return_bus_width / 4.0) {
+		v->return_bw =dcn_bw_min2(v->return_bw, v->return_bandwidth_to_dcn * 4.0 * (1.0 - v->urgent_latency / ((v->rob_buffer_size_in_kbyte - v->pixel_chunk_size_in_kbyte) * 1024.0 / (v->return_bandwidth_to_dcn - v->dcfclk * v->return_bus_width / 4.0) + v->urgent_latency)));
+	}
+	v->critical_compression = 2.0 * v->return_bus_width * v->dcfclk * v->urgent_latency / (v->return_bandwidth_to_dcn * v->urgent_latency + (v->rob_buffer_size_in_kbyte - v->pixel_chunk_size_in_kbyte) * 1024.0);
+	if (v->dcc_enabled_any_plane == dcn_bw_yes && v->critical_compression > 1.0 && v->critical_compression < 4.0) {
+		v->return_bw =dcn_bw_min2(v->return_bw, dcn_bw_pow(4.0 * v->return_bandwidth_to_dcn * (v->rob_buffer_size_in_kbyte - v->pixel_chunk_size_in_kbyte) * 1024.0 * v->return_bus_width * v->dcfclk * v->urgent_latency / (v->return_bandwidth_to_dcn * v->urgent_latency + (v->rob_buffer_size_in_kbyte - v->pixel_chunk_size_in_kbyte) * 1024.0), 2));
+	}
+	v->return_bandwidth_to_dcn =dcn_bw_min2(v->return_bus_width * v->dcfclk, v->fabric_and_dram_bandwidth * 1000.0);
+	if (v->dcc_enabled_any_plane == dcn_bw_yes && v->return_bandwidth_to_dcn > v->dcfclk * v->return_bus_width / 4.0) {
+		v->return_bw =dcn_bw_min2(v->return_bw, v->return_bandwidth_to_dcn * 4.0 * (1.0 - v->urgent_latency / ((v->rob_buffer_size_in_kbyte - v->pixel_chunk_size_in_kbyte) * 1024.0 / (v->return_bandwidth_to_dcn - v->dcfclk * v->return_bus_width / 4.0) + v->urgent_latency)));
+	}
+	v->critical_compression = 2.0 * v->return_bus_width * v->dcfclk * v->urgent_latency / (v->return_bandwidth_to_dcn * v->urgent_latency + (v->rob_buffer_size_in_kbyte - v->pixel_chunk_size_in_kbyte) * 1024.0);
+	if (v->dcc_enabled_any_plane == dcn_bw_yes && v->critical_compression > 1.0 && v->critical_compression < 4.0) {
+		v->return_bw =dcn_bw_min2(v->return_bw, dcn_bw_pow(4.0 * v->return_bandwidth_to_dcn * (v->rob_buffer_size_in_kbyte - v->pixel_chunk_size_in_kbyte) * 1024.0 * v->return_bus_width * v->dcfclk * v->urgent_latency / (v->return_bandwidth_to_dcn * v->urgent_latency + (v->rob_buffer_size_in_kbyte - v->pixel_chunk_size_in_kbyte) * 1024.0), 2));
+	}
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->source_scan[k] == dcn_bw_hor) {
+			v->swath_width_y[k] = v->viewport_width[k] / v->dpp_per_plane[k];
+		}
+		else {
+			v->swath_width_y[k] = v->viewport_height[k] / v->dpp_per_plane[k];
+		}
+	}
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->source_pixel_format[k] == dcn_bw_rgb_sub_64) {
+			v->byte_per_pixel_dety[k] = 8.0;
+			v->byte_per_pixel_detc[k] = 0.0;
+		}
+		else if (v->source_pixel_format[k] == dcn_bw_rgb_sub_32) {
+			v->byte_per_pixel_dety[k] = 4.0;
+			v->byte_per_pixel_detc[k] = 0.0;
+		}
+		else if (v->source_pixel_format[k] == dcn_bw_rgb_sub_16) {
+			v->byte_per_pixel_dety[k] = 2.0;
+			v->byte_per_pixel_detc[k] = 0.0;
+		}
+		else if (v->source_pixel_format[k] == dcn_bw_yuv420_sub_8) {
+			v->byte_per_pixel_dety[k] = 1.0;
+			v->byte_per_pixel_detc[k] = 2.0;
+		}
+		else {
+			v->byte_per_pixel_dety[k] = 4.0f / 3.0f;
+			v->byte_per_pixel_detc[k] = 8.0f / 3.0f;
+		}
+	}
+	v->total_data_read_bandwidth = 0.0;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		v->read_bandwidth_plane_luma[k] = v->swath_width_y[k] * v->dpp_per_plane[k] *dcn_bw_ceil2(v->byte_per_pixel_dety[k], 1.0) / (v->htotal[k] / v->pixel_clock[k]) * v->v_ratio[k];
+		v->read_bandwidth_plane_chroma[k] = v->swath_width_y[k] / 2.0 * v->dpp_per_plane[k] *dcn_bw_ceil2(v->byte_per_pixel_detc[k], 2.0) / (v->htotal[k] / v->pixel_clock[k]) * v->v_ratio[k] / 2.0;
+		v->total_data_read_bandwidth = v->total_data_read_bandwidth + v->read_bandwidth_plane_luma[k] + v->read_bandwidth_plane_chroma[k];
+	}
+	v->total_active_dpp = 0.0;
+	v->total_dcc_active_dpp = 0.0;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		v->total_active_dpp = v->total_active_dpp + v->dpp_per_plane[k];
+		if (v->dcc_enable[k] == dcn_bw_yes) {
+			v->total_dcc_active_dpp = v->total_dcc_active_dpp + v->dpp_per_plane[k];
+		}
+	}
+	v->urgent_round_trip_and_out_of_order_latency = (v->round_trip_ping_latency_cycles + 32.0) / v->dcfclk + v->urgent_out_of_order_return_per_channel * v->number_of_channels / v->return_bw;
+	v->last_pixel_of_line_extra_watermark = 0.0;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->v_ratio[k] <= 1.0) {
+			v->display_pipe_line_delivery_time_luma[k] = v->swath_width_y[k] * v->dpp_per_plane[k] / v->h_ratio[k] / v->pixel_clock[k];
+		}
+		else {
+			v->display_pipe_line_delivery_time_luma[k] = v->swath_width_y[k] / v->pscl_throughput[k] / v->dppclk;
+		}
+		v->data_fabric_line_delivery_time_luma = v->swath_width_y[k] * v->swath_height_y[k] *dcn_bw_ceil2(v->byte_per_pixel_dety[k], 1.0) / (v->return_bw * v->read_bandwidth_plane_luma[k] / v->dpp_per_plane[k] / v->total_data_read_bandwidth);
+		v->last_pixel_of_line_extra_watermark =dcn_bw_max2(v->last_pixel_of_line_extra_watermark, v->data_fabric_line_delivery_time_luma - v->display_pipe_line_delivery_time_luma[k]);
+		if (v->byte_per_pixel_detc[k] == 0.0) {
+			v->display_pipe_line_delivery_time_chroma[k] = 0.0;
+		}
+		else {
+			if (v->v_ratio[k] / 2.0 <= 1.0) {
+				v->display_pipe_line_delivery_time_chroma[k] = v->swath_width_y[k] / 2.0 * v->dpp_per_plane[k] / (v->h_ratio[k] / 2.0) / v->pixel_clock[k];
+			}
+			else {
+				v->display_pipe_line_delivery_time_chroma[k] = v->swath_width_y[k] / 2.0 / v->pscl_throughput_chroma[k] / v->dppclk;
+			}
+			v->data_fabric_line_delivery_time_chroma = v->swath_width_y[k] / 2.0 * v->swath_height_c[k] *dcn_bw_ceil2(v->byte_per_pixel_detc[k], 2.0) / (v->return_bw * v->read_bandwidth_plane_chroma[k] / v->dpp_per_plane[k] / v->total_data_read_bandwidth);
+			v->last_pixel_of_line_extra_watermark =dcn_bw_max2(v->last_pixel_of_line_extra_watermark, v->data_fabric_line_delivery_time_chroma - v->display_pipe_line_delivery_time_chroma[k]);
+		}
+	}
+	v->urgent_extra_latency = v->urgent_round_trip_and_out_of_order_latency + (v->total_active_dpp * v->pixel_chunk_size_in_kbyte + v->total_dcc_active_dpp * v->meta_chunk_size) * 1024.0 / v->return_bw;
+	if (v->pte_enable == dcn_bw_yes) {
+		v->urgent_extra_latency = v->urgent_extra_latency + v->total_active_dpp * v->pte_chunk_size * 1024.0 / v->return_bw;
+	}
+	v->urgent_watermark = v->urgent_latency + v->last_pixel_of_line_extra_watermark + v->urgent_extra_latency;
+	v->ptemeta_urgent_watermark = v->urgent_watermark + 2.0 * v->urgent_latency;
+	/*nb p-state/dram clock change watermark*/
+
+	v->dram_clock_change_watermark = v->dram_clock_change_latency + v->urgent_watermark;
+	v->total_active_writeback = 0.0;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->output[k] == dcn_bw_writeback) {
+			v->total_active_writeback = v->total_active_writeback + 1.0;
+		}
+	}
+	if (v->total_active_writeback <= 1.0) {
+		v->writeback_dram_clock_change_watermark = v->dram_clock_change_latency + v->write_back_latency;
+	}
+	else {
+		v->writeback_dram_clock_change_watermark = v->dram_clock_change_latency + v->write_back_latency + v->writeback_chunk_size * 1024.0 / 32.0 / v->socclk;
+	}
+	/*stutter efficiency*/
+
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		v->lines_in_dety[k] = v->det_buffer_size_y[k] / v->byte_per_pixel_dety[k] / v->swath_width_y[k];
+		v->lines_in_dety_rounded_down_to_swath[k] =dcn_bw_floor2(v->lines_in_dety[k], v->swath_height_y[k]);
+		v->full_det_buffering_time_y[k] = v->lines_in_dety_rounded_down_to_swath[k] * (v->htotal[k] / v->pixel_clock[k]) / v->v_ratio[k];
+		if (v->byte_per_pixel_detc[k] > 0.0) {
+			v->lines_in_detc[k] = v->det_buffer_size_c[k] / v->byte_per_pixel_detc[k] / (v->swath_width_y[k] / 2.0);
+			v->lines_in_detc_rounded_down_to_swath[k] =dcn_bw_floor2(v->lines_in_detc[k], v->swath_height_c[k]);
+			v->full_det_buffering_time_c[k] = v->lines_in_detc_rounded_down_to_swath[k] * (v->htotal[k] / v->pixel_clock[k]) / (v->v_ratio[k] / 2.0);
+		}
+		else {
+			v->lines_in_detc[k] = 0.0;
+			v->lines_in_detc_rounded_down_to_swath[k] = 0.0;
+			v->full_det_buffering_time_c[k] = 999999.0;
+		}
+	}
+	v->min_full_det_buffering_time = 999999.0;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->full_det_buffering_time_y[k] < v->min_full_det_buffering_time) {
+			v->min_full_det_buffering_time = v->full_det_buffering_time_y[k];
+			v->frame_time_for_min_full_det_buffering_time = v->vtotal[k] * v->htotal[k] / v->pixel_clock[k];
+		}
+		if (v->full_det_buffering_time_c[k] < v->min_full_det_buffering_time) {
+			v->min_full_det_buffering_time = v->full_det_buffering_time_c[k];
+			v->frame_time_for_min_full_det_buffering_time = v->vtotal[k] * v->htotal[k] / v->pixel_clock[k];
+		}
+	}
+	v->average_read_bandwidth_gbyte_per_second = 0.0;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->dcc_enable[k] == dcn_bw_yes) {
+			v->average_read_bandwidth_gbyte_per_second = v->average_read_bandwidth_gbyte_per_second + v->read_bandwidth_plane_luma[k] / v->dcc_rate[k] / 1000.0 + v->read_bandwidth_plane_chroma[k] / v->dcc_rate[k] / 1000.0;
+		}
+		else {
+			v->average_read_bandwidth_gbyte_per_second = v->average_read_bandwidth_gbyte_per_second + v->read_bandwidth_plane_luma[k] / 1000.0 + v->read_bandwidth_plane_chroma[k] / 1000.0;
+		}
+		if (v->dcc_enable[k] == dcn_bw_yes) {
+			v->average_read_bandwidth_gbyte_per_second = v->average_read_bandwidth_gbyte_per_second + v->read_bandwidth_plane_luma[k] / 1000.0 / 256.0 + v->read_bandwidth_plane_chroma[k] / 1000.0 / 256.0;
+		}
+		if (v->pte_enable == dcn_bw_yes) {
+			v->average_read_bandwidth_gbyte_per_second = v->average_read_bandwidth_gbyte_per_second + v->read_bandwidth_plane_luma[k] / 1000.0 / 512.0 + v->read_bandwidth_plane_chroma[k] / 1000.0 / 512.0;
+		}
+	}
+	v->part_of_burst_that_fits_in_rob =dcn_bw_min2(v->min_full_det_buffering_time * v->total_data_read_bandwidth, v->rob_buffer_size_in_kbyte * 1024.0 * v->total_data_read_bandwidth / (v->average_read_bandwidth_gbyte_per_second * 1000.0));
+	v->stutter_burst_time = v->part_of_burst_that_fits_in_rob * (v->average_read_bandwidth_gbyte_per_second * 1000.0) / v->total_data_read_bandwidth / v->return_bw + (v->min_full_det_buffering_time * v->total_data_read_bandwidth - v->part_of_burst_that_fits_in_rob) / (v->dcfclk * 64.0);
+	if (v->total_active_writeback == 0.0) {
+		v->stutter_efficiency_not_including_vblank = (1.0 - (v->sr_exit_time + v->stutter_burst_time) / v->min_full_det_buffering_time) * 100.0;
+	}
+	else {
+		v->stutter_efficiency_not_including_vblank = 0.0;
+	}
+	v->smallest_vblank = 999999.0;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->synchronized_vblank == dcn_bw_yes || v->number_of_active_planes == 1) {
+			v->v_blank_time = (v->vtotal[k] - v->vactive[k]) * v->htotal[k] / v->pixel_clock[k];
+		}
+		else {
+			v->v_blank_time = 0.0;
+		}
+		v->smallest_vblank =dcn_bw_min2(v->smallest_vblank, v->v_blank_time);
+	}
+	v->stutter_efficiency = (v->stutter_efficiency_not_including_vblank / 100.0 * (v->frame_time_for_min_full_det_buffering_time - v->smallest_vblank) + v->smallest_vblank) / v->frame_time_for_min_full_det_buffering_time * 100.0;
+	/*dcfclk deep sleep*/
+
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->byte_per_pixel_detc[k] > 0.0) {
+			v->dcfclk_deep_sleep_per_plane[k] =dcn_bw_max2(1.1 * v->swath_width_y[k] *dcn_bw_ceil2(v->byte_per_pixel_dety[k], 1.0) / 32.0 / v->display_pipe_line_delivery_time_luma[k], 1.1 * v->swath_width_y[k] / 2.0 *dcn_bw_ceil2(v->byte_per_pixel_detc[k], 2.0) / 32.0 / v->display_pipe_line_delivery_time_chroma[k]);
+		}
+		else {
+			v->dcfclk_deep_sleep_per_plane[k] = 1.1 * v->swath_width_y[k] *dcn_bw_ceil2(v->byte_per_pixel_dety[k], 1.0) / 64.0 / v->display_pipe_line_delivery_time_luma[k];
+		}
+		v->dcfclk_deep_sleep_per_plane[k] =dcn_bw_max2(v->dcfclk_deep_sleep_per_plane[k], v->pixel_clock[k] / 16.0);
+	}
+	v->dcf_clk_deep_sleep = 8.0;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		v->dcf_clk_deep_sleep =dcn_bw_max2(v->dcf_clk_deep_sleep, v->dcfclk_deep_sleep_per_plane[k]);
+	}
+	/*stutter watermark*/
+
+	v->stutter_exit_watermark = v->sr_exit_time + v->last_pixel_of_line_extra_watermark + v->urgent_extra_latency + 10.0 / v->dcf_clk_deep_sleep;
+	v->stutter_enter_plus_exit_watermark = v->sr_enter_plus_exit_time + v->last_pixel_of_line_extra_watermark + v->urgent_extra_latency;
+	/*urgent latency supported*/
+
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		v->effective_det_plus_lb_lines_luma =dcn_bw_floor2(v->lines_in_dety[k] +dcn_bw_min2(v->lines_in_dety[k] * v->dppclk * v->byte_per_pixel_dety[k] * v->pscl_throughput[k] / (v->return_bw / v->dpp_per_plane[k]), v->effective_lb_latency_hiding_source_lines_luma), v->swath_height_y[k]);
+		v->urgent_latency_support_us_luma = v->effective_det_plus_lb_lines_luma * (v->htotal[k] / v->pixel_clock[k]) / v->v_ratio[k] - v->effective_det_plus_lb_lines_luma * v->swath_width_y[k] * v->byte_per_pixel_dety[k] / (v->return_bw / v->dpp_per_plane[k]);
+		if (v->byte_per_pixel_detc[k] > 0.0) {
+			v->effective_det_plus_lb_lines_chroma =dcn_bw_floor2(v->lines_in_detc[k] +dcn_bw_min2(v->lines_in_detc[k] * v->dppclk * v->byte_per_pixel_detc[k] * v->pscl_throughput_chroma[k] / (v->return_bw / v->dpp_per_plane[k]), v->effective_lb_latency_hiding_source_lines_chroma), v->swath_height_c[k]);
+			v->urgent_latency_support_us_chroma = v->effective_det_plus_lb_lines_chroma * (v->htotal[k] / v->pixel_clock[k]) / (v->v_ratio[k] / 2.0) - v->effective_det_plus_lb_lines_chroma * (v->swath_width_y[k] / 2.0) * v->byte_per_pixel_detc[k] / (v->return_bw / v->dpp_per_plane[k]);
+			v->urgent_latency_support_us[k] =dcn_bw_min2(v->urgent_latency_support_us_luma, v->urgent_latency_support_us_chroma);
+		}
+		else {
+			v->urgent_latency_support_us[k] = v->urgent_latency_support_us_luma;
+		}
+	}
+	v->min_urgent_latency_support_us = 999999.0;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		v->min_urgent_latency_support_us =dcn_bw_min2(v->min_urgent_latency_support_us, v->urgent_latency_support_us[k]);
+	}
+	/*non-urgent latency tolerance*/
+
+	v->non_urgent_latency_tolerance = v->min_urgent_latency_support_us - v->urgent_watermark;
+	/*prefetch*/
+
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if ((v->source_pixel_format[k] == dcn_bw_rgb_sub_64 || v->source_pixel_format[k] == dcn_bw_rgb_sub_32 || v->source_pixel_format[k] == dcn_bw_rgb_sub_16)) {
+			if (v->source_surface_mode[k] == dcn_bw_sw_linear) {
+				v->block_height256_bytes_y = 1.0;
+			}
+			else if (v->source_pixel_format[k] == dcn_bw_rgb_sub_64) {
+				v->block_height256_bytes_y = 4.0;
+			}
+			else {
+				v->block_height256_bytes_y = 8.0;
+			}
+			v->block_height256_bytes_c = 0.0;
+		}
+		else {
+			if (v->source_surface_mode[k] == dcn_bw_sw_linear) {
+				v->block_height256_bytes_y = 1.0;
+				v->block_height256_bytes_c = 1.0;
+			}
+			else if (v->source_pixel_format[k] == dcn_bw_yuv420_sub_8) {
+				v->block_height256_bytes_y = 16.0;
+				v->block_height256_bytes_c = 8.0;
+			}
+			else {
+				v->block_height256_bytes_y = 8.0;
+				v->block_height256_bytes_c = 8.0;
+			}
+		}
+		if (v->dcc_enable[k] == dcn_bw_yes) {
+			v->meta_request_width_y = 64.0 * 256.0 /dcn_bw_ceil2(v->byte_per_pixel_dety[k], 1.0) / (8.0 * v->block_height256_bytes_y);
+			v->meta_surf_width_y =dcn_bw_ceil2(v->swath_width_y[k] - 1.0, v->meta_request_width_y) + v->meta_request_width_y;
+			v->meta_surf_height_y =dcn_bw_ceil2(v->viewport_height[k] - 1.0, 8.0 * v->block_height256_bytes_y) + 8.0 * v->block_height256_bytes_y;
+			if (v->pte_enable == dcn_bw_yes) {
+				v->meta_pte_bytes_frame_y = (dcn_bw_ceil2((v->meta_surf_width_y * v->meta_surf_height_y *dcn_bw_ceil2(v->byte_per_pixel_dety[k], 1.0) / 256.0 - 4096.0) / 8.0 / 4096.0, 1.0) + 1) * 64.0;
+			}
+			else {
+				v->meta_pte_bytes_frame_y = 0.0;
+			}
+			if (v->source_scan[k] == dcn_bw_hor) {
+				v->meta_row_byte_y = v->meta_surf_width_y * 8.0 * v->block_height256_bytes_y *dcn_bw_ceil2(v->byte_per_pixel_dety[k], 1.0) / 256.0;
+			}
+			else {
+				v->meta_row_byte_y = v->meta_surf_height_y * v->meta_request_width_y *dcn_bw_ceil2(v->byte_per_pixel_dety[k], 1.0) / 256.0;
+			}
+		}
+		else {
+			v->meta_pte_bytes_frame_y = 0.0;
+			v->meta_row_byte_y = 0.0;
+		}
+		if (v->pte_enable == dcn_bw_yes) {
+			if (v->source_surface_mode[k] == dcn_bw_sw_linear) {
+				v->macro_tile_size_byte_y = 256.0;
+				v->macro_tile_height_y = 1.0;
+			}
+			else if (v->source_surface_mode[k] == dcn_bw_sw_4_kb_s || v->source_surface_mode[k] == dcn_bw_sw_4_kb_s_x || v->source_surface_mode[k] == dcn_bw_sw_4_kb_d || v->source_surface_mode[k] == dcn_bw_sw_4_kb_d_x) {
+				v->macro_tile_size_byte_y = 4096.0;
+				v->macro_tile_height_y = 4.0 * v->block_height256_bytes_y;
+			}
+			else if (v->source_surface_mode[k] == dcn_bw_sw_64_kb_s || v->source_surface_mode[k] == dcn_bw_sw_64_kb_s_t || v->source_surface_mode[k] == dcn_bw_sw_64_kb_s_x || v->source_surface_mode[k] == dcn_bw_sw_64_kb_d || v->source_surface_mode[k] == dcn_bw_sw_64_kb_d_t || v->source_surface_mode[k] == dcn_bw_sw_64_kb_d_x) {
+				v->macro_tile_size_byte_y = 64.0 * 1024;
+				v->macro_tile_height_y = 16.0 * v->block_height256_bytes_y;
+			}
+			else {
+				v->macro_tile_size_byte_y = 256.0 * 1024;
+				v->macro_tile_height_y = 32.0 * v->block_height256_bytes_y;
+			}
+			if (v->macro_tile_size_byte_y <= 65536.0) {
+				v->pixel_pte_req_height_y = v->macro_tile_height_y;
+			}
+			else {
+				v->pixel_pte_req_height_y = 16.0 * v->block_height256_bytes_y;
+			}
+			v->pixel_pte_req_width_y = 4096.0 /dcn_bw_ceil2(v->byte_per_pixel_dety[k], 1.0) / v->pixel_pte_req_height_y * 8;
+			if (v->source_surface_mode[k] == dcn_bw_sw_linear) {
+				v->pixel_pte_bytes_per_row_y = 64.0 * (dcn_bw_ceil2((v->swath_width_y[k] *dcn_bw_min2(128.0, dcn_bw_pow(2.0,dcn_bw_floor2(dcn_bw_log(v->pte_buffer_size_in_requests * v->pixel_pte_req_width_y / v->swath_width_y[k], 2.0), 1.0))) - 1.0) / v->pixel_pte_req_width_y, 1.0) + 1);
+			}
+			else if (v->source_scan[k] == dcn_bw_hor) {
+				v->pixel_pte_bytes_per_row_y = 64.0 * (dcn_bw_ceil2((v->swath_width_y[k] - 1.0) / v->pixel_pte_req_width_y, 1.0) + 1);
+			}
+			else {
+				v->pixel_pte_bytes_per_row_y = 64.0 * (dcn_bw_ceil2((v->viewport_height[k] - 1.0) / v->pixel_pte_req_height_y, 1.0) + 1);
+			}
+		}
+		else {
+			v->pixel_pte_bytes_per_row_y = 0.0;
+		}
+		if ((v->source_pixel_format[k] != dcn_bw_rgb_sub_64 && v->source_pixel_format[k] != dcn_bw_rgb_sub_32 && v->source_pixel_format[k] != dcn_bw_rgb_sub_16)) {
+			if (v->dcc_enable[k] == dcn_bw_yes) {
+				v->meta_request_width_c = 64.0 * 256.0 /dcn_bw_ceil2(v->byte_per_pixel_detc[k], 2.0) / (8.0 * v->block_height256_bytes_c);
+				v->meta_surf_width_c =dcn_bw_ceil2(v->swath_width_y[k] / 2.0 - 1.0, v->meta_request_width_c) + v->meta_request_width_c;
+				v->meta_surf_height_c =dcn_bw_ceil2(v->viewport_height[k] / 2.0 - 1.0, 8.0 * v->block_height256_bytes_c) + 8.0 * v->block_height256_bytes_c;
+				if (v->pte_enable == dcn_bw_yes) {
+					v->meta_pte_bytes_frame_c = (dcn_bw_ceil2((v->meta_surf_width_c * v->meta_surf_height_c *dcn_bw_ceil2(v->byte_per_pixel_detc[k], 2.0) / 256.0 - 4096.0) / 8.0 / 4096.0, 1.0) + 1) * 64.0;
+				}
+				else {
+					v->meta_pte_bytes_frame_c = 0.0;
+				}
+				if (v->source_scan[k] == dcn_bw_hor) {
+					v->meta_row_byte_c = v->meta_surf_width_c * 8.0 * v->block_height256_bytes_c *dcn_bw_ceil2(v->byte_per_pixel_detc[k], 2.0) / 256.0;
+				}
+				else {
+					v->meta_row_byte_c = v->meta_surf_height_c * v->meta_request_width_c *dcn_bw_ceil2(v->byte_per_pixel_detc[k], 2.0) / 256.0;
+				}
+			}
+			else {
+				v->meta_pte_bytes_frame_c = 0.0;
+				v->meta_row_byte_c = 0.0;
+			}
+			if (v->pte_enable == dcn_bw_yes) {
+				if (v->source_surface_mode[k] == dcn_bw_sw_linear) {
+					v->macro_tile_size_bytes_c = 256.0;
+					v->macro_tile_height_c = 1.0;
+				}
+				else if (v->source_surface_mode[k] == dcn_bw_sw_4_kb_s || v->source_surface_mode[k] == dcn_bw_sw_4_kb_s_x || v->source_surface_mode[k] == dcn_bw_sw_4_kb_d || v->source_surface_mode[k] == dcn_bw_sw_4_kb_d_x) {
+					v->macro_tile_size_bytes_c = 4096.0;
+					v->macro_tile_height_c = 4.0 * v->block_height256_bytes_c;
+				}
+				else if (v->source_surface_mode[k] == dcn_bw_sw_64_kb_s || v->source_surface_mode[k] == dcn_bw_sw_64_kb_s_t || v->source_surface_mode[k] == dcn_bw_sw_64_kb_s_x || v->source_surface_mode[k] == dcn_bw_sw_64_kb_d || v->source_surface_mode[k] == dcn_bw_sw_64_kb_d_t || v->source_surface_mode[k] == dcn_bw_sw_64_kb_d_x) {
+					v->macro_tile_size_bytes_c = 64.0 * 1024;
+					v->macro_tile_height_c = 16.0 * v->block_height256_bytes_c;
+				}
+				else {
+					v->macro_tile_size_bytes_c = 256.0 * 1024;
+					v->macro_tile_height_c = 32.0 * v->block_height256_bytes_c;
+				}
+				if (v->macro_tile_size_bytes_c <= 65536.0) {
+					v->pixel_pte_req_height_c = v->macro_tile_height_c;
+				}
+				else {
+					v->pixel_pte_req_height_c = 16.0 * v->block_height256_bytes_c;
+				}
+				v->pixel_pte_req_width_c = 4096.0 /dcn_bw_ceil2(v->byte_per_pixel_detc[k], 2.0) / v->pixel_pte_req_height_c * 8;
+				if (v->source_surface_mode[k] == dcn_bw_sw_linear) {
+					v->pixel_pte_bytes_per_row_c = 64.0 * (dcn_bw_ceil2((v->swath_width_y[k] / 2.0 * dcn_bw_min2(128.0, dcn_bw_pow(2.0,dcn_bw_floor2(dcn_bw_log(v->pte_buffer_size_in_requests * v->pixel_pte_req_width_c / (v->swath_width_y[k] / 2.0), 2.0), 1.0))) - 1.0) / v->pixel_pte_req_width_c, 1.0) + 1);
+				}
+				else if (v->source_scan[k] == dcn_bw_hor) {
+					v->pixel_pte_bytes_per_row_c = 64.0 * (dcn_bw_ceil2((v->swath_width_y[k] / 2.0 - 1.0) / v->pixel_pte_req_width_c, 1.0) + 1);
+				}
+				else {
+					v->pixel_pte_bytes_per_row_c = 64.0 * (dcn_bw_ceil2((v->viewport_height[k] / 2.0 - 1.0) / v->pixel_pte_req_height_c, 1.0) + 1);
+				}
+			}
+			else {
+				v->pixel_pte_bytes_per_row_c = 0.0;
+			}
+		}
+		else {
+			v->pixel_pte_bytes_per_row_c = 0.0;
+			v->meta_pte_bytes_frame_c = 0.0;
+			v->meta_row_byte_c = 0.0;
+		}
+		v->pixel_pte_bytes_per_row[k] = v->pixel_pte_bytes_per_row_y + v->pixel_pte_bytes_per_row_c;
+		v->meta_pte_bytes_frame[k] = v->meta_pte_bytes_frame_y + v->meta_pte_bytes_frame_c;
+		v->meta_row_byte[k] = v->meta_row_byte_y + v->meta_row_byte_c;
+		v->v_init_pre_fill_y[k] =dcn_bw_floor2((v->v_ratio[k] + v->vtaps[k] + 1.0 + v->interlace_output[k] * 0.5 * v->v_ratio[k]) / 2.0, 1.0);
+		v->max_num_swath_y[k] =dcn_bw_ceil2((v->v_init_pre_fill_y[k] - 1.0) / v->swath_height_y[k], 1.0) + 1;
+		if (v->v_init_pre_fill_y[k] > 1.0) {
+			v->max_partial_swath_y =dcn_bw_mod((v->v_init_pre_fill_y[k] - 2.0), v->swath_height_y[k]);
+		}
+		else {
+			v->max_partial_swath_y =dcn_bw_mod((v->v_init_pre_fill_y[k] + v->swath_height_y[k] - 2.0), v->swath_height_y[k]);
+		}
+		v->max_partial_swath_y =dcn_bw_max2(1.0, v->max_partial_swath_y);
+		v->prefetch_source_lines_y[k] = v->max_num_swath_y[k] * v->swath_height_y[k] + v->max_partial_swath_y;
+		if ((v->source_pixel_format[k] != dcn_bw_rgb_sub_64 && v->source_pixel_format[k] != dcn_bw_rgb_sub_32 && v->source_pixel_format[k] != dcn_bw_rgb_sub_16)) {
+			v->v_init_pre_fill_c[k] =dcn_bw_floor2((v->v_ratio[k] / 2.0 + v->vtaps[k] + 1.0 + v->interlace_output[k] * 0.5 * v->v_ratio[k] / 2.0) / 2.0, 1.0);
+			v->max_num_swath_c[k] =dcn_bw_ceil2((v->v_init_pre_fill_c[k] - 1.0) / v->swath_height_c[k], 1.0) + 1;
+			if (v->v_init_pre_fill_c[k] > 1.0) {
+				v->max_partial_swath_c =dcn_bw_mod((v->v_init_pre_fill_c[k] - 2.0), v->swath_height_c[k]);
+			}
+			else {
+				v->max_partial_swath_c =dcn_bw_mod((v->v_init_pre_fill_c[k] + v->swath_height_c[k] - 2.0), v->swath_height_c[k]);
+			}
+			v->max_partial_swath_c =dcn_bw_max2(1.0, v->max_partial_swath_c);
+		}
+		else {
+			v->max_num_swath_c[k] = 0.0;
+			v->max_partial_swath_c = 0.0;
+		}
+		v->prefetch_source_lines_c[k] = v->max_num_swath_c[k] * v->swath_height_c[k] + v->max_partial_swath_c;
+	}
+	v->t_calc = 24.0 / v->dcf_clk_deep_sleep;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->can_vstartup_lines_exceed_vsync_plus_back_porch_lines_minus_one == dcn_bw_yes) {
+			v->max_vstartup_lines[k] = v->vtotal[k] - v->vactive[k] - 1.0;
+		}
+		else {
+			v->max_vstartup_lines[k] = v->v_sync_plus_back_porch[k] - 1.0;
+		}
+	}
+	v->next_prefetch_mode = 0.0;
+	do {
+		v->v_startup_lines = 13.0;
+		do {
+			v->planes_with_room_to_increase_vstartup_prefetch_bw_less_than_active_bw = dcn_bw_yes;
+			v->planes_with_room_to_increase_vstartup_vratio_prefetch_more_than4 = dcn_bw_no;
+			v->planes_with_room_to_increase_vstartup_destination_line_times_for_prefetch_less_than2 = dcn_bw_no;
+			v->v_ratio_prefetch_more_than4 = dcn_bw_no;
+			v->destination_line_times_for_prefetch_less_than2 = dcn_bw_no;
+			v->prefetch_mode = v->next_prefetch_mode;
+			for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+				v->dstx_after_scaler = 90.0 * v->pixel_clock[k] / v->dppclk + 42.0 * v->pixel_clock[k] / v->dispclk;
+				if (v->dpp_per_plane[k] > 1.0) {
+					v->dstx_after_scaler = v->dstx_after_scaler + v->scaler_rec_out_width[k] / 2.0;
+				}
+				if (v->output_format[k] == dcn_bw_420) {
+					v->dsty_after_scaler = 1.0;
+				}
+				else {
+					v->dsty_after_scaler = 0.0;
+				}
+				v->v_update_offset_pix[k] = dcn_bw_ceil2(v->htotal[k] / 4.0, 1.0);
+				v->total_repeater_delay_time = v->max_inter_dcn_tile_repeaters * (2.0 / v->dppclk + 3.0 / v->dispclk);
+				v->v_update_width_pix[k] = (14.0 / v->dcf_clk_deep_sleep + 12.0 / v->dppclk + v->total_repeater_delay_time) * v->pixel_clock[k];
+				v->v_ready_offset_pix[k] = dcn_bw_max2(150.0 / v->dppclk, v->total_repeater_delay_time + 20.0 / v->dcf_clk_deep_sleep + 10.0 / v->dppclk) * v->pixel_clock[k];
+				v->t_setup = (v->v_update_offset_pix[k] + v->v_update_width_pix[k] + v->v_ready_offset_pix[k]) / v->pixel_clock[k];
+				v->v_startup[k] =dcn_bw_min2(v->v_startup_lines, v->max_vstartup_lines[k]);
+				if (v->prefetch_mode == 0.0) {
+					v->t_wait =dcn_bw_max3(v->dram_clock_change_latency + v->urgent_latency, v->sr_enter_plus_exit_time, v->urgent_latency);
+				}
+				else if (v->prefetch_mode == 1.0) {
+					v->t_wait =dcn_bw_max2(v->sr_enter_plus_exit_time, v->urgent_latency);
+				}
+				else {
+					v->t_wait = v->urgent_latency;
+				}
+				v->destination_lines_for_prefetch[k] =dcn_bw_floor2(4.0 * (v->v_startup[k] - v->t_wait / (v->htotal[k] / v->pixel_clock[k]) - (v->t_calc + v->t_setup) / (v->htotal[k] / v->pixel_clock[k]) - (v->dsty_after_scaler + v->dstx_after_scaler / v->htotal[k]) + 0.125), 1.0) / 4;
+				if (v->destination_lines_for_prefetch[k] > 0.0) {
+					v->prefetch_bandwidth[k] = (v->meta_pte_bytes_frame[k] + 2.0 * v->meta_row_byte[k] + 2.0 * v->pixel_pte_bytes_per_row[k] + v->prefetch_source_lines_y[k] * v->swath_width_y[k] *dcn_bw_ceil2(v->byte_per_pixel_dety[k], 1.0) + v->prefetch_source_lines_c[k] * v->swath_width_y[k] / 2.0 *dcn_bw_ceil2(v->byte_per_pixel_detc[k], 2.0)) / (v->destination_lines_for_prefetch[k] * v->htotal[k] / v->pixel_clock[k]);
+				}
+				else {
+					v->prefetch_bandwidth[k] = 999999.0;
+				}
+			}
+			v->bandwidth_available_for_immediate_flip = v->return_bw;
+			for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+				v->bandwidth_available_for_immediate_flip = v->bandwidth_available_for_immediate_flip -dcn_bw_max2(v->read_bandwidth_plane_luma[k] + v->read_bandwidth_plane_chroma[k], v->prefetch_bandwidth[k]);
+			}
+			v->tot_immediate_flip_bytes = 0.0;
+			for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+				if (v->immediate_flip_supported == dcn_bw_yes && (v->source_pixel_format[k] != dcn_bw_yuv420_sub_8 && v->source_pixel_format[k] != dcn_bw_yuv420_sub_10)) {
+					v->tot_immediate_flip_bytes = v->tot_immediate_flip_bytes + v->meta_pte_bytes_frame[k] + v->meta_row_byte[k] + v->pixel_pte_bytes_per_row[k];
+				}
+			}
+			v->max_rd_bandwidth = 0.0;
+			for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+				if (v->pte_enable == dcn_bw_yes && v->dcc_enable[k] == dcn_bw_yes) {
+					if (v->immediate_flip_supported == dcn_bw_yes && (v->source_pixel_format[k] != dcn_bw_yuv420_sub_8 && v->source_pixel_format[k] != dcn_bw_yuv420_sub_10)) {
+						v->time_for_fetching_meta_pte =dcn_bw_max5(v->meta_pte_bytes_frame[k] / v->prefetch_bandwidth[k], v->meta_pte_bytes_frame[k] * v->tot_immediate_flip_bytes / (v->bandwidth_available_for_immediate_flip * (v->meta_pte_bytes_frame[k] + v->meta_row_byte[k] + v->pixel_pte_bytes_per_row[k])), v->urgent_extra_latency, v->urgent_latency, v->htotal[k] / v->pixel_clock[k] / 4.0);
+					}
+					else {
+						v->time_for_fetching_meta_pte =dcn_bw_max3(v->meta_pte_bytes_frame[k] / v->prefetch_bandwidth[k], v->urgent_extra_latency, v->htotal[k] / v->pixel_clock[k] / 4.0);
+					}
+				}
+				else {
+					v->time_for_fetching_meta_pte = v->htotal[k] / v->pixel_clock[k] / 4.0;
+				}
+				v->destination_lines_to_request_vm_inv_blank[k] =dcn_bw_floor2(4.0 * (v->time_for_fetching_meta_pte / (v->htotal[k] / v->pixel_clock[k]) + 0.125), 1.0) / 4;
+				if ((v->pte_enable == dcn_bw_yes || v->dcc_enable[k] == dcn_bw_yes)) {
+					if (v->immediate_flip_supported == dcn_bw_yes && (v->source_pixel_format[k] != dcn_bw_yuv420_sub_8 && v->source_pixel_format[k] != dcn_bw_yuv420_sub_10)) {
+						v->time_for_fetching_row_in_vblank =dcn_bw_max5((v->meta_row_byte[k] + v->pixel_pte_bytes_per_row[k]) / v->prefetch_bandwidth[k], (v->meta_row_byte[k] + v->pixel_pte_bytes_per_row[k]) * v->tot_immediate_flip_bytes / (v->bandwidth_available_for_immediate_flip * (v->meta_pte_bytes_frame[k] + v->meta_row_byte[k] + v->pixel_pte_bytes_per_row[k])), v->urgent_extra_latency, 2.0 * v->urgent_latency, v->htotal[k] / v->pixel_clock[k] - v->time_for_fetching_meta_pte);
+					}
+					else {
+						v->time_for_fetching_row_in_vblank =dcn_bw_max3((v->meta_row_byte[k] + v->pixel_pte_bytes_per_row[k]) / v->prefetch_bandwidth[k], v->urgent_extra_latency, v->htotal[k] / v->pixel_clock[k] - v->time_for_fetching_meta_pte);
+					}
+				}
+				else {
+					v->time_for_fetching_row_in_vblank =dcn_bw_max2(v->urgent_extra_latency - v->time_for_fetching_meta_pte, v->htotal[k] / v->pixel_clock[k] - v->time_for_fetching_meta_pte);
+				}
+				v->destination_lines_to_request_row_in_vblank[k] =dcn_bw_floor2(4.0 * (v->time_for_fetching_row_in_vblank / (v->htotal[k] / v->pixel_clock[k]) + 0.125), 1.0) / 4;
+				v->lines_to_request_prefetch_pixel_data = v->destination_lines_for_prefetch[k] - v->destination_lines_to_request_vm_inv_blank[k] - v->destination_lines_to_request_row_in_vblank[k];
+				if (v->lines_to_request_prefetch_pixel_data > 0.0) {
+					v->v_ratio_prefetch_y[k] = v->prefetch_source_lines_y[k] / v->lines_to_request_prefetch_pixel_data;
+					if ((v->swath_height_y[k] > 4.0)) {
+						if (v->lines_to_request_prefetch_pixel_data > (v->v_init_pre_fill_y[k] - 3.0) / 2.0) {
+							v->v_ratio_prefetch_y[k] =dcn_bw_max2(v->v_ratio_prefetch_y[k], v->max_num_swath_y[k] * v->swath_height_y[k] / (v->lines_to_request_prefetch_pixel_data - (v->v_init_pre_fill_y[k] - 3.0) / 2.0));
+						}
+						else {
+							v->v_ratio_prefetch_y[k] = 999999.0;
+						}
+					}
+				}
+				else {
+					v->v_ratio_prefetch_y[k] = 999999.0;
+				}
+				v->v_ratio_prefetch_y[k] =dcn_bw_max2(v->v_ratio_prefetch_y[k], 1.0);
+				if (v->lines_to_request_prefetch_pixel_data > 0.0) {
+					v->v_ratio_prefetch_c[k] = v->prefetch_source_lines_c[k] / v->lines_to_request_prefetch_pixel_data;
+					if ((v->swath_height_c[k] > 4.0)) {
+						if (v->lines_to_request_prefetch_pixel_data > (v->v_init_pre_fill_c[k] - 3.0) / 2.0) {
+							v->v_ratio_prefetch_c[k] =dcn_bw_max2(v->v_ratio_prefetch_c[k], v->max_num_swath_c[k] * v->swath_height_c[k] / (v->lines_to_request_prefetch_pixel_data - (v->v_init_pre_fill_c[k] - 3.0) / 2.0));
+						}
+						else {
+							v->v_ratio_prefetch_c[k] = 999999.0;
+						}
+					}
+				}
+				else {
+					v->v_ratio_prefetch_c[k] = 999999.0;
+				}
+				v->v_ratio_prefetch_c[k] =dcn_bw_max2(v->v_ratio_prefetch_c[k], 1.0);
+				if (v->lines_to_request_prefetch_pixel_data > 0.0) {
+					v->required_prefetch_pix_data_bw = v->dpp_per_plane[k] * (v->prefetch_source_lines_y[k] / v->lines_to_request_prefetch_pixel_data *dcn_bw_ceil2(v->byte_per_pixel_dety[k], 1.0) + v->prefetch_source_lines_c[k] / v->lines_to_request_prefetch_pixel_data *dcn_bw_ceil2(v->byte_per_pixel_detc[k], 2.0) / 2.0) * v->swath_width_y[k] / (v->htotal[k] / v->pixel_clock[k]);
+				}
+				else {
+					v->required_prefetch_pix_data_bw = 999999.0;
+				}
+				v->max_rd_bandwidth = v->max_rd_bandwidth +dcn_bw_max2(v->read_bandwidth_plane_luma[k] + v->read_bandwidth_plane_chroma[k], v->required_prefetch_pix_data_bw);
+				if (v->immediate_flip_supported == dcn_bw_yes && (v->source_pixel_format[k] != dcn_bw_yuv420_sub_8 && v->source_pixel_format[k] != dcn_bw_yuv420_sub_10)) {
+					v->max_rd_bandwidth = v->max_rd_bandwidth +dcn_bw_max2(v->meta_pte_bytes_frame[k] / (v->destination_lines_to_request_vm_inv_blank[k] * v->htotal[k] / v->pixel_clock[k]), (v->meta_row_byte[k] + v->pixel_pte_bytes_per_row[k]) / (v->destination_lines_to_request_row_in_vblank[k] * v->htotal[k] / v->pixel_clock[k]));
+				}
+				if (v->v_ratio_prefetch_y[k] > 4.0 || v->v_ratio_prefetch_c[k] > 4.0) {
+					v->v_ratio_prefetch_more_than4 = dcn_bw_yes;
+				}
+				if (v->destination_lines_for_prefetch[k] < 2.0) {
+					v->destination_line_times_for_prefetch_less_than2 = dcn_bw_yes;
+				}
+				if (v->max_vstartup_lines[k] > v->v_startup_lines) {
+					if (v->required_prefetch_pix_data_bw > (v->read_bandwidth_plane_luma[k] + v->read_bandwidth_plane_chroma[k])) {
+						v->planes_with_room_to_increase_vstartup_prefetch_bw_less_than_active_bw = dcn_bw_no;
+					}
+					if (v->v_ratio_prefetch_y[k] > 4.0 || v->v_ratio_prefetch_c[k] > 4.0) {
+						v->planes_with_room_to_increase_vstartup_vratio_prefetch_more_than4 = dcn_bw_yes;
+					}
+					if (v->destination_lines_for_prefetch[k] < 2.0) {
+						v->planes_with_room_to_increase_vstartup_destination_line_times_for_prefetch_less_than2 = dcn_bw_yes;
+					}
+				}
+			}
+			if (v->max_rd_bandwidth <= v->return_bw && v->v_ratio_prefetch_more_than4 == dcn_bw_no && v->destination_line_times_for_prefetch_less_than2 == dcn_bw_no) {
+				v->prefetch_mode_supported = dcn_bw_yes;
+			}
+			else {
+				v->prefetch_mode_supported = dcn_bw_no;
+			}
+			v->v_startup_lines = v->v_startup_lines + 1.0;
+		} while (!(v->prefetch_mode_supported == dcn_bw_yes || (v->planes_with_room_to_increase_vstartup_prefetch_bw_less_than_active_bw == dcn_bw_yes && v->planes_with_room_to_increase_vstartup_vratio_prefetch_more_than4 == dcn_bw_no && v->planes_with_room_to_increase_vstartup_destination_line_times_for_prefetch_less_than2 == dcn_bw_no)));
+		v->next_prefetch_mode = v->next_prefetch_mode + 1.0;
+	} while (!(v->prefetch_mode_supported == dcn_bw_yes || v->prefetch_mode == 2.0));
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->v_ratio_prefetch_y[k] <= 1.0) {
+			v->display_pipe_line_delivery_time_luma_prefetch[k] = v->swath_width_y[k] * v->dpp_per_plane[k] / v->h_ratio[k] / v->pixel_clock[k];
+		}
+		else {
+			v->display_pipe_line_delivery_time_luma_prefetch[k] = v->swath_width_y[k] / v->pscl_throughput[k] / v->dppclk;
+		}
+		if (v->byte_per_pixel_detc[k] == 0.0) {
+			v->display_pipe_line_delivery_time_chroma_prefetch[k] = 0.0;
+		}
+		else {
+			if (v->v_ratio_prefetch_c[k] <= 1.0) {
+				v->display_pipe_line_delivery_time_chroma_prefetch[k] = v->swath_width_y[k] * v->dpp_per_plane[k] / v->h_ratio[k] / v->pixel_clock[k];
+			}
+			else {
+				v->display_pipe_line_delivery_time_chroma_prefetch[k] = v->swath_width_y[k] / v->pscl_throughput[k] / v->dppclk;
+			}
+		}
+	}
+	/*min ttuv_blank*/
+
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->prefetch_mode == 0.0) {
+			v->allow_dram_clock_change_during_vblank[k] = dcn_bw_yes;
+			v->allow_dram_self_refresh_during_vblank[k] = dcn_bw_yes;
+			v->min_ttuv_blank[k] = v->t_calc +dcn_bw_max3(v->dram_clock_change_watermark, v->stutter_enter_plus_exit_watermark, v->urgent_watermark);
+		}
+		else if (v->prefetch_mode == 1.0) {
+			v->allow_dram_clock_change_during_vblank[k] = dcn_bw_no;
+			v->allow_dram_self_refresh_during_vblank[k] = dcn_bw_yes;
+			v->min_ttuv_blank[k] = v->t_calc +dcn_bw_max2(v->stutter_enter_plus_exit_watermark, v->urgent_watermark);
+		}
+		else {
+			v->allow_dram_clock_change_during_vblank[k] = dcn_bw_no;
+			v->allow_dram_self_refresh_during_vblank[k] = dcn_bw_no;
+			v->min_ttuv_blank[k] = v->t_calc + v->urgent_watermark;
+		}
+	}
+	/*nb p-state/dram clock change support*/
+
+	v->active_dp_ps = 0.0;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		v->active_dp_ps = v->active_dp_ps + v->dpp_per_plane[k];
+	}
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		v->lb_latency_hiding_source_lines_y =dcn_bw_min2(v->max_line_buffer_lines,dcn_bw_floor2(v->line_buffer_size / v->lb_bit_per_pixel[k] / (v->swath_width_y[k] /dcn_bw_max2(v->h_ratio[k], 1.0)), 1.0)) - (v->vtaps[k] - 1.0);
+		v->lb_latency_hiding_source_lines_c =dcn_bw_min2(v->max_line_buffer_lines,dcn_bw_floor2(v->line_buffer_size / v->lb_bit_per_pixel[k] / (v->swath_width_y[k] / 2.0 /dcn_bw_max2(v->h_ratio[k] / 2.0, 1.0)), 1.0)) - (v->vta_pschroma[k] - 1.0);
+		v->effective_lb_latency_hiding_y = v->lb_latency_hiding_source_lines_y / v->v_ratio[k] * (v->htotal[k] / v->pixel_clock[k]);
+		v->effective_lb_latency_hiding_c = v->lb_latency_hiding_source_lines_c / (v->v_ratio[k] / 2.0) * (v->htotal[k] / v->pixel_clock[k]);
+		if (v->swath_width_y[k] > 2.0 * v->dpp_output_buffer_pixels) {
+			v->dpp_output_buffer_lines_y = v->dpp_output_buffer_pixels / v->swath_width_y[k];
+		}
+		else if (v->swath_width_y[k] > v->dpp_output_buffer_pixels) {
+			v->dpp_output_buffer_lines_y = 0.5;
+		}
+		else {
+			v->dpp_output_buffer_lines_y = 1.0;
+		}
+		if (v->swath_width_y[k] / 2.0 > 2.0 * v->dpp_output_buffer_pixels) {
+			v->dpp_output_buffer_lines_c = v->dpp_output_buffer_pixels / (v->swath_width_y[k] / 2.0);
+		}
+		else if (v->swath_width_y[k] / 2.0 > v->dpp_output_buffer_pixels) {
+			v->dpp_output_buffer_lines_c = 0.5;
+		}
+		else {
+			v->dpp_output_buffer_lines_c = 1.0;
+		}
+		v->dppopp_buffering_y = (v->htotal[k] / v->pixel_clock[k]) * (v->dpp_output_buffer_lines_y + v->opp_output_buffer_lines);
+		v->max_det_buffering_time_y = v->full_det_buffering_time_y[k] + (v->lines_in_dety[k] - v->lines_in_dety_rounded_down_to_swath[k]) / v->swath_height_y[k] * (v->htotal[k] / v->pixel_clock[k]);
+		v->active_dram_clock_change_latency_margin_y = v->dppopp_buffering_y + v->effective_lb_latency_hiding_y + v->max_det_buffering_time_y - v->dram_clock_change_watermark;
+		if (v->active_dp_ps > 1.0) {
+			v->active_dram_clock_change_latency_margin_y = v->active_dram_clock_change_latency_margin_y - (1.0 - 1.0 / (v->active_dp_ps - 1.0)) * v->swath_height_y[k] * (v->htotal[k] / v->pixel_clock[k]);
+		}
+		if (v->byte_per_pixel_detc[k] > 0.0) {
+			v->dppopp_buffering_c = (v->htotal[k] / v->pixel_clock[k]) * (v->dpp_output_buffer_lines_c + v->opp_output_buffer_lines);
+			v->max_det_buffering_time_c = v->full_det_buffering_time_c[k] + (v->lines_in_detc[k] - v->lines_in_detc_rounded_down_to_swath[k]) / v->swath_height_c[k] * (v->htotal[k] / v->pixel_clock[k]);
+			v->active_dram_clock_change_latency_margin_c = v->dppopp_buffering_c + v->effective_lb_latency_hiding_c + v->max_det_buffering_time_c - v->dram_clock_change_watermark;
+			if (v->active_dp_ps > 1.0) {
+				v->active_dram_clock_change_latency_margin_c = v->active_dram_clock_change_latency_margin_c - (1.0 - 1.0 / (v->active_dp_ps - 1.0)) * v->swath_height_c[k] * (v->htotal[k] / v->pixel_clock[k]);
+			}
+			v->active_dram_clock_change_latency_margin[k] =dcn_bw_min2(v->active_dram_clock_change_latency_margin_y, v->active_dram_clock_change_latency_margin_c);
+		}
+		else {
+			v->active_dram_clock_change_latency_margin[k] = v->active_dram_clock_change_latency_margin_y;
+		}
+		if (v->output_format[k] == dcn_bw_444) {
+			v->writeback_dram_clock_change_latency_margin = (v->writeback_luma_buffer_size + v->writeback_chroma_buffer_size) * 1024.0 / (v->scaler_rec_out_width[k] / (v->htotal[k] / v->pixel_clock[k]) * 4.0) - v->writeback_dram_clock_change_watermark;
+		}
+		else {
+			v->writeback_dram_clock_change_latency_margin =dcn_bw_min2(v->writeback_luma_buffer_size, 2.0 * v->writeback_chroma_buffer_size) * 1024.0 / (v->scaler_rec_out_width[k] / (v->htotal[k] / v->pixel_clock[k])) - v->writeback_dram_clock_change_watermark;
+		}
+		if (v->output[k] == dcn_bw_writeback) {
+			v->active_dram_clock_change_latency_margin[k] =dcn_bw_min2(v->active_dram_clock_change_latency_margin[k], v->writeback_dram_clock_change_latency_margin);
+		}
+	}
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->allow_dram_clock_change_during_vblank[k] == dcn_bw_yes) {
+			v->v_blank_dram_clock_change_latency_margin[k] = (v->vtotal[k] - v->scaler_recout_height[k]) * (v->htotal[k] / v->pixel_clock[k]) -dcn_bw_max2(v->dram_clock_change_watermark, v->writeback_dram_clock_change_watermark);
+		}
+		else {
+			v->v_blank_dram_clock_change_latency_margin[k] = 0.0;
+		}
+	}
+	v->min_active_dram_clock_change_margin = 999999.0;
+	v->v_blank_of_min_active_dram_clock_change_margin = 999999.0;
+	v->second_min_active_dram_clock_change_margin = 999999.0;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->active_dram_clock_change_latency_margin[k] < v->min_active_dram_clock_change_margin) {
+			v->second_min_active_dram_clock_change_margin = v->min_active_dram_clock_change_margin;
+			v->min_active_dram_clock_change_margin = v->active_dram_clock_change_latency_margin[k];
+			v->v_blank_of_min_active_dram_clock_change_margin = v->v_blank_dram_clock_change_latency_margin[k];
+		}
+		else if (v->active_dram_clock_change_latency_margin[k] < v->second_min_active_dram_clock_change_margin) {
+			v->second_min_active_dram_clock_change_margin = v->active_dram_clock_change_latency_margin[k];
+		}
+	}
+	v->min_vblank_dram_clock_change_margin = 999999.0;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->min_vblank_dram_clock_change_margin > v->v_blank_dram_clock_change_latency_margin[k]) {
+			v->min_vblank_dram_clock_change_margin = v->v_blank_dram_clock_change_latency_margin[k];
+		}
+	}
+	if (v->synchronized_vblank == dcn_bw_yes || v->number_of_active_planes == 1) {
+		v->dram_clock_change_margin =dcn_bw_max2(v->min_active_dram_clock_change_margin, v->min_vblank_dram_clock_change_margin);
+	}
+	else if (v->v_blank_of_min_active_dram_clock_change_margin > v->min_active_dram_clock_change_margin) {
+		v->dram_clock_change_margin =dcn_bw_min2(v->second_min_active_dram_clock_change_margin, v->v_blank_of_min_active_dram_clock_change_margin);
+	}
+	else {
+		v->dram_clock_change_margin = v->min_active_dram_clock_change_margin;
+	}
+	if (v->min_active_dram_clock_change_margin > 0.0) {
+		v->dram_clock_change_support = dcn_bw_supported_in_v_active;
+	}
+	else if (v->dram_clock_change_margin > 0.0) {
+		v->dram_clock_change_support = dcn_bw_supported_in_v_blank;
+	}
+	else {
+		v->dram_clock_change_support = dcn_bw_not_supported;
+	}
+	/*maximum bandwidth used*/
+
+	v->wr_bandwidth = 0.0;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->output[k] == dcn_bw_writeback && v->output_format[k] == dcn_bw_444) {
+			v->wr_bandwidth = v->wr_bandwidth + v->scaler_rec_out_width[k] / (v->htotal[k] / v->pixel_clock[k]) * 4.0;
+		}
+		else if (v->output[k] == dcn_bw_writeback) {
+			v->wr_bandwidth = v->wr_bandwidth + v->scaler_rec_out_width[k] / (v->htotal[k] / v->pixel_clock[k]) * 1.5;
+		}
+	}
+	v->max_used_bw = v->max_rd_bandwidth + v->wr_bandwidth;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dml/calcs/dcn_calc_auto.h b/drivers/gpu/drm/amd/display/dc/dml/calcs/dcn_calc_auto.h
new file mode 100644
index 000000000000..ce35de79a6c7
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dml/calcs/dcn_calc_auto.h
@@ -0,0 +1,38 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef _DCN_CALC_AUTO_H_
+#define _DCN_CALC_AUTO_H_
+
+#include "dc.h"
+#include "dcn_calcs.h"
+
+void scaler_settings_calculation(struct dcn_bw_internal_vars *v);
+void mode_support_and_system_configuration(struct dcn_bw_internal_vars *v);
+void display_pipe_configuration(struct dcn_bw_internal_vars *v);
+void dispclkdppclkdcfclk_deep_sleep_prefetch_parameters_watermarks_and_performance_calculation(
+		struct dcn_bw_internal_vars *v);
+
+#endif /* _DCN_CALC_AUTO_H_ */
diff --git a/drivers/gpu/drm/amd/display/dc/dml/calcs/dcn_calc_math.c b/drivers/gpu/drm/amd/display/dc/dml/calcs/dcn_calc_math.c
new file mode 100644
index 000000000000..cac72413a097
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dml/calcs/dcn_calc_math.c
@@ -0,0 +1,145 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "os_types.h"
+#include "dcn_calc_math.h"
+
+#define isNaN(number) ((number) != (number))
+
+/*
+ * NOTE:
+ *   This file is gcc-parseable HW gospel, coming straight from HW engineers.
+ *
+ * It doesn't adhere to Linux kernel style and sometimes will do things in odd
+ * ways. Unless there is something clearly wrong with it the code should
+ * remain as-is as it provides us with a guarantee from HW that it is correct.
+ */
+
+float dcn_bw_mod(const float arg1, const float arg2)
+{
+	if (isNaN(arg1))
+		return arg2;
+	if (isNaN(arg2))
+		return arg1;
+	return arg1 - arg1 * ((int) (arg1 / arg2));
+}
+
+float dcn_bw_min2(const float arg1, const float arg2)
+{
+	if (isNaN(arg1))
+		return arg2;
+	if (isNaN(arg2))
+		return arg1;
+	return arg1 < arg2 ? arg1 : arg2;
+}
+
+unsigned int dcn_bw_max(const unsigned int arg1, const unsigned int arg2)
+{
+	return arg1 > arg2 ? arg1 : arg2;
+}
+float dcn_bw_max2(const float arg1, const float arg2)
+{
+	if (isNaN(arg1))
+		return arg2;
+	if (isNaN(arg2))
+		return arg1;
+	return arg1 > arg2 ? arg1 : arg2;
+}
+
+float dcn_bw_floor2(const float arg, const float significance)
+{
+	ASSERT(significance != 0);
+
+	return ((int) (arg / significance)) * significance;
+}
+float dcn_bw_floor(const float arg)
+{
+	return ((int) (arg));
+}
+
+float dcn_bw_ceil(const float arg)
+{
+	return (int) (arg + 0.99999);
+}
+
+float dcn_bw_ceil2(const float arg, const float significance)
+{
+	ASSERT(significance != 0);
+
+	return ((int) (arg / significance + 0.99999)) * significance;
+}
+
+float dcn_bw_max3(float v1, float v2, float v3)
+{
+	return v3 > dcn_bw_max2(v1, v2) ? v3 : dcn_bw_max2(v1, v2);
+}
+
+float dcn_bw_max5(float v1, float v2, float v3, float v4, float v5)
+{
+	return dcn_bw_max3(v1, v2, v3) > dcn_bw_max2(v4, v5) ? dcn_bw_max3(v1, v2, v3) : dcn_bw_max2(v4, v5);
+}
+
+float dcn_bw_pow(float a, float exp)
+{
+	float temp;
+	/*ASSERT(exp == (int)exp);*/
+	if ((int)exp == 0)
+		return 1;
+	temp = dcn_bw_pow(a, (int)(exp / 2));
+	if (((int)exp % 2) == 0) {
+		return temp * temp;
+	} else {
+		if ((int)exp > 0)
+			return a * temp * temp;
+		else
+			return (temp * temp) / a;
+	}
+}
+
+double dcn_bw_fabs(double a)
+{
+	if (a > 0)
+		return (a);
+	else
+		return (-a);
+}
+
+
+float dcn_bw_log(float a, float b)
+{
+	int * const exp_ptr = (int *)(&a);
+	int x = *exp_ptr;
+	const int log_2 = ((x >> 23) & 255) - 128;
+	x &= ~(255 << 23);
+	x += 127 << 23;
+	*exp_ptr = x;
+
+	a = ((-1.0f / 3) * a + 2) * a - 2.0f / 3;
+
+	if (b > 2.00001 || b < 1.99999)
+		return (a + log_2) / dcn_bw_log(b, 2);
+	else
+		return (a + log_2);
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dml/calcs/dcn_calcs.c b/drivers/gpu/drm/amd/display/dc/dml/calcs/dcn_calcs.c
new file mode 100644
index 000000000000..74962791302f
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dml/calcs/dcn_calcs.c
@@ -0,0 +1,1619 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ * Copyright 2019 Raptor Engineering, LLC
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dm_services.h"
+#include "dc.h"
+#include "dcn_calcs.h"
+#include "dcn_calc_auto.h"
+#include "dal_asic_id.h"
+#include "resource.h"
+#include "resource/dcn10/dcn10_resource.h"
+#include "dcn10/dcn10_hubbub.h"
+#include "dml/dml1_display_rq_dlg_calc.h"
+
+#include "dcn_calc_math.h"
+
+#define DC_LOGGER \
+	dc->ctx->logger
+
+#define WM_SET_COUNT 4
+#define WM_A 0
+#define WM_B 1
+#define WM_C 2
+#define WM_D 3
+
+/*
+ * NOTE:
+ *   This file is gcc-parseable HW gospel, coming straight from HW engineers.
+ *
+ * It doesn't adhere to Linux kernel style and sometimes will do things in odd
+ * ways. Unless there is something clearly wrong with it the code should
+ * remain as-is as it provides us with a guarantee from HW that it is correct.
+ */
+
+/* Defaults from spreadsheet rev#247.
+ * RV2 delta: dram_clock_change_latency, max_num_dpp
+ */
+const struct dcn_soc_bounding_box dcn10_soc_defaults = {
+		/* latencies */
+		.sr_exit_time = 17, /*us*/
+		.sr_enter_plus_exit_time = 19, /*us*/
+		.urgent_latency = 4, /*us*/
+		.dram_clock_change_latency = 17, /*us*/
+		.write_back_latency = 12, /*us*/
+		.percent_of_ideal_drambw_received_after_urg_latency = 80, /*%*/
+
+		/* below default clocks derived from STA target base on
+		 * slow-slow corner + 10% margin with voltages aligned to FCLK.
+		 *
+		 * Use these value if fused value doesn't make sense as earlier
+		 * part don't have correct value fused */
+		/* default DCF CLK DPM on RV*/
+		.dcfclkv_max0p9 = 655,	/* MHz, = 3600/5.5 */
+		.dcfclkv_nom0p8 = 626,	/* MHz, = 3600/5.75 */
+		.dcfclkv_mid0p72 = 600,	/* MHz, = 3600/6, bypass */
+		.dcfclkv_min0p65 = 300,	/* MHz, = 3600/12, bypass */
+
+		/* default DISP CLK voltage state on RV */
+		.max_dispclk_vmax0p9 = 1108,	/* MHz, = 3600/3.25 */
+		.max_dispclk_vnom0p8 = 1029,	/* MHz, = 3600/3.5 */
+		.max_dispclk_vmid0p72 = 960,	/* MHz, = 3600/3.75 */
+		.max_dispclk_vmin0p65 = 626,	/* MHz, = 3600/5.75 */
+
+		/* default DPP CLK voltage state on RV */
+		.max_dppclk_vmax0p9 = 720,	/* MHz, = 3600/5 */
+		.max_dppclk_vnom0p8 = 686,	/* MHz, = 3600/5.25 */
+		.max_dppclk_vmid0p72 = 626,	/* MHz, = 3600/5.75 */
+		.max_dppclk_vmin0p65 = 400,	/* MHz, = 3600/9 */
+
+		/* default PHY CLK voltage state on RV */
+		.phyclkv_max0p9 = 900, /*MHz*/
+		.phyclkv_nom0p8 = 847, /*MHz*/
+		.phyclkv_mid0p72 = 800, /*MHz*/
+		.phyclkv_min0p65 = 600, /*MHz*/
+
+		/* BW depend on FCLK, MCLK, # of channels */
+		/* dual channel BW */
+		.fabric_and_dram_bandwidth_vmax0p9 = 38.4f, /*GB/s*/
+		.fabric_and_dram_bandwidth_vnom0p8 = 34.133f, /*GB/s*/
+		.fabric_and_dram_bandwidth_vmid0p72 = 29.866f, /*GB/s*/
+		.fabric_and_dram_bandwidth_vmin0p65 = 12.8f, /*GB/s*/
+		/* single channel BW
+		.fabric_and_dram_bandwidth_vmax0p9 = 19.2f,
+		.fabric_and_dram_bandwidth_vnom0p8 = 17.066f,
+		.fabric_and_dram_bandwidth_vmid0p72 = 14.933f,
+		.fabric_and_dram_bandwidth_vmin0p65 = 12.8f,
+		*/
+
+		.number_of_channels = 2,
+
+		.socclk = 208, /*MHz*/
+		.downspreading = 0.5f, /*%*/
+		.round_trip_ping_latency_cycles = 128, /*DCFCLK Cycles*/
+		.urgent_out_of_order_return_per_channel = 256, /*bytes*/
+		.vmm_page_size = 4096, /*bytes*/
+		.return_bus_width = 64, /*bytes*/
+		.max_request_size = 256, /*bytes*/
+
+		/* Depends on user class (client vs embedded, workstation, etc) */
+		.percent_disp_bw_limit = 0.3f /*%*/
+};
+
+const struct dcn_ip_params dcn10_ip_defaults = {
+		.rob_buffer_size_in_kbyte = 64,
+		.det_buffer_size_in_kbyte = 164,
+		.dpp_output_buffer_pixels = 2560,
+		.opp_output_buffer_lines = 1,
+		.pixel_chunk_size_in_kbyte = 8,
+		.pte_enable = dcn_bw_yes,
+		.pte_chunk_size = 2, /*kbytes*/
+		.meta_chunk_size = 2, /*kbytes*/
+		.writeback_chunk_size = 2, /*kbytes*/
+		.odm_capability = dcn_bw_no,
+		.dsc_capability = dcn_bw_no,
+		.line_buffer_size = 589824, /*bit*/
+		.max_line_buffer_lines = 12,
+		.is_line_buffer_bpp_fixed = dcn_bw_no,
+		.line_buffer_fixed_bpp = dcn_bw_na,
+		.writeback_luma_buffer_size = 12, /*kbytes*/
+		.writeback_chroma_buffer_size = 8, /*kbytes*/
+		.max_num_dpp = 4,
+		.max_num_writeback = 2,
+		.max_dchub_topscl_throughput = 4, /*pixels/dppclk*/
+		.max_pscl_tolb_throughput = 2, /*pixels/dppclk*/
+		.max_lb_tovscl_throughput = 4, /*pixels/dppclk*/
+		.max_vscl_tohscl_throughput = 4, /*pixels/dppclk*/
+		.max_hscl_ratio = 4,
+		.max_vscl_ratio = 4,
+		.max_hscl_taps = 8,
+		.max_vscl_taps = 8,
+		.pte_buffer_size_in_requests = 42,
+		.dispclk_ramping_margin = 1, /*%*/
+		.under_scan_factor = 1.11f,
+		.max_inter_dcn_tile_repeaters = 8,
+		.can_vstartup_lines_exceed_vsync_plus_back_porch_lines_minus_one = dcn_bw_no,
+		.bug_forcing_luma_and_chroma_request_to_same_size_fixed = dcn_bw_no,
+		.dcfclk_cstate_latency = 10 /*TODO clone of something else? sr_enter_plus_exit_time?*/
+};
+
+static enum dcn_bw_defs tl_sw_mode_to_bw_defs(enum swizzle_mode_values sw_mode)
+{
+	switch (sw_mode) {
+	case DC_SW_LINEAR:
+		return dcn_bw_sw_linear;
+	case DC_SW_4KB_S:
+		return dcn_bw_sw_4_kb_s;
+	case DC_SW_4KB_D:
+		return dcn_bw_sw_4_kb_d;
+	case DC_SW_64KB_S:
+		return dcn_bw_sw_64_kb_s;
+	case DC_SW_64KB_D:
+		return dcn_bw_sw_64_kb_d;
+	case DC_SW_VAR_S:
+		return dcn_bw_sw_var_s;
+	case DC_SW_VAR_D:
+		return dcn_bw_sw_var_d;
+	case DC_SW_64KB_S_T:
+		return dcn_bw_sw_64_kb_s_t;
+	case DC_SW_64KB_D_T:
+		return dcn_bw_sw_64_kb_d_t;
+	case DC_SW_4KB_S_X:
+		return dcn_bw_sw_4_kb_s_x;
+	case DC_SW_4KB_D_X:
+		return dcn_bw_sw_4_kb_d_x;
+	case DC_SW_64KB_S_X:
+		return dcn_bw_sw_64_kb_s_x;
+	case DC_SW_64KB_D_X:
+		return dcn_bw_sw_64_kb_d_x;
+	case DC_SW_VAR_S_X:
+		return dcn_bw_sw_var_s_x;
+	case DC_SW_VAR_D_X:
+		return dcn_bw_sw_var_d_x;
+	case DC_SW_256B_S:
+	case DC_SW_256_D:
+	case DC_SW_256_R:
+	case DC_SW_4KB_R:
+	case DC_SW_64KB_R:
+	case DC_SW_VAR_R:
+	case DC_SW_4KB_R_X:
+	case DC_SW_64KB_R_X:
+	case DC_SW_VAR_R_X:
+	default:
+		BREAK_TO_DEBUGGER(); /*not in formula*/
+		return dcn_bw_sw_4_kb_s;
+	}
+}
+
+static int tl_lb_bpp_to_int(enum lb_pixel_depth depth)
+{
+	switch (depth) {
+	case LB_PIXEL_DEPTH_18BPP:
+		return 18;
+	case LB_PIXEL_DEPTH_24BPP:
+		return 24;
+	case LB_PIXEL_DEPTH_30BPP:
+		return 30;
+	case LB_PIXEL_DEPTH_36BPP:
+		return 36;
+	default:
+		return 30;
+	}
+}
+
+static enum dcn_bw_defs tl_pixel_format_to_bw_defs(enum surface_pixel_format format)
+{
+	switch (format) {
+	case SURFACE_PIXEL_FORMAT_GRPH_ARGB1555:
+	case SURFACE_PIXEL_FORMAT_GRPH_RGB565:
+		return dcn_bw_rgb_sub_16;
+	case SURFACE_PIXEL_FORMAT_GRPH_ARGB8888:
+	case SURFACE_PIXEL_FORMAT_GRPH_ABGR8888:
+	case SURFACE_PIXEL_FORMAT_GRPH_ARGB2101010:
+	case SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010:
+	case SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010_XR_BIAS:
+		return dcn_bw_rgb_sub_32;
+	case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616:
+	case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616:
+	case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616F:
+	case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F:
+		return dcn_bw_rgb_sub_64;
+	case SURFACE_PIXEL_FORMAT_VIDEO_420_YCbCr:
+	case SURFACE_PIXEL_FORMAT_VIDEO_420_YCrCb:
+		return dcn_bw_yuv420_sub_8;
+	case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCbCr:
+	case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCrCb:
+		return dcn_bw_yuv420_sub_10;
+	default:
+		return dcn_bw_rgb_sub_32;
+	}
+}
+
+enum source_macro_tile_size swizzle_mode_to_macro_tile_size(enum swizzle_mode_values sw_mode)
+{
+	switch (sw_mode) {
+	/* for 4/8/16 high tiles */
+	case DC_SW_LINEAR:
+		return dm_4k_tile;
+	case DC_SW_4KB_S:
+	case DC_SW_4KB_S_X:
+		return dm_4k_tile;
+	case DC_SW_64KB_S:
+	case DC_SW_64KB_S_X:
+	case DC_SW_64KB_S_T:
+		return dm_64k_tile;
+	case DC_SW_VAR_S:
+	case DC_SW_VAR_S_X:
+		return dm_256k_tile;
+
+	/* For 64bpp 2 high tiles */
+	case DC_SW_4KB_D:
+	case DC_SW_4KB_D_X:
+		return dm_4k_tile;
+	case DC_SW_64KB_D:
+	case DC_SW_64KB_D_X:
+	case DC_SW_64KB_D_T:
+		return dm_64k_tile;
+	case DC_SW_VAR_D:
+	case DC_SW_VAR_D_X:
+		return dm_256k_tile;
+
+	case DC_SW_4KB_R:
+	case DC_SW_4KB_R_X:
+		return dm_4k_tile;
+	case DC_SW_64KB_R:
+	case DC_SW_64KB_R_X:
+		return dm_64k_tile;
+	case DC_SW_VAR_R:
+	case DC_SW_VAR_R_X:
+		return dm_256k_tile;
+
+	/* Unsupported swizzle modes for dcn */
+	case DC_SW_256B_S:
+	default:
+		ASSERT(0); /* Not supported */
+		return 0;
+	}
+}
+
+static void pipe_ctx_to_e2e_pipe_params (
+		const struct pipe_ctx *pipe,
+		struct _vcs_dpi_display_pipe_params_st *input)
+{
+	input->src.is_hsplit = false;
+
+	/* stereo can never be split */
+	if (pipe->plane_state->stereo_format == PLANE_STEREO_FORMAT_SIDE_BY_SIDE ||
+	    pipe->plane_state->stereo_format == PLANE_STEREO_FORMAT_TOP_AND_BOTTOM) {
+		/* reset the split group if it was already considered split. */
+		input->src.hsplit_grp = pipe->pipe_idx;
+	} else if (pipe->top_pipe != NULL && pipe->top_pipe->plane_state == pipe->plane_state) {
+		input->src.is_hsplit = true;
+	} else if (pipe->bottom_pipe != NULL && pipe->bottom_pipe->plane_state == pipe->plane_state) {
+		input->src.is_hsplit = true;
+	}
+
+	if (pipe->plane_res.dpp->ctx->dc->debug.optimized_watermark) {
+		/*
+		 * this method requires us to always re-calculate watermark when dcc change
+		 * between flip.
+		 */
+		input->src.dcc = pipe->plane_state->dcc.enable ? 1 : 0;
+	} else {
+		/*
+		 * allow us to disable dcc on the fly without re-calculating WM
+		 *
+		 * extra overhead for DCC is quite small.  for 1080p WM without
+		 * DCC is only 0.417us lower (urgent goes from 6.979us to 6.562us)
+		 */
+		unsigned int bpe;
+
+		input->src.dcc = pipe->plane_res.dpp->ctx->dc->res_pool->hubbub->funcs->
+			dcc_support_pixel_format(pipe->plane_state->format, &bpe) ? 1 : 0;
+	}
+	input->src.dcc_rate            = 1;
+	input->src.meta_pitch          = pipe->plane_state->dcc.meta_pitch;
+	input->src.source_scan         = dm_horz;
+	input->src.sw_mode             = pipe->plane_state->tiling_info.gfx9.swizzle;
+
+	input->src.viewport_width      = pipe->plane_res.scl_data.viewport.width;
+	input->src.viewport_height     = pipe->plane_res.scl_data.viewport.height;
+	input->src.data_pitch          = pipe->plane_res.scl_data.viewport.width;
+	input->src.data_pitch_c        = pipe->plane_res.scl_data.viewport.width;
+	input->src.cur0_src_width      = 128; /* TODO: Cursor calcs, not curently stored */
+	input->src.cur0_bpp            = 32;
+
+	input->src.macro_tile_size = swizzle_mode_to_macro_tile_size(pipe->plane_state->tiling_info.gfx9.swizzle);
+
+	switch (pipe->plane_state->rotation) {
+	case ROTATION_ANGLE_0:
+	case ROTATION_ANGLE_180:
+		input->src.source_scan = dm_horz;
+		break;
+	case ROTATION_ANGLE_90:
+	case ROTATION_ANGLE_270:
+		input->src.source_scan = dm_vert;
+		break;
+	default:
+		ASSERT(0); /* Not supported */
+		break;
+	}
+
+	/* TODO: Fix pixel format mappings */
+	switch (pipe->plane_state->format) {
+	case SURFACE_PIXEL_FORMAT_VIDEO_420_YCbCr:
+	case SURFACE_PIXEL_FORMAT_VIDEO_420_YCrCb:
+		input->src.source_format = dm_420_8;
+		input->src.viewport_width_c    = input->src.viewport_width / 2;
+		input->src.viewport_height_c   = input->src.viewport_height / 2;
+		break;
+	case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCbCr:
+	case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCrCb:
+		input->src.source_format = dm_420_10;
+		input->src.viewport_width_c    = input->src.viewport_width / 2;
+		input->src.viewport_height_c   = input->src.viewport_height / 2;
+		break;
+	case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616:
+	case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616:
+	case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616F:
+	case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F:
+		input->src.source_format = dm_444_64;
+		input->src.viewport_width_c    = input->src.viewport_width;
+		input->src.viewport_height_c   = input->src.viewport_height;
+		break;
+	case SURFACE_PIXEL_FORMAT_GRPH_RGBE_ALPHA:
+		input->src.source_format = dm_rgbe_alpha;
+		input->src.viewport_width_c    = input->src.viewport_width;
+		input->src.viewport_height_c   = input->src.viewport_height;
+		break;
+	default:
+		input->src.source_format = dm_444_32;
+		input->src.viewport_width_c    = input->src.viewport_width;
+		input->src.viewport_height_c   = input->src.viewport_height;
+		break;
+	}
+
+	input->scale_taps.htaps                = pipe->plane_res.scl_data.taps.h_taps;
+	input->scale_ratio_depth.hscl_ratio    = pipe->plane_res.scl_data.ratios.horz.value/4294967296.0;
+	input->scale_ratio_depth.vscl_ratio    = pipe->plane_res.scl_data.ratios.vert.value/4294967296.0;
+	input->scale_ratio_depth.vinit =  pipe->plane_res.scl_data.inits.v.value/4294967296.0;
+	if (input->scale_ratio_depth.vinit < 1.0)
+			input->scale_ratio_depth.vinit = 1;
+	input->scale_taps.vtaps = pipe->plane_res.scl_data.taps.v_taps;
+	input->scale_taps.vtaps_c = pipe->plane_res.scl_data.taps.v_taps_c;
+	input->scale_taps.htaps_c              = pipe->plane_res.scl_data.taps.h_taps_c;
+	input->scale_ratio_depth.hscl_ratio_c  = pipe->plane_res.scl_data.ratios.horz_c.value/4294967296.0;
+	input->scale_ratio_depth.vscl_ratio_c  = pipe->plane_res.scl_data.ratios.vert_c.value/4294967296.0;
+	input->scale_ratio_depth.vinit_c       = pipe->plane_res.scl_data.inits.v_c.value/4294967296.0;
+	if (input->scale_ratio_depth.vinit_c < 1.0)
+			input->scale_ratio_depth.vinit_c = 1;
+	switch (pipe->plane_res.scl_data.lb_params.depth) {
+	case LB_PIXEL_DEPTH_30BPP:
+		input->scale_ratio_depth.lb_depth = 30; break;
+	case LB_PIXEL_DEPTH_36BPP:
+		input->scale_ratio_depth.lb_depth = 36; break;
+	default:
+		input->scale_ratio_depth.lb_depth = 24; break;
+	}
+
+
+	input->dest.vactive        = pipe->stream->timing.v_addressable + pipe->stream->timing.v_border_top
+			+ pipe->stream->timing.v_border_bottom;
+
+	input->dest.recout_width   = pipe->plane_res.scl_data.recout.width;
+	input->dest.recout_height  = pipe->plane_res.scl_data.recout.height;
+
+	input->dest.full_recout_width   = pipe->plane_res.scl_data.recout.width;
+	input->dest.full_recout_height  = pipe->plane_res.scl_data.recout.height;
+
+	input->dest.htotal         = pipe->stream->timing.h_total;
+	input->dest.hblank_start   = input->dest.htotal - pipe->stream->timing.h_front_porch;
+	input->dest.hblank_end     = input->dest.hblank_start
+			- pipe->stream->timing.h_addressable
+			- pipe->stream->timing.h_border_left
+			- pipe->stream->timing.h_border_right;
+
+	input->dest.vtotal         = pipe->stream->timing.v_total;
+	input->dest.vblank_start   = input->dest.vtotal - pipe->stream->timing.v_front_porch;
+	input->dest.vblank_end     = input->dest.vblank_start
+			- pipe->stream->timing.v_addressable
+			- pipe->stream->timing.v_border_bottom
+			- pipe->stream->timing.v_border_top;
+	input->dest.pixel_rate_mhz = pipe->stream->timing.pix_clk_100hz/10000.0;
+	input->dest.vstartup_start = pipe->pipe_dlg_param.vstartup_start;
+	input->dest.vupdate_offset = pipe->pipe_dlg_param.vupdate_offset;
+	input->dest.vupdate_offset = pipe->pipe_dlg_param.vupdate_offset;
+	input->dest.vupdate_width = pipe->pipe_dlg_param.vupdate_width;
+
+}
+
+static void dcn_bw_calc_rq_dlg_ttu(
+		const struct dc *dc,
+		const struct dcn_bw_internal_vars *v,
+		struct pipe_ctx *pipe,
+		int in_idx)
+{
+	struct display_mode_lib *dml = (struct display_mode_lib *)(&dc->dml);
+	struct _vcs_dpi_display_dlg_regs_st *dlg_regs = &pipe->dlg_regs;
+	struct _vcs_dpi_display_ttu_regs_st *ttu_regs = &pipe->ttu_regs;
+	struct _vcs_dpi_display_rq_regs_st *rq_regs = &pipe->rq_regs;
+	struct _vcs_dpi_display_rq_params_st *rq_param = &pipe->dml_rq_param;
+	struct _vcs_dpi_display_dlg_sys_params_st *dlg_sys_param = &pipe->dml_dlg_sys_param;
+	struct _vcs_dpi_display_e2e_pipe_params_st *input = &pipe->dml_input;
+	float total_active_bw = 0;
+	float total_prefetch_bw = 0;
+	int total_flip_bytes = 0;
+	int i;
+
+	memset(dlg_regs, 0, sizeof(*dlg_regs));
+	memset(ttu_regs, 0, sizeof(*ttu_regs));
+	memset(rq_regs, 0, sizeof(*rq_regs));
+	memset(rq_param, 0, sizeof(*rq_param));
+	memset(dlg_sys_param, 0, sizeof(*dlg_sys_param));
+	memset(input, 0, sizeof(*input));
+
+	for (i = 0; i < number_of_planes; i++) {
+		total_active_bw += v->read_bandwidth[i];
+		total_prefetch_bw += v->prefetch_bandwidth[i];
+		total_flip_bytes += v->total_immediate_flip_bytes[i];
+	}
+	dlg_sys_param->total_flip_bw = v->return_bw - dcn_bw_max2(total_active_bw, total_prefetch_bw);
+	if (dlg_sys_param->total_flip_bw < 0.0)
+		dlg_sys_param->total_flip_bw = 0;
+
+	dlg_sys_param->t_mclk_wm_us = v->dram_clock_change_watermark;
+	dlg_sys_param->t_sr_wm_us = v->stutter_enter_plus_exit_watermark;
+	dlg_sys_param->t_urg_wm_us = v->urgent_watermark;
+	dlg_sys_param->t_extra_us = v->urgent_extra_latency;
+	dlg_sys_param->deepsleep_dcfclk_mhz = v->dcf_clk_deep_sleep;
+	dlg_sys_param->total_flip_bytes = total_flip_bytes;
+
+	pipe_ctx_to_e2e_pipe_params(pipe, &input->pipe);
+	input->clks_cfg.dcfclk_mhz = v->dcfclk;
+	input->clks_cfg.dispclk_mhz = v->dispclk;
+	input->clks_cfg.dppclk_mhz = v->dppclk;
+	input->clks_cfg.refclk_mhz = dc->res_pool->ref_clocks.dchub_ref_clock_inKhz / 1000.0;
+	input->clks_cfg.socclk_mhz = v->socclk;
+	input->clks_cfg.voltage = v->voltage_level;
+//	dc->dml.logger = pool->base.logger;
+	input->dout.output_format = (v->output_format[in_idx] == dcn_bw_420) ? dm_420 : dm_444;
+	input->dout.output_type  = (v->output[in_idx] == dcn_bw_hdmi) ? dm_hdmi : dm_dp;
+	//input[in_idx].dout.output_standard;
+
+	/*todo: soc->sr_enter_plus_exit_time??*/
+
+	dml1_rq_dlg_get_rq_params(dml, rq_param, &input->pipe.src);
+	dml1_extract_rq_regs(dml, rq_regs, rq_param);
+	dml1_rq_dlg_get_dlg_params(
+			dml,
+			dlg_regs,
+			ttu_regs,
+			&rq_param->dlg,
+			dlg_sys_param,
+			input,
+			true,
+			true,
+			v->pte_enable == dcn_bw_yes,
+			pipe->plane_state->flip_immediate);
+}
+
+static void split_stream_across_pipes(
+		struct resource_context *res_ctx,
+		const struct resource_pool *pool,
+		struct pipe_ctx *primary_pipe,
+		struct pipe_ctx *secondary_pipe)
+{
+	int pipe_idx = secondary_pipe->pipe_idx;
+
+	if (!primary_pipe->plane_state)
+		return;
+
+	*secondary_pipe = *primary_pipe;
+
+	secondary_pipe->pipe_idx = pipe_idx;
+	secondary_pipe->plane_res.mi = pool->mis[secondary_pipe->pipe_idx];
+	secondary_pipe->plane_res.hubp = pool->hubps[secondary_pipe->pipe_idx];
+	secondary_pipe->plane_res.ipp = pool->ipps[secondary_pipe->pipe_idx];
+	secondary_pipe->plane_res.xfm = pool->transforms[secondary_pipe->pipe_idx];
+	secondary_pipe->plane_res.dpp = pool->dpps[secondary_pipe->pipe_idx];
+	secondary_pipe->plane_res.mpcc_inst = pool->dpps[secondary_pipe->pipe_idx]->inst;
+	if (primary_pipe->bottom_pipe) {
+		ASSERT(primary_pipe->bottom_pipe != secondary_pipe);
+		secondary_pipe->bottom_pipe = primary_pipe->bottom_pipe;
+		secondary_pipe->bottom_pipe->top_pipe = secondary_pipe;
+	}
+	primary_pipe->bottom_pipe = secondary_pipe;
+	secondary_pipe->top_pipe = primary_pipe;
+
+	resource_build_scaling_params(primary_pipe);
+	resource_build_scaling_params(secondary_pipe);
+}
+
+#if 0
+static void calc_wm_sets_and_perf_params(
+		struct dc_state *context,
+		struct dcn_bw_internal_vars *v)
+{
+	/* Calculate set A last to keep internal var state consistent for required config */
+	if (v->voltage_level < 2) {
+		v->fabric_and_dram_bandwidth_per_state[1] = v->fabric_and_dram_bandwidth_vnom0p8;
+		v->fabric_and_dram_bandwidth_per_state[0] = v->fabric_and_dram_bandwidth_vnom0p8;
+		v->fabric_and_dram_bandwidth = v->fabric_and_dram_bandwidth_vnom0p8;
+		dispclkdppclkdcfclk_deep_sleep_prefetch_parameters_watermarks_and_performance_calculation(v);
+
+		context->bw_ctx.bw.dcn.watermarks.b.cstate_pstate.cstate_exit_ns =
+			v->stutter_exit_watermark * 1000;
+		context->bw_ctx.bw.dcn.watermarks.b.cstate_pstate.cstate_enter_plus_exit_ns =
+				v->stutter_enter_plus_exit_watermark * 1000;
+		context->bw_ctx.bw.dcn.watermarks.b.cstate_pstate.pstate_change_ns =
+				v->dram_clock_change_watermark * 1000;
+		context->bw_ctx.bw.dcn.watermarks.b.pte_meta_urgent_ns = v->ptemeta_urgent_watermark * 1000;
+		context->bw_ctx.bw.dcn.watermarks.b.urgent_ns = v->urgent_watermark * 1000;
+
+		v->dcfclk_per_state[1] = v->dcfclkv_nom0p8;
+		v->dcfclk_per_state[0] = v->dcfclkv_nom0p8;
+		v->dcfclk = v->dcfclkv_nom0p8;
+		dispclkdppclkdcfclk_deep_sleep_prefetch_parameters_watermarks_and_performance_calculation(v);
+
+		context->bw_ctx.bw.dcn.watermarks.c.cstate_pstate.cstate_exit_ns =
+			v->stutter_exit_watermark * 1000;
+		context->bw_ctx.bw.dcn.watermarks.c.cstate_pstate.cstate_enter_plus_exit_ns =
+				v->stutter_enter_plus_exit_watermark * 1000;
+		context->bw_ctx.bw.dcn.watermarks.c.cstate_pstate.pstate_change_ns =
+				v->dram_clock_change_watermark * 1000;
+		context->bw_ctx.bw.dcn.watermarks.c.pte_meta_urgent_ns = v->ptemeta_urgent_watermark * 1000;
+		context->bw_ctx.bw.dcn.watermarks.c.urgent_ns = v->urgent_watermark * 1000;
+	}
+
+	if (v->voltage_level < 3) {
+		v->fabric_and_dram_bandwidth_per_state[2] = v->fabric_and_dram_bandwidth_vmax0p9;
+		v->fabric_and_dram_bandwidth_per_state[1] = v->fabric_and_dram_bandwidth_vmax0p9;
+		v->fabric_and_dram_bandwidth_per_state[0] = v->fabric_and_dram_bandwidth_vmax0p9;
+		v->fabric_and_dram_bandwidth = v->fabric_and_dram_bandwidth_vmax0p9;
+		v->dcfclk_per_state[2] = v->dcfclkv_max0p9;
+		v->dcfclk_per_state[1] = v->dcfclkv_max0p9;
+		v->dcfclk_per_state[0] = v->dcfclkv_max0p9;
+		v->dcfclk = v->dcfclkv_max0p9;
+		dispclkdppclkdcfclk_deep_sleep_prefetch_parameters_watermarks_and_performance_calculation(v);
+
+		context->bw_ctx.bw.dcn.watermarks.d.cstate_pstate.cstate_exit_ns =
+			v->stutter_exit_watermark * 1000;
+		context->bw_ctx.bw.dcn.watermarks.d.cstate_pstate.cstate_enter_plus_exit_ns =
+				v->stutter_enter_plus_exit_watermark * 1000;
+		context->bw_ctx.bw.dcn.watermarks.d.cstate_pstate.pstate_change_ns =
+				v->dram_clock_change_watermark * 1000;
+		context->bw_ctx.bw.dcn.watermarks.d.pte_meta_urgent_ns = v->ptemeta_urgent_watermark * 1000;
+		context->bw_ctx.bw.dcn.watermarks.d.urgent_ns = v->urgent_watermark * 1000;
+	}
+
+	v->fabric_and_dram_bandwidth_per_state[2] = v->fabric_and_dram_bandwidth_vnom0p8;
+	v->fabric_and_dram_bandwidth_per_state[1] = v->fabric_and_dram_bandwidth_vmid0p72;
+	v->fabric_and_dram_bandwidth_per_state[0] = v->fabric_and_dram_bandwidth_vmin0p65;
+	v->fabric_and_dram_bandwidth = v->fabric_and_dram_bandwidth_per_state[v->voltage_level];
+	v->dcfclk_per_state[2] = v->dcfclkv_nom0p8;
+	v->dcfclk_per_state[1] = v->dcfclkv_mid0p72;
+	v->dcfclk_per_state[0] = v->dcfclkv_min0p65;
+	v->dcfclk = v->dcfclk_per_state[v->voltage_level];
+	dispclkdppclkdcfclk_deep_sleep_prefetch_parameters_watermarks_and_performance_calculation(v);
+
+	context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.cstate_exit_ns =
+		v->stutter_exit_watermark * 1000;
+	context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.cstate_enter_plus_exit_ns =
+			v->stutter_enter_plus_exit_watermark * 1000;
+	context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.pstate_change_ns =
+			v->dram_clock_change_watermark * 1000;
+	context->bw_ctx.bw.dcn.watermarks.a.pte_meta_urgent_ns = v->ptemeta_urgent_watermark * 1000;
+	context->bw_ctx.bw.dcn.watermarks.a.urgent_ns = v->urgent_watermark * 1000;
+	if (v->voltage_level >= 2) {
+		context->bw_ctx.bw.dcn.watermarks.b = context->bw_ctx.bw.dcn.watermarks.a;
+		context->bw_ctx.bw.dcn.watermarks.c = context->bw_ctx.bw.dcn.watermarks.a;
+	}
+	if (v->voltage_level >= 3)
+		context->bw_ctx.bw.dcn.watermarks.d = context->bw_ctx.bw.dcn.watermarks.a;
+}
+#endif
+
+static bool dcn_bw_apply_registry_override(struct dc *dc)
+{
+	bool updated = false;
+
+	if ((int)(dc->dcn_soc->sr_exit_time * 1000) != dc->debug.sr_exit_time_ns
+			&& dc->debug.sr_exit_time_ns) {
+		updated = true;
+		dc->dcn_soc->sr_exit_time = dc->debug.sr_exit_time_ns / 1000.0;
+	}
+
+	if ((int)(dc->dcn_soc->sr_enter_plus_exit_time * 1000)
+				!= dc->debug.sr_enter_plus_exit_time_ns
+			&& dc->debug.sr_enter_plus_exit_time_ns) {
+		updated = true;
+		dc->dcn_soc->sr_enter_plus_exit_time =
+				dc->debug.sr_enter_plus_exit_time_ns / 1000.0;
+	}
+
+	if ((int)(dc->dcn_soc->urgent_latency * 1000) != dc->debug.urgent_latency_ns
+			&& dc->debug.urgent_latency_ns) {
+		updated = true;
+		dc->dcn_soc->urgent_latency = dc->debug.urgent_latency_ns / 1000.0;
+	}
+
+	if ((int)(dc->dcn_soc->percent_of_ideal_drambw_received_after_urg_latency * 1000)
+				!= dc->debug.percent_of_ideal_drambw
+			&& dc->debug.percent_of_ideal_drambw) {
+		updated = true;
+		dc->dcn_soc->percent_of_ideal_drambw_received_after_urg_latency =
+				dc->debug.percent_of_ideal_drambw;
+	}
+
+	if ((int)(dc->dcn_soc->dram_clock_change_latency * 1000)
+				!= dc->debug.dram_clock_change_latency_ns
+			&& dc->debug.dram_clock_change_latency_ns) {
+		updated = true;
+		dc->dcn_soc->dram_clock_change_latency =
+				dc->debug.dram_clock_change_latency_ns / 1000.0;
+	}
+
+	return updated;
+}
+
+static void hack_disable_optional_pipe_split(struct dcn_bw_internal_vars *v)
+{
+	/*
+	 * disable optional pipe split by lower dispclk bounding box
+	 * at DPM0
+	 */
+	v->max_dispclk[0] = v->max_dppclk_vmin0p65;
+}
+
+static void hack_force_pipe_split(struct dcn_bw_internal_vars *v,
+		unsigned int pixel_rate_100hz)
+{
+	float pixel_rate_mhz = pixel_rate_100hz / 10000.0;
+
+	/*
+	 * force enabling pipe split by lower dpp clock for DPM0 to just
+	 * below the specify pixel_rate, so bw calc would split pipe.
+	 */
+	if (pixel_rate_mhz < v->max_dppclk[0])
+		v->max_dppclk[0] = pixel_rate_mhz;
+}
+
+static void hack_bounding_box(struct dcn_bw_internal_vars *v,
+		struct dc_debug_options *dbg,
+		struct dc_state *context)
+{
+	int i;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
+
+		/**
+		 * Workaround for avoiding pipe-split in cases where we'd split
+		 * planes that are too small, resulting in splits that aren't
+		 * valid for the scaler.
+		 */
+		if (pipe->plane_state &&
+		    (pipe->plane_state->dst_rect.width <= 16 ||
+		     pipe->plane_state->dst_rect.height <= 16 ||
+		     pipe->plane_state->src_rect.width <= 16 ||
+		     pipe->plane_state->src_rect.height <= 16)) {
+			hack_disable_optional_pipe_split(v);
+			return;
+		}
+	}
+
+	if (dbg->pipe_split_policy == MPC_SPLIT_AVOID)
+		hack_disable_optional_pipe_split(v);
+
+	if (dbg->pipe_split_policy == MPC_SPLIT_AVOID_MULT_DISP &&
+		context->stream_count >= 2)
+		hack_disable_optional_pipe_split(v);
+
+	if (context->stream_count == 1 &&
+			dbg->force_single_disp_pipe_split)
+		hack_force_pipe_split(v, context->streams[0]->timing.pix_clk_100hz);
+}
+
+static unsigned int get_highest_allowed_voltage_level(bool is_vmin_only_asic)
+{
+	/* for low power RV2 variants, the highest voltage level we want is 0 */
+	if (is_vmin_only_asic)
+		return 0;
+	else	/* we are ok with all levels */
+		return 4;
+}
+
+bool dcn_validate_bandwidth(
+		struct dc *dc,
+		struct dc_state *context,
+		enum dc_validate_mode validate_mode)
+{
+	/*
+	 * we want a breakdown of the various stages of validation, which the
+	 * perf_trace macro doesn't support
+	 */
+	BW_VAL_TRACE_SETUP();
+
+	const struct resource_pool *pool = dc->res_pool;
+	struct dcn_bw_internal_vars *v = &context->dcn_bw_vars;
+	int i, input_idx, k;
+	int vesa_sync_start, asic_blank_end, asic_blank_start;
+	bool bw_limit_pass;
+	float bw_limit;
+
+	PERFORMANCE_TRACE_START();
+
+	BW_VAL_TRACE_COUNT();
+
+	if (dcn_bw_apply_registry_override(dc))
+		dcn_bw_sync_calcs_and_dml(dc);
+
+	memset(v, 0, sizeof(*v));
+
+	v->sr_exit_time = dc->dcn_soc->sr_exit_time;
+	v->sr_enter_plus_exit_time = dc->dcn_soc->sr_enter_plus_exit_time;
+	v->urgent_latency = dc->dcn_soc->urgent_latency;
+	v->write_back_latency = dc->dcn_soc->write_back_latency;
+	v->percent_of_ideal_drambw_received_after_urg_latency =
+			dc->dcn_soc->percent_of_ideal_drambw_received_after_urg_latency;
+
+	v->dcfclkv_min0p65 = dc->dcn_soc->dcfclkv_min0p65;
+	v->dcfclkv_mid0p72 = dc->dcn_soc->dcfclkv_mid0p72;
+	v->dcfclkv_nom0p8 = dc->dcn_soc->dcfclkv_nom0p8;
+	v->dcfclkv_max0p9 = dc->dcn_soc->dcfclkv_max0p9;
+
+	v->max_dispclk_vmin0p65 = dc->dcn_soc->max_dispclk_vmin0p65;
+	v->max_dispclk_vmid0p72 = dc->dcn_soc->max_dispclk_vmid0p72;
+	v->max_dispclk_vnom0p8 = dc->dcn_soc->max_dispclk_vnom0p8;
+	v->max_dispclk_vmax0p9 = dc->dcn_soc->max_dispclk_vmax0p9;
+
+	v->max_dppclk_vmin0p65 = dc->dcn_soc->max_dppclk_vmin0p65;
+	v->max_dppclk_vmid0p72 = dc->dcn_soc->max_dppclk_vmid0p72;
+	v->max_dppclk_vnom0p8 = dc->dcn_soc->max_dppclk_vnom0p8;
+	v->max_dppclk_vmax0p9 = dc->dcn_soc->max_dppclk_vmax0p9;
+
+	v->socclk = dc->dcn_soc->socclk;
+
+	v->fabric_and_dram_bandwidth_vmin0p65 = dc->dcn_soc->fabric_and_dram_bandwidth_vmin0p65;
+	v->fabric_and_dram_bandwidth_vmid0p72 = dc->dcn_soc->fabric_and_dram_bandwidth_vmid0p72;
+	v->fabric_and_dram_bandwidth_vnom0p8 = dc->dcn_soc->fabric_and_dram_bandwidth_vnom0p8;
+	v->fabric_and_dram_bandwidth_vmax0p9 = dc->dcn_soc->fabric_and_dram_bandwidth_vmax0p9;
+
+	v->phyclkv_min0p65 = dc->dcn_soc->phyclkv_min0p65;
+	v->phyclkv_mid0p72 = dc->dcn_soc->phyclkv_mid0p72;
+	v->phyclkv_nom0p8 = dc->dcn_soc->phyclkv_nom0p8;
+	v->phyclkv_max0p9 = dc->dcn_soc->phyclkv_max0p9;
+
+	v->downspreading = dc->dcn_soc->downspreading;
+	v->round_trip_ping_latency_cycles = dc->dcn_soc->round_trip_ping_latency_cycles;
+	v->urgent_out_of_order_return_per_channel = dc->dcn_soc->urgent_out_of_order_return_per_channel;
+	v->number_of_channels = dc->dcn_soc->number_of_channels;
+	v->vmm_page_size = dc->dcn_soc->vmm_page_size;
+	v->dram_clock_change_latency = dc->dcn_soc->dram_clock_change_latency;
+	v->return_bus_width = dc->dcn_soc->return_bus_width;
+
+	v->rob_buffer_size_in_kbyte = dc->dcn_ip->rob_buffer_size_in_kbyte;
+	v->det_buffer_size_in_kbyte = dc->dcn_ip->det_buffer_size_in_kbyte;
+	v->dpp_output_buffer_pixels = dc->dcn_ip->dpp_output_buffer_pixels;
+	v->opp_output_buffer_lines = dc->dcn_ip->opp_output_buffer_lines;
+	v->pixel_chunk_size_in_kbyte = dc->dcn_ip->pixel_chunk_size_in_kbyte;
+	v->pte_enable = dc->dcn_ip->pte_enable;
+	v->pte_chunk_size = dc->dcn_ip->pte_chunk_size;
+	v->meta_chunk_size = dc->dcn_ip->meta_chunk_size;
+	v->writeback_chunk_size = dc->dcn_ip->writeback_chunk_size;
+	v->odm_capability = dc->dcn_ip->odm_capability;
+	v->dsc_capability = dc->dcn_ip->dsc_capability;
+	v->line_buffer_size = dc->dcn_ip->line_buffer_size;
+	v->is_line_buffer_bpp_fixed = dc->dcn_ip->is_line_buffer_bpp_fixed;
+	v->line_buffer_fixed_bpp = dc->dcn_ip->line_buffer_fixed_bpp;
+	v->max_line_buffer_lines = dc->dcn_ip->max_line_buffer_lines;
+	v->writeback_luma_buffer_size = dc->dcn_ip->writeback_luma_buffer_size;
+	v->writeback_chroma_buffer_size = dc->dcn_ip->writeback_chroma_buffer_size;
+	v->max_num_dpp = dc->dcn_ip->max_num_dpp;
+	v->max_num_writeback = dc->dcn_ip->max_num_writeback;
+	v->max_dchub_topscl_throughput = dc->dcn_ip->max_dchub_topscl_throughput;
+	v->max_pscl_tolb_throughput = dc->dcn_ip->max_pscl_tolb_throughput;
+	v->max_lb_tovscl_throughput = dc->dcn_ip->max_lb_tovscl_throughput;
+	v->max_vscl_tohscl_throughput = dc->dcn_ip->max_vscl_tohscl_throughput;
+	v->max_hscl_ratio = dc->dcn_ip->max_hscl_ratio;
+	v->max_vscl_ratio = dc->dcn_ip->max_vscl_ratio;
+	v->max_hscl_taps = dc->dcn_ip->max_hscl_taps;
+	v->max_vscl_taps = dc->dcn_ip->max_vscl_taps;
+	v->under_scan_factor = dc->dcn_ip->under_scan_factor;
+	v->pte_buffer_size_in_requests = dc->dcn_ip->pte_buffer_size_in_requests;
+	v->dispclk_ramping_margin = dc->dcn_ip->dispclk_ramping_margin;
+	v->max_inter_dcn_tile_repeaters = dc->dcn_ip->max_inter_dcn_tile_repeaters;
+	v->can_vstartup_lines_exceed_vsync_plus_back_porch_lines_minus_one =
+			dc->dcn_ip->can_vstartup_lines_exceed_vsync_plus_back_porch_lines_minus_one;
+	v->bug_forcing_luma_and_chroma_request_to_same_size_fixed =
+			dc->dcn_ip->bug_forcing_luma_and_chroma_request_to_same_size_fixed;
+
+	v->voltage[5] = dcn_bw_no_support;
+	v->voltage[4] = dcn_bw_v_max0p9;
+	v->voltage[3] = dcn_bw_v_max0p9;
+	v->voltage[2] = dcn_bw_v_nom0p8;
+	v->voltage[1] = dcn_bw_v_mid0p72;
+	v->voltage[0] = dcn_bw_v_min0p65;
+	v->fabric_and_dram_bandwidth_per_state[5] = v->fabric_and_dram_bandwidth_vmax0p9;
+	v->fabric_and_dram_bandwidth_per_state[4] = v->fabric_and_dram_bandwidth_vmax0p9;
+	v->fabric_and_dram_bandwidth_per_state[3] = v->fabric_and_dram_bandwidth_vmax0p9;
+	v->fabric_and_dram_bandwidth_per_state[2] = v->fabric_and_dram_bandwidth_vnom0p8;
+	v->fabric_and_dram_bandwidth_per_state[1] = v->fabric_and_dram_bandwidth_vmid0p72;
+	v->fabric_and_dram_bandwidth_per_state[0] = v->fabric_and_dram_bandwidth_vmin0p65;
+	v->dcfclk_per_state[5] = v->dcfclkv_max0p9;
+	v->dcfclk_per_state[4] = v->dcfclkv_max0p9;
+	v->dcfclk_per_state[3] = v->dcfclkv_max0p9;
+	v->dcfclk_per_state[2] = v->dcfclkv_nom0p8;
+	v->dcfclk_per_state[1] = v->dcfclkv_mid0p72;
+	v->dcfclk_per_state[0] = v->dcfclkv_min0p65;
+	v->max_dispclk[5] = v->max_dispclk_vmax0p9;
+	v->max_dispclk[4] = v->max_dispclk_vmax0p9;
+	v->max_dispclk[3] = v->max_dispclk_vmax0p9;
+	v->max_dispclk[2] = v->max_dispclk_vnom0p8;
+	v->max_dispclk[1] = v->max_dispclk_vmid0p72;
+	v->max_dispclk[0] = v->max_dispclk_vmin0p65;
+	v->max_dppclk[5] = v->max_dppclk_vmax0p9;
+	v->max_dppclk[4] = v->max_dppclk_vmax0p9;
+	v->max_dppclk[3] = v->max_dppclk_vmax0p9;
+	v->max_dppclk[2] = v->max_dppclk_vnom0p8;
+	v->max_dppclk[1] = v->max_dppclk_vmid0p72;
+	v->max_dppclk[0] = v->max_dppclk_vmin0p65;
+	v->phyclk_per_state[5] = v->phyclkv_max0p9;
+	v->phyclk_per_state[4] = v->phyclkv_max0p9;
+	v->phyclk_per_state[3] = v->phyclkv_max0p9;
+	v->phyclk_per_state[2] = v->phyclkv_nom0p8;
+	v->phyclk_per_state[1] = v->phyclkv_mid0p72;
+	v->phyclk_per_state[0] = v->phyclkv_min0p65;
+	v->synchronized_vblank = dcn_bw_no;
+	v->ta_pscalculation = dcn_bw_override;
+	v->allow_different_hratio_vratio = dcn_bw_yes;
+
+	for (i = 0, input_idx = 0; i < pool->pipe_count; i++) {
+		struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
+
+		if (!pipe->stream)
+			continue;
+		/* skip all but first of split pipes */
+		if (pipe->top_pipe && pipe->top_pipe->plane_state == pipe->plane_state)
+			continue;
+
+		v->underscan_output[input_idx] = false; /* taken care of in recout already*/
+		v->interlace_output[input_idx] = false;
+
+		v->htotal[input_idx] = pipe->stream->timing.h_total;
+		v->vtotal[input_idx] = pipe->stream->timing.v_total;
+		v->vactive[input_idx] = pipe->stream->timing.v_addressable +
+				pipe->stream->timing.v_border_top + pipe->stream->timing.v_border_bottom;
+		v->v_sync_plus_back_porch[input_idx] = pipe->stream->timing.v_total
+				- v->vactive[input_idx]
+				- pipe->stream->timing.v_front_porch;
+		v->pixel_clock[input_idx] = pipe->stream->timing.pix_clk_100hz/10000.0;
+		if (pipe->stream->timing.timing_3d_format == TIMING_3D_FORMAT_HW_FRAME_PACKING)
+			v->pixel_clock[input_idx] *= 2;
+		if (!pipe->plane_state) {
+			v->dcc_enable[input_idx] = dcn_bw_yes;
+			v->source_pixel_format[input_idx] = dcn_bw_rgb_sub_32;
+			v->source_surface_mode[input_idx] = dcn_bw_sw_4_kb_s;
+			v->lb_bit_per_pixel[input_idx] = 30;
+			v->viewport_width[input_idx] = pipe->stream->timing.h_addressable;
+			v->viewport_height[input_idx] = pipe->stream->timing.v_addressable;
+			/*
+			 * for cases where we have no plane, we want to validate up to 1080p
+			 * source size because here we are only interested in if the output
+			 * timing is supported or not. if we cannot support native resolution
+			 * of the high res display, we still want to support lower res up scale
+			 * to native
+			 */
+			if (v->viewport_width[input_idx] > 1920)
+				v->viewport_width[input_idx] = 1920;
+			if (v->viewport_height[input_idx] > 1080)
+				v->viewport_height[input_idx] = 1080;
+			v->scaler_rec_out_width[input_idx] = v->viewport_width[input_idx];
+			v->scaler_recout_height[input_idx] = v->viewport_height[input_idx];
+			v->override_hta_ps[input_idx] = 1;
+			v->override_vta_ps[input_idx] = 1;
+			v->override_hta_pschroma[input_idx] = 1;
+			v->override_vta_pschroma[input_idx] = 1;
+			v->source_scan[input_idx] = dcn_bw_hor;
+
+		} else {
+			v->viewport_height[input_idx] =  pipe->plane_res.scl_data.viewport.height;
+			v->viewport_width[input_idx] = pipe->plane_res.scl_data.viewport.width;
+			v->scaler_rec_out_width[input_idx] = pipe->plane_res.scl_data.recout.width;
+			v->scaler_recout_height[input_idx] = pipe->plane_res.scl_data.recout.height;
+			if (pipe->bottom_pipe && pipe->bottom_pipe->plane_state == pipe->plane_state) {
+				if (pipe->plane_state->rotation % 2 == 0) {
+					int viewport_end = pipe->plane_res.scl_data.viewport.width
+							+ pipe->plane_res.scl_data.viewport.x;
+					int viewport_b_end = pipe->bottom_pipe->plane_res.scl_data.viewport.width
+							+ pipe->bottom_pipe->plane_res.scl_data.viewport.x;
+
+					if (viewport_end > viewport_b_end)
+						v->viewport_width[input_idx] = viewport_end
+							- pipe->bottom_pipe->plane_res.scl_data.viewport.x;
+					else
+						v->viewport_width[input_idx] = viewport_b_end
+									- pipe->plane_res.scl_data.viewport.x;
+				} else  {
+					int viewport_end = pipe->plane_res.scl_data.viewport.height
+						+ pipe->plane_res.scl_data.viewport.y;
+					int viewport_b_end = pipe->bottom_pipe->plane_res.scl_data.viewport.height
+						+ pipe->bottom_pipe->plane_res.scl_data.viewport.y;
+
+					if (viewport_end > viewport_b_end)
+						v->viewport_height[input_idx] = viewport_end
+							- pipe->bottom_pipe->plane_res.scl_data.viewport.y;
+					else
+						v->viewport_height[input_idx] = viewport_b_end
+									- pipe->plane_res.scl_data.viewport.y;
+				}
+				v->scaler_rec_out_width[input_idx] = pipe->plane_res.scl_data.recout.width
+						+ pipe->bottom_pipe->plane_res.scl_data.recout.width;
+			}
+
+			if (pipe->plane_state->rotation % 2 == 0) {
+				ASSERT(pipe->plane_res.scl_data.ratios.horz.value != dc_fixpt_one.value
+					|| v->scaler_rec_out_width[input_idx] == v->viewport_width[input_idx]);
+				ASSERT(pipe->plane_res.scl_data.ratios.vert.value != dc_fixpt_one.value
+					|| v->scaler_recout_height[input_idx] == v->viewport_height[input_idx]);
+			} else {
+				ASSERT(pipe->plane_res.scl_data.ratios.horz.value != dc_fixpt_one.value
+					|| v->scaler_recout_height[input_idx] == v->viewport_width[input_idx]);
+				ASSERT(pipe->plane_res.scl_data.ratios.vert.value != dc_fixpt_one.value
+					|| v->scaler_rec_out_width[input_idx] == v->viewport_height[input_idx]);
+			}
+
+			if (dc->debug.optimized_watermark) {
+				/*
+				 * this method requires us to always re-calculate watermark when dcc change
+				 * between flip.
+				 */
+				v->dcc_enable[input_idx] = pipe->plane_state->dcc.enable ? dcn_bw_yes : dcn_bw_no;
+			} else {
+				/*
+				 * allow us to disable dcc on the fly without re-calculating WM
+				 *
+				 * extra overhead for DCC is quite small.  for 1080p WM without
+				 * DCC is only 0.417us lower (urgent goes from 6.979us to 6.562us)
+				 */
+				unsigned int bpe;
+
+				v->dcc_enable[input_idx] = dc->res_pool->hubbub->funcs->dcc_support_pixel_format(
+						pipe->plane_state->format, &bpe) ? dcn_bw_yes : dcn_bw_no;
+			}
+
+			v->source_pixel_format[input_idx] = tl_pixel_format_to_bw_defs(
+					pipe->plane_state->format);
+			v->source_surface_mode[input_idx] = tl_sw_mode_to_bw_defs(
+					pipe->plane_state->tiling_info.gfx9.swizzle);
+			v->lb_bit_per_pixel[input_idx] = tl_lb_bpp_to_int(pipe->plane_res.scl_data.lb_params.depth);
+			v->override_hta_ps[input_idx] = pipe->plane_res.scl_data.taps.h_taps;
+			v->override_vta_ps[input_idx] = pipe->plane_res.scl_data.taps.v_taps;
+			v->override_hta_pschroma[input_idx] = pipe->plane_res.scl_data.taps.h_taps_c;
+			v->override_vta_pschroma[input_idx] = pipe->plane_res.scl_data.taps.v_taps_c;
+			/*
+			 * Spreadsheet doesn't handle taps_c is one properly,
+			 * need to force Chroma to always be scaled to pass
+			 * bandwidth validation.
+			 */
+			if (v->override_hta_pschroma[input_idx] == 1)
+				v->override_hta_pschroma[input_idx] = 2;
+			if (v->override_vta_pschroma[input_idx] == 1)
+				v->override_vta_pschroma[input_idx] = 2;
+			v->source_scan[input_idx] = (pipe->plane_state->rotation % 2) ? dcn_bw_vert : dcn_bw_hor;
+		}
+		if (v->is_line_buffer_bpp_fixed == dcn_bw_yes)
+			v->lb_bit_per_pixel[input_idx] = v->line_buffer_fixed_bpp;
+		v->dcc_rate[input_idx] = 1; /*TODO: Worst case? does this change?*/
+		v->output_format[input_idx] = pipe->stream->timing.pixel_encoding ==
+				PIXEL_ENCODING_YCBCR420 ? dcn_bw_420 : dcn_bw_444;
+		v->output[input_idx] = pipe->stream->signal ==
+				SIGNAL_TYPE_HDMI_TYPE_A ? dcn_bw_hdmi : dcn_bw_dp;
+		v->output_deep_color[input_idx] = dcn_bw_encoder_8bpc;
+		if (v->output[input_idx] == dcn_bw_hdmi) {
+			switch (pipe->stream->timing.display_color_depth) {
+			case COLOR_DEPTH_101010:
+				v->output_deep_color[input_idx] = dcn_bw_encoder_10bpc;
+				break;
+			case COLOR_DEPTH_121212:
+				v->output_deep_color[input_idx]  = dcn_bw_encoder_12bpc;
+				break;
+			case COLOR_DEPTH_161616:
+				v->output_deep_color[input_idx]  = dcn_bw_encoder_16bpc;
+				break;
+			default:
+				break;
+			}
+		}
+
+		input_idx++;
+	}
+	v->number_of_active_planes = input_idx;
+
+	scaler_settings_calculation(v);
+
+	hack_bounding_box(v, &dc->debug, context);
+
+	mode_support_and_system_configuration(v);
+
+	/* Unhack dppclk: dont bother with trying to pipe split if we cannot maintain dpm0 */
+	if (v->voltage_level != 0
+			&& context->stream_count == 1
+			&& dc->debug.force_single_disp_pipe_split) {
+		v->max_dppclk[0] = v->max_dppclk_vmin0p65;
+		mode_support_and_system_configuration(v);
+	}
+
+	if (v->voltage_level == 0 &&
+			(dc->debug.sr_exit_time_dpm0_ns
+				|| dc->debug.sr_enter_plus_exit_time_dpm0_ns)) {
+
+		if (dc->debug.sr_enter_plus_exit_time_dpm0_ns)
+			v->sr_enter_plus_exit_time =
+				dc->debug.sr_enter_plus_exit_time_dpm0_ns / 1000.0f;
+		if (dc->debug.sr_exit_time_dpm0_ns)
+			v->sr_exit_time =  dc->debug.sr_exit_time_dpm0_ns / 1000.0f;
+		context->bw_ctx.dml.soc.sr_enter_plus_exit_time_us = v->sr_enter_plus_exit_time;
+		context->bw_ctx.dml.soc.sr_exit_time_us = v->sr_exit_time;
+		mode_support_and_system_configuration(v);
+	}
+
+	display_pipe_configuration(v);
+
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->source_scan[k] == dcn_bw_hor)
+			v->swath_width_y[k] = v->viewport_width[k] / v->dpp_per_plane[k];
+		else
+			v->swath_width_y[k] = v->viewport_height[k] / v->dpp_per_plane[k];
+	}
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		if (v->source_pixel_format[k] == dcn_bw_rgb_sub_64) {
+			v->byte_per_pixel_dety[k] = 8.0;
+			v->byte_per_pixel_detc[k] = 0.0;
+		} else if (v->source_pixel_format[k] == dcn_bw_rgb_sub_32) {
+			v->byte_per_pixel_dety[k] = 4.0;
+			v->byte_per_pixel_detc[k] = 0.0;
+		} else if (v->source_pixel_format[k] == dcn_bw_rgb_sub_16) {
+			v->byte_per_pixel_dety[k] = 2.0;
+			v->byte_per_pixel_detc[k] = 0.0;
+		} else if (v->source_pixel_format[k] == dcn_bw_yuv420_sub_8) {
+			v->byte_per_pixel_dety[k] = 1.0;
+			v->byte_per_pixel_detc[k] = 2.0;
+		} else {
+			v->byte_per_pixel_dety[k] = 4.0f / 3.0f;
+			v->byte_per_pixel_detc[k] = 8.0f / 3.0f;
+		}
+	}
+
+	v->total_data_read_bandwidth = 0.0;
+	for (k = 0; k <= v->number_of_active_planes - 1; k++) {
+		v->read_bandwidth_plane_luma[k] = v->swath_width_y[k] * v->dpp_per_plane[k] *
+				dcn_bw_ceil2(v->byte_per_pixel_dety[k], 1.0) / (v->htotal[k] / v->pixel_clock[k]) * v->v_ratio[k];
+		v->read_bandwidth_plane_chroma[k] = v->swath_width_y[k] / 2.0 * v->dpp_per_plane[k] *
+				dcn_bw_ceil2(v->byte_per_pixel_detc[k], 2.0) / (v->htotal[k] / v->pixel_clock[k]) * v->v_ratio[k] / 2.0;
+		v->total_data_read_bandwidth = v->total_data_read_bandwidth +
+				v->read_bandwidth_plane_luma[k] + v->read_bandwidth_plane_chroma[k];
+	}
+
+	BW_VAL_TRACE_END_VOLTAGE_LEVEL();
+
+	if (v->voltage_level != number_of_states_plus_one && validate_mode == DC_VALIDATE_MODE_AND_PROGRAMMING) {
+		float bw_consumed = v->total_bandwidth_consumed_gbyte_per_second;
+
+		if (bw_consumed < v->fabric_and_dram_bandwidth_vmin0p65)
+			bw_consumed = v->fabric_and_dram_bandwidth_vmin0p65;
+		else if (bw_consumed < v->fabric_and_dram_bandwidth_vmid0p72)
+			bw_consumed = v->fabric_and_dram_bandwidth_vmid0p72;
+		else if (bw_consumed < v->fabric_and_dram_bandwidth_vnom0p8)
+			bw_consumed = v->fabric_and_dram_bandwidth_vnom0p8;
+		else
+			bw_consumed = v->fabric_and_dram_bandwidth_vmax0p9;
+
+		if (bw_consumed < v->fabric_and_dram_bandwidth)
+			if (dc->debug.voltage_align_fclk)
+				bw_consumed = v->fabric_and_dram_bandwidth;
+
+		display_pipe_configuration(v);
+		/*calc_wm_sets_and_perf_params(context, v);*/
+		/* Only 1 set is used by dcn since no noticeable
+		 * performance improvement was measured and due to hw bug DEGVIDCN10-254
+		 */
+		dispclkdppclkdcfclk_deep_sleep_prefetch_parameters_watermarks_and_performance_calculation(v);
+
+		context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.cstate_exit_ns =
+			v->stutter_exit_watermark * 1000;
+		context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.cstate_enter_plus_exit_ns =
+				v->stutter_enter_plus_exit_watermark * 1000;
+		context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.pstate_change_ns =
+				v->dram_clock_change_watermark * 1000;
+		context->bw_ctx.bw.dcn.watermarks.a.pte_meta_urgent_ns = v->ptemeta_urgent_watermark * 1000;
+		context->bw_ctx.bw.dcn.watermarks.a.urgent_ns = v->urgent_watermark * 1000;
+		context->bw_ctx.bw.dcn.watermarks.b = context->bw_ctx.bw.dcn.watermarks.a;
+		context->bw_ctx.bw.dcn.watermarks.c = context->bw_ctx.bw.dcn.watermarks.a;
+		context->bw_ctx.bw.dcn.watermarks.d = context->bw_ctx.bw.dcn.watermarks.a;
+
+		context->bw_ctx.bw.dcn.clk.fclk_khz = (int)(bw_consumed * 1000000 /
+				(ddr4_dram_factor_single_Channel * v->number_of_channels));
+		if (bw_consumed == v->fabric_and_dram_bandwidth_vmin0p65)
+			context->bw_ctx.bw.dcn.clk.fclk_khz = (int)(bw_consumed * 1000000 / 32);
+
+		context->bw_ctx.bw.dcn.clk.dcfclk_deep_sleep_khz = (int)(v->dcf_clk_deep_sleep * 1000);
+		context->bw_ctx.bw.dcn.clk.dcfclk_khz = (int)(v->dcfclk * 1000);
+
+		context->bw_ctx.bw.dcn.clk.dispclk_khz = (int)(v->dispclk * 1000);
+		if (dc->debug.max_disp_clk == true)
+			context->bw_ctx.bw.dcn.clk.dispclk_khz = (int)(dc->dcn_soc->max_dispclk_vmax0p9 * 1000);
+
+		if (context->bw_ctx.bw.dcn.clk.dispclk_khz <
+				dc->debug.min_disp_clk_khz) {
+			context->bw_ctx.bw.dcn.clk.dispclk_khz =
+					dc->debug.min_disp_clk_khz;
+		}
+
+		context->bw_ctx.bw.dcn.clk.dppclk_khz = context->bw_ctx.bw.dcn.clk.dispclk_khz /
+				v->dispclk_dppclk_ratio;
+		context->bw_ctx.bw.dcn.clk.phyclk_khz = v->phyclk_per_state[v->voltage_level];
+		switch (v->voltage_level) {
+		case 0:
+			context->bw_ctx.bw.dcn.clk.max_supported_dppclk_khz =
+					(int)(dc->dcn_soc->max_dppclk_vmin0p65 * 1000);
+			break;
+		case 1:
+			context->bw_ctx.bw.dcn.clk.max_supported_dppclk_khz =
+					(int)(dc->dcn_soc->max_dppclk_vmid0p72 * 1000);
+			break;
+		case 2:
+			context->bw_ctx.bw.dcn.clk.max_supported_dppclk_khz =
+					(int)(dc->dcn_soc->max_dppclk_vnom0p8 * 1000);
+			break;
+		default:
+			context->bw_ctx.bw.dcn.clk.max_supported_dppclk_khz =
+					(int)(dc->dcn_soc->max_dppclk_vmax0p9 * 1000);
+			break;
+		}
+
+		BW_VAL_TRACE_END_WATERMARKS();
+
+		for (i = 0, input_idx = 0; i < pool->pipe_count; i++) {
+			struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
+
+			/* skip inactive pipe */
+			if (!pipe->stream)
+				continue;
+			/* skip all but first of split pipes */
+			if (pipe->top_pipe && pipe->top_pipe->plane_state == pipe->plane_state)
+				continue;
+
+			pipe->pipe_dlg_param.vupdate_width = v->v_update_width_pix[input_idx];
+			pipe->pipe_dlg_param.vupdate_offset = v->v_update_offset_pix[input_idx];
+			pipe->pipe_dlg_param.vready_offset = v->v_ready_offset_pix[input_idx];
+			pipe->pipe_dlg_param.vstartup_start = v->v_startup[input_idx];
+
+			pipe->pipe_dlg_param.htotal = pipe->stream->timing.h_total;
+			pipe->pipe_dlg_param.vtotal = pipe->stream->timing.v_total;
+			vesa_sync_start = pipe->stream->timing.v_addressable +
+						pipe->stream->timing.v_border_bottom +
+						pipe->stream->timing.v_front_porch;
+
+			asic_blank_end = (pipe->stream->timing.v_total -
+						vesa_sync_start -
+						pipe->stream->timing.v_border_top)
+			* (pipe->stream->timing.flags.INTERLACE ? 1 : 0);
+
+			asic_blank_start = asic_blank_end +
+						(pipe->stream->timing.v_border_top +
+						pipe->stream->timing.v_addressable +
+						pipe->stream->timing.v_border_bottom)
+			* (pipe->stream->timing.flags.INTERLACE ? 1 : 0);
+
+			pipe->pipe_dlg_param.vblank_start = asic_blank_start;
+			pipe->pipe_dlg_param.vblank_end = asic_blank_end;
+
+			if (pipe->plane_state) {
+				struct pipe_ctx *hsplit_pipe = pipe->bottom_pipe;
+
+				pipe->plane_state->update_flags.bits.full_update = 1;
+
+				if (v->dpp_per_plane[input_idx] == 2 ||
+					((pipe->stream->view_format ==
+					  VIEW_3D_FORMAT_SIDE_BY_SIDE ||
+					  pipe->stream->view_format ==
+					  VIEW_3D_FORMAT_TOP_AND_BOTTOM) &&
+					(pipe->stream->timing.timing_3d_format ==
+					 TIMING_3D_FORMAT_TOP_AND_BOTTOM ||
+					 pipe->stream->timing.timing_3d_format ==
+					 TIMING_3D_FORMAT_SIDE_BY_SIDE))) {
+					if (hsplit_pipe && hsplit_pipe->plane_state == pipe->plane_state) {
+						/* update previously split pipe */
+						hsplit_pipe->pipe_dlg_param.vupdate_width = v->v_update_width_pix[input_idx];
+						hsplit_pipe->pipe_dlg_param.vupdate_offset = v->v_update_offset_pix[input_idx];
+						hsplit_pipe->pipe_dlg_param.vready_offset = v->v_ready_offset_pix[input_idx];
+						hsplit_pipe->pipe_dlg_param.vstartup_start = v->v_startup[input_idx];
+
+						hsplit_pipe->pipe_dlg_param.htotal = pipe->stream->timing.h_total;
+						hsplit_pipe->pipe_dlg_param.vtotal = pipe->stream->timing.v_total;
+						hsplit_pipe->pipe_dlg_param.vblank_start = pipe->pipe_dlg_param.vblank_start;
+						hsplit_pipe->pipe_dlg_param.vblank_end = pipe->pipe_dlg_param.vblank_end;
+					} else {
+						/* pipe not split previously needs split */
+						hsplit_pipe = resource_find_free_secondary_pipe_legacy(&context->res_ctx, pool, pipe);
+						ASSERT(hsplit_pipe);
+						split_stream_across_pipes(&context->res_ctx, pool, pipe, hsplit_pipe);
+					}
+
+					dcn_bw_calc_rq_dlg_ttu(dc, v, hsplit_pipe, input_idx);
+				} else if (hsplit_pipe && hsplit_pipe->plane_state == pipe->plane_state) {
+					/* merge previously split pipe */
+					pipe->bottom_pipe = hsplit_pipe->bottom_pipe;
+					if (hsplit_pipe->bottom_pipe)
+						hsplit_pipe->bottom_pipe->top_pipe = pipe;
+					hsplit_pipe->plane_state = NULL;
+					hsplit_pipe->stream = NULL;
+					hsplit_pipe->top_pipe = NULL;
+					hsplit_pipe->bottom_pipe = NULL;
+					/* Clear plane_res and stream_res */
+					memset(&hsplit_pipe->plane_res, 0, sizeof(hsplit_pipe->plane_res));
+					memset(&hsplit_pipe->stream_res, 0, sizeof(hsplit_pipe->stream_res));
+					resource_build_scaling_params(pipe);
+				}
+				/* for now important to do this after pipe split for building e2e params */
+				dcn_bw_calc_rq_dlg_ttu(dc, v, pipe, input_idx);
+			}
+
+			input_idx++;
+		}
+	} else if (v->voltage_level == number_of_states_plus_one) {
+		BW_VAL_TRACE_SKIP(fail);
+	} else if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING) {
+		BW_VAL_TRACE_SKIP(fast);
+	}
+
+	if (v->voltage_level == 0) {
+		context->bw_ctx.dml.soc.sr_enter_plus_exit_time_us =
+				dc->dcn_soc->sr_enter_plus_exit_time;
+		context->bw_ctx.dml.soc.sr_exit_time_us = dc->dcn_soc->sr_exit_time;
+	}
+
+	/*
+	 * BW limit is set to prevent display from impacting other system functions
+	 */
+
+	bw_limit = dc->dcn_soc->percent_disp_bw_limit * v->fabric_and_dram_bandwidth_vmax0p9;
+	bw_limit_pass = (v->total_data_read_bandwidth / 1000.0) < bw_limit;
+
+	PERFORMANCE_TRACE_END();
+	BW_VAL_TRACE_FINISH();
+
+	if (bw_limit_pass && v->voltage_level <= get_highest_allowed_voltage_level(dc->config.is_vmin_only_asic))
+		return true;
+	else
+		return false;
+}
+
+void dcn_bw_update_from_pplib_fclks(
+	struct dc *dc,
+	struct dm_pp_clock_levels_with_voltage *fclks)
+{
+	unsigned vmin0p65_idx, vmid0p72_idx, vnom0p8_idx, vmax0p9_idx;
+
+	ASSERT(fclks->num_levels);
+
+	vmin0p65_idx = 0;
+	vmid0p72_idx = fclks->num_levels > 2 ? fclks->num_levels - 3 : 0;
+	vnom0p8_idx = fclks->num_levels > 1 ? fclks->num_levels - 2 : 0;
+	vmax0p9_idx = fclks->num_levels > 0 ? fclks->num_levels - 1 : 0;
+
+	dc->dcn_soc->fabric_and_dram_bandwidth_vmin0p65 =
+		32 * (fclks->data[vmin0p65_idx].clocks_in_khz / 1000.0) / 1000.0;
+	dc->dcn_soc->fabric_and_dram_bandwidth_vmid0p72 =
+		dc->dcn_soc->number_of_channels *
+		(fclks->data[vmid0p72_idx].clocks_in_khz / 1000.0)
+		* ddr4_dram_factor_single_Channel / 1000.0;
+	dc->dcn_soc->fabric_and_dram_bandwidth_vnom0p8 =
+		dc->dcn_soc->number_of_channels *
+		(fclks->data[vnom0p8_idx].clocks_in_khz / 1000.0)
+		* ddr4_dram_factor_single_Channel / 1000.0;
+	dc->dcn_soc->fabric_and_dram_bandwidth_vmax0p9 =
+		dc->dcn_soc->number_of_channels *
+		(fclks->data[vmax0p9_idx].clocks_in_khz / 1000.0)
+		* ddr4_dram_factor_single_Channel / 1000.0;
+}
+
+void dcn_bw_update_from_pplib_dcfclks(
+	struct dc *dc,
+	struct dm_pp_clock_levels_with_voltage *dcfclks)
+{
+	if (dcfclks->num_levels >= 3) {
+		dc->dcn_soc->dcfclkv_min0p65 = dcfclks->data[0].clocks_in_khz / 1000.0;
+		dc->dcn_soc->dcfclkv_mid0p72 = dcfclks->data[dcfclks->num_levels - 3].clocks_in_khz / 1000.0;
+		dc->dcn_soc->dcfclkv_nom0p8 = dcfclks->data[dcfclks->num_levels - 2].clocks_in_khz / 1000.0;
+		dc->dcn_soc->dcfclkv_max0p9 = dcfclks->data[dcfclks->num_levels - 1].clocks_in_khz / 1000.0;
+	}
+}
+
+void dcn_get_soc_clks(
+	struct dc *dc,
+	int *min_fclk_khz,
+	int *min_dcfclk_khz,
+	int *socclk_khz)
+{
+	*min_fclk_khz = dc->dcn_soc->fabric_and_dram_bandwidth_vmin0p65 * 1000000 / 32;
+	*min_dcfclk_khz = dc->dcn_soc->dcfclkv_min0p65 * 1000;
+	*socclk_khz = dc->dcn_soc->socclk * 1000;
+}
+
+void dcn_bw_notify_pplib_of_wm_ranges(
+	struct dc *dc,
+	int min_fclk_khz,
+	int min_dcfclk_khz,
+	int socclk_khz)
+{
+	struct pp_smu_funcs_rv *pp = NULL;
+	struct pp_smu_wm_range_sets ranges = {0};
+	const int overdrive = 5000000; /* 5 GHz to cover Overdrive */
+
+	if (dc->res_pool->pp_smu)
+		pp = &dc->res_pool->pp_smu->rv_funcs;
+	if (!pp || !pp->set_wm_ranges)
+		return;
+
+	/* Now notify PPLib/SMU about which Watermarks sets they should select
+	 * depending on DPM state they are in. And update BW MGR GFX Engine and
+	 * Memory clock member variables for Watermarks calculations for each
+	 * Watermark Set. Only one watermark set for dcn1 due to hw bug DEGVIDCN10-254.
+	 */
+	/* SOCCLK does not affect anytihng but writeback for DCN so for now we dont
+	 * care what the value is, hence min to overdrive level
+	 */
+	ranges.num_reader_wm_sets = WM_SET_COUNT;
+	ranges.num_writer_wm_sets = WM_SET_COUNT;
+	ranges.reader_wm_sets[0].wm_inst = WM_A;
+	ranges.reader_wm_sets[0].min_drain_clk_mhz = min_dcfclk_khz / 1000;
+	ranges.reader_wm_sets[0].max_drain_clk_mhz = overdrive / 1000;
+	ranges.reader_wm_sets[0].min_fill_clk_mhz = min_fclk_khz / 1000;
+	ranges.reader_wm_sets[0].max_fill_clk_mhz = overdrive / 1000;
+	ranges.writer_wm_sets[0].wm_inst = WM_A;
+	ranges.writer_wm_sets[0].min_fill_clk_mhz = socclk_khz / 1000;
+	ranges.writer_wm_sets[0].max_fill_clk_mhz = overdrive / 1000;
+	ranges.writer_wm_sets[0].min_drain_clk_mhz = min_fclk_khz / 1000;
+	ranges.writer_wm_sets[0].max_drain_clk_mhz = overdrive / 1000;
+
+	if (dc->debug.pplib_wm_report_mode == WM_REPORT_OVERRIDE) {
+		ranges.reader_wm_sets[0].wm_inst = WM_A;
+		ranges.reader_wm_sets[0].min_drain_clk_mhz = 300;
+		ranges.reader_wm_sets[0].max_drain_clk_mhz = 5000;
+		ranges.reader_wm_sets[0].min_fill_clk_mhz = 800;
+		ranges.reader_wm_sets[0].max_fill_clk_mhz = 5000;
+		ranges.writer_wm_sets[0].wm_inst = WM_A;
+		ranges.writer_wm_sets[0].min_fill_clk_mhz = 200;
+		ranges.writer_wm_sets[0].max_fill_clk_mhz = 5000;
+		ranges.writer_wm_sets[0].min_drain_clk_mhz = 800;
+		ranges.writer_wm_sets[0].max_drain_clk_mhz = 5000;
+	}
+
+	ranges.reader_wm_sets[1] = ranges.writer_wm_sets[0];
+	ranges.reader_wm_sets[1].wm_inst = WM_B;
+
+	ranges.reader_wm_sets[2] = ranges.writer_wm_sets[0];
+	ranges.reader_wm_sets[2].wm_inst = WM_C;
+
+	ranges.reader_wm_sets[3] = ranges.writer_wm_sets[0];
+	ranges.reader_wm_sets[3].wm_inst = WM_D;
+
+	/* Notify PP Lib/SMU which Watermarks to use for which clock ranges */
+	pp->set_wm_ranges(&pp->pp_smu, &ranges);
+}
+
+void dcn_bw_sync_calcs_and_dml(struct dc *dc)
+{
+	DC_LOG_BANDWIDTH_CALCS("sr_exit_time: %f ns\n"
+			"sr_enter_plus_exit_time: %f ns\n"
+			"urgent_latency: %f ns\n"
+			"write_back_latency: %f ns\n"
+			"percent_of_ideal_drambw_received_after_urg_latency: %f %%\n"
+			"max_request_size: %d bytes\n"
+			"dcfclkv_max0p9: %f kHz\n"
+			"dcfclkv_nom0p8: %f kHz\n"
+			"dcfclkv_mid0p72: %f kHz\n"
+			"dcfclkv_min0p65: %f kHz\n"
+			"max_dispclk_vmax0p9: %f kHz\n"
+			"max_dispclk_vnom0p8: %f kHz\n"
+			"max_dispclk_vmid0p72: %f kHz\n"
+			"max_dispclk_vmin0p65: %f kHz\n"
+			"max_dppclk_vmax0p9: %f kHz\n"
+			"max_dppclk_vnom0p8: %f kHz\n"
+			"max_dppclk_vmid0p72: %f kHz\n"
+			"max_dppclk_vmin0p65: %f kHz\n"
+			"socclk: %f kHz\n"
+			"fabric_and_dram_bandwidth_vmax0p9: %f MB/s\n"
+			"fabric_and_dram_bandwidth_vnom0p8: %f MB/s\n"
+			"fabric_and_dram_bandwidth_vmid0p72: %f MB/s\n"
+			"fabric_and_dram_bandwidth_vmin0p65: %f MB/s\n"
+			"phyclkv_max0p9: %f kHz\n"
+			"phyclkv_nom0p8: %f kHz\n"
+			"phyclkv_mid0p72: %f kHz\n"
+			"phyclkv_min0p65: %f kHz\n"
+			"downspreading: %f %%\n"
+			"round_trip_ping_latency_cycles: %d DCFCLK Cycles\n"
+			"urgent_out_of_order_return_per_channel: %d Bytes\n"
+			"number_of_channels: %d\n"
+			"vmm_page_size: %d Bytes\n"
+			"dram_clock_change_latency: %f ns\n"
+			"return_bus_width: %d Bytes\n",
+			dc->dcn_soc->sr_exit_time * 1000,
+			dc->dcn_soc->sr_enter_plus_exit_time * 1000,
+			dc->dcn_soc->urgent_latency * 1000,
+			dc->dcn_soc->write_back_latency * 1000,
+			dc->dcn_soc->percent_of_ideal_drambw_received_after_urg_latency,
+			dc->dcn_soc->max_request_size,
+			dc->dcn_soc->dcfclkv_max0p9 * 1000,
+			dc->dcn_soc->dcfclkv_nom0p8 * 1000,
+			dc->dcn_soc->dcfclkv_mid0p72 * 1000,
+			dc->dcn_soc->dcfclkv_min0p65 * 1000,
+			dc->dcn_soc->max_dispclk_vmax0p9 * 1000,
+			dc->dcn_soc->max_dispclk_vnom0p8 * 1000,
+			dc->dcn_soc->max_dispclk_vmid0p72 * 1000,
+			dc->dcn_soc->max_dispclk_vmin0p65 * 1000,
+			dc->dcn_soc->max_dppclk_vmax0p9 * 1000,
+			dc->dcn_soc->max_dppclk_vnom0p8 * 1000,
+			dc->dcn_soc->max_dppclk_vmid0p72 * 1000,
+			dc->dcn_soc->max_dppclk_vmin0p65 * 1000,
+			dc->dcn_soc->socclk * 1000,
+			dc->dcn_soc->fabric_and_dram_bandwidth_vmax0p9 * 1000,
+			dc->dcn_soc->fabric_and_dram_bandwidth_vnom0p8 * 1000,
+			dc->dcn_soc->fabric_and_dram_bandwidth_vmid0p72 * 1000,
+			dc->dcn_soc->fabric_and_dram_bandwidth_vmin0p65 * 1000,
+			dc->dcn_soc->phyclkv_max0p9 * 1000,
+			dc->dcn_soc->phyclkv_nom0p8 * 1000,
+			dc->dcn_soc->phyclkv_mid0p72 * 1000,
+			dc->dcn_soc->phyclkv_min0p65 * 1000,
+			dc->dcn_soc->downspreading * 100,
+			dc->dcn_soc->round_trip_ping_latency_cycles,
+			dc->dcn_soc->urgent_out_of_order_return_per_channel,
+			dc->dcn_soc->number_of_channels,
+			dc->dcn_soc->vmm_page_size,
+			dc->dcn_soc->dram_clock_change_latency * 1000,
+			dc->dcn_soc->return_bus_width);
+	DC_LOG_BANDWIDTH_CALCS("rob_buffer_size_in_kbyte: %f\n"
+			"det_buffer_size_in_kbyte: %f\n"
+			"dpp_output_buffer_pixels: %f\n"
+			"opp_output_buffer_lines: %f\n"
+			"pixel_chunk_size_in_kbyte: %f\n"
+			"pte_enable: %d\n"
+			"pte_chunk_size: %d kbytes\n"
+			"meta_chunk_size: %d kbytes\n"
+			"writeback_chunk_size: %d kbytes\n"
+			"odm_capability: %d\n"
+			"dsc_capability: %d\n"
+			"line_buffer_size: %d bits\n"
+			"max_line_buffer_lines: %d\n"
+			"is_line_buffer_bpp_fixed: %d\n"
+			"line_buffer_fixed_bpp: %d\n"
+			"writeback_luma_buffer_size: %d kbytes\n"
+			"writeback_chroma_buffer_size: %d kbytes\n"
+			"max_num_dpp: %d\n"
+			"max_num_writeback: %d\n"
+			"max_dchub_topscl_throughput: %d pixels/dppclk\n"
+			"max_pscl_tolb_throughput: %d pixels/dppclk\n"
+			"max_lb_tovscl_throughput: %d pixels/dppclk\n"
+			"max_vscl_tohscl_throughput: %d pixels/dppclk\n"
+			"max_hscl_ratio: %f\n"
+			"max_vscl_ratio: %f\n"
+			"max_hscl_taps: %d\n"
+			"max_vscl_taps: %d\n"
+			"pte_buffer_size_in_requests: %d\n"
+			"dispclk_ramping_margin: %f %%\n"
+			"under_scan_factor: %f %%\n"
+			"max_inter_dcn_tile_repeaters: %d\n"
+			"can_vstartup_lines_exceed_vsync_plus_back_porch_lines_minus_one: %d\n"
+			"bug_forcing_luma_and_chroma_request_to_same_size_fixed: %d\n"
+			"dcfclk_cstate_latency: %d\n",
+			dc->dcn_ip->rob_buffer_size_in_kbyte,
+			dc->dcn_ip->det_buffer_size_in_kbyte,
+			dc->dcn_ip->dpp_output_buffer_pixels,
+			dc->dcn_ip->opp_output_buffer_lines,
+			dc->dcn_ip->pixel_chunk_size_in_kbyte,
+			dc->dcn_ip->pte_enable,
+			dc->dcn_ip->pte_chunk_size,
+			dc->dcn_ip->meta_chunk_size,
+			dc->dcn_ip->writeback_chunk_size,
+			dc->dcn_ip->odm_capability,
+			dc->dcn_ip->dsc_capability,
+			dc->dcn_ip->line_buffer_size,
+			dc->dcn_ip->max_line_buffer_lines,
+			dc->dcn_ip->is_line_buffer_bpp_fixed,
+			dc->dcn_ip->line_buffer_fixed_bpp,
+			dc->dcn_ip->writeback_luma_buffer_size,
+			dc->dcn_ip->writeback_chroma_buffer_size,
+			dc->dcn_ip->max_num_dpp,
+			dc->dcn_ip->max_num_writeback,
+			dc->dcn_ip->max_dchub_topscl_throughput,
+			dc->dcn_ip->max_pscl_tolb_throughput,
+			dc->dcn_ip->max_lb_tovscl_throughput,
+			dc->dcn_ip->max_vscl_tohscl_throughput,
+			dc->dcn_ip->max_hscl_ratio,
+			dc->dcn_ip->max_vscl_ratio,
+			dc->dcn_ip->max_hscl_taps,
+			dc->dcn_ip->max_vscl_taps,
+			dc->dcn_ip->pte_buffer_size_in_requests,
+			dc->dcn_ip->dispclk_ramping_margin,
+			dc->dcn_ip->under_scan_factor * 100,
+			dc->dcn_ip->max_inter_dcn_tile_repeaters,
+			dc->dcn_ip->can_vstartup_lines_exceed_vsync_plus_back_porch_lines_minus_one,
+			dc->dcn_ip->bug_forcing_luma_and_chroma_request_to_same_size_fixed,
+			dc->dcn_ip->dcfclk_cstate_latency);
+
+	dc->dml.soc.sr_exit_time_us = dc->dcn_soc->sr_exit_time;
+	dc->dml.soc.sr_enter_plus_exit_time_us = dc->dcn_soc->sr_enter_plus_exit_time;
+	dc->dml.soc.urgent_latency_us = dc->dcn_soc->urgent_latency;
+	dc->dml.soc.writeback_latency_us = dc->dcn_soc->write_back_latency;
+	dc->dml.soc.ideal_dram_bw_after_urgent_percent =
+			dc->dcn_soc->percent_of_ideal_drambw_received_after_urg_latency;
+	dc->dml.soc.max_request_size_bytes = dc->dcn_soc->max_request_size;
+	dc->dml.soc.downspread_percent = dc->dcn_soc->downspreading;
+	dc->dml.soc.round_trip_ping_latency_dcfclk_cycles =
+			dc->dcn_soc->round_trip_ping_latency_cycles;
+	dc->dml.soc.urgent_out_of_order_return_per_channel_bytes =
+			dc->dcn_soc->urgent_out_of_order_return_per_channel;
+	dc->dml.soc.num_chans = dc->dcn_soc->number_of_channels;
+	dc->dml.soc.vmm_page_size_bytes = dc->dcn_soc->vmm_page_size;
+	dc->dml.soc.dram_clock_change_latency_us = dc->dcn_soc->dram_clock_change_latency;
+	dc->dml.soc.return_bus_width_bytes = dc->dcn_soc->return_bus_width;
+
+	dc->dml.ip.rob_buffer_size_kbytes = dc->dcn_ip->rob_buffer_size_in_kbyte;
+	dc->dml.ip.det_buffer_size_kbytes = dc->dcn_ip->det_buffer_size_in_kbyte;
+	dc->dml.ip.dpp_output_buffer_pixels = dc->dcn_ip->dpp_output_buffer_pixels;
+	dc->dml.ip.opp_output_buffer_lines = dc->dcn_ip->opp_output_buffer_lines;
+	dc->dml.ip.pixel_chunk_size_kbytes = dc->dcn_ip->pixel_chunk_size_in_kbyte;
+	dc->dml.ip.pte_enable = dc->dcn_ip->pte_enable == dcn_bw_yes;
+	dc->dml.ip.pte_chunk_size_kbytes = dc->dcn_ip->pte_chunk_size;
+	dc->dml.ip.meta_chunk_size_kbytes = dc->dcn_ip->meta_chunk_size;
+	dc->dml.ip.writeback_chunk_size_kbytes = dc->dcn_ip->writeback_chunk_size;
+	dc->dml.ip.line_buffer_size_bits = dc->dcn_ip->line_buffer_size;
+	dc->dml.ip.max_line_buffer_lines = dc->dcn_ip->max_line_buffer_lines;
+	dc->dml.ip.IsLineBufferBppFixed = dc->dcn_ip->is_line_buffer_bpp_fixed == dcn_bw_yes;
+	dc->dml.ip.LineBufferFixedBpp = dc->dcn_ip->line_buffer_fixed_bpp;
+	dc->dml.ip.writeback_luma_buffer_size_kbytes = dc->dcn_ip->writeback_luma_buffer_size;
+	dc->dml.ip.writeback_chroma_buffer_size_kbytes = dc->dcn_ip->writeback_chroma_buffer_size;
+	dc->dml.ip.max_num_dpp = dc->dcn_ip->max_num_dpp;
+	dc->dml.ip.max_num_wb = dc->dcn_ip->max_num_writeback;
+	dc->dml.ip.max_dchub_pscl_bw_pix_per_clk = dc->dcn_ip->max_dchub_topscl_throughput;
+	dc->dml.ip.max_pscl_lb_bw_pix_per_clk = dc->dcn_ip->max_pscl_tolb_throughput;
+	dc->dml.ip.max_lb_vscl_bw_pix_per_clk = dc->dcn_ip->max_lb_tovscl_throughput;
+	dc->dml.ip.max_vscl_hscl_bw_pix_per_clk = dc->dcn_ip->max_vscl_tohscl_throughput;
+	dc->dml.ip.max_hscl_ratio = dc->dcn_ip->max_hscl_ratio;
+	dc->dml.ip.max_vscl_ratio = dc->dcn_ip->max_vscl_ratio;
+	dc->dml.ip.max_hscl_taps = dc->dcn_ip->max_hscl_taps;
+	dc->dml.ip.max_vscl_taps = dc->dcn_ip->max_vscl_taps;
+	/*pte_buffer_size_in_requests missing in dml*/
+	dc->dml.ip.dispclk_ramp_margin_percent = dc->dcn_ip->dispclk_ramping_margin;
+	dc->dml.ip.underscan_factor = dc->dcn_ip->under_scan_factor;
+	dc->dml.ip.max_inter_dcn_tile_repeaters = dc->dcn_ip->max_inter_dcn_tile_repeaters;
+	dc->dml.ip.can_vstartup_lines_exceed_vsync_plus_back_porch_lines_minus_one =
+		dc->dcn_ip->can_vstartup_lines_exceed_vsync_plus_back_porch_lines_minus_one == dcn_bw_yes;
+	dc->dml.ip.bug_forcing_LC_req_same_size_fixed =
+		dc->dcn_ip->bug_forcing_luma_and_chroma_request_to_same_size_fixed == dcn_bw_yes;
+	dc->dml.ip.dcfclk_cstate_latency = dc->dcn_ip->dcfclk_cstate_latency;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dc_features.h b/drivers/gpu/drm/amd/display/dc/dml/dc_features.h
new file mode 100644
index 000000000000..6e669a2c5b2d
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dml/dc_features.h
@@ -0,0 +1,566 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+#ifndef __DC_FEATURES_H__
+#define __DC_FEATURES_H__
+
+// local features
+#define DC__PRESENT 1
+#define DC__PRESENT__1 1
+#define DC__NUM_DPP 4
+
+/**
+ * @DC__VOLTAGE_STATES:
+ *
+ * Define the maximum amount of states supported by the ASIC. Every ASIC has a
+ * specific number of states; this macro defines the maximum number of states.
+ */
+#define DC__VOLTAGE_STATES 40
+#define DC__NUM_DPP__4 1
+#define DC__NUM_DPP__0_PRESENT 1
+#define DC__NUM_DPP__1_PRESENT 1
+#define DC__NUM_DPP__2_PRESENT 1
+#define DC__NUM_DPP__3_PRESENT 1
+#define DC__NUM_DPP__MAX 8
+#define DC__NUM_DPP__MAX__8 1
+#define DC__PIPE_10BIT 0
+#define DC__PIPE_10BIT__0 1
+#define DC__PIPE_10BIT__MAX 1
+#define DC__PIPE_10BIT__MAX__1 1
+#define DC__NUM_OPP 4
+#define DC__NUM_OPP__4 1
+#define DC__NUM_OPP__0_PRESENT 1
+#define DC__NUM_OPP__1_PRESENT 1
+#define DC__NUM_OPP__2_PRESENT 1
+#define DC__NUM_OPP__3_PRESENT 1
+#define DC__NUM_OPP__MAX 6
+#define DC__NUM_OPP__MAX__6 1
+#define DC__NUM_DSC 0
+#define DC__NUM_DSC__0 1
+#define DC__NUM_DSC__MAX 6
+#define DC__NUM_DSC__MAX__6 1
+#define DC__NUM_ABM 1
+#define DC__NUM_ABM__1 1
+#define DC__NUM_ABM__0_PRESENT 1
+#define DC__NUM_ABM__MAX 2
+#define DC__NUM_ABM__MAX__2 1
+#define DC__ODM_PRESENT 0
+#define DC__ODM_PRESENT__0 1
+#define DC__NUM_OTG 4
+#define DC__NUM_OTG__4 1
+#define DC__NUM_OTG__0_PRESENT 1
+#define DC__NUM_OTG__1_PRESENT 1
+#define DC__NUM_OTG__2_PRESENT 1
+#define DC__NUM_OTG__3_PRESENT 1
+#define DC__NUM_OTG__MAX 6
+#define DC__NUM_OTG__MAX__6 1
+#define DC__NUM_DWB 2
+#define DC__NUM_DWB__2 1
+#define DC__NUM_DWB__0_PRESENT 1
+#define DC__NUM_DWB__1_PRESENT 1
+#define DC__NUM_DWB__MAX 2
+#define DC__NUM_DWB__MAX__2 1
+#define DC__NUM_DIG 4
+#define DC__NUM_DIG__4 1
+#define DC__NUM_DIG__0_PRESENT 1
+#define DC__NUM_DIG__1_PRESENT 1
+#define DC__NUM_DIG__2_PRESENT 1
+#define DC__NUM_DIG__3_PRESENT 1
+#define DC__NUM_DIG__MAX 6
+#define DC__NUM_DIG__MAX__6 1
+#define DC__NUM_AUX 4
+#define DC__NUM_AUX__4 1
+#define DC__NUM_AUX__0_PRESENT 1
+#define DC__NUM_AUX__1_PRESENT 1
+#define DC__NUM_AUX__2_PRESENT 1
+#define DC__NUM_AUX__3_PRESENT 1
+#define DC__NUM_AUX__MAX 6
+#define DC__NUM_AUX__MAX__6 1
+#define DC__NUM_AUDIO_STREAMS 4
+#define DC__NUM_AUDIO_STREAMS__4 1
+#define DC__NUM_AUDIO_STREAMS__0_PRESENT 1
+#define DC__NUM_AUDIO_STREAMS__1_PRESENT 1
+#define DC__NUM_AUDIO_STREAMS__2_PRESENT 1
+#define DC__NUM_AUDIO_STREAMS__3_PRESENT 1
+#define DC__NUM_AUDIO_STREAMS__MAX 8
+#define DC__NUM_AUDIO_STREAMS__MAX__8 1
+#define DC__NUM_AUDIO_ENDPOINTS 6
+#define DC__NUM_AUDIO_ENDPOINTS__6 1
+#define DC__NUM_AUDIO_ENDPOINTS__0_PRESENT 1
+#define DC__NUM_AUDIO_ENDPOINTS__1_PRESENT 1
+#define DC__NUM_AUDIO_ENDPOINTS__2_PRESENT 1
+#define DC__NUM_AUDIO_ENDPOINTS__3_PRESENT 1
+#define DC__NUM_AUDIO_ENDPOINTS__4_PRESENT 1
+#define DC__NUM_AUDIO_ENDPOINTS__5_PRESENT 1
+#define DC__NUM_AUDIO_ENDPOINTS__MAX 8
+#define DC__NUM_AUDIO_ENDPOINTS__MAX__8 1
+#define DC__NUM_AUDIO_INPUT_STREAMS 0
+#define DC__NUM_AUDIO_INPUT_STREAMS__0 1
+#define DC__NUM_AUDIO_INPUT_STREAMS__MAX 8
+#define DC__NUM_AUDIO_INPUT_STREAMS__MAX__8 1
+#define DC__NUM_AUDIO_INPUT_ENDPOINTS 0
+#define DC__NUM_AUDIO_INPUT_ENDPOINTS__0 1
+#define DC__NUM_AUDIO_INPUT_ENDPOINTS__MAX 8
+#define DC__NUM_AUDIO_INPUT_ENDPOINTS__MAX__8 1
+#define DC__NUM_CURSOR 1
+#define DC__NUM_CURSOR__1 1
+#define DC__NUM_CURSOR__0_PRESENT 1
+#define DC__NUM_CURSOR__MAX 2
+#define DC__NUM_CURSOR__MAX__2 1
+#define DC__DIGITAL_BYPASS_PRESENT 0
+#define DC__DIGITAL_BYPASS_PRESENT__0 1
+#define DC__HCID_HWMAJVER 1
+#define DC__HCID_HWMAJVER__1 1
+#define DC__HCID_HWMINVER 0
+#define DC__HCID_HWMINVER__0 1
+#define DC__HCID_HWREV 0
+#define DC__HCID_HWREV__0 1
+#define DC__ROMSTRAP_PRESENT 0
+#define DC__ROMSTRAP_PRESENT__0 1
+#define DC__NUM_RBBMIF_DECODES 30
+#define DC__NUM_RBBMIF_DECODES__30 1
+#define DC__NUM_DBG_REGS 36
+#define DC__NUM_DBG_REGS__36 1
+#define DC__NUM_PIPES_UNDERLAY 0
+#define DC__NUM_PIPES_UNDERLAY__0 1
+#define DC__NUM_PIPES_UNDERLAY__MAX 2
+#define DC__NUM_PIPES_UNDERLAY__MAX__2 1
+#define DC__NUM_VCE_ENGINE 1
+#define DC__NUM_VCE_ENGINE__1 1
+#define DC__NUM_VCE_ENGINE__0_PRESENT 1
+#define DC__NUM_VCE_ENGINE__MAX 2
+#define DC__NUM_VCE_ENGINE__MAX__2 1
+#define DC__OTG_EXTERNAL_SYNC_PRESENT 0
+#define DC__OTG_EXTERNAL_SYNC_PRESENT__0 1
+#define DC__OTG_CRC_PRESENT 1
+#define DC__OTG_CRC_PRESENT__1 1
+#define DC__VIP_PRESENT 0
+#define DC__VIP_PRESENT__0 1
+#define DC__DTMTEST_PRESENT 0
+#define DC__DTMTEST_PRESENT__0 1
+#define DC__POWER_GATE_PRESENT 1
+#define DC__POWER_GATE_PRESENT__1 1
+#define DC__MEM_PG 1
+#define DC__MEM_PG__1 1
+#define DC__FMT_SRC_SEL_PRESENT 0
+#define DC__FMT_SRC_SEL_PRESENT__0 1
+#define DC__DIG_FEATURES__HDMI_PRESENT 1
+#define DC__DIG_FEATURES__HDMI_PRESENT__1 1
+#define DC__DIG_FEATURES__DP_PRESENT 1
+#define DC__DIG_FEATURES__DP_PRESENT__1 1
+#define DC__DIG_FEATURES__DP_MST_PRESENT 1
+#define DC__DIG_FEATURES__DP_MST_PRESENT__1 1
+#define DC__DIG_LP_FEATURES__HDMI_PRESENT 0
+#define DC__DIG_LP_FEATURES__HDMI_PRESENT__0 1
+#define DC__DIG_LP_FEATURES__DP_PRESENT 1
+#define DC__DIG_LP_FEATURES__DP_PRESENT__1 1
+#define DC__DIG_LP_FEATURES__DP_MST_PRESENT 0
+#define DC__DIG_LP_FEATURES__DP_MST_PRESENT__0 1
+#define DC__DIG_RESYNC_FIFO_SIZE 14
+#define DC__DIG_RESYNC_FIFO_SIZE__14 1
+#define DC__DIG_RESYNC_FIFO_SIZE__0_PRESENT 1
+#define DC__DIG_RESYNC_FIFO_SIZE__1_PRESENT 1
+#define DC__DIG_RESYNC_FIFO_SIZE__2_PRESENT 1
+#define DC__DIG_RESYNC_FIFO_SIZE__3_PRESENT 1
+#define DC__DIG_RESYNC_FIFO_SIZE__4_PRESENT 1
+#define DC__DIG_RESYNC_FIFO_SIZE__5_PRESENT 1
+#define DC__DIG_RESYNC_FIFO_SIZE__6_PRESENT 1
+#define DC__DIG_RESYNC_FIFO_SIZE__7_PRESENT 1
+#define DC__DIG_RESYNC_FIFO_SIZE__8_PRESENT 1
+#define DC__DIG_RESYNC_FIFO_SIZE__9_PRESENT 1
+#define DC__DIG_RESYNC_FIFO_SIZE__10_PRESENT 1
+#define DC__DIG_RESYNC_FIFO_SIZE__11_PRESENT 1
+#define DC__DIG_RESYNC_FIFO_SIZE__12_PRESENT 1
+#define DC__DIG_RESYNC_FIFO_SIZE__13_PRESENT 1
+#define DC__DIG_RESYNC_FIFO_SIZE__MAX 16
+#define DC__DIG_RESYNC_FIFO_SIZE__MAX__16 1
+#define DC__DAC_RESYNC_FIFO_SIZE 12
+#define DC__DAC_RESYNC_FIFO_SIZE__12 1
+#define DC__DAC_RESYNC_FIFO_SIZE__0_PRESENT 1
+#define DC__DAC_RESYNC_FIFO_SIZE__1_PRESENT 1
+#define DC__DAC_RESYNC_FIFO_SIZE__2_PRESENT 1
+#define DC__DAC_RESYNC_FIFO_SIZE__3_PRESENT 1
+#define DC__DAC_RESYNC_FIFO_SIZE__4_PRESENT 1
+#define DC__DAC_RESYNC_FIFO_SIZE__5_PRESENT 1
+#define DC__DAC_RESYNC_FIFO_SIZE__6_PRESENT 1
+#define DC__DAC_RESYNC_FIFO_SIZE__7_PRESENT 1
+#define DC__DAC_RESYNC_FIFO_SIZE__8_PRESENT 1
+#define DC__DAC_RESYNC_FIFO_SIZE__9_PRESENT 1
+#define DC__DAC_RESYNC_FIFO_SIZE__10_PRESENT 1
+#define DC__DAC_RESYNC_FIFO_SIZE__11_PRESENT 1
+#define DC__DAC_RESYNC_FIFO_SIZE__MAX 16
+#define DC__DAC_RESYNC_FIFO_SIZE__MAX__16 1
+#define DC__DVO_RESYNC_FIFO_SIZE 12
+#define DC__DVO_RESYNC_FIFO_SIZE__12 1
+#define DC__DVO_RESYNC_FIFO_SIZE__0_PRESENT 1
+#define DC__DVO_RESYNC_FIFO_SIZE__1_PRESENT 1
+#define DC__DVO_RESYNC_FIFO_SIZE__2_PRESENT 1
+#define DC__DVO_RESYNC_FIFO_SIZE__3_PRESENT 1
+#define DC__DVO_RESYNC_FIFO_SIZE__4_PRESENT 1
+#define DC__DVO_RESYNC_FIFO_SIZE__5_PRESENT 1
+#define DC__DVO_RESYNC_FIFO_SIZE__6_PRESENT 1
+#define DC__DVO_RESYNC_FIFO_SIZE__7_PRESENT 1
+#define DC__DVO_RESYNC_FIFO_SIZE__8_PRESENT 1
+#define DC__DVO_RESYNC_FIFO_SIZE__9_PRESENT 1
+#define DC__DVO_RESYNC_FIFO_SIZE__10_PRESENT 1
+#define DC__DVO_RESYNC_FIFO_SIZE__11_PRESENT 1
+#define DC__DVO_RESYNC_FIFO_SIZE__MAX 16
+#define DC__DVO_RESYNC_FIFO_SIZE__MAX__16 1
+#define DC__MEM_CDC_PRESENT 1
+#define DC__MEM_CDC_PRESENT__1 1
+#define DC__NUM_HPD 4
+#define DC__NUM_HPD__4 1
+#define DC__NUM_HPD__0_PRESENT 1
+#define DC__NUM_HPD__1_PRESENT 1
+#define DC__NUM_HPD__2_PRESENT 1
+#define DC__NUM_HPD__3_PRESENT 1
+#define DC__NUM_HPD__MAX 6
+#define DC__NUM_HPD__MAX__6 1
+#define DC__NUM_DDC_PAIRS 4
+#define DC__NUM_DDC_PAIRS__4 1
+#define DC__NUM_DDC_PAIRS__0_PRESENT 1
+#define DC__NUM_DDC_PAIRS__1_PRESENT 1
+#define DC__NUM_DDC_PAIRS__2_PRESENT 1
+#define DC__NUM_DDC_PAIRS__3_PRESENT 1
+#define DC__NUM_DDC_PAIRS__MAX 6
+#define DC__NUM_DDC_PAIRS__MAX__6 1
+#define DC__NUM_AUDIO_PLL 0
+#define DC__NUM_AUDIO_PLL__0 1
+#define DC__NUM_AUDIO_PLL__MAX 2
+#define DC__NUM_AUDIO_PLL__MAX__2 1
+#define DC__NUM_PIXEL_PLL 1
+#define DC__NUM_PIXEL_PLL__1 1
+#define DC__NUM_PIXEL_PLL__0_PRESENT 1
+#define DC__NUM_PIXEL_PLL__MAX 4
+#define DC__NUM_PIXEL_PLL__MAX__4 1
+#define DC__NUM_CASCADED_PLL 0
+#define DC__NUM_CASCADED_PLL__0 1
+#define DC__NUM_CASCADED_PLL__MAX 3
+#define DC__NUM_CASCADED_PLL__MAX__3 1
+#define DC__PIXCLK_FROM_PHYPLL 1
+#define DC__PIXCLK_FROM_PHYPLL__1 1
+#define DC__NB_STUTTER_MODE_PRESENT 0
+#define DC__NB_STUTTER_MODE_PRESENT__0 1
+#define DC__I2S0_AND_SPDIF0_PRESENT 0
+#define DC__I2S0_AND_SPDIF0_PRESENT__0 1
+#define DC__I2S1_PRESENT 0
+#define DC__I2S1_PRESENT__0 1
+#define DC__SPDIF1_PRESENT 0
+#define DC__SPDIF1_PRESENT__0 1
+#define DC__DSI_PRESENT 0
+#define DC__DSI_PRESENT__0 1
+#define DC__DACA_PRESENT 0
+#define DC__DACA_PRESENT__0 1
+#define DC__DACB_PRESENT 0
+#define DC__DACB_PRESENT__0 1
+#define DC__NUM_PIPES 4
+#define DC__NUM_PIPES__4 1
+#define DC__NUM_PIPES__0_PRESENT 1
+#define DC__NUM_PIPES__1_PRESENT 1
+#define DC__NUM_PIPES__2_PRESENT 1
+#define DC__NUM_PIPES__3_PRESENT 1
+#define DC__NUM_PIPES__MAX 6
+#define DC__NUM_PIPES__MAX__6 1
+#define DC__NUM_DIG_LP 0
+#define DC__NUM_DIG_LP__0 1
+#define DC__NUM_DIG_LP__MAX 2
+#define DC__NUM_DIG_LP__MAX__2 1
+#define DC__DPDEBUG_PRESENT 0
+#define DC__DPDEBUG_PRESENT__0 1
+#define DC__DISPLAY_WB_PRESENT 1
+#define DC__DISPLAY_WB_PRESENT__1 1
+#define DC__NUM_CWB 0
+#define DC__NUM_CWB__0 1
+#define DC__NUM_CWB__MAX 2
+#define DC__NUM_CWB__MAX__2 1
+#define DC__MVP_PRESENT 0
+#define DC__MVP_PRESENT__0 1
+#define DC__DVO_PRESENT 0
+#define DC__DVO_PRESENT__0 1
+#define DC__ABM_PRESENT 0
+#define DC__ABM_PRESENT__0 1
+#define DC__BPHYC_PLL_PRESENT 0
+#define DC__BPHYC_PLL_PRESENT__0 1
+#define DC__BPHYC_UNIPHY_PRESENT 0
+#define DC__BPHYC_UNIPHY_PRESENT__0 1
+#define DC__PHY_BROADCAST_PRESENT 0
+#define DC__PHY_BROADCAST_PRESENT__0 1
+#define DC__NUM_OF_DCRX_SD 0
+#define DC__NUM_OF_DCRX_SD__0 1
+#define DC__DVO_17BIT_MAPPING 0
+#define DC__DVO_17BIT_MAPPING__0 1
+#define DC__AVSYNC_PRESENT 0
+#define DC__AVSYNC_PRESENT__0 1
+#define DC__NUM_OF_DCRX_PORTS 0
+#define DC__NUM_OF_DCRX_PORTS__0 1
+#define DC__NUM_OF_DCRX_PORTS__MAX 1
+#define DC__NUM_OF_DCRX_PORTS__MAX__1 1
+#define DC__NUM_PHY 4
+#define DC__NUM_PHY__4 1
+#define DC__NUM_PHY__0_PRESENT 1
+#define DC__NUM_PHY__1_PRESENT 1
+#define DC__NUM_PHY__2_PRESENT 1
+#define DC__NUM_PHY__3_PRESENT 1
+#define DC__NUM_PHY__MAX 7
+#define DC__NUM_PHY__MAX__7 1
+#define DC__NUM_PHY_LP 0
+#define DC__NUM_PHY_LP__0 1
+#define DC__NUM_PHY_LP__MAX 2
+#define DC__NUM_PHY_LP__MAX__2 1
+#define DC__SYNC_CELL vid_sync_gf14lpp
+#define DC__SYNC_CELL__VID_SYNC_GF14LPP 1
+#define DC__USE_NEW_VSS 1
+#define DC__USE_NEW_VSS__1 1
+#define DC__SYNC_CELL_DISPCLK_NUM_LATCHES 6
+#define DC__SYNC_CELL_DISPCLK_NUM_LATCHES__6 1
+#define DC__SYNC_CELL_DVOCLK_NUM_LATCHES 6
+#define DC__SYNC_CELL_DVOCLK_NUM_LATCHES__6 1
+#define DC__SYNC_CELL_PIXCLK_NUM_LATCHES 6
+#define DC__SYNC_CELL_PIXCLK_NUM_LATCHES__6 1
+#define DC__SYNC_CELL_SYMCLK_NUM_LATCHES 6
+#define DC__SYNC_CELL_SYMCLK_NUM_LATCHES__6 1
+#define DC__SYNC_CELL_DPPCLK_NUM_LATCHES 6
+#define DC__SYNC_CELL_DPPCLK_NUM_LATCHES__6 1
+#define DC__SYNC_CELL_DPREFCLK_NUM_LATCHES 6
+#define DC__SYNC_CELL_DPREFCLK_NUM_LATCHES__6 1
+#define DC__SYNC_CELL_REFCLK_NUM_LATCHES 6
+#define DC__SYNC_CELL_REFCLK_NUM_LATCHES__6 1
+#define DC__SYNC_CELL_PCIE_REFCLK_NUM_LATCHES 6
+#define DC__SYNC_CELL_PCIE_REFCLK_NUM_LATCHES__6 1
+#define DC__SYNC_CELL_MVPCLK_NUM_LATCHES 6
+#define DC__SYNC_CELL_MVPCLK_NUM_LATCHES__6 1
+#define DC__SYNC_CELL_SCLK_NUM_LATCHES 6
+#define DC__SYNC_CELL_SCLK_NUM_LATCHES__6 1
+#define DC__SYNC_CELL_DCEFCLK_NUM_LATCHES 6
+#define DC__SYNC_CELL_DCEFCLK_NUM_LATCHES__6 1
+#define DC__SYNC_CELL_AMCLK_NUM_LATCHES 6
+#define DC__SYNC_CELL_AMCLK_NUM_LATCHES__6 1
+#define DC__SYNC_CELL_DSICLK_NUM_LATCHES 6
+#define DC__SYNC_CELL_DSICLK_NUM_LATCHES__6 1
+#define DC__SYNC_CELL_BYTECLK_NUM_LATCHES 6
+#define DC__SYNC_CELL_BYTECLK_NUM_LATCHES__6 1
+#define DC__SYNC_CELL_ESCCLK_NUM_LATCHES 6
+#define DC__SYNC_CELL_ESCCLK_NUM_LATCHES__6 1
+#define DC__SYNC_CELL_DB_CLK_NUM_LATCHES 6
+#define DC__SYNC_CELL_DB_CLK_NUM_LATCHES__6 1
+#define UNIPHYA_PRESENT 1
+#define UNIPHYA_PRESENT__1 1
+#define DC__UNIPHYA_PRESENT 1
+#define DC__UNIPHYA_PRESENT__1 1
+#define UNIPHYB_PRESENT 1
+#define UNIPHYB_PRESENT__1 1
+#define DC__UNIPHYB_PRESENT 1
+#define DC__UNIPHYB_PRESENT__1 1
+#define UNIPHYC_PRESENT 1
+#define UNIPHYC_PRESENT__1 1
+#define DC__UNIPHYC_PRESENT 1
+#define DC__UNIPHYC_PRESENT__1 1
+#define UNIPHYD_PRESENT 1
+#define UNIPHYD_PRESENT__1 1
+#define DC__UNIPHYD_PRESENT 1
+#define DC__UNIPHYD_PRESENT__1 1
+#define UNIPHYE_PRESENT 0
+#define UNIPHYE_PRESENT__0 1
+#define DC__UNIPHYE_PRESENT 0
+#define DC__UNIPHYE_PRESENT__0 1
+#define UNIPHYF_PRESENT 0
+#define UNIPHYF_PRESENT__0 1
+#define DC__UNIPHYF_PRESENT 0
+#define DC__UNIPHYF_PRESENT__0 1
+#define UNIPHYG_PRESENT 0
+#define UNIPHYG_PRESENT__0 1
+#define DC__UNIPHYG_PRESENT 0
+#define DC__UNIPHYG_PRESENT__0 1
+#define DC__TMDS_LINK tmds_link_dual
+#define DC__TMDS_LINK__TMDS_LINK_DUAL 1
+#define DC__WBSCL_PIXBW 8
+#define DC__WBSCL_PIXBW__8 1
+#define DC__DWB_CSC_PRESENT 0
+#define DC__DWB_CSC_PRESENT__0 1
+#define DC__DWB_LUMA_SCL_PRESENT 0
+#define DC__DWB_LUMA_SCL_PRESENT__0 1
+#define DC__DENTIST_INTERFACE_PRESENT 1
+#define DC__DENTIST_INTERFACE_PRESENT__1 1
+#define DC__GENERICA_PRESENT 1
+#define DC__GENERICA_PRESENT__1 1
+#define DC__GENERICB_PRESENT 1
+#define DC__GENERICB_PRESENT__1 1
+#define DC__GENERICC_PRESENT 0
+#define DC__GENERICC_PRESENT__0 1
+#define DC__GENERICD_PRESENT 0
+#define DC__GENERICD_PRESENT__0 1
+#define DC__GENERICE_PRESENT 0
+#define DC__GENERICE_PRESENT__0 1
+#define DC__GENERICF_PRESENT 0
+#define DC__GENERICF_PRESENT__0 1
+#define DC__GENERICG_PRESENT 0
+#define DC__GENERICG_PRESENT__0 1
+#define DC__UNIPHY_VOLTAGE_MODE 1
+#define DC__UNIPHY_VOLTAGE_MODE__1 1
+#define DC__BLON_TYPE dedicated
+#define DC__BLON_TYPE__DEDICATED 1
+#define DC__UNIPHY_STAGGER_CH_PRESENT 1
+#define DC__UNIPHY_STAGGER_CH_PRESENT__1 1
+#define DC__XDMA_PRESENT 0
+#define DC__XDMA_PRESENT__0 1
+#define XDMA__PRESENT 0
+#define XDMA__PRESENT__0 1
+#define DC__DP_MEM_PG 0
+#define DC__DP_MEM_PG__0 1
+#define DP__MEM_PG 0
+#define DP__MEM_PG__0 1
+#define DC__AFMT_MEM_PG 0
+#define DC__AFMT_MEM_PG__0 1
+#define AFMT__MEM_PG 0
+#define AFMT__MEM_PG__0 1
+#define DC__HDMI_MEM_PG 0
+#define DC__HDMI_MEM_PG__0 1
+#define HDMI__MEM_PG 0
+#define HDMI__MEM_PG__0 1
+#define DC__I2C_MEM_PG 0
+#define DC__I2C_MEM_PG__0 1
+#define I2C__MEM_PG 0
+#define I2C__MEM_PG__0 1
+#define DC__DSCL_MEM_PG 0
+#define DC__DSCL_MEM_PG__0 1
+#define DSCL__MEM_PG 0
+#define DSCL__MEM_PG__0 1
+#define DC__CM_MEM_PG 0
+#define DC__CM_MEM_PG__0 1
+#define CM__MEM_PG 0
+#define CM__MEM_PG__0 1
+#define DC__OBUF_MEM_PG 0
+#define DC__OBUF_MEM_PG__0 1
+#define OBUF__MEM_PG 0
+#define OBUF__MEM_PG__0 1
+#define DC__WBIF_MEM_PG 1
+#define DC__WBIF_MEM_PG__1 1
+#define WBIF__MEM_PG 1
+#define WBIF__MEM_PG__1 1
+#define DC__VGA_MEM_PG 0
+#define DC__VGA_MEM_PG__0 1
+#define VGA__MEM_PG 0
+#define VGA__MEM_PG__0 1
+#define DC__FMT_MEM_PG 0
+#define DC__FMT_MEM_PG__0 1
+#define FMT__MEM_PG 0
+#define FMT__MEM_PG__0 1
+#define DC__ODM_MEM_PG 0
+#define DC__ODM_MEM_PG__0 1
+#define ODM__MEM_PG 0
+#define ODM__MEM_PG__0 1
+#define DC__DSI_MEM_PG 0
+#define DC__DSI_MEM_PG__0 1
+#define DSI__MEM_PG 0
+#define DSI__MEM_PG__0 1
+#define DC__AZ_MEM_PG 1
+#define DC__AZ_MEM_PG__1 1
+#define AZ__MEM_PG 1
+#define AZ__MEM_PG__1 1
+#define DC__WBSCL_MEM1P1024X64QS_MEM_PG 1
+#define DC__WBSCL_MEM1P1024X64QS_MEM_PG__1 1
+#define WBSCL_MEM1P1024X64QS__MEM_PG 1
+#define WBSCL_MEM1P1024X64QS__MEM_PG__1 1
+#define DC__WBSCL_MEM1P528X64QS_MEM_PG 1
+#define DC__WBSCL_MEM1P528X64QS_MEM_PG__1 1
+#define WBSCL_MEM1P528X64QS__MEM_PG 1
+#define WBSCL_MEM1P528X64QS__MEM_PG__1 1
+#define DC__DMCU_MEM1P1024X32BQS_MEM_PG 1
+#define DC__DMCU_MEM1P1024X32BQS_MEM_PG__1 1
+#define DMCU_MEM1P1024X32BQS__MEM_PG 1
+#define DMCU_MEM1P1024X32BQS__MEM_PG__1 1
+#define DC__HUBBUB_SDP_TAG_INT_MEM_PG 0
+#define DC__HUBBUB_SDP_TAG_INT_MEM_PG__0 1
+#define HUBBUB_SDP_TAG_INT__MEM_PG 0
+#define HUBBUB_SDP_TAG_INT__MEM_PG__0 1
+#define DC__HUBBUB_SDP_TAG_EXT_MEM_PG 0
+#define DC__HUBBUB_SDP_TAG_EXT_MEM_PG__0 1
+#define HUBBUB_SDP_TAG_EXT__MEM_PG 0
+#define HUBBUB_SDP_TAG_EXT__MEM_PG__0 1
+#define DC__HUBBUB_RET_ZERO_MEM_PG 0
+#define DC__HUBBUB_RET_ZERO_MEM_PG__0 1
+#define HUBBUB_RET_ZERO__MEM_PG 0
+#define HUBBUB_RET_ZERO__MEM_PG__0 1
+#define DC__HUBBUB_RET_ROB_MEM_PG 0
+#define DC__HUBBUB_RET_ROB_MEM_PG__0 1
+#define HUBBUB_RET_ROB__MEM_PG 0
+#define HUBBUB_RET_ROB__MEM_PG__0 1
+#define DC__HUBPRET_CUR_ROB_MEM_PG 0
+#define DC__HUBPRET_CUR_ROB_MEM_PG__0 1
+#define HUBPRET_CUR_ROB__MEM_PG 0
+#define HUBPRET_CUR_ROB__MEM_PG__0 1
+#define DC__HUBPRET_CUR_CDC_MEM_PG 0
+#define DC__HUBPRET_CUR_CDC_MEM_PG__0 1
+#define HUBPRET_CUR_CDC__MEM_PG 0
+#define HUBPRET_CUR_CDC__MEM_PG__0 1
+#define DC__HUBPREQ_MPTE_MEM_PG 0
+#define DC__HUBPREQ_MPTE_MEM_PG__0 1
+#define HUBPREQ_MPTE__MEM_PG 0
+#define HUBPREQ_MPTE__MEM_PG__0 1
+#define DC__HUBPREQ_META_MEM_PG 0
+#define DC__HUBPREQ_META_MEM_PG__0 1
+#define HUBPREQ_META__MEM_PG 0
+#define HUBPREQ_META__MEM_PG__0 1
+#define DC__HUBPREQ_DPTE_MEM_PG 0
+#define DC__HUBPREQ_DPTE_MEM_PG__0 1
+#define HUBPREQ_DPTE__MEM_PG 0
+#define HUBPREQ_DPTE__MEM_PG__0 1
+#define DC__HUBPRET_DET_MEM_PG 0
+#define DC__HUBPRET_DET_MEM_PG__0 1
+#define HUBPRET_DET__MEM_PG 0
+#define HUBPRET_DET__MEM_PG__0 1
+#define DC__HUBPRET_PIX_CDC_MEM_PG 0
+#define DC__HUBPRET_PIX_CDC_MEM_PG__0 1
+#define HUBPRET_PIX_CDC__MEM_PG 0
+#define HUBPRET_PIX_CDC__MEM_PG__0 1
+#define DC__TOP_BLKS__DCCG 1
+#define DC__TOP_BLKS__DCHUBBUB 1
+#define DC__TOP_BLKS__DCHUBP 1
+#define DC__TOP_BLKS__HDA 1
+#define DC__TOP_BLKS__DIO 1
+#define DC__TOP_BLKS__DCIO 1
+#define DC__TOP_BLKS__DMU 1
+#define DC__TOP_BLKS__DPP 1
+#define DC__TOP_BLKS__MPC 1
+#define DC__TOP_BLKS__OPP 1
+#define DC__TOP_BLKS__OPTC 1
+#define DC__TOP_BLKS__MMHUBBUB 1
+#define DC__TOP_BLKS__WB 1
+#define DC__TOP_BLKS__MAX 13
+#define DC__TOP_BLKS__MAX__13 1
+#define DC__DCHUBP_DPP_SF_PIXEL_CREDITS 9
+#define DC__DCHUBP_DPP_SF_PIXEL_CREDITS__9 1
+#define DC__DPP_MPC_SF_PIXEL_CREDITS 9
+#define DC__DPP_MPC_SF_PIXEL_CREDITS__9 1
+#define DC__MPC_OPP_SF_PIXEL_CREDITS 8
+#define DC__MPC_OPP_SF_PIXEL_CREDITS__8 1
+#define DC__OPP_OPTC_SF_PIXEL_CREDITS 8
+#define DC__OPP_OPTC_SF_PIXEL_CREDITS__8 1
+#define DC__SFR_SFT_ROUND_TRIP_DELAY 5
+#define DC__SFR_SFT_ROUND_TRIP_DELAY__5 1
+#define DC__REPEATER_PROJECT_MAX 8
+#define DC__REPEATER_PROJECT_MAX__8 1
+#define DC__SURFACE_422_CAPABLE 0
+#define DC__SURFACE_422_CAPABLE__0 1
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn10/dcn10_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn10/dcn10_fpu.c
new file mode 100644
index 000000000000..c5e84190c17a
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn10/dcn10_fpu.c
@@ -0,0 +1,159 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dcn10/dcn10_resource.h"
+
+#include "dcn10_fpu.h"
+#include "resource.h"
+#include "amdgpu_dm/dc_fpu.h"
+
+/**
+ * DOC: DCN10 FPU manipulation Overview
+ *
+ * The DCN architecture relies on FPU operations, which require special
+ * compilation flags and the use of kernel_fpu_begin/end functions; ideally, we
+ * want to avoid spreading FPU access across multiple files. With this idea in
+ * mind, this file aims to centralize DCN10 functions that require FPU access
+ * in a single place. Code in this file follows the following code pattern:
+ *
+ * 1. Functions that use FPU operations should be isolated in static functions.
+ * 2. The FPU functions should have the noinline attribute to ensure anything
+ *    that deals with FP register is contained within this call.
+ * 3. All function that needs to be accessed outside this file requires a
+ *    public interface that not uses any FPU reference.
+ * 4. Developers **must not** use DC_FP_START/END in this file, but they need
+ *    to ensure that the caller invokes it before access any function available
+ *    in this file. For this reason, public functions in this file must invoke
+ *    dc_assert_fp_enabled();
+ *
+ * Let's expand a little bit more the idea in the code pattern. To fully
+ * isolate FPU operations in a single place, we must avoid situations where
+ * compilers spill FP values to registers due to FP enable in a specific C
+ * file. Note that even if we isolate all FPU functions in a single file and
+ * call its interface from other files, the compiler might enable the use of
+ * FPU before we call DC_FP_START. Nevertheless, it is the programmer's
+ * responsibility to invoke DC_FP_START/END in the correct place. To highlight
+ * situations where developers forgot to use the FP protection before calling
+ * the DC FPU interface functions, we introduce a helper that checks if the
+ * function is invoked under FP protection. If not, it will trigger a kernel
+ * warning.
+ */
+
+struct _vcs_dpi_ip_params_st dcn1_0_ip = {
+	.rob_buffer_size_kbytes = 64,
+	.det_buffer_size_kbytes = 164,
+	.dpte_buffer_size_in_pte_reqs_luma = 42,
+	.dpp_output_buffer_pixels = 2560,
+	.opp_output_buffer_lines = 1,
+	.pixel_chunk_size_kbytes = 8,
+	.pte_enable = 1,
+	.pte_chunk_size_kbytes = 2,
+	.meta_chunk_size_kbytes = 2,
+	.writeback_chunk_size_kbytes = 2,
+	.line_buffer_size_bits = 589824,
+	.max_line_buffer_lines = 12,
+	.IsLineBufferBppFixed = 0,
+	.LineBufferFixedBpp = -1,
+	.writeback_luma_buffer_size_kbytes = 12,
+	.writeback_chroma_buffer_size_kbytes = 8,
+	.max_num_dpp = 4,
+	.max_num_wb = 2,
+	.max_dchub_pscl_bw_pix_per_clk = 4,
+	.max_pscl_lb_bw_pix_per_clk = 2,
+	.max_lb_vscl_bw_pix_per_clk = 4,
+	.max_vscl_hscl_bw_pix_per_clk = 4,
+	.max_hscl_ratio = 4,
+	.max_vscl_ratio = 4,
+	.hscl_mults = 4,
+	.vscl_mults = 4,
+	.max_hscl_taps = 8,
+	.max_vscl_taps = 8,
+	.dispclk_ramp_margin_percent = 1,
+	.underscan_factor = 1.10,
+	.min_vblank_lines = 14,
+	.dppclk_delay_subtotal = 90,
+	.dispclk_delay_subtotal = 42,
+	.dcfclk_cstate_latency = 10,
+	.max_inter_dcn_tile_repeaters = 8,
+	.can_vstartup_lines_exceed_vsync_plus_back_porch_lines_minus_one = 0,
+	.bug_forcing_LC_req_same_size_fixed = 0,
+};
+
+struct _vcs_dpi_soc_bounding_box_st dcn1_0_soc = {
+	.sr_exit_time_us = 9.0,
+	.sr_enter_plus_exit_time_us = 11.0,
+	.urgent_latency_us = 4.0,
+	.writeback_latency_us = 12.0,
+	.ideal_dram_bw_after_urgent_percent = 80.0,
+	.max_request_size_bytes = 256,
+	.downspread_percent = 0.5,
+	.dram_page_open_time_ns = 50.0,
+	.dram_rw_turnaround_time_ns = 17.5,
+	.dram_return_buffer_per_channel_bytes = 8192,
+	.round_trip_ping_latency_dcfclk_cycles = 128,
+	.urgent_out_of_order_return_per_channel_bytes = 256,
+	.channel_interleave_bytes = 256,
+	.num_banks = 8,
+	.num_chans = 2,
+	.vmm_page_size_bytes = 4096,
+	.dram_clock_change_latency_us = 17.0,
+	.writeback_dram_clock_change_latency_us = 23.0,
+	.return_bus_width_bytes = 64,
+};
+
+void dcn10_resource_construct_fp(struct dc *dc)
+{
+	dc_assert_fp_enabled();
+	if (dc->ctx->dce_version == DCN_VERSION_1_01) {
+		struct dcn_soc_bounding_box *dcn_soc = dc->dcn_soc;
+		struct dcn_ip_params *dcn_ip = dc->dcn_ip;
+		struct display_mode_lib *dml = &dc->dml;
+
+		dml->ip.max_num_dpp = 3;
+		/* TODO how to handle 23.84? */
+		dcn_soc->dram_clock_change_latency = 23;
+		dcn_ip->max_num_dpp = 3;
+	}
+	if (ASICREV_IS_RV1_F0(dc->ctx->asic_id.hw_internal_rev)) {
+		dc->dcn_soc->urgent_latency = 3;
+		dc->debug.disable_dmcu = true;
+		dc->dcn_soc->fabric_and_dram_bandwidth_vmax0p9 = 41.60f;
+	}
+
+	dc->dcn_soc->number_of_channels = dc->ctx->asic_id.vram_width / ddr4_dram_width;
+	ASSERT(dc->dcn_soc->number_of_channels < 3);
+	if (dc->dcn_soc->number_of_channels == 0)/*old sbios bug*/
+		dc->dcn_soc->number_of_channels = 2;
+
+	if (dc->dcn_soc->number_of_channels == 1) {
+		dc->dcn_soc->fabric_and_dram_bandwidth_vmax0p9 = 19.2f;
+		dc->dcn_soc->fabric_and_dram_bandwidth_vnom0p8 = 17.066f;
+		dc->dcn_soc->fabric_and_dram_bandwidth_vmid0p72 = 14.933f;
+		dc->dcn_soc->fabric_and_dram_bandwidth_vmin0p65 = 12.8f;
+		if (ASICREV_IS_RV1_F0(dc->ctx->asic_id.hw_internal_rev))
+			dc->dcn_soc->fabric_and_dram_bandwidth_vmax0p9 = 20.80f;
+	}
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn10/dcn10_fpu.h b/drivers/gpu/drm/amd/display/dc/dml/dcn10/dcn10_fpu.h
new file mode 100644
index 000000000000..1bf6b12f5663
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn10/dcn10_fpu.h
@@ -0,0 +1,32 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DCN10_FPU_H__
+#define __DCN10_FPU_H__
+
+void dcn10_resource_construct_fp(struct dc *dc);
+
+#endif /* __DCN10_FPU_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn20/dcn20_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn20/dcn20_fpu.c
new file mode 100644
index 000000000000..7aaf13bbd4e4
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn20/dcn20_fpu.c
@@ -0,0 +1,2559 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "resource.h"
+#include "clk_mgr.h"
+#include "dchubbub.h"
+#include "dcn20/dcn20_resource.h"
+#include "dcn21/dcn21_resource.h"
+#include "clk_mgr/dcn21/rn_clk_mgr.h"
+#include "link_service.h"
+#include "dcn20_fpu.h"
+#include "dc_state_priv.h"
+
+#define DC_LOGGER \
+	dc->ctx->logger
+#define DC_LOGGER_INIT(logger)
+
+#ifndef MAX
+#define MAX(X, Y) ((X) > (Y) ? (X) : (Y))
+#endif
+#ifndef MIN
+#define MIN(X, Y) ((X) < (Y) ? (X) : (Y))
+#endif
+
+/* Constant */
+#define LPDDR_MEM_RETRAIN_LATENCY 4.977 /* Number obtained from LPDDR4 Training Counter Requirement doc */
+
+/**
+ * DOC: DCN2x FPU manipulation Overview
+ *
+ * The DCN architecture relies on FPU operations, which require special
+ * compilation flags and the use of kernel_fpu_begin/end functions; ideally, we
+ * want to avoid spreading FPU access across multiple files. With this idea in
+ * mind, this file aims to centralize all DCN20 and DCN2.1 (DCN2x) functions
+ * that require FPU access in a single place. Code in this file follows the
+ * following code pattern:
+ *
+ * 1. Functions that use FPU operations should be isolated in static functions.
+ * 2. The FPU functions should have the noinline attribute to ensure anything
+ *    that deals with FP register is contained within this call.
+ * 3. All function that needs to be accessed outside this file requires a
+ *    public interface that not uses any FPU reference.
+ * 4. Developers **must not** use DC_FP_START/END in this file, but they need
+ *    to ensure that the caller invokes it before access any function available
+ *    in this file. For this reason, public functions in this file must invoke
+ *    dc_assert_fp_enabled();
+ *
+ * Let's expand a little bit more the idea in the code pattern. To fully
+ * isolate FPU operations in a single place, we must avoid situations where
+ * compilers spill FP values to registers due to FP enable in a specific C
+ * file. Note that even if we isolate all FPU functions in a single file and
+ * call its interface from other files, the compiler might enable the use of
+ * FPU before we call DC_FP_START. Nevertheless, it is the programmer's
+ * responsibility to invoke DC_FP_START/END in the correct place. To highlight
+ * situations where developers forgot to use the FP protection before calling
+ * the DC FPU interface functions, we introduce a helper that checks if the
+ * function is invoked under FP protection. If not, it will trigger a kernel
+ * warning.
+ */
+
+struct _vcs_dpi_ip_params_st dcn2_0_ip = {
+	.odm_capable = 1,
+	.gpuvm_enable = 0,
+	.hostvm_enable = 0,
+	.gpuvm_max_page_table_levels = 4,
+	.hostvm_max_page_table_levels = 4,
+	.hostvm_cached_page_table_levels = 0,
+	.pte_group_size_bytes = 2048,
+	.num_dsc = 6,
+	.rob_buffer_size_kbytes = 168,
+	.det_buffer_size_kbytes = 164,
+	.dpte_buffer_size_in_pte_reqs_luma = 84,
+	.pde_proc_buffer_size_64k_reqs = 48,
+	.dpp_output_buffer_pixels = 2560,
+	.opp_output_buffer_lines = 1,
+	.pixel_chunk_size_kbytes = 8,
+	.pte_chunk_size_kbytes = 2,
+	.meta_chunk_size_kbytes = 2,
+	.writeback_chunk_size_kbytes = 2,
+	.line_buffer_size_bits = 789504,
+	.is_line_buffer_bpp_fixed = 0,
+	.line_buffer_fixed_bpp = 0,
+	.dcc_supported = true,
+	.max_line_buffer_lines = 12,
+	.writeback_luma_buffer_size_kbytes = 12,
+	.writeback_chroma_buffer_size_kbytes = 8,
+	.writeback_chroma_line_buffer_width_pixels = 4,
+	.writeback_max_hscl_ratio = 1,
+	.writeback_max_vscl_ratio = 1,
+	.writeback_min_hscl_ratio = 1,
+	.writeback_min_vscl_ratio = 1,
+	.writeback_max_hscl_taps = 12,
+	.writeback_max_vscl_taps = 12,
+	.writeback_line_buffer_luma_buffer_size = 0,
+	.writeback_line_buffer_chroma_buffer_size = 14643,
+	.cursor_buffer_size = 8,
+	.cursor_chunk_size = 2,
+	.max_num_otg = 6,
+	.max_num_dpp = 6,
+	.max_num_wb = 1,
+	.max_dchub_pscl_bw_pix_per_clk = 4,
+	.max_pscl_lb_bw_pix_per_clk = 2,
+	.max_lb_vscl_bw_pix_per_clk = 4,
+	.max_vscl_hscl_bw_pix_per_clk = 4,
+	.max_hscl_ratio = 8,
+	.max_vscl_ratio = 8,
+	.hscl_mults = 4,
+	.vscl_mults = 4,
+	.max_hscl_taps = 8,
+	.max_vscl_taps = 8,
+	.dispclk_ramp_margin_percent = 1,
+	.underscan_factor = 1.10,
+	.min_vblank_lines = 32, //
+	.dppclk_delay_subtotal = 77, //
+	.dppclk_delay_scl_lb_only = 16,
+	.dppclk_delay_scl = 50,
+	.dppclk_delay_cnvc_formatter = 8,
+	.dppclk_delay_cnvc_cursor = 6,
+	.dispclk_delay_subtotal = 87, //
+	.dcfclk_cstate_latency = 10, // SRExitTime
+	.max_inter_dcn_tile_repeaters = 8,
+	.xfc_supported = true,
+	.xfc_fill_bw_overhead_percent = 10.0,
+	.xfc_fill_constant_bytes = 0,
+	.number_of_cursors = 1,
+};
+
+struct _vcs_dpi_ip_params_st dcn2_0_nv14_ip = {
+	.odm_capable = 1,
+	.gpuvm_enable = 0,
+	.hostvm_enable = 0,
+	.gpuvm_max_page_table_levels = 4,
+	.hostvm_max_page_table_levels = 4,
+	.hostvm_cached_page_table_levels = 0,
+	.num_dsc = 5,
+	.rob_buffer_size_kbytes = 168,
+	.det_buffer_size_kbytes = 164,
+	.dpte_buffer_size_in_pte_reqs_luma = 84,
+	.dpte_buffer_size_in_pte_reqs_chroma = 42,//todo
+	.dpp_output_buffer_pixels = 2560,
+	.opp_output_buffer_lines = 1,
+	.pixel_chunk_size_kbytes = 8,
+	.pte_enable = 1,
+	.max_page_table_levels = 4,
+	.pte_chunk_size_kbytes = 2,
+	.meta_chunk_size_kbytes = 2,
+	.writeback_chunk_size_kbytes = 2,
+	.line_buffer_size_bits = 789504,
+	.is_line_buffer_bpp_fixed = 0,
+	.line_buffer_fixed_bpp = 0,
+	.dcc_supported = true,
+	.max_line_buffer_lines = 12,
+	.writeback_luma_buffer_size_kbytes = 12,
+	.writeback_chroma_buffer_size_kbytes = 8,
+	.writeback_chroma_line_buffer_width_pixels = 4,
+	.writeback_max_hscl_ratio = 1,
+	.writeback_max_vscl_ratio = 1,
+	.writeback_min_hscl_ratio = 1,
+	.writeback_min_vscl_ratio = 1,
+	.writeback_max_hscl_taps = 12,
+	.writeback_max_vscl_taps = 12,
+	.writeback_line_buffer_luma_buffer_size = 0,
+	.writeback_line_buffer_chroma_buffer_size = 14643,
+	.cursor_buffer_size = 8,
+	.cursor_chunk_size = 2,
+	.max_num_otg = 5,
+	.max_num_dpp = 5,
+	.max_num_wb = 1,
+	.max_dchub_pscl_bw_pix_per_clk = 4,
+	.max_pscl_lb_bw_pix_per_clk = 2,
+	.max_lb_vscl_bw_pix_per_clk = 4,
+	.max_vscl_hscl_bw_pix_per_clk = 4,
+	.max_hscl_ratio = 8,
+	.max_vscl_ratio = 8,
+	.hscl_mults = 4,
+	.vscl_mults = 4,
+	.max_hscl_taps = 8,
+	.max_vscl_taps = 8,
+	.dispclk_ramp_margin_percent = 1,
+	.underscan_factor = 1.10,
+	.min_vblank_lines = 32, //
+	.dppclk_delay_subtotal = 77, //
+	.dppclk_delay_scl_lb_only = 16,
+	.dppclk_delay_scl = 50,
+	.dppclk_delay_cnvc_formatter = 8,
+	.dppclk_delay_cnvc_cursor = 6,
+	.dispclk_delay_subtotal = 87, //
+	.dcfclk_cstate_latency = 10, // SRExitTime
+	.max_inter_dcn_tile_repeaters = 8,
+	.xfc_supported = true,
+	.xfc_fill_bw_overhead_percent = 10.0,
+	.xfc_fill_constant_bytes = 0,
+	.ptoi_supported = 0,
+	.number_of_cursors = 1,
+};
+
+struct _vcs_dpi_soc_bounding_box_st dcn2_0_soc = {
+	/* Defaults that get patched on driver load from firmware. */
+	.clock_limits = {
+			{
+				.state = 0,
+				.dcfclk_mhz = 560.0,
+				.fabricclk_mhz = 560.0,
+				.dispclk_mhz = 513.0,
+				.dppclk_mhz = 513.0,
+				.phyclk_mhz = 540.0,
+				.socclk_mhz = 560.0,
+				.dscclk_mhz = 171.0,
+				.dram_speed_mts = 8960.0,
+			},
+			{
+				.state = 1,
+				.dcfclk_mhz = 694.0,
+				.fabricclk_mhz = 694.0,
+				.dispclk_mhz = 642.0,
+				.dppclk_mhz = 642.0,
+				.phyclk_mhz = 600.0,
+				.socclk_mhz = 694.0,
+				.dscclk_mhz = 214.0,
+				.dram_speed_mts = 11104.0,
+			},
+			{
+				.state = 2,
+				.dcfclk_mhz = 875.0,
+				.fabricclk_mhz = 875.0,
+				.dispclk_mhz = 734.0,
+				.dppclk_mhz = 734.0,
+				.phyclk_mhz = 810.0,
+				.socclk_mhz = 875.0,
+				.dscclk_mhz = 245.0,
+				.dram_speed_mts = 14000.0,
+			},
+			{
+				.state = 3,
+				.dcfclk_mhz = 1000.0,
+				.fabricclk_mhz = 1000.0,
+				.dispclk_mhz = 1100.0,
+				.dppclk_mhz = 1100.0,
+				.phyclk_mhz = 810.0,
+				.socclk_mhz = 1000.0,
+				.dscclk_mhz = 367.0,
+				.dram_speed_mts = 16000.0,
+			},
+			{
+				.state = 4,
+				.dcfclk_mhz = 1200.0,
+				.fabricclk_mhz = 1200.0,
+				.dispclk_mhz = 1284.0,
+				.dppclk_mhz = 1284.0,
+				.phyclk_mhz = 810.0,
+				.socclk_mhz = 1200.0,
+				.dscclk_mhz = 428.0,
+				.dram_speed_mts = 16000.0,
+			},
+			/*Extra state, no dispclk ramping*/
+			{
+				.state = 5,
+				.dcfclk_mhz = 1200.0,
+				.fabricclk_mhz = 1200.0,
+				.dispclk_mhz = 1284.0,
+				.dppclk_mhz = 1284.0,
+				.phyclk_mhz = 810.0,
+				.socclk_mhz = 1200.0,
+				.dscclk_mhz = 428.0,
+				.dram_speed_mts = 16000.0,
+			},
+		},
+	.num_states = 5,
+	.sr_exit_time_us = 8.6,
+	.sr_enter_plus_exit_time_us = 10.9,
+	.urgent_latency_us = 4.0,
+	.urgent_latency_pixel_data_only_us = 4.0,
+	.urgent_latency_pixel_mixed_with_vm_data_us = 4.0,
+	.urgent_latency_vm_data_only_us = 4.0,
+	.urgent_out_of_order_return_per_channel_pixel_only_bytes = 4096,
+	.urgent_out_of_order_return_per_channel_pixel_and_vm_bytes = 4096,
+	.urgent_out_of_order_return_per_channel_vm_only_bytes = 4096,
+	.pct_ideal_dram_sdp_bw_after_urgent_pixel_only = 40.0,
+	.pct_ideal_dram_sdp_bw_after_urgent_pixel_and_vm = 40.0,
+	.pct_ideal_dram_sdp_bw_after_urgent_vm_only = 40.0,
+	.max_avg_sdp_bw_use_normal_percent = 40.0,
+	.max_avg_dram_bw_use_normal_percent = 40.0,
+	.writeback_latency_us = 12.0,
+	.ideal_dram_bw_after_urgent_percent = 40.0,
+	.max_request_size_bytes = 256,
+	.dram_channel_width_bytes = 2,
+	.fabric_datapath_to_dcn_data_return_bytes = 64,
+	.dcn_downspread_percent = 0.5,
+	.downspread_percent = 0.38,
+	.dram_page_open_time_ns = 50.0,
+	.dram_rw_turnaround_time_ns = 17.5,
+	.dram_return_buffer_per_channel_bytes = 8192,
+	.round_trip_ping_latency_dcfclk_cycles = 131,
+	.urgent_out_of_order_return_per_channel_bytes = 256,
+	.channel_interleave_bytes = 256,
+	.num_banks = 8,
+	.num_chans = 16,
+	.vmm_page_size_bytes = 4096,
+	.dram_clock_change_latency_us = 404.0,
+	.dummy_pstate_latency_us = 5.0,
+	.writeback_dram_clock_change_latency_us = 23.0,
+	.return_bus_width_bytes = 64,
+	.dispclk_dppclk_vco_speed_mhz = 3850,
+	.xfc_bus_transport_time_us = 20,
+	.xfc_xbuf_latency_tolerance_us = 4,
+	.use_urgent_burst_bw = 0
+};
+
+struct _vcs_dpi_soc_bounding_box_st dcn2_0_nv14_soc = {
+	.clock_limits = {
+			{
+				.state = 0,
+				.dcfclk_mhz = 560.0,
+				.fabricclk_mhz = 560.0,
+				.dispclk_mhz = 513.0,
+				.dppclk_mhz = 513.0,
+				.phyclk_mhz = 540.0,
+				.socclk_mhz = 560.0,
+				.dscclk_mhz = 171.0,
+				.dram_speed_mts = 8960.0,
+			},
+			{
+				.state = 1,
+				.dcfclk_mhz = 694.0,
+				.fabricclk_mhz = 694.0,
+				.dispclk_mhz = 642.0,
+				.dppclk_mhz = 642.0,
+				.phyclk_mhz = 600.0,
+				.socclk_mhz = 694.0,
+				.dscclk_mhz = 214.0,
+				.dram_speed_mts = 11104.0,
+			},
+			{
+				.state = 2,
+				.dcfclk_mhz = 875.0,
+				.fabricclk_mhz = 875.0,
+				.dispclk_mhz = 734.0,
+				.dppclk_mhz = 734.0,
+				.phyclk_mhz = 810.0,
+				.socclk_mhz = 875.0,
+				.dscclk_mhz = 245.0,
+				.dram_speed_mts = 14000.0,
+			},
+			{
+				.state = 3,
+				.dcfclk_mhz = 1000.0,
+				.fabricclk_mhz = 1000.0,
+				.dispclk_mhz = 1100.0,
+				.dppclk_mhz = 1100.0,
+				.phyclk_mhz = 810.0,
+				.socclk_mhz = 1000.0,
+				.dscclk_mhz = 367.0,
+				.dram_speed_mts = 16000.0,
+			},
+			{
+				.state = 4,
+				.dcfclk_mhz = 1200.0,
+				.fabricclk_mhz = 1200.0,
+				.dispclk_mhz = 1284.0,
+				.dppclk_mhz = 1284.0,
+				.phyclk_mhz = 810.0,
+				.socclk_mhz = 1200.0,
+				.dscclk_mhz = 428.0,
+				.dram_speed_mts = 16000.0,
+			},
+			/*Extra state, no dispclk ramping*/
+			{
+				.state = 5,
+				.dcfclk_mhz = 1200.0,
+				.fabricclk_mhz = 1200.0,
+				.dispclk_mhz = 1284.0,
+				.dppclk_mhz = 1284.0,
+				.phyclk_mhz = 810.0,
+				.socclk_mhz = 1200.0,
+				.dscclk_mhz = 428.0,
+				.dram_speed_mts = 16000.0,
+			},
+		},
+	.num_states = 5,
+	.sr_exit_time_us = 11.6,
+	.sr_enter_plus_exit_time_us = 13.9,
+	.urgent_latency_us = 4.0,
+	.urgent_latency_pixel_data_only_us = 4.0,
+	.urgent_latency_pixel_mixed_with_vm_data_us = 4.0,
+	.urgent_latency_vm_data_only_us = 4.0,
+	.urgent_out_of_order_return_per_channel_pixel_only_bytes = 4096,
+	.urgent_out_of_order_return_per_channel_pixel_and_vm_bytes = 4096,
+	.urgent_out_of_order_return_per_channel_vm_only_bytes = 4096,
+	.pct_ideal_dram_sdp_bw_after_urgent_pixel_only = 40.0,
+	.pct_ideal_dram_sdp_bw_after_urgent_pixel_and_vm = 40.0,
+	.pct_ideal_dram_sdp_bw_after_urgent_vm_only = 40.0,
+	.max_avg_sdp_bw_use_normal_percent = 40.0,
+	.max_avg_dram_bw_use_normal_percent = 40.0,
+	.writeback_latency_us = 12.0,
+	.ideal_dram_bw_after_urgent_percent = 40.0,
+	.max_request_size_bytes = 256,
+	.dram_channel_width_bytes = 2,
+	.fabric_datapath_to_dcn_data_return_bytes = 64,
+	.dcn_downspread_percent = 0.5,
+	.downspread_percent = 0.38,
+	.dram_page_open_time_ns = 50.0,
+	.dram_rw_turnaround_time_ns = 17.5,
+	.dram_return_buffer_per_channel_bytes = 8192,
+	.round_trip_ping_latency_dcfclk_cycles = 131,
+	.urgent_out_of_order_return_per_channel_bytes = 256,
+	.channel_interleave_bytes = 256,
+	.num_banks = 8,
+	.num_chans = 8,
+	.vmm_page_size_bytes = 4096,
+	.dram_clock_change_latency_us = 404.0,
+	.dummy_pstate_latency_us = 5.0,
+	.writeback_dram_clock_change_latency_us = 23.0,
+	.return_bus_width_bytes = 64,
+	.dispclk_dppclk_vco_speed_mhz = 3850,
+	.xfc_bus_transport_time_us = 20,
+	.xfc_xbuf_latency_tolerance_us = 4,
+	.use_urgent_burst_bw = 0
+};
+
+struct _vcs_dpi_soc_bounding_box_st dcn2_0_nv12_soc = {
+	.clock_limits = {
+		{
+			.state = 0,
+			.dcfclk_mhz = 560.0,
+			.fabricclk_mhz = 560.0,
+			.dispclk_mhz = 513.0,
+			.dppclk_mhz = 513.0,
+			.phyclk_mhz = 540.0,
+			.socclk_mhz = 560.0,
+			.dscclk_mhz = 171.0,
+			.dram_speed_mts = 1069.0,
+		},
+		{
+			.state = 1,
+			.dcfclk_mhz = 694.0,
+			.fabricclk_mhz = 694.0,
+			.dispclk_mhz = 642.0,
+			.dppclk_mhz = 642.0,
+			.phyclk_mhz = 600.0,
+			.socclk_mhz = 694.0,
+			.dscclk_mhz = 214.0,
+			.dram_speed_mts = 1324.0,
+		},
+		{
+			.state = 2,
+			.dcfclk_mhz = 875.0,
+			.fabricclk_mhz = 875.0,
+			.dispclk_mhz = 734.0,
+			.dppclk_mhz = 734.0,
+			.phyclk_mhz = 810.0,
+			.socclk_mhz = 875.0,
+			.dscclk_mhz = 245.0,
+			.dram_speed_mts = 1670.0,
+		},
+		{
+			.state = 3,
+			.dcfclk_mhz = 1000.0,
+			.fabricclk_mhz = 1000.0,
+			.dispclk_mhz = 1100.0,
+			.dppclk_mhz = 1100.0,
+			.phyclk_mhz = 810.0,
+			.socclk_mhz = 1000.0,
+			.dscclk_mhz = 367.0,
+			.dram_speed_mts = 2000.0,
+		},
+		{
+			.state = 4,
+			.dcfclk_mhz = 1200.0,
+			.fabricclk_mhz = 1200.0,
+			.dispclk_mhz = 1284.0,
+			.dppclk_mhz = 1284.0,
+			.phyclk_mhz = 810.0,
+			.socclk_mhz = 1200.0,
+			.dscclk_mhz = 428.0,
+			.dram_speed_mts = 2000.0,
+		},
+		{
+			.state = 5,
+			.dcfclk_mhz = 1200.0,
+			.fabricclk_mhz = 1200.0,
+			.dispclk_mhz = 1284.0,
+			.dppclk_mhz = 1284.0,
+			.phyclk_mhz = 810.0,
+			.socclk_mhz = 1200.0,
+			.dscclk_mhz = 428.0,
+			.dram_speed_mts = 2000.0,
+		},
+	},
+
+	.num_states = 5,
+	.sr_exit_time_us = 1.9,
+	.sr_enter_plus_exit_time_us = 4.4,
+	.urgent_latency_us = 3.0,
+	.urgent_latency_pixel_data_only_us = 4.0,
+	.urgent_latency_pixel_mixed_with_vm_data_us = 4.0,
+	.urgent_latency_vm_data_only_us = 4.0,
+	.urgent_out_of_order_return_per_channel_pixel_only_bytes = 4096,
+	.urgent_out_of_order_return_per_channel_pixel_and_vm_bytes = 4096,
+	.urgent_out_of_order_return_per_channel_vm_only_bytes = 4096,
+	.pct_ideal_dram_sdp_bw_after_urgent_pixel_only = 40.0,
+	.pct_ideal_dram_sdp_bw_after_urgent_pixel_and_vm = 40.0,
+	.pct_ideal_dram_sdp_bw_after_urgent_vm_only = 40.0,
+	.max_avg_sdp_bw_use_normal_percent = 40.0,
+	.max_avg_dram_bw_use_normal_percent = 40.0,
+	.writeback_latency_us = 12.0,
+	.ideal_dram_bw_after_urgent_percent = 40.0,
+	.max_request_size_bytes = 256,
+	.dram_channel_width_bytes = 16,
+	.fabric_datapath_to_dcn_data_return_bytes = 64,
+	.dcn_downspread_percent = 0.5,
+	.downspread_percent = 0.5,
+	.dram_page_open_time_ns = 50.0,
+	.dram_rw_turnaround_time_ns = 17.5,
+	.dram_return_buffer_per_channel_bytes = 8192,
+	.round_trip_ping_latency_dcfclk_cycles = 131,
+	.urgent_out_of_order_return_per_channel_bytes = 4096,
+	.channel_interleave_bytes = 256,
+	.num_banks = 8,
+	.num_chans = 16,
+	.vmm_page_size_bytes = 4096,
+	.dram_clock_change_latency_us = 45.0,
+	.writeback_dram_clock_change_latency_us = 23.0,
+	.return_bus_width_bytes = 64,
+	.dispclk_dppclk_vco_speed_mhz = 3850,
+	.xfc_bus_transport_time_us = 20,
+	.xfc_xbuf_latency_tolerance_us = 50,
+	.use_urgent_burst_bw = 0,
+};
+
+struct _vcs_dpi_ip_params_st dcn2_1_ip = {
+	.odm_capable = 1,
+	.gpuvm_enable = 1,
+	.hostvm_enable = 1,
+	.gpuvm_max_page_table_levels = 1,
+	.hostvm_max_page_table_levels = 4,
+	.hostvm_cached_page_table_levels = 2,
+	.num_dsc = 3,
+	.rob_buffer_size_kbytes = 168,
+	.det_buffer_size_kbytes = 164,
+	.dpte_buffer_size_in_pte_reqs_luma = 44,
+	.dpte_buffer_size_in_pte_reqs_chroma = 42,//todo
+	.dpp_output_buffer_pixels = 2560,
+	.opp_output_buffer_lines = 1,
+	.pixel_chunk_size_kbytes = 8,
+	.pte_enable = 1,
+	.max_page_table_levels = 4,
+	.pte_chunk_size_kbytes = 2,
+	.meta_chunk_size_kbytes = 2,
+	.min_meta_chunk_size_bytes = 256,
+	.writeback_chunk_size_kbytes = 2,
+	.line_buffer_size_bits = 789504,
+	.is_line_buffer_bpp_fixed = 0,
+	.line_buffer_fixed_bpp = 0,
+	.dcc_supported = true,
+	.max_line_buffer_lines = 12,
+	.writeback_luma_buffer_size_kbytes = 12,
+	.writeback_chroma_buffer_size_kbytes = 8,
+	.writeback_chroma_line_buffer_width_pixels = 4,
+	.writeback_max_hscl_ratio = 1,
+	.writeback_max_vscl_ratio = 1,
+	.writeback_min_hscl_ratio = 1,
+	.writeback_min_vscl_ratio = 1,
+	.writeback_max_hscl_taps = 12,
+	.writeback_max_vscl_taps = 12,
+	.writeback_line_buffer_luma_buffer_size = 0,
+	.writeback_line_buffer_chroma_buffer_size = 14643,
+	.cursor_buffer_size = 8,
+	.cursor_chunk_size = 2,
+	.max_num_otg = 4,
+	.max_num_dpp = 4,
+	.max_num_wb = 1,
+	.max_dchub_pscl_bw_pix_per_clk = 4,
+	.max_pscl_lb_bw_pix_per_clk = 2,
+	.max_lb_vscl_bw_pix_per_clk = 4,
+	.max_vscl_hscl_bw_pix_per_clk = 4,
+	.max_hscl_ratio = 4,
+	.max_vscl_ratio = 4,
+	.hscl_mults = 4,
+	.vscl_mults = 4,
+	.max_hscl_taps = 8,
+	.max_vscl_taps = 8,
+	.dispclk_ramp_margin_percent = 1,
+	.underscan_factor = 1.10,
+	.min_vblank_lines = 32, //
+	.dppclk_delay_subtotal = 77, //
+	.dppclk_delay_scl_lb_only = 16,
+	.dppclk_delay_scl = 50,
+	.dppclk_delay_cnvc_formatter = 8,
+	.dppclk_delay_cnvc_cursor = 6,
+	.dispclk_delay_subtotal = 87, //
+	.dcfclk_cstate_latency = 10, // SRExitTime
+	.max_inter_dcn_tile_repeaters = 8,
+
+	.xfc_supported = false,
+	.xfc_fill_bw_overhead_percent = 10.0,
+	.xfc_fill_constant_bytes = 0,
+	.ptoi_supported = 0,
+	.number_of_cursors = 1,
+};
+
+struct _vcs_dpi_soc_bounding_box_st dcn2_1_soc = {
+	.clock_limits = {
+			{
+				.state = 0,
+				.dcfclk_mhz = 400.0,
+				.fabricclk_mhz = 400.0,
+				.dispclk_mhz = 600.0,
+				.dppclk_mhz = 400.00,
+				.phyclk_mhz = 600.0,
+				.socclk_mhz = 278.0,
+				.dscclk_mhz = 205.67,
+				.dram_speed_mts = 1600.0,
+			},
+			{
+				.state = 1,
+				.dcfclk_mhz = 464.52,
+				.fabricclk_mhz = 800.0,
+				.dispclk_mhz = 654.55,
+				.dppclk_mhz = 626.09,
+				.phyclk_mhz = 600.0,
+				.socclk_mhz = 278.0,
+				.dscclk_mhz = 205.67,
+				.dram_speed_mts = 1600.0,
+			},
+			{
+				.state = 2,
+				.dcfclk_mhz = 514.29,
+				.fabricclk_mhz = 933.0,
+				.dispclk_mhz = 757.89,
+				.dppclk_mhz = 685.71,
+				.phyclk_mhz = 600.0,
+				.socclk_mhz = 278.0,
+				.dscclk_mhz = 287.67,
+				.dram_speed_mts = 1866.0,
+			},
+			{
+				.state = 3,
+				.dcfclk_mhz = 576.00,
+				.fabricclk_mhz = 1067.0,
+				.dispclk_mhz = 847.06,
+				.dppclk_mhz = 757.89,
+				.phyclk_mhz = 600.0,
+				.socclk_mhz = 715.0,
+				.dscclk_mhz = 318.334,
+				.dram_speed_mts = 2134.0,
+			},
+			{
+				.state = 4,
+				.dcfclk_mhz = 626.09,
+				.fabricclk_mhz = 1200.0,
+				.dispclk_mhz = 900.00,
+				.dppclk_mhz = 847.06,
+				.phyclk_mhz = 810.0,
+				.socclk_mhz = 953.0,
+				.dscclk_mhz = 300.0,
+				.dram_speed_mts = 2400.0,
+			},
+			{
+				.state = 5,
+				.dcfclk_mhz = 685.71,
+				.fabricclk_mhz = 1333.0,
+				.dispclk_mhz = 1028.57,
+				.dppclk_mhz = 960.00,
+				.phyclk_mhz = 810.0,
+				.socclk_mhz = 278.0,
+				.dscclk_mhz = 342.86,
+				.dram_speed_mts = 2666.0,
+			},
+			{
+				.state = 6,
+				.dcfclk_mhz = 757.89,
+				.fabricclk_mhz = 1467.0,
+				.dispclk_mhz = 1107.69,
+				.dppclk_mhz = 1028.57,
+				.phyclk_mhz = 810.0,
+				.socclk_mhz = 715.0,
+				.dscclk_mhz = 369.23,
+				.dram_speed_mts = 3200.0,
+			},
+			{
+				.state = 7,
+				.dcfclk_mhz = 847.06,
+				.fabricclk_mhz = 1600.0,
+				.dispclk_mhz = 1395.0,
+				.dppclk_mhz = 1285.00,
+				.phyclk_mhz = 1325.0,
+				.socclk_mhz = 953.0,
+				.dscclk_mhz = 489.0,
+				.dram_speed_mts = 4266.0,
+			},
+			/*Extra state, no dispclk ramping*/
+			{
+				.state = 8,
+				.dcfclk_mhz = 847.06,
+				.fabricclk_mhz = 1600.0,
+				.dispclk_mhz = 1395.0,
+				.dppclk_mhz = 1285.0,
+				.phyclk_mhz = 1325.0,
+				.socclk_mhz = 953.0,
+				.dscclk_mhz = 489.0,
+				.dram_speed_mts = 4266.0,
+			},
+
+		},
+
+	.sr_exit_time_us = 12.5,
+	.sr_enter_plus_exit_time_us = 17.0,
+	.urgent_latency_us = 4.0,
+	.urgent_latency_pixel_data_only_us = 4.0,
+	.urgent_latency_pixel_mixed_with_vm_data_us = 4.0,
+	.urgent_latency_vm_data_only_us = 4.0,
+	.urgent_out_of_order_return_per_channel_pixel_only_bytes = 4096,
+	.urgent_out_of_order_return_per_channel_pixel_and_vm_bytes = 4096,
+	.urgent_out_of_order_return_per_channel_vm_only_bytes = 4096,
+	.pct_ideal_dram_sdp_bw_after_urgent_pixel_only = 80.0,
+	.pct_ideal_dram_sdp_bw_after_urgent_pixel_and_vm = 75.0,
+	.pct_ideal_dram_sdp_bw_after_urgent_vm_only = 40.0,
+	.max_avg_sdp_bw_use_normal_percent = 60.0,
+	.max_avg_dram_bw_use_normal_percent = 100.0,
+	.writeback_latency_us = 12.0,
+	.max_request_size_bytes = 256,
+	.dram_channel_width_bytes = 4,
+	.fabric_datapath_to_dcn_data_return_bytes = 32,
+	.dcn_downspread_percent = 0.5,
+	.downspread_percent = 0.38,
+	.dram_page_open_time_ns = 50.0,
+	.dram_rw_turnaround_time_ns = 17.5,
+	.dram_return_buffer_per_channel_bytes = 8192,
+	.round_trip_ping_latency_dcfclk_cycles = 128,
+	.urgent_out_of_order_return_per_channel_bytes = 4096,
+	.channel_interleave_bytes = 256,
+	.num_banks = 8,
+	.num_chans = 4,
+	.vmm_page_size_bytes = 4096,
+	.dram_clock_change_latency_us = 23.84,
+	.return_bus_width_bytes = 64,
+	.dispclk_dppclk_vco_speed_mhz = 3600,
+	.xfc_bus_transport_time_us = 4,
+	.xfc_xbuf_latency_tolerance_us = 4,
+	.use_urgent_burst_bw = 1,
+	.num_states = 8
+};
+
+struct wm_table ddr4_wm_table_gs = {
+	.entries = {
+		{
+			.wm_inst = WM_A,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 7.09,
+			.sr_enter_plus_exit_time_us = 8.14,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_B,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 10.12,
+			.sr_enter_plus_exit_time_us = 11.48,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_C,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 10.12,
+			.sr_enter_plus_exit_time_us = 11.48,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_D,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 10.12,
+			.sr_enter_plus_exit_time_us = 11.48,
+			.valid = true,
+		},
+	}
+};
+
+struct wm_table lpddr4_wm_table_gs = {
+	.entries = {
+		{
+			.wm_inst = WM_A,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 5.32,
+			.sr_enter_plus_exit_time_us = 6.38,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_B,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 9.82,
+			.sr_enter_plus_exit_time_us = 11.196,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_C,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 9.89,
+			.sr_enter_plus_exit_time_us = 11.24,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_D,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 9.748,
+			.sr_enter_plus_exit_time_us = 11.102,
+			.valid = true,
+		},
+	}
+};
+
+struct wm_table lpddr4_wm_table_with_disabled_ppt = {
+	.entries = {
+		{
+			.wm_inst = WM_A,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 8.32,
+			.sr_enter_plus_exit_time_us = 9.38,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_B,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 9.82,
+			.sr_enter_plus_exit_time_us = 11.196,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_C,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 9.89,
+			.sr_enter_plus_exit_time_us = 11.24,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_D,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 9.748,
+			.sr_enter_plus_exit_time_us = 11.102,
+			.valid = true,
+		},
+	}
+};
+
+struct wm_table ddr4_wm_table_rn = {
+	.entries = {
+		{
+			.wm_inst = WM_A,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 11.90,
+			.sr_enter_plus_exit_time_us = 12.80,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_B,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 13.18,
+			.sr_enter_plus_exit_time_us = 14.30,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_C,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 13.18,
+			.sr_enter_plus_exit_time_us = 14.30,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_D,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 13.18,
+			.sr_enter_plus_exit_time_us = 14.30,
+			.valid = true,
+		},
+	}
+};
+
+struct wm_table ddr4_1R_wm_table_rn = {
+	.entries = {
+		{
+			.wm_inst = WM_A,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 13.90,
+			.sr_enter_plus_exit_time_us = 14.80,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_B,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 13.90,
+			.sr_enter_plus_exit_time_us = 14.80,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_C,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 13.90,
+			.sr_enter_plus_exit_time_us = 14.80,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_D,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.72,
+			.sr_exit_time_us = 13.90,
+			.sr_enter_plus_exit_time_us = 14.80,
+			.valid = true,
+		},
+	}
+};
+
+struct wm_table lpddr4_wm_table_rn = {
+	.entries = {
+		{
+			.wm_inst = WM_A,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 7.32,
+			.sr_enter_plus_exit_time_us = 8.38,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_B,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 9.82,
+			.sr_enter_plus_exit_time_us = 11.196,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_C,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 9.89,
+			.sr_enter_plus_exit_time_us = 11.24,
+			.valid = true,
+		},
+		{
+			.wm_inst = WM_D,
+			.wm_type = WM_TYPE_PSTATE_CHG,
+			.pstate_latency_us = 11.65333,
+			.sr_exit_time_us = 9.748,
+			.sr_enter_plus_exit_time_us = 11.102,
+			.valid = true,
+		},
+	}
+};
+
+void dcn20_populate_dml_writeback_from_context(struct dc *dc,
+					       struct resource_context *res_ctx,
+					       display_e2e_pipe_params_st *pipes)
+{
+	int pipe_cnt, i;
+
+	dc_assert_fp_enabled();
+
+	for (i = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) {
+		struct dc_writeback_info *wb_info = &res_ctx->pipe_ctx[i].stream->writeback_info[0];
+
+		if (!res_ctx->pipe_ctx[i].stream)
+			continue;
+
+		/* Set writeback information */
+		pipes[pipe_cnt].dout.wb_enable = (wb_info->wb_enabled == true) ? 1 : 0;
+		pipes[pipe_cnt].dout.num_active_wb++;
+		pipes[pipe_cnt].dout.wb.wb_src_height = wb_info->dwb_params.cnv_params.crop_height;
+		pipes[pipe_cnt].dout.wb.wb_src_width = wb_info->dwb_params.cnv_params.crop_width;
+		pipes[pipe_cnt].dout.wb.wb_dst_width = wb_info->dwb_params.dest_width;
+		pipes[pipe_cnt].dout.wb.wb_dst_height = wb_info->dwb_params.dest_height;
+		pipes[pipe_cnt].dout.wb.wb_htaps_luma = 1;
+		pipes[pipe_cnt].dout.wb.wb_vtaps_luma = 1;
+		pipes[pipe_cnt].dout.wb.wb_htaps_chroma = wb_info->dwb_params.scaler_taps.h_taps_c;
+		pipes[pipe_cnt].dout.wb.wb_vtaps_chroma = wb_info->dwb_params.scaler_taps.v_taps_c;
+		pipes[pipe_cnt].dout.wb.wb_hratio = 1.0;
+		pipes[pipe_cnt].dout.wb.wb_vratio = 1.0;
+		if (wb_info->dwb_params.out_format == dwb_scaler_mode_yuv420) {
+			if (wb_info->dwb_params.output_depth == DWB_OUTPUT_PIXEL_DEPTH_8BPC)
+				pipes[pipe_cnt].dout.wb.wb_pixel_format = dm_420_8;
+			else
+				pipes[pipe_cnt].dout.wb.wb_pixel_format = dm_420_10;
+		} else {
+			pipes[pipe_cnt].dout.wb.wb_pixel_format = dm_444_32;
+		}
+
+		pipe_cnt++;
+	}
+}
+
+void dcn20_fpu_set_wb_arb_params(struct mcif_arb_params *wb_arb_params,
+				 struct dc_state *context,
+				 display_e2e_pipe_params_st *pipes,
+				 int pipe_cnt, int i)
+{
+	int k;
+
+	dc_assert_fp_enabled();
+
+	for (k = 0; k < sizeof(wb_arb_params->cli_watermark)/sizeof(wb_arb_params->cli_watermark[0]); k++) {
+		wb_arb_params->cli_watermark[k] = get_wm_writeback_urgent(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+		wb_arb_params->pstate_watermark[k] = get_wm_writeback_dram_clock_change(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	}
+	wb_arb_params->time_per_pixel = 16.0 * 1000 / (context->res_ctx.pipe_ctx[i].stream->phy_pix_clk / 1000); /* 4 bit fraction, ms */
+}
+
+static bool is_dtbclk_required(struct dc *dc, struct dc_state *context)
+{
+	int i;
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		if (!context->res_ctx.pipe_ctx[i].stream)
+			continue;
+		if (dc->link_srv->dp_is_128b_132b_signal(&context->res_ctx.pipe_ctx[i]))
+			return true;
+	}
+	return false;
+}
+
+static enum dcn_zstate_support_state  decide_zstate_support(struct dc *dc, struct dc_state *context)
+{
+	int plane_count;
+	int i;
+
+	plane_count = 0;
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		if (context->res_ctx.pipe_ctx[i].plane_state)
+			plane_count++;
+	}
+
+	/*
+	 * Z9 and Z10 allowed cases:
+	 * 	1. 0 Planes enabled
+	 * 	2. single eDP, on link 0, 1 plane and stutter period > 5ms
+	 * Z10 only cases:
+	 * 	1. single eDP, on link 0, 1 plane and stutter period >= 5ms
+	 * Z8 cases:
+	 * 	1. stutter period sufficient
+	 * Zstate not allowed cases:
+	 * 	1. Everything else
+	 */
+	if (plane_count == 0)
+		return DCN_ZSTATE_SUPPORT_ALLOW;
+	else if (context->stream_count == 1 &&  context->streams[0]->signal == SIGNAL_TYPE_EDP) {
+		struct dc_link *link = context->streams[0]->sink->link;
+		struct dc_stream_status *stream_status = &context->stream_status[0];
+		int minmum_z8_residency = dc->debug.minimum_z8_residency_time > 0 ? dc->debug.minimum_z8_residency_time : 1000;
+		bool allow_z8 = context->bw_ctx.dml.vba.StutterPeriod > (double)minmum_z8_residency;
+		bool is_pwrseq0 = (link && link->link_index == 0);
+		bool is_psr = (link && (link->psr_settings.psr_version == DC_PSR_VERSION_1 ||
+						link->psr_settings.psr_version == DC_PSR_VERSION_SU_1) && !link->panel_config.psr.disable_psr);
+		bool is_replay = link && link->replay_settings.replay_feature_enabled;
+
+		/* Don't support multi-plane configurations */
+		if (stream_status->plane_count > 1)
+			return DCN_ZSTATE_SUPPORT_DISALLOW;
+
+		if (is_pwrseq0 && context->bw_ctx.dml.vba.StutterPeriod > 5000.0)
+			return DCN_ZSTATE_SUPPORT_ALLOW;
+		else if (is_pwrseq0 && (is_psr || is_replay))
+			return DCN_ZSTATE_SUPPORT_ALLOW_Z8_Z10_ONLY;
+		else
+			return allow_z8 ? DCN_ZSTATE_SUPPORT_ALLOW_Z8_ONLY : DCN_ZSTATE_SUPPORT_DISALLOW;
+	} else {
+		return DCN_ZSTATE_SUPPORT_DISALLOW;
+	}
+}
+
+static void dcn20_adjust_freesync_v_startup(
+		const struct dc_crtc_timing *dc_crtc_timing, int *vstartup_start)
+{
+	struct dc_crtc_timing patched_crtc_timing;
+	uint32_t asic_blank_end   = 0;
+	uint32_t asic_blank_start = 0;
+	uint32_t newVstartup	  = 0;
+
+	patched_crtc_timing = *dc_crtc_timing;
+
+	if (patched_crtc_timing.flags.INTERLACE == 1) {
+		if (patched_crtc_timing.v_front_porch < 2)
+			patched_crtc_timing.v_front_porch = 2;
+	} else {
+		if (patched_crtc_timing.v_front_porch < 1)
+			patched_crtc_timing.v_front_porch = 1;
+	}
+
+	/* blank_start = frame end - front porch */
+	asic_blank_start = patched_crtc_timing.v_total -
+					patched_crtc_timing.v_front_porch;
+
+	/* blank_end = blank_start - active */
+	asic_blank_end = asic_blank_start -
+					patched_crtc_timing.v_border_bottom -
+					patched_crtc_timing.v_addressable -
+					patched_crtc_timing.v_border_top;
+
+	/* The newVStartUp is 1 line before vsync point */
+	newVstartup = asic_blank_end + 1;
+
+	*vstartup_start = ((newVstartup > *vstartup_start) ? newVstartup : *vstartup_start);
+}
+
+void dcn20_calculate_dlg_params(struct dc *dc,
+				struct dc_state *context,
+				display_e2e_pipe_params_st *pipes,
+				int pipe_cnt,
+				int vlevel)
+{
+	int i, pipe_idx, active_hubp_count = 0;
+
+	dc_assert_fp_enabled();
+
+	/* Writeback MCIF_WB arbitration parameters */
+	dc->res_pool->funcs->set_mcif_arb_params(dc, context, pipes, pipe_cnt);
+
+	context->bw_ctx.bw.dcn.clk.dispclk_khz = context->bw_ctx.dml.vba.DISPCLK * 1000;
+	context->bw_ctx.bw.dcn.clk.dcfclk_khz = context->bw_ctx.dml.vba.DCFCLK * 1000;
+	context->bw_ctx.bw.dcn.clk.socclk_khz = context->bw_ctx.dml.vba.SOCCLK * 1000;
+	context->bw_ctx.bw.dcn.clk.dramclk_khz = context->bw_ctx.dml.vba.DRAMSpeed * 1000 / 16;
+
+	if (dc->debug.min_dram_clk_khz > context->bw_ctx.bw.dcn.clk.dramclk_khz)
+		context->bw_ctx.bw.dcn.clk.dramclk_khz = dc->debug.min_dram_clk_khz;
+
+	context->bw_ctx.bw.dcn.clk.dcfclk_deep_sleep_khz = context->bw_ctx.dml.vba.DCFCLKDeepSleep * 1000;
+	context->bw_ctx.bw.dcn.clk.fclk_khz = context->bw_ctx.dml.vba.FabricClock * 1000;
+	context->bw_ctx.bw.dcn.clk.p_state_change_support =
+		context->bw_ctx.dml.vba.DRAMClockChangeSupport[vlevel][context->bw_ctx.dml.vba.maxMpcComb]
+							!= dm_dram_clock_change_unsupported;
+
+	/* Pstate change might not be supported by hardware, but it might be
+	 * possible with firmware driven vertical blank stretching.
+	 */
+	context->bw_ctx.bw.dcn.clk.p_state_change_support |= context->bw_ctx.bw.dcn.clk.fw_based_mclk_switching;
+
+	context->bw_ctx.bw.dcn.clk.dppclk_khz = 0;
+
+	context->bw_ctx.bw.dcn.clk.dtbclk_en = is_dtbclk_required(dc, context);
+
+	if (context->bw_ctx.bw.dcn.clk.dispclk_khz < dc->debug.min_disp_clk_khz)
+		context->bw_ctx.bw.dcn.clk.dispclk_khz = dc->debug.min_disp_clk_khz;
+
+	for (i = 0, pipe_idx = 0; i < dc->res_pool->pipe_count; i++) {
+		if (!context->res_ctx.pipe_ctx[i].stream)
+			continue;
+		if (context->res_ctx.pipe_ctx[i].plane_state)
+			active_hubp_count++;
+		pipes[pipe_idx].pipe.dest.vstartup_start = get_vstartup(&context->bw_ctx.dml, pipes, pipe_cnt, pipe_idx);
+		pipes[pipe_idx].pipe.dest.vupdate_offset = get_vupdate_offset(&context->bw_ctx.dml, pipes, pipe_cnt, pipe_idx);
+		pipes[pipe_idx].pipe.dest.vupdate_width = get_vupdate_width(&context->bw_ctx.dml, pipes, pipe_cnt, pipe_idx);
+		pipes[pipe_idx].pipe.dest.vready_offset = get_vready_offset(&context->bw_ctx.dml, pipes, pipe_cnt, pipe_idx);
+
+		if (dc_state_get_pipe_subvp_type(context, &context->res_ctx.pipe_ctx[i]) == SUBVP_PHANTOM) {
+			// Phantom pipe requires that DET_SIZE = 0 and no unbounded requests
+			context->res_ctx.pipe_ctx[i].det_buffer_size_kb = 0;
+			context->res_ctx.pipe_ctx[i].unbounded_req = false;
+		} else {
+			context->res_ctx.pipe_ctx[i].det_buffer_size_kb = context->bw_ctx.dml.ip.det_buffer_size_kbytes;
+			context->res_ctx.pipe_ctx[i].unbounded_req = pipes[pipe_idx].pipe.src.unbounded_req_mode;
+		}
+
+		if (context->bw_ctx.bw.dcn.clk.dppclk_khz < pipes[pipe_idx].clks_cfg.dppclk_mhz * 1000)
+			context->bw_ctx.bw.dcn.clk.dppclk_khz = pipes[pipe_idx].clks_cfg.dppclk_mhz * 1000;
+		context->res_ctx.pipe_ctx[i].plane_res.bw.dppclk_khz =
+						pipes[pipe_idx].clks_cfg.dppclk_mhz * 1000;
+		context->res_ctx.pipe_ctx[i].pipe_dlg_param = pipes[pipe_idx].pipe.dest;
+		if (dc->ctx->dce_version < DCN_VERSION_3_1 &&
+		    context->res_ctx.pipe_ctx[i].stream->adaptive_sync_infopacket.valid)
+			dcn20_adjust_freesync_v_startup(
+				&context->res_ctx.pipe_ctx[i].stream->timing,
+				&context->res_ctx.pipe_ctx[i].pipe_dlg_param.vstartup_start);
+
+		pipe_idx++;
+	}
+	/* If DCN isn't making memory requests we can allow pstate change */
+	if (!active_hubp_count) {
+		context->bw_ctx.bw.dcn.clk.p_state_change_support = true;
+	}
+	/*save a original dppclock copy*/
+	context->bw_ctx.bw.dcn.clk.bw_dppclk_khz = context->bw_ctx.bw.dcn.clk.dppclk_khz;
+	context->bw_ctx.bw.dcn.clk.bw_dispclk_khz = context->bw_ctx.bw.dcn.clk.dispclk_khz;
+	context->bw_ctx.bw.dcn.clk.max_supported_dppclk_khz = context->bw_ctx.dml.soc.clock_limits[vlevel].dppclk_mhz * 1000;
+	context->bw_ctx.bw.dcn.clk.max_supported_dispclk_khz = context->bw_ctx.dml.soc.clock_limits[vlevel].dispclk_mhz * 1000;
+
+	context->bw_ctx.bw.dcn.compbuf_size_kb = context->bw_ctx.dml.ip.config_return_buffer_size_in_kbytes
+						- context->bw_ctx.dml.ip.det_buffer_size_kbytes * pipe_idx;
+
+	for (i = 0, pipe_idx = 0; i < dc->res_pool->pipe_count; i++) {
+		bool cstate_en = context->bw_ctx.dml.vba.PrefetchMode[vlevel][context->bw_ctx.dml.vba.maxMpcComb] != 2;
+
+		if (!context->res_ctx.pipe_ctx[i].stream)
+			continue;
+
+		/* cstate disabled on 201 */
+		if (dc->ctx->dce_version == DCN_VERSION_2_01)
+			cstate_en = false;
+
+		context->bw_ctx.dml.funcs.rq_dlg_get_dlg_reg(&context->bw_ctx.dml,
+				&context->res_ctx.pipe_ctx[i].dlg_regs,
+				&context->res_ctx.pipe_ctx[i].ttu_regs,
+				pipes,
+				pipe_cnt,
+				pipe_idx,
+				cstate_en,
+				context->bw_ctx.bw.dcn.clk.p_state_change_support,
+				false, false, true);
+
+		context->bw_ctx.dml.funcs.rq_dlg_get_rq_reg(&context->bw_ctx.dml,
+				&context->res_ctx.pipe_ctx[i].rq_regs,
+				&pipes[pipe_idx].pipe);
+		pipe_idx++;
+	}
+	context->bw_ctx.bw.dcn.clk.zstate_support = decide_zstate_support(dc, context);
+}
+
+static void swizzle_to_dml_params(
+		enum swizzle_mode_values swizzle,
+		unsigned int *sw_mode)
+{
+	switch (swizzle) {
+	case DC_SW_LINEAR:
+		*sw_mode = dm_sw_linear;
+		break;
+	case DC_SW_4KB_S:
+		*sw_mode = dm_sw_4kb_s;
+		break;
+	case DC_SW_4KB_S_X:
+		*sw_mode = dm_sw_4kb_s_x;
+		break;
+	case DC_SW_4KB_D:
+		*sw_mode = dm_sw_4kb_d;
+		break;
+	case DC_SW_4KB_D_X:
+		*sw_mode = dm_sw_4kb_d_x;
+		break;
+	case DC_SW_64KB_S:
+		*sw_mode = dm_sw_64kb_s;
+		break;
+	case DC_SW_64KB_S_X:
+		*sw_mode = dm_sw_64kb_s_x;
+		break;
+	case DC_SW_64KB_S_T:
+		*sw_mode = dm_sw_64kb_s_t;
+		break;
+	case DC_SW_64KB_D:
+		*sw_mode = dm_sw_64kb_d;
+		break;
+	case DC_SW_64KB_D_X:
+		*sw_mode = dm_sw_64kb_d_x;
+		break;
+	case DC_SW_64KB_D_T:
+		*sw_mode = dm_sw_64kb_d_t;
+		break;
+	case DC_SW_64KB_R_X:
+		*sw_mode = dm_sw_64kb_r_x;
+		break;
+	case DC_SW_VAR_S:
+		*sw_mode = dm_sw_var_s;
+		break;
+	case DC_SW_VAR_S_X:
+		*sw_mode = dm_sw_var_s_x;
+		break;
+	case DC_SW_VAR_D:
+		*sw_mode = dm_sw_var_d;
+		break;
+	case DC_SW_VAR_D_X:
+		*sw_mode = dm_sw_var_d_x;
+		break;
+	case DC_SW_VAR_R_X:
+		*sw_mode = dm_sw_var_r_x;
+		break;
+	default:
+		ASSERT(0); /* Not supported */
+		break;
+	}
+}
+
+int dcn20_populate_dml_pipes_from_context(struct dc *dc,
+					  struct dc_state *context,
+					  display_e2e_pipe_params_st *pipes,
+					  enum dc_validate_mode validate_mode)
+{
+	int pipe_cnt, i;
+	bool synchronized_vblank = true;
+	struct resource_context *res_ctx = &context->res_ctx;
+
+	dc_assert_fp_enabled();
+
+	for (i = 0, pipe_cnt = -1; i < dc->res_pool->pipe_count; i++) {
+		if (!res_ctx->pipe_ctx[i].stream)
+			continue;
+
+		if (pipe_cnt < 0) {
+			pipe_cnt = i;
+			continue;
+		}
+
+		if (res_ctx->pipe_ctx[pipe_cnt].stream == res_ctx->pipe_ctx[i].stream)
+			continue;
+
+		if (dc->debug.disable_timing_sync ||
+			(!resource_are_streams_timing_synchronizable(
+				res_ctx->pipe_ctx[pipe_cnt].stream,
+				res_ctx->pipe_ctx[i].stream) &&
+			!resource_are_vblanks_synchronizable(
+				res_ctx->pipe_ctx[pipe_cnt].stream,
+				res_ctx->pipe_ctx[i].stream))) {
+			synchronized_vblank = false;
+			break;
+		}
+	}
+
+	for (i = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) {
+		struct dc_crtc_timing *timing = &res_ctx->pipe_ctx[i].stream->timing;
+		unsigned int v_total;
+		unsigned int front_porch;
+		int output_bpc;
+		struct audio_check aud_check = {0};
+
+		if (!res_ctx->pipe_ctx[i].stream)
+			continue;
+
+		v_total = timing->v_total;
+		front_porch = timing->v_front_porch;
+
+		/* todo:
+		pipes[pipe_cnt].pipe.src.dynamic_metadata_enable = 0;
+		pipes[pipe_cnt].pipe.src.dcc = 0;
+		pipes[pipe_cnt].pipe.src.vm = 0;*/
+
+		pipes[pipe_cnt].clks_cfg.refclk_mhz = dc->res_pool->ref_clocks.dchub_ref_clock_inKhz / 1000.0;
+
+		pipes[pipe_cnt].pipe.dest.use_maximum_vstartup = dc->ctx->dce_version == DCN_VERSION_2_01;
+
+		pipes[pipe_cnt].dout.dsc_enable = res_ctx->pipe_ctx[i].stream->timing.flags.DSC;
+		/* todo: rotation?*/
+		pipes[pipe_cnt].dout.dsc_slices = res_ctx->pipe_ctx[i].stream->timing.dsc_cfg.num_slices_h;
+		if (res_ctx->pipe_ctx[i].stream->use_dynamic_meta) {
+			pipes[pipe_cnt].pipe.src.dynamic_metadata_enable = true;
+			/* 1/2 vblank */
+			pipes[pipe_cnt].pipe.src.dynamic_metadata_lines_before_active =
+				(v_total - timing->v_addressable
+					- timing->v_border_top - timing->v_border_bottom) / 2;
+			/* 36 bytes dp, 32 hdmi */
+			pipes[pipe_cnt].pipe.src.dynamic_metadata_xmit_bytes =
+				dc_is_dp_signal(res_ctx->pipe_ctx[i].stream->signal) ? 36 : 32;
+		}
+		pipes[pipe_cnt].pipe.src.dcc = false;
+		pipes[pipe_cnt].pipe.src.dcc_rate = 1;
+		pipes[pipe_cnt].pipe.dest.synchronized_vblank_all_planes = synchronized_vblank;
+		pipes[pipe_cnt].pipe.dest.synchronize_timings = synchronized_vblank;
+		pipes[pipe_cnt].pipe.dest.hblank_start = timing->h_total - timing->h_front_porch;
+		pipes[pipe_cnt].pipe.dest.hblank_end = pipes[pipe_cnt].pipe.dest.hblank_start
+				- timing->h_addressable
+				- timing->h_border_left
+				- timing->h_border_right;
+		pipes[pipe_cnt].pipe.dest.vblank_start = v_total - front_porch;
+		pipes[pipe_cnt].pipe.dest.vblank_end = pipes[pipe_cnt].pipe.dest.vblank_start
+				- timing->v_addressable
+				- timing->v_border_top
+				- timing->v_border_bottom;
+		pipes[pipe_cnt].pipe.dest.htotal = timing->h_total;
+		pipes[pipe_cnt].pipe.dest.vtotal = v_total;
+		pipes[pipe_cnt].pipe.dest.hactive =
+			timing->h_addressable + timing->h_border_left + timing->h_border_right;
+		pipes[pipe_cnt].pipe.dest.vactive =
+			timing->v_addressable + timing->v_border_top + timing->v_border_bottom;
+		pipes[pipe_cnt].pipe.dest.interlaced = timing->flags.INTERLACE;
+		pipes[pipe_cnt].pipe.dest.pixel_rate_mhz = timing->pix_clk_100hz/10000.0;
+		if (timing->timing_3d_format == TIMING_3D_FORMAT_HW_FRAME_PACKING)
+			pipes[pipe_cnt].pipe.dest.pixel_rate_mhz *= 2;
+		pipes[pipe_cnt].pipe.dest.otg_inst = res_ctx->pipe_ctx[i].stream_res.tg->inst;
+		pipes[pipe_cnt].dout.dp_lanes = 4;
+		pipes[pipe_cnt].dout.dp_rate = dm_dp_rate_na;
+		pipes[pipe_cnt].dout.is_virtual = 0;
+		pipes[pipe_cnt].pipe.dest.vtotal_min = res_ctx->pipe_ctx[i].stream->adjust.v_total_min;
+		pipes[pipe_cnt].pipe.dest.vtotal_max = res_ctx->pipe_ctx[i].stream->adjust.v_total_max;
+		switch (resource_get_odm_slice_count(&res_ctx->pipe_ctx[i])) {
+		case 2:
+			pipes[pipe_cnt].pipe.dest.odm_combine = dm_odm_combine_mode_2to1;
+			break;
+		case 4:
+			pipes[pipe_cnt].pipe.dest.odm_combine = dm_odm_combine_mode_4to1;
+			break;
+		default:
+			pipes[pipe_cnt].pipe.dest.odm_combine = dm_odm_combine_mode_disabled;
+		}
+		pipes[pipe_cnt].pipe.src.hsplit_grp = res_ctx->pipe_ctx[i].pipe_idx;
+		if (res_ctx->pipe_ctx[i].top_pipe && res_ctx->pipe_ctx[i].top_pipe->plane_state
+				== res_ctx->pipe_ctx[i].plane_state) {
+			struct pipe_ctx *first_pipe = res_ctx->pipe_ctx[i].top_pipe;
+			int split_idx = 0;
+
+			while (first_pipe->top_pipe && first_pipe->top_pipe->plane_state
+					== res_ctx->pipe_ctx[i].plane_state) {
+				first_pipe = first_pipe->top_pipe;
+				split_idx++;
+			}
+			/* Treat 4to1 mpc combine as an mpo of 2 2-to-1 combines */
+			if (split_idx == 0)
+				pipes[pipe_cnt].pipe.src.hsplit_grp = first_pipe->pipe_idx;
+			else if (split_idx == 1)
+				pipes[pipe_cnt].pipe.src.hsplit_grp = res_ctx->pipe_ctx[i].pipe_idx;
+			else if (split_idx == 2)
+				pipes[pipe_cnt].pipe.src.hsplit_grp = res_ctx->pipe_ctx[i].top_pipe->pipe_idx;
+		} else if (res_ctx->pipe_ctx[i].prev_odm_pipe) {
+			struct pipe_ctx *first_pipe = res_ctx->pipe_ctx[i].prev_odm_pipe;
+
+			while (first_pipe->prev_odm_pipe)
+				first_pipe = first_pipe->prev_odm_pipe;
+			pipes[pipe_cnt].pipe.src.hsplit_grp = first_pipe->pipe_idx;
+		}
+
+		switch (res_ctx->pipe_ctx[i].stream->signal) {
+		case SIGNAL_TYPE_DISPLAY_PORT_MST:
+		case SIGNAL_TYPE_DISPLAY_PORT:
+			pipes[pipe_cnt].dout.output_type = dm_dp;
+			if (dc->link_srv->dp_is_128b_132b_signal(&res_ctx->pipe_ctx[i]))
+				pipes[pipe_cnt].dout.output_type = dm_dp2p0;
+			break;
+		case SIGNAL_TYPE_EDP:
+			pipes[pipe_cnt].dout.output_type = dm_edp;
+			break;
+		case SIGNAL_TYPE_HDMI_TYPE_A:
+		case SIGNAL_TYPE_DVI_SINGLE_LINK:
+		case SIGNAL_TYPE_DVI_DUAL_LINK:
+			pipes[pipe_cnt].dout.output_type = dm_hdmi;
+			break;
+		default:
+			/* In case there is no signal, set dp with 4 lanes to allow max config */
+			pipes[pipe_cnt].dout.is_virtual = 1;
+			pipes[pipe_cnt].dout.output_type = dm_dp;
+			pipes[pipe_cnt].dout.dp_lanes = 4;
+		}
+
+		switch (res_ctx->pipe_ctx[i].stream->timing.display_color_depth) {
+		case COLOR_DEPTH_666:
+			output_bpc = 6;
+			break;
+		case COLOR_DEPTH_888:
+			output_bpc = 8;
+			break;
+		case COLOR_DEPTH_101010:
+			output_bpc = 10;
+			break;
+		case COLOR_DEPTH_121212:
+			output_bpc = 12;
+			break;
+		case COLOR_DEPTH_141414:
+			output_bpc = 14;
+			break;
+		case COLOR_DEPTH_161616:
+			output_bpc = 16;
+			break;
+		case COLOR_DEPTH_999:
+			output_bpc = 9;
+			break;
+		case COLOR_DEPTH_111111:
+			output_bpc = 11;
+			break;
+		default:
+			output_bpc = 8;
+			break;
+		}
+
+		switch (res_ctx->pipe_ctx[i].stream->timing.pixel_encoding) {
+		case PIXEL_ENCODING_RGB:
+		case PIXEL_ENCODING_YCBCR444:
+			pipes[pipe_cnt].dout.output_format = dm_444;
+			pipes[pipe_cnt].dout.output_bpp = output_bpc * 3;
+			break;
+		case PIXEL_ENCODING_YCBCR420:
+			pipes[pipe_cnt].dout.output_format = dm_420;
+			pipes[pipe_cnt].dout.output_bpp = (output_bpc * 3.0) / 2;
+			break;
+		case PIXEL_ENCODING_YCBCR422:
+			if (res_ctx->pipe_ctx[i].stream->timing.flags.DSC &&
+			    !res_ctx->pipe_ctx[i].stream->timing.dsc_cfg.ycbcr422_simple)
+				pipes[pipe_cnt].dout.output_format = dm_n422;
+			else
+				pipes[pipe_cnt].dout.output_format = dm_s422;
+			pipes[pipe_cnt].dout.output_bpp = output_bpc * 2;
+			break;
+		default:
+			pipes[pipe_cnt].dout.output_format = dm_444;
+			pipes[pipe_cnt].dout.output_bpp = output_bpc * 3;
+		}
+
+		if (res_ctx->pipe_ctx[i].stream->timing.flags.DSC)
+			pipes[pipe_cnt].dout.output_bpp = res_ctx->pipe_ctx[i].stream->timing.dsc_cfg.bits_per_pixel / 16.0;
+
+		/* todo: default max for now, until there is logic reflecting this in dc*/
+		pipes[pipe_cnt].dout.dsc_input_bpc = 12;
+		/*fill up the audio sample rate (unit in kHz)*/
+		get_audio_check(&res_ctx->pipe_ctx[i].stream->audio_info, &aud_check);
+		pipes[pipe_cnt].dout.max_audio_sample_rate = aud_check.max_audiosample_rate / 1000;
+		/*
+		 * For graphic plane, cursor number is 1, nv12 is 0
+		 * bw calculations due to cursor on/off
+		 */
+		if (res_ctx->pipe_ctx[i].plane_state &&
+				(res_ctx->pipe_ctx[i].plane_state->address.type == PLN_ADDR_TYPE_VIDEO_PROGRESSIVE ||
+				dc_state_get_pipe_subvp_type(context, &res_ctx->pipe_ctx[i]) == SUBVP_PHANTOM))
+			pipes[pipe_cnt].pipe.src.num_cursors = 0;
+		else
+			pipes[pipe_cnt].pipe.src.num_cursors = dc->dml.ip.number_of_cursors;
+
+		pipes[pipe_cnt].pipe.src.cur0_src_width = 256;
+		pipes[pipe_cnt].pipe.src.cur0_bpp = dm_cur_32bit;
+
+		if (!res_ctx->pipe_ctx[i].plane_state) {
+			pipes[pipe_cnt].pipe.src.is_hsplit = pipes[pipe_cnt].pipe.dest.odm_combine != dm_odm_combine_mode_disabled;
+			pipes[pipe_cnt].pipe.src.source_scan = dm_horz;
+			pipes[pipe_cnt].pipe.src.source_rotation = dm_rotation_0;
+			pipes[pipe_cnt].pipe.src.sw_mode = dm_sw_4kb_s;
+			pipes[pipe_cnt].pipe.src.macro_tile_size = dm_64k_tile;
+			pipes[pipe_cnt].pipe.src.viewport_width = timing->h_addressable;
+			if (pipes[pipe_cnt].pipe.src.viewport_width > 1920)
+				pipes[pipe_cnt].pipe.src.viewport_width = 1920;
+			pipes[pipe_cnt].pipe.src.viewport_height = timing->v_addressable;
+			if (pipes[pipe_cnt].pipe.src.viewport_height > 1080)
+				pipes[pipe_cnt].pipe.src.viewport_height = 1080;
+			pipes[pipe_cnt].pipe.src.surface_height_y = pipes[pipe_cnt].pipe.src.viewport_height;
+			pipes[pipe_cnt].pipe.src.surface_width_y = pipes[pipe_cnt].pipe.src.viewport_width;
+			pipes[pipe_cnt].pipe.src.surface_height_c = pipes[pipe_cnt].pipe.src.viewport_height;
+			pipes[pipe_cnt].pipe.src.surface_width_c = pipes[pipe_cnt].pipe.src.viewport_width;
+			pipes[pipe_cnt].pipe.src.data_pitch = ((pipes[pipe_cnt].pipe.src.viewport_width + 255) / 256) * 256;
+			pipes[pipe_cnt].pipe.src.source_format = dm_444_32;
+			pipes[pipe_cnt].pipe.src.cur0_src_width = 0;
+			pipes[pipe_cnt].pipe.src.cur1_src_width = 0;
+			pipes[pipe_cnt].pipe.dest.recout_width = pipes[pipe_cnt].pipe.src.viewport_width; /*vp_width/hratio*/
+			pipes[pipe_cnt].pipe.dest.recout_height = pipes[pipe_cnt].pipe.src.viewport_height; /*vp_height/vratio*/
+			pipes[pipe_cnt].pipe.dest.full_recout_width = pipes[pipe_cnt].pipe.dest.recout_width;  /*when is_hsplit != 1*/
+			pipes[pipe_cnt].pipe.dest.full_recout_height = pipes[pipe_cnt].pipe.dest.recout_height; /*when is_hsplit != 1*/
+			pipes[pipe_cnt].pipe.scale_ratio_depth.lb_depth = dm_lb_16;
+			pipes[pipe_cnt].pipe.scale_ratio_depth.hscl_ratio = 1.0;
+			pipes[pipe_cnt].pipe.scale_ratio_depth.vscl_ratio = 1.0;
+			pipes[pipe_cnt].pipe.scale_ratio_depth.scl_enable = 0; /*Lb only or Full scl*/
+			pipes[pipe_cnt].pipe.scale_taps.htaps = 1;
+			pipes[pipe_cnt].pipe.scale_taps.vtaps = 1;
+			pipes[pipe_cnt].pipe.dest.vtotal_min = v_total;
+			pipes[pipe_cnt].pipe.dest.vtotal_max = v_total;
+
+			if (pipes[pipe_cnt].pipe.dest.odm_combine == dm_odm_combine_mode_2to1) {
+				pipes[pipe_cnt].pipe.src.viewport_width /= 2;
+				pipes[pipe_cnt].pipe.dest.recout_width /= 2;
+			} else if (pipes[pipe_cnt].pipe.dest.odm_combine == dm_odm_combine_mode_4to1) {
+				pipes[pipe_cnt].pipe.src.viewport_width /= 4;
+				pipes[pipe_cnt].pipe.dest.recout_width /= 4;
+			}
+		} else {
+			struct dc_plane_state *pln = res_ctx->pipe_ctx[i].plane_state;
+			struct scaler_data *scl = &res_ctx->pipe_ctx[i].plane_res.scl_data;
+
+			pipes[pipe_cnt].pipe.src.immediate_flip = pln->flip_immediate;
+			pipes[pipe_cnt].pipe.src.is_hsplit = (res_ctx->pipe_ctx[i].bottom_pipe && res_ctx->pipe_ctx[i].bottom_pipe->plane_state == pln)
+					|| (res_ctx->pipe_ctx[i].top_pipe && res_ctx->pipe_ctx[i].top_pipe->plane_state == pln)
+					|| pipes[pipe_cnt].pipe.dest.odm_combine != dm_odm_combine_mode_disabled;
+
+			/* stereo is not split */
+			if (pln->stereo_format == PLANE_STEREO_FORMAT_SIDE_BY_SIDE ||
+			    pln->stereo_format == PLANE_STEREO_FORMAT_TOP_AND_BOTTOM) {
+				pipes[pipe_cnt].pipe.src.is_hsplit = false;
+				pipes[pipe_cnt].pipe.src.hsplit_grp = res_ctx->pipe_ctx[i].pipe_idx;
+			}
+
+			pipes[pipe_cnt].pipe.src.source_scan = pln->rotation == ROTATION_ANGLE_90
+					|| pln->rotation == ROTATION_ANGLE_270 ? dm_vert : dm_horz;
+			switch (pln->rotation) {
+			case ROTATION_ANGLE_0:
+				pipes[pipe_cnt].pipe.src.source_rotation = dm_rotation_0;
+				break;
+			case ROTATION_ANGLE_90:
+				pipes[pipe_cnt].pipe.src.source_rotation = dm_rotation_90;
+				break;
+			case ROTATION_ANGLE_180:
+				pipes[pipe_cnt].pipe.src.source_rotation = dm_rotation_180;
+				break;
+			case ROTATION_ANGLE_270:
+				pipes[pipe_cnt].pipe.src.source_rotation = dm_rotation_270;
+				break;
+			default:
+				break;
+			}
+
+			pipes[pipe_cnt].pipe.src.viewport_y_y = scl->viewport.y;
+			pipes[pipe_cnt].pipe.src.viewport_y_c = scl->viewport_c.y;
+			pipes[pipe_cnt].pipe.src.viewport_x_y = scl->viewport.x;
+			pipes[pipe_cnt].pipe.src.viewport_x_c = scl->viewport_c.x;
+			pipes[pipe_cnt].pipe.src.viewport_width = scl->viewport.width;
+			pipes[pipe_cnt].pipe.src.viewport_width_c = scl->viewport_c.width;
+			pipes[pipe_cnt].pipe.src.viewport_height = scl->viewport.height;
+			pipes[pipe_cnt].pipe.src.viewport_height_c = scl->viewport_c.height;
+			pipes[pipe_cnt].pipe.src.viewport_width_max = pln->src_rect.width;
+			pipes[pipe_cnt].pipe.src.viewport_height_max = pln->src_rect.height;
+			pipes[pipe_cnt].pipe.src.surface_width_y = pln->plane_size.surface_size.width;
+			pipes[pipe_cnt].pipe.src.surface_height_y = pln->plane_size.surface_size.height;
+			pipes[pipe_cnt].pipe.src.surface_width_c = pln->plane_size.chroma_size.width;
+			pipes[pipe_cnt].pipe.src.surface_height_c = pln->plane_size.chroma_size.height;
+			if (pln->format == SURFACE_PIXEL_FORMAT_GRPH_RGBE_ALPHA
+					|| pln->format >= SURFACE_PIXEL_FORMAT_VIDEO_BEGIN) {
+				pipes[pipe_cnt].pipe.src.data_pitch = pln->plane_size.surface_pitch;
+				pipes[pipe_cnt].pipe.src.data_pitch_c = pln->plane_size.chroma_pitch;
+				pipes[pipe_cnt].pipe.src.meta_pitch = pln->dcc.meta_pitch;
+				pipes[pipe_cnt].pipe.src.meta_pitch_c = pln->dcc.meta_pitch_c;
+			} else {
+				pipes[pipe_cnt].pipe.src.data_pitch = pln->plane_size.surface_pitch;
+				pipes[pipe_cnt].pipe.src.meta_pitch = pln->dcc.meta_pitch;
+			}
+			pipes[pipe_cnt].pipe.src.dcc = pln->dcc.enable;
+			pipes[pipe_cnt].pipe.dest.recout_width = scl->recout.width;
+			pipes[pipe_cnt].pipe.dest.recout_height = scl->recout.height;
+			pipes[pipe_cnt].pipe.dest.full_recout_height = scl->recout.height;
+			pipes[pipe_cnt].pipe.dest.full_recout_width = scl->recout.width;
+			if (pipes[pipe_cnt].pipe.dest.odm_combine == dm_odm_combine_mode_2to1)
+				pipes[pipe_cnt].pipe.dest.full_recout_width *= 2;
+			else if (pipes[pipe_cnt].pipe.dest.odm_combine == dm_odm_combine_mode_4to1)
+				pipes[pipe_cnt].pipe.dest.full_recout_width *= 4;
+			else {
+				struct pipe_ctx *split_pipe = res_ctx->pipe_ctx[i].bottom_pipe;
+
+				while (split_pipe && split_pipe->plane_state == pln) {
+					pipes[pipe_cnt].pipe.dest.full_recout_width += split_pipe->plane_res.scl_data.recout.width;
+					split_pipe = split_pipe->bottom_pipe;
+				}
+				split_pipe = res_ctx->pipe_ctx[i].top_pipe;
+				while (split_pipe && split_pipe->plane_state == pln) {
+					pipes[pipe_cnt].pipe.dest.full_recout_width += split_pipe->plane_res.scl_data.recout.width;
+					split_pipe = split_pipe->top_pipe;
+				}
+			}
+
+			pipes[pipe_cnt].pipe.scale_ratio_depth.lb_depth = dm_lb_16;
+			pipes[pipe_cnt].pipe.scale_ratio_depth.hscl_ratio = (double) scl->ratios.horz.value / (1ULL<<32);
+			pipes[pipe_cnt].pipe.scale_ratio_depth.hscl_ratio_c = (double) scl->ratios.horz_c.value / (1ULL<<32);
+			pipes[pipe_cnt].pipe.scale_ratio_depth.vscl_ratio = (double) scl->ratios.vert.value / (1ULL<<32);
+			pipes[pipe_cnt].pipe.scale_ratio_depth.vscl_ratio_c = (double) scl->ratios.vert_c.value / (1ULL<<32);
+			pipes[pipe_cnt].pipe.scale_ratio_depth.scl_enable =
+					scl->ratios.vert.value != dc_fixpt_one.value
+					|| scl->ratios.horz.value != dc_fixpt_one.value
+					|| scl->ratios.vert_c.value != dc_fixpt_one.value
+					|| scl->ratios.horz_c.value != dc_fixpt_one.value /*Lb only or Full scl*/
+					|| dc->debug.always_scale; /*support always scale*/
+			pipes[pipe_cnt].pipe.scale_taps.htaps = scl->taps.h_taps;
+			pipes[pipe_cnt].pipe.scale_taps.htaps_c = scl->taps.h_taps_c;
+			pipes[pipe_cnt].pipe.scale_taps.vtaps = scl->taps.v_taps;
+			pipes[pipe_cnt].pipe.scale_taps.vtaps_c = scl->taps.v_taps_c;
+
+			pipes[pipe_cnt].pipe.src.macro_tile_size =
+					swizzle_mode_to_macro_tile_size(pln->tiling_info.gfx9.swizzle);
+			swizzle_to_dml_params(pln->tiling_info.gfx9.swizzle,
+					&pipes[pipe_cnt].pipe.src.sw_mode);
+
+			switch (pln->format) {
+			case SURFACE_PIXEL_FORMAT_VIDEO_420_YCbCr:
+			case SURFACE_PIXEL_FORMAT_VIDEO_420_YCrCb:
+				pipes[pipe_cnt].pipe.src.source_format = dm_420_8;
+				break;
+			case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCbCr:
+			case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCrCb:
+				pipes[pipe_cnt].pipe.src.source_format = dm_420_10;
+				break;
+			case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616:
+			case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616:
+			case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616F:
+			case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F:
+				pipes[pipe_cnt].pipe.src.source_format = dm_444_64;
+				break;
+			case SURFACE_PIXEL_FORMAT_GRPH_ARGB1555:
+			case SURFACE_PIXEL_FORMAT_GRPH_RGB565:
+				pipes[pipe_cnt].pipe.src.source_format = dm_444_16;
+				break;
+			case SURFACE_PIXEL_FORMAT_GRPH_PALETA_256_COLORS:
+				pipes[pipe_cnt].pipe.src.source_format = dm_444_8;
+				break;
+			case SURFACE_PIXEL_FORMAT_GRPH_RGBE_ALPHA:
+				pipes[pipe_cnt].pipe.src.source_format = dm_rgbe_alpha;
+				break;
+			default:
+				pipes[pipe_cnt].pipe.src.source_format = dm_444_32;
+				break;
+			}
+		}
+
+		pipe_cnt++;
+	}
+
+	/* populate writeback information */
+	dc->res_pool->funcs->populate_dml_writeback_from_context(dc, res_ctx, pipes);
+
+	return pipe_cnt;
+}
+
+void dcn20_calculate_wm(struct dc *dc, struct dc_state *context,
+			display_e2e_pipe_params_st *pipes,
+			int *out_pipe_cnt,
+			int *pipe_split_from,
+			int vlevel,
+			enum dc_validate_mode validate_mode)
+{
+	int pipe_cnt, i, pipe_idx;
+
+	dc_assert_fp_enabled();
+
+	for (i = 0, pipe_idx = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) {
+		if (!context->res_ctx.pipe_ctx[i].stream)
+			continue;
+
+		pipes[pipe_cnt].clks_cfg.refclk_mhz = dc->res_pool->ref_clocks.dchub_ref_clock_inKhz / 1000.0;
+		pipes[pipe_cnt].clks_cfg.dispclk_mhz = context->bw_ctx.dml.vba.RequiredDISPCLK[vlevel][context->bw_ctx.dml.vba.maxMpcComb];
+
+		if (pipe_split_from[i] < 0) {
+			pipes[pipe_cnt].clks_cfg.dppclk_mhz =
+					context->bw_ctx.dml.vba.RequiredDPPCLK[vlevel][context->bw_ctx.dml.vba.maxMpcComb][pipe_idx];
+			if (context->bw_ctx.dml.vba.BlendingAndTiming[pipe_idx] == pipe_idx)
+				pipes[pipe_cnt].pipe.dest.odm_combine =
+						context->bw_ctx.dml.vba.ODMCombineEnabled[pipe_idx];
+			else
+				pipes[pipe_cnt].pipe.dest.odm_combine = 0;
+			pipe_idx++;
+		} else {
+			pipes[pipe_cnt].clks_cfg.dppclk_mhz =
+					context->bw_ctx.dml.vba.RequiredDPPCLK[vlevel][context->bw_ctx.dml.vba.maxMpcComb][pipe_split_from[i]];
+			if (context->bw_ctx.dml.vba.BlendingAndTiming[pipe_split_from[i]] == pipe_split_from[i])
+				pipes[pipe_cnt].pipe.dest.odm_combine =
+						context->bw_ctx.dml.vba.ODMCombineEnabled[pipe_split_from[i]];
+			else
+				pipes[pipe_cnt].pipe.dest.odm_combine = 0;
+		}
+
+		if (dc->config.forced_clocks) {
+			pipes[pipe_cnt].clks_cfg.dispclk_mhz = context->bw_ctx.dml.soc.clock_limits[0].dispclk_mhz;
+			pipes[pipe_cnt].clks_cfg.dppclk_mhz = context->bw_ctx.dml.soc.clock_limits[0].dppclk_mhz;
+		}
+		if (dc->debug.min_disp_clk_khz > pipes[pipe_cnt].clks_cfg.dispclk_mhz * 1000)
+			pipes[pipe_cnt].clks_cfg.dispclk_mhz = dc->debug.min_disp_clk_khz / 1000.0;
+		if (dc->debug.min_dpp_clk_khz > pipes[pipe_cnt].clks_cfg.dppclk_mhz * 1000)
+			pipes[pipe_cnt].clks_cfg.dppclk_mhz = dc->debug.min_dpp_clk_khz / 1000.0;
+
+		pipe_cnt++;
+	}
+
+	if (pipe_cnt != pipe_idx) {
+		if (dc->res_pool->funcs->populate_dml_pipes)
+			pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc,
+				context, pipes, validate_mode);
+		else
+			pipe_cnt = dcn20_populate_dml_pipes_from_context(dc,
+				context, pipes, validate_mode);
+	}
+
+	*out_pipe_cnt = pipe_cnt;
+
+	pipes[0].clks_cfg.voltage = vlevel;
+	pipes[0].clks_cfg.dcfclk_mhz = context->bw_ctx.dml.soc.clock_limits[vlevel].dcfclk_mhz;
+	pipes[0].clks_cfg.socclk_mhz = context->bw_ctx.dml.soc.clock_limits[vlevel].socclk_mhz;
+
+	/* only pipe 0 is read for voltage and dcf/soc clocks */
+	if (vlevel < 1) {
+		pipes[0].clks_cfg.voltage = 1;
+		pipes[0].clks_cfg.dcfclk_mhz = context->bw_ctx.dml.soc.clock_limits[1].dcfclk_mhz;
+		pipes[0].clks_cfg.socclk_mhz = context->bw_ctx.dml.soc.clock_limits[1].socclk_mhz;
+	}
+	context->bw_ctx.bw.dcn.watermarks.b.urgent_ns = get_wm_urgent(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.b.cstate_pstate.cstate_enter_plus_exit_ns = get_wm_stutter_enter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.b.cstate_pstate.cstate_exit_ns = get_wm_stutter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.b.cstate_pstate.pstate_change_ns = get_wm_dram_clock_change(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.b.pte_meta_urgent_ns = get_wm_memory_trip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.b.frac_urg_bw_nom = get_fraction_of_urgent_bandwidth(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.b.frac_urg_bw_flip = get_fraction_of_urgent_bandwidth_imm_flip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.b.urgent_latency_ns = get_urgent_latency(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+
+	if (vlevel < 2) {
+		pipes[0].clks_cfg.voltage = 2;
+		pipes[0].clks_cfg.dcfclk_mhz = context->bw_ctx.dml.soc.clock_limits[2].dcfclk_mhz;
+		pipes[0].clks_cfg.socclk_mhz = context->bw_ctx.dml.soc.clock_limits[2].socclk_mhz;
+	}
+	context->bw_ctx.bw.dcn.watermarks.c.urgent_ns = get_wm_urgent(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.c.cstate_pstate.cstate_enter_plus_exit_ns = get_wm_stutter_enter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.c.cstate_pstate.cstate_exit_ns = get_wm_stutter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.c.cstate_pstate.pstate_change_ns = get_wm_dram_clock_change(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.c.pte_meta_urgent_ns = get_wm_memory_trip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.c.frac_urg_bw_nom = get_fraction_of_urgent_bandwidth(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.c.frac_urg_bw_flip = get_fraction_of_urgent_bandwidth_imm_flip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+
+	if (vlevel < 3) {
+		pipes[0].clks_cfg.voltage = 3;
+		pipes[0].clks_cfg.dcfclk_mhz = context->bw_ctx.dml.soc.clock_limits[2].dcfclk_mhz;
+		pipes[0].clks_cfg.socclk_mhz = context->bw_ctx.dml.soc.clock_limits[2].socclk_mhz;
+	}
+	context->bw_ctx.bw.dcn.watermarks.d.urgent_ns = get_wm_urgent(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.d.cstate_pstate.cstate_enter_plus_exit_ns = get_wm_stutter_enter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.d.cstate_pstate.cstate_exit_ns = get_wm_stutter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.d.cstate_pstate.pstate_change_ns = get_wm_dram_clock_change(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.d.pte_meta_urgent_ns = get_wm_memory_trip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.d.frac_urg_bw_nom = get_fraction_of_urgent_bandwidth(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.d.frac_urg_bw_flip = get_fraction_of_urgent_bandwidth_imm_flip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+
+	pipes[0].clks_cfg.voltage = vlevel;
+	pipes[0].clks_cfg.dcfclk_mhz = context->bw_ctx.dml.soc.clock_limits[vlevel].dcfclk_mhz;
+	pipes[0].clks_cfg.socclk_mhz = context->bw_ctx.dml.soc.clock_limits[vlevel].socclk_mhz;
+	context->bw_ctx.bw.dcn.watermarks.a.urgent_ns = get_wm_urgent(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.cstate_enter_plus_exit_ns = get_wm_stutter_enter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.cstate_exit_ns = get_wm_stutter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.pstate_change_ns = get_wm_dram_clock_change(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.a.pte_meta_urgent_ns = get_wm_memory_trip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.a.frac_urg_bw_nom = get_fraction_of_urgent_bandwidth(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+	context->bw_ctx.bw.dcn.watermarks.a.frac_urg_bw_flip = get_fraction_of_urgent_bandwidth_imm_flip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
+}
+
+void dcn20_update_bounding_box(struct dc *dc,
+			       struct _vcs_dpi_soc_bounding_box_st *bb,
+			       struct pp_smu_nv_clock_table *max_clocks,
+			       unsigned int *uclk_states,
+			       unsigned int num_states)
+{
+	int num_calculated_states = 0;
+	int min_dcfclk = 0;
+	int i;
+
+	dc_assert_fp_enabled();
+
+	if (num_states == 0)
+		return;
+
+	memset(bb->clock_limits, 0, sizeof(bb->clock_limits));
+
+	if (dc->bb_overrides.min_dcfclk_mhz > 0) {
+		min_dcfclk = dc->bb_overrides.min_dcfclk_mhz;
+	} else {
+		if (ASICREV_IS_NAVI12_P(dc->ctx->asic_id.hw_internal_rev))
+			min_dcfclk = 310;
+		else
+			// Accounting for SOC/DCF relationship, we can go as high as
+			// 506Mhz in Vmin.
+			min_dcfclk = 506;
+	}
+
+	for (i = 0; i < num_states; i++) {
+		int min_fclk_required_by_uclk;
+		bb->clock_limits[i].state = i;
+		bb->clock_limits[i].dram_speed_mts = uclk_states[i] * 16 / 1000;
+
+		// FCLK:UCLK ratio is 1.08
+		min_fclk_required_by_uclk = div_u64(((unsigned long long)uclk_states[i]) * 1080,
+			1000000);
+
+		bb->clock_limits[i].fabricclk_mhz = (min_fclk_required_by_uclk < min_dcfclk) ?
+				min_dcfclk : min_fclk_required_by_uclk;
+
+		bb->clock_limits[i].socclk_mhz = (bb->clock_limits[i].fabricclk_mhz > max_clocks->socClockInKhz / 1000.0) ?
+				max_clocks->socClockInKhz / 1000 : bb->clock_limits[i].fabricclk_mhz;
+
+		bb->clock_limits[i].dcfclk_mhz = (bb->clock_limits[i].fabricclk_mhz > max_clocks->dcfClockInKhz / 1000.0) ?
+				max_clocks->dcfClockInKhz / 1000 : bb->clock_limits[i].fabricclk_mhz;
+
+		bb->clock_limits[i].dispclk_mhz = max_clocks->displayClockInKhz / 1000;
+		bb->clock_limits[i].dppclk_mhz = max_clocks->displayClockInKhz / 1000;
+		bb->clock_limits[i].dscclk_mhz = max_clocks->displayClockInKhz / (1000 * 3);
+
+		bb->clock_limits[i].phyclk_mhz = max_clocks->phyClockInKhz / 1000;
+
+		num_calculated_states++;
+	}
+
+	bb->clock_limits[num_calculated_states - 1].socclk_mhz = max_clocks->socClockInKhz / 1000;
+	bb->clock_limits[num_calculated_states - 1].fabricclk_mhz = max_clocks->socClockInKhz / 1000;
+	bb->clock_limits[num_calculated_states - 1].dcfclk_mhz = max_clocks->dcfClockInKhz / 1000;
+
+	bb->num_states = num_calculated_states;
+
+	// Duplicate the last state, DML always an extra state identical to max state to work
+	memcpy(&bb->clock_limits[num_calculated_states], &bb->clock_limits[num_calculated_states - 1], sizeof(struct _vcs_dpi_voltage_scaling_st));
+	bb->clock_limits[num_calculated_states].state = bb->num_states;
+}
+
+void dcn20_cap_soc_clocks(struct _vcs_dpi_soc_bounding_box_st *bb,
+			  struct pp_smu_nv_clock_table max_clocks)
+{
+	int i;
+
+	dc_assert_fp_enabled();
+
+	// First pass - cap all clocks higher than the reported max
+	for (i = 0; i < bb->num_states; i++) {
+		if ((bb->clock_limits[i].dcfclk_mhz > (max_clocks.dcfClockInKhz / 1000.0))
+				&& max_clocks.dcfClockInKhz != 0)
+			bb->clock_limits[i].dcfclk_mhz = (max_clocks.dcfClockInKhz / 1000);
+
+		if ((bb->clock_limits[i].dram_speed_mts > (max_clocks.uClockInKhz / 1000.0) * 16)
+						&& max_clocks.uClockInKhz != 0)
+			bb->clock_limits[i].dram_speed_mts = (max_clocks.uClockInKhz / 1000) * 16;
+
+		if ((bb->clock_limits[i].fabricclk_mhz > (max_clocks.fabricClockInKhz / 1000.0))
+						&& max_clocks.fabricClockInKhz != 0)
+			bb->clock_limits[i].fabricclk_mhz = (max_clocks.fabricClockInKhz / 1000);
+
+		if ((bb->clock_limits[i].dispclk_mhz > (max_clocks.displayClockInKhz / 1000.0))
+						&& max_clocks.displayClockInKhz != 0)
+			bb->clock_limits[i].dispclk_mhz = (max_clocks.displayClockInKhz / 1000);
+
+		if ((bb->clock_limits[i].dppclk_mhz > (max_clocks.dppClockInKhz / 1000.0))
+						&& max_clocks.dppClockInKhz != 0)
+			bb->clock_limits[i].dppclk_mhz = (max_clocks.dppClockInKhz / 1000);
+
+		if ((bb->clock_limits[i].phyclk_mhz > (max_clocks.phyClockInKhz / 1000.0))
+						&& max_clocks.phyClockInKhz != 0)
+			bb->clock_limits[i].phyclk_mhz = (max_clocks.phyClockInKhz / 1000);
+
+		if ((bb->clock_limits[i].socclk_mhz > (max_clocks.socClockInKhz / 1000.0))
+						&& max_clocks.socClockInKhz != 0)
+			bb->clock_limits[i].socclk_mhz = (max_clocks.socClockInKhz / 1000);
+
+		if ((bb->clock_limits[i].dscclk_mhz > (max_clocks.dscClockInKhz / 1000.0))
+						&& max_clocks.dscClockInKhz != 0)
+			bb->clock_limits[i].dscclk_mhz = (max_clocks.dscClockInKhz / 1000);
+	}
+
+	// Second pass - remove all duplicate clock states
+	for (i = bb->num_states - 1; i > 1; i--) {
+		bool duplicate = true;
+
+		if (bb->clock_limits[i-1].dcfclk_mhz != bb->clock_limits[i].dcfclk_mhz)
+			duplicate = false;
+		if (bb->clock_limits[i-1].dispclk_mhz != bb->clock_limits[i].dispclk_mhz)
+			duplicate = false;
+		if (bb->clock_limits[i-1].dppclk_mhz != bb->clock_limits[i].dppclk_mhz)
+			duplicate = false;
+		if (bb->clock_limits[i-1].dram_speed_mts != bb->clock_limits[i].dram_speed_mts)
+			duplicate = false;
+		if (bb->clock_limits[i-1].dscclk_mhz != bb->clock_limits[i].dscclk_mhz)
+			duplicate = false;
+		if (bb->clock_limits[i-1].fabricclk_mhz != bb->clock_limits[i].fabricclk_mhz)
+			duplicate = false;
+		if (bb->clock_limits[i-1].phyclk_mhz != bb->clock_limits[i].phyclk_mhz)
+			duplicate = false;
+		if (bb->clock_limits[i-1].socclk_mhz != bb->clock_limits[i].socclk_mhz)
+			duplicate = false;
+
+		if (duplicate)
+			bb->num_states--;
+	}
+}
+
+void dcn20_patch_bounding_box(struct dc *dc, struct _vcs_dpi_soc_bounding_box_st *bb)
+{
+	dc_assert_fp_enabled();
+
+	if ((int)(bb->sr_exit_time_us * 1000) != dc->bb_overrides.sr_exit_time_ns
+			&& dc->bb_overrides.sr_exit_time_ns) {
+		bb->sr_exit_time_us = dc->bb_overrides.sr_exit_time_ns / 1000.0;
+	}
+
+	if ((int)(bb->sr_enter_plus_exit_time_us * 1000)
+				!= dc->bb_overrides.sr_enter_plus_exit_time_ns
+			&& dc->bb_overrides.sr_enter_plus_exit_time_ns) {
+		bb->sr_enter_plus_exit_time_us =
+				dc->bb_overrides.sr_enter_plus_exit_time_ns / 1000.0;
+	}
+
+	if ((int)(bb->sr_exit_z8_time_us * 1000)
+				!= dc->bb_overrides.sr_exit_z8_time_ns
+			&& dc->bb_overrides.sr_exit_z8_time_ns) {
+		bb->sr_exit_z8_time_us = dc->bb_overrides.sr_exit_z8_time_ns / 1000.0;
+	}
+
+	if ((int)(bb->sr_enter_plus_exit_z8_time_us * 1000)
+				!= dc->bb_overrides.sr_enter_plus_exit_z8_time_ns
+			&& dc->bb_overrides.sr_enter_plus_exit_z8_time_ns) {
+		bb->sr_enter_plus_exit_z8_time_us = dc->bb_overrides.sr_enter_plus_exit_z8_time_ns / 1000.0;
+	}
+	if ((int)(bb->urgent_latency_us * 1000) != dc->bb_overrides.urgent_latency_ns
+			&& dc->bb_overrides.urgent_latency_ns) {
+		bb->urgent_latency_us = dc->bb_overrides.urgent_latency_ns / 1000.0;
+	}
+
+	if ((int)(bb->dram_clock_change_latency_us * 1000)
+				!= dc->bb_overrides.dram_clock_change_latency_ns
+			&& dc->bb_overrides.dram_clock_change_latency_ns) {
+		bb->dram_clock_change_latency_us =
+				dc->bb_overrides.dram_clock_change_latency_ns / 1000.0;
+	}
+
+	if ((int)(bb->dummy_pstate_latency_us * 1000)
+				!= dc->bb_overrides.dummy_clock_change_latency_ns
+			&& dc->bb_overrides.dummy_clock_change_latency_ns) {
+		bb->dummy_pstate_latency_us =
+				dc->bb_overrides.dummy_clock_change_latency_ns / 1000.0;
+	}
+}
+
+static bool dcn20_validate_bandwidth_internal(struct dc *dc, struct dc_state *context,
+		enum dc_validate_mode validate_mode, display_e2e_pipe_params_st *pipes)
+{
+	bool out = false;
+
+	BW_VAL_TRACE_SETUP();
+
+	int vlevel = 0;
+	int pipe_split_from[MAX_PIPES];
+	int pipe_cnt = 0;
+	DC_LOGGER_INIT(dc->ctx->logger);
+
+	BW_VAL_TRACE_COUNT();
+
+	out = dcn20_fast_validate_bw(dc, context, pipes, &pipe_cnt, pipe_split_from, &vlevel, validate_mode);
+
+	if (pipe_cnt == 0)
+		goto validate_out;
+
+	if (!out)
+		goto validate_fail;
+
+	BW_VAL_TRACE_END_VOLTAGE_LEVEL();
+
+	if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING) {
+		BW_VAL_TRACE_SKIP(fast);
+		goto validate_out;
+	}
+
+	dcn20_calculate_wm(dc, context, pipes, &pipe_cnt, pipe_split_from, vlevel, validate_mode);
+	dcn20_calculate_dlg_params(dc, context, pipes, pipe_cnt, vlevel);
+
+	BW_VAL_TRACE_END_WATERMARKS();
+
+	goto validate_out;
+
+validate_fail:
+	DC_LOG_WARNING("Mode Validation Warning: %s failed validation.\n",
+		dml_get_status_message(context->bw_ctx.dml.vba.ValidationStatus[context->bw_ctx.dml.vba.soc.num_states]));
+
+	BW_VAL_TRACE_SKIP(fail);
+	out = false;
+
+validate_out:
+
+	BW_VAL_TRACE_FINISH();
+
+	return out;
+}
+
+bool dcn20_validate_bandwidth_fp(struct dc *dc, struct dc_state *context,
+				 enum dc_validate_mode validate_mode, display_e2e_pipe_params_st *pipes)
+{
+	bool voltage_supported = false;
+	bool full_pstate_supported = false;
+	bool dummy_pstate_supported = false;
+	double p_state_latency_us;
+
+	dc_assert_fp_enabled();
+
+	p_state_latency_us = context->bw_ctx.dml.soc.dram_clock_change_latency_us;
+	context->bw_ctx.dml.soc.disable_dram_clock_change_vactive_support =
+		dc->debug.disable_dram_clock_change_vactive_support;
+	context->bw_ctx.dml.soc.allow_dram_clock_one_display_vactive =
+		dc->debug.enable_dram_clock_change_one_display_vactive;
+
+	/*Unsafe due to current pipe merge and split logic*/
+	ASSERT(context != dc->current_state);
+
+	if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING)
+		return dcn20_validate_bandwidth_internal(dc, context, validate_mode, pipes);
+
+	// Best case, we support full UCLK switch latency
+	voltage_supported = dcn20_validate_bandwidth_internal(dc, context, DC_VALIDATE_MODE_AND_PROGRAMMING, pipes);
+	full_pstate_supported = context->bw_ctx.bw.dcn.clk.p_state_change_support;
+
+	if (context->bw_ctx.dml.soc.dummy_pstate_latency_us == 0 ||
+		(voltage_supported && full_pstate_supported)) {
+		context->bw_ctx.bw.dcn.clk.p_state_change_support = full_pstate_supported;
+		goto restore_dml_state;
+	}
+
+	// Fallback: Try to only support G6 temperature read latency
+	context->bw_ctx.dml.soc.dram_clock_change_latency_us = context->bw_ctx.dml.soc.dummy_pstate_latency_us;
+
+	memset(pipes, 0, dc->res_pool->pipe_count * sizeof(display_e2e_pipe_params_st));
+	voltage_supported = dcn20_validate_bandwidth_internal(dc, context, DC_VALIDATE_MODE_AND_PROGRAMMING, pipes);
+	dummy_pstate_supported = context->bw_ctx.bw.dcn.clk.p_state_change_support;
+
+	if (voltage_supported && (dummy_pstate_supported || !(context->stream_count))) {
+		context->bw_ctx.bw.dcn.clk.p_state_change_support = false;
+		goto restore_dml_state;
+	}
+
+	// ERROR: fallback is supposed to always work.
+	ASSERT(false);
+
+restore_dml_state:
+	context->bw_ctx.dml.soc.dram_clock_change_latency_us = p_state_latency_us;
+	return voltage_supported;
+}
+
+void dcn20_fpu_set_wm_ranges(int i,
+			     struct pp_smu_wm_range_sets *ranges,
+			     struct _vcs_dpi_soc_bounding_box_st *loaded_bb)
+{
+	dc_assert_fp_enabled();
+
+	ranges->reader_wm_sets[i].min_fill_clk_mhz = (i > 0) ? (loaded_bb->clock_limits[i - 1].dram_speed_mts / 16) + 1 : 0;
+	ranges->reader_wm_sets[i].max_fill_clk_mhz = loaded_bb->clock_limits[i].dram_speed_mts / 16;
+}
+
+void dcn20_fpu_adjust_dppclk(struct vba_vars_st *v,
+			     int vlevel,
+			     int max_mpc_comb,
+			     int pipe_idx,
+			     bool is_validating_bw)
+{
+	dc_assert_fp_enabled();
+
+	if (is_validating_bw)
+		v->RequiredDPPCLK[vlevel][max_mpc_comb][pipe_idx] *= 2;
+	else
+		v->RequiredDPPCLK[vlevel][max_mpc_comb][pipe_idx] /= 2;
+}
+
+int dcn21_populate_dml_pipes_from_context(struct dc *dc,
+					  struct dc_state *context,
+					  display_e2e_pipe_params_st *pipes,
+					  enum dc_validate_mode validate_mode)
+{
+	uint32_t pipe_cnt;
+	int i;
+
+	dc_assert_fp_enabled();
+
+	pipe_cnt = dcn20_populate_dml_pipes_from_context(dc, context, pipes, validate_mode);
+
+	for (i = 0; i < pipe_cnt; i++) {
+
+		pipes[i].pipe.src.hostvm = dc->res_pool->hubbub->riommu_active;
+		pipes[i].pipe.src.gpuvm = 1;
+	}
+
+	return pipe_cnt;
+}
+
+static void patch_bounding_box(struct dc *dc, struct _vcs_dpi_soc_bounding_box_st *bb)
+{
+	int i;
+
+	if (dc->bb_overrides.sr_exit_time_ns) {
+		for (i = 0; i < WM_SET_COUNT; i++) {
+			  dc->clk_mgr->bw_params->wm_table.entries[i].sr_exit_time_us =
+					  dc->bb_overrides.sr_exit_time_ns / 1000.0;
+		}
+	}
+
+	if (dc->bb_overrides.sr_enter_plus_exit_time_ns) {
+		for (i = 0; i < WM_SET_COUNT; i++) {
+			  dc->clk_mgr->bw_params->wm_table.entries[i].sr_enter_plus_exit_time_us =
+					  dc->bb_overrides.sr_enter_plus_exit_time_ns / 1000.0;
+		}
+	}
+
+	if (dc->bb_overrides.urgent_latency_ns) {
+		bb->urgent_latency_us = dc->bb_overrides.urgent_latency_ns / 1000.0;
+	}
+
+	if (dc->bb_overrides.dram_clock_change_latency_ns) {
+		for (i = 0; i < WM_SET_COUNT; i++) {
+			dc->clk_mgr->bw_params->wm_table.entries[i].pstate_latency_us =
+				dc->bb_overrides.dram_clock_change_latency_ns / 1000.0;
+		}
+	}
+}
+
+static void calculate_wm_set_for_vlevel(int vlevel,
+					struct wm_range_table_entry *table_entry,
+					struct dcn_watermarks *wm_set,
+					struct display_mode_lib *dml,
+					display_e2e_pipe_params_st *pipes,
+					int pipe_cnt)
+{
+	double dram_clock_change_latency_cached = dml->soc.dram_clock_change_latency_us;
+
+	ASSERT(vlevel < dml->soc.num_states);
+	/* only pipe 0 is read for voltage and dcf/soc clocks */
+	pipes[0].clks_cfg.voltage = vlevel;
+	pipes[0].clks_cfg.dcfclk_mhz = dml->soc.clock_limits[vlevel].dcfclk_mhz;
+	pipes[0].clks_cfg.socclk_mhz = dml->soc.clock_limits[vlevel].socclk_mhz;
+
+	dml->soc.dram_clock_change_latency_us = table_entry->pstate_latency_us;
+	dml->soc.sr_exit_time_us = table_entry->sr_exit_time_us;
+	dml->soc.sr_enter_plus_exit_time_us = table_entry->sr_enter_plus_exit_time_us;
+
+	wm_set->urgent_ns = get_wm_urgent(dml, pipes, pipe_cnt) * 1000;
+	wm_set->cstate_pstate.cstate_enter_plus_exit_ns = get_wm_stutter_enter_exit(dml, pipes, pipe_cnt) * 1000;
+	wm_set->cstate_pstate.cstate_exit_ns = get_wm_stutter_exit(dml, pipes, pipe_cnt) * 1000;
+	wm_set->cstate_pstate.pstate_change_ns = get_wm_dram_clock_change(dml, pipes, pipe_cnt) * 1000;
+	wm_set->pte_meta_urgent_ns = get_wm_memory_trip(dml, pipes, pipe_cnt) * 1000;
+	wm_set->frac_urg_bw_nom = get_fraction_of_urgent_bandwidth(dml, pipes, pipe_cnt) * 1000;
+	wm_set->frac_urg_bw_flip = get_fraction_of_urgent_bandwidth_imm_flip(dml, pipes, pipe_cnt) * 1000;
+	wm_set->urgent_latency_ns = get_urgent_latency(dml, pipes, pipe_cnt) * 1000;
+	dml->soc.dram_clock_change_latency_us = dram_clock_change_latency_cached;
+}
+
+static void dcn21_calculate_wm(struct dc *dc, struct dc_state *context,
+			display_e2e_pipe_params_st *pipes,
+			int *out_pipe_cnt,
+			int *pipe_split_from,
+			int vlevel_req,
+			enum dc_validate_mode validate_mode)
+{
+	int pipe_cnt, i, pipe_idx;
+	int vlevel, vlevel_max;
+	struct wm_range_table_entry *table_entry;
+	struct clk_bw_params *bw_params = dc->clk_mgr->bw_params;
+
+	ASSERT(bw_params);
+
+	patch_bounding_box(dc, &context->bw_ctx.dml.soc);
+
+	for (i = 0, pipe_idx = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) {
+			if (!context->res_ctx.pipe_ctx[i].stream)
+				continue;
+
+			pipes[pipe_cnt].clks_cfg.refclk_mhz = dc->res_pool->ref_clocks.dchub_ref_clock_inKhz / 1000.0;
+			pipes[pipe_cnt].clks_cfg.dispclk_mhz = context->bw_ctx.dml.vba.RequiredDISPCLK[vlevel_req][context->bw_ctx.dml.vba.maxMpcComb];
+
+			if (pipe_split_from[i] < 0) {
+				pipes[pipe_cnt].clks_cfg.dppclk_mhz =
+						context->bw_ctx.dml.vba.RequiredDPPCLK[vlevel_req][context->bw_ctx.dml.vba.maxMpcComb][pipe_idx];
+				if (context->bw_ctx.dml.vba.BlendingAndTiming[pipe_idx] == pipe_idx)
+					pipes[pipe_cnt].pipe.dest.odm_combine =
+							context->bw_ctx.dml.vba.ODMCombineEnablePerState[vlevel_req][pipe_idx];
+				else
+					pipes[pipe_cnt].pipe.dest.odm_combine = 0;
+				pipe_idx++;
+			} else {
+				pipes[pipe_cnt].clks_cfg.dppclk_mhz =
+						context->bw_ctx.dml.vba.RequiredDPPCLK[vlevel_req][context->bw_ctx.dml.vba.maxMpcComb][pipe_split_from[i]];
+				if (context->bw_ctx.dml.vba.BlendingAndTiming[pipe_split_from[i]] == pipe_split_from[i])
+					pipes[pipe_cnt].pipe.dest.odm_combine =
+							context->bw_ctx.dml.vba.ODMCombineEnablePerState[vlevel_req][pipe_split_from[i]];
+				else
+					pipes[pipe_cnt].pipe.dest.odm_combine = 0;
+			}
+			pipe_cnt++;
+	}
+
+	if (pipe_cnt != pipe_idx) {
+		if (dc->res_pool->funcs->populate_dml_pipes)
+			pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc,
+				context, pipes, validate_mode);
+		else
+			pipe_cnt = dcn21_populate_dml_pipes_from_context(dc,
+				context, pipes, validate_mode);
+	}
+
+	*out_pipe_cnt = pipe_cnt;
+
+	vlevel_max = bw_params->clk_table.num_entries - 1;
+
+
+	/* WM Set D */
+	table_entry = &bw_params->wm_table.entries[WM_D];
+	if (table_entry->wm_type == WM_TYPE_RETRAINING)
+		vlevel = 0;
+	else
+		vlevel = vlevel_max;
+	calculate_wm_set_for_vlevel(vlevel, table_entry, &context->bw_ctx.bw.dcn.watermarks.d,
+						&context->bw_ctx.dml, pipes, pipe_cnt);
+	/* WM Set C */
+	table_entry = &bw_params->wm_table.entries[WM_C];
+	vlevel = MIN(MAX(vlevel_req, 3), vlevel_max);
+	calculate_wm_set_for_vlevel(vlevel, table_entry, &context->bw_ctx.bw.dcn.watermarks.c,
+						&context->bw_ctx.dml, pipes, pipe_cnt);
+	/* WM Set B */
+	table_entry = &bw_params->wm_table.entries[WM_B];
+	vlevel = MIN(MAX(vlevel_req, 2), vlevel_max);
+	calculate_wm_set_for_vlevel(vlevel, table_entry, &context->bw_ctx.bw.dcn.watermarks.b,
+						&context->bw_ctx.dml, pipes, pipe_cnt);
+
+	/* WM Set A */
+	table_entry = &bw_params->wm_table.entries[WM_A];
+	vlevel = MIN(vlevel_req, vlevel_max);
+	calculate_wm_set_for_vlevel(vlevel, table_entry, &context->bw_ctx.bw.dcn.watermarks.a,
+						&context->bw_ctx.dml, pipes, pipe_cnt);
+}
+
+bool dcn21_validate_bandwidth_fp(struct dc *dc, struct dc_state *context,
+				 enum dc_validate_mode validate_mode, display_e2e_pipe_params_st *pipes)
+{
+	bool out = false;
+
+	BW_VAL_TRACE_SETUP();
+
+	int vlevel = 0;
+	int pipe_split_from[MAX_PIPES];
+	int pipe_cnt = 0;
+	DC_LOGGER_INIT(dc->ctx->logger);
+
+	BW_VAL_TRACE_COUNT();
+
+	dc_assert_fp_enabled();
+
+	/*Unsafe due to current pipe merge and split logic*/
+	ASSERT(context != dc->current_state);
+
+	out = dcn21_fast_validate_bw(dc, context, pipes, &pipe_cnt, pipe_split_from, &vlevel, validate_mode);
+
+	if (pipe_cnt == 0)
+		goto validate_out;
+
+	if (!out)
+		goto validate_fail;
+
+	BW_VAL_TRACE_END_VOLTAGE_LEVEL();
+
+	if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING) {
+		BW_VAL_TRACE_SKIP(fast);
+		goto validate_out;
+	}
+
+	dcn21_calculate_wm(dc, context, pipes, &pipe_cnt, pipe_split_from, vlevel, validate_mode);
+	dcn20_calculate_dlg_params(dc, context, pipes, pipe_cnt, vlevel);
+
+	BW_VAL_TRACE_END_WATERMARKS();
+
+	goto validate_out;
+
+validate_fail:
+	DC_LOG_WARNING("Mode Validation Warning: %s failed validation.\n",
+			dml_get_status_message(context->bw_ctx.dml.vba.ValidationStatus[context->bw_ctx.dml.vba.soc.num_states]));
+
+	BW_VAL_TRACE_SKIP(fail);
+	out = false;
+
+validate_out:
+
+	BW_VAL_TRACE_FINISH();
+
+	return out;
+}
+
+static struct _vcs_dpi_voltage_scaling_st construct_low_pstate_lvl(struct clk_limit_table *clk_table, unsigned int high_voltage_lvl)
+{
+	struct _vcs_dpi_voltage_scaling_st low_pstate_lvl = {0};
+	int i;
+
+	low_pstate_lvl.state = 1;
+	low_pstate_lvl.dcfclk_mhz = clk_table->entries[0].dcfclk_mhz;
+	low_pstate_lvl.fabricclk_mhz = clk_table->entries[0].fclk_mhz;
+	low_pstate_lvl.socclk_mhz = clk_table->entries[0].socclk_mhz;
+	low_pstate_lvl.dram_speed_mts = clk_table->entries[0].memclk_mhz * 2;
+
+	low_pstate_lvl.dispclk_mhz = dcn2_1_soc.clock_limits[high_voltage_lvl].dispclk_mhz;
+	low_pstate_lvl.dppclk_mhz = dcn2_1_soc.clock_limits[high_voltage_lvl].dppclk_mhz;
+	low_pstate_lvl.dram_bw_per_chan_gbps = dcn2_1_soc.clock_limits[high_voltage_lvl].dram_bw_per_chan_gbps;
+	low_pstate_lvl.dscclk_mhz = dcn2_1_soc.clock_limits[high_voltage_lvl].dscclk_mhz;
+	low_pstate_lvl.dtbclk_mhz = dcn2_1_soc.clock_limits[high_voltage_lvl].dtbclk_mhz;
+	low_pstate_lvl.phyclk_d18_mhz = dcn2_1_soc.clock_limits[high_voltage_lvl].phyclk_d18_mhz;
+	low_pstate_lvl.phyclk_mhz = dcn2_1_soc.clock_limits[high_voltage_lvl].phyclk_mhz;
+
+	for (i = clk_table->num_entries; i > 1; i--)
+		clk_table->entries[i] = clk_table->entries[i-1];
+	clk_table->entries[1] = clk_table->entries[0];
+	clk_table->num_entries++;
+
+	return low_pstate_lvl;
+}
+
+void dcn21_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_params)
+{
+	struct _vcs_dpi_voltage_scaling_st *s = dc->scratch.update_bw_bounding_box.clock_limits;
+	struct dcn21_resource_pool *pool = TO_DCN21_RES_POOL(dc->res_pool);
+	struct clk_limit_table *clk_table = &bw_params->clk_table;
+	unsigned int i, closest_clk_lvl = 0, k = 0;
+	int j;
+
+	dc_assert_fp_enabled();
+
+	dcn2_1_ip.max_num_otg = pool->base.res_cap->num_timing_generator;
+	dcn2_1_ip.max_num_dpp = pool->base.pipe_count;
+	dcn2_1_soc.num_chans = bw_params->num_channels;
+
+	ASSERT(clk_table->num_entries);
+	/* Copy dcn2_1_soc.clock_limits to clock_limits to avoid copying over null states later */
+	memcpy(s, dcn2_1_soc.clock_limits, sizeof(dcn2_1_soc.clock_limits));
+
+	for (i = 0; i < clk_table->num_entries; i++) {
+		/* loop backwards*/
+		for (closest_clk_lvl = 0, j = dcn2_1_soc.num_states - 1; j >= 0; j--) {
+			if ((unsigned int) dcn2_1_soc.clock_limits[j].dcfclk_mhz <= clk_table->entries[i].dcfclk_mhz) {
+				closest_clk_lvl = j;
+				break;
+			}
+		}
+
+		/* clk_table[1] is reserved for min DF PState.  skip here to fill in later. */
+		if (i == 1)
+			k++;
+
+		s[k].state = k;
+		s[k].dcfclk_mhz = clk_table->entries[i].dcfclk_mhz;
+		s[k].fabricclk_mhz = clk_table->entries[i].fclk_mhz;
+		s[k].socclk_mhz = clk_table->entries[i].socclk_mhz;
+		s[k].dram_speed_mts = clk_table->entries[i].memclk_mhz * 2;
+
+		s[k].dispclk_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].dispclk_mhz;
+		s[k].dppclk_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].dppclk_mhz;
+		s[k].dram_bw_per_chan_gbps =
+			dcn2_1_soc.clock_limits[closest_clk_lvl].dram_bw_per_chan_gbps;
+		s[k].dscclk_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].dscclk_mhz;
+		s[k].dtbclk_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].dtbclk_mhz;
+		s[k].phyclk_d18_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].phyclk_d18_mhz;
+		s[k].phyclk_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].phyclk_mhz;
+
+		k++;
+	}
+
+	memcpy(&dcn2_1_soc.clock_limits, s, sizeof(dcn2_1_soc.clock_limits));
+
+	if (clk_table->num_entries) {
+		dcn2_1_soc.num_states = clk_table->num_entries + 1;
+		/* fill in min DF PState */
+		dcn2_1_soc.clock_limits[1] = construct_low_pstate_lvl(clk_table, closest_clk_lvl);
+		/* duplicate last level */
+		dcn2_1_soc.clock_limits[dcn2_1_soc.num_states] = dcn2_1_soc.clock_limits[dcn2_1_soc.num_states - 1];
+		dcn2_1_soc.clock_limits[dcn2_1_soc.num_states].state = dcn2_1_soc.num_states;
+	}
+
+	dml_init_instance(&dc->dml, &dcn2_1_soc, &dcn2_1_ip, DML_PROJECT_DCN21);
+}
+
+void dcn21_clk_mgr_set_bw_params_wm_table(struct clk_bw_params *bw_params)
+{
+	dc_assert_fp_enabled();
+
+	bw_params->wm_table.entries[WM_D].pstate_latency_us = LPDDR_MEM_RETRAIN_LATENCY;
+	bw_params->wm_table.entries[WM_D].wm_inst = WM_D;
+	bw_params->wm_table.entries[WM_D].wm_type = WM_TYPE_RETRAINING;
+	bw_params->wm_table.entries[WM_D].valid = true;
+}
+
+void dcn201_populate_dml_writeback_from_context_fpu(struct dc *dc,
+						    struct resource_context *res_ctx,
+						    display_e2e_pipe_params_st *pipes)
+{
+	int pipe_cnt, i, j;
+	double max_calc_writeback_dispclk;
+	double writeback_dispclk;
+	struct writeback_st dout_wb = {0};
+
+	dc_assert_fp_enabled();
+
+	for (i = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) {
+		struct dc_stream_state *stream = res_ctx->pipe_ctx[i].stream;
+
+		if (!stream)
+			continue;
+		max_calc_writeback_dispclk = 0;
+
+		/* Set writeback information */
+		pipes[pipe_cnt].dout.wb_enable = 0;
+		pipes[pipe_cnt].dout.num_active_wb = 0;
+		for (j = 0; j < stream->num_wb_info; j++) {
+			struct dc_writeback_info *wb_info = &stream->writeback_info[j];
+
+			if (wb_info->wb_enabled && wb_info->writeback_source_plane &&
+					(wb_info->writeback_source_plane == res_ctx->pipe_ctx[i].plane_state)) {
+				pipes[pipe_cnt].dout.wb_enable = 1;
+				pipes[pipe_cnt].dout.num_active_wb++;
+				dout_wb.wb_src_height = wb_info->dwb_params.cnv_params.crop_en ?
+					wb_info->dwb_params.cnv_params.crop_height :
+					wb_info->dwb_params.cnv_params.src_height;
+				dout_wb.wb_src_width = wb_info->dwb_params.cnv_params.crop_en ?
+					wb_info->dwb_params.cnv_params.crop_width :
+					wb_info->dwb_params.cnv_params.src_width;
+				dout_wb.wb_dst_width = wb_info->dwb_params.dest_width;
+				dout_wb.wb_dst_height = wb_info->dwb_params.dest_height;
+				dout_wb.wb_htaps_luma = wb_info->dwb_params.scaler_taps.h_taps;
+				dout_wb.wb_vtaps_luma = wb_info->dwb_params.scaler_taps.v_taps;
+				dout_wb.wb_htaps_chroma = wb_info->dwb_params.scaler_taps.h_taps_c;
+				dout_wb.wb_vtaps_chroma = wb_info->dwb_params.scaler_taps.v_taps_c;
+				dout_wb.wb_hratio = wb_info->dwb_params.cnv_params.crop_en ?
+					(double)wb_info->dwb_params.cnv_params.crop_width /
+						(double)wb_info->dwb_params.dest_width :
+					(double)wb_info->dwb_params.cnv_params.src_width /
+						(double)wb_info->dwb_params.dest_width;
+				dout_wb.wb_vratio = wb_info->dwb_params.cnv_params.crop_en ?
+					(double)wb_info->dwb_params.cnv_params.crop_height /
+						(double)wb_info->dwb_params.dest_height :
+					(double)wb_info->dwb_params.cnv_params.src_height /
+						(double)wb_info->dwb_params.dest_height;
+				if (wb_info->dwb_params.out_format == dwb_scaler_mode_yuv420) {
+					if (wb_info->dwb_params.output_depth == DWB_OUTPUT_PIXEL_DEPTH_8BPC)
+						dout_wb.wb_pixel_format = dm_420_8;
+					else
+						dout_wb.wb_pixel_format = dm_420_10;
+				} else
+					dout_wb.wb_pixel_format = dm_444_32;
+
+				/* Workaround for cases where multiple writebacks are connected to same plane
+				 * In which case, need to compute worst case and set the associated writeback parameters
+				 * This workaround is necessary due to DML computation assuming only 1 set of writeback
+				 * parameters per pipe */
+				writeback_dispclk = CalculateWriteBackDISPCLK(
+						dout_wb.wb_pixel_format,
+						pipes[pipe_cnt].pipe.dest.pixel_rate_mhz,
+						dout_wb.wb_hratio,
+						dout_wb.wb_vratio,
+						dout_wb.wb_htaps_luma,
+						dout_wb.wb_vtaps_luma,
+						dout_wb.wb_htaps_chroma,
+						dout_wb.wb_vtaps_chroma,
+						dout_wb.wb_dst_width,
+						pipes[pipe_cnt].pipe.dest.htotal,
+						2);
+
+				if (writeback_dispclk > max_calc_writeback_dispclk) {
+					max_calc_writeback_dispclk = writeback_dispclk;
+					pipes[pipe_cnt].dout.wb = dout_wb;
+				}
+			}
+		}
+
+		pipe_cnt++;
+	}
+
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn20/dcn20_fpu.h b/drivers/gpu/drm/amd/display/dc/dml/dcn20/dcn20_fpu.h
new file mode 100644
index 000000000000..aed00039ca62
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn20/dcn20_fpu.h
@@ -0,0 +1,89 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2021 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+#include "core_types.h"
+
+#ifndef __DCN20_FPU_H__
+#define __DCN20_FPU_H__
+
+void dcn20_populate_dml_writeback_from_context(struct dc *dc,
+					       struct resource_context *res_ctx,
+					       display_e2e_pipe_params_st *pipes);
+
+void dcn20_fpu_set_wb_arb_params(struct mcif_arb_params *wb_arb_params,
+				 struct dc_state *context,
+				 display_e2e_pipe_params_st *pipes,
+				 int pipe_cnt, int i);
+void dcn20_calculate_dlg_params(struct dc *dc,
+				struct dc_state *context,
+				display_e2e_pipe_params_st *pipes,
+				int pipe_cnt,
+				int vlevel);
+int dcn20_populate_dml_pipes_from_context(struct dc *dc,
+					  struct dc_state *context,
+					  display_e2e_pipe_params_st *pipes,
+					  enum dc_validate_mode validate_mode);
+void dcn20_calculate_wm(struct dc *dc,
+			struct dc_state *context,
+			display_e2e_pipe_params_st *pipes,
+			int *out_pipe_cnt,
+			int *pipe_split_from,
+			int vlevel,
+			enum dc_validate_mode validate_mode);
+void dcn20_cap_soc_clocks(struct _vcs_dpi_soc_bounding_box_st *bb,
+			  struct pp_smu_nv_clock_table max_clocks);
+void dcn20_update_bounding_box(struct dc *dc,
+			       struct _vcs_dpi_soc_bounding_box_st *bb,
+			       struct pp_smu_nv_clock_table *max_clocks,
+			       unsigned int *uclk_states,
+			       unsigned int num_states);
+void dcn20_patch_bounding_box(struct dc *dc,
+			      struct _vcs_dpi_soc_bounding_box_st *bb);
+bool dcn20_validate_bandwidth_fp(struct dc *dc, struct dc_state *context,
+				 enum dc_validate_mode validate_mode, display_e2e_pipe_params_st *pipes);
+void dcn20_fpu_set_wm_ranges(int i,
+			     struct pp_smu_wm_range_sets *ranges,
+			     struct _vcs_dpi_soc_bounding_box_st *loaded_bb);
+void dcn20_fpu_adjust_dppclk(struct vba_vars_st *v,
+			     int vlevel,
+			     int max_mpc_comb,
+			     int pipe_idx,
+			     bool is_validating_bw);
+
+int dcn21_populate_dml_pipes_from_context(struct dc *dc,
+					  struct dc_state *context,
+					  display_e2e_pipe_params_st *pipes,
+					  enum dc_validate_mode validate_mode);
+bool dcn21_validate_bandwidth_fp(struct dc *dc, struct dc_state *context, enum
+				 dc_validate_mode, display_e2e_pipe_params_st *pipes);
+void dcn21_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_params);
+
+void dcn21_clk_mgr_set_bw_params_wm_table(struct clk_bw_params *bw_params);
+
+void dcn201_populate_dml_writeback_from_context_fpu(struct dc *dc,
+						struct resource_context *res_ctx,
+						display_e2e_pipe_params_st *pipes);
+
+#endif /* __DCN20_FPU_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_mode_vba_20.c b/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_mode_vba_20.c
new file mode 100644
index 000000000000..0c8c4a080c50
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_mode_vba_20.c
@@ -0,0 +1,5115 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "../display_mode_lib.h"
+#include "display_mode_vba_20.h"
+#include "../dml_inline_defs.h"
+
+/*
+ * NOTE:
+ *   This file is gcc-parseable HW gospel, coming straight from HW engineers.
+ *
+ * It doesn't adhere to Linux kernel style and sometimes will do things in odd
+ * ways. Unless there is something clearly wrong with it the code should
+ * remain as-is as it provides us with a guarantee from HW that it is correct.
+ */
+
+#define BPP_INVALID 0
+#define BPP_BLENDED_PIPE 0xffffffff
+#define DCN20_MAX_420_IMAGE_WIDTH 4096
+
+static double adjust_ReturnBW(
+		struct display_mode_lib *mode_lib,
+		double ReturnBW,
+		bool DCCEnabledAnyPlane,
+		double ReturnBandwidthToDCN);
+static unsigned int dscceComputeDelay(
+		unsigned int bpc,
+		double bpp,
+		unsigned int sliceWidth,
+		unsigned int numSlices,
+		enum output_format_class pixelFormat);
+static unsigned int dscComputeDelay(enum output_format_class pixelFormat);
+// Super monster function with some 45 argument
+static bool CalculatePrefetchSchedule(
+		struct display_mode_lib *mode_lib,
+		double DPPCLK,
+		double DISPCLK,
+		double PixelClock,
+		double DCFCLKDeepSleep,
+		unsigned int DSCDelay,
+		unsigned int DPPPerPlane,
+		bool ScalerEnabled,
+		unsigned int NumberOfCursors,
+		double DPPCLKDelaySubtotal,
+		double DPPCLKDelaySCL,
+		double DPPCLKDelaySCLLBOnly,
+		double DPPCLKDelayCNVCFormater,
+		double DPPCLKDelayCNVCCursor,
+		double DISPCLKDelaySubtotal,
+		unsigned int ScalerRecoutWidth,
+		enum output_format_class OutputFormat,
+		unsigned int VBlank,
+		unsigned int HTotal,
+		unsigned int MaxInterDCNTileRepeaters,
+		unsigned int VStartup,
+		unsigned int PageTableLevels,
+		bool GPUVMEnable,
+		bool DynamicMetadataEnable,
+		unsigned int DynamicMetadataLinesBeforeActiveRequired,
+		unsigned int DynamicMetadataTransmittedBytes,
+		bool DCCEnable,
+		double UrgentLatencyPixelDataOnly,
+		double UrgentExtraLatency,
+		double TCalc,
+		unsigned int PDEAndMetaPTEBytesFrame,
+		unsigned int MetaRowByte,
+		unsigned int PixelPTEBytesPerRow,
+		double PrefetchSourceLinesY,
+		unsigned int SwathWidthY,
+		double BytePerPixelDETY,
+		double VInitPreFillY,
+		unsigned int MaxNumSwathY,
+		double PrefetchSourceLinesC,
+		double BytePerPixelDETC,
+		double VInitPreFillC,
+		unsigned int MaxNumSwathC,
+		unsigned int SwathHeightY,
+		unsigned int SwathHeightC,
+		double TWait,
+		bool XFCEnabled,
+		double XFCRemoteSurfaceFlipDelay,
+		bool InterlaceEnable,
+		bool ProgressiveToInterlaceUnitInOPP,
+		double *DSTXAfterScaler,
+		double *DSTYAfterScaler,
+		double *DestinationLinesForPrefetch,
+		double *PrefetchBandwidth,
+		double *DestinationLinesToRequestVMInVBlank,
+		double *DestinationLinesToRequestRowInVBlank,
+		double *VRatioPrefetchY,
+		double *VRatioPrefetchC,
+		double *RequiredPrefetchPixDataBW,
+		unsigned int *VStartupRequiredWhenNotEnoughTimeForDynamicMetadata,
+		double *Tno_bw,
+		unsigned int *VUpdateOffsetPix,
+		double *VUpdateWidthPix,
+		double *VReadyOffsetPix);
+static double RoundToDFSGranularityUp(double Clock, double VCOSpeed);
+static double RoundToDFSGranularityDown(double Clock, double VCOSpeed);
+static double CalculatePrefetchSourceLines(
+		struct display_mode_lib *mode_lib,
+		double VRatio,
+		double vtaps,
+		bool Interlace,
+		bool ProgressiveToInterlaceUnitInOPP,
+		unsigned int SwathHeight,
+		unsigned int ViewportYStart,
+		double *VInitPreFill,
+		unsigned int *MaxNumSwath);
+static unsigned int CalculateVMAndRowBytes(
+		struct display_mode_lib *mode_lib,
+		bool DCCEnable,
+		unsigned int BlockHeight256Bytes,
+		unsigned int BlockWidth256Bytes,
+		enum source_format_class SourcePixelFormat,
+		unsigned int SurfaceTiling,
+		unsigned int BytePerPixel,
+		enum scan_direction_class ScanDirection,
+		unsigned int ViewportWidth,
+		unsigned int ViewportHeight,
+		unsigned int SwathWidthY,
+		bool GPUVMEnable,
+		unsigned int VMMPageSize,
+		unsigned int PTEBufferSizeInRequestsLuma,
+		unsigned int PDEProcessingBufIn64KBReqs,
+		unsigned int Pitch,
+		unsigned int DCCMetaPitch,
+		unsigned int *MacroTileWidth,
+		unsigned int *MetaRowByte,
+		unsigned int *PixelPTEBytesPerRow,
+		bool *PTEBufferSizeNotExceeded,
+		unsigned int *dpte_row_height,
+		unsigned int *meta_row_height);
+static double CalculateTWait(
+		unsigned int PrefetchMode,
+		double DRAMClockChangeLatency,
+		double UrgentLatencyPixelDataOnly,
+		double SREnterPlusExitTime);
+static double CalculateRemoteSurfaceFlipDelay(
+		struct display_mode_lib *mode_lib,
+		double VRatio,
+		double SwathWidth,
+		double Bpp,
+		double LineTime,
+		double XFCTSlvVupdateOffset,
+		double XFCTSlvVupdateWidth,
+		double XFCTSlvVreadyOffset,
+		double XFCXBUFLatencyTolerance,
+		double XFCFillBWOverhead,
+		double XFCSlvChunkSize,
+		double XFCBusTransportTime,
+		double TCalc,
+		double TWait,
+		double *SrcActiveDrainRate,
+		double *TInitXFill,
+		double *TslvChk);
+static void CalculateActiveRowBandwidth(
+		bool GPUVMEnable,
+		enum source_format_class SourcePixelFormat,
+		double VRatio,
+		bool DCCEnable,
+		double LineTime,
+		unsigned int MetaRowByteLuma,
+		unsigned int MetaRowByteChroma,
+		unsigned int meta_row_height_luma,
+		unsigned int meta_row_height_chroma,
+		unsigned int PixelPTEBytesPerRowLuma,
+		unsigned int PixelPTEBytesPerRowChroma,
+		unsigned int dpte_row_height_luma,
+		unsigned int dpte_row_height_chroma,
+		double *meta_row_bw,
+		double *dpte_row_bw,
+		double *qual_row_bw);
+static void CalculateFlipSchedule(
+		struct display_mode_lib *mode_lib,
+		double UrgentExtraLatency,
+		double UrgentLatencyPixelDataOnly,
+		unsigned int GPUVMMaxPageTableLevels,
+		bool GPUVMEnable,
+		double BandwidthAvailableForImmediateFlip,
+		unsigned int TotImmediateFlipBytes,
+		enum source_format_class SourcePixelFormat,
+		unsigned int ImmediateFlipBytes,
+		double LineTime,
+		double VRatio,
+		double Tno_bw,
+		double PDEAndMetaPTEBytesFrame,
+		unsigned int MetaRowByte,
+		unsigned int PixelPTEBytesPerRow,
+		bool DCCEnable,
+		unsigned int dpte_row_height,
+		unsigned int meta_row_height,
+		double qual_row_bw,
+		double *DestinationLinesToRequestVMInImmediateFlip,
+		double *DestinationLinesToRequestRowInImmediateFlip,
+		double *final_flip_bw,
+		bool *ImmediateFlipSupportedForPipe);
+static double CalculateWriteBackDelay(
+		enum source_format_class WritebackPixelFormat,
+		double WritebackHRatio,
+		double WritebackVRatio,
+		unsigned int WritebackLumaHTaps,
+		unsigned int WritebackLumaVTaps,
+		unsigned int WritebackChromaHTaps,
+		unsigned int WritebackChromaVTaps,
+		unsigned int WritebackDestinationWidth);
+
+static void dml20_DisplayPipeConfiguration(struct display_mode_lib *mode_lib);
+static void dml20_DISPCLKDPPCLKDCFCLKDeepSleepPrefetchParametersWatermarksAndPerformanceCalculation(
+		struct display_mode_lib *mode_lib);
+
+void dml20_recalculate(struct display_mode_lib *mode_lib)
+{
+	ModeSupportAndSystemConfiguration(mode_lib);
+	mode_lib->vba.FabricAndDRAMBandwidth = dml_min(
+		mode_lib->vba.DRAMSpeed * mode_lib->vba.NumberOfChannels * mode_lib->vba.DRAMChannelWidth,
+		mode_lib->vba.FabricClock * mode_lib->vba.FabricDatapathToDCNDataReturn) / 1000.0;
+	PixelClockAdjustmentForProgressiveToInterlaceUnit(mode_lib);
+	dml20_DisplayPipeConfiguration(mode_lib);
+	dml20_DISPCLKDPPCLKDCFCLKDeepSleepPrefetchParametersWatermarksAndPerformanceCalculation(mode_lib);
+}
+
+static double adjust_ReturnBW(
+		struct display_mode_lib *mode_lib,
+		double ReturnBW,
+		bool DCCEnabledAnyPlane,
+		double ReturnBandwidthToDCN)
+{
+	double CriticalCompression;
+
+	if (DCCEnabledAnyPlane
+			&& ReturnBandwidthToDCN
+					> mode_lib->vba.DCFCLK * mode_lib->vba.ReturnBusWidth / 4.0)
+		ReturnBW =
+				dml_min(
+						ReturnBW,
+						ReturnBandwidthToDCN * 4
+								* (1.0
+										- mode_lib->vba.UrgentLatencyPixelDataOnly
+												/ ((mode_lib->vba.ROBBufferSizeInKByte
+														- mode_lib->vba.PixelChunkSizeInKByte)
+														* 1024
+														/ ReturnBandwidthToDCN
+														- mode_lib->vba.DCFCLK
+																* mode_lib->vba.ReturnBusWidth
+																/ 4)
+										+ mode_lib->vba.UrgentLatencyPixelDataOnly));
+
+	CriticalCompression = 2.0 * mode_lib->vba.ReturnBusWidth * mode_lib->vba.DCFCLK
+			* mode_lib->vba.UrgentLatencyPixelDataOnly
+			/ (ReturnBandwidthToDCN * mode_lib->vba.UrgentLatencyPixelDataOnly
+					+ (mode_lib->vba.ROBBufferSizeInKByte
+							- mode_lib->vba.PixelChunkSizeInKByte)
+							* 1024);
+
+	if (DCCEnabledAnyPlane && CriticalCompression > 1.0 && CriticalCompression < 4.0)
+		ReturnBW =
+				dml_min(
+						ReturnBW,
+						4.0 * ReturnBandwidthToDCN
+								* (mode_lib->vba.ROBBufferSizeInKByte
+										- mode_lib->vba.PixelChunkSizeInKByte)
+								* 1024
+								* mode_lib->vba.ReturnBusWidth
+								* mode_lib->vba.DCFCLK
+								* mode_lib->vba.UrgentLatencyPixelDataOnly
+								/ dml_pow(
+										(ReturnBandwidthToDCN
+												* mode_lib->vba.UrgentLatencyPixelDataOnly
+												+ (mode_lib->vba.ROBBufferSizeInKByte
+														- mode_lib->vba.PixelChunkSizeInKByte)
+														* 1024),
+										2));
+
+	return ReturnBW;
+}
+
+static unsigned int dscceComputeDelay(
+		unsigned int bpc,
+		double bpp,
+		unsigned int sliceWidth,
+		unsigned int numSlices,
+		enum output_format_class pixelFormat)
+{
+	// valid bpc         = source bits per component in the set of {8, 10, 12}
+	// valid bpp         = increments of 1/16 of a bit
+	//                    min = 6/7/8 in N420/N422/444, respectively
+	//                    max = such that compression is 1:1
+	//valid sliceWidth  = number of pixels per slice line, must be less than or equal to 5184/numSlices (or 4096/numSlices in 420 mode)
+	//valid numSlices   = number of slices in the horiziontal direction per DSC engine in the set of {1, 2, 3, 4}
+	//valid pixelFormat = pixel/color format in the set of {:N444_RGB, :S422, :N422, :N420}
+
+	// fixed value
+	unsigned int rcModelSize = 8192;
+
+	// N422/N420 operate at 2 pixels per clock
+	unsigned int pixelsPerClock, lstall, D, initalXmitDelay, w, s, ix, wx, p, l0, a, ax, l,
+			Delay, pixels;
+
+	if (pixelFormat == dm_n422 || pixelFormat == dm_420)
+		pixelsPerClock = 2;
+	// #all other modes operate at 1 pixel per clock
+	else
+		pixelsPerClock = 1;
+
+	//initial transmit delay as per PPS
+	initalXmitDelay = dml_round(rcModelSize / 2.0 / bpp / pixelsPerClock);
+
+	//compute ssm delay
+	if (bpc == 8)
+		D = 81;
+	else if (bpc == 10)
+		D = 89;
+	else
+		D = 113;
+
+	//divide by pixel per cycle to compute slice width as seen by DSC
+	w = sliceWidth / pixelsPerClock;
+
+	//422 mode has an additional cycle of delay
+	if (pixelFormat == dm_s422)
+		s = 1;
+	else
+		s = 0;
+
+	//main calculation for the dscce
+	ix = initalXmitDelay + 45;
+	wx = (w + 2) / 3;
+	p = 3 * wx - w;
+	l0 = ix / w;
+	a = ix + p * l0;
+	ax = (a + 2) / 3 + D + 6 + 1;
+	l = (ax + wx - 1) / wx;
+	if ((ix % w) == 0 && p != 0)
+		lstall = 1;
+	else
+		lstall = 0;
+	Delay = l * wx * (numSlices - 1) + ax + s + lstall + 22;
+
+	//dsc processes 3 pixel containers per cycle and a container can contain 1 or 2 pixels
+	pixels = Delay * 3 * pixelsPerClock;
+	return pixels;
+}
+
+static unsigned int dscComputeDelay(enum output_format_class pixelFormat)
+{
+	unsigned int Delay = 0;
+
+	if (pixelFormat == dm_420) {
+		//   sfr
+		Delay = Delay + 2;
+		//   dsccif
+		Delay = Delay + 0;
+		//   dscc - input deserializer
+		Delay = Delay + 3;
+		//   dscc gets pixels every other cycle
+		Delay = Delay + 2;
+		//   dscc - input cdc fifo
+		Delay = Delay + 12;
+		//   dscc gets pixels every other cycle
+		Delay = Delay + 13;
+		//   dscc - cdc uncertainty
+		Delay = Delay + 2;
+		//   dscc - output cdc fifo
+		Delay = Delay + 7;
+		//   dscc gets pixels every other cycle
+		Delay = Delay + 3;
+		//   dscc - cdc uncertainty
+		Delay = Delay + 2;
+		//   dscc - output serializer
+		Delay = Delay + 1;
+		//   sft
+		Delay = Delay + 1;
+	} else if (pixelFormat == dm_n422) {
+		//   sfr
+		Delay = Delay + 2;
+		//   dsccif
+		Delay = Delay + 1;
+		//   dscc - input deserializer
+		Delay = Delay + 5;
+		//  dscc - input cdc fifo
+		Delay = Delay + 25;
+		//   dscc - cdc uncertainty
+		Delay = Delay + 2;
+		//   dscc - output cdc fifo
+		Delay = Delay + 10;
+		//   dscc - cdc uncertainty
+		Delay = Delay + 2;
+		//   dscc - output serializer
+		Delay = Delay + 1;
+		//   sft
+		Delay = Delay + 1;
+	} else {
+		//   sfr
+		Delay = Delay + 2;
+		//   dsccif
+		Delay = Delay + 0;
+		//   dscc - input deserializer
+		Delay = Delay + 3;
+		//   dscc - input cdc fifo
+		Delay = Delay + 12;
+		//   dscc - cdc uncertainty
+		Delay = Delay + 2;
+		//   dscc - output cdc fifo
+		Delay = Delay + 7;
+		//   dscc - output serializer
+		Delay = Delay + 1;
+		//   dscc - cdc uncertainty
+		Delay = Delay + 2;
+		//   sft
+		Delay = Delay + 1;
+	}
+
+	return Delay;
+}
+
+static bool CalculatePrefetchSchedule(
+		struct display_mode_lib *mode_lib,
+		double DPPCLK,
+		double DISPCLK,
+		double PixelClock,
+		double DCFCLKDeepSleep,
+		unsigned int DSCDelay,
+		unsigned int DPPPerPlane,
+		bool ScalerEnabled,
+		unsigned int NumberOfCursors,
+		double DPPCLKDelaySubtotal,
+		double DPPCLKDelaySCL,
+		double DPPCLKDelaySCLLBOnly,
+		double DPPCLKDelayCNVCFormater,
+		double DPPCLKDelayCNVCCursor,
+		double DISPCLKDelaySubtotal,
+		unsigned int ScalerRecoutWidth,
+		enum output_format_class OutputFormat,
+		unsigned int VBlank,
+		unsigned int HTotal,
+		unsigned int MaxInterDCNTileRepeaters,
+		unsigned int VStartup,
+		unsigned int PageTableLevels,
+		bool GPUVMEnable,
+		bool DynamicMetadataEnable,
+		unsigned int DynamicMetadataLinesBeforeActiveRequired,
+		unsigned int DynamicMetadataTransmittedBytes,
+		bool DCCEnable,
+		double UrgentLatencyPixelDataOnly,
+		double UrgentExtraLatency,
+		double TCalc,
+		unsigned int PDEAndMetaPTEBytesFrame,
+		unsigned int MetaRowByte,
+		unsigned int PixelPTEBytesPerRow,
+		double PrefetchSourceLinesY,
+		unsigned int SwathWidthY,
+		double BytePerPixelDETY,
+		double VInitPreFillY,
+		unsigned int MaxNumSwathY,
+		double PrefetchSourceLinesC,
+		double BytePerPixelDETC,
+		double VInitPreFillC,
+		unsigned int MaxNumSwathC,
+		unsigned int SwathHeightY,
+		unsigned int SwathHeightC,
+		double TWait,
+		bool XFCEnabled,
+		double XFCRemoteSurfaceFlipDelay,
+		bool InterlaceEnable,
+		bool ProgressiveToInterlaceUnitInOPP,
+		double *DSTXAfterScaler,
+		double *DSTYAfterScaler,
+		double *DestinationLinesForPrefetch,
+		double *PrefetchBandwidth,
+		double *DestinationLinesToRequestVMInVBlank,
+		double *DestinationLinesToRequestRowInVBlank,
+		double *VRatioPrefetchY,
+		double *VRatioPrefetchC,
+		double *RequiredPrefetchPixDataBW,
+		unsigned int *VStartupRequiredWhenNotEnoughTimeForDynamicMetadata,
+		double *Tno_bw,
+		unsigned int *VUpdateOffsetPix,
+		double *VUpdateWidthPix,
+		double *VReadyOffsetPix)
+{
+	bool MyError = false;
+	unsigned int DPPCycles, DISPCLKCycles;
+	double DSTTotalPixelsAfterScaler, TotalRepeaterDelayTime;
+	double Tdm, LineTime, Tsetup;
+	double dst_y_prefetch_equ;
+	double Tsw_oto;
+	double prefetch_bw_oto;
+	double Tvm_oto;
+	double Tr0_oto;
+	double Tpre_oto;
+	double dst_y_prefetch_oto;
+	double TimeForFetchingMetaPTE = 0;
+	double TimeForFetchingRowInVBlank = 0;
+	double LinesToRequestPrefetchPixelData = 0;
+
+	if (ScalerEnabled)
+		DPPCycles = DPPCLKDelaySubtotal + DPPCLKDelaySCL;
+	else
+		DPPCycles = DPPCLKDelaySubtotal + DPPCLKDelaySCLLBOnly;
+
+	DPPCycles = DPPCycles + DPPCLKDelayCNVCFormater + NumberOfCursors * DPPCLKDelayCNVCCursor;
+
+	DISPCLKCycles = DISPCLKDelaySubtotal;
+
+	if (DPPCLK == 0.0 || DISPCLK == 0.0)
+		return true;
+
+	*DSTXAfterScaler = DPPCycles * PixelClock / DPPCLK + DISPCLKCycles * PixelClock / DISPCLK
+			+ DSCDelay;
+
+	if (DPPPerPlane > 1)
+		*DSTXAfterScaler = *DSTXAfterScaler + ScalerRecoutWidth;
+
+	if (OutputFormat == dm_420 || (InterlaceEnable && ProgressiveToInterlaceUnitInOPP))
+		*DSTYAfterScaler = 1;
+	else
+		*DSTYAfterScaler = 0;
+
+	DSTTotalPixelsAfterScaler = ((double) (*DSTYAfterScaler * HTotal)) + *DSTXAfterScaler;
+	*DSTYAfterScaler = dml_floor(DSTTotalPixelsAfterScaler / HTotal, 1);
+	*DSTXAfterScaler = DSTTotalPixelsAfterScaler - ((double) (*DSTYAfterScaler * HTotal));
+
+	*VUpdateOffsetPix = dml_ceil(HTotal / 4.0, 1);
+	TotalRepeaterDelayTime = MaxInterDCNTileRepeaters * (2.0 / DPPCLK + 3.0 / DISPCLK);
+	*VUpdateWidthPix = (14.0 / DCFCLKDeepSleep + 12.0 / DPPCLK + TotalRepeaterDelayTime)
+			* PixelClock;
+
+	*VReadyOffsetPix = dml_max(
+			150.0 / DPPCLK,
+			TotalRepeaterDelayTime + 20.0 / DCFCLKDeepSleep + 10.0 / DPPCLK)
+			* PixelClock;
+
+	Tsetup = (double) (*VUpdateOffsetPix + *VUpdateWidthPix + *VReadyOffsetPix) / PixelClock;
+
+	LineTime = (double) HTotal / PixelClock;
+
+	if (DynamicMetadataEnable) {
+		double Tdmbf, Tdmec, Tdmsks;
+
+		Tdm = dml_max(0.0, UrgentExtraLatency - TCalc);
+		Tdmbf = DynamicMetadataTransmittedBytes / 4.0 / DISPCLK;
+		Tdmec = LineTime;
+		if (DynamicMetadataLinesBeforeActiveRequired == 0)
+			Tdmsks = VBlank * LineTime / 2.0;
+		else
+			Tdmsks = DynamicMetadataLinesBeforeActiveRequired * LineTime;
+		if (InterlaceEnable && !ProgressiveToInterlaceUnitInOPP)
+			Tdmsks = Tdmsks / 2;
+		if (VStartup * LineTime
+				< Tsetup + TWait + UrgentExtraLatency + Tdmbf + Tdmec + Tdmsks) {
+			MyError = true;
+			*VStartupRequiredWhenNotEnoughTimeForDynamicMetadata = (Tsetup + TWait
+					+ UrgentExtraLatency + Tdmbf + Tdmec + Tdmsks) / LineTime;
+		} else
+			*VStartupRequiredWhenNotEnoughTimeForDynamicMetadata = 0.0;
+	} else
+		Tdm = 0;
+
+	if (GPUVMEnable) {
+		if (PageTableLevels == 4)
+			*Tno_bw = UrgentExtraLatency + UrgentLatencyPixelDataOnly;
+		else if (PageTableLevels == 3)
+			*Tno_bw = UrgentExtraLatency;
+		else
+			*Tno_bw = 0;
+	} else if (DCCEnable)
+		*Tno_bw = LineTime;
+	else
+		*Tno_bw = LineTime / 4;
+
+	dst_y_prefetch_equ = VStartup - dml_max(TCalc + TWait, XFCRemoteSurfaceFlipDelay) / LineTime
+			- (Tsetup + Tdm) / LineTime
+			- (*DSTYAfterScaler + *DSTXAfterScaler / HTotal);
+
+	Tsw_oto = dml_max(PrefetchSourceLinesY, PrefetchSourceLinesC) * LineTime;
+
+	prefetch_bw_oto = (MetaRowByte + PixelPTEBytesPerRow
+			+ PrefetchSourceLinesY * SwathWidthY * dml_ceil(BytePerPixelDETY, 1)
+			+ PrefetchSourceLinesC * SwathWidthY / 2 * dml_ceil(BytePerPixelDETC, 2))
+			/ Tsw_oto;
+
+	if (GPUVMEnable == true) {
+		Tvm_oto =
+				dml_max(
+						*Tno_bw + PDEAndMetaPTEBytesFrame / prefetch_bw_oto,
+						dml_max(
+								UrgentExtraLatency
+										+ UrgentLatencyPixelDataOnly
+												* (PageTableLevels
+														- 1),
+								LineTime / 4.0));
+	} else
+		Tvm_oto = LineTime / 4.0;
+
+	if ((GPUVMEnable == true || DCCEnable == true)) {
+		Tr0_oto = dml_max(
+				(MetaRowByte + PixelPTEBytesPerRow) / prefetch_bw_oto,
+				dml_max(UrgentLatencyPixelDataOnly, dml_max(LineTime - Tvm_oto, LineTime / 4)));
+	} else
+		Tr0_oto = LineTime - Tvm_oto;
+
+	Tpre_oto = Tvm_oto + Tr0_oto + Tsw_oto;
+
+	dst_y_prefetch_oto = Tpre_oto / LineTime;
+
+	if (dst_y_prefetch_oto < dst_y_prefetch_equ)
+		*DestinationLinesForPrefetch = dst_y_prefetch_oto;
+	else
+		*DestinationLinesForPrefetch = dst_y_prefetch_equ;
+
+	*DestinationLinesForPrefetch = dml_floor(4.0 * (*DestinationLinesForPrefetch + 0.125), 1)
+			/ 4;
+
+	dml_print("DML: VStartup: %d\n", VStartup);
+	dml_print("DML: TCalc: %f\n", TCalc);
+	dml_print("DML: TWait: %f\n", TWait);
+	dml_print("DML: XFCRemoteSurfaceFlipDelay: %f\n", XFCRemoteSurfaceFlipDelay);
+	dml_print("DML: LineTime: %f\n", LineTime);
+	dml_print("DML: Tsetup: %f\n", Tsetup);
+	dml_print("DML: Tdm: %f\n", Tdm);
+	dml_print("DML: DSTYAfterScaler: %f\n", *DSTYAfterScaler);
+	dml_print("DML: DSTXAfterScaler: %f\n", *DSTXAfterScaler);
+	dml_print("DML: HTotal: %d\n", HTotal);
+
+	*PrefetchBandwidth = 0;
+	*DestinationLinesToRequestVMInVBlank = 0;
+	*DestinationLinesToRequestRowInVBlank = 0;
+	*VRatioPrefetchY = 0;
+	*VRatioPrefetchC = 0;
+	*RequiredPrefetchPixDataBW = 0;
+	if (*DestinationLinesForPrefetch > 1) {
+		*PrefetchBandwidth = (PDEAndMetaPTEBytesFrame + 2 * MetaRowByte
+				+ 2 * PixelPTEBytesPerRow
+				+ PrefetchSourceLinesY * SwathWidthY * dml_ceil(BytePerPixelDETY, 1)
+				+ PrefetchSourceLinesC * SwathWidthY / 2
+						* dml_ceil(BytePerPixelDETC, 2))
+				/ (*DestinationLinesForPrefetch * LineTime - *Tno_bw);
+		if (GPUVMEnable) {
+			TimeForFetchingMetaPTE =
+					dml_max(
+							*Tno_bw
+									+ (double) PDEAndMetaPTEBytesFrame
+											/ *PrefetchBandwidth,
+							dml_max(
+									UrgentExtraLatency
+											+ UrgentLatencyPixelDataOnly
+													* (PageTableLevels
+															- 1),
+									LineTime / 4));
+		} else {
+			if (NumberOfCursors > 0 || XFCEnabled)
+				TimeForFetchingMetaPTE = LineTime / 4;
+			else
+				TimeForFetchingMetaPTE = 0.0;
+		}
+
+		if ((GPUVMEnable == true || DCCEnable == true)) {
+			TimeForFetchingRowInVBlank =
+					dml_max(
+							(MetaRowByte + PixelPTEBytesPerRow)
+									/ *PrefetchBandwidth,
+							dml_max(
+									UrgentLatencyPixelDataOnly,
+									dml_max(
+											LineTime
+													- TimeForFetchingMetaPTE,
+											LineTime
+													/ 4.0)));
+		} else {
+			if (NumberOfCursors > 0 || XFCEnabled)
+				TimeForFetchingRowInVBlank = LineTime - TimeForFetchingMetaPTE;
+			else
+				TimeForFetchingRowInVBlank = 0.0;
+		}
+
+		*DestinationLinesToRequestVMInVBlank = dml_floor(
+				4.0 * (TimeForFetchingMetaPTE / LineTime + 0.125),
+				1) / 4.0;
+
+		*DestinationLinesToRequestRowInVBlank = dml_floor(
+				4.0 * (TimeForFetchingRowInVBlank / LineTime + 0.125),
+				1) / 4.0;
+
+		LinesToRequestPrefetchPixelData =
+				*DestinationLinesForPrefetch
+						- ((NumberOfCursors > 0 || GPUVMEnable
+								|| DCCEnable) ?
+								(*DestinationLinesToRequestVMInVBlank
+										+ *DestinationLinesToRequestRowInVBlank) :
+								0.0);
+
+		if (LinesToRequestPrefetchPixelData > 0) {
+
+			*VRatioPrefetchY = (double) PrefetchSourceLinesY
+					/ LinesToRequestPrefetchPixelData;
+			*VRatioPrefetchY = dml_max(*VRatioPrefetchY, 1.0);
+			if ((SwathHeightY > 4) && (VInitPreFillY > 3)) {
+				if (LinesToRequestPrefetchPixelData > (VInitPreFillY - 3.0) / 2.0) {
+					*VRatioPrefetchY =
+							dml_max(
+									(double) PrefetchSourceLinesY
+											/ LinesToRequestPrefetchPixelData,
+									(double) MaxNumSwathY
+											* SwathHeightY
+											/ (LinesToRequestPrefetchPixelData
+													- (VInitPreFillY
+															- 3.0)
+															/ 2.0));
+					*VRatioPrefetchY = dml_max(*VRatioPrefetchY, 1.0);
+				} else {
+					MyError = true;
+					*VRatioPrefetchY = 0;
+				}
+			}
+
+			*VRatioPrefetchC = (double) PrefetchSourceLinesC
+					/ LinesToRequestPrefetchPixelData;
+			*VRatioPrefetchC = dml_max(*VRatioPrefetchC, 1.0);
+
+			if ((SwathHeightC > 4)) {
+				if (LinesToRequestPrefetchPixelData > (VInitPreFillC - 3.0) / 2.0) {
+					*VRatioPrefetchC =
+							dml_max(
+									*VRatioPrefetchC,
+									(double) MaxNumSwathC
+											* SwathHeightC
+											/ (LinesToRequestPrefetchPixelData
+													- (VInitPreFillC
+															- 3.0)
+															/ 2.0));
+					*VRatioPrefetchC = dml_max(*VRatioPrefetchC, 1.0);
+				} else {
+					MyError = true;
+					*VRatioPrefetchC = 0;
+				}
+			}
+
+			*RequiredPrefetchPixDataBW =
+					DPPPerPlane
+							* ((double) PrefetchSourceLinesY
+									/ LinesToRequestPrefetchPixelData
+									* dml_ceil(
+											BytePerPixelDETY,
+											1)
+									+ (double) PrefetchSourceLinesC
+											/ LinesToRequestPrefetchPixelData
+											* dml_ceil(
+													BytePerPixelDETC,
+													2)
+											/ 2)
+							* SwathWidthY / LineTime;
+		} else {
+			MyError = true;
+			*VRatioPrefetchY = 0;
+			*VRatioPrefetchC = 0;
+			*RequiredPrefetchPixDataBW = 0;
+		}
+
+	} else {
+		MyError = true;
+	}
+
+	if (MyError) {
+		*PrefetchBandwidth = 0;
+		*DestinationLinesToRequestVMInVBlank = 0;
+		*DestinationLinesToRequestRowInVBlank = 0;
+		*DestinationLinesForPrefetch = 0;
+		*VRatioPrefetchY = 0;
+		*VRatioPrefetchC = 0;
+		*RequiredPrefetchPixDataBW = 0;
+	}
+
+	return MyError;
+}
+
+static double RoundToDFSGranularityUp(double Clock, double VCOSpeed)
+{
+	return VCOSpeed * 4 / dml_floor(VCOSpeed * 4 / Clock, 1);
+}
+
+static double RoundToDFSGranularityDown(double Clock, double VCOSpeed)
+{
+	return VCOSpeed * 4 / dml_ceil(VCOSpeed * 4 / Clock, 1);
+}
+
+static double CalculatePrefetchSourceLines(
+		struct display_mode_lib *mode_lib,
+		double VRatio,
+		double vtaps,
+		bool Interlace,
+		bool ProgressiveToInterlaceUnitInOPP,
+		unsigned int SwathHeight,
+		unsigned int ViewportYStart,
+		double *VInitPreFill,
+		unsigned int *MaxNumSwath)
+{
+	unsigned int MaxPartialSwath;
+
+	if (ProgressiveToInterlaceUnitInOPP)
+		*VInitPreFill = dml_floor((VRatio + vtaps + 1) / 2.0, 1);
+	else
+		*VInitPreFill = dml_floor((VRatio + vtaps + 1 + Interlace * 0.5 * VRatio) / 2.0, 1);
+
+	if (!mode_lib->vba.IgnoreViewportPositioning) {
+
+		*MaxNumSwath = dml_ceil((*VInitPreFill - 1.0) / SwathHeight, 1) + 1.0;
+
+		if (*VInitPreFill > 1.0)
+			MaxPartialSwath = (unsigned int) (*VInitPreFill - 2) % SwathHeight;
+		else
+			MaxPartialSwath = (unsigned int) (*VInitPreFill + SwathHeight - 2)
+					% SwathHeight;
+		MaxPartialSwath = dml_max(1U, MaxPartialSwath);
+
+	} else {
+
+		if (ViewportYStart != 0)
+			dml_print(
+					"WARNING DML: using viewport y position of 0 even though actual viewport y position is non-zero in prefetch source lines calculation\n");
+
+		*MaxNumSwath = dml_ceil(*VInitPreFill / SwathHeight, 1);
+
+		if (*VInitPreFill > 1.0)
+			MaxPartialSwath = (unsigned int) (*VInitPreFill - 1) % SwathHeight;
+		else
+			MaxPartialSwath = (unsigned int) (*VInitPreFill + SwathHeight - 1)
+					% SwathHeight;
+	}
+
+	return *MaxNumSwath * SwathHeight + MaxPartialSwath;
+}
+
+static unsigned int CalculateVMAndRowBytes(
+		struct display_mode_lib *mode_lib,
+		bool DCCEnable,
+		unsigned int BlockHeight256Bytes,
+		unsigned int BlockWidth256Bytes,
+		enum source_format_class SourcePixelFormat,
+		unsigned int SurfaceTiling,
+		unsigned int BytePerPixel,
+		enum scan_direction_class ScanDirection,
+		unsigned int ViewportWidth,
+		unsigned int ViewportHeight,
+		unsigned int SwathWidth,
+		bool GPUVMEnable,
+		unsigned int VMMPageSize,
+		unsigned int PTEBufferSizeInRequestsLuma,
+		unsigned int PDEProcessingBufIn64KBReqs,
+		unsigned int Pitch,
+		unsigned int DCCMetaPitch,
+		unsigned int *MacroTileWidth,
+		unsigned int *MetaRowByte,
+		unsigned int *PixelPTEBytesPerRow,
+		bool *PTEBufferSizeNotExceeded,
+		unsigned int *dpte_row_height,
+		unsigned int *meta_row_height)
+{
+	unsigned int MetaRequestHeight;
+	unsigned int MetaRequestWidth;
+	unsigned int MetaSurfWidth;
+	unsigned int MetaSurfHeight;
+	unsigned int MPDEBytesFrame;
+	unsigned int MetaPTEBytesFrame;
+	unsigned int DCCMetaSurfaceBytes;
+
+	unsigned int MacroTileSizeBytes;
+	unsigned int MacroTileHeight;
+	unsigned int DPDE0BytesFrame;
+	unsigned int ExtraDPDEBytesFrame;
+	unsigned int PDEAndMetaPTEBytesFrame;
+
+	if (DCCEnable == true) {
+		MetaRequestHeight = 8 * BlockHeight256Bytes;
+		MetaRequestWidth = 8 * BlockWidth256Bytes;
+		if (ScanDirection == dm_horz) {
+			*meta_row_height = MetaRequestHeight;
+			MetaSurfWidth = dml_ceil((double) SwathWidth - 1, MetaRequestWidth)
+					+ MetaRequestWidth;
+			*MetaRowByte = MetaSurfWidth * MetaRequestHeight * BytePerPixel / 256.0;
+		} else {
+			*meta_row_height = MetaRequestWidth;
+			MetaSurfHeight = dml_ceil((double) SwathWidth - 1, MetaRequestHeight)
+					+ MetaRequestHeight;
+			*MetaRowByte = MetaSurfHeight * MetaRequestWidth * BytePerPixel / 256.0;
+		}
+		if (ScanDirection == dm_horz) {
+			DCCMetaSurfaceBytes = DCCMetaPitch
+					* (dml_ceil(ViewportHeight - 1, 64 * BlockHeight256Bytes)
+							+ 64 * BlockHeight256Bytes) * BytePerPixel
+					/ 256;
+		} else {
+			DCCMetaSurfaceBytes = DCCMetaPitch
+					* (dml_ceil(
+							(double) ViewportHeight - 1,
+							64 * BlockHeight256Bytes)
+							+ 64 * BlockHeight256Bytes) * BytePerPixel
+					/ 256;
+		}
+		if (GPUVMEnable == true) {
+			MetaPTEBytesFrame = (dml_ceil(
+					(double) (DCCMetaSurfaceBytes - VMMPageSize)
+							/ (8 * VMMPageSize),
+					1) + 1) * 64;
+			MPDEBytesFrame = 128 * (mode_lib->vba.GPUVMMaxPageTableLevels - 1);
+		} else {
+			MetaPTEBytesFrame = 0;
+			MPDEBytesFrame = 0;
+		}
+	} else {
+		MetaPTEBytesFrame = 0;
+		MPDEBytesFrame = 0;
+		*MetaRowByte = 0;
+	}
+
+	if (SurfaceTiling == dm_sw_linear || SurfaceTiling == dm_sw_gfx7_2d_thin_gl || SurfaceTiling == dm_sw_gfx7_2d_thin_l_vp) {
+		MacroTileSizeBytes = 256;
+		MacroTileHeight = BlockHeight256Bytes;
+	} else if (SurfaceTiling == dm_sw_4kb_s || SurfaceTiling == dm_sw_4kb_s_x
+			|| SurfaceTiling == dm_sw_4kb_d || SurfaceTiling == dm_sw_4kb_d_x) {
+		MacroTileSizeBytes = 4096;
+		MacroTileHeight = 4 * BlockHeight256Bytes;
+	} else if (SurfaceTiling == dm_sw_64kb_s || SurfaceTiling == dm_sw_64kb_s_t
+			|| SurfaceTiling == dm_sw_64kb_s_x || SurfaceTiling == dm_sw_64kb_d
+			|| SurfaceTiling == dm_sw_64kb_d_t || SurfaceTiling == dm_sw_64kb_d_x
+			|| SurfaceTiling == dm_sw_64kb_r_x) {
+		MacroTileSizeBytes = 65536;
+		MacroTileHeight = 16 * BlockHeight256Bytes;
+	} else {
+		MacroTileSizeBytes = 262144;
+		MacroTileHeight = 32 * BlockHeight256Bytes;
+	}
+	*MacroTileWidth = MacroTileSizeBytes / BytePerPixel / MacroTileHeight;
+
+	if (GPUVMEnable == true && mode_lib->vba.GPUVMMaxPageTableLevels > 1) {
+		if (ScanDirection == dm_horz) {
+			DPDE0BytesFrame =
+					64
+							* (dml_ceil(
+									((Pitch
+											* (dml_ceil(
+													ViewportHeight
+															- 1,
+													MacroTileHeight)
+													+ MacroTileHeight)
+											* BytePerPixel)
+											- MacroTileSizeBytes)
+											/ (8
+													* 2097152),
+									1) + 1);
+		} else {
+			DPDE0BytesFrame =
+					64
+							* (dml_ceil(
+									((Pitch
+											* (dml_ceil(
+													(double) SwathWidth
+															- 1,
+													MacroTileHeight)
+													+ MacroTileHeight)
+											* BytePerPixel)
+											- MacroTileSizeBytes)
+											/ (8
+													* 2097152),
+									1) + 1);
+		}
+		ExtraDPDEBytesFrame = 128 * (mode_lib->vba.GPUVMMaxPageTableLevels - 2);
+	} else {
+		DPDE0BytesFrame = 0;
+		ExtraDPDEBytesFrame = 0;
+	}
+
+	PDEAndMetaPTEBytesFrame = MetaPTEBytesFrame + MPDEBytesFrame + DPDE0BytesFrame
+			+ ExtraDPDEBytesFrame;
+
+	if (GPUVMEnable == true) {
+		unsigned int PTERequestSize;
+		unsigned int PixelPTEReqHeight;
+		unsigned int PixelPTEReqWidth;
+		double FractionOfPTEReturnDrop;
+		unsigned int EffectivePDEProcessingBufIn64KBReqs;
+
+		if (SurfaceTiling == dm_sw_linear) {
+			PixelPTEReqHeight = 1;
+			PixelPTEReqWidth = 8.0 * VMMPageSize / BytePerPixel;
+			PTERequestSize = 64;
+			FractionOfPTEReturnDrop = 0;
+		} else if (MacroTileSizeBytes == 4096) {
+			PixelPTEReqHeight = MacroTileHeight;
+			PixelPTEReqWidth = 8 * *MacroTileWidth;
+			PTERequestSize = 64;
+			if (ScanDirection == dm_horz)
+				FractionOfPTEReturnDrop = 0;
+			else
+				FractionOfPTEReturnDrop = 7.0 / 8;
+		} else if (VMMPageSize == 4096 && MacroTileSizeBytes > 4096) {
+			PixelPTEReqHeight = 16 * BlockHeight256Bytes;
+			PixelPTEReqWidth = 16 * BlockWidth256Bytes;
+			PTERequestSize = 128;
+			FractionOfPTEReturnDrop = 0;
+		} else {
+			PixelPTEReqHeight = MacroTileHeight;
+			PixelPTEReqWidth = 8 * *MacroTileWidth;
+			PTERequestSize = 64;
+			FractionOfPTEReturnDrop = 0;
+		}
+
+		if (SourcePixelFormat == dm_420_8 || SourcePixelFormat == dm_420_10)
+			EffectivePDEProcessingBufIn64KBReqs = PDEProcessingBufIn64KBReqs / 2;
+		else
+			EffectivePDEProcessingBufIn64KBReqs = PDEProcessingBufIn64KBReqs;
+
+		if (SurfaceTiling == dm_sw_linear) {
+			*dpte_row_height =
+					dml_min(
+							128,
+							1
+									<< (unsigned int) dml_floor(
+											dml_log2(
+													dml_min(
+															(double) PTEBufferSizeInRequestsLuma
+																	* PixelPTEReqWidth,
+															EffectivePDEProcessingBufIn64KBReqs
+																	* 65536.0
+																	/ BytePerPixel)
+															/ Pitch),
+											1));
+			*PixelPTEBytesPerRow = PTERequestSize
+					* (dml_ceil(
+							(double) (Pitch * *dpte_row_height - 1)
+									/ PixelPTEReqWidth,
+							1) + 1);
+		} else if (ScanDirection == dm_horz) {
+			*dpte_row_height = PixelPTEReqHeight;
+			*PixelPTEBytesPerRow = PTERequestSize
+					* (dml_ceil(((double) SwathWidth - 1) / PixelPTEReqWidth, 1)
+							+ 1);
+		} else {
+			*dpte_row_height = dml_min(PixelPTEReqWidth, *MacroTileWidth);
+			*PixelPTEBytesPerRow = PTERequestSize
+					* (dml_ceil(
+							((double) SwathWidth - 1)
+									/ PixelPTEReqHeight,
+							1) + 1);
+		}
+		if (*PixelPTEBytesPerRow * (1 - FractionOfPTEReturnDrop)
+				<= 64 * PTEBufferSizeInRequestsLuma) {
+			*PTEBufferSizeNotExceeded = true;
+		} else {
+			*PTEBufferSizeNotExceeded = false;
+		}
+	} else {
+		*PixelPTEBytesPerRow = 0;
+		*PTEBufferSizeNotExceeded = true;
+	}
+
+	return PDEAndMetaPTEBytesFrame;
+}
+
+static void dml20_DISPCLKDPPCLKDCFCLKDeepSleepPrefetchParametersWatermarksAndPerformanceCalculation(
+		struct display_mode_lib *mode_lib)
+{
+	unsigned int j, k;
+
+	mode_lib->vba.WritebackDISPCLK = 0.0;
+	mode_lib->vba.DISPCLKWithRamping = 0;
+	mode_lib->vba.DISPCLKWithoutRamping = 0;
+	mode_lib->vba.GlobalDPPCLK = 0.0;
+
+	// dml_ml->vba.DISPCLK and dml_ml->vba.DPPCLK Calculation
+	//
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.WritebackEnable[k]) {
+			mode_lib->vba.WritebackDISPCLK =
+					dml_max(
+							mode_lib->vba.WritebackDISPCLK,
+							CalculateWriteBackDISPCLK(
+									mode_lib->vba.WritebackPixelFormat[k],
+									mode_lib->vba.PixelClock[k],
+									mode_lib->vba.WritebackHRatio[k],
+									mode_lib->vba.WritebackVRatio[k],
+									mode_lib->vba.WritebackLumaHTaps[k],
+									mode_lib->vba.WritebackLumaVTaps[k],
+									mode_lib->vba.WritebackChromaHTaps[k],
+									mode_lib->vba.WritebackChromaVTaps[k],
+									mode_lib->vba.WritebackDestinationWidth[k],
+									mode_lib->vba.HTotal[k],
+									mode_lib->vba.WritebackChromaLineBufferWidth));
+		}
+	}
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.HRatio[k] > 1) {
+			mode_lib->vba.PSCL_THROUGHPUT_LUMA[k] = dml_min(
+					mode_lib->vba.MaxDCHUBToPSCLThroughput,
+					mode_lib->vba.MaxPSCLToLBThroughput
+							* mode_lib->vba.HRatio[k]
+							/ dml_ceil(
+									mode_lib->vba.htaps[k]
+											/ 6.0,
+									1));
+		} else {
+			mode_lib->vba.PSCL_THROUGHPUT_LUMA[k] = dml_min(
+					mode_lib->vba.MaxDCHUBToPSCLThroughput,
+					mode_lib->vba.MaxPSCLToLBThroughput);
+		}
+
+		mode_lib->vba.DPPCLKUsingSingleDPPLuma =
+				mode_lib->vba.PixelClock[k]
+						* dml_max(
+								mode_lib->vba.vtaps[k] / 6.0
+										* dml_min(
+												1.0,
+												mode_lib->vba.HRatio[k]),
+								dml_max(
+										mode_lib->vba.HRatio[k]
+												* mode_lib->vba.VRatio[k]
+												/ mode_lib->vba.PSCL_THROUGHPUT_LUMA[k],
+										1.0));
+
+		if ((mode_lib->vba.htaps[k] > 6 || mode_lib->vba.vtaps[k] > 6)
+				&& mode_lib->vba.DPPCLKUsingSingleDPPLuma
+						< 2 * mode_lib->vba.PixelClock[k]) {
+			mode_lib->vba.DPPCLKUsingSingleDPPLuma = 2 * mode_lib->vba.PixelClock[k];
+		}
+
+		if ((mode_lib->vba.SourcePixelFormat[k] != dm_420_8
+				&& mode_lib->vba.SourcePixelFormat[k] != dm_420_10)) {
+			mode_lib->vba.PSCL_THROUGHPUT_CHROMA[k] = 0.0;
+			mode_lib->vba.DPPCLKUsingSingleDPP[k] =
+					mode_lib->vba.DPPCLKUsingSingleDPPLuma;
+		} else {
+			if (mode_lib->vba.HRatio[k] > 1) {
+				mode_lib->vba.PSCL_THROUGHPUT_CHROMA[k] =
+						dml_min(
+								mode_lib->vba.MaxDCHUBToPSCLThroughput,
+								mode_lib->vba.MaxPSCLToLBThroughput
+										* mode_lib->vba.HRatio[k]
+										/ 2
+										/ dml_ceil(
+												mode_lib->vba.HTAPsChroma[k]
+														/ 6.0,
+												1.0));
+			} else {
+				mode_lib->vba.PSCL_THROUGHPUT_CHROMA[k] = dml_min(
+						mode_lib->vba.MaxDCHUBToPSCLThroughput,
+						mode_lib->vba.MaxPSCLToLBThroughput);
+			}
+			mode_lib->vba.DPPCLKUsingSingleDPPChroma =
+					mode_lib->vba.PixelClock[k]
+							* dml_max(
+									mode_lib->vba.VTAPsChroma[k]
+											/ 6.0
+											* dml_min(
+													1.0,
+													mode_lib->vba.HRatio[k]
+															/ 2),
+									dml_max(
+											mode_lib->vba.HRatio[k]
+													* mode_lib->vba.VRatio[k]
+													/ 4
+													/ mode_lib->vba.PSCL_THROUGHPUT_CHROMA[k],
+											1.0));
+
+			if ((mode_lib->vba.HTAPsChroma[k] > 6 || mode_lib->vba.VTAPsChroma[k] > 6)
+					&& mode_lib->vba.DPPCLKUsingSingleDPPChroma
+							< 2 * mode_lib->vba.PixelClock[k]) {
+				mode_lib->vba.DPPCLKUsingSingleDPPChroma = 2
+						* mode_lib->vba.PixelClock[k];
+			}
+
+			mode_lib->vba.DPPCLKUsingSingleDPP[k] = dml_max(
+					mode_lib->vba.DPPCLKUsingSingleDPPLuma,
+					mode_lib->vba.DPPCLKUsingSingleDPPChroma);
+		}
+	}
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.BlendingAndTiming[k] != k)
+			continue;
+		if (mode_lib->vba.ODMCombineEnabled[k]) {
+			mode_lib->vba.DISPCLKWithRamping =
+					dml_max(
+							mode_lib->vba.DISPCLKWithRamping,
+							mode_lib->vba.PixelClock[k] / 2
+									* (1
+											+ mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading
+													/ 100)
+									* (1
+											+ mode_lib->vba.DISPCLKRampingMargin
+													/ 100));
+			mode_lib->vba.DISPCLKWithoutRamping =
+					dml_max(
+							mode_lib->vba.DISPCLKWithoutRamping,
+							mode_lib->vba.PixelClock[k] / 2
+									* (1
+											+ mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading
+													/ 100));
+		} else if (!mode_lib->vba.ODMCombineEnabled[k]) {
+			mode_lib->vba.DISPCLKWithRamping =
+					dml_max(
+							mode_lib->vba.DISPCLKWithRamping,
+							mode_lib->vba.PixelClock[k]
+									* (1
+											+ mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading
+													/ 100)
+									* (1
+											+ mode_lib->vba.DISPCLKRampingMargin
+													/ 100));
+			mode_lib->vba.DISPCLKWithoutRamping =
+					dml_max(
+							mode_lib->vba.DISPCLKWithoutRamping,
+							mode_lib->vba.PixelClock[k]
+									* (1
+											+ mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading
+													/ 100));
+		}
+	}
+
+	mode_lib->vba.DISPCLKWithRamping = dml_max(
+			mode_lib->vba.DISPCLKWithRamping,
+			mode_lib->vba.WritebackDISPCLK);
+	mode_lib->vba.DISPCLKWithoutRamping = dml_max(
+			mode_lib->vba.DISPCLKWithoutRamping,
+			mode_lib->vba.WritebackDISPCLK);
+
+	ASSERT(mode_lib->vba.DISPCLKDPPCLKVCOSpeed != 0);
+	mode_lib->vba.DISPCLKWithRampingRoundedToDFSGranularity = RoundToDFSGranularityUp(
+			mode_lib->vba.DISPCLKWithRamping,
+			mode_lib->vba.DISPCLKDPPCLKVCOSpeed);
+	mode_lib->vba.DISPCLKWithoutRampingRoundedToDFSGranularity = RoundToDFSGranularityUp(
+			mode_lib->vba.DISPCLKWithoutRamping,
+			mode_lib->vba.DISPCLKDPPCLKVCOSpeed);
+	mode_lib->vba.MaxDispclkRoundedToDFSGranularity = RoundToDFSGranularityDown(
+			mode_lib->vba.soc.clock_limits[mode_lib->vba.soc.num_states].dispclk_mhz,
+			mode_lib->vba.DISPCLKDPPCLKVCOSpeed);
+	if (mode_lib->vba.DISPCLKWithoutRampingRoundedToDFSGranularity
+			> mode_lib->vba.MaxDispclkRoundedToDFSGranularity) {
+		mode_lib->vba.DISPCLK_calculated =
+				mode_lib->vba.DISPCLKWithoutRampingRoundedToDFSGranularity;
+	} else if (mode_lib->vba.DISPCLKWithRampingRoundedToDFSGranularity
+			> mode_lib->vba.MaxDispclkRoundedToDFSGranularity) {
+		mode_lib->vba.DISPCLK_calculated = mode_lib->vba.MaxDispclkRoundedToDFSGranularity;
+	} else {
+		mode_lib->vba.DISPCLK_calculated =
+				mode_lib->vba.DISPCLKWithRampingRoundedToDFSGranularity;
+	}
+	DTRACE("   dispclk_mhz (calculated) = %f", mode_lib->vba.DISPCLK_calculated);
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.DPPPerPlane[k] == 0) {
+			mode_lib->vba.DPPCLK_calculated[k] = 0;
+		} else {
+			mode_lib->vba.DPPCLK_calculated[k] = mode_lib->vba.DPPCLKUsingSingleDPP[k]
+					/ mode_lib->vba.DPPPerPlane[k]
+					* (1 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100);
+		}
+		mode_lib->vba.GlobalDPPCLK = dml_max(
+				mode_lib->vba.GlobalDPPCLK,
+				mode_lib->vba.DPPCLK_calculated[k]);
+	}
+	mode_lib->vba.GlobalDPPCLK = RoundToDFSGranularityUp(
+			mode_lib->vba.GlobalDPPCLK,
+			mode_lib->vba.DISPCLKDPPCLKVCOSpeed);
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		mode_lib->vba.DPPCLK_calculated[k] = mode_lib->vba.GlobalDPPCLK / 255
+				* dml_ceil(
+						mode_lib->vba.DPPCLK_calculated[k] * 255
+								/ mode_lib->vba.GlobalDPPCLK,
+						1);
+		DTRACE("   dppclk_mhz[%i] (calculated) = %f", k, mode_lib->vba.DPPCLK_calculated[k]);
+	}
+
+	// Urgent Watermark
+	mode_lib->vba.DCCEnabledAnyPlane = false;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k)
+		if (mode_lib->vba.DCCEnable[k])
+			mode_lib->vba.DCCEnabledAnyPlane = true;
+
+	mode_lib->vba.ReturnBandwidthToDCN = dml_min(
+			mode_lib->vba.ReturnBusWidth * mode_lib->vba.DCFCLK,
+			mode_lib->vba.FabricAndDRAMBandwidth * 1000)
+			* mode_lib->vba.PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelDataOnly / 100;
+
+	mode_lib->vba.ReturnBW = mode_lib->vba.ReturnBandwidthToDCN;
+	mode_lib->vba.ReturnBW = adjust_ReturnBW(
+			mode_lib,
+			mode_lib->vba.ReturnBW,
+			mode_lib->vba.DCCEnabledAnyPlane,
+			mode_lib->vba.ReturnBandwidthToDCN);
+
+	// Let's do this calculation again??
+	mode_lib->vba.ReturnBandwidthToDCN = dml_min(
+			mode_lib->vba.ReturnBusWidth * mode_lib->vba.DCFCLK,
+			mode_lib->vba.FabricAndDRAMBandwidth * 1000);
+	mode_lib->vba.ReturnBW = adjust_ReturnBW(
+			mode_lib,
+			mode_lib->vba.ReturnBW,
+			mode_lib->vba.DCCEnabledAnyPlane,
+			mode_lib->vba.ReturnBandwidthToDCN);
+
+	DTRACE("   dcfclk_mhz         = %f", mode_lib->vba.DCFCLK);
+	DTRACE("   return_bw_to_dcn   = %f", mode_lib->vba.ReturnBandwidthToDCN);
+	DTRACE("   return_bus_bw      = %f", mode_lib->vba.ReturnBW);
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		bool MainPlaneDoesODMCombine = false;
+
+		if (mode_lib->vba.SourceScan[k] == dm_horz)
+			mode_lib->vba.SwathWidthSingleDPPY[k] = mode_lib->vba.ViewportWidth[k];
+		else
+			mode_lib->vba.SwathWidthSingleDPPY[k] = mode_lib->vba.ViewportHeight[k];
+
+		if (mode_lib->vba.ODMCombineEnabled[k] == dm_odm_combine_mode_2to1)
+			MainPlaneDoesODMCombine = true;
+		for (j = 0; j < mode_lib->vba.NumberOfActivePlanes; ++j)
+			if (mode_lib->vba.BlendingAndTiming[k] == j
+					&& mode_lib->vba.ODMCombineEnabled[k] == dm_odm_combine_mode_2to1)
+				MainPlaneDoesODMCombine = true;
+
+		if (MainPlaneDoesODMCombine == true)
+			mode_lib->vba.SwathWidthY[k] = dml_min(
+					(double) mode_lib->vba.SwathWidthSingleDPPY[k],
+					dml_round(
+							mode_lib->vba.HActive[k] / 2.0
+									* mode_lib->vba.HRatio[k]));
+		else {
+			if (mode_lib->vba.DPPPerPlane[k] == 0) {
+				mode_lib->vba.SwathWidthY[k] = 0;
+			} else {
+				mode_lib->vba.SwathWidthY[k] = mode_lib->vba.SwathWidthSingleDPPY[k]
+						/ mode_lib->vba.DPPPerPlane[k];
+			}
+		}
+	}
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.SourcePixelFormat[k] == dm_444_64) {
+			mode_lib->vba.BytePerPixelDETY[k] = 8;
+			mode_lib->vba.BytePerPixelDETC[k] = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_32) {
+			mode_lib->vba.BytePerPixelDETY[k] = 4;
+			mode_lib->vba.BytePerPixelDETC[k] = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_16) {
+			mode_lib->vba.BytePerPixelDETY[k] = 2;
+			mode_lib->vba.BytePerPixelDETC[k] = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_8) {
+			mode_lib->vba.BytePerPixelDETY[k] = 1;
+			mode_lib->vba.BytePerPixelDETC[k] = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_8) {
+			mode_lib->vba.BytePerPixelDETY[k] = 1;
+			mode_lib->vba.BytePerPixelDETC[k] = 2;
+		} else { // dm_420_10
+			mode_lib->vba.BytePerPixelDETY[k] = 4.0 / 3.0;
+			mode_lib->vba.BytePerPixelDETC[k] = 8.0 / 3.0;
+		}
+	}
+
+	mode_lib->vba.TotalDataReadBandwidth = 0.0;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		mode_lib->vba.ReadBandwidthPlaneLuma[k] = mode_lib->vba.SwathWidthSingleDPPY[k]
+				* dml_ceil(mode_lib->vba.BytePerPixelDETY[k], 1)
+				/ (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k])
+				* mode_lib->vba.VRatio[k];
+		mode_lib->vba.ReadBandwidthPlaneChroma[k] = mode_lib->vba.SwathWidthSingleDPPY[k]
+				/ 2 * dml_ceil(mode_lib->vba.BytePerPixelDETC[k], 2)
+				/ (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k])
+				* mode_lib->vba.VRatio[k] / 2;
+		DTRACE(
+				"   read_bw[%i] = %fBps",
+				k,
+				mode_lib->vba.ReadBandwidthPlaneLuma[k]
+						+ mode_lib->vba.ReadBandwidthPlaneChroma[k]);
+		mode_lib->vba.TotalDataReadBandwidth += mode_lib->vba.ReadBandwidthPlaneLuma[k]
+				+ mode_lib->vba.ReadBandwidthPlaneChroma[k];
+	}
+
+	mode_lib->vba.TotalDCCActiveDPP = 0;
+	mode_lib->vba.TotalActiveDPP = 0;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		mode_lib->vba.TotalActiveDPP = mode_lib->vba.TotalActiveDPP
+				+ mode_lib->vba.DPPPerPlane[k];
+		if (mode_lib->vba.DCCEnable[k])
+			mode_lib->vba.TotalDCCActiveDPP = mode_lib->vba.TotalDCCActiveDPP
+					+ mode_lib->vba.DPPPerPlane[k];
+	}
+
+	mode_lib->vba.UrgentRoundTripAndOutOfOrderLatency =
+			(mode_lib->vba.RoundTripPingLatencyCycles + 32) / mode_lib->vba.DCFCLK
+					+ mode_lib->vba.UrgentOutOfOrderReturnPerChannelPixelDataOnly
+							* mode_lib->vba.NumberOfChannels
+							/ mode_lib->vba.ReturnBW;
+
+	mode_lib->vba.LastPixelOfLineExtraWatermark = 0;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		double DataFabricLineDeliveryTimeLuma, DataFabricLineDeliveryTimeChroma;
+
+		if (mode_lib->vba.VRatio[k] <= 1.0)
+			mode_lib->vba.DisplayPipeLineDeliveryTimeLuma[k] =
+					(double) mode_lib->vba.SwathWidthY[k]
+							* mode_lib->vba.DPPPerPlane[k]
+							/ mode_lib->vba.HRatio[k]
+							/ mode_lib->vba.PixelClock[k];
+		else
+			mode_lib->vba.DisplayPipeLineDeliveryTimeLuma[k] =
+					(double) mode_lib->vba.SwathWidthY[k]
+							/ mode_lib->vba.PSCL_THROUGHPUT_LUMA[k]
+							/ mode_lib->vba.DPPCLK[k];
+
+		DataFabricLineDeliveryTimeLuma = mode_lib->vba.SwathWidthSingleDPPY[k]
+				* mode_lib->vba.SwathHeightY[k]
+				* dml_ceil(mode_lib->vba.BytePerPixelDETY[k], 1)
+				/ (mode_lib->vba.ReturnBW * mode_lib->vba.ReadBandwidthPlaneLuma[k]
+						/ mode_lib->vba.TotalDataReadBandwidth);
+		mode_lib->vba.LastPixelOfLineExtraWatermark = dml_max(
+				mode_lib->vba.LastPixelOfLineExtraWatermark,
+				DataFabricLineDeliveryTimeLuma
+						- mode_lib->vba.DisplayPipeLineDeliveryTimeLuma[k]);
+
+		if (mode_lib->vba.BytePerPixelDETC[k] == 0)
+			mode_lib->vba.DisplayPipeLineDeliveryTimeChroma[k] = 0.0;
+		else if (mode_lib->vba.VRatio[k] / 2.0 <= 1.0)
+			mode_lib->vba.DisplayPipeLineDeliveryTimeChroma[k] =
+					mode_lib->vba.SwathWidthY[k] / 2.0
+							* mode_lib->vba.DPPPerPlane[k]
+							/ (mode_lib->vba.HRatio[k] / 2.0)
+							/ mode_lib->vba.PixelClock[k];
+		else
+			mode_lib->vba.DisplayPipeLineDeliveryTimeChroma[k] =
+					mode_lib->vba.SwathWidthY[k] / 2.0
+							/ mode_lib->vba.PSCL_THROUGHPUT_CHROMA[k]
+							/ mode_lib->vba.DPPCLK[k];
+
+		DataFabricLineDeliveryTimeChroma = mode_lib->vba.SwathWidthSingleDPPY[k] / 2.0
+				* mode_lib->vba.SwathHeightC[k]
+				* dml_ceil(mode_lib->vba.BytePerPixelDETC[k], 2)
+				/ (mode_lib->vba.ReturnBW
+						* mode_lib->vba.ReadBandwidthPlaneChroma[k]
+						/ mode_lib->vba.TotalDataReadBandwidth);
+		mode_lib->vba.LastPixelOfLineExtraWatermark =
+				dml_max(
+						mode_lib->vba.LastPixelOfLineExtraWatermark,
+						DataFabricLineDeliveryTimeChroma
+								- mode_lib->vba.DisplayPipeLineDeliveryTimeChroma[k]);
+	}
+
+	mode_lib->vba.UrgentExtraLatency = mode_lib->vba.UrgentRoundTripAndOutOfOrderLatency
+			+ (mode_lib->vba.TotalActiveDPP * mode_lib->vba.PixelChunkSizeInKByte
+					+ mode_lib->vba.TotalDCCActiveDPP
+							* mode_lib->vba.MetaChunkSize) * 1024.0
+					/ mode_lib->vba.ReturnBW;
+
+	if (mode_lib->vba.GPUVMEnable)
+		mode_lib->vba.UrgentExtraLatency += mode_lib->vba.TotalActiveDPP
+				* mode_lib->vba.PTEGroupSize / mode_lib->vba.ReturnBW;
+
+	mode_lib->vba.UrgentWatermark = mode_lib->vba.UrgentLatencyPixelDataOnly
+			+ mode_lib->vba.LastPixelOfLineExtraWatermark
+			+ mode_lib->vba.UrgentExtraLatency;
+
+	DTRACE("   urgent_extra_latency = %fus", mode_lib->vba.UrgentExtraLatency);
+	DTRACE("   wm_urgent = %fus", mode_lib->vba.UrgentWatermark);
+
+	mode_lib->vba.UrgentLatency = mode_lib->vba.UrgentLatencyPixelDataOnly;
+
+	mode_lib->vba.TotalActiveWriteback = 0;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.WritebackEnable[k])
+			mode_lib->vba.TotalActiveWriteback = mode_lib->vba.TotalActiveWriteback + mode_lib->vba.ActiveWritebacksPerPlane[k];
+	}
+
+	if (mode_lib->vba.TotalActiveWriteback <= 1)
+		mode_lib->vba.WritebackUrgentWatermark = mode_lib->vba.WritebackLatency;
+	else
+		mode_lib->vba.WritebackUrgentWatermark = mode_lib->vba.WritebackLatency
+				+ mode_lib->vba.WritebackChunkSize * 1024.0 / 32
+						/ mode_lib->vba.SOCCLK;
+
+	DTRACE("   wm_wb_urgent = %fus", mode_lib->vba.WritebackUrgentWatermark);
+
+	// NB P-State/DRAM Clock Change Watermark
+	mode_lib->vba.DRAMClockChangeWatermark = mode_lib->vba.DRAMClockChangeLatency
+			+ mode_lib->vba.UrgentWatermark;
+
+	DTRACE("   wm_pstate_change = %fus", mode_lib->vba.DRAMClockChangeWatermark);
+
+	DTRACE("   calculating wb pstate watermark");
+	DTRACE("      total wb outputs %d", mode_lib->vba.TotalActiveWriteback);
+	DTRACE("      socclk frequency %f Mhz", mode_lib->vba.SOCCLK);
+
+	if (mode_lib->vba.TotalActiveWriteback <= 1)
+		mode_lib->vba.WritebackDRAMClockChangeWatermark =
+				mode_lib->vba.DRAMClockChangeLatency
+						+ mode_lib->vba.WritebackLatency;
+	else
+		mode_lib->vba.WritebackDRAMClockChangeWatermark =
+				mode_lib->vba.DRAMClockChangeLatency
+						+ mode_lib->vba.WritebackLatency
+						+ mode_lib->vba.WritebackChunkSize * 1024.0 / 32
+								/ mode_lib->vba.SOCCLK;
+
+	DTRACE("   wm_wb_pstate %fus", mode_lib->vba.WritebackDRAMClockChangeWatermark);
+
+	// Stutter Efficiency
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		mode_lib->vba.LinesInDETY[k] = mode_lib->vba.DETBufferSizeY[k]
+				/ mode_lib->vba.BytePerPixelDETY[k] / mode_lib->vba.SwathWidthY[k];
+		mode_lib->vba.LinesInDETYRoundedDownToSwath[k] = dml_floor(
+				mode_lib->vba.LinesInDETY[k],
+				mode_lib->vba.SwathHeightY[k]);
+		mode_lib->vba.FullDETBufferingTimeY[k] =
+				mode_lib->vba.LinesInDETYRoundedDownToSwath[k]
+						* (mode_lib->vba.HTotal[k]
+								/ mode_lib->vba.PixelClock[k])
+						/ mode_lib->vba.VRatio[k];
+		if (mode_lib->vba.BytePerPixelDETC[k] > 0) {
+			mode_lib->vba.LinesInDETC[k] = mode_lib->vba.DETBufferSizeC[k]
+					/ mode_lib->vba.BytePerPixelDETC[k]
+					/ (mode_lib->vba.SwathWidthY[k] / 2);
+			mode_lib->vba.LinesInDETCRoundedDownToSwath[k] = dml_floor(
+					mode_lib->vba.LinesInDETC[k],
+					mode_lib->vba.SwathHeightC[k]);
+			mode_lib->vba.FullDETBufferingTimeC[k] =
+					mode_lib->vba.LinesInDETCRoundedDownToSwath[k]
+							* (mode_lib->vba.HTotal[k]
+									/ mode_lib->vba.PixelClock[k])
+							/ (mode_lib->vba.VRatio[k] / 2);
+		} else {
+			mode_lib->vba.LinesInDETC[k] = 0;
+			mode_lib->vba.LinesInDETCRoundedDownToSwath[k] = 0;
+			mode_lib->vba.FullDETBufferingTimeC[k] = 999999;
+		}
+	}
+
+	mode_lib->vba.MinFullDETBufferingTime = 999999.0;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.FullDETBufferingTimeY[k]
+				< mode_lib->vba.MinFullDETBufferingTime) {
+			mode_lib->vba.MinFullDETBufferingTime =
+					mode_lib->vba.FullDETBufferingTimeY[k];
+			mode_lib->vba.FrameTimeForMinFullDETBufferingTime =
+					(double) mode_lib->vba.VTotal[k] * mode_lib->vba.HTotal[k]
+							/ mode_lib->vba.PixelClock[k];
+		}
+		if (mode_lib->vba.FullDETBufferingTimeC[k]
+				< mode_lib->vba.MinFullDETBufferingTime) {
+			mode_lib->vba.MinFullDETBufferingTime =
+					mode_lib->vba.FullDETBufferingTimeC[k];
+			mode_lib->vba.FrameTimeForMinFullDETBufferingTime =
+					(double) mode_lib->vba.VTotal[k] * mode_lib->vba.HTotal[k]
+							/ mode_lib->vba.PixelClock[k];
+		}
+	}
+
+	mode_lib->vba.AverageReadBandwidthGBytePerSecond = 0.0;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.DCCEnable[k]) {
+			mode_lib->vba.AverageReadBandwidthGBytePerSecond =
+					mode_lib->vba.AverageReadBandwidthGBytePerSecond
+							+ mode_lib->vba.ReadBandwidthPlaneLuma[k]
+									/ mode_lib->vba.DCCRate[k]
+									/ 1000
+							+ mode_lib->vba.ReadBandwidthPlaneChroma[k]
+									/ mode_lib->vba.DCCRate[k]
+									/ 1000;
+		} else {
+			mode_lib->vba.AverageReadBandwidthGBytePerSecond =
+					mode_lib->vba.AverageReadBandwidthGBytePerSecond
+							+ mode_lib->vba.ReadBandwidthPlaneLuma[k]
+									/ 1000
+							+ mode_lib->vba.ReadBandwidthPlaneChroma[k]
+									/ 1000;
+		}
+		if (mode_lib->vba.DCCEnable[k]) {
+			mode_lib->vba.AverageReadBandwidthGBytePerSecond =
+					mode_lib->vba.AverageReadBandwidthGBytePerSecond
+							+ mode_lib->vba.ReadBandwidthPlaneLuma[k]
+									/ 1000 / 256
+							+ mode_lib->vba.ReadBandwidthPlaneChroma[k]
+									/ 1000 / 256;
+		}
+		if (mode_lib->vba.GPUVMEnable) {
+			mode_lib->vba.AverageReadBandwidthGBytePerSecond =
+					mode_lib->vba.AverageReadBandwidthGBytePerSecond
+							+ mode_lib->vba.ReadBandwidthPlaneLuma[k]
+									/ 1000 / 512
+							+ mode_lib->vba.ReadBandwidthPlaneChroma[k]
+									/ 1000 / 512;
+		}
+	}
+
+	mode_lib->vba.PartOfBurstThatFitsInROB =
+			dml_min(
+					mode_lib->vba.MinFullDETBufferingTime
+							* mode_lib->vba.TotalDataReadBandwidth,
+					mode_lib->vba.ROBBufferSizeInKByte * 1024
+							* mode_lib->vba.TotalDataReadBandwidth
+							/ (mode_lib->vba.AverageReadBandwidthGBytePerSecond
+									* 1000));
+	mode_lib->vba.StutterBurstTime = mode_lib->vba.PartOfBurstThatFitsInROB
+			* (mode_lib->vba.AverageReadBandwidthGBytePerSecond * 1000)
+			/ mode_lib->vba.TotalDataReadBandwidth / mode_lib->vba.ReturnBW
+			+ (mode_lib->vba.MinFullDETBufferingTime
+					* mode_lib->vba.TotalDataReadBandwidth
+					- mode_lib->vba.PartOfBurstThatFitsInROB)
+					/ (mode_lib->vba.DCFCLK * 64);
+	if (mode_lib->vba.TotalActiveWriteback == 0) {
+		mode_lib->vba.StutterEfficiencyNotIncludingVBlank = (1
+				- (mode_lib->vba.SRExitTime + mode_lib->vba.StutterBurstTime)
+						/ mode_lib->vba.MinFullDETBufferingTime) * 100;
+	} else {
+		mode_lib->vba.StutterEfficiencyNotIncludingVBlank = 0;
+	}
+
+	mode_lib->vba.SmallestVBlank = 999999;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.SynchronizedVBlank || mode_lib->vba.NumberOfActivePlanes == 1) {
+			mode_lib->vba.VBlankTime = (double) (mode_lib->vba.VTotal[k]
+					- mode_lib->vba.VActive[k]) * mode_lib->vba.HTotal[k]
+					/ mode_lib->vba.PixelClock[k];
+		} else {
+			mode_lib->vba.VBlankTime = 0;
+		}
+		mode_lib->vba.SmallestVBlank = dml_min(
+				mode_lib->vba.SmallestVBlank,
+				mode_lib->vba.VBlankTime);
+	}
+
+	mode_lib->vba.StutterEfficiency = (mode_lib->vba.StutterEfficiencyNotIncludingVBlank / 100
+			* (mode_lib->vba.FrameTimeForMinFullDETBufferingTime
+					- mode_lib->vba.SmallestVBlank)
+			+ mode_lib->vba.SmallestVBlank)
+			/ mode_lib->vba.FrameTimeForMinFullDETBufferingTime * 100;
+
+	// dml_ml->vba.DCFCLK Deep Sleep
+	mode_lib->vba.DCFCLKDeepSleep = 8.0;
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; k++) {
+		if (mode_lib->vba.BytePerPixelDETC[k] > 0) {
+			mode_lib->vba.DCFCLKDeepSleepPerPlane[k] =
+					dml_max(
+							1.1 * mode_lib->vba.SwathWidthY[k]
+									* dml_ceil(
+											mode_lib->vba.BytePerPixelDETY[k],
+											1) / 32
+									/ mode_lib->vba.DisplayPipeLineDeliveryTimeLuma[k],
+							1.1 * mode_lib->vba.SwathWidthY[k] / 2.0
+									* dml_ceil(
+											mode_lib->vba.BytePerPixelDETC[k],
+											2) / 32
+									/ mode_lib->vba.DisplayPipeLineDeliveryTimeChroma[k]);
+		} else
+			mode_lib->vba.DCFCLKDeepSleepPerPlane[k] = 1.1 * mode_lib->vba.SwathWidthY[k]
+					* dml_ceil(mode_lib->vba.BytePerPixelDETY[k], 1) / 64.0
+					/ mode_lib->vba.DisplayPipeLineDeliveryTimeLuma[k];
+		mode_lib->vba.DCFCLKDeepSleepPerPlane[k] = dml_max(
+				mode_lib->vba.DCFCLKDeepSleepPerPlane[k],
+				mode_lib->vba.PixelClock[k] / 16.0);
+		mode_lib->vba.DCFCLKDeepSleep = dml_max(
+				mode_lib->vba.DCFCLKDeepSleep,
+				mode_lib->vba.DCFCLKDeepSleepPerPlane[k]);
+
+		DTRACE(
+				"   dcfclk_deepsleep_per_plane[%i] = %fMHz",
+				k,
+				mode_lib->vba.DCFCLKDeepSleepPerPlane[k]);
+	}
+
+	DTRACE("   dcfclk_deepsleep_mhz = %fMHz", mode_lib->vba.DCFCLKDeepSleep);
+
+	// Stutter Watermark
+	mode_lib->vba.StutterExitWatermark = mode_lib->vba.SRExitTime
+			+ mode_lib->vba.LastPixelOfLineExtraWatermark
+			+ mode_lib->vba.UrgentExtraLatency + 10 / mode_lib->vba.DCFCLKDeepSleep;
+	mode_lib->vba.StutterEnterPlusExitWatermark = mode_lib->vba.SREnterPlusExitTime
+			+ mode_lib->vba.LastPixelOfLineExtraWatermark
+			+ mode_lib->vba.UrgentExtraLatency;
+
+	DTRACE("   wm_cstate_exit       = %fus", mode_lib->vba.StutterExitWatermark);
+	DTRACE("   wm_cstate_enter_exit = %fus", mode_lib->vba.StutterEnterPlusExitWatermark);
+
+	// Urgent Latency Supported
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		mode_lib->vba.EffectiveDETPlusLBLinesLuma =
+				dml_floor(
+						mode_lib->vba.LinesInDETY[k]
+								+ dml_min(
+										mode_lib->vba.LinesInDETY[k]
+												* mode_lib->vba.DPPCLK[k]
+												* mode_lib->vba.BytePerPixelDETY[k]
+												* mode_lib->vba.PSCL_THROUGHPUT_LUMA[k]
+												/ (mode_lib->vba.ReturnBW
+														/ mode_lib->vba.DPPPerPlane[k]),
+										(double) mode_lib->vba.EffectiveLBLatencyHidingSourceLinesLuma),
+						mode_lib->vba.SwathHeightY[k]);
+
+		mode_lib->vba.UrgentLatencySupportUsLuma = mode_lib->vba.EffectiveDETPlusLBLinesLuma
+				* (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k])
+				/ mode_lib->vba.VRatio[k]
+				- mode_lib->vba.EffectiveDETPlusLBLinesLuma
+						* mode_lib->vba.SwathWidthY[k]
+						* mode_lib->vba.BytePerPixelDETY[k]
+						/ (mode_lib->vba.ReturnBW
+								/ mode_lib->vba.DPPPerPlane[k]);
+
+		if (mode_lib->vba.BytePerPixelDETC[k] > 0) {
+			mode_lib->vba.EffectiveDETPlusLBLinesChroma =
+					dml_floor(
+							mode_lib->vba.LinesInDETC[k]
+									+ dml_min(
+											mode_lib->vba.LinesInDETC[k]
+													* mode_lib->vba.DPPCLK[k]
+													* mode_lib->vba.BytePerPixelDETC[k]
+													* mode_lib->vba.PSCL_THROUGHPUT_CHROMA[k]
+													/ (mode_lib->vba.ReturnBW
+															/ mode_lib->vba.DPPPerPlane[k]),
+											(double) mode_lib->vba.EffectiveLBLatencyHidingSourceLinesChroma),
+							mode_lib->vba.SwathHeightC[k]);
+			mode_lib->vba.UrgentLatencySupportUsChroma =
+					mode_lib->vba.EffectiveDETPlusLBLinesChroma
+							* (mode_lib->vba.HTotal[k]
+									/ mode_lib->vba.PixelClock[k])
+							/ (mode_lib->vba.VRatio[k] / 2)
+							- mode_lib->vba.EffectiveDETPlusLBLinesChroma
+									* (mode_lib->vba.SwathWidthY[k]
+											/ 2)
+									* mode_lib->vba.BytePerPixelDETC[k]
+									/ (mode_lib->vba.ReturnBW
+											/ mode_lib->vba.DPPPerPlane[k]);
+			mode_lib->vba.UrgentLatencySupportUs[k] = dml_min(
+					mode_lib->vba.UrgentLatencySupportUsLuma,
+					mode_lib->vba.UrgentLatencySupportUsChroma);
+		} else {
+			mode_lib->vba.UrgentLatencySupportUs[k] =
+					mode_lib->vba.UrgentLatencySupportUsLuma;
+		}
+	}
+
+	mode_lib->vba.MinUrgentLatencySupportUs = 999999;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		mode_lib->vba.MinUrgentLatencySupportUs = dml_min(
+				mode_lib->vba.MinUrgentLatencySupportUs,
+				mode_lib->vba.UrgentLatencySupportUs[k]);
+	}
+
+	// Non-Urgent Latency Tolerance
+	mode_lib->vba.NonUrgentLatencyTolerance = mode_lib->vba.MinUrgentLatencySupportUs
+			- mode_lib->vba.UrgentWatermark;
+
+	// DSCCLK
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if ((mode_lib->vba.BlendingAndTiming[k] != k) || !mode_lib->vba.DSCEnabled[k]) {
+			mode_lib->vba.DSCCLK_calculated[k] = 0.0;
+		} else {
+			if (mode_lib->vba.OutputFormat[k] == dm_420
+					|| mode_lib->vba.OutputFormat[k] == dm_n422)
+				mode_lib->vba.DSCFormatFactor = 2;
+			else
+				mode_lib->vba.DSCFormatFactor = 1;
+			if (mode_lib->vba.ODMCombineEnabled[k])
+				mode_lib->vba.DSCCLK_calculated[k] =
+						mode_lib->vba.PixelClockBackEnd[k] / 6
+								/ mode_lib->vba.DSCFormatFactor
+								/ (1
+										- mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading
+												/ 100);
+			else
+				mode_lib->vba.DSCCLK_calculated[k] =
+						mode_lib->vba.PixelClockBackEnd[k] / 3
+								/ mode_lib->vba.DSCFormatFactor
+								/ (1
+										- mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading
+												/ 100);
+		}
+	}
+
+	// DSC Delay
+	// TODO
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		double bpp = mode_lib->vba.OutputBpp[k];
+		unsigned int slices = mode_lib->vba.NumberOfDSCSlices[k];
+
+		if (mode_lib->vba.DSCEnabled[k] && bpp != 0) {
+			if (!mode_lib->vba.ODMCombineEnabled[k]) {
+				mode_lib->vba.DSCDelay[k] =
+						dscceComputeDelay(
+								mode_lib->vba.DSCInputBitPerComponent[k],
+								bpp,
+								dml_ceil(
+										(double) mode_lib->vba.HActive[k]
+												/ mode_lib->vba.NumberOfDSCSlices[k],
+										1),
+								slices,
+								mode_lib->vba.OutputFormat[k])
+								+ dscComputeDelay(
+										mode_lib->vba.OutputFormat[k]);
+			} else {
+				mode_lib->vba.DSCDelay[k] =
+						2
+								* (dscceComputeDelay(
+										mode_lib->vba.DSCInputBitPerComponent[k],
+										bpp,
+										dml_ceil(
+												(double) mode_lib->vba.HActive[k]
+														/ mode_lib->vba.NumberOfDSCSlices[k],
+												1),
+										slices / 2.0,
+										mode_lib->vba.OutputFormat[k])
+										+ dscComputeDelay(
+												mode_lib->vba.OutputFormat[k]));
+			}
+			mode_lib->vba.DSCDelay[k] = mode_lib->vba.DSCDelay[k]
+					* mode_lib->vba.PixelClock[k]
+					/ mode_lib->vba.PixelClockBackEnd[k];
+		} else {
+			mode_lib->vba.DSCDelay[k] = 0;
+		}
+	}
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k)
+		for (j = 0; j < mode_lib->vba.NumberOfActivePlanes; ++j) // NumberOfPlanes
+			if (j != k && mode_lib->vba.BlendingAndTiming[k] == j
+					&& mode_lib->vba.DSCEnabled[j])
+				mode_lib->vba.DSCDelay[k] = mode_lib->vba.DSCDelay[j];
+
+	// Prefetch
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		unsigned int PDEAndMetaPTEBytesFrameY;
+		unsigned int PixelPTEBytesPerRowY;
+		unsigned int MetaRowByteY;
+		unsigned int MetaRowByteC;
+		unsigned int PDEAndMetaPTEBytesFrameC;
+		unsigned int PixelPTEBytesPerRowC;
+
+		Calculate256BBlockSizes(
+				mode_lib->vba.SourcePixelFormat[k],
+				mode_lib->vba.SurfaceTiling[k],
+				dml_ceil(mode_lib->vba.BytePerPixelDETY[k], 1),
+				dml_ceil(mode_lib->vba.BytePerPixelDETC[k], 2),
+				&mode_lib->vba.BlockHeight256BytesY[k],
+				&mode_lib->vba.BlockHeight256BytesC[k],
+				&mode_lib->vba.BlockWidth256BytesY[k],
+				&mode_lib->vba.BlockWidth256BytesC[k]);
+		PDEAndMetaPTEBytesFrameY = CalculateVMAndRowBytes(
+				mode_lib,
+				mode_lib->vba.DCCEnable[k],
+				mode_lib->vba.BlockHeight256BytesY[k],
+				mode_lib->vba.BlockWidth256BytesY[k],
+				mode_lib->vba.SourcePixelFormat[k],
+				mode_lib->vba.SurfaceTiling[k],
+				dml_ceil(mode_lib->vba.BytePerPixelDETY[k], 1),
+				mode_lib->vba.SourceScan[k],
+				mode_lib->vba.ViewportWidth[k],
+				mode_lib->vba.ViewportHeight[k],
+				mode_lib->vba.SwathWidthY[k],
+				mode_lib->vba.GPUVMEnable,
+				mode_lib->vba.VMMPageSize,
+				mode_lib->vba.PTEBufferSizeInRequestsLuma,
+				mode_lib->vba.PDEProcessingBufIn64KBReqs,
+				mode_lib->vba.PitchY[k],
+				mode_lib->vba.DCCMetaPitchY[k],
+				&mode_lib->vba.MacroTileWidthY[k],
+				&MetaRowByteY,
+				&PixelPTEBytesPerRowY,
+				&mode_lib->vba.PTEBufferSizeNotExceeded[mode_lib->vba.VoltageLevel][0],
+				&mode_lib->vba.dpte_row_height[k],
+				&mode_lib->vba.meta_row_height[k]);
+		mode_lib->vba.PrefetchSourceLinesY[k] = CalculatePrefetchSourceLines(
+				mode_lib,
+				mode_lib->vba.VRatio[k],
+				mode_lib->vba.vtaps[k],
+				mode_lib->vba.Interlace[k],
+				mode_lib->vba.ProgressiveToInterlaceUnitInOPP,
+				mode_lib->vba.SwathHeightY[k],
+				mode_lib->vba.ViewportYStartY[k],
+				&mode_lib->vba.VInitPreFillY[k],
+				&mode_lib->vba.MaxNumSwathY[k]);
+
+		if ((mode_lib->vba.SourcePixelFormat[k] != dm_444_64
+				&& mode_lib->vba.SourcePixelFormat[k] != dm_444_32
+				&& mode_lib->vba.SourcePixelFormat[k] != dm_444_16
+				&& mode_lib->vba.SourcePixelFormat[k] != dm_444_8)) {
+			PDEAndMetaPTEBytesFrameC =
+					CalculateVMAndRowBytes(
+							mode_lib,
+							mode_lib->vba.DCCEnable[k],
+							mode_lib->vba.BlockHeight256BytesC[k],
+							mode_lib->vba.BlockWidth256BytesC[k],
+							mode_lib->vba.SourcePixelFormat[k],
+							mode_lib->vba.SurfaceTiling[k],
+							dml_ceil(
+									mode_lib->vba.BytePerPixelDETC[k],
+									2),
+							mode_lib->vba.SourceScan[k],
+							mode_lib->vba.ViewportWidth[k] / 2,
+							mode_lib->vba.ViewportHeight[k] / 2,
+							mode_lib->vba.SwathWidthY[k] / 2,
+							mode_lib->vba.GPUVMEnable,
+							mode_lib->vba.VMMPageSize,
+							mode_lib->vba.PTEBufferSizeInRequestsLuma,
+							mode_lib->vba.PDEProcessingBufIn64KBReqs,
+							mode_lib->vba.PitchC[k],
+							0,
+							&mode_lib->vba.MacroTileWidthC[k],
+							&MetaRowByteC,
+							&PixelPTEBytesPerRowC,
+							&mode_lib->vba.PTEBufferSizeNotExceeded[mode_lib->vba.VoltageLevel][0],
+							&mode_lib->vba.dpte_row_height_chroma[k],
+							&mode_lib->vba.meta_row_height_chroma[k]);
+			mode_lib->vba.PrefetchSourceLinesC[k] = CalculatePrefetchSourceLines(
+					mode_lib,
+					mode_lib->vba.VRatio[k] / 2,
+					mode_lib->vba.VTAPsChroma[k],
+					mode_lib->vba.Interlace[k],
+					mode_lib->vba.ProgressiveToInterlaceUnitInOPP,
+					mode_lib->vba.SwathHeightC[k],
+					mode_lib->vba.ViewportYStartC[k],
+					&mode_lib->vba.VInitPreFillC[k],
+					&mode_lib->vba.MaxNumSwathC[k]);
+		} else {
+			PixelPTEBytesPerRowC = 0;
+			PDEAndMetaPTEBytesFrameC = 0;
+			MetaRowByteC = 0;
+			mode_lib->vba.MaxNumSwathC[k] = 0;
+			mode_lib->vba.PrefetchSourceLinesC[k] = 0;
+		}
+
+		mode_lib->vba.PixelPTEBytesPerRow[k] = PixelPTEBytesPerRowY + PixelPTEBytesPerRowC;
+		mode_lib->vba.PDEAndMetaPTEBytesFrame[k] = PDEAndMetaPTEBytesFrameY
+				+ PDEAndMetaPTEBytesFrameC;
+		mode_lib->vba.MetaRowByte[k] = MetaRowByteY + MetaRowByteC;
+
+		CalculateActiveRowBandwidth(
+				mode_lib->vba.GPUVMEnable,
+				mode_lib->vba.SourcePixelFormat[k],
+				mode_lib->vba.VRatio[k],
+				mode_lib->vba.DCCEnable[k],
+				mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k],
+				MetaRowByteY,
+				MetaRowByteC,
+				mode_lib->vba.meta_row_height[k],
+				mode_lib->vba.meta_row_height_chroma[k],
+				PixelPTEBytesPerRowY,
+				PixelPTEBytesPerRowC,
+				mode_lib->vba.dpte_row_height[k],
+				mode_lib->vba.dpte_row_height_chroma[k],
+				&mode_lib->vba.meta_row_bw[k],
+				&mode_lib->vba.dpte_row_bw[k],
+				&mode_lib->vba.qual_row_bw[k]);
+	}
+
+	mode_lib->vba.TCalc = 24.0 / mode_lib->vba.DCFCLKDeepSleep;
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.BlendingAndTiming[k] == k) {
+			if (mode_lib->vba.WritebackEnable[k] == true) {
+				mode_lib->vba.WritebackDelay[mode_lib->vba.VoltageLevel][k] =
+						mode_lib->vba.WritebackLatency
+								+ CalculateWriteBackDelay(
+										mode_lib->vba.WritebackPixelFormat[k],
+										mode_lib->vba.WritebackHRatio[k],
+										mode_lib->vba.WritebackVRatio[k],
+										mode_lib->vba.WritebackLumaHTaps[k],
+										mode_lib->vba.WritebackLumaVTaps[k],
+										mode_lib->vba.WritebackChromaHTaps[k],
+										mode_lib->vba.WritebackChromaVTaps[k],
+										mode_lib->vba.WritebackDestinationWidth[k])
+										/ mode_lib->vba.DISPCLK;
+			} else
+				mode_lib->vba.WritebackDelay[mode_lib->vba.VoltageLevel][k] = 0;
+			for (j = 0; j < mode_lib->vba.NumberOfActivePlanes; ++j) {
+				if (mode_lib->vba.BlendingAndTiming[j] == k
+						&& mode_lib->vba.WritebackEnable[j] == true) {
+					mode_lib->vba.WritebackDelay[mode_lib->vba.VoltageLevel][k] =
+							dml_max(
+									mode_lib->vba.WritebackDelay[mode_lib->vba.VoltageLevel][k],
+									mode_lib->vba.WritebackLatency
+											+ CalculateWriteBackDelay(
+													mode_lib->vba.WritebackPixelFormat[j],
+													mode_lib->vba.WritebackHRatio[j],
+													mode_lib->vba.WritebackVRatio[j],
+													mode_lib->vba.WritebackLumaHTaps[j],
+													mode_lib->vba.WritebackLumaVTaps[j],
+													mode_lib->vba.WritebackChromaHTaps[j],
+													mode_lib->vba.WritebackChromaVTaps[j],
+													mode_lib->vba.WritebackDestinationWidth[j])
+													/ mode_lib->vba.DISPCLK);
+				}
+			}
+		}
+	}
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k)
+		for (j = 0; j < mode_lib->vba.NumberOfActivePlanes; ++j)
+			if (mode_lib->vba.BlendingAndTiming[k] == j)
+				mode_lib->vba.WritebackDelay[mode_lib->vba.VoltageLevel][k] =
+						mode_lib->vba.WritebackDelay[mode_lib->vba.VoltageLevel][j];
+
+	mode_lib->vba.VStartupLines = 13;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		mode_lib->vba.MaxVStartupLines[k] =
+				mode_lib->vba.VTotal[k] - mode_lib->vba.VActive[k]
+						- dml_max(
+								1.0,
+								dml_ceil(
+										mode_lib->vba.WritebackDelay[mode_lib->vba.VoltageLevel][k]
+												/ (mode_lib->vba.HTotal[k]
+														/ mode_lib->vba.PixelClock[k]),
+										1));
+	}
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k)
+		mode_lib->vba.MaximumMaxVStartupLines = dml_max(
+				mode_lib->vba.MaximumMaxVStartupLines,
+				mode_lib->vba.MaxVStartupLines[k]);
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		mode_lib->vba.cursor_bw[k] = 0.0;
+		for (j = 0; j < mode_lib->vba.NumberOfCursors[k]; ++j)
+			mode_lib->vba.cursor_bw[k] += mode_lib->vba.CursorWidth[k][j]
+					* mode_lib->vba.CursorBPP[k][j] / 8.0
+					/ (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k])
+					* mode_lib->vba.VRatio[k];
+	}
+
+	do {
+		double MaxTotalRDBandwidth = 0;
+		bool DestinationLineTimesForPrefetchLessThan2 = false;
+		bool VRatioPrefetchMoreThan4 = false;
+		bool prefetch_vm_bw_valid = true;
+		bool prefetch_row_bw_valid = true;
+		double TWait = CalculateTWait(
+				mode_lib->vba.PrefetchMode[mode_lib->vba.VoltageLevel][mode_lib->vba.maxMpcComb],
+				mode_lib->vba.DRAMClockChangeLatency,
+				mode_lib->vba.UrgentLatencyPixelDataOnly,
+				mode_lib->vba.SREnterPlusExitTime);
+
+		for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+			if (mode_lib->vba.XFCEnabled[k] == true) {
+				mode_lib->vba.XFCRemoteSurfaceFlipDelay =
+						CalculateRemoteSurfaceFlipDelay(
+								mode_lib,
+								mode_lib->vba.VRatio[k],
+								mode_lib->vba.SwathWidthY[k],
+								dml_ceil(
+										mode_lib->vba.BytePerPixelDETY[k],
+										1),
+								mode_lib->vba.HTotal[k]
+										/ mode_lib->vba.PixelClock[k],
+								mode_lib->vba.XFCTSlvVupdateOffset,
+								mode_lib->vba.XFCTSlvVupdateWidth,
+								mode_lib->vba.XFCTSlvVreadyOffset,
+								mode_lib->vba.XFCXBUFLatencyTolerance,
+								mode_lib->vba.XFCFillBWOverhead,
+								mode_lib->vba.XFCSlvChunkSize,
+								mode_lib->vba.XFCBusTransportTime,
+								mode_lib->vba.TCalc,
+								TWait,
+								&mode_lib->vba.SrcActiveDrainRate,
+								&mode_lib->vba.TInitXFill,
+								&mode_lib->vba.TslvChk);
+			} else {
+				mode_lib->vba.XFCRemoteSurfaceFlipDelay = 0;
+			}
+			mode_lib->vba.ErrorResult[k] =
+					CalculatePrefetchSchedule(
+							mode_lib,
+							mode_lib->vba.DPPCLK[k],
+							mode_lib->vba.DISPCLK,
+							mode_lib->vba.PixelClock[k],
+							mode_lib->vba.DCFCLKDeepSleep,
+							mode_lib->vba.DSCDelay[k],
+							mode_lib->vba.DPPPerPlane[k],
+							mode_lib->vba.ScalerEnabled[k],
+							mode_lib->vba.NumberOfCursors[k],
+							mode_lib->vba.DPPCLKDelaySubtotal,
+							mode_lib->vba.DPPCLKDelaySCL,
+							mode_lib->vba.DPPCLKDelaySCLLBOnly,
+							mode_lib->vba.DPPCLKDelayCNVCFormater,
+							mode_lib->vba.DPPCLKDelayCNVCCursor,
+							mode_lib->vba.DISPCLKDelaySubtotal,
+							(unsigned int) (mode_lib->vba.SwathWidthY[k]
+									/ mode_lib->vba.HRatio[k]),
+							mode_lib->vba.OutputFormat[k],
+							mode_lib->vba.VTotal[k]
+									- mode_lib->vba.VActive[k],
+							mode_lib->vba.HTotal[k],
+							mode_lib->vba.MaxInterDCNTileRepeaters,
+							dml_min(
+									mode_lib->vba.VStartupLines,
+									mode_lib->vba.MaxVStartupLines[k]),
+							mode_lib->vba.GPUVMMaxPageTableLevels,
+							mode_lib->vba.GPUVMEnable,
+							mode_lib->vba.DynamicMetadataEnable[k],
+							mode_lib->vba.DynamicMetadataLinesBeforeActiveRequired[k],
+							mode_lib->vba.DynamicMetadataTransmittedBytes[k],
+							mode_lib->vba.DCCEnable[k],
+							mode_lib->vba.UrgentLatencyPixelDataOnly,
+							mode_lib->vba.UrgentExtraLatency,
+							mode_lib->vba.TCalc,
+							mode_lib->vba.PDEAndMetaPTEBytesFrame[k],
+							mode_lib->vba.MetaRowByte[k],
+							mode_lib->vba.PixelPTEBytesPerRow[k],
+							mode_lib->vba.PrefetchSourceLinesY[k],
+							mode_lib->vba.SwathWidthY[k],
+							mode_lib->vba.BytePerPixelDETY[k],
+							mode_lib->vba.VInitPreFillY[k],
+							mode_lib->vba.MaxNumSwathY[k],
+							mode_lib->vba.PrefetchSourceLinesC[k],
+							mode_lib->vba.BytePerPixelDETC[k],
+							mode_lib->vba.VInitPreFillC[k],
+							mode_lib->vba.MaxNumSwathC[k],
+							mode_lib->vba.SwathHeightY[k],
+							mode_lib->vba.SwathHeightC[k],
+							TWait,
+							mode_lib->vba.XFCEnabled[k],
+							mode_lib->vba.XFCRemoteSurfaceFlipDelay,
+							mode_lib->vba.Interlace[k],
+							mode_lib->vba.ProgressiveToInterlaceUnitInOPP,
+							&mode_lib->vba.DSTXAfterScaler[k],
+							&mode_lib->vba.DSTYAfterScaler[k],
+							&mode_lib->vba.DestinationLinesForPrefetch[k],
+							&mode_lib->vba.PrefetchBandwidth[k],
+							&mode_lib->vba.DestinationLinesToRequestVMInVBlank[k],
+							&mode_lib->vba.DestinationLinesToRequestRowInVBlank[k],
+							&mode_lib->vba.VRatioPrefetchY[k],
+							&mode_lib->vba.VRatioPrefetchC[k],
+							&mode_lib->vba.RequiredPrefetchPixDataBWLuma[k],
+							&mode_lib->vba.VStartupRequiredWhenNotEnoughTimeForDynamicMetadata,
+							&mode_lib->vba.Tno_bw[k],
+							&mode_lib->vba.VUpdateOffsetPix[k],
+							&mode_lib->vba.VUpdateWidthPix[k],
+							&mode_lib->vba.VReadyOffsetPix[k]);
+			if (mode_lib->vba.BlendingAndTiming[k] == k) {
+				mode_lib->vba.VStartup[k] = dml_min(
+						mode_lib->vba.VStartupLines,
+						mode_lib->vba.MaxVStartupLines[k]);
+				if (mode_lib->vba.VStartupRequiredWhenNotEnoughTimeForDynamicMetadata
+						!= 0) {
+					mode_lib->vba.VStartup[k] =
+							mode_lib->vba.VStartupRequiredWhenNotEnoughTimeForDynamicMetadata;
+				}
+			} else {
+				mode_lib->vba.VStartup[k] =
+						dml_min(
+								mode_lib->vba.VStartupLines,
+								mode_lib->vba.MaxVStartupLines[mode_lib->vba.BlendingAndTiming[k]]);
+			}
+		}
+
+		for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+
+			if (mode_lib->vba.PDEAndMetaPTEBytesFrame[k] == 0)
+				mode_lib->vba.prefetch_vm_bw[k] = 0;
+			else if (mode_lib->vba.DestinationLinesToRequestVMInVBlank[k] > 0) {
+				mode_lib->vba.prefetch_vm_bw[k] =
+						(double) mode_lib->vba.PDEAndMetaPTEBytesFrame[k]
+								/ (mode_lib->vba.DestinationLinesToRequestVMInVBlank[k]
+										* mode_lib->vba.HTotal[k]
+										/ mode_lib->vba.PixelClock[k]);
+			} else {
+				mode_lib->vba.prefetch_vm_bw[k] = 0;
+				prefetch_vm_bw_valid = false;
+			}
+			if (mode_lib->vba.MetaRowByte[k] + mode_lib->vba.PixelPTEBytesPerRow[k]
+					== 0)
+				mode_lib->vba.prefetch_row_bw[k] = 0;
+			else if (mode_lib->vba.DestinationLinesToRequestRowInVBlank[k] > 0) {
+				mode_lib->vba.prefetch_row_bw[k] =
+						(double) (mode_lib->vba.MetaRowByte[k]
+								+ mode_lib->vba.PixelPTEBytesPerRow[k])
+								/ (mode_lib->vba.DestinationLinesToRequestRowInVBlank[k]
+										* mode_lib->vba.HTotal[k]
+										/ mode_lib->vba.PixelClock[k]);
+			} else {
+				mode_lib->vba.prefetch_row_bw[k] = 0;
+				prefetch_row_bw_valid = false;
+			}
+
+			MaxTotalRDBandwidth =
+					MaxTotalRDBandwidth + mode_lib->vba.cursor_bw[k]
+							+ dml_max(
+									mode_lib->vba.prefetch_vm_bw[k],
+									dml_max(
+											mode_lib->vba.prefetch_row_bw[k],
+											dml_max(
+													mode_lib->vba.ReadBandwidthPlaneLuma[k]
+															+ mode_lib->vba.ReadBandwidthPlaneChroma[k],
+													mode_lib->vba.RequiredPrefetchPixDataBWLuma[k])
+													+ mode_lib->vba.meta_row_bw[k]
+													+ mode_lib->vba.dpte_row_bw[k]));
+
+			if (mode_lib->vba.DestinationLinesForPrefetch[k] < 2)
+				DestinationLineTimesForPrefetchLessThan2 = true;
+			if (mode_lib->vba.VRatioPrefetchY[k] > 4
+					|| mode_lib->vba.VRatioPrefetchC[k] > 4)
+				VRatioPrefetchMoreThan4 = true;
+		}
+
+		if (MaxTotalRDBandwidth <= mode_lib->vba.ReturnBW && prefetch_vm_bw_valid
+				&& prefetch_row_bw_valid && !VRatioPrefetchMoreThan4
+				&& !DestinationLineTimesForPrefetchLessThan2)
+			mode_lib->vba.PrefetchModeSupported = true;
+		else {
+			mode_lib->vba.PrefetchModeSupported = false;
+			dml_print(
+					"DML: CalculatePrefetchSchedule ***failed***. Bandwidth violation. Results are NOT valid\n");
+		}
+
+		if (mode_lib->vba.PrefetchModeSupported == true) {
+			double final_flip_bw[DC__NUM_DPP__MAX];
+			unsigned int ImmediateFlipBytes[DC__NUM_DPP__MAX];
+			double total_dcn_read_bw_with_flip = 0;
+
+			mode_lib->vba.BandwidthAvailableForImmediateFlip = mode_lib->vba.ReturnBW;
+			for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+				mode_lib->vba.BandwidthAvailableForImmediateFlip =
+						mode_lib->vba.BandwidthAvailableForImmediateFlip
+								- mode_lib->vba.cursor_bw[k]
+								- dml_max(
+										mode_lib->vba.ReadBandwidthPlaneLuma[k]
+												+ mode_lib->vba.ReadBandwidthPlaneChroma[k]
+												+ mode_lib->vba.qual_row_bw[k],
+										mode_lib->vba.PrefetchBandwidth[k]);
+			}
+
+			for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+				ImmediateFlipBytes[k] = 0;
+				if ((mode_lib->vba.SourcePixelFormat[k] != dm_420_8
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_420_10)) {
+					ImmediateFlipBytes[k] =
+							mode_lib->vba.PDEAndMetaPTEBytesFrame[k]
+									+ mode_lib->vba.MetaRowByte[k]
+									+ mode_lib->vba.PixelPTEBytesPerRow[k];
+				}
+			}
+			mode_lib->vba.TotImmediateFlipBytes = 0;
+			for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+				if ((mode_lib->vba.SourcePixelFormat[k] != dm_420_8
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_420_10)) {
+					mode_lib->vba.TotImmediateFlipBytes =
+							mode_lib->vba.TotImmediateFlipBytes
+									+ ImmediateFlipBytes[k];
+				}
+			}
+			for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+				CalculateFlipSchedule(
+						mode_lib,
+						mode_lib->vba.UrgentExtraLatency,
+						mode_lib->vba.UrgentLatencyPixelDataOnly,
+						mode_lib->vba.GPUVMMaxPageTableLevels,
+						mode_lib->vba.GPUVMEnable,
+						mode_lib->vba.BandwidthAvailableForImmediateFlip,
+						mode_lib->vba.TotImmediateFlipBytes,
+						mode_lib->vba.SourcePixelFormat[k],
+						ImmediateFlipBytes[k],
+						mode_lib->vba.HTotal[k]
+								/ mode_lib->vba.PixelClock[k],
+						mode_lib->vba.VRatio[k],
+						mode_lib->vba.Tno_bw[k],
+						mode_lib->vba.PDEAndMetaPTEBytesFrame[k],
+						mode_lib->vba.MetaRowByte[k],
+						mode_lib->vba.PixelPTEBytesPerRow[k],
+						mode_lib->vba.DCCEnable[k],
+						mode_lib->vba.dpte_row_height[k],
+						mode_lib->vba.meta_row_height[k],
+						mode_lib->vba.qual_row_bw[k],
+						&mode_lib->vba.DestinationLinesToRequestVMInImmediateFlip[k],
+						&mode_lib->vba.DestinationLinesToRequestRowInImmediateFlip[k],
+						&final_flip_bw[k],
+						&mode_lib->vba.ImmediateFlipSupportedForPipe[k]);
+			}
+			for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+				total_dcn_read_bw_with_flip =
+						total_dcn_read_bw_with_flip
+								+ mode_lib->vba.cursor_bw[k]
+								+ dml_max(
+										mode_lib->vba.prefetch_vm_bw[k],
+										dml_max(
+												mode_lib->vba.prefetch_row_bw[k],
+												final_flip_bw[k]
+														+ dml_max(
+																mode_lib->vba.ReadBandwidthPlaneLuma[k]
+																		+ mode_lib->vba.ReadBandwidthPlaneChroma[k],
+																mode_lib->vba.RequiredPrefetchPixDataBWLuma[k])));
+			}
+			mode_lib->vba.ImmediateFlipSupported = true;
+			if (total_dcn_read_bw_with_flip > mode_lib->vba.ReturnBW) {
+				mode_lib->vba.ImmediateFlipSupported = false;
+			}
+			for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+				if (mode_lib->vba.ImmediateFlipSupportedForPipe[k] == false) {
+					mode_lib->vba.ImmediateFlipSupported = false;
+				}
+			}
+		} else {
+			mode_lib->vba.ImmediateFlipSupported = false;
+		}
+
+		for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+			if (mode_lib->vba.ErrorResult[k]) {
+				mode_lib->vba.PrefetchModeSupported = false;
+				dml_print(
+						"DML: CalculatePrefetchSchedule ***failed***. Prefetch schedule violation. Results are NOT valid\n");
+			}
+		}
+
+		mode_lib->vba.VStartupLines = mode_lib->vba.VStartupLines + 1;
+	} while (!((mode_lib->vba.PrefetchModeSupported
+			&& (!mode_lib->vba.ImmediateFlipSupport
+					|| mode_lib->vba.ImmediateFlipSupported))
+			|| mode_lib->vba.MaximumMaxVStartupLines < mode_lib->vba.VStartupLines));
+
+	//Display Pipeline Delivery Time in Prefetch
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.VRatioPrefetchY[k] <= 1) {
+			mode_lib->vba.DisplayPipeLineDeliveryTimeLumaPrefetch[k] =
+					mode_lib->vba.SwathWidthY[k] * mode_lib->vba.DPPPerPlane[k]
+							/ mode_lib->vba.HRatio[k]
+							/ mode_lib->vba.PixelClock[k];
+		} else {
+			mode_lib->vba.DisplayPipeLineDeliveryTimeLumaPrefetch[k] =
+					mode_lib->vba.SwathWidthY[k]
+							/ mode_lib->vba.PSCL_THROUGHPUT_LUMA[k]
+							/ mode_lib->vba.DPPCLK[k];
+		}
+		if (mode_lib->vba.BytePerPixelDETC[k] == 0) {
+			mode_lib->vba.DisplayPipeLineDeliveryTimeChromaPrefetch[k] = 0;
+		} else {
+			if (mode_lib->vba.VRatioPrefetchC[k] <= 1) {
+				mode_lib->vba.DisplayPipeLineDeliveryTimeChromaPrefetch[k] =
+						mode_lib->vba.SwathWidthY[k]
+								* mode_lib->vba.DPPPerPlane[k]
+								/ mode_lib->vba.HRatio[k]
+								/ mode_lib->vba.PixelClock[k];
+			} else {
+				mode_lib->vba.DisplayPipeLineDeliveryTimeChromaPrefetch[k] =
+						mode_lib->vba.SwathWidthY[k]
+								/ mode_lib->vba.PSCL_THROUGHPUT_LUMA[k]
+								/ mode_lib->vba.DPPCLK[k];
+			}
+		}
+	}
+
+	// Min TTUVBlank
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.PrefetchMode[mode_lib->vba.VoltageLevel][mode_lib->vba.maxMpcComb] == 0) {
+			mode_lib->vba.AllowDRAMClockChangeDuringVBlank[k] = true;
+			mode_lib->vba.AllowDRAMSelfRefreshDuringVBlank[k] = true;
+			mode_lib->vba.MinTTUVBlank[k] = dml_max(
+					mode_lib->vba.DRAMClockChangeWatermark,
+					dml_max(
+							mode_lib->vba.StutterEnterPlusExitWatermark,
+							mode_lib->vba.UrgentWatermark));
+		} else if (mode_lib->vba.PrefetchMode[mode_lib->vba.VoltageLevel][mode_lib->vba.maxMpcComb] == 1) {
+			mode_lib->vba.AllowDRAMClockChangeDuringVBlank[k] = false;
+			mode_lib->vba.AllowDRAMSelfRefreshDuringVBlank[k] = true;
+			mode_lib->vba.MinTTUVBlank[k] = dml_max(
+					mode_lib->vba.StutterEnterPlusExitWatermark,
+					mode_lib->vba.UrgentWatermark);
+		} else {
+			mode_lib->vba.AllowDRAMClockChangeDuringVBlank[k] = false;
+			mode_lib->vba.AllowDRAMSelfRefreshDuringVBlank[k] = false;
+			mode_lib->vba.MinTTUVBlank[k] = mode_lib->vba.UrgentWatermark;
+		}
+		if (!mode_lib->vba.DynamicMetadataEnable[k])
+			mode_lib->vba.MinTTUVBlank[k] = mode_lib->vba.TCalc
+					+ mode_lib->vba.MinTTUVBlank[k];
+	}
+
+	// DCC Configuration
+	mode_lib->vba.ActiveDPPs = 0;
+	// NB P-State/DRAM Clock Change Support
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		mode_lib->vba.ActiveDPPs = mode_lib->vba.ActiveDPPs + mode_lib->vba.DPPPerPlane[k];
+	}
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		double EffectiveLBLatencyHidingY;
+		double EffectiveLBLatencyHidingC;
+		double DPPOutputBufferLinesY;
+		double DPPOutputBufferLinesC;
+		double DPPOPPBufferingY;
+		double MaxDETBufferingTimeY;
+		double ActiveDRAMClockChangeLatencyMarginY;
+
+		mode_lib->vba.LBLatencyHidingSourceLinesY =
+				dml_min(
+						mode_lib->vba.MaxLineBufferLines,
+						(unsigned int) dml_floor(
+								(double) mode_lib->vba.LineBufferSize
+										/ mode_lib->vba.LBBitPerPixel[k]
+										/ (mode_lib->vba.SwathWidthY[k]
+												/ dml_max(
+														mode_lib->vba.HRatio[k],
+														1.0)),
+								1)) - (mode_lib->vba.vtaps[k] - 1);
+
+		mode_lib->vba.LBLatencyHidingSourceLinesC =
+				dml_min(
+						mode_lib->vba.MaxLineBufferLines,
+						(unsigned int) dml_floor(
+								(double) mode_lib->vba.LineBufferSize
+										/ mode_lib->vba.LBBitPerPixel[k]
+										/ (mode_lib->vba.SwathWidthY[k]
+												/ 2.0
+												/ dml_max(
+														mode_lib->vba.HRatio[k]
+																/ 2,
+														1.0)),
+								1))
+						- (mode_lib->vba.VTAPsChroma[k] - 1);
+
+		EffectiveLBLatencyHidingY = mode_lib->vba.LBLatencyHidingSourceLinesY
+				/ mode_lib->vba.VRatio[k]
+				* (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k]);
+
+		EffectiveLBLatencyHidingC = mode_lib->vba.LBLatencyHidingSourceLinesC
+				/ (mode_lib->vba.VRatio[k] / 2)
+				* (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k]);
+
+		if (mode_lib->vba.SwathWidthY[k] > 2 * mode_lib->vba.DPPOutputBufferPixels) {
+			DPPOutputBufferLinesY = mode_lib->vba.DPPOutputBufferPixels
+					/ mode_lib->vba.SwathWidthY[k];
+		} else if (mode_lib->vba.SwathWidthY[k] > mode_lib->vba.DPPOutputBufferPixels) {
+			DPPOutputBufferLinesY = 0.5;
+		} else {
+			DPPOutputBufferLinesY = 1;
+		}
+
+		if (mode_lib->vba.SwathWidthY[k] / 2 > 2 * mode_lib->vba.DPPOutputBufferPixels) {
+			DPPOutputBufferLinesC = mode_lib->vba.DPPOutputBufferPixels
+					/ (mode_lib->vba.SwathWidthY[k] / 2);
+		} else if (mode_lib->vba.SwathWidthY[k] / 2 > mode_lib->vba.DPPOutputBufferPixels) {
+			DPPOutputBufferLinesC = 0.5;
+		} else {
+			DPPOutputBufferLinesC = 1;
+		}
+
+		DPPOPPBufferingY = (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k])
+				* (DPPOutputBufferLinesY + mode_lib->vba.OPPOutputBufferLines);
+		MaxDETBufferingTimeY = mode_lib->vba.FullDETBufferingTimeY[k]
+				+ (mode_lib->vba.LinesInDETY[k]
+						- mode_lib->vba.LinesInDETYRoundedDownToSwath[k])
+						/ mode_lib->vba.SwathHeightY[k]
+						* (mode_lib->vba.HTotal[k]
+								/ mode_lib->vba.PixelClock[k]);
+
+		ActiveDRAMClockChangeLatencyMarginY = DPPOPPBufferingY + EffectiveLBLatencyHidingY
+				+ MaxDETBufferingTimeY - mode_lib->vba.DRAMClockChangeWatermark;
+
+		if (mode_lib->vba.ActiveDPPs > 1) {
+			ActiveDRAMClockChangeLatencyMarginY =
+					ActiveDRAMClockChangeLatencyMarginY
+							- (1 - 1 / (mode_lib->vba.ActiveDPPs - 1))
+									* mode_lib->vba.SwathHeightY[k]
+									* (mode_lib->vba.HTotal[k]
+											/ mode_lib->vba.PixelClock[k]);
+		}
+
+		if (mode_lib->vba.BytePerPixelDETC[k] > 0) {
+			double DPPOPPBufferingC = (mode_lib->vba.HTotal[k]
+					/ mode_lib->vba.PixelClock[k])
+					* (DPPOutputBufferLinesC
+							+ mode_lib->vba.OPPOutputBufferLines);
+			double MaxDETBufferingTimeC =
+					mode_lib->vba.FullDETBufferingTimeC[k]
+							+ (mode_lib->vba.LinesInDETC[k]
+									- mode_lib->vba.LinesInDETCRoundedDownToSwath[k])
+									/ mode_lib->vba.SwathHeightC[k]
+									* (mode_lib->vba.HTotal[k]
+											/ mode_lib->vba.PixelClock[k]);
+			double ActiveDRAMClockChangeLatencyMarginC = DPPOPPBufferingC
+					+ EffectiveLBLatencyHidingC + MaxDETBufferingTimeC
+					- mode_lib->vba.DRAMClockChangeWatermark;
+
+			if (mode_lib->vba.ActiveDPPs > 1) {
+				ActiveDRAMClockChangeLatencyMarginC =
+						ActiveDRAMClockChangeLatencyMarginC
+								- (1
+										- 1
+												/ (mode_lib->vba.ActiveDPPs
+														- 1))
+										* mode_lib->vba.SwathHeightC[k]
+										* (mode_lib->vba.HTotal[k]
+												/ mode_lib->vba.PixelClock[k]);
+			}
+			mode_lib->vba.ActiveDRAMClockChangeLatencyMargin[k] = dml_min(
+					ActiveDRAMClockChangeLatencyMarginY,
+					ActiveDRAMClockChangeLatencyMarginC);
+		} else {
+			mode_lib->vba.ActiveDRAMClockChangeLatencyMargin[k] =
+					ActiveDRAMClockChangeLatencyMarginY;
+		}
+
+		if (mode_lib->vba.WritebackEnable[k]) {
+			double WritebackDRAMClockChangeLatencyMargin;
+
+			if (mode_lib->vba.WritebackPixelFormat[k] == dm_444_32) {
+				WritebackDRAMClockChangeLatencyMargin =
+						(double) (mode_lib->vba.WritebackInterfaceLumaBufferSize
+								+ mode_lib->vba.WritebackInterfaceChromaBufferSize)
+								/ (mode_lib->vba.WritebackDestinationWidth[k]
+										* mode_lib->vba.WritebackDestinationHeight[k]
+										/ (mode_lib->vba.WritebackSourceHeight[k]
+												* mode_lib->vba.HTotal[k]
+												/ mode_lib->vba.PixelClock[k])
+										* 4)
+								- mode_lib->vba.WritebackDRAMClockChangeWatermark;
+			} else if (mode_lib->vba.WritebackPixelFormat[k] == dm_420_10) {
+				WritebackDRAMClockChangeLatencyMargin =
+						dml_min(
+								(double) mode_lib->vba.WritebackInterfaceLumaBufferSize
+										* 8.0 / 10,
+								2.0
+										* mode_lib->vba.WritebackInterfaceChromaBufferSize
+										* 8 / 10)
+								/ (mode_lib->vba.WritebackDestinationWidth[k]
+										* mode_lib->vba.WritebackDestinationHeight[k]
+										/ (mode_lib->vba.WritebackSourceHeight[k]
+												* mode_lib->vba.HTotal[k]
+												/ mode_lib->vba.PixelClock[k]))
+								- mode_lib->vba.WritebackDRAMClockChangeWatermark;
+			} else {
+				WritebackDRAMClockChangeLatencyMargin =
+						dml_min(
+								(double) mode_lib->vba.WritebackInterfaceLumaBufferSize,
+								2.0
+										* mode_lib->vba.WritebackInterfaceChromaBufferSize)
+								/ (mode_lib->vba.WritebackDestinationWidth[k]
+										* mode_lib->vba.WritebackDestinationHeight[k]
+										/ (mode_lib->vba.WritebackSourceHeight[k]
+												* mode_lib->vba.HTotal[k]
+												/ mode_lib->vba.PixelClock[k]))
+								- mode_lib->vba.WritebackDRAMClockChangeWatermark;
+			}
+			mode_lib->vba.ActiveDRAMClockChangeLatencyMargin[k] = dml_min(
+					mode_lib->vba.ActiveDRAMClockChangeLatencyMargin[k],
+					WritebackDRAMClockChangeLatencyMargin);
+		}
+	}
+
+	mode_lib->vba.MinActiveDRAMClockChangeMargin = 999999;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.ActiveDRAMClockChangeLatencyMargin[k]
+				< mode_lib->vba.MinActiveDRAMClockChangeMargin) {
+			mode_lib->vba.MinActiveDRAMClockChangeMargin =
+					mode_lib->vba.ActiveDRAMClockChangeLatencyMargin[k];
+		}
+	}
+
+	mode_lib->vba.MinActiveDRAMClockChangeLatencySupported =
+			mode_lib->vba.MinActiveDRAMClockChangeMargin
+					+ mode_lib->vba.DRAMClockChangeLatency;
+
+	if (mode_lib->vba.DRAMClockChangeSupportsVActive &&
+			mode_lib->vba.MinActiveDRAMClockChangeMargin > 60) {
+		mode_lib->vba.DRAMClockChangeWatermark += 25;
+		mode_lib->vba.DRAMClockChangeSupport[0][0] = dm_dram_clock_change_vactive;
+	} else {
+		if (mode_lib->vba.SynchronizedVBlank || mode_lib->vba.NumberOfActivePlanes == 1) {
+			mode_lib->vba.DRAMClockChangeSupport[0][0] = dm_dram_clock_change_vblank;
+			for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+				if (!mode_lib->vba.AllowDRAMClockChangeDuringVBlank[k]) {
+					mode_lib->vba.DRAMClockChangeSupport[0][0] =
+							dm_dram_clock_change_unsupported;
+				}
+			}
+		} else {
+			mode_lib->vba.DRAMClockChangeSupport[0][0] = dm_dram_clock_change_unsupported;
+		}
+	}
+	for (k = 0; k <= mode_lib->vba.soc.num_states; k++)
+		for (j = 0; j < 2; j++)
+			mode_lib->vba.DRAMClockChangeSupport[k][j] = mode_lib->vba.DRAMClockChangeSupport[0][0];
+
+	//XFC Parameters:
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.XFCEnabled[k] == true) {
+			double TWait;
+
+			mode_lib->vba.XFCSlaveVUpdateOffset[k] = mode_lib->vba.XFCTSlvVupdateOffset;
+			mode_lib->vba.XFCSlaveVupdateWidth[k] = mode_lib->vba.XFCTSlvVupdateWidth;
+			mode_lib->vba.XFCSlaveVReadyOffset[k] = mode_lib->vba.XFCTSlvVreadyOffset;
+			TWait = CalculateTWait(
+					mode_lib->vba.PrefetchMode[mode_lib->vba.VoltageLevel][mode_lib->vba.maxMpcComb],
+					mode_lib->vba.DRAMClockChangeLatency,
+					mode_lib->vba.UrgentLatencyPixelDataOnly,
+					mode_lib->vba.SREnterPlusExitTime);
+			mode_lib->vba.XFCRemoteSurfaceFlipDelay = CalculateRemoteSurfaceFlipDelay(
+					mode_lib,
+					mode_lib->vba.VRatio[k],
+					mode_lib->vba.SwathWidthY[k],
+					dml_ceil(mode_lib->vba.BytePerPixelDETY[k], 1),
+					mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k],
+					mode_lib->vba.XFCTSlvVupdateOffset,
+					mode_lib->vba.XFCTSlvVupdateWidth,
+					mode_lib->vba.XFCTSlvVreadyOffset,
+					mode_lib->vba.XFCXBUFLatencyTolerance,
+					mode_lib->vba.XFCFillBWOverhead,
+					mode_lib->vba.XFCSlvChunkSize,
+					mode_lib->vba.XFCBusTransportTime,
+					mode_lib->vba.TCalc,
+					TWait,
+					&mode_lib->vba.SrcActiveDrainRate,
+					&mode_lib->vba.TInitXFill,
+					&mode_lib->vba.TslvChk);
+			mode_lib->vba.XFCRemoteSurfaceFlipLatency[k] =
+					dml_floor(
+							mode_lib->vba.XFCRemoteSurfaceFlipDelay
+									/ (mode_lib->vba.HTotal[k]
+											/ mode_lib->vba.PixelClock[k]),
+							1);
+			mode_lib->vba.XFCTransferDelay[k] =
+					dml_ceil(
+							mode_lib->vba.XFCBusTransportTime
+									/ (mode_lib->vba.HTotal[k]
+											/ mode_lib->vba.PixelClock[k]),
+							1);
+			mode_lib->vba.XFCPrechargeDelay[k] =
+					dml_ceil(
+							(mode_lib->vba.XFCBusTransportTime
+									+ mode_lib->vba.TInitXFill
+									+ mode_lib->vba.TslvChk)
+									/ (mode_lib->vba.HTotal[k]
+											/ mode_lib->vba.PixelClock[k]),
+							1);
+			mode_lib->vba.InitFillLevel = mode_lib->vba.XFCXBUFLatencyTolerance
+					* mode_lib->vba.SrcActiveDrainRate;
+			mode_lib->vba.FinalFillMargin =
+					(mode_lib->vba.DestinationLinesToRequestVMInVBlank[k]
+							+ mode_lib->vba.DestinationLinesToRequestRowInVBlank[k])
+							* mode_lib->vba.HTotal[k]
+							/ mode_lib->vba.PixelClock[k]
+							* mode_lib->vba.SrcActiveDrainRate
+							+ mode_lib->vba.XFCFillConstant;
+			mode_lib->vba.FinalFillLevel = mode_lib->vba.XFCRemoteSurfaceFlipDelay
+					* mode_lib->vba.SrcActiveDrainRate
+					+ mode_lib->vba.FinalFillMargin;
+			mode_lib->vba.RemainingFillLevel = dml_max(
+					0.0,
+					mode_lib->vba.FinalFillLevel - mode_lib->vba.InitFillLevel);
+			mode_lib->vba.TFinalxFill = mode_lib->vba.RemainingFillLevel
+					/ (mode_lib->vba.SrcActiveDrainRate
+							* mode_lib->vba.XFCFillBWOverhead / 100);
+			mode_lib->vba.XFCPrefetchMargin[k] =
+					mode_lib->vba.XFCRemoteSurfaceFlipDelay
+							+ mode_lib->vba.TFinalxFill
+							+ (mode_lib->vba.DestinationLinesToRequestVMInVBlank[k]
+									+ mode_lib->vba.DestinationLinesToRequestRowInVBlank[k])
+									* mode_lib->vba.HTotal[k]
+									/ mode_lib->vba.PixelClock[k];
+		} else {
+			mode_lib->vba.XFCSlaveVUpdateOffset[k] = 0;
+			mode_lib->vba.XFCSlaveVupdateWidth[k] = 0;
+			mode_lib->vba.XFCSlaveVReadyOffset[k] = 0;
+			mode_lib->vba.XFCRemoteSurfaceFlipLatency[k] = 0;
+			mode_lib->vba.XFCPrechargeDelay[k] = 0;
+			mode_lib->vba.XFCTransferDelay[k] = 0;
+			mode_lib->vba.XFCPrefetchMargin[k] = 0;
+		}
+	}
+	{
+		unsigned int VStartupMargin = 0;
+		bool FirstMainPlane = true;
+
+		for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+			if (mode_lib->vba.BlendingAndTiming[k] == k) {
+				unsigned int Margin = (mode_lib->vba.MaxVStartupLines[k] - mode_lib->vba.VStartup[k])
+						* mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k];
+
+				if (FirstMainPlane) {
+					VStartupMargin = Margin;
+					FirstMainPlane = false;
+				} else
+					VStartupMargin = dml_min(VStartupMargin, Margin);
+		}
+
+		if (mode_lib->vba.UseMaximumVStartup) {
+			if (mode_lib->vba.VTotal_Max[k] == mode_lib->vba.VTotal[k]) {
+				//only use max vstart if it is not drr or lateflip.
+				mode_lib->vba.VStartup[k] = mode_lib->vba.MaxVStartupLines[mode_lib->vba.BlendingAndTiming[k]];
+			}
+		}
+	}
+}
+}
+
+static void dml20_DisplayPipeConfiguration(struct display_mode_lib *mode_lib)
+{
+	double BytePerPixDETY;
+	double BytePerPixDETC;
+	double Read256BytesBlockHeightY;
+	double Read256BytesBlockHeightC;
+	double Read256BytesBlockWidthY;
+	double Read256BytesBlockWidthC;
+	double MaximumSwathHeightY;
+	double MaximumSwathHeightC;
+	double MinimumSwathHeightY;
+	double MinimumSwathHeightC;
+	double SwathWidth;
+	double SwathWidthGranularityY;
+	double SwathWidthGranularityC;
+	double RoundedUpMaxSwathSizeBytesY;
+	double RoundedUpMaxSwathSizeBytesC;
+	unsigned int j, k;
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		bool MainPlaneDoesODMCombine = false;
+
+		if (mode_lib->vba.SourcePixelFormat[k] == dm_444_64) {
+			BytePerPixDETY = 8;
+			BytePerPixDETC = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_32) {
+			BytePerPixDETY = 4;
+			BytePerPixDETC = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_16) {
+			BytePerPixDETY = 2;
+			BytePerPixDETC = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_8) {
+			BytePerPixDETY = 1;
+			BytePerPixDETC = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_8) {
+			BytePerPixDETY = 1;
+			BytePerPixDETC = 2;
+		} else {
+			BytePerPixDETY = 4.0 / 3.0;
+			BytePerPixDETC = 8.0 / 3.0;
+		}
+
+		if ((mode_lib->vba.SourcePixelFormat[k] == dm_444_64
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_32
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_16
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_8)) {
+			if (mode_lib->vba.SurfaceTiling[k] == dm_sw_linear) {
+				Read256BytesBlockHeightY = 1;
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_64) {
+				Read256BytesBlockHeightY = 4;
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_32
+					|| mode_lib->vba.SourcePixelFormat[k] == dm_444_16) {
+				Read256BytesBlockHeightY = 8;
+			} else {
+				Read256BytesBlockHeightY = 16;
+			}
+			Read256BytesBlockWidthY = 256 / dml_ceil(BytePerPixDETY, 1)
+					/ Read256BytesBlockHeightY;
+			Read256BytesBlockHeightC = 0;
+			Read256BytesBlockWidthC = 0;
+		} else {
+			if (mode_lib->vba.SurfaceTiling[k] == dm_sw_linear) {
+				Read256BytesBlockHeightY = 1;
+				Read256BytesBlockHeightC = 1;
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_8) {
+				Read256BytesBlockHeightY = 16;
+				Read256BytesBlockHeightC = 8;
+			} else {
+				Read256BytesBlockHeightY = 8;
+				Read256BytesBlockHeightC = 8;
+			}
+			Read256BytesBlockWidthY = 256 / dml_ceil(BytePerPixDETY, 1)
+					/ Read256BytesBlockHeightY;
+			Read256BytesBlockWidthC = 256 / dml_ceil(BytePerPixDETC, 2)
+					/ Read256BytesBlockHeightC;
+		}
+
+		if (mode_lib->vba.SourceScan[k] == dm_horz) {
+			MaximumSwathHeightY = Read256BytesBlockHeightY;
+			MaximumSwathHeightC = Read256BytesBlockHeightC;
+		} else {
+			MaximumSwathHeightY = Read256BytesBlockWidthY;
+			MaximumSwathHeightC = Read256BytesBlockWidthC;
+		}
+
+		if ((mode_lib->vba.SourcePixelFormat[k] == dm_444_64
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_32
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_16
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_8)) {
+			if (mode_lib->vba.SurfaceTiling[k] == dm_sw_linear
+					|| (mode_lib->vba.SourcePixelFormat[k] == dm_444_64
+							&& (mode_lib->vba.SurfaceTiling[k]
+									== dm_sw_4kb_s
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_4kb_s_x
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_64kb_s
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_64kb_s_t
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_64kb_s_x
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_var_s
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_var_s_x)
+							&& mode_lib->vba.SourceScan[k] == dm_horz)) {
+				MinimumSwathHeightY = MaximumSwathHeightY;
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_8
+					&& mode_lib->vba.SourceScan[k] != dm_horz) {
+				MinimumSwathHeightY = MaximumSwathHeightY;
+			} else {
+				MinimumSwathHeightY = MaximumSwathHeightY / 2.0;
+			}
+			MinimumSwathHeightC = MaximumSwathHeightC;
+		} else {
+			if (mode_lib->vba.SurfaceTiling[k] == dm_sw_linear) {
+				MinimumSwathHeightY = MaximumSwathHeightY;
+				MinimumSwathHeightC = MaximumSwathHeightC;
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_8
+					&& mode_lib->vba.SourceScan[k] == dm_horz) {
+				MinimumSwathHeightY = MaximumSwathHeightY / 2.0;
+				MinimumSwathHeightC = MaximumSwathHeightC;
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_10
+					&& mode_lib->vba.SourceScan[k] == dm_horz) {
+				MinimumSwathHeightC = MaximumSwathHeightC / 2.0;
+				MinimumSwathHeightY = MaximumSwathHeightY;
+			} else {
+				MinimumSwathHeightY = MaximumSwathHeightY;
+				MinimumSwathHeightC = MaximumSwathHeightC;
+			}
+		}
+
+		if (mode_lib->vba.SourceScan[k] == dm_horz) {
+			SwathWidth = mode_lib->vba.ViewportWidth[k];
+		} else {
+			SwathWidth = mode_lib->vba.ViewportHeight[k];
+		}
+
+		if (mode_lib->vba.ODMCombineEnabled[k] == dm_odm_combine_mode_2to1) {
+			MainPlaneDoesODMCombine = true;
+		}
+		for (j = 0; j < mode_lib->vba.NumberOfActivePlanes; ++j) {
+			if (mode_lib->vba.BlendingAndTiming[k] == j
+					&& mode_lib->vba.ODMCombineEnabled[k] == dm_odm_combine_mode_2to1) {
+				MainPlaneDoesODMCombine = true;
+			}
+		}
+
+		if (MainPlaneDoesODMCombine == true) {
+			SwathWidth = dml_min(
+					SwathWidth,
+					mode_lib->vba.HActive[k] / 2.0 * mode_lib->vba.HRatio[k]);
+		} else {
+			if (mode_lib->vba.DPPPerPlane[k] == 0)
+				SwathWidth = 0;
+			else
+				SwathWidth = SwathWidth / mode_lib->vba.DPPPerPlane[k];
+		}
+
+		SwathWidthGranularityY = 256 / dml_ceil(BytePerPixDETY, 1) / MaximumSwathHeightY;
+		RoundedUpMaxSwathSizeBytesY = (dml_ceil(
+				(double) (SwathWidth - 1),
+				SwathWidthGranularityY) + SwathWidthGranularityY) * BytePerPixDETY
+				* MaximumSwathHeightY;
+		if (mode_lib->vba.SourcePixelFormat[k] == dm_420_10) {
+			RoundedUpMaxSwathSizeBytesY = dml_ceil(RoundedUpMaxSwathSizeBytesY, 256)
+					+ 256;
+		}
+		if (MaximumSwathHeightC > 0) {
+			SwathWidthGranularityC = 256.0 / dml_ceil(BytePerPixDETC, 2)
+					/ MaximumSwathHeightC;
+			RoundedUpMaxSwathSizeBytesC = (dml_ceil(
+					(double) (SwathWidth / 2.0 - 1),
+					SwathWidthGranularityC) + SwathWidthGranularityC)
+					* BytePerPixDETC * MaximumSwathHeightC;
+			if (mode_lib->vba.SourcePixelFormat[k] == dm_420_10) {
+				RoundedUpMaxSwathSizeBytesC = dml_ceil(
+						RoundedUpMaxSwathSizeBytesC,
+						256) + 256;
+			}
+		} else
+			RoundedUpMaxSwathSizeBytesC = 0.0;
+
+		if (RoundedUpMaxSwathSizeBytesY + RoundedUpMaxSwathSizeBytesC
+				<= mode_lib->vba.DETBufferSizeInKByte[0] * 1024.0 / 2.0) {
+			mode_lib->vba.SwathHeightY[k] = MaximumSwathHeightY;
+			mode_lib->vba.SwathHeightC[k] = MaximumSwathHeightC;
+		} else {
+			mode_lib->vba.SwathHeightY[k] = MinimumSwathHeightY;
+			mode_lib->vba.SwathHeightC[k] = MinimumSwathHeightC;
+		}
+
+		if (mode_lib->vba.SwathHeightC[k] == 0) {
+			mode_lib->vba.DETBufferSizeY[k] = mode_lib->vba.DETBufferSizeInKByte[0] * 1024;
+			mode_lib->vba.DETBufferSizeC[k] = 0;
+		} else if (mode_lib->vba.SwathHeightY[k] <= mode_lib->vba.SwathHeightC[k]) {
+			mode_lib->vba.DETBufferSizeY[k] = mode_lib->vba.DETBufferSizeInKByte[0]
+					* 1024.0 / 2;
+			mode_lib->vba.DETBufferSizeC[k] = mode_lib->vba.DETBufferSizeInKByte[0]
+					* 1024.0 / 2;
+		} else {
+			mode_lib->vba.DETBufferSizeY[k] = mode_lib->vba.DETBufferSizeInKByte[0]
+					* 1024.0 * 2 / 3;
+			mode_lib->vba.DETBufferSizeC[k] = mode_lib->vba.DETBufferSizeInKByte[0]
+					* 1024.0 / 3;
+		}
+	}
+}
+
+static double CalculateTWait(
+		unsigned int PrefetchMode,
+		double DRAMClockChangeLatency,
+		double UrgentLatencyPixelDataOnly,
+		double SREnterPlusExitTime)
+{
+	if (PrefetchMode == 0) {
+		return dml_max(
+				DRAMClockChangeLatency + UrgentLatencyPixelDataOnly,
+				dml_max(SREnterPlusExitTime, UrgentLatencyPixelDataOnly));
+	} else if (PrefetchMode == 1) {
+		return dml_max(SREnterPlusExitTime, UrgentLatencyPixelDataOnly);
+	} else {
+		return UrgentLatencyPixelDataOnly;
+	}
+}
+
+static double CalculateRemoteSurfaceFlipDelay(
+		struct display_mode_lib *mode_lib,
+		double VRatio,
+		double SwathWidth,
+		double Bpp,
+		double LineTime,
+		double XFCTSlvVupdateOffset,
+		double XFCTSlvVupdateWidth,
+		double XFCTSlvVreadyOffset,
+		double XFCXBUFLatencyTolerance,
+		double XFCFillBWOverhead,
+		double XFCSlvChunkSize,
+		double XFCBusTransportTime,
+		double TCalc,
+		double TWait,
+		double *SrcActiveDrainRate,
+		double *TInitXFill,
+		double *TslvChk)
+{
+	double TSlvSetup, AvgfillRate, result;
+
+	*SrcActiveDrainRate = VRatio * SwathWidth * Bpp / LineTime;
+	TSlvSetup = XFCTSlvVupdateOffset + XFCTSlvVupdateWidth + XFCTSlvVreadyOffset;
+	*TInitXFill = XFCXBUFLatencyTolerance / (1 + XFCFillBWOverhead / 100);
+	AvgfillRate = *SrcActiveDrainRate * (1 + XFCFillBWOverhead / 100);
+	*TslvChk = XFCSlvChunkSize / AvgfillRate;
+	dml_print(
+			"DML::CalculateRemoteSurfaceFlipDelay: SrcActiveDrainRate: %f\n",
+			*SrcActiveDrainRate);
+	dml_print("DML::CalculateRemoteSurfaceFlipDelay: TSlvSetup: %f\n", TSlvSetup);
+	dml_print("DML::CalculateRemoteSurfaceFlipDelay: TInitXFill: %f\n", *TInitXFill);
+	dml_print("DML::CalculateRemoteSurfaceFlipDelay: AvgfillRate: %f\n", AvgfillRate);
+	dml_print("DML::CalculateRemoteSurfaceFlipDelay: TslvChk: %f\n", *TslvChk);
+	result = 2 * XFCBusTransportTime + TSlvSetup + TCalc + TWait + *TslvChk + *TInitXFill; // TODO: This doesn't seem to match programming guide
+	dml_print("DML::CalculateRemoteSurfaceFlipDelay: RemoteSurfaceFlipDelay: %f\n", result);
+	return result;
+}
+
+static double CalculateWriteBackDelay(
+		enum source_format_class WritebackPixelFormat,
+		double WritebackHRatio,
+		double WritebackVRatio,
+		unsigned int WritebackLumaHTaps,
+		unsigned int WritebackLumaVTaps,
+		unsigned int WritebackChromaHTaps,
+		unsigned int WritebackChromaVTaps,
+		unsigned int WritebackDestinationWidth)
+{
+	double CalculateWriteBackDelay =
+			dml_max(
+					dml_ceil(WritebackLumaHTaps / 4.0, 1) / WritebackHRatio,
+					WritebackLumaVTaps * dml_ceil(1.0 / WritebackVRatio, 1)
+							* dml_ceil(
+									WritebackDestinationWidth
+											/ 4.0,
+									1)
+							+ dml_ceil(1.0 / WritebackVRatio, 1)
+									* (dml_ceil(
+											WritebackLumaVTaps
+													/ 4.0,
+											1) + 4));
+
+	if (WritebackPixelFormat != dm_444_32) {
+		CalculateWriteBackDelay =
+				dml_max(
+						CalculateWriteBackDelay,
+						dml_max(
+								dml_ceil(
+										WritebackChromaHTaps
+												/ 2.0,
+										1)
+										/ (2
+												* WritebackHRatio),
+								WritebackChromaVTaps
+										* dml_ceil(
+												1
+														/ (2
+																* WritebackVRatio),
+												1)
+										* dml_ceil(
+												WritebackDestinationWidth
+														/ 2.0
+														/ 2.0,
+												1)
+										+ dml_ceil(
+												1
+														/ (2
+																* WritebackVRatio),
+												1)
+												* (dml_ceil(
+														WritebackChromaVTaps
+																/ 4.0,
+														1)
+														+ 4)));
+	}
+	return CalculateWriteBackDelay;
+}
+
+static void CalculateActiveRowBandwidth(
+		bool GPUVMEnable,
+		enum source_format_class SourcePixelFormat,
+		double VRatio,
+		bool DCCEnable,
+		double LineTime,
+		unsigned int MetaRowByteLuma,
+		unsigned int MetaRowByteChroma,
+		unsigned int meta_row_height_luma,
+		unsigned int meta_row_height_chroma,
+		unsigned int PixelPTEBytesPerRowLuma,
+		unsigned int PixelPTEBytesPerRowChroma,
+		unsigned int dpte_row_height_luma,
+		unsigned int dpte_row_height_chroma,
+		double *meta_row_bw,
+		double *dpte_row_bw,
+		double *qual_row_bw)
+{
+	if (DCCEnable != true) {
+		*meta_row_bw = 0;
+	} else if (SourcePixelFormat == dm_420_8 || SourcePixelFormat == dm_420_10) {
+		*meta_row_bw = VRatio * MetaRowByteLuma / (meta_row_height_luma * LineTime)
+				+ VRatio / 2 * MetaRowByteChroma
+						/ (meta_row_height_chroma * LineTime);
+	} else {
+		*meta_row_bw = VRatio * MetaRowByteLuma / (meta_row_height_luma * LineTime);
+	}
+
+	if (GPUVMEnable != true) {
+		*dpte_row_bw = 0;
+	} else if (SourcePixelFormat == dm_420_8 || SourcePixelFormat == dm_420_10) {
+		*dpte_row_bw = VRatio * PixelPTEBytesPerRowLuma / (dpte_row_height_luma * LineTime)
+				+ VRatio / 2 * PixelPTEBytesPerRowChroma
+						/ (dpte_row_height_chroma * LineTime);
+	} else {
+		*dpte_row_bw = VRatio * PixelPTEBytesPerRowLuma / (dpte_row_height_luma * LineTime);
+	}
+
+	if ((SourcePixelFormat == dm_420_8 || SourcePixelFormat == dm_420_10)) {
+		*qual_row_bw = *meta_row_bw + *dpte_row_bw;
+	} else {
+		*qual_row_bw = 0;
+	}
+}
+
+static void CalculateFlipSchedule(
+		struct display_mode_lib *mode_lib,
+		double UrgentExtraLatency,
+		double UrgentLatencyPixelDataOnly,
+		unsigned int GPUVMMaxPageTableLevels,
+		bool GPUVMEnable,
+		double BandwidthAvailableForImmediateFlip,
+		unsigned int TotImmediateFlipBytes,
+		enum source_format_class SourcePixelFormat,
+		unsigned int ImmediateFlipBytes,
+		double LineTime,
+		double VRatio,
+		double Tno_bw,
+		double PDEAndMetaPTEBytesFrame,
+		unsigned int MetaRowByte,
+		unsigned int PixelPTEBytesPerRow,
+		bool DCCEnable,
+		unsigned int dpte_row_height,
+		unsigned int meta_row_height,
+		double qual_row_bw,
+		double *DestinationLinesToRequestVMInImmediateFlip,
+		double *DestinationLinesToRequestRowInImmediateFlip,
+		double *final_flip_bw,
+		bool *ImmediateFlipSupportedForPipe)
+{
+	double min_row_time = 0.0;
+
+	if (SourcePixelFormat == dm_420_8 || SourcePixelFormat == dm_420_10) {
+		*DestinationLinesToRequestVMInImmediateFlip = 0.0;
+		*DestinationLinesToRequestRowInImmediateFlip = 0.0;
+		*final_flip_bw = qual_row_bw;
+		*ImmediateFlipSupportedForPipe = true;
+	} else {
+		double TimeForFetchingMetaPTEImmediateFlip;
+		double TimeForFetchingRowInVBlankImmediateFlip;
+
+		if (GPUVMEnable == true) {
+			mode_lib->vba.ImmediateFlipBW[0] = BandwidthAvailableForImmediateFlip
+					* ImmediateFlipBytes / TotImmediateFlipBytes;
+			TimeForFetchingMetaPTEImmediateFlip =
+					dml_max(
+							Tno_bw
+									+ PDEAndMetaPTEBytesFrame
+											/ mode_lib->vba.ImmediateFlipBW[0],
+							dml_max(
+									UrgentExtraLatency
+											+ UrgentLatencyPixelDataOnly
+													* (GPUVMMaxPageTableLevels
+															- 1),
+									LineTime / 4.0));
+		} else {
+			TimeForFetchingMetaPTEImmediateFlip = 0;
+		}
+
+		*DestinationLinesToRequestVMInImmediateFlip = dml_floor(
+				4.0 * (TimeForFetchingMetaPTEImmediateFlip / LineTime + 0.125),
+				1) / 4.0;
+
+		if ((GPUVMEnable || DCCEnable)) {
+			mode_lib->vba.ImmediateFlipBW[0] = BandwidthAvailableForImmediateFlip
+					* ImmediateFlipBytes / TotImmediateFlipBytes;
+			TimeForFetchingRowInVBlankImmediateFlip = dml_max(
+					(MetaRowByte + PixelPTEBytesPerRow)
+							/ mode_lib->vba.ImmediateFlipBW[0],
+					dml_max(UrgentLatencyPixelDataOnly, LineTime / 4.0));
+		} else {
+			TimeForFetchingRowInVBlankImmediateFlip = 0;
+		}
+
+		*DestinationLinesToRequestRowInImmediateFlip = dml_floor(
+				4.0 * (TimeForFetchingRowInVBlankImmediateFlip / LineTime + 0.125),
+				1) / 4.0;
+
+		if (GPUVMEnable == true) {
+			*final_flip_bw =
+					dml_max(
+							PDEAndMetaPTEBytesFrame
+									/ (*DestinationLinesToRequestVMInImmediateFlip
+											* LineTime),
+							(MetaRowByte + PixelPTEBytesPerRow)
+									/ (TimeForFetchingRowInVBlankImmediateFlip
+											* LineTime));
+		} else if (MetaRowByte + PixelPTEBytesPerRow > 0) {
+			*final_flip_bw = (MetaRowByte + PixelPTEBytesPerRow)
+					/ (TimeForFetchingRowInVBlankImmediateFlip * LineTime);
+		} else {
+			*final_flip_bw = 0;
+		}
+
+		if (GPUVMEnable && !DCCEnable)
+			min_row_time = dpte_row_height * LineTime / VRatio;
+		else if (!GPUVMEnable && DCCEnable)
+			min_row_time = meta_row_height * LineTime / VRatio;
+		else
+			min_row_time = dml_min(dpte_row_height, meta_row_height) * LineTime
+					/ VRatio;
+
+		if (*DestinationLinesToRequestVMInImmediateFlip >= 8
+				|| *DestinationLinesToRequestRowInImmediateFlip >= 16
+				|| TimeForFetchingMetaPTEImmediateFlip
+						+ 2 * TimeForFetchingRowInVBlankImmediateFlip
+						> min_row_time)
+			*ImmediateFlipSupportedForPipe = false;
+		else
+			*ImmediateFlipSupportedForPipe = true;
+	}
+}
+
+static unsigned int TruncToValidBPP(
+		double DecimalBPP,
+		bool DSCEnabled,
+		enum output_encoder_class Output,
+		enum output_format_class Format,
+		unsigned int DSCInputBitPerComponent)
+{
+	if (Output == dm_hdmi) {
+		if (Format == dm_420) {
+			if (DecimalBPP >= 18)
+				return 18;
+			else if (DecimalBPP >= 15)
+				return 15;
+			else if (DecimalBPP >= 12)
+				return 12;
+			else
+				return BPP_INVALID;
+		} else if (Format == dm_444) {
+			if (DecimalBPP >= 36)
+				return 36;
+			else if (DecimalBPP >= 30)
+				return 30;
+			else if (DecimalBPP >= 24)
+				return 24;
+			else if (DecimalBPP >= 18)
+				return 18;
+			else
+				return BPP_INVALID;
+		} else {
+			if (DecimalBPP / 1.5 >= 24)
+				return 24;
+			else if (DecimalBPP / 1.5 >= 20)
+				return 20;
+			else if (DecimalBPP / 1.5 >= 16)
+				return 16;
+			else
+				return BPP_INVALID;
+		}
+	} else {
+		if (DSCEnabled) {
+			if (Format == dm_420) {
+				if (DecimalBPP < 6)
+					return BPP_INVALID;
+				else if (DecimalBPP >= 1.5 * DSCInputBitPerComponent - 1.0 / 16)
+					return 1.5 * DSCInputBitPerComponent - 1.0 / 16;
+				else
+					return dml_floor(16 * DecimalBPP, 1) / 16;
+			} else if (Format == dm_n422) {
+				if (DecimalBPP < 7)
+					return BPP_INVALID;
+				else if (DecimalBPP >= 2 * DSCInputBitPerComponent - 1.0 / 16)
+					return 2 * DSCInputBitPerComponent - 1.0 / 16;
+				else
+					return dml_floor(16 * DecimalBPP, 1) / 16;
+			} else {
+				if (DecimalBPP < 8)
+					return BPP_INVALID;
+				else if (DecimalBPP >= 3 * DSCInputBitPerComponent - 1.0 / 16)
+					return 3 * DSCInputBitPerComponent - 1.0 / 16;
+				else
+					return dml_floor(16 * DecimalBPP, 1) / 16;
+			}
+		} else if (Format == dm_420) {
+			if (DecimalBPP >= 18)
+				return 18;
+			else if (DecimalBPP >= 15)
+				return 15;
+			else if (DecimalBPP >= 12)
+				return 12;
+			else
+				return BPP_INVALID;
+		} else if (Format == dm_s422 || Format == dm_n422) {
+			if (DecimalBPP >= 24)
+				return 24;
+			else if (DecimalBPP >= 20)
+				return 20;
+			else if (DecimalBPP >= 16)
+				return 16;
+			else
+				return BPP_INVALID;
+		} else {
+			if (DecimalBPP >= 36)
+				return 36;
+			else if (DecimalBPP >= 30)
+				return 30;
+			else if (DecimalBPP >= 24)
+				return 24;
+			else if (DecimalBPP >= 18)
+				return 18;
+			else
+				return BPP_INVALID;
+		}
+	}
+}
+
+void dml20_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_lib)
+{
+	struct vba_vars_st *locals = &mode_lib->vba;
+
+	int i;
+	unsigned int j, k, m;
+
+	/*MODE SUPPORT, VOLTAGE STATE AND SOC CONFIGURATION*/
+
+	/*Scale Ratio, taps Support Check*/
+
+	mode_lib->vba.ScaleRatioAndTapsSupport = true;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.ScalerEnabled[k] == false
+				&& ((mode_lib->vba.SourcePixelFormat[k] != dm_444_64
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_444_32
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_444_16
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_mono_16
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_mono_8)
+						|| mode_lib->vba.HRatio[k] != 1.0
+						|| mode_lib->vba.htaps[k] != 1.0
+						|| mode_lib->vba.VRatio[k] != 1.0
+						|| mode_lib->vba.vtaps[k] != 1.0)) {
+			mode_lib->vba.ScaleRatioAndTapsSupport = false;
+		} else if (mode_lib->vba.vtaps[k] < 1.0 || mode_lib->vba.vtaps[k] > 8.0
+				|| mode_lib->vba.htaps[k] < 1.0 || mode_lib->vba.htaps[k] > 8.0
+				|| (mode_lib->vba.htaps[k] > 1.0
+						&& (mode_lib->vba.htaps[k] % 2) == 1)
+				|| mode_lib->vba.HRatio[k] > mode_lib->vba.MaxHSCLRatio
+				|| mode_lib->vba.VRatio[k] > mode_lib->vba.MaxVSCLRatio
+				|| mode_lib->vba.HRatio[k] > mode_lib->vba.htaps[k]
+				|| mode_lib->vba.VRatio[k] > mode_lib->vba.vtaps[k]
+				|| (mode_lib->vba.SourcePixelFormat[k] != dm_444_64
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_444_32
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_444_16
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_mono_16
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_mono_8
+						&& (mode_lib->vba.HRatio[k] / 2.0
+								> mode_lib->vba.HTAPsChroma[k]
+								|| mode_lib->vba.VRatio[k] / 2.0
+										> mode_lib->vba.VTAPsChroma[k]))) {
+			mode_lib->vba.ScaleRatioAndTapsSupport = false;
+		}
+	}
+	/*Source Format, Pixel Format and Scan Support Check*/
+
+	mode_lib->vba.SourceFormatPixelAndScanSupport = true;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if ((mode_lib->vba.SurfaceTiling[k] == dm_sw_linear
+				&& mode_lib->vba.SourceScan[k] != dm_horz)
+				|| ((mode_lib->vba.SurfaceTiling[k] == dm_sw_4kb_d
+						|| mode_lib->vba.SurfaceTiling[k] == dm_sw_4kb_d_x
+						|| mode_lib->vba.SurfaceTiling[k] == dm_sw_64kb_d
+						|| mode_lib->vba.SurfaceTiling[k] == dm_sw_64kb_d_t
+						|| mode_lib->vba.SurfaceTiling[k] == dm_sw_64kb_d_x
+						|| mode_lib->vba.SurfaceTiling[k] == dm_sw_var_d
+						|| mode_lib->vba.SurfaceTiling[k] == dm_sw_var_d_x)
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_444_64)
+				|| (mode_lib->vba.SurfaceTiling[k] == dm_sw_64kb_r_x
+						&& (mode_lib->vba.SourcePixelFormat[k] == dm_mono_8
+								|| mode_lib->vba.SourcePixelFormat[k]
+										== dm_420_8
+								|| mode_lib->vba.SourcePixelFormat[k]
+										== dm_420_10))
+				|| (((mode_lib->vba.SurfaceTiling[k] == dm_sw_gfx7_2d_thin_gl
+						|| mode_lib->vba.SurfaceTiling[k]
+								== dm_sw_gfx7_2d_thin_l_vp)
+						&& !((mode_lib->vba.SourcePixelFormat[k]
+								== dm_444_64
+								|| mode_lib->vba.SourcePixelFormat[k]
+										== dm_444_32)
+								&& mode_lib->vba.SourceScan[k]
+										== dm_horz
+								&& mode_lib->vba.SupportGFX7CompatibleTilingIn32bppAnd64bpp
+										== true
+								&& mode_lib->vba.DCCEnable[k]
+										== false))
+						|| (mode_lib->vba.DCCEnable[k] == true
+								&& (mode_lib->vba.SurfaceTiling[k]
+										== dm_sw_linear
+										|| mode_lib->vba.SourcePixelFormat[k]
+												== dm_420_8
+										|| mode_lib->vba.SourcePixelFormat[k]
+												== dm_420_10)))) {
+			mode_lib->vba.SourceFormatPixelAndScanSupport = false;
+		}
+	}
+	/*Bandwidth Support Check*/
+
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.SourcePixelFormat[k] == dm_444_64) {
+			locals->BytePerPixelInDETY[k] = 8.0;
+			locals->BytePerPixelInDETC[k] = 0.0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_32) {
+			locals->BytePerPixelInDETY[k] = 4.0;
+			locals->BytePerPixelInDETC[k] = 0.0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_16
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_mono_16) {
+			locals->BytePerPixelInDETY[k] = 2.0;
+			locals->BytePerPixelInDETC[k] = 0.0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_mono_8) {
+			locals->BytePerPixelInDETY[k] = 1.0;
+			locals->BytePerPixelInDETC[k] = 0.0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_8) {
+			locals->BytePerPixelInDETY[k] = 1.0;
+			locals->BytePerPixelInDETC[k] = 2.0;
+		} else {
+			locals->BytePerPixelInDETY[k] = 4.0 / 3;
+			locals->BytePerPixelInDETC[k] = 8.0 / 3;
+		}
+		if (mode_lib->vba.SourceScan[k] == dm_horz) {
+			locals->SwathWidthYSingleDPP[k] = mode_lib->vba.ViewportWidth[k];
+		} else {
+			locals->SwathWidthYSingleDPP[k] = mode_lib->vba.ViewportHeight[k];
+		}
+	}
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		locals->ReadBandwidthLuma[k] = locals->SwathWidthYSingleDPP[k] * dml_ceil(locals->BytePerPixelInDETY[k], 1.0)
+				/ (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k]) * mode_lib->vba.VRatio[k];
+		locals->ReadBandwidthChroma[k] = locals->SwathWidthYSingleDPP[k] / 2 * dml_ceil(locals->BytePerPixelInDETC[k], 2.0)
+				/ (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k]) * mode_lib->vba.VRatio[k] / 2.0;
+		locals->ReadBandwidth[k] = locals->ReadBandwidthLuma[k] + locals->ReadBandwidthChroma[k];
+	}
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.WritebackEnable[k] == true
+				&& mode_lib->vba.WritebackPixelFormat[k] == dm_444_32) {
+			locals->WriteBandwidth[k] = mode_lib->vba.WritebackDestinationWidth[k]
+					* mode_lib->vba.WritebackDestinationHeight[k]
+					/ (mode_lib->vba.WritebackSourceHeight[k]
+							* mode_lib->vba.HTotal[k]
+							/ mode_lib->vba.PixelClock[k]) * 4.0;
+		} else if (mode_lib->vba.WritebackEnable[k] == true
+				&& mode_lib->vba.WritebackPixelFormat[k] == dm_420_10) {
+			locals->WriteBandwidth[k] = mode_lib->vba.WritebackDestinationWidth[k]
+					* mode_lib->vba.WritebackDestinationHeight[k]
+					/ (mode_lib->vba.WritebackSourceHeight[k]
+							* mode_lib->vba.HTotal[k]
+							/ mode_lib->vba.PixelClock[k]) * 3.0;
+		} else if (mode_lib->vba.WritebackEnable[k] == true) {
+			locals->WriteBandwidth[k] = mode_lib->vba.WritebackDestinationWidth[k]
+					* mode_lib->vba.WritebackDestinationHeight[k]
+					/ (mode_lib->vba.WritebackSourceHeight[k]
+							* mode_lib->vba.HTotal[k]
+							/ mode_lib->vba.PixelClock[k]) * 1.5;
+		} else {
+			locals->WriteBandwidth[k] = 0.0;
+		}
+	}
+	mode_lib->vba.DCCEnabledInAnyPlane = false;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.DCCEnable[k] == true) {
+			mode_lib->vba.DCCEnabledInAnyPlane = true;
+		}
+	}
+	mode_lib->vba.UrgentLatency = mode_lib->vba.UrgentLatencyPixelDataOnly;
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		locals->FabricAndDRAMBandwidthPerState[i] = dml_min(
+				mode_lib->vba.DRAMSpeedPerState[i] * mode_lib->vba.NumberOfChannels
+						* mode_lib->vba.DRAMChannelWidth,
+				mode_lib->vba.FabricClockPerState[i]
+						* mode_lib->vba.FabricDatapathToDCNDataReturn) / 1000;
+		locals->ReturnBWToDCNPerState = dml_min(locals->ReturnBusWidth * locals->DCFCLKPerState[i],
+				locals->FabricAndDRAMBandwidthPerState[i] * 1000)
+				* locals->PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelDataOnly / 100;
+
+		locals->ReturnBWPerState[i][0] = locals->ReturnBWToDCNPerState;
+
+		if (locals->DCCEnabledInAnyPlane == true && locals->ReturnBWToDCNPerState > locals->DCFCLKPerState[i] * locals->ReturnBusWidth / 4) {
+			locals->ReturnBWPerState[i][0] = dml_min(locals->ReturnBWPerState[i][0],
+					locals->ReturnBWToDCNPerState * 4 * (1 - locals->UrgentLatency /
+					((locals->ROBBufferSizeInKByte - locals->PixelChunkSizeInKByte) * 1024
+					/ (locals->ReturnBWToDCNPerState - locals->DCFCLKPerState[i]
+					* locals->ReturnBusWidth / 4) + locals->UrgentLatency)));
+		}
+		locals->CriticalPoint = 2 * locals->ReturnBusWidth * locals->DCFCLKPerState[i] *
+				locals->UrgentLatency / (locals->ReturnBWToDCNPerState * locals->UrgentLatency
+				+ (locals->ROBBufferSizeInKByte - locals->PixelChunkSizeInKByte) * 1024);
+
+		if (locals->DCCEnabledInAnyPlane && locals->CriticalPoint > 1 && locals->CriticalPoint < 4) {
+			locals->ReturnBWPerState[i][0] = dml_min(locals->ReturnBWPerState[i][0],
+				4 * locals->ReturnBWToDCNPerState *
+				(locals->ROBBufferSizeInKByte - locals->PixelChunkSizeInKByte) * 1024
+				* locals->ReturnBusWidth * locals->DCFCLKPerState[i] * locals->UrgentLatency /
+				dml_pow((locals->ReturnBWToDCNPerState * locals->UrgentLatency
+				+ (locals->ROBBufferSizeInKByte - locals->PixelChunkSizeInKByte) * 1024), 2));
+		}
+
+		locals->ReturnBWToDCNPerState = dml_min(locals->ReturnBusWidth *
+				locals->DCFCLKPerState[i], locals->FabricAndDRAMBandwidthPerState[i] * 1000);
+
+		if (locals->DCCEnabledInAnyPlane == true && locals->ReturnBWToDCNPerState > locals->DCFCLKPerState[i] * locals->ReturnBusWidth / 4) {
+			locals->ReturnBWPerState[i][0] = dml_min(locals->ReturnBWPerState[i][0],
+					locals->ReturnBWToDCNPerState * 4 * (1 - locals->UrgentLatency /
+					((locals->ROBBufferSizeInKByte - locals->PixelChunkSizeInKByte) * 1024
+					/ (locals->ReturnBWToDCNPerState - locals->DCFCLKPerState[i]
+					* locals->ReturnBusWidth / 4) + locals->UrgentLatency)));
+		}
+		locals->CriticalPoint = 2 * locals->ReturnBusWidth * locals->DCFCLKPerState[i] *
+				locals->UrgentLatency / (locals->ReturnBWToDCNPerState * locals->UrgentLatency
+				+ (locals->ROBBufferSizeInKByte - locals->PixelChunkSizeInKByte) * 1024);
+
+		if (locals->DCCEnabledInAnyPlane && locals->CriticalPoint > 1 && locals->CriticalPoint < 4) {
+			locals->ReturnBWPerState[i][0] = dml_min(locals->ReturnBWPerState[i][0],
+				4 * locals->ReturnBWToDCNPerState *
+				(locals->ROBBufferSizeInKByte - locals->PixelChunkSizeInKByte) * 1024
+				* locals->ReturnBusWidth * locals->DCFCLKPerState[i] * locals->UrgentLatency /
+				dml_pow((locals->ReturnBWToDCNPerState * locals->UrgentLatency
+				+ (locals->ROBBufferSizeInKByte - locals->PixelChunkSizeInKByte) * 1024), 2));
+		}
+	}
+	/*Writeback Latency support check*/
+
+	mode_lib->vba.WritebackLatencySupport = true;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.WritebackEnable[k] == true) {
+			if (mode_lib->vba.WritebackPixelFormat[k] == dm_444_32) {
+				if (locals->WriteBandwidth[k]
+						> (mode_lib->vba.WritebackInterfaceLumaBufferSize
+								+ mode_lib->vba.WritebackInterfaceChromaBufferSize)
+								/ mode_lib->vba.WritebackLatency) {
+					mode_lib->vba.WritebackLatencySupport = false;
+				}
+			} else {
+				if (locals->WriteBandwidth[k]
+						> 1.5
+								* dml_min(
+										mode_lib->vba.WritebackInterfaceLumaBufferSize,
+										2.0
+												* mode_lib->vba.WritebackInterfaceChromaBufferSize)
+								/ mode_lib->vba.WritebackLatency) {
+					mode_lib->vba.WritebackLatencySupport = false;
+				}
+			}
+		}
+	}
+	/*Re-ordering Buffer Support Check*/
+
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		locals->UrgentRoundTripAndOutOfOrderLatencyPerState[i] =
+				(mode_lib->vba.RoundTripPingLatencyCycles + 32.0) / mode_lib->vba.DCFCLKPerState[i]
+				+ locals->UrgentOutOfOrderReturnPerChannel * mode_lib->vba.NumberOfChannels / locals->ReturnBWPerState[i][0];
+		if ((mode_lib->vba.ROBBufferSizeInKByte - mode_lib->vba.PixelChunkSizeInKByte) * 1024.0 / locals->ReturnBWPerState[i][0]
+				> locals->UrgentRoundTripAndOutOfOrderLatencyPerState[i]) {
+			locals->ROBSupport[i][0] = true;
+		} else {
+			locals->ROBSupport[i][0] = false;
+		}
+	}
+	/*Writeback Mode Support Check*/
+
+	mode_lib->vba.TotalNumberOfActiveWriteback = 0;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.WritebackEnable[k] == true) {
+			if (mode_lib->vba.ActiveWritebacksPerPlane[k] == 0)
+				mode_lib->vba.ActiveWritebacksPerPlane[k] = 1;
+			mode_lib->vba.TotalNumberOfActiveWriteback =
+					mode_lib->vba.TotalNumberOfActiveWriteback
+							+ mode_lib->vba.ActiveWritebacksPerPlane[k];
+		}
+	}
+	mode_lib->vba.WritebackModeSupport = true;
+	if (mode_lib->vba.TotalNumberOfActiveWriteback > mode_lib->vba.MaxNumWriteback) {
+		mode_lib->vba.WritebackModeSupport = false;
+	}
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.WritebackEnable[k] == true
+				&& mode_lib->vba.Writeback10bpc420Supported != true
+				&& mode_lib->vba.WritebackPixelFormat[k] == dm_420_10) {
+			mode_lib->vba.WritebackModeSupport = false;
+		}
+	}
+	/*Writeback Scale Ratio and Taps Support Check*/
+
+	mode_lib->vba.WritebackScaleRatioAndTapsSupport = true;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.WritebackEnable[k] == true) {
+			if (mode_lib->vba.WritebackLumaAndChromaScalingSupported == false
+					&& (mode_lib->vba.WritebackHRatio[k] != 1.0
+							|| mode_lib->vba.WritebackVRatio[k] != 1.0)) {
+				mode_lib->vba.WritebackScaleRatioAndTapsSupport = false;
+			}
+			if (mode_lib->vba.WritebackHRatio[k] > mode_lib->vba.WritebackMaxHSCLRatio
+					|| mode_lib->vba.WritebackVRatio[k]
+							> mode_lib->vba.WritebackMaxVSCLRatio
+					|| mode_lib->vba.WritebackHRatio[k]
+							< mode_lib->vba.WritebackMinHSCLRatio
+					|| mode_lib->vba.WritebackVRatio[k]
+							< mode_lib->vba.WritebackMinVSCLRatio
+					|| mode_lib->vba.WritebackLumaHTaps[k]
+							> mode_lib->vba.WritebackMaxHSCLTaps
+					|| mode_lib->vba.WritebackLumaVTaps[k]
+							> mode_lib->vba.WritebackMaxVSCLTaps
+					|| mode_lib->vba.WritebackHRatio[k]
+							> mode_lib->vba.WritebackLumaHTaps[k]
+					|| mode_lib->vba.WritebackVRatio[k]
+							> mode_lib->vba.WritebackLumaVTaps[k]
+					|| (mode_lib->vba.WritebackLumaHTaps[k] > 2.0
+							&& ((mode_lib->vba.WritebackLumaHTaps[k] % 2)
+									== 1))
+					|| (mode_lib->vba.WritebackPixelFormat[k] != dm_444_32
+							&& (mode_lib->vba.WritebackChromaHTaps[k]
+									> mode_lib->vba.WritebackMaxHSCLTaps
+									|| mode_lib->vba.WritebackChromaVTaps[k]
+											> mode_lib->vba.WritebackMaxVSCLTaps
+									|| 2.0
+											* mode_lib->vba.WritebackHRatio[k]
+											> mode_lib->vba.WritebackChromaHTaps[k]
+									|| 2.0
+											* mode_lib->vba.WritebackVRatio[k]
+											> mode_lib->vba.WritebackChromaVTaps[k]
+									|| (mode_lib->vba.WritebackChromaHTaps[k] > 2.0
+										&& ((mode_lib->vba.WritebackChromaHTaps[k] % 2) == 1))))) {
+				mode_lib->vba.WritebackScaleRatioAndTapsSupport = false;
+			}
+			if (mode_lib->vba.WritebackVRatio[k] < 1.0) {
+				mode_lib->vba.WritebackLumaVExtra =
+						dml_max(1.0 - 2.0 / dml_ceil(1.0 / mode_lib->vba.WritebackVRatio[k], 1.0), 0.0);
+			} else {
+				mode_lib->vba.WritebackLumaVExtra = -1;
+			}
+			if ((mode_lib->vba.WritebackPixelFormat[k] == dm_444_32
+					&& mode_lib->vba.WritebackLumaVTaps[k]
+							> (mode_lib->vba.WritebackLineBufferLumaBufferSize
+									+ mode_lib->vba.WritebackLineBufferChromaBufferSize)
+									/ 3.0
+									/ mode_lib->vba.WritebackDestinationWidth[k]
+									- mode_lib->vba.WritebackLumaVExtra)
+					|| (mode_lib->vba.WritebackPixelFormat[k] == dm_420_8
+							&& mode_lib->vba.WritebackLumaVTaps[k]
+									> mode_lib->vba.WritebackLineBufferLumaBufferSize
+											* 8.0 / 10.0 / mode_lib->vba.WritebackDestinationWidth[k]
+											- mode_lib->vba.WritebackLumaVExtra)
+					|| (mode_lib->vba.WritebackPixelFormat[k] == dm_420_10
+							&& mode_lib->vba.WritebackLumaVTaps[k]
+									> mode_lib->vba.WritebackLineBufferLumaBufferSize
+											* 8.0 / 10.0
+											/ mode_lib->vba.WritebackDestinationWidth[k]
+											- mode_lib->vba.WritebackLumaVExtra)) {
+				mode_lib->vba.WritebackScaleRatioAndTapsSupport = false;
+			}
+			if (2.0 * mode_lib->vba.WritebackVRatio[k] < 1) {
+				mode_lib->vba.WritebackChromaVExtra = 0.0;
+			} else {
+				mode_lib->vba.WritebackChromaVExtra = -1;
+			}
+			if ((mode_lib->vba.WritebackPixelFormat[k] == dm_420_8
+					&& mode_lib->vba.WritebackChromaVTaps[k]
+							> mode_lib->vba.WritebackLineBufferChromaBufferSize
+									* 8.0 / 10.0 / mode_lib->vba.WritebackDestinationWidth[k]
+									- mode_lib->vba.WritebackChromaVExtra)
+					|| (mode_lib->vba.WritebackPixelFormat[k] == dm_420_10
+							&& mode_lib->vba.WritebackChromaVTaps[k]
+									> mode_lib->vba.WritebackLineBufferChromaBufferSize
+											* 8.0 / 10.0
+											/ mode_lib->vba.WritebackDestinationWidth[k]
+											- mode_lib->vba.WritebackChromaVExtra)) {
+				mode_lib->vba.WritebackScaleRatioAndTapsSupport = false;
+			}
+		}
+	}
+	/*Maximum DISPCLK/DPPCLK Support check*/
+
+	mode_lib->vba.WritebackRequiredDISPCLK = 0.0;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.WritebackEnable[k] == true) {
+			mode_lib->vba.WritebackRequiredDISPCLK =
+					dml_max(
+							mode_lib->vba.WritebackRequiredDISPCLK,
+							CalculateWriteBackDISPCLK(
+									mode_lib->vba.WritebackPixelFormat[k],
+									mode_lib->vba.PixelClock[k],
+									mode_lib->vba.WritebackHRatio[k],
+									mode_lib->vba.WritebackVRatio[k],
+									mode_lib->vba.WritebackLumaHTaps[k],
+									mode_lib->vba.WritebackLumaVTaps[k],
+									mode_lib->vba.WritebackChromaHTaps[k],
+									mode_lib->vba.WritebackChromaVTaps[k],
+									mode_lib->vba.WritebackDestinationWidth[k],
+									mode_lib->vba.HTotal[k],
+									mode_lib->vba.WritebackChromaLineBufferWidth));
+		}
+	}
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.HRatio[k] > 1.0) {
+			locals->PSCL_FACTOR[k] = dml_min(
+					mode_lib->vba.MaxDCHUBToPSCLThroughput,
+					mode_lib->vba.MaxPSCLToLBThroughput
+							* mode_lib->vba.HRatio[k]
+							/ dml_ceil(
+									mode_lib->vba.htaps[k]
+											/ 6.0,
+									1.0));
+		} else {
+			locals->PSCL_FACTOR[k] = dml_min(
+					mode_lib->vba.MaxDCHUBToPSCLThroughput,
+					mode_lib->vba.MaxPSCLToLBThroughput);
+		}
+		if (locals->BytePerPixelInDETC[k] == 0.0) {
+			locals->PSCL_FACTOR_CHROMA[k] = 0.0;
+			locals->MinDPPCLKUsingSingleDPP[k] =
+					mode_lib->vba.PixelClock[k]
+							* dml_max3(
+									mode_lib->vba.vtaps[k] / 6.0
+											* dml_min(
+													1.0,
+													mode_lib->vba.HRatio[k]),
+									mode_lib->vba.HRatio[k]
+											* mode_lib->vba.VRatio[k]
+											/ locals->PSCL_FACTOR[k],
+									1.0);
+			if ((mode_lib->vba.htaps[k] > 6.0 || mode_lib->vba.vtaps[k] > 6.0)
+					&& locals->MinDPPCLKUsingSingleDPP[k]
+							< 2.0 * mode_lib->vba.PixelClock[k]) {
+				locals->MinDPPCLKUsingSingleDPP[k] = 2.0
+						* mode_lib->vba.PixelClock[k];
+			}
+		} else {
+			if (mode_lib->vba.HRatio[k] / 2.0 > 1.0) {
+				locals->PSCL_FACTOR_CHROMA[k] =
+						dml_min(
+								mode_lib->vba.MaxDCHUBToPSCLThroughput,
+								mode_lib->vba.MaxPSCLToLBThroughput
+										* mode_lib->vba.HRatio[k]
+										/ 2.0
+										/ dml_ceil(
+												mode_lib->vba.HTAPsChroma[k]
+														/ 6.0,
+												1.0));
+			} else {
+				locals->PSCL_FACTOR_CHROMA[k] = dml_min(
+						mode_lib->vba.MaxDCHUBToPSCLThroughput,
+						mode_lib->vba.MaxPSCLToLBThroughput);
+			}
+			locals->MinDPPCLKUsingSingleDPP[k] =
+					mode_lib->vba.PixelClock[k]
+							* dml_max5(
+									mode_lib->vba.vtaps[k] / 6.0
+											* dml_min(
+													1.0,
+													mode_lib->vba.HRatio[k]),
+									mode_lib->vba.HRatio[k]
+											* mode_lib->vba.VRatio[k]
+											/ locals->PSCL_FACTOR[k],
+									mode_lib->vba.VTAPsChroma[k]
+											/ 6.0
+											* dml_min(
+													1.0,
+													mode_lib->vba.HRatio[k]
+															/ 2.0),
+									mode_lib->vba.HRatio[k]
+											* mode_lib->vba.VRatio[k]
+											/ 4.0
+											/ locals->PSCL_FACTOR_CHROMA[k],
+									1.0);
+			if ((mode_lib->vba.htaps[k] > 6.0 || mode_lib->vba.vtaps[k] > 6.0
+					|| mode_lib->vba.HTAPsChroma[k] > 6.0
+					|| mode_lib->vba.VTAPsChroma[k] > 6.0)
+					&& locals->MinDPPCLKUsingSingleDPP[k]
+							< 2.0 * mode_lib->vba.PixelClock[k]) {
+				locals->MinDPPCLKUsingSingleDPP[k] = 2.0
+						* mode_lib->vba.PixelClock[k];
+			}
+		}
+	}
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		Calculate256BBlockSizes(
+				mode_lib->vba.SourcePixelFormat[k],
+				mode_lib->vba.SurfaceTiling[k],
+				dml_ceil(locals->BytePerPixelInDETY[k], 1.0),
+				dml_ceil(locals->BytePerPixelInDETC[k], 2.0),
+				&locals->Read256BlockHeightY[k],
+				&locals->Read256BlockHeightC[k],
+				&locals->Read256BlockWidthY[k],
+				&locals->Read256BlockWidthC[k]);
+		if (mode_lib->vba.SourceScan[k] == dm_horz) {
+			locals->MaxSwathHeightY[k] = locals->Read256BlockHeightY[k];
+			locals->MaxSwathHeightC[k] = locals->Read256BlockHeightC[k];
+		} else {
+			locals->MaxSwathHeightY[k] = locals->Read256BlockWidthY[k];
+			locals->MaxSwathHeightC[k] = locals->Read256BlockWidthC[k];
+		}
+		if ((mode_lib->vba.SourcePixelFormat[k] == dm_444_64
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_32
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_16
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_mono_16
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_mono_8)) {
+			if (mode_lib->vba.SurfaceTiling[k] == dm_sw_linear
+					|| (mode_lib->vba.SourcePixelFormat[k] == dm_444_64
+							&& (mode_lib->vba.SurfaceTiling[k]
+									== dm_sw_4kb_s
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_4kb_s_x
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_64kb_s
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_64kb_s_t
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_64kb_s_x
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_var_s
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_var_s_x)
+							&& mode_lib->vba.SourceScan[k] == dm_horz)) {
+				locals->MinSwathHeightY[k] = locals->MaxSwathHeightY[k];
+			} else {
+				locals->MinSwathHeightY[k] = locals->MaxSwathHeightY[k]
+						/ 2.0;
+			}
+			locals->MinSwathHeightC[k] = locals->MaxSwathHeightC[k];
+		} else {
+			if (mode_lib->vba.SurfaceTiling[k] == dm_sw_linear) {
+				locals->MinSwathHeightY[k] = locals->MaxSwathHeightY[k];
+				locals->MinSwathHeightC[k] = locals->MaxSwathHeightC[k];
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_8
+					&& mode_lib->vba.SourceScan[k] == dm_horz) {
+				locals->MinSwathHeightY[k] = locals->MaxSwathHeightY[k]
+						/ 2.0;
+				locals->MinSwathHeightC[k] = locals->MaxSwathHeightC[k];
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_10
+					&& mode_lib->vba.SourceScan[k] == dm_horz) {
+				locals->MinSwathHeightC[k] = locals->MaxSwathHeightC[k]
+						/ 2.0;
+				locals->MinSwathHeightY[k] = locals->MaxSwathHeightY[k];
+			} else {
+				locals->MinSwathHeightY[k] = locals->MaxSwathHeightY[k];
+				locals->MinSwathHeightC[k] = locals->MaxSwathHeightC[k];
+			}
+		}
+		if (mode_lib->vba.SurfaceTiling[k] == dm_sw_linear) {
+			mode_lib->vba.MaximumSwathWidthSupport = 8192.0;
+		} else {
+			mode_lib->vba.MaximumSwathWidthSupport = 5120.0;
+		}
+		mode_lib->vba.MaximumSwathWidthInDETBuffer =
+				dml_min(
+						mode_lib->vba.MaximumSwathWidthSupport,
+						mode_lib->vba.DETBufferSizeInKByte[0] * 1024.0 / 2.0
+								/ (locals->BytePerPixelInDETY[k]
+										* locals->MinSwathHeightY[k]
+										+ locals->BytePerPixelInDETC[k]
+												/ 2.0
+												* locals->MinSwathHeightC[k]));
+		if (locals->BytePerPixelInDETC[k] == 0.0) {
+			mode_lib->vba.MaximumSwathWidthInLineBuffer =
+					mode_lib->vba.LineBufferSize
+							* dml_max(mode_lib->vba.HRatio[k], 1.0)
+							/ mode_lib->vba.LBBitPerPixel[k]
+							/ (mode_lib->vba.vtaps[k]
+									+ dml_max(
+											dml_ceil(
+													mode_lib->vba.VRatio[k],
+													1.0)
+													- 2,
+											0.0));
+		} else {
+			mode_lib->vba.MaximumSwathWidthInLineBuffer =
+					dml_min(
+							mode_lib->vba.LineBufferSize
+									* dml_max(
+											mode_lib->vba.HRatio[k],
+											1.0)
+									/ mode_lib->vba.LBBitPerPixel[k]
+									/ (mode_lib->vba.vtaps[k]
+											+ dml_max(
+													dml_ceil(
+															mode_lib->vba.VRatio[k],
+															1.0)
+															- 2,
+													0.0)),
+							2.0 * mode_lib->vba.LineBufferSize
+									* dml_max(
+											mode_lib->vba.HRatio[k]
+													/ 2.0,
+											1.0)
+									/ mode_lib->vba.LBBitPerPixel[k]
+									/ (mode_lib->vba.VTAPsChroma[k]
+											+ dml_max(
+													dml_ceil(
+															mode_lib->vba.VRatio[k]
+																	/ 2.0,
+															1.0)
+															- 2,
+													0.0)));
+		}
+		locals->MaximumSwathWidth[k] = dml_min(
+				mode_lib->vba.MaximumSwathWidthInDETBuffer,
+				mode_lib->vba.MaximumSwathWidthInLineBuffer);
+	}
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		for (j = 0; j < 2; j++) {
+			mode_lib->vba.MaxDispclkRoundedDownToDFSGranularity = RoundToDFSGranularityDown(
+				mode_lib->vba.MaxDispclk[i],
+				mode_lib->vba.DISPCLKDPPCLKVCOSpeed);
+			mode_lib->vba.MaxDppclkRoundedDownToDFSGranularity = RoundToDFSGranularityDown(
+				mode_lib->vba.MaxDppclk[i],
+				mode_lib->vba.DISPCLKDPPCLKVCOSpeed);
+			locals->RequiredDISPCLK[i][j] = 0.0;
+			locals->DISPCLK_DPPCLK_Support[i][j] = true;
+			for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+				mode_lib->vba.PlaneRequiredDISPCLKWithoutODMCombine =
+						mode_lib->vba.PixelClock[k] * (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0)
+								* (1.0 + mode_lib->vba.DISPCLKRampingMargin / 100.0);
+				if (mode_lib->vba.PlaneRequiredDISPCLKWithoutODMCombine >= mode_lib->vba.MaxDispclk[i]
+						&& i == mode_lib->vba.soc.num_states)
+					mode_lib->vba.PlaneRequiredDISPCLKWithoutODMCombine = mode_lib->vba.PixelClock[k]
+							* (1 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0);
+
+				mode_lib->vba.PlaneRequiredDISPCLKWithODMCombine = mode_lib->vba.PixelClock[k] / 2
+					* (1 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) * (1 + mode_lib->vba.DISPCLKRampingMargin / 100.0);
+				if (mode_lib->vba.PlaneRequiredDISPCLKWithODMCombine >= mode_lib->vba.MaxDispclk[i]
+						&& i == mode_lib->vba.soc.num_states)
+					mode_lib->vba.PlaneRequiredDISPCLKWithODMCombine = mode_lib->vba.PixelClock[k] / 2
+							* (1 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0);
+
+				locals->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_disabled;
+				mode_lib->vba.PlaneRequiredDISPCLK = mode_lib->vba.PlaneRequiredDISPCLKWithoutODMCombine;
+				if (mode_lib->vba.ODMCapability) {
+					if (locals->PlaneRequiredDISPCLKWithoutODMCombine > mode_lib->vba.MaxDispclkRoundedDownToDFSGranularity) {
+						locals->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_2to1;
+						mode_lib->vba.PlaneRequiredDISPCLK = mode_lib->vba.PlaneRequiredDISPCLKWithODMCombine;
+					} else if (locals->HActive[k] > DCN20_MAX_420_IMAGE_WIDTH && locals->OutputFormat[k] == dm_420) {
+						locals->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_2to1;
+						mode_lib->vba.PlaneRequiredDISPCLK = mode_lib->vba.PlaneRequiredDISPCLKWithODMCombine;
+					}
+				}
+
+				if (locals->MinDPPCLKUsingSingleDPP[k] * (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) <= mode_lib->vba.MaxDppclkRoundedDownToDFSGranularity
+						&& locals->SwathWidthYSingleDPP[k] <= locals->MaximumSwathWidth[k]
+						&& locals->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_disabled) {
+					locals->NoOfDPP[i][j][k] = 1;
+					locals->RequiredDPPCLK[i][j][k] =
+						locals->MinDPPCLKUsingSingleDPP[k] * (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0);
+				} else {
+					locals->NoOfDPP[i][j][k] = 2;
+					locals->RequiredDPPCLK[i][j][k] =
+						locals->MinDPPCLKUsingSingleDPP[k] * (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) / 2.0;
+				}
+				locals->RequiredDISPCLK[i][j] = dml_max(
+						locals->RequiredDISPCLK[i][j],
+						mode_lib->vba.PlaneRequiredDISPCLK);
+				if ((locals->MinDPPCLKUsingSingleDPP[k] / locals->NoOfDPP[i][j][k] * (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0)
+						> mode_lib->vba.MaxDppclkRoundedDownToDFSGranularity)
+						|| (mode_lib->vba.PlaneRequiredDISPCLK > mode_lib->vba.MaxDispclkRoundedDownToDFSGranularity)) {
+					locals->DISPCLK_DPPCLK_Support[i][j] = false;
+				}
+			}
+			locals->TotalNumberOfActiveDPP[i][j] = 0.0;
+			for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++)
+				locals->TotalNumberOfActiveDPP[i][j] = locals->TotalNumberOfActiveDPP[i][j] + locals->NoOfDPP[i][j][k];
+			if (j == 1) {
+				while (locals->TotalNumberOfActiveDPP[i][j] < mode_lib->vba.MaxNumDPP
+						&& locals->TotalNumberOfActiveDPP[i][j] < 2 * mode_lib->vba.NumberOfActivePlanes) {
+					double BWOfNonSplitPlaneOfMaximumBandwidth;
+					unsigned int NumberOfNonSplitPlaneOfMaximumBandwidth;
+
+					BWOfNonSplitPlaneOfMaximumBandwidth = 0;
+					NumberOfNonSplitPlaneOfMaximumBandwidth = 0;
+					for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+						if (locals->ReadBandwidth[k] > BWOfNonSplitPlaneOfMaximumBandwidth && locals->NoOfDPP[i][j][k] == 1) {
+							BWOfNonSplitPlaneOfMaximumBandwidth = locals->ReadBandwidth[k];
+							NumberOfNonSplitPlaneOfMaximumBandwidth = k;
+						}
+					}
+					locals->NoOfDPP[i][j][NumberOfNonSplitPlaneOfMaximumBandwidth] = 2;
+					locals->RequiredDPPCLK[i][j][NumberOfNonSplitPlaneOfMaximumBandwidth] =
+						locals->MinDPPCLKUsingSingleDPP[NumberOfNonSplitPlaneOfMaximumBandwidth]
+							* (1 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100) / 2;
+					locals->TotalNumberOfActiveDPP[i][j] = locals->TotalNumberOfActiveDPP[i][j] + 1;
+				}
+			}
+			if (locals->TotalNumberOfActiveDPP[i][j] > mode_lib->vba.MaxNumDPP) {
+				locals->RequiredDISPCLK[i][j] = 0.0;
+				locals->DISPCLK_DPPCLK_Support[i][j] = true;
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+					locals->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_disabled;
+					if (locals->SwathWidthYSingleDPP[k] <= locals->MaximumSwathWidth[k]) {
+						locals->NoOfDPP[i][j][k] = 1;
+						locals->RequiredDPPCLK[i][j][k] = locals->MinDPPCLKUsingSingleDPP[k]
+							* (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0);
+					} else {
+						locals->NoOfDPP[i][j][k] = 2;
+						locals->RequiredDPPCLK[i][j][k] = locals->MinDPPCLKUsingSingleDPP[k]
+										* (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) / 2.0;
+					}
+					if (i != mode_lib->vba.soc.num_states) {
+						mode_lib->vba.PlaneRequiredDISPCLK =
+								mode_lib->vba.PixelClock[k]
+										* (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0)
+										* (1.0 + mode_lib->vba.DISPCLKRampingMargin / 100.0);
+					} else {
+						mode_lib->vba.PlaneRequiredDISPCLK = mode_lib->vba.PixelClock[k]
+							* (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0);
+					}
+					locals->RequiredDISPCLK[i][j] = dml_max(
+							locals->RequiredDISPCLK[i][j],
+							mode_lib->vba.PlaneRequiredDISPCLK);
+					if (locals->MinDPPCLKUsingSingleDPP[k] / locals->NoOfDPP[i][j][k] * (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0)
+							> mode_lib->vba.MaxDppclkRoundedDownToDFSGranularity
+							|| mode_lib->vba.PlaneRequiredDISPCLK > mode_lib->vba.MaxDispclkRoundedDownToDFSGranularity)
+						locals->DISPCLK_DPPCLK_Support[i][j] = false;
+				}
+				locals->TotalNumberOfActiveDPP[i][j] = 0.0;
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++)
+					locals->TotalNumberOfActiveDPP[i][j] = locals->TotalNumberOfActiveDPP[i][j] + locals->NoOfDPP[i][j][k];
+			}
+			locals->RequiredDISPCLK[i][j] = dml_max(
+					locals->RequiredDISPCLK[i][j],
+					mode_lib->vba.WritebackRequiredDISPCLK);
+			if (mode_lib->vba.MaxDispclkRoundedDownToDFSGranularity
+					< mode_lib->vba.WritebackRequiredDISPCLK) {
+				locals->DISPCLK_DPPCLK_Support[i][j] = false;
+			}
+		}
+	}
+	/*Viewport Size Check*/
+
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		locals->ViewportSizeSupport[i][0] = true;
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			if (locals->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_2to1) {
+				if (dml_min(locals->SwathWidthYSingleDPP[k], dml_round(mode_lib->vba.HActive[k] / 2.0 * mode_lib->vba.HRatio[k]))
+						> locals->MaximumSwathWidth[k]) {
+					locals->ViewportSizeSupport[i][0] = false;
+				}
+			} else {
+				if (locals->SwathWidthYSingleDPP[k] / 2.0 > locals->MaximumSwathWidth[k]) {
+					locals->ViewportSizeSupport[i][0] = false;
+				}
+			}
+		}
+	}
+	/*Total Available Pipes Support Check*/
+
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		for (j = 0; j < 2; j++) {
+			if (locals->TotalNumberOfActiveDPP[i][j] <= mode_lib->vba.MaxNumDPP)
+				locals->TotalAvailablePipesSupport[i][j] = true;
+			else
+				locals->TotalAvailablePipesSupport[i][j] = false;
+		}
+	}
+	/*Total Available OTG Support Check*/
+
+	mode_lib->vba.TotalNumberOfActiveOTG = 0.0;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.BlendingAndTiming[k] == k) {
+			mode_lib->vba.TotalNumberOfActiveOTG = mode_lib->vba.TotalNumberOfActiveOTG
+					+ 1.0;
+		}
+	}
+	if (mode_lib->vba.TotalNumberOfActiveOTG <= mode_lib->vba.MaxNumOTG) {
+		mode_lib->vba.NumberOfOTGSupport = true;
+	} else {
+		mode_lib->vba.NumberOfOTGSupport = false;
+	}
+	/*Display IO and DSC Support Check*/
+
+	mode_lib->vba.NonsupportedDSCInputBPC = false;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (!(mode_lib->vba.DSCInputBitPerComponent[k] == 12.0
+				|| mode_lib->vba.DSCInputBitPerComponent[k] == 10.0
+				|| mode_lib->vba.DSCInputBitPerComponent[k] == 8.0)) {
+			mode_lib->vba.NonsupportedDSCInputBPC = true;
+		}
+	}
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			locals->RequiresDSC[i][k] = 0;
+			locals->RequiresFEC[i][k] = 0;
+			if (mode_lib->vba.BlendingAndTiming[k] == k) {
+				if (mode_lib->vba.Output[k] == dm_hdmi) {
+					locals->RequiresDSC[i][k] = 0;
+					locals->RequiresFEC[i][k] = 0;
+					locals->OutputBppPerState[i][k] = TruncToValidBPP(
+							dml_min(600.0, mode_lib->vba.PHYCLKPerState[i]) / mode_lib->vba.PixelClockBackEnd[k] * 24,
+							false,
+							mode_lib->vba.Output[k],
+							mode_lib->vba.OutputFormat[k],
+							mode_lib->vba.DSCInputBitPerComponent[k]);
+				} else if (mode_lib->vba.Output[k] == dm_dp
+						|| mode_lib->vba.Output[k] == dm_edp) {
+					if (mode_lib->vba.Output[k] == dm_edp) {
+						mode_lib->vba.EffectiveFECOverhead = 0.0;
+					} else {
+						mode_lib->vba.EffectiveFECOverhead =
+								mode_lib->vba.FECOverhead;
+					}
+					if (mode_lib->vba.PHYCLKPerState[i] >= 270.0) {
+						mode_lib->vba.Outbpp = TruncToValidBPP(
+								(1.0 - mode_lib->vba.Downspreading / 100.0) * 270.0
+								* mode_lib->vba.OutputLinkDPLanes[k] / mode_lib->vba.PixelClockBackEnd[k] * 8.0,
+								false,
+								mode_lib->vba.Output[k],
+								mode_lib->vba.OutputFormat[k],
+								mode_lib->vba.DSCInputBitPerComponent[k]);
+						mode_lib->vba.OutbppDSC = TruncToValidBPP(
+								(1.0 - mode_lib->vba.Downspreading / 100.0) * (1.0 - mode_lib->vba.EffectiveFECOverhead / 100.0) * 270.0
+								* mode_lib->vba.OutputLinkDPLanes[k] / mode_lib->vba.PixelClockBackEnd[k] * 8.0,
+								true,
+								mode_lib->vba.Output[k],
+								mode_lib->vba.OutputFormat[k],
+								mode_lib->vba.DSCInputBitPerComponent[k]);
+						if (mode_lib->vba.DSCEnabled[k] == true) {
+							locals->RequiresDSC[i][k] = true;
+							if (mode_lib->vba.Output[k] == dm_dp) {
+								locals->RequiresFEC[i][k] = true;
+							} else {
+								locals->RequiresFEC[i][k] = false;
+							}
+							mode_lib->vba.Outbpp = mode_lib->vba.OutbppDSC;
+						} else {
+							locals->RequiresDSC[i][k] = false;
+							locals->RequiresFEC[i][k] = false;
+						}
+						locals->OutputBppPerState[i][k] = mode_lib->vba.Outbpp;
+					}
+					if (mode_lib->vba.Outbpp == BPP_INVALID && mode_lib->vba.PHYCLKPerState[i] >= 540.0) {
+						mode_lib->vba.Outbpp = TruncToValidBPP(
+								(1.0 - mode_lib->vba.Downspreading / 100.0) * 540.0
+								* mode_lib->vba.OutputLinkDPLanes[k] / mode_lib->vba.PixelClockBackEnd[k] * 8.0,
+									false,
+									mode_lib->vba.Output[k],
+									mode_lib->vba.OutputFormat[k],
+									mode_lib->vba.DSCInputBitPerComponent[k]);
+						mode_lib->vba.OutbppDSC = TruncToValidBPP(
+								(1.0 - mode_lib->vba.Downspreading / 100.0) * (1.0 - mode_lib->vba.EffectiveFECOverhead / 100.0) * 540.0
+								* mode_lib->vba.OutputLinkDPLanes[k] / mode_lib->vba.PixelClockBackEnd[k] * 8.0,
+								true,
+								mode_lib->vba.Output[k],
+								mode_lib->vba.OutputFormat[k],
+								mode_lib->vba.DSCInputBitPerComponent[k]);
+						if (mode_lib->vba.DSCEnabled[k] == true) {
+							locals->RequiresDSC[i][k] = true;
+							if (mode_lib->vba.Output[k] == dm_dp) {
+								locals->RequiresFEC[i][k] = true;
+							} else {
+								locals->RequiresFEC[i][k] = false;
+							}
+							mode_lib->vba.Outbpp = mode_lib->vba.OutbppDSC;
+						} else {
+							locals->RequiresDSC[i][k] = false;
+							locals->RequiresFEC[i][k] = false;
+						}
+						locals->OutputBppPerState[i][k] = mode_lib->vba.Outbpp;
+					}
+					if (mode_lib->vba.Outbpp == BPP_INVALID
+							&& mode_lib->vba.PHYCLKPerState[i]
+									>= 810.0) {
+						mode_lib->vba.Outbpp = TruncToValidBPP(
+								(1.0 - mode_lib->vba.Downspreading / 100.0) * 810.0
+								* mode_lib->vba.OutputLinkDPLanes[k] / mode_lib->vba.PixelClockBackEnd[k] * 8.0,
+								false,
+								mode_lib->vba.Output[k],
+								mode_lib->vba.OutputFormat[k],
+								mode_lib->vba.DSCInputBitPerComponent[k]);
+						mode_lib->vba.OutbppDSC = TruncToValidBPP(
+								(1.0 - mode_lib->vba.Downspreading / 100.0) * (1.0 - mode_lib->vba.EffectiveFECOverhead / 100.0) * 810.0
+								* mode_lib->vba.OutputLinkDPLanes[k] / mode_lib->vba.PixelClockBackEnd[k] * 8.0,
+								true,
+								mode_lib->vba.Output[k],
+								mode_lib->vba.OutputFormat[k],
+								mode_lib->vba.DSCInputBitPerComponent[k]);
+						if (mode_lib->vba.DSCEnabled[k] == true || mode_lib->vba.Outbpp == BPP_INVALID) {
+							locals->RequiresDSC[i][k] = true;
+							if (mode_lib->vba.Output[k] == dm_dp) {
+								locals->RequiresFEC[i][k] = true;
+							} else {
+								locals->RequiresFEC[i][k] = false;
+							}
+							mode_lib->vba.Outbpp = mode_lib->vba.OutbppDSC;
+						} else {
+							locals->RequiresDSC[i][k] = false;
+							locals->RequiresFEC[i][k] = false;
+						}
+						locals->OutputBppPerState[i][k] =
+								mode_lib->vba.Outbpp;
+					}
+				}
+			} else {
+				locals->OutputBppPerState[i][k] = BPP_BLENDED_PIPE;
+			}
+		}
+	}
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		locals->DIOSupport[i] = true;
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			if (!mode_lib->vba.skip_dio_check[k]
+					&& (locals->OutputBppPerState[i][k] == BPP_INVALID
+						|| (mode_lib->vba.OutputFormat[k] == dm_420
+							&& mode_lib->vba.Interlace[k] == true
+							&& mode_lib->vba.ProgressiveToInterlaceUnitInOPP == true))) {
+				locals->DIOSupport[i] = false;
+			}
+		}
+	}
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			locals->DSCCLKRequiredMoreThanSupported[i] = false;
+			if (mode_lib->vba.BlendingAndTiming[k] == k) {
+				if ((mode_lib->vba.Output[k] == dm_dp
+						|| mode_lib->vba.Output[k] == dm_edp)) {
+					if (mode_lib->vba.OutputFormat[k] == dm_420
+							|| mode_lib->vba.OutputFormat[k]
+									== dm_n422) {
+						mode_lib->vba.DSCFormatFactor = 2;
+					} else {
+						mode_lib->vba.DSCFormatFactor = 1;
+					}
+					if (locals->RequiresDSC[i][k] == true) {
+						if (locals->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_2to1) {
+							if (mode_lib->vba.PixelClockBackEnd[k] / 6.0 / mode_lib->vba.DSCFormatFactor
+									> (1.0 - mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) * mode_lib->vba.MaxDSCCLK[i]) {
+								locals->DSCCLKRequiredMoreThanSupported[i] =
+										true;
+							}
+						} else {
+							if (mode_lib->vba.PixelClockBackEnd[k] / 3.0 / mode_lib->vba.DSCFormatFactor
+									> (1.0 - mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) * mode_lib->vba.MaxDSCCLK[i]) {
+								locals->DSCCLKRequiredMoreThanSupported[i] =
+										true;
+							}
+						}
+					}
+				}
+			}
+		}
+	}
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		locals->NotEnoughDSCUnits[i] = false;
+		mode_lib->vba.TotalDSCUnitsRequired = 0.0;
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			if (locals->RequiresDSC[i][k] == true) {
+				if (locals->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_2to1) {
+					mode_lib->vba.TotalDSCUnitsRequired =
+							mode_lib->vba.TotalDSCUnitsRequired + 2.0;
+				} else {
+					mode_lib->vba.TotalDSCUnitsRequired =
+							mode_lib->vba.TotalDSCUnitsRequired + 1.0;
+				}
+			}
+		}
+		if (mode_lib->vba.TotalDSCUnitsRequired > mode_lib->vba.NumberOfDSC) {
+			locals->NotEnoughDSCUnits[i] = true;
+		}
+	}
+	/*DSC Delay per state*/
+
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			if (mode_lib->vba.BlendingAndTiming[k] != k) {
+				mode_lib->vba.slices = 0;
+			} else if (locals->RequiresDSC[i][k] == 0
+					|| locals->RequiresDSC[i][k] == false) {
+				mode_lib->vba.slices = 0;
+			} else if (mode_lib->vba.PixelClockBackEnd[k] > 3200.0) {
+				mode_lib->vba.slices = dml_ceil(
+						mode_lib->vba.PixelClockBackEnd[k] / 400.0,
+						4.0);
+			} else if (mode_lib->vba.PixelClockBackEnd[k] > 1360.0) {
+				mode_lib->vba.slices = 8.0;
+			} else if (mode_lib->vba.PixelClockBackEnd[k] > 680.0) {
+				mode_lib->vba.slices = 4.0;
+			} else if (mode_lib->vba.PixelClockBackEnd[k] > 340.0) {
+				mode_lib->vba.slices = 2.0;
+			} else {
+				mode_lib->vba.slices = 1.0;
+			}
+			if (locals->OutputBppPerState[i][k] == BPP_BLENDED_PIPE
+					|| locals->OutputBppPerState[i][k] == BPP_INVALID) {
+				mode_lib->vba.bpp = 0.0;
+			} else {
+				mode_lib->vba.bpp = locals->OutputBppPerState[i][k];
+			}
+			if (locals->RequiresDSC[i][k] == true && mode_lib->vba.bpp != 0.0) {
+				if (locals->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_disabled) {
+					locals->DSCDelayPerState[i][k] =
+							dscceComputeDelay(
+									mode_lib->vba.DSCInputBitPerComponent[k],
+									mode_lib->vba.bpp,
+									dml_ceil(
+											mode_lib->vba.HActive[k]
+													/ mode_lib->vba.slices,
+											1.0),
+									mode_lib->vba.slices,
+									mode_lib->vba.OutputFormat[k])
+									+ dscComputeDelay(
+											mode_lib->vba.OutputFormat[k]);
+				} else {
+					locals->DSCDelayPerState[i][k] =
+							2.0 * (dscceComputeDelay(
+											mode_lib->vba.DSCInputBitPerComponent[k],
+											mode_lib->vba.bpp,
+											dml_ceil(mode_lib->vba.HActive[k] / mode_lib->vba.slices, 1.0),
+											mode_lib->vba.slices / 2,
+											mode_lib->vba.OutputFormat[k])
+									+ dscComputeDelay(mode_lib->vba.OutputFormat[k]));
+				}
+				locals->DSCDelayPerState[i][k] =
+						locals->DSCDelayPerState[i][k] * mode_lib->vba.PixelClock[k] / mode_lib->vba.PixelClockBackEnd[k];
+			} else {
+				locals->DSCDelayPerState[i][k] = 0.0;
+			}
+		}
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			for (m = 0; m <= mode_lib->vba.NumberOfActivePlanes - 1; m++) {
+				for (j = 0; j <= mode_lib->vba.NumberOfActivePlanes - 1; j++) {
+					if (mode_lib->vba.BlendingAndTiming[k] == m && locals->RequiresDSC[i][m] == true)
+						locals->DSCDelayPerState[i][k] = locals->DSCDelayPerState[i][m];
+				}
+			}
+		}
+	}
+
+	//Prefetch Check
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		for (j = 0; j < 2; j++) {
+			for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+				if (locals->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_2to1)
+					locals->SwathWidthYPerState[i][j][k] = dml_min(locals->SwathWidthYSingleDPP[k], dml_round(locals->HActive[k] / 2 * locals->HRatio[k]));
+				else
+					locals->SwathWidthYPerState[i][j][k] = locals->SwathWidthYSingleDPP[k] / locals->NoOfDPP[i][j][k];
+				locals->SwathWidthGranularityY = 256  / dml_ceil(locals->BytePerPixelInDETY[k], 1) / locals->MaxSwathHeightY[k];
+				locals->RoundedUpMaxSwathSizeBytesY = (dml_ceil(locals->SwathWidthYPerState[i][j][k] - 1, locals->SwathWidthGranularityY)
+						+ locals->SwathWidthGranularityY) * locals->BytePerPixelInDETY[k] * locals->MaxSwathHeightY[k];
+				if (locals->SourcePixelFormat[k] == dm_420_10) {
+					locals->RoundedUpMaxSwathSizeBytesY = dml_ceil(locals->RoundedUpMaxSwathSizeBytesY, 256) + 256;
+				}
+				if (locals->MaxSwathHeightC[k] > 0) {
+					locals->SwathWidthGranularityC = 256 / dml_ceil(locals->BytePerPixelInDETC[k], 2) / locals->MaxSwathHeightC[k];
+
+					locals->RoundedUpMaxSwathSizeBytesC = (dml_ceil(locals->SwathWidthYPerState[i][j][k] / 2 - 1, locals->SwathWidthGranularityC)
+					+ locals->SwathWidthGranularityC) * locals->BytePerPixelInDETC[k] * locals->MaxSwathHeightC[k];
+				}
+				if (locals->SourcePixelFormat[k] == dm_420_10) {
+					locals->RoundedUpMaxSwathSizeBytesC = dml_ceil(locals->RoundedUpMaxSwathSizeBytesC, 256)  + 256;
+				} else {
+					locals->RoundedUpMaxSwathSizeBytesC = 0;
+				}
+
+				if (locals->RoundedUpMaxSwathSizeBytesY + locals->RoundedUpMaxSwathSizeBytesC <= locals->DETBufferSizeInKByte[0] * 1024.0 / 2) {
+					locals->SwathHeightYPerState[i][j][k] = locals->MaxSwathHeightY[k];
+					locals->SwathHeightCPerState[i][j][k] = locals->MaxSwathHeightC[k];
+				} else {
+					locals->SwathHeightYPerState[i][j][k] = locals->MinSwathHeightY[k];
+					locals->SwathHeightCPerState[i][j][k] = locals->MinSwathHeightC[k];
+				}
+
+				if (locals->BytePerPixelInDETC[k] == 0) {
+					locals->LinesInDETLuma = locals->DETBufferSizeInKByte[0] * 1024 / locals->BytePerPixelInDETY[k] / locals->SwathWidthYPerState[i][j][k];
+					locals->LinesInDETChroma = 0;
+				} else if (locals->SwathHeightYPerState[i][j][k] <= locals->SwathHeightCPerState[i][j][k]) {
+					locals->LinesInDETLuma = locals->DETBufferSizeInKByte[0] * 1024 / 2 / locals->BytePerPixelInDETY[k] /
+							locals->SwathWidthYPerState[i][j][k];
+					locals->LinesInDETChroma = locals->DETBufferSizeInKByte[0] * 1024 / 2 / locals->BytePerPixelInDETC[k] / (locals->SwathWidthYPerState[i][j][k] / 2);
+				} else {
+					locals->LinesInDETLuma = locals->DETBufferSizeInKByte[0] * 1024 * 2 / 3 / locals->BytePerPixelInDETY[k] / locals->SwathWidthYPerState[i][j][k];
+					locals->LinesInDETChroma = locals->DETBufferSizeInKByte[0] * 1024 / 3 / locals->BytePerPixelInDETY[k] / (locals->SwathWidthYPerState[i][j][k] / 2);
+				}
+
+				locals->EffectiveLBLatencyHidingSourceLinesLuma = dml_min(locals->MaxLineBufferLines,
+					dml_floor(locals->LineBufferSize / locals->LBBitPerPixel[k] / (locals->SwathWidthYPerState[i][j][k]
+					/ dml_max(locals->HRatio[k], 1)), 1)) - (locals->vtaps[k] - 1);
+
+				locals->EffectiveLBLatencyHidingSourceLinesChroma =  dml_min(locals->MaxLineBufferLines,
+						dml_floor(locals->LineBufferSize / locals->LBBitPerPixel[k]
+						/ (locals->SwathWidthYPerState[i][j][k] / 2
+						/ dml_max(locals->HRatio[k] / 2, 1)), 1)) - (locals->VTAPsChroma[k] - 1);
+
+				locals->EffectiveDETLBLinesLuma = dml_floor(locals->LinesInDETLuma +  dml_min(
+						locals->LinesInDETLuma * locals->RequiredDISPCLK[i][j] * locals->BytePerPixelInDETY[k] *
+						locals->PSCL_FACTOR[k] / locals->ReturnBWPerState[i][0],
+						locals->EffectiveLBLatencyHidingSourceLinesLuma),
+						locals->SwathHeightYPerState[i][j][k]);
+				if (locals->LinesInDETChroma) {
+					locals->EffectiveDETLBLinesChroma = dml_floor(locals->LinesInDETChroma +
+						    dml_min(locals->LinesInDETChroma * locals->RequiredDISPCLK[i][j] *
+						    locals->BytePerPixelInDETC[k] *
+							locals->PSCL_FACTOR_CHROMA[k] / locals->ReturnBWPerState[i][0],
+							locals->EffectiveLBLatencyHidingSourceLinesChroma),
+							locals->SwathHeightCPerState[i][j][k]);
+				} else {
+					locals->EffectiveDETLBLinesChroma = 0;
+				}
+
+				if (locals->BytePerPixelInDETC[k] == 0) {
+					locals->UrgentLatencySupportUsPerState[i][j][k] = locals->EffectiveDETLBLinesLuma * (locals->HTotal[k] / locals->PixelClock[k])
+							/ locals->VRatio[k] - locals->EffectiveDETLBLinesLuma * locals->SwathWidthYPerState[i][j][k] *
+								dml_ceil(locals->BytePerPixelInDETY[k], 1) / (locals->ReturnBWPerState[i][0] / locals->NoOfDPP[i][j][k]);
+				} else {
+					locals->UrgentLatencySupportUsPerState[i][j][k] = dml_min(
+						locals->EffectiveDETLBLinesLuma * (locals->HTotal[k] / locals->PixelClock[k])
+						/ locals->VRatio[k] - locals->EffectiveDETLBLinesLuma * locals->SwathWidthYPerState[i][j][k] *
+						dml_ceil(locals->BytePerPixelInDETY[k], 1) / (locals->ReturnBWPerState[i][0] / locals->NoOfDPP[i][j][k]),
+							locals->EffectiveDETLBLinesChroma * (locals->HTotal[k] / locals->PixelClock[k]) / (locals->VRatio[k] / 2) -
+							locals->EffectiveDETLBLinesChroma * locals->SwathWidthYPerState[i][j][k] / 2 *
+							dml_ceil(locals->BytePerPixelInDETC[k], 2) / (locals->ReturnBWPerState[i][0] / locals->NoOfDPP[i][j][k]));
+				}
+			}
+		}
+	}
+
+	for (i = 0; i <= locals->soc.num_states; i++) {
+		for (j = 0; j < 2; j++) {
+			locals->UrgentLatencySupport[i][j] = true;
+			for (k = 0; k < locals->NumberOfActivePlanes; k++) {
+				if (locals->UrgentLatencySupportUsPerState[i][j][k] < locals->UrgentLatency)
+					locals->UrgentLatencySupport[i][j] = false;
+			}
+		}
+	}
+
+
+	/*Prefetch Check*/
+	for (i = 0; i <= locals->soc.num_states; i++) {
+		for (j = 0; j < 2; j++) {
+			locals->TotalNumberOfDCCActiveDPP[i][j] = 0;
+			for (k = 0; k < locals->NumberOfActivePlanes; k++) {
+				if (locals->DCCEnable[k] == true) {
+					locals->TotalNumberOfDCCActiveDPP[i][j] =
+						locals->TotalNumberOfDCCActiveDPP[i][j] + locals->NoOfDPP[i][j][k];
+				}
+			}
+		}
+	}
+
+	CalculateMinAndMaxPrefetchMode(locals->AllowDRAMSelfRefreshOrDRAMClockChangeInVblank, &locals->MinPrefetchMode, &locals->MaxPrefetchMode);
+
+	for (i = 0; i <= locals->soc.num_states; i++) {
+		for (j = 0; j < 2; j++) {
+			for (k = 0; k < locals->NumberOfActivePlanes; k++) {
+				locals->NoOfDPPThisState[k] = locals->NoOfDPP[i][j][k];
+				locals->RequiredDPPCLKThisState[k] = locals->RequiredDPPCLK[i][j][k];
+				locals->SwathHeightYThisState[k] = locals->SwathHeightYPerState[i][j][k];
+				locals->SwathHeightCThisState[k] = locals->SwathHeightCPerState[i][j][k];
+				locals->SwathWidthYThisState[k] = locals->SwathWidthYPerState[i][j][k];
+				mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0] = dml_max(
+						mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0],
+						mode_lib->vba.PixelClock[k] / 16.0);
+				if (mode_lib->vba.BytePerPixelInDETC[k] == 0.0) {
+					if (mode_lib->vba.VRatio[k] <= 1.0) {
+						mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0] =
+								dml_max(
+										mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0],
+										1.1
+												* dml_ceil(
+														mode_lib->vba.BytePerPixelInDETY[k],
+														1.0)
+												/ 64.0
+												* mode_lib->vba.HRatio[k]
+												* mode_lib->vba.PixelClock[k]
+												/ mode_lib->vba.NoOfDPP[i][j][k]);
+					} else {
+						mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0] =
+								dml_max(
+										mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0],
+										1.1
+												* dml_ceil(
+														mode_lib->vba.BytePerPixelInDETY[k],
+														1.0)
+												/ 64.0
+												* mode_lib->vba.PSCL_FACTOR[k]
+												* mode_lib->vba.RequiredDPPCLK[i][j][k]);
+					}
+				} else {
+					if (mode_lib->vba.VRatio[k] <= 1.0) {
+						mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0] =
+								dml_max(
+										mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0],
+										1.1
+												* dml_ceil(
+														mode_lib->vba.BytePerPixelInDETY[k],
+														1.0)
+												/ 32.0
+												* mode_lib->vba.HRatio[k]
+												* mode_lib->vba.PixelClock[k]
+												/ mode_lib->vba.NoOfDPP[i][j][k]);
+					} else {
+						mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0] =
+								dml_max(
+										mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0],
+										1.1
+												* dml_ceil(
+														mode_lib->vba.BytePerPixelInDETY[k],
+														1.0)
+												/ 32.0
+												* mode_lib->vba.PSCL_FACTOR[k]
+												* mode_lib->vba.RequiredDPPCLK[i][j][k]);
+					}
+					if (mode_lib->vba.VRatio[k] / 2.0 <= 1.0) {
+						mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0] =
+								dml_max(
+										mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0],
+										1.1
+												* dml_ceil(
+														mode_lib->vba.BytePerPixelInDETC[k],
+														2.0)
+												/ 32.0
+												* mode_lib->vba.HRatio[k]
+												/ 2.0
+												* mode_lib->vba.PixelClock[k]
+												/ mode_lib->vba.NoOfDPP[i][j][k]);
+					} else {
+						mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0] =
+								dml_max(
+										mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0],
+										1.1
+												* dml_ceil(
+														mode_lib->vba.BytePerPixelInDETC[k],
+														2.0)
+												/ 32.0
+												* mode_lib->vba.PSCL_FACTOR_CHROMA[k]
+												* mode_lib->vba.RequiredDPPCLK[i][j][k]);
+					}
+				}
+			}
+			for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+				mode_lib->vba.PDEAndMetaPTEBytesPerFrameY = CalculateVMAndRowBytes(
+						mode_lib,
+						mode_lib->vba.DCCEnable[k],
+						mode_lib->vba.Read256BlockHeightY[k],
+						mode_lib->vba.Read256BlockWidthY[k],
+						mode_lib->vba.SourcePixelFormat[k],
+						mode_lib->vba.SurfaceTiling[k],
+						dml_ceil(mode_lib->vba.BytePerPixelInDETY[k], 1.0),
+						mode_lib->vba.SourceScan[k],
+						mode_lib->vba.ViewportWidth[k],
+						mode_lib->vba.ViewportHeight[k],
+						mode_lib->vba.SwathWidthYPerState[i][j][k],
+						mode_lib->vba.GPUVMEnable,
+						mode_lib->vba.VMMPageSize,
+						mode_lib->vba.PTEBufferSizeInRequestsLuma,
+						mode_lib->vba.PDEProcessingBufIn64KBReqs,
+						mode_lib->vba.PitchY[k],
+						mode_lib->vba.DCCMetaPitchY[k],
+						&mode_lib->vba.MacroTileWidthY[k],
+						&mode_lib->vba.MetaRowBytesY,
+						&mode_lib->vba.DPTEBytesPerRowY,
+						&mode_lib->vba.PTEBufferSizeNotExceededY[i][j][k],
+						&mode_lib->vba.dpte_row_height[k],
+						&mode_lib->vba.meta_row_height[k]);
+				mode_lib->vba.PrefetchLinesY[0][0][k] = CalculatePrefetchSourceLines(
+						mode_lib,
+						mode_lib->vba.VRatio[k],
+						mode_lib->vba.vtaps[k],
+						mode_lib->vba.Interlace[k],
+						mode_lib->vba.ProgressiveToInterlaceUnitInOPP,
+						mode_lib->vba.SwathHeightYPerState[i][j][k],
+						mode_lib->vba.ViewportYStartY[k],
+						&mode_lib->vba.PrefillY[k],
+						&mode_lib->vba.MaxNumSwY[k]);
+				if ((mode_lib->vba.SourcePixelFormat[k] != dm_444_64
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_444_32
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_444_16
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_mono_16
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_mono_8)) {
+					mode_lib->vba.PDEAndMetaPTEBytesPerFrameC = CalculateVMAndRowBytes(
+							mode_lib,
+							mode_lib->vba.DCCEnable[k],
+							mode_lib->vba.Read256BlockHeightY[k],
+							mode_lib->vba.Read256BlockWidthY[k],
+							mode_lib->vba.SourcePixelFormat[k],
+							mode_lib->vba.SurfaceTiling[k],
+							dml_ceil(mode_lib->vba.BytePerPixelInDETC[k], 2.0),
+							mode_lib->vba.SourceScan[k],
+							mode_lib->vba.ViewportWidth[k] / 2.0,
+							mode_lib->vba.ViewportHeight[k] / 2.0,
+							mode_lib->vba.SwathWidthYPerState[i][j][k] / 2.0,
+							mode_lib->vba.GPUVMEnable,
+							mode_lib->vba.VMMPageSize,
+							mode_lib->vba.PTEBufferSizeInRequestsLuma,
+							mode_lib->vba.PDEProcessingBufIn64KBReqs,
+							mode_lib->vba.PitchC[k],
+							0.0,
+							&mode_lib->vba.MacroTileWidthC[k],
+							&mode_lib->vba.MetaRowBytesC,
+							&mode_lib->vba.DPTEBytesPerRowC,
+							&mode_lib->vba.PTEBufferSizeNotExceededC[i][j][k],
+							&mode_lib->vba.dpte_row_height_chroma[k],
+							&mode_lib->vba.meta_row_height_chroma[k]);
+					mode_lib->vba.PrefetchLinesC[0][0][k] = CalculatePrefetchSourceLines(
+							mode_lib,
+							mode_lib->vba.VRatio[k] / 2.0,
+							mode_lib->vba.VTAPsChroma[k],
+							mode_lib->vba.Interlace[k],
+							mode_lib->vba.ProgressiveToInterlaceUnitInOPP,
+							mode_lib->vba.SwathHeightCPerState[i][j][k],
+							mode_lib->vba.ViewportYStartC[k],
+							&mode_lib->vba.PrefillC[k],
+							&mode_lib->vba.MaxNumSwC[k]);
+				} else {
+					mode_lib->vba.PDEAndMetaPTEBytesPerFrameC = 0.0;
+					mode_lib->vba.MetaRowBytesC = 0.0;
+					mode_lib->vba.DPTEBytesPerRowC = 0.0;
+					locals->PrefetchLinesC[0][0][k] = 0.0;
+					locals->PTEBufferSizeNotExceededC[i][j][k] = true;
+					locals->PTEBufferSizeInRequestsForLuma = mode_lib->vba.PTEBufferSizeInRequestsLuma + mode_lib->vba.PTEBufferSizeInRequestsChroma;
+				}
+				locals->PDEAndMetaPTEBytesPerFrame[0][0][k] =
+						mode_lib->vba.PDEAndMetaPTEBytesPerFrameY + mode_lib->vba.PDEAndMetaPTEBytesPerFrameC;
+				locals->MetaRowBytes[0][0][k] = mode_lib->vba.MetaRowBytesY + mode_lib->vba.MetaRowBytesC;
+				locals->DPTEBytesPerRow[0][0][k] = mode_lib->vba.DPTEBytesPerRowY + mode_lib->vba.DPTEBytesPerRowC;
+
+				CalculateActiveRowBandwidth(
+						mode_lib->vba.GPUVMEnable,
+						mode_lib->vba.SourcePixelFormat[k],
+						mode_lib->vba.VRatio[k],
+						mode_lib->vba.DCCEnable[k],
+						mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k],
+						mode_lib->vba.MetaRowBytesY,
+						mode_lib->vba.MetaRowBytesC,
+						mode_lib->vba.meta_row_height[k],
+						mode_lib->vba.meta_row_height_chroma[k],
+						mode_lib->vba.DPTEBytesPerRowY,
+						mode_lib->vba.DPTEBytesPerRowC,
+						mode_lib->vba.dpte_row_height[k],
+						mode_lib->vba.dpte_row_height_chroma[k],
+						&mode_lib->vba.meta_row_bw[k],
+						&mode_lib->vba.dpte_row_bw[k],
+						&mode_lib->vba.qual_row_bw[k]);
+			}
+			mode_lib->vba.ExtraLatency =
+					mode_lib->vba.UrgentRoundTripAndOutOfOrderLatencyPerState[i]
+							+ (mode_lib->vba.TotalNumberOfActiveDPP[i][j]
+									* mode_lib->vba.PixelChunkSizeInKByte
+									+ mode_lib->vba.TotalNumberOfDCCActiveDPP[i][j]
+											* mode_lib->vba.MetaChunkSize)
+									* 1024.0
+									/ mode_lib->vba.ReturnBWPerState[i][0];
+			if (mode_lib->vba.GPUVMEnable == true) {
+				mode_lib->vba.ExtraLatency = mode_lib->vba.ExtraLatency
+						+ mode_lib->vba.TotalNumberOfActiveDPP[i][j]
+								* mode_lib->vba.PTEGroupSize
+								/ mode_lib->vba.ReturnBWPerState[i][0];
+			}
+			mode_lib->vba.TimeCalc = 24.0 / mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0];
+
+			for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+				if (mode_lib->vba.BlendingAndTiming[k] == k) {
+					if (mode_lib->vba.WritebackEnable[k] == true) {
+						locals->WritebackDelay[i][k] = mode_lib->vba.WritebackLatency
+								+ CalculateWriteBackDelay(
+										mode_lib->vba.WritebackPixelFormat[k],
+										mode_lib->vba.WritebackHRatio[k],
+										mode_lib->vba.WritebackVRatio[k],
+										mode_lib->vba.WritebackLumaHTaps[k],
+										mode_lib->vba.WritebackLumaVTaps[k],
+										mode_lib->vba.WritebackChromaHTaps[k],
+										mode_lib->vba.WritebackChromaVTaps[k],
+										mode_lib->vba.WritebackDestinationWidth[k]) / locals->RequiredDISPCLK[i][j];
+					} else {
+						locals->WritebackDelay[i][k] = 0.0;
+					}
+					for (m = 0; m <= mode_lib->vba.NumberOfActivePlanes - 1; m++) {
+						if (mode_lib->vba.BlendingAndTiming[m] == k
+								&& mode_lib->vba.WritebackEnable[m]
+										== true) {
+							locals->WritebackDelay[i][k] = dml_max(locals->WritebackDelay[i][k],
+											mode_lib->vba.WritebackLatency + CalculateWriteBackDelay(
+													mode_lib->vba.WritebackPixelFormat[m],
+													mode_lib->vba.WritebackHRatio[m],
+													mode_lib->vba.WritebackVRatio[m],
+													mode_lib->vba.WritebackLumaHTaps[m],
+													mode_lib->vba.WritebackLumaVTaps[m],
+													mode_lib->vba.WritebackChromaHTaps[m],
+													mode_lib->vba.WritebackChromaVTaps[m],
+													mode_lib->vba.WritebackDestinationWidth[m]) / locals->RequiredDISPCLK[i][j]);
+						}
+					}
+				}
+			}
+			for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+				for (m = 0; m <= mode_lib->vba.NumberOfActivePlanes - 1; m++) {
+					if (mode_lib->vba.BlendingAndTiming[k] == m) {
+						locals->WritebackDelay[i][k] = locals->WritebackDelay[i][m];
+					}
+				}
+			}
+			for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+				for (m = 0; m < locals->NumberOfCursors[k]; m++)
+					locals->cursor_bw[k] = locals->NumberOfCursors[k] * locals->CursorWidth[k][m] * locals->CursorBPP[k][m]
+						/ 8 / (locals->HTotal[k] / locals->PixelClock[k]) * locals->VRatio[k];
+			}
+
+			for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+				locals->MaximumVStartup[0][0][k] = mode_lib->vba.VTotal[k] - mode_lib->vba.VActive[k]
+					- dml_max(1.0, dml_ceil(locals->WritebackDelay[i][k] / (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k]), 1.0));
+			}
+
+			mode_lib->vba.NextPrefetchMode = mode_lib->vba.MinPrefetchMode;
+			do {
+				mode_lib->vba.PrefetchMode[i][j] = mode_lib->vba.NextPrefetchMode;
+				mode_lib->vba.NextPrefetchMode = mode_lib->vba.NextPrefetchMode + 1;
+
+				mode_lib->vba.TWait = CalculateTWait(
+						mode_lib->vba.PrefetchMode[i][j],
+						mode_lib->vba.DRAMClockChangeLatency,
+						mode_lib->vba.UrgentLatency,
+						mode_lib->vba.SREnterPlusExitTime);
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+
+					if (mode_lib->vba.XFCEnabled[k] == true) {
+						mode_lib->vba.XFCRemoteSurfaceFlipDelay =
+								CalculateRemoteSurfaceFlipDelay(
+										mode_lib,
+										mode_lib->vba.VRatio[k],
+										locals->SwathWidthYPerState[i][j][k],
+										dml_ceil(locals->BytePerPixelInDETY[k], 1.0),
+										mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k],
+										mode_lib->vba.XFCTSlvVupdateOffset,
+										mode_lib->vba.XFCTSlvVupdateWidth,
+										mode_lib->vba.XFCTSlvVreadyOffset,
+										mode_lib->vba.XFCXBUFLatencyTolerance,
+										mode_lib->vba.XFCFillBWOverhead,
+										mode_lib->vba.XFCSlvChunkSize,
+										mode_lib->vba.XFCBusTransportTime,
+										mode_lib->vba.TimeCalc,
+										mode_lib->vba.TWait,
+										&mode_lib->vba.SrcActiveDrainRate,
+										&mode_lib->vba.TInitXFill,
+										&mode_lib->vba.TslvChk);
+					} else {
+						mode_lib->vba.XFCRemoteSurfaceFlipDelay = 0.0;
+					}
+					mode_lib->vba.IsErrorResult[i][j][k] =
+							CalculatePrefetchSchedule(
+									mode_lib,
+									mode_lib->vba.RequiredDPPCLK[i][j][k],
+									mode_lib->vba.RequiredDISPCLK[i][j],
+									mode_lib->vba.PixelClock[k],
+									mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0],
+									mode_lib->vba.DSCDelayPerState[i][k],
+									mode_lib->vba.NoOfDPP[i][j][k],
+									mode_lib->vba.ScalerEnabled[k],
+									mode_lib->vba.NumberOfCursors[k],
+									mode_lib->vba.DPPCLKDelaySubtotal,
+									mode_lib->vba.DPPCLKDelaySCL,
+									mode_lib->vba.DPPCLKDelaySCLLBOnly,
+									mode_lib->vba.DPPCLKDelayCNVCFormater,
+									mode_lib->vba.DPPCLKDelayCNVCCursor,
+									mode_lib->vba.DISPCLKDelaySubtotal,
+									mode_lib->vba.SwathWidthYPerState[i][j][k]
+											/ mode_lib->vba.HRatio[k],
+									mode_lib->vba.OutputFormat[k],
+									mode_lib->vba.VTotal[k]
+											- mode_lib->vba.VActive[k],
+									mode_lib->vba.HTotal[k],
+									mode_lib->vba.MaxInterDCNTileRepeaters,
+									mode_lib->vba.MaximumVStartup[0][0][k],
+									mode_lib->vba.GPUVMMaxPageTableLevels,
+									mode_lib->vba.GPUVMEnable,
+									mode_lib->vba.DynamicMetadataEnable[k],
+									mode_lib->vba.DynamicMetadataLinesBeforeActiveRequired[k],
+									mode_lib->vba.DynamicMetadataTransmittedBytes[k],
+									mode_lib->vba.DCCEnable[k],
+									mode_lib->vba.UrgentLatencyPixelDataOnly,
+									mode_lib->vba.ExtraLatency,
+									mode_lib->vba.TimeCalc,
+									mode_lib->vba.PDEAndMetaPTEBytesPerFrame[0][0][k],
+									mode_lib->vba.MetaRowBytes[0][0][k],
+									mode_lib->vba.DPTEBytesPerRow[0][0][k],
+									mode_lib->vba.PrefetchLinesY[0][0][k],
+									mode_lib->vba.SwathWidthYPerState[i][j][k],
+									mode_lib->vba.BytePerPixelInDETY[k],
+									mode_lib->vba.PrefillY[k],
+									mode_lib->vba.MaxNumSwY[k],
+									mode_lib->vba.PrefetchLinesC[0][0][k],
+									mode_lib->vba.BytePerPixelInDETC[k],
+									mode_lib->vba.PrefillC[k],
+									mode_lib->vba.MaxNumSwC[k],
+									mode_lib->vba.SwathHeightYPerState[i][j][k],
+									mode_lib->vba.SwathHeightCPerState[i][j][k],
+									mode_lib->vba.TWait,
+									mode_lib->vba.XFCEnabled[k],
+									mode_lib->vba.XFCRemoteSurfaceFlipDelay,
+									mode_lib->vba.Interlace[k],
+									mode_lib->vba.ProgressiveToInterlaceUnitInOPP,
+									mode_lib->vba.DSTXAfterScaler,
+									mode_lib->vba.DSTYAfterScaler,
+									&mode_lib->vba.LineTimesForPrefetch[k],
+									&mode_lib->vba.PrefetchBW[k],
+									&mode_lib->vba.LinesForMetaPTE[k],
+									&mode_lib->vba.LinesForMetaAndDPTERow[k],
+									&mode_lib->vba.VRatioPreY[i][j][k],
+									&mode_lib->vba.VRatioPreC[i][j][k],
+									&mode_lib->vba.RequiredPrefetchPixelDataBWLuma[i][j][k],
+									&mode_lib->vba.VStartupRequiredWhenNotEnoughTimeForDynamicMetadata,
+									&mode_lib->vba.Tno_bw[k],
+									&mode_lib->vba.VUpdateOffsetPix[k],
+									&mode_lib->vba.VUpdateWidthPix[k],
+									&mode_lib->vba.VReadyOffsetPix[k]);
+				}
+				mode_lib->vba.MaximumReadBandwidthWithoutPrefetch = 0.0;
+				mode_lib->vba.MaximumReadBandwidthWithPrefetch = 0.0;
+				locals->prefetch_vm_bw_valid = true;
+				locals->prefetch_row_bw_valid = true;
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+					if (locals->PDEAndMetaPTEBytesPerFrame[0][0][k] == 0)
+						locals->prefetch_vm_bw[k] = 0;
+					else if (locals->LinesForMetaPTE[k] > 0)
+						locals->prefetch_vm_bw[k] = locals->PDEAndMetaPTEBytesPerFrame[0][0][k]
+							/ (locals->LinesForMetaPTE[k] * locals->HTotal[k] / locals->PixelClock[k]);
+					else {
+						locals->prefetch_vm_bw[k] = 0;
+						locals->prefetch_vm_bw_valid = false;
+					}
+					if (locals->MetaRowBytes[0][0][k] + locals->DPTEBytesPerRow[0][0][k] == 0)
+						locals->prefetch_row_bw[k] = 0;
+					else if (locals->LinesForMetaAndDPTERow[k] > 0)
+						locals->prefetch_row_bw[k] = (locals->MetaRowBytes[0][0][k] + locals->DPTEBytesPerRow[0][0][k])
+							/ (locals->LinesForMetaAndDPTERow[k] * locals->HTotal[k] / locals->PixelClock[k]);
+					else {
+						locals->prefetch_row_bw[k] = 0;
+						locals->prefetch_row_bw_valid = false;
+					}
+
+					mode_lib->vba.MaximumReadBandwidthWithoutPrefetch = mode_lib->vba.MaximumReadBandwidthWithPrefetch
+							+ mode_lib->vba.cursor_bw[k] + mode_lib->vba.ReadBandwidth[k] + mode_lib->vba.meta_row_bw[k] + mode_lib->vba.dpte_row_bw[k];
+					mode_lib->vba.MaximumReadBandwidthWithPrefetch =
+							mode_lib->vba.MaximumReadBandwidthWithPrefetch
+									+ mode_lib->vba.cursor_bw[k]
+									+ dml_max3(
+											mode_lib->vba.prefetch_vm_bw[k],
+											mode_lib->vba.prefetch_row_bw[k],
+											dml_max(mode_lib->vba.ReadBandwidth[k],
+											mode_lib->vba.RequiredPrefetchPixelDataBWLuma[i][j][k])
+											+ mode_lib->vba.meta_row_bw[k] + mode_lib->vba.dpte_row_bw[k]);
+				}
+				locals->BandwidthWithoutPrefetchSupported[i][0] = true;
+				if (mode_lib->vba.MaximumReadBandwidthWithoutPrefetch > locals->ReturnBWPerState[i][0]) {
+					locals->BandwidthWithoutPrefetchSupported[i][0] = false;
+				}
+
+				locals->PrefetchSupported[i][j] = true;
+				if (mode_lib->vba.MaximumReadBandwidthWithPrefetch > locals->ReturnBWPerState[i][0]) {
+					locals->PrefetchSupported[i][j] = false;
+				}
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+					if (locals->LineTimesForPrefetch[k] < 2.0
+							|| locals->LinesForMetaPTE[k] >= 8.0
+							|| locals->LinesForMetaAndDPTERow[k] >= 16.0
+							|| mode_lib->vba.IsErrorResult[i][j][k] == true) {
+						locals->PrefetchSupported[i][j] = false;
+					}
+				}
+				locals->VRatioInPrefetchSupported[i][j] = true;
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+					if (locals->VRatioPreY[i][j][k] > 4.0
+							|| locals->VRatioPreC[i][j][k] > 4.0
+							|| mode_lib->vba.IsErrorResult[i][j][k] == true) {
+						locals->VRatioInPrefetchSupported[i][j] = false;
+					}
+				}
+			} while ((locals->PrefetchSupported[i][j] != true || locals->VRatioInPrefetchSupported[i][j] != true)
+					&& mode_lib->vba.NextPrefetchMode < mode_lib->vba.MaxPrefetchMode);
+
+			if (mode_lib->vba.PrefetchSupported[i][j] == true
+					&& mode_lib->vba.VRatioInPrefetchSupported[i][j] == true) {
+				mode_lib->vba.BandwidthAvailableForImmediateFlip =
+						mode_lib->vba.ReturnBWPerState[i][0];
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+					mode_lib->vba.BandwidthAvailableForImmediateFlip =
+							mode_lib->vba.BandwidthAvailableForImmediateFlip
+									- mode_lib->vba.cursor_bw[k]
+									- dml_max(
+											mode_lib->vba.ReadBandwidth[k] + mode_lib->vba.qual_row_bw[k],
+											mode_lib->vba.PrefetchBW[k]);
+				}
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+					mode_lib->vba.ImmediateFlipBytes[k] = 0.0;
+					if ((mode_lib->vba.SourcePixelFormat[k] != dm_420_8
+							&& mode_lib->vba.SourcePixelFormat[k] != dm_420_10)) {
+						mode_lib->vba.ImmediateFlipBytes[k] =
+								mode_lib->vba.PDEAndMetaPTEBytesPerFrame[0][0][k]
+										+ mode_lib->vba.MetaRowBytes[0][0][k]
+										+ mode_lib->vba.DPTEBytesPerRow[0][0][k];
+					}
+				}
+				mode_lib->vba.TotImmediateFlipBytes = 0.0;
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+					if ((mode_lib->vba.SourcePixelFormat[k] != dm_420_8
+							&& mode_lib->vba.SourcePixelFormat[k] != dm_420_10)) {
+						mode_lib->vba.TotImmediateFlipBytes =
+								mode_lib->vba.TotImmediateFlipBytes
+										+ mode_lib->vba.ImmediateFlipBytes[k];
+					}
+				}
+
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+					CalculateFlipSchedule(
+							mode_lib,
+							mode_lib->vba.ExtraLatency,
+							mode_lib->vba.UrgentLatencyPixelDataOnly,
+							mode_lib->vba.GPUVMMaxPageTableLevels,
+							mode_lib->vba.GPUVMEnable,
+							mode_lib->vba.BandwidthAvailableForImmediateFlip,
+							mode_lib->vba.TotImmediateFlipBytes,
+							mode_lib->vba.SourcePixelFormat[k],
+							mode_lib->vba.ImmediateFlipBytes[k],
+							mode_lib->vba.HTotal[k]
+									/ mode_lib->vba.PixelClock[k],
+							mode_lib->vba.VRatio[k],
+							mode_lib->vba.Tno_bw[k],
+							mode_lib->vba.PDEAndMetaPTEBytesPerFrame[0][0][k],
+							mode_lib->vba.MetaRowBytes[0][0][k],
+							mode_lib->vba.DPTEBytesPerRow[0][0][k],
+							mode_lib->vba.DCCEnable[k],
+							mode_lib->vba.dpte_row_height[k],
+							mode_lib->vba.meta_row_height[k],
+							mode_lib->vba.qual_row_bw[k],
+							&mode_lib->vba.DestinationLinesToRequestVMInImmediateFlip[k],
+							&mode_lib->vba.DestinationLinesToRequestRowInImmediateFlip[k],
+							&mode_lib->vba.final_flip_bw[k],
+							&mode_lib->vba.ImmediateFlipSupportedForPipe[k]);
+				}
+				mode_lib->vba.total_dcn_read_bw_with_flip = 0.0;
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+					mode_lib->vba.total_dcn_read_bw_with_flip =
+							mode_lib->vba.total_dcn_read_bw_with_flip
+									+ mode_lib->vba.cursor_bw[k]
+									+ dml_max3(
+											mode_lib->vba.prefetch_vm_bw[k],
+											mode_lib->vba.prefetch_row_bw[k],
+											mode_lib->vba.final_flip_bw[k]
+													+ dml_max(
+															mode_lib->vba.ReadBandwidth[k],
+															mode_lib->vba.RequiredPrefetchPixelDataBWLuma[i][j][k]));
+				}
+				mode_lib->vba.ImmediateFlipSupportedForState[i][j] = true;
+				if (mode_lib->vba.total_dcn_read_bw_with_flip
+						> mode_lib->vba.ReturnBWPerState[i][0]) {
+					mode_lib->vba.ImmediateFlipSupportedForState[i][j] = false;
+				}
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+					if (mode_lib->vba.ImmediateFlipSupportedForPipe[k] == false) {
+						mode_lib->vba.ImmediateFlipSupportedForState[i][j] = false;
+					}
+				}
+			} else {
+				mode_lib->vba.ImmediateFlipSupportedForState[i][j] = false;
+			}
+		}
+	}
+
+	/*Vertical Active BW support*/
+	mode_lib->vba.MaxTotalVActiveRDBandwidth = 0;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; k++)
+		mode_lib->vba.MaxTotalVActiveRDBandwidth = mode_lib->vba.MaxTotalVActiveRDBandwidth + mode_lib->vba.ReadBandwidth[k];
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		mode_lib->vba.MaxTotalVerticalActiveAvailableBandwidth[i][0] = dml_min(mode_lib->vba.ReturnBusWidth *
+				mode_lib->vba.DCFCLKPerState[i], mode_lib->vba.FabricAndDRAMBandwidthPerState[i] * 1000) *
+				mode_lib->vba.MaxAveragePercentOfIdealDRAMBWDisplayCanUseInNormalSystemOperation / 100;
+		if (mode_lib->vba.MaxTotalVActiveRDBandwidth <= mode_lib->vba.MaxTotalVerticalActiveAvailableBandwidth[i][0])
+			mode_lib->vba.TotalVerticalActiveBandwidthSupport[i][0] = true;
+		else
+			mode_lib->vba.TotalVerticalActiveBandwidthSupport[i][0] = false;
+	}
+
+	/*PTE Buffer Size Check*/
+
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		for (j = 0; j < 2; j++) {
+			locals->PTEBufferSizeNotExceeded[i][j] = true;
+			for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+				if (locals->PTEBufferSizeNotExceededY[i][j][k] == false
+						|| locals->PTEBufferSizeNotExceededC[i][j][k] == false) {
+					locals->PTEBufferSizeNotExceeded[i][j] = false;
+				}
+			}
+		}
+	}
+	/*Cursor Support Check*/
+	mode_lib->vba.CursorSupport = true;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		for (j = 0; j < 2; j++) {
+			if (mode_lib->vba.CursorWidth[k][j] > 0.0) {
+				if (dml_floor(
+						dml_floor(
+								mode_lib->vba.CursorBufferSize
+										- mode_lib->vba.CursorChunkSize,
+								mode_lib->vba.CursorChunkSize) * 1024.0
+								/ (mode_lib->vba.CursorWidth[k][j]
+										* mode_lib->vba.CursorBPP[k][j]
+										/ 8.0),
+						1.0)
+						* (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k])
+						/ mode_lib->vba.VRatio[k] < mode_lib->vba.UrgentLatencyPixelDataOnly
+						|| (mode_lib->vba.CursorBPP[k][j] == 64.0
+								&& mode_lib->vba.Cursor64BppSupport == false)) {
+					mode_lib->vba.CursorSupport = false;
+				}
+			}
+		}
+	}
+	/*Valid Pitch Check*/
+
+	mode_lib->vba.PitchSupport = true;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		locals->AlignedYPitch[k] = dml_ceil(
+				dml_max(mode_lib->vba.PitchY[k], mode_lib->vba.ViewportWidth[k]),
+				locals->MacroTileWidthY[k]);
+		if (locals->AlignedYPitch[k] > mode_lib->vba.PitchY[k]) {
+			mode_lib->vba.PitchSupport = false;
+		}
+		if (mode_lib->vba.DCCEnable[k] == true) {
+			locals->AlignedDCCMetaPitch[k] = dml_ceil(
+					dml_max(
+							mode_lib->vba.DCCMetaPitchY[k],
+							mode_lib->vba.ViewportWidth[k]),
+					64.0 * locals->Read256BlockWidthY[k]);
+		} else {
+			locals->AlignedDCCMetaPitch[k] = mode_lib->vba.DCCMetaPitchY[k];
+		}
+		if (locals->AlignedDCCMetaPitch[k] > mode_lib->vba.DCCMetaPitchY[k]) {
+			mode_lib->vba.PitchSupport = false;
+		}
+		if (mode_lib->vba.SourcePixelFormat[k] != dm_444_64
+				&& mode_lib->vba.SourcePixelFormat[k] != dm_444_32
+				&& mode_lib->vba.SourcePixelFormat[k] != dm_444_16
+				&& mode_lib->vba.SourcePixelFormat[k] != dm_mono_16
+				&& mode_lib->vba.SourcePixelFormat[k] != dm_mono_8) {
+			locals->AlignedCPitch[k] = dml_ceil(
+					dml_max(
+							mode_lib->vba.PitchC[k],
+							mode_lib->vba.ViewportWidth[k] / 2.0),
+					locals->MacroTileWidthC[k]);
+		} else {
+			locals->AlignedCPitch[k] = mode_lib->vba.PitchC[k];
+		}
+		if (locals->AlignedCPitch[k] > mode_lib->vba.PitchC[k]) {
+			mode_lib->vba.PitchSupport = false;
+		}
+	}
+	/*Mode Support, Voltage State and SOC Configuration*/
+
+	for (i = mode_lib->vba.soc.num_states; i >= 0; i--) {
+		for (j = 0; j < 2; j++) {
+			enum dm_validation_status status = DML_VALIDATION_OK;
+
+			if (mode_lib->vba.ScaleRatioAndTapsSupport != true) {
+				status = DML_FAIL_SCALE_RATIO_TAP;
+			} else if (mode_lib->vba.SourceFormatPixelAndScanSupport != true) {
+				status = DML_FAIL_SOURCE_PIXEL_FORMAT;
+			} else if (locals->ViewportSizeSupport[i][0] != true) {
+				status = DML_FAIL_VIEWPORT_SIZE;
+			} else if (locals->DIOSupport[i] != true) {
+				status = DML_FAIL_DIO_SUPPORT;
+			} else if (locals->NotEnoughDSCUnits[i] != false) {
+				status = DML_FAIL_NOT_ENOUGH_DSC;
+			} else if (locals->DSCCLKRequiredMoreThanSupported[i] != false) {
+				status = DML_FAIL_DSC_CLK_REQUIRED;
+			} else if (locals->UrgentLatencySupport[i][j] != true) {
+				status = DML_FAIL_URGENT_LATENCY;
+			} else if (locals->ROBSupport[i][0] != true) {
+				status = DML_FAIL_REORDERING_BUFFER;
+			} else if (locals->DISPCLK_DPPCLK_Support[i][j] != true) {
+				status = DML_FAIL_DISPCLK_DPPCLK;
+			} else if (locals->TotalAvailablePipesSupport[i][j] != true) {
+				status = DML_FAIL_TOTAL_AVAILABLE_PIPES;
+			} else if (mode_lib->vba.NumberOfOTGSupport != true) {
+				status = DML_FAIL_NUM_OTG;
+			} else if (mode_lib->vba.WritebackModeSupport != true) {
+				status = DML_FAIL_WRITEBACK_MODE;
+			} else if (mode_lib->vba.WritebackLatencySupport != true) {
+				status = DML_FAIL_WRITEBACK_LATENCY;
+			} else if (mode_lib->vba.WritebackScaleRatioAndTapsSupport != true) {
+				status = DML_FAIL_WRITEBACK_SCALE_RATIO_TAP;
+			} else if (mode_lib->vba.CursorSupport != true) {
+				status = DML_FAIL_CURSOR_SUPPORT;
+			} else if (mode_lib->vba.PitchSupport != true) {
+				status = DML_FAIL_PITCH_SUPPORT;
+			} else if (locals->PrefetchSupported[i][j] != true) {
+				status = DML_FAIL_PREFETCH_SUPPORT;
+			} else if (locals->TotalVerticalActiveBandwidthSupport[i][0] != true) {
+				status = DML_FAIL_TOTAL_V_ACTIVE_BW;
+			} else if (locals->VRatioInPrefetchSupported[i][j] != true) {
+				status = DML_FAIL_V_RATIO_PREFETCH;
+			} else if (locals->PTEBufferSizeNotExceeded[i][j] != true) {
+				status = DML_FAIL_PTE_BUFFER_SIZE;
+			} else if (mode_lib->vba.NonsupportedDSCInputBPC != false) {
+				status = DML_FAIL_DSC_INPUT_BPC;
+			}
+
+			if (status == DML_VALIDATION_OK) {
+				locals->ModeSupport[i][j] = true;
+			} else {
+				locals->ModeSupport[i][j] = false;
+			}
+			locals->ValidationStatus[i] = status;
+		}
+	}
+	{
+		unsigned int MaximumMPCCombine = 0;
+		mode_lib->vba.VoltageLevel = mode_lib->vba.soc.num_states + 1;
+		for (i = mode_lib->vba.VoltageOverrideLevel; i <= mode_lib->vba.soc.num_states; i++) {
+			if (locals->ModeSupport[i][0] == true || locals->ModeSupport[i][1] == true) {
+				mode_lib->vba.VoltageLevel = i;
+				if (locals->ModeSupport[i][1] == true && (locals->ModeSupport[i][0] == false
+						|| mode_lib->vba.WhenToDoMPCCombine == dm_mpc_always_when_possible)) {
+					MaximumMPCCombine = 1;
+				} else {
+					MaximumMPCCombine = 0;
+				}
+				break;
+			}
+		}
+		mode_lib->vba.ImmediateFlipSupport =
+			locals->ImmediateFlipSupportedForState[mode_lib->vba.VoltageLevel][MaximumMPCCombine];
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			mode_lib->vba.DPPPerPlane[k] = locals->NoOfDPP[mode_lib->vba.VoltageLevel][MaximumMPCCombine][k];
+			locals->DPPCLK[k] = locals->RequiredDPPCLK[mode_lib->vba.VoltageLevel][MaximumMPCCombine][k];
+		}
+		mode_lib->vba.DISPCLK = locals->RequiredDISPCLK[mode_lib->vba.VoltageLevel][MaximumMPCCombine];
+		mode_lib->vba.maxMpcComb = MaximumMPCCombine;
+	}
+	mode_lib->vba.DCFCLK = mode_lib->vba.DCFCLKPerState[mode_lib->vba.VoltageLevel];
+	mode_lib->vba.DRAMSpeed = mode_lib->vba.DRAMSpeedPerState[mode_lib->vba.VoltageLevel];
+	mode_lib->vba.FabricClock = mode_lib->vba.FabricClockPerState[mode_lib->vba.VoltageLevel];
+	mode_lib->vba.SOCCLK = mode_lib->vba.SOCCLKPerState[mode_lib->vba.VoltageLevel];
+	mode_lib->vba.ReturnBW = locals->ReturnBWPerState[mode_lib->vba.VoltageLevel][0];
+	mode_lib->vba.FabricAndDRAMBandwidth = locals->FabricAndDRAMBandwidthPerState[mode_lib->vba.VoltageLevel];
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.BlendingAndTiming[k] == k) {
+			mode_lib->vba.ODMCombineEnabled[k] =
+					locals->ODMCombineEnablePerState[mode_lib->vba.VoltageLevel][k];
+		} else {
+			mode_lib->vba.ODMCombineEnabled[k] = 0;
+		}
+		mode_lib->vba.DSCEnabled[k] =
+				locals->RequiresDSC[mode_lib->vba.VoltageLevel][k];
+		mode_lib->vba.OutputBpp[k] =
+				locals->OutputBppPerState[mode_lib->vba.VoltageLevel][k];
+	}
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_mode_vba_20.h b/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_mode_vba_20.h
new file mode 100644
index 000000000000..92b6805f4342
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_mode_vba_20.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef _DCN20_DISPLAY_MODE_VBA_H_
+#define _DCN20_DISPLAY_MODE_VBA_H_
+
+void dml20_recalculate(struct display_mode_lib *mode_lib);
+void dml20_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_lib);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_mode_vba_20v2.c b/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_mode_vba_20v2.c
new file mode 100644
index 000000000000..c935903b68e1
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_mode_vba_20v2.c
@@ -0,0 +1,5231 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "../display_mode_lib.h"
+#include "display_mode_vba_20v2.h"
+#include "../dml_inline_defs.h"
+
+/*
+ * NOTE:
+ *   This file is gcc-parseable HW gospel, coming straight from HW engineers.
+ *
+ * It doesn't adhere to Linux kernel style and sometimes will do things in odd
+ * ways. Unless there is something clearly wrong with it the code should
+ * remain as-is as it provides us with a guarantee from HW that it is correct.
+ */
+
+#define BPP_INVALID 0
+#define BPP_BLENDED_PIPE 0xffffffff
+#define DCN20_MAX_DSC_IMAGE_WIDTH 5184
+#define DCN20_MAX_420_IMAGE_WIDTH 4096
+
+static double adjust_ReturnBW(
+		struct display_mode_lib *mode_lib,
+		double ReturnBW,
+		bool DCCEnabledAnyPlane,
+		double ReturnBandwidthToDCN);
+static unsigned int dscceComputeDelay(
+		unsigned int bpc,
+		double bpp,
+		unsigned int sliceWidth,
+		unsigned int numSlices,
+		enum output_format_class pixelFormat);
+static unsigned int dscComputeDelay(enum output_format_class pixelFormat);
+static bool CalculateDelayAfterScaler(
+		struct display_mode_lib *mode_lib,
+		double ReturnBW,
+		double ReadBandwidthPlaneLuma,
+		double ReadBandwidthPlaneChroma,
+		double TotalDataReadBandwidth,
+		double DisplayPipeLineDeliveryTimeLuma,
+		double DisplayPipeLineDeliveryTimeChroma,
+		double DPPCLK,
+		double DISPCLK,
+		double PixelClock,
+		unsigned int DSCDelay,
+		unsigned int DPPPerPlane,
+		bool ScalerEnabled,
+		unsigned int NumberOfCursors,
+		double DPPCLKDelaySubtotal,
+		double DPPCLKDelaySCL,
+		double DPPCLKDelaySCLLBOnly,
+		double DPPCLKDelayCNVCFormater,
+		double DPPCLKDelayCNVCCursor,
+		double DISPCLKDelaySubtotal,
+		unsigned int ScalerRecoutWidth,
+		enum output_format_class OutputFormat,
+		unsigned int HTotal,
+		unsigned int SwathWidthSingleDPPY,
+		double BytePerPixelDETY,
+		double BytePerPixelDETC,
+		unsigned int SwathHeightY,
+		unsigned int SwathHeightC,
+		bool Interlace,
+		bool ProgressiveToInterlaceUnitInOPP,
+		double *DSTXAfterScaler,
+		double *DSTYAfterScaler
+		);
+// Super monster function with some 45 argument
+static bool CalculatePrefetchSchedule(
+		struct display_mode_lib *mode_lib,
+		double DPPCLK,
+		double DISPCLK,
+		double PixelClock,
+		double DCFCLKDeepSleep,
+		unsigned int DPPPerPlane,
+		unsigned int NumberOfCursors,
+		unsigned int VBlank,
+		unsigned int HTotal,
+		unsigned int MaxInterDCNTileRepeaters,
+		unsigned int VStartup,
+		unsigned int PageTableLevels,
+		bool GPUVMEnable,
+		bool DynamicMetadataEnable,
+		unsigned int DynamicMetadataLinesBeforeActiveRequired,
+		unsigned int DynamicMetadataTransmittedBytes,
+		bool DCCEnable,
+		double UrgentLatencyPixelDataOnly,
+		double UrgentExtraLatency,
+		double TCalc,
+		unsigned int PDEAndMetaPTEBytesFrame,
+		unsigned int MetaRowByte,
+		unsigned int PixelPTEBytesPerRow,
+		double PrefetchSourceLinesY,
+		unsigned int SwathWidthY,
+		double BytePerPixelDETY,
+		double VInitPreFillY,
+		unsigned int MaxNumSwathY,
+		double PrefetchSourceLinesC,
+		double BytePerPixelDETC,
+		double VInitPreFillC,
+		unsigned int MaxNumSwathC,
+		unsigned int SwathHeightY,
+		unsigned int SwathHeightC,
+		double TWait,
+		bool XFCEnabled,
+		double XFCRemoteSurfaceFlipDelay,
+		bool InterlaceEnable,
+		bool ProgressiveToInterlaceUnitInOPP,
+		double DSTXAfterScaler,
+		double DSTYAfterScaler,
+		double *DestinationLinesForPrefetch,
+		double *PrefetchBandwidth,
+		double *DestinationLinesToRequestVMInVBlank,
+		double *DestinationLinesToRequestRowInVBlank,
+		double *VRatioPrefetchY,
+		double *VRatioPrefetchC,
+		double *RequiredPrefetchPixDataBW,
+		double *Tno_bw,
+		unsigned int *VUpdateOffsetPix,
+		double *VUpdateWidthPix,
+		double *VReadyOffsetPix);
+static double RoundToDFSGranularityUp(double Clock, double VCOSpeed);
+static double RoundToDFSGranularityDown(double Clock, double VCOSpeed);
+static double CalculatePrefetchSourceLines(
+		struct display_mode_lib *mode_lib,
+		double VRatio,
+		double vtaps,
+		bool Interlace,
+		bool ProgressiveToInterlaceUnitInOPP,
+		unsigned int SwathHeight,
+		unsigned int ViewportYStart,
+		double *VInitPreFill,
+		unsigned int *MaxNumSwath);
+static unsigned int CalculateVMAndRowBytes(
+		struct display_mode_lib *mode_lib,
+		bool DCCEnable,
+		unsigned int BlockHeight256Bytes,
+		unsigned int BlockWidth256Bytes,
+		enum source_format_class SourcePixelFormat,
+		unsigned int SurfaceTiling,
+		unsigned int BytePerPixel,
+		enum scan_direction_class ScanDirection,
+		unsigned int ViewportWidth,
+		unsigned int ViewportHeight,
+		unsigned int SwathWidthY,
+		bool GPUVMEnable,
+		unsigned int VMMPageSize,
+		unsigned int PTEBufferSizeInRequestsLuma,
+		unsigned int PDEProcessingBufIn64KBReqs,
+		unsigned int Pitch,
+		unsigned int DCCMetaPitch,
+		unsigned int *MacroTileWidth,
+		unsigned int *MetaRowByte,
+		unsigned int *PixelPTEBytesPerRow,
+		bool *PTEBufferSizeNotExceeded,
+		unsigned int *dpte_row_height,
+		unsigned int *meta_row_height);
+static double CalculateTWait(
+		unsigned int PrefetchMode,
+		double DRAMClockChangeLatency,
+		double UrgentLatencyPixelDataOnly,
+		double SREnterPlusExitTime);
+static double CalculateRemoteSurfaceFlipDelay(
+		struct display_mode_lib *mode_lib,
+		double VRatio,
+		double SwathWidth,
+		double Bpp,
+		double LineTime,
+		double XFCTSlvVupdateOffset,
+		double XFCTSlvVupdateWidth,
+		double XFCTSlvVreadyOffset,
+		double XFCXBUFLatencyTolerance,
+		double XFCFillBWOverhead,
+		double XFCSlvChunkSize,
+		double XFCBusTransportTime,
+		double TCalc,
+		double TWait,
+		double *SrcActiveDrainRate,
+		double *TInitXFill,
+		double *TslvChk);
+static void CalculateActiveRowBandwidth(
+		bool GPUVMEnable,
+		enum source_format_class SourcePixelFormat,
+		double VRatio,
+		bool DCCEnable,
+		double LineTime,
+		unsigned int MetaRowByteLuma,
+		unsigned int MetaRowByteChroma,
+		unsigned int meta_row_height_luma,
+		unsigned int meta_row_height_chroma,
+		unsigned int PixelPTEBytesPerRowLuma,
+		unsigned int PixelPTEBytesPerRowChroma,
+		unsigned int dpte_row_height_luma,
+		unsigned int dpte_row_height_chroma,
+		double *meta_row_bw,
+		double *dpte_row_bw,
+		double *qual_row_bw);
+static void CalculateFlipSchedule(
+		struct display_mode_lib *mode_lib,
+		double UrgentExtraLatency,
+		double UrgentLatencyPixelDataOnly,
+		unsigned int GPUVMMaxPageTableLevels,
+		bool GPUVMEnable,
+		double BandwidthAvailableForImmediateFlip,
+		unsigned int TotImmediateFlipBytes,
+		enum source_format_class SourcePixelFormat,
+		unsigned int ImmediateFlipBytes,
+		double LineTime,
+		double VRatio,
+		double Tno_bw,
+		double PDEAndMetaPTEBytesFrame,
+		unsigned int MetaRowByte,
+		unsigned int PixelPTEBytesPerRow,
+		bool DCCEnable,
+		unsigned int dpte_row_height,
+		unsigned int meta_row_height,
+		double qual_row_bw,
+		double *DestinationLinesToRequestVMInImmediateFlip,
+		double *DestinationLinesToRequestRowInImmediateFlip,
+		double *final_flip_bw,
+		bool *ImmediateFlipSupportedForPipe);
+static double CalculateWriteBackDelay(
+		enum source_format_class WritebackPixelFormat,
+		double WritebackHRatio,
+		double WritebackVRatio,
+		unsigned int WritebackLumaHTaps,
+		unsigned int WritebackLumaVTaps,
+		unsigned int WritebackChromaHTaps,
+		unsigned int WritebackChromaVTaps,
+		unsigned int WritebackDestinationWidth);
+
+static void dml20v2_DisplayPipeConfiguration(struct display_mode_lib *mode_lib);
+static void dml20v2_DISPCLKDPPCLKDCFCLKDeepSleepPrefetchParametersWatermarksAndPerformanceCalculation(
+		struct display_mode_lib *mode_lib);
+
+void dml20v2_recalculate(struct display_mode_lib *mode_lib)
+{
+	ModeSupportAndSystemConfiguration(mode_lib);
+	mode_lib->vba.FabricAndDRAMBandwidth = dml_min(
+		mode_lib->vba.DRAMSpeed * mode_lib->vba.NumberOfChannels * mode_lib->vba.DRAMChannelWidth,
+		mode_lib->vba.FabricClock * mode_lib->vba.FabricDatapathToDCNDataReturn) / 1000.0;
+	PixelClockAdjustmentForProgressiveToInterlaceUnit(mode_lib);
+	dml20v2_DisplayPipeConfiguration(mode_lib);
+	dml20v2_DISPCLKDPPCLKDCFCLKDeepSleepPrefetchParametersWatermarksAndPerformanceCalculation(mode_lib);
+}
+
+static double adjust_ReturnBW(
+		struct display_mode_lib *mode_lib,
+		double ReturnBW,
+		bool DCCEnabledAnyPlane,
+		double ReturnBandwidthToDCN)
+{
+	double CriticalCompression;
+
+	if (DCCEnabledAnyPlane
+			&& ReturnBandwidthToDCN
+					> mode_lib->vba.DCFCLK * mode_lib->vba.ReturnBusWidth / 4.0)
+		ReturnBW =
+				dml_min(
+						ReturnBW,
+						ReturnBandwidthToDCN * 4
+								* (1.0
+										- mode_lib->vba.UrgentLatencyPixelDataOnly
+												/ ((mode_lib->vba.ROBBufferSizeInKByte
+														- mode_lib->vba.PixelChunkSizeInKByte)
+														* 1024
+														/ ReturnBandwidthToDCN
+														- mode_lib->vba.DCFCLK
+																* mode_lib->vba.ReturnBusWidth
+																/ 4)
+										+ mode_lib->vba.UrgentLatencyPixelDataOnly));
+
+	CriticalCompression = 2.0 * mode_lib->vba.ReturnBusWidth * mode_lib->vba.DCFCLK
+			* mode_lib->vba.UrgentLatencyPixelDataOnly
+			/ (ReturnBandwidthToDCN * mode_lib->vba.UrgentLatencyPixelDataOnly
+					+ (mode_lib->vba.ROBBufferSizeInKByte
+							- mode_lib->vba.PixelChunkSizeInKByte)
+							* 1024);
+
+	if (DCCEnabledAnyPlane && CriticalCompression > 1.0 && CriticalCompression < 4.0)
+		ReturnBW =
+				dml_min(
+						ReturnBW,
+						4.0 * ReturnBandwidthToDCN
+								* (mode_lib->vba.ROBBufferSizeInKByte
+										- mode_lib->vba.PixelChunkSizeInKByte)
+								* 1024
+								* mode_lib->vba.ReturnBusWidth
+								* mode_lib->vba.DCFCLK
+								* mode_lib->vba.UrgentLatencyPixelDataOnly
+								/ dml_pow(
+										(ReturnBandwidthToDCN
+												* mode_lib->vba.UrgentLatencyPixelDataOnly
+												+ (mode_lib->vba.ROBBufferSizeInKByte
+														- mode_lib->vba.PixelChunkSizeInKByte)
+														* 1024),
+										2));
+
+	return ReturnBW;
+}
+
+static unsigned int dscceComputeDelay(
+		unsigned int bpc,
+		double bpp,
+		unsigned int sliceWidth,
+		unsigned int numSlices,
+		enum output_format_class pixelFormat)
+{
+	// valid bpc         = source bits per component in the set of {8, 10, 12}
+	// valid bpp         = increments of 1/16 of a bit
+	//                    min = 6/7/8 in N420/N422/444, respectively
+	//                    max = such that compression is 1:1
+	//valid sliceWidth  = number of pixels per slice line, must be less than or equal to 5184/numSlices (or 4096/numSlices in 420 mode)
+	//valid numSlices   = number of slices in the horiziontal direction per DSC engine in the set of {1, 2, 3, 4}
+	//valid pixelFormat = pixel/color format in the set of {:N444_RGB, :S422, :N422, :N420}
+
+	// fixed value
+	unsigned int rcModelSize = 8192;
+
+	// N422/N420 operate at 2 pixels per clock
+	unsigned int pixelsPerClock, lstall, D, initalXmitDelay, w, s, ix, wx, p, l0, a, ax, l,
+			Delay, pixels;
+
+	if (pixelFormat == dm_n422 || pixelFormat == dm_420)
+		pixelsPerClock = 2;
+	// #all other modes operate at 1 pixel per clock
+	else
+		pixelsPerClock = 1;
+
+	//initial transmit delay as per PPS
+	initalXmitDelay = dml_round(rcModelSize / 2.0 / bpp / pixelsPerClock);
+
+	//compute ssm delay
+	if (bpc == 8)
+		D = 81;
+	else if (bpc == 10)
+		D = 89;
+	else
+		D = 113;
+
+	//divide by pixel per cycle to compute slice width as seen by DSC
+	w = sliceWidth / pixelsPerClock;
+
+	//422 mode has an additional cycle of delay
+	if (pixelFormat == dm_s422)
+		s = 1;
+	else
+		s = 0;
+
+	//main calculation for the dscce
+	ix = initalXmitDelay + 45;
+	wx = (w + 2) / 3;
+	p = 3 * wx - w;
+	l0 = ix / w;
+	a = ix + p * l0;
+	ax = (a + 2) / 3 + D + 6 + 1;
+	l = (ax + wx - 1) / wx;
+	if ((ix % w) == 0 && p != 0)
+		lstall = 1;
+	else
+		lstall = 0;
+	Delay = l * wx * (numSlices - 1) + ax + s + lstall + 22;
+
+	//dsc processes 3 pixel containers per cycle and a container can contain 1 or 2 pixels
+	pixels = Delay * 3 * pixelsPerClock;
+	return pixels;
+}
+
+static unsigned int dscComputeDelay(enum output_format_class pixelFormat)
+{
+	unsigned int Delay = 0;
+
+	if (pixelFormat == dm_420) {
+		//   sfr
+		Delay = Delay + 2;
+		//   dsccif
+		Delay = Delay + 0;
+		//   dscc - input deserializer
+		Delay = Delay + 3;
+		//   dscc gets pixels every other cycle
+		Delay = Delay + 2;
+		//   dscc - input cdc fifo
+		Delay = Delay + 12;
+		//   dscc gets pixels every other cycle
+		Delay = Delay + 13;
+		//   dscc - cdc uncertainty
+		Delay = Delay + 2;
+		//   dscc - output cdc fifo
+		Delay = Delay + 7;
+		//   dscc gets pixels every other cycle
+		Delay = Delay + 3;
+		//   dscc - cdc uncertainty
+		Delay = Delay + 2;
+		//   dscc - output serializer
+		Delay = Delay + 1;
+		//   sft
+		Delay = Delay + 1;
+	} else if (pixelFormat == dm_n422) {
+		//   sfr
+		Delay = Delay + 2;
+		//   dsccif
+		Delay = Delay + 1;
+		//   dscc - input deserializer
+		Delay = Delay + 5;
+		//  dscc - input cdc fifo
+		Delay = Delay + 25;
+		//   dscc - cdc uncertainty
+		Delay = Delay + 2;
+		//   dscc - output cdc fifo
+		Delay = Delay + 10;
+		//   dscc - cdc uncertainty
+		Delay = Delay + 2;
+		//   dscc - output serializer
+		Delay = Delay + 1;
+		//   sft
+		Delay = Delay + 1;
+	} else {
+		//   sfr
+		Delay = Delay + 2;
+		//   dsccif
+		Delay = Delay + 0;
+		//   dscc - input deserializer
+		Delay = Delay + 3;
+		//   dscc - input cdc fifo
+		Delay = Delay + 12;
+		//   dscc - cdc uncertainty
+		Delay = Delay + 2;
+		//   dscc - output cdc fifo
+		Delay = Delay + 7;
+		//   dscc - output serializer
+		Delay = Delay + 1;
+		//   dscc - cdc uncertainty
+		Delay = Delay + 2;
+		//   sft
+		Delay = Delay + 1;
+	}
+
+	return Delay;
+}
+
+static bool CalculateDelayAfterScaler(
+		struct display_mode_lib *mode_lib,
+		double ReturnBW,
+		double ReadBandwidthPlaneLuma,
+		double ReadBandwidthPlaneChroma,
+		double TotalDataReadBandwidth,
+		double DisplayPipeLineDeliveryTimeLuma,
+		double DisplayPipeLineDeliveryTimeChroma,
+		double DPPCLK,
+		double DISPCLK,
+		double PixelClock,
+		unsigned int DSCDelay,
+		unsigned int DPPPerPlane,
+		bool ScalerEnabled,
+		unsigned int NumberOfCursors,
+		double DPPCLKDelaySubtotal,
+		double DPPCLKDelaySCL,
+		double DPPCLKDelaySCLLBOnly,
+		double DPPCLKDelayCNVCFormater,
+		double DPPCLKDelayCNVCCursor,
+		double DISPCLKDelaySubtotal,
+		unsigned int ScalerRecoutWidth,
+		enum output_format_class OutputFormat,
+		unsigned int HTotal,
+		unsigned int SwathWidthSingleDPPY,
+		double BytePerPixelDETY,
+		double BytePerPixelDETC,
+		unsigned int SwathHeightY,
+		unsigned int SwathHeightC,
+		bool Interlace,
+		bool ProgressiveToInterlaceUnitInOPP,
+		double *DSTXAfterScaler,
+		double *DSTYAfterScaler
+		)
+{
+	unsigned int DPPCycles, DISPCLKCycles;
+	double DataFabricLineDeliveryTimeLuma;
+	double DataFabricLineDeliveryTimeChroma;
+	double DSTTotalPixelsAfterScaler;
+
+	DataFabricLineDeliveryTimeLuma = SwathWidthSingleDPPY * SwathHeightY * dml_ceil(BytePerPixelDETY, 1) / (mode_lib->vba.ReturnBW * ReadBandwidthPlaneLuma / TotalDataReadBandwidth);
+	mode_lib->vba.LastPixelOfLineExtraWatermark = dml_max(mode_lib->vba.LastPixelOfLineExtraWatermark, DataFabricLineDeliveryTimeLuma - DisplayPipeLineDeliveryTimeLuma);
+
+	if (BytePerPixelDETC != 0) {
+		DataFabricLineDeliveryTimeChroma = SwathWidthSingleDPPY / 2 * SwathHeightC * dml_ceil(BytePerPixelDETC, 2) / (mode_lib->vba.ReturnBW * ReadBandwidthPlaneChroma / TotalDataReadBandwidth);
+		mode_lib->vba.LastPixelOfLineExtraWatermark = dml_max(mode_lib->vba.LastPixelOfLineExtraWatermark, DataFabricLineDeliveryTimeChroma - DisplayPipeLineDeliveryTimeChroma);
+	}
+
+	if (ScalerEnabled)
+		DPPCycles = DPPCLKDelaySubtotal + DPPCLKDelaySCL;
+	else
+		DPPCycles = DPPCLKDelaySubtotal + DPPCLKDelaySCLLBOnly;
+
+	DPPCycles = DPPCycles + DPPCLKDelayCNVCFormater + NumberOfCursors * DPPCLKDelayCNVCCursor;
+
+	DISPCLKCycles = DISPCLKDelaySubtotal;
+
+	if (DPPCLK == 0.0 || DISPCLK == 0.0)
+		return true;
+
+	*DSTXAfterScaler = DPPCycles * PixelClock / DPPCLK + DISPCLKCycles * PixelClock / DISPCLK
+			+ DSCDelay;
+
+	if (DPPPerPlane > 1)
+		*DSTXAfterScaler = *DSTXAfterScaler + ScalerRecoutWidth;
+
+	if (OutputFormat == dm_420 || (Interlace && ProgressiveToInterlaceUnitInOPP))
+		*DSTYAfterScaler = 1;
+	else
+		*DSTYAfterScaler = 0;
+
+	DSTTotalPixelsAfterScaler = ((double) (*DSTYAfterScaler * HTotal)) + *DSTXAfterScaler;
+	*DSTYAfterScaler = dml_floor(DSTTotalPixelsAfterScaler / HTotal, 1);
+	*DSTXAfterScaler = DSTTotalPixelsAfterScaler - ((double) (*DSTYAfterScaler * HTotal));
+
+	return true;
+}
+
+static bool CalculatePrefetchSchedule(
+		struct display_mode_lib *mode_lib,
+		double DPPCLK,
+		double DISPCLK,
+		double PixelClock,
+		double DCFCLKDeepSleep,
+		unsigned int DPPPerPlane,
+		unsigned int NumberOfCursors,
+		unsigned int VBlank,
+		unsigned int HTotal,
+		unsigned int MaxInterDCNTileRepeaters,
+		unsigned int VStartup,
+		unsigned int PageTableLevels,
+		bool GPUVMEnable,
+		bool DynamicMetadataEnable,
+		unsigned int DynamicMetadataLinesBeforeActiveRequired,
+		unsigned int DynamicMetadataTransmittedBytes,
+		bool DCCEnable,
+		double UrgentLatencyPixelDataOnly,
+		double UrgentExtraLatency,
+		double TCalc,
+		unsigned int PDEAndMetaPTEBytesFrame,
+		unsigned int MetaRowByte,
+		unsigned int PixelPTEBytesPerRow,
+		double PrefetchSourceLinesY,
+		unsigned int SwathWidthY,
+		double BytePerPixelDETY,
+		double VInitPreFillY,
+		unsigned int MaxNumSwathY,
+		double PrefetchSourceLinesC,
+		double BytePerPixelDETC,
+		double VInitPreFillC,
+		unsigned int MaxNumSwathC,
+		unsigned int SwathHeightY,
+		unsigned int SwathHeightC,
+		double TWait,
+		bool XFCEnabled,
+		double XFCRemoteSurfaceFlipDelay,
+		bool InterlaceEnable,
+		bool ProgressiveToInterlaceUnitInOPP,
+		double DSTXAfterScaler,
+		double DSTYAfterScaler,
+		double *DestinationLinesForPrefetch,
+		double *PrefetchBandwidth,
+		double *DestinationLinesToRequestVMInVBlank,
+		double *DestinationLinesToRequestRowInVBlank,
+		double *VRatioPrefetchY,
+		double *VRatioPrefetchC,
+		double *RequiredPrefetchPixDataBW,
+		double *Tno_bw,
+		unsigned int *VUpdateOffsetPix,
+		double *VUpdateWidthPix,
+		double *VReadyOffsetPix)
+{
+	bool MyError = false;
+	double TotalRepeaterDelayTime;
+	double Tdm, LineTime, Tsetup;
+	double dst_y_prefetch_equ;
+	double Tsw_oto;
+	double prefetch_bw_oto;
+	double Tvm_oto;
+	double Tr0_oto;
+	double Tpre_oto;
+	double dst_y_prefetch_oto;
+	double TimeForFetchingMetaPTE = 0;
+	double TimeForFetchingRowInVBlank = 0;
+	double LinesToRequestPrefetchPixelData = 0;
+
+	*VUpdateOffsetPix = dml_ceil(HTotal / 4.0, 1);
+	TotalRepeaterDelayTime = MaxInterDCNTileRepeaters * (2.0 / DPPCLK + 3.0 / DISPCLK);
+	*VUpdateWidthPix = (14.0 / DCFCLKDeepSleep + 12.0 / DPPCLK + TotalRepeaterDelayTime)
+			* PixelClock;
+
+	*VReadyOffsetPix = dml_max(
+			150.0 / DPPCLK,
+			TotalRepeaterDelayTime + 20.0 / DCFCLKDeepSleep + 10.0 / DPPCLK)
+			* PixelClock;
+
+	Tsetup = (double) (*VUpdateOffsetPix + *VUpdateWidthPix + *VReadyOffsetPix) / PixelClock;
+
+	LineTime = (double) HTotal / PixelClock;
+
+	if (DynamicMetadataEnable) {
+		double Tdmbf, Tdmec, Tdmsks;
+
+		Tdm = dml_max(0.0, UrgentExtraLatency - TCalc);
+		Tdmbf = DynamicMetadataTransmittedBytes / 4.0 / DISPCLK;
+		Tdmec = LineTime;
+		if (DynamicMetadataLinesBeforeActiveRequired == 0)
+			Tdmsks = VBlank * LineTime / 2.0;
+		else
+			Tdmsks = DynamicMetadataLinesBeforeActiveRequired * LineTime;
+		if (InterlaceEnable && !ProgressiveToInterlaceUnitInOPP)
+			Tdmsks = Tdmsks / 2;
+		if (VStartup * LineTime
+				< Tsetup + TWait + UrgentExtraLatency + Tdmbf + Tdmec + Tdmsks) {
+			MyError = true;
+		}
+	} else
+		Tdm = 0;
+
+	if (GPUVMEnable) {
+		if (PageTableLevels == 4)
+			*Tno_bw = UrgentExtraLatency + UrgentLatencyPixelDataOnly;
+		else if (PageTableLevels == 3)
+			*Tno_bw = UrgentExtraLatency;
+		else
+			*Tno_bw = 0;
+	} else if (DCCEnable)
+		*Tno_bw = LineTime;
+	else
+		*Tno_bw = LineTime / 4;
+
+	dst_y_prefetch_equ = VStartup - dml_max(TCalc + TWait, XFCRemoteSurfaceFlipDelay) / LineTime
+			- (Tsetup + Tdm) / LineTime
+			- (DSTYAfterScaler + DSTXAfterScaler / HTotal);
+
+	Tsw_oto = dml_max(PrefetchSourceLinesY, PrefetchSourceLinesC) * LineTime;
+
+	prefetch_bw_oto = (MetaRowByte + PixelPTEBytesPerRow
+			+ PrefetchSourceLinesY * SwathWidthY * dml_ceil(BytePerPixelDETY, 1)
+			+ PrefetchSourceLinesC * SwathWidthY / 2 * dml_ceil(BytePerPixelDETC, 2))
+			/ Tsw_oto;
+
+	if (GPUVMEnable == true) {
+		Tvm_oto =
+				dml_max(
+						*Tno_bw + PDEAndMetaPTEBytesFrame / prefetch_bw_oto,
+						dml_max(
+								UrgentExtraLatency
+										+ UrgentLatencyPixelDataOnly
+												* (PageTableLevels
+														- 1),
+								LineTime / 4.0));
+	} else
+		Tvm_oto = LineTime / 4.0;
+
+	if ((GPUVMEnable == true || DCCEnable == true)) {
+		Tr0_oto = dml_max(
+				(MetaRowByte + PixelPTEBytesPerRow) / prefetch_bw_oto,
+				dml_max(UrgentLatencyPixelDataOnly, dml_max(LineTime - Tvm_oto, LineTime / 4)));
+	} else
+		Tr0_oto = LineTime - Tvm_oto;
+
+	Tpre_oto = Tvm_oto + Tr0_oto + Tsw_oto;
+
+	dst_y_prefetch_oto = Tpre_oto / LineTime;
+
+	if (dst_y_prefetch_oto < dst_y_prefetch_equ)
+		*DestinationLinesForPrefetch = dst_y_prefetch_oto;
+	else
+		*DestinationLinesForPrefetch = dst_y_prefetch_equ;
+
+	*DestinationLinesForPrefetch = dml_floor(4.0 * (*DestinationLinesForPrefetch + 0.125), 1)
+			/ 4;
+
+	dml_print("DML: VStartup: %d\n", VStartup);
+	dml_print("DML: TCalc: %f\n", TCalc);
+	dml_print("DML: TWait: %f\n", TWait);
+	dml_print("DML: XFCRemoteSurfaceFlipDelay: %f\n", XFCRemoteSurfaceFlipDelay);
+	dml_print("DML: LineTime: %f\n", LineTime);
+	dml_print("DML: Tsetup: %f\n", Tsetup);
+	dml_print("DML: Tdm: %f\n", Tdm);
+	dml_print("DML: DSTYAfterScaler: %f\n", DSTYAfterScaler);
+	dml_print("DML: DSTXAfterScaler: %f\n", DSTXAfterScaler);
+	dml_print("DML: HTotal: %d\n", HTotal);
+
+	*PrefetchBandwidth = 0;
+	*DestinationLinesToRequestVMInVBlank = 0;
+	*DestinationLinesToRequestRowInVBlank = 0;
+	*VRatioPrefetchY = 0;
+	*VRatioPrefetchC = 0;
+	*RequiredPrefetchPixDataBW = 0;
+	if (*DestinationLinesForPrefetch > 1) {
+		*PrefetchBandwidth = (PDEAndMetaPTEBytesFrame + 2 * MetaRowByte
+				+ 2 * PixelPTEBytesPerRow
+				+ PrefetchSourceLinesY * SwathWidthY * dml_ceil(BytePerPixelDETY, 1)
+				+ PrefetchSourceLinesC * SwathWidthY / 2
+						* dml_ceil(BytePerPixelDETC, 2))
+				/ (*DestinationLinesForPrefetch * LineTime - *Tno_bw);
+		if (GPUVMEnable) {
+			TimeForFetchingMetaPTE =
+					dml_max(
+							*Tno_bw
+									+ (double) PDEAndMetaPTEBytesFrame
+											/ *PrefetchBandwidth,
+							dml_max(
+									UrgentExtraLatency
+											+ UrgentLatencyPixelDataOnly
+													* (PageTableLevels
+															- 1),
+									LineTime / 4));
+		} else {
+			if (NumberOfCursors > 0 || XFCEnabled)
+				TimeForFetchingMetaPTE = LineTime / 4;
+			else
+				TimeForFetchingMetaPTE = 0.0;
+		}
+
+		if ((GPUVMEnable == true || DCCEnable == true)) {
+			TimeForFetchingRowInVBlank =
+					dml_max(
+							(MetaRowByte + PixelPTEBytesPerRow)
+									/ *PrefetchBandwidth,
+							dml_max(
+									UrgentLatencyPixelDataOnly,
+									dml_max(
+											LineTime
+													- TimeForFetchingMetaPTE,
+											LineTime
+													/ 4.0)));
+		} else {
+			if (NumberOfCursors > 0 || XFCEnabled)
+				TimeForFetchingRowInVBlank = LineTime - TimeForFetchingMetaPTE;
+			else
+				TimeForFetchingRowInVBlank = 0.0;
+		}
+
+		*DestinationLinesToRequestVMInVBlank = dml_floor(
+				4.0 * (TimeForFetchingMetaPTE / LineTime + 0.125),
+				1) / 4.0;
+
+		*DestinationLinesToRequestRowInVBlank = dml_floor(
+				4.0 * (TimeForFetchingRowInVBlank / LineTime + 0.125),
+				1) / 4.0;
+
+		LinesToRequestPrefetchPixelData =
+				*DestinationLinesForPrefetch
+						- ((NumberOfCursors > 0 || GPUVMEnable
+								|| DCCEnable) ?
+								(*DestinationLinesToRequestVMInVBlank
+										+ *DestinationLinesToRequestRowInVBlank) :
+								0.0);
+
+		if (LinesToRequestPrefetchPixelData > 0) {
+
+			*VRatioPrefetchY = (double) PrefetchSourceLinesY
+					/ LinesToRequestPrefetchPixelData;
+			*VRatioPrefetchY = dml_max(*VRatioPrefetchY, 1.0);
+			if ((SwathHeightY > 4) && (VInitPreFillY > 3)) {
+				if (LinesToRequestPrefetchPixelData > (VInitPreFillY - 3.0) / 2.0) {
+					*VRatioPrefetchY =
+							dml_max(
+									(double) PrefetchSourceLinesY
+											/ LinesToRequestPrefetchPixelData,
+									(double) MaxNumSwathY
+											* SwathHeightY
+											/ (LinesToRequestPrefetchPixelData
+													- (VInitPreFillY
+															- 3.0)
+															/ 2.0));
+					*VRatioPrefetchY = dml_max(*VRatioPrefetchY, 1.0);
+				} else {
+					MyError = true;
+					*VRatioPrefetchY = 0;
+				}
+			}
+
+			*VRatioPrefetchC = (double) PrefetchSourceLinesC
+					/ LinesToRequestPrefetchPixelData;
+			*VRatioPrefetchC = dml_max(*VRatioPrefetchC, 1.0);
+
+			if ((SwathHeightC > 4)) {
+				if (LinesToRequestPrefetchPixelData > (VInitPreFillC - 3.0) / 2.0) {
+					*VRatioPrefetchC =
+							dml_max(
+									*VRatioPrefetchC,
+									(double) MaxNumSwathC
+											* SwathHeightC
+											/ (LinesToRequestPrefetchPixelData
+													- (VInitPreFillC
+															- 3.0)
+															/ 2.0));
+					*VRatioPrefetchC = dml_max(*VRatioPrefetchC, 1.0);
+				} else {
+					MyError = true;
+					*VRatioPrefetchC = 0;
+				}
+			}
+
+			*RequiredPrefetchPixDataBW =
+					DPPPerPlane
+							* ((double) PrefetchSourceLinesY
+									/ LinesToRequestPrefetchPixelData
+									* dml_ceil(
+											BytePerPixelDETY,
+											1)
+									+ (double) PrefetchSourceLinesC
+											/ LinesToRequestPrefetchPixelData
+											* dml_ceil(
+													BytePerPixelDETC,
+													2)
+											/ 2)
+							* SwathWidthY / LineTime;
+		} else {
+			MyError = true;
+			*VRatioPrefetchY = 0;
+			*VRatioPrefetchC = 0;
+			*RequiredPrefetchPixDataBW = 0;
+		}
+
+	} else {
+		MyError = true;
+	}
+
+	if (MyError) {
+		*PrefetchBandwidth = 0;
+		*DestinationLinesToRequestVMInVBlank = 0;
+		*DestinationLinesToRequestRowInVBlank = 0;
+		*DestinationLinesForPrefetch = 0;
+		*VRatioPrefetchY = 0;
+		*VRatioPrefetchC = 0;
+		*RequiredPrefetchPixDataBW = 0;
+	}
+
+	return MyError;
+}
+
+static double RoundToDFSGranularityUp(double Clock, double VCOSpeed)
+{
+	return VCOSpeed * 4 / dml_floor(VCOSpeed * 4 / Clock, 1);
+}
+
+static double RoundToDFSGranularityDown(double Clock, double VCOSpeed)
+{
+	return VCOSpeed * 4 / dml_ceil(VCOSpeed * 4 / Clock, 1);
+}
+
+static double CalculatePrefetchSourceLines(
+		struct display_mode_lib *mode_lib,
+		double VRatio,
+		double vtaps,
+		bool Interlace,
+		bool ProgressiveToInterlaceUnitInOPP,
+		unsigned int SwathHeight,
+		unsigned int ViewportYStart,
+		double *VInitPreFill,
+		unsigned int *MaxNumSwath)
+{
+	unsigned int MaxPartialSwath;
+
+	if (ProgressiveToInterlaceUnitInOPP)
+		*VInitPreFill = dml_floor((VRatio + vtaps + 1) / 2.0, 1);
+	else
+		*VInitPreFill = dml_floor((VRatio + vtaps + 1 + Interlace * 0.5 * VRatio) / 2.0, 1);
+
+	if (!mode_lib->vba.IgnoreViewportPositioning) {
+
+		*MaxNumSwath = dml_ceil((*VInitPreFill - 1.0) / SwathHeight, 1) + 1.0;
+
+		if (*VInitPreFill > 1.0)
+			MaxPartialSwath = (unsigned int) (*VInitPreFill - 2) % SwathHeight;
+		else
+			MaxPartialSwath = (unsigned int) (*VInitPreFill + SwathHeight - 2)
+					% SwathHeight;
+		MaxPartialSwath = dml_max(1U, MaxPartialSwath);
+
+	} else {
+
+		if (ViewportYStart != 0)
+			dml_print(
+					"WARNING DML: using viewport y position of 0 even though actual viewport y position is non-zero in prefetch source lines calculation\n");
+
+		*MaxNumSwath = dml_ceil(*VInitPreFill / SwathHeight, 1);
+
+		if (*VInitPreFill > 1.0)
+			MaxPartialSwath = (unsigned int) (*VInitPreFill - 1) % SwathHeight;
+		else
+			MaxPartialSwath = (unsigned int) (*VInitPreFill + SwathHeight - 1)
+					% SwathHeight;
+	}
+
+	return *MaxNumSwath * SwathHeight + MaxPartialSwath;
+}
+
+static unsigned int CalculateVMAndRowBytes(
+		struct display_mode_lib *mode_lib,
+		bool DCCEnable,
+		unsigned int BlockHeight256Bytes,
+		unsigned int BlockWidth256Bytes,
+		enum source_format_class SourcePixelFormat,
+		unsigned int SurfaceTiling,
+		unsigned int BytePerPixel,
+		enum scan_direction_class ScanDirection,
+		unsigned int ViewportWidth,
+		unsigned int ViewportHeight,
+		unsigned int SwathWidth,
+		bool GPUVMEnable,
+		unsigned int VMMPageSize,
+		unsigned int PTEBufferSizeInRequestsLuma,
+		unsigned int PDEProcessingBufIn64KBReqs,
+		unsigned int Pitch,
+		unsigned int DCCMetaPitch,
+		unsigned int *MacroTileWidth,
+		unsigned int *MetaRowByte,
+		unsigned int *PixelPTEBytesPerRow,
+		bool *PTEBufferSizeNotExceeded,
+		unsigned int *dpte_row_height,
+		unsigned int *meta_row_height)
+{
+	unsigned int MetaRequestHeight;
+	unsigned int MetaRequestWidth;
+	unsigned int MetaSurfWidth;
+	unsigned int MetaSurfHeight;
+	unsigned int MPDEBytesFrame;
+	unsigned int MetaPTEBytesFrame;
+	unsigned int DCCMetaSurfaceBytes;
+
+	unsigned int MacroTileSizeBytes;
+	unsigned int MacroTileHeight;
+	unsigned int DPDE0BytesFrame;
+	unsigned int ExtraDPDEBytesFrame;
+	unsigned int PDEAndMetaPTEBytesFrame;
+
+	if (DCCEnable == true) {
+		MetaRequestHeight = 8 * BlockHeight256Bytes;
+		MetaRequestWidth = 8 * BlockWidth256Bytes;
+		if (ScanDirection == dm_horz) {
+			*meta_row_height = MetaRequestHeight;
+			MetaSurfWidth = dml_ceil((double) SwathWidth - 1, MetaRequestWidth)
+					+ MetaRequestWidth;
+			*MetaRowByte = MetaSurfWidth * MetaRequestHeight * BytePerPixel / 256.0;
+		} else {
+			*meta_row_height = MetaRequestWidth;
+			MetaSurfHeight = dml_ceil((double) SwathWidth - 1, MetaRequestHeight)
+					+ MetaRequestHeight;
+			*MetaRowByte = MetaSurfHeight * MetaRequestWidth * BytePerPixel / 256.0;
+		}
+		if (ScanDirection == dm_horz) {
+			DCCMetaSurfaceBytes = DCCMetaPitch
+					* (dml_ceil(ViewportHeight - 1, 64 * BlockHeight256Bytes)
+							+ 64 * BlockHeight256Bytes) * BytePerPixel
+					/ 256;
+		} else {
+			DCCMetaSurfaceBytes = DCCMetaPitch
+					* (dml_ceil(
+							(double) ViewportHeight - 1,
+							64 * BlockHeight256Bytes)
+							+ 64 * BlockHeight256Bytes) * BytePerPixel
+					/ 256;
+		}
+		if (GPUVMEnable == true) {
+			MetaPTEBytesFrame = (dml_ceil(
+					(double) (DCCMetaSurfaceBytes - VMMPageSize)
+							/ (8 * VMMPageSize),
+					1) + 1) * 64;
+			MPDEBytesFrame = 128 * (mode_lib->vba.GPUVMMaxPageTableLevels - 1);
+		} else {
+			MetaPTEBytesFrame = 0;
+			MPDEBytesFrame = 0;
+		}
+	} else {
+		MetaPTEBytesFrame = 0;
+		MPDEBytesFrame = 0;
+		*MetaRowByte = 0;
+	}
+
+	if (SurfaceTiling == dm_sw_linear || SurfaceTiling == dm_sw_gfx7_2d_thin_gl || SurfaceTiling == dm_sw_gfx7_2d_thin_l_vp) {
+		MacroTileSizeBytes = 256;
+		MacroTileHeight = BlockHeight256Bytes;
+	} else if (SurfaceTiling == dm_sw_4kb_s || SurfaceTiling == dm_sw_4kb_s_x
+			|| SurfaceTiling == dm_sw_4kb_d || SurfaceTiling == dm_sw_4kb_d_x) {
+		MacroTileSizeBytes = 4096;
+		MacroTileHeight = 4 * BlockHeight256Bytes;
+	} else if (SurfaceTiling == dm_sw_64kb_s || SurfaceTiling == dm_sw_64kb_s_t
+			|| SurfaceTiling == dm_sw_64kb_s_x || SurfaceTiling == dm_sw_64kb_d
+			|| SurfaceTiling == dm_sw_64kb_d_t || SurfaceTiling == dm_sw_64kb_d_x
+			|| SurfaceTiling == dm_sw_64kb_r_x) {
+		MacroTileSizeBytes = 65536;
+		MacroTileHeight = 16 * BlockHeight256Bytes;
+	} else {
+		MacroTileSizeBytes = 262144;
+		MacroTileHeight = 32 * BlockHeight256Bytes;
+	}
+	*MacroTileWidth = MacroTileSizeBytes / BytePerPixel / MacroTileHeight;
+
+	if (GPUVMEnable == true && mode_lib->vba.GPUVMMaxPageTableLevels > 1) {
+		if (ScanDirection == dm_horz) {
+			DPDE0BytesFrame =
+					64
+							* (dml_ceil(
+									((Pitch
+											* (dml_ceil(
+													ViewportHeight
+															- 1,
+													MacroTileHeight)
+													+ MacroTileHeight)
+											* BytePerPixel)
+											- MacroTileSizeBytes)
+											/ (8
+													* 2097152),
+									1) + 1);
+		} else {
+			DPDE0BytesFrame =
+					64
+							* (dml_ceil(
+									((Pitch
+											* (dml_ceil(
+													(double) SwathWidth
+															- 1,
+													MacroTileHeight)
+													+ MacroTileHeight)
+											* BytePerPixel)
+											- MacroTileSizeBytes)
+											/ (8
+													* 2097152),
+									1) + 1);
+		}
+		ExtraDPDEBytesFrame = 128 * (mode_lib->vba.GPUVMMaxPageTableLevels - 2);
+	} else {
+		DPDE0BytesFrame = 0;
+		ExtraDPDEBytesFrame = 0;
+	}
+
+	PDEAndMetaPTEBytesFrame = MetaPTEBytesFrame + MPDEBytesFrame + DPDE0BytesFrame
+			+ ExtraDPDEBytesFrame;
+
+	if (GPUVMEnable == true) {
+		unsigned int PTERequestSize;
+		unsigned int PixelPTEReqHeight;
+		unsigned int PixelPTEReqWidth;
+		double FractionOfPTEReturnDrop;
+		unsigned int EffectivePDEProcessingBufIn64KBReqs;
+
+		if (SurfaceTiling == dm_sw_linear) {
+			PixelPTEReqHeight = 1;
+			PixelPTEReqWidth = 8.0 * VMMPageSize / BytePerPixel;
+			PTERequestSize = 64;
+			FractionOfPTEReturnDrop = 0;
+		} else if (MacroTileSizeBytes == 4096) {
+			PixelPTEReqHeight = MacroTileHeight;
+			PixelPTEReqWidth = 8 * *MacroTileWidth;
+			PTERequestSize = 64;
+			if (ScanDirection == dm_horz)
+				FractionOfPTEReturnDrop = 0;
+			else
+				FractionOfPTEReturnDrop = 7.0 / 8;
+		} else if (VMMPageSize == 4096 && MacroTileSizeBytes > 4096) {
+			PixelPTEReqHeight = 16 * BlockHeight256Bytes;
+			PixelPTEReqWidth = 16 * BlockWidth256Bytes;
+			PTERequestSize = 128;
+			FractionOfPTEReturnDrop = 0;
+		} else {
+			PixelPTEReqHeight = MacroTileHeight;
+			PixelPTEReqWidth = 8 * *MacroTileWidth;
+			PTERequestSize = 64;
+			FractionOfPTEReturnDrop = 0;
+		}
+
+		if (SourcePixelFormat == dm_420_8 || SourcePixelFormat == dm_420_10)
+			EffectivePDEProcessingBufIn64KBReqs = PDEProcessingBufIn64KBReqs / 2;
+		else
+			EffectivePDEProcessingBufIn64KBReqs = PDEProcessingBufIn64KBReqs;
+
+		if (SurfaceTiling == dm_sw_linear) {
+			*dpte_row_height =
+					dml_min(
+							128,
+							1
+									<< (unsigned int) dml_floor(
+											dml_log2(
+													dml_min(
+															(double) PTEBufferSizeInRequestsLuma
+																	* PixelPTEReqWidth,
+															EffectivePDEProcessingBufIn64KBReqs
+																	* 65536.0
+																	/ BytePerPixel)
+															/ Pitch),
+											1));
+			*PixelPTEBytesPerRow = PTERequestSize
+					* (dml_ceil(
+							(double) (Pitch * *dpte_row_height - 1)
+									/ PixelPTEReqWidth,
+							1) + 1);
+		} else if (ScanDirection == dm_horz) {
+			*dpte_row_height = PixelPTEReqHeight;
+			*PixelPTEBytesPerRow = PTERequestSize
+					* (dml_ceil(((double) SwathWidth - 1) / PixelPTEReqWidth, 1)
+							+ 1);
+		} else {
+			*dpte_row_height = dml_min(PixelPTEReqWidth, *MacroTileWidth);
+			*PixelPTEBytesPerRow = PTERequestSize
+					* (dml_ceil(
+							((double) SwathWidth - 1)
+									/ PixelPTEReqHeight,
+							1) + 1);
+		}
+		if (*PixelPTEBytesPerRow * (1 - FractionOfPTEReturnDrop)
+				<= 64 * PTEBufferSizeInRequestsLuma) {
+			*PTEBufferSizeNotExceeded = true;
+		} else {
+			*PTEBufferSizeNotExceeded = false;
+		}
+	} else {
+		*PixelPTEBytesPerRow = 0;
+		*PTEBufferSizeNotExceeded = true;
+	}
+
+	return PDEAndMetaPTEBytesFrame;
+}
+
+static void dml20v2_DISPCLKDPPCLKDCFCLKDeepSleepPrefetchParametersWatermarksAndPerformanceCalculation(
+		struct display_mode_lib *mode_lib)
+{
+	unsigned int j, k;
+
+	mode_lib->vba.WritebackDISPCLK = 0.0;
+	mode_lib->vba.DISPCLKWithRamping = 0;
+	mode_lib->vba.DISPCLKWithoutRamping = 0;
+	mode_lib->vba.GlobalDPPCLK = 0.0;
+
+	// dml_ml->vba.DISPCLK and dml_ml->vba.DPPCLK Calculation
+	//
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.WritebackEnable[k]) {
+			mode_lib->vba.WritebackDISPCLK =
+					dml_max(
+							mode_lib->vba.WritebackDISPCLK,
+							CalculateWriteBackDISPCLK(
+									mode_lib->vba.WritebackPixelFormat[k],
+									mode_lib->vba.PixelClock[k],
+									mode_lib->vba.WritebackHRatio[k],
+									mode_lib->vba.WritebackVRatio[k],
+									mode_lib->vba.WritebackLumaHTaps[k],
+									mode_lib->vba.WritebackLumaVTaps[k],
+									mode_lib->vba.WritebackChromaHTaps[k],
+									mode_lib->vba.WritebackChromaVTaps[k],
+									mode_lib->vba.WritebackDestinationWidth[k],
+									mode_lib->vba.HTotal[k],
+									mode_lib->vba.WritebackChromaLineBufferWidth));
+		}
+	}
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.HRatio[k] > 1) {
+			mode_lib->vba.PSCL_THROUGHPUT_LUMA[k] = dml_min(
+					mode_lib->vba.MaxDCHUBToPSCLThroughput,
+					mode_lib->vba.MaxPSCLToLBThroughput
+							* mode_lib->vba.HRatio[k]
+							/ dml_ceil(
+									mode_lib->vba.htaps[k]
+											/ 6.0,
+									1));
+		} else {
+			mode_lib->vba.PSCL_THROUGHPUT_LUMA[k] = dml_min(
+					mode_lib->vba.MaxDCHUBToPSCLThroughput,
+					mode_lib->vba.MaxPSCLToLBThroughput);
+		}
+
+		mode_lib->vba.DPPCLKUsingSingleDPPLuma =
+				mode_lib->vba.PixelClock[k]
+						* dml_max(
+								mode_lib->vba.vtaps[k] / 6.0
+										* dml_min(
+												1.0,
+												mode_lib->vba.HRatio[k]),
+								dml_max(
+										mode_lib->vba.HRatio[k]
+												* mode_lib->vba.VRatio[k]
+												/ mode_lib->vba.PSCL_THROUGHPUT_LUMA[k],
+										1.0));
+
+		if ((mode_lib->vba.htaps[k] > 6 || mode_lib->vba.vtaps[k] > 6)
+				&& mode_lib->vba.DPPCLKUsingSingleDPPLuma
+						< 2 * mode_lib->vba.PixelClock[k]) {
+			mode_lib->vba.DPPCLKUsingSingleDPPLuma = 2 * mode_lib->vba.PixelClock[k];
+		}
+
+		if ((mode_lib->vba.SourcePixelFormat[k] != dm_420_8
+				&& mode_lib->vba.SourcePixelFormat[k] != dm_420_10)) {
+			mode_lib->vba.PSCL_THROUGHPUT_CHROMA[k] = 0.0;
+			mode_lib->vba.DPPCLKUsingSingleDPP[k] =
+					mode_lib->vba.DPPCLKUsingSingleDPPLuma;
+		} else {
+			if (mode_lib->vba.HRatio[k] > 1) {
+				mode_lib->vba.PSCL_THROUGHPUT_CHROMA[k] =
+						dml_min(
+								mode_lib->vba.MaxDCHUBToPSCLThroughput,
+								mode_lib->vba.MaxPSCLToLBThroughput
+										* mode_lib->vba.HRatio[k]
+										/ 2
+										/ dml_ceil(
+												mode_lib->vba.HTAPsChroma[k]
+														/ 6.0,
+												1.0));
+			} else {
+				mode_lib->vba.PSCL_THROUGHPUT_CHROMA[k] = dml_min(
+						mode_lib->vba.MaxDCHUBToPSCLThroughput,
+						mode_lib->vba.MaxPSCLToLBThroughput);
+			}
+			mode_lib->vba.DPPCLKUsingSingleDPPChroma =
+					mode_lib->vba.PixelClock[k]
+							* dml_max(
+									mode_lib->vba.VTAPsChroma[k]
+											/ 6.0
+											* dml_min(
+													1.0,
+													mode_lib->vba.HRatio[k]
+															/ 2),
+									dml_max(
+											mode_lib->vba.HRatio[k]
+													* mode_lib->vba.VRatio[k]
+													/ 4
+													/ mode_lib->vba.PSCL_THROUGHPUT_CHROMA[k],
+											1.0));
+
+			if ((mode_lib->vba.HTAPsChroma[k] > 6 || mode_lib->vba.VTAPsChroma[k] > 6)
+					&& mode_lib->vba.DPPCLKUsingSingleDPPChroma
+							< 2 * mode_lib->vba.PixelClock[k]) {
+				mode_lib->vba.DPPCLKUsingSingleDPPChroma = 2
+						* mode_lib->vba.PixelClock[k];
+			}
+
+			mode_lib->vba.DPPCLKUsingSingleDPP[k] = dml_max(
+					mode_lib->vba.DPPCLKUsingSingleDPPLuma,
+					mode_lib->vba.DPPCLKUsingSingleDPPChroma);
+		}
+	}
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.BlendingAndTiming[k] != k)
+			continue;
+		if (mode_lib->vba.ODMCombineEnabled[k]) {
+			mode_lib->vba.DISPCLKWithRamping =
+					dml_max(
+							mode_lib->vba.DISPCLKWithRamping,
+							mode_lib->vba.PixelClock[k] / 2
+									* (1
+											+ mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading
+													/ 100)
+									* (1
+											+ mode_lib->vba.DISPCLKRampingMargin
+													/ 100));
+			mode_lib->vba.DISPCLKWithoutRamping =
+					dml_max(
+							mode_lib->vba.DISPCLKWithoutRamping,
+							mode_lib->vba.PixelClock[k] / 2
+									* (1
+											+ mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading
+													/ 100));
+		} else if (!mode_lib->vba.ODMCombineEnabled[k]) {
+			mode_lib->vba.DISPCLKWithRamping =
+					dml_max(
+							mode_lib->vba.DISPCLKWithRamping,
+							mode_lib->vba.PixelClock[k]
+									* (1
+											+ mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading
+													/ 100)
+									* (1
+											+ mode_lib->vba.DISPCLKRampingMargin
+													/ 100));
+			mode_lib->vba.DISPCLKWithoutRamping =
+					dml_max(
+							mode_lib->vba.DISPCLKWithoutRamping,
+							mode_lib->vba.PixelClock[k]
+									* (1
+											+ mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading
+													/ 100));
+		}
+	}
+
+	mode_lib->vba.DISPCLKWithRamping = dml_max(
+			mode_lib->vba.DISPCLKWithRamping,
+			mode_lib->vba.WritebackDISPCLK);
+	mode_lib->vba.DISPCLKWithoutRamping = dml_max(
+			mode_lib->vba.DISPCLKWithoutRamping,
+			mode_lib->vba.WritebackDISPCLK);
+
+	ASSERT(mode_lib->vba.DISPCLKDPPCLKVCOSpeed != 0);
+	mode_lib->vba.DISPCLKWithRampingRoundedToDFSGranularity = RoundToDFSGranularityUp(
+			mode_lib->vba.DISPCLKWithRamping,
+			mode_lib->vba.DISPCLKDPPCLKVCOSpeed);
+	mode_lib->vba.DISPCLKWithoutRampingRoundedToDFSGranularity = RoundToDFSGranularityUp(
+			mode_lib->vba.DISPCLKWithoutRamping,
+			mode_lib->vba.DISPCLKDPPCLKVCOSpeed);
+	mode_lib->vba.MaxDispclkRoundedToDFSGranularity = RoundToDFSGranularityDown(
+			mode_lib->vba.soc.clock_limits[mode_lib->vba.soc.num_states].dispclk_mhz,
+			mode_lib->vba.DISPCLKDPPCLKVCOSpeed);
+	if (mode_lib->vba.DISPCLKWithoutRampingRoundedToDFSGranularity
+			> mode_lib->vba.MaxDispclkRoundedToDFSGranularity) {
+		mode_lib->vba.DISPCLK_calculated =
+				mode_lib->vba.DISPCLKWithoutRampingRoundedToDFSGranularity;
+	} else if (mode_lib->vba.DISPCLKWithRampingRoundedToDFSGranularity
+			> mode_lib->vba.MaxDispclkRoundedToDFSGranularity) {
+		mode_lib->vba.DISPCLK_calculated = mode_lib->vba.MaxDispclkRoundedToDFSGranularity;
+	} else {
+		mode_lib->vba.DISPCLK_calculated =
+				mode_lib->vba.DISPCLKWithRampingRoundedToDFSGranularity;
+	}
+	DTRACE("   dispclk_mhz (calculated) = %f", mode_lib->vba.DISPCLK_calculated);
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.DPPPerPlane[k] == 0) {
+			mode_lib->vba.DPPCLK_calculated[k] = 0;
+		} else {
+			mode_lib->vba.DPPCLK_calculated[k] = mode_lib->vba.DPPCLKUsingSingleDPP[k]
+					/ mode_lib->vba.DPPPerPlane[k]
+					* (1 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100);
+		}
+		mode_lib->vba.GlobalDPPCLK = dml_max(
+				mode_lib->vba.GlobalDPPCLK,
+				mode_lib->vba.DPPCLK_calculated[k]);
+	}
+	mode_lib->vba.GlobalDPPCLK = RoundToDFSGranularityUp(
+			mode_lib->vba.GlobalDPPCLK,
+			mode_lib->vba.DISPCLKDPPCLKVCOSpeed);
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		mode_lib->vba.DPPCLK_calculated[k] = mode_lib->vba.GlobalDPPCLK / 255
+				* dml_ceil(
+						mode_lib->vba.DPPCLK_calculated[k] * 255
+								/ mode_lib->vba.GlobalDPPCLK,
+						1);
+		DTRACE("   dppclk_mhz[%i] (calculated) = %f", k, mode_lib->vba.DPPCLK_calculated[k]);
+	}
+
+	// Urgent Watermark
+	mode_lib->vba.DCCEnabledAnyPlane = false;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k)
+		if (mode_lib->vba.DCCEnable[k])
+			mode_lib->vba.DCCEnabledAnyPlane = true;
+
+	mode_lib->vba.ReturnBandwidthToDCN = dml_min(
+			mode_lib->vba.ReturnBusWidth * mode_lib->vba.DCFCLK,
+			mode_lib->vba.FabricAndDRAMBandwidth * 1000)
+			* mode_lib->vba.PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelDataOnly / 100;
+
+	mode_lib->vba.ReturnBW = mode_lib->vba.ReturnBandwidthToDCN;
+	mode_lib->vba.ReturnBW = adjust_ReturnBW(
+			mode_lib,
+			mode_lib->vba.ReturnBW,
+			mode_lib->vba.DCCEnabledAnyPlane,
+			mode_lib->vba.ReturnBandwidthToDCN);
+
+	// Let's do this calculation again??
+	mode_lib->vba.ReturnBandwidthToDCN = dml_min(
+			mode_lib->vba.ReturnBusWidth * mode_lib->vba.DCFCLK,
+			mode_lib->vba.FabricAndDRAMBandwidth * 1000);
+	mode_lib->vba.ReturnBW = adjust_ReturnBW(
+			mode_lib,
+			mode_lib->vba.ReturnBW,
+			mode_lib->vba.DCCEnabledAnyPlane,
+			mode_lib->vba.ReturnBandwidthToDCN);
+
+	DTRACE("   dcfclk_mhz         = %f", mode_lib->vba.DCFCLK);
+	DTRACE("   return_bw_to_dcn   = %f", mode_lib->vba.ReturnBandwidthToDCN);
+	DTRACE("   return_bus_bw      = %f", mode_lib->vba.ReturnBW);
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		bool MainPlaneDoesODMCombine = false;
+
+		if (mode_lib->vba.SourceScan[k] == dm_horz)
+			mode_lib->vba.SwathWidthSingleDPPY[k] = mode_lib->vba.ViewportWidth[k];
+		else
+			mode_lib->vba.SwathWidthSingleDPPY[k] = mode_lib->vba.ViewportHeight[k];
+
+		if (mode_lib->vba.ODMCombineEnabled[k] == dm_odm_combine_mode_2to1)
+			MainPlaneDoesODMCombine = true;
+		for (j = 0; j < mode_lib->vba.NumberOfActivePlanes; ++j)
+			if (mode_lib->vba.BlendingAndTiming[k] == j
+					&& mode_lib->vba.ODMCombineEnabled[k] == dm_odm_combine_mode_2to1)
+				MainPlaneDoesODMCombine = true;
+
+		if (MainPlaneDoesODMCombine == true)
+			mode_lib->vba.SwathWidthY[k] = dml_min(
+					(double) mode_lib->vba.SwathWidthSingleDPPY[k],
+					dml_round(
+							mode_lib->vba.HActive[k] / 2.0
+									* mode_lib->vba.HRatio[k]));
+		else {
+			if (mode_lib->vba.DPPPerPlane[k] == 0) {
+				mode_lib->vba.SwathWidthY[k] = 0;
+			} else {
+				mode_lib->vba.SwathWidthY[k] = mode_lib->vba.SwathWidthSingleDPPY[k]
+						/ mode_lib->vba.DPPPerPlane[k];
+			}
+		}
+	}
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.SourcePixelFormat[k] == dm_444_64) {
+			mode_lib->vba.BytePerPixelDETY[k] = 8;
+			mode_lib->vba.BytePerPixelDETC[k] = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_32) {
+			mode_lib->vba.BytePerPixelDETY[k] = 4;
+			mode_lib->vba.BytePerPixelDETC[k] = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_16) {
+			mode_lib->vba.BytePerPixelDETY[k] = 2;
+			mode_lib->vba.BytePerPixelDETC[k] = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_8) {
+			mode_lib->vba.BytePerPixelDETY[k] = 1;
+			mode_lib->vba.BytePerPixelDETC[k] = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_8) {
+			mode_lib->vba.BytePerPixelDETY[k] = 1;
+			mode_lib->vba.BytePerPixelDETC[k] = 2;
+		} else { // dm_420_10
+			mode_lib->vba.BytePerPixelDETY[k] = 4.0 / 3.0;
+			mode_lib->vba.BytePerPixelDETC[k] = 8.0 / 3.0;
+		}
+	}
+
+	mode_lib->vba.TotalDataReadBandwidth = 0.0;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		mode_lib->vba.ReadBandwidthPlaneLuma[k] = mode_lib->vba.SwathWidthSingleDPPY[k]
+				* dml_ceil(mode_lib->vba.BytePerPixelDETY[k], 1)
+				/ (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k])
+				* mode_lib->vba.VRatio[k];
+		mode_lib->vba.ReadBandwidthPlaneChroma[k] = mode_lib->vba.SwathWidthSingleDPPY[k]
+				/ 2 * dml_ceil(mode_lib->vba.BytePerPixelDETC[k], 2)
+				/ (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k])
+				* mode_lib->vba.VRatio[k] / 2;
+		DTRACE(
+				"   read_bw[%i] = %fBps",
+				k,
+				mode_lib->vba.ReadBandwidthPlaneLuma[k]
+						+ mode_lib->vba.ReadBandwidthPlaneChroma[k]);
+		mode_lib->vba.TotalDataReadBandwidth += mode_lib->vba.ReadBandwidthPlaneLuma[k]
+				+ mode_lib->vba.ReadBandwidthPlaneChroma[k];
+	}
+
+	mode_lib->vba.TotalDCCActiveDPP = 0;
+	mode_lib->vba.TotalActiveDPP = 0;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		mode_lib->vba.TotalActiveDPP = mode_lib->vba.TotalActiveDPP
+				+ mode_lib->vba.DPPPerPlane[k];
+		if (mode_lib->vba.DCCEnable[k])
+			mode_lib->vba.TotalDCCActiveDPP = mode_lib->vba.TotalDCCActiveDPP
+					+ mode_lib->vba.DPPPerPlane[k];
+	}
+
+	mode_lib->vba.UrgentRoundTripAndOutOfOrderLatency =
+			(mode_lib->vba.RoundTripPingLatencyCycles + 32) / mode_lib->vba.DCFCLK
+					+ mode_lib->vba.UrgentOutOfOrderReturnPerChannelPixelDataOnly
+							* mode_lib->vba.NumberOfChannels
+							/ mode_lib->vba.ReturnBW;
+
+	mode_lib->vba.LastPixelOfLineExtraWatermark = 0;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+			if (mode_lib->vba.VRatio[k] <= 1.0)
+				mode_lib->vba.DisplayPipeLineDeliveryTimeLuma[k] =
+						(double) mode_lib->vba.SwathWidthY[k]
+								* mode_lib->vba.DPPPerPlane[k]
+								/ mode_lib->vba.HRatio[k]
+								/ mode_lib->vba.PixelClock[k];
+			else
+				mode_lib->vba.DisplayPipeLineDeliveryTimeLuma[k] =
+						(double) mode_lib->vba.SwathWidthY[k]
+								/ mode_lib->vba.PSCL_THROUGHPUT_LUMA[k]
+								/ mode_lib->vba.DPPCLK[k];
+
+			if (mode_lib->vba.BytePerPixelDETC[k] == 0)
+				mode_lib->vba.DisplayPipeLineDeliveryTimeChroma[k] = 0.0;
+			else if (mode_lib->vba.VRatio[k] / 2.0 <= 1.0)
+				mode_lib->vba.DisplayPipeLineDeliveryTimeChroma[k] =
+						mode_lib->vba.SwathWidthY[k] / 2.0
+								* mode_lib->vba.DPPPerPlane[k]
+								/ (mode_lib->vba.HRatio[k] / 2.0)
+								/ mode_lib->vba.PixelClock[k];
+			else
+				mode_lib->vba.DisplayPipeLineDeliveryTimeChroma[k] =
+						mode_lib->vba.SwathWidthY[k] / 2.0
+								/ mode_lib->vba.PSCL_THROUGHPUT_CHROMA[k]
+								/ mode_lib->vba.DPPCLK[k];
+		}
+
+	mode_lib->vba.UrgentExtraLatency = mode_lib->vba.UrgentRoundTripAndOutOfOrderLatency
+			+ (mode_lib->vba.TotalActiveDPP * mode_lib->vba.PixelChunkSizeInKByte
+					+ mode_lib->vba.TotalDCCActiveDPP
+							* mode_lib->vba.MetaChunkSize) * 1024.0
+					/ mode_lib->vba.ReturnBW;
+
+	if (mode_lib->vba.GPUVMEnable)
+		mode_lib->vba.UrgentExtraLatency += mode_lib->vba.TotalActiveDPP
+				* mode_lib->vba.PTEGroupSize / mode_lib->vba.ReturnBW;
+
+	mode_lib->vba.UrgentWatermark = mode_lib->vba.UrgentLatencyPixelDataOnly
+			+ mode_lib->vba.LastPixelOfLineExtraWatermark
+			+ mode_lib->vba.UrgentExtraLatency;
+
+	DTRACE("   urgent_extra_latency = %fus", mode_lib->vba.UrgentExtraLatency);
+	DTRACE("   wm_urgent = %fus", mode_lib->vba.UrgentWatermark);
+
+	mode_lib->vba.UrgentLatency = mode_lib->vba.UrgentLatencyPixelDataOnly;
+
+	mode_lib->vba.TotalActiveWriteback = 0;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.WritebackEnable[k])
+			mode_lib->vba.TotalActiveWriteback = mode_lib->vba.TotalActiveWriteback + mode_lib->vba.ActiveWritebacksPerPlane[k];
+	}
+
+	if (mode_lib->vba.TotalActiveWriteback <= 1)
+		mode_lib->vba.WritebackUrgentWatermark = mode_lib->vba.WritebackLatency;
+	else
+		mode_lib->vba.WritebackUrgentWatermark = mode_lib->vba.WritebackLatency
+				+ mode_lib->vba.WritebackChunkSize * 1024.0 / 32
+						/ mode_lib->vba.SOCCLK;
+
+	DTRACE("   wm_wb_urgent = %fus", mode_lib->vba.WritebackUrgentWatermark);
+
+	// NB P-State/DRAM Clock Change Watermark
+	mode_lib->vba.DRAMClockChangeWatermark = mode_lib->vba.DRAMClockChangeLatency
+			+ mode_lib->vba.UrgentWatermark;
+
+	DTRACE("   wm_pstate_change = %fus", mode_lib->vba.DRAMClockChangeWatermark);
+
+	DTRACE("   calculating wb pstate watermark");
+	DTRACE("      total wb outputs %d", mode_lib->vba.TotalActiveWriteback);
+	DTRACE("      socclk frequency %f Mhz", mode_lib->vba.SOCCLK);
+
+	if (mode_lib->vba.TotalActiveWriteback <= 1)
+		mode_lib->vba.WritebackDRAMClockChangeWatermark =
+				mode_lib->vba.DRAMClockChangeLatency
+						+ mode_lib->vba.WritebackLatency;
+	else
+		mode_lib->vba.WritebackDRAMClockChangeWatermark =
+				mode_lib->vba.DRAMClockChangeLatency
+						+ mode_lib->vba.WritebackLatency
+						+ mode_lib->vba.WritebackChunkSize * 1024.0 / 32
+								/ mode_lib->vba.SOCCLK;
+
+	DTRACE("   wm_wb_pstate %fus", mode_lib->vba.WritebackDRAMClockChangeWatermark);
+
+	// Stutter Efficiency
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		mode_lib->vba.LinesInDETY[k] = mode_lib->vba.DETBufferSizeY[k]
+				/ mode_lib->vba.BytePerPixelDETY[k] / mode_lib->vba.SwathWidthY[k];
+		mode_lib->vba.LinesInDETYRoundedDownToSwath[k] = dml_floor(
+				mode_lib->vba.LinesInDETY[k],
+				mode_lib->vba.SwathHeightY[k]);
+		mode_lib->vba.FullDETBufferingTimeY[k] =
+				mode_lib->vba.LinesInDETYRoundedDownToSwath[k]
+						* (mode_lib->vba.HTotal[k]
+								/ mode_lib->vba.PixelClock[k])
+						/ mode_lib->vba.VRatio[k];
+		if (mode_lib->vba.BytePerPixelDETC[k] > 0) {
+			mode_lib->vba.LinesInDETC[k] = mode_lib->vba.DETBufferSizeC[k]
+					/ mode_lib->vba.BytePerPixelDETC[k]
+					/ (mode_lib->vba.SwathWidthY[k] / 2);
+			mode_lib->vba.LinesInDETCRoundedDownToSwath[k] = dml_floor(
+					mode_lib->vba.LinesInDETC[k],
+					mode_lib->vba.SwathHeightC[k]);
+			mode_lib->vba.FullDETBufferingTimeC[k] =
+					mode_lib->vba.LinesInDETCRoundedDownToSwath[k]
+							* (mode_lib->vba.HTotal[k]
+									/ mode_lib->vba.PixelClock[k])
+							/ (mode_lib->vba.VRatio[k] / 2);
+		} else {
+			mode_lib->vba.LinesInDETC[k] = 0;
+			mode_lib->vba.LinesInDETCRoundedDownToSwath[k] = 0;
+			mode_lib->vba.FullDETBufferingTimeC[k] = 999999;
+		}
+	}
+
+	mode_lib->vba.MinFullDETBufferingTime = 999999.0;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.FullDETBufferingTimeY[k]
+				< mode_lib->vba.MinFullDETBufferingTime) {
+			mode_lib->vba.MinFullDETBufferingTime =
+					mode_lib->vba.FullDETBufferingTimeY[k];
+			mode_lib->vba.FrameTimeForMinFullDETBufferingTime =
+					(double) mode_lib->vba.VTotal[k] * mode_lib->vba.HTotal[k]
+							/ mode_lib->vba.PixelClock[k];
+		}
+		if (mode_lib->vba.FullDETBufferingTimeC[k]
+				< mode_lib->vba.MinFullDETBufferingTime) {
+			mode_lib->vba.MinFullDETBufferingTime =
+					mode_lib->vba.FullDETBufferingTimeC[k];
+			mode_lib->vba.FrameTimeForMinFullDETBufferingTime =
+					(double) mode_lib->vba.VTotal[k] * mode_lib->vba.HTotal[k]
+							/ mode_lib->vba.PixelClock[k];
+		}
+	}
+
+	mode_lib->vba.AverageReadBandwidthGBytePerSecond = 0.0;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.DCCEnable[k]) {
+			mode_lib->vba.AverageReadBandwidthGBytePerSecond =
+					mode_lib->vba.AverageReadBandwidthGBytePerSecond
+							+ mode_lib->vba.ReadBandwidthPlaneLuma[k]
+									/ mode_lib->vba.DCCRate[k]
+									/ 1000
+							+ mode_lib->vba.ReadBandwidthPlaneChroma[k]
+									/ mode_lib->vba.DCCRate[k]
+									/ 1000;
+		} else {
+			mode_lib->vba.AverageReadBandwidthGBytePerSecond =
+					mode_lib->vba.AverageReadBandwidthGBytePerSecond
+							+ mode_lib->vba.ReadBandwidthPlaneLuma[k]
+									/ 1000
+							+ mode_lib->vba.ReadBandwidthPlaneChroma[k]
+									/ 1000;
+		}
+		if (mode_lib->vba.DCCEnable[k]) {
+			mode_lib->vba.AverageReadBandwidthGBytePerSecond =
+					mode_lib->vba.AverageReadBandwidthGBytePerSecond
+							+ mode_lib->vba.ReadBandwidthPlaneLuma[k]
+									/ 1000 / 256
+							+ mode_lib->vba.ReadBandwidthPlaneChroma[k]
+									/ 1000 / 256;
+		}
+		if (mode_lib->vba.GPUVMEnable) {
+			mode_lib->vba.AverageReadBandwidthGBytePerSecond =
+					mode_lib->vba.AverageReadBandwidthGBytePerSecond
+							+ mode_lib->vba.ReadBandwidthPlaneLuma[k]
+									/ 1000 / 512
+							+ mode_lib->vba.ReadBandwidthPlaneChroma[k]
+									/ 1000 / 512;
+		}
+	}
+
+	mode_lib->vba.PartOfBurstThatFitsInROB =
+			dml_min(
+					mode_lib->vba.MinFullDETBufferingTime
+							* mode_lib->vba.TotalDataReadBandwidth,
+					mode_lib->vba.ROBBufferSizeInKByte * 1024
+							* mode_lib->vba.TotalDataReadBandwidth
+							/ (mode_lib->vba.AverageReadBandwidthGBytePerSecond
+									* 1000));
+	mode_lib->vba.StutterBurstTime = mode_lib->vba.PartOfBurstThatFitsInROB
+			* (mode_lib->vba.AverageReadBandwidthGBytePerSecond * 1000)
+			/ mode_lib->vba.TotalDataReadBandwidth / mode_lib->vba.ReturnBW
+			+ (mode_lib->vba.MinFullDETBufferingTime
+					* mode_lib->vba.TotalDataReadBandwidth
+					- mode_lib->vba.PartOfBurstThatFitsInROB)
+					/ (mode_lib->vba.DCFCLK * 64);
+	if (mode_lib->vba.TotalActiveWriteback == 0) {
+		mode_lib->vba.StutterEfficiencyNotIncludingVBlank = (1
+				- (mode_lib->vba.SRExitTime + mode_lib->vba.StutterBurstTime)
+						/ mode_lib->vba.MinFullDETBufferingTime) * 100;
+	} else {
+		mode_lib->vba.StutterEfficiencyNotIncludingVBlank = 0;
+	}
+
+	mode_lib->vba.SmallestVBlank = 999999;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.SynchronizedVBlank || mode_lib->vba.NumberOfActivePlanes == 1) {
+			mode_lib->vba.VBlankTime = (double) (mode_lib->vba.VTotal[k]
+					- mode_lib->vba.VActive[k]) * mode_lib->vba.HTotal[k]
+					/ mode_lib->vba.PixelClock[k];
+		} else {
+			mode_lib->vba.VBlankTime = 0;
+		}
+		mode_lib->vba.SmallestVBlank = dml_min(
+				mode_lib->vba.SmallestVBlank,
+				mode_lib->vba.VBlankTime);
+	}
+
+	mode_lib->vba.StutterEfficiency = (mode_lib->vba.StutterEfficiencyNotIncludingVBlank / 100
+			* (mode_lib->vba.FrameTimeForMinFullDETBufferingTime
+					- mode_lib->vba.SmallestVBlank)
+			+ mode_lib->vba.SmallestVBlank)
+			/ mode_lib->vba.FrameTimeForMinFullDETBufferingTime * 100;
+
+	// dml_ml->vba.DCFCLK Deep Sleep
+	mode_lib->vba.DCFCLKDeepSleep = 8.0;
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; k++) {
+		if (mode_lib->vba.BytePerPixelDETC[k] > 0) {
+			mode_lib->vba.DCFCLKDeepSleepPerPlane[k] =
+					dml_max(
+							1.1 * mode_lib->vba.SwathWidthY[k]
+									* dml_ceil(
+											mode_lib->vba.BytePerPixelDETY[k],
+											1) / 32
+									/ mode_lib->vba.DisplayPipeLineDeliveryTimeLuma[k],
+							1.1 * mode_lib->vba.SwathWidthY[k] / 2.0
+									* dml_ceil(
+											mode_lib->vba.BytePerPixelDETC[k],
+											2) / 32
+									/ mode_lib->vba.DisplayPipeLineDeliveryTimeChroma[k]);
+		} else
+			mode_lib->vba.DCFCLKDeepSleepPerPlane[k] = 1.1 * mode_lib->vba.SwathWidthY[k]
+					* dml_ceil(mode_lib->vba.BytePerPixelDETY[k], 1) / 64.0
+					/ mode_lib->vba.DisplayPipeLineDeliveryTimeLuma[k];
+		mode_lib->vba.DCFCLKDeepSleepPerPlane[k] = dml_max(
+				mode_lib->vba.DCFCLKDeepSleepPerPlane[k],
+				mode_lib->vba.PixelClock[k] / 16.0);
+		mode_lib->vba.DCFCLKDeepSleep = dml_max(
+				mode_lib->vba.DCFCLKDeepSleep,
+				mode_lib->vba.DCFCLKDeepSleepPerPlane[k]);
+
+		DTRACE(
+				"   dcfclk_deepsleep_per_plane[%i] = %fMHz",
+				k,
+				mode_lib->vba.DCFCLKDeepSleepPerPlane[k]);
+	}
+
+	DTRACE("   dcfclk_deepsleep_mhz = %fMHz", mode_lib->vba.DCFCLKDeepSleep);
+
+	// Stutter Watermark
+	mode_lib->vba.StutterExitWatermark = mode_lib->vba.SRExitTime
+			+ mode_lib->vba.LastPixelOfLineExtraWatermark
+			+ mode_lib->vba.UrgentExtraLatency + 10 / mode_lib->vba.DCFCLKDeepSleep;
+	mode_lib->vba.StutterEnterPlusExitWatermark = mode_lib->vba.SREnterPlusExitTime
+			+ mode_lib->vba.LastPixelOfLineExtraWatermark
+			+ mode_lib->vba.UrgentExtraLatency;
+
+	DTRACE("   wm_cstate_exit       = %fus", mode_lib->vba.StutterExitWatermark);
+	DTRACE("   wm_cstate_enter_exit = %fus", mode_lib->vba.StutterEnterPlusExitWatermark);
+
+	// Urgent Latency Supported
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		mode_lib->vba.EffectiveDETPlusLBLinesLuma =
+				dml_floor(
+						mode_lib->vba.LinesInDETY[k]
+								+ dml_min(
+										mode_lib->vba.LinesInDETY[k]
+												* mode_lib->vba.DPPCLK[k]
+												* mode_lib->vba.BytePerPixelDETY[k]
+												* mode_lib->vba.PSCL_THROUGHPUT_LUMA[k]
+												/ (mode_lib->vba.ReturnBW
+														/ mode_lib->vba.DPPPerPlane[k]),
+										(double) mode_lib->vba.EffectiveLBLatencyHidingSourceLinesLuma),
+						mode_lib->vba.SwathHeightY[k]);
+
+		mode_lib->vba.UrgentLatencySupportUsLuma = mode_lib->vba.EffectiveDETPlusLBLinesLuma
+				* (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k])
+				/ mode_lib->vba.VRatio[k]
+				- mode_lib->vba.EffectiveDETPlusLBLinesLuma
+						* mode_lib->vba.SwathWidthY[k]
+						* mode_lib->vba.BytePerPixelDETY[k]
+						/ (mode_lib->vba.ReturnBW
+								/ mode_lib->vba.DPPPerPlane[k]);
+
+		if (mode_lib->vba.BytePerPixelDETC[k] > 0) {
+			mode_lib->vba.EffectiveDETPlusLBLinesChroma =
+					dml_floor(
+							mode_lib->vba.LinesInDETC[k]
+									+ dml_min(
+											mode_lib->vba.LinesInDETC[k]
+													* mode_lib->vba.DPPCLK[k]
+													* mode_lib->vba.BytePerPixelDETC[k]
+													* mode_lib->vba.PSCL_THROUGHPUT_CHROMA[k]
+													/ (mode_lib->vba.ReturnBW
+															/ mode_lib->vba.DPPPerPlane[k]),
+											(double) mode_lib->vba.EffectiveLBLatencyHidingSourceLinesChroma),
+							mode_lib->vba.SwathHeightC[k]);
+			mode_lib->vba.UrgentLatencySupportUsChroma =
+					mode_lib->vba.EffectiveDETPlusLBLinesChroma
+							* (mode_lib->vba.HTotal[k]
+									/ mode_lib->vba.PixelClock[k])
+							/ (mode_lib->vba.VRatio[k] / 2)
+							- mode_lib->vba.EffectiveDETPlusLBLinesChroma
+									* (mode_lib->vba.SwathWidthY[k]
+											/ 2)
+									* mode_lib->vba.BytePerPixelDETC[k]
+									/ (mode_lib->vba.ReturnBW
+											/ mode_lib->vba.DPPPerPlane[k]);
+			mode_lib->vba.UrgentLatencySupportUs[k] = dml_min(
+					mode_lib->vba.UrgentLatencySupportUsLuma,
+					mode_lib->vba.UrgentLatencySupportUsChroma);
+		} else {
+			mode_lib->vba.UrgentLatencySupportUs[k] =
+					mode_lib->vba.UrgentLatencySupportUsLuma;
+		}
+	}
+
+	mode_lib->vba.MinUrgentLatencySupportUs = 999999;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		mode_lib->vba.MinUrgentLatencySupportUs = dml_min(
+				mode_lib->vba.MinUrgentLatencySupportUs,
+				mode_lib->vba.UrgentLatencySupportUs[k]);
+	}
+
+	// Non-Urgent Latency Tolerance
+	mode_lib->vba.NonUrgentLatencyTolerance = mode_lib->vba.MinUrgentLatencySupportUs
+			- mode_lib->vba.UrgentWatermark;
+
+	// DSCCLK
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if ((mode_lib->vba.BlendingAndTiming[k] != k) || !mode_lib->vba.DSCEnabled[k]) {
+			mode_lib->vba.DSCCLK_calculated[k] = 0.0;
+		} else {
+			if (mode_lib->vba.OutputFormat[k] == dm_420
+					|| mode_lib->vba.OutputFormat[k] == dm_n422)
+				mode_lib->vba.DSCFormatFactor = 2;
+			else
+				mode_lib->vba.DSCFormatFactor = 1;
+			if (mode_lib->vba.ODMCombineEnabled[k])
+				mode_lib->vba.DSCCLK_calculated[k] =
+						mode_lib->vba.PixelClockBackEnd[k] / 6
+								/ mode_lib->vba.DSCFormatFactor
+								/ (1
+										- mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading
+												/ 100);
+			else
+				mode_lib->vba.DSCCLK_calculated[k] =
+						mode_lib->vba.PixelClockBackEnd[k] / 3
+								/ mode_lib->vba.DSCFormatFactor
+								/ (1
+										- mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading
+												/ 100);
+		}
+	}
+
+	// DSC Delay
+	// TODO
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		double bpp = mode_lib->vba.OutputBpp[k];
+		unsigned int slices = mode_lib->vba.NumberOfDSCSlices[k];
+
+		if (mode_lib->vba.DSCEnabled[k] && bpp != 0) {
+			if (!mode_lib->vba.ODMCombineEnabled[k]) {
+				mode_lib->vba.DSCDelay[k] =
+						dscceComputeDelay(
+								mode_lib->vba.DSCInputBitPerComponent[k],
+								bpp,
+								dml_ceil(
+										(double) mode_lib->vba.HActive[k]
+												/ mode_lib->vba.NumberOfDSCSlices[k],
+										1),
+								slices,
+								mode_lib->vba.OutputFormat[k])
+								+ dscComputeDelay(
+										mode_lib->vba.OutputFormat[k]);
+			} else {
+				mode_lib->vba.DSCDelay[k] =
+						2
+								* (dscceComputeDelay(
+										mode_lib->vba.DSCInputBitPerComponent[k],
+										bpp,
+										dml_ceil(
+												(double) mode_lib->vba.HActive[k]
+														/ mode_lib->vba.NumberOfDSCSlices[k],
+												1),
+										slices / 2.0,
+										mode_lib->vba.OutputFormat[k])
+										+ dscComputeDelay(
+												mode_lib->vba.OutputFormat[k]));
+			}
+			mode_lib->vba.DSCDelay[k] = mode_lib->vba.DSCDelay[k]
+					* mode_lib->vba.PixelClock[k]
+					/ mode_lib->vba.PixelClockBackEnd[k];
+		} else {
+			mode_lib->vba.DSCDelay[k] = 0;
+		}
+	}
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k)
+		for (j = 0; j < mode_lib->vba.NumberOfActivePlanes; ++j) // NumberOfPlanes
+			if (j != k && mode_lib->vba.BlendingAndTiming[k] == j
+					&& mode_lib->vba.DSCEnabled[j])
+				mode_lib->vba.DSCDelay[k] = mode_lib->vba.DSCDelay[j];
+
+	// Prefetch
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		unsigned int PDEAndMetaPTEBytesFrameY;
+		unsigned int PixelPTEBytesPerRowY;
+		unsigned int MetaRowByteY;
+		unsigned int MetaRowByteC;
+		unsigned int PDEAndMetaPTEBytesFrameC;
+		unsigned int PixelPTEBytesPerRowC;
+
+		Calculate256BBlockSizes(
+				mode_lib->vba.SourcePixelFormat[k],
+				mode_lib->vba.SurfaceTiling[k],
+				dml_ceil(mode_lib->vba.BytePerPixelDETY[k], 1),
+				dml_ceil(mode_lib->vba.BytePerPixelDETC[k], 2),
+				&mode_lib->vba.BlockHeight256BytesY[k],
+				&mode_lib->vba.BlockHeight256BytesC[k],
+				&mode_lib->vba.BlockWidth256BytesY[k],
+				&mode_lib->vba.BlockWidth256BytesC[k]);
+		PDEAndMetaPTEBytesFrameY = CalculateVMAndRowBytes(
+				mode_lib,
+				mode_lib->vba.DCCEnable[k],
+				mode_lib->vba.BlockHeight256BytesY[k],
+				mode_lib->vba.BlockWidth256BytesY[k],
+				mode_lib->vba.SourcePixelFormat[k],
+				mode_lib->vba.SurfaceTiling[k],
+				dml_ceil(mode_lib->vba.BytePerPixelDETY[k], 1),
+				mode_lib->vba.SourceScan[k],
+				mode_lib->vba.ViewportWidth[k],
+				mode_lib->vba.ViewportHeight[k],
+				mode_lib->vba.SwathWidthY[k],
+				mode_lib->vba.GPUVMEnable,
+				mode_lib->vba.VMMPageSize,
+				mode_lib->vba.PTEBufferSizeInRequestsLuma,
+				mode_lib->vba.PDEProcessingBufIn64KBReqs,
+				mode_lib->vba.PitchY[k],
+				mode_lib->vba.DCCMetaPitchY[k],
+				&mode_lib->vba.MacroTileWidthY[k],
+				&MetaRowByteY,
+				&PixelPTEBytesPerRowY,
+				&mode_lib->vba.PTEBufferSizeNotExceeded[mode_lib->vba.VoltageLevel][0],
+				&mode_lib->vba.dpte_row_height[k],
+				&mode_lib->vba.meta_row_height[k]);
+		mode_lib->vba.PrefetchSourceLinesY[k] = CalculatePrefetchSourceLines(
+				mode_lib,
+				mode_lib->vba.VRatio[k],
+				mode_lib->vba.vtaps[k],
+				mode_lib->vba.Interlace[k],
+				mode_lib->vba.ProgressiveToInterlaceUnitInOPP,
+				mode_lib->vba.SwathHeightY[k],
+				mode_lib->vba.ViewportYStartY[k],
+				&mode_lib->vba.VInitPreFillY[k],
+				&mode_lib->vba.MaxNumSwathY[k]);
+
+		if ((mode_lib->vba.SourcePixelFormat[k] != dm_444_64
+				&& mode_lib->vba.SourcePixelFormat[k] != dm_444_32
+				&& mode_lib->vba.SourcePixelFormat[k] != dm_444_16
+				&& mode_lib->vba.SourcePixelFormat[k] != dm_444_8)) {
+			PDEAndMetaPTEBytesFrameC =
+					CalculateVMAndRowBytes(
+							mode_lib,
+							mode_lib->vba.DCCEnable[k],
+							mode_lib->vba.BlockHeight256BytesC[k],
+							mode_lib->vba.BlockWidth256BytesC[k],
+							mode_lib->vba.SourcePixelFormat[k],
+							mode_lib->vba.SurfaceTiling[k],
+							dml_ceil(
+									mode_lib->vba.BytePerPixelDETC[k],
+									2),
+							mode_lib->vba.SourceScan[k],
+							mode_lib->vba.ViewportWidth[k] / 2,
+							mode_lib->vba.ViewportHeight[k] / 2,
+							mode_lib->vba.SwathWidthY[k] / 2,
+							mode_lib->vba.GPUVMEnable,
+							mode_lib->vba.VMMPageSize,
+							mode_lib->vba.PTEBufferSizeInRequestsLuma,
+							mode_lib->vba.PDEProcessingBufIn64KBReqs,
+							mode_lib->vba.PitchC[k],
+							0,
+							&mode_lib->vba.MacroTileWidthC[k],
+							&MetaRowByteC,
+							&PixelPTEBytesPerRowC,
+							&mode_lib->vba.PTEBufferSizeNotExceeded[mode_lib->vba.VoltageLevel][0],
+							&mode_lib->vba.dpte_row_height_chroma[k],
+							&mode_lib->vba.meta_row_height_chroma[k]);
+			mode_lib->vba.PrefetchSourceLinesC[k] = CalculatePrefetchSourceLines(
+					mode_lib,
+					mode_lib->vba.VRatio[k] / 2,
+					mode_lib->vba.VTAPsChroma[k],
+					mode_lib->vba.Interlace[k],
+					mode_lib->vba.ProgressiveToInterlaceUnitInOPP,
+					mode_lib->vba.SwathHeightC[k],
+					mode_lib->vba.ViewportYStartC[k],
+					&mode_lib->vba.VInitPreFillC[k],
+					&mode_lib->vba.MaxNumSwathC[k]);
+		} else {
+			PixelPTEBytesPerRowC = 0;
+			PDEAndMetaPTEBytesFrameC = 0;
+			MetaRowByteC = 0;
+			mode_lib->vba.MaxNumSwathC[k] = 0;
+			mode_lib->vba.PrefetchSourceLinesC[k] = 0;
+		}
+
+		mode_lib->vba.PixelPTEBytesPerRow[k] = PixelPTEBytesPerRowY + PixelPTEBytesPerRowC;
+		mode_lib->vba.PDEAndMetaPTEBytesFrame[k] = PDEAndMetaPTEBytesFrameY
+				+ PDEAndMetaPTEBytesFrameC;
+		mode_lib->vba.MetaRowByte[k] = MetaRowByteY + MetaRowByteC;
+
+		CalculateActiveRowBandwidth(
+				mode_lib->vba.GPUVMEnable,
+				mode_lib->vba.SourcePixelFormat[k],
+				mode_lib->vba.VRatio[k],
+				mode_lib->vba.DCCEnable[k],
+				mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k],
+				MetaRowByteY,
+				MetaRowByteC,
+				mode_lib->vba.meta_row_height[k],
+				mode_lib->vba.meta_row_height_chroma[k],
+				PixelPTEBytesPerRowY,
+				PixelPTEBytesPerRowC,
+				mode_lib->vba.dpte_row_height[k],
+				mode_lib->vba.dpte_row_height_chroma[k],
+				&mode_lib->vba.meta_row_bw[k],
+				&mode_lib->vba.dpte_row_bw[k],
+				&mode_lib->vba.qual_row_bw[k]);
+	}
+
+	mode_lib->vba.TCalc = 24.0 / mode_lib->vba.DCFCLKDeepSleep;
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.BlendingAndTiming[k] == k) {
+			if (mode_lib->vba.WritebackEnable[k] == true) {
+				mode_lib->vba.WritebackDelay[mode_lib->vba.VoltageLevel][k] =
+						mode_lib->vba.WritebackLatency
+								+ CalculateWriteBackDelay(
+										mode_lib->vba.WritebackPixelFormat[k],
+										mode_lib->vba.WritebackHRatio[k],
+										mode_lib->vba.WritebackVRatio[k],
+										mode_lib->vba.WritebackLumaHTaps[k],
+										mode_lib->vba.WritebackLumaVTaps[k],
+										mode_lib->vba.WritebackChromaHTaps[k],
+										mode_lib->vba.WritebackChromaVTaps[k],
+										mode_lib->vba.WritebackDestinationWidth[k])
+										/ mode_lib->vba.DISPCLK;
+			} else
+				mode_lib->vba.WritebackDelay[mode_lib->vba.VoltageLevel][k] = 0;
+			for (j = 0; j < mode_lib->vba.NumberOfActivePlanes; ++j) {
+				if (mode_lib->vba.BlendingAndTiming[j] == k
+						&& mode_lib->vba.WritebackEnable[j] == true) {
+					mode_lib->vba.WritebackDelay[mode_lib->vba.VoltageLevel][k] =
+							dml_max(
+									mode_lib->vba.WritebackDelay[mode_lib->vba.VoltageLevel][k],
+									mode_lib->vba.WritebackLatency
+											+ CalculateWriteBackDelay(
+													mode_lib->vba.WritebackPixelFormat[j],
+													mode_lib->vba.WritebackHRatio[j],
+													mode_lib->vba.WritebackVRatio[j],
+													mode_lib->vba.WritebackLumaHTaps[j],
+													mode_lib->vba.WritebackLumaVTaps[j],
+													mode_lib->vba.WritebackChromaHTaps[j],
+													mode_lib->vba.WritebackChromaVTaps[j],
+													mode_lib->vba.WritebackDestinationWidth[j])
+													/ mode_lib->vba.DISPCLK);
+				}
+			}
+		}
+	}
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k)
+		for (j = 0; j < mode_lib->vba.NumberOfActivePlanes; ++j)
+			if (mode_lib->vba.BlendingAndTiming[k] == j)
+				mode_lib->vba.WritebackDelay[mode_lib->vba.VoltageLevel][k] =
+						mode_lib->vba.WritebackDelay[mode_lib->vba.VoltageLevel][j];
+
+	mode_lib->vba.VStartupLines = 13;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		mode_lib->vba.MaxVStartupLines[k] =
+				mode_lib->vba.VTotal[k] - mode_lib->vba.VActive[k]
+						- dml_max(
+								1.0,
+								dml_ceil(
+										mode_lib->vba.WritebackDelay[mode_lib->vba.VoltageLevel][k]
+												/ (mode_lib->vba.HTotal[k]
+														/ mode_lib->vba.PixelClock[k]),
+										1));
+	}
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k)
+		mode_lib->vba.MaximumMaxVStartupLines = dml_max(
+				mode_lib->vba.MaximumMaxVStartupLines,
+				mode_lib->vba.MaxVStartupLines[k]);
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		mode_lib->vba.cursor_bw[k] = 0.0;
+		for (j = 0; j < mode_lib->vba.NumberOfCursors[k]; ++j)
+			mode_lib->vba.cursor_bw[k] += mode_lib->vba.CursorWidth[k][j]
+					* mode_lib->vba.CursorBPP[k][j] / 8.0
+					/ (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k])
+					* mode_lib->vba.VRatio[k];
+	}
+
+	do {
+		double MaxTotalRDBandwidth = 0;
+		bool DestinationLineTimesForPrefetchLessThan2 = false;
+		bool VRatioPrefetchMoreThan4 = false;
+		bool prefetch_vm_bw_valid = true;
+		bool prefetch_row_bw_valid = true;
+		double TWait = CalculateTWait(
+				mode_lib->vba.PrefetchMode[mode_lib->vba.VoltageLevel][mode_lib->vba.maxMpcComb],
+				mode_lib->vba.DRAMClockChangeLatency,
+				mode_lib->vba.UrgentLatencyPixelDataOnly,
+				mode_lib->vba.SREnterPlusExitTime);
+
+		for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+			if (mode_lib->vba.XFCEnabled[k] == true) {
+				mode_lib->vba.XFCRemoteSurfaceFlipDelay =
+						CalculateRemoteSurfaceFlipDelay(
+								mode_lib,
+								mode_lib->vba.VRatio[k],
+								mode_lib->vba.SwathWidthY[k],
+								dml_ceil(
+										mode_lib->vba.BytePerPixelDETY[k],
+										1),
+								mode_lib->vba.HTotal[k]
+										/ mode_lib->vba.PixelClock[k],
+								mode_lib->vba.XFCTSlvVupdateOffset,
+								mode_lib->vba.XFCTSlvVupdateWidth,
+								mode_lib->vba.XFCTSlvVreadyOffset,
+								mode_lib->vba.XFCXBUFLatencyTolerance,
+								mode_lib->vba.XFCFillBWOverhead,
+								mode_lib->vba.XFCSlvChunkSize,
+								mode_lib->vba.XFCBusTransportTime,
+								mode_lib->vba.TCalc,
+								TWait,
+								&mode_lib->vba.SrcActiveDrainRate,
+								&mode_lib->vba.TInitXFill,
+								&mode_lib->vba.TslvChk);
+			} else {
+				mode_lib->vba.XFCRemoteSurfaceFlipDelay = 0;
+			}
+
+			CalculateDelayAfterScaler(mode_lib, mode_lib->vba.ReturnBW, mode_lib->vba.ReadBandwidthPlaneLuma[k], mode_lib->vba.ReadBandwidthPlaneChroma[k], mode_lib->vba.TotalDataReadBandwidth,
+					mode_lib->vba.DisplayPipeLineDeliveryTimeLuma[k], mode_lib->vba.DisplayPipeLineDeliveryTimeChroma[k],
+					mode_lib->vba.DPPCLK[k], mode_lib->vba.DISPCLK, mode_lib->vba.PixelClock[k], mode_lib->vba.DSCDelay[k], mode_lib->vba.DPPPerPlane[k], mode_lib->vba.ScalerEnabled[k], mode_lib->vba.NumberOfCursors[k],
+					mode_lib->vba.DPPCLKDelaySubtotal, mode_lib->vba.DPPCLKDelaySCL, mode_lib->vba.DPPCLKDelaySCLLBOnly, mode_lib->vba.DPPCLKDelayCNVCFormater, mode_lib->vba.DPPCLKDelayCNVCCursor, mode_lib->vba.DISPCLKDelaySubtotal,
+					mode_lib->vba.SwathWidthY[k] / mode_lib->vba.HRatio[k], mode_lib->vba.OutputFormat[k], mode_lib->vba.HTotal[k],
+					mode_lib->vba.SwathWidthSingleDPPY[k], mode_lib->vba.BytePerPixelDETY[k], mode_lib->vba.BytePerPixelDETC[k], mode_lib->vba.SwathHeightY[k], mode_lib->vba.SwathHeightC[k], mode_lib->vba.Interlace[k],
+					mode_lib->vba.ProgressiveToInterlaceUnitInOPP, &mode_lib->vba.DSTXAfterScaler[k], &mode_lib->vba.DSTYAfterScaler[k]);
+
+			mode_lib->vba.ErrorResult[k] =
+					CalculatePrefetchSchedule(
+							mode_lib,
+							mode_lib->vba.DPPCLK[k],
+							mode_lib->vba.DISPCLK,
+							mode_lib->vba.PixelClock[k],
+							mode_lib->vba.DCFCLKDeepSleep,
+							mode_lib->vba.DPPPerPlane[k],
+							mode_lib->vba.NumberOfCursors[k],
+							mode_lib->vba.VTotal[k]
+									- mode_lib->vba.VActive[k],
+							mode_lib->vba.HTotal[k],
+							mode_lib->vba.MaxInterDCNTileRepeaters,
+							dml_min(
+									mode_lib->vba.VStartupLines,
+									mode_lib->vba.MaxVStartupLines[k]),
+							mode_lib->vba.GPUVMMaxPageTableLevels,
+							mode_lib->vba.GPUVMEnable,
+							mode_lib->vba.DynamicMetadataEnable[k],
+							mode_lib->vba.DynamicMetadataLinesBeforeActiveRequired[k],
+							mode_lib->vba.DynamicMetadataTransmittedBytes[k],
+							mode_lib->vba.DCCEnable[k],
+							mode_lib->vba.UrgentLatencyPixelDataOnly,
+							mode_lib->vba.UrgentExtraLatency,
+							mode_lib->vba.TCalc,
+							mode_lib->vba.PDEAndMetaPTEBytesFrame[k],
+							mode_lib->vba.MetaRowByte[k],
+							mode_lib->vba.PixelPTEBytesPerRow[k],
+							mode_lib->vba.PrefetchSourceLinesY[k],
+							mode_lib->vba.SwathWidthY[k],
+							mode_lib->vba.BytePerPixelDETY[k],
+							mode_lib->vba.VInitPreFillY[k],
+							mode_lib->vba.MaxNumSwathY[k],
+							mode_lib->vba.PrefetchSourceLinesC[k],
+							mode_lib->vba.BytePerPixelDETC[k],
+							mode_lib->vba.VInitPreFillC[k],
+							mode_lib->vba.MaxNumSwathC[k],
+							mode_lib->vba.SwathHeightY[k],
+							mode_lib->vba.SwathHeightC[k],
+							TWait,
+							mode_lib->vba.XFCEnabled[k],
+							mode_lib->vba.XFCRemoteSurfaceFlipDelay,
+							mode_lib->vba.Interlace[k],
+							mode_lib->vba.ProgressiveToInterlaceUnitInOPP,
+							mode_lib->vba.DSTXAfterScaler[k],
+							mode_lib->vba.DSTYAfterScaler[k],
+							&mode_lib->vba.DestinationLinesForPrefetch[k],
+							&mode_lib->vba.PrefetchBandwidth[k],
+							&mode_lib->vba.DestinationLinesToRequestVMInVBlank[k],
+							&mode_lib->vba.DestinationLinesToRequestRowInVBlank[k],
+							&mode_lib->vba.VRatioPrefetchY[k],
+							&mode_lib->vba.VRatioPrefetchC[k],
+							&mode_lib->vba.RequiredPrefetchPixDataBWLuma[k],
+							&mode_lib->vba.Tno_bw[k],
+							&mode_lib->vba.VUpdateOffsetPix[k],
+							&mode_lib->vba.VUpdateWidthPix[k],
+							&mode_lib->vba.VReadyOffsetPix[k]);
+
+			if (mode_lib->vba.BlendingAndTiming[k] == k) {
+				mode_lib->vba.VStartup[k] = dml_min(
+						mode_lib->vba.VStartupLines,
+						mode_lib->vba.MaxVStartupLines[k]);
+				if (mode_lib->vba.VStartupRequiredWhenNotEnoughTimeForDynamicMetadata
+						!= 0) {
+					mode_lib->vba.VStartup[k] =
+							mode_lib->vba.VStartupRequiredWhenNotEnoughTimeForDynamicMetadata;
+				}
+			} else {
+				mode_lib->vba.VStartup[k] =
+						dml_min(
+								mode_lib->vba.VStartupLines,
+								mode_lib->vba.MaxVStartupLines[mode_lib->vba.BlendingAndTiming[k]]);
+			}
+		}
+
+		for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+
+			if (mode_lib->vba.PDEAndMetaPTEBytesFrame[k] == 0)
+				mode_lib->vba.prefetch_vm_bw[k] = 0;
+			else if (mode_lib->vba.DestinationLinesToRequestVMInVBlank[k] > 0) {
+				mode_lib->vba.prefetch_vm_bw[k] =
+						(double) mode_lib->vba.PDEAndMetaPTEBytesFrame[k]
+								/ (mode_lib->vba.DestinationLinesToRequestVMInVBlank[k]
+										* mode_lib->vba.HTotal[k]
+										/ mode_lib->vba.PixelClock[k]);
+			} else {
+				mode_lib->vba.prefetch_vm_bw[k] = 0;
+				prefetch_vm_bw_valid = false;
+			}
+			if (mode_lib->vba.MetaRowByte[k] + mode_lib->vba.PixelPTEBytesPerRow[k]
+					== 0)
+				mode_lib->vba.prefetch_row_bw[k] = 0;
+			else if (mode_lib->vba.DestinationLinesToRequestRowInVBlank[k] > 0) {
+				mode_lib->vba.prefetch_row_bw[k] =
+						(double) (mode_lib->vba.MetaRowByte[k]
+								+ mode_lib->vba.PixelPTEBytesPerRow[k])
+								/ (mode_lib->vba.DestinationLinesToRequestRowInVBlank[k]
+										* mode_lib->vba.HTotal[k]
+										/ mode_lib->vba.PixelClock[k]);
+			} else {
+				mode_lib->vba.prefetch_row_bw[k] = 0;
+				prefetch_row_bw_valid = false;
+			}
+
+			MaxTotalRDBandwidth =
+					MaxTotalRDBandwidth + mode_lib->vba.cursor_bw[k]
+							+ dml_max(
+									mode_lib->vba.prefetch_vm_bw[k],
+									dml_max(
+											mode_lib->vba.prefetch_row_bw[k],
+											dml_max(
+													mode_lib->vba.ReadBandwidthPlaneLuma[k]
+															+ mode_lib->vba.ReadBandwidthPlaneChroma[k],
+													mode_lib->vba.RequiredPrefetchPixDataBWLuma[k])
+													+ mode_lib->vba.meta_row_bw[k]
+													+ mode_lib->vba.dpte_row_bw[k]));
+
+			if (mode_lib->vba.DestinationLinesForPrefetch[k] < 2)
+				DestinationLineTimesForPrefetchLessThan2 = true;
+			if (mode_lib->vba.VRatioPrefetchY[k] > 4
+					|| mode_lib->vba.VRatioPrefetchC[k] > 4)
+				VRatioPrefetchMoreThan4 = true;
+		}
+
+		if (MaxTotalRDBandwidth <= mode_lib->vba.ReturnBW && prefetch_vm_bw_valid
+				&& prefetch_row_bw_valid && !VRatioPrefetchMoreThan4
+				&& !DestinationLineTimesForPrefetchLessThan2)
+			mode_lib->vba.PrefetchModeSupported = true;
+		else {
+			mode_lib->vba.PrefetchModeSupported = false;
+			dml_print(
+					"DML: CalculatePrefetchSchedule ***failed***. Bandwidth violation. Results are NOT valid\n");
+		}
+
+		if (mode_lib->vba.PrefetchModeSupported == true) {
+			double final_flip_bw[DC__NUM_DPP__MAX];
+			unsigned int ImmediateFlipBytes[DC__NUM_DPP__MAX];
+			double total_dcn_read_bw_with_flip = 0;
+
+			mode_lib->vba.BandwidthAvailableForImmediateFlip = mode_lib->vba.ReturnBW;
+			for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+				mode_lib->vba.BandwidthAvailableForImmediateFlip =
+						mode_lib->vba.BandwidthAvailableForImmediateFlip
+								- mode_lib->vba.cursor_bw[k]
+								- dml_max(
+										mode_lib->vba.ReadBandwidthPlaneLuma[k]
+												+ mode_lib->vba.ReadBandwidthPlaneChroma[k]
+												+ mode_lib->vba.qual_row_bw[k],
+										mode_lib->vba.PrefetchBandwidth[k]);
+			}
+
+			for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+				ImmediateFlipBytes[k] = 0;
+				if ((mode_lib->vba.SourcePixelFormat[k] != dm_420_8
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_420_10)) {
+					ImmediateFlipBytes[k] =
+							mode_lib->vba.PDEAndMetaPTEBytesFrame[k]
+									+ mode_lib->vba.MetaRowByte[k]
+									+ mode_lib->vba.PixelPTEBytesPerRow[k];
+				}
+			}
+			mode_lib->vba.TotImmediateFlipBytes = 0;
+			for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+				if ((mode_lib->vba.SourcePixelFormat[k] != dm_420_8
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_420_10)) {
+					mode_lib->vba.TotImmediateFlipBytes =
+							mode_lib->vba.TotImmediateFlipBytes
+									+ ImmediateFlipBytes[k];
+				}
+			}
+			for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+				CalculateFlipSchedule(
+						mode_lib,
+						mode_lib->vba.UrgentExtraLatency,
+						mode_lib->vba.UrgentLatencyPixelDataOnly,
+						mode_lib->vba.GPUVMMaxPageTableLevels,
+						mode_lib->vba.GPUVMEnable,
+						mode_lib->vba.BandwidthAvailableForImmediateFlip,
+						mode_lib->vba.TotImmediateFlipBytes,
+						mode_lib->vba.SourcePixelFormat[k],
+						ImmediateFlipBytes[k],
+						mode_lib->vba.HTotal[k]
+								/ mode_lib->vba.PixelClock[k],
+						mode_lib->vba.VRatio[k],
+						mode_lib->vba.Tno_bw[k],
+						mode_lib->vba.PDEAndMetaPTEBytesFrame[k],
+						mode_lib->vba.MetaRowByte[k],
+						mode_lib->vba.PixelPTEBytesPerRow[k],
+						mode_lib->vba.DCCEnable[k],
+						mode_lib->vba.dpte_row_height[k],
+						mode_lib->vba.meta_row_height[k],
+						mode_lib->vba.qual_row_bw[k],
+						&mode_lib->vba.DestinationLinesToRequestVMInImmediateFlip[k],
+						&mode_lib->vba.DestinationLinesToRequestRowInImmediateFlip[k],
+						&final_flip_bw[k],
+						&mode_lib->vba.ImmediateFlipSupportedForPipe[k]);
+			}
+			for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+				total_dcn_read_bw_with_flip =
+						total_dcn_read_bw_with_flip
+								+ mode_lib->vba.cursor_bw[k]
+								+ dml_max(
+										mode_lib->vba.prefetch_vm_bw[k],
+										dml_max(
+												mode_lib->vba.prefetch_row_bw[k],
+												final_flip_bw[k]
+														+ dml_max(
+																mode_lib->vba.ReadBandwidthPlaneLuma[k]
+																		+ mode_lib->vba.ReadBandwidthPlaneChroma[k],
+																mode_lib->vba.RequiredPrefetchPixDataBWLuma[k])));
+			}
+			mode_lib->vba.ImmediateFlipSupported = true;
+			if (total_dcn_read_bw_with_flip > mode_lib->vba.ReturnBW) {
+				mode_lib->vba.ImmediateFlipSupported = false;
+			}
+			for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+				if (mode_lib->vba.ImmediateFlipSupportedForPipe[k] == false) {
+					mode_lib->vba.ImmediateFlipSupported = false;
+				}
+			}
+		} else {
+			mode_lib->vba.ImmediateFlipSupported = false;
+		}
+
+		for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+			if (mode_lib->vba.ErrorResult[k]) {
+				mode_lib->vba.PrefetchModeSupported = false;
+				dml_print(
+						"DML: CalculatePrefetchSchedule ***failed***. Prefetch schedule violation. Results are NOT valid\n");
+			}
+		}
+
+		mode_lib->vba.VStartupLines = mode_lib->vba.VStartupLines + 1;
+	} while (!((mode_lib->vba.PrefetchModeSupported
+			&& (!mode_lib->vba.ImmediateFlipSupport
+					|| mode_lib->vba.ImmediateFlipSupported))
+			|| mode_lib->vba.MaximumMaxVStartupLines < mode_lib->vba.VStartupLines));
+
+	//Display Pipeline Delivery Time in Prefetch
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.VRatioPrefetchY[k] <= 1) {
+			mode_lib->vba.DisplayPipeLineDeliveryTimeLumaPrefetch[k] =
+					mode_lib->vba.SwathWidthY[k] * mode_lib->vba.DPPPerPlane[k]
+							/ mode_lib->vba.HRatio[k]
+							/ mode_lib->vba.PixelClock[k];
+		} else {
+			mode_lib->vba.DisplayPipeLineDeliveryTimeLumaPrefetch[k] =
+					mode_lib->vba.SwathWidthY[k]
+							/ mode_lib->vba.PSCL_THROUGHPUT_LUMA[k]
+							/ mode_lib->vba.DPPCLK[k];
+		}
+		if (mode_lib->vba.BytePerPixelDETC[k] == 0) {
+			mode_lib->vba.DisplayPipeLineDeliveryTimeChromaPrefetch[k] = 0;
+		} else {
+			if (mode_lib->vba.VRatioPrefetchC[k] <= 1) {
+				mode_lib->vba.DisplayPipeLineDeliveryTimeChromaPrefetch[k] =
+						mode_lib->vba.SwathWidthY[k]
+								* mode_lib->vba.DPPPerPlane[k]
+								/ mode_lib->vba.HRatio[k]
+								/ mode_lib->vba.PixelClock[k];
+			} else {
+				mode_lib->vba.DisplayPipeLineDeliveryTimeChromaPrefetch[k] =
+						mode_lib->vba.SwathWidthY[k]
+								/ mode_lib->vba.PSCL_THROUGHPUT_LUMA[k]
+								/ mode_lib->vba.DPPCLK[k];
+			}
+		}
+	}
+
+	// Min TTUVBlank
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.PrefetchMode[mode_lib->vba.VoltageLevel][mode_lib->vba.maxMpcComb] == 0) {
+			mode_lib->vba.AllowDRAMClockChangeDuringVBlank[k] = true;
+			mode_lib->vba.AllowDRAMSelfRefreshDuringVBlank[k] = true;
+			mode_lib->vba.MinTTUVBlank[k] = dml_max(
+					mode_lib->vba.DRAMClockChangeWatermark,
+					dml_max(
+							mode_lib->vba.StutterEnterPlusExitWatermark,
+							mode_lib->vba.UrgentWatermark));
+		} else if (mode_lib->vba.PrefetchMode[mode_lib->vba.VoltageLevel][mode_lib->vba.maxMpcComb] == 1) {
+			mode_lib->vba.AllowDRAMClockChangeDuringVBlank[k] = false;
+			mode_lib->vba.AllowDRAMSelfRefreshDuringVBlank[k] = true;
+			mode_lib->vba.MinTTUVBlank[k] = dml_max(
+					mode_lib->vba.StutterEnterPlusExitWatermark,
+					mode_lib->vba.UrgentWatermark);
+		} else {
+			mode_lib->vba.AllowDRAMClockChangeDuringVBlank[k] = false;
+			mode_lib->vba.AllowDRAMSelfRefreshDuringVBlank[k] = false;
+			mode_lib->vba.MinTTUVBlank[k] = mode_lib->vba.UrgentWatermark;
+		}
+		if (!mode_lib->vba.DynamicMetadataEnable[k])
+			mode_lib->vba.MinTTUVBlank[k] = mode_lib->vba.TCalc
+					+ mode_lib->vba.MinTTUVBlank[k];
+	}
+
+	// DCC Configuration
+	mode_lib->vba.ActiveDPPs = 0;
+	// NB P-State/DRAM Clock Change Support
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		mode_lib->vba.ActiveDPPs = mode_lib->vba.ActiveDPPs + mode_lib->vba.DPPPerPlane[k];
+	}
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		double EffectiveLBLatencyHidingY;
+		double EffectiveLBLatencyHidingC;
+		double DPPOutputBufferLinesY;
+		double DPPOutputBufferLinesC;
+		double DPPOPPBufferingY;
+		double MaxDETBufferingTimeY;
+		double ActiveDRAMClockChangeLatencyMarginY;
+
+		mode_lib->vba.LBLatencyHidingSourceLinesY =
+				dml_min(
+						mode_lib->vba.MaxLineBufferLines,
+						(unsigned int) dml_floor(
+								(double) mode_lib->vba.LineBufferSize
+										/ mode_lib->vba.LBBitPerPixel[k]
+										/ (mode_lib->vba.SwathWidthY[k]
+												/ dml_max(
+														mode_lib->vba.HRatio[k],
+														1.0)),
+								1)) - (mode_lib->vba.vtaps[k] - 1);
+
+		mode_lib->vba.LBLatencyHidingSourceLinesC =
+				dml_min(
+						mode_lib->vba.MaxLineBufferLines,
+						(unsigned int) dml_floor(
+								(double) mode_lib->vba.LineBufferSize
+										/ mode_lib->vba.LBBitPerPixel[k]
+										/ (mode_lib->vba.SwathWidthY[k]
+												/ 2.0
+												/ dml_max(
+														mode_lib->vba.HRatio[k]
+																/ 2,
+														1.0)),
+								1))
+						- (mode_lib->vba.VTAPsChroma[k] - 1);
+
+		EffectiveLBLatencyHidingY = mode_lib->vba.LBLatencyHidingSourceLinesY
+				/ mode_lib->vba.VRatio[k]
+				* (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k]);
+
+		EffectiveLBLatencyHidingC = mode_lib->vba.LBLatencyHidingSourceLinesC
+				/ (mode_lib->vba.VRatio[k] / 2)
+				* (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k]);
+
+		if (mode_lib->vba.SwathWidthY[k] > 2 * mode_lib->vba.DPPOutputBufferPixels) {
+			DPPOutputBufferLinesY = mode_lib->vba.DPPOutputBufferPixels
+					/ mode_lib->vba.SwathWidthY[k];
+		} else if (mode_lib->vba.SwathWidthY[k] > mode_lib->vba.DPPOutputBufferPixels) {
+			DPPOutputBufferLinesY = 0.5;
+		} else {
+			DPPOutputBufferLinesY = 1;
+		}
+
+		if (mode_lib->vba.SwathWidthY[k] / 2 > 2 * mode_lib->vba.DPPOutputBufferPixels) {
+			DPPOutputBufferLinesC = mode_lib->vba.DPPOutputBufferPixels
+					/ (mode_lib->vba.SwathWidthY[k] / 2);
+		} else if (mode_lib->vba.SwathWidthY[k] / 2 > mode_lib->vba.DPPOutputBufferPixels) {
+			DPPOutputBufferLinesC = 0.5;
+		} else {
+			DPPOutputBufferLinesC = 1;
+		}
+
+		DPPOPPBufferingY = (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k])
+				* (DPPOutputBufferLinesY + mode_lib->vba.OPPOutputBufferLines);
+		MaxDETBufferingTimeY = mode_lib->vba.FullDETBufferingTimeY[k]
+				+ (mode_lib->vba.LinesInDETY[k]
+						- mode_lib->vba.LinesInDETYRoundedDownToSwath[k])
+						/ mode_lib->vba.SwathHeightY[k]
+						* (mode_lib->vba.HTotal[k]
+								/ mode_lib->vba.PixelClock[k]);
+
+		ActiveDRAMClockChangeLatencyMarginY = DPPOPPBufferingY + EffectiveLBLatencyHidingY
+				+ MaxDETBufferingTimeY - mode_lib->vba.DRAMClockChangeWatermark;
+
+		if (mode_lib->vba.ActiveDPPs > 1) {
+			ActiveDRAMClockChangeLatencyMarginY =
+					ActiveDRAMClockChangeLatencyMarginY
+							- (1 - 1 / (mode_lib->vba.ActiveDPPs - 1))
+									* mode_lib->vba.SwathHeightY[k]
+									* (mode_lib->vba.HTotal[k]
+											/ mode_lib->vba.PixelClock[k]);
+		}
+
+		if (mode_lib->vba.BytePerPixelDETC[k] > 0) {
+			double DPPOPPBufferingC = (mode_lib->vba.HTotal[k]
+					/ mode_lib->vba.PixelClock[k])
+					* (DPPOutputBufferLinesC
+							+ mode_lib->vba.OPPOutputBufferLines);
+			double MaxDETBufferingTimeC =
+					mode_lib->vba.FullDETBufferingTimeC[k]
+							+ (mode_lib->vba.LinesInDETC[k]
+									- mode_lib->vba.LinesInDETCRoundedDownToSwath[k])
+									/ mode_lib->vba.SwathHeightC[k]
+									* (mode_lib->vba.HTotal[k]
+											/ mode_lib->vba.PixelClock[k]);
+			double ActiveDRAMClockChangeLatencyMarginC = DPPOPPBufferingC
+					+ EffectiveLBLatencyHidingC + MaxDETBufferingTimeC
+					- mode_lib->vba.DRAMClockChangeWatermark;
+
+			if (mode_lib->vba.ActiveDPPs > 1) {
+				ActiveDRAMClockChangeLatencyMarginC =
+						ActiveDRAMClockChangeLatencyMarginC
+								- (1
+										- 1
+												/ (mode_lib->vba.ActiveDPPs
+														- 1))
+										* mode_lib->vba.SwathHeightC[k]
+										* (mode_lib->vba.HTotal[k]
+												/ mode_lib->vba.PixelClock[k]);
+			}
+			mode_lib->vba.ActiveDRAMClockChangeLatencyMargin[k] = dml_min(
+					ActiveDRAMClockChangeLatencyMarginY,
+					ActiveDRAMClockChangeLatencyMarginC);
+		} else {
+			mode_lib->vba.ActiveDRAMClockChangeLatencyMargin[k] =
+					ActiveDRAMClockChangeLatencyMarginY;
+		}
+
+		if (mode_lib->vba.WritebackEnable[k]) {
+			double WritebackDRAMClockChangeLatencyMargin;
+
+			if (mode_lib->vba.WritebackPixelFormat[k] == dm_444_32) {
+				WritebackDRAMClockChangeLatencyMargin =
+						(double) (mode_lib->vba.WritebackInterfaceLumaBufferSize
+								+ mode_lib->vba.WritebackInterfaceChromaBufferSize)
+								/ (mode_lib->vba.WritebackDestinationWidth[k]
+										* mode_lib->vba.WritebackDestinationHeight[k]
+										/ (mode_lib->vba.WritebackSourceHeight[k]
+												* mode_lib->vba.HTotal[k]
+												/ mode_lib->vba.PixelClock[k])
+										* 4)
+								- mode_lib->vba.WritebackDRAMClockChangeWatermark;
+			} else if (mode_lib->vba.WritebackPixelFormat[k] == dm_420_10) {
+				WritebackDRAMClockChangeLatencyMargin =
+						dml_min(
+								(double) mode_lib->vba.WritebackInterfaceLumaBufferSize
+										* 8.0 / 10,
+								2.0
+										* mode_lib->vba.WritebackInterfaceChromaBufferSize
+										* 8 / 10)
+								/ (mode_lib->vba.WritebackDestinationWidth[k]
+										* mode_lib->vba.WritebackDestinationHeight[k]
+										/ (mode_lib->vba.WritebackSourceHeight[k]
+												* mode_lib->vba.HTotal[k]
+												/ mode_lib->vba.PixelClock[k]))
+								- mode_lib->vba.WritebackDRAMClockChangeWatermark;
+			} else {
+				WritebackDRAMClockChangeLatencyMargin =
+						dml_min(
+								(double) mode_lib->vba.WritebackInterfaceLumaBufferSize,
+								2.0
+										* mode_lib->vba.WritebackInterfaceChromaBufferSize)
+								/ (mode_lib->vba.WritebackDestinationWidth[k]
+										* mode_lib->vba.WritebackDestinationHeight[k]
+										/ (mode_lib->vba.WritebackSourceHeight[k]
+												* mode_lib->vba.HTotal[k]
+												/ mode_lib->vba.PixelClock[k]))
+								- mode_lib->vba.WritebackDRAMClockChangeWatermark;
+			}
+			mode_lib->vba.ActiveDRAMClockChangeLatencyMargin[k] = dml_min(
+					mode_lib->vba.ActiveDRAMClockChangeLatencyMargin[k],
+					WritebackDRAMClockChangeLatencyMargin);
+		}
+	}
+
+	{
+	float SecondMinActiveDRAMClockChangeMarginOneDisplayInVBLank = 999999;
+	int PlaneWithMinActiveDRAMClockChangeMargin = -1;
+
+	mode_lib->vba.MinActiveDRAMClockChangeMargin = 999999;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.ActiveDRAMClockChangeLatencyMargin[k]
+				< mode_lib->vba.MinActiveDRAMClockChangeMargin) {
+			mode_lib->vba.MinActiveDRAMClockChangeMargin =
+					mode_lib->vba.ActiveDRAMClockChangeLatencyMargin[k];
+			if (mode_lib->vba.BlendingAndTiming[k] == k) {
+				PlaneWithMinActiveDRAMClockChangeMargin = k;
+			} else {
+				for (j = 0; j < mode_lib->vba.NumberOfActivePlanes; ++j) {
+					if (mode_lib->vba.BlendingAndTiming[k] == j) {
+						PlaneWithMinActiveDRAMClockChangeMargin = j;
+					}
+				}
+			}
+		}
+	}
+
+	mode_lib->vba.MinActiveDRAMClockChangeLatencySupported =
+			mode_lib->vba.MinActiveDRAMClockChangeMargin
+					+ mode_lib->vba.DRAMClockChangeLatency;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (!((k == PlaneWithMinActiveDRAMClockChangeMargin) && (mode_lib->vba.BlendingAndTiming[k] == k))
+				&& !(mode_lib->vba.BlendingAndTiming[k] == PlaneWithMinActiveDRAMClockChangeMargin)
+				&& mode_lib->vba.ActiveDRAMClockChangeLatencyMargin[k]
+						< SecondMinActiveDRAMClockChangeMarginOneDisplayInVBLank) {
+			SecondMinActiveDRAMClockChangeMarginOneDisplayInVBLank =
+					mode_lib->vba.ActiveDRAMClockChangeLatencyMargin[k];
+		}
+	}
+
+	if (mode_lib->vba.DRAMClockChangeSupportsVActive &&
+			mode_lib->vba.MinActiveDRAMClockChangeMargin > 60) {
+		mode_lib->vba.DRAMClockChangeWatermark += 25;
+
+		for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+			if (mode_lib->vba.PrefetchMode[mode_lib->vba.VoltageLevel][mode_lib->vba.maxMpcComb] == 0) {
+				if (mode_lib->vba.DRAMClockChangeWatermark >
+					dml_max(mode_lib->vba.StutterEnterPlusExitWatermark, mode_lib->vba.UrgentWatermark))
+					mode_lib->vba.MinTTUVBlank[k] += 25;
+			}
+		}
+
+		mode_lib->vba.DRAMClockChangeSupport[0][0] = dm_dram_clock_change_vactive;
+	} else if (mode_lib->vba.DummyPStateCheck &&
+			mode_lib->vba.MinActiveDRAMClockChangeMargin > 0) {
+		mode_lib->vba.DRAMClockChangeSupport[0][0] = dm_dram_clock_change_vactive;
+	} else {
+		if ((mode_lib->vba.SynchronizedVBlank
+				|| mode_lib->vba.NumberOfActivePlanes == 1
+				|| (SecondMinActiveDRAMClockChangeMarginOneDisplayInVBLank > 0 &&
+						mode_lib->vba.AllowDramClockChangeOneDisplayVactive))
+					&& mode_lib->vba.PrefetchMode[mode_lib->vba.VoltageLevel][mode_lib->vba.maxMpcComb] == 0) {
+			mode_lib->vba.DRAMClockChangeSupport[0][0] = dm_dram_clock_change_vblank;
+			for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+				if (!mode_lib->vba.AllowDRAMClockChangeDuringVBlank[k]) {
+					mode_lib->vba.DRAMClockChangeSupport[0][0] =
+							dm_dram_clock_change_unsupported;
+				}
+			}
+		} else {
+			mode_lib->vba.DRAMClockChangeSupport[0][0] = dm_dram_clock_change_unsupported;
+		}
+	}
+	}
+	for (k = 0; k <= mode_lib->vba.soc.num_states; k++)
+		for (j = 0; j < 2; j++)
+			mode_lib->vba.DRAMClockChangeSupport[k][j] = mode_lib->vba.DRAMClockChangeSupport[0][0];
+
+	//XFC Parameters:
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.XFCEnabled[k] == true) {
+			double TWait;
+
+			mode_lib->vba.XFCSlaveVUpdateOffset[k] = mode_lib->vba.XFCTSlvVupdateOffset;
+			mode_lib->vba.XFCSlaveVupdateWidth[k] = mode_lib->vba.XFCTSlvVupdateWidth;
+			mode_lib->vba.XFCSlaveVReadyOffset[k] = mode_lib->vba.XFCTSlvVreadyOffset;
+			TWait = CalculateTWait(
+					mode_lib->vba.PrefetchMode[mode_lib->vba.VoltageLevel][mode_lib->vba.maxMpcComb],
+					mode_lib->vba.DRAMClockChangeLatency,
+					mode_lib->vba.UrgentLatencyPixelDataOnly,
+					mode_lib->vba.SREnterPlusExitTime);
+			mode_lib->vba.XFCRemoteSurfaceFlipDelay = CalculateRemoteSurfaceFlipDelay(
+					mode_lib,
+					mode_lib->vba.VRatio[k],
+					mode_lib->vba.SwathWidthY[k],
+					dml_ceil(mode_lib->vba.BytePerPixelDETY[k], 1),
+					mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k],
+					mode_lib->vba.XFCTSlvVupdateOffset,
+					mode_lib->vba.XFCTSlvVupdateWidth,
+					mode_lib->vba.XFCTSlvVreadyOffset,
+					mode_lib->vba.XFCXBUFLatencyTolerance,
+					mode_lib->vba.XFCFillBWOverhead,
+					mode_lib->vba.XFCSlvChunkSize,
+					mode_lib->vba.XFCBusTransportTime,
+					mode_lib->vba.TCalc,
+					TWait,
+					&mode_lib->vba.SrcActiveDrainRate,
+					&mode_lib->vba.TInitXFill,
+					&mode_lib->vba.TslvChk);
+			mode_lib->vba.XFCRemoteSurfaceFlipLatency[k] =
+					dml_floor(
+							mode_lib->vba.XFCRemoteSurfaceFlipDelay
+									/ (mode_lib->vba.HTotal[k]
+											/ mode_lib->vba.PixelClock[k]),
+							1);
+			mode_lib->vba.XFCTransferDelay[k] =
+					dml_ceil(
+							mode_lib->vba.XFCBusTransportTime
+									/ (mode_lib->vba.HTotal[k]
+											/ mode_lib->vba.PixelClock[k]),
+							1);
+			mode_lib->vba.XFCPrechargeDelay[k] =
+					dml_ceil(
+							(mode_lib->vba.XFCBusTransportTime
+									+ mode_lib->vba.TInitXFill
+									+ mode_lib->vba.TslvChk)
+									/ (mode_lib->vba.HTotal[k]
+											/ mode_lib->vba.PixelClock[k]),
+							1);
+			mode_lib->vba.InitFillLevel = mode_lib->vba.XFCXBUFLatencyTolerance
+					* mode_lib->vba.SrcActiveDrainRate;
+			mode_lib->vba.FinalFillMargin =
+					(mode_lib->vba.DestinationLinesToRequestVMInVBlank[k]
+							+ mode_lib->vba.DestinationLinesToRequestRowInVBlank[k])
+							* mode_lib->vba.HTotal[k]
+							/ mode_lib->vba.PixelClock[k]
+							* mode_lib->vba.SrcActiveDrainRate
+							+ mode_lib->vba.XFCFillConstant;
+			mode_lib->vba.FinalFillLevel = mode_lib->vba.XFCRemoteSurfaceFlipDelay
+					* mode_lib->vba.SrcActiveDrainRate
+					+ mode_lib->vba.FinalFillMargin;
+			mode_lib->vba.RemainingFillLevel = dml_max(
+					0.0,
+					mode_lib->vba.FinalFillLevel - mode_lib->vba.InitFillLevel);
+			mode_lib->vba.TFinalxFill = mode_lib->vba.RemainingFillLevel
+					/ (mode_lib->vba.SrcActiveDrainRate
+							* mode_lib->vba.XFCFillBWOverhead / 100);
+			mode_lib->vba.XFCPrefetchMargin[k] =
+					mode_lib->vba.XFCRemoteSurfaceFlipDelay
+							+ mode_lib->vba.TFinalxFill
+							+ (mode_lib->vba.DestinationLinesToRequestVMInVBlank[k]
+									+ mode_lib->vba.DestinationLinesToRequestRowInVBlank[k])
+									* mode_lib->vba.HTotal[k]
+									/ mode_lib->vba.PixelClock[k];
+		} else {
+			mode_lib->vba.XFCSlaveVUpdateOffset[k] = 0;
+			mode_lib->vba.XFCSlaveVupdateWidth[k] = 0;
+			mode_lib->vba.XFCSlaveVReadyOffset[k] = 0;
+			mode_lib->vba.XFCRemoteSurfaceFlipLatency[k] = 0;
+			mode_lib->vba.XFCPrechargeDelay[k] = 0;
+			mode_lib->vba.XFCTransferDelay[k] = 0;
+			mode_lib->vba.XFCPrefetchMargin[k] = 0;
+		}
+	}
+	{
+		unsigned int VStartupMargin = 0;
+		bool FirstMainPlane = true;
+
+		for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+			if (mode_lib->vba.BlendingAndTiming[k] == k) {
+				unsigned int Margin = (mode_lib->vba.MaxVStartupLines[k] - mode_lib->vba.VStartup[k])
+						* mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k];
+
+				if (FirstMainPlane) {
+					VStartupMargin = Margin;
+					FirstMainPlane = false;
+				} else
+					VStartupMargin = dml_min(VStartupMargin, Margin);
+		}
+
+		if (mode_lib->vba.UseMaximumVStartup) {
+			if (mode_lib->vba.VTotal_Max[k] == mode_lib->vba.VTotal[k]) {
+				//only use max vstart if it is not drr or lateflip.
+				mode_lib->vba.VStartup[k] = mode_lib->vba.MaxVStartupLines[mode_lib->vba.BlendingAndTiming[k]];
+			}
+		}
+	}
+}
+}
+
+static void dml20v2_DisplayPipeConfiguration(struct display_mode_lib *mode_lib)
+{
+	double BytePerPixDETY;
+	double BytePerPixDETC;
+	double Read256BytesBlockHeightY;
+	double Read256BytesBlockHeightC;
+	double Read256BytesBlockWidthY;
+	double Read256BytesBlockWidthC;
+	double MaximumSwathHeightY;
+	double MaximumSwathHeightC;
+	double MinimumSwathHeightY;
+	double MinimumSwathHeightC;
+	double SwathWidth;
+	double SwathWidthGranularityY;
+	double SwathWidthGranularityC;
+	double RoundedUpMaxSwathSizeBytesY;
+	double RoundedUpMaxSwathSizeBytesC;
+	unsigned int j, k;
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		bool MainPlaneDoesODMCombine = false;
+
+		if (mode_lib->vba.SourcePixelFormat[k] == dm_444_64) {
+			BytePerPixDETY = 8;
+			BytePerPixDETC = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_32) {
+			BytePerPixDETY = 4;
+			BytePerPixDETC = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_16) {
+			BytePerPixDETY = 2;
+			BytePerPixDETC = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_8) {
+			BytePerPixDETY = 1;
+			BytePerPixDETC = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_8) {
+			BytePerPixDETY = 1;
+			BytePerPixDETC = 2;
+		} else {
+			BytePerPixDETY = 4.0 / 3.0;
+			BytePerPixDETC = 8.0 / 3.0;
+		}
+
+		if ((mode_lib->vba.SourcePixelFormat[k] == dm_444_64
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_32
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_16
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_8)) {
+			if (mode_lib->vba.SurfaceTiling[k] == dm_sw_linear) {
+				Read256BytesBlockHeightY = 1;
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_64) {
+				Read256BytesBlockHeightY = 4;
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_32
+					|| mode_lib->vba.SourcePixelFormat[k] == dm_444_16) {
+				Read256BytesBlockHeightY = 8;
+			} else {
+				Read256BytesBlockHeightY = 16;
+			}
+			Read256BytesBlockWidthY = 256 / dml_ceil(BytePerPixDETY, 1)
+					/ Read256BytesBlockHeightY;
+			Read256BytesBlockHeightC = 0;
+			Read256BytesBlockWidthC = 0;
+		} else {
+			if (mode_lib->vba.SurfaceTiling[k] == dm_sw_linear) {
+				Read256BytesBlockHeightY = 1;
+				Read256BytesBlockHeightC = 1;
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_8) {
+				Read256BytesBlockHeightY = 16;
+				Read256BytesBlockHeightC = 8;
+			} else {
+				Read256BytesBlockHeightY = 8;
+				Read256BytesBlockHeightC = 8;
+			}
+			Read256BytesBlockWidthY = 256 / dml_ceil(BytePerPixDETY, 1)
+					/ Read256BytesBlockHeightY;
+			Read256BytesBlockWidthC = 256 / dml_ceil(BytePerPixDETC, 2)
+					/ Read256BytesBlockHeightC;
+		}
+
+		if (mode_lib->vba.SourceScan[k] == dm_horz) {
+			MaximumSwathHeightY = Read256BytesBlockHeightY;
+			MaximumSwathHeightC = Read256BytesBlockHeightC;
+		} else {
+			MaximumSwathHeightY = Read256BytesBlockWidthY;
+			MaximumSwathHeightC = Read256BytesBlockWidthC;
+		}
+
+		if ((mode_lib->vba.SourcePixelFormat[k] == dm_444_64
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_32
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_16
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_8)) {
+			if (mode_lib->vba.SurfaceTiling[k] == dm_sw_linear
+					|| (mode_lib->vba.SourcePixelFormat[k] == dm_444_64
+							&& (mode_lib->vba.SurfaceTiling[k]
+									== dm_sw_4kb_s
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_4kb_s_x
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_64kb_s
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_64kb_s_t
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_64kb_s_x
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_var_s
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_var_s_x)
+							&& mode_lib->vba.SourceScan[k] == dm_horz)) {
+				MinimumSwathHeightY = MaximumSwathHeightY;
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_8
+					&& mode_lib->vba.SourceScan[k] != dm_horz) {
+				MinimumSwathHeightY = MaximumSwathHeightY;
+			} else {
+				MinimumSwathHeightY = MaximumSwathHeightY / 2.0;
+			}
+			MinimumSwathHeightC = MaximumSwathHeightC;
+		} else {
+			if (mode_lib->vba.SurfaceTiling[k] == dm_sw_linear) {
+				MinimumSwathHeightY = MaximumSwathHeightY;
+				MinimumSwathHeightC = MaximumSwathHeightC;
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_8
+					&& mode_lib->vba.SourceScan[k] == dm_horz) {
+				MinimumSwathHeightY = MaximumSwathHeightY / 2.0;
+				MinimumSwathHeightC = MaximumSwathHeightC;
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_10
+					&& mode_lib->vba.SourceScan[k] == dm_horz) {
+				MinimumSwathHeightC = MaximumSwathHeightC / 2.0;
+				MinimumSwathHeightY = MaximumSwathHeightY;
+			} else {
+				MinimumSwathHeightY = MaximumSwathHeightY;
+				MinimumSwathHeightC = MaximumSwathHeightC;
+			}
+		}
+
+		if (mode_lib->vba.SourceScan[k] == dm_horz) {
+			SwathWidth = mode_lib->vba.ViewportWidth[k];
+		} else {
+			SwathWidth = mode_lib->vba.ViewportHeight[k];
+		}
+
+		if (mode_lib->vba.ODMCombineEnabled[k] == dm_odm_combine_mode_2to1) {
+			MainPlaneDoesODMCombine = true;
+		}
+		for (j = 0; j < mode_lib->vba.NumberOfActivePlanes; ++j) {
+			if (mode_lib->vba.BlendingAndTiming[k] == j
+					&& mode_lib->vba.ODMCombineEnabled[k] == dm_odm_combine_mode_2to1) {
+				MainPlaneDoesODMCombine = true;
+			}
+		}
+
+		if (MainPlaneDoesODMCombine == true) {
+			SwathWidth = dml_min(
+					SwathWidth,
+					mode_lib->vba.HActive[k] / 2.0 * mode_lib->vba.HRatio[k]);
+		} else {
+			if (mode_lib->vba.DPPPerPlane[k] == 0)
+				SwathWidth = 0;
+			else
+				SwathWidth = SwathWidth / mode_lib->vba.DPPPerPlane[k];
+		}
+
+		SwathWidthGranularityY = 256 / dml_ceil(BytePerPixDETY, 1) / MaximumSwathHeightY;
+		RoundedUpMaxSwathSizeBytesY = (dml_ceil(
+				(double) (SwathWidth - 1),
+				SwathWidthGranularityY) + SwathWidthGranularityY) * BytePerPixDETY
+				* MaximumSwathHeightY;
+		if (mode_lib->vba.SourcePixelFormat[k] == dm_420_10) {
+			RoundedUpMaxSwathSizeBytesY = dml_ceil(RoundedUpMaxSwathSizeBytesY, 256)
+					+ 256;
+		}
+		if (MaximumSwathHeightC > 0) {
+			SwathWidthGranularityC = 256.0 / dml_ceil(BytePerPixDETC, 2)
+					/ MaximumSwathHeightC;
+			RoundedUpMaxSwathSizeBytesC = (dml_ceil(
+					(double) (SwathWidth / 2.0 - 1),
+					SwathWidthGranularityC) + SwathWidthGranularityC)
+					* BytePerPixDETC * MaximumSwathHeightC;
+			if (mode_lib->vba.SourcePixelFormat[k] == dm_420_10) {
+				RoundedUpMaxSwathSizeBytesC = dml_ceil(
+						RoundedUpMaxSwathSizeBytesC,
+						256) + 256;
+			}
+		} else
+			RoundedUpMaxSwathSizeBytesC = 0.0;
+
+		if (RoundedUpMaxSwathSizeBytesY + RoundedUpMaxSwathSizeBytesC
+				<= mode_lib->vba.DETBufferSizeInKByte[0] * 1024.0 / 2.0) {
+			mode_lib->vba.SwathHeightY[k] = MaximumSwathHeightY;
+			mode_lib->vba.SwathHeightC[k] = MaximumSwathHeightC;
+		} else {
+			mode_lib->vba.SwathHeightY[k] = MinimumSwathHeightY;
+			mode_lib->vba.SwathHeightC[k] = MinimumSwathHeightC;
+		}
+
+		if (mode_lib->vba.SwathHeightC[k] == 0) {
+			mode_lib->vba.DETBufferSizeY[k] = mode_lib->vba.DETBufferSizeInKByte[0] * 1024;
+			mode_lib->vba.DETBufferSizeC[k] = 0;
+		} else if (mode_lib->vba.SwathHeightY[k] <= mode_lib->vba.SwathHeightC[k]) {
+			mode_lib->vba.DETBufferSizeY[k] = mode_lib->vba.DETBufferSizeInKByte[0]
+					* 1024.0 / 2;
+			mode_lib->vba.DETBufferSizeC[k] = mode_lib->vba.DETBufferSizeInKByte[0]
+					* 1024.0 / 2;
+		} else {
+			mode_lib->vba.DETBufferSizeY[k] = mode_lib->vba.DETBufferSizeInKByte[0]
+					* 1024.0 * 2 / 3;
+			mode_lib->vba.DETBufferSizeC[k] = mode_lib->vba.DETBufferSizeInKByte[0]
+					* 1024.0 / 3;
+		}
+	}
+}
+
+static double CalculateTWait(
+		unsigned int PrefetchMode,
+		double DRAMClockChangeLatency,
+		double UrgentLatencyPixelDataOnly,
+		double SREnterPlusExitTime)
+{
+	if (PrefetchMode == 0) {
+		return dml_max(
+				DRAMClockChangeLatency + UrgentLatencyPixelDataOnly,
+				dml_max(SREnterPlusExitTime, UrgentLatencyPixelDataOnly));
+	} else if (PrefetchMode == 1) {
+		return dml_max(SREnterPlusExitTime, UrgentLatencyPixelDataOnly);
+	} else {
+		return UrgentLatencyPixelDataOnly;
+	}
+}
+
+static double CalculateRemoteSurfaceFlipDelay(
+		struct display_mode_lib *mode_lib,
+		double VRatio,
+		double SwathWidth,
+		double Bpp,
+		double LineTime,
+		double XFCTSlvVupdateOffset,
+		double XFCTSlvVupdateWidth,
+		double XFCTSlvVreadyOffset,
+		double XFCXBUFLatencyTolerance,
+		double XFCFillBWOverhead,
+		double XFCSlvChunkSize,
+		double XFCBusTransportTime,
+		double TCalc,
+		double TWait,
+		double *SrcActiveDrainRate,
+		double *TInitXFill,
+		double *TslvChk)
+{
+	double TSlvSetup, AvgfillRate, result;
+
+	*SrcActiveDrainRate = VRatio * SwathWidth * Bpp / LineTime;
+	TSlvSetup = XFCTSlvVupdateOffset + XFCTSlvVupdateWidth + XFCTSlvVreadyOffset;
+	*TInitXFill = XFCXBUFLatencyTolerance / (1 + XFCFillBWOverhead / 100);
+	AvgfillRate = *SrcActiveDrainRate * (1 + XFCFillBWOverhead / 100);
+	*TslvChk = XFCSlvChunkSize / AvgfillRate;
+	dml_print(
+			"DML::CalculateRemoteSurfaceFlipDelay: SrcActiveDrainRate: %f\n",
+			*SrcActiveDrainRate);
+	dml_print("DML::CalculateRemoteSurfaceFlipDelay: TSlvSetup: %f\n", TSlvSetup);
+	dml_print("DML::CalculateRemoteSurfaceFlipDelay: TInitXFill: %f\n", *TInitXFill);
+	dml_print("DML::CalculateRemoteSurfaceFlipDelay: AvgfillRate: %f\n", AvgfillRate);
+	dml_print("DML::CalculateRemoteSurfaceFlipDelay: TslvChk: %f\n", *TslvChk);
+	result = 2 * XFCBusTransportTime + TSlvSetup + TCalc + TWait + *TslvChk + *TInitXFill; // TODO: This doesn't seem to match programming guide
+	dml_print("DML::CalculateRemoteSurfaceFlipDelay: RemoteSurfaceFlipDelay: %f\n", result);
+	return result;
+}
+
+static double CalculateWriteBackDelay(
+		enum source_format_class WritebackPixelFormat,
+		double WritebackHRatio,
+		double WritebackVRatio,
+		unsigned int WritebackLumaHTaps,
+		unsigned int WritebackLumaVTaps,
+		unsigned int WritebackChromaHTaps,
+		unsigned int WritebackChromaVTaps,
+		unsigned int WritebackDestinationWidth)
+{
+	double CalculateWriteBackDelay =
+			dml_max(
+					dml_ceil(WritebackLumaHTaps / 4.0, 1) / WritebackHRatio,
+					WritebackLumaVTaps * dml_ceil(1.0 / WritebackVRatio, 1)
+							* dml_ceil(
+									WritebackDestinationWidth
+											/ 4.0,
+									1)
+							+ dml_ceil(1.0 / WritebackVRatio, 1)
+									* (dml_ceil(
+											WritebackLumaVTaps
+													/ 4.0,
+											1) + 4));
+
+	if (WritebackPixelFormat != dm_444_32) {
+		CalculateWriteBackDelay =
+				dml_max(
+						CalculateWriteBackDelay,
+						dml_max(
+								dml_ceil(
+										WritebackChromaHTaps
+												/ 2.0,
+										1)
+										/ (2
+												* WritebackHRatio),
+								WritebackChromaVTaps
+										* dml_ceil(
+												1
+														/ (2
+																* WritebackVRatio),
+												1)
+										* dml_ceil(
+												WritebackDestinationWidth
+														/ 2.0
+														/ 2.0,
+												1)
+										+ dml_ceil(
+												1
+														/ (2
+																* WritebackVRatio),
+												1)
+												* (dml_ceil(
+														WritebackChromaVTaps
+																/ 4.0,
+														1)
+														+ 4)));
+	}
+	return CalculateWriteBackDelay;
+}
+
+static void CalculateActiveRowBandwidth(
+		bool GPUVMEnable,
+		enum source_format_class SourcePixelFormat,
+		double VRatio,
+		bool DCCEnable,
+		double LineTime,
+		unsigned int MetaRowByteLuma,
+		unsigned int MetaRowByteChroma,
+		unsigned int meta_row_height_luma,
+		unsigned int meta_row_height_chroma,
+		unsigned int PixelPTEBytesPerRowLuma,
+		unsigned int PixelPTEBytesPerRowChroma,
+		unsigned int dpte_row_height_luma,
+		unsigned int dpte_row_height_chroma,
+		double *meta_row_bw,
+		double *dpte_row_bw,
+		double *qual_row_bw)
+{
+	if (DCCEnable != true) {
+		*meta_row_bw = 0;
+	} else if (SourcePixelFormat == dm_420_8 || SourcePixelFormat == dm_420_10) {
+		*meta_row_bw = VRatio * MetaRowByteLuma / (meta_row_height_luma * LineTime)
+				+ VRatio / 2 * MetaRowByteChroma
+						/ (meta_row_height_chroma * LineTime);
+	} else {
+		*meta_row_bw = VRatio * MetaRowByteLuma / (meta_row_height_luma * LineTime);
+	}
+
+	if (GPUVMEnable != true) {
+		*dpte_row_bw = 0;
+	} else if (SourcePixelFormat == dm_420_8 || SourcePixelFormat == dm_420_10) {
+		*dpte_row_bw = VRatio * PixelPTEBytesPerRowLuma / (dpte_row_height_luma * LineTime)
+				+ VRatio / 2 * PixelPTEBytesPerRowChroma
+						/ (dpte_row_height_chroma * LineTime);
+	} else {
+		*dpte_row_bw = VRatio * PixelPTEBytesPerRowLuma / (dpte_row_height_luma * LineTime);
+	}
+
+	if ((SourcePixelFormat == dm_420_8 || SourcePixelFormat == dm_420_10)) {
+		*qual_row_bw = *meta_row_bw + *dpte_row_bw;
+	} else {
+		*qual_row_bw = 0;
+	}
+}
+
+static void CalculateFlipSchedule(
+		struct display_mode_lib *mode_lib,
+		double UrgentExtraLatency,
+		double UrgentLatencyPixelDataOnly,
+		unsigned int GPUVMMaxPageTableLevels,
+		bool GPUVMEnable,
+		double BandwidthAvailableForImmediateFlip,
+		unsigned int TotImmediateFlipBytes,
+		enum source_format_class SourcePixelFormat,
+		unsigned int ImmediateFlipBytes,
+		double LineTime,
+		double VRatio,
+		double Tno_bw,
+		double PDEAndMetaPTEBytesFrame,
+		unsigned int MetaRowByte,
+		unsigned int PixelPTEBytesPerRow,
+		bool DCCEnable,
+		unsigned int dpte_row_height,
+		unsigned int meta_row_height,
+		double qual_row_bw,
+		double *DestinationLinesToRequestVMInImmediateFlip,
+		double *DestinationLinesToRequestRowInImmediateFlip,
+		double *final_flip_bw,
+		bool *ImmediateFlipSupportedForPipe)
+{
+	double min_row_time = 0.0;
+
+	if (SourcePixelFormat == dm_420_8 || SourcePixelFormat == dm_420_10) {
+		*DestinationLinesToRequestVMInImmediateFlip = 0.0;
+		*DestinationLinesToRequestRowInImmediateFlip = 0.0;
+		*final_flip_bw = qual_row_bw;
+		*ImmediateFlipSupportedForPipe = true;
+	} else {
+		double TimeForFetchingMetaPTEImmediateFlip;
+		double TimeForFetchingRowInVBlankImmediateFlip;
+
+		if (GPUVMEnable == true) {
+			mode_lib->vba.ImmediateFlipBW[0] = BandwidthAvailableForImmediateFlip
+					* ImmediateFlipBytes / TotImmediateFlipBytes;
+			TimeForFetchingMetaPTEImmediateFlip =
+					dml_max(
+							Tno_bw
+									+ PDEAndMetaPTEBytesFrame
+											/ mode_lib->vba.ImmediateFlipBW[0],
+							dml_max(
+									UrgentExtraLatency
+											+ UrgentLatencyPixelDataOnly
+													* (GPUVMMaxPageTableLevels
+															- 1),
+									LineTime / 4.0));
+		} else {
+			TimeForFetchingMetaPTEImmediateFlip = 0;
+		}
+
+		*DestinationLinesToRequestVMInImmediateFlip = dml_floor(
+				4.0 * (TimeForFetchingMetaPTEImmediateFlip / LineTime + 0.125),
+				1) / 4.0;
+
+		if ((GPUVMEnable == true || DCCEnable == true)) {
+			mode_lib->vba.ImmediateFlipBW[0] = BandwidthAvailableForImmediateFlip
+					* ImmediateFlipBytes / TotImmediateFlipBytes;
+			TimeForFetchingRowInVBlankImmediateFlip = dml_max(
+					(MetaRowByte + PixelPTEBytesPerRow)
+							/ mode_lib->vba.ImmediateFlipBW[0],
+					dml_max(UrgentLatencyPixelDataOnly, LineTime / 4.0));
+		} else {
+			TimeForFetchingRowInVBlankImmediateFlip = 0;
+		}
+
+		*DestinationLinesToRequestRowInImmediateFlip = dml_floor(
+				4.0 * (TimeForFetchingRowInVBlankImmediateFlip / LineTime + 0.125),
+				1) / 4.0;
+
+		if (GPUVMEnable == true) {
+			*final_flip_bw =
+					dml_max(
+							PDEAndMetaPTEBytesFrame
+									/ (*DestinationLinesToRequestVMInImmediateFlip
+											* LineTime),
+							(MetaRowByte + PixelPTEBytesPerRow)
+									/ (TimeForFetchingRowInVBlankImmediateFlip
+											* LineTime));
+		} else if (MetaRowByte + PixelPTEBytesPerRow > 0) {
+			*final_flip_bw = (MetaRowByte + PixelPTEBytesPerRow)
+					/ (TimeForFetchingRowInVBlankImmediateFlip * LineTime);
+		} else {
+			*final_flip_bw = 0;
+		}
+
+		if (GPUVMEnable && !DCCEnable)
+			min_row_time = dpte_row_height * LineTime / VRatio;
+		else if (!GPUVMEnable && DCCEnable)
+			min_row_time = meta_row_height * LineTime / VRatio;
+		else
+			min_row_time = dml_min(dpte_row_height, meta_row_height) * LineTime
+					/ VRatio;
+
+		if (*DestinationLinesToRequestVMInImmediateFlip >= 8
+				|| *DestinationLinesToRequestRowInImmediateFlip >= 16
+				|| TimeForFetchingMetaPTEImmediateFlip
+						+ 2 * TimeForFetchingRowInVBlankImmediateFlip
+						> min_row_time)
+			*ImmediateFlipSupportedForPipe = false;
+		else
+			*ImmediateFlipSupportedForPipe = true;
+	}
+}
+
+static unsigned int TruncToValidBPP(
+		double DecimalBPP,
+		double DesiredBPP,
+		bool DSCEnabled,
+		enum output_encoder_class Output,
+		enum output_format_class Format,
+		unsigned int DSCInputBitPerComponent)
+{
+	if (Output == dm_hdmi) {
+		if (Format == dm_420) {
+			if (DecimalBPP >= 18 && (DesiredBPP == 0 || DesiredBPP == 18))
+				return 18;
+			else if (DecimalBPP >= 15 && (DesiredBPP == 0 || DesiredBPP == 15))
+				return 15;
+			else if (DecimalBPP >= 12 && (DesiredBPP == 0 || DesiredBPP == 12))
+				return 12;
+			else
+				return BPP_INVALID;
+		} else if (Format == dm_444) {
+			if (DecimalBPP >= 36 && (DesiredBPP == 0 || DesiredBPP == 36))
+				return 36;
+			else if (DecimalBPP >= 30 && (DesiredBPP == 0 || DesiredBPP == 30))
+				return 30;
+			else if (DecimalBPP >= 24 && (DesiredBPP == 0 || DesiredBPP == 24))
+				return 24;
+			else if (DecimalBPP >= 18 && (DesiredBPP == 0 || DesiredBPP == 18))
+				return 18;
+			else
+				return BPP_INVALID;
+		} else {
+			if (DecimalBPP / 1.5 >= 24 && (DesiredBPP == 0 || DesiredBPP == 24))
+				return 24;
+			else if (DecimalBPP / 1.5 >= 20 && (DesiredBPP == 0 || DesiredBPP == 20))
+				return 20;
+			else if (DecimalBPP / 1.5 >= 16 && (DesiredBPP == 0 || DesiredBPP == 16))
+				return 16;
+			else
+				return BPP_INVALID;
+		}
+	} else {
+		if (DSCEnabled) {
+			if (Format == dm_420) {
+				if (DesiredBPP == 0) {
+					if (DecimalBPP < 6)
+						return BPP_INVALID;
+					else if (DecimalBPP >= 1.5 * DSCInputBitPerComponent - 1.0 / 16.0)
+						return 1.5 * DSCInputBitPerComponent - 1.0 / 16.0;
+					else
+						return dml_floor(16 * DecimalBPP, 1) / 16.0;
+				} else {
+					if (DecimalBPP < 6
+							|| DesiredBPP < 6
+							|| DesiredBPP > 1.5 * DSCInputBitPerComponent - 1.0 / 16.0
+							|| DecimalBPP < DesiredBPP) {
+						return BPP_INVALID;
+					} else {
+						return DesiredBPP;
+					}
+				}
+			} else if (Format == dm_n422) {
+				if (DesiredBPP == 0) {
+					if (DecimalBPP < 7)
+						return BPP_INVALID;
+					else if (DecimalBPP >= 2 * DSCInputBitPerComponent - 1.0 / 16.0)
+						return 2 * DSCInputBitPerComponent - 1.0 / 16.0;
+					else
+						return dml_floor(16 * DecimalBPP, 1) / 16.0;
+				} else {
+					if (DecimalBPP < 7
+							|| DesiredBPP < 7
+							|| DesiredBPP > 2 * DSCInputBitPerComponent - 1.0 / 16.0
+							|| DecimalBPP < DesiredBPP) {
+						return BPP_INVALID;
+					} else {
+						return DesiredBPP;
+					}
+				}
+			} else {
+				if (DesiredBPP == 0) {
+					if (DecimalBPP < 8)
+						return BPP_INVALID;
+					else if (DecimalBPP >= 3 * DSCInputBitPerComponent - 1.0 / 16.0)
+						return 3 * DSCInputBitPerComponent - 1.0 / 16.0;
+					else
+						return dml_floor(16 * DecimalBPP, 1) / 16.0;
+				} else {
+					if (DecimalBPP < 8
+							|| DesiredBPP < 8
+							|| DesiredBPP > 3 * DSCInputBitPerComponent - 1.0 / 16.0
+							|| DecimalBPP < DesiredBPP) {
+						return BPP_INVALID;
+					} else {
+						return DesiredBPP;
+					}
+				}
+			}
+		} else if (Format == dm_420) {
+			if (DecimalBPP >= 18 && (DesiredBPP == 0 || DesiredBPP == 18))
+				return 18;
+			else if (DecimalBPP >= 15 && (DesiredBPP == 0 || DesiredBPP == 15))
+				return 15;
+			else if (DecimalBPP >= 12 && (DesiredBPP == 0 || DesiredBPP == 12))
+				return 12;
+			else
+				return BPP_INVALID;
+		} else if (Format == dm_s422 || Format == dm_n422) {
+			if (DecimalBPP >= 24 && (DesiredBPP == 0 || DesiredBPP == 24))
+				return 24;
+			else if (DecimalBPP >= 20 && (DesiredBPP == 0 || DesiredBPP == 20))
+				return 20;
+			else if (DecimalBPP >= 16 && (DesiredBPP == 0 || DesiredBPP == 16))
+				return 16;
+			else
+				return BPP_INVALID;
+		} else {
+			if (DecimalBPP >= 36 && (DesiredBPP == 0 || DesiredBPP == 36))
+				return 36;
+			else if (DecimalBPP >= 30 && (DesiredBPP == 0 || DesiredBPP == 30))
+				return 30;
+			else if (DecimalBPP >= 24 && (DesiredBPP == 0 || DesiredBPP == 24))
+				return 24;
+			else if (DecimalBPP >= 18 && (DesiredBPP == 0 || DesiredBPP == 18))
+				return 18;
+			else
+				return BPP_INVALID;
+		}
+	}
+}
+
+void dml20v2_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_lib)
+{
+	struct vba_vars_st *locals = &mode_lib->vba;
+
+	int i;
+	unsigned int j, k, m;
+
+	/*MODE SUPPORT, VOLTAGE STATE AND SOC CONFIGURATION*/
+
+	/*Scale Ratio, taps Support Check*/
+
+	mode_lib->vba.ScaleRatioAndTapsSupport = true;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.ScalerEnabled[k] == false
+				&& ((mode_lib->vba.SourcePixelFormat[k] != dm_444_64
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_444_32
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_444_16
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_mono_16
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_mono_8)
+						|| mode_lib->vba.HRatio[k] != 1.0
+						|| mode_lib->vba.htaps[k] != 1.0
+						|| mode_lib->vba.VRatio[k] != 1.0
+						|| mode_lib->vba.vtaps[k] != 1.0)) {
+			mode_lib->vba.ScaleRatioAndTapsSupport = false;
+		} else if (mode_lib->vba.vtaps[k] < 1.0 || mode_lib->vba.vtaps[k] > 8.0
+				|| mode_lib->vba.htaps[k] < 1.0 || mode_lib->vba.htaps[k] > 8.0
+				|| (mode_lib->vba.htaps[k] > 1.0
+						&& (mode_lib->vba.htaps[k] % 2) == 1)
+				|| mode_lib->vba.HRatio[k] > mode_lib->vba.MaxHSCLRatio
+				|| mode_lib->vba.VRatio[k] > mode_lib->vba.MaxVSCLRatio
+				|| mode_lib->vba.HRatio[k] > mode_lib->vba.htaps[k]
+				|| mode_lib->vba.VRatio[k] > mode_lib->vba.vtaps[k]
+				|| (mode_lib->vba.SourcePixelFormat[k] != dm_444_64
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_444_32
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_444_16
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_mono_16
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_mono_8
+						&& (mode_lib->vba.HRatio[k] / 2.0
+								> mode_lib->vba.HTAPsChroma[k]
+								|| mode_lib->vba.VRatio[k] / 2.0
+										> mode_lib->vba.VTAPsChroma[k]))) {
+			mode_lib->vba.ScaleRatioAndTapsSupport = false;
+		}
+	}
+	/*Source Format, Pixel Format and Scan Support Check*/
+
+	mode_lib->vba.SourceFormatPixelAndScanSupport = true;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if ((mode_lib->vba.SurfaceTiling[k] == dm_sw_linear
+				&& mode_lib->vba.SourceScan[k] != dm_horz)
+				|| ((mode_lib->vba.SurfaceTiling[k] == dm_sw_4kb_d
+						|| mode_lib->vba.SurfaceTiling[k] == dm_sw_4kb_d_x
+						|| mode_lib->vba.SurfaceTiling[k] == dm_sw_64kb_d
+						|| mode_lib->vba.SurfaceTiling[k] == dm_sw_64kb_d_t
+						|| mode_lib->vba.SurfaceTiling[k] == dm_sw_64kb_d_x
+						|| mode_lib->vba.SurfaceTiling[k] == dm_sw_var_d
+						|| mode_lib->vba.SurfaceTiling[k] == dm_sw_var_d_x)
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_444_64)
+				|| (mode_lib->vba.SurfaceTiling[k] == dm_sw_64kb_r_x
+						&& (mode_lib->vba.SourcePixelFormat[k] == dm_mono_8
+								|| mode_lib->vba.SourcePixelFormat[k]
+										== dm_420_8
+								|| mode_lib->vba.SourcePixelFormat[k]
+										== dm_420_10))
+				|| (((mode_lib->vba.SurfaceTiling[k] == dm_sw_gfx7_2d_thin_gl
+						|| mode_lib->vba.SurfaceTiling[k]
+								== dm_sw_gfx7_2d_thin_l_vp)
+						&& !((mode_lib->vba.SourcePixelFormat[k]
+								== dm_444_64
+								|| mode_lib->vba.SourcePixelFormat[k]
+										== dm_444_32)
+								&& mode_lib->vba.SourceScan[k]
+										== dm_horz
+								&& mode_lib->vba.SupportGFX7CompatibleTilingIn32bppAnd64bpp
+										== true
+								&& mode_lib->vba.DCCEnable[k]
+										== false))
+						|| (mode_lib->vba.DCCEnable[k] == true
+								&& (mode_lib->vba.SurfaceTiling[k]
+										== dm_sw_linear
+										|| mode_lib->vba.SourcePixelFormat[k]
+												== dm_420_8
+										|| mode_lib->vba.SourcePixelFormat[k]
+												== dm_420_10)))) {
+			mode_lib->vba.SourceFormatPixelAndScanSupport = false;
+		}
+	}
+	/*Bandwidth Support Check*/
+
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.SourcePixelFormat[k] == dm_444_64) {
+			locals->BytePerPixelInDETY[k] = 8.0;
+			locals->BytePerPixelInDETC[k] = 0.0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_32) {
+			locals->BytePerPixelInDETY[k] = 4.0;
+			locals->BytePerPixelInDETC[k] = 0.0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_16
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_mono_16) {
+			locals->BytePerPixelInDETY[k] = 2.0;
+			locals->BytePerPixelInDETC[k] = 0.0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_mono_8) {
+			locals->BytePerPixelInDETY[k] = 1.0;
+			locals->BytePerPixelInDETC[k] = 0.0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_8) {
+			locals->BytePerPixelInDETY[k] = 1.0;
+			locals->BytePerPixelInDETC[k] = 2.0;
+		} else {
+			locals->BytePerPixelInDETY[k] = 4.0 / 3;
+			locals->BytePerPixelInDETC[k] = 8.0 / 3;
+		}
+		if (mode_lib->vba.SourceScan[k] == dm_horz) {
+			locals->SwathWidthYSingleDPP[k] = mode_lib->vba.ViewportWidth[k];
+		} else {
+			locals->SwathWidthYSingleDPP[k] = mode_lib->vba.ViewportHeight[k];
+		}
+	}
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		locals->ReadBandwidthLuma[k] = locals->SwathWidthYSingleDPP[k] * dml_ceil(locals->BytePerPixelInDETY[k], 1.0)
+				/ (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k]) * mode_lib->vba.VRatio[k];
+		locals->ReadBandwidthChroma[k] = locals->SwathWidthYSingleDPP[k] / 2 * dml_ceil(locals->BytePerPixelInDETC[k], 2.0)
+				/ (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k]) * mode_lib->vba.VRatio[k] / 2.0;
+		locals->ReadBandwidth[k] = locals->ReadBandwidthLuma[k] + locals->ReadBandwidthChroma[k];
+	}
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.WritebackEnable[k] == true
+				&& mode_lib->vba.WritebackPixelFormat[k] == dm_444_32) {
+			locals->WriteBandwidth[k] = mode_lib->vba.WritebackDestinationWidth[k]
+					* mode_lib->vba.WritebackDestinationHeight[k]
+					/ (mode_lib->vba.WritebackSourceHeight[k]
+							* mode_lib->vba.HTotal[k]
+							/ mode_lib->vba.PixelClock[k]) * 4.0;
+		} else if (mode_lib->vba.WritebackEnable[k] == true
+				&& mode_lib->vba.WritebackPixelFormat[k] == dm_420_10) {
+			locals->WriteBandwidth[k] = mode_lib->vba.WritebackDestinationWidth[k]
+					* mode_lib->vba.WritebackDestinationHeight[k]
+					/ (mode_lib->vba.WritebackSourceHeight[k]
+							* mode_lib->vba.HTotal[k]
+							/ mode_lib->vba.PixelClock[k]) * 3.0;
+		} else if (mode_lib->vba.WritebackEnable[k] == true) {
+			locals->WriteBandwidth[k] = mode_lib->vba.WritebackDestinationWidth[k]
+					* mode_lib->vba.WritebackDestinationHeight[k]
+					/ (mode_lib->vba.WritebackSourceHeight[k]
+							* mode_lib->vba.HTotal[k]
+							/ mode_lib->vba.PixelClock[k]) * 1.5;
+		} else {
+			locals->WriteBandwidth[k] = 0.0;
+		}
+	}
+	mode_lib->vba.DCCEnabledInAnyPlane = false;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.DCCEnable[k] == true) {
+			mode_lib->vba.DCCEnabledInAnyPlane = true;
+		}
+	}
+	mode_lib->vba.UrgentLatency = mode_lib->vba.UrgentLatencyPixelDataOnly;
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		locals->FabricAndDRAMBandwidthPerState[i] = dml_min(
+				mode_lib->vba.DRAMSpeedPerState[i] * mode_lib->vba.NumberOfChannels
+						* mode_lib->vba.DRAMChannelWidth,
+				mode_lib->vba.FabricClockPerState[i]
+						* mode_lib->vba.FabricDatapathToDCNDataReturn) / 1000;
+		locals->ReturnBWToDCNPerState = dml_min(locals->ReturnBusWidth * locals->DCFCLKPerState[i],
+				locals->FabricAndDRAMBandwidthPerState[i] * 1000)
+				* locals->PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelDataOnly / 100;
+
+		locals->ReturnBWPerState[i][0] = locals->ReturnBWToDCNPerState;
+
+		if (locals->DCCEnabledInAnyPlane == true && locals->ReturnBWToDCNPerState > locals->DCFCLKPerState[i] * locals->ReturnBusWidth / 4) {
+			locals->ReturnBWPerState[i][0] = dml_min(locals->ReturnBWPerState[i][0],
+					locals->ReturnBWToDCNPerState * 4 * (1 - locals->UrgentLatency /
+					((locals->ROBBufferSizeInKByte - locals->PixelChunkSizeInKByte) * 1024
+					/ (locals->ReturnBWToDCNPerState - locals->DCFCLKPerState[i]
+					* locals->ReturnBusWidth / 4) + locals->UrgentLatency)));
+		}
+		locals->CriticalPoint = 2 * locals->ReturnBusWidth * locals->DCFCLKPerState[i] *
+				locals->UrgentLatency / (locals->ReturnBWToDCNPerState * locals->UrgentLatency
+				+ (locals->ROBBufferSizeInKByte - locals->PixelChunkSizeInKByte) * 1024);
+
+		if (locals->DCCEnabledInAnyPlane && locals->CriticalPoint > 1 && locals->CriticalPoint < 4) {
+			locals->ReturnBWPerState[i][0] = dml_min(locals->ReturnBWPerState[i][0],
+				4 * locals->ReturnBWToDCNPerState *
+				(locals->ROBBufferSizeInKByte - locals->PixelChunkSizeInKByte) * 1024
+				* locals->ReturnBusWidth * locals->DCFCLKPerState[i] * locals->UrgentLatency /
+				dml_pow((locals->ReturnBWToDCNPerState * locals->UrgentLatency
+				+ (locals->ROBBufferSizeInKByte - locals->PixelChunkSizeInKByte) * 1024), 2));
+		}
+
+		locals->ReturnBWToDCNPerState = dml_min(locals->ReturnBusWidth *
+				locals->DCFCLKPerState[i], locals->FabricAndDRAMBandwidthPerState[i] * 1000);
+
+		if (locals->DCCEnabledInAnyPlane == true && locals->ReturnBWToDCNPerState > locals->DCFCLKPerState[i] * locals->ReturnBusWidth / 4) {
+			locals->ReturnBWPerState[i][0] = dml_min(locals->ReturnBWPerState[i][0],
+					locals->ReturnBWToDCNPerState * 4 * (1 - locals->UrgentLatency /
+					((locals->ROBBufferSizeInKByte - locals->PixelChunkSizeInKByte) * 1024
+					/ (locals->ReturnBWToDCNPerState - locals->DCFCLKPerState[i]
+					* locals->ReturnBusWidth / 4) + locals->UrgentLatency)));
+		}
+		locals->CriticalPoint = 2 * locals->ReturnBusWidth * locals->DCFCLKPerState[i] *
+				locals->UrgentLatency / (locals->ReturnBWToDCNPerState * locals->UrgentLatency
+				+ (locals->ROBBufferSizeInKByte - locals->PixelChunkSizeInKByte) * 1024);
+
+		if (locals->DCCEnabledInAnyPlane && locals->CriticalPoint > 1 && locals->CriticalPoint < 4) {
+			locals->ReturnBWPerState[i][0] = dml_min(locals->ReturnBWPerState[i][0],
+				4 * locals->ReturnBWToDCNPerState *
+				(locals->ROBBufferSizeInKByte - locals->PixelChunkSizeInKByte) * 1024
+				* locals->ReturnBusWidth * locals->DCFCLKPerState[i] * locals->UrgentLatency /
+				dml_pow((locals->ReturnBWToDCNPerState * locals->UrgentLatency
+				+ (locals->ROBBufferSizeInKByte - locals->PixelChunkSizeInKByte) * 1024), 2));
+		}
+	}
+	/*Writeback Latency support check*/
+
+	mode_lib->vba.WritebackLatencySupport = true;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.WritebackEnable[k] == true) {
+			if (mode_lib->vba.WritebackPixelFormat[k] == dm_444_32) {
+				if (locals->WriteBandwidth[k]
+						> (mode_lib->vba.WritebackInterfaceLumaBufferSize
+								+ mode_lib->vba.WritebackInterfaceChromaBufferSize)
+								/ mode_lib->vba.WritebackLatency) {
+					mode_lib->vba.WritebackLatencySupport = false;
+				}
+			} else {
+				if (locals->WriteBandwidth[k]
+						> 1.5
+								* dml_min(
+										mode_lib->vba.WritebackInterfaceLumaBufferSize,
+										2.0
+												* mode_lib->vba.WritebackInterfaceChromaBufferSize)
+								/ mode_lib->vba.WritebackLatency) {
+					mode_lib->vba.WritebackLatencySupport = false;
+				}
+			}
+		}
+	}
+	/*Re-ordering Buffer Support Check*/
+
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		locals->UrgentRoundTripAndOutOfOrderLatencyPerState[i] =
+				(mode_lib->vba.RoundTripPingLatencyCycles + 32.0) / mode_lib->vba.DCFCLKPerState[i]
+				+ locals->UrgentOutOfOrderReturnPerChannel * mode_lib->vba.NumberOfChannels / locals->ReturnBWPerState[i][0];
+		if ((mode_lib->vba.ROBBufferSizeInKByte - mode_lib->vba.PixelChunkSizeInKByte) * 1024.0 / locals->ReturnBWPerState[i][0]
+				> locals->UrgentRoundTripAndOutOfOrderLatencyPerState[i]) {
+			locals->ROBSupport[i][0] = true;
+		} else {
+			locals->ROBSupport[i][0] = false;
+		}
+	}
+	/*Writeback Mode Support Check*/
+
+	mode_lib->vba.TotalNumberOfActiveWriteback = 0;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.WritebackEnable[k] == true) {
+			if (mode_lib->vba.ActiveWritebacksPerPlane[k] == 0)
+				mode_lib->vba.ActiveWritebacksPerPlane[k] = 1;
+			mode_lib->vba.TotalNumberOfActiveWriteback =
+					mode_lib->vba.TotalNumberOfActiveWriteback
+							+ mode_lib->vba.ActiveWritebacksPerPlane[k];
+		}
+	}
+	mode_lib->vba.WritebackModeSupport = true;
+	if (mode_lib->vba.TotalNumberOfActiveWriteback > mode_lib->vba.MaxNumWriteback) {
+		mode_lib->vba.WritebackModeSupport = false;
+	}
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.WritebackEnable[k] == true
+				&& mode_lib->vba.Writeback10bpc420Supported != true
+				&& mode_lib->vba.WritebackPixelFormat[k] == dm_420_10) {
+			mode_lib->vba.WritebackModeSupport = false;
+		}
+	}
+	/*Writeback Scale Ratio and Taps Support Check*/
+
+	mode_lib->vba.WritebackScaleRatioAndTapsSupport = true;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.WritebackEnable[k] == true) {
+			if (mode_lib->vba.WritebackLumaAndChromaScalingSupported == false
+					&& (mode_lib->vba.WritebackHRatio[k] != 1.0
+							|| mode_lib->vba.WritebackVRatio[k] != 1.0)) {
+				mode_lib->vba.WritebackScaleRatioAndTapsSupport = false;
+			}
+			if (mode_lib->vba.WritebackHRatio[k] > mode_lib->vba.WritebackMaxHSCLRatio
+					|| mode_lib->vba.WritebackVRatio[k]
+							> mode_lib->vba.WritebackMaxVSCLRatio
+					|| mode_lib->vba.WritebackHRatio[k]
+							< mode_lib->vba.WritebackMinHSCLRatio
+					|| mode_lib->vba.WritebackVRatio[k]
+							< mode_lib->vba.WritebackMinVSCLRatio
+					|| mode_lib->vba.WritebackLumaHTaps[k]
+							> mode_lib->vba.WritebackMaxHSCLTaps
+					|| mode_lib->vba.WritebackLumaVTaps[k]
+							> mode_lib->vba.WritebackMaxVSCLTaps
+					|| mode_lib->vba.WritebackHRatio[k]
+							> mode_lib->vba.WritebackLumaHTaps[k]
+					|| mode_lib->vba.WritebackVRatio[k]
+							> mode_lib->vba.WritebackLumaVTaps[k]
+					|| (mode_lib->vba.WritebackLumaHTaps[k] > 2.0
+							&& ((mode_lib->vba.WritebackLumaHTaps[k] % 2)
+									== 1))
+					|| (mode_lib->vba.WritebackPixelFormat[k] != dm_444_32
+							&& (mode_lib->vba.WritebackChromaHTaps[k]
+									> mode_lib->vba.WritebackMaxHSCLTaps
+									|| mode_lib->vba.WritebackChromaVTaps[k]
+											> mode_lib->vba.WritebackMaxVSCLTaps
+									|| 2.0
+											* mode_lib->vba.WritebackHRatio[k]
+											> mode_lib->vba.WritebackChromaHTaps[k]
+									|| 2.0
+											* mode_lib->vba.WritebackVRatio[k]
+											> mode_lib->vba.WritebackChromaVTaps[k]
+									|| (mode_lib->vba.WritebackChromaHTaps[k] > 2.0
+										&& ((mode_lib->vba.WritebackChromaHTaps[k] % 2) == 1))))) {
+				mode_lib->vba.WritebackScaleRatioAndTapsSupport = false;
+			}
+			if (mode_lib->vba.WritebackVRatio[k] < 1.0) {
+				mode_lib->vba.WritebackLumaVExtra =
+						dml_max(1.0 - 2.0 / dml_ceil(1.0 / mode_lib->vba.WritebackVRatio[k], 1.0), 0.0);
+			} else {
+				mode_lib->vba.WritebackLumaVExtra = -1;
+			}
+			if ((mode_lib->vba.WritebackPixelFormat[k] == dm_444_32
+					&& mode_lib->vba.WritebackLumaVTaps[k]
+							> (mode_lib->vba.WritebackLineBufferLumaBufferSize
+									+ mode_lib->vba.WritebackLineBufferChromaBufferSize)
+									/ 3.0
+									/ mode_lib->vba.WritebackDestinationWidth[k]
+									- mode_lib->vba.WritebackLumaVExtra)
+					|| (mode_lib->vba.WritebackPixelFormat[k] == dm_420_8
+							&& mode_lib->vba.WritebackLumaVTaps[k]
+									> mode_lib->vba.WritebackLineBufferLumaBufferSize
+											* 8.0 / 10.0 / mode_lib->vba.WritebackDestinationWidth[k]
+											- mode_lib->vba.WritebackLumaVExtra)
+					|| (mode_lib->vba.WritebackPixelFormat[k] == dm_420_10
+							&& mode_lib->vba.WritebackLumaVTaps[k]
+									> mode_lib->vba.WritebackLineBufferLumaBufferSize
+											* 8.0 / 10.0
+											/ mode_lib->vba.WritebackDestinationWidth[k]
+											- mode_lib->vba.WritebackLumaVExtra)) {
+				mode_lib->vba.WritebackScaleRatioAndTapsSupport = false;
+			}
+			if (2.0 * mode_lib->vba.WritebackVRatio[k] < 1) {
+				mode_lib->vba.WritebackChromaVExtra = 0.0;
+			} else {
+				mode_lib->vba.WritebackChromaVExtra = -1;
+			}
+			if ((mode_lib->vba.WritebackPixelFormat[k] == dm_420_8
+					&& mode_lib->vba.WritebackChromaVTaps[k]
+							> mode_lib->vba.WritebackLineBufferChromaBufferSize
+									* 8.0 / 10.0 / mode_lib->vba.WritebackDestinationWidth[k]
+									- mode_lib->vba.WritebackChromaVExtra)
+					|| (mode_lib->vba.WritebackPixelFormat[k] == dm_420_10
+							&& mode_lib->vba.WritebackChromaVTaps[k]
+									> mode_lib->vba.WritebackLineBufferChromaBufferSize
+											* 8.0 / 10.0
+											/ mode_lib->vba.WritebackDestinationWidth[k]
+											- mode_lib->vba.WritebackChromaVExtra)) {
+				mode_lib->vba.WritebackScaleRatioAndTapsSupport = false;
+			}
+		}
+	}
+	/*Maximum DISPCLK/DPPCLK Support check*/
+
+	mode_lib->vba.WritebackRequiredDISPCLK = 0.0;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.WritebackEnable[k] == true) {
+			mode_lib->vba.WritebackRequiredDISPCLK =
+					dml_max(
+							mode_lib->vba.WritebackRequiredDISPCLK,
+							CalculateWriteBackDISPCLK(
+									mode_lib->vba.WritebackPixelFormat[k],
+									mode_lib->vba.PixelClock[k],
+									mode_lib->vba.WritebackHRatio[k],
+									mode_lib->vba.WritebackVRatio[k],
+									mode_lib->vba.WritebackLumaHTaps[k],
+									mode_lib->vba.WritebackLumaVTaps[k],
+									mode_lib->vba.WritebackChromaHTaps[k],
+									mode_lib->vba.WritebackChromaVTaps[k],
+									mode_lib->vba.WritebackDestinationWidth[k],
+									mode_lib->vba.HTotal[k],
+									mode_lib->vba.WritebackChromaLineBufferWidth));
+		}
+	}
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.HRatio[k] > 1.0) {
+			locals->PSCL_FACTOR[k] = dml_min(
+					mode_lib->vba.MaxDCHUBToPSCLThroughput,
+					mode_lib->vba.MaxPSCLToLBThroughput
+							* mode_lib->vba.HRatio[k]
+							/ dml_ceil(
+									mode_lib->vba.htaps[k]
+											/ 6.0,
+									1.0));
+		} else {
+			locals->PSCL_FACTOR[k] = dml_min(
+					mode_lib->vba.MaxDCHUBToPSCLThroughput,
+					mode_lib->vba.MaxPSCLToLBThroughput);
+		}
+		if (locals->BytePerPixelInDETC[k] == 0.0) {
+			locals->PSCL_FACTOR_CHROMA[k] = 0.0;
+			locals->MinDPPCLKUsingSingleDPP[k] =
+					mode_lib->vba.PixelClock[k]
+							* dml_max3(
+									mode_lib->vba.vtaps[k] / 6.0
+											* dml_min(
+													1.0,
+													mode_lib->vba.HRatio[k]),
+									mode_lib->vba.HRatio[k]
+											* mode_lib->vba.VRatio[k]
+											/ locals->PSCL_FACTOR[k],
+									1.0);
+			if ((mode_lib->vba.htaps[k] > 6.0 || mode_lib->vba.vtaps[k] > 6.0)
+					&& locals->MinDPPCLKUsingSingleDPP[k]
+							< 2.0 * mode_lib->vba.PixelClock[k]) {
+				locals->MinDPPCLKUsingSingleDPP[k] = 2.0
+						* mode_lib->vba.PixelClock[k];
+			}
+		} else {
+			if (mode_lib->vba.HRatio[k] / 2.0 > 1.0) {
+				locals->PSCL_FACTOR_CHROMA[k] =
+						dml_min(
+								mode_lib->vba.MaxDCHUBToPSCLThroughput,
+								mode_lib->vba.MaxPSCLToLBThroughput
+										* mode_lib->vba.HRatio[k]
+										/ 2.0
+										/ dml_ceil(
+												mode_lib->vba.HTAPsChroma[k]
+														/ 6.0,
+												1.0));
+			} else {
+				locals->PSCL_FACTOR_CHROMA[k] = dml_min(
+						mode_lib->vba.MaxDCHUBToPSCLThroughput,
+						mode_lib->vba.MaxPSCLToLBThroughput);
+			}
+			locals->MinDPPCLKUsingSingleDPP[k] =
+					mode_lib->vba.PixelClock[k]
+							* dml_max5(
+									mode_lib->vba.vtaps[k] / 6.0
+											* dml_min(
+													1.0,
+													mode_lib->vba.HRatio[k]),
+									mode_lib->vba.HRatio[k]
+											* mode_lib->vba.VRatio[k]
+											/ locals->PSCL_FACTOR[k],
+									mode_lib->vba.VTAPsChroma[k]
+											/ 6.0
+											* dml_min(
+													1.0,
+													mode_lib->vba.HRatio[k]
+															/ 2.0),
+									mode_lib->vba.HRatio[k]
+											* mode_lib->vba.VRatio[k]
+											/ 4.0
+											/ locals->PSCL_FACTOR_CHROMA[k],
+									1.0);
+			if ((mode_lib->vba.htaps[k] > 6.0 || mode_lib->vba.vtaps[k] > 6.0
+					|| mode_lib->vba.HTAPsChroma[k] > 6.0
+					|| mode_lib->vba.VTAPsChroma[k] > 6.0)
+					&& locals->MinDPPCLKUsingSingleDPP[k]
+							< 2.0 * mode_lib->vba.PixelClock[k]) {
+				locals->MinDPPCLKUsingSingleDPP[k] = 2.0
+						* mode_lib->vba.PixelClock[k];
+			}
+		}
+	}
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		Calculate256BBlockSizes(
+				mode_lib->vba.SourcePixelFormat[k],
+				mode_lib->vba.SurfaceTiling[k],
+				dml_ceil(locals->BytePerPixelInDETY[k], 1.0),
+				dml_ceil(locals->BytePerPixelInDETC[k], 2.0),
+				&locals->Read256BlockHeightY[k],
+				&locals->Read256BlockHeightC[k],
+				&locals->Read256BlockWidthY[k],
+				&locals->Read256BlockWidthC[k]);
+		if (mode_lib->vba.SourceScan[k] == dm_horz) {
+			locals->MaxSwathHeightY[k] = locals->Read256BlockHeightY[k];
+			locals->MaxSwathHeightC[k] = locals->Read256BlockHeightC[k];
+		} else {
+			locals->MaxSwathHeightY[k] = locals->Read256BlockWidthY[k];
+			locals->MaxSwathHeightC[k] = locals->Read256BlockWidthC[k];
+		}
+		if ((mode_lib->vba.SourcePixelFormat[k] == dm_444_64
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_32
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_16
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_mono_16
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_mono_8)) {
+			if (mode_lib->vba.SurfaceTiling[k] == dm_sw_linear
+					|| (mode_lib->vba.SourcePixelFormat[k] == dm_444_64
+							&& (mode_lib->vba.SurfaceTiling[k]
+									== dm_sw_4kb_s
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_4kb_s_x
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_64kb_s
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_64kb_s_t
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_64kb_s_x
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_var_s
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_var_s_x)
+							&& mode_lib->vba.SourceScan[k] == dm_horz)) {
+				locals->MinSwathHeightY[k] = locals->MaxSwathHeightY[k];
+			} else {
+				locals->MinSwathHeightY[k] = locals->MaxSwathHeightY[k]
+						/ 2.0;
+			}
+			locals->MinSwathHeightC[k] = locals->MaxSwathHeightC[k];
+		} else {
+			if (mode_lib->vba.SurfaceTiling[k] == dm_sw_linear) {
+				locals->MinSwathHeightY[k] = locals->MaxSwathHeightY[k];
+				locals->MinSwathHeightC[k] = locals->MaxSwathHeightC[k];
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_8
+					&& mode_lib->vba.SourceScan[k] == dm_horz) {
+				locals->MinSwathHeightY[k] = locals->MaxSwathHeightY[k]
+						/ 2.0;
+				locals->MinSwathHeightC[k] = locals->MaxSwathHeightC[k];
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_10
+					&& mode_lib->vba.SourceScan[k] == dm_horz) {
+				locals->MinSwathHeightC[k] = locals->MaxSwathHeightC[k]
+						/ 2.0;
+				locals->MinSwathHeightY[k] = locals->MaxSwathHeightY[k];
+			} else {
+				locals->MinSwathHeightY[k] = locals->MaxSwathHeightY[k];
+				locals->MinSwathHeightC[k] = locals->MaxSwathHeightC[k];
+			}
+		}
+		if (mode_lib->vba.SurfaceTiling[k] == dm_sw_linear) {
+			mode_lib->vba.MaximumSwathWidthSupport = 8192.0;
+		} else {
+			mode_lib->vba.MaximumSwathWidthSupport = 5120.0;
+		}
+		mode_lib->vba.MaximumSwathWidthInDETBuffer =
+				dml_min(
+						mode_lib->vba.MaximumSwathWidthSupport,
+						mode_lib->vba.DETBufferSizeInKByte[0] * 1024.0 / 2.0
+								/ (locals->BytePerPixelInDETY[k]
+										* locals->MinSwathHeightY[k]
+										+ locals->BytePerPixelInDETC[k]
+												/ 2.0
+												* locals->MinSwathHeightC[k]));
+		if (locals->BytePerPixelInDETC[k] == 0.0) {
+			mode_lib->vba.MaximumSwathWidthInLineBuffer =
+					mode_lib->vba.LineBufferSize
+							* dml_max(mode_lib->vba.HRatio[k], 1.0)
+							/ mode_lib->vba.LBBitPerPixel[k]
+							/ (mode_lib->vba.vtaps[k]
+									+ dml_max(
+											dml_ceil(
+													mode_lib->vba.VRatio[k],
+													1.0)
+													- 2,
+											0.0));
+		} else {
+			mode_lib->vba.MaximumSwathWidthInLineBuffer =
+					dml_min(
+							mode_lib->vba.LineBufferSize
+									* dml_max(
+											mode_lib->vba.HRatio[k],
+											1.0)
+									/ mode_lib->vba.LBBitPerPixel[k]
+									/ (mode_lib->vba.vtaps[k]
+											+ dml_max(
+													dml_ceil(
+															mode_lib->vba.VRatio[k],
+															1.0)
+															- 2,
+													0.0)),
+							2.0 * mode_lib->vba.LineBufferSize
+									* dml_max(
+											mode_lib->vba.HRatio[k]
+													/ 2.0,
+											1.0)
+									/ mode_lib->vba.LBBitPerPixel[k]
+									/ (mode_lib->vba.VTAPsChroma[k]
+											+ dml_max(
+													dml_ceil(
+															mode_lib->vba.VRatio[k]
+																	/ 2.0,
+															1.0)
+															- 2,
+													0.0)));
+		}
+		locals->MaximumSwathWidth[k] = dml_min(
+				mode_lib->vba.MaximumSwathWidthInDETBuffer,
+				mode_lib->vba.MaximumSwathWidthInLineBuffer);
+	}
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		double MaxMaxDispclkRoundedDown = RoundToDFSGranularityDown(
+			mode_lib->vba.MaxDispclk[mode_lib->vba.soc.num_states],
+			mode_lib->vba.DISPCLKDPPCLKVCOSpeed);
+
+		for (j = 0; j < 2; j++) {
+			mode_lib->vba.MaxDispclkRoundedDownToDFSGranularity = RoundToDFSGranularityDown(
+				mode_lib->vba.MaxDispclk[i],
+				mode_lib->vba.DISPCLKDPPCLKVCOSpeed);
+			mode_lib->vba.MaxDppclkRoundedDownToDFSGranularity = RoundToDFSGranularityDown(
+				mode_lib->vba.MaxDppclk[i],
+				mode_lib->vba.DISPCLKDPPCLKVCOSpeed);
+			locals->RequiredDISPCLK[i][j] = 0.0;
+			locals->DISPCLK_DPPCLK_Support[i][j] = true;
+			for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+				mode_lib->vba.PlaneRequiredDISPCLKWithoutODMCombine =
+						mode_lib->vba.PixelClock[k] * (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0)
+								* (1.0 + mode_lib->vba.DISPCLKRampingMargin / 100.0);
+				if (mode_lib->vba.PlaneRequiredDISPCLKWithoutODMCombine >= mode_lib->vba.MaxDispclk[i]
+						&& i == mode_lib->vba.soc.num_states)
+					mode_lib->vba.PlaneRequiredDISPCLKWithoutODMCombine = mode_lib->vba.PixelClock[k]
+							* (1 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0);
+
+				mode_lib->vba.PlaneRequiredDISPCLKWithODMCombine = mode_lib->vba.PixelClock[k] / 2
+					* (1 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) * (1 + mode_lib->vba.DISPCLKRampingMargin / 100.0);
+				if (mode_lib->vba.PlaneRequiredDISPCLKWithODMCombine >= mode_lib->vba.MaxDispclk[i]
+						&& i == mode_lib->vba.soc.num_states)
+					mode_lib->vba.PlaneRequiredDISPCLKWithODMCombine = mode_lib->vba.PixelClock[k] / 2
+							* (1 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0);
+
+				locals->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_disabled;
+				mode_lib->vba.PlaneRequiredDISPCLK = mode_lib->vba.PlaneRequiredDISPCLKWithoutODMCombine;
+				if (mode_lib->vba.ODMCapability) {
+					if (locals->PlaneRequiredDISPCLKWithoutODMCombine > MaxMaxDispclkRoundedDown) {
+						locals->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_2to1;
+						mode_lib->vba.PlaneRequiredDISPCLK = mode_lib->vba.PlaneRequiredDISPCLKWithODMCombine;
+					} else if (locals->DSCEnabled[k] && (locals->HActive[k] > DCN20_MAX_DSC_IMAGE_WIDTH)) {
+						locals->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_2to1;
+						mode_lib->vba.PlaneRequiredDISPCLK = mode_lib->vba.PlaneRequiredDISPCLKWithODMCombine;
+					} else if (locals->HActive[k] > DCN20_MAX_420_IMAGE_WIDTH && locals->OutputFormat[k] == dm_420) {
+						locals->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_2to1;
+						mode_lib->vba.PlaneRequiredDISPCLK = mode_lib->vba.PlaneRequiredDISPCLKWithODMCombine;
+					}
+				}
+
+				if (locals->MinDPPCLKUsingSingleDPP[k] * (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) <= mode_lib->vba.MaxDppclkRoundedDownToDFSGranularity
+						&& locals->SwathWidthYSingleDPP[k] <= locals->MaximumSwathWidth[k]
+						&& locals->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_disabled) {
+					locals->NoOfDPP[i][j][k] = 1;
+					locals->RequiredDPPCLK[i][j][k] =
+						locals->MinDPPCLKUsingSingleDPP[k] * (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0);
+				} else {
+					locals->NoOfDPP[i][j][k] = 2;
+					locals->RequiredDPPCLK[i][j][k] =
+						locals->MinDPPCLKUsingSingleDPP[k] * (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) / 2.0;
+				}
+				locals->RequiredDISPCLK[i][j] = dml_max(
+						locals->RequiredDISPCLK[i][j],
+						mode_lib->vba.PlaneRequiredDISPCLK);
+				if ((locals->MinDPPCLKUsingSingleDPP[k] / locals->NoOfDPP[i][j][k] * (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0)
+						> mode_lib->vba.MaxDppclkRoundedDownToDFSGranularity)
+						|| (mode_lib->vba.PlaneRequiredDISPCLK > mode_lib->vba.MaxDispclkRoundedDownToDFSGranularity)) {
+					locals->DISPCLK_DPPCLK_Support[i][j] = false;
+				}
+			}
+			locals->TotalNumberOfActiveDPP[i][j] = 0.0;
+			for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++)
+				locals->TotalNumberOfActiveDPP[i][j] = locals->TotalNumberOfActiveDPP[i][j] + locals->NoOfDPP[i][j][k];
+			if (j == 1) {
+				while (locals->TotalNumberOfActiveDPP[i][j] < mode_lib->vba.MaxNumDPP
+						&& locals->TotalNumberOfActiveDPP[i][j] < 2 * mode_lib->vba.NumberOfActivePlanes) {
+					double BWOfNonSplitPlaneOfMaximumBandwidth;
+					unsigned int NumberOfNonSplitPlaneOfMaximumBandwidth;
+
+					BWOfNonSplitPlaneOfMaximumBandwidth = 0;
+					NumberOfNonSplitPlaneOfMaximumBandwidth = 0;
+					for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+						if (locals->ReadBandwidth[k] > BWOfNonSplitPlaneOfMaximumBandwidth && locals->NoOfDPP[i][j][k] == 1) {
+							BWOfNonSplitPlaneOfMaximumBandwidth = locals->ReadBandwidth[k];
+							NumberOfNonSplitPlaneOfMaximumBandwidth = k;
+						}
+					}
+					locals->NoOfDPP[i][j][NumberOfNonSplitPlaneOfMaximumBandwidth] = 2;
+					locals->RequiredDPPCLK[i][j][NumberOfNonSplitPlaneOfMaximumBandwidth] =
+						locals->MinDPPCLKUsingSingleDPP[NumberOfNonSplitPlaneOfMaximumBandwidth]
+							* (1 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100) / 2;
+					locals->TotalNumberOfActiveDPP[i][j] = locals->TotalNumberOfActiveDPP[i][j] + 1;
+				}
+			}
+			if (locals->TotalNumberOfActiveDPP[i][j] > mode_lib->vba.MaxNumDPP) {
+				locals->RequiredDISPCLK[i][j] = 0.0;
+				locals->DISPCLK_DPPCLK_Support[i][j] = true;
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+					locals->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_disabled;
+					if (locals->SwathWidthYSingleDPP[k] <= locals->MaximumSwathWidth[k]) {
+						locals->NoOfDPP[i][j][k] = 1;
+						locals->RequiredDPPCLK[i][j][k] = locals->MinDPPCLKUsingSingleDPP[k]
+							* (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0);
+					} else {
+						locals->NoOfDPP[i][j][k] = 2;
+						locals->RequiredDPPCLK[i][j][k] = locals->MinDPPCLKUsingSingleDPP[k]
+										* (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) / 2.0;
+					}
+					if (i != mode_lib->vba.soc.num_states) {
+						mode_lib->vba.PlaneRequiredDISPCLK =
+								mode_lib->vba.PixelClock[k]
+										* (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0)
+										* (1.0 + mode_lib->vba.DISPCLKRampingMargin / 100.0);
+					} else {
+						mode_lib->vba.PlaneRequiredDISPCLK = mode_lib->vba.PixelClock[k]
+							* (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0);
+					}
+					locals->RequiredDISPCLK[i][j] = dml_max(
+							locals->RequiredDISPCLK[i][j],
+							mode_lib->vba.PlaneRequiredDISPCLK);
+					if (locals->MinDPPCLKUsingSingleDPP[k] / locals->NoOfDPP[i][j][k] * (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0)
+							> mode_lib->vba.MaxDppclkRoundedDownToDFSGranularity
+							|| mode_lib->vba.PlaneRequiredDISPCLK > mode_lib->vba.MaxDispclkRoundedDownToDFSGranularity)
+						locals->DISPCLK_DPPCLK_Support[i][j] = false;
+				}
+				locals->TotalNumberOfActiveDPP[i][j] = 0.0;
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++)
+					locals->TotalNumberOfActiveDPP[i][j] = locals->TotalNumberOfActiveDPP[i][j] + locals->NoOfDPP[i][j][k];
+			}
+			locals->RequiredDISPCLK[i][j] = dml_max(
+					locals->RequiredDISPCLK[i][j],
+					mode_lib->vba.WritebackRequiredDISPCLK);
+			if (mode_lib->vba.MaxDispclkRoundedDownToDFSGranularity
+					< mode_lib->vba.WritebackRequiredDISPCLK) {
+				locals->DISPCLK_DPPCLK_Support[i][j] = false;
+			}
+		}
+	}
+	/*Viewport Size Check*/
+
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		locals->ViewportSizeSupport[i][0] = true;
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			if (locals->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_2to1) {
+				if (dml_min(locals->SwathWidthYSingleDPP[k], dml_round(mode_lib->vba.HActive[k] / 2.0 * mode_lib->vba.HRatio[k]))
+						> locals->MaximumSwathWidth[k]) {
+					locals->ViewportSizeSupport[i][0] = false;
+				}
+			} else {
+				if (locals->SwathWidthYSingleDPP[k] / 2.0 > locals->MaximumSwathWidth[k]) {
+					locals->ViewportSizeSupport[i][0] = false;
+				}
+			}
+		}
+	}
+	/*Total Available Pipes Support Check*/
+
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		for (j = 0; j < 2; j++) {
+			if (locals->TotalNumberOfActiveDPP[i][j] <= mode_lib->vba.MaxNumDPP)
+				locals->TotalAvailablePipesSupport[i][j] = true;
+			else
+				locals->TotalAvailablePipesSupport[i][j] = false;
+		}
+	}
+	/*Total Available OTG Support Check*/
+
+	mode_lib->vba.TotalNumberOfActiveOTG = 0.0;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.BlendingAndTiming[k] == k) {
+			mode_lib->vba.TotalNumberOfActiveOTG = mode_lib->vba.TotalNumberOfActiveOTG
+					+ 1.0;
+		}
+	}
+	if (mode_lib->vba.TotalNumberOfActiveOTG <= mode_lib->vba.MaxNumOTG) {
+		mode_lib->vba.NumberOfOTGSupport = true;
+	} else {
+		mode_lib->vba.NumberOfOTGSupport = false;
+	}
+	/*Display IO and DSC Support Check*/
+
+	mode_lib->vba.NonsupportedDSCInputBPC = false;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (!(mode_lib->vba.DSCInputBitPerComponent[k] == 12.0
+				|| mode_lib->vba.DSCInputBitPerComponent[k] == 10.0
+				|| mode_lib->vba.DSCInputBitPerComponent[k] == 8.0)) {
+			mode_lib->vba.NonsupportedDSCInputBPC = true;
+		}
+	}
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			locals->RequiresDSC[i][k] = 0;
+			locals->RequiresFEC[i][k] = 0;
+			if (mode_lib->vba.BlendingAndTiming[k] == k) {
+				if (mode_lib->vba.Output[k] == dm_hdmi) {
+					locals->RequiresDSC[i][k] = 0;
+					locals->RequiresFEC[i][k] = 0;
+					locals->OutputBppPerState[i][k] = TruncToValidBPP(
+							dml_min(600.0, mode_lib->vba.PHYCLKPerState[i]) / mode_lib->vba.PixelClockBackEnd[k] * 24,
+							mode_lib->vba.ForcedOutputLinkBPP[k],
+							false,
+							mode_lib->vba.Output[k],
+							mode_lib->vba.OutputFormat[k],
+							mode_lib->vba.DSCInputBitPerComponent[k]);
+				} else if (mode_lib->vba.Output[k] == dm_dp
+						|| mode_lib->vba.Output[k] == dm_edp) {
+					if (mode_lib->vba.Output[k] == dm_edp) {
+						mode_lib->vba.EffectiveFECOverhead = 0.0;
+					} else {
+						mode_lib->vba.EffectiveFECOverhead =
+								mode_lib->vba.FECOverhead;
+					}
+					if (mode_lib->vba.PHYCLKPerState[i] >= 270.0) {
+						mode_lib->vba.Outbpp = TruncToValidBPP(
+								(1.0 - mode_lib->vba.Downspreading / 100.0) * 270.0
+								* mode_lib->vba.OutputLinkDPLanes[k] / mode_lib->vba.PixelClockBackEnd[k] * 8.0,
+								mode_lib->vba.ForcedOutputLinkBPP[k],
+								false,
+								mode_lib->vba.Output[k],
+								mode_lib->vba.OutputFormat[k],
+								mode_lib->vba.DSCInputBitPerComponent[k]);
+						mode_lib->vba.OutbppDSC = TruncToValidBPP(
+								(1.0 - mode_lib->vba.Downspreading / 100.0) * (1.0 - mode_lib->vba.EffectiveFECOverhead / 100.0) * 270.0
+								* mode_lib->vba.OutputLinkDPLanes[k] / mode_lib->vba.PixelClockBackEnd[k] * 8.0,
+								mode_lib->vba.ForcedOutputLinkBPP[k],
+								true,
+								mode_lib->vba.Output[k],
+								mode_lib->vba.OutputFormat[k],
+								mode_lib->vba.DSCInputBitPerComponent[k]);
+						if (mode_lib->vba.DSCEnabled[k] == true) {
+							locals->RequiresDSC[i][k] = true;
+							if (mode_lib->vba.Output[k] == dm_dp) {
+								locals->RequiresFEC[i][k] = true;
+							} else {
+								locals->RequiresFEC[i][k] = false;
+							}
+							mode_lib->vba.Outbpp = mode_lib->vba.OutbppDSC;
+						} else {
+							locals->RequiresDSC[i][k] = false;
+							locals->RequiresFEC[i][k] = false;
+						}
+						locals->OutputBppPerState[i][k] = mode_lib->vba.Outbpp;
+					}
+					if (mode_lib->vba.Outbpp == BPP_INVALID && mode_lib->vba.PHYCLKPerState[i] >= 540.0) {
+						mode_lib->vba.Outbpp = TruncToValidBPP(
+								(1.0 - mode_lib->vba.Downspreading / 100.0) * 540.0
+								* mode_lib->vba.OutputLinkDPLanes[k] / mode_lib->vba.PixelClockBackEnd[k] * 8.0,
+								mode_lib->vba.ForcedOutputLinkBPP[k],
+									false,
+									mode_lib->vba.Output[k],
+									mode_lib->vba.OutputFormat[k],
+									mode_lib->vba.DSCInputBitPerComponent[k]);
+						mode_lib->vba.OutbppDSC = TruncToValidBPP(
+								(1.0 - mode_lib->vba.Downspreading / 100.0) * (1.0 - mode_lib->vba.EffectiveFECOverhead / 100.0) * 540.0
+								* mode_lib->vba.OutputLinkDPLanes[k] / mode_lib->vba.PixelClockBackEnd[k] * 8.0,
+								mode_lib->vba.ForcedOutputLinkBPP[k],
+								true,
+								mode_lib->vba.Output[k],
+								mode_lib->vba.OutputFormat[k],
+								mode_lib->vba.DSCInputBitPerComponent[k]);
+						if (mode_lib->vba.DSCEnabled[k] == true) {
+							locals->RequiresDSC[i][k] = true;
+							if (mode_lib->vba.Output[k] == dm_dp) {
+								locals->RequiresFEC[i][k] = true;
+							} else {
+								locals->RequiresFEC[i][k] = false;
+							}
+							mode_lib->vba.Outbpp = mode_lib->vba.OutbppDSC;
+						} else {
+							locals->RequiresDSC[i][k] = false;
+							locals->RequiresFEC[i][k] = false;
+						}
+						locals->OutputBppPerState[i][k] = mode_lib->vba.Outbpp;
+					}
+					if (mode_lib->vba.Outbpp == BPP_INVALID
+							&& mode_lib->vba.PHYCLKPerState[i]
+									>= 810.0) {
+						mode_lib->vba.Outbpp = TruncToValidBPP(
+								(1.0 - mode_lib->vba.Downspreading / 100.0) * 810.0
+								* mode_lib->vba.OutputLinkDPLanes[k] / mode_lib->vba.PixelClockBackEnd[k] * 8.0,
+								mode_lib->vba.ForcedOutputLinkBPP[k],
+								false,
+								mode_lib->vba.Output[k],
+								mode_lib->vba.OutputFormat[k],
+								mode_lib->vba.DSCInputBitPerComponent[k]);
+						mode_lib->vba.OutbppDSC = TruncToValidBPP(
+								(1.0 - mode_lib->vba.Downspreading / 100.0) * (1.0 - mode_lib->vba.EffectiveFECOverhead / 100.0) * 810.0
+								* mode_lib->vba.OutputLinkDPLanes[k] / mode_lib->vba.PixelClockBackEnd[k] * 8.0,
+								mode_lib->vba.ForcedOutputLinkBPP[k],
+								true,
+								mode_lib->vba.Output[k],
+								mode_lib->vba.OutputFormat[k],
+								mode_lib->vba.DSCInputBitPerComponent[k]);
+						if (mode_lib->vba.DSCEnabled[k] == true || mode_lib->vba.Outbpp == BPP_INVALID) {
+							locals->RequiresDSC[i][k] = true;
+							if (mode_lib->vba.Output[k] == dm_dp) {
+								locals->RequiresFEC[i][k] = true;
+							} else {
+								locals->RequiresFEC[i][k] = false;
+							}
+							mode_lib->vba.Outbpp = mode_lib->vba.OutbppDSC;
+						} else {
+							locals->RequiresDSC[i][k] = false;
+							locals->RequiresFEC[i][k] = false;
+						}
+						locals->OutputBppPerState[i][k] =
+								mode_lib->vba.Outbpp;
+					}
+				}
+			} else {
+				locals->OutputBppPerState[i][k] = BPP_BLENDED_PIPE;
+			}
+		}
+	}
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		locals->DIOSupport[i] = true;
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			if (!mode_lib->vba.skip_dio_check[k]
+					&& (locals->OutputBppPerState[i][k] == BPP_INVALID
+						|| (mode_lib->vba.OutputFormat[k] == dm_420
+							&& mode_lib->vba.Interlace[k] == true
+							&& mode_lib->vba.ProgressiveToInterlaceUnitInOPP == true))) {
+				locals->DIOSupport[i] = false;
+			}
+		}
+	}
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			locals->DSCCLKRequiredMoreThanSupported[i] = false;
+			if (mode_lib->vba.BlendingAndTiming[k] == k) {
+				if ((mode_lib->vba.Output[k] == dm_dp
+						|| mode_lib->vba.Output[k] == dm_edp)) {
+					if (mode_lib->vba.OutputFormat[k] == dm_420
+							|| mode_lib->vba.OutputFormat[k]
+									== dm_n422) {
+						mode_lib->vba.DSCFormatFactor = 2;
+					} else {
+						mode_lib->vba.DSCFormatFactor = 1;
+					}
+					if (locals->RequiresDSC[i][k] == true) {
+						if (locals->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_2to1) {
+							if (mode_lib->vba.PixelClockBackEnd[k] / 6.0 / mode_lib->vba.DSCFormatFactor
+									> (1.0 - mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) * mode_lib->vba.MaxDSCCLK[i]) {
+								locals->DSCCLKRequiredMoreThanSupported[i] =
+										true;
+							}
+						} else {
+							if (mode_lib->vba.PixelClockBackEnd[k] / 3.0 / mode_lib->vba.DSCFormatFactor
+									> (1.0 - mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) * mode_lib->vba.MaxDSCCLK[i]) {
+								locals->DSCCLKRequiredMoreThanSupported[i] =
+										true;
+							}
+						}
+					}
+				}
+			}
+		}
+	}
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		locals->NotEnoughDSCUnits[i] = false;
+		mode_lib->vba.TotalDSCUnitsRequired = 0.0;
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			if (locals->RequiresDSC[i][k] == true) {
+				if (locals->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_2to1) {
+					mode_lib->vba.TotalDSCUnitsRequired =
+							mode_lib->vba.TotalDSCUnitsRequired + 2.0;
+				} else {
+					mode_lib->vba.TotalDSCUnitsRequired =
+							mode_lib->vba.TotalDSCUnitsRequired + 1.0;
+				}
+			}
+		}
+		if (mode_lib->vba.TotalDSCUnitsRequired > mode_lib->vba.NumberOfDSC) {
+			locals->NotEnoughDSCUnits[i] = true;
+		}
+	}
+	/*DSC Delay per state*/
+
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			if (mode_lib->vba.BlendingAndTiming[k] != k) {
+				mode_lib->vba.slices = 0;
+			} else if (locals->RequiresDSC[i][k] == 0
+					|| locals->RequiresDSC[i][k] == false) {
+				mode_lib->vba.slices = 0;
+			} else if (mode_lib->vba.PixelClockBackEnd[k] > 3200.0) {
+				mode_lib->vba.slices = dml_ceil(
+						mode_lib->vba.PixelClockBackEnd[k] / 400.0,
+						4.0);
+			} else if (mode_lib->vba.PixelClockBackEnd[k] > 1360.0) {
+				mode_lib->vba.slices = 8.0;
+			} else if (mode_lib->vba.PixelClockBackEnd[k] > 680.0) {
+				mode_lib->vba.slices = 4.0;
+			} else if (mode_lib->vba.PixelClockBackEnd[k] > 340.0) {
+				mode_lib->vba.slices = 2.0;
+			} else {
+				mode_lib->vba.slices = 1.0;
+			}
+			if (locals->OutputBppPerState[i][k] == BPP_BLENDED_PIPE
+					|| locals->OutputBppPerState[i][k] == BPP_INVALID) {
+				mode_lib->vba.bpp = 0.0;
+			} else {
+				mode_lib->vba.bpp = locals->OutputBppPerState[i][k];
+			}
+			if (locals->RequiresDSC[i][k] == true && mode_lib->vba.bpp != 0.0) {
+				if (locals->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_disabled) {
+					locals->DSCDelayPerState[i][k] =
+							dscceComputeDelay(
+									mode_lib->vba.DSCInputBitPerComponent[k],
+									mode_lib->vba.bpp,
+									dml_ceil(
+											mode_lib->vba.HActive[k]
+													/ mode_lib->vba.slices,
+											1.0),
+									mode_lib->vba.slices,
+									mode_lib->vba.OutputFormat[k])
+									+ dscComputeDelay(
+											mode_lib->vba.OutputFormat[k]);
+				} else {
+					locals->DSCDelayPerState[i][k] =
+							2.0 * (dscceComputeDelay(
+											mode_lib->vba.DSCInputBitPerComponent[k],
+											mode_lib->vba.bpp,
+											dml_ceil(mode_lib->vba.HActive[k] / mode_lib->vba.slices, 1.0),
+											mode_lib->vba.slices / 2,
+											mode_lib->vba.OutputFormat[k])
+									+ dscComputeDelay(mode_lib->vba.OutputFormat[k]));
+				}
+				locals->DSCDelayPerState[i][k] =
+						locals->DSCDelayPerState[i][k] * mode_lib->vba.PixelClock[k] / mode_lib->vba.PixelClockBackEnd[k];
+			} else {
+				locals->DSCDelayPerState[i][k] = 0.0;
+			}
+		}
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			for (m = 0; m <= mode_lib->vba.NumberOfActivePlanes - 1; m++) {
+				for (j = 0; j <= mode_lib->vba.NumberOfActivePlanes - 1; j++) {
+					if (mode_lib->vba.BlendingAndTiming[k] == m && locals->RequiresDSC[i][m] == true)
+						locals->DSCDelayPerState[i][k] = locals->DSCDelayPerState[i][m];
+				}
+			}
+		}
+	}
+
+	//Prefetch Check
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		for (j = 0; j < 2; j++) {
+			for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+				if (locals->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_2to1)
+					locals->SwathWidthYPerState[i][j][k] = dml_min(locals->SwathWidthYSingleDPP[k], dml_round(locals->HActive[k] / 2 * locals->HRatio[k]));
+				else
+					locals->SwathWidthYPerState[i][j][k] = locals->SwathWidthYSingleDPP[k] / locals->NoOfDPP[i][j][k];
+				locals->SwathWidthGranularityY = 256  / dml_ceil(locals->BytePerPixelInDETY[k], 1) / locals->MaxSwathHeightY[k];
+				locals->RoundedUpMaxSwathSizeBytesY = (dml_ceil(locals->SwathWidthYPerState[i][j][k] - 1, locals->SwathWidthGranularityY)
+						+ locals->SwathWidthGranularityY) * locals->BytePerPixelInDETY[k] * locals->MaxSwathHeightY[k];
+				if (locals->SourcePixelFormat[k] == dm_420_10) {
+					locals->RoundedUpMaxSwathSizeBytesY = dml_ceil(locals->RoundedUpMaxSwathSizeBytesY, 256) + 256;
+				}
+				if (locals->MaxSwathHeightC[k] > 0) {
+					locals->SwathWidthGranularityC = 256 / dml_ceil(locals->BytePerPixelInDETC[k], 2) / locals->MaxSwathHeightC[k];
+
+					locals->RoundedUpMaxSwathSizeBytesC = (dml_ceil(locals->SwathWidthYPerState[i][j][k] / 2 - 1, locals->SwathWidthGranularityC)
+					+ locals->SwathWidthGranularityC) * locals->BytePerPixelInDETC[k] * locals->MaxSwathHeightC[k];
+				}
+				if (locals->SourcePixelFormat[k] == dm_420_10) {
+					locals->RoundedUpMaxSwathSizeBytesC = dml_ceil(locals->RoundedUpMaxSwathSizeBytesC, 256)  + 256;
+				} else {
+					locals->RoundedUpMaxSwathSizeBytesC = 0;
+				}
+
+				if (locals->RoundedUpMaxSwathSizeBytesY + locals->RoundedUpMaxSwathSizeBytesC <= locals->DETBufferSizeInKByte[0] * 1024.0 / 2) {
+					locals->SwathHeightYPerState[i][j][k] = locals->MaxSwathHeightY[k];
+					locals->SwathHeightCPerState[i][j][k] = locals->MaxSwathHeightC[k];
+				} else {
+					locals->SwathHeightYPerState[i][j][k] = locals->MinSwathHeightY[k];
+					locals->SwathHeightCPerState[i][j][k] = locals->MinSwathHeightC[k];
+				}
+
+				if (locals->BytePerPixelInDETC[k] == 0) {
+					locals->LinesInDETLuma = locals->DETBufferSizeInKByte[0] * 1024 / locals->BytePerPixelInDETY[k] / locals->SwathWidthYPerState[i][j][k];
+					locals->LinesInDETChroma = 0;
+				} else if (locals->SwathHeightYPerState[i][j][k] <= locals->SwathHeightCPerState[i][j][k]) {
+					locals->LinesInDETLuma = locals->DETBufferSizeInKByte[0] * 1024 / 2 / locals->BytePerPixelInDETY[k] /
+							locals->SwathWidthYPerState[i][j][k];
+					locals->LinesInDETChroma = locals->DETBufferSizeInKByte[0] * 1024 / 2 / locals->BytePerPixelInDETC[k] / (locals->SwathWidthYPerState[i][j][k] / 2);
+				} else {
+					locals->LinesInDETLuma = locals->DETBufferSizeInKByte[0] * 1024 * 2 / 3 / locals->BytePerPixelInDETY[k] / locals->SwathWidthYPerState[i][j][k];
+					locals->LinesInDETChroma = locals->DETBufferSizeInKByte[0] * 1024 / 3 / locals->BytePerPixelInDETY[k] / (locals->SwathWidthYPerState[i][j][k] / 2);
+				}
+
+				locals->EffectiveLBLatencyHidingSourceLinesLuma = dml_min(locals->MaxLineBufferLines,
+					dml_floor(locals->LineBufferSize / locals->LBBitPerPixel[k] / (locals->SwathWidthYPerState[i][j][k]
+					/ dml_max(locals->HRatio[k], 1)), 1)) - (locals->vtaps[k] - 1);
+
+				locals->EffectiveLBLatencyHidingSourceLinesChroma =  dml_min(locals->MaxLineBufferLines,
+						dml_floor(locals->LineBufferSize / locals->LBBitPerPixel[k]
+						/ (locals->SwathWidthYPerState[i][j][k] / 2
+						/ dml_max(locals->HRatio[k] / 2, 1)), 1)) - (locals->VTAPsChroma[k] - 1);
+
+				locals->EffectiveDETLBLinesLuma = dml_floor(locals->LinesInDETLuma +  dml_min(
+						locals->LinesInDETLuma * locals->RequiredDISPCLK[i][j] * locals->BytePerPixelInDETY[k] *
+						locals->PSCL_FACTOR[k] / locals->ReturnBWPerState[i][0],
+						locals->EffectiveLBLatencyHidingSourceLinesLuma),
+						locals->SwathHeightYPerState[i][j][k]);
+
+
+				if (locals->BytePerPixelInDETC[k] == 0) {
+					locals->UrgentLatencySupportUsPerState[i][j][k] = locals->EffectiveDETLBLinesLuma * (locals->HTotal[k] / locals->PixelClock[k])
+							/ locals->VRatio[k] - locals->EffectiveDETLBLinesLuma * locals->SwathWidthYPerState[i][j][k] *
+								dml_ceil(locals->BytePerPixelInDETY[k], 1) / (locals->ReturnBWPerState[i][0] / locals->NoOfDPP[i][j][k]);
+				} else {
+					locals->EffectiveDETLBLinesChroma = dml_floor(locals->LinesInDETChroma + dml_min(
+							locals->LinesInDETChroma * locals->RequiredDISPCLK[i][j] * locals->BytePerPixelInDETC[k] *
+							locals->PSCL_FACTOR_CHROMA[k] / locals->ReturnBWPerState[i][0],
+							locals->EffectiveLBLatencyHidingSourceLinesChroma),
+							locals->SwathHeightCPerState[i][j][k]);
+					locals->UrgentLatencySupportUsPerState[i][j][k] = dml_min(
+						locals->EffectiveDETLBLinesLuma * (locals->HTotal[k] / locals->PixelClock[k])
+						/ locals->VRatio[k] - locals->EffectiveDETLBLinesLuma * locals->SwathWidthYPerState[i][j][k] *
+						dml_ceil(locals->BytePerPixelInDETY[k], 1) / (locals->ReturnBWPerState[i][0] / locals->NoOfDPP[i][j][k]),
+							locals->EffectiveDETLBLinesChroma * (locals->HTotal[k] / locals->PixelClock[k]) / (locals->VRatio[k] / 2) -
+							locals->EffectiveDETLBLinesChroma * locals->SwathWidthYPerState[i][j][k] / 2 *
+							dml_ceil(locals->BytePerPixelInDETC[k], 2) / (locals->ReturnBWPerState[i][0] / locals->NoOfDPP[i][j][k]));
+				}
+			}
+		}
+	}
+
+	for (i = 0; i <= locals->soc.num_states; i++) {
+		for (j = 0; j < 2; j++) {
+			locals->UrgentLatencySupport[i][j] = true;
+			for (k = 0; k < locals->NumberOfActivePlanes; k++) {
+				if (locals->UrgentLatencySupportUsPerState[i][j][k] < locals->UrgentLatency)
+					locals->UrgentLatencySupport[i][j] = false;
+			}
+		}
+	}
+
+
+	/*Prefetch Check*/
+	for (i = 0; i <= locals->soc.num_states; i++) {
+		for (j = 0; j < 2; j++) {
+			locals->TotalNumberOfDCCActiveDPP[i][j] = 0;
+			for (k = 0; k < locals->NumberOfActivePlanes; k++) {
+				if (locals->DCCEnable[k] == true) {
+					locals->TotalNumberOfDCCActiveDPP[i][j] =
+						locals->TotalNumberOfDCCActiveDPP[i][j] + locals->NoOfDPP[i][j][k];
+				}
+			}
+		}
+	}
+
+	CalculateMinAndMaxPrefetchMode(locals->AllowDRAMSelfRefreshOrDRAMClockChangeInVblank, &locals->MinPrefetchMode, &locals->MaxPrefetchMode);
+
+	locals->MaxTotalVActiveRDBandwidth = 0;
+	for (k = 0; k < locals->NumberOfActivePlanes; k++) {
+		locals->MaxTotalVActiveRDBandwidth = locals->MaxTotalVActiveRDBandwidth + locals->ReadBandwidth[k];
+	}
+
+	for (i = 0; i <= locals->soc.num_states; i++) {
+		for (j = 0; j < 2; j++) {
+			for (k = 0; k < locals->NumberOfActivePlanes; k++) {
+				locals->NoOfDPPThisState[k] = locals->NoOfDPP[i][j][k];
+				locals->RequiredDPPCLKThisState[k] = locals->RequiredDPPCLK[i][j][k];
+				locals->SwathHeightYThisState[k] = locals->SwathHeightYPerState[i][j][k];
+				locals->SwathHeightCThisState[k] = locals->SwathHeightCPerState[i][j][k];
+				locals->SwathWidthYThisState[k] = locals->SwathWidthYPerState[i][j][k];
+				mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0] = dml_max(
+						mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0],
+						mode_lib->vba.PixelClock[k] / 16.0);
+				if (mode_lib->vba.BytePerPixelInDETC[k] == 0.0) {
+					if (mode_lib->vba.VRatio[k] <= 1.0) {
+						mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0] =
+								dml_max(
+										mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0],
+										1.1
+												* dml_ceil(
+														mode_lib->vba.BytePerPixelInDETY[k],
+														1.0)
+												/ 64.0
+												* mode_lib->vba.HRatio[k]
+												* mode_lib->vba.PixelClock[k]
+												/ mode_lib->vba.NoOfDPP[i][j][k]);
+					} else {
+						mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0] =
+								dml_max(
+										mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0],
+										1.1
+												* dml_ceil(
+														mode_lib->vba.BytePerPixelInDETY[k],
+														1.0)
+												/ 64.0
+												* mode_lib->vba.PSCL_FACTOR[k]
+												* mode_lib->vba.RequiredDPPCLK[i][j][k]);
+					}
+				} else {
+					if (mode_lib->vba.VRatio[k] <= 1.0) {
+						mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0] =
+								dml_max(
+										mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0],
+										1.1
+												* dml_ceil(
+														mode_lib->vba.BytePerPixelInDETY[k],
+														1.0)
+												/ 32.0
+												* mode_lib->vba.HRatio[k]
+												* mode_lib->vba.PixelClock[k]
+												/ mode_lib->vba.NoOfDPP[i][j][k]);
+					} else {
+						mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0] =
+								dml_max(
+										mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0],
+										1.1
+												* dml_ceil(
+														mode_lib->vba.BytePerPixelInDETY[k],
+														1.0)
+												/ 32.0
+												* mode_lib->vba.PSCL_FACTOR[k]
+												* mode_lib->vba.RequiredDPPCLK[i][j][k]);
+					}
+					if (mode_lib->vba.VRatio[k] / 2.0 <= 1.0) {
+						mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0] =
+								dml_max(
+										mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0],
+										1.1
+												* dml_ceil(
+														mode_lib->vba.BytePerPixelInDETC[k],
+														2.0)
+												/ 32.0
+												* mode_lib->vba.HRatio[k]
+												/ 2.0
+												* mode_lib->vba.PixelClock[k]
+												/ mode_lib->vba.NoOfDPP[i][j][k]);
+					} else {
+						mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0] =
+								dml_max(
+										mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0],
+										1.1
+												* dml_ceil(
+														mode_lib->vba.BytePerPixelInDETC[k],
+														2.0)
+												/ 32.0
+												* mode_lib->vba.PSCL_FACTOR_CHROMA[k]
+												* mode_lib->vba.RequiredDPPCLK[i][j][k]);
+					}
+				}
+			}
+			for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+				mode_lib->vba.PDEAndMetaPTEBytesPerFrameY = CalculateVMAndRowBytes(
+						mode_lib,
+						mode_lib->vba.DCCEnable[k],
+						mode_lib->vba.Read256BlockHeightY[k],
+						mode_lib->vba.Read256BlockWidthY[k],
+						mode_lib->vba.SourcePixelFormat[k],
+						mode_lib->vba.SurfaceTiling[k],
+						dml_ceil(mode_lib->vba.BytePerPixelInDETY[k], 1.0),
+						mode_lib->vba.SourceScan[k],
+						mode_lib->vba.ViewportWidth[k],
+						mode_lib->vba.ViewportHeight[k],
+						mode_lib->vba.SwathWidthYPerState[i][j][k],
+						mode_lib->vba.GPUVMEnable,
+						mode_lib->vba.VMMPageSize,
+						mode_lib->vba.PTEBufferSizeInRequestsLuma,
+						mode_lib->vba.PDEProcessingBufIn64KBReqs,
+						mode_lib->vba.PitchY[k],
+						mode_lib->vba.DCCMetaPitchY[k],
+						&mode_lib->vba.MacroTileWidthY[k],
+						&mode_lib->vba.MetaRowBytesY,
+						&mode_lib->vba.DPTEBytesPerRowY,
+						&mode_lib->vba.PTEBufferSizeNotExceededY[i][j][k],
+						&mode_lib->vba.dpte_row_height[k],
+						&mode_lib->vba.meta_row_height[k]);
+				mode_lib->vba.PrefetchLinesY[0][0][k] = CalculatePrefetchSourceLines(
+						mode_lib,
+						mode_lib->vba.VRatio[k],
+						mode_lib->vba.vtaps[k],
+						mode_lib->vba.Interlace[k],
+						mode_lib->vba.ProgressiveToInterlaceUnitInOPP,
+						mode_lib->vba.SwathHeightYPerState[i][j][k],
+						mode_lib->vba.ViewportYStartY[k],
+						&mode_lib->vba.PrefillY[k],
+						&mode_lib->vba.MaxNumSwY[k]);
+				if ((mode_lib->vba.SourcePixelFormat[k] != dm_444_64
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_444_32
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_444_16
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_mono_16
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_mono_8)) {
+					mode_lib->vba.PDEAndMetaPTEBytesPerFrameC = CalculateVMAndRowBytes(
+							mode_lib,
+							mode_lib->vba.DCCEnable[k],
+							mode_lib->vba.Read256BlockHeightY[k],
+							mode_lib->vba.Read256BlockWidthY[k],
+							mode_lib->vba.SourcePixelFormat[k],
+							mode_lib->vba.SurfaceTiling[k],
+							dml_ceil(mode_lib->vba.BytePerPixelInDETC[k], 2.0),
+							mode_lib->vba.SourceScan[k],
+							mode_lib->vba.ViewportWidth[k] / 2.0,
+							mode_lib->vba.ViewportHeight[k] / 2.0,
+							mode_lib->vba.SwathWidthYPerState[i][j][k] / 2.0,
+							mode_lib->vba.GPUVMEnable,
+							mode_lib->vba.VMMPageSize,
+							mode_lib->vba.PTEBufferSizeInRequestsLuma,
+							mode_lib->vba.PDEProcessingBufIn64KBReqs,
+							mode_lib->vba.PitchC[k],
+							0.0,
+							&mode_lib->vba.MacroTileWidthC[k],
+							&mode_lib->vba.MetaRowBytesC,
+							&mode_lib->vba.DPTEBytesPerRowC,
+							&mode_lib->vba.PTEBufferSizeNotExceededC[i][j][k],
+							&mode_lib->vba.dpte_row_height_chroma[k],
+							&mode_lib->vba.meta_row_height_chroma[k]);
+					mode_lib->vba.PrefetchLinesC[0][0][k] = CalculatePrefetchSourceLines(
+							mode_lib,
+							mode_lib->vba.VRatio[k] / 2.0,
+							mode_lib->vba.VTAPsChroma[k],
+							mode_lib->vba.Interlace[k],
+							mode_lib->vba.ProgressiveToInterlaceUnitInOPP,
+							mode_lib->vba.SwathHeightCPerState[i][j][k],
+							mode_lib->vba.ViewportYStartC[k],
+							&mode_lib->vba.PrefillC[k],
+							&mode_lib->vba.MaxNumSwC[k]);
+				} else {
+					mode_lib->vba.PDEAndMetaPTEBytesPerFrameC = 0.0;
+					mode_lib->vba.MetaRowBytesC = 0.0;
+					mode_lib->vba.DPTEBytesPerRowC = 0.0;
+					locals->PrefetchLinesC[0][0][k] = 0.0;
+					locals->PTEBufferSizeNotExceededC[i][j][k] = true;
+					locals->PTEBufferSizeInRequestsForLuma = mode_lib->vba.PTEBufferSizeInRequestsLuma + mode_lib->vba.PTEBufferSizeInRequestsChroma;
+				}
+				locals->PDEAndMetaPTEBytesPerFrame[0][0][k] =
+						mode_lib->vba.PDEAndMetaPTEBytesPerFrameY + mode_lib->vba.PDEAndMetaPTEBytesPerFrameC;
+				locals->MetaRowBytes[0][0][k] = mode_lib->vba.MetaRowBytesY + mode_lib->vba.MetaRowBytesC;
+				locals->DPTEBytesPerRow[0][0][k] = mode_lib->vba.DPTEBytesPerRowY + mode_lib->vba.DPTEBytesPerRowC;
+
+				CalculateActiveRowBandwidth(
+						mode_lib->vba.GPUVMEnable,
+						mode_lib->vba.SourcePixelFormat[k],
+						mode_lib->vba.VRatio[k],
+						mode_lib->vba.DCCEnable[k],
+						mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k],
+						mode_lib->vba.MetaRowBytesY,
+						mode_lib->vba.MetaRowBytesC,
+						mode_lib->vba.meta_row_height[k],
+						mode_lib->vba.meta_row_height_chroma[k],
+						mode_lib->vba.DPTEBytesPerRowY,
+						mode_lib->vba.DPTEBytesPerRowC,
+						mode_lib->vba.dpte_row_height[k],
+						mode_lib->vba.dpte_row_height_chroma[k],
+						&mode_lib->vba.meta_row_bw[k],
+						&mode_lib->vba.dpte_row_bw[k],
+						&mode_lib->vba.qual_row_bw[k]);
+			}
+			mode_lib->vba.ExtraLatency =
+					mode_lib->vba.UrgentRoundTripAndOutOfOrderLatencyPerState[i]
+							+ (mode_lib->vba.TotalNumberOfActiveDPP[i][j]
+									* mode_lib->vba.PixelChunkSizeInKByte
+									+ mode_lib->vba.TotalNumberOfDCCActiveDPP[i][j]
+											* mode_lib->vba.MetaChunkSize)
+									* 1024.0
+									/ mode_lib->vba.ReturnBWPerState[i][0];
+			if (mode_lib->vba.GPUVMEnable == true) {
+				mode_lib->vba.ExtraLatency = mode_lib->vba.ExtraLatency
+						+ mode_lib->vba.TotalNumberOfActiveDPP[i][j]
+								* mode_lib->vba.PTEGroupSize
+								/ mode_lib->vba.ReturnBWPerState[i][0];
+			}
+			mode_lib->vba.TimeCalc = 24.0 / mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0];
+
+			for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+				if (mode_lib->vba.BlendingAndTiming[k] == k) {
+					if (mode_lib->vba.WritebackEnable[k] == true) {
+						locals->WritebackDelay[i][k] = mode_lib->vba.WritebackLatency
+								+ CalculateWriteBackDelay(
+										mode_lib->vba.WritebackPixelFormat[k],
+										mode_lib->vba.WritebackHRatio[k],
+										mode_lib->vba.WritebackVRatio[k],
+										mode_lib->vba.WritebackLumaHTaps[k],
+										mode_lib->vba.WritebackLumaVTaps[k],
+										mode_lib->vba.WritebackChromaHTaps[k],
+										mode_lib->vba.WritebackChromaVTaps[k],
+										mode_lib->vba.WritebackDestinationWidth[k]) / locals->RequiredDISPCLK[i][j];
+					} else {
+						locals->WritebackDelay[i][k] = 0.0;
+					}
+					for (m = 0; m <= mode_lib->vba.NumberOfActivePlanes - 1; m++) {
+						if (mode_lib->vba.BlendingAndTiming[m] == k
+								&& mode_lib->vba.WritebackEnable[m]
+										== true) {
+							locals->WritebackDelay[i][k] = dml_max(locals->WritebackDelay[i][k],
+											mode_lib->vba.WritebackLatency + CalculateWriteBackDelay(
+													mode_lib->vba.WritebackPixelFormat[m],
+													mode_lib->vba.WritebackHRatio[m],
+													mode_lib->vba.WritebackVRatio[m],
+													mode_lib->vba.WritebackLumaHTaps[m],
+													mode_lib->vba.WritebackLumaVTaps[m],
+													mode_lib->vba.WritebackChromaHTaps[m],
+													mode_lib->vba.WritebackChromaVTaps[m],
+													mode_lib->vba.WritebackDestinationWidth[m]) / locals->RequiredDISPCLK[i][j]);
+						}
+					}
+				}
+			}
+			for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+				for (m = 0; m <= mode_lib->vba.NumberOfActivePlanes - 1; m++) {
+					if (mode_lib->vba.BlendingAndTiming[k] == m) {
+						locals->WritebackDelay[i][k] = locals->WritebackDelay[i][m];
+					}
+				}
+			}
+			for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+				for (m = 0; m < locals->NumberOfCursors[k]; m++)
+					locals->cursor_bw[k] = locals->NumberOfCursors[k] * locals->CursorWidth[k][m] * locals->CursorBPP[k][m]
+						/ 8 / (locals->HTotal[k] / locals->PixelClock[k]) * locals->VRatio[k];
+			}
+
+			for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+				locals->MaximumVStartup[0][0][k] = mode_lib->vba.VTotal[k] - mode_lib->vba.VActive[k]
+					- dml_max(1.0, dml_ceil(locals->WritebackDelay[i][k] / (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k]), 1.0));
+			}
+
+			mode_lib->vba.NextPrefetchMode = mode_lib->vba.MinPrefetchMode;
+			do {
+				mode_lib->vba.PrefetchMode[i][j] = mode_lib->vba.NextPrefetchMode;
+				mode_lib->vba.NextPrefetchMode = mode_lib->vba.NextPrefetchMode + 1;
+
+				mode_lib->vba.TWait = CalculateTWait(
+						mode_lib->vba.PrefetchMode[i][j],
+						mode_lib->vba.DRAMClockChangeLatency,
+						mode_lib->vba.UrgentLatency,
+						mode_lib->vba.SREnterPlusExitTime);
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+
+					if (mode_lib->vba.XFCEnabled[k] == true) {
+						mode_lib->vba.XFCRemoteSurfaceFlipDelay =
+								CalculateRemoteSurfaceFlipDelay(
+										mode_lib,
+										mode_lib->vba.VRatio[k],
+										locals->SwathWidthYPerState[i][j][k],
+										dml_ceil(locals->BytePerPixelInDETY[k], 1.0),
+										mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k],
+										mode_lib->vba.XFCTSlvVupdateOffset,
+										mode_lib->vba.XFCTSlvVupdateWidth,
+										mode_lib->vba.XFCTSlvVreadyOffset,
+										mode_lib->vba.XFCXBUFLatencyTolerance,
+										mode_lib->vba.XFCFillBWOverhead,
+										mode_lib->vba.XFCSlvChunkSize,
+										mode_lib->vba.XFCBusTransportTime,
+										mode_lib->vba.TimeCalc,
+										mode_lib->vba.TWait,
+										&mode_lib->vba.SrcActiveDrainRate,
+										&mode_lib->vba.TInitXFill,
+										&mode_lib->vba.TslvChk);
+					} else {
+						mode_lib->vba.XFCRemoteSurfaceFlipDelay = 0.0;
+					}
+
+					CalculateDelayAfterScaler(mode_lib, mode_lib->vba.ReturnBWPerState[i][0], mode_lib->vba.ReadBandwidthLuma[k], mode_lib->vba.ReadBandwidthChroma[k], mode_lib->vba.MaxTotalVActiveRDBandwidth,
+						mode_lib->vba.DisplayPipeLineDeliveryTimeLuma[k], mode_lib->vba.DisplayPipeLineDeliveryTimeChroma[k],
+						mode_lib->vba.RequiredDPPCLK[i][j][k], mode_lib->vba.RequiredDISPCLK[i][j], mode_lib->vba.PixelClock[k], mode_lib->vba.DSCDelayPerState[i][k], mode_lib->vba.NoOfDPP[i][j][k], mode_lib->vba.ScalerEnabled[k], mode_lib->vba.NumberOfCursors[k],
+						mode_lib->vba.DPPCLKDelaySubtotal, mode_lib->vba.DPPCLKDelaySCL, mode_lib->vba.DPPCLKDelaySCLLBOnly, mode_lib->vba.DPPCLKDelayCNVCFormater, mode_lib->vba.DPPCLKDelayCNVCCursor, mode_lib->vba.DISPCLKDelaySubtotal,
+						mode_lib->vba.SwathWidthYPerState[i][j][k] / mode_lib->vba.HRatio[k], mode_lib->vba.OutputFormat[k], mode_lib->vba.HTotal[k],
+						mode_lib->vba.SwathWidthYSingleDPP[k], mode_lib->vba.BytePerPixelInDETY[k], mode_lib->vba.BytePerPixelInDETC[k], mode_lib->vba.SwathHeightYThisState[k], mode_lib->vba.SwathHeightCThisState[k], mode_lib->vba.Interlace[k], mode_lib->vba.ProgressiveToInterlaceUnitInOPP,
+						&mode_lib->vba.DSTXAfterScaler[k], &mode_lib->vba.DSTYAfterScaler[k]);
+
+					mode_lib->vba.IsErrorResult[i][j][k] =
+							CalculatePrefetchSchedule(
+									mode_lib,
+									mode_lib->vba.RequiredDPPCLK[i][j][k],
+									mode_lib->vba.RequiredDISPCLK[i][j],
+									mode_lib->vba.PixelClock[k],
+									mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0],
+									mode_lib->vba.NoOfDPP[i][j][k],
+									mode_lib->vba.NumberOfCursors[k],
+									mode_lib->vba.VTotal[k]
+											- mode_lib->vba.VActive[k],
+									mode_lib->vba.HTotal[k],
+									mode_lib->vba.MaxInterDCNTileRepeaters,
+									mode_lib->vba.MaximumVStartup[0][0][k],
+									mode_lib->vba.GPUVMMaxPageTableLevels,
+									mode_lib->vba.GPUVMEnable,
+									mode_lib->vba.DynamicMetadataEnable[k],
+									mode_lib->vba.DynamicMetadataLinesBeforeActiveRequired[k],
+									mode_lib->vba.DynamicMetadataTransmittedBytes[k],
+									mode_lib->vba.DCCEnable[k],
+									mode_lib->vba.UrgentLatencyPixelDataOnly,
+									mode_lib->vba.ExtraLatency,
+									mode_lib->vba.TimeCalc,
+									mode_lib->vba.PDEAndMetaPTEBytesPerFrame[0][0][k],
+									mode_lib->vba.MetaRowBytes[0][0][k],
+									mode_lib->vba.DPTEBytesPerRow[0][0][k],
+									mode_lib->vba.PrefetchLinesY[0][0][k],
+									mode_lib->vba.SwathWidthYPerState[i][j][k],
+									mode_lib->vba.BytePerPixelInDETY[k],
+									mode_lib->vba.PrefillY[k],
+									mode_lib->vba.MaxNumSwY[k],
+									mode_lib->vba.PrefetchLinesC[0][0][k],
+									mode_lib->vba.BytePerPixelInDETC[k],
+									mode_lib->vba.PrefillC[k],
+									mode_lib->vba.MaxNumSwC[k],
+									mode_lib->vba.SwathHeightYPerState[i][j][k],
+									mode_lib->vba.SwathHeightCPerState[i][j][k],
+									mode_lib->vba.TWait,
+									mode_lib->vba.XFCEnabled[k],
+									mode_lib->vba.XFCRemoteSurfaceFlipDelay,
+									mode_lib->vba.Interlace[k],
+									mode_lib->vba.ProgressiveToInterlaceUnitInOPP,
+									mode_lib->vba.DSTXAfterScaler[k],
+									mode_lib->vba.DSTYAfterScaler[k],
+									&mode_lib->vba.LineTimesForPrefetch[k],
+									&mode_lib->vba.PrefetchBW[k],
+									&mode_lib->vba.LinesForMetaPTE[k],
+									&mode_lib->vba.LinesForMetaAndDPTERow[k],
+									&mode_lib->vba.VRatioPreY[i][j][k],
+									&mode_lib->vba.VRatioPreC[i][j][k],
+									&mode_lib->vba.RequiredPrefetchPixelDataBWLuma[i][j][k],
+									&mode_lib->vba.Tno_bw[k],
+									&mode_lib->vba.VUpdateOffsetPix[k],
+									&mode_lib->vba.VUpdateWidthPix[k],
+									&mode_lib->vba.VReadyOffsetPix[k]);
+				}
+				mode_lib->vba.MaximumReadBandwidthWithoutPrefetch = 0.0;
+				mode_lib->vba.MaximumReadBandwidthWithPrefetch = 0.0;
+				locals->prefetch_vm_bw_valid = true;
+				locals->prefetch_row_bw_valid = true;
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+					if (locals->PDEAndMetaPTEBytesPerFrame[0][0][k] == 0)
+						locals->prefetch_vm_bw[k] = 0;
+					else if (locals->LinesForMetaPTE[k] > 0)
+						locals->prefetch_vm_bw[k] = locals->PDEAndMetaPTEBytesPerFrame[0][0][k]
+							/ (locals->LinesForMetaPTE[k] * locals->HTotal[k] / locals->PixelClock[k]);
+					else {
+						locals->prefetch_vm_bw[k] = 0;
+						locals->prefetch_vm_bw_valid = false;
+					}
+					if (locals->MetaRowBytes[0][0][k] + locals->DPTEBytesPerRow[0][0][k] == 0)
+						locals->prefetch_row_bw[k] = 0;
+					else if (locals->LinesForMetaAndDPTERow[k] > 0)
+						locals->prefetch_row_bw[k] = (locals->MetaRowBytes[0][0][k] + locals->DPTEBytesPerRow[0][0][k])
+							/ (locals->LinesForMetaAndDPTERow[k] * locals->HTotal[k] / locals->PixelClock[k]);
+					else {
+						locals->prefetch_row_bw[k] = 0;
+						locals->prefetch_row_bw_valid = false;
+					}
+
+					mode_lib->vba.MaximumReadBandwidthWithoutPrefetch = mode_lib->vba.MaximumReadBandwidthWithPrefetch
+							+ mode_lib->vba.cursor_bw[k] + mode_lib->vba.ReadBandwidth[k] + mode_lib->vba.meta_row_bw[k] + mode_lib->vba.dpte_row_bw[k];
+					mode_lib->vba.MaximumReadBandwidthWithPrefetch =
+							mode_lib->vba.MaximumReadBandwidthWithPrefetch
+									+ mode_lib->vba.cursor_bw[k]
+									+ dml_max3(
+											mode_lib->vba.prefetch_vm_bw[k],
+											mode_lib->vba.prefetch_row_bw[k],
+											dml_max(mode_lib->vba.ReadBandwidth[k],
+											mode_lib->vba.RequiredPrefetchPixelDataBWLuma[i][j][k])
+											+ mode_lib->vba.meta_row_bw[k] + mode_lib->vba.dpte_row_bw[k]);
+				}
+				locals->BandwidthWithoutPrefetchSupported[i][0] = true;
+				if (mode_lib->vba.MaximumReadBandwidthWithoutPrefetch > locals->ReturnBWPerState[i][0]) {
+					locals->BandwidthWithoutPrefetchSupported[i][0] = false;
+				}
+
+				locals->PrefetchSupported[i][j] = true;
+				if (mode_lib->vba.MaximumReadBandwidthWithPrefetch > locals->ReturnBWPerState[i][0]) {
+					locals->PrefetchSupported[i][j] = false;
+				}
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+					if (locals->LineTimesForPrefetch[k] < 2.0
+							|| locals->LinesForMetaPTE[k] >= 8.0
+							|| locals->LinesForMetaAndDPTERow[k] >= 16.0
+							|| mode_lib->vba.IsErrorResult[i][j][k] == true) {
+						locals->PrefetchSupported[i][j] = false;
+					}
+				}
+				locals->VRatioInPrefetchSupported[i][j] = true;
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+					if (locals->VRatioPreY[i][j][k] > 4.0
+							|| locals->VRatioPreC[i][j][k] > 4.0
+							|| mode_lib->vba.IsErrorResult[i][j][k] == true) {
+						locals->VRatioInPrefetchSupported[i][j] = false;
+					}
+				}
+			} while ((locals->PrefetchSupported[i][j] != true || locals->VRatioInPrefetchSupported[i][j] != true)
+					&& mode_lib->vba.NextPrefetchMode < mode_lib->vba.MaxPrefetchMode);
+
+			if (mode_lib->vba.PrefetchSupported[i][j] == true
+					&& mode_lib->vba.VRatioInPrefetchSupported[i][j] == true) {
+				mode_lib->vba.BandwidthAvailableForImmediateFlip =
+						mode_lib->vba.ReturnBWPerState[i][0];
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+					mode_lib->vba.BandwidthAvailableForImmediateFlip =
+							mode_lib->vba.BandwidthAvailableForImmediateFlip
+									- mode_lib->vba.cursor_bw[k]
+									- dml_max(
+											mode_lib->vba.ReadBandwidth[k] + mode_lib->vba.qual_row_bw[k],
+											mode_lib->vba.PrefetchBW[k]);
+				}
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+					mode_lib->vba.ImmediateFlipBytes[k] = 0.0;
+					if ((mode_lib->vba.SourcePixelFormat[k] != dm_420_8
+							&& mode_lib->vba.SourcePixelFormat[k] != dm_420_10)) {
+						mode_lib->vba.ImmediateFlipBytes[k] =
+								mode_lib->vba.PDEAndMetaPTEBytesPerFrame[0][0][k]
+										+ mode_lib->vba.MetaRowBytes[0][0][k]
+										+ mode_lib->vba.DPTEBytesPerRow[0][0][k];
+					}
+				}
+				mode_lib->vba.TotImmediateFlipBytes = 0.0;
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+					if ((mode_lib->vba.SourcePixelFormat[k] != dm_420_8
+							&& mode_lib->vba.SourcePixelFormat[k] != dm_420_10)) {
+						mode_lib->vba.TotImmediateFlipBytes =
+								mode_lib->vba.TotImmediateFlipBytes
+										+ mode_lib->vba.ImmediateFlipBytes[k];
+					}
+				}
+
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+					CalculateFlipSchedule(
+							mode_lib,
+							mode_lib->vba.ExtraLatency,
+							mode_lib->vba.UrgentLatencyPixelDataOnly,
+							mode_lib->vba.GPUVMMaxPageTableLevels,
+							mode_lib->vba.GPUVMEnable,
+							mode_lib->vba.BandwidthAvailableForImmediateFlip,
+							mode_lib->vba.TotImmediateFlipBytes,
+							mode_lib->vba.SourcePixelFormat[k],
+							mode_lib->vba.ImmediateFlipBytes[k],
+							mode_lib->vba.HTotal[k]
+									/ mode_lib->vba.PixelClock[k],
+							mode_lib->vba.VRatio[k],
+							mode_lib->vba.Tno_bw[k],
+							mode_lib->vba.PDEAndMetaPTEBytesPerFrame[0][0][k],
+							mode_lib->vba.MetaRowBytes[0][0][k],
+							mode_lib->vba.DPTEBytesPerRow[0][0][k],
+							mode_lib->vba.DCCEnable[k],
+							mode_lib->vba.dpte_row_height[k],
+							mode_lib->vba.meta_row_height[k],
+							mode_lib->vba.qual_row_bw[k],
+							&mode_lib->vba.DestinationLinesToRequestVMInImmediateFlip[k],
+							&mode_lib->vba.DestinationLinesToRequestRowInImmediateFlip[k],
+							&mode_lib->vba.final_flip_bw[k],
+							&mode_lib->vba.ImmediateFlipSupportedForPipe[k]);
+				}
+				mode_lib->vba.total_dcn_read_bw_with_flip = 0.0;
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+					mode_lib->vba.total_dcn_read_bw_with_flip =
+							mode_lib->vba.total_dcn_read_bw_with_flip
+									+ mode_lib->vba.cursor_bw[k]
+									+ dml_max3(
+											mode_lib->vba.prefetch_vm_bw[k],
+											mode_lib->vba.prefetch_row_bw[k],
+											mode_lib->vba.final_flip_bw[k]
+													+ dml_max(
+															mode_lib->vba.ReadBandwidth[k],
+															mode_lib->vba.RequiredPrefetchPixelDataBWLuma[i][j][k]));
+				}
+				mode_lib->vba.ImmediateFlipSupportedForState[i][j] = true;
+				if (mode_lib->vba.total_dcn_read_bw_with_flip
+						> mode_lib->vba.ReturnBWPerState[i][0]) {
+					mode_lib->vba.ImmediateFlipSupportedForState[i][j] = false;
+				}
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+					if (mode_lib->vba.ImmediateFlipSupportedForPipe[k] == false) {
+						mode_lib->vba.ImmediateFlipSupportedForState[i][j] = false;
+					}
+				}
+			} else {
+				mode_lib->vba.ImmediateFlipSupportedForState[i][j] = false;
+			}
+		}
+	}
+
+	/*Vertical Active BW support*/
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		mode_lib->vba.MaxTotalVerticalActiveAvailableBandwidth[i][0] = dml_min(mode_lib->vba.ReturnBusWidth *
+				mode_lib->vba.DCFCLKPerState[i], mode_lib->vba.FabricAndDRAMBandwidthPerState[i] * 1000) *
+				mode_lib->vba.MaxAveragePercentOfIdealDRAMBWDisplayCanUseInNormalSystemOperation / 100;
+		if (mode_lib->vba.MaxTotalVActiveRDBandwidth <= mode_lib->vba.MaxTotalVerticalActiveAvailableBandwidth[i][0])
+			mode_lib->vba.TotalVerticalActiveBandwidthSupport[i][0] = true;
+		else
+			mode_lib->vba.TotalVerticalActiveBandwidthSupport[i][0] = false;
+	}
+
+	/*PTE Buffer Size Check*/
+
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		for (j = 0; j < 2; j++) {
+			locals->PTEBufferSizeNotExceeded[i][j] = true;
+			for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+				if (locals->PTEBufferSizeNotExceededY[i][j][k] == false
+						|| locals->PTEBufferSizeNotExceededC[i][j][k] == false) {
+					locals->PTEBufferSizeNotExceeded[i][j] = false;
+				}
+			}
+		}
+	}
+	/*Cursor Support Check*/
+	mode_lib->vba.CursorSupport = true;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		for (j = 0; j < 2; j++) {
+			if (mode_lib->vba.CursorWidth[k][j] > 0.0) {
+				if (dml_floor(
+						dml_floor(
+								mode_lib->vba.CursorBufferSize
+										- mode_lib->vba.CursorChunkSize,
+								mode_lib->vba.CursorChunkSize) * 1024.0
+								/ (mode_lib->vba.CursorWidth[k][j]
+										* mode_lib->vba.CursorBPP[k][j]
+										/ 8.0),
+						1.0)
+						* (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k])
+						/ mode_lib->vba.VRatio[k] < mode_lib->vba.UrgentLatencyPixelDataOnly
+						|| (mode_lib->vba.CursorBPP[k][j] == 64.0
+								&& mode_lib->vba.Cursor64BppSupport == false)) {
+					mode_lib->vba.CursorSupport = false;
+				}
+			}
+		}
+	}
+	/*Valid Pitch Check*/
+
+	mode_lib->vba.PitchSupport = true;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		locals->AlignedYPitch[k] = dml_ceil(
+				dml_max(mode_lib->vba.PitchY[k], mode_lib->vba.ViewportWidth[k]),
+				locals->MacroTileWidthY[k]);
+		if (locals->AlignedYPitch[k] > mode_lib->vba.PitchY[k]) {
+			mode_lib->vba.PitchSupport = false;
+		}
+		if (mode_lib->vba.DCCEnable[k] == true) {
+			locals->AlignedDCCMetaPitch[k] = dml_ceil(
+					dml_max(
+							mode_lib->vba.DCCMetaPitchY[k],
+							mode_lib->vba.ViewportWidth[k]),
+					64.0 * locals->Read256BlockWidthY[k]);
+		} else {
+			locals->AlignedDCCMetaPitch[k] = mode_lib->vba.DCCMetaPitchY[k];
+		}
+		if (locals->AlignedDCCMetaPitch[k] > mode_lib->vba.DCCMetaPitchY[k]) {
+			mode_lib->vba.PitchSupport = false;
+		}
+		if (mode_lib->vba.SourcePixelFormat[k] != dm_444_64
+				&& mode_lib->vba.SourcePixelFormat[k] != dm_444_32
+				&& mode_lib->vba.SourcePixelFormat[k] != dm_444_16
+				&& mode_lib->vba.SourcePixelFormat[k] != dm_mono_16
+				&& mode_lib->vba.SourcePixelFormat[k] != dm_mono_8) {
+			locals->AlignedCPitch[k] = dml_ceil(
+					dml_max(
+							mode_lib->vba.PitchC[k],
+							mode_lib->vba.ViewportWidth[k] / 2.0),
+					locals->MacroTileWidthC[k]);
+		} else {
+			locals->AlignedCPitch[k] = mode_lib->vba.PitchC[k];
+		}
+		if (locals->AlignedCPitch[k] > mode_lib->vba.PitchC[k]) {
+			mode_lib->vba.PitchSupport = false;
+		}
+	}
+	/*Mode Support, Voltage State and SOC Configuration*/
+
+	for (i = mode_lib->vba.soc.num_states; i >= 0; i--) {
+		for (j = 0; j < 2; j++) {
+			enum dm_validation_status status = DML_VALIDATION_OK;
+
+			if (mode_lib->vba.ScaleRatioAndTapsSupport != true) {
+				status = DML_FAIL_SCALE_RATIO_TAP;
+			} else if (mode_lib->vba.SourceFormatPixelAndScanSupport != true) {
+				status = DML_FAIL_SOURCE_PIXEL_FORMAT;
+			} else if (locals->ViewportSizeSupport[i][0] != true) {
+				status = DML_FAIL_VIEWPORT_SIZE;
+			} else if (locals->DIOSupport[i] != true) {
+				status = DML_FAIL_DIO_SUPPORT;
+			} else if (locals->NotEnoughDSCUnits[i] != false) {
+				status = DML_FAIL_NOT_ENOUGH_DSC;
+			} else if (locals->DSCCLKRequiredMoreThanSupported[i] != false) {
+				status = DML_FAIL_DSC_CLK_REQUIRED;
+			} else if (locals->UrgentLatencySupport[i][j] != true) {
+				status = DML_FAIL_URGENT_LATENCY;
+			} else if (locals->ROBSupport[i][0] != true) {
+				status = DML_FAIL_REORDERING_BUFFER;
+			} else if (locals->DISPCLK_DPPCLK_Support[i][j] != true) {
+				status = DML_FAIL_DISPCLK_DPPCLK;
+			} else if (locals->TotalAvailablePipesSupport[i][j] != true) {
+				status = DML_FAIL_TOTAL_AVAILABLE_PIPES;
+			} else if (mode_lib->vba.NumberOfOTGSupport != true) {
+				status = DML_FAIL_NUM_OTG;
+			} else if (mode_lib->vba.WritebackModeSupport != true) {
+				status = DML_FAIL_WRITEBACK_MODE;
+			} else if (mode_lib->vba.WritebackLatencySupport != true) {
+				status = DML_FAIL_WRITEBACK_LATENCY;
+			} else if (mode_lib->vba.WritebackScaleRatioAndTapsSupport != true) {
+				status = DML_FAIL_WRITEBACK_SCALE_RATIO_TAP;
+			} else if (mode_lib->vba.CursorSupport != true) {
+				status = DML_FAIL_CURSOR_SUPPORT;
+			} else if (mode_lib->vba.PitchSupport != true) {
+				status = DML_FAIL_PITCH_SUPPORT;
+			} else if (locals->PrefetchSupported[i][j] != true) {
+				status = DML_FAIL_PREFETCH_SUPPORT;
+			} else if (locals->TotalVerticalActiveBandwidthSupport[i][0] != true) {
+				status = DML_FAIL_TOTAL_V_ACTIVE_BW;
+			} else if (locals->VRatioInPrefetchSupported[i][j] != true) {
+				status = DML_FAIL_V_RATIO_PREFETCH;
+			} else if (locals->PTEBufferSizeNotExceeded[i][j] != true) {
+				status = DML_FAIL_PTE_BUFFER_SIZE;
+			} else if (mode_lib->vba.NonsupportedDSCInputBPC != false) {
+				status = DML_FAIL_DSC_INPUT_BPC;
+			}
+
+			if (status == DML_VALIDATION_OK) {
+				locals->ModeSupport[i][j] = true;
+			} else {
+				locals->ModeSupport[i][j] = false;
+			}
+			locals->ValidationStatus[i] = status;
+		}
+	}
+	{
+		unsigned int MaximumMPCCombine = 0;
+		mode_lib->vba.VoltageLevel = mode_lib->vba.soc.num_states + 1;
+		for (i = mode_lib->vba.VoltageOverrideLevel; i <= mode_lib->vba.soc.num_states; i++) {
+			if (locals->ModeSupport[i][0] == true || locals->ModeSupport[i][1] == true) {
+				mode_lib->vba.VoltageLevel = i;
+				if (locals->ModeSupport[i][1] == true && (locals->ModeSupport[i][0] == false
+						|| mode_lib->vba.WhenToDoMPCCombine == dm_mpc_always_when_possible)) {
+					MaximumMPCCombine = 1;
+				} else {
+					MaximumMPCCombine = 0;
+				}
+				break;
+			}
+		}
+		mode_lib->vba.ImmediateFlipSupport =
+			locals->ImmediateFlipSupportedForState[mode_lib->vba.VoltageLevel][MaximumMPCCombine];
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			mode_lib->vba.DPPPerPlane[k] = locals->NoOfDPP[mode_lib->vba.VoltageLevel][MaximumMPCCombine][k];
+			locals->DPPCLK[k] = locals->RequiredDPPCLK[mode_lib->vba.VoltageLevel][MaximumMPCCombine][k];
+		}
+		mode_lib->vba.DISPCLK = locals->RequiredDISPCLK[mode_lib->vba.VoltageLevel][MaximumMPCCombine];
+		mode_lib->vba.maxMpcComb = MaximumMPCCombine;
+	}
+	mode_lib->vba.DCFCLK = mode_lib->vba.DCFCLKPerState[mode_lib->vba.VoltageLevel];
+	mode_lib->vba.DRAMSpeed = mode_lib->vba.DRAMSpeedPerState[mode_lib->vba.VoltageLevel];
+	mode_lib->vba.FabricClock = mode_lib->vba.FabricClockPerState[mode_lib->vba.VoltageLevel];
+	mode_lib->vba.SOCCLK = mode_lib->vba.SOCCLKPerState[mode_lib->vba.VoltageLevel];
+	mode_lib->vba.ReturnBW = locals->ReturnBWPerState[mode_lib->vba.VoltageLevel][0];
+	mode_lib->vba.FabricAndDRAMBandwidth = locals->FabricAndDRAMBandwidthPerState[mode_lib->vba.VoltageLevel];
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.BlendingAndTiming[k] == k) {
+			mode_lib->vba.ODMCombineEnabled[k] =
+					locals->ODMCombineEnablePerState[mode_lib->vba.VoltageLevel][k];
+		} else {
+			mode_lib->vba.ODMCombineEnabled[k] = 0;
+		}
+		mode_lib->vba.DSCEnabled[k] =
+				locals->RequiresDSC[mode_lib->vba.VoltageLevel][k];
+		mode_lib->vba.OutputBpp[k] =
+				locals->OutputBppPerState[mode_lib->vba.VoltageLevel][k];
+	}
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_mode_vba_20v2.h b/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_mode_vba_20v2.h
new file mode 100644
index 000000000000..a989d3ca1e99
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_mode_vba_20v2.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef _DCN20V2_DISPLAY_MODE_VBA_H_
+#define _DCN20V2_DISPLAY_MODE_VBA_H_
+
+void dml20v2_recalculate(struct display_mode_lib *mode_lib);
+void dml20v2_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_lib);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_rq_dlg_calc_20.c b/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_rq_dlg_calc_20.c
new file mode 100644
index 000000000000..9c58ff1069d6
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_rq_dlg_calc_20.c
@@ -0,0 +1,1665 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "../display_mode_lib.h"
+#include "../display_mode_vba.h"
+#include "display_rq_dlg_calc_20.h"
+
+// Function: dml20_rq_dlg_get_rq_params
+//  Calculate requestor related parameters that register definition agnostic
+//  (i.e. this layer does try to separate real values from register definition)
+// Input:
+//  pipe_src_param - pipe source configuration (e.g. vp, pitch, etc.)
+// Output:
+//  rq_param - values that can be used to setup RQ (e.g. swath_height, plane1_addr, etc.)
+//
+static void dml20_rq_dlg_get_rq_params(
+		struct display_mode_lib *mode_lib,
+		display_rq_params_st *rq_param,
+		const display_pipe_source_params_st *pipe_src_param);
+
+// Function: dml20_rq_dlg_get_dlg_params
+//  Calculate deadline related parameters
+//
+static void dml20_rq_dlg_get_dlg_params(struct display_mode_lib *mode_lib,
+		const display_e2e_pipe_params_st *e2e_pipe_param,
+		const unsigned int num_pipes,
+		const unsigned int pipe_idx,
+		display_dlg_regs_st *disp_dlg_regs,
+		display_ttu_regs_st *disp_ttu_regs,
+		const display_rq_dlg_params_st *rq_dlg_param,
+		const display_dlg_sys_params_st *dlg_sys_param,
+		const bool cstate_en,
+		const bool pstate_en);
+/*
+ * NOTE:
+ *   This file is gcc-parseable HW gospel, coming straight from HW engineers.
+ *
+ * It doesn't adhere to Linux kernel style and sometimes will do things in odd
+ * ways. Unless there is something clearly wrong with it the code should
+ * remain as-is as it provides us with a guarantee from HW that it is correct.
+ */
+
+static void calculate_ttu_cursor(struct display_mode_lib *mode_lib,
+		double *refcyc_per_req_delivery_pre_cur,
+		double *refcyc_per_req_delivery_cur,
+		double refclk_freq_in_mhz,
+		double ref_freq_to_pix_freq,
+		double hscale_pixel_rate_l,
+		double hscl_ratio,
+		double vratio_pre_l,
+		double vratio_l,
+		unsigned int cur_width,
+		enum cursor_bpp cur_bpp);
+
+#include "../dml_inline_defs.h"
+
+static unsigned int get_bytes_per_element(enum source_format_class source_format, bool is_chroma)
+{
+	unsigned int ret_val = 1;
+
+	if (source_format == dm_444_16) {
+		if (!is_chroma)
+			ret_val = 2;
+	} else if (source_format == dm_444_32) {
+		if (!is_chroma)
+			ret_val = 4;
+	} else if (source_format == dm_444_64) {
+		if (!is_chroma)
+			ret_val = 8;
+	} else if (source_format == dm_420_8) {
+		if (is_chroma)
+			ret_val = 2;
+		else
+			ret_val = 1;
+	} else if (source_format == dm_420_10) {
+		if (is_chroma)
+			ret_val = 4;
+		else
+			ret_val = 2;
+	} else if (source_format == dm_444_8) {
+		ret_val = 1;
+	}
+	return ret_val;
+}
+
+static bool is_dual_plane(enum source_format_class source_format)
+{
+	bool ret_val = false;
+
+	if ((source_format == dm_420_8) || (source_format == dm_420_10))
+		ret_val = true;
+
+	return ret_val;
+}
+
+static double get_refcyc_per_delivery(struct display_mode_lib *mode_lib,
+		double refclk_freq_in_mhz,
+		double pclk_freq_in_mhz,
+		bool odm_combine,
+		unsigned int recout_width,
+		unsigned int hactive,
+		double vratio,
+		double hscale_pixel_rate,
+		unsigned int delivery_width,
+		unsigned int req_per_swath_ub)
+{
+	double refcyc_per_delivery = 0.0;
+
+	if (vratio <= 1.0) {
+		if (odm_combine)
+			refcyc_per_delivery = (double) refclk_freq_in_mhz
+					* dml_min((double) recout_width, (double) hactive / 2.0)
+					/ pclk_freq_in_mhz / (double) req_per_swath_ub;
+		else
+			refcyc_per_delivery = (double) refclk_freq_in_mhz * (double) recout_width
+					/ pclk_freq_in_mhz / (double) req_per_swath_ub;
+	} else {
+		refcyc_per_delivery = (double) refclk_freq_in_mhz * (double) delivery_width
+				/ (double) hscale_pixel_rate / (double) req_per_swath_ub;
+	}
+
+	dml_print("DML_DLG: %s: refclk_freq_in_mhz = %3.2f\n", __func__, refclk_freq_in_mhz);
+	dml_print("DML_DLG: %s: pclk_freq_in_mhz   = %3.2f\n", __func__, pclk_freq_in_mhz);
+	dml_print("DML_DLG: %s: recout_width       = %d\n", __func__, recout_width);
+	dml_print("DML_DLG: %s: vratio             = %3.2f\n", __func__, vratio);
+	dml_print("DML_DLG: %s: req_per_swath_ub   = %d\n", __func__, req_per_swath_ub);
+	dml_print("DML_DLG: %s: refcyc_per_delivery= %3.2f\n", __func__, refcyc_per_delivery);
+
+	return refcyc_per_delivery;
+
+}
+
+static unsigned int get_blk_size_bytes(const enum source_macro_tile_size tile_size)
+{
+	if (tile_size == dm_256k_tile)
+		return (256 * 1024);
+	else if (tile_size == dm_64k_tile)
+		return (64 * 1024);
+	else
+		return (4 * 1024);
+}
+
+static void extract_rq_sizing_regs(struct display_mode_lib *mode_lib,
+		display_data_rq_regs_st *rq_regs,
+		const display_data_rq_sizing_params_st *rq_sizing)
+{
+	dml_print("DML_DLG: %s: rq_sizing param\n", __func__);
+	print__data_rq_sizing_params_st(mode_lib, rq_sizing);
+
+	rq_regs->chunk_size = dml_log2(rq_sizing->chunk_bytes) - 10;
+
+	if (rq_sizing->min_chunk_bytes == 0)
+		rq_regs->min_chunk_size = 0;
+	else
+		rq_regs->min_chunk_size = dml_log2(rq_sizing->min_chunk_bytes) - 8 + 1;
+
+	rq_regs->meta_chunk_size = dml_log2(rq_sizing->meta_chunk_bytes) - 10;
+	if (rq_sizing->min_meta_chunk_bytes == 0)
+		rq_regs->min_meta_chunk_size = 0;
+	else
+		rq_regs->min_meta_chunk_size = dml_log2(rq_sizing->min_meta_chunk_bytes) - 6 + 1;
+
+	rq_regs->dpte_group_size = dml_log2(rq_sizing->dpte_group_bytes) - 6;
+	rq_regs->mpte_group_size = dml_log2(rq_sizing->mpte_group_bytes) - 6;
+}
+
+static void extract_rq_regs(struct display_mode_lib *mode_lib,
+		display_rq_regs_st *rq_regs,
+		const display_rq_params_st *rq_param)
+{
+	unsigned int detile_buf_size_in_bytes = mode_lib->ip.det_buffer_size_kbytes * 1024;
+	unsigned int detile_buf_plane1_addr = 0;
+
+	extract_rq_sizing_regs(mode_lib, &(rq_regs->rq_regs_l), &rq_param->sizing.rq_l);
+
+	rq_regs->rq_regs_l.pte_row_height_linear = dml_floor(dml_log2(rq_param->dlg.rq_l.dpte_row_height),
+			1) - 3;
+
+	if (rq_param->yuv420) {
+		extract_rq_sizing_regs(mode_lib, &(rq_regs->rq_regs_c), &rq_param->sizing.rq_c);
+		rq_regs->rq_regs_c.pte_row_height_linear = dml_floor(dml_log2(rq_param->dlg.rq_c.dpte_row_height),
+				1) - 3;
+	}
+
+	rq_regs->rq_regs_l.swath_height = dml_log2(rq_param->dlg.rq_l.swath_height);
+	rq_regs->rq_regs_c.swath_height = dml_log2(rq_param->dlg.rq_c.swath_height);
+
+	// TODO: take the max between luma, chroma chunk size?
+	// okay for now, as we are setting chunk_bytes to 8kb anyways
+	if (rq_param->sizing.rq_l.chunk_bytes >= 32 * 1024) { //32kb
+		rq_regs->drq_expansion_mode = 0;
+	} else {
+		rq_regs->drq_expansion_mode = 2;
+	}
+	rq_regs->prq_expansion_mode = 1;
+	rq_regs->mrq_expansion_mode = 1;
+	rq_regs->crq_expansion_mode = 1;
+
+	if (rq_param->yuv420) {
+		if ((double) rq_param->misc.rq_l.stored_swath_bytes
+				/ (double) rq_param->misc.rq_c.stored_swath_bytes <= 1.5) {
+			detile_buf_plane1_addr = (detile_buf_size_in_bytes / 2.0 / 64.0); // half to chroma
+		} else {
+			detile_buf_plane1_addr = dml_round_to_multiple((unsigned int) ((2.0 * detile_buf_size_in_bytes) / 3.0),
+					256,
+					0) / 64.0; // 2/3 to chroma
+		}
+	}
+	rq_regs->plane1_base_address = detile_buf_plane1_addr;
+}
+
+static void handle_det_buf_split(struct display_mode_lib *mode_lib,
+		display_rq_params_st *rq_param,
+		const display_pipe_source_params_st *pipe_src_param)
+{
+	unsigned int total_swath_bytes = 0;
+	unsigned int swath_bytes_l = 0;
+	unsigned int swath_bytes_c = 0;
+	unsigned int full_swath_bytes_packed_l = 0;
+	unsigned int full_swath_bytes_packed_c = 0;
+	bool req128_l = false;
+	bool req128_c = false;
+	bool surf_linear = (pipe_src_param->sw_mode == dm_sw_linear);
+	bool surf_vert = (pipe_src_param->source_scan == dm_vert);
+	unsigned int log2_swath_height_l = 0;
+	unsigned int log2_swath_height_c = 0;
+	unsigned int detile_buf_size_in_bytes = mode_lib->ip.det_buffer_size_kbytes * 1024;
+
+	full_swath_bytes_packed_l = rq_param->misc.rq_l.full_swath_bytes;
+	full_swath_bytes_packed_c = rq_param->misc.rq_c.full_swath_bytes;
+
+	if (rq_param->yuv420_10bpc) {
+		full_swath_bytes_packed_l = dml_round_to_multiple(rq_param->misc.rq_l.full_swath_bytes * 2 / 3,
+				256,
+				1) + 256;
+		full_swath_bytes_packed_c = dml_round_to_multiple(rq_param->misc.rq_c.full_swath_bytes * 2 / 3,
+				256,
+				1) + 256;
+	}
+
+	if (rq_param->yuv420) {
+		total_swath_bytes = 2 * full_swath_bytes_packed_l + 2 * full_swath_bytes_packed_c;
+
+		if (total_swath_bytes <= detile_buf_size_in_bytes) { //full 256b request
+			req128_l = false;
+			req128_c = false;
+			swath_bytes_l = full_swath_bytes_packed_l;
+			swath_bytes_c = full_swath_bytes_packed_c;
+		} else { //128b request (for luma only for yuv420 8bpc)
+			req128_l = true;
+			req128_c = false;
+			swath_bytes_l = full_swath_bytes_packed_l / 2;
+			swath_bytes_c = full_swath_bytes_packed_c;
+		}
+		// Note: assumption, the config that pass in will fit into
+		//       the detiled buffer.
+	} else {
+		total_swath_bytes = 2 * full_swath_bytes_packed_l;
+
+		if (total_swath_bytes <= detile_buf_size_in_bytes)
+			req128_l = false;
+		else
+			req128_l = true;
+
+		swath_bytes_l = total_swath_bytes;
+		swath_bytes_c = 0;
+	}
+	rq_param->misc.rq_l.stored_swath_bytes = swath_bytes_l;
+	rq_param->misc.rq_c.stored_swath_bytes = swath_bytes_c;
+
+	if (surf_linear) {
+		log2_swath_height_l = 0;
+		log2_swath_height_c = 0;
+	} else {
+		unsigned int swath_height_l;
+		unsigned int swath_height_c;
+
+		if (!surf_vert) {
+			swath_height_l = rq_param->misc.rq_l.blk256_height;
+			swath_height_c = rq_param->misc.rq_c.blk256_height;
+		} else {
+			swath_height_l = rq_param->misc.rq_l.blk256_width;
+			swath_height_c = rq_param->misc.rq_c.blk256_width;
+		}
+
+		if (swath_height_l > 0)
+			log2_swath_height_l = dml_log2(swath_height_l);
+
+		if (req128_l && log2_swath_height_l > 0)
+			log2_swath_height_l -= 1;
+
+		if (swath_height_c > 0)
+			log2_swath_height_c = dml_log2(swath_height_c);
+	}
+
+	rq_param->dlg.rq_l.swath_height = 1 << log2_swath_height_l;
+	rq_param->dlg.rq_c.swath_height = 1 << log2_swath_height_c;
+
+	dml_print("DML_DLG: %s: req128_l = %0d\n", __func__, req128_l);
+	dml_print("DML_DLG: %s: req128_c = %0d\n", __func__, req128_c);
+	dml_print("DML_DLG: %s: full_swath_bytes_packed_l = %0d\n",
+			__func__,
+			full_swath_bytes_packed_l);
+	dml_print("DML_DLG: %s: full_swath_bytes_packed_c = %0d\n",
+			__func__,
+			full_swath_bytes_packed_c);
+}
+
+static void get_meta_and_pte_attr(struct display_mode_lib *mode_lib,
+		display_data_rq_dlg_params_st *rq_dlg_param,
+		display_data_rq_misc_params_st *rq_misc_param,
+		display_data_rq_sizing_params_st *rq_sizing_param,
+		unsigned int vp_width,
+		unsigned int vp_height,
+		unsigned int data_pitch,
+		unsigned int meta_pitch,
+		unsigned int source_format,
+		unsigned int tiling,
+		unsigned int macro_tile_size,
+		unsigned int source_scan,
+		unsigned int is_chroma)
+{
+	bool surf_linear = (tiling == dm_sw_linear);
+	bool surf_vert = (source_scan == dm_vert);
+
+	unsigned int bytes_per_element;
+	unsigned int bytes_per_element_y = get_bytes_per_element((enum source_format_class)(source_format),
+			false);
+	unsigned int bytes_per_element_c = get_bytes_per_element((enum source_format_class)(source_format),
+			true);
+
+	unsigned int blk256_width = 0;
+	unsigned int blk256_height = 0;
+
+	unsigned int blk256_width_y = 0;
+	unsigned int blk256_height_y = 0;
+	unsigned int blk256_width_c = 0;
+	unsigned int blk256_height_c = 0;
+	unsigned int log2_bytes_per_element;
+	unsigned int log2_blk256_width;
+	unsigned int log2_blk256_height;
+	unsigned int blk_bytes;
+	unsigned int log2_blk_bytes;
+	unsigned int log2_blk_height;
+	unsigned int log2_blk_width;
+	unsigned int log2_meta_req_bytes;
+	unsigned int log2_meta_req_height;
+	unsigned int log2_meta_req_width;
+	unsigned int meta_req_width;
+	unsigned int meta_req_height;
+	unsigned int log2_meta_row_height;
+	unsigned int meta_row_width_ub;
+	unsigned int log2_meta_chunk_bytes;
+	unsigned int log2_meta_chunk_height;
+
+	//full sized meta chunk width in unit of data elements
+	unsigned int log2_meta_chunk_width;
+	unsigned int log2_min_meta_chunk_bytes;
+	unsigned int min_meta_chunk_width;
+	unsigned int meta_chunk_width;
+	unsigned int meta_chunk_per_row_int;
+	unsigned int meta_row_remainder;
+	unsigned int meta_chunk_threshold;
+	unsigned int meta_blk_bytes;
+	unsigned int meta_blk_height;
+	unsigned int meta_blk_width;
+	unsigned int meta_surface_bytes;
+	unsigned int vmpg_bytes;
+	unsigned int meta_pte_req_per_frame_ub;
+	unsigned int meta_pte_bytes_per_frame_ub;
+	const unsigned int log2_vmpg_bytes = dml_log2(mode_lib->soc.vmm_page_size_bytes);
+	const unsigned int dpte_buf_in_pte_reqs = mode_lib->ip.dpte_buffer_size_in_pte_reqs_luma;
+	const unsigned int pde_proc_buffer_size_64k_reqs =
+			mode_lib->ip.pde_proc_buffer_size_64k_reqs;
+
+	unsigned int log2_vmpg_height = 0;
+	unsigned int log2_vmpg_width = 0;
+	unsigned int log2_dpte_req_height_ptes = 0;
+	unsigned int log2_dpte_req_height = 0;
+	unsigned int log2_dpte_req_width = 0;
+	unsigned int log2_dpte_row_height_linear = 0;
+	unsigned int log2_dpte_row_height = 0;
+	unsigned int log2_dpte_group_width = 0;
+	unsigned int dpte_row_width_ub = 0;
+	unsigned int dpte_req_height = 0;
+	unsigned int dpte_req_width = 0;
+	unsigned int dpte_group_width = 0;
+	unsigned int log2_dpte_group_bytes = 0;
+	unsigned int log2_dpte_group_length = 0;
+	unsigned int pde_buf_entries;
+	bool yuv420 = (source_format == dm_420_8 || source_format == dm_420_10);
+
+	Calculate256BBlockSizes((enum source_format_class)(source_format),
+			(enum dm_swizzle_mode)(tiling),
+			bytes_per_element_y,
+			bytes_per_element_c,
+			&blk256_height_y,
+			&blk256_height_c,
+			&blk256_width_y,
+			&blk256_width_c);
+
+	if (!is_chroma) {
+		blk256_width = blk256_width_y;
+		blk256_height = blk256_height_y;
+		bytes_per_element = bytes_per_element_y;
+	} else {
+		blk256_width = blk256_width_c;
+		blk256_height = blk256_height_c;
+		bytes_per_element = bytes_per_element_c;
+	}
+
+	log2_bytes_per_element = dml_log2(bytes_per_element);
+
+	dml_print("DML_DLG: %s: surf_linear        = %d\n", __func__, surf_linear);
+	dml_print("DML_DLG: %s: surf_vert          = %d\n", __func__, surf_vert);
+	dml_print("DML_DLG: %s: blk256_width       = %d\n", __func__, blk256_width);
+	dml_print("DML_DLG: %s: blk256_height      = %d\n", __func__, blk256_height);
+
+	log2_blk256_width = dml_log2((double) blk256_width);
+	log2_blk256_height = dml_log2((double) blk256_height);
+	blk_bytes = surf_linear ?
+			256 : get_blk_size_bytes((enum source_macro_tile_size) macro_tile_size);
+	log2_blk_bytes = dml_log2((double) blk_bytes);
+
+	// remember log rule
+	// "+" in log is multiply
+	// "-" in log is divide
+	// "/2" is like square root
+	// blk is vertical biased
+	if (tiling != dm_sw_linear)
+		log2_blk_height = log2_blk256_height
+				+ dml_ceil((double) (log2_blk_bytes - 8) / 2.0, 1);
+	else
+		log2_blk_height = 0;  // blk height of 1
+
+	log2_blk_width = log2_blk_bytes - log2_bytes_per_element - log2_blk_height;
+
+	if (!surf_vert) {
+		rq_dlg_param->swath_width_ub = dml_round_to_multiple(vp_width - 1, blk256_width, 1)
+				+ blk256_width;
+		rq_dlg_param->req_per_swath_ub = rq_dlg_param->swath_width_ub >> log2_blk256_width;
+	} else {
+		rq_dlg_param->swath_width_ub = dml_round_to_multiple(vp_height - 1, blk256_height, 1)
+				+ blk256_height;
+		rq_dlg_param->req_per_swath_ub = rq_dlg_param->swath_width_ub >> log2_blk256_height;
+	}
+
+	if (!surf_vert)
+		rq_misc_param->full_swath_bytes = rq_dlg_param->swath_width_ub * blk256_height
+				* bytes_per_element;
+	else
+		rq_misc_param->full_swath_bytes = rq_dlg_param->swath_width_ub * blk256_width
+				* bytes_per_element;
+
+	rq_misc_param->blk256_height = blk256_height;
+	rq_misc_param->blk256_width = blk256_width;
+
+	// -------
+	// meta
+	// -------
+	log2_meta_req_bytes = 6; // meta request is 64b and is 8x8byte meta element
+
+	// each 64b meta request for dcn is 8x8 meta elements and
+	// a meta element covers one 256b block of the data surface.
+	log2_meta_req_height = log2_blk256_height + 3; // meta req is 8x8 byte, each byte represent 1 blk256
+	log2_meta_req_width = log2_meta_req_bytes + 8 - log2_bytes_per_element
+			- log2_meta_req_height;
+	meta_req_width = 1 << log2_meta_req_width;
+	meta_req_height = 1 << log2_meta_req_height;
+
+	// the dimensions of a meta row are meta_row_width x meta_row_height in elements.
+	// calculate upper bound of the meta_row_width
+	if (!surf_vert) {
+		log2_meta_row_height = log2_meta_req_height;
+		meta_row_width_ub = dml_round_to_multiple(vp_width - 1, meta_req_width, 1)
+				+ meta_req_width;
+		rq_dlg_param->meta_req_per_row_ub = meta_row_width_ub / meta_req_width;
+	} else {
+		log2_meta_row_height = log2_meta_req_width;
+		meta_row_width_ub = dml_round_to_multiple(vp_height - 1, meta_req_height, 1)
+				+ meta_req_height;
+		rq_dlg_param->meta_req_per_row_ub = meta_row_width_ub / meta_req_height;
+	}
+	rq_dlg_param->meta_bytes_per_row_ub = rq_dlg_param->meta_req_per_row_ub * 64;
+
+	rq_dlg_param->meta_row_height = 1 << log2_meta_row_height;
+
+	log2_meta_chunk_bytes = dml_log2(rq_sizing_param->meta_chunk_bytes);
+	log2_meta_chunk_height = log2_meta_row_height;
+
+	//full sized meta chunk width in unit of data elements
+	log2_meta_chunk_width = log2_meta_chunk_bytes + 8 - log2_bytes_per_element
+			- log2_meta_chunk_height;
+	log2_min_meta_chunk_bytes = dml_log2(rq_sizing_param->min_meta_chunk_bytes);
+	min_meta_chunk_width = 1
+			<< (log2_min_meta_chunk_bytes + 8 - log2_bytes_per_element
+					- log2_meta_chunk_height);
+	meta_chunk_width = 1 << log2_meta_chunk_width;
+	meta_chunk_per_row_int = (unsigned int) (meta_row_width_ub / meta_chunk_width);
+	meta_row_remainder = meta_row_width_ub % meta_chunk_width;
+	meta_chunk_threshold = 0;
+	meta_blk_bytes = 4096;
+	meta_blk_height = blk256_height * 64;
+	meta_blk_width = meta_blk_bytes * 256 / bytes_per_element / meta_blk_height;
+	meta_surface_bytes = meta_pitch
+			* (dml_round_to_multiple(vp_height - 1, meta_blk_height, 1) + meta_blk_height)
+			* bytes_per_element / 256;
+	vmpg_bytes = mode_lib->soc.vmm_page_size_bytes;
+	meta_pte_req_per_frame_ub = (dml_round_to_multiple(meta_surface_bytes - vmpg_bytes,
+			8 * vmpg_bytes,
+			1) + 8 * vmpg_bytes) / (8 * vmpg_bytes);
+	meta_pte_bytes_per_frame_ub = meta_pte_req_per_frame_ub * 64; //64B mpte request
+	rq_dlg_param->meta_pte_bytes_per_frame_ub = meta_pte_bytes_per_frame_ub;
+
+	dml_print("DML_DLG: %s: meta_blk_height             = %d\n", __func__, meta_blk_height);
+	dml_print("DML_DLG: %s: meta_blk_width              = %d\n", __func__, meta_blk_width);
+	dml_print("DML_DLG: %s: meta_surface_bytes          = %d\n", __func__, meta_surface_bytes);
+	dml_print("DML_DLG: %s: meta_pte_req_per_frame_ub   = %d\n",
+			__func__,
+			meta_pte_req_per_frame_ub);
+	dml_print("DML_DLG: %s: meta_pte_bytes_per_frame_ub = %d\n",
+			__func__,
+			meta_pte_bytes_per_frame_ub);
+
+	if (!surf_vert)
+		meta_chunk_threshold = 2 * min_meta_chunk_width - meta_req_width;
+	else
+		meta_chunk_threshold = 2 * min_meta_chunk_width - meta_req_height;
+
+	if (meta_row_remainder <= meta_chunk_threshold)
+		rq_dlg_param->meta_chunks_per_row_ub = meta_chunk_per_row_int + 1;
+	else
+		rq_dlg_param->meta_chunks_per_row_ub = meta_chunk_per_row_int + 2;
+
+	// ------
+	// dpte
+	// ------
+	if (surf_linear) {
+		log2_vmpg_height = 0;   // one line high
+	} else {
+		log2_vmpg_height = (log2_vmpg_bytes - 8) / 2 + log2_blk256_height;
+	}
+	log2_vmpg_width = log2_vmpg_bytes - log2_bytes_per_element - log2_vmpg_height;
+
+	// only 3 possible shapes for dpte request in dimensions of ptes: 8x1, 4x2, 2x4.
+	if (surf_linear) { //one 64B PTE request returns 8 PTEs
+		log2_dpte_req_height_ptes = 0;
+		log2_dpte_req_width = log2_vmpg_width + 3;
+		log2_dpte_req_height = 0;
+	} else if (log2_blk_bytes == 12) { //4KB tile means 4kB page size
+		//one 64B req gives 8x1 PTEs for 4KB tile
+		log2_dpte_req_height_ptes = 0;
+		log2_dpte_req_width = log2_blk_width + 3;
+		log2_dpte_req_height = log2_blk_height + 0;
+	} else if ((log2_blk_bytes >= 16) && (log2_vmpg_bytes == 12)) { // tile block >= 64KB
+		//two 64B reqs of 2x4 PTEs give 16 PTEs to cover 64KB
+		log2_dpte_req_height_ptes = 4;
+		log2_dpte_req_width = log2_blk256_width + 4; // log2_64KB_width
+		log2_dpte_req_height = log2_blk256_height + 4; // log2_64KB_height
+	} else { //64KB page size and must 64KB tile block
+		 //one 64B req gives 8x1 PTEs for 64KB tile
+		log2_dpte_req_height_ptes = 0;
+		log2_dpte_req_width = log2_blk_width + 3;
+		log2_dpte_req_height = log2_blk_height + 0;
+	}
+
+	// The dpte request dimensions in data elements is dpte_req_width x dpte_req_height
+	// log2_vmpg_width is how much 1 pte represent, now calculating how much a 64b pte req represent
+	// That depends on the pte shape (i.e. 8x1, 4x2, 2x4)
+	//log2_dpte_req_height    = log2_vmpg_height + log2_dpte_req_height_ptes;
+	//log2_dpte_req_width     = log2_vmpg_width + log2_dpte_req_width_ptes;
+	dpte_req_height = 1 << log2_dpte_req_height;
+	dpte_req_width = 1 << log2_dpte_req_width;
+
+	// calculate pitch dpte row buffer can hold
+	// round the result down to a power of two.
+	pde_buf_entries = yuv420 ? (pde_proc_buffer_size_64k_reqs >> 1) : pde_proc_buffer_size_64k_reqs;
+	if (surf_linear) {
+		unsigned int dpte_row_height;
+
+		log2_dpte_row_height_linear = dml_floor(dml_log2(dml_min(64 * 1024 * pde_buf_entries
+										/ bytes_per_element,
+								dpte_buf_in_pte_reqs
+										* dpte_req_width)
+								/ data_pitch),
+				1);
+
+		ASSERT(log2_dpte_row_height_linear >= 3);
+
+		if (log2_dpte_row_height_linear > 7)
+			log2_dpte_row_height_linear = 7;
+
+		log2_dpte_row_height = log2_dpte_row_height_linear;
+		// For linear, the dpte row is pitch dependent and the pte requests wrap at the pitch boundary.
+		// the dpte_row_width_ub is the upper bound of data_pitch*dpte_row_height in elements with this unique buffering.
+		dpte_row_height = 1 << log2_dpte_row_height;
+		dpte_row_width_ub = dml_round_to_multiple(data_pitch * dpte_row_height - 1,
+				dpte_req_width,
+				1) + dpte_req_width;
+		rq_dlg_param->dpte_req_per_row_ub = dpte_row_width_ub / dpte_req_width;
+	} else {
+		// the upper bound of the dpte_row_width without dependency on viewport position follows.
+		// for tiled mode, row height is the same as req height and row store up to vp size upper bound
+		if (!surf_vert) {
+			log2_dpte_row_height = log2_dpte_req_height;
+			dpte_row_width_ub = dml_round_to_multiple(vp_width - 1, dpte_req_width, 1)
+					+ dpte_req_width;
+			rq_dlg_param->dpte_req_per_row_ub = dpte_row_width_ub / dpte_req_width;
+		} else {
+			log2_dpte_row_height =
+					(log2_blk_width < log2_dpte_req_width) ?
+							log2_blk_width : log2_dpte_req_width;
+			dpte_row_width_ub = dml_round_to_multiple(vp_height - 1, dpte_req_height, 1)
+					+ dpte_req_height;
+			rq_dlg_param->dpte_req_per_row_ub = dpte_row_width_ub / dpte_req_height;
+		}
+	}
+	if (log2_blk_bytes >= 16 && log2_vmpg_bytes == 12) // tile block >= 64KB
+		rq_dlg_param->dpte_bytes_per_row_ub = rq_dlg_param->dpte_req_per_row_ub * 128; //2*64B dpte request
+	else
+		rq_dlg_param->dpte_bytes_per_row_ub = rq_dlg_param->dpte_req_per_row_ub * 64; //64B dpte request
+
+	rq_dlg_param->dpte_row_height = 1 << log2_dpte_row_height;
+
+	// the dpte_group_bytes is reduced for the specific case of vertical
+	// access of a tile surface that has dpte request of 8x1 ptes.
+	if (!surf_linear && (log2_dpte_req_height_ptes == 0) && surf_vert) //reduced, in this case, will have page fault within a group
+		rq_sizing_param->dpte_group_bytes = 512;
+	else
+		//full size
+		rq_sizing_param->dpte_group_bytes = 2048;
+
+	//since pte request size is 64byte, the number of data pte requests per full sized group is as follows.
+	log2_dpte_group_bytes = dml_log2(rq_sizing_param->dpte_group_bytes);
+	log2_dpte_group_length = log2_dpte_group_bytes - 6; //length in 64b requests
+
+	// full sized data pte group width in elements
+	if (!surf_vert)
+		log2_dpte_group_width = log2_dpte_group_length + log2_dpte_req_width;
+	else
+		log2_dpte_group_width = log2_dpte_group_length + log2_dpte_req_height;
+
+	//But if the tile block >=64KB and the page size is 4KB, then each dPTE request is 2*64B
+	if ((log2_blk_bytes >= 16) && (log2_vmpg_bytes == 12)) // tile block >= 64KB
+		log2_dpte_group_width = log2_dpte_group_width - 1;
+
+	dpte_group_width = 1 << log2_dpte_group_width;
+
+	// since dpte groups are only aligned to dpte_req_width and not dpte_group_width,
+	// the upper bound for the dpte groups per row is as follows.
+	rq_dlg_param->dpte_groups_per_row_ub = dml_ceil((double) dpte_row_width_ub / dpte_group_width,
+			1);
+}
+
+static void get_surf_rq_param(struct display_mode_lib *mode_lib,
+		display_data_rq_sizing_params_st *rq_sizing_param,
+		display_data_rq_dlg_params_st *rq_dlg_param,
+		display_data_rq_misc_params_st *rq_misc_param,
+		const display_pipe_source_params_st *pipe_src_param,
+		bool is_chroma)
+{
+	unsigned int vp_width = 0;
+	unsigned int vp_height = 0;
+	unsigned int data_pitch = 0;
+	unsigned int meta_pitch = 0;
+	unsigned int ppe = 1;
+
+	// TODO check if ppe apply for both luma and chroma in 422 case
+	if (is_chroma) {
+		vp_width = pipe_src_param->viewport_width_c / ppe;
+		vp_height = pipe_src_param->viewport_height_c;
+		data_pitch = pipe_src_param->data_pitch_c;
+		meta_pitch = pipe_src_param->meta_pitch_c;
+	} else {
+		vp_width = pipe_src_param->viewport_width / ppe;
+		vp_height = pipe_src_param->viewport_height;
+		data_pitch = pipe_src_param->data_pitch;
+		meta_pitch = pipe_src_param->meta_pitch;
+	}
+
+	rq_sizing_param->chunk_bytes = 8192;
+
+	if (rq_sizing_param->chunk_bytes == 64 * 1024)
+		rq_sizing_param->min_chunk_bytes = 0;
+	else
+		rq_sizing_param->min_chunk_bytes = 1024;
+
+	rq_sizing_param->meta_chunk_bytes = 2048;
+	rq_sizing_param->min_meta_chunk_bytes = 256;
+
+	rq_sizing_param->mpte_group_bytes = 2048;
+
+	get_meta_and_pte_attr(mode_lib,
+			rq_dlg_param,
+			rq_misc_param,
+			rq_sizing_param,
+			vp_width,
+			vp_height,
+			data_pitch,
+			meta_pitch,
+			pipe_src_param->source_format,
+			pipe_src_param->sw_mode,
+			pipe_src_param->macro_tile_size,
+			pipe_src_param->source_scan,
+			is_chroma);
+}
+
+static void dml20_rq_dlg_get_rq_params(struct display_mode_lib *mode_lib,
+		display_rq_params_st *rq_param,
+		const display_pipe_source_params_st *pipe_src_param)
+{
+	// get param for luma surface
+	rq_param->yuv420 = pipe_src_param->source_format == dm_420_8
+			|| pipe_src_param->source_format == dm_420_10;
+	rq_param->yuv420_10bpc = pipe_src_param->source_format == dm_420_10;
+
+	get_surf_rq_param(mode_lib,
+			&(rq_param->sizing.rq_l),
+			&(rq_param->dlg.rq_l),
+			&(rq_param->misc.rq_l),
+			pipe_src_param,
+			0);
+
+	if (is_dual_plane((enum source_format_class)(pipe_src_param->source_format))) {
+		// get param for chroma surface
+		get_surf_rq_param(mode_lib,
+				&(rq_param->sizing.rq_c),
+				&(rq_param->dlg.rq_c),
+				&(rq_param->misc.rq_c),
+				pipe_src_param,
+				1);
+	}
+
+	// calculate how to split the det buffer space between luma and chroma
+	handle_det_buf_split(mode_lib, rq_param, pipe_src_param);
+	print__rq_params_st(mode_lib, rq_param);
+}
+
+void dml20_rq_dlg_get_rq_reg(struct display_mode_lib *mode_lib,
+		display_rq_regs_st *rq_regs,
+		const display_pipe_params_st *pipe_param)
+{
+	display_rq_params_st rq_param = {0};
+
+	memset(rq_regs, 0, sizeof(*rq_regs));
+	dml20_rq_dlg_get_rq_params(mode_lib, &rq_param, &pipe_param->src);
+	extract_rq_regs(mode_lib, rq_regs, &rq_param);
+
+	print__rq_regs_st(mode_lib, rq_regs);
+}
+
+// Note: currently taken in as is.
+// Nice to decouple code from hw register implement and extract code that are repeated for luma and chroma.
+static void dml20_rq_dlg_get_dlg_params(struct display_mode_lib *mode_lib,
+		const display_e2e_pipe_params_st *e2e_pipe_param,
+		const unsigned int num_pipes,
+		const unsigned int pipe_idx,
+		display_dlg_regs_st *disp_dlg_regs,
+		display_ttu_regs_st *disp_ttu_regs,
+		const display_rq_dlg_params_st *rq_dlg_param,
+		const display_dlg_sys_params_st *dlg_sys_param,
+		const bool cstate_en,
+		const bool pstate_en)
+{
+	const display_pipe_source_params_st *src = &e2e_pipe_param[pipe_idx].pipe.src;
+	const display_pipe_dest_params_st *dst = &e2e_pipe_param[pipe_idx].pipe.dest;
+	const display_output_params_st *dout = &e2e_pipe_param[pipe_idx].dout;
+	const display_clocks_and_cfg_st *clks = &e2e_pipe_param[pipe_idx].clks_cfg;
+	const scaler_ratio_depth_st *scl = &e2e_pipe_param[pipe_idx].pipe.scale_ratio_depth;
+	const scaler_taps_st *taps = &e2e_pipe_param[pipe_idx].pipe.scale_taps;
+
+	// -------------------------
+	// Section 1.15.2.1: OTG dependent Params
+	// -------------------------
+	// Timing
+	unsigned int htotal = dst->htotal;
+//    unsigned int hblank_start = dst.hblank_start; // TODO: Remove
+	unsigned int hblank_end = dst->hblank_end;
+	unsigned int vblank_start = dst->vblank_start;
+	unsigned int vblank_end = dst->vblank_end;
+	unsigned int min_vblank = mode_lib->ip.min_vblank_lines;
+
+	double dppclk_freq_in_mhz = clks->dppclk_mhz;
+	double dispclk_freq_in_mhz = clks->dispclk_mhz;
+	double refclk_freq_in_mhz = clks->refclk_mhz;
+	double pclk_freq_in_mhz = dst->pixel_rate_mhz;
+	bool interlaced = dst->interlaced;
+
+	double ref_freq_to_pix_freq = refclk_freq_in_mhz / pclk_freq_in_mhz;
+
+	double min_dcfclk_mhz;
+	double t_calc_us;
+	double min_ttu_vblank;
+
+	double min_dst_y_ttu_vblank;
+	unsigned int dlg_vblank_start;
+	bool dual_plane;
+	unsigned int access_dir;
+	unsigned int vp_height_l;
+	unsigned int vp_width_l;
+	unsigned int vp_height_c;
+	unsigned int vp_width_c;
+
+	// Scaling
+	unsigned int htaps_l;
+	unsigned int htaps_c;
+	double hratio_l;
+	double hratio_c;
+	double vratio_l;
+	double vratio_c;
+	bool scl_enable;
+
+	double line_time_in_us;
+	//    double vinit_l;
+	//    double vinit_c;
+	//    double vinit_bot_l;
+	//    double vinit_bot_c;
+
+	//    unsigned int swath_height_l;
+	unsigned int swath_width_ub_l;
+	//    unsigned int dpte_bytes_per_row_ub_l;
+	unsigned int dpte_groups_per_row_ub_l;
+	//    unsigned int meta_pte_bytes_per_frame_ub_l;
+	//    unsigned int meta_bytes_per_row_ub_l;
+
+	//    unsigned int swath_height_c;
+	unsigned int swath_width_ub_c;
+	//   unsigned int dpte_bytes_per_row_ub_c;
+	unsigned int dpte_groups_per_row_ub_c;
+
+	unsigned int meta_chunks_per_row_ub_l;
+	unsigned int meta_chunks_per_row_ub_c;
+	unsigned int vupdate_offset;
+	unsigned int vupdate_width;
+	unsigned int vready_offset;
+
+	unsigned int dppclk_delay_subtotal;
+	unsigned int dispclk_delay_subtotal;
+	unsigned int pixel_rate_delay_subtotal;
+
+	unsigned int vstartup_start;
+	unsigned int dst_x_after_scaler;
+	unsigned int dst_y_after_scaler;
+	double line_wait;
+	double dst_y_prefetch;
+	double dst_y_per_vm_vblank;
+	double dst_y_per_row_vblank;
+	double dst_y_per_vm_flip;
+	double dst_y_per_row_flip;
+	double min_dst_y_per_vm_vblank;
+	double min_dst_y_per_row_vblank;
+	double lsw;
+	double vratio_pre_l;
+	double vratio_pre_c;
+	unsigned int req_per_swath_ub_l;
+	unsigned int req_per_swath_ub_c;
+	unsigned int meta_row_height_l;
+	unsigned int meta_row_height_c;
+	unsigned int swath_width_pixels_ub_l;
+	unsigned int swath_width_pixels_ub_c;
+	unsigned int scaler_rec_in_width_l;
+	unsigned int scaler_rec_in_width_c;
+	unsigned int dpte_row_height_l;
+	unsigned int dpte_row_height_c;
+	double hscale_pixel_rate_l;
+	double hscale_pixel_rate_c;
+	double min_hratio_fact_l;
+	double min_hratio_fact_c;
+	double refcyc_per_line_delivery_pre_l;
+	double refcyc_per_line_delivery_pre_c;
+	double refcyc_per_line_delivery_l;
+	double refcyc_per_line_delivery_c;
+
+	double refcyc_per_req_delivery_pre_l;
+	double refcyc_per_req_delivery_pre_c;
+	double refcyc_per_req_delivery_l;
+	double refcyc_per_req_delivery_c;
+
+	unsigned int full_recout_width;
+	double refcyc_per_req_delivery_pre_cur0;
+	double refcyc_per_req_delivery_cur0;
+	double refcyc_per_req_delivery_pre_cur1;
+	double refcyc_per_req_delivery_cur1;
+
+	memset(disp_dlg_regs, 0, sizeof(*disp_dlg_regs));
+	memset(disp_ttu_regs, 0, sizeof(*disp_ttu_regs));
+
+	dml_print("DML_DLG: %s:  cstate_en = %d\n", __func__, cstate_en);
+	dml_print("DML_DLG: %s:  pstate_en = %d\n", __func__, pstate_en);
+
+	dml_print("DML_DLG: %s: dppclk_freq_in_mhz     = %3.2f\n", __func__, dppclk_freq_in_mhz);
+	dml_print("DML_DLG: %s: dispclk_freq_in_mhz    = %3.2f\n", __func__, dispclk_freq_in_mhz);
+	dml_print("DML_DLG: %s: refclk_freq_in_mhz     = %3.2f\n", __func__, refclk_freq_in_mhz);
+	dml_print("DML_DLG: %s: pclk_freq_in_mhz       = %3.2f\n", __func__, pclk_freq_in_mhz);
+	dml_print("DML_DLG: %s: interlaced             = %d\n", __func__, interlaced);
+	ASSERT(ref_freq_to_pix_freq < 4.0);
+
+	disp_dlg_regs->ref_freq_to_pix_freq =
+			(unsigned int) (ref_freq_to_pix_freq * dml_pow(2, 19));
+	disp_dlg_regs->refcyc_per_htotal = (unsigned int) (ref_freq_to_pix_freq * (double) htotal
+			* dml_pow(2, 8));
+	disp_dlg_regs->dlg_vblank_end = interlaced ? (vblank_end / 2) : vblank_end; // 15 bits
+	disp_dlg_regs->refcyc_h_blank_end = (unsigned int) ((double) hblank_end
+			* (double) ref_freq_to_pix_freq);
+	ASSERT(disp_dlg_regs->refcyc_h_blank_end < (unsigned int) dml_pow(2, 13));
+
+	min_dcfclk_mhz = dlg_sys_param->deepsleep_dcfclk_mhz;
+	t_calc_us = get_tcalc(mode_lib, e2e_pipe_param, num_pipes);
+	min_ttu_vblank = get_min_ttu_vblank(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
+
+	min_dst_y_ttu_vblank = min_ttu_vblank * pclk_freq_in_mhz / (double) htotal;
+	dlg_vblank_start = interlaced ? (vblank_start / 2) : vblank_start;
+
+	disp_dlg_regs->min_dst_y_next_start = (unsigned int) ((double) dlg_vblank_start * dml_pow(2, 2));
+	ASSERT(disp_dlg_regs->min_dst_y_next_start < (unsigned int) dml_pow(2, 18));
+
+	dml_print("DML_DLG: %s: min_dcfclk_mhz                         = %3.2f\n",
+			__func__,
+			min_dcfclk_mhz);
+	dml_print("DML_DLG: %s: min_ttu_vblank                         = %3.2f\n",
+			__func__,
+			min_ttu_vblank);
+	dml_print("DML_DLG: %s: min_dst_y_ttu_vblank                   = %3.2f\n",
+			__func__,
+			min_dst_y_ttu_vblank);
+	dml_print("DML_DLG: %s: t_calc_us                              = %3.2f\n",
+			__func__,
+			t_calc_us);
+	dml_print("DML_DLG: %s: disp_dlg_regs->min_dst_y_next_start    = 0x%0x\n",
+			__func__,
+			disp_dlg_regs->min_dst_y_next_start);
+	dml_print("DML_DLG: %s: ref_freq_to_pix_freq                   = %3.2f\n",
+			__func__,
+			ref_freq_to_pix_freq);
+
+	// -------------------------
+	// Section 1.15.2.2: Prefetch, Active and TTU
+	// -------------------------
+	// Prefetch Calc
+	// Source
+//             dcc_en              = src.dcc;
+	dual_plane = is_dual_plane((enum source_format_class)(src->source_format));
+	access_dir = (src->source_scan == dm_vert); // vp access direction: horizontal or vertical accessed
+//      bytes_per_element_l = get_bytes_per_element(source_format_class(src.source_format), 0);
+//      bytes_per_element_c = get_bytes_per_element(source_format_class(src.source_format), 1);
+	vp_height_l = src->viewport_height;
+	vp_width_l = src->viewport_width;
+	vp_height_c = src->viewport_height_c;
+	vp_width_c = src->viewport_width_c;
+
+	// Scaling
+	htaps_l = taps->htaps;
+	htaps_c = taps->htaps_c;
+	hratio_l = scl->hscl_ratio;
+	hratio_c = scl->hscl_ratio_c;
+	vratio_l = scl->vscl_ratio;
+	vratio_c = scl->vscl_ratio_c;
+	scl_enable = scl->scl_enable;
+
+	line_time_in_us = (htotal / pclk_freq_in_mhz);
+//     vinit_l         = scl.vinit;
+//     vinit_c         = scl.vinit_c;
+//     vinit_bot_l     = scl.vinit_bot;
+//     vinit_bot_c     = scl.vinit_bot_c;
+
+//    unsigned int swath_height_l                 = rq_dlg_param->rq_l.swath_height;
+	swath_width_ub_l = rq_dlg_param->rq_l.swath_width_ub;
+//    unsigned int dpte_bytes_per_row_ub_l        = rq_dlg_param->rq_l.dpte_bytes_per_row_ub;
+	dpte_groups_per_row_ub_l = rq_dlg_param->rq_l.dpte_groups_per_row_ub;
+//    unsigned int meta_pte_bytes_per_frame_ub_l  = rq_dlg_param->rq_l.meta_pte_bytes_per_frame_ub;
+//    unsigned int meta_bytes_per_row_ub_l        = rq_dlg_param->rq_l.meta_bytes_per_row_ub;
+
+//    unsigned int swath_height_c                 = rq_dlg_param->rq_c.swath_height;
+	swath_width_ub_c = rq_dlg_param->rq_c.swath_width_ub;
+	//   dpte_bytes_per_row_ub_c        = rq_dlg_param->rq_c.dpte_bytes_per_row_ub;
+	dpte_groups_per_row_ub_c = rq_dlg_param->rq_c.dpte_groups_per_row_ub;
+
+	meta_chunks_per_row_ub_l = rq_dlg_param->rq_l.meta_chunks_per_row_ub;
+	meta_chunks_per_row_ub_c = rq_dlg_param->rq_c.meta_chunks_per_row_ub;
+	vupdate_offset = dst->vupdate_offset;
+	vupdate_width = dst->vupdate_width;
+	vready_offset = dst->vready_offset;
+
+	dppclk_delay_subtotal = mode_lib->ip.dppclk_delay_subtotal;
+	dispclk_delay_subtotal = mode_lib->ip.dispclk_delay_subtotal;
+
+	if (scl_enable)
+		dppclk_delay_subtotal += mode_lib->ip.dppclk_delay_scl;
+	else
+		dppclk_delay_subtotal += mode_lib->ip.dppclk_delay_scl_lb_only;
+
+	dppclk_delay_subtotal += mode_lib->ip.dppclk_delay_cnvc_formatter
+			+ src->num_cursors * mode_lib->ip.dppclk_delay_cnvc_cursor;
+
+	if (dout->dsc_enable) {
+		double dsc_delay = get_dsc_delay(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
+
+		dispclk_delay_subtotal += dsc_delay;
+	}
+
+	pixel_rate_delay_subtotal = dppclk_delay_subtotal * pclk_freq_in_mhz / dppclk_freq_in_mhz
+			+ dispclk_delay_subtotal * pclk_freq_in_mhz / dispclk_freq_in_mhz;
+
+	vstartup_start = dst->vstartup_start;
+	if (interlaced) {
+		if (vstartup_start / 2.0
+				- (double) (vready_offset + vupdate_width + vupdate_offset) / htotal
+				<= vblank_end / 2.0)
+			disp_dlg_regs->vready_after_vcount0 = 1;
+		else
+			disp_dlg_regs->vready_after_vcount0 = 0;
+	} else {
+		if (vstartup_start
+				- (double) (vready_offset + vupdate_width + vupdate_offset) / htotal
+				<= vblank_end)
+			disp_dlg_regs->vready_after_vcount0 = 1;
+		else
+			disp_dlg_regs->vready_after_vcount0 = 0;
+	}
+
+	// TODO: Where is this coming from?
+	if (interlaced)
+		vstartup_start = vstartup_start / 2;
+
+	// TODO: What if this min_vblank doesn't match the value in the dml_config_settings.cpp?
+	if (vstartup_start >= min_vblank) {
+		dml_print("WARNING: DML_DLG: %s: vblank_start=%d vblank_end=%d\n",
+				__func__,
+				vblank_start,
+				vblank_end);
+		dml_print("WARNING: DML_DLG: %s: vstartup_start=%d should be less than min_vblank=%d\n",
+				__func__,
+				vstartup_start,
+				min_vblank);
+		min_vblank = vstartup_start + 1;
+		dml_print("WARNING: DML_DLG: %s: vstartup_start=%d should be less than min_vblank=%d\n",
+				__func__,
+				vstartup_start,
+				min_vblank);
+	}
+
+	dst_x_after_scaler = get_dst_x_after_scaler(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
+	dst_y_after_scaler = get_dst_y_after_scaler(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
+
+	dml_print("DML_DLG: %s: htotal                                 = %d\n", __func__, htotal);
+	dml_print("DML_DLG: %s: pixel_rate_delay_subtotal              = %d\n",
+			__func__,
+			pixel_rate_delay_subtotal);
+	dml_print("DML_DLG: %s: dst_x_after_scaler                     = %d\n",
+			__func__,
+			dst_x_after_scaler);
+	dml_print("DML_DLG: %s: dst_y_after_scaler                     = %d\n",
+			__func__,
+			dst_y_after_scaler);
+
+	// Lwait
+	line_wait = mode_lib->soc.urgent_latency_us;
+	if (cstate_en)
+		line_wait = dml_max(mode_lib->soc.sr_enter_plus_exit_time_us, line_wait);
+	if (pstate_en)
+		line_wait = dml_max(mode_lib->soc.dram_clock_change_latency_us
+						+ mode_lib->soc.urgent_latency_us,
+				line_wait);
+	line_wait = line_wait / line_time_in_us;
+
+	dst_y_prefetch = get_dst_y_prefetch(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
+	dml_print("DML_DLG: %s: dst_y_prefetch (after rnd) = %3.2f\n", __func__, dst_y_prefetch);
+
+	dst_y_per_vm_vblank = get_dst_y_per_vm_vblank(mode_lib,
+			e2e_pipe_param,
+			num_pipes,
+			pipe_idx);
+	dst_y_per_row_vblank = get_dst_y_per_row_vblank(mode_lib,
+			e2e_pipe_param,
+			num_pipes,
+			pipe_idx);
+	dst_y_per_vm_flip = get_dst_y_per_vm_flip(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
+	dst_y_per_row_flip = get_dst_y_per_row_flip(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
+
+	min_dst_y_per_vm_vblank = 8.0;
+	min_dst_y_per_row_vblank = 16.0;
+
+	// magic!
+	if (htotal <= 75) {
+		min_vblank = 300;
+		min_dst_y_per_vm_vblank = 100.0;
+		min_dst_y_per_row_vblank = 100.0;
+	}
+
+	dml_print("DML_DLG: %s: dst_y_per_vm_vblank    = %3.2f\n", __func__, dst_y_per_vm_vblank);
+	dml_print("DML_DLG: %s: dst_y_per_row_vblank   = %3.2f\n", __func__, dst_y_per_row_vblank);
+
+	ASSERT(dst_y_per_vm_vblank < min_dst_y_per_vm_vblank);
+	ASSERT(dst_y_per_row_vblank < min_dst_y_per_row_vblank);
+
+	ASSERT(dst_y_prefetch > (dst_y_per_vm_vblank + dst_y_per_row_vblank));
+	lsw = dst_y_prefetch - (dst_y_per_vm_vblank + dst_y_per_row_vblank);
+
+	dml_print("DML_DLG: %s: lsw = %3.2f\n", __func__, lsw);
+
+	vratio_pre_l = get_vratio_prefetch_l(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
+	vratio_pre_c = get_vratio_prefetch_c(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
+
+	dml_print("DML_DLG: %s: vratio_pre_l=%3.2f\n", __func__, vratio_pre_l);
+	dml_print("DML_DLG: %s: vratio_pre_c=%3.2f\n", __func__, vratio_pre_c);
+
+	// Active
+	req_per_swath_ub_l = rq_dlg_param->rq_l.req_per_swath_ub;
+	req_per_swath_ub_c = rq_dlg_param->rq_c.req_per_swath_ub;
+	meta_row_height_l = rq_dlg_param->rq_l.meta_row_height;
+	meta_row_height_c = rq_dlg_param->rq_c.meta_row_height;
+	swath_width_pixels_ub_l = 0;
+	swath_width_pixels_ub_c = 0;
+	scaler_rec_in_width_l = 0;
+	scaler_rec_in_width_c = 0;
+	dpte_row_height_l = rq_dlg_param->rq_l.dpte_row_height;
+	dpte_row_height_c = rq_dlg_param->rq_c.dpte_row_height;
+
+	swath_width_pixels_ub_l = swath_width_ub_l;
+	swath_width_pixels_ub_c = swath_width_ub_c;
+
+	if (htaps_l <= 1)
+		min_hratio_fact_l = 2.0;
+	else if (htaps_l <= 6) {
+		if ((hratio_l * 2.0) > 4.0)
+			min_hratio_fact_l = 4.0;
+		else
+			min_hratio_fact_l = hratio_l * 2.0;
+	} else {
+		if (hratio_l > 4.0)
+			min_hratio_fact_l = 4.0;
+		else
+			min_hratio_fact_l = hratio_l;
+	}
+
+	hscale_pixel_rate_l = min_hratio_fact_l * dppclk_freq_in_mhz;
+
+	if (htaps_c <= 1)
+		min_hratio_fact_c = 2.0;
+	else if (htaps_c <= 6) {
+		if ((hratio_c * 2.0) > 4.0)
+			min_hratio_fact_c = 4.0;
+		else
+			min_hratio_fact_c = hratio_c * 2.0;
+	} else {
+		if (hratio_c > 4.0)
+			min_hratio_fact_c = 4.0;
+		else
+			min_hratio_fact_c = hratio_c;
+	}
+
+	hscale_pixel_rate_c = min_hratio_fact_c * dppclk_freq_in_mhz;
+
+	refcyc_per_line_delivery_pre_l = 0.;
+	refcyc_per_line_delivery_pre_c = 0.;
+	refcyc_per_line_delivery_l = 0.;
+	refcyc_per_line_delivery_c = 0.;
+
+	refcyc_per_req_delivery_pre_l = 0.;
+	refcyc_per_req_delivery_pre_c = 0.;
+	refcyc_per_req_delivery_l = 0.;
+	refcyc_per_req_delivery_c = 0.;
+
+	full_recout_width = 0;
+	// In ODM
+	if (src->is_hsplit) {
+		// This "hack"  is only allowed (and valid) for MPC combine. In ODM
+		// combine, you MUST specify the full_recout_width...according to Oswin
+		if (dst->full_recout_width == 0 && !dst->odm_combine) {
+			dml_print("DML_DLG: %s: Warning: full_recout_width not set in hsplit mode\n",
+					__func__);
+			full_recout_width = dst->recout_width * 2; // assume half split for dcn1
+		} else
+			full_recout_width = dst->full_recout_width;
+	} else
+		full_recout_width = dst->recout_width;
+
+	// As of DCN2, mpc_combine and odm_combine are mutually exclusive
+	refcyc_per_line_delivery_pre_l = get_refcyc_per_delivery(mode_lib,
+			refclk_freq_in_mhz,
+			pclk_freq_in_mhz,
+			dst->odm_combine,
+			full_recout_width,
+			dst->hactive,
+			vratio_pre_l,
+			hscale_pixel_rate_l,
+			swath_width_pixels_ub_l,
+			1); // per line
+
+	refcyc_per_line_delivery_l = get_refcyc_per_delivery(mode_lib,
+			refclk_freq_in_mhz,
+			pclk_freq_in_mhz,
+			dst->odm_combine,
+			full_recout_width,
+			dst->hactive,
+			vratio_l,
+			hscale_pixel_rate_l,
+			swath_width_pixels_ub_l,
+			1); // per line
+
+	dml_print("DML_DLG: %s: full_recout_width              = %d\n",
+			__func__,
+			full_recout_width);
+	dml_print("DML_DLG: %s: hscale_pixel_rate_l            = %3.2f\n",
+			__func__,
+			hscale_pixel_rate_l);
+	dml_print("DML_DLG: %s: refcyc_per_line_delivery_pre_l = %3.2f\n",
+			__func__,
+			refcyc_per_line_delivery_pre_l);
+	dml_print("DML_DLG: %s: refcyc_per_line_delivery_l     = %3.2f\n",
+			__func__,
+			refcyc_per_line_delivery_l);
+
+	if (dual_plane) {
+		refcyc_per_line_delivery_pre_c = get_refcyc_per_delivery(mode_lib,
+				refclk_freq_in_mhz,
+				pclk_freq_in_mhz,
+				dst->odm_combine,
+				full_recout_width,
+				dst->hactive,
+				vratio_pre_c,
+				hscale_pixel_rate_c,
+				swath_width_pixels_ub_c,
+				1); // per line
+
+		refcyc_per_line_delivery_c = get_refcyc_per_delivery(mode_lib,
+				refclk_freq_in_mhz,
+				pclk_freq_in_mhz,
+				dst->odm_combine,
+				full_recout_width,
+				dst->hactive,
+				vratio_c,
+				hscale_pixel_rate_c,
+				swath_width_pixels_ub_c,
+				1);  // per line
+
+		dml_print("DML_DLG: %s: refcyc_per_line_delivery_pre_c = %3.2f\n",
+				__func__,
+				refcyc_per_line_delivery_pre_c);
+		dml_print("DML_DLG: %s: refcyc_per_line_delivery_c     = %3.2f\n",
+				__func__,
+				refcyc_per_line_delivery_c);
+	}
+
+	// TTU - Luma / Chroma
+	if (access_dir) {  // vertical access
+		scaler_rec_in_width_l = vp_height_l;
+		scaler_rec_in_width_c = vp_height_c;
+	} else {
+		scaler_rec_in_width_l = vp_width_l;
+		scaler_rec_in_width_c = vp_width_c;
+	}
+
+	refcyc_per_req_delivery_pre_l = get_refcyc_per_delivery(mode_lib,
+			refclk_freq_in_mhz,
+			pclk_freq_in_mhz,
+			dst->odm_combine,
+			full_recout_width,
+			dst->hactive,
+			vratio_pre_l,
+			hscale_pixel_rate_l,
+			scaler_rec_in_width_l,
+			req_per_swath_ub_l);  // per req
+	refcyc_per_req_delivery_l = get_refcyc_per_delivery(mode_lib,
+			refclk_freq_in_mhz,
+			pclk_freq_in_mhz,
+			dst->odm_combine,
+			full_recout_width,
+			dst->hactive,
+			vratio_l,
+			hscale_pixel_rate_l,
+			scaler_rec_in_width_l,
+			req_per_swath_ub_l);  // per req
+
+	dml_print("DML_DLG: %s: refcyc_per_req_delivery_pre_l = %3.2f\n",
+			__func__,
+			refcyc_per_req_delivery_pre_l);
+	dml_print("DML_DLG: %s: refcyc_per_req_delivery_l     = %3.2f\n",
+			__func__,
+			refcyc_per_req_delivery_l);
+
+	ASSERT(refcyc_per_req_delivery_pre_l < dml_pow(2, 13));
+	ASSERT(refcyc_per_req_delivery_l < dml_pow(2, 13));
+
+	if (dual_plane) {
+		refcyc_per_req_delivery_pre_c = get_refcyc_per_delivery(mode_lib,
+				refclk_freq_in_mhz,
+				pclk_freq_in_mhz,
+				dst->odm_combine,
+				full_recout_width,
+				dst->hactive,
+				vratio_pre_c,
+				hscale_pixel_rate_c,
+				scaler_rec_in_width_c,
+				req_per_swath_ub_c);  // per req
+		refcyc_per_req_delivery_c = get_refcyc_per_delivery(mode_lib,
+				refclk_freq_in_mhz,
+				pclk_freq_in_mhz,
+				dst->odm_combine,
+				full_recout_width,
+				dst->hactive,
+				vratio_c,
+				hscale_pixel_rate_c,
+				scaler_rec_in_width_c,
+				req_per_swath_ub_c);  // per req
+
+		dml_print("DML_DLG: %s: refcyc_per_req_delivery_pre_c = %3.2f\n",
+				__func__,
+				refcyc_per_req_delivery_pre_c);
+		dml_print("DML_DLG: %s: refcyc_per_req_delivery_c     = %3.2f\n",
+				__func__,
+				refcyc_per_req_delivery_c);
+
+		ASSERT(refcyc_per_req_delivery_pre_c < dml_pow(2, 13));
+		ASSERT(refcyc_per_req_delivery_c < dml_pow(2, 13));
+	}
+
+	// TTU - Cursor
+	refcyc_per_req_delivery_pre_cur0 = 0.0;
+	refcyc_per_req_delivery_cur0 = 0.0;
+	if (src->num_cursors > 0) {
+		calculate_ttu_cursor(mode_lib,
+				&refcyc_per_req_delivery_pre_cur0,
+				&refcyc_per_req_delivery_cur0,
+				refclk_freq_in_mhz,
+				ref_freq_to_pix_freq,
+				hscale_pixel_rate_l,
+				scl->hscl_ratio,
+				vratio_pre_l,
+				vratio_l,
+				src->cur0_src_width,
+				(enum cursor_bpp)(src->cur0_bpp));
+	}
+
+	refcyc_per_req_delivery_pre_cur1 = 0.0;
+	refcyc_per_req_delivery_cur1 = 0.0;
+	if (src->num_cursors > 1) {
+		calculate_ttu_cursor(mode_lib,
+				&refcyc_per_req_delivery_pre_cur1,
+				&refcyc_per_req_delivery_cur1,
+				refclk_freq_in_mhz,
+				ref_freq_to_pix_freq,
+				hscale_pixel_rate_l,
+				scl->hscl_ratio,
+				vratio_pre_l,
+				vratio_l,
+				src->cur1_src_width,
+				(enum cursor_bpp)(src->cur1_bpp));
+	}
+
+	// TTU - Misc
+	// all hard-coded
+
+	// Assignment to register structures
+	disp_dlg_regs->dst_y_after_scaler = dst_y_after_scaler; // in terms of line
+	disp_dlg_regs->refcyc_x_after_scaler = dst_x_after_scaler * ref_freq_to_pix_freq; // in terms of refclk
+	ASSERT(disp_dlg_regs->refcyc_x_after_scaler < (unsigned int) dml_pow(2, 13));
+	disp_dlg_regs->dst_y_prefetch = (unsigned int) (dst_y_prefetch * dml_pow(2, 2));
+	disp_dlg_regs->dst_y_per_vm_vblank = (unsigned int) (dst_y_per_vm_vblank * dml_pow(2, 2));
+	disp_dlg_regs->dst_y_per_row_vblank = (unsigned int) (dst_y_per_row_vblank * dml_pow(2, 2));
+	disp_dlg_regs->dst_y_per_vm_flip = (unsigned int) (dst_y_per_vm_flip * dml_pow(2, 2));
+	disp_dlg_regs->dst_y_per_row_flip = (unsigned int) (dst_y_per_row_flip * dml_pow(2, 2));
+
+	disp_dlg_regs->vratio_prefetch = (unsigned int) (vratio_pre_l * dml_pow(2, 19));
+	disp_dlg_regs->vratio_prefetch_c = (unsigned int) (vratio_pre_c * dml_pow(2, 19));
+
+	disp_dlg_regs->refcyc_per_pte_group_vblank_l =
+			(unsigned int) (dst_y_per_row_vblank * (double) htotal
+					* ref_freq_to_pix_freq / (double) dpte_groups_per_row_ub_l);
+	ASSERT(disp_dlg_regs->refcyc_per_pte_group_vblank_l < (unsigned int) dml_pow(2, 13));
+
+	if (dual_plane) {
+		disp_dlg_regs->refcyc_per_pte_group_vblank_c = (unsigned int) (dst_y_per_row_vblank
+				* (double) htotal * ref_freq_to_pix_freq
+				/ (double) dpte_groups_per_row_ub_c);
+		ASSERT(disp_dlg_regs->refcyc_per_pte_group_vblank_c
+						< (unsigned int) dml_pow(2, 13));
+	}
+
+	disp_dlg_regs->refcyc_per_meta_chunk_vblank_l =
+			(unsigned int) (dst_y_per_row_vblank * (double) htotal
+					* ref_freq_to_pix_freq / (double) meta_chunks_per_row_ub_l);
+	ASSERT(disp_dlg_regs->refcyc_per_meta_chunk_vblank_l < (unsigned int) dml_pow(2, 13));
+
+	disp_dlg_regs->refcyc_per_meta_chunk_vblank_c =
+			disp_dlg_regs->refcyc_per_meta_chunk_vblank_l; // dcc for 4:2:0 is not supported in dcn1.0.  assigned to be the same as _l for now
+
+	disp_dlg_regs->refcyc_per_pte_group_flip_l = (unsigned int) (dst_y_per_row_flip * htotal
+			* ref_freq_to_pix_freq) / dpte_groups_per_row_ub_l;
+	disp_dlg_regs->refcyc_per_meta_chunk_flip_l = (unsigned int) (dst_y_per_row_flip * htotal
+			* ref_freq_to_pix_freq) / meta_chunks_per_row_ub_l;
+
+	if (dual_plane) {
+		disp_dlg_regs->refcyc_per_pte_group_flip_c = (unsigned int) (dst_y_per_row_flip
+				* htotal * ref_freq_to_pix_freq) / dpte_groups_per_row_ub_c;
+		disp_dlg_regs->refcyc_per_meta_chunk_flip_c = (unsigned int) (dst_y_per_row_flip
+				* htotal * ref_freq_to_pix_freq) / meta_chunks_per_row_ub_c;
+	}
+
+	disp_dlg_regs->dst_y_per_pte_row_nom_l = (unsigned int) ((double) dpte_row_height_l
+			/ (double) vratio_l * dml_pow(2, 2));
+	ASSERT(disp_dlg_regs->dst_y_per_pte_row_nom_l < (unsigned int) dml_pow(2, 17));
+
+	if (dual_plane) {
+		disp_dlg_regs->dst_y_per_pte_row_nom_c = (unsigned int) ((double) dpte_row_height_c
+				/ (double) vratio_c * dml_pow(2, 2));
+		if (disp_dlg_regs->dst_y_per_pte_row_nom_c >= (unsigned int) dml_pow(2, 17)) {
+			dml_print("DML_DLG: %s: Warning dst_y_per_pte_row_nom_c %u larger than supported by register format U15.2 %u\n",
+					__func__,
+					disp_dlg_regs->dst_y_per_pte_row_nom_c,
+					(unsigned int) dml_pow(2, 17) - 1);
+		}
+	}
+
+	disp_dlg_regs->dst_y_per_meta_row_nom_l = (unsigned int) ((double) meta_row_height_l
+			/ (double) vratio_l * dml_pow(2, 2));
+	ASSERT(disp_dlg_regs->dst_y_per_meta_row_nom_l < (unsigned int) dml_pow(2, 17));
+
+	disp_dlg_regs->dst_y_per_meta_row_nom_c = disp_dlg_regs->dst_y_per_meta_row_nom_l; // TODO: dcc for 4:2:0 is not supported in dcn1.0.  assigned to be the same as _l for now
+
+	disp_dlg_regs->refcyc_per_pte_group_nom_l = (unsigned int) ((double) dpte_row_height_l
+			/ (double) vratio_l * (double) htotal * ref_freq_to_pix_freq
+			/ (double) dpte_groups_per_row_ub_l);
+	if (disp_dlg_regs->refcyc_per_pte_group_nom_l >= (unsigned int) dml_pow(2, 23))
+		disp_dlg_regs->refcyc_per_pte_group_nom_l = dml_pow(2, 23) - 1;
+	disp_dlg_regs->refcyc_per_meta_chunk_nom_l = (unsigned int) ((double) meta_row_height_l
+			/ (double) vratio_l * (double) htotal * ref_freq_to_pix_freq
+			/ (double) meta_chunks_per_row_ub_l);
+	if (disp_dlg_regs->refcyc_per_meta_chunk_nom_l >= (unsigned int) dml_pow(2, 23))
+		disp_dlg_regs->refcyc_per_meta_chunk_nom_l = dml_pow(2, 23) - 1;
+
+	if (dual_plane) {
+		disp_dlg_regs->refcyc_per_pte_group_nom_c =
+				(unsigned int) ((double) dpte_row_height_c / (double) vratio_c
+						* (double) htotal * ref_freq_to_pix_freq
+						/ (double) dpte_groups_per_row_ub_c);
+		if (disp_dlg_regs->refcyc_per_pte_group_nom_c >= (unsigned int) dml_pow(2, 23))
+			disp_dlg_regs->refcyc_per_pte_group_nom_c = dml_pow(2, 23) - 1;
+
+		// TODO: Is this the right calculation? Does htotal need to be halved?
+		disp_dlg_regs->refcyc_per_meta_chunk_nom_c =
+				(unsigned int) ((double) meta_row_height_c / (double) vratio_c
+						* (double) htotal * ref_freq_to_pix_freq
+						/ (double) meta_chunks_per_row_ub_c);
+		if (disp_dlg_regs->refcyc_per_meta_chunk_nom_c >= (unsigned int) dml_pow(2, 23))
+			disp_dlg_regs->refcyc_per_meta_chunk_nom_c = dml_pow(2, 23) - 1;
+	}
+
+	disp_dlg_regs->refcyc_per_line_delivery_pre_l = (unsigned int) dml_floor(refcyc_per_line_delivery_pre_l,
+			1);
+	disp_dlg_regs->refcyc_per_line_delivery_l = (unsigned int) dml_floor(refcyc_per_line_delivery_l,
+			1);
+	ASSERT(disp_dlg_regs->refcyc_per_line_delivery_pre_l < (unsigned int) dml_pow(2, 13));
+	ASSERT(disp_dlg_regs->refcyc_per_line_delivery_l < (unsigned int) dml_pow(2, 13));
+
+	disp_dlg_regs->refcyc_per_line_delivery_pre_c = (unsigned int) dml_floor(refcyc_per_line_delivery_pre_c,
+			1);
+	disp_dlg_regs->refcyc_per_line_delivery_c = (unsigned int) dml_floor(refcyc_per_line_delivery_c,
+			1);
+	ASSERT(disp_dlg_regs->refcyc_per_line_delivery_pre_c < (unsigned int) dml_pow(2, 13));
+	ASSERT(disp_dlg_regs->refcyc_per_line_delivery_c < (unsigned int) dml_pow(2, 13));
+
+	disp_dlg_regs->chunk_hdl_adjust_cur0 = 3;
+	disp_dlg_regs->dst_y_offset_cur0 = 0;
+	disp_dlg_regs->chunk_hdl_adjust_cur1 = 3;
+	disp_dlg_regs->dst_y_offset_cur1 = 0;
+
+	disp_dlg_regs->dst_y_delta_drq_limit = 0x7fff; // off
+
+	disp_ttu_regs->refcyc_per_req_delivery_pre_l = (unsigned int) (refcyc_per_req_delivery_pre_l
+			* dml_pow(2, 10));
+	disp_ttu_regs->refcyc_per_req_delivery_l = (unsigned int) (refcyc_per_req_delivery_l
+			* dml_pow(2, 10));
+	disp_ttu_regs->refcyc_per_req_delivery_pre_c = (unsigned int) (refcyc_per_req_delivery_pre_c
+			* dml_pow(2, 10));
+	disp_ttu_regs->refcyc_per_req_delivery_c = (unsigned int) (refcyc_per_req_delivery_c
+			* dml_pow(2, 10));
+	disp_ttu_regs->refcyc_per_req_delivery_pre_cur0 =
+			(unsigned int) (refcyc_per_req_delivery_pre_cur0 * dml_pow(2, 10));
+	disp_ttu_regs->refcyc_per_req_delivery_cur0 = (unsigned int) (refcyc_per_req_delivery_cur0
+			* dml_pow(2, 10));
+	disp_ttu_regs->refcyc_per_req_delivery_pre_cur1 =
+			(unsigned int) (refcyc_per_req_delivery_pre_cur1 * dml_pow(2, 10));
+	disp_ttu_regs->refcyc_per_req_delivery_cur1 = (unsigned int) (refcyc_per_req_delivery_cur1
+			* dml_pow(2, 10));
+	disp_ttu_regs->qos_level_low_wm = 0;
+	ASSERT(disp_ttu_regs->qos_level_low_wm < dml_pow(2, 14));
+	disp_ttu_regs->qos_level_high_wm = (unsigned int) (4.0 * (double) htotal
+			* ref_freq_to_pix_freq);
+	/*ASSERT(disp_ttu_regs->qos_level_high_wm < dml_pow(2, 14));*/
+
+	disp_ttu_regs->qos_level_flip = 14;
+	disp_ttu_regs->qos_level_fixed_l = 8;
+	disp_ttu_regs->qos_level_fixed_c = 8;
+	disp_ttu_regs->qos_level_fixed_cur0 = 8;
+	disp_ttu_regs->qos_ramp_disable_l = 0;
+	disp_ttu_regs->qos_ramp_disable_c = 0;
+	disp_ttu_regs->qos_ramp_disable_cur0 = 0;
+
+	disp_ttu_regs->min_ttu_vblank = min_ttu_vblank * refclk_freq_in_mhz;
+	ASSERT(disp_ttu_regs->min_ttu_vblank < dml_pow(2, 24));
+
+	print__ttu_regs_st(mode_lib, disp_ttu_regs);
+	print__dlg_regs_st(mode_lib, disp_dlg_regs);
+}
+
+void dml20_rq_dlg_get_dlg_reg(struct display_mode_lib *mode_lib,
+		display_dlg_regs_st *dlg_regs,
+		display_ttu_regs_st *ttu_regs,
+		const display_e2e_pipe_params_st *e2e_pipe_param,
+		const unsigned int num_pipes,
+		const unsigned int pipe_idx,
+		const bool cstate_en,
+		const bool pstate_en,
+		const bool vm_en,
+		const bool ignore_viewport_pos,
+		const bool immediate_flip_support)
+{
+	display_rq_params_st rq_param = {0};
+	display_dlg_sys_params_st dlg_sys_param = {0};
+
+	// Get watermark and Tex.
+	dlg_sys_param.t_urg_wm_us = get_wm_urgent(mode_lib, e2e_pipe_param, num_pipes);
+	dlg_sys_param.deepsleep_dcfclk_mhz = get_clk_dcf_deepsleep(mode_lib,
+			e2e_pipe_param,
+			num_pipes);
+	dlg_sys_param.t_extra_us = get_urgent_extra_latency(mode_lib, e2e_pipe_param, num_pipes);
+	dlg_sys_param.mem_trip_us = get_wm_memory_trip(mode_lib, e2e_pipe_param, num_pipes);
+	dlg_sys_param.t_mclk_wm_us = get_wm_dram_clock_change(mode_lib, e2e_pipe_param, num_pipes);
+	dlg_sys_param.t_sr_wm_us = get_wm_stutter_enter_exit(mode_lib, e2e_pipe_param, num_pipes);
+	dlg_sys_param.total_flip_bw = get_total_immediate_flip_bw(mode_lib,
+			e2e_pipe_param,
+			num_pipes);
+	dlg_sys_param.total_flip_bytes = get_total_immediate_flip_bytes(mode_lib,
+			e2e_pipe_param,
+			num_pipes);
+
+	print__dlg_sys_params_st(mode_lib, &dlg_sys_param);
+
+	// system parameter calculation done
+
+	dml_print("DML_DLG: Calculation for pipe[%d] start\n\n", pipe_idx);
+	dml20_rq_dlg_get_rq_params(mode_lib, &rq_param, &e2e_pipe_param[pipe_idx].pipe.src);
+	dml20_rq_dlg_get_dlg_params(mode_lib,
+			e2e_pipe_param,
+			num_pipes,
+			pipe_idx,
+			dlg_regs,
+			ttu_regs,
+			&rq_param.dlg,
+			&dlg_sys_param,
+			cstate_en,
+			pstate_en);
+	dml_print("DML_DLG: Calculation for pipe[%d] end\n", pipe_idx);
+}
+
+static void calculate_ttu_cursor(struct display_mode_lib *mode_lib,
+		double *refcyc_per_req_delivery_pre_cur,
+		double *refcyc_per_req_delivery_cur,
+		double refclk_freq_in_mhz,
+		double ref_freq_to_pix_freq,
+		double hscale_pixel_rate_l,
+		double hscl_ratio,
+		double vratio_pre_l,
+		double vratio_l,
+		unsigned int cur_width,
+		enum cursor_bpp cur_bpp)
+{
+	unsigned int cur_src_width = cur_width;
+	unsigned int cur_req_size = 0;
+	unsigned int cur_req_width = 0;
+	double cur_width_ub = 0.0;
+	double cur_req_per_width = 0.0;
+	double hactive_cur = 0.0;
+
+	ASSERT(cur_src_width <= 256);
+
+	*refcyc_per_req_delivery_pre_cur = 0.0;
+	*refcyc_per_req_delivery_cur = 0.0;
+	if (cur_src_width > 0) {
+		unsigned int cur_bit_per_pixel = 0;
+
+		if (cur_bpp == dm_cur_2bit) {
+			cur_req_size = 64; // byte
+			cur_bit_per_pixel = 2;
+		} else { // 32bit
+			cur_bit_per_pixel = 32;
+			if (cur_src_width >= 1 && cur_src_width <= 16)
+				cur_req_size = 64;
+			else if (cur_src_width >= 17 && cur_src_width <= 31)
+				cur_req_size = 128;
+			else
+				cur_req_size = 256;
+		}
+
+		cur_req_width = (double) cur_req_size / ((double) cur_bit_per_pixel / 8.0);
+		cur_width_ub = dml_ceil((double) cur_src_width / (double) cur_req_width, 1)
+				* (double) cur_req_width;
+		cur_req_per_width = cur_width_ub / (double) cur_req_width;
+		hactive_cur = (double) cur_src_width / hscl_ratio; // TODO: oswin to think about what to do for cursor
+
+		if (vratio_pre_l <= 1.0) {
+			*refcyc_per_req_delivery_pre_cur = hactive_cur * ref_freq_to_pix_freq
+					/ (double) cur_req_per_width;
+		} else {
+			*refcyc_per_req_delivery_pre_cur = (double) refclk_freq_in_mhz
+					* (double) cur_src_width / hscale_pixel_rate_l
+					/ (double) cur_req_per_width;
+		}
+
+		ASSERT(*refcyc_per_req_delivery_pre_cur < dml_pow(2, 13));
+
+		if (vratio_l <= 1.0) {
+			*refcyc_per_req_delivery_cur = hactive_cur * ref_freq_to_pix_freq
+					/ (double) cur_req_per_width;
+		} else {
+			*refcyc_per_req_delivery_cur = (double) refclk_freq_in_mhz
+					* (double) cur_src_width / hscale_pixel_rate_l
+					/ (double) cur_req_per_width;
+		}
+
+		dml_print("DML_DLG: %s: cur_req_width                     = %d\n",
+				__func__,
+				cur_req_width);
+		dml_print("DML_DLG: %s: cur_width_ub                      = %3.2f\n",
+				__func__,
+				cur_width_ub);
+		dml_print("DML_DLG: %s: cur_req_per_width                 = %3.2f\n",
+				__func__,
+				cur_req_per_width);
+		dml_print("DML_DLG: %s: hactive_cur                       = %3.2f\n",
+				__func__,
+				hactive_cur);
+		dml_print("DML_DLG: %s: refcyc_per_req_delivery_pre_cur   = %3.2f\n",
+				__func__,
+				*refcyc_per_req_delivery_pre_cur);
+		dml_print("DML_DLG: %s: refcyc_per_req_delivery_cur       = %3.2f\n",
+				__func__,
+				*refcyc_per_req_delivery_cur);
+
+		ASSERT(*refcyc_per_req_delivery_cur < dml_pow(2, 13));
+	}
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_rq_dlg_calc_20.h b/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_rq_dlg_calc_20.h
new file mode 100644
index 000000000000..8b23867e97c1
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_rq_dlg_calc_20.h
@@ -0,0 +1,73 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DML20_DISPLAY_RQ_DLG_CALC_H__
+#define __DML20_DISPLAY_RQ_DLG_CALC_H__
+
+#include "../display_rq_dlg_helpers.h"
+
+struct display_mode_lib;
+
+
+// Function: dml_rq_dlg_get_rq_reg
+//  Main entry point for test to get the register values out of this DML class.
+//  This function calls <get_rq_param> and <extract_rq_regs> functions to calculate
+//  and then populate the rq_regs struct
+// Input:
+//  pipe_src_param - pipe source configuration (e.g. vp, pitch, etc.)
+// Output:
+//  rq_regs - struct that holds all the RQ registers field value.
+//            See also: <display_rq_regs_st>
+void dml20_rq_dlg_get_rq_reg(
+		struct display_mode_lib *mode_lib,
+		display_rq_regs_st *rq_regs,
+		const display_pipe_params_st *pipe_param);
+
+
+// Function: dml_rq_dlg_get_dlg_reg
+//   Calculate and return DLG and TTU register struct given the system setting
+// Output:
+//  dlg_regs - output DLG register struct
+//  ttu_regs - output DLG TTU register struct
+// Input:
+//  e2e_pipe_param - "compacted" array of e2e pipe param struct
+//  num_pipes - num of active "pipe" or "route"
+//  pipe_idx - index that identifies the e2e_pipe_param that corresponding to this dlg
+//  cstate - 0: when calculate min_ttu_vblank it is assumed cstate is not required. 1: Normal mode, cstate is considered.
+//           Added for legacy or unrealistic timing tests.
+void dml20_rq_dlg_get_dlg_reg(
+		struct display_mode_lib *mode_lib,
+		display_dlg_regs_st *dlg_regs,
+		display_ttu_regs_st *ttu_regs,
+		const display_e2e_pipe_params_st *e2e_pipe_param,
+		const unsigned int num_pipes,
+		const unsigned int pipe_idx,
+		const bool cstate_en,
+		const bool pstate_en,
+		const bool vm_en,
+		const bool ignore_viewport_pos,
+		const bool immediate_flip_support);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_rq_dlg_calc_20v2.c b/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_rq_dlg_calc_20v2.c
new file mode 100644
index 000000000000..570e6e39eb45
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_rq_dlg_calc_20v2.c
@@ -0,0 +1,1666 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "../display_mode_lib.h"
+#include "../display_mode_vba.h"
+#include "display_rq_dlg_calc_20v2.h"
+
+// Function: dml20v2_rq_dlg_get_rq_params
+//  Calculate requestor related parameters that register definition agnostic
+//  (i.e. this layer does try to separate real values from register definition)
+// Input:
+//  pipe_src_param - pipe source configuration (e.g. vp, pitch, etc.)
+// Output:
+//  rq_param - values that can be used to setup RQ (e.g. swath_height, plane1_addr, etc.)
+//
+static void dml20v2_rq_dlg_get_rq_params(
+		struct display_mode_lib *mode_lib,
+		display_rq_params_st * rq_param,
+		const display_pipe_source_params_st *pipe_src_param);
+
+// Function: dml20v2_rq_dlg_get_dlg_params
+//  Calculate deadline related parameters
+//
+static void dml20v2_rq_dlg_get_dlg_params(struct display_mode_lib *mode_lib,
+		const display_e2e_pipe_params_st *e2e_pipe_param,
+		const unsigned int num_pipes,
+		const unsigned int pipe_idx,
+		display_dlg_regs_st *disp_dlg_regs,
+		display_ttu_regs_st *disp_ttu_regs,
+		const display_rq_dlg_params_st *rq_dlg_param,
+		const display_dlg_sys_params_st *dlg_sys_param,
+		const bool cstate_en,
+		const bool pstate_en);
+/*
+ * NOTE:
+ *   This file is gcc-parseable HW gospel, coming straight from HW engineers.
+ *
+ * It doesn't adhere to Linux kernel style and sometimes will do things in odd
+ * ways. Unless there is something clearly wrong with it the code should
+ * remain as-is as it provides us with a guarantee from HW that it is correct.
+ */
+
+static void calculate_ttu_cursor(struct display_mode_lib *mode_lib,
+		double *refcyc_per_req_delivery_pre_cur,
+		double *refcyc_per_req_delivery_cur,
+		double refclk_freq_in_mhz,
+		double ref_freq_to_pix_freq,
+		double hscale_pixel_rate_l,
+		double hscl_ratio,
+		double vratio_pre_l,
+		double vratio_l,
+		unsigned int cur_width,
+		enum cursor_bpp cur_bpp);
+
+#include "../dml_inline_defs.h"
+
+static unsigned int get_bytes_per_element(enum source_format_class source_format, bool is_chroma)
+{
+	unsigned int ret_val = 1;
+
+	if (source_format == dm_444_16) {
+		if (!is_chroma)
+			ret_val = 2;
+	} else if (source_format == dm_444_32) {
+		if (!is_chroma)
+			ret_val = 4;
+	} else if (source_format == dm_444_64) {
+		if (!is_chroma)
+			ret_val = 8;
+	} else if (source_format == dm_420_8) {
+		if (is_chroma)
+			ret_val = 2;
+		else
+			ret_val = 1;
+	} else if (source_format == dm_420_10) {
+		if (is_chroma)
+			ret_val = 4;
+		else
+			ret_val = 2;
+	} else if (source_format == dm_444_8) {
+		ret_val = 1;
+	}
+	return ret_val;
+}
+
+static bool is_dual_plane(enum source_format_class source_format)
+{
+	bool ret_val = false;
+
+	if ((source_format == dm_420_8) || (source_format == dm_420_10))
+		ret_val = true;
+
+	return ret_val;
+}
+
+static double get_refcyc_per_delivery(struct display_mode_lib *mode_lib,
+		double refclk_freq_in_mhz,
+		double pclk_freq_in_mhz,
+		bool odm_combine,
+		unsigned int recout_width,
+		unsigned int hactive,
+		double vratio,
+		double hscale_pixel_rate,
+		unsigned int delivery_width,
+		unsigned int req_per_swath_ub)
+{
+	double refcyc_per_delivery = 0.0;
+
+	if (vratio <= 1.0) {
+		if (odm_combine)
+			refcyc_per_delivery = (double) refclk_freq_in_mhz
+					* dml_min((double) recout_width, (double) hactive / 2.0)
+					/ pclk_freq_in_mhz / (double) req_per_swath_ub;
+		else
+			refcyc_per_delivery = (double) refclk_freq_in_mhz * (double) recout_width
+					/ pclk_freq_in_mhz / (double) req_per_swath_ub;
+	} else {
+		refcyc_per_delivery = (double) refclk_freq_in_mhz * (double) delivery_width
+				/ (double) hscale_pixel_rate / (double) req_per_swath_ub;
+	}
+
+	dml_print("DML_DLG: %s: refclk_freq_in_mhz = %3.2f\n", __func__, refclk_freq_in_mhz);
+	dml_print("DML_DLG: %s: pclk_freq_in_mhz   = %3.2f\n", __func__, pclk_freq_in_mhz);
+	dml_print("DML_DLG: %s: recout_width       = %d\n", __func__, recout_width);
+	dml_print("DML_DLG: %s: vratio             = %3.2f\n", __func__, vratio);
+	dml_print("DML_DLG: %s: req_per_swath_ub   = %d\n", __func__, req_per_swath_ub);
+	dml_print("DML_DLG: %s: refcyc_per_delivery= %3.2f\n", __func__, refcyc_per_delivery);
+
+	return refcyc_per_delivery;
+
+}
+
+static unsigned int get_blk_size_bytes(const enum source_macro_tile_size tile_size)
+{
+	if (tile_size == dm_256k_tile)
+		return (256 * 1024);
+	else if (tile_size == dm_64k_tile)
+		return (64 * 1024);
+	else
+		return (4 * 1024);
+}
+
+static void extract_rq_sizing_regs(struct display_mode_lib *mode_lib,
+		display_data_rq_regs_st *rq_regs,
+		const display_data_rq_sizing_params_st *rq_sizing)
+{
+	dml_print("DML_DLG: %s: rq_sizing param\n", __func__);
+	print__data_rq_sizing_params_st(mode_lib, rq_sizing);
+
+	rq_regs->chunk_size = dml_log2(rq_sizing->chunk_bytes) - 10;
+
+	if (rq_sizing->min_chunk_bytes == 0)
+		rq_regs->min_chunk_size = 0;
+	else
+		rq_regs->min_chunk_size = dml_log2(rq_sizing->min_chunk_bytes) - 8 + 1;
+
+	rq_regs->meta_chunk_size = dml_log2(rq_sizing->meta_chunk_bytes) - 10;
+	if (rq_sizing->min_meta_chunk_bytes == 0)
+		rq_regs->min_meta_chunk_size = 0;
+	else
+		rq_regs->min_meta_chunk_size = dml_log2(rq_sizing->min_meta_chunk_bytes) - 6 + 1;
+
+	rq_regs->dpte_group_size = dml_log2(rq_sizing->dpte_group_bytes) - 6;
+	rq_regs->mpte_group_size = dml_log2(rq_sizing->mpte_group_bytes) - 6;
+}
+
+static void extract_rq_regs(struct display_mode_lib *mode_lib,
+		display_rq_regs_st *rq_regs,
+		const display_rq_params_st *rq_param)
+{
+	unsigned int detile_buf_size_in_bytes = mode_lib->ip.det_buffer_size_kbytes * 1024;
+	unsigned int detile_buf_plane1_addr = 0;
+
+	extract_rq_sizing_regs(mode_lib, &(rq_regs->rq_regs_l), &rq_param->sizing.rq_l);
+
+	rq_regs->rq_regs_l.pte_row_height_linear = dml_floor(dml_log2(rq_param->dlg.rq_l.dpte_row_height),
+			1) - 3;
+
+	if (rq_param->yuv420) {
+		extract_rq_sizing_regs(mode_lib, &(rq_regs->rq_regs_c), &rq_param->sizing.rq_c);
+		rq_regs->rq_regs_c.pte_row_height_linear = dml_floor(dml_log2(rq_param->dlg.rq_c.dpte_row_height),
+				1) - 3;
+	}
+
+	rq_regs->rq_regs_l.swath_height = dml_log2(rq_param->dlg.rq_l.swath_height);
+	rq_regs->rq_regs_c.swath_height = dml_log2(rq_param->dlg.rq_c.swath_height);
+
+	// TODO: take the max between luma, chroma chunk size?
+	// okay for now, as we are setting chunk_bytes to 8kb anyways
+	if (rq_param->sizing.rq_l.chunk_bytes >= 32 * 1024) { //32kb
+		rq_regs->drq_expansion_mode = 0;
+	} else {
+		rq_regs->drq_expansion_mode = 2;
+	}
+	rq_regs->prq_expansion_mode = 1;
+	rq_regs->mrq_expansion_mode = 1;
+	rq_regs->crq_expansion_mode = 1;
+
+	if (rq_param->yuv420) {
+		if ((double) rq_param->misc.rq_l.stored_swath_bytes
+				/ (double) rq_param->misc.rq_c.stored_swath_bytes <= 1.5) {
+			detile_buf_plane1_addr = (detile_buf_size_in_bytes / 2.0 / 64.0); // half to chroma
+		} else {
+			detile_buf_plane1_addr = dml_round_to_multiple((unsigned int) ((2.0 * detile_buf_size_in_bytes) / 3.0),
+					256,
+					0) / 64.0; // 2/3 to chroma
+		}
+	}
+	rq_regs->plane1_base_address = detile_buf_plane1_addr;
+}
+
+static void handle_det_buf_split(struct display_mode_lib *mode_lib,
+		display_rq_params_st *rq_param,
+		const display_pipe_source_params_st *pipe_src_param)
+{
+	unsigned int total_swath_bytes = 0;
+	unsigned int swath_bytes_l = 0;
+	unsigned int swath_bytes_c = 0;
+	unsigned int full_swath_bytes_packed_l = 0;
+	unsigned int full_swath_bytes_packed_c = 0;
+	bool req128_l = false;
+	bool req128_c = false;
+	bool surf_linear = (pipe_src_param->sw_mode == dm_sw_linear);
+	bool surf_vert = (pipe_src_param->source_scan == dm_vert);
+	unsigned int log2_swath_height_l = 0;
+	unsigned int log2_swath_height_c = 0;
+	unsigned int detile_buf_size_in_bytes = mode_lib->ip.det_buffer_size_kbytes * 1024;
+
+	full_swath_bytes_packed_l = rq_param->misc.rq_l.full_swath_bytes;
+	full_swath_bytes_packed_c = rq_param->misc.rq_c.full_swath_bytes;
+
+	if (rq_param->yuv420_10bpc) {
+		full_swath_bytes_packed_l = dml_round_to_multiple(rq_param->misc.rq_l.full_swath_bytes * 2 / 3,
+				256,
+				1) + 256;
+		full_swath_bytes_packed_c = dml_round_to_multiple(rq_param->misc.rq_c.full_swath_bytes * 2 / 3,
+				256,
+				1) + 256;
+	}
+
+	if (rq_param->yuv420) {
+		total_swath_bytes = 2 * full_swath_bytes_packed_l + 2 * full_swath_bytes_packed_c;
+
+		if (total_swath_bytes <= detile_buf_size_in_bytes) { //full 256b request
+			req128_l = false;
+			req128_c = false;
+			swath_bytes_l = full_swath_bytes_packed_l;
+			swath_bytes_c = full_swath_bytes_packed_c;
+		} else { //128b request (for luma only for yuv420 8bpc)
+			req128_l = true;
+			req128_c = false;
+			swath_bytes_l = full_swath_bytes_packed_l / 2;
+			swath_bytes_c = full_swath_bytes_packed_c;
+		}
+		// Note: assumption, the config that pass in will fit into
+		//       the detiled buffer.
+	} else {
+		total_swath_bytes = 2 * full_swath_bytes_packed_l;
+
+		if (total_swath_bytes <= detile_buf_size_in_bytes)
+			req128_l = false;
+		else
+			req128_l = true;
+
+		swath_bytes_l = total_swath_bytes;
+		swath_bytes_c = 0;
+	}
+	rq_param->misc.rq_l.stored_swath_bytes = swath_bytes_l;
+	rq_param->misc.rq_c.stored_swath_bytes = swath_bytes_c;
+
+	if (surf_linear) {
+		log2_swath_height_l = 0;
+		log2_swath_height_c = 0;
+	} else {
+		unsigned int swath_height_l;
+		unsigned int swath_height_c;
+
+		if (!surf_vert) {
+			swath_height_l = rq_param->misc.rq_l.blk256_height;
+			swath_height_c = rq_param->misc.rq_c.blk256_height;
+		} else {
+			swath_height_l = rq_param->misc.rq_l.blk256_width;
+			swath_height_c = rq_param->misc.rq_c.blk256_width;
+		}
+
+		if (swath_height_l > 0)
+			log2_swath_height_l = dml_log2(swath_height_l);
+
+		if (req128_l && log2_swath_height_l > 0)
+			log2_swath_height_l -= 1;
+
+		if (swath_height_c > 0)
+			log2_swath_height_c = dml_log2(swath_height_c);
+	}
+
+	rq_param->dlg.rq_l.swath_height = 1 << log2_swath_height_l;
+	rq_param->dlg.rq_c.swath_height = 1 << log2_swath_height_c;
+
+	dml_print("DML_DLG: %s: req128_l = %0d\n", __func__, req128_l);
+	dml_print("DML_DLG: %s: req128_c = %0d\n", __func__, req128_c);
+	dml_print("DML_DLG: %s: full_swath_bytes_packed_l = %0d\n",
+			__func__,
+			full_swath_bytes_packed_l);
+	dml_print("DML_DLG: %s: full_swath_bytes_packed_c = %0d\n",
+			__func__,
+			full_swath_bytes_packed_c);
+}
+
+static void get_meta_and_pte_attr(struct display_mode_lib *mode_lib,
+		display_data_rq_dlg_params_st *rq_dlg_param,
+		display_data_rq_misc_params_st *rq_misc_param,
+		display_data_rq_sizing_params_st *rq_sizing_param,
+		unsigned int vp_width,
+		unsigned int vp_height,
+		unsigned int data_pitch,
+		unsigned int meta_pitch,
+		unsigned int source_format,
+		unsigned int tiling,
+		unsigned int macro_tile_size,
+		unsigned int source_scan,
+		unsigned int is_chroma)
+{
+	bool surf_linear = (tiling == dm_sw_linear);
+	bool surf_vert = (source_scan == dm_vert);
+
+	unsigned int bytes_per_element;
+	unsigned int bytes_per_element_y = get_bytes_per_element((enum source_format_class)(source_format),
+			false);
+	unsigned int bytes_per_element_c = get_bytes_per_element((enum source_format_class)(source_format),
+			true);
+
+	unsigned int blk256_width = 0;
+	unsigned int blk256_height = 0;
+
+	unsigned int blk256_width_y = 0;
+	unsigned int blk256_height_y = 0;
+	unsigned int blk256_width_c = 0;
+	unsigned int blk256_height_c = 0;
+	unsigned int log2_bytes_per_element;
+	unsigned int log2_blk256_width;
+	unsigned int log2_blk256_height;
+	unsigned int blk_bytes;
+	unsigned int log2_blk_bytes;
+	unsigned int log2_blk_height;
+	unsigned int log2_blk_width;
+	unsigned int log2_meta_req_bytes;
+	unsigned int log2_meta_req_height;
+	unsigned int log2_meta_req_width;
+	unsigned int meta_req_width;
+	unsigned int meta_req_height;
+	unsigned int log2_meta_row_height;
+	unsigned int meta_row_width_ub;
+	unsigned int log2_meta_chunk_bytes;
+	unsigned int log2_meta_chunk_height;
+
+	//full sized meta chunk width in unit of data elements
+	unsigned int log2_meta_chunk_width;
+	unsigned int log2_min_meta_chunk_bytes;
+	unsigned int min_meta_chunk_width;
+	unsigned int meta_chunk_width;
+	unsigned int meta_chunk_per_row_int;
+	unsigned int meta_row_remainder;
+	unsigned int meta_chunk_threshold;
+	unsigned int meta_blk_bytes;
+	unsigned int meta_blk_height;
+	unsigned int meta_blk_width;
+	unsigned int meta_surface_bytes;
+	unsigned int vmpg_bytes;
+	unsigned int meta_pte_req_per_frame_ub;
+	unsigned int meta_pte_bytes_per_frame_ub;
+	const unsigned int log2_vmpg_bytes = dml_log2(mode_lib->soc.vmm_page_size_bytes);
+	const unsigned int dpte_buf_in_pte_reqs = mode_lib->ip.dpte_buffer_size_in_pte_reqs_luma;
+	const unsigned int pde_proc_buffer_size_64k_reqs =
+			mode_lib->ip.pde_proc_buffer_size_64k_reqs;
+
+	unsigned int log2_vmpg_height = 0;
+	unsigned int log2_vmpg_width = 0;
+	unsigned int log2_dpte_req_height_ptes = 0;
+	unsigned int log2_dpte_req_height = 0;
+	unsigned int log2_dpte_req_width = 0;
+	unsigned int log2_dpte_row_height_linear = 0;
+	unsigned int log2_dpte_row_height = 0;
+	unsigned int log2_dpte_group_width = 0;
+	unsigned int dpte_row_width_ub = 0;
+	unsigned int dpte_req_height = 0;
+	unsigned int dpte_req_width = 0;
+	unsigned int dpte_group_width = 0;
+	unsigned int log2_dpte_group_bytes = 0;
+	unsigned int log2_dpte_group_length = 0;
+	unsigned int pde_buf_entries;
+	bool yuv420 = (source_format == dm_420_8 || source_format == dm_420_10);
+
+	Calculate256BBlockSizes((enum source_format_class)(source_format),
+			(enum dm_swizzle_mode)(tiling),
+			bytes_per_element_y,
+			bytes_per_element_c,
+			&blk256_height_y,
+			&blk256_height_c,
+			&blk256_width_y,
+			&blk256_width_c);
+
+	if (!is_chroma) {
+		blk256_width = blk256_width_y;
+		blk256_height = blk256_height_y;
+		bytes_per_element = bytes_per_element_y;
+	} else {
+		blk256_width = blk256_width_c;
+		blk256_height = blk256_height_c;
+		bytes_per_element = bytes_per_element_c;
+	}
+
+	log2_bytes_per_element = dml_log2(bytes_per_element);
+
+	dml_print("DML_DLG: %s: surf_linear        = %d\n", __func__, surf_linear);
+	dml_print("DML_DLG: %s: surf_vert          = %d\n", __func__, surf_vert);
+	dml_print("DML_DLG: %s: blk256_width       = %d\n", __func__, blk256_width);
+	dml_print("DML_DLG: %s: blk256_height      = %d\n", __func__, blk256_height);
+
+	log2_blk256_width = dml_log2((double) blk256_width);
+	log2_blk256_height = dml_log2((double) blk256_height);
+	blk_bytes = surf_linear ?
+			256 : get_blk_size_bytes((enum source_macro_tile_size) macro_tile_size);
+	log2_blk_bytes = dml_log2((double) blk_bytes);
+
+	// remember log rule
+	// "+" in log is multiply
+	// "-" in log is divide
+	// "/2" is like square root
+	// blk is vertical biased
+	if (tiling != dm_sw_linear)
+		log2_blk_height = log2_blk256_height
+				+ dml_ceil((double) (log2_blk_bytes - 8) / 2.0, 1);
+	else
+		log2_blk_height = 0;  // blk height of 1
+
+	log2_blk_width = log2_blk_bytes - log2_bytes_per_element - log2_blk_height;
+
+	if (!surf_vert) {
+		rq_dlg_param->swath_width_ub = dml_round_to_multiple(vp_width - 1, blk256_width, 1)
+				+ blk256_width;
+		rq_dlg_param->req_per_swath_ub = rq_dlg_param->swath_width_ub >> log2_blk256_width;
+	} else {
+		rq_dlg_param->swath_width_ub = dml_round_to_multiple(vp_height - 1, blk256_height, 1)
+				+ blk256_height;
+		rq_dlg_param->req_per_swath_ub = rq_dlg_param->swath_width_ub >> log2_blk256_height;
+	}
+
+	if (!surf_vert)
+		rq_misc_param->full_swath_bytes = rq_dlg_param->swath_width_ub * blk256_height
+				* bytes_per_element;
+	else
+		rq_misc_param->full_swath_bytes = rq_dlg_param->swath_width_ub * blk256_width
+				* bytes_per_element;
+
+	rq_misc_param->blk256_height = blk256_height;
+	rq_misc_param->blk256_width = blk256_width;
+
+	// -------
+	// meta
+	// -------
+	log2_meta_req_bytes = 6; // meta request is 64b and is 8x8byte meta element
+
+	// each 64b meta request for dcn is 8x8 meta elements and
+	// a meta element covers one 256b block of the data surface.
+	log2_meta_req_height = log2_blk256_height + 3; // meta req is 8x8 byte, each byte represent 1 blk256
+	log2_meta_req_width = log2_meta_req_bytes + 8 - log2_bytes_per_element
+			- log2_meta_req_height;
+	meta_req_width = 1 << log2_meta_req_width;
+	meta_req_height = 1 << log2_meta_req_height;
+
+	// the dimensions of a meta row are meta_row_width x meta_row_height in elements.
+	// calculate upper bound of the meta_row_width
+	if (!surf_vert) {
+		log2_meta_row_height = log2_meta_req_height;
+		meta_row_width_ub = dml_round_to_multiple(vp_width - 1, meta_req_width, 1)
+				+ meta_req_width;
+		rq_dlg_param->meta_req_per_row_ub = meta_row_width_ub / meta_req_width;
+	} else {
+		log2_meta_row_height = log2_meta_req_width;
+		meta_row_width_ub = dml_round_to_multiple(vp_height - 1, meta_req_height, 1)
+				+ meta_req_height;
+		rq_dlg_param->meta_req_per_row_ub = meta_row_width_ub / meta_req_height;
+	}
+	rq_dlg_param->meta_bytes_per_row_ub = rq_dlg_param->meta_req_per_row_ub * 64;
+
+	rq_dlg_param->meta_row_height = 1 << log2_meta_row_height;
+
+	log2_meta_chunk_bytes = dml_log2(rq_sizing_param->meta_chunk_bytes);
+	log2_meta_chunk_height = log2_meta_row_height;
+
+	//full sized meta chunk width in unit of data elements
+	log2_meta_chunk_width = log2_meta_chunk_bytes + 8 - log2_bytes_per_element
+			- log2_meta_chunk_height;
+	log2_min_meta_chunk_bytes = dml_log2(rq_sizing_param->min_meta_chunk_bytes);
+	min_meta_chunk_width = 1
+			<< (log2_min_meta_chunk_bytes + 8 - log2_bytes_per_element
+					- log2_meta_chunk_height);
+	meta_chunk_width = 1 << log2_meta_chunk_width;
+	meta_chunk_per_row_int = (unsigned int) (meta_row_width_ub / meta_chunk_width);
+	meta_row_remainder = meta_row_width_ub % meta_chunk_width;
+	meta_chunk_threshold = 0;
+	meta_blk_bytes = 4096;
+	meta_blk_height = blk256_height * 64;
+	meta_blk_width = meta_blk_bytes * 256 / bytes_per_element / meta_blk_height;
+	meta_surface_bytes = meta_pitch
+			* (dml_round_to_multiple(vp_height - 1, meta_blk_height, 1) + meta_blk_height)
+			* bytes_per_element / 256;
+	vmpg_bytes = mode_lib->soc.vmm_page_size_bytes;
+	meta_pte_req_per_frame_ub = (dml_round_to_multiple(meta_surface_bytes - vmpg_bytes,
+			8 * vmpg_bytes,
+			1) + 8 * vmpg_bytes) / (8 * vmpg_bytes);
+	meta_pte_bytes_per_frame_ub = meta_pte_req_per_frame_ub * 64; //64B mpte request
+	rq_dlg_param->meta_pte_bytes_per_frame_ub = meta_pte_bytes_per_frame_ub;
+
+	dml_print("DML_DLG: %s: meta_blk_height             = %d\n", __func__, meta_blk_height);
+	dml_print("DML_DLG: %s: meta_blk_width              = %d\n", __func__, meta_blk_width);
+	dml_print("DML_DLG: %s: meta_surface_bytes          = %d\n", __func__, meta_surface_bytes);
+	dml_print("DML_DLG: %s: meta_pte_req_per_frame_ub   = %d\n",
+			__func__,
+			meta_pte_req_per_frame_ub);
+	dml_print("DML_DLG: %s: meta_pte_bytes_per_frame_ub = %d\n",
+			__func__,
+			meta_pte_bytes_per_frame_ub);
+
+	if (!surf_vert)
+		meta_chunk_threshold = 2 * min_meta_chunk_width - meta_req_width;
+	else
+		meta_chunk_threshold = 2 * min_meta_chunk_width - meta_req_height;
+
+	if (meta_row_remainder <= meta_chunk_threshold)
+		rq_dlg_param->meta_chunks_per_row_ub = meta_chunk_per_row_int + 1;
+	else
+		rq_dlg_param->meta_chunks_per_row_ub = meta_chunk_per_row_int + 2;
+
+	// ------
+	// dpte
+	// ------
+	if (surf_linear) {
+		log2_vmpg_height = 0;   // one line high
+	} else {
+		log2_vmpg_height = (log2_vmpg_bytes - 8) / 2 + log2_blk256_height;
+	}
+	log2_vmpg_width = log2_vmpg_bytes - log2_bytes_per_element - log2_vmpg_height;
+
+	// only 3 possible shapes for dpte request in dimensions of ptes: 8x1, 4x2, 2x4.
+	if (surf_linear) { //one 64B PTE request returns 8 PTEs
+		log2_dpte_req_height_ptes = 0;
+		log2_dpte_req_width = log2_vmpg_width + 3;
+		log2_dpte_req_height = 0;
+	} else if (log2_blk_bytes == 12) { //4KB tile means 4kB page size
+		//one 64B req gives 8x1 PTEs for 4KB tile
+		log2_dpte_req_height_ptes = 0;
+		log2_dpte_req_width = log2_blk_width + 3;
+		log2_dpte_req_height = log2_blk_height + 0;
+	} else if ((log2_blk_bytes >= 16) && (log2_vmpg_bytes == 12)) { // tile block >= 64KB
+		//two 64B reqs of 2x4 PTEs give 16 PTEs to cover 64KB
+		log2_dpte_req_height_ptes = 4;
+		log2_dpte_req_width = log2_blk256_width + 4; // log2_64KB_width
+		log2_dpte_req_height = log2_blk256_height + 4; // log2_64KB_height
+	} else { //64KB page size and must 64KB tile block
+		 //one 64B req gives 8x1 PTEs for 64KB tile
+		log2_dpte_req_height_ptes = 0;
+		log2_dpte_req_width = log2_blk_width + 3;
+		log2_dpte_req_height = log2_blk_height + 0;
+	}
+
+	// The dpte request dimensions in data elements is dpte_req_width x dpte_req_height
+	// log2_vmpg_width is how much 1 pte represent, now calculating how much a 64b pte req represent
+	// That depends on the pte shape (i.e. 8x1, 4x2, 2x4)
+	//log2_dpte_req_height    = log2_vmpg_height + log2_dpte_req_height_ptes;
+	//log2_dpte_req_width     = log2_vmpg_width + log2_dpte_req_width_ptes;
+	dpte_req_height = 1 << log2_dpte_req_height;
+	dpte_req_width = 1 << log2_dpte_req_width;
+
+	// calculate pitch dpte row buffer can hold
+	// round the result down to a power of two.
+	pde_buf_entries = yuv420 ? (pde_proc_buffer_size_64k_reqs >> 1) : pde_proc_buffer_size_64k_reqs;
+	if (surf_linear) {
+		unsigned int dpte_row_height;
+
+		log2_dpte_row_height_linear = dml_floor(dml_log2(dml_min(64 * 1024 * pde_buf_entries
+										/ bytes_per_element,
+								dpte_buf_in_pte_reqs
+										* dpte_req_width)
+								/ data_pitch),
+				1);
+
+		ASSERT(log2_dpte_row_height_linear >= 3);
+
+		if (log2_dpte_row_height_linear > 7)
+			log2_dpte_row_height_linear = 7;
+
+		log2_dpte_row_height = log2_dpte_row_height_linear;
+		// For linear, the dpte row is pitch dependent and the pte requests wrap at the pitch boundary.
+		// the dpte_row_width_ub is the upper bound of data_pitch*dpte_row_height in elements with this unique buffering.
+		dpte_row_height = 1 << log2_dpte_row_height;
+		dpte_row_width_ub = dml_round_to_multiple(data_pitch * dpte_row_height - 1,
+				dpte_req_width,
+				1) + dpte_req_width;
+		rq_dlg_param->dpte_req_per_row_ub = dpte_row_width_ub / dpte_req_width;
+	} else {
+		// the upper bound of the dpte_row_width without dependency on viewport position follows.
+		// for tiled mode, row height is the same as req height and row store up to vp size upper bound
+		if (!surf_vert) {
+			log2_dpte_row_height = log2_dpte_req_height;
+			dpte_row_width_ub = dml_round_to_multiple(vp_width - 1, dpte_req_width, 1)
+					+ dpte_req_width;
+			rq_dlg_param->dpte_req_per_row_ub = dpte_row_width_ub / dpte_req_width;
+		} else {
+			log2_dpte_row_height =
+					(log2_blk_width < log2_dpte_req_width) ?
+							log2_blk_width : log2_dpte_req_width;
+			dpte_row_width_ub = dml_round_to_multiple(vp_height - 1, dpte_req_height, 1)
+					+ dpte_req_height;
+			rq_dlg_param->dpte_req_per_row_ub = dpte_row_width_ub / dpte_req_height;
+		}
+	}
+	if (log2_blk_bytes >= 16 && log2_vmpg_bytes == 12) // tile block >= 64KB
+		rq_dlg_param->dpte_bytes_per_row_ub = rq_dlg_param->dpte_req_per_row_ub * 128; //2*64B dpte request
+	else
+		rq_dlg_param->dpte_bytes_per_row_ub = rq_dlg_param->dpte_req_per_row_ub * 64; //64B dpte request
+
+	rq_dlg_param->dpte_row_height = 1 << log2_dpte_row_height;
+
+	// the dpte_group_bytes is reduced for the specific case of vertical
+	// access of a tile surface that has dpte request of 8x1 ptes.
+	if (!surf_linear && (log2_dpte_req_height_ptes == 0) && surf_vert) //reduced, in this case, will have page fault within a group
+		rq_sizing_param->dpte_group_bytes = 512;
+	else
+		//full size
+		rq_sizing_param->dpte_group_bytes = 2048;
+
+	//since pte request size is 64byte, the number of data pte requests per full sized group is as follows.
+	log2_dpte_group_bytes = dml_log2(rq_sizing_param->dpte_group_bytes);
+	log2_dpte_group_length = log2_dpte_group_bytes - 6; //length in 64b requests
+
+	// full sized data pte group width in elements
+	if (!surf_vert)
+		log2_dpte_group_width = log2_dpte_group_length + log2_dpte_req_width;
+	else
+		log2_dpte_group_width = log2_dpte_group_length + log2_dpte_req_height;
+
+	//But if the tile block >=64KB and the page size is 4KB, then each dPTE request is 2*64B
+	if ((log2_blk_bytes >= 16) && (log2_vmpg_bytes == 12)) // tile block >= 64KB
+		log2_dpte_group_width = log2_dpte_group_width - 1;
+
+	dpte_group_width = 1 << log2_dpte_group_width;
+
+	// since dpte groups are only aligned to dpte_req_width and not dpte_group_width,
+	// the upper bound for the dpte groups per row is as follows.
+	rq_dlg_param->dpte_groups_per_row_ub = dml_ceil((double) dpte_row_width_ub / dpte_group_width,
+			1);
+}
+
+static void get_surf_rq_param(struct display_mode_lib *mode_lib,
+		display_data_rq_sizing_params_st *rq_sizing_param,
+		display_data_rq_dlg_params_st *rq_dlg_param,
+		display_data_rq_misc_params_st *rq_misc_param,
+		const display_pipe_source_params_st *pipe_src_param,
+		bool is_chroma)
+{
+	unsigned int vp_width = 0;
+	unsigned int vp_height = 0;
+	unsigned int data_pitch = 0;
+	unsigned int meta_pitch = 0;
+	unsigned int ppe = 1;
+
+	// TODO check if ppe apply for both luma and chroma in 422 case
+	if (is_chroma) {
+		vp_width = pipe_src_param->viewport_width_c / ppe;
+		vp_height = pipe_src_param->viewport_height_c;
+		data_pitch = pipe_src_param->data_pitch_c;
+		meta_pitch = pipe_src_param->meta_pitch_c;
+	} else {
+		vp_width = pipe_src_param->viewport_width / ppe;
+		vp_height = pipe_src_param->viewport_height;
+		data_pitch = pipe_src_param->data_pitch;
+		meta_pitch = pipe_src_param->meta_pitch;
+	}
+
+	rq_sizing_param->chunk_bytes = 8192;
+
+	if (rq_sizing_param->chunk_bytes == 64 * 1024)
+		rq_sizing_param->min_chunk_bytes = 0;
+	else
+		rq_sizing_param->min_chunk_bytes = 1024;
+
+	rq_sizing_param->meta_chunk_bytes = 2048;
+	rq_sizing_param->min_meta_chunk_bytes = 256;
+
+	rq_sizing_param->mpte_group_bytes = 2048;
+
+	get_meta_and_pte_attr(mode_lib,
+			rq_dlg_param,
+			rq_misc_param,
+			rq_sizing_param,
+			vp_width,
+			vp_height,
+			data_pitch,
+			meta_pitch,
+			pipe_src_param->source_format,
+			pipe_src_param->sw_mode,
+			pipe_src_param->macro_tile_size,
+			pipe_src_param->source_scan,
+			is_chroma);
+}
+
+static void dml20v2_rq_dlg_get_rq_params(struct display_mode_lib *mode_lib,
+		display_rq_params_st *rq_param,
+		const display_pipe_source_params_st *pipe_src_param)
+{
+	// get param for luma surface
+	rq_param->yuv420 = pipe_src_param->source_format == dm_420_8
+			|| pipe_src_param->source_format == dm_420_10;
+	rq_param->yuv420_10bpc = pipe_src_param->source_format == dm_420_10;
+
+	get_surf_rq_param(mode_lib,
+			&(rq_param->sizing.rq_l),
+			&(rq_param->dlg.rq_l),
+			&(rq_param->misc.rq_l),
+			pipe_src_param,
+			0);
+
+	if (is_dual_plane((enum source_format_class)(pipe_src_param->source_format))) {
+		// get param for chroma surface
+		get_surf_rq_param(mode_lib,
+				&(rq_param->sizing.rq_c),
+				&(rq_param->dlg.rq_c),
+				&(rq_param->misc.rq_c),
+				pipe_src_param,
+				1);
+	}
+
+	// calculate how to split the det buffer space between luma and chroma
+	handle_det_buf_split(mode_lib, rq_param, pipe_src_param);
+	print__rq_params_st(mode_lib, rq_param);
+}
+
+void dml20v2_rq_dlg_get_rq_reg(struct display_mode_lib *mode_lib,
+		display_rq_regs_st *rq_regs,
+		const display_pipe_params_st *pipe_param)
+{
+	display_rq_params_st rq_param = {0};
+
+	memset(rq_regs, 0, sizeof(*rq_regs));
+	dml20v2_rq_dlg_get_rq_params(mode_lib, &rq_param, &pipe_param->src);
+	extract_rq_regs(mode_lib, rq_regs, &rq_param);
+
+	print__rq_regs_st(mode_lib, rq_regs);
+}
+
+// Note: currently taken in as is.
+// Nice to decouple code from hw register implement and extract code that are repeated for luma and chroma.
+static void dml20v2_rq_dlg_get_dlg_params(struct display_mode_lib *mode_lib,
+		const display_e2e_pipe_params_st *e2e_pipe_param,
+		const unsigned int num_pipes,
+		const unsigned int pipe_idx,
+		display_dlg_regs_st *disp_dlg_regs,
+		display_ttu_regs_st *disp_ttu_regs,
+		const display_rq_dlg_params_st *rq_dlg_param,
+		const display_dlg_sys_params_st *dlg_sys_param,
+		const bool cstate_en,
+		const bool pstate_en)
+{
+	const display_pipe_source_params_st *src = &e2e_pipe_param[pipe_idx].pipe.src;
+	const display_pipe_dest_params_st *dst = &e2e_pipe_param[pipe_idx].pipe.dest;
+	const display_output_params_st *dout = &e2e_pipe_param[pipe_idx].dout;
+	const display_clocks_and_cfg_st *clks = &e2e_pipe_param[pipe_idx].clks_cfg;
+	const scaler_ratio_depth_st *scl = &e2e_pipe_param[pipe_idx].pipe.scale_ratio_depth;
+	const scaler_taps_st *taps = &e2e_pipe_param[pipe_idx].pipe.scale_taps;
+
+	// -------------------------
+	// Section 1.15.2.1: OTG dependent Params
+	// -------------------------
+	// Timing
+	unsigned int htotal = dst->htotal;
+//    unsigned int hblank_start = dst.hblank_start; // TODO: Remove
+	unsigned int hblank_end = dst->hblank_end;
+	unsigned int vblank_start = dst->vblank_start;
+	unsigned int vblank_end = dst->vblank_end;
+	unsigned int min_vblank = mode_lib->ip.min_vblank_lines;
+
+	double dppclk_freq_in_mhz = clks->dppclk_mhz;
+	double dispclk_freq_in_mhz = clks->dispclk_mhz;
+	double refclk_freq_in_mhz = clks->refclk_mhz;
+	double pclk_freq_in_mhz = dst->pixel_rate_mhz;
+	bool interlaced = dst->interlaced;
+
+	double ref_freq_to_pix_freq = refclk_freq_in_mhz / pclk_freq_in_mhz;
+
+	double min_dcfclk_mhz;
+	double t_calc_us;
+	double min_ttu_vblank;
+
+	double min_dst_y_ttu_vblank;
+	unsigned int dlg_vblank_start;
+	bool dual_plane;
+	unsigned int access_dir;
+	unsigned int vp_height_l;
+	unsigned int vp_width_l;
+	unsigned int vp_height_c;
+	unsigned int vp_width_c;
+
+	// Scaling
+	unsigned int htaps_l;
+	unsigned int htaps_c;
+	double hratio_l;
+	double hratio_c;
+	double vratio_l;
+	double vratio_c;
+	bool scl_enable;
+
+	double line_time_in_us;
+	//    double vinit_l;
+	//    double vinit_c;
+	//    double vinit_bot_l;
+	//    double vinit_bot_c;
+
+	//    unsigned int swath_height_l;
+	unsigned int swath_width_ub_l;
+	//    unsigned int dpte_bytes_per_row_ub_l;
+	unsigned int dpte_groups_per_row_ub_l;
+	//    unsigned int meta_pte_bytes_per_frame_ub_l;
+	//    unsigned int meta_bytes_per_row_ub_l;
+
+	//    unsigned int swath_height_c;
+	unsigned int swath_width_ub_c;
+	//   unsigned int dpte_bytes_per_row_ub_c;
+	unsigned int dpte_groups_per_row_ub_c;
+
+	unsigned int meta_chunks_per_row_ub_l;
+	unsigned int meta_chunks_per_row_ub_c;
+	unsigned int vupdate_offset;
+	unsigned int vupdate_width;
+	unsigned int vready_offset;
+
+	unsigned int dppclk_delay_subtotal;
+	unsigned int dispclk_delay_subtotal;
+	unsigned int pixel_rate_delay_subtotal;
+
+	unsigned int vstartup_start;
+	unsigned int dst_x_after_scaler;
+	unsigned int dst_y_after_scaler;
+	double line_wait;
+	double dst_y_prefetch;
+	double dst_y_per_vm_vblank;
+	double dst_y_per_row_vblank;
+	double dst_y_per_vm_flip;
+	double dst_y_per_row_flip;
+	double min_dst_y_per_vm_vblank;
+	double min_dst_y_per_row_vblank;
+	double lsw;
+	double vratio_pre_l;
+	double vratio_pre_c;
+	unsigned int req_per_swath_ub_l;
+	unsigned int req_per_swath_ub_c;
+	unsigned int meta_row_height_l;
+	unsigned int meta_row_height_c;
+	unsigned int swath_width_pixels_ub_l;
+	unsigned int swath_width_pixels_ub_c;
+	unsigned int scaler_rec_in_width_l;
+	unsigned int scaler_rec_in_width_c;
+	unsigned int dpte_row_height_l;
+	unsigned int dpte_row_height_c;
+	double hscale_pixel_rate_l;
+	double hscale_pixel_rate_c;
+	double min_hratio_fact_l;
+	double min_hratio_fact_c;
+	double refcyc_per_line_delivery_pre_l;
+	double refcyc_per_line_delivery_pre_c;
+	double refcyc_per_line_delivery_l;
+	double refcyc_per_line_delivery_c;
+
+	double refcyc_per_req_delivery_pre_l;
+	double refcyc_per_req_delivery_pre_c;
+	double refcyc_per_req_delivery_l;
+	double refcyc_per_req_delivery_c;
+
+	unsigned int full_recout_width;
+	double refcyc_per_req_delivery_pre_cur0;
+	double refcyc_per_req_delivery_cur0;
+	double refcyc_per_req_delivery_pre_cur1;
+	double refcyc_per_req_delivery_cur1;
+
+	memset(disp_dlg_regs, 0, sizeof(*disp_dlg_regs));
+	memset(disp_ttu_regs, 0, sizeof(*disp_ttu_regs));
+
+	dml_print("DML_DLG: %s:  cstate_en = %d\n", __func__, cstate_en);
+	dml_print("DML_DLG: %s:  pstate_en = %d\n", __func__, pstate_en);
+
+	dml_print("DML_DLG: %s: dppclk_freq_in_mhz     = %3.2f\n", __func__, dppclk_freq_in_mhz);
+	dml_print("DML_DLG: %s: dispclk_freq_in_mhz    = %3.2f\n", __func__, dispclk_freq_in_mhz);
+	dml_print("DML_DLG: %s: refclk_freq_in_mhz     = %3.2f\n", __func__, refclk_freq_in_mhz);
+	dml_print("DML_DLG: %s: pclk_freq_in_mhz       = %3.2f\n", __func__, pclk_freq_in_mhz);
+	dml_print("DML_DLG: %s: interlaced             = %d\n", __func__, interlaced);
+	ASSERT(ref_freq_to_pix_freq < 4.0);
+
+	disp_dlg_regs->ref_freq_to_pix_freq =
+			(unsigned int) (ref_freq_to_pix_freq * dml_pow(2, 19));
+	disp_dlg_regs->refcyc_per_htotal = (unsigned int) (ref_freq_to_pix_freq * (double) htotal
+			* dml_pow(2, 8));
+	disp_dlg_regs->dlg_vblank_end = interlaced ? (vblank_end / 2) : vblank_end; // 15 bits
+	disp_dlg_regs->refcyc_h_blank_end = (unsigned int) ((double) hblank_end
+			* (double) ref_freq_to_pix_freq);
+	ASSERT(disp_dlg_regs->refcyc_h_blank_end < (unsigned int) dml_pow(2, 13));
+
+	min_dcfclk_mhz = dlg_sys_param->deepsleep_dcfclk_mhz;
+	t_calc_us = get_tcalc(mode_lib, e2e_pipe_param, num_pipes);
+	min_ttu_vblank = get_min_ttu_vblank(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
+
+	min_dst_y_ttu_vblank = min_ttu_vblank * pclk_freq_in_mhz / (double) htotal;
+	dlg_vblank_start = interlaced ? (vblank_start / 2) : vblank_start;
+
+	disp_dlg_regs->min_dst_y_next_start = (unsigned int) (((double) dlg_vblank_start
+			+ min_dst_y_ttu_vblank) * dml_pow(2, 2));
+	ASSERT(disp_dlg_regs->min_dst_y_next_start < (unsigned int) dml_pow(2, 18));
+
+	dml_print("DML_DLG: %s: min_dcfclk_mhz                         = %3.2f\n",
+			__func__,
+			min_dcfclk_mhz);
+	dml_print("DML_DLG: %s: min_ttu_vblank                         = %3.2f\n",
+			__func__,
+			min_ttu_vblank);
+	dml_print("DML_DLG: %s: min_dst_y_ttu_vblank                   = %3.2f\n",
+			__func__,
+			min_dst_y_ttu_vblank);
+	dml_print("DML_DLG: %s: t_calc_us                              = %3.2f\n",
+			__func__,
+			t_calc_us);
+	dml_print("DML_DLG: %s: disp_dlg_regs->min_dst_y_next_start    = 0x%0x\n",
+			__func__,
+			disp_dlg_regs->min_dst_y_next_start);
+	dml_print("DML_DLG: %s: ref_freq_to_pix_freq                   = %3.2f\n",
+			__func__,
+			ref_freq_to_pix_freq);
+
+	// -------------------------
+	// Section 1.15.2.2: Prefetch, Active and TTU
+	// -------------------------
+	// Prefetch Calc
+	// Source
+//             dcc_en              = src.dcc;
+	dual_plane = is_dual_plane((enum source_format_class)(src->source_format));
+	access_dir = (src->source_scan == dm_vert); // vp access direction: horizontal or vertical accessed
+//      bytes_per_element_l = get_bytes_per_element(source_format_class(src.source_format), 0);
+//      bytes_per_element_c = get_bytes_per_element(source_format_class(src.source_format), 1);
+	vp_height_l = src->viewport_height;
+	vp_width_l = src->viewport_width;
+	vp_height_c = src->viewport_height_c;
+	vp_width_c = src->viewport_width_c;
+
+	// Scaling
+	htaps_l = taps->htaps;
+	htaps_c = taps->htaps_c;
+	hratio_l = scl->hscl_ratio;
+	hratio_c = scl->hscl_ratio_c;
+	vratio_l = scl->vscl_ratio;
+	vratio_c = scl->vscl_ratio_c;
+	scl_enable = scl->scl_enable;
+
+	line_time_in_us = (htotal / pclk_freq_in_mhz);
+//     vinit_l         = scl.vinit;
+//     vinit_c         = scl.vinit_c;
+//     vinit_bot_l     = scl.vinit_bot;
+//     vinit_bot_c     = scl.vinit_bot_c;
+
+//    unsigned int swath_height_l                 = rq_dlg_param->rq_l.swath_height;
+	swath_width_ub_l = rq_dlg_param->rq_l.swath_width_ub;
+//    unsigned int dpte_bytes_per_row_ub_l        = rq_dlg_param->rq_l.dpte_bytes_per_row_ub;
+	dpte_groups_per_row_ub_l = rq_dlg_param->rq_l.dpte_groups_per_row_ub;
+//    unsigned int meta_pte_bytes_per_frame_ub_l  = rq_dlg_param->rq_l.meta_pte_bytes_per_frame_ub;
+//    unsigned int meta_bytes_per_row_ub_l        = rq_dlg_param->rq_l.meta_bytes_per_row_ub;
+
+//    unsigned int swath_height_c                 = rq_dlg_param->rq_c.swath_height;
+	swath_width_ub_c = rq_dlg_param->rq_c.swath_width_ub;
+	//   dpte_bytes_per_row_ub_c        = rq_dlg_param->rq_c.dpte_bytes_per_row_ub;
+	dpte_groups_per_row_ub_c = rq_dlg_param->rq_c.dpte_groups_per_row_ub;
+
+	meta_chunks_per_row_ub_l = rq_dlg_param->rq_l.meta_chunks_per_row_ub;
+	meta_chunks_per_row_ub_c = rq_dlg_param->rq_c.meta_chunks_per_row_ub;
+	vupdate_offset = dst->vupdate_offset;
+	vupdate_width = dst->vupdate_width;
+	vready_offset = dst->vready_offset;
+
+	dppclk_delay_subtotal = mode_lib->ip.dppclk_delay_subtotal;
+	dispclk_delay_subtotal = mode_lib->ip.dispclk_delay_subtotal;
+
+	if (scl_enable)
+		dppclk_delay_subtotal += mode_lib->ip.dppclk_delay_scl;
+	else
+		dppclk_delay_subtotal += mode_lib->ip.dppclk_delay_scl_lb_only;
+
+	dppclk_delay_subtotal += mode_lib->ip.dppclk_delay_cnvc_formatter
+			+ src->num_cursors * mode_lib->ip.dppclk_delay_cnvc_cursor;
+
+	if (dout->dsc_enable) {
+		double dsc_delay = get_dsc_delay(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
+
+		dispclk_delay_subtotal += dsc_delay;
+	}
+
+	pixel_rate_delay_subtotal = dppclk_delay_subtotal * pclk_freq_in_mhz / dppclk_freq_in_mhz
+			+ dispclk_delay_subtotal * pclk_freq_in_mhz / dispclk_freq_in_mhz;
+
+	vstartup_start = dst->vstartup_start;
+	if (interlaced) {
+		if (vstartup_start / 2.0
+				- (double) (vready_offset + vupdate_width + vupdate_offset) / htotal
+				<= vblank_end / 2.0)
+			disp_dlg_regs->vready_after_vcount0 = 1;
+		else
+			disp_dlg_regs->vready_after_vcount0 = 0;
+	} else {
+		if (vstartup_start
+				- (double) (vready_offset + vupdate_width + vupdate_offset) / htotal
+				<= vblank_end)
+			disp_dlg_regs->vready_after_vcount0 = 1;
+		else
+			disp_dlg_regs->vready_after_vcount0 = 0;
+	}
+
+	// TODO: Where is this coming from?
+	if (interlaced)
+		vstartup_start = vstartup_start / 2;
+
+	// TODO: What if this min_vblank doesn't match the value in the dml_config_settings.cpp?
+	if (vstartup_start >= min_vblank) {
+		dml_print("WARNING: DML_DLG: %s: vblank_start=%d vblank_end=%d\n",
+				__func__,
+				vblank_start,
+				vblank_end);
+		dml_print("WARNING: DML_DLG: %s: vstartup_start=%d should be less than min_vblank=%d\n",
+				__func__,
+				vstartup_start,
+				min_vblank);
+		min_vblank = vstartup_start + 1;
+		dml_print("WARNING: DML_DLG: %s: vstartup_start=%d should be less than min_vblank=%d\n",
+				__func__,
+				vstartup_start,
+				min_vblank);
+	}
+
+	dst_x_after_scaler = get_dst_x_after_scaler(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
+	dst_y_after_scaler = get_dst_y_after_scaler(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
+
+	dml_print("DML_DLG: %s: htotal                                 = %d\n", __func__, htotal);
+	dml_print("DML_DLG: %s: pixel_rate_delay_subtotal              = %d\n",
+			__func__,
+			pixel_rate_delay_subtotal);
+	dml_print("DML_DLG: %s: dst_x_after_scaler                     = %d\n",
+			__func__,
+			dst_x_after_scaler);
+	dml_print("DML_DLG: %s: dst_y_after_scaler                     = %d\n",
+			__func__,
+			dst_y_after_scaler);
+
+	// Lwait
+	line_wait = mode_lib->soc.urgent_latency_us;
+	if (cstate_en)
+		line_wait = dml_max(mode_lib->soc.sr_enter_plus_exit_time_us, line_wait);
+	if (pstate_en)
+		line_wait = dml_max(mode_lib->soc.dram_clock_change_latency_us
+						+ mode_lib->soc.urgent_latency_us,
+				line_wait);
+	line_wait = line_wait / line_time_in_us;
+
+	dst_y_prefetch = get_dst_y_prefetch(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
+	dml_print("DML_DLG: %s: dst_y_prefetch (after rnd) = %3.2f\n", __func__, dst_y_prefetch);
+
+	dst_y_per_vm_vblank = get_dst_y_per_vm_vblank(mode_lib,
+			e2e_pipe_param,
+			num_pipes,
+			pipe_idx);
+	dst_y_per_row_vblank = get_dst_y_per_row_vblank(mode_lib,
+			e2e_pipe_param,
+			num_pipes,
+			pipe_idx);
+	dst_y_per_vm_flip = get_dst_y_per_vm_flip(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
+	dst_y_per_row_flip = get_dst_y_per_row_flip(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
+
+	min_dst_y_per_vm_vblank = 8.0;
+	min_dst_y_per_row_vblank = 16.0;
+
+	// magic!
+	if (htotal <= 75) {
+		min_vblank = 300;
+		min_dst_y_per_vm_vblank = 100.0;
+		min_dst_y_per_row_vblank = 100.0;
+	}
+
+	dml_print("DML_DLG: %s: dst_y_per_vm_vblank    = %3.2f\n", __func__, dst_y_per_vm_vblank);
+	dml_print("DML_DLG: %s: dst_y_per_row_vblank   = %3.2f\n", __func__, dst_y_per_row_vblank);
+
+	ASSERT(dst_y_per_vm_vblank < min_dst_y_per_vm_vblank);
+	ASSERT(dst_y_per_row_vblank < min_dst_y_per_row_vblank);
+
+	ASSERT(dst_y_prefetch > (dst_y_per_vm_vblank + dst_y_per_row_vblank));
+	lsw = dst_y_prefetch - (dst_y_per_vm_vblank + dst_y_per_row_vblank);
+
+	dml_print("DML_DLG: %s: lsw = %3.2f\n", __func__, lsw);
+
+	vratio_pre_l = get_vratio_prefetch_l(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
+	vratio_pre_c = get_vratio_prefetch_c(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
+
+	dml_print("DML_DLG: %s: vratio_pre_l=%3.2f\n", __func__, vratio_pre_l);
+	dml_print("DML_DLG: %s: vratio_pre_c=%3.2f\n", __func__, vratio_pre_c);
+
+	// Active
+	req_per_swath_ub_l = rq_dlg_param->rq_l.req_per_swath_ub;
+	req_per_swath_ub_c = rq_dlg_param->rq_c.req_per_swath_ub;
+	meta_row_height_l = rq_dlg_param->rq_l.meta_row_height;
+	meta_row_height_c = rq_dlg_param->rq_c.meta_row_height;
+	swath_width_pixels_ub_l = 0;
+	swath_width_pixels_ub_c = 0;
+	scaler_rec_in_width_l = 0;
+	scaler_rec_in_width_c = 0;
+	dpte_row_height_l = rq_dlg_param->rq_l.dpte_row_height;
+	dpte_row_height_c = rq_dlg_param->rq_c.dpte_row_height;
+
+	swath_width_pixels_ub_l = swath_width_ub_l;
+	swath_width_pixels_ub_c = swath_width_ub_c;
+
+	if (htaps_l <= 1)
+		min_hratio_fact_l = 2.0;
+	else if (htaps_l <= 6) {
+		if ((hratio_l * 2.0) > 4.0)
+			min_hratio_fact_l = 4.0;
+		else
+			min_hratio_fact_l = hratio_l * 2.0;
+	} else {
+		if (hratio_l > 4.0)
+			min_hratio_fact_l = 4.0;
+		else
+			min_hratio_fact_l = hratio_l;
+	}
+
+	hscale_pixel_rate_l = min_hratio_fact_l * dppclk_freq_in_mhz;
+
+	if (htaps_c <= 1)
+		min_hratio_fact_c = 2.0;
+	else if (htaps_c <= 6) {
+		if ((hratio_c * 2.0) > 4.0)
+			min_hratio_fact_c = 4.0;
+		else
+			min_hratio_fact_c = hratio_c * 2.0;
+	} else {
+		if (hratio_c > 4.0)
+			min_hratio_fact_c = 4.0;
+		else
+			min_hratio_fact_c = hratio_c;
+	}
+
+	hscale_pixel_rate_c = min_hratio_fact_c * dppclk_freq_in_mhz;
+
+	refcyc_per_line_delivery_pre_l = 0.;
+	refcyc_per_line_delivery_pre_c = 0.;
+	refcyc_per_line_delivery_l = 0.;
+	refcyc_per_line_delivery_c = 0.;
+
+	refcyc_per_req_delivery_pre_l = 0.;
+	refcyc_per_req_delivery_pre_c = 0.;
+	refcyc_per_req_delivery_l = 0.;
+	refcyc_per_req_delivery_c = 0.;
+
+	full_recout_width = 0;
+	// In ODM
+	if (src->is_hsplit) {
+		// This "hack"  is only allowed (and valid) for MPC combine. In ODM
+		// combine, you MUST specify the full_recout_width...according to Oswin
+		if (dst->full_recout_width == 0 && !dst->odm_combine) {
+			dml_print("DML_DLG: %s: Warning: full_recout_width not set in hsplit mode\n",
+					__func__);
+			full_recout_width = dst->recout_width * 2; // assume half split for dcn1
+		} else
+			full_recout_width = dst->full_recout_width;
+	} else
+		full_recout_width = dst->recout_width;
+
+	// As of DCN2, mpc_combine and odm_combine are mutually exclusive
+	refcyc_per_line_delivery_pre_l = get_refcyc_per_delivery(mode_lib,
+			refclk_freq_in_mhz,
+			pclk_freq_in_mhz,
+			dst->odm_combine,
+			full_recout_width,
+			dst->hactive,
+			vratio_pre_l,
+			hscale_pixel_rate_l,
+			swath_width_pixels_ub_l,
+			1); // per line
+
+	refcyc_per_line_delivery_l = get_refcyc_per_delivery(mode_lib,
+			refclk_freq_in_mhz,
+			pclk_freq_in_mhz,
+			dst->odm_combine,
+			full_recout_width,
+			dst->hactive,
+			vratio_l,
+			hscale_pixel_rate_l,
+			swath_width_pixels_ub_l,
+			1); // per line
+
+	dml_print("DML_DLG: %s: full_recout_width              = %d\n",
+			__func__,
+			full_recout_width);
+	dml_print("DML_DLG: %s: hscale_pixel_rate_l            = %3.2f\n",
+			__func__,
+			hscale_pixel_rate_l);
+	dml_print("DML_DLG: %s: refcyc_per_line_delivery_pre_l = %3.2f\n",
+			__func__,
+			refcyc_per_line_delivery_pre_l);
+	dml_print("DML_DLG: %s: refcyc_per_line_delivery_l     = %3.2f\n",
+			__func__,
+			refcyc_per_line_delivery_l);
+
+	if (dual_plane) {
+		refcyc_per_line_delivery_pre_c = get_refcyc_per_delivery(mode_lib,
+				refclk_freq_in_mhz,
+				pclk_freq_in_mhz,
+				dst->odm_combine,
+				full_recout_width,
+				dst->hactive,
+				vratio_pre_c,
+				hscale_pixel_rate_c,
+				swath_width_pixels_ub_c,
+				1); // per line
+
+		refcyc_per_line_delivery_c = get_refcyc_per_delivery(mode_lib,
+				refclk_freq_in_mhz,
+				pclk_freq_in_mhz,
+				dst->odm_combine,
+				full_recout_width,
+				dst->hactive,
+				vratio_c,
+				hscale_pixel_rate_c,
+				swath_width_pixels_ub_c,
+				1);  // per line
+
+		dml_print("DML_DLG: %s: refcyc_per_line_delivery_pre_c = %3.2f\n",
+				__func__,
+				refcyc_per_line_delivery_pre_c);
+		dml_print("DML_DLG: %s: refcyc_per_line_delivery_c     = %3.2f\n",
+				__func__,
+				refcyc_per_line_delivery_c);
+	}
+
+	// TTU - Luma / Chroma
+	if (access_dir) {  // vertical access
+		scaler_rec_in_width_l = vp_height_l;
+		scaler_rec_in_width_c = vp_height_c;
+	} else {
+		scaler_rec_in_width_l = vp_width_l;
+		scaler_rec_in_width_c = vp_width_c;
+	}
+
+	refcyc_per_req_delivery_pre_l = get_refcyc_per_delivery(mode_lib,
+			refclk_freq_in_mhz,
+			pclk_freq_in_mhz,
+			dst->odm_combine,
+			full_recout_width,
+			dst->hactive,
+			vratio_pre_l,
+			hscale_pixel_rate_l,
+			scaler_rec_in_width_l,
+			req_per_swath_ub_l);  // per req
+	refcyc_per_req_delivery_l = get_refcyc_per_delivery(mode_lib,
+			refclk_freq_in_mhz,
+			pclk_freq_in_mhz,
+			dst->odm_combine,
+			full_recout_width,
+			dst->hactive,
+			vratio_l,
+			hscale_pixel_rate_l,
+			scaler_rec_in_width_l,
+			req_per_swath_ub_l);  // per req
+
+	dml_print("DML_DLG: %s: refcyc_per_req_delivery_pre_l = %3.2f\n",
+			__func__,
+			refcyc_per_req_delivery_pre_l);
+	dml_print("DML_DLG: %s: refcyc_per_req_delivery_l     = %3.2f\n",
+			__func__,
+			refcyc_per_req_delivery_l);
+
+	ASSERT(refcyc_per_req_delivery_pre_l < dml_pow(2, 13));
+	ASSERT(refcyc_per_req_delivery_l < dml_pow(2, 13));
+
+	if (dual_plane) {
+		refcyc_per_req_delivery_pre_c = get_refcyc_per_delivery(mode_lib,
+				refclk_freq_in_mhz,
+				pclk_freq_in_mhz,
+				dst->odm_combine,
+				full_recout_width,
+				dst->hactive,
+				vratio_pre_c,
+				hscale_pixel_rate_c,
+				scaler_rec_in_width_c,
+				req_per_swath_ub_c);  // per req
+		refcyc_per_req_delivery_c = get_refcyc_per_delivery(mode_lib,
+				refclk_freq_in_mhz,
+				pclk_freq_in_mhz,
+				dst->odm_combine,
+				full_recout_width,
+				dst->hactive,
+				vratio_c,
+				hscale_pixel_rate_c,
+				scaler_rec_in_width_c,
+				req_per_swath_ub_c);  // per req
+
+		dml_print("DML_DLG: %s: refcyc_per_req_delivery_pre_c = %3.2f\n",
+				__func__,
+				refcyc_per_req_delivery_pre_c);
+		dml_print("DML_DLG: %s: refcyc_per_req_delivery_c     = %3.2f\n",
+				__func__,
+				refcyc_per_req_delivery_c);
+
+		ASSERT(refcyc_per_req_delivery_pre_c < dml_pow(2, 13));
+		ASSERT(refcyc_per_req_delivery_c < dml_pow(2, 13));
+	}
+
+	// TTU - Cursor
+	refcyc_per_req_delivery_pre_cur0 = 0.0;
+	refcyc_per_req_delivery_cur0 = 0.0;
+	if (src->num_cursors > 0) {
+		calculate_ttu_cursor(mode_lib,
+				&refcyc_per_req_delivery_pre_cur0,
+				&refcyc_per_req_delivery_cur0,
+				refclk_freq_in_mhz,
+				ref_freq_to_pix_freq,
+				hscale_pixel_rate_l,
+				scl->hscl_ratio,
+				vratio_pre_l,
+				vratio_l,
+				src->cur0_src_width,
+				(enum cursor_bpp)(src->cur0_bpp));
+	}
+
+	refcyc_per_req_delivery_pre_cur1 = 0.0;
+	refcyc_per_req_delivery_cur1 = 0.0;
+	if (src->num_cursors > 1) {
+		calculate_ttu_cursor(mode_lib,
+				&refcyc_per_req_delivery_pre_cur1,
+				&refcyc_per_req_delivery_cur1,
+				refclk_freq_in_mhz,
+				ref_freq_to_pix_freq,
+				hscale_pixel_rate_l,
+				scl->hscl_ratio,
+				vratio_pre_l,
+				vratio_l,
+				src->cur1_src_width,
+				(enum cursor_bpp)(src->cur1_bpp));
+	}
+
+	// TTU - Misc
+	// all hard-coded
+
+	// Assignment to register structures
+	disp_dlg_regs->dst_y_after_scaler = dst_y_after_scaler; // in terms of line
+	disp_dlg_regs->refcyc_x_after_scaler = dst_x_after_scaler * ref_freq_to_pix_freq; // in terms of refclk
+	ASSERT(disp_dlg_regs->refcyc_x_after_scaler < (unsigned int) dml_pow(2, 13));
+	disp_dlg_regs->dst_y_prefetch = (unsigned int) (dst_y_prefetch * dml_pow(2, 2));
+	disp_dlg_regs->dst_y_per_vm_vblank = (unsigned int) (dst_y_per_vm_vblank * dml_pow(2, 2));
+	disp_dlg_regs->dst_y_per_row_vblank = (unsigned int) (dst_y_per_row_vblank * dml_pow(2, 2));
+	disp_dlg_regs->dst_y_per_vm_flip = (unsigned int) (dst_y_per_vm_flip * dml_pow(2, 2));
+	disp_dlg_regs->dst_y_per_row_flip = (unsigned int) (dst_y_per_row_flip * dml_pow(2, 2));
+
+	disp_dlg_regs->vratio_prefetch = (unsigned int) (vratio_pre_l * dml_pow(2, 19));
+	disp_dlg_regs->vratio_prefetch_c = (unsigned int) (vratio_pre_c * dml_pow(2, 19));
+
+	disp_dlg_regs->refcyc_per_pte_group_vblank_l =
+			(unsigned int) (dst_y_per_row_vblank * (double) htotal
+					* ref_freq_to_pix_freq / (double) dpte_groups_per_row_ub_l);
+	ASSERT(disp_dlg_regs->refcyc_per_pte_group_vblank_l < (unsigned int) dml_pow(2, 13));
+
+	if (dual_plane) {
+		disp_dlg_regs->refcyc_per_pte_group_vblank_c = (unsigned int) (dst_y_per_row_vblank
+				* (double) htotal * ref_freq_to_pix_freq
+				/ (double) dpte_groups_per_row_ub_c);
+		ASSERT(disp_dlg_regs->refcyc_per_pte_group_vblank_c
+						< (unsigned int) dml_pow(2, 13));
+	}
+
+	disp_dlg_regs->refcyc_per_meta_chunk_vblank_l =
+			(unsigned int) (dst_y_per_row_vblank * (double) htotal
+					* ref_freq_to_pix_freq / (double) meta_chunks_per_row_ub_l);
+	ASSERT(disp_dlg_regs->refcyc_per_meta_chunk_vblank_l < (unsigned int) dml_pow(2, 13));
+
+	disp_dlg_regs->refcyc_per_meta_chunk_vblank_c =
+			disp_dlg_regs->refcyc_per_meta_chunk_vblank_l; // dcc for 4:2:0 is not supported in dcn1.0.  assigned to be the same as _l for now
+
+	disp_dlg_regs->refcyc_per_pte_group_flip_l = (unsigned int) (dst_y_per_row_flip * htotal
+			* ref_freq_to_pix_freq) / dpte_groups_per_row_ub_l;
+	disp_dlg_regs->refcyc_per_meta_chunk_flip_l = (unsigned int) (dst_y_per_row_flip * htotal
+			* ref_freq_to_pix_freq) / meta_chunks_per_row_ub_l;
+
+	if (dual_plane) {
+		disp_dlg_regs->refcyc_per_pte_group_flip_c = (unsigned int) (dst_y_per_row_flip
+				* htotal * ref_freq_to_pix_freq) / dpte_groups_per_row_ub_c;
+		disp_dlg_regs->refcyc_per_meta_chunk_flip_c = (unsigned int) (dst_y_per_row_flip
+				* htotal * ref_freq_to_pix_freq) / meta_chunks_per_row_ub_c;
+	}
+
+	disp_dlg_regs->dst_y_per_pte_row_nom_l = (unsigned int) ((double) dpte_row_height_l
+			/ (double) vratio_l * dml_pow(2, 2));
+	ASSERT(disp_dlg_regs->dst_y_per_pte_row_nom_l < (unsigned int) dml_pow(2, 17));
+
+	if (dual_plane) {
+		disp_dlg_regs->dst_y_per_pte_row_nom_c = (unsigned int) ((double) dpte_row_height_c
+				/ (double) vratio_c * dml_pow(2, 2));
+		if (disp_dlg_regs->dst_y_per_pte_row_nom_c >= (unsigned int) dml_pow(2, 17)) {
+			dml_print("DML_DLG: %s: Warning dst_y_per_pte_row_nom_c %u larger than supported by register format U15.2 %u\n",
+					__func__,
+					disp_dlg_regs->dst_y_per_pte_row_nom_c,
+					(unsigned int) dml_pow(2, 17) - 1);
+		}
+	}
+
+	disp_dlg_regs->dst_y_per_meta_row_nom_l = (unsigned int) ((double) meta_row_height_l
+			/ (double) vratio_l * dml_pow(2, 2));
+	ASSERT(disp_dlg_regs->dst_y_per_meta_row_nom_l < (unsigned int) dml_pow(2, 17));
+
+	disp_dlg_regs->dst_y_per_meta_row_nom_c = disp_dlg_regs->dst_y_per_meta_row_nom_l; // TODO: dcc for 4:2:0 is not supported in dcn1.0.  assigned to be the same as _l for now
+
+	disp_dlg_regs->refcyc_per_pte_group_nom_l = (unsigned int) ((double) dpte_row_height_l
+			/ (double) vratio_l * (double) htotal * ref_freq_to_pix_freq
+			/ (double) dpte_groups_per_row_ub_l);
+	if (disp_dlg_regs->refcyc_per_pte_group_nom_l >= (unsigned int) dml_pow(2, 23))
+		disp_dlg_regs->refcyc_per_pte_group_nom_l = dml_pow(2, 23) - 1;
+	disp_dlg_regs->refcyc_per_meta_chunk_nom_l = (unsigned int) ((double) meta_row_height_l
+			/ (double) vratio_l * (double) htotal * ref_freq_to_pix_freq
+			/ (double) meta_chunks_per_row_ub_l);
+	if (disp_dlg_regs->refcyc_per_meta_chunk_nom_l >= (unsigned int) dml_pow(2, 23))
+		disp_dlg_regs->refcyc_per_meta_chunk_nom_l = dml_pow(2, 23) - 1;
+
+	if (dual_plane) {
+		disp_dlg_regs->refcyc_per_pte_group_nom_c =
+				(unsigned int) ((double) dpte_row_height_c / (double) vratio_c
+						* (double) htotal * ref_freq_to_pix_freq
+						/ (double) dpte_groups_per_row_ub_c);
+		if (disp_dlg_regs->refcyc_per_pte_group_nom_c >= (unsigned int) dml_pow(2, 23))
+			disp_dlg_regs->refcyc_per_pte_group_nom_c = dml_pow(2, 23) - 1;
+
+		// TODO: Is this the right calculation? Does htotal need to be halved?
+		disp_dlg_regs->refcyc_per_meta_chunk_nom_c =
+				(unsigned int) ((double) meta_row_height_c / (double) vratio_c
+						* (double) htotal * ref_freq_to_pix_freq
+						/ (double) meta_chunks_per_row_ub_c);
+		if (disp_dlg_regs->refcyc_per_meta_chunk_nom_c >= (unsigned int) dml_pow(2, 23))
+			disp_dlg_regs->refcyc_per_meta_chunk_nom_c = dml_pow(2, 23) - 1;
+	}
+
+	disp_dlg_regs->refcyc_per_line_delivery_pre_l = (unsigned int) dml_floor(refcyc_per_line_delivery_pre_l,
+			1);
+	disp_dlg_regs->refcyc_per_line_delivery_l = (unsigned int) dml_floor(refcyc_per_line_delivery_l,
+			1);
+	ASSERT(disp_dlg_regs->refcyc_per_line_delivery_pre_l < (unsigned int) dml_pow(2, 13));
+	ASSERT(disp_dlg_regs->refcyc_per_line_delivery_l < (unsigned int) dml_pow(2, 13));
+
+	disp_dlg_regs->refcyc_per_line_delivery_pre_c = (unsigned int) dml_floor(refcyc_per_line_delivery_pre_c,
+			1);
+	disp_dlg_regs->refcyc_per_line_delivery_c = (unsigned int) dml_floor(refcyc_per_line_delivery_c,
+			1);
+	ASSERT(disp_dlg_regs->refcyc_per_line_delivery_pre_c < (unsigned int) dml_pow(2, 13));
+	ASSERT(disp_dlg_regs->refcyc_per_line_delivery_c < (unsigned int) dml_pow(2, 13));
+
+	disp_dlg_regs->chunk_hdl_adjust_cur0 = 3;
+	disp_dlg_regs->dst_y_offset_cur0 = 0;
+	disp_dlg_regs->chunk_hdl_adjust_cur1 = 3;
+	disp_dlg_regs->dst_y_offset_cur1 = 0;
+
+	disp_dlg_regs->dst_y_delta_drq_limit = 0x7fff; // off
+
+	disp_ttu_regs->refcyc_per_req_delivery_pre_l = (unsigned int) (refcyc_per_req_delivery_pre_l
+			* dml_pow(2, 10));
+	disp_ttu_regs->refcyc_per_req_delivery_l = (unsigned int) (refcyc_per_req_delivery_l
+			* dml_pow(2, 10));
+	disp_ttu_regs->refcyc_per_req_delivery_pre_c = (unsigned int) (refcyc_per_req_delivery_pre_c
+			* dml_pow(2, 10));
+	disp_ttu_regs->refcyc_per_req_delivery_c = (unsigned int) (refcyc_per_req_delivery_c
+			* dml_pow(2, 10));
+	disp_ttu_regs->refcyc_per_req_delivery_pre_cur0 =
+			(unsigned int) (refcyc_per_req_delivery_pre_cur0 * dml_pow(2, 10));
+	disp_ttu_regs->refcyc_per_req_delivery_cur0 = (unsigned int) (refcyc_per_req_delivery_cur0
+			* dml_pow(2, 10));
+	disp_ttu_regs->refcyc_per_req_delivery_pre_cur1 =
+			(unsigned int) (refcyc_per_req_delivery_pre_cur1 * dml_pow(2, 10));
+	disp_ttu_regs->refcyc_per_req_delivery_cur1 = (unsigned int) (refcyc_per_req_delivery_cur1
+			* dml_pow(2, 10));
+	disp_ttu_regs->qos_level_low_wm = 0;
+	ASSERT(disp_ttu_regs->qos_level_low_wm < dml_pow(2, 14));
+	disp_ttu_regs->qos_level_high_wm = (unsigned int) (4.0 * (double) htotal
+			* ref_freq_to_pix_freq);
+	/*ASSERT(disp_ttu_regs->qos_level_high_wm < dml_pow(2, 14));*/
+
+	disp_ttu_regs->qos_level_flip = 14;
+	disp_ttu_regs->qos_level_fixed_l = 8;
+	disp_ttu_regs->qos_level_fixed_c = 8;
+	disp_ttu_regs->qos_level_fixed_cur0 = 8;
+	disp_ttu_regs->qos_ramp_disable_l = 0;
+	disp_ttu_regs->qos_ramp_disable_c = 0;
+	disp_ttu_regs->qos_ramp_disable_cur0 = 0;
+
+	disp_ttu_regs->min_ttu_vblank = min_ttu_vblank * refclk_freq_in_mhz;
+	ASSERT(disp_ttu_regs->min_ttu_vblank < dml_pow(2, 24));
+
+	print__ttu_regs_st(mode_lib, disp_ttu_regs);
+	print__dlg_regs_st(mode_lib, disp_dlg_regs);
+}
+
+void dml20v2_rq_dlg_get_dlg_reg(struct display_mode_lib *mode_lib,
+		display_dlg_regs_st *dlg_regs,
+		display_ttu_regs_st *ttu_regs,
+		const display_e2e_pipe_params_st *e2e_pipe_param,
+		const unsigned int num_pipes,
+		const unsigned int pipe_idx,
+		const bool cstate_en,
+		const bool pstate_en,
+		const bool vm_en,
+		const bool ignore_viewport_pos,
+		const bool immediate_flip_support)
+{
+	display_rq_params_st rq_param = {0};
+	display_dlg_sys_params_st dlg_sys_param = {0};
+
+	// Get watermark and Tex.
+	dlg_sys_param.t_urg_wm_us = get_wm_urgent(mode_lib, e2e_pipe_param, num_pipes);
+	dlg_sys_param.deepsleep_dcfclk_mhz = get_clk_dcf_deepsleep(mode_lib,
+			e2e_pipe_param,
+			num_pipes);
+	dlg_sys_param.t_extra_us = get_urgent_extra_latency(mode_lib, e2e_pipe_param, num_pipes);
+	dlg_sys_param.mem_trip_us = get_wm_memory_trip(mode_lib, e2e_pipe_param, num_pipes);
+	dlg_sys_param.t_mclk_wm_us = get_wm_dram_clock_change(mode_lib, e2e_pipe_param, num_pipes);
+	dlg_sys_param.t_sr_wm_us = get_wm_stutter_enter_exit(mode_lib, e2e_pipe_param, num_pipes);
+	dlg_sys_param.total_flip_bw = get_total_immediate_flip_bw(mode_lib,
+			e2e_pipe_param,
+			num_pipes);
+	dlg_sys_param.total_flip_bytes = get_total_immediate_flip_bytes(mode_lib,
+			e2e_pipe_param,
+			num_pipes);
+
+	print__dlg_sys_params_st(mode_lib, &dlg_sys_param);
+
+	// system parameter calculation done
+
+	dml_print("DML_DLG: Calculation for pipe[%d] start\n\n", pipe_idx);
+	dml20v2_rq_dlg_get_rq_params(mode_lib, &rq_param, &e2e_pipe_param[pipe_idx].pipe.src);
+	dml20v2_rq_dlg_get_dlg_params(mode_lib,
+			e2e_pipe_param,
+			num_pipes,
+			pipe_idx,
+			dlg_regs,
+			ttu_regs,
+			&rq_param.dlg,
+			&dlg_sys_param,
+			cstate_en,
+			pstate_en);
+	dml_print("DML_DLG: Calculation for pipe[%d] end\n", pipe_idx);
+}
+
+static void calculate_ttu_cursor(struct display_mode_lib *mode_lib,
+		double *refcyc_per_req_delivery_pre_cur,
+		double *refcyc_per_req_delivery_cur,
+		double refclk_freq_in_mhz,
+		double ref_freq_to_pix_freq,
+		double hscale_pixel_rate_l,
+		double hscl_ratio,
+		double vratio_pre_l,
+		double vratio_l,
+		unsigned int cur_width,
+		enum cursor_bpp cur_bpp)
+{
+	unsigned int cur_src_width = cur_width;
+	unsigned int cur_req_size = 0;
+	unsigned int cur_req_width = 0;
+	double cur_width_ub = 0.0;
+	double cur_req_per_width = 0.0;
+	double hactive_cur = 0.0;
+
+	ASSERT(cur_src_width <= 256);
+
+	*refcyc_per_req_delivery_pre_cur = 0.0;
+	*refcyc_per_req_delivery_cur = 0.0;
+	if (cur_src_width > 0) {
+		unsigned int cur_bit_per_pixel = 0;
+
+		if (cur_bpp == dm_cur_2bit) {
+			cur_req_size = 64; // byte
+			cur_bit_per_pixel = 2;
+		} else { // 32bit
+			cur_bit_per_pixel = 32;
+			if (cur_src_width >= 1 && cur_src_width <= 16)
+				cur_req_size = 64;
+			else if (cur_src_width >= 17 && cur_src_width <= 31)
+				cur_req_size = 128;
+			else
+				cur_req_size = 256;
+		}
+
+		cur_req_width = (double) cur_req_size / ((double) cur_bit_per_pixel / 8.0);
+		cur_width_ub = dml_ceil((double) cur_src_width / (double) cur_req_width, 1)
+				* (double) cur_req_width;
+		cur_req_per_width = cur_width_ub / (double) cur_req_width;
+		hactive_cur = (double) cur_src_width / hscl_ratio; // TODO: oswin to think about what to do for cursor
+
+		if (vratio_pre_l <= 1.0) {
+			*refcyc_per_req_delivery_pre_cur = hactive_cur * ref_freq_to_pix_freq
+					/ (double) cur_req_per_width;
+		} else {
+			*refcyc_per_req_delivery_pre_cur = (double) refclk_freq_in_mhz
+					* (double) cur_src_width / hscale_pixel_rate_l
+					/ (double) cur_req_per_width;
+		}
+
+		ASSERT(*refcyc_per_req_delivery_pre_cur < dml_pow(2, 13));
+
+		if (vratio_l <= 1.0) {
+			*refcyc_per_req_delivery_cur = hactive_cur * ref_freq_to_pix_freq
+					/ (double) cur_req_per_width;
+		} else {
+			*refcyc_per_req_delivery_cur = (double) refclk_freq_in_mhz
+					* (double) cur_src_width / hscale_pixel_rate_l
+					/ (double) cur_req_per_width;
+		}
+
+		dml_print("DML_DLG: %s: cur_req_width                     = %d\n",
+				__func__,
+				cur_req_width);
+		dml_print("DML_DLG: %s: cur_width_ub                      = %3.2f\n",
+				__func__,
+				cur_width_ub);
+		dml_print("DML_DLG: %s: cur_req_per_width                 = %3.2f\n",
+				__func__,
+				cur_req_per_width);
+		dml_print("DML_DLG: %s: hactive_cur                       = %3.2f\n",
+				__func__,
+				hactive_cur);
+		dml_print("DML_DLG: %s: refcyc_per_req_delivery_pre_cur   = %3.2f\n",
+				__func__,
+				*refcyc_per_req_delivery_pre_cur);
+		dml_print("DML_DLG: %s: refcyc_per_req_delivery_cur       = %3.2f\n",
+				__func__,
+				*refcyc_per_req_delivery_cur);
+
+		ASSERT(*refcyc_per_req_delivery_cur < dml_pow(2, 13));
+	}
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_rq_dlg_calc_20v2.h b/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_rq_dlg_calc_20v2.h
new file mode 100644
index 000000000000..2b4e46ea1c3d
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn20/display_rq_dlg_calc_20v2.h
@@ -0,0 +1,73 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __DML20V2_DISPLAY_RQ_DLG_CALC_H__
+#define __DML20V2_DISPLAY_RQ_DLG_CALC_H__
+
+#include "../display_rq_dlg_helpers.h"
+
+struct display_mode_lib;
+
+
+// Function: dml_rq_dlg_get_rq_reg
+//  Main entry point for test to get the register values out of this DML class.
+//  This function calls <get_rq_param> and <extract_rq_regs> functions to calculate
+//  and then populate the rq_regs struct
+// Input:
+//  pipe_src_param - pipe source configuration (e.g. vp, pitch, etc.)
+// Output:
+//  rq_regs - struct that holds all the RQ registers field value.
+//            See also: <display_rq_regs_st>
+void dml20v2_rq_dlg_get_rq_reg(
+		struct display_mode_lib *mode_lib,
+		display_rq_regs_st *rq_regs,
+		const display_pipe_params_st *pipe_param);
+
+
+// Function: dml_rq_dlg_get_dlg_reg
+//   Calculate and return DLG and TTU register struct given the system setting
+// Output:
+//  dlg_regs - output DLG register struct
+//  ttu_regs - output DLG TTU register struct
+// Input:
+//  e2e_pipe_param - "compacted" array of e2e pipe param struct
+//  num_pipes - num of active "pipe" or "route"
+//  pipe_idx - index that identifies the e2e_pipe_param that corresponding to this dlg
+//  cstate - 0: when calculate min_ttu_vblank it is assumed cstate is not required. 1: Normal mode, cstate is considered.
+//           Added for legacy or unrealistic timing tests.
+void dml20v2_rq_dlg_get_dlg_reg(
+		struct display_mode_lib *mode_lib,
+		display_dlg_regs_st *dlg_regs,
+		display_ttu_regs_st *ttu_regs,
+		const display_e2e_pipe_params_st *e2e_pipe_param,
+		const unsigned int num_pipes,
+		const unsigned int pipe_idx,
+		const bool cstate_en,
+		const bool pstate_en,
+		const bool vm_en,
+		const bool ignore_viewport_pos,
+		const bool immediate_flip_support);
+
+#endif
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn21/display_mode_vba_21.c b/drivers/gpu/drm/amd/display/dc/dml/dcn21/display_mode_vba_21.c
new file mode 100644
index 000000000000..cd8cca651419
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn21/display_mode_vba_21.c
@@ -0,0 +1,6150 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+#include "../display_mode_lib.h"
+#include "../dml_inline_defs.h"
+#include "../display_mode_vba.h"
+#include "display_mode_vba_21.h"
+
+
+/*
+ * NOTE:
+ *   This file is gcc-parsable HW gospel, coming straight from HW engineers.
+ *
+ * It doesn't adhere to Linux kernel style and sometimes will do things in odd
+ * ways. Unless there is something clearly wrong with it the code should
+ * remain as-is as it provides us with a guarantee from HW that it is correct.
+ */
+typedef struct {
+	double DPPCLK;
+	double DISPCLK;
+	double PixelClock;
+	double DCFCLKDeepSleep;
+	unsigned int DPPPerPlane;
+	bool ScalerEnabled;
+	enum scan_direction_class SourceScan;
+	unsigned int BlockWidth256BytesY;
+	unsigned int BlockHeight256BytesY;
+	unsigned int BlockWidth256BytesC;
+	unsigned int BlockHeight256BytesC;
+	unsigned int InterlaceEnable;
+	unsigned int NumberOfCursors;
+	unsigned int VBlank;
+	unsigned int HTotal;
+} Pipe;
+
+typedef struct {
+	bool Enable;
+	unsigned int MaxPageTableLevels;
+	unsigned int CachedPageTableLevels;
+} HostVM;
+
+#define BPP_INVALID 0
+#define BPP_BLENDED_PIPE 0xffffffff
+#define DCN21_MAX_DSC_IMAGE_WIDTH 5184
+#define DCN21_MAX_420_IMAGE_WIDTH 4096
+
+static void DisplayPipeConfiguration(struct display_mode_lib *mode_lib);
+static void DISPCLKDPPCLKDCFCLKDeepSleepPrefetchParametersWatermarksAndPerformanceCalculation(
+		struct display_mode_lib *mode_lib);
+static unsigned int dscceComputeDelay(
+		unsigned int bpc,
+		double bpp,
+		unsigned int sliceWidth,
+		unsigned int numSlices,
+		enum output_format_class pixelFormat);
+static unsigned int dscComputeDelay(enum output_format_class pixelFormat);
+// Super monster function with some 45 argument
+static bool CalculatePrefetchSchedule(
+		struct display_mode_lib *mode_lib,
+		double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData,
+		double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly,
+		Pipe *myPipe,
+		unsigned int DSCDelay,
+		double DPPCLKDelaySubtotal,
+		double DPPCLKDelaySCL,
+		double DPPCLKDelaySCLLBOnly,
+		double DPPCLKDelayCNVCFormater,
+		double DPPCLKDelayCNVCCursor,
+		double DISPCLKDelaySubtotal,
+		unsigned int ScalerRecoutWidth,
+		enum output_format_class OutputFormat,
+		unsigned int MaxInterDCNTileRepeaters,
+		unsigned int VStartup,
+		unsigned int MaxVStartup,
+		unsigned int GPUVMPageTableLevels,
+		bool GPUVMEnable,
+		HostVM *myHostVM,
+		bool DynamicMetadataEnable,
+		int DynamicMetadataLinesBeforeActiveRequired,
+		unsigned int DynamicMetadataTransmittedBytes,
+		bool DCCEnable,
+		double UrgentLatency,
+		double UrgentExtraLatency,
+		double TCalc,
+		unsigned int PDEAndMetaPTEBytesFrame,
+		unsigned int MetaRowByte,
+		unsigned int PixelPTEBytesPerRow,
+		double PrefetchSourceLinesY,
+		unsigned int SwathWidthY,
+		double BytePerPixelDETY,
+		double VInitPreFillY,
+		unsigned int MaxNumSwathY,
+		double PrefetchSourceLinesC,
+		double BytePerPixelDETC,
+		double VInitPreFillC,
+		unsigned int MaxNumSwathC,
+		unsigned int SwathHeightY,
+		unsigned int SwathHeightC,
+		double TWait,
+		bool XFCEnabled,
+		double XFCRemoteSurfaceFlipDelay,
+		bool ProgressiveToInterlaceUnitInOPP,
+		double *DSTXAfterScaler,
+		double *DSTYAfterScaler,
+		double *DestinationLinesForPrefetch,
+		double *PrefetchBandwidth,
+		double *DestinationLinesToRequestVMInVBlank,
+		double *DestinationLinesToRequestRowInVBlank,
+		double *VRatioPrefetchY,
+		double *VRatioPrefetchC,
+		double *RequiredPrefetchPixDataBWLuma,
+		double *RequiredPrefetchPixDataBWChroma,
+		unsigned int *VStartupRequiredWhenNotEnoughTimeForDynamicMetadata,
+		double *Tno_bw,
+		double *prefetch_vmrow_bw,
+		unsigned int *swath_width_luma_ub,
+		unsigned int *swath_width_chroma_ub,
+		unsigned int *VUpdateOffsetPix,
+		double *VUpdateWidthPix,
+		double *VReadyOffsetPix);
+static double RoundToDFSGranularityUp(double Clock, double VCOSpeed);
+static double RoundToDFSGranularityDown(double Clock, double VCOSpeed);
+static double CalculateDCCConfiguration(
+		bool                 DCCEnabled,
+		bool                 DCCProgrammingAssumesScanDirectionUnknown,
+		unsigned int         ViewportWidth,
+		unsigned int         ViewportHeight,
+		unsigned int         DETBufferSize,
+		unsigned int         RequestHeight256Byte,
+		unsigned int         SwathHeight,
+		enum dm_swizzle_mode TilingFormat,
+		unsigned int         BytePerPixel,
+		enum scan_direction_class ScanOrientation,
+		unsigned int        *MaxUncompressedBlock,
+		unsigned int        *MaxCompressedBlock,
+		unsigned int        *Independent64ByteBlock);
+static double CalculatePrefetchSourceLines(
+		struct display_mode_lib *mode_lib,
+		double VRatio,
+		double vtaps,
+		bool Interlace,
+		bool ProgressiveToInterlaceUnitInOPP,
+		unsigned int SwathHeight,
+		unsigned int ViewportYStart,
+		double *VInitPreFill,
+		unsigned int *MaxNumSwath);
+static unsigned int CalculateVMAndRowBytes(
+		struct display_mode_lib *mode_lib,
+		bool DCCEnable,
+		unsigned int BlockHeight256Bytes,
+		unsigned int BlockWidth256Bytes,
+		enum source_format_class SourcePixelFormat,
+		unsigned int SurfaceTiling,
+		unsigned int BytePerPixel,
+		enum scan_direction_class ScanDirection,
+		unsigned int ViewportWidth,
+		unsigned int ViewportHeight,
+		unsigned int SwathWidthY,
+		bool GPUVMEnable,
+		bool HostVMEnable,
+		unsigned int HostVMMaxPageTableLevels,
+		unsigned int HostVMCachedPageTableLevels,
+		unsigned int VMMPageSize,
+		unsigned int PTEBufferSizeInRequests,
+		unsigned int Pitch,
+		unsigned int DCCMetaPitch,
+		unsigned int *MacroTileWidth,
+		unsigned int *MetaRowByte,
+		unsigned int *PixelPTEBytesPerRow,
+		bool *PTEBufferSizeNotExceeded,
+		unsigned int *dpte_row_width_ub,
+		unsigned int *dpte_row_height,
+		unsigned int *MetaRequestWidth,
+		unsigned int *MetaRequestHeight,
+		unsigned int *meta_row_width,
+		unsigned int *meta_row_height,
+		unsigned int *vm_group_bytes,
+		unsigned int *dpte_group_bytes,
+		unsigned int *PixelPTEReqWidth,
+		unsigned int *PixelPTEReqHeight,
+		unsigned int *PTERequestSize,
+		unsigned int *DPDE0BytesFrame,
+		unsigned int *MetaPTEBytesFrame);
+
+static double CalculateTWait(
+		unsigned int PrefetchMode,
+		double DRAMClockChangeLatency,
+		double UrgentLatency,
+		double SREnterPlusExitTime);
+static double CalculateRemoteSurfaceFlipDelay(
+		struct display_mode_lib *mode_lib,
+		double VRatio,
+		double SwathWidth,
+		double Bpp,
+		double LineTime,
+		double XFCTSlvVupdateOffset,
+		double XFCTSlvVupdateWidth,
+		double XFCTSlvVreadyOffset,
+		double XFCXBUFLatencyTolerance,
+		double XFCFillBWOverhead,
+		double XFCSlvChunkSize,
+		double XFCBusTransportTime,
+		double TCalc,
+		double TWait,
+		double *SrcActiveDrainRate,
+		double *TInitXFill,
+		double *TslvChk);
+static void CalculateActiveRowBandwidth(
+		bool GPUVMEnable,
+		enum source_format_class SourcePixelFormat,
+		double VRatio,
+		bool DCCEnable,
+		double LineTime,
+		unsigned int MetaRowByteLuma,
+		unsigned int MetaRowByteChroma,
+		unsigned int meta_row_height_luma,
+		unsigned int meta_row_height_chroma,
+		unsigned int PixelPTEBytesPerRowLuma,
+		unsigned int PixelPTEBytesPerRowChroma,
+		unsigned int dpte_row_height_luma,
+		unsigned int dpte_row_height_chroma,
+		double *meta_row_bw,
+		double *dpte_row_bw);
+static void CalculateFlipSchedule(
+		struct display_mode_lib *mode_lib,
+		double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData,
+		double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly,
+		double UrgentExtraLatency,
+		double UrgentLatency,
+		unsigned int GPUVMMaxPageTableLevels,
+		bool HostVMEnable,
+		unsigned int HostVMMaxPageTableLevels,
+		unsigned int HostVMCachedPageTableLevels,
+		bool GPUVMEnable,
+		double PDEAndMetaPTEBytesPerFrame,
+		double MetaRowBytes,
+		double DPTEBytesPerRow,
+		double BandwidthAvailableForImmediateFlip,
+		unsigned int TotImmediateFlipBytes,
+		enum source_format_class SourcePixelFormat,
+		double LineTime,
+		double VRatio,
+		double Tno_bw,
+		bool DCCEnable,
+		unsigned int dpte_row_height,
+		unsigned int meta_row_height,
+		unsigned int dpte_row_height_chroma,
+		unsigned int meta_row_height_chroma,
+		double *DestinationLinesToRequestVMInImmediateFlip,
+		double *DestinationLinesToRequestRowInImmediateFlip,
+		double *final_flip_bw,
+		bool *ImmediateFlipSupportedForPipe);
+static double CalculateWriteBackDelay(
+		enum source_format_class WritebackPixelFormat,
+		double WritebackHRatio,
+		double WritebackVRatio,
+		unsigned int WritebackLumaHTaps,
+		unsigned int WritebackLumaVTaps,
+		unsigned int WritebackChromaHTaps,
+		unsigned int WritebackChromaVTaps,
+		unsigned int WritebackDestinationWidth);
+static void CalculateWatermarksAndDRAMSpeedChangeSupport(
+		struct display_mode_lib *mode_lib,
+		unsigned int PrefetchMode,
+		unsigned int NumberOfActivePlanes,
+		unsigned int MaxLineBufferLines,
+		unsigned int LineBufferSize,
+		unsigned int DPPOutputBufferPixels,
+		unsigned int DETBufferSizeInKByte,
+		unsigned int WritebackInterfaceLumaBufferSize,
+		unsigned int WritebackInterfaceChromaBufferSize,
+		double DCFCLK,
+		double UrgentOutOfOrderReturn,
+		double ReturnBW,
+		bool GPUVMEnable,
+		int dpte_group_bytes[],
+		unsigned int MetaChunkSize,
+		double UrgentLatency,
+		double ExtraLatency,
+		double WritebackLatency,
+		double WritebackChunkSize,
+		double SOCCLK,
+		double DRAMClockChangeLatency,
+		double SRExitTime,
+		double SREnterPlusExitTime,
+		double DCFCLKDeepSleep,
+		int DPPPerPlane[],
+		bool DCCEnable[],
+		double DPPCLK[],
+		double SwathWidthSingleDPPY[],
+		unsigned int SwathHeightY[],
+		double ReadBandwidthPlaneLuma[],
+		unsigned int SwathHeightC[],
+		double ReadBandwidthPlaneChroma[],
+		unsigned int LBBitPerPixel[],
+		double SwathWidthY[],
+		double HRatio[],
+		unsigned int vtaps[],
+		unsigned int VTAPsChroma[],
+		double VRatio[],
+		unsigned int HTotal[],
+		double PixelClock[],
+		unsigned int BlendingAndTiming[],
+		double BytePerPixelDETY[],
+		double BytePerPixelDETC[],
+		bool WritebackEnable[],
+		enum source_format_class WritebackPixelFormat[],
+		double WritebackDestinationWidth[],
+		double WritebackDestinationHeight[],
+		double WritebackSourceHeight[],
+		enum clock_change_support *DRAMClockChangeSupport,
+		double *UrgentWatermark,
+		double *WritebackUrgentWatermark,
+		double *DRAMClockChangeWatermark,
+		double *WritebackDRAMClockChangeWatermark,
+		double *StutterExitWatermark,
+		double *StutterEnterPlusExitWatermark,
+		double *MinActiveDRAMClockChangeLatencySupported);
+static void CalculateDCFCLKDeepSleep(
+		struct display_mode_lib *mode_lib,
+		unsigned int NumberOfActivePlanes,
+		double BytePerPixelDETY[],
+		double BytePerPixelDETC[],
+		double VRatio[],
+		double SwathWidthY[],
+		int DPPPerPlane[],
+		double HRatio[],
+		double PixelClock[],
+		double PSCL_THROUGHPUT[],
+		double PSCL_THROUGHPUT_CHROMA[],
+		double DPPCLK[],
+		double *DCFCLKDeepSleep);
+static void CalculateDETBufferSize(
+		unsigned int DETBufferSizeInKByte,
+		unsigned int SwathHeightY,
+		unsigned int SwathHeightC,
+		unsigned int *DETBufferSizeY,
+		unsigned int *DETBufferSizeC);
+static void CalculateUrgentBurstFactor(
+		unsigned int DETBufferSizeInKByte,
+		unsigned int SwathHeightY,
+		unsigned int SwathHeightC,
+		unsigned int SwathWidthY,
+		double LineTime,
+		double UrgentLatency,
+		double CursorBufferSize,
+		unsigned int CursorWidth,
+		unsigned int CursorBPP,
+		double VRatio,
+		double VRatioPreY,
+		double VRatioPreC,
+		double BytePerPixelInDETY,
+		double BytePerPixelInDETC,
+		double *UrgentBurstFactorCursor,
+		double *UrgentBurstFactorCursorPre,
+		double *UrgentBurstFactorLuma,
+		double *UrgentBurstFactorLumaPre,
+		double *UrgentBurstFactorChroma,
+		double *UrgentBurstFactorChromaPre,
+		unsigned int *NotEnoughUrgentLatencyHiding,
+		unsigned int *NotEnoughUrgentLatencyHidingPre);
+
+static void CalculatePixelDeliveryTimes(
+		unsigned int           NumberOfActivePlanes,
+		double                 VRatio[],
+		double                 VRatioPrefetchY[],
+		double                 VRatioPrefetchC[],
+		unsigned int           swath_width_luma_ub[],
+		unsigned int           swath_width_chroma_ub[],
+		int                    DPPPerPlane[],
+		double                 HRatio[],
+		double                 PixelClock[],
+		double                 PSCL_THROUGHPUT[],
+		double                 PSCL_THROUGHPUT_CHROMA[],
+		double                 DPPCLK[],
+		double                 BytePerPixelDETC[],
+		enum scan_direction_class SourceScan[],
+		unsigned int           BlockWidth256BytesY[],
+		unsigned int           BlockHeight256BytesY[],
+		unsigned int           BlockWidth256BytesC[],
+		unsigned int           BlockHeight256BytesC[],
+		double                 DisplayPipeLineDeliveryTimeLuma[],
+		double                 DisplayPipeLineDeliveryTimeChroma[],
+		double                 DisplayPipeLineDeliveryTimeLumaPrefetch[],
+		double                 DisplayPipeLineDeliveryTimeChromaPrefetch[],
+		double                 DisplayPipeRequestDeliveryTimeLuma[],
+		double                 DisplayPipeRequestDeliveryTimeChroma[],
+		double                 DisplayPipeRequestDeliveryTimeLumaPrefetch[],
+		double                 DisplayPipeRequestDeliveryTimeChromaPrefetch[]);
+
+static void CalculateMetaAndPTETimes(
+		unsigned int           NumberOfActivePlanes,
+		bool                   GPUVMEnable,
+		unsigned int           MetaChunkSize,
+		unsigned int           MinMetaChunkSizeBytes,
+		unsigned int           GPUVMMaxPageTableLevels,
+		unsigned int           HTotal[],
+		double                 VRatio[],
+		double                 VRatioPrefetchY[],
+		double                 VRatioPrefetchC[],
+		double                 DestinationLinesToRequestRowInVBlank[],
+		double                 DestinationLinesToRequestRowInImmediateFlip[],
+		double                 DestinationLinesToRequestVMInVBlank[],
+		double                 DestinationLinesToRequestVMInImmediateFlip[],
+		bool                   DCCEnable[],
+		double                 PixelClock[],
+		double                 BytePerPixelDETY[],
+		double                 BytePerPixelDETC[],
+		enum scan_direction_class SourceScan[],
+		unsigned int           dpte_row_height[],
+		unsigned int           dpte_row_height_chroma[],
+		unsigned int           meta_row_width[],
+		unsigned int           meta_row_height[],
+		unsigned int           meta_req_width[],
+		unsigned int           meta_req_height[],
+		int                   dpte_group_bytes[],
+		unsigned int           PTERequestSizeY[],
+		unsigned int           PTERequestSizeC[],
+		unsigned int           PixelPTEReqWidthY[],
+		unsigned int           PixelPTEReqHeightY[],
+		unsigned int           PixelPTEReqWidthC[],
+		unsigned int           PixelPTEReqHeightC[],
+		unsigned int           dpte_row_width_luma_ub[],
+		unsigned int           dpte_row_width_chroma_ub[],
+		unsigned int           vm_group_bytes[],
+		unsigned int           dpde0_bytes_per_frame_ub_l[],
+		unsigned int           dpde0_bytes_per_frame_ub_c[],
+		unsigned int           meta_pte_bytes_per_frame_ub_l[],
+		unsigned int           meta_pte_bytes_per_frame_ub_c[],
+		double                 DST_Y_PER_PTE_ROW_NOM_L[],
+		double                 DST_Y_PER_PTE_ROW_NOM_C[],
+		double                 DST_Y_PER_META_ROW_NOM_L[],
+		double                 TimePerMetaChunkNominal[],
+		double                 TimePerMetaChunkVBlank[],
+		double                 TimePerMetaChunkFlip[],
+		double                 time_per_pte_group_nom_luma[],
+		double                 time_per_pte_group_vblank_luma[],
+		double                 time_per_pte_group_flip_luma[],
+		double                 time_per_pte_group_nom_chroma[],
+		double                 time_per_pte_group_vblank_chroma[],
+		double                 time_per_pte_group_flip_chroma[],
+		double                 TimePerVMGroupVBlank[],
+		double                 TimePerVMGroupFlip[],
+		double                 TimePerVMRequestVBlank[],
+		double                 TimePerVMRequestFlip[]);
+
+static double CalculateExtraLatency(
+		double UrgentRoundTripAndOutOfOrderLatency,
+		int TotalNumberOfActiveDPP,
+		int PixelChunkSizeInKByte,
+		int TotalNumberOfDCCActiveDPP,
+		int MetaChunkSize,
+		double ReturnBW,
+		bool GPUVMEnable,
+		bool HostVMEnable,
+		int NumberOfActivePlanes,
+		int NumberOfDPP[],
+		int dpte_group_bytes[],
+		double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData,
+		double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly,
+		int HostVMMaxPageTableLevels,
+		int HostVMCachedPageTableLevels);
+
+void dml21_recalculate(struct display_mode_lib *mode_lib)
+{
+	ModeSupportAndSystemConfiguration(mode_lib);
+	PixelClockAdjustmentForProgressiveToInterlaceUnit(mode_lib);
+	DisplayPipeConfiguration(mode_lib);
+	DISPCLKDPPCLKDCFCLKDeepSleepPrefetchParametersWatermarksAndPerformanceCalculation(mode_lib);
+}
+
+static unsigned int dscceComputeDelay(
+		unsigned int bpc,
+		double bpp,
+		unsigned int sliceWidth,
+		unsigned int numSlices,
+		enum output_format_class pixelFormat)
+{
+	// valid bpc         = source bits per component in the set of {8, 10, 12}
+	// valid bpp         = increments of 1/16 of a bit
+	//                    min = 6/7/8 in N420/N422/444, respectively
+	//                    max = such that compression is 1:1
+	//valid sliceWidth  = number of pixels per slice line, must be less than or equal to 5184/numSlices (or 4096/numSlices in 420 mode)
+	//valid numSlices   = number of slices in the horiziontal direction per DSC engine in the set of {1, 2, 3, 4}
+	//valid pixelFormat = pixel/color format in the set of {:N444_RGB, :S422, :N422, :N420}
+
+	// fixed value
+	unsigned int rcModelSize = 8192;
+
+	// N422/N420 operate at 2 pixels per clock
+	unsigned int pixelsPerClock, lstall, D, initalXmitDelay, w, S, ix, wx, p, l0, a, ax, l,
+			Delay, pixels;
+
+	if (pixelFormat == dm_n422 || pixelFormat == dm_420)
+		pixelsPerClock = 2;
+	// #all other modes operate at 1 pixel per clock
+	else
+		pixelsPerClock = 1;
+
+	//initial transmit delay as per PPS
+	initalXmitDelay = dml_round(rcModelSize / 2.0 / bpp / pixelsPerClock);
+
+	//compute ssm delay
+	if (bpc == 8)
+		D = 81;
+	else if (bpc == 10)
+		D = 89;
+	else
+		D = 113;
+
+	//divide by pixel per cycle to compute slice width as seen by DSC
+	w = sliceWidth / pixelsPerClock;
+
+	//422 mode has an additional cycle of delay
+	if (pixelFormat == dm_s422)
+		S = 1;
+	else
+		S = 0;
+
+	//main calculation for the dscce
+	ix = initalXmitDelay + 45;
+	wx = (w + 2) / 3;
+	p = 3 * wx - w;
+	l0 = ix / w;
+	a = ix + p * l0;
+	ax = (a + 2) / 3 + D + 6 + 1;
+	l = (ax + wx - 1) / wx;
+	if ((ix % w) == 0 && p != 0)
+		lstall = 1;
+	else
+		lstall = 0;
+	Delay = l * wx * (numSlices - 1) + ax + S + lstall + 22;
+
+	//dsc processes 3 pixel containers per cycle and a container can contain 1 or 2 pixels
+	pixels = Delay * 3 * pixelsPerClock;
+	return pixels;
+}
+
+static unsigned int dscComputeDelay(enum output_format_class pixelFormat)
+{
+	unsigned int Delay = 0;
+
+	if (pixelFormat == dm_420) {
+		//   sfr
+		Delay = Delay + 2;
+		//   dsccif
+		Delay = Delay + 0;
+		//   dscc - input deserializer
+		Delay = Delay + 3;
+		//   dscc gets pixels every other cycle
+		Delay = Delay + 2;
+		//   dscc - input cdc fifo
+		Delay = Delay + 12;
+		//   dscc gets pixels every other cycle
+		Delay = Delay + 13;
+		//   dscc - cdc uncertainty
+		Delay = Delay + 2;
+		//   dscc - output cdc fifo
+		Delay = Delay + 7;
+		//   dscc gets pixels every other cycle
+		Delay = Delay + 3;
+		//   dscc - cdc uncertainty
+		Delay = Delay + 2;
+		//   dscc - output serializer
+		Delay = Delay + 1;
+		//   sft
+		Delay = Delay + 1;
+	} else if (pixelFormat == dm_n422) {
+		//   sfr
+		Delay = Delay + 2;
+		//   dsccif
+		Delay = Delay + 1;
+		//   dscc - input deserializer
+		Delay = Delay + 5;
+		//  dscc - input cdc fifo
+		Delay = Delay + 25;
+		//   dscc - cdc uncertainty
+		Delay = Delay + 2;
+		//   dscc - output cdc fifo
+		Delay = Delay + 10;
+		//   dscc - cdc uncertainty
+		Delay = Delay + 2;
+		//   dscc - output serializer
+		Delay = Delay + 1;
+		//   sft
+		Delay = Delay + 1;
+	} else {
+		//   sfr
+		Delay = Delay + 2;
+		//   dsccif
+		Delay = Delay + 0;
+		//   dscc - input deserializer
+		Delay = Delay + 3;
+		//   dscc - input cdc fifo
+		Delay = Delay + 12;
+		//   dscc - cdc uncertainty
+		Delay = Delay + 2;
+		//   dscc - output cdc fifo
+		Delay = Delay + 7;
+		//   dscc - output serializer
+		Delay = Delay + 1;
+		//   dscc - cdc uncertainty
+		Delay = Delay + 2;
+		//   sft
+		Delay = Delay + 1;
+	}
+
+	return Delay;
+}
+
+static bool CalculatePrefetchSchedule(
+		struct display_mode_lib *mode_lib,
+		double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData,
+		double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly,
+		Pipe *myPipe,
+		unsigned int DSCDelay,
+		double DPPCLKDelaySubtotal,
+		double DPPCLKDelaySCL,
+		double DPPCLKDelaySCLLBOnly,
+		double DPPCLKDelayCNVCFormater,
+		double DPPCLKDelayCNVCCursor,
+		double DISPCLKDelaySubtotal,
+		unsigned int ScalerRecoutWidth,
+		enum output_format_class OutputFormat,
+		unsigned int MaxInterDCNTileRepeaters,
+		unsigned int VStartup,
+		unsigned int MaxVStartup,
+		unsigned int GPUVMPageTableLevels,
+		bool GPUVMEnable,
+		HostVM *myHostVM,
+		bool DynamicMetadataEnable,
+		int DynamicMetadataLinesBeforeActiveRequired,
+		unsigned int DynamicMetadataTransmittedBytes,
+		bool DCCEnable,
+		double UrgentLatency,
+		double UrgentExtraLatency,
+		double TCalc,
+		unsigned int PDEAndMetaPTEBytesFrame,
+		unsigned int MetaRowByte,
+		unsigned int PixelPTEBytesPerRow,
+		double PrefetchSourceLinesY,
+		unsigned int SwathWidthY,
+		double BytePerPixelDETY,
+		double VInitPreFillY,
+		unsigned int MaxNumSwathY,
+		double PrefetchSourceLinesC,
+		double BytePerPixelDETC,
+		double VInitPreFillC,
+		unsigned int MaxNumSwathC,
+		unsigned int SwathHeightY,
+		unsigned int SwathHeightC,
+		double TWait,
+		bool XFCEnabled,
+		double XFCRemoteSurfaceFlipDelay,
+		bool ProgressiveToInterlaceUnitInOPP,
+		double *DSTXAfterScaler,
+		double *DSTYAfterScaler,
+		double *DestinationLinesForPrefetch,
+		double *PrefetchBandwidth,
+		double *DestinationLinesToRequestVMInVBlank,
+		double *DestinationLinesToRequestRowInVBlank,
+		double *VRatioPrefetchY,
+		double *VRatioPrefetchC,
+		double *RequiredPrefetchPixDataBWLuma,
+		double *RequiredPrefetchPixDataBWChroma,
+		unsigned int *VStartupRequiredWhenNotEnoughTimeForDynamicMetadata,
+		double *Tno_bw,
+		double *prefetch_vmrow_bw,
+		unsigned int *swath_width_luma_ub,
+		unsigned int *swath_width_chroma_ub,
+		unsigned int *VUpdateOffsetPix,
+		double *VUpdateWidthPix,
+		double *VReadyOffsetPix)
+{
+	bool MyError = false;
+	unsigned int DPPCycles, DISPCLKCycles;
+	double DSTTotalPixelsAfterScaler, TotalRepeaterDelayTime;
+	double Tdm, LineTime, Tsetup;
+	double dst_y_prefetch_equ;
+	double Tsw_oto;
+	double prefetch_bw_oto;
+	double Tvm_oto;
+	double Tr0_oto;
+	double Tvm_oto_lines;
+	double Tr0_oto_lines;
+	double Tsw_oto_lines;
+	double dst_y_prefetch_oto;
+	double TimeForFetchingMetaPTE = 0;
+	double TimeForFetchingRowInVBlank = 0;
+	double LinesToRequestPrefetchPixelData = 0;
+	double HostVMInefficiencyFactor;
+	unsigned int HostVMDynamicLevels;
+
+	if (GPUVMEnable == true && myHostVM->Enable == true) {
+		HostVMInefficiencyFactor =
+				PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData
+						/ PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly;
+		HostVMDynamicLevels = myHostVM->MaxPageTableLevels
+				- myHostVM->CachedPageTableLevels;
+	} else {
+		HostVMInefficiencyFactor = 1;
+		HostVMDynamicLevels = 0;
+	}
+
+	if (myPipe->ScalerEnabled)
+		DPPCycles = DPPCLKDelaySubtotal + DPPCLKDelaySCL;
+	else
+		DPPCycles = DPPCLKDelaySubtotal + DPPCLKDelaySCLLBOnly;
+
+	DPPCycles = DPPCycles + DPPCLKDelayCNVCFormater + myPipe->NumberOfCursors * DPPCLKDelayCNVCCursor;
+
+	DISPCLKCycles = DISPCLKDelaySubtotal;
+
+	if (myPipe->DPPCLK == 0.0 || myPipe->DISPCLK == 0.0)
+		return true;
+
+	*DSTXAfterScaler = DPPCycles * myPipe->PixelClock / myPipe->DPPCLK
+			+ DISPCLKCycles * myPipe->PixelClock / myPipe->DISPCLK + DSCDelay;
+
+	if (myPipe->DPPPerPlane > 1)
+		*DSTXAfterScaler = *DSTXAfterScaler + ScalerRecoutWidth;
+
+	if (OutputFormat == dm_420 || (myPipe->InterlaceEnable && ProgressiveToInterlaceUnitInOPP))
+		*DSTYAfterScaler = 1;
+	else
+		*DSTYAfterScaler = 0;
+
+	DSTTotalPixelsAfterScaler = ((double) (*DSTYAfterScaler * myPipe->HTotal)) + *DSTXAfterScaler;
+	*DSTYAfterScaler = dml_floor(DSTTotalPixelsAfterScaler / myPipe->HTotal, 1);
+	*DSTXAfterScaler = DSTTotalPixelsAfterScaler - ((double) (*DSTYAfterScaler * myPipe->HTotal));
+
+	*VUpdateOffsetPix = dml_ceil(myPipe->HTotal / 4.0, 1);
+	TotalRepeaterDelayTime = MaxInterDCNTileRepeaters * (2.0 / myPipe->DPPCLK + 3.0 / myPipe->DISPCLK);
+	*VUpdateWidthPix = (14.0 / myPipe->DCFCLKDeepSleep + 12.0 / myPipe->DPPCLK + TotalRepeaterDelayTime)
+			* myPipe->PixelClock;
+
+	*VReadyOffsetPix = dml_max(
+			150.0 / myPipe->DPPCLK,
+			TotalRepeaterDelayTime + 20.0 / myPipe->DCFCLKDeepSleep + 10.0 / myPipe->DPPCLK)
+			* myPipe->PixelClock;
+
+	Tsetup = (double) (*VUpdateOffsetPix + *VUpdateWidthPix + *VReadyOffsetPix) / myPipe->PixelClock;
+
+	LineTime = (double) myPipe->HTotal / myPipe->PixelClock;
+
+	if (DynamicMetadataEnable) {
+		double Tdmbf, Tdmec, Tdmsks;
+
+		Tdm = dml_max(0.0, UrgentExtraLatency - TCalc);
+		Tdmbf = DynamicMetadataTransmittedBytes / 4.0 / myPipe->DISPCLK;
+		Tdmec = LineTime;
+		if (DynamicMetadataLinesBeforeActiveRequired == -1)
+			Tdmsks = myPipe->VBlank * LineTime / 2.0;
+		else
+			Tdmsks = DynamicMetadataLinesBeforeActiveRequired * LineTime;
+		if (myPipe->InterlaceEnable && !ProgressiveToInterlaceUnitInOPP)
+			Tdmsks = Tdmsks / 2;
+		if (VStartup * LineTime
+				< Tsetup + TWait + UrgentExtraLatency + Tdmbf + Tdmec + Tdmsks) {
+			MyError = true;
+			*VStartupRequiredWhenNotEnoughTimeForDynamicMetadata = (Tsetup + TWait
+					+ UrgentExtraLatency + Tdmbf + Tdmec + Tdmsks) / LineTime;
+		} else
+			*VStartupRequiredWhenNotEnoughTimeForDynamicMetadata = 0.0;
+	} else
+		Tdm = 0;
+
+	if (GPUVMEnable) {
+		if (GPUVMPageTableLevels >= 3)
+			*Tno_bw = UrgentExtraLatency + UrgentLatency * ((GPUVMPageTableLevels - 2) * (myHostVM->MaxPageTableLevels + 1) - 1);
+		else
+			*Tno_bw = 0;
+	} else if (!DCCEnable)
+		*Tno_bw = LineTime;
+	else
+		*Tno_bw = LineTime / 4;
+
+	dst_y_prefetch_equ = VStartup - dml_max(TCalc + TWait, XFCRemoteSurfaceFlipDelay) / LineTime
+			- (Tsetup + Tdm) / LineTime
+			- (*DSTYAfterScaler + *DSTXAfterScaler / myPipe->HTotal);
+
+	Tsw_oto = dml_max(PrefetchSourceLinesY, PrefetchSourceLinesC) * LineTime;
+
+	if (myPipe->SourceScan == dm_horz) {
+		*swath_width_luma_ub = dml_ceil(SwathWidthY - 1, myPipe->BlockWidth256BytesY) + myPipe->BlockWidth256BytesY;
+		if (myPipe->BlockWidth256BytesC > 0)
+			*swath_width_chroma_ub = dml_ceil(SwathWidthY / 2 - 1, myPipe->BlockWidth256BytesC) + myPipe->BlockWidth256BytesC;
+	} else {
+		*swath_width_luma_ub = dml_ceil(SwathWidthY - 1, myPipe->BlockHeight256BytesY) + myPipe->BlockHeight256BytesY;
+		if (myPipe->BlockHeight256BytesC > 0)
+			*swath_width_chroma_ub = dml_ceil(SwathWidthY / 2 - 1, myPipe->BlockHeight256BytesC) + myPipe->BlockHeight256BytesC;
+	}
+
+	prefetch_bw_oto = (PrefetchSourceLinesY * *swath_width_luma_ub * dml_ceil(BytePerPixelDETY, 1) + PrefetchSourceLinesC * *swath_width_chroma_ub * dml_ceil(BytePerPixelDETC, 2)) / Tsw_oto;
+
+
+	if (GPUVMEnable == true) {
+		Tvm_oto = dml_max(*Tno_bw + PDEAndMetaPTEBytesFrame * HostVMInefficiencyFactor / prefetch_bw_oto,
+				dml_max(UrgentExtraLatency + UrgentLatency * (GPUVMPageTableLevels * (HostVMDynamicLevels + 1) - 1),
+					LineTime / 4.0));
+	} else
+		Tvm_oto = LineTime / 4.0;
+
+	if ((GPUVMEnable == true || DCCEnable == true)) {
+		Tr0_oto = dml_max(
+				(MetaRowByte + PixelPTEBytesPerRow * HostVMInefficiencyFactor) / prefetch_bw_oto,
+				dml_max(UrgentLatency * (HostVMDynamicLevels + 1), dml_max(LineTime - Tvm_oto, LineTime / 4)));
+	} else
+		Tr0_oto = (LineTime - Tvm_oto) / 2.0;
+
+	Tvm_oto_lines = dml_ceil(4 * Tvm_oto / LineTime, 1) / 4.0;
+	Tr0_oto_lines = dml_ceil(4 * Tr0_oto / LineTime, 1) / 4.0;
+	Tsw_oto_lines = dml_ceil(4 * Tsw_oto / LineTime, 1) / 4.0;
+	dst_y_prefetch_oto = Tvm_oto_lines + 2 * Tr0_oto_lines + Tsw_oto_lines + 0.75;
+
+	dst_y_prefetch_equ = dml_floor(4.0 * (dst_y_prefetch_equ + 0.125), 1) / 4.0;
+
+	if (dst_y_prefetch_oto < dst_y_prefetch_equ)
+		*DestinationLinesForPrefetch = dst_y_prefetch_oto;
+	else
+		*DestinationLinesForPrefetch = dst_y_prefetch_equ;
+
+	// Limit to prevent overflow in DST_Y_PREFETCH register
+	*DestinationLinesForPrefetch = dml_min(*DestinationLinesForPrefetch, 63.75);
+
+	dml_print("DML: VStartup: %d\n", VStartup);
+	dml_print("DML: TCalc: %f\n", TCalc);
+	dml_print("DML: TWait: %f\n", TWait);
+	dml_print("DML: XFCRemoteSurfaceFlipDelay: %f\n", XFCRemoteSurfaceFlipDelay);
+	dml_print("DML: LineTime: %f\n", LineTime);
+	dml_print("DML: Tsetup: %f\n", Tsetup);
+	dml_print("DML: Tdm: %f\n", Tdm);
+	dml_print("DML: DSTYAfterScaler: %f\n", *DSTYAfterScaler);
+	dml_print("DML: DSTXAfterScaler: %f\n", *DSTXAfterScaler);
+	dml_print("DML: HTotal: %d\n", myPipe->HTotal);
+
+	*PrefetchBandwidth = 0;
+	*DestinationLinesToRequestVMInVBlank = 0;
+	*DestinationLinesToRequestRowInVBlank = 0;
+	*VRatioPrefetchY = 0;
+	*VRatioPrefetchC = 0;
+	*RequiredPrefetchPixDataBWLuma = 0;
+	if (*DestinationLinesForPrefetch > 1) {
+		double PrefetchBandwidth1 = (PDEAndMetaPTEBytesFrame * HostVMInefficiencyFactor + 2 * MetaRowByte
+				+ 2 * PixelPTEBytesPerRow * HostVMInefficiencyFactor
+				+ PrefetchSourceLinesY * *swath_width_luma_ub * dml_ceil(BytePerPixelDETY, 1)
+				+ PrefetchSourceLinesC * *swath_width_chroma_ub * dml_ceil(BytePerPixelDETC, 2))
+				/ (*DestinationLinesForPrefetch * LineTime - *Tno_bw);
+
+		double PrefetchBandwidth2 = (PDEAndMetaPTEBytesFrame *
+				HostVMInefficiencyFactor + PrefetchSourceLinesY *
+				*swath_width_luma_ub * dml_ceil(BytePerPixelDETY, 1) +
+				PrefetchSourceLinesC * *swath_width_chroma_ub *
+				dml_ceil(BytePerPixelDETC, 2)) /
+				(*DestinationLinesForPrefetch * LineTime - *Tno_bw - 2 *
+				UrgentLatency * (1 + HostVMDynamicLevels));
+
+		double PrefetchBandwidth3 = (2 * MetaRowByte + 2 * PixelPTEBytesPerRow
+				* HostVMInefficiencyFactor + PrefetchSourceLinesY *
+				*swath_width_luma_ub * dml_ceil(BytePerPixelDETY, 1) +
+				PrefetchSourceLinesC * *swath_width_chroma_ub *
+				dml_ceil(BytePerPixelDETC, 2)) /
+				(*DestinationLinesForPrefetch * LineTime -
+				UrgentExtraLatency - UrgentLatency * (GPUVMPageTableLevels
+				* (HostVMDynamicLevels + 1) - 1));
+
+		double PrefetchBandwidth4 = (PrefetchSourceLinesY * *swath_width_luma_ub *
+				dml_ceil(BytePerPixelDETY, 1) + PrefetchSourceLinesC *
+				*swath_width_chroma_ub * dml_ceil(BytePerPixelDETC, 2)) /
+				(*DestinationLinesForPrefetch * LineTime -
+				UrgentExtraLatency - UrgentLatency * (GPUVMPageTableLevels
+				* (HostVMDynamicLevels + 1) - 1) - 2 * UrgentLatency *
+				(1 + HostVMDynamicLevels));
+
+		if (VStartup == MaxVStartup && (PrefetchBandwidth1 > 4 * prefetch_bw_oto) && (*DestinationLinesForPrefetch - dml_ceil(Tsw_oto_lines, 1) / 4.0 - 0.75) * LineTime - *Tno_bw > 0) {
+			PrefetchBandwidth1 = (PDEAndMetaPTEBytesFrame * HostVMInefficiencyFactor + 2 * MetaRowByte + 2 * PixelPTEBytesPerRow * HostVMInefficiencyFactor) / ((*DestinationLinesForPrefetch - dml_ceil(Tsw_oto_lines, 1) / 4.0 - 0.75) * LineTime - *Tno_bw);
+		}
+		if (*Tno_bw + PDEAndMetaPTEBytesFrame * HostVMInefficiencyFactor / PrefetchBandwidth1 >= UrgentExtraLatency + UrgentLatency * (GPUVMPageTableLevels * (HostVMDynamicLevels + 1) - 1) && (MetaRowByte + PixelPTEBytesPerRow * HostVMInefficiencyFactor) / PrefetchBandwidth1 >= UrgentLatency * (1 + HostVMDynamicLevels)) {
+			*PrefetchBandwidth = PrefetchBandwidth1;
+		} else if (*Tno_bw + PDEAndMetaPTEBytesFrame * HostVMInefficiencyFactor / PrefetchBandwidth2 >= UrgentExtraLatency + UrgentLatency * (GPUVMPageTableLevels * (HostVMDynamicLevels + 1) - 1) && (MetaRowByte + PixelPTEBytesPerRow * HostVMInefficiencyFactor) / PrefetchBandwidth2 < UrgentLatency * (1 + HostVMDynamicLevels)) {
+			*PrefetchBandwidth = PrefetchBandwidth2;
+		} else if (*Tno_bw + PDEAndMetaPTEBytesFrame * HostVMInefficiencyFactor / PrefetchBandwidth3 < UrgentExtraLatency + UrgentLatency * (GPUVMPageTableLevels * (HostVMDynamicLevels + 1) - 1) && (MetaRowByte + PixelPTEBytesPerRow * HostVMInefficiencyFactor) / PrefetchBandwidth3 >= UrgentLatency * (1 + HostVMDynamicLevels)) {
+			*PrefetchBandwidth = PrefetchBandwidth3;
+		} else {
+			*PrefetchBandwidth = PrefetchBandwidth4;
+		}
+
+		if (GPUVMEnable) {
+			TimeForFetchingMetaPTE = dml_max(*Tno_bw + (double) PDEAndMetaPTEBytesFrame * HostVMInefficiencyFactor / *PrefetchBandwidth,
+					dml_max(UrgentExtraLatency + UrgentLatency * (GPUVMPageTableLevels * (HostVMDynamicLevels + 1) - 1), LineTime / 4));
+		} else {
+// 5/30/2018 - This was an optimization requested from Sy but now NumberOfCursors is no longer a factor
+//             so if this needs to be reinstated, then it should be officially done in the VBA code as well.
+//			if (mode_lib->NumberOfCursors > 0 || XFCEnabled)
+				TimeForFetchingMetaPTE = LineTime / 4;
+//			else
+//				TimeForFetchingMetaPTE = 0.0;
+		}
+
+		if ((GPUVMEnable == true || DCCEnable == true)) {
+			TimeForFetchingRowInVBlank =
+					dml_max(
+							(MetaRowByte + PixelPTEBytesPerRow * HostVMInefficiencyFactor)
+									/ *PrefetchBandwidth,
+							dml_max(
+									UrgentLatency * (1 + HostVMDynamicLevels),
+									dml_max(
+											(LineTime
+													- TimeForFetchingMetaPTE) / 2.0,
+											LineTime
+													/ 4.0)));
+		} else {
+// See note above dated 5/30/2018
+//			if (NumberOfCursors > 0 || XFCEnabled)
+				TimeForFetchingRowInVBlank = (LineTime - TimeForFetchingMetaPTE) / 2.0;
+//			else // TODO: Did someone else add this??
+//				TimeForFetchingRowInVBlank = 0.0;
+		}
+
+		*DestinationLinesToRequestVMInVBlank = dml_ceil(4.0 * TimeForFetchingMetaPTE / LineTime, 1.0) / 4.0;
+
+		*DestinationLinesToRequestRowInVBlank = dml_ceil(4.0 * TimeForFetchingRowInVBlank / LineTime, 1.0) / 4.0;
+
+		LinesToRequestPrefetchPixelData = *DestinationLinesForPrefetch
+// See note above dated 5/30/2018
+//						- ((NumberOfCursors > 0 || GPUVMEnable || DCCEnable) ?
+						- ((GPUVMEnable || DCCEnable) ?
+								(*DestinationLinesToRequestVMInVBlank + 2 * *DestinationLinesToRequestRowInVBlank) :
+								0.0); // TODO: Did someone else add this??
+
+		if (LinesToRequestPrefetchPixelData > 0) {
+
+			*VRatioPrefetchY = (double) PrefetchSourceLinesY
+					/ LinesToRequestPrefetchPixelData;
+			*VRatioPrefetchY = dml_max(*VRatioPrefetchY, 1.0);
+			if ((SwathHeightY > 4) && (VInitPreFillY > 3)) {
+				if (LinesToRequestPrefetchPixelData > (VInitPreFillY - 3.0) / 2.0) {
+					*VRatioPrefetchY =
+							dml_max(
+									(double) PrefetchSourceLinesY
+											/ LinesToRequestPrefetchPixelData,
+									(double) MaxNumSwathY
+											* SwathHeightY
+											/ (LinesToRequestPrefetchPixelData
+													- (VInitPreFillY
+															- 3.0)
+															/ 2.0));
+					*VRatioPrefetchY = dml_max(*VRatioPrefetchY, 1.0);
+				} else {
+					MyError = true;
+					*VRatioPrefetchY = 0;
+				}
+			}
+
+			*VRatioPrefetchC = (double) PrefetchSourceLinesC
+					/ LinesToRequestPrefetchPixelData;
+			*VRatioPrefetchC = dml_max(*VRatioPrefetchC, 1.0);
+
+			if ((SwathHeightC > 4)) {
+				if (LinesToRequestPrefetchPixelData > (VInitPreFillC - 3.0) / 2.0) {
+					*VRatioPrefetchC =
+							dml_max(
+									*VRatioPrefetchC,
+									(double) MaxNumSwathC
+											* SwathHeightC
+											/ (LinesToRequestPrefetchPixelData
+													- (VInitPreFillC
+															- 3.0)
+															/ 2.0));
+					*VRatioPrefetchC = dml_max(*VRatioPrefetchC, 1.0);
+				} else {
+					MyError = true;
+					*VRatioPrefetchC = 0;
+				}
+			}
+
+			*RequiredPrefetchPixDataBWLuma = myPipe->DPPPerPlane
+					* (double) PrefetchSourceLinesY / LinesToRequestPrefetchPixelData
+					* dml_ceil(BytePerPixelDETY, 1)
+					* *swath_width_luma_ub / LineTime;
+			*RequiredPrefetchPixDataBWChroma = myPipe->DPPPerPlane
+					* (double) PrefetchSourceLinesC / LinesToRequestPrefetchPixelData
+					* dml_ceil(BytePerPixelDETC, 2)
+					* *swath_width_chroma_ub / LineTime;
+		} else {
+			MyError = true;
+			*VRatioPrefetchY = 0;
+			*VRatioPrefetchC = 0;
+			*RequiredPrefetchPixDataBWLuma = 0;
+			*RequiredPrefetchPixDataBWChroma = 0;
+		}
+
+		dml_print("DML: Tvm: %fus\n", TimeForFetchingMetaPTE);
+		dml_print("DML: Tr0: %fus\n", TimeForFetchingRowInVBlank);
+		dml_print("DML: Tsw: %fus\n", (double)(*DestinationLinesForPrefetch) * LineTime - TimeForFetchingMetaPTE - TimeForFetchingRowInVBlank);
+		dml_print("DML: Tpre: %fus\n", (double)(*DestinationLinesForPrefetch) * LineTime);
+		dml_print("DML: row_bytes = dpte_row_bytes (per_pipe) = PixelPTEBytesPerRow = : %d\n", PixelPTEBytesPerRow);
+
+	} else {
+		MyError = true;
+	}
+
+	{
+		double prefetch_vm_bw;
+		double prefetch_row_bw;
+
+		if (PDEAndMetaPTEBytesFrame == 0) {
+			prefetch_vm_bw = 0;
+		} else if (*DestinationLinesToRequestVMInVBlank > 0) {
+			prefetch_vm_bw = PDEAndMetaPTEBytesFrame * HostVMInefficiencyFactor / (*DestinationLinesToRequestVMInVBlank * LineTime);
+		} else {
+			prefetch_vm_bw = 0;
+			MyError = true;
+		}
+		if (MetaRowByte + PixelPTEBytesPerRow == 0) {
+			prefetch_row_bw = 0;
+		} else if (*DestinationLinesToRequestRowInVBlank > 0) {
+			prefetch_row_bw = (MetaRowByte + PixelPTEBytesPerRow * HostVMInefficiencyFactor) / (*DestinationLinesToRequestRowInVBlank * LineTime);
+		} else {
+			prefetch_row_bw = 0;
+			MyError = true;
+		}
+
+		*prefetch_vmrow_bw = dml_max(prefetch_vm_bw, prefetch_row_bw);
+	}
+
+	if (MyError) {
+		*PrefetchBandwidth = 0;
+		*DestinationLinesToRequestVMInVBlank = 0;
+		*DestinationLinesToRequestRowInVBlank = 0;
+		*DestinationLinesForPrefetch = 0;
+		*VRatioPrefetchY = 0;
+		*VRatioPrefetchC = 0;
+		*RequiredPrefetchPixDataBWLuma = 0;
+		*RequiredPrefetchPixDataBWChroma = 0;
+	}
+
+	return MyError;
+}
+
+static double RoundToDFSGranularityUp(double Clock, double VCOSpeed)
+{
+	return VCOSpeed * 4 / dml_floor(VCOSpeed * 4 / Clock, 1);
+}
+
+static double RoundToDFSGranularityDown(double Clock, double VCOSpeed)
+{
+	return VCOSpeed * 4 / dml_ceil(VCOSpeed * 4 / Clock, 1);
+}
+
+static double CalculateDCCConfiguration(
+		bool DCCEnabled,
+		bool DCCProgrammingAssumesScanDirectionUnknown,
+		unsigned int ViewportWidth,
+		unsigned int ViewportHeight,
+		unsigned int DETBufferSize,
+		unsigned int RequestHeight256Byte,
+		unsigned int SwathHeight,
+		enum dm_swizzle_mode TilingFormat,
+		unsigned int BytePerPixel,
+		enum scan_direction_class ScanOrientation,
+		unsigned int *MaxUncompressedBlock,
+		unsigned int *MaxCompressedBlock,
+		unsigned int *Independent64ByteBlock)
+{
+	double MaximumDCCCompressionSurface = 0.0;
+	enum {
+		REQ_256Bytes,
+		REQ_128BytesNonContiguous,
+		REQ_128BytesContiguous,
+		REQ_NA
+	} Request = REQ_NA;
+
+	if (DCCEnabled == true) {
+		if (DCCProgrammingAssumesScanDirectionUnknown == true) {
+			if (DETBufferSize >= RequestHeight256Byte * ViewportWidth * BytePerPixel
+					&& DETBufferSize
+							>= 256 / RequestHeight256Byte
+									* ViewportHeight) {
+				Request = REQ_256Bytes;
+			} else if ((DETBufferSize
+					< RequestHeight256Byte * ViewportWidth * BytePerPixel
+					&& (BytePerPixel == 2 || BytePerPixel == 4))
+					|| (DETBufferSize
+							< 256 / RequestHeight256Byte
+									* ViewportHeight
+							&& BytePerPixel == 8
+							&& (TilingFormat == dm_sw_4kb_d
+									|| TilingFormat
+											== dm_sw_4kb_d_x
+									|| TilingFormat
+											== dm_sw_var_d
+									|| TilingFormat
+											== dm_sw_var_d_x
+									|| TilingFormat
+											== dm_sw_64kb_d
+									|| TilingFormat
+											== dm_sw_64kb_d_x
+									|| TilingFormat
+											== dm_sw_64kb_d_t
+									|| TilingFormat
+											== dm_sw_64kb_r_x))) {
+				Request = REQ_128BytesNonContiguous;
+			} else {
+				Request = REQ_128BytesContiguous;
+			}
+		} else {
+			if (BytePerPixel == 1) {
+				if (ScanOrientation == dm_vert || SwathHeight == 16) {
+					Request = REQ_256Bytes;
+				} else {
+					Request = REQ_128BytesContiguous;
+				}
+			} else if (BytePerPixel == 2) {
+				if ((ScanOrientation == dm_vert && SwathHeight == 16) || (ScanOrientation != dm_vert && SwathHeight == 8)) {
+					Request = REQ_256Bytes;
+				} else if (ScanOrientation == dm_vert) {
+					Request = REQ_128BytesContiguous;
+				} else {
+					Request = REQ_128BytesNonContiguous;
+				}
+			} else if (BytePerPixel == 4) {
+				if (SwathHeight == 8) {
+					Request = REQ_256Bytes;
+				} else if (ScanOrientation == dm_vert) {
+					Request = REQ_128BytesContiguous;
+				} else {
+					Request = REQ_128BytesNonContiguous;
+				}
+			} else if (BytePerPixel == 8) {
+				if (TilingFormat == dm_sw_4kb_d || TilingFormat == dm_sw_4kb_d_x
+						|| TilingFormat == dm_sw_var_d
+						|| TilingFormat == dm_sw_var_d_x
+						|| TilingFormat == dm_sw_64kb_d
+						|| TilingFormat == dm_sw_64kb_d_x
+						|| TilingFormat == dm_sw_64kb_d_t
+						|| TilingFormat == dm_sw_64kb_r_x) {
+					if ((ScanOrientation == dm_vert && SwathHeight == 8)
+							|| (ScanOrientation != dm_vert
+									&& SwathHeight == 4)) {
+						Request = REQ_256Bytes;
+					} else if (ScanOrientation != dm_vert) {
+						Request = REQ_128BytesContiguous;
+					} else {
+						Request = REQ_128BytesNonContiguous;
+					}
+				} else {
+					if (ScanOrientation != dm_vert || SwathHeight == 8) {
+						Request = REQ_256Bytes;
+					} else {
+						Request = REQ_128BytesContiguous;
+					}
+				}
+			}
+		}
+	} else {
+		Request = REQ_NA;
+	}
+
+	if (Request == REQ_256Bytes) {
+		*MaxUncompressedBlock = 256;
+		*MaxCompressedBlock = 256;
+		*Independent64ByteBlock = false;
+		MaximumDCCCompressionSurface = 4.0;
+	} else if (Request == REQ_128BytesContiguous) {
+		*MaxUncompressedBlock = 128;
+		*MaxCompressedBlock = 128;
+		*Independent64ByteBlock = false;
+		MaximumDCCCompressionSurface = 2.0;
+	} else if (Request == REQ_128BytesNonContiguous) {
+		*MaxUncompressedBlock = 256;
+		*MaxCompressedBlock = 64;
+		*Independent64ByteBlock = true;
+		MaximumDCCCompressionSurface = 4.0;
+	} else {
+		*MaxUncompressedBlock = 0;
+		*MaxCompressedBlock = 0;
+		*Independent64ByteBlock = 0;
+		MaximumDCCCompressionSurface = 0.0;
+	}
+
+	return MaximumDCCCompressionSurface;
+}
+
+static double CalculatePrefetchSourceLines(
+		struct display_mode_lib *mode_lib,
+		double VRatio,
+		double vtaps,
+		bool Interlace,
+		bool ProgressiveToInterlaceUnitInOPP,
+		unsigned int SwathHeight,
+		unsigned int ViewportYStart,
+		double *VInitPreFill,
+		unsigned int *MaxNumSwath)
+{
+	unsigned int MaxPartialSwath;
+
+	if (ProgressiveToInterlaceUnitInOPP)
+		*VInitPreFill = dml_floor((VRatio + vtaps + 1) / 2.0, 1);
+	else
+		*VInitPreFill = dml_floor((VRatio + vtaps + 1 + Interlace * 0.5 * VRatio) / 2.0, 1);
+
+	if (!mode_lib->vba.IgnoreViewportPositioning) {
+
+		*MaxNumSwath = dml_ceil((*VInitPreFill - 1.0) / SwathHeight, 1) + 1.0;
+
+		if (*VInitPreFill > 1.0)
+			MaxPartialSwath = (unsigned int) (*VInitPreFill - 2) % SwathHeight;
+		else
+			MaxPartialSwath = (unsigned int) (*VInitPreFill + SwathHeight - 2)
+					% SwathHeight;
+		MaxPartialSwath = dml_max(1U, MaxPartialSwath);
+
+	} else {
+
+		if (ViewportYStart != 0)
+			dml_print(
+					"WARNING DML: using viewport y position of 0 even though actual viewport y position is non-zero in prefetch source lines calculation\n");
+
+		*MaxNumSwath = dml_ceil(*VInitPreFill / SwathHeight, 1);
+
+		if (*VInitPreFill > 1.0)
+			MaxPartialSwath = (unsigned int) (*VInitPreFill - 1) % SwathHeight;
+		else
+			MaxPartialSwath = (unsigned int) (*VInitPreFill + SwathHeight - 1)
+					% SwathHeight;
+	}
+
+	return *MaxNumSwath * SwathHeight + MaxPartialSwath;
+}
+
+static unsigned int CalculateVMAndRowBytes(
+		struct display_mode_lib *mode_lib,
+		bool DCCEnable,
+		unsigned int BlockHeight256Bytes,
+		unsigned int BlockWidth256Bytes,
+		enum source_format_class SourcePixelFormat,
+		unsigned int SurfaceTiling,
+		unsigned int BytePerPixel,
+		enum scan_direction_class ScanDirection,
+		unsigned int ViewportWidth,
+		unsigned int ViewportHeight,
+		unsigned int SwathWidth,
+		bool GPUVMEnable,
+		bool HostVMEnable,
+		unsigned int HostVMMaxPageTableLevels,
+		unsigned int HostVMCachedPageTableLevels,
+		unsigned int VMMPageSize,
+		unsigned int PTEBufferSizeInRequests,
+		unsigned int Pitch,
+		unsigned int DCCMetaPitch,
+		unsigned int *MacroTileWidth,
+		unsigned int *MetaRowByte,
+		unsigned int *PixelPTEBytesPerRow,
+		bool *PTEBufferSizeNotExceeded,
+		unsigned int *dpte_row_width_ub,
+		unsigned int *dpte_row_height,
+		unsigned int *MetaRequestWidth,
+		unsigned int *MetaRequestHeight,
+		unsigned int *meta_row_width,
+		unsigned int *meta_row_height,
+		unsigned int *vm_group_bytes,
+		unsigned int *dpte_group_bytes,
+		unsigned int *PixelPTEReqWidth,
+		unsigned int *PixelPTEReqHeight,
+		unsigned int *PTERequestSize,
+		unsigned int *DPDE0BytesFrame,
+		unsigned int *MetaPTEBytesFrame)
+{
+	unsigned int MPDEBytesFrame;
+	unsigned int DCCMetaSurfaceBytes;
+	unsigned int MacroTileSizeBytes;
+	unsigned int MacroTileHeight;
+	unsigned int ExtraDPDEBytesFrame;
+	unsigned int PDEAndMetaPTEBytesFrame;
+	unsigned int PixelPTEReqHeightPTEs = 0;
+
+	if (DCCEnable == true) {
+		*MetaRequestHeight = 8 * BlockHeight256Bytes;
+		*MetaRequestWidth = 8 * BlockWidth256Bytes;
+		if (ScanDirection == dm_horz) {
+			*meta_row_height = *MetaRequestHeight;
+			*meta_row_width = dml_ceil((double) SwathWidth - 1, *MetaRequestWidth)
+					+ *MetaRequestWidth;
+			*MetaRowByte = *meta_row_width * *MetaRequestHeight * BytePerPixel / 256.0;
+		} else {
+			*meta_row_height = *MetaRequestWidth;
+			*meta_row_width = dml_ceil((double) SwathWidth - 1, *MetaRequestHeight)
+					+ *MetaRequestHeight;
+			*MetaRowByte = *meta_row_width * *MetaRequestWidth * BytePerPixel / 256.0;
+		}
+		if (ScanDirection == dm_horz) {
+			DCCMetaSurfaceBytes = DCCMetaPitch
+					* (dml_ceil(ViewportHeight - 1, 64 * BlockHeight256Bytes)
+							+ 64 * BlockHeight256Bytes) * BytePerPixel
+					/ 256;
+		} else {
+			DCCMetaSurfaceBytes = DCCMetaPitch
+					* (dml_ceil(
+							(double) ViewportHeight - 1,
+							64 * BlockHeight256Bytes)
+							+ 64 * BlockHeight256Bytes) * BytePerPixel
+					/ 256;
+		}
+		if (GPUVMEnable == true) {
+			*MetaPTEBytesFrame = (dml_ceil(
+					(double) (DCCMetaSurfaceBytes - VMMPageSize)
+							/ (8 * VMMPageSize),
+					1) + 1) * 64;
+			MPDEBytesFrame = 128 * ((mode_lib->vba.GPUVMMaxPageTableLevels + 1) * (mode_lib->vba.HostVMMaxPageTableLevels + 1) - 2);
+		} else {
+			*MetaPTEBytesFrame = 0;
+			MPDEBytesFrame = 0;
+		}
+	} else {
+		*MetaPTEBytesFrame = 0;
+		MPDEBytesFrame = 0;
+		*MetaRowByte = 0;
+	}
+
+	if (SurfaceTiling == dm_sw_linear || SurfaceTiling == dm_sw_gfx7_2d_thin_gl || SurfaceTiling == dm_sw_gfx7_2d_thin_l_vp) {
+		MacroTileSizeBytes = 256;
+		MacroTileHeight = BlockHeight256Bytes;
+	} else if (SurfaceTiling == dm_sw_4kb_s || SurfaceTiling == dm_sw_4kb_s_x
+			|| SurfaceTiling == dm_sw_4kb_d || SurfaceTiling == dm_sw_4kb_d_x) {
+		MacroTileSizeBytes = 4096;
+		MacroTileHeight = 4 * BlockHeight256Bytes;
+	} else if (SurfaceTiling == dm_sw_64kb_s || SurfaceTiling == dm_sw_64kb_s_t
+			|| SurfaceTiling == dm_sw_64kb_s_x || SurfaceTiling == dm_sw_64kb_d
+			|| SurfaceTiling == dm_sw_64kb_d_t || SurfaceTiling == dm_sw_64kb_d_x
+			|| SurfaceTiling == dm_sw_64kb_r_x) {
+		MacroTileSizeBytes = 65536;
+		MacroTileHeight = 16 * BlockHeight256Bytes;
+	} else {
+		MacroTileSizeBytes = 262144;
+		MacroTileHeight = 32 * BlockHeight256Bytes;
+	}
+	*MacroTileWidth = MacroTileSizeBytes / BytePerPixel / MacroTileHeight;
+
+	if (GPUVMEnable == true && (mode_lib->vba.GPUVMMaxPageTableLevels + 1) * (mode_lib->vba.HostVMMaxPageTableLevels + 1) > 2) {
+		if (ScanDirection == dm_horz) {
+			*DPDE0BytesFrame = 64 * (dml_ceil(((Pitch * (dml_ceil(ViewportHeight - 1, MacroTileHeight) + MacroTileHeight) * BytePerPixel) - MacroTileSizeBytes) / (8 * 2097152), 1) + 1);
+		} else {
+			*DPDE0BytesFrame = 64 * (dml_ceil(((Pitch * (dml_ceil((double) SwathWidth - 1, MacroTileHeight) + MacroTileHeight) * BytePerPixel) - MacroTileSizeBytes) / (8 * 2097152), 1) + 1);
+		}
+		ExtraDPDEBytesFrame = 128 * ((mode_lib->vba.GPUVMMaxPageTableLevels + 1) * (mode_lib->vba.HostVMMaxPageTableLevels + 1) - 3);
+	} else {
+		*DPDE0BytesFrame = 0;
+		ExtraDPDEBytesFrame = 0;
+	}
+
+	PDEAndMetaPTEBytesFrame = *MetaPTEBytesFrame + MPDEBytesFrame + *DPDE0BytesFrame
+			+ ExtraDPDEBytesFrame;
+
+	if (HostVMEnable == true) {
+		PDEAndMetaPTEBytesFrame = PDEAndMetaPTEBytesFrame * (1 + 8 * (HostVMMaxPageTableLevels - HostVMCachedPageTableLevels));
+	}
+
+	if (GPUVMEnable == true) {
+		double FractionOfPTEReturnDrop;
+
+		if (SurfaceTiling == dm_sw_linear) {
+			PixelPTEReqHeightPTEs = 1;
+			*PixelPTEReqHeight = 1;
+			*PixelPTEReqWidth = 8.0 * VMMPageSize / BytePerPixel;
+			*PTERequestSize = 64;
+			FractionOfPTEReturnDrop = 0;
+		} else if (MacroTileSizeBytes == 4096) {
+			PixelPTEReqHeightPTEs = 1;
+			*PixelPTEReqHeight = MacroTileHeight;
+			*PixelPTEReqWidth = 8 * *MacroTileWidth;
+			*PTERequestSize = 64;
+			if (ScanDirection == dm_horz)
+				FractionOfPTEReturnDrop = 0;
+			else
+				FractionOfPTEReturnDrop = 7.0 / 8;
+		} else if (VMMPageSize == 4096 && MacroTileSizeBytes > 4096) {
+			PixelPTEReqHeightPTEs = 16;
+			*PixelPTEReqHeight = 16 * BlockHeight256Bytes;
+			*PixelPTEReqWidth = 16 * BlockWidth256Bytes;
+			*PTERequestSize = 128;
+			FractionOfPTEReturnDrop = 0;
+		} else {
+			PixelPTEReqHeightPTEs = 1;
+			*PixelPTEReqHeight = MacroTileHeight;
+			*PixelPTEReqWidth = 8 * *MacroTileWidth;
+			*PTERequestSize = 64;
+			FractionOfPTEReturnDrop = 0;
+		}
+
+		if (SurfaceTiling == dm_sw_linear) {
+			*dpte_row_height = dml_min(128,
+					1 << (unsigned int) dml_floor(
+						dml_log2(
+							(double) PTEBufferSizeInRequests * *PixelPTEReqWidth / Pitch),
+						1));
+			*dpte_row_width_ub = (dml_ceil((double) (Pitch * *dpte_row_height - 1) / *PixelPTEReqWidth, 1) + 1) * *PixelPTEReqWidth;
+			*PixelPTEBytesPerRow = *dpte_row_width_ub / *PixelPTEReqWidth * *PTERequestSize;
+		} else if (ScanDirection == dm_horz) {
+			*dpte_row_height = *PixelPTEReqHeight;
+			*dpte_row_width_ub = (dml_ceil((double) (SwathWidth - 1) / *PixelPTEReqWidth, 1) + 1) * *PixelPTEReqWidth;
+			*PixelPTEBytesPerRow = *dpte_row_width_ub / *PixelPTEReqWidth * *PTERequestSize;
+		} else {
+			*dpte_row_height = dml_min(*PixelPTEReqWidth, *MacroTileWidth);
+			*dpte_row_width_ub = (dml_ceil((double) (SwathWidth - 1) / *PixelPTEReqHeight, 1) + 1) * *PixelPTEReqHeight;
+			*PixelPTEBytesPerRow = *dpte_row_width_ub / *PixelPTEReqHeight * *PTERequestSize;
+		}
+		if (*PixelPTEBytesPerRow * (1 - FractionOfPTEReturnDrop)
+				<= 64 * PTEBufferSizeInRequests) {
+			*PTEBufferSizeNotExceeded = true;
+		} else {
+			*PTEBufferSizeNotExceeded = false;
+		}
+	} else {
+		*PixelPTEBytesPerRow = 0;
+		*PTEBufferSizeNotExceeded = true;
+	}
+	dml_print("DML: vm_bytes = meta_pte_bytes_per_frame (per_pipe) = MetaPTEBytesFrame = : %d\n", *MetaPTEBytesFrame);
+
+	if (HostVMEnable == true) {
+		*PixelPTEBytesPerRow = *PixelPTEBytesPerRow * (1 + 8 * (HostVMMaxPageTableLevels - HostVMCachedPageTableLevels));
+	}
+
+	if (HostVMEnable == true) {
+		*vm_group_bytes = 512;
+		*dpte_group_bytes = 512;
+	} else if (GPUVMEnable == true) {
+		*vm_group_bytes = 2048;
+		if (SurfaceTiling != dm_sw_linear && PixelPTEReqHeightPTEs == 1 && ScanDirection != dm_horz) {
+			*dpte_group_bytes = 512;
+		} else {
+			*dpte_group_bytes = 2048;
+		}
+	} else {
+		*vm_group_bytes = 0;
+		*dpte_group_bytes = 0;
+	}
+
+	return PDEAndMetaPTEBytesFrame;
+}
+
+static void DISPCLKDPPCLKDCFCLKDeepSleepPrefetchParametersWatermarksAndPerformanceCalculation(
+		struct display_mode_lib *mode_lib)
+{
+	struct vba_vars_st *locals = &mode_lib->vba;
+	unsigned int j, k;
+
+	mode_lib->vba.WritebackDISPCLK = 0.0;
+	mode_lib->vba.DISPCLKWithRamping = 0;
+	mode_lib->vba.DISPCLKWithoutRamping = 0;
+	mode_lib->vba.GlobalDPPCLK = 0.0;
+
+	// DISPCLK and DPPCLK Calculation
+	//
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.WritebackEnable[k]) {
+			mode_lib->vba.WritebackDISPCLK =
+					dml_max(
+							mode_lib->vba.WritebackDISPCLK,
+							CalculateWriteBackDISPCLK(
+									mode_lib->vba.WritebackPixelFormat[k],
+									mode_lib->vba.PixelClock[k],
+									mode_lib->vba.WritebackHRatio[k],
+									mode_lib->vba.WritebackVRatio[k],
+									mode_lib->vba.WritebackLumaHTaps[k],
+									mode_lib->vba.WritebackLumaVTaps[k],
+									mode_lib->vba.WritebackChromaHTaps[k],
+									mode_lib->vba.WritebackChromaVTaps[k],
+									mode_lib->vba.WritebackDestinationWidth[k],
+									mode_lib->vba.HTotal[k],
+									mode_lib->vba.WritebackChromaLineBufferWidth));
+		}
+	}
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.HRatio[k] > 1) {
+			locals->PSCL_THROUGHPUT_LUMA[k] = dml_min(
+					mode_lib->vba.MaxDCHUBToPSCLThroughput,
+					mode_lib->vba.MaxPSCLToLBThroughput
+							* mode_lib->vba.HRatio[k]
+							/ dml_ceil(
+									mode_lib->vba.htaps[k]
+											/ 6.0,
+									1));
+		} else {
+			locals->PSCL_THROUGHPUT_LUMA[k] = dml_min(
+					mode_lib->vba.MaxDCHUBToPSCLThroughput,
+					mode_lib->vba.MaxPSCLToLBThroughput);
+		}
+
+		mode_lib->vba.DPPCLKUsingSingleDPPLuma =
+				mode_lib->vba.PixelClock[k]
+						* dml_max(
+								mode_lib->vba.vtaps[k] / 6.0
+										* dml_min(
+												1.0,
+												mode_lib->vba.HRatio[k]),
+								dml_max(
+										mode_lib->vba.HRatio[k]
+												* mode_lib->vba.VRatio[k]
+												/ locals->PSCL_THROUGHPUT_LUMA[k],
+										1.0));
+
+		if ((mode_lib->vba.htaps[k] > 6 || mode_lib->vba.vtaps[k] > 6)
+				&& mode_lib->vba.DPPCLKUsingSingleDPPLuma
+						< 2 * mode_lib->vba.PixelClock[k]) {
+			mode_lib->vba.DPPCLKUsingSingleDPPLuma = 2 * mode_lib->vba.PixelClock[k];
+		}
+
+		if ((mode_lib->vba.SourcePixelFormat[k] != dm_420_8
+				&& mode_lib->vba.SourcePixelFormat[k] != dm_420_10)) {
+			locals->PSCL_THROUGHPUT_CHROMA[k] = 0.0;
+			locals->DPPCLKUsingSingleDPP[k] =
+					mode_lib->vba.DPPCLKUsingSingleDPPLuma;
+		} else {
+			if (mode_lib->vba.HRatio[k] > 1) {
+				locals->PSCL_THROUGHPUT_CHROMA[k] =
+						dml_min(
+								mode_lib->vba.MaxDCHUBToPSCLThroughput,
+								mode_lib->vba.MaxPSCLToLBThroughput
+										* mode_lib->vba.HRatio[k]
+										/ 2
+										/ dml_ceil(
+												mode_lib->vba.HTAPsChroma[k]
+														/ 6.0,
+												1.0));
+			} else {
+				locals->PSCL_THROUGHPUT_CHROMA[k] = dml_min(
+						mode_lib->vba.MaxDCHUBToPSCLThroughput,
+						mode_lib->vba.MaxPSCLToLBThroughput);
+			}
+			mode_lib->vba.DPPCLKUsingSingleDPPChroma =
+					mode_lib->vba.PixelClock[k]
+							* dml_max(
+									mode_lib->vba.VTAPsChroma[k]
+											/ 6.0
+											* dml_min(
+													1.0,
+													mode_lib->vba.HRatio[k]
+															/ 2),
+									dml_max(
+											mode_lib->vba.HRatio[k]
+													* mode_lib->vba.VRatio[k]
+													/ 4
+													/ locals->PSCL_THROUGHPUT_CHROMA[k],
+											1.0));
+
+			if ((mode_lib->vba.HTAPsChroma[k] > 6 || mode_lib->vba.VTAPsChroma[k] > 6)
+					&& mode_lib->vba.DPPCLKUsingSingleDPPChroma
+							< 2 * mode_lib->vba.PixelClock[k]) {
+				mode_lib->vba.DPPCLKUsingSingleDPPChroma = 2
+						* mode_lib->vba.PixelClock[k];
+			}
+
+			locals->DPPCLKUsingSingleDPP[k] = dml_max(
+					mode_lib->vba.DPPCLKUsingSingleDPPLuma,
+					mode_lib->vba.DPPCLKUsingSingleDPPChroma);
+		}
+	}
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.BlendingAndTiming[k] != k)
+			continue;
+		if (mode_lib->vba.ODMCombineEnabled[k]) {
+			mode_lib->vba.DISPCLKWithRamping =
+					dml_max(
+							mode_lib->vba.DISPCLKWithRamping,
+							mode_lib->vba.PixelClock[k] / 2
+									* (1
+											+ mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading
+													/ 100)
+									* (1
+											+ mode_lib->vba.DISPCLKRampingMargin
+													/ 100));
+			mode_lib->vba.DISPCLKWithoutRamping =
+					dml_max(
+							mode_lib->vba.DISPCLKWithoutRamping,
+							mode_lib->vba.PixelClock[k] / 2
+									* (1
+											+ mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading
+													/ 100));
+		} else if (!mode_lib->vba.ODMCombineEnabled[k]) {
+			mode_lib->vba.DISPCLKWithRamping =
+					dml_max(
+							mode_lib->vba.DISPCLKWithRamping,
+							mode_lib->vba.PixelClock[k]
+									* (1
+											+ mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading
+													/ 100)
+									* (1
+											+ mode_lib->vba.DISPCLKRampingMargin
+													/ 100));
+			mode_lib->vba.DISPCLKWithoutRamping =
+					dml_max(
+							mode_lib->vba.DISPCLKWithoutRamping,
+							mode_lib->vba.PixelClock[k]
+									* (1
+											+ mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading
+													/ 100));
+		}
+	}
+
+	mode_lib->vba.DISPCLKWithRamping = dml_max(
+			mode_lib->vba.DISPCLKWithRamping,
+			mode_lib->vba.WritebackDISPCLK);
+	mode_lib->vba.DISPCLKWithoutRamping = dml_max(
+			mode_lib->vba.DISPCLKWithoutRamping,
+			mode_lib->vba.WritebackDISPCLK);
+
+	ASSERT(mode_lib->vba.DISPCLKDPPCLKVCOSpeed != 0);
+	mode_lib->vba.DISPCLKWithRampingRoundedToDFSGranularity = RoundToDFSGranularityUp(
+			mode_lib->vba.DISPCLKWithRamping,
+			mode_lib->vba.DISPCLKDPPCLKVCOSpeed);
+	mode_lib->vba.DISPCLKWithoutRampingRoundedToDFSGranularity = RoundToDFSGranularityUp(
+			mode_lib->vba.DISPCLKWithoutRamping,
+			mode_lib->vba.DISPCLKDPPCLKVCOSpeed);
+	mode_lib->vba.MaxDispclkRoundedToDFSGranularity = RoundToDFSGranularityDown(
+			mode_lib->vba.soc.clock_limits[mode_lib->vba.soc.num_states - 1].dispclk_mhz,
+			mode_lib->vba.DISPCLKDPPCLKVCOSpeed);
+	if (mode_lib->vba.DISPCLKWithoutRampingRoundedToDFSGranularity
+			> mode_lib->vba.MaxDispclkRoundedToDFSGranularity) {
+		mode_lib->vba.DISPCLK_calculated =
+				mode_lib->vba.DISPCLKWithoutRampingRoundedToDFSGranularity;
+	} else if (mode_lib->vba.DISPCLKWithRampingRoundedToDFSGranularity
+			> mode_lib->vba.MaxDispclkRoundedToDFSGranularity) {
+		mode_lib->vba.DISPCLK_calculated = mode_lib->vba.MaxDispclkRoundedToDFSGranularity;
+	} else {
+		mode_lib->vba.DISPCLK_calculated =
+				mode_lib->vba.DISPCLKWithRampingRoundedToDFSGranularity;
+	}
+	DTRACE("   dispclk_mhz (calculated) = %f", mode_lib->vba.DISPCLK_calculated);
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		mode_lib->vba.DPPCLK_calculated[k] = locals->DPPCLKUsingSingleDPP[k]
+				/ mode_lib->vba.DPPPerPlane[k]
+				* (1 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100);
+		mode_lib->vba.GlobalDPPCLK = dml_max(
+				mode_lib->vba.GlobalDPPCLK,
+				mode_lib->vba.DPPCLK_calculated[k]);
+	}
+	mode_lib->vba.GlobalDPPCLK = RoundToDFSGranularityUp(
+			mode_lib->vba.GlobalDPPCLK,
+			mode_lib->vba.DISPCLKDPPCLKVCOSpeed);
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		mode_lib->vba.DPPCLK_calculated[k] = mode_lib->vba.GlobalDPPCLK / 255
+				* dml_ceil(
+						mode_lib->vba.DPPCLK_calculated[k] * 255
+								/ mode_lib->vba.GlobalDPPCLK,
+						1);
+		DTRACE("   dppclk_mhz[%i] (calculated) = %f", k, mode_lib->vba.DPPCLK_calculated[k]);
+	}
+
+	// Urgent and B P-State/DRAM Clock Change Watermark
+	DTRACE("   dcfclk_mhz         = %f", mode_lib->vba.DCFCLK);
+	DTRACE("   return_bw_to_dcn   = %f", mode_lib->vba.ReturnBandwidthToDCN);
+	DTRACE("   return_bus_bw      = %f", mode_lib->vba.ReturnBW);
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		bool MainPlaneDoesODMCombine = false;
+
+		if (mode_lib->vba.SourceScan[k] == dm_horz)
+			locals->SwathWidthSingleDPPY[k] = mode_lib->vba.ViewportWidth[k];
+		else
+			locals->SwathWidthSingleDPPY[k] = mode_lib->vba.ViewportHeight[k];
+
+		if (mode_lib->vba.ODMCombineEnabled[k] == dm_odm_combine_mode_2to1)
+			MainPlaneDoesODMCombine = true;
+		for (j = 0; j < mode_lib->vba.NumberOfActivePlanes; ++j)
+			if (mode_lib->vba.BlendingAndTiming[k] == j
+					&& mode_lib->vba.ODMCombineEnabled[k] == dm_odm_combine_mode_2to1)
+				MainPlaneDoesODMCombine = true;
+
+		if (MainPlaneDoesODMCombine == true)
+			locals->SwathWidthY[k] = dml_min(
+					(double) locals->SwathWidthSingleDPPY[k],
+					dml_round(
+							mode_lib->vba.HActive[k] / 2.0
+									* mode_lib->vba.HRatio[k]));
+		else
+			locals->SwathWidthY[k] = locals->SwathWidthSingleDPPY[k]
+					/ mode_lib->vba.DPPPerPlane[k];
+	}
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.SourcePixelFormat[k] == dm_444_64) {
+			locals->BytePerPixelDETY[k] = 8;
+			locals->BytePerPixelDETC[k] = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_32) {
+			locals->BytePerPixelDETY[k] = 4;
+			locals->BytePerPixelDETC[k] = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_16 || mode_lib->vba.SourcePixelFormat[k] == dm_mono_16) {
+			locals->BytePerPixelDETY[k] = 2;
+			locals->BytePerPixelDETC[k] = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_8 || mode_lib->vba.SourcePixelFormat[k] == dm_mono_8) {
+			locals->BytePerPixelDETY[k] = 1;
+			locals->BytePerPixelDETC[k] = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_8) {
+			locals->BytePerPixelDETY[k] = 1;
+			locals->BytePerPixelDETC[k] = 2;
+		} else { // dm_420_10
+			locals->BytePerPixelDETY[k] = 4.0 / 3.0;
+			locals->BytePerPixelDETC[k] = 8.0 / 3.0;
+		}
+	}
+
+	mode_lib->vba.TotalDataReadBandwidth = 0.0;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		locals->ReadBandwidthPlaneLuma[k] = locals->SwathWidthSingleDPPY[k]
+				* dml_ceil(locals->BytePerPixelDETY[k], 1)
+				/ (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k])
+				* mode_lib->vba.VRatio[k];
+		locals->ReadBandwidthPlaneChroma[k] = locals->SwathWidthSingleDPPY[k]
+				/ 2 * dml_ceil(locals->BytePerPixelDETC[k], 2)
+				/ (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k])
+				* mode_lib->vba.VRatio[k] / 2;
+		DTRACE(
+				"   read_bw[%i] = %fBps",
+				k,
+				locals->ReadBandwidthPlaneLuma[k]
+						+ locals->ReadBandwidthPlaneChroma[k]);
+		mode_lib->vba.TotalDataReadBandwidth += locals->ReadBandwidthPlaneLuma[k]
+				+ locals->ReadBandwidthPlaneChroma[k];
+	}
+
+	// DCFCLK Deep Sleep
+	CalculateDCFCLKDeepSleep(
+		mode_lib,
+		mode_lib->vba.NumberOfActivePlanes,
+		locals->BytePerPixelDETY,
+		locals->BytePerPixelDETC,
+		mode_lib->vba.VRatio,
+		locals->SwathWidthY,
+		mode_lib->vba.DPPPerPlane,
+		mode_lib->vba.HRatio,
+		mode_lib->vba.PixelClock,
+		locals->PSCL_THROUGHPUT_LUMA,
+		locals->PSCL_THROUGHPUT_CHROMA,
+		locals->DPPCLK,
+		&mode_lib->vba.DCFCLKDeepSleep);
+
+	// DSCCLK
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if ((mode_lib->vba.BlendingAndTiming[k] != k) || !mode_lib->vba.DSCEnabled[k]) {
+			locals->DSCCLK_calculated[k] = 0.0;
+		} else {
+			if (mode_lib->vba.OutputFormat[k] == dm_420
+					|| mode_lib->vba.OutputFormat[k] == dm_n422)
+				mode_lib->vba.DSCFormatFactor = 2;
+			else
+				mode_lib->vba.DSCFormatFactor = 1;
+			if (mode_lib->vba.ODMCombineEnabled[k])
+				locals->DSCCLK_calculated[k] =
+						mode_lib->vba.PixelClockBackEnd[k] / 6
+								/ mode_lib->vba.DSCFormatFactor
+								/ (1
+										- mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading
+												/ 100);
+			else
+				locals->DSCCLK_calculated[k] =
+						mode_lib->vba.PixelClockBackEnd[k] / 3
+								/ mode_lib->vba.DSCFormatFactor
+								/ (1
+										- mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading
+												/ 100);
+		}
+	}
+
+	// DSC Delay
+	// TODO
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		double bpp = mode_lib->vba.OutputBpp[k];
+		unsigned int slices = mode_lib->vba.NumberOfDSCSlices[k];
+
+		if (mode_lib->vba.DSCEnabled[k] && bpp != 0) {
+			if (!mode_lib->vba.ODMCombineEnabled[k]) {
+				locals->DSCDelay[k] =
+						dscceComputeDelay(
+								mode_lib->vba.DSCInputBitPerComponent[k],
+								bpp,
+								dml_ceil(
+										(double) mode_lib->vba.HActive[k]
+												/ mode_lib->vba.NumberOfDSCSlices[k],
+										1),
+								slices,
+								mode_lib->vba.OutputFormat[k])
+								+ dscComputeDelay(
+										mode_lib->vba.OutputFormat[k]);
+			} else {
+				locals->DSCDelay[k] =
+						2
+								* (dscceComputeDelay(
+										mode_lib->vba.DSCInputBitPerComponent[k],
+										bpp,
+										dml_ceil(
+												(double) mode_lib->vba.HActive[k]
+														/ mode_lib->vba.NumberOfDSCSlices[k],
+												1),
+										slices / 2.0,
+										mode_lib->vba.OutputFormat[k])
+										+ dscComputeDelay(
+												mode_lib->vba.OutputFormat[k]));
+			}
+			locals->DSCDelay[k] = locals->DSCDelay[k]
+					* mode_lib->vba.PixelClock[k]
+					/ mode_lib->vba.PixelClockBackEnd[k];
+		} else {
+			locals->DSCDelay[k] = 0;
+		}
+	}
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k)
+		for (j = 0; j < mode_lib->vba.NumberOfActivePlanes; ++j) // NumberOfPlanes
+			if (j != k && mode_lib->vba.BlendingAndTiming[k] == j
+					&& mode_lib->vba.DSCEnabled[j])
+				locals->DSCDelay[k] = locals->DSCDelay[j];
+
+	// Prefetch
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		unsigned int PDEAndMetaPTEBytesFrameY;
+		unsigned int PixelPTEBytesPerRowY;
+		unsigned int MetaRowByteY;
+		unsigned int MetaRowByteC;
+		unsigned int PDEAndMetaPTEBytesFrameC;
+		unsigned int PixelPTEBytesPerRowC;
+		bool         PTEBufferSizeNotExceededY;
+		bool         PTEBufferSizeNotExceededC;
+
+		Calculate256BBlockSizes(
+				mode_lib->vba.SourcePixelFormat[k],
+				mode_lib->vba.SurfaceTiling[k],
+				dml_ceil(locals->BytePerPixelDETY[k], 1),
+				dml_ceil(locals->BytePerPixelDETC[k], 2),
+				&locals->BlockHeight256BytesY[k],
+				&locals->BlockHeight256BytesC[k],
+				&locals->BlockWidth256BytesY[k],
+				&locals->BlockWidth256BytesC[k]);
+
+		locals->PrefetchSourceLinesY[k] = CalculatePrefetchSourceLines(
+				mode_lib,
+				mode_lib->vba.VRatio[k],
+				mode_lib->vba.vtaps[k],
+				mode_lib->vba.Interlace[k],
+				mode_lib->vba.ProgressiveToInterlaceUnitInOPP,
+				mode_lib->vba.SwathHeightY[k],
+				mode_lib->vba.ViewportYStartY[k],
+				&locals->VInitPreFillY[k],
+				&locals->MaxNumSwathY[k]);
+
+		if ((mode_lib->vba.SourcePixelFormat[k] != dm_444_64
+				&& mode_lib->vba.SourcePixelFormat[k] != dm_444_32
+				&& mode_lib->vba.SourcePixelFormat[k] != dm_444_16
+				&& mode_lib->vba.SourcePixelFormat[k] != dm_444_8)) {
+			PDEAndMetaPTEBytesFrameC =
+					CalculateVMAndRowBytes(
+							mode_lib,
+							mode_lib->vba.DCCEnable[k],
+							locals->BlockHeight256BytesC[k],
+							locals->BlockWidth256BytesC[k],
+							mode_lib->vba.SourcePixelFormat[k],
+							mode_lib->vba.SurfaceTiling[k],
+							dml_ceil(
+									locals->BytePerPixelDETC[k],
+									2),
+							mode_lib->vba.SourceScan[k],
+							mode_lib->vba.ViewportWidth[k] / 2,
+							mode_lib->vba.ViewportHeight[k] / 2,
+							locals->SwathWidthY[k] / 2,
+							mode_lib->vba.GPUVMEnable,
+							mode_lib->vba.HostVMEnable,
+							mode_lib->vba.HostVMMaxPageTableLevels,
+							mode_lib->vba.HostVMCachedPageTableLevels,
+							mode_lib->vba.VMMPageSize,
+							mode_lib->vba.PTEBufferSizeInRequestsChroma,
+							mode_lib->vba.PitchC[k],
+							mode_lib->vba.DCCMetaPitchC[k],
+							&locals->MacroTileWidthC[k],
+							&MetaRowByteC,
+							&PixelPTEBytesPerRowC,
+							&PTEBufferSizeNotExceededC,
+							&locals->dpte_row_width_chroma_ub[k],
+							&locals->dpte_row_height_chroma[k],
+							&locals->meta_req_width_chroma[k],
+							&locals->meta_req_height_chroma[k],
+							&locals->meta_row_width_chroma[k],
+							&locals->meta_row_height_chroma[k],
+							&locals->vm_group_bytes_chroma,
+							&locals->dpte_group_bytes_chroma,
+							&locals->PixelPTEReqWidthC[k],
+							&locals->PixelPTEReqHeightC[k],
+							&locals->PTERequestSizeC[k],
+							&locals->dpde0_bytes_per_frame_ub_c[k],
+							&locals->meta_pte_bytes_per_frame_ub_c[k]);
+
+			locals->PrefetchSourceLinesC[k] = CalculatePrefetchSourceLines(
+					mode_lib,
+					mode_lib->vba.VRatio[k] / 2,
+					mode_lib->vba.VTAPsChroma[k],
+					mode_lib->vba.Interlace[k],
+					mode_lib->vba.ProgressiveToInterlaceUnitInOPP,
+					mode_lib->vba.SwathHeightC[k],
+					mode_lib->vba.ViewportYStartC[k],
+					&locals->VInitPreFillC[k],
+					&locals->MaxNumSwathC[k]);
+		} else {
+			PixelPTEBytesPerRowC = 0;
+			PDEAndMetaPTEBytesFrameC = 0;
+			MetaRowByteC = 0;
+			locals->MaxNumSwathC[k] = 0;
+			locals->PrefetchSourceLinesC[k] = 0;
+			locals->PTEBufferSizeInRequestsForLuma = mode_lib->vba.PTEBufferSizeInRequestsLuma + mode_lib->vba.PTEBufferSizeInRequestsChroma;
+		}
+
+		PDEAndMetaPTEBytesFrameY = CalculateVMAndRowBytes(
+				mode_lib,
+				mode_lib->vba.DCCEnable[k],
+				locals->BlockHeight256BytesY[k],
+				locals->BlockWidth256BytesY[k],
+				mode_lib->vba.SourcePixelFormat[k],
+				mode_lib->vba.SurfaceTiling[k],
+				dml_ceil(locals->BytePerPixelDETY[k], 1),
+				mode_lib->vba.SourceScan[k],
+				mode_lib->vba.ViewportWidth[k],
+				mode_lib->vba.ViewportHeight[k],
+				locals->SwathWidthY[k],
+				mode_lib->vba.GPUVMEnable,
+				mode_lib->vba.HostVMEnable,
+				mode_lib->vba.HostVMMaxPageTableLevels,
+				mode_lib->vba.HostVMCachedPageTableLevels,
+				mode_lib->vba.VMMPageSize,
+				locals->PTEBufferSizeInRequestsForLuma,
+				mode_lib->vba.PitchY[k],
+				mode_lib->vba.DCCMetaPitchY[k],
+				&locals->MacroTileWidthY[k],
+				&MetaRowByteY,
+				&PixelPTEBytesPerRowY,
+				&PTEBufferSizeNotExceededY,
+				&locals->dpte_row_width_luma_ub[k],
+				&locals->dpte_row_height[k],
+				&locals->meta_req_width[k],
+				&locals->meta_req_height[k],
+				&locals->meta_row_width[k],
+				&locals->meta_row_height[k],
+				&locals->vm_group_bytes[k],
+				&locals->dpte_group_bytes[k],
+				&locals->PixelPTEReqWidthY[k],
+				&locals->PixelPTEReqHeightY[k],
+				&locals->PTERequestSizeY[k],
+				&locals->dpde0_bytes_per_frame_ub_l[k],
+				&locals->meta_pte_bytes_per_frame_ub_l[k]);
+
+		locals->PixelPTEBytesPerRow[k] = PixelPTEBytesPerRowY + PixelPTEBytesPerRowC;
+		locals->PDEAndMetaPTEBytesFrame[k] = PDEAndMetaPTEBytesFrameY
+				+ PDEAndMetaPTEBytesFrameC;
+		locals->MetaRowByte[k] = MetaRowByteY + MetaRowByteC;
+
+		CalculateActiveRowBandwidth(
+				mode_lib->vba.GPUVMEnable,
+				mode_lib->vba.SourcePixelFormat[k],
+				mode_lib->vba.VRatio[k],
+				mode_lib->vba.DCCEnable[k],
+				mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k],
+				MetaRowByteY,
+				MetaRowByteC,
+				locals->meta_row_height[k],
+				locals->meta_row_height_chroma[k],
+				PixelPTEBytesPerRowY,
+				PixelPTEBytesPerRowC,
+				locals->dpte_row_height[k],
+				locals->dpte_row_height_chroma[k],
+				&locals->meta_row_bw[k],
+				&locals->dpte_row_bw[k]);
+	}
+
+	mode_lib->vba.TotalDCCActiveDPP = 0;
+	mode_lib->vba.TotalActiveDPP = 0;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		mode_lib->vba.TotalActiveDPP = mode_lib->vba.TotalActiveDPP
+				+ mode_lib->vba.DPPPerPlane[k];
+		if (mode_lib->vba.DCCEnable[k])
+			mode_lib->vba.TotalDCCActiveDPP = mode_lib->vba.TotalDCCActiveDPP
+					+ mode_lib->vba.DPPPerPlane[k];
+	}
+
+	mode_lib->vba.UrgentOutOfOrderReturnPerChannel = dml_max3(
+			mode_lib->vba.UrgentOutOfOrderReturnPerChannelPixelDataOnly,
+			mode_lib->vba.UrgentOutOfOrderReturnPerChannelPixelMixedWithVMData,
+			mode_lib->vba.UrgentOutOfOrderReturnPerChannelVMDataOnly);
+
+	mode_lib->vba.UrgentRoundTripAndOutOfOrderLatency =
+			(mode_lib->vba.RoundTripPingLatencyCycles + 32) / mode_lib->vba.DCFCLK
+					+ mode_lib->vba.UrgentOutOfOrderReturnPerChannel
+							* mode_lib->vba.NumberOfChannels
+							/ mode_lib->vba.ReturnBW;
+
+	mode_lib->vba.UrgentExtraLatency = CalculateExtraLatency(
+			mode_lib->vba.UrgentRoundTripAndOutOfOrderLatency,
+			mode_lib->vba.TotalActiveDPP,
+			mode_lib->vba.PixelChunkSizeInKByte,
+			mode_lib->vba.TotalDCCActiveDPP,
+			mode_lib->vba.MetaChunkSize,
+			mode_lib->vba.ReturnBW,
+			mode_lib->vba.GPUVMEnable,
+			mode_lib->vba.HostVMEnable,
+			mode_lib->vba.NumberOfActivePlanes,
+			mode_lib->vba.DPPPerPlane,
+			locals->dpte_group_bytes,
+			mode_lib->vba.PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData,
+			mode_lib->vba.PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly,
+			mode_lib->vba.HostVMMaxPageTableLevels,
+			mode_lib->vba.HostVMCachedPageTableLevels);
+
+
+	mode_lib->vba.TCalc = 24.0 / mode_lib->vba.DCFCLKDeepSleep;
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.BlendingAndTiming[k] == k) {
+			if (mode_lib->vba.WritebackEnable[k] == true) {
+				locals->WritebackDelay[mode_lib->vba.VoltageLevel][k] =
+						mode_lib->vba.WritebackLatency
+								+ CalculateWriteBackDelay(
+										mode_lib->vba.WritebackPixelFormat[k],
+										mode_lib->vba.WritebackHRatio[k],
+										mode_lib->vba.WritebackVRatio[k],
+										mode_lib->vba.WritebackLumaHTaps[k],
+										mode_lib->vba.WritebackLumaVTaps[k],
+										mode_lib->vba.WritebackChromaHTaps[k],
+										mode_lib->vba.WritebackChromaVTaps[k],
+										mode_lib->vba.WritebackDestinationWidth[k])
+										/ mode_lib->vba.DISPCLK;
+			} else
+				locals->WritebackDelay[mode_lib->vba.VoltageLevel][k] = 0;
+			for (j = 0; j < mode_lib->vba.NumberOfActivePlanes; ++j) {
+				if (mode_lib->vba.BlendingAndTiming[j] == k
+						&& mode_lib->vba.WritebackEnable[j] == true) {
+					locals->WritebackDelay[mode_lib->vba.VoltageLevel][k] =
+							dml_max(
+									locals->WritebackDelay[mode_lib->vba.VoltageLevel][k],
+									mode_lib->vba.WritebackLatency
+											+ CalculateWriteBackDelay(
+													mode_lib->vba.WritebackPixelFormat[j],
+													mode_lib->vba.WritebackHRatio[j],
+													mode_lib->vba.WritebackVRatio[j],
+													mode_lib->vba.WritebackLumaHTaps[j],
+													mode_lib->vba.WritebackLumaVTaps[j],
+													mode_lib->vba.WritebackChromaHTaps[j],
+													mode_lib->vba.WritebackChromaVTaps[j],
+													mode_lib->vba.WritebackDestinationWidth[j])
+													/ mode_lib->vba.DISPCLK);
+				}
+			}
+		}
+	}
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k)
+		for (j = 0; j < mode_lib->vba.NumberOfActivePlanes; ++j)
+			if (mode_lib->vba.BlendingAndTiming[k] == j)
+				locals->WritebackDelay[mode_lib->vba.VoltageLevel][k] =
+						locals->WritebackDelay[mode_lib->vba.VoltageLevel][j];
+
+	mode_lib->vba.VStartupLines = 13;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		locals->MaxVStartupLines[k] = mode_lib->vba.VTotal[k] - mode_lib->vba.VActive[k] - dml_max(1.0, dml_ceil(locals->WritebackDelay[mode_lib->vba.VoltageLevel][k] / (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k]), 1));
+	}
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k)
+		locals->MaximumMaxVStartupLines = dml_max(locals->MaximumMaxVStartupLines, locals->MaxVStartupLines[k]);
+
+	// We don't really care to iterate between the various prefetch modes
+	//mode_lib->vba.PrefetchERROR = CalculateMinAndMaxPrefetchMode(mode_lib->vba.AllowDRAMSelfRefreshOrDRAMClockChangeInVblank, &mode_lib->vba.MinPrefetchMode, &mode_lib->vba.MaxPrefetchMode);
+	mode_lib->vba.UrgentLatency = dml_max3(mode_lib->vba.UrgentLatencyPixelDataOnly, mode_lib->vba.UrgentLatencyPixelMixedWithVMData, mode_lib->vba.UrgentLatencyVMDataOnly);
+
+	do {
+		double MaxTotalRDBandwidth = 0;
+		double MaxTotalRDBandwidthNoUrgentBurst = 0;
+		bool DestinationLineTimesForPrefetchLessThan2 = false;
+		bool VRatioPrefetchMoreThan4 = false;
+		double TWait = CalculateTWait(
+				mode_lib->vba.PrefetchMode[mode_lib->vba.VoltageLevel][mode_lib->vba.maxMpcComb],
+				mode_lib->vba.DRAMClockChangeLatency,
+				mode_lib->vba.UrgentLatency,
+				mode_lib->vba.SREnterPlusExitTime);
+
+		for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+			Pipe   myPipe;
+			HostVM myHostVM;
+
+			if (mode_lib->vba.XFCEnabled[k] == true) {
+				mode_lib->vba.XFCRemoteSurfaceFlipDelay =
+						CalculateRemoteSurfaceFlipDelay(
+								mode_lib,
+								mode_lib->vba.VRatio[k],
+								locals->SwathWidthY[k],
+								dml_ceil(
+										locals->BytePerPixelDETY[k],
+										1),
+								mode_lib->vba.HTotal[k]
+										/ mode_lib->vba.PixelClock[k],
+								mode_lib->vba.XFCTSlvVupdateOffset,
+								mode_lib->vba.XFCTSlvVupdateWidth,
+								mode_lib->vba.XFCTSlvVreadyOffset,
+								mode_lib->vba.XFCXBUFLatencyTolerance,
+								mode_lib->vba.XFCFillBWOverhead,
+								mode_lib->vba.XFCSlvChunkSize,
+								mode_lib->vba.XFCBusTransportTime,
+								mode_lib->vba.TCalc,
+								TWait,
+								&mode_lib->vba.SrcActiveDrainRate,
+								&mode_lib->vba.TInitXFill,
+								&mode_lib->vba.TslvChk);
+			} else {
+				mode_lib->vba.XFCRemoteSurfaceFlipDelay = 0;
+			}
+
+			myPipe.DPPCLK = locals->DPPCLK[k];
+			myPipe.DISPCLK = mode_lib->vba.DISPCLK;
+			myPipe.PixelClock = mode_lib->vba.PixelClock[k];
+			myPipe.DCFCLKDeepSleep = mode_lib->vba.DCFCLKDeepSleep;
+			myPipe.DPPPerPlane = mode_lib->vba.DPPPerPlane[k];
+			myPipe.ScalerEnabled = mode_lib->vba.ScalerEnabled[k];
+			myPipe.SourceScan = mode_lib->vba.SourceScan[k];
+			myPipe.BlockWidth256BytesY = locals->BlockWidth256BytesY[k];
+			myPipe.BlockHeight256BytesY = locals->BlockHeight256BytesY[k];
+			myPipe.BlockWidth256BytesC = locals->BlockWidth256BytesC[k];
+			myPipe.BlockHeight256BytesC = locals->BlockHeight256BytesC[k];
+			myPipe.InterlaceEnable = mode_lib->vba.Interlace[k];
+			myPipe.NumberOfCursors = mode_lib->vba.NumberOfCursors[k];
+			myPipe.VBlank = mode_lib->vba.VTotal[k] - mode_lib->vba.VActive[k];
+			myPipe.HTotal = mode_lib->vba.HTotal[k];
+
+
+			myHostVM.Enable = mode_lib->vba.HostVMEnable;
+			myHostVM.MaxPageTableLevels = mode_lib->vba.HostVMMaxPageTableLevels;
+			myHostVM.CachedPageTableLevels = mode_lib->vba.HostVMCachedPageTableLevels;
+
+			mode_lib->vba.ErrorResult[k] =
+					CalculatePrefetchSchedule(
+							mode_lib,
+							mode_lib->vba.PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData,
+							mode_lib->vba.PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly,
+							&myPipe,
+							locals->DSCDelay[k],
+							mode_lib->vba.DPPCLKDelaySubtotal,
+							mode_lib->vba.DPPCLKDelaySCL,
+							mode_lib->vba.DPPCLKDelaySCLLBOnly,
+							mode_lib->vba.DPPCLKDelayCNVCFormater,
+							mode_lib->vba.DPPCLKDelayCNVCCursor,
+							mode_lib->vba.DISPCLKDelaySubtotal,
+							(unsigned int) (locals->SwathWidthY[k]
+									/ mode_lib->vba.HRatio[k]),
+							mode_lib->vba.OutputFormat[k],
+							mode_lib->vba.MaxInterDCNTileRepeaters,
+							dml_min(mode_lib->vba.VStartupLines, locals->MaxVStartupLines[k]),
+							locals->MaxVStartupLines[k],
+							mode_lib->vba.GPUVMMaxPageTableLevels,
+							mode_lib->vba.GPUVMEnable,
+							&myHostVM,
+							mode_lib->vba.DynamicMetadataEnable[k],
+							mode_lib->vba.DynamicMetadataLinesBeforeActiveRequired[k],
+							mode_lib->vba.DynamicMetadataTransmittedBytes[k],
+							mode_lib->vba.DCCEnable[k],
+							mode_lib->vba.UrgentLatency,
+							mode_lib->vba.UrgentExtraLatency,
+							mode_lib->vba.TCalc,
+							locals->PDEAndMetaPTEBytesFrame[k],
+							locals->MetaRowByte[k],
+							locals->PixelPTEBytesPerRow[k],
+							locals->PrefetchSourceLinesY[k],
+							locals->SwathWidthY[k],
+							locals->BytePerPixelDETY[k],
+							locals->VInitPreFillY[k],
+							locals->MaxNumSwathY[k],
+							locals->PrefetchSourceLinesC[k],
+							locals->BytePerPixelDETC[k],
+							locals->VInitPreFillC[k],
+							locals->MaxNumSwathC[k],
+							mode_lib->vba.SwathHeightY[k],
+							mode_lib->vba.SwathHeightC[k],
+							TWait,
+							mode_lib->vba.XFCEnabled[k],
+							mode_lib->vba.XFCRemoteSurfaceFlipDelay,
+							mode_lib->vba.ProgressiveToInterlaceUnitInOPP,
+							&locals->DSTXAfterScaler[k],
+							&locals->DSTYAfterScaler[k],
+							&locals->DestinationLinesForPrefetch[k],
+							&locals->PrefetchBandwidth[k],
+							&locals->DestinationLinesToRequestVMInVBlank[k],
+							&locals->DestinationLinesToRequestRowInVBlank[k],
+							&locals->VRatioPrefetchY[k],
+							&locals->VRatioPrefetchC[k],
+							&locals->RequiredPrefetchPixDataBWLuma[k],
+							&locals->RequiredPrefetchPixDataBWChroma[k],
+							&locals->VStartupRequiredWhenNotEnoughTimeForDynamicMetadata,
+							&locals->Tno_bw[k],
+							&locals->prefetch_vmrow_bw[k],
+							&locals->swath_width_luma_ub[k],
+							&locals->swath_width_chroma_ub[k],
+							&mode_lib->vba.VUpdateOffsetPix[k],
+							&mode_lib->vba.VUpdateWidthPix[k],
+							&mode_lib->vba.VReadyOffsetPix[k]);
+			if (mode_lib->vba.BlendingAndTiming[k] == k) {
+				locals->VStartup[k] = dml_min(
+						mode_lib->vba.VStartupLines,
+						locals->MaxVStartupLines[k]);
+				if (locals->VStartupRequiredWhenNotEnoughTimeForDynamicMetadata
+						!= 0) {
+					locals->VStartup[k] =
+							locals->VStartupRequiredWhenNotEnoughTimeForDynamicMetadata;
+				}
+			} else {
+				locals->VStartup[k] =
+						dml_min(
+								mode_lib->vba.VStartupLines,
+								locals->MaxVStartupLines[mode_lib->vba.BlendingAndTiming[k]]);
+			}
+		}
+
+		for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+			unsigned int m;
+
+			locals->cursor_bw[k] = 0;
+			locals->cursor_bw_pre[k] = 0;
+			for (m = 0; m < mode_lib->vba.NumberOfCursors[k]; m++) {
+				locals->cursor_bw[k] += mode_lib->vba.CursorWidth[k][m] * mode_lib->vba.CursorBPP[k][m] / 8.0 / (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k]) * mode_lib->vba.VRatio[k];
+				locals->cursor_bw_pre[k] += mode_lib->vba.CursorWidth[k][m] * mode_lib->vba.CursorBPP[k][m] / 8.0 / (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k]) * locals->VRatioPrefetchY[k];
+			}
+
+			CalculateUrgentBurstFactor(
+					mode_lib->vba.DETBufferSizeInKByte[0],
+					mode_lib->vba.SwathHeightY[k],
+					mode_lib->vba.SwathHeightC[k],
+					locals->SwathWidthY[k],
+					mode_lib->vba.HTotal[k] /
+					mode_lib->vba.PixelClock[k],
+					mode_lib->vba.UrgentLatency,
+					mode_lib->vba.CursorBufferSize,
+					mode_lib->vba.CursorWidth[k][0] + mode_lib->vba.CursorWidth[k][1],
+					dml_max(mode_lib->vba.CursorBPP[k][0], mode_lib->vba.CursorBPP[k][1]),
+					mode_lib->vba.VRatio[k],
+					locals->VRatioPrefetchY[k],
+					locals->VRatioPrefetchC[k],
+					locals->BytePerPixelDETY[k],
+					locals->BytePerPixelDETC[k],
+					&locals->UrgentBurstFactorCursor[k],
+					&locals->UrgentBurstFactorCursorPre[k],
+					&locals->UrgentBurstFactorLuma[k],
+					&locals->UrgentBurstFactorLumaPre[k],
+					&locals->UrgentBurstFactorChroma[k],
+					&locals->UrgentBurstFactorChromaPre[k],
+					&locals->NotEnoughUrgentLatencyHiding[0][0],
+					&locals->NotEnoughUrgentLatencyHidingPre);
+
+			if (mode_lib->vba.UseUrgentBurstBandwidth == false) {
+				locals->UrgentBurstFactorLuma[k] = 1;
+				locals->UrgentBurstFactorChroma[k] = 1;
+				locals->UrgentBurstFactorCursor[k] = 1;
+				locals->UrgentBurstFactorLumaPre[k] = 1;
+				locals->UrgentBurstFactorChromaPre[k] = 1;
+				locals->UrgentBurstFactorCursorPre[k] = 1;
+			}
+
+			MaxTotalRDBandwidth = MaxTotalRDBandwidth +
+				dml_max3(locals->prefetch_vmrow_bw[k],
+					locals->ReadBandwidthPlaneLuma[k] * locals->UrgentBurstFactorLuma[k]
+					+ locals->ReadBandwidthPlaneChroma[k] * locals->UrgentBurstFactorChroma[k] + locals->cursor_bw[k]
+					* locals->UrgentBurstFactorCursor[k] + locals->meta_row_bw[k] + locals->dpte_row_bw[k],
+					locals->RequiredPrefetchPixDataBWLuma[k] * locals->UrgentBurstFactorLumaPre[k] + locals->RequiredPrefetchPixDataBWChroma[k]
+					* locals->UrgentBurstFactorChromaPre[k] + locals->cursor_bw_pre[k] * locals->UrgentBurstFactorCursorPre[k]);
+
+			MaxTotalRDBandwidthNoUrgentBurst = MaxTotalRDBandwidthNoUrgentBurst +
+				dml_max3(locals->prefetch_vmrow_bw[k],
+					locals->ReadBandwidthPlaneLuma[k] + locals->ReadBandwidthPlaneChroma[k] + locals->cursor_bw[k]
+					+ locals->meta_row_bw[k] + locals->dpte_row_bw[k],
+					locals->RequiredPrefetchPixDataBWLuma[k] + locals->RequiredPrefetchPixDataBWChroma[k] + locals->cursor_bw_pre[k]);
+
+			if (locals->DestinationLinesForPrefetch[k] < 2)
+				DestinationLineTimesForPrefetchLessThan2 = true;
+			if (locals->VRatioPrefetchY[k] > 4 || locals->VRatioPrefetchC[k] > 4)
+				VRatioPrefetchMoreThan4 = true;
+		}
+		mode_lib->vba.FractionOfUrgentBandwidth = MaxTotalRDBandwidthNoUrgentBurst / mode_lib->vba.ReturnBW;
+
+		if (MaxTotalRDBandwidth <= mode_lib->vba.ReturnBW && locals->NotEnoughUrgentLatencyHiding[0][0] == 0 &&
+				locals->NotEnoughUrgentLatencyHidingPre == 0 && !VRatioPrefetchMoreThan4
+				&& !DestinationLineTimesForPrefetchLessThan2)
+			mode_lib->vba.PrefetchModeSupported = true;
+		else {
+			mode_lib->vba.PrefetchModeSupported = false;
+			dml_print(
+					"DML: CalculatePrefetchSchedule ***failed***. Bandwidth violation. Results are NOT valid\n");
+		}
+
+		if (mode_lib->vba.PrefetchModeSupported == true) {
+			mode_lib->vba.BandwidthAvailableForImmediateFlip = mode_lib->vba.ReturnBW;
+			for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+				mode_lib->vba.BandwidthAvailableForImmediateFlip =
+						mode_lib->vba.BandwidthAvailableForImmediateFlip
+							- dml_max(
+								locals->ReadBandwidthPlaneLuma[k] * locals->UrgentBurstFactorLuma[k]
+								+ locals->ReadBandwidthPlaneChroma[k] * locals->UrgentBurstFactorChroma[k]
+								+ locals->cursor_bw[k] * locals->UrgentBurstFactorCursor[k],
+								locals->RequiredPrefetchPixDataBWLuma[k] * locals->UrgentBurstFactorLumaPre[k] +
+								locals->RequiredPrefetchPixDataBWChroma[k] * locals->UrgentBurstFactorChromaPre[k] +
+								locals->cursor_bw_pre[k] * locals->UrgentBurstFactorCursorPre[k]);
+			}
+
+			mode_lib->vba.TotImmediateFlipBytes = 0;
+			for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+				mode_lib->vba.TotImmediateFlipBytes = mode_lib->vba.TotImmediateFlipBytes + locals->PDEAndMetaPTEBytesFrame[k] + locals->MetaRowByte[k] + locals->PixelPTEBytesPerRow[k];
+			}
+			for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+				CalculateFlipSchedule(
+						mode_lib,
+						mode_lib->vba.PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData,
+						mode_lib->vba.PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly,
+						mode_lib->vba.UrgentExtraLatency,
+						mode_lib->vba.UrgentLatency,
+						mode_lib->vba.GPUVMMaxPageTableLevels,
+						mode_lib->vba.HostVMEnable,
+						mode_lib->vba.HostVMMaxPageTableLevels,
+						mode_lib->vba.HostVMCachedPageTableLevels,
+						mode_lib->vba.GPUVMEnable,
+						locals->PDEAndMetaPTEBytesFrame[k],
+						locals->MetaRowByte[k],
+						locals->PixelPTEBytesPerRow[k],
+						mode_lib->vba.BandwidthAvailableForImmediateFlip,
+						mode_lib->vba.TotImmediateFlipBytes,
+						mode_lib->vba.SourcePixelFormat[k],
+						mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k],
+						mode_lib->vba.VRatio[k],
+						locals->Tno_bw[k],
+						mode_lib->vba.DCCEnable[k],
+						locals->dpte_row_height[k],
+						locals->meta_row_height[k],
+						locals->dpte_row_height_chroma[k],
+						locals->meta_row_height_chroma[k],
+						&locals->DestinationLinesToRequestVMInImmediateFlip[k],
+						&locals->DestinationLinesToRequestRowInImmediateFlip[k],
+						&locals->final_flip_bw[k],
+						&locals->ImmediateFlipSupportedForPipe[k]);
+			}
+			mode_lib->vba.total_dcn_read_bw_with_flip = 0.0;
+			mode_lib->vba.total_dcn_read_bw_with_flip_no_urgent_burst = 0.0;
+			for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+				mode_lib->vba.total_dcn_read_bw_with_flip =
+						mode_lib->vba.total_dcn_read_bw_with_flip + dml_max3(
+							locals->prefetch_vmrow_bw[k],
+							locals->final_flip_bw[k] + locals->ReadBandwidthLuma[k] * locals->UrgentBurstFactorLuma[k]
+							+ locals->ReadBandwidthChroma[k] * locals->UrgentBurstFactorChroma[k] + locals->cursor_bw[k] * locals->UrgentBurstFactorCursor[k],
+							locals->final_flip_bw[k] + locals->RequiredPrefetchPixDataBWLuma[k] * locals->UrgentBurstFactorLumaPre[k]
+							+ locals->RequiredPrefetchPixDataBWChroma[k] * locals->UrgentBurstFactorChromaPre[k]
+							+ locals->cursor_bw_pre[k] * locals->UrgentBurstFactorCursorPre[k]);
+				mode_lib->vba.total_dcn_read_bw_with_flip_no_urgent_burst =
+				mode_lib->vba.total_dcn_read_bw_with_flip_no_urgent_burst +
+					dml_max3(locals->prefetch_vmrow_bw[k],
+						locals->final_flip_bw[k] + locals->ReadBandwidthPlaneLuma[k] + locals->ReadBandwidthPlaneChroma[k] + locals->cursor_bw[k],
+						locals->final_flip_bw[k] + locals->RequiredPrefetchPixDataBWLuma[k] + locals->RequiredPrefetchPixDataBWChroma[k] + locals->cursor_bw_pre[k]);
+
+			}
+			mode_lib->vba.FractionOfUrgentBandwidthImmediateFlip = mode_lib->vba.total_dcn_read_bw_with_flip_no_urgent_burst / mode_lib->vba.ReturnBW;
+
+			mode_lib->vba.ImmediateFlipSupported = true;
+			if (mode_lib->vba.total_dcn_read_bw_with_flip > mode_lib->vba.ReturnBW) {
+				mode_lib->vba.ImmediateFlipSupported = false;
+			}
+			for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+				if (locals->ImmediateFlipSupportedForPipe[k] == false) {
+					mode_lib->vba.ImmediateFlipSupported = false;
+				}
+			}
+		} else {
+			mode_lib->vba.ImmediateFlipSupported = false;
+		}
+
+		for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+			if (mode_lib->vba.ErrorResult[k]) {
+				mode_lib->vba.PrefetchModeSupported = false;
+				dml_print(
+						"DML: CalculatePrefetchSchedule ***failed***. Prefetch schedule violation. Results are NOT valid\n");
+			}
+		}
+
+		mode_lib->vba.VStartupLines = mode_lib->vba.VStartupLines + 1;
+	} while (!((mode_lib->vba.PrefetchModeSupported
+			&& ((!mode_lib->vba.ImmediateFlipSupport && !mode_lib->vba.HostVMEnable)
+					|| mode_lib->vba.ImmediateFlipSupported))
+			|| locals->MaximumMaxVStartupLines < mode_lib->vba.VStartupLines));
+
+	//Watermarks and NB P-State/DRAM Clock Change Support
+	{
+		enum clock_change_support DRAMClockChangeSupport; // dummy
+		CalculateWatermarksAndDRAMSpeedChangeSupport(
+				mode_lib,
+				mode_lib->vba.PrefetchMode[mode_lib->vba.VoltageLevel][mode_lib->vba.maxMpcComb],
+				mode_lib->vba.NumberOfActivePlanes,
+				mode_lib->vba.MaxLineBufferLines,
+				mode_lib->vba.LineBufferSize,
+				mode_lib->vba.DPPOutputBufferPixels,
+				mode_lib->vba.DETBufferSizeInKByte[0],
+				mode_lib->vba.WritebackInterfaceLumaBufferSize,
+				mode_lib->vba.WritebackInterfaceChromaBufferSize,
+				mode_lib->vba.DCFCLK,
+				mode_lib->vba.UrgentOutOfOrderReturnPerChannel * mode_lib->vba.NumberOfChannels,
+				mode_lib->vba.ReturnBW,
+				mode_lib->vba.GPUVMEnable,
+				locals->dpte_group_bytes,
+				mode_lib->vba.MetaChunkSize,
+				mode_lib->vba.UrgentLatency,
+				mode_lib->vba.UrgentExtraLatency,
+				mode_lib->vba.WritebackLatency,
+				mode_lib->vba.WritebackChunkSize,
+				mode_lib->vba.SOCCLK,
+				mode_lib->vba.DRAMClockChangeLatency,
+				mode_lib->vba.SRExitTime,
+				mode_lib->vba.SREnterPlusExitTime,
+				mode_lib->vba.DCFCLKDeepSleep,
+				mode_lib->vba.DPPPerPlane,
+				mode_lib->vba.DCCEnable,
+				locals->DPPCLK,
+				locals->SwathWidthSingleDPPY,
+				mode_lib->vba.SwathHeightY,
+				locals->ReadBandwidthPlaneLuma,
+				mode_lib->vba.SwathHeightC,
+				locals->ReadBandwidthPlaneChroma,
+				mode_lib->vba.LBBitPerPixel,
+				locals->SwathWidthY,
+				mode_lib->vba.HRatio,
+				mode_lib->vba.vtaps,
+				mode_lib->vba.VTAPsChroma,
+				mode_lib->vba.VRatio,
+				mode_lib->vba.HTotal,
+				mode_lib->vba.PixelClock,
+				mode_lib->vba.BlendingAndTiming,
+				locals->BytePerPixelDETY,
+				locals->BytePerPixelDETC,
+				mode_lib->vba.WritebackEnable,
+				mode_lib->vba.WritebackPixelFormat,
+				mode_lib->vba.WritebackDestinationWidth,
+				mode_lib->vba.WritebackDestinationHeight,
+				mode_lib->vba.WritebackSourceHeight,
+				&DRAMClockChangeSupport,
+				&mode_lib->vba.UrgentWatermark,
+				&mode_lib->vba.WritebackUrgentWatermark,
+				&mode_lib->vba.DRAMClockChangeWatermark,
+				&mode_lib->vba.WritebackDRAMClockChangeWatermark,
+				&mode_lib->vba.StutterExitWatermark,
+				&mode_lib->vba.StutterEnterPlusExitWatermark,
+				&mode_lib->vba.MinActiveDRAMClockChangeLatencySupported);
+	}
+
+
+	//Display Pipeline Delivery Time in Prefetch, Groups
+	CalculatePixelDeliveryTimes(
+		mode_lib->vba.NumberOfActivePlanes,
+		mode_lib->vba.VRatio,
+		locals->VRatioPrefetchY,
+		locals->VRatioPrefetchC,
+		locals->swath_width_luma_ub,
+		locals->swath_width_chroma_ub,
+		mode_lib->vba.DPPPerPlane,
+		mode_lib->vba.HRatio,
+		mode_lib->vba.PixelClock,
+		locals->PSCL_THROUGHPUT_LUMA,
+		locals->PSCL_THROUGHPUT_CHROMA,
+		locals->DPPCLK,
+		locals->BytePerPixelDETC,
+		mode_lib->vba.SourceScan,
+		locals->BlockWidth256BytesY,
+		locals->BlockHeight256BytesY,
+		locals->BlockWidth256BytesC,
+		locals->BlockHeight256BytesC,
+		locals->DisplayPipeLineDeliveryTimeLuma,
+		locals->DisplayPipeLineDeliveryTimeChroma,
+		locals->DisplayPipeLineDeliveryTimeLumaPrefetch,
+		locals->DisplayPipeLineDeliveryTimeChromaPrefetch,
+		locals->DisplayPipeRequestDeliveryTimeLuma,
+		locals->DisplayPipeRequestDeliveryTimeChroma,
+		locals->DisplayPipeRequestDeliveryTimeLumaPrefetch,
+		locals->DisplayPipeRequestDeliveryTimeChromaPrefetch);
+
+	CalculateMetaAndPTETimes(
+		mode_lib->vba.NumberOfActivePlanes,
+		mode_lib->vba.GPUVMEnable,
+		mode_lib->vba.MetaChunkSize,
+		mode_lib->vba.MinMetaChunkSizeBytes,
+		mode_lib->vba.GPUVMMaxPageTableLevels,
+		mode_lib->vba.HTotal,
+		mode_lib->vba.VRatio,
+		locals->VRatioPrefetchY,
+		locals->VRatioPrefetchC,
+		locals->DestinationLinesToRequestRowInVBlank,
+		locals->DestinationLinesToRequestRowInImmediateFlip,
+		locals->DestinationLinesToRequestVMInVBlank,
+		locals->DestinationLinesToRequestVMInImmediateFlip,
+		mode_lib->vba.DCCEnable,
+		mode_lib->vba.PixelClock,
+		locals->BytePerPixelDETY,
+		locals->BytePerPixelDETC,
+		mode_lib->vba.SourceScan,
+		locals->dpte_row_height,
+		locals->dpte_row_height_chroma,
+		locals->meta_row_width,
+		locals->meta_row_height,
+		locals->meta_req_width,
+		locals->meta_req_height,
+		locals->dpte_group_bytes,
+		locals->PTERequestSizeY,
+		locals->PTERequestSizeC,
+		locals->PixelPTEReqWidthY,
+		locals->PixelPTEReqHeightY,
+		locals->PixelPTEReqWidthC,
+		locals->PixelPTEReqHeightC,
+		locals->dpte_row_width_luma_ub,
+		locals->dpte_row_width_chroma_ub,
+		locals->vm_group_bytes,
+		locals->dpde0_bytes_per_frame_ub_l,
+		locals->dpde0_bytes_per_frame_ub_c,
+		locals->meta_pte_bytes_per_frame_ub_l,
+		locals->meta_pte_bytes_per_frame_ub_c,
+		locals->DST_Y_PER_PTE_ROW_NOM_L,
+		locals->DST_Y_PER_PTE_ROW_NOM_C,
+		locals->DST_Y_PER_META_ROW_NOM_L,
+		locals->TimePerMetaChunkNominal,
+		locals->TimePerMetaChunkVBlank,
+		locals->TimePerMetaChunkFlip,
+		locals->time_per_pte_group_nom_luma,
+		locals->time_per_pte_group_vblank_luma,
+		locals->time_per_pte_group_flip_luma,
+		locals->time_per_pte_group_nom_chroma,
+		locals->time_per_pte_group_vblank_chroma,
+		locals->time_per_pte_group_flip_chroma,
+		locals->TimePerVMGroupVBlank,
+		locals->TimePerVMGroupFlip,
+		locals->TimePerVMRequestVBlank,
+		locals->TimePerVMRequestFlip);
+
+
+	// Min TTUVBlank
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.PrefetchMode[mode_lib->vba.VoltageLevel][mode_lib->vba.maxMpcComb] == 0) {
+			locals->AllowDRAMClockChangeDuringVBlank[k] = true;
+			locals->AllowDRAMSelfRefreshDuringVBlank[k] = true;
+			locals->MinTTUVBlank[k] = dml_max(
+					mode_lib->vba.DRAMClockChangeWatermark,
+					dml_max(
+							mode_lib->vba.StutterEnterPlusExitWatermark,
+							mode_lib->vba.UrgentWatermark));
+		} else if (mode_lib->vba.PrefetchMode[mode_lib->vba.VoltageLevel][mode_lib->vba.maxMpcComb] == 1) {
+			locals->AllowDRAMClockChangeDuringVBlank[k] = false;
+			locals->AllowDRAMSelfRefreshDuringVBlank[k] = true;
+			locals->MinTTUVBlank[k] = dml_max(
+					mode_lib->vba.StutterEnterPlusExitWatermark,
+					mode_lib->vba.UrgentWatermark);
+		} else {
+			locals->AllowDRAMClockChangeDuringVBlank[k] = false;
+			locals->AllowDRAMSelfRefreshDuringVBlank[k] = false;
+			locals->MinTTUVBlank[k] = mode_lib->vba.UrgentWatermark;
+		}
+		if (!mode_lib->vba.DynamicMetadataEnable[k])
+			locals->MinTTUVBlank[k] = mode_lib->vba.TCalc
+					+ locals->MinTTUVBlank[k];
+	}
+
+	// DCC Configuration
+	mode_lib->vba.ActiveDPPs = 0;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		locals->MaximumDCCCompressionYSurface[k] = CalculateDCCConfiguration(
+			mode_lib->vba.DCCEnable[k],
+			false, // We should always know the direction DCCProgrammingAssumesScanDirectionUnknown,
+			mode_lib->vba.ViewportWidth[k],
+			mode_lib->vba.ViewportHeight[k],
+			mode_lib->vba.DETBufferSizeInKByte[0] * 1024,
+			locals->BlockHeight256BytesY[k],
+			mode_lib->vba.SwathHeightY[k],
+			mode_lib->vba.SurfaceTiling[k],
+			locals->BytePerPixelDETY[k],
+			mode_lib->vba.SourceScan[k],
+			&locals->DCCYMaxUncompressedBlock[k],
+			&locals->DCCYMaxCompressedBlock[k],
+			&locals->DCCYIndependent64ByteBlock[k]);
+	}
+
+	//XFC Parameters:
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.XFCEnabled[k] == true) {
+			double TWait;
+
+			locals->XFCSlaveVUpdateOffset[k] = mode_lib->vba.XFCTSlvVupdateOffset;
+			locals->XFCSlaveVupdateWidth[k] = mode_lib->vba.XFCTSlvVupdateWidth;
+			locals->XFCSlaveVReadyOffset[k] = mode_lib->vba.XFCTSlvVreadyOffset;
+			TWait = CalculateTWait(
+					mode_lib->vba.PrefetchMode[mode_lib->vba.VoltageLevel][mode_lib->vba.maxMpcComb],
+					mode_lib->vba.DRAMClockChangeLatency,
+					mode_lib->vba.UrgentLatency,
+					mode_lib->vba.SREnterPlusExitTime);
+			mode_lib->vba.XFCRemoteSurfaceFlipDelay = CalculateRemoteSurfaceFlipDelay(
+					mode_lib,
+					mode_lib->vba.VRatio[k],
+					locals->SwathWidthY[k],
+					dml_ceil(locals->BytePerPixelDETY[k], 1),
+					mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k],
+					mode_lib->vba.XFCTSlvVupdateOffset,
+					mode_lib->vba.XFCTSlvVupdateWidth,
+					mode_lib->vba.XFCTSlvVreadyOffset,
+					mode_lib->vba.XFCXBUFLatencyTolerance,
+					mode_lib->vba.XFCFillBWOverhead,
+					mode_lib->vba.XFCSlvChunkSize,
+					mode_lib->vba.XFCBusTransportTime,
+					mode_lib->vba.TCalc,
+					TWait,
+					&mode_lib->vba.SrcActiveDrainRate,
+					&mode_lib->vba.TInitXFill,
+					&mode_lib->vba.TslvChk);
+			locals->XFCRemoteSurfaceFlipLatency[k] =
+					dml_floor(
+							mode_lib->vba.XFCRemoteSurfaceFlipDelay
+									/ (mode_lib->vba.HTotal[k]
+											/ mode_lib->vba.PixelClock[k]),
+							1);
+			locals->XFCTransferDelay[k] =
+					dml_ceil(
+							mode_lib->vba.XFCBusTransportTime
+									/ (mode_lib->vba.HTotal[k]
+											/ mode_lib->vba.PixelClock[k]),
+							1);
+			locals->XFCPrechargeDelay[k] =
+					dml_ceil(
+							(mode_lib->vba.XFCBusTransportTime
+									+ mode_lib->vba.TInitXFill
+									+ mode_lib->vba.TslvChk)
+									/ (mode_lib->vba.HTotal[k]
+											/ mode_lib->vba.PixelClock[k]),
+							1);
+			mode_lib->vba.InitFillLevel = mode_lib->vba.XFCXBUFLatencyTolerance
+					* mode_lib->vba.SrcActiveDrainRate;
+			mode_lib->vba.FinalFillMargin =
+					(locals->DestinationLinesToRequestVMInVBlank[k]
+							+ locals->DestinationLinesToRequestRowInVBlank[k])
+							* mode_lib->vba.HTotal[k]
+							/ mode_lib->vba.PixelClock[k]
+							* mode_lib->vba.SrcActiveDrainRate
+							+ mode_lib->vba.XFCFillConstant;
+			mode_lib->vba.FinalFillLevel = mode_lib->vba.XFCRemoteSurfaceFlipDelay
+					* mode_lib->vba.SrcActiveDrainRate
+					+ mode_lib->vba.FinalFillMargin;
+			mode_lib->vba.RemainingFillLevel = dml_max(
+					0.0,
+					mode_lib->vba.FinalFillLevel - mode_lib->vba.InitFillLevel);
+			mode_lib->vba.TFinalxFill = mode_lib->vba.RemainingFillLevel
+					/ (mode_lib->vba.SrcActiveDrainRate
+							* mode_lib->vba.XFCFillBWOverhead / 100);
+			locals->XFCPrefetchMargin[k] =
+					mode_lib->vba.XFCRemoteSurfaceFlipDelay
+							+ mode_lib->vba.TFinalxFill
+							+ (locals->DestinationLinesToRequestVMInVBlank[k]
+									+ locals->DestinationLinesToRequestRowInVBlank[k])
+									* mode_lib->vba.HTotal[k]
+									/ mode_lib->vba.PixelClock[k];
+		} else {
+			locals->XFCSlaveVUpdateOffset[k] = 0;
+			locals->XFCSlaveVupdateWidth[k] = 0;
+			locals->XFCSlaveVReadyOffset[k] = 0;
+			locals->XFCRemoteSurfaceFlipLatency[k] = 0;
+			locals->XFCPrechargeDelay[k] = 0;
+			locals->XFCTransferDelay[k] = 0;
+			locals->XFCPrefetchMargin[k] = 0;
+		}
+	}
+
+	// Stutter Efficiency
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		CalculateDETBufferSize(
+			mode_lib->vba.DETBufferSizeInKByte[0],
+			mode_lib->vba.SwathHeightY[k],
+			mode_lib->vba.SwathHeightC[k],
+			&locals->DETBufferSizeY[k],
+			&locals->DETBufferSizeC[k]);
+
+		locals->LinesInDETY[k] = (double)locals->DETBufferSizeY[k]
+				/ locals->BytePerPixelDETY[k] / locals->SwathWidthY[k];
+		locals->LinesInDETYRoundedDownToSwath[k] = dml_floor(
+				locals->LinesInDETY[k],
+				mode_lib->vba.SwathHeightY[k]);
+		locals->FullDETBufferingTimeY[k] =
+				locals->LinesInDETYRoundedDownToSwath[k]
+						* (mode_lib->vba.HTotal[k]
+								/ mode_lib->vba.PixelClock[k])
+						/ mode_lib->vba.VRatio[k];
+	}
+
+	mode_lib->vba.StutterPeriod = 999999.0;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (locals->FullDETBufferingTimeY[k] < mode_lib->vba.StutterPeriod) {
+			mode_lib->vba.StutterPeriod = locals->FullDETBufferingTimeY[k];
+			mode_lib->vba.FrameTimeForMinFullDETBufferingTime =
+				(double) mode_lib->vba.VTotal[k] * mode_lib->vba.HTotal[k]
+				/ mode_lib->vba.PixelClock[k];
+			locals->BytePerPixelYCriticalPlane = dml_ceil(locals->BytePerPixelDETY[k], 1);
+			locals->SwathWidthYCriticalPlane = locals->SwathWidthY[k];
+			locals->LinesToFinishSwathTransferStutterCriticalPlane =
+				mode_lib->vba.SwathHeightY[k] - (locals->LinesInDETY[k] - locals->LinesInDETYRoundedDownToSwath[k]);
+		}
+	}
+
+	mode_lib->vba.AverageReadBandwidth = 0.0;
+	mode_lib->vba.TotalRowReadBandwidth = 0.0;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		unsigned int DCCRateLimit;
+
+		if (mode_lib->vba.DCCEnable[k]) {
+			if (locals->DCCYMaxCompressedBlock[k] == 256)
+				DCCRateLimit = 4;
+			else
+				DCCRateLimit = 2;
+
+			mode_lib->vba.AverageReadBandwidth =
+					mode_lib->vba.AverageReadBandwidth
+							+ (locals->ReadBandwidthPlaneLuma[k] + locals->ReadBandwidthPlaneChroma[k]) /
+								dml_min(mode_lib->vba.DCCRate[k], DCCRateLimit);
+		} else {
+			mode_lib->vba.AverageReadBandwidth =
+					mode_lib->vba.AverageReadBandwidth
+							+ locals->ReadBandwidthPlaneLuma[k]
+							+ locals->ReadBandwidthPlaneChroma[k];
+		}
+		mode_lib->vba.TotalRowReadBandwidth = mode_lib->vba.TotalRowReadBandwidth +
+		locals->meta_row_bw[k] + locals->dpte_row_bw[k];
+	}
+
+	mode_lib->vba.AverageDCCCompressionRate = mode_lib->vba.TotalDataReadBandwidth / mode_lib->vba.AverageReadBandwidth;
+
+	mode_lib->vba.PartOfBurstThatFitsInROB =
+			dml_min(
+					mode_lib->vba.StutterPeriod
+							* mode_lib->vba.TotalDataReadBandwidth,
+					mode_lib->vba.ROBBufferSizeInKByte * 1024
+							* mode_lib->vba.AverageDCCCompressionRate);
+	mode_lib->vba.StutterBurstTime = mode_lib->vba.PartOfBurstThatFitsInROB
+			/ mode_lib->vba.AverageDCCCompressionRate / mode_lib->vba.ReturnBW
+			+ (mode_lib->vba.StutterPeriod * mode_lib->vba.TotalDataReadBandwidth
+					- mode_lib->vba.PartOfBurstThatFitsInROB)
+					/ (mode_lib->vba.DCFCLK * 64)
+					+ mode_lib->vba.StutterPeriod * mode_lib->vba.TotalRowReadBandwidth / mode_lib->vba.ReturnBW;
+	mode_lib->vba.StutterBurstTime = dml_max(
+		mode_lib->vba.StutterBurstTime,
+		(locals->LinesToFinishSwathTransferStutterCriticalPlane * locals->BytePerPixelYCriticalPlane *
+		locals->SwathWidthYCriticalPlane / mode_lib->vba.ReturnBW)
+	);
+
+	mode_lib->vba.TotalActiveWriteback = 0;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.WritebackEnable[k] == true) {
+			mode_lib->vba.TotalActiveWriteback = mode_lib->vba.TotalActiveWriteback + 1;
+		}
+	}
+
+	if (mode_lib->vba.TotalActiveWriteback == 0) {
+		mode_lib->vba.StutterEfficiencyNotIncludingVBlank = (1
+				- (mode_lib->vba.SRExitTime + mode_lib->vba.StutterBurstTime)
+						/ mode_lib->vba.StutterPeriod) * 100;
+	} else {
+		mode_lib->vba.StutterEfficiencyNotIncludingVBlank = 0;
+	}
+
+	mode_lib->vba.SmallestVBlank = 999999;
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		if (mode_lib->vba.SynchronizedVBlank || mode_lib->vba.NumberOfActivePlanes == 1) {
+			mode_lib->vba.VBlankTime = (double) (mode_lib->vba.VTotal[k]
+					- mode_lib->vba.VActive[k]) * mode_lib->vba.HTotal[k]
+					/ mode_lib->vba.PixelClock[k];
+		} else {
+			mode_lib->vba.VBlankTime = 0;
+		}
+		mode_lib->vba.SmallestVBlank = dml_min(
+				mode_lib->vba.SmallestVBlank,
+				mode_lib->vba.VBlankTime);
+	}
+
+	mode_lib->vba.StutterEfficiency = (mode_lib->vba.StutterEfficiencyNotIncludingVBlank / 100
+			* (mode_lib->vba.FrameTimeForMinFullDETBufferingTime
+					- mode_lib->vba.SmallestVBlank)
+			+ mode_lib->vba.SmallestVBlank)
+			/ mode_lib->vba.FrameTimeForMinFullDETBufferingTime * 100;
+}
+
+static void DisplayPipeConfiguration(struct display_mode_lib *mode_lib)
+{
+	// Display Pipe Configuration
+	double BytePerPixDETY;
+	double BytePerPixDETC;
+	double Read256BytesBlockHeightY;
+	double Read256BytesBlockHeightC;
+	double Read256BytesBlockWidthY;
+	double Read256BytesBlockWidthC;
+	double MaximumSwathHeightY;
+	double MaximumSwathHeightC;
+	double MinimumSwathHeightY;
+	double MinimumSwathHeightC;
+	double SwathWidth;
+	double SwathWidthGranularityY;
+	double SwathWidthGranularityC;
+	double RoundedUpMaxSwathSizeBytesY;
+	double RoundedUpMaxSwathSizeBytesC;
+	unsigned int j, k;
+
+	for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+		bool MainPlaneDoesODMCombine = false;
+
+		if (mode_lib->vba.SourcePixelFormat[k] == dm_444_64) {
+			BytePerPixDETY = 8;
+			BytePerPixDETC = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_32) {
+			BytePerPixDETY = 4;
+			BytePerPixDETC = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_16) {
+			BytePerPixDETY = 2;
+			BytePerPixDETC = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_8) {
+			BytePerPixDETY = 1;
+			BytePerPixDETC = 0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_8) {
+			BytePerPixDETY = 1;
+			BytePerPixDETC = 2;
+		} else {
+			BytePerPixDETY = 4.0 / 3.0;
+			BytePerPixDETC = 8.0 / 3.0;
+		}
+
+		if ((mode_lib->vba.SourcePixelFormat[k] == dm_444_64
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_32
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_16
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_8)) {
+			if (mode_lib->vba.SurfaceTiling[k] == dm_sw_linear) {
+				Read256BytesBlockHeightY = 1;
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_64) {
+				Read256BytesBlockHeightY = 4;
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_32
+					|| mode_lib->vba.SourcePixelFormat[k] == dm_444_16) {
+				Read256BytesBlockHeightY = 8;
+			} else {
+				Read256BytesBlockHeightY = 16;
+			}
+			Read256BytesBlockWidthY = 256 / dml_ceil(BytePerPixDETY, 1)
+					/ Read256BytesBlockHeightY;
+			Read256BytesBlockHeightC = 0;
+			Read256BytesBlockWidthC = 0;
+		} else {
+			if (mode_lib->vba.SurfaceTiling[k] == dm_sw_linear) {
+				Read256BytesBlockHeightY = 1;
+				Read256BytesBlockHeightC = 1;
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_8) {
+				Read256BytesBlockHeightY = 16;
+				Read256BytesBlockHeightC = 8;
+			} else {
+				Read256BytesBlockHeightY = 8;
+				Read256BytesBlockHeightC = 8;
+			}
+			Read256BytesBlockWidthY = 256 / dml_ceil(BytePerPixDETY, 1)
+					/ Read256BytesBlockHeightY;
+			Read256BytesBlockWidthC = 256 / dml_ceil(BytePerPixDETC, 2)
+					/ Read256BytesBlockHeightC;
+		}
+
+		if (mode_lib->vba.SourceScan[k] == dm_horz) {
+			MaximumSwathHeightY = Read256BytesBlockHeightY;
+			MaximumSwathHeightC = Read256BytesBlockHeightC;
+		} else {
+			MaximumSwathHeightY = Read256BytesBlockWidthY;
+			MaximumSwathHeightC = Read256BytesBlockWidthC;
+		}
+
+		if ((mode_lib->vba.SourcePixelFormat[k] == dm_444_64
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_32
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_16
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_8)) {
+			if (mode_lib->vba.SurfaceTiling[k] == dm_sw_linear
+					|| (mode_lib->vba.SourcePixelFormat[k] == dm_444_64
+							&& (mode_lib->vba.SurfaceTiling[k]
+									== dm_sw_4kb_s
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_4kb_s_x
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_64kb_s
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_64kb_s_t
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_64kb_s_x
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_var_s
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_var_s_x)
+							&& mode_lib->vba.SourceScan[k] == dm_horz)) {
+				MinimumSwathHeightY = MaximumSwathHeightY;
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_8
+					&& mode_lib->vba.SourceScan[k] != dm_horz) {
+				MinimumSwathHeightY = MaximumSwathHeightY;
+			} else {
+				MinimumSwathHeightY = MaximumSwathHeightY / 2.0;
+			}
+			MinimumSwathHeightC = MaximumSwathHeightC;
+		} else {
+			if (mode_lib->vba.SurfaceTiling[k] == dm_sw_linear) {
+				MinimumSwathHeightY = MaximumSwathHeightY;
+				MinimumSwathHeightC = MaximumSwathHeightC;
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_8
+					&& mode_lib->vba.SourceScan[k] == dm_horz) {
+				MinimumSwathHeightY = MaximumSwathHeightY / 2.0;
+				MinimumSwathHeightC = MaximumSwathHeightC;
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_10
+					&& mode_lib->vba.SourceScan[k] == dm_horz) {
+				MinimumSwathHeightC = MaximumSwathHeightC / 2.0;
+				MinimumSwathHeightY = MaximumSwathHeightY;
+			} else {
+				MinimumSwathHeightY = MaximumSwathHeightY;
+				MinimumSwathHeightC = MaximumSwathHeightC;
+			}
+		}
+
+		if (mode_lib->vba.SourceScan[k] == dm_horz) {
+			SwathWidth = mode_lib->vba.ViewportWidth[k];
+		} else {
+			SwathWidth = mode_lib->vba.ViewportHeight[k];
+		}
+
+		if (mode_lib->vba.ODMCombineEnabled[k] == dm_odm_combine_mode_2to1) {
+			MainPlaneDoesODMCombine = true;
+		}
+		for (j = 0; j < mode_lib->vba.NumberOfActivePlanes; ++j) {
+			if (mode_lib->vba.BlendingAndTiming[k] == j
+					&& mode_lib->vba.ODMCombineEnabled[k] == dm_odm_combine_mode_2to1) {
+				MainPlaneDoesODMCombine = true;
+			}
+		}
+
+		if (MainPlaneDoesODMCombine == true) {
+			SwathWidth = dml_min(
+					SwathWidth,
+					mode_lib->vba.HActive[k] / 2.0 * mode_lib->vba.HRatio[k]);
+		} else {
+			SwathWidth = SwathWidth / mode_lib->vba.DPPPerPlane[k];
+		}
+
+		SwathWidthGranularityY = 256 / dml_ceil(BytePerPixDETY, 1) / MaximumSwathHeightY;
+		RoundedUpMaxSwathSizeBytesY = (dml_ceil(
+				(double) (SwathWidth - 1),
+				SwathWidthGranularityY) + SwathWidthGranularityY) * BytePerPixDETY
+				* MaximumSwathHeightY;
+		if (mode_lib->vba.SourcePixelFormat[k] == dm_420_10) {
+			RoundedUpMaxSwathSizeBytesY = dml_ceil(RoundedUpMaxSwathSizeBytesY, 256)
+					+ 256;
+		}
+		if (MaximumSwathHeightC > 0) {
+			SwathWidthGranularityC = 256.0 / dml_ceil(BytePerPixDETC, 2)
+					/ MaximumSwathHeightC;
+			RoundedUpMaxSwathSizeBytesC = (dml_ceil(
+					(double) (SwathWidth / 2.0 - 1),
+					SwathWidthGranularityC) + SwathWidthGranularityC)
+					* BytePerPixDETC * MaximumSwathHeightC;
+			if (mode_lib->vba.SourcePixelFormat[k] == dm_420_10) {
+				RoundedUpMaxSwathSizeBytesC = dml_ceil(
+						RoundedUpMaxSwathSizeBytesC,
+						256) + 256;
+			}
+		} else
+			RoundedUpMaxSwathSizeBytesC = 0.0;
+
+		if (RoundedUpMaxSwathSizeBytesY + RoundedUpMaxSwathSizeBytesC
+				<= mode_lib->vba.DETBufferSizeInKByte[0] * 1024.0 / 2.0) {
+			mode_lib->vba.SwathHeightY[k] = MaximumSwathHeightY;
+			mode_lib->vba.SwathHeightC[k] = MaximumSwathHeightC;
+		} else {
+			mode_lib->vba.SwathHeightY[k] = MinimumSwathHeightY;
+			mode_lib->vba.SwathHeightC[k] = MinimumSwathHeightC;
+		}
+
+		CalculateDETBufferSize(
+				mode_lib->vba.DETBufferSizeInKByte[0],
+				mode_lib->vba.SwathHeightY[k],
+				mode_lib->vba.SwathHeightC[k],
+				&mode_lib->vba.DETBufferSizeY[k],
+				&mode_lib->vba.DETBufferSizeC[k]);
+	}
+}
+
+static double CalculateTWait(
+		unsigned int PrefetchMode,
+		double DRAMClockChangeLatency,
+		double UrgentLatency,
+		double SREnterPlusExitTime)
+{
+	if (PrefetchMode == 0) {
+		return dml_max(
+				DRAMClockChangeLatency + UrgentLatency,
+				dml_max(SREnterPlusExitTime, UrgentLatency));
+	} else if (PrefetchMode == 1) {
+		return dml_max(SREnterPlusExitTime, UrgentLatency);
+	} else {
+		return UrgentLatency;
+	}
+}
+
+static double CalculateRemoteSurfaceFlipDelay(
+		struct display_mode_lib *mode_lib,
+		double VRatio,
+		double SwathWidth,
+		double Bpp,
+		double LineTime,
+		double XFCTSlvVupdateOffset,
+		double XFCTSlvVupdateWidth,
+		double XFCTSlvVreadyOffset,
+		double XFCXBUFLatencyTolerance,
+		double XFCFillBWOverhead,
+		double XFCSlvChunkSize,
+		double XFCBusTransportTime,
+		double TCalc,
+		double TWait,
+		double *SrcActiveDrainRate,
+		double *TInitXFill,
+		double *TslvChk)
+{
+	double TSlvSetup, AvgfillRate, result;
+
+	*SrcActiveDrainRate = VRatio * SwathWidth * Bpp / LineTime;
+	TSlvSetup = XFCTSlvVupdateOffset + XFCTSlvVupdateWidth + XFCTSlvVreadyOffset;
+	*TInitXFill = XFCXBUFLatencyTolerance / (1 + XFCFillBWOverhead / 100);
+	AvgfillRate = *SrcActiveDrainRate * (1 + XFCFillBWOverhead / 100);
+	*TslvChk = XFCSlvChunkSize / AvgfillRate;
+	dml_print(
+			"DML::CalculateRemoteSurfaceFlipDelay: SrcActiveDrainRate: %f\n",
+			*SrcActiveDrainRate);
+	dml_print("DML::CalculateRemoteSurfaceFlipDelay: TSlvSetup: %f\n", TSlvSetup);
+	dml_print("DML::CalculateRemoteSurfaceFlipDelay: TInitXFill: %f\n", *TInitXFill);
+	dml_print("DML::CalculateRemoteSurfaceFlipDelay: AvgfillRate: %f\n", AvgfillRate);
+	dml_print("DML::CalculateRemoteSurfaceFlipDelay: TslvChk: %f\n", *TslvChk);
+	result = 2 * XFCBusTransportTime + TSlvSetup + TCalc + TWait + *TslvChk + *TInitXFill; // TODO: This doesn't seem to match programming guide
+	dml_print("DML::CalculateRemoteSurfaceFlipDelay: RemoteSurfaceFlipDelay: %f\n", result);
+	return result;
+}
+
+static double CalculateWriteBackDelay(
+		enum source_format_class WritebackPixelFormat,
+		double WritebackHRatio,
+		double WritebackVRatio,
+		unsigned int WritebackLumaHTaps,
+		unsigned int WritebackLumaVTaps,
+		unsigned int WritebackChromaHTaps,
+		unsigned int WritebackChromaVTaps,
+		unsigned int WritebackDestinationWidth)
+{
+	double CalculateWriteBackDelay =
+			dml_max(
+					dml_ceil(WritebackLumaHTaps / 4.0, 1) / WritebackHRatio,
+					WritebackLumaVTaps * dml_ceil(1.0 / WritebackVRatio, 1)
+							* dml_ceil(
+									WritebackDestinationWidth
+											/ 4.0,
+									1)
+							+ dml_ceil(1.0 / WritebackVRatio, 1)
+									* (dml_ceil(
+											WritebackLumaVTaps
+													/ 4.0,
+											1) + 4));
+
+	if (WritebackPixelFormat != dm_444_32) {
+		CalculateWriteBackDelay =
+				dml_max(
+						CalculateWriteBackDelay,
+						dml_max(
+								dml_ceil(
+										WritebackChromaHTaps
+												/ 2.0,
+										1)
+										/ (2
+												* WritebackHRatio),
+								WritebackChromaVTaps
+										* dml_ceil(
+												1
+														/ (2
+																* WritebackVRatio),
+												1)
+										* dml_ceil(
+												WritebackDestinationWidth
+														/ 2.0
+														/ 2.0,
+												1)
+										+ dml_ceil(
+												1
+														/ (2
+																* WritebackVRatio),
+												1)
+												* (dml_ceil(
+														WritebackChromaVTaps
+																/ 4.0,
+														1)
+														+ 4)));
+	}
+	return CalculateWriteBackDelay;
+}
+
+static void CalculateActiveRowBandwidth(
+		bool GPUVMEnable,
+		enum source_format_class SourcePixelFormat,
+		double VRatio,
+		bool DCCEnable,
+		double LineTime,
+		unsigned int MetaRowByteLuma,
+		unsigned int MetaRowByteChroma,
+		unsigned int meta_row_height_luma,
+		unsigned int meta_row_height_chroma,
+		unsigned int PixelPTEBytesPerRowLuma,
+		unsigned int PixelPTEBytesPerRowChroma,
+		unsigned int dpte_row_height_luma,
+		unsigned int dpte_row_height_chroma,
+		double *meta_row_bw,
+		double *dpte_row_bw)
+{
+	if (DCCEnable != true) {
+		*meta_row_bw = 0;
+	} else if (SourcePixelFormat == dm_420_8 || SourcePixelFormat == dm_420_10) {
+		*meta_row_bw = VRatio * MetaRowByteLuma / (meta_row_height_luma * LineTime)
+				+ VRatio / 2 * MetaRowByteChroma
+						/ (meta_row_height_chroma * LineTime);
+	} else {
+		*meta_row_bw = VRatio * MetaRowByteLuma / (meta_row_height_luma * LineTime);
+	}
+
+	if (GPUVMEnable != true) {
+		*dpte_row_bw = 0;
+	} else if (SourcePixelFormat == dm_420_8 || SourcePixelFormat == dm_420_10) {
+		*dpte_row_bw = VRatio * PixelPTEBytesPerRowLuma / (dpte_row_height_luma * LineTime)
+				+ VRatio / 2 * PixelPTEBytesPerRowChroma
+						/ (dpte_row_height_chroma * LineTime);
+	} else {
+		*dpte_row_bw = VRatio * PixelPTEBytesPerRowLuma / (dpte_row_height_luma * LineTime);
+	}
+}
+
+static void CalculateFlipSchedule(
+		struct display_mode_lib *mode_lib,
+		double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData,
+		double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly,
+		double UrgentExtraLatency,
+		double UrgentLatency,
+		unsigned int GPUVMMaxPageTableLevels,
+		bool HostVMEnable,
+		unsigned int HostVMMaxPageTableLevels,
+		unsigned int HostVMCachedPageTableLevels,
+		bool GPUVMEnable,
+		double PDEAndMetaPTEBytesPerFrame,
+		double MetaRowBytes,
+		double DPTEBytesPerRow,
+		double BandwidthAvailableForImmediateFlip,
+		unsigned int TotImmediateFlipBytes,
+		enum source_format_class SourcePixelFormat,
+		double LineTime,
+		double VRatio,
+		double Tno_bw,
+		bool DCCEnable,
+		unsigned int dpte_row_height,
+		unsigned int meta_row_height,
+		unsigned int dpte_row_height_chroma,
+		unsigned int meta_row_height_chroma,
+		double *DestinationLinesToRequestVMInImmediateFlip,
+		double *DestinationLinesToRequestRowInImmediateFlip,
+		double *final_flip_bw,
+		bool *ImmediateFlipSupportedForPipe)
+{
+	double min_row_time = 0.0;
+	unsigned int HostVMDynamicLevels;
+	double TimeForFetchingMetaPTEImmediateFlip;
+	double TimeForFetchingRowInVBlankImmediateFlip;
+	double ImmediateFlipBW = 1.0;
+	double HostVMInefficiencyFactor;
+	double VRatioClamped;
+
+	if (GPUVMEnable == true && HostVMEnable == true) {
+		HostVMInefficiencyFactor =
+				PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData
+						/ PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly;
+		HostVMDynamicLevels = HostVMMaxPageTableLevels - HostVMCachedPageTableLevels;
+	} else {
+		HostVMInefficiencyFactor = 1;
+		HostVMDynamicLevels = 0;
+	}
+
+	ImmediateFlipBW = (PDEAndMetaPTEBytesPerFrame + MetaRowBytes + DPTEBytesPerRow)
+			* BandwidthAvailableForImmediateFlip / TotImmediateFlipBytes;
+
+	if (GPUVMEnable == true) {
+		TimeForFetchingMetaPTEImmediateFlip = dml_max3(
+			Tno_bw + PDEAndMetaPTEBytesPerFrame * HostVMInefficiencyFactor / ImmediateFlipBW,
+			UrgentExtraLatency + UrgentLatency * (GPUVMMaxPageTableLevels * (HostVMDynamicLevels + 1) - 1),
+			LineTime / 4.0);
+	} else {
+		TimeForFetchingMetaPTEImmediateFlip = 0;
+	}
+
+	*DestinationLinesToRequestVMInImmediateFlip = dml_ceil(4.0 * (TimeForFetchingMetaPTEImmediateFlip / LineTime), 1) / 4.0;
+	if ((GPUVMEnable == true || DCCEnable == true)) {
+		TimeForFetchingRowInVBlankImmediateFlip = dml_max3((MetaRowBytes + DPTEBytesPerRow) * HostVMInefficiencyFactor / ImmediateFlipBW, UrgentLatency * (HostVMDynamicLevels + 1), LineTime / 4);
+	} else {
+		TimeForFetchingRowInVBlankImmediateFlip = 0;
+	}
+
+	*DestinationLinesToRequestRowInImmediateFlip = dml_ceil(4.0 * (TimeForFetchingRowInVBlankImmediateFlip / LineTime), 1) / 4.0;
+	*final_flip_bw = dml_max(PDEAndMetaPTEBytesPerFrame * HostVMInefficiencyFactor / (*DestinationLinesToRequestVMInImmediateFlip * LineTime), (MetaRowBytes + DPTEBytesPerRow) * HostVMInefficiencyFactor / (*DestinationLinesToRequestRowInImmediateFlip * LineTime));
+	VRatioClamped = (VRatio < 1.0) ? 1.0 : VRatio;
+	if (SourcePixelFormat == dm_420_8 || SourcePixelFormat == dm_420_10) {
+		if (GPUVMEnable == true && DCCEnable != true) {
+			min_row_time = dml_min(
+					dpte_row_height * LineTime / VRatioClamped,
+					dpte_row_height_chroma * LineTime / (VRatioClamped / 2));
+		} else if (GPUVMEnable != true && DCCEnable == true) {
+			min_row_time = dml_min(
+					meta_row_height * LineTime / VRatioClamped,
+					meta_row_height_chroma * LineTime / (VRatioClamped / 2));
+		} else {
+			min_row_time = dml_min4(
+					dpte_row_height * LineTime / VRatioClamped,
+					meta_row_height * LineTime / VRatioClamped,
+					dpte_row_height_chroma * LineTime / (VRatioClamped / 2),
+					meta_row_height_chroma * LineTime / (VRatioClamped / 2));
+		}
+	} else {
+		if (GPUVMEnable == true && DCCEnable != true) {
+			min_row_time = dpte_row_height * LineTime / VRatioClamped;
+		} else if (GPUVMEnable != true && DCCEnable == true) {
+			min_row_time = meta_row_height * LineTime / VRatioClamped;
+		} else {
+			min_row_time = dml_min(
+					dpte_row_height * LineTime / VRatioClamped,
+					meta_row_height * LineTime / VRatioClamped);
+		}
+	}
+
+	if (*DestinationLinesToRequestVMInImmediateFlip >= 32
+			|| *DestinationLinesToRequestRowInImmediateFlip >= 16
+			|| TimeForFetchingMetaPTEImmediateFlip + 2 * TimeForFetchingRowInVBlankImmediateFlip > min_row_time) {
+		*ImmediateFlipSupportedForPipe = false;
+	} else {
+		*ImmediateFlipSupportedForPipe = true;
+	}
+}
+
+static unsigned int TruncToValidBPP(
+		double DecimalBPP,
+		double DesiredBPP,
+		bool DSCEnabled,
+		enum output_encoder_class Output,
+		enum output_format_class Format,
+		unsigned int DSCInputBitPerComponent)
+{
+	if (Output == dm_hdmi) {
+		if (Format == dm_420) {
+			if (DecimalBPP >= 18 && (DesiredBPP == 0 || DesiredBPP == 18))
+				return 18;
+			else if (DecimalBPP >= 15 && (DesiredBPP == 0 || DesiredBPP == 15))
+				return 15;
+			else if (DecimalBPP >= 12 && (DesiredBPP == 0 || DesiredBPP == 12))
+				return 12;
+			else
+				return BPP_INVALID;
+		} else if (Format == dm_444) {
+			if (DecimalBPP >= 36 && (DesiredBPP == 0 || DesiredBPP == 36))
+				return 36;
+			else if (DecimalBPP >= 30 && (DesiredBPP == 0 || DesiredBPP == 30))
+				return 30;
+			else if (DecimalBPP >= 24 && (DesiredBPP == 0 || DesiredBPP == 24))
+				return 24;
+			else if (DecimalBPP >= 18 && (DesiredBPP == 0 || DesiredBPP == 18))
+				return 18;
+			else
+				return BPP_INVALID;
+		} else {
+			if (DecimalBPP / 1.5 >= 24 && (DesiredBPP == 0 || DesiredBPP == 24))
+				return 24;
+			else if (DecimalBPP / 1.5 >= 20 && (DesiredBPP == 0 || DesiredBPP == 20))
+				return 20;
+			else if (DecimalBPP / 1.5 >= 16 && (DesiredBPP == 0 || DesiredBPP == 16))
+				return 16;
+			else
+				return BPP_INVALID;
+		}
+	} else {
+		if (DSCEnabled) {
+			if (Format == dm_420) {
+				if (DesiredBPP == 0) {
+					if (DecimalBPP < 6)
+						return BPP_INVALID;
+					else if (DecimalBPP >= 1.5 * DSCInputBitPerComponent - 1.0 / 16.0)
+						return 1.5 * DSCInputBitPerComponent - 1.0 / 16.0;
+					else
+						return dml_floor(16 * DecimalBPP, 1) / 16.0;
+				} else {
+					if (DecimalBPP < 6
+							|| DesiredBPP < 6
+							|| DesiredBPP > 1.5 * DSCInputBitPerComponent - 1.0 / 16.0
+							|| DecimalBPP < DesiredBPP) {
+						return BPP_INVALID;
+					} else {
+						return DesiredBPP;
+					}
+				}
+			} else if (Format == dm_n422) {
+				if (DesiredBPP == 0) {
+					if (DecimalBPP < 7)
+						return BPP_INVALID;
+					else if (DecimalBPP >= 2 * DSCInputBitPerComponent - 1.0 / 16.0)
+						return 2 * DSCInputBitPerComponent - 1.0 / 16.0;
+					else
+						return dml_floor(16 * DecimalBPP, 1) / 16.0;
+				} else {
+					if (DecimalBPP < 7
+							|| DesiredBPP < 7
+							|| DesiredBPP > 2 * DSCInputBitPerComponent - 1.0 / 16.0
+							|| DecimalBPP < DesiredBPP) {
+						return BPP_INVALID;
+					} else {
+						return DesiredBPP;
+					}
+				}
+			} else {
+				if (DesiredBPP == 0) {
+					if (DecimalBPP < 8)
+						return BPP_INVALID;
+					else if (DecimalBPP >= 3 * DSCInputBitPerComponent - 1.0 / 16.0)
+						return 3 * DSCInputBitPerComponent - 1.0 / 16.0;
+					else
+						return dml_floor(16 * DecimalBPP, 1) / 16.0;
+				} else {
+					if (DecimalBPP < 8
+							|| DesiredBPP < 8
+							|| DesiredBPP > 3 * DSCInputBitPerComponent - 1.0 / 16.0
+							|| DecimalBPP < DesiredBPP) {
+						return BPP_INVALID;
+					} else {
+						return DesiredBPP;
+					}
+				}
+			}
+		} else if (Format == dm_420) {
+			if (DecimalBPP >= 18 && (DesiredBPP == 0 || DesiredBPP == 18))
+				return 18;
+			else if (DecimalBPP >= 15 && (DesiredBPP == 0 || DesiredBPP == 15))
+				return 15;
+			else if (DecimalBPP >= 12 && (DesiredBPP == 0 || DesiredBPP == 12))
+				return 12;
+			else
+				return BPP_INVALID;
+		} else if (Format == dm_s422 || Format == dm_n422) {
+			if (DecimalBPP >= 24 && (DesiredBPP == 0 || DesiredBPP == 24))
+				return 24;
+			else if (DecimalBPP >= 20 && (DesiredBPP == 0 || DesiredBPP == 20))
+				return 20;
+			else if (DecimalBPP >= 16 && (DesiredBPP == 0 || DesiredBPP == 16))
+				return 16;
+			else
+				return BPP_INVALID;
+		} else {
+			if (DecimalBPP >= 36 && (DesiredBPP == 0 || DesiredBPP == 36))
+				return 36;
+			else if (DecimalBPP >= 30 && (DesiredBPP == 0 || DesiredBPP == 30))
+				return 30;
+			else if (DecimalBPP >= 24 && (DesiredBPP == 0 || DesiredBPP == 24))
+				return 24;
+			else if (DecimalBPP >= 18 && (DesiredBPP == 0 || DesiredBPP == 18))
+				return 18;
+			else
+				return BPP_INVALID;
+		}
+	}
+}
+
+
+static noinline void CalculatePrefetchSchedulePerPlane(
+		struct display_mode_lib *mode_lib,
+		int i,
+		unsigned j,
+		unsigned k)
+{
+	struct vba_vars_st *locals = &mode_lib->vba;
+	Pipe myPipe;
+	HostVM myHostVM;
+
+	if (mode_lib->vba.XFCEnabled[k] == true) {
+		mode_lib->vba.XFCRemoteSurfaceFlipDelay =
+				CalculateRemoteSurfaceFlipDelay(
+						mode_lib,
+						mode_lib->vba.VRatio[k],
+						locals->SwathWidthYThisState[k],
+						dml_ceil(locals->BytePerPixelInDETY[k], 1.0),
+						mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k],
+						mode_lib->vba.XFCTSlvVupdateOffset,
+						mode_lib->vba.XFCTSlvVupdateWidth,
+						mode_lib->vba.XFCTSlvVreadyOffset,
+						mode_lib->vba.XFCXBUFLatencyTolerance,
+						mode_lib->vba.XFCFillBWOverhead,
+						mode_lib->vba.XFCSlvChunkSize,
+						mode_lib->vba.XFCBusTransportTime,
+						mode_lib->vba.TimeCalc,
+						mode_lib->vba.TWait,
+						&mode_lib->vba.SrcActiveDrainRate,
+						&mode_lib->vba.TInitXFill,
+						&mode_lib->vba.TslvChk);
+	} else {
+		mode_lib->vba.XFCRemoteSurfaceFlipDelay = 0.0;
+	}
+
+	myPipe.DPPCLK = locals->RequiredDPPCLK[i][j][k];
+	myPipe.DISPCLK = locals->RequiredDISPCLK[i][j];
+	myPipe.PixelClock = mode_lib->vba.PixelClock[k];
+	myPipe.DCFCLKDeepSleep = mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0];
+	myPipe.DPPPerPlane = locals->NoOfDPP[i][j][k];
+	myPipe.ScalerEnabled = mode_lib->vba.ScalerEnabled[k];
+	myPipe.SourceScan = mode_lib->vba.SourceScan[k];
+	myPipe.BlockWidth256BytesY = locals->Read256BlockWidthY[k];
+	myPipe.BlockHeight256BytesY = locals->Read256BlockHeightY[k];
+	myPipe.BlockWidth256BytesC = locals->Read256BlockWidthC[k];
+	myPipe.BlockHeight256BytesC = locals->Read256BlockHeightC[k];
+	myPipe.InterlaceEnable = mode_lib->vba.Interlace[k];
+	myPipe.NumberOfCursors = mode_lib->vba.NumberOfCursors[k];
+	myPipe.VBlank = mode_lib->vba.VTotal[k] - mode_lib->vba.VActive[k];
+	myPipe.HTotal = mode_lib->vba.HTotal[k];
+
+
+	myHostVM.Enable = mode_lib->vba.HostVMEnable;
+	myHostVM.MaxPageTableLevels = mode_lib->vba.HostVMMaxPageTableLevels;
+	myHostVM.CachedPageTableLevels = mode_lib->vba.HostVMCachedPageTableLevels;
+
+
+	mode_lib->vba.IsErrorResult[i][j][k] = CalculatePrefetchSchedule(
+			mode_lib,
+			mode_lib->vba.PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData,
+			mode_lib->vba.PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly,
+			&myPipe,
+			locals->DSCDelayPerState[i][k],
+			mode_lib->vba.DPPCLKDelaySubtotal,
+			mode_lib->vba.DPPCLKDelaySCL,
+			mode_lib->vba.DPPCLKDelaySCLLBOnly,
+			mode_lib->vba.DPPCLKDelayCNVCFormater,
+			mode_lib->vba.DPPCLKDelayCNVCCursor,
+			mode_lib->vba.DISPCLKDelaySubtotal,
+			locals->SwathWidthYThisState[k] / mode_lib->vba.HRatio[k],
+			mode_lib->vba.OutputFormat[k],
+			mode_lib->vba.MaxInterDCNTileRepeaters,
+			dml_min(mode_lib->vba.MaxVStartup, locals->MaximumVStartup[0][0][k]),
+			locals->MaximumVStartup[0][0][k],
+			mode_lib->vba.GPUVMMaxPageTableLevels,
+			mode_lib->vba.GPUVMEnable,
+			&myHostVM,
+			mode_lib->vba.DynamicMetadataEnable[k],
+			mode_lib->vba.DynamicMetadataLinesBeforeActiveRequired[k],
+			mode_lib->vba.DynamicMetadataTransmittedBytes[k],
+			mode_lib->vba.DCCEnable[k],
+			mode_lib->vba.UrgentLatency,
+			mode_lib->vba.ExtraLatency,
+			mode_lib->vba.TimeCalc,
+			locals->PDEAndMetaPTEBytesPerFrame[0][0][k],
+			locals->MetaRowBytes[0][0][k],
+			locals->DPTEBytesPerRow[0][0][k],
+			locals->PrefetchLinesY[0][0][k],
+			locals->SwathWidthYThisState[k],
+			locals->BytePerPixelInDETY[k],
+			locals->PrefillY[k],
+			locals->MaxNumSwY[k],
+			locals->PrefetchLinesC[0][0][k],
+			locals->BytePerPixelInDETC[k],
+			locals->PrefillC[k],
+			locals->MaxNumSwC[k],
+			locals->SwathHeightYThisState[k],
+			locals->SwathHeightCThisState[k],
+			mode_lib->vba.TWait,
+			mode_lib->vba.XFCEnabled[k],
+			mode_lib->vba.XFCRemoteSurfaceFlipDelay,
+			mode_lib->vba.ProgressiveToInterlaceUnitInOPP,
+			&locals->dst_x_after_scaler,
+			&locals->dst_y_after_scaler,
+			&locals->LineTimesForPrefetch[k],
+			&locals->PrefetchBW[k],
+			&locals->LinesForMetaPTE[k],
+			&locals->LinesForMetaAndDPTERow[k],
+			&locals->VRatioPreY[i][j][k],
+			&locals->VRatioPreC[i][j][k],
+			&locals->RequiredPrefetchPixelDataBWLuma[i][j][k],
+			&locals->RequiredPrefetchPixelDataBWChroma[i][j][k],
+			&locals->VStartupRequiredWhenNotEnoughTimeForDynamicMetadata,
+			&locals->Tno_bw[k],
+			&locals->prefetch_vmrow_bw[k],
+			locals->swath_width_luma_ub,
+			locals->swath_width_chroma_ub,
+			&mode_lib->vba.VUpdateOffsetPix[k],
+			&mode_lib->vba.VUpdateWidthPix[k],
+			&mode_lib->vba.VReadyOffsetPix[k]);
+}
+void dml21_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_lib)
+{
+	struct vba_vars_st *locals = &mode_lib->vba;
+
+	int i;
+	unsigned int j, k, m;
+
+	/*MODE SUPPORT, VOLTAGE STATE AND SOC CONFIGURATION*/
+
+	/*Scale Ratio, taps Support Check*/
+
+	mode_lib->vba.ScaleRatioAndTapsSupport = true;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.ScalerEnabled[k] == false
+				&& ((mode_lib->vba.SourcePixelFormat[k] != dm_444_64
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_444_32
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_444_16
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_mono_16
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_mono_8)
+						|| mode_lib->vba.HRatio[k] != 1.0
+						|| mode_lib->vba.htaps[k] != 1.0
+						|| mode_lib->vba.VRatio[k] != 1.0
+						|| mode_lib->vba.vtaps[k] != 1.0)) {
+			mode_lib->vba.ScaleRatioAndTapsSupport = false;
+		} else if (mode_lib->vba.vtaps[k] < 1.0 || mode_lib->vba.vtaps[k] > 8.0
+				|| mode_lib->vba.htaps[k] < 1.0 || mode_lib->vba.htaps[k] > 8.0
+				|| (mode_lib->vba.htaps[k] > 1.0
+						&& (mode_lib->vba.htaps[k] % 2) == 1)
+				|| mode_lib->vba.HRatio[k] > mode_lib->vba.MaxHSCLRatio
+				|| mode_lib->vba.VRatio[k] > mode_lib->vba.MaxVSCLRatio
+				|| mode_lib->vba.HRatio[k] > mode_lib->vba.htaps[k]
+				|| mode_lib->vba.VRatio[k] > mode_lib->vba.vtaps[k]
+				|| (mode_lib->vba.SourcePixelFormat[k] != dm_444_64
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_444_32
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_444_16
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_mono_16
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_mono_8
+						&& (mode_lib->vba.HRatio[k] / 2.0
+								> mode_lib->vba.HTAPsChroma[k]
+								|| mode_lib->vba.VRatio[k] / 2.0
+										> mode_lib->vba.VTAPsChroma[k]))) {
+			mode_lib->vba.ScaleRatioAndTapsSupport = false;
+		}
+	}
+	/*Source Format, Pixel Format and Scan Support Check*/
+
+	mode_lib->vba.SourceFormatPixelAndScanSupport = true;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if ((mode_lib->vba.SurfaceTiling[k] == dm_sw_linear
+				&& mode_lib->vba.SourceScan[k] != dm_horz)
+				|| ((mode_lib->vba.SurfaceTiling[k] == dm_sw_4kb_d
+						|| mode_lib->vba.SurfaceTiling[k] == dm_sw_4kb_d_x
+						|| mode_lib->vba.SurfaceTiling[k] == dm_sw_64kb_d
+						|| mode_lib->vba.SurfaceTiling[k] == dm_sw_64kb_d_t
+						|| mode_lib->vba.SurfaceTiling[k] == dm_sw_64kb_d_x
+						|| mode_lib->vba.SurfaceTiling[k] == dm_sw_var_d
+						|| mode_lib->vba.SurfaceTiling[k] == dm_sw_var_d_x)
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_444_64)
+				|| (mode_lib->vba.SurfaceTiling[k] == dm_sw_64kb_r_x
+						&& (mode_lib->vba.SourcePixelFormat[k] == dm_mono_8
+								|| mode_lib->vba.SourcePixelFormat[k]
+										== dm_420_8
+								|| mode_lib->vba.SourcePixelFormat[k]
+										== dm_420_10))
+				|| (((mode_lib->vba.SurfaceTiling[k] == dm_sw_gfx7_2d_thin_gl
+						|| mode_lib->vba.SurfaceTiling[k]
+								== dm_sw_gfx7_2d_thin_l_vp)
+						&& !((mode_lib->vba.SourcePixelFormat[k]
+								== dm_444_64
+								|| mode_lib->vba.SourcePixelFormat[k]
+										== dm_444_32)
+								&& mode_lib->vba.SourceScan[k]
+										== dm_horz
+								&& mode_lib->vba.SupportGFX7CompatibleTilingIn32bppAnd64bpp
+										== true
+								&& mode_lib->vba.DCCEnable[k]
+										== false))
+						|| (mode_lib->vba.DCCEnable[k] == true
+								&& (mode_lib->vba.SurfaceTiling[k]
+										== dm_sw_linear
+										|| mode_lib->vba.SourcePixelFormat[k]
+												== dm_420_8
+										|| mode_lib->vba.SourcePixelFormat[k]
+												== dm_420_10)))) {
+			mode_lib->vba.SourceFormatPixelAndScanSupport = false;
+		}
+	}
+	/*Bandwidth Support Check*/
+
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.SourcePixelFormat[k] == dm_444_64) {
+			locals->BytePerPixelInDETY[k] = 8.0;
+			locals->BytePerPixelInDETC[k] = 0.0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_32) {
+			locals->BytePerPixelInDETY[k] = 4.0;
+			locals->BytePerPixelInDETC[k] = 0.0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_444_16
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_mono_16) {
+			locals->BytePerPixelInDETY[k] = 2.0;
+			locals->BytePerPixelInDETC[k] = 0.0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_mono_8) {
+			locals->BytePerPixelInDETY[k] = 1.0;
+			locals->BytePerPixelInDETC[k] = 0.0;
+		} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_8) {
+			locals->BytePerPixelInDETY[k] = 1.0;
+			locals->BytePerPixelInDETC[k] = 2.0;
+		} else {
+			locals->BytePerPixelInDETY[k] = 4.0 / 3;
+			locals->BytePerPixelInDETC[k] = 8.0 / 3;
+		}
+		if (mode_lib->vba.SourceScan[k] == dm_horz) {
+			locals->SwathWidthYSingleDPP[k] = mode_lib->vba.ViewportWidth[k];
+		} else {
+			locals->SwathWidthYSingleDPP[k] = mode_lib->vba.ViewportHeight[k];
+		}
+	}
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		locals->ReadBandwidthLuma[k] = locals->SwathWidthYSingleDPP[k] * dml_ceil(locals->BytePerPixelInDETY[k], 1.0)
+				/ (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k]) * mode_lib->vba.VRatio[k];
+		locals->ReadBandwidthChroma[k] = locals->SwathWidthYSingleDPP[k] / 2 * dml_ceil(locals->BytePerPixelInDETC[k], 2.0)
+				/ (mode_lib->vba.HTotal[k] / mode_lib->vba.PixelClock[k]) * mode_lib->vba.VRatio[k] / 2.0;
+		locals->ReadBandwidth[k] = locals->ReadBandwidthLuma[k] + locals->ReadBandwidthChroma[k];
+	}
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.WritebackEnable[k] == true
+				&& mode_lib->vba.WritebackPixelFormat[k] == dm_444_32) {
+			locals->WriteBandwidth[k] = mode_lib->vba.WritebackDestinationWidth[k]
+					* mode_lib->vba.WritebackDestinationHeight[k]
+					/ (mode_lib->vba.WritebackSourceHeight[k]
+							* mode_lib->vba.HTotal[k]
+							/ mode_lib->vba.PixelClock[k]) * 4.0;
+		} else if (mode_lib->vba.WritebackEnable[k] == true
+				&& mode_lib->vba.WritebackPixelFormat[k] == dm_420_10) {
+			locals->WriteBandwidth[k] = mode_lib->vba.WritebackDestinationWidth[k]
+					* mode_lib->vba.WritebackDestinationHeight[k]
+					/ (mode_lib->vba.WritebackSourceHeight[k]
+							* mode_lib->vba.HTotal[k]
+							/ mode_lib->vba.PixelClock[k]) * 3.0;
+		} else if (mode_lib->vba.WritebackEnable[k] == true) {
+			locals->WriteBandwidth[k] = mode_lib->vba.WritebackDestinationWidth[k]
+					* mode_lib->vba.WritebackDestinationHeight[k]
+					/ (mode_lib->vba.WritebackSourceHeight[k]
+							* mode_lib->vba.HTotal[k]
+							/ mode_lib->vba.PixelClock[k]) * 1.5;
+		} else {
+			locals->WriteBandwidth[k] = 0.0;
+		}
+	}
+	mode_lib->vba.DCCEnabledInAnyPlane = false;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.DCCEnable[k] == true) {
+			mode_lib->vba.DCCEnabledInAnyPlane = true;
+		}
+	}
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		locals->IdealSDPPortBandwidthPerState[i][0] = dml_min3(
+				mode_lib->vba.ReturnBusWidth * mode_lib->vba.DCFCLKPerState[i],
+				mode_lib->vba.DRAMSpeedPerState[i] * mode_lib->vba.NumberOfChannels
+						* mode_lib->vba.DRAMChannelWidth,
+				mode_lib->vba.FabricClockPerState[i]
+						* mode_lib->vba.FabricDatapathToDCNDataReturn);
+		if (mode_lib->vba.HostVMEnable == false) {
+			locals->ReturnBWPerState[i][0] = locals->IdealSDPPortBandwidthPerState[i][0]
+					* mode_lib->vba.PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelDataOnly / 100.0;
+		} else {
+			locals->ReturnBWPerState[i][0] = locals->IdealSDPPortBandwidthPerState[i][0]
+					* mode_lib->vba.PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData / 100.0;
+		}
+	}
+	/*Writeback Latency support check*/
+
+	mode_lib->vba.WritebackLatencySupport = true;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.WritebackEnable[k] == true) {
+			if (mode_lib->vba.WritebackPixelFormat[k] == dm_444_32) {
+				if (locals->WriteBandwidth[k]
+						> (mode_lib->vba.WritebackInterfaceLumaBufferSize
+								+ mode_lib->vba.WritebackInterfaceChromaBufferSize)
+								/ mode_lib->vba.WritebackLatency) {
+					mode_lib->vba.WritebackLatencySupport = false;
+				}
+			} else {
+				if (locals->WriteBandwidth[k]
+						> 1.5
+								* dml_min(
+										mode_lib->vba.WritebackInterfaceLumaBufferSize,
+										2.0
+												* mode_lib->vba.WritebackInterfaceChromaBufferSize)
+								/ mode_lib->vba.WritebackLatency) {
+					mode_lib->vba.WritebackLatencySupport = false;
+				}
+			}
+		}
+	}
+	/*Re-ordering Buffer Support Check*/
+
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		locals->UrgentRoundTripAndOutOfOrderLatencyPerState[i] =
+				(mode_lib->vba.RoundTripPingLatencyCycles + 32.0) / mode_lib->vba.DCFCLKPerState[i]
+				+ dml_max3(mode_lib->vba.UrgentOutOfOrderReturnPerChannelPixelDataOnly,
+						mode_lib->vba.UrgentOutOfOrderReturnPerChannelPixelMixedWithVMData,
+						mode_lib->vba.UrgentOutOfOrderReturnPerChannelVMDataOnly)
+					* mode_lib->vba.NumberOfChannels / locals->ReturnBWPerState[i][0];
+		if ((mode_lib->vba.ROBBufferSizeInKByte - mode_lib->vba.PixelChunkSizeInKByte) * 1024.0 / locals->ReturnBWPerState[i][0]
+				> locals->UrgentRoundTripAndOutOfOrderLatencyPerState[i]) {
+			locals->ROBSupport[i][0] = true;
+		} else {
+			locals->ROBSupport[i][0] = false;
+		}
+	}
+	/*Writeback Mode Support Check*/
+
+	mode_lib->vba.TotalNumberOfActiveWriteback = 0;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.WritebackEnable[k] == true) {
+			if (mode_lib->vba.ActiveWritebacksPerPlane[k] == 0)
+				mode_lib->vba.ActiveWritebacksPerPlane[k] = 1;
+			mode_lib->vba.TotalNumberOfActiveWriteback =
+					mode_lib->vba.TotalNumberOfActiveWriteback
+							+ mode_lib->vba.ActiveWritebacksPerPlane[k];
+		}
+	}
+	mode_lib->vba.WritebackModeSupport = true;
+	if (mode_lib->vba.TotalNumberOfActiveWriteback > mode_lib->vba.MaxNumWriteback) {
+		mode_lib->vba.WritebackModeSupport = false;
+	}
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.WritebackEnable[k] == true
+				&& mode_lib->vba.Writeback10bpc420Supported != true
+				&& mode_lib->vba.WritebackPixelFormat[k] == dm_420_10) {
+			mode_lib->vba.WritebackModeSupport = false;
+		}
+	}
+	/*Writeback Scale Ratio and Taps Support Check*/
+
+	mode_lib->vba.WritebackScaleRatioAndTapsSupport = true;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.WritebackEnable[k] == true) {
+			if (mode_lib->vba.WritebackLumaAndChromaScalingSupported == false
+					&& (mode_lib->vba.WritebackHRatio[k] != 1.0
+							|| mode_lib->vba.WritebackVRatio[k] != 1.0)) {
+				mode_lib->vba.WritebackScaleRatioAndTapsSupport = false;
+			}
+			if (mode_lib->vba.WritebackHRatio[k] > mode_lib->vba.WritebackMaxHSCLRatio
+					|| mode_lib->vba.WritebackVRatio[k]
+							> mode_lib->vba.WritebackMaxVSCLRatio
+					|| mode_lib->vba.WritebackHRatio[k]
+							< mode_lib->vba.WritebackMinHSCLRatio
+					|| mode_lib->vba.WritebackVRatio[k]
+							< mode_lib->vba.WritebackMinVSCLRatio
+					|| mode_lib->vba.WritebackLumaHTaps[k]
+							> mode_lib->vba.WritebackMaxHSCLTaps
+					|| mode_lib->vba.WritebackLumaVTaps[k]
+							> mode_lib->vba.WritebackMaxVSCLTaps
+					|| mode_lib->vba.WritebackHRatio[k]
+							> mode_lib->vba.WritebackLumaHTaps[k]
+					|| mode_lib->vba.WritebackVRatio[k]
+							> mode_lib->vba.WritebackLumaVTaps[k]
+					|| (mode_lib->vba.WritebackLumaHTaps[k] > 2.0
+							&& ((mode_lib->vba.WritebackLumaHTaps[k] % 2)
+									== 1))
+					|| (mode_lib->vba.WritebackPixelFormat[k] != dm_444_32
+							&& (mode_lib->vba.WritebackChromaHTaps[k]
+									> mode_lib->vba.WritebackMaxHSCLTaps
+									|| mode_lib->vba.WritebackChromaVTaps[k]
+											> mode_lib->vba.WritebackMaxVSCLTaps
+									|| 2.0
+											* mode_lib->vba.WritebackHRatio[k]
+											> mode_lib->vba.WritebackChromaHTaps[k]
+									|| 2.0
+											* mode_lib->vba.WritebackVRatio[k]
+											> mode_lib->vba.WritebackChromaVTaps[k]
+									|| (mode_lib->vba.WritebackChromaHTaps[k] > 2.0
+										&& ((mode_lib->vba.WritebackChromaHTaps[k] % 2) == 1))))) {
+				mode_lib->vba.WritebackScaleRatioAndTapsSupport = false;
+			}
+			if (mode_lib->vba.WritebackVRatio[k] < 1.0) {
+				mode_lib->vba.WritebackLumaVExtra =
+						dml_max(1.0 - 2.0 / dml_ceil(1.0 / mode_lib->vba.WritebackVRatio[k], 1.0), 0.0);
+			} else {
+				mode_lib->vba.WritebackLumaVExtra = -1;
+			}
+			if ((mode_lib->vba.WritebackPixelFormat[k] == dm_444_32
+					&& mode_lib->vba.WritebackLumaVTaps[k]
+							> (mode_lib->vba.WritebackLineBufferLumaBufferSize
+									+ mode_lib->vba.WritebackLineBufferChromaBufferSize)
+									/ 3.0
+									/ mode_lib->vba.WritebackDestinationWidth[k]
+									- mode_lib->vba.WritebackLumaVExtra)
+					|| (mode_lib->vba.WritebackPixelFormat[k] == dm_420_8
+							&& mode_lib->vba.WritebackLumaVTaps[k]
+									> mode_lib->vba.WritebackLineBufferLumaBufferSize
+											* 8.0 / 10.0 / mode_lib->vba.WritebackDestinationWidth[k]
+											- mode_lib->vba.WritebackLumaVExtra)
+					|| (mode_lib->vba.WritebackPixelFormat[k] == dm_420_10
+							&& mode_lib->vba.WritebackLumaVTaps[k]
+									> mode_lib->vba.WritebackLineBufferLumaBufferSize
+											* 8.0 / 10.0
+											/ mode_lib->vba.WritebackDestinationWidth[k]
+											- mode_lib->vba.WritebackLumaVExtra)) {
+				mode_lib->vba.WritebackScaleRatioAndTapsSupport = false;
+			}
+			if (2.0 * mode_lib->vba.WritebackVRatio[k] < 1) {
+				mode_lib->vba.WritebackChromaVExtra = 0.0;
+			} else {
+				mode_lib->vba.WritebackChromaVExtra = -1;
+			}
+			if ((mode_lib->vba.WritebackPixelFormat[k] == dm_420_8
+					&& mode_lib->vba.WritebackChromaVTaps[k]
+							> mode_lib->vba.WritebackLineBufferChromaBufferSize
+									* 8.0 / 10.0 / mode_lib->vba.WritebackDestinationWidth[k]
+									- mode_lib->vba.WritebackChromaVExtra)
+					|| (mode_lib->vba.WritebackPixelFormat[k] == dm_420_10
+							&& mode_lib->vba.WritebackChromaVTaps[k]
+									> mode_lib->vba.WritebackLineBufferChromaBufferSize
+											* 8.0 / 10.0
+											/ mode_lib->vba.WritebackDestinationWidth[k]
+											- mode_lib->vba.WritebackChromaVExtra)) {
+				mode_lib->vba.WritebackScaleRatioAndTapsSupport = false;
+			}
+		}
+	}
+	/*Maximum DISPCLK/DPPCLK Support check*/
+
+	mode_lib->vba.WritebackRequiredDISPCLK = 0.0;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.WritebackEnable[k] == true) {
+			mode_lib->vba.WritebackRequiredDISPCLK =
+					dml_max(
+							mode_lib->vba.WritebackRequiredDISPCLK,
+							CalculateWriteBackDISPCLK(
+									mode_lib->vba.WritebackPixelFormat[k],
+									mode_lib->vba.PixelClock[k],
+									mode_lib->vba.WritebackHRatio[k],
+									mode_lib->vba.WritebackVRatio[k],
+									mode_lib->vba.WritebackLumaHTaps[k],
+									mode_lib->vba.WritebackLumaVTaps[k],
+									mode_lib->vba.WritebackChromaHTaps[k],
+									mode_lib->vba.WritebackChromaVTaps[k],
+									mode_lib->vba.WritebackDestinationWidth[k],
+									mode_lib->vba.HTotal[k],
+									mode_lib->vba.WritebackChromaLineBufferWidth));
+		}
+	}
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.HRatio[k] > 1.0) {
+			locals->PSCL_FACTOR[k] = dml_min(
+					mode_lib->vba.MaxDCHUBToPSCLThroughput,
+					mode_lib->vba.MaxPSCLToLBThroughput
+							* mode_lib->vba.HRatio[k]
+							/ dml_ceil(
+									mode_lib->vba.htaps[k]
+											/ 6.0,
+									1.0));
+		} else {
+			locals->PSCL_FACTOR[k] = dml_min(
+					mode_lib->vba.MaxDCHUBToPSCLThroughput,
+					mode_lib->vba.MaxPSCLToLBThroughput);
+		}
+		if (locals->BytePerPixelInDETC[k] == 0.0) {
+			locals->PSCL_FACTOR_CHROMA[k] = 0.0;
+			locals->MinDPPCLKUsingSingleDPP[k] =
+					mode_lib->vba.PixelClock[k]
+							* dml_max3(
+									mode_lib->vba.vtaps[k] / 6.0
+											* dml_min(
+													1.0,
+													mode_lib->vba.HRatio[k]),
+									mode_lib->vba.HRatio[k]
+											* mode_lib->vba.VRatio[k]
+											/ locals->PSCL_FACTOR[k],
+									1.0);
+			if ((mode_lib->vba.htaps[k] > 6.0 || mode_lib->vba.vtaps[k] > 6.0)
+					&& locals->MinDPPCLKUsingSingleDPP[k]
+							< 2.0 * mode_lib->vba.PixelClock[k]) {
+				locals->MinDPPCLKUsingSingleDPP[k] = 2.0
+						* mode_lib->vba.PixelClock[k];
+			}
+		} else {
+			if (mode_lib->vba.HRatio[k] / 2.0 > 1.0) {
+				locals->PSCL_FACTOR_CHROMA[k] =
+						dml_min(
+								mode_lib->vba.MaxDCHUBToPSCLThroughput,
+								mode_lib->vba.MaxPSCLToLBThroughput
+										* mode_lib->vba.HRatio[k]
+										/ 2.0
+										/ dml_ceil(
+												mode_lib->vba.HTAPsChroma[k]
+														/ 6.0,
+												1.0));
+			} else {
+				locals->PSCL_FACTOR_CHROMA[k] = dml_min(
+						mode_lib->vba.MaxDCHUBToPSCLThroughput,
+						mode_lib->vba.MaxPSCLToLBThroughput);
+			}
+			locals->MinDPPCLKUsingSingleDPP[k] =
+					mode_lib->vba.PixelClock[k]
+							* dml_max5(
+									mode_lib->vba.vtaps[k] / 6.0
+											* dml_min(
+													1.0,
+													mode_lib->vba.HRatio[k]),
+									mode_lib->vba.HRatio[k]
+											* mode_lib->vba.VRatio[k]
+											/ locals->PSCL_FACTOR[k],
+									mode_lib->vba.VTAPsChroma[k]
+											/ 6.0
+											* dml_min(
+													1.0,
+													mode_lib->vba.HRatio[k]
+															/ 2.0),
+									mode_lib->vba.HRatio[k]
+											* mode_lib->vba.VRatio[k]
+											/ 4.0
+											/ locals->PSCL_FACTOR_CHROMA[k],
+									1.0);
+			if ((mode_lib->vba.htaps[k] > 6.0 || mode_lib->vba.vtaps[k] > 6.0
+					|| mode_lib->vba.HTAPsChroma[k] > 6.0
+					|| mode_lib->vba.VTAPsChroma[k] > 6.0)
+					&& locals->MinDPPCLKUsingSingleDPP[k]
+							< 2.0 * mode_lib->vba.PixelClock[k]) {
+				locals->MinDPPCLKUsingSingleDPP[k] = 2.0
+						* mode_lib->vba.PixelClock[k];
+			}
+		}
+	}
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		Calculate256BBlockSizes(
+				mode_lib->vba.SourcePixelFormat[k],
+				mode_lib->vba.SurfaceTiling[k],
+				dml_ceil(locals->BytePerPixelInDETY[k], 1.0),
+				dml_ceil(locals->BytePerPixelInDETC[k], 2.0),
+				&locals->Read256BlockHeightY[k],
+				&locals->Read256BlockHeightC[k],
+				&locals->Read256BlockWidthY[k],
+				&locals->Read256BlockWidthC[k]);
+		if (mode_lib->vba.SourceScan[k] == dm_horz) {
+			locals->MaxSwathHeightY[k] = locals->Read256BlockHeightY[k];
+			locals->MaxSwathHeightC[k] = locals->Read256BlockHeightC[k];
+		} else {
+			locals->MaxSwathHeightY[k] = locals->Read256BlockWidthY[k];
+			locals->MaxSwathHeightC[k] = locals->Read256BlockWidthC[k];
+		}
+		if ((mode_lib->vba.SourcePixelFormat[k] == dm_444_64
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_32
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_444_16
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_mono_16
+				|| mode_lib->vba.SourcePixelFormat[k] == dm_mono_8)) {
+			if (mode_lib->vba.SurfaceTiling[k] == dm_sw_linear
+					|| (mode_lib->vba.SourcePixelFormat[k] == dm_444_64
+							&& (mode_lib->vba.SurfaceTiling[k]
+									== dm_sw_4kb_s
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_4kb_s_x
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_64kb_s
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_64kb_s_t
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_64kb_s_x
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_var_s
+									|| mode_lib->vba.SurfaceTiling[k]
+											== dm_sw_var_s_x)
+							&& mode_lib->vba.SourceScan[k] == dm_horz)) {
+				locals->MinSwathHeightY[k] = locals->MaxSwathHeightY[k];
+			} else {
+				locals->MinSwathHeightY[k] = locals->MaxSwathHeightY[k]
+						/ 2.0;
+			}
+			locals->MinSwathHeightC[k] = locals->MaxSwathHeightC[k];
+		} else {
+			if (mode_lib->vba.SurfaceTiling[k] == dm_sw_linear) {
+				locals->MinSwathHeightY[k] = locals->MaxSwathHeightY[k];
+				locals->MinSwathHeightC[k] = locals->MaxSwathHeightC[k];
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_8
+					&& mode_lib->vba.SourceScan[k] == dm_horz) {
+				locals->MinSwathHeightY[k] = locals->MaxSwathHeightY[k]
+						/ 2.0;
+				locals->MinSwathHeightC[k] = locals->MaxSwathHeightC[k];
+			} else if (mode_lib->vba.SourcePixelFormat[k] == dm_420_10
+					&& mode_lib->vba.SourceScan[k] == dm_horz) {
+				locals->MinSwathHeightC[k] = locals->MaxSwathHeightC[k]
+						/ 2.0;
+				locals->MinSwathHeightY[k] = locals->MaxSwathHeightY[k];
+			} else {
+				locals->MinSwathHeightY[k] = locals->MaxSwathHeightY[k];
+				locals->MinSwathHeightC[k] = locals->MaxSwathHeightC[k];
+			}
+		}
+		if (mode_lib->vba.SurfaceTiling[k] == dm_sw_linear) {
+			mode_lib->vba.MaximumSwathWidthSupport = 8192.0;
+		} else {
+			mode_lib->vba.MaximumSwathWidthSupport = 5120.0;
+		}
+		mode_lib->vba.MaximumSwathWidthInDETBuffer =
+				dml_min(
+						mode_lib->vba.MaximumSwathWidthSupport,
+						mode_lib->vba.DETBufferSizeInKByte[0] * 1024.0 / 2.0
+								/ (locals->BytePerPixelInDETY[k]
+										* locals->MinSwathHeightY[k]
+										+ locals->BytePerPixelInDETC[k]
+												/ 2.0
+												* locals->MinSwathHeightC[k]));
+		if (locals->BytePerPixelInDETC[k] == 0.0) {
+			mode_lib->vba.MaximumSwathWidthInLineBuffer =
+					mode_lib->vba.LineBufferSize
+							* dml_max(mode_lib->vba.HRatio[k], 1.0)
+							/ mode_lib->vba.LBBitPerPixel[k]
+							/ (mode_lib->vba.vtaps[k]
+									+ dml_max(
+											dml_ceil(
+													mode_lib->vba.VRatio[k],
+													1.0)
+													- 2,
+											0.0));
+		} else {
+			mode_lib->vba.MaximumSwathWidthInLineBuffer =
+					dml_min(
+							mode_lib->vba.LineBufferSize
+									* dml_max(
+											mode_lib->vba.HRatio[k],
+											1.0)
+									/ mode_lib->vba.LBBitPerPixel[k]
+									/ (mode_lib->vba.vtaps[k]
+											+ dml_max(
+													dml_ceil(
+															mode_lib->vba.VRatio[k],
+															1.0)
+															- 2,
+													0.0)),
+							2.0 * mode_lib->vba.LineBufferSize
+									* dml_max(
+											mode_lib->vba.HRatio[k]
+													/ 2.0,
+											1.0)
+									/ mode_lib->vba.LBBitPerPixel[k]
+									/ (mode_lib->vba.VTAPsChroma[k]
+											+ dml_max(
+													dml_ceil(
+															mode_lib->vba.VRatio[k]
+																	/ 2.0,
+															1.0)
+															- 2,
+													0.0)));
+		}
+		locals->MaximumSwathWidth[k] = dml_min(
+				mode_lib->vba.MaximumSwathWidthInDETBuffer,
+				mode_lib->vba.MaximumSwathWidthInLineBuffer);
+	}
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		double MaxMaxDispclkRoundedDown = RoundToDFSGranularityDown(
+			mode_lib->vba.MaxDispclk[mode_lib->vba.soc.num_states],
+			mode_lib->vba.DISPCLKDPPCLKVCOSpeed);
+
+		for (j = 0; j < 2; j++) {
+			mode_lib->vba.MaxDispclkRoundedDownToDFSGranularity = RoundToDFSGranularityDown(
+				mode_lib->vba.MaxDispclk[i],
+				mode_lib->vba.DISPCLKDPPCLKVCOSpeed);
+			mode_lib->vba.MaxDppclkRoundedDownToDFSGranularity = RoundToDFSGranularityDown(
+				mode_lib->vba.MaxDppclk[i],
+				mode_lib->vba.DISPCLKDPPCLKVCOSpeed);
+			locals->RequiredDISPCLK[i][j] = 0.0;
+			locals->DISPCLK_DPPCLK_Support[i][j] = true;
+			for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+				mode_lib->vba.PlaneRequiredDISPCLKWithoutODMCombine =
+						mode_lib->vba.PixelClock[k]
+								* (1.0
+										+ mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading
+												/ 100.0)
+								* (1.0
+										+ mode_lib->vba.DISPCLKRampingMargin
+												/ 100.0);
+				if (mode_lib->vba.PlaneRequiredDISPCLKWithoutODMCombine >= mode_lib->vba.MaxDispclk[i]
+						&& i == mode_lib->vba.soc.num_states)
+					mode_lib->vba.PlaneRequiredDISPCLKWithoutODMCombine = mode_lib->vba.PixelClock[k]
+							* (1 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0);
+
+				mode_lib->vba.PlaneRequiredDISPCLKWithODMCombine = mode_lib->vba.PixelClock[k] / 2
+					* (1 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) * (1 + mode_lib->vba.DISPCLKRampingMargin / 100.0);
+				if (mode_lib->vba.PlaneRequiredDISPCLKWithODMCombine >= mode_lib->vba.MaxDispclk[i]
+						&& i == mode_lib->vba.soc.num_states)
+					mode_lib->vba.PlaneRequiredDISPCLKWithODMCombine = mode_lib->vba.PixelClock[k] / 2
+							* (1 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0);
+
+				locals->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_disabled;
+				mode_lib->vba.PlaneRequiredDISPCLK = mode_lib->vba.PlaneRequiredDISPCLKWithoutODMCombine;
+				if (mode_lib->vba.ODMCapability) {
+					if (locals->PlaneRequiredDISPCLKWithoutODMCombine > MaxMaxDispclkRoundedDown) {
+						locals->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_2to1;
+						mode_lib->vba.PlaneRequiredDISPCLK = mode_lib->vba.PlaneRequiredDISPCLKWithODMCombine;
+					} else if (locals->DSCEnabled[k] && (locals->HActive[k] > DCN21_MAX_DSC_IMAGE_WIDTH)) {
+						locals->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_2to1;
+						mode_lib->vba.PlaneRequiredDISPCLK = mode_lib->vba.PlaneRequiredDISPCLKWithODMCombine;
+					} else if (locals->HActive[k] > DCN21_MAX_420_IMAGE_WIDTH && locals->OutputFormat[k] == dm_420) {
+						locals->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_2to1;
+						mode_lib->vba.PlaneRequiredDISPCLK = mode_lib->vba.PlaneRequiredDISPCLKWithODMCombine;
+					}
+				}
+
+				if (locals->MinDPPCLKUsingSingleDPP[k] * (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) <= mode_lib->vba.MaxDppclkRoundedDownToDFSGranularity
+						&& locals->SwathWidthYSingleDPP[k] <= locals->MaximumSwathWidth[k]
+						&& locals->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_disabled) {
+					locals->NoOfDPP[i][j][k] = 1;
+					locals->RequiredDPPCLK[i][j][k] =
+						locals->MinDPPCLKUsingSingleDPP[k] * (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0);
+				} else {
+					locals->NoOfDPP[i][j][k] = 2;
+					locals->RequiredDPPCLK[i][j][k] =
+						locals->MinDPPCLKUsingSingleDPP[k] * (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) / 2.0;
+				}
+				locals->RequiredDISPCLK[i][j] = dml_max(
+						locals->RequiredDISPCLK[i][j],
+						mode_lib->vba.PlaneRequiredDISPCLK);
+				if ((locals->MinDPPCLKUsingSingleDPP[k] / locals->NoOfDPP[i][j][k] * (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0)
+						> mode_lib->vba.MaxDppclkRoundedDownToDFSGranularity)
+						|| (mode_lib->vba.PlaneRequiredDISPCLK > mode_lib->vba.MaxDispclkRoundedDownToDFSGranularity)) {
+					locals->DISPCLK_DPPCLK_Support[i][j] = false;
+				}
+			}
+			locals->TotalNumberOfActiveDPP[i][j] = 0.0;
+			for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++)
+				locals->TotalNumberOfActiveDPP[i][j] = locals->TotalNumberOfActiveDPP[i][j] + locals->NoOfDPP[i][j][k];
+			if (j == 1) {
+				while (locals->TotalNumberOfActiveDPP[i][j] < mode_lib->vba.MaxNumDPP
+						&& locals->TotalNumberOfActiveDPP[i][j] < 2 * mode_lib->vba.NumberOfActivePlanes) {
+					double BWOfNonSplitPlaneOfMaximumBandwidth;
+					unsigned int NumberOfNonSplitPlaneOfMaximumBandwidth;
+
+					BWOfNonSplitPlaneOfMaximumBandwidth = 0;
+					NumberOfNonSplitPlaneOfMaximumBandwidth = 0;
+					for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+						if (locals->ReadBandwidth[k] > BWOfNonSplitPlaneOfMaximumBandwidth && locals->NoOfDPP[i][j][k] == 1) {
+							BWOfNonSplitPlaneOfMaximumBandwidth = locals->ReadBandwidth[k];
+							NumberOfNonSplitPlaneOfMaximumBandwidth = k;
+						}
+					}
+					locals->NoOfDPP[i][j][NumberOfNonSplitPlaneOfMaximumBandwidth] = 2;
+					locals->RequiredDPPCLK[i][j][NumberOfNonSplitPlaneOfMaximumBandwidth] =
+						locals->MinDPPCLKUsingSingleDPP[NumberOfNonSplitPlaneOfMaximumBandwidth]
+							* (1 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100) / 2;
+					locals->TotalNumberOfActiveDPP[i][j] = locals->TotalNumberOfActiveDPP[i][j] + 1;
+				}
+			}
+			if (locals->TotalNumberOfActiveDPP[i][j] > mode_lib->vba.MaxNumDPP) {
+				locals->RequiredDISPCLK[i][j] = 0.0;
+				locals->DISPCLK_DPPCLK_Support[i][j] = true;
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+					locals->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_disabled;
+					if (locals->SwathWidthYSingleDPP[k] <= locals->MaximumSwathWidth[k]) {
+						locals->NoOfDPP[i][j][k] = 1;
+						locals->RequiredDPPCLK[i][j][k] = locals->MinDPPCLKUsingSingleDPP[k]
+							* (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0);
+					} else {
+						locals->NoOfDPP[i][j][k] = 2;
+						locals->RequiredDPPCLK[i][j][k] = locals->MinDPPCLKUsingSingleDPP[k]
+										* (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) / 2.0;
+					}
+					if (i != mode_lib->vba.soc.num_states) {
+						mode_lib->vba.PlaneRequiredDISPCLK =
+								mode_lib->vba.PixelClock[k]
+										* (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0)
+										* (1.0 + mode_lib->vba.DISPCLKRampingMargin / 100.0);
+					} else {
+						mode_lib->vba.PlaneRequiredDISPCLK = mode_lib->vba.PixelClock[k]
+							* (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0);
+					}
+					locals->RequiredDISPCLK[i][j] = dml_max(
+							locals->RequiredDISPCLK[i][j],
+							mode_lib->vba.PlaneRequiredDISPCLK);
+					if (locals->MinDPPCLKUsingSingleDPP[k] / locals->NoOfDPP[i][j][k] * (1.0 + mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0)
+							> mode_lib->vba.MaxDppclkRoundedDownToDFSGranularity
+							|| mode_lib->vba.PlaneRequiredDISPCLK > mode_lib->vba.MaxDispclkRoundedDownToDFSGranularity)
+						locals->DISPCLK_DPPCLK_Support[i][j] = false;
+				}
+				locals->TotalNumberOfActiveDPP[i][j] = 0.0;
+				for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++)
+					locals->TotalNumberOfActiveDPP[i][j] = locals->TotalNumberOfActiveDPP[i][j] + locals->NoOfDPP[i][j][k];
+			}
+			locals->RequiredDISPCLK[i][j] = dml_max(
+					locals->RequiredDISPCLK[i][j],
+					mode_lib->vba.WritebackRequiredDISPCLK);
+			if (mode_lib->vba.MaxDispclkRoundedDownToDFSGranularity
+					< mode_lib->vba.WritebackRequiredDISPCLK) {
+				locals->DISPCLK_DPPCLK_Support[i][j] = false;
+			}
+		}
+	}
+	/*Viewport Size Check*/
+
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		locals->ViewportSizeSupport[i][0] = true;
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			if (locals->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_2to1) {
+				if (dml_min(locals->SwathWidthYSingleDPP[k], dml_round(mode_lib->vba.HActive[k] / 2.0 * mode_lib->vba.HRatio[k]))
+						> locals->MaximumSwathWidth[k]) {
+					locals->ViewportSizeSupport[i][0] = false;
+				}
+			} else {
+				if (locals->SwathWidthYSingleDPP[k] / 2.0 > locals->MaximumSwathWidth[k]) {
+					locals->ViewportSizeSupport[i][0] = false;
+				}
+			}
+		}
+	}
+	/*Total Available Pipes Support Check*/
+
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		for (j = 0; j < 2; j++) {
+			if (locals->TotalNumberOfActiveDPP[i][j] <= mode_lib->vba.MaxNumDPP)
+				locals->TotalAvailablePipesSupport[i][j] = true;
+			else
+				locals->TotalAvailablePipesSupport[i][j] = false;
+		}
+	}
+	/*Total Available OTG Support Check*/
+
+	mode_lib->vba.TotalNumberOfActiveOTG = 0.0;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (mode_lib->vba.BlendingAndTiming[k] == k) {
+			mode_lib->vba.TotalNumberOfActiveOTG = mode_lib->vba.TotalNumberOfActiveOTG
+					+ 1.0;
+		}
+	}
+	if (mode_lib->vba.TotalNumberOfActiveOTG <= mode_lib->vba.MaxNumOTG) {
+		mode_lib->vba.NumberOfOTGSupport = true;
+	} else {
+		mode_lib->vba.NumberOfOTGSupport = false;
+	}
+	/*Display IO and DSC Support Check*/
+
+	mode_lib->vba.NonsupportedDSCInputBPC = false;
+	for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+		if (!(mode_lib->vba.DSCInputBitPerComponent[k] == 12.0
+				|| mode_lib->vba.DSCInputBitPerComponent[k] == 10.0
+				|| mode_lib->vba.DSCInputBitPerComponent[k] == 8.0)) {
+			mode_lib->vba.NonsupportedDSCInputBPC = true;
+		}
+	}
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			locals->RequiresDSC[i][k] = false;
+			locals->RequiresFEC[i][k] = 0;
+			if (mode_lib->vba.BlendingAndTiming[k] == k) {
+				if (mode_lib->vba.Output[k] == dm_hdmi) {
+					locals->RequiresDSC[i][k] = false;
+					locals->RequiresFEC[i][k] = 0;
+					locals->OutputBppPerState[i][k] = TruncToValidBPP(
+							dml_min(600.0, mode_lib->vba.PHYCLKPerState[i]) / mode_lib->vba.PixelClockBackEnd[k] * 24,
+							mode_lib->vba.ForcedOutputLinkBPP[k],
+							false,
+							mode_lib->vba.Output[k],
+							mode_lib->vba.OutputFormat[k],
+							mode_lib->vba.DSCInputBitPerComponent[k]);
+				} else if (mode_lib->vba.Output[k] == dm_dp
+						|| mode_lib->vba.Output[k] == dm_edp) {
+					if (mode_lib->vba.Output[k] == dm_edp) {
+						mode_lib->vba.EffectiveFECOverhead = 0.0;
+					} else {
+						mode_lib->vba.EffectiveFECOverhead =
+								mode_lib->vba.FECOverhead;
+					}
+					if (mode_lib->vba.PHYCLKPerState[i] >= 270.0) {
+						mode_lib->vba.Outbpp = TruncToValidBPP(
+								(1.0 - mode_lib->vba.Downspreading / 100.0) * 270.0
+								* mode_lib->vba.OutputLinkDPLanes[k] / mode_lib->vba.PixelClockBackEnd[k] * 8.0,
+								mode_lib->vba.ForcedOutputLinkBPP[k],
+								false,
+								mode_lib->vba.Output[k],
+								mode_lib->vba.OutputFormat[k],
+								mode_lib->vba.DSCInputBitPerComponent[k]);
+						mode_lib->vba.OutbppDSC = TruncToValidBPP(
+								(1.0 - mode_lib->vba.Downspreading / 100.0) * (1.0 - mode_lib->vba.EffectiveFECOverhead / 100.0) * 270.0
+								* mode_lib->vba.OutputLinkDPLanes[k] / mode_lib->vba.PixelClockBackEnd[k] * 8.0,
+								mode_lib->vba.ForcedOutputLinkBPP[k],
+								true,
+								mode_lib->vba.Output[k],
+								mode_lib->vba.OutputFormat[k],
+								mode_lib->vba.DSCInputBitPerComponent[k]);
+						if (mode_lib->vba.DSCEnabled[k] == true) {
+							locals->RequiresDSC[i][k] = true;
+							if (mode_lib->vba.Output[k] == dm_dp) {
+								locals->RequiresFEC[i][k] = true;
+							} else {
+								locals->RequiresFEC[i][k] = false;
+							}
+							mode_lib->vba.Outbpp = mode_lib->vba.OutbppDSC;
+						} else {
+							locals->RequiresDSC[i][k] = false;
+							locals->RequiresFEC[i][k] = false;
+						}
+						locals->OutputBppPerState[i][k] = mode_lib->vba.Outbpp;
+					}
+					if (mode_lib->vba.Outbpp == BPP_INVALID && mode_lib->vba.PHYCLKPerState[i] >= 540.0) {
+						mode_lib->vba.Outbpp = TruncToValidBPP(
+								(1.0 - mode_lib->vba.Downspreading / 100.0) * 540.0
+								* mode_lib->vba.OutputLinkDPLanes[k] / mode_lib->vba.PixelClockBackEnd[k] * 8.0,
+								mode_lib->vba.ForcedOutputLinkBPP[k],
+									false,
+									mode_lib->vba.Output[k],
+									mode_lib->vba.OutputFormat[k],
+									mode_lib->vba.DSCInputBitPerComponent[k]);
+						mode_lib->vba.OutbppDSC = TruncToValidBPP(
+								(1.0 - mode_lib->vba.Downspreading / 100.0) * (1.0 - mode_lib->vba.EffectiveFECOverhead / 100.0) * 540.0
+								* mode_lib->vba.OutputLinkDPLanes[k] / mode_lib->vba.PixelClockBackEnd[k] * 8.0,
+								mode_lib->vba.ForcedOutputLinkBPP[k],
+								true,
+								mode_lib->vba.Output[k],
+								mode_lib->vba.OutputFormat[k],
+								mode_lib->vba.DSCInputBitPerComponent[k]);
+						if (mode_lib->vba.DSCEnabled[k] == true) {
+							locals->RequiresDSC[i][k] = true;
+							if (mode_lib->vba.Output[k] == dm_dp) {
+								locals->RequiresFEC[i][k] = true;
+							} else {
+								locals->RequiresFEC[i][k] = false;
+							}
+							mode_lib->vba.Outbpp = mode_lib->vba.OutbppDSC;
+						} else {
+							locals->RequiresDSC[i][k] = false;
+							locals->RequiresFEC[i][k] = false;
+						}
+						locals->OutputBppPerState[i][k] = mode_lib->vba.Outbpp;
+					}
+					if (mode_lib->vba.Outbpp == BPP_INVALID
+							&& mode_lib->vba.PHYCLKPerState[i]
+									>= 810.0) {
+						mode_lib->vba.Outbpp = TruncToValidBPP(
+								(1.0 - mode_lib->vba.Downspreading / 100.0) * 810.0
+								* mode_lib->vba.OutputLinkDPLanes[k] / mode_lib->vba.PixelClockBackEnd[k] * 8.0,
+								mode_lib->vba.ForcedOutputLinkBPP[k],
+								false,
+								mode_lib->vba.Output[k],
+								mode_lib->vba.OutputFormat[k],
+								mode_lib->vba.DSCInputBitPerComponent[k]);
+						mode_lib->vba.OutbppDSC = TruncToValidBPP(
+								(1.0 - mode_lib->vba.Downspreading / 100.0) * (1.0 - mode_lib->vba.EffectiveFECOverhead / 100.0) * 810.0
+								* mode_lib->vba.OutputLinkDPLanes[k] / mode_lib->vba.PixelClockBackEnd[k] * 8.0,
+								mode_lib->vba.ForcedOutputLinkBPP[k],
+								true,
+								mode_lib->vba.Output[k],
+								mode_lib->vba.OutputFormat[k],
+								mode_lib->vba.DSCInputBitPerComponent[k]);
+						if (mode_lib->vba.DSCEnabled[k] == true || mode_lib->vba.Outbpp == BPP_INVALID) {
+							locals->RequiresDSC[i][k] = true;
+							if (mode_lib->vba.Output[k] == dm_dp) {
+								locals->RequiresFEC[i][k] = true;
+							} else {
+								locals->RequiresFEC[i][k] = false;
+							}
+							mode_lib->vba.Outbpp = mode_lib->vba.OutbppDSC;
+						} else {
+							locals->RequiresDSC[i][k] = false;
+							locals->RequiresFEC[i][k] = false;
+						}
+						locals->OutputBppPerState[i][k] =
+								mode_lib->vba.Outbpp;
+					}
+				}
+			} else {
+				locals->OutputBppPerState[i][k] = BPP_BLENDED_PIPE;
+			}
+		}
+	}
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		locals->DIOSupport[i] = true;
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			if (!mode_lib->vba.skip_dio_check[k]
+					&& (locals->OutputBppPerState[i][k] == BPP_INVALID
+						|| (mode_lib->vba.OutputFormat[k] == dm_420
+							&& mode_lib->vba.Interlace[k] == true
+							&& mode_lib->vba.ProgressiveToInterlaceUnitInOPP == true))) {
+				locals->DIOSupport[i] = false;
+			}
+		}
+	}
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			locals->DSCCLKRequiredMoreThanSupported[i] = false;
+			if (mode_lib->vba.BlendingAndTiming[k] == k) {
+				if ((mode_lib->vba.Output[k] == dm_dp
+						|| mode_lib->vba.Output[k] == dm_edp)) {
+					if (mode_lib->vba.OutputFormat[k] == dm_420
+							|| mode_lib->vba.OutputFormat[k]
+									== dm_n422) {
+						mode_lib->vba.DSCFormatFactor = 2;
+					} else {
+						mode_lib->vba.DSCFormatFactor = 1;
+					}
+					if (locals->RequiresDSC[i][k] == true) {
+						if (locals->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_2to1) {
+							if (mode_lib->vba.PixelClockBackEnd[k] / 6.0 / mode_lib->vba.DSCFormatFactor
+									> (1.0 - mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) * mode_lib->vba.MaxDSCCLK[i]) {
+								locals->DSCCLKRequiredMoreThanSupported[i] =
+										true;
+							}
+						} else {
+							if (mode_lib->vba.PixelClockBackEnd[k] / 3.0 / mode_lib->vba.DSCFormatFactor
+									> (1.0 - mode_lib->vba.DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) * mode_lib->vba.MaxDSCCLK[i]) {
+								locals->DSCCLKRequiredMoreThanSupported[i] =
+										true;
+							}
+						}
+					}
+				}
+			}
+		}
+	}
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		locals->NotEnoughDSCUnits[i] = false;
+		mode_lib->vba.TotalDSCUnitsRequired = 0.0;
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			if (locals->RequiresDSC[i][k] == true) {
+				if (locals->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_2to1) {
+					mode_lib->vba.TotalDSCUnitsRequired =
+							mode_lib->vba.TotalDSCUnitsRequired + 2.0;
+				} else {
+					mode_lib->vba.TotalDSCUnitsRequired =
+							mode_lib->vba.TotalDSCUnitsRequired + 1.0;
+				}
+			}
+		}
+		if (mode_lib->vba.TotalDSCUnitsRequired > mode_lib->vba.NumberOfDSC) {
+			locals->NotEnoughDSCUnits[i] = true;
+		}
+	}
+	/*DSC Delay per state*/
+
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			if (mode_lib->vba.BlendingAndTiming[k] != k) {
+				mode_lib->vba.slices = 0;
+			} else if (locals->RequiresDSC[i][k] == 0
+					|| locals->RequiresDSC[i][k] == false) {
+				mode_lib->vba.slices = 0;
+			} else if (mode_lib->vba.PixelClockBackEnd[k] > 3200.0) {
+				mode_lib->vba.slices = dml_ceil(
+						mode_lib->vba.PixelClockBackEnd[k] / 400.0,
+						4.0);
+			} else if (mode_lib->vba.PixelClockBackEnd[k] > 1360.0) {
+				mode_lib->vba.slices = 8.0;
+			} else if (mode_lib->vba.PixelClockBackEnd[k] > 680.0) {
+				mode_lib->vba.slices = 4.0;
+			} else if (mode_lib->vba.PixelClockBackEnd[k] > 340.0) {
+				mode_lib->vba.slices = 2.0;
+			} else {
+				mode_lib->vba.slices = 1.0;
+			}
+			if (locals->OutputBppPerState[i][k] == BPP_BLENDED_PIPE
+					|| locals->OutputBppPerState[i][k] == BPP_INVALID) {
+				mode_lib->vba.bpp = 0.0;
+			} else {
+				mode_lib->vba.bpp = locals->OutputBppPerState[i][k];
+			}
+			if (locals->RequiresDSC[i][k] == true && mode_lib->vba.bpp != 0.0) {
+				if (locals->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_disabled) {
+					locals->DSCDelayPerState[i][k] =
+							dscceComputeDelay(
+									mode_lib->vba.DSCInputBitPerComponent[k],
+									mode_lib->vba.bpp,
+									dml_ceil(
+											mode_lib->vba.HActive[k]
+													/ mode_lib->vba.slices,
+											1.0),
+									mode_lib->vba.slices,
+									mode_lib->vba.OutputFormat[k])
+									+ dscComputeDelay(
+											mode_lib->vba.OutputFormat[k]);
+				} else {
+					locals->DSCDelayPerState[i][k] =
+							2.0 * (dscceComputeDelay(
+											mode_lib->vba.DSCInputBitPerComponent[k],
+											mode_lib->vba.bpp,
+											dml_ceil(mode_lib->vba.HActive[k] / mode_lib->vba.slices, 1.0),
+											mode_lib->vba.slices / 2,
+											mode_lib->vba.OutputFormat[k])
+									+ dscComputeDelay(mode_lib->vba.OutputFormat[k]));
+				}
+				locals->DSCDelayPerState[i][k] =
+						locals->DSCDelayPerState[i][k] * mode_lib->vba.PixelClock[k] / mode_lib->vba.PixelClockBackEnd[k];
+			} else {
+				locals->DSCDelayPerState[i][k] = 0.0;
+			}
+		}
+		for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+			for (m = 0; m <= mode_lib->vba.NumberOfActivePlanes - 1; m++) {
+				for (j = 0; j <= mode_lib->vba.NumberOfActivePlanes - 1; j++) {
+					if (mode_lib->vba.BlendingAndTiming[k] == m && locals->RequiresDSC[i][m] == true)
+						locals->DSCDelayPerState[i][k] = locals->DSCDelayPerState[i][m];
+				}
+			}
+		}
+	}
+
+	//Prefetch Check
+	for (i = 0; i <= mode_lib->vba.soc.num_states; ++i) {
+		for (j = 0; j <= 1; ++j) {
+			locals->TotalNumberOfDCCActiveDPP[i][j] = 0;
+			for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) {
+				if (mode_lib->vba.DCCEnable[k] == true)
+					locals->TotalNumberOfDCCActiveDPP[i][j] = locals->TotalNumberOfDCCActiveDPP[i][j] + locals->NoOfDPP[i][j][k];
+			}
+		}
+	}
+
+	mode_lib->vba.UrgentLatency = dml_max3(
+			mode_lib->vba.UrgentLatencyPixelDataOnly,
+			mode_lib->vba.UrgentLatencyPixelMixedWithVMData,
+			mode_lib->vba.UrgentLatencyVMDataOnly);
+	mode_lib->vba.PrefetchERROR = CalculateMinAndMaxPrefetchMode(
+			mode_lib->vba.AllowDRAMSelfRefreshOrDRAMClockChangeInVblank,
+			&mode_lib->vba.MinPrefetchMode,
+			&mode_lib->vba.MaxPrefetchMode);
+
+	for (i = 0; i <= mode_lib->vba.soc.num_states; i++) {
+		for (j = 0; j < 2; j++) {
+			for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+				locals->RequiredDPPCLKThisState[k] = locals->RequiredDPPCLK[i][j][k];
+				locals->NoOfDPPThisState[k]        = locals->NoOfDPP[i][j][k];
+				if (locals->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_2to1) {
+					locals->SwathWidthYThisState[k] =
+						dml_min(locals->SwathWidthYSingleDPP[k], dml_round(mode_lib->vba.HActive[k] / 2.0 * mode_lib->vba.HRatio[k]));
+				} else {
+					locals->SwathWidthYThisState[k] = locals->SwathWidthYSingleDPP[k] / locals->NoOfDPP[i][j][k];
+				}
+				mode_lib->vba.SwathWidthGranularityY = 256.0
+					/ dml_ceil(locals->BytePerPixelInDETY[k], 1.0)
+					/ locals->MaxSwathHeightY[k];
+				mode_lib->vba.RoundedUpMaxSwathSizeBytesY =
+					(dml_ceil(locals->SwathWidthYThisState[k] - 1.0, mode_lib->vba.SwathWidthGranularityY)
+					+ mode_lib->vba.SwathWidthGranularityY) * locals->BytePerPixelInDETY[k] * locals->MaxSwathHeightY[k];
+				if (mode_lib->vba.SourcePixelFormat[k] == dm_420_10) {
+					mode_lib->vba.RoundedUpMaxSwathSizeBytesY = dml_ceil(
+							mode_lib->vba.RoundedUpMaxSwathSizeBytesY,
+							256.0) + 256;
+				}
+				if (locals->MaxSwathHeightC[k] > 0.0) {
+					mode_lib->vba.SwathWidthGranularityC = 256.0 / dml_ceil(locals->BytePerPixelInDETC[k], 2.0) / locals->MaxSwathHeightC[k];
+					mode_lib->vba.RoundedUpMaxSwathSizeBytesC = (dml_ceil(locals->SwathWidthYThisState[k] / 2.0 - 1.0, mode_lib->vba.SwathWidthGranularityC)
+							+ mode_lib->vba.SwathWidthGranularityC) * locals->BytePerPixelInDETC[k] * locals->MaxSwathHeightC[k];
+					if (mode_lib->vba.SourcePixelFormat[k] == dm_420_10) {
+						mode_lib->vba.RoundedUpMaxSwathSizeBytesC = dml_ceil(mode_lib->vba.RoundedUpMaxSwathSizeBytesC, 256.0) + 256;
+					}
+				} else {
+					mode_lib->vba.RoundedUpMaxSwathSizeBytesC = 0.0;
+				}
+				if (mode_lib->vba.RoundedUpMaxSwathSizeBytesY + mode_lib->vba.RoundedUpMaxSwathSizeBytesC
+						<= mode_lib->vba.DETBufferSizeInKByte[0] * 1024.0 / 2.0) {
+					locals->SwathHeightYThisState[k] = locals->MaxSwathHeightY[k];
+					locals->SwathHeightCThisState[k] = locals->MaxSwathHeightC[k];
+				} else {
+					locals->SwathHeightYThisState[k] =
+							locals->MinSwathHeightY[k];
+					locals->SwathHeightCThisState[k] =
+							locals->MinSwathHeightC[k];
+				}
+			}
+
+			CalculateDCFCLKDeepSleep(
+					mode_lib,
+					mode_lib->vba.NumberOfActivePlanes,
+					locals->BytePerPixelInDETY,
+					locals->BytePerPixelInDETC,
+					mode_lib->vba.VRatio,
+					locals->SwathWidthYThisState,
+					locals->NoOfDPPThisState,
+					mode_lib->vba.HRatio,
+					mode_lib->vba.PixelClock,
+					locals->PSCL_FACTOR,
+					locals->PSCL_FACTOR_CHROMA,
+					locals->RequiredDPPCLKThisState,
+					&mode_lib->vba.ProjectedDCFCLKDeepSleep[0][0]);
+
+			for (k = 0; k <= mode_lib->vba.NumberOfActivePlanes - 1; k++) {
+				if ((mode_lib->vba.SourcePixelFormat[k] != dm_444_64
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_444_32
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_444_16
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_mono_16
+						&& mode_lib->vba.SourcePixelFormat[k] != dm_mono_8)) {
+					mode_lib->vba.PDEAndMetaPTEBytesPerFrameC = CalculateVMAndRowBytes(
+							mode_lib,
+							mode_lib->vba.DCCEnable[k],
+							locals->Read256BlockHeightC[k],
+							locals->Read256BlockWidthC[k],
+							mode_lib->vba.SourcePixelFormat[k],
+							mode_lib->vba.SurfaceTiling[k],
+							dml_ceil(locals->BytePerPixelInDETC[k], 2.0),
+							mode_lib->vba.SourceScan[k],
+							mode_lib->vba.ViewportWidth[k] / 2.0,
+							mode_lib->vba.ViewportHeight[k] / 2.0,
+							locals->SwathWidthYThisState[k] / 2.0,
+							mode_lib->vba.GPUVMEnable,
+							mode_lib->vba.HostVMEnable,
+							mode_lib->vba.HostVMMaxPageTableLevels,
+							mode_lib->vba.HostVMCachedPageTableLevels,
+							mode_lib->vba.VMMPageSize,
+							mode_lib->vba.PTEBufferSizeInRequestsChroma,
+							mode_lib->vba.PitchC[k],
+							0.0,
+							&locals->MacroTileWidthC[k],
+							&mode_lib->vba.MetaRowBytesC,
+							&mode_lib->vba.DPTEBytesPerRowC,
+							&locals->PTEBufferSizeNotExceededC[i][j][k],
+							locals->dpte_row_width_chroma_ub,
+							&locals->dpte_row_height_chroma[k],
+							&locals->meta_req_width_chroma[k],
+							&locals->meta_req_height_chroma[k],
+							&locals->meta_row_width_chroma[k],
+							&locals->meta_row_height_chroma[k],
+							&locals->vm_group_bytes_chroma,
+							&locals->dpte_group_bytes_chroma,
+							locals->PixelPTEReqWidthC,
+							locals->PixelPTEReqHeightC,
+							locals->PTERequestSizeC,
+							locals->dpde0_bytes_per_frame_ub_c,
+							locals->meta_pte_bytes_per_frame_ub_c);
+					locals->PrefetchLinesC[0][0][k] = CalculatePrefetchSourceLines(
+							mode_lib,
+							mode_lib->vba.VRatio[k]/2,
+							mode_lib->vba.VTAPsChroma[k],
+							mode_lib->vba.Interlace[k],
+							mode_lib->vba.ProgressiveToInterlaceUnitInOPP,
+							locals->SwathHeightCThisState[k],
+							mode_lib->vba.ViewportYStartC[k],
+							&locals->PrefillC[k],
+							&locals->MaxNumSwC[k]);
+					locals->PTEBufferSizeInRequestsForLuma = mode_lib->vba.PTEBufferSizeInRequestsLuma;
+				} else {
+					mode_lib->vba.PDEAndMetaPTEBytesPerFrameC = 0.0;
+					mode_lib->vba.MetaRowBytesC = 0.0;
+					mode_lib->vba.DPTEBytesPerRowC = 0.0;
+					locals->PrefetchLinesC[0][0][k] = 0.0;
+					locals->PTEBufferSizeNotExceededC[i][j][k] = true;
+					locals->PTEBufferSizeInRequestsForLuma = mode_lib->vba.PTEBufferSizeInRequestsLuma + mode_lib->vba.PTEBufferSizeInRequestsChroma;
+				}
+				mode_lib->vba.PDEAndMetaPTEBytesPerFrameY = CalculateVMAndRowBytes(
+						mode_lib,
+						mode_lib->vba.DCCEnable[k],
+						locals->Read256BlockHeightY[k],
+						locals->Read256BlockWidthY[k],
+						mode_lib->vba.SourcePixelFormat[k],
+						mode_lib->vba.SurfaceTiling[k],
+						dml_ceil(locals->BytePerPixelInDETY[k], 1.0),
+						mode_lib->vba.SourceScan[k],
+						mode_lib->vba.ViewportWidth[k],
+						mode_lib->vba.ViewportHeight[k],
+						locals->SwathWidthYThisState[k],
+						mode_lib->vba.GPUVMEnable,
+						mode_lib->vba.HostVMEnable,
+						mode_lib->vba.HostVMMaxPageTableLevels,
+						mode_lib->vba.HostVMCachedPageTableLevels,
+						mode_lib->vba.VMMPageSize,
+						locals->PTEBufferSizeInRequestsForLuma,
+						mode_lib->vba.PitchY[k],
+						mode_lib->vba.DCCMetaPitchY[k],
+						&locals->MacroTileWidthY[k],
+						&mode_lib->vba.MetaRowBytesY,